From 4d698495eae6912db94dcdedb0c3b01c63143646 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 11:16:07 +0300
Subject: [PATCH 001/242] gguf : init

---
 .gitignore             |  1 +
 Makefile               |  7 +++++--
 examples/gguf/gguf.cpp | 34 ++++++++++++++++++++++++++++++++++
 ggml.h                 | 25 ++++++++++++++++++++++++-
 4 files changed, 64 insertions(+), 3 deletions(-)
 create mode 100644 examples/gguf/gguf.cpp

diff --git a/.gitignore b/.gitignore
index c1ab6bb6d08a3..abe8e28cb7687 100644
--- a/.gitignore
+++ b/.gitignore
@@ -45,6 +45,7 @@ models-mnt
 /server
 /Pipfile
 /embd-input-test
+/gguf
 /libllama.so
 build-info.h
 arm_neon.h
diff --git a/Makefile b/Makefile
index fb7c27cd972bb..e19acfbb24a3f 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test gguf
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0
@@ -330,7 +330,7 @@ libllama.so: llama.o ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) -shared -fPIC -o $@ $^ $(LDFLAGS)
 
 clean:
-	rm -vf *.o *.so *.dll main quantize quantize-stats perplexity embedding benchmark-matmult save-load-state server simple vdot train-text-from-scratch embd-input-test build-info.h $(TEST_TARGETS)
+	rm -vf *.o *.so *.dll main quantize quantize-stats perplexity embedding benchmark-matmult save-load-state server simple vdot train-text-from-scratch embd-input-test gguf build-info.h $(TEST_TARGETS)
 
 #
 # Examples
@@ -370,6 +370,9 @@ $(LIB_PRE)embdinput$(DSO_EXT): examples/embd-input/embd-input.h examples/embd-in
 embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-test.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %$(DSO_EXT),$(filter-out %.h,$(filter-out %.hpp,$^))) -o $@ $(LDFLAGS) -L. -lembdinput
 
+gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o $(OBJS)
+	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
+
 train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
new file mode 100644
index 0000000000000..602de519adb1e
--- /dev/null
+++ b/examples/gguf/gguf.cpp
@@ -0,0 +1,34 @@
+#include "ggml.h"
+
+#include <cstdio>
+#include <string>
+
+bool gguf_write(const std::string & fname) {
+
+
+    return true;
+}
+
+bool gguf_read(const std::string & fname) {
+    return true;
+}
+
+int main(int argc, char ** argv) {
+    if (argc < 3) {
+        fprintf(stdout, "usage: %s data.gguf r|w\n", argv[0]);
+        return -1;
+    }
+
+    const std::string fname(argv[1]);
+    const std::string mode(argv[2]);
+
+    GGML_ASSERT((mode == "r" || mode == "w") && "mode must be r or w");
+
+    if (mode == "w") {
+        GGML_ASSERT(gguf_write(fname) && "failed to write gguf file");
+    } else if (mode == "r") {
+        GGML_ASSERT(gguf_read(fname)  && "failed to read gguf file");
+    }
+
+    return 0;
+}
diff --git a/ggml.h b/ggml.h
index 9919cce7c263f..2e700c9a06645 100644
--- a/ggml.h
+++ b/ggml.h
@@ -190,6 +190,9 @@
 #define GGML_FILE_MAGIC   0x67676d6c // "ggml"
 #define GGML_FILE_VERSION 1
 
+#define GGUF_FILE_MAGIC   0x47475546 // "GGUF"
+#define GGUF_FILE_VERSION 1
+
 #define GGML_QNT_VERSION        2    // bump this on quantization format changes
 #define GGML_QNT_VERSION_FACTOR 1000 // do not change this
 
@@ -202,7 +205,6 @@
 #define GGML_MAX_OP_PARAMS     32
 #define GGML_DEFAULT_N_THREADS 4
 
-
 #define GGML_EXIT_SUCCESS 0
 #define GGML_EXIT_ABORTED 1
 
@@ -1611,6 +1613,27 @@ extern "C" {
 
     GGML_API size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, int start, int n, int64_t * hist);
 
+    //
+    // gguf
+    //
+
+    enum gguf_metadata_value_type {
+        GGUF_METADATA_VALUE_TYPE_UINT8   = 0,
+        GGUF_METADATA_VALUE_TYPE_INT8    = 1,
+        GGUF_METADATA_VALUE_TYPE_UINT16  = 2,
+        GGUF_METADATA_VALUE_TYPE_INT16   = 3,
+        GGUF_METADATA_VALUE_TYPE_UINT32  = 4,
+        GGUF_METADATA_VALUE_TYPE_INT32   = 5,
+        GGUF_METADATA_VALUE_TYPE_FLOAT32 = 6,
+        GGUF_METADATA_VALUE_TYPE_BOOL    = 7,
+        GGUF_METADATA_VALUE_TYPE_STRING  = 8,
+        GGUF_METADATA_VALUE_TYPE_ARRAY   = 9,
+    };
+
+    struct gguf_string {
+        uint32_t n;
+        char * data;
+    };
     //
     // system info
     //

From bae6b125f6d6148d3bebb774f1b73ecc67dc0051 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Wed, 26 Jul 2023 11:17:05 +0300
Subject: [PATCH 002/242] wip : implement GGUF (#2397)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* Add LLAMA_DEFAULT_RMS_EPS so we can change the default (#2384)

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>

* WIP: python class to write GGUF, incomplete C apı for reading

---------

Co-authored-by: Kawrakow <48489457+ikawrakow@users.noreply.github.com>
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
---
 constants.py |  32 +++++++
 gguf.c       | 192 ++++++++++++++++++++++++++++++++++++++
 gguf.py      | 257 +++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 481 insertions(+)
 create mode 100644 constants.py
 create mode 100644 gguf.c
 create mode 100644 gguf.py

diff --git a/constants.py b/constants.py
new file mode 100644
index 0000000000000..7c7456403aaeb
--- /dev/null
+++ b/constants.py
@@ -0,0 +1,32 @@
+GGUF_MAGIC = 0x47475546
+GGUF_VERSION = 1
+
+# general
+KEY_GENERAL_ARCHITECTURE = "general.architecture"
+KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
+KEY_GENERAL_NAME = "general.name"
+KEY_GENERAL_AUTHOR = "general.author"
+KEY_GENERAL_URL = "general.url"
+KEY_GENERAL_DESCRIPTION = "general.description"
+KEY_GENERAL_FILE_TYPE = "general.file_type"
+KEY_GENERAL_LICENSE = "general.license"
+KEY_GENERAL_SOURCE_URL = "general.source.url"
+KEY_GENERAL_SOURCE_HF_REPO = "general.source.hugginface.repository"
+
+# LLM
+KEY_LLM_CONTEXT_LENGTH = "{llm}.context_length"
+KEY_LLM_EMBEDDING_LENGTH = "{llm}.embedding_length"
+KEY_LLM_LAYER_COUNT = "{llm}.layer_count"
+KEY_LLM_FEED_FORWARD_LENGTH = "{llm}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL = "{llm}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT = "{llm}.tensor_data_layout"
+
+# attention
+KEY_ATTENTION_HEAD_COUNT = "{llm}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV = "{llm}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS = "{llm}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV = "{llm}.attention.clamp_kqv"
+
+# RoPE
+KEY_ROPE_DIMENSION_COUNT = "{llm}.rope.dimension_count"
+KEY_ROPE_SCALE = "{llm}.rope.scale"
diff --git a/gguf.c b/gguf.c
new file mode 100644
index 0000000000000..54b31d411f35b
--- /dev/null
+++ b/gguf.c
@@ -0,0 +1,192 @@
+// TODO: convert to proper gguf.h gguf.c structure, now I'm trying to be fast as much as possible,
+// and everything is in this file for quick debugging.
+
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <stdbool.h>
+
+
+enum ggml_type {
+    GGML_TYPE_F32  = 0,
+    GGML_TYPE_F16  = 1,
+    GGML_TYPE_Q4_0 = 2,
+    GGML_TYPE_Q4_1 = 3,
+    // GGML_TYPE_Q4_2 = 4, support has been removed
+    // GGML_TYPE_Q4_3 (5) support has been removed
+    GGML_TYPE_Q5_0 = 6,
+    GGML_TYPE_Q5_1 = 7,
+    GGML_TYPE_Q8_0 = 8,
+    GGML_TYPE_Q8_1 = 9,
+    // k-quantizations
+    GGML_TYPE_Q2_K = 10,
+    GGML_TYPE_Q3_K = 11,
+    GGML_TYPE_Q4_K = 12,
+    GGML_TYPE_Q5_K = 13,
+    GGML_TYPE_Q6_K = 14,
+    GGML_TYPE_Q8_K = 15,
+    GGML_TYPE_I8,
+    GGML_TYPE_I16,
+    GGML_TYPE_I32,
+    GGML_TYPE_COUNT,
+};
+
+enum gguf_metadata_value_type {
+    GGUF_METADATA_VALUE_TYPE_UINT8 = 0,
+    GGUF_METADATA_VALUE_TYPE_INT8 = 1,
+    GGUF_METADATA_VALUE_TYPE_UINT16 = 2,
+    GGUF_METADATA_VALUE_TYPE_INT16 = 3,
+    GGUF_METADATA_VALUE_TYPE_UINT32 = 4,
+    GGUF_METADATA_VALUE_TYPE_INT32 = 5,
+    GGUF_METADATA_VALUE_TYPE_FLOAT32 = 6,
+    GGUF_METADATA_VALUE_TYPE_BOOL = 7,
+    GGUF_METADATA_VALUE_TYPE_STRING = 8,
+    GGUF_METADATA_VALUE_TYPE_ARRAY = 9,
+};
+
+struct gguf_string_t {
+    uint32_t len;
+    char * string;
+};
+
+union gguf_metadata_value_t;
+
+// Union definition for gguf_metadata_value_t
+union gguf_metadata_value_t {
+    uint8_t uint8;
+    int8_t int8;
+    uint16_t uint16;
+    int16_t int16;
+    uint32_t uint32;
+    int32_t int32;
+    float float32;
+    bool bool_;
+    struct gguf_string_t string;
+    struct {
+        uint32_t len;
+        enum gguf_metadata_value_type type;
+        union gguf_metadata_value_t * array;
+    } array;
+};
+
+
+struct gguf_metadata_kv_t {
+    struct gguf_string_t key;
+    uint32_t value_len;
+    enum gguf_metadata_value_type value_type;
+    union gguf_metadata_value_t* value;
+};
+
+struct gguf_header_t {
+    uint32_t magic;
+    uint32_t version;
+    uint32_t tensor_count;
+    uint32_t metadata_kv_count;
+    struct gguf_metadata_kv_t * metadata_kv;
+};
+
+struct gguf_tensor_info_t {
+    struct gguf_string_t name;
+    uint32_t n_dimensions;
+    uint32_t dimensions[];
+};
+
+struct gguf_file_t {
+    struct gguf_header_t header;
+    uint8_t tensor_data[];
+};
+
+void read_gguf_file(const char * file_path, struct gguf_file_t * gguf_file) {
+    FILE* file = fopen(file_path, "rb");
+    if (file == NULL) {
+        printf("Error opening the file.\n");
+        return;
+    }
+
+    fread(&gguf_file->header.magic, sizeof(uint32_t), 1, file);
+
+    // Verify magic and version
+    if (gguf_file->header.magic != 0x47475546) {
+        printf("Invalid magic number. Not a valid GGUF file.\n");
+        fclose(file);
+        return;
+    }
+
+    fread(&gguf_file->header.version, sizeof(uint32_t), 1, file);
+
+    if (gguf_file->header.version != 1) {
+        printf("Unsupported version. Expected version 1.\n");
+        fclose(file);
+        return;
+    }
+
+    fread(&gguf_file->header.tensor_count, sizeof(uint32_t), 1, file);
+    fread(&gguf_file->header.metadata_kv_count, sizeof(uint32_t), 1, file);
+
+    printf("Magic: %x\n", gguf_file->header.magic);
+    printf("Version: %d\n", gguf_file->header.version);
+    printf("Tensor Count: %d\n", gguf_file->header.tensor_count);
+    printf("Metadata Key-Value Count: %d\n", gguf_file->header.metadata_kv_count);
+
+    gguf_file->header.metadata_kv = (struct gguf_metadata_kv_t*)malloc(gguf_file->header.metadata_kv_count * sizeof(struct gguf_metadata_kv_t));
+
+    for (int i = 0; i < gguf_file->header.metadata_kv_count; i++) {
+        struct gguf_metadata_kv_t* kv = &gguf_file->header.metadata_kv[i];
+        fread(&kv->key.len, sizeof(uint32_t), 1, file);
+        kv->key.string = (char*)malloc(kv->key.len ); // Allocate memory for the key string
+        fread(kv->key.string, sizeof(char), kv->key.len, file);
+        //kv->key.string[kv->key.len] = '\0'; // Null-terminate the key string
+
+        fread(&kv->value_type, sizeof(uint32_t), 1, file);
+
+        printf("Metadata Value Type: %d\n", kv->value_type);
+        printf("Metadata Key: %s\n", kv->key.string);
+
+        // Read metadata value according to its type using reinterpret_cast
+        switch (kv->value_type) {
+            case GGUF_METADATA_VALUE_TYPE_UINT32:
+            kv->value = (uint32_t *) malloc(sizeof(uint32_t));
+            fread(kv->value, sizeof(uint32_t), 1, file);
+            printf("value: %d\n", kv->value->uint32);
+            break;
+            case GGUF_METADATA_VALUE_TYPE_FLOAT32:
+            kv->value = (float *)malloc(sizeof(float));
+            fread(kv->value, sizeof(float), 1, file);
+            printf("value: %f\n", (float)kv->value->float32);
+            break;
+            case GGUF_METADATA_VALUE_TYPE_STRING:
+            fread(&kv->value_len, sizeof(uint32_t), 1, file);
+            printf("value len: %d\n", kv->value_len);
+kv->value = (char *)malloc(sizeof(char) * kv->value_len); // Allocate memory for the value string
+fread(kv->value, sizeof(char), kv->value_len, file);
+        printf("value: %s\n", (char *)kv->value);
+        break;
+            // ... (handle other types in a similar manner)
+            default:
+                printf("Unsupported metadata value type.\n");
+                fclose(file);
+                return;
+        }
+    }
+
+    // TODO: handle reading tensor data
+
+    fclose(file);
+}
+
+void gguf_free(struct gguf_file_t * gguf_file) {
+    // Free allocated memory for key strings avd values
+    for (int i = 0; i < gguf_file->header.metadata_kv_count; i++) {
+        free(gguf_file->header.metadata_kv[i].key.string);
+        free(gguf_file->header.metadata_kv[i].value);
+    }
+    free(gguf_file->header.metadata_kv);
+}
+
+int main() {
+    const char* file_path = "example.gguf";
+    struct gguf_file_t gguf_file;
+    read_gguf_file(file_path, &gguf_file);
+    gguf_free(&gguf_file);
+    return 0;
+}
diff --git a/gguf.py b/gguf.py
new file mode 100644
index 0000000000000..dfd5ba5bf8490
--- /dev/null
+++ b/gguf.py
@@ -0,0 +1,257 @@
+"""TODOs
+1. Implement writing tensor data with alignment.
+2. Implement writers for known architectures, LLaMA in particular.
+3. Add docstrings from the format specs.
+4. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
+"""
+
+import struct
+from enum import IntEnum
+from typing import List, Any
+import constants
+
+
+class GGMLQuantizationType(IntEnum):
+    F32 = 0
+    F16 = 1
+    QR_0 = 2
+    Q4_1 = 3
+    # Q4_2 = 4 # support has been removed
+    # Q4_3 = 5 # support has been removed
+    Q5_0 = 6
+    Q5_1 = 7
+    Q8_0 = 8
+    Q8_1 = 9
+    Q2_K = 10
+    Q3_K = 11
+    Q4_K = 12
+    Q5_K = 13
+    Q6_K = 14
+    Q8_K = 15
+
+
+class GGUFValueType(IntEnum):
+    UINT8 = 0
+    INT8 = 1
+    UINT16 = 2
+    INT16 = 3
+    UINT32 = 4
+    INT32 = 5
+    FLOAT32 = 6
+    BOOL = 7
+    STRING = 8
+    ARRAY = 9
+
+    @staticmethod
+    def get_type(value):
+        if isinstance(value, str):
+            return GGUFValueType.STRING
+        elif isinstance(value, list):
+            return GGUFValueType.ARRAY
+        elif isinstance(value, float):
+            return GGUFValueType.FLOAT32
+        elif isinstance(value, bool):
+            return GGUFValueType.BOOL
+        else:
+            return GGUFValueType.INT32
+
+
+class GGUFWriter:
+    def __init__(self, buffered_writer):
+        self.buffered_writer = buffered_writer
+
+    def write_header(self, tensor_count: int, metadata_kv_count: int):
+        self.buffered_writer.write(struct.pack("<I", constants.GGUF_MAGIC))
+        self.buffered_writer.write(struct.pack("<I", constants.GGUF_VERSION))
+        self.buffered_writer.write(struct.pack("<I", tensor_count))
+        self.buffered_writer.write(struct.pack("<I", metadata_kv_count))
+
+    @classmethod
+    def open(cls, path: str) -> "GGUFWriter":
+        f = open(path, "wb")
+        return cls(f)
+
+    def write_key(self, key: str, value_type: GGUFValueType):
+        encoded_key = key.encode("utf8")
+        self.buffered_writer.write(struct.pack("<I", len(encoded_key)))
+        self.buffered_writer.write(encoded_key)
+        self.buffered_writer.write(struct.pack("<I", value_type))
+
+    def write_uint8(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.UINT8)
+        self.buffered_writer.write(struct.pack("<B", value))
+
+    def write_int8(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.INT8)
+        self.buffered_writer.write(struct.pack("<b", value))
+
+    def write_uint16(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.UINT16)
+        self.buffered_writer.write(struct.pack("<H", value))
+
+    def write_int16(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.INT16)
+        self.buffered_writer.write(struct.pack("<h", value))
+
+    def write_uint32(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.UINT32)
+        self.buffered_writer.write(struct.pack("<I", value))
+
+    def write_int32(self, key: str, value: int):
+        self.write_key(key, GGUFValueType.INT32)
+        self.buffered_writer.write(struct.pack("<i", value))
+
+    def write_float32(self, key: str, value: float):
+        self.write_key(key, GGUFValueType.FLOAT32)
+        self.buffered_writer.write(struct.pack("<f", value))
+
+    def write_bool(self, key: str, value: bool):
+        self.write_key(key, GGUFValueType.BOOL)
+        self.buffered_writer.write(struct.pack("<?", value))
+
+    def write_string(self, key: str, value: str):
+        self.write_key(key, GGUFValueType.STRING)
+        encoded_string = value.encode('utf-8')
+        self.buffered_writer.write(struct.pack("<I", len(encoded_string)))
+        self.buffered_writer.write(encoded_string)
+
+    def write_array(self, key: str, value: list):
+        if not isinstance(value, list):
+            raise ValueError("Value must be a list for array type")
+
+        self.write_key(key, GGUFValueType.ARRAY)
+
+        self.buffered_writer.write(struct.pack("<I", len(value)))
+
+        for item in value:
+            self.write_value(item)
+
+    def write_value(self: str, value: Any):
+        value_type = GGUFValueType.get_type(value)
+        self.buffered_writer.write(struct.pack("<I", value_type))
+
+        if value_type == GGUFValueType.UINT8:
+            self.buffered_writer.write(struct.pack("<B", value))
+        elif value_type == GGUFValueType.INT8:
+            self.buffered_writer.write(struct.pack("<b", value))
+        elif value_type == GGUFValueType.UINT16:
+            self.buffered_writer.write(struct.pack("<H", value))
+        elif value_type == GGUFValueType.INT16:
+            self.buffered_writer.write(struct.pack("<h", value))
+        elif value_type == GGUFValueType.UINT32:
+            self.buffered_writer.write(struct.pack("<I", value))
+        elif value_type == GGUFValueType.INT32:
+            self.buffered_writer.write(struct.pack("<i", value))
+        elif value_type == GGUFValueType.FLOAT32:
+            self.buffered_writer.write(struct.pack("<f", value))
+        elif value_type == GGUFValueType.BOOL:
+            self.buffered_writer.write(struct.pack("?", value))
+        elif value_type == GGUFValueType.STRING:
+            encoded_value = value.encode("utf8")
+            self.buffered_writer.write(struct.pack("<I", len(encoded_value)))
+            self.buffered_writer.write(encoded_value)
+        elif value_type == GGUFValueType.ARRAY:
+            self.buffered_writer.write(struct.pack("<I", len(value)))
+            for item in value:
+                self.write_value(item)
+        else:
+            raise ValueError("Invalid GGUF metadata value type")
+
+    def flush(self):
+        self.buffered_writer.flush()
+
+    def close(self):
+        self.buffered_writer.close()
+
+    def write_architecture(self, architecture: str):
+        self.write_string(constants.KEY_GENERAL_ARCHITECTURE,
+                          architecture)
+
+    def write_author(self, author: str):
+        self.write_string(constants.KEY_GENERAL_AUTHOR, author)
+
+    def write_url(self, url: str):
+        self.write_string(constants.KEY_GENERAL_URL, url)
+
+    def write_description(self, description: str):
+        self.write_string(constants.KEY_GENERAL_DESCRIPTION, description)
+
+    def write_file_type(self, file_type: str):
+        self.write_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
+
+    def write_source_url(self, url: str):
+        self.write_string(constants.KEY_GENERAL_SOURCE_URL, url)
+
+    def write_source_hf_repo(self, repo: str):
+        self.write_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
+
+    def write_name(self, name: str):
+        self.write_string(constants.KEY_GENERAL_NAME, name)
+
+    def write_quantization_version(self, quantization_version: GGMLQuantizationType):
+        self.write_uint32(
+            constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
+
+    def write_context_length(self, llm: str, length: int):
+        self.write_uint32(
+            constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
+
+    def write_embedding_length(self, llm: str, length: int):
+        self.write_uint32(
+            constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
+
+    def write_layer_count(self, llm: str, length: int):
+        self.write_uint32(
+            constants.KEY_LLM_LAYER_COUNT.format(llm=llm), length)
+
+    def write_feed_forward_length(self, llm: str, length: int):
+        self.write_uint32(
+            constants.KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
+
+    def write_parallel_residual(self, llm: str, use: bool):
+        self.write_bool(
+            constants.KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
+
+    def write_tensor_data_layout(self, llm: str, layout: str):
+        self.write_string(
+            constants.KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
+
+    def write_head_count(self, llm: str, count: int):
+        self.write_uint32(
+            constants.KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
+
+    def write_head_count_kv(self, llm: str, count: int):
+        self.write_uint32(
+            constants.KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
+
+    def write_max_alibi_bias(self, llm: str, bias: float):
+        self.write_float32(
+            constants.KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
+
+    def write_clamp_kqv(self, llm: str, value: float):
+        self.write_float32(
+            constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
+
+    def write_rope_dimension_count(self, llm: str, count: int):
+        self.write_uint32(
+            constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
+
+    def write_rope_scale(self, llm: str, value:  float):
+        self.write_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
+
+
+# Example usage:
+if __name__ == "__main__":
+    # Example usage with a file
+    gguf_writer = GGUFWriter.open("example.gguf")
+    gguf_writer.write_header(0, 3)
+
+gguf_writer.write_architecture("llama")
+gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
+gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
+# Write an array of integers
+#gguf_writer.write_array("simple_array", [1, 2, 3, 4])
+# Write a nested array
+#gguf_writer.write_array("nested", [1, "nested", [2, 3]])
+
+gguf_writer.close()

From 7e82d25f40386540c2c15226300ad998ecd871ea Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 11:26:14 +0300
Subject: [PATCH 003/242] ci : disable CI temporary to not waste energy

---
 .github/ISSUE_TEMPLATE/custom.md   | 185 ---------
 .github/workflows/build.yml        | 632 -----------------------------
 .github/workflows/docker.yml       |  65 ---
 .github/workflows/editorconfig.yml |  17 -
 .github/workflows/tidy-post.yml    |  20 -
 .github/workflows/tidy-review.yml  |  23 --
 6 files changed, 942 deletions(-)
 delete mode 100644 .github/ISSUE_TEMPLATE/custom.md
 delete mode 100644 .github/workflows/build.yml
 delete mode 100644 .github/workflows/docker.yml
 delete mode 100644 .github/workflows/editorconfig.yml
 delete mode 100644 .github/workflows/tidy-post.yml
 delete mode 100644 .github/workflows/tidy-review.yml

diff --git a/.github/ISSUE_TEMPLATE/custom.md b/.github/ISSUE_TEMPLATE/custom.md
deleted file mode 100644
index 8fd9553567780..0000000000000
--- a/.github/ISSUE_TEMPLATE/custom.md
+++ /dev/null
@@ -1,185 +0,0 @@
----
-name: Issue and enhancement template
-about: Used to report issues and request enhancements for llama.cpp
-title: "[User] Insert summary of your issue or enhancement.."
-labels: ''
-assignees: ''
-
----
-
-# Prerequisites
-
-Please answer the following questions for yourself before submitting an issue.
-
-- [ ] I am running the latest code. Development is very rapid so there are no tagged versions as of now.
-- [ ] I carefully followed the [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md).
-- [ ] I [searched using keywords relevant to my issue](https://docs.github.com/en/issues/tracking-your-work-with-issues/filtering-and-searching-issues-and-pull-requests) to make sure that I am creating a new issue that is not already open (or closed).
-- [ ] I reviewed the [Discussions](https://github.com/ggerganov/llama.cpp/discussions), and have a new bug or useful enhancement to share.
-
-# Expected Behavior
-
-Please provide a detailed written description of what you were trying to do, and what you expected `llama.cpp` to do.
-
-# Current Behavior
-
-Please provide a detailed written description of what `llama.cpp` did, instead.
-
-# Environment and Context
-
-Please provide detailed information about your computer setup. This is important in case the issue is not reproducible except for under certain specific conditions.
-
-* Physical (or virtual) hardware you are using, e.g. for Linux:
-
-`$ lscpu`
-
-* Operating System, e.g. for Linux:
-
-`$ uname -a`
-
-* SDK version, e.g. for Linux:
-
-```
-$ python3 --version
-$ make --version
-$ g++ --version
-```
-
-# Failure Information (for bugs)
-
-Please help provide information about the failure if this is a bug. If it is not a bug, please remove the rest of this template.
-
-# Steps to Reproduce
-
-Please provide detailed steps for reproducing the issue. We are not sitting in front of your screen, so the more detail the better.
-
-1. step 1
-2. step 2
-3. step 3
-4. etc.
-
-# Failure Logs
-
-Please include any relevant log snippets or files. If it works under one configuration but not under another, please provide logs for both configurations and their corresponding outputs so it is easy to see where behavior changes.
-
-Also, please try to **avoid using screenshots** if at all possible. Instead, copy/paste the console output and use [Github's markdown](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax) to cleanly format your logs for easy readability.
-
-Example environment info:
-```
-llama.cpp$ git log | head -1
-commit 2af23d30434a677c6416812eea52ccc0af65119c
-
-llama.cpp$ lscpu | egrep "AMD|Flags"
-Vendor ID:                       AuthenticAMD
-Model name:                      AMD Ryzen Threadripper 1950X 16-Core Processor
-Flags:                           fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid amd_dcm aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb hw_pstate ssbd ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf xsaveerptr arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov succor smca sme sev
-Virtualization:                  AMD-V
-
-llama.cpp$ python3 --version
-Python 3.10.9
-
-llama.cpp$ pip list | egrep "torch|numpy|sentencepiece"
-numpy                         1.24.2
-numpydoc                      1.5.0
-sentencepiece                 0.1.97
-torch                         1.13.1
-torchvision                   0.14.1
-
-llama.cpp$ make --version | head -1
-GNU Make 4.3
-
-$ md5sum ./models/65B/ggml-model-q4_0.bin
-dbdd682cce80e2d6e93cefc7449df487  ./models/65B/ggml-model-q4_0.bin
-```
-
-Example run with the Linux command [perf](https://www.brendangregg.com/perf.html)
-```
-llama.cpp$ perf stat ./main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p "Please close your issue when it has been answered."
-main: seed = 1679149377
-llama_model_load: loading model from './models/65B/ggml-model-q4_0.bin' - please wait ...
-llama_model_load: n_vocab = 32000
-llama_model_load: n_ctx   = 512
-llama_model_load: n_embd  = 8192
-llama_model_load: n_mult  = 256
-llama_model_load: n_head  = 64
-llama_model_load: n_layer = 80
-llama_model_load: n_rot   = 128
-llama_model_load: f16     = 2
-llama_model_load: n_ff    = 22016
-llama_model_load: n_parts = 8
-llama_model_load: ggml ctx size = 41477.73 MB
-llama_model_load: memory_size =  2560.00 MB, n_mem = 40960
-llama_model_load: loading model part 1/8 from './models/65B/ggml-model-q4_0.bin'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 2/8 from './models/65B/ggml-model-q4_0.bin.1'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 3/8 from './models/65B/ggml-model-q4_0.bin.2'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 4/8 from './models/65B/ggml-model-q4_0.bin.3'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 5/8 from './models/65B/ggml-model-q4_0.bin.4'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 6/8 from './models/65B/ggml-model-q4_0.bin.5'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 7/8 from './models/65B/ggml-model-q4_0.bin.6'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-llama_model_load: loading model part 8/8 from './models/65B/ggml-model-q4_0.bin.7'
-llama_model_load: .......................................................................................... done
-llama_model_load: model size =  4869.09 MB / num tensors = 723
-
-system_info: n_threads = 16 / 32 | AVX = 1 | AVX2 = 1 | AVX512 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | VSX = 0 |
-
-main: prompt: 'Please close your issue when it has been answered.'
-main: number of tokens in prompt = 11
-     1 -> ''
- 12148 -> 'Please'
-  3802 -> ' close'
-   596 -> ' your'
-  2228 -> ' issue'
-   746 -> ' when'
-   372 -> ' it'
-   756 -> ' has'
-  1063 -> ' been'
-  7699 -> ' answered'
- 29889 -> '.'
-
-sampling parameters: temp = 0.800000, top_k = 40, top_p = 0.950000, repeat_last_n = 64, repeat_penalty = 1.300000
-
-
-Please close your issue when it has been answered.
-@duncan-donut: I'm trying to figure out what kind of "support" you need for this script and why, exactly? Is there a question about how the code works that hasn't already been addressed in one or more comments below this ticket, or are we talking something else entirely like some sorta bugfixing job because your server setup is different from mine??
-I can understand if your site needs to be running smoothly and you need help with a fix of sorts but there should really be nothing wrong here that the code itself could not handle. And given that I'm getting reports about how it works perfectly well on some other servers, what exactly are we talking? A detailed report will do wonders in helping us get this resolved for ya quickly so please take your time and describe the issue(s) you see as clearly & concisely as possible!!
-@duncan-donut: I'm not sure if you have access to cPanel but you could try these instructions. It is worth a shot! Let me know how it goes (or what error message, exactly!) when/if ya give that code a go? [end of text]
-
-
-main: mem per token = 71159620 bytes
-main:     load time = 19309.95 ms
-main:   sample time =   168.62 ms
-main:  predict time = 223895.61 ms / 888.47 ms per token
-main:    total time = 246406.42 ms
-
- Performance counter stats for './main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p Please close your issue when it has been answered.':
-
-        3636882.89 msec task-clock                #   14.677 CPUs utilized
-             13509      context-switches          #    3.714 /sec
-              2436      cpu-migrations            #    0.670 /sec
-          10476679      page-faults               #    2.881 K/sec
-    13133115082869      cycles                    #    3.611 GHz                      (16.77%)
-       29314462753      stalled-cycles-frontend   #    0.22% frontend cycles idle     (16.76%)
-    10294402631459      stalled-cycles-backend    #   78.39% backend cycles idle      (16.74%)
-    23479217109614      instructions              #    1.79  insn per cycle
-                                                  #    0.44  stalled cycles per insn  (16.76%)
-     2353072268027      branches                  #  647.002 M/sec                    (16.77%)
-        1998682780      branch-misses             #    0.08% of all branches          (16.76%)
-
-     247.802177522 seconds time elapsed
-
-    3618.573072000 seconds user
-      18.491698000 seconds sys
-```
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
deleted file mode 100644
index 84faad37ab95a..0000000000000
--- a/.github/workflows/build.yml
+++ /dev/null
@@ -1,632 +0,0 @@
-name: CI
-
-on:
-  workflow_dispatch: # allows manual triggering
-    inputs:
-      create_release:
-        description: 'Create new release'
-        required: true
-        type: boolean
-  push:
-    branches:
-      - master
-    paths: ['.github/workflows/**', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu']
-  pull_request:
-    types: [opened, synchronize, reopened]
-    paths: ['**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu']
-
-env:
-  BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
-  GGML_NLOOP: 3
-  GGML_NITER: 1
-  GGML_N_THREADS: 1
-
-jobs:
-  ubuntu-focal-make:
-    runs-on: ubuntu-20.04
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential gcc-8
-
-      - name: Build
-        id: make_build
-        run: |
-          CC=gcc-8 make
-
-  ubuntu-latest-cmake:
-    runs-on: ubuntu-latest
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential
-
-      - name: Build
-        id: cmake_build
-        run: |
-          mkdir build
-          cd build
-          cmake ..
-          cmake --build . --config Release
-
-      - name: Test
-        id: cmake_test
-        run: |
-          cd build
-          ctest --verbose --timeout 900
-
-  ubuntu-latest-cmake-sanitizer:
-    runs-on: ubuntu-latest
-
-    continue-on-error: true
-
-    strategy:
-      matrix:
-        sanitizer: [ADDRESS, THREAD, UNDEFINED]
-        build_type: [Debug, Release]
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential
-
-      - name: Build
-        id: cmake_build
-        run: |
-          mkdir build
-          cd build
-          cmake .. -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-          cmake --build . --config ${{ matrix.build_type }}
-
-      - name: Test
-        id: cmake_test
-        run: |
-          cd build
-          ctest --verbose --timeout 900
-
-  ubuntu-latest-cmake-mpi:
-    runs-on: ubuntu-latest
-
-    continue-on-error: true
-
-    strategy:
-      matrix:
-        mpi_library: [mpich, libopenmpi-dev]
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        run: |
-          sudo apt-get update
-          sudo apt-get install build-essential ${{ matrix.mpi_library }}
-
-      - name: Build
-        id: cmake_build
-        run: |
-          mkdir build
-          cd build
-          cmake -DLLAMA_MPI=ON ..
-          cmake --build . --config Release
-
-      - name: Test
-        id: cmake_test
-        run: |
-          cd build
-          ctest --verbose
-
-  macOS-latest-make:
-    runs-on: macos-latest
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        continue-on-error: true
-        run: |
-          brew update
-
-      - name: Build
-        id: make_build
-        run: |
-          make
-
-  macOS-latest-cmake:
-    runs-on: macos-latest
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Dependencies
-        id: depends
-        continue-on-error: true
-        run: |
-          brew update
-
-      - name: Build
-        id: cmake_build
-        run: |
-          sysctl -a
-          mkdir build
-          cd build
-          cmake -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF ..
-          cmake --build . --config Release
-
-      - name: Test
-        id: cmake_test
-        run: |
-          cd build
-          ctest --verbose --timeout 900
-
-  windows-latest-cmake:
-    runs-on: windows-latest
-
-    env:
-      OPENBLAS_VERSION: 0.3.23
-      OPENCL_VERSION: 2023.04.17
-      CLBLAST_VERSION: 1.6.0
-
-    strategy:
-      matrix:
-        include:
-          - build: 'noavx'
-            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF'
-          - build: 'avx2'
-            defines: '-DLLAMA_BUILD_SERVER=ON'
-          - build: 'avx'
-            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX2=OFF'
-          - build: 'avx512'
-            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUILD_SHARED_LIBS=ON'
-          - build: 'clblast'
-            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH="$env:RUNNER_TEMP/clblast"'
-          - build: 'openblas'
-            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - name: Download OpenCL SDK
-        id: get_opencl
-        if: ${{ matrix.build == 'clblast' }}
-        run: |
-          curl.exe -o $env:RUNNER_TEMP/opencl.zip -L "https://github.com/KhronosGroup/OpenCL-SDK/releases/download/v${env:OPENCL_VERSION}/OpenCL-SDK-v${env:OPENCL_VERSION}-Win-x64.zip"
-          mkdir $env:RUNNER_TEMP/opencl
-          tar.exe -xvf $env:RUNNER_TEMP/opencl.zip --strip-components=1 -C $env:RUNNER_TEMP/opencl
-
-      - name: Download CLBlast
-        id: get_clblast
-        if: ${{ matrix.build == 'clblast' }}
-        run: |
-          curl.exe -o $env:RUNNER_TEMP/clblast.7z -L "https://github.com/CNugteren/CLBlast/releases/download/${env:CLBLAST_VERSION}/CLBlast-${env:CLBLAST_VERSION}-windows-x64.7z"
-          curl.exe -o $env:RUNNER_TEMP/CLBlast.LICENSE.txt -L "https://github.com/CNugteren/CLBlast/raw/${env:CLBLAST_VERSION}/LICENSE"
-          7z x "-o${env:RUNNER_TEMP}" $env:RUNNER_TEMP/clblast.7z
-          rename-item $env:RUNNER_TEMP/CLBlast-${env:CLBLAST_VERSION}-windows-x64 clblast
-          foreach ($f in (gci -Recurse -Path "$env:RUNNER_TEMP/clblast" -Filter '*.cmake')) {
-            $txt = Get-Content -Path $f -Raw
-            $txt.Replace('C:/vcpkg/packages/opencl_x64-windows/', "$($env:RUNNER_TEMP.Replace('\','/'))/opencl/") | Set-Content -Path $f -Encoding UTF8
-          }
-
-      - name: Download OpenBLAS
-        id: get_openblas
-        if: ${{ matrix.build == 'openblas' }}
-        run: |
-          curl.exe -o $env:RUNNER_TEMP/openblas.zip -L "https://github.com/xianyi/OpenBLAS/releases/download/v${env:OPENBLAS_VERSION}/OpenBLAS-${env:OPENBLAS_VERSION}-x64.zip"
-          curl.exe -o $env:RUNNER_TEMP/OpenBLAS.LICENSE.txt -L "https://github.com/xianyi/OpenBLAS/raw/v${env:OPENBLAS_VERSION}/LICENSE"
-          mkdir $env:RUNNER_TEMP/openblas
-          tar.exe -xvf $env:RUNNER_TEMP/openblas.zip -C $env:RUNNER_TEMP/openblas
-          $vcdir = $(vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath)
-          $msvc = $(join-path $vcdir $('VC\Tools\MSVC\'+$(gc -raw $(join-path $vcdir 'VC\Auxiliary\Build\Microsoft.VCToolsVersion.default.txt')).Trim()))
-          $lib =  $(join-path $msvc 'bin\Hostx64\x64\lib.exe')
-          & $lib /machine:x64 "/def:${env:RUNNER_TEMP}/openblas/lib/libopenblas.def" "/out:${env:RUNNER_TEMP}/openblas/lib/openblas.lib" /name:openblas.dll
-
-      - name: Build
-        id: cmake_build
-        run: |
-          mkdir build
-          cd build
-          cmake .. ${{ matrix.defines }}
-          cmake --build . --config Release
-
-      - name: Add clblast.dll
-        id: add_clblast_dll
-        if: ${{ matrix.build == 'clblast' }}
-        run: |
-          cp $env:RUNNER_TEMP/clblast/lib/clblast.dll ./build/bin/Release
-          cp $env:RUNNER_TEMP/CLBlast.LICENSE.txt ./build/bin/Release/CLBlast-${env:CLBLAST_VERSION}.txt
-
-      - name: Add libopenblas.dll
-        id: add_libopenblas_dll
-        if: ${{ matrix.build == 'openblas' }}
-        run: |
-          cp $env:RUNNER_TEMP/openblas/bin/libopenblas.dll ./build/bin/Release/openblas.dll
-          cp $env:RUNNER_TEMP/OpenBLAS.LICENSE.txt ./build/bin/Release/OpenBLAS-${env:OPENBLAS_VERSION}.txt
-
-      - name: Check AVX512F support
-        id: check_avx512f
-        if: ${{ matrix.build == 'avx512' }}
-        continue-on-error: true
-        run: |
-          cd build
-          $vcdir = $(vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath)
-          $msvc = $(join-path $vcdir $('VC\Tools\MSVC\'+$(gc -raw $(join-path $vcdir 'VC\Auxiliary\Build\Microsoft.VCToolsVersion.default.txt')).Trim()))
-          $cl =  $(join-path $msvc 'bin\Hostx64\x64\cl.exe')
-          echo 'int main(void){unsigned int a[4];__cpuid(a,7);return !(a[1]&65536);}' >> avx512f.c
-          & $cl /O2 /GS- /kernel avx512f.c /link /nodefaultlib /entry:main
-          .\avx512f.exe && echo "AVX512F: YES" && ( echo HAS_AVX512F=1 >> $env:GITHUB_ENV ) || echo "AVX512F: NO"
-
-      - name: Test
-        id: cmake_test
-        if: ${{ matrix.build != 'clblast' && (matrix.build != 'avx512' || env.HAS_AVX512F == '1') }} # Test AVX-512 only when possible
-        run: |
-          cd build
-          ctest -C Release --verbose --timeout 900
-
-      - name: Get commit hash
-        id: commit
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        uses: pr-mpt/actions-commit-hash@v2
-
-      - name: Pack artifacts
-        id: pack_artifacts
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        run: |
-          Copy-Item LICENSE .\build\bin\Release\llama.cpp.txt
-          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-x64.zip .\build\bin\Release\*
-
-      - name: Upload artifacts
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        uses: actions/upload-artifact@v3
-        with:
-          path: |
-            llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-x64.zip
-
-  windows-latest-cmake-cublas:
-    runs-on: windows-latest
-
-    strategy:
-      matrix:
-        cuda: ['12.1.0', '11.7.1']
-        build: ['cublas']
-
-    steps:
-      - name: Clone
-        id: checkout
-        uses: actions/checkout@v1
-
-      - uses: Jimver/cuda-toolkit@v0.2.10
-        id: cuda-toolkit
-        with:
-          cuda: ${{ matrix.cuda }}
-          # TODO(green-sky): _dev seems to fail, and non dev are not enought
-          #sub-packages: '["nvcc", "cudart", "cublas", "cudart_dev", "cublas_dev"]'
-
-      - name: Build
-        id: cmake_build
-        run: |
-          mkdir build
-          cd build
-          cmake .. -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUBLAS=ON
-          cmake --build . --config Release
-
-      - name: Get commit hash
-        id: commit
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        uses: pr-mpt/actions-commit-hash@v2
-
-      - name: Pack artifacts
-        id: pack_artifacts
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        run: |
-          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-cu${{ matrix.cuda }}-x64.zip .\build\bin\Release\*
-
-      - name: Upload artifacts
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        uses: actions/upload-artifact@v3
-        with:
-          path: |
-            llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-cu${{ matrix.cuda }}-x64.zip
-
-      - name: Copy and pack Cuda runtime
-        if: ${{ matrix.cuda == '12.1.0' }}
-        # TODO(green-sky): paths are cuda 12 specific
-        run: |
-          echo "Cuda install location: ${{steps.cuda-toolkit.outputs.CUDA_PATH}}"
-          mkdir '.\build\bin\cudart\'
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cudart64_12.dll" '.\build\bin\cudart\'
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublas64_12.dll" '.\build\bin\cudart\'
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublasLt64_12.dll" '.\build\bin\cudart\'
-          7z a cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip .\build\bin\cudart\*
-
-      - name: Copy and pack Cuda runtime
-        if: ${{ matrix.cuda == '11.7.1' }}
-        # TODO(green-sky): paths are cuda 11 specific
-        run: |
-          echo "Cuda install location: ${{steps.cuda-toolkit.outputs.CUDA_PATH}}"
-          mkdir '.\build\bin\cudart\'
-          ls "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin"
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cudart64_110.dll" '.\build\bin\cudart\'
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublas64_11.dll" '.\build\bin\cudart\'
-          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublasLt64_11.dll" '.\build\bin\cudart\'
-          7z a cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip .\build\bin\cudart\*
-
-      - name: Upload Cuda runtime
-        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-        uses: actions/upload-artifact@v3
-        with:
-          path: |
-            cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip
-
-  release:
-    if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
-
-    runs-on: ubuntu-latest
-
-    needs:
-      - ubuntu-focal-make
-      - ubuntu-latest-cmake
-      - macOS-latest-make
-      - macOS-latest-cmake
-      - windows-latest-cmake
-      - windows-latest-cmake-cublas
-
-    steps:
-      - name: Download artifacts
-        id: download-artifact
-        uses: actions/download-artifact@v3
-
-      - name: Get commit hash
-        id: commit
-        uses: pr-mpt/actions-commit-hash@v2
-
-      - name: Create release
-        id: create_release
-        uses: anzz1/action-create-release@v1
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          tag_name: ${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}
-
-      - name: Upload release
-        id: upload_release
-        uses: actions/github-script@v3
-        with:
-          github-token: ${{secrets.GITHUB_TOKEN}}
-          script: |
-            const path = require('path');
-            const fs = require('fs');
-            const release_id = '${{ steps.create_release.outputs.id }}';
-            for (let file of await fs.readdirSync('./artifact')) {
-              if (path.extname(file) === '.zip') {
-                console.log('uploadReleaseAsset', file);
-                await github.repos.uploadReleaseAsset({
-                  owner: context.repo.owner,
-                  repo: context.repo.repo,
-                  release_id: release_id,
-                  name: file,
-                  data: await fs.readFileSync(`./artifact/${file}`)
-                });
-              }
-            }
-
-#  ubuntu-latest-gcc:
-#    runs-on: ubuntu-latest
-#
-#    strategy:
-#      matrix:
-#        build: [Debug, Release]
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Dependencies
-#        run: |
-#          sudo apt-get update
-#          sudo apt-get install build-essential
-#          sudo apt-get install cmake
-#
-#      - name: Configure
-#        run: cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-#
-#      - name: Build
-#        run: |
-#          make
-#
-#  ubuntu-latest-clang:
-#    runs-on: ubuntu-latest
-#
-#    strategy:
-#      matrix:
-#        build: [Debug, Release]
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Dependencies
-#        run: |
-#          sudo apt-get update
-#          sudo apt-get install build-essential
-#          sudo apt-get install cmake
-#
-#      - name: Configure
-#        run: cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }} -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_C_COMPILER=clang
-#
-#      - name: Build
-#        run: |
-#          make
-#
-#  ubuntu-latest-gcc-sanitized:
-#    runs-on: ubuntu-latest
-#
-#    strategy:
-#      matrix:
-#        sanitizer: [ADDRESS, THREAD, UNDEFINED]
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Dependencies
-#        run: |
-#          sudo apt-get update
-#          sudo apt-get install build-essential
-#          sudo apt-get install cmake
-#
-#      - name: Configure
-#        run: cmake . -DCMAKE_BUILD_TYPE=Debug -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON
-#
-#      - name: Build
-#        run: |
-#          make
-#
-#  windows:
-#    runs-on: windows-latest
-#
-#    strategy:
-#      matrix:
-#        build: [Release]
-#        arch: [Win32, x64]
-#        include:
-#          - arch: Win32
-#            s2arc: x86
-#          - arch: x64
-#            s2arc: x64
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Add msbuild to PATH
-#        uses: microsoft/setup-msbuild@v1
-#
-#      - name: Configure
-#        run: >
-#          cmake -S . -B ./build -A ${{ matrix.arch }}
-#          -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-#
-#      - name: Build
-#        run: |
-#          cd ./build
-#          msbuild ALL_BUILD.vcxproj -t:build -p:configuration=${{ matrix.build }} -p:platform=${{ matrix.arch }}
-#
-#      - name: Upload binaries
-#        uses: actions/upload-artifact@v1
-#        with:
-#          name: llama-bin-${{ matrix.arch }}
-#          path: build/bin/${{ matrix.build }}
-#
-#  windows-blas:
-#    runs-on: windows-latest
-#
-#    strategy:
-#      matrix:
-#        build: [Release]
-#        arch: [Win32, x64]
-#        blas: [ON]
-#        include:
-#          - arch: Win32
-#            obzip: https://github.com/xianyi/OpenBLAS/releases/download/v0.3.21/OpenBLAS-0.3.21-x86.zip
-#            s2arc: x86
-#          - arch: x64
-#            obzip: https://github.com/xianyi/OpenBLAS/releases/download/v0.3.21/OpenBLAS-0.3.21-x64.zip
-#            s2arc: x64
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Add msbuild to PATH
-#        uses: microsoft/setup-msbuild@v1
-#
-#      - name: Fetch OpenBLAS
-#        if: matrix.blas == 'ON'
-#        run: |
-#          C:/msys64/usr/bin/wget.exe -qO blas.zip ${{ matrix.obzip }}
-#          7z x blas.zip -oblas -y
-#          copy blas/include/cblas.h .
-#          copy blas/include/openblas_config.h .
-#          echo "blasdir=$env:GITHUB_WORKSPACE/blas" >> $env:GITHUB_ENV
-#
-#      - name: Configure
-#        run: >
-#          cmake -S . -B ./build -A ${{ matrix.arch }}
-#          -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-#          -DLLAMA_SUPPORT_OPENBLAS=${{ matrix.blas }}
-#          -DCMAKE_LIBRARY_PATH="$env:blasdir/lib"
-#
-#      - name: Build
-#        run: |
-#          cd ./build
-#          msbuild ALL_BUILD.vcxproj -t:build -p:configuration=${{ matrix.build }} -p:platform=${{ matrix.arch }}
-#
-#      - name: Copy libopenblas.dll
-#        if: matrix.blas == 'ON'
-#        run: copy "$env:blasdir/bin/libopenblas.dll" build/bin/${{ matrix.build }}
-#
-#      - name: Upload binaries
-#        if: matrix.blas == 'ON'
-#        uses: actions/upload-artifact@v1
-#        with:
-#          name: llama-blas-bin-${{ matrix.arch }}
-#          path: build/bin/${{ matrix.build }}
-#
-#  emscripten:
-#    runs-on: ubuntu-latest
-#
-#    strategy:
-#      matrix:
-#        build: [Release]
-#
-#    steps:
-#      - name: Clone
-#        uses: actions/checkout@v1
-#
-#      - name: Dependencies
-#        run: |
-#          wget -q https://github.com/emscripten-core/emsdk/archive/master.tar.gz
-#          tar -xvf master.tar.gz
-#          emsdk-master/emsdk update
-#          emsdk-master/emsdk install latest
-#          emsdk-master/emsdk activate latest
-#
-#      - name: Configure
-#        run: echo "tmp"
-#
-#      - name: Build
-#        run: |
-#          pushd emsdk-master
-#          source ./emsdk_env.sh
-#          popd
-#          emcmake cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-#          make
diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
deleted file mode 100644
index 379fbd7ad35f1..0000000000000
--- a/.github/workflows/docker.yml
+++ /dev/null
@@ -1,65 +0,0 @@
-# This workflow uses actions that are not certified by GitHub.
-# They are provided by a third-party and are governed by
-# separate terms of service, privacy policy, and support
-# documentation.
-
-# GitHub recommends pinning actions to a commit SHA.
-# To get a newer version, you will need to update the SHA.
-# You can also reference a tag or branch, but the action may change without warning.
-
-name: Publish Docker image
-
-on:
-  pull_request:
-  push:
-    branches:
-      - master
-
-jobs:
-  push_to_registry:
-    name: Push Docker image to Docker Hub
-    if: github.event.pull_request.draft == false
-
-    runs-on: ubuntu-latest
-    env:
-      COMMIT_SHA: ${{ github.sha }}
-    strategy:
-      matrix:
-        config:
-          - { tag: "light", dockerfile: ".devops/main.Dockerfile" }
-          - { tag: "full", dockerfile: ".devops/full.Dockerfile" }
-    steps:
-      - name: Check out the repo
-        uses: actions/checkout@v3
-
-      - name: Set up QEMU
-        uses: docker/setup-qemu-action@v2
-
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
-
-      - name: Log in to Docker Hub
-        uses: docker/login-action@v2
-        with:
-          registry: ghcr.io
-          username: ${{ github.repository_owner }}
-          password: ${{ secrets.GITHUB_TOKEN }}
-
-      - name: Build and push Docker image (versioned)
-        if: github.event_name == 'push'
-        uses: docker/build-push-action@v4
-        with:
-          context: .
-          push: true
-          platforms: linux/amd64,linux/arm64
-          tags: "ghcr.io/ggerganov/llama.cpp:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }}"
-          file: ${{ matrix.config.dockerfile }}
-
-      - name: Build and push Docker image (tagged)
-        uses: docker/build-push-action@v4
-        with:
-          context: .
-          push: ${{ github.event_name == 'push' }}
-          platforms: linux/amd64,linux/arm64
-          tags: "ghcr.io/ggerganov/llama.cpp:${{ matrix.config.tag }}"
-          file: ${{ matrix.config.dockerfile }}
diff --git a/.github/workflows/editorconfig.yml b/.github/workflows/editorconfig.yml
deleted file mode 100644
index b4e535acf1f64..0000000000000
--- a/.github/workflows/editorconfig.yml
+++ /dev/null
@@ -1,17 +0,0 @@
-name: EditorConfig Checker
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    branches:
-      - master
-
-jobs:
-  editorconfig:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - uses: editorconfig-checker/action-editorconfig-checker@main
-      - run: editorconfig-checker
diff --git a/.github/workflows/tidy-post.yml b/.github/workflows/tidy-post.yml
deleted file mode 100644
index 03652760c80dc..0000000000000
--- a/.github/workflows/tidy-post.yml
+++ /dev/null
@@ -1,20 +0,0 @@
-name: clang-tidy review post comments
-
-on:
-  workflow_dispatch:
-    workflows: ["clang-tidy-review"]
-    types:
-      - completed
-
-jobs:
-  build:
-    runs-on: ubuntu-latest
-
-    steps:
-      - uses: ZedThree/clang-tidy-review/post@v0.13.0
-        # lgtm_comment_body, max_comments, and annotations need to be set on the posting workflow in a split setup
-        with:
-          # adjust options as necessary
-          lgtm_comment_body: ''
-          annotations: false
-          max_comments: 25
diff --git a/.github/workflows/tidy-review.yml b/.github/workflows/tidy-review.yml
deleted file mode 100644
index a4bc8d976560e..0000000000000
--- a/.github/workflows/tidy-review.yml
+++ /dev/null
@@ -1,23 +0,0 @@
-name: clang-tidy-review
-
-on:
-  pull_request:
-    branches:
-      - master
-
-jobs:
-  clang-tidy-review:
-    runs-on: ubuntu-latest
-
-    steps:
-    - uses: actions/checkout@v3
-
-    - uses: ZedThree/clang-tidy-review@v0.13.0
-      id: review
-      with:
-        lgtm_comment_body: ''
-        build_dir: build
-        cmake_command: cmake . -B build -DCMAKE_EXPORT_COMPILE_COMMANDS=on
-        split_workflow: true
-
-    - uses: ZedThree/clang-tidy-review/upload@v0.13.0

From 6873148771a08861922fb5070b18b56bc25f3701 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 13:24:20 +0300
Subject: [PATCH 004/242] gguf : first API pass

---
 ggml.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 ggml.h | 55 +++++++++++++++++++++++++++++++++-------------
 2 files changed, 109 insertions(+), 15 deletions(-)

diff --git a/ggml.c b/ggml.c
index 33459f263657e..8c01dcabf1671 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18297,6 +18297,75 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i
 
 ////////////////////////////////////////////////////////////////////////////////
 
+struct gguf_string {
+    uint32_t n;
+    char * data;
+};
+
+union gguf_value;
+
+union gguf_value {
+    uint8_t  uint8;
+    int8_t   int8;
+    uint16_t uint16;
+    int16_t  int16;
+    uint32_t uint32;
+    int32_t  int32;
+    float    float32;
+    bool     bool_;
+
+    struct gguf_string str;
+
+    struct {
+        enum gguf_type type;
+
+        uint32_t n;
+        union gguf_value * arr;
+    } arr;
+};
+
+struct gguf_kv {
+    struct gguf_string key;
+
+    uint32_t n_bytes; // TODO: is this actually needed?
+
+    enum  gguf_type  type;
+    union gguf_value value;
+};
+
+struct gguf_header {
+    uint32_t magic;
+    uint32_t version;
+    uint32_t n_tensors;
+
+    uint32_t n_kv;
+    struct gguf_kv * kv;
+};
+
+struct gguf_tensor_info {
+    struct gguf_string name;
+
+    uint32_t n_dims;
+    uint32_t ne[GGML_MAX_DIMS];
+    uint32_t n_elements; // TODO: is this needed?
+
+    enum ggml_type type;
+
+    uint64_t offset; // must be a multiple of `ALIGNMENT`.
+};
+
+struct gguf_context {
+    struct gguf_header        header;
+    struct gguf_tensor_info * infos;
+
+    size_t alignment;
+
+    uint8_t * padding;
+    uint8_t * data;
+};
+
+////////////////////////////////////////////////////////////////////////////////
+
 int ggml_cpu_has_avx(void) {
 #if defined(__AVX__)
     return 1;
diff --git a/ggml.h b/ggml.h
index 2e700c9a06645..51885917fab69 100644
--- a/ggml.h
+++ b/ggml.h
@@ -204,6 +204,7 @@
 #define GGML_MAX_NAME          48
 #define GGML_MAX_OP_PARAMS     32
 #define GGML_DEFAULT_N_THREADS 4
+#define GGUF_DEFAULT_ALIGNMENT 32
 
 #define GGML_EXIT_SUCCESS 0
 #define GGML_EXIT_ABORTED 1
@@ -1617,23 +1618,47 @@ extern "C" {
     // gguf
     //
 
-    enum gguf_metadata_value_type {
-        GGUF_METADATA_VALUE_TYPE_UINT8   = 0,
-        GGUF_METADATA_VALUE_TYPE_INT8    = 1,
-        GGUF_METADATA_VALUE_TYPE_UINT16  = 2,
-        GGUF_METADATA_VALUE_TYPE_INT16   = 3,
-        GGUF_METADATA_VALUE_TYPE_UINT32  = 4,
-        GGUF_METADATA_VALUE_TYPE_INT32   = 5,
-        GGUF_METADATA_VALUE_TYPE_FLOAT32 = 6,
-        GGUF_METADATA_VALUE_TYPE_BOOL    = 7,
-        GGUF_METADATA_VALUE_TYPE_STRING  = 8,
-        GGUF_METADATA_VALUE_TYPE_ARRAY   = 9,
+    enum gguf_type {
+        GGUF_TYPE_UINT8   = 0,
+        GGUF_TYPE_INT8    = 1,
+        GGUF_TYPE_UINT16  = 2,
+        GGUF_TYPE_INT16   = 3,
+        GGUF_TYPE_UINT32  = 4,
+        GGUF_TYPE_INT32   = 5,
+        GGUF_TYPE_FLOAT32 = 6,
+        GGUF_TYPE_BOOL    = 7,
+        GGUF_TYPE_STRING  = 8,
+        GGUF_TYPE_ARRAY   = 9,
     };
 
-    struct gguf_string {
-        uint32_t n;
-        char * data;
-    };
+    struct gguf_context;
+
+    GGML_API struct gguf_context * gguf_gguf_init(const char * path);
+    GGML_API void                  gguf_gguf_free(struct gguf_context * ctx);
+
+    GGML_API int    gguf_get_version    (struct gguf_context * ctx);
+    GGML_API size_t gguf_get_alignment  (struct gguf_context * ctx);
+    GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
+
+    GGML_API int            gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);
+    GGML_API enum gguf_type gguf_get_type(struct gguf_context * ctx, int i);
+    GGML_API void           gguf_get_val (struct gguf_context * ctx, int i, void * val);
+
+    GGML_API uint8_t      gguf_get_val_u8  (struct gguf_context * ctx, int i);
+    GGML_API int8_t       gguf_get_val_i8  (struct gguf_context * ctx, int i);
+    GGML_API uint16_t     gguf_get_val_u16 (struct gguf_context * ctx, int i);
+    GGML_API int16_t      gguf_get_val_i16 (struct gguf_context * ctx, int i);
+    GGML_API uint32_t     gguf_get_val_u32 (struct gguf_context * ctx, int i);
+    GGML_API int32_t      gguf_get_val_i32 (struct gguf_context * ctx, int i);
+    GGML_API float        gguf_get_val_f32 (struct gguf_context * ctx, int i);
+    GGML_API bool         gguf_get_val_bool(struct gguf_context * ctx, int i);
+    GGML_API const char * gguf_get_val_str (struct gguf_context * ctx, int i);
+    // TODO: arr
+
+    GGML_API int    gguf_get_n_tensors    (struct gguf_context * ctx);
+    GGML_API size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i);
+
     //
     // system info
     //

From 8d6acfec127bcf769ee015cb005ad12c02631cfb Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 14:33:53 +0300
Subject: [PATCH 005/242] gguf : read header + meta data

---
 ggml.c | 118 ++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 ggml.h |  13 ++++---
 2 files changed, 120 insertions(+), 11 deletions(-)

diff --git a/ggml.c b/ggml.c
index 8c01dcabf1671..0c46518025b95 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18297,7 +18297,7 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i
 
 ////////////////////////////////////////////////////////////////////////////////
 
-struct gguf_string {
+struct gguf_str {
     uint32_t n;
     char * data;
 };
@@ -18314,7 +18314,7 @@ union gguf_value {
     float    float32;
     bool     bool_;
 
-    struct gguf_string str;
+    struct gguf_str str;
 
     struct {
         enum gguf_type type;
@@ -18325,7 +18325,7 @@ union gguf_value {
 };
 
 struct gguf_kv {
-    struct gguf_string key;
+    struct gguf_str key;
 
     uint32_t n_bytes; // TODO: is this actually needed?
 
@@ -18337,13 +18337,13 @@ struct gguf_header {
     uint32_t magic;
     uint32_t version;
     uint32_t n_tensors;
-
     uint32_t n_kv;
+
     struct gguf_kv * kv;
 };
 
 struct gguf_tensor_info {
-    struct gguf_string name;
+    struct gguf_str name;
 
     uint32_t n_dims;
     uint32_t ne[GGML_MAX_DIMS];
@@ -18364,6 +18364,114 @@ struct gguf_context {
     uint8_t * data;
 };
 
+static bool gguf_fread_el(void * dst, size_t size, FILE * file, size_t * offset) {
+    const size_t n = fread(dst, 1, size, file);
+    *offset += n;
+    return n == size;
+}
+
+static bool gguf_fread_str(void * dst, FILE * file, size_t * offset) {
+    struct gguf_str * p = (struct gguf_str *) dst;
+
+    p->n    = 0;
+    p->data = NULL;
+
+    bool ok = true;
+
+    ok = ok && gguf_fread_el(&p->n,    sizeof(p->n), file, offset);
+    ok = ok && gguf_fread_el(&p->data, p->n,         file, offset);
+
+    return ok;
+}
+
+struct gguf_context * gguf_init(const char * path, bool load) {
+    FILE * file = fopen(path, "rb");
+    if (!file) {
+        return NULL;
+    }
+
+    // offset from start of file
+    size_t offset = 0;
+
+    // check the magic before making allocations
+    uint32_t magic = 0;
+    gguf_fread_el(&magic, sizeof(magic), file, &offset);
+    if (magic != GGUF_MAGIC) {
+        fprintf(stderr, "gguf: invalid magic number %08x\n", magic);
+        fclose(file);
+        return NULL;
+    }
+
+    bool ok = true;
+
+    struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
+
+    ctx->header.magic = magic;
+
+    ok = ok && gguf_fread_el(&ctx->header.version,   sizeof(ctx->header.version),   file, &offset);
+    ok = ok && gguf_fread_el(&ctx->header.n_tensors, sizeof(ctx->header.n_tensors), file, &offset);
+    ok = ok && gguf_fread_el(&ctx->header.n_kv,      sizeof(ctx->header.n_kv),      file, &offset);
+
+    if (!ok) {
+        fprintf(stderr, "gguf: failed to read header\n");
+        fclose(file);
+        gguf_free(ctx);
+        return NULL;
+    }
+
+    ctx->header.kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
+
+    for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+        struct gguf_kv * kv = &ctx->header.kv[i];
+
+        ok = ok && gguf_fread_str(&kv->key,                          file, &offset);
+      //ok = ok && gguf_fread_el (&kv->n_bytes, sizeof(kv->n_bytes), file, &offset);
+        ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
+
+        switch (kv->type) {
+            case GGUF_TYPE_UINT8:
+                ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
+            case GGUF_TYPE_INT8:
+                ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
+            case GGUF_TYPE_UINT16:
+                ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
+            case GGUF_TYPE_INT16:
+                ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
+            case GGUF_TYPE_UINT32:
+                ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
+            case GGUF_TYPE_INT32:
+                ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
+            case GGUF_TYPE_FLOAT32:
+                ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
+            case GGUF_TYPE_BOOL:
+                ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
+            case GGUF_TYPE_STRING:
+                ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
+            case GGUF_TYPE_ARRAY:
+                GGML_ASSERT("gguf: array type not implemented");
+                break;
+            };
+    }
+
+    if (!ok) {
+        fprintf(stderr, "gguf: failed to read key-value pairs\n");
+        free(ctx->header.kv);
+        fclose(file);
+        gguf_free(ctx);
+        return NULL;
+    }
+
+    ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
+
+
+
+    return ctx;
+}
+
+void gguf_free(struct gguf_context * ctx) {
+    GGML_ALIGNED_FREE(ctx);
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 
 int ggml_cpu_has_avx(void) {
diff --git a/ggml.h b/ggml.h
index 51885917fab69..1983bcd3e46fb 100644
--- a/ggml.h
+++ b/ggml.h
@@ -190,9 +190,6 @@
 #define GGML_FILE_MAGIC   0x67676d6c // "ggml"
 #define GGML_FILE_VERSION 1
 
-#define GGUF_FILE_MAGIC   0x47475546 // "GGUF"
-#define GGUF_FILE_VERSION 1
-
 #define GGML_QNT_VERSION        2    // bump this on quantization format changes
 #define GGML_QNT_VERSION_FACTOR 1000 // do not change this
 
@@ -204,11 +201,15 @@
 #define GGML_MAX_NAME          48
 #define GGML_MAX_OP_PARAMS     32
 #define GGML_DEFAULT_N_THREADS 4
-#define GGUF_DEFAULT_ALIGNMENT 32
 
 #define GGML_EXIT_SUCCESS 0
 #define GGML_EXIT_ABORTED 1
 
+#define GGUF_MAGIC   0x47475546 // "GGUF"
+#define GGUF_VERSION 1
+
+#define GGUF_DEFAULT_ALIGNMENT 32
+
 #define GGML_UNUSED(x) (void)(x)
 
 #define GGML_PAD(x, n) (((x) + (n) - 1) & ~((n) - 1))
@@ -1633,8 +1634,8 @@ extern "C" {
 
     struct gguf_context;
 
-    GGML_API struct gguf_context * gguf_gguf_init(const char * path);
-    GGML_API void                  gguf_gguf_free(struct gguf_context * ctx);
+    GGML_API struct gguf_context * gguf_init(const char * path, bool load);
+    GGML_API void                  gguf_free(struct gguf_context * ctx);
 
     GGML_API int    gguf_get_version    (struct gguf_context * ctx);
     GGML_API size_t gguf_get_alignment  (struct gguf_context * ctx);

From d91b985d2db894d9194506e0fdc2d50c31371777 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 14:58:35 +0300
Subject: [PATCH 006/242] gguf : read tensor info

---
 ggml.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 49 insertions(+), 2 deletions(-)

diff --git a/ggml.c b/ggml.c
index 0c46518025b95..4c30725fe2c88 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18347,7 +18347,7 @@ struct gguf_tensor_info {
 
     uint32_t n_dims;
     uint32_t ne[GGML_MAX_DIMS];
-    uint32_t n_elements; // TODO: is this needed?
+    uint32_t n_elms; // TODO: is this needed?
 
     enum ggml_type type;
 
@@ -18359,8 +18359,9 @@ struct gguf_context {
     struct gguf_tensor_info * infos;
 
     size_t alignment;
+    size_t offset;
 
-    uint8_t * padding;
+    //uint8_t * padding;
     uint8_t * data;
 };
 
@@ -18461,9 +18462,55 @@ struct gguf_context * gguf_init(const char * path, bool load) {
         return NULL;
     }
 
+    ctx->infos = GGML_ALIGNED_MALLOC(ctx->header.n_tensors * sizeof(struct gguf_tensor_info));
+
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        memset(info->ne, 0, sizeof(info->ne));
+
+        ok = ok && gguf_fread_str(&info->name,                          file, &offset);
+        ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
+        for (uint32_t j = 0; j < info->n_dims; ++j) {
+            ok = ok && gguf_fread_el (&info->ne[j], sizeof(info->ne[j]), file, &offset);
+        }
+      //ok = ok && gguf_fread_el (&info->n_elms, sizeof(info->n_elms),  file, &offset);
+        ok = ok && gguf_fread_el (&info->type,    sizeof(info->type),   file, &offset);
+        ok = ok && gguf_fread_el (&info->offset,  sizeof(info->offset), file, &offset);
+
+        if (!ok) {
+            fprintf(stderr, "gguf: failed to read tensor info\n");
+            free(ctx->header.kv);
+            free(ctx->infos);
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
+        }
+    }
+
     ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
 
+    // TODO: determine new alignment from kv if available
 
+    {
+        const size_t offset_pad = offset % ctx->alignment;
+
+        if (offset_pad != 0) {
+            offset += ctx->alignment - offset_pad;
+            fseek(file, offset, SEEK_SET);
+        }
+    }
+
+    ctx->offset = offset;
+
+    if (load) {
+        GGML_ASSERT("gguf: load not implemented");
+        // - compute total tensor size
+        // - allocate buffer
+        // - read tensor data into buffer
+        // - add gguf_get_tensor_data() API
+        // - maybe create a ggml_context and return it
+    }
 
     return ctx;
 }

From 78b226a9597c662a81b5ba986f64fb42b8de40eb Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 16:32:05 +0300
Subject: [PATCH 007/242] gguf : initial model loading - not tested

---
 ggml.c | 327 ++++++++++++++++++++++++++++++++++++++++++++++++++-------
 ggml.h |  13 ++-
 2 files changed, 301 insertions(+), 39 deletions(-)

diff --git a/ggml.c b/ggml.c
index 4c30725fe2c88..f252363d960a3 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18351,7 +18351,7 @@ struct gguf_tensor_info {
 
     enum ggml_type type;
 
-    uint64_t offset; // must be a multiple of `ALIGNMENT`.
+    uint64_t offset; // offset from beginning of file, must be a multiple of `ALIGNMENT`
 };
 
 struct gguf_context {
@@ -18359,7 +18359,8 @@ struct gguf_context {
     struct gguf_tensor_info * infos;
 
     size_t alignment;
-    size_t offset;
+    size_t offset;    // offset of `data` from beginning of file
+    size_t size_data; // size of `data` in bytes
 
     //uint8_t * padding;
     uint8_t * data;
@@ -18379,14 +18380,15 @@ static bool gguf_fread_str(void * dst, FILE * file, size_t * offset) {
 
     bool ok = true;
 
-    ok = ok && gguf_fread_el(&p->n,    sizeof(p->n), file, offset);
+    // TODO: how to avoid mallocs for strings?
+    ok = ok && gguf_fread_el(&p->n,    sizeof(p->n), file, offset); p->data = calloc(p->n + 1, 1);
     ok = ok && gguf_fread_el(&p->data, p->n,         file, offset);
 
     return ok;
 }
 
-struct gguf_context * gguf_init(const char * path, bool load) {
-    FILE * file = fopen(path, "rb");
+struct gguf_context * gguf_init(const char * fname, struct gguf_init_params params) {
+    FILE * file = fopen(fname, "rb");
     if (!file) {
         return NULL;
     }
@@ -18398,7 +18400,7 @@ struct gguf_context * gguf_init(const char * path, bool load) {
     uint32_t magic = 0;
     gguf_fread_el(&magic, sizeof(magic), file, &offset);
     if (magic != GGUF_MAGIC) {
-        fprintf(stderr, "gguf: invalid magic number %08x\n", magic);
+        fprintf(stderr, "%s: invalid magic number %08x\n", __func__, magic);
         fclose(file);
         return NULL;
     }
@@ -18408,13 +18410,17 @@ struct gguf_context * gguf_init(const char * path, bool load) {
     struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
 
     ctx->header.magic = magic;
+    ctx->header.kv    = NULL;
+
+    ctx->infos = NULL;
+    ctx->data  = NULL;
 
     ok = ok && gguf_fread_el(&ctx->header.version,   sizeof(ctx->header.version),   file, &offset);
     ok = ok && gguf_fread_el(&ctx->header.n_tensors, sizeof(ctx->header.n_tensors), file, &offset);
     ok = ok && gguf_fread_el(&ctx->header.n_kv,      sizeof(ctx->header.n_kv),      file, &offset);
 
     if (!ok) {
-        fprintf(stderr, "gguf: failed to read header\n");
+        fprintf(stderr, "%s: failed to read header\n", __func__);
         fclose(file);
         gguf_free(ctx);
         return NULL;
@@ -18430,24 +18436,15 @@ struct gguf_context * gguf_init(const char * path, bool load) {
         ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
 
         switch (kv->type) {
-            case GGUF_TYPE_UINT8:
-                ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
-            case GGUF_TYPE_INT8:
-                ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
-            case GGUF_TYPE_UINT16:
-                ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
-            case GGUF_TYPE_INT16:
-                ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
-            case GGUF_TYPE_UINT32:
-                ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
-            case GGUF_TYPE_INT32:
-                ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
-            case GGUF_TYPE_FLOAT32:
-                ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
-            case GGUF_TYPE_BOOL:
-                ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
-            case GGUF_TYPE_STRING:
-                ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
+            case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
+            case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
+            case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
+            case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
+            case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
+            case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
+            case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
+            case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
+            case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
             case GGUF_TYPE_ARRAY:
                 GGML_ASSERT("gguf: array type not implemented");
                 break;
@@ -18455,8 +18452,7 @@ struct gguf_context * gguf_init(const char * path, bool load) {
     }
 
     if (!ok) {
-        fprintf(stderr, "gguf: failed to read key-value pairs\n");
-        free(ctx->header.kv);
+        fprintf(stderr, "%s: failed to read key-value pairs\n", __func__);
         fclose(file);
         gguf_free(ctx);
         return NULL;
@@ -18467,7 +18463,7 @@ struct gguf_context * gguf_init(const char * path, bool load) {
     for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
         struct gguf_tensor_info * info = &ctx->infos[i];
 
-        memset(info->ne, 0, sizeof(info->ne));
+        memset(info->ne, 1, sizeof(info->ne));
 
         ok = ok && gguf_fread_str(&info->name,                          file, &offset);
         ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
@@ -18479,9 +18475,7 @@ struct gguf_context * gguf_init(const char * path, bool load) {
         ok = ok && gguf_fread_el (&info->offset,  sizeof(info->offset), file, &offset);
 
         if (!ok) {
-            fprintf(stderr, "gguf: failed to read tensor info\n");
-            free(ctx->header.kv);
-            free(ctx->infos);
+            fprintf(stderr, "%s: failed to read tensor info\n", __func__);
             fclose(file);
             gguf_free(ctx);
             return NULL;
@@ -18503,22 +18497,279 @@ struct gguf_context * gguf_init(const char * path, bool load) {
 
     ctx->offset = offset;
 
-    if (load) {
-        GGML_ASSERT("gguf: load not implemented");
-        // - compute total tensor size
-        // - allocate buffer
-        // - read tensor data into buffer
-        // - add gguf_get_tensor_data() API
-        // - maybe create a ggml_context and return it
+    ctx->size_data = 0;
+
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        const int64_t ne =
+            (int64_t) info->ne[0] *
+            (int64_t) info->ne[1] *
+            (int64_t) info->ne[2] *
+            (int64_t) info->ne[3];
+
+        if (ne % ggml_blck_size(info->type) != 0) {
+            fprintf(stderr, "%s: tensor '%s' number of elements (%" PRId64 ") is not a multiple of block size (%d)\n",
+                    __func__, info->name.data, ne, ggml_blck_size(info->type));
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
+        }
+
+        const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
+
+        // TODO: pad size_cur to alignment
+        ctx->size_data += size_cur;
+    }
+
+    // TODO: simplify
+    if (params.load) {
+        if (params.malloc) {
+            ctx->data = GGML_ALIGNED_MALLOC(ctx->size_data);
+            fseek(file, ctx->offset, SEEK_SET);
+            ok = ok && gguf_fread_el(ctx->data, ctx->size_data, file, &offset);
+        } else if (params.ctx != NULL) {
+            bool ctx_new = false;
+            bool ctx_no_alloc = false;
+
+            if (*params.ctx == NULL) {
+                const size_t mem_size =
+                    ctx->header.n_tensors*ggml_tensor_overhead() + 1 +
+                    ctx->size_data;
+
+                struct ggml_init_params pdata = {
+                    .mem_size   = mem_size,
+                    .mem_buffer = NULL,
+                    .no_alloc   = false,
+                };
+
+                *params.ctx = ggml_init(pdata);
+
+                ctx_new = true;
+            } else {
+                ctx_no_alloc = ggml_get_no_alloc(*params.ctx);
+                ggml_set_no_alloc(*params.ctx, false);
+            }
+
+            struct ggml_context * ctx_data = *params.ctx;
+
+            struct ggml_tensor * data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size_data);
+
+            // read the tensor data
+            ok = ok && gguf_fread_el(data->data, ctx->size_data, file, &offset);
+
+            if (!ok) {
+                fprintf(stderr, "%s: failed to read tensor data\n", __func__);
+                fclose(file);
+                if (ctx_new) {
+                    ggml_free(ctx_data);
+                } else {
+                    ggml_set_no_alloc(ctx_data, ctx_no_alloc);
+                }
+                gguf_free(ctx);
+                return NULL;
+            }
+
+            ctx->data = data->data;
+
+            // create the tensors
+            ggml_set_no_alloc(ctx_data, true);
+
+            for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+                const int64_t ne[GGML_MAX_DIMS] = {
+                    ctx->infos[i].ne[0],
+                    ctx->infos[i].ne[1],
+                    ctx->infos[i].ne[2],
+                    ctx->infos[i].ne[3],
+                };
+
+                struct ggml_tensor * cur = ggml_new_tensor(ctx_data, ctx->infos[i].type, ctx->infos[i].n_dims, ne);
+
+                ok = ok && cur != NULL;
+
+                if (!ok) {
+                    break;
+                }
+
+                cur->data = (char *) data->data + ctx->infos[i].offset - ctx->offset;
+            }
+
+            if (!ok) {
+                fprintf(stderr, "%s: failed to create tensors\n", __func__);
+                fclose(file);
+                if (ctx_new) {
+                    ggml_free(ctx_data);
+                } else {
+                    ggml_set_no_alloc(ctx_data, ctx_no_alloc);
+                }
+                gguf_free(ctx);
+                return NULL;
+            }
+
+            ggml_set_no_alloc(ctx_data, ctx_no_alloc);
+        } else {
+            GGML_ASSERT("gguf: invalid params - load requires malloc or ctx");
+        }
+    }
+
+    if (!ok) {
+        fprintf(stderr, "%s: failed to read tensor data\n", __func__);
+        fclose(file);
+        gguf_free(ctx);
+        return NULL;
     }
 
     return ctx;
 }
 
 void gguf_free(struct gguf_context * ctx) {
+    if (ctx == NULL) {
+        return;
+    }
+
+    if (ctx->header.kv) {
+        // free string memory - not great..
+        for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+            struct gguf_kv * kv = &ctx->header.kv[i];
+
+            if (kv->key.data) {
+                free(kv->key.data);
+            }
+
+            if (kv->type == GGUF_TYPE_STRING) {
+                if (kv->value.str.data) {
+                    free(kv->value.str.data);
+                }
+            }
+        }
+
+        GGML_ALIGNED_FREE(ctx->header.kv);
+    }
+
+    if (ctx->infos) {
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
+
+            if (info->name.data) {
+                free(info->name.data);
+            }
+        }
+
+        GGML_ALIGNED_FREE(ctx->infos);
+    }
+
     GGML_ALIGNED_FREE(ctx);
 }
 
+int gguf_get_version(struct gguf_context * ctx) {
+    return ctx->header.version;
+}
+
+size_t gguf_get_alignment(struct gguf_context * ctx) {
+    return ctx->alignment;
+}
+
+size_t gguf_get_data_offset(struct gguf_context * ctx) {
+    return ctx->offset;
+}
+
+void * gguf_get_data(struct gguf_context * ctx) {
+    return ctx->data;
+}
+
+int gguf_get_n_kv(struct gguf_context * ctx) {
+    return ctx->header.n_kv;
+}
+
+const char * gguf_get_key(struct gguf_context * ctx, int i) {
+    return ctx->header.kv[i].key.data;
+}
+
+enum gguf_type gguf_get_type(struct gguf_context * ctx, int i) {
+    return ctx->header.kv[i].type;
+}
+
+void gguf_get_val(struct gguf_context * ctx, int i, void * val) {
+    struct gguf_kv * kv = &ctx->header.kv[i];
+
+    switch (kv->type) {
+        case GGUF_TYPE_UINT8:   memcpy(val, &kv->value.uint8,    sizeof(uint8_t));  break;
+        case GGUF_TYPE_INT8:    memcpy(val, &kv->value.int8,     sizeof(int8_t));   break;
+        case GGUF_TYPE_UINT16:  memcpy(val, &kv->value.uint16,   sizeof(uint16_t)); break;
+        case GGUF_TYPE_INT16:   memcpy(val, &kv->value.int16,    sizeof(int16_t));  break;
+        case GGUF_TYPE_UINT32:  memcpy(val, &kv->value.uint32,   sizeof(uint32_t)); break;
+        case GGUF_TYPE_INT32:   memcpy(val, &kv->value.int32,    sizeof(int32_t));  break;
+        case GGUF_TYPE_FLOAT32: memcpy(val, &kv->value.float32,  sizeof(float));    break;
+        case GGUF_TYPE_BOOL:    memcpy(val, &kv->value.bool_,    sizeof(bool));     break;
+        case GGUF_TYPE_STRING:  memcpy(val, &kv->value.str.data, sizeof(char *));   break;
+        default:
+            GGML_ASSERT("gguf: not implemented");
+            break;
+    }
+}
+
+uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
+    uint8_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+int8_t gguf_get_val_i8(struct gguf_context * ctx, int i) {
+    int8_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+uint16_t gguf_get_val_u16(struct gguf_context * ctx, int i) {
+    uint16_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+int16_t gguf_get_val_i16(struct gguf_context * ctx, int i) {
+    int16_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+uint32_t gguf_get_val_u32(struct gguf_context * ctx, int i) {
+    uint32_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+int32_t gguf_get_val_i32(struct gguf_context * ctx, int i) {
+    int32_t val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+float gguf_get_val_f32(struct gguf_context * ctx, int i) {
+    float val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+bool gguf_get_val_bool(struct gguf_context * ctx, int i) {
+    bool val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+const char * gguf_get_val_str (struct gguf_context * ctx, int i) {
+    char * val;
+    gguf_get_val(ctx, i, &val);
+    return val;
+}
+
+int gguf_get_n_tensors(struct gguf_context * ctx) {
+    return ctx->header.n_tensors;
+}
+
+size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i) {
+    return ctx->infos[i].offset;
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 
 int ggml_cpu_has_avx(void) {
diff --git a/ggml.h b/ggml.h
index 1983bcd3e46fb..fac0f5e687e23 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1634,12 +1634,23 @@ extern "C" {
 
     struct gguf_context;
 
-    GGML_API struct gguf_context * gguf_init(const char * path, bool load);
+    struct gguf_init_params {
+        bool load;   // load the tensor data
+        bool malloc; // if false, use the provided ggml_context to allocate the tensor data
+                     //           it no ggml_context is provided, it will be created
+                     // if true,  use malloc to allocate the tensor data
+
+        struct ggml_context ** ctx;
+    };
+
+    GGML_API struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_params params);
+    //GGML_API struct gguf_context * gguf_init_from_buffer(..);
     GGML_API void                  gguf_free(struct gguf_context * ctx);
 
     GGML_API int    gguf_get_version    (struct gguf_context * ctx);
     GGML_API size_t gguf_get_alignment  (struct gguf_context * ctx);
     GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
+    GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
     GGML_API int            gguf_get_n_kv(struct gguf_context * ctx);
     GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);

From 860c9c63ce204dea31fe994ef370c569f7a75596 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 16:36:03 +0300
Subject: [PATCH 008/242] gguf : add gguf_get_tensor_name()

---
 ggml.c | 4 ++++
 ggml.h | 1 +
 2 files changed, 5 insertions(+)

diff --git a/ggml.c b/ggml.c
index f252363d960a3..0304750623148 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18770,6 +18770,10 @@ size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i) {
     return ctx->infos[i].offset;
 }
 
+char * gguf_get_tensor_name(struct gguf_context * ctx, int i) {
+    return ctx->infos[i].name.data;
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 
 int ggml_cpu_has_avx(void) {
diff --git a/ggml.h b/ggml.h
index fac0f5e687e23..1a748d8d82082 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1670,6 +1670,7 @@ extern "C" {
 
     GGML_API int    gguf_get_n_tensors    (struct gguf_context * ctx);
     GGML_API size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i);
+    GGML_API char * gguf_get_tensor_name  (struct gguf_context * ctx, int i);
 
     //
     // system info

From cb871fa022aa7a8b72c3a616f7ac7e8e9f1748d9 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 18:48:52 +0300
Subject: [PATCH 009/242] gguf : do not support passing existing ggml_context
 to gguf_init

---
 ggml.c | 51 ++++++++++++++++-----------------------------------
 ggml.h |  5 ++---
 2 files changed, 18 insertions(+), 38 deletions(-)

diff --git a/ggml.c b/ggml.c
index 0304750623148..1b14c37905b56 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18388,6 +18388,8 @@ static bool gguf_fread_str(void * dst, FILE * file, size_t * offset) {
 }
 
 struct gguf_context * gguf_init(const char * fname, struct gguf_init_params params) {
+    GGML_ASSERT(!params.load || params.malloc || params.ctx != NULL);
+
     FILE * file = fopen(fname, "rb");
     if (!file) {
         return NULL;
@@ -18518,8 +18520,7 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
 
         const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
 
-        // TODO: pad size_cur to alignment
-        ctx->size_data += size_cur;
+        ctx->size_data += GGML_PAD(size_cur, ctx->alignment);
     }
 
     // TODO: simplify
@@ -18528,28 +18529,18 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
             ctx->data = GGML_ALIGNED_MALLOC(ctx->size_data);
             fseek(file, ctx->offset, SEEK_SET);
             ok = ok && gguf_fread_el(ctx->data, ctx->size_data, file, &offset);
-        } else if (params.ctx != NULL) {
-            bool ctx_new = false;
-            bool ctx_no_alloc = false;
-
-            if (*params.ctx == NULL) {
-                const size_t mem_size =
-                    ctx->header.n_tensors*ggml_tensor_overhead() + 1 +
-                    ctx->size_data;
-
-                struct ggml_init_params pdata = {
-                    .mem_size   = mem_size,
-                    .mem_buffer = NULL,
-                    .no_alloc   = false,
-                };
+        } else {
+            const size_t mem_size =
+                ctx->header.n_tensors*ggml_tensor_overhead() + 1 +
+                ctx->size_data;
 
-                *params.ctx = ggml_init(pdata);
+            struct ggml_init_params pdata = {
+                .mem_size   = mem_size,
+                .mem_buffer = NULL,
+                .no_alloc   = false,
+            };
 
-                ctx_new = true;
-            } else {
-                ctx_no_alloc = ggml_get_no_alloc(*params.ctx);
-                ggml_set_no_alloc(*params.ctx, false);
-            }
+            *params.ctx = ggml_init(pdata);
 
             struct ggml_context * ctx_data = *params.ctx;
 
@@ -18561,11 +18552,7 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
             if (!ok) {
                 fprintf(stderr, "%s: failed to read tensor data\n", __func__);
                 fclose(file);
-                if (ctx_new) {
-                    ggml_free(ctx_data);
-                } else {
-                    ggml_set_no_alloc(ctx_data, ctx_no_alloc);
-                }
+                ggml_free(ctx_data);
                 gguf_free(ctx);
                 return NULL;
             }
@@ -18597,18 +18584,12 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
             if (!ok) {
                 fprintf(stderr, "%s: failed to create tensors\n", __func__);
                 fclose(file);
-                if (ctx_new) {
-                    ggml_free(ctx_data);
-                } else {
-                    ggml_set_no_alloc(ctx_data, ctx_no_alloc);
-                }
+                ggml_free(ctx_data);
                 gguf_free(ctx);
                 return NULL;
             }
 
-            ggml_set_no_alloc(ctx_data, ctx_no_alloc);
-        } else {
-            GGML_ASSERT("gguf: invalid params - load requires malloc or ctx");
+            ggml_set_no_alloc(ctx_data, false);
         }
     }
 
diff --git a/ggml.h b/ggml.h
index 1a748d8d82082..7d5514ba394f5 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1636,9 +1636,8 @@ extern "C" {
 
     struct gguf_init_params {
         bool load;   // load the tensor data
-        bool malloc; // if false, use the provided ggml_context to allocate the tensor data
-                     //           it no ggml_context is provided, it will be created
-                     // if true,  use malloc to allocate the tensor data
+        bool malloc; // if false, create a ggml_context and allocate the tensor data in it
+                     // if  true, use malloc to allocate the tensor data instead
 
         struct ggml_context ** ctx;
     };

From d313c0fa33dc284e23a88fae90b1f94cb0ff6f5c Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 18:53:57 +0300
Subject: [PATCH 010/242] gguf : simplify gguf_get_val

---
 ggml.c | 68 +++++++++++++++++++---------------------------------------
 ggml.h | 20 +++--------------
 2 files changed, 25 insertions(+), 63 deletions(-)

diff --git a/ggml.c b/ggml.c
index 1b14c37905b56..e68e91e18646b 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18297,6 +18297,19 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i
 
 ////////////////////////////////////////////////////////////////////////////////
 
+enum gguf_type {
+    GGUF_TYPE_UINT8   = 0,
+    GGUF_TYPE_INT8    = 1,
+    GGUF_TYPE_UINT16  = 2,
+    GGUF_TYPE_INT16   = 3,
+    GGUF_TYPE_UINT32  = 4,
+    GGUF_TYPE_INT32   = 5,
+    GGUF_TYPE_FLOAT32 = 6,
+    GGUF_TYPE_BOOL    = 7,
+    GGUF_TYPE_STRING  = 8,
+    GGUF_TYPE_ARRAY   = 9,
+};
+
 struct gguf_str {
     uint32_t n;
     char * data;
@@ -18670,77 +18683,40 @@ enum gguf_type gguf_get_type(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].type;
 }
 
-void gguf_get_val(struct gguf_context * ctx, int i, void * val) {
-    struct gguf_kv * kv = &ctx->header.kv[i];
-
-    switch (kv->type) {
-        case GGUF_TYPE_UINT8:   memcpy(val, &kv->value.uint8,    sizeof(uint8_t));  break;
-        case GGUF_TYPE_INT8:    memcpy(val, &kv->value.int8,     sizeof(int8_t));   break;
-        case GGUF_TYPE_UINT16:  memcpy(val, &kv->value.uint16,   sizeof(uint16_t)); break;
-        case GGUF_TYPE_INT16:   memcpy(val, &kv->value.int16,    sizeof(int16_t));  break;
-        case GGUF_TYPE_UINT32:  memcpy(val, &kv->value.uint32,   sizeof(uint32_t)); break;
-        case GGUF_TYPE_INT32:   memcpy(val, &kv->value.int32,    sizeof(int32_t));  break;
-        case GGUF_TYPE_FLOAT32: memcpy(val, &kv->value.float32,  sizeof(float));    break;
-        case GGUF_TYPE_BOOL:    memcpy(val, &kv->value.bool_,    sizeof(bool));     break;
-        case GGUF_TYPE_STRING:  memcpy(val, &kv->value.str.data, sizeof(char *));   break;
-        default:
-            GGML_ASSERT("gguf: not implemented");
-            break;
-    }
-}
-
 uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
-    uint8_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.uint8;
 }
 
 int8_t gguf_get_val_i8(struct gguf_context * ctx, int i) {
-    int8_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.int8;
 }
 
 uint16_t gguf_get_val_u16(struct gguf_context * ctx, int i) {
-    uint16_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.uint16;
 }
 
 int16_t gguf_get_val_i16(struct gguf_context * ctx, int i) {
-    int16_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.int16;
 }
 
 uint32_t gguf_get_val_u32(struct gguf_context * ctx, int i) {
-    uint32_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.uint32;
 }
 
 int32_t gguf_get_val_i32(struct gguf_context * ctx, int i) {
-    int32_t val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.int32;
 }
 
 float gguf_get_val_f32(struct gguf_context * ctx, int i) {
-    float val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.float32;
 }
 
 bool gguf_get_val_bool(struct gguf_context * ctx, int i) {
-    bool val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.bool_;
 }
 
 const char * gguf_get_val_str (struct gguf_context * ctx, int i) {
-    char * val;
-    gguf_get_val(ctx, i, &val);
-    return val;
+    return ctx->header.kv[i].value.str.data;
 }
 
 int gguf_get_n_tensors(struct gguf_context * ctx) {
diff --git a/ggml.h b/ggml.h
index 7d5514ba394f5..75a41a28f6653 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1619,19 +1619,6 @@ extern "C" {
     // gguf
     //
 
-    enum gguf_type {
-        GGUF_TYPE_UINT8   = 0,
-        GGUF_TYPE_INT8    = 1,
-        GGUF_TYPE_UINT16  = 2,
-        GGUF_TYPE_INT16   = 3,
-        GGUF_TYPE_UINT32  = 4,
-        GGUF_TYPE_INT32   = 5,
-        GGUF_TYPE_FLOAT32 = 6,
-        GGUF_TYPE_BOOL    = 7,
-        GGUF_TYPE_STRING  = 8,
-        GGUF_TYPE_ARRAY   = 9,
-    };
-
     struct gguf_context;
 
     struct gguf_init_params {
@@ -1651,10 +1638,9 @@ extern "C" {
     GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
     GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
-    GGML_API int            gguf_get_n_kv(struct gguf_context * ctx);
-    GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);
-    GGML_API enum gguf_type gguf_get_type(struct gguf_context * ctx, int i);
-    GGML_API void           gguf_get_val (struct gguf_context * ctx, int i, void * val);
+    GGML_API int          gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API const char * gguf_get_key (struct gguf_context * ctx, int i);
+    GGML_API void         gguf_get_val (struct gguf_context * ctx, int i, void * val);
 
     GGML_API uint8_t      gguf_get_val_u8  (struct gguf_context * ctx, int i);
     GGML_API int8_t       gguf_get_val_i8  (struct gguf_context * ctx, int i);

From e46870f5af3432212982645fcd1cc59b8e106734 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 18:55:32 +0300
Subject: [PATCH 011/242] gguf : gguf.c is now part of ggml.c

---
 gguf.c | 192 ---------------------------------------------------------
 1 file changed, 192 deletions(-)
 delete mode 100644 gguf.c

diff --git a/gguf.c b/gguf.c
deleted file mode 100644
index 54b31d411f35b..0000000000000
--- a/gguf.c
+++ /dev/null
@@ -1,192 +0,0 @@
-// TODO: convert to proper gguf.h gguf.c structure, now I'm trying to be fast as much as possible,
-// and everything is in this file for quick debugging.
-
-#include <stdio.h>
-#include <stdint.h>
-#include <stdlib.h>
-#include <stdbool.h>
-
-
-enum ggml_type {
-    GGML_TYPE_F32  = 0,
-    GGML_TYPE_F16  = 1,
-    GGML_TYPE_Q4_0 = 2,
-    GGML_TYPE_Q4_1 = 3,
-    // GGML_TYPE_Q4_2 = 4, support has been removed
-    // GGML_TYPE_Q4_3 (5) support has been removed
-    GGML_TYPE_Q5_0 = 6,
-    GGML_TYPE_Q5_1 = 7,
-    GGML_TYPE_Q8_0 = 8,
-    GGML_TYPE_Q8_1 = 9,
-    // k-quantizations
-    GGML_TYPE_Q2_K = 10,
-    GGML_TYPE_Q3_K = 11,
-    GGML_TYPE_Q4_K = 12,
-    GGML_TYPE_Q5_K = 13,
-    GGML_TYPE_Q6_K = 14,
-    GGML_TYPE_Q8_K = 15,
-    GGML_TYPE_I8,
-    GGML_TYPE_I16,
-    GGML_TYPE_I32,
-    GGML_TYPE_COUNT,
-};
-
-enum gguf_metadata_value_type {
-    GGUF_METADATA_VALUE_TYPE_UINT8 = 0,
-    GGUF_METADATA_VALUE_TYPE_INT8 = 1,
-    GGUF_METADATA_VALUE_TYPE_UINT16 = 2,
-    GGUF_METADATA_VALUE_TYPE_INT16 = 3,
-    GGUF_METADATA_VALUE_TYPE_UINT32 = 4,
-    GGUF_METADATA_VALUE_TYPE_INT32 = 5,
-    GGUF_METADATA_VALUE_TYPE_FLOAT32 = 6,
-    GGUF_METADATA_VALUE_TYPE_BOOL = 7,
-    GGUF_METADATA_VALUE_TYPE_STRING = 8,
-    GGUF_METADATA_VALUE_TYPE_ARRAY = 9,
-};
-
-struct gguf_string_t {
-    uint32_t len;
-    char * string;
-};
-
-union gguf_metadata_value_t;
-
-// Union definition for gguf_metadata_value_t
-union gguf_metadata_value_t {
-    uint8_t uint8;
-    int8_t int8;
-    uint16_t uint16;
-    int16_t int16;
-    uint32_t uint32;
-    int32_t int32;
-    float float32;
-    bool bool_;
-    struct gguf_string_t string;
-    struct {
-        uint32_t len;
-        enum gguf_metadata_value_type type;
-        union gguf_metadata_value_t * array;
-    } array;
-};
-
-
-struct gguf_metadata_kv_t {
-    struct gguf_string_t key;
-    uint32_t value_len;
-    enum gguf_metadata_value_type value_type;
-    union gguf_metadata_value_t* value;
-};
-
-struct gguf_header_t {
-    uint32_t magic;
-    uint32_t version;
-    uint32_t tensor_count;
-    uint32_t metadata_kv_count;
-    struct gguf_metadata_kv_t * metadata_kv;
-};
-
-struct gguf_tensor_info_t {
-    struct gguf_string_t name;
-    uint32_t n_dimensions;
-    uint32_t dimensions[];
-};
-
-struct gguf_file_t {
-    struct gguf_header_t header;
-    uint8_t tensor_data[];
-};
-
-void read_gguf_file(const char * file_path, struct gguf_file_t * gguf_file) {
-    FILE* file = fopen(file_path, "rb");
-    if (file == NULL) {
-        printf("Error opening the file.\n");
-        return;
-    }
-
-    fread(&gguf_file->header.magic, sizeof(uint32_t), 1, file);
-
-    // Verify magic and version
-    if (gguf_file->header.magic != 0x47475546) {
-        printf("Invalid magic number. Not a valid GGUF file.\n");
-        fclose(file);
-        return;
-    }
-
-    fread(&gguf_file->header.version, sizeof(uint32_t), 1, file);
-
-    if (gguf_file->header.version != 1) {
-        printf("Unsupported version. Expected version 1.\n");
-        fclose(file);
-        return;
-    }
-
-    fread(&gguf_file->header.tensor_count, sizeof(uint32_t), 1, file);
-    fread(&gguf_file->header.metadata_kv_count, sizeof(uint32_t), 1, file);
-
-    printf("Magic: %x\n", gguf_file->header.magic);
-    printf("Version: %d\n", gguf_file->header.version);
-    printf("Tensor Count: %d\n", gguf_file->header.tensor_count);
-    printf("Metadata Key-Value Count: %d\n", gguf_file->header.metadata_kv_count);
-
-    gguf_file->header.metadata_kv = (struct gguf_metadata_kv_t*)malloc(gguf_file->header.metadata_kv_count * sizeof(struct gguf_metadata_kv_t));
-
-    for (int i = 0; i < gguf_file->header.metadata_kv_count; i++) {
-        struct gguf_metadata_kv_t* kv = &gguf_file->header.metadata_kv[i];
-        fread(&kv->key.len, sizeof(uint32_t), 1, file);
-        kv->key.string = (char*)malloc(kv->key.len ); // Allocate memory for the key string
-        fread(kv->key.string, sizeof(char), kv->key.len, file);
-        //kv->key.string[kv->key.len] = '\0'; // Null-terminate the key string
-
-        fread(&kv->value_type, sizeof(uint32_t), 1, file);
-
-        printf("Metadata Value Type: %d\n", kv->value_type);
-        printf("Metadata Key: %s\n", kv->key.string);
-
-        // Read metadata value according to its type using reinterpret_cast
-        switch (kv->value_type) {
-            case GGUF_METADATA_VALUE_TYPE_UINT32:
-            kv->value = (uint32_t *) malloc(sizeof(uint32_t));
-            fread(kv->value, sizeof(uint32_t), 1, file);
-            printf("value: %d\n", kv->value->uint32);
-            break;
-            case GGUF_METADATA_VALUE_TYPE_FLOAT32:
-            kv->value = (float *)malloc(sizeof(float));
-            fread(kv->value, sizeof(float), 1, file);
-            printf("value: %f\n", (float)kv->value->float32);
-            break;
-            case GGUF_METADATA_VALUE_TYPE_STRING:
-            fread(&kv->value_len, sizeof(uint32_t), 1, file);
-            printf("value len: %d\n", kv->value_len);
-kv->value = (char *)malloc(sizeof(char) * kv->value_len); // Allocate memory for the value string
-fread(kv->value, sizeof(char), kv->value_len, file);
-        printf("value: %s\n", (char *)kv->value);
-        break;
-            // ... (handle other types in a similar manner)
-            default:
-                printf("Unsupported metadata value type.\n");
-                fclose(file);
-                return;
-        }
-    }
-
-    // TODO: handle reading tensor data
-
-    fclose(file);
-}
-
-void gguf_free(struct gguf_file_t * gguf_file) {
-    // Free allocated memory for key strings avd values
-    for (int i = 0; i < gguf_file->header.metadata_kv_count; i++) {
-        free(gguf_file->header.metadata_kv[i].key.string);
-        free(gguf_file->header.metadata_kv[i].value);
-    }
-    free(gguf_file->header.metadata_kv);
-}
-
-int main() {
-    const char* file_path = "example.gguf";
-    struct gguf_file_t gguf_file;
-    read_gguf_file(file_path, &gguf_file);
-    gguf_free(&gguf_file);
-    return 0;
-}

From 5628ec71636e0b390213caa4c273d3ef8bbd7459 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 20:04:22 +0300
Subject: [PATCH 012/242] gguf : read / write sample models

---
 examples/gguf/gguf.cpp | 323 ++++++++++++++++++++++++++++++++++++++++-
 ggml.c                 | 117 ++++++++-------
 ggml.h                 |   5 +-
 3 files changed, 385 insertions(+), 60 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 602de519adb1e..a5c442ac581f5 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,15 +1,326 @@
 #include "ggml.h"
 
 #include <cstdio>
+#include <cinttypes>
 #include <string>
+#include <sstream>
+#include <fstream>
+#include <vector>
 
-bool gguf_write(const std::string & fname) {
+enum gguf_type {
+    GGUF_TYPE_UINT8   = 0,
+    GGUF_TYPE_INT8    = 1,
+    GGUF_TYPE_UINT16  = 2,
+    GGUF_TYPE_INT16   = 3,
+    GGUF_TYPE_UINT32  = 4,
+    GGUF_TYPE_INT32   = 5,
+    GGUF_TYPE_FLOAT32 = 6,
+    GGUF_TYPE_BOOL    = 7,
+    GGUF_TYPE_STRING  = 8,
+    GGUF_TYPE_ARRAY   = 9,
+};
 
+template<typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
+
+void gguf_ex_write_str(std::ofstream & fout, const std::string & val) {
+    const int32_t n = val.size();
+    fout.write((const char *) &n, sizeof(n));
+    fout.write(val.c_str(), n);
+}
+
+void gguf_ex_write_i32(std::ofstream & fout, int32_t val) {
+    fout.write((const char *) &val, sizeof(val));
+}
+
+void gguf_ex_write_u64(std::ofstream & fout, size_t val) {
+    fout.write((const char *) &val, sizeof(val));
+}
+
+template<typename T>
+void gguf_ex_write_param(std::ofstream & fout, const std::string & key, enum gguf_type type, const T & val) {
+    gguf_ex_write_str(fout, key);
+    fout.write((const char *) &type, sizeof(type));
+    fout.write((const char *) &val,  sizeof(val));
+
+    fprintf(stdout, "%s: write param: %s = %s\n", __func__, key.c_str(), to_string(val).c_str());
+}
+
+template<>
+void gguf_ex_write_param<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::string & val) {
+    gguf_ex_write_str(fout, key);
+    fout.write((const char *) &type, sizeof(type));
+
+    const int32_t n = val.size();
+    fout.write((const char *) &n, sizeof(n));
+    fout.write(val.c_str(), n);
+}
+
+bool gguf_ex_write(const std::string & fname) {
+    std::ofstream fout(fname.c_str(), std::ios::binary);
+
+    {
+        const int32_t magic = GGUF_MAGIC;
+        fout.write((const char *) &magic, sizeof(magic));
+    }
+
+    {
+        const int32_t version = GGUF_VERSION;
+        fout.write((const char *) &version, sizeof(version));
+    }
+
+    const int n_tensors = 10;
+    const int n_kv = 9;
+
+    fout.write((const char*) &n_tensors, sizeof(n_tensors));
+    fout.write((const char*) &n_kv, sizeof(n_kv));
+
+    fprintf(stdout, "%s: write header\n", __func__);
+
+    // kv data
+    {
+        gguf_ex_write_param< uint8_t>(fout, "some.parameter.uint8",   GGUF_TYPE_UINT8,   0x12);
+        gguf_ex_write_param<  int8_t>(fout, "some.parameter.int8",    GGUF_TYPE_INT8,   -0x13);
+        gguf_ex_write_param<uint16_t>(fout, "some.parameter.uint16",  GGUF_TYPE_UINT16,  0x1234);
+        gguf_ex_write_param< int16_t>(fout, "some.parameter.int16",   GGUF_TYPE_INT16,  -0x1235);
+        gguf_ex_write_param<uint32_t>(fout, "some.parameter.uint32",  GGUF_TYPE_UINT32,  0x12345678);
+        gguf_ex_write_param< int32_t>(fout, "some.parameter.int32",   GGUF_TYPE_INT32,  -0x12345679);
+
+        gguf_ex_write_param<float>   (fout, "some.parameter.float32", GGUF_TYPE_FLOAT32, 0.123456789f);
+        gguf_ex_write_param<bool>    (fout, "some.parameter.bool",    GGUF_TYPE_BOOL,    true);
+
+        gguf_ex_write_param<std::string>(fout, "some.parameter.string",  GGUF_TYPE_STRING,  "hello world");
+    }
+
+    uint64_t offset_tensor = 0;
+
+    struct ggml_init_params params = {
+        /*.mem_size   =*/ 128ull*1024ull*1024ull,
+        /*.mem_buffer =*/ NULL,
+        /*.no_alloc   =*/ false,
+    };
+
+    struct ggml_context * ctx_data = ggml_init(params);
+
+    // tensor infos
+    for (int i = 0; i < n_tensors; ++i) {
+        const std::string name = "tensor_" + to_string(i);
+
+        int64_t ne[GGML_MAX_DIMS] = { 1 };
+        int32_t n_dims = rand() % GGML_MAX_DIMS + 1;
+
+        for (int j = 0; j < n_dims; ++j) {
+            ne[j] = rand() % 10 + 1;
+        }
+
+        struct ggml_tensor * cur = ggml_new_tensor(ctx_data, GGML_TYPE_F32, n_dims, ne);
+        ggml_set_name(cur, name.c_str());
+
+        {
+            float * data = (float *) cur->data;
+            for (int j = 0; j < ggml_nelements(cur); ++j) {
+                data[j] = 100 + i;
+            }
+        }
+
+        fprintf(stdout, "%s: tensor: %s, %d dims, ne = [", __func__, name.c_str(), n_dims);
+        for (int j = 0; j < 4; ++j) {
+            fprintf(stdout, "%s%3d", j == 0 ? "" : ", ", (int) cur->ne[j]);
+        }
+        fprintf(stdout, "], offset_tensor = %6" PRIu64 "\n", offset_tensor);
+
+        gguf_ex_write_str(fout, name);
+        gguf_ex_write_i32(fout, n_dims);
+        for (int j = 0; j < n_dims; ++j) {
+            gguf_ex_write_i32(fout, cur->ne[j]);
+        }
+        gguf_ex_write_i32(fout, cur->type);
+        gguf_ex_write_u64(fout, offset_tensor);
+
+        offset_tensor += GGML_PAD(ggml_nbytes(cur), GGUF_DEFAULT_ALIGNMENT);
+    }
+
+    const uint64_t offset_data = GGML_PAD((uint64_t) fout.tellp(), GGUF_DEFAULT_ALIGNMENT);
+
+    fprintf(stdout, "%s: data offset = %" PRIu64 "\n", __func__, offset_data);
+
+    {
+        const size_t pad = offset_data - fout.tellp();
+
+        for (size_t j = 0; j < pad; ++j) {
+            fout.put(0);
+        }
+    }
+
+    for (int i = 0; i < n_tensors; ++i) {
+        fprintf(stdout, "%s: writing tensor %d data\n", __func__, i);
+
+        const std::string name = "tensor_" + to_string(i);
+
+        struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name.c_str());
+
+        fout.write((const char *) cur->data, ggml_nbytes(cur));
+
+        {
+            const size_t pad = GGML_PAD(ggml_nbytes(cur), GGUF_DEFAULT_ALIGNMENT) - ggml_nbytes(cur);
+
+            for (size_t j = 0; j < pad; ++j) {
+                fout.put(0);
+            }
+        }
+    }
+
+    fout.close();
+
+    fprintf(stdout, "%s: wrote file '%s;\n", __func__, fname.c_str());
+
+    ggml_free(ctx_data);
+
+    return true;
+}
+
+// just read tensor info
+bool gguf_ex_read_0(const std::string & fname) {
+    struct gguf_init_params params = {
+        /*.no_alloc = */ false,
+        /*.ctx      = */ NULL,
+    };
+
+    struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
+
+    fprintf(stdout, "%s: version:      %d\n", __func__, gguf_get_version(ctx));
+    fprintf(stdout, "%s: alignment:   %zu\n", __func__, gguf_get_alignment(ctx));
+    fprintf(stdout, "%s: data offset: %zu\n", __func__, gguf_get_data_offset(ctx));
+
+    // kv
+    {
+        const int n_kv = gguf_get_n_kv(ctx);
+
+        fprintf(stdout, "%s: n_kv: %d\n", __func__, n_kv);
+
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(ctx, i);
+
+            fprintf(stdout, "%s: kv[%d]: key = %s\n", __func__, i, key);
+        }
+    }
+
+    // tensor info
+    {
+        const int n_tensors = gguf_get_n_tensors(ctx);
+
+        fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            const char * name = gguf_get_tensor_name(ctx, i);
+            const size_t offset = gguf_get_tensor_offset(ctx, i);
+
+            fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
+        }
+    }
+
+    return true;
+}
+
+// read and create ggml_context containing the tensors and their data
+bool gguf_ex_read_1(const std::string & fname) {
+    struct ggml_context * ctx_data = NULL;
+
+    struct gguf_init_params params = {
+        /*.no_alloc = */ false,
+        /*.ctx      = */ &ctx_data,
+    };
+
+    struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
+
+    fprintf(stdout, "%s: version:      %d\n", __func__, gguf_get_version(ctx));
+    fprintf(stdout, "%s: alignment:   %zu\n", __func__, gguf_get_alignment(ctx));
+    fprintf(stdout, "%s: data offset: %zu\n", __func__, gguf_get_data_offset(ctx));
+
+    // kv
+    {
+        const int n_kv = gguf_get_n_kv(ctx);
+
+        fprintf(stdout, "%s: n_kv: %d\n", __func__, n_kv);
+
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(ctx, i);
+
+            fprintf(stdout, "%s: kv[%d]: key = %s\n", __func__, i, key);
+        }
+    }
+
+    // tensor info
+    {
+        const int n_tensors = gguf_get_n_tensors(ctx);
+
+        fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            const char * name = gguf_get_tensor_name(ctx, i);
+            const size_t offset = gguf_get_tensor_offset(ctx, i);
+
+            fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
+        }
+    }
+
+    // data
+    {
+        const int n_tensors = gguf_get_n_tensors(ctx);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            fprintf(stdout, "%s: reading tensor %d data\n", __func__, i);
+
+            const std::string name = "tensor_" + to_string(i);
+
+            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name.c_str());
+
+            fprintf(stdout, "%s: tensor[%d]: n_dims = %d, name = %s, data = %p\n",
+                    __func__, i, cur->n_dims, cur->name, cur->data);
+
+            // check data
+            {
+                const float * data = (const float *) cur->data;
+                for (int j = 0; j < ggml_nelements(cur); ++j) {
+                    if (data[j] != 100 + i) {
+                        fprintf(stderr, "%s: tensor[%d]: data[%d] = %f\n", __func__, i, j, data[j]);
+                        return false;
+                    }
+                }
+            }
+        }
+    }
+
+    fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
+
+    ggml_free(ctx_data);
+    gguf_free(ctx);
 
     return true;
 }
 
-bool gguf_read(const std::string & fname) {
+// read just the tensor info and mmap the data in user code
+bool gguf_ex_read_2(const std::string & fname) {
+    struct ggml_context * ctx_data = NULL;
+
+    struct gguf_init_params params = {
+        /*.no_alloc = */ true,
+        /*.ctx      = */ &ctx_data,
+    };
+
+    struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
+
+    // TODO: mmap based on tensor infos
+
+    fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
+
+    ggml_free(ctx_data);
+    gguf_free(ctx);
+
     return true;
 }
 
@@ -20,14 +331,16 @@ int main(int argc, char ** argv) {
     }
 
     const std::string fname(argv[1]);
-    const std::string mode(argv[2]);
+    const std::string mode (argv[2]);
 
     GGML_ASSERT((mode == "r" || mode == "w") && "mode must be r or w");
 
     if (mode == "w") {
-        GGML_ASSERT(gguf_write(fname) && "failed to write gguf file");
+        GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
     } else if (mode == "r") {
-        GGML_ASSERT(gguf_read(fname)  && "failed to read gguf file");
+        GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
+        GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
+        GGML_ASSERT(gguf_ex_read_2(fname) && "failed to read gguf file");
     }
 
     return 0;
diff --git a/ggml.c b/ggml.c
index e68e91e18646b..5736c800ebe69 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18364,7 +18364,7 @@ struct gguf_tensor_info {
 
     enum ggml_type type;
 
-    uint64_t offset; // offset from beginning of file, must be a multiple of `ALIGNMENT`
+    uint64_t offset; // offset from start of `data`, must be a multiple of `ALIGNMENT`
 };
 
 struct gguf_context {
@@ -18385,9 +18385,7 @@ static bool gguf_fread_el(void * dst, size_t size, FILE * file, size_t * offset)
     return n == size;
 }
 
-static bool gguf_fread_str(void * dst, FILE * file, size_t * offset) {
-    struct gguf_str * p = (struct gguf_str *) dst;
-
+static bool gguf_fread_str(struct gguf_str * p, FILE * file, size_t * offset) {
     p->n    = 0;
     p->data = NULL;
 
@@ -18395,14 +18393,12 @@ static bool gguf_fread_str(void * dst, FILE * file, size_t * offset) {
 
     // TODO: how to avoid mallocs for strings?
     ok = ok && gguf_fread_el(&p->n,    sizeof(p->n), file, offset); p->data = calloc(p->n + 1, 1);
-    ok = ok && gguf_fread_el(&p->data, p->n,         file, offset);
+    ok = ok && gguf_fread_el( p->data, p->n,         file, offset);
 
     return ok;
 }
 
-struct gguf_context * gguf_init(const char * fname, struct gguf_init_params params) {
-    GGML_ASSERT(!params.load || params.malloc || params.ctx != NULL);
-
+struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_params params) {
     FILE * file = fopen(fname, "rb");
     if (!file) {
         return NULL;
@@ -18446,10 +18442,14 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
     for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
         struct gguf_kv * kv = &ctx->header.kv[i];
 
+        //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
+
         ok = ok && gguf_fread_str(&kv->key,                          file, &offset);
       //ok = ok && gguf_fread_el (&kv->n_bytes, sizeof(kv->n_bytes), file, &offset);
         ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
 
+        //fprintf(stderr, "%s: reading kv with key %s\n", __func__, kv->key.data);
+
         switch (kv->type) {
             case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
             case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
@@ -18461,9 +18461,13 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
             case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
             case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
             case GGUF_TYPE_ARRAY:
-                GGML_ASSERT("gguf: array type not implemented");
-                break;
-            };
+                                    GGML_ASSERT("gguf: array type not implemented");
+                                    break;
+        };
+
+        if (!ok) {
+            break;
+        }
     }
 
     if (!ok) {
@@ -18478,12 +18482,14 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
     for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
         struct gguf_tensor_info * info = &ctx->infos[i];
 
-        memset(info->ne, 1, sizeof(info->ne));
+        for (int j = 0; j < GGML_MAX_DIMS; ++j) {
+            info->ne[j] = 1;
+        }
 
         ok = ok && gguf_fread_str(&info->name,                          file, &offset);
         ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
         for (uint32_t j = 0; j < info->n_dims; ++j) {
-            ok = ok && gguf_fread_el (&info->ne[j], sizeof(info->ne[j]), file, &offset);
+            ok = ok && gguf_fread_el(&info->ne[j], sizeof(info->ne[j]), file, &offset);
         }
       //ok = ok && gguf_fread_el (&info->n_elms, sizeof(info->n_elms),  file, &offset);
         ok = ok && gguf_fread_el (&info->type,    sizeof(info->type),   file, &offset);
@@ -18536,28 +18542,30 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
         ctx->size_data += GGML_PAD(size_cur, ctx->alignment);
     }
 
+    // load the tensor data
     // TODO: simplify
-    if (params.load) {
-        if (params.malloc) {
-            ctx->data = GGML_ALIGNED_MALLOC(ctx->size_data);
-            fseek(file, ctx->offset, SEEK_SET);
-            ok = ok && gguf_fread_el(ctx->data, ctx->size_data, file, &offset);
-        } else {
-            const size_t mem_size =
-                ctx->header.n_tensors*ggml_tensor_overhead() + 1 +
-                ctx->size_data;
+    if (params.ctx != NULL) {
+        const size_t mem_size =
+            params.no_alloc ?
+            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() :
+            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() + ctx->size_data;
+
+        struct ggml_init_params pdata = {
+            .mem_size   = mem_size,
+            .mem_buffer = NULL,
+            .no_alloc   = params.no_alloc,
+        };
 
-            struct ggml_init_params pdata = {
-                .mem_size   = mem_size,
-                .mem_buffer = NULL,
-                .no_alloc   = false,
-            };
+        *params.ctx = ggml_init(pdata);
+
+        struct ggml_context * ctx_data = *params.ctx;
 
-            *params.ctx = ggml_init(pdata);
+        struct ggml_tensor * data = NULL;
 
-            struct ggml_context * ctx_data = *params.ctx;
+        if (params.no_alloc == false) {
+            data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size_data);
 
-            struct ggml_tensor * data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size_data);
+            ok = ok && data != NULL;
 
             // read the tensor data
             ok = ok && gguf_fread_el(data->data, ctx->size_data, file, &offset);
@@ -18571,39 +18579,44 @@ struct gguf_context * gguf_init(const char * fname, struct gguf_init_params para
             }
 
             ctx->data = data->data;
+        }
 
-            // create the tensors
-            ggml_set_no_alloc(ctx_data, true);
+        ggml_set_no_alloc(ctx_data, true);
 
-            for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
-                const int64_t ne[GGML_MAX_DIMS] = {
-                    ctx->infos[i].ne[0],
-                    ctx->infos[i].ne[1],
-                    ctx->infos[i].ne[2],
-                    ctx->infos[i].ne[3],
-                };
+        // create the tensors
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            const int64_t ne[GGML_MAX_DIMS] = {
+                ctx->infos[i].ne[0],
+                ctx->infos[i].ne[1],
+                ctx->infos[i].ne[2],
+                ctx->infos[i].ne[3],
+            };
 
-                struct ggml_tensor * cur = ggml_new_tensor(ctx_data, ctx->infos[i].type, ctx->infos[i].n_dims, ne);
+            struct ggml_tensor * cur = ggml_new_tensor(ctx_data, ctx->infos[i].type, ctx->infos[i].n_dims, ne);
 
-                ok = ok && cur != NULL;
+            ok = ok && cur != NULL;
 
-                if (!ok) {
-                    break;
-                }
+            ggml_set_name(cur, ctx->infos[i].name.data);
 
-                cur->data = (char *) data->data + ctx->infos[i].offset - ctx->offset;
+            if (!ok) {
+                break;
             }
 
-            if (!ok) {
-                fprintf(stderr, "%s: failed to create tensors\n", __func__);
-                fclose(file);
-                ggml_free(ctx_data);
-                gguf_free(ctx);
-                return NULL;
+            if (params.no_alloc == false) {
+              //cur->data = (char *) data->data + ctx->infos[i].offset - ctx->offset; // offset from start of file
+                cur->data = (char *) data->data + ctx->infos[i].offset;               // offset from data
             }
+        }
 
-            ggml_set_no_alloc(ctx_data, false);
+        if (!ok) {
+            fprintf(stderr, "%s: failed to create tensors\n", __func__);
+            fclose(file);
+            ggml_free(ctx_data);
+            gguf_free(ctx);
+            return NULL;
         }
+
+        ggml_set_no_alloc(ctx_data, params.no_alloc);
     }
 
     if (!ok) {
diff --git a/ggml.h b/ggml.h
index 75a41a28f6653..e0abbbfdd81e3 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1622,10 +1622,9 @@ extern "C" {
     struct gguf_context;
 
     struct gguf_init_params {
-        bool load;   // load the tensor data
-        bool malloc; // if false, create a ggml_context and allocate the tensor data in it
-                     // if  true, use malloc to allocate the tensor data instead
+        bool no_alloc;
 
+        // if not NULL, create a ggml_context and allocate the tensor data in it
         struct ggml_context ** ctx;
     };
 

From d8491fc7e3545a4447555cb9e4487994bc98c024 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 26 Jul 2023 22:56:26 +0300
Subject: [PATCH 013/242] gguf : add comments

---
 ggml.c | 212 +++++++++++++++++++++++++++++++--------------------------
 1 file changed, 114 insertions(+), 98 deletions(-)

diff --git a/ggml.c b/ggml.c
index 5736c800ebe69..b005fd8893e58 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18407,99 +18407,112 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
     // offset from start of file
     size_t offset = 0;
 
-    // check the magic before making allocations
     uint32_t magic = 0;
-    gguf_fread_el(&magic, sizeof(magic), file, &offset);
-    if (magic != GGUF_MAGIC) {
-        fprintf(stderr, "%s: invalid magic number %08x\n", __func__, magic);
-        fclose(file);
-        return NULL;
+
+    // check the magic before making allocations
+    {
+        gguf_fread_el(&magic, sizeof(magic), file, &offset);
+
+        if (magic != GGUF_MAGIC) {
+            fprintf(stderr, "%s: invalid magic number %08x\n", __func__, magic);
+            fclose(file);
+            return NULL;
+        }
     }
 
     bool ok = true;
 
     struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
 
-    ctx->header.magic = magic;
-    ctx->header.kv    = NULL;
+    // read the header
+    {
+        ctx->header.magic = magic;
+        ctx->header.kv    = NULL;
 
-    ctx->infos = NULL;
-    ctx->data  = NULL;
+        ctx->infos = NULL;
+        ctx->data  = NULL;
 
-    ok = ok && gguf_fread_el(&ctx->header.version,   sizeof(ctx->header.version),   file, &offset);
-    ok = ok && gguf_fread_el(&ctx->header.n_tensors, sizeof(ctx->header.n_tensors), file, &offset);
-    ok = ok && gguf_fread_el(&ctx->header.n_kv,      sizeof(ctx->header.n_kv),      file, &offset);
+        ok = ok && gguf_fread_el(&ctx->header.version,   sizeof(ctx->header.version),   file, &offset);
+        ok = ok && gguf_fread_el(&ctx->header.n_tensors, sizeof(ctx->header.n_tensors), file, &offset);
+        ok = ok && gguf_fread_el(&ctx->header.n_kv,      sizeof(ctx->header.n_kv),      file, &offset);
 
-    if (!ok) {
-        fprintf(stderr, "%s: failed to read header\n", __func__);
-        fclose(file);
-        gguf_free(ctx);
-        return NULL;
+        if (!ok) {
+            fprintf(stderr, "%s: failed to read header\n", __func__);
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
+        }
     }
 
-    ctx->header.kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
-
-    for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
-        struct gguf_kv * kv = &ctx->header.kv[i];
-
-        //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
+    // read the kv pairs
+    {
+        ctx->header.kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
 
-        ok = ok && gguf_fread_str(&kv->key,                          file, &offset);
-      //ok = ok && gguf_fread_el (&kv->n_bytes, sizeof(kv->n_bytes), file, &offset);
-        ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
+        for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+            struct gguf_kv * kv = &ctx->header.kv[i];
 
-        //fprintf(stderr, "%s: reading kv with key %s\n", __func__, kv->key.data);
+            //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
+
+            ok = ok && gguf_fread_str(&kv->key,                          file, &offset);
+          //ok = ok && gguf_fread_el (&kv->n_bytes, sizeof(kv->n_bytes), file, &offset);
+            ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
+
+            //fprintf(stderr, "%s: reading kv with key %s\n", __func__, kv->key.data);
+
+            switch (kv->type) {
+                case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
+                case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
+                case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
+                case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
+                case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
+                case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
+                case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
+                case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
+                case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
+                case GGUF_TYPE_ARRAY:
+                                        GGML_ASSERT("gguf: array type not implemented");
+                                        break;
+            };
 
-        switch (kv->type) {
-            case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
-            case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
-            case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
-            case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
-            case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
-            case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
-            case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
-            case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
-            case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
-            case GGUF_TYPE_ARRAY:
-                                    GGML_ASSERT("gguf: array type not implemented");
-                                    break;
-        };
+            if (!ok) {
+                break;
+            }
+        }
 
         if (!ok) {
-            break;
+            fprintf(stderr, "%s: failed to read key-value pairs\n", __func__);
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
         }
     }
 
-    if (!ok) {
-        fprintf(stderr, "%s: failed to read key-value pairs\n", __func__);
-        fclose(file);
-        gguf_free(ctx);
-        return NULL;
-    }
-
-    ctx->infos = GGML_ALIGNED_MALLOC(ctx->header.n_tensors * sizeof(struct gguf_tensor_info));
+    // read the tensor infos
+    {
+        ctx->infos = GGML_ALIGNED_MALLOC(ctx->header.n_tensors * sizeof(struct gguf_tensor_info));
 
-    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
-        struct gguf_tensor_info * info = &ctx->infos[i];
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
 
-        for (int j = 0; j < GGML_MAX_DIMS; ++j) {
-            info->ne[j] = 1;
-        }
+            for (int j = 0; j < GGML_MAX_DIMS; ++j) {
+                info->ne[j] = 1;
+            }
 
-        ok = ok && gguf_fread_str(&info->name,                          file, &offset);
-        ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
-        for (uint32_t j = 0; j < info->n_dims; ++j) {
-            ok = ok && gguf_fread_el(&info->ne[j], sizeof(info->ne[j]), file, &offset);
-        }
-      //ok = ok && gguf_fread_el (&info->n_elms, sizeof(info->n_elms),  file, &offset);
-        ok = ok && gguf_fread_el (&info->type,    sizeof(info->type),   file, &offset);
-        ok = ok && gguf_fread_el (&info->offset,  sizeof(info->offset), file, &offset);
+            ok = ok && gguf_fread_str(&info->name,                          file, &offset);
+            ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
+            for (uint32_t j = 0; j < info->n_dims; ++j) {
+                ok = ok && gguf_fread_el(&info->ne[j], sizeof(info->ne[j]), file, &offset);
+            }
+          //ok = ok && gguf_fread_el (&info->n_elms, sizeof(info->n_elms),  file, &offset);
+            ok = ok && gguf_fread_el (&info->type,   sizeof(info->type),    file, &offset);
+            ok = ok && gguf_fread_el (&info->offset, sizeof(info->offset),  file, &offset);
 
-        if (!ok) {
-            fprintf(stderr, "%s: failed to read tensor info\n", __func__);
-            fclose(file);
-            gguf_free(ctx);
-            return NULL;
+            if (!ok) {
+                fprintf(stderr, "%s: failed to read tensor info\n", __func__);
+                fclose(file);
+                gguf_free(ctx);
+                return NULL;
+            }
         }
     }
 
@@ -18507,6 +18520,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     // TODO: determine new alignment from kv if available
 
+    // we require the data section to be aligned, so take into account any padding
     {
         const size_t offset_pad = offset % ctx->alignment;
 
@@ -18516,38 +18530,46 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         }
     }
 
+    // store the current file offset - this is where the data section starts
     ctx->offset = offset;
 
-    ctx->size_data = 0;
+    // compute the total size of the data section, taking into account the alignment
+    {
 
-    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
-        struct gguf_tensor_info * info = &ctx->infos[i];
+        ctx->size_data = 0;
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
 
-        const int64_t ne =
-            (int64_t) info->ne[0] *
-            (int64_t) info->ne[1] *
-            (int64_t) info->ne[2] *
-            (int64_t) info->ne[3];
+            const int64_t ne =
+                (int64_t) info->ne[0] *
+                (int64_t) info->ne[1] *
+                (int64_t) info->ne[2] *
+                (int64_t) info->ne[3];
 
-        if (ne % ggml_blck_size(info->type) != 0) {
-            fprintf(stderr, "%s: tensor '%s' number of elements (%" PRId64 ") is not a multiple of block size (%d)\n",
-                    __func__, info->name.data, ne, ggml_blck_size(info->type));
-            fclose(file);
-            gguf_free(ctx);
-            return NULL;
-        }
+            if (ne % ggml_blck_size(info->type) != 0) {
+                fprintf(stderr, "%s: tensor '%s' number of elements (%" PRId64 ") is not a multiple of block size (%d)\n",
+                        __func__, info->name.data, ne, ggml_blck_size(info->type));
+                fclose(file);
+                gguf_free(ctx);
+                return NULL;
+            }
 
-        const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
+            const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
 
-        ctx->size_data += GGML_PAD(size_cur, ctx->alignment);
+            ctx->size_data += GGML_PAD(size_cur, ctx->alignment);
+        }
     }
 
-    // load the tensor data
-    // TODO: simplify
+    // load the tensor data only if requested
     if (params.ctx != NULL) {
+        // if the provided gguf_context is no_alloc, then we create "empty" tensors and do not read the binary blob
+        // otherwise, we load the binary blob into the created ggml_context as well, and point the "data" members of
+        // the ggml_tensor structs to the appropriate locations in the binary blob
+
+        // compute the exact size needed for the new ggml_context
         const size_t mem_size =
             params.no_alloc ?
-            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() :
+            (ctx->header.n_tensors    )*ggml_tensor_overhead() :
             (ctx->header.n_tensors + 1)*ggml_tensor_overhead() + ctx->size_data;
 
         struct ggml_init_params pdata = {
@@ -18567,7 +18589,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
             ok = ok && data != NULL;
 
-            // read the tensor data
+            // read the binary blob with the tensor data
             ok = ok && gguf_fread_el(data->data, ctx->size_data, file, &offset);
 
             if (!ok) {
@@ -18602,6 +18624,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
                 break;
             }
 
+            // point the data member to the appropriate location in the binary blob using the tensor infos
             if (params.no_alloc == false) {
               //cur->data = (char *) data->data + ctx->infos[i].offset - ctx->offset; // offset from start of file
                 cur->data = (char *) data->data + ctx->infos[i].offset;               // offset from data
@@ -18609,7 +18632,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         }
 
         if (!ok) {
-            fprintf(stderr, "%s: failed to create tensors\n", __func__);
+            fprintf(stderr, "%s: failed to read the tensor data\n", __func__);
             fclose(file);
             ggml_free(ctx_data);
             gguf_free(ctx);
@@ -18619,13 +18642,6 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         ggml_set_no_alloc(ctx_data, params.no_alloc);
     }
 
-    if (!ok) {
-        fprintf(stderr, "%s: failed to read tensor data\n", __func__);
-        fclose(file);
-        gguf_free(ctx);
-        return NULL;
-    }
-
     return ctx;
 }
 

From c85d3178b3165137c1f7d0b454db2b2b5efd7b18 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 27 Jul 2023 10:29:29 +0300
Subject: [PATCH 014/242] refactor : reduce code duplication and better API
 (#2415)

---
 gguf.py | 57 +++++++++++++++++++++++++--------------------------------
 1 file changed, 25 insertions(+), 32 deletions(-)

diff --git a/gguf.py b/gguf.py
index dfd5ba5bf8490..991bbe2f32efc 100644
--- a/gguf.py
+++ b/gguf.py
@@ -71,63 +71,56 @@ def open(cls, path: str) -> "GGUFWriter":
         f = open(path, "wb")
         return cls(f)
 
-    def write_key(self, key: str, value_type: GGUFValueType):
-        encoded_key = key.encode("utf8")
-        self.buffered_writer.write(struct.pack("<I", len(encoded_key)))
-        self.buffered_writer.write(encoded_key)
-        self.buffered_writer.write(struct.pack("<I", value_type))
+    def write_key(self, key: str):
+        self.write_value(key, GGUFValueType.STRING)
 
     def write_uint8(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.UINT8)
-        self.buffered_writer.write(struct.pack("<B", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.UINT8)
 
     def write_int8(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.INT8)
-        self.buffered_writer.write(struct.pack("<b", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.INT8)
 
     def write_uint16(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.UINT16)
-        self.buffered_writer.write(struct.pack("<H", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.UINT16)
 
     def write_int16(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.INT16)
-        self.buffered_writer.write(struct.pack("<h", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.INT16)
 
     def write_uint32(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.UINT32)
-        self.buffered_writer.write(struct.pack("<I", value))
+        self.write_key(key)
+        self.write(value, GGUFValueType.UINT32)
 
     def write_int32(self, key: str, value: int):
-        self.write_key(key, GGUFValueType.INT32)
-        self.buffered_writer.write(struct.pack("<i", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.INT32)
 
     def write_float32(self, key: str, value: float):
-        self.write_key(key, GGUFValueType.FLOAT32)
-        self.buffered_writer.write(struct.pack("<f", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.FLOAT32)
 
     def write_bool(self, key: str, value: bool):
-        self.write_key(key, GGUFValueType.BOOL)
-        self.buffered_writer.write(struct.pack("<?", value))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.BOOL)
 
     def write_string(self, key: str, value: str):
-        self.write_key(key, GGUFValueType.STRING)
-        encoded_string = value.encode('utf-8')
-        self.buffered_writer.write(struct.pack("<I", len(encoded_string)))
-        self.buffered_writer.write(encoded_string)
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.STRING)
 
     def write_array(self, key: str, value: list):
         if not isinstance(value, list):
             raise ValueError("Value must be a list for array type")
 
-        self.write_key(key, GGUFValueType.ARRAY)
-
-        self.buffered_writer.write(struct.pack("<I", len(value)))
+        self.write_key(key)
+        self.write_value(value, GGUFValueType.ARRAY)
 
-        for item in value:
-            self.write_value(item)
+    def write_value(self: str, value: Any, value_type: GGUFValueType = None):
+        if value_type is None:
+            value_type = GGUFValueType.get_type(value)
 
-    def write_value(self: str, value: Any):
-        value_type = GGUFValueType.get_type(value)
         self.buffered_writer.write(struct.pack("<I", value_type))
 
         if value_type == GGUFValueType.UINT8:

From d89533dff60c758cc59401e5e260855c540008a3 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 27 Jul 2023 11:10:34 +0300
Subject: [PATCH 015/242] gguf : expose the gguf_type enum through the API for
 now

---
 examples/gguf/gguf.cpp | 13 -------------
 ggml.c                 | 13 -------------
 ggml.h                 | 14 ++++++++++++++
 3 files changed, 14 insertions(+), 26 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index a5c442ac581f5..d6a0691d016eb 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -7,19 +7,6 @@
 #include <fstream>
 #include <vector>
 
-enum gguf_type {
-    GGUF_TYPE_UINT8   = 0,
-    GGUF_TYPE_INT8    = 1,
-    GGUF_TYPE_UINT16  = 2,
-    GGUF_TYPE_INT16   = 3,
-    GGUF_TYPE_UINT32  = 4,
-    GGUF_TYPE_INT32   = 5,
-    GGUF_TYPE_FLOAT32 = 6,
-    GGUF_TYPE_BOOL    = 7,
-    GGUF_TYPE_STRING  = 8,
-    GGUF_TYPE_ARRAY   = 9,
-};
-
 template<typename T>
 static std::string to_string(const T & val) {
     std::stringstream ss;
diff --git a/ggml.c b/ggml.c
index b005fd8893e58..ebdb6536f81fa 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18297,19 +18297,6 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i
 
 ////////////////////////////////////////////////////////////////////////////////
 
-enum gguf_type {
-    GGUF_TYPE_UINT8   = 0,
-    GGUF_TYPE_INT8    = 1,
-    GGUF_TYPE_UINT16  = 2,
-    GGUF_TYPE_INT16   = 3,
-    GGUF_TYPE_UINT32  = 4,
-    GGUF_TYPE_INT32   = 5,
-    GGUF_TYPE_FLOAT32 = 6,
-    GGUF_TYPE_BOOL    = 7,
-    GGUF_TYPE_STRING  = 8,
-    GGUF_TYPE_ARRAY   = 9,
-};
-
 struct gguf_str {
     uint32_t n;
     char * data;
diff --git a/ggml.h b/ggml.h
index e0abbbfdd81e3..91588895ce70e 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1619,6 +1619,20 @@ extern "C" {
     // gguf
     //
 
+    // TODO: can be removed if the API is extended for writing
+    enum gguf_type {
+        GGUF_TYPE_UINT8   = 0,
+        GGUF_TYPE_INT8    = 1,
+        GGUF_TYPE_UINT16  = 2,
+        GGUF_TYPE_INT16   = 3,
+        GGUF_TYPE_UINT32  = 4,
+        GGUF_TYPE_INT32   = 5,
+        GGUF_TYPE_FLOAT32 = 6,
+        GGUF_TYPE_BOOL    = 7,
+        GGUF_TYPE_STRING  = 8,
+        GGUF_TYPE_ARRAY   = 9,
+    };
+
     struct gguf_context;
 
     struct gguf_init_params {

From d2b6ca13ad25a55e64bdd8287e773393fa54d212 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 27 Jul 2023 14:53:07 +0300
Subject: [PATCH 016/242] gguf : add array support

---
 examples/gguf/gguf.cpp | 87 +++++++++++++++++++++++++++++++++++-------
 ggml.c                 | 64 ++++++++++++++++++++++++++++---
 ggml.h                 |  4 +-
 3 files changed, 135 insertions(+), 20 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index d6a0691d016eb..c3494a343a6ec 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -29,7 +29,7 @@ void gguf_ex_write_u64(std::ofstream & fout, size_t val) {
 }
 
 template<typename T>
-void gguf_ex_write_param(std::ofstream & fout, const std::string & key, enum gguf_type type, const T & val) {
+void gguf_ex_write_val(std::ofstream & fout, const std::string & key, enum gguf_type type, const T & val) {
     gguf_ex_write_str(fout, key);
     fout.write((const char *) &type, sizeof(type));
     fout.write((const char *) &val,  sizeof(val));
@@ -38,13 +38,65 @@ void gguf_ex_write_param(std::ofstream & fout, const std::string & key, enum ggu
 }
 
 template<>
-void gguf_ex_write_param<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::string & val) {
+void gguf_ex_write_val<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::string & val) {
     gguf_ex_write_str(fout, key);
     fout.write((const char *) &type, sizeof(type));
 
     const int32_t n = val.size();
     fout.write((const char *) &n, sizeof(n));
     fout.write(val.c_str(), n);
+
+    fprintf(stdout, "%s: write param: %s = %s\n", __func__, key.c_str(), val.c_str());
+}
+
+template<typename T>
+void gguf_ex_write_arr(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::vector<T> & val) {
+    gguf_ex_write_str(fout, key);
+    {
+        const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+        fout.write((const char *) &tarr, sizeof(tarr));
+    }
+
+    const int32_t n = val.size();
+    fout.write((const char *) &type, sizeof(type));
+    fout.write((const char *) &n,    sizeof(n));
+    fout.write((const char *) val.data(), n * sizeof(T));
+
+    fprintf(stdout, "%s: write param: %s = [", __func__, key.c_str());
+    for (int i = 0; i < n; ++i) {
+        fprintf(stdout, "%s", to_string(val[i]).c_str());
+        if (i < n - 1) {
+            fprintf(stdout, ", ");
+        }
+    }
+    fprintf(stdout, "]\n");
+}
+
+template<>
+void gguf_ex_write_arr<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
+    gguf_ex_write_str(fout, key);
+    {
+        const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+        fout.write((const char *) &tarr, sizeof(tarr));
+    }
+
+    const int32_t n = val.size();
+    fout.write((const char *) &type, sizeof(type));
+    fout.write((const char *) &n,    sizeof(n));
+    for (int i = 0; i < n; ++i) {
+        const int32_t nstr = val[i].size();
+        fout.write((const char *) &nstr, sizeof(nstr));
+        fout.write(val[i].c_str(), nstr);
+    }
+
+    fprintf(stdout, "%s: write param: %s = [", __func__, key.c_str());
+    for (int i = 0; i < n; ++i) {
+        fprintf(stdout, "%s", val[i].c_str());
+        if (i < n - 1) {
+            fprintf(stdout, ", ");
+        }
+    }
+    fprintf(stdout, "]\n");
 }
 
 bool gguf_ex_write(const std::string & fname) {
@@ -60,8 +112,9 @@ bool gguf_ex_write(const std::string & fname) {
         fout.write((const char *) &version, sizeof(version));
     }
 
+    // NOTE: these have to match the output below!
     const int n_tensors = 10;
-    const int n_kv = 9;
+    const int n_kv      = 12;
 
     fout.write((const char*) &n_tensors, sizeof(n_tensors));
     fout.write((const char*) &n_kv, sizeof(n_kv));
@@ -70,17 +123,21 @@ bool gguf_ex_write(const std::string & fname) {
 
     // kv data
     {
-        gguf_ex_write_param< uint8_t>(fout, "some.parameter.uint8",   GGUF_TYPE_UINT8,   0x12);
-        gguf_ex_write_param<  int8_t>(fout, "some.parameter.int8",    GGUF_TYPE_INT8,   -0x13);
-        gguf_ex_write_param<uint16_t>(fout, "some.parameter.uint16",  GGUF_TYPE_UINT16,  0x1234);
-        gguf_ex_write_param< int16_t>(fout, "some.parameter.int16",   GGUF_TYPE_INT16,  -0x1235);
-        gguf_ex_write_param<uint32_t>(fout, "some.parameter.uint32",  GGUF_TYPE_UINT32,  0x12345678);
-        gguf_ex_write_param< int32_t>(fout, "some.parameter.int32",   GGUF_TYPE_INT32,  -0x12345679);
+        gguf_ex_write_val< uint8_t>(fout, "some.parameter.uint8",   GGUF_TYPE_UINT8,   0x12);
+        gguf_ex_write_val<  int8_t>(fout, "some.parameter.int8",    GGUF_TYPE_INT8,   -0x13);
+        gguf_ex_write_val<uint16_t>(fout, "some.parameter.uint16",  GGUF_TYPE_UINT16,  0x1234);
+        gguf_ex_write_val< int16_t>(fout, "some.parameter.int16",   GGUF_TYPE_INT16,  -0x1235);
+        gguf_ex_write_val<uint32_t>(fout, "some.parameter.uint32",  GGUF_TYPE_UINT32,  0x12345678);
+        gguf_ex_write_val< int32_t>(fout, "some.parameter.int32",   GGUF_TYPE_INT32,  -0x12345679);
 
-        gguf_ex_write_param<float>   (fout, "some.parameter.float32", GGUF_TYPE_FLOAT32, 0.123456789f);
-        gguf_ex_write_param<bool>    (fout, "some.parameter.bool",    GGUF_TYPE_BOOL,    true);
+        gguf_ex_write_val<float>   (fout, "some.parameter.float32", GGUF_TYPE_FLOAT32, 0.123456789f);
+        gguf_ex_write_val<bool>    (fout, "some.parameter.bool",    GGUF_TYPE_BOOL,    true);
 
-        gguf_ex_write_param<std::string>(fout, "some.parameter.string",  GGUF_TYPE_STRING,  "hello world");
+        gguf_ex_write_val<std::string>(fout, "some.parameter.string",  GGUF_TYPE_STRING,  "hello world");
+
+        gguf_ex_write_arr<int16_t>    (fout, "some.parameter.arr.i16", GGUF_TYPE_INT16,   { 1, 2, 3, 4, });
+        gguf_ex_write_arr<float>      (fout, "some.parameter.arr.f32", GGUF_TYPE_FLOAT32, { 3.145f, 2.718f, 1.414f, });
+        gguf_ex_write_arr<std::string>(fout, "some.parameter.arr.str", GGUF_TYPE_STRING,  { "hello", "world", "!" });
     }
 
     uint64_t offset_tensor = 0;
@@ -203,13 +260,15 @@ bool gguf_ex_read_0(const std::string & fname) {
         fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
 
         for (int i = 0; i < n_tensors; ++i) {
-            const char * name = gguf_get_tensor_name(ctx, i);
+            const char * name   = gguf_get_tensor_name  (ctx, i);
             const size_t offset = gguf_get_tensor_offset(ctx, i);
 
             fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
         }
     }
 
+    gguf_free(ctx);
+
     return true;
 }
 
@@ -248,7 +307,7 @@ bool gguf_ex_read_1(const std::string & fname) {
         fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
 
         for (int i = 0; i < n_tensors; ++i) {
-            const char * name = gguf_get_tensor_name(ctx, i);
+            const char * name   = gguf_get_tensor_name  (ctx, i);
             const size_t offset = gguf_get_tensor_offset(ctx, i);
 
             fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
diff --git a/ggml.c b/ggml.c
index ebdb6536f81fa..96c7ebd347c1b 100644
--- a/ggml.c
+++ b/ggml.c
@@ -3698,7 +3698,6 @@ static const size_t GGML_TYPE_SIZE[GGML_TYPE_COUNT] = {
 };
 static_assert(GGML_TYPE_COUNT == 19, "GGML_TYPE_SIZE is outdated");
 
-
 static const char * GGML_TYPE_NAME[GGML_TYPE_COUNT] = {
     [GGML_TYPE_F32]  = "f32",
     [GGML_TYPE_F16]  = "f16",
@@ -18302,7 +18301,19 @@ struct gguf_str {
     char * data;
 };
 
-union gguf_value;
+static const size_t GGUF_TYPE_SIZE[GGUF_TYPE_COUNT] = {
+    [GGUF_TYPE_UINT8]   = sizeof(uint8_t),
+    [GGUF_TYPE_INT8]    = sizeof(int8_t),
+    [GGUF_TYPE_UINT16]  = sizeof(uint16_t),
+    [GGUF_TYPE_INT16]   = sizeof(int16_t),
+    [GGUF_TYPE_UINT32]  = sizeof(uint32_t),
+    [GGUF_TYPE_INT32]   = sizeof(int32_t),
+    [GGUF_TYPE_FLOAT32] = sizeof(float),
+    [GGUF_TYPE_BOOL]    = sizeof(bool),
+    [GGUF_TYPE_STRING]  = sizeof(struct gguf_str),
+    [GGUF_TYPE_ARRAY]   = 0, // undefined
+};
+static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
 
 union gguf_value {
     uint8_t  uint8;
@@ -18320,7 +18331,7 @@ union gguf_value {
         enum gguf_type type;
 
         uint32_t n;
-        union gguf_value * arr;
+        void * data;
     } arr;
 };
 
@@ -18457,8 +18468,35 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
                 case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
                 case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
                 case GGUF_TYPE_ARRAY:
-                                        GGML_ASSERT("gguf: array type not implemented");
-                                        break;
+                    {
+                        ok = ok && gguf_fread_el(&kv->value.arr.type, sizeof(kv->value.arr.type), file, &offset);
+                        ok = ok && gguf_fread_el(&kv->value.arr.n,    sizeof(kv->value.arr.n),    file, &offset);
+
+                        switch (kv->value.arr.type) {
+                            case GGUF_TYPE_UINT8:
+                            case GGUF_TYPE_INT8:
+                            case GGUF_TYPE_UINT16:
+                            case GGUF_TYPE_INT16:
+                            case GGUF_TYPE_UINT32:
+                            case GGUF_TYPE_INT32:
+                            case GGUF_TYPE_FLOAT32:
+                            case GGUF_TYPE_BOOL:
+                                {
+                                    kv->value.arr.data = malloc(kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type]);
+                                    ok = ok && gguf_fread_el(kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type], file, &offset);
+                                } break;
+                            case GGUF_TYPE_STRING:
+                                {
+                                    kv->value.arr.data = malloc(kv->value.arr.n * sizeof(struct gguf_str));
+                                    for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                                        ok = ok && gguf_fread_str(&((struct gguf_str *) kv->value.arr.data)[j], file, &offset);
+                                    }
+                                } break;
+                            case GGUF_TYPE_ARRAY:
+                            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
+                        };
+                    } break;
+                case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
             };
 
             if (!ok) {
@@ -18629,6 +18667,8 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         ggml_set_no_alloc(ctx_data, params.no_alloc);
     }
 
+    fclose(file);
+
     return ctx;
 }
 
@@ -18651,6 +18691,20 @@ void gguf_free(struct gguf_context * ctx) {
                     free(kv->value.str.data);
                 }
             }
+
+            if (kv->type == GGUF_TYPE_ARRAY) {
+                if (kv->value.arr.data) {
+                    if (kv->value.arr.type == GGUF_TYPE_STRING) {
+                        for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                            struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[j];
+                            if (str->data) {
+                                free(str->data);
+                            }
+                        }
+                    }
+                    free(kv->value.arr.data);
+                }
+            }
         }
 
         GGML_ALIGNED_FREE(ctx->header.kv);
diff --git a/ggml.h b/ggml.h
index 91588895ce70e..e857b3f14435f 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1631,6 +1631,7 @@ extern "C" {
         GGUF_TYPE_BOOL    = 7,
         GGUF_TYPE_STRING  = 8,
         GGUF_TYPE_ARRAY   = 9,
+        GGUF_TYPE_COUNT,       // marks the end of the enum
     };
 
     struct gguf_context;
@@ -1664,7 +1665,8 @@ extern "C" {
     GGML_API float        gguf_get_val_f32 (struct gguf_context * ctx, int i);
     GGML_API bool         gguf_get_val_bool(struct gguf_context * ctx, int i);
     GGML_API const char * gguf_get_val_str (struct gguf_context * ctx, int i);
-    // TODO: arr
+    GGML_API int          gguf_get_arr_n   (struct gguf_context * ctx, int i);
+    GGML_API void         gguf_get_arr_data(struct gguf_context * ctx, int i, void * data);
 
     GGML_API int    gguf_get_n_tensors    (struct gguf_context * ctx);
     GGML_API size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i);

From 158be8f7f4f0f34ef8057a87f0e35a3b2af2ed4d Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 27 Jul 2023 15:37:06 +0300
Subject: [PATCH 017/242] gguf.py : some code style changes

---
 constants.py |  44 ++++++++--------
 gguf.py      | 142 +++++++++++++++++++++++++--------------------------
 2 files changed, 92 insertions(+), 94 deletions(-)

diff --git a/constants.py b/constants.py
index 7c7456403aaeb..3a97460e5dc5e 100644
--- a/constants.py
+++ b/constants.py
@@ -1,32 +1,32 @@
-GGUF_MAGIC = 0x47475546
+GGUF_MAGIC   = 0x47475546
 GGUF_VERSION = 1
 
 # general
-KEY_GENERAL_ARCHITECTURE = "general.architecture"
+KEY_GENERAL_ARCHITECTURE         = "general.architecture"
 KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
-KEY_GENERAL_NAME = "general.name"
-KEY_GENERAL_AUTHOR = "general.author"
-KEY_GENERAL_URL = "general.url"
-KEY_GENERAL_DESCRIPTION = "general.description"
-KEY_GENERAL_FILE_TYPE = "general.file_type"
-KEY_GENERAL_LICENSE = "general.license"
-KEY_GENERAL_SOURCE_URL = "general.source.url"
-KEY_GENERAL_SOURCE_HF_REPO = "general.source.hugginface.repository"
+KEY_GENERAL_NAME                 = "general.name"
+KEY_GENERAL_AUTHOR               = "general.author"
+KEY_GENERAL_URL                  = "general.url"
+KEY_GENERAL_DESCRIPTION          = "general.description"
+KEY_GENERAL_FILE_TYPE            = "general.file_type"
+KEY_GENERAL_LICENSE              = "general.license"
+KEY_GENERAL_SOURCE_URL           = "general.source.url"
+KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
 
 # LLM
-KEY_LLM_CONTEXT_LENGTH = "{llm}.context_length"
-KEY_LLM_EMBEDDING_LENGTH = "{llm}.embedding_length"
-KEY_LLM_LAYER_COUNT = "{llm}.layer_count"
-KEY_LLM_FEED_FORWARD_LENGTH = "{llm}.feed_forward_length"
-KEY_LLM_USE_PARALLEL_RESIDUAL = "{llm}.use_parallel_residual"
-KEY_LLM_TENSOR_DATA_LAYOUT = "{llm}.tensor_data_layout"
+KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
+KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
+KEY_LLM_LAYER_COUNT              = "{llm}.layer_count"
+KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
 
 # attention
-KEY_ATTENTION_HEAD_COUNT = "{llm}.attention.head_count"
-KEY_ATTENTION_HEAD_COUNT_KV = "{llm}.attention.head_count_kv"
-KEY_ATTENTION_MAX_ALIBI_BIAS = "{llm}.attention.max_alibi_bias"
-KEY_ATTENTION_CLAMP_KQV = "{llm}.attention.clamp_kqv"
+KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
 
 # RoPE
-KEY_ROPE_DIMENSION_COUNT = "{llm}.rope.dimension_count"
-KEY_ROPE_SCALE = "{llm}.rope.scale"
+KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
+KEY_ROPE_SCALE                   = "{llm}.rope.scale"
diff --git a/gguf.py b/gguf.py
index 991bbe2f32efc..764ae9a9df27c 100644
--- a/gguf.py
+++ b/gguf.py
@@ -6,14 +6,13 @@
 """
 
 import struct
+import constants
 from enum import IntEnum
 from typing import List, Any
-import constants
-
 
 class GGMLQuantizationType(IntEnum):
-    F32 = 0
-    F16 = 1
+    F32  = 0
+    F16  = 1
     QR_0 = 2
     Q4_1 = 3
     # Q4_2 = 4 # support has been removed
@@ -31,31 +30,30 @@ class GGMLQuantizationType(IntEnum):
 
 
 class GGUFValueType(IntEnum):
-    UINT8 = 0
-    INT8 = 1
-    UINT16 = 2
-    INT16 = 3
-    UINT32 = 4
-    INT32 = 5
+    UINT8   = 0
+    INT8    = 1
+    UINT16  = 2
+    INT16   = 3
+    UINT32  = 4
+    INT32   = 5
     FLOAT32 = 6
-    BOOL = 7
-    STRING = 8
-    ARRAY = 9
+    BOOL    = 7
+    STRING  = 8
+    ARRAY   = 9
 
     @staticmethod
-    def get_type(value):
-        if isinstance(value, str):
+    def get_type(val):
+        if isinstance(val, str):
             return GGUFValueType.STRING
-        elif isinstance(value, list):
+        elif isinstance(val, list):
             return GGUFValueType.ARRAY
-        elif isinstance(value, float):
+        elif isinstance(val, float):
             return GGUFValueType.FLOAT32
-        elif isinstance(value, bool):
+        elif isinstance(val, bool):
             return GGUFValueType.BOOL
         else:
             return GGUFValueType.INT32
 
-
 class GGUFWriter:
     def __init__(self, buffered_writer):
         self.buffered_writer = buffered_writer
@@ -72,81 +70,81 @@ def open(cls, path: str) -> "GGUFWriter":
         return cls(f)
 
     def write_key(self, key: str):
-        self.write_value(key, GGUFValueType.STRING)
+        self.write_val(key, GGUFValueType.STRING)
 
-    def write_uint8(self, key: str, value: int):
+    def write_uint8(self, key: str, val: int):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.UINT8)
+        self.write_val(val, GGUFValueType.UINT8)
 
-    def write_int8(self, key: str, value: int):
+    def write_int8(self, key: str, val: int):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.INT8)
+        self.write_val(val, GGUFValueType.INT8)
 
-    def write_uint16(self, key: str, value: int):
+    def write_uint16(self, key: str, val: int):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.UINT16)
+        self.write_val(val, GGUFValueType.UINT16)
 
-    def write_int16(self, key: str, value: int):
+    def write_int16(self, key: str, val: int):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.INT16)
+        self.write_val(val, GGUFValueType.INT16)
 
-    def write_uint32(self, key: str, value: int):
+    def write_uint32(self, key: str, val: int):
         self.write_key(key)
-        self.write(value, GGUFValueType.UINT32)
+        self.write_val(val, GGUFValueType.UINT32)
 
-    def write_int32(self, key: str, value: int):
+    def write_int32(self, key: str, val: int):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.INT32)
+        self.write_val(val, GGUFValueType.INT32)
 
-    def write_float32(self, key: str, value: float):
+    def write_float32(self, key: str, val: float):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.FLOAT32)
+        self.write_val(val, GGUFValueType.FLOAT32)
 
-    def write_bool(self, key: str, value: bool):
+    def write_bool(self, key: str, val: bool):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.BOOL)
+        self.write_val(val, GGUFValueType.BOOL)
 
-    def write_string(self, key: str, value: str):
+    def write_string(self, key: str, val: str):
         self.write_key(key)
-        self.write_value(value, GGUFValueType.STRING)
+        self.write_val(val, GGUFValueType.STRING)
 
-    def write_array(self, key: str, value: list):
-        if not isinstance(value, list):
+    def write_array(self, key: str, val: list):
+        if not isinstance(val, list):
             raise ValueError("Value must be a list for array type")
 
         self.write_key(key)
-        self.write_value(value, GGUFValueType.ARRAY)
-
-    def write_value(self: str, value: Any, value_type: GGUFValueType = None):
-        if value_type is None:
-            value_type = GGUFValueType.get_type(value)
-
-        self.buffered_writer.write(struct.pack("<I", value_type))
-
-        if value_type == GGUFValueType.UINT8:
-            self.buffered_writer.write(struct.pack("<B", value))
-        elif value_type == GGUFValueType.INT8:
-            self.buffered_writer.write(struct.pack("<b", value))
-        elif value_type == GGUFValueType.UINT16:
-            self.buffered_writer.write(struct.pack("<H", value))
-        elif value_type == GGUFValueType.INT16:
-            self.buffered_writer.write(struct.pack("<h", value))
-        elif value_type == GGUFValueType.UINT32:
-            self.buffered_writer.write(struct.pack("<I", value))
-        elif value_type == GGUFValueType.INT32:
-            self.buffered_writer.write(struct.pack("<i", value))
-        elif value_type == GGUFValueType.FLOAT32:
-            self.buffered_writer.write(struct.pack("<f", value))
-        elif value_type == GGUFValueType.BOOL:
-            self.buffered_writer.write(struct.pack("?", value))
-        elif value_type == GGUFValueType.STRING:
-            encoded_value = value.encode("utf8")
-            self.buffered_writer.write(struct.pack("<I", len(encoded_value)))
-            self.buffered_writer.write(encoded_value)
-        elif value_type == GGUFValueType.ARRAY:
-            self.buffered_writer.write(struct.pack("<I", len(value)))
-            for item in value:
-                self.write_value(item)
+        self.write_val(val, GGUFValueType.ARRAY)
+
+    def write_val(self: str, val: Any, vtype: GGUFValueType = None):
+        if vtype is None:
+            vtype = GGUFValueType.get_type(val)
+
+        self.buffered_writer.write(struct.pack("<I", vtype))
+
+        if vtype == GGUFValueType.UINT8:
+            self.buffered_writer.write(struct.pack("<B", val))
+        elif vtype == GGUFValueType.INT8:
+            self.buffered_writer.write(struct.pack("<b", val))
+        elif vtype == GGUFValueType.UINT16:
+            self.buffered_writer.write(struct.pack("<H", val))
+        elif vtype == GGUFValueType.INT16:
+            self.buffered_writer.write(struct.pack("<h", val))
+        elif vtype == GGUFValueType.UINT32:
+            self.buffered_writer.write(struct.pack("<I", val))
+        elif vtype == GGUFValueType.INT32:
+            self.buffered_writer.write(struct.pack("<i", val))
+        elif vtype == GGUFValueType.FLOAT32:
+            self.buffered_writer.write(struct.pack("<f", val))
+        elif vtype == GGUFValueType.BOOL:
+            self.buffered_writer.write(struct.pack("?", val))
+        elif vtype == GGUFValueType.STRING:
+            encoded_val = val.encode("utf8")
+            self.buffered_writer.write(struct.pack("<I", len(encoded_val)))
+            self.buffered_writer.write(encoded_val)
+        elif vtype == GGUFValueType.ARRAY:
+            self.buffered_writer.write(struct.pack("<I", len(val)))
+            for item in val:
+                self.write_val(item)
         else:
             raise ValueError("Invalid GGUF metadata value type")
 

From 68f53485e4787262489320b22c786889b220f9cf Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 27 Jul 2023 15:56:53 +0300
Subject: [PATCH 018/242] convert.py : start a new simplified implementation by
 removing old stuff

---
 convert-new.py | 961 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 961 insertions(+)
 create mode 100755 convert-new.py

diff --git a/convert-new.py b/convert-new.py
new file mode 100755
index 0000000000000..dacad693fea1c
--- /dev/null
+++ b/convert-new.py
@@ -0,0 +1,961 @@
+#!/usr/bin/env python
+
+import argparse
+import concurrent.futures
+import copy
+import enum
+import faulthandler
+import functools
+import io
+import itertools
+import json
+import math
+import mmap
+import pickle
+import re
+import signal
+import struct
+import sys
+import zipfile
+import numpy as np
+
+from abc import ABCMeta, abstractmethod
+from dataclasses import dataclass
+from pathlib import Path
+from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List, Literal, Optional, Sequence, Tuple, TypeVar, Union)
+from sentencepiece import SentencePieceProcessor  # type: ignore
+
+if TYPE_CHECKING:
+    from typing_extensions import TypeAlias
+
+if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
+    faulthandler.register(signal.SIGUSR1)
+
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+
+@dataclass(frozen=True)
+class UnquantizedDataType:
+    name: str
+
+DT_F16  = UnquantizedDataType('F16')
+DT_F32  = UnquantizedDataType('F32')
+DT_I32  = UnquantizedDataType('I32')
+DT_BF16 = UnquantizedDataType('BF16')
+
+DataType = Union[UnquantizedDataType]
+
+DATA_TYPE_TO_FTYPE: Dict[DataType, int] = {
+    DT_F32: 0,
+    DT_F16: 1,
+}
+
+FTYPE_TO_DATA_TYPE: Dict[int, DataType] = \
+    {ftype: dtype for (dtype, ftype) in DATA_TYPE_TO_FTYPE.items()}
+
+DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
+    DT_BF16: np.dtype(np.uint16),
+    DT_F16:  np.dtype(np.float16),
+    DT_F32:  np.dtype(np.float32),
+    DT_I32:  np.dtype(np.int32),
+}
+
+NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
+    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
+
+class GGMLFileType(enum.Enum):
+    AllF32    = 0
+    MostlyF16 = 1  # except 1d tensors
+
+    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
+        if len(tensor.shape) == 1:
+            # 1D tensors are always F32.
+            return DT_F32
+        elif self == GGMLFileType.AllF32:
+            return DT_F32
+        elif self == GGMLFileType.MostlyF16:
+            return DT_F16
+        else:
+            raise ValueError(self)
+
+# TODO: this is LLaMA specific
+def make_tensors_list() -> List[str]:
+    ret = [
+        'tok_embeddings.weight',
+        'norm.weight',
+        'output.weight',
+    ]
+    for i in range(80):  # maximum number of layer
+        ret += [
+            f'layers.{i}.attention.wq.weight',
+            f'layers.{i}.attention.wk.weight',
+            f'layers.{i}.attention.wv.weight',
+            f'layers.{i}.attention.wo.weight',
+            f'layers.{i}.attention_norm.weight',
+            f'layers.{i}.feed_forward.w1.weight',
+            f'layers.{i}.feed_forward.w2.weight',
+            f'layers.{i}.feed_forward.w3.weight',
+            f'layers.{i}.ffn_norm.weight',
+        ]
+    return ret
+
+# TODO: this should be generalized for non-LLaMA models
+TENSORS_LIST = make_tensors_list()
+TENSORS_SET = set(TENSORS_LIST)
+
+def find_n_mult(n_ff: int, n_embd: int) -> int:
+    # hardcoded magic range
+    for n_mult in range(256, 1, -1):
+        calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
+        if calc_ff == n_ff:
+            return n_mult
+    raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
+
+
+@dataclass
+class Params:
+    n_vocab: int
+    n_embd:  int
+    n_mult:  int
+    n_head:  int
+    n_layer: int
+
+    @staticmethod
+    def guessed(model: 'LazyModel') -> 'Params':
+        # try transformer naming first
+        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
+
+        # try transformer naming first
+        if "model.layers.0.self_attn.q_proj.weight" in model:
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
+        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
+        else:
+            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
+
+        if n_layer < 1:
+            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
+
+        n_head=n_embd // 128 # guessed
+
+        return Params(
+            n_vocab = n_vocab,
+            n_embd  = n_embd,
+            n_mult  = 256,
+            n_head  = n_head,
+            n_layer = n_layer,
+        )
+
+    @staticmethod
+    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab = config["vocab_size"];
+        n_embd  = config["hidden_size"];
+        n_head  = config["num_attention_heads"];
+        n_layer = config["num_hidden_layers"];
+        n_ff    = config["intermediate_size"];
+
+        n_mult = find_n_mult(n_ff, n_embd);
+
+        return Params(
+            n_vocab = n_vocab,
+            n_embd  = n_embd,
+            n_mult  = n_mult,
+            n_head  = n_head,
+            n_layer = n_layer,
+        )
+
+    # LLaMA v2 70B params.json
+    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
+    @staticmethod
+    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab = config["vocab_size"];
+        n_embd  = config["dim"];
+        n_head  = config["n_heads"];
+        n_layer = config["n_layers"];
+        n_mult  = config["multiple_of"];
+
+        if n_vocab == -1:
+            n_vocab = model["tok_embeddings.weight"].shape[0]
+
+        return Params(
+            n_vocab = n_vocab,
+            n_embd  = n_embd,
+            n_mult  = n_mult,
+            n_head  = n_head,
+            n_layer = n_layer,
+        )
+
+    @staticmethod
+    def load(model_plus: 'ModelPlus') -> 'Params':
+        hf_config_path   = model_plus.paths[0].parent / "config.json"
+        orig_config_path = model_plus.paths[0].parent / "params.json"
+
+        if hf_config_path.exists():
+            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
+        elif orig_config_path.exists():
+            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
+        else:
+            params = Params.guessed(model_plus.model)
+
+        print(f'params: n_vocab:{params.n_vocab} n_embd:{params.n_embd} n_mult:{params.n_mult} n_head:{params.n_head} n_layer:{params.n_layer}')
+        return params
+
+
+class SentencePieceVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vocabtype: Optional[str]) -> None:
+        self.vocabtype = vocabtype
+        if self.vocabtype == "bpe":
+            self.sentencepiece_tokenizer = json.loads(open(str(fname_tokenizer)).read())
+        else:
+            self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens))
+        else:
+            added_tokens = {}
+
+        if self.vocabtype == "bpe":
+            vocab_size: int = len(self.sentencepiece_tokenizer)
+        else:
+            vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
+
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids   = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
+
+    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.sentencepiece_tokenizer
+        if self.vocabtype == "bpe":
+            from transformers.models.gpt2 import tokenization_gpt2
+            byte_encoder = tokenization_gpt2.bytes_to_unicode()
+            byte_decoder = {v: k for k, v in byte_encoder.items()}
+            for i, item in enumerate(tokenizer):
+                text: bytes
+                text = b''.join([x.to_bytes(1, byteorder='big') for x in [byte_decoder[y] for y in item]])
+                score: float = -i
+                yield text, score
+        else:
+            for i in range(tokenizer.vocab_size()):
+                text: bytes
+                if tokenizer.is_unknown(i):
+                    text = " \u2047 ".encode("utf-8")
+                elif tokenizer.is_control(i):
+                    text = b""
+                elif tokenizer.is_byte(i):
+                    piece = tokenizer.id_to_piece(i)
+                    if len(piece) != 6:
+                        raise Exception(f"Invalid token: {piece}")
+                    byte_value = int(piece[3:-1], 16)
+                    text = struct.pack("B", byte_value)
+                else:
+                    text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
+                score: float = tokenizer.get_score(i)
+                yield text, score
+
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.sentencepiece_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+
+class GGMLVocab:
+    def __init__(self, tokens: List[Tuple[bytes, float]]):
+        self.tokens = tokens
+        self.vocab_size = len(tokens)
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        return self.tokens
+
+    def __repr__(self) -> str:
+        return f"<GGMLVocab with {self.vocab_size} tokens>"
+
+
+Vocab = Union[SentencePieceVocab, GGMLVocab]
+
+
+def permute(weights: NDArray, n_head: int) -> NDArray:
+    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
+                   .swapaxes(1, 2)
+                   .reshape(weights.shape))
+
+
+class Tensor(metaclass=ABCMeta):
+    data_type: DataType
+
+    @abstractmethod
+    def astype(self, data_type: DataType) -> 'Tensor': ...
+    @abstractmethod
+    def permute(self, n_head: int) -> 'Tensor': ...
+    @abstractmethod
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def part(self, n_part: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
+
+
+def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
+    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
+    fp32_arr = bf16_arr.astype(np.uint32) << 16
+    return fp32_arr.view(np.float32)
+
+
+class UnquantizedTensor(Tensor):
+    def __init__(self, ndarray: NDArray) -> None:
+        assert isinstance(ndarray, np.ndarray)
+        self.ndarray = ndarray
+        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
+
+    def astype(self, data_type: DataType) -> Tensor:
+        dtype = DATA_TYPE_TO_NUMPY[data_type]
+        if self.data_type == DT_BF16:
+            self.ndarray = bf16_to_fp32(self.ndarray)
+        return UnquantizedTensor(self.ndarray.astype(dtype))
+
+    def to_ggml(self) -> 'UnquantizedTensor':
+        return self
+
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
+
+    def part(self, n_part: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
+
+    def permute(self, n_head: int) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head))
+
+
+def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
+    tensor = lazy_tensor.load()
+    assert isinstance(tensor, UnquantizedTensor)
+
+    # double-check:
+    actual_shape = list(tensor.ndarray.shape)
+    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
+    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
+        if convert:
+            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
+        else:
+            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
+
+    return tensor.ndarray
+
+
+GGMLCompatibleTensor = Union[UnquantizedTensor]
+
+
+class DeferredPermutedTensor(Tensor):
+    def __init__(self, base: Tensor, n_head: int) -> None:
+        self.base = base
+        self.n_head = n_head
+        self.data_type = self.base.data_type
+
+    def astype(self, data_type: DataType) -> Tensor:
+        return self.base.astype(data_type).permute(self.n_head)
+
+    def to_ggml(self) -> GGMLCompatibleTensor:
+        return self.base.to_ggml().permute(self.n_head)
+
+    def permute(self, n_head: int) -> Tensor:
+        raise Exception("shouldn't permute twice")
+
+
+@dataclass
+class LazyTensor:
+    _load: Callable[[], Tensor]
+    shape: List[int]
+    data_type: DataType
+    description: str
+
+    def load(self) -> Tensor:
+        ret = self._load()
+        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
+        return ret
+
+    def astype(self, data_type: DataType) -> 'LazyTensor':
+        self.validate_conversion_to(data_type)
+
+        def load() -> Tensor:
+            return self.load().astype(data_type)
+        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
+
+    def validate_conversion_to(self, data_type: DataType) -> None:
+        if data_type == self.data_type:
+            return
+
+
+LazyModel = Dict[str, LazyTensor]
+
+
+@dataclass
+class ModelPlus:
+    model: LazyModel
+    paths: List[Path]  # Where this was read from.
+    format: Literal['ggml', 'torch', 'safetensors']
+    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
+
+
+def merge_sharded(models: List[LazyModel]) -> LazyModel:
+    # Original LLaMA models have each file contain one part of each tensor.
+    # Use a dict instead of a set to preserve order.
+    names = {name: None for model in models for name in model}
+
+    def convert(name: str) -> LazyTensor:
+        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
+        if len(lazy_tensors) == 1:
+            # only one file; don't go through this procedure since there might
+            # be quantized tensors
+            return lazy_tensors[0]
+        if len(lazy_tensors[0].shape) == 1:
+            # the tensor is just duplicated in every file
+            return lazy_tensors[0]
+        if name.startswith('tok_embeddings.') or \
+           name.endswith('.attention.wo.weight') or \
+           name.endswith('.feed_forward.w2.weight'):
+            # split by columns
+            axis = 1
+        else:
+            # split by rows
+            axis = 0
+        concatenated_shape = list(lazy_tensors[0].shape)
+        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
+
+        def load() -> UnquantizedTensor:
+            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
+            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
+            return UnquantizedTensor(concatenated)
+        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
+        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
+    return {name: convert(name) for name in names}
+
+
+def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
+    formats = set(mp.format for mp in models_plus)
+    assert len(formats) == 1, "different formats?"
+    format = formats.pop()
+    paths = [path for mp in models_plus for path in mp.paths]
+    # Use the first non-None vocab, if any.
+    try:
+        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
+    except StopIteration:
+        vocab = None
+
+    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
+        # Transformers models put different tensors in different files, but
+        # don't split indivdual tensors between files.
+        model: LazyModel = {}
+        for mp in models_plus:
+            model.update(mp.model)
+    else:
+        model = merge_sharded([mp.model for mp in models_plus])
+
+    return ModelPlus(model, paths, format, vocab)
+
+
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute(n_head)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
+
+def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute_part(n_part, n_head)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
+
+def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().part(n_part)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
+
+def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
+    out: LazyModel = {}
+    out["tok_embeddings.weight"] = model["model.embed_tokens.weight"]
+    out["norm.weight"]           = model["model.norm.weight"]
+    out["output.weight"]         = model["lm_head.weight"]
+
+    for i in itertools.count():
+        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
+            out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
+            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head)
+            out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
+        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
+            out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)
+            out[f"layers.{i}.attention.wk.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head)
+            out[f"layers.{i}.attention.wv.weight"] = part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
+        else:
+            break
+
+        out[f"layers.{i}.attention.wo.weight"] = model[f"model.layers.{i}.self_attn.o_proj.weight"]
+
+        out[f"layers.{i}.feed_forward.w1.weight"] = model[f"model.layers.{i}.mlp.gate_proj.weight"]
+        out[f"layers.{i}.feed_forward.w2.weight"] = model[f"model.layers.{i}.mlp.down_proj.weight"]
+        out[f"layers.{i}.feed_forward.w3.weight"] = model[f"model.layers.{i}.mlp.up_proj.weight"]
+
+        out[f"layers.{i}.attention_norm.weight"] = model[f"model.layers.{i}.input_layernorm.weight"]
+        out[f"layers.{i}.ffn_norm.weight"]       = model[f"model.layers.{i}.post_attention_layernorm.weight"]
+    return out
+
+
+# Functionality that simulates `torch.load` but where individual tensors are
+# only loaded into memory on demand, not all at once.
+# PyTorch can't do this natively as of time of writing:
+# - https://github.com/pytorch/pytorch/issues/64327
+# This allows us to de-shard without multiplying RAM usage, and also
+# conveniently drops the PyTorch dependency (though we still need numpy).
+
+
+@dataclass
+class LazyStorageKind:
+    data_type: DataType
+
+
+@dataclass
+class LazyStorage:
+    load: Callable[[int, int], NDArray]
+    kind: LazyStorageKind
+    description: str
+
+
+class LazyUnpickler(pickle.Unpickler):
+    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
+        super().__init__(fp)
+        self.data_base_path = data_base_path
+        self.zip_file = zip_file
+
+    def persistent_load(self, pid: Any) -> Any:
+        assert pid[0] == 'storage'
+        assert isinstance(pid[1], LazyStorageKind)
+        data_type = pid[1].data_type
+        filename_stem = pid[2]
+        filename = self.data_base_path + '/' + filename_stem
+        info = self.zip_file.getinfo(filename)
+
+        def load(offset: int, elm_count: int) -> NDArray:
+            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
+            if dtype is None:
+                raise Exception("tensor stored in unsupported format")
+            fp = self.zip_file.open(info)
+            fp.seek(offset * dtype.itemsize)
+            size = elm_count * dtype.itemsize
+            data = fp.read(size)
+            assert len(data) == size
+            return np.frombuffer(data, dtype)
+        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
+        return LazyStorage(load=load, kind=pid[1], description=description)
+
+    # @staticmethod
+    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
+                               # pyright: ignore[reportSelfClsParameterName]
+                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
+        assert isinstance(storage, LazyStorage)
+
+        def load() -> UnquantizedTensor:
+            elm_count = stride[0] * size[0]
+            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
+        description = f'pickled storage_offset={storage_offset} in {storage.description}'
+        return LazyTensor(load, list(size), storage.kind.data_type, description)
+
+    # @staticmethod
+    def rebuild_from_type_v2(func, new_type, args, state):
+        return func(*args)
+
+    CLASSES: Dict[Any, Any] = {
+        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
+        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
+        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
+        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
+        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
+        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
+        ('torch', 'Tensor'): LazyTensor,
+    }
+
+    def find_class(self, module: str, name: str) -> Any:
+        if not module.startswith('torch'):
+            return super().find_class(module, name)
+        return self.CLASSES[(module, name)]
+
+
+def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
+    zf = zipfile.ZipFile(outer_fp)
+    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
+    assert len(pickle_paths) == 1, pickle_paths
+    pickle_fp = zf.open(pickle_paths[0], 'r')
+    unpickler = LazyUnpickler(pickle_fp,
+                              data_base_path=pickle_paths[0][:-4],
+                              zip_file=zf)
+    model = unpickler.load()
+    as_dict = dict(model.items())
+    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
+
+
+SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
+    'BF16': DT_BF16,
+    'F16': DT_F16,
+    'F32': DT_F32,
+    'I32': DT_I32,
+}
+
+
+def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
+    header_size, = struct.unpack('<Q', fp.read(8))
+    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
+    # Use mmap for the actual data to avoid race conditions with the file offset.
+    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
+    byte_buf = mapped[8 + header_size:]
+
+    def convert(info: Dict[str, Any]) -> LazyTensor:
+        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
+        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
+        shape: List[int] = info['shape']
+        begin, end = info['data_offsets']
+        assert 0 <= begin <= end <= len(byte_buf)
+        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
+        buf = byte_buf[begin:end]
+
+        def load() -> UnquantizedTensor:
+            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
+        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
+        return LazyTensor(load, shape, data_type, description)
+    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
+    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
+
+
+def must_read(fp: IO[bytes], length: int) -> bytes:
+    ret = fp.read(length)
+    if len(ret) < length:
+        raise Exception("unexpectedly reached end of file")
+    return ret
+
+
+@functools.lru_cache(maxsize=None)
+def lazy_load_file(path: Path) -> ModelPlus:
+    fp = open(path, 'rb')
+    first8 = fp.read(8)
+    fp.seek(0)
+    if first8[:2] == b'PK':
+        # A zip file, i.e. PyTorch format
+        return lazy_load_torch_file(fp, path)
+    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
+        # Probably safetensors
+        return lazy_load_safetensors_file(fp, path)
+    else:
+        raise ValueError(f"unknown format: {path}")
+
+
+In = TypeVar('In')
+Out = TypeVar('Out')
+
+
+def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
+    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
+    fast enough, this will stop calling `func` at some point rather than
+    letting results pile up in memory.  Specifically, there is a max of one
+    output value buffered per thread.'''
+    with concurrent.futures.ThreadPoolExecutor() as executor:
+        futures: List[concurrent.futures.Future[Out]] = []
+        items_rev = list(iterable)[::-1]
+        for i in range(min(concurrency, len(items_rev))):
+            futures.append(executor.submit(func, items_rev.pop()))
+        while futures:
+            result = futures.pop(0).result()
+            if items_rev:
+                futures.append(executor.submit(func, items_rev.pop()))
+            yield result
+
+
+def check_vocab_size(params: Params, vocab: Vocab) -> None:
+    if params.n_vocab != vocab.vocab_size:
+        # GGMLVocab comes from the same file as the model so shouldn't mismatch:
+        assert isinstance(vocab, SentencePieceVocab)
+        if params.n_vocab == vocab.vocab_size_base:
+            print("Ignoring added_tokens.json since model matches vocab size without it.")
+            vocab.added_tokens_list = []
+            vocab.vocab_size = vocab.vocab_size_base
+            return
+        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
+        if vocab.fname_added_tokens is not None:
+            msg += f" combined with {vocab.fname_added_tokens}"
+        msg += f" has {vocab.vocab_size})."
+        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
+            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
+        raise Exception(msg)
+
+
+class OutputFile:
+    def __init__(self, fname_out: Path) -> None:
+        self.fout = open(fname_out, "wb")
+
+    def write_file_header(self, params: Params, file_type: GGMLFileType) -> None:
+        self.fout.write(b"ggjt"[::-1])  # magic
+        values = [
+            1,  # file version
+            params.n_vocab,
+            params.n_embd,
+            params.n_mult,
+            params.n_head,
+            params.n_layer,
+            params.n_embd // params.n_head,  # rot (obsolete)
+            file_type.value,
+        ]
+        self.fout.write(struct.pack("i" * len(values), *values))
+
+    def write_tensor_header(self, name: str, shape: Sequence[int], data_type: DataType) -> None:
+        sname = name.encode('utf-8')
+        self.fout.write(struct.pack("iii", len(shape), len(sname), DATA_TYPE_TO_FTYPE[data_type]))
+        self.fout.write(struct.pack("i" * len(shape), *shape[::-1]))
+        self.fout.write(sname)
+        self.fout.seek((self.fout.tell() + 31) & -32)
+
+    def write_vocab(self, vocab: Vocab) -> None:
+        for text, score in vocab.all_tokens():
+            self.fout.write(struct.pack("i", len(text)))
+            self.fout.write(text)
+            self.fout.write(struct.pack("f", score))
+
+    @staticmethod
+    def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
+        of = OutputFile(fname_out)
+        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0)
+        of = OutputFile(fname_out)
+        of.write_file_header(params, file_type=GGMLFileType.AllF32)
+        of.write_vocab(vocab)
+        of.fout.close()
+
+    @staticmethod
+    def write_all(fname_out: Path, params: Params, file_type: GGMLFileType, model: LazyModel, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+        of = OutputFile(fname_out)
+        of.write_file_header(params, file_type)
+        print("Writing vocab...")
+        of.write_vocab(vocab)
+
+        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
+            name, lazy_tensor = item
+            return lazy_tensor.load().to_ggml().ndarray
+
+        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8)
+        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
+            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
+            padi = len(str(len(model)))
+            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
+            of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
+            ndarray.tofile(of.fout)
+        of.fout.close()
+
+
+def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
+    wq_type = model["layers.0.attention.wq.weight"].data_type
+    if output_type_str == "f32" or (output_type_str is None and wq_type in (DT_F32, DT_BF16)):
+        return GGMLFileType.AllF32
+    if output_type_str == "f16" or (output_type_str is None and wq_type == DT_F16):
+        return GGMLFileType.MostlyF16
+    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
+    raise Exception(f"Unexpected combination of types: {name_to_type}")
+
+
+def do_necessary_conversions(model: LazyModel, params: Params) -> LazyModel:
+    if "lm_head.weight" in model:
+        model = convert_transformers_to_orig(model, params)
+    model = filter_and_sort_tensors(model)
+
+    return model
+
+
+def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
+    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
+            for (name, tensor) in model.items()}
+
+
+def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the nth path in the model.
+    '''
+    # Support the following patterns:
+    patterns: List[Tuple[str, str]] = [
+        # - x.00.pth, x.01.pth, etc.
+        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
+        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
+        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
+        # x.bin, x.bin.1, etc.
+        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
+    ]
+    for regex, replacement in patterns:
+        if re.search(regex, path.name):
+            new_path = path.with_name(re.sub(regex, replacement, path.name))
+            if new_path.exists():
+                return new_path
+    return None
+
+
+def find_multifile_paths(path: Path) -> List[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the whole list of paths in the model.
+    '''
+    ret: List[Path] = []
+    for i in itertools.count():
+        nth_path = nth_multifile_path(path, i)
+        if nth_path is None:
+            break
+        ret.append(nth_path)
+    if not ret:
+        # No matches.  This should only happen if the file was named, e.g.,
+        # foo.0, and there was no file named foo.  Oh well, try to process it
+        # as a single file.
+        return [path]
+    return ret
+
+
+def load_some_model(path: Path) -> ModelPlus:
+    '''Load a model of any supported format.'''
+    # Be extra-friendly and accept either a file or a directory:
+    if path.is_dir():
+        # Check if it's a set of safetensors files first
+        files = list(path.glob("model-00001-of-*.safetensors"))
+        if not files:
+            # Try the PyTorch patterns too, with lower priority
+            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
+            files = [file for glob in globs for file in path.glob(glob)]
+        if not files:
+            # Try GGML too, but with lower priority, since if both a non-GGML
+            # model and a GGML model exist in the same directory, we assume the
+            # latter was converted from the former.
+            files = list(path.glob("ggml-model*.bin*"))
+        if not files:
+            raise Exception(f"Can't find model in directory {path}")
+        if len(files) > 1:
+            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
+        path = files[0]
+
+    paths = find_multifile_paths(path)
+    models_plus: List[ModelPlus] = []
+    for path in paths:
+        print(f"Loading model file {path}")
+        models_plus.append(lazy_load_file(path))
+
+    model_plus = merge_multifile_models(models_plus)
+    return model_plus
+
+
+def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
+    return {name: model[name] for name in TENSORS_LIST if name in model}
+
+
+def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
+    print(f"vocabtype: {vocabtype}")
+    # Be extra-friendly and accept either a file or a directory.  Also, if it's
+    # a directory, it might be the model directory, and tokenizer.model might
+    # be in the parent of that.
+    if path.is_dir():
+        vocab_file = "tokenizer.model"
+        if vocabtype == 'bpe':
+          vocab_file = "vocab.json"
+        path2 = path / vocab_file
+        # Use `.parent` instead of /.. to handle the symlink case better.
+        path3 = path.parent / vocab_file
+        if path2.exists():
+            path = path2
+        elif path3.exists():
+            path = path3
+        else:
+            raise FileNotFoundError(
+                f"Could not find tokenizer.model in {path} or its parent; "
+                "if it's in another directory, pass the directory as --vocab-dir")
+    added_tokens_path = path.parent / "added_tokens.json"
+    print(f"Loading vocab file {path}")
+    return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None,
+                              vocabtype)
+
+
+def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
+    namestr = {
+        GGMLFileType.AllF32:    "f32",
+        GGMLFileType.MostlyF16: "f16",
+    }[file_type]
+    ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
+    if ret in model_paths:
+        sys.stderr.write(
+            f"Error: Default output path ({ret}) would overwrite the input. "
+            "Please explicitly specify a path using --outfile.\n")
+        sys.exit(1)
+    return ret
+
+
+def do_dump_model(model_plus: ModelPlus) -> None:
+    print(f"model_plus.paths = {model_plus.paths!r}")
+    print(f"model_plus.format = {model_plus.format!r}")
+    print(f"model_plus.vocab = {model_plus.vocab!r}")
+    for name, lazy_tensor in model_plus.model.items():
+        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
+
+
+def main(args_in: Optional[List[str]] = None) -> None:
+    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
+    parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
+    parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
+    parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
+    parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)")
+    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
+    parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
+    parser.add_argument("model", type=Path,
+                        help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
+    parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
+    args = parser.parse_args(args_in)
+
+    vocab: Vocab
+    if args.dump_single:
+        model_plus = lazy_load_file(args.model)
+        do_dump_model(model_plus)
+    elif args.vocab_only:
+        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
+        assert args.outfile, "need --outfile if using --vocab-only"
+        outfile = args.outfile
+        OutputFile.write_vocab_only(outfile, vocab)
+        print(f"Wrote {outfile}")
+    else:
+        model_plus = load_some_model(args.model)
+        if args.dump:
+            do_dump_model(model_plus)
+            return
+        if model_plus.vocab is not None and args.vocab_dir is None:
+            vocab = model_plus.vocab
+        else:
+            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
+            vocab = load_vocab(vocab_dir, args.vocabtype)
+        params = Params.load(model_plus)
+        model = model_plus.model
+        model = do_necessary_conversions(model, params)
+        output_type = pick_output_type(model, args.outtype)
+        model = convert_to_output_type(model, output_type)
+        outfile = args.outfile or default_outfile(model_plus.paths, output_type)
+        OutputFile.write_all(outfile, params, output_type, model, vocab)
+        print(f"Wrote {outfile}")
+
+
+if __name__ == '__main__':
+    main()

From d2bb3ac10b025e9c2e475a0bc9527a7286b63810 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 27 Jul 2023 16:36:35 +0300
Subject: [PATCH 019/242] convert.py : remove GGML vocab + other obsolete stuff

---
 convert-new.py | 46 +++++++++++++++++-----------------------------
 1 file changed, 17 insertions(+), 29 deletions(-)

diff --git a/convert-new.py b/convert-new.py
index dacad693fea1c..bf85cb55177fa 100755
--- a/convert-new.py
+++ b/convert-new.py
@@ -278,19 +278,7 @@ def __repr__(self) -> str:
         return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
 
 
-class GGMLVocab:
-    def __init__(self, tokens: List[Tuple[bytes, float]]):
-        self.tokens = tokens
-        self.vocab_size = len(tokens)
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        return self.tokens
-
-    def __repr__(self) -> str:
-        return f"<GGMLVocab with {self.vocab_size} tokens>"
-
-
-Vocab = Union[SentencePieceVocab, GGMLVocab]
+Vocab = Union[SentencePieceVocab]
 
 
 def permute(weights: NDArray, n_head: int) -> NDArray:
@@ -691,7 +679,6 @@ def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], conc
 
 def check_vocab_size(params: Params, vocab: Vocab) -> None:
     if params.n_vocab != vocab.vocab_size:
-        # GGMLVocab comes from the same file as the model so shouldn't mismatch:
         assert isinstance(vocab, SentencePieceVocab)
         if params.n_vocab == vocab.vocab_size_base:
             print("Ignoring added_tokens.json since model matches vocab size without it.")
@@ -874,7 +861,7 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
     if path.is_dir():
         vocab_file = "tokenizer.model"
         if vocabtype == 'bpe':
-          vocab_file = "vocab.json"
+            vocab_file = "vocab.json"
         path2 = path / vocab_file
         # Use `.parent` instead of /.. to handle the symlink case better.
         path3 = path.parent / vocab_file
@@ -916,15 +903,14 @@ def do_dump_model(model_plus: ModelPlus) -> None:
 
 def main(args_in: Optional[List[str]] = None) -> None:
     parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
-    parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
-    parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
-    parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
-    parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)")
-    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
-    parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
-    parser.add_argument("model", type=Path,
-                        help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
+    parser.add_argument("--dump",        action="store_true",    help="don't convert, just show what's in the model")
+    parser.add_argument("--dump-single", action="store_true",    help="don't convert, just show what's in a single model file")
+    parser.add_argument("--vocab-only",  action="store_true",    help="extract only the vocab")
+    parser.add_argument("--outtype",     choices=["f32", "f16"], help="output format (default: based on input)")
+    parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
+    parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
+    parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
+    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)")
     args = parser.parse_args(args_in)
 
     vocab: Vocab
@@ -947,12 +933,14 @@ def main(args_in: Optional[List[str]] = None) -> None:
         else:
             vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
             vocab = load_vocab(vocab_dir, args.vocabtype)
-        params = Params.load(model_plus)
-        model = model_plus.model
-        model = do_necessary_conversions(model, params)
+
+        params      = Params.load(model_plus)
+        model       = model_plus.model
+        model       = do_necessary_conversions(model, params)
         output_type = pick_output_type(model, args.outtype)
-        model = convert_to_output_type(model, output_type)
-        outfile = args.outfile or default_outfile(model_plus.paths, output_type)
+        model       = convert_to_output_type(model, output_type)
+        outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
+
         OutputFile.write_all(outfile, params, output_type, model, vocab)
         print(f"Wrote {outfile}")
 

From 11ef380c2a3a68289fa4fda0403b69ff0f59ddb2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 28 Jul 2023 11:34:16 +0300
Subject: [PATCH 020/242] GGUF : write tensor (#2426)

* WIP: Write tensor

* GGUF : Support writing tensors in Python

* refactor : rm unused import and upd todos

* fix : fix errors upd writing example

* rm example.gguf

* gitignore *.gguf

* undo formatting
---
 .gitignore   |   1 +
 constants.py |   5 ++-
 gguf.py      | 103 ++++++++++++++++++++++++++++++++++-----------------
 3 files changed, 73 insertions(+), 36 deletions(-)

diff --git a/.gitignore b/.gitignore
index abe8e28cb7687..fd41f26cc0d20 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,6 +1,7 @@
 *.o
 *.a
 *.so
+*.gguf
 .DS_Store
 .build/
 .cache/
diff --git a/constants.py b/constants.py
index 3a97460e5dc5e..34880bb20db9c 100644
--- a/constants.py
+++ b/constants.py
@@ -1,5 +1,6 @@
-GGUF_MAGIC   = 0x47475546
-GGUF_VERSION = 1
+GGUF_MAGIC             = 0x47475546
+GGUF_VERSION           = 1
+GGUF_DEFAULT_ALIGNMENT = 32
 
 # general
 KEY_GENERAL_ARCHITECTURE         = "general.architecture"
diff --git a/gguf.py b/gguf.py
index 764ae9a9df27c..c5b2174c9070a 100644
--- a/gguf.py
+++ b/gguf.py
@@ -1,14 +1,16 @@
 """TODOs
-1. Implement writing tensor data with alignment.
-2. Implement writers for known architectures, LLaMA in particular.
-3. Add docstrings from the format specs.
-4. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
+1. Implement writers for known architectures, LLaMA in particular.
+2. Add docstrings from the format specs.
+3. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
 """
 
 import struct
 import constants
 from enum import IntEnum
-from typing import List, Any
+from typing import Any, IO, List
+
+import numpy as np
+
 
 class GGMLQuantizationType(IntEnum):
     F32  = 0
@@ -54,15 +56,18 @@ def get_type(val):
         else:
             return GGUFValueType.INT32
 
+
 class GGUFWriter:
-    def __init__(self, buffered_writer):
-        self.buffered_writer = buffered_writer
+    def __init__(self, fout: IO):
+        self.fout = fout
+        self.offset_tensor = 0
+        self.tensors: List[np.ndarray] = []
 
     def write_header(self, tensor_count: int, metadata_kv_count: int):
-        self.buffered_writer.write(struct.pack("<I", constants.GGUF_MAGIC))
-        self.buffered_writer.write(struct.pack("<I", constants.GGUF_VERSION))
-        self.buffered_writer.write(struct.pack("<I", tensor_count))
-        self.buffered_writer.write(struct.pack("<I", metadata_kv_count))
+        self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
+        self.fout.write(struct.pack("<I", constants.GGUF_VERSION))
+        self.fout.write(struct.pack("<I", tensor_count))
+        self.fout.write(struct.pack("<I", metadata_kv_count))
 
     @classmethod
     def open(cls, path: str) -> "GGUFWriter":
@@ -119,40 +124,69 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None):
         if vtype is None:
             vtype = GGUFValueType.get_type(val)
 
-        self.buffered_writer.write(struct.pack("<I", vtype))
+        self.fout.write(struct.pack("<I", vtype))
 
         if vtype == GGUFValueType.UINT8:
-            self.buffered_writer.write(struct.pack("<B", val))
+            self.fout.write(struct.pack("<B", val))
         elif vtype == GGUFValueType.INT8:
-            self.buffered_writer.write(struct.pack("<b", val))
+            self.fout.write(struct.pack("<b", val))
         elif vtype == GGUFValueType.UINT16:
-            self.buffered_writer.write(struct.pack("<H", val))
+            self.fout.write(struct.pack("<H", val))
         elif vtype == GGUFValueType.INT16:
-            self.buffered_writer.write(struct.pack("<h", val))
+            self.fout.write(struct.pack("<h", val))
         elif vtype == GGUFValueType.UINT32:
-            self.buffered_writer.write(struct.pack("<I", val))
+            self.fout.write(struct.pack("<I", val))
         elif vtype == GGUFValueType.INT32:
-            self.buffered_writer.write(struct.pack("<i", val))
+            self.fout.write(struct.pack("<i", val))
         elif vtype == GGUFValueType.FLOAT32:
-            self.buffered_writer.write(struct.pack("<f", val))
+            self.fout.write(struct.pack("<f", val))
         elif vtype == GGUFValueType.BOOL:
-            self.buffered_writer.write(struct.pack("?", val))
+            self.fout.write(struct.pack("?", val))
         elif vtype == GGUFValueType.STRING:
             encoded_val = val.encode("utf8")
-            self.buffered_writer.write(struct.pack("<I", len(encoded_val)))
-            self.buffered_writer.write(encoded_val)
+            self.fout.write(struct.pack("<I", len(encoded_val)))
+            self.fout.write(encoded_val)
         elif vtype == GGUFValueType.ARRAY:
-            self.buffered_writer.write(struct.pack("<I", len(val)))
+            self.fout.write(struct.pack("<I", len(val)))
             for item in val:
                 self.write_val(item)
         else:
             raise ValueError("Invalid GGUF metadata value type")
 
+    @staticmethod
+    def ggml_pad(x: int, n: int) -> int:
+        return ((x + n - 1) // n) * n
+
+    def write_tensor_info(self, name: str, tensor: np.ndarray):
+        self.write_val(name, GGUFValueType.STRING)
+        n_dims = len(tensor.shape)
+        self.write_val(n_dims, GGUFValueType.INT32)
+        for i in range(n_dims):
+            self.write_val(tensor.shape[n_dims - 1 - i], GGUFValueType.INT32)
+
+        assert tensor.dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
+        dtype = GGMLQuantizationType.F32 if tensor.dtype == np.float32 else GGMLQuantizationType.F16
+        self.write_val(dtype, GGUFValueType.INT32)
+        self.fout.write(struct.pack("<Q", self.offset_tensor))
+        self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT)
+
+        offset_data = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT)
+        pad = offset_data - self.fout.tell()
+        self.fout.write(bytes([0] * pad))
+
+        self.tensors.append(tensor)
+
+    def write_tensors(self):
+        for tensor in self.tensors:
+            tensor.tofile(self.fout)
+            pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
+            self.fout.write(bytes([0] * pad))
+
     def flush(self):
-        self.buffered_writer.flush()
+        self.fout.flush()
 
     def close(self):
-        self.buffered_writer.close()
+        self.fout.close()
 
     def write_architecture(self, architecture: str):
         self.write_string(constants.KEY_GENERAL_ARCHITECTURE,
@@ -235,14 +269,15 @@ def write_rope_scale(self, llm: str, value:  float):
 if __name__ == "__main__":
     # Example usage with a file
     gguf_writer = GGUFWriter.open("example.gguf")
-    gguf_writer.write_header(0, 3)
-
-gguf_writer.write_architecture("llama")
-gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
-gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
-# Write an array of integers
-#gguf_writer.write_array("simple_array", [1, 2, 3, 4])
-# Write a nested array
-#gguf_writer.write_array("nested", [1, "nested", [2, 3]])
+    gguf_writer.write_header(2, 3)
+
+    gguf_writer.write_architecture("llama")
+    gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
+    gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
+    tensor1 = np.random.random(size=(7, 10)).astype(np.float32)
+    tensor2 = np.random.random(size=(16, 12)).astype(np.float16)
+    gguf_writer.write_tensor_info("tensor1", tensor1)
+    gguf_writer.write_tensor_info("tensor2", tensor2)
+    gguf_writer.write_tensors()
 
 gguf_writer.close()

From 3492f848d73ddb3211d34572d3b1d3a04ffd0d3f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 28 Jul 2023 22:45:24 +0200
Subject: [PATCH 021/242] gguf : add gguf_find_key (#2438)

* gguf.cpp : find key example

* ggml.h : add gguf_find_key

* ggml.c : add gguf_find_key
---
 examples/gguf/gguf.cpp | 14 ++++++++++++++
 ggml.c                 | 15 +++++++++++++++
 ggml.h                 |  1 +
 3 files changed, 30 insertions(+)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index c3494a343a6ec..35a870a8343f5 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -253,6 +253,20 @@ bool gguf_ex_read_0(const std::string & fname) {
         }
     }
 
+    // find kv string
+    {
+        char findkey[32];
+        sprintf(findkey, "some.parameter.string");
+
+        int keyidx = gguf_find_key(ctx, findkey);
+        if (keyidx == -1) {
+            fprintf(stdout, "%s: find key: %s not found.\n", __func__, findkey);
+        } else {
+            const char * key_value = gguf_get_val_str(ctx, keyidx);
+            fprintf(stdout, "%s: find key: %s found, kv[%d] value = %s\n", __func__, findkey, keyidx, key_value);
+        }
+    }
+
     // tensor info
     {
         const int n_tensors = gguf_get_n_tensors(ctx);
diff --git a/ggml.c b/ggml.c
index 96c7ebd347c1b..d402daa678b90 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18745,6 +18745,21 @@ int gguf_get_n_kv(struct gguf_context * ctx) {
     return ctx->header.n_kv;
 }
 
+int gguf_find_key(struct gguf_context * ctx, const char * key) {
+    // return -1 if key not found
+    const int n_kv = gguf_get_n_kv(ctx);
+    int keyfound   = -1;
+
+    for (int i = 0; i < n_kv; ++i) {
+        if (strcmp(key, gguf_get_key(ctx, i)) == 0) {
+            keyfound = i;
+            break;
+        }
+    }
+
+    return keyfound;
+}
+
 const char * gguf_get_key(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].key.data;
 }
diff --git a/ggml.h b/ggml.h
index e857b3f14435f..a72c82a333cf5 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1653,6 +1653,7 @@ extern "C" {
     GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
     GGML_API int          gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API int          gguf_find_key(struct gguf_context * ctx, const char * key);
     GGML_API const char * gguf_get_key (struct gguf_context * ctx, int i);
     GGML_API void         gguf_get_val (struct gguf_context * ctx, int i, void * val);
 

From 1495735aaced6ca90209c7b3716ac77376b90d45 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 00:26:22 +0300
Subject: [PATCH 022/242] gguf : fix writing tensors

---
 gguf.py | 56 ++++++++++++++++++++++++++++++++------------------------
 1 file changed, 32 insertions(+), 24 deletions(-)

diff --git a/gguf.py b/gguf.py
index c5b2174c9070a..86c122a5e77d1 100644
--- a/gguf.py
+++ b/gguf.py
@@ -13,9 +13,9 @@
 
 
 class GGMLQuantizationType(IntEnum):
-    F32  = 0
-    F16  = 1
-    QR_0 = 2
+    F32 = 0
+    F16 = 1
+    Q4_0 = 2
     Q4_1 = 3
     # Q4_2 = 4 # support has been removed
     # Q4_3 = 5 # support has been removed
@@ -32,16 +32,16 @@ class GGMLQuantizationType(IntEnum):
 
 
 class GGUFValueType(IntEnum):
-    UINT8   = 0
-    INT8    = 1
-    UINT16  = 2
-    INT16   = 3
-    UINT32  = 4
-    INT32   = 5
+    UINT8 = 0
+    INT8 = 1
+    UINT16 = 2
+    INT16 = 3
+    UINT32 = 4
+    INT32 = 5
     FLOAT32 = 6
-    BOOL    = 7
-    STRING  = 8
-    ARRAY   = 9
+    BOOL = 7
+    STRING = 8
+    ARRAY = 9
 
     @staticmethod
     def get_type(val):
@@ -75,7 +75,9 @@ def open(cls, path: str) -> "GGUFWriter":
         return cls(f)
 
     def write_key(self, key: str):
-        self.write_val(key, GGUFValueType.STRING)
+        encoded_key = key.encode("utf8")
+        self.fout.write(struct.pack("<I", len(encoded_key)))
+        self.fout.write(encoded_key)
 
     def write_uint8(self, key: str, val: int):
         self.write_key(key)
@@ -158,29 +160,35 @@ def ggml_pad(x: int, n: int) -> int:
         return ((x + n - 1) // n) * n
 
     def write_tensor_info(self, name: str, tensor: np.ndarray):
-        self.write_val(name, GGUFValueType.STRING)
+        self.write_key(name)
         n_dims = len(tensor.shape)
-        self.write_val(n_dims, GGUFValueType.INT32)
+        self.fout.write(struct.pack("<i", n_dims))
         for i in range(n_dims):
-            self.write_val(tensor.shape[n_dims - 1 - i], GGUFValueType.INT32)
+            self.fout.write(struct.pack("<i", tensor.shape[n_dims - 1 - i]))
 
         assert tensor.dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
         dtype = GGMLQuantizationType.F32 if tensor.dtype == np.float32 else GGMLQuantizationType.F16
-        self.write_val(dtype, GGUFValueType.INT32)
+        self.fout.write(struct.pack("<i", dtype))
         self.fout.write(struct.pack("<Q", self.offset_tensor))
         self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT)
 
-        offset_data = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT)
-        pad = offset_data - self.fout.tell()
-        self.fout.write(bytes([0] * pad))
+        self.flush()
 
         self.tensors.append(tensor)
 
     def write_tensors(self):
+        offset_data = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT)
+        pad = offset_data - self.fout.tell()
+        print(f"pad: {pad}")
+        if pad != 0:
+            self.fout.write(bytes([0] * pad))
+
         for tensor in self.tensors:
             tensor.tofile(self.fout)
             pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
-            self.fout.write(bytes([0] * pad))
+            print(f"pad: {pad}")
+            if pad != 0:
+                self.fout.write(bytes([0] * pad))
 
     def flush(self):
         self.fout.flush()
@@ -274,10 +282,10 @@ def write_rope_scale(self, llm: str, value:  float):
     gguf_writer.write_architecture("llama")
     gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
     gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
-    tensor1 = np.random.random(size=(7, 10)).astype(np.float32)
-    tensor2 = np.random.random(size=(16, 12)).astype(np.float16)
+    tensor1 = np.ones((7, 8, 3), dtype=np.float32)
+    tensor2 = np.ones((7, 8, 3), dtype=np.float32)
     gguf_writer.write_tensor_info("tensor1", tensor1)
     gguf_writer.write_tensor_info("tensor2", tensor2)
     gguf_writer.write_tensors()
 
-gguf_writer.close()
+    gguf_writer.close()

From 08dc8fd88428e2fb6ca689d6283b2999677a5f32 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 10:24:46 +0300
Subject: [PATCH 023/242] gguf : do not hardcode tensor names to read

---
 examples/gguf/gguf.cpp | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 35a870a8343f5..84388c219f5ef 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -296,7 +296,7 @@ bool gguf_ex_read_1(const std::string & fname) {
     };
 
     struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
-
+    
     fprintf(stdout, "%s: version:      %d\n", __func__, gguf_get_version(ctx));
     fprintf(stdout, "%s: alignment:   %zu\n", __func__, gguf_get_alignment(ctx));
     fprintf(stdout, "%s: data offset: %zu\n", __func__, gguf_get_data_offset(ctx));
@@ -335,9 +335,9 @@ bool gguf_ex_read_1(const std::string & fname) {
         for (int i = 0; i < n_tensors; ++i) {
             fprintf(stdout, "%s: reading tensor %d data\n", __func__, i);
 
-            const std::string name = "tensor_" + to_string(i);
+            const char * name = gguf_get_tensor_name(ctx, i);
 
-            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name.c_str());
+            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
 
             fprintf(stdout, "%s: tensor[%d]: n_dims = %d, name = %s, data = %p\n",
                     __func__, i, cur->n_dims, cur->name, cur->data);

From 06f423a8e1ca6c4c1d348a5c6003bbab05644807 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 10:26:26 +0300
Subject: [PATCH 024/242] gguf : write sample tensors to read

---
 gguf.py | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/gguf.py b/gguf.py
index 86c122a5e77d1..25d99ae93dbcd 100644
--- a/gguf.py
+++ b/gguf.py
@@ -179,14 +179,12 @@ def write_tensor_info(self, name: str, tensor: np.ndarray):
     def write_tensors(self):
         offset_data = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT)
         pad = offset_data - self.fout.tell()
-        print(f"pad: {pad}")
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
         for tensor in self.tensors:
             tensor.tofile(self.fout)
             pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
-            print(f"pad: {pad}")
             if pad != 0:
                 self.fout.write(bytes([0] * pad))
 
@@ -282,10 +280,10 @@ def write_rope_scale(self, llm: str, value:  float):
     gguf_writer.write_architecture("llama")
     gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
     gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
-    tensor1 = np.ones((7, 8, 3), dtype=np.float32)
-    tensor2 = np.ones((7, 8, 3), dtype=np.float32)
-    gguf_writer.write_tensor_info("tensor1", tensor1)
-    gguf_writer.write_tensor_info("tensor2", tensor2)
+    tensor1 = np.ones((32,), dtype=np.float32) * 100.0
+    tensor2 = np.ones((32,), dtype=np.float32) * 101.0
+    gguf_writer.write_tensor_info("tensor0", tensor1)
+    gguf_writer.write_tensor_info("tensor1", tensor2)
     gguf_writer.write_tensors()
 
     gguf_writer.close()

From d54f53ca51a35af80bc6bf15849042c717587038 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 12:04:45 +0300
Subject: [PATCH 025/242] gguf : add tokenization constants

---
 constants.py | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/constants.py b/constants.py
index 34880bb20db9c..024bf7b03bbe1 100644
--- a/constants.py
+++ b/constants.py
@@ -31,3 +31,16 @@
 # RoPE
 KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
 KEY_ROPE_SCALE                   = "{llm}.rope.scale"
+
+# tokenization
+KEY_TOKENIZER_MODEL   = "tokenizer.ggml.model"
+KEY_TOKENIZER_LIST    = "tokenizer.ggml.tokens"
+KEY_TOKENIZER_SCORES  = "tokenizer.ggml.scores"
+KEY_TOKENIZER_MERGES  = "tokenizer.ggml.merges"
+KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.seperator_token_id"
+KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
+KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json"
+KEY_TOKENIZER_RWKV    = "tokenizer.rwkv.world"

From 999431c4b651ec4faa2b8715ce5f32ceeb3b9cb4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 11:20:05 +0200
Subject: [PATCH 026/242] quick and dirty conversion example

---
 convert-llama-h5-to-gguf.py | 231 ++++++++++++++++++++++++++++++++++++
 1 file changed, 231 insertions(+)
 create mode 100644 convert-llama-h5-to-gguf.py

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
new file mode 100644
index 0000000000000..29daed3b3cf48
--- /dev/null
+++ b/convert-llama-h5-to-gguf.py
@@ -0,0 +1,231 @@
+# Quick and dirty HF llama --> gguf conversion, GQA/70b wont work
+
+import gguf
+import sys
+import struct
+import json
+import numpy as np
+from typing import List
+from pathlib import Path
+from transformers import AutoModelForCausalLM
+from sentencepiece import SentencePieceProcessor
+
+
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+def permute(weights: NDArray, n_head: int) -> NDArray:
+    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
+                   .swapaxes(1, 2)
+                   .reshape(weights.shape))
+
+if len(sys.argv) < 3:
+    print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
+    print("  ftype == 0 -> float32")
+    print("  ftype == 1 -> float16")
+    sys.exit(1)
+
+
+# output in the same directory as the model
+dir_model = sys.argv[1]
+fname_out = sys.argv[1] + "/ggml-model.bin"
+
+
+# possible tensor data types
+#   ftype == 0 -> float32
+#   ftype == 1 -> float16
+#
+# map from ftype to string
+ftype_str = ["f32", "f16"]
+
+ftype = 1
+if len(sys.argv) > 2:
+    ftype = int(sys.argv[2])
+    if ftype < 0 or ftype > 1:
+        print("Invalid ftype: " + str(ftype))
+        sys.exit(1)
+    fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+
+model = AutoModelForCausalLM.from_pretrained( dir_model, low_cpu_mem_usage=True, trust_remote_code=True )
+list_vars = model.state_dict()
+
+# count tensors to be converted
+tensor_count = 0
+for name in list_vars.keys():
+    # we don't need these
+    if name.endswith(".rotary_emb.inv_freq"):
+        continue
+    tensor_count += 1
+
+#fout = open(fname_out, "wb")
+gguf_writer = gguf.GGUFWriter.open(fname_out)
+
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+# This mmust be changed when adding/deleting kv
+kv_count = 13
+
+print("tensors " + str(tensor_count) + " kv " + str(kv_count) )
+
+print("write gguf header")
+
+gguf_writer.write_header(tensor_count, kv_count)
+
+print("write gguf hparams")
+
+llm_arch = "llama"
+
+gguf_writer.write_name("llama2-7b")
+gguf_writer.write_description("gguf test model")
+gguf_writer.write_architecture(llm_arch)
+gguf_writer.write_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.write_embedding_length(llm_arch, hparams["hidden_size"])
+gguf_writer.write_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
+gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
+gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
+gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparams["rms_norm_eps"])
+
+
+# TOKENIZATION
+
+tokens: List[str] = []
+scores: List[float] = []
+
+if Path( dir_model + "/tokenizer.model").is_file():
+    # vocab type SPIECE
+    print( "Adding sentencepiece tokenizer vocab." )
+    tokenizer = SentencePieceProcessor( dir_model + "/tokenizer.model" )
+
+    # output vocab_size followed by all piece/score pairs
+    outbytes: bytes
+    outbytes = b""
+    outbytes += struct.pack("I", tokenizer.vocab_size())
+
+    for i in range(tokenizer.vocab_size()):
+        text: bytes
+        if tokenizer.is_unknown(i):
+            text = " \u2047 ".encode("utf-8")
+        elif tokenizer.is_control(i):
+            text = b""
+        if tokenizer.is_byte(i):
+            piece = tokenizer.id_to_piece(i)
+            if len(piece) != 6:
+                raise Exception(f"Invalid token: {piece}")
+            byte_value = int(piece[3:-1], 16)
+            text = struct.pack("B", byte_value)
+        else:
+            text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
+        score: float = tokenizer.get_score(i)
+
+        tokens.append( str(text) );
+        scores.append( score );
+
+print("write gguf tokens")
+
+gguf_writer.write_string("tokenizer.ggml.model", "llama")
+gguf_writer.write_array("tokenizer.ggml.tokens",tokens)
+gguf_writer.write_array("tokenizer.ggml.scores",scores)
+
+# TENSORS
+
+
+# tensor info
+print("write gguf tensor info")
+
+for name in list_vars.keys():
+    data = list_vars[name].squeeze().numpy()
+
+    # we don't need these
+    if name.endswith(".rotary_emb.inv_freq"):
+        continue
+
+    # permute these
+    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+        data = permute( data, hparams["num_attention_heads"] )
+
+    # chnage tensor name
+
+    if name == "model.embed_tokens.weight":
+        name = "tok_embeddings.weight"
+    elif name == "model.norm.weight":
+        name = "norm.weight"
+    elif name == "lm_head.weight":
+        name = "output.weight"
+    else:
+        for i in range(80):  # maximum number of layers
+            if name == "model.layers." + str(i) + ".input_layernorm.weight":
+                name = "layers." + str(i) + ".attention_norm.weight"
+                break
+            if name == "model.layers." + str(i) + ".self_attn.q_proj.weight":
+                name = "layers." + str(i) + ".attention.wq.weight"
+                break
+            if name == "model.layers." + str(i) + ".self_attn.k_proj.weight":
+                name = "layers." + str(i) + ".attention.wk.weight"
+                break
+            if name == "model.layers." + str(i) + ".self_attn.v_proj.weight":
+                name = "layers." + str(i) + ".attention.wv.weight"
+                break
+            if name == "model.layers." + str(i) + ".self_attn.o_proj.weight":
+                name = "layers." + str(i) + ".attention.wo.weight"
+                break
+            if name == "model.layers." + str(i) + ".post_attention_layernorm.weight":
+                name = "layers." + str(i) + ".ffn_norm.weight"
+                break
+            if name == "model.layers." + str(i) + ".mlp.gate_proj.weight":
+                name = "layers." + str(i) + ".feed_forward.w1.weight"
+                break
+            if name == "model.layers." + str(i) + ".mlp.down_proj.weight":
+                name = "layers." + str(i) + ".feed_forward.w2.weight"
+                break
+            if name == "model.layers." + str(i) + ".mlp.up_proj.weight":
+                name = "layers." + str(i) + ".feed_forward.w3.weight"
+                break
+
+    gguf_writer.write_tensor_info(name, data)
+
+
+# tensor data
+print("write gguf tensor data")
+
+for name in list_vars.keys():
+    data = list_vars[name].squeeze().numpy()
+    print("Process tensor: " + name + " with shape: ", data.shape)
+
+    # we don't need these
+    if name.endswith(".rotary_emb.inv_freq"):
+        print("  Skip tensor: " + name)
+        continue
+
+    ## permute these
+    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+        print("  Permute tensor: " + name)
+        data = permute( data, hparams["num_attention_heads"] )
+
+    n_dims = len(data.shape)
+
+    # ftype == 0 -> float32, ftype == 1 -> float16
+    ftype_cur = 0
+    if ftype != 0:
+        if name.endswith(".weight") and n_dims == 2:
+            print("  Converting to float16")
+            data = data.astype(np.float16)
+            ftype_cur = 1
+        else:
+            print("  Converting to float32")
+            data = data.astype(np.float32)
+            ftype_cur = 0
+    else:
+        if data.dtype != np.float32:
+            print("  Converting to float32")
+            data = data.astype(np.float32)
+            ftype_cur = 0
+
+    gguf_writer.write_tensor_padding()
+    gguf_writer.write_tensor(data)
+
+gguf_writer.close()
+
+
+print("Done. Output file: " + fname_out)
+print("")

From ea5f9ad2ca4311dcad93f43bd020a88a72fc0cdb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 12:25:43 +0300
Subject: [PATCH 027/242] gguf : fix writing gguf arrays

---
 gguf.py | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/gguf.py b/gguf.py
index 25d99ae93dbcd..ac376606ffaf7 100644
--- a/gguf.py
+++ b/gguf.py
@@ -122,11 +122,12 @@ def write_array(self, key: str, val: list):
         self.write_key(key)
         self.write_val(val, GGUFValueType.ARRAY)
 
-    def write_val(self: str, val: Any, vtype: GGUFValueType = None):
+    def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: bool = True):
         if vtype is None:
             vtype = GGUFValueType.get_type(val)
 
-        self.fout.write(struct.pack("<I", vtype))
+        if write_vtype:
+            self.fout.write(struct.pack("<I", vtype))
 
         if vtype == GGUFValueType.UINT8:
             self.fout.write(struct.pack("<B", val))
@@ -150,8 +151,10 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None):
             self.fout.write(encoded_val)
         elif vtype == GGUFValueType.ARRAY:
             self.fout.write(struct.pack("<I", len(val)))
+            # TODO: verify that all elements are of the same type
+            self.fout.write(struct.pack("<I", GGUFValueType.get_type(val[0])))
             for item in val:
-                self.write_val(item)
+                self.write_val(item, write_vtype=False)
         else:
             raise ValueError("Invalid GGUF metadata value type")
 
@@ -177,8 +180,7 @@ def write_tensor_info(self, name: str, tensor: np.ndarray):
         self.tensors.append(tensor)
 
     def write_tensors(self):
-        offset_data = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT)
-        pad = offset_data - self.fout.tell()
+        pad = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT) - self.fout.tell()
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 

From 93f7f7aef712f9247a16c277ee55aa24121d3e24 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 12:34:35 +0300
Subject: [PATCH 028/242] gguf : write tensors one by one and code reuse

---
 gguf.py | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

diff --git a/gguf.py b/gguf.py
index ac376606ffaf7..c5a8f5b909630 100644
--- a/gguf.py
+++ b/gguf.py
@@ -61,7 +61,6 @@ class GGUFWriter:
     def __init__(self, fout: IO):
         self.fout = fout
         self.offset_tensor = 0
-        self.tensors: List[np.ndarray] = []
 
     def write_header(self, tensor_count: int, metadata_kv_count: int):
         self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
@@ -75,9 +74,7 @@ def open(cls, path: str) -> "GGUFWriter":
         return cls(f)
 
     def write_key(self, key: str):
-        encoded_key = key.encode("utf8")
-        self.fout.write(struct.pack("<I", len(encoded_key)))
-        self.fout.write(encoded_key)
+        self.write_val(key, GGUFValueType.STRING, write_vtype=False)
 
     def write_uint8(self, key: str, val: int):
         self.write_key(key)
@@ -179,16 +176,15 @@ def write_tensor_info(self, name: str, tensor: np.ndarray):
 
         self.tensors.append(tensor)
 
-    def write_tensors(self):
+    def write_tensor(self, tensor: np.ndarray):
         pad = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT) - self.fout.tell()
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
-        for tensor in self.tensors:
-            tensor.tofile(self.fout)
-            pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
-            if pad != 0:
-                self.fout.write(bytes([0] * pad))
+        tensor.tofile(self.fout)
+        pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
+        if pad != 0:
+            self.fout.write(bytes([0] * pad))
 
     def flush(self):
         self.fout.flush()

From 0c219fb5b56d05d54f10811826a768fe039136e2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 12:42:54 +0300
Subject: [PATCH 029/242] gguf : fix writing gguf arrays

---
 gguf.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index c5a8f5b909630..de75b54d43137 100644
--- a/gguf.py
+++ b/gguf.py
@@ -147,9 +147,10 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: boo
             self.fout.write(struct.pack("<I", len(encoded_val)))
             self.fout.write(encoded_val)
         elif vtype == GGUFValueType.ARRAY:
+            ltype = set([GGUFValueType.get_type(item) for item in val])
+            assert len(ltype) == 1, "All items in a GGUF array should be of the same type"
+            self.fout.write(struct.pack("<I", ltype[0]))
             self.fout.write(struct.pack("<I", len(val)))
-            # TODO: verify that all elements are of the same type
-            self.fout.write(struct.pack("<I", GGUFValueType.get_type(val[0])))
             for item in val:
                 self.write_val(item, write_vtype=False)
         else:

From c861e234f4aecc1b851a7bbff0cb38a74361caf8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 12:49:01 +0300
Subject: [PATCH 030/242] gguf : write tensors one by one

---
 gguf.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gguf.py b/gguf.py
index de75b54d43137..1888b6c2dec8f 100644
--- a/gguf.py
+++ b/gguf.py
@@ -283,6 +283,7 @@ def write_rope_scale(self, llm: str, value:  float):
     tensor2 = np.ones((32,), dtype=np.float32) * 101.0
     gguf_writer.write_tensor_info("tensor0", tensor1)
     gguf_writer.write_tensor_info("tensor1", tensor2)
-    gguf_writer.write_tensors()
+    gguf_writer.write_tensor(tensor1)
+    gguf_writer.write_tensor(tensor2)
 
     gguf_writer.close()

From 8a76dd8a85bf2fd5c0c727c7915c923973d13564 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 13:17:28 +0300
Subject: [PATCH 031/242] gguf : write tensors one by one

---
 gguf.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index 1888b6c2dec8f..cd727f3ea7a84 100644
--- a/gguf.py
+++ b/gguf.py
@@ -175,8 +175,6 @@ def write_tensor_info(self, name: str, tensor: np.ndarray):
 
         self.flush()
 
-        self.tensors.append(tensor)
-
     def write_tensor(self, tensor: np.ndarray):
         pad = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT) - self.fout.tell()
         if pad != 0:

From cc3dd7f0420de545ca39b3f0ea6ca47273e5a439 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 13:30:22 +0300
Subject: [PATCH 032/242] gguf : write tokenizer data

---
 gguf.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/gguf.py b/gguf.py
index cd727f3ea7a84..e0b4b88fb11b8 100644
--- a/gguf.py
+++ b/gguf.py
@@ -267,6 +267,15 @@ def write_rope_dimension_count(self, llm: str, count: int):
     def write_rope_scale(self, llm: str, value:  float):
         self.write_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
 
+    def write_tokenizer_model(self, model: str):
+        self.write_string(constants.KEY_TOKENIZER_MODEL, model)
+
+    def write_token_list(self, tokens: List[str]):
+        self.write_array(constants.KEY_TOKENIZER_LIST, tokens)
+
+    def write_token_scores(self, scores: List[float]):
+        self.write_array(constants.KEY_TOKENIZER_SCORES, scores)
+
 
 # Example usage:
 if __name__ == "__main__":

From 0317c41d9877d6583b8469710fc0aefcf4ba2965 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 13:31:07 +0300
Subject: [PATCH 033/242] gguf : upd gguf conversion script

---
 convert-llama-h5-to-gguf.py | 31 ++++++++++++++++---------------
 1 file changed, 16 insertions(+), 15 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 29daed3b3cf48..439a6da307f88 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -12,11 +12,14 @@
 
 
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+
+
 def permute(weights: NDArray, n_head: int) -> NDArray:
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
                    .swapaxes(1, 2)
                    .reshape(weights.shape))
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
@@ -45,7 +48,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
 
-model = AutoModelForCausalLM.from_pretrained( dir_model, low_cpu_mem_usage=True, trust_remote_code=True )
+model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
 
 # count tensors to be converted
@@ -56,7 +59,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         continue
     tensor_count += 1
 
-#fout = open(fname_out, "wb")
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
@@ -65,7 +67,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 # This mmust be changed when adding/deleting kv
 kv_count = 13
 
-print("tensors " + str(tensor_count) + " kv " + str(kv_count) )
+print("tensors " + str(tensor_count) + " kv " + str(kv_count))
 
 print("write gguf header")
 
@@ -92,10 +94,10 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 tokens: List[str] = []
 scores: List[float] = []
 
-if Path( dir_model + "/tokenizer.model").is_file():
+if Path(dir_model + "/tokenizer.model").is_file():
     # vocab type SPIECE
-    print( "Adding sentencepiece tokenizer vocab." )
-    tokenizer = SentencePieceProcessor( dir_model + "/tokenizer.model" )
+    print("Adding sentencepiece tokenizer vocab.")
+    tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
     # output vocab_size followed by all piece/score pairs
     outbytes: bytes
@@ -118,14 +120,14 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
             text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
         score: float = tokenizer.get_score(i)
 
-        tokens.append( str(text) );
-        scores.append( score );
+        tokens.append(str(text))
+        scores.append(score)
 
 print("write gguf tokens")
 
-gguf_writer.write_string("tokenizer.ggml.model", "llama")
-gguf_writer.write_array("tokenizer.ggml.tokens",tokens)
-gguf_writer.write_array("tokenizer.ggml.scores",scores)
+gguf_writer.write_tokenizer_model("llama")
+gguf_writer.write_token_list(tokens)
+gguf_writer.write_token_scores(scores)
 
 # TENSORS
 
@@ -142,7 +144,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
     # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute( data, hparams["num_attention_heads"] )
+        data = permute(data, hparams["num_attention_heads"])
 
     # chnage tensor name
 
@@ -197,10 +199,10 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         print("  Skip tensor: " + name)
         continue
 
-    ## permute these
+    # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
         print("  Permute tensor: " + name)
-        data = permute( data, hparams["num_attention_heads"] )
+        data = permute(data, hparams["num_attention_heads"])
 
     n_dims = len(data.shape)
 
@@ -221,7 +223,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
             data = data.astype(np.float32)
             ftype_cur = 0
 
-    gguf_writer.write_tensor_padding()
     gguf_writer.write_tensor(data)
 
 gguf_writer.close()

From 8ad7cd49fb0979aeafa39e42ce5f335a4ed53e35 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 16:47:00 +0200
Subject: [PATCH 034/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 439a6da307f88..d36e6da9a30ce 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -120,7 +120,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
             text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
         score: float = tokenizer.get_score(i)
 
-        tokens.append(str(text))
+        tokens.append(text)
         scores.append(score)
 
 print("write gguf tokens")
@@ -184,6 +184,22 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
                 name = "layers." + str(i) + ".feed_forward.w3.weight"
                 break
 
+    n_dims = len(data.shape)
+
+    # ftype == 0 -> float32, ftype == 1 -> float16
+    ftype_cur = 0
+    if ftype != 0:
+        if name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+            ftype_cur = 1
+        else:
+            data = data.astype(np.float32)
+            ftype_cur = 0
+    else:
+        if data.dtype != np.float32:
+            data = data.astype(np.float32)
+            ftype_cur = 0
+
     gguf_writer.write_tensor_info(name, data)
 
 

From 0f5e57f01d8913457f954e9b4d1ff9b9471b375a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 29 Jul 2023 19:56:06 +0300
Subject: [PATCH 035/242] gguf : handle already encoded string

---
 gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index e0b4b88fb11b8..88d7e43fb7fd3 100644
--- a/gguf.py
+++ b/gguf.py
@@ -45,7 +45,7 @@ class GGUFValueType(IntEnum):
 
     @staticmethod
     def get_type(val):
-        if isinstance(val, str):
+        if isinstance(val, str) or isinstance(val, bytes):
             return GGUFValueType.STRING
         elif isinstance(val, list):
             return GGUFValueType.ARRAY
@@ -143,7 +143,7 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: boo
         elif vtype == GGUFValueType.BOOL:
             self.fout.write(struct.pack("?", val))
         elif vtype == GGUFValueType.STRING:
-            encoded_val = val.encode("utf8")
+            encoded_val = val.encode("utf8") if isinstance(val, str) else val
             self.fout.write(struct.pack("<I", len(encoded_val)))
             self.fout.write(encoded_val)
         elif vtype == GGUFValueType.ARRAY:

From 34469b9ea79e453445bf3434c534ccd51acb8abe Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 20:36:06 +0200
Subject: [PATCH 036/242] ggml.h : get array str and f32

---
 ggml.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/ggml.h b/ggml.h
index a72c82a333cf5..0296508517e8e 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1657,6 +1657,9 @@ extern "C" {
     GGML_API const char * gguf_get_key (struct gguf_context * ctx, int i);
     GGML_API void         gguf_get_val (struct gguf_context * ctx, int i, void * val);
 
+    GGML_API const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i);
+    GGML_API float        gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i);
+
     GGML_API uint8_t      gguf_get_val_u8  (struct gguf_context * ctx, int i);
     GGML_API int8_t       gguf_get_val_i8  (struct gguf_context * ctx, int i);
     GGML_API uint16_t     gguf_get_val_u16 (struct gguf_context * ctx, int i);

From 2c22e3bcdb25d718761a2c2e69ef1dd894f747fa Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 20:37:47 +0200
Subject: [PATCH 037/242] ggml.c : get arr str and f32

---
 ggml.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/ggml.c b/ggml.c
index d402daa678b90..af3a9ce71c664 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18768,6 +18768,16 @@ enum gguf_type gguf_get_type(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].type;
 }
 
+const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
+    struct gguf_kv * kv = &ctx->header.kv[key_id];
+    struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];
+    return str->data;
+}
+
+float gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i) {
+    return ((float *) ctx->header.kv[key_id].value.arr.data)[i];
+}
+
 uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].value.uint8;
 }

From 9577821487909aac1ee33654ac0e9ade9a63f886 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 21:29:07 +0200
Subject: [PATCH 038/242] gguf.py : support any type

---
 gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index 88d7e43fb7fd3..7e15e5a777d16 100644
--- a/gguf.py
+++ b/gguf.py
@@ -270,10 +270,10 @@ def write_rope_scale(self, llm: str, value:  float):
     def write_tokenizer_model(self, model: str):
         self.write_string(constants.KEY_TOKENIZER_MODEL, model)
 
-    def write_token_list(self, tokens: List[str]):
+    def write_token_list(self, tokens: List):
         self.write_array(constants.KEY_TOKENIZER_LIST, tokens)
 
-    def write_token_scores(self, scores: List[float]):
+    def write_token_scores(self, scores: List:
         self.write_array(constants.KEY_TOKENIZER_SCORES, scores)
 
 

From 06c3e4a1a7eec3db5ef69c9f31dfa55b22c4e778 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 29 Jul 2023 21:38:01 +0200
Subject: [PATCH 039/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index d36e6da9a30ce..ba9e9f677305d 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -91,11 +91,13 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 # TOKENIZATION
 
+print("write gguf tokenizer")
+
 tokens: List[str] = []
 scores: List[float] = []
 
 if Path(dir_model + "/tokenizer.model").is_file():
-    # vocab type SPIECE
+    # vocab type sentencepiece
     print("Adding sentencepiece tokenizer vocab.")
     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
@@ -123,15 +125,12 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         tokens.append(text)
         scores.append(score)
 
-print("write gguf tokens")
-
 gguf_writer.write_tokenizer_model("llama")
 gguf_writer.write_token_list(tokens)
 gguf_writer.write_token_scores(scores)
 
 # TENSORS
 
-
 # tensor info
 print("write gguf tensor info")
 

From 32e037ffbe7af7be1bb620b6fe64861fe6de2843 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sun, 30 Jul 2023 01:01:13 +0300
Subject: [PATCH 040/242] gguf : fix set is not subscriptable

---
 gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index 7e15e5a777d16..3bcc15f8bc755 100644
--- a/gguf.py
+++ b/gguf.py
@@ -149,7 +149,7 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: boo
         elif vtype == GGUFValueType.ARRAY:
             ltype = set([GGUFValueType.get_type(item) for item in val])
             assert len(ltype) == 1, "All items in a GGUF array should be of the same type"
-            self.fout.write(struct.pack("<I", ltype[0]))
+            self.fout.write(struct.pack("<I", list(ltype)[0]))
             self.fout.write(struct.pack("<I", len(val)))
             for item in val:
                 self.write_val(item, write_vtype=False)
@@ -273,7 +273,7 @@ def write_tokenizer_model(self, model: str):
     def write_token_list(self, tokens: List):
         self.write_array(constants.KEY_TOKENIZER_LIST, tokens)
 
-    def write_token_scores(self, scores: List:
+    def write_token_scores(self, scores: List[float]):
         self.write_array(constants.KEY_TOKENIZER_SCORES, scores)
 
 

From 87c34e4dd4f546bc40ad567d2c4e3672e76e6fad Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sun, 30 Jul 2023 01:09:22 +0300
Subject: [PATCH 041/242] gguf : update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index ba9e9f677305d..0ea5e21aa0002 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -5,13 +5,13 @@
 import struct
 import json
 import numpy as np
-from typing import List
+from typing import Any, List
 from pathlib import Path
 from transformers import AutoModelForCausalLM
 from sentencepiece import SentencePieceProcessor
 
 
-NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+NDArray = np.ndarray[Any, Any]
 
 
 def permute(weights: NDArray, n_head: int) -> NDArray:

From 0790c121aad79e35e89413d5031818413944d6d4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 14:46:36 +0200
Subject: [PATCH 042/242] constants.py : add layer norm eps

---
 constants.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/constants.py b/constants.py
index 024bf7b03bbe1..14f11123b1280 100644
--- a/constants.py
+++ b/constants.py
@@ -27,6 +27,8 @@
 KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
 KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
 KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
+KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
+KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
 
 # RoPE
 KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"

From ccd81a751bfd6f313d5bea7ea20cd2eee3ee53b0 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 14:48:14 +0200
Subject: [PATCH 043/242] gguf.py : add layer norm eps and merges

---
 gguf.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/gguf.py b/gguf.py
index 3bcc15f8bc755..99ea817028c70 100644
--- a/gguf.py
+++ b/gguf.py
@@ -260,6 +260,14 @@ def write_clamp_kqv(self, llm: str, value: float):
         self.write_float32(
             constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
 
+    def write_layer_norm_eps(self, llm: str, value: float):
+        self.write_float32(
+            constants.KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
+
+    def write_layer_norm_rms_eps(self, llm: str, value: float):
+        self.write_float32(
+            constants.KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
+
     def write_rope_dimension_count(self, llm: str, count: int):
         self.write_uint32(
             constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
@@ -273,6 +281,9 @@ def write_tokenizer_model(self, model: str):
     def write_token_list(self, tokens: List):
         self.write_array(constants.KEY_TOKENIZER_LIST, tokens)
 
+    def write_token_merges(self, merges: List):
+        self.write_array(constants.KEY_TOKENIZER_MERGES, merges)
+
     def write_token_scores(self, scores: List[float]):
         self.write_array(constants.KEY_TOKENIZER_SCORES, scores)
 

From b4676ee4479985741b17d9f0323d8d229fff285e Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 14:51:37 +0200
Subject: [PATCH 044/242] ggml.h : increase GGML_MAX_NAME to 64

---
 ggml.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml.h b/ggml.h
index 0296508517e8e..f939c30672387 100644
--- a/ggml.h
+++ b/ggml.h
@@ -198,7 +198,7 @@
 #define GGML_MAX_PARAMS        256
 #define GGML_MAX_CONTEXTS      64
 #define GGML_MAX_SRC           6
-#define GGML_MAX_NAME          48
+#define GGML_MAX_NAME          64
 #define GGML_MAX_OP_PARAMS     32
 #define GGML_DEFAULT_N_THREADS 4
 

From b19c11750b20877561ddbec93a3c1f3bdd655d08 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 14:58:50 +0200
Subject: [PATCH 045/242] ggml.c : add gguf_get_arr_n

---
 ggml.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/ggml.c b/ggml.c
index af3a9ce71c664..1b2f071e71a3d 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18778,6 +18778,10 @@ float gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i) {
     return ((float *) ctx->header.kv[key_id].value.arr.data)[i];
 }
 
+int gguf_get_arr_n(struct gguf_context * ctx, int i) {
+    return ctx->header.kv[i].value.arr.n;
+}
+
 uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].value.uint8;
 }

From 4ed98bf1ab5766a844df8bca92879ee793b7cf50 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 15:01:47 +0200
Subject: [PATCH 046/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 0ea5e21aa0002..926dd265b4a4c 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -87,6 +87,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
 gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
 gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparams["rms_norm_eps"])
+gguf_writer.write_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
 
 
 # TOKENIZATION

From e9192b0135c02201bc03702c16f12b27a5ab580d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 15:05:37 +0200
Subject: [PATCH 047/242] add gptneox gguf example

---
 convert-gptneox-h5-to-gguf.py | 173 ++++++++++
 gptneox-common.cpp            | 601 ++++++++++++++++++++++++++++++++++
 gptneox-common.h              | 125 +++++++
 3 files changed, 899 insertions(+)
 create mode 100644 convert-gptneox-h5-to-gguf.py
 create mode 100644 gptneox-common.cpp
 create mode 100644 gptneox-common.h

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
new file mode 100644
index 0000000000000..cbade7079be8b
--- /dev/null
+++ b/convert-gptneox-h5-to-gguf.py
@@ -0,0 +1,173 @@
+# Quick and dirty HF gptneox--> gguf conversion
+
+import gguf
+import sys
+import struct
+import json
+import numpy as np
+from typing import Any, List
+from pathlib import Path
+from transformers import AutoModelForCausalLM
+
+
+if len(sys.argv) < 3:
+    print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
+    print("  ftype == 0 -> float32")
+    print("  ftype == 1 -> float16")
+    sys.exit(1)
+
+
+# output in the same directory as the model
+dir_model = sys.argv[1]
+fname_out = sys.argv[1] + "/ggml-model.bin"
+
+
+# possible tensor data types
+#   ftype == 0 -> float32
+#   ftype == 1 -> float16
+#
+# map from ftype to string
+ftype_str = ["f32", "f16"]
+
+ftype = 1
+if len(sys.argv) > 2:
+    ftype = int(sys.argv[2])
+    if ftype < 0 or ftype > 1:
+        print("Invalid ftype: " + str(ftype))
+        sys.exit(1)
+    fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+
+model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
+list_vars = model.state_dict()
+
+# count tensors to be converted
+tensor_count = 0
+for name in list_vars.keys():
+    # we don't need these
+    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
+        continue
+    tensor_count += 1
+
+gguf_writer = gguf.GGUFWriter.open(fname_out)
+
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+# This mmust be changed when adding/deleting kv
+kv_count = 14
+
+print("tensors " + str(tensor_count) + " kv " + str(kv_count))
+
+print("write gguf header")
+
+gguf_writer.write_header(tensor_count, kv_count)
+
+print("write gguf hparams")
+
+llm_arch = "gptneox"
+
+gguf_writer.write_name("pythia-70b-deduped")
+gguf_writer.write_description("gguf test model")
+gguf_writer.write_architecture(llm_arch)
+gguf_writer.write_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.write_embedding_length(llm_arch, hparams["hidden_size"])
+gguf_writer.write_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
+gguf_writer.write_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
+gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
+gguf_writer.write_parallel_residual(llm_arch, hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
+gguf_writer.write_layer_norm_eps(llm_arch, hparams["layer_norm_eps"])
+
+# TOKENIZATION
+
+print("write gguf tokenizer")
+
+tokens: List[str] = []
+merges: List[str] = []
+
+if Path(dir_model + "/tokenizer.json").is_file():
+    # vocab type gpt2
+    print("Adding gpt2 tokenizer vocab")
+
+    with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
+        tokenizer = json.load(f)
+
+    for key in tokenizer["model"]["vocab"]:
+        tokens.append(key)
+
+    merges = tokenizer["model"]["merges"]
+
+gguf_writer.write_tokenizer_model("gpt2")
+gguf_writer.write_token_list(tokens)
+gguf_writer.write_token_merges(merges)
+
+# TENSORS
+
+# tensor info
+print("write gguf tensor info")
+
+for name in list_vars.keys():
+    data = list_vars[name].squeeze().numpy()
+
+    # we don't need these
+    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
+        continue
+
+    n_dims = len(data.shape)
+
+    # ftype == 0 -> float32, ftype == 1 -> float16
+    ftype_cur = 0
+    if ftype != 0:
+        if name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+            ftype_cur = 1
+        else:
+            data = data.astype(np.float32)
+            ftype_cur = 0
+    else:
+        if data.dtype != np.float32:
+            data = data.astype(np.float32)
+            ftype_cur = 0
+
+    gguf_writer.write_tensor_info(name, data)
+
+
+# tensor data
+print("write gguf tensor data")
+
+for name in list_vars.keys():
+    data = list_vars[name].squeeze().numpy()
+    print("Process tensor: " + name + " with shape: ", data.shape)
+
+    # we don't need these
+    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
+        print("  Skip tensor: " + name)
+        continue
+
+    n_dims = len(data.shape)
+
+    # ftype == 0 -> float32, ftype == 1 -> float16
+    ftype_cur = 0
+    if ftype != 0:
+        if name.endswith(".weight") and n_dims == 2:
+            print("  Converting to float16")
+            data = data.astype(np.float16)
+            ftype_cur = 1
+        else:
+            print("  Converting to float32")
+            data = data.astype(np.float32)
+            ftype_cur = 0
+    else:
+        if data.dtype != np.float32:
+            print("  Converting to float32")
+            data = data.astype(np.float32)
+            ftype_cur = 0
+
+    gguf_writer.write_tensor(data)
+
+gguf_writer.close()
+
+
+print("Done. Output file: " + fname_out)
+print("")
diff --git a/gptneox-common.cpp b/gptneox-common.cpp
new file mode 100644
index 0000000000000..9dee0cb9ce8b2
--- /dev/null
+++ b/gptneox-common.cpp
@@ -0,0 +1,601 @@
+#include "gptneox-common.h"
+
+#include <cmath>
+#include <cstring>
+#include <fstream>
+#include <regex>
+#include <locale>
+#include <codecvt>
+#include <sstream>
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+// Function to check if the next argument exists
+std::string get_next_arg(int& i, int argc, char** argv, const std::string& flag, gpt_params& params) {
+    if (i + 1 < argc && argv[i + 1][0] != '-') {
+        return argv[++i];
+    } else {
+        fprintf(stderr, "error: %s requires one argument.\n", flag.c_str());
+        gpt_print_usage(argc, argv, params);
+        exit(0);
+    }
+}
+
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
+    for (int i = 1; i < argc; i++) {
+        std::string arg = argv[i];
+
+        if (arg == "-s" || arg == "--seed") {
+            params.seed = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-t" || arg == "--threads") {
+            params.n_threads = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-ngl" || arg == "--gpu-layers" || arg == "--n-gpu-layers") {
+            params.n_gpu_layers = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-p" || arg == "--prompt") {
+            params.prompt = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-n" || arg == "--n_predict") {
+            params.n_predict = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_k") {
+            params.top_k = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_p") {
+            params.top_p = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--temp") {
+            params.temp = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-last-n") {
+            params.repeat_last_n = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-penalty") {
+            params.repeat_penalty = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-b" || arg == "--batch_size") {
+            params.n_batch= std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-m" || arg == "--model") {
+            params.model = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-i" || arg == "--interactive") {
+            params.interactive = true;
+        } else if (arg == "-ip" || arg == "--interactive-port") {
+            params.interactive = true;
+            params.interactive_port = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-h" || arg == "--help") {
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        } else if (arg == "-f" || arg == "--file") {
+            get_next_arg(i, argc, argv, arg, params);
+            std::ifstream file(argv[i]);
+            if (!file) {
+                fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
+                break;
+            }
+            std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
+            if (params.prompt.back() == '\n') {
+                params.prompt.pop_back();
+            }
+        } else if (arg == "-tt" || arg == "--token_test") {
+            params.token_test = get_next_arg(i, argc, argv, arg, params);
+        }
+        else {
+            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        }
+    }
+
+    return true;
+}
+
+void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
+    fprintf(stderr, "usage: %s [options]\n", argv[0]);
+    fprintf(stderr, "\n");
+    fprintf(stderr, "options:\n");
+    fprintf(stderr, "  -h, --help            show this help message and exit\n");
+    fprintf(stderr, "  -s SEED, --seed SEED  RNG seed (default: -1)\n");
+    fprintf(stderr, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
+    fprintf(stderr, "  -ngl N, --gpu-layers N  number of layers to offload to GPU on supported models (default: %d)\n", params.n_gpu_layers);
+    fprintf(stderr, "  -p PROMPT, --prompt PROMPT\n");
+    fprintf(stderr, "                        prompt to start generation with (default: random)\n");
+    fprintf(stderr, "  -f FNAME, --file FNAME\n");
+    fprintf(stderr, "                        load prompt from a file\n");
+    fprintf(stderr, "  -tt TOKEN_TEST, --token_test TOKEN_TEST\n");
+    fprintf(stderr, "                        test tokenization\n");
+    fprintf(stderr, "  -n N, --n_predict N   number of tokens to predict (default: %d)\n", params.n_predict);
+    fprintf(stderr, "  --top_k N             top-k sampling (default: %d)\n", params.top_k);
+    fprintf(stderr, "  --top_p N             top-p sampling (default: %.1f)\n", params.top_p);
+    fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
+    fprintf(stderr, "  --repeat-last-n N     last n tokens to consider for penalize (default: %d, 0 = disabled)\n", params.repeat_last_n);
+    fprintf(stderr, "  --repeat-penalty N    penalize repeat sequence of tokens (default: %.2f, 1.0 = disabled)\n", (double)params.repeat_penalty);
+    fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
+    fprintf(stderr, "  -m FNAME, --model FNAME\n");
+    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
+    fprintf(stderr, "\n");
+}
+
+std::string gpt_random_prompt(std::mt19937 & rng) {
+    const int r = rng() % 10;
+    switch (r) {
+        case 0: return "So";
+        case 1: return "Once upon a time";
+        case 2: return "When";
+        case 3: return "The";
+        case 4: return "After";
+        case 5: return "If";
+        case 6: return "import";
+        case 7: return "He";
+        case 8: return "She";
+        case 9: return "They";
+        default: return "To";
+    }
+
+    return "The";
+}
+
+std::string trim(const std::string & s) {
+    std::regex e("^\\s+|\\s+$");
+    return std::regex_replace(s, e, "");
+}
+
+std::string replace(const std::string & s, const std::string & from, const std::string & to) {
+    std::string result = s;
+    size_t pos = 0;
+    while ((pos = result.find(from, pos)) != std::string::npos) {
+        result.replace(pos, from.length(), to);
+        pos += to.length();
+    }
+    return result;
+}
+
+void gpt_vocab::add_special_token(const std::string & token) {
+    special_tokens.push_back(token);
+}
+
+std::map<std::string, int32_t> json_parse(const std::string & fname) {
+    std::map<std::string, int32_t> result;
+
+    // read file into string
+    std::string json;
+    {
+        std::ifstream ifs(fname);
+        if (!ifs) {
+            fprintf(stderr, "Failed to open %s\n", fname.c_str());
+            exit(1);
+        }
+
+        json = std::string((std::istreambuf_iterator<char>(ifs)),
+                (std::istreambuf_iterator<char>()));
+    }
+
+    if (json[0] != '{') {
+        return result;
+    }
+
+    // parse json
+    {
+        bool has_key  = false;
+        bool in_token = false;
+
+        std::string str_key = "";
+        std::string str_val = "";
+
+        int n = json.size();
+        for (int i = 1; i < n; ++i) {
+            if (!in_token) {
+                if (json[i] == ' ') continue;
+                if (json[i] == '"') {
+                    in_token = true;
+                    continue;
+                }
+            } else {
+                if (json[i] == '\\' && i+1 < n) {
+                    if (has_key == false) {
+                        str_key += json[i];
+                    } else {
+                        str_val += json[i];
+                    }
+                    ++i;
+                } else if (json[i] == '"') {
+                    if (has_key == false) {
+                        has_key = true;
+                        ++i;
+                        while (json[i] == ' ') ++i;
+                        ++i; // :
+                        while (json[i] == ' ') ++i;
+                        if (json[i] != '\"') {
+                            while (json[i] != ',' && json[i] != '}') {
+                                str_val += json[i++];
+                            }
+                            has_key = false;
+                        } else {
+                            in_token = true;
+                            continue;
+                        }
+                    } else {
+                        has_key = false;
+                    }
+
+                    str_key = ::replace(str_key, "\\u0120", " " ); // \u0120 -> space
+                    str_key = ::replace(str_key, "\\u010a", "\n"); // \u010a -> new line
+                    str_key = ::replace(str_key, "\\\"",    "\""); // \\\"   -> "
+
+                    try {
+                        result[str_key] = std::stoi(str_val);
+                    } catch (...) {
+                        //fprintf(stderr, "%s: ignoring key '%s' with value '%s'\n", fname.c_str(), str_key.c_str(), str_val.c_str());
+
+                    }
+                    str_key = "";
+                    str_val = "";
+                    in_token = false;
+                    continue;
+                }
+                if (has_key == false) {
+                    str_key += json[i];
+                } else {
+                    str_val += json[i];
+                }
+            }
+        }
+    }
+
+    return result;
+}
+
+std::string convert_to_utf8(const std::wstring & input) {
+    std::wstring_convert<std::codecvt_utf8<wchar_t>> converter;
+    return converter.to_bytes(input);
+}
+
+
+std::wstring convert_to_wstring(const std::string & input) {
+    std::wstring_convert<std::codecvt_utf8<wchar_t>> converter;
+    return converter.from_bytes(input);
+}
+
+void gpt_split_words(std::string str, std::vector<std::string>& words) {
+    const std::string pattern = R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)";
+    const std::regex re(pattern);
+    std::smatch m;
+
+    while (std::regex_search(str, m, re)) {
+        for (auto x : m) {
+            words.push_back(x);
+        }
+        str = m.suffix();
+    }
+}
+
+std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text) {
+    std::vector<std::string> words;
+
+    // first split the text into words
+    {
+        std::string str = text;
+
+        // Generate the subpattern from the special_tokens vector if it's not empty
+        if (!vocab.special_tokens.empty()) {
+            const std::regex escape(R"([\[\\\^\$\.\|\?\*\+\(\)\{\}])");
+            std::string special_tokens_subpattern;
+            for (const auto & token : vocab.special_tokens) {
+                if (!special_tokens_subpattern.empty()) {
+                    special_tokens_subpattern += "|";
+                }
+                special_tokens_subpattern += std::regex_replace(token, escape, R"(\$&)");
+            }
+
+            std::regex re(special_tokens_subpattern);
+            std::smatch m;
+            // Split the text by special tokens.
+            while (std::regex_search(str, m, re)) {
+                // Split the substrings in-between special tokens into words.
+                gpt_split_words(m.prefix(), words);
+                // Add matched special tokens as words.
+                for (auto x : m) {
+                    words.push_back(x);
+                }
+                str = m.suffix();
+            }
+            // Remaining text without special tokens will be handled below.
+        }
+
+        gpt_split_words(str, words);
+    }
+
+    // find the longest token that forms each word in words:
+    std::vector<gpt_vocab::id> tokens;
+    for (const auto & word : words) {
+        for (int i = 0; i < (int) word.size(); ){
+            for (int j = word.size() - 1; j >= i; j--){
+                auto cand = word.substr(i, j-i+1);
+                auto it = vocab.token_to_id.find(cand);
+                if (it != vocab.token_to_id.end()){ // word.substr(i, j-i+1) in vocab
+                    tokens.push_back(it->second);
+                    i = j + 1;
+                    break;
+                }
+                else if (j == i){ // word.substr(i, 1) has no matching
+                    fprintf(stderr, "%s: unknown token '%s'\n", __func__, word.substr(i, 1).data());
+                    i++;
+                }
+            }
+        }
+    }
+
+    return tokens;
+}
+
+std::vector<gpt_vocab::id> parse_tokens_from_string(const std::string& input, char delimiter) {
+    std::vector<gpt_vocab::id> output;
+    std::stringstream ss(input);
+    std::string token;
+
+    while (std::getline(ss, token, delimiter)) {
+        output.push_back(std::stoi(token));
+    }
+
+    return output;
+}
+
+std::map<std::string, std::vector<gpt_vocab::id>> extract_tests_from_file(const std::string & fpath_test){
+    if (fpath_test.empty()){
+        fprintf(stderr, "%s : No test file found.\n", __func__);
+        return std::map<std::string, std::vector<gpt_vocab::id>>();
+    }
+
+    std::map<std::string, std::vector<gpt_vocab::id>> tests;
+
+    auto fin = std::ifstream(fpath_test, std::ios_base::in);
+    const char * delimeter = " => ";
+    const char del_tok = ',';
+    std::string line;
+    while (std::getline(fin, line)) {
+        size_t delimiterPos = line.find(delimeter);
+        if (delimiterPos != std::string::npos) {
+            std::string text = line.substr(0, delimiterPos);
+            std::string s_tokens = line.substr(delimiterPos + std::strlen(delimeter));
+            tests[text] = parse_tokens_from_string(s_tokens, del_tok);
+        }
+    }
+    return tests;
+}
+
+void test_gpt_tokenizer(gpt_vocab & vocab, const std::string & fpath_test){
+    std::map<std::string, std::vector<gpt_vocab::id>> tests = extract_tests_from_file(fpath_test);
+
+    size_t n_fails = 0;
+
+    for (const auto & test : tests) {
+        std::vector<gpt_vocab::id> tokens = gpt_tokenize(vocab, test.first);
+
+        if (tokens != test.second){
+            n_fails++;
+
+            // print out failure cases
+            fprintf(stderr, "%s : failed test: '%s'\n", __func__, test.first.c_str());
+            fprintf(stderr, "%s : tokens in hf:   ", __func__);
+            for (const auto & t : test.second) {
+                fprintf(stderr, "%s(%d), ", vocab.id_to_token[t].c_str(), t);
+            }
+            fprintf(stderr, "\n");
+            fprintf(stderr, "%s : tokens in ggml: ", __func__);
+            for (const auto & t : tokens) {
+                fprintf(stderr, "%s(%d), ", vocab.id_to_token[t].c_str(), t);
+            }
+            fprintf(stderr, "\n");
+        }
+    }
+
+    fprintf(stderr, "%s : %zu tests failed out of %zu tests.\n", __func__, n_fails, tests.size());
+}
+
+bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab) {
+    printf("%s: loading vocab from '%s'\n", __func__, fname.c_str());
+
+    vocab.token_to_id = ::json_parse(fname);
+
+    for (const auto & kv : vocab.token_to_id) {
+        vocab.id_to_token[kv.second] = kv.first;
+    }
+
+    printf("%s: vocab size = %d\n", __func__, (int) vocab.token_to_id.size());
+
+    // print the vocabulary
+    //for (auto kv : vocab.token_to_id) {
+    //    printf("'%s' -> %d\n", kv.first.data(), kv.second);
+    //}
+
+    return true;
+}
+
+gpt_vocab::id gpt_sample_top_k_top_p(
+        const gpt_vocab & vocab,
+        const float * logits,
+        int    top_k,
+        double top_p,
+        double temp,
+        std::mt19937 & rng) {
+    int n_logits = vocab.id_to_token.size();
+
+    std::vector<std::pair<double, gpt_vocab::id>> logits_id;
+    logits_id.reserve(n_logits);
+
+    {
+        const double scale = 1.0/temp;
+        for (int i = 0; i < n_logits; ++i) {
+            logits_id.push_back(std::make_pair(logits[i]*scale, i));
+        }
+    }
+
+    // find the top K tokens
+    std::partial_sort(
+            logits_id.begin(),
+            logits_id.begin() + top_k, logits_id.end(),
+            [](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
+        return a.first > b.first;
+    });
+
+    logits_id.resize(top_k);
+
+    double maxl = -INFINITY;
+    for (const auto & kv : logits_id) {
+        maxl = std::max(maxl, kv.first);
+    }
+
+    // compute probs for the top K tokens
+    std::vector<double> probs;
+    probs.reserve(logits_id.size());
+
+    double sum = 0.0;
+    for (const auto & kv : logits_id) {
+        double p = exp(kv.first - maxl);
+        probs.push_back(p);
+        sum += p;
+    }
+
+    // normalize the probs
+    for (auto & p : probs) {
+        p /= sum;
+    }
+
+    if (top_p < 1.0f) {
+        double cumsum = 0.0f;
+        for (int i = 0; i < top_k; i++) {
+            cumsum += probs[i];
+            if (cumsum >= top_p) {
+                top_k = i + 1;
+                probs.resize(top_k);
+                logits_id.resize(top_k);
+                break;
+            }
+        }
+
+        cumsum = 1.0/cumsum;
+        for (int i = 0; i < (int) probs.size(); i++) {
+            probs[i] *= cumsum;
+        }
+    }
+
+    //printf("\n");
+    //for (int i = 0; i < (int) probs.size(); i++) {
+    //    printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
+    //}
+    //exit(0);
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    int idx = dist(rng);
+
+    return logits_id[idx].second;
+}
+
+gpt_vocab::id gpt_sample_top_k_top_p_repeat(
+        const gpt_vocab & vocab,
+        const float * logits,
+        const int32_t * last_n_tokens_data,
+        size_t last_n_tokens_data_size,
+        int    top_k,
+        double top_p,
+        double temp,
+        int repeat_last_n,
+        float repeat_penalty,
+        std::mt19937 & rng) {
+
+    int n_logits = vocab.id_to_token.size();
+
+    const auto * plogits = logits;
+
+    const auto last_n_tokens = std::vector<int32_t>(last_n_tokens_data, last_n_tokens_data + last_n_tokens_data_size);
+
+    if (temp <= 0) {
+        // select the token with the highest logit directly
+        float max_logit = plogits[0];
+        gpt_vocab::id max_id = 0;
+
+        for (int i = 1; i < n_logits; ++i) {
+            if (plogits[i] > max_logit) {
+                max_logit = plogits[i];
+                max_id = i;
+            }
+        }
+        return max_id;
+    }
+
+
+    std::vector<std::pair<double, gpt_vocab::id>> logits_id;
+    logits_id.reserve(n_logits);
+
+    {
+        const float scale = 1.0f/temp;
+        for (int i = 0; i < n_logits; ++i) {
+            // repetition penalty from ctrl paper (https://arxiv.org/abs/1909.05858)
+            // credit https://github.com/facebookresearch/llama/compare/main...shawwn:llama:main
+            if (repeat_last_n > 0 && std::find(last_n_tokens.end()-repeat_last_n, last_n_tokens.end(), i) != last_n_tokens.end()) {
+                // if score < 0 then repetition penalty has to multiplied to reduce the previous token probability
+                if (plogits[i] < 0.0f) {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale*repeat_penalty, i));
+                } else {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale/repeat_penalty, i));
+                }
+            } else {
+                logits_id.push_back(std::make_pair(plogits[i]*scale, i));
+            }
+        }
+    }
+
+    // find the top K tokens
+    std::partial_sort(
+            logits_id.begin(),
+            logits_id.begin() + top_k, logits_id.end(),
+            [](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
+        return a.first > b.first;
+    });
+
+    logits_id.resize(top_k);
+
+    double maxl = -INFINITY;
+    for (const auto & kv : logits_id) {
+        maxl = std::max(maxl, kv.first);
+    }
+
+    // compute probs for the top K tokens
+    std::vector<double> probs;
+    probs.reserve(logits_id.size());
+
+    double sum = 0.0;
+    for (const auto & kv : logits_id) {
+        double p = exp(kv.first - maxl);
+        probs.push_back(p);
+        sum += p;
+    }
+
+    // normalize the probs
+    for (auto & p : probs) {
+        p /= sum;
+    }
+
+    if (top_p < 1.0f) {
+        double cumsum = 0.0f;
+        for (int i = 0; i < top_k; i++) {
+            cumsum += probs[i];
+            if (cumsum >= top_p) {
+                top_k = i + 1;
+                probs.resize(top_k);
+                logits_id.resize(top_k);
+                break;
+            }
+        }
+
+        cumsum = 1.0/cumsum;
+        for (int i = 0; i < (int) probs.size(); i++) {
+            probs[i] *= cumsum;
+        }
+    }
+
+//    printf("\n");
+//    for (int i = 0; i < (int) probs.size(); i++) {
+//    for (int i = 0; i < 10; i++) {
+//        printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
+//    }
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    int idx = dist(rng);
+
+    return logits_id[idx].second;
+
+}
diff --git a/gptneox-common.h b/gptneox-common.h
new file mode 100644
index 0000000000000..60e5650c129ab
--- /dev/null
+++ b/gptneox-common.h
@@ -0,0 +1,125 @@
+// Various helper functions and utilities
+
+#pragma once
+
+#include <string>
+#include <map>
+#include <vector>
+#include <random>
+#include <thread>
+
+//
+// CLI argument parsing
+//
+
+struct gpt_params {
+    int32_t seed      = -1;  // RNG seed
+    int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
+    int32_t n_predict = 200; // new tokens to predict
+    int32_t n_batch   = 8;   // batch size for prompt processing
+
+    // sampling parameters
+    int32_t top_k          = 40;
+    float   top_p          = 0.9f;
+    float   temp           = 0.9f;
+    int32_t repeat_last_n  = 64;
+    float   repeat_penalty = 1.00f;
+
+    std::string model      = "models/gpt-2-117M/ggml-model.bin"; // model path
+    std::string prompt     = "";
+    std::string token_test = "";
+
+    bool    interactive      = false;
+    int32_t interactive_port = -1;
+
+    int32_t n_gpu_layers     = 0;
+};
+
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params);
+
+void gpt_print_usage(int argc, char ** argv, const gpt_params & params);
+
+std::string gpt_random_prompt(std::mt19937 & rng);
+
+//
+// Vocab utils
+//
+
+std::string trim(const std::string & s);
+
+std::string replace(
+        const std::string & s,
+        const std::string & from,
+        const std::string & to);
+
+struct gpt_vocab {
+    using id    = int32_t;
+    using token = std::string;
+
+    std::map<token, id> token_to_id;
+    std::map<id, token> id_to_token;
+    std::vector<std::string> special_tokens;
+
+    void add_special_token(const std::string & token);
+};
+
+// poor-man's JSON parsing
+std::map<std::string, int32_t> json_parse(const std::string & fname);
+
+std::string convert_to_utf8(const std::wstring & input);
+
+std::wstring convert_to_wstring(const std::string & input);
+
+void gpt_split_words(std::string str, std::vector<std::string>& words);
+
+// split text into tokens
+//
+// ref: https://github.com/openai/gpt-2/blob/a74da5d99abaaba920de8131d64da2862a8f213b/src/encoder.py#L53
+//
+// Regex (Python):
+// r"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+"""
+//
+// Regex (C++):
+// R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)"
+//
+std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text);
+
+// test outputs of gpt_tokenize
+//
+//   - compare with tokens generated by the huggingface tokenizer
+//   - test cases are chosen based on the model's main language (under 'prompt' directory)
+//   - if all sentences are tokenized identically, print 'All tests passed.'
+//   - otherwise, print sentence, huggingface tokens, ggml tokens
+//
+void test_gpt_tokenizer(gpt_vocab & vocab, const std::string & fpath_test);
+
+// load the tokens from encoder.json
+bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab);
+
+// sample next token given probabilities for each embedding
+//
+//   - consider only the top K tokens
+//   - from them, consider only the top tokens with cumulative probability > P
+//
+// TODO: not sure if this implementation is correct
+// TODO: temperature is not implemented
+//
+gpt_vocab::id gpt_sample_top_k_top_p(
+        const gpt_vocab & vocab,
+        const float * logits,
+        int    top_k,
+        double top_p,
+        double temp,
+        std::mt19937 & rng);
+
+gpt_vocab::id gpt_sample_top_k_top_p_repeat(
+        const gpt_vocab & vocab,
+        const float * logits,
+        const int32_t * last_n_tokens_data,
+        size_t last_n_tokens_data_size,
+        int    top_k,
+        double top_p,
+        double temp,
+        int repeat_last_n,
+        float repeat_penalty,
+        std::mt19937 & rng);

From f175b05872462c5a3f14f072945884acef7861f8 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 15:08:37 +0200
Subject: [PATCH 048/242] Makefile : add gptneox gguf example

---
 Makefile | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/Makefile b/Makefile
index e19acfbb24a3f..7d37b66a698ec 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test gguf
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test gguf gptneox-main
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0
@@ -329,6 +329,9 @@ grammar-parser.o: examples/grammar-parser.cpp examples/grammar-parser.h
 libllama.so: llama.o ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) -shared -fPIC -o $@ $^ $(LDFLAGS)
 
+gptneox-common.o: gptneox-common.cpp gptneox-common.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
 clean:
 	rm -vf *.o *.so *.dll main quantize quantize-stats perplexity embedding benchmark-matmult save-load-state server simple vdot train-text-from-scratch embd-input-test gguf build-info.h $(TEST_TARGETS)
 
@@ -373,6 +376,9 @@ embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-te
 gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
+gptneox-main: gptneox-main.cpp gptneox-common.o ggml.o $(OBJS)
+	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
+
 train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 

From 2fabc176ced5d5cd13458a7856fe67e9189e5b64 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 16:28:08 +0200
Subject: [PATCH 049/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 926dd265b4a4c..7c867619bc659 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -65,7 +65,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     hparams = json.load(f)
 
 # This mmust be changed when adding/deleting kv
-kv_count = 13
+kv_count = 14
 
 print("tensors " + str(tensor_count) + " kv " + str(kv_count))
 

From 30c4ea47e6930c5bf9877d7fd7c468557e960fb4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 16:59:26 +0200
Subject: [PATCH 050/242] add gptneox gguf example

---
 gptneox-main.cpp | 812 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 812 insertions(+)
 create mode 100644 gptneox-main.cpp

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
new file mode 100644
index 0000000000000..02fbc9fba35a5
--- /dev/null
+++ b/gptneox-main.cpp
@@ -0,0 +1,812 @@
+#include "ggml.h"
+
+#include "gptneox-common.h"
+
+#include <cassert>
+#include <cmath>
+#include <cstdio>
+#include <cstring>
+#include <cinttypes>
+#include <fstream>
+#include <map>
+#include <string>
+#include <vector>
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+// default hparams
+struct gpt_neox_hparams {
+    size_t n_merges = 0;
+    size_t n_vocab  = 0;
+    int32_t n_ctx    = 0;
+    int32_t n_embd   = 0;
+    int32_t n_head   = 0;
+    int32_t n_layer  = 0;
+    int32_t n_rot    = 0; // rotary_pct * (n_embd / n_head)
+    bool par_res = true;
+    float norm_eps = 1e-5;
+};
+
+struct gpt_neox_layer {
+    // pre normalization
+    struct ggml_tensor * ln_1_g;
+    struct ggml_tensor * ln_1_b;
+
+    // attention
+    struct ggml_tensor * c_attn_attn_w;
+    struct ggml_tensor * c_attn_attn_b;
+
+    struct ggml_tensor * c_attn_proj_w;
+    struct ggml_tensor * c_attn_proj_b;
+
+    // post normalization
+    struct ggml_tensor * ln_2_g;
+    struct ggml_tensor * ln_2_b;
+
+    // ff
+    struct ggml_tensor * c_mlp_fc_w;
+    struct ggml_tensor * c_mlp_fc_b;
+
+    struct ggml_tensor * c_mlp_proj_w;
+    struct ggml_tensor * c_mlp_proj_b;
+};
+
+struct gpt_neox_model {
+    gpt_neox_hparams hparams;
+
+    // normalization
+    struct ggml_tensor * ln_f_g;
+    struct ggml_tensor * ln_f_b;
+
+    struct ggml_tensor * wte; // position embedding
+
+    struct ggml_tensor * lmh_g; // language model head
+
+    std::vector<gpt_neox_layer> layers;
+
+    // key + value memory
+    struct ggml_tensor * memory_k;
+    struct ggml_tensor * memory_v;
+
+    //
+    struct gguf_context * ggufctx;
+    struct ggml_context * ctx;
+    struct ggml_context * kvctx;
+
+    std::map<std::string, struct ggml_tensor *> tensors;
+};
+
+struct ggml_tensor * get_tensor_ex( struct ggml_context * ctx, std::string name){
+
+    struct ggml_tensor * cur = ggml_get_tensor(ctx, name.c_str());
+    if( cur == NULL ) {
+        fprintf(stdout, "%s: tensor '%s' not found!\n", __func__, name.c_str());
+    } else {
+//        fprintf(stdout, "%s: n_dims = %d, name = '%s'\n", __func__, cur->n_dims, cur->name);
+    }
+
+    return cur;
+}
+
+// load the model's weights from a file
+bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_vocab & vocab) {
+    printf("%s: loading model from '%s'..\n", __func__, fname.c_str());
+
+    model.ctx = NULL;
+
+    struct gguf_init_params ggufparams = {
+        /*.no_alloc = */ false,
+        /*.ctx      = */ &model.ctx,
+    };
+
+    auto & ggufctx = model.ggufctx;
+
+    ggufctx  = gguf_init_from_file(fname.c_str(), ggufparams);
+
+    if (!ggufctx) {
+        fprintf(stderr, "%s: gguf_init_from_file() failed\n", __func__);
+        return false;
+    }
+
+    fprintf(stdout, "%s: gguf version     = %d\n", __func__, gguf_get_version(ggufctx));
+    fprintf(stdout, "%s: gguf alignment   = %zu\n", __func__, gguf_get_alignment(ggufctx));
+    fprintf(stdout, "%s: gguf data offset = %zu\n", __func__, gguf_get_data_offset(ggufctx));
+
+    // print all kv
+    if( false )
+    {
+        const int n_kv = gguf_get_n_kv(ggufctx);
+
+        fprintf(stdout, "%s: n_kv: %d\n", __func__, n_kv);
+
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(ggufctx, i);
+
+            fprintf(stdout, "%s: kv[%d]: key = %s\n", __func__, i, key);
+        }
+    }
+
+    // print some standard metadata
+    {
+        int keyidx;
+
+        keyidx = gguf_find_key(ggufctx, "general.name");
+        if (keyidx != -1) { fprintf(stdout, "%s: model name         = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.description");
+        if (keyidx != -1) { fprintf(stdout, "%s: model description  = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.author");
+        if (keyidx != -1) { fprintf(stdout, "%s: model author       = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.license");
+        if (keyidx != -1) { fprintf(stdout, "%s: model license      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.architecture");
+        if (keyidx != -1) { fprintf(stdout, "%s: model architecture = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+    }
+
+    // check required metadata
+    {
+        int keyidx;
+
+        keyidx = gguf_find_key(ggufctx, "general.architecture");
+        if (keyidx != -1) {
+            if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "gptneox") != 0) {
+                fprintf(stdout, "%s: model architecture not supported!\n", __func__);
+                return false;
+            }
+        } else {
+            fprintf(stdout, "%s: gguf model architecture not found!\n", __func__);
+            return false;
+        }
+
+    }
+
+    // load hparams
+    {
+        auto & hparams = model.hparams;
+
+        bool ok = true;
+        int keyidx;
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.context_length");
+                  if (keyidx != -1) { hparams.n_ctx = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.embedding_length");
+                  if (keyidx != -1) { hparams.n_embd = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.attention.head_count");
+                  if (keyidx != -1) { hparams.n_head = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.layer_count");
+                  if (keyidx != -1) { hparams.n_layer = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.rope.dimension_count");
+                  if (keyidx != -1) { hparams.n_rot = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.use_parallel_residual");
+                  if (keyidx != -1) { hparams.par_res = gguf_get_val_bool(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.attention.layer_norm_epsilon");
+                  if (keyidx != -1) { hparams.norm_eps= gguf_get_val_f32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (!ok) {
+            fprintf(stderr, "%s: required hparam missing!\n", __func__);
+            return false;
+        }
+
+        printf("%s: n_ctx    = %d\n", __func__, hparams.n_ctx);
+        printf("%s: n_embd   = %d\n", __func__, hparams.n_embd);
+        printf("%s: n_head   = %d\n", __func__, hparams.n_head);
+        printf("%s: n_layer  = %d\n", __func__, hparams.n_layer);
+        printf("%s: n_rot    = %d\n", __func__, hparams.n_rot);
+        printf("%s: par_res  = %d\n", __func__, hparams.par_res);
+        printf("%s: norm_eps = %g\n", __func__, hparams.norm_eps);
+
+    }
+
+    // load vocab
+    {
+
+        // TODO: implement a better bpe tokenizer, utilizing merges and handles unicode
+
+        auto & hparams = model.hparams;
+
+        int keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.model");
+
+        if (keyidx != -1) {
+            if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "gpt2") != 0) {
+                fprintf(stdout, "%s: tokenizer model not supported!\n", __func__);
+                return false;
+            }
+        } else {
+            fprintf(stdout, "%s: tokenizer model not found!\n", __func__);
+            return false;
+        }
+
+
+        int tokens_keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.tokens");
+
+        if (tokens_keyidx == -1) {
+            fprintf(stdout, "%s: gpt2 tokenizer vocab not found!\n", __func__);
+            return false;
+        }
+
+        int merges_keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.merges");
+
+        if (merges_keyidx == -1) {
+            fprintf(stdout, "%s: gpt2 tokenizer merges not found!\n", __func__);
+            return false;
+        }
+
+        hparams.n_vocab = gguf_get_arr_n(ggufctx,tokens_keyidx);
+        hparams.n_merges = gguf_get_arr_n(ggufctx,merges_keyidx);
+
+        fprintf(stdout, "%s: gpt2 tokenizer vocab  = %zu\n", __func__, hparams.n_vocab);
+        fprintf(stdout, "%s: gpt2 tokenizer merges = %zu\n", __func__, hparams.n_merges);
+
+        for (size_t i = 0; i < hparams.n_vocab; i++) {
+            std::string word = gguf_get_arr_str(ggufctx, tokens_keyidx, i);
+
+
+            // TEMP until a better bpe tokenizer is implemented
+            word = replace(word, "Ġ", " ");
+            word = replace(word, "Ċ", "\n");
+
+
+            vocab.token_to_id[word] = i;
+            vocab.id_to_token[i] = word;
+        }
+
+
+    }
+
+
+    auto & ctx = model.ctx;
+    size_t ctx_size = ggml_get_mem_size(ctx);
+
+    printf("%s: ggml ctx size = %6.2f MB\n", __func__, ctx_size/(1024.0*1024.0));
+
+    // print tensor info
+    if( false )
+    {
+        const int n_tensors = gguf_get_n_tensors(ggufctx);
+
+        fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            const char * name   = gguf_get_tensor_name  (ggufctx, i);
+            const size_t offset = gguf_get_tensor_offset(ggufctx, i);
+
+            fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
+        }
+    }
+
+
+    // prepare memory for the weights
+    {
+        const int n_layer = model.hparams.n_layer;
+
+        model.layers.resize(n_layer);
+
+        model.wte    = ggml_get_tensor(ctx, "gpt_neox.embed_in.weight");
+        model.ln_f_g = ggml_get_tensor(ctx, "gpt_neox.final_layer_norm.weight");
+        model.ln_f_b = ggml_get_tensor(ctx, "gpt_neox.final_layer_norm.bias");
+        model.lmh_g  = ggml_get_tensor(ctx, "embed_out.weight");
+
+        // map by name
+        model.tensors["gpt_neox.embed_in.weight"] = model.wte;
+        model.tensors["gpt_neox.final_layer_norm.weight"] = model.ln_f_g;
+        model.tensors["gpt_neox.final_layer_norm.bias"]   = model.ln_f_b;
+        model.tensors["embed_out.weight"] = model.lmh_g;
+
+        for (int i = 0; i < n_layer; ++i) {
+            auto & layer = model.layers[i];
+
+            layer.ln_1_g          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".input_layernorm.weight" );
+            layer.ln_1_b          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".input_layernorm.bias" );
+
+            layer.c_attn_attn_w   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.weight" );
+            layer.c_attn_attn_b   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.bias" );
+
+            layer.c_attn_proj_w   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.dense.weight" );
+            layer.c_attn_proj_b   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.dense.bias" );
+
+            layer.ln_2_g          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.weight" );
+            layer.ln_2_b          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.bias");
+
+            layer.c_mlp_fc_w      = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.weight" );
+            layer.c_mlp_fc_b      = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.bias" );
+
+            layer.c_mlp_proj_w    = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.weight" );
+            layer.c_mlp_proj_b    = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.bias" );
+
+            // map by name
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".input_layernorm.weight"] = layer.ln_1_g;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".input_layernorm.bias"]   = layer.ln_1_b;
+
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.weight"] = layer.c_attn_attn_w;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.bias"]   = layer.c_attn_attn_b;
+
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.dense.weight"] = layer.c_attn_proj_w;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.dense.bias"]   = layer.c_attn_proj_b;
+
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.weight"] = layer.ln_2_g;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.bias"]   = layer.ln_2_b;
+
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.weight"] = layer.c_mlp_fc_w;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.bias"]   = layer.c_mlp_fc_b;
+
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.weight"] = layer.c_mlp_proj_w;
+            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.bias"]   = layer.c_mlp_proj_b;
+        }
+    }
+
+    // key + value memory
+    {
+        const auto & kvctx = model.kvctx;
+        const auto & hparams = model.hparams;
+
+        const int n_embd  = hparams.n_embd;
+        const int n_layer = hparams.n_layer;
+        const int n_ctx   = hparams.n_ctx;
+
+        const int64_t n_mem      = n_layer*n_ctx;
+        const int64_t n_elements = n_embd*n_mem;
+
+        // create the ggml context
+        {
+            struct ggml_init_params params = {
+                /*.mem_size   =*/ size_t(n_elements*4+ggml_tensor_overhead()*2),
+                /*.mem_buffer =*/ NULL,
+                /*.no_alloc   =*/ false,
+            };
+
+            model.kvctx = ggml_init(params);
+            if (!model.kvctx) {
+                fprintf(stderr, "%s: kv ggml_init() failed\n", __func__);
+                return false;
+            }
+
+        }
+
+
+        model.memory_k = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_elements);
+        model.memory_v = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_elements);
+
+        const size_t memory_size = ggml_nbytes(model.memory_k) + ggml_nbytes(model.memory_v);
+
+        printf("%s: memory_size = %8.2f MB, n_mem = %" PRId64 "\n", __func__, memory_size/1024.0/1024.0, n_mem);
+    }
+
+    return true;
+}
+
+
+// feed-forward network
+ggml_tensor * gpt_neox_ff(
+        const gpt_neox_layer &layer,
+        ggml_context * ctx0,
+        ggml_tensor * inp) {
+    ggml_tensor * cur = ggml_norm(ctx0, inp);
+
+    cur = ggml_add(ctx0,
+        ggml_mul(ctx0,
+            ggml_repeat(ctx0, layer.ln_2_g, cur),
+            cur),
+        ggml_repeat(ctx0, layer.ln_2_b, cur));
+
+    cur = ggml_mul_mat(ctx0,
+            layer.c_mlp_fc_w,
+            cur);
+
+    cur = ggml_add(ctx0,
+            ggml_repeat(ctx0, layer.c_mlp_fc_b, cur),
+            cur);
+
+    // GELU activation
+    cur = ggml_gelu(ctx0, cur);
+
+    // projection
+    // cur = proj_w*cur + proj_b
+    cur = ggml_mul_mat(ctx0,
+            layer.c_mlp_proj_w,
+            cur);
+
+    cur = ggml_add(ctx0,
+            ggml_repeat(ctx0, layer.c_mlp_proj_b, cur),
+            cur);
+    return cur;
+}
+
+// evaluate the transformer
+//
+//   - model:     the model
+//   - n_threads: number of threads to use
+//   - n_past:    the context size so far
+//   - embd_inp:  the embeddings of the tokens in the context
+//   - embd_w:    the predicted logits for the next token
+//
+bool gpt_neox_eval(
+        const gpt_neox_model & model,
+        const int n_threads,
+        const int n_past,
+        const std::vector<gpt_vocab::id> & embd_inp,
+              std::vector<float>         & embd_w,
+              size_t                     & mem_per_token) {
+    const int N = embd_inp.size();
+
+    const auto & hparams = model.hparams;
+
+    const int n_embd  = hparams.n_embd;
+    const int n_layer = hparams.n_layer;
+    const int n_ctx   = hparams.n_ctx;
+    const int n_head  = hparams.n_head;
+    const int n_vocab = hparams.n_vocab;
+    const int n_rot   = hparams.n_rot;
+
+    static size_t buf_size = 256u*1024*1024;
+    static void * buf = malloc(buf_size);
+
+    // use 2 scratch buffers
+    // TODO: very hacky solution - reimplement in a more elegant way
+    static size_t scr0_size = 256u*1024*1024;
+    static void * scr0 = malloc(scr0_size);
+
+    static size_t scr1_size = 256u*1024*1024;
+    static void * scr1 = malloc(scr1_size);
+
+    if (mem_per_token > 0 && mem_per_token*N > buf_size) {
+        const size_t buf_size_new = 1.1*(mem_per_token*N); // add 10% to account for ggml object overhead
+        //printf("\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, buf_size, buf_size_new);
+
+        // reallocate
+        buf_size = buf_size_new;
+        buf = realloc(buf, buf_size);
+        if (buf == nullptr) {
+            fprintf(stderr, "%s: failed to allocate %zu bytes\n", __func__, buf_size);
+            return false;
+        }
+    }
+
+    struct ggml_init_params params = {
+        /*.mem_size   =*/ buf_size,
+        /*.mem_buffer =*/ buf,
+        /*.no_alloc   =*/ false,
+    };
+
+    struct ggml_context * ctx0 = ggml_init(params);
+    struct ggml_cgraph gf = {};
+
+    struct ggml_tensor * embd = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
+    memcpy(embd->data, embd_inp.data(), N*ggml_element_size(embd));
+
+
+    // wte
+    struct ggml_tensor * inpL = ggml_get_rows(ctx0, model.wte, embd);
+
+    for (int il = 0; il < n_layer; ++il) {
+        struct ggml_tensor * cur;
+
+        ggml_set_scratch(ctx0, { 0, scr0_size, scr0, });
+
+        // self-attention
+        {
+            {
+                cur = ggml_norm(ctx0, inpL);
+
+                cur = ggml_add(ctx0,
+                        ggml_mul(ctx0,
+                            ggml_repeat(ctx0, model.layers[il].ln_1_g, cur),
+                            cur),
+                        ggml_repeat(ctx0, model.layers[il].ln_1_b, cur));
+            }
+
+            // compute QKV
+            {
+
+                cur = ggml_mul_mat(ctx0,
+                        model.layers[il].c_attn_attn_w,
+                        cur);
+
+                cur = ggml_add(ctx0,
+                        ggml_repeat(ctx0, model.layers[il].c_attn_attn_b, cur),
+                        cur);
+            }
+
+            struct ggml_tensor * Qcur = ggml_cont(ctx0, ggml_view_3d(ctx0, cur, n_embd/n_head, n_head, N, cur->nb[1]/n_head, cur->nb[1], 0*sizeof(float)*n_embd/n_head));
+            struct ggml_tensor * Kcur = ggml_cont(ctx0, ggml_view_3d(ctx0, cur, n_embd/n_head, n_head, N, cur->nb[1]/n_head, cur->nb[1], 1*sizeof(float)*n_embd/n_head));
+            struct ggml_tensor * Vcur = ggml_cont(ctx0, ggml_view_3d(ctx0, cur, n_embd/n_head, n_head, N, cur->nb[1]/n_head, cur->nb[1], 2*sizeof(float)*n_embd/n_head));
+
+            // using mode = 2 for GPT-NeoX mode
+            Qcur = ggml_rope_inplace(ctx0, Qcur, n_past, n_rot, 2, 0);
+            Kcur = ggml_rope_inplace(ctx0, Kcur, n_past, n_rot, 2, 0);
+
+            // store key and value to memory
+            {
+                Vcur = ggml_transpose(ctx0, ggml_reshape_2d(ctx0, Vcur, n_embd, N));
+
+                struct ggml_tensor * k = ggml_view_1d(ctx0, model.memory_k, N*n_embd, (ggml_element_size(model.memory_k)*n_embd)*(il*n_ctx + n_past));
+                struct ggml_tensor * v = ggml_view_2d(ctx0, model.memory_v, N, n_embd,
+                        (   n_ctx)*ggml_element_size(model.memory_v),
+                        (il*n_ctx)*ggml_element_size(model.memory_v)*n_embd + n_past*ggml_element_size(model.memory_v));
+
+                ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Kcur, k));
+                ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Vcur, v));
+            }
+
+            // Q = Qcur.contiguous().view(n_embd/n_head, n_head, N).permute(0, 2, 1, 3)
+            struct ggml_tensor * Q =
+                ggml_permute(ctx0,
+                        Qcur,
+                        0, 2, 1, 3);
+
+            // K = Kmem.view(n_embd/n_head, n_head, n_past + N).permute(0, 2, 1, 3)
+            struct ggml_tensor * K =
+                ggml_permute(ctx0,
+                        ggml_reshape_3d(ctx0,
+                            ggml_view_1d(ctx0, model.memory_k, (n_past + N)*n_embd, il*n_ctx*ggml_element_size(model.memory_k)*n_embd),
+                            n_embd/n_head, n_head, n_past + N),
+                        0, 2, 1, 3);
+
+            // K * Q
+            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
+
+            // KQ_scaled = KQ / sqrt(n_embd/n_head)
+            struct ggml_tensor * KQ_scaled =
+                ggml_scale_inplace(ctx0,
+                        KQ,
+                        ggml_new_f32(ctx0, 1.0f/sqrt(float(n_embd)/n_head))
+                        );
+
+            // KQ_masked = mask_past(KQ_scaled)
+            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
+
+            // KQ = soft_max(KQ_masked)
+            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
+
+            // V_trans = Vmem.view(n_embd/n_head, n_head, n_past + N).permute(1, 2, 0, 3).contiguous()
+            struct ggml_tensor * V =
+                ggml_view_3d(ctx0, model.memory_v,
+                        n_past + N, n_embd/n_head, n_head,
+                        n_ctx*ggml_element_size(model.memory_v),
+                        n_ctx*ggml_element_size(model.memory_v)*n_embd/n_head,
+                        il*n_ctx*ggml_element_size(model.memory_v)*n_embd);
+
+            // KQV = transpose(V) * KQ_soft_max
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
+
+            // KQV_merged = KQV.permute(0, 2, 1, 3)
+            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
+
+            // cur = KQV_merged.contiguous().view(n_embd, N)
+            cur = ggml_cpy(ctx0,
+                    KQV_merged,
+                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+
+            // projection
+            {
+                cur = ggml_mul_mat(ctx0,
+                        model.layers[il].c_attn_proj_w,
+                        cur);
+
+                cur = ggml_add(ctx0, ggml_repeat(ctx0, model.layers[il].c_attn_proj_b, cur), cur);
+            }
+        }
+
+        ggml_set_scratch(ctx0, { 0, scr1_size, scr1, });
+
+        if (hparams.par_res == 0) {
+            struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpL);
+
+            cur = gpt_neox_ff(model.layers[il], ctx0, inpFF);
+
+            // input for next layer
+            inpL = ggml_add(ctx0, cur, inpFF);
+        } else {
+            struct ggml_tensor * inpFF = cur;
+
+            // this is independent of the self-attention result, so it could be done in parallel to the self-attention
+            // note here we pass inpL instead of cur
+            cur = gpt_neox_ff(model.layers[il], ctx0, inpL);
+
+            // layer input + FF
+            cur  = ggml_add(ctx0, cur, inpFF);
+
+            // input for next layer
+            inpL = ggml_add(ctx0, cur, inpL);
+        }
+    }
+
+    ggml_set_scratch(ctx0, { 0, scr0_size, scr0, });
+
+    // norm
+    {
+        inpL = ggml_norm(ctx0, inpL);
+
+        // inpL = ln_f_g*inpL + ln_f_b
+        inpL = ggml_add(ctx0,
+                ggml_mul(ctx0,
+                    ggml_repeat(ctx0, model.ln_f_g, inpL),
+                    inpL),
+                ggml_repeat(ctx0, model.ln_f_b, inpL));
+    }
+
+    ggml_set_scratch(ctx0, { 0, 0, nullptr, });
+
+    // lm_head
+    {
+        inpL = ggml_mul_mat(ctx0, model.lmh_g, inpL);
+
+        //inpL = ggml_add(ctx0,
+        //        ggml_repeat(ctx0, model.lmh_b, inpL),
+        //        inpL);
+    }
+
+    // logits -> probs
+    //inpL = ggml_soft_max_inplace(ctx0, inpL);
+
+    // run the computation
+    ggml_build_forward_expand(&gf, inpL);
+    ggml_graph_compute_with_ctx(ctx0, &gf, n_threads);
+
+    //if (n_past%100 == 0) {
+    //    ggml_graph_print   (&gf);
+    //    ggml_graph_dump_dot(&gf, NULL, "gpt-2.dot");
+    //}
+
+    //embd_w.resize(n_vocab*N);
+    //memcpy(embd_w.data(), ggml_get_data(inpL), sizeof(float)*n_vocab*N);
+
+    // return result for just the last token
+    embd_w.resize(n_vocab);
+    memcpy(embd_w.data(), (float *) ggml_get_data(inpL) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
+
+    if (mem_per_token == 0) {
+        mem_per_token = ggml_used_mem(ctx0)/N;
+    }
+    //printf("used_mem = %zu\n", ggml_used_mem(ctx0));
+
+    ggml_free(ctx0);
+
+    return true;
+}
+
+int main(int argc, char ** argv) {
+    ggml_time_init();
+
+    const int64_t t_main_start_us = ggml_time_us();
+
+    gpt_params params;
+
+    if (gpt_params_parse(argc, argv, params) == false) {
+        return 1;
+    }
+
+    if (params.seed < 0) {
+        params.seed = time(NULL);
+    }
+
+    printf("%s: seed = %d\n", __func__, params.seed);
+
+    std::mt19937 rng(params.seed);
+    if (params.prompt.empty()) {
+        params.prompt = gpt_random_prompt(rng);
+    }
+
+    int64_t t_load_us = 0;
+
+    gpt_vocab vocab;
+    gpt_neox_model model;
+
+    // load the model
+    {
+        const int64_t t_start_us = ggml_time_us();
+
+        if (!gpt_neox_model_load(params.model, model, vocab)) {
+            fprintf(stderr, "%s: failed to load model from '%s'\n", __func__, params.model.c_str());
+            return 1;
+        }
+
+        t_load_us = ggml_time_us() - t_start_us;
+
+    }
+
+    int n_past = 0;
+
+    int64_t t_sample_us  = 0;
+    int64_t t_predict_us = 0;
+
+    std::vector<float> logits;
+
+    // tokenize the prompt
+    std::vector<gpt_vocab::id> embd_inp = ::gpt_tokenize(vocab, params.prompt);
+
+    params.n_predict = std::min(params.n_predict, model.hparams.n_ctx - (int) embd_inp.size());
+
+    printf("%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
+    for (int i = 0; i < embd_inp.size(); i++) {
+        printf("%s: token[%d] = %6d, %s\n", __func__, i, embd_inp[i], vocab.id_to_token.at(embd_inp[i]).c_str());
+    }
+    printf("\n");
+
+    std::vector<gpt_vocab::id> embd;
+
+    // determine the required inference memory per token:
+    size_t mem_per_token = 0;
+    gpt_neox_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
+
+    for (int i = embd.size(); i < embd_inp.size() + params.n_predict; i++) {
+        // predict
+        if (embd.size() > 0) {
+            const int64_t t_start_us = ggml_time_us();
+
+            if (!gpt_neox_eval(model, params.n_threads, n_past, embd, logits, mem_per_token)) {
+                printf("Failed to predict\n");
+                return 1;
+            }
+
+            t_predict_us += ggml_time_us() - t_start_us;
+        }
+
+        n_past += embd.size();
+        embd.clear();
+
+        if (i >= embd_inp.size()) {
+            // sample next token
+            const int   top_k = params.top_k;
+            const float top_p = params.top_p;
+            const float temp  = params.temp;
+
+            const int n_vocab = model.hparams.n_vocab;
+
+            gpt_vocab::id id = 0;
+
+            {
+                const int64_t t_start_sample_us = ggml_time_us();
+
+                id = gpt_sample_top_k_top_p(vocab, logits.data() + (logits.size() - n_vocab), top_k, top_p, temp, rng);
+
+                t_sample_us += ggml_time_us() - t_start_sample_us;
+            }
+
+            // add it to the context
+            embd.push_back(id);
+        } else {
+            // if here, it means we are still processing the input prompt
+            for (int k = i; k < embd_inp.size(); k++) {
+                embd.push_back(embd_inp[k]);
+                if (embd.size() > params.n_batch) {
+                    break;
+                }
+            }
+            i += embd.size() - 1;
+        }
+
+        // display text
+        for (auto id : embd) {
+            printf("%s", vocab.id_to_token[id].c_str());
+        }
+        fflush(stdout);
+
+        // end of text token
+        if (embd.back() == 0) {
+            break;
+        }
+    }
+
+    // report timing
+    {
+        const int64_t t_main_end_us = ggml_time_us();
+
+        printf("\n\n");
+        printf("%s: mem per token = %8zu bytes\n", __func__, mem_per_token);
+        printf("%s:     load time = %8.2f ms\n", __func__, t_load_us/1000.0f);
+        printf("%s:   sample time = %8.2f ms\n", __func__, t_sample_us/1000.0f);
+        printf("%s:  predict time = %8.2f ms / %.2f ms per token\n", __func__, t_predict_us/1000.0f, t_predict_us/1000.0f/n_past);
+        printf("%s:    total time = %8.2f ms\n", __func__, (t_main_end_us - t_main_start_us)/1000.0f);
+    }
+
+    ggml_free(model.ctx);
+
+    return 0;
+}

From 068a8e0fbe126b7cbd1e4cf2fc125c8246513315 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 17:29:56 +0200
Subject: [PATCH 051/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 7c867619bc659..3e7529b2faa89 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -64,8 +64,8 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
     hparams = json.load(f)
 
-# This mmust be changed when adding/deleting kv
-kv_count = 14
+# This must be changed when adding/deleting kv
+kv_count = 13
 
 print("tensors " + str(tensor_count) + " kv " + str(kv_count))
 
@@ -86,7 +86,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
 gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
-gguf_writer.write_float32(llm_arch + ".attention.layer_norm_rms_epsilon", hparams["rms_norm_eps"])
 gguf_writer.write_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
 
 

From 2a0914673cdd57e0388b69711f04f7b9029d7bfe Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 30 Jul 2023 17:31:11 +0200
Subject: [PATCH 052/242] Update convert-gptneox-h5-to-gguf.py

---
 convert-gptneox-h5-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index cbade7079be8b..12b6cc960775a 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -54,7 +54,7 @@
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
     hparams = json.load(f)
 
-# This mmust be changed when adding/deleting kv
+# This must be changed when adding/deleting kv
 kv_count = 14
 
 print("tensors " + str(tensor_count) + " kv " + str(kv_count))

From 4f5b6224beb7b1a0291843cb118b148d5b6e7a6c Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 31 Jul 2023 03:00:20 +0200
Subject: [PATCH 053/242] Update convert-gptneox-h5-to-gguf.py

---
 convert-gptneox-h5-to-gguf.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 12b6cc960775a..2661995096979 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -37,6 +37,12 @@
         sys.exit(1)
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+if hparams["architectures"][0] != "GPTNeoXForCausalLM":
+    print("Model architecture not supported: " + hparams["architectures"][0] )
+    sys.exit()
 
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
@@ -51,9 +57,6 @@
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
-    hparams = json.load(f)
-
 # This must be changed when adding/deleting kv
 kv_count = 14
 

From 6b3a7b9f4f726a7a3474e1538d74d3e8d6a92541 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 31 Jul 2023 03:02:00 +0200
Subject: [PATCH 054/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 3e7529b2faa89..0451ffe232b85 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -47,6 +47,12 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         sys.exit(1)
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
+if hparams["architectures"][0] != "LlamaForCausalLM":
+    print("Model architecture not supported: " + hparams["architectures"][0] )
+    sys.exit()
+    
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
 
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
@@ -61,9 +67,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
-    hparams = json.load(f)
-
 # This must be changed when adding/deleting kv
 kv_count = 13
 

From 7aa0a0e7f7ff7d7809a289475f77999dcfe093b4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Mon, 31 Jul 2023 09:59:36 +0300
Subject: [PATCH 055/242] gguf : support custom alignment value

---
 constants.py | 1 +
 ggml.c       | 5 ++++-
 gguf.py      | 6 +++++-
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/constants.py b/constants.py
index 14f11123b1280..d9e110b73fd63 100644
--- a/constants.py
+++ b/constants.py
@@ -5,6 +5,7 @@
 # general
 KEY_GENERAL_ARCHITECTURE         = "general.architecture"
 KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
+KEY_GENERAL_ALIGNMENT            = "general.alignment"
 KEY_GENERAL_NAME                 = "general.name"
 KEY_GENERAL_AUTHOR               = "general.author"
 KEY_GENERAL_URL                  = "general.url"
diff --git a/ggml.c b/ggml.c
index 1b2f071e71a3d..157199118b282 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18543,7 +18543,10 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
 
-    // TODO: determine new alignment from kv if available
+    int alignment_idx = gguf_find_key(ctx, "general.alignment");
+    if (alignment_idx != -1) {
+        ctx->alignment = gguf_get_u32(ctx, alignment_idx);
+    }
 
     // we require the data section to be aligned, so take into account any padding
     {
diff --git a/gguf.py b/gguf.py
index 99ea817028c70..8e2c771baebc5 100644
--- a/gguf.py
+++ b/gguf.py
@@ -220,6 +220,9 @@ def write_quantization_version(self, quantization_version: GGMLQuantizationType)
         self.write_uint32(
             constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
 
+    def write_custom_alignment(self, alignment: int):
+        self.write_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
+
     def write_context_length(self, llm: str, length: int):
         self.write_uint32(
             constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
@@ -292,11 +295,12 @@ def write_token_scores(self, scores: List[float]):
 if __name__ == "__main__":
     # Example usage with a file
     gguf_writer = GGUFWriter.open("example.gguf")
-    gguf_writer.write_header(2, 3)
+    gguf_writer.write_header(2, 4)
 
     gguf_writer.write_architecture("llama")
     gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
     gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
+    gguf_writer.write_custom_alignment(64)
     tensor1 = np.ones((32,), dtype=np.float32) * 100.0
     tensor2 = np.ones((32,), dtype=np.float32) * 101.0
     gguf_writer.write_tensor_info("tensor0", tensor1)

From b26f5b2e43eb3e8b1f5948f16eafe22b362e6c26 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Mon, 31 Jul 2023 16:23:54 +0300
Subject: [PATCH 056/242] gguf : fix typo in function call

---
 ggml.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml.c b/ggml.c
index 157199118b282..33a90f5371f7d 100644
--- a/ggml.c
+++ b/ggml.c
@@ -18545,7 +18545,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     int alignment_idx = gguf_find_key(ctx, "general.alignment");
     if (alignment_idx != -1) {
-        ctx->alignment = gguf_get_u32(ctx, alignment_idx);
+        ctx->alignment = gguf_get_val_u32(ctx, alignment_idx);
     }
 
     // we require the data section to be aligned, so take into account any padding

From bb42aefaeb115a882d839a3c658ab119b692e075 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Mon, 31 Jul 2023 17:46:12 +0300
Subject: [PATCH 057/242] gguf : mmap tensor data example

---
 examples/gguf/gguf.cpp | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 84388c219f5ef..deb4a83592779 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,4 +1,5 @@
 #include "ggml.h"
+#include "llama-util.h"
 
 #include <cstdio>
 #include <cinttypes>
@@ -376,7 +377,31 @@ bool gguf_ex_read_2(const std::string & fname) {
 
     // TODO: mmap based on tensor infos
 
-    fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
+    
+    struct llama_file file(fname.c_str(), "rb");
+    llama_mmap data_mmap(&file, 0, false);
+    const int n_tensors = gguf_get_n_tensors(ctx);
+
+    for (int i = 0; i < n_tensors; ++i) {
+        const char * name             = gguf_get_tensor_name(ctx, i);
+        const size_t offset      = gguf_get_data_offset(ctx) + gguf_get_tensor_offset(ctx, i);
+        struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
+
+        cur->data = static_cast<char *>(data_mmap.addr) + offset;
+
+        // print first 10 elements
+    const float * data = (const float *) cur->data;
+                
+        printf("%s data[:10] : ", name);
+
+        for (int j = 0; j < 10; ++j) {
+            printf("%f ", data[j]);
+        }
+
+        printf("\n\n");
+    }
+
+fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
 
     ggml_free(ctx_data);
     gguf_free(ctx);

From f3de876a1212767c9ba9223de26472b40f9043ff Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Mon, 31 Jul 2023 23:58:29 +0300
Subject: [PATCH 058/242] fix : update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 0451ffe232b85..412d334fb8920 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -46,13 +46,13 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         print("Invalid ftype: " + str(ftype))
         sys.exit(1)
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+    
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
 
 if hparams["architectures"][0] != "LlamaForCausalLM":
     print("Model architecture not supported: " + hparams["architectures"][0] )
     sys.exit()
-    
-with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
-    hparams = json.load(f)
 
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()

From da4900e8359aba5923ac4f005608b292b77c45da Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 31 Jul 2023 23:04:03 +0200
Subject: [PATCH 059/242] Update convert-llama-h5-to-gguf.py

---
 convert-llama-h5-to-gguf.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 412d334fb8920..7f16905959b73 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -104,11 +104,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     print("Adding sentencepiece tokenizer vocab.")
     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
-    # output vocab_size followed by all piece/score pairs
-    outbytes: bytes
-    outbytes = b""
-    outbytes += struct.pack("I", tokenizer.vocab_size())
-
     for i in range(tokenizer.vocab_size()):
         text: bytes
         if tokenizer.is_unknown(i):

From e7a741695c5a23ef6d591555e7cc255f9b79d690 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 14:30:00 +0200
Subject: [PATCH 060/242] convert-gptneox-h5-to-gguf.py : Special tokens

---
 convert-gptneox-h5-to-gguf.py | 42 +++++++++++++++++++++++++++++++----
 1 file changed, 38 insertions(+), 4 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 2661995096979..f255cb7d7ee13 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -58,7 +58,7 @@
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
 # This must be changed when adding/deleting kv
-kv_count = 14
+kv_count = 17
 
 print("tensors " + str(tensor_count) + " kv " + str(kv_count))
 
@@ -101,9 +101,43 @@
 
     merges = tokenizer["model"]["merges"]
 
-gguf_writer.write_tokenizer_model("gpt2")
-gguf_writer.write_token_list(tokens)
-gguf_writer.write_token_merges(merges)
+    gguf_writer.write_tokenizer_model("gpt2")
+    gguf_writer.write_token_list(tokens)
+    gguf_writer.write_token_merges(merges)
+
+    if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
+        print("Adding special token ids")
+
+        with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
+            tokenizer_config = json.load(f)
+
+        # find special token ids
+
+        if "bos_token" in tokenizer_config:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["bos_token"]:
+                    gguf_writer.write_uint32("tokenizer.ggml.bos_token_id", key["id"] )
+
+        if "eos_token" in tokenizer_config:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["eos_token"]:
+                    gguf_writer.write_uint32("tokenizer.ggml.eos_token_id", key["id"] )
+
+        if "unk_token" in tokenizer_config:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["unk_token"]:
+                    gguf_writer.write_uint32("tokenizer.ggml.unknown_token_id", key["id"] )
+
+        if "sep_token" in tokenizer_config:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["sep_token"]:
+                    gguf_writer.write_uint32("tokenizer.ggml.separator_token_id", key["id"] )
+
+        if "pad_token" in tokenizer_config:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["pad_token"]:
+                    gguf_writer.write_uint32("tokenizer.ggml.padding_token_id", key["id"] )
+
 
 # TENSORS
 

From c77fabb1f9b77d6665065da494cb918af71c1dc0 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 14:32:53 +0200
Subject: [PATCH 061/242] gptneox-main.cpp : special tokens

---
 gptneox-main.cpp | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 02fbc9fba35a5..fa95cb7722a11 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -257,6 +257,11 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
             vocab.id_to_token[i] = word;
         }
 
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       printf("bos id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       printf("eos id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   printf("unk id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { printf("sep id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   printf("pad id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
 
     }
 

From 36a36c32a3814c036b6290735ba9d4c9111653c7 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 14:44:28 +0200
Subject: [PATCH 062/242] Update gptneox-main.cpp

---
 gptneox-main.cpp | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index fa95cb7722a11..235fca7ce7094 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -716,6 +716,9 @@ int main(int argc, char ** argv) {
 
     }
 
+    uint32_t eos_token_id = 0;
+    int keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {  eos_token_id = gguf_get_val_u32(ggufctx, keyidx); }    
+
     int n_past = 0;
 
     int64_t t_sample_us  = 0;
@@ -794,7 +797,7 @@ int main(int argc, char ** argv) {
         fflush(stdout);
 
         // end of text token
-        if (embd.back() == 0) {
+        if (embd.back() == eos_token_id) {
             break;
         }
     }

From ff1cb02397002f874bde7d91256bb31a6729a577 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 23:17:21 +0200
Subject: [PATCH 063/242] constants.py : special tokens

---
 constants.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/constants.py b/constants.py
index d9e110b73fd63..ae6e719fb555e 100644
--- a/constants.py
+++ b/constants.py
@@ -47,3 +47,8 @@
 KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
 KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json"
 KEY_TOKENIZER_RWKV    = "tokenizer.rwkv.world"
+KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.separator_token_id"
+KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"

From 49380a23a32067b392dc1f81857c2129471456a9 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 23:37:48 +0200
Subject: [PATCH 064/242] gguf.py : accumulate kv and tensor info data +
 special tokens

---
 gguf.py | 284 +++++++++++++++++++++++++++++++-------------------------
 1 file changed, 159 insertions(+), 125 deletions(-)

diff --git a/gguf.py b/gguf.py
index 8e2c771baebc5..88e2bee0733b1 100644
--- a/gguf.py
+++ b/gguf.py
@@ -61,98 +61,113 @@ class GGUFWriter:
     def __init__(self, fout: IO):
         self.fout = fout
         self.offset_tensor = 0
+        self.kv_data = b""
+        self.kv_data_count = 0
+        self.ti_data = b""
+        self.ti_data_count = 0
 
-    def write_header(self, tensor_count: int, metadata_kv_count: int):
+    def write_header_to_file(self):
         self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
         self.fout.write(struct.pack("<I", constants.GGUF_VERSION))
-        self.fout.write(struct.pack("<I", tensor_count))
-        self.fout.write(struct.pack("<I", metadata_kv_count))
+        self.fout.write(struct.pack("<I", self.ti_data_count))
+        self.fout.write(struct.pack("<I", self.kv_data_count))
+        self.flush()
+#        print("tensors " + str(self.ti_data_count) + " kv " + str(self.kv_data_count))
+
+    def write_kv_data_to_file(self):
+        self.fout.write(self.kv_data)
+        self.flush()
+
+    def write_ti_data_to_file(self):
+        self.fout.write(self.ti_data)
+        self.flush()
 
     @classmethod
     def open(cls, path: str) -> "GGUFWriter":
         f = open(path, "wb")
         return cls(f)
 
-    def write_key(self, key: str):
-        self.write_val(key, GGUFValueType.STRING, write_vtype=False)
+    def add_key(self, key: str):
+        self.add_val(key, GGUFValueType.STRING, add_vtype=False)
 
-    def write_uint8(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.UINT8)
+    def add_uint8(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT8)
 
-    def write_int8(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.INT8)
+    def add_int8(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT8)
 
-    def write_uint16(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.UINT16)
+    def add_uint16(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT16)
 
-    def write_int16(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.INT16)
+    def add_int16(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT16)
 
-    def write_uint32(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.UINT32)
+    def add_uint32(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT32)
 
-    def write_int32(self, key: str, val: int):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.INT32)
+    def add_int32(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT32)
 
-    def write_float32(self, key: str, val: float):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.FLOAT32)
+    def add_float32(self, key: str, val: float):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.FLOAT32)
 
-    def write_bool(self, key: str, val: bool):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.BOOL)
+    def add_bool(self, key: str, val: bool):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.BOOL)
 
-    def write_string(self, key: str, val: str):
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.STRING)
+    def add_string(self, key: str, val: str):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.STRING)
 
-    def write_array(self, key: str, val: list):
+    def add_array(self, key: str, val: list):
         if not isinstance(val, list):
             raise ValueError("Value must be a list for array type")
 
-        self.write_key(key)
-        self.write_val(val, GGUFValueType.ARRAY)
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.ARRAY)
 
-    def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: bool = True):
+    def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool = True):
         if vtype is None:
             vtype = GGUFValueType.get_type(val)
 
-        if write_vtype:
-            self.fout.write(struct.pack("<I", vtype))
+        if add_vtype:
+            self.kv_data += struct.pack("<I", vtype)
+            self.kv_data_count += 1;
 
         if vtype == GGUFValueType.UINT8:
-            self.fout.write(struct.pack("<B", val))
+            self.kv_data += struct.pack("<B", val)
         elif vtype == GGUFValueType.INT8:
-            self.fout.write(struct.pack("<b", val))
+            self.kv_data += struct.pack("<b", val)
         elif vtype == GGUFValueType.UINT16:
-            self.fout.write(struct.pack("<H", val))
+            self.kv_data += struct.pack("<H", val)
         elif vtype == GGUFValueType.INT16:
-            self.fout.write(struct.pack("<h", val))
+            self.kv_data += struct.pack("<h", val)
         elif vtype == GGUFValueType.UINT32:
-            self.fout.write(struct.pack("<I", val))
+            self.kv_data += struct.pack("<I", val)
         elif vtype == GGUFValueType.INT32:
-            self.fout.write(struct.pack("<i", val))
+            self.kv_data += struct.pack("<i", val)
         elif vtype == GGUFValueType.FLOAT32:
-            self.fout.write(struct.pack("<f", val))
+            self.kv_data += struct.pack("<f", val)
         elif vtype == GGUFValueType.BOOL:
-            self.fout.write(struct.pack("?", val))
+            self.kv_data += struct.pack("?", val)
         elif vtype == GGUFValueType.STRING:
             encoded_val = val.encode("utf8") if isinstance(val, str) else val
-            self.fout.write(struct.pack("<I", len(encoded_val)))
-            self.fout.write(encoded_val)
+            self.kv_data += struct.pack("<I", len(encoded_val))
+            self.kv_data += encoded_val
         elif vtype == GGUFValueType.ARRAY:
             ltype = set([GGUFValueType.get_type(item) for item in val])
             assert len(ltype) == 1, "All items in a GGUF array should be of the same type"
-            self.fout.write(struct.pack("<I", list(ltype)[0]))
-            self.fout.write(struct.pack("<I", len(val)))
+            self.kv_data += struct.pack("<I", list(ltype)[0])
+            self.kv_data += struct.pack("<I", len(val))
             for item in val:
-                self.write_val(item, write_vtype=False)
+                self.add_val(item, add_vtype=False)
         else:
             raise ValueError("Invalid GGUF metadata value type")
 
@@ -160,22 +175,23 @@ def write_val(self: str, val: Any, vtype: GGUFValueType = None, write_vtype: boo
     def ggml_pad(x: int, n: int) -> int:
         return ((x + n - 1) // n) * n
 
-    def write_tensor_info(self, name: str, tensor: np.ndarray):
-        self.write_key(name)
+    def add_tensor_info(self, name: str, tensor: np.ndarray):
+        encoded_name = name.encode("utf8")
+        self.ti_data += struct.pack("<I", len(encoded_name))
+        self.ti_data += encoded_name
         n_dims = len(tensor.shape)
-        self.fout.write(struct.pack("<i", n_dims))
+        self.ti_data += struct.pack("<I", n_dims)
         for i in range(n_dims):
-            self.fout.write(struct.pack("<i", tensor.shape[n_dims - 1 - i]))
+            self.ti_data += struct.pack("<I", tensor.shape[n_dims - 1 - i])
 
         assert tensor.dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
         dtype = GGMLQuantizationType.F32 if tensor.dtype == np.float32 else GGMLQuantizationType.F16
-        self.fout.write(struct.pack("<i", dtype))
-        self.fout.write(struct.pack("<Q", self.offset_tensor))
+        self.ti_data += struct.pack("<I", dtype)
+        self.ti_data += struct.pack("<Q", self.offset_tensor)
         self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT)
+        self.ti_data_count += 1
 
-        self.flush()
-
-    def write_tensor(self, tensor: np.ndarray):
+    def write_tensor_to_file(self, tensor: np.ndarray):
         pad = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT) - self.fout.tell()
         if pad != 0:
             self.fout.write(bytes([0] * pad))
@@ -191,121 +207,139 @@ def flush(self):
     def close(self):
         self.fout.close()
 
-    def write_architecture(self, architecture: str):
-        self.write_string(constants.KEY_GENERAL_ARCHITECTURE,
+    def add_architecture(self, architecture: str):
+        self.add_string(constants.KEY_GENERAL_ARCHITECTURE,
                           architecture)
 
-    def write_author(self, author: str):
-        self.write_string(constants.KEY_GENERAL_AUTHOR, author)
+    def add_author(self, author: str):
+        self.add_string(constants.KEY_GENERAL_AUTHOR, author)
 
-    def write_url(self, url: str):
-        self.write_string(constants.KEY_GENERAL_URL, url)
+    def add_url(self, url: str):
+        self.add_string(constants.KEY_GENERAL_URL, url)
 
-    def write_description(self, description: str):
-        self.write_string(constants.KEY_GENERAL_DESCRIPTION, description)
+    def add_description(self, description: str):
+        self.add_string(constants.KEY_GENERAL_DESCRIPTION, description)
 
-    def write_file_type(self, file_type: str):
-        self.write_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
+    def add_file_type(self, file_type: str):
+        self.add_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
 
-    def write_source_url(self, url: str):
-        self.write_string(constants.KEY_GENERAL_SOURCE_URL, url)
+    def add_source_url(self, url: str):
+        self.add_string(constants.KEY_GENERAL_SOURCE_URL, url)
 
-    def write_source_hf_repo(self, repo: str):
-        self.write_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
+    def add_source_hf_repo(self, repo: str):
+        self.add_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
 
-    def write_name(self, name: str):
-        self.write_string(constants.KEY_GENERAL_NAME, name)
+    def add_name(self, name: str):
+        self.add_string(constants.KEY_GENERAL_NAME, name)
 
-    def write_quantization_version(self, quantization_version: GGMLQuantizationType):
-        self.write_uint32(
+    def add_quantization_version(self, quantization_version: GGMLQuantizationType):
+        self.add_uint32(
             constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
 
-    def write_custom_alignment(self, alignment: int):
-        self.write_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
+    def add_custom_alignment(self, alignment: int):
+        self.add_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
 
-    def write_context_length(self, llm: str, length: int):
-        self.write_uint32(
+    def add_context_length(self, llm: str, length: int):
+        self.add_uint32(
             constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
 
-    def write_embedding_length(self, llm: str, length: int):
-        self.write_uint32(
+    def add_embedding_length(self, llm: str, length: int):
+        self.add_uint32(
             constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
 
-    def write_layer_count(self, llm: str, length: int):
-        self.write_uint32(
+    def add_layer_count(self, llm: str, length: int):
+        self.add_uint32(
             constants.KEY_LLM_LAYER_COUNT.format(llm=llm), length)
 
-    def write_feed_forward_length(self, llm: str, length: int):
-        self.write_uint32(
+    def add_feed_forward_length(self, llm: str, length: int):
+        self.add_uint32(
             constants.KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
 
-    def write_parallel_residual(self, llm: str, use: bool):
-        self.write_bool(
+    def add_parallel_residual(self, llm: str, use: bool):
+        self.add_bool(
             constants.KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
 
-    def write_tensor_data_layout(self, llm: str, layout: str):
-        self.write_string(
+    def add_tensor_data_layout(self, llm: str, layout: str):
+        self.add_string(
             constants.KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
 
-    def write_head_count(self, llm: str, count: int):
-        self.write_uint32(
+    def add_head_count(self, llm: str, count: int):
+        self.add_uint32(
             constants.KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
 
-    def write_head_count_kv(self, llm: str, count: int):
-        self.write_uint32(
+    def add_head_count_kv(self, llm: str, count: int):
+        self.add_uint32(
             constants.KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
 
-    def write_max_alibi_bias(self, llm: str, bias: float):
-        self.write_float32(
+    def add_max_alibi_bias(self, llm: str, bias: float):
+        self.add_float32(
             constants.KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
 
-    def write_clamp_kqv(self, llm: str, value: float):
-        self.write_float32(
+    def add_clamp_kqv(self, llm: str, value: float):
+        self.add_float32(
             constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
 
-    def write_layer_norm_eps(self, llm: str, value: float):
-        self.write_float32(
+    def add_layer_norm_eps(self, llm: str, value: float):
+        self.add_float32(
             constants.KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
 
-    def write_layer_norm_rms_eps(self, llm: str, value: float):
-        self.write_float32(
+    def add_layer_norm_rms_eps(self, llm: str, value: float):
+        self.add_float32(
             constants.KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
 
-    def write_rope_dimension_count(self, llm: str, count: int):
-        self.write_uint32(
+    def add_rope_dimension_count(self, llm: str, count: int):
+        self.add_uint32(
             constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
 
-    def write_rope_scale(self, llm: str, value:  float):
-        self.write_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
+    def add_rope_scale(self, llm: str, value:  float):
+        self.add_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
+
+    def add_tokenizer_model(self, model: str):
+        self.add_string(constants.KEY_TOKENIZER_MODEL, model)
+
+    def add_token_list(self, tokens: List):
+        self.add_array(constants.KEY_TOKENIZER_LIST, tokens)
+
+    def add_token_merges(self, merges: List):
+        self.add_array(constants.KEY_TOKENIZER_MERGES, merges)
+
+    def add_token_scores(self, scores: List[float]):
+        self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
+    
+    def add_bos_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_BOS_ID, id)
 
-    def write_tokenizer_model(self, model: str):
-        self.write_string(constants.KEY_TOKENIZER_MODEL, model)
+    def add_eos_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_EOS_ID, id)
 
-    def write_token_list(self, tokens: List):
-        self.write_array(constants.KEY_TOKENIZER_LIST, tokens)
+    def add_unk_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_UNK_ID, id)
 
-    def write_token_merges(self, merges: List):
-        self.write_array(constants.KEY_TOKENIZER_MERGES, merges)
+    def add_sep_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_SEP_ID, id)
 
-    def write_token_scores(self, scores: List[float]):
-        self.write_array(constants.KEY_TOKENIZER_SCORES, scores)
+    def add_pad_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_PAD_ID, id)
 
 
 # Example usage:
 if __name__ == "__main__":
     # Example usage with a file
     gguf_writer = GGUFWriter.open("example.gguf")
-    gguf_writer.write_header(2, 4)
 
-    gguf_writer.write_architecture("llama")
-    gguf_writer.write_uint32("answer", 42)  # Write a 32-bit integer
-    gguf_writer.write_float32("answer_in_float", 42.0)  # Write a 32-bit float
-    gguf_writer.write_custom_alignment(64)
+    gguf_writer.add_architecture("llama")
+    gguf_writer.add_uint32("answer", 42)  # Write a 32-bit integer
+    gguf_writer.add_float32("answer_in_float", 42.0)  # Write a 32-bit float
+    gguf_writer.add_custom_alignment(64)
     tensor1 = np.ones((32,), dtype=np.float32) * 100.0
     tensor2 = np.ones((32,), dtype=np.float32) * 101.0
-    gguf_writer.write_tensor_info("tensor0", tensor1)
-    gguf_writer.write_tensor_info("tensor1", tensor2)
-    gguf_writer.write_tensor(tensor1)
-    gguf_writer.write_tensor(tensor2)
+    gguf_writer.add_tensor_info("tensor0", tensor1)
+    gguf_writer.add_tensor_info("tensor1", tensor2)
+
+    gguf_writer.write_header_to_file()
+    gguf_writer.write_kv_data_to_file()
+    gguf_writer.write_ti_data_to_file()
+    gguf_writer.write_tensor_to_file(tensor1)
+    gguf_writer.write_tensor_to_file(tensor2)
 
     gguf_writer.close()

From 1b4f9c8eb9586b83bfb1b8bd634a1c448017ab25 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 1 Aug 2023 23:40:50 +0200
Subject: [PATCH 065/242] convert-gptneox-h5-to-gguf.py : accumulate kv and ti
 + special tokens

---
 convert-gptneox-h5-to-gguf.py | 88 ++++++++++++++++-------------------
 1 file changed, 39 insertions(+), 49 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index f255cb7d7ee13..4f9d41487d22f 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -47,51 +47,35 @@
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
 
-# count tensors to be converted
-tensor_count = 0
-for name in list_vars.keys():
-    # we don't need these
-    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-        continue
-    tensor_count += 1
-
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-# This must be changed when adding/deleting kv
-kv_count = 17
-
-print("tensors " + str(tensor_count) + " kv " + str(kv_count))
-
-print("write gguf header")
-
-gguf_writer.write_header(tensor_count, kv_count)
 
-print("write gguf hparams")
+print("gguf: add key-values, metadata")
 
 llm_arch = "gptneox"
 
-gguf_writer.write_name("pythia-70b-deduped")
-gguf_writer.write_description("gguf test model")
-gguf_writer.write_architecture(llm_arch)
-gguf_writer.write_context_length(llm_arch, hparams["max_position_embeddings"])
-gguf_writer.write_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.write_layer_count(llm_arch, hparams["num_hidden_layers"])
-gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
-gguf_writer.write_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
-gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
-gguf_writer.write_parallel_residual(llm_arch, hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
-gguf_writer.write_layer_norm_eps(llm_arch, hparams["layer_norm_eps"])
+gguf_writer.add_name("pythia-70b-deduped")
+gguf_writer.add_description("gguf test model")
+gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
+gguf_writer.add_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
+gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
+gguf_writer.add_parallel_residual(llm_arch, hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
+gguf_writer.add_layer_norm_eps(llm_arch, hparams["layer_norm_eps"])
 
 # TOKENIZATION
 
-print("write gguf tokenizer")
+print("gguf: add key-values, tokenizer")
 
 tokens: List[str] = []
 merges: List[str] = []
 
 if Path(dir_model + "/tokenizer.json").is_file():
     # vocab type gpt2
-    print("Adding gpt2 tokenizer vocab")
+    print("gguf: adding gpt2 tokenizer vocab")
 
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
         tokenizer = json.load(f)
@@ -101,12 +85,12 @@
 
     merges = tokenizer["model"]["merges"]
 
-    gguf_writer.write_tokenizer_model("gpt2")
-    gguf_writer.write_token_list(tokens)
-    gguf_writer.write_token_merges(merges)
+    gguf_writer.add_tokenizer_model("gpt2")
+    gguf_writer.add_token_list(tokens)
+    gguf_writer.add_token_merges(merges)
 
     if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
-        print("Adding special token ids")
+        print("gguf: adding special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
             tokenizer_config = json.load(f)
@@ -116,33 +100,33 @@
         if "bos_token" in tokenizer_config:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["bos_token"]:
-                    gguf_writer.write_uint32("tokenizer.ggml.bos_token_id", key["id"] )
+                    gguf_writer.add_bos_token_id(key["id"])
 
         if "eos_token" in tokenizer_config:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["eos_token"]:
-                    gguf_writer.write_uint32("tokenizer.ggml.eos_token_id", key["id"] )
+                    gguf_writer.add_eos_token_id(key["id"])
 
         if "unk_token" in tokenizer_config:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["unk_token"]:
-                    gguf_writer.write_uint32("tokenizer.ggml.unknown_token_id", key["id"] )
+                    gguf_writer.add_unk_token_id(key["id"])
 
         if "sep_token" in tokenizer_config:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["sep_token"]:
-                    gguf_writer.write_uint32("tokenizer.ggml.separator_token_id", key["id"] )
+                    gguf_writer.add_sep_token_id(key["id"])
 
         if "pad_token" in tokenizer_config:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["pad_token"]:
-                    gguf_writer.write_uint32("tokenizer.ggml.padding_token_id", key["id"] )
+                    gguf_writer.add_pad_token_id(key["id"])
 
 
 # TENSORS
 
 # tensor info
-print("write gguf tensor info")
+print("gguf: add gguf tensor info")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -167,19 +151,25 @@
             data = data.astype(np.float32)
             ftype_cur = 0
 
-    gguf_writer.write_tensor_info(name, data)
+    gguf_writer.add_tensor_info(name, data)
 
+print("gguf: write header")
+gguf_writer.write_header_to_file()
+print("gguf: write key-values")
+gguf_writer.write_kv_data_to_file()
+print("gguf: write tensor info")
+gguf_writer.write_ti_data_to_file()
 
 # tensor data
-print("write gguf tensor data")
+print("gguf: write tensor data")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
-    print("Process tensor: " + name + " with shape: ", data.shape)
+#    print("Process tensor: " + name + " with shape: ", data.shape)
 
     # we don't need these
     if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-        print("  Skip tensor: " + name)
+#        print("  Skip tensor: " + name)
         continue
 
     n_dims = len(data.shape)
@@ -188,23 +178,23 @@
     ftype_cur = 0
     if ftype != 0:
         if name.endswith(".weight") and n_dims == 2:
-            print("  Converting to float16")
+#            print("  Converting to float16")
             data = data.astype(np.float16)
             ftype_cur = 1
         else:
-            print("  Converting to float32")
+#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
     else:
         if data.dtype != np.float32:
-            print("  Converting to float32")
+#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
 
-    gguf_writer.write_tensor(data)
+    gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 
 
-print("Done. Output file: " + fname_out)
+print("gguf: conversion done, output file: " + fname_out)
 print("")

From cf365fbc2022ca25ead5568dd81b4b63349d2c7d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Wed, 2 Aug 2023 11:13:56 +0300
Subject: [PATCH 066/242] gguf : gguf counterpart of llama-util.h

---
 examples/gguf/gguf.cpp |  11 +-
 gguf-util.h            | 451 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 455 insertions(+), 7 deletions(-)
 create mode 100644 gguf-util.h

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index deb4a83592779..671f8748f5e13 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,5 +1,5 @@
 #include "ggml.h"
-#include "llama-util.h"
+#include "gguf-util.h"
 
 #include <cstdio>
 #include <cinttypes>
@@ -375,11 +375,8 @@ bool gguf_ex_read_2(const std::string & fname) {
 
     struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
 
-    // TODO: mmap based on tensor infos
-
-    
-    struct llama_file file(fname.c_str(), "rb");
-    llama_mmap data_mmap(&file, 0, false);
+    struct gguf_file file(fname.c_str(), "rb");
+    gguf_mmap data_mmap(&file, 0, false);
     const int n_tensors = gguf_get_n_tensors(ctx);
 
     for (int i = 0; i < n_tensors; ++i) {
@@ -405,7 +402,7 @@ fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data
 
     ggml_free(ctx_data);
     gguf_free(ctx);
-
+    
     return true;
 }
 
diff --git a/gguf-util.h b/gguf-util.h
new file mode 100644
index 0000000000000..23f411bb0246b
--- /dev/null
+++ b/gguf-util.h
@@ -0,0 +1,451 @@
+// GGUF counterpart of llama-util.h.
+// we may consider making it a part of ggml.c once GGUF work is complete.
+// Contains wrappers around OS interfaces.
+
+#ifndef GGUF_UTIL_H
+#define GGUF_UTIL_H
+#include "ggml.h"
+#include <cstdio>
+#include <cstdint>
+#include <cerrno>
+#include <cstring>
+#include <cstdarg>
+#include <cstdlib>
+#include <climits>
+
+#include <string>
+#include <vector>
+#include <stdexcept>
+
+#ifdef __has_include
+    #if __has_include(<unistd.h>)
+        #include <unistd.h>
+        #if defined(_POSIX_MAPPED_FILES)
+            #include <sys/mman.h>
+        #endif
+        #if defined(_POSIX_MEMLOCK_RANGE)
+            #include <sys/resource.h>
+        #endif
+    #endif
+#endif
+
+#if defined(_WIN32)
+    #define WIN32_LEAN_AND_MEAN
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <windows.h>
+    #include <io.h>
+    #include <stdio.h> // for _fseeki64
+#endif
+
+#ifdef __GNUC__
+#ifdef __MINGW32__
+__attribute__((format(gnu_printf, 1, 2)))
+#else
+__attribute__((format(printf, 1, 2)))
+#endif
+#endif
+static std::string format(const char * fmt, ...) {
+    va_list ap, ap2;
+    va_start(ap, fmt);
+    va_copy(ap2, ap);
+    int size = vsnprintf(NULL, 0, fmt, ap);
+    GGML_ASSERT(size >= 0 && size < INT_MAX);
+    std::vector<char> buf(size + 1);
+    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
+    GGML_ASSERT(size2 == size);
+    va_end(ap2);
+    va_end(ap);
+    return std::string(buf.data(), size);
+}
+
+// TODO: can we merge this one and gguf_context?
+struct gguf_file {
+    // use FILE * so we don't have to re-open the file to mmap
+    FILE * fp;
+    size_t size;
+
+    gguf_file(const char * fname, const char * mode) {
+        fp = std::fopen(fname, mode);
+        if (fp == NULL) {
+            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
+        }
+        seek(0, SEEK_END);
+        size = tell();
+        seek(0, SEEK_SET);
+    }
+
+    size_t tell() const {
+#ifdef _WIN32
+        __int64 ret = _ftelli64(fp);
+#else
+        long ret = std::ftell(fp);
+#endif
+        GGML_ASSERT(ret != -1); // this really shouldn't fail
+        return (size_t) ret;
+    }
+
+    void seek(size_t offset, int whence) {
+#ifdef _WIN32
+        int ret = _fseeki64(fp, (__int64) offset, whence);
+#else
+        int ret = std::fseek(fp, (long) offset, whence);
+#endif
+        GGML_ASSERT(ret == 0); // same
+    }
+};
+
+#if defined(_WIN32)
+static std::string gguf_format_win_err(DWORD err) {
+    LPSTR buf;
+    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
+                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
+    if (!size) {
+        return "FormatMessageA failed";
+    }
+    std::string ret(buf, size);
+    LocalFree(buf);
+    return ret;
+}
+#endif
+
+struct gguf_mmap {
+    void * addr;
+    size_t size;
+
+    gguf_mmap(const gguf_mmap &) = delete;
+
+#ifdef _POSIX_MAPPED_FILES
+    static constexpr bool SUPPORTED = true;
+
+    gguf_mmap(struct gguf_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
+        size = file->size;
+        int fd = fileno(file->fp);
+        int flags = MAP_SHARED;
+        // prefetch/readahead impairs performance on NUMA systems
+        if (numa) { prefetch = 0; }
+#ifdef __linux__
+        if (prefetch) { flags |= MAP_POPULATE; }
+#endif
+        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
+        if (addr == MAP_FAILED) {
+            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
+        }
+
+        if (prefetch > 0) {
+            // Advise the kernel to preload the mapped memory
+            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
+                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+        if (numa) {
+            // advise the kernel not to use readahead
+            // (because the next page might not belong on the same node)
+            if (madvise(addr, file->size, MADV_RANDOM)) {
+                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+    }
+
+    ~gguf_mmap() {
+        munmap(addr, size);
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    gguf_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) numa;
+
+        size = file->size;
+
+        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
+
+        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
+        DWORD error = GetLastError();
+
+        if (hMapping == NULL) {
+            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
+        error = GetLastError();
+        CloseHandle(hMapping);
+
+        if (addr == NULL) {
+            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+        if (prefetch) {
+            // Advise the kernel to preload the mapped memory
+            WIN32_MEMORY_RANGE_ENTRY range;
+            range.VirtualAddress = addr;
+            range.NumberOfBytes = (SIZE_T)size;
+            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
+                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+            }
+        }
+        #else
+        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
+        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
+    }
+
+    ~gguf_mmap() {
+        if (!UnmapViewOfFile(addr)) {
+            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    gguf_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
+        (void) prefetch;
+        (void) numa;
+
+        throw std::runtime_error(std::string("mmap not supported"));
+    }
+#endif
+};
+
+// Represents some region of memory being locked using mlock or VirtualLock;
+// will automatically unlock on destruction.
+struct gguf_mlock {
+    void * addr = NULL;
+    size_t size = 0;
+    bool failed_already = false;
+
+    gguf_mlock() {}
+    gguf_mlock(const gguf_mlock &) = delete;
+
+    ~gguf_mlock() {
+        if (size) {
+            raw_unlock(addr, size);
+        }
+    }
+
+    void init(void * ptr) {
+        GGML_ASSERT(addr == NULL && size == 0);
+        addr = ptr;
+    }
+
+    void grow_to(size_t target_size) {
+        GGML_ASSERT(addr);
+        if (failed_already) {
+            return;
+        }
+        size_t granularity = lock_granularity();
+        target_size = (target_size + granularity - 1) & ~(granularity - 1);
+        if (target_size > size) {
+            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
+                size = target_size;
+            } else {
+                failed_already = true;
+            }
+        }
+    }
+
+#ifdef _POSIX_MEMLOCK_RANGE
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        return (size_t) sysconf(_SC_PAGESIZE);
+    }
+
+    #ifdef __APPLE__
+        #define MLOCK_SUGGESTION \
+            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
+            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
+    #else
+        #define MLOCK_SUGGESTION \
+            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
+    #endif
+
+    bool raw_lock(const void * addr, size_t size) {
+        if (!mlock(addr, size)) {
+            return true;
+        } else {
+            char* errmsg = std::strerror(errno);
+            bool suggest = (errno == ENOMEM);
+
+            // Check if the resource limit is fine after all
+            struct rlimit lock_limit;
+            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
+                suggest = false;
+            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
+                suggest = false;
+
+            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
+                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
+            return false;
+        }
+    }
+
+    #undef MLOCK_SUGGESTION
+
+    void raw_unlock(void * addr, size_t size) {
+        if (munlock(addr, size)) {
+            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
+        }
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        SYSTEM_INFO si;
+        GetSystemInfo(&si);
+        return (size_t) si.dwPageSize;
+    }
+
+    bool raw_lock(void * ptr, size_t len) {
+        for (int tries = 1; ; tries++) {
+            if (VirtualLock(ptr, len)) {
+                return true;
+            }
+            if (tries == 2) {
+                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
+                    len, size, llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+
+            // It failed but this was only the first try; increase the working
+            // set size and try again.
+            SIZE_T min_ws_size, max_ws_size;
+            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
+                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+            // Per MSDN: "The maximum number of pages that a process can lock
+            // is equal to the number of pages in its minimum working set minus
+            // a small overhead."
+            // Hopefully a megabyte is enough overhead:
+            size_t increment = len + 1048576;
+            // The minimum must be <= the maximum, so we need to increase both:
+            min_ws_size += increment;
+            max_ws_size += increment;
+            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
+                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+        }
+    }
+
+    void raw_unlock(void * ptr, size_t len) {
+        if (!VirtualUnlock(ptr, len)) {
+            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
+                    gguf_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    size_t lock_granularity() {
+        return (size_t) 65536;
+    }
+
+    bool raw_lock(const void * addr, size_t len) {
+        fprintf(stderr, "warning: mlock not supported on this system\n");
+        return false;
+    }
+
+    void raw_unlock(const void * addr, size_t len) {}
+#endif
+};
+
+// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
+struct gguf_buffer {
+    uint8_t * addr = NULL;
+    size_t size = 0;
+
+    gguf_buffer() = default;
+
+    void resize(size_t len) {
+#ifdef GGML_USE_METAL
+        free(addr);
+        int result = posix_memalign((void **) &addr, getpagesize(), len);
+        if (result == 0) {
+            memset(addr, 0, len);
+        }
+        else {
+            addr = NULL;
+        }
+#else
+        delete[] addr;
+        addr = new uint8_t[len];
+#endif
+        size = len;
+    }
+
+    ~gguf_buffer() {
+#ifdef GGML_USE_METAL
+        free(addr);
+#else
+        delete[] addr;
+#endif
+        addr = NULL;
+    }
+
+    // disable copy and move
+    gguf_buffer(const gguf_buffer&) = delete;
+    gguf_buffer(gguf_buffer&&) = delete;
+    gguf_buffer& operator=(const gguf_buffer&) = delete;
+    gguf_buffer& operator=(gguf_buffer&&) = delete;
+};
+
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+struct gguf_ctx_buffer {
+    uint8_t * addr = NULL;
+    bool is_cuda;
+    size_t size = 0;
+
+    gguf_ctx_buffer() = default;
+
+    void resize(size_t size) {
+        free();
+
+        addr = (uint8_t *) ggml_cuda_host_malloc(size);
+        if (addr) {
+            is_cuda = true;
+        }
+        else {
+            // fall back to pageable memory
+            addr = new uint8_t[size];
+            is_cuda = false;
+        }
+        this->size = size;
+    }
+
+    void free() {
+        if (addr) {
+            if (is_cuda) {
+                ggml_cuda_host_free(addr);
+            }
+            else {
+                delete[] addr;
+            }
+        }
+        addr = NULL;
+    }
+
+    ~gguf_ctx_buffer() {
+        free();
+    }
+
+    // disable copy and move
+    gguf_ctx_buffer(const gguf_ctx_buffer&) = delete;
+    gguf_ctx_buffer(gguf_ctx_buffer&&) = delete;
+    gguf_ctx_buffer& operator=(const gguf_ctx_buffer&) = delete;
+    gguf_ctx_buffer& operator=(gguf_ctx_buffer&&) = delete;
+};
+#else
+typedef gguf_buffer gguf_ctx_buffer;
+#endif
+
+#endif

From c3a65c4bbe1cd64cd6255e70b5c7563fb977959f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Wed, 2 Aug 2023 11:16:23 +0300
Subject: [PATCH 067/242] gguf-util.h : update note

---
 gguf-util.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gguf-util.h b/gguf-util.h
index 23f411bb0246b..74d6e61f792d0 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -1,5 +1,6 @@
 // GGUF counterpart of llama-util.h.
 // we may consider making it a part of ggml.c once GGUF work is complete.
+// this will require extra work to migrate this to pure C.
 // Contains wrappers around OS interfaces.
 
 #ifndef GGUF_UTIL_H

From e1e9b285477670d0ee69474f3e5b339a5b4741bf Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 2 Aug 2023 11:15:33 +0200
Subject: [PATCH 068/242] convert-llama-h5-to-gguf.py : accumulate kv / ti +
 special tokens

---
 convert-llama-h5-to-gguf.py | 120 +++++++++++++++++++++++-------------
 1 file changed, 76 insertions(+), 44 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 7f16905959b73..7a5dae77a249b 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -11,8 +11,10 @@
 from sentencepiece import SentencePieceProcessor
 
 
-NDArray = np.ndarray[Any, Any]
+#NDArray = np.ndarray[Any, Any]
 
+# compatible with python < 3.9
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
 def permute(weights: NDArray, n_head: int) -> NDArray:
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
@@ -57,51 +59,36 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
 
-# count tensors to be converted
-tensor_count = 0
-for name in list_vars.keys():
-    # we don't need these
-    if name.endswith(".rotary_emb.inv_freq"):
-        continue
-    tensor_count += 1
-
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-# This must be changed when adding/deleting kv
-kv_count = 13
 
-print("tensors " + str(tensor_count) + " kv " + str(kv_count))
-
-print("write gguf header")
-
-gguf_writer.write_header(tensor_count, kv_count)
-
-print("write gguf hparams")
+print("gguf: add key-values, metadata")
 
 llm_arch = "llama"
 
-gguf_writer.write_name("llama2-7b")
-gguf_writer.write_description("gguf test model")
-gguf_writer.write_architecture(llm_arch)
-gguf_writer.write_context_length(llm_arch, hparams["max_position_embeddings"])
-gguf_writer.write_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.write_layer_count(llm_arch, hparams["num_hidden_layers"])
-gguf_writer.write_feed_forward_length(llm_arch, hparams["intermediate_size"])
-gguf_writer.write_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
-gguf_writer.write_head_count(llm_arch, hparams["num_attention_heads"])
-gguf_writer.write_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
+gguf_writer.add_name("llama2-7b")
+gguf_writer.add_description("gguf test model")
+gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
+gguf_writer.add_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
+gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
+gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
 
 
 # TOKENIZATION
 
-print("write gguf tokenizer")
+print("gguf: add key-values, tokenizer")
 
 tokens: List[str] = []
 scores: List[float] = []
 
 if Path(dir_model + "/tokenizer.model").is_file():
     # vocab type sentencepiece
-    print("Adding sentencepiece tokenizer vocab.")
+    print("gguf: adding sentencepiece tokenizer vocab")
+
     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
     for i in range(tokenizer.vocab_size()):
@@ -123,14 +110,52 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         tokens.append(text)
         scores.append(score)
 
-gguf_writer.write_tokenizer_model("llama")
-gguf_writer.write_token_list(tokens)
-gguf_writer.write_token_scores(scores)
+    gguf_writer.add_tokenizer_model("llama")
+    gguf_writer.add_token_list(tokens)
+    gguf_writer.add_token_scores(scores)
+
+if Path(dir_model + "/tokenizer.json").is_file():
+    with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
+        tokenizer = json.load(f)
+
+    if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
+        print("gguf: adding special token ids")
+
+        with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
+            tokenizer_config = json.load(f)
+
+        # find special token ids
+
+        if "bos_token" in tokenizer_config and tokenizer_config["bos_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["bos_token"] or key["content"] == tokenizer_config["bos_token"]["content"]:
+                    gguf_writer.add_bos_token_id(key["id"])
+
+        if "eos_token" in tokenizer_config and tokenizer_config["eos_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["eos_token"] or key["content"] == tokenizer_config["eos_token"]["content"]:
+                    gguf_writer.add_eos_token_id(key["id"])
+
+        if "unk_token" in tokenizer_config and tokenizer_config["unk_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["unk_token"] or key["content"] == tokenizer_config["unk_token"]["content"]:
+                    gguf_writer.add_unk_token_id(key["id"])
+
+        if "sep_token" in tokenizer_config and tokenizer_config["sep_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["sep_token"] or key["content"] == tokenizer_config["sep_token"]["content"]:
+                    gguf_writer.add_sep_token_id(key["id"])
+
+        if "pad_token" in tokenizer_config and tokenizer_config["pad_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["pad_token"] or key["content"] == tokenizer_config["pad_token"]["content"]:
+                    gguf_writer.add_pad_token_id(key["id"])
+
 
 # TENSORS
 
 # tensor info
-print("write gguf tensor info")
+print("gguf: add gguf tensor info")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -197,24 +222,31 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
             data = data.astype(np.float32)
             ftype_cur = 0
 
-    gguf_writer.write_tensor_info(name, data)
+    gguf_writer.add_tensor_info(name, data)
+
 
+print("gguf: write header")
+gguf_writer.write_header_to_file()
+print("gguf: write key-values")
+gguf_writer.write_kv_data_to_file()
+print("gguf: write tensor info")
+gguf_writer.write_ti_data_to_file()
 
 # tensor data
-print("write gguf tensor data")
+print("gguf: write tensor data")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
-    print("Process tensor: " + name + " with shape: ", data.shape)
+#    print("Process tensor: " + name + " with shape: ", data.shape)
 
     # we don't need these
     if name.endswith(".rotary_emb.inv_freq"):
-        print("  Skip tensor: " + name)
+#        print("  Skip tensor: " + name)
         continue
 
     # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        print("  Permute tensor: " + name)
+#        print("  Permute tensor: " + name)
         data = permute(data, hparams["num_attention_heads"])
 
     n_dims = len(data.shape)
@@ -223,23 +255,23 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     ftype_cur = 0
     if ftype != 0:
         if name.endswith(".weight") and n_dims == 2:
-            print("  Converting to float16")
+#            print("  Converting to float16")
             data = data.astype(np.float16)
             ftype_cur = 1
         else:
-            print("  Converting to float32")
+#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
     else:
         if data.dtype != np.float32:
-            print("  Converting to float32")
+#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
 
-    gguf_writer.write_tensor(data)
+    gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 
 
-print("Done. Output file: " + fname_out)
+print("gguf: conversion done, output file: " + fname_out)
 print("")

From c5ba5efda2a3b3b9ee3d42d908dde2b92dc80bc1 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 2 Aug 2023 11:26:07 +0200
Subject: [PATCH 069/242] convert-llama-h5-to-gguf.py : special tokens

---
 convert-llama-h5-to-gguf.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 7a5dae77a249b..67b3c55d4e35f 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -128,27 +128,27 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
         if "bos_token" in tokenizer_config and tokenizer_config["bos_token"] != None:
             for key in tokenizer["added_tokens"]:
-                if key["content"] == tokenizer_config["bos_token"] or key["content"] == tokenizer_config["bos_token"]["content"]:
+                if key["content"] == tokenizer_config["bos_token"]["content"]:
                     gguf_writer.add_bos_token_id(key["id"])
 
         if "eos_token" in tokenizer_config and tokenizer_config["eos_token"] != None:
             for key in tokenizer["added_tokens"]:
-                if key["content"] == tokenizer_config["eos_token"] or key["content"] == tokenizer_config["eos_token"]["content"]:
+                if key["content"] == tokenizer_config["eos_token"]["content"]:
                     gguf_writer.add_eos_token_id(key["id"])
 
         if "unk_token" in tokenizer_config and tokenizer_config["unk_token"] != None:
             for key in tokenizer["added_tokens"]:
-                if key["content"] == tokenizer_config["unk_token"] or key["content"] == tokenizer_config["unk_token"]["content"]:
+                if key["content"] == tokenizer_config["unk_token"]["content"]:
                     gguf_writer.add_unk_token_id(key["id"])
 
         if "sep_token" in tokenizer_config and tokenizer_config["sep_token"] != None:
             for key in tokenizer["added_tokens"]:
-                if key["content"] == tokenizer_config["sep_token"] or key["content"] == tokenizer_config["sep_token"]["content"]:
+                if key["content"] == tokenizer_config["sep_token"]["content"]:
                     gguf_writer.add_sep_token_id(key["id"])
 
         if "pad_token" in tokenizer_config and tokenizer_config["pad_token"] != None:
             for key in tokenizer["added_tokens"]:
-                if key["content"] == tokenizer_config["pad_token"] or key["content"] == tokenizer_config["pad_token"]["content"]:
+                if key["content"] == tokenizer_config["pad_token"]["content"]:
                     gguf_writer.add_pad_token_id(key["id"])
 
 

From 23abbe8e00b14d529778195c404005f0740e6d6b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 03:51:43 +0200
Subject: [PATCH 070/242] Delete gptneox-common.cpp

---
 gptneox-common.cpp | 601 ---------------------------------------------
 1 file changed, 601 deletions(-)
 delete mode 100644 gptneox-common.cpp

diff --git a/gptneox-common.cpp b/gptneox-common.cpp
deleted file mode 100644
index 9dee0cb9ce8b2..0000000000000
--- a/gptneox-common.cpp
+++ /dev/null
@@ -1,601 +0,0 @@
-#include "gptneox-common.h"
-
-#include <cmath>
-#include <cstring>
-#include <fstream>
-#include <regex>
-#include <locale>
-#include <codecvt>
-#include <sstream>
-
-#if defined(_MSC_VER)
-#pragma warning(disable: 4244 4267) // possible loss of data
-#endif
-
-// Function to check if the next argument exists
-std::string get_next_arg(int& i, int argc, char** argv, const std::string& flag, gpt_params& params) {
-    if (i + 1 < argc && argv[i + 1][0] != '-') {
-        return argv[++i];
-    } else {
-        fprintf(stderr, "error: %s requires one argument.\n", flag.c_str());
-        gpt_print_usage(argc, argv, params);
-        exit(0);
-    }
-}
-
-bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
-    for (int i = 1; i < argc; i++) {
-        std::string arg = argv[i];
-
-        if (arg == "-s" || arg == "--seed") {
-            params.seed = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-t" || arg == "--threads") {
-            params.n_threads = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-ngl" || arg == "--gpu-layers" || arg == "--n-gpu-layers") {
-            params.n_gpu_layers = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-p" || arg == "--prompt") {
-            params.prompt = get_next_arg(i, argc, argv, arg, params);
-        } else if (arg == "-n" || arg == "--n_predict") {
-            params.n_predict = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "--top_k") {
-            params.top_k = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "--top_p") {
-            params.top_p = std::stof(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "--temp") {
-            params.temp = std::stof(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "--repeat-last-n") {
-            params.repeat_last_n = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "--repeat-penalty") {
-            params.repeat_penalty = std::stof(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-b" || arg == "--batch_size") {
-            params.n_batch= std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-m" || arg == "--model") {
-            params.model = get_next_arg(i, argc, argv, arg, params);
-        } else if (arg == "-i" || arg == "--interactive") {
-            params.interactive = true;
-        } else if (arg == "-ip" || arg == "--interactive-port") {
-            params.interactive = true;
-            params.interactive_port = std::stoi(get_next_arg(i, argc, argv, arg, params));
-        } else if (arg == "-h" || arg == "--help") {
-            gpt_print_usage(argc, argv, params);
-            exit(0);
-        } else if (arg == "-f" || arg == "--file") {
-            get_next_arg(i, argc, argv, arg, params);
-            std::ifstream file(argv[i]);
-            if (!file) {
-                fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
-                break;
-            }
-            std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
-            if (params.prompt.back() == '\n') {
-                params.prompt.pop_back();
-            }
-        } else if (arg == "-tt" || arg == "--token_test") {
-            params.token_test = get_next_arg(i, argc, argv, arg, params);
-        }
-        else {
-            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
-            gpt_print_usage(argc, argv, params);
-            exit(0);
-        }
-    }
-
-    return true;
-}
-
-void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
-    fprintf(stderr, "usage: %s [options]\n", argv[0]);
-    fprintf(stderr, "\n");
-    fprintf(stderr, "options:\n");
-    fprintf(stderr, "  -h, --help            show this help message and exit\n");
-    fprintf(stderr, "  -s SEED, --seed SEED  RNG seed (default: -1)\n");
-    fprintf(stderr, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
-    fprintf(stderr, "  -ngl N, --gpu-layers N  number of layers to offload to GPU on supported models (default: %d)\n", params.n_gpu_layers);
-    fprintf(stderr, "  -p PROMPT, --prompt PROMPT\n");
-    fprintf(stderr, "                        prompt to start generation with (default: random)\n");
-    fprintf(stderr, "  -f FNAME, --file FNAME\n");
-    fprintf(stderr, "                        load prompt from a file\n");
-    fprintf(stderr, "  -tt TOKEN_TEST, --token_test TOKEN_TEST\n");
-    fprintf(stderr, "                        test tokenization\n");
-    fprintf(stderr, "  -n N, --n_predict N   number of tokens to predict (default: %d)\n", params.n_predict);
-    fprintf(stderr, "  --top_k N             top-k sampling (default: %d)\n", params.top_k);
-    fprintf(stderr, "  --top_p N             top-p sampling (default: %.1f)\n", params.top_p);
-    fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
-    fprintf(stderr, "  --repeat-last-n N     last n tokens to consider for penalize (default: %d, 0 = disabled)\n", params.repeat_last_n);
-    fprintf(stderr, "  --repeat-penalty N    penalize repeat sequence of tokens (default: %.2f, 1.0 = disabled)\n", (double)params.repeat_penalty);
-    fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
-    fprintf(stderr, "  -m FNAME, --model FNAME\n");
-    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
-    fprintf(stderr, "\n");
-}
-
-std::string gpt_random_prompt(std::mt19937 & rng) {
-    const int r = rng() % 10;
-    switch (r) {
-        case 0: return "So";
-        case 1: return "Once upon a time";
-        case 2: return "When";
-        case 3: return "The";
-        case 4: return "After";
-        case 5: return "If";
-        case 6: return "import";
-        case 7: return "He";
-        case 8: return "She";
-        case 9: return "They";
-        default: return "To";
-    }
-
-    return "The";
-}
-
-std::string trim(const std::string & s) {
-    std::regex e("^\\s+|\\s+$");
-    return std::regex_replace(s, e, "");
-}
-
-std::string replace(const std::string & s, const std::string & from, const std::string & to) {
-    std::string result = s;
-    size_t pos = 0;
-    while ((pos = result.find(from, pos)) != std::string::npos) {
-        result.replace(pos, from.length(), to);
-        pos += to.length();
-    }
-    return result;
-}
-
-void gpt_vocab::add_special_token(const std::string & token) {
-    special_tokens.push_back(token);
-}
-
-std::map<std::string, int32_t> json_parse(const std::string & fname) {
-    std::map<std::string, int32_t> result;
-
-    // read file into string
-    std::string json;
-    {
-        std::ifstream ifs(fname);
-        if (!ifs) {
-            fprintf(stderr, "Failed to open %s\n", fname.c_str());
-            exit(1);
-        }
-
-        json = std::string((std::istreambuf_iterator<char>(ifs)),
-                (std::istreambuf_iterator<char>()));
-    }
-
-    if (json[0] != '{') {
-        return result;
-    }
-
-    // parse json
-    {
-        bool has_key  = false;
-        bool in_token = false;
-
-        std::string str_key = "";
-        std::string str_val = "";
-
-        int n = json.size();
-        for (int i = 1; i < n; ++i) {
-            if (!in_token) {
-                if (json[i] == ' ') continue;
-                if (json[i] == '"') {
-                    in_token = true;
-                    continue;
-                }
-            } else {
-                if (json[i] == '\\' && i+1 < n) {
-                    if (has_key == false) {
-                        str_key += json[i];
-                    } else {
-                        str_val += json[i];
-                    }
-                    ++i;
-                } else if (json[i] == '"') {
-                    if (has_key == false) {
-                        has_key = true;
-                        ++i;
-                        while (json[i] == ' ') ++i;
-                        ++i; // :
-                        while (json[i] == ' ') ++i;
-                        if (json[i] != '\"') {
-                            while (json[i] != ',' && json[i] != '}') {
-                                str_val += json[i++];
-                            }
-                            has_key = false;
-                        } else {
-                            in_token = true;
-                            continue;
-                        }
-                    } else {
-                        has_key = false;
-                    }
-
-                    str_key = ::replace(str_key, "\\u0120", " " ); // \u0120 -> space
-                    str_key = ::replace(str_key, "\\u010a", "\n"); // \u010a -> new line
-                    str_key = ::replace(str_key, "\\\"",    "\""); // \\\"   -> "
-
-                    try {
-                        result[str_key] = std::stoi(str_val);
-                    } catch (...) {
-                        //fprintf(stderr, "%s: ignoring key '%s' with value '%s'\n", fname.c_str(), str_key.c_str(), str_val.c_str());
-
-                    }
-                    str_key = "";
-                    str_val = "";
-                    in_token = false;
-                    continue;
-                }
-                if (has_key == false) {
-                    str_key += json[i];
-                } else {
-                    str_val += json[i];
-                }
-            }
-        }
-    }
-
-    return result;
-}
-
-std::string convert_to_utf8(const std::wstring & input) {
-    std::wstring_convert<std::codecvt_utf8<wchar_t>> converter;
-    return converter.to_bytes(input);
-}
-
-
-std::wstring convert_to_wstring(const std::string & input) {
-    std::wstring_convert<std::codecvt_utf8<wchar_t>> converter;
-    return converter.from_bytes(input);
-}
-
-void gpt_split_words(std::string str, std::vector<std::string>& words) {
-    const std::string pattern = R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)";
-    const std::regex re(pattern);
-    std::smatch m;
-
-    while (std::regex_search(str, m, re)) {
-        for (auto x : m) {
-            words.push_back(x);
-        }
-        str = m.suffix();
-    }
-}
-
-std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text) {
-    std::vector<std::string> words;
-
-    // first split the text into words
-    {
-        std::string str = text;
-
-        // Generate the subpattern from the special_tokens vector if it's not empty
-        if (!vocab.special_tokens.empty()) {
-            const std::regex escape(R"([\[\\\^\$\.\|\?\*\+\(\)\{\}])");
-            std::string special_tokens_subpattern;
-            for (const auto & token : vocab.special_tokens) {
-                if (!special_tokens_subpattern.empty()) {
-                    special_tokens_subpattern += "|";
-                }
-                special_tokens_subpattern += std::regex_replace(token, escape, R"(\$&)");
-            }
-
-            std::regex re(special_tokens_subpattern);
-            std::smatch m;
-            // Split the text by special tokens.
-            while (std::regex_search(str, m, re)) {
-                // Split the substrings in-between special tokens into words.
-                gpt_split_words(m.prefix(), words);
-                // Add matched special tokens as words.
-                for (auto x : m) {
-                    words.push_back(x);
-                }
-                str = m.suffix();
-            }
-            // Remaining text without special tokens will be handled below.
-        }
-
-        gpt_split_words(str, words);
-    }
-
-    // find the longest token that forms each word in words:
-    std::vector<gpt_vocab::id> tokens;
-    for (const auto & word : words) {
-        for (int i = 0; i < (int) word.size(); ){
-            for (int j = word.size() - 1; j >= i; j--){
-                auto cand = word.substr(i, j-i+1);
-                auto it = vocab.token_to_id.find(cand);
-                if (it != vocab.token_to_id.end()){ // word.substr(i, j-i+1) in vocab
-                    tokens.push_back(it->second);
-                    i = j + 1;
-                    break;
-                }
-                else if (j == i){ // word.substr(i, 1) has no matching
-                    fprintf(stderr, "%s: unknown token '%s'\n", __func__, word.substr(i, 1).data());
-                    i++;
-                }
-            }
-        }
-    }
-
-    return tokens;
-}
-
-std::vector<gpt_vocab::id> parse_tokens_from_string(const std::string& input, char delimiter) {
-    std::vector<gpt_vocab::id> output;
-    std::stringstream ss(input);
-    std::string token;
-
-    while (std::getline(ss, token, delimiter)) {
-        output.push_back(std::stoi(token));
-    }
-
-    return output;
-}
-
-std::map<std::string, std::vector<gpt_vocab::id>> extract_tests_from_file(const std::string & fpath_test){
-    if (fpath_test.empty()){
-        fprintf(stderr, "%s : No test file found.\n", __func__);
-        return std::map<std::string, std::vector<gpt_vocab::id>>();
-    }
-
-    std::map<std::string, std::vector<gpt_vocab::id>> tests;
-
-    auto fin = std::ifstream(fpath_test, std::ios_base::in);
-    const char * delimeter = " => ";
-    const char del_tok = ',';
-    std::string line;
-    while (std::getline(fin, line)) {
-        size_t delimiterPos = line.find(delimeter);
-        if (delimiterPos != std::string::npos) {
-            std::string text = line.substr(0, delimiterPos);
-            std::string s_tokens = line.substr(delimiterPos + std::strlen(delimeter));
-            tests[text] = parse_tokens_from_string(s_tokens, del_tok);
-        }
-    }
-    return tests;
-}
-
-void test_gpt_tokenizer(gpt_vocab & vocab, const std::string & fpath_test){
-    std::map<std::string, std::vector<gpt_vocab::id>> tests = extract_tests_from_file(fpath_test);
-
-    size_t n_fails = 0;
-
-    for (const auto & test : tests) {
-        std::vector<gpt_vocab::id> tokens = gpt_tokenize(vocab, test.first);
-
-        if (tokens != test.second){
-            n_fails++;
-
-            // print out failure cases
-            fprintf(stderr, "%s : failed test: '%s'\n", __func__, test.first.c_str());
-            fprintf(stderr, "%s : tokens in hf:   ", __func__);
-            for (const auto & t : test.second) {
-                fprintf(stderr, "%s(%d), ", vocab.id_to_token[t].c_str(), t);
-            }
-            fprintf(stderr, "\n");
-            fprintf(stderr, "%s : tokens in ggml: ", __func__);
-            for (const auto & t : tokens) {
-                fprintf(stderr, "%s(%d), ", vocab.id_to_token[t].c_str(), t);
-            }
-            fprintf(stderr, "\n");
-        }
-    }
-
-    fprintf(stderr, "%s : %zu tests failed out of %zu tests.\n", __func__, n_fails, tests.size());
-}
-
-bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab) {
-    printf("%s: loading vocab from '%s'\n", __func__, fname.c_str());
-
-    vocab.token_to_id = ::json_parse(fname);
-
-    for (const auto & kv : vocab.token_to_id) {
-        vocab.id_to_token[kv.second] = kv.first;
-    }
-
-    printf("%s: vocab size = %d\n", __func__, (int) vocab.token_to_id.size());
-
-    // print the vocabulary
-    //for (auto kv : vocab.token_to_id) {
-    //    printf("'%s' -> %d\n", kv.first.data(), kv.second);
-    //}
-
-    return true;
-}
-
-gpt_vocab::id gpt_sample_top_k_top_p(
-        const gpt_vocab & vocab,
-        const float * logits,
-        int    top_k,
-        double top_p,
-        double temp,
-        std::mt19937 & rng) {
-    int n_logits = vocab.id_to_token.size();
-
-    std::vector<std::pair<double, gpt_vocab::id>> logits_id;
-    logits_id.reserve(n_logits);
-
-    {
-        const double scale = 1.0/temp;
-        for (int i = 0; i < n_logits; ++i) {
-            logits_id.push_back(std::make_pair(logits[i]*scale, i));
-        }
-    }
-
-    // find the top K tokens
-    std::partial_sort(
-            logits_id.begin(),
-            logits_id.begin() + top_k, logits_id.end(),
-            [](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
-        return a.first > b.first;
-    });
-
-    logits_id.resize(top_k);
-
-    double maxl = -INFINITY;
-    for (const auto & kv : logits_id) {
-        maxl = std::max(maxl, kv.first);
-    }
-
-    // compute probs for the top K tokens
-    std::vector<double> probs;
-    probs.reserve(logits_id.size());
-
-    double sum = 0.0;
-    for (const auto & kv : logits_id) {
-        double p = exp(kv.first - maxl);
-        probs.push_back(p);
-        sum += p;
-    }
-
-    // normalize the probs
-    for (auto & p : probs) {
-        p /= sum;
-    }
-
-    if (top_p < 1.0f) {
-        double cumsum = 0.0f;
-        for (int i = 0; i < top_k; i++) {
-            cumsum += probs[i];
-            if (cumsum >= top_p) {
-                top_k = i + 1;
-                probs.resize(top_k);
-                logits_id.resize(top_k);
-                break;
-            }
-        }
-
-        cumsum = 1.0/cumsum;
-        for (int i = 0; i < (int) probs.size(); i++) {
-            probs[i] *= cumsum;
-        }
-    }
-
-    //printf("\n");
-    //for (int i = 0; i < (int) probs.size(); i++) {
-    //    printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
-    //}
-    //exit(0);
-
-    std::discrete_distribution<> dist(probs.begin(), probs.end());
-    int idx = dist(rng);
-
-    return logits_id[idx].second;
-}
-
-gpt_vocab::id gpt_sample_top_k_top_p_repeat(
-        const gpt_vocab & vocab,
-        const float * logits,
-        const int32_t * last_n_tokens_data,
-        size_t last_n_tokens_data_size,
-        int    top_k,
-        double top_p,
-        double temp,
-        int repeat_last_n,
-        float repeat_penalty,
-        std::mt19937 & rng) {
-
-    int n_logits = vocab.id_to_token.size();
-
-    const auto * plogits = logits;
-
-    const auto last_n_tokens = std::vector<int32_t>(last_n_tokens_data, last_n_tokens_data + last_n_tokens_data_size);
-
-    if (temp <= 0) {
-        // select the token with the highest logit directly
-        float max_logit = plogits[0];
-        gpt_vocab::id max_id = 0;
-
-        for (int i = 1; i < n_logits; ++i) {
-            if (plogits[i] > max_logit) {
-                max_logit = plogits[i];
-                max_id = i;
-            }
-        }
-        return max_id;
-    }
-
-
-    std::vector<std::pair<double, gpt_vocab::id>> logits_id;
-    logits_id.reserve(n_logits);
-
-    {
-        const float scale = 1.0f/temp;
-        for (int i = 0; i < n_logits; ++i) {
-            // repetition penalty from ctrl paper (https://arxiv.org/abs/1909.05858)
-            // credit https://github.com/facebookresearch/llama/compare/main...shawwn:llama:main
-            if (repeat_last_n > 0 && std::find(last_n_tokens.end()-repeat_last_n, last_n_tokens.end(), i) != last_n_tokens.end()) {
-                // if score < 0 then repetition penalty has to multiplied to reduce the previous token probability
-                if (plogits[i] < 0.0f) {
-                    logits_id.push_back(std::make_pair(plogits[i]*scale*repeat_penalty, i));
-                } else {
-                    logits_id.push_back(std::make_pair(plogits[i]*scale/repeat_penalty, i));
-                }
-            } else {
-                logits_id.push_back(std::make_pair(plogits[i]*scale, i));
-            }
-        }
-    }
-
-    // find the top K tokens
-    std::partial_sort(
-            logits_id.begin(),
-            logits_id.begin() + top_k, logits_id.end(),
-            [](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
-        return a.first > b.first;
-    });
-
-    logits_id.resize(top_k);
-
-    double maxl = -INFINITY;
-    for (const auto & kv : logits_id) {
-        maxl = std::max(maxl, kv.first);
-    }
-
-    // compute probs for the top K tokens
-    std::vector<double> probs;
-    probs.reserve(logits_id.size());
-
-    double sum = 0.0;
-    for (const auto & kv : logits_id) {
-        double p = exp(kv.first - maxl);
-        probs.push_back(p);
-        sum += p;
-    }
-
-    // normalize the probs
-    for (auto & p : probs) {
-        p /= sum;
-    }
-
-    if (top_p < 1.0f) {
-        double cumsum = 0.0f;
-        for (int i = 0; i < top_k; i++) {
-            cumsum += probs[i];
-            if (cumsum >= top_p) {
-                top_k = i + 1;
-                probs.resize(top_k);
-                logits_id.resize(top_k);
-                break;
-            }
-        }
-
-        cumsum = 1.0/cumsum;
-        for (int i = 0; i < (int) probs.size(); i++) {
-            probs[i] *= cumsum;
-        }
-    }
-
-//    printf("\n");
-//    for (int i = 0; i < (int) probs.size(); i++) {
-//    for (int i = 0; i < 10; i++) {
-//        printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
-//    }
-
-    std::discrete_distribution<> dist(probs.begin(), probs.end());
-    int idx = dist(rng);
-
-    return logits_id[idx].second;
-
-}

From 6691aa879755d4142a88a62466856b31998ad0d4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 03:52:01 +0200
Subject: [PATCH 071/242] Delete gptneox-common.h

---
 gptneox-common.h | 125 -----------------------------------------------
 1 file changed, 125 deletions(-)
 delete mode 100644 gptneox-common.h

diff --git a/gptneox-common.h b/gptneox-common.h
deleted file mode 100644
index 60e5650c129ab..0000000000000
--- a/gptneox-common.h
+++ /dev/null
@@ -1,125 +0,0 @@
-// Various helper functions and utilities
-
-#pragma once
-
-#include <string>
-#include <map>
-#include <vector>
-#include <random>
-#include <thread>
-
-//
-// CLI argument parsing
-//
-
-struct gpt_params {
-    int32_t seed      = -1;  // RNG seed
-    int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
-    int32_t n_predict = 200; // new tokens to predict
-    int32_t n_batch   = 8;   // batch size for prompt processing
-
-    // sampling parameters
-    int32_t top_k          = 40;
-    float   top_p          = 0.9f;
-    float   temp           = 0.9f;
-    int32_t repeat_last_n  = 64;
-    float   repeat_penalty = 1.00f;
-
-    std::string model      = "models/gpt-2-117M/ggml-model.bin"; // model path
-    std::string prompt     = "";
-    std::string token_test = "";
-
-    bool    interactive      = false;
-    int32_t interactive_port = -1;
-
-    int32_t n_gpu_layers     = 0;
-};
-
-bool gpt_params_parse(int argc, char ** argv, gpt_params & params);
-
-void gpt_print_usage(int argc, char ** argv, const gpt_params & params);
-
-std::string gpt_random_prompt(std::mt19937 & rng);
-
-//
-// Vocab utils
-//
-
-std::string trim(const std::string & s);
-
-std::string replace(
-        const std::string & s,
-        const std::string & from,
-        const std::string & to);
-
-struct gpt_vocab {
-    using id    = int32_t;
-    using token = std::string;
-
-    std::map<token, id> token_to_id;
-    std::map<id, token> id_to_token;
-    std::vector<std::string> special_tokens;
-
-    void add_special_token(const std::string & token);
-};
-
-// poor-man's JSON parsing
-std::map<std::string, int32_t> json_parse(const std::string & fname);
-
-std::string convert_to_utf8(const std::wstring & input);
-
-std::wstring convert_to_wstring(const std::string & input);
-
-void gpt_split_words(std::string str, std::vector<std::string>& words);
-
-// split text into tokens
-//
-// ref: https://github.com/openai/gpt-2/blob/a74da5d99abaaba920de8131d64da2862a8f213b/src/encoder.py#L53
-//
-// Regex (Python):
-// r"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+"""
-//
-// Regex (C++):
-// R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)"
-//
-std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text);
-
-// test outputs of gpt_tokenize
-//
-//   - compare with tokens generated by the huggingface tokenizer
-//   - test cases are chosen based on the model's main language (under 'prompt' directory)
-//   - if all sentences are tokenized identically, print 'All tests passed.'
-//   - otherwise, print sentence, huggingface tokens, ggml tokens
-//
-void test_gpt_tokenizer(gpt_vocab & vocab, const std::string & fpath_test);
-
-// load the tokens from encoder.json
-bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab);
-
-// sample next token given probabilities for each embedding
-//
-//   - consider only the top K tokens
-//   - from them, consider only the top tokens with cumulative probability > P
-//
-// TODO: not sure if this implementation is correct
-// TODO: temperature is not implemented
-//
-gpt_vocab::id gpt_sample_top_k_top_p(
-        const gpt_vocab & vocab,
-        const float * logits,
-        int    top_k,
-        double top_p,
-        double temp,
-        std::mt19937 & rng);
-
-gpt_vocab::id gpt_sample_top_k_top_p_repeat(
-        const gpt_vocab & vocab,
-        const float * logits,
-        const int32_t * last_n_tokens_data,
-        size_t last_n_tokens_data_size,
-        int    top_k,
-        double top_p,
-        double temp,
-        int repeat_last_n,
-        float repeat_penalty,
-        std::mt19937 & rng);

From 2922280a1a6615e66902ecaf2f1dddc7e605b94a Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 03:55:23 +0200
Subject: [PATCH 072/242] convert-gptneox-h5-to-gguf.py : gpt2bpe tokenizer

---
 convert-gptneox-h5-to-gguf.py | 98 +++++++++++++++++++++++++----------
 1 file changed, 72 insertions(+), 26 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 4f9d41487d22f..066ed0da07697 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -1,14 +1,36 @@
 # Quick and dirty HF gptneox--> gguf conversion
 
 import gguf
+import os
 import sys
 import struct
 import json
 import numpy as np
 from typing import Any, List
 from pathlib import Path
-from transformers import AutoModelForCausalLM
-
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
+def bytes_to_unicode():
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a significant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1))
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8+n)
+            n += 1
+    cs = [chr(n) for n in cs]
+    return dict(zip(bs, cs))
 
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
@@ -20,7 +42,7 @@
 # output in the same directory as the model
 dir_model = sys.argv[1]
 fname_out = sys.argv[1] + "/ggml-model.bin"
-
+last_dir = os.path.basename(os.path.normpath(dir_model))
 
 # possible tensor data types
 #   ftype == 0 -> float32
@@ -37,6 +59,8 @@
         sys.exit(1)
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
+print("gguf: loading model "+last_dir)
+
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
     hparams = json.load(f)
 
@@ -44,17 +68,17 @@
     print("Model architecture not supported: " + hparams["architectures"][0] )
     sys.exit()
 
+
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
 list_vars = model.state_dict()
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-
-print("gguf: add key-values, metadata")
+print("gguf: add metadata")
 
 llm_arch = "gptneox"
 
-gguf_writer.add_name("pythia-70b-deduped")
+gguf_writer.add_name(last_dir)
 gguf_writer.add_description("gguf test model")
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
@@ -68,28 +92,55 @@
 
 # TOKENIZATION
 
-print("gguf: add key-values, tokenizer")
+print("gguf: add tokenizer")
 
 tokens: List[str] = []
 merges: List[str] = []
 
+
 if Path(dir_model + "/tokenizer.json").is_file():
-    # vocab type gpt2
-    print("gguf: adding gpt2 tokenizer vocab")
+    # gpt2 tokenizer
+    gguf_writer.add_tokenizer_model("gpt2")
+
+    print("gguf: adding gpt2 tokenizer merges")
 
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
-        tokenizer = json.load(f)
+        tokenizer_json = json.load(f)
+    merges = tokenizer_json["model"]["merges"]
 
-    for key in tokenizer["model"]["vocab"]:
-        tokens.append(key)
+    gguf_writer.add_token_merges(merges)
 
-    merges = tokenizer["model"]["merges"]
+    print("gguf: adding gpt2 tokenizer vocab")
+
+    vocab_size = len( tokenizer_json["model"]["vocab"] )
+
+    # from ggllm.cpp falcon_convert.py
+    tokenizer = AutoTokenizer.from_pretrained(dir_model)
+
+    reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()}
+    byte_encoder = bytes_to_unicode()
+    byte_decoder = {v:k for k, v in byte_encoder.items()}
+
+    for i in range(vocab_size):
+        if i in reverse_vocab:
+            try:
+                text = bytearray([byte_decoder[c] for c in reverse_vocab[i]])
+            except KeyError:
+                text = bytearray()
+                for c in reverse_vocab[i]:
+                    if ord(c) < 256:  # single byte character
+                        text.append(byte_decoder[ord(c)])
+                    else:  # multibyte special token character
+                        text.extend(c.encode('utf-8'))
+        else:
+            print(f"Key {i} not in tokenizer vocabulary. Padding with an arbitrary token.")
+            padding_token = f"[PAD{i}]".encode("utf8")
+            text = bytearray(padding_token)
+        tokens.append(text)
 
-    gguf_writer.add_tokenizer_model("gpt2")
     gguf_writer.add_token_list(tokens)
-    gguf_writer.add_token_merges(merges)
 
-    if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
+    if "added_tokens" in tokenizer_json and Path(dir_model + "/tokenizer_config.json").is_file():
         print("gguf: adding special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
@@ -98,27 +149,27 @@
         # find special token ids
 
         if "bos_token" in tokenizer_config:
-            for key in tokenizer["added_tokens"]:
+            for key in tokenizer_json["added_tokens"]:
                 if key["content"] == tokenizer_config["bos_token"]:
                     gguf_writer.add_bos_token_id(key["id"])
 
         if "eos_token" in tokenizer_config:
-            for key in tokenizer["added_tokens"]:
+            for key in tokenizer_json["added_tokens"]:
                 if key["content"] == tokenizer_config["eos_token"]:
                     gguf_writer.add_eos_token_id(key["id"])
 
         if "unk_token" in tokenizer_config:
-            for key in tokenizer["added_tokens"]:
+            for key in tokenizer_json["added_tokens"]:
                 if key["content"] == tokenizer_config["unk_token"]:
                     gguf_writer.add_unk_token_id(key["id"])
 
         if "sep_token" in tokenizer_config:
-            for key in tokenizer["added_tokens"]:
+            for key in tokenizer_json["added_tokens"]:
                 if key["content"] == tokenizer_config["sep_token"]:
                     gguf_writer.add_sep_token_id(key["id"])
 
         if "pad_token" in tokenizer_config:
-            for key in tokenizer["added_tokens"]:
+            for key in tokenizer_json["added_tokens"]:
                 if key["content"] == tokenizer_config["pad_token"]:
                     gguf_writer.add_pad_token_id(key["id"])
 
@@ -165,11 +216,9 @@
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
-#    print("Process tensor: " + name + " with shape: ", data.shape)
 
     # we don't need these
     if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-#        print("  Skip tensor: " + name)
         continue
 
     n_dims = len(data.shape)
@@ -178,16 +227,13 @@
     ftype_cur = 0
     if ftype != 0:
         if name.endswith(".weight") and n_dims == 2:
-#            print("  Converting to float16")
             data = data.astype(np.float16)
             ftype_cur = 1
         else:
-#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
     else:
         if data.dtype != np.float32:
-#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
 

From e6f19ba2404b21e84e1f9dbc5d1b7002a67694e6 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 03:56:37 +0200
Subject: [PATCH 073/242] gptneox-main.cpp : gpt2 bpe tokenizer

---
 gptneox-main.cpp | 372 +++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 329 insertions(+), 43 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 235fca7ce7094..0161d9c2ee7e8 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -1,6 +1,5 @@
 #include "ggml.h"
-
-#include "gptneox-common.h"
+#include "cmpnct_gpt2bpe.hpp"
 
 #include <cassert>
 #include <cmath>
@@ -11,6 +10,8 @@
 #include <map>
 #include <string>
 #include <vector>
+#include <thread>
+#include <random>
 
 #if defined(_MSC_VER)
 #pragma warning(disable: 4244 4267) // possible loss of data
@@ -20,11 +21,11 @@
 struct gpt_neox_hparams {
     size_t n_merges = 0;
     size_t n_vocab  = 0;
-    int32_t n_ctx    = 0;
-    int32_t n_embd   = 0;
-    int32_t n_head   = 0;
-    int32_t n_layer  = 0;
-    int32_t n_rot    = 0; // rotary_pct * (n_embd / n_head)
+    uint32_t n_ctx    = 0;
+    uint32_t n_embd   = 0;
+    uint32_t n_head   = 0;
+    uint32_t n_layer  = 0;
+    uint32_t n_rot    = 0; // rotary_pct * (n_embd / n_head)
     bool par_res = true;
     float norm_eps = 1e-5;
 };
@@ -78,6 +79,241 @@ struct gpt_neox_model {
     std::map<std::string, struct ggml_tensor *> tensors;
 };
 
+struct gpt_params {
+    int32_t seed      = -1;  // RNG seed
+    int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
+    uint32_t n_predict = 200; // new tokens to predict
+    uint32_t n_batch   = 512;   // batch size for prompt processing
+
+    // sampling parameters
+    int32_t top_k          = 40;
+    float top_p            = 1.0f;
+    float temp             = 0.8f;
+    int32_t repeat_last_n  = 64;
+    float repeat_penalty   = 1.02f;
+
+    std::string model      = ""; // model path
+    std::string prompt     = "";
+
+    std::string token_test = "";
+    bool    interactive      = false;
+    int32_t interactive_port = -1;
+    int32_t n_gpu_layers     = 0;
+};
+
+void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
+    fprintf(stderr, "usage: %s [options]\n", argv[0]);
+    fprintf(stderr, "\n");
+    fprintf(stderr, "options:\n");
+    fprintf(stderr, "  -h, --help            show this help message and exit\n");
+    fprintf(stderr, "  -s SEED, --seed SEED  RNG seed (default: -1)\n");
+    fprintf(stderr, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
+    fprintf(stderr, "  -ngl N, --gpu-layers N  number of layers to offload to GPU on supported models (default: %d)\n", params.n_gpu_layers);
+    fprintf(stderr, "  -p PROMPT, --prompt PROMPT\n");
+    fprintf(stderr, "                        prompt to start generation with (default: random)\n");
+    fprintf(stderr, "  -f FNAME, --file FNAME\n");
+    fprintf(stderr, "                        load prompt from a file\n");
+    fprintf(stderr, "  -tt TOKEN_TEST, --token_test TOKEN_TEST\n");
+    fprintf(stderr, "                        test tokenization\n");
+    fprintf(stderr, "  -n N, --n_predict N   number of tokens to predict (default: %d)\n", params.n_predict);
+    fprintf(stderr, "  --top_k N             top-k sampling, 0 = n_vocab (default: %d)\n", params.top_k);
+    fprintf(stderr, "  --top_p N             top-p sampling (default: %.1f)\n", params.top_p);
+    fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
+    fprintf(stderr, "  --repeat-last-n N     last n tokens to consider for penalize (default: %d, 0 = disabled)\n", params.repeat_last_n);
+    fprintf(stderr, "  --repeat-penalty N    penalize repeat sequence of tokens (default: %.2f, 1.0 = disabled)\n", (double)params.repeat_penalty);
+    fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
+    fprintf(stderr, "  -m FNAME, --model FNAME\n");
+    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
+    fprintf(stderr, "\n");
+}
+
+// Function to check if the next argument exists
+std::string get_next_arg(int& i, int argc, char** argv, const std::string& flag, gpt_params& params) {
+    if (i + 1 < argc && argv[i + 1][0] != '-') {
+        return argv[++i];
+    } else {
+        fprintf(stderr, "error: %s requires one argument.\n", flag.c_str());
+        gpt_print_usage(argc, argv, params);
+        exit(0);
+    }
+}
+
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
+    for (int i = 1; i < argc; i++) {
+        std::string arg = argv[i];
+
+        if (arg == "-s" || arg == "--seed") {
+            params.seed = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-t" || arg == "--threads") {
+            params.n_threads = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-ngl" || arg == "--gpu-layers" || arg == "--n-gpu-layers") {
+            params.n_gpu_layers = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-p" || arg == "--prompt") {
+            params.prompt = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-n" || arg == "--n_predict") {
+            params.n_predict = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_k") {
+            params.top_k = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_p") {
+            params.top_p = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--temp") {
+            params.temp = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-last-n") {
+            params.repeat_last_n = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-penalty") {
+            params.repeat_penalty = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-b" || arg == "--batch_size") {
+            params.n_batch= std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-m" || arg == "--model") {
+            params.model = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-i" || arg == "--interactive") {
+            params.interactive = true;
+        } else if (arg == "-ip" || arg == "--interactive-port") {
+            params.interactive = true;
+            params.interactive_port = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-h" || arg == "--help") {
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        } else if (arg == "-f" || arg == "--file") {
+            get_next_arg(i, argc, argv, arg, params);
+            std::ifstream file(argv[i]);
+            if (!file) {
+                fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
+                break;
+            }
+            std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
+            if (params.prompt.back() == '\n') {
+                params.prompt.pop_back();
+            }
+        } else if (arg == "-tt" || arg == "--token_test") {
+            params.token_test = get_next_arg(i, argc, argv, arg, params);
+        }
+        else {
+            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        }
+    }
+
+    return true;
+}
+
+gpt2bpe_vocab::id sample_top_k_top_p_repeat(
+        const gpt2bpe_vocab & vocab,
+        const float * logits,
+        const int32_t * last_n_tokens_data,
+        size_t last_n_tokens_data_size,
+        int    top_k,
+        double top_p,
+        double temp,
+        int repeat_last_n,
+        float repeat_penalty,
+        std::mt19937 & rng) {
+
+    int n_logits = vocab.id_to_token.size();
+
+    const auto * plogits = logits;
+
+    const auto last_n_tokens = std::vector<int32_t>(last_n_tokens_data, last_n_tokens_data + last_n_tokens_data_size);
+
+    if (temp <= 0) {
+        // select the token with the highest logit directly
+        float max_logit = plogits[0];
+        gpt2bpe_vocab::id max_id = 0;
+
+        for (int i = 1; i < n_logits; ++i) {
+            if (plogits[i] > max_logit) {
+                max_logit = plogits[i];
+                max_id = i;
+            }
+        }
+        return max_id;
+    }
+
+
+    std::vector<std::pair<double, gpt2bpe_vocab::id>> logits_id;
+    logits_id.reserve(n_logits);
+
+    {
+        const float scale = 1.0f/temp;
+        for (int i = 0; i < n_logits; ++i) {
+            // repetition penalty from ctrl paper (https://arxiv.org/abs/1909.05858)
+            // credit https://github.com/facebookresearch/llama/compare/main...shawwn:llama:main
+            if (repeat_last_n > 0 && std::find(last_n_tokens.end()-repeat_last_n, last_n_tokens.end(), i) != last_n_tokens.end()) {
+                // if score < 0 then repetition penalty has to multiplied to reduce the previous token probability
+                if (plogits[i] < 0.0f) {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale*repeat_penalty, i));
+                } else {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale/repeat_penalty, i));
+                }
+            } else {
+                logits_id.push_back(std::make_pair(plogits[i]*scale, i));
+            }
+        }
+    }
+
+    // find the top K tokens
+    std::partial_sort(
+            logits_id.begin(),
+            logits_id.begin() + top_k, logits_id.end(),
+            [](const std::pair<double, gpt2bpe_vocab::id> & a, const std::pair<double, gpt2bpe_vocab::id> & b) {
+        return a.first > b.first;
+    });
+
+    logits_id.resize(top_k);
+
+    double maxl = -INFINITY;
+    for (const auto & kv : logits_id) {
+        maxl = std::max(maxl, kv.first);
+    }
+
+    // compute probs for the top K tokens
+    std::vector<double> probs;
+    probs.reserve(logits_id.size());
+
+    double sum = 0.0;
+    for (const auto & kv : logits_id) {
+        double p = exp(kv.first - maxl);
+        probs.push_back(p);
+        sum += p;
+    }
+
+    // normalize the probs
+    for (auto & p : probs) {
+        p /= sum;
+    }
+
+    if (top_p < 1.0f) {
+        double cumsum = 0.0f;
+        for (int i = 0; i < top_k; i++) {
+            cumsum += probs[i];
+            if (cumsum >= top_p) {
+                top_k = i + 1;
+                probs.resize(top_k);
+                logits_id.resize(top_k);
+                break;
+            }
+        }
+
+        cumsum = 1.0/cumsum;
+        for (int i = 0; i < (int) probs.size(); i++) {
+            probs[i] *= cumsum;
+        }
+    }
+
+//    printf("\n");
+//    for (int i = 0; i < (int) probs.size(); i++) {
+//    for (int i = 0; i < 10; i++) {
+//        printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
+//    }
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    int idx = dist(rng);
+
+    return logits_id[idx].second;
+
+}
+
 struct ggml_tensor * get_tensor_ex( struct ggml_context * ctx, std::string name){
 
     struct ggml_tensor * cur = ggml_get_tensor(ctx, name.c_str());
@@ -91,7 +327,7 @@ struct ggml_tensor * get_tensor_ex( struct ggml_context * ctx, std::string name)
 }
 
 // load the model's weights from a file
-bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_vocab & vocab) {
+bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2bpe_vocab & vocab) {
     printf("%s: loading model from '%s'..\n", __func__, fname.c_str());
 
     model.ctx = NULL;
@@ -115,7 +351,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
     fprintf(stdout, "%s: gguf data offset = %zu\n", __func__, gguf_get_data_offset(ggufctx));
 
     // print all kv
-    if( false )
+    #if 0
     {
         const int n_kv = gguf_get_n_kv(ggufctx);
 
@@ -127,6 +363,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
             fprintf(stdout, "%s: kv[%d]: key = %s\n", __func__, i, key);
         }
     }
+    #endif
 
     // print some standard metadata
     {
@@ -249,20 +486,47 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
 
 
             // TEMP until a better bpe tokenizer is implemented
-            word = replace(word, "Ġ", " ");
-            word = replace(word, "Ċ", "\n");
+//            word = replace(word, "Ġ", " ");
+//            word = replace(word, "Ċ", "\n");
 
+//            printf("token %d = '%s'\n",i,word.c_str() );
 
             vocab.token_to_id[word] = i;
             vocab.id_to_token[i] = word;
+
         }
 
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       printf("bos id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       printf("eos id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   printf("unk id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { printf("sep id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   printf("pad id = %d\n", gguf_get_val_u32(ggufctx, keyidx) ); }
+        std::vector<std::pair<std::string, std::string>> bpe_merges;
+
+        for (size_t i = 0; i < hparams.n_merges; i++) {
 
+            std::string word = gguf_get_arr_str(ggufctx, merges_keyidx, i);
+
+            // Split the merges
+            std::string first, second;
+            size_t pos = word.find(' ', 1); // Start the search from the second character
+            if (pos != std::string::npos) {
+                first = word.substr(0, pos);
+                second = word.substr(pos + 1);
+            }
+
+            bpe_merges.push_back(std::make_pair(first, second));
+        }
+
+        vocab.populate_bpe_ranks(bpe_merges);
+
+
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_bos=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_eos=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_unk=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_sep=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_pad=true; }
+
+        if( vocab.special_have_bos ) { fprintf(stdout, "%s: bos token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
+        if( vocab.special_have_eos ) { fprintf(stdout, "%s: eos token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
+        if( vocab.special_have_unk ) { fprintf(stdout, "%s: unk token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
+        if( vocab.special_have_sep ) { fprintf(stdout, "%s: sep token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
+        if( vocab.special_have_pad ) { fprintf(stdout, "%s: pad token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
     }
 
 
@@ -272,7 +536,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
     printf("%s: ggml ctx size = %6.2f MB\n", __func__, ctx_size/(1024.0*1024.0));
 
     // print tensor info
-    if( false )
+    #if 0
     {
         const int n_tensors = gguf_get_n_tensors(ggufctx);
 
@@ -285,7 +549,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt_
             fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
         }
     }
-
+    #endif
 
     // prepare memory for the weights
     {
@@ -435,7 +699,7 @@ bool gpt_neox_eval(
         const gpt_neox_model & model,
         const int n_threads,
         const int n_past,
-        const std::vector<gpt_vocab::id> & embd_inp,
+        const std::vector<gpt2bpe_vocab::id> & embd_inp,
               std::vector<float>         & embd_w,
               size_t                     & mem_per_token) {
     const int N = embd_inp.size();
@@ -687,20 +951,9 @@ int main(int argc, char ** argv) {
         return 1;
     }
 
-    if (params.seed < 0) {
-        params.seed = time(NULL);
-    }
-
-    printf("%s: seed = %d\n", __func__, params.seed);
-
-    std::mt19937 rng(params.seed);
-    if (params.prompt.empty()) {
-        params.prompt = gpt_random_prompt(rng);
-    }
-
     int64_t t_load_us = 0;
 
-    gpt_vocab vocab;
+    gpt2bpe_vocab vocab;
     gpt_neox_model model;
 
     // load the model
@@ -716,8 +969,29 @@ int main(int argc, char ** argv) {
 
     }
 
-    uint32_t eos_token_id = 0;
-    int keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {  eos_token_id = gguf_get_val_u32(ggufctx, keyidx); }    
+    if (params.seed < 0) {
+        params.seed = time(NULL);
+    }
+
+    if (params.top_k == 0) {
+        params.top_k = model.hparams.n_vocab;
+    }
+
+    printf("%s: seed           = %d\n",   __func__, params.seed);
+    printf("%s: temp           = %.3f\n", __func__, params.temp);
+    printf("%s: top_k          = %d\n",   __func__, params.top_k);
+    printf("%s: top_p          = %.3f\n", __func__, params.top_p);
+    printf("%s: repeat_last_n  = %d\n",   __func__, params.repeat_last_n);
+    printf("%s: repeat_penalty = %.3f\n", __func__, params.repeat_penalty);
+
+    std::mt19937 rng(params.seed);
+
+    if (params.prompt.empty()) {
+        params.prompt = "Once upon";
+    }
+
+    std::vector<int32_t> last_n_tokens(model.hparams.n_ctx);
+    std::fill(last_n_tokens.begin(), last_n_tokens.end(), 0);
 
     int n_past = 0;
 
@@ -727,23 +1001,29 @@ int main(int argc, char ** argv) {
     std::vector<float> logits;
 
     // tokenize the prompt
-    std::vector<gpt_vocab::id> embd_inp = ::gpt_tokenize(vocab, params.prompt);
+    std::vector<gpt2bpe_vocab::id> embd_inp = gpt2bpe_tokenize(vocab, params.prompt,false, false);
 
     params.n_predict = std::min(params.n_predict, model.hparams.n_ctx - (int) embd_inp.size());
 
     printf("%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
-    for (int i = 0; i < embd_inp.size(); i++) {
-        printf("%s: token[%d] = %6d, %s\n", __func__, i, embd_inp[i], vocab.id_to_token.at(embd_inp[i]).c_str());
+//    for (size_t i = 0; i < embd_inp.size(); i++) {
+//        printf("%s: token[%zu] = %6d, %s\n", __func__, i, embd_inp[i], vocab.id_to_token[embd_inp[i]].c_str());
+//    }
+
+    if( model.hparams.n_ctx < params.n_predict+embd_inp.size() ) {
+        params.n_predict = model.hparams.n_ctx-embd_inp.size();
     }
+
+    printf("%s: n_predict = %d\n", __func__, params.n_predict);
     printf("\n");
 
-    std::vector<gpt_vocab::id> embd;
+    std::vector<gpt2bpe_vocab::id> embd;
 
     // determine the required inference memory per token:
     size_t mem_per_token = 0;
     gpt_neox_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
 
-    for (int i = embd.size(); i < embd_inp.size() + params.n_predict; i++) {
+    for (size_t i = embd.size(); i < embd_inp.size() + params.n_predict; i++) {
         // predict
         if (embd.size() > 0) {
             const int64_t t_start_us = ggml_time_us();
@@ -764,15 +1044,21 @@ int main(int argc, char ** argv) {
             const int   top_k = params.top_k;
             const float top_p = params.top_p;
             const float temp  = params.temp;
+            const int repeat_last_n = params.repeat_last_n;
+            const float repeat_penalty = params.repeat_penalty;
 
             const int n_vocab = model.hparams.n_vocab;
 
-            gpt_vocab::id id = 0;
+            gpt2bpe_vocab::id id = 0;
 
             {
                 const int64_t t_start_sample_us = ggml_time_us();
 
-                id = gpt_sample_top_k_top_p(vocab, logits.data() + (logits.size() - n_vocab), top_k, top_p, temp, rng);
+//                id = sample_top_k_top_p(vocab, logits.data() + (logits.size() - n_vocab), top_k, top_p, temp, repeat_last_n, repeat_penalty, rng);
+                id = sample_top_k_top_p_repeat(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens.data(), last_n_tokens.size(), top_k, top_p, temp, repeat_last_n, repeat_penalty, rng);
+
+                last_n_tokens.erase(last_n_tokens.begin());
+                last_n_tokens.push_back(id);
 
                 t_sample_us += ggml_time_us() - t_start_sample_us;
             }
@@ -781,7 +1067,7 @@ int main(int argc, char ** argv) {
             embd.push_back(id);
         } else {
             // if here, it means we are still processing the input prompt
-            for (int k = i; k < embd_inp.size(); k++) {
+            for (size_t k = i; k < embd_inp.size(); k++) {
                 embd.push_back(embd_inp[k]);
                 if (embd.size() > params.n_batch) {
                     break;
@@ -792,12 +1078,12 @@ int main(int argc, char ** argv) {
 
         // display text
         for (auto id : embd) {
-            printf("%s", vocab.id_to_token[id].c_str());
+            printf("%s", vocab.id_to_token[id].c_str()  );
         }
         fflush(stdout);
 
         // end of text token
-        if (embd.back() == eos_token_id) {
+        if (vocab.special_have_eos && embd.back() == vocab.special_eos_id) {
             break;
         }
     }

From 5d98989cf647b39b993cfd6dcf271cc282eea58f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 03:58:44 +0200
Subject: [PATCH 074/242] gpt2 bpe tokenizer (handles merges and unicode)

---
 cmpnct_gpt2bpe.hpp | 1011 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 1011 insertions(+)
 create mode 100644 cmpnct_gpt2bpe.hpp

diff --git a/cmpnct_gpt2bpe.hpp b/cmpnct_gpt2bpe.hpp
new file mode 100644
index 0000000000000..0743a294ac0ff
--- /dev/null
+++ b/cmpnct_gpt2bpe.hpp
@@ -0,0 +1,1011 @@
+#ifndef CMPNCT_GPT2BPE
+#define CMPNCT_GPT2BPE
+
+#include <vector>
+#include <string>
+#include <algorithm>
+#include <utility>
+#include <iostream>
+#include <map>
+#include <unordered_map>
+#include <queue>
+#include <cstring>
+
+
+/**
+ * https://github.com/cmp-nct/ggllm.cpp
+ * Minimal library for high performance handling and categorization of UTF8 strings and characters
+ * Using std::string
+ */
+
+enum CNCTCharType {
+    DIGIT,          // a numerical char in any language
+    LETTER,         // a letter in any language
+    WHITESPACE,     // any form of whitespace
+    ACCENT_MARK,    // letter modifiers like ´ in é
+    PUNCTUATION,    // punctuation including brackets
+    SYMBOL,         // math, currency, other symbols
+    CONTROL,        // control characters
+    MIXED,          // a mix of the above
+    UNIDENTIFIED    // something more exotic like emoji or separators
+};
+
+struct CNCTUnicode;
+
+struct CNCTString {
+    std::string str;
+    size_t utf8_chars;
+
+    CNCTCharType char_type=UNIDENTIFIED;
+    bool is_sequential=false;
+
+    size_t seq_offset_bytes=0;
+    size_t seq_offset_utf8_chars=0;
+
+    bool operator==(const std::string &other) const;
+    bool operator==(const char other) const;
+    bool operator==(const CNCTString &other) const;
+    CNCTString &operator+=(const std::string &other);
+    CNCTString &operator+=(const char other);
+    friend CNCTString operator+(CNCTString lhs, const std::string &rhs);
+    friend CNCTString operator+(CNCTString lhs, const char rhs);
+    CNCTString& operator+=(const CNCTString& other);
+    friend CNCTString operator+(CNCTString lhs, const CNCTString& rhs);
+};
+
+struct CNCTUnicode {
+    static bool check_code_range(int c, const std::vector<std::pair<int, int>>& ranges);
+    static CNCTCharType get_code_type(int c);
+    static CNCTCharType get_code_type(const std::string &utf8_char);
+    static int utf8_len(const char c);
+    static int strlen_utf8(std::string src);
+    static std::vector<std::string> split_utf8(const std::string &src);
+    static std::vector<CNCTString> split_utf8_enhanced(const std::string &src);
+    static CNCTCharType string_identify(const std::string& str);
+    static bool string_test(const std::string& str, CNCTCharType chartype);
+};
+
+static const std::vector<std::pair<int, int>> digit_ranges = {
+{0x30, 0x39}, {0xB2, 0xB3}, {0xB9, 0xB9}, {0x660, 0x669}, {0x6F0, 0x6F9}, {0x7C0, 0x7C9}, {0x966, 0x96F}, {0x9E6, 0x9EF}, {0xA66, 0xA6F}, {0xAE6, 0xAEF}, {0xB66, 0xB6F}, {0xBE6, 0xBEF}, {0xC66, 0xC6F}, {0xCE6, 0xCEF}, {0xD66, 0xD6F}, {0xDE6, 0xDEF}, {0xE50, 0xE59}, {0xED0, 0xED9}, {0xF20, 0xF29}, {0x1040, 0x1049}, {0x1090, 0x1099}, {0x1369, 0x1371}, {0x17E0, 0x17E9}, {0x1810, 0x1819}, {0x1946, 0x194F}, {0x19D0, 0x19DA}, {0x1A80, 0x1A89}, {0x1A90, 0x1A99}, {0x1B50, 0x1B59}, {0x1BB0, 0x1BB9}, {0x1C40, 0x1C49}, {0x1C50, 0x1C59}, {0x2070, 0x2070}, {0x2074, 0x2079}, {0x2080, 0x2089}, {0x2460, 0x2468}, {0x2474, 0x247C}, {0x2488, 0x2490}, {0x24EA, 0x24EA}, {0x24F5, 0x24FD}, {0x24FF, 0x24FF}, {0x2776, 0x277E}, {0x2780, 0x2788}, {0x278A, 0x2792}, {0xA620, 0xA629}, {0xA8D0, 0xA8D9}, {0xA900, 0xA909}, {0xA9D0, 0xA9D9}, {0xA9F0, 0xA9F9}, {0xAA50, 0xAA59}, {0xABF0, 0xABF9}, {0xFF10, 0xFF19}, {0x104A0, 0x104A9}, {0x10A40, 0x10A43}, {0x10D30, 0x10D39}, {0x10E60, 0x10E68}, {0x11052, 0x1105A}, {0x11066, 0x1106F}, {0x110F0, 0x110F9}, {0x11136, 0x1113F}, {0x111D0, 0x111D9}, {0x112F0, 0x112F9}, {0x11450, 0x11459}, {0x114D0, 0x114D9}, {0x11650, 0x11659}, {0x116C0, 0x116C9}, {0x11730, 0x11739}, {0x118E0, 0x118E9}, {0x11950, 0x11959}, {0x11C50, 0x11C59}, {0x11D50, 0x11D59}, {0x11DA0, 0x11DA9}, {0x16A60, 0x16A69}, {0x16B50, 0x16B59}, {0x1D7CE, 0x1D7FF}, {0x1E140, 0x1E149}, {0x1E2F0, 0x1E2F9}, {0x1E950, 0x1E959}, {0x1F100, 0x1F10A}, {0x1FBF0, 0x1FBF9}, };
+
+static const std::vector<std::pair<int, int>> letter_ranges = {
+{0x41, 0x5A}, {0x61, 0x7A}, {0xAA, 0xAA}, {0xB5, 0xB5}, {0xBA, 0xBA}, {0xC0, 0xD6}, {0xD8, 0xF6}, {0xF8, 0x2C1}, {0x2C6, 0x2D1}, {0x2E0, 0x2E4}, {0x2EC, 0x2EC}, {0x2EE, 0x2EE}, {0x370, 0x374}, {0x376, 0x377}, {0x37A, 0x37D}, {0x37F, 0x37F}, {0x386, 0x386}, {0x388, 0x38A}, {0x38C, 0x38C}, {0x38E, 0x3A1}, {0x3A3, 0x3F5}, {0x3F7, 0x481}, {0x48A, 0x52F}, {0x531, 0x556}, {0x559, 0x559}, {0x560, 0x588}, {0x5D0, 0x5EA}, {0x5EF, 0x5F2}, {0x620, 0x64A}, {0x66E, 0x66F}, {0x671, 0x6D3}, {0x6D5, 0x6D5}, {0x6E5, 0x6E6}, {0x6EE, 0x6EF}, {0x6FA, 0x6FC}, {0x6FF, 0x6FF}, {0x710, 0x710}, {0x712, 0x72F}, {0x74D, 0x7A5}, {0x7B1, 0x7B1}, {0x7CA, 0x7EA}, {0x7F4, 0x7F5}, {0x7FA, 0x7FA}, {0x800, 0x815}, {0x81A, 0x81A}, {0x824, 0x824}, {0x828, 0x828}, {0x840, 0x858}, {0x860, 0x86A}, {0x8A0, 0x8B4}, {0x8B6, 0x8C7}, {0x904, 0x939}, {0x93D, 0x93D}, {0x950, 0x950}, {0x958, 0x961}, {0x971, 0x980}, {0x985, 0x98C}, {0x98F, 0x990}, {0x993, 0x9A8}, {0x9AA, 0x9B0}, {0x9B2, 0x9B2}, {0x9B6, 0x9B9}, {0x9BD, 0x9BD}, {0x9CE, 0x9CE}, {0x9DC, 0x9DD}, {0x9DF, 0x9E1}, {0x9F0, 0x9F1}, {0x9FC, 0x9FC}, {0xA05, 0xA0A}, {0xA0F, 0xA10}, {0xA13, 0xA28}, {0xA2A, 0xA30}, {0xA32, 0xA33}, {0xA35, 0xA36}, {0xA38, 0xA39}, {0xA59, 0xA5C}, {0xA5E, 0xA5E}, {0xA72, 0xA74}, {0xA85, 0xA8D}, {0xA8F, 0xA91}, {0xA93, 0xAA8}, {0xAAA, 0xAB0}, {0xAB2, 0xAB3}, {0xAB5, 0xAB9}, {0xABD, 0xABD}, {0xAD0, 0xAD0}, {0xAE0, 0xAE1}, {0xAF9, 0xAF9}, {0xB05, 0xB0C}, {0xB0F, 0xB10}, {0xB13, 0xB28}, {0xB2A, 0xB30}, {0xB32, 0xB33}, {0xB35, 0xB39}, {0xB3D, 0xB3D}, {0xB5C, 0xB5D}, {0xB5F, 0xB61}, {0xB71, 0xB71}, {0xB83, 0xB83}, {0xB85, 0xB8A}, {0xB8E, 0xB90}, {0xB92, 0xB95}, {0xB99, 0xB9A}, {0xB9C, 0xB9C}, {0xB9E, 0xB9F}, {0xBA3, 0xBA4}, {0xBA8, 0xBAA}, {0xBAE, 0xBB9}, {0xBD0, 0xBD0}, {0xC05, 0xC0C}, {0xC0E, 0xC10}, {0xC12, 0xC28}, {0xC2A, 0xC39}, {0xC3D, 0xC3D}, {0xC58, 0xC5A}, {0xC60, 0xC61}, {0xC80, 0xC80}, {0xC85, 0xC8C}, {0xC8E, 0xC90}, {0xC92, 0xCA8}, {0xCAA, 0xCB3}, {0xCB5, 0xCB9}, {0xCBD, 0xCBD}, {0xCDE, 0xCDE}, {0xCE0, 0xCE1}, {0xCF1, 0xCF2}, {0xD04, 0xD0C}, {0xD0E, 0xD10}, {0xD12, 0xD3A}, {0xD3D, 0xD3D}, {0xD4E, 0xD4E}, {0xD54, 0xD56}, {0xD5F, 0xD61}, {0xD7A, 0xD7F}, {0xD85, 0xD96}, {0xD9A, 0xDB1}, {0xDB3, 0xDBB}, {0xDBD, 0xDBD}, {0xDC0, 0xDC6}, {0xE01, 0xE30}, {0xE32, 0xE33}, {0xE40, 0xE46}, {0xE81, 0xE82}, {0xE84, 0xE84}, {0xE86, 0xE8A}, {0xE8C, 0xEA3}, {0xEA5, 0xEA5}, {0xEA7, 0xEB0}, {0xEB2, 0xEB3}, {0xEBD, 0xEBD}, {0xEC0, 0xEC4}, {0xEC6, 0xEC6}, {0xEDC, 0xEDF}, {0xF00, 0xF00}, {0xF40, 0xF47}, {0xF49, 0xF6C}, {0xF88, 0xF8C}, {0x1000, 0x102A}, {0x103F, 0x103F}, {0x1050, 0x1055}, {0x105A, 0x105D}, {0x1061, 0x1061}, {0x1065, 0x1066}, {0x106E, 0x1070}, {0x1075, 0x1081}, {0x108E, 0x108E}, {0x10A0, 0x10C5}, {0x10C7, 0x10C7}, {0x10CD, 0x10CD}, {0x10D0, 0x10FA}, {0x10FC, 0x1248}, {0x124A, 0x124D}, {0x1250, 0x1256}, {0x1258, 0x1258}, {0x125A, 0x125D}, {0x1260, 0x1288}, {0x128A, 0x128D}, {0x1290, 0x12B0}, {0x12B2, 0x12B5}, {0x12B8, 0x12BE}, {0x12C0, 0x12C0}, {0x12C2, 0x12C5}, {0x12C8, 0x12D6}, {0x12D8, 0x1310}, {0x1312, 0x1315}, {0x1318, 0x135A}, {0x1380, 0x138F}, {0x13A0, 0x13F5}, {0x13F8, 0x13FD}, {0x1401, 0x166C}, {0x166F, 0x167F}, {0x1681, 0x169A}, {0x16A0, 0x16EA}, {0x16F1, 0x16F8}, {0x1700, 0x170C}, {0x170E, 0x1711}, {0x1720, 0x1731}, {0x1740, 0x1751}, {0x1760, 0x176C}, {0x176E, 0x1770}, {0x1780, 0x17B3}, {0x17D7, 0x17D7}, {0x17DC, 0x17DC}, {0x1820, 0x1878}, {0x1880, 0x1884}, {0x1887, 0x18A8}, {0x18AA, 0x18AA}, {0x18B0, 0x18F5}, {0x1900, 0x191E}, {0x1950, 0x196D}, {0x1970, 0x1974}, {0x1980, 0x19AB}, {0x19B0, 0x19C9}, {0x1A00, 0x1A16}, {0x1A20, 0x1A54}, {0x1AA7, 0x1AA7}, {0x1B05, 0x1B33}, {0x1B45, 0x1B4B}, {0x1B83, 0x1BA0}, {0x1BAE, 0x1BAF}, {0x1BBA, 0x1BE5}, {0x1C00, 0x1C23}, {0x1C4D, 0x1C4F}, {0x1C5A, 0x1C7D}, {0x1C80, 0x1C88}, {0x1C90, 0x1CBA}, {0x1CBD, 0x1CBF}, {0x1CE9, 0x1CEC}, {0x1CEE, 0x1CF3}, {0x1CF5, 0x1CF6}, {0x1CFA, 0x1CFA}, {0x1D00, 0x1DBF}, {0x1E00, 0x1F15}, {0x1F18, 0x1F1D}, {0x1F20, 0x1F45}, {0x1F48, 0x1F4D}, {0x1F50, 0x1F57}, {0x1F59, 0x1F59}, {0x1F5B, 0x1F5B}, {0x1F5D, 0x1F5D}, {0x1F5F, 0x1F7D}, {0x1F80, 0x1FB4}, {0x1FB6, 0x1FBC}, {0x1FBE, 0x1FBE}, {0x1FC2, 0x1FC4}, {0x1FC6, 0x1FCC}, {0x1FD0, 0x1FD3}, {0x1FD6, 0x1FDB}, {0x1FE0, 0x1FEC}, {0x1FF2, 0x1FF4}, {0x1FF6, 0x1FFC}, {0x2071, 0x2071}, {0x207F, 0x207F}, {0x2090, 0x209C}, {0x2102, 0x2102}, {0x2107, 0x2107}, {0x210A, 0x2113}, {0x2115, 0x2115}, {0x2119, 0x211D}, {0x2124, 0x2124}, {0x2126, 0x2126}, {0x2128, 0x2128}, {0x212A, 0x212D}, {0x212F, 0x2139}, {0x213C, 0x213F}, {0x2145, 0x2149}, {0x214E, 0x214E}, {0x2183, 0x2184}, {0x2C00, 0x2C2E}, {0x2C30, 0x2C5E}, {0x2C60, 0x2CE4}, {0x2CEB, 0x2CEE}, {0x2CF2, 0x2CF3}, {0x2D00, 0x2D25}, {0x2D27, 0x2D27}, {0x2D2D, 0x2D2D}, {0x2D30, 0x2D67}, {0x2D6F, 0x2D6F}, {0x2D80, 0x2D96}, {0x2DA0, 0x2DA6}, {0x2DA8, 0x2DAE}, {0x2DB0, 0x2DB6}, {0x2DB8, 0x2DBE}, {0x2DC0, 0x2DC6}, {0x2DC8, 0x2DCE}, {0x2DD0, 0x2DD6}, {0x2DD8, 0x2DDE}, {0x2E2F, 0x2E2F}, {0x3005, 0x3006}, {0x3031, 0x3035}, {0x303B, 0x303C}, {0x3041, 0x3096}, {0x309D, 0x309F}, {0x30A1, 0x30FA}, {0x30FC, 0x30FF}, {0x3105, 0x312F}, {0x3131, 0x318E}, {0x31A0, 0x31BF}, {0x31F0, 0x31FF}, {0x3400, 0x4DBF}, {0x4E00, 0x9FFC}, {0xA000, 0xA48C}, {0xA4D0, 0xA4FD}, {0xA500, 0xA60C}, {0xA610, 0xA61F}, {0xA62A, 0xA62B}, {0xA640, 0xA66E}, {0xA67F, 0xA69D}, {0xA6A0, 0xA6E5}, {0xA717, 0xA71F}, {0xA722, 0xA788}, {0xA78B, 0xA7BF}, {0xA7C2, 0xA7CA}, {0xA7F5, 0xA801}, {0xA803, 0xA805}, {0xA807, 0xA80A}, {0xA80C, 0xA822}, {0xA840, 0xA873}, {0xA882, 0xA8B3}, {0xA8F2, 0xA8F7}, {0xA8FB, 0xA8FB}, {0xA8FD, 0xA8FE}, {0xA90A, 0xA925}, {0xA930, 0xA946}, {0xA960, 0xA97C}, {0xA984, 0xA9B2}, {0xA9CF, 0xA9CF}, {0xA9E0, 0xA9E4}, {0xA9E6, 0xA9EF}, {0xA9FA, 0xA9FE}, {0xAA00, 0xAA28}, {0xAA40, 0xAA42}, {0xAA44, 0xAA4B}, {0xAA60, 0xAA76}, {0xAA7A, 0xAA7A}, {0xAA7E, 0xAAAF}, {0xAAB1, 0xAAB1}, {0xAAB5, 0xAAB6}, {0xAAB9, 0xAABD}, {0xAAC0, 0xAAC0}, {0xAAC2, 0xAAC2}, {0xAADB, 0xAADD}, {0xAAE0, 0xAAEA}, {0xAAF2, 0xAAF4}, {0xAB01, 0xAB06}, {0xAB09, 0xAB0E}, {0xAB11, 0xAB16}, {0xAB20, 0xAB26}, {0xAB28, 0xAB2E}, {0xAB30, 0xAB5A}, {0xAB5C, 0xAB69}, {0xAB70, 0xABE2}, {0xAC00, 0xD7A3}, {0xD7B0, 0xD7C6}, {0xD7CB, 0xD7FB}, {0xF900, 0xFA6D}, {0xFA70, 0xFAD9}, {0xFB00, 0xFB06}, {0xFB13, 0xFB17}, {0xFB1D, 0xFB1D}, {0xFB1F, 0xFB28}, {0xFB2A, 0xFB36}, {0xFB38, 0xFB3C}, {0xFB3E, 0xFB3E}, {0xFB40, 0xFB41}, {0xFB43, 0xFB44}, {0xFB46, 0xFBB1}, {0xFBD3, 0xFD3D}, {0xFD50, 0xFD8F}, {0xFD92, 0xFDC7}, {0xFDF0, 0xFDFB}, {0xFE70, 0xFE74}, {0xFE76, 0xFEFC}, {0xFF21, 0xFF3A}, {0xFF41, 0xFF5A}, {0xFF66, 0xFFBE}, {0xFFC2, 0xFFC7}, {0xFFCA, 0xFFCF}, {0xFFD2, 0xFFD7}, {0xFFDA, 0xFFDC}, {0x10000, 0x1000B}, {0x1000D, 0x10026}, {0x10028, 0x1003A}, {0x1003C, 0x1003D}, {0x1003F, 0x1004D}, {0x10050, 0x1005D}, {0x10080, 0x100FA}, {0x10280, 0x1029C}, {0x102A0, 0x102D0}, {0x10300, 0x1031F}, {0x1032D, 0x10340}, {0x10342, 0x10349}, {0x10350, 0x10375}, {0x10380, 0x1039D}, {0x103A0, 0x103C3}, {0x103C8, 0x103CF}, {0x10400, 0x1049D}, {0x104B0, 0x104D3}, {0x104D8, 0x104FB}, {0x10500, 0x10527}, {0x10530, 0x10563}, {0x10600, 0x10736}, {0x10740, 0x10755}, {0x10760, 0x10767}, {0x10800, 0x10805}, {0x10808, 0x10808}, {0x1080A, 0x10835}, {0x10837, 0x10838}, {0x1083C, 0x1083C}, {0x1083F, 0x10855}, {0x10860, 0x10876}, {0x10880, 0x1089E}, {0x108E0, 0x108F2}, {0x108F4, 0x108F5}, {0x10900, 0x10915}, {0x10920, 0x10939}, {0x10980, 0x109B7}, {0x109BE, 0x109BF}, {0x10A00, 0x10A00}, {0x10A10, 0x10A13}, {0x10A15, 0x10A17}, {0x10A19, 0x10A35}, {0x10A60, 0x10A7C}, {0x10A80, 0x10A9C}, {0x10AC0, 0x10AC7}, {0x10AC9, 0x10AE4}, {0x10B00, 0x10B35}, {0x10B40, 0x10B55}, {0x10B60, 0x10B72}, {0x10B80, 0x10B91}, {0x10C00, 0x10C48}, {0x10C80, 0x10CB2}, {0x10CC0, 0x10CF2}, {0x10D00, 0x10D23}, {0x10E80, 0x10EA9}, {0x10EB0, 0x10EB1}, {0x10F00, 0x10F1C}, {0x10F27, 0x10F27}, {0x10F30, 0x10F45}, {0x10FB0, 0x10FC4}, {0x10FE0, 0x10FF6}, {0x11003, 0x11037}, {0x11083, 0x110AF}, {0x110D0, 0x110E8}, {0x11103, 0x11126}, {0x11144, 0x11144}, {0x11147, 0x11147}, {0x11150, 0x11172}, {0x11176, 0x11176}, {0x11183, 0x111B2}, {0x111C1, 0x111C4}, {0x111DA, 0x111DA}, {0x111DC, 0x111DC}, {0x11200, 0x11211}, {0x11213, 0x1122B}, {0x11280, 0x11286}, {0x11288, 0x11288}, {0x1128A, 0x1128D}, {0x1128F, 0x1129D}, {0x1129F, 0x112A8}, {0x112B0, 0x112DE}, {0x11305, 0x1130C}, {0x1130F, 0x11310}, {0x11313, 0x11328}, {0x1132A, 0x11330}, {0x11332, 0x11333}, {0x11335, 0x11339}, {0x1133D, 0x1133D}, {0x11350, 0x11350}, {0x1135D, 0x11361}, {0x11400, 0x11434}, {0x11447, 0x1144A}, {0x1145F, 0x11461}, {0x11480, 0x114AF}, {0x114C4, 0x114C5}, {0x114C7, 0x114C7}, {0x11580, 0x115AE}, {0x115D8, 0x115DB}, {0x11600, 0x1162F}, {0x11644, 0x11644}, {0x11680, 0x116AA}, {0x116B8, 0x116B8}, {0x11700, 0x1171A}, {0x11800, 0x1182B}, {0x118A0, 0x118DF}, {0x118FF, 0x11906}, {0x11909, 0x11909}, {0x1190C, 0x11913}, {0x11915, 0x11916}, {0x11918, 0x1192F}, {0x1193F, 0x1193F}, {0x11941, 0x11941}, {0x119A0, 0x119A7}, {0x119AA, 0x119D0}, {0x119E1, 0x119E1}, {0x119E3, 0x119E3}, {0x11A00, 0x11A00}, {0x11A0B, 0x11A32}, {0x11A3A, 0x11A3A}, {0x11A50, 0x11A50}, {0x11A5C, 0x11A89}, {0x11A9D, 0x11A9D}, {0x11AC0, 0x11AF8}, {0x11C00, 0x11C08}, {0x11C0A, 0x11C2E}, {0x11C40, 0x11C40}, {0x11C72, 0x11C8F}, {0x11D00, 0x11D06}, {0x11D08, 0x11D09}, {0x11D0B, 0x11D30}, {0x11D46, 0x11D46}, {0x11D60, 0x11D65}, {0x11D67, 0x11D68}, {0x11D6A, 0x11D89}, {0x11D98, 0x11D98}, {0x11EE0, 0x11EF2}, {0x11FB0, 0x11FB0}, {0x12000, 0x12399}, {0x12480, 0x12543}, {0x13000, 0x1342E}, {0x14400, 0x14646}, {0x16800, 0x16A38}, {0x16A40, 0x16A5E}, {0x16AD0, 0x16AED}, {0x16B00, 0x16B2F}, {0x16B40, 0x16B43}, {0x16B63, 0x16B77}, {0x16B7D, 0x16B8F}, {0x16E40, 0x16E7F}, {0x16F00, 0x16F4A}, {0x16F50, 0x16F50}, {0x16F93, 0x16F9F}, {0x16FE0, 0x16FE1}, {0x16FE3, 0x16FE3}, {0x17000, 0x187F7}, {0x18800, 0x18CD5}, {0x18D00, 0x18D08}, {0x1B000, 0x1B11E}, {0x1B150, 0x1B152}, {0x1B164, 0x1B167}, {0x1B170, 0x1B2FB}, {0x1BC00, 0x1BC6A}, {0x1BC70, 0x1BC7C}, {0x1BC80, 0x1BC88}, {0x1BC90, 0x1BC99}, {0x1D400, 0x1D454}, {0x1D456, 0x1D49C}, {0x1D49E, 0x1D49F}, {0x1D4A2, 0x1D4A2}, {0x1D4A5, 0x1D4A6}, {0x1D4A9, 0x1D4AC}, {0x1D4AE, 0x1D4B9}, {0x1D4BB, 0x1D4BB}, {0x1D4BD, 0x1D4C3}, {0x1D4C5, 0x1D505}, {0x1D507, 0x1D50A}, {0x1D50D, 0x1D514}, {0x1D516, 0x1D51C}, {0x1D51E, 0x1D539}, {0x1D53B, 0x1D53E}, {0x1D540, 0x1D544}, {0x1D546, 0x1D546}, {0x1D54A, 0x1D550}, {0x1D552, 0x1D6A5}, {0x1D6A8, 0x1D6C0}, {0x1D6C2, 0x1D6DA}, {0x1D6DC, 0x1D6FA}, {0x1D6FC, 0x1D714}, {0x1D716, 0x1D734}, {0x1D736, 0x1D74E}, {0x1D750, 0x1D76E}, {0x1D770, 0x1D788}, {0x1D78A, 0x1D7A8}, {0x1D7AA, 0x1D7C2}, {0x1D7C4, 0x1D7CB}, {0x1E100, 0x1E12C}, {0x1E137, 0x1E13D}, {0x1E14E, 0x1E14E}, {0x1E2C0, 0x1E2EB}, {0x1E800, 0x1E8C4}, {0x1E900, 0x1E943}, {0x1E94B, 0x1E94B}, {0x1EE00, 0x1EE03}, {0x1EE05, 0x1EE1F}, {0x1EE21, 0x1EE22}, {0x1EE24, 0x1EE24}, {0x1EE27, 0x1EE27}, {0x1EE29, 0x1EE32}, {0x1EE34, 0x1EE37}, {0x1EE39, 0x1EE39}, {0x1EE3B, 0x1EE3B}, {0x1EE42, 0x1EE42}, {0x1EE47, 0x1EE47}, {0x1EE49, 0x1EE49}, {0x1EE4B, 0x1EE4B}, {0x1EE4D, 0x1EE4F}, {0x1EE51, 0x1EE52}, {0x1EE54, 0x1EE54}, {0x1EE57, 0x1EE57}, {0x1EE59, 0x1EE59}, {0x1EE5B, 0x1EE5B}, {0x1EE5D, 0x1EE5D}, {0x1EE5F, 0x1EE5F}, {0x1EE61, 0x1EE62}, {0x1EE64, 0x1EE64}, {0x1EE67, 0x1EE6A}, {0x1EE6C, 0x1EE72}, {0x1EE74, 0x1EE77}, {0x1EE79, 0x1EE7C}, {0x1EE7E, 0x1EE7E}, {0x1EE80, 0x1EE89}, {0x1EE8B, 0x1EE9B}, {0x1EEA1, 0x1EEA3}, {0x1EEA5, 0x1EEA9}, {0x1EEAB, 0x1EEBB}, {0x20000, 0x2A6DD}, {0x2A700, 0x2B734}, {0x2B740, 0x2B81D}, {0x2B820, 0x2CEA1}, {0x2CEB0, 0x2EBE0}, {0x2F800, 0x2FA1D}, {0x30000, 0x3134A}, };
+
+static const std::vector<std::pair<int, int>> whitespace_ranges = {
+{0x9, 0xD}, {0x1C, 0x20}, {0x85, 0x85}, {0xA0, 0xA0}, {0x1680, 0x1680}, {0x2000, 0x200A}, {0x2028, 0x2029}, {0x202F, 0x202F}, {0x205F, 0x205F}, {0x3000, 0x3000}, };
+
+static const std::vector<std::pair<int, int>> accent_mark_ranges = {
+{0x300, 0x36F}, {0x483, 0x489}, {0x591, 0x5BD}, {0x5BF, 0x5BF}, {0x5C1, 0x5C2}, {0x5C4, 0x5C5}, {0x5C7, 0x5C7}, {0x610, 0x61A}, {0x64B, 0x65F}, {0x670, 0x670}, {0x6D6, 0x6DC}, {0x6DF, 0x6E4}, {0x6E7, 0x6E8}, {0x6EA, 0x6ED}, {0x711, 0x711}, {0x730, 0x74A}, {0x7A6, 0x7B0}, {0x7EB, 0x7F3}, {0x7FD, 0x7FD}, {0x816, 0x819}, {0x81B, 0x823}, {0x825, 0x827}, {0x829, 0x82D}, {0x859, 0x85B}, {0x8D3, 0x8E1}, {0x8E3, 0x903}, {0x93A, 0x93C}, {0x93E, 0x94F}, {0x951, 0x957}, {0x962, 0x963}, {0x981, 0x983}, {0x9BC, 0x9BC}, {0x9BE, 0x9C4}, {0x9C7, 0x9C8}, {0x9CB, 0x9CD}, {0x9D7, 0x9D7}, {0x9E2, 0x9E3}, {0x9FE, 0x9FE}, {0xA01, 0xA03}, {0xA3C, 0xA3C}, {0xA3E, 0xA42}, {0xA47, 0xA48}, {0xA4B, 0xA4D}, {0xA51, 0xA51}, {0xA70, 0xA71}, {0xA75, 0xA75}, {0xA81, 0xA83}, {0xABC, 0xABC}, {0xABE, 0xAC5}, {0xAC7, 0xAC9}, {0xACB, 0xACD}, {0xAE2, 0xAE3}, {0xAFA, 0xAFF}, {0xB01, 0xB03}, {0xB3C, 0xB3C}, {0xB3E, 0xB44}, {0xB47, 0xB48}, {0xB4B, 0xB4D}, {0xB55, 0xB57}, {0xB62, 0xB63}, {0xB82, 0xB82}, {0xBBE, 0xBC2}, {0xBC6, 0xBC8}, {0xBCA, 0xBCD}, {0xBD7, 0xBD7}, {0xC00, 0xC04}, {0xC3E, 0xC44}, {0xC46, 0xC48}, {0xC4A, 0xC4D}, {0xC55, 0xC56}, {0xC62, 0xC63}, {0xC81, 0xC83}, {0xCBC, 0xCBC}, {0xCBE, 0xCC4}, {0xCC6, 0xCC8}, {0xCCA, 0xCCD}, {0xCD5, 0xCD6}, {0xCE2, 0xCE3}, {0xD00, 0xD03}, {0xD3B, 0xD3C}, {0xD3E, 0xD44}, {0xD46, 0xD48}, {0xD4A, 0xD4D}, {0xD57, 0xD57}, {0xD62, 0xD63}, {0xD81, 0xD83}, {0xDCA, 0xDCA}, {0xDCF, 0xDD4}, {0xDD6, 0xDD6}, {0xDD8, 0xDDF}, {0xDF2, 0xDF3}, {0xE31, 0xE31}, {0xE34, 0xE3A}, {0xE47, 0xE4E}, {0xEB1, 0xEB1}, {0xEB4, 0xEBC}, {0xEC8, 0xECD}, {0xF18, 0xF19}, {0xF35, 0xF35}, {0xF37, 0xF37}, {0xF39, 0xF39}, {0xF3E, 0xF3F}, {0xF71, 0xF84}, {0xF86, 0xF87}, {0xF8D, 0xF97}, {0xF99, 0xFBC}, {0xFC6, 0xFC6}, {0x102B, 0x103E}, {0x1056, 0x1059}, {0x105E, 0x1060}, {0x1062, 0x1064}, {0x1067, 0x106D}, {0x1071, 0x1074}, {0x1082, 0x108D}, {0x108F, 0x108F}, {0x109A, 0x109D}, {0x135D, 0x135F}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753}, {0x1772, 0x1773}, {0x17B4, 0x17D3}, {0x17DD, 0x17DD}, {0x180B, 0x180D}, {0x1885, 0x1886}, {0x18A9, 0x18A9}, {0x1920, 0x192B}, {0x1930, 0x193B}, {0x1A17, 0x1A1B}, {0x1A55, 0x1A5E}, {0x1A60, 0x1A7C}, {0x1A7F, 0x1A7F}, {0x1AB0, 0x1AC0}, {0x1B00, 0x1B04}, {0x1B34, 0x1B44}, {0x1B6B, 0x1B73}, {0x1B80, 0x1B82}, {0x1BA1, 0x1BAD}, {0x1BE6, 0x1BF3}, {0x1C24, 0x1C37}, {0x1CD0, 0x1CD2}, {0x1CD4, 0x1CE8}, {0x1CED, 0x1CED}, {0x1CF4, 0x1CF4}, {0x1CF7, 0x1CF9}, {0x1DC0, 0x1DF9}, {0x1DFB, 0x1DFF}, {0x20D0, 0x20F0}, {0x2CEF, 0x2CF1}, {0x2D7F, 0x2D7F}, {0x2DE0, 0x2DFF}, {0x302A, 0x302F}, {0x3099, 0x309A}, {0xA66F, 0xA672}, {0xA674, 0xA67D}, {0xA69E, 0xA69F}, {0xA6F0, 0xA6F1}, {0xA802, 0xA802}, {0xA806, 0xA806}, {0xA80B, 0xA80B}, {0xA823, 0xA827}, {0xA82C, 0xA82C}, {0xA880, 0xA881}, {0xA8B4, 0xA8C5}, {0xA8E0, 0xA8F1}, {0xA8FF, 0xA8FF}, {0xA926, 0xA92D}, {0xA947, 0xA953}, {0xA980, 0xA983}, {0xA9B3, 0xA9C0}, {0xA9E5, 0xA9E5}, {0xAA29, 0xAA36}, {0xAA43, 0xAA43}, {0xAA4C, 0xAA4D}, {0xAA7B, 0xAA7D}, {0xAAB0, 0xAAB0}, {0xAAB2, 0xAAB4}, {0xAAB7, 0xAAB8}, {0xAABE, 0xAABF}, {0xAAC1, 0xAAC1}, {0xAAEB, 0xAAEF}, {0xAAF5, 0xAAF6}, {0xABE3, 0xABEA}, {0xABEC, 0xABED}, {0xFB1E, 0xFB1E}, {0xFE00, 0xFE0F}, {0xFE20, 0xFE2F}, {0x101FD, 0x101FD}, {0x102E0, 0x102E0}, {0x10376, 0x1037A}, {0x10A01, 0x10A03}, {0x10A05, 0x10A06}, {0x10A0C, 0x10A0F}, {0x10A38, 0x10A3A}, {0x10A3F, 0x10A3F}, {0x10AE5, 0x10AE6}, {0x10D24, 0x10D27}, {0x10EAB, 0x10EAC}, {0x10F46, 0x10F50}, {0x11000, 0x11002}, {0x11038, 0x11046}, {0x1107F, 0x11082}, {0x110B0, 0x110BA}, {0x11100, 0x11102}, {0x11127, 0x11134}, {0x11145, 0x11146}, {0x11173, 0x11173}, {0x11180, 0x11182}, {0x111B3, 0x111C0}, {0x111C9, 0x111CC}, {0x111CE, 0x111CF}, {0x1122C, 0x11237}, {0x1123E, 0x1123E}, {0x112DF, 0x112EA}, {0x11300, 0x11303}, {0x1133B, 0x1133C}, {0x1133E, 0x11344}, {0x11347, 0x11348}, {0x1134B, 0x1134D}, {0x11357, 0x11357}, {0x11362, 0x11363}, {0x11366, 0x1136C}, {0x11370, 0x11374}, {0x11435, 0x11446}, {0x1145E, 0x1145E}, {0x114B0, 0x114C3}, {0x115AF, 0x115B5}, {0x115B8, 0x115C0}, {0x115DC, 0x115DD}, {0x11630, 0x11640}, {0x116AB, 0x116B7}, {0x1171D, 0x1172B}, {0x1182C, 0x1183A}, {0x11930, 0x11935}, {0x11937, 0x11938}, {0x1193B, 0x1193E}, {0x11940, 0x11940}, {0x11942, 0x11943}, {0x119D1, 0x119D7}, {0x119DA, 0x119E0}, {0x119E4, 0x119E4}, {0x11A01, 0x11A0A}, {0x11A33, 0x11A39}, {0x11A3B, 0x11A3E}, {0x11A47, 0x11A47}, {0x11A51, 0x11A5B}, {0x11A8A, 0x11A99}, {0x11C2F, 0x11C36}, {0x11C38, 0x11C3F}, {0x11C92, 0x11CA7}, {0x11CA9, 0x11CB6}, {0x11D31, 0x11D36}, {0x11D3A, 0x11D3A}, {0x11D3C, 0x11D3D}, {0x11D3F, 0x11D45}, {0x11D47, 0x11D47}, {0x11D8A, 0x11D8E}, {0x11D90, 0x11D91}, {0x11D93, 0x11D97}, {0x11EF3, 0x11EF6}, {0x16AF0, 0x16AF4}, {0x16B30, 0x16B36}, {0x16F4F, 0x16F4F}, {0x16F51, 0x16F87}, {0x16F8F, 0x16F92}, {0x16FE4, 0x16FE4}, {0x16FF0, 0x16FF1}, {0x1BC9D, 0x1BC9E}, {0x1D165, 0x1D169}, {0x1D16D, 0x1D172}, {0x1D17B, 0x1D182}, {0x1D185, 0x1D18B}, {0x1D1AA, 0x1D1AD}, {0x1D242, 0x1D244}, {0x1DA00, 0x1DA36}, {0x1DA3B, 0x1DA6C}, {0x1DA75, 0x1DA75}, {0x1DA84, 0x1DA84}, {0x1DA9B, 0x1DA9F}, {0x1DAA1, 0x1DAAF}, {0x1E000, 0x1E006}, {0x1E008, 0x1E018}, {0x1E01B, 0x1E021}, {0x1E023, 0x1E024}, {0x1E026, 0x1E02A}, {0x1E130, 0x1E136}, {0x1E2EC, 0x1E2EF}, {0x1E8D0, 0x1E8D6}, {0x1E944, 0x1E94A}, {0xE0100, 0xE01EF}, };
+
+static const std::vector<std::pair<int, int>> punctuation_ranges = {
+{0x21, 0x23}, {0x25, 0x2A}, {0x2C, 0x2F}, {0x3A, 0x3B}, {0x3F, 0x40}, {0x5B, 0x5D}, {0x5F, 0x5F}, {0x7B, 0x7B}, {0x7D, 0x7D}, {0xA1, 0xA1}, {0xA7, 0xA7}, {0xAB, 0xAB}, {0xB6, 0xB7}, {0xBB, 0xBB}, {0xBF, 0xBF}, {0x37E, 0x37E}, {0x387, 0x387}, {0x55A, 0x55F}, {0x589, 0x58A}, {0x5BE, 0x5BE}, {0x5C0, 0x5C0}, {0x5C3, 0x5C3}, {0x5C6, 0x5C6}, {0x5F3, 0x5F4}, {0x609, 0x60A}, {0x60C, 0x60D}, {0x61B, 0x61B}, {0x61E, 0x61F}, {0x66A, 0x66D}, {0x6D4, 0x6D4}, {0x700, 0x70D}, {0x7F7, 0x7F9}, {0x830, 0x83E}, {0x85E, 0x85E}, {0x964, 0x965}, {0x970, 0x970}, {0x9FD, 0x9FD}, {0xA76, 0xA76}, {0xAF0, 0xAF0}, {0xC77, 0xC77}, {0xC84, 0xC84}, {0xDF4, 0xDF4}, {0xE4F, 0xE4F}, {0xE5A, 0xE5B}, {0xF04, 0xF12}, {0xF14, 0xF14}, {0xF3A, 0xF3D}, {0xF85, 0xF85}, {0xFD0, 0xFD4}, {0xFD9, 0xFDA}, {0x104A, 0x104F}, {0x10FB, 0x10FB}, {0x1360, 0x1368}, {0x1400, 0x1400}, {0x166E, 0x166E}, {0x169B, 0x169C}, {0x16EB, 0x16ED}, {0x1735, 0x1736}, {0x17D4, 0x17D6}, {0x17D8, 0x17DA}, {0x1800, 0x180A}, {0x1944, 0x1945}, {0x1A1E, 0x1A1F}, {0x1AA0, 0x1AA6}, {0x1AA8, 0x1AAD}, {0x1B5A, 0x1B60}, {0x1BFC, 0x1BFF}, {0x1C3B, 0x1C3F}, {0x1C7E, 0x1C7F}, {0x1CC0, 0x1CC7}, {0x1CD3, 0x1CD3}, {0x2010, 0x2027}, {0x2030, 0x2043}, {0x2045, 0x2051}, {0x2053, 0x205E}, {0x207D, 0x207E}, {0x208D, 0x208E}, {0x2308, 0x230B}, {0x2329, 0x232A}, {0x2768, 0x2775}, {0x27C5, 0x27C6}, {0x27E6, 0x27EF}, {0x2983, 0x2998}, {0x29D8, 0x29DB}, {0x29FC, 0x29FD}, {0x2CF9, 0x2CFC}, {0x2CFE, 0x2CFF}, {0x2D70, 0x2D70}, {0x2E00, 0x2E2E}, {0x2E30, 0x2E4F}, {0x2E52, 0x2E52}, {0x3001, 0x3003}, {0x3008, 0x3011}, {0x3014, 0x301F}, {0x3030, 0x3030}, {0x303D, 0x303D}, {0x30A0, 0x30A0}, {0x30FB, 0x30FB}, {0xA4FE, 0xA4FF}, {0xA60D, 0xA60F}, {0xA673, 0xA673}, {0xA67E, 0xA67E}, {0xA6F2, 0xA6F7}, {0xA874, 0xA877}, {0xA8CE, 0xA8CF}, {0xA8F8, 0xA8FA}, {0xA8FC, 0xA8FC}, {0xA92E, 0xA92F}, {0xA95F, 0xA95F}, {0xA9C1, 0xA9CD}, {0xA9DE, 0xA9DF}, {0xAA5C, 0xAA5F}, {0xAADE, 0xAADF}, {0xAAF0, 0xAAF1}, {0xABEB, 0xABEB}, {0xFD3E, 0xFD3F}, {0xFE10, 0xFE19}, {0xFE30, 0xFE52}, {0xFE54, 0xFE61}, {0xFE63, 0xFE63}, {0xFE68, 0xFE68}, {0xFE6A, 0xFE6B}, {0xFF01, 0xFF03}, {0xFF05, 0xFF0A}, {0xFF0C, 0xFF0F}, {0xFF1A, 0xFF1B}, {0xFF1F, 0xFF20}, {0xFF3B, 0xFF3D}, {0xFF3F, 0xFF3F}, {0xFF5B, 0xFF5B}, {0xFF5D, 0xFF5D}, {0xFF5F, 0xFF65}, {0x10100, 0x10102}, {0x1039F, 0x1039F}, {0x103D0, 0x103D0}, {0x1056F, 0x1056F}, {0x10857, 0x10857}, {0x1091F, 0x1091F}, {0x1093F, 0x1093F}, {0x10A50, 0x10A58}, {0x10A7F, 0x10A7F}, {0x10AF0, 0x10AF6}, {0x10B39, 0x10B3F}, {0x10B99, 0x10B9C}, {0x10EAD, 0x10EAD}, {0x10F55, 0x10F59}, {0x11047, 0x1104D}, {0x110BB, 0x110BC}, {0x110BE, 0x110C1}, {0x11140, 0x11143}, {0x11174, 0x11175}, {0x111C5, 0x111C8}, {0x111CD, 0x111CD}, {0x111DB, 0x111DB}, {0x111DD, 0x111DF}, {0x11238, 0x1123D}, {0x112A9, 0x112A9}, {0x1144B, 0x1144F}, {0x1145A, 0x1145B}, {0x1145D, 0x1145D}, {0x114C6, 0x114C6}, {0x115C1, 0x115D7}, {0x11641, 0x11643}, {0x11660, 0x1166C}, {0x1173C, 0x1173E}, {0x1183B, 0x1183B}, {0x11944, 0x11946}, {0x119E2, 0x119E2}, {0x11A3F, 0x11A46}, {0x11A9A, 0x11A9C}, {0x11A9E, 0x11AA2}, {0x11C41, 0x11C45}, {0x11C70, 0x11C71}, {0x11EF7, 0x11EF8}, {0x11FFF, 0x11FFF}, {0x12470, 0x12474}, {0x16A6E, 0x16A6F}, {0x16AF5, 0x16AF5}, {0x16B37, 0x16B3B}, {0x16B44, 0x16B44}, {0x16E97, 0x16E9A}, {0x16FE2, 0x16FE2}, {0x1BC9F, 0x1BC9F}, {0x1DA87, 0x1DA8B}, {0x1E95E, 0x1E95F}, };
+
+static const std::vector<std::pair<int, int>> symbol_ranges = {
+{0x24, 0x24}, {0x2B, 0x2B}, {0x3C, 0x3E}, {0x5E, 0x5E}, {0x60, 0x60}, {0x7C, 0x7C}, {0x7E, 0x7E}, {0xA2, 0xA6}, {0xA8, 0xA9}, {0xAC, 0xAC}, {0xAE, 0xB1}, {0xB4, 0xB4}, {0xB8, 0xB8}, {0xD7, 0xD7}, {0xF7, 0xF7}, {0x2C2, 0x2C5}, {0x2D2, 0x2DF}, {0x2E5, 0x2EB}, {0x2ED, 0x2ED}, {0x2EF, 0x2FF}, {0x375, 0x375}, {0x384, 0x385}, {0x3F6, 0x3F6}, {0x482, 0x482}, {0x58D, 0x58F}, {0x606, 0x608}, {0x60B, 0x60B}, {0x60E, 0x60F}, {0x6DE, 0x6DE}, {0x6E9, 0x6E9}, {0x6FD, 0x6FE}, {0x7F6, 0x7F6}, {0x7FE, 0x7FF}, {0x9F2, 0x9F3}, {0x9FA, 0x9FB}, {0xAF1, 0xAF1}, {0xB70, 0xB70}, {0xBF3, 0xBFA}, {0xC7F, 0xC7F}, {0xD4F, 0xD4F}, {0xD79, 0xD79}, {0xE3F, 0xE3F}, {0xF01, 0xF03}, {0xF13, 0xF13}, {0xF15, 0xF17}, {0xF1A, 0xF1F}, {0xF34, 0xF34}, {0xF36, 0xF36}, {0xF38, 0xF38}, {0xFBE, 0xFC5}, {0xFC7, 0xFCC}, {0xFCE, 0xFCF}, {0xFD5, 0xFD8}, {0x109E, 0x109F}, {0x1390, 0x1399}, {0x166D, 0x166D}, {0x17DB, 0x17DB}, {0x1940, 0x1940}, {0x19DE, 0x19FF}, {0x1B61, 0x1B6A}, {0x1B74, 0x1B7C}, {0x1FBD, 0x1FBD}, {0x1FBF, 0x1FC1}, {0x1FCD, 0x1FCF}, {0x1FDD, 0x1FDF}, {0x1FED, 0x1FEF}, {0x1FFD, 0x1FFE}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207A, 0x207C}, {0x208A, 0x208C}, {0x20A0, 0x20BF}, {0x2100, 0x2101}, {0x2103, 0x2106}, {0x2108, 0x2109}, {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211E, 0x2123}, {0x2125, 0x2125}, {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212E, 0x212E}, {0x213A, 0x213B}, {0x2140, 0x2144}, {0x214A, 0x214D}, {0x214F, 0x214F}, {0x218A, 0x218B}, {0x2190, 0x2307}, {0x230C, 0x2328}, {0x232B, 0x2426}, {0x2440, 0x244A}, {0x249C, 0x24E9}, {0x2500, 0x2767}, {0x2794, 0x27C4}, {0x27C7, 0x27E5}, {0x27F0, 0x2982}, {0x2999, 0x29D7}, {0x29DC, 0x29FB}, {0x29FE, 0x2B73}, {0x2B76, 0x2B95}, {0x2B97, 0x2BFF}, {0x2CE5, 0x2CEA}, {0x2E50, 0x2E51}, {0x2E80, 0x2E99}, {0x2E9B, 0x2EF3}, {0x2F00, 0x2FD5}, {0x2FF0, 0x2FFB}, {0x3004, 0x3004}, {0x3012, 0x3013}, {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303E, 0x303F}, {0x309B, 0x309C}, {0x3190, 0x3191}, {0x3196, 0x319F}, {0x31C0, 0x31E3}, {0x3200, 0x321E}, {0x322A, 0x3247}, {0x3250, 0x3250}, {0x3260, 0x327F}, {0x328A, 0x32B0}, {0x32C0, 0x33FF}, {0x4DC0, 0x4DFF}, {0xA490, 0xA4C6}, {0xA700, 0xA716}, {0xA720, 0xA721}, {0xA789, 0xA78A}, {0xA828, 0xA82B}, {0xA836, 0xA839}, {0xAA77, 0xAA79}, {0xAB5B, 0xAB5B}, {0xAB6A, 0xAB6B}, {0xFB29, 0xFB29}, {0xFBB2, 0xFBC1}, {0xFDFC, 0xFDFD}, {0xFE62, 0xFE62}, {0xFE64, 0xFE66}, {0xFE69, 0xFE69}, {0xFF04, 0xFF04}, {0xFF0B, 0xFF0B}, {0xFF1C, 0xFF1E}, {0xFF3E, 0xFF3E}, {0xFF40, 0xFF40}, {0xFF5C, 0xFF5C}, {0xFF5E, 0xFF5E}, {0xFFE0, 0xFFE6}, {0xFFE8, 0xFFEE}, {0xFFFC, 0xFFFD}, {0x10137, 0x1013F}, {0x10179, 0x10189}, {0x1018C, 0x1018E}, {0x10190, 0x1019C}, {0x101A0, 0x101A0}, {0x101D0, 0x101FC}, {0x10877, 0x10878}, {0x10AC8, 0x10AC8}, {0x1173F, 0x1173F}, {0x11FD5, 0x11FF1}, {0x16B3C, 0x16B3F}, {0x16B45, 0x16B45}, {0x1BC9C, 0x1BC9C}, {0x1D000, 0x1D0F5}, {0x1D100, 0x1D126}, {0x1D129, 0x1D164}, {0x1D16A, 0x1D16C}, {0x1D183, 0x1D184}, {0x1D18C, 0x1D1A9}, {0x1D1AE, 0x1D1E8}, {0x1D200, 0x1D241}, {0x1D245, 0x1D245}, {0x1D300, 0x1D356}, {0x1D6C1, 0x1D6C1}, {0x1D6DB, 0x1D6DB}, {0x1D6FB, 0x1D6FB}, {0x1D715, 0x1D715}, {0x1D735, 0x1D735}, {0x1D74F, 0x1D74F}, {0x1D76F, 0x1D76F}, {0x1D789, 0x1D789}, {0x1D7A9, 0x1D7A9}, {0x1D7C3, 0x1D7C3}, {0x1D800, 0x1D9FF}, {0x1DA37, 0x1DA3A}, {0x1DA6D, 0x1DA74}, {0x1DA76, 0x1DA83}, {0x1DA85, 0x1DA86}, {0x1E14F, 0x1E14F}, {0x1E2FF, 0x1E2FF}, {0x1ECAC, 0x1ECAC}, {0x1ECB0, 0x1ECB0}, {0x1ED2E, 0x1ED2E}, {0x1EEF0, 0x1EEF1}, {0x1F000, 0x1F02B}, {0x1F030, 0x1F093}, {0x1F0A0, 0x1F0AE}, {0x1F0B1, 0x1F0BF}, {0x1F0C1, 0x1F0CF}, {0x1F0D1, 0x1F0F5}, {0x1F10D, 0x1F1AD}, {0x1F1E6, 0x1F202}, {0x1F210, 0x1F23B}, {0x1F240, 0x1F248}, {0x1F250, 0x1F251}, {0x1F260, 0x1F265}, {0x1F300, 0x1F6D7}, {0x1F6E0, 0x1F6EC}, {0x1F6F0, 0x1F6FC}, {0x1F700, 0x1F773}, {0x1F780, 0x1F7D8}, {0x1F7E0, 0x1F7EB}, {0x1F800, 0x1F80B}, {0x1F810, 0x1F847}, {0x1F850, 0x1F859}, {0x1F860, 0x1F887}, {0x1F890, 0x1F8AD}, {0x1F8B0, 0x1F8B1}, {0x1F900, 0x1F978}, {0x1F97A, 0x1F9CB}, {0x1F9CD, 0x1FA53}, {0x1FA60, 0x1FA6D}, {0x1FA70, 0x1FA74}, {0x1FA78, 0x1FA7A}, {0x1FA80, 0x1FA86}, {0x1FA90, 0x1FAA8}, {0x1FAB0, 0x1FAB6}, {0x1FAC0, 0x1FAC2}, {0x1FAD0, 0x1FAD6}, {0x1FB00, 0x1FB92}, {0x1FB94, 0x1FBCA}, };
+
+static const std::vector<std::pair<int, int>> control_ranges = {
+{0x0, 0x8}, {0xE, 0x1B}, {0x7F, 0x84}, {0x86, 0x9F}, {0xAD, 0xAD}, {0x378, 0x379}, {0x380, 0x383}, {0x38B, 0x38B}, {0x38D, 0x38D}, {0x3A2, 0x3A2}, {0x530, 0x530}, {0x557, 0x558}, {0x58B, 0x58C}, {0x590, 0x590}, {0x5C8, 0x5CF}, {0x5EB, 0x5EE}, {0x5F5, 0x605}, {0x61C, 0x61D}, {0x6DD, 0x6DD}, {0x70E, 0x70F}, {0x74B, 0x74C}, {0x7B2, 0x7BF}, {0x7FB, 0x7FC}, {0x82E, 0x82F}, {0x83F, 0x83F}, {0x85C, 0x85D}, {0x85F, 0x85F}, {0x86B, 0x89F}, {0x8B5, 0x8B5}, {0x8C8, 0x8D2}, {0x8E2, 0x8E2}, {0x984, 0x984}, {0x98D, 0x98E}, {0x991, 0x992}, {0x9A9, 0x9A9}, {0x9B1, 0x9B1}, {0x9B3, 0x9B5}, {0x9BA, 0x9BB}, {0x9C5, 0x9C6}, {0x9C9, 0x9CA}, {0x9CF, 0x9D6}, {0x9D8, 0x9DB}, {0x9DE, 0x9DE}, {0x9E4, 0x9E5}, {0x9FF, 0xA00}, {0xA04, 0xA04}, {0xA0B, 0xA0E}, {0xA11, 0xA12}, {0xA29, 0xA29}, {0xA31, 0xA31}, {0xA34, 0xA34}, {0xA37, 0xA37}, {0xA3A, 0xA3B}, {0xA3D, 0xA3D}, {0xA43, 0xA46}, {0xA49, 0xA4A}, {0xA4E, 0xA50}, {0xA52, 0xA58}, {0xA5D, 0xA5D}, {0xA5F, 0xA65}, {0xA77, 0xA80}, {0xA84, 0xA84}, {0xA8E, 0xA8E}, {0xA92, 0xA92}, {0xAA9, 0xAA9}, {0xAB1, 0xAB1}, {0xAB4, 0xAB4}, {0xABA, 0xABB}, {0xAC6, 0xAC6}, {0xACA, 0xACA}, {0xACE, 0xACF}, {0xAD1, 0xADF}, {0xAE4, 0xAE5}, {0xAF2, 0xAF8}, {0xB00, 0xB00}, {0xB04, 0xB04}, {0xB0D, 0xB0E}, {0xB11, 0xB12}, {0xB29, 0xB29}, {0xB31, 0xB31}, {0xB34, 0xB34}, {0xB3A, 0xB3B}, {0xB45, 0xB46}, {0xB49, 0xB4A}, {0xB4E, 0xB54}, {0xB58, 0xB5B}, {0xB5E, 0xB5E}, {0xB64, 0xB65}, {0xB78, 0xB81}, {0xB84, 0xB84}, {0xB8B, 0xB8D}, {0xB91, 0xB91}, {0xB96, 0xB98}, {0xB9B, 0xB9B}, {0xB9D, 0xB9D}, {0xBA0, 0xBA2}, {0xBA5, 0xBA7}, {0xBAB, 0xBAD}, {0xBBA, 0xBBD}, {0xBC3, 0xBC5}, {0xBC9, 0xBC9}, {0xBCE, 0xBCF}, {0xBD1, 0xBD6}, {0xBD8, 0xBE5}, {0xBFB, 0xBFF}, {0xC0D, 0xC0D}, {0xC11, 0xC11}, {0xC29, 0xC29}, {0xC3A, 0xC3C}, {0xC45, 0xC45}, {0xC49, 0xC49}, {0xC4E, 0xC54}, {0xC57, 0xC57}, {0xC5B, 0xC5F}, {0xC64, 0xC65}, {0xC70, 0xC76}, {0xC8D, 0xC8D}, {0xC91, 0xC91}, {0xCA9, 0xCA9}, {0xCB4, 0xCB4}, {0xCBA, 0xCBB}, {0xCC5, 0xCC5}, {0xCC9, 0xCC9}, {0xCCE, 0xCD4}, {0xCD7, 0xCDD}, {0xCDF, 0xCDF}, {0xCE4, 0xCE5}, {0xCF0, 0xCF0}, {0xCF3, 0xCFF}, {0xD0D, 0xD0D}, {0xD11, 0xD11}, {0xD45, 0xD45}, {0xD49, 0xD49}, {0xD50, 0xD53}, {0xD64, 0xD65}, {0xD80, 0xD80}, {0xD84, 0xD84}, {0xD97, 0xD99}, {0xDB2, 0xDB2}, {0xDBC, 0xDBC}, {0xDBE, 0xDBF}, {0xDC7, 0xDC9}, {0xDCB, 0xDCE}, {0xDD5, 0xDD5}, {0xDD7, 0xDD7}, {0xDE0, 0xDE5}, {0xDF0, 0xDF1}, {0xDF5, 0xE00}, {0xE3B, 0xE3E}, {0xE5C, 0xE80}, {0xE83, 0xE83}, {0xE85, 0xE85}, {0xE8B, 0xE8B}, {0xEA4, 0xEA4}, {0xEA6, 0xEA6}, {0xEBE, 0xEBF}, {0xEC5, 0xEC5}, {0xEC7, 0xEC7}, {0xECE, 0xECF}, {0xEDA, 0xEDB}, {0xEE0, 0xEFF}, {0xF48, 0xF48}, {0xF6D, 0xF70}, {0xF98, 0xF98}, {0xFBD, 0xFBD}, {0xFCD, 0xFCD}, {0xFDB, 0xFFF}, {0x10C6, 0x10C6}, {0x10C8, 0x10CC}, {0x10CE, 0x10CF}, {0x1249, 0x1249}, {0x124E, 0x124F}, {0x1257, 0x1257}, {0x1259, 0x1259}, {0x125E, 0x125F}, {0x1289, 0x1289}, {0x128E, 0x128F}, {0x12B1, 0x12B1}, {0x12B6, 0x12B7}, {0x12BF, 0x12BF}, {0x12C1, 0x12C1}, {0x12C6, 0x12C7}, {0x12D7, 0x12D7}, {0x1311, 0x1311}, {0x1316, 0x1317}, {0x135B, 0x135C}, {0x137D, 0x137F}, {0x139A, 0x139F}, {0x13F6, 0x13F7}, {0x13FE, 0x13FF}, {0x169D, 0x169F}, {0x16F9, 0x16FF}, {0x170D, 0x170D}, {0x1715, 0x171F}, {0x1737, 0x173F}, {0x1754, 0x175F}, {0x176D, 0x176D}, {0x1771, 0x1771}, {0x1774, 0x177F}, {0x17DE, 0x17DF}, {0x17EA, 0x17EF}, {0x17FA, 0x17FF}, {0x180E, 0x180F}, {0x181A, 0x181F}, {0x1879, 0x187F}, {0x18AB, 0x18AF}, {0x18F6, 0x18FF}, {0x191F, 0x191F}, {0x192C, 0x192F}, {0x193C, 0x193F}, {0x1941, 0x1943}, {0x196E, 0x196F}, {0x1975, 0x197F}, {0x19AC, 0x19AF}, {0x19CA, 0x19CF}, {0x19DB, 0x19DD}, {0x1A1C, 0x1A1D}, {0x1A5F, 0x1A5F}, {0x1A7D, 0x1A7E}, {0x1A8A, 0x1A8F}, {0x1A9A, 0x1A9F}, {0x1AAE, 0x1AAF}, {0x1AC1, 0x1AFF}, {0x1B4C, 0x1B4F}, {0x1B7D, 0x1B7F}, {0x1BF4, 0x1BFB}, {0x1C38, 0x1C3A}, {0x1C4A, 0x1C4C}, {0x1C89, 0x1C8F}, {0x1CBB, 0x1CBC}, {0x1CC8, 0x1CCF}, {0x1CFB, 0x1CFF}, {0x1DFA, 0x1DFA}, {0x1F16, 0x1F17}, {0x1F1E, 0x1F1F}, {0x1F46, 0x1F47}, {0x1F4E, 0x1F4F}, {0x1F58, 0x1F58}, {0x1F5A, 0x1F5A}, {0x1F5C, 0x1F5C}, {0x1F5E, 0x1F5E}, {0x1F7E, 0x1F7F}, {0x1FB5, 0x1FB5}, {0x1FC5, 0x1FC5}, {0x1FD4, 0x1FD5}, {0x1FDC, 0x1FDC}, {0x1FF0, 0x1FF1}, {0x1FF5, 0x1FF5}, {0x1FFF, 0x1FFF}, {0x200B, 0x200F}, {0x202A, 0x202E}, {0x2060, 0x206F}, {0x2072, 0x2073}, {0x208F, 0x208F}, {0x209D, 0x209F}, {0x20C0, 0x20CF}, {0x20F1, 0x20FF}, {0x218C, 0x218F}, {0x2427, 0x243F}, {0x244B, 0x245F}, {0x2B74, 0x2B75}, {0x2B96, 0x2B96}, {0x2C2F, 0x2C2F}, {0x2C5F, 0x2C5F}, {0x2CF4, 0x2CF8}, {0x2D26, 0x2D26}, {0x2D28, 0x2D2C}, {0x2D2E, 0x2D2F}, {0x2D68, 0x2D6E}, {0x2D71, 0x2D7E}, {0x2D97, 0x2D9F}, {0x2DA7, 0x2DA7}, {0x2DAF, 0x2DAF}, {0x2DB7, 0x2DB7}, {0x2DBF, 0x2DBF}, {0x2DC7, 0x2DC7}, {0x2DCF, 0x2DCF}, {0x2DD7, 0x2DD7}, {0x2DDF, 0x2DDF}, {0x2E53, 0x2E7F}, {0x2E9A, 0x2E9A}, {0x2EF4, 0x2EFF}, {0x2FD6, 0x2FEF}, {0x2FFC, 0x2FFF}, {0x3040, 0x3040}, {0x3097, 0x3098}, {0x3100, 0x3104}, {0x3130, 0x3130}, {0x318F, 0x318F}, {0x31E4, 0x31EF}, {0x321F, 0x321F}, {0x9FFD, 0x9FFF}, {0xA48D, 0xA48F}, {0xA4C7, 0xA4CF}, {0xA62C, 0xA63F}, {0xA6F8, 0xA6FF}, {0xA7C0, 0xA7C1}, {0xA7CB, 0xA7F4}, {0xA82D, 0xA82F}, {0xA83A, 0xA83F}, {0xA878, 0xA87F}, {0xA8C6, 0xA8CD}, {0xA8DA, 0xA8DF}, {0xA954, 0xA95E}, {0xA97D, 0xA97F}, {0xA9CE, 0xA9CE}, {0xA9DA, 0xA9DD}, {0xA9FF, 0xA9FF}, {0xAA37, 0xAA3F}, {0xAA4E, 0xAA4F}, {0xAA5A, 0xAA5B}, {0xAAC3, 0xAADA}, {0xAAF7, 0xAB00}, {0xAB07, 0xAB08}, {0xAB0F, 0xAB10}, {0xAB17, 0xAB1F}, {0xAB27, 0xAB27}, {0xAB2F, 0xAB2F}, {0xAB6C, 0xAB6F}, {0xABEE, 0xABEF}, {0xABFA, 0xABFF}, {0xD7A4, 0xD7AF}, {0xD7C7, 0xD7CA}, {0xD7FC, 0xF8FF}, {0xFA6E, 0xFA6F}, {0xFADA, 0xFAFF}, {0xFB07, 0xFB12}, {0xFB18, 0xFB1C}, {0xFB37, 0xFB37}, {0xFB3D, 0xFB3D}, {0xFB3F, 0xFB3F}, {0xFB42, 0xFB42}, {0xFB45, 0xFB45}, {0xFBC2, 0xFBD2}, {0xFD40, 0xFD4F}, {0xFD90, 0xFD91}, {0xFDC8, 0xFDEF}, {0xFDFE, 0xFDFF}, {0xFE1A, 0xFE1F}, {0xFE53, 0xFE53}, {0xFE67, 0xFE67}, {0xFE6C, 0xFE6F}, {0xFE75, 0xFE75}, {0xFEFD, 0xFF00}, {0xFFBF, 0xFFC1}, {0xFFC8, 0xFFC9}, {0xFFD0, 0xFFD1}, {0xFFD8, 0xFFD9}, {0xFFDD, 0xFFDF}, {0xFFE7, 0xFFE7}, {0xFFEF, 0xFFFB}, {0xFFFE, 0xFFFF}, {0x1000C, 0x1000C}, {0x10027, 0x10027}, {0x1003B, 0x1003B}, {0x1003E, 0x1003E}, {0x1004E, 0x1004F}, {0x1005E, 0x1007F}, {0x100FB, 0x100FF}, {0x10103, 0x10106}, {0x10134, 0x10136}, {0x1018F, 0x1018F}, {0x1019D, 0x1019F}, {0x101A1, 0x101CF}, {0x101FE, 0x1027F}, {0x1029D, 0x1029F}, {0x102D1, 0x102DF}, {0x102FC, 0x102FF}, {0x10324, 0x1032C}, {0x1034B, 0x1034F}, {0x1037B, 0x1037F}, {0x1039E, 0x1039E}, {0x103C4, 0x103C7}, {0x103D6, 0x103FF}, {0x1049E, 0x1049F}, {0x104AA, 0x104AF}, {0x104D4, 0x104D7}, {0x104FC, 0x104FF}, {0x10528, 0x1052F}, {0x10564, 0x1056E}, {0x10570, 0x105FF}, {0x10737, 0x1073F}, {0x10756, 0x1075F}, {0x10768, 0x107FF}, {0x10806, 0x10807}, {0x10809, 0x10809}, {0x10836, 0x10836}, {0x10839, 0x1083B}, {0x1083D, 0x1083E}, {0x10856, 0x10856}, {0x1089F, 0x108A6}, {0x108B0, 0x108DF}, {0x108F3, 0x108F3}, {0x108F6, 0x108FA}, {0x1091C, 0x1091E}, {0x1093A, 0x1093E}, {0x10940, 0x1097F}, {0x109B8, 0x109BB}, {0x109D0, 0x109D1}, {0x10A04, 0x10A04}, {0x10A07, 0x10A0B}, {0x10A14, 0x10A14}, {0x10A18, 0x10A18}, {0x10A36, 0x10A37}, {0x10A3B, 0x10A3E}, {0x10A49, 0x10A4F}, {0x10A59, 0x10A5F}, {0x10AA0, 0x10ABF}, {0x10AE7, 0x10AEA}, {0x10AF7, 0x10AFF}, {0x10B36, 0x10B38}, {0x10B56, 0x10B57}, {0x10B73, 0x10B77}, {0x10B92, 0x10B98}, {0x10B9D, 0x10BA8}, {0x10BB0, 0x10BFF}, {0x10C49, 0x10C7F}, {0x10CB3, 0x10CBF}, {0x10CF3, 0x10CF9}, {0x10D28, 0x10D2F}, {0x10D3A, 0x10E5F}, {0x10E7F, 0x10E7F}, {0x10EAA, 0x10EAA}, {0x10EAE, 0x10EAF}, {0x10EB2, 0x10EFF}, {0x10F28, 0x10F2F}, {0x10F5A, 0x10FAF}, {0x10FCC, 0x10FDF}, {0x10FF7, 0x10FFF}, {0x1104E, 0x11051}, {0x11070, 0x1107E}, {0x110BD, 0x110BD}, {0x110C2, 0x110CF}, {0x110E9, 0x110EF}, {0x110FA, 0x110FF}, {0x11135, 0x11135}, {0x11148, 0x1114F}, {0x11177, 0x1117F}, {0x111E0, 0x111E0}, {0x111F5, 0x111FF}, {0x11212, 0x11212}, {0x1123F, 0x1127F}, {0x11287, 0x11287}, {0x11289, 0x11289}, {0x1128E, 0x1128E}, {0x1129E, 0x1129E}, {0x112AA, 0x112AF}, {0x112EB, 0x112EF}, {0x112FA, 0x112FF}, {0x11304, 0x11304}, {0x1130D, 0x1130E}, {0x11311, 0x11312}, {0x11329, 0x11329}, {0x11331, 0x11331}, {0x11334, 0x11334}, {0x1133A, 0x1133A}, {0x11345, 0x11346}, {0x11349, 0x1134A}, {0x1134E, 0x1134F}, {0x11351, 0x11356}, {0x11358, 0x1135C}, {0x11364, 0x11365}, {0x1136D, 0x1136F}, {0x11375, 0x113FF}, {0x1145C, 0x1145C}, {0x11462, 0x1147F}, {0x114C8, 0x114CF}, {0x114DA, 0x1157F}, {0x115B6, 0x115B7}, {0x115DE, 0x115FF}, {0x11645, 0x1164F}, {0x1165A, 0x1165F}, {0x1166D, 0x1167F}, {0x116B9, 0x116BF}, {0x116CA, 0x116FF}, {0x1171B, 0x1171C}, {0x1172C, 0x1172F}, {0x11740, 0x117FF}, {0x1183C, 0x1189F}, {0x118F3, 0x118FE}, {0x11907, 0x11908}, {0x1190A, 0x1190B}, {0x11914, 0x11914}, {0x11917, 0x11917}, {0x11936, 0x11936}, {0x11939, 0x1193A}, {0x11947, 0x1194F}, {0x1195A, 0x1199F}, {0x119A8, 0x119A9}, {0x119D8, 0x119D9}, {0x119E5, 0x119FF}, {0x11A48, 0x11A4F}, {0x11AA3, 0x11ABF}, {0x11AF9, 0x11BFF}, {0x11C09, 0x11C09}, {0x11C37, 0x11C37}, {0x11C46, 0x11C4F}, {0x11C6D, 0x11C6F}, {0x11C90, 0x11C91}, {0x11CA8, 0x11CA8}, {0x11CB7, 0x11CFF}, {0x11D07, 0x11D07}, {0x11D0A, 0x11D0A}, {0x11D37, 0x11D39}, {0x11D3B, 0x11D3B}, {0x11D3E, 0x11D3E}, {0x11D48, 0x11D4F}, {0x11D5A, 0x11D5F}, {0x11D66, 0x11D66}, {0x11D69, 0x11D69}, {0x11D8F, 0x11D8F}, {0x11D92, 0x11D92}, {0x11D99, 0x11D9F}, {0x11DAA, 0x11EDF}, {0x11EF9, 0x11FAF}, {0x11FB1, 0x11FBF}, {0x11FF2, 0x11FFE}, {0x1239A, 0x123FF}, {0x1246F, 0x1246F}, {0x12475, 0x1247F}, {0x12544, 0x12FFF}, {0x1342F, 0x143FF}, {0x14647, 0x167FF}, {0x16A39, 0x16A3F}, {0x16A5F, 0x16A5F}, {0x16A6A, 0x16A6D}, {0x16A70, 0x16ACF}, {0x16AEE, 0x16AEF}, {0x16AF6, 0x16AFF}, {0x16B46, 0x16B4F}, {0x16B5A, 0x16B5A}, {0x16B62, 0x16B62}, {0x16B78, 0x16B7C}, {0x16B90, 0x16E3F}, {0x16E9B, 0x16EFF}, {0x16F4B, 0x16F4E}, {0x16F88, 0x16F8E}, {0x16FA0, 0x16FDF}, {0x16FE5, 0x16FEF}, {0x16FF2, 0x16FFF}, {0x187F8, 0x187FF}, {0x18CD6, 0x18CFF}, {0x18D09, 0x1AFFF}, {0x1B11F, 0x1B14F}, {0x1B153, 0x1B163}, {0x1B168, 0x1B16F}, {0x1B2FC, 0x1BBFF}, {0x1BC6B, 0x1BC6F}, {0x1BC7D, 0x1BC7F}, {0x1BC89, 0x1BC8F}, {0x1BC9A, 0x1BC9B}, {0x1BCA0, 0x1CFFF}, {0x1D0F6, 0x1D0FF}, {0x1D127, 0x1D128}, {0x1D173, 0x1D17A}, {0x1D1E9, 0x1D1FF}, {0x1D246, 0x1D2DF}, {0x1D2F4, 0x1D2FF}, {0x1D357, 0x1D35F}, {0x1D379, 0x1D3FF}, {0x1D455, 0x1D455}, {0x1D49D, 0x1D49D}, {0x1D4A0, 0x1D4A1}, {0x1D4A3, 0x1D4A4}, {0x1D4A7, 0x1D4A8}, {0x1D4AD, 0x1D4AD}, {0x1D4BA, 0x1D4BA}, {0x1D4BC, 0x1D4BC}, {0x1D4C4, 0x1D4C4}, {0x1D506, 0x1D506}, {0x1D50B, 0x1D50C}, {0x1D515, 0x1D515}, {0x1D51D, 0x1D51D}, {0x1D53A, 0x1D53A}, {0x1D53F, 0x1D53F}, {0x1D545, 0x1D545}, {0x1D547, 0x1D549}, {0x1D551, 0x1D551}, {0x1D6A6, 0x1D6A7}, {0x1D7CC, 0x1D7CD}, {0x1DA8C, 0x1DA9A}, {0x1DAA0, 0x1DAA0}, {0x1DAB0, 0x1DFFF}, {0x1E007, 0x1E007}, {0x1E019, 0x1E01A}, {0x1E022, 0x1E022}, {0x1E025, 0x1E025}, {0x1E02B, 0x1E0FF}, {0x1E12D, 0x1E12F}, {0x1E13E, 0x1E13F}, {0x1E14A, 0x1E14D}, {0x1E150, 0x1E2BF}, {0x1E2FA, 0x1E2FE}, {0x1E300, 0x1E7FF}, {0x1E8C5, 0x1E8C6}, {0x1E8D7, 0x1E8FF}, {0x1E94C, 0x1E94F}, {0x1E95A, 0x1E95D}, {0x1E960, 0x1EC70}, {0x1ECB5, 0x1ED00}, {0x1ED3E, 0x1EDFF}, {0x1EE04, 0x1EE04}, {0x1EE20, 0x1EE20}, {0x1EE23, 0x1EE23}, {0x1EE25, 0x1EE26}, {0x1EE28, 0x1EE28}, {0x1EE33, 0x1EE33}, {0x1EE38, 0x1EE38}, {0x1EE3A, 0x1EE3A}, {0x1EE3C, 0x1EE41}, {0x1EE43, 0x1EE46}, {0x1EE48, 0x1EE48}, {0x1EE4A, 0x1EE4A}, {0x1EE4C, 0x1EE4C}, {0x1EE50, 0x1EE50}, {0x1EE53, 0x1EE53}, {0x1EE55, 0x1EE56}, {0x1EE58, 0x1EE58}, {0x1EE5A, 0x1EE5A}, {0x1EE5C, 0x1EE5C}, {0x1EE5E, 0x1EE5E}, {0x1EE60, 0x1EE60}, {0x1EE63, 0x1EE63}, {0x1EE65, 0x1EE66}, {0x1EE6B, 0x1EE6B}, {0x1EE73, 0x1EE73}, {0x1EE78, 0x1EE78}, {0x1EE7D, 0x1EE7D}, {0x1EE7F, 0x1EE7F}, {0x1EE8A, 0x1EE8A}, {0x1EE9C, 0x1EEA0}, {0x1EEA4, 0x1EEA4}, {0x1EEAA, 0x1EEAA}, {0x1EEBC, 0x1EEEF}, {0x1EEF2, 0x1EFFF}, {0x1F02C, 0x1F02F}, {0x1F094, 0x1F09F}, {0x1F0AF, 0x1F0B0}, {0x1F0C0, 0x1F0C0}, {0x1F0D0, 0x1F0D0}, {0x1F0F6, 0x1F0FF}, {0x1F1AE, 0x1F1E5}, {0x1F203, 0x1F20F}, {0x1F23C, 0x1F23F}, {0x1F249, 0x1F24F}, {0x1F252, 0x1F25F}, {0x1F266, 0x1F2FF}, {0x1F6D8, 0x1F6DF}, {0x1F6ED, 0x1F6EF}, {0x1F6FD, 0x1F6FF}, {0x1F774, 0x1F77F}, {0x1F7D9, 0x1F7DF}, {0x1F7EC, 0x1F7FF}, {0x1F80C, 0x1F80F}, {0x1F848, 0x1F84F}, {0x1F85A, 0x1F85F}, {0x1F888, 0x1F88F}, {0x1F8AE, 0x1F8AF}, {0x1F8B2, 0x1F8FF}, {0x1F979, 0x1F979}, {0x1F9CC, 0x1F9CC}, {0x1FA54, 0x1FA5F}, {0x1FA6E, 0x1FA6F}, {0x1FA75, 0x1FA77}, {0x1FA7B, 0x1FA7F}, {0x1FA87, 0x1FA8F}, {0x1FAA9, 0x1FAAF}, {0x1FAB7, 0x1FABF}, {0x1FAC3, 0x1FACF}, {0x1FAD7, 0x1FAFF}, {0x1FB93, 0x1FB93}, {0x1FBCB, 0x1FBEF}, {0x1FBFA, 0x1FFFF}, {0x2A6DE, 0x2A6FF}, {0x2B735, 0x2B73F}, {0x2B81E, 0x2B81F}, {0x2CEA2, 0x2CEAF}, {0x2EBE1, 0x2F7FF}, {0x2FA1E, 0x2FFFF}, {0x3134B, 0xE00FF}, {0xE01F0, 0x10FFFF}, };
+
+//String
+bool CNCTString::operator==(const std::string& other) const {
+    return str.compare(other) == 0;
+}
+bool CNCTString::operator==(const char other) const {
+    return str.compare(std::string(1, other)) == 0;
+}
+bool CNCTString::operator==(const CNCTString& other) const {
+    return str.compare(other.str) == 0;
+}
+// + operators
+CNCTString& CNCTString::operator+=(const std::string& other) {
+    str += other;
+    int new_len = CNCTUnicode::strlen_utf8(other);
+    utf8_chars += new_len;
+    char_type = CNCTUnicode::string_identify(str);
+    seq_offset_bytes += other.size();
+    seq_offset_utf8_chars += new_len;
+    return *this;
+}
+
+CNCTString& CNCTString::operator+=(const char other) {
+    std::string str = std::string(1, other);
+    *this += str;
+    return *this;
+}
+
+CNCTString& CNCTString::operator+=(const CNCTString& other) {
+    str += other.str;
+    utf8_chars += other.utf8_chars;
+    char_type = CNCTUnicode::string_identify(str);
+    seq_offset_bytes += other.str.size();
+    seq_offset_utf8_chars += other.utf8_chars;
+    return *this;
+}
+
+struct CRCompare {
+    bool operator()(const std::pair<int, int>& p, int i) {
+        return p.second < i;
+    }
+    bool operator()(int i, const std::pair<int, int>& p) {
+        return i < p.first;
+    }
+};
+
+// binary search for code range
+bool CNCTUnicode::check_code_range(int c, const std::vector<std::pair<int, int>> &ranges)
+{
+    auto it = std::upper_bound(ranges.begin(), ranges.end(), c, CRCompare());
+    if (it != ranges.begin()) {
+        --it;
+    }
+    return c >= it->first && c <= it->second;
+}
+
+// these are binary searches, it takes only a few operations
+CNCTCharType CNCTUnicode::get_code_type(int c)
+{
+    if (check_code_range(c, letter_ranges))
+        return LETTER;
+    if (check_code_range(c, digit_ranges))
+        return DIGIT;
+    if (check_code_range(c, whitespace_ranges))
+        return WHITESPACE;
+    if (check_code_range(c, punctuation_ranges))
+        return PUNCTUATION;
+    if (check_code_range(c, symbol_ranges))
+        return SYMBOL;
+    if (check_code_range(c, accent_mark_ranges))
+        return ACCENT_MARK;
+    if (check_code_range(c, control_ranges))
+        return CONTROL;
+    return UNIDENTIFIED;
+}
+
+static int utf8_to_unicode(const std::string& utf8_char) {
+    int c = 0;
+    int len = (int)utf8_char.size();
+    if (len == 1) {
+        c = utf8_char[0];
+    } else if (len == 2) {
+        c = ((utf8_char[0] & 0x1F) << 6) | (utf8_char[1] & 0x3F);
+    } else if (len == 3) {
+        c = ((utf8_char[0] & 0x0F) << 12) | ((utf8_char[1] & 0x3F) << 6) | (utf8_char[2] & 0x3F);
+    } else if (len == 4) {
+        c = ((utf8_char[0] & 0x07) << 18) | ((utf8_char[1] & 0x3F) << 12) | ((utf8_char[2] & 0x3F) << 6) | (utf8_char[3] & 0x3F);
+    }
+    return c;
+}
+
+CNCTCharType CNCTUnicode::get_code_type(const std::string &utf8_char)
+{
+    return get_code_type(utf8_to_unicode(utf8_char));
+}
+
+int CNCTUnicode::utf8_len(const char c)
+{
+    if ((c & 0x80) == 0)
+        return 1; // ASCII character
+    if ((c & 0xE0) == 0xC0)
+        return 2; // 2-byte character
+    if ((c & 0xF0) == 0xE0)
+        return 3; // 3-byte character
+    if ((c & 0xF0) == 0xF0)
+        return 4; // 4-byte character
+    return 1;     // not valid utf8
+    // static const uint8_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    // return lookup[static_cast<uint8_t>(c) >> 4];
+}
+
+int CNCTUnicode::strlen_utf8(const std::string src)
+{
+    int len = 0;
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
+    {
+        int char_len = utf8_len(*it);
+        if (char_len > 1)
+        {
+            it += char_len - 1;
+        }
+        len += 1;
+    }
+    return len;
+}
+
+// split a string into unicode strings
+std::vector<std::string> CNCTUnicode::split_utf8(const std::string &src)
+{
+    std::vector<std::string> result;
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
+    {
+        int char_len = utf8_len(*it);
+        std::string str(it, it + char_len);
+        result.push_back(str);
+        if (char_len > 1)
+        {
+            it += char_len - 1;
+        }
+    }
+    return result;
+}
+
+// split a string into unicode strings (CNCTString) with sequence information
+std::vector<CNCTString> CNCTUnicode::split_utf8_enhanced(const std::string &src)
+{
+    std::vector<CNCTString> result;
+    int seq_offset_bytes=0;
+    int seq_offset_utf8_chars=0;
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
+    {
+        int char_len = utf8_len(*it);
+        std::string str(it, it + char_len);
+        CNCTString cnct_str;
+        cnct_str.seq_offset_bytes = seq_offset_bytes;
+        cnct_str.seq_offset_utf8_chars = seq_offset_utf8_chars;
+        cnct_str.str = str;
+        cnct_str.utf8_chars = 1;
+        cnct_str.char_type = get_code_type(str);
+        #if 0
+        switch (cnct_str.char_type)
+        {
+        case DIGIT:
+            printf("%s = DIGIT\n", str.c_str());
+            break;
+        case LETTER:
+            printf("%s = LETTER\n", str.c_str());
+            break;
+        case WHITESPACE:
+            printf("%s = WHITESPACE\n", str.c_str());
+            break;
+        case PUNCTUATION:
+            printf("%s = PUNCTUATION\n", str.c_str());
+            break;
+        case UNIDENTIFIED:
+            printf("%s = UNIDENTIFIED\n", str.c_str());
+            break;
+        case SYMBOL:
+            printf("%s = SYMBOL\n", str.c_str());
+            break;
+        case CONTROL:
+            printf("%s = CONTROL\n", str.c_str());
+            break;
+        }
+        #endif
+
+
+        result.push_back(cnct_str);
+        seq_offset_bytes += char_len;
+        seq_offset_utf8_chars += 1;
+        if (char_len > 1)
+        {
+            it += char_len - 1;
+        }
+
+    }
+    return result;
+}
+
+// return the type of the string
+CNCTCharType CNCTUnicode::string_identify(const std::string &str)
+{
+    CNCTCharType result = UNIDENTIFIED;
+    std::string::const_iterator it = str.begin();
+    while (it != str.end())
+    {
+        int len = utf8_len(*it);
+        int c = 0;
+        for (int i = 0; i < len && it != str.end(); ++i, ++it)
+        {
+            c = (c << 8) | static_cast<unsigned char>(*it);
+        }
+        switch (get_code_type(c))
+        {
+        case DIGIT:
+            if (result == UNIDENTIFIED)
+            {
+                result = DIGIT;
+            }
+            else if (result != DIGIT)
+            {
+                return MIXED;
+            }
+            break;
+        case LETTER:
+            if (result == UNIDENTIFIED)
+            {
+                result = LETTER;
+            }
+            else if (result != LETTER)
+            {
+                return MIXED;
+            }
+            break;
+        case WHITESPACE:
+            if (result == UNIDENTIFIED)
+            {
+                result = WHITESPACE;
+            }
+            else if (result != WHITESPACE)
+            {
+                return MIXED;
+            }
+            break;
+        case PUNCTUATION:
+            if (result == UNIDENTIFIED)
+            {
+                result = PUNCTUATION;
+            }
+            else if (result != PUNCTUATION)
+            {
+                return MIXED;
+            }
+            break;
+        default:
+            return MIXED;
+            break;
+        }
+    }
+    return result;
+}
+
+// verify the content of a string
+bool CNCTUnicode::string_test(const std::string &str, CNCTCharType chartype)
+{
+    std::string::const_iterator it = str.begin();
+    while (it != str.end())
+    {
+        int len = utf8_len(*it);
+        int c = 0;
+        for (int i = 0; i < len && it != str.end(); ++i, ++it)
+        {
+            c = (c << 8) | static_cast<unsigned char>(*it);
+        }
+        if (get_code_type(c) != chartype)
+        {
+            return false;
+        }
+    }
+    return true;
+}
+
+// Ported from libfalcon.cpp (https://github.com/cmp-nct/ggllm.cpp)
+
+std::string replaceAll(std::string str, const std::string& from, const std::string& to) {
+    size_t start_pos = 0;
+    while((start_pos = str.find(from, start_pos)) != std::string::npos) {
+        str.replace(start_pos, from.length(), to);
+        start_pos += to.length(); // Handles case where 'to' is a substring of 'from'
+    }
+    return str;
+}
+
+struct TrieNode {
+    std::map<char, TrieNode*> map;
+    int32_t Id = -1;
+};
+
+struct Trie {
+    TrieNode *root;
+
+    Trie() : root(new TrieNode()) {}
+    ~Trie() {
+        if(root)
+        deleteTrie(root);
+    }
+        // Move constructor
+    Trie(Trie&& other) noexcept : root(other.root) {
+        other.root = nullptr;
+    }
+
+    // Move assignment operator
+    Trie& operator=(Trie&& other) noexcept {
+        if (this != &other) {
+            if(root)
+                deleteTrie(root);
+            root = other.root;
+            other.root = nullptr;
+        }
+        return *this;
+    }
+
+    void insert(const std::string &token, int32_t Id) {
+        TrieNode* current = root;
+        for(auto ch : token) {
+            if(current->map.find(ch) == current->map.end())
+                current->map[ch] = new TrieNode();
+            current = current->map[ch];
+        }
+        current->Id = Id;
+    }
+
+    void reset()
+    {
+        deleteTrie(root);
+        root = new TrieNode();
+    }
+
+private:
+    void deleteTrie(TrieNode* node) {
+        for(auto &it: node->map) {
+            deleteTrie(it.second);
+        }
+        delete node;
+    }
+
+};
+
+struct gpt2bpe_vocab {
+    using id = int32_t;
+    using token = std::string;
+    std::map<std::string, uint32_t> max_token_length; // max length, for each 2byte prefix
+
+    std::map<std::pair<std::string,std::string>, int> bpe_ranks;
+    std::vector<std::pair<std::string, std::string>> bpe_merges;
+    std::map<std::string, int> special_tokens;
+
+    id special_bos_id = 0;
+    id special_eos_id = 0;
+    id special_unk_id = 0;
+    id special_sep_id = 0;
+    id special_pad_id = 0;
+
+    bool special_have_bos = false;
+    bool special_have_eos = false;
+    bool special_have_unk = false;
+    bool special_have_sep = false;
+    bool special_have_pad = false;
+
+    std::unordered_map<token, id> token_to_id;
+    std::unordered_map<id, token> id_to_token;
+
+    Trie trie; // highspeed access to tokens by prefix tree
+
+    // populate trie from map
+    void populate_trie_from_map() {
+        trie.reset();
+        for (const auto& pair : token_to_id) {
+            trie.insert(pair.first, pair.second);
+            if (pair.first.size() >= 2)
+            {
+                std::string prefix = pair.first.substr(0, 2);
+                max_token_length[prefix] = std::max(max_token_length[prefix], (uint32_t)pair.first.size());
+            }
+        }
+    }
+    // populate token ranks map
+    int populate_bpe_ranks(std::vector<std::pair<std::string, std::string>> bpe_merges_) {
+        for (int i = 0; i < (int)bpe_merges_.size(); i++) {
+            bpe_ranks.emplace(bpe_merges_[i], i);
+        }
+        bpe_merges = bpe_merges_;
+
+        // populate special tokens too (0-11 and if available 65024++)
+
+        #if 0
+        for (int i = 0; i < 12; i++) {
+            special_tokens[id_to_token[i].tok] = i;
+        }
+        for (int i = 65024; i < (int)id_to_token.size(); i++) {
+            special_tokens[id_to_token[i].tok] = i;
+        }
+        #endif
+
+        // token_to_id["</s>"] = 11; // bugfix for TII instruct training (blocks stopwords)
+        // special_tokens["</s>"] = 11; // bugfix for TII instruct training (blocks stopwords)
+
+
+        return bpe_merges_.size();
+    }
+
+    // Trim whitespace characters from the beginning and end of the string
+    void trim(std::string& str) {
+        // Remove whitespace characters from the beginning of the string
+        str.erase(str.begin(), std::find_if(str.begin(), str.end(), [](int ch) {
+            return !std::isspace(ch);
+        }));
+
+        // Remove whitespace characters from the end of the string
+        str.erase(std::find_if(str.rbegin(), str.rend(), [](int ch) {
+            return !std::isspace(ch);
+        }).base(), str.end());
+    }
+
+    // removed, merges loaded from gguf model file:
+    // requires the standard HF type tokenizer.json (pretty printed)
+    // std::vector<std::pair<std::string, std::string>> parse_json_to_bpe_merges(const std::string& filename) {
+
+    // get max token length available for a prefix of 2 bytes (string at least 2 bytes long)
+    int get_max_token_length(const std::string& string) const {
+        if (string.size() < 2)
+            return -1;
+        std::string prefix = string.substr(0, 2);
+        if (max_token_length.find(prefix) == max_token_length.end())
+            return 0;
+        return max_token_length.at(prefix);
+    }
+
+    // function to find if two tokens match in bpe_rank, return rank or -1
+    int find_bpe_rank(const std::string& token1, const std::string& token2) const
+    {
+        std::string left_token = token1;
+        std::string right_token = token2;
+        left_token = replaceAll(left_token, " ", "Ġ");
+        left_token = replaceAll(left_token, "\n", "Ċ");
+        right_token = replaceAll(right_token, " ", "Ġ");
+        right_token = replaceAll(right_token, "\n", "Ċ");
+
+        auto it = bpe_ranks.find(std::make_pair(left_token, right_token));
+        if (it == bpe_ranks.end())
+            return -1;
+        return it->second;
+    }
+
+    std::pair<gpt2bpe_vocab::id, std::string> find_longest_match(const std::string& snippet) const {
+        TrieNode* current = trie.root;
+        gpt2bpe_vocab::id last_matched_id = -1;
+        std::string last_matched_token = "";
+        std::string current_token = "";
+        for (auto ch : snippet) {
+            if (current->map.find(ch) == current->map.end()) {
+                break;
+            }
+            current = current->map[ch];
+            current_token += ch;
+            if (current->Id != -1) {
+                last_matched_id = current->Id;
+                last_matched_token = current_token;
+            }
+        }
+        return {last_matched_id, last_matched_token};
+    }
+
+};
+
+
+//
+// tokenizer - bpe type, gpt2 tokenization compatible
+//
+
+struct ggllm_bpe_symbol {
+    using index = int;
+    index prev;
+    index next;
+    const char * text;
+    size_t n;
+};
+
+static_assert(std::is_trivially_copyable<ggllm_bpe_symbol>::value, "ggllm_bpe_symbol is not trivially copyable");
+
+struct ggllm_bpe_bigram {
+    struct comparator
+    {
+        bool operator()(ggllm_bpe_bigram & l, ggllm_bpe_bigram & r)
+        {
+            return l.rank > r.rank || (l.rank == r.rank && l.left > r.left);
+        }
+    };
+
+    using queue_storage = std::vector<ggllm_bpe_bigram>;
+    using queue = std::priority_queue<ggllm_bpe_bigram, queue_storage, comparator>;
+    ggllm_bpe_symbol::index left;
+    ggllm_bpe_symbol::index right;
+    std::string text;
+    int rank;
+    size_t size;
+};
+
+struct gpt2bpe_tokenizer {
+    gpt2bpe_tokenizer(const gpt2bpe_vocab & vocab, bool g2ws_): vocab_(vocab) { flag_g2ws = g2ws_; }
+
+    void tokenize(const std::string & text, std::vector<gpt2bpe_vocab::id> & output) {
+        int final_prev_index = -1;
+        // auto start = ggml_time_us();
+        auto word_collection = bpe_gpt2_preprocess(text);
+        // auto end = ggml_time_us();
+        // fprintf(stderr, "%s: preprocessing took %0.3f ms\n", __func__, (end - start) / 1000.0);
+
+
+        symbols_final.clear();
+
+        for (auto & word : word_collection)
+        {
+            work_queue_ = ggllm_bpe_bigram::queue();
+            symbols_.clear();
+            bool is_special = false;
+            for (auto it = vocab_.special_tokens.begin(); it != vocab_.special_tokens.end(); ++it)
+            {
+                std::string special_token = it->first;
+                if (word.compare(special_token) == 0)
+                {
+                    ggllm_bpe_symbol sym;
+                    sym.text = word.c_str();
+                    sym.n = word.size();
+                    sym.prev = -1;
+                    sym.next = -1;
+                    symbols_.emplace_back(sym);
+                    is_special = true;
+                    break;
+                }
+            }
+
+            int index = 0;
+            size_t offset = 0;
+            if (!is_special)
+            {
+
+                while (offset < word.size())
+                {
+                    ggllm_bpe_symbol sym;
+                    size_t char_len = std::min(word.size() - offset, (size_t) CNCTUnicode::utf8_len(word[offset]));
+                    sym.text = word.c_str() + offset;
+                    sym.n = 1;
+                    sym.n = char_len;
+                    offset += sym.n;
+                    sym.prev = index - 1;
+                    sym.next = offset == word.size() ? -1 : index + 1;
+                    index++;
+                    symbols_.emplace_back(sym);
+                }
+                for (size_t i = 1; i < symbols_.size(); ++i) {
+                    add_new_bigram(i - 1, i);
+                }
+            }
+            // build token(s)
+            while (!work_queue_.empty())
+            {
+                auto bigram = work_queue_.top();
+                work_queue_.pop();
+
+                auto & left_symbol = symbols_[bigram.left];
+                auto & right_symbol = symbols_[bigram.right];
+
+                if (left_symbol.n == 0 || right_symbol.n == 0) {
+                    continue;
+                }
+                std::string left_token = std::string(left_symbol.text, left_symbol.n);
+                std::string right_token = std::string(right_symbol.text, right_symbol.n);
+                if (left_token + right_token != bigram.text) {
+                    continue;  // Skip this bigram if it's outdated
+                }
+
+                // merge the right sym into the left one
+                left_symbol.n += right_symbol.n;
+                right_symbol.n = 0;
+
+                // remove the right sym from the chain
+                left_symbol.next = right_symbol.next;
+                if (right_symbol.next >= 0) {
+                    symbols_[right_symbol.next].prev = bigram.left;
+                }
+
+                add_new_bigram(left_symbol.prev, bigram.left);  // left side of current symbol
+                add_new_bigram(bigram.left, left_symbol.next);  // right side of current symbol
+            }
+            // add the fnished tokens to the final list keeping correct order for next and prev
+
+            for (auto & sym : symbols_)
+            {
+                if (sym.n > 0)
+                {
+                    sym.prev = final_prev_index;
+                    sym.next = -1;
+                    if (final_prev_index != -1)
+                    {
+                        symbols_final[final_prev_index].next = symbols_final.size();
+                    }
+                    symbols_final.emplace_back(sym);
+                    final_prev_index = symbols_final.size() - 1;
+                }
+            }
+        }
+
+
+        symbols_ = symbols_final;
+        if (symbols_.size())
+        for (int i = 0; i != -1; i = symbols_[i].next) {
+            auto & symbol = symbols_[i];
+            if (symbol.n == 0) {
+                continue;
+            }
+            std::string str = std::string(symbol.text, symbol.n);
+            std::string str_decoded = decode_token(str);
+            auto token = vocab_.token_to_id.find(str_decoded);
+
+            if (token == vocab_.token_to_id.end()) {
+                for (auto j = str_decoded.begin(); j != str_decoded.end(); ++j) {
+                    std::string byte_str(1, *j);
+                    auto token_multibyte = vocab_.token_to_id.find(byte_str);
+                    if (token_multibyte == vocab_.token_to_id.end()) {
+                        fprintf(stderr,"ERROR: byte not found in vocab: '%s'\n", byte_str.c_str());
+                    }
+                    output.push_back((*token_multibyte).second);
+                }
+            } else {
+                output.push_back((*token).second);
+            }
+        }
+    }
+
+private:
+    void add_new_bigram(int left, int right)
+    {
+        if (left == -1 || right == -1)  return;
+
+        std::string left_token = std::string(symbols_[left].text, symbols_[left].n);
+        std::string right_token = std::string(symbols_[right].text, symbols_[right].n);
+
+        int rank_found = -1;
+        rank_found = vocab_.find_bpe_rank(left_token, right_token);
+
+        if (rank_found < 0) {
+            return;
+        }
+
+        ggllm_bpe_bigram bigram;
+        bigram.left = left;
+        bigram.right = right;
+        bigram.rank = rank_found;
+        bigram.size = left_token.size() + right_token.size();
+        bigram.text = left_token + right_token;
+        work_queue_.push(bigram);
+    }
+
+    std::unordered_map<unsigned char, std::string> bytes_to_unicode() {
+        static std::unordered_map<unsigned char, std::string> hex_map = { { 0x21, "\x21" }, { 0x22, "\x22" }, { 0x23, "\x23" }, { 0x24, "\x24" }, { 0x25, "\x25" }, { 0x26, "\x26" }, { 0x27, "\x27" }, { 0x28, "\x28" }, { 0x29, "\x29" }, { 0x2A, "\x2A" }, { 0x2B, "\x2B" }, { 0x2C, "\x2C" }, { 0x2D, "\x2D" }, { 0x2E, "\x2E" }, { 0x2F, "\x2F" }, { 0x30, "\x30" }, { 0x31, "\x31" }, { 0x32, "\x32" }, { 0x33, "\x33" }, { 0x34, "\x34" }, { 0x35, "\x35" }, { 0x36, "\x36" }, { 0x37, "\x37" }, { 0x38, "\x38" }, { 0x39, "\x39" }, { 0x3A, "\x3A" }, { 0x3B, "\x3B" }, { 0x3C, "\x3C" }, { 0x3D, "\x3D" }, { 0x3E, "\x3E" }, { 0x3F, "\x3F" }, { 0x40, "\x40" }, { 0x41, "\x41" }, { 0x42, "\x42" }, { 0x43, "\x43" }, { 0x44, "\x44" }, { 0x45, "\x45" }, { 0x46, "\x46" }, { 0x47, "\x47" }, { 0x48, "\x48" }, { 0x49, "\x49" }, { 0x4A, "\x4A" }, { 0x4B, "\x4B" }, { 0x4C, "\x4C" }, { 0x4D, "\x4D" }, { 0x4E, "\x4E" }, { 0x4F, "\x4F" }, { 0x50, "\x50" }, { 0x51, "\x51" }, { 0x52, "\x52" }, { 0x53, "\x53" }, { 0x54, "\x54" }, { 0x55, "\x55" }, { 0x56, "\x56" }, { 0x57, "\x57" }, { 0x58, "\x58" }, { 0x59, "\x59" }, { 0x5A, "\x5A" }, { 0x5B, "\x5B" }, { 0x5C, "\x5C" }, { 0x5D, "\x5D" }, { 0x5E, "\x5E" }, { 0x5F, "\x5F" }, { 0x60, "\x60" }, { 0x61, "\x61" }, { 0x62, "\x62" }, { 0x63, "\x63" }, { 0x64, "\x64" }, { 0x65, "\x65" }, { 0x66, "\x66" }, { 0x67, "\x67" }, { 0x68, "\x68" }, { 0x69, "\x69" }, { 0x6A, "\x6A" }, { 0x6B, "\x6B" }, { 0x6C, "\x6C" }, { 0x6D, "\x6D" }, { 0x6E, "\x6E" }, { 0x6F, "\x6F" }, { 0x70, "\x70" }, { 0x71, "\x71" }, { 0x72, "\x72" }, { 0x73, "\x73" }, { 0x74, "\x74" }, { 0x75, "\x75" }, { 0x76, "\x76" }, { 0x77, "\x77" }, { 0x78, "\x78" }, { 0x79, "\x79" }, { 0x7A, "\x7A" }, { 0x7B, "\x7B" }, { 0x7C, "\x7C" }, { 0x7D, "\x7D" }, { 0x7E, "\x7E" }, { 0xA1, "\xC2\xA1" }, { 0xA2, "\xC2\xA2" }, { 0xA3, "\xC2\xA3" }, { 0xA4, "\xC2\xA4" }, { 0xA5, "\xC2\xA5" }, { 0xA6, "\xC2\xA6" }, { 0xA7, "\xC2\xA7" }, { 0xA8, "\xC2\xA8" }, { 0xA9, "\xC2\xA9" }, { 0xAA, "\xC2\xAA" }, { 0xAB, "\xC2\xAB" }, { 0xAC, "\xC2\xAC" }, { 0xAE, "\xC2\xAE" }, { 0xAF, "\xC2\xAF" }, { 0xB0, "\xC2\xB0" }, { 0xB1, "\xC2\xB1" }, { 0xB2, "\xC2\xB2" }, { 0xB3, "\xC2\xB3" }, { 0xB4, "\xC2\xB4" }, { 0xB5, "\xC2\xB5" }, { 0xB6, "\xC2\xB6" }, { 0xB7, "\xC2\xB7" }, { 0xB8, "\xC2\xB8" }, { 0xB9, "\xC2\xB9" }, { 0xBA, "\xC2\xBA" }, { 0xBB, "\xC2\xBB" }, { 0xBC, "\xC2\xBC" }, { 0xBD, "\xC2\xBD" }, { 0xBE, "\xC2\xBE" }, { 0xBF, "\xC2\xBF" }, { 0xC0, "\xC3\x80" }, { 0xC1, "\xC3\x81" }, { 0xC2, "\xC3\x82" }, { 0xC3, "\xC3\x83" }, { 0xC4, "\xC3\x84" }, { 0xC5, "\xC3\x85" }, { 0xC6, "\xC3\x86" }, { 0xC7, "\xC3\x87" }, { 0xC8, "\xC3\x88" }, { 0xC9, "\xC3\x89" }, { 0xCA, "\xC3\x8A" }, { 0xCB, "\xC3\x8B" }, { 0xCC, "\xC3\x8C" }, { 0xCD, "\xC3\x8D" }, { 0xCE, "\xC3\x8E" }, { 0xCF, "\xC3\x8F" }, { 0xD0, "\xC3\x90" }, { 0xD1, "\xC3\x91" }, { 0xD2, "\xC3\x92" }, { 0xD3, "\xC3\x93" }, { 0xD4, "\xC3\x94" }, { 0xD5, "\xC3\x95" }, { 0xD6, "\xC3\x96" }, { 0xD7, "\xC3\x97" }, { 0xD8, "\xC3\x98" }, { 0xD9, "\xC3\x99" }, { 0xDA, "\xC3\x9A" }, { 0xDB, "\xC3\x9B" }, { 0xDC, "\xC3\x9C" }, { 0xDD, "\xC3\x9D" }, { 0xDE, "\xC3\x9E" }, { 0xDF, "\xC3\x9F" }, { 0xE0, "\xC3\xA0" }, { 0xE1, "\xC3\xA1" }, { 0xE2, "\xC3\xA2" }, { 0xE3, "\xC3\xA3" }, { 0xE4, "\xC3\xA4" }, { 0xE5, "\xC3\xA5" }, { 0xE6, "\xC3\xA6" }, { 0xE7, "\xC3\xA7" }, { 0xE8, "\xC3\xA8" }, { 0xE9, "\xC3\xA9" }, { 0xEA, "\xC3\xAA" }, { 0xEB, "\xC3\xAB" }, { 0xEC, "\xC3\xAC" }, { 0xED, "\xC3\xAD" }, { 0xEE, "\xC3\xAE" }, { 0xEF, "\xC3\xAF" }, { 0xF0, "\xC3\xB0" }, { 0xF1, "\xC3\xB1" }, { 0xF2, "\xC3\xB2" }, { 0xF3, "\xC3\xB3" }, { 0xF4, "\xC3\xB4" }, { 0xF5, "\xC3\xB5" }, { 0xF6, "\xC3\xB6" }, { 0xF7, "\xC3\xB7" }, { 0xF8, "\xC3\xB8" }, { 0xF9, "\xC3\xB9" }, { 0xFA, "\xC3\xBA" }, { 0xFB, "\xC3\xBB" }, { 0xFC, "\xC3\xBC" }, { 0xFD, "\xC3\xBD" }, { 0xFE, "\xC3\xBE" }, { 0xFF, "\xC3\xBF" }, { 0x00, "\xC4\x80" }, { 0x01, "\xC4\x81" }, { 0x02, "\xC4\x82" }, { 0x03, "\xC4\x83" }, { 0x04, "\xC4\x84" }, { 0x05, "\xC4\x85" }, { 0x06, "\xC4\x86" }, { 0x07, "\xC4\x87" }, { 0x08, "\xC4\x88" }, { 0x09, "\xC4\x89" }, { 0x0A, "\xC4\x8A" }, { 0x0B, "\xC4\x8B" }, { 0x0C, "\xC4\x8C" }, { 0x0D, "\xC4\x8D" }, { 0x0E, "\xC4\x8E" }, { 0x0F, "\xC4\x8F" }, { 0x10, "\xC4\x90" }, { 0x11, "\xC4\x91" }, { 0x12, "\xC4\x92" }, { 0x13, "\xC4\x93" }, { 0x14, "\xC4\x94" }, { 0x15, "\xC4\x95" }, { 0x16, "\xC4\x96" }, { 0x17, "\xC4\x97" }, { 0x18, "\xC4\x98" }, { 0x19, "\xC4\x99" }, { 0x1A, "\xC4\x9A" }, { 0x1B, "\xC4\x9B" }, { 0x1C, "\xC4\x9C" }, { 0x1D, "\xC4\x9D" }, { 0x1E, "\xC4\x9E" }, { 0x1F, "\xC4\x9F" }, { 0x20, "\xC4\xA0" }, { 0x7F, "\xC4\xA1" }, { 0x80, "\xC4\xA2" }, { 0x81, "\xC4\xA3" }, { 0x82, "\xC4\xA4" }, { 0x83, "\xC4\xA5" }, { 0x84, "\xC4\xA6" }, { 0x85, "\xC4\xA7" }, { 0x86, "\xC4\xA8" }, { 0x87, "\xC4\xA9" }, { 0x88, "\xC4\xAA" }, { 0x89, "\xC4\xAB" }, { 0x8A, "\xC4\xAC" }, { 0x8B, "\xC4\xAD" }, { 0x8C, "\xC4\xAE" }, { 0x8D, "\xC4\xAF" }, { 0x8E, "\xC4\xB0" }, { 0x8F, "\xC4\xB1" }, { 0x90, "\xC4\xB2" }, { 0x91, "\xC4\xB3" }, { 0x92, "\xC4\xB4" }, { 0x93, "\xC4\xB5" }, { 0x94, "\xC4\xB6" }, { 0x95, "\xC4\xB7" }, { 0x96, "\xC4\xB8" }, { 0x97, "\xC4\xB9" }, { 0x98, "\xC4\xBA" }, { 0x99, "\xC4\xBB" }, { 0x9A, "\xC4\xBC" }, { 0x9B, "\xC4\xBD" }, { 0x9C, "\xC4\xBE" }, { 0x9D, "\xC4\xBF" }, { 0x9E, "\xC5\x80" }, { 0x9F, "\xC5\x81" }, { 0xA0, "\xC5\x82" }, { 0xAD, "\xC5\x83" }};
+        return hex_map;
+    }
+
+    std::unordered_map<std::string, unsigned char> unicode_to_bytes() {
+        static std::unordered_map<std::string, unsigned char> hex_map = { { "\x21", 0x21 }, { "\x22", 0x22 }, { "\x23", 0x23 }, { "\x24", 0x24 }, { "\x25", 0x25 }, { "\x26", 0x26 }, { "\x27", 0x27 }, { "\x28", 0x28 }, { "\x29", 0x29 }, { "\x2A", 0x2A }, { "\x2B", 0x2B }, { "\x2C", 0x2C }, { "\x2D", 0x2D }, { "\x2E", 0x2E }, { "\x2F", 0x2F }, { "\x30", 0x30 }, { "\x31", 0x31 }, { "\x32", 0x32 }, { "\x33", 0x33 }, { "\x34", 0x34 }, { "\x35", 0x35 }, { "\x36", 0x36 }, { "\x37", 0x37 }, { "\x38", 0x38 }, { "\x39", 0x39 }, { "\x3A", 0x3A }, { "\x3B", 0x3B }, { "\x3C", 0x3C }, { "\x3D", 0x3D }, { "\x3E", 0x3E }, { "\x3F", 0x3F }, { "\x40", 0x40 }, { "\x41", 0x41 }, { "\x42", 0x42 }, { "\x43", 0x43 }, { "\x44", 0x44 }, { "\x45", 0x45 }, { "\x46", 0x46 }, { "\x47", 0x47 }, { "\x48", 0x48 }, { "\x49", 0x49 }, { "\x4A", 0x4A }, { "\x4B", 0x4B }, { "\x4C", 0x4C }, { "\x4D", 0x4D }, { "\x4E", 0x4E }, { "\x4F", 0x4F }, { "\x50", 0x50 }, { "\x51", 0x51 }, { "\x52", 0x52 }, { "\x53", 0x53 }, { "\x54", 0x54 }, { "\x55", 0x55 }, { "\x56", 0x56 }, { "\x57", 0x57 }, { "\x58", 0x58 }, { "\x59", 0x59 }, { "\x5A", 0x5A }, { "\x5B", 0x5B }, { "\x5C", 0x5C }, { "\x5D", 0x5D }, { "\x5E", 0x5E }, { "\x5F", 0x5F }, { "\x60", 0x60 }, { "\x61", 0x61 }, { "\x62", 0x62 }, { "\x63", 0x63 }, { "\x64", 0x64 }, { "\x65", 0x65 }, { "\x66", 0x66 }, { "\x67", 0x67 }, { "\x68", 0x68 }, { "\x69", 0x69 }, { "\x6A", 0x6A }, { "\x6B", 0x6B }, { "\x6C", 0x6C }, { "\x6D", 0x6D }, { "\x6E", 0x6E }, { "\x6F", 0x6F }, { "\x70", 0x70 }, { "\x71", 0x71 }, { "\x72", 0x72 }, { "\x73", 0x73 }, { "\x74", 0x74 }, { "\x75", 0x75 }, { "\x76", 0x76 }, { "\x77", 0x77 }, { "\x78", 0x78 }, { "\x79", 0x79 }, { "\x7A", 0x7A }, { "\x7B", 0x7B }, { "\x7C", 0x7C }, { "\x7D", 0x7D }, { "\x7E", 0x7E }, { "\xC2\xA1", 0xA1 }, { "\xC2\xA2", 0xA2 }, { "\xC2\xA3", 0xA3 }, { "\xC2\xA4", 0xA4 }, { "\xC2\xA5", 0xA5 }, { "\xC2\xA6", 0xA6 }, { "\xC2\xA7", 0xA7 }, { "\xC2\xA8", 0xA8 }, { "\xC2\xA9", 0xA9 }, { "\xC2\xAA", 0xAA }, { "\xC2\xAB", 0xAB }, { "\xC2\xAC", 0xAC }, { "\xC2\xAE", 0xAE }, { "\xC2\xAF", 0xAF }, { "\xC2\xB0", 0xB0 }, { "\xC2\xB1", 0xB1 }, { "\xC2\xB2", 0xB2 }, { "\xC2\xB3", 0xB3 }, { "\xC2\xB4", 0xB4 }, { "\xC2\xB5", 0xB5 }, { "\xC2\xB6", 0xB6 }, { "\xC2\xB7", 0xB7 }, { "\xC2\xB8", 0xB8 }, { "\xC2\xB9", 0xB9 }, { "\xC2\xBA", 0xBA }, { "\xC2\xBB", 0xBB }, { "\xC2\xBC", 0xBC }, { "\xC2\xBD", 0xBD }, { "\xC2\xBE", 0xBE }, { "\xC2\xBF", 0xBF }, { "\xC3\x80", 0xC0 }, { "\xC3\x81", 0xC1 }, { "\xC3\x82", 0xC2 }, { "\xC3\x83", 0xC3 }, { "\xC3\x84", 0xC4 }, { "\xC3\x85", 0xC5 }, { "\xC3\x86", 0xC6 }, { "\xC3\x87", 0xC7 }, { "\xC3\x88", 0xC8 }, { "\xC3\x89", 0xC9 }, { "\xC3\x8A", 0xCA }, { "\xC3\x8B", 0xCB }, { "\xC3\x8C", 0xCC }, { "\xC3\x8D", 0xCD }, { "\xC3\x8E", 0xCE }, { "\xC3\x8F", 0xCF }, { "\xC3\x90", 0xD0 }, { "\xC3\x91", 0xD1 }, { "\xC3\x92", 0xD2 }, { "\xC3\x93", 0xD3 }, { "\xC3\x94", 0xD4 }, { "\xC3\x95", 0xD5 }, { "\xC3\x96", 0xD6 }, { "\xC3\x97", 0xD7 }, { "\xC3\x98", 0xD8 }, { "\xC3\x99", 0xD9 }, { "\xC3\x9A", 0xDA }, { "\xC3\x9B", 0xDB }, { "\xC3\x9C", 0xDC }, { "\xC3\x9D", 0xDD }, { "\xC3\x9E", 0xDE }, { "\xC3\x9F", 0xDF }, { "\xC3\xA0", 0xE0 }, { "\xC3\xA1", 0xE1 }, { "\xC3\xA2", 0xE2 }, { "\xC3\xA3", 0xE3 }, { "\xC3\xA4", 0xE4 }, { "\xC3\xA5", 0xE5 }, { "\xC3\xA6", 0xE6 }, { "\xC3\xA7", 0xE7 }, { "\xC3\xA8", 0xE8 }, { "\xC3\xA9", 0xE9 }, { "\xC3\xAA", 0xEA }, { "\xC3\xAB", 0xEB }, { "\xC3\xAC", 0xEC }, { "\xC3\xAD", 0xED }, { "\xC3\xAE", 0xEE }, { "\xC3\xAF", 0xEF }, { "\xC3\xB0", 0xF0 }, { "\xC3\xB1", 0xF1 }, { "\xC3\xB2", 0xF2 }, { "\xC3\xB3", 0xF3 }, { "\xC3\xB4", 0xF4 }, { "\xC3\xB5", 0xF5 }, { "\xC3\xB6", 0xF6 }, { "\xC3\xB7", 0xF7 }, { "\xC3\xB8", 0xF8 }, { "\xC3\xB9", 0xF9 }, { "\xC3\xBA", 0xFA }, { "\xC3\xBB", 0xFB }, { "\xC3\xBC", 0xFC }, { "\xC3\xBD", 0xFD }, { "\xC3\xBE", 0xFE }, { "\xC3\xBF", 0xFF }, { "\xC4\x80", 0x00 }, { "\xC4\x81", 0x01 }, { "\xC4\x82", 0x02 }, { "\xC4\x83", 0x03 }, { "\xC4\x84", 0x04 }, { "\xC4\x85", 0x05 }, { "\xC4\x86", 0x06 }, { "\xC4\x87", 0x07 }, { "\xC4\x88", 0x08 }, { "\xC4\x89", 0x09 }, { "\xC4\x8A", 0x0A }, { "\xC4\x8B", 0x0B }, { "\xC4\x8C", 0x0C }, { "\xC4\x8D", 0x0D }, { "\xC4\x8E", 0x0E }, { "\xC4\x8F", 0x0F }, { "\xC4\x90", 0x10 }, { "\xC4\x91", 0x11 }, { "\xC4\x92", 0x12 }, { "\xC4\x93", 0x13 }, { "\xC4\x94", 0x14 }, { "\xC4\x95", 0x15 }, { "\xC4\x96", 0x16 }, { "\xC4\x97", 0x17 }, { "\xC4\x98", 0x18 }, { "\xC4\x99", 0x19 }, { "\xC4\x9A", 0x1A }, { "\xC4\x9B", 0x1B }, { "\xC4\x9C", 0x1C }, { "\xC4\x9D", 0x1D }, { "\xC4\x9E", 0x1E }, { "\xC4\x9F", 0x1F }, { "\xC4\xA0", 0x20 }, { "\xC4\xA1", 0x7F }, { "\xC4\xA2", 0x80 }, { "\xC4\xA3", 0x81 }, { "\xC4\xA4", 0x82 }, { "\xC4\xA5", 0x83 }, { "\xC4\xA6", 0x84 }, { "\xC4\xA7", 0x85 }, { "\xC4\xA8", 0x86 }, { "\xC4\xA9", 0x87 }, { "\xC4\xAA", 0x88 }, { "\xC4\xAB", 0x89 }, { "\xC4\xAC", 0x8A }, { "\xC4\xAD", 0x8B }, { "\xC4\xAE", 0x8C }, { "\xC4\xAF", 0x8D }, { "\xC4\xB0", 0x8E }, { "\xC4\xB1", 0x8F }, { "\xC4\xB2", 0x90 }, { "\xC4\xB3", 0x91 }, { "\xC4\xB4", 0x92 }, { "\xC4\xB5", 0x93 }, { "\xC4\xB6", 0x94 }, { "\xC4\xB7", 0x95 }, { "\xC4\xB8", 0x96 }, { "\xC4\xB9", 0x97 }, { "\xC4\xBA", 0x98 }, { "\xC4\xBB", 0x99 }, { "\xC4\xBC", 0x9A }, { "\xC4\xBD", 0x9B }, { "\xC4\xBE", 0x9C }, { "\xC4\xBF", 0x9D }, { "\xC5\x80", 0x9E }, { "\xC5\x81", 0x9F }, { "\xC5\x82", 0xA0 }, { "\xC5\x83", 0xAD }};
+        return hex_map;
+    }
+
+    // len must be available
+    bool inline str_is_equal(const char* str1, const char* str2, size_t len) {
+        for (size_t i = 0; i < len; ++i) {
+            if (str1[i] != str2[i]) {
+                return false;
+            }
+        }
+        return true;
+    }
+
+    std::vector<std::string> bpe_gpt2_preprocess(const std::string& text)
+    {
+        static std::unordered_map< unsigned char, std::string> byte_encoder = bytes_to_unicode();
+        std::vector<std::string> bpe_words;
+        std::vector<std::string> bpe_encoded_words;
+
+
+        std::string token="";
+        const char *raw_text_p = text.c_str();
+        // GPT2 system regex:  's|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+
+        bool collecting_numeric = false;
+        bool collecting_letter = false;
+        bool collecting_special = false;
+        bool collecting_whitespace_lookahead = false;
+        bool collecting=false;
+
+        std::vector<CNCTString> text_utf;
+        text_utf.reserve(text.size());
+        bpe_words.reserve(text.size());
+        bpe_encoded_words.reserve(text.size());
+
+        text_utf = CNCTUnicode::split_utf8_enhanced(text);
+        std::map<std::string, int> special_tokens = vocab_.special_tokens;
+        int smallest_len_special_tokens = 0;
+        if (special_tokens.size())
+        {
+            smallest_len_special_tokens = special_tokens.begin()->first.size();
+            for (auto it = special_tokens.begin(); it != special_tokens.end(); ++it)
+            {
+                if (it->first.size() < (size_t)smallest_len_special_tokens)
+                    smallest_len_special_tokens = it->first.size();
+            }
+        }
+
+        for (int i = 0; i < (int)text_utf.size(); i++)
+        {
+            const CNCTString &utf_char = text_utf[i];
+            bool split_condition = false;
+            const char *text_pos = raw_text_p + utf_char.seq_offset_bytes;
+            int bytes_remain = strlen(text_pos);
+            // forward backward lookups
+            const CNCTString &utf_char_next = (i+1 < (int)text_utf.size()) ? text_utf[i+1] : CNCTString();
+            const CNCTString &utf_char_next_next = (i+2 < (int)text_utf.size()) ? text_utf[i+2] : CNCTString();
+            // const CNCTString &utf_char_prev = (i > 0) ? text_utf[i-1] : CNCTString();
+
+        // handling special tokens
+            bool special_token_found = false;
+            if (bytes_remain >= (int)smallest_len_special_tokens)
+            for (auto it = special_tokens.begin(); it != special_tokens.end(); ++it)
+            {
+                if ((bytes_remain) < (int)it->first.size())
+                    continue;
+
+                if (str_is_equal(text_pos, it->first.c_str(), it->first.size()))
+                {
+                    if (token.size())
+                    {
+                        bpe_words.emplace_back(token); // push previous content as token
+                        token.clear();
+                        collecting = false;
+                        collecting_letter = false;
+                        collecting_numeric = false;
+                        collecting_special = false;
+                        collecting_whitespace_lookahead = false;
+                    }
+
+                    bpe_words.emplace_back(it->first); // push special token as token
+
+                    // we now advance i until the token is fulfilled by the utf_chars
+                    int st_bytes = (int)it->first.size();
+                    for (;st_bytes;st_bytes -= text_utf[i++].str.size());
+                    i--;
+                    special_token_found = true;
+                    break;
+                }
+            }
+
+            if (special_token_found)   continue;
+
+
+            // handling contractions
+            if (!split_condition && bytes_remain >= 2)
+            {
+                // 's|'t|'m|'d
+                if (utf_char == '\'' && (utf_char_next == 's' || utf_char_next == 't' || utf_char_next == 'm' || utf_char_next == 'd'))
+                    split_condition = true;
+                if (split_condition)
+                {
+                    if (token.size())
+                        bpe_words.emplace_back(token); // push previous content as token
+                    token = utf_char.str + utf_char_next.str;
+                    bpe_words.emplace_back(token);
+                    token="";
+                    i++;
+                    continue;
+                }
+            }
+            if (!split_condition && bytes_remain >= 3)
+            {
+                // 're|'ve|'ll
+                if (utf_char == '\'' && (
+                    (utf_char_next == 'r' || utf_char_next_next == 'e') ||
+                    (utf_char_next == 'v' || utf_char_next_next == 'e') ||
+                    (utf_char_next == 'l' || utf_char_next_next == 'l'))
+                )
+                    split_condition = true;
+                if (split_condition)
+                {
+                    // current token + next token can be defined
+                    if (token.size())
+                        bpe_words.emplace_back(token); // push previous content as token
+                    token = utf_char.str + utf_char_next.str + utf_char_next_next.str;
+                    bpe_words.emplace_back(token); // the contraction
+                    token="";
+                    i+=2;
+                    continue;
+                }
+            }
+
+            if (!split_condition && !collecting)
+            {
+                if (utf_char.char_type == CNCTCharType::LETTER || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::LETTER))
+                {
+                    collecting_letter = true;
+                    collecting = true;
+                }
+                else if (utf_char.char_type == CNCTCharType::DIGIT || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::DIGIT))
+                {
+                    collecting_numeric = true;
+                    collecting = true;
+                }
+                else if (
+                    ((utf_char.char_type != CNCTCharType::LETTER && utf_char.char_type != CNCTCharType::DIGIT) && (utf_char.char_type != CNCTCharType::WHITESPACE)) ||
+                    (!token.size() && utf_char==" " && utf_char_next.char_type != CNCTCharType::LETTER && utf_char_next.char_type != CNCTCharType::DIGIT && utf_char_next.char_type != CNCTCharType::WHITESPACE)
+                    )
+                {
+                    collecting_special = true;
+                    collecting = true;
+                }
+                else if (utf_char.char_type == CNCTCharType::WHITESPACE && utf_char_next.char_type == CNCTCharType::WHITESPACE)
+                {
+                    collecting_whitespace_lookahead = true;
+                    collecting = true;
+                } else if (utf_char.char_type == CNCTCharType::WHITESPACE)
+                {
+                    split_condition = true;
+                }
+            } else
+            if (!split_condition && collecting)
+            {
+                if (collecting_letter && utf_char.char_type != CNCTCharType::LETTER)
+                {
+                    split_condition = true;
+                }
+                else if (collecting_numeric && utf_char.char_type != CNCTCharType::DIGIT)
+                {
+                    split_condition = true;
+                }
+                else if (collecting_special && (utf_char.char_type == CNCTCharType::LETTER || utf_char.char_type == CNCTCharType::DIGIT || utf_char.char_type == CNCTCharType::WHITESPACE))
+                {
+                    split_condition = true;
+                }
+                else if (collecting_whitespace_lookahead && utf_char_next.char_type != CNCTCharType::WHITESPACE)
+                {
+                    split_condition = true;
+                }
+            }
+            if(utf_char_next.str.size() == 0)
+            {
+                split_condition = true; // final
+                token += utf_char.str;
+            }
+
+            if (split_condition)
+            {
+                if (token.size())
+                    bpe_words.emplace_back(token);
+                token = utf_char.str;
+                collecting = false;
+                collecting_letter = false;
+                collecting_numeric = false;
+                collecting_special = false;
+                collecting_whitespace_lookahead = false;
+            } else token += utf_char.str;
+        }
+
+        for (std::string& word : bpe_words)
+        {
+            std::string encoded_token="";
+            for (char& c : word)
+            {
+                encoded_token += byte_encoder[c];
+            }
+            bpe_encoded_words.emplace_back(encoded_token);
+        }
+
+        return bpe_encoded_words;
+    }
+
+    // decoder (for one token)
+    std::string decode_token(const std::string& token)
+    {
+        static std::unordered_map< std::string, unsigned char> byte_decoder = unicode_to_bytes();
+        std::string decoded_token="";
+        auto unicode_seqeunces = CNCTUnicode::split_utf8(token);
+        for (auto& unicode_sequence : unicode_seqeunces)
+        {
+            decoded_token += byte_decoder[unicode_sequence];
+
+        }
+
+        return decoded_token;
+    }
+
+
+    const gpt2bpe_vocab & vocab_;
+    std::vector<ggllm_bpe_symbol> symbols_;
+    std::vector<ggllm_bpe_symbol> symbols_final;
+    ggllm_bpe_bigram::queue work_queue_;
+    bool flag_g2ws=false;
+};
+
+static std::vector<gpt2bpe_vocab::id> gpt2bpe_tokenize(const gpt2bpe_vocab & vocab, const std::string & text, bool bos, bool g2ws ) {
+    gpt2bpe_tokenizer tokenizer(vocab, g2ws);
+    std::vector<gpt2bpe_vocab::id> output;
+
+    if (text.empty()) {
+        return output;
+    }
+
+    if (bos && vocab.special_have_bos) {
+        output.push_back(vocab.special_bos_id);
+    }
+
+    tokenizer.tokenize(text, output);
+    return output;
+}
+
+
+#endif // CMPNCT_GPT2BPE

From fb0b24370563824203f7619615a9d57a33dee57f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 04:02:10 +0200
Subject: [PATCH 075/242] Makefile : remove gptneox-common

---
 Makefile | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/Makefile b/Makefile
index 7d37b66a698ec..90563d7813e67 100644
--- a/Makefile
+++ b/Makefile
@@ -329,9 +329,6 @@ grammar-parser.o: examples/grammar-parser.cpp examples/grammar-parser.h
 libllama.so: llama.o ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) -shared -fPIC -o $@ $^ $(LDFLAGS)
 
-gptneox-common.o: gptneox-common.cpp gptneox-common.h
-	$(CXX) $(CXXFLAGS) -c $< -o $@
-
 clean:
 	rm -vf *.o *.so *.dll main quantize quantize-stats perplexity embedding benchmark-matmult save-load-state server simple vdot train-text-from-scratch embd-input-test gguf build-info.h $(TEST_TARGETS)
 
@@ -376,7 +373,7 @@ embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-te
 gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
-gptneox-main: gptneox-main.cpp gptneox-common.o ggml.o $(OBJS)
+gptneox-main: gptneox-main.cpp ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
 train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o $(OBJS)

From 278ada9572c645a8353edce256453a755ed1743e Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 04:07:57 +0200
Subject: [PATCH 076/242] gguf.py : bytesarray for gpt2bpe tokenizer

---
 gguf.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/gguf.py b/gguf.py
index 88e2bee0733b1..1e4f8ea3b235f 100644
--- a/gguf.py
+++ b/gguf.py
@@ -10,7 +10,7 @@
 from typing import Any, IO, List
 
 import numpy as np
-
+import sys
 
 class GGMLQuantizationType(IntEnum):
     F32 = 0
@@ -45,7 +45,7 @@ class GGUFValueType(IntEnum):
 
     @staticmethod
     def get_type(val):
-        if isinstance(val, str) or isinstance(val, bytes):
+        if isinstance(val, str) or isinstance(val, bytes) or isinstance(val, bytearray):
             return GGUFValueType.STRING
         elif isinstance(val, list):
             return GGUFValueType.ARRAY
@@ -53,8 +53,11 @@ def get_type(val):
             return GGUFValueType.FLOAT32
         elif isinstance(val, bool):
             return GGUFValueType.BOOL
-        else:
+        elif isinstance(val, int):
             return GGUFValueType.INT32
+        else:
+            print("Unknown type: "+str(type(val)))
+            sys.exit()
 
 
 class GGUFWriter:

From db5618ad9919ca4e4fb29b6fa6509923368f4755 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 4 Aug 2023 04:57:51 +0200
Subject: [PATCH 077/242] cmpnct_gpt2bpe.hpp : comments

---
 cmpnct_gpt2bpe.hpp | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/cmpnct_gpt2bpe.hpp b/cmpnct_gpt2bpe.hpp
index 0743a294ac0ff..ac5f8c672a305 100644
--- a/cmpnct_gpt2bpe.hpp
+++ b/cmpnct_gpt2bpe.hpp
@@ -11,12 +11,15 @@
 #include <queue>
 #include <cstring>
 
+//-----
+// Unicode GPT2 Byte Pair Encoding Tokenizer
+// Adapted from https://github.com/cmp-nct/ggllm.cpp
+//-----
 
-/**
- * https://github.com/cmp-nct/ggllm.cpp
- * Minimal library for high performance handling and categorization of UTF8 strings and characters
- * Using std::string
- */
+// Unicode library (from cmpnct_unicode.cpp)
+
+// Minimal library for high performance handling and categorization of UTF8 strings and characters
+// Using std::string
 
 enum CNCTCharType {
     DIGIT,          // a numerical char in any language
@@ -367,7 +370,7 @@ bool CNCTUnicode::string_test(const std::string &str, CNCTCharType chartype)
     return true;
 }
 
-// Ported from libfalcon.cpp (https://github.com/cmp-nct/ggllm.cpp)
+// llama.cpp GPT2 vocab (from libfalcon.cpp)
 
 std::string replaceAll(std::string str, const std::string& from, const std::string& to) {
     size_t start_pos = 0;

From 4357e692ace6a5afa6c9a1e2a597795b838e999f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 7 Aug 2023 13:51:26 +0200
Subject: [PATCH 078/242] gguf.py : use custom alignment if present

---
 gguf.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/gguf.py b/gguf.py
index 1e4f8ea3b235f..5eb21ee05fb6f 100644
--- a/gguf.py
+++ b/gguf.py
@@ -64,6 +64,7 @@ class GGUFWriter:
     def __init__(self, fout: IO):
         self.fout = fout
         self.offset_tensor = 0
+        self.data_alignment = constants.GGUF_DEFAULT_ALIGNMENT
         self.kv_data = b""
         self.kv_data_count = 0
         self.ti_data = b""
@@ -191,16 +192,17 @@ def add_tensor_info(self, name: str, tensor: np.ndarray):
         dtype = GGMLQuantizationType.F32 if tensor.dtype == np.float32 else GGMLQuantizationType.F16
         self.ti_data += struct.pack("<I", dtype)
         self.ti_data += struct.pack("<Q", self.offset_tensor)
-        self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT)
+        self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment)
         self.ti_data_count += 1
 
     def write_tensor_to_file(self, tensor: np.ndarray):
-        pad = GGUFWriter.ggml_pad(self.fout.tell(), constants.GGUF_DEFAULT_ALIGNMENT) - self.fout.tell()
+        pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
         tensor.tofile(self.fout)
-        pad = GGUFWriter.ggml_pad(tensor.nbytes, constants.GGUF_DEFAULT_ALIGNMENT) - tensor.nbytes
+        
+        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
@@ -240,6 +242,7 @@ def add_quantization_version(self, quantization_version: GGMLQuantizationType):
             constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
 
     def add_custom_alignment(self, alignment: int):
+        self.data_alignment = alignment
         self.add_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
 
     def add_context_length(self, llm: str, length: int):

From 8083ae347a1c2aef99435716f9106ad31ea78fa5 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 7 Aug 2023 19:02:18 +0300
Subject: [PATCH 079/242] gguf : minor stuff

---
 examples/gguf/gguf.cpp | 11 +++++------
 ggml.c                 |  4 ----
 2 files changed, 5 insertions(+), 10 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 671f8748f5e13..a1b8edc71f373 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -256,10 +256,9 @@ bool gguf_ex_read_0(const std::string & fname) {
 
     // find kv string
     {
-        char findkey[32];
-        sprintf(findkey, "some.parameter.string");
+        const char * findkey = "some.parameter.string";
 
-        int keyidx = gguf_find_key(ctx, findkey);
+        const int keyidx = gguf_find_key(ctx, findkey);
         if (keyidx == -1) {
             fprintf(stdout, "%s: find key: %s not found.\n", __func__, findkey);
         } else {
@@ -297,7 +296,7 @@ bool gguf_ex_read_1(const std::string & fname) {
     };
 
     struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
-    
+
     fprintf(stdout, "%s: version:      %d\n", __func__, gguf_get_version(ctx));
     fprintf(stdout, "%s: alignment:   %zu\n", __func__, gguf_get_alignment(ctx));
     fprintf(stdout, "%s: data offset: %zu\n", __func__, gguf_get_data_offset(ctx));
@@ -388,7 +387,7 @@ bool gguf_ex_read_2(const std::string & fname) {
 
         // print first 10 elements
     const float * data = (const float *) cur->data;
-                
+
         printf("%s data[:10] : ", name);
 
         for (int j = 0; j < 10; ++j) {
@@ -402,7 +401,7 @@ fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data
 
     ggml_free(ctx_data);
     gguf_free(ctx);
-    
+
     return true;
 }
 
diff --git a/ggml.c b/ggml.c
index d3d745333434b..df8dce7e89915 100644
--- a/ggml.c
+++ b/ggml.c
@@ -19031,10 +19031,6 @@ const char * gguf_get_key(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].key.data;
 }
 
-enum gguf_type gguf_get_type(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].type;
-}
-
 const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
     struct gguf_kv * kv = &ctx->header.kv[key_id];
     struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];

From 65559a23c8df919bae0ffc82c48c94759b2d53a1 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 7 Aug 2023 22:28:43 +0200
Subject: [PATCH 080/242] Update gptneox-main.cpp

---
 gptneox-main.cpp | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 0161d9c2ee7e8..eecd59678a3eb 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -443,9 +443,6 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
     // load vocab
     {
-
-        // TODO: implement a better bpe tokenizer, utilizing merges and handles unicode
-
         auto & hparams = model.hparams;
 
         int keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.model");
@@ -484,11 +481,6 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         for (size_t i = 0; i < hparams.n_vocab; i++) {
             std::string word = gguf_get_arr_str(ggufctx, tokens_keyidx, i);
 
-
-            // TEMP until a better bpe tokenizer is implemented
-//            word = replace(word, "Ġ", " ");
-//            word = replace(word, "Ċ", "\n");
-
 //            printf("token %d = '%s'\n",i,word.c_str() );
 
             vocab.token_to_id[word] = i;
@@ -1054,7 +1046,6 @@ int main(int argc, char ** argv) {
             {
                 const int64_t t_start_sample_us = ggml_time_us();
 
-//                id = sample_top_k_top_p(vocab, logits.data() + (logits.size() - n_vocab), top_k, top_p, temp, repeat_last_n, repeat_penalty, rng);
                 id = sample_top_k_top_p_repeat(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens.data(), last_n_tokens.size(), top_k, top_p, temp, repeat_last_n, repeat_penalty, rng);
 
                 last_n_tokens.erase(last_n_tokens.begin());

From ece4fc185edf8677fe158a52b52002258b32833b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 9 Aug 2023 00:48:33 +0200
Subject: [PATCH 081/242] map tensor names

---
 gguf_tensor_map.py | 96 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)
 create mode 100644 gguf_tensor_map.py

diff --git a/gguf_tensor_map.py b/gguf_tensor_map.py
new file mode 100644
index 0000000000000..4fba633b2bf12
--- /dev/null
+++ b/gguf_tensor_map.py
@@ -0,0 +1,96 @@
+# Recommended mapping of model tensor names for storage in gguf
+
+def get_tensor_map( n_blocks : int):
+    tensor_map = {}
+    # Token embeddings
+    mapped_to = "transformer.token_embd"
+    tensor_map["gpt_neox.embed_in"] = mapped_to           # gptneox
+    tensor_map["transformer.wte"] = mapped_to             # gpt2 mpt
+    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
+    tensor_map["model.embed_tokens"] = mapped_to          # llama-hf
+    tensor_map["tok_embeddings"] = mapped_to              # llama-pth
+    # Position embeddings
+    mapped_to = "transformer.pos_embd"
+    tensor_map["transformer.wpe"] = mapped_to # gpt2
+    # Output norm
+    mapped_to = "transformer.output_norm"
+    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
+    tensor_map["transformer.ln_f"] = mapped_to          # gpt2 falcon
+    tensor_map["transformer.norm_f"] = mapped_to        # mpt
+    tensor_map["model.norm"] = mapped_to                # llama-hf
+    tensor_map["norm"] = mapped_to                      # llama-pth
+    # Output
+    mapped_to = "transformer.output"
+    tensor_map["embed_out"] = mapped_to # gptneox
+    tensor_map["lm_head"] = mapped_to   # gpt2 mpt falcon llama-hf
+    tensor_map["output"] = mapped_to    # llama-pth
+    # Attention and fee-forward layer blocks
+    for i in range(0,n_blocks):
+        # Attention norm 1
+        mapped_to = "transformer.blocks."+str(i)+".attn_norm_1"
+        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to             # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to      # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to  # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to          # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to   # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to          # llama-pth
+        # Attention norm 2
+        mapped_to = "transformer.blocks."+str(i)+".attn_norm_2"
+        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
+        # Attention query-key-value
+        mapped_to = "transformer.blocks."+str(i)+".attn_qkv"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"] = mapped_to     # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_attn"] = mapped_to                     # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"] = mapped_to                  # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to  # falcon
+        # Attention query
+        mapped_to = "transformer.blocks."+str(i)+".attn_q"
+        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wq"] = mapped_to           # llama-pth
+        # Attention key
+        mapped_to = "transformer.blocks."+str(i)+".attn_k"
+        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wk"] = mapped_to           # llama-pth
+        # Attention value
+        mapped_to = "transformer.blocks."+str(i)+".attn_v"
+        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wv"] = mapped_to           # llama-pth
+        # Attention output
+        mapped_to = "transformer.blocks."+str(i)+".attn_output"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"] = mapped_to    # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_proj"] = mapped_to          # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"] = mapped_to   # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".self_attn.o_proj"] = mapped_to      # llama-hf
+        tensor_map["layers."+str(i)+".attention.wo"] = mapped_to                # llama-pth
+        # Feed-forward norm
+        mapped_to = "transformer.blocks."+str(i)+".ffn_norm"
+        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to                       # gpt2
+        tensor_map[" transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
+        tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to    # llama-hf
+        tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to                          # llama-pth
+        # Feed-forward up
+        mapped_to = "transformer.blocks."+str(i)+".ffn_up"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_fc"] = mapped_to            # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"] = mapped_to    # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"] = mapped_to   # falcon
+        tensor_map["model.layers."+str(i)+".mlp.up_proj"] = mapped_to          # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w3"] = mapped_to            # llama-pth
+        # Feed-forward gate
+        mapped_to = "transformer.blocks."+str(i)+".ffn_gate"
+        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w1"] = mapped_to     # llama-pth
+        # Feed-forward down
+        mapped_to = "transformer.blocks."+str(i)+".ffn_down"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_proj"] = mapped_to          # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"] = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"] = mapped_to   # falcon
+        tensor_map["model.layers."+str(i)+".mlp.down_proj"] = mapped_to        # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w2"] = mapped_to            # llama-pth
+
+    return tensor_map
+

From f4d137d98cf770d205cd788947cb990012593bd8 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 9 Aug 2023 00:50:11 +0200
Subject: [PATCH 082/242] convert-gptneox-h5-to-gguf.py : map tensor names

---
 convert-gptneox-h5-to-gguf.py | 36 ++++++++++++++++++++++++-----------
 1 file changed, 25 insertions(+), 11 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 066ed0da07697..22508bd3dae56 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -1,6 +1,7 @@
 # Quick and dirty HF gptneox--> gguf conversion
 
 import gguf
+import gguf_tensor_map as tmap
 import os
 import sys
 import struct
@@ -32,6 +33,7 @@ def bytes_to_unicode():
     cs = [chr(n) for n in cs]
     return dict(zip(bs, cs))
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
@@ -74,16 +76,17 @@ def bytes_to_unicode():
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
-print("gguf: add metadata")
+print("gguf: get model metadata")
 
-llm_arch = "gptneox"
+llm_arch    = "gptneox"
+block_count = hparams["num_hidden_layers"]
 
 gguf_writer.add_name(last_dir)
 gguf_writer.add_description("gguf test model")
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.add_layer_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.add_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
 gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
@@ -92,7 +95,7 @@ def bytes_to_unicode():
 
 # TOKENIZATION
 
-print("gguf: add tokenizer")
+print("gguf: get tokenizer metadata")
 
 tokens: List[str] = []
 merges: List[str] = []
@@ -102,7 +105,7 @@ def bytes_to_unicode():
     # gpt2 tokenizer
     gguf_writer.add_tokenizer_model("gpt2")
 
-    print("gguf: adding gpt2 tokenizer merges")
+    print("gguf: get gpt2 tokenizer merges")
 
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
         tokenizer_json = json.load(f)
@@ -110,7 +113,7 @@ def bytes_to_unicode():
 
     gguf_writer.add_token_merges(merges)
 
-    print("gguf: adding gpt2 tokenizer vocab")
+    print("gguf: get gpt2 tokenizer vocab")
 
     vocab_size = len( tokenizer_json["model"]["vocab"] )
 
@@ -141,7 +144,7 @@ def bytes_to_unicode():
     gguf_writer.add_token_list(tokens)
 
     if "added_tokens" in tokenizer_json and Path(dir_model + "/tokenizer_config.json").is_file():
-        print("gguf: adding special token ids")
+        print("gguf: get special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
             tokenizer_config = json.load(f)
@@ -176,8 +179,10 @@ def bytes_to_unicode():
 
 # TENSORS
 
+tensor_map = tmap.get_tensor_map(block_count)
+
 # tensor info
-print("gguf: add gguf tensor info")
+print("gguf: get tensor metadata")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -186,6 +191,15 @@ def bytes_to_unicode():
     if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
         continue
 
+    # map tensor names
+    if name.endswith(".weight") and name[:-7] in tensor_map:
+        name = tensor_map[name[:-7]] + ".weight"
+    elif name.endswith(".bias") and name[:-5] in tensor_map:
+        name = tensor_map[name[:-5]] + ".bias"
+    else:
+        print( "Can not map tensor '" + name + "'" )
+        sys.exit()
+
     n_dims = len(data.shape)
 
     # ftype == 0 -> float32, ftype == 1 -> float16
@@ -206,9 +220,9 @@ def bytes_to_unicode():
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
-print("gguf: write key-values")
+print("gguf: write metadata")
 gguf_writer.write_kv_data_to_file()
-print("gguf: write tensor info")
+print("gguf: write tensor metadata")
 gguf_writer.write_ti_data_to_file()
 
 # tensor data
@@ -242,5 +256,5 @@ def bytes_to_unicode():
 gguf_writer.close()
 
 
-print("gguf: conversion done, output file: " + fname_out)
+print("gguf: model successfully exported to '" + fname_out + "'" )
 print("")

From 7d5f4522ddc2717e878966bf357dee790c33e860 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 9 Aug 2023 00:52:16 +0200
Subject: [PATCH 083/242] convert-llama-h5-to-gguf.py : map tensor names

---
 convert-llama-h5-to-gguf.py | 85 +++++++++++++------------------------
 1 file changed, 29 insertions(+), 56 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 67b3c55d4e35f..0b477a1336570 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -1,6 +1,8 @@
 # Quick and dirty HF llama --> gguf conversion, GQA/70b wont work
 
 import gguf
+import gguf_tensor_map as tmap
+import os
 import sys
 import struct
 import json
@@ -12,7 +14,6 @@
 
 
 #NDArray = np.ndarray[Any, Any]
-
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
@@ -32,6 +33,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 # output in the same directory as the model
 dir_model = sys.argv[1]
 fname_out = sys.argv[1] + "/ggml-model.bin"
+last_dir = os.path.basename(os.path.normpath(dir_model))
 
 
 # possible tensor data types
@@ -48,6 +50,8 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         print("Invalid ftype: " + str(ftype))
         sys.exit(1)
     fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+print("gguf: loading model "+last_dir)
     
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
     hparams = json.load(f)
@@ -62,32 +66,34 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
 
-print("gguf: add key-values, metadata")
+print("gguf: get model metadata")
 
-llm_arch = "llama"
+llm_arch    = "llama"
+head_count  = hparams["num_attention_heads"]
+block_count = hparams["num_hidden_layers"]
 
 gguf_writer.add_name("llama2-7b")
 gguf_writer.add_description("gguf test model")
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_layer_count(llm_arch, hparams["num_hidden_layers"])
+gguf_writer.add_layer_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
-gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
+gguf_writer.add_head_count(llm_arch, head_count)
 gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
 
 
 # TOKENIZATION
 
-print("gguf: add key-values, tokenizer")
+print("gguf: get tokenizer metadata")
 
 tokens: List[str] = []
 scores: List[float] = []
 
 if Path(dir_model + "/tokenizer.model").is_file():
     # vocab type sentencepiece
-    print("gguf: adding sentencepiece tokenizer vocab")
+    print("gguf: get sentencepiece tokenizer vocab and scores")
 
     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
@@ -119,7 +125,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         tokenizer = json.load(f)
 
     if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
-        print("gguf: adding special token ids")
+        print("gguf: get special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
             tokenizer_config = json.load(f)
@@ -154,8 +160,10 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 # TENSORS
 
+tensor_map = tmap.get_tensor_map(block_count)
+
 # tensor info
-print("gguf: add gguf tensor info")
+print("gguf: get tensor metadata")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -166,45 +174,16 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
     # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute(data, hparams["num_attention_heads"])
-
-    # chnage tensor name
+        data = permute(data,head_count)
 
-    if name == "model.embed_tokens.weight":
-        name = "tok_embeddings.weight"
-    elif name == "model.norm.weight":
-        name = "norm.weight"
-    elif name == "lm_head.weight":
-        name = "output.weight"
+    # map tensor names
+    if name.endswith(".weight") and name[:-7] in tensor_map:
+        name = tensor_map[name[:-7]] + ".weight"
+    elif name.endswith(".bias") and name[:-5] in tensor_map:
+        name = tensor_map[name[:-5]] + ".bias"
     else:
-        for i in range(80):  # maximum number of layers
-            if name == "model.layers." + str(i) + ".input_layernorm.weight":
-                name = "layers." + str(i) + ".attention_norm.weight"
-                break
-            if name == "model.layers." + str(i) + ".self_attn.q_proj.weight":
-                name = "layers." + str(i) + ".attention.wq.weight"
-                break
-            if name == "model.layers." + str(i) + ".self_attn.k_proj.weight":
-                name = "layers." + str(i) + ".attention.wk.weight"
-                break
-            if name == "model.layers." + str(i) + ".self_attn.v_proj.weight":
-                name = "layers." + str(i) + ".attention.wv.weight"
-                break
-            if name == "model.layers." + str(i) + ".self_attn.o_proj.weight":
-                name = "layers." + str(i) + ".attention.wo.weight"
-                break
-            if name == "model.layers." + str(i) + ".post_attention_layernorm.weight":
-                name = "layers." + str(i) + ".ffn_norm.weight"
-                break
-            if name == "model.layers." + str(i) + ".mlp.gate_proj.weight":
-                name = "layers." + str(i) + ".feed_forward.w1.weight"
-                break
-            if name == "model.layers." + str(i) + ".mlp.down_proj.weight":
-                name = "layers." + str(i) + ".feed_forward.w2.weight"
-                break
-            if name == "model.layers." + str(i) + ".mlp.up_proj.weight":
-                name = "layers." + str(i) + ".feed_forward.w3.weight"
-                break
+        print( "Can not map tensor '" + name + "'" )
+        sys.exit()
 
     n_dims = len(data.shape)
 
@@ -227,9 +206,9 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
-print("gguf: write key-values")
+print("gguf: write metadata")
 gguf_writer.write_kv_data_to_file()
-print("gguf: write tensor info")
+print("gguf: write tensor metadata")
 gguf_writer.write_ti_data_to_file()
 
 # tensor data
@@ -237,17 +216,14 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
-#    print("Process tensor: " + name + " with shape: ", data.shape)
 
     # we don't need these
     if name.endswith(".rotary_emb.inv_freq"):
-#        print("  Skip tensor: " + name)
         continue
 
     # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-#        print("  Permute tensor: " + name)
-        data = permute(data, hparams["num_attention_heads"])
+        data = permute(data, head_count)
 
     n_dims = len(data.shape)
 
@@ -255,16 +231,13 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     ftype_cur = 0
     if ftype != 0:
         if name.endswith(".weight") and n_dims == 2:
-#            print("  Converting to float16")
             data = data.astype(np.float16)
             ftype_cur = 1
         else:
-#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
     else:
         if data.dtype != np.float32:
-#            print("  Converting to float32")
             data = data.astype(np.float32)
             ftype_cur = 0
 
@@ -273,5 +246,5 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.close()
 
 
-print("gguf: conversion done, output file: " + fname_out)
+print("gguf: model successfully exported to '" + fname_out + "'" )
 print("")

From 0246d0dd6f8f57b64b696a33362ecf287d33ef63 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 9 Aug 2023 00:54:21 +0200
Subject: [PATCH 084/242] gptneox-main.cpp : map tensor names

---
 gptneox-main.cpp | 66 +++++++++++++++++++++++++-----------------------
 1 file changed, 34 insertions(+), 32 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index eecd59678a3eb..1667c4d545114 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -549,56 +549,58 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
         model.layers.resize(n_layer);
 
-        model.wte    = ggml_get_tensor(ctx, "gpt_neox.embed_in.weight");
-        model.ln_f_g = ggml_get_tensor(ctx, "gpt_neox.final_layer_norm.weight");
-        model.ln_f_b = ggml_get_tensor(ctx, "gpt_neox.final_layer_norm.bias");
-        model.lmh_g  = ggml_get_tensor(ctx, "embed_out.weight");
+        model.wte    = ggml_get_tensor(ctx, "transformer.token_embd.weight");
+        model.ln_f_g = ggml_get_tensor(ctx, "transformer.output_norm.weight");
+        model.ln_f_b = ggml_get_tensor(ctx, "transformer.output_norm.bias");
+        model.lmh_g  = ggml_get_tensor(ctx, "transformer.output.weight");
 
         // map by name
-        model.tensors["gpt_neox.embed_in.weight"] = model.wte;
-        model.tensors["gpt_neox.final_layer_norm.weight"] = model.ln_f_g;
-        model.tensors["gpt_neox.final_layer_norm.bias"]   = model.ln_f_b;
-        model.tensors["embed_out.weight"] = model.lmh_g;
+        model.tensors["transformer.token_embd.weight"] = model.wte;
+        model.tensors["transformer.output_norm.weight"] = model.ln_f_g;
+        model.tensors["transformer.output_norm.bias"]   = model.ln_f_b;
+        model.tensors["transformer.output.weight"] = model.lmh_g;
 
         for (int i = 0; i < n_layer; ++i) {
             auto & layer = model.layers[i];
 
-            layer.ln_1_g          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".input_layernorm.weight" );
-            layer.ln_1_b          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".input_layernorm.bias" );
+            std::string blocknamestart = "transformer.blocks." + std::to_string(i) + ".";
 
-            layer.c_attn_attn_w   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.weight" );
-            layer.c_attn_attn_b   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.bias" );
+            layer.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm_1.weight" );
+            layer.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm_1.bias" );
 
-            layer.c_attn_proj_w   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.dense.weight" );
-            layer.c_attn_proj_b   = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".attention.dense.bias" );
+            layer.c_attn_attn_w   = get_tensor_ex(ctx, blocknamestart + "attn_qkv.weight" );
+            layer.c_attn_attn_b   = get_tensor_ex(ctx ,blocknamestart + "attn_qkv.bias" );
 
-            layer.ln_2_g          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.weight" );
-            layer.ln_2_b          = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.bias");
+            layer.c_attn_proj_w   = get_tensor_ex(ctx, blocknamestart + "attn_output.weight" );
+            layer.c_attn_proj_b   = get_tensor_ex(ctx, blocknamestart + "attn_output.bias" );
 
-            layer.c_mlp_fc_w      = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.weight" );
-            layer.c_mlp_fc_b      = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.bias" );
+            layer.ln_2_g          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.weight" );
+            layer.ln_2_b          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.bias");
 
-            layer.c_mlp_proj_w    = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.weight" );
-            layer.c_mlp_proj_b    = get_tensor_ex(ctx, "gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.bias" );
+            layer.c_mlp_fc_w      = get_tensor_ex(ctx, blocknamestart + "ffn_up.weight" );
+            layer.c_mlp_fc_b      = get_tensor_ex(ctx, blocknamestart + "ffn_up.bias" );
+
+            layer.c_mlp_proj_w    = get_tensor_ex(ctx, blocknamestart + "ffn_down.weight" );
+            layer.c_mlp_proj_b    = get_tensor_ex(ctx, blocknamestart + "ffn_down.bias" );
 
             // map by name
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".input_layernorm.weight"] = layer.ln_1_g;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".input_layernorm.bias"]   = layer.ln_1_b;
+            model.tensors[blocknamestart + "attn_norm_1.weight"] = layer.ln_1_g;
+            model.tensors[blocknamestart + "attn_norm_1.bias"]   = layer.ln_1_b;
 
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.weight"] = layer.c_attn_attn_w;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.query_key_value.bias"]   = layer.c_attn_attn_b;
+            model.tensors[blocknamestart + "attn_qkv.weight"] = layer.c_attn_attn_w;
+            model.tensors[blocknamestart + "attn_qkv.bias"]   = layer.c_attn_attn_b;
 
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.dense.weight"] = layer.c_attn_proj_w;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".attention.dense.bias"]   = layer.c_attn_proj_b;
+            model.tensors[blocknamestart + "attn_output.weight"] = layer.c_attn_proj_w;
+            model.tensors[blocknamestart + "attn_output.bias"]   = layer.c_attn_proj_b;
 
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.weight"] = layer.ln_2_g;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".post_attention_layernorm.bias"]   = layer.ln_2_b;
+            model.tensors[blocknamestart + "ffn_norm.weight"] = layer.ln_2_g;
+            model.tensors[blocknamestart + "ffn_norm.bias"]   = layer.ln_2_b;
 
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.weight"] = layer.c_mlp_fc_w;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_h_to_4h.bias"]   = layer.c_mlp_fc_b;
+            model.tensors[blocknamestart + "ffn_up.weight"] = layer.c_mlp_fc_w;
+            model.tensors[blocknamestart + "ffn_up.bias"]   = layer.c_mlp_fc_b;
 
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.weight"] = layer.c_mlp_proj_w;
-            model.tensors["gpt_neox.layers." + std::to_string(i) + ".mlp.dense_4h_to_h.bias"]   = layer.c_mlp_proj_b;
+            model.tensors[blocknamestart + "ffn_down.weight"] = layer.c_mlp_proj_w;
+            model.tensors[blocknamestart + "ffn_down.bias"]   = layer.c_mlp_proj_b;
         }
     }
 

From 1c4d8bf98145e2a4c0955e154d74aa579c6d19ec Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 16:52:08 +0300
Subject: [PATCH 085/242] gguf : start implementing libllama in GGUF (WIP)

---
 Makefile                            |    8 +-
 examples/gguf/gguf-llama-simple.cpp |  182 ++
 gguf-llama-simple                   |  Bin 0 -> 607488 bytes
 gguf-llama.cpp                      | 4060 +++++++++++++++++++++++++++
 gguf-llama.h                        |  468 +++
 5 files changed, 4717 insertions(+), 1 deletion(-)
 create mode 100644 examples/gguf/gguf-llama-simple.cpp
 create mode 100644 gguf-llama-simple
 create mode 100644 gguf-llama.cpp
 create mode 100644 gguf-llama.h

diff --git a/Makefile b/Makefile
index a3600e4f215a5..f5922c95d2f98 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test gguf gptneox-main
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch simple server embd-input-test gguf gguf-llama-simple gptneox-main
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0
@@ -337,6 +337,9 @@ OBJS += ggml-alloc.o
 llama.o: llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h llama.h llama-util.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
+gguf-llama.o: gguf-llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h gguf-llama.h gguf-util.h
+	$(CXX) $(CXXFLAGS) -c $< -o $@
+
 common.o: examples/common.cpp examples/common.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
@@ -393,6 +396,9 @@ embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-te
 gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
+gguf-llama-simple: examples/gguf/gguf-llama-simple.cpp                            build-info.h ggml.o gguf-llama.o common.o $(OBJS)
+	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
+
 gptneox-main: gptneox-main.cpp ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
diff --git a/examples/gguf/gguf-llama-simple.cpp b/examples/gguf/gguf-llama-simple.cpp
new file mode 100644
index 0000000000000..35c3c818349be
--- /dev/null
+++ b/examples/gguf/gguf-llama-simple.cpp
@@ -0,0 +1,182 @@
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+
+#include "common.h"
+#include "gguf-llama.h"
+#include "build-info.h"
+
+#include <cassert>
+#include <cinttypes>
+#include <cmath>
+#include <cstdio>
+#include <cstring>
+#include <ctime>
+#include <fstream>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
+#include <signal.h>
+#include <unistd.h>
+#elif defined (_WIN32)
+#define WIN32_LEAN_AND_MEAN
+#define NOMINMAX
+#include <windows.h>
+#include <signal.h>
+#endif
+
+
+
+int main(int argc, char ** argv)
+{
+    gpt_params params;
+
+    //---------------------------------
+    // Print help :
+    //---------------------------------
+
+    if ( argc == 1 || argv[1][0] == '-' )
+    {
+        printf( "usage: %s MODEL_PATH [PROMPT]\n" , argv[0] );
+        return 1 ;
+    }
+
+    //---------------------------------
+    // Load parameters :
+    //---------------------------------
+
+    if ( argc >= 2 )
+    {
+        params.model = argv[1];
+    }
+
+    if ( argc >= 3 )
+    {
+        params.prompt = argv[2];
+    }
+
+    if ( params.prompt.empty() )
+    {
+        params.prompt = "Hello my name is";
+    }
+
+    //---------------------------------
+    // Init LLM :
+    //---------------------------------
+
+    llama_backend_init(params.numa);
+
+    llama_context_params ctx_params = llama_context_default_params();
+
+    llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
+    
+    if ( model == NULL )
+    {
+        fprintf( stderr , "%s: error: unable to load model\n" , __func__ );
+        return 1;
+    }
+
+    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
+
+    //---------------------------------
+    // Tokenize the prompt :
+    //---------------------------------
+
+    std::vector<llama_token> tokens_list;
+    tokens_list = ::llama_tokenize( ctx , params.prompt , true );
+
+    const int max_context_size     = llama_n_ctx( ctx );
+    const int max_tokens_list_size = max_context_size - 4 ;
+
+    if ( (int)tokens_list.size() > max_tokens_list_size )
+    {
+        fprintf( stderr , "%s: error: prompt too long (%d tokens, max %d)\n" ,
+             __func__ , (int)tokens_list.size() , max_tokens_list_size );
+        return 1;
+    }
+
+    fprintf( stderr, "\n\n" );
+
+    // Print the tokens from the prompt :
+
+    for( auto id : tokens_list )
+    {
+        printf( "%s" , llama_token_to_str( ctx , id ) );
+    }
+
+    fflush(stdout);
+
+
+    //---------------------------------
+    // Main prediction loop :
+    //---------------------------------
+
+    // The LLM keeps a contextual cache memory of previous token evaluation.
+    // Usually, once this cache is full, it is required to recompute a compressed context based on previous
+    // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
+    // example, we will just stop the loop once this cache is full or once an end of stream is detected.
+
+    while ( llama_get_kv_cache_token_count( ctx ) < max_context_size )
+    {
+        //---------------------------------
+        // Evaluate the tokens :
+        //---------------------------------
+
+        if ( llama_eval( ctx , tokens_list.data() , int(tokens_list.size()) , llama_get_kv_cache_token_count( ctx ) , params.n_threads ) )
+        {
+            fprintf( stderr,  "%s : failed to eval\n" , __func__ );
+            return 1;
+        }
+
+        tokens_list.clear();
+
+        //---------------------------------
+        // Select the best prediction :
+        //---------------------------------
+
+        llama_token new_token_id = 0;
+
+        auto logits  = llama_get_logits( ctx );
+        auto n_vocab = llama_n_vocab( ctx ); // the size of the LLM vocabulary (in tokens)
+
+        std::vector<llama_token_data> candidates;
+        candidates.reserve( n_vocab );
+
+        for( llama_token token_id = 0 ; token_id < n_vocab ; token_id++ )
+        {
+            candidates.emplace_back( llama_token_data{ token_id , logits[ token_id ] , 0.0f } );
+        }
+
+        llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
+
+        // Select it using the "Greedy sampling" method :
+        new_token_id = llama_sample_token_greedy( ctx , &candidates_p );
+
+
+        // is it an end of stream ?
+        if ( new_token_id == llama_token_eos() )
+        {
+            fprintf(stderr, " [end of text]\n");
+            break;
+        }
+
+        // Print the new token :
+        printf( "%s" , llama_token_to_str( ctx , new_token_id ) );
+        fflush( stdout );
+
+        // Push this new token for next evaluation :
+        tokens_list.push_back( new_token_id );
+
+    } // wend of main loop
+
+    llama_free( ctx );
+    llama_free_model( model );
+
+    llama_backend_free();
+
+    return 0;
+}
+
+// EOF
diff --git a/gguf-llama-simple b/gguf-llama-simple
new file mode 100644
index 0000000000000000000000000000000000000000..d7600282e3315a4bc805a4f6c01550d58bdb80a5
GIT binary patch
literal 607488
zcmeFa3tUxI7C(N#i=t`w`YL+giW;0=;1e?IA`j2ii;|LQTA&C*ArM>?O%uCNo^E%d
zvKectvB%8xm>Hd>2lOyd5wsUtQ&v-<S=}3HXpW+l{Jv}Nea_t+Icn4V{-4kP{~!Fw
z-s^kT*=O&y)?Rz>wf8wJPxGWlMnow3PcP+ag($~Nj=01H*SNkSrzkEZQ8@*_&r{A-
zdIPoLze`*-rv;Jxf}TbTzFtV_zNf`?b9%Pm(^I{Ml+t}xKb;f!6+KlPoTmGtD4WS?
zb9&L2)DlI}Q~5sjJXarnqbcn_h}TR{<@=~a$925Lj=b(^j-3}5UD%&zL_J9ybYJ|9
zoIid+_tYumrKkKVmkHH#lelh9ow!CnPWC%2!^|(e^XR_lfAIF1)1pEA20hhfe=Y9g
z5)%cUAQ8<|ri$lT$)%$oFGx?hbQR@_qQbeOM_y4h@3Nx8;)=@4DicRvHhSdc<t3L7
zXZAql6Yqq{*RcY~zJw?Y`uE5Y536_83S~uvQbpJ!`2Q&WZ_9gUfxY^iuYzwJSh@3|
z?XLi`{Mmb|q6pD+<4J)Sd%{aC_&Z<+J@E&C5A0SlC(6(bd&0*;F+G(t9R|`9uEFej
z!rd12zvPsj`R{>{dn)H!RJSMGZc(3oP)tw!Pg}_6EsN(S!4P^X$A)(GB)3Ht?Hvtg
z(Nj6Ez_5D4>n+NeZBhSgEy_=YetRnaJPZ3-Y$4|lE%*;u;4TaKKWBkwqo4Lv&r*x_
z_O;OadJFzHEbO__LhnCW$lnh+_oVj^E$WkHVL#Vd)bmUW{7sAcJZ@n>KUm-!2J|fF
z&n@s_7W$2~z}Hy#f#<+~K@a}tVT<-UEXsFT)U&TeIk^_N(?Tz!E$m^ZMLFMC;8QH@
z;c*N8wHE%cw?#cQi++8X1^@jPa%-}v&ms%_Hj94RYJqRH(Bn1>{1S`jeq@0^41dy7
zf4s$lzn6vFR$1`xwD22mSm^g`i+;DoLO!+V*FDw0z``DE7IN#bsOL=<{&t^5eb!mX
zA>IOCW`Vy9yr*`(WFh}YEc#uY#kdw>@!a7S_J67cKE|S)D=g&ZwBY~F0&lbMhXXC}
z6pM0JL+?HL&-X0&&$B51MGL$7(Za3<V{+TmxSVLgKgFUxM=ku%ehYi3wkT(X#dGhp
zn73SG(H~<h?BNcJdfsSJ&v|%mPkI?)QO_qV%DKg2e80g$Uq>zUdmifJ#DDY8wHEcc
z+Ct7lE%@KIsLyzd`UEWG-_N4_Wfp#HwuK#zx6tF67Uh2qKhu+6`pSYI#rG)Bu@?AS
z7V_+{c<wlh=N4Po;nx=9MGNH9Q@eh!@SlfJ&z|hoXHoxlG2T^O+ugEgb=;GGP_Vu(
z_s+|`<Pus_k5;m?=PxKJ&Mx=PDf4D$E7|U>jO=-NWqI=p%e{GJSs5usCB=DJbLJN1
z@oPO^DfJeV<;|Iwor~XIFlOt(EqRM|P={_VDJ)j9Z<svIn=o?jobtll?85wVFK%4m
z&IOi`TQH|A+gmoL&|B`#O-oBjNSitd%ww{%b1N$o5)y{<n{Z2E@qFBpkUeb<F{a_>
z5!vpN^6V+4dBqD#=Aj-pOrCU{JCAu7FKA3|$%4{3WqE0tlX7`QM;Ddk&MC@EO-Ne^
zV48P$e0DZkT(T&;D6e?Fw;(&OtgNJrvrqGuq24_-WGdB9G+4AZD?2MbJ8jzN?5Wd6
zK=e0cO`qmX%B#%HEA<wZ2q9Uv+oJYK*%{e$%F7Gq7ej8-MpCiIliu-4DTN&8krWmP
zF>oy>jKU+hy>NgeA74>icxy!-6kq0^JS}^4_B1anfv$|4mzO`MqR0!YD$4WbxuLMM
zw9NE{3)7e^lVD{F3X8!^vPD%#^FoKm!w%*xf*!JSONw(V%F6PJa~CrOOrBO)R-QJs
zkn<-NLgkQaUg~gmHOp-hVpbN_;v`Isl~*{bJTLE-`KbBG{L+$gcS1sT_5$=6R5061
zRVKq7QC3mxEnJYt6@Nk#wGc8&gH6xLo?n5MX3xpJwW6?$8SS(Z(qdt8d0v^<y&x@Z
z>a>!q|4@-B!%2a#1ZH5eB_zuI$hrgaK7l3*(@)BtIv3qDFHh7jdrtAZ?6SPMbBg8^
z=jP4Lgme-OmSXfmrkiQ;uyN6ZlhOqqh<cN#&0QMZZ=~M+3X2!cDJq2Q%H~%r$Sd~p
zE(rS=N$w~uGq;E<%;-d;<VMe}$nUAA3NJuLzPLCSg)9*Ef2`W|=UICv3)SnkiVmL;
zt`mHqMV*-8{Rh5mR9SwZ=)<BBMt2$ME-v&6mnPeV*}qz}AR!(CD$6S^nv<KCwqOAy
zK@WhpouB7L<TlS*veu4pO!idkn!uM$8<m|!-h;cftm!gySc(`HETkv~3e+<zOTPDK
zbp=vI@f;>iC?sKcHW@fZvHZd!G&*feVR?2594GY`*1Oq2!9<e)Gh`I*fXx<RRH06V
zc0iaky3LtexG;eRm-52w+#-1KZ13Vy>S*~Ty04gvfw?$4caAr=fRz~@ipm#bz+S*T
z61hh5gC>jZ?r-LnlzG7?3*sboKe{Zhyl@HBPUGXeqD<}@{-4l1YjeVAS&q4L%DuX*
zxhE~dK!fpsw@nNuVn7o8?muXAVrfY!^p<;zp;)Fs8WyAxJ9~a*Wp-&^S$Rn@9Fuo(
z_QH5nBMU>KKCbB_Bd?uFLE-Kwsy&isOU(8D2UVUkEj)!WJmAl=buqGKA2aD1o{a&;
zV#+nGkoktZN=)K%#rRb?l?`C9Na(JFk2d-McY?{1;C366rA!*k$g$6x!z_>Hc%yQu
zr=a^`yvWF2R5&lM7{wQQx#+-<n2q^|cTQmuI!tC64c%pV1#`*^vP(;f3Ue1HjKpol
zd3iLH5e$!}JN=WHIT(5?9Oca`Dw=OR))4zi_AF!WxnO}ldn;blqvsYQ@_%3!G`HXr
z^|YzfO6H}?^1^;9JieAxc(Y6Lv&-fb&(F)wUqIvQ5^lIZw_)_q1r=;gC=C{vT&bjl
zxr@Dd<(ZQf>Owb4|5?`B!~9%sfEcX*gHd#(c@Whn^5pX^#+tOSs|io(c@3GsF}0T5
zxHM;)mj~sr$qaf+4^t_~9~Y8z7g$s>AA?VL0m2H;r}<gfEn&&GfUowvbBYV{72GpC
zUn#$}%!}U`T^K;WSguIPEh*1eiqV1R<YRWO<W-jDE2Sli@+pTKQQZ98-0X5HGC|3{
zu6PmV7}-<v$}tC{i(;imm*M@WZp9_pw1i@-8hssDj_#Nb!IGCOE>(0GMv_-a+_#vZ
zied(${9N9a{QRPd@&YBFMq5;`ys$ERLEZwm-TB4J0>O>?7ZoY+H^sT7i<Qy}Z#n&<
zB_=9Xo?CzppIchNJh_s~%9mi0BHimPftV;MQWh+jQ>yU$78m9f%~SF*-6>bFI>gPy
zgax7KQ5YF23%v@egqAA#w5m`RutzK`FD}L6m4xq|2Mbm5sZPX6mCQ#yFnF;CR^W+x
z(O8t1QM;%HWOv0SO8%m<LM*pY&pGqJIB#JgZe-2KMzv^xf};78crS;zA)h4}8nY{D
zNuz+Rtf+7Sujhi2g?Wb1%kzrzmASbK&|GhsNSKx3UebYB0WO$>)jXyiSfk}D6H-#L
zM?mKjJnr!+*~2d%o)pfG31=@KW#mU(K9c8W%)p%^FCQ*$7G#3BOD7Y%Gmq{@CM0wt
zla!PhGm^)<vlA{KA!YT4>ksUH)9~SP89a}M==DH0+>V5i6mf}Edf{9`B>qJq*IVg>
z-}J9H`$tzJk>hkU%SS7H#T|t9WA_qYFD07&r?Mh(?xvs8pV586-y3%fK-x<=m0gWe
z>`19>`F|wtIfY%P68MvO$-MO5IJ81_8EEkIMvSPpLGjY59+9kugGCt*&{T&=We9##
z8jXK8ly(ZMGeTU-P@t!=tJFSfStPDg9ds!XML1z3Ed&(t6M=szh_&hQyv&OiBa%5(
znN0w@-b*=vm>ZEU{MYXLEgpB8G7kV@cQM+blnA=w%tAzR`zlN5B7P$9tQ=8(JZSG3
zbbW|&XE%PE#D5Cz;bk`<E$#drnlM<oU+~Xo^dRMNK^HK3pt4TT3mM&Cc|*{17=4QJ
zj-a*W!%$v7<x@ewxAqo90;837L04>}9mhV(K|$Zj?uXP6L<FRiF*-^aBxvuj`w;=`
zrJOH<j1iD)gJ>_<mESS>*p<r!|LKK=kmjjMqM%o?>jRYWg68szG~}V5XTn3+Plq@!
z79-9%>G--C{JPVGw+TGXgs&6js3ts~4v64qh6!)Y<K^d=@OAkdFR;K%O?cIQUVfDc
zk3Ycil_or=7q{SA6Ry0%`B$6pQo;YU33u6eIqOV#&R~w$nej$meuD`w75vV8E_q%4
zt|7dfI1^qvl;iOxymd1#KhcC&34Yatm!8JU$u!}$r*nLU2`_zzm!D(8*9iVn6TU8n
zms4rNGtc07l?ktUpO?SVjGxK**O>6OvpD{=33r{t@dgu~`2u-0MQJhN@h@|{)r7Bk
zgyTC*xMMIj2JjO!;cZQvzr%!QZsB;R2`?4TRpxisd(Kc^zQcq&TX{J#CcNq+jyp|w
zt$1#n32%tu<tLi(H9L7ZE)$*^<hW|W8^m)nO?d5Dy!;#!-tZ$Yr@(|OQJlZjgx8Aa
zR+{j2=W_m96Rrro#20jz=k@oB_L}g**&KJ7@C5=_O?a8Wb4>UmfftzYr2;QC;VT4Q
zX~ORmc$EqF3w)&suN8Q$34c)Ft4;XB0$*do{~++EP54s+UuVLf5qO;m|BJvIO!zv1
zx0>+9)4BY2nD7?_f13%f6L`>szb5bw6W%EBP7~fFaHX)jooo?!v<ZJ(;0_bMP2e#m
z{GS4Mn(&VW9%sTo6L`D{-zD%w6W%UxmkI9>xN5==2|Uw;9~Jlv6W;Geu9qAW9wYDq
z6Fx-Xr6&9wfmfRFO9WnJ!bb>vr3t@E;I$@vyueqR@N|K%G2s&h{<I07BJg!4{Ca`c
znef>HZ!qEW1m0r8iv`|l!WRjAhY7C{c$*2oTi`(xzFOcNCj8F=?=<1-1g_lN-Tv1L
zJlceB6u85Lzb)_>6aJpSohJMvfybHfodS<H;UR%1n(%!BcbV|R0#{AA?Iv!|nI?RI
zz-O57a|E7a!Y>whfe9Za@KO_=An-~PK33pWCOlo>D^2)Rf!CVwn*_evgx@UiH70zS
zz#B~Xg92|c;V%fh)r3DR@EsO-n+bnR@CQx!O9JmO;jak1(}b@VxN=K(yWJr0XcN9!
z;0_c1rodxN_;!IiP51`_k2B%_5_r4`-zD%w6aKBhT_*gXz*Q3-l`ZVngr6qx876#?
zz;jIaxdJaR;a3Q})B>+G;Uffpl?fjs@RcS!Mc}n2+%53cCj44~uQB0c1^%=NpDplp
zCValY>rA*;;0-3cO5iOf{62xVn(&7NzQcq+EATcG{+hspCcH)99VYxEfp?nlodQ>i
zy4(L=fk&J0Ljrf0a3zP^Rg4KgMc_^oewx7JO!)Z%k2m3$3Ovz-Uny{x2~QWeYQkp<
zJkx~d34DeLFBf=@2~QSyfeBwK@KO`*6L_Tw4+y-<gg+$kl_vbp0<SgUZwS23gzpe^
ziwQ5?fp8LjT1|MZz}rl?vz+U<(}c%}_tB#Dl$EL|>qI#Y6W%ItrwLyt%Fpz6w?mz7
z5YMI6JiVJ`{D^YQxZZBUr{h1x#OOz_uNe<Y<?|IK=Z}-hm+{OJUQfcfBsu>K2_MWM
z_B%(y=^Zour$EBb&_Sf768@NkS4ud&r>6f@NqC$NB3&urDH2{Q;gcnNwS;F&_!<en
zUBaK1@M;NPC*i-5@Hz>vmGB&)chZZD7g*p8QaSfa<+Mon0}|dU;SWmq4hjFAgttlf
zY6%ZY_(KxjA>j{8_zbE3%OpIel-m#WDr&R-<CJh7apG6vBpgR_jh}c4=aF)LDN(|C
z&*ivF!g&oiu1Yup-^Nd-gyWS1<7bA1BM@!;<VZM<6dOMU5{|&K@lz_{2uB+~l@g94
zt;SE4gzJ5pF0Pbt5#na|)JnK~&ULkf4>3wtlr<7QRKlN@@Y5uGorIq*;dK%oBjF7a
zF8i?-2|rWfZ<X+~Bz%X2>+eL<<u(aFN8%4k__-3^A>nlPT>t5maHkGZdTZEj&zJCM
z3BN$X9TI+_gvUsDtb{uy{2~dDlkkfrJYK?oCE<w@K1{-068>umS0(%s3D1=9OC@}U
zgkL7%ITC)kgcnHo6%t-5;qek)Dd7ncUM1nfC48lXkC5<M31=IT;@WBnA0_dxk?_$H
z{<MUTk??gAo+#mU5}qXC4HAB(gttgIoz2&OS|$7{9YnfA!pBK?n}lC2;Xw(%M#4KJ
z+$G_i5}qvKN?F+c$4hv$gr`clL&DP}JVwIPCEO|D6C^xN!c_^6mvFa)CrbE433o~O
zBnej~+#}(c5}qO9GbDV9gy%?jri2$r__Y#VD&bQlyi&raNqCinXG!=<38!}&^q*P@
zzg`ECu9k4I^~kQSk#Mo~$nd8n{06C<brL>H!s{e_wuCoG_>B_YBH=elc&mixNcav3
zpCjRI5<XYLgA$%A;T;k_Pr^GTJWs-v^058qOL(+|&zEqAgcnG7jD#0TxKqM!mhd<U
zzeU31CA>(&6D53sgu5iXSi)5aFOl#}2``oK84`Z0gy%?jnS>Wec)5g^O1M|TD<!-_
z!mA{Fp@grL@I?|{E8&$AzFNW;OZXZIUn1d8OZaUPzD~lIN_d@wFPHEJ39pjy771S=
z;jI#WhlKBtaG!*?NqCKf2POPY3Ga~byCl3*!dFVTBIfgbfN2sQE#Y@dxI@DI5*{Pr
z_ei)?!UGZ>C*k)>c)Wz)C*g?_zDmMf68>8W|6lfh68N74{wIO|N#K7H_@4y+|0{v-
z?dKj*ecwi_{;0p77p<r@4c^GER@JvDdLt7`*Qi~9yDr*|-}dud$PvwE_Mx^e{9N=g
z(Gg${wK94;(UC+qFnTM|y@+1N=nX_u=rXj1(XS94MRYBrpC`IE(N&ComS~!Og-RLy
z1kto83FR>QVWRsIoyq9?iS9?Vi_v!zeG1X>jJ|{D{zN+&y@coiL^~K=Ml^*cLkgpB
zCfZJP$58+m%_W*bk)bw5-$3+0qFWg~m1qh*h8h?>k?6rhuVeIhqK6Q@hS66NO`*q7
zEu)7MO(Dln6{CMmG=&;NrHsCSXbLfgau|Ik(Pt2y$>>2uQ%Eu7Vst;EDU=wBXLKad
z6haI+8GYnj&=fihIT*d4XbKsI6h?nTG=&O79Y0a~cM<I*x{c8v6Ma6>t&HAI^aVsW
zFnTM|7ZSaW(Hn@S5MXExqhBHVBBE;<{XEeZ6J5pVXNmq5(WQ)jg6KG+a~S<F(Zh(&
zWc2++|C(qQqwgmA5~AZ7eFxE(676L45~437+QH~DqA8RYQW$+R(N_@N@gvhe(G<E1
zwK4hzq7#U2W%N{{DO49~VDv<yDMS}q$LR4yk0g2xqpu{ILUN&6Mh_>NLUExgM*o^<
z3c-a+8GQlK6nYEgF#1fQDdZN)Wb`1SDbyBnF}feo6k-d-GdhxJ3ay2lj6QM>XbP!?
z9E{#iG=<Va3ZuUvnnGxyjvtu*iFOg)#^{fUPA0mQ(c6h0PjmyLw-TK~^g2dwAeusE
zp*4(th3GV*YZ?7K(dk52G5T4eClFoA=qHF)iOym4!$iA@&SdocL{B8z#pt_<o<wv!
zqwgTvL$s68ONh=O+QH~Dq9+rrF#2Yqrx4w7gz2AX3U!6r7<~iL*Am^z=&3|gXe-pf
z=!ryANGr6C(c_8EB6<y@uOym6SfN@*4<~v$(N&E8HPIBx3Y9YY0-`B|70O}snM70Q
zDwN6SK}1u?D&%5xKcXpA6^dtcB+(S23ON~l<ZRFsnhH4>y`N|bNre<fe?xRO(H)1G
z{)x^Zx{c8v6FrCMRz`0pdM?oojNVFgF45~4y@6;7J%!dV`W2${h^}Sy^F-$pUB&2U
ziJnh%DWjhtng*ay4x=9?x{&BhM&D2L%|yExeK*m!5FO9xJBThK+R5l8L@yxP!RRug
zDO40v7=1I*B}8|0GW`=xp`lP4qi-PkR-#)OJ(Xw*1%(<IJ&|Y%0fp8vdOXoyqSr9`
zN}?&`6RKtOaH1CyUB&2M6HOtWP${D?Aeurup&Uk^Ni>CYLYa&nL^OqRLM}%4Bbq`u
zp?F3|5>26-kdx6z&H}xRXa}SB6MZ|;3ZuUvnnF6EjzdiUL{|~r#^{fUUO{v#qqh@%
z2hk0T-b%EO=yi<VKy)?HYZ(0s(KSTZa{kkzmHgVSi?(-5ah?54Qv>JJ;|^8(JlF>R
ztop|dI~`<T=p79GY*y$Qm~FoOJbHwp5Pw|75XQfb^B+dR;Ge4VFP8afK9sP>t!=wW
z)pn`A-5r@(o<P)BpY+Bp14>vjt_9_G4J|`42Wr*8xM5UKAnF22k~+OZF_oH3kHA!_
zZ&#cByeg{4Mv;R5@R`IvYCk2zj;dOR+PL=`wXw688nId3cGNor1zaWyi0*1bQ{XN2
z`>B3a<6gwQO2sADslIU!4@FkhzV!BoC~gNHtjDk&oCj5HjtYQ|xDoeuyFauD7e#%p
zOTIpNdh&J2S<}=N-^TQ!qQ2LDQUkBW22tx^1}0*@pCY_{ReyPO)zz0(j8Xj~(1we?
z>a8R+1kXL#)%AJ=7>ygw?u$5CP+hT!%9@otJNd@so079{-01r$a_NP{RR{Lq#qg}Y
zpL&%IM6nJ+I|(KKs|Frngz9@O7O!Zbm3GMLpZGd6X?!nKWq;|l*uf|~IOld7d+r*l
z=x%(7O7-*NReyDC8$eY{jdeBs=%|h*7G&;;ZD5&!o>*toj{|M7O`VH^H@{_5+}dWS
zt0ESX_YHws{E2wzt@O~qP(KnhwHGSKW{4+Wx<@-i519EaJ)p^>y%1Z4$E$&*u>~p~
zsC`^dq6>a>NpHpD2k7z7i!D|Csj-z*^GuKST`*%>l#&{7{pitldWIcV8$*$*e`wp_
zXxz{*IjLpY1*qqDyIDQIK;qHfMO9}&N=z%EDyGot*p=+L<Nghn^(0jGl9ET;5?p{G
zJ|pRC@2@?9Q6u8|pnxCzBj~}lzuQ0@LOoI7m;+FZZ==nV^hHG*WOO<z02u|ZfP?U8
zMX@v5PfbhO>+vIJucj|&p#+b%G*%71enk{G?_m{4YOx32;}?T}1fDVMM|b15k*Uzu
zx2O}l=V+?7OKS`s92ce3G}voC5bWu%kO((X#m+)jYr@5qAzZXqS3rhp?x$+N_9zO0
zbgyy+PlqE>wMJmG)WF#*w7qEwU2g4dkM>mTYT|(=YLT|b=GUfbNlixvCfk~m(T%7Q
zdsJUze|J~2C+VoY`fL=wX~hzU?KL&&ql))D!`jt4a*Fs_t|o1?SI<Np+}daE#-ovH
z-E+_KACG-)hezvh`;JDIN0GfN69Yry)%D_XKULeOChcE1+Wnf2cp^H=lQL>{d#^%e
zQe)>$4A}li)nTo1qm!vxQ_~R#uC$-6YTLTqSk~;;-om{Vn-ca+35;Brt}T60+4T65
zfvGRr#y6+F$l9gWd~*A#Drsde>@GMRBU?tz9&a>Bbj!cHx+VrZFDjD)ws~q`b`4sz
z$Kw~W@^e{9QmOd{4yN6fuqR=Q3gs85!$NA~VX6J9eQk@Xd16a9!<3=^rLi;AfS2r>
zOycfFo06<;4&MJaN}H0CHruQ3rM41Zs&+7_;R0ODQy2AtrpWy2_CWytsK*`L>grKn
z>Lr=(q-O8HG;K^qV7^VQ*;8>%Xdyg2eK{B{Xbuh}K5b*F77x>L`!?C=37Kk5mp3Z-
zPgt}^n};&A&vC20`YIY5AdJ^=A#?>BD55Ia1Ec8^#k|L$R$K9YQ*aE4Py3Oc`DQ2*
z%p}ICd$<Rx0(USiXj2mQ1kYV!Q!;A0?A8B5m-J{|!5gXgcR`>%XYlr*2j`$rc$)K(
zpt%o0-=V9}G4Z%Z=y8erS-y($mm<%)R{NRlG3}?d4{0C7$y0b@YhT#zC3oh7tQGG?
z?JC28wD)hH)jqQwqS;JVj=G=2nJO+K56s;CP4E}!^}#=}4)k5{Z{Op|yrfS^dTQ_y
zLUi^k1p6Ye(=&Dabb)6B&%azBirF~EhMTNunX2|fb06wQ!JEI2QnVCXQ)gV%^3QuK
zO`QW%qSZPGkuASfEhVO@<N7UtCpnvba75igJU<LfiEmCxRFtY~)Idrpm|SXeib`FL
z7&Dtwa_Ac63p{~xS93~b9J>}5HC8-hsap5;TlZY!0<3Djd)U>%{#BIXDHnmO+4l%2
z`oZ&jYmj;ZPg2D^+LKQsQJa11kkHR&AFFV)uZ59Rt5;i*RRg{@N;Ds%nr*Q$jQlEL
zd$5dkYktAsKUUTJ#M;#9h+4?2sHm^tJ6T1S&n+xgwM+*XcF@g!cFSfPIzzxmXUR}{
z2QD--qN4bg;$~kjz{Mqtic!T*T%zd7W?vr?5AJdDdmNNB`(g;i&t`_xbr-)L&#xyM
z*9oWVIsAGizdpmbPB>k!;@3<0^-AM9;dFfszh29)uQsj|j`m;Q0JZtn@v9BuDsJ0B
zh~{hI7h1)I%?zhvJNfk>zusY7CmgMdhNq_XI*`(%^U-vIaEV?GjcE2=z(~d<UJ50q
zi|Hbm;wathyNsBcee|+6{ZOEcW;vMSY=qBI{jZW`crc<y1#j)3u6iiA1y7^Ce=Yic
zE!qDx7{BU&gVj{j8=fpG=)w&sfwfSym*H_#kpCYf5|97YIwTnwoNBJ5OwB=Z;)jAZ
zelN957csl{+wWMJr5vxKgyHXVJd@)&2L2ky<2mj!@TWNL;JDMk@8LKZAfBrjxR>K{
zCt%N=&2iR$0<72Y`lJF!Pt8>Qw=oxex|pAO{MQ8I=$eKBtE-{|*IlZAnyNPL>XmRX
zn6am;i%e!yW97i!?3!<5shaeISDi&V3GLYn4}>4X3Aufn9gv=Pu-kt%<vmG<yzRlw
zU~AtkWDJ?wYo5j!7kUoASUDSw^@OfJPhgQl^)GkOaB`MZKGnAprG@J|7S)Xw)qU*S
zZq*%&NmV@6wjZzUHMr6AJgpP!jz06F6%7Bj6&q{G7jF9E*8U9<o4#0lPJ_~P`1Xd4
z((`M!+kZns*IIo?aE3c!OXy*u8htz4)Bu)2zFiGz2@RgWfPdgcVyp``J+W+rtq=Y7
z^l>!(nBLM9yng%K?aPMUKV$_XAzVh<G|uFtWOko;c5%n`yV=E9jghDuBl|zJ>G7ol
zhosZ*)O05$^U__Ec+zv+z6O^kX+uSgn$+lx!>~6DH8JFyYJb@_1+y22?G8+ZY`;;v
z8r7thiXbM^SjM=Nge}yvX!||P7Y5W$x3<OI80zKjYE08wl6`G`RNvu5*6@eWaJO$~
zXTlaqK3E0V?%njovU6;cX+@BFuak(>$qr7rIOXJ24yWRYs=l`h<#E8pfi(t@$$<s~
z$WfY(mQ>%{W&ouSeWT0mYwWbwRASi6!0Ipt@&}tyA8<swyILyRlc^;wy^@n!mtCTv
z_HzA?F#g3<fAHJFtV?$%ZL<fu@I+w4wt6uCJ+w`YXoSl9crVlkVJ6zUxb19ua`8}G
zy5jC?P`kEPd<ntzvEQ*C1xU~9fuEq%yKLbyQqv=8!U-^125kcDFN3`W=#ar^0S=MD
zegce<!Ttg~TLw>snj4}$NmwvsK(u5khCQKgbdw+UI$~%t8vO=bl@k2xH*k-2pNxFz
z!#h4ZD@8q26zfn6J+aY2ws2}y8-rMx9~q=Zw5ky;aP~1LDDOsA-n)mUeeuWNZ*qpq
zbB(uS@Y2icV_jb8ln{9=!(Y`=O*V$<OyAML<-bDAAYiM4XISt9CdvM+MyIM>?<Cu9
zZ2ECvA64s|Nm9Th^MxpVIs5z<Y?8SV2_hPl)W40@JtmxS@JhkKL5c||-Qoc>68Xke
z&W*-Gt0oac9)c+??EV*G8z4Rpma_5Gy~6cdPeT9(B&;$}^50N$SIamuO1p0c3Sd(A
zXa_yok>Con6pQY`{sE`s30;la06S9O;pnng`~#wqU2(Vzg)Pd%Q&*dxs@rq-=Upv7
zkA`>oc?0o0|A4h~qOr8BaM1IfM1I9#3Nw_w?2o#SU9nexfZEn<u~)x?6qEBC`qE76
zoRQ`lq$TWbKfBf+bpvh#+X%3^yEbVX+)3}-eUFiJ5I}0&bqdygPc&DbrZ@rNr)En%
zCEoWufxn!lU^<wvr~S0B{JNU_1Nx)Hq_@kTlG>K4?M&8!km&4oNupF!xrJ9ihiKcY
z$^IIAIz|bo5y@JBa>-gzjne*PbJR0ApzzZ^3-hIg)3L(1V;7mls2hLPM+uhTehOP%
zjb)RneZ8J`QB-Y1@MW@U;OExG_fwP+wpja1p4iCzzKsYv1;)J?iCgCp3nDc4e2sL)
zCbAqU`U69}q!$@tH27y@9kYHIie^!WW0*r9p^3QRNThDLoo&o=`{F)W?tCQmTd^jC
zG>W0p;LI->|5k&)h4X)e1pH%&pZ-h&_^&4ZVVwVAgMT&We}?n-A^tkfe=6}G-Yvxw
zed7wV@f{@l0}U9C5SQ~Wic$TitNtm`9{&V{vc_G9+Y=f>wlAnVxfIOyu+}-^A-=%4
zN?e4n1B@rwwu{DrgWk(vUgrX0{QiY8ZvSa+|K#Wle>x`ppQG9@6Qh5j1I!hlp%}hi
z9F@+B;Zx9?Lid9yzgG26h{5_BRlzSO{fZ-(W%v_ZXDDNe+erLzsJ%*ScC;ff>XuJY
zMZEyVMf|o8aH)eX<&OL<OhWCuUv2!x>FtxST|IOV!GP_D4q`l4?GrHGG}u*NTL+hq
zn!A~m=0_NBsGXN>+lu^?tg!xfv%<Er!g^C-`!=GieYm-r>{Ma%pGDyHWA+38#OUUK
zp(1`;<Hr12M!yVt?!KeZx7iT$8Q=zY)Ba!gU(s~rCa|y{qy3D3%wh7H;rp9S^q-km
zW!L?EP~!FB=heLhXQtYp?6luc7NBJMy6oj=)ZL2!Ilywen%j&U?&KAoNgjAOIv@`y
z#(#!t2ZkmifnELutGxPB2eZpReFZza4+^ZUqbaxK@1dV~b?k&h`|s#+E|fGOp7b<e
zBgLY}+O}0>LVbx`V}Vf_RMYM%>d#aJ<y2hU-dE_U;w+To!c9|RXm&M#Zg;9l6IA=E
zhK&iBP#W^0!q-KpNmJr!zI(k(P1?4ok2>~xl~;)B<50(@J1U&W#UK|`aT;<?<eU`<
zNyECvrpJk5yN@rvZ&g~$C>{33sEcL~fmgo$+}e(=0cRi|+=NbQT<ycIF8`ux+<{pL
zXxpp*jr(X6cQEf4{2G!&A8NzJw1k76z(pU-I#t=UYw?Vz06urqba;7cY`iCMS;H&`
z?z*gg)*!Ia76JNivi;@ero+Vz;2ylAsdLzDM23$HJgaHf%&2j6FNNd!6E%XUT{8z<
z!BEH0OOS5bHRqYFO-E)n2LFiaHXWG}HI!YQN%xp0>ewN0&BF9$Bj#KvsWUE>o-{M+
z>vs{=`HJ1{81MlI6w}zW|0b-0<0wpx(rHS*>5D>R&cDfrGA>r|1hgyoEPwQXhv{BX
z>`mMB$8Tlpge?l))BkcMxRki}eMf7ENPP{#>U23quV$=>A}P@8UI8nQz@YyrRzQ-V
zZ{OLCEMerhM0nCy;%2~-4+dZRyc<UbI5tP!aKjKKBELCmE~OEhqi&*fb5yo~XD}Go
zuNT*+^Xs#CvWh2%XF~NJgqW&eGYOwSo(&PvO2YPpEmc=VDqe5Z(#RaeyR7Q9$ObSV
zGOXb@C@0yNz4`%8RdMQ3PPK9B&zy2$A^^_6b81bPqKUb>Vgo*>rl>29(4bx658<x|
z!kUJ-F~&!!5gS4@<g40N)weYQ@Kga~{SqQi#;-)mSA30;?L{2#)gABEp2cxncih%~
z7008x<5BIgsA6w5uqu+;(0&@{g6j&`!`IkH+p;+lb?c8R??ef+Bck1Ydxn30v?p+@
z-R=7dLF+aIl0EjceH92DrN+WJQ5%=WI@VLxqwNpg^C?X+yn}t6{k=ndou^ft0gv{V
zZ+k247zA-g`zX*a5RIn~U4qQqZJ$Tqwz=ttnNgLPZ({U%79d8Gw~(xh2!8TOS64y<
z9-C}`X^O+6jX*Fj+P*5qv2xPDm6HZF9Y#Fpr-3Ph>DLGZ`7&9ZPzC4H_KNXv7vEfu
zWv#XY!ywlD`hJs`{l0XB9DO35_QDQ&=+Gu4@V)Mj;d|o;(iQ+bbgjYl8_x9y&NY^}
zhJz~sopDhx0=69d_FpWZVE9Lh<rsb{8&!c9i<8D%RL~7Hc%V)4XJZ)O@-JY+wl36=
zb`p+|aRs8Xbb24BpVH}eP9M<eFFBn`qYK^tIj2|Y^hccDt<&#udL;S`<A0me)jHkG
z>GyTIk<+m>OQP~$<+N9)U*z;AonFi7AvDb+{%1HnPp6;c^b0!uD5q@{$0YvWaXL$<
z@8k4SI_>B50iCYlbSkZH=>FxLUZvBEIlWt_%Q-!=x8PsE>1v&x&*}GddJd;!`w0G7
zoc8MUb)4R$(^EJ-BwFyhIXzFOQ#k#CPLJcXjaC;_|1q4-(&+?FKc&-`aQc8wU&QIu
ze&YV~IK4`z&*1cKou(~Q*w@HY1pfd|SL<{iPQS0y5uA?gFZh3eW5NAio&GncH|g|u
zoE|bj@PEbWc{=?mr(e+N_c?7lRq(&b=`5Yz!0D%Sx}MVqbb1}9Q|;pZKXZDOPCv%!
z-8%gsr$^GJ3(1@Qa4FhTt<yeEzpv9vI2}7s@ZZX5uTIkuN8G<jr?WXdWRT#W&gpqN
zox$lB1YI)^blo*@W+7TwW5tj_0F6hXUr`SKl@NV|unvk)LiBmZIveHsQZ9;ev^uWC
z5?%?>e5j71{GoQr(HbwblXBRNS3)0B4huIW^fu*Aq1;Bw^`{(dLe~wT+*-<=O1VE$
zj<&r*zo#5I|ImGugELe@cTsK-<(5%yFy+cAH-vJ9lp9L9Y|5QRx$7v$&M;1*90pJ&
zluWrZC^v?3XHxDm%AG~I3n_Ot<<6koIh3<g?p(@6QSLm-{eWL}PRi}0-1(IInsOIV
z?o-NLNV)eY7fZP<l)H#>Z&2=H%DqUrUs3Kk%EeLcamvv$HS{~m(ONhZpxh;t^HJ_n
z{AyoXcNrxWb(d38Tz3T}^XuX%$*D`AWM<uPN~YG0pkz|rNJ>)cMo}`ZZZsvM>&8%W
zMO`8#adk<QoL_e(C1=)+rDSm3Rh0Cv8%If0-PK4C26Cy5Z5Yi5Jm1+zp<SV>+Z+m=
z*SIY@xOPWZ7Y3k~%puAw3W=rS2LBYNTC>HA1HZ}9YOe@3+n2^GHINmHMf#LD)weuR
z@kXisi1x^Y2Gw7Nwdk!5Y)?gEdnUqIpV?oUFc51jV?1=NT-1%m3Lsj(uKN3!)}xDJ
zu#b$8d-WY?560$i-k~9SfDhBz_KRv0w!05~?@rp_wollguPoU@GrdK{iWoO~Gkmwl
zC>8s#I>8$J+=>3wGvg-vv(8kp4c|J^zBctt2m2kvemlqehn(rgg0?jy_w4(y`1JJs
zm=>QEzfIVleCYf3RjSq^*B8O^_EV(xvG{+owk=uvo<g{1A(rl6<fJWw>ZeeDw|{!H
zyVq$f=w7vapn|`o0)`2R#JdPJjPv-fk4N;+lhkUzle@@qA0o1Z8Ok|0oPlXAYNT!B
z6OM-M*n@Zx-5q}BeaxLE`)xa*4@_h7A*0E;Lmp=J-%j<nSO0?;A95c$g4Q)8`wmCi
z0|d2KQM_N&v1T@+M_3O=(Vt7CMY@O9lCvY7o}~BechQjQ@h^|g2n+#wgVU{@eY#sq
zJw1ASz&5t&@ORi{S+ab*<7vCS=2u`<0~2C00=H9C!edYE%+OBvAP7kcS0gr}!4-o%
zfd^>zjS7$~GVG}vv=PlI<i(U!ZBjI1-%+!hvC*1BjtHUf<tYw&3}xc!S0?=`#V=ni
zzwUb)sRx^Yf@~FOhe#EA8e(_9AxDy`y#p~g%3?eLTcmqf3xdS)?iHO?;I;c`0f2gC
zXpQbe+uRZR+}eIO1<IYKc)?`<$i~T<ccTMQayO#n-?>?o{B7K>22vvnCi;goPSmm*
zS|%b28%?=Z$~h>vgK{yHYonZ#azV<)QLck>@s#VNT;fD+NCS^!Z?1Tzy(C%dz-~r|
zCvZ<BHQo~#@^2)Fbnk*f4;u09O^@GRVoRXko&+Z)SqUymW+&uqT5<dG)PyyZcoG^Y
z$x3LW<m?2uZ*Nsb(*BA@cT$JF<`7%tMcw*lZ)Kvt{|;4~6tDVI5X>5Sl%U*^WM@Xi
zM}0^=bW}uq>}wk_U8m?nN(|eCN5+Ya+BIPa=HzTIi>Bi4h*njb5GNVpdpO?^D#OtU
z>UR4LlfCC%;HlcyWHu5)eFN=xJj0ql6<ccWslG2Dk(7u{J88-O#j?n#+cxx8e%!Qc
z*~|eKG*aNL>8Fxy`2Ud}KbpJ$r#M%2BaMi2_y0T2bx$KX_qqG`#JN^Asx13uoa>Rs
z0%YM|9GDwayvaBl=I+OptvrusB{fE%uxM9f)8W#Vy*!J0+}_|$`q*B5I_gdF88{hR
zBW#6sxV{fX0L231ioN<W<g&NeGURF`aCJk`^Xp$X78q;Z#{$FOid!N!V4bv@M`qiQ
z7%t(j@37@Z*xw?!I0R=0cCuJ13_3Uq=hS9~_dnQv2V=pm2Jyiqkl_bM&|y1cL&Naj
z-S|B?h6cC4lm6&AyXh#MTwyoq@&$h!uL8<v{2rYAfidjFIW2zE7JgG4i(gSZRQuko
z9i@R|<%B_-Ck#Z?!Z9HZqY5HYaFsBWOtpCejf2>)x5c6-V|}-CJxw@_R#u=)EEp#b
zN?tj6AU1y;*j~gIaC#g9r#RCy8KpF9pXs4dni(B?-VrrNRMK~Nko~T8sH+<Xdi=|s
zFvX>@G3+-AL;G=7DFY|l5;JhRu!ZUp7}^Y@Q*#l!jM&z-6UTdgGYj|QNZN0vA#FPH
z>&;qcs<zL!^So4Tzi(F?%rmCN9axl;(A9pv$-k&LX%?|JBjP)3gWIbI<9;Y7*>}`g
zHj0hcY}544H9TqHNgE~o?}W2}-`-I1FE=)pFr>dg_J`AI18sM~05g&{SL|<JL$!My
z>Z97xl8tJo$A<Ly5A7jqhgn^(!2vAaSBQo!TilSU9rAtMCyn``&Wym6oV0}P$JF25
zwIL%Sg!;Raw%V(w;R&dGba`}I!Y83BNDGv!ykEj6?bG@E12{>*us7zUW-=$0I+Ho6
z)S1jl*=D+Z8?jORS;fE9n(f}-;asX)`yN}rWtVolA48HY8!@YBchZiEeHqxepz~IF
z80VP$@>4Whw!3R%1`7H<G#M~S<T>;LlE{-t=@^&;VPVY{t9G@Z#V-3xdsSa2cF~r`
zM%#TWVSt#Pr_e39wN%_n^M9dc2n}k6zgQM`+vXmnc9!9<C3$VIS6_%i{>$?EnB=vE
z<n^9gdyjV0ey+Sm9xJcgAOn&Y&GNh3HZ^+bW_MsYtfQfQa4osMi0_y^G}x<Cz@L_|
zrJFrtP`VfPObP6n64>+Y%Wd~!a~$@(mF#(&Vb3Yr$a}H3pON%V#ew!G)L~b};%wv|
z7#4g&YHSRR;6KXtK&aEBeXZ@KXw~I!(n*Z=jEFCBLXSgd0dhxtYS5>c=pVLrzus=5
zKQYm78nj#cls}>l5Cy}W7*L`;z2?CkCSlL4CI{B1`e(xr*L*@}eI4G5;b)rM5zXQL
zfO8|fka844ZlUYVigv{1;L75T32%Lfzb?zeTmNig<H}b^Fs?kdALGiO;4tCmCt<`0
zjz(d;|4oC#cCe_n4Uwypbi%2z(NH8jAAEemLBk1m;6Ri8rAdPjse=noalwVBU^8hF
z#-vH)fSc&Vstq-F%nHG=vHdM+e_8UsYjBzQ-$k_9gsn3CLP-<I`Fi{%i5~wO!u76x
zU3R?(?Z3&ytz?=iQmw<}c4z2r_e^8FI-$SCY>E6WCColOcQX65^tX#|@8)k&rEdP#
z6l-U`cHhbQ+M(Dlk^BYxYwrIW{3|5zs&JCD0Ac>M>ByxGIQ6XIoMYD(-&cL?F9qqi
ziK^iwhTZodtP3tSlkURZ%f#IkNcoMjyqKHqd2kw%5C0>NM<Y<ef7!qO_s1i&*bf1a
z5&MuzEynx-qj4Mau?@P9J+IsNDf!!$|H$9IL;iLj4Y6$gfZ?*6za9Cr{O$UtWBl!N
zfMkFBdlT(9(YKlCTTJvUgEsuF3g{R6+g>Q-ME-XEF5TbeH86j>5DEP4<$K_7Kj!}S
zlNR{f&6{+8>+g+-fEt+W&^BX$484w{jNSAn&Vn-U=4alMZ^aKrM4*j)9adDdmyh)o
zTU&{(_`LPx22XY?KH(r1aKZCYDO!|ui@#)IgWK1MDe#lb!4IfLKvB)Lt-x`oE(|)$
z<8AhuztbiEU966|@4!B7E3vbUYmxSv_1&)F@SaNh<yX^r#Y=4~I9r6h=JoDu>uu3S
z3B2ig>Gn5lD!+sWQ_wE_U_X1TKWb;2u7%G@3W1>=NT7wM+o6S{&;kqWVcAt@FuueY
zn>pjTSSi*YCdS{g2SnX#Fa|i|W1O)S>a9Nj#`aPaS7>m}=UkPXOC_!^iEARbQVgzS
z&NY>DjVG>eh>NzI&M~;o;#`+<u5rZm9k?)Z-?tf-v~&YA7-PO{=!+>OFu;kw|6<k`
z#WgX{-w+lZ27KGf8M@9$;&k1-4hZw(blp&!s_;9YzSp`bV&#zNZe@k<%h!ge{L(Sn
z(}qw5{81Nvu50gnGVj39;YgspZ$fDK1&Af<+B@=@&e-`HW9-WrbKU^sIm9Tmx6NRD
zmov5@L5YKiF$RoGd(RnM&v33+IM>|}e0^WyBJI@}Tt3dVigQ&H*Qvxs+RHY$Zsc6W
zoNGC84F(sq_a1s%aD4;s?&0~Jp})V9NCN}@^5U`jd$pUiF^e+kE-Y`+-@d+5oDVh>
z*4(WKJNg?G-e72r9c?puH1)agWoYi7-R{OERsnz1{7-e2<h{yNvJeURKm3uM`c$-A
z*WJ|y<5iq7gEKw_71m!vj6!!q4aUKoF_tq<C&nvD-qiEH`<MRQJ+H9m9z}xZz5}t>
zUqW1@yDbJ+Gw0gDxmt)Tp14SNPa0g0bFSw(*GA$R1up3B^o`J6<Qvdkt)Va4B7nYB
z{QW0$f7@shm)zG_gwlKz#E`tXpYfh~X@ly&mwVd}R@w#{T4QBlA{D-xui38Um!Y)}
zQ4N|rgfDBh`*77Cwe}NTet)fJ@>`Dt@>__vIP2-#6uRE-G#INn<Aa>>TWF;|ml%cK
zZZsHYamFIf_!u$XM6xEmjWM`JajtaEbsBL^AuiI}K!eN9xtyHKL0nnHMS5%hSZ_m!
zhN!^MBS_GOQ;2IOxS+S}M(Azg>(HCwFGXK5_AAVErf+%Xh>9rxge;6Y=>Nu3eEcb^
z_*44PQx@^397d-O?tYEO`_#Z)q!y!mHlD9&ChQd#eg$s<<7OXiUC}88Hu%cxReJ4~
zm=S-bw?ee6*m!z%E|EfME()co)eT;}$TSS^yWzJdHV416VoUM+k=QESh|E*$ZAN7N
z61#>YFT}3nnb%?)kU^vxi&2z_m~M4!8zCU@mX!n9ct#@l!D~&SHWwg|wmrBUgAX08
zSh0z!I17yw=NZ5EkEWLz0+SpbKMTeOcjE*bdj;$PV#Va;0O?X|hE~pC%ia{cg_llo
z<wT^3NL@%>NL8eGNv%T!>u7c&#fNCW1urLOVqYQ#FMxQ*y0wF<KRdeJ&eyXUcn5><
z3zL1zu?y6Z;s4ddz-t#$C2-6*_|tlJBK9Ls#247c!tM&46N%Q}*rE0quRpV>q{m-e
zX|xR|d?`}emo=@q*+nnQ;EkI--m{baW0GmRNjr>lH#oF%i1>DXNg~*kq0O&!Yk#O>
zHRv?f>|+0op9eiweowKR@d}+TzZb~<nEXCMlpJz8uKbdl(^>7@&FPhRFwE&OUI9a_
zBL11MrM<r_-|%`oQ}vH$NBQhEv?a{qM+e;+9=$Ne<DVFV7u7v|TW})Y=^aEbtz)a{
zFfAK~hVBNt)<J=MT1VJxmVkEqZ+Ck9)0{Yp?&(YC-q~H9Y=es5l|^@TvhY8<D-|@3
zykduJR-8v`@buk5ciF4Y0uGw){DCnN;0~las<(Lall`NUF~QF8-%0)w!4tgp8=Qdy
zyY$Lu;J3mS3ouDaOW4E3c{gJ6Zhtz&iPLk)z#da$6On;IqR$}%i%gAGlW_owR;1Y4
zb^C6wR4PtUeYaKOeAt`xq@;K3HMEoLN&3)U(}0ZY#fXfg9rl_RkQcSWBYBO;IyT}!
zFe(KrqB338d%Wi+`$xtm`?24Ox4XF=brI)7jI${N@77jkh?mCQ+9R<maUYKaF#{Ga
z)~9L>Z0C|hQ~f3F<oKiB_^0kGU;hX5mG2;duY4DqZS@~>FE#>xDR^!jHemLIJgkRC
ztu$`Cli&6`e%pf>A?w@tZ9l$>+rEC8oL?2{qmZd6p_gEyyg&AAU--I!Z;1mF$qcZ=
zvOvXvx`rs<ZBArI@@yQk!;odYJU!k&8jk*>b!fv_+kSh^D7xaGjswzAmVNE?L>1ep
zHRlm(Uz_f9C#9?Q8hS+yX3KSk#SL`kErWK*PK4Rq+TC8_cA4xYktSuV@7;)f20ZQL
zjHMSu0mW7%AQGCMuqXV$ihIL$@G4G`!48;pi5tA{8kXQjhz<UP*};8`6R%MBU<AU0
z-AH5Bw)b>xzx#Kl?KULP_Kye#*7w4h3^sl~XK+2kxnAL1LE^H33zm1@D^PB~m*nx2
z>vuHlit9J?W~Iv;8xzD0P~9d6wwk*6>##qeog|Oeg!k^bD}j5VH7tA#eo>fjJR*G4
z9pqlza4q(=8?bldMB11EWyaC#JonN|n<V};|HWuf>)(1Slg8TG>^05MF?<DWJYuJ8
z==E#acK8$F?QpN14i4;hC*i1K=pk_PZAN-Uh+-{dKzPz-_RbKwD-?viP%@t)D!OL3
zz2<pjd}DD2_FcLGEBGqBpN{Tdjh0J32ZIET9Alt%1lv9*)i#IyPRH}Q9Sy}-1GpVc
z5I$#ua5)o%$C;p}X(d_7TFDFwWgzf5VK!wlk;zGJo-hxUP>@o%6JJUp>0^7s(cLpW
z3OGaiHbZ-RT?AZMf2t<Ew(UqbNRGvY`r$PWY&%ekwB}STCrc~9JNE_MAB9qkN1@UE
z6|b}VfXB+D68C<_jJdrRq(I|dhW2t5ZpQG(omD4HB*qv5Z`{Z$2gk}b{6n+3-*Ify
z4fs?j68bz33G{sp#8)2=OVekaU$^RvU&0i?cn}E~lQx3!3bJ&Xbv7DYZ*VTM8*sgd
zUR|F+Tr}(ay}|Vm=X#cN4WY7z6Bo@oD-5o3&Q;C1rV`iLe4QRbhY7y_Jf9C{-W<XD
z#{rBoJ@t>C{Ud#UB~xor{gW{QPjuL8=))6wx0#G%^%G<4H57xVE~1W|<h0iuAvyae
z<9Pj~ID5^vOzK@^rkF409y2TkQ$J;z3fpEDys9xKF2tu2PTI}U*|YX(-TZ42^RH|u
z=7Sy?$F_g5M`FaIJB~Nv{h4u|?1+aKrXGN5K7L0xh8=%l#t=dRHGhpkt9~97h$>8e
z5voc0J9A#tZ}su&PQ#xaM<4&6nGdNrV%=>%Gy_I#=za;l=y1}yr_+Z2o%zr|KGv;+
z$~>NRXpHk@t)r*;(Au|jQ~B$&%v9DRfvLPf@fy4pO{Q|;KVT}Su9fFQmBwug`E7Uc
z+iEa!)<3{+`xgB^_~GAv=6vY?et*yU$fWam=pBa6@5C2XPFm-5dg5gI`%auT<TG5l
z<9Hf<nQ@-1Mn!-BW~;8m;L}Wrhmk;uhcH;wzrwX~6S{iP{nz94_mf!foxy5H>%C~S
z!+O29??cw!|7^YY0=rpS?`<LzV{^_B_6Sa9y+^e>;d<{R+B=L@h1%Qqxb2PjCGFkx
z0SQZL@B7c_?LEBr<l0NM`){;2iB*N#OJ}03^*fO9{fF!M_y1{X?`|?$rr*9N*<PyM
z&uB09_1PJjqdbN|@9(f(drZyP#=Ws^pTw$0?WdFA$7w&~Gqs=QG{%Gu?_Y-P%lEA_
z{CvEM9<}08ypTp;Sxfcb9-Zv#id>3!g>a-s4V1?u2d?j;Z(`7UHTBQpz3dTGO?>1`
zJCv#&!R&!9)Z#oGrm3hXO&6^zTEO^@t*BF~sJ{PC=O3IZThC8Ib@c;gIKPk{gYyd%
z6+;Yyoh_lsv$m5@oulvw)MEhZp*Fi%h(*;_R}tyk<eWvzu!);>|D-;}{L@S%@J|D`
z!9P_FWrNK6zra5|^R(oj^!hj>DsrhlzT4v>DrT}(Yeglz0KzT^&FQ*iFJ;k6fHE<Z
zp)i~{97l7m8{cE~q(l5x^?ag9ujiU4Sv~)b1oh1L6YA+lJ=w%}sll~`bNM+}3SMxm
zzn8bB<2khElRu+1;dpC!{fAQ^IDy0?ND~n!#MD-6ks?SfQUr-&KUzyTg2bW${_>a%
zEcRL4$%r7axYH3HK{}Q0VsWSaEHHGGTJ68hnc=?)rii!`iy%?7ip8BMdW2}z4sh|f
z)A#JI7SJASy=M6HdEBXu$DRH_IFCEwOJ4|U5#UCA(4DlU;tV%V&kk`X{ZL`gNZM4<
zk3Odrm7yJIzk!0Hn0`Yq;hAvrHa1egfaskH@LI-x9%Y;-J2Hy7&H<`k^&GyK+PG0y
zZ{y=kz1xvMy;ZQr`X{;K#-PUqFZdG_SKG6`CPQBnf2O|NEKWG>XX=Z`31|IGeepP9
z?#b!PL;CV0eNZvPlk_?1tL;SkIvr<l`Ao)f0)6dympFfkz6Nj5^)+w})7J$^ps%l<
zfWF?~`nn4qBUttn=_~ASE#uWQS)H2E|1(FR`qq9ko-vyIX4q%4^OKT&X#WiRcXN#V
zOPn%;tJVs#PWaSU7^~^2@S<{1Gwe^S;^kgeiZq|7CTnNo>obVgpJ@Jsag_<@gyZXV
zr~qrY(PDkaWB=RKWT^&`Mh$*-5;aKHinAzcPbxaL7C1RipKW!W%>G0nt4OoT-Y%js
z!HpBt)>^-kv76){);bI4oml<{-ZHh`M%K&Z&%dv6{QU~Kvgw(}wx8R>O&d?f-V0e(
zMEfbYeuDNh_Fvq7IyNe6-}VQi{n&muk^S-Z^UD6>_Oq&p_ESjt1np<+zqtK4QQE!z
zj~VU9{>90)pI7!5x1Uu-w4cJGCul!o|HbX!Vs1ZXyTbmlop5sP=av1%?PpaH?Wa)N
z3EI!te{uV9%&oirA2HgGeT$Q8Kd<aBZa=GvXg>v$PSAeF{)^k+Y;OPWjrO;lZ2Ng-
ze{uU+RYdzK^m2msGxlHH{>|q0KWwz0wi!;aU(Nj=uk0^wKdXvpKMRT+fBa+YzqtK4
zF52DxA2QmHO_7spKd<aBZa=GvXg>>*9KZdH{TH_%$J@HMf3?wmY&D!*`*~%L+kR>O
z8D3Aa<-6G0Ee&ry(C$Zg(KVITkT%@$g57b$k#UUkWGC?mN8THrA7TAYX9F(9)YzCI
z(n9=w93jJ_u|s5Dh_hh9IWOt6{<Bs=Zh@hfBEhVGH=X}naT*I2-1i7({YzKV@;dxH
ztNHLeJg=KRyua1;@esoQC(;MaM5(UUGcmE>&(<H<tkPd0!An{AHf_FqXr>|fA)On-
z8_(CGM*JlZu)An!fI|dWknHxJC3dJhIC{=b(6AlvKj1!k+Z-o?aD=A&2^Rj1UBp>-
zy`YP(?S3Y@eMlg>H~tLKi7mP5zlZ3Oe@EZj#ruXvT@4NCbNI{gz;05z`axZ4x!83$
zk<@tG@K*G|bT*f#)*VlRsfzS{MI%0f*AD5>5kD@#ESL=Ffkvtt1-}s5W-)_fP91$s
ze77(@7;j0ByRY)9iq;e1{>0wgq)*I^9ghviUt!@8?qL3jPVCD&5wKDFcIbz95U|0U
zzZ66n66!>qC}4xHzENl-L{;>c<AurF5j*Qp`*!je$_Bzg<3<<+7#jevjfb)6iT1lb
zC#u`on$1Aap~1ofUn~GdK{?uBr?*97lKtt?2(IAZ$COsKpMjWR7Z1|q;9wqKwqx0R
z+#WxgaUQS7#~CZ48#hvn>Chn-WO4Bz%T3SgHaNSM+28^su)zrrzy|+??%RF|j@69j
z%n1hb6`a|{nG=ZlQ{LDAh;AONUDeffjDE2BJ06?ApQ1(jhHyHzfX8DCcngWo*aD_g
z!#Erh?;blHqJ57j{S=ITa_K&ROu9IBLGJ|nry-^9J>akaa(HhyEM1RTx{+__(nXbz
zE8QLIiSuNo3qjq&1r_~wU8=qBWl}vI38Z=+I(Yp-k}7S<?^>sGwf%<oe<a}AbsxA6
zadA#Z*AFH?fd1d}{_Zb2kL11wkNv-S9%=RKOw!@K`lrcm*xLGSipihsc_jMa$_dXS
zUBVy6&LfE%EY2fEFwPU7M@n9+E9ja4Q&1)nD5wbWk@|sn22<l%2G<#!>(`uX25}7n
z7kt~j4?>Mg?!yRZoag;z=aCTGr!z=!dhEQI_5r0hhBTF4_@3^dU-Sa_^f)?LcF!fu
zCw^~#DK&Nm8|Il8syKzYEPHhq#QXOM^cq`+){HYZ6#4a!fN!d9V6P5)r|QE@GY-wc
zF@+Czen^(f6#63q_I#MZX+&iBX+&gP>@*@W{4^r_doJy#CF~)ezetovXL<gDTCjsc
zsvj;g8P&;3!^xl-X*fPOI~g522OocpoyYVW16~~Hu_@kjx}P|s13idWHw4c?$$SKc
zp6SqG6>g#PXDgBN!-w*D9_&KU8deD%!GQ|&&Gf(Os!#PZ)nA7Ms^5X&Nc}in^-hE9
zT+Vd`=lY4b#0b6$oj17rUN-rW=Rf-Snpk?E@7VKanKWkkC*zG#obd8bR&Q1^gPlJa
zytT(%kH`$}kgwCxQ|aE<!q-?lG}Cnb5!3aa{SDjir|liY$YFB$n@9NNbTj)a@7ex7
zu-6Qw{f!p#;`~?-{^UygK)yB9huNTaZ+P+>KGGPRQP0dRBk(u<{404Zt=Xe}0~Oy*
zTMQ@RLezktZcnaSPl@8i-?SJt20<RYT+ItgMnTN1=p#<F-G7Z?RNW7>w$#C>VsU#k
zza3vEL_H!+w^uaM#Lv>~<cSO{?O)I8miF;VW@+t6U}<B})cOaw`^bdn2#yVq4VnB$
zKCFkIXFQ2`dS6r@GqG^*on(Z1uEs_I9m+cs60nZAw!F;sVRbNEwr{Er-XRDtCUhU-
zV})$?=zmyb)B|@5r5=GWUqmRMyAkU%_E5lm2-0AS6<^=RJu7ez&HJhH$Ie>$-y!H0
ze}(tNiCj-$=!BZyikppqi<yhuhVOY`;|-fec<<Z3w%HRnXA@X5a&7-$M>03C&-=0?
znH_F?^?xw?_!n*W5oxa(fl{yq7ej6jhd*09_)}3e=Q3~-`D6SEfD|0f>`xE#C&&8-
z;^CM~Ucfl6#bXyIf@=f5==1=49{ZvbKLUzxM>XSlhtNiLOd6y<T}{TZ_L+1Hv@;{=
z2Yb!?U{U?o(h1a$JxMe1Jes|NYDSeBk*37aL(o3@pB#s?rS_U8V)5Tn>AyCbtz-Dx
z5D;<jy%$YZvq_@^IN0Q-cOs&&eL;qE7Vft;oO@XrmP>VXgMa>6?*{Kd;8;iQMuKi&
z!^7%#lEbC5j%y9BzjChioa+GkO#Lq1Eo%nXU7Tw*=h{tNU+^yV6M9oHsL^q!YE&KW
z!FYdyoj=&doSZng?ew8a%uMKeZHA4o?**Yd4PArlp4|S=GoA5IKr<WJ$B*rj+hJS6
zOww#A0ULphk<Rda6BrJAi4JcW0NCnDngpHw9ygH*3N_)4TCyejsvD_^25@{NjJ>kL
z4n;sy>|8`J_63u=hN9Jmx~5~#;Kb@``$rG`oCYalEkWl5F8GtKGG{eYSppKM?Ap5^
z=A*hYe*B}(b%dlC7<vllx`(*BxH3xbfih;T?B?IP>4QEDLxYacN9_t-AGC9B=!5rJ
zJi>7Y>0=1j2m7j+ag3V!>_<W$I8_bN(_0Yeuh7Rd(nm`MzW-A55(Ak&wtJGMK_5>M
z(oG-QR&=9Qc5<K8M{4oV83$Q4>>vpHR8i2619siK@6tOZ#k`JS{)rVcft6wEN5?2e
zdrBAiT@VE1v6|!&h^j+3s6QQZCpK@*HMr(*t}@QmAJwXl0T+7rhj&9U8}5W+B>#zX
zC-h+xNgqm;u8-4YeN1CM6wj~1bx-cuTp!(i+6L%j7o;bA+6LWykv?XTK3X&IMYoza
z!uo(un+1Ka9&@Zdwi!Bk-p~pCO)BG~PNL6n#n8S@aQ`}!V!nWd8a*Z!y9_5*%->iU
z$0|lTxq}|l@r16Jy(F=~&@K`T)`fUreS)r-tp-;M=lY0q)e_fmu9&Ogxq`!L*x-Mx
z|AKzD%lgT=UDppkFfYcqS-O7OaNU!871z&^@ECW<owO19*(>xzq`jsM6z?y$k$!e$
zBz>-TBGzAyc#>{|e(H(sSp97G;JpI-+QYbX1b<}IUh^{ec-NuNe;H~*4i>=u8`l-h
z&x7e}O}ZKbr()=;5<3DX)>R}c<3zejeO%Yo_~lGj(~v+{<>)2#8M>~{HMq{^T$gdK
zC{(L{GS}7pcR^RneB|Av`J2hVy#;-dBcmt=oqJQoi(mm8KC;t*&k{D!D?|_v_wrZ&
z%_N7(<PzD4T#ju4dXaD<&T}G|HzD(E2#$tkgerS@+3&z#^@Y*|tdsswMe4vYBeSy~
zwMG}m2r{O?&{VP+yqiw3AF<SV296@Q>JI+?K)3i0-U59)gI=N0o9R!-5XQn`St4LV
z2f^qYi|j;L9LsPKOg4+a!Tsr2RJ@x%okYEbiJ%la10;f*FyZ;qjki%Uw30-K!-fBN
z9gd5H{v2SmVj-2!{$e4OL}LhKuPmMW2VE>@FJof43<<>Y1I2KJSnfwR4K7<@(tkJk
z#9`H)fps;v0YCo4q{;?)xl0%)<?x5|*io;5EnYW`5jZ->zHJx02o>)pl~H7uOe!~E
zcYvglN2}}{blo;uX&1P)DNH`tZ0jx`OOt7Q&O$hHtP$=m;xSz~UAHmeIF>Tue1<5H
zE}V*L2q&is{Z^WPIk_O{@30n7Ounb@@)^bvrwd^;34wVDCo0%Y2=QdIObFLudm$_Y
z5&&LP%8Jc|g=D$4>k+k>;({a$D-kjN9y-pe&ss>KyU+je5nTvJmM|fl0$IR9K0-)P
z7eXnVLvYq|vJhi^*Udgu-w~VL_ab^lvi6SeF!I}xPsJY{^>u>SOBWbCAd*-<OikKv
zul_aVlfJW8Pedjy>0NvE%?#OVufB^R@8DCf$fPE{oeFo`-(LM1D7SV1^DNdoT@=fQ
zKVt9lW#F&dlG#y2C^5MH8Io|58-L_jEY07BZXsg+=aRmoSU#)_uN$V}H8FZERZGKr
zVySp7RZDYG2GO82d}3wPH{?gO@A!Q$<Gy5EbfjvBwbXm?wiZr@;+siY>ME9ZYJ0WR
z7g#n9-yiF*eXgax#>jZE<G3Q81YVYLfs<w6XkK&RLx2~-!E9>0eR-N79E^xhZcfWQ
zhZ+Wd;8MKjd$bU(fRg<cO0xgDXo@H$BBDg+cw3&<Ri1`Dia%E2t~6~K!c-MAlC=d0
zQr(#2!QYJb_3#1ze?P37!S|KS3~Y;;8AQNe)E~w`5z2)U(P`c?7`Jl9PmsV~euRAM
zj}YUPobgG6@o~=hJZC&aj6V?LK+d?tV65bf8fV;3j6V`%kect0y4GNv!WnZo;~ru>
zN{oCO78PeOUd$QCaK`tDv5Odg%S-HfNN@E|i&(2q<&5>js36?J-gEoN;QEkr1(Bf1
zPY_oGanaiE9|qUoIoAfx<tMI4;-a8>z~J(8u17eRm$-U?3$xk?xSQa&w=sv@Lw`eL
zjWz@MW)>Ywmc9<K`A=l`2wprV%>DzYk<UMPEa~6aGB}C&MI@@j;ujRiYUKCCV7Cx&
zk)}A1QUn2ghQ&b8iP94(zCm9Mr!%)_i_J!3>yhm_Uc#z|X-$6Rv0IMf9cwH95`(BP
zy}bM(o<MIb)8TEtErYEZ@s2ZX%ZPWJ*YPaz{1I8-*g^K1N7$|6ZDoC*rV1e61w+Fo
zIoJ_!n&V0PUB-7@Z}}sMmbSUiqq~mtUU@R(Jl?j99NNXU%u>EBbL;PPYb>c?)>wrE
zc2)z&T%XBp<I82Rjn>7->|b&Dus0X+H{rN^*qe(8ee!p}*_(@;Z70cs^L&78vhQjL
z=q=<_?M>CUm?XWMDBo=)!jFjCKLPufDcHZH<*3`YoL<ux;rTm(u<$(o3elvP@D>-o
z13)iX<8y=flp#Df?64tt(NoMZLyAry{H0gK$hj{Qt_^`tVm<OCOqC{O3*gLNuL669
z_9BJH-P)tINVqGL$8=25=miFRB%ce^`CDC}=X#kyuRsEUjzp)ZzXbwiUGm6-I#*{o
z<Lb+~t|hJ_F7{%mC^)kcU9zWmj1vN<*X%Vsn@<h!8RM9AX9UfZn|>HH8GnSaqJKjN
zdr1R(<0;X;&R$F;_6+R)?Z(7Fc<|n6GiC-5&-p0JI$-tL6U-R9zY+z<-hFI$8ZA4Q
zx9kceX#bP&a`kuemaT%H2`*p6XMe}^zbt+xx2LlQvrq*(dyoNt!;T}*Akm}d7qFUJ
z9X*J}ZO0xxpjj*aj4Ye8-n>fJM?o3W$6_R?^#qE@h&k(LOHk`A3nhJo=OdZo{4|bB
zNZ%OI!@EyCPLh*O6Tbj$8!NOA5Ya!8dlUzXXJGWg+uP})z3gKp>})~zqZT9C^N#QK
z!si8!Bk%S#7V^IExbl7hw;f;JNAB0<-C4@*|5hgN4=EfY<XsGp6P#J`AIN(op5IO0
zsjxvq-uVdm|6F<V_FBn%(;r07IL<J0lIOh_v!bXo#xfCGeUIi^{gKE09*wbnV54`B
zwi(R|l58^umT$=FN^DG=P*$)<sj@xT<5Q3IARlAV82+BQ)Stq~RZi+W*m(W~i0vYX
z?RW!a4C6fBKq+rVbw3YgT|a$rrvO`c+m5#Oq^-M0A7x?)Z~gxjw$Fxr+BJBW_5)60
z>yGWpov4+=*`rRrm7$4UI8Tw;Y$HYDLt~Dpk#p#;)j0+@k<y>3Rc3CK-Y25(f8fZ^
z%6@Q2$@;q^$=X|K^x+9?nzK_`_$=K(tfBO!lfyWng&{ao@p^=Nqz#oja2yWfZg?B+
z{t~zawmN+o!Fw`670V{1iAWQXx{$h%sz}v#{B;WUnWig6Ikb7c3YAXP7G$MrdH4**
zve{S&*|fAAcCL%6mNt)N`%|_63!aO?PYWLWg_YAWIiPh;AjA0n)9<-U!_%-1e6)&q
z*|J8SR;m7|^M9i|w)2XZV~a<Ef#iAkxB4rw8^?zK9|Ahp;aeD2KhCuS;Tqljm3ZO)
zZYYKOGv?!@9~uEjKdZ92eyWfLbC1&B1u*m%oED~uKbXR4ZogU6YPMI-#6sD1u*p`j
zFjadqS$j8Gdmocr`Jh@uaL>b#JVx8a^evH8t#SswC8Ev9K_~kaHDA0c>Cx6zfs{wn
zvyJAR$(wfx5(t6py}p`+K#P}y_vk`6a5ED^6z6(?xN3A!wHjRSaIVjhP+5rK=%Px+
z+$(rRNze0%X_;iGPf(4u@0nfVtJOHx$-m~Ly`8Lmzz%lEOPcQ=qMj=43in!|n`0cd
z8^|0H5D`Z@CNOiPP!X9U1xnJiMeJiQ%)V@NgmA)a3bDnZ67J>+l}Iq5l;VnTIr>bb
z&96)4vqC17?~tIqx1vMTKdwvV6@#meb8X>VD{wHh{*T~-L%F;RQaPho8b8H(N!~xP
zE5pTb4;PKmJM~%1-qj=oX=FG`%4B`=!oTZ6u-=hI<8870_+(fP=`(2dGK|T=>SJaw
zjZf~@h44`U6T&x0AcQAyg0p@DZ{Cz!(Yz}c$o)gNZ~hwa5?WJy!x~5HHCk5uiwPdz
zn8RXnYzj`^vAr2ChOdZ|eFUuL6J%4hx8cm-(%_RFDS`gisJUCzUXvW|+->+v0&EBJ
z5dGb7cP`?$_ya0@i?a`%hCJLJMO!xqRQ&CA%@&ISoIJ3LwE}&>iiLC83%;0yOY4+n
zzhnK)Uj1kGe522y!_gn8GIcn-qD%`DeFCoA&l*h^-DC5a=++{E=zOr?`ZltkQV4j7
z!Bxq*G|rX03S3_j7xmCAgKH}1%H>>cDr+ZkQF!5UgX>bxHI8$gLg585mwN)8GZ?^8
zz;5yW@OUcDtBSLaQy?V1<fd*)pM>K}ZlVT&f^Jd~g32M)1hOu}H#_&My>RX(Hw!^x
zNjV*AE<Q`%(_gpQ7uHQoN?-_-OC`im38mdiaOZaL5|VKSjQ%Kczcj?8*<Z<tPS}n-
zjybTh5V%Kma6lG1h96)VEKEtiHmHdfmybRu^!p&wZvlL24*68L)*OmDYwA2CrJ7m*
zjbCy+jmryAXy7k*={kQlkLmnXB+z*|ovl3OETPRi4X$d=^&sb}MYpc+LtLcI9E0m7
z&Q-#>#!)~(X!Cvay5Jjy$N1Yctr1-dtM3MPWAGYxM1y-<Ff{*1<NGPy`zQ0k4>3`^
zNnu-k5DM7NQdd)N8`h%sLTR}z$96FKo7Df7bQtbmC=b>mcdLtW@|K3N>yH`6F6}|a
zVm+&Or)%%jMLT646KxI>h&F~cGm>?oU1D$z<6KvAu4r1$3!(j?2tvEN;8=e}=f`M1
zCEW2oR0?Af1}p4&<Y3Y=BNbuY%!mdO$~JXatJ*8Yk%1ijc{9F|53hqi4aU8WZrK6b
zXmzzax0T-zlac$Lbi<%D?b9^vE#~6*y$+ouqy7|sEKN6R*fR582vm_!X(mXvyzPE2
z)S^_alz(Te5_X)EgYS)z5zl}TBgxU!BG~d}lAOa|R`@^Ey$yU6)zv?qzy^W_C((dt
zQKPOF6cJiz$phE~HaIJ@fEopCwA28iq9Sf2Vt^#OOIU}Es93etimiRbN-Om>AZP*!
z0bh92`raDxZPx__MF@!G_x+wbGrOAvYWqC?pU*#^Pj+VRJ#$~qIrrRi&pG!_JI(2{
zppEHIt`<!1!v&cBE1d$<Y)m&eP%kN{tqSV*FlpMD9tt@V&ATSqUNs?c-V-IT9)1n+
zd0tqp%rSX71L6}RToVz_NciJpAEV*mG<EW&7y|LlYl+Hs;I$xie)}DDp1L<O9*4KV
zUZsZzxXX?gn-x!(oCg6GWEchH#?v^naAKw&{!WFrS0S7YOsWMRK{mVy`2JC2VYmp+
z%AyJU_F(Wm9!#2IveLs}k`fo<O)j=|bATKEgAR79za;EdDJvVRTfy<r{4wrl1Z?zA
zhsF<Q<{Vsr{u&7LRn<iQK(rbk{wY%M!yNdZDfkN&e6L#*gjfL)5xs9(yYa8$C7ts~
zj|V-x;VoV&px;Q`%27%35s=}KgqB3FSp<?8Zb)2v8rFhhjpbA<8xx)+7<73rDPV-N
z+-0YGq3FY&c#<+Jf<)!Yx_BE&M0lKTNLL5e6^cMDcLGh}_26cH1JWF1PQda&x0eU&
zkPBVO2;YKumnSOa{XcSgHo`9u|9bd|`*|@L3lmr-DQ{!y?vx@ID-6F`WmDu0==h+>
zC%AwjS<`{xKY}74{zZ@{(V<iMVFtUo_<)>h@xYW8$EKiE&c2=4+o&%Y;q?Y+V&HS{
zT8MTgOtRkaW^ec{@qXd(Mj0j>k8sGx*?1%QkdNM^Dg+o#-k1o=?6%H{)q=aE@6Jgd
z*6DWmygW{5M9tmto-cn4R5O-n<kS3HBxM!zOFNe$gI^xSF&$cHf!utB!#m_=Tpqqy
zAa}R+!=4J<>G?cN?||Ihg0x<+2unCp#5GFI7+9-9uwmy(>P-E0#2StK9xFZ>^;%fx
z2B$%6yR+<BR*6R=T&KvqQjz(7A#?I1l5~hkFjrCb&;pyX2d@;$9($Eg_D7gCR;>qR
zV{~g`P`>T}Zc~6eZ~+UPPQdL1T&w{9>;NuTfU6W>Hv+y+K+Ba@<p3^FfJ+o$ycB?M
z5YTd^O>zJ)SAf?jz^w%QI{__M+W8LPc?$3X1^6rh-y~p^XfrdqIe;z&*h>LENWixU
zXt~n<G2ceddnJONow%Ua0Rp~FK+Bc(j05<T0$i^EuOr|)02E!f)`42ApdM0C#e{m7
zP}Fs2I8f6S)EotM0ioU_6t$yK4pf1Hx>!M-#^t>=(Laab7<~l8P1#F0xA()`&h^J-
zq!vDOb^#RGn|QC*8*;|--;L7k%9(*NrjEophVkw)wRe;KSl<|k`GW$GnE8*9Y=i>F
zyHg`7M80_oW;1)d`B8)#Yw>Qf)9J`XY_p5N7c)095Z-D;PwSkVgOG3WW8_pCR<r@%
z)iikKk7_EQR8z5RCKR`i0je450|uDOZAcIge&EL5UC0=y`#A#Jjl0_sTPL&N@4+(E
zEG(-7#@=8uQHRr+pa|QV+G7<q*Zk=U;hN`g0oOz@FRki=fFEIoDhFzTf?A@W$_Uk!
zP-KQ*I#81p)O8B#j2b{?5{k?))Pee`g7PS+<GEC~nBnoO!3;~Uvf_V)7eM{8YrzGU
zzE>q(ZL$%im#$NysT5z|<XQrY#-z2Pt$$sNU^gnYGu5dqKRE(+#JLf<-NuYLSupb;
zF2GFH6twdkyU*7+P}3CDtqN)&V;|4A`&{Ti{X#)aP*A7dnCSDfzeb;zU6~ZWg2^2n
z7LQJndxY9c7!j`Q2ka)8J*QC04nbe6*sM-$$&>Ix=e`B;=Q0TQp}BT{e>X|`y8;*V
zcP^K$7F}}K#zQ4CB|7#s#@k%VjT@bW=(iwZEyX0Xn=}_|NH4e>K)$1zgnTp!!34z4
zbu<B4@%TVU%J+y9>tpDz_GZp=M25W1^|6>SxPeg3wqvk-9*qf4oh6I4=<v~)^bWE{
z>Q{$k3pGxVy}JDgvY$n}NbeB4VxOvhJbLGN0fsz`zKdy^kt;k@oq;PnQ=M@uV<c?F
zhoBECt3y#;{MSn5wRUmONm#@!H|sr!ij#33k;XL{;c`-*@6vXET02eGdh1HO1#h@@
zAn`P4y3bb1`wyupm{BGv3e~RoBCtbkxy=rXFBf)r2p6!!?=bzWS_;Mn8+Q2>*kRWd
zq+`Za4t$A%zd^wd=Q`P9pY0gY(Y1h2(WlghZDU2Sw#t4YCw}WF{2%JT|35>Yj>7+V
z;a~Cp9J|B6yG-#vF64h$PVEllumguC{~P>|ptMx}pXl&EoDJWX|Jk`C@_!P)OIReC
zoKYvqE}&Zy%tU^+j#D3n)=`X1#?r|WBR@w?WTy7ZChLuCyrJ;OCDXrTq{U#2qFZge
zjxH9w7UKfEo&npdjn`EWDbYtKDS0pS$9{SYKjqI#*w-z$P&AA%H}G^~;5|T|6S49l
zTn=i~F=*__-!kwa`V2!4$BDDcUfhU*xs!wC?eT#rQ_i2l*E^3SZzS^a6>E_k{tn;W
z_D{27Wx2ZbSzoyce;IAAW)C1b?%Rs$5xw>nxXu0^!$G(Z5XS-0NXda?(M59LSagvb
z7g~~Inn;decs+iDwQla~E;}woh%DfOSwaq3zqR%qi@HhnYZGK=DOfDG7a`UQC-a4H
z69~UcH~V6}eAGCP6k+n79Wrx)-)4gmmk1m9Z~+_qh%04FK0F1{5Pj(K_VU5@$0lQ!
zJTVdRgbq@pWmWAk8X~(1PyCsViGKs1-sU2U8P-+;F&%<g5>8n9r;o}H8@L{POQ!Mf
z0tedsnW0@;?I9Fegl~2D+q+N@z8L*WCHs-IGj{EzN&e0c;9gKt1MH0(@CG6kaRcV?
zxC?Lt?m$eWVsCDE^<kl=_PXYy+W1s}lvB4%)U7NcLs75HE<u2y0i0c6W;MdIrec!e
zyLw=RTMoW&|BVi@sC=_P@XvW>mwc$h-YeVj109eG=g+p;;k*gL4j14;c7WY<l?PFj
zqAT`rpt>ulJ__nnF3>EdnFp~Eol#8L!Is<SNfck|L$r&9d<XN4`$=?Mg+E7!OUsim
zK9(FRoHL8g1yNG|Gjuz0B?%eHKEHl9+h{m@yrALdxBv~g>>$yAS@yw+=wQvI>|iDP
zXZxEw_b<aHj9EYo5ctI5w}pXLi|P790}>aqe3`*N+frweyzP4UgXsD5tx;sFG{3kQ
zatU7vqO<+#-*6o9(;r9EXdssm!m7gtn+-s)UeSdALbKxwL^2tWyP1qggP~6g(u-F3
zJY<yUtZWM!K@In@c@{U`4+u+6fg3xNS`=*SL~2P9Q&~iKJ){;sEsRRN;VbPq3sGtb
z<tN#qS~R#Vc2gT^o)NjpCXMM6(maR@NOK!Zr>o}M9D0odHBCX?s-QfCvN-g;%Rs;l
zuurM|wMp}}FT7dNOD4Mg31Za}FA2eIO&rm?DlF>Qv&JLj8llYw*z3QSj00_a;WCFT
z?PMJy_7a#$mgK<5?Kj#Oxy=xa+<^-)at&93#0cp2B?e;j-3jQcvp)FuzK;R}8sxl}
zHnjcT@OKDCxA^xW#lMrlHC+4wJMtM_D2eT60t(R?h?N8F6Vq$?xFjL^+j|tQ^e~nm
zm|t;)W)4qu7-H1CP{&mlPQy)c_<Hpz5Zq5~9z9_=>W-OYN^-CP;hM=!1!c#>S(+4T
z$uNq-=aMI1G>eSAd9KCS%K(8UQtz@DdqPqbMZtakzDiPs-%;lQ$?JFGbqoiQvIZ=L
z*y7`cnSdJxU#LWhM%Xf-KCLWkG?%6__`Voy{ABPIGi`GI`640b3%G!s_rqYiYK_g{
z3mvGSg1T2hl@iKg@Pn6t=%0+U^!FtGR>VZCs#1v(#QZIq+6jOE4Z&+ET2TWTP5Ap$
z80^0<f0K4pmW0Vkuy#z|VM4c6JS(^v$hq@+8#xiLAm{hE06DkA<|xPkDZ8@UU-*=|
zlHm8G{wLO-Ytag&%_i2LkF#ZTrO5j8ck0Pnf2t*3>L;7Zl@f0s(Cvgi%SJBp*6ebk
zkAcr4P}tGpxpaa2{*g}c8jk<#Y@FPttMQKuaIy$OXw_b8{A1`x-!+oQzt{&7n5bkw
zZ8wq}<x@5bgw68~ACp%$9e4e3*RQNZML1`t<IL~C?S0bAZpH;if?tOV&I2!{>Su8f
zyFO=*94<Bzh1>Qn7@1iYB5f$q;6dYvP&NZ}=C9Y<omp|AbY?9s=*$Zc6sxijATJ8Y
zA_r=rg1S#ZJxQpZfI`)W#-q!hjdkYdOE--1r>oW1$;mbmHghWiVKK_f<;1gdox=O8
z!v9MZK9z-^t_pvch2Md~lle=ILx9uAJU_bSr1UgzX}A37{BGX-z8lIi&_OX@ZJuHG
z@6|EVzklF@{(bUG6dt-jT0V+}pU1+J{TqyTp`9fB?ds*#UD5{ciylOr8|GzxJ@8kk
z&+HnipWmYmXA=0=V=8yPAb1tcNS8bH7OQ%vsd{fns`q7xis-W_wB32i`0yi1MRB3r
zT5Zw-Bm8OfF|@MKbh+W|tBvOZro`>Sg5`?A@5N7M>y!L5dfR#(-tYWZ<_+<g{`WIQ
z?eNEvbPP)rk4BU%c0^#4L-75~N(v;GJKn$=4&c}_Dk|mhe-&^qsct}(#&W~gk52_O
z<hE^KG=<9uKkCN2X>r2~hjY0ede47eQTRL2IIw8e!(T+>uc7U5hj0ze{Yn3XDLVR_
zJ6So;_GsJe4ID-Qx++Io*mgXc5*c$C<uzi{5Vq)Jr^Y+x=m~8b1Fa+HY%$C~vO>dr
zLijq;^Mqf3!1f2K={~|6`}~ig$}Mt=P3%k@e$Sm!NU!(bDF#xHd3~{A&MQxZ>^TSC
z4Dees)L~!&GZdWDFf&a;m5dob6BlfI(}FK%*{ZZPK~cxD%mfVU_h8SREA}X$Y{Y26
zY|K2V$iZ%{=zFjP#@>VZ$7y?Sr|o+9^XT=pHmXbrgU~)LxDmiOs;k@qi!IG$JyzFK
zUmsw<?lWQi!gtJqVR`+R6f96eLl5?*+59_a-zdI(Fj{^2pdWO9P#h~P*}Y@9;m=dO
zRL;V5^b<Jg37ptc$MV$E{RLzchQ9>&#=gd}o=0e1Eu3bG4@|^J(w80o4vxvvON))t
zUrsVgZ!a-QOQ*m;QLdX4vUO9(X66DkLr2?6sC#@+SPTh!Y2hW&uFb#-oM^O+urRbc
zTag`iILFLdw#g<uD7u!WJrV(L+`JK4YHz@s@7SnrhGzllJ7Siw_^WXD!6sLWVbs<(
ztB8X)vky2-XCBm$KUD(SYhH%>KOpzNwwk<vks`zje-Qhy+Yymbwb((1Md^Pc4kne-
zj5zb5Jdm=@KR~$F2!8=S8N4ZaF=B<Z=Mbr9C1V%#y6AvJp*H}-yQDGrXq{EEjYiN3
zC1@xib{YHy?W#aexGif_bR1TM9je%5TJ5%2%&OvaRAHF6gK`iau%%$=t3rBkSxMG7
zERbj=N?r;LUg3y5!<K)Aa1p36<*~QF#f>$9wcyKm;*M5~HWUdW3%K+jMKGlg$|=ez
zTvT*?;fA6dxDiX-1>owgY>19;dK4jv9;TvfXhyXCfPFpIrfKX1%0t13T7kkXQ7b<}
zI4#J_GLJ`@Il5VdcSmQV#3FobH>WEjYcCkmEiLv(ypd?<8SUE%ByxA=rf3f=B>(r@
zdCN`Tqn&eqifJ(17F`HMn)rjv=!3(5gbrxgmn#(!tgP(Vgo8xAlI&V=Jf2bom=WHA
z*M^a8`Ut3!L0Zs_Cm|;#2E~{Qd>HGIQe0Yx;wTSqhh;eS%7Ns1-@;Ss0Ja%`TE*NO
zrQ?u9m>H9!EM&=0BDoi0?fY0I=@WjE^M>%SFwj3^r)MPb`#DKNDq8v1PPqL&IFl@q
zl{}z+iZ&`arQ}ld>kXtb)COVZp(skR_XpsESPNnVjx@%ud$N;Sr*&TI+E!JoEtnNO
zbgULIb?Ih-5`*SX+cS8Z!`>rqV0Rh^*NZjySm4HrXU8!I?JZrhKk`_4A?a8yj7*pI
z9NfQ<_t>`#NwFi=l~EI=zbw{EM^BPLiW}&(7@LsoQ>;K5m_a1EsV6>cMD1gES1ojp
zmX?-ZglXw%+;}nL4ezFfw&GrW8gv}w{}{;saghJQSf~x*mxY($8;g1TH>@M-3NMX*
zRSzs+LZEneq(>b)h<0F3p&^h5`B{_2gcR9sBa{qw(oBU;`JMtgh5|-UK=2gi1Z+t-
zF?F3PIQr~*zQGEtxE`>EA613lm7@X-GzW)N;b;!|cxt(v4}+%CGa*A!jU0sXQ4U@w
zE-WJ4o+F$0v3QhWx4ln7k+IA@D2YwK3Jn2u^~k-Jp?w)GiL#`;2VMlBdD8rLq%RPk
zDE(RQh29B$afQ1QNZ=%gTEO4Y=+pRMQ!%;VW#NLi*jot6b<w}V^CU7n7frTx04>;!
z9e^p#?5#(79+QtI4R>u5isHbN9@OsSQ?6sOakiL=Y+q^C-6+mb5z|`mSxiy<o+}TA
zQctmJyq1tufBL4qq@rjkKs#fzkd<gj1)gwXQ8A(@hT+7q0-PqLlNZh2{E~1s>XBx}
zE`WyR!~B7Hcb66%gSX7llgzOtftId;54)O|P6_baRTHtjy+5XsW=~Mg;AqC)dooNj
z_T8hXG$1JtWXA`vJnnuT++hY+;sWC)dhttWUM}XHL_hwqm(2sG1Ftf)KG(zVM3*-L
zkjD}v`1%<D3QHt7Az4OVj;K0JVvv#C1(XztCNgrE<nr8-Co7R{eXZ4zktC))NyG|K
zQ34So@dWIdV-avVy2N9S#kfjzs2-|+d`tpb>j)j#iETC683zZbohhH<$`ZglAE3Y2
zrA61gf{tNmSi%#^PV9xR(HSQe4<ZESCo>=xABN;<!N;LvnWKvXtzERhNf2r9QH1IX
zwR;h1Yzmmd=%-Leyy0lH|2xFp234dMyiJHcW>6vMlZn$H80&=yl<&0QdZ0XTpsN-<
zC^ucSP(Q$WK~-p)7|o3Ex2c7IW>gy;xf-PU4&Z2y9%${M)f_;B!1cwLiJfIj0!2H&
zYCqN4>Jx|aoiJX}q3V3uCH6!LyZjYz+YP|LME~?~EPC6Q)`S8k67dqd6>x*=WsM@>
zNwNxfrs{5u;CG`t#jynGgT-5H9LtZKGHED|`!oB#?lb$sg16QBu9h0V!Ao6SJ#OXj
zMD)fwIVG8Cu@Z>+#PMz0G3!DE@aUs4edM#t^-SS0th#)6`ELSHG|0S6qyx6mOM;Jw
zEfpb{*`*{x@==1i93}V&;8@RYJYqso#T4KzAA$ilYN0BopbXSgq-&wscvQ^wd2H&Q
zWZ5tUPf8NYRP%Xs3MS!L5mFQM8bKI)qLj5oGQB=P9IHbw3IOc%nUG}pvpkG-0_Nx_
z)cBM1V{p+NMDD0ILOK2-wt6?4X=`Px@RZtw<dI*Znt+j?gZg9t5IJl!tPgT1&j;4+
z2ZXExwLpz+zVPcf@8@lskFHDO@@hi#kE>)L=EZ-KV0d{y=~eq3Z@2-kT@w9Oof^E5
z-Ha5IY>~Z*1!B}(Kpb&-PB~+#1bfg{P7pqhEi`H?T7+Ou`OTBxeubvb8J+&~F)aPp
z2M%CJGR_Zk6GkNH&9fP;<p&0&FT5?f;yK79nA{YA&n!a!Vb_rEuSZSq_`)Bcf>!&r
z!b=J#a?s=!UOFNAu<B1<{QTqsl3upo0s`>jrO~UNPeFiWQ>(Fv8LbtE6IQD-VVb{F
zLX7M3;EpdyWJnbp^Tp!{_{*~|Ni4z4Wkkni+590&F6-w_f0?K<#}Y4s62qM8&h$?g
zywKSM&VtT>!w)E^!k<|vdEi-~<a#>~vmg|#v7MkqO2ca@C|Rs}m=}lUtO{oVNhpze
z6gDOR_mt&ihc1H=xp<2p0ua%VxT{8&$Kt#A#a&y~3H92KA@>&2JQOL|V^F9>O6WMq
zc!WCI-}LK+dl5N0_LCqE|HN2t0I2BVuO&$fxsW13it00fy>M?ysd~P0QY;Gq_UD2N
z@ufc=P~QsiATf*1OR64FSiO`cFCkGH$3(;@^Cj)c*MWvta3x>80+K^J{(J(4MU>8F
z;30NkCpx=N@q8!?mX8>ZCN6?JDGKf5Xepn66HIOK6B71M%V2ivL}euNG-g(o`_ueU
z-JDcjXr3>6!at+Wxo8;e4#Vgc7)G1@n_!z<eW`5|ts9KFC(>&ZF0hGigNagXqOns1
zL9KdKBiBfdpI<Dh70%mp)!U|edo*1FIk2e4KA>DRGcI)iFIIq8E5LCu@2<KSruo=&
z3h*2Uu%80VSAZS@UP8c#0!()R+XhL)vlZa@VF3ICtDdR=-@e4|;hPHZGhEQahZ(i?
zC|n`2;R^6c2k<Wn@I?hU1e5ZrIWYXij#q#S9l)RhyjKC176Fj{pI8i28+v`E1Nch?
zI8y<-S@mXCU9SL#IDkJ<fTI=Qw+ISe^(1?Jp8`DA0qmgwPga223HUStZ&ZLgF1CB!
za=xHv4=(8S;{<%0fJF*$l>_*^0^FzoD+%}<oCmSK3UG-7SgQaZR)7-;xSD`pi;k3W
zjRQDM0p6+r2NLjfQs!j^c!2{rLIL^|U<LvE6YvoQ*vkPtK>@lI;QKV7uOS}G72wVZ
zcCSAlD82p`7xemR0@f37yaHVB05&MV?F#S?0;Ul?XDh(pI)L{pz{eHfuL+pWUQ>?)
zt#5V!Z&HAj3UC+!GYI&G0@NMAF$!>^0@MiDgMd#cz|$PSQxxDp1^6i<Dp&O+AY2iw
zdf#}v*L%;CUS}x4wFJx|;A92(ngjTX0(>7Apl2xodl7Ju0({f~{DT5~S^=Ky0bpMO
zc2|J6JAk(-z&jLR&S(IhN<97(x+vOsi32!60sdM6b|v6x1hjT4^mhP%tN@28!0$!@
zFpq%NPK7iF@G#d0kzSetY(^Zhd`Z{dsqoe~8$EwlfLu?Y*T=K!SuD}osqj|^aD@U~
zs{jwO>YE8@?NkUkfB^-#Q~`cKz}W<h%DkFU;s9Qu0IyeojRd@ffYwfhpE`hp72qfZ
zcn|@{t1>tgsKEe_$2fqQ3h*QaxQ&1xf-(}E{;|*Q@{V%_H(%p|E<Zu2PYA`}^cNkd
zH45rA1r;LHUP3WA{eB1PHwx-81@$XJwE_xDkgOr_UhO|0D~i)9(ay99x;ZD09v43>
zG?{wW-nyyjk?O5rSZLdGZqm*1*z`0iPY=6tV1>%bT*(~VWb<v_NTj+a;>s-CSFe>_
zp{H-tBNO`=5d<$|64X4<OJPAApQm@dFjG8Yv|sF-6uTOYpc0W(-K}@&RqLdJ$jIvo
ziH@O9;euv|&cl>zW;|!u*m~w1!B!J4z}AC+TD2A$GC<GyDSFsTdg#H@fbn;T$Am9a
zho4q_6n|#RxtXggv?)~ol<Mtft>#aXGd|OZ%!Mgx#L;?O7#X4!?AxGeA<@iKul2|*
z^txWxJYVVI<JfD%ye4yeWXK517l@r@(Q5}{6^4C?;eSRqA4*FbZMsG%m!vnpWJ6}z
zbOYNrtf*#2ZCW^T8oBV7jDT?PG^#-GbO$cz{FMU)PqkxhJT2}oczOsI;K>gt!ISXI
z3<qkuf|{eC7F+-*%gg%yU|_Qu*thtlj<{N%&}S0y0^HB4Jg=ecYcUa}Y2!cF1I^j1
z`xJmK`=YZ^NTGRLXp>g^-;mga;po_i3pya|U>8MPO?W{bGJS}AY52%BE;1s-G$V3e
zw!2$e;hI<5Pz=<q_>W+_%io}DV|Ir>4;R0XtG6}l`3JP%2$Two%Hd3N!i2V$0|!TH
zHH#qoe5J;{X}<KBm`n@XJ~RtT@}uF;HDfEH%QVm5<1d)d&j0v^R{b!li~m74D=Tqh
zC*Hxth+!^bqHZpsYE@{a>k(I_5$U=17v0l}Ors&slYfjBEG1>Qqb|~G?^)f_#%fQx
zUWN`iiV8c15PteW`rUSRx3qDkS$8B`#BN1H#Ud6vUCLO(+RWpnU4bk~s$|Wu33DmY
zY<-|Zz^sD~;v(l2RG^#(A+#|iP|jUQCj&Qzd3CnOJO**ST99@J1hgP$8h{E{;V~`t
zAbzcm8s;+9(H!W*!Li5)90agNU<Oh`8^^8Bytoe?dsx0)%w}E81<rQ#b3z|IeS1uO
z_1kJUk{Wj8c~_yLG&Cm7NXPLB2T%2K`OCir#_7Zr){B8*U1;XTB7gMO*goEvQbu4{
zx>l1W-p>MJ)Z@;KeGPpZM5#<b;o~o`&FX!;4Ik!tS0i0&^QCX|HXO+F20lDw<YSpN
z{$;fJ6}u)w8;y8?&0cNHp4f7fXciO)>JQ-XG;P%1V-GL^f<m~0_{TPBQ2qokrPs_=
z2W?~779Z~bA@`dzQ*tL{#|H`j#m1sE3xn&>Hm=3FLgb2`zj;7X7scy(KE7r%=BwC=
zd}9@`ofa4dyg!Q@4?b{H7+7~&VPHo~k@n;pzS8YjqFj>a3GZn5{4`Jcj=+w?I<}T|
zh0m45Fw(a-?)cmOb<VI+HN;+oi%(2Cu_x$H#4MnNqYW+H0`=T^p1uLAynW&8vEG9N
zVss`d#t3a6lLe`qccCq9P6gM4!_UVB(&n>^AxHjRkeEC7kFf#20rmjcO#!}0z`qgD
znmac;fUhdRf8c@=?<L@6g&2{uQ15aF>c17#^9rgPCfZdGp*Wd4s~xCH1$CE#I)KS>
z)$a+#xwFK9x<Wx+ub_4k>R~`ZxNe1TjlR&25hqT3;AMLF6PaVIE&Y?k#s4seTj`i=
zALB9B;X%g^Fh90~&(DNu;6o&s-68W!5RBiuef$SLj<L}{3a9aI$qdYegJM^pqBb{e
z!KY~C^F=oPo&|pae;aTCS{AdBvw;>g@?18uHyVkRupTbjs)y%o#DAq8E?cB5+&M){
zJmI2c_<yvyXgS9o1aW~EjxLB{xN6YYmfha|*qgPN-qw{y<c3NkTvT5OgKWj6zR3JO
zFF+ok-m|EpOY;->@HQh_074@CD+uzN=n)yb4<02Wd?0!N69DiLIs?3nFc)v283w&8
zx0bz%^I?R!eBm4M!v1J00AO|tjNFW4#k*+LkDx$v<OVz#k(++uZP?YdFz`uhv<jW1
zbA;*yOr_DBA7h~QdKh3_*~5lO<30}zuE+a>H(_4(|6B%P{P;wB^|l5a{^37W4{z2R
zzUXS?e{svs#vOb1WB%TvH+<R6+hfR~_`A`es0$UrdWX(LpFdaiWFYWB>&Ii~w6z81
z)pw~F36BmoN49j;BYg`B*YrWf?eH_!9C0=#!;G_431d)#H_!+R>PKZC<K>NQl}j+e
zdT{i3WZK(FOZdQ!LxVSQt8-fXq?CA}K!Z!iry7E%fR3uo=){+^Pvp-){&?|o%iHnk
z*7tG38w)F2Ml@c%YX1grc$2pw)@^(b*Q<otUa0a4D>tThmj8Srze^|iT3{BM5Z+Qy
z`Pqnq$}L?RcQouj?8Wzv8lv6C_88F^yO=FBdww+vbTs>io1?SbNQewQbr_hK!2s*P
zzBBv7in3gUe8xsV>gF(jY{(OKe+w6|`_uiv?n?@V-A4dE`s1@$a>o4*{BIQeV+uY%
z_&+N6EW+<+#n}Jrz)x53a}<0I;jdNj?}PBsW(7aYf&ZC;zfi${_G7?br{Gr*{yqht
z>A-hW@FyzxR|$WEg0CR_HGsGDs}y}|7{b_~wPNzg)Xf>02%Vw2G{_wK-0!=m1rB$S
z4+Ut|<ZoCF88|v9;u!>yzJDLK4{$&3KJK|~&>KEH)`+;Ppi5ECnIq@oeIv2}W9dtb
zrO#TULxDja{G`Ux!D!Q&ZEd5?;8|&T7^@gK!^TDypOprF4zxlIsK)>uynT7P|8&!J
zmy{m^(;c|wc+n#ocO2Ni3Ezvv*vZ-u{g$}k$T@6;zlzR7cPRrtM+qG2H(<=tr<z&#
zBDe$2iJ9(fn9CC5vj}2|`MMdR9?7;~iLy2i_ibAxCy3%sKyeHT+1;kQwb1>*2R-K5
zNQLhXkv~i@5jO(t<>3bF5jgM)1BY=ozQmG1*(==)`{>J=Jn4G^IPRH;PDra9@e}%o
zJ&|01BhG@6Ez#j$tEvjyg6@?l7sp-Q6m-|&uIa$H1vpH5?gI6wx`jZy;e)jZKbYey
z-4`Fi>OAI@ent6v{a<^ApRO(D=|-_}_>p)*m<v^ZOVhpi8??x)D0rMXbx-CPv*OuY
zzTI@-%lf8+3ku-VLQoUW`bK+uHa^Jq;voMM*m1nXA;tEZUr}Xi1J$aK$V^maXWqz6
ztAuqaB@lI130?rP;_uO8n?_m?%S;f~q7z6I1Ve$xr66{2P>qgh7q|%TsnPot@w9bN
z83>s-##`D5dwN5r=?Uf9?FD-ybW!M_)1!o_0T2?$_2H%;{6vvApe3Qb(Ffsrh&AH@
z`)Yf#g57Ne{p#3dtf_`23Doe0z^8`?Z-0qdn1t>9H+o=Q7ieBu@KluH0ekzx4G@xD
zePHjNlzLkz&&GqZjfjhfIzfDIbd51x=}CaAQvl;nK<XVDNvW3)sW<mDU?wuOp9<ws
zQtws_lGsk@VCwraSg%q)&}#2vyTjk&d=Dcs@{yA${m%NYle?wGe`>cEt3wQO7}fK)
z>Dq<c(0ok~?AVPdrb`~C!>;6~(v9By&GW`xWG?+Wvl0bjc7(lo^+pzMVo73Nq(e>I
zi2qU47p>(AS07Kh-tv|XS<3Qze8sb{%I*~5vMzpS@)9*^ydwS<fw27w*lzd_5ySRb
z^ldoVbQr}YTpR<qhqd??CqiHY8vN;QuX&svsrU(;dRw%brDz#GJqyDFkLk%e5iZ;l
zGclr1w2clY_91^phU9YvUN%fmf6E)WyidyN`vPxxexA4MIIsxC2g7(ih>m06LRBN1
zhfrfg*knu{Vb3vF1Ful2Zovg}%tf$K+ZKhl&wz|xe=1&0|4NUH+$#3mp?v^>2`k<=
zMITJ)g3^7Dd5x<m^zr<25$`VJd%U^ki_A1w_aiTDMHgrU=#SzlV!drmzWkkK|0pug
z<|}96mFY{yn-yOmbk_ksg!LhSkF|hX5DJO!lsJ67zQg=sYFLg;Karj4)rL!G)BG7l
zgWtnO@Aw%Qk6d)Z(~9qFUD&J#>Ol+We(8<gw)IB-JO09p%;NVTd&wko$t10vGRd}7
zcW&9A#eW-LF1yLO%_BUO`~5PW{4WlIK`$MHo--c%G_WX!LR`&`UvhQ~?_V|ZhHI_?
zVy{L|F=2h6GCxw8P{3r4LH;P@n`2!d*%l`c$>%cpSfsv`oZ1VisOfL2uxIV85Ani5
zG{R|7go6`74H$0SPAASzIY`(@FE&n(ocFOjo%fDh&e|-AL$8-huRqA8=Q7F5cu+1q
zf5%IBKj#P9qTq_XtPo2Dlo&xy;U&1Kv~O;ao9MN~N}zE<;C2^k9(l!2I9H$dH7;81
zbL5%Gd1pgq0&+pWi_G9h<R>%JV;0~zy4)?MJ^_Bm*V38}WfvgaN@t{_K1?@f>AG1b
zV?IMTFR&9vF^{`IzGT55i9yiz;iHt(!_j2KZ>Y~bydyYioM2ZKD4&p~g<i-)O9FM3
z?3fmMT|F&QPjNhX+Zqux=JMt@YKtdA_B$v(hXYAa-0KQ-7*xrWOD|n6J%4!`5uPzw
zE<G>f1qU+KgA{(;7hZ~+>2f2;o5)*Y@gn<Xq<u5UzB$vr>22Tiux}2X%Gaa&q)m-e
z0>gphd@kN1XRZAi#`muHIjQS2ryHfGpc`#XdTCcZze$U@xX3d-@k|mz2O+YyzzhWk
z``?8S*`S7wA+V$)0d&q`1;Ba|?lEzLgGZs2iUgOd_Dzr)Y<ma9c&XXNV_wACn$1zH
zt%<dbVr>_pwrZ)ZYF&cf#coswqIXu^Tfl{??lblc9B5ZZvU1k=U#)J2Q(bwx>X!Tv
z)g6F$RQ0Y$yXr(;{NL`~9Zq$>#RQRz%w<1B?|$S|cW!ERmCL0%C}5+$gJAsMW6sC>
z+W1`cBh|rJ3ltb<e2hLBsga=z0cMAlL_atQY0zte<1!&2!W(e5F@?RO<c)TLH}lM{
z`0WeD;YC`JWyOxO?xS~;Y!RH*oSJ;Qm7JHFJk?6>o0{yik_Wwr79`648Imb(wc37A
zEM(gH#9&?Vi`uQ<kDzaP&9xg*pe)eY*iS%(BncSV1~qe>xk5yBAyy*nC$1CNQ(QA~
zbthWNZsxI=-=mn+WPbD{TGJsJyD0ZWXR%TVZ|4cLMT0ofvlAVan4~??i}-R<@1Z~J
zi4Ik-_R+)gDi)rdy{dKg>OqhnwLwvnS23S-_Uape>QL)T>eYVMt0I~@Hu!<gUJdfq
z$XZG^LHg<f+_5LuOUArnm6~UzUS`#Vs3OsIq1s~m0DfqEEdnX9ricvv#c`OAp!B|p
z``3A2I^;JA-4_P+?`SW@;(A@|5N>6*7?weX!4mhAIE;d6U};Qt$wcEaY)|wJ3<ob9
z6{TJI#o9<0yd|5-$C)#~#iH|jbHrkl(+NkEbx0PDxQey@OO6=EmlGU8zK7k8tv*$~
z|1a306~j8|ef8=`Nc`7qv9_~Um#bI*1zS{i_Ua9MwKKLTmW=jnae>qnssxu{ba>6b
zEW(}7T+12AXa0#ZgU<{tlgnS0;{xerfWP}`!H@6+^%Ls!se`vC>=pX(;Oo@K{3G->
zF3QTjFn@~h)qPs+S<nz5J%)Q;&-9-pnQxZuiD^hpw&zu?OUp*!Q*N>};0L-P|N2mP
zB6`_$YR-ixZ1{@xQr}pf-kRTl`c4oH8Gf7Ery$G**HeNy_G*3j%=J<~fY+h^I{}iM
zpCQ#_E8+RdUf6-3*_7^9Q$l56^$rji3YMt(;6(M6x$6|7dJV2$Vhmr?vAtZ(vAtb7
zw$~V5k<#7|AV@oHZ_gvN*J?_!H05LeO{ys;11+*1GloCfu|1%Pv}ZOdYllB$_|%m4
z@Z9SEq&<zjvASU#GeZaQ?Y@2fGS#v?sW6FN&va}b=SAY{EnM5RPalqLt>iC@zRv*U
zKWU#E?UPvfbH9~k7Wlk3ssjBiejA>|ZxCi^&-}4M-xU6e{~VQfR)6f>gEZLS8&b(u
znb;{g4q7rc4X~HyH{^1i(5o1iqHrs>Zlt_4>;#MqGvfgVK3BoJae?EZ1AL}}zt(}j
z<`^lz92cxSjz0lQ|GuLv_t9tvs!%~)qM&jJbqb;AK03vL`jLVfprEn{bt<5+(7fw-
zm=P+DlMTxbeUsMfRXe##q`3lTl>8khNLbMl74Cj8=lm2eVp_Tz1wuGSmu6EMK$R-y
z?o7)IBHwz7<|B(<iPPaX#^xE^;&|BY(KGl)Bvb%~<rG~&3#pC@(>B0@$h8aXh@85y
zT;u|wHOJ+So(g9Y_J2_9X~E-Ala8Il1^nJt0c^T2LQ*Lmioxwq4I?ZLHLTl+WSM>u
zFKL87j7C9JJwK|2c1eKZXVIT)jKTg3;SdI#R&{_{pceWL*@c05F4!u2qpl*e3^SuI
zYgEKFnC^D?Rl0e*V)q_QwEaBc*@*04?50o-<HXDM9ZK)KLq2$J#Qdm-)1_1OsM8y0
z>f__i+?gEjyLc?d6dp%cWX^N(KwwN`Sw6EYS8BmSUuY3H(--HqGet+r0IzxF5I$h(
zT{HswC_Q|b#5Q0VHVg~A8G<z8>5iY1EDbp3azbN6MRt+FU*Alu$SSGL;JRoxu1uBC
z5+h=#7ZiJ|npPms2rnn@aJmsn_plj-(AXvy@S&d({xuY-#Zk@*x>?u{p5YcKHH9?o
z71D2Bm}mCDXu2c=OfL}%kzSwjUWFuBpEW4!21REN&llJC@8MY;Q@Gz{2A1<)?ZU={
zr3m!ynb)5KHD3?A+~Cj!IVdaAYiD-1v~WzVxb6eJu~zoxf1!o`c$gFryrgX?!1m6-
z8X;nu7Mu;aHh5Q&ISpJ<<|-1*UzX)H=K=FrA_MDvedd*dc^#481AOMZA)Jk|@bv-d
zWgZ?5G|5{UB95?I18A;r&+@?dRH22sg0@iXk-O5$)q?!8N~A7rcU#*?PJ7(Bf&h)6
z2wdpk-)8EX(W;x>^>EO=3~%9O%2O7t1-twkFH@1g%QC*@<rXA--<p@Gm)GE;YJTy1
zz8s_ZxSjm{mz5gkkCSl}dLQ(H1nmR^p~FsA`~HPs!m{3K-}=xfCu@=FSUoQ6n1StS
zGpH0#DeD32eSevsliF*LJYI`D-ge}vL?OP&Ji{A#XfbXhV?IInh6ijjV$J|CWm6g%
z+_g|o-`&<?vEMMdM_$D?y4YiQ!}7nWyzn#P+riKU=T~^}8F7j{j^0-hU8b9hB|EH7
zh~Lx>zjoQ;YCi-Sic0v+RQDpjcdVYih$A;X%(mYj34U)4g8Gl2zQoRFTc<2;Yol1`
z&Y$slwxmc%j_j_uFDRyfgG`{k2}2e)u>KkDDYz+s$k2`OvP!gpy;#O;co7M~N0Jeb
zBpDy&8eYUyO9(MlfWwQHSwNM%h8HcDM{aHidty*(bjqDQ2#L|2)a8egN6*tE8Sj!5
z+LKvZaerev$ueXK4?XU&8$QE4%%$7hGVhn=Di3BnH(7s>eFAWKP#TSZIBZ0{aFjtP
zZLo!suH;bvo+5{)W+mj1cC5%DJ%5FW94)j2$_+8x4^9CNPREWeG4r}=2-s)+u~Jqa
zlqCv9wCNa-2eL22?ZEUzCgk2;fPzmRCBPW~=;{Dy3gC4Bs6r|M!-=|ICU}slG{E0D
z(^h)+t2!T5<rM?g@}Os|Oo54GOMAvcPSQg4+70TpQ}`Mdad5o_(LeiqISE#&+Sv*b
z2KQ3a0c?EyE+^>Tif6g0w{Mo&H<k9y6#J&YzR9z1((IcSRHVwWZ<g6NmFfmNcmY5?
zgTHzB<=ipy!85vHPyU?fr#+x_{uBwnQQcVBRaV=u%I^FT&ET)06mg23S1+^8dH2gX
zEhEbS9=|b#U(Cz1byMv-_L*6SJm&9L;^oE<Uq5FdDc;y{={fH;eVcvO3wtxtn!dfN
z5&y3+cWYDpsyFce9<(h_c;8BwoKo>!7f>nCm@RHKEaf-ME4s+^e+`CTSdY>m#=6Mr
zOV@lG&G{4b^sGab^Zhu4@pCP7Gbzk-O}*IHhkdgQd(b@FVv~_>-U3I)$Q;$%`%ww4
zQS(O`k)g*T4fX|_elWs+t7dk@b|K8oEWqsNYJ?{32_FXEmEc{mGtl(jT!GJi?$aZu
zd<bncp2PDVy@3sx-m1g9K`Q^*7n!ah(1QG*k2Qy!%}&>&_ZfftIs;ts!OHZghQi{n
zp<&0v7lI)|en||GSONGpNq-IO#OaN1h}(~v>T6Vl!*4}D&p@}SKYl{506#7@vI1E=
zfki2gP==z<g5b!Q#SCoANLc0`R2kPXBRe@GfQ&+VN$FfF&}xsv+7WJ&HQ*0e?FZTT
zdXZ}ru_2=Pn_y7q`t?E<3Vo!syV&kuN4I?D)Gsp;l63|)6>UmdFC##2OQw;3b-8~t
zLxu7;mhCFUL4<y&D|ew}=+Hl%dD9$T2i-Pp=1i_=@G<WmtY?QFfhrH0+}{mNtz07~
zlt8s1r}qr@;W*D~(1L$PG2VzezeK&h6R+d7U-LEWlg;eHePr3(tcP=-{{(Px*<;^d
z6KY|7jd&r(1G%Ozv3J34#S@S78P4;LwA#Pp(a0~&^q(uZNU{fnx9Ry?5t<gcINivf
z`buWmF7CSwJj&A0t06Z^->#WGZCdO!FsiE1hYI!Z%V^{pP`5)PgTDkM_a){@OA(aY
z1aQuUZ0!KNjMGT!qm_6Ud2+Ik7Q6)71h^Rpx4U(!!oy^M#eB$>_DqxZ%$aemKXxb5
z^tzeXUNggg^~^GA+=B=*Y+72(kE}v-Jj^p%?dgg$KFXwgKIQHqK<=)%$C~e7>QBtT
z7kLjv4ESd)pi+2dI4tYrnG?EOJTo5gR<CK1jlwjKe#yaz71R)E)~su9ydIrYVPZD<
zA8>UfKNmP8zyGvU((mamq@Rvg;(a(?70UjBWQA`7ISF>KsDU;MLeG~>unCNsSjovR
zLvlI61zKI&oo9e9$=1V>XX91&ZYX$-r0DET;UC}={^)oX{LTiJa<z?uV3#|h@5bKE
zTd?Z*%!57f^M>0j@-CVz5naCpV&l3!9w8uaIl}_3zo4zH>EIRhsu#=f0%Xqh1V8De
zGcB?lN>5!cAE}WMz0JsvX|*NnAqZ}ZmJiswi58=E(z_TIgi4NRwOeVFjhw})6P!>~
zKt_p$I1m!!WpF(NYG|LX6}^eiNP9KUtB`GYdfOL{`S=vc(0)Wl8+sfsng{Z3Xm?2&
zDHqLyF#z*E*CyF?yJjXv#$AX#@P^;U(3l3vn<Sxrf|g;&@Ryl-;Dt&w2-4?${E9?W
zrA2y8>xTAj!n-|yLEXHEa3UaKm)YdRw?K+)$@Jz=D#wtw<Qdf3!TW6bfVcK%>iHL!
z`%l&pI+DM;Y*+lar2Z2ap_H4@<!u%M5afz^Y<&U&$4Ml`(4M>~1ay;>k#ad20+t}e
z1wFV1L`P4lX`-h)@SD=p+p$`;qX7C6QRY1aRh=D2@tvU8{}=r|6|Dx9PvJ%Nx3{E>
zl*`fjyR47&cTSr0m-=>R{WV?aFL@fM89>yS_Ym~|tiJ=%YV`L!UQ~b2mXwilIa+^L
zA^-)QsA)S)`f_2>S$}`m75$|R7N8@D8uK24{-5>t0<;?a9m$L8@6RM<q+E{H-+Js~
zMknsWN*VoKhE8<W-zU1Dzb@nf)F5iidkFe}*57exHTvu0MfG=#q>Ply(fYf!m-P3s
z@1(z66QaLEu*gaozb|0bw+a%C{VgGC%zFr0XyT9vt@gJt3@6JY59JZ$)iENE_LS{{
z1ma=DYCVJF2-=W;dASyxh8NXdpt4x(ay-}yKwpd<1kKzFv4Im8PvMH)vA;49z4Ss%
z>Fc$~UNxnEz$x7evs5J-mF6EUZD(X^9s&%p!aIZotO(j==%&|PDsV=4fTiC27&ySO
z9lmgjm-esdZw~<qrHM52k7-}pUs0MZ%*I;<5;RfZo3v&p73j4J6mik9z$;mxB`ppU
z6fGoZqQFqQ0$aaTEtK)tvA`@AXbs1|SOpR^QQ(kv1wLjKxC~Hf9SeM+EyLmW601Oh
zCJOurR_^33Uv3p>anyxSKHB4PIt#U!s=HMv!4rknwktFpg^C{+D}(d#c0z5S&TuAZ
z0mUS0wt_*iGy?#}#86(K*cLz=Lu@<98{bP_Z`2d(B8C<A%(a*C$9}~Zd-cTyzJ&$f
zp_REFVg0Iuhufkv_Z{W|rU`xwx8giazZd3cF;q6S_RE;D2TRk|92{VoJrzBg-dvEI
zp59#0C;P`V7x&9p^cnJD_~~mIoeD-Y7nA^yOX>Ain#cV<RO3S`wO__JyPYDLq>5}h
z1k3-sED0gjhPxF8qape#7|7FDEuK)WCl|MD)OuAyfv&IGS^k$P7&?TQ#43~flZb~5
z?8J%c>SF?k3cWxoE5I)-Q}|V*2z8Ii#Lu{X__<u?petw|GXrI#@XKdy7*n2Zj>%*U
zdaxWdLNg7mgQoynHQf@l@gXRdIQx&5guV=>3kNDu0-^BDw-k`*q0=N^jkr$o-&gr+
z0CkeTM&+yV&`JJ-Du3CL>aS4wic>pzzeMG)#QTgB!3q_TF<3#i@5d0p!3~5HwCey0
ziCQU_4ne-=PheA+xC=a)8tsc!f{Q`M?;{H#zt1OyJp>zpYCG;FwiXXZ!S{+&1)p0{
zd#Ctbtn%MJQvOVp-*TjUL*+-0ls`b_??(Pn@#O-X!dELmeWlYz7-4ruwElzfLg4@{
zTb<^wRr&n<61z)V1oU}}$0&qpGrnH|G4TG#K({EMJ^%vsh`&xkw1s*P#E(e5Q*m-h
zD)s!cY|6!Fs?e2=k>YRfkvW)Spp!b=F^UQce-pLbm*A1_Q_E~s3n~8t){>=aS@Ofy
z^544!V=X^yEd{EU`+wM4-u+Vg!V&cY(#5Z8S@y%$a-6D#qyGo&%knP-W0cwdX)W`d
zQF+^uM&&3~3&j%;B?|I0ssCYhTJSF@MXD5E*O98+vP(L)@_*2=W1NoBfs9DGj^gH8
zRZE@KvCuW_7;Nf}21Ol$FT)UtZ3zvwUk8R497%budYcmYKg5Rrd$h4>r(l9|{Rgb&
zSE?3H3_oBk2jbEfHM1nP`<OpzC)q^>u-g&owfz%JGo?eGN|N&b8gVIfj4eRTTx})k
zx6Xd);!D+G5&YO&sI6=uwglusabCk(Vaa6}O2)%@P=P4-(qbRr>EO+<xA~{*ks_S3
z@Tj}~Fzt;W<7Fz5(1Nq|Jp4>3fd1G%svdx!4&F|s@EV-|iPEM>lp~H={3<#W%!qUG
zVKFab>|^wsp0Gl8E0iSxzFvyWxGiK)`S56Je-a9A)lPnKKo7reeZmfbaerP0%7#Bo
zxHPB^gnNnV!7pK0`KU1!WI8~6=A-V7XqI70um_^*tau=tZXG@AbI_D~X8#$z0|N;@
z1$Bgoo{wYq%YKRNPS$}VpGD6h`K^#V?7`e2i_NI#;WqpG1SsW;s0&bn4h*t)5wznA
zN8#{chn!(~W$YImvS!Ah?)KtlAgLAU^$T3ExVi8k7B|+pA&+91M(>Puq<=EL%5j#a
zfyjb-4uJ40diWbXyvAL}r^uL0a~uxBOV@ECuna@^<QEAXmI}CP!Dh4uHfh)lLL+@#
z3ALmDQuh>SU|@vB?gDiabn`e@oR^V}5TlRq4sQG4_HA{`Gc#UOw>&dr9d4EPWD4pV
zya`@7c++J2b<%oGjHp_0G@E7?zmi!w7st893ai>Ep|r?E6yhz#W}2mYXJHL(&hDc<
zIU8Fkn{mm*f4{=?eeu=byX|JsUBNB@X)94aKdO*EXAq8H4Z3HskF5hUu6`>ct#B>Z
z>ssI}18E1d7rSSw%#B*$cqE~e5~~zWVycrTpOVzcA&!0JU$k8rkzu6m+gblB>cKAQ
z88*AL?GSd!{z%wm7|oj9#$dl@1=<AE0>QA8kVI+z3~%Y1@e}c*=U+Cf>^Kh&eDn-o
z>SlPpAHh~LbaNglt|*Edo5om3T2xOR{@`(r{2Q<pF?{6=4{RbVN)R`Ks4zp7s8%;4
zi{x^zQXrpIy4TueNHok4k!6?>&W}Z*?H+I>^F86Ea{ByyoI7$a1|xilcXQzKGe_GE
zWKtawD(I}TwLju6r0&U9qYr{ZD+Ok9(Cv{dIOF?)zS2HJy%ls<N*4M&!A@C%8=OqS
z+9)(<>1LI>4~po}tsOw%mbXzFoYZ!1J>RfD=%^h3Ub~#2yA`?KL{ZD}$P)=r3?~r}
z#^@kii$ZPlD`1R!FMeY3ctA+9g7g2%Fjq8^>k+8M!!GCXpqP0)>~dZ|eE(qn>xfR&
zOJOcSxC*zYg9Vnr#0GCWN%aReqFt)y=D41Lv)z}!zwBdA`qw(dlDR?zL42uhD$zjx
z^_Xjf7}&&(!T`cPXtfXGmxK0aHa{!%q!Y!0Rc`tCaLSh1U;o=%W?>>a_zy6-wTYni
zpSW+giC}J<+C&h$6*q?Yj7T%=b}a|7Gye2CXS_MlKb;I`=u)>^=xd2YfKpI`p=RL{
zy}MCdZI}$o&!itMC)9|}@>-88BX!7kaMk!vFpFk9upc7}ui;zM0C81S;@oAJaS+!g
zywndID0Mw!ampIEUCLIx<v*@DMAqTo27F!;0%RD#!4S)lO9$a|eS}8>EiH{Z=JnYM
zsk_u&0Me5!XXvF{eWh=qf4!Qvqko^5)MFHY>re!t>j52fx8SKL{4En2Bq0y41GYh6
zTTo8|b|qkug&~0kN&fDXe4hCuW2cya2T^*q++gesuv0Q|b5W%4Lnt9_Y{c^rf;8Ur
z_J7j=;>hshBd!NHH&x;ulc6vCNm2MvVc_G_3=Dfp2Yx%}ZD#hz*5ajN-u0%#Dfo<=
zk{?^{^R&Cy>E}M9Mq<@r7F3|!y`doU&4NYGxJLlU6aHu5(-s}Cv>;QS5Nr611PU0?
zM*1JtJG?gx?-@9(r2ppjd4Bf-Px#%y4h+<#vdiDVK&(W;+T9Ia`1J2ym-q6bV)uOe
zjQz0@;~~Lm&`miaI4?lyUhHSQyWWelmhW!JYghznB~>n!Lk_$+a)>f*2GaDvN0pqO
zDGwWW+}0<f4IWgX#uDU$=_oPEc^h;OLJfFg#HETRT!}j)vc^3Hq*5tS6ypJt4?<4-
zO3OcQ^K*j#m_kmE{_VrVZL)nIXG=t%*+C^Y=w=~^eMeQi0|LmP+C!4UDz#wm-e@@>
z*)Wb0!bLpk>EWUoc4-`er6@2u4prbNR~@a|2r$OM&d_ShaSO!*g73H`^oI{$A#%}Q
zI?i0Plk(1M-b7dPP4vLuL<js$i16P9c~`m~a&936Qj-5-K{*6o^cmRi9BF6de^d4;
zrCoBp@ul*;Rvs)1jR{S_toMwJvH9M?^|pL#Y%nIfDiI)<!(U^vr!tQZa^X{fdcuWI
zQ`FNGEV1&8{N1w~AUN+w(S`Y6P`EAviYQn~5eSP%ZDFy(5*BMjSa8H_wUAk-+O<YT
zvu^uFa3#NJk-L#i+%Mu4b3MmV-fWy*tdg*Q%5J|g*<j0!$;E84F}XywRdr3WDsm*N
zA_rM)zZd&ia6)`gh93*(;7s$Mk-Gk9Y`8-3gL$RW+QG0B6rO)2QRoq*x)wIBD_U+N
z5m@#t%wtLdm6BGV($G4hw?NXSsI)2V(E4ZKGi71|GBL_AB(#VG6t_o-RvjdUtYicD
zOC$ae4}xl33-Mzt%)@W-Le^WdQ0hgC(MsY57vct&?}xTO<FKRz#?DG!uR)3uAKu4_
zk0hjh7TSfHpK()2Yqh^d>k!~XnU?0yLh$Y_Ha}wHHd;>!q4wM0C09567~Q1(-K729
zrZgRzT;H^J@}{Ppleag0Jo!BcBUg9)7~KJuQ`{Y3IVIge=V&GAyo!akping2n-Aj%
z?|Bf-{%t&e5}8|>fK9#ZOo#q79Yk%I1#8FwZ7-MMoWcX$DGSKHHrHy^+-N2Jci{2l
zS?IDJJvF$)GroWz#d&J@Ed<ZPEaw?_c+Pg7afjzz=NWf+?&Cb;4$pm^=f2=;I5*d6
zveAlcI$mw}N$6WLUaH7%F8pFlSLF%w@xrG9^@Izbrl_aHc$a~$W-Kj|%yWs+d8N&U
zGCu73Fg{fk#jFD3Q<cp|ROCn&#;2+sSsb4@d=BH2XIt7s^F$f%mrUkMV6Cd8@hH5t
z5(8xZ)YSb@<e!?xb-v}QCRY^_JT+i>-||G)B@O(~G#deuHbteiM+3(vhU5x&i%6-z
z2^X}JQd)IYQnDkzr9JpDYQU)3#Wi5m?2;O(2Q5SkhzneZ3tYY*n*N-ErVir~{lEC}
zc1Xi`)M{&iFbqSg2v|X37*1P)sTF*k7=Iv7wkwUlMp_;(#c9M)c1fB-N(S?Cj11=G
z6lX9mrzArL9(NZ?X)QzqBDV9lX+h4!&cNHZuDkzCd-SRC4$5Y`DoI~Y2&cZuaEq(q
zrd)msuMR#_;Q8RGbg8#_!BeJs;ssAx>WLRT@$@Cs!wa5r)e|px>Z6``!Bb!LWDl^i
zMTqHKkmk<;&MH#mg~H!*_7an^;4vkixO1iR*GMPMvWwG+v+R=eDNT@J>G(kgevyt}
zBvXcQEFgM(vB)qRl_Iii83t5h%YIKKjwA=wvl*;C$)3jUU=To<Ewr$^%NAypfSx`7
z$nb)RDi_}MEL6y?f&jyV!fh3#7akO9t021YpfFnnxrGOX*eVDuJSe<YL3ZK6LTf|W
zT}`Xc89M_*tC1?zp!W}Bos!+Nego9RudoQ9WuhOLwdt#?c2l|EOc6JZ1K#n6H#Ybu
z7l+b-(j2+)iFn2nE_@=o@q`PXrl==J!o8%1wpu*fB3_G7wA*qmA?mJnlB^kb`Vyyp
z|Kt)?<5X4EOuMQy)+l)uIg(eAgFGei)|I`aCr_;>qP&YH8|=^A99%4k_qm0pInkc)
zDHJ`_WTEJ*tLllFjiQA<C>$_<6fk7Zub4D*15Jl=@UwSvaX>f?zq}{E@t5~g<le!`
zCXxH8^IemulOh1aonmKd0z1}>L48E(-#fWPDxE6T&6Mh-Sxi2Z!{kFbNVeyJO(_Bn
z5jhUMG52K!z(iRA3pZa-u$_@1%lN=|nOfjuG>(D>#tvxrc)%iVCD-`aQ-k(^&-xIF
zc5m$I+@^iwBbaKQhDpK5R|_rwkCBw^dYZPgQH>dCv)42oytuvtSlK&+>jU5AX@Tn?
zWWmwc9)lYS1-3y}20Bz_N{HC=$g$TiXwRZ|xi&chwpo$q>AsB1XN$5)fFRCUXiKze
z#rljit(yDI5-nq{=V=!k9btB$8?#Z_k5HK^)1_cC33IK$ND+Y9|BellsbI1QQzS5l
z(E*n8x&xD?U|^9Sm~pzm9JFAbbYQX-OfF%*TgUe7vtUdICRf4W<D>&K-VvCuESPH?
zm_7;ykt_o<o+V863$XCVVgOY037a;|0(U*Fpalyx0*#Owh!{maphx8k2Ck8OZdps@
zC-?qB$wxG4QhsvJe=PY3lTOM{?sa#`$99{f{Nx^g*1+Ci6K_&}a&Ma@|F!Q_e(Y7;
zke@qcu;lUb>LlpXuTG*q{pzG*=+m!GD(S}IPBlA6)a)E4WA5pO@yEHRxoi{HkS#ze
zzM<mH-|Dcl|M;e_CSiSHPbW~dzaZ1!U4M+nJ>P%);QCk%Ns$<@9re+QY>y}r*-DAX
zR!T%Rl!&H7SD{gRuVSNiUd2Xzd=(l+`KewxlHSo=wwz;&4c1|Y%zq0X;|05?=pBO{
zy`#hY2+gDE;8j@0G+?MXTF58%JejKWKcCt^nIuJCXB#2Rm5E|7X{mV=Wvb~jgy|D$
zDH=)$vnbM1^pp@LQ>1Y+b+nZb=2WDm=qw>jtw>AJTtb*(k=9Os(L%*wEX)xQdyCy^
zU7<I_(?tH^{N)M~#r*Y8_5||;-@NdtKt18Yrzz^mnI{_5><LpM(QPR}TI2~5qT`yi
zGAS7ic~lxg!l}h8v(p+IN~Ap{lMObZWO6Z^P%^nh)vcN;S(wyS>yV}9jb^SNTAlTS
za4j3=KQ+ORjSaJ<d<glVHG}Nr4DuM@EKUxJLCcTRC}I(`NE9Sz{^^XLT_7*?I~Upu
z{mzB<LceoO!KFm_)?Xs5KC?tveQF6<9eFwOT8?@xN4=JV*KGaoPwn+TS9_Kom|$tA
zedI?XX>JKPnGB@5iAAi#&&0D<&6DC}ePzLC8$om<=`~;qWiTzrBU#y=AXyiQii70|
ziR4`*Di)R}B$9cNsCZbOkVx)DqGDosLL%80iHeKm35n!iBq}xz+WMjv@(CNa%`fK(
zg>A5}@q6gQAGGXiu$I5C>}xmDzIG$+Yd6xqcB8Vdc}tsp5qFRWC))ZEY--uA6^Qx6
z?i9`)hv3+iJuJJ*J}yfCQmiuZ-zNAc*{5K88W0wD7F1TPmLCWZO(Ow+RDZ&PS5y57
zSO44o1oX}I7Mu!2s6W+_P!?i{L_F9|zi;nQD1HHoB>G(}`y^652~cupf#;(73fq$~
z5F2Eh4wgi?s%zR^3>^pg5Of=x4#KgPEt=23ZCcHC<a1!}<~4+KQkB)#w&(iCQ#vFY
zSkoKA8y^5v9?IK~v+396Z(eW~3eBp)LSUoR;8e$|Xq<-*ux-xuO_+;nU^1ZP7$i6x
zV+*q*^>_lnMy=*uq?qEJXsZv)6;@wDYm2ubE8gFpLT!0p$$|o3<6DTaE5B}QTL~Al
z@}_#U=T?YY6~~LGSI+B;a}7Tq2z4&~d?9<nJ(*s1w(XhB^Wb<T4{D@4ksq>DI%$59
zrz|5r0Z{N&LKc31KP9kQhsIm}%C&D;{z_E=+>~l97@=`fH}7qc%d=6rtR*>$;0r^4
zXak@_GSrGRpK|TdAz4qK1`R9pUQ#X_mfJ7k?pq>-sj~wCtfoiDb6AOYlRSru0Nl<u
z$QHskXgg2lm*GqYoHlyA*pGN*qHQlP0kev|+-J_Ge{eqigY)ShoUi<Y-qJ>Iq;F;N
zcutJuYV1CRewX8VLe4_{lH++N^_oi5AHR>Xa}g0~R<oPBc}J#PEN3Y^q5w<sm$t}b
zb#^Kc#|KL2lbXskD_D{pO`CtK>fQ-TKmDKfj~c7%i9|0A1il=hf2qz-Hb$#`0Kbq>
z*uHiP^hs=b$bwoLAND=v)t2T|dG#M18H^Jz^-}-6;9%#_;5$TAg*iAX1(W6F7{eKL
z_>vrDQyoI>L9HV=J2{l2e|n7SO`BG`P&$Y`AQnF*`(JJtXFh8AUuMcqMjR#%0g8<q
z@+))17JOIxetR)yL`Kl*3=!(guU->h?W05HOsM+eka=F#ir<ud5`Np0zT2K3;tMJN
z*c&)t_hWw{<dhP56o%wDh7HmADJys7vNkxl(E$UYJ!%}G2s~3laUf~6<TrfFJ8(Ep
ztG=B%;XOwG1A5f~1%`8EX3yk@4O;X_*1gD2(*qw?4#w9#=Jm<A`Pqy#*cU1$W}V32
z;uG_DDLJu#m#HUmO_0?aKgGTC@LSTGYl5l0p^toyHNiI8qfk6H1D72m>n-(Y+oN!(
zba#J$@$&URJ&F0XVJ?OE?unQpa<Xh2R7}ZQoIasdKZI7wd)uCo_NBl>4S?y?Wxa1G
zJCp%Kx+>66wkV?}%SjM>lu?uIBv1uG8M#h^*rkk`K2Czzri_}tPC`Q8xB>bG2nB>f
z(iH&tg<;t-F@GXXuGu|tQ&V)}_NJDJ?>YU2EO*!7XQtSD{WU;hc16vU#vNDIz2CUw
zvbw)F?(o%ZZro8+$K#aYRC{~&t4$wIf@5P692}GC8$X(NQbT%E>!gOJZ!56fzM>w3
zf)#u6cWaT~lcM+IEP4)}{o8eXBL}YDz^C2)C)xG{&Y6unZeE4YR=}9Lx$Fx)aIpLA
z8MeNIaRqW5geBSqsK+-p?znc<%Xo%!0z`Pc0k^=WWJzjvA~n}eEw1T<*B;*{0DTFt
z4t;?mqM>Y=Af<I+1D(h#(c0O)n&w_y558s}xYs1qw&1MzJPzx-HX)=8UPOX5mG?^p
zntw$qg6;N$SAk&u0>iuj#k~9r9B({2jy?JNwaDW{ZfSF8*zN1km&}xkVAxZE;F~P`
zTls)W!7!6;6f1sII9BNj$tqo8nfErD6-+9g@3-+hyBp0F@iUvCrMqVlt^c(V)-BY=
zFc*bf3;PtWoZ0CcqcmxPsr52^h8c?fuP{2VBJCHuX8-~dDn_faV2KhHICK^~2M7sa
z<V+>r)q;0}o(Q$XJ4&qiU3dj$=7jPbO3%NnJrPkE=vOWDt5)=@R`jbr>KOUorC)XO
z0O^E&XdlpDoisv(2Gwea@ZP9SqF;3q{i>7bSFMJLdiUt`JB^g9+>jU$7R{1r_4t#R
z4q@*ljydUJeVZ{))c2eaH4X;e=%sJ@N;eBpB&3g%j?lx20AvdNuma-;u~8gWymCWQ
z+9_`X1`IRR&T!&l`XB|y%0al`e6?2rW*=&TGfmnyA#E$tio&~n5KS;{<BX+Qh?~*(
zZ#B|4QkAX;{}cvVX*Hq}wgiuQc$?8bW=&%VE6HqFWVjdb=))P|J%JBfc#>NSO63D0
zVAM%Oh(b-wefCR;UWw`bw-JVGgp?u?P6x|}C)^PDP@;5pqeKMv79o;`<rBRB%X(L|
zmk~09Bp2<nr&m?QV)tWgaK!a2E|^|hD%qzc0=yBAXse1Q0`jI1M?rUAmW;>cc%f}0
z61G|iv+$N4(WkKb-BMaTc5kVe6uv;J+WFMeA!Q|h2GV_zQBzc0%rdEau?p$I+fhJ5
zsnBY#K{d*6h4Zpxx{jAAf0f{zxW}uh_6?5ABirYW;;>p*hOcrcD$!x3bOks7Utqf(
zEdh1j&jfB**o|ZUsoAhGDe_h9aS4ajPDfdoeKYhIh_(=!R_t}Scj4Y+IjCr1!-H~A
z(X@sKQ3zm(kOxr)V1bYaQ3POlkOxr$U~!NKTLCCr0{qQS^E>#PVd0N>v;5ug@5zB%
z2t^?B5R1SIf)Q{zx{arS@hF|3?8#)5(Uoy3#|iO9e-re7%zsISoh1B}r{oK+FwX<n
zw%}lvmnVfGO!V@kP*g+Y08-(oCdEpF8DH`gmTKDBWy->kCQ2qzC*ra@aA_Vy3!V_i
zIo;Nhn%<uHu`-}J`ZRGhK<0D5n$IgRpBI7$p);+!tPWGfI$vE0riyh%b;Xz}){U$)
zs-crA)v2NB{Yi)i5%p;kRi~HJn|4onx#{bQP5hT>o!<JweGPDzPR}I6S;v^+Fq2u9
z{VNv^^J`aSRg1>7sSNm3(T&|_D|{j!ST6x4*WLU;E?V^KRh%4<22B*NK44&Gz*8~s
z35a6M4U(6w@&up7z^CL1)0F_9c&`LAgh2L1nJ7ir4xxC!XY0V8e}5b{rJCNS0gqoz
zUTTSPgiX?Xi<HHd7{@%Fhm_`^grK3b9MKMbWp$y(7^al1$TiS&ghwme4VNnvbXx7X
z_(g00C8Y3se46lgdwXXS;ww6d&mpfL-A3P#A|DLI8wOXRDkU0>^agw#w!8l<htX|)
zujwG;i^zM9VP5!s_N{9CCFkQ_2?AQ&<w0>fr2=vkvr{VILGe1J0v;5rQ!3y=aXO^}
z9+K_o-u(KqB}f%U7pE)uqyT73n?H@l^h(^Nk)O#TWy_Pj(hv(|VJ1>YDCaa2!O%>k
z$WT6NCQea_B0%7=wwe7=5Bz_uPm4T-nsjn7S#>v(upF#*(m>m6h<-;tp*?672N5n)
zQ70}l?OM`U8Qs&k&^?XI_i|k~lszgRdsLx(jH9#agMN(u-}ZQl7n@A|G1uo6XeF#Y
zI4VfldvFLoj#zf$BcJ)0_S+-bd-T#KZ=@td{nX}<_V$nw*p6@8Lp<rbY<)F;pG6;g
zea@J0=&~{(A|t1-<c@>0+Hlco>D{Ik_b2p^9MyjN(sQZykW2nE`z`j$E={tAEEa2s
zsqop}Ci3W?luhJO*&5ctUMt3s`+tLP0PULZ2bySDOZgsPvOVO|i_8^b4>1sbkK-8U
zW%DPGKaPEkf`TRqe6UdB@3F5!%pdM=l#5tg<&3`nv?zUV{E;O|R*!UmU|k1YKv_5J
zp@XaPL2>1{Sj@mp<pVb%x##LfGbF?}4J_T>7-EysR*5ksaxmD797dUDb8ys5#4m=$
zN?fo`;dy2<4{e`+Rq8zCjE^61{`pFaB<7#qXv+Uz<{ui_IjvwK;yS-)!5=fyZoL_I
zORXT^fm^kj3B3fs-aVoWj#`Zu$<SUm0{<|4Z$+=Q;7Fvw{;pCH#DT@8TE)b!IDpM|
zwk)g1JKYhi0JBT196w?)Amx&5Eu4~^`AtMxC{09~$kuGXNYm_!&`}qZ&B+D<TPSoi
zzIgIM!4J_WY7l;eI;Ln39?E-Y0jS@xxIll1|5y9|JhvbH>DS7Rs=3%vX|=#P*lZZP
zmpcwqkTCm3N1iH4nq3TMoMW<Ky>!IUW~17Wv>=7sEs>tIAcd2Sb{q9>VnIsbd>;s%
zQThmk%UmcF8$q+#-DtUiWdSSGPoL+?)LLSgmQ_KGKU!{Vv{=;oReNf+qt_2%zX6$O
zzY#L!2_}=~qa=gaCY)E6(GJtMq_j`$JHYM_X1^i2?<czVK8$E0WxqKGXt%}&&^-t9
z3QRaEp*`L^m~rM}Op;(2oIOJ*(y`}=83(d<36ufU2c>){PG{l%*h|dF0KB1mZgd4%
zJ~t9_C7>4!Kw%O=TCxt@a@D%4_f+&L&iDmNN5*0|Z2(eY%||ETZf;I)6(lZg0crCu
z0EF0p8XL+7uDdlCx%cKFr#Zy$ee*B=BUA3_k7F8y1Ygx6zd>OT99%D>kv);VjOn$f
zu-T=dL3n_^2xI*w>WQ=X7|Fc@Dy}^@Xtf(4vy^#!J!;a!ZyNo-ur|VI)k@)fi%y^=
z$Tj*m^QdeX3EC?8h<#?TN8inq$Pl{z)AX*-$if^S(F10dph5~7&q%o#BlGxFJhFh7
zl95wzDRB)$q(q4^3|a8^4?`CG{llg-9m?@sunfP(1@-tXzMuuaeJ{`hUm{-pK-uG*
zHd~~7hou=t{+6<=*vUwNvxe=`aWv-sw~X{H!r?>N1uh7sS-67PgKTh`jO}N6DRKP`
zGS>AJj>Yn}Z7lmVo@oQ?fwu0l3ec)$YDeA61(zf4T+;r1&+{LvB(8T#Jb!bcOwaRC
zl}7a%X|m{key2*KmW?!7dq2NhrBT^Nnk>Pe->cH7b0bYw;?K9LG^*Z6lLh(nhg2Fh
zaHMfv?s{HF6r2}^@X%$L8z@WrNxO&CDw4kh1OQP+e|*GINUBw@K#~`>sQxX?O!HQ?
zR^noj`Zu`MBO@$QuaP~0bb;Emu9CGqez~^Cg=>3UxVEnXsa=g#AhiowAhiowAT`21
z9>G$m`!ZgP?uZkFV&A=mm%iP-2(R*C^#8YnSE1<!K732+`fzS+R_qv(8=Qs&Z@AUF
zf2%KjqmcVdwkT=+{}OTHLL6d_>qVSs=55<l_L(=(Y7=!Gy9MOTu#l6y9wbs-iT6D(
zD)PRh$h*cx^1i6Z`;sE>8X@nCio7o=@~#o`zNpCik|OUKA@7Tdye}#8t`YLSsL1=0
zBJUa@?~97OFDde_5%RvM2bw|NRxNZn$cqk8zoQ{bMdPZWh@n}_(bB4Dcc+|Rm82bj
zekuKZ1WTS8QUeYHQUeYHQX_jrP2(S~tSec#tnm+*)fF$SZ~TL=&RE#e_=lo8*FrDs
z3qH7J8j`gHt3h#u6HXK!$wu=!QyF9j7OikvWG8-wgd-B%I-=Gdy)<ynO0~a1cxF9M
zZNeG2GG77Cu}bi5DZRQ@2))!Ay}5S~FN)|R?o9ZDjORZ_sl^zyJ8pZxc<N)ie2k}}
zNb_LO?zs5@&r>_`U|_Tgz;4Oet8(^APOHjk#X~U$Cf+H2>JT1WXzH~Ops6x0_mtg*
z1_;sG=EKCW)FM*u9mxuT7=fJgM4Jv}&wL7kG28zX1Y>r^Q!PyxbWhP$_6c2O|D>zz
zEt-8`T-^YF*+1bgOEUU^8X*1=Og}WF3HHxpWBRG#AXwsHqK$GEI+6PuV&r9ha;a@g
z*%wp~kl~<v_RPds%Bm#>?jk-*?gBokOzrsfs@L&s+(leIPG4H_UBIOBj~RCnk5($M
zs8UPr0uF!u_`873x{LX*;Jduw{B32+I0Lt!aU9!K=ChooN!8^@%HJQ}qZwP;>K}kM
zxgjih;d?nLinaGv+`xcC>$GLCtZ>hW&!WK+IwvxfJ_%27uWRn3mN0W-Ul=ZDPMiwi
zhRaL{H(ZJ#+!`RPy6|^u7i1w`y$iA+$GX54p*APfu~0Q(!R3vL^!&qG@M~19!<xST
zO+Ec^!M^71euSv0eb>qJQ*O=GduH~T*sJ9?YzCBvYO%a+Dchk34r8Acto~Xx`;oBE
zVfW}Me*-vHaIdRZ@JkiVPOuDe#c&N<AjOVRE|Eg%g}1U3DU@AE$xWnCav`NpB874b
zDSZ<umVJ1G7UFrr;d{wip-mR^!<f&d54PJt7l9q2cd|Tc+mOFm3;K~`iqod6@&<7G
zHEA5)TBtp_K^3UqPlJh)eO4${WJC=bAs%NAqo%DBvBK{l@BB*MRY2aIPD^SFFK}8`
z`|1W%H&z$bO@%3_dL+!M1&!51VOE`zkaurPde0hO+pX@zvy<D;>N(n8Pj8A&dc7%L
zv7P^7Xr)A!(RLw;J_{%@BCZ;H>LQoSQugb<bu^*ouC}JGguRup*?8}k(_<=x6t6hr
z(9YDV??goug3AHwi?}M2+ecEB)XAhawJu5{Yv0w4>n5)%S<M+PySQ#AO|8YNIky3D
zx0qp9E54}PtMY^=O6powp76uex<kyHx>|T47af?4Lcj)`unv@k_0sL^^xJ3|MsDtX
zGQs1avEjO_a+gSw;`yETWY9Q4At`;JAfG>r*2H&{1ZFi4^shu#8f4<s)yo>!O<O*-
zZkY_pI!eo_tLs%-y-K6(oVvP2rM0NE7Noh+`Dx4Dbxd*Bp}CNtAI$!<MOq=-b%-Q)
zB;Ibc3%11Hq2>;45(*mO2KH?b;6$FN_ThKtYPP|@nzfEt-M9k{UbT#unXBq~@vmy(
zWyGqcLpjL7FEa3pbo`E3#kOJYFZ-*3&<0?L?zl9ooj6GooFRiM_`)GJca0^@T{RYn
zEnk`rX7^nJ!ad&iZcsR<@7)mV6s(`cs_`Pd++w+fSHeS^EeM2NL6w48x$sGFDi=Np
zHs#`<oLfmm=2qG!*#^PjVzks*QVIrVWA>wU7PAg9m9oCH>cf5^R^*>Nl{NS$50Ij*
zrbrgVuc~~2)W`4fA-;L5gB-uwA31k2?5*gk_%FD&+{sgNt0`ebE<pmFldzX4slz$M
z5;y{*q|!LNqXdu=JzOIZ)g0JRlWRIrwK!dog%j(KM~<#<RVjU|O6glwO5dtV$e*N8
z7>lwN<>NgRzd`rvOy@%UMymXwNIru6$-?l3g7rU^KQmB8XY$8T@~4>c=e7s3Fa)7+
zDf#1}{DH=$<c~r51D#CC9}ndZ^F;m_lt0W<^AY6_v@RuoT<E~H0PG-tT0jc$xC_yZ
zwlK=hx(^|=7=qX&)<N)GLk`_7<jobClZMe-OnC#ZgpxP7NE+nL|6}jn<D)9>{_!l?
zz(NBjsR5$}40fYQ3zXE*CIXs;jhw}^Tr^6ss4+o{N(CYD0ZlNFL|B(qYSC(IEmYfw
zKDAX_FL>=P>_P&95H5m<hKrJAxnGiSOTO>-%sG2WLh#$y^L_pP`SBuq=JJ_2pP4iB
zxy{UHn1&$>L0}bTF5p0Kn1=DoG|c0q6g0MMCFmokN$2$D@lNQ?^F$#E6-jDzIzs_D
zZ10o!;diQ_H=ub450-Bgzf(U*BMQ?#$m3(i2Soyv5BSS0JeY+CvJ9Go^X(mIj#9lv
zr!=$hnN@HyeLNDKD10Eof!@sfYoeexi4l79oG30$7}$yh)8}$jv`(xZK~lRJc9CVm
z_)Y2U*@6aG0pHyRr_>)W5W1~2riS|1K3<48e2slvg{P48W`Z;tn*bNG=N5b<Hak@6
zy(pi$p^?Jb>}l{C+|U5W{Aj~cU=f8s*7N$H3$;FJ?Aktg?Aks>feXpzUii^^Avw9A
z7m|~w_CkjRu@~^OGE^jFz835S^oh4Js;}~&RbA(}9Z;$MF%YGoC=rtG97Oj;y&p;h
z_M=5s`>|h*XYHW_&wdE>8T4GwN65|-^<3;1VWF4ubUo6AKFX8z2n#)wr|J<F`X^7+
zBP{ezo~B2b`exKQ)R$y;&c@{&XSK||`mT}LQ%S|A>pizDZ2vJsTqsY{5;>ZZ3KBV*
zlBOmWaZs5R{%Q)dk_!}OB~MlKep?UvJpvU<cZ+&=iH#EQNx=P+{q*(yYSQzt_W`-y
zSt0slDRSib4&*Z|7xt6UGcrcG1~DPt$fFsE3GqfQ>?0<~i#&{hm>@6m6b52KyphK*
z5Yxq6tVeUVdnxvwd`)Zc&ga2`J$>)SVhvKTzxB!&+uABC2E>)7^LT+50}LCK3Hc8R
zIR=nOp*On2zvYHX`i`S>w}|gJK*P~f>`)1h1f}W>h(>TS4(HiEuEsP7p2p;|eO$?4
zPz3Q#Y*6Et{K-sw1V3;s_Lo)NF#vOLK;kth55V~yJm86L&wKDh7l8R*T+Ue;hMF&-
zo|RC4Ce#Oj!Y(#6=EN)T`10lNV*PQ)+;5a&b8WJDZn*{w*a23ZV&x^Q+yvxsU2qS!
z!bo+ExY0v-<}C1ax&pqI1iR}5uZm1_fkfD?sc>wU%s^7sZo%(7W3H*$tvJ>I68n-(
zUTjJBYNl3)XXMLv_$3^I%C&gj6oQeOiuqzw-kos5wZ3Yf;%MVeYhTcez($dN@)GH%
zE|Grv66t3yk=}HP^yW*Xw_GCq>?P9AT_XMbCDJckBK_he(pxW)-qv}#(n+$TYTftM
zwpyiz7YwQKQ>|(ANR}I-QfpC`3`J__Hb==<_>A25%$4j|eq!_FUz1;R#HKhnIPl3m
zW-n*H1$Hl+xEJSSV-5<UT8C+m({#a+d!&3lt_rexY;xS8n90jhFT#aoMr4Mq5#y=q
zT$-}v1#?`S?@)s3+t&kzoz#S1E=u4S$NItj2nh}IT+)DCgef+p7clYc%pqyeUGN;9
zfoJefK!Wl!0NMifDe%T(pQ8213yduE`)%`o;Vr&xenytpU!0p02xJ$2%CtI>R_8Cy
z%Gv1mWj%VrTs-GvOZuk6fq6T|cSCu5kF<b)N_mf>PHXWOUI?`QqOJHODsJ=dv*lb2
z_#SOG7oS2Mt%n0=cg*hv6VdmjiDEUjmq47wDhe>e5?v2-Ag5q+z&jZ?B+eb$qCAiN
z)z8%xPrAvK<Sm{w)Rj~p7*?>ec*{`B>J26&_e}DMViv#c+IugNm)J#KQWtqCUF4;9
zk=MVAytFRztX<?~bdfi>i@af7<Yjk}m(xYw$S(4*DeRKp!QMrl+C`q$Mc%|N@(Q}h
zE9@d~Y8QD$UF1#cB5!&Zc{94mo7F{LX%~5OyT~iQMBa^4eEu;HCf1t#W9AL0i}RPv
z>)QuA{)etB>#n#zJdblUzT$gYf^Q*2F(5|1{(rz<uedaS^)++@Wv+~3u-x_>VXzM@
z{r~xEM4tHnhQFr#Z}ZnmgTF5LuKX3l1ZKxh7;Ijg;IGrcd362?rkapi3BC#Dnu4cb
zvwiz81;>VvnqL2;WV1s)Mu6<~eYbdy^5Gf$0&F%~-}H~JNL|k#FV;y?df;bj575TJ
zQ#bo<*;DHMHtW=#4DAjKF5FJ0TQ8Vyy}vj!XGg$VxQA)GMcQtE@t~Zq{Jueto&qEO
zyCwZ=GU9v8sIvf!m^FhcMc3?Q&0f^Ju^l_#1cn`8S`^z~>-Xh8dK@*sV@Y4fnr}%2
zdrk?n=P{UB(d_xgyEg}Jyw~Tq{q$}$<EQs}4gMPw2hWX(C$)=%x5nhBwu^&@#^k59
zi-T9j<byxP#K99|;=|g-!TVzJbK1qh<6`ph?PB8KWifH^ub4P^R!n?iyEu4LOnzaz
zICxM@KE7~F9K0qb4*n7o2TzHK&uSM3?}*8t8;;lFdrs{^9&z_-GKxCn;Ar<x?^*93
zb6?^X!5r%27~UzuyW{+}%=@+m=FRViz%h)8xkJ~JW1NM;=eON;&z9gZge1oZ+lONK
zg$fnR=Bl-M@(h)%z-pmR_B_-tcoP?LHI%O;97n;45Nc3jAGA=*pupj}Y8xENd-A)J
zWu#@*+diS5Y5OQH&V=$atU0`;aai>H^o+g+FL12Zh4;itY@A%In%1bMJ>rsj*imsW
z%27LRDzFg=st8!Y3FEJNSR@fed~U6GH#l35J%8$7UDw<?=rjMm4b5%7gJ%C|N3-8H
zsMhWopXeX`uVydKsQ1cut+!R}>yxLv+otvW+~Er**tRs<sv4ej_N;|6_;K8=ZT0}P
z;rZRT=2_J`P<6G={tMUOs@C+=xc*Mp!(0={<7t=-bwq&`N5!d~y4hyfeugPaz7Lqt
z1V_nE&6BO`GHu|Gp0Uth+MvIzV;7XbHNN>>{NZNxqpE2w6k#?UPQ76eI4J6Ln7-Cb
z&DsbkvE?`q0aYhTlR<BwGfY0U1lK&oT%x6~j|h6ZJ|RqRg+BA!#|?evuLv3X%!3FE
zeda|bLZ6xSlq9&Se2$vP=e(H-)A_8J*?U&Gc=lZ^p97Z5XZk8UgSX)8QU3ucJ`zZo
zlHCJ8<vsEOiG`<02%7|nZSoiQ%sCS<7hYgmn@DT(7n^b}7H_GDqw=!%V%+X^#rr*$
z^tCOBSav+c`?UPNo{#P@`)xgUAIA%_n*5`Ck3He{nI8QdiKg8xOl<RyhDZJ4y(t;c
zn@o8743g{Kq2}U8%hNi(FpyGsm`EHENF4DOC+8dyNSp~HQ4>L;k3gd1`_B7)$&YRa
z63M$8iNq2A=)PkQ7vG+ck$$w*Kc*u0s0n$Yqkh}GS!eKT1n&?*%;iVAfeG{rGXeM-
zBojdIJ)TtF1DGUQ3cP+>(&QtY?nlFFS|3+ZQ}K%V3!tp+HCqa5i{F|bx4I6BTA!Xt
z&7!QOc*SFJ_g*qPvFq%luCr6R&Q9$*yMNc&X<cVqyUxz&I(u-}*~7Zd&h9!pr|ayI
zU1#TYoo(+rTkSeq>pFX4*VzSKXBT#zJ+<rXqOP;2b)7xE>+Bg_XV2<7yR_@<xm{<M
zUpl+jloM0|TLout6Dr_M;O}#Xo*_qn50`QJZN>K-4gMV=!O@NN+*p2Ya0aqc!N|*W
zMvf@clB&+a!m5#vE69-+Ru>M*gNsQ{4zR8``O;Ya?Q4h!|ITa2yj|-X)1!EIjJ^uq
zo#)mzng!z~fB1jQuT#OV%dn*>hF6n|#_;Ov$g4HAWLGq=7WTk){F?U0Jk?VgpvfBT
z93>gGs4$vQ<4{K|qef4$ByEn-S6wa@{5e^-H&(4pqRo*yk$H~6qFZN!%fj`4ccf4!
zD&9NalGOVnq>@<=`EO=JD|&Ulc#irTpJ38!_ym*QP%55%_si#i^YWQ~MH!g%M0_TG
zo``(qUtN)UP%`auvYm3UojkuSaq5{s;uLUh^VExA;b)>)c)&d60yy}^C=Px}=6wrH
z8Q+z23tlbWpH{rLJR`~8BVN89?<<~Eo;LMJySE0GO*w$QIo||Ye{nz<K@S9e{pe|P
zAbIz(;yK{mM_Zs|4JCu7ln~9JLxEk7ZZiko-;KS06<G)TtM49rpm=NFjP!3?{c9?6
zzcB@V9Xb%0H|y|GBXEZZoSpx`X|U_mFuPWz&Ni6rx)dhW@88`L`2M{?zpdBZn83a6
zJtERA?c#~;;?UV*;8WVgp|i#0_iq<ZYZte+i$iCNDL=Sf96DP}es;S!bheoMk?rCb
zFERP{c5&!zG5K1%ICQp{{DO9I=xj0hQ`^Oh+Qp}}i$iCNDL<oK96DP}erdZnbT&Qz
zjF^n}9@i94njd%fG5+4$gj~#)UiX}&t`?@~Ywk<@#^m2WZ@_`Lz=Hb@1Tc1vLsmV8
zaWQvjGpYOnXluoj9*euDHTamJxGjyL@1W6jwg`=v>t`=YLXJ(`R36;fSVkj}FNI11
zN^en~`4Iqw#s;skTu58v@Z=?s%%{>t+;y?@pSC<~?QN#&0iA{wcz%WoI?zEZX8ZOg
zsGh+#hbKGZR?o<f?4*YPRpVz^Q+drj&j%{+7WBU8+L7mN?UCm_YR-#e(41Gf)&liw
zhggFnNbBW6+Aa?g*ENV*<uhZ=G)<{_fchKtxxlq*?oOqe=c%&lf)ns?kO_nrpP;X`
zTW>PjjnyX8T44}d-#C_BfPE!;O|CmRbI+8oZny3P@^2J3+d8a1RiYMYA3F##Li<?N
zx^nUZEE91RYVQnCa+t|XzrvHd9I*!V4$?rg3e(@#149lX5cHxZ{d^C2M04O%s9x3J
z27%%^`5Ec{mcU1aZP+DbY5N|CrCc!*SG5Vi*DeDHw*+wl(0>`g?@Pea9tQ#!XI?Bm
z<;+OG&;tLe&3tteU)?0=^t)FB5Onm~Fdh9C!zgebB(yqETW}n+#@u=)u>Rg=G`YOT
zl7P3MB^+BC6I&J&dod>Va!hP_OzgFo*vgpL>oKu6V`6W|#8$<`K8lHXV`Bc8SZz$~
z)0i06<f2-;Ehe@zCiZ1atRW_LASTur6FVFeJ024|9TRJbiCu__jqDMQtiR`kf4>lA
zLF^#P?w{B9rnrFb801z-=N`mfg}^>gxgrMI-FGeNUBT^i7TDHDS(FXah34AeaP5HS
zQcqteq#d=yyihf5Zfmf?&g?8{@tEZSl@}eRwX}nHvY*0c3e9z{apn`aq5LF-E48E!
zC!wCfu7L5lQ_Svyhp-k)c^bk_*B1YZ`7_PW`CqLVV4jS7>#@<i%BNtX;OzbRt}V8z
zy7>HDUxhW_RWJO(u6SGk&(nd;5t#D*`M7Gra}k3XiTSQI`I%dtxaPdM-eFqpsM_D7
zbO83RtiX!v)8qr*IRC5j`kE)iIof=Vwhd+9IN&EqS;!Mt4o|`OKB`B123s`&Z+9Jt
z6tIsw87C7PqZ3c-iHD;ToAgA^ApL#zaV>h{>FC6BdSXj-;srhNLUdxQNHn`5Y36b0
zH6OfhV_9GLA2oULRf3Ph$lv6<MyaVpz+hUs+0X2Cc+9tmuQKu(?1+cW3vOo|uARmu
zkwO2Swl-1v;Z90dqV!W;mHzgCUi!|CrLjBJld=?J1-DI!;W84#rKuwZ^@sF2Ms%#h
zoCgdr*f6Rtf&VnP#}xQq7ggHknGWNh7skIv-^B!klxJ=ssTUg=u*J@|F9G)(jlR`m
zzD}<j<MbNkseS;m@wQ5KRO2oj2*!n5*Ck}tJ+L&ZK9+wg@0KQAbT#LB+rV#+f!{I+
ze!GtRwpnd6pCUx-Wg*%w1CgE7B>>KAfYo9zMa{#4)$$VBYqi|7O3kf^3%#Qn_fGy4
z4fEitN1hJAmLOk?1uXJmZ-FtBFY#%9yW1Sc;yei1p*Xj8_!`WYsp7SZvWhlvnhsSw
z=zWv3P$wssK;ox~2?B}zo!3x*`5KmI$QsslR>Q;IM=pcFTZhdsvWp?H7AIcd{6Uv4
zAirQMOp7s>NMf<q>x1XI>wCRc%>ohi^sU1Kd%Zrdz%H&ERQPEv0UiMSbr|qj3AkJW
z`Usdyz#Apt^f2H967aty;NJ*%GXaw&;LTycQ4(;x1bm)=`2=i4Z{gMb!hk6fFkJ#Z
zNx)kPD2_0?6OR1aXiR9PXv_sX(3l4acpCx5*+qAPKMc590&bOnc?7%%fFbNcPk1E^
z^|FNeyM(%$P}2xC4p4K$P>)HdMH1>X?A`A_L?|nuZVN+Alu-9cs7-__1Qf2YUHf0y
zD)-i$*0yN*e!<~Y95b$MklnMfCGX(egt*4TWw`qQgra&<Z|1FAXSZsA!uKn;YGSqJ
z1MoMD17D*Ll|l&B7M#81j<I8vn!6HNVy%iIMMjaLPm&P)AP+x{TV}Vm;TZ81RV|67
zeXCmfm0g8Fabf}donEQBr&^0bKSBH~<2?}gHv+0>)J5?ae$7J2py!hkHWnwuy=iK^
z?$Oq^tljJ4`C@h57;eA<@Dt|1*~#9(+2k=v5lpdzwDC${s#dj_m6{T4as;3ylFt>R
zJOZ73%mBIxK$_>_#0U^>xg^MEVUQoN9#?>O3Rd#nuBlj-@r+5Lnp0?V=VPN^Lp*3}
zWhcFXJAE4iXh4U)_y()c><jc?XgIpy2;=OFe<4u3rP=a?KX86Z!dv%=p4pE?e6<90
z5PBqK7J39+ZpKgxO+kgK%Vevnpe$2vO;f7B0awVX)38J~F_wPwZH1~km8x0}tWmCU
zo{^_UuQT)LSlqaE_EQ!HqI=?OV=cyr({+(|JDUQxxUHvb7#t=6)A$h@HqM1&j;c0u
z*(8yu;#93`9~{9oKEu&v3TU|vWj`C|DL<7GU2(8+rp)H2vMI~fjdt5lVST!#*JD2j
zUJ24mc?fb^I9XX_LaOTNy?;8oq68OGnQGe_@~hj>UJwGlAXUBqF-&(Tqm7B-SD5&U
zAg>TakA&CzL}@yok18z&yePd1+X#Wbfnaa}07VJfJ)=qhE&Y8jhf8!5@0(zh;045@
z#AtMKDm$6IoyI^^r(k_AjU9IfsxkU|+Nsp24!;#2o*kaX4p(Xp8|A~D9{dnVK(ci9
z_(+GaNxVK9x38l9y^Uu;8}x>ijl-r%a}*8$1H{${1RFBmGvib$zXXg?Etc7t_))6=
zh+3<>&G0`s=b_-_o_K9mTP)z5&-2~)oAJq(JqY@#ykUi7(u0k^1Z8lz@-4)2sCcu)
zUVgGE^351De6cwG7WKVw$umEAcZ^el=iq{44ZSu|Ou}4*IsO;+<$mP`pl5U64O{Tn
zD%NTnk7KOzwYWvR^aDDuZ_-?FjRUCA4-9)ZS^$5-s0`x8nj}Z=+Br*!Wuw!5zXjrb
zf7?29e-wywth4uTKUIv6_o8SXCyUK(77hb>lE*VwpE9mThz5#_DHCbGTW-*YV!;t`
zJ1P-Vr$zKaFMYig_|e${kq+KY(FgN6&mE^d+>mn#(lhQ?5z_M`lHsACFb{!Qga!-J
z!(Pl#UAEM!d6u}cTOF>$!FVyv`rb_-(xV?D;P;*SkswFc-Yv*c0n$K@9>W9VsODjk
zqnN5n-&qB>5g?Uz>%^=&VPSY<DyaI|OzHhW+7F0AhUT6n36lEot{n98z`)rI3FP<9
zOSTV8L9Y))UF`LN?Ry<``g(lWFoDW<TL(tXmmi?{=<R{Bw<ku8Humxn%qG#x4Q0a}
zp7Mb)j3n5&NoI5YE6cn#w>B`Trmn1C=)v%LD;!*qsik)~S)MBa6LHBwKkO_HWk0Yq
zdR`rr<9lAaO0iQs05z`4Z+7HvpVMchscb#mXCIRNQ;n1Ef}+=A3KPSO=o-&94rXP7
z)%`aCTyW4;K5&?XVNa2bStN`_si_h$euwFZ({$LFKKc++tMH+6G~7fF-*5}3asM<1
zbcVVvKP?Rvx$_f0<P-PUm#}1^=FTG1gN3dVZN(i7Jz?=*daJv1s0Ea!@f1cmI9^{z
zRUkvlJyQ0qXosjWbgjB-C&upes1BFeMlwo!j&V;NilJ~fh*X2>YXm172)c;~{eYd~
zg^^dNnOhyEFG<W?-)7Y}hGxJuj_ZEnISlp1xt*BbbxQSlgf;gasp`XLbMmuxtDXTX
zrcs0&ey#*7A8-*MIe92ImOtXcoveCC^9j{`N1E!fC$*`~U#NpttI*|KWy2g*YjSXa
zQ>o_ZX#c{B;r6RjU~I~T$LO=CeSx#HYL6D5ylBaTOQ5s!y<^?wS7(ebN&FC%YCUV&
zo>FXc@v1L5n$NN=KMc+UpC$rfegxYAwz=T_*u_#a3Aqz+!$fPO0*GDcIRh$4KOA=U
zj_4ev=4*Vm;1T>pTIj4fpj10Bqakx;SME2_JE*UrI~JNM+hEK{bivNrT{WGk-mNP6
zwJJ_eFGu54_%mFIC;u&$4u^!}8$>yyak^(F4pbo)IQOL43KgL7xka340f)<Mg>;l%
zy}PVjb>|NwbP;R~)*=X1QaZ#J1*h;frY*zRrR@1K=?xseyI0jB3*OM=00t$LGiEI2
zu-F$fM*!}m_=RmSRfJYu`8l?=`N<|RjH`0u%B~5H#~VP6@aEVorJC+;D12t9?lEx2
zR53g+CD8gLFmz8$a?gTDC-DS!Y|@jz0fM*#9k8l<7LEXXycutl(rx_o*0!}ZT;-&X
z*X+2mG$H`3<OK$TsJb1gYP=0yAg;a85$vWc{09nsD0`tv^~^xmfG=l+uEu&L?^BT^
z9XF2UsJUyDs!BjRke2HI69(KcO(HznT`)`??Jgb0epOvT9pz7-<5yTG;Kv~shJ=4d
zkRJcb);8PPMNwEk+bP!L;_X@RfwG7l9DERC46o0&EzVDtB!Lu({i8ex`xl+$KrQ*O
z+R&Juma5fEJn)f6fHROD>_>lr5eE;TN`Nm3ga1>4=SlEH2_A9+W_lRLEMfW+26qcI
ze*2K9H27aY4S+X?0c{fS6jUOB{TpE1EtS@IKi`f}MhHhP6CgeShn!p8+ST}|!;591
zPedWMYf-q|5R?ny>Z*h<tBmGVN$92W6;s2Y-2jSD(Ri1O4L#Pl5Dm>b!s90k{HqX-
zIthaMmW1njT8Q_3e~$+g%;~81{aYPEi;F|ujiLLAdBW5%{9O|MVF{l@cvZr0#^-4K
z5b#m-1h>E7Bqd&Nn{2z&c89ItcGY!Wbw8OTGFBs_>WRcSrTSk8(dlBU!~LYC@mq`&
zh(Yc#iI9Mls^^fYxt~ay;IY1YtGoOlC~RA4LZcT@W8C(GDG&zBQkjv)jI|RyGY+OS
z{)uA^no(6t&$54oGygilQ+_bjR&}$b>?NCftQjog*S3!=;_XL5SMa`W80xD05WO}k
zRWtE@9qtL1aqf|WSoPjUP1HKp9NK|-@<dYONW>9HWv<OV#)A5W+dh%?wS^Saw~qC#
z&2i*@HV3-at*NLhi_qUR#-S{Vgm$xn#*+{5yL}!0Ou@ZKSQ_p&yZgz+4}ne_cGu0n
z(cuCI`DGFiwTL4`VfyXB0vW8sji*sJaOs2g04T$mi~CPP@pe~8+92<?-DA5KCxS2x
z$0n(`PXKh-ZLdvTgD&9){BV89PeVd<JfmGaCpzvPo7ld*Q9ovSt%aHG>W|J(Q%lx^
zAD97av9!aF<(ZE~67yo@RpO(y!@oa1O1M4ByKK;<-*?&k_byZZGWnuEqwoRV;R>Vs
z(`f(Y@Dc5A-yZi^(WTM#rSf||TnAQQw+w3!4}7BYJA98-WD9E|-}`?%p5+x7h4F14
zZ~wmb`TyJT%e_w=oSy%H%>{-$*_)qL7wyW3`+HWDCt05s?xmhi6Yiy+&f}x}=_2}<
z>hBWnrCe#ky_74D?xpN5_*?N|5!x5)UrIWb>ge168*=3>@Gq(*;H_Rb26LEN9Imqt
zIKwX;6g!_V$lwlw{-YLP0hpe_jyT{MZ1!V@KN02wNZf7!=L8Gr`$$`0=HTT02Jk<?
zTYlvQ2XH^YTYiy~By)JETH_T5%;Bl(0^V@|zXZJH7hv!y3BLpj^mD`OaKG=l<nG{T
zYr^`9?9az9-JgG|xWYOPlExK-9e!MzvWLNG<C6H8;Q$(U;|dj#gf3;0c$AJC#+;2J
z$APeOmEY*&zeoG<>Z*^iryJ7bT|7A1p`YVRM{1tFGk0~D+leCc!>+v7e>5+VOZ**!
zHvfXzQxbm{IC)oH;C!J^{{&y)6U0leDqyjhti{mxQ7BQCU(BE^znJ9_)kb$?)hW!<
zYzRLzC1o+|4IH{_y~u|PI_5J?z##t6l*CB+?os6tjxI0otMZz0->1AkK5+gnZ-?)t
zE;vjfOl;%oq0m2wuQ6f4Iri0@9@tV~!%ksdsL-N$7SS({?~u7_A2xlb)MHIN;U5#A
za9^C+do7-jcCYGN<3z#*I6hboSj;AL>eV~D%e+U8yh`MWMgMrDFfgQ5)63r>VG688
zqPl@bG*3itHlk@Fnj@nrRuTKDk=Y=kKQ^LX5w#l8<sv%3h*tWGE3O*e4^ak2^+lj~
z-^Glp60R^py_1blU&ROw=xv13zh{Jo^fW@ldl;cn2@DmpKDJeadYg<;-!?Q`qz$-e
zgwoF&p&@6D(C}s>H0lgOq5-G%z<^VFVE9P{9#}e7dGBS*qYxut*?8o<e}T<kZmBZK
z=kpk0jQ@AMe7=v-#Q25;`P_pM#P~6s8m_njS18|7J63u3Wy>5^XUl<O7Tdf>HWc4h
zk@=kY1AkRkTDhgh<gc>f`P**(cw3rg`D=V2Fz0xJKRyeQ56yfQST-gE-kEa(iJ6Fe
z-CaIU^x*#T`oNIgUk1`!acrsl7rXtokvQ0d*B1Y@BJ<V$&-i^qX4jhoLt3|E>-kl?
zahW=@W;5#(!sE!u$SL*MVmA%L#g={Nm*Ju*uLSzv`85mG%R;O#<rUGIQqh`uk8G8A
zJY|y47tsRZ_doIS`4Nz3e3OoMgO2yf2;Mq+vDlX<&9GtvS@xaZgt6)W7d9b~a_7OQ
zW{4L4rCl@9zI!vuVw<r@Z^lb{GybMG<72%UpXtrmt2g7+Wt;K3n8W{BORU(k_g%jS
z&!TW(>EC0q-jLtv4S8K}$R~P3HtP-9r#IyEWg9ZX+VG#XVw%<a-J3yGF{&9Zy&1pP
zoAGzO8F0iYhSe6m8T<8SocRvTSpJ_jqw>2qgHE}ln(?&WjF<Ih{6lYsS8qnW-i!l!
zGny{jjMuG#{Quui5Vql~@SMU7PrTF&pBLK(x88<7=xumIZ-Y;7!&bcw-{@^<?z#=c
z<6syMeQuk(oZH1&*%^2443|lHr8vuugwE;H<FVK;(XoF;$No(n`_(%3+jQ)|)v<5s
z8he4t-Y_Z>eSyy2E{KyT^+mXp-heMUZ@?3=4e;m<_@mx{xAX?A(HpQ`Z@@vl0cXEM
z1NMheiE6<9%QYaHKZquO*?9xXV;k^{-hkzL1OBNuz^^x8hu(ljy#eRGLj&3~2Z7Ev
zmuo=u@E0Y%>bwCJu?=`uZ@>z@0sqn)5YQX&x!!=F-hlJpp#kk@fIz1!{#KmaehwsG
zdJfEwZNO5!0e{jP@V4H7wR!_~>J2!gH{il$8bDM!%mJd)<s4wCm^r`_mzV=j#x~&B
zdIMh78}N?afOUEUcIgcW=?%E}9U9PK4iKF#=KvcJGY43r(>XBJ0+tbG$<OK7zoujV
zu8w`Jj{O%p_J?)sTfYPL9VP&A>2d-P`<My95}i(fsg?y%*#Aby{?9u0@9Egr>Dceq
zu|J|?-}W7_?=bp_OIM>mkR3C00`|@a&Q#07D9o4XnEyq`e3g#*dL8pGb<B?n%u_58
z#M?>K*>}ba7JFyIBqC}XlJ#mJy(vs+&Cdk<ck&46Ro;jkZvLeBtuq<hHWa^gF=N{r
z#{wT0qP5)h@Q;L>wYkt&{ZFJoUF81N$!aE+hH$~3N6W&J&G5MOLNhE(IDeDW1A8|i
z5Nq#H-n$XXNY-uL0!Mc541CkND{ujN0yc$PKM8zEWg&UW^1u;l2}fGDAg>;&^+=sW
zEimi0UeU#j0^ZIFBxB1Td<((<9h~XAU{j}BEKD0jC~VIP{{G#zX!|#<MpbR_xwH)6
zzGK+)?(3$S;6BaVhLv<rLjSt%LPPfF2~D~0m-0E_G@jgl0<AvrMk?`E8UY-w{snhU
zu`Xr<4`2yw#*N(qbv%!||Bjraroi^y!NB$lhXSoLv*1A6auIeA%Mm!xHXjLWKNCXh
z5xIL6wv+g<6&;SMaCGdC{{7;@+2e(9t2r;BR$e`Ou3v579Lx-c^(NB3XnQ~tb^mtu
zfSaT20o<P9I}~5FuK^$TIobnWxmnr+-o+EzKbmn>-`S(0Oaib6^s3eE0rwdhm8b`x
zh3Th8>e}sinPr#(t~8=~qQ)E}nkJ%IGOC*aer#lN09@7AYDB#vI>3l77twx3RG0yV
zD2NJsKp#CYsy71C9`HRQ)Vrq<>f6Hz4M;FT>G4Krh{*^IZ$rC8Eu$_nB<um_jZp8i
zMyPMI5gKsD2#G!sC5D_Z(uN;1LZgl#BwBJ<4-5z)5H((fJ>WiaBiI8%{soi#uUn`(
z!Xz-sKf~gJNr2Bkz$C!uTQCXm8AuQXOZ0-L{0kiZw=L9h^n$x!D{%NrEpAl6=O6V7
zj`SeKkZJ!yWz1Wb02FVy7}#>`h)me67lxssxZ;MjD6zS?9#)0H9|Y=-H5ETs@uTOm
z7Yb$lejS1HG&6i;p=7UH96a5m<zYp<e7<9r&$<NpgfT-{CSYeMuE_cfW{DFCf#JLT
z)VMeMi|<&uLy$IF6H-12d`zQ-Z}{vj=JpcRQ!D_VW_PUIM3*VnqJZ_z&%;zP@del*
z?%c{kbrDNL;tK+Kp3)&X-#|0S5(@=;P;^>Go&iPY^G~4Ye7*~c&gXhibUwcVMdwo*
z8|r}fHfbpkc+=pKM*~L~jg(h(<bg~_hTg6PY2UpCzl?1GC40RE&+09BRd2z2dJAAn
z5}4D>!xl(8L&p}pCaeqpX(Reuzk3^|$F_mOz21hUdK+HT+ptP+!>4*1X!2njq=}(p
z8>U$s{?i6bwR*pM3%EcYHI69V>n-@T-hw~tEr3N&V7yUp0j)i3fwV4k)`I2#c?&AP
zdkY?lZ2`r5y#>$dE%=Mx0=Q8Y7}M-V%xUJ~QyLaJwm|N(`Og}W(qSSCJ7dR*%rgK{
zAK=Ld`oGc9he=hSDNUZ5h2}P*P4f<)(x%WJeONF$ox&ZZ+NddP?tt~@os*v#(O7Gi
zO1;OI=_s$%QKq4jDAKUTCrvqgN<%_Bl!clXZOM>yr#Ne5XZDN;)}kEbGu^bC_<|_0
zrE|o86^r-+9dVei1!8~I5vOUBh|-kCC+#<UN&`Y?h(~L0QHXcc;o2d-wLRjOAo%o3
z8y+W`WucDv3p(O7EfYbSGWn$Oj8B?w_>_i&&Jd5*+M*Eeiqwl9-`*baOAK)CgNQ;L
z3kCuqTJ?z-jmdn{c*!R%W_;3Q!>2SDbcT4e!WM;iR|8xi9?fCJP`ShypB0OEwT?KA
z`9zGyV?JrF<deoSK544qQyL1|BTi+mlQEurxiQXiF=L!1E-}W%5{)4nXk14T2+@2`
z#Aq(&lQv2|X(Z#5CK^7afuJ+QI~n6dqsuW)#AC)dOI%`%&yMYJE}sw~n(O(bftXL4
zC;6mtj8B?o_>{JR&Jgcpj1!G6$2bv>8RIN*i7_r$v2<xnvv9$L2+=^#C#}MK(k#g*
zjbeP#B*UjP3v`BfCu5vwbUDU}c+41QiA#)eW5G&@k(z}IBt(cNc|K_n=9A_~K4}c&
zlh&B5v_xqYXoolL4`I>(Q~Jox#)lM#Bhs(`PN<mr0$k!h0w3KO@|RRxwMmM-Cm}~?
z|3g><)(dOEYFGo7L$MRqfX57Lz^h1svgpF!Z`)Y|PQI^O1C9)Z24k$B>-zsVGqmH;
z-2!KJZ-y?BiuJ!g!36LWRNi{tI<k!#|MI{Y$nQvk-g*DZPm$!H4wyAAk;P&_y&XV%
zLjA|t5UKt@-Cq5V=*PkHe~;1sh3@|ww2HdF?_f75{{J@a(GUiJAD~%Kjr*<{CRF1A
z=kbiv&tmm|?MA3mk0%X=0btVm_Ic*O7gMbBgdM;fYX>+4{og!e9yWB)P!Og3Z-nBX
zLdE|tA&c0`qTi>1i_;qmagWx#ge~$4^>h8U1Ww{A^(Zp{?Ga)lz*QzH_QHQOwz+zG
zPlf|^s`oelM7Pu2Hbh!8#-lMnp>RMh5*SDyDRYJt$mR~8A(|WMKeV?8zum|MP|FI(
zcJmjH%C9A7IAMmC13hb0K73hLWZm1ncusmo`i7Rk@YYZLzVz8!;c0S6M*6N6*tUc+
zR2Rse@>!rfIk7vr^?E&~b<blR8;#1=va-p2L|LOUE-Dw#8I_T~r3DMKLL)RP)4Hdz
zHgDjgDOmqbF7IB1{tUY_tGNlpPuPSa_Ml(96*C^f_8^{vHtazd6BrkZR8vb<8@EgR
z7y6N7;#oXlBWigU-TTdBG;2*szd`T7t(XPae-s`c*@k^++>~r#A8Nw5D8I3L;C;G-
zS(BgBVhYTA#C~J9;@iqHJl?>l(CWab(ly1mf1Hu-Rf^{X;5Xp_EJwNi(W5N2{?+*w
zTvs;S+!hF&xrn&|huh9o>vYReKMtUp?_xQU{(mg(#>@3=5XOt5kBZ*c{#Z76AQ~mc
zOHY}QJ_HGX9Fi|uYmZ}lF(@!VvKO=)D(0x65*a3zK>F_e#dC&dr0;FvFxqP7_xsXc
zoWJrb437R7Mi@19ExJUPPc1%KA0(f$nt&J}N!W+5TU4`4RHF|NS#Aw$TZ?jbIXZli
zIEP2reNN5I#_))Y_u%eWd!e8g+%GG~h6OYOM;y#g#?Xv|Z%djnL5_t1zvOezWAyGb
zWysa&ItGScCHrvHUk69gOF3RlIHWH&6@i{0oQMIF^}g*w;PIK-jkr1@^HJJ4hUb7N
z%zLEZMsx9Pzsm4nD@JH*U`XjUbj5afb<IfM(c&*2V(u9je&(~ljZ<uTAC87^TL0?d
zysyufA)1fDviSd=+Q{=Kg5EsC{-D2*wMD!D$=3kCEM^?)!O@^E|0T9$h36aEhdVg-
zRSdE#-7svu94grGb{cMLFC^cxT;||r6HH+}%`4NsADz}9)2@z6>#kNEtrX)QcXhtR
zNgna1|3iIzSFP<ve<moLns5aUKTDtbSqKS#CKqQ8*n@8pdL)Xz(-oBu)Z_^Mnn+bJ
zt1L%&t)!(X4sm!!fEyDLhhv?Kzc_-wroi@xkK);kzXO4d4}-v$;M?E-GtA#7uZwXs
z;zG{_a6t~NZtymehTrlYdH9_MANvIb-hft63j@`?zJT2bOw|Kay?k)yIP6|8+;%6_
z!Z}qAz4Jca4<+EIH*n)Ub#&=gCtSKouioR~Ji(TA44z%*4)xLH&p;sz&c3HMn4%B;
zF#Q|CK0B6B)n+P#1DK!>h}v8&@KV}pd*6x(u3`1`ZGB4&^xofQiXn{oe~a?o)LvhT
zU2VT(xF^Ec)%H8~c|AQ{Ozrg*AJ$&{ozWBbN8WEa#qh|Uw26a=I?cNSXU<@0fC~k`
zTDenH3Bx+(gshU`DP~b6q=~yNewffLoLpu<3SuB(SfSt=waVKKyA17)TrW=gqJg;b
z&{R4ZX9+QDIYeP29SMix2|q_Ocs)zvDkPY2Kl2n^K?Q{C0<AL(x^=_vR8u!pY1v0?
zA+g+Tt`8hJ)Atxhju2tWR;|WBzy}F*ZK#*@_SW!t1pScybuc0qq@y<Y*IB_QxXUDT
zYLEwsqSr5s@pz#9m^>iHpfNJ#s1)PR7?T%&(H6L))gN%Se&)B0YV~0N;an2NL#r36
zBTgpyZMjHg5Z*vV>Rimi53opcF8Q7dfjtlJ!B{>R_@?Abbj8nhV=NEESOy8<m6NMW
z=V91dIgV$Daa=0KG4G1_vS262GHzPFcP9q49<b{HI5@j%rzx=N-rev!pRgw0?<)e>
z<Nema;UIiWe{jz(z8<t|7_JdD#%)t8^<9qJ!wAXAUX7lLlk(4xa4t&Hqcjk>Xdgxc
ztg??s$!Fg)Khzo8X+4mBJ#vCc_&}1raTqk|{);p?a;49qPlXpFNF#4|@M6@)Yc3ik
z3<MOwXj2M$yoNc9)tXPI&=rv=51zIGZ5muyhxoSQInIpqtu2Aq3!Cs&EZ*<Ym!0ma
zn~u4`=_>}cF{@5w=@)e%Ydl=5(3cU0#UO9kp-~gC_*vlNR=A+87{m+p1~nnp6bQ67
z0X(<a6sT)$4xF9UdejKqAp$KRUG-CrqQf6K5cO%5>Rsp$9JikVe@L4P==a5Z=1ide
zUO<9sEuoJjgvxs?6=`0?jJ#zc&xpMkhJ3jL<nkEE*TRr1J3zi31Nmkc^6d_gt70HO
z3PXB3K>A}KYr~MAc7WU*1Gy~>xw8Z0mobnHVaNj=ARA*K4~HRTKSuZG=@`hCFyw`H
zb;5HZKMFg({+=K|DBeH02`-S1lI~xJ<IjOZ_nZk1U?6(`x9!f`C*tr54*{)uYsA>%
z)WQFruewh9_nmKoTa(5_^+E5XH0Xf16)DsAK~~}Xi#RVB#V>J3)Eby7af^LZ=>#pe
zsdOCgR$HqoZ=6(pM~6Jv#L}6^RWRvoIJWfX@y4;KS0M#KUU(Z1P~;g?aBOMRP2!l-
zRl{*?$(+R_eoadCMoe<eC12ujzgvkY-vz&+UpL+i8jrNa(P=-HX<XP2!(A!UUW`tQ
zmubtR(~g0-puUySX}ghTtC~MCt_;_O&z~DtI@Wy!_FE^##Z}C=RerJ=uyJKqV0C<!
z4J~ms<Cxh>99VZ>p_Z)HT%~I)4)^%EI7NwGii>OWy5d!w5&DJgo;B{<&3Zi#gzHH`
zJ#AQNmh}{|p4-i$B6I-uc%%Fn97LBDpgZq$8r@j{K?dD<KOX4LHK^nL2W5AzzX9F(
z&QHXtvFpO{RtY~`!aqd#A_@OH!q)&^$Ws$>G)k$4QivgN+&z9`TV2)JQOcqZ%}DU{
zec^h%wIuOr)wHE;uj+miO$6NP!eD>}^tgMe_&_dSVepf&ohJYCmz^djJ<)0MS)8Qn
zr0!x!rJW>i_<baqn4SM~Z2TAa+D0*v|F_uqhQG(gcfS=I54{%~|2>@1N7whTH#Yv`
z+DJUAXHIX8BomE-&tv1;jz!{8#fISMrt!9@<R+Yw?j-q1T*BE&a?TB%Ci_Non*7k{
z4$1DtoNhP@Qy6?5e%OsyupKW?h>agEjExVT5*wcy6_09AS#d1b_8GD9GfUdV{}LPj
z^Xrj#RJEVH(P{F*y^&-#;`-yU@&7(;igT+4RcE^?Rb@DWjI*{ibvEUBuhDY-6-K{c
z<Y&WGNvFA*yOMNCb1z<whw3t08(W5Tiz{4XyudDAP$ymhfBbTe#k@eTLX;g02VaFa
zbr+e6y&jCbnT`1Z*HuR6e}F^ODqa?y_JX?`Fv<(3MRnQ3OS;81_W&CKS$KySM;BdJ
z*tXZm6x7zhGphOsOzB)z_K+&UqTlFRS7>W%kad+#)Nyl9HskQ8hBpFq8svJw=<h3q
z^9o64WYgUf%yyi`)7*uCRdv|?vc9sPM|vo#W%r?FMkg3`MF#e%7nv*KVRR=>)8X+d
zybh@0jzOT+HnD(jNO*6oL7v}%j1PLb77vi;vDF~A+j9lC`!VjIYwY(UL7o?d;cF!P
z3le_(ZGb;>qkum@1n}Qx3i!!k_&X&00}_7Y2*AH6;WrTeT?v1E7(QLX|4hRFf$*<N
z_}>%$DZop4ILaUFg2TMSS9NVw-Q$PZ-F@IY;gf3<;wHHJZ-Puf2SV`4w@aycU5twH
zIPZ8%6663ol0AL1A#YD8F+UCNp!J03xkG8CiFB+0$-q~d7<z(E4j3+b6vgsTY<wzZ
zh7yEsNmFxgQOgQ(?0%U0dyWz2O=X(q9+QKtF-gvn6j^7zsIv#_JQ+%0btj<*mR*Zd
zd2lo{29CMCWhpF!>c*&LJ#D%03NdI1ypq?zQEe%X>cfjbs#1L%tPL)0O-jw}H*;P-
z0UsVX?+@nzszZ5W9gYmAsqWk1Aw{#|B>yt280X3x>rr*Bh9+z7J5$jH0Nn{7yEP5s
z5dAeEn1l+BYjl3D!EA!(EEHPJm(N9DE%)Em^ubGwDYO%)(1BAPj*h}RlY1`uRL!GH
z1|i<K^HkMUE$<5k0Ny%lnC33hfix?o4I$`np+uh!xl`eo9sUmBNTb%m*1^}y7?0i5
z_L8l+#kE>_V~c6CJ-61TIC0%dwRM&Fl&IPcmoh9%Y^H1QX@X`7zgco$h*EPDwZVBH
zdhGy0&fIfK&29wL5vB0xTeF$LD%>6)r#ur+(E;H!gqxwbjaVjLqj(M>f-hAL3`<U`
zxMo_sy~LNf1p=wsTu+1=h>!_gwoF`eU9~r!T{9h{3LUtEH=!>Fga(b@^|qC=8IoJs
zVWO_)UHocEumsc~3q!fuT8ZblsJIt#_!4dcNf$jIx;u7%BVUn*Y2bjEi{Np9^P*&)
zzaw7g&mvyxn=0a;@di~ITXEA0p{ZuVN0o+4XJ;wZ=Rnvspup;gipa%tC+h_Iq9C3v
zBCgko{x4+f#S1Zl=4RE0)EIeunYES$aFYz)D8`A&j`0P|CaNZS1~XO5Dosn|t8kkP
zhQkj*(OI!5jLsLGhYPCNadD*&@N*Z^Eydid-Sn@Ns^JDrj7xdyC}KZ#xIbnM8n#q&
z%z{43=(iefN1bpT$X441Eq<DHy2Er##fVBO+sIEX-j=zGcd0j`csyT$0beTJ=I1F3
ztw4`HS{jh+$oy3G%rm|`bq!o)TS^}uZ(ElmUw-p)s|u)YkV-J9;DD^!n5h^KxG)EK
zrl`C(!{t5_&XiDxm;ozg`X<%=vgj=hA6I@5x~jegjyOJH169*!s_AQo`yCO9e_QxU
z!L@a&>rmsAbU07Y!Q%DWV_^Ai9d<}GEU}CMd@%xOjs;Ys0nrtDbzfP@CrFs_;!2cO
z<6pLV@pM%AFDSzP9b6AyuDMIXR4xxh${$VTKte#V&>=dByAz-WiF+)=m_$z3W>U5^
zlwJ+*xg5X^d?BVK{x++*Ta=n&uo#^M_r;PO*-6;F!rfF$c`Yl(%q^hBb~K)x0N;%W
zBMdyM#poab!}xm`#O?_|j{y)|6(Il!LF{G<F@f|-slHm!&?cM5-ly%Q=CfK!Eier7
zvZz9vDQL?*VON}++~1=oZHqH^CoxFl%umA2#V88bK{XRtMeess%`FHxa|24vC<HiG
z*+P)9VVpaNR?6rgMD4klYKo^HhpqC)G^@n{Pr;y@g*KA2Eow7<b^40ilt7?3sBAht
z17G2&1`#B+a(#v{e1OU6>S*3YKkIpF6wwlc*KkXWYB^{@#OguBLPZ_v7YU0_zZwC_
zX>*)U5-FX2NqPqQwM-{kf|^LG2Kprlu~PkWD94;xl2lbovWIUlmy|@(vt@E_yFSuM
zmUL>y;UV6M=W=3OFJ?S@KQ%Pz()7$L@RU?uCuuq}1XbHCGBvy%Jf#v5hwF$^-HRV5
zOxW-R_@#CtH?Fjp*JCe5h_iY*p709=!jly9e!*C!BO+6?KEpA+=*98LK{`FdOm*Ns
zEsVH4(lfp!^9#`!zKowXUIQ)*hY}Z1l2u57-zwGfxS(SA-adTe&&ZuX#rV;cH*jsc
zAYqS3NEkVto{@&K5&8vwA~QZRh#XjCwL!?Xka*8!TSX6HmRiE=^{PR?J_pHmxX!2%
z61EsfY9{!^+t>I)8cCR->}tGxhoJB}xhu8K(gA-J4G10|3pHJb8XF6x#zKL3iI)xn
zl;%3h?gagM+xjN)!P_~y;f&9qUtfTJk=I0I&DvDTny^Yx)`a_#LzH$<NI3Fx;CQ`=
z1>2fr!8CN86jGlR1*3&jrMd?ui!P{wB*}R2Z@3!u))Cojm70&i#4)%yo>bh3KmpOy
zcP#3GpRa|{2;8LCb0V0Fkm^ZTcqMd!5rg5x`X@NYui^e^(p~JgM`5*OB(Ya}o|D)w
z#V&q&JZla<EkbYz#&*<<L=)h7S2Q3)&6WIKsYw?t)49L$#v0}v2Zl<`_w+VHptGO>
zo0OWfXE;5Miz@Oa#7NUh^_)Sn@h5|ih>(!c*$p*+#l=AE^<lsof~zFbO3ia3Str@b
z8&bYvHVKVVGY8prHx^J6s{5gfMhwP6&Amv|TaS+x(`Tdg*PjTE(qE5e`$3b~-AWBF
z?4w^eem+&r1rJbaxQs=-CW_QyOf6Dsu4F1?PD?kXrshhD{9dJc>|jG=eQ~JFrg%CH
z-mVStw2)y=M|g7W$y!r&mt>rpsTvJ@4QRiYbIl8sC~<LWgUL{Kft25*Wo`n*Y0^6m
zL8oc0GZVrDBz`0}sU}!n5q7Q3+yNayZ3YjO!jNXFB{M6m`c}$04i`LS%I<VR(O`Yc
z`Nrn2Si!^yH(*zI%ntmGWd#`sJj=i^NV(UmsNzKwqlBAvBzPUJRuDpa^?ook`UkH$
z_R~mLOlgU9g%jbhq@Jf|B%!~u87`@pR10dLP-t{oT%vae`+~1tj8{7{*McFQfN7MZ
zPKbztxa!^eB74c<+T^JEwv}~GC+QZmi}DBHp-d+BaG2^;6S|#W1e}F^uA0umy?$03
zx!!K@y&I(d3V-_!I8&rFyhO|a6r~K`!q-3)N4G-$g)}~+xeY!eHv>00SG0H?dAtPd
zOfqPPtJdlI6bs#muS5;}T11$VPhltr@*tgYG6#<Movt8U?K6Qa1`ZaTrbef0Bg=@g
zBQ(n0zaUG%yGWYp)6UZ?o&!nB)f$i%u*w^f&1vyqZ>&KOLw!Kd3$d5QU3*!8y+sir
zCxi>%sjKCn`PB44e(RiUGm<oX%*6mys(%Cu#sQxM-Hn@I^0N*!fts3xf>-))r)wAb
zWTV4Xj|Y8On%1+f9K={-ypD>Haj?FcI7suPfBqY+uP&JPh<OQzt@6euhiMne!9|WJ
zRfN3x=43cy5l;N3pa&dnYqSwlt*TPvV1M9q;j4hmJjsdMwoV-(nGQcRL`r*0mX}|-
zJ~#l-6mY`(4d4$I4G8r&_M^cka+SZE7LWPrga}tn*{qW$5p%dd9x^~+{&<qzgF)Ci
z>W?Q2)Kw^Pd1Vs~#LIqWAYC@B3xq@02yrZeFGgyjXpHU;s1w|<9-f#B+!nxDIeWDW
z-1Kf0xG6Q=VmkIDWVJAkX<gtLTn|WrQ_vd@y$vqQ{bp)BXbL_X2U;yZ2F5}mCsZXO
zRM5P5R{S`48aE0F_{him06F}4=ftPur(d!OP+De7xDyW<2$rq`P-yo9OO2Tj%8>Ip
z3hz$p`}8vPah#LLKgs=#2&W-D2jMVfdqB@1WxI>v5{!0nwb4mXHpuMDr1n=LjTXX%
zAQLOuw9L;z)1mT7f<y{NtpXHGpOaK6Nx{Z@QXl}Nz#%ut6$Wx0+d4{bAEZMeeADlv
zq*xjlUIXfPaOO!)3d*IrcBz;#pj`jN)en4!C`Lj-R?h@}A2N+b@dO#ufN6t>NU5BJ
zj1~dBh$Kqbbwa~q<`Z?TLOP*@wSXjwCWN`TID^KrLLvYXred6FP%KGPow?@~&(C8>
z7`Pawy<lHn@fH{n*bq^jPOpeA$x-N5^hNZ1CT`3SZk?rO<1}IPy(TE2({#Zosg77;
ztpuPTTS|2Ss2VOfF_AXD$z=4yN<5XC_4v)Ivo)<26mELyABYQI0lvD9RKJ$sB%4n6
zb0QqSOpI!$X|>Z;i>Bb-gEbfpQ>_y<(;BC1Cz96EG?MBxO5=WEl1IyEOx<LMs|65{
zL*TV6F;#Rj2D~$Ot5P#i%uui($o)!<2{GOh1<I19)RY^f4?{Lg_s}~SFW8@R05g;G
zE2}PaZFGMD0Gpz^IZW_Lw#|{dNvY<&z>*8Nf5+xHU7rCRcyC{ghal+ZoUX4C#%P6I
z#~Clge5dK011@!Gjw?bd`v35OF|Phs>Jv4oOeJR{*6<#!V_cRg=4NNuhOWZkuEx8`
zU3Q1QCtD%b7h^iwwE^9>2kuJ6dnfXOLl?ZzzVVg7%n|<w;*V6WR<r+$K(<r5J#dRF
zWdt$9H%k)TdJdgqAgI&?5ep+Z@RCRd{eY901>cVZp8(;D;x8QM{2CbN#<Zyz65Ij0
z(}6xkeKdSAPK+i00Gu=PypEe%26`tx_b}4Cy35eJPU`#PW$NSAXqc_gp>%6Zo3KM3
zM*~Q}TSRw)C@qn86<Q)79-N`oc9tP#SKoG&^hH*n6G}=TEKH%U5zmSj7+Q%CS^;xp
zguhX7<5%N5I1e$84Z9>USd1hGUJLD=rNz62STv4vA^vqS@u)*DM=r8;Vt^+rE4|O@
z+C+O~IgN!gtkWM@>Ttiu*FlAp0;Ojpr0{)>+moS7n!bVY(JA%xiDlb~QYB#3%w6mv
zu!BO_9Z{9C@D{v1lqUF>4%8k(+9O{VM9h)7QT0rcayrJPu<XOgJgJ)zfw5+i$%t4G
zsHG8r4JQ4ig(fV&==iIqO{(dum?nH5O&ISfO^3-*G*OK&g{7Q8x@A*u>Y*e!KR10F
z1(l~)1)HbrjdcKt%1>fGMhawfTywFnaV4_J*NadM{_{?B6<-FPyP<PaUJb&I3qiZN
z8A@nbsiA=82^FjiYhs4_C6uzFvfJQp4_6Z;LUk3cxZnRlT->UxxE>G&UaeIYet8|z
zJ!YIS=sob-?iiTUd=_e?JzJ#5Z|Z)Gzoa%}9IgotXPhp*xd}lm)I7u*Fi0BEqhh3a
z@l>99DTFbE<VwU4eHu~SEcc|y(fL2ZT8X#?1ExQz#Cxe)%%#Z=!}>*(mRbwU64GA8
zCc^UAL2*%q)&j|ut38E~1~wBCDvKE^%jSs6f*~SpP}pKldS0J(L{Of3-Tipfc#!j@
zkvKr(ft_kIf~pA%94#7HUISuE^&7Z5iX#7X5KcLlDVXm>9;eK5JlPOY5>HMYgCEi@
zwq$=y;nWG`MUO|maWTt7j)Mpw#w?z!0vNCvA{X*`C6bhCjob#8)X(J=3U-zgqI||G
z<Wl(CjCN)napvwTPj_bS)A>nW*-^DEppKYfRZBMkAoO*ZUvPAbH&A#2dtJ_I(7Ch1
z1O))Zieg$}XfSj@pda0Tpm@TN-U1C#8}YXIoJw^H<OAxUIYJDC>LVQ<d!hQIK`>xf
zqQacge`6~SCM;`!4lxV;i|;8eRqLVQ8VTkkIb4k*9So~Qb5RJqNOQai-SfG;woJek
ziTl@j!1THr?N@4e%HQY!rDhCbMkgpWMW#sC{Tr=g*L{w*iQWrd4_IAa5c-ZcG$6uX
z>mAqzs{vW8c&=5de+*csYX^jzGY->NQoRsX<zoUVh%ldNuxs*KI&8SjmH47LnEw7X
zxGyeO$G0rsz_skh2A*Zt8K1WUj-h^%9|C9cdm=9ReYk(ZsDQNMeK2A;O;8jeKYuPz
z$?4E_KM!}^RWV&>iR!vCFnBrQI-nHXFSt5U3GR#pX+MmN7Y)MRQuH^z5J#97LCS}<
zcvk#0c$9oTW_%fM>oC5RB|gENBV!!QULWJ@ImRKUb)Z-Wzm1_-I}z0>)<%&dDVC4n
z5?CBG2<K@UgmK;I!U?YvEzTJbEqwM<2GPQmfH0|)VkAQi5VSxnF-ZD^l_qE+P11e>
zg`_@oqxfF%`c2`X5!Hna*QQhKk`}(fbX;?-g*w2A)qn)0+6v?}*D0X%vMl8?IGwSQ
zDJ@qdH|#V)DYIY(Gvox!I;u@*7Yt*97)8%#TI%^R;Ky5E&6M;_bDh^fjF_c`iP6m%
z82q5@7fw0PO+qn&g`aI9EYQT7uyLP(UKJ?`0Yo1`&zT;whv|_-4HRWMF#w&wA{2xr
z2lzEgyvQm|8=#xXg>qoB7)eU?{UA5sf+lJ!1Y+Z6NqY;NuClqdCa+LPQcG6|tx-^v
zVbn)fvQqae0$oU-B)vlK5UL24+KkDw7UnB_?|?IY5n*9!MC+i0xD&4ms~}@siNg%~
zlO-mR5aaU%3QEn22-$%t=Jz57;|u)czmFP8glJ?F94pw2DFk_o0^=!2eLvLYUs6r6
zh(a|)HLU^Ha^!AMs<{VKbH6JvlNyNX+71M9RgY%6=yZJ!)Gvx96D?jWsUkJ3fsA!$
zch7fa|3so*cB3vp=Mi4MEo^^L%d&J4KO@ROAaM>2f<3_~5lW$a7oktMJ{bJ!KSnX%
zojovuqEd4T-O~x0KhX(>S~ld8Xa;>K8>a8|!G9XWsxJ87G8lXuL`!~&;LOX@FHEG0
z^dRZc(oS9q@q?3N$qIvB$ss}IL{L(-6!Z$piRd5kX@xOH>T_Mx_u^&hla}XIye5j4
z=Y#<mU=YG-pZR((>7b-`TM-W1XV&T&w9kCNFs4qbamPjpt_OjwhI^I8Vwx`Am&n~8
zoTrd*!Pgv_n=oe}%n1`TZJF)`Vs=%vT9n_2Wm*d^ljH?qYi()R+;m2*`lbaJxnz{D
zcAcu)4@n>L0cD?+Vp6>E8Hw3)gf(hVSOip?aluT=^GJcPQh?j3>3zBPuqLxO=%L~X
zE9qPF#R1wW17@8zV;ml_F9vN6UB@+7fem_J=Rhiix#O%$nyP&bXgFxNfKuLACi6h&
zmW%f-)1@=Xy;RrwEOZle1u@52<x+r!ek$jyCglDoAy3aI7LuB+*SKQ#fFPx>TPwvY
zbg7SPk?|ktuj67RkP?MxVP*~NpU}`~`YrB5;=n=|=BTh{qaPVU%Zik0I%LSN1~Y(N
z0z0w9yS4}_Hl48Qcu(R=Tvi032mQGZf5GCD+cWZx7oXgdk%zlv&-R0#5?r-&j>jHk
z@I2!<HF35du13f2N&HsiHVLpMm{wDg>1pX)z~>*CYx1wK`A1&Wjc_<L;qzx_#rsz{
z{MjZP=k;eNCx}Q^cSdrB)8!_fu;yjM!XMJexBUyslSjUf3)C!uJ=^vNwr>k*xfhk1
zJ4tF?K{Z#f9^+fsaWyKJ6dN(ALp=?73AgmsaeDyTo5VSg7gst;bu~k(L}x=g9q;MC
z5>sEN56XL25`XJ5{LcTOqiwwkVXkV8S<Bs8w%?h##hJMsSCGTmc@<v5PEXP@>ztl>
z@t6UvPUxt3?r?bKQ9W%%coSucoQOO$20ipWo|VEIRy>|;U1>wYh)BY;m1crm5uIXT
z3Sb0UE9T9gwlWdAqgjI<iL;l!3uXTkJOiRd)vfv*mY}%D;^$X!_J?;=e6@!JAgsE5
zhNvzG@uFUXG^}Dch5|q+;#g0UG+@191$E%&Tnw|+@XR;VI1WAcOJX8<65k|@V1+_N
z;^0M(tRCmly2rDo%^Z(at!eFOgwK{%XJ0I8l2gv%I_;=B&>#th&2`3B^$mRJ%_<U`
zlqm1a9p}l$h5)cWNAn(Bn6(`hlQQ(~gUh4w?0X%aIL0$&z#Z6)$UyqxV=^%0Ry}2S
zJp$y#^&}Z=Hz$?8DhQ)XBasE&Y`{~ge&$PZUumr2?4m^$;Ly`Fxtne=!7*dmMm_0G
z1RbWGM0l35-pG9#5Ew(bhQ49Q%VByEMYZrDrqhrCEGkhKjIx4=NW$g7DA&maaKX}^
zH0@xCieCW&n!d0k<3(Kfnne{CL<ZGDfTsC~pD5P;36={zg|?Alv`=C`=UsRyZe+@d
zE4^LFYnVtI*zzLPgvs*7c#nChBX_e>^;1eaV$zv*F+Xn{oaAxaQWdwoKsJ^wPmY!U
znGSv<=W7}V=ES(t5t?f?Ub90rogD8OxfI(Gu@gtm@4(qdZsb(G+0j<3!p4fL7O|tK
ztkDTsJrf(JS}<F2EBA09O>EP&%o7ldjyp_EvOMS6O`M4%m)`10fnT=AlSX!DynLyd
zc_WkX^<*3ehhyT2?cz!8;wkOoshq(hll+*$<xBgEi^^d;tMaudHID(k;IBbZmcjss
z&&S?Im#aiM1O4`Er)eLq=D0ZX9FDvBM91m$Dv-ZzLtxalt;MrH&dA$YJUft)*HAqB
z?-_ZUi)X)+kynen!^N|o&&X@^+blT1L<bh%ugLsO_H}++>f{6AxOE_E`m%WTpHYk7
z7B^XVzZZ}r`-roHVaUXOsB^pijuqu`xFMF6E<~l~fqq-sWa+`4FX=xNl^&6=FJ4g^
zH@PkhnmE*NOPYKnT<hR|fw{@ko_-U*rB5%%ulDq2{FXoM#qZRoj~DMPw<O!2ZbfXy
z(@7$R4JlQ=v_N26D-c@i$o&+%X3>RC*SA{kG_IrMT|}?`13&uiWk0NK+o(-7fH+6C
zIE+Oasb1_Bhb*MZZOhk-d?DsS^a$C-{c&Qu$myM|O>P&F);gu80cnxiq{R-4BuEuf
znUC47FFNej3*PVq_QPYiQ(PNHAcWHpwjvxB*RIqvD6V~v;TU3#O#@+ZO_1v4n9gc@
zaV<i$BjVcowS@E}f-Z3vHr25xf#{s-(QXo|)=q*|wI;Qr*j`&&t$i^~ctW&;_(0Ll
zpwy7-NCh39?3*JrIx$S6*B_BIdY^m_xC_@}3;8a67D6P}LvB7qfo?Z-z@ku^AhUux
zQ?CSRK2EHpDA%-uo$`@P0$t`la7eMber`I)R#g%lNun__$%`aznleOCki<e>1Panj
z$fcqLHrIxRlIdcfg)XE~Wi_(jmUcDM4(fLpeRl}r{IU=tVGd&5<N%o6VjEzkcqKkl
z0jIfK5z|0H5znKsk))W$Mk)X4mAg*Ldk6Wsc%5Ay^me}rlL9P+)3p&;GK-<lYyeG|
zFgZZ(fw0%ZX*xUJlU?Y<LRQs%Aj2A#pH(aX3Y#7P?sewYDpg~n2{d-^McMyApPjA~
zj^-`cf}m69Tb!=dps70?rg~84Ld~@X)Y;%?5N~e8IvvTg!-RWc)^#I!4hr(ThU=aX
zkHp^hB8MQ<gVAda*Dj|)rgsST5Rsoy-%SsUc8x5&)ibiY_&`{*GIz(CmA$Z=i02~?
zNf?`xM54K~L|L*Oq!8#W?Gk%+Ry<K80+o?Dz@@yf4<tkjOcppS4`R;!7eU_d$?xOD
z9uis&#Lf<WONS{083UyJH2%(moUg`T9hi~S{{+<r?UvMhBdEFGc6NRP8;P^#^B!sR
zk2%|~0X_*Tau1mNreyzM6Vd@8!`)=I<O4DZjt#=u@Osg{bV5{icRd@`T1`Eeu`~9?
zJH?B4N56Py)QgdgqWw+cm7D#hK?sXiZbk+p3B<czAgpJLSIX=ZywddelLdDs@nO9a
z95N}eaoZVC_=}+Mfxz%>8@1f+N)3&TTtCVUiSM9PH$yWH-VZ__CGj4vm(fb}IjH3x
zM0JU`UgSvXx|ZQ6BC630NCE?icMC)g%tvrwN686CW)r^SNyGL>#tcc<CF$nlZ8q0t
zxOTu<)O}b$XKD8gE?Qu2qmo0c5Q`St9r*gV5~%<4#7Xt^PgYGK)su3Ykem*wCM<iY
z-TB5;Ofz2kifNk`8q1xgBpt$bs4quASd*ivjx%T$jbtDqi&-y-2;L4!abv`a_|+D2
zDp?KpY76pCXtA{%6&R!j;_q$I^IgN5p(KBxOM>VcMqCJ$+%|@{L-N}|YXI!^u-eT)
z$Dk!)A@wFq2c4>kB{AgFQZSK<eGw&4$fgiOL)S`s2-jx{xczsFYQjadkomT&xhJr7
zlGRkT#m^Z4^Gc1E`vWXW&C_T?XhRfzV&PP52zdy*maufO6rx*_JsrLs<_+-1(ExA1
z?Hn#OgK7SU(P<a369{R8qSN5)gK5`A(#*Gk`>i%RO<y{4KQDXDVcH(Xq?+pu_}1H0
z(;?8q@O%!OOM?0<Uj8bq|I_gi1DXj&?Db^_f{JKUSV_y>95ZBpgs47buVIc5Up-b2
znhgFW>@-Wo_(Ql*Ejf;xzo~Q@Ow3Vo){)r?!hG5ow6;uGQ;uU(Bte$577%_|T3AWc
zR(WyOhvQACuwX%Cl=?RLPB%tkkEz@kiP-|nD>g=6%fNK8D{?o#eQGG1y3}S?DwoU~
z8N~3$@fl(EovVt)ay4xC9{d@Hja~+|%{A85xL33gmQ0w?Zh8Q^z$u==kV6zIjlLf4
zMfEgo%Q*tmw!Z46Z)#Dh>0^@xR}v2>#%v?_tt`PQIIDD;&`~&?Ihl2NJF;P2^^Cll
ztGnw@Uc|RS8#U8@M;lgLj%m3V-f}&h-3XP#;rfa%VMoK*g({VW9510X*iWTi)(ONQ
zKybI?a+DHF;*W|Y@nQHZ$EmboN&HbRiHiu9#2=;2K}5uoIGev1McJ65@OU)#qnkGI
zNQ>Tt9%ur@f*%^0TnT43#RtHovYGo&T;wu%6iA*9l4gK-0TPuD&8d}Z-5GE#z#3MF
zY#<|gORj}$;L}(OKPA_~??sD)IJzl>Ukbw4Z<JvQ!ara*id3s5=ScpOu;mo(7JL2k
z)RMR5A_%Av_9?)}!U@dDT6O{Wb1R5AXA!pw!wShjRYQYVoaCS?qG$%w%`nYVSOgx8
zE4S}?s5{OugnrIu%;kO<8vFUlj<aJid^2!Ri4~|M>=Z#kggKP3|Mk<b+^(zHfF?&6
zR`C`Er=+;~e-&IrbFbjIaC!!hbecZJt>!<EroVi%g$rGHDF->XWq{6Z6#Jtbrcb%Z
z3t_VeJx9<y%>^M)h}o>XNGS6$^-20V9kdG@nm}1g{~)kQo6|R|Ey&p*1g`M@38yFf
z_gEbX+sFK?D{ipgv6&C;js~i4bp0f(bt#KP6b|=9)cmXjI?l(`JK*RCusbIo5Si9z
zMfCXOIuT9$qu*DaT=2qjKI|{R#Z7tgv=<uqNPGb#Hw_krvTEr48up(jDb+t?bMZNn
z)Xa0huw3&bzL+KCEr@R^^99}hE!6L1^ZPWY=bkG&sG80P+c31mMhQ)M)r%?EJudE6
zkGT?u1kS{9qlA4m#Qi7HeVetOxT%5HqmQDHjKih8QXD&2gE@%1j&rXD_N+zhYxCE0
zpoZ;hGZEEE_cR1C-u9?sO9bcutq&!24%1o~&)@@u<4|xw&TB3~h1VD5V2>9E{y5)@
zKu>YsF?AA@$-}1ZEbmgrCmK6|R8u4OofJ8EV|9RBHkIS0{rMXZG~@h@=E5w8ykcwH
zMuP)ZE%#tQn4@iD*$%#bBKA{}v!`;~kr#tQ#Zje7Jj)g_8w1nJVU|aOdZV7*fTvPZ
zz^OdF4&C6|EEHxNrYVN~tT<zaJTnftOycCe31NY%!;_zcj=;s~gBPJQHgWFc>C=8O
z=LycTYT^VJ5I0T-Y!+(~PS+`C=2kE~EEc|OtpqHrJVFL@X70fLaa4zvQd8l+&CSfO
z`lcasGdQ3Z5oGAQ=<!N)vA!DNG3OdQl5=i4d6J;&${XK0Jm@ldxg0+H0TK$D_KHHI
z$X8fB3B+8EKOlNs#Hp~Fu($CwNz_e#>WUGXky9H*Gb9W@d?BgR+=`WwY$(tC8DkJ~
z+9<J~3h$~Eb?ed}>63m^1ss?BPrPv&MJl`wJ)l&VaQdJR7Bg38f+QTn*f<uUX;U;3
zEl3JEV&p4+c29OH_BPpF=Wxidjix_AdN^RjIXJQnjvBlfFXi!ERjIyB%vrI|M_l3V
zfGa?@4$}?g=JeybE#4M<74MJedoW52gB%y8N3aLnV4p-Ms5fYn8id<9TA%81dqT=n
z!ik^}H9sSPj`Puf5=jqfjwC<N2;ZPaH1U<dh2(<gL86lF&!^#c+Vgq%)t)bsv}b$S
zV$Bs4=;5Mi=o_FTnyG>P4|1c@&S!oha4s1;$svE>ChW4t;8wus^D~v>7o1`z`<i3e
zft@a{HFkY>HwkXIePYX)n3GAOV^K0stop9AwS8K)SIz`uhv=VzFHz}nC%n?VJ4Ra$
zyxqKfzzofz@89u&V}$#Y;fSC(4a3FT`28an+Zr!6l0_#x5{924;VUG(@4OkeqYo7D
z`wjzsqlCXH3_nuBJ0<+D34hl$0{)+b|AU0@6NXQe@B<|L_9WnQn}lCT_+Lr*uM!OW
zzr@KI;C}=U;QwdB-zxFHjqsx+{6E6*uS@ukC43p-|3}t;E#Z4h`1xV@atZH|@cD$_
zBinm21o%A%MgNQs!)p@$ZVBI)@Y`hls|o*i2|q9l-(SKHk?`Lrfd7}Q|2e`}06yk>
z$M}Qbu}{0ZANE=IQ5M}dfCFX6s!uEE74+53sw*45Fvu(@P%Bj{5=dcgNpg(1B?-&p
zS98|hlj<CCPpVS$jfl-pbB>sgozI^mrn(=?aEy2iyQ1qD8(~A4e-moN6u`eDLh}Io
zH<9}&;8rk1ZWDFFm<YB%VQjX-EdsVe_BAz_<Hn6J*DE#GVMC%2^3<xwhndMC@HQI~
zHTR=ATJEF6ltm>XX3KM8Pt~HQ1iVc(M0`3NT(km-t3GW*Av9j8T8*eZHyPWHSrRn6
zG7h^HZv?Bc^cTeYybb5pwG!z#WVAS*Y->L2T8-q*rp>v>73bRA)rv=o4(Lkv_#Ad;
z5dwJ(6e0i&aIF7UPvUc!N_$&ypwOKtSP)Ef+9ogs!HgzgLc_eIf_GM}l|Cpu=C?@3
z($tbexLz5dj2h_M6dFOgp<H8$1qUE3>=vQ8O~kv!n(jor8#1BvEFp=m3-(sCG!dYg
zf{}R4{gAf6U_@01uOa!Sph-UE<ug;*Hdi62d8WNpghRjj{sMi&3oYKD%2JA?vIXw2
zoCQsrBoWW!(GbBe<hH5|u3SRs1^!BP9EEah%aeOZORqwhB2gQ*rHOxG`sHdQw{J$j
zS%q;%t{#@`#1<Y%L6FiL>Gm2&fqL#zeZe%<CEr=vgVmjvjz&)LhU!Bq)^}++E8@XT
zvhG6Y5eQ&KNO%YM+z~I$))-z8b7UCRtz@IP4<^R}M<HS@#o^k6!0UpgAxn-O1YXt+
zCTqB5X(`L$>p&P2{E3~-4zlC*u2Ue!s^*qH3v5;A)4;F19O4)SUk48YOxF)p*HI^q
zo_g#<P?=of)J$iwgagrPy_ySOFS5*>>oja`G##fvMV27ybDeRz0>PIMjp%FGC+so*
z4Cxv+{&?&?ig3sTpN=1z{0IIb?TeS=S6l4GZ^7aQ{PtfA$+^(J1mbjQp|%9#^o+uS
zB@m~l74~03p;~0Y4=}SJcQXsVf|qbTMa|t*_CGWeK$`X9B~WEnfyo&5Nt<hnFeCV|
zM5mc{i7VQ{cd{2v@Ysh#;A(VwQYvxSL70HC|3`@83t@06G(W{<H1kuC5O6ho*|%tq
zT$`JM>NT9c$@Dor_UezECTy%cf~MAB$-v&De^yg8e}p~7{^kL2+(P+C#zUkv$}~QZ
z)+o{r%QQZac37kxmuY+;jeZyHZ=RNEd?4+#NP`<>zJm{>wTQF}GK~+UT|k=oO(u!o
zWb>OK2NoKu#Qucf0$3&UYwDHifo9rQ&S6um2WMq%?(RG-I|1Zd_;>j$(n!Yzt(C<5
zslFtY1~H#2Vg@mPNW{ixI7f^Jk)Dbe2dza?bSE<fMIR$Vr0BUK<N)wb5Heih{6riq
z#%Qm~0bcJAhpM^5x~XoGbFTB&Y%$W5nkEF@|7CVRX@QV@J|FW~bK8c&1w<}fK(s-j
zF&)>P+&HRQa3uJt)VMg%b<cX9_l1OSfrS4F!=MVPdlG1$v1l?0<WHAF-IG*?D>39q
zUO{6g3nDoe0YE{wMgSmK>EI;7qQU4KA#fuAwHG!p1Yk>`)}kLY(zhYG|IHbumTZ!w
z8S@x((C*rSv%n4*Y$-Jg;_dL{!TL)245E#LzGFp9Ad#nfMvE92-2nA2D2t#N%TcRg
zG74dztWU)Xxu6mvM>jty)oWTvJZlZ&*$ZJwJbSomCAFmS?j-2r=gi6kt6#14{6acl
zilvKA*BMYUm;j*P2_jdhmF#o6_9IbPQ@OYLBWooPlT=got<%*20I3xq0A>U*>_R{o
zXUD`BGOMnUR`vqyDUycHv0x4dzeZJDIO2YRR2XF^ct&2Q#bf#3VOr-j?csJOC>Gd3
zq_4+gt4g`;9A-x1hYXHA2PgZ9AHo8f2uI4WgZ>}t&ILTG>gxL$2@-8|5*szPsH2TG
zv}mK!N(yZg3_62n6fM=D6eCcjt#3bsq@p4Pk|^UiG+wG`twn1sty;x9)z)wc1P}#8
zLGcn5H8H4w2m!f#zyI3jOfmu6+V_2)@BQ*TnX}J6>+G}7K6|hGTAM`5wO1bM4c?Ju
zkXl1N<j&85-02{9cS7#|9LU{*99SLz4m{&?;3W7p;wCUZTpyQEQcqgU?~A~p>lXF!
zLaz@=O&W}7gw+fj4Xck%6Cqbze2;=6S=;hJ>4Q{==+!&<cLkRYl72Zisbbl}yT@1N
zCNth2J#g+B=z)pWT~zZz+d|M^EJ+d=m%@sbd>-7=5A*3q@~px*Xu#9lS?lx>yTB!!
z+pLHRxu~M;I6)J{wQZfF9-aMhfY4KPv&uJvuogtBReFg%)EGS_TLtvgv;w9a3y35$
z#Ay*@p%?VU9m(}Du_NEHhBg~rOuUaCffH4Zr3OPmlYFx>Sctz{VuaRU6&#jkA->!d
zzo+?!Z1@MP_}=J_XksmWLWkSh$u~&n#wZ=d=hF)Mq)i^|-LUAmJPl!hMMv3LfJKMd
zS%5`P=d24BJyjPCi=JR5VbQNyNm%p<N&yya+oGoiSai=JG}8@>e)Y?KUAVrC;=heW
ziSHQ)ys&7U9^zxsiT1$%zp!YoaO&PzbiW^cF)VtG!iWDq#iC38Z7kXae=eLlzS6sH
zO9t-P8+SS=d^Njc*bKbspmy#J)fHz7DEGR(VY=YU9|!ny*uTV=Q@$v^G*mey<MA1&
za&_CApvnqIGd~gXt3M%EOZ>b@EIBeFEv>_n)rXkIHoEOIfT|B|>jjjlA-B1^wtY!=
zon9PIPDUf&7+@ssYBaj6yAEqx2~<U*(Fn9XtdWW?SYnP^84!b0`jeH;0ZaVOO66dQ
zn<)iYV!=jX32Cb(4u|XCKS&M1Xn`wmPe5TGuJ1uGA2~LYSBS-035$gqkMx8{cT<=<
zlwpCSP4qJmj*Ek8OJblNU=Zw70s{4;N>P<6qcBJ+zzz+vYY4G}>{1!5TM4Y&ZZyuJ
z=8Dor@)I`3YIkSL3TzZG&D{hJ$WA^Iw=<poko5(!Lwd#3`yb<Y>NvdDD7x27K4J`4
z5>lD`n=7KS@RHMla!xt13iy!DRR0QV-#(gzHDdl6^{(Q7oKOS{R;99p9hfsKw+bY6
zx|Lg0MpQw(j!@TIun$w`r-$s%Jyy5Y;5q`f&gxbe9NbI$08>!->>IuGhX7%3mSE6V
z&(!^#H`uv91a%uMj<?x6fE_zpBh^3h**fC98=>Xi0cJs{9wxf(tOsOHA;V2$FyDR{
z%(qx#O_#|_H8*JTCSGhm+s*dP<z3sXc_6J*Us>L@$(jd)30vN^NmZkS-}Xu3#ZFd2
zx{SQ)0|Q>79b#_GDCUPo;||L_oR%bj!&5&`0h=b=fa4nF=c>raDH;&55S0MP5kL&a
z*z($4tj2*w;o2XEPz{<JLr!_0k7J1qv2CkK+!CvOzgKl7X2bnuK1=m`NVbPaG&!({
zIKOCU4K5N?;Wv$wPYBM$OZDX(e&Qsy@x`A1_bijTo|YVz{SP$kvC=&???ywrWI`y3
z)$YXO^Xnj}I1lWzL3!A4Oh{$_RRNEhAJHDYc@pWua_^VrLenE2nimhPjHL?x9L4cp
zljEeVCSM~TKC-@PM`gtvev$h4r3U$u8h)uUS@RL3OKSL~#w5*0kS(d<ml~5ZA3?ID
zhF@w-%6tU5k{W)gF&Xm_IAUt}rN$)8N2ti2!?}|$W2IZd^@<*78w@s-q!eKUVaK<c
z_l1m%p~V)2xCVnrjNtgGpH1O)!qX-QWdq+|o^GKv3aRX~`|n)JicQ@u5V^H!g@&gM
z0TfO@<n?g5H{yJQWbdMQs_4sJk9*D36AQh@bN82FsD*oYI39ZVTTGplSMmE<I(WZt
z)M41~TXg9Cdwwe`?&K0Wi95N3PU223p_90iOZgRdaS5HoU0gyZaTk}+N!*2NRm3ZO
zv*x~NVppe5qJB0n-*y@jwvf;$Km1s7=6Na`vfVcp@?}FG`?&0&!ofMP_&mcQ9?E}G
zxsf-M#953cq5z+D9~IT^#AjUyGZemMeq{c1H(}&gial9SN*yw46;6<g%4^q?p39-J
z4z=c^u8r2REN-dOVUV0BIUZpLUAVxxLr&m`<zo1e_ssKM)=MdfAlM8}ond#i(2Sr?
zsPmz|{iXTY2yA^H=R#Kd_l4RbhT8n_<$oKy9ZU$cv0MLL>_%8x7wmR!fZafP|1oy!
z+G5RLedJ~+4{U6LlCWE3-`LHi8Yzo|4284)1$G0YT_#d$1{{SY<0s&-Tr{)-72Xu*
z+{%Z!OqMKGm;9O#qI(yG>kivbi^}68<1884Bz;+G;8&$a)G~_^_ZfQg>{M^GE-s6x
z7g`nzts9c+{arb4JY_H+d!f&Crmr5C%>9BEW9cj9;hPqK_BdeQekFX<yl81h87G75
zJ`2~c=E3F9az|EKWV`QTNr&pEGcSX6t=Xuih^<WZK1z8wG$&)BWflx%j^@%}YASb;
zZ*jz1hEw;89AD$O-Z^;Vb-vMj7CXB7o{t*;JM%`DPr2^|{?^}%S2wX0e54F*CC0eG
zDSXAKyD!T+e-(9WN8LRp%Y4XNM%_Ik%Zzrdqwc|*Jh#6&rc?gL%>^ZP`#slu^XA?;
z=ATY=-<yl_s9rJZ<ZDj2d0^qf@?ty_PS%|fgC2j)u{Rg6gA*?5R^7XJ)G7ZJDSRt5
z>XaWx3fu8*{C43BI4Ab59P{j~g{K`?C{M+L$`Q}9!B4+R?^#?i`&Je_yRaPRgr{fe
zFyd)0fwP|G(xm)?r`ath?T{WV-5#!A4kC?Vt|ohFjxcDr?q}ev_8vrBN&4Ql#vE1S
z_3NuI^AfMif*pQ{NOzH@qoScUlnisFWFQ)VSRd>XP{WnX4=V3;oDTKHMk-Ho@s?1h
zfyssQ22~D#VvzK34`l5VeB*20LxNAw1Ub@$r@cwmMWM<Z592}T4KMV@kW}tJ0$aYL
zh-aqy@h~69r+(D~Nt`gJGxnpOi{h!XLmZ#yf44<YRe+38eN4J_%fh<{RSxVgh3}*r
zIZ<7Q)hX7WI!;u#)9Q2+b(|1iack3j|IUFo;sKiNg<8GRRW*NVJBi*hZ0UDsd)nw(
z&wJW>vdConoVD0&hjTb!J6!iBw#nm;G27%}E3mzVo-^Ct_%!@D2S3%pe^2ngh#35P
zF95%CnZXZA!<RYua~%BDg8$gTKOy+L9sD6__&f)HjDtT@@N*sfwSvFg!N1)a^uJ}1
z^?wrw`oB!@r@H>1C-`y)|417CVFy3Y!9OMV{to_l!RI^pX=(T=4*o_5f0y6~IQXs4
z1OFD8i?I!#n}#3i;4gOYR|<ZhgI^^0IS&5VG<?LtpXA_&3EuKdHn!d>_^A$l^ZUX3
zZfvsm4LSI43H}GJ{YwOYj)R|{hHrH6D>*PeM+&~o!GA~a$2j<#((nxq{%!|<(Hp>*
zJNW$tziElR@2E6<m4m<B!S^9%Z0@fe{L9OMpXcCDPQ#Zt_;LroeG~B4IQaVnf1`un
zFAbmN;PV}P-s`|$?dHQT1%I)FU-e!vJ}Va5_`Jn|_x<!W;3qlw(*=K$gTE&YKg+?-
zaqvf~{oM}!Fu{i${P;BdI0rw~!RH7*%k}>~mP+f&W;g%S@MRAE90&i2`V(^SPYC{Q
z2Y*N!KF`4)<KR<*Z}#WE;4gRZZ@(M7Zwo9BcH~VQc;AVFU+Un`6MVUYe<Th6u!EoH
z;71F7nS(!G@c9mYS{i<egTK+i$B6fv+uQYjD@&#It;KHsr{RY>_=_F<5rTiuwZBO4
zbAWHiu-}__n|QGzc~p7ov>mk_2S_r_F-a?2D_3PN`HMVnz)#U!E?4rN<mrXjZIJ$-
z<qU0a1bqTpggJJ0da*Y^nuI?H5-~Xvt-V?~B^T(t_Ub~4Pq|`*;)AYOOz|FyzUrZV
zehA>H3)QhqyFDG;EnCgBpCU?N&?Izp3+`}(>(`Mt&FP-$X^fuM(wz2S@CKYiPiyS#
z7xeU}^z>3Y8%a-x)6?Oc`JJ5Xit6M9S5zm*x}rMyV~T-(#8P&`VWA&EArh{CE01<K
zCurQx)t~6-!2a|B%8}PmU<{D$2@Qg8ep-tTV}aDA2k}CK34xuFEnSt!TQoTYt_=GH
zg)-v|DMVGkk|c+U<ds|(W0$;=>yl=ZS8`oRYJ3-eXHbeJpLU-ZDXSu}+^H{yA~@H!
z934%x>nn|m_E~_fs7fq-1EU^Im0Nkb%INKRDDG3`rh=Gab!a=J8cGeK4mW3WSd#-M
zJ;8?Ec0`Y0Fr1ggZ~tLl7O#fuhnO14`d>~*Wmc2chuF*OF0xCP6O*<%eBCsi47fZR
zzHT<9z@=hsxONGr6{San>o*F5samYAl++GDzjF}ZN6G@*=;HhwH17;FJaq-Wd2Dcz
zrwuN4m~(~gaA6uGxmWxexr_v{GW#>`XW6);cXBQkWr2$Yf<11fsh*S9^d$XOE*hL0
ziZmLrq_>f0;h7svUQrlJT_f*`?PLr7hd26FH{9a}*MF!n8XBH)WiJg)tv*T9j=Zd<
zjy`V82PZzNNVM`z;^Z3QJ80}<eHJ4p6*e~Q%y(w}Xlleisx;1p?E|~;w{ZAvUg*PU
z=p(t)GkXf1cf>?_Ly)7y(yES=1(AUId*J0r7ZYfJzO2+SsCss^=cn!W6^XUdAd-5X
z0Wp7N6WODqTk&t;<(BH6KFu261GdW}LS7?d*^W)wx!QX+$kSBDj^gM{Z}bOvq1k7M
zI_sV+Ex3A0DCzHfBBCzWM=a@lB^k*o9-?n2DJFzZC?4GYOF3KXNsZ-%U{5&Q<dmk4
zV&6f8?pC#rSH4ihm!_S0&KvQPhNjmlD{5!)>(w^$JFK>a-`-?0%U3l%b5!F?NA=G9
z-nHx?itFy;8e;GX#C3`@zEjtSgzFOUaq015^U%udI#cY_;awWTM7E*rQXm5V3c?4-
zrpMqPC+;<|(L12k)Nfc-Nof|_wyV9D{Bz5<ed@`oD*x~0x@2SQ9k5bk?7U5O_}kh<
zXT}p-`0jF%^iw4gPQmslPW>gf@W5gkVIUbjeZpd<TrU(|<Jl-l&Ol20^gfyu`^F?E
zi-$Ins>SQGI68HYrgpgYAaF6+bQ8_xWN1x%7TZR6OFJIuy{d=doqBdE_v<*3$5MUa
zw=YvtQ!>khR>o7gb7gzYBOFNjDEuiZbN?)c*?Um2B=OSLnszVrF?pFaDX~5abMVtY
zR{I%hu;)wwX-5wf54|rVUMlxlaro+z+)qL)qN(hsq^f7<6Y?#iZ1NhU>GE#mm$GzR
z-i`cHmUhd#QJOBVL3%CkM(Mb`25Gds8>QXy8l=neZj@fjYmgSpyHOgg+y;udH{wp2
z6C;zt?QQ+&mlk9*67gY5dSS>C`8JiEt0X;IvvDPJQ})s>>vQ4V)s@50FL4-weu={v
z^h+H2O23ppY^EuZM$9xN(wLd1MCv=!DUsII;BEb;Oliax)Jq-K_8{mg*+c#q#VEAR
zALEn*g;h?^6E+ojh6I)zVB{GRD0Tm_c6J^lP%8fdI}3>Q{+#(F@SCnE5*Tqsk-$S;
zQ6%s{iUA3%l#75!pt)m|sSD1K2o)rIS#XZtUBxslG(!Qjuzw~kbl4bL7*eymZy}1Y
zPf5rfO(umc!%yau9B^vs{=%EtH~K{KDkFlCX!0tsiXwoMSAl1YEaI2zCC}*Zds&=O
zra}odC{Lsq@f|FDA~{nd2T=vJmM2=vEiCTcogyt!sGNCG*aL>9JuKvmD@X5~_Ae_u
zi1p3mI3<Kzoz^0ETBu-=?+6z8=BVt$J6PnbR~GpS5Tn0Gfm`I;-6G!<of%6gABPb+
zrvfTrkxTU+=&zk-M_zJT<PjUUY<WSfZ+k?mCDr$EDWS_^p(d0z-#4inbq&|M%unJa
zy`Su0eO4AE7QS<NE-D1P9#)yx!?4jivh2Df-F;r)>8|hoyz9H&bqUP-xW8aDjtvV*
zRq}4Qz8dDovhv^0H20)w`iw*;qt$8L=Q>#T=s2si?xV`^m-gSofnq4@wP0>W=NDXg
zXu%KhZJw_<jMv)ESDWlxhPy$Mk?FcVMYncCsXP4*m_tgORh>br&^~YT{SCMViQx;%
z%g?CPatKwd$DcJ*p;r}&5B4}xC5kN4{sArq6Q$gS8nZ=6K#dD<Fz7;!4^2~J91PO$
z7tmvKPBbq9-$St#Hv>9=o84L?#+!^7gHns*i8sRa%1;U{-r5#1t}+8XT#hT3d^H?;
z7;R?(diar@1?b@n&V2N6nJWrCT<nTM59hg}(8HM&1N2au;b*{Z+ZpI#WpIwCmWuy2
zdU)H<&MoxtvdRuUfD}Ta`!H}~Dzz4pa@4}KLxJxpkAFsnHSTZ*{IbItu%QlTfD<g&
zbHG14oM5)j!HE{(1TwmS6R<A94F@;@rYgV?=yMKg=?8-PT-0!!`2+lqpoSp-ad9lM
z0CX?{zF(-pWrCAw>vK_q3z;HV<!}*gJoTH<y`Lg9(I4+9)KxeEdDna%KDcIw_IvlP
z`8<4Z%}#gy^YFnnyWREA!3V>xf$z11DCC))5FmudS-hYv)Pi8Kt1}V8!~h|r*UP_!
z4@S`>nxn#(dqO#-i975uljOh70WQDSb4bpo9BZ1j8=gUOz5<U@;9Fk`;NN@H;J0xA
z{=((J_jB+g1b>Eue<}_Cgo9t~;QzD?_}@B7zOUd9ckt8G@Yg!{+Z_C7>d&1He&^G`
zzyGB5r!ozHo`b*C!Iuj@=KAxT;GcBx#cBBC9sH>dzC!SxgTGVo*E#r()xrC2eZ=0k
zmxDiA@Nox!nc&ZN@XcxXMGk%q2gYZI-uEmAKTz;rcknaQ@V7enKRNi91V7Zl=L&w?
z6ZXDi)9{x#_$wUz0>Pi-;9vV2@QWRM|1|t}9Q+v$zNg^Nb?^@e{x%1nlZM~l!5{A6
zw=Dra<lrX>{!$0O_Qha)UY=v)^F9aO_e{0FpM&=Vf2xC@orb^9!9VHXZ&Uj(xbNFr
z@Vy-T#5DXb9sG3;zI`e1D;)fWxxlY^+{P!GhCkiGpYPzq(j?yI`tyw7|K#8c)9{Bm
z_^&(oH>FvtbMQ9^{t5@*x+-|z_a1ihp9Al^wHf$7yZ($2{231ZsWki(4t}wNUnls#
zIQYJTKit7jPs3m9;BRyAJv2TAu0K0jDy{Fc8wKmJG7W#8gTK_lf28&gbMVgz{z(U4
zoQ6N%!Jq2jUlIJ_4*pKTU+3UERtE38^&vO^9sC5rzwP>enc&ZN@XcxXMGk%q2gYZl
z;MY6&fr9_KgP)m(ztzG2$-&12|E`1275uhGcLnlfhW>6g97pZ<kqrh(vyfP(jog#J
z%tLvC5YUslxKa5t5dyYBXxWK85*=_RLO>g5cxs{tyAVCtI?mWowL}{;ozVUOeI8KJ
zM5}k`9`f}jc1GD~4??pDZRgJ)<of7>Y33N>xLi3R*}kjM(#1F;jpjDruVXVc;rer_
z&^CSWOxv;g5n5vp{CN~gMf4f_>6Pe7wUebxJ5ELI066cx{h~y{)E8LAdAb{$AwIA@
zsNl|(vC{LR;rdU-fs8&4w-xyjjXteXg)#~wROnBkDw@bHs=azlR`ven$?T%`T>VOf
zIxv)#^=AaCJvG{mP=lGNbBvAlin0eI7qEx>`FZi=MR>QK9f>7>RUAv6QKIH^1}Bdn
zlB|B0c(oW_FfbbE1V13Ieb5SeNZjTKHbks#4|(B1YmHDNfS}ilCz@lmt*ULLmhQ4M
zPOP>~6^f|!M2zQoIDN#GCgU{RQhykwUTvPHXw%c!zN?3hks;jP#HTzPS;u9s#Te(M
z4r;nTJB#ghiNHa;u<dqv1y~~I8$=nd$oaHa53=3O+RJ=pVa(6b8h)sq5nsapGX9%8
zP3z&hix53Iw@=<SoD-=pHpWWHd3MHpR4CzSZ40LhBvM7fwO~ViQQX(g(-+@JbH9|H
z?3aalClw-e^==cJ8q>NF*DGH)vX92YwYRGq^5wFWIn43DQbChVL{N1gWPVRNzg?zr
z#cPat7hf#r3)1Mz11t|7a({~P<gdlJODU39j%1d+)tx@7gM<cHK_}jO$$}z;5FCN|
z@DvVFp@i5b!gWjljrG|zBvo+;^G$wjrbtD>N!4LiHB_ZQY1*ubfSZOit7u_$A0?cK
zIV)6n&lDctet3iL3o%`tLEa``(mBjn`x<Vn4wjY{JyJ*em!xlTPEi}Z(QCcLtBfs^
z()ZAh9`J#A=pQ7JXIJ}ysIV&Z>9&1}s?Lv8A_U3p3c><{p<K;qjzLGG$RpEt$jy2t
zUEwpjY@<XLg{QTu#`J~cTVQ9z5F7C_GVxGLEYuNaCz8abc%ON(soSRj7@nq7!U8BZ
zt<zf5IxTY6|EP292~X2QNWk@7FNEhdqe;+wNjx<WZyqJnAO!~wSVQq-cm6xqN{N!^
zT`MI@-mq(>M9J%YEx(l&*KrB|o$I)S|IT$>!hh$wg?Hn>!(AM77YE(Nq4%{hRDeqP
zbYC1zCG%u#R?Wd}zht7g@sc5&SIryThaDhyN4Jsh<qdZF=~aCZnxOCU%4eo}Kcz0`
zkU9~aSfRxrIGq^I_^|3Heb`n0ltSZv)Q6SXe>NrJY+TV}r61Lx`})|{k9;krFdC#W
zLp}DPef&4PKD%R~9VFSq;p}gE+C;*b@RM5W<Yd#2jf?geSmd(dcabM~Ud3(FlvOk+
zDk;C>CQhi5!mOgfMI+tB302aYRW!I@q?^n|>^7;-DjHlY(oLKIB`sP-g9}Bv$(+P)
zlQONM!9^n7WL{#oNvBrP-~y3uGB>f?sL5`ycfLt#wrJ_g;rfvv<F*5MN8gPQ;s8!*
z3UCV8to2KO-L<E!pG5=pJb!6Gpno*m2=pZ!Ah<6-3jzHj1iei@C2$HSrUCmp!0$Q0
zBO3v{SAe%Vz@BNq9uBa;0Y0}Hz+PWtNsgnrH<tv@ef>}N+>bc$+)Ht4n!B9l+J^u&
zHx2cqgKBnAYXtSYpl~<sd2<@-MhA6|gSuT^TLBcI(7hgp3D|r;p}3j$JB7Cz{eKrN
zo0Gkbl&L<GltpW=6mfmW6mv#rcdM8@JrYe$?k_K`wHg36TQAfP;_zgsh4U4n47G5s
zcG2X|Bs1F`jW%Qwq@C}NlcjxsujG|^!ing`uEa&PoX5y!<>x3cv(+<2v#bGGuKYAM
zwY%`>b*fe37Qp6E=^4tf=HqK+n8A621O<L1rKJ;3i|%Nh6U*gz3f-@L3$*-^1om;t
zTVY;oQk9vh3xwi)++=@3(@#f>sQ>}j6-`m694#*a+Tm;I>aaH<?aOHwf=%$hPZebK
zinbpW#2fl<x&4)qKuDX!D&UTPiLcXY!A^Ybxy?VT!Pg30{ZfFhiB^?zumh~@?@f|e
zNZA=Yu(`19Ag9hLIF{H!*(F?M@fg}(L9gP8cIsVpD3$&<Q-e=&l}is4HBvCYSPYWJ
z#~5V_Py38|A$#05Ny)RwN6_ruuvVB}$bNI2M5a@BOaTymKEFaxP=i*efsgE;rWQM|
zd2xoP-LDg)9{@&ZRM?$-U@xNq{FVg%s#u8KouGd4P_r@sR)S~Efd~7--S}^khlPVj
z9u^M6$iu>+_pRXn5fwKJ|Bvu)7XBYG>}KKr5xsBbQhr4Pm%#rGTmt_$a0&e1fDaG&
z|7GF-PlW%+3IA)qn*jf_MKJikqR$)Jn`+yON{J@?k5<n**g_xAPG!$jKZ`xEay<0G
z*{Om%v?)r7qLU#EYedE(IJW&>g`rS{wEhBwn3w$!zaz5$V#rYhAqqjjAByyGzTE-F
zDjGF=$qhyj+PyyOw5vJ;^c;qnJC%@eibhp2yRGmlGtFWm*Q^{10>e@34i(BQTD2Q}
zkpJ+AAY4^$#Lj`3Dn}pWM@$tYfhtEI<VZ{vB!MbNALL0)6(oTwM<3)$Ocf-7Dn}pW
zOH36cfhtEI<V>8GjT`X@KTehWLyMpfpZN6QcC45}3}2Q(&BqMAC_BvHQrtNv*2fG&
z32tGV0+r!NXc3R4FhhkZG443uSLEX|%j_bvqx@x~wYz+OS-FqHM)}UN8yQ&37}{h>
z$b1J`V}g9+d5eY~)B^%3H@8<gB?^I{Japc(9YDx?_Q;ved-mHL&3kstt-kl{XTK3i
z>HQ#p&SUmm|H`*tq0otc*yyw?KOsFRS#SvL^lfJTl<CAruPLxqaY_G-e~gY^OV-y6
z{?W%rdat3f_X;WPkG)sX^Rf;rQ;^HX0_Ow@@$vg;k{XMrbjjegRfy0`GjHW3K1d^^
zn2*DJD&#v03J(b(0p@(0$>TiH2C~o$Im!VOk#QZ%%jo@p<K-FifhWlaAE$9nAued#
z2~CJ5Hj{N5e?(YRw!v8>-`6r0>URY-mJJoPK@*8?F`JKrnJQ2<c<1@Jaya*cAbMgv
zNtgD9itH6ybOgC#2oCT{*WlDTUz_dRG|wyDhEwaJc<Dm<wwBi|4cEWGw=e&PK4V#g
zL@Ct@ED>rzA)O<9P!_IVtm#3k5!l=ud8lhLkxOuhZ)<Gfl=pFYIMbf4nuiVlbN;QP
zrcNpbhwifDuv>TAam1}ktc<s-@7an6?69o!l?EBVS1-#FG~@SLK>}tnDu3hVqJajk
z<jQ^fSq5jSzj<@NTytohhRf@~0_XmE2+;j{8}uP^c`b78ujk6=HG5%52weAA*5^^D
z{3uemGi2$Vi~5c_r#dp?<58y!juh?=jXLMD$cR1igngd!n8=7F3r{=J9AS$_tXg=?
zr2NZno2A2~+ZuK7Zfnt@;5MQ8AQiJmlJELQLGh%hlIJT-i%qGA<@&1K3%cM(a_qH#
zm7jfnNaQbV7v;QU^WybOs(*p~`c!;i&q@`{F$#dk>VC>H9yu#jL=DmvyGMb!NIcZ2
zWb053C`PM-Ii~P^6*!W|wgU48jp#3OFz^u!j$|!}m2Rq8PYgx3Za6>L4v?Wt7kTkM
zTVkPYI8Ece@}#_Cg@=^g9n4SK!-7W?J0esROHGzOVHbvZMYL#w;~)@Eou};qUV?%t
z7j9^2JT(~|#Y4N|p^uey3l{=*m109Z76nT~J3PriJLwLv{h$#8_s4`j=`NxCNq2zk
zsp(}k!uXT!0M}EqpVbKA54@vkzJnNe7rVRTNBd?x^lDtcZ^TMptGTc37~%V5&r$yf
znD44PjrqQu1I%|HX$j94i!SDSU>fSX4(dz?H3Sv^+@k~~=KHWTRDpv!&Owz5>S&-~
zzOTO*=DTK=G;qkv8Tz+khST`()kN%(Gns&XoY0is=4L9Q$zK%4lEVuXIDCM4`B$Sd
zXmzntgzOcx{TEOQAok<@Mslj?z<Ba<u-XdmfQ_aNLA_=XS-V}FUw@$dBxPGI&hP7X
z+eq{)R~DN+u_L(6ynBcjS9oB^ZZqeeZFbv6BL48$qqFZ(iVpZG8kE<pPHqCLtRyeF
zW=dUT3%&C`Ws58x5cPF<Dw5eII$EAsezb`uQurYd?0NlFCkjBhOyx#b23Nanbz{jZ
zi4}Bt7arF1G;^L$LHxW6v+N!RGt-r=tp-=Nwx+9Gy9c2H)=Z2#guqGD%BUv%sZ^sv
zwn~rCVn-(lk*O8&107a)hISWID)br3#urWA=sIqaN6g6~gA+{^iFrsIu-+n%{bPO1
zJ%EU@v#NYP;A|@(Kn$@0-6O-c)P<=bNEDVeX1gzkoNW~mZD9e+kV7L@Q6}fFvF)d4
z;v+d>U<LE@?vR{CZz072{HXtmWM8pYgcA^XC8dx_B&4)4=OVE@6RR&3K_?Utj}yXD
zor}_?Doa_!=HIe3b_O&5#1F%Rksw%RZ7@^E_$KV#vPG&LH|wc(5rXX(8^1twV3)_#
z+-|{spOz<9m3Mte&2Rk(zinNgQFEtn5#L(g^%<&Op@pQ5iCfU8ciY-JmN{<9tb0v+
z1=jMn1d<HY2qZQBj1<c;%hfy<q~>uO>L^Mun&?{;o+gYb9GOR2A}{n7`ErDY$v}!g
z^fFNiQ35FXYzzFTrY$jWIdZUlf&jLX=JW4VA^~lm!XZ3uHoxHL1KqwAVli@y;A8R?
zZ822a^QCgLZ37&hh1N^ox6;WYCjXcw%k*E8e@v4l{ny=-hD-l-clB@&{ny=>4Ig$L
zqS2V)zW!?}m!ODKxrF>Pl}pG!Q>Fj%r2i6Y-QOu4YrfQW3v!0^i)>SrSkwo>`eVqP
z$c1E|pQ?WvhEy)$h(<uO(Ou(Ewj#!EHBx&OpPkA+P(orVd$x!iZYqI)SYGyQ{6mJL
z2D_W^$UiILu;d-Q8cn1M6kzfc8_Q}&3)YLyc=Uh}O|f7*W2x*1B;ZteJ^B^FHxU*|
z>?AafP3IvZC=&moku75zE1r>8F=M?RSaH1$`4!W3pt99f>OiFqRBmvUI#8(tl^b29
z4pizurNTyNQwJ(7h@@hM;zp>{fl7-ashA=2Ih8t4sY4#g|ESBpURBxC$;+POmA(+J
zKTG<sBVa=e6HP|zB_Ej?HKzW`#6%|dpm4-61A|WCS{1-Ere^kXYzLbN$U9csVLRB=
z?$HKU^Zd<gDwUSD_N^9qwn}|u0?uMXYdt*Bn9K{tJ%GSewlx3(D?@jOh;vJkHc1h_
z!km7dRpe2D50XuEi8?1Iov1T}qlr3KCVWvR&o7+tfBeF?w^C^Tf#%!i;I})Av~WHj
z#r@EP9Bdz}r1N$tZkUha$exu(aW2|9KymC*4lZ-#lfXZpd<xGr2Pp0wAM=6YR=1yh
zNr2C~v{~~&7kt*GMFsvj^4qn+numNW2R>Wva3&jc!xI_}&rf5{gMEH5jpnp3_MJYO
zdnL~9hk=rw%|0nRNvtcTbLp|VVmd#enxv5^S!1L_5+b5uhw&_8umzc`(!T6iiCa}s
zA_X<G@_e~#faET=(EtPDPC`)DRxqGb7=_rQhj4ynlSBPOTABrCr?SIRpR<L-%hVuP
z9joM`HZ4ONNU<!o75?|Rc!;>`oyb&(QRu^8L&93<m2TzG7%g2(C}Bk%PGiqA#Z0m)
zkF51^g%Q4cZ3@c+IwDrSBwXJFWSkk!v*GSRKc;007-GaSUWzC-qSQ38J$cD(Wzw3g
zmBIi2&&gXR5&dtBCp@P2@G(yW@r1_|3?K0XqEo-&-V=yUeTNr3ApvU4qqB4v@hF#|
zA&+tisp`>K?V_|)6<6T5;tBDI!{8Y&eak*bxUP@Hs#E>>tnHJU;!4E>i`IUO+hqxd
zW&4i7_v<2mk=hQW8&ioLpsRlwC)JL*NfJy>0^kBZ;wmrnYScs<kdxx{$Es;Qx>uBg
zYA%jNDwX}P$yvHutRJ#xuMgy{smNQikWGfWU6$@)!xA!ad!=vHtoL=Dg=pT0MjRz6
z3!Q48%^o?HjLh^#pt{pAp_^lnq&$0IE)Tq+$YI3uwDlS(0!B)<deck&QX-Y|U~R*;
zgguKm!?|0U<_{lOB4RM<=`F4b+A%0y*Wv0Q9n=*%xSgsSIM=H3R5kEvIa66Y;tt{w
zA8|I5ugwX>r(=F?;?v0-5T8D}50PiI6M1gC4UuQs&E{nB$-@EsrWppmpM(D!yKm3^
zn}fep@aH=C1!?$s4t^B}+Fz;mCp!331%IrAzc~$mql3T4!9OARD;<0<!Ee54FZ+ks
zT9d!-OuHtV{X_Nplu_;g&X-XhSMdwmKfL5sU&#J}hUao-^YF;OWB(w-Frvz1W@_Eb
z{-Im}vVWkHgFXBl`v-YSUg$J{D2M?}J=S6g6zbz!K`?$O;77)f)EQP7+T98=U3364
z35Ab1D@b4&*%;_C%@8t??2ahdkWC0aYR@D7!ey1~AFiu5DWf?M0x+9nrzTBE`ypNR
zDKhjS{+d@bYXCK!Z?5(Z*lgzNgGIG>x90}<0)6Awe`fz!rO1L#`-h{8&GtH-_7Bgt
zo_!!NdhKoha58B*DW4S??+44Xnm^C}5mhG1&|0&9kkxxZI*G+|JXy#9Rut<<VE-_t
z$hUtOt_x4QPhCv*vn-^Az73?x*+6jP9wFtU*+4)j&B9(ov4NbluAymX9*HRUB@YF^
zq@m!q_g&=299i*O@L2Uo@3)QQ(;PYMw=Lw;9NGJ~mQRzggL&X>!VWkLBkX`f@0sXb
z$fx;w&0k`P55@kErAK_+nuFSo<71ls1Ge3Sor*r>4Q(eTBp&+AWPln&vvi97LFSFg
z$EemDtMrRRfRnhyu(|Q)!nuW*KmIE7M-)=q#r^>j%YG8{H8T5A5E@BD*hL1)U-KS$
zHnp*glkYX<(4CSyML{FiArHU;#rw2Nqs9oy2=ahXuFYc^LF7J;AmYzrKPg@zIxr*y
z)ca|DThk7#5tuwi?rr~|s=NJzs_ym=s=C`hsOoP2psKt5gQ{-!kC9#NAF<E1e+V<k
z{^2mg=h{Cs6r#X#UAbMWAE%S-yHpq(*gw8(_KzwCwb1f~>`CVfnWCFvh-+IwDNe~?
z`9c<#Ydc(!3|K;@4l}62?m0ULvlRDYo9wn5;6W$;%E<tyb2J&C<{D^<5J-C;C(lZ*
zj%GN$bou1rx}!zJKoP!XAwP{&4)XCsfKOVZW?k3|pQx|V>H{27>E-zMt={NnIBu|p
zGpM?J{b~CJsn3nqp3NA<5+4BObOoLzDD*S)XWB1Z>Rv-bPjMHw`}T`xil%VRP>Ll1
zxuUScYA2o=#~w*gygr)Qej!+CTxoP#BJ6|x2~X|peAX8@Ml61+!16$iLoFK%eIO(<
z5n0>J7FaXTM0^5woudY1#fYJLpeVCNACsA%i<Z6t8yzS5QdWzt&T6q*40hcz_Wz@k
z&S+5)VK=rLidxib$4Jp@@yp@*M(TtN3Nc%ptTkO?vNyT*DM#r3N$cRDZJS;`ZXM8)
zE7_@+4foi!w_{;>e}6N^bFlzq4{U1mw{cv>>A-&Aq`ZE2EnG0Tup`v;+{kTBJM;TV
zHJsjBmpmdh5BWNK-c)1wcQ8TwH@onQfxbSo@16LDP&#rv32o%W>(^EP7y9{LWIy2(
z%(1uN{1B-$Koz~I!G9;7=t%49TYY_fGp3n9A9Gb+(aj6zT`)2t9<1nAZCMV_Nf6n3
zgr3Z9kogHlQvNQIdvZ|MlM4w2JWoem&qfNcvqDe8Q?QnV8%R;lIwwNrg{T~aIu-Tw
zX$|TQ=7MHKYGeqoH%ZVU`*dg-SFr<V8Za49gQUL)oy;49(t-{b**>CUOp!rdN7Jh#
zi`f47s{Apg#-IxP1lWGA&h!>V*9)qbz3L3q7O<NiV=nMQ?U)7Pr1AKD+c#kMQMKQH
zU%*HHxz6~=zj1&s9FNcO+yjezwS9dWs@6f>;-GFuFF$uB+@PuLFG@rG*g^fmK}{3X
zD#tUvh-JL>AJbj>`p)m&MSowrq0pGeVkB5!!F@=yc41*R&QUyK+ApTQaq?>f1?LHk
z<kT;jOp21T`kO;c5I<syVKX2NW{<VRymGR|#iiFo*~@kx(bUK)FEt*C18dWkSmLGj
z<1*yS7~5|ss}BBMyVx!#juxoyQrNz7v^L{mM4I`mKy1XbLtmUd3}c<%7PTL-!bG4E
zaQn1^t76+$#WD9NX7gx&e$IowKG3vdIFujl(}M&_HAVL{<8M~Gqu2PUa>&ZOu8N8W
zvODjvZDuPT$my)faWJ`^HMy=Pud^o4)f9Br6u6qg&YD73)4Q{#ceJ*N0L<2K{TOM&
z-h>M=^>^p@R5Dxs(*6^0KNi@VXMw$W!)lEms%_diys>G^@Fh(fhOcT`KYX3|9XSAU
z@UC@C5NcB~LD#w%-~~00li<d8FR1?u4{p0y3~jRK+kXfqa?4bk$eTDYk^lTxCSN}$
zGB%_0u4nR<Uz?e4p>{*m@zdx-a%|>Z)FVCf@R?m0?VNe|&N9KG$#TuTa`|MH7c=*;
zo>+A(kGZ!zvBdHEh0tE8A|xfToUNv7-!irzZRZ|EBU#G(^!PBpJoa!Ytp2rTNhF#a
zimTG>X6B8}nrJf0be?JFFo1cz{1j0}hCuui!XVT(tr(x)`ZJOo{yh)~<jqS+?4+K(
zg?Ep^9@sh{xy!m?%-oYG2+(YzDhk)V1G$sPTJhNWrj8Tmg{N(z6HPnL!5xQ*`Q&<c
zt<|p8Zsl6*Mn+X{<N#V7zy=4Pn_FMV0NCIFHadU}0{CkNz(xnK#R0r8fK&#6iQW~D
zbvS^x1#npgfXU$%kCBp_7pQ$*0B2?Zm{6Ymm|dT?QM2epfV#}r)~{y3XT0^a1DW)l
zBa-pL)3g^~=e3M?omP->E#oz(De-XUeHm{!?Wl}v880{OYZ=!v-m8%J3YyIB@`>w@
z)QTbf0h~Gl;Z*3HCRwp23B2dS1g7e|Bsgc^7)_hvuIst=Gf;T?qd%@aCB4#5!*%^=
zOtdTeQH|~RwZZc%0+5=w+VbpyJx`hyEZA>Nwgp?pfd$*~7Z&VDw_yLsT5WBZwrkHm
z=)<v0)LIzUaQ*j<Lpz-Si$R!8-{MP>mX$EdL6}bUT_FYpSPa2@+GN+xBEVt@Ce$Z6
z^C^thc^3-9o?Mn<RiOxh@suJXB1FGL6ozHLtUR$8rk52ivx$&MUcgV`kl7!|52HQ)
zqDAW^FX<0mc$Y{Jug86=aRKS+^)D2geP`=Sh)hj8@+*D^{wFQlo8W)UF6_+)&Y^WL
zv9;!pZRhaGL3o?)3dZl_Nj832Fzt+A%PoxG&l$hKdA2zXxX1yn;Q-*T{|sP4Bwo(5
z_oSg_IjA`f>cZQA`kA0)zn_$by2?Q%9n|T9suq;&_vfXdhB>I8IH-YwssRdIytIx_
z`}mZ-?f2m0<o*hwuX~#z$F5TULK+{3>mH+weyCX4_BHZ_QCVJYwpzZj61H4xCzH9d
zhvXidR1k)5rsB(Vh0lcg=Pd9mR_^5KAb&7))On`GQk7&@Alu@5y$CMt@KTd|>p|?2
z^_@(5rq4k_M6&BjVqhT^%ljRsn;d;ap2Hi4>pnFVyx^p17sf88p)Z@f&aFzChjz4d
zY4sy9((xyDoz$qknCqAvAl%O?=V{Zl(DFE&pq6%2Ki{<J{3WC@aPEaS0g^#qwdRKX
z&Mamle%IRc!3C?DKB-*V^!5cW>gcX+P_>F9ez%b;>kQ(P%4P2Amf$9L+L3v>v-94p
zt}ScF;y9ZB`m33gab>_CMB(hzh~H%?puKZ`PrWLy;15BYC~xDX@6~+Jb|}DZ1%~Th
z)mrQFWy|=o_OZe<u5*_dNYIjf@EWV}zwua3(%<+(PV{+cs!Z?*MeZ9@<GcehzcF_F
zgSAQH33J(}r~O+2{nT>vzb`i;XxLsiCO_Htg;I*P*`n9?rEda~oZEi5^leNHrCSL#
zD`_kfqNMtM>@j2p;Ul&mf|va~;?7CK$S01Ke#ZBOf-+?KzT+=%Ab+=a>1Jgy@lx83
zz`GP#pz~yF*Ap}romX~rJwY(&yt2FVB==Ik9IIW(=WD5Huumt3_N>&HpWw!*{SQ6M
z_Wc`B5z%1JlRxDQt{cuLAwi=hBxvM|f&KigZWg|C<)L7Jf~L<dX!MgoSgK|)8d$U&
zAJy}uI|BUmt%(l*a}fT&7bMZ*ki;Ls53LQ8y5avy<I<kW4)FgPN?jy{%1(Zbq&uK;
zbF5znkdM{3yP~pS9p2L>U179T#Nu0a1{#qV5iJ$P804yoIfG(cpfj-l?czXlvL@fi
zWSo3MMRFuKfGpX5_&V5x@O3A7kO@nz70Zc)c7+&b+;5Y2B=9bu(NGGYAu)5w86pnx
z)M(Jq>*a8bC~`X3lNH8oN&62v*OTlf{bSmM)9x}OADbpCH-<KUU+J-h2s2;Lq!MJb
z9-1%vV1Kk9Y{9V7OW?}DInQM`CEcIc1ZUd*RVR-~-r+`aJg|TA8nLE1)ycTIt&0vx
zQAe0fVyU`yEc|3Vq4bR#_2tC|W}mFSVeyfZ*u3gFhD1M9M&<F>82)S8k>~woJ-@^L
z(%Q87f+fOtV27o4)U^#(tLqyLYv}rCmGkYsE$KT1*^zO^_8+viBd_2uU7K|BN_)-p
zc>Ra$nZrCL*8Z<>H37aA((@kA3YvnQiI;v_)7GYZ4N+&A$3Pz{3*WM+Eb=^FH?i7x
zT1?f}m{?HzPA~S|jct45tkl3JKA*V|zxs2hQ`^kZbZW=^%(tsQ_0w!2M^&CsKxI?&
z`CE1Jz1ujE=y}W~4PZ$7dS0?Go)f#g0b}$z?Mr?|dWpol$R0Qa;^x!0h|hMEk*Ci2
zK{9J)4}c}omzz^o2WDPt_-Z>hVkRalXgu*suy4HlRbOOd^mQE9bKJo3Q;xevz0D5t
zXs}Oj9<}+JWB$-DH0tfgyH$*Tu()i6K|FFzpFi}=A2pstt(g_%Z_TXuP4|ja_lo+=
z3i`8{{=9%Lqs94ODB&!Gr|uBGQ2Tx-Er}(T#I~)Bu_GD!qT|>*n~A;>orls9m1UJL
zh8VMU_uzO!_K}xsK8p5Pi9<PB>Xobv9IS8-6MN@{+5|7x->fY3JE1{~v#ngdI6F^u
z_-X!UMM6pEb>J`Q%d02<CP05B6Ab;8a{&Eq|2^pM%g4dK<r|vlc|aPlmjf(xfb&xT
zwif}|O1I#;T5btid}F+|_%R1sTqM9v0`$YKo=gKi?f@4#z^MY<EI>c(YDOCH1_$^%
z2Y90Zw*qLK<JdIRB@XHe2URVoZGsZ#I4BKO=Ah1TP!|ZQLr~%zd#9nk;-J3ipvnZb
z9Vj@*hN~b7SN~Efq%?n#As;yU05$n%Q~^#y>;Wue<Kz~21dy|s1<qCxStaIiGCPS1
zsSg4-8j%|kWgi$cCOf`?9>DUF_}Cq~FW!H~J!!=5cn=V}<2_0SC(ut_>u~#JMW~J5
z=@swuS~NAz5c_u;I`O$U9Qa6_Hc;>iKih`p!snoJu^3<hwl4k#(>}oG>;mlL^9tef
z`Jac+wTA<(h2{&N7kA?G9EZ<ehI<h{KfVh-{{&>Zd1SM2dalFi#4>`@w^u(XhV&$G
z`mSE%uWS$|pDlboV)kC#r(HAV)2?Umo^~zGdfHi<^RzQ#JSp{_-IeWh_<V!#`MS@=
z=fl9~J|k@S{1YFai-qB(8gW{4`PT!yzDViR0=(V~N-kWH_-tRu{m5NEG30*Sp_e$E
zenQ0<SNQh2(P`xVy;D`W;F=>=gcgpvX81u*aVqpKdTIEbuj%mna^ZKQ)O^{YY5d+6
z;P=LT;P*!1cXoMV()#$_&;ZUf!tXh0{Jt6-k3I5))$cYPxB5elTRH9?_11&Ms~3-Y
z>yO2&Ul{c^7<^?K#~0*}dK(PxD#nBKvq9-#@VC-9zTg;3!|f_g2I*&00S3>60E1^j
zkgYot;?tnQwXv}Lgw-n-mLIp8?Er=4qc+9C@^6Oga)sq{JJ)+BzTYl<|F!P;9(g){
zzfNpVN-rPVFEDJsy%XCzeg_scBy0~Kke?mW-Pw6y`Y%#&`2C3hzn}Z-02>eeFT=(c
za{wD3aR-=e2>2aLwx|Y7Hs|uRK62O(jkCQCG1cd1bH&S!EWhZ-#>t}95Q1ta?`R1s
zj=VG$LIj>-xQC!oC*=%IG{#d=xXAfvx_x1GWN$1#JMwzIQ-vGf@fq?X{0sDQ)1XA(
zlcjmUc#M^+bV2nLc%+j)B-LWGBbL&}i9Sk2H|$YUi;Z{`4(_6*=zEq9PHd|P58hP~
z9<qz7W#y?~gs5It9v=L0d3eaj?j|_9i<S;fP2Nv;2;5a~^zT~AT|IOu^DcNkyDK}m
zOW^KK7{H5kSI@ogf*Z8E_V0dI!K407y}I5cHK=xytv{0K7T30DVw>ZMH4Wkq3!czD
zf*+jNLdV}N51+XK%o5x4Ld~0Pe*=>E6Y+}Ej?lNL-Hv<_R*A%AwePiHbL-i3vEhw_
zUgs#5@6cR6g$hUh;i7Ek-1(MHPuR(6(~^%JU-&D4qKR#Mur$4Es#!h1>6+C`Iljno
z9mjV$emH9Ur^RJkf|YuAzmrFe-_^YWj&5Nm#D1g3!_j3{fDbY&zz3NX;DgKxIJ&MC
zO`ooQp=sCZ#bmPg`86dCsXI}sKNpG<TFeGFF_}O}m$iw>Cn^@(w$=hj;nx^A6mEO4
zoGrl`TJw?D=Vj+*{DBNteKvyTHbq05*)w#xAiMFuKD(D1c;Mn}BC4~p7H600K={lT
zE^Ws2&{6YX+vUz;c4abH*jK>FlX_)32Nrf2@qcqCx`q8AYr1vqe@VGjr^jZ-r%dPF
zY@r`SE*Pi6>_bVh2Slnw;W0$S{hb>RG4Grh`!w1nRB~LAYfiaUgzKN-9D4+W<Se+5
zbkaiZif!AjErcw}MEsmW0yk*`AnkZn?{160&_uW`o_JNhp*g{Jk@<Qkk^?b<Y9Oz@
zl4oKbVVfY%YvAX~i#@i2wkJA=puT3i*VVrA94Pb=Sd<hldf!Z$TzyfLO!aDbzT)eH
z0X=@K4d|&H7|_aE2DFX=ZM#~jWn8HI@m4g6Ug>JUhIgU6n0z}sq)xuYm&$KI**Ez*
z;=CqLBg-#3KvpyIy3*7Af}(}G5J=Cv!_X1`p8P50i;>uG7lY;{h#PPb?8wD=3>e&B
z6%M}gu#npTNpw_{+co)tEryjM+qR&dMTFSmLI{bgk#R9XPF7h2RYv8A!qTDSqX#Dz
z#<k($Ek#ma#rv!w7ENCT<rOML-2JQMjWE<p{k%${1q>;UzBd&0w?{cQV)=KMaIQdo
z33rN?F+UvOimm<ux9>B(oz8Xfg_TlqmDxikvUj9d%}Ou%K+zPg$3plED=d>dkMI>$
z;kqAy)YOfMyw^sVzQ;y!gqsDdvzT+KL|q%F-9D%I)kAehv2z)X5YVppEbTQso9l$h
zY2Om{(hCuO5cT8n)rE(t#P$$V^g0H8FY$>Si)IS66!A62r*m5|qeJ&QL1!AdPHIK<
zOAi2%aGCEWkG&Quh=5h+)VMsO60a*x1All?yu`ZpZ#eodT`S`r!Gl#@(L9J_1|e`q
zUp!nlkcZPWOqcT*@fgxpuIMIu3g4}V=r)DYGKhGP;65W=-yn@Qx;|}-q^}eWO@h1U
zAuH_uI)A^ERz0^v4Jd6i{#V`r7G$_?lwLdhP`h>&)f2VVbq*hwy3dsGw#Dk6DJr8*
zeNYqX-A0Zm2EOaR9}kzGD|!KYd&M0qG$4lmH_vsR-k(MXJY&;3T<2+>Fs<w!`N1;c
zda8ywsw1i;du83J$Eh(7LYKTuZ{65_o+0Ekz0&iB0%B9npS7Wv<iTPSmG4}sR&kJ_
zH&p^e?Tz9I0L}P#X>+))zZ;2W<sG&@FwG}&QIEH_l~#wyNsWY-#yGp?nqxC<b=@A4
z*n~C9%=?}6RP&N7$;28Dc&um&H@f*&&K6DMi=6A?^Kx9X=Y8`AYdgrCQ~M@Ejo_rr
zK44|Ozh*8daB!bx=*42&@Tq?fxuRF~k$enC55!Mo{~G>9VI}BS_bjVnJG0%=BjMM?
zHg>VZi{WY7pUZu}uJ9^X{5o4VS4_`#XF2XH$DQT6vs`zU=g#DO%zXv!tiYWWx-<DI
zbFH^K>ka=6AM{*JqD}md<}a_8!7mb&lvj~bq*9)jQlwJeu#_T|@<ybxIOJC(bd5L(
zUCSSq(6#*D2^uDg!wWTyagZ#DyvH)qS>gKmJf-bQexL?lsSBt<;V7d9eK<f3-opQQ
z?&DAc(>ZLZ4WKr|N<oBsIH(6PG|jCQlynZyr=gy6P~y92>$NL^x>-=tIm}K&-RGd5
zbWqC#bqi2%>b-vkLGN|RzVp4U<t>b!7ZQ`{=y^X%gy$juqZu9oJtwShiRBCrC^Me&
zivOIaW$GxiYQptGRNO{Rv#TKJ+-B+A2~kyKKlo3#9nyBaf7`ZltEst08jN^yP{eMT
zXHSBNlYh!ra;&P(7ZE2RSWr<D;~r9`T>up)Yy2yA4^-UkcUkjIize}@_!&@fy=8YQ
zZV|H8tmB@erv)G6VsWdfL&tMqsQ-F3gE)^tY`YOYEauFr11YK-C8nFV9AE6AOr>H{
zs)RYm(zpIm7Iuppjxv=ST{)^-hsOMEDRbh<@nx~(&!XkY1L3(3QKlNh%m7s?u5=c6
zi5Lhu3wh48W3^}&`qXONg~Ii}R$VIl?;O<kqek)yqjRq}Hf*W$cRp!VblawADT<)#
zgW5;I4!M(>^UVSRHfHz0MKv)mv5_*-Fikbz6DNVT-R*=ZXT^E4Ra#_@)8v@DI4sq?
zxA!|~YwOI>yDaT{Wm_T2O2pdmj4%ZYE!*D;*fV!&uHgX!FSD7V9*#=L()5dOh|X)-
zPs3AIk?4T8Rxuh+>?BKWhxOA-^=;BiilZQ;T?C_n)Aj+~_-ARE`kN8UAFxoH0{&(J
z>-(SgkCwg~uKR`h03M!V7qB^i`oE2SBcSU>SKZiqA%%6&$x5#jV(Y$=D1h#*#!GEa
zjz02q&L%|s-5wQ`RMxhPs=fLLiR}v1{d(d}g|KXHJ#Ex31h{tz>mNYgXIJ+v)e*)N
zoFCooJlySk&&8SWA9p){GkwlZks@FkQunBZT#~X1&%0RZ2Ins&E;R~)S!>MIC-KB?
zX3WP*{LAogN%)pPPK3$CtF5&(;DtVwjM?6HpP@+BO+5qXBBRAvHS#zuqM_0vjuZQd
zI50*LHkZM~%_x)HqS<sR1)d#Y?)0T-6YFafj7k2A)iA^It!6zD))f4yw##DtQ9QCp
ze#6tU5-a0Q1FA(dG48fEso=>lL$^5An3IT3M4AWVpG&dPif^l^A@WiMDK)0^*&JCR
zZ=%Q{DhXEWWk>{!^D_MOVu|NO&owh*eOAX)6RR{mClcaVrTUH3pG15Cq}PuVqf!il
z_~oZKw)by{5_O$N(WbNy#S|>0ij)cZW;s@TJvzX3KU{Z!-kFx{Hj9w=d&yIZ)fDC}
zE=MS88R>r4Eh?YV+xUUr0(zKpAtpVK4c4(Y^SN@`f0X721YDvPWH;tZYO(d#Ihq(c
zXE-b5xY#I1wXaT#)xkWZPS2t$TqhSJhVg@FVvU-#QKfU!tto2BfUY7FW2o3Wd!!~G
zu<o4<rcyBQn|e2)%qot0M?Fr1te#=#F7LuW(ZE<;-7{a&pHUTgg>f7SG*DM3QjEj$
z(bb_88DC>D-EDktXyq-&pUJtFem3Y<xH<RiTZ(4toIa=>f!YuQx0d(TxrSKxt+1ne
zH2m+mo&4`C-6{SzSTT1N%~loC4pq(E=Y<|~TscRztA0NWhy){4=9O-I>FhQ=aXME#
zG!U$S5=kUebPfT7Y~!S^mVH3Iic}T!u31H3^)_NiS8&u@=_wjnb%>EYQ8TH^2GdPp
z$z@}48`hE@CzlY0u@TsC-LW=|K+~<I>Xshi$q20V2g~XnA<Hig>*F+=+^ccLhDyXX
z@g`~)au{^s*N2h*R39@yGtU%F(THfVFm>ah?Vj5wH%fwXkKVL$HC$IgZ&<ZE)c+DM
z#Ic%%8X^`C`J1FCLKkHK;4xV(X$x*wQ~?KW>mDKn+aI|{#S**ZcGf#A`B7>%<T0c(
z`11yP)Nmd}MyT<yFN;8+TiF5R6!~eF?r?jgNL(#eDzwtw;o6<tauS4q7lKN%bBfH@
z$`bj!si2N%+L4$2w2~qI9bd<MezhC0{BM=tF`ifvo^}<F>l~lG;Hjv1I@?;{mlpV?
z26tTA<d+(LslgqW8h)wamm1u0so|FzeyPD7ml}Sl;g=fRajD^#8h)w49hVw@slgw2
zG5!i{6lwmrFF+=`(ZlZf#1_tje}7z#wwS%;Xqz|eaXH%Njd(nZ16v1kjRc)@xJH7`
zIb0(_=NvBO+a*|gU4pgOC0KhU17nY#Ej54C_6orsJ>h!o%G36Jqgy?HFfrf}zjvPT
zh-YztN8DeEo~+|R)J)HwN2H++cTmL+YSlQP8sQ|3wg2Rbpsh_mvbOefP@!J{wL(y0
z?U$#amN}@`IPk=c=;G(D1Pa!^d?c)W-wVtmJw5;TVNX07tvm@dP|=kn4~eFZ9R5R|
zJ0SZ-@+QNakx7@VGUt`W;agdWjh<2>%ORJ1&Jm{Pi!Lp)ZRP45Qu=WbRTQN9tTL4O
zx@&%%Ot|%5<pLEEmPxZ$dR8=Ce}L7*N-C21XAQ{ZiO0AKIKScCdyaeyS9+yf$gvCO
z=!vgJONVpsR$&&Y6bB5)Ir3m#FD3m|PYQ}MsoyDdk(5hh3<sBwIExJ!3FLtQSYGFl
zLE_@o!L|5+vwSFDGLYkB7l)e#9{FXGzsg~MueEHjh6$je)NLxGL8%BeR+Kig!`i&^
z(#C6=>eo%4SB^LLO!1;5R5{^0WIkW`_LYSPeRu|Ai+4)I$EWCX*p>@dmq31wB}dA?
za&4@~U52=mmun0*dOha2)5)L{lu*=lcy2E%xubQiUj*H}>_6eKbYx>z)AlBQSX1@d
z`wQM#!Z6NiJMxDWVQx5oUhBvo%PlkS(6i_k%Wi(t2O}Hth9?bm(+49O7ry<|+qtW0
zGrsH{#dK0qboPDb<o0gOZ-jA*(R)axmLn5xL|iSOj+YA8p8{HGd&;ekqicdWeWd1A
z>d+H8Fk9<2r{DQ{rvZO!bpW{KTm$Us0H+CXy#Rd!{_-^7G6(n?2U`4<0N(}BR@|S`
zP=9n#PdKPs@kg2ao}eU5PfbJp%0V?as7kf<zM!<?E=)rWcThibP{-n_JNNKnTX8S2
z*joQ~9xHBN>uu@=Bk>T5aZac%)C6S$GmhiD0hJ8en}=&oV+`XPM>M44-2R7)U40~q
z95EA&A6$xr>mP8JAQp9wEAsDRHs7dAkfAbvN-OKQx{xXC>hWd%PJ;LA4shITo{=?q
zg^GGy*5oEjuSBo%Bq>HFKkn|rn;#GUG6_nBG#pKqMb(cyD<g^G{{_J50Hg(80}!F6
z#T>n7VMXHg+IK;=pM8V4+)rbp7c^}rOzX3ei{y00to&Zy;Owk=ga*gnG<}xutLnJ9
zlF1C!4$t^6sn7dt#B45QvDS*7Dc&hOL#e1@sjFdnCl!tL-Opl)Z75gCq})-FSVv4@
zJn<T`<j#u3n<UtlVGCkirq}4Rdq^t#`(znad;O^)@0MOZk9E)HT$dAt;CL&sw+ftC
zW6UTIj-}2aLJ^-Tas7?4RGGyq78~=#JtFp!62+{<VExZdO?s-zOT5f_{N7O&SkEk4
z?W(+k=A1_U#YR0oi^of*n8ht^Q~0^0K}@NtVQD0qx{>!eAhHF~{y1I5I|9J56!DS}
zpuPZ}z2>|t9cGn5HA<ITX;6zL$quzzpjPOy_INlV&5m&0LuSr*Fc$gk?v&C}jhXMT
znxClg(37P$!qJYoqAqF9=+)V=n4#WCNds7B9df%S8God{+C^0QwCv!VRyG={^MxDU
z)5$9__i`<?7GrZcCdfp+T(P|dFXzi5Kyh!YJ@W=%!13EgG42+ayB@Cltxy<GHfY|Y
z#QH8KkV-YIr}P&toJe*Coeb4jMeNEjrQ4&@13~h>Il31HGSBslnhyHVKB5c$cKoqo
z@bu<W{4P${Q64VBwESSw@wd<91m72WpdaEP=gTD*L_DdeIGlRc`Ywh$d0n%bK<XWf
z?!?6@Px!KVFD_|o;8H+;qN#!$SMB5rW}|ogdVYspzfXV0>(l;>*YD-eczuRHBLRMa
zKcw1B`!imj>CZTi_^ap5pRpPv3NG!@6#A;rJh{8UFVmCU8-s?BBVwg5)wBo5!qB4U
zCo9mq_LvYhZ=NM>zqNNPWBBEI#7n9`cw;{wAb5X!Q~Y*DtPq);L5Tpa!0TM2fESN{
zjZ(L3`1JVKC}mtLxZb~1Fr(|0C=ra>d^{mvkJI(Kq%$hG-j^@*>N9Y(Mn#>bC-kpi
zde9b)c0NTx%%#g}-e~$HjX4)DoHy~r#_?ZLuuA)nZ2UAgpJFU<yJ*dWQfcMrT$n<W
z(I#>eSZBZw0u%&3n!!oaXL&x}wRM{Gu`GSAl;k-}jXp9Grh!chg&>iVI*7vFGJM$d
zSxr0h^{X;Imy{J8H*7kwa+F&r^G-YS3#Kz3Bm`(1zoeq{^>At<dKyjiMw%8WdYEHm
z{Do(xvJcm8?o&9;^4s4se1x?*##-yE*4$%Ny)$3cJM;Vc?LS|$M9m7BtJyM+;InBw
zP10O{Nf_zzUO4Zn;knoGo1c3fIzJdk{Kd})R1R7*4$r|<s;=&~>>gxIi1zs;mO2fL
z!2pnd$)`3J+V^e;&T{3o>)G1tot^ByD*K{QLyf;}i<^ClFpaHLsX0V##pk82=&jFE
z2w%8D$aIR}Q|IT#5=w~K5lfws%cW+mYc1ti%6qa(N;a2=j~Vm5=f4+xOup>JoEY_R
z_ELx1%zowQhm;1wa@Z?H4h(521|-m=r0GXMy;z+h7D9OzEb^F93i<gdh-eemGvUfs
zy5@miXLAhKU2Yo}-r$ooce7r8h2D*K6lT>?-(1>iXoPiyJmkEWx)+zZOYF@=M>u@#
zM|(}i_mk2Pd$9FcEQ(RW#~dA@gOCAx0@I;IPvRwQTHs2^8);_P8@aV149s4ON~cAm
z(hbbyo^YLUxjr&yZ(r%dbf@-f;0>%jZU712;J*%-Cm?3pQ8mGIG@{^kRB8WVm-(k2
zO;^t{NrJMhXw>0-)Ip;HT}VO|gCEd>aH_+|fZt9hB`ARs(9VQhp9~~;OhfelJ^qiD
zF8A?2Y5sm~sK0YP?i2ruX@J017xn)<N>I{c!l6DTxSQAPY|qGmURs5!{~h#TSxq8d
zzuwH*)#dbU2J}D(Acygo{zd44NXx&V2Qa~!&4is@(VB&arae`Q29>>=jv@vpmIn9(
zWHObLOkzOldLfNQ%~R0>(E;p5W)FyrzraX>!`co13BF$D@U`JZ@U_q-tL|-(Fp9Ts
z`k;N{>n8!NNW6;1J(^f7{o&YXpEc1`_NiKhT3smO!K7nwPf3yeAEYaPWFMWzQ<ptf
zg_*c%{yE#yJo%CR5l{81k}3MbDpZq>hy;SrPfz+q*lovSr;SU_|3ZC7-zO^ufprT#
zvZgzHc=2ZE_FSd}9r#!{Rd`_Lvltnp-;RHA?YGdF^rtX@=G-x*v+*h#S4wt=pcMaP
zF)@}=PmK#y^asldHaC1G1OCVKnVsAv9+UX3&`+!@*VKw6VTO5FUqY+}fy%{;CY;Le
zV$Gk%INAheOV+8KvxRXHdS;07v_>^#QoX8$G`A2-WglfJBE;YQH~57zy(Z({sFzFg
z3zf{g6aQQM!V7NxX`&bh&E$Itl#27iCCXngJoj4T53Wt~2TyX<*n(V1rEZ$6Aw?_R
z@z{KJ$NRj&D}Tp$g!Ig_X_h(ja5u@AbhE$Bq)X4YvF#@cDM$koOg1UeqyF?7YM(72
zxBG2${!YxSF%nRKV@{o{{mXACnxYr$oKj|8vsUhPi=%TYRj0Hn1n(u>MKx6=vrC@#
zBq@VVUFJj=<3w!oL<#eaTyv{Gd7jFgJm1#j!Cy)9XUdsw{xtUMI)BD?pFc66K`@<P
zdQW8hj{_X@t#1AtUmV{^{D2_CznKsIH<?sy4IZqIed3p1f`RKBoOnYKw&Veh)xO`A
z+uK`bP%Rlw+A<7+;u18bQW(S%%fUpgvDA-qx%7O1gg`$#@{a7qz%;Ere|@0h5L?y3
zjqOU#>-=x@J%wca@0kZ5VgjSD0cg~vP7Oc?cEU`{R~Ocp4OF=9`45FPTyy%&#Q`=_
zf7OteSgvtmCJDK)&Ov?cKQn}!=AZTPO3XOr3+=o#_{#jLD%lQJs{?Mu__G$;RF7<-
z38h`8jS{eaZnT&@HH06RLus-Hs#9atDZ7C!#f-~BlTAH{S&}iFBM36DE4`np2^g#G
zfcrt*nsBhs{UoWCeLpncRRr?I{|ou|1@S*yk$+PJ$vOro_`ZAdKQDKafBh|mc|#Iq
z;o3T(H@RDk_gU5U8(y;8_kmGN?7(~2K{U|*uv^peFS~7ZkudFATK;9*t?t*-@-O>u
zWnL?|HIP|v>v{!CH}da`#_v>rCzh)I4X?W?ni{4f`L!=me+n9cdvDCRSHl154E|;T
zKI|<v4ZXygE55JvNnO+^d~HoM%{hj}=3t;iS!@%31iJ^`unx6W+v)t!*1`$TNL}S*
z5Hwk>Z*V~~dZJb#T|2U-1hq3uP9V}WX*G7?jaU!?wQG}yuBI04E)HLN`3KBjcqYx}
zio_=h#z7aUK-GLpnc{S?IO_oys9|#^lu1v3NppRB4qABf)K+6~7w*kf&<vR&glXU9
z2^|wY5@KyVfezb4jr_L1qFWD$)hal8OZ6jr-}}`r_ih@TdG9}mUhsvWfAfW)Fa1K$
zrC$j8OJ4~3^@@FsNnYpZhf{Y!{B0CJurF|b<{fe=%Nr3`IS&4=Svij8fbn9_IhYcT
zE%wb4A6ytfwZd>?G}+Dps5yjY&Am`iGD|E>L(O+kFFL5d3<IhdC=7x92Vs`@+V|z-
zm+7xR5$a&3h|Na{^PY|^O^HT!e^neYnDca)VCN{z1;WVR<pL`Gh{^Rn-(cxSp0f+5
zLm`5V0R~}zctW*E>zd=D8CoR&!W7|aUu!37_qb{56@1(Z?#r6IT*WC?oZrP7;rciE
zA+ZDYpyb&8&L)9p0O^JxvrfkMl4b5fa$J<98yyxM{i_`EmJ27c1YfLu$7~R*qN5kY
zM{jO28$`bI=RvBR?~D=o9&3MdWux(KO5-5zhgWN*4N>D+97PPr>AhR*1U00PNFyiV
z8Q<xN;sS5T8Ou=^RbbdmWgjF}#;3wsC9yusW2r$9U8mT_HGgnoO?mD6;D)!+d!Q6X
zZ6x;H$U0C%`GB!LO>8A~5Ra`h_Y~4_i7LxJu_B(jpf_y9$MSW_0yb`G16yJ#y6wF8
z5QiLLj-@8uKphxjUHhm`dTIXM?)}7CjH3}k=R5W7c(?d3ivLBGUFBT@Bf>Ui+a9Y<
zhPHSE?zAr%uFEog1VkN0%BukbB+ThfS7<;t+Nrx)%AZO#4WZ&i^prrwF8;*&lz3k5
z?^s~O8?EK$_OpF?$!Y1un=`~vfal3qry|Cw2b?w^uamUiXQ_HD?G4G$<Rw3arowew
zt;a3`vA>=<n`(`msj0#bDcLP&&D4r9*8L{O=}iLtPegq$vDdf^Ka<32<7=?a8TYKO
zJ5-ZqI&I*v63}}hB{hMwbRSwC$jKO`W?iVtLch{X0ufy<jt+2wV>s0Va?`mcrl@OC
zHYf}~R4snpS7`Ct_9jY(@VwTSbZosQe8uO9p5~@*^0iUdXj(}m`s{Vu2S|7p==eVv
zzBFAH40eJpS5>oduNv$H-C(d8bgRK`kWrI+eb)*U$oPw}lY#t2Ao20~J=MpdHo3vn
zCg4B%n>#(@4kZD3q2_&kuSyoWV+s7#?ps<Mb)V(N@MCnX9U56pQH`Wkhnk~U(?T>u
zAl((MFO@<ku)G(6YAP`eK@fWOEY8jvA2wb5!RhbYnglP43C5|>O<XBohH6z!r+wNu
zXwKK+`U&ezYu0hPY{4s^F3bW%>_;O<oBe@)<qPTCUm5S@9Zi5dCO~BTrNSV1G4Ysg
z-~T_6r<ObUIjZT(BZhKhO&u%yA$c&BeybDTn0!@d@^l?*R<6ez+P0a#9jsZ4_qk!s
zqYl;zvuK18_96_hrapxQG__SEULwgzSG;LnXc{LLst^PkcPzVK)u~<@#cda~d6m#6
z`9pl8hy@=vpOqSMdsP>-sR@AjlA-F~J|dl8$Jb@cAJ3-?phF=6?ZZfwLoR8&sn0Gf
z5(|Bz$z%uu78pTi1<i-+UhGhG&`z*9?xo1?LWU9G=^yIlXMEYmzfsC(*PIZfo+M?o
z9*!~^=F-P|ph<JlMBgGcw?2&JMUfeiiKh?5+EHhPV$7%tzEn@M?3_))Z}IH{OFeoz
z#d;cI$>EYsPg~T}h#sw`sn{+JZJDrPjv7c1HRjq<Hw}$}`Se(FR5~`Z)F027!+%B{
zKz(R8NRMyQqDG$_)iPqIk=}`WBR#JZ=}GO*;sR3!=uY#c*Mu+m=<Wf|h?S1GQ<kW+
zH~z21ZyIGB3n!scLzQO|P5p<}$Aj)Lm|``QK4xF|P73z-3I5$1+Z7A1J<Kq8NyBz*
zdJ|wfVIzn9yP>)dNR4NxuDBD`edTMQIvo5kJAT;^NLnAz`dcugkMHQS@Y~8`9e&e7
zDiMO1&7&gwL~lye1Cg&Be@US4>HXi<_sD2!_CR3)JzmIxg*OfSgfo3N{I^g1bUvV<
zCDgzu>BE@ge3p*)fjd)1R;en=jDw5!c?km@ti8`gXdfJxxF+?}syHE1N~0^yY&11#
zDtc!E+s|pFWv8fi$Yxk4zSN?$MXETwPEJhAEUKu-XMENH$KD$p+ZfsuaAE<1BHCe#
zmhPK88~+Sk2edrZDiv;T74_D}2Q10578(#cqsL%`6F1c!v-P4*7<S!*ipOWNv{IvD
zvYs*d7-fT=thJ|{PKrxYQD`(=r`?xvw&B4mG+d~0{Pq+TjXEPLB%)i*o5XLBp6D4)
zm)$pmo`7`KGAbvWY!UNG5#lTe;gqEt<YgLkE6<uJFBb7~RodH$rQp=((WJTFS$<?+
zX^^o(6vPNW;5%$ZS$P?pz_OG4<$+tpi9JIR=K792_c1{{Vi9s$ultU{9JKi8E|fl&
z%I=G1O-jNp^zlE@k9X2{kZ^io`HRpo4`wFS2hw1SKbL0@^znz&$A1C)dJODh>q*Nv
zs?2#4;hx!G$L7br*AD=Wgu~%>7kRWd0HZ5;rZt}WkuT5S=kFHFS~B%jbs@^rd8$v8
zh5y`kEt}<TiFE4ETqB)WG&IdMh0O@&yN@%yG|}CxuF@bOC_pd+dIZyfd!|;X%}vca
zCJo+U%uSi5_0|&6G)G0<0?L)==t(!0oV{wedy|$56HfEpr%imRUK0-S3Dcvr<v$<#
zgvThE8<kS~e?5H~210jCcnL^Px=cB<@Rz0stbl#<<9idQoPan*T>@eJvG=v~$W!h!
z;X}#F;8s{CWt!3(%5<PnrUNuTTEBFP=n-~sErAli4=M*F;I|tIQ97w{pR~vy|DD6V
z`dK)a*e_+do7gX9Ii1)q<t!^hU>fmoSvww7^qe6kP?BsdgicJ?j2UpRQ{U60s_+fZ
zLs*1SZI%h6f+NM2X;KN7A;D^%Nqeb@Duv=~JAXMqtKs?~)`maxyK<#&qLa~E%?M{3
zujOcn$cnLQaHRn&TCe?S+fMt)Xq|N>ocdNTtsaPrCsOx<BXveVP*xE8&(6kS6K$R5
z!C6>u=t}1zWQ{hu>C-tdKzponeh4QydSdY{Ep$^R0viy;x7_QG#g?F(okZ<+>=z_H
zV!2ul2$rkgl)wC0uI_G025PYDnkOfrtuKwfhL|{N9~ThIqH_iR1opAj!DzctsZ%~l
zWdjjmfSgVg5wk`NQ8?Gnz<?HrC64Vq#=e1Qi8x#AuD%`ErbFug&@Lg=v^!5Es&XGB
zYGM}>m0l-46KB$_^AU1Xs&lCv_s#TDfl6I?gi$FzMVdsd|CS?BMY@DAsyBsM<5TaO
zpp3uJiL0rKBgwjHRLtS4|A3yD{Z%wv#AF?`!VA3{=ttzO&f40+I^!Gp@}lgo0X-=m
zoOoGAxwI$<Jz+|qG>@l#V#Fj6;Z8IACG*XF?=JG&z5c3Nt(7vJudlD<FN=WdVxO`0
zz2}?%)sfVQe-v5J-TL&iIg<2Nltn3c)A-QY^To{BQXhje0n-2-f5c1_cNR_I#&F$_
zY=*tUOlX$6F~@BAoLR;SfF&A~89&sGf8C4z?B*_033-}N!z4yH>4;C;Yl2T}+UDq~
zqBs9P$wx+dwo3n#COy(u?StMtU-SOn^ri%U5CYSvr54pvdze88w!&i9YI#pDZEgR$
zzjB}*rs+`<wvH2<uQ)}n6xc5}koO{BI`>IP!Z%#Xd~Ms0AF+sU9GaaqVL|JJzPt9c
zKNH72Ec_5I8f>jr)rHGxpesV0CK_lZYPvXnX9qW|GEqa5F2d)v*#~#a7bWmaxZ3eG
zf@ePx;9JBWMJd22Nt|T9`TR;3bW_+3-PqRlV8y_N&a(QaB}Vx}hc+@uok*tf_^wE1
zYk`ksO43N?R}Kpu^sFxV8ZRg?38aletG{fl;Qfd6cDdKXXZE{d)$X4pe1qaQZPQ3%
z^L(<(m`Hn*A*$5R<AZ4dNBr2!7QTs=Trplro!G(YfWq~o^^Ny`pKt6>knnZS?x*?W
zbj<9T!Dj-e^TqWs#qN?Hx8DJdu_OxLLs2!0M?{GvgadU5B85qayA)J_lK(Qa`s|7u
zBG*)7`}wwB^?JfJ<|Ptrt3#jJOkIy|RfAZ_%b+)U7vv<gUgQb#6-#_RJTB3LuK+My
zclvuGK(0lJ%2TKfY05-|fXal5P*6GPriR?{>)XBwFRk|~)U}BgtCYe22^VTI>wXs8
zB=C<WU+1g}gfQo(+z?1(1Xw1`mBb4{<jF|1G?rM%-aM@r7Kxt&qj?I=#`&}*tsuoj
zI#zRFmYxS=Q~J`RhxPJ>`f3yHx%+Hi%RHNA|3!;fo2%63Ghz(%PO;jB`{^@9)qajS
zk{X-Oo3iJM^z~3{Pi9+G&*Az_%#f{aHiRrdmnH%eu{u{pnlV7U#5;WL5io-CSQ{r5
zYDAW9hwfv3h9`1aCjZ{77ECAITJbdv%VTL@??zn`g7SLY>kkNl{M#ly+qTsB#hyR^
zK~H7|?&#nC>5*oByov+n#|v;mnmeYL0F++D4La5>a}qlZy0)s+Jf^QJf_Yp*mzhwq
zLTvQsZvpdHVh{6If*$2!5g;ZDv#FXl1Bb|y!{v1#<-@}EL0XjT*+u(qLTeC_p8VO&
zuyU8~r^IYrp+=*MD|T~u+B+=UVD5(Nf5$+86@!77QRpeEs_#I(oE~RFnUs5}*xyu6
zE^|k_(Mz!Z99u^-b{_}xES$Pf)S3Rwq@k#Dm(%`Q)I4l8{=_9{cxmhX$8<%CZ-qhr
ze{@1i>p1Xb`%Dd#Evi{|FHD&Rq)HPvV(3P}#HD2Tv$<AcbJ?G3l{UNmxmFboIfu^9
zNMrdeY2g)W;kgCS&Ro(m(R;nmdj$ri;N`A~{TnaF99`PFD$@1Ea`a;N`Y(3dalF`V
zf%B1i%<Hh~v1tF(hlf+jtN-tI;GB#OTz4=XnB{liFm>Q&%;)t3gIQkt6`SR^aA1~u
z1VqmLEwkK=>=&h>e(a!r;h+X<E&L29jO;IB&uIO}QU7HBr>Ps7h43A_hQD_-(XYs?
z@0DRqlWewP9Gj=Fo#PD{iy1!Oug02PXw|y4GCClNCH_<FAG(EzyNIn1$GQy|L`2{#
zo!3x#81EvsZh%bkf46HQFZPJP2B)I0e$#jObwkypJOXI)TE7l9$kox*c<Q=X^Yw1s
z0}JV&v*h2W`w<qRS(8`mI5cbWbIyie=eTWx3M(hijV5z>qjL~(#qlJspfCbcY4WuG
zvegexu=nAPZ|ow3q9rzZLwxk7?Bg5j1D1z$`R$l$rmm=r&Is3tR!xTj#c<mfCx;vs
z(Kk^30?JiJoL)9`gm5^^U9teav0NevUTq6bRN!Q_hu~G)vQwuMIkk&NS72->RHF}A
zjY-^Tl{U@qe4Jr>o}Vjc#uAvqtMU$g?@OGFy0Y-dcPV(uSQ+^SqiV6gI+;9!rCsOD
zG08KmzQpRUrv4dcI7(hk^iuZCf}3T%$Eyi#U4trSQo+iDV#g9~WDWR+?vffenwp#&
zOO4HCONc6Ia)sc$)YVnd)S1jepRrAyMJS`QY|CGvnEQsLyrkVA)Wdexu~gJfh1V#P
z;<-`c>r>Vdm;6qPTAkMLKFl!&7KQBS{iL~Vuz`G_XdFjqhvPD&Hk)Y|ZZSK5)5Jb^
z7ClUbUTQXXwq45uojfjnrdjP`&xg$dTQ_no8hSkzdPPc$MqPupljuTksfnQ$Wae6H
zm*7Vofzr{?F`%=p4WGXX-RUE40u;H0-T4&Z91YhUCd!y_Pt*|2wk|dyBed!{HVmOw
zb-F{3vdn^R6f_ze)fQmxvprB0)!h!fRhQ(5@4uUoNB2E-XR$$URKqO>={kl<j|d<&
zX>>+cd5wUdwKvymQNLElv;gD3KJ_AqdVoeo$2`A_Hga6g;aoaW4Qi*wF7O@j9j$@e
zXzjSfN7@dK1!yfI+*(4f4EYvM6ul4Iz=eGKC<Cq*JdPb|iTW(v&rv<lhq9h=v>x*&
zIW{B&MBZXAr<cy7N<p8r(xTS*;z~CaM^eKrZfLk}o_)SWtaj`o?i(3*LA<J3Bd?{B
z?Ad2xrdFyBF!ya9D;leFF(WKiyL5{!RLuu_AKebP`#`r8l^FO6fe)o<Idbf_2VC!{
zSBG_reI3rn*_iq72C5&ci|!Mdals!2FuE~2HZ*Epw*hHG#zp8jFQ;+jWSmai=K+iP
zkgawRG{(8(tI7n{{D0Yd7x<{EbMHR`6A3mt6Ac=VSYsR8M5zs}ZAqnVA_LCojAEPC
z#G*|J_0Sd-A+e}{fh5Xw7{q$1SBkCe!Cut*+6rnlfs6(e1w_Rg2C$eAR6vD*K>pv~
zv-V69w5NT~dEfv0|9?I&A2NIIwV%EA+UxS%*YkMA4IH}`@#FL%o}U*1R~D<my<$Ls
zo~+6k5mllUUSrqTa_p$R-;TdcH<0Z|cE=bzGs0>c1{LVD$uWwxN^%p+O79O779Q@!
zA&d+H;kD&J2a<nfD79J^s!HQBy&Q50QBm8UKthb!Um}RRU$RVthJdGMaGX9Owf2H{
zaIHs+-le`DAdje2_q&GbMK0c#K9QY4hv~EJ^OE_4`|2<SOGbYKgAu*%1AZ&7lkdta
zmXGM8sui3^=eqr{FI26>kM-5eoo&^XbMdA1G8^ehdCv~KPYL&l7Jr?K7GM2>6>P2^
zKG(8`xO_<4&@SUo+Y)H57oUXh%QzA2so~`5bBzJT$Q4fd=OUP=^M_Jr?m`;3*6Q0;
zoqMa*Coi5Cs=CBvm%Uv7D()~lQmZb0I`wE(i#C3=P8(t>^a%^ujgL}`ZUvh^Tr{8V
zl>5tHk2l`WEBGkrfy^j&cwMCMHGdyn%>Gig=>j}*08*c717<2Pv=vf$YxY3HD>eII
zk-W@2MEG*nnFU92zd}hXzouzpv(rJdrrla*j;dBtQz;PZt}t<z=N!~z3p9Q`o0YIG
z6n~FxLHLjFc}WSe@w=<D^w<c?E!x*JBj($Q{fZ69{O09aE+oR2ZC5BAZN(Vnnlkr#
z>XZn|^nOZ;ea-N+zEaOeZ+(m2ir%`zU3A(7X^?bhyIm+(GZ#DTLRp%*U{dAf-s;YK
z?1p8=jNZE6E-Wb~qo;~<@vCMV^_uz{+m=@>+mU+V!qut8mK)$XX=T@rZGLNZOfv&n
zYq>?x>6^dXt0z3P7+ZyYb>3p<Eq2~v=N+66e~8oJ4=FZIv~26Nm1E0@59g?OAbE#<
z|INliNYX0g`wOT2(3Uf&_D-$JDlV*Ey=f9}+Fm{7A=Q4mYIn_0@%}s&la1Hj7cSc)
z`=aMJ0y9}kXG3Y~KBqn}q|aHAec!XI8sGDMdHJ{6W^wYPSWvh}c(rCrVD3B`E8(U=
zN0=vNTP|mw2<e3F__&Ar4=U7E8q-N>TCRXs!tUl)zA;?3E6_MyQ_QQT(5t4{tESXI
zRgL0*85QxpDl<;mI;J2u*<}`rzZ)ufPBrXVN1T33b^aB7A0I|Y1~7WWx1rznDc(!&
zytaF$Rps75c$d7~8?^bX_qqD5VG^vWN7A#hjStkAGb@)zeII%DM1_QjCdjWwKCPec
zPz~-tVW%tnz!mC1;dWQp;|g`4kmvNcGW7DeCF+ePcz{OzTxIqP9N>aeXjkAY!fP}E
zxC#~Fl|l>jnwxNCit|cYuEMrP%bs)jaZ<ncT;Vi}&vLUkd3rD%9{~Hpt&5o~`qjCn
zkA8KoNu*!)o8QVQbE0uM^LMj1e;r3BWTTlWZZfRt>T$H9AEGybWG60Jnt;W+o0;D=
zWKjl*7&+)S|6F^ppIaLy+1p3x?XH<BW`DXhZG^QaQnv2U`5kDg*DKLO7h>-KUxNvR
zeJtNkSGDdXN1xiapoOez()P(<^bz_aSosJq3Ql;0J_r^)lKP-3ss&zE$@+S$Y69!)
z^{OH*I~CJ7Z`}i(S7ggBQns_6lgx4pG@lO<?z+sBpf?*ABBaWM1F1YU5?^zzC^AL`
zqmNQ!u=3Fsz5h{a3>H0_I*=EAg5tc&Cn(OF@C3zqMNbg!tB@Ql^@JQ=HQ-%8BoBMn
z7cE$0_SzBI42m>U&mV3wdj|((_V&*qjsLFLr17ywBaN2^r1Q$YCj<X02mhdhpUgJJ
zWswqtA13%L2Y*Ed{&EN3;NUMf8TdcC`u~0u@N147kiTutcB49SqlYySUP*`Yfm)3#
z6gvjvHqK;ngzIsoDq(ew8>N%UoRsUq9Gi3FO#I8@Jeruu()mdN@wu#yLQbp>)RtG5
za$%2$IpJ|*&d%gn5C!rj4=RHG6+H7}-d({fVPR;QVry}2FI=jY!gm$ei=90Bpvs7N
zrHZQ3k$P&<GD70h(y7e|CH^g(%syp^KkM8xM&lAOsi?HR*tW;ZZqmk<h5L1{HvbL_
zr@zkP%^ovyU1ZO$?);}iMftphKEiy}Jw_3HcV)^@2k=4Mz!j+e^CI|RYD_8iP<j2a
zx-JJ%iz}vNYT)J;nwcLdBj9ZvJ1%2oomv6tEpD(Y5XS?Y2zwSp%Ti>U*H!Z5sfRIr
zC>#|z!~c4q`3*`5*2>0%{hXy$gnrRAbrCVfFm{u+dT`*nQ)KoA-CuteC(@^>zXI3)
zfU88Haqpl&(?a3;Wd5CW(6SS=vl4Y{3Y(eLPI#=3Rg}FSXqsX$xqrp;pgkv&@PA<6
zR2ohmaiv;ToBQX!IK6=;ah@(#Z+wL!oZc{-0!xj|H>^RwZj(v;e|Ee2wejf!-VzRk
zdcuhz8h4Wk_*~_`<z2}mf**TJ-4J_2x8K45^UQ5FLQ-t=U=>lvR!8H{^L$UZ43+ky
zv><Dle!_wCp5c$cq07TVU0~rO6mgkM4KEPlp!;xDSP_4L#ImZ1$R?p2Wa$j>;RZ~+
z=I+A;iBDWIbfN=z)uxRBe1wj_n<gqtL>o!^Xb1_;{fA4<iQD&va3H#YHtxr5FH(jh
zYV2*(IjDVQug)3~D|>s^=c93Of@Y3FCSt6K4&A301e#1rQY`AOZapxkO{=+2qt*2p
zu!FVDdVj$K0Fj9yu;6!gLN*3zlB0=h3ovph4k|mJB|orGa<nuiJAW$a@J7k#A^Z@P
zYZH8>XH+y*{lQRtS1kTwMK6J0w<&}x`hAq$tLqUjwSS?Ps1M%L+$oo!R48jf3-#6C
zBN~tcWpH6~0U&@Fo*1kP2?U3J(+aIRWG)bs`eHl=#^}(VaBF9wG+9fiQ7gWaW8D!l
zKDtQVE#03YQ-nRLtQ#VWf<z(1gt4hd|4xQi-@xB1(i^PN{nie!iIaAtNyijgNO?>~
z>_C(>rx8%sEasTW*VwomB<9LuWt#)Zukj&~<k<nI425yC%IQ@Q<a|$t1#zTpF#3=V
z)pwfb&(X;23u*=wgh_d@fm`~1md^&Z(~0&Ade*#Xd{d;cEf^_do-L0AqU_da&j_a_
z%<#>Gqpn0WAvD&|c6C`@KIiHp=KPB8qdfYDH@bR?xzB!9Z{LWf9d?Q6&`x`9v}_bS
zcR->CfzgSxKeWZ}lyUCGI%gMbd)SRl9SFZ&>!r8*^o6ea592TOf3W^(LeJ>6I}|%O
z{ak&!lDIP7fd|r5=4MajV>#a_ce{idW`dgQjq98tu16CfR5qmkv(BdlHLi<k1{RLR
z-(ZlZh$}rnUWx8prY;38u&*pTg;B!o3iGg5ck=ffcM$?r;CnxU6^u?1;fW6Ra+Jt8
zX0cA>u=ye}q%|{h7`!H)VX%-J=5Zt?NmvHHu3i+NTG+z~gX1&v0j3ay;5R`#Gl|sZ
z=$+OMh;{Uw^wd*u7S>8l=czdgf~1DBP^!`^DUI|+N5-oUG&cjuCt!ILJQicpVin9k
zARcH5Kkyl4MQOg|YlDn033Grinf7V;hB@#VUA25DfWZvlkq+=g2l!n9&K2O-9pKxa
zP1pE$pVhdV12v8kV50!NREkez03UaNs~zC+0&EhX#*kOvk^#KQ0p9BXXJgP<cD(?-
zREkqFfIo47*Eqmw0-Ps6FB0RJ3}A%={I&x;A4j2O2>`p%L1mwifjZtnjc`z9s&2lZ
z&IW4FXVTyMVU~UGVGiokf=UVs^Fa364Ak=us)GZ)GFVUxfWjIxcBuHj>_hmorF25#
zFlcA<%p4{l%S~B=&bSax;cW^OE!4#U9XeFEpCQmFw?=RgD?101$G182vhMy&5RKPp
zo8n!ZDrs4)Yz$VLCj_0ijH{o5c00IA3~B6vBj|?OH@6TxKkz^|M5>j)lqK(8#|^M+
zB=J}~?Sm#-)O?V&n%j>6o$nVrJI*LQ_)Qtn#+3_s(ZOODLAnHMHLkLT6R?D%-<%s*
z6(<HpiUBMtz=t`pTjk-}^Kw|7TFlR|n{a49+8jxIho@L&(b|9a*9|d3?HB}rJJ=%>
zEo+Yu`4m(cNGhDa8223ZI_}f5jxX0noQ`nq>ToT)_aw)HtBacn8Q>i0#1k#%L)D(r
zz=H4lv$D!wl<Bu%)_07H6Rs|fl(d=`{d~t)gVY;%CPMPNWUfzmSr{dp^f%Y&NL4`@
z6&7;Qbsx>>lgBV@hKFu2>{_IAWrc{)=Y8lHEa%X1cOmpU6z_2SMn91HH$1E>gl*na
zY&Yk`+$SKXQZd)THY+N$05Ct(RrDafmg0HDxt1gM+pVbvJQ;rHm>QTb*aqCa``Q~x
zSmNXe?GxI*jW3G$dSZ!$=1SYIXBGH;=Ov?gHPOU%DnV0a4z?Nn2QiF;(Rqv8md{!e
zC3Q*bOj4KVKw*n3M4_ib9VqN@g&nR?2dNi$IybJO+;}=f+wa6>_<F=%<OG`KkH(I4
zc`}p+ntv&w@X+~-UT}rtRZdboE(a;mdI_x-)?Li8ZdJeG8P+#EPl=p5{GS;Qf{Udo
zsMxC+b;B~_c2vLq%Z$|1^z%&UlyxNba}f{?(|A;Uq!XieVQ@Lerf7pT*+wVKFM~x1
zij;`&C1RqnJicBNQ3UX);sxa&K6d0gN`2|t7wJvWD+Fwc5wi)q0}{+Lfu_U6eE|=T
zVq{KKPm|MQ{7B-{Z*+IC3pbYdO=f_HoR<upUK33YA3U;`cr%RBV{SZ`F1As!1BGrp
zmuevMqIRIrjptG~o^}Z4K4#;&)W&nEMDB3ed((br$(<zi%%fo_#%|(ZKAgPyXlbBu
znDmd1p*rrLpH}Uuq&*xa&;kyVcqfMuyrq6gC5_`Sf#z_SL<2aCuvhCAE!8a!y2W9H
zwMQxd+T-|yv>=B2(XUL|01k<kyczR-a9+}X*Llg|cSL=!u}{YFcW8`)xp%hBz3RT)
z+Z3)i_cnGglWDVltL3jOdohr7<{Ouf+*>t?|DoPCHo!wS%Rx7>_zRLB8n_iHc_EfO
z9b2%*Fd3awBb=j-5`%0bnG9V=A&95@i~ZN{e>&d-O)ue6-EYpb4joF8By@!AFW!7d
zAUb5aACsBx+F_Mh@0t1TkkC|RLD$4b>j!4?&{Y`HCKn>Sz25JKuH^5+Wan3chT$!3
z&~e-C>|<H)n)UL~L4#@EJs8!iqDuUm>whh6&HrAZf}B9pe)Cw`N^M>o0N+Nbu>B}5
z&=es}7fzdDvy4@w?<JiQ33nIkFkwlH4wII2=rCeQ+pn2V_bt|8!hJ0|OuDZ_hY|My
zkf&Q5bc=&-aTsx5q%l<(E9+)HvEV(oAEr_zE7G~umV6jZ<~|foj(7x3xcoy&m)^FV
z`9n~6=Fgw;{Em<^eQVcA-XhM4KZieYxN%cY_YulxF2dJzgmhCPd^yvOLNe{+R~wi2
z)bGP{nPfuoUBp}JT~+^4_sYhep@Hk)hcYdDb})S8Z(66*A=!J5%kV#Y21Cw<9o92g
zoXrzOa5hh#Vw}xSv*2vLw|{?cV-IOKAEXK%d?NGU<L<%L9C)x<54O1nzoZ9G;6dOQ
zWZ>sJ_`4nao}+=Ubnt@&|ISASe{lwWf`h-*!LJwmcOCq?-u=Cue{=AsW#C6R_-{D)
zYX$#32mfoqCmsC3g7k;)-)$d$goFRS;4g6S(*%E!gMTgq|Ez<5g9G2Y6DOZ#7drSe
z1^)#HzbFHLmxKS4ga4!8FLLll3I4!7``*hk@KYT8EU*5ffS>5#Hy;FkwS)g=27a`I
zk2(0&g0FS(4+;Km2Y+M+KG(sY;NV~U4DhQR{9M6b3cNJRnf2W-e*c{9;s(rvD4fGI
zaMAaFEhSZ)%?Vcql@Idlr2j8Kfu~RdSlIhH0KoS`io_Nw{({|E6KMW-DYMX2!U(p6
z%f^Kv>w+~mJamVNylC=>(eDo?j=q0wpn0M}M|jH>Fq3sqI+)sMa@eCl#A@?@AqC4&
zr}N7*ou71dId#LjN_bGh=`Je>G#{szn4a={8sxYltXHHA>y;iZaeg7hab2|Rd>Zr)
zC9yJ+=^?NMnore}Wh(;BpWy_1m#Zt5INJ;Q|0#;)*^Kw}H9=O(i#*L~0jE*{a$3mg
zlkT*b(}y|j=hLMA?PI{jz=#KSSpZg$1)x2X1)x^Cjx8bImQWd4fn%>!l)XOdiAbVR
zWCfH`C|NTYuH6QU`X@IWFtru&_k{Yk6XByHR*T;)GAe=kCl!rP?axD|R@~@tV1XIA
zYgdT4&_A8gl6DJZD0N@Sny9Z`H{k!^H4rJb$CC9mauH^{iE{Wt_vbZz`sNS#N)!%n
zROKBJk}eN4<r!wPq+OB3LuxZdwz@M#j6@M~%+=kU7oglXLxaq`@Ad2liPFf_{vhoo
z-VJScqP7dd%<TJ`nM5-4H?_Vxv`~tF{pwumfBouQMgaY;ul6tG6t5<;&+H=}FaZeB
z18LNgWkTJ~J}|0q+V?o^_kCXd>#L(yXQkCPT(!B^su-(-y=u7CA13}3_C$06!rzmj
z4I4zWeD4Lc-y!vCwf~pcTl~L--^IJP{@t!+#)WvjHCMe=QTB2m837eX;@vTFBi6nZ
zu6<2|RQ7bBsTu??Jx*q<$k6$&h9?*)dC8O;rm{?4&amOd+g<bz#)~+NWMX?g&_p)1
z-hRXM?f-O`ywG}M386sTJSj@cKQ>|+z6ToQTvOv@#r}xMos8cX2~T@S{}RcW{w1*T
z`f5dQUSB<2@s>jt=2^w`uf<y)Zt<2g@$^Qif64FGRLlULoy;X%X1Vq69QE%UDG*66
z_imTvPg2ia4RaK!Z41|Shf21PA&`WV5#PIEqV*f2RX0VwSHQyS5to(2Nk@1Otl&GQ
zJWTWBxcv(XT_`rh?z%5|Ja@kyd`*g1(t<MwEDmbHZcS_EnoZ^I$D<|qrjo>RaA5A+
zW`Y8vgiAiet8TDUBpTAfbD$DwB3aaI3!LY3SUo25mPeDLZ?RIcgF9gZ*F@y+GyonN
z7bC`Ic8rPpvISqjiBPodFAef(a`rvG-+17C93xLwV`_h(>1=D<q09L5VZJ}eeAW);
zQBfZoK)mlP{jmRce!O2!4x@f{&lU84uIGdQ{5gB$RPV`f|0kdL&z}r5eL<Lr_Lh53
zBc1iNw_(83U5Yo+Z((&I+J#Hd_|fr@qV?fq-LW-c-@fk8ehmM?rYQ})Z=bEFqp>Z@
zs-YP_o|6IVPxu5wCG#CaCG#CaCG-6W&3Ch3s#sP`p|0U(za(7|_$2-`Hk!H{7Ycto
zs`>q>c8Qc7V16rWGd$Sg!8Fg4^RDqc$8ms%xffrUWv{>kq%~}t0%^cE-!Z@)8~~h&
zS#sH{0GfvFu?*BQ2eryUorVc++3SLmhV7ON)J+cRUI*2rm%b(_Y1l5$KwavfW;>{_
zO2d|Ws_~&)m^_`&?R4x(AAK3XpDba{8-Mb7=qf@7NjJjcOU}3gMcF&}+{6;+vHRg1
z6lHI66-k_jHtrk<`&zsA1IYOIQM;|sJ%U9g#*`)o!xb;x#`mA$Cl67B-Ps>bT%+9(
zPjPQ_H}`7u=c3Sv4qan<o})PT7?7t)Od91G&wc?1@^qm23<|<1!k9FB!(}~z=2Iw$
zCVr~biB=1HA>1s?!nIQN9jVjAS);<OTL&rgd-DOm3NFyo#NVe)+bQ75wUM$5U|2SC
zMdqKAzfZWaF%t}Lg=>FW(|FCKth&Q068@s@T>YYxY?3bbK5%<)y%M#k?*{T7ky`Td
zR3J8CD|9yF8EMtt)U;}Eaz$cDTkGu0(MFOtJ(l>Mk}a*Bxa+{EZ)NX%b8uGb!z)*k
zzP>eeVET^2i8~JzPwOa*CQrDeYj76Q&g|S3sRNe}yKT$htckn+UVNoXbi&icGuzv?
zTsmlF%fwwD7MHhFBxV;^fZN)(%qna3rS?+a-l^^H>V%HOmNZ5Rua?Bo^urkK02w1Z
zHb#u4|An^Jy|dZd+8s@P;|5+Es~!DIV-?K|N2aM@IG$frcaBDEPdB>_Oe-2mJR(BN
z#}H@YftCC{!yl|qyEn?IgV)|w)Bp%6ts%JFup#gIDi>y|!YwWmIIAVx*g|HfjvW!$
z#_Rve&4^>0*K(Q+eUp?<2T9vR)xLJ7LhNbsRwd=*&^WpXmF6VQVoqlePK+*+4dEuW
z%Vl~R4|j*pH4IZ~?@JSk!fjjj9s|AgDP!r0!tUHE##fwFFwlD0Ad{g}><pbk=oH+q
z;psGS<4yJ{z*$+T7{MCbC*!emed|qeUNHe;qcA@?Oc;SCSt<H@rjE2Tn2REN_H_@v
zWTE;6);n6uTd-@9osgR)YitBEc3W&%-Jf#uyr6f6ZzBmMBPudz5lZ-~a4nC%Ku__Z
z+>!%>ex=UDHK6fe;j}NRO)#3T#C%;bd1rg76yLg-_Pj5Tp}n26*T$@I7;~TaMA1Z1
z(by6am>vl4c~Ksrjc*O98xu?B9#0+TB~M?bT{sHbhoW=sdC6R^o#z%={yhE9oU1PL
z-`F<is@eWqvDCzf5V}1^43NeHLjrS^KQ0{qFj{gT;(IOT>qdF>m`l;;Ya!k!+|E)t
z*_2vkj(>o84=sF4nt;$vZ`nnsyTEIYJM2i%9lZ9qkUt+6c<pf^e?Bhc3L07{e?Bg7
z+v7t1d|d2L?Y+#8K982(B!4|Imp68epo#tIRV+C=C<dkw?k|=sI$nKRU>*7_tN`(+
z=ghu5^vGfzqK~xbQ29uQ4#OYe_uA0Ix^->zVcoj6@?qV&cKE}X$nbmGQ{NiF8s3OG
zG7DVFo4fwR)bXYWn2!aMN-r2@h*<vp+xPdT4qoooTGtuQBj;N~(-i6(ZySX=mje{)
ziGzc)mK_JZNRzT-GJyFG@MH&=-3wp=fJVx8<fN;6cdJ#`%Yo`1kv3qOphU`6WuV#}
z)GH2Z)?lD21SL{-ZwBhu4(fk6sNd_2V}KfoBk<2UhrI_i%j(|W+mC)VZiiDR0y7Q_
z?<WpRj^Ed~Z)l(~!E$#ZA{zo^?MO@)a+@wRH@#3l#cWew@DUTbBJrIdJR!~w{su_X
zpCJ`Y@Z}Zp51HzoeK_o7#ItKhFJ(1eGb%FJ!*^GX;(AYH@Lz>r#M6*Nq@s5{5(c5V
zVlWb^u3z$w<oJV)`vwK({usI;B61C7rWu+DLQA(($&90kbHpm2qm8BKNTNQcc&ywP
zzHJR16Xf9q;|$I)8djfC5vLZ7U~2M^-q`rccx&UP6R~C!%e(`wwNye&PAoZ<P({7?
z)I=v`XLS$n!%s9e;yWml1yT8${?50YS3BZSUcU|R%jrW#zSi|?%FKb|?Bs}3QRCXb
zjJ$||{N>O^YjlWS1d}zcJ=J-^x{2#%mHZ;V$u4o&woCtf2ed<n2^VGAO~Ulr&7%2s
z$<|xD%)Lv>=E_BjbqFnL(P7e}4jpnAg&SAkg!Wo}6S7m7E@~P@-;ycS<iU!-BU3KQ
zVg(2Nf7Ckm-xg-Ge%S`}gOZ{><-t+m#<o!dsW*hth@g=}&66?3%`LNErk^2r2!qY*
zX0KAG>mtV^E0F}VAaTD@J~vNH{aO)US%F=x9jX#j24N^kq@*oPRSKfXSv6_W!G>=(
zRZbx}K|xwSiyXNJ)EYw-Zeaw@tv&tsj0boXbXZ!U%%cRwQ?l=rA;W0NhVVBU7^Fbs
z!~7~}0a|Vf3AnxI<tY1w$;VxHeA=H>J#%MjUtZ{LTtZfaNi~XpbnlR9*VzV2@ZHtH
z(4w~-fV+_YbLeh!W)3Zq|8wYWb7l@LlK*q)?py~Z|L4%%L3bes=+NEy?m`aGp}Pxo
z;a_Cu<PBYc3h9;lg<YRwR7^^f(|a$9fM96pVt%oZ@LRc*ITajk28qy<HXWi*+H|OV
z(x$`kC!J5?{`!<Q-hwL&!IgW3EB`ENlE^;OpT<_@n+;o?&jD;TlpT-DK0m^k<-cwN
zz1*{PfA7M9_=T-&1jCEOcN5{y4AV{WowoJ+dvsS7yw1<-o|YafQNgA?S(xt-Go84q
zAbiF(FwXPs3J<?Cro%YTwyO%_fR2H2p6afg=58+RpVQp^f#PVQS_q~Eh278&abY8z
z!fu;$hPq254F~q2_@pLXb=C7!x<PVuqX}^G*U+FZS$#Mfo-u|771>ME4LZi|Md0Mm
zp+Pxz&#4lJ?c8V$LeDqGfPOrRRq-_3=nS>Ab_AQ;@DH;F)m<aMG#Znto82`GXt5>M
zw&sm6<nm->3cndG+Yo48tpW)HWfkz>Pugt+hojxM`|MTc*tf@O{qCbNxyu7=j`z_p
zhpq0aYaNec|J9o@^n0&@3+;@Sbq1Of170=DZl8mU5;-FmK|rrYjL=s-k3zJuk0lE<
zT?MA)+IUpe-86%Ot|U+B%d7#--?RG0=*ynR#vW)Ms{;G7F?`wAR6QwaMWSBq%%9}T
zs@S7l>b<>zrXQJ5a&8dbu69kh>^lt7i-3#Q9TaHZ!KoxKzT>>0ek3HE=g7gSJSnn3
z06)Z<8-#k`$-zQKhfL6Nm<Un;5+dNw5JwDpM!ptJekFD>=>OmP=_iB(d@n`IFoLdu
zB1LOQ-_G|C4(w@+MvXl=)6P8v-~W2nm;ILvMAp}}&8e#*sp*`VdH(;1G`11Ed`mrP
z)L$o(a^0eV&yNc{GU5_Tf_gu{_7q6^KV+6F>!{z<HHH4^V{ih^NVH6!G3A*t9ZEqo
z@qGz@*nQ9~V}gN_OGK+X!o)MoxfFF@?1?5k`3fsZ2y-V75v6WNTJk)Ca~oo2a`#Z3
z!+OJZm%RXI^#V*EqR8eQBX@URNirL~!>j*y@D-87xhRYgDYfJ$X*8UmLb_4zomIf=
z&I;;BQq@_|lE8w?c}!#X@AmjFO)!C)pw)9^1rK1eb2@_lgzh_10^^(Hf*A=Z;bMVK
zm^?;cp3y~4q+~@nISygPUlfz7GC9sf6Pf;E%E`5_``?T{${yx}y@uYytGpa&dQ~$W
zicxxY?eN9)&jduE<cLojo%y1+s^TduXGs{`r((W?(XtN%ji2WkSrI7?MtmEi$k~lY
zQwEC>Zx5C9;1z~0Fs6{jkRAn)?c*2jfxi4L%f^1GwhwTj*rw2-w@~3B1Y9VBpv(Z)
zUfGL7kFo!OeGHFv_Ir%7In$~_OKr14h+77@T@azAwpk(c7@HyV^bS{OyA^b0g&(-W
zAXj;z$87gR=&>GGnC}X0_eAKi{jRXU7256z|6{IfxXS#G5&VTYPZsQRX?xf05)sn=
z24QB61^n3U$v^kZ42B+*B{Ud)P?peO<%6<>28Tb$`l$+$8k@zW)Yx-Ke@$xaIg_fC
z8v7pjAbTv&J(lMl%NzcnWQADS`}MyiK&0p&u073HivHz^{(S<UlU)j<>m2k?d(9}~
zw>Ur%KgUv9b`}(|Yokqy>>(MzYzKIZ13ZBBZ`ou4{>1@q+@G%T^;fONcQ{bv`vMFL
zaJ~clO9t?NIKVasxJrOO7NF;YH$MaTa|d{*1N^)IBLeh%@M<%FKXic89N<ALcFTSu
zK+gy7>lwgN4lv>XeNw>PCcsWDu|#%W1~A|N3mxF952$gR06ibP&X3Yxz4aCQ>OCCz
z>U#wE?*jCE@Se;7{>=eC=K$YP8!G`cvFi2=)Gr;>5(l+GP~R1l#H#5Ts4E=QTnBZJ
zpso^>#Hv^Z>Kq3(!9jglO}iQ>eDGQ~!Swv`4bu>2)@MKdXU=wQpfUa63-A+-V75T>
zF3$Q{-~@3fAAK2Y$4@|tJH&*X@e^2%Kn`b(vt`%&f=<2N7H~*s!V`^gMa9Bnqg)t-
z@3{x0(7VD;uh!{&J0)3F*6b$D^a3<KGlVUg@$R<RYl+Lsal8vh6K8Xr$k7?g5|*nQ
z9m62Sc|}rF3?+0nL3t*jQm(_cgqRc#%uPW#Y|BP6j5oCB51h<(Ho;{6pRjnKp^)h=
z`CZ(c>kiA2gGsw44!etV9@Ion)@9iZ?G*uP94a0!{aSLm9}le$RQDR5!aZ)B&gDuN
zMy<4gqrx(A`k)o9k%vo%P8&BL4DZ2#-)+7gi65B+hvM!rBCwh9%Xluy58bSN&M8fP
zfLLRyv_Wn$)ybZMUW$mWUlp1#UwwS2y$qFbH2O+x1iN&c5gv>YQFj}7RNu1Lj=0dP
zr82{kt!iX?KXQ+5>iu#>fwW*OXgB+3)Q6;_*A@htzGFRUCR6PbFjXdImeU96HL8Ka
z)fx9-%63pDl?=j{Hc`+I^Q>gHKd2UU>XGEkulf~1ivj9f{wg&4Nd|zfI{;U;oHL@x
zbZ{VikA@tRY%x|DHN>jK=}T3Hlf!f5@unBbJGwzf3q9-kqx)8zn=JYprA!sKsa^(H
z-=2}Xqw&q1XY*e4TbGH--GlPzF>`!|hg{I1p5{$cHAwDFA)nt`8!g!k+vP$lOOqtm
z#9kPW1F!7If0^gf^rYhP9D8p%+L@RBgQZibHfv7FYp;c)e5BhscEiH;n;%VH;K-*G
z!mu&bBZf`=Y!lal^LQ@bN~cM=?Qz1fNK#MJV>Bg_tSmyMKSkE=`9-!XDLJ}?#*U}4
z`o2=AB2t{wqn(Df@Nt)WR1V<y7!=PAM3Ue269D5h9y;E8s1Wb(V%LK?>3S3+ppX%y
zA20JSP3@f>I(0v^c2JbfEyaNarQXPu^XOnbx~+5d2KuYO%5l_$3o=t=u~K+8?j#Zx
zJFs{8*Z~u`MX5qzPkMxP3MZ~Zh2CnvtK)mvSS(N2((s-)&rSN@p$oH6X|C20J*G79
zSFS=;f8g>H7H~wJQGN2?+L6)X?{|Mz-sf6p?BKNZ7xQhA<N7F@ae=0l*s=_D1<M1!
zNR{WTt~(Mnp#|6J+DZKj<IlmbZYDXS42SUpTx`<>37@NQAuOokh;KE~4P#~R+C$S%
zBu*(22=TC=o=;7-%;Gz{eoI}(KdBLgG_{+*U3PaMKR$8Si^b)8Cr(>kT;3|K-i1Qh
zKW@#$qh^)dlsm*SDo*_NtW$3Cf6HKd1Vur~Z~Wh!xQoa73Ua@hdU5^s)avygFsFLz
z@6}u=#4fBsFNp69G`-2}vDf#_M}m_^&{7Nro2y?+v+CHhu$|A|8u7i(coB4pi+5wb
zH*l>O`Hu9lmyIZrR`ybG3|+BaaTm+ptKS+ad6z(A;dqKYVz80Zd6(IrV<lvg&(uh8
ziI{I=-}|MMP<>bDou+<9Q#K;sPV9yh7<nVy9Zl*Hi~*c?Bst<Rr4%w$6in^Q8v*Oi
z!yX8C&<X#)ew&Ut#YR}WEDZ4mSu$!GmoerDsZ|=E?m^0DeRO?CMu<dq(fes0<};XB
z(1L!M13XOk0T_k1U=-4P)<?V2fO`=X0X*CRK8!(c*<S?c@mcFKfNLD!CJxkiI_A@5
zPYSS;$;qql%K-j|16=9=j}>5x06jjdAp<zu0p8#M_hI#1)+#`c&$=K3_+1Bhi35C1
zfXe|iK5JwK>WdDl+(ESpYK5T0XMHLIHPk`<n}a&76R4D+#Ap3|XS!FmJa4`70S7+z
zee~wbo&pL!s~Vit8Ct&|9A2I^9)7C+$ke`I_Dd@UXH|7Yzvok5+?}cCPFK$#T|JZb
zQ%~_1te!m8v;T#G{F>EwxT0cIb!EA_&d=0!o~x_I)m5OnM!C9DFH_fp)YXsQYurBB
zn7vn_FrKs8MLJDf%w)o#nasVLfy4C(RY!;$itvsR6EC)M<VI&)9YMM0qK01Md#FMF
zO^}H*ODhsb;huIk?vzf^Q%rd*F%3KDg;MrB#m&OiQgr3s*w5G%ka)fmIlu1cuHEpH
zX6`bvzmqHK#fEw|bV#PM7b0c2y{)nGZ%c!|DpqzjEqFqFDC`{;mG=Y9FX;;XQS)j6
z$|tRj6SuYG>|&B)=?b2wpr5LE@*ylP8*iPHS*x7PBJq|!KeKk;+Zyt<hRRmrXSTAU
ztbNuKVuT7q@l}nR*=MmYR=ZYu1ISJg1-}SQPGG?l_|CC-1<FN&gyHAFk>PAMHcrIm
zx5-!edaPuBD890B6HX`?6C)+hRG{To9bw$f!cb>UF-^Xxi8ZU0DF|43v}858?v7|(
z4zWThA5NZdZZyEs%{~hoyIcaRx{oGKlcFav!dwQL0!BJb^+1=50`;~VS{^d}7*bsJ
z6mRIhcp&{aH(3`T$AfM;dA)Vw4<;<YAg2SSI2|y;>9E5cFudsslba4O2@`)Xe1Y-;
zDI0Q&5&{*#701KAz2@(``3J+Di|?L#B2~dUMiW2gOL3{Aah>UKlSVr9NicLPnWRL<
z3i!`X)}7{0y>h`@3b%P&TK)#1Ul(a?85EP|^j=_vwd>~WPp!G&9pbw_HEnpR{es;T
z=G9$i!v5)VR6IS^{Y~X3S~cPQwk<!--R*1Ja@7gDvr<>3c4wi~I(@pTIsL)ZstXUM
zUaNZh<IgwqXty4H#~yu0kAB$q=!geX>n{9|N4IuWF$igX&uhMs{;01@gE2m<?pvt!
z?S0p$TRr`ARR5Z)f6c0Y4PH6*ewG@>oUqE@pz<wh*N8c(_b+}uwdKOs;9Ux~VA&XN
z-rnl38toqF$1kQ{ubyI#H2T#!TKPQv>KxrX=Y%Gx9wCpL5uSg|%hOiE^1zQWNJJYk
zLWB82Z4VbtyR`3J+O?1~a6WnOd%e2O-Z7KfCsg}YyK9Du_vfh?Wuy1L1M&~VcS<iP
z$HAbOH)i^8b9K?5q0mKc679mjhplJ5O7SKILa1z2AbAq{IJffLs%njvtYrGJX($$d
zE?V-udMW$MPq2!JK7T4l@fA7U!-UTx>`V2QP@rio=2pV6*+%w=7v_kpcG37V%+=O&
zlhMPl!kGmv!;ho%sjj>or4QyVJ8*^WM^`#<g$hVlI#lQ1u8aKJ$*F(|?1gEqvG_+Y
z=7rO~!--penfj_HxEdz8M@QJB-mBcLJa;Q^gw>}_7}%SiA+u?q>13uLn{BXP-!rW`
zf01TQ{vvj2#L8YJ-;?4}TF-Teon@E{?5i8vSxxFl@j`N1U{DGhw}Piy!zDY}ln{kt
z5O1!8D8?H>*gKKeN^5ux{hIy4<HkxC9I=WMamzW~$FkmtmJ*eh+r}tjV~|WJyfG%u
z-F-C5yt?Yj>lf=V;rbRGCSBj5!-(spGqYQ4I?%0|d3K9U2iLRyp$5zKz3ocbd3#1*
z$Cp;;UuWNY9RnK!XnWVs^t~nQ{A~&ES+l#<S68b~gYwxX6<nr7!FI6k<j9(lyKS~)
z^AR>4DdHjjd2wogp8ayJ>1)57YyH@7b?%=zWiJD|yl~k|(^e{c9-a<-@T|R(xXOK2
z?fs9zWH9+;ENr5)MV-wC5GEL-*M^a_p^v})FTTHDwa|8bnyMB4`!9YE2p3O<K}Kr=
z&Ex3MuHW<+2m0&IGz+!(DPy7j%mEf^H0GyeV_;W|X`7jWy4pb{9MrM+LM;mkN=)1L
zGEkKc>PHT0tm+yoC^2oN8K~17)VCbeo?SrQ=oqAXpMycVVU_1T*Vmu@_@z19moNav
zGM#uhBjNP_<JnOvPe+9r!7yWV9tVSF8!Y$=#uwYuDL_R%Cx4e!(eA*F1i2GVTvkZ8
zBiG2u|DbVD<#13zn56H+HaZR}0AwsN6%J}VC<cy&2>=`Qg;#)x)hb2rCwcf6;Nkxw
zHJYu%B(p$sitAWnMnNP|UXH1NS5sS9ywM-*PK|6B?4yCtYrdTl?0MKMT`L~q+C?%s
z-&#=)XzQV`&v#Z9c{j5kpst`x`+Hprys|*^k6p#k_w+Uc7&?rP=UaLIO$Oi6E6aoq
znlI$k*g=EBWZtZ;7UJfjXGGxUzHSdoWGj1${lQ$t%H9k#N9?iff#z~62?P06&eHxE
zKND*he=1sgf%rT;>AFKGqxzUU?e$z2@~s00!sM-0x->9esO06qg3&x0itlLb=ol+E
zlZlu!!_GDralYGP_|RS*h$il9XHG<G+hes`$AZyn2@L85(?*eRC=?ey{IIV#o=@BV
z<RJ2@SOLwBRncVrNbz~2B74@x;%~=%n{+jq_!JoZq#s*I{dRR?G&yNx4LDlLykd6c
zo|DYQ+=)jTF_AFxbe)}?azc%A=7h-~aGaVrRhrT&*r+eU5lFw*pvHPiH#Xi$hg#Zw
z0sBSya`L0b;xAb^b}xE&FR`$g(PS0!LL9#l^Su!Ey{b~wflz`M2=u1CobYTpiqIk@
zVl4h{plKi0PsuZQUre-E?69t>jtlS`Ul)nLAydhuTfp}-gHvu1QV-VM0=}OaoOw%D
z+uVzmPPln-+uZLit-QIVZEkF7^yZGXxnr05Z;r50AyT${nt~~<B&-tgrm9-HvoBew
zD!0r!4ll``s@Bv;Gh6f*D%&A%IQ3s3d7l~vztGfxWlvu-Y5;$<FPM8r+wv=x`R~M(
z8e#X?J7Pm751Cfy@W1Xu-zx9$vuW-2+&kNr|9DySPCrGVJ92pN_RyWmvKGA~SNEfL
z=GvX0yMu0lJNfQTJ{OgD6zERnodpIqoZ6_sbEkN-{>G>UpMSH$LZ_+yHVjM~oEQ#B
zTiKwp<mo4p2S7p2=5ZM-d8y)W9J3-W+@6{Brd8A5;CuLuV3a=Ppg!d=fj;F>bThr4
z7rjwKkym-6h9YmmjT(x)q8ruSv9h)G@F%O)9mQGoL%WvpiiL8KFL_s?Q}&pfAv^Wz
zPVC2TVpj|1{z3FN_kQ;3Mu?QQon45;@59zOZ3g~yefM=9{ez2EcE8_UndL8fe~!Cy
z0m|;rbysHVi{2k}S7zjk-k<NT%(@r7zrbCYT`zk7aP*@X_B!k1>d@&&(<R-r2lA^T
z%k9=xjvn*<!Ak3t-~{WG;3Vsm;D{Sj+b66|y*1&L)aw&ArPfdAavf*==RoruXx@lS
z|NS}U_R$n))qOf6zXXXQxNAbY^^x<TTvOkv28L0zE~b;GACBAB7ETyHjla;w%VHa*
zYUFiNb8d3Pv8JcQq^p7%iVsruF1sADypE3&;DdbuAF+!%QugZf<8d8O_v?XD7ce%%
zzw+sJFU^nW`$O*gqwf1F-S-c_kz3UvE#126sFv=O>Pju$N!1frx=b7Q7*mRSno``G
znKwMs7r%v#c5}(vfddHK_QJ98T@@v-vWg>nUL?Q?S!9OPeWu@hVwuS2sM^mW%=0hu
z)@T=jN6`^}iyCY=`{=s|AB-+yiKd?-{$}V7u5m@Pg}aCqO9{V}X!V8`v1}>fml9jT
zi&(pqNIMwPIWR0>O8BKji#fE2RZI!Ll-P=1G~7z~rNmp-_{>})e(6SIw!uFR!2}u?
z)BLW(F-d_c&kN&yo)fMm41z)vYE!V{V;+wV8+Neo`v^U#0F6Cj$XQl_x$>6`m$awH
z4<?QMu-!92fwPD(gJ;2>HNG{f>R4)8xv+O7$#~gY57sWhv4ulp&yZ=6Q=tNl?4<pu
z`<nwE`ivgxk1s#@`JtKTKhd)uo^2+M`p#y3HLNx2$ArX$WcD%Z;o0tnXG<#!zmowx
z-vLf^fbU@dUA9Pop0e<i3}CSXEOUUV4FLX1yy8^e>(4+9a!^M*sP|t0YO$cicW!(m
z-MZJsp(lsE!-3Y_q?g_!DDj<3Gf;nZP)|FkYo$`$0Tjx@5pD3D$F@jW*ucmfk`Eq1
zRfJu%Qhi_Qe9|7u^?cG^At8I3hxsjq3XCJz2vOd#fNkM+tTtQ)n-`SB()^E<z#Uae
zaRow6p!xa}<w;XjS&=B8AO%aLw!FqHkut%wUL7o-q~GvlCg*&Y<CghRO89%Mfa!P%
zVL0CTW{is_rj|z&Gs2O?R56x?u|!RAC;>-x9;Q(*8aLlC8KyN9--?it3?12fdn7S~
z8Xh+@T`V!VJd&79lrq?_H6cD+>Y9*m4gMyvr(}4fVG8sut5;j=Pfl6Mq8hf~egySx
z7q9G`vx03<gsW0Kfmriffd#+#6rQ>;J*d&*tM<Hl(yJp|P{7r$3xo(zk7*^Gykam8
zrQtvXC!vSEFzGkD@ZM!NTm10Af*6(90>t=SVdt1?X~)5E>*MwaX&xT77R7hAcAe<k
zxpuF1A9$w2+SX()uK6o>)4l${&7{bHIO2cluL#`LQjxPJG|vl~-1~fZ?R(miq@3!+
zeBcc%Xj0I)IO&jYi*Tt4p`@Tm|9R|^|EfAeHfPR%277oGCMmIu=+n+262|mMa`+gT
z^u!nts*PJ))j_iG5kGlMT1>%bNQUupj>1DSm_KZGK3Vb<BvmnCmZ(J>*VxfTuj~31
z6%VA3;w6~|_ZBH;t8H#|k7#Ok>b<^B#og+#(=KscQw8YTCT3-a>Ly(vr+Tf-rY=M<
z32<(<sKEsCkhycb8YPR{T^0*A=KDy4pzyHRxs|g|b0E-kqol6*-;q$>Q`1|hUf-{#
z(HHmHx2T{;@8z3S0ehkpB5g9K3W=jPX$e-@Ewh-bNn!XLI!6r_ldNyhSm?8&jJ>{;
zM6waA$TYlI4>0iMT~eqI>@T{d9N;)%_!Q&;=rETSG0yclN<ekkcuLu3pbo3D({1X}
z8at&&-FPYd{$z?slBgk=YIi?ShqYTN`T6d$0O2vE>ceV@-C2oPmU1`HRDlIeLc%DU
zteQ6)XfBgvP}`<QPiY}}51YQ2!^jl)N-&kZSo$Tq!Yoh0v?~TiYtOgZHuvmWu7Wkp
zr<*OpK4h-Ra0sSOW-yXTlf`S*RMQuuF#AyF4V;UjO!oBvp|0V=pZe$*?d-&!Z!y{@
z?!0nn&|PL(zngalgF$OGbTfEJvQ44JPam#hgWYs}q9K<ora7VRvbkaRU5{!6#VBiJ
z)!CBvW@vad2s(<hRU^S%SoamOova3wbziwqb5^I?2Sg8c=oBM6(bql8eL4!eULc50
zgYaVa1Jv$aH`oz_s-D!RX=4!Yw#w0sO|}_@g&yr@yeAnpOF=nF@Pso1^A%r^G*P$F
z;z06eCY4I#z?d~~U$a9m>I<*t!pz8YI79PAUno8gk~rV6914He`=DHxGr>>=`RIyq
z8DJP1$mPY$a2fg?hm5`oD0a!I;SX|CL~Oi-(gL1q?F>B1<`tz<4)O>UGt!-Bq{~L7
z%YvP6FEco-9|jccJO&`~B;zGF7f2_r;ea%mN;2?BZZMqu#`#~)R;9T&a9zqt{tahx
zR>;}#8}wOf5EqYIGK|#n>djm@f1b9@S3I`0ZOi$&Z<%iWEvx|5>nL3IvI<vO;mRy(
z^B1e$>oty|yWacTesk3@oR%;c49hL%71gVNn#BII%|CqXS=F%8tHB@&wC8#i_m;7z
z@BCaOQ)EDF%a|KlvQ(A7RL#^X*j3$Py>3itQ805Yc9UMlGZ3ZQ3jR`q<TjyZEu>jk
z%yAh|pa36xhKdPCO<-2#Khr`VR6p8r*?|S+(%egW8yskSoZcV#8vZl$IKdG_j?7)<
zSgj7>pVjZhh7BB9PDdLJxB#NfE!Q5^)d|b0NS$Fh#lFS`VPDqJE-311t|Vg@3WYUL
zB-}sGj?6^4JTd6~`N_VSB=v}<6I4rhPai6Mr3jeoAAjcqk9hOO_;DuJMip$hfCu_!
zi>87byttM`;E{a%NWPIj(oLRUYLspETz_`bU&dg*N@q_g3NjrYLzPgSvA?vfxN6vN
z!sf4YQ)P|rEw_Pg0U>wkwm&yLl86EL@RE;A8|c}43CU8A1m=%s-KHm}heL800UJVS
zgy){4sgHlIr)z<mp_k0a`39rfH3K<8{kee$h2{$-&v`LQ>8TyOD6>9gD5ZDQX`Ky!
z^}5SM6AB)%0!=NZ+9YL%&zr<zX1rKx7E41t?>aB3&ZD#PF&uy8r7ybPyOn~_s6q5t
z{&ktIEAYDRhq))jels}hNX>N2SAG5V<J|1i(>GeOQk#B=fQVz&NzC20&e#5ONFxtd
zBW+B*7V$yhM5Z}#d0#oJe{+ieL38>>vR_mBhA`cX&`I9x;(HiF4M`7Qt^sOn!{)VO
zzv4nPK6NnAyjTQvPrpU<Cc1r1@FhH;NSX@5)XBM$5)^B&<Pz3l1!!(4@CN4XhrMr3
z&*Nj%35U*O!l$;V61fv96KR@7_vD9!3o42dsK=o8VJq}8(@#0gjnyLzmCyR-`pi&$
ze3t*izQOgxe__`Fbo{~)8GV7{drsWbZ!6ZEwi(ciJ`=l5-%>9#)tsXoS=6KoB_>xP
zZ3H8=FNQJo&rj$c#G}x-l#1OcWSF}t!yuoST$!kv5cUKZbzNKQAp(rJA#dIH2`yA#
zuG)A{d01q;5U4&0W@<5N^3qV^ytEWELzRGaoD{SBlcks%q?j4h_P8<8Ce5^mpny)A
zX-`WtL3%+_%xad$u=aaADdsVzvlyWzyh*0H1<x{z;;*D-m~BpmSrrIbifI<x)yOdG
zkYP-2SW34EMQ+fsY`*bAfdz3*L}@9(xU)DX%Xui?g|rg#b*06Xc72)iQ<%DYg<q_{
zFp_vBO--ALDB!R1<5jr=;j%_{3+fz2oh7Ot97Jovz2UW_DS^K;ZY4{#Hf}<W@jM{m
za_MW2wX_}MaE}jYtMAqL-kDkwgihiGF<}qVfbP*spa6~*+6uG1SR<qw3hby<btJ`v
z4JeilM9DGXk+>CTx(Sj+_z0?h+&yHXiC17xQ@~u(qY{6K5us-4J(2h-vC7tqYOxgU
zQ@hWHzVKr8vEp6x!Bukgm@2NlfkMso8RF1|3iW&oM~tWnEvY?zyP{_}?i2&5Gr8(J
z)bUZ@hPF+sPHAD`H5uDZcQ-l7jz~2gyWFE%VP#b-n1b|FE?T1+GCAbypu^NONh=L@
z#5qn4;#v(F`C9kq6z^<>^uWS8dU=PA$_bkapD}@1-83BfLyWsnvc?Dds2Mx+Sxz)p
z14)YnBddWpV*IF)vor{r1^8V$8mwtAc5eDGruDt{4d{To$}}V=mfK6`>80$Ojk8g(
z55(y^Jrcus<QTf&{=gyV{&=#U^H`cDiFQGgOhp(;T<>}w>L@HYkT#(|4dyP<Z{-qD
zxJSnwF`|;;$7N6(SCz;4?e&))`ue-9pzrnRcj|RUbM%7f5_1i#T#~D?fou+V84!8H
zPs+Samz5{MyrrP5jw@9zdO|Tbwm?L>SzEAST1dDtJ57Vwcr8j%bsGV`bfG#G+M>Rf
z5_r7GbU4uTDXmS#c(x%Q>{5EEyA&m$kJ>EZaeXo0PIq?J-|uxhjSq@9xR-+TYxPm3
z9<M8_bvYYe4&kEYKwhZaw`zpkIs*^9q@Fo+04Blc!brW&Ylj~WA8cIq8m=l2XH4xY
z+$VL7@tD(5{=ocY(9v*XlaWZ%!)V^a?xOKd`maU(LO*U$f8cWvP5Q4BdMHFkaw|F#
zyt5{-Zna4aqZnF}n{Z5i3|Cro=t-XW7=P>1t!dw!hjb)(i<l7GRg(E+JH-j)W)sJ?
z*kR8`SzInhPt=_8K$(NJC*TX}ag%T5OAfB6vT9a@K@kppt(Uo6%bgoq>H^T>Dz~Fo
zjUr;Y%9VN!!Zb?Dy~ZAKO%onfYtwzw71u{*=ri9kkGmCk?b6DH0(6qAn_o>P{CQt`
z8+Ri<`%+1K_N8nM&@<vi4IP4y;eRyj;A6VXiRJc<6NvUo5%_rqj|7_TF{<GZpuWOh
z?t^TA?m=OymheG=7@CCN3pIifwY)`HJRL)8-<=lSafomSU#xGE4gj;}U0x<OuO;-L
z5UKcH(7wZ<GH%xBr(9Yf`yjfqP+jRX49;*S(O-R_V*_=lYfJ5#4GI{6*%R?2^<s{i
zK2ZKAu`tt>XLOurM~U|IzAa0X0xTk@M9|S%$T3Sb0c49?LVuv?9Q7Iv&`V6}r{k#4
zL)L-$*Eqr7A?*9@c`tn*NW<y{S!lIOye~KLKmxzoByB*09f{wo*DmFP_3yS7aN@KS
zj070L(^qqWO+cK}g8&bZ@1N@Y;_p2y5;iQ7(|NHgXmTRA!z_oV%Yro4;QXCmwX&dU
z?vwKO1!p3vEU(6Wz90Ufzi-o<6DGa!e6E<}{8D~~7YpSGRk_xY=4`%u^F!6!xIk#a
zxN2Jo^G1}V9tHbg*j>vf<sZl6A~DOW(J@Pd0S<qbvYQRRN|0YAyq#N;+E)FnlihWB
zl7@c`%`q?i)gW>5dLs<52MFwL$K`(ZKH`MRFm^o;PCEqO2*qnWyDmjCw$xb9e!zgk
zW)y~#=fhlWQ`|e$Y|Kp?O>Gn6e24Wu;hZw8RZlGG|CTV{RN}QP1mWS}=0p!6R$;(3
z_-@6LM21Y}3iB~~IFWMkOG43%R0!V?G+k4$yBaP=pH{#{#An&Qc4%r5jg}V{TB#wl
zPAi3v>JgUgup8~d0&DEX^C(5d1TEnz(6sA)`F3rfC9V#wrxx0gF4NmorfDe5XLT_3
zty=MyPSJ|zMvQph>I5E81cr1cIMxkXU==w%rx3z}Tz4g?^^%2v_Hd<?wP~8_!GnXA
z>NysN;Pnxc?b=fk0$*j<k2~ZJeAuq<btsyeTl8FSOT_nqr7DAw)O5ftA8tXK*Cr1W
zE@Co7g!UXVWc$=7aDxr1oUb-+5QHNvq!-YA<R|k|Z3qMUHqbHpDD|0|tm9&Px=+99
z@S67tAidk{zWWqY``dGADuZu?-FKfR#%~_iE>~r7_fH4Ndt#^i=K_cES6w8#Rvr3A
zhw>cm6t>g5bzg0=q}aOF6jxPvmZIvMpls{o>8AJwPt{q&G&1%ynPE>Cqn;9OKrKsO
zg22N=cZOTL3hR!NGKyu=inZt3VdifuXQU6UAg~~8Iu}omrF?VNK<1N>BK_^L$@FT2
z3itx`rihfan2BJ$$NDI>a;FK^0??O(HSRTPrpO--Z{t**vA>dc6~HA_43#xeMxNZv
z+^{V)m&z#2)@tL1#5Uj)Elz@|z#DV)#<JBRP<Ixm1e`JY@JB-;5?jI|X%07wor1A0
zsAxPec=|DjSzVKP#1K%1A4qS6)B1l)6DJ6W7U3w>Zu&L&L*tFlzxNOxv4MhtxcQIn
z&B@$isXc_JBMG+;p5fsu98dT|ks|yLaPGW1m|x?u%l%oyi|ND(ApgzP6NHNM^qXcD
zm7QcfW%(D<sui7&{pOHXZKFVAp%2g^)(25tMt~U5;$*bgpKfCE^sgJs`opu6p;AkZ
zmucG+$F{z~YlmN);pE8=XzhqJw)w+w+C~T(Q!w#QF(@$cLfF2Hva%`5hN_zxh5}Sw
z8i^l_#h<1Qy0~-KuMhc<lkCe-B>QcL|DMR&?d+Kv(0r+Yc!%Z)1l*gz1P2G9(6zaX
z&rbRmYu0EaJc=oi+7PTL1^tr0(9n<pIF+oj?85b#!HnXJG-I6m8*QyfGr3gUC()=A
z6lif$QLt;cBXODe_0Q2POTFs`bA8DE*%N6qd;MKzX1|UDX7+Ec#DKp0iy(1Zsfp}!
zGk{SCc##8q>tO&J1n6xoIynRQSqFH!1Khg|!0QF*c?<0STl&@e?zC5b$^o9J8hcNJ
zCQik||565Oy@UEY2O9W$Rd*QGnYX}$8L0am)RPYC7lIlhs3Pk6c?N2pgSy>8O%c>Q
zpzszr$Q0=8`W0TPY5Xz3KUGFb(ZLRl9?O0MN9QuI84FOO0+cU0uFqxQMG7Uh2)rWk
zlghD)v#Z7?uAZP7qDSTSDI%+R%sFT^P5Q1V=Nio72||f!Sp3h#;y<z2v-TTWq6NZh
z0}cMM#F;_i6X90QOQT?<Qce(Dz`&8jh2>lzGBz3M+Orqk8)_mDX9Me&Ab_4LKowWg
zsWngyQ}C2lrtT4q3s9o%y11S5>|;6ayi98!Y9)MaE&-df_1NA|(xy?Fp$N2QYJFhA
zUx8JGdVK?z|DntHMrYg~4@AXocBUdH%#bX>(t#^RLAKv~1e_6rr>@Xbq6F0OG3{6Z
zwWPIU4FE*ryVUpA4BH~@^?h>mnbC|(v1dMo!qIWjWNrmQZ8Hxm8V5X^y0J!*OFKo~
z<;J*q5}@R=w7hCO+--V4JY1aJ*d7Z>E>Z}q`YX`%7j%6J=qd_0RiF&YLTjcg6ZKG;
zFlrwUnwF7>F1laT#J&bX(qm7~cT5sYSUFF1|A?-c!3lZjL<B%uW^?qQX>FLBV!;G0
z5?J6^VM?;>&Ljv&g00E;4w8XF`4fCQ*~K0`<Bx6>n?4ef<t-YqsIRka({fLiEL1_f
z6U_vvOCagiJZb{by(@=@+bBqr=8##V)lf5D+P+h9<oI6kCDsxgD8jM$R>A>)b%6f6
z7s=Qy8#CC*?F7}1Z<V&O;s&e96=>Q`BSZ0?P;!Q}ZVkL==2M(haS}*wF*4zZxP;X4
z>7*d&++gHQM_`SA*buqeUd)5j<#KkfG3O&&y6#X_rhjv5hE#Wv`1*bjee&F-BElDV
zybP0Rlo9($cX`xkGS7sHayxIb0?24Cwju0c%>OAxxgzsI9ci@~<KJPIrW?aLBy-Ht
zVV-p;@JMK&H3r2x>Y!yCW!va)Tl4I%e;Z&_OIu{jFqG<Ce&-=A+-U_~3!z0`n<9O-
zp;8)iYPvBpRh15~4P{z`eYp+OY-XpGvj(d*=VwfOYK?3l$7zj~(%V{}xbC%})vJ_J
z<Ry_lElMV-Ndzyl2@c-i=L|2rK>?4T;B$6BiRFY4bpxsAt0beh@UY~=rK*IZxVk`-
z?YCl=jdXl9dM}*EV(t*djHz5lDnYDEn#>wUwo5ZmjBw{ocX-`3${Wz-6lkhy_%g~p
zPxav2kbXbT<E$7&ERV|cS(StZwDQsoRCv--&&KrB(VmSd)9gcSO_Pd7Y2;sT+k#o}
ztJn9L1s$8GzQ!zgLi$FFnFZ<Z&@dWH|8(7iyqwv>Ij2peu%C(PP%>AMxbS0S&c?>b
zTnN2U;)ihIKgOFaP1~lWY1>kDvh+hn*fKVtG-2C8|D_3Al_P8p`*2<^y4Gm`SL5cE
zp>4t-eY9;9S0PyU;tXN?J>Q)sY~SR(^A=X0tVHdAnw3bZ`A@bIv6Ppn#n3giJg{J+
z@CiEmLeAH5E^%)jcZ4_=3i7rvt<0eVG4zOfK1iXFw^4!S@0i)g>C)(V=7*jTtI092
zV6N(B0jLvjuR~w)F8Nx*adWT}t|Ngqf|v^^f*DG%);eiwsZC>Wc5?JMY(YlTq+nHl
zF_WQK2+-&&BGKe<PhYWj<U2&Uq_w5IUHZPs-Cr3f&&Z{Y=gQ$=$@WE*T2Yv=^p&{E
zw5!cc&q=e)zB&?Q3YQk@gDm_9dKS;j#Aw5kJj133(~+QS)#9{<ryEE&=U5{YJ_RgY
z&^3p4OTW;-k#M~%ZFOW{pSdqf_np4fDquE42B|=6Me9~e&&V#<ReA+Db{^$wLXc}C
zeQ%>yfUrb;TSfY8pqsKxu609cq|Yo&uqnlk?tu*)vC+t@CVIImO+JLa3Voe|P#YLB
zYYQEuwh-b@)6beChdI|G<Pxk*Umi&R(5$eYB;4|a<JAL%s8kvwh5)%I(DVv(Cf=Ex
zEYhciF!LEsMsX5IKCS7*u-3@BWXKJEOmhp;=LOh-(^+Z_dff1#*q`y!rM{a~M6k`?
zF2z$Fpr3H0-eNprAUhC?w`G9S@-6hwab6;~a&c<woOFwaE=$w&&k8cs1uQDGs<Cni
z)SMACR2{U=i`-{ISgC$1OZ^!+s#J)NjV{PhC7vt^y5q94#0lD+Hy`ruV-I<^=ZxXK
zMZ`l}AA-4s_Nu}|Uqw^<y@@Mcguf7k$}J>03$bP<hZTB+|Hi?rQ74;4O5+8=(rPw>
zr4Q6xbl>fb6LYnNa?NfV9y;PdgJXC+b<QTOJITINMY0+gC$WGvOu~m9x8`|8sQ+oC
zh}K51Jf|?)-Y6VF*J%4Kb-EWs63xOP$|Y+H*|#AY-^*K^ek?+B-0ho<;97=srWfzT
ziI6&T<$YD@OZOODgQH5V49>Bh%7MY*vhrB^EB$fqW4p*mEZ?RE8UQZ55-Sx`URUw<
z*3J`s?Q46&wek+8RP&o=?F=XL$LRw~X#zPYoT8d@5GJ`4{p0iw>w0bJkkwVD@#M+%
zx!LTe6Vu&8_0rRnQepR`g(7q+A})=FbFyB;>&kUhe4mGnqxOtbQ|<1}o&t_QHV-@N
zHzQ7Vv-moR5vt7cs{$Kn`ZY!=Eqtv(Et1<*1HPaXh3(Y6%Cn#ABfBpYLe9wlv}p#V
z71Ge5#MC^4)=C)!-YZo5vbZe`HGKU9@%5O`K^7o4Vw686;Y3(6&~&ti+=Nkt`OD3Q
zo@VZYl<NC6Nh^|mA5GP3C9V~y{RO&DcNVG(7u(0UH>VMuEZm@cw<!$6&UZcjSe<DT
zOZFY94AEaM)br9~>c~rLR5{mHFGqWs+!DAiG+)epwBLr?Ty3FvyJySMy%4Ys9B2L_
zBLk$kRMqqJ3i@v7UQ=wEmmTrYTCP2|Ya$(jmz}4xk*nQ!N;IbSeh`z|qlKz=3NxwQ
zOm~O-<!k3u?E}V3l0b{#>AA^<ibWu_hJ+idM!qBHe&fsPL=0fbraDJ1JcPrnu>zdu
za-Hvzo3#3@4b<oHoUq}!r7W;u?tDD&EQm0HrPnzT!|do7IkhM^_7@OTNk?BLa+@5E
z>t!FK-si|=E|sIeWvM|L!BV+vA&XIgtuumTG5yO91m=&l8RkCUGv__1?*_f1$46du
z{a_ELu^Ki{&F0A-(p;)0Dfq}=80@YVA-sOG-pz!r18}aeW5sSv7--czk7HoLmp=rB
zz{1CO2!HGl{y0dgfN=6>8O~~lW~^q7<E)UO(>#+F(-KvOfk3aVI+s9a=Oqbz3#!pz
zYaSE9jLVmz1U>*<<BGNDm~)cD=a|(%-%Hr@^pZI@dHQuiQaL;cn>Lp~8zWbCH2{^?
zM`?2QvJDV7eP%0QScxUSm%>hB{k3UlukAMfeWpJ*Vgz5%JT(}o&BNxuZ%7J#Ev3p|
zNa}2Gn|$fD34kTiuG-_o>ODdhviRz>o4>7nGA1dW9J5!g(<(PFd=RDT3qup=wo!Ka
zVHj6D;RIF918NZ)%w#B1tFbK+8A5nfovHQTNsVH8I&=-x20Jo;ICkB`&3@~FUTCA3
z#O!?xpjU%<18y1V<Y+U{j%XCr6nd@4u3>o$4#NaTVRl)O;Ptjx&@dS1huwAyVM$`>
zfs6IL5AOPAZ>z*+q41?V9ca3csi@B_2S)D{qaXFHY1_2UQ;giEm&Mo6f5sqI*`<3o
zbAheAo?e+LRT%QMFdg)T>f-UL62w_<^uR72HoPnLt;c%J(Ft{Pxjp2*RlPXX-s8Sm
zeLvamyKhfhu$;w-=o4n~YA^`+aBA+J>ima9MhZOp(|Fy-uC&-YT8*^YobmNKA!%y{
zKw%F^%~bgeyOv$vS)vGD3Aqrd4hNb(M<0^2#xq5QLut-+PKI;!qdQaY?of3ic24B^
zM2plhU5bwY?$%0|jViF<D~5Zp)rh!id;V1Q?`^6~+Bo|%I$juDOS@bnZ7t?K*3nEZ
z`y!6|Lg57KVBpjZlC{(=pnpfiO>;N_K8lg4;YRlr>HbuqbABnL96sLsKKg9IK4hVp
z!gvT;-X<z>jFs5koq5=yRkK#9W~DuYUXKr#J)?=+KvAG+tW8R)(ab5ONYf~zd=_?u
z<DXenGd&<@G9^{;=&{~CL&mDz+|aaYp-VJHm)g<sQ~lPXagVv3x5N)&B-<#^3f;hC
z5jsbhP;1>ap7C>vtZ?EMcPIU&FVTa!)fSkavwV0+qfr^g@g5k*E4ecPq9^N><OH3k
z)f%#1O%}7CKMkU}qI3VeLmGCfHO%7+89n<h8rEVBq)RfbP3BTLjlq0uO$jHA6(*tp
z-15g;bw#EPQ6CQ0v|uWaq1po&&6WJolp>%?%(=c92?1@@;|vtce=#HAQ>SR{kv7*T
zn+8={sKu1=nO2o5jKsG`<Es&#F>ZApf9MCEO#y-<u>m4+q|d)g_N|GmeJ4T)Kn6ks
z(oY=~)ZF0ng_ShvZ_{nFa7E)?yxsBn>IWZIFF|%#1?C+hTUUc!-fSdH-DKvjc~<B=
zFEZw?DUt{V_FL=Kc@vI+^7%_LotN%2nY^syx>iEZGxnK(73&*f>B@3;uWNskHa^Xo
zW5%Zi955@5W2@t`XG_f#^gi>Vb7Qj?uq4A@_p=8zZl94jtg+|l!1Wa%BgH6*f7smO
zEiBaRfu`d@(T%bS9RommBWb?0{)t%yiCAI0omd!82Obz(e0#;7UL@<BP~u0Wk;eV_
z|Bf&OcTtedqm`(~73n`h5tmg87<Vb>;^N9Dr1q9TD58n_QPISu<*D~iY3Y3X=Y&k%
za@D6=hNeCm(Xx!8R8Fovt*ih*C+?hCa6$HW>IP@=Ms^;Y!gR(3he>>KD_h;X^)cK5
z)92X>0p%uAVdodPZCstrzk=r10t-TeF^x3$1{R#b37CX<*$YZJ#YY|;TY01wE`i#$
zpW3gp$&p$%I?Gq1_E{ZAVK35gJfcdDPI3?jNUzaEt2GB@Vn-OjJF7g<^fEmv?U3Zj
zQ34L9w#hcdVqh0WcaYhJxG8*!?qhr%R(xGb2(AK!b%)4b{FHpcpR;K-yz@^i4pN*D
z@Ftu*qTwLhOJun&g#W$IKSyGye?;^QbCZIT)$XmSb$3S-f6{s-4IxV?8vl^$r)V?(
zVpq!kgxB@V<e4|&vvw+7oD608Xr{0R6sNO{chZdR>(eu05FZsy{3tUc+PoRD-_3|w
z%!ppXHUA#4a3Vb;Iy57mrYJolx;g)cGvY<w`OjuV#k5bH5npE^SO@o;5wijHX2kK_
zK4eCGi3vAW8-z<Z@n*!QIsF%A1U{BFBQ||(MtJe&4w(;6GZ6oc`7r9x`Ed7j`dFI+
z%dfbPY4OQ(;J$x(4*cqq=fGEZkLJKl%bx8GC?EDTQX{m@=%-P~V!dHe8cn@4)6mr0
z96(dQx(_sU6E<=z)=@B2=hx<F46>JG;D6`fA93*Cyao752R~f!Lmd2!4E&W2{(1-h
zlRJR_u7ls+1pLblR)07HKhD8l;NZ^{{P!IE<AVPW2VanZKgPj--obzFX5fc8`1ykW
znS<}VJN>;|ud(ml!-4O8^>*Noaqtraf0lz^nt}hTgMZq=KOy*I9sCHv9}B$4&kVrd
z#UB2yW9cB!H&5i4UX(SQhZ`+U`Y-KtwnC;quG7T1g{*lZb@EdjCU_+XIVn+wNS)Z;
zLE~ek9{4mR%%w9c`8m5PTzhQ}_d(KU^Y?4vil`SyodA|#t(-VJ4%jUJK>NAL{FA_N
z&&LRv#15EqNFTxR>54dUXx=`Nh;ODgjh1|*U@2?$v*Ent%uf@9WYS3l#KR40inK7W
z)7{4t%hL4Mgus3_=35n2aNjk7<Xo*xa@L`#ZZ$y;e{$G<#npR58R7~Fi5Z#mQ3I_E
zn5Gr%sij9yrk!kpkL-CNf+@Ih)8M*O2kIxJ5uNlGwp?DcDfQ|PItc%gSM-n~8TGEL
z-$Ke!a>^=oSB8FnBD^ca1*xx+Ne`Ka(4@cQ3@o*c0F(X#!`B}+S3aT9vfcHIyJVm4
z$DdmCjDFXozzR;fVX+QVZfMb=?uHH>M&FQn{i4mO^%reRJ$uovRNF=SQm_ADOKSZO
zx>L{o;QdtF4?Y4aZ_<1{nm1*>9?h$puSfGn&&Tpu2>)GwPXvbsFd$igoBu{bx-R8s
z*>m2KSJtfJz9p}G+pOZvccfQA#|&EqJ2|il?)xpXdZU}w-(vxHR?M<h`A`P_K?mRJ
z;9C{~|5pe9Il<>S__-PQ1_ytmga57IA93)zSumZOW)95XG-rDc)1mRI2{1G#02hf9
zcK@SrZAFzOIVIm?>wDjl;Vwtcmuq`1*{A>$6_t%Wg91&b5z7m9NEV_T7#!}nJTrl&
zq9g4N@xv%={5ri=qmwdD8aRnO!F1KX1|^$0Nzn>v!wtGa=PDG(A366{>nctYgYAX+
zBzTQ6YYA!#e-KU(Yy%-@RD4hCdnfw#tnKAuiuv$WB!7Hj<KH`k1AcZawE_-zf%;Kr
z<vK%?dM3gmz8@Ju5uX}qc_c89aVAKvpaKO=M-kL{>orboPS219jiQ2>FhUIHpfLOp
zjSgoSg(K#L%8=$@eb%t}TVtptgd9}$gF2E2A(9Lob5J+Kz7~})wsQYbs(GrOBf+Q2
z=TUCiOW}J|*rP&K7GFc+7+vh(0-u(l;nq$5k-fw)KjI@o<#HT$^rbHJ`Uz+L)Cx{M
z^D;Yr;X-%($~p)5jcpv;CeEtPUiK5_rPKj2COQ2yLt2G}Y8o=;5<9}Pb&ngMzmy%J
zMfYSk)H4!Mxi*EKdrWnrHN?`b9S=&3{h0y)7<uW{fL{!}v}Bz+N|*ky6GGTyfi6we
z(kb3f9-np01|TKuScjrnYSFdTR)-28Ibxi!C9YD^K=5P(x1A~^*gbk6e$(U)Vsf?Y
zZsA6z)gT4FXV#OYqfSb7*9U>7qoiVJ+?O-$a<!;fC_czL%@Nw#lN@~*p*ZzH_900>
z8J1`c&dX_QI>y)w>4;bv#)6kfp|m3wICD*;Y+s=9yDSLVVAjx56!W;gTXx>YZ%}Tk
zhK=DpFX8{G?$(%x2?IPz?HCWXUtysz%_<T}`pXB5&siviFTNxcM-8?DK{wp{yz<#8
zkI*2#Czc#DM4?wj3JN2V|N7TxN{J}WRFUL0hw+}h=+R=ypZa*B3qw~dS)0x2CVcl}
z$?pu}bS<Wb2pK$pC7kJVk>vLWbNY%Lcz8ki$d*eMj=`tp;!7&x`9nCXym%0&mDyQ#
z?pN_3J3Z`@c>Z8J%i-)WJ3Cyr{dSs5p?m&F6}t6C%vTUqdz1*QcXXW$UJjfX(kwyD
znK(@ufEm1Cpz%e`g<dOpGj($JTK<T-=Ty_!Gbc{_ZE<<W#A$yjF7Irsxcc(bXSY=h
zTi?}Ial-mdZMA!k>F~AvaPRD`ZDVGY5$vF1@9d7YF?)~MY^U7<==SN}b}M*W_W-NC
zYFJ%8+`NusvwUrzwc{W=em85d9WToA+x?5Ovh8|S)?s%1QPvPL_~oQdUjJI^3+oB_
zcp%Vxfrc*L9WC1yXgUK`Lf7m3FildkE`OQ7jr{$czwMff^Z4F=^wGX-hka#FTg4IU
zcTSvhFz0gr#rmopb{@UhH*w3XoMEA{YVqD<cGw5)&J<b)>_~$Qav&3@-CbOsDIDBZ
zk-dJes_aR_@o{^P>9MMMGgV~|06ENoOq_NLNfXk~kzj#Fy4I;x(M+ohf=jP?13=QP
z@}98v46MZ3v%|H7O0qR*#H_Yn@Sd{v^nnavf(6?uzOa5xTgA!iH)uw5Xl{^7Z)2eO
zUM5_)Y*(P^4;D<|F6VI&e{=V8NaF&H(J0g$jy^hq+4K&ck4E_NiajDZQYRvk=X3JO
zM6yPAJR&)ZbB{>g$*D&q-}{V+q(~Haw=hj4%MT%vsAVgXlhZ_UU65LN1_JnBBa(O&
zbzV{TiA3^mN7Kqfh~%GD_J5E_zVstUB;Vx-zg>O836mU=JmKFRk^G7~e&tUN@Ea)s
zib&pE`%j6aQ-UfQy5XPy0)70YZPNdrqmNkmq=|!!{x^u@7ty!;tMu?QX?i&2e;Yky
zLiQ(y<!NH5#I0eHVuRisG2C5u>Oj(!A%-8Lf5gjFKG(op(ewW;^zZwke<FI2LeW2w
z!nA4}5;&0mfS&z}^w3d3(L+ZW|1UlK$Mo=(|4R=Ckh1?T(!;OFj4#nJ2ZCi(ub7kY
zQWm-XHH6c0;nyiPl83(L0)aR!PCcg{$=iSYKa~ro%GR}zddDWOII*H}tI39!5tjTD
z$R1LjWW%&4pJy~N@z4av{<2|B=hf^y6}qVzC>vgKBz2=1$G=IhHSsVUSa1QwLRE-|
zGE}aOVR31<+n#S!GXE9>PfnGy7yI$HB9M6USMu~`SCwZRML=rYicsjdJSQm*5Dbfw
z2H(0Vd*p5jhZ`?;bZ<9DqkAV`lp$yE=#tN*xtB=%pfySY-&inK#$F<vYweYvzTlCM
zS^(8fb<9h8%9HnqdrPN8i7VbT@d$nY%X|;Pbny$Ys?(X_h{NLfMq7^{;&=iv#cKrV
zThpJaEwzWeMrbdBB-Sv#75zm%*CqnTr=`BH_SqjCeUlxLz>>&&7%|fFo-=4^GvWBO
zyodecA4+`Dr2nv(gr>B_m%qr9_>gt)G8s0O9DXO7HPS*@N{6#{A@RMc%qor32S|Jf
zCv19Fp4jgjkt1^ua|9{T#J;Cwe%`@{qjjkJl5T~eE6Y8xFN$=X^v`kjXQhHG`UQ^O
zh03(%>xhszwp#+9zA7g$cRze;EICb5AAt@{>N97Tz6$jTIQo+WCOJ(~ADcv;izUyL
z)JJOE_E_?2NqtyFkOeQ4)b~2R6_MnHqI$1G^<v4N45pO5cDPnJsjuJskt&e(;hvbC
zWop4+_@$CB@-CD(J+El6yQ9SEeQ?;bGIuQRSkVv%R-hY)>BA@eVz@G&6&*%=ZOMWP
zvAv*cD4(3sPpb2Rwo*Dqwy-3&@wYQACrSQOzPBFB_tuh{5zL*NO0D0|N#)$2C3c%e
zVz=BUQ-7mSdERuH|0I5%Iiz;$Cbir9;j*XdAB|{Zk`_55+7uV}KaxfmwtM>o58Dk<
zaM_PzyJIL$W4ls;IBa(gryjOD5{=eBVer9r&e>sXa%u+K4PpD*e+%2OEu(Y(WMMnb
z`e8eA#q1C!{W!MUq`VK-5&tK!-HvL9?Q$;ku-(_HgzY365h|PGq1evj+(Z55>%WBW
z%qwn<bxXADr9jg*IsO07<GV_+@c$~ZlM5`u8v3n|A-l2x?f);}yMMsZ|1ZFI#?kj7
zyNmxBva=A|eYi<TERF1(J^6nTzKi}td>2XnBpb}O9Z#`;gzv5vz7s<1$K|K#jl*{X
z5ZOOLcxkjY0N;V)G6?hk7x?Zp=|gI-@h7I|kQb!1_x3$FVz$$fdn<f<$Ci;8E=G*E
zRVDa6f+qtD$`q>(y@tZ0eSvfK6Jb{SvsjJB9(H;ET@JOuzzsukfNPzED`LE7&zcHf
zYy8E=zYifBJL+(@=mZ+K!D|qj=nvwvGVlMAdw&b>@2T+hj4fNk`<I8}%f0u9pm|dQ
z3m(pMjVPxPOE`}vuA&k5c#TlZXIgNW7B`J9kCMgsV5D&?DOaEJyVkT+#9yqi_VieL
zN&}7J1GUTaK-aK9|0eyjE=k~<5LoaTuSM#XWByT#J_VMQ7wxDB%;G0?3(eSQ%^+iH
z*Mb4f_(wgHvhFM)CkDe4!o`Tj+!E~To;c?kBX3tcV`SYUBS>-JhK(NHAJ|8#t5AbR
zH@%bYra<GFG>8!!qS|YDulWYcfWmaBp{t#vIN$HRUVU-6&Rz7{D2^yoY)dXZpxiEl
zVT$*MlQVL$FOZW8wuf{sL&;EEac;7*f-qT=ipKvx?7a(ol-0HWoq<7O8=Z;9Hny>i
zn)U?67AjV9q;C@(=#0)NYFYzLYno6?TUrqkixx4EM486H;IXY*t!QlztynL$iuVLE
za<Nt}UJ!2qyd*}Hn-H#fzrVGgnM?w0J?H$-`*}a_=gsGnd7fuI>)F?}*IsMwwby3A
zzJPo0+mILTNtI=TB)o7|U)d47*~$Ok__tjh`J60U7V{BeZ%cO(=b<S6gckvT0gnKg
zKS);Rv!*KI#tgNssMtT_2py_74SNb}x7=p(gV1$5@!jyE)rO!~F^SN1JsxoLmB)n;
z5*%=U2Ig@N=)W!lc7~*dR(}-F6$Fnv32hi?tdM0!;$D;;41xQQW{IcDnXKotORU~4
zn5N}<I8)F+uP<lJt9a)`eQ$?LR@H6lI!a0X<4z$HhRxL)Yt&l9=5pPqItCYc+do$N
zzuPURWz2W~ynoh75m8D1ztlgG8kWpZ39QSdZ`O;dQ{x6%jiI}N(+-R<>7u{uKGj78
zB>IQL22aR@lEQBR%9+FYd=KFAFqwVZ8XxgZk+x0#^AeR0_<4SfQS?BLMv(>%|E67N
z@I1g_;J;Tg8bK#n|0vpqH+cLucorAcgAiyuJylHWpIzM-rN$3OtEJ|9hRP#5863GG
zr$K}s>9MC1+=zKZXHw=esw5fsAJLZA6Br`65BOe`2f<S|MC$s6$Tk}ygj3NFxdD&;
z`xzn`<AYS@lghY^aq(W4{Ea_OyH{Iu12;6jrTbL7v)ib)=#X)NvpDRQb~inrMqfaq
zFUV|kd_w;_?S4wC{r-9T|3!Q|JwE<l-u@w0V5f%S8+LDK+7g(5l!{)H#3i=*BAjKG
zMQfPf-`M>|Ss=UvTSX1swMM?2#ij@|HAomCreXP{5dxzwwnhn0z_zhiqWr=UG2cqF
z-$&y6MBR1(0B#QKkQ`vYUG9Gz(zO%0V(}MnA0!x920zL?p-a1X-0_wf^tJTrW2aWG
z#u#&K?RU6*bICjH2I_DiF5g`8h68PNpsjiuQ>hm(2{QFN*mA}#!3|-VoK&VSPQF&L
z);q^*r4H{}<x6g@Se4u{Yo-0~t!R@QkN<JG{adJWbe5)zMA5x$?P1Rt6Ih_h?EAg=
z8LbKfDCUfC$rwcJ=t8vXiX(?{cA9JZ%On!Hwm&yNT=QNizRMbZm)G#<G`1H2rzjeJ
zJq^0<+(h{W*6@VnaN#(t12(u02wgx2AT;R%#Jb|X_EN|noC4U$VY&c8m-_byk|Ms)
zXD|0(+Se<~Vv6<np7rO)>W^=)Ruk7=lF=XE?r@-t{`j_aMC~Q6KfdkygYSl4(yu%2
z{g~bHXS&1oRxML^l&d?4cOK&`xU(kXDto=xAKK-Y9m7R{PAQZ(KLjngv}u#NWji{4
z{Jq^9>~OtB^bt%d6lcdvYD~mG*!@Nz{E>A`tcJlNF`;iT+*;)ZLt6(s*;Aw8#;Qo$
zMxM!w^WZcd-i-M+iyHnH#)BNUZj%#Mu=X}NVFhR1#xm=yp#L^C@Z{SUwr`3psl2^a
zZCrjk-SM?0(c58Kj`lfs_rF2=KNv{+Uj=|iiyZ7qI9{c%kq@Ym6GKG%d%adZCZm--
z>OW}Izgvw=Cq-%wPyY+7twsMy(Q4S*JT(9o(B_-ja3_YPVUXtEM*lkrkMX}m|7!o=
zxc2{zYyaQ4_UB|i+W(G)?MrF@J6hY9(*AcaD!#TP=Z;kKw|&g!?@#Mf4xx3g{RhxG
zWNr--Xj1%_t|R;sP2PoidpVbdO$KjUGctn~H*8{Ky{SY-&KdkT8t;`;*Y?h3-`wAO
zexm5h(fGR#@F0MgkDV9Z^IZd;ZyfMEcfj-X0ngRxXIQ7p;(S1E*&AP#=wQgIWCI6#
zQvIea##~K93N#MK-JYG1lBOcQH+H|?ul&2^xrwQ$(`OmH)0?b&uWRoh)G-dtw6=IR
zobYc+wMC(|g*AiX+4R-Foa*+gKY&PP{RceXIN*8ifamE0o~s8uU&wR6$``8gQ>i>*
zJ2l4WT#Q<{6;c<)!o_tp;(N19)$il~>GBV@<UcH<`rTdoup^<{dew&%4qtuN<I~?S
zyJ%ql1L)WP1D<aj@H}_G^Yj7F)dQX{<oWRR?_d2dYhl1TU)8tTAGZFM{D)Nk{jU9V
zwi^jzH@@dH$3k|)+l+;@{kyOJ7pgcI!y$s9X`;UuXml_GEe-k2A&~eXke?j_`Til0
zi!vZKW4R$Q1~A;iMniVV>w%_k_Yp*K?H6_2q7K((92h0Eq{}~;4`aS}t^YX`uZJe+
z?|Q2<Qqu>P#@Fv&k0iJ(T=RUSW?Kl)Yu~a^{Po1Fp+3&;D~tD(?Oxy8+I^IErj#&U
zBIwlqF%n<NT!)+WHt!V2(Pi-uyFV*|qfb9TlFC`YT1%{Cdu^Mf3q12q!51qXAF5(y
zS?{ysO4iryjIRp~d$kPbgXt@~1K#`iIs+^zi!Uo1_UfhViQ@d#fAU|}u&k$UL-)M?
zwcBd7W4fVsTSM{PWxi(yRxQU>EztNpywla}d5*-@k(~tt>vI9?Cmco1#?-4<dR6*2
zt5PWr@Q2jr;K2F-_#f3L-nn~yAe?l475~te98ecy?A_@K9b|o_e}(e@LH|7WG5Uvt
zFQ>0k_c(nszD|HgUH4eLlmAZt^jQBa*PbcF0eYyn%(wh7wJPqbRdKpj9szbAC;lzT
zJtw1oR{!h%`9DH$II<r<HRZ_u>2u7Hy_gG*?59n^(R{{N*vq|2^H%eatv^m7eh@Ll
zUM?W!2(ea(9Nr<s<!QvELww00&c6c0tAvOvyAXeyM!d}--s=#*FT`6$f_T0|tV<)#
zaEOf#acnh+uZ#fkiw-fGMm)zMe%m3=!tv|T4MIGiX{dR~$!WwB9pYCU;!lOxCBzpT
z;@%&nYTXmIS|8~U@A)2x?+Ni9hqxw<_`E~x;6n4fEW~ah&UJ{3(ujX>h>tkL%j!U!
ztmc{M5U)uiUhNR)JH)6`yj#tq-7Hk*TWQ2{hxmPmc<rSi{$7Yf9pY(e#8VyO7>77h
zhz|+z4NX(cL;PvPK@Ra)hxpej5dSK~zdFQMf0$~XSH5A*^Bxy!y-|n{3-Nk~_+T3G
zeuvoR5U&y9C?VE3#CRI<I)`|(Lwpopy+_Xw;=en@sx;zv9pYsU@g5<DgqZ6P&rBnp
z;Sk3=#7UQcI8KP2Of%H_=rm%ELp;GD=BbS1h3LgkeS1o()|<y$t#@*v)(<MhULkt%
zQ=dp9KIRagb%^idnD}TGeQohmf0f3%$zlE8VQmyvKv+6$by^zhrw;2HhxH$LeLNZz
zRuNd|q_M&dYof!NprYl1#cs5x&ST%mAJ6$m@l!SFYReUyH%1+WuF19pIrL2A-W+<a
z<R;uOTN^rV(tD0s#fRrULF4<HhuWTwcd&|a9zDX}?EG-^1;KDUkJ4R${|z#3S-f`~
zOGcsQpB8fcQ2RO9t-vbw6rvNBf`he8XQ#?RWD$kozEt;HRj{&hUKci(RdA_lF4K{%
zrOjnkyuP2;q2}*R4mDps#kP8&+T!$++V6q+h-JpkJ0nhD8zI9gjxLz`wSbWk;`9{H
zx~CP1&)NN$B5sf6RzV3qsVj@)gSp1nvKNdTBF*=&Ab0ufh?Ziqt;>O>W^JtI`S?o>
z@A$O$prTB2!RCXq5$=8i+|}ToofuW5RhbFV#KmaiZm0ISi7`hi+JkfY98-j>)D!j*
z-h*gud?S0}G;qk!g&b3E#q1b4hIxND^O^i{%%uiz5pG-c4d>lha)iK8qXrts%C=_f
z+1~Dh!JhMu7dHS=J>|mV0stwXfRZAUFZH+2NCIZR1(~srUpZN8AZm;bG)_bG(^69v
z$Jy=;o$3NqsleK20ab7pmY2^I3aSR-b_6-}j~~jrn|VhT>MRb57@@3YrVeutpOse>
zXuN`)aG*IAihvyIg%5T(y%mU6_`rSIcEXm77QF>)Ay?%-Q+DiQQ}O}!RICJ%lILoh
z#>KZnq{(r<o|#?I5-4-_CKQq+=G1&&-33IS;;tgA_BoCVB5zWzC&igh#aqKA$=Q#%
zlj$pQN=n6}v?v3?)M`)Iw@zi8z*5a|{)(*bYk5%Zj~~#!NN&EZk&<<_ElsU6PasZz
z$?LN}74<Dk?hX3y@YZ>jvBGrWL2MYCeamJ(M|=rzo;ei9e_>YlebMGJ%uE*)H~-up
zYCbpn>_jNLm)-4qpfg`jsHC+l5NowgiHX$nv$L|;;gbk*O|3Wh?cPSl9(Ig{o4-Xv
z%loSNf>PZ~;Km^EfQ$0S#XpdHUaaQzSj{tL4o?N%W>8i|o39nYn0f23aDr&f^9sL6
z-;FEzAl#f&q;a9)9Y~ynDNIIoJngoIcdwQsF?a-Xe|vB^F&5h+oo@<|w+Tz>GJ1$L
z%d!<gd3j{ud{!vMz}wDAM6>wHizJcAnXa|YY1r!7(v@?$Uf4Kl;J^whRY(x@tRqr<
z$zqUgd)lAXjc<wm()vTnWkAw7E1fEY&Oa^n4nY;a(K#lHM`G1#d|1^YGYtTUR6Ft6
z>sckxJb`dVTVz+alPxu<!_5UnI?qZ8iK6Y0FxF}<Sl9|&yrN_i4m3(1Pq|7$y;?gY
z2!v7=Q=BO-XjaE*8px#tR~xV&OzAI7rSDo%ET$e1;Z3CGvlWa8C^+BXOr}g3__Dt#
z(}iHUJ+se-IjP6SZ6fCvq5@oXmdXN&(+klBZY!z>H4<NsJ-++PneCa}H>uUE+mD(t
z8(zq+C7cWjiXtU%1{%LZPNq7cs3kiuUYqQ=_*I$Y)BPbU@8n5l|D9Ta{r5XdUUwj~
z|4yyI;P{;-FFKGJey3L0Wz;0Scup^hDrBN;mrX2MvSa$ao+-XHxwE2G=6Fsl?Wt%>
zsiw)jlUid<TWZHkWn=~Qk>M3ulg(Mo-Ng2`=@WAnC$~>pncP=Fpr~b$lGp2AlTvJ}
z<N1SUe7WlrRH?80w)0pcFQ1*cAEN3Rx%IsdJTmpWds7)qowUC%2fg8u69ve6>ZF5&
zRhXE!SotR3o3u)A0qLzk6mF6bT;X{7Zr_)K`hRI9Z&>D!l)PFur>l}8W$Lcsnm1ki
zRCTD0VNG%zf)z+?WPGc26yStGyeGVSP1N^_8^8XmG=69HjbDZggOq62yEK1}AZ(W#
zxeSqI=dk56oESA*Z>EJ08K5F=xO3vP2~RLUM_YrX=|RM+@uIa7h}Om>i<dSonTgcg
zjG7arR50fmros;~DCHvuNPMI(0x1RP9-c|B3V^xL(zm#ujK^mb8Hlib^+0RuMe7T>
zqaF$u!`xuA8!d!Hf;K4*Hxo~T6FqwZ$ERrA5KG7m#Oyv?6Fv*CtJ+`bxW2%BV-T*`
ztP@QXJ*EMT2(?>|3ViPL-Q<o*D^162jP$lfeJ|VkgTJWuKaH@6A+kcmdb(~zuvrA#
z$kS;P<-gQPu$OSW4|fpE+T@OkRUW~<>JjXo!P6IlUw+A6_kr#w?(h4=-sz0H%3mto
z@`(o(0*K^DD!FOWI(vW6y%&t<yosQAIqLC}jIRgv|Jn-V=rGWv_1~@_Kad0R9cD0_
z1td<oDS+jt49*`C=V$W3^vbqq1uLP05mZblj5Ze)rR{4mPC^s`Oo+gB;_D7sFO0@}
z$e|sh+gxI%jofkwGjbb5()4GXrquDHaT)xm)bXQn8T_c!@uP8$AC;#0(d2Pi9z$w{
zA#G)S@GWc@?^pZ-hQvV5^)C)fu6aFt*6^Z0ll)2GrV5v<r9ss<&~&;1+oGIjSM27F
zCX1iTq6eg(t<AH2vsP1*VSiXFohIjo4r+0A{tC(26KGw0Gilp!qCCLQsJ}w}%pH3O
zTujE@#RJA?FM$Q=KF$)6w+rQSW3qNZgphW`1dVU#J~@NG%PX2OJ0a$Kg+(+)d8;S+
z+9feojN{KLrpyV-Al?27b)fj;I}`$DbZsjgN<5#!>1TT5t^LkfxzWcLx=X9Obm)@v
zI02w&@1DBXBJt&slJ?q%QuZ7F<Fsb1WPjcE?uAlm+Eu~kA&*W<>57kxHeGQA7j(sX
zsi|8~$WlsWOB!*$L%hQwe&+%ZuN9)FRQ@21_<e_Xg+u)Hw?Mp3iXhYYkg_z^*B#ab
zhqZkISicpPl*-Shv5t3Gr#P$=mFXSAl2ZB6#8e%2oN0CNJFF+9pl$;TrSgZRD3#@7
z6vCr4>zifYx~So*DzwJ!#zWVHn#Wdz&Kg@8Xv+1cqIc9~f4l6v-%&h;P;;a_-2BuE
zYq9YxeSS;FGOlj^R5WqI_s$^4z<uMTEsR$%kMXh```7X69bUz13jTvl7ttd(gxlUL
ztQ|UXZDjY3?jbtZXh)=EDciA^;>S(cmX58nTTByeT08qXJPV0*@UY6}>+vK}M#w2}
z*(*kmQxJf$KH?rU?BUGsrQD4}@oPGdu`w>*Rz^Gx_|cHpP8pol{VC_4uTTn!oCS)d
z(C}|U8@U$JBh*ozdmqB`k1?6wlC`yK$1e9T=<U6xH(IlUgloSNia#gL+Z{~v8|bt5
zx!M!tESC33vgb={uGk{~m%Ihwt?J&`BJ{F`9{-F$_uXo5dBTh<*-`uJzAt`us_x4l
zA$ana*8J=Y_rPhQJbc|f5TBHU+q=)bv;}~dc*aLpJUb9<pI^~75NrtruKgJm?|apy
z^*`5w+7w|em}5rj)e(<147p_(O^DJ(FZ8wFyO>CkK-t&be?0o*8i1jAp}Um2OI4(y
ztuR)yqVBaAhpLq<b>op1j;n={hogABZSDT8>$ssZw}#Z~efj;&^7H!ZbLsw4pn28C
zEd#ioTcMtVNKsIB$U>{$`!u~r`%SONc_f8S-__v>*^@o}ku*Q)doSGRBZ`wk+U`N&
zQPK}NnGcF!oqm&-`H-LaP~aXQgKWs4uBGcN8eb)(7Ofa?QZ2sXaP?=$*5TSa$J#sU
z+B+Jz_KwC6-`>jJANN`t{s+%!`(}z~Jbk+HjMZG=d!M-wp7GVwjc1$zyXY)r3pMx)
z()c$x{DlsGyExIA4u6R7x1MSE-%I0P;_#<A{9+mF@*Mt)XMx`ee!qA^QyQ+~2<WCI
zGVR3IG+Z^6hgW$BHD3Uax}Yl1bike_R))@+I62VtdED31XDT$bPwBrP)qEwa%hXqs
zCBn_(U*pBUTE)MPx|1kthv8pqc=g4R;$N^0@vl{V{A-)zU)v%jt?;kbH2<25){Yem
zjR`?DCNA|8y4jc-x@opt;O1+gt)Um@c8jXVG;Zd!Qb|2%cuo>(RY_6Q=NQA~J{XF#
zj&>jCcAe5XPSN6yz+OB~$?Y)1@pnfPIj-^CAs7IAu?p_iwBJ0wRLO)xGH(qH#b0iC
z7f!1D65)byd_7EhnGlG#oAXPJneMTD^RFSQI5_CRbJ18OzXrpJykAJo4w}rMnV$p*
zIf{fHXqZE;S|g%D2o%Tx@CZK2mi{aJDpR(lQ<5&~XsazKDPjrgXkqR3`m8E!M+YMG
zqhYf<V2=FHN2@*2K--X*CGx=^Z1A^Tl^bnBML=L)8PjH8wQ<Sn+8fK3hniHQER8UH
z!w7BWOg3rx!km8I49k6YL+yjOGnF;$Wt-O3d}Hai&w&U0=UWoW$ih8!_IrB2-_zE9
zPdocP_4Ip^u9r$P=RduNNJD#ewB+5od0~!HZ@dH&IWH0VJ+7N?rT7!&$-zCuK8^IY
zM@!zR>+X`jhC2R&RL38LRjA`roF%QpHEc}Ddg{D}rBza3ION6?^Acm8RCAfK(LEA=
zuP#D~p^NBxS@MFYc31@TH6dlx>rvP1L^7n;>8(h7EsK>N|FU@Kdm{u)*5L`Y_oM8D
z8}?<)7jux`<}Pfv3zh>tb?A~aH_{+^xlQuh3jFFa<hMD}jY^hDeoImJXwiG)T^MI*
zho;_rDhpPih>jbj3fYnPvPi?`4v%;zQEk;HF)E?z5@RK4)%F}eMS%DPO`VJUJA7_5
zQH?Nvb^ef>PGS<&Zp;k%)D0v??`TW0|KNzS^U(KE)c28&da?mh(i3Q?rf{+l6#D9h
zjcnb;a3_QlE~-i(yY|oFhCOVfIG&l<8#Y}Nx#t`kj-?hJ?fVM!|5-=>upE5U_9)=A
zZ!ZT}M<M&gN<F)9#<6N_S>U=SrsWkt{mQSV4$ZOl56!Xm56!Xm_s>!LvsF^_^?FN&
zKpVB6x9a!t2j<!9-26?CJX5#T`uAYrjG-=LH}})OcaqKAR(I*p#c%yPF#SN|sgQC;
z`;z{g%=EiH$$Pcb2w3%ssBh!Qwf*Xo+@Bj-&}zRO{Q4KzZ6TCih*7q#BTCRju?L2G
zpz-Tsk1{lmu&tbu&zAe=DLY%NUsboQt00quj*Gv9ZFKOs_{#*2+n?g0oW9Y=L(fSJ
zHy$dvb8;c{GTs`jkKbrKscpiF79k?+Pci~c;7qVgemata88osSKjjcF5r(>TAU{=7
zSRT-@Fza#h^`ZQf?Ygp9;!Q5=!)UQ^H0w3S6rH>!pH!j?by*HXRH?mvxxYnC5{*9}
ziT@4byGg)EXZ0E~*xz-iqwmdF$?LUCNA7}s`e&TY5kIwW!&DxD9FVLt`je3xD|x0?
z0$VWeT9a$vPwrJPS|=}SA8w>%ckPp~xM2B1J53Be#Xa=c!|Co}zdel6L*7G{PJ8z^
z<iFvimGQ|H&kcauw^y|HgStcl*Zv&(O26oG)3%#HQM^5lg2v#OiM8cKn2D5dZ+AE@
z_r^?c0<yVBJvrjM#MC)e(8io-;(LyneB*Nv$7tq5(Zhd<f1I>C%|Ax|L;j&qp(+0H
zL+(@dr9=1#%6Vr09FCvd_W$z}ag~3~Prm%G`H2*()TA6`=_%9p!hGBYEoo`Si$cJe
z1&)T~bgU6s+|B0F%>J_ZSw<kJ6Kdozga<UTnVfAeLuzvVfFlRI$$6C~=lm2s{Ez47
zR?W{Wp)LDI^Yf7JMN<03CI4pn#WXJH7mtlY{=eRn|GzBx|5Vc#MyBz<?C?t+{%?ez
zaQMF$ev8BRrSW^eXz2$XezoxDJN!D~S2_GO;Z%8^claG#D9_o#zro>;6aEN?e`gy1
z4u^lg!=G{y_?I~R<Ak3Deun-s=lw;<<ROOta0wI$;l^y;xxzZ;G|a;CdU&XoUbGQC
z(723WTh71pR9UTO2V*5y;#&TMJzbg~EBRi2py_3v#5P8Ro5M^AE26_Xw9ZveN`ZNs
z_0lc*`-Z3RSog;T;4|KHY1loVo_ZfGnMo$G)N3Be<p(@Q8m^SXK$C8D&|E{$;J~#D
z?TMvB;+|gjggzp37XqIJD)3lK^p=%*(MqYpB})U>JuXadFX8SiG<n3;hpDa%{Az@7
z^Epg~E(D|tnm(2mYqYrMz)94dw&m4)8grLBIw0IU4aA4sc^S>6hNvSkgjmW;5YZe`
zSUNN#>|q7%-feG;)=cv^Ps@%r|ELh3!}rm#o=b#AYTIf&iECiszBhcYmpm7Ub+xzc
zJ!QF#;GQqF1?Fz13dy|}w3c)S67N!tt~aet=LI7r5gPS_em!@jwE;aBZLUB)T<`Uq
zRxJYaj?+tejs{QnjCzXasP~=q9M9iRr4ud*lgTyd*Ck<k?iL=so_j>M>7IMup1hvx
z)JL>wr`jI!#52P6+#&-s7Io;Vo_kA}{d%tMIBdxx?*0n);kwct>M+*2@_uU?*OiOx
z-iY|nu3Ts@T~{jTRKKpAnd-{X8CW;l{eQJ9H|{-!**Ddh^7AhY%vDg)RF579nRcBA
zptI<-{k`hr!6trkhC|}~!3`U=SaQ?_pCZfw)|-U?$Y>1)oTWi1=%sP^w8GqN^TAC;
z3u(-FPm0->#rNS|(qYO|%?ih_Yj*3v-#c2hkSv0T)@%#cung-R2%xC%EYMO-*!=Lq
zHhK)M@=<gcK|b8r4<{;o5mxC+-V6uMf0M&j!o&836Moa@m`7TWM}AMbS>a)v+qEq`
z3=wlR>++ErLV2wgB}}t6)Kbju)m|vGDFTsizOJZ_Vqw#?!<P|W1&kf!4V#Fdytv4^
zY1oV5!BhF3XX_Bw&I;{I6ff=_Ex6KJ2$1#F8{aN|;j|T>rUBo6;e-yZoGe8I9V~U>
z#CPzbVew6al+v{*p<#IIxU+_$=!9NC)_N-lpHZw3%MS^^Em`0I&{Cx@?cf;2nBj0R
zOT=nB(+aj6j3?~)8JfrYj3X96-OBrO66GyN>!F20=wXB&bO4PW@GjtF8G8t|5G06-
z@gPm{3%BM`BTivWuA8+55;@}+dTzbAx4o<cr|6#tk9zJnW~v$vk5Pv(<Bw+7R%(pF
zo?z?ftZMnuDrUk@DH`4JKi>L&I8lbD68u%~QxrgOQfuRcAEpOWrZ}cCh7;c~<Ox<M
zzbPm}D0YeCz|>eY6J$9Lbf`p$v#BW*p#<oI6^3`V)o}_i+Ul28!UJa>S3MVa&sj=`
zhFR^wb3FLXN-N<fpmAeSmF_9D+8EeO!72ocFLnoIB*t0(6Wlk%<%Lff?$?GB16^ks
zZ6(>%S)V%LYZ0#%YF6y5`>a0wG*u;8JUN4Dq0qdOEfsatKJ(%SaZHXhbEXYi-aEjA
z+4acIa9d|#?WaU>w?uXy=&#Q}uUf(QNc+rZ%Gtv@tY(E+Y6X_H=k(by9^0ZvZnQJd
zj#4W7HIl;JpKQ^`*amn=d!mrY{c1EkFG?0WUXbDfBw9H8=V7ATu+2E1x)e_TuD&x8
zPkrY@!ejGfBWY;EIzhF6Vfy#vUzq+q`4^^t7yZK7<5%I&pkmY1(UP0oNQcD~*5;W$
zXlC?!^x>tEl9%eXIbQ)y5v9A57#MHY!i@G$OoesMzs6jCjH%ze0vt8IaMcDn2B=t>
zK<yQrt%BM0A$N3@PTxsIO{~{v;TkoEeA<HQE*+AR?QU+e)Ar=><y(vAPf#OdHLuP|
z<ejc|FoI>pop!3iU+8qG2%8FV+=FQNHx-F?)z%Y4)E!Dv5}17-I^C#y3E9i3=Ja&?
zm0dZBNLrrXo$F-k+FmyE*zzW(*|tHElCD6aX$TW-O`AwcLz*&o+TzqvBrcVkSn_$p
zbi?N2ykYHS#TTz_FDtzG)%LP{PL3`MUc9BfEc@bJ?PXaP_g3FIE9>Ic>N{(*E?((v
zSM8(c>ZMl~A6s5|pn4P6%1H;R@A^sMl4lL7r?2qT?cp5sa1I*IG7smV;cWJBdOe(8
zd!9V0*WT{4-`5P5S7|5(lMJ@QVBW(z=VG%gF!xes#p(-c3oH1YRaiki{QAkOpn?Sz
zRJEvLZTHc({_3i!`k#IE@jD39n++1TFK}C2@j`pq$+OlbPo4GRI80@U7PKXrdeU0X
zuJ_fQtl-uQMIccBG+H*RZg$tVLCLT$hzeYf(%y8rdRZ`i7Q#W{906AV)P4JjR|N}x
z-ClFzi3Pt7Y7~f;^rHImqWp=xA!}+dtYV#VZdFpLoCA%UkYUBTr$vWtKR;1kXZ<|k
zoJ2vcd>utd=O?DtVhY`rsjm@la%$Z{qGHaPevqRBqcaX#A)dBE+z@mb-)9+v8m-F$
zb<A`yhtCjV5s1g1SactV1^E+;Zm_}$`!%~`S!qzTWP9B$T_H-R_U;>;n?(Fl8=}qg
zY?M)74NlCpkn2sJ{0!F$6P!uMA{E*PZe&J1w^<=-EIRAFCy^sQ^c%6r6YT3gVno!p
zE|Q2IZ-cNZDs$KMHq8)nAnbbuTfL1=p)i$V_R{MANK<e7=)Jw{)JX(mGv=ys=i?;P
z-(z)pCQ|Z2pz&ym7^`6`vKw}i+-=yo^>-U~Ifor>Ec-h9Y5O2H8eg!w{iLgpops^r
z_OVAT`Otwn9H>K}cO7V}18uFI%h>Z@pf_&{H0r|Y3$HF*@)DvS2Gga}?-@w>fz0}N
zq`%JSeI50i#rWXI2>#Oyr1r8?RORbU!s2wb+DX#fbrRw*Skr*^fk+C@NkYbVKqCd=
z<Vh0??r1MNNo0F!#X4K(TUob3K7<+pDe}cP?RSiu?oX%O1v6gge0SOZq3dS}*mBNG
zkZ~+4ieRU+Tt&aJy~^-D-JY{oIYO`GXs`D3s>r?CpZv_kB3@CB_Oj1Rdd=P+wD$(I
z?+=2M`F&RE`{Cyd_}o>fp^bAWo((kBN-}>FA|KEm^1O31Jk_1er0OITBJqt}tMQPm
zomk$oFgYt%zq)U=`&PT}u=@_~u}99rkSMb;mg4CvOQy<di}M$>FJ2JV7!Wyg4J5RP
zprM5U=LwU93#b1eBWH1mpl2`j@~@p(NzSzsC$}iK;mXYwOJHVfz_W6PTmE5!=rK}=
zN&ke&uUWMxX7IcAva2RC5pl`oBJVS-@bA^nDjH5XZ!&Qx(0HawTf-WjcwD^-%aA3x
zLf~jTRa`6}`)XAw%$SNK@^a{GVwk^LJ@P>DWpAg~4-|y?w97t0KsVxVf22s<TLk*r
z{v!uxRkz(<{K9g<m{txHPdw^;@T%_wxTgBf-xt5IvU=_Bi_ctIz4rFvGgsjX{gsve
z<biQOe`(M}@tJD{T`TC?>a|S<Jy3Y&20>pH^wsLM9~CNvErn-p67(%W->QCK)-a@;
z16)txdXVd>Tzk0=&vk;1m0RUe@m^Qq-;S01ol`SRpS#Iq<!A{K32jk#Vc&R7&MbaO
z@(oIv8j<TfuF=F9zx$7)*=q2Z8k3e>NTUs%@fAWEg5PxwP_Mt`w^;Jt)l1N%XYXJ+
zQnGh_Z+B-(z7UyYxfb`kuZ^J7>jlYEE-9b;mi>0x@B8+<)qZ!{Z;$=%=QnRI=37=i
z+v@NVTv=PF;ufp8zKvl_{_wmOVYQevx2A3f1(&KxJ=`WC^k<5~SUH?*!Wuyy-(QtG
zMr6q-TpLK-NaEESXJ-dPEeD+J^1!U@FKrFXRW&_1hS#~g_R!QfV9YqBdd=*tO8UDj
zXRVK(JYj8Ca!+!V{0?)@g9nVcp}lR;q?ePQt=L4Cvs!w*@8OOTk-ThX56q_5(EKeP
zLPJb!Vxg+&$;1CFTleh^)ep?hnzTB3a>WJ#4lR42;YUjKEPgI+`S|G_Tk)Qk-oQdG
zg5&<PbTXRblNH;$t^~brJRVkxJQ>$%_%71asTD8k=cV?t)4!{EX42aBu~+ABFg7_m
zlEI|J9<`o+xADjQ#*TEawKV<eUKViu-Yx#Pe;|Jpzo~~H{%icu<yQnJY)uwToJ#6N
z6Gtg^#}5VOJRrccaz}amaNR%Shi#`AKWyvA4^_{bp@KfuS|Mvrp|IhG1KBe^F|xJm
z??xnGq{d^8U-vBx2=Q;?(=65b8EEt6bKY_U%kOdJW4B{10@j%c+v#BMXTo+l*w#!~
zkAqn_vb9+*y}b@*(a2J;{SIcq$WpL_g5}NWOD=CLwQ0@9moCusO^6c>ZD=iBe7)1N
zT_x6_R+dEv-)(#$lJGYUTt6kvmW(tnkmq%TDd-)E;BNa@Os5n6ZmYzEM$_}<zmZL1
z*DAA4wR6HhPbQ2B!dRkac5#;kD|9oF^uZDK#_5PGYnsX@(bUrs*O=<p3IH!^mNR82
ztMY7tEhfTBHOv>wGG$S0)VB&_#4##|GaeM6{Y^b)gaFyRXMN;oZ7QlDXE`w->^fN;
zN#_{CeOzxgoQUu5cwBTx@PBAOME^5&4(0Pskwsy0=COI7h!7~os;>GFjF-r}z7-S&
zAJahMst@hB?0mTFhg2m~_d8y?U#@h&BM+zhF(R|%e_VdSYE!$)zbSREc*&}2DSzb`
zN!=@6@}5KK@KB`f6)#!pp-9^+Ub4xdZ1qq)rSBpsd&Nuk+V87E$-BVA$THY=OC;|i
zdtm#N$wKbDvFbG_doMd>FPT%Zy}d0@QXaZqayY8qu^IfyswSbg=IZ0O`;zle2T}J@
z;&0oJQTKj4>qQAzfy6>so9TP9F4UcDf|I7#DKsp;Uz%6DekAsno*&(L;4+}$^~~9?
z-flVs<{;|c7*qGu*{1H@$y)K#?VEpiVxH7J>;R_gZJU}uG4I#O9(?A9RUmlZ<c=#&
zXWQ6gx|Fl0j6-w#)<I^esTUq(s+yoNc%LPS^7epIdRi%I8v7QwV=e@9!k`QtPpX@g
z%Nvq^lwv${(mKnPZDU~!EDO7uO{PuP&&JxBTijUt=Od}l+LAevCgWxm-^kGNHcsj6
zS-iFSMv|FaY@~T3>25o6ob&C}6dE+L?hwoHJz}91qCMVZwz`-PgXL{z%9E{GN2L=6
zT|><G0S1>szqaX(ZLuk*&wj#AP{!vXHW6tnQeit{Yu-@2vMekbi<oj+z=PA*&W=9(
zGu}@Rotyr_(D4p*NVk_!df95ky<=;~qR;(g$p;S9;XvqfKXD<x%FDJo&{k%|W5<-~
z#T$y}l{dD%ddL#!UG7Kgr{6n(wvUm3)|a1|av<%szT_gPzuU-ptt00gxUkdam~W?K
z)qZ_stBCS!bAEVQ<T1AWPImb`Z5uXDh4>`DW5cdnZw@Ib(jPpdxa%*FeOkxRctD4{
zLqnmzyeRq%DFh#suKV$O`*N^%p1r}|?QcI3?d|F0J)*>uCanWFCgf;UFv9`=BuZ%2
zumn6}7Vh`~-zQZCqG6(g1e&HxNIC#DcXl4DnAMd<g<3#a`ZmtSiL(bcblBoa9^3r2
zc!I%ICX+DBA&CoY0YyqLy9GD0gn~g24GP!C)B?&lJGxN|C{w~UJ&~FnZtWym!zKdF
zI#@8=f{6?|q~k4^MB=+5!w!by2kh87o8HMYnke<*hZGB(zbxkKhz(m0h29Pvt75-Z
z6X{^Xwko9N?lm%a<m((~8%Uhw3`kBpo|;W<WdGLczD&vWx@V&4q%`=kZe)F91M4Po
zX$(h(y%h0%AVXv;SW&R&ag`fzs(cwsC11GxAkD>6$zQpSi@$+45^1duC#JAgf~}bJ
zqbj1eE|6Z*@K!?dd2iA7z4h5)tA!c_YO5A+?Hi<<iKpGL-`$+?MQ%9#s!EF_(v0AC
zUwcYeWux6X#t^y>%hScMO|P1mc}O!`CE>b4IZBM8PBbgIJDU;rQW5OA*l?shr#)t9
z1b{-SG9<PF>mjwqFu{A7T3e9Rsj)FS?8UOv0>pX&S>ob$ro+EcxsNbh?c$E#?TG=4
zos|%qu<Y*LU(X!hT0fbf!}ZR3!mNW@Jg6WEqi5NjIxIc+X`u0~O^nibPioO1mwUHp
zaI~in#(UgyK@A0C!eWKsMd7F?a4p^>*J9gaUi+!^GC?+aSDN3X*9)@6oh8#on=iyP
z;p!Vp_-_oA+-5^*cCEKgAWAHp{w?8%0HVINBjF^Ai|T<((4PF*Li?4ZKlw4q_(hM=
z>4o8v19cFgXYgYwLJ8G=u0Ovi!#hkGt?9Yz>bANCf-3Tf$vrv2grJ^505geA&8L@y
zH)2XcZ`M823o8QM%t5!gbwmheJ$Z$j-6}%hzPpR=w~spch)4sv_jN>TCNnMRW!M8G
zrVg%(`1YNXnDv<|k?d9_!oFk7_o63ukm!Iw4ld%|GytsPeFhI`76y2b#Fx)}byaS`
z&2p?2!-|!>7HC|_%Ser_lJqD+q9<yP#C+{9NS_Djd--rWGp;rR_;dS-MAW-uA-ZvK
zFm%U3yGHMj&A&L9cSrl6Axmm+Us!z%=ESMDAFMv6u8@#>z17EH)~~!BuCR<EJv;re
z_OhG`L`ToQ{%ZHp8S<7^6%epq!@gKpfm!(ES=a*k3E^Cgi1k_6vW$vfQ_<B3(L}+g
zM`A1X){Dxp-#y-5ef*;ZOGNf1dpNIYu=d+Ag;jBE)2qxm-MMaA(F3ipscErWq8oM~
zmZ*I42}a^$w&X$%#OzGdFS_jH$KK+1>SLYlV^6$g^5bu{kIj8_^5dPrDj$2leeChK
zR6hQ`p)oNFVXM8_>E7(LH$Co64-Zox+pjlMAD88^;4u=rrQmU$*eCK`&^6pmk?H(x
zBA$(20!xYdRy*w{V?M_kS84V^{W>e%tXyZSt28B4zqJz!WU(tSeKcCKclyduNu{>i
z-E9mnnmDa#jvqbvKa4`em&i6;IDL95TM3_|)d4}8M--ASpO?QZgR&1w@vW6UtU@g0
z_i*~~%jm-lnO5b7)GEy=tM)UH1o}MmJh7nOsJh;>4!&>JK?!g7gBWO(tDtl;<0uSf
z#>r}>qj?DHp!Qdpb&zp%lg$FgL9h*OHrwF){mXyc3m+v<cmZ?IFN<Gzrg|gBzrUL?
z@VK$N24NBcy1`6>82B&)E;HlbvnZTtV<2Wfpcn(!37RqnV)g@yG4NGEQ^r8dfI#0a
zJae<4o-uHnjDe3GsNN=H;A025Vhnt&mn+6V<7_epo}u!{4)~4^5&R&6s@C*%jSVW9
zMU|r^cUue8nZtKq|I93YlZ_YZyH;T(zHsX`G6X)LcB9c+01lmjUpI!pH@mNtsSN}2
z)i|)5{m(JW?ymHBRmV->8!WDa+(*VR`tm#Jl1l7<`t7ve_w9G9{qD5i9{b(TZ-Hn3
zTQ2+GF4_NB)=?>aY_ImuHTxe}DwI}pW&e}Vm0|y@H*AbYhb~KE|N9j7E}6vqB7CI0
z4Xn`df!WQXnE2enrEGJ10&{O9VYi5B-{`$8h0zvg0^Bp>OVw+x$f~55OEOJ>du0L~
zGpo8yR>dc?JY!s+@$8_NN@|(Za+^N0Urx@zmgB5|djbvLkSO{a$|HN-zsNJHJg>-_
zv__`K!<RhJphziQU*qS$k{-4@FTH`~`%jqEzbsf^V+^j?+f~aaZawwTo@y`qY{dq6
zeg$^!lPliP&zrLUX&zu<2>ah=#|AMDrPyGK|0i$Du>a{^Qh|PTFWEr9xAnLGrT8EA
zVA@z|rWoGksSMH#uT{N}$?sf#1#r67WWmH*QZJY|rqBK-Fb9DH<{WV2x8V};J7lIb
ztNVOLdhKm!EA=>k=jNsPow%>eFC%tocsqks)H6sK-;+?$X+=k&g`mT6e5jC476DVN
zL^SA409Z{H5*y2XE*L@72`Id}&HO%07`d4Q<Pz^SEvsiISBYJWu%X4FiMN?Niqae<
z#4-j84gd2Kl_v)!ft+E(zi&RkXZ$YX-o?lywhkKcZ3+9jBff6MCC(2tPWs4{WH!pp
zmwhBt3`v=mMs7*-8%6XjzjtAK8Dhx2t<`N3L+<UhYxG`@*!;<o+C>Yc|4v=h%IQjo
zAB#Aw7crz@QMe(ALVRHQ6YXs|GA$#0B#SCGbmwOFr^_CFc!MvwFs%=7P{|8Xil+n;
zr!Z{HG&!7N$)1{bPkZ$bA1%0#E$=Mq)@_jw83crINOgwJIH9Y@<cmyTVD4Kf`fp>K
zz&%@OqTu9vcD64)@s{#Gz16<-_(#kCr1OJ=mG>Yk{P32_KWUp5nEM@yeNb?Zd!=1k
zyz;wO9LuY;vh{BApZ2>%awKCE{D~$&uK1uX-Zk0EG^D8jou(PR<BSGEz=jjY-NXf7
z;wUc2kACG0%vG!8A=5mhGL1ONA^zAQzMK!@rK3Q+(;=RoMjYu7$2!DEg!ls?&UA>M
zNF!!D#N!>}uZ1{8h+a&-jh{`G@y$I}#;sf^W1SFxBt-m~RmLT0#D^W?a)&rYh(8vh
z7nARXG-ARb{?;LWTZmJI=*8r_G>tgfA=Wv>^MrVr5WSduUrQsN<q)F|@ykM-CPXhL
zUtSvV7>9VWLp&)L#GeUqqNyB1x<8YubyttodM_7hJ@gn5X9{tILtL6hY;}lh9AcMJ
zyi$l>Ouk#vh`)A-iyY#!LcB_dUQE7eX~ds8#A_U4H;$r@K7_+~Dkk4I(um_6;<vmM
zh4@z?dNKJvmqsjbh^INkdxZF~5WSduA03~n^^V<EYrjL>Cd4H|^kVY;EseO!A->9m
z=6OSij|$O?$@jZ7;++ohL5KLP5FZnw<D5gTN+Zs6h;fHlEyO2<=*8r_AdPswL#%R$
zXAAKuA$s8xiqnW+bckm<#IFWGOoC|P69%QRvK&^f!}?qhtY?I!@CmQwry6DhavuY=
zn+x?FEUe|iQuu_2(^wBVtfa&0RE{gaB7DLrz59AQkK4D8=@<FBpTB=6oV<BZgW~ah
zkgFken>o=$n|~UNmi!n`?IVv7^r!jZk{{!!UBnY7p%Hx#zItw9plOU=JLmmLc6*il
z_Gj9ydE)<yTSf5$`jFlJT<+m(xs8_mi1fGV4xh#5b8s)mXC!J)=>B7s?3MWx>m+W?
zn<QqQ<70S?l}w>5-?lrI@ha{j4L=w5&D@3?ERTn|jW(Y<I?}L*JpgO$>7r5&_zPV3
zo*uA#Eu!}FwJ(-NEj>a8AXz8~D@NQnoS+Q>s|8oWD+I24i7(xL?LgzApuWZxFvM1>
z=5TSe<{SP{^TpXrt{)MP?(YOX%T?gBz44`Z<FEFu4wbAR_E{UNU5Tmsw)vWTSx)k`
zc!aae<YQz-?2uD;a9q4w{?;*P9;}Gf;FZaS4DWm^x1A3&BbJ+4==?RSRGSr6fi?v_
zqDZB-#o4)*arkkR(^Ie^tfnHNcpG6RWcgdclWoT#Y}rC3HEmpc4M!i%50^a8B`1;?
zGmpK?fg3S`%ibqhJ5q8GE2Ejhxc=w)iQ0t)vBc!jM@JGlqm`({cda-TvTk>NY2e1?
zUO|=sFFgsByma2IA~suk;Rc%<RBn9`j5aT@`<k`RCNq;S<bsab5x4S2)8xX#*!g{%
znT7O<bS$#f<{m+V`e0wf=72(i@m#NzhrJ?`=S=HwJ(JKvu90D^&auMaeKJ@jD3vSs
z=k+a(<?SaJwp{Jzj5~C6xOpbEban{tOLaemG*UY>Oc+CAGp*Ecsc{OiA_Gw}ltep6
zZZ6RdgSkY*8Mvi(kKijKMmCDk5*4piwXkJN7rs)})KPbAb9d%UQDz0$R_D=P8gz>T
zi~tZxlvW9QHA#uy=UI|tu`oHKr|p&o$pAe<N*q^tx%`<Cj8gVnZDpB9Vrge)3*ea6
z2c^>8ugvi)A$X2D)lSdk8fct+6i=8ukKAf5%g37e^X2d@;oS$j1I`{GgXqg6*Y4gW
zO3lXHE+Sh(Sr})3SMyuqvPkl#Eqr8``794p8d%L`wYXW;xt}kHYjDl*u5M=04EdX+
zKROgIFu&a<+f+|_4v34Lqci)OC0wgNwT7kAu-O#Y6K*9=EFrakTvJOC<T^Y}*jm*R
zjYMR7|G?lPA0E$1)Mm34_T_M*ykvweAE@Q5+kC_#!~t=Y1bJ#SA)JVom;+*%6|2(+
z$lr=*Cj%wM9kY<&@0r}23sv+cqb12e<NIW%wFa0@75T5RpbWoOduppT1sE7$%DM2R
z`?LKn;@zFUT9>KkdGVzT)ZsuKF5aDiwmOhqMxF2COYx%UJPR-7+__)rAMcKpSo=%o
zr{WwM1aF*HW<HN#u6S=c>~;m!qJL|n&i?Itnbka-@5ZyRc^Lu9f;qp??Gz#+Ow9R3
zDng#3ah7z--w>0oMELcj!$Q@=NXy-MX6Vg;YPDCBQE6;g7dG3UzJzb|kF<^(w%qDz
z#Vj%j{b;XcI9K<VaZiqOm7~4euXlOw9rvuJ%}vG3v-bzxd%^5`o(Y?m`riMe^yhx6
zmHD;4h%_YqG5ot-yt1t<X5n6{@m8;Hr-!Gd7F`Knx8Nt)=wMB}A(&2K;KtTZgo=jN
z9?cHutNQdg?a0Sx0VFpbllzG<%1`vVLVnAyE9JNHx~k;<3jehPER0@TNTjgxYfFh4
zR(UPpTw&)LcCKOPf~}=Wd3`-7w_IAikmO+pd->HFi%T_<6}*p7J)4HsYR7LMJ?vHf
zdzBkxeVt5%y{qcF2tt6Z_fZlUKtEEILf0xDMlgD9r+vtME}-Q0iqN%w_n-(Fps}+X
zn4KxigM)1~SdPO}v<&ce8Z78wrj3QJ?J<}gG|Yp;W5tm4@*SSyX9QJjKm`Kfeq!z~
zZGon5LCTP?dTODXVty~bBU>Xx`U%W^f_rsBkXCAO37bPtYgW4gvxc^3IiPv;q<}j;
zpt<y<fV(`P`ShfKJsvQ1#-OFS*8`@`7&PF1512Y*(0~UG=)V?&rcNjX<_{r+nOh&5
zG^rS&L39YUCmZ0-!yI%Swmlo4N&3_LcldTQ-<to9JPY3OD7V$x<5t))RGK{_Fmh$r
zOdeA9pOLMAgMqo_G>(|s%z^#C>FQCT@%bzhj*5;C>XJ9UedC_lYj9SNWhM9g^p*C_
zdp@-Wm%gh@SQ6jZUKz{s8G^eX<h|y@^6}XYHaO$G|Ip8eWV}AYz2>{}Sk_RzSlKV*
zmFX<Wf2C!-G9zPqO*t*cw&b>U?k&eQD{!mNfh<SwwdF|2&~5D@a!l^MCNw_CZ{B$N
zB&hw7b{>~TT|F(!sQbFL95NDv(St&Cc|grxXYH4*hRd;r<Akt1pIT+vyYl{BfaiVf
zZF^?7wlCZBsZ`MfNa|hEhx=*K53@aA$vr=93nz}s<2kvf-G70PhoRoX<#;8zb!C9t
zr#Or&3BP;P+dObUU%S7|&(q7Pr^pbVR;HeQcmz+6d!PMqC@k*JJc#mIsSS<bS0l*Z
z6kSGyNp%;h5&9;~6nm6&NGES<iw4kB;1U0Mu=R0Sm&X0)5p^R{vXb-C)3f{{YfNuL
z6f5tV@E01)nXY9Al=bdr!I0P{m~#uu>b4q>6il?t!2B~!_((;=jhk33i6gA9BxaQA
zM#4fZ!UA%_2z$oPw<?^NTH<FzP+7`<J0%ae`8_e_2J=%e{!?Ocjh=8z!oGuMC_E5l
z`ml}8nt8`H_0GQ9_M3Kkv2O?ZZ$KEwPM+=JP0re`IsSDAa`7f-4HD=@2XgTyXC0%b
zQM2{J;@?i5b&?lvayfcs@4t#SIZpASzvRvR;igw><>s%I?Ikoq3%d%&oFjgi^BDg6
zvtn>3rr*c<CVaxSJzMumlo-f`>i^YzG+1A&k!PIP^*3*hKlEi_?)|(Y{aG<%<t~FL
zu4JhU?41jGYnexx8knIUOU>{0-q@vssesX478chEXzz)Ji+td|XBx@;-p%lB%<#_z
znr6@fU7w>wMxg2Wk+YGsR^CFju&)wD0bBBj!s%ftI4rwc7$hRR;$0|`fyyQHo6sbI
zF&t<(2RO<cF!q${oy^B;!uXyy=%gV$g5vL`W@tr{9$|CzxO7s9QFv8KluV4dP7ARd
zX3HLsbou1z<G+F53jch5bNx4XvEAO0kZZH}${jL$E9GItxOS^sEYHqWJPG++Tq}o2
zsUz%El07zb|6BY<@7L@edXW2EL73ei1j0SP;Mm;$pklY!JrJAOALJJZzd&qWe~@1w
z`~tCA{Xu?#@C(G|^auF`B99>d{bcKZkZ1q>1cQOQYzs7Le_5B0dPIrMOKGKHv6`;{
zCpbRvkhX6lK85Bjlt4AFRRYz#4t@*fNuHW7d1}7NQ}a!pnlE|EuyYMN*RXR7<`IVj
zd1{s9snsIt4syW#@It7g>*w2P%{lFz^j-hm?}{XPsxzcs#?ffbD*ThT%-A%W3&y6e
z%2N28tiqnfb$l8z<Pg8<5E}=BSRq8u;(9_F@v{zbxI_Gb5GM-Jv$*aYlFEAfX3Ki8
zLp(=_lZ5D5T%S!Nu5^eSxKPH^gjgv=&*FMl8gZdRe83?V3h_Ha^enDdrV)SU5a&6>
zJRyEph@QoDLK^Wrhj@uY93;d`gy>mZ|1FL91&8=mhq!AHh*d)LEUpLrsao%6n;X`>
zqa5N}LaY{|+u$(d#WdnNhxis3YW<85CkxTDxGqj3{>dRe?hqFV@%uvbEUpb{#5oS}
zMu+$dA^t#!p2c-i8nMD5{?H*#72*^jdKTA_X~Zu(#8QWNo)CX1M9rjFE3?yxM>xb!
zJH!z}{E-kni|d<%Q_b`GMr)oAxKQgY*v=pQi4Z-D>%(cpha6(kA+BX<;nAOhXcpI&
zG}e5Fb%(=xL|B&zOBUB3rLm?stScSXZNi!+ELmJbX{@mh>mrBs$)l;m8nCdqF6qMJ
z`n$JfaUEc9Ww?c~UtF*D;JS!w!~exdy|++8$az7c!(k7+FuyEuj-oYyvQwxvuS80)
zW8G$aBoa9Pl_;V`T+W1WNTi6Wz`UDC!HdT6pxs_2OY3Ugq8=+FFE+<JfTAVmU^pFe
zV&BWCMdBy`;tZ{4**kKGl$2s^{U(nRi3<Hp4ld@$j{!xR$FuOIAS(?*z|Nwcjx?WF
z8ZKE8xb7wcBGkyh+QA#yh$?(woONC}aNTcs?H%fJ^XGY4bB({LweB2bjzH6O03^5x
zJ4|d!LQxhY`ION@kFFpk?k_0hXLY!3(;%X3!{BJ!J_gs7_tripk#7}7o3FqYLoq2F
z;mt^Fe5vlu+n#8ey4MaewUUZRj2dt~!)bFE-0QvF>Mp#;XmC?PA`wWanrDhr)Dopd
z9>6URv5)5SvpMHO4ImQ~hT)}Ub|OZsIf9KDoybd14Hs+vA;BW?Tcgt4kJf#ayD!zf
zHn^s~%jbUgo_Mk>-p(N+?}mJDhf0#zmpj`nNJuahBt+)9DQZMaWfQn+6G3?T(Fzz+
zR`UL=!mblON3@V#o%^tC@J5DNq46$|7{n{BMQf>c*9qj!S+2f)aw~1?C=~sbhHJ3L
zI*-h7Vgii;S>9B{5`|1%8$$8P3iC8suCSXMHg|;NOUmzR&&|{X<F>ntTB-XWl8V&4
zAOlwg6zvb!V25QzAXc+i$2b(m;(H>)0Ie~Sr+@ec7ewV=uUmwds&HbOFH*9d0Yae1
zt(?sLOWv?lUG!VYHdELG(Ih}**xxW~OXf4HAV;oHSmL2_rLlr)<ix1?z_#jHo86e^
zkNWHyJXxm`xfgc?+-dO{h}Lj+GckSCx}1kG>?JcIhuijiZe0A}d5PhFRc7k9B_zSJ
z`t9~m#RFSBhL!3~Jr62~<b&9-1L56k-L?Q4vDAnaNA~dU7o)xd(KyCG<T`!8bp)A4
z8s6^Ey+k294>J@$(D3$9p^fI_Q2fP)w=tf}UKM{XGHi=Tp#|JXd|O1ZQ4q1J%vZN9
zdEkYkcC4j-2d}xlsGe^D3>mb}0F6aemYCdtqw!tgVOU*h6=3%Cfu5v}gyIJq-ub*d
z=Uvx4RJ1~697Z@~*AG}Ye_JK&X;ug4j}*p~jGXtL$fgH)q@le^j;>BRRqYdb_flqD
zA4lD)k7t5}8|1B|#E{}a12mh`gm=RsuXh$x&sf5@3HNRE)~H@CvX8I7bs<nu7f!m-
zTPyV@U!6#UOmQM|;J#IYZUu_(RsIQAJwGTb(xx+xv8|H`u6tj&6j)y*1P(|rYQrRG
zzSt_>JTOHeVKqeJTP)YGrlcYx1gg^{&|;9!`;=ZxG33QX#ndGd?_@*zX_OIzMrv`#
zOtvKd31Mehc^@jOCmCF^&J=73Ftxu8t3-~GpK4JI9NNFV3S8B?SP#xMIc2<NNrh|$
zcP`4Po+@V|=U(>_yQD0`>zx5Skz?(u#5)8hvlLz-&&ip?*v>?P_`q<kle@XCXdyT1
z+l7>)yD+ui6uw)byfEaQg;>w$fyOuyjAR3(ois$l@t49!u90CsO#2JFHXj2&L4jB&
zIIh3`!@(T!ZM)rxoY9)~oRg&$wfu1WMWTWd-#-?AU9AdleZyEj`6-?-qy3c#CnhCu
z6(f(EEms}t<b~rg?Mce~XqA;l%%|V7)#}qo0ldXFqmYc7P9V`C;8A{F1HImD*cD!}
zQXoAo3tVqJotuYjTNrG_Xq<1L%pq0E2)9e;6h|wc;w`4#48pmI=r9oAgqEIj)~5$Q
z5hWHT;MIU*J{At~b2W@wl{q@XzFFaYqu!GDq;rDr07?|1AvUSC=oJIWnnESjU<mu*
zMHbU&k-diX^Nku%T28jRe?~3X%02bIt*DizjB*S~762AF--sf{Y~?<jD5EHi&e56j
zT0RLs)D1@`aSYGQFcZ{Lh22-A^(8&y-wthf^EpSWj1K!eiwakjJL<dJtx0iJT6(th
zd2a>hs*aKQnI@iKM<%7FKFcxCScCCA+I$(3N23jnNMdRe4f2rj>QKo`v+F$L^#~Fw
zoPM$Jrm51tPN(MeP;vBA9AVM#^ee}(O7uG%anl-Ghd!I4<{*}hQX1&5?gB!~U%gd3
z<+s|YS>MnHeAq!2sf8xGYKp4b?94?XFbwEg8<_?edd#X)=|wtGg`9rE^R!$Z_AOVq
zg%-V_G4gc~m96-VV7#KdcMoHPW$N#_YzmjsE_s0kb#0~$1{(ju_putDZnH%cx_U#?
ztFyHY;xoxTAGRj9eV9H#Nils>_FLlp1Si)k-cN99z2f}@YwH#7Cpf0Q{jTpUsU+0z
zUEf?%PN?6zVoRcg`n_x15*O+>!d|Dg%%e%Vhazt)FKoF34owP`><A>vP|795YI2mj
zY1w%R|K|_ckJX0hsJ?yE<&R=#=xvSA=J=C)LI$dJDIc@a@DZwY_ovkd_3DGZ7q9U`
zEm9E~_Kc*C{8++YA4?Q8Xbh-DU50(_Nq*4=Tc0Rz=xkql<t@>A>|{TDG+OT#78>4f
zUwZj1p?X_Vh&J#}C}x2QHQ+#Y`7Qo>nT(<hJ?@p3G=k*~IF4O@OL@H(GlKpGQY4KW
zx8U#${$Qghr|i3kcm6ErUtVmaJHXzVTs+ZP9Q3PutQ6QFTJ0WN1@|4?`>|N$vWK>z
z4HMoT`Ina?!$D+{Y}5{gB0fp(5}q|*i%=v**klMj6O%|S32L@v&eFo^lbvscuTvwE
z1MTa1pXHTj@kCib7EhD~Wbs7VOa7^01iOn)e<IxxS+#|k^qhQfA|cmh&0%z_zgz3x
zuYbZcG$x`iLDb5<O5bCvU7?<q427S$)MOK%TE!-_HGF(sBiAMmUUt2SOx)aL!jxLH
zufN%^bi$j<5v{housQkB3$J1fop{<$9pEDXTLk!8Qh=YlupB^d_4bd7&up#UzNPrg
z6>=N*qEKvq+n}!(pSfC4&+&Zwiw1qJ_{{Z!dXDGYlLmdV_{<JLJ;(FQaXhbFaG?5f
z9M3Bk9OQZ`*Iur}a|;$||E$V#!b+9pqZgi~#Oqj@2{esnAgRt-Y8_%ws6pz!<gMq-
z9M&;T@%?S$rnMyS44qLt5(o2P9o;iw-O-XYSFgtiKYPdWkmLx$OZ4|IUjmRQJ3|;N
z`lw&~CHGzunuCpxU+jAPV$tIl`yId7?D)k_$1k?IOZ;;j!7s~*FENkuXo^&PiIuV+
zUIN)VJHGtOrEjC|{L(s}*}|})KqnFLVJ0I>uD|%Lk!v|`Vf=)nu}_@v^$8pZ_Kp7K
zT~*1tTyD-5R!puuVGH~z`H>bvMqRFfIs7llwUa(dzFP4P6lmd7@KWC!NPLm(lY4`;
z3wBymzL{ToXZjV%6&LRUtF}d!$=U@yNuV=|lOHGj?uwWAa%zh#n6(S`r_-!XrzuP3
zs0B(g&~Ukommz+{A4s3o!rYu)R{mWo|E|9BkNQh;)ue-*l=b`}C7a*%J(6hoSl@51
zc+<YmEJe4BmrwDd=Ik3k1Dl_`hm1}_p!+J8^1qD#_T;|cQ3ZnS_%0{9bH3ju4vTV(
z?i(7&Y@}h1_ULsdJY*s4L%+Hgf77q-#ozS1b7G!2OkN*<6X%zj8&1?VWf`*zBpRR`
zCEt&KDKFz+ZgH~3(p6592X;=JY^e`d>aG|9v#rgt9&Y)`u#4i0VK>daGTUSC#2hV!
z=eXIbwGM?0Nsl2oyCfX{5Yq_a;}WQt(~vrQ*Y0C~uf=U6f}50>JD1W-vw}`I=f;Wa
z%#MNigZRj?Jqfi{8W7BbGi_j?L}(02KDPnF)@s(v>;qW7K(vnp(?jG36C7?bGf7O2
zEx4w=6vJfR_#Y%J=R(PhtGSP8hqjCO6>GlH%^Yq`F4p|89^2<$o$vpg>(-j{69t15
zxvYir0fN!LAT(UPDEffJid<a`hzKE|gbaYK4v0V@phOOU9RlY3MX;O)1j>1U5M}IP
zJ2?FhA$-mJAx?z|o5Ul#MqwbKcs$yA=XH8uu9%?_9FzXPQzUJ&l1}H%U&^4I@^PvB
zi<M#WUo6+*<XX3={6UIdx2W<#3R|~m@`DtwZqe`uRTRDAqE}q>ip%f^UHNv_{XwPh
zrnrm(Ha0VaBEIcud20^x&8_WsUHQl8?O9rY;X+JpHV)Rk)pbc~xb=-UYhU&TIfX7T
zf3Zj85*9?)(05nbWvGiq!i^k|NdQ96mK3HZ%|D0FH1n6*v$y+zmRShA|M=b?oV<YX
z7@WF*fdIE+3<k$6&|)DQ%Gx&}3$`$zkp){i+BYE!wzxPCd#5KeWkFM|Ih&C%{TH`-
zrIE6YMDlTpbT|nTf#&M(mF-%J(;+4<>QVb0ae<jO0Q&d|D^V?_k{=N#@Ly@=bbx2^
zPcgoLkDBBMz8I?B%AYZ+>v7irY5Bn5UB3tFv64*uz+AH&(~g=Vy%CtNCiT6Sl8?xr
z{a1T~<#RjRHz6v{d8>UB(xhg~U}*09l<XFpG(p?y5Nz#h?oRhclS_1Nk9(u}qI~Xt
zy(!m(8uZU4v0L1<Ds^OmP4zm*tf80lJ8&Jk#@KZq{n$`YRh)FmMj6())W62)g2w2A
z@W1yf*CgMnd?oo><;LU-m0ihy%n!BV{-`2!I)3;?%P`%3Dn9bu_R1A7Mea;*T3**l
z`KA|}yzcsoGn4}j_hze^VQvl<DOnb1xCNLNIoGBaIXMwlZb+X``Kur)GIvk(H;{vq
z8s&oYj$6EZ*m}+xl)Mgdef6F<->4VVKT-C_?VV_{3@44SUdcTW?!BI}x3*q6mtVf3
zpTOZ49Iwv|LYDzi$A#w7W8m-$M7<Z9OSb{x7l`XZ0RA~V`%}B`+<Ik7>ecG)%s}JG
zbm~76eK*6;KAz~?ViM$sPcX}5%Db~2pika?QGov^<lS0N-mUfI-C9rH9n~lAzM`^t
z@@|Bn@%`l8Xo>R)4K%$0(OlZ%<{!)yDGDDz|DL|$)RX_B{uD^e?c)Qc5`8a~kBJl}
zpE>>PzoZX6@9jr3dE%r2<EOtqq(T0^hDe2xgT-Dk6GXd&9~jX*!3lx_Ppr3FcA8;j
zP`MQ+*W0p90v#G-e6WHm(JZ^a;KqZUuj!c#+!33;H=`Wy9u71vAU$b`9V8ZO{*BG~
zcU_&2^#Q}e_s&m@`Xn3BY!?3mlP0tHoi;)0G6x~x4@{u~+H|Rl0hv$*w8>K!fH{A#
z`TkDLr#W{<8rlmZC9h5Y12a87>642p0*Q-ll6Tftaq#t&2o3PJT#Ua_mcf)o#@AK~
zO^c@akuoT!d|WCi5tqq-uv~|eEA#zb6rK70E(*(he;37LzQ0RF(JL-`#YL~U48Kd|
zi*e!>^Zi3ER22nUL9VnFg-scNIa|sm{(S@Jv3(BnKcdE!f6RGU(|^vx#3qfE?5umI
z>vBF|tF832K%xlqmvpobosNdZ_HE;T>gZOwFr0`6t88B(tAf_?FX_P5+dLx_CLswv
z3n><-{$#WVd-RF!7xm%pvB~D;Xvxcg#w$o6=Jo=pLD@P=7yneAN~2@GHW&QV+wV^c
zMi<a%^dc9UL>B|ngaTS~>0&?{R6uJyT>x@6A~2nhNPjO-5Yz}EGyi~%NCl$1b$+c#
zwVu*vboU)RT6gQ+B9+*Rrg~bhTl*-3a#x~CSBA+h*Wu(!fB%Z2)8D_Mu=Mw@C?5U&
zD-}hrxabuZz2Y+bS1w5F3+ivCeYJF7e^c62E4%)VBB9?8PA2`)-}9`$e_{Ro3ncIu
zDuvhIi26#nQ&nXls+FsWAc6TezUSO<biytc4dslP(8>P~c_!TQ5mj=Xo`wOLSh$?A
z#C-cqm`F>$&S(h-!kwoy>;z=g0;S-4KIRSwB*;;|f3)MC)dXBf%cpJh?~@B@=S5MA
z^TDt7fJUEebugsD+~{Mi4%Xpd$cVr?987(*2#>G*xjD9ZUM^*SG!@P4P=Ulmx>u82
zT4bM>>R5xC9Nn)x<&=jaR8j;klb^JVrannVbSTB+Qb~!p5ROEb;mVyi{PKcd;QS7^
z@<dAZ)%`Kbj$M;qTw^!tykkc{ZS>dUh+luqS;8Ehqo5WM4h4R<OXE3{{Y&!3+_axq
zmSlv~Nk&22^Bg4F!~TgO#OR$983@VuCzc~dzcsPI4i^)4u3_8W=>1+;2*u2MFfCRE
zn)b5!xT{k~Hw|1{*SWfQLj3!t!H|r$l8%NU;KLH(f|G`C3;q&g&%{Z?b9Tv1BL*B|
zp+o#FGw7rH&H&MKcIn)hO7WeiEydkjNbxNp9uT7E?DBLP@kxjHcZaw}hzEt}IlKHS
zjd+to{JlebN{GEe^qgIOl1BWoL!9LhzbV8)Uj@-~b_t~s$2!D|9O74nI9P~oC=K~+
z8u56Cc#1<j?@bW>LVViB+>q@ZsakJ+%4+R%h+h`skwW~HLtL3gT<#FpbD?=&ehI`J
z5Y5@;cWJCU9oB;m>*<%lI!ah_cDXu@^>c@Hy~Da&SVs#>&Mp_HvA*fBE_GOOVFkd#
z+2z@1aCUik={^nNf%^-MU9sB6LRoE5sd>6I9+BEYXNR%~qt9!@Q5F`I=F$qS{>*G@
zoy}Hk1C$Gp55R1>+3dI~6l6pg(8Gq43YJ+R(C-WsZJvNf?b_MD)Ke*^*x*0H8ASGN
zqqkNpgH6rMN@<J=Z?v6-y}#1nEYkabsVj?y`g#PGi50eI3>YjCpN_3uC==cggTig^
z5Rc#a`jW`Ak@!mRpO3U{WX<?hZLX91%A4V~&7f`sm6<KVgB*FZ7SA(I5?W(dDlN3i
zqgCYdbdaC2r?gnhLE?xQWlst`a*NzJhzNx-LW_(ez}IFiTErRI)iTqxnlnXSLsCGx
zr?~z}1yW)^kZQ$#ki~A_moKU5c8X*=nQm?E=x#IZ5?N1Dg=D%2?u^_nGKA+Pfh;sA
zC!tQ2(r}7+6pj>H#On}}9Vp?vHR7GFwRp9fS8`LR*EMk8R2qk;QmchaQtm323s+^V
zbz@mp)H+r?@Wr-i2}&L8o|c+FR6iSms$XiwVX=9#Mtq!l;aWqr>``)rb!Ym{B4<J%
zsFvAwyIZvka?3%1s%5@Gcd3>{Fj(Sh`B8ZALsm<*da;lL_q`rwnP<Hfl)bcqA^NJi
zSW9l^Jqx4ul2u~8q!w1#bQ?&lkjsEFR2x_s;`?-hh)SkrP+`>!R{ppx@NOkd_C?&F
zq%~T?gO{ujXqSwBpLklnJv5R{f0}PFU}+l}Z7@M;X))EKAsVj&P{jA5(qZX&v77is
z9Cw>&N2o(J(=@|M2Z!*&Db^eGB`v3p2sBEb)Y5(6zHMp+t8ik>5xOr`zO1IO;y;>O
z)mYz(SZCU^r#DdDDF%1t&|<dAr|PTdfy7ao%S9uC(jCRRTG0ddwaH#tXy|U4S@G+t
zV)28q_=jxRQm@_5+wLFtlTQbGiS=tojz(5fKK<n3n^N1zSVi@{9`n5x^F5<D87xaz
zMIo08QK7O4m?bV6Uk-aXpf*?pP)0P`U?I0xH>upE>E&FOyu9P%BRa3Vn1jWm@eO>;
znMJsk@CCw@1n^-mhv)Rx)7^oFXVvLAK)e**{StOn33e?YDz^41Y}YM>r42l<02_h-
z*O)UDBZnwEy(@`l(Qpv+6x_t9CNTGNYGWS}9-~E7EwU-Jm<?iqtQaA!MsU`k*(|ia
zz`a>3TFm0GKz0g09|1N4MT_6=%{0-HZTB)tgj#aA_e%#S9`IVugZLxFgR@IXI5DcY
zUpy}4ou^FRdCKISr#g^&3Z6ow33=ioU}P`AxdooQwWp4=lnHLTD$r;y(FzI1>~`fV
zsfqVzkJ!XJj|=l@DIUm=)+4|$yKQ0cbgq9~uAEN)$k;#M@cvAj((XhR&N#(;X8-w&
zGt%X$(mWoo3O|$TczjXdyWAA&hJBjVjLhjwN@ER>AFlaPIqqv=?;M{`=w7>7yNR5*
z)ooYpbn*AXN4{rAdw;^)?Yg@A2)&PER%dovntJ_d?{x@AOk~ddtZP2gRZgF}w9?SP
z{I3Hv)4z^qR)os=r*Y#^VXWDBN;JM38r&aldv_2nZn9Hmh1-gX2F2o=b;M{W{th7?
zoo`2oD0zu*RC!JooDo@uTLHsjWUD;-@Rr2>(0%3r{ZxY>ETbI)`Up|sgov+4-fg83
zU$4&VhX4Lr{TH~eM=2m$q;~x(@o??uyxNv?m~y<0V~6c8t&gYn`xchr0vNbY&Z%Sq
z#}vCGfozqn)S`0YBq&40=aoSEd)X^|T`}K>v6AiRH?&Wb4)*%$7~NZ61yb~ax1+KQ
z!R*BU>2PILXt$_M9KW`|aQg6Q{6kuGul7qY>hrLE+_zelFJ^Q{*@K=M;BVLfkHxn}
z;x9#ZzZCPmqs?_+3O5(%6NYG|rqbOR<<<E64;$aV))wsE=j*ma7~iq@Gl97y)VqQE
zph}8V<>Uh!jB-cF_5a%LwKXy~^lQc~qf~n$d0^m*3%CbHuGoNk2-<^u6+p|g2e~Tn
zkk11<{p`NLjq?|Xcej|n%xqute<TWFs{N$j+`3Tu&8-aVVEL^K>|nvI8t=CN&W+vz
zIJf*3z_|st=zyO{$*UUg8!Q%LjO1-k592-lVrsl!_ZJ)QH*;aUPv~I0KjFsv0fu^K
z=Oh1MyyrJuHQ{t3PUJ!K@pa8(%W-}_GH{*$Q<4&Qu=(r_mczm|*vEY<dOOM*&cd}j
z$jemBh%fVDr1|VZ`ugmA{S>naWB&05oe|Wc+fjCZE;H{%c+TFVSv=DGxGrV!t&nVd
zD<KLv&R4u7LNJ8HB~`IRT@}J1Dr5eT`G3GCk(;BM?B?qH<7ROaiSO5z{p{|~T-p*L
zMxr8q6U6n!i0^N)lIH`B3Y$^Z&{H`5i;Pu@1x632?BK%la+KZ8*^yyOBXjoFQ;0yr
z>->(~65l$kGZKF<vU_ce8Hhj(wJCo-IyhLcIN4GDin_Y;Uc`%JxOYX}yUb!kr_Ztd
ztM-1q_g)`VF81?<2QmMZ%Ej4z5c6TFT%6Mf(RN(E3id&Q4wBag(N<h=^7|n94pPtu
zfe@2SIGur}$&7`rH=&jR{yE9_%PSie@;kW!MY?=yLkGX38ZZRpR$eQ!K<?ygWfsVt
zdacX?x$J&wXoUus)!|M2dbnXs5j60esBkVnjwHTFC#OiFKw(i1A&Dal>PTWQky#={
zGWpje0W0rM6G;yBe)(>!+4wJiTUT!v9~P4QFFXZ6H>Z_xV<a-Xs5}2~^f8kXh(2Z<
zGjdDUClM$z_{(7^CJ)M7I&w>jWahpC$z+b_+Itwd!CCj<4;KvGgFjp_=bq|qlZ%(W
z;j_!TzUnnRt#%i-!1gRa7^>bDEneCUMyPn{4lZMgm+s>-s(9(MTxQ){yp!KQ7w^Y=
z<-X!}e*akfH-%Cx{z!ooi_OcWIGOBeQCfJu6hEi#s{OO|;jBNmwwKMy4gI-8mmJQt
zPtu3oZvV3NBRkp<;ohJR`_Cr>N5bIRP`~*?BsXet^|te}m#(6mv*W?9jr$#eCAa-F
zd+FZ=I5POP*1PBlM$4@9Xj#c<Aq>G~sqFgZpF`8@NvAdb$lx9?9ii1O_GjjMI6NC`
zwTo%ozVaOo&qiSF;+%hlhl5#?1y9D41dOixFd{pWuY`nBBtI;0IFdisppN8++Wp5N
z`NML-nRj|z@G`rB3tq}inhR=6Z^nFnC_kKc^#6H&c=5mBhZi5h4^xmnewc#v@xv6P
zj~}KWef%&5>EnkfNFP5;K|T&Yyx92R#l{aWHhy^V|BxR}p}SMWk#p4Hh+`LS>5*nf
z9Piove}_2CPXSgq;SR?N#r;k=n};-M9DmsPaV9x_-1+e+*yG{m$6{}No^x}um*y;X
ztl$gJ{ML>9$?1{*rsm1O+&YGj+0**)uZRaXYIK8Z!02Khsq{hp9&}JYXtoC(+z*=L
zLH+%pK@U2lA2iQ{9?=h)??H$5gZ9yn{KvZPgMST}Pph|0EnXUJIKXGzls6o7H<b;&
z?xv`LE4+I%zke$3=J%fB9sDjX-sdJ;<JVK&oLlSW+%f4n_s+gKSEmkjeTg5(fKG$1
zQZ%;c6NjU*8x4AHP@%G~<-RQbJm<tCXPK>_QXZ(-H^a?89*u}Z&__==`?elmkHz1J
z47MdojC7g&g*BI;|973=e-O{BJCTER7GS&9@7}>^5It*nQ4GNmn@o1y@4JG^)r7`H
zP+Y3CdyhO^Y2PxaEA4vjGfS(zUtw&6+1;Olp8MqzXuJpH0hNd~*NjHkdz(tc2LH}1
zN3RmY-Rhg%n;W_vA#Pcu<jvX#HLdt>OLCI<&)(_GrB1fZY;VPo%WoT$)xFLYJ};y2
zUps8!|IMIoIDgje|J%cvs-{QpXIXzbocyvR`{R$_k0amaEl$37$!d+o^7OdPlyCdJ
zSCf?^-=@DXeqN4zn}!%aFGs#jLyVu7Bj2VW#?Q+a=k-C1pO+)wrg4m)moF~pgY=EZ
ze<<IgQz75tkAr->KR6lrc7JeqL-PGg@W;6Xf1FG3$GL=4<XNC}@@}OYdz0PR8=f9}
zf9)H4PipK<PmjG}8d)wZ-;|-o$V(A<Il!G>$4o&KDswB2+PjOS8*h)4%m@dX{vAlT
zdB%uP^I)88A3$z8_pDI_?>HBDwE1id`sZRdnK3#RUxhA@A0f^fU$%D_6++NNvOb$s
zP5=^XK06pa>uh$tzNYu=PHkFeH){h;tL*(n#nH1aDh@O~!BcAJ>3ra5&H4W3YqC+u
z*}S=y&71$;d&}<LaLsCRTov-I3YBnpPHb(N7R}EGf1Fgh@%qr|ZLww>Q6YE<k?m|T
zBryS-WnNJkCxAzXwMH;a(F8=)&C3bLcfd_vi6v@3pA+@5+j3Jh@wKj#vr$vOu<2x+
zi$BBaZAALpuCQ+hCpJYfwPptzUm8L~p=%7Dkqx)%x(kWhLQ<7xX}v#0pjnH~gY)ic
zIq&9OYy<fwULy1EbU0q9pTpzyI2<qX&*5<_vX%dyf1H*7&x%hKm3a7lrUwEZ1sdf)
z`G1r5F5ppB*Z+4CCK5F`69tWm5;fWcP=ke*2xuY+p3#Y5MTJU{g0}dp7lau=<r0`g
z7{^hnw$*CY*0w*bwpyzeP)Y(B67YuiTP=vUGY%-ALb&GreD|K2Ob~7B^M9ZJ`#f(R
zm~+lP>+H+gYp=c6Wv@-l;D<-X`u|DszEv?iJTjTItG+|$n)*H?X7me~5z)ZchF(i-
z&QRDUIxR(EV?YH<h5b_4Q1dRE!ukqv844T8GSO1l5YoSn!uptZ*%a1;luinJK<`Ci
zzcx2e*w4-TG=+tEN>kV|p8t-*Hlb@Y6!rl(|Hmk7!EuG6u=?Ymu(Gc5&Hm}_CzSsG
zXL?f?r|9i;Q1Ny2cC2}qO>c!lT!!A(vHY|2b`0rXM{l|2T{gXah67F~y^Yg*(c1;)
z26{V(o4RP4-iGp&rnf({JpFGt+HBwry#@8*|0#}kM4y91Z@xbNDZOojq5c1p-kfkt
zaaV%^n#|=*X-(!bC<xb!7r3Cwtbu<mnSnMh9k1n=8NfR%;2$jD@k;?5@ErhC@mj9R
z0RGGZ&a!}g1bCVNQ}J4AGJt1UzzZzk+9d$~uK-i=T8_>D9%%vlTEOQ8_@DsiL&Y>^
z>+*DCHvh&nCf5Sa5#W^q9BTnrX8>C*;G0}%^$h~NN`Ms>@cs<oy%z9M3wZ030Df11
z-7VmZ3}DCt-e>`TBEW|Pm~vq`F9R5`fIqT;!vuJ>08=h3CuRUouz&+C;KtPeej%rr
zlncx5f23Qzlch6Vc(4UrDZsub0hn@Oc_jn*k_Ftrg;xJsfTIPNa$)&X2Jk@(xWocp
zE5JDdOu4Yk%K*-`fQu}kBftj*=xVq{^T%WWFS39?wSavD__6>~E-a^J08h4nXIQ{(
zID0PHEx=#e91q9<dMscs3;4VMPXdp|g=PJ+G;aQNx53S4TxjzD3hKv#k_*ey4Af!^
zwc0|>6Vw}mk_*cn8K~PV)B_gk5<%Sq6h^*+-=p;Z{JwqSwTv>xy+^3Hj4hGvf#^?I
zLQeFBrgL1IKRU9QC?>_BEBw*3OZ-Ddl!PwzM=vh(54or;bfM+8!82INHuHZ4{wSwa
zN8&m^z#pBijo)ATyO~Wae51$L{M~*fxvM7famOp>T!BrY*BJT*okL$PSFO*a5flHl
zLCou1K+IE*)0mvWsV4t619hK;ddxyiW*^^@5};^uuLo#y4o&W~XWI5Iq!y{_U+a&2
zVa$}?M=@IRgSBGLxIiTCkG$ZIyiqNm0$fZyYS}w!drhi+pJ&>~UVFZM#gEZiEd+6T
ze%9aZRy|_a^K`ss#LIU>E5F*}*Pc$d_?5d%i$CB(i|<)QTh6y_@&Ast47^v}SJgG%
zPCF%-Ui=XvX(~F--w#%YzM>fCG;T09=05|)K1O7RyLsnC$0Ji*R7E9MnsA9Xl1LO8
z8Esy7Dlc+4%UxMShgVcb52Gqa{R#!r5D%X>lCEjFPI1QxDMGm9-|(_-QqS9va{>px
zY^=530WbTvBj7X#4@+FY7djzLkKV>wn{P2wcNNk_n=ICi6(;e55)xjntGs~Xk0pih
z=)PQUW2L=S?DJFXZ3(w0*xNF0i@B|1t3%@;rLn)Eu|GE)remRqIaF3R<lHiEV=uiY
z#t2c&jTtoBy(W5gK_L3m;&6Lju#&($Kf`uRz}UtgOG^=}%_XWiB8>^a!|7-y29NOF
zNwJAV_v}*mIepZhj~wgyX_gbik@`&}^6n_s3cTT=JWD+{;N`>`ZWScJB{M{>&k1j%
zVCuOFrru6Ca%{E*xozoa>doHRT0W?o)T9k|1yt&_z$7BiY^seUbQojZq}OU9Nn@D~
zl#b?TMt|h3RBV|F8yZ^PIsTZ3{V9rTgKL&`@RupKe65YO&WWF&)|Kxy4qCZRBq{%s
zCO&h@iA+v71JzW{TYdELV)LR+ep&UpCA@-Sx1@tgV7uoXws938H!$<ukc!`KorliJ
z9OzhBpZ0hnAVfV~V}nNNVS>IcQACpU6vM9ldTU;f@AM5?a}yfQri(k+;+Ow6_fE`X
zsT}D0RjF7lBB(6wxWvqq9UYT5e!Kx9_Dq&<S+ShQe8~#4e@d9GGG=nbup&=`ZYyM?
zCx;68uLy<S2B$!MoIaU`nC6{lD3>|Wa~M^&s!pV^lnTw01JawOGQAfPwr~#Y`bO~t
zCwiV(1U^J^S8%#dI~q6F_OeyUN!s7P<>J+eO)N%J`qTi?lzQ9uMqfn@FRfYhfxWo)
zBC9oto9#uGYM$EdMb_y{%_cJJfUk#|IKMV{BrbqI$F+ty1&dJHUFw<d?`W#4Yz=KS
z6{4vJXf+%Q6nNx0^&oW)L_MW-v7SCv51)bKGWIUAFJkzc<8p04MQ?+S?k3$gj;*+|
z$hDvH`PA>k*K-aYz~gQ+d;9U#C3#v?@{+FdSbW1cHg;DRzGL>|zLOKbF20hg9Mf{x
zAvtFgoOC$%ix2Y{xxaTfzTt=OvqSc|uVm~EPeSk}Gxm;WooIhue1$Tbk(JrcQknG(
zbLi{aWcF;Lj8D`2-Zof1=}oi>JcxEQGX&4Y2o5dci=1lXw1JdIaE9u_CR6z)Z6<Dl
z7JGE3JnBwh;2ygV^~hY=a~`Fwpfs{5>Uldmdh>r6fu?tjZ_e!KrKr*Pc8(S~mN-#}
zDz{CsCiR!G@1*wI;v(>$@9$l$_yuffwKJACb`uQ~@z(1~OEtnvre2<1u4q}WzxDoJ
zvcHq>MPP$-f=6fQhs|1r+Gti7Hqi-Igg#}ccyGCUw{hLTM<%?dMCZYq5Gh7KosA5Q
z+%aA7A)7~j>y4mt_n3e-;*Bv|p2sLK2{9+N5!oc0J<zzXn~lgeu!A-tGv?Q(>Uf0J
z4=){KHo|#SKQ`5A|FJ!aPv|KlEYe^!QJwu^^x+T7neK6B2dTt84@*Z~xheF4nsO@M
zK~Zs{3?Yf1?QFFCOm3MAK~v&Oep34Q6i0*R@R4gvL*T-|bZ4?s21`;IOgm8>xk#9{
z1mmx4nO3AS({!8owRzczMQydhc_~mi0oOV?33tIK+S30s{`ZP`{kVLJ4Fz=8$J@{i
zW|N_Z?C7iPz}CjZF7`|sOSF$>wkI|~HMUpSo#@)#iDqs}x2LzLV}cVM$iT_$59a&-
zukBA{^`dyHhWB=#c8XahG^r<Mnu$r<o+fLe+Y-*pNjyRCWsiSRA#D_MX#UzS|IU|L
z`)3tj!Z(`Tn8fT?%w1T~8gpgWCh;bes`s1CDcE|i*aLd^Mdk}VwRd^io%#;WmX)QA
zPq8YruzFj(d+<A^qU^GHOf+aBRFH4s!gNL^PQ{C%q*!3$XYDX>mQ0Bi0HaH8x9cZS
zXBPj16nuqu=y&uTo^(Fx1}QmkM}d+p1SSYhl}|OvzV)R)vX4ExJ{EU=+q-MK<jA=;
zRIj63ONdZ*$AE1qdy=p4@pK;AEL6G0wg($u@x}Se<}uLmcwy5?JI(4`Oxmph`SC~7
z^1#ZNkq2Jo!Xo_AKOhe@5AKvS@5uoE+5$do0SBX=Sh4~@BWX_0Km{#SqlH@bXP{OJ
zN|NTN4AglR>QW1J#X~@~2uhM>-waflg{rhrTcn=oTWQ3udw+qb68X8=Eb$e8(s>(Q
zbvuhOD~j#*X(5cXwbqH&Rzxq<%=~8-a|j!+cpHz>+vtV(j@C)oSmDHusfdD-NkeZp
zzMnhO!iKYbyM&EH6E0?Ud=d5J<*Z||zzbQF)k@Y_V-$$1OyX==U^q#)7Fpv7rLZ2X
z2t@Eow2?noleS;+=`r)-Jo3D#iYhf5@#$`pc#*`Xm#z2|F?lk4{n099D4%W5J*3t}
zryx8{P<vJdY(oMEW*XvG)mxC7t_4(5lODZ|lgx)RF(*}3VMzHnP`TdQc%Dg|pi)~1
zVM<Gel0qcWifEB7?*}{wqGPqvB#P&x-WF?bcVg{xoVdVRU*N!+o-ifZOoXwq#evE<
zy^RMeg)B;(%4$m4jq6BV<(KlZukbeZ<Yisu4hvORxxw3*XEIPMqG%P$f+{4h&-g@!
zU~l7x=Hs`pQE}rZ4ou@K+&DvOfm*FQ`1Ppmc{tVJ&_A@)+mM@1bS6#p1SXwRKy!Cd
zn|7oDFAIER&NZe%-E}zkj%2>D#>q3XxxFsIq_W-d*3s3RN%dq86<~vhA!RIv6v=IS
zO=K6*EIGFwN$T~wNgq4J2;>!n`Ej89jk-v4U1Y6)``WtP^^!!i6t0bYSWOs8f8@=&
za#xA>lUNsN_lMW-O-n0vk+#~%d&p-<OvD!rl)osU!$&v}Jv;u$J9XvHQUu&DfA4HY
zIMPS~P+D8mlbta;@ACEf-dwQ@-!a>OF-~op%*}PhCkMH4Vnd$<7y`DNNE~vRKm-0N
zG0Lm~V&{UTG`lT)5@QU&Y--ju6Vy<_{@5$fmP)jd6N##+*p*03R{W9pR3Ry^2{5y{
z*^X~H-Lm3PDri<HjJ4c)8%9VY!AuMx{8c#zqUU1_RbWEK3DVAHg~7c=sMd<}F|OT-
zuK>d@+nk(~#ov(x5ASW$wP3v;z%KkhzUI>4k`K(UYkoJH-_7Rt3-jA<es}XbaEU#L
zdS@sdU7-($fGuU~!GP82`4OmFwPmCd`7DHAdo9(r$DcC*xi<@^MuRNw%lw@};25=c
zloR#CAnejQ5G(vS0@s}D{kVU48?{X$-h6R~HuVfGt=`s}D=%;-co};cnBWWl(o#K`
z(}0?1p3*#RQ2g<6+gjegHajrUDmty#(+3z?;w)|-BM3R)5wxQD^dD}HuN<?trJD8T
zLmV>HDIZXtDJ7m9zdio!54VYo_SQCTo5N8;HS|0(#kMSb$7{ycJa#`Z{Pe4PycoG}
zGWYm0a=)H?qQcLz0pW><wrwhV*3`B1{+o0VUA&^Ua<A5SW!~5=q^Kz4oIkl_>Q(7-
zwZwtFTQx#P-m>M_j18DFN1M`mt2A3JC1a{5c~(*H#|TJ|ALG`{et!(fDl~XZW<A-`
ze9qXR^O)O<hR!n*MoVbpxv_zHhWru?i14lwZ&>oKP6E2Jv?Os8_nr9T9HhgMg4tlB
zhQ)^F+75JL`KRR{l9P1er;S_HGW=SJe8bHFG0oRugT@3sidoy6kDyvj4=rk`KK+Mn
z@xP5BJP|$gC~ky#PURy;v*B&{6plwf1|H>b(=-80J*lO-;^Nf;CtcL>s3nYlByY}M
z|4{x5Dxb4TD1S9__jU1?(&gXkvE>(1{%M?`{ID&*ONQR?74X}{KqK>3(Nifw4N-%Q
z%^BZ(gPUm{1hGv2og15#C(rCI<B?6>zTjf*^!CkK+|+U$F0hN&>r%K_`@Mt9%_i_!
z%k`?LX1QJ!6)yLOTTANLH_V}RZ)zQkO0A(;Lw%7MDLSRDm1F+ct>zDZ+@0>t(6?4?
z0~PH9$7I6W$^YL8ci>?*pU+jp-lB-A^I}7b3~2~HxZndm;MYF5=0Um?-mtlcH`1u^
zpnX96+6Qd}3NDa#g*Vy>#IJqOR-oVlqt+O>fct?9wCqJZ3;mpl=o2VxS^<c=O#0h%
zV>Njbvi5_=|2YP9@e8)%<Hmq2e%TtmyGY5qi<AtaNCr=fVPDYuF8Tkb@MNO>Wslzt
zV+P)+ItU%A!lxSjIw$W`>LF~3t(QzYf(>FCvx#?v=1zP%kx%L(aV9-sYl0@M+|QiK
zz0;265Jl!M?%U-iZN9CGtf-5)iMz-_g_J0fl4hOGCN#Yn8JpF2LJBrBzIXeAkJ2#T
ztVd~>ujWx2W@tMwp7fUHn(@47sUA9>H!alz$BXPC8@-xUWL|*{y#gH?MxY}Z`cuk<
zS;JrJ1{s1<{JgsSGs1*r(awn?*Ut9FclriC7+*7f;9rEazsNQL4Iy}Tut$u%=En^k
z)#M>fnlRm3uLdF}0@N5kZSBoY0=Cc5-1PcC_$c$y_~3&dh%EF=i<z59F_flXQkbVm
zG32ISQkb(yF*K)NQkcI;F@&dIQkcs~G1RADQkd6BF(jy8-@r#nph>jmLGS|^;^uAS
zOGdU<J-}2D7w=2W0gjmsQZ#An5C4VmL~@4zr^wIne?x>KMfm?>KscxodJI%<pSl`$
zFAg!q8;cog=w$a-n<7}}n^J^Cc)sGFq~cxL^5FNBR5Nz8$-TF7x7AQG{N5DZJ4OFH
zMgK-kQ-G<%HRIZp9z2YAd}D`-{_&Yqw96Lhlh}9CnCis|ssNMTizYH48f43H#cJVX
z=wHREM#H4Ij5AEicbQ>QzW&Qwjz4*E0M>T=iHmDsZNEEa@hn)|?|LmBxY+Wq?Zj(+
zVoXQ;nK9eqEn|o~F|_on=lv1oP|>%x=f>N{Fd@&{rvISN8`IRuqM&>5R8_LV%$det
zi1}+n^%6UxA<m6jF+B)y_i0Ba9_U1OD$zUtR@2z8us+kj;#jlbeAT!iEg*uH|1Ldk
zs9Z9OC*FzoZK%F0#R6x+{0vB%2ikXr3C@E3nRn)Z?i%>t#Ok|rA(w#*$nq}B2h|A?
z;lfP0&*^k@7-DxA;xyq~%aIHLR^^B%i2i)0<9(*%ebB?uajo%z<6eySANNMQblkdc
zq>o*SpOxm1(jIjk@6gAJaeLD=A|5FgZKG$IRWnmxC9!=_ta!t_z-1tp9fuH8vp2kv
zM~-CMcS7pahsPO#H)_$*by7k<m1ZbwIJ8K6O{>n01@n3XALOjQoxWL|-nV9KP`a5S
zrL5rI@C%*l2Mqjz6vHocjvp!T3sMZf;9e>43sMZf&>4QDz%NKK{6Z)AkpjOU#qbO6
zRa+9$m9^8B<rj(LsAy);YRoV~uory8{2mx<=J%ShW_}+!*39n%#}+O(Q@>diXhJhn
zzgZP9rKP9-KW4hj@ZH8c<zsoTj!Ggv2U!1#s^r)Ard1_(&N8Z!hq$ol8iWOP$@kGm
zq^<YYW&p3Tfb|ye;5z^uEWniY-pK%tuz(j^K-pxMR03#pD#vG_N-Wd>3-x_Loh~Tp
zRCfO+UAvt(nA#m|p$4hYA%c=l<@F5Is}|}*F0|n{4+8ZZP?)mMX+Wnks9xsZn(Xtf
zN14+on9rkU!BHm`*F`q4@Mhk=UcRAkaQ}lcd*%uTU$e8C;ML=$F?fYl$wg&4@wrjk
zr=93^c%TkIpg8rqyqt4}4BhH&bVv{A$PZ~sVkbYvi}NO3`6SXKGPm88YwWyV5>Ob8
z#gTwZ_SPfX0F!C8v=uD@CFtNw%0&>D4nbom5S>GrC$WrvngE064H=3~<5iPBLay2`
zQrI7mirWI6J5SqWS2pDxo_Gk*ng}WBBJU&tbGH;1D@~CLNT5he+#{UO?AppNZ2k^*
ztnE3{Fa*a2COMpFT~&bbHP0AdojduwjjgOTm22^NT~SlsHs`x!vcu_S4rNir0^M#g
zvfre&ym%uJd7-OKKk>+1&Sm+EW5cI+r>S8hm;}>%xX{m6{hWU8f2`@}iSy~_x;Z-5
zCI258_-8Eq^A^5H@ck_OX@Wn-!rzjCzuCh7#==(%{$vZEC-_aX(amM^!%qJQC%mz&
zCOUC+EgPd3{-PWBjJ)EHZ1YEO2DV0X)JWG9gs&fy6U?uUc26G6k6c`zrn?^!N<&^+
zUe-BKS8`HLaWxyk(UoDi9#MAJc0vu_po4oZA^+4tJJY$v@uCHR+c*0k{^&)%>dIGV
zKZz0BxD-cTuPOfsRf$=`y%GM1Kx9qHqKtfp<x(VgqYaaB;39u+f&hxMS$p<8jy>Dj
zK0K6GaGNaD`kPpw{*D)dWp$OW%po4u4I)zZU;j*0$F}D^0lV)3cVO)cNV3Z4=r*la
z^1<}_Rnf$7Vy=8W_zSX_i`4SU?5euhw7j~?*XRD&#2R~j&fAH3ke?Z^0bF&-dNW0-
zh#hr=Ip}6Kys8M#eSANw>6?LS?(?Jym~-F1t8sy?f%g{ePy2Vkg$W%q=*V?EdI=Ws
zw;Ao|VyBgWk1{s(L*dVpw5E>D{qzf5M4Qfm@ki&BNRd^{+c!wKsLa7VyzylSArQTw
zB6^q;8}@z>3Lnz*4R50fQ9<(oT!EYW|Agc+Quy;M^_^iQTb#;PSjko}XVrL|gV^|i
z8P~(B_;wef2O2{XsI5cEtBlH0)&b6GLeY9Sp_;CuRA-0>_m4lB^y$V?h3b5v7n6qm
zO3J~M%C!)C*Y@p9^}&OabEP;|Fu>{LCEpkYQ?1+_`YE?)N&Mm>mpIlx)NkYS(wpy6
zEe=wFOSS2MZ)WbA(!wA|JQ!ZJCE4Aws4Y~~?>!s$pC(4g`pwkAZMcEWCD@`$3lopY
z))~!T^Q$zEXahs=h_|@FBmAh@m+a}^IZgdG19+bWe9Qu##@w-FJAh`Ix*-E~orSu|
zLe&zUb&2=n)HHQr2I>L}b(w`4A}F7rG)<kFfjZejonfKo-wf1lN@SXP0M6pxavkg?
zBb#fytrWM(T7T()&}ff~?xKZ}kyXfn`F*)Z&^^0E4}0!358Y)BG2_RP$dZExOwQ6A
z$ZO?de#kUa#$&mrgs*+R0Jc^Y+2uZQEBMBm6mKt=2Co&*4rQ!W9%@a*6*P~$h_@Gc
zO5G!V1s9IAy6f&X%)E+2I1jia@!LhaIxg9_J^MZ<9;TqI_IBZ~#kK}T-JRC6eJbAj
z)Bz1}(0geV&S<Q!=<jnt!GKK|l{LbA$?tT<Uri*0#x#zn%r!UzJ>@CY_qBZQeX1$a
z+?4QmpW5P|RQTBA1!ZemCQklxGY9RI=5BwzYTK&9&{=`VdK7;C$g3ItNgjoz*su<*
zZvN<i@Ox)gANT?rxLC7K{V-6uC$xw?^<DapZ%M~TO#B5+LO~q@_F^ble{=ne*#^a`
z2}Jfd&3p2k@SZ%=V~%%t+==xrbCy`hjy_K1vyS)dw}_<Yc+YxE=bT=RK3fXm9QO#w
z$RCN<MuLAPxC?wNSXwaW5Q0c>BohZkxfjpkee6m$CxHnabx}ar4!3!x=1$3%kyC}9
zy4WloNj298QosCV9vdwzx*A_%X_@Q^hPxCn_rB!AG%a{kYpp7g!E)j$9_Sk2zox-|
z6Lf?ltadrgyJ+*SOq*8*Vgrji+w8A=A>ciG3&Car-m|u-;)fEp#^~_f6R5pE(vno5
z&AM3sd#YPK_oy1Q-|U}L`#*LfoG(NBVIJCQBCHiSk#+8>86sHFb2$yZs^Hw1=LsiL
z>(gNC;$KZbg<%I?o|ohPA;br<VS6s){x0s5c}bc8hm!6M4h9+v+$kp4b?z97OV58<
z^gM5)bfTit3TN>}7n;9(c&InrrrDjdw9+nONZ<_%^5n$jkz)f~P5xLPyQi807<}i%
zdY6D=#vB2~BOBcBse0`2E71=#Q?Rrwd4k~w{!H%-FTj0|7wm|utK9BwsH6ZhoX?5&
zZfBQafQk+FHmX1)N}n4moC^uI;WN=zK<J`2AMpUAC)AaH=q$$6tUu^C?rqphMubrE
zH+OVHZ*Tk3?;YOCfop>^ed$g0Wq{}{wl4{Q1mV-4ritza)76CdIjuE&)MWZ`)0{(0
zAF-PG#JvL&KQA_G538Wo0%FmKgg<uons)aKL~ba+#Pub})!a-Y6(MGDs-=yGQ`Lmd
zu|<A%qTT$FT5=gtP!PU>)1YIQuwUw6ATo`H2D2Oeyg&LqfBBdDG<px&)kQY?n|F7E
zNf5OqZ(6<?xg5Dv;w%+*`19fy%TzyuGG7eju6GvO<PI-5Xcwi@Qw6~gNHC3d&ok|b
zOVdx>mu}}oYCbfG=p~(;5`~J08Mi!~7rX33TGCRWRU34@-TgGEL2*i5<R`)h$cl^r
zAD|-B)W%*^A#pS#T8hR@33KmF{&m2nY~*`yohylQAwQXUz^8ejM7@(qnxLd9N|GkL
z-dwQ+1)`;#>OBtL5+l-IAU3OXLb6En8%!l$7b|UH$8YGmOfFT*WfULdxG=OdZ|Vul
z3%GaBq&a~IMBU6mGBi-}u&~)V`azsqmtt>H1c8QjlRsLu0^x)LW^AqY-T^Z4W^{1l
zYj9tjwyPs?JqTE?$;@5Pqij%&`vNnaO^CPqu<kLbRq4l}4+4Z{ia2Em)f&86YjwE6
zXA|HC(Q6H7cwhH(?fuL5a6e7=k(J5AlgB4%0q)wVYUI0)$`wl7e--5BB=apRA;IUK
zd|)m_yw2^#NH^m*i+_YS((qR-|A-!7NCzME0@=44B)-)_Ann!BLt^Le3GdAFhS!nC
zl9sW1gOBr=c=DIG2;~??in{w#Q))I{w|Ez@a)A?k?6yo5C84~<aC>$EQQqnzKlN#d
zfD?LmN`b$8Tkv2r#cXxAL$-CfU)C{+Bu`Hl8LatgS^Io{5EFQ0tNS3Rs*B7haKgh2
ziebP`Hk#GT=t@UZaF;iUA0R}F-BpEm<gxID2_ZURdXk@L^gtB&RZg^bGsht?Xo97K
zlh!{C`dbB8TM2oHQ49{<Y8n-YmSBOWh$iDUV^%$#)D9xV_Lg+Az306P{GDu%E(neQ
z7w#>l(+PJ7kH&=3E}lx~w+14Hi7lTZdPWYY4xf=Xt*<|_k?3>bA@o%%y_2Yir}&#c
zhN1A4X3rGl)Ytu5G<25EqTvn0fA;#D_ri4dW|;0Of^b!2Gu`KzKoi9Og5e}|mVAS@
zk-%>@QDBbyFDk?&DijPpH{$t1##cLUeyVxX%WGm;e<36~jhn<{K67s}jkrw9hG1#&
z$k@=0roL*%{C|)shhqc}@kfqz!UzHdiHDU`{_#H5Fa5UFR|%Qw_n|79Ga}=@)V$w0
zU+i=*V~m{_3w{Y>Sk*aSyl{;^xH46DSmY*12$t5lKGra;a9~;xJ9aR4mV}M)N_dO4
z-$#E2k*qY_<qNnAYEOnO+=Y0{i1hAN6p*-+J60vxSoPleM{!dV0AW+QzTR5{V#XKp
z12?&cb*ww=DzS!PzZl2;0DB+N{fXSWQ?BG?H2<ZW(t^*6la1i>PlONFY<K<w>7p-;
z6X~KBLUb$mC|Qs^An8djS5lP?=DRId3lr<yM;PZ_<P8|*cdIY1zUTsnFq6?4IPXem
z-y8BKk2Cs!@Ty|<FaquwVyd6H*OODy#kF!;ab{`x6Vd|F?}`=Z+rzXNvG}7q%=g~R
z_otihfzEtgX6trI_4w*b6G!v#b@Bm@A=}yQ@BTEb#>A`Z=Mt2s5%~ySID=kKEu-q9
zr=A<@J>4G*{w9b0)XQ*ndD#hXox_L~7cFsH5uzDIMc$j=g}lU6a9p@jYp%<XWn)(r
zN@W9EOeA|pa*<lAJ@FlL`~_@N2}Isw{R(?+00N$KNnPx+MwyJNeKJ4SdZwROEv6@i
z;E#GRCW?cm-1%`UuQDr~Pu(X)$pX~kgsZPQQk#{3==X^eD1h}#HJQYBT<!0PJ9xE+
z;=MP@9?c>U5e0px6FcXN=+xbjsqpEkUpSG|fIDlW6InCk!$cK;G}AD?5^uwm<mivp
z7J`OkX}W*9v@bTC9q8)<m7Bc{XR3XXm9*^abjwCZhnse>jIhm$yd5b2B<NKH%5W>I
z&`;feF&r|jNjJ~QWE=W(;$sk-O@Gm|3u)d+HBU;kx4bw0kA_F&?UeoNKWHB-9=c5i
zt|}9@(%UefLj1ADCN7#Eh*8}$Hc;6vdH7HEuVusM+g5v*zSnK)$m(cJfGy28_!?Wx
z&CF98B~?!^(fTnq<%{Ud-I1B#cP989Ohv1}@8%gFCPBlh$RY9e!$x@g^Wx9(Ng(1n
z;hlxvuy*-XuR7b8>krSBvHl}y)F1sm=p5lOYLMcV>avSp9`KG1Ey~6R2cDrslIG)B
z0LQ^n-G>+voWLX9l5!}jIkq6)cg$KeS!|{fcPeD9J#`hHmfnUrl!L3U&l?U?1D&C{
zLWc_HqJ??Uk!!%o0Vc}xVl_8%uJAfeXKHD_-bdT7AHRa1BGi9`o@y!#RK6JcFu+C?
z6O5)@nR3#&rdYll*QLHYeN~T8U1zb$gz8m2($YM2JPxr%Rf~C1)slJ9qQ|wF$F+L=
zXXf#rdX$aLek-d^l?-Y2QRW(3Pco*7O%qrmwukD?e#$I<LGaU5mjHu{9@2jhP8=;@
z3qh4fo!1Udrk_C9rN+-iUqt8Zj?AHd=g_}r05XF9eSOA<k%Q>kvpgJZT^Mx=^*u_E
zQ>r}E@6IkSHua0>^}8e2^ZoUFe>xx<e_I)U(?oI{JL$yc6f7--Je=(tyo=EB)f|Ao
z15w?*^T$!Q2J8%@xuO<B`_&%bb4~tF#-irgb4`yqTooz*Wz2wMk8v+1aM;#8WwSog
z<H>jNNLxfco%5Lv3#|G){_d;W1X;B!zI#NQvEh=`!Nb~#Q)ido?ir@_w(+3hx$o_T
zgUs{rLB&nE@u6UxuLSsj$G(^}`^Z(fCUGl?1I^bQk@7s}M910N(jHpKSrOjGTbKwX
zs~sl){uL;^9Kr<V;&0EE<gTl^jQ*<2dUvot@?7LKm^^clJD>0UkyV^TUhN&Zu`c@4
zf||+?Ya;8t0n7NEJH3r9P$t}-7h0TDh%J@P?ju5l`8VQB;<*g{P9x*;Gxzb&S#mY#
z0~GiT=mI&Dd*Ef~#<1FGl6W<HlK9jey+u>RZl;J&%(&J9fhhtVFH*Ori0XeZz|0(>
zhzS|}TAH8kEb>d&3`c_lGhb-`Jk?=R8n@4UVYw>g13U4Qq5?4bj;y&N@;{m@h;(MB
ziULg)omqz-N)(fKnx99zF;G&og<+>ZOS4luAOCxPx;xj~@FQ|ePZc`;^=sM}8^L6P
zTG;hA^wSrSRTQ3?FKSqyYr-h-OM3B?vh?3~zWB=(X1*{@N;fh2V0L+$A9ngu^FtuA
z#e1XVIkNf>=7&*0(ub_<=rc_X#EuQqX<1W)@0+HE(O*9`oS4S<x0)ZO1<c$afdtt_
zUXsXMQdW80o3CdjRleED4W|ap;4I-t^U=QiE%?N~yzM*iwv7JLiT%IJ-?VhB9(}<m
z*j|47`FT0hLa0EDZfh5ex~A0Qp3XrM?vto<=tp)7BYt2)K>ohCX{nfw-LaJ!KTc&^
z@C-CC@ttElhs)12yrPi2o2P!Hb7kN2N6t$jY7#jH%2C&*5oPvDvw~cS`XB`sdMo*n
zs5gs#yYL%?{!{YLQmqdX(+STh=r&C@x~zx`<c5_`-Vn4!;_F`b3S2bY4^Ya~MXzBX
z5`jIz+2q5#jf?Ze5Dce)>W%dESCaf3HNFf|)L)<TVx@nk)xdr1j+&%pEG4A0dJ}%E
zJbx;6#$JErM(=_UUyGlX`kO!CsDO{%W6c-Ho2f5O;R^>3pWO9H&wxz>noveP>{Pym
z1x~aD+qp}_GWvZ#dgdkEV)7mSyw(R(Pk`|KPHglq3!TbWyx}<SNQr8yayXGEtbH^J
zCH{u7E568i2LAAS-GZmEeE5Nu4?Rc53jKsf#!6?R|IWL9oQd=8xn1FT!YjIQGIVHz
ztb^Y9Jt-O8^$!rU{xNu;{(vY(un-tGg6Y4=bGsh78#h9(9W}Rm^vvzJ5pvq7xxGNo
z+>SfIJQ8klg$Ezk3UqFvClimT5s@{TKu`REoj`|iWs_99;^WR9+v|^`jGy?}V}8Vq
zng7jvxxvhppJ;+Jeosu3X7}vw5pPap!n-|qN}icRdFXrltdsIm(__EeYfj16VX6k`
z%9HY2-mfkGD7R(hg~e_zC)s_7*{1Yqxm26u#P+<3wS8wOW^94xG#B(-<q2=Y_t3#O
z%{$PH@5o!KUcwsi5-sh<@=KmT+Y)~?EqmR;4POpHO*nk-F6%J?A9t>6F$|4++QaW<
zPnYt0<<m9%);xU~?>?WnHB>ftD|fG&yH~i|WA1jxXO%TgoXXpH{JC-OZ~)({@fNE5
z%s7{7k6R|2)97Uwor*47R>kj?%UE%>V)WY^iaM2BgWr>vmf6VZY09!`QSHFr=(cgq
z>LDB=Vsi-KTGs5K)S>a~&16xl1*O{N;nbi@U}<WY>7wHNXY2B^{Vn6WM+a1w(WWk=
z$Ehx(FIzT&8kzJ|9Zfnl&D0TJimk!ECMfcC$}!EaRlg_MKQOwD`f+ErAZ=xT91G?0
zPrMC(0{`MoPUKTVAMl|ApfHZ*T^0IV>|zGFd2f7F9THx#8!#F&%R67QD_bq{<nPa*
z34gpEYF}pDx7zkS@0<mR+=Rw8mgsU>W0fvfHcp6d98b`-@dRBPPtdjT33&xs*dhyC
zWMPYdWtEO+VQc77ofq8j0Hs@bkoFxC-#T8cJ0xB2WJ0rEx`l1rPu%iU!433~uSZP)
z8rnL(DXTG4&;5pms7@TKXfLWiPpUtcOr74LPQk;{^(pGaAJH!0E^aU&N>AIp?6y2@
zsbh9X9<dL9Fk21Uf=AI092TvaVEV>D<y%uR^{h#Hl%YRzu|)qMBIFFWPO8>^T65&y
zXeLm!SO+6>>s|AY#4I^?DVx9ZBJrSJ#D*P=3C4Zol3g7G-r;Evdt}{vd16%~UDy%Q
zh0Osn*@KiM#gd#dFa4BcS=dIxj`yj5)_gN>qr7l5EJ`Is=C3{nN(zthXog_Xi}0{(
zK!<zqXUr})qNQ;n4W;#b;5RlW#Ti)0lT;MQ@*6~^#6EmlQ>bLfPWmoltBh0+Ak<TA
zO7poRsF*=;0V84=8gL^#f+j0HuA15=nm1a|QBuuMAlDp6&MVhZkj#1&SUswrv8CBq
z<A+jxt2Z^H?6CA>Rj1w*5e=E%?zJ%{+X3$+D-41%^F0-;Wh4<b*d9p3It^WOxTTO?
zRNinMoDm3G<V2x4b_$fZ*H1{dg*u}-F}$XIG`0d6WFK`R@A}I*LXdM)X@j?6Wp_1z
z^LJ3aqCL&)qCLf~)~P*Bn_P8zLdxWNg=yfFjN#SNL#sulF(FJbZYqD5^03TdH+oSZ
zLeG8bzB)c#OA(I@TW&SPWWPK&*8lJRnCCt$qW(y0Vh_R}*1fJd&3!5v&sa%>ybhGY
z#^%=XygDnjpVBYW-aT}Zwhl)F=s%-u)MIYefVl~k(ke&|f}8$2ardUagyqQSLv6=8
z%Twx*gCHSm!j;NEbip!aaI)kli;zx?YAS6;ix)yV=4`h@3Oy$(BV4R^l@2!Yck5kR
zM;AO7=!bQ59|ds50W4)Ad;BtbFHUP=a3^z7nP_z~^;9CUlsS<d@Q&#N71A@;o%Ve#
zeojXu`h%gz*w8As_YXBP;>nXz#<%o0GoHiU&we(Qh@j8$WoXraO(SDhUX?=|&WhCh
zSt1q|He)hRnT+X{l*zbMJRm)ue>GOhWbEe>ATee&f?do^(lcQ;K%Fex+~7{^B|e}B
zwc;wMEB`Xco~NM*I~Zy|ai1Rt!#F|~jWpsL7nt6rL$`eu8FeZ~hZJe3t2#a18>AfV
zXIVjNv71qxel|vxPU-14=mCZAD5+`l-B<KDuSF~;OM*~4#%MT9kRUrsXNxquL%od;
z!G*}yv{0MoE4&R05$3UJSc7z=!~LByjHK6|JKe`GVzo+9eFE(5W8(v4!$28@MSbT?
zwm?8fiG=nx`6Hhts>zE<rJI=@@flmcnGN-EsW|fQc5vs=uB)*<7LBm0vtG+nH^*#C
z-JDp*jh&D8yqZAkur<DSE~Llrm0y<d8~n10)TDW44Y10`Q;QswUS!u@@W@X2)rsV4
z;|JzN<cZLhZ1pfzl9drJfxVh8wz}fIdy9mq0`JY&Ym7yI1hu26jn$zI2$WV;KSL_S
zfe6Q%9AA4gxRTXb<uz}>7>q$nCx8(}CK7j-K*DT?kmMX8Kbak1c&z?feaymSZ1&?^
z`C-m^(cp*XeSu+pkHb6)f6Z`in_kXf`H+$+dRIDY_s3>vtP9`)lk}(O1AkQ7>B9N?
z3_)f3i3~9~aEAX1AJJ;LM;FeQMon-tJ_$qyNG?UQT4DTS=u?CV$@qr+(O|vC-8lU~
z^nC2OS5+*0N}qcdbVt(hM{7;VH`<apV~S=?KZ#MQ31ENa6G=8*z7shzHO_vNV@_od
z)&nTa8n*9*lA?7q{IldS$ivh$5Dn2Q!!i4@wH56(P%jB8f{g+rBjsk{?<Q4edN~Sa
zmV0U4H8}s&J)S{taQC+^^^11T;17Mu^jrDq@_xVgOI!{HYztzI_4E_93SMrl%k!FN
z>Sg4GZP1`NiWzUAi`b1l(}~6mG`KH7_3860)NdhCc&Ich5E#!Wv(=z*ku_b?SqZG!
z$be41ct)~kXO@X)cZe&Ao69n?j3+1N6ePREi*RCSwj)Q78^qV_GuXRbkkKhPqM`by
z5{pbF9#bWta_LT6GwTf}@_i>d!`A6BNZ1k(bvgpOF}1bL$jV`Q3{`Y?|L^zv#AM+p
zHf&ZktFYo9z<FZBrs%%V-d~F7<K}QLdAlH4q$SyR%}+94m;PFwb#c~Nc>$S_;!C;j
z_}srL1zrckOtH}WNYOfO!Ns!1t0!`-X)L_a7vC{gDmbH>{CF}m`OGp7x`<dEyj1U6
zk|AnKy4LQE#QS`m(*An((f+bc&}e^~0Fm}r)`l$Y@8scDc&+E!sr~(g=4WYt@8*#u
z{eP<Og}-;v_nPPZ>3hwKef7Oc{HngU;WJb3eJl(^P!SjlgU4DJG~7G;Tcq^w><yZ;
zvg{2CxRT))ajXLWH|-6Yln>SP@cvJrm25?w&)6?{C6~uceLL6xkQr2xhv>^d%75;R
zyqwSh-so0QjpFIOYH7<X*gb9|xzp6q(wD^vF~|^RG5!d^xM@A%j~>Cy<ni9z4yOKb
z20tw;+P>W8G=GZxxysL3l-dz~aZ~c>MaXt*(AUiOfcPSogUH2r6UZrmZkjtH+H*e2
zA}4|`$9raet*tByUS3;yz9(4AZ$WS*zlFh(et3h^@w`90ax*@QxpUw2<H_d_@9Gx3
zgx|d2cz$z3_w#k=Zg#(Swx16@#FEe8gXaR|ws12>E6^eBv}TULeA#B2z4K)uk8d5j
zCUG`5s?3|pK{qd^ZoG{;o-EmfhQE}LXyxH*B#1i)UC@d`P8&*D0K<E`1^Z~dAFSjz
zH}qcg^g5)N%E6&Gl4qyJcl30BWQ$21S{gj9y0S8K1i=ZCW$F1-pS(wMFrngkrO)ro
zY7hF#2t9R9s{Yre>whIT((;;RoY#HgGLehBkGk?lGI{?9g2~{glU{Jl<s?&Zv<s49
zMT?M_ilnt7bjbBmimnj>y(?V`Owa)V68aF9khxd}y44h^H4V0H2~dc2KDr(1lca|b
ziY)`7iR7}Jocc3OQU$A0C&n5yd1P9@9%=5V)|$Vr@*{79Y#}2j&3ck~A5O}`&pEJN
zp54JR?yb8x9!qA*iF)3IW<w_>XLlL@HG<o|y(?63tP?G$8FF1g@OHWK;BM<Uldff6
z+MLdxx^niNaYuhUczCirGyj+)@i0NW9qLtGxjodB(2;Ce_%tV|x*yen3i;gR<P49Z
z^@%tItyQwf<y2cWL|P6&ckD(^%gsr6$VOACQ>WGts@#r7etWZlR*bJMT3D*Bl}zH}
zN_|MSBdk9f28*}*HZff9Yl`o>t<9YI)8pOKrDhqsre(vAm$Vt3>0dn(TqX0>{M94R
z0^*A_{EB>WN4(_RNeWWY7C4ZQgkbk5?zOvzf%-b4{Z=8iIJrjUSr7WpGwH!60@1<M
zLq_7{C{kQh7d=yp^n3l2=7P`mbdKPYeo&66$&OCU64K20S8}yUw!7R|BtMh*4zH_M
zK|fmbX})|XSmbVy?zx{_^inE#NTOJ>NAGO@=S2IsP5?+0FOBbnkWnQYf;OMS%6Pyw
z_h;w8=^#lg%TkLJRjs!mESU1$L-b|1WlWX?lX0Faa7S$r>92F&9Qif=w`L-d{{BmO
zjSMi#@|x1g?C8K<;9PW>tY^lRP6-~@IsRDlI73Dicz>}X*h3a%U_!}c-s8CoN0)lz
zJ3eb-9f0d%MoD5GvpmPbvt{ld%AkB&lco{eb!w{?XF9&Lo+~ldGvmG0Q_Sc?$@HmC
zuE3TbfEN#$O-Rgk%?vD}ok@Y>-F{9HT`VhV^p|$K$Wvv!cFJFe=Xbke+w;#o_gv+c
z*=X1YMx5xtt*F<6f%-FYg7?esu)_W2TCM1aUFHno2-IM8e8)BT^~ZP4<uq3HAa0`w
zO%MejNH1CAXTUb{3O!LjtJI6DjRsdjW;JCy3iAPZuNi<}9=%#&<<3j!Ph~m=e+dpZ
zWiNg3=s;vP9W|Sgf(IEML2wRk5(kly)~CKedE8mNnJ50{5=741u*C4ZO-RW2ych*R
z#@!Z){~s2#qq|PltlT<-2E3H<!R;>1!7y|&4(`>`0R?$It8^wu9!fLY{ISP<!*AT+
zI7d+xpP4+q%g0l($gIiwKGcMBm%sT#5BB2zBO^ompBtIgp9y)>oSxWZ0?Jk~0o?Z>
zR^&Rh4}PJ8mDSmDTyAx1DN(9B+j6<is%FF+^~g;aK8zG#i6rrbWMvSAFwK@Cb)T5H
z&%I5I-g-MSQe<y$=x#{lLU6mvDQBnMs?E;4y&Yuiy-}M5p$$08s>t^6$KLA2K2ieS
z+SWih72Ym9=m6Py%}u0B5I+sScO!<+`@U|jEAG|sJ($+sYwnUC5!WBr^K(TLKLfN5
zoilcqx%?P=0#8?Tn9H@f6Ssa?R`piKSzz>0q`do;l-ItJ^87w2R82ct)GBXdh$={O
zUF2vZ=D_K#HIFmA>R}G7ODdZ~u9jq@q{D$DP4?%ed=5IvvgK@%rS;u|u{>!Wr2)~X
z2vmL^I#pZ^7P-#7OYgp-IZ4ll;gxume2(N89b*0A{s-`+n`VrmWl3!!Xc>rnGNG0?
zC(FW@F8KS&%d;QLk{4w^9+^Bl^{BBFG7%Mgx`P5Vp|8bTO%-$*V(#Z7vKq85*&{on
zweD+rYd*R=_0iw>2zhxiRFu6x$cf%+`N}-YS0a|LG-J{1cExJTSLPHtlU_il2y5}Q
ztsKzwwd<8)B4s<#LkWV-^z$MG8TCZ2Go+g5`*}8N+;#5ur^VVv6!`lc;54^)^DoO*
z2ud4h!;@fJT?TjLs}GXC4F2PsQTTlDazFdoa#2(;0QVx3yAO~jeS%V!RZx3TH>1CB
z*dJimJLUM|X$3@ftjIcoe|I0JcN{czdV0wHz542RskKozr**i6>|!fW2=1C6r=2m+
zI@35~j{K={#_YHnXUuSce6HNZP=x!yaBDiSWRmX=rvS;8{on*y{I%<P&#0TLghlpH
zGG)y%DxJ5G0i5V15T}*{7@C{WY5U<Jw&xa^F6+0*bYLTg%@eOu40bRwsdxtZB%JeL
z#2GdA8#QV=+a_4<5XsG&s-Ff{EULm;o?@?koLK+6s-H3Q*gFU*(gzz&)bT7ca~XPC
zw@6nXAcjVGjWA26R3<|B1+sr4z3B;tZmi<ohCL{O{Sk3>tl3?;dLebeLbzi_H#8$U
z7TbH)cB3;LSQ<2BmEH$*G<8R5jt7Xg0vX}KbG%WVKMG4WW^cJPgKIS1pSS_~BBL)|
zPM;t@JqjI{`eXeU7#tV`P<E!(iQ{QM$|Mn@mVg!B`G<?q#zshp@*UtTMT4uqGySnZ
zFBsICT<5qo)v=I1!jjoNctYp=&VmtNhf~_(TQ6;5KNlWeO<?VX(8s9Mh{CfBq@5e<
z`4<*n4I)jc%%o2nC2^wjqu3NZujODTG8wT20QM@N%jpCf!2q2iR~bCjGGC6^0uTlg
zA3svwct^Zo>JgjLQ;BlAOqC(XnOKW_PH|S^UOw-w#NJtn_z^a-bRyol76Qi0LFPrk
z<SNUCP1&Us5We2Fr>L-0X!G35dY8BE+YAJN3rpi$$F%XL$a8Nh{`)Hf-XVv=F8ZnT
zr{kO5dGIvzVV95omE_2yZ8FNeAVpe<)|SOi^UiM0GI1~cu}76?zD$~d=we?_jR=>S
zx5SzBF7ywhI}YD>Z^NPVu}bUikBvlvTIDqF#ixB$dt`g_XL%hE+)m>_xzBz>Ds*5c
za?XCqj*1u6zk7Sz{`JQO-k06K2lCDN+5P*H6T2?Ee|G|aw{h(J?EZZNAoZ_5yMK2A
zr2h4%`}Ymozt^dMjTczDe|H9A)#_h=rhfz85ri%{U;XRP^snDLVl7Ref4A8FH6s^!
zBRCVdyv_~==T;xD4wFFR`FpB=cQVL;P4(|Q`uFd~Wb}FH6+TLJaAy6R{*?axEhF1}
zXhw7TqklEBZwUbleOs`!%6rSTW*H=fd(=li>R8(D16C(=Pz}Cqcxw)<V&Z#zH&|bQ
ziX%)*s=c>0`zt%Ra3WZ9mG?HHj_%^pI$(8e1N*Y8`L?QLueWh9H<3@nAN3JklyA|h
z=%1%rUECL3rHgyrdXgM7?i9OPI`Ls{viNyR^DKqJ;WxOvM3-61t6HiF+Om9t0y#a+
z&*-O1_#O9j6~C7~J%O+-&$mpx>cQu=eyU1NoQeZy(k!1&!`Cv<m%wGSkK`2Gv`Z!_
zM0t5ZpS+L-c4P}myB9j0G=qJQjQ@Vz_LdD-J-FR~Jx*@Hz60d8u`L^HKudG)ahqFK
zTt!eNm0{uhUxD+>{R1W7bJny%R7)S@->8;ZV&s%R594;W<7<!N^}}w#q0BZo7|mMh
z!CG(?HfkaW*4Po|xmDd9jc*;-*0SNihuWA^iV77}5ETo4Lv5kA89ew6-AK0dr`(ht
zM<HYQiQxzPd=E{XBR^ta&8;4C4nzrbo61_RT>DCFHGr>N_A3*pr{-3Tt3b3c`_(~a
zAct4rTGST0Np?u%)VfJ3=kiXcR@RJ1O0?}Fqp}uFj2^}myKkyihJWhI%5cOGEzgGC
zqWjaiPw&_0qMt5O7A*I6RIy@6&ks&?ft~di+F5_0nf0|;YZkj>iLyW}j?LkAKIyW)
zz`yndv`EqsG3=!RvP#E)lWF{uzNkL{@9Azj-$+i@5pKc@5Ww@^205&XPgK;g?P<~`
zEuxlD2m1U2%>`ZbKSr0?(Z&YqY190O{3(_)amz;qHxQn##Z-=iNO^oNz7wtxrS+!p
z62CFU)Rf`>HQrO^PBkxZ8d2}z9N)6xiY0C4bJLx23{dz0C9lE30EdE&p?S)(F7kzz
z2^C3EME=S#3~)PdagOCf__~`#7-N^I-W{ew+RFl^C1`If5_}W41dA{3ahM=gLQi+{
z8HRE7XfGpa>o~xu4`(-yZ$dFi{oJz+K>xMI1<&Hn^NPlf47|S{C6dZGGDg-V(ca3>
z@Hla!xO+YCjngDsXfS6*wmR|?-|Lxm8Gg>4zQW7XYv`7VPj74WH@o?S{%~OWLVxFa
zq;j$w9A7#4i{Pb|Wx>lUCl?0C`8%{KsX%wS){oX<?mwA5^O!U7Vrp_FY_atJE6bJK
zAje7W%)^<1!>L0(`Tj&83b`JWy0UGxEr}+f$xSkTn!ZS)kZn?LR28@nK;|~a^IP8$
z=CrkeH&9~BifjLT0O-b12!wdu2<n_pgowwtWIMrjZBG~a9)lMR##Z-ia*$B%6Cyuy
z7e6a=Bv#za0{R`z7LD(SXz>X?RZr=?^-ovo+h#u$V(n(dyd|*ZU1?Vv%vZ(<CwY?e
zZ-%!VLwkyH{n}05j3WySP<t}A*)5rM(f*~fHNPe;M1jA&qjP0!!T}6d{;6;PYV4l$
z&SKDBmQhUf!9tMQS^S)00;G2q_hF;oda)<%ls9e??vBACm?kZcn^M&ic0CWG2-Fg(
zKza+-GurZ3hi;;LsPq<WPsUFLM|mavk}MRmH&*x{8~*M$HbsLGK1>|Js3+?ItCLu^
z+azV;Tf-6`wZa-V0M(i#P1-1w{>9-Z?h)U~MP4BW`<-&lkM^JE{y_>^M0H<EdDszt
z&}g?ZduDr51BuOL63_D8d{szhI(hQ9e}7kpA&YFiAnbJHiLle;AU=!@`>yWSai7r_
zmbr@%Nt0d#<>Wyo)gU92>+i@_?Di%IwR605p)dI7>-2P>bRfSd&{F2WH&OgBLXD#U
zIcV(Gt`^BLVlTx}jAbTHx4W>{um!=$Sm<>;zu_Lj>R*-R4-@&MW#Zl2*87_`GJ44n
zy^j;SqFCCP9=|I!tO-_>J)RjtwVT}$pa-^#8mXmu<|zlTtrw!wtKDwDeP7K_y((~h
z`@Xt1^{S`)CCEtez4nWnQ<<K}Yk$XcFw+X6E5Iz+Dsbx!P{UUpS%7gxHRxGQ8$6fL
z;(16^xxv#VR|SXj+buN1sJ=K)OLZ-9=TfTW5l^1L@=8rHeKcd37<kg0Z&L!aNladt
z{@6U)9EN3bn3m2##{RdgmcRV@pp(!b!H>m!lt~_7Tvb$x&F9PSrEAhJosX>lr=>p#
zO_a;NXea2tg=PhQ<<D)D#~}nN0;p+~4Eb5|$FeY&q(2L;PL>NbVvljd9^-|kBAvL(
zfZl5Ygk*fBV94ywHf4rxwiCqqRqX0FQ^hv?9hfRcQj`5o6=VOisRI4b|F!uc%viO3
zzvQ^o{D6X*`C(w^{2+XL-7BO^&EWe#nIA3~l^Ut17{xC(dqyTURgSirmJdHqPP1Y|
zbs;|RPr6#xc21`oR&MKP{;Zp{^X#7+hT=32xneFJ@kuZGNG3RcWQU*GF4&9uR-q%9
z$xcmtBxbxHAnShxVpU%-V~(DBNOBNmiDmLMtvqTpj=kJpJ8DjP$A@KvrcrEYFxEj$
zzghY(cD`YAWK3RfH>7FX<cjG(RoRjP&+^RpAKgo&<Nj*hm)ZNNx*uror|Z7T-e0f#
zfVrQp?QRZL9<8)7CJmlHV<D>ZWZ3LKk#w3WY5Fg<IZn^EwCT6;tXVx(;`Cd1F4eOw
zb^1*_t5;JcPrs4pp?bEZ&zPTVFw^^f`R}}q4>ELR&S4YqUUwpz9Zb)m3+PIM|BsY8
zCwO?;e$((EYAZU&7;upO^P~8ZI#ITQ4vO?Mio0K4U=(*hVEwmb`9Vf;*YRBxcb}Z9
zc>o_+fAp4rry6}ru5`>v!P3cddN`AwwMx=~Z7kuiqgoeK_sGA?%<+iT_@lKmR;h3k
zzT=z&U?Qrl!Oc|)3Mi5alRSm^DpJ`MCy(6+AB02k1cqR=@HX7RS1Nx@ZIpvU_&N}s
z$@XohB&G3`VF#7xD9tl<l`nc5nv90ehPO}Zh#};$K-4~yFZ;9s0y)gpea!ga0ECYN
z(a8jvsLL@?CUU(CCaSK4O&HLjP2M?mIA%SbTh-7JEMyT7c_X~{0K$biwUw89ymzcf
zCaa^Ng7Dt_&>IFQRH7HtJsgpBICq8K+wJzn&i)48$K>L(IW>oObo1WPT!D~&38OO8
zu+9Kg^1vzomy;E$;RWx4PfbI=gn@Q*D&GvAF1_pfsctPnsp-m1uR@5%^Q+Z(fNpgU
zH2_zZmQ{ym=j2U0G4W4L6wM!^kObKJ#R8rN@Xmim?~P`^d{3vht1(|)CnRFh+c#PR
zZhM{n#$FCoLM8Z;^*6>f*I2CT$<^C%v+|A&eT_?wTV`rztqA8u`hPCd6HD!NQeRS|
zPuz=7v5O^QPz?NB0TAq>X|0mAUr_{#zNDeFp6r+kFmSPYQQgrp1*I6qH_x5-G<=V=
z62A5}ypI{my@ebSZ*e!^om9bd7bJh3z5dYr4&hzLL_ZeY8{}r(qTwt>)sS*EtS9AI
z-bsx9%60f<Qx#mYHTxAO56#H8iWlIo{Fj_)Q}G54q7F$nEzhcmDP*4Gv{o4vy{y}e
zzz%oA$wC70K_uKcve67zD4a3LUR~;&vftpIma^aciSM%WCkI4>P|w8AQD84JS@4WY
zi6TR=KD-|IM2Lsf_@VyP!P_XWzJyjo1|>LE7Gw7)3C?zxoJGa@!BF&KvXeVc%Y@9b
zv!ALxuSKja%gPPIdfQ1etT#_G4C}-548vM>5e(}iEC5&<&Vw@C%6{Y!8x|nl-PG6I
z*B-_Fd%BP2-<HW|k<I5fHlIImS<+@1-vP>Ji{M!tf2T|Q2vLLSBv#>b`zv?W=x(nS
zSxuO%WH2+TBKtg1XWquiEEq8XxffUO>R3LQ2lq#SqP=GqQvL=Uf8)E0a$2iOs%Si9
zlmFS-B()b+m9|z51e$x}s#1lyy=nr_dRZ?_t*T$FM<-Ta44jk<sM?A~PSNkJ^LS%!
znz-?|E?Cb+fBwh<SC`fWo4F`|rs|>v?c7wiE+|j}elX@sIm!T6VbU0nPc@Xj$Nj>Z
z(rW^--<PUT8TIvzp^e`ksfMnT5pq|VT8J!UO}r*p)L!4zQ#BnJ_}CL>CK&LtJIajf
z2Ad2YUPJ{Pwo91oV~>>HPxi819BsUpx_h5Gcws@3tpsU~HGvM(1i;u!*^?<$Rd>it
zWv%k$=F~<n_tZv1g<3Dvq7y-GnW<O@wT#Si%Mog$vwIRJ=c*chYHMla)w&tJu)&;c
zi4Ov-vY}LX;(oC1g+2GZGeutkNqO_u^K$u2e%u=@6;*kkdd;D!rpZW@HEpUEV<ix~
ziZ#YE^@MT0aH8Ku%Rb5-{<;9V#dm8l!Jgd;bwpiH{nir8cJ}1JBx9hU`ZS`pELz(;
z=6n?1ar}(qi3QmU%5f$Y-sI%I)bXD9VsWd^tjIAMm*m)R`>4>NspNbm>nC}Zl*ivo
zCBw0E<aLmf?3YSmqRUyLpCr!6CP$VOIZO1D>~2z`htv>xD>|a5d}g*qMiA#`FY?|O
zp|is~@<V4a_<Gf*2k{)18&vdRE=vl?VnCCqCC1n#l3gi=F^I6H6x+ui&tyTAUxCAW
zvb#EsM8?L=rS*V7u!jXo`zypqVe|TKCY%L=0ns%R$DyTT6V=#e9F7WRZZ@a$=DyJJ
zf^m@AQvNzqz|27O<c6AV^x|ydZgqHOAYA4@OI)LKCUJB^ZZoZl_1M}=^&E7JSse}9
z&HhPDZML%O>00wGolYMPIz%vf>8{Fp!;~`Uz|5ORvfi+|9^}cqxh?Ar6&ci#F3O#m
z^`<9pc4pp;&w7IlJ?P8Ko0_aQy?L`a^QLds8)mve$;=yH<_%grbs?P$z&Z;MX{KSD
z%59TvN0h~P{;&y%mo3C&x$g6+_awe*6T!_)K769fhy7C@#xoz@(d9jBVw3SR_I+LC
zQ}Vtx1>&EC5QJOtp{Bo1g>42Vc^3L}o08p$0u+VqQ~Iy?67mYLnVHhRXFfec!DBkw
z!tDom!$rLqUUvd9gzp(C^BF08cD?<qm;I~*-UVUg%{@vKHW2Y>>Ve_*?%u{>hjS22
zd-v&wCXeNPw-8Z_&HL^%4q2k5iuVDzZc;@3Glpldg83`S#0Z)q2G*0woP(hY+<QyK
zv9`P4I|<g&I(!N+)cS@3_wtW5`oD`c!TMmt;WnC5dH`7UBwG~H%(KC8M}@-8cQ>EN
z%Z5$CUNBjv(8&z=tSMCcrC-3ls_QBs8I237)1$Fj{36!tRW4|$!I+1pC;yITSOK14
zxRkVoFe~Et(~<nX_o)lU*n8Hy{&}rcl+(abkW(BaBfqzXDKU7457(Mi0dLJ-*2WUh
zLaztllbS$i`^1;F7u?WF1d$#mpT!KfYb-|wjwPPTpce2>HGVlVFNY4V|Bi@_wY^=w
zolo1fu(|FyO|R@}hxOH^4#2`6>*}0PK3<o}GyEKPps54@tSeEw>+w$e_#-_s@7RgO
zE4@VjmsKNZ)Sp@os&!)Nqo}GTpD|n)-cdS1kARq~7RvD*J_{7zp%Ab=nbK9M7|Stj
zK#-&Ay{v1h;qjho8c`?`NslVx<UCVaPo<3a=QfJl%tI}(=ycAV?8F*M?<Ya^Fm;6!
zRzXeRCRKS>sa4FF$I?p~AX-sHt@0mZFSly09H?B5vvk@iGcLd4r;~$DWE6feH(QA-
zHAk-U6adA#Vn=zlfMf#@{^JDYuXGH6CIlf{f>I|iMSPLTh}sHAn3}@PvKcBJu-d(_
z1+?<vU@f3oKT2nLzA}SwO^ebkG=%$o^9IkuuQYt|)z?hDO0`!j)u7gf(*NaRTEo3M
zmXo@3RyC9=(As`mq6y0V;`lBtF%|D@$=5aGmX)>{RaworNjdGe8B_Eo-LQ1mExslc
zoDvG2f@&|-sf&3?bw@+#AE=y@YDf?6X~>_s?6)DBCf$A%5`cQjM3CR6YnHN9Ro6J_
zHU%K6`^lOv`iq8Es%?Qvb=p0o5IVPXlm-bua%5yRhj$&aoU^p@miGtiACoQ{DnFTq
zKi{YF!bB^b#{wqVYx-g5#zqrmE>`%G^bu@E36zduu}=sn`mOqzDhf6+GT|ul2Ng4M
zxDMt@EMer0297GF#>8z8w>Ubxy+Oy)2clQyEcNMQeD|V*elrwK7V9ziXwAU(3z*nC
zs!+Wc(LhiGw&5@x>+ux()!nBcI3dxHA|9@Wy5Vwbf_>6nl1`MJZuS7VF6X1|@a1u4
z79pA0%Oe0$pay(vmAWTqRh68AMe}fiQI;={DG^pV?Z&2=PK(}S3J-v6<)CG23DIMH
zQ;$`HB!n`&&t!VAeGKwTM5Lo6-j(ZkxKJ;<{*`xemOU%I8s6#tppQyL1+E<EY`BBg
zze;kPC9+5I*F=?a*=pC`%X?!frzb9DVlya>4f+K;;oW7u(R-)4L#LKuxfxUUs{}Hq
z1aSv5rYbVrLDelY+XqqoV2^|=oY{nTPo(C=CSgc|6a+TJRI@9>zGKq91;P%=@T)Ee
zD>De|X%Tkx*Wl~;WB((3<)W@^Kg@e$9;0=suxR>g(42F@)~!eHA6sVNPF4P$*n)Th
zrJdM{j56rk2VsXSw+L&?LfAp3o?Q{P?HJYlTOjORGnb0NcR|>948jhx2vbDwjDJce
zeQkv_{sa0-<Lh4x+AMrcVtc%M=~3T+ud~#quf-Q+wi)E_A8l{2%n_Zf%|hENrn+6x
zw%nwB3$*=FBbvJHgEo!e+)$xK+hIl@(uuDX5XXOkucc@rvhd{~mbj-L`3?9wRE_y|
z`1)T|cf!}3S@?RuRJSX>MBm?v%%-O}#eajZ+aQkr0$;x*=6@ExPG(u}b{z2y_}ZZS
zzZPHnC#*R*)e2GTvk-N?sb^P2{luhw3&N^PF^PQ;^`DH#i4e!v;;Y&9WTN7t18(y+
zOob_BCeYZRLs|K|FCM;sd>L;4xbpAB*T3iX=P$L$a<h<CXDZnhS(PU3TOg|>#Ub`V
zR$ns;54WT67;*dk@~_j1zX4xe=11ZVwIA-iF~CA%=@A)@zoiz8{ifIcF=jY^y-MiB
z*ghN|zs#BYr?TQD7I9`bMrxYuX=>XQaXBXKTOjUZjbkdd58{q9i0fey*VEcVI{8-~
zgXwGOYk&AELy41xueWNz*Ht~g0biq3!dLJmH2Ko2iv7-#?%yrG+Or7l14vnq+3T>i
zChc3`>+dO6@n7NVWk}<{!j}gnTo%5X+3oBOJM0_qRigHM1z&fR)@x$)eLa0Ge#+u2
zCylSP1mIwm)54o&+3!v5v+yU2ofhZ3TWH?d)iY9px8Y(=esiainyBMWMu#cJfFFns
zDjmSW9L2>4NKz7m3tSx_u8<x^fQj}v`c^MGV-5ti+5vCFfpD9;m_3}3u)fZ`6-~59
zUi&kh+?CZ)g>%V&_ZGHa$zC^LTjlvBRo;76ad0;=6#`q>AhMeCIAcAQ0%ADbVsw^6
zMwNJP=0N%!r`wn^?^F4Ec-he8ZCF8JPPEqxf^B>mI};$)Q9N^E`Hz$KlSSlzPC9>1
zt{%OW{9ji7y>2u4uk|*JQT`73kL=9<<W&A0Hvhe`RQ~-t^WQ+)H|4)V-{4l>(2f??
z<o_8r13FN3T$5Afy`xo4sbw`^Al>G^>IL>kZr`RCDF1pcqZT^dY%OBiOZ;Ce9n2PV
zFe}l)^m?em=wMc$gIQrx*Tg>`YyP8=xj@_MW4%0V&web={nYfl>E1~NItWTaZM`z*
z<oSkt$U{<<7G_S&3OLD&zruf0rxrYZsK!THQC~}Tbw)vN_V`6=OEh+fhF{-CZY<zi
zCIp83zO<ePZ^Nzp)=4O}FwqAE;|zIlEtlYrh>bzLYT1xFu(VDjmzMNkZ@#I=Ad+C=
zM7ny<44+#1KW%ELW1|}6tkX}$3hn+rbpMAeEJ&PBqU`--IFs-gP$uC++}LV@s9tJ}
zTm~z)@dKKSJ}Tg?X@~Y5_A0ikf-tchgQ6W~bhwL}NkluPmpQ>Bq~?xyNJ4wYz;)b_
z`fT`=v~IADtQG?72kp9cQ6`@6v&zIo5yq?0(w+5dC|#d|ev@lT{pf9+oy0VEGe4Ng
z=12pF=6xk!)8DfWHvJv>E`oF{zjiR%`PZOXgz2rYBKKXdpoU$d)#%Ud{Yc$6+WX<U
zzs}x&U-wtp`+mCrvAyr3`wQ*;VY>I*`|i3QX76`t^)|rX4^es}h1;#m&mJPChv)F1
zVZ&5a%Z8}mX;6h!qiJOnOlKssjgI*k^bzBZVZJ~Y(if1PXayrS%bA%hCw9$p9r1eh
z$#ORpF<IVjvmBVpav)hIXUIq!8#MSJgTua628Wk$S(2bI>i8FE-o1<BY{4Bp(rr0{
z9kN+%`N>xc4yOw~$`<@|Kiih4NvB8p-{eI;pc0CUfF7TKh`xL;PWtiPMGd7I7)top
zDXE0Vl(5H_|7Rua0b&seWb<SMrjy2C43z=S>ZK)1WsVV%oWZ6hpnm$SE`t^t;9fgU
z?Uv(~+Rpj|Np1lRG!&lrt>VtahV@bEZB$g1+eDh`a%j3P>B_x#@;;S&ObJ<)dvM`5
zR_?cqMG9ME>Nw~I1KewzDz`u#92<5gN$$_Uxvkt~R0Ib`%&)wgrFUoYE^!!KHk16C
zlFO95n~|FwYa$04(n;1OuTR94m^on}^%WeV%IY>QRk>eP6>``d-zChC!jdPiPtHt#
zYsuM(rF4ZxJz7jQ==DvlL%zgrI>UmS6cO~{BBa=?(kbq4dKRl2F^ffV2{|_=k4_$v
zEX2oipNb~#$Yf-{ZOVKr6=Y(BzC9~hmAo){LGtwEU|M``^0Z_{vNkz9**9619F`oS
z!(EbtlHW}pl`Kx4oa~o8H0ev8m^^{f`j~?kj!YhrZdyv^)5CSBZSsK5<SMHdGi~k9
zJKL-@3exy9uX|(~dRV%$jrUSn*#+huBy9J1yi8*q6D8~I){gUKO54H4T?Q?zII@CK
zSO%@S7r^flPkba2Xibi{l}R48|4RL;FjFbv_|8Wyj)RfpiK!3hgQ)<eTV+~cyJ4c4
z9QJ3fsaCGkyPTS2{5?cChwAJJ_xRsK&XIMOB#s1bfBP5aZ8Yv!nz07KqIK|Q?2*#V
z7#)_kV=QoU`A(d&HRoC$m*{7)76SXoLu&{`cC0gCjdR7C%Ol;dWOH@F9`<`jMz3U7
z*uHP}+BZZ=+xJa}ebep?o5BDJvgPcPV+X9UdFB%Cv{1(xdXl#~t>JpAD4<p0h1^+H
z_nsmXx41R@n3DJlM4ndjR)w3CfXdZ(tL88a8qqdV;0xh5xsqMG0Q8Ea0yJ{ER-~?@
zQr9V|>!Q^4vD9^K>bf~~^?@Ywed;=jYioE4zv&WYrrz9{x-L##-{eYkf^d6Bh+}=k
zEYF7?1F^qVkTv>IY5Owh&(^1XE&f{!4!ib@#eYl5*WtgVEQS9g_r-spg&hraz%8;i
z2+8UJDVi&msu!R%)%lnMcXia@`fygw!iUSKc>V0*n9+oI?^Cm@>n5GeVsj*G&9jXq
zYeq4KEVI(|Pa4^q1!h}d633f8x&v)?sauWqnzYtH3`))^p+`S<m*tBl!y?i^<N@_u
zYdD8;^k+BQBF_vav_=(USOO}KGPU8JW$kQ_$Z{D8yQ(r#KQ2PVXfBlrUlVVDs3I&@
zujRrpuRd7jL!}qv#7FuFE~J-}E~fc&_#2^3*+e|AMWmKxD>dHhtvERlw{WMs@O%2e
z#I!RsM?=!DuHuy@kVtsFY7`sPjj}cR%hcL+em;y1+FmKEM2EYUM<vh1ghL5Rx=PlZ
z-1V@nT_=cbRZiUU85@6O8a)(`{VdbyAK4p48|HJtG<vycEBb({?Zlq;fx%h~zddAQ
z^l)@8LP3%<cX^1cau2*u{Wu~=uY;wEQ15>4cc44=w+Vdbz76V~=-+Hs!)oR+;my$N
z=wSEMH>qeYTO+p1R64+ZG=-0>BsT#Bc7A!5z8nUO4#=X6JZ=f~?Vq=*Ud23+mknGe
zU|%QjYrkc}xt4vz9K16M?pMFu4Etd3fUTJpy4aSsyM;7HS5J4_iT#y_*>EQJ;KZf_
zw3|Fj-=p(9ry%g%Q!jxpcmSJcrd*<ZXHgQ|y`Yi3oJ6|z<<A$scjFE;N>8KHPZfGP
zDgDH3GVD0M)WG2$Uq63jeP(&or<uychB+&A8MI@r;Cj7%5-X5=>cyMGRcOM)y{(4U
zvBPqe*rG&;vNtu_U;TRm)LH+?uJm_9Z3|py(ko76t%)Sryq=mN9W>=*HkSJn2$EM4
z>V>KUNGhQxa`rj;B}LxYHrBVZ7r02k_1yukU}oWaS#lTOYHfY*w`|z8ttlHa=O7uV
zHMa=Tal-HIcADQiR1qt3a(uiSMRukZW1Pjem-=thKbuSW3g@LyoZJ-zCFh$Bd;{{T
zl-~p`IwkgF4j=a6oc-b4{R{RH-oDYK%b+F$)a!~9LHHfQXKdKJB)K!Ybwzla5-(69
z2!FUh2+!Cz()99AU^?+lPS5J27XVJv%j0Z>=%km1^pl~Ni_%Y`mml-x{^{j2PV)kw
z*jB&N!1ekQ?96>lwTlhANQExR>stM4CElS#s=twOkWDX;yVEIE(<_j#CFuZuqPy9v
zQ^K!8Ld<X{yus?6=2VK1k)_YC61w#T`(b^40j9xOO6iL^+dDc|)oW-H?x4jX_Jz4*
z;V)}{(7cUEkM!UjrAGL~Pf#8y@)ML#1q&T-Lyw~bKxl`(?khPdPHJTl<0pu?Zu|rd
zd`c=oa4$1szrkjc5#Db$*~>oK+vqWM+%1B;aS%+Y9L^NdGAf6ZGY1(cf7wO@=oALa
z&zxOWc(~qRy%^oWrfOL(jD%{yg^|<fwI8dqUW+uy%{-B*6w$`M$z$;^x!DzBw8;!t
zX3ew>7HyNJ5FKvsVcu}GnDyJO9o=(kybo;lhG+4pwg|T@bz}KAL33wCJ&#d=LJjy)
zjjQJ#<UV7_915`Bh?`WPxXu$6dmNnbvb)&u)F)E5_j^j~$x=?kWlr>H7})zBE1R~P
z-rMFzaKKRb?E>{~H2<t)G;wJo!FGbY;5}qr;8v<&>Zi<{=ol8`Fe4{A*;kEmj-<+D
z@M;~A8yj>rf?yyvY$+GMP7dtCzn91$4WTmuQPWQu)dN<ej_>gPqPaSktye2?E=mSV
zmq3yh3@g9eYB+3rN3gk?DB2b#^v~t>z_>{|lkHA%8MXHmRZY-Om3~V2$%$|IY&9XQ
zZnkcttdvS<wElS6Q0h~QNvz<iKlK)Esz!H(oB^S)7hzYfbt3%<igutT=SXvPthRe?
z<ECn_KTd2HDQ*BdE9Tjx9BRtl*r@J^#s0N~Xds6MUREP>^T`K%(qccUx1W4yKUv@P
zlUXX1xqeiHQ2Te3KE`858|CQCEz`^$<!AtSpyqN;(;v6eAL&n?{0*f|yeFt$vQF5~
z8qe`9S2htH>t>Tde8;O6JIYOh?P9L(5-_SI57tksq!Sjr4O@%I5n0#~<B0Ft3D$|w
zVv3OyNs2V;XO77so?FzJLZmx?H#82$LnK+Aj&Bns6P%r(FgkWl$4==G8VjL{bPf=g
z&*?X&K<YllKx9N(A=<5&D>}9hTCtS4pBhus6Q;F9g5vn){LYCG#pqk`+^gwbCz68U
zNH>K;Z1)@Q$3ALBS{{vMi|C=fJ3E$W!eV{LNccz#&ej!(ejV;P;vNtRLFfq&hM&Yv
zXF#*R-)Jft-df-sxW;?O%E$2NyV(ro`s)tN2`P>ix$I%B=N?U&H2D`zCFs7Hyz6)6
zGJwsFTW=%tM^5Fq(n4>;6PdDG_0@+v%~uVtkZi_=&9R$Sz%RT1D=wfx)UciRLSz;F
z5^matPa_&e<H5LKBT35(^*+|)2l&-ElRWb{47_7~WOF}D#IwFz`Gg_ZIel@ZyrNBK
zIxAFI^CxHGfLoIL5@FZN+h}@W<VAcZy^Z%NOT0{S62IWi=<nrG<cK%SoVaCqYUNvu
zw|?i!cUU*HSS)#ew__*UY(Gd`isOa8Vn>AVS-!&Ca6Su6Hj#PI^gU7Bjc8w^NvIzT
z5hAJUIKje5YwT=V=keZn10^SZ$B*GG7A<FjGSIR}tihn=?){-f1ID}HFty?Tq3vDZ
zqpGg{?*tME8l0d|Q;U+;s6<hrg-S#;fdFS@0u-@gMWsL$D+-S!fR(FBgwx{)Ew<Qd
ztF}H>tMyVZ#d{(N38)Cj)f<;8oN?r;LO>z!_qX?%$t2P~&+~t6J}~E;ecx-Zz1G@m
zuf28vmaAxJOR#978n9?m{%^suoPsS_bcAN^jEtbkO9d))$cd&`^Jy*&Oc~L^1*Yli
znHXw%*_F0!i6L#|YVJM{oD57VZzp#}V|LlBipGwis-4K%uK~%l_)YfF&VmW|Xchs;
zH<5g{?rBc^78_6SkD7yf#jpB$Zo!FUXD5?Q|6#Ly)oSD>ooGicbjLK_yd2NTa2vAl
zrLzumkydFr<YGoSbjJa`vC*)Vno61SqsbO}((JD~qR&|XUw>xOOU}4F6FSukeLo@{
zcJG#J&YdCGgax9~^bHT2a?n}!-Q>ylPWaXqW{y?#SWz#P8^@P%GBnSdk>_t_Q-pGd
zDOsL3B`-;TilkCchyqbdE>OA_aiyI!t`sI?%tjh&%tbF2EKC_Q^5R`yQy+BNCa<_+
zP9A#^OXxKIFh-=$=_V_a9$MN$8EM@pOJn!3u8oaH3Rm<G?R{*fl=wv=!QVH{3{~!>
znihG3eTG<fvMS-NXptl1_a>U6plJjN0DDcFH$^(gO@1DrbT`Rs<^lG^%5r0)G%R$$
zG3Ni;Ppc|6BPFLoc@YrHA%7XF1LxqghEIuK?WLdbPHb1T3{rJIQFYe(EYMIhxDODh
zKVzq?RB062>pdAK@8S&p*p~dusn-I=Gz-H;cJh~jCiew4cORpxc-C|((v~BOYI}~;
zw$j-CGHyL77$+>crTB|}2vniUzexa)Aj7yRw%?tG%NC0c<4mxG!h*{3ID9zTn`lht
zMl1HU5&Q_2KW{wT6lwdc>71N$sR@Ehc>H>2yd3gDFu}H$tp##4`vtQG3a8`nsLRnD
zDg6;O%t5y)bj4v+JR{eoA{uN^iQBqCGK*i5YkI1|QzWa7)+z)(+vWO`+M1>;6o!Lh
zQxL|rzbi<xME2S@jljEn9R328;OOPo?=RjD>EN~w7$_k9GI@6mf@SC2q@hXTK%bMo
z3HN1*1AV?KIWc*d?<zCP>Np|FMvbv9E%^Z6$(x|&xN-cHgetf2&XG*E)8bkg%xk)b
zfY9wv@Qcm2)}r_x%iuDdx2AGJB9aMYOv1u9L$|xi%ZX(iibl9kqhi}$Zf1WAL-33@
zd#$b8yyv4-d8f8!R!+vc+M3yuBkS_?F=}0rzQ(MRq==h;u2hJum#jEyeG%_t){iB6
zPQJ3tnV>9sC<~9V=Kcci;(9b(_;%T8@{eKE(#HqBcTvb?2_w0Rq<n{7<hX)M#diu%
zzQ-YUS;8)&`3aO2=@R_tVjY1Eq8}PGJs(gy2*hZJT0Tb<V|$o;9J+K0KPbB>sb5BS
zqQc8zraBnO(ZcdixtSsv?^G0`@OxI%R`%Y7o6C2}9em=>py(IE7#w&+?NO$%cbWvE
zt<<ez?=qW+{UhBaLM=zvwrii#XKoPQqEOmxvWwV8FRCIB_Y08#U07Xmma8|K<vjIg
zh4vpRpNvdnb2Z}=q*IAZ&aVsdg2+8Yi(<A3-C3?;nOGd<ezfTl;lr*I^gW`aFjf+Z
z#ES965V~u)QYBH3ml<k;6AdJP)oi{Eng$1#iL}iL#K9=J)W%1YlHuCQlh3F03E^4F
zQ%m4#eS`0g(l+oxpI7rpD^mD<X8j@AHJ2vI8%hCAGtkF3wk>XYs~98byYdymO_LS1
zWa5BKxPZI3?Rq6rfDWgzt?n(Zc%N9dtK}9Km(eLxvsZ)SZ$D8JxsbbApeVZiBs7HU
z;W?CRkK#J%y80(Eb4Bq&m2bc%M1UqHrmzP7H=k)3X#wG}$4SwAKGkA}IaIv8*9!T!
zb&#f&Y45bJ41=8!>mk5L&%HLgZQ8W%mqs(<3^vYS%DOJH_=vMzgOBnX!#<8Z^n!9R
zehw#yzo3i2s0n9k6nYK_H8h&Ry#&crl$=S)^1)|ttAX0|mn6v>_Y1uOvPlFkUGFp_
zCi<CGcXW=!QkW$WQBYil%n;O<^HorSrk-?yrK$Z?RzOn|v@~3C(TB{cYmnzOM#>g;
zGd{F#{z6suZiise^4L@*V*S`{F4fVmgMI_pZzgNgTjhI%G&6*{dE@qN%(|_(gG@Mf
zoM0s8z=VU~hGr9v{_o@)+l_I%)vy3t>5XOww+e(R*XyO`5QIE=-QEB^L~$u1tN+do
zPT~4cMO0mnj(W>jq*So3LSE)bfAe*S4dIHdZOguey?*xUNP1*FhBm!d>*~WM{6*Vq
z4?9uMB0P+)cG71C@!rhUS^COy+LrOgCk}9Cu7?Si@y2JaK6CYnDF}BPLQ7wXX}V~g
zQL%lmW<TS;?}`<5u`AZrZQJvnVgJqYCPO(v_u@=8_u@>}y*NiF&-`L6*Z=14ty{lt
z(%!tfx7L;IEvl<qH*@b;#@ZxLxkl&c93?-pE@Lmw-XbL5wQdy2KVCOxFZ6vmrv~<y
z|Bk3?f>!E@r?|qJbk@o>b%&-j?AA-|wnMXVvd6Lg6>!v4e!~ebR3Mk5yO}*=+`fym
zZY#+@(Hu&}gSzpjwC%;SGdAI-3!y`Y;=qTexIl2azxY#zvx!6SUmH;~mW6NJQY$YG
zhAeM1R2T%)NE@sBx;;XMU{|t=*ZT~i!niBG_b_fGP6*BaQx^KBw`Vbx|GX(waW`K=
zy}Y`^vx^*lI5Gvd3RS+x{cPe`as$pz4XP873I8FAF?s`i8TZ;bv8PCiw$CwpG8g03
zq<E>A7i{9UG|(=^I&@|AdKP_SbLVu&JO0(P>4UulGTMJ6dHlaZYK&MLd;c?2H7n&;
zSfBd);#N8E59~DIIe$G)JSXebb5aVpNCn)W0ubcVV)s%2J{ZH%iw>x7C~G^{5e{<A
z%%5n<$@?@VQurkv`77!goq0zjcihrVv{;(*XrwJHbXBNA%q*;XtewJTQ*RR?<?43X
zr<r3U_(P|ke2AsO^F_vXh`Q$x40XD`eOpX<n#B6e&f!^1Rf)?c+BA&$4!zA_8!<TW
z$FK{s2r))4na6t!t4?l}+QqF>+)DL{rU%g*oM`qt^aVLb_=ISf@flMf@)<Hr&Zy5c
z9mafy3^S`!#sNt!kpsvuvpS7BU}Tsv2Q($}$uVaFIrK{oev_#Ggy=(67c?EQv-ugf
z?`YO9O7c&w`|`w&BaVYA9pZ4`p(~{}LRVJ%&s#HVFwyCMBbyLT^wha}xF1@Ku8Dz@
z(f+@*Va9Dj4(tBEDp(gy(IxMuEGJrWD(j-3=R~}5xon-`8#m(pBG2(I?Jk}_0zgF4
z4>{f~$h~4x>H9dz>v#ijcKVYdflpwIq;HCNBb`Y4>WFvgh)8+^ij9aj0Ik$7#^N)A
z!#%g)G#jM2bV@jVRTvkJ-hdfALU(~c20g~o(yzLy^<}8eqkWz?@5~Ra4y{lYp|VCs
zJ;L{1^Ss*#`VqtcpKpJ)f4Hf(=EbvDG2sOqrus5D<g<rgPc1S<ZUF=jE8(_x&#r5x
z1E@raxXYAqjEHg*vHk4Y);Y$Qh)>S0Zyhnv)bYvLcJ<Rb_KYC5B!6P-sN<3u1=-fI
zpYBxCRjj()I(Ahu);UJrr;@P(U+cWTPR0tztz*lASXS<>qpnOwjc6Tpak5Mirq;1}
z$&7ieqcVvy<kr7YbU87#nWYloI0hyY+ljnJq7!r|vMXMkt^al-Z<nAtwDNY1<ZYdc
zI{g#MVT0#8dP1J((n)qD$_9V!5b>1^fW(A-nke-Hr;w1j#UbLn$9PFOL_8Pul8S6F
zGm5>7Na0Rudcl0YY&HZi=OyG}LcrsBwc5Z**%k-qzub_X<}XoibEwAe&zrTzJ4Gr&
zI=P785_1Ut!RwJ`TN>oO3UG||+~t`f&Rq^<Xg#~dKR|e_m!?*<|E_OCqx;Y&-nZ$!
zK<~I<#H(=!bc=1wMe{^H0jq#_ya?Hc+`puv%r?Xe+vRev%VW95{4SeW^ukyg6{CU3
zIk<9+_MfV(2gp22nRQKVwEwk)nW724&IFsJP9o{Z=A^w-lJ+Cm!lKs>%5u@t&dKwu
z%>G@U`$?DpinTqu^SZAA<xxs7Kq)d^=(4S?1<E$UW51;3^zH+NnclsJ;|+S}cQC!X
zo(f*r1D}ue2f0i;Bpm$ZW78DFu%)6pppCC+pXDiIWSev3XJ2d1DT(GhZR+eYGH8za
zZ#(L=D*YZ?dQWLX0}QaG-~aW}Uk05`XD_p*f8NG+_H)AH-?!!aAKBJN`z&Sr*?RQv
zP>9#3o6S=l<f%4OBeO=I1~#v4HL!V;ohgi07jy(Ra~Sk^J`lV&nd)~o)n9C?aZ0rl
z2r}#;@ZLYHx^N$JGRzB5KE`Kk6{RFh!z_wB(O|u2%+jVY^jPSX|0F&({%;eO+Jt7K
zQBr>S$k;Z@wq}L01#Gq_8vvAJ(^>?bPY`OaWbW7_>Rpg08LA1Dsm$|LvdkG{1r!t0
ziZRwrKd0zt7eBT(lQTAkgo*E9t5(~gv1l?}1>dQ$XL$qd*i$w-<~2tq0td#Xr38Kr
zTII(Bs8H2h@H&2+-=E?HxUxi<IGrZ92<T|q2Z6%b0s#m{#+<GSYh~!qb~U17Lz)J@
zg{9vg+Y1M~Jfnxq{p7Qv5E|uAfQT%m#IzU&L@ZT}*(d#Z|DP%SSrB`vB~Ua(v2iJ3
zserjQ86M9I3Z&Vhe=U*YN-zi;sDS(fQd6offsb>st7%9|cPl$dlnN)yPV~I8ox<fD
zHOOOBJ;_10F0zEM?{3vnT*W3(B{omM9%}<lj`&&jBmS+RFa}Rbz{G+zS*!+P5p^o=
zVt*mPieJzw3wjjk3zMdgLK^ijK??jA?l;QXl;s92rV^k?+l6udPJem|gp^DJ<&Ps!
zW#s6ioHBH%yrYJTz`69u&$&aGVsSB4*&qj9y9{PDBTwZrg>ED?3CC9z@*MNf$mcEW
z6*etzlZLR?K;e%3>K%|xrFZVwqhJ$0VxXXtGYx*-&x4IYkn4R~>F+c;Oslj~>(UIy
z9(WNwNSgnm*aCup0v)f&KTw&G$7^P&s<#yWL2~WL_hh|b^Zbx*&U@Je{B+Xw*N3Xk
zYLzsLqHWUSNy-@l1+Ynw-74KWT42-tQ;<oiY;`hb1u-QwP$BqrIlCI=fb!Mtmjj(^
z5T0it3bA8Rt3d0RrK5$z5=;XSLpRcf*N80gquYdhb)5W5D}r!ypVhrvs2AndCWcX=
zI^+)lmZJk-K?03lj9T3|)lf=n?DyIc@}g|%sR`BHB(fs43Y3J#KdS&dMDw!vB0N`t
zy1-3gd3-3d!C~a_vHB*_M4n1#5>4Tau(=9IWL&^UsrQd+LQ1`dg%Mo%R(6KOIF*sq
zmx5L(@%=6U3>8gx5}{8;bJ&+0lGkKS;)JTS>0<YdiuACn5R!=|(?C@3poEQigb@3t
zwWiI~o#s8jm;P*AUoyszyKWJ7DphUK)|x@q(H2zvqx=KVfaZ{A^B*FgzpYi^n+klB
zz++%1o3&Tu(6Xg!pJ^j)eu{Uf&ZxB67wUMAK3P4t+;2(N>VD`?5hY@meSAs&D74!d
z;lkxJw`epLkW=G2Lqjo_5B-4<3TH$0Zqr`V89OQbGu6uR_t|g?$OYR4MPa?Ps+*Lr
z;>(0vF@-Xp6w0>lAz^+ZF}Q6KGtEsQNYv+(GrgSiqNpxPaQ2cFKebtZ1QI8vCU@le
zM?SFNvSpwbR#lX#hFG&5`>tM^@(XLqB95o7M(?)$E?iyqa~%b;KtPRRkX{8DWSpz6
zun&24vdQd`si>Go4EnRXZKdqEjS=yi!NzxbIX1ht!UgT)gIeg9umn0*4bh<gz`+8s
zA^t;kR5kf02fZ}C19GNKs32w}ER#%@L=lXk66%kgY(wVxblCC8^siP)6zdZyqYA-o
z`QVJ7*0(!v(&Sq-{96ghRA^=XE*{nunG;pV0t1Q;q9A5l{;wTW3fCAvWU__RYX$qc
z<O;iQl@@#)x_zaP7TAvfX9d6_o{_c_aP{RrZblb3%K?@sDlxQR$vZ-^Acsm7DkKyN
za!!~pXU*j~7?ow`gCboL_9^llzC%}=Z-39ChRs-Mmij74^%%HCOHfCE^`m85h3~P4
zOv|+ype%Tg4ou3}rn6{ItNNQ(X}=-aBs4){oV4dF+a4LYeJiPNS9?(BZ}&&GXph;(
z#`TIFLG3UWZVui4jP&Yw#2hOWB`z0;SOO<HGTJ`@vIa78Brc2H`O18e3|VqiEY^1<
ze-tX?P}Q4!Xx54#IOFfke^yUWvnR&KlQ1aP0#Vbb&#O%=9#<Ma1$+Dx8CGLDy~V#?
z3T_;iR@8Gn>CUapd~5DC{C$D+nFf`ou>Jn23ZaRO3ZjWK{Cm|zH&=sRWQGjdswam~
z4WP$LktTdKHhadL(8L2{g#zQvAKH<(*%I@Cfuk51?{S`-%ToV`^W^*e?VIiR+v!x9
z-`BLlB2r@Bm@s1}MEk2zI)T@jp*u?07%)b}BCDnu$ga^o_Z|gY{Whlb7@~UHS~>@P
zn5G^Z5|botsXbmw1<gIS^OY+T38lw7H7k?Z?li%|brq)`pKp5n9eWV`dm`{j``BLZ
z#eO5$%jW$<jYxcx?Y4?mnO~=fLD<*3eI<KO$!=S)soReoq1z=JT6Vi~jp=siH)y=A
zw3F<=S%H0SnqftzBrcekcZsRQnsd9<!Y#0SETD{dc|z5`rS8YbxR`^8Dpb_xS^Pim
zC0lD{kx*Ren&WN<@i3AXqV+{5ql>*KmlzeewUel`mtzL#2&EG~>Cd{>8m|TeFFV|U
zcP)!_sLO2;<Mz+a>H}2ow8$i+U{;EgC>IWOcFz#$<1)0M<3@8717n6pyUuZH{+Wph
z&1oYXZ-J&zX>{gk+z1gl4Pp|>?>qNaEblrwUU-VNygRKZ?1tiJI9@sicc&Gxq6Sks
z(n{uk)%US@Lce_)9-z4Ck0zNCi;)cdciHed0J;OXbbJ%%HC)8uvvf#$au^c5{NGQ>
z!J=KsY+s*}|H=dCHNqO`)c?aA%);O(iHyy2@Z>K+#(%ui5*c@j=6JbDuqSAb!MC5&
zhKhb{_q~o%L5t96pLY&Z$Xa<fNTdw<CQbmAnwbxkvRG}-C?AVT{ovhy2a<nd^nB_1
z|7V!=qz48j+U^%jP&3k<<J6r$uWg2jB;a>2HH_$r`(w?)AJc~7ii-itf5F0SDRqY^
z<rQqQ!d@xL{DE8}Sre*y2sxM&;JM*`gL6Ywzcp*7nJfS-u{p1Q3O9LQ&6r669Gngo
zCtM_6@p^h%d7Im?jUt<W1cQC!eOBz}31t{c1``()d2ew_k-t-l0#cL~eq1)#B$Gok
z_g(*4<I5x9W4~0gHZ2O;?(ZL%Qqavxu@#Fe?ow4lkP%eMtHzN;P|6RL=4G9zj|MP<
zypNf0zY`Q&P)Fp`{PmxDNQSnSZ$YL*m1h8+W)EUt$}MsCNI{$neZmFrPrFVt;Sn-o
z4-hHb8mf?Wp^iJffF?nIa~v4~Yz^IB33ZDU;+}si`~RkO>06wTbI2O*9D*>)9!W)E
zO2<6+|H6cT?!<=jF*I7zhx7j*a`8%|u=pO5#b%3)dYNb9vOuai$&L)bSqF`N?xzRR
zS;hDdf)vuXq$SotvB_q*GxzRpk?Uk~#n!8<+T)jKGh_|QgY=z=+DyM=$D1lUK4UT$
zs!WQ82=z&`wZ^|l=NumO79{G`$UsdcmT?Ju)mjRO>4%2j6f_~)=VKa+M(J<lChx9w
z=+&;U%?u~g?6NhuHq8p9`GHMyMlwydP4hFG=D$gU(Jy2`bTdYOQ0(mrM`xajURqt8
zuo{&G?L*IIL2B-zK`as)&8G+2FGP8QzS00=cO?CBJJ^Uj8UhKt;>GqjLm=%WFwOtj
z!=RZE(?5sZP|4ws&<_5J94|0Rd5Qppa<c*ik$%S~GU$Cg#DHGl>0a8_Z6hR88oPdP
zWG1R498+iH`%Rqnw(EB`FtU_L%heC4Mr5MVJgrZJ;PUI6-xg7^OZmXrwZzT`NZC{k
zu=Mgs+j8v1GvD}BNYeAQ^9b?}00Wlf%oPswrH(2~kavNks<ZNburx4~x;lv2L=mdh
z*VDxO#l&b0Lhu3`9I9NzmCU|JRO}AkY<<#+&LAk_erBfR>CCGLdoJE>>{N2CHpRW~
zUqOQSdK&#L(`YjvK)gN;pBvoz(Psv?eg#EsF&|p=1RhM$G5rL(sEf%4W}d{u>Y_$a
zK>W-gwvgcX@xe!tP3rGA6?G@wD~ty4JN4zbz}Qv-jIdE^2pqn`+KPPSjtui1MF~f^
zpU3D(nzo4s=7harB&y6d-=Gi*b@L}ILp7V`-~Z-eT}(8uR*JRMeI$c<Pw%w@)6*b^
z_MPr0#Q(kYeqP7;Cp7GwK&R#v)^3Y)>DrgqWd4Ab1G<x#M`)g|X5f=kol=AAk#?+7
z8TK5$W)Wsz4;cj#6f&QR+A$zTa69&nTKl)u(S!c|DkgPDZPh}7r`av<c@Odp223;c
z-Ip&-lEHC2IFT?|N<qITD#Bjy{!4&ThuMlyOX*qlHzy@*k&>F*-t&S|lQU;=Q0g&e
zPVbabl|E6bl7<VvqR(qhr<(kgJY#JzLoxG3tzT+lEdCBT7XB_FL418O{eqPAHytbe
z^(K9Qd&liFXgX;x4|g*(^ij~ro<_QsbXo#N1O5kHv69<|{zEGPeNyUZdXX-r`Jo#D
zO%y3y$5DU%a|>c4s{uE&`ZKfF2NoLrzKH@vuvD=BM->1q3gkQgycKEyS_Bk!q34UR
z5Wdfx4bh^!fhrf+gEOLWt2HdF-gO$e(j)@J(75kt_f#S{%K)xdo;de5#OmI5BWn%~
zLHdYz{rqRy=m<UE09M(h?cHig#j+{DU((a%`h;Tlq!yj+D2i>d?pfNI!&iMS<^fv(
zdT$6RqWyQbWqL&W^i#C^F5%DE@EXGXgLI2D&*?p`cQ5n4oto!Bq|fi`%y>M^{tugE
zT^OAg?qVTE5jhm$e`)jHs=QCyyknDjud#V=w0U1<NMHD+&AX1F^k1`i`z!BMn>Rn1
z_j@*PX9Q~MEl&@&WvTZb<*l%J*Gt~MK=1A{_Ag>LiXfpsU@o;D=;Qn*kR;xgXz{`Q
ze8lh-qvWw-`vlyv{2)k{8!<zwh##bn5`Z1st)C(K8K56+qsGLO?LBFUYFQdq)qsMW
zYT+kWKfmWQeqt<JX^)S0h_{cW>-Ar>HdJ-HvTxJRX8o{)NQ=GAPtdzo>?u_Ft`exG
z_~^uSn6$>qg$d<iec37gVSp7sDLzu&9;M{7Qj)iZ*#!R$CBK?7e;gkbRNb2Y&4{t=
zP1Eg6z`Yd{tisn?T1=9paVOF#9x_+08>b|(KM@>cg1uR}svJrDYs|NNpaNUfTkf>3
zt-LgTxqpg~;p;_|ab-df)2f`ZrT&fv<!RaUvg7=>3AXqGA8>SLtw^Z?rtvhf$t|gm
zRCW{^;{VkU9ZSmnj+5wN)XFY?8XvlUZ-#0kP{`#Et>XitdKq&cd(+Dt?=>B*_ZNTM
z)WpGaHnTS?h1c=8k2E7&>|PEB8R6(R>)~#(yO+0t*8X4K2zD>8;;#7?`k2f73zG%U
zwFN)@&#x7HfC)>%JyQ$*o(e8aDcI0=Z@%0yEKBZh$Q4P&QRCCXpIfi4E$32i_xPod
zQ;d9R=fCt{@gMge0x-~r&(EX5j<>K8C_CN~`9g5o&hKj;OXPgRDOe4!&FYbsp0;dJ
zi~YVC%S1iDpo<zlt#ACt1w%fT8dUz!zYi{>)$i6KjrFtnt9eRX-o#UXNXUw!=0JW#
zHFvI@RvV;5VBSh(q_8>?8dA++I2ElT<c!+ysj1jSOv-ThAj%Kd3O{T*t?pD+r0^gt
zJha?4P9?mkeHr%G`jJ{jMcu_Va<%o8m&PA1G`hEF)Y<>8QQVngTJ%Rv5YA>KT12@2
z;xa1mxC%m>!j6^`eN_uhFts@Ums<LoTKayimcITdYH>VHpC=$BuDH-ZLi`?MgG?U-
zSTn296v%Fg>(!Ccv@JUeS=$bcfr@{HV8Z#et!>aHv_niSl5DH-O~|!+4|{_hg`1)F
zGFOrs<IEfdRhy(xTj;6eG?8JuGPu(;DYH-O#0*{|nf%8YA_|sljDgqJD;4fYPp&_8
zwL2yI(`(e;KU+kuFc|DD&INVD1$YVBhCy!|$KBvy;a_u<%l4kjW#}C@nq?l(<>*oH
zdKrqjXOvfx>*Z$9UNW=5Gv07?^re7<SclpSb+gYL9-VMzR=5Bq?xg&jvM%J*Okpox
z{!VhqT$=us<5d`^AfBhG#dR}XBjYD#JL8M8obiu|N`&3){9+2o3Rl!-n<H#k00H8c
zHfbtqf6xD$(XERU(C<a+v^HpXzen%<F@GN5fjfau^E8L5`Dk8B3Q~3{`K;Lls?C!!
zCzvQNK61OPxcow(M_W~?@jSw23@2JRwCi`<M-1(l5^*1BFa_89U$TQ>4)oTeK_Hv_
zwwP?%2gza?hn*>&{Yu%p*s28z*~d!Juz>s#cWI)SXSET*_|8?ah>sYrKcyCksZ1<&
zx0QyTd@JH&R0<Q46C5Mw*>>&ydD;j+{}LA96}4cS+1#-Os4Zk<5Vn#BBi=0%D_NMS
zMUH!<)ZI(b%Kq`iHoMtC;Y@(k!pxp2{-j>$h}cgSiGVjjuKh>LNNcuKlw_<QRuUU^
zm!bftrUQyDD`kg&O(1F}1?mD4P652+Fx-C6->R<_dRNyqINHTBkZEgovl3&Y`KbK<
zSHCrt`?cDH1x`7jG-X&+H;Co5t*~gi*dTt}#lplCnvHA8>A{UAqSq~!k(qVxi`Pl4
zT}I<!ajP-@L%h6xC!*;jwp9=>3qfZ;|22)5k!K7wInIB*IuHgN#>kMAuH4nEBmwD#
z<R0hWmrNCa$$!_BgugE+!F4V9$wb=6bw}`%i1$Oz20)fx&aW;Ch_zp>oP-GX{egT$
z5HL^;a>g^__QusvquH@JRdPq3<L-b(^cT4SMVK4e`NoHo_Giv8`!o0%ztfb?{@+Og
zxN1Vc(5}qos>;hC3w*$w7O`8939SpqM|2X0D64;A8vW^Y+d2&^`ux2-+G-h_#tk&^
zF^CM1ww^3$*yq4SfXlNMqvfW);7Am@1Ml$+l~Z479&K`DDL{9vYASU<>C$NLej->C
zUV&~_)Ms(DZ!WW-e5VtAMla3qB7q2mfQP#P-+eq=H3SwlOL@hjrk506UZ2VDwjlgA
zi&bdoJwf;lf_s^3NSz@3w#WcV5Prh|%%XDftExZqqSlESPoFaVX8=el;cs23h~5Jv
z)So&fmg3KT31Z|uo^L3*XZ$!NCWkr0)kY`0sw&Z?l!BEa^rgU1<X`H)%SkIPfuQIV
z91>}lez0KGwC_aRk0a4PTV}A+zvn^q`J#MIvc^xGnsaG1`!(Ig(cW=A!(?=WtM-f!
z`=<fq;oe10K*P-}^eT$XWYl>w+R_-RI*qmrjZS)kql!&sT{wO9is_KM+u#23;igNY
zlU`v+2BmqW9CWCIDX@1u1g0S2WQJ2{e6@wD&Y%$5zEtBoegs$EjN<TFg13ONV=-FD
zj*Q<HuoC}zvsWg)BtV?23PC8ni}dIZF0zP6l8cyOq#!PUWwHJMQI?C`tUcO_Bo{GL
zGvFehy{PeR!9|Qj1^?Ad$??~ILITP}NnriWBFaCZ{ibx^@jSyudV$HMTH)icg4H^j
z1<`HR8aZLt_kR@>Fhb>~vZRSBg_0-2hAj^%%XdogJCWca6S#ae^dyBrw?E-Lt|oLY
zMT!%=*B1f{BQe{6&jcLgb55WbuEPv164lRz{gVNEHp*4DhmzTrQ><YiGGaKF``5%i
z1et(G{QPFI3zLMF@Ar>W63ae@6P1{tN%nEvotAx=36RP@_8H|_l6`1Y1e+;N+evht
zDPibWKqLHRi%hk?2cs2Jok5?oT~BgHq~PPkEU~i5|HMLc2kHD@sy?HQ6q#<xMrLnF
zu#uV18aCp-Z`eq0jICdo2OEjO`e<$D<;jGX*@R<l!tW{Jog|F2o0@rkGS#^@)rB_I
zca`csrMi?<9g?Zq+f<!xs)0(iQK@9N{7-g1EQj0njA`2L_e|5$mFgK%!Qrl8>iGkf
z!q2`H@9bDEX9)9g{`)T^xEui1kocz%q6O;Tnd{dr!rvOAS1z1$v{9_i#6$NZ*QTXC
zlg>s_MQtul)y$$b{~zS?I$lm|B3>_ZiaBtu`IoJ%L$Gt>#HE~Tz5z+|5740atLD73
zE&TqM)piv=l)~BhplEzr7O_6A_%vq_UUJ+9cMXiw;qpIPDJz9y9&=xkb3#}UR&OBr
zRm=R_q~|)PVzp?8ThYK{xp5<zo=DMhecWQbH<}_#m}p}RCq=di-CZp=1;453+YqYy
z3SgGFZy57dZdCrt=)-h-n4Ar=^VJ4p8DvD}za=SQxS>c`iFb3hH>|TaJGaF3@nf*B
z*sa!{y@-&!BWmwV)al?sw<buDSQ$iU%#@OTgY!aFE2-VPxvO_g2HTl~xmgcSbR~M_
zt~_2R=#^{ncpYmFHhF`Q3oSJa!xqPCUO-p!!tm;ehACjrkuD%r8Y@Uhq}CeXl6?`u
zvE)&yEO{Kxu9TG+plxmLkmP2UM_B#!%$s?Nf7ht5Fb62><UeO#%ClGqjy9#DIBihr
z<HzLpgf(cHI71S(Q+BgvFRRm?ub!v`Y+v>LI8ciw>F6EWOb$JX9#57(Ad-Lg^MpkE
ztdOP=@!&U&q5_=C3wzVi{vsjo_kVh!_I7JIo-Qdl4^#R}WFJPV?%G2`*xu~#`bbsP
zMm!_9<L-fz#r6KlRL8q-2&$6$XHcCfKT#U~i?USMomLFysD0>3F(3f4*YtA~SOj0_
z>0q8GHk7^^A2`JNiq#dAV8B-1$L*V@L!GCYCwEoKqPFhvkE$8ZjvN1+j+N~-ykn|*
zpE6S76mkda3X0pLP+htlE!kzpay0?gD0D=xwXrC`Z=nyWuadJ;4M_s*uy$iKNSMF-
zkOr6ujlLT|0N^|T>|cQWGC*ZVoK}H$emi5MrHN0Hwo1~OB<#zN&_&!3?GcxHqq=KQ
z$>Tp_e9;A|x8gU*-*G^xhmfkQJDiYI@A|iJTSsy$ezQ_lr=&s*HmM@!PLH6hKPc7j
zNflJIrNt(^B4;Hdg%RCS8!}n(SEp1Vyir49shu0AR2L*m!EJ=tWnaG*gcEWZ2=q~+
z6G<dMR%=8!*j!QP2+K!A3R!It7OJdnr#Xn^+UL|b<POA8T4-W>Wkp<(c3*#RuNh{F
z+U;LHrv#rxq|gjC&*ThMJ~KlNa{CS}(epIURV!w2ar518$SJK;_&U4yxB#m(pUqxt
z#bmpi1>i&0t$@2D80K!6BP54p+9!1K+%N3FNx~T6c%7j!Z<wqH{MY5jiq*G)$S5UF
zOC~PMW#uFTJ&8Vj{rQ|kEp@*#i&J6H1xmg4a}`~pV#{(VcMNYPk3X8^$&z!F_AjJ0
zk>);ZWF?M%S||NxH&1K(#b=8=X@Q$qgO<44OWc^bnG(Bo@K`hxOGiJ;n0N%jylMVZ
zF)AreRAPt)<v+#HsOY}QqW@))&lOqXsF`LVGEPPA)@C|=M3Uo@X8M_{CK`mAnhNU+
zmAFQU1%U(V!6Q_T%!r)kch{|sB4b*;d%bN<jlvt#_%_O18Pvow&gvZ?l11#JBr%l!
z%vBcM_<(%fH|)Nx>peuOoM`shNQZA4GNpO~5+TCodp%MHW1SHH+li_i?;gpQ5+2r@
zovk_~RJv!Dr%M&(zhc7c71mcGtD!WI+Qt^CmXM0d>+mfb^z%ajiTQ-A4YMe9{Zint
zm8KYKB^;W81mWMc-_imZ!2o&9`P=f46UB*Xyb(f{`%A}5bVRum+`GV>zg(##qAd4+
z-yEya>$UcUHe@Yc%6nR|x!>4-`1}+Y{0>!u--U{UF5Dv#T`*XK?wnrGQ%jx%`mGeg
za?BrOgk;DKCx=nXB(PT_AO@l3z}mBHEAf+Zj3(sMgk;}OM7QDp8fBKPNZrm)C52z?
zApRB`56)X&;-?$l7fjgoN;14f5f7%)!4-->KN+vS2vz(Z7o@b~2_-r%naF>#Fa=Kc
zDsq?E(>Dl_#Fa<cXn+)e1@_!d3skt%MAi}cc%U**!KTSXS_lh|vLY^yXCP)119*!n
z;MG)$?M1XzDDxI@Pl9dBKWNcaw5S<uyJWN_1}K#J;$!0bCF74Z41YK`CH)GeKi;Ot
z`2_<zLk)2J`C6lhxqK+mC6>y<9#Ec*cv}$j45(^ZR-Q@yt|)V!e<f{7lD|8Y{$bL$
z<QG*7TJVdix&J$UK{fs<8{igE%Pm3+-p7M0gZoxXj$uAy)r$M(*p%$M{ts+QBMT*h
z9wyk+TyBgIU^uw?k@#u7f7fmS@`@sp?0|nhUJk{iqkWzV65M1GOox(C&UXJOe#?x3
zL7#hr1Q8Ng%c26ykJVvuTm7}-yxnaI!+9ytUyXR2=#WTG16lqh|BW!!7ZsxdGv4E|
zk^t;>TL}r5b+~`xwV+m%ZEY$i2y7bdubN31B`d6j>h-y~YF%>e+*k3elx^N7|F`qL
zY=Y#DX=?=14_xC@Jv_cW0&b~yQN7MDFnf0UM?Il47CCj&4b5HJonNdKn0;s*BHkkk
zb-cNOQYlgpvr44{QYu{nukuS-!|UqR_|Siux^UmrAz~<85le=yi43JvbB9JF&mpEX
z7_=`tfgKE06wfr@G}XW4Q^vDq2ihp{^?1>=|2a!GSv0MBHmz{RuYl#Oa-zK#1ciuD
z4*e^0(|g6Lv#R=1A&!NFYct9wK&c~v;;b+>^%N)i2PKomry9S-zxFQmwqN|jX*olq
z*-r*av#1Ku%vuz0Rrg!A?vLpJ^xzZ!?MF#a@k$oy)5^}Z!rM!54K1hA%g6XnN^5#f
zW&Q)iJLy&Tz*w599_KgyQ=9l=t&7m^c?^%!VV~USm9>}qyaVxetTfBC{HRj#Hu<%Q
zvdbQt-Oc||%hDeZ+X6{?rdP@RKUP^<4!6{UOsh*M8_HOkpcI(*hEX=`-GZ{=$C-iR
zVnj04Fq`Tsn`)m@{Y<H(xHvnRs=H0q$EMn>RR2XP1mizGhO)qY*ck6s0)({I`hfkZ
zDuDz3dDz{3T_rG4s@Z?95=fq3D89mKJkXcvbf>+@cH<)g^Rjx$1B|Rx@qCLfE1-BO
z$5s#x^zJ`yeAsO9E6EaQ9Ez_gzIK@*;9LBg2-bS_^2dYorADk+)_i`Vg+4+YR;o@X
zs!dfE6Idc(*83y^$skGE96|kDM^Gv7+5Z-)V@pVj(2B;KTTYicrMnXaGtiZEp*~q5
z5#8PJBxyKsJ-w^!G;?n0ID2lXG)?ZXMPh}XJpsHh!cn%K%4S?obEUgv6zxmkyZ4If
zgp+fr9s<&d>OgzIc_c%14mPvr!ZvN$_(7Ei$_flnbrg9H6<DAOiOmhc%1P6Zi`jxu
z7aVngexL$Ymh4#;nJ4PRai)K&X(B${?B@x=CswgozbgG#e^DQ1QYsxmC%Qaba5T&=
zX-Pk>ItVRn$Dm<M7pnXLFYKCFPle&oz~#n`blxr3vLbW}4m<M>&!sQ7uq2_O+qU*v
z8=UXL5qicFej-b7hn_QPiM*qaSu$2W)ECP?`eK4Rjap1_r!kAi;-e~0KB`uji|6VE
z9qc0xu9bs(2DD;F`?Awo>ZeM)jD;m`hUYZZI8Cd{_i~!AedzWmIjw!E*Lk7a+4Dv%
z%y36}89cLiX7SACnawkYCwKiV?98(>&s?6lJiGGjYJG|{l)n(i3MccPn#Alr@hvmE
z^LX^C78tu|1^__}@!9Femt44Jcbnpa$TRgiY-B+FD^Rcm&qniVg}2RwgZ%h??B7S8
zL1sN9JY0qi0Ya7;7Bh@LCaW22f6_cVzKpnZ5N2zouVNv$c_NqATqd^eD;X#jL9ymb
zZRMOZKTnSuzwHgxZ@u~En<x+G(uy>1NR~`ulJnaelHKB){L;~RMOox1QQg6Jw*A%?
ztu5MHsEQ&5a#q^*h83!W0|}8DwlMB$qNZYL?^aZ7Yg^HnF*R!t%9jTe%3TM8dcDV|
z9k1bDM(!{-V?$7C?Yv*)l;q~sKKP4H8NVLxmE?9G=9X;8Y7y%ww&d3tiXNe82EjDR
zW}afBvlKl;(YYznmATUC+7#JgJfwx^EmN~f<!fw;gpR5eWhHFW{CZe4dp!|p!=u?Q
zn{FGpprH+|roCCTmwQy2tIV?PJ<*g`QNv_gQ@&9qLBJ11n72?6jd=6Sh<Gnpu+=-b
z*u0_1fUQ0hsTbI?0H^`m+Z8*T!S*PvNC8{R`OpkMQ~bY#UrOw;;Fl78Echj({}=FU
z`d9GV*%E%qGE2S=zl}-wfq#-QKt9Xw__km)*k>#>lGwLzHr5u%Guf`=Wi)Wv0`^Wz
zF}K7#D~a#AF9lWxo2=dT<=J;gEspE#%n$HnQ}<=3_WoDT%m316vYqZUFe@CrLmKmN
z^sYtb@mmQ6;pp>{Jn&I>^K$*Hl+Sm)tkBQuy4a)@wf`*dTg{~qgB#=^f872f1;5IM
zgG;-4wEK>lar=(+y)yF#)p?}Tz?@?5jz=hvgQA9KhDsaUp*?GBelhzF$u@O|rc`em
zw|`b{|IBU+$#H1Iz}`BQQc<fLTtrM#=d<OPqqI%XlWQW>$@z&+HnL+uw^$kaU7EW&
zhHo2MTl3l3roXyQ=M9q>XJpEdQ(F^1`wqdt`WYuE*8~Y3$sin|kxH;h265r$tLhtF
zR9}iKn%WqXeSC<1S?oRW6k!!d+Jqu;osuox6!FeD{blV%_s+N8p-Q<CPg}7pSzf^>
z9IpR1>f!FEi8Kp~cJEJe3U=e!Ta?U%=3@5#hP*uXy^H?FuUTCezw#Xlc-rPHaSuJ!
z)*nSSljFU!KQOuS61m#6%oWt<#>mvIIn#S4Q};xsx1ueoI%=zOH<Rg;<}M6uXl{F!
z+0^MNwtzfzb!-{QjQt~5nbkil0&n(g0)e8{WnWim*bRM=d{%sUu^I5O+}rbbuDz(W
z9X<|l7*)k5yk?qtbUgV^nZTj(^FO9Gzo;MnM7=)0dJ#4k?Re{di^q-0FEY1W!(Z5z
z(f-sZ3W|1olpYl?MPVydj<HJn6Gcdh$%(j(kB68dkp=gTE45VVOi(AYQ62BVrgEFw
zni-wihhJeMTYuG#Hv*OBDB0>{IrT+Z3S3^4LzAVqNcR!h+Jx?SAV+og=}p~^yYoe5
z_OH5E)lQHd$OM^rq}Cf!k%XCwd&ZATRGdL$Y<;K*Th&*l47yw(+JD5!p)~)?-;1dl
zLk$_tz4-@yUX*XmHeNMf>=8#0!q8;P$ndX(`I#3QP!2{n2C&Y1=$N3Kdzk^{4a)^e
zPS++udBp>gg8#QrO!Vd7!)^+@Z@{jxE(Lbo=-W3Bwr_t6u+{#q1=xObUn<z<n=gUQ
zKx_t|{y+N~h;?3SAokUtjtQ}0BMrnZp%cYR*{n)J?2P-rd5B%gjv^3SUYi23UHAMa
zAhxqJ5c@Snry$yR@VXh%E;nCEhz;h`uercLZ2R*DVk;j!Cd9h_$Uy95I#F!yLr6^5
z&p^{}8prZsu0ZVWWhoGQX~};AV$Yof#Kux|3dDMunVOiIrw8*kz_HGJ`uG3PK<ugK
z48-nv;Fu8mWVnIY=bVFW39%Z`^qYp*(c=xo#!*?4qPf5QPe82VL?Ct^MW;Zl$)H>U
zV*3IF3?R0hPk&Y)1F_l98i-wY|1lx9beMtI2Q`9NDoa=>+WZaEu}z$*2FH3*SrTGb
z{N_Iau^)8;Vqe|e0%C6&luJPDwEzJFh&{!pe^qY-vB)z9V!i)(Oo-i4Y9MxhwIG&C
z`L8w<`J1Kuf7h;lw9nDkQXqB)-TLM!f43YUwsLU`h}~;YE&;J<fPewSX7lM6{=h)&
zl)oE@wfVy_AvP*vAXZK%ikE@{Ny>i?jOLrhu{)7JfY_S|<VhU+?AQMZi0wK7h}}cc
zDYNnhu(_G?UuC|Mlt03!fAaSY#QdiW#8%&XOo)ANsDaopI?)ox4uPiMG{mkm&O!S8
zUN%q($*ksA{|Sh_)DehXN6{$|yTDA+1dbI1Q#Qb{Q~311=xHGKJgPzDu|NL)m=N0=
zHW16F6D@J99yI-?Ayx#T1ji=Iuqy$vg^T_Z5OcGESZ|6>fmo)QqzQ-}3=lAYn9ryG
z&iMvnl}{LmP5Rw2A@-_cAa;Z`T=CNElv(*3Ly^B(I(7;xW+3)MDoaA_>V^Lah+T0!
z5NkuxDG+<dpj-lCbpZkf5PP0a|LOA##76$jK<vVMjtQ|lhZu-$Vn@2=tQ>DB@;3{y
z&siY>u?#9pLhPKo{u2;8<2WF;I@$ta4;qw9Kx}b<fC0oR`Sf2q*FY@qaRafeCC7x=
zPX-%^J%KKyCB%Ax!{79*T*cM_5L?T(LlR<#?)*<c?6WK&_D70N!Li8(<q{D4X@GzM
z#76Sz|DlJ0*zUg?h;8`IF(KA{kbzhwooFeKHG!u8{~;E^-k2jxsA=waOgiOgpFd+L
zkObFC?>_;q+V+5JGDW9=s|ZwXW|ZGIUr9!Jy7|n!lYFSg#wOM6QnfEpVFRFYGOypU
z8!c}%bE*=$nPc?eLJ2+UGe`2!-}21M^yYZ|Cp;?1cNqj(^rr~lqy57~`J))g*uAEx
z{<Bz~`p>AX(f<FX_lNEK&-8w;eZNxg3+;QU-Ye{TvEHZK_da^R(Y_bx{aX8es@^ZN
z@168M#J*?hy`OzQ#DdR1-@fnC`|0+5o8C|6J(i}iv%3t!uu#=W1jNkG$A*%d4G0@0
z<EfbWS=l=314WsiCtF9ot|;?!ck8H!6{Vl}%oc4}sBrzn*<4OGp!~M@>^LWBn*)9x
zzbQT-HQ>tln0T+$04F{)o|hVMLHxov3ZdpQ&WslXpLraJEJr6~W?rajs33#NCzyDf
z1v$1?YBT}g)JLHIc`x?emM|=TC%z%2$WUch;{D$V+9%qS=0LhhVrTh1u@C8NO2GWs
zTP*@YmA9}95tDURbE;6~)e8J`t3XGAY`M2c`2zw2AlvR$epewsPc1E^kkP3jNAPbE
zADkKzR|tw`;1`s$Ss~v`4XIZM3b^Jp&k^F^^Gm_Z!ZTF$I6*P}G^ayrR+dPkeXSAf
zYlW&fHr*7{PpUq9X|?*-)@(wd@l0#6AN&Y$c89f7d#8Q~+CtKL1GrSAjpJfGAF5m<
zj)-CDT*Y8(IV0jtX{>83;5@E2WJeIp&P<`>-N3yhdqR~cpV2O}rDqO$#?M=16_Vb>
ztVq9!Iibq;Nnh$-jzQYh9M~z!%Ha%Fj>&mDIg7n^QRhg-q4d!FX^4BJ-pOI_+AQz*
zLEPiy9*B3N{eMp%RJo$8(3_p3yKutp-iW)r*j-c6)ULe5o1InBlv(zEBsw_Dq$}TE
zI4mP{_v`U@ljq9`n3mCNZT!{3VcDU@ug9M^+d<9y8x@UhG1BkJjgSKv<cixeu(KV_
zoBYYjio=QY!v;50DYR+s2z%N&v0h5%<%A2?J4>+}(BGUeCqXKAAc*P4teT^`TDf^k
zcvzHO@Pu=$_Tj=eIkdeklD-$uR@=;7H@~5hguQnJm9VV1YIbJ$OBQH`)kQpMiLK#W
z{ANMA$Z<3B!v%G%8X)V3h`Y<FIg;VPhm4JY6B<|-iK4e{-2l!PB^xl$^m0R@-fq-e
z>h6zE#&uN9j*Ls?iAyyOb4T?_yVM=ihf`X+W^*D0bDDQJL99*xp@tf~8}WMJL9{N1
z;}7PHsoD4Cq@TOde{{Zjb56v)F)Lj2QARlZ(}+8s{dJBp#Xi7!Hf9esA7@0Oy|LEn
zi~@R0cU9wfUGr5{z45E|3APIUR^~*6Lq{M6PH{{DQl)QU;`k+|l1;KaC=!A?wkpmz
zO+XQ?P3YKvQ5P6kQGKMM5e$?ti0O;>a{gSP5I@uIM~0)**m<a2Tb9i(nEQIgp>}22
z@t(yMvq6-*tK**(d&9CS4rP{aRHE{%L<GU0`tlr`18AJ@u#?OwKqeCw51LSUnovEd
z3Xs-C3U)a!B#_97UdfdFAT<9uniO{Tgln3(Y|SsbR0dDp9WvA~hOc?Uq8ByFjG`xC
zkyR1kf4NF9!Eyt4cx|TsAoixpN5Sufvp9pf>U`LT%s;|4JKLI^^5LFH9=iRHq%JMk
z87|xssw|;$-3(WAgqyokkjRO4Wl%P=i+xcNmqJcx&}JP}3k}#T|3sB8b#Q;1_+Fa7
zi8e7_oEhUX)m`+6qXwMen0<n=qQg|=u%D?U70Vguz-4=-=?CVE8och|s4i`k6-g3&
zV+Fn`DqjX3c{1=AD@>aqR&?fxl35bY@)Z;1=4aVY>b$f6kBj9*>se#fsR7p>cewCq
zsIn9;6p?E!K#~(@6O?(}KI~2=V3Xrr#Q-jYe2+%D6D`_89#debGMkL0as_ZRY3ja;
z^t&ymycWM#c7E2U&d>T#rM4|)zh3iIhPj9St2R?J%_L<bJ<oa7(vqh34xTUy_fPBT
zyqE~#KBB33hvLW&NAAjg7=e>;MVd%NS*3w-*xe+Sv!+FsTi%kDNRX?1r5wRA?}kRl
zoRK!%Ejc4?s5|KlAP-3L^HT?9uI3OuP7cbR3Nb3}IJuwzcdld7lgmbbMkJl%;|_Be
zu*J5(k`mC>X<|}iUu3e5;jBAzZ(^kIv+`jeiU`6M{~i;^7Lql>af8ny`}qrf&Z`yg
zN(!l0H}7v(EaPnTXL3%^PwDsgmy$m2Yh1=sL-`-$4}o~}xdrZp7S3h%$6S6eT-owR
zgjRFP`-7QUF?9LlA!C;I`s0|A?Lc+IzTRaZq<`@|+7j*m<0%M<`{$bXOZC3PzF(~O
zckTPR<n>qC_cIm#I`6XL68uz|yutpSPQ_jo_F;}VSwCa-bCpSlK%gE?GLE?wJS;W%
z^7!=B&%yDDsh=0buhhLA)D}zvV<vO{?P~H&jx5xjEV5uhUf#_JAJiooFWI{C0*%|J
z>h?i(TLOyO^S<w!&t$@~js6Bgl=@Xtx3srt-fx}vsU&K9d|miERwh~#%?Y&#Z!@^!
z7;`YJX6FJAUhH`cUhd9+z=l|t@Ps^+Uu{F|$rwU#pkhN_w#TQIO12Sk717|{wMt{k
zst=ALmgr=N2rxfI>M?T~(xNxy7?}-0EpmDV&LCIPr~cyYp^7&su$X%<qLb5QubW1%
z1!LVqAl8>NN8rU)6i`5pWgr|SbBWG1KTp;{?pIE9AeBLSCukCFV}IY=f>@@F-7VPJ
z{f%j?YlN9tgbd9Y1i*WfD!n`x3p+e<+W~S1@^ZY6+{aHdw4eI7`_-!7<P_l<nzcM5
z3krVF?<*{XfX%Db2qY&FQfTJK_a-p^pSK#!U;m`R{P%ghZ~<*Lrk1ZIQ@vtSJ!w;|
zRjLa~#r*gIxaFTVTQzaa?x4R+jbU;>#&k@vpK2kG=F%L;Yv&XWM)b7UAxjf4iz(Q{
zoJNJ7ys8K)_&XJZz4JH&pXC(za%_D|q!?8s6dRPziA2vh@OPODH2MBas*cXguvu|y
zGQf-cjTpH%o1{{j>QV~g<T+B?Ez2X(+j1gqKh2LE|Gz2146U$S=8&g}%LIdzMVHLk
zBgRSb?m$T7fwDZ~HQ~6>^B`0)e)2)PGVS&=GdNEn$ONeqW5<990?7fnT5IpVbVBw5
zw6zHQN`Y6!r`-de&+O0ZBrEmMdtf7^#cHd6x8Ju(i+9lkOxo{MaTm@Mm~q5AT_@^K
zLbQ`TUL7dG;=XEz__Q|^gRuf<{J<hv61eG(cVpjh;jU1{{ge*zFAm+kyzDe(|8|gl
z@GN>mdZi=}w=_6~V*WnYbLL05%Q~yPx8s5845cgyXlkCC>J+TR-ABz~(32yUm>yxI
zspbp@iImm5Ezj|$c6YqXdX^OI;<Qz<>nr}ubksWObh6hjEI>n5hbSwib=o4wld%Fr
zV@xF+uADV%4&%#acU!Do@2XL~-cZ`o!QGGfW6muYVslAo;h|8)JBo{5v_>YJ70dbP
z)r5H`eliZO4s+o081|I3vN4jLeaLC5H};g_ipKW0{InU~j@P@<N$<@Cfn|+cJz91M
z0F*Uy>uA{_T2R)=g`;JMa`l#{H@aKa*j;aQx~#FM-g@e-uioSh97*tS`J(`nOyaib
zl957{`7l{VBvknwaa%j6Pu<2s`gp#J-Mz&XTUpF+j1+K7&x#a$sC{tu<ct;ijjX^7
zZ}#L-D~j|Ovtn%RgR|CUG}Jy=zAn;0)J^L~H4yc)bz>Tcnmxzlnqza#vAO1uD~t_G
zr10(Xzmy7}bfI@_I0K_qu(*Y>?ED7}L&?;=8PjEt!+DF@zkpX~LNVgq_>pn%$r3hB
zbSnCfTmov^>w7klDp6jh|8o@vVHpX5xo(mChxzSqH9=R!h7s42|2W*<?acvpQ*uDW
ziWzz7<(aXwj3TzEd(&2%zkWuL|AepSX9Qx?jj`xl@FR6S&B^nydoWzkgijG6S?JDg
zMqWq7fxoZG+2Q)S&w)g^;7B|_xgKSxL=K)K0xM~9Rx9USAJf@z#qRXb?YBS?kcge7
zkNwsD)o%at04pLt!-;k-!dP!HC$PZg=-|(J8MQd0uJOE;ybW%gb`r0vL-TK?MBW;v
zb%+mHWL_@x24+DbKt&^vdV{hI0~rK&Y%7}N4kAiKWl&n&fUg;!Rphj@p%AcW5*_cj
z%u*){?qbg9A1`*}I(*1|Ojf*mtD1^!O?V=<6%~_G6%nvk*Mcvi<pVSD9N!TwWDUmX
zZ?>u*vp&vPN^OAoX>GQ;Gd<2-?D^SteMUf0qZ2LqeEJtS@(ESkC-lcYDGLcBE35@G
z>u5&4Sl?Q$>r>HCB@>$yyVZ=vF(BU7i8>q>U+Jv*#EH&ma@Oo|Dpt^P+7pRpgoi~l
zDk9PBkyyTZVRvo(4A~Qxx?gY%<t7gLclAo#30{Vd)O#!#K4EM{wuVQ0j~wQ8eFmG_
zoRVj<N?A1*Zk#>3)Lo|ni0(0Mcr<^)0B_P`Sp(cjf6f}_^>}QU+vCsjx?Om5)_KD*
z=5I3`){>5&l$*FPDJSw6!hDZufP%N_ub_H}FHP8gK7!;BF5FOlQZYK2V)uruQm<#J
z+e=b-PO1BTxaQNm(!vd)yGj5;ZOyGWw}yXd^k$CC?7D<n&ua>=Ig+Sz)MG-tHK99>
zC!-VX@fiJ@hF6vWZsZvtTZ(!h`x#nNTDX4BOR*TEg6H{;v(ADH;PYWj@zutPZ`mS{
zdZe4N*qsR?1F6Mu+e6v2H#?p|@4+|;1LcN&E(9F*6)K?5wJ0js>~Pw&rm2ndVq%tX
zjPEd6zdiG2CwhghT)ZR_EjgJMH$wm7=QPhh>cd1iJlGqvIBT#w#sj{c7Z1Y*KJ2cE
z6uv&I=kVyL4sC|=lRiA!b)<$VaxdBlz0Np~*Xb`u80>!p_TjzO79FThFYg_R4&hvY
zf61lMneC4{Ynnv0yTMvUvM@a{9dsEv*o#zTmAH|Ys1d|*GhUK1qh?Q@JVe~}IS{Hl
zbo1Mh=m1nMHQ_a};nB`v)o!OakX}QQVbRWu1+qcjPbiBnlzJzkC0QJCSHg+Xr&nV(
z{wV2^{cRyXw5C@+w8oqQIU0dpIhI~It?8A{w6~^LPbg<a9Gpf>`#1C|9L@f%rtmOB
zu=F<Sw<#3syWkKMYe|A)ogBZs)%+(x!?X*F5&9rtc)e@qh>#US!iwF&Oz=UhDB3~8
za*7QRBQ7i|R+hkjRiOOCM8L?8eh7*M1uJ%kfv2QC0TPyDNSG)bG^{U8NXk#GYxasw
z#k$49*6WI-iMAS=hCd$Dfn<Jlc`)eHiomsq|G&_;Nepg+zR?}gH>lay30#gLaH4r`
zPHej7Xv>L*i0()nN}Y}IHZA;(Q83crUI$N6z?_pKjk%#H3uB}y#XvPGUW4IjrM)-o
zz89|fvQ0DB;>h2oaCB%>I6Cks%<GcUXi3|nrS9wDHHX6$4Un!3Y(lf=N2sKckCBmR
z=lR@|_3hA|izSlQ#K=*0Jo6WE>|p%7BtCN+fQW1Ue0S1(Lowz9ewt$#MB&CzMGu-J
zpbrt64au0c!+bDL&f*@yu-hIrT(LVO3!-4?wJ8s-8umubhi5FLw2b+Vo3T(n`U=08
z#Z?(&I^Yv=QU|V;&bX=$YC9ilOCu<r*Xgf@-4Eg~B->YfU>OuwO5?`zPQ?aW27snu
zOPf-uz&_>z?L5?(p}XX1of-dDOMD4OuSUAr?5v3~&BIhTYDqw7+ZaNdu>{ft1fb?6
z?xZC&ml>R~g!`#<%h>d46dkcm;D~^-w|O{f$Gvo*8v4{W3`+~UE1)l>(b0f6Zde*s
zq1L15&SCVfOG@393H@iO3(5Z)$Sphj9PJ5NMxr-bLQ6}Qz|A(67QQq4C#CNC_*ur@
zn)o^Jj+@dv8Vf&rk`{Mfa;e+-NlGp)+%oGt`g1P)Z05NDxW`BV_(`#|IfCz#2EH3x
zz_&WRyh|wpXO4FnBl%;>J1OjP&*a+);DQBr4GMs}C>(Hu(&!{{13ClrQo;Hi1J=$<
zhDR@&!&b-ux8ya7Y;&ob@fyBJ#J*mR9oD2{VCBxI;!xY-(65&j=Ph&G&*G<}KLvd!
zl(@a;m$);NC|R*A+bMh}R3U{bwG9lAai`u{Y~jBj?Ot1O5kU-(USZh4pvw2ko^FLP
zRn@bv5LbS?)U5@@I3FX?2LzE!5&>L>K|E<NWZLPMMl+t2>;P^6>4$|IbmQT$yN*_;
zPp_q|X0F68rO%WY)0T&h*X`{Xel@{JXPuG4p08WPxzr+N1AJ?tnQlywio@Bn(Y7ZU
zH0ul)^zdc|y`KYbaMLgj<8a`xiVWYLd3`t<hAnGakLtic9A*OM8+oUzgxc@`Z_NE!
z?wExz$FBEB+^)>On$p76v%Vj(^RJ`>G^4ATfA@nPU5kVs_h(FJkT%6%Y^K*3>lYuW
zfgw{LgeP_h0P6rlJ`^r&DDS4;43Czi!P>Kn26KvyNyN`EFPr+Z7lvWH9D&tJ=B~ED
zMxbZ$#9xwLGk;k7uMDsGBq1G)x<8npmOppBe>k+Q7-jJQcg%d~&Uc2nUFVm&tR@))
z=4W;65zr~YpPKpeh_L$wYlZN<BlAF-(48GbTjeflBj-|_D7yzPl5>4sqXrEgeY(L1
zvz|CCS44j&+$<@M%vJuoR+uHl9$55IzLDWF5dU!-n>Ezyyc;~sSinzbEf7{a-t0y`
zvX@TdjO2MMr=0?yjQ`A(&o}%x?0rk*68zIQU(-$wyB9m|j~mTxfT{3tPY5$9r8`$M
zlzt_p1?h)1TJ=N<Jl${MFP5!~eG<p7TNL&#R@;_Bal07*&)$U`dPIJ%V<mZB#?tA#
zBhhh<&G{o<aU&~-Moon5xoX|?PuW=uRs0hmmkMHTVf?S`66t846CHs(zedEhsli#}
z7ayq03RNAZnNGz@1pBw+ol4y|hPxRbIu!!kT$o*{mjS^%goa@DWf75@FXMSB{T&{S
ze5VaA4Ob0!FZz(o;li3(=YU<`8Sc$o#hJRusvMm$%XqPN-i-4yemU)JWk`>|l^FkU
z;fnI(OTFw3rEc~rO&Zh^cZpw9m&tg0OZ+R1UWq&&M;;&ZZKlzeRQ;hT`o+*4MP#EH
zt4iI?;leNHtTUH+nC#8)Lbe?0c4kh-=-&hM92tUob)a<i4s__ew(4I=2m04}`EYVc
zFfFWB|1vrZ_eL$(nZe~b0)563Nyi!U)jy_cYX2H$dm7OtmVT9L{xB^8Ll*5nMA55E
z)+M1k#+vqbUS8@p02bOmk^l^8I?Os9L50&>$z?-xf5M!FWM*Z}{s4;9Xym(LAjEQH
zmVJ?eeRiF|2G*fYGbglO8AP<xyf*&b)cFm6II&Hs*LA~ix9bL0by^af3llrB)Elz_
zMm}nT9nas@*3jQyO?z7zQu^C#?SY!C^0SV$9Ki4OMJI;6nMV;iFWO+%1xEBQq?bTm
zK$^cOMgC8L%(qBV>UL&>)r$F>5%VPv;9t+2H}|?lyg6Bs!Y|9thFHW;YK9N{H|&pm
znX*4}_RqCH6828h{!2$RDV&NczBYCe!R9UH@E`Q~|6cye8;)82GX($3@^NSzosNEN
z4jZ{mW#37eUlH#Te+F4mqoVHO{4(1-F#h1KlRsW>y=^cUo!Tj?Vp&gz-X3TBbV+la
ziTU<_@6X|h$L!BP3H}%TvFFQ>oNN64<umN*NilqHc#b)MiMMT>SX>_Q*Z?o6yEJ;y
zbIiju|7l8;9~<v<^)OPf9qgEWywM-;@ISwi$eJT<l6!IENsYQ1;p0gMA%f>Wc;+N0
ziFC<-yqtZfORqyGmgHa3ml=f^aL)8kqY#0Ok9JqZa@3_3!(U{^pVM+9Y_=&%ndx8Q
zQn&X^9^^6Nz0_S)lHJp)0=*<Q{FX5^k?2b@iXy2VM1`t)lE(2~^7(eW#e4Oz*VerA
zv*w`$R5nGl!fX+k+iGr6f!;IKWTDaq9;Q<EK$<HZMjD))KBZ!5NN38?zM;Py=t(^(
zdn4kt-<ybEr1-xC@%@OWjs=9NBc0(!VQ*TN|6>}bO^w~a(y6ady#Ba_X@-SJ7Xh+<
z;EW)2gy<D9p~^oIqA}4Fdy3rb9Ou&RvhgaKU1Xi`iu=DMCjOE1N2+B3U#0F2qvu^M
zHs+lf#?~bq9UR>EbiI<>`*ITHb#GDLU{hW&L<a6RP)<yX#{h&TaD@Wj<R@sMc7)!U
zCl8!G{fcooqv`8M#?!HApS=sUwZf87E;t>7H3=xPx20n0_#yk<LG}`xy`#<kbZYhv
zWRLw;$<Y9JMapT$PTjY_POfB~*V!vzr+}%lpR1z^HpIS<_4%}p$cj(PPfCX44(DXP
z{ELE{pGidWF0o_b(s?O}pZ=-lQ@gNx4Kk}a{7d=~W)^a2y}xwp-nY;Xi$f;oN3R2g
zKguf#-M8EcmEspAvJZ;1qCBPxlaB8<B5!6pY&g1KsmaxrgBX4(2-ln$dTHm^Yh$|z
z!!%dE1a&|e1z6;GX+$z^i;Yu25Lc9pTV~@F5XALN#yxD~6cEJaCF49Br+^?XCmA={
z#wj3(OH0OGX5$nP#O;;e7t@CeY@7muxUI>!@7Op61aYgAaZM9U-xUzVRVU;AY2y?S
z#66XatG96q2;v?{#yw`^6cEHMO2#d;aS90H<|gB&+BgLSaZ{3USJ*fO1aV`Nas6zZ
z0)jXv8Mg%_hs5+u2CT9HImv)mY=AtxT2OYaG+Fj01K!gNsjpCg{_GVl7{dkrIY#)$
zjb&r?XV(V)UJSl_-5mD<vF}puRh!a+Qaw&;(GbnQ7FUZ+Jzeyw`D(Fudh?B~l${x|
z&Ptz{CCS;4&<BmW%w3Nj#xZ;vLEJcPY0sjBxi<turXk`1$oevE>NKQh{g5Ja(LL{u
z0<hH?GKB<buZbF){YNu5Ys20J5CLt7eEwb2yoh_c2s874i{FbD7A?4ObU>W0#s>~b
zcltMwH{Mnou+iJ5`29FjsSYpxUd#S{_5;4_p)QKXKR;+{{C(83EK-a>3_u_?V#`pJ
z-gD=S#+_lwh8Fa|nb)q!FpU{b1g!wt-5kedNm4zQJmuS#asN(YFu3Ic*<@;#-<`pd
zTorqb!Dz`)?f#Lx4Rm+kVriEkUF>P2lyba_{Wy7gt*F`+2@N{bYlXJ>ep6GkEu@VH
z%I;-vD&wd*YHj~oa1#oLW)4pC82!;6ZC9lW7ak0U1~6dt_7r#~FsrIAKclE}S6LTW
zh0<{90oJGbh`ETtXMavd>%|4$x=mc5uRm*y;R1uNrbiqq#GW2tS}rgEjG?&lr{lF`
zxQ`56S?fn^hAuY4`D751iET7j5Nl6XdxP24N{6{#7n|)v9@Mk+ZrLJWkZNmM`ax4u
z(Q~MW$`rE2-$eFa7zFm2Y%1-NqX*$ROeJIUqMTTHh+CF-R2xqghJ2T3zXxL@AnX<f
zEmy7!udy?x4bcgF4;%zc43^(`ayp`mnf=+t;0%I9<pF~{Q@WRWKVwlaC?`^|6&LoI
zE%qbWV%{IR<78nEyEE7{a$EAu9x&>X9z6|zI}G}VQJgreQwmp>_Yjlm;IE`T49IE5
zE^-emdiPyGH!&&LT^-B%J~JstgLj(X<t{7bHVDAg9XD_Oy&tk>ik9{09>HLTQ~ERI
zKaemvnRkHg!Nmb=nRPe=g`HpdeoOG)3a%LN=KMlq+U4n6EqKLG<3nx5|9zFI7)=00
zPEQm`VcD$;>yapInM(1xOb7~_KsZeF(qG9#kpE+lu=U85b@b<e9dQ`=a*aOSq5blY
zZ=Cv!7v|V5VzCoHGssSl1HV5P_(jUzvvs)^Bd}yR=!a;8e+sEAMr2x(DRAOU;Hm_>
zX;&(HxAL%fe8jszba!y6HyZ-nBvJ7ad`D;b=KvWIoMjXdt{-g3UN|~C%in*h7TZ5G
z<V|Id#1>P}-)jslq54<r2`hg`qI2-vGhMbOy2>nd9rW9uv=E-{I!GcYYoXo=@v|2d
zEmDo$d9m_k+0s_B<yQZABIBDA0%6MiBj~4OoP&TC@qX&FKY3Q}fVnT9wGBK>EBnrX
z(DRk2Z{>BE_wwm`d1|#bKK2~gpAr(P)c!&7+&O*I%GzVlG~~)S1JPvQ9roJu`f46|
zn@03x-*3PITjGG_m8Ywgbk*CUuqzUUEhx!vSJrJ%==nKU4yZ3VJ@-tKN4y(+8hDPf
zrj<{P_f<u&Fgh(O!k&w|OqY(Gvv8RA&^fk8*j;=?Va0RtMV2O_veTbr3e3poJ<)X9
zYbs42yx`??wvry5ygYEU7ylXR^+wsX#lQPX48s=8%x`a~uwjBb{E!O6IBDJyne3Fm
z6ZW2;D?x$u6G)H#&hZ@Iope0LJ~Pz0ex`Gw=O^X2EC0bT@7aE~od?Q_is#OziQm$2
zuK+pNIO8>?g(heEfCZ2DQwx*+agrJq|8&}dJbSryXO|ZYzGQ$6{t`+=eJ#&j6KQ&{
z19(J%8Wj-?<-*jtpP&C@k{G`9J5#<4*7_MgD5uy|9Uy9XAiQBW*OZGmV0YJN%w-}^
zU52@EnrP;etQ%e!O9Am;Sh{oW6{>le|MBI9FFm1JqJ4fvGGkGMJ3rmPJb$C5Xu1@8
zi~lGXZRlB%sin)dGE2|45f10tu1{0<){&;}wF^w$BUSg0say03VEDz<ZHNh$Ef5)m
zhE(pN`R=b8ouDNN?(#mFGzbq7t?_Tx_eur`zbI`ke{&*WKjZ^b)kr~gq;Pepax2Ik
zwqjRJV_WBigo2)(ZXCxnwy}s48n~K$v^LJGm)pP0aN$0V6|BM|99I3SFyM}GJU}+w
zp(kUdRV!rKoF}8pTp3;Flzp2`1D&V==ob}5+;{wQ=t9_iH{PaKb=aE)bJO9CufhfE
zuvSKaifKb2=4CWs*42<>@@h|(4WtIAU}ISc_6(u<=h1q9iy1A9?aEQ3-URQa01QV^
zS5WI#sVl+-UzBw>9s_Zw|Av18a7LG{OY4T}a6nt6M~o%M=P-pdgkvz;gyw_(6WGj6
z-xYBSbVviuS5sM$+GW19rc3C1aA}hz?NBLW?Bs;qQZ(G$hg*;D9R5%%=OW9YA!{sO
zuMaKF3sq1eKY4RPwg4Q?Q@|j6JIOUqaC}CY*n1@7!SZ*kh`&NeDZ3cQMSF9jR5Pd^
zGiuadD+DEH865Dq5|)vF83pp}yG)(0_rJ}DdYR?Vp(Ot>10s}fV<z1e|0m{I?{EK1
z`o9+Vgut_si#-kE?1ffqSC9=>v`p(u-A`q1?s$`#h(pjrp%8D(;C6P$?VUe$th3bO
zAx7ukNe(AkbaVs<MFp`);kMB2IFm?=6n>1wDlXvPVvDdKT(}%G+v}vqK(oElCRA<%
zprE=jN7wv}cxd0flVY-()fHm}uj}dy4#kwZJ4*`=PAxG<GB6P-V45Q67#rGbL+9|2
znLeIoA1FI9Qo#L*#P0X^2@qayH2Mdzkju}Cbx}t-mm+Kc%Dp*##kTY+Bw#6i_b_?b
zDDz(=8`inDHMl6lS{}dP5%)H*hQm6#iWJVku_Ds@Q6;9`Is{OfelU@$>LF6a=JC_3
zy40-?7i>f??5KR`K-o!*nj?GP!+xo0BY~z0noB!n8r+9)b(d@JRJ6JgE;v{Qe(Y4X
zZxKxtmqTdjv&X-~pnzzMY=waYy$q)>kGNBXL8%5bG`hK+Ww>{zXPl_`q-C<$bDD|K
z{+nlkY=t8cY7+Qu^(P5&CaLiH8EHhejtD2Ma6*Zw;IH`?`a#Pd&_W}gG?j-T_Sn<*
z3^Ok?{iPJ5c{DhWxi>IRS0+Y>p{K1W>&q%GUYM$X5tCj(df^C`fd63MK8$KR935hG
zUt9efiHu_(1XXH}{!@xEhp`Mtr(q5;7=Y>C@E>ExeOa#f)9DdeTzf@E;_h%5!=S*!
z?ZG(5!VYqsQpQHjcTMWS-4WQbKq|^Ufu^=&sQN~{@kMI<u)c&0?ModFbvPz{U(!1y
zM-#cT?7O9C-aA-*+>Wx7TIl1R2b}SbBqxFmP5#4trm$QVUCHwd#$MidCxb9u!eO+x
zDW0i3cTP@P`Gr9FwbKajyZlI0x}I|}?V7>|w}iio5Cg{I`MKcew3D2++<A7uPXmkK
z$Cx04aRJ(>F8|?S%3aI)?xx|UoZ^a$xDYFP#+jF-r<H%-EVN$BQaz{IBDxt97mlhq
zg+j_`=8=oVBhR?(b{?T2TVbR3-elP*mRfM+kLMBL?&r1xE);};cA=c*19Q@2WzZQ?
z-xQ>d_PLSMq%i!el_8q>2#*(bQ!{zCG1C5<G_Y9$7riqV3Lzgt#Li*KtW{SkZ)|W1
zhmS3nbb+rpqzfxId_j*PqO>fRCL0@XofyjU?;|2uTY@`V{FyY=2n#2f7It}dk`)%7
zB|KP3(0?5sXu<vKi1t^HXwCgUr1hFcqVIP9s};T0?H72al^rRE1BB=-(<E7x8H6ps
zrtS3yn)_F~{fz_r@Fn{I2mGe?FjrCRa@<3)^N_y`Th{$O88AzUsH4+DRU?Rqy~>YM
zv%8(s#DW)_pY40mqK@R;PNtK7sP53@jt6g)pTx;sn(VVr`$^`ze@M?;Cug2)pB=lI
z@IBRpbvfOJoq7(>u)Df$+n)F8Ht+eUZq1%uwZ-4w^HFVaC%tvpbD*|3eNS?~d)$Mw
z(kK8=kK-OJPusJCx7FhwyeVzZTHZE|d+=vzl=Ic#t>YfNI*ro48tjjI@ba`hyLj6>
z?oUU0Eb7{R+=HcQdk*k%?xOsVN&BZG=Pl}SvWf29ZQO%{()Ki&@a;zm7L|lL@ZK<P
z`;oI3MNaN8F1<hX^L5UmQ5Ps?HSawZ&Afz|KD1}txLM!*RmQ0qprFY%Zr1mHm3@Xj
zZQI6eC%)^s$#~OD{8Q$wqfnyq`<S%*zQ|km-nf0c^45Jc?$;Nmoz*bz*FQ`<Yt6X7
zUzc{)d*l9oP1;!>joVk3w{F+CUVkW9w{={v-<-WJHm+CnoOS!h^{VQzjuTJyX{e~e
z71et~RXR6ZThnFTo3%BW>o(NwT8G2TS%oXp>-Mc%S$Am4n%#P-tvNLNt-6obRo8vM
z|Mlytm8K@W?$Ej#(lqSWi_%nMxm9*n!k?_;{g>l-w~h#>@2!}Vf&OdI{&4yi?5LyP
zswv+#Zp{;Ex9R>M(8}Ue*;F&5q{K@zld7&1X6E-s5zhVd&5Zn{UD~+d`n52Ar&%H!
zz0)jKz~xzf`-rB+9{>JQW+iIOJg1@`$hrm6>r~v9#p^n;cBaA<W8Z|;GRO8x#Z1E#
z$pPZsp~?;<kz?Fl@vbTQ`*7Q99BA-ZlVk_OCZc=txP5eE8CN6G4T)9=fzykdHo`nc
zAhQKT)?3&>`mj4}G%Hi|43*X81ycU!3oxO|4}}u+`V{RkA*S7Ki0I&CVyOUZgTrkH
z7lqxyqvPFV&%gm)Z(0#6j9zR1Kla`QKC0?!_?|!lg9IihXi(5d2T2e#QB+byGmyX;
zoggaJU=c7NV#R_m0WSd(l31q0D7Cir)>f@=y*zEb)C-_u5@r&>3l|l+hy+kL!+<1Q
z6K?tbYo9Y`CJAcW=lR~geZP+%nVGZpK5O6BUVE*z*Irv*KaV!cR#df=<sUJ+?0%3}
zxM7R<W)x_?g;EJ|%Ap&|F9B>$<?=rK&Ts#7MvF9#7YVXYn%H^iBrLUsR6vp#@$6Hh
znoWlhipie!bH!w@Lkd{B4W?kkW`2N%=##IbRsN*U?~1<iX`Xte?q2oEMe+(4ikq*D
zkXN#dSDv?CdDeL4v4_+v2ZfbSq%pAavn79mk$<L@e}<92(8&L5zLtiK{JZEReVyCj
z*$~R?>*eP*eSC}6pp8DoijN$6k<gB^Re1b&*|nwFkL(;Q|7FQA>~x|GRdH5iR*|-(
z_P$>7ANvww94i8v4zh<ja_e}t2^p9O{+{dx#{wCDq<oLB!R;8i=DNUuY_9!79M(Mv
zO>*=_gty_y`TNKx+Ek@pM}sA%y1zX_1X2o0k~<?9xe#ftJ?_%TySwG!*a7_uO~gMg
zRh(px+yh0UF<`?h4%twX4jbsYSE@6qoZ)>%GFN;p<~_T4TJw~g{#s9;SrSGbMSFY~
zgNS=X(DO|FgQc0Xw@3{kWD2E?#LJQhh;KiLfS5`NQTgyEn(rH9&%AW+Myldw0m1s^
ztH^}vJifqSzBsuzh%<N<;(x+-V6^f58}aAh`S&x7?>kF)xEF_`jG|}xnERR={X8(t
zp_SFuyE71O4{3pl6!L1mytP^&kM%fjrM4uEQZxDEsS;?Vs2E&UjSC#N`ZcfjOv3%2
zrih?Cw@w|X>!}5b>L3@+)7>r{2DQ_l58!%{O})KvL{H$PwmMEH;p&l?8HHndDb40?
zQ8J3LLY=M?AuVvHuIzGh=CqamJ{T3Ch2pMR`n)D8yoU*+woT;V4vKds(h8cq#ITls
zZ8UkP22K8Q<6nC%`zYjfN@BY^hB3tSI??w%87s7^<3holw5DyS^InXwi6`kIA-fX&
zZrXu#7ilm4et}+3=LZIS$&i1El#%$QxaO-944P8N(VaQmoiekRlKJ4t^jQYnS)8bq
z4l5e7P<9`(B@`tqVsgCQi7SEWa)eHf5y-Lgyphs(Lg2VzyD39m2MmCdH*`VnmudO9
zzCyC;qN*3+jVUxaG9{8PyDjnj+m~Oj+_IIz-bnF{nlGmwg<V_-Y&Epe)Z@{-G4^Wd
zrv9$^R(gBp`|jm!4Ka0P(Oang2_mNB4*n*Y>&0E5{BI!1s*L1E>DwI`lSGHiOxL(|
z(l-<NeoRLiF(xUK+sD!Jb32M0Y;)D*c1(d^dpf4MeV!x@ufmQ}GR<l5_7NngGGmg`
ziNxGd(h%VqivC0mI-oVQi@Z#Olem0Aw4ql<NLQ&-q8EBNkRjcc#sHSM1BpS+=MoKX
zaLn4jLbVf}swT3E3A*KSCg_}8q3T+F|8SY*yY*W^sjnCE#ZItUvCVDzWx{UBv}`AK
z0&7Rl)GtQMB$lfV{qu9^m!dnwn9ZtkRPPKfR5vWmvzRVNiik52Z)qMt(NTQ~TR{s;
z^R%d?xi(K2NQrb@)IR|~ia39VKa(Nm@9+dqrak(xHumU;B<m(>!)+7zkBo&sgvap_
zwNS^Jru-9Q^7Uv>;4XXZNHu)XU?(2mUD&<uN|PN+sh#*=m9Lj-Vwn)Wa@^3gk04c?
zsqjg>61cvj{4Xl|K5rM?cG5V1kmefFu{ak`9gfj_10FBHj;!~-&;8e>Wv$O%>G*v$
z#Pm6{XMIsPd*$5jo+0<7V$WbXJ4z?|HF*JV!Qba+@Aang+chLlJP$&Vu5t4L-Q)w|
ze)_L@NA%E+5L6EN-ag5V4`SEwZP$<H2ua<Ot~GN*$35xck-|jFT6oNf?!p2(5L7bC
zJrs-2wb@2%>?r700xrOzxmI7aJ)T;WzXa)LTj|qUr5kGl#KQ)RY!4vo!0NW?UsDD4
zM$kQv>PsL~R8yF5QJrF?aW(`9kpp_yqM@@{QR}?`ZKef3(8{tgV<V4}TvgN{Q>a5h
zcnTN`R_Hg-?(ZIdHMkge8C+|twMmHtae@GBx~!bm#X?z#vwlWUSv!5J@xt<CA9E}i
z!A5d^_6e2=7XQyi1qStTXMcj`+u%vqfdLG2(^k!}vqGh~5<4B`-4R3Z?NaNv{`oLw
zl|Pv)vV%<Xy}}_`hEg15?a2k!utc87$4<&hu4Qb7ig5O7pPzk_7-G~8X+!aMNX?yr
z)f1x;y9kVZvQyv81mVzvQ%jhNc(K*nEkA>RD!Yv^Df)Y)VmxVneQ0n;{bA(ujrc-3
zX1e4)5SCD-3nqSAxc+2%aZT0{&f}e$pLmdu+=FTAGe><+l;AsTXR@qv@k99qs%go`
z*&p!uGGT)Z&dQ=g;f|^?825+_G@ak^O>Cr2;ZXotm0EPc0*R%>&LyX9>vMg_%62$l
zhV|S5kX~Gc5VUt9YHn<h<s@ou?X$*LrSVm#j;Ge5E8$L6Q}m?=BkoyQN7Ve)e985q
z_fq#m`5bcgZ6(x-w};h*B1J#J)RMvRBz!X}P<#?L4_Brq$pr`fijJ0CGtFG?g#Dt;
zIGD%+-U1EE%lSSGKo5wqms5<2iDNB`8hS+_XS0=?Uua?@M|0hQt3R?jJ%0TQR&mny
zkbLGIN;AGNKgp9h1`S1Dy0h1K&ocR|2S*Sbneb;WqTB~C0qu5YZ1eV2!~d~Va2a#O
z%?+=fgnMzLu5Dyc?w`f|DH|h`5HGQky6oXm%vR1Iw0Acn45?}|1|OOgTpwS*5cKgW
zAGx0@`uJ4Q$EVb-8v0mQR7c$$<;{$!RX;@^7@DDv-N0GHF7dL4{e7sc5r%)E+HN;*
zBf{0a$H3uaSnv(f<B9a7>&7M6f>p)cM)V3brYBwfQ+dKxceSf_ga>ls#`<}q-GPTo
z_=(ar8k1asfTN4lh>2vJA?_&ofxkU#C4ldK!TRt*Kqx+6-WFASUYv#cBC!Ib^KN%B
zx@_$UMzP}tR>_5LBZRHd!@79ZTcZsBWRRQnC)v{x7Wg`IUis&m{*#`-d^Y9hq|2&0
z6>}Y;(amY5_v7bRw^&w53k=u9A0SD8@EReo%aui%mHJj*_dX=$n+#q&3vVHhVP6t{
ztMO#6r%$w5PcJ4hvdOalimoi19v7kun*vHFa>!~Oz?w=8w(5-xRfHR7E|;2D#FdLb
zU2MJ8q6z}`dUlEDr|6ZNjyEH+dgM68IxJT1KptyDxw_3(6d!dd6^ec2{Mh`~{Ki$A
z%{5%G<Z+Zg$$MgNn<(njXbB`xK%d)cfY_n|fRyEnRWTJQ(OA~GbMINn#y>h53DrQV
zkEceO|J`Q=4!%s5P<;>Q$cez@=S<bTc7QYFZ0!AVWoil8JieWtjDx~9i4#E(CTpFz
zz)`lCbj}F5GmwMJ)peGJdPzRQ2|Q1#Tb;1wy0$tkYo$OJJW6}iLej`W5>)YivR5PK
zuOqY%V*WZ7EsgkVD|y*bKA%rS`rq(H{Ch}5Ss-)eTF7UdB><&2Uz6{Q6_JUG-#H!*
z@eCLG8^@u|;6zKzoF#mRWjTqHFQ~!$Kxikrs6Oh)$2ifI_@3Y`0UCkO-c%(J;^F$h
z)qjp=*SK@Mk_^0aF7IH(kq(oph)Rl*%$Ck({n~3qB{d^Dy)i_l5?w2uCGWz0mP88@
zn6#Od;bDOpvl7}Mno<gd{*^T9$fWyL-$a3~Kps)MSRFU9FpJrxtg)-(7uR5WTe@a0
zUH7LdM_DFMGWz;UpUbU;SVYiIH2?i6Q5w@P{epyqjZ&bDrF2(;jDDufqUQ(GF5G8E
z6WqeLS=4@CrM{J^K&}}~?`pQVFu{u`Pm|tsmIwx<{+^L*;`$fKB~fyrCOC8b2>isB
z_}a0!S~Ss}gcaKsmUB8W`X?{AZvi*EZ<FfDuXPef@-$hE8jY*@&Wyv)aw1p4%f9De
zX~z@PBA<QR{2_Bwyf@_s?m|56Ck<0U&kAG!U|w~EDoL4{ChHqcRVgTagJjbCkmxZM
zxP3Pu{)13hjYmZ_8SHqwGdkb3i&^z#t;TjfT3<`w8~HHm$sKcPVV&Yz%5SvY_mvnq
z`ecioB3om4y1D%h(GX?iIE|U6`QLMv@T#QaFwZVVO_YC+WK(A3VA3FRv5GI@(4qXB
zkGSY48$##l_X}nzo<VMXz9?&pN|7tu+`b9v{AEISf2R(EMR{eO8zU!sjhr;!CN*8i
zO!Fk*Q#>J4u^z5u%gL0*Vq^{?8jC0_(^q8Azqp7AD2gn4ZKkp#;u6flGe@6tB{hQL
zU?(}UDOq-$-Zpm~yv`i@J4hLl%*!e=nTOhA31zW7o=|kI05c6WyVZSe$mmjUIM-%<
z?NvsPQiM{p89GDleowJjHMIJFi50o(RNN*BM{^t8&A0|+DpRPIy<hPOmKukUJdH7(
z+HL0JXw?hS-{ts95nqRL!)(rAqz-OmV5}7|y4&)LD-2c@xZ-1W2UNR}anV#T1x7eZ
z9^G(}Y1SFGpf+5dzgbq`9>e((_;BFH-c7cJ?UE=qImVbtco^_K-3s5ka0i9&CIjEW
zn+$vhfk&nxeeAo7O%2BAGR8=sJBStshJFA{zmRf_5^I6NYAIv}J4Soi6;W-Vw1KTk
z%a4}!HO0vNL1t+cQrgJKB%#Q=Baq1o{KF)QYHH**f&C=X-wYp$F`5zZ=IW18pU6;2
zWFoJKTrMA%$Ol3zyi4-24yK8gl%(HDW^rsxia@kT9Y^(9)Lv>iTIV_g3dDbwI;i+2
zEXL;2eu>bfQranWo1FRR1ti`@HEumu@lDv2y4DT*z<jw;PJ%D3v!>rlNBM7=exf(G
z-h@*QDy}6;YH|#(!~$irJX8|g*@w}bzI7*5qB(tQ>{$wh3SCN|M@(9FFKRj-3Ux^=
ze~|a*`s30(-yU&G@%WrhiFY9$m0AKC?|$ro43|EU+L#RPuqPWVj`B~WFJ#L=*t$J(
z=3<$=tSKV%ILg)1X&4$2`n%=_rm(f}5fO@fsAqmuD!e^f#ZwZa5*j$k;<B5EynT^u
z8dD55q7f-J;svDn@{ruw?oKCT9wwm?3}_zP$8o#NilM<$Qbx$lj(Isz=?mZLIr>5&
zKLoJs6POU`Zh?7$g|Fp7D$ju~NBO4|DG~c3z2oI`Bz)CA6~ggNX}#4xM?Pn)P+Ozz
z>!6x<wy^8i`V(s(d#kE_SHXYH_KlSGjf}Q$WVC&-(zy1;7`C);WVC$-399zJbh)(e
zS`~3I+P=qusrE^egxc+*>Z|MzkYuh)qCjJ^G!E?xbtJ6BxX8?CyZsRu`WokxbWfJh
zO|2fx6j9_zQCH26`xa8fB#1MjEM7gqZ_?Ka{TZ1dOrAsPyH+n=+tMDQ>}RHXFe{mu
zJUoHvEUA}z{MUH=Q@}V`wr6yA*@>2+-PS9Nn4*<ukl3QX7F+ZQ!xrsebv10!%9wvx
zs;%Uem@!(Bmbg#M%Y>#_%J9kxnzHZkbe#RIRkCM`RdUui{g%swq1B2>7(MR<0_j8P
z{AD8Ry@<IRtLJQ7nWtpBGY*MQ3Ub;GeE~BrKLJa_L$dZ6ggQKjo{pU{gD_X<KfY8k
zOW2ABK9;1frs~3w-r#)|pXngSi%tIXH`OdL8b<lgGeEwN|0vgM@mHU1U)J$%N3DMG
zC8{ZIKLLe1A(bN;bulF)R_xCrXBZOEL6`3{(J+uS@@&`n@ir5;?GMzPVM_iScs^_P
zKJP<rBc=!wUF`dqkwwTl=*iyUs1TvhligHwfqTfkDMdYH@6uqzXo5eY;(~Gqd2nuH
zpK+fY>H?{6rCG6}Hw2l&<y{gPVcH`s`~-~wcV~Cm+!SQ}r{eIO*9y<-7@m(lkWtnv
zAZM+XUoyc*r)os|z_%E;8v6AUIn_CXnsx)68et9@R%kV=@Qq<!kA`%+w%}V))yZVq
z;zPu#@+y4*C9sDjMrh&4E#?{WO|AaZkK-puoW9G+_0N=F1D2gysCGoU=2Ls2)btkE
zdzGF^7rvC>g?rRNFltk$S{H9IIi)T-Q>T^D@vW0VK?46m#+Qeq4K9M4`6e+Fx-z;=
z;_Ab}F@~?bdi146TyD!H2xrXk2wKWvRUM`K9)}>!*K3_1^$lurH*?@l3}%KURa|GJ
z#SaWn1TOkN`>8g4tPnMZ_dA$m!aA0!?^@mU5%K9Peqrzmt?X2SqkOsQROnMA`>P!l
z4Lpwu3{a**u@Ap!&^Uz|fhFIeL{4mQC-U$VvSPMon^On|H%VM0E92sesXibMZ6dvc
zhZ@w#>Gie3OYh}$31_5%RC_udP>xV%X03N0JtCRuzoK9yWy<2L^{yD?4Z_@LoG<ZZ
zu^IA(R(6fe`zkD>Hu7h2?xHrCsy3}^Rxw4EDWr_R@Jz#m@y<Z$v004i5KC5Ep(k7{
z{Z*@H@NKNN!_)O-A#6IcyXw?!`ZV=js~=iz+}C?ECuZO$s2hxp5`}I9Io<x#@QwUf
zEbjFqW1AcYJ^r^vm-P6T%Vd_Vj<X||#;teA1Ip=BdA?caF*|@N^7V?iqoCCNsJ>rD
zKJd^)b|WUrVp52DIT|ZYfoa8Y(30pfk19o4psa*qjb3<NMlpKg$~Y1F2S6>?UnyoG
zP}6?%hc@<`dq|A4-za(Q=F#FOENc<_5Ug+yO;rXf{ev;mvn&;U-0d>Bc$kyNM4|AW
z&u2SFg-@zRW|x0<8h1QviTe4sft-01h3Wfn$7JD<a|?{r^m}<Kn>X54bTQnni{x{>
zJ*z;!m2zCZ={R=cPtiPlqG+vZ(lBL|fA_qRQR`Z9U*%Td!`_X%*OEKf^M41iO@$dP
zuyWmE<eFsUDv?}&Cl><vM~saAPKLF=6q8p+o2xN$HrxILgYjgX)UprHnnXx8jOH5Q
zC^64q!kKl4<?TA37<5`zl_xvwSg?fN@ML3L4`YsD!{8%1faB2W3>VirS+1|aN}`{^
zAbX&09NlmY(0s|x$bi|?al*u`<GqXJ(jK`(SyRMbcrslUQ3-Hvk6bT-ItO5<7m-y>
zY#6J)g&Y$40>h6S_7?40@5ubOw(=kBFO%ou`~)0KXW^7;L`ye$k{Lz~73RckNIT;}
z7L(;<J!?B(`h!o+mkaq3Z@(+sJ5@X&QehHy?S0GLyn<Szd^pPg7tD^XO;L{LSxkha
z`PXrS%2H-K2))jUk<2NJ^L*W%F1&T11!TB<QI9?TiNx;<!ty7=gh$w$tC<guhpDex
zSu#!{Ik6j`e62fy7<=Ke?e|~jPflP=P!GbMg#EdJn@%y+eV@7<xj23z?s<kxOfCA2
z+A~7>Luv_XO9;^@|K_tBaY7;imS<sUlbWYVqtcmA=oszMF%Gwg$&YP`nEX-adHkmh
zTU@<mi^C-ZhvGou=a2{d_UfEmw2=-Fpv#BLdI*JNpM<^};Tv<<M#PuHiKBeQhq>~4
z+F_CK#7iS{9)g{Wt4LJ?n7*R|lMk~u5=>=`MfFZGyh48^lz_iD#6myiA^{Uf2AwZN
zGl>2f8xd5QiJ}3g%Zi@HI2_fY;}CvYHd7cRvCE{0wIV;`knCmA@NW<{+sz#8;)*yd
zWaj<QuD~3qYd;$=`W@KI*Z@!_4t1oB)d_b6$EzWCcJ0g!2vTY8A$jSX;o<%Kz}&P*
z2IsZ$`o}o!+Yhm)fxv$=A<g$AU!Nt|%vZek`O`g<X5e~NfTY~4-zq~h>0^><V=~Ud
z-jylBxCy4weKw6+%Tu$rN=rD3mRhtFLk8&q0sGF^YCFnE_`PcrecqMaSid%rP27db
zuT)|Q4p<wW3LQu*m~R0LUy-Zh_&qzf=`N8&%7Rq%4q`Ex{iJ%_MLd&4POYf-9OdF8
zrC~z9lYvjf_A{*p*@=O{361SAs60pPKgci02slxsiT=%qd7b!r={wXs69W(uKY2Wu
zui=^llp}>ibI!LR^NaGvKWl#;Hfu)mKhRegL$17<>(m5Ej#zhKEmA^`UQItJ?;<G^
zWj;wCeUY$@eU5Sw4CRGXjx})+#~mq-ibVQE^Jlm-_Qs})nkHw^Mq_fcYc(-?F)?IH
z@$UjH`$HziCh-7{hWLLOGGadbUPePdOl4Uoq<4q=65oOH#SRQP!?bqjS210fB=4};
zCQ2XBi~IHGRr{opzHE>0K0LS+|0wPcVF^`W?N6v-!Fxse8}Apy8~72lpXzwTCA&$q
zk~h7dAnxAAVEzno=xbp{X>xl=u8mM}m!oYYYQEd-?&hF2iO4b^IRYKjt6fJp-X3(J
ztE`6G6(U8S6ZK2VfSqukMwFcx7kbb{k1vAX4LKO=<I#BiP;aUhjn03890<r>H>bVE
zu?wuq(IXWX&}F(b`!2dm8M5^nr|Pae-=IG-V9MmI`b&8t?a@2J3EY8Rf0UYw4OR&s
zGkHg~+~fPPGzX)~s~+Fa6swSa5wl=g_#8$HB`+TH{69zD02DVijz=fb0ji@}r{j*6
z;vhZrfo?1$J^BbQBFv_;JV9S;ps)HL36S%*{bK%MtS_?JgYi;)TRxE-a=w8lj7uNz
z7_Z!^^JH-sa1ZA88e^l}_jwsOl~1Qj2{PKG8V+VF76=bz_kYso68Dm$9_vy(?I#m+
z7=v%RXn4?%ljC=W!rfb@AVT{tCg2@ZQwEd;O@h;USOmUL<z667_KVR+tZXuS3}Y&C
zEVqT|H^a|F%eYOOltXu)rTMSJf_zhs5vHQOwmi<O>wuPBGdqiza8Bx%2=6E%xKD0X
zn)0KCQl$V&!ECXDA8eIu(-Vc&z|J{}BuaFJRHmgLkg_7iWhqocv`xQ`Ss%wg{0N@(
z8+-d1)e?ElvFHy>)X=ecE?o=Ur&AGkz5m7&tci`9|3<>85+@K1QI193s1vN4d3Ix?
zgwG}-S5;H{7AKn5p}u4}7L^Kn30$YbxzRaiuxg(YEo6Z52Ux{<H1`eg%TfL@D^s{C
zrjiIz_os9xvh@Akx>ve=E7=U0JB%o~HsUIFC#y%|vGp>7ovyV@IH_?P(ONIrJm+;j
zKP~#(gMT6J>Dp0&UK{iLXXXt#YyOjwr?OTZsZJ>3-1AJrynTBuI}(=>=IxklLSHDM
z6FI6TAsYv|Irv=9s*}5xCZ=?!t=+n>#jSVVouQ_A)uIc?zZn0e^7;N7d4bA{6|bAT
zuo87uRJYle(1$R(diOe=_^dgE{#~m=@U4~luAWoH=enZK4|Z~8cboqvvH>nz$?j0o
zhQyVOy=Nt3FJT$Moy?C$z6nvoXq8pzW;~OaJj?mOR0#=(+@Rc8qXL;{$XzMk4AxF`
zcxXuKn5VYHMY%p7pBvfbPel~Bhm~iOY2Hvm5<AoKk86q$kVJWnb9Kcpz`Cla7Kf9;
z@nX@(aDQa6Mp#)`SebYn?i0>05|MuN0-0$mXzv!(!}*@T)tI~S8}xZrnDwI*6IUpC
zh~-w~p>5^`3B6zus=l%;AUWuMaJhZwOZ}03?xldl6|&<VcL<{D!FeNf3F~=7J?6+0
z^a3mLNJyC;$N+JpDYsPN6+x?d78fwoRT3~SN6_VX#Q(>ayFv)!92s8J`g-O{!t8+<
zQ)tNsZ5?6e5@+MSvqBB;3S)Sy)$pG2pgr3$|7|fv3pN>DT+6$8RxrAX(X2?8Rf)!T
zX9~A+ustss?)pxz)9WuX*h))}8oRDJPj#%iR;Cm7aEx8OSb@YCLSFpWJ7Dc-6y^Q=
zuox92>F8lhBe4|Zb`pr_V}vRgl_^Ycidd{7Da=>X-i5vVSP@%ln~Hko51Eou#POxs
zG90pcXP+;YF7ITb49H?MR%w&WVxvMu?<?b!An<+=xjG)-rGl>`3S)}Rdy)AzrY&^Q
z*xPGjZ?9vISxh2jEgaWG?>x#9Yx<MK<_G#-oBry0?*9S()v5QtyT59Y+uHP(@DpV?
zIte{Qb!GhHEoV4FE!EGuH%@Gk(Ay$jWqhRA#D2`iO-ftF8(u`rAM%DDYRNkrpkpPg
zw)BL*0PX875Q$KWC&AjD%KP9=alnq;ru^WG-^WeB$fLAS@qZ_m;w0#oE&SghQ%h;i
zqd=5qS!!_~NPJCDB1QfcddFA1&8+<3dS>N?ApWQ9mou2g*e?gPuwQPG-fd~Wya-!n
z)P5Ofuk72>UU_TGuo=r#w8i~BR?JB_P5P>}PHm4h{;(DE@8`%=*`}Yzl+k}nOT_B-
z2S`_dN7Og>GrSXT+!g-$FWED{pl8tzn97DdbGMY;%4+zp*)u1NkRJHoZO`1DD%925
zo|$#FbY87KmT$|Rd9}g8^<34%+w^DDcdhPx+puTG@iR^mX#NSbXCom1w58Ft8rMok
z`7Nx~;uI=oa`vyjKOg|%X5s<@U;|_E{?<4~Alt3^88xOU1b5nf%D7=I<wOP6GiY5t
zE+Xa;0#@|daY6<$oAQx7JV91ZV?&b=?6rByE{%I?J|1IN^6l|)gdY8iCH9{=e=ngb
z|E2Ty>~|J@O5k1*B5L&!ioo#b^cgPpB1oWL22J<+IL}Pccz%Iet^cTptjY254$ZAK
z>!wUua5IitTWjoJ;(#Kj&|s#<ZEg~*&U+PpQrGHVD`IGWT_Cwm4xR|-tO}!HH=~N;
zcDue=!wkk{G_%f8Zrn#BKa+3=V*1HC5-t}W;#=$HV2mfD>bk(7N^m|b7r87BmKNWZ
z_=tRZ9w8EG@oi%sNt3T}k7Nd)EAA%uw-h62rpcGOzoobv_qWI;E%~_UV(F~*X5jm4
z>DTB39Ye=%s|+0{fM!`O_eIv=Ou9<U#B+9PLgRslQlp!UIM@wO7k&+okWJ|c^3z<d
z)$T}O=#`ux*1tGQ=6el$yeNW2Xs8llvZ#;Pwx59&=r7XO2GARcbXHDnDhB0OdVmzk
zRzx;MKXLOzX`0tc<SH{NeOaSp&Y7BQ<Uh<PXjopu98<&Ej=TOUsupUxwX(H8dFXgS
zg5gcxP;?M!uaOLl!~Kp03#2hsd)hNFX5TY-wZR>8d#<GCqxLc`$w}$HMn1=_p_AvP
zOkR^CU(?o14DB1gF7(a#O`)yhw}(C-9|qh_W$k8U?Pg@{M%L(fye{L>BK~aXcuJ?x
z!SO2@56&=)*Ly=zL|fWq3hu%T%3h?VvL}a!xF;|R_4a!HZcaBfQBU@sS>rejkfvpP
z3_Fr>y0yFFznnsojA^Yxi#{onMZC=s44FN<bvxOyW2Z7>aIF}g1_zMubQv;w9~UV>
zN)ReE*5PT|QhxBS{#wnanRfrE-?PlFo(YD(O?5Zf4~~te#6M9(H3HSw?Q5pxgg}{z
z5vLGZMJX0jw79Ez`vj{V>ed0{5k=3$ri=x*RwA4^Dt1zCxPwximF*@C$1!b5lyefJ
z!cl$<itw;!Bxj~D+$f(E?m}I`U8q<Nwlgc^uCXNO+ICDg7@p-qR>lC;!kda3@Qt)k
zcR<%v6QF7N#R#$0CfT;K|I10can3aJI@GVkk?dNSfRUg5Q;Z^8)b*tLOJ_#G%ATY7
z`eVnjV-@~}nieu$XmqrR<>^Av9~i^m?1=PlrLRiYz<@z7IIOo0P+gJ=I_778Q-p8k
zW%(J0;yVWQZ#!A!Jb_S=SWiwyjara>y3DogCNX7wD0hN6hHazvMt?ZUD|i#pZYy6l
z{W!yF^-1u#hq`gVkiJvhpv+l2^ubJOBg?43G>#uqSy3b~5W){)uZvaB><hvrZH@Is
ztohOQRix1&k(8``u>$8m8hK7*r?V{nnkk{bmvPe8^Bg8W5&1>J@8xjsajG_CSQ=J2
z@jO4WDJkoc1lyLYo%*J($C|gW<FP5}MP3z1tw2y*ziQZ&<RDkps!gl5gg1S=rF!e;
zueNW+yzBTiyKCF8)h-uPZ@@()62!9AYLm6A68$2b3H&a52(ouA;d=}-yvH%TQP>pS
z(`8e$&9<06#uZTYG>gVjUG8r!Y}jq*5cYAo2#L=z&a~i;@KRwB1^^kC-+07r#d^81
z%iRMLY+J`U)71;<?&dW`uY`_|=MeZMNwX<Jn1f;%FbK|2u9S_Z!HV(m+qa>Wzmo(E
zs5mn395iAOB-YRV7OlMAt{vT=C2WXn*D_>_o|zTQF;4bSXi^HLgc@#LrSGTTBEO0m
zS47|R|3Zbd8Y&U491H6CgvE|MC#zDvk^%#8@8?ICMdpv*Pou5~r0)VCaz17Hv|2l#
z-r96Ni)*#QS)Zr3uWrx7nA)j@vp!C5|8e`mS)ugy8~ECn-hNyA&`-8(59MzOhw`>G
zgznwaR5<JX^uBeCg|psD@2m6m*Yv*ieEqNVzKxB+s-d$!<JjO8vp(l@;H(XN_MP=b
zLTKpz^4h7%H0+A~i)t6jkAeFiuYE~=^xgkV?Mi+W9@voH@590apQQKuwD3SpIw$=P
ze4E~HXXyR?dkfc{%$Tl+-r9eNU;U>yg#Nn!1ivnxevH?Hg*ToYFntBbDhAZ@;jNXk
z6~(nVAvx=UJ6C0jHna{xa8$@ynI+p^vp*?^``0D}cQ&0}lc1H#LH82P|2sQ-0-FEP
z6uw-(wXV`Svd?#S5_6U3@?T6PmY8ot*{+16UzP3>B%UnHeDB!grP3XO!uLt(E_p@k
zBqk8;@^@y)E33<PBplsbx<g*kvN->1lv64kL`qEsfo}sRcI?GFN_QCVt4p&k|K0t(
zA6KB3wSfYqdP}9s!5t5bO?s-o4EY0NyF6+B=<(8VBdPbJKw9y;(6Q;$*3|PkXiaeE
zjgGJFp>+)C!L4@2F0@GoYSaD~)~sa-U^MLgR`YnA@C==%p9meBU%W;S?sPl8v4@zx
z>Lp!X`rLS_PQCPr@lr&+w3nBrtzl8CN+dqUF0Hg_rjTvLY}k)hTcz6nqerEYn*SW%
zhG^$~Vszd_XhL<}h1Ev4Nk%amX?RsUs-}TI5v>T;5$;R~{;VqB%D#Us_;CV`j~zKM
z;HxU#zKEa*a25xNGFPkgGf80iU2$6d9-lD%L0K^T_p?OFI>o(rhTqgdyOIG^5Cu;s
z!=>DCoDvrh?rZ6+9H;R9aB-*^C$<#PVK6VTb%C2xkhXgKX=*!$Z9EW9B?Df{s1QZx
zMT(;G5tUR<B}gHTibv%o!+#3y$wg$WY}9L;oEYBO+3=tGJ$dr+`4j1*(7@<hX;!f4
zb-_lJz~+5BGCXRpqQ*dqE*IRadGKnRw;+n7cWWdc#Uc6F@NW_`I_#CrsbgvfNypy+
z86@~dg18;EG7n%wz(QMzyU7Y8Fs*)uh_9F*^so5E4Fa&yTK%S%(3Rul8|+g@3}>y4
zlAoQFTAE6S1nx)hxSwzq*u;L$vMm>-vwPSFnF8ld`UlxZ&S5_X_r$T1!4$Vg2MV<K
zn4M4NgrYOpKTLGWEu*r{W`umwD?3TOWCsHq2kr^7e}Vf)nv&gOeoRrH0qXp>j#9)n
z{cLKe|0-EkQ-89H0c=p;_&dEAw;vaCt@&a9uF(wff1rQYlO2=}BK}>=zrx-90h%Gk
zky6z^+w@K9yH>yB&;P1_SFJG@p&G*;@&zUS&)7r$J^o!^Kt^rtAr~e|zt!q)zAbym
z0AnubgH)$((-*1lTD|-KYGV&MjeYl$L^TxuUVC^3t^N1f!*~Db^!8nRf{7z_vc(p{
z@ph&z-ojtDUWC%9zbp?&qyDn+=s15_6&%lx|JZbPSS)|pj%xGrH2$)SguS7#p3YzP
zAXFd=iR?y1{beV#@^fwBFFQ=>|7txk{Besi1vmQ%_{)0lh~jANEBlGsP!RuJ?^=&?
z{|Q|2p>iifZk7v&S-QrFqOncSp(^HjAp23cydv9F={=8g7&hMJHQjPr#F-6(p8C9W
zzZhwFKW;l4v)BfM<L`n_#olc(efDGW14Zo06`|h_#v=;8KN-`yIEnXB$|0M~z=ZF=
zthj^leKEe{r|YYc#j&^Y74HP&@90|CLHIt$?37Q@%J#xOlvWyJ{4rAv4hwNU+o}a`
zC&)b7rk_teb>Vly-<~rUgC_M|tCv0pe`~3)pN77kw*8+$U%#dk{$2ES&yU2Z@gJqH
zP3wPvzU<Hb5PdzYCiTBaUv9=bF22hT)7Mm?mVb)AZfR~(^tGHf^=yM_>n^n(Y}21K
z*M$rJ*qXlB?{4GoKjc&swZ#Lq1^?pEV~l_CVDT^bR2={MfEnr${`KGY_rIK`{QLa<
zf5R;M-{SAz>qsl&Y_We0^SkHw@moUi{GoL35|kq?V4mJZ{V0(0+7l72*#=NoL*zQA
z0=XZx@mluX%bg`MKFVXqpY+&ya^(OT##A`827~rEHC#OLBH-Y+C-9~^qMNlP;%(ut
z;qe*AjwSY*E*w+Se2Wra1Fse^jwtJ^pdiin7j-(B_)aR<Xe$>-#l`bi{>dA!QCofF
z3GA&a%8s)Q@~xws8*OaMWPbyW%cTR4or^DXpv%A}(Xlt;lDz3ng0w9;hYXk|$s<v&
z$#7XVmhivI)Dee3v1F!eep-VgNfoCwR&zU%IHsy&8e@oH-zB<V(vGg-GON*O0@!pk
zpPjQA>Ht;^L5pP`;+NM`>T9!AasUcrQmW%&zf@A7;sBM-sl^WmQ~dbjO$cUkn5sp|
z=L%+E91y|g7U>@6|H6JvJ9c}zu@z-hfEu@~fOL{7hXrey)Jq?%_yeU#Y^NlB)k*pr
z^@v)C_o8yre@nsMGmcq9A<qg1mGZQtfK%+8`uJl_L=$tA%cihVZ^%V$EpgyKmByEw
zqnM*}a5wQ;-ocVclmLFHi057h>*I*wk@KVLv)Is9QUnVyKbM~YVxr4CmakQVEwzw^
zSFcb_piA8Pkz9LPQqlR+j|H;loW2y~MmFK`>H1hs;7I$VxsLLOIFGDvfMCM+@#`NT
z#~+jTpCiX5P12#9I>!cEbgu4(p354OFgaFK26xOlTKR4w6?Ein^aNfoq7a~FPo-bv
ze1bFr!|Fuygu=aaaCF};*IB31IcLbIokg|K5le$IiQ@#GR`wU^PaUc<;`w+vpFm`%
z*xgih`iCSEpEoe{cM_@UW2&xil-~K2Iuh3#iAXwD?|wO*hy=1otrW`3Fh(!=dIeZC
ze_MLd^m`;lp;*TbgrIQ8P{@$r6BbW=MSnjHo}wiNibqTo9YCRXQ)R}}Z74iT3_aom
z*|sY0hb69^O0-D-6S`M}m}P?K&mE1{w4Yhb9~rq;{fYYp8v~I18@$3k6O3c)WmuSx
z2npu`TG&04TmK}4jqFRU)3U1_>i#iyi$^-%?rwLL92-}($sM>cfuvb$^Bl{QQ}O~g
zzThf5mO7`~s7cAcawn{EHE(y-Bxa6_yyGfs>f^nZ&wfQ61=KI{f=U4$*C+7mLQiG_
zr9}b)J|pt5e02#I(^UD@DLm8s)knEm3$F|7N17~oI8I(H-CK|(BlM}}!zJmYaD$18
znaiqBKy!^K9@1awL=w=mL=J^FJVA;^C<8gSGX?f6&jSZ){94Q}sMA(c?v&F{HW!-~
zDm18L(b8VL6ewHFT$M8J=l2ZtyWF{vS9lz4KCV@doMa^14B*1?rMQ(wHr;%zG#*!)
zkH*qgR!^ytSvOMGnvZqH<4fkFShc~NqTbDANeYx@ns^#X!rEDoG#mqHsA+z=GljNm
z{-m3`v#f}v7woT*$)%3)$dQrQGC7)#yP7hn!T@KayYgrA4{*AgYx0PQ8u;Kmj98Ou
z@R1GS_Z$(|_5#_@+jv-3fo5fEa1j8>w_QS6oT<qDan-3RRSPUqeXbWUX@w-yI7u!}
z+<_~efFxzl@7qI0%53<)ky1*E{)GviV1e%waHR=OweZ|7;FnEsf(7m;tl}{ftihuV
z+}V=*ArqWsfiD*DWE1>V=V<P81e|Yz*IQtRfUh*cPAswp?#=?fzy!Z#fjbD;Zh|{n
z-~(BJ8x9(cueWgjR&sAP!KGI2y^?!{3I3S{-Y($3nBbHy(VDCm@NZ1;AFbTW1w79L
z%YK7V@4pN9E)$$#A$eNBqfPLc7Wi=iXPe-$7I>C``<UQi7Wf_kcQ(NfTi`ndeCUAD
z_|+D8qJX!W;I3Vx6}&~jH73|@fkz7XRTF%<1-?qaPnh6I7Wg6omzv-wEpSf(|I7rh
zvA|~v_+}GqcSLKFEa0n6@HG~=lYsl0V6O#E5^z@&{E7u`V&dvY8;r*9w7~TOt~0?s
zQ=&!c0$yc;Z?M3-1^k8y-erNm6!6m~c$5WRDd2Jwe3b=WF5r7ju%lbF$iE5r785+(
z0zWI@YfbRQ7I=|>(@k)93p`iA-AwQt3w*bLkJTHEpJ;(^6Yx$GJlg_aC*Y4v@XHqX
zS^>Xhf=ey%#RC3=3I3S{?jzt|nBZSo;By2#!vsHLfx8QMtO@p7;C2EYW`b|Gz(+0x
ze6b0hYJnRB+}#8hTHsv*KDpm${5T8zg@AXN;9D&4N&&Ai!FO2T<pN$}g0Hi{e-ZF=
zCirp-oG11Er3s#AfwKiX%LLzLfz_amGr<cjaFOKBGr{*+;P(Z5i3$Fb1%5-o=a}F{
z7WkBaZ6^3B3w%Jp;ctyTDYL*=%iLLSf^W7c_EG`AYl8n^fzt&1f(f=;NKynGFu@)R
z+)lu=O>nG&C!{75Oz@>vZZ!cl6P#*+cS!Cm6I@_{HwyS%6MT~e{#d~6O>nFxYF_T$
zXEZ+6-)dfdZi2hA!)Gw+C4ywR366Eu3j%)81W&Vyd`iHJO>m#oD3XT-{D2AWY=LJB
zxX=Xu<;-aA=>opq1pnCrPZjVW6a1_NzFok*P4L?m_$C2&Fu@;LU`3ky_o~KM3_;mN
z=<scHWq+k1fH(rL5Fs>w(yr_L-fxh6-gAm1WGIK{v6zb9req_3j((&(aV3dKj*4`!
zAXq0ZMl@QERLM@7I&UU7)!ncE{Zf&k3{3hh6dZQ22riDwI@uH$m36ehUz83nT9p)o
zwbW650~tj4*}|*J6pI2$C?>VlmJvlP%3mrb*DG1{M9?xS$<ZIo0}7=&PIm;Hvg_uk
z_=;aBn90tC_HZ4)IsZsy3zPuA9?OXsAVzCGu2*kNaV&URB)g#{<a3m9^%Slj;cxhA
zVZVsB=k6Xq(4Ie$LE6*#A8F4N*0Jwy4~7Y|Js0vT+8%6}(w^uAGWtwtKw3v%M78ck
zs)N)iM8DD1`nWyPy+MDvxw+}vuljVEDBAjd5<S@*(Q=f3#FeNnHZm(#A%S|c(L$7Y
zTMIm{`q@$b4jFtzVK+ux%l@KhaQK45*qYQo>wJz2$+yLuE#7gI?>&S%K9Z`3cknCP
zyPZL9Q44Lt2>((l?UfE)MkO_Wu~Ta$A0L4Gg`WdStV<@)_dnbv?(p|?$(OJdvrEG8
z3JxR^*Va+7j^ENf^b33K^5EObWnC!!_M${-vu`-3XU)DrPG3Uz%Aq{F{!{AYAD$?u
zBAJMf)^jVA-j70fA-t0Z@kA=yQNp>UiCN8|lhc!aOniL4yWGV0l*NS&bJM+^Yfh@~
zG1r_@-{Y=nR^QHR3LEBU%vZ_t`}0lm{EPV}dH#U;$=7IQ`%5%Vjx1H-&eu}hOEVnh
zvZar^L{f^B#PuSneA9_q+3tD{WXW|uF_$39v3rxdfv?m2<Zeb7(J$0}%Z~CA@I}VH
zmVlymb@&_}qQiMLY^n8dV(wDI8PR2o-gAng)oMi3=$7!B1N2D(JN7s*RYj6jY}ijY
zgb4bQYbF%!ujw!_vPTBlUQ=>DI1<9Y=Z$E|)o|?7^SR3*m(%e%HHMCg3B34S%uds5
z`b(;zF#7#MnI**u&1YzVXVi=<IzthtJ|RLGSUg7VrXd}qAx1qfQ1v|A5Uc086h{x?
zN*IER8d!K!3|Oi9OT5kl*Y?RQ6|0eulb+B=Dj_GATo}R(S|encqx=~V>j$Cg@MCdB
zyu*V2okirGzKB0uXca*yd8>$`=fZ>IFyH=tm?OJRTlw4==KJF?J1T~=u!J|)$GWB&
zA^tn-9XLJqU1?UmIk0Ed+fnfi-cI4-IPA}TANEV5*sVSvsHzRIaY8L8vjys33hE2v
zP>=XN)SrE0c8i62y{fZ`dQc10mkR2d{jolYpi6ycpFGe8^_+OrgIl2f3?pdx9#EU4
ziLo3P-P_{Injcq|R94np_^?5mT_}qOj`^GwUq*4bk+n2B35UjEh^^#8*`It(FPK6D
zIM!Ss!)A0r$wH~f>js)Q$)z5JHr8}wXpOp*wnVld9$ASXo7@K3pS}~>pq9w4h({(n
z$>DR_AiL!|k?r0Y-%Mkz9e#e4L`xS7zFIn-4vTAPhwsGqyB7HLClo5P+ZGA32iqW9
ziZFKiZo94}vg_lKJub)wwm~-SJCU8-5#Lh%RPVR}eL|4QEtxHwdBJxgd#wdBquY`d
zMqzrT&j>ydMqA=rgSqGQEuGi`pMFPt<z5hE!`mQR@SVucYKg3SJhEp6nMk43%o4e;
z0u@q%RW;a>*d6tAVKU)xL{09v(YJqR$xm*N?_;A|4r59**8*J0qO&3<cXZKr$}r2^
zVvBC5SY?;Wha<l#`dXy0Wk3Fc#m1_vIY*6&mfE5MdQUOsMrZ9xS{<u^u~%no92Kex
zVEy!M4IkUny16~Sv$R+0+BUaCthuQX(J`!EUWbvgRWFONSAUcF94>29?DJ$leIqYF
zO|g%jYxQ!k@Y!Nmk1P4swk3<zS^u$=+{q}}s%HPTnZ;tQcM-N@2np#t+rn$Z7HjSe
zAR)YzwIIIOI(wT2R|=*Tba(i%*xPC9?T2{V#NSr|eij2<r~qSQfIbSK#Q;B2fU9DF
zUJ7tg43MD!Jz@Z(6^Q`iD}Aw1Y0cle`$=R?f9K`5&{!3yh@GhX^HFI(l~s%ccU*#B
z5|)t=UaDK|=&y==o^>p~gjM(~LkAe@-#8&9*jtqFv>>`q5X}Hl{CmH$-h0q|@3y!K
zd@S!}%X@=k6}VVc;6f@uX}RWGx5vHJ{sP`=FK?Y7ObMATFf)yd%XGbD`buc!3zdnP
zT2nGWp{@Zy9pjioM1llQii(TT=`Jg3@mA@!_srdqH`QvbAH@<8o)SYV96cdCJqGBf
z0OMl-y8`65DR}007TjJ6ww7{(D@_oK!4ro1=POMCXSfb$E&X>-oNBhT_hQ-AF)R8v
zDk_R;NppA|8Cy2`2)k3UMoZWbq~jd@mtV#THgs8#a6VA@GioBaa@pEed)M2(lC1KG
zio)SI;(Cx(wzbkN+l8RW(XUY57~5ZVR6Nbv5N)62IY;j%9Ks);XHKg;se11%t@GSW
z9=WdWoU+D&-cDtW7dt9))N51~5$L0{&21(d@`YR!evUMP9+Fzkr)4=MCkHwnCvZjj
zBB$gD)!%CDNrjF-wyLmUcBd2xU167y&8=F$J#K__Cl3T97A2Hv;;5L)?G0g$${3uy
zrsN_9%NF)X&}f4l;q3v<42c1p3eX=w{QS9*z3o=~cP$z9LEE70_7oDI+1WPpM#<b!
zGH>8Yn)u8|Mc|5GKZ$#-Z)HToe<7!F{ps?}(%p;HhMf38Nbv0|)-?(P=&^?{{r8Mn
zcqYYSVj_S9Aw&5sVJyb*rhWsdui@)&?BGdM>NxswR4jvwh2<l<trzKf#K;Pw0;;EU
z$C7tN4YhgumAoq)o6`r|`6_u=*eeg8sRv=cJc!VD1Z|lZlU!d4eZ8BahM-JTN@b-I
zk4|Hoc{3N(u+7}BrG4e!yWiZ&*l#|=+HdZ1n*C<|{*5y2-W_c;tsvU8*Q};ZGMd(l
z^8)_~?Rn`7p|Z03r6Hv&W<=ZbN7bH6JOYlMrG_e%_T(;1zDgo{n*IQ5Wz+v54rUU{
zD!HSpq-k8yR(D{u@&))T*ReeDoV<XBFF;f3oRm?M5`FH3m9FNxn#8NdMc#9jHTLli
z;<H~-2O#|-uLvAmxtuLHCwF1)Rmu_IaiE<dkMMa$WFenjA`AHJ6MjG}w#1y2YxZ_+
zkQThBwvV-dxrMJ-<a?WBtjrz$8?YhB92hN3bE4`0Lk8wZX;7R6;`~ln*0jTFrDX*_
z|J(%I@*Mg@E$Mnk#djZ9ZUYj_`7Y*4<W}A|TfH&(I{z3kLtimW=%7-6LYO9XcnJi0
zXI0gty@v!hx+@nMkn^#gD(BBt&KTmZApU^yu<4(vb_{)u^WJ)+j2Cw$%Fcwx`2C9f
z&QWg<T_kxvGvH4MJl%xL?_O%GlkRO<{1N6S72PlF28*b;Luh_(Regj>9Daz7rfolt
zzB0fWK2JdL8Sf-xR<nKywG4Nr!eLo(!|nMH+jl(M!PO^S%^$Nr?x8~At@6s}^08Jv
zR?*^cP(D83qcsLc#q*L$;=hHT<3ptz*u-2^Q7#F7`6%NfGCy1_zhy0p%!=F>nI4`d
zKc|N86tH*)hPl(n7CuQ)k(<IdNaFPocVuL1yqy2Dh4B=yg)f#5jEc6f7_h^=<l`JZ
zB4>q#->Dvmbc`g0+wl~KMo4oB#XxzEimlSVujS)Q`S?OU*7FgUUo}U3(ZcV^$D5}~
z{hcKIK|X%lCbej;I>{fI-x6ifl*o)$kTH?*k+A}cQt6$a;ze~`BoA2RXOU|nS4X(}
zJ0#b1+q_*PDUkuuC&M~?It?jV&w<n<Ns(xa7E-~Mlk)<Z3lyo`Z{9$l6iSJooTok)
zbIg8y7j#9)fLvD0TW)59<5Q3SioymtT|`J>VnJQ8{A@+Qmq1y>iv2+Bih32;iDbe|
zd5hHLXej3N)gW^t;m?vV*h;w6N~n>9Oe^7RE8#0i7-c1#q+$lD?hrssxYxh767nQr
zg7wCSR>C4l=xZgsVI@2%36=t(KWQa=Aqk_cH%hDo2lIFNcq`#9E8!=Spjio{RKhg`
z$)wK~7?b{()lEUkjNWf{(}fn6Rgz#SF1p=HI3@`<S#RuD-7h^p2!0mVgbh~0Oi37P
zz45-4@b{K)ykI3XNWvKF4ZoEzP?&OT@Mc;G`H~Rpzp++=UlOjjP~}(&%O&APE1{p2
za8we+t#uhQ8UG9`p-~dDt%SpHF^JKBvz?AT`#eRA&q2HjmJDo+0SiQ7DKY>C?13n(
zJFvG6*qsK9=orH@B|HWx?O-#nCFs{jbK_!m#Ry>V?W<zBi_P3KjNHAgl(G^kYbHKz
zB-&z$<*~#sjKrOv8MXhlSo0|Ag5IjAj|?ysD-+9{iSS+$soSfu#3y2je=!mtizSK~
z&U|y1kyxY>g>nX>k}hZO)$H+SEb7dagvnMynM%0E|LbP}QNLf3OjitumGVNYlxh;G
zpCgv|Y%KARk$3=>XH4bGQC*i`MFwNgds<bvNfKg&WwR0%OG0e^?N%Ktt=S+6KebSO
zVkPuUBjFA!;VmnnP!cY)5}vUViX|b<N+`1u$|WJjl76NVu8|R1Np7>B|3)Typo?mx
zBn}^Brff4(4z5?@Eh&S|ly<Ccyt7HANNcVHuzadj1%rMh;bBR*(n{EEjphrIP+%o|
zVkLYn3Dd2Fw_+nB%Y<2hOZ%t_6j&9w89-c9=UeD}k`U|8DF!;t|7%gwp^gtphdR0%
zjr8A)7JWrecASA7TW3yFqAXu=mjT-aOs^Vq3TJRH*DgpC|3FII5GPWC{%DIFTO`Nb
zR+DyE^fbUpLcW!-%1XFH5+1Y?Ub7NbNWu+P!tbnvW0G)%l`zjrIEN!<%k!)RS-*{X
z=So71m^~yQ%-y>uD%y^d^hwr>1FRQcl!R-ogtM%KPb48W5sz7;b1ov$a!iWmyxDGa
zHiNJh7@gaLbUNI(R_bOf(i{yV7}EG~k2T`Ln`<Lpc-dZjP>*Z)d5>O>lo`9N+E*Ri
zaq!40iDSe?mfQiB>?8a`S!1U;eb7<f;qp&yQer755}NBY<3{c&+*fe(^zc4aJV(=f
z?FBMsdU#6=B32W7ih9No-%uxBWcbOf3_To3NP2`8OB5vzM#wiNlJ+%;c77U1+i)pY
z$iaQ<)H{D3el~_(iko_;E&N&wkZX$fPHNJ9rj1i_cJz)msHC3mDG|bpMls)g?~Li{
z9k0BD*~(Tk{Or@dr`;(zpnGzh*(QfkqX#5QqRP_!bXjhu(vk&5AeJotzA^u5&X<4Y
z@_piIPskjiCRXV~9Tk;iC-|_OV?5qEe3gV;8N@9UBF~D6Li2TZ>O((*79=uTd+xM-
z4f}?1AbG@Uq0;VLYI-=*g<h9?V~Ea3NIs7Fa&Z10j#&*Nb#HzKcj8JAopvaVoJZr<
ziDUW`7eP0Z>gA#tclM!KT|60_Xy2yah+9L%F}n+4h29~ovX@9$T(sWyTzX(=!y(rC
z$5u(zut}=WP)YjC4@DL|s^12w#{36F9_7>+?bWkJpMtJOB!f1QyK=hwh9^sq&=D-)
z-55>dAbfXsKrSK~o`&1Y!>J6?!r?jcv#0v$a``*>+RL--TP_?vOn!>VdExLp`6=e;
zg~Lb4Pcd;W96nrriYa{I@R7vZNf|MSNGwE3ineO}3T0l2tFN3F9dk36Wy%U|)LaY5
zwOuoBh*m$>IY!k#*E!B=0Q$1gf?VfhqX}|#9)+tm7_Ra1sM=BP{1wQhEgYQ^sA^5Q
z^D%+)s9IF+{EhjjT2=1+t@)^0R_^?@`KVe~?i5AGY9S6<ZvS${R&ecFOp;hYx$247
zNTd1mIA?IJl?1Xm(Jj#_p5tSFe~McoaVq}>nT6v1f%kBvlUMJ4U%l$(V!)ugY`JKt
zHgA8o|84Qjcl&=^&lf>#ydC*1-uo{9C_=k8>#l@m+}Mlm_V|7zw=&3>lm?~O#mb<<
zmHCst`C6F)-2O*YV=x)8E^-6y9elwm$3t_<u<(BCd_OhvWgGdflzjc*>%TGb!EC1+
z8B&c5Jrwco*5_885?6e^=)F_w9FEMba3R`!^t@6mz5@43%_ofi9CZu3rebd)-B#^r
zk5%{2HL+tnoYZP@3;Q6md12Sersl;mV+QlXyia6FoBKr0BQdfm+9l?BJ_wX$KwT>0
z@n5UY2?{@6k?T(rV>-Q!^%$npOc;-3#VOc7<t_~G#o8n>NEKb`KdbCmA20E?k9fIz
zrz<z0x6k$cD0&XrxD_NY^dTf4y)#W=d7m!7o0;-uLu^K7&Kgdq*J4vgtxt;osVOfN
z`)Y-%Sk(KB&>7OXjWkK~CptYDjovhWXF2iG*Q`p1id9AJhzn{gd+K(Qz|cm7M7=De
z>akaT&oS#EIjM$D8>8bRywyL)t__)!!e!-h-;%GZhrscHVO*_N>Fws>x*TlyzELTW
zuHkLyyb?N*vlHDgpI;|%{@Q7^!|c+JHN!a2S2og~w!(_X7ae~U>G8N{5{H@QEB@Yz
zsK$i<?Fp}!v#*@<|BPjj_$+lyE3OrsnkxxMk8<x-4DP*&%H$5Y&+aInOL-DM#ovkG
z*WMp%lkTG?^_qzmCqDzNoSQ#8a-LNm11V6B3uPZEs*DUau1HLjVKb<~<#+frdu<N;
z?o7TVGCg7UvGep{tO3TcM_4n*6=Sz4QJ?>m_I4NY_{ZAY#HfE~d)wZ}@|2of++Xx7
ze#D=rmH2o*qO5SYA*0PjjM!SDfuZ|W$>JDNw;jfZ^pVq0>OwSKMd9H_0=Y1Y2ZHDr
zVSNbjK!liRSf6DwSpL*}i7@WJo(cQ4_W?I?Na5!82zQLz%-A`ag!2iInQK{!F7OPQ
z#C2#Y5+ep9$a9C@oyrS)e@P)cPh#EOAs89)XtW{X`Wmr;K<0fXV#Ifd;V;S(+#R*D
zhY0U5S@SC#y<6c}8;@g+Y$f5IFj(%)k$o`3&?#YkxSFL({D{-^ixoMpc!O;ikrv})
zIS{#re6{)$6Hmu~e>lF)4BFf#zRiD8k5=)23A1Fy|7~@D8AsFq#rw<X*?%$e@6r#H
zAf@gvW0P5bQV}5WAB4n_)0)&b@gF9{$G4FM!R<TrK_>mFcc_-Ev4}=9h-fr}h(=dC
z^Zk>BAox(RE&4*Ft-@QZ2K_%3ejr>)VT`36v5nC-;u;yF8M}&5g|*nW+-Si_W`|yF
ztPT2V)x2$bAG2AnjE_ISZpH5~MX`VAezuNk_8~Q!;_^B4=>Jsw9vRpF-SLdip@-YX
z@44fB>CIZ*$F~)~XPPncx>xn?HvJ>@U8|3|UFJ}{egeTj6~yk}AaU9e{+c<_pX@!i
z>|}FM590NwxQC2Popl*{gUAU*XGBK0eEVdXzNW0HxhNZHmmDKf9cAK>!AZ|%N5w7t
ziVT83vC21(7WFXM<qKt1N~*_^+bDRddE4>00y!S9&aCDbBX4PeS;$c|N8*W|Emi0i
zt%9Sh8yTnt%cjSF5BI9vnXn93OLhtB(NX+2v#@_Q`@*<|?c1#NS*!FjrRAKoPLXQx
z?y!Xe($zg*9H);7Eh83+cfdWOvJR`sO&627+48|!b0ybMqnMb0;dDisbih5*0aIcf
z@EnzKm7Qqz-W49BxQo$#I>0PFiel#bR^PaMLGCds62;A^A6cb)UTCF{M0@K5omM?~
zt)u(^g-IB7$~WuPLFyHma5~gh4I(+l0eeQpf>FMsA{W_K>dqN`5Tiqan@Ys|`QIah
z+*o7wzB_PltKQeLKX*L*7Q5taMqn}1&?=4h!;g~7s)saNRiP3lClum6m*V3I!BKIL
zq_8*>8SErJ)`itvs|2AmlhDFzc`+uBy37k<iC;p5h^x#f-daX|tLu_Kk<sIlAfgJS
zs5_I(8rv-eW+Nc&N}_hNM(rUI=Z`h+J5c|)eJgJde(D5Z-NT7{AzwQ;m8<)_i3Lae
zhW%RA?j(Kod%{zLa=+`>?y6l$T&^-G<$7N-bLA%lzRn}wM77s(eP9rQNzYcXAiAl0
zK_%3L^6NmK&$nOi^9&;3kc#K6<WpA-alIv&H*p>A&ga4%tmY$ABCa_4cKh<{kumvV
z_0<~rZq$4^^@OZwq+PZe8eA4i(*jzW9B%exle*H|Gv9YFZ)x>}GcI}`3lBjy?<b6H
z<UKOitJ{s0c3c+9B;u<SXweJRt7m9bE&U@YFRwA*H@}`cidA{8nvSYKcVJ|iyh>Pb
zlue^R2)dq_-9_(Fpt>-H0^gMa<+h?E{mgfTnh+>FzBG@IAS7Hb?Me7X^G$7iVSs88
zWc*QN4VsG^kw^b#iD^HJ%Ckyx`=R?!8MXh|`vYdZBdfzUealkT+FNJ|@_;^GEg9SN
zx6P&Fnp@-7+JE`}fZtJuvQnAr?Tb?OfAIYQFG4MC*V~4-q!VlPo^Qvkw<nevJ)5X{
zc$+>>eb?$=|KxP*?Z4}Ok0T5SlAhw>laSL8*0^sd(&~PXdKyS9arA)iyWe9qxzI!8
z=EN5FdmNxAtouFgq`-gTevfUtnWHWc$Lm+6dzGJs?{Vj|aH`T{HaM|k(Rw%^vY<#T
zn5ZPBN~cA}PXZCACarUDtmy>Ni=O2L1z8b=hy$I1ycLDiAzLWOi&01=kgP2PH`IKF
zmtr_XAyNB+oV&Wu>vXi^h*ppNcrz3p+?U=r*V%PXVMCFmq$HBk`>;wmncnxqz2+O;
zIsor`Qo#?W_YLkPK#uZw5uT<6s|Y3BA;rvlRyz~xnNh6W6AKT<W9!kO@SwtCyqM`O
zt+ipU7KgL~t(l8)r#VN%2FGYao>6_ob;KQUl}Vu`%f)6bfmH*GoR##DM|^$6jwi0L
zco(h;HD+KlBqelgLKU`I420Z{MxJAYTQ6`2Zq13|5hQC(B$M?U1cSU?V!SPHMqi(-
zULWla{H}!ZTVkj)F-+CFL@~^0i(#t5pn;*X4TgFX!!-RvO#iA56Wd~#rZD7!!A4@s
zI@lO?>53V;FN$GwTMRQ4h8!^Dw80=-X@bG4kBnlFO!Q)WKX?^}tEC^>U=T@IFx;;v
zMKNTy#c;pEFj)Gb4Tek;L$SVsZ4=cGX>BpE3C>Usl71k6%YMjF7#KqOSf3X~lF}N9
zkwM|fl-_8AXS9i@T<}~O#baxWM=~fp1Ef#d;F)ORStxkEg~nB*>eWke9czT6QFzj&
zXWHPIVd9ZpdHv5(JaujH7~A&rk5l@G$<nfaq=ue=VN}=0NAZYty2U^n8B~4Jq<`Ar
zS#08wy?*`dD4xo;cnoI%>eEa5rwyKGO*|ry>7S6S`scN_cqD_W5BEm=uCfiD*GxR(
zE}%awc#cg-W^3S*F8P0i{XN`Xja<sWn|TVYAHPn!XT=y{oa^ySe#R+NCAjmB3!ic3
z@GHgmDYTnr{1hV0B<Bl?u>^OHz3>IaB*-a0XBau<=S0aF+<ANImIRewde+EP$uBd%
zG!M*UFRV0>3V9ltZ2U4a)kibE;H=}PLRy&++B>$Azfai(fi;plw6}nVf)#}a=A~y<
zs?YVI@U%*{cIGE`2pt$7iu8)fx6<rrfG$!cY~C@pQwsRRQ<2YcPL0pW&RjmHvE_0s
z#p}G9&tm7`(1WniL3|RHo6j-M0ep_*B7g_U@8okD8!``)zZai)B8Lu4=(6_r)Ns#)
zRiPdHZRT%1f2;Yc=5HB)Z}9gDf6w#x6n~HL_e=g3@OLcbai@%@LU62`4CfsE6|^!^
zoP?ofvGClpbIg?CzTAsO8~c%$)XknqYL98E-0?vf;>ys8v5TA~g6PCpSqQAIquWA<
z3J9rB*R3)-Fw``Ss|@a|mw@bDP9b|2{`#=JZ{kIJf~EB5MWM~(xutK8?xYsi2gY4=
zlPdfsu}Enn9q&xOXtdlJI-0<KS^FYqbKboyl;Xyjw)CWotfRb`An-27J0}FkTFfE@
z#kvFos?*_kyRxuho~=LL@MGng$E4rV@}hPREqA;#gej5e)Sn}}t8!l=J8Yp?R@J~@
zRdIpm)in7^d{rRbouLy{^w9X#G!_9w=;TlV4+Yf<AbcToVhlOPtmbBe$DMUdpWu$7
z?A5%TU9CXEg8&U(ZFCO_L#yNaXny&CO7MnzqDr#!CBjb3zn1n9LOyD*8FMq18iwUu
z?tm=5)a7f;4`^J=Sa~Yfx3a8`kh_f?XPFidCe5d^kyr65<E5gy2_(FlCwJNQM(zQX
zXKVjJu20NekSJfC1@_Rf+t)MiSA-7mePBYR0ZwlDn9}mGd&@_tHvZjSEg#d&$Bk6p
zo-Zfs#e3G>TuGI^8KVM;3*;81IQ<o>Nl9F2)@zHqnH!Q;7VTktHy8DLu&;<K70r%E
zi1TDaqVcU^@2&rTLElP@xwSH!--U6KNxHfJ8p(L345T!2bN^kT9pg4bj$3?asg>?#
zZt;@KqFlZjSJ^J^OWi_$_pL}u5O@4MOwj|BJt4HL|H<Hv8(-!kiXZpq_96GnwRw(L
zoAVNmxyx3x%g_FbPzrRlbliZaLhHxv3GEnrIJCOp5C@14mwv<D4Ytw{fgzi-4;QzS
zxNG}eRm~R!ci0Z>!$th+p1Ij4y&W-C^>n?SKw_@r)suqD`&43R^|%8dI~A%fI1#LR
zVBQH=Dd!;O+gznV!RN|8RNUU>J48+Le7hCWVYUM&Ksvc+LY^EGchS=$78dtR0Cji}
z57IH=i`f<m9UBj`xP22XPFk@NvPlL=4;U>+1-@14;Cw2ydE9!cusO7&p!YqA5<Fnu
z)|PE@<bUh(ee0?^MU(8RNrPz8VXH}W-`3!cyO*wTyt>boutz!(f*W65*if8$YmYUo
zQ7&J#t87QdvS$0N<WSwM)M1MvUSnJuO^`*u!yIo9=n2+ug3sUmvhyh(r}a$Wc^`HB
zQee}bN>sphk_q5e0Zx9(u7K@Z=1f+=#FjZz6fmh}&h84>p=Hig1?<=|XD=xe+JM@>
za#ZBg?fLA5UyeIRPcF=T3d@>U{4D#)aN2W?pbajdJ?}~@<j!|uI8b@+uiAL)*P+d~
zZ+0E2;ta{1?A$SCJsxHAKeT6jCHJd2%6CD2Er@tvWjl+A`-Lv+9|{(_Uk(PVF6+NL
zSnYbbt~LGs&7|K!PeF1AVW&cJCmHLLEywzwq2FTYw|d<1(1T-ZLq9GE1ziu!3;wt1
zw+)4=8OOXG3$H9lT9u$EG}N-zKS-f(&_O%ySXvjQ(98N)!zEZ;pwQj}al&Xpp^S%6
zD3iwQ^hOyXK<51)kOcJkM*n@m=kIuVK$Je;=zlo0xd58I<7H;HBAa%S3E(aPPP6E<
zMb0RFw#XT!&lWkO^w}b3ls;SJ>?LLXv-BzLSwMT<l~!2vxtY7w4Eo$+&}T@rxhP$J
zYju&yTnDDJq?^q3(x`s@{~`L!Xr_0X=KYV-X8^CD|DQh9eESFJv#+S7-%XzjPpY$c
zviP5lK4<=4&Od!8{^#j)-{SuZ=<`54_x^vAe^#JDSoFC^@lO$s;Gd$~_0W$AeO5-Y
z;^c>5)y)3KML<0fI)D&?e7ajLWCT{LLOfk<cz`JJ^p}XI5f)1kPY;MhdC*n5(~u}v
z7!u_{kth%5`nFb`ydXGi^noq1nsstzN4$w67WV2yg3Ix0M9HVW6!|nNEF2STr(C7G
z3~V(9wj+Y=h$}(Q^&J+3)uRu5i+Jkolz@cjC?AGsG;HCtP6^?`{8H;eC8B46pzr?6
z(4KLhQrBu(*z21RuqmP1Rl36{q1q_nK&~TyL#}Uwh@aKkfdgRg-YGYGgEw*5!a<$r
zkQ$LaS=p&Yy$C?)1?I}$<K{<DA!^EYDao_5h?!Z<>Y!a&of0~;jQvFRS`)M7)jCkU
z>|DlkZZ_LvGyCgG;siKtS)zgxr+sGsIwf)foxIGhpzT}bO;*svR(VqtG^thI?h4wW
zRo+wu?bs@>k~_IE)~;nYIx6z$QZ^sF{k4Q9Es(e@YrQF-pmTJiV|J_7ES4t7#kbiT
zicpVG3iMmohxXjQsjy*k>aATeaWKMy)Y)`Kyd4EK&fgh-f&ydb6vySLwf+?-SUflI
zX;Nl-oc`E1NXZY}l@Q$VKx)!ftZiuLI0A9OVBy??9!|Ciwf7-)8*kb|^X3y`U#!_F
zW_@MvB2~L*;~0zp#l10F%+Xi9CI$@*WXTG4lfFa%VpB;DJutoZxnK;<n+_g#z|%Vg
zrH9gj)pN6WkgqYlQ{`)1Zv~v(J3X{+|GunBoTfzocDy~a*>|*Tw<Gje|D&O&#vZBq
zHj$sFpvC6U(pjIU1zkn!R|Z{IY*`(275{XV7!QufM{}?+|D`p+x^)e@?ro7JnlLKO
z@%Clkg*@oGv&Bno@s7#~x^7=z8FUTZQY~fEf%z}hN{J&TY6(pEZ%abYo!W<p@hx%N
z+LfJ3bW{w44yaw0J54@sOlOx{{TiDo&vy)#wztqnFqR~Dlp6-60}iMT2ulYv{m*oO
zpilfR^#8pskoL2u-OlWTa9NYte%WwI73Yq!;28Fx*&|7GIx7Bz#{m9k1uac5i^KPz
zv&Zq*n|+^_HX*p#=FuzqC=vr@@>sFO;a*YIXuqd&jP}G&WB#JC^8n&f4o%CI4k^*;
zk=-@6nZ?ewnj(B_fjNmdkxx{_uI5R|XU>y}JQgIqEsss}apn%XGH=B^@wIpw=BSgh
zGJG+8Qtc?csSK=Y%j)VQSmb2Os$AJ@jr}F=?!fY)Hm;2H*jCe<>c$PVcjZ#A8oN%V
zY*%$nGm+Qxvaw@Tt~P_BNK<HTi+!s%otc2Ftt*8j$K1N|;Fy-?ZVrk&ho+U*sO_vW
zM(XFDA(NT@<y|o4czlhr<20HA<#_dt@=e1vA}0!`!s-*bw`Kd`V%1`abaG@NSashC
zE<1l<z8w=7w*j$+62B4FQU2a&>J@4nLv+x>hAH;yRS8(XJ}hkbd8d@MS~HOmYjRZR
zL%A8Gtdif&RHzC+QeMWl8V&|Jw-}O|+*R8<1QW29INViriDJI;I1&rD7uzYZL*eVa
z>|?46*PTxh$%XgL^QAQMF6Q9E=e+hYs|#<OZJV%7-pO?c(sE}dq*eG_vAlHSJllj1
zjkIKyRwrp$q_KFPF`>ywg1_;an7&og9_+#Ujg;;xMXabQWj$8otwsv>hx1NaG^HB5
z@~1{hFO`xhDX&XGg?EgyPdH|zrKz->Xxfg#gz@%hSs0qZm=aA}U6?T29xW|hrLC13
z3eM`ngn4%Qx^$jBWt^k4^i;+;XXnBTC)+5mw5x4GlbM!mrk$bEb`(B$zHP#OGp&o6
z)=8zUFMRG2+k|h-w612FU8PkQK6j06!j~$|>u{(vud`#}a~>)0xly(Wp9<`f9=AMf
zvyBRrCkn9@b#VBfM_DL*y@Mi-#$S14FfX!KuT3aiccs1hE51h9tH0sv7JKy`zM!J|
z1SrT}eJVkSv#^0dU!Ba@`%=CvZ0KOC{)i0@@FYTvli3hxTr4;nrgW0zvu#3yl1xH6
z8@1A14O7lYVJW`IMyWu{)wTxE&M6xUe|4peN+_rr{DvvX0v%zi{@Q@5fp3`7MWDCX
zsG`cN2EAcQSAmYVQDX&F1Ku#j0rd4LHfeyACPS@inW$>HQ2}L)6)-ans0LSRYvg&&
zs65qFuRLo!Qa|;$*m$I3>aoOlq((fl9Sd6O!BkD^Lc_2X?iA@4@r)HdY1e*cwQ#`5
zEv9_HvZV7D3RVs4e?Y|F6`}QGzm*?*@<chV#01*wjL?Cx>r@VYY?j20p*>?a$dAuO
zPIi|yF~8-EjtHN*kDLf?9=kHMY}_i5I#*j#=dt#-99wY**E0Ebl^#3Ue64LRk0<z+
z7n2_uv2e^8&Ac4u<$dyU#1gTOn=ij>yj-m$*wW*I!x6ZWW!m+w;0a&M&p3W=+Dm#w
z?Il>mjoR)sYddIRX!Y1#!yL<dbqc-G|2PjpI&uxm#$P`SR^5Nes@WeEHjGQXbzor`
z-MN13{@@PR%X`R%8Zf>gw5;HW)H9|EsCv##bs1GFKHnAiFR5zwR&ROaqA81AX8l#g
z=k72nZRK&4kE8n319R8dUDChbNFBC29u~ht>hy90CD+jTmDrU(WHdg)v|r_TxG%|~
zVz`>wElU52zt6(mfGy6eSnnwBM-U>jBYG)zA6xJK_5QeN!GCOjoWpfkKhz(xe=9`!
z57Wo6(*MN%kUb4ZECZ&*t`96?MZ{zg;f`z_8uYXK=nwbTj(>JOomui9*k5WoZ6q=}
z<P$;1vRU(I1;7}+j1Ybjtm=PShJv6Z45i!r@9BU8ExG30b4-3&+}jm+N*GFR_7?Ag
zNZ*#sbf>{gV_6*KKNW_utpCB*3`ODye-A@(Du&X%lf_NW0zi9Z<2lWJ&zC>X4>-?v
z)#jc*m2tPZ`7u#`C@N~aWUUAQf0bv}AzxU&$5&%{xoq)rh^jfErgD4YFuUiXiB)?N
zeO2qiCEHI0cieY$wXeEtM|%?bOhw<2p!41dEKZBcmAJXAU=_OAD#!9wWla~%J-e)F
zd+~XwLe*(x>^SGoa=}!NZRrJ+O48hKmex^HEdAjUE*(c!PF<MXA+pfg52J!zO6F3*
z1LG@2IezasLXC6q7eRN5%a?opG|mWfNng+C`hl|;equ<q$64`kfwS?e;~dToA$eo!
zNOd~!eu!ItmZ<h_#1NF5HnCDq^ld8J-kvAVg;RZ-*X=3UaVl7N&(Q`~<*r0m$*xS7
zy~b5iQ|EZ+q*UetD&ul2uL-(R*~WUx*sE9Tx2y_P(#g|kWU#b2CHYnP9K)VGif<(w
z@hd}n$5+ZjkA~eePd4J&kXL}O1t1&oY{&yB=<x*req{4x?8q1N*r-5k#siOT!yEhZ
zp@wOdp{5f_?Q~&~53nmQ5?^>HUgyw<!B2|uV#}(0BVjoyT!M(Ih271A2fIBou%*7y
zTv^tbM!Y_kBe!z>4itxWTIne^qHtZfkAUEgw-*h%fv1wD#Nuw+q(+y$`_&~?Cl(se
zmM=?Qsfy%%sYPCaMt&owW!ts(8Hm$2vByO9%%SZuZhdOFO;$zhVrjWz85@NVgZv)4
zoZl_1E?(r$dGp0mbI&(FAC9S|-!vJ(@qcV;F5G_7K0PVO8(3Ay!6&(|_viLQUrlrl
z=f_lxG(^mvp6m?nysyv7c7+G#+Kbm#>Fv~Gd%WNKF|H4c?d2p>-zPzZ`Dx^5Y`f)s
z8QghyyHArt;VnIW1-$TJk-cw^-}0*weYd+4)ni_AKlvQf&n{o%`lZO%<bJ8a1#`2e
zftN3DzjXO3?w1*CH#aM{-{7DtYyYa(5?o8@FRpoz^RWp`nOqz9aMcoqZ3G7pD(gi3
z>c3V?*c0kMc5Sfg!ddImg07#e-&r_wuKhth=<2;?SK<2E_KQ9)ys>viWVhH{z77t<
z1uE$Jf7p8$_$aGuZ}<rW42Vonu+d_DpovbD37G&&xMo5k6PXDy2?`1hGsz@ONoIz*
z0P%o?0<}3Y%4uzT-uHM~+G8*Fv#083-`3V@0y6@F$W2A#1@IQc3wS|-z`Xyp_I_rb
znPdPxp7;BG@2>+p&wlpW*R|K$d+oK?-p{xNZ<FHXnux`|_bnOIn-UX|_EV%=Rg^e5
zQ6QIQDUYXXNJ*DeLPH7kJ$t(&mb~saTzl{2sCoHE-=xYQblie{T*PA1ez?q&vd%&`
z)pGZCe2EgISk^In0PdQD9%K-)+2Pe@5Bg!Xd6|*^wiJRE%$CX4I+)GZ=4Mzk_AuUv
zz~lSVhNo|dt=oLiwG61*``?Wx!_y<;Wkm=>0#)1(-!PV<5bOy_T?5vg0)9pTKNYcP
zMKP|*?4@f``)G#bdgI8qKT49%ll4PqKkTa?@}}1h5qXk+s8Hiu+h7X7cfk4NL!s0U
z6<RxPVk7R!V4|EsgCP3znwr*G=*XNwFF<&Sk_9wYOd{kRErb*>9Oq>%B%4E!&rep!
zd^TyR+O;?vwN$^>Li}2YPYbC+2p!;o)ijwJ9(fFZ#P=(fLGObRi}XvEyoSoM>9SA{
z;7bk<a&($-7DvynDH|w^poPi`+9T+jq^gRcSZ2ZiTWF;;=tgAzK0~W?8YvCBI}Y{7
zRZSCf*`^GiKc&^DZ7K#Z!}>R2g!zn$_owM}nY573Mrtz2g<fSvkMC77nt^7c6%oi9
z6OYJSEn@&Cy$fzwNTeq{6}6fYkL;NwQer(TOp~AtDv9<u_E2hD?#S%uF$Wep`I&5U
zid4z@RLOjDdOm!GYk-nO${b1w^J1B_wNFyNOgg&faV_a$)SR^IB8Sns9Fj}jD3hu^
zD@0PEaG-X_Q>$rHKPpq__7rQWcMGhM?VKnZSdG^{x$NCCwXx^R0}|(*A~jjSiqyt<
zYUn|elOL1G$Mrmg<hHhnxpO-v;#k(ZnZz-++;NzQhHcA0=`h-K9dF9od`VGIQHDc0
z#AUK5$aOr_2HU@r##2Bo8FrpFj&dGC$%mUShk?BDf!tQa!dx|`HrEe(Q+k%D;<*@>
zrp-Cf@@7OoTd+5JI{P8!L)e+_$IBzE1C4%A`yz}4pU4&5*F5zl7Do2o+)*|G_pYFS
zQOJY22e}Y3t0ml_9P)O;i=vz`evQ1fI;RYpI~3FjkthNQ2*xX`r#{QMbw!UaJ5Gw(
zth`I=-c*ziAl|%Wn~;fA#a$WE<ALMg5zMb;xkEp2J$`qVxopBf5^$e*0|UWkAwRDA
z8AP_WXhUgR=7i{}%YkW#zJq>*7(GsaNbS710rmAGM%jb`f6GDTFbW*SU8t?_Vm<M!
zb=?-6^*W0?4nF=8qw2b|16Sk|yH6(Dw8GQ0r&wrdeS2m}>pL@#&S7yXG2d`RL}}uP
znT2g#kb5m1x8k@xq%iD!<z8;-{9pjmL)P30fwZLa!vV;tup}#r?I0FAfA>=oJO5V%
z%Z6<bzq_Df7_>4Vw0fkUfOrGG)cCrm1<`4Uyf>FxPy%aFB;1XJBwXfTb-8MO?JqV#
z9NmNNd{FCKGAM?1GdKf%yB~e~NUj(on~OzH)A$JTnJ#*|cJ$REBN!tiBeFq8GQf*p
z)JIZ9q=V;-G=?H&{sI=OT_A!jAc6>LT1=EMT~I>V=&?rvbc{0f^T8O^_y(N#{GKE~
z8nt<-5UOsS{D}Pqo_Vfb#xn+nzlydqN54hu6=?7))<$VmttfuYu>OlIMVZqv?rj{_
zgWc_)r;YFV_D}S?b>j#04CR3Arrco_9cAM)A#fwVC$Sq7;}4WQ-3cG0?IXF!s$nC(
zscRcF2!G24VgEUbo;Fz8c7T#b=U6(Af)3sS9UQfQ5G<WX@Or4QJgdvzQ8rJ3rSRTJ
z*Y~b_6`p<Mf#KnTZoFh$S`Qlp^!DoxW8L=fTd@5eaSej1G7_+xb+Cp^0wQ}YVG=Nl
zfCJM2hY=8KqkdCn6L3%(;79@vP6Ny#;6-VGqX~#F5%*iBO28p$fVl)5ng(bj;KgZx
z;|Mq`4X}`am!ttsB;fEgz{vzOqybJNpsqXb(sx-20Y@ZKCIQO`cxf7-m4KI}0on*S
zG7YebfTPj?YYBLH8sK~aek%=dApvvJ02dSRiZsA_0$!O0xP*Y;P6KQr;OI0!KLM{w
z0}K)HJ86I`33zoH;3@#2bHD4lSD>hV4lh|?YGOSOJOuvTN9=l-*!4FwSOa9u(8qa&
z;I9(==j$bWKpOZDC468SI4l)7e;Qcgb<pKcV;+@1C=Hy>@eq7)8aN5f1ivT^oDOmj
zJP8N7@;6ENkTmHZlJKEv;J=gbi_^d#k?>(@;3PIt{VquZe?r2Cr-A=P!VPKQPf2)o
z8hEFKk4OWLNcg2`;M*kpvNZ6W5<W5woFp`A|EM(Z{Spp&rC<6vAmQIi1Ak4zG0*jz
z{-A_kkp}*@gkPBk-Xr1PP6Piy!bhipACvH_(!j9+&-nRH8u&>8FS(i~Zy*wa)?L>f
zNboPL5tRzq1wgc?WjgK0)3$tUEB4*jp6vwGu<mXs_Hh_y;EEjB(XSem2|1y5P*zFn
zdowY@%sg7m@<T0$+XiKox9xzuVCZOxt$Xh*4522Mbbbuk@x8*btT(BWmd^KSiiZU8
z9HhV)%%onxIJCW@V`T;`00tWF@u4^rv5*B4+aOG$Z=s+=Gmo?b0F{CMG9yi2a526Z
zr_-k->kB@@ci58cNnk_Jk=ZF#24#xg-fHa!ee?wnUvvTbf*1G7bv|({iL`@$K_N-{
zf<I{?g)Frq&TqAlYz~3Gz}+bG*`(!jhZbj}ma0Vy@oOP7wU8====BBk6%G1ctI!9K
z-c{%XPLF_20P^&DwkM#&)&J@t9l~nkj2+l7U-Pi<-htVuK@?YEzOj#%Zulgo48n=)
zr5v<6|8xX-h$T4|J@K>;h_oLb9kp3YD}z2?M$02u!tDG2x&_<7Lasv<$glshf2E}t
zWkVC8mFed+{grv2mL^S^Rb1xAv}Jm=^nJ>FG;Nu;YiZJy>E|+cr!8}emR{nRuzE}@
zZ6Ec>?|mX`i2~sE8j<F@HT{-(@4$`dzkZZ6M4(bR^FpSk6rl}~{**I&w_TWWhSDc1
zXZ}^B{EEsMEuSj1os>K|JylvhCFfDjXi5Dt>FAzOTGI0=XSAdnWm2{0)vdDK2hp(e
zC}*_PkIK}!Jr8QBcZ<~LQqE|pcgxhqo;$VFowDq6DQC3g$7J$xJy#<+DQE1S6V%md
zbu&M8o=-OueSsI(|Nqp@Y<>R1bu&4LtLbJAZs<=pa{_`&PZXe3-Hhew93UB+f7sBA
zulITX?^zYQe$daXfn5cB1I7N|WN26Fb527&gQbR4&lIyh2I?3J&&*+MOh?UEQO10m
zDxIQ?Il5RVV|;y;F(0fyw=!lelrew)uP9?4$I<25iC1H{5gDaZ#)wF&h?GtlBO;BV
zNSS|5ql}?AWfR99*^o{dqtm~v{ose_FEXiyvR`L3>c*x3Ut|S6N!(%-`m`)t$CArn
zMooJH8yq}0LtTdne*W~$^m-Qid5e!4llLpRKgk$&ZIoEhIu)53)(_x)2dL<5o?{%0
zdquKHH;L<1G166b)K7pl^-WduW-NtXX9B6mOK@2k3?gxJ0)kde7|=7G0_eDVE)1~;
z9r_+s0$YZQvFkVKEu2V}+D_76L+MlXSjOW6V25A^<-#D&-gXctZ;sGcWpwmP^2%QN
zgX8Adm3!6R@z++=8?xfvYV?(5`!l1jbiWI@`R(Yby9->O+?sXGj_B)0Kc>0J_d)dZ
z9XJya{vB3~t455aM2oZHsgS^%#YxYZ%mtks(q+ffX{C9y&z-J>)8+I{mlIF-CRg({
zRI{6^iG}N`5w~M6uM-TqbwKB#0iA~ju;R1hrV`qIw!&OP$i6uAWmrew4Ys%Kq8rAj
zUva5j2)EmHj4LE@)dm_WS#dRew+O5|1I%4JBQsYB#$|E)0RjUwlHdzxWR|zR2HrRP
z1zTXTjFamQLQKva+}SfA7c%leDBN^1@~d1L&a~cO%fN+#XDin2${=O0tqt{Dfy<XC
z=0Z)EdjpLn$UJl-^jz=GJZej&N2=aJ0NY-HC~dGjMMal))KANx)YoXr-f}2=k7goh
z;tiJ0k07R#t-%|%fuK_?9bj>`4TH>PJa<PQUx3px3-;Q@evNJFs@y`0;YT<~j_)TL
zezc{c_+!J5I_<@~OAJ4XTGqc}*gz|h61B&$0oH{qAnOS373xGa`egUp)Z84<Ou_j{
z=2}T<NEIiRtazvQN1Kfrk?E71o4y(~`qC#kH)%N*|Ai*yBQTM|1xKy;i%Yd_ZD(}+
zhK1NLTyA*C0A94Lw+%XDG5ln+rR#8J`H<(x9thTxQo%C%9>ldN8M?@4(?(u+4=wR*
z3bv!;Sz}sAll8)TNTq6122yNuH`N0p;QNN3Z0UL{vuwx?bTjxx)??v4P$W0qLx$R!
zu%uB98zzH-DzIF%#<p>%A+McO_wDfPBF~-#jof?BdSJme6`Qc24O{Upq6&NQt4t!G
zsUK}AS${M<q)a_vSntPnhPgADevu{c)rA#0)w7oIwZ(#JmJfmX<;}W!Udc_$(8VQm
zCF0(ti+j6<twhFztu&D}WF;cwR{3<1hcRa2xS9rB+1p+|@<X%@y9G1@C7*~$*`GPG
z6d#7y(M$4rwi3YP^{l6p=M!;6B6U40?)w5;YWof!UPe=$xa<q_j-2iAaS}Qo@q<oO
zj%OfqCp}c2`TlJ_&wSc_V3-Trw2+O65B|#;^v5lTgLP%DpI;x+UTLz-8T6ufh4DS6
z9)v6y^vnXDKE;h<M}AH-l4d`}izO9HS9tEtEfJmvau*BFgSoBbd41U@nSejR_lUSY
zecqUN@Ni`9-D`5M0q9e#G7)l!L+Y<lqi2@wrmb?*W49ye8Q*jd?u&4J8a3uco1$k<
zZjFAvY|nt`YTuya2*>l1HiGZ_lQC~wQ=}_=ay!CC&5KYN7O&5JfY+-U_!Xx<cAP}^
zpHlL?0EfkobF9Cn73ksSY!a1+)w%G9+f4v^P<(-unxkt){WT?ho3WUiky>qiHD&c&
z`x4c1?bB6f6RH!x*36cxT4P<rz45alYax0-mjDXbX2levTBq7Dp0EGw2wEQUF%cEj
zr(+`YB~E!vgpJLjy&rcTT`cT>gnbv;i_!s-l}0kSr>l8JLXdgjse|b1Xiu%U%*l1U
zAZznBob|iQvDvlkh9TL{r9B$*EG!6M+lQV1=*oO#nS3;4WIsnkhPoabhr<x!qyWmn
zNdfJ6h&b<}J0bEOES};gL{wB!cS2;RfQb_#g#_#Wgh+`7>T^QGuR&hLokA~)>~h5J
zs0N`EBCiUFI3co;Ac+$qUkaGYsTOOnbDa<or$G2Tip@YLMDVd&PRGY$==4bH2@%6N
zPKXTGoe=rLB~F%TCqypMoe=p*!n6}2!*nM^UT0XnJRwq#6Dhx3sMH>I{qoNE36Y7l
zg!(4sw!L&tgs6=y7=MqgOFk#EC&Tp^&U_rm;1eS6;(-$)&*LlwPIX+$=R|ghmu%M_
z920>#0zro3z(dyNUE+mrKw-Mufs-QzZP_?Rk~lds6gC16@nI;OGZ~t_pO1~mBOFW?
z$GsP4@8$y|N01+ma4<m}_YOlS9T<66gc2=8>G;W|;=~gl8+lE@h#DA{t-&gl&3gr`
zRfD0<bW(~BmQXWL5>W)wt;!t+m?3*N2QJnEhw1{e5s25}GL9l%FOjduz~O+R#+<`J
z@rq{g>u&KnRK9kKSE>q~TiKaGHl$4kTDN6SG;j)<eUKlbkfp;Le@87(-MJU54!*hK
zuI%YQM$plPiShl^f|HOQajX*efJ6=#KMX5^1uzdNe)vmyp2X6+MU`htUg20c#qbtI
zIEa`KONt+UGjXP*1I24+O5VqrlJEC9Q*yN;&Xmv-HqwXWnUaHemUrB)YUfH8;#|r5
z3wn`3jLvI(nK)M>Gx(dxV6!||Lirdv3NX7{{ziEn`7XR<-}>sE`=Yngx$5je)>{}G
zr{tMJy7eexQE%%$IxnFwu<$qEaF5IG=<gSFMl6?J_f}-{6%TKTSgwP0W9bzS?}-K=
zZ(oYLc8V6f5^-Je@BtuY!A=4_m%syk<VeHAX2QdKZ0WV!y(3*SzLa?2x(ggwI{r<C
z-8fFVGyxZ-IM8FcZvM*=SUc{I-z;-lBI7A~nsNB%w9Mxz!hvNiW_sM5)bU8Jz*2-V
zgsJ<uq2=axEJI-XcL}frIc@Kep^Pqc$%)Olc<01GP!R^!g6D|id+-+N8oOW%t}{a0
zPZn%5tlLjvkS4Z9ccOXw$vg+M3ck7{?3VehN^i70z4C9Iq8OJ-g^OPxv)n<mC}?wd
z`ZgR;o-VGY6MwgBM$3Po7&y^L<-?5RSGb!A9l2&MI`XbN-;Y>~w-Ie~N5bU#=$wtR
z|4uD92&#@mEY;V28reJqRQ=DeV1?%}x^GXk8V7c+TW~1i8uIW#>e-jbw?EN+#6?>7
z{pg$Mz8@_(1R8joXyEN{uJ;UD?+t@T^xVV&vgh_v&(T-UBVAYGh}7?K`Uo9|>3Rp;
zYld}~5oq)3oahC1pw@5a?5?v<#=GmvJNM!=HFwu;yVhNNmN&t7Hk_{#{Fd5ZTeSW<
z#Oy}>#q2gju-h)`ubuxl^j94;Q3+0)hd1UlEZDUE+K2vnJJ|S*_f_>@;(bM|B?$wT
zNHkgK1jVCQU`Z6$r{XF*Y(PSVrk%H%)*9Ws;A1*JgO%|X*qGyv!Yv4jBF8SQSpKk+
z1gC=Mmg7#s+ML@a?U(~%`0E1PAyEAAr-+Bkx9-F+Kc9#BQ;AbFG@`8@BWH+J*XobR
ztOT_my<W(-mcNR5KujdBSpLcbU;N08Vf`FL=Gzdz)D8oF#1(qq@Pk5tP{W6NQH`xQ
zpNHEra6a#o1<!+1a7K^L<5^laWAYoBfl9t;iFIL$eGl&SBcJo>i8Ac)ZP;#D|LA9I
z!Ue?fVUlkYU;lRV|9N~xzmIcJ^5|q7TSe!_7M(XrhpjIdTdf#W1-@E8{o447SO)Y@
zUe5~hYO44O=o+h$mr+k%EocedhRF3zrmqA2)7O7DzN&DtMlfc2`XciInzB;(@qGB2
zX8r#HUqAZh>5CchTIgedC|s<n=`Z5r)i8jP@G}zQ)n?58m*W1reugVaPorm_13$f%
zwTHw7J-s|!fu4Fv09$1&@4)9qI%DO<`wZ)o=?MqkNbu!&s}ERk>Wu{7L~I(bB3+mE
zgCO&{8SqkOz%C3|!+5xgk5#?kJ`Bb@j~p3|EinD~MB-E@jxu1?kLz&wt_BLa>LxV%
z6zpxW?Xlo{bi~f)sn1nOAAt*P*Iz>m6B7T*piZF6a}`Jv@W6temO$qa1)*hmKvO))
z^K0Um8vpI2+jX(HUbFt1PLx0w;~24__npN217mC4mm9UN{gM_^IDd^3DAEl2mmb=A
z!$)u<7Ui}WcrtDby|f%l*8%?o`4tE6enEe0f1y(K;|K4~Rb@n+KWZaHLU{B??ly4H
zxsKd@j2Jl9wSF|MosZD{q)<8by$4s{A4&UT2Y>_GA1kE&u`Eq#cos%5eCH5$K%~Yn
z3l}>dUTvIz$(cVwoFLz*FErm($2DKxQTXT{l0w&^v&J>@Bmj9I;BRPh=!3a_l*(SL
zz2Qaebn`zcBCS@>0ip!D>Ihp6e}3xf-FPEX`OXqLKX=yX6G_>d$PhihQRWR@;-l=P
zvvvniEo`7oe^fjt$#+wyc+dOcTr{|2{nedFLA|yO0t&uP@ECfPuZ5xQw8?V3j#%!x
zJ9Aj{`M_RWId#|FPYr9OL<qDEQ=^BLeUTacm+o$CQGGE09?0YQP)52Qi)MBAxc2=a
zEAM06dz-&(PiC}h*~=6o7TL18_|puWf}7Os+V}UYyw^@*<ud8nh_m<Z&5>EXmv?1E
zX00yn%z)-EEAQ=-5sM7(%!t(WUcL}#7dYS8$!N<<3v+WQ@_`KOKV`ejcH`OY^!{81
z<rbwf_R=2&_Fn!1r-uqKgR<O3@t~T6H$u>8MzqDdnUd|J*L_@xJsFYj7OR<AfW2hP
zrE4=M7IG<^>#GdR9zcm+r9`iCqHd8$8Ay2_pm5x>(c*oBYL7a6cZ@o0G9G4in94hI
zIgvV)QJBd=#(`9?y^Nl{q`XuGbP#3sGG(@(3qg!J<dBmYukjw>U}84MHsYcYpyE@)
z>r0g8MNwW3=YsZe38>DC_)<`9u8|XEUqo5$MNMCdd^cYmnGIMd7a5|M!oeJ*4B$s0
zKB|S26bzYsQMB!JH{9EBAHvZ&h%wzj`WUQVu_p-+-EcMLZB9BlGlElx1QGA*AYz`$
zkvbHZGnio8IbPO9ln}avRd>)%+L#Zc(-w5mA(Rstc%5`2a;W$26Op%iFaIJVI&r~v
z3F+#+`&eX4@8!p7<8Ne6^pOR}vAFl{<t)+dR7&=M=u->2QGpNXH52$2*pEJe)hu}@
z58`fB2GdUJ==-pidp+_W#p*azi7G;&58X#BfIQy8DH__J8}l(|2+)J~(I6Uq4T<=z
z6>qRT%yK=}Y&^T)x~><SOaBk{C$ZH;8^<}FPjXf`<~FI6cGMvM5|gbO)){><u!k8g
zG)yIKD;st&dSuxN;<mkDyA#Aq-O<OEy$oi1KCAm6v&(y6HZyGoY+v>gML0ps_P(Gd
z*S=_0-u{z>zcYltQ-rGC48mOuEG0M*DCF>y%q+N~NU-4-#BA7QiPWtwepxcxJmc9e
zA|>!EayujV49wO^6t@-m_Fle9GFz0=AqcsjrBY94M0>rbh~9Qk<Q+`XV8;KjsMDC)
zU{XNLW^|cR%5#+U9tH|#+e?YS0Vo`QDUo2d-GbJcgzLBtAmmJ9HexWzY;PetF&lzZ
zX148&p6!C!7|Lvd0fPAzg&+n|hF~_X@pi#%6q}e0l{`w!_B^G54v{h!%!b$^eedPZ
zOJ<`)g4v!!O?NTd$C-^P3T6Y(5wqbT36%M4ua4K~#yBtt{=`6gG#-uVq_=KPRhY>v
z3J5_(7zmq(3}&-VjwaY_yWl${ej5Tuj0oaGX0s2eV?Uyf{5S&)*eN0C#UAR#Ba+RY
zSb&qxE#BuiOEp7sH0lmsi_md4F#B6(U}|7QAfGspy{OWEQ13CXjl2&^ul*E|D|iia
z2{Dx9HJCV%ZglmScR53V9uU0tHWKk0@tW(gu<`6p;yCNt-c8_A!-hw|Ze#_V2~LEG
zgOyEJBD@a-)0HgTCx4Qq2oeoaHZGYTkF?{8OWk~(ykF@BFm=C@^N#OVYC?t|C!)IE
zhg`v%j@L}XFoByFe)?N61CO#x&sA1<NVSZ%&NuxNy`0LP^yXANJ8z2}xi%|Reu!8C
zxaOb~lQ@RLZ^1cK{Anz~AK&8WxcPgdx$|F*j-G!wI&%K|(Q&o=A}&+$i&oFJT*6lX
z=0y(-2X2&Tt^~=jVH7qxIcy|_jZ6+3O<@xk{H#?xf3ZnCf7>aZe>_Ofbk&e$<x-rK
z^2bqFO8FBhETw!PlFR3TcjAIysAv}I^V<^f{NrMJo~L{ayM4;H^;15!Z~BzKP?p~+
z%KOD8S$?N1|KJ6bU*AvpP5qP~>ZklwvV2}JQ2S}&K<yXLKQ5-{dD_3epYreRr~C){
zDgQxPeyb?&7n@}HowEFc7f}8~{gnSmKjlBsPx(*D^7#WB)PCsEx&7k#$HnwKPx~YN
zl)tl|^7r>s{%f-QR#DzBHp%ik#q<7yspU_^S3xv3l=T00KjnYWPx+s5S+zSNu1Uqu
zbJ}}{pGFuaAJsNIlO`L;z83QtTrJw--}6zzJ5PAK@&kLag}1I{)V;&;Nd_(T$giGy
za3Xc?S5KXlNImhZryiO}UG~*e4@;!3{pzWQCsNmc_0+U|DSIsR)l<{=NI3Pq!-;FZ
zdIlpC84x>v^$adgWI%lT)icORWIzo4)iby<kpXe;SI=N{A_HRSub#no5*ZNhe)SC0
zL<YpvUp<4doPp(L9J;@2MCNd-{G1;yjVj(P_7a}Nz=MT4c8DUD$+vZC`a3dKx&Jf#
zU3mY=sIm4S+Zyc5aNROvNH%T-z7k)7Qfqf}=yM{JwrUYd=}sfX0fy~}rx?qyk@xox
zf-IH^S&TL}!t0|e3w9bd+>EnMhJP-j6Hh`4iJl(u)bN9-EqgjO&nC^&Dm<%(FGhq_
z!%O(Dioer_Z(ToJAxz+xjDihkNJ)po>YDsDE|cXieDC&R^Y$a(1#-lG2jnaq(f`%z
zF}QV=9<_#H^jK&ZNsoHNXvmw};Yg6Bg(C^ccqhS@7Lp6_K4LiBtyG~qscsF6MRvI&
zyKy4Bi6Xmc7m%Hlc}M(o6Pkq#f8Hsczd4Af$h%DBZ4-Idio6$INZy1l<oQ#gK+3yB
zJb$woPmy=M$h%489TIu3x{$nEW!{@)-kmb<gEH@X4eMnK9x#x6L0La&;9&^9hYT!B
zoFCoyDI`-+8^rVHCF1#;#duPl_Zl9GmqT)dD2E3dE{A1?3n-^mma|Eg(<#e27%wLh
zFNfp>Q4S9&Tn@_y7f_BuG6J<nJbzvyp1)a)r)baH@p4Em5asZ|!R3$)aGw5&ApiSX
zWjUK<Ii2G9gM)b5+MXqYxk0onuomwztp5_3+C1Cooz(o9hRTiy3?wkvI(|T_DO>SN
zhILP4bye9>Ysj=0!<_t4ykWN>NxzAWOq<dh>8ietjH=;A4Cf$R`JV`?gkd(KeVG)N
z#bNk#l`h^e4m*<+md#;jlfp)F81(+Sv^fYH2K|aIXf!Ul!c`-kSruzfDrlPc-qMjz
z<=8>@QyiI!y&~QEkG}^kKM%ebS1pkN-r5)^zBh&}Dqv3E_Hx(9S-2^@+U9wW@JF|%
z(IE@90R6<5*C+Jv=ogsTll@y}0mHJP0A6#Jl)q!DwVN-n^#Gj+A9_$Z5^0HN_9z?y
zU)U>z*2p*aKcOLlrnT31)t|Bb5k20$1Gl2#!29Y#*vj1Pb36^J%Y6f)FD~1cNr8K#
z&oA3In2bM9{u`qIp}RZseRt-<f=qa^WDie<cIf+yGV2R+5rEJ4%*$*lD2u#xM%n%m
z0!~D}Z_f-B)RVR70i=(VmMdGxO!TOH*%|%&=iRPPCK;<=L$K*H3f^3<yn~lX7lG61
z<!r=qw{P|vu1^*jtD!Pmv}`xNfiMpmtO)L?e=ag$Qs%t{>p_REPs)weZ$!H6%XSau
z+M!+#6+A?&u(tX02MUNRwMsrx5FvJuHGHDrZHl!ytbFO(i%ec2?`h8OHS)d#zS%3w
z$LlWFCt;L?H8(yskDY6%eZ>TExcPT<hY}_Wnhb!{@b7lhUGO*tLp^LS{=l%o46=ce
z<O;fMdH&|eTP=ggwrxH>iF-MAPUtyeAP)I1j;k4He$vxZ3DFboB-iL^IQUXePfZDW
zihg<Npf2h*BByK}eS)4~+bZcv`^dpeBB#A`v7OUC`?HGnxhJ98@otB?HxzNZqrLuD
z=Fz=+UHB>(P?2@_b9C+AcIa!IW6_pg|K4nX>j@Q}*{|VG`F?-~>>b8b^#EPH_fsUL
zx5g*vuWvu@{4@G4>HJqx$-HBH^SHQ{Zw+Fdbbb1cF|XSIt7MoJN0;p%5Otr~68#gM
z#5r<$R5PMP|FCR7u0+4v2U?^DQQ%LOK_?G?2qcKds)BmgW1kzlNx7`<MMq}taGC#Z
z?Cy$`PS1=)Hrq2#xXeE>b{}!Q)oh%+89KgEnIChdMqLcva6R_AvHNgj^P<eRuvcO1
zJ_TaTJRN<#cf0GcKS{{DczxIP*dL7DpHO~ZM7qj*p<lK4e$M$^^%ga*XtfHe|NH0L
zIV|sEEiB^tWV*5X1r9a!Qt0D|c_#XIV!%HwdvO5TxNl(e*UMhajQ%_LkqL4q`18fV
zu=Dmk#Z*rXeMC?>xNh%9(RY^Z8$^LmFo~1zDJE|&X42zS%&$<)vl11q>6u$yW;9?=
zq|}u8qRag6#_mHn1*KN+ifq0L7(&-{gCX3RZ@M0P(b)Zd<nX-AkHLVwJ&Xp%3dY*<
z-cv5~uZ{R<1d#h|^zXe$sp0QES5dQ3!M9M++~-}7!~X&B2ELz1&8PQwn7ekdA6FSQ
z8>@F<BEZKtn6bpPl?_!hUjKp%k}w*o*gOOcgT1_Z_8JR2Sz<52RB73(B#FH?>iNqL
z{?d4BA?ENnZ%MvllTM81+Ia0ZN(NJP<Mqk>dR(?e^{h%6wVBJBZ08)b*9_>nnPbqX
z4Q|Y|w9n15wBI%iAEEgubM3nZ{MugV`gp+F_X_b_duS$pL%vvhm;ujreKcV0y9CIi
zaB%OoAukcL?t<m&T|O}oGV>0==KwPvD4^9Iy}vLJ=*2Vu<u26i7~2m)9S=s$jvo4)
z)W|K{L0q6kjNePIL!<v)8}KnxEK(+|z*p>|pMU;5!sR59OW~$92q(|<pOVLdvAmE*
z?wvGp+X~8}C!w0|nl8ulNMykD%(8;|h-;BDY6B9uKCy!sDE(EO9u-<q1)g{bJkfo^
zwHG9`oo1Ow1UGc$b$>>$PcTdHXui$$i3xPabqUJ4<oz>wkk=Z&>l3%JdKX&uGUfP%
zNX)YX+QB5ci>AB>1ryOE1RF6UbuWGBVHhl!7Sf>Cf(621&tDGmo;Dx(z?E72<(9wf
z!GQzXg4NUMtDw`PS$sTz^thM?A{o&G^GSbx#U5NChf8P?CVZG0_C!B{M(EV#urG$G
zXJVT0C(&`-{z6aBG+$`+yd{O6X*ZbbfcXUU98WLi2TVoS0Eg6&Js4gvBd%<nHlAjQ
zM=FSZi9f~ELU+01(ygB$cvd^E+nUvG8=p0+y>@&WiUqAQ&Gw<#Auq*gmZX)R(wr?x
zce5wsry8Yh=R03DdR>c2U_7p|TV0HO6_D)_+L>tCPGe+fd?BWfPjP<p^SvZs{J+<t
z!9VGKm*k6kegx{;hZ8XGo`jGCdP2m30G^1r<|%_7{;`CfI7w4LLne*y%R$lel<9u}
zj0RTQ_$mz1pAMi8^}Oz?`GYa<jg#O290OZKGM4+y4rGTbT(9^TR<ND{`}v^CyC);1
zd6|`EoNGOgiMuG*Nh>H<D*8U@8TD)OP(mtf;dnpY&-#w522&$4>B^%ify9;Vs5vG+
z0H1f&pgp@#<Et{CUXRXY?%5gHJSp=x50Quilsqw>Xc6-ZCK<3^bQG?9t-(2;H?YS(
zPr2k4DihT)ErEyeB4#O?R~S34eJG!hl9%~&ViV@Boh-xcARPVtZ4Lxq$|0Qj1Us3Y
zzvF`oBXE=w0Yq02lSp3Wr4KlX$M#o1dGN}yeeJU(Hi2!3PFZ@Y6Kn%=C9)-ENjuQL
zKfhDTQptlo$?JOlpz&TWPhFZMPp#d$8)YX%D%ARlL19o1DvHC!{5m*Sz81#yCnR6d
z6~ZL1C&^cVn{e`+<nUwB<BRuTPR{b>{U<&WhM6x*d5zyq@%MN5eNz1WBYvMAe}9=q
zuoaJ@`%h%LoG<`c&$c73e}S}e5@N&&Q0Cpf``O|Wo}X307REIK5>O`%pbY%2pX{3e
z^lfyF!1R59^TCZK)QLX^S&%q03DoQw0hwnz5n`wy({&kWy$cZ2GFW^dY*j(7Yb{lv
zpo|h;_B_swon!`t4CaKGH>&u4gl{&fnY;M!%dTI+f53IwuZ?+oK5zx_@&?IQXGwh6
z>MELK%-j|^Y$Exi+}PU-JgeQ&cRqiUJa5w<ES4bDxgMKt>^&7dzGS<rD4YfMp}#Nk
z-xIFK;eUYzi#G{Tjk|%U&oU);HZ1#W0Io7N_U>?v*Z_3ljI;_l)3x@D+TH2;@xN2c
zU6=iYA2i9}i0(XypRRv_40ME0@*+3<A+hxM@$1IC9>hPR=57Cg+AS9zK#f>xkWBa~
zVM@*;go%jjF#hgje7wj)4=!Iu$UZ{EXJntzdn|HzP$mYCd6^%&9$RGWJwzMHz3&r4
ze~JA(wUe0>m|LxW5{>p^k%HM0YnIjOyK=n(Ea6al*jNptaJ@jagC{CXT$(@yPJ}>}
zjk${UCiQbw!G+0GDRWWWcp2wG682S9WJA^VrM2ygo~?6wV^$~>+UI0jhitQU^{Ant
z*0y7~Nc)T9M+cMVi{o?*FMqeSZL@XAbJni+RothG5rLyv+xA#9j<?Us!RLwZdmW*C
zE&jTbnTB;V0#$770%gaZraxv<I9Tz4<k=N%rz&vF*xpflCB9OI^Tx1qyQE<wu6u%o
zo2~fOP~P13X;<2dPudMPZLKJNL7dZl!D9GX#FF(v_+@)rZ`;|UyVsqK8P+$UOl#d)
z@pY!m%J#C0)ryXhSvR-UUaa1nZ@btSHpjNL&B`f$cC`_gP+n1}mk+ZQ9P^8w98q%2
zS~n*LrXJbW;uluEXK(w#HdNl)zOL|+0gCnd%b=;X8fF6nKjwgSk*xTPzi5m2ChKWi
z#~t(gG&lJ^j&)lz(ZUj2@ymwwq`0uR?X<Vg*==wCr-Sz5ZD?#)Me%mg*h6S+)ROg~
z-B7;MQ1%6Iv0|VdAED}ljKLi%GJp@{aqOIt25_XQY2z3+EJ7c3P$|7fs;Rrs@w(2v
z|GTI#bzu8d<K|=lH&d-0lQ!ebleJ^i5DcB>&YnliASDjJG@ru<BHY$7h`>*-140q!
zhltY?H1*Q?q^+}gf5F<0j~%SYCL$YhOe4T;L|rF9fbe_)T0N!_;3q_Yp^F4KlYtp~
ztZj2uYy0dhjx^gypJ>Af$n#5W4Tya|W*1yr@JtE1v!}eUH&)TM)z(&MZJVEEYg>$O
zvaQVds}V4KVt(s5EWKsTY;DJELk@~6!EfGM@x^CHv&zV4`QkHM+d(d`%QobYwd)91
ziXseRXhp_L_O=zOH6K^=JB)N{RZ`8J7Kg5WHFTT8Gr)~PdMQ!ChsG02Zn18~wZK4N
z+e%c@FDj{uN{&S(pItW@FBx0c72`p_t?Q;~zFCaezf$swwP)$EV(4G!K~+0}rf3yE
zo~ZZ<QSlSjj_*@UYuhZfBL8Ju+dmntZ3phKaJ7@|A&-2!pjWSNGyJc6s7m;B<a(}9
zV!XlkBdrx}pJB+cbv&cOFygP-g9e~wF(CZb>q*y<nM(u6gF{ip?N6KM-+kujRdYI~
z8LVyJniazb)3`rL=~>)`kKJdYS?#NKX0_i|Rn}4Zjs?R_;L>%cuE5>*Kw?~ewzi!*
z`fZ)Q1@W#3e=ouRW$mR^<sDX}@68NeYF+n!Yw~#yj(??f?dBh%Io9?|_3;Cnk3MSN
zn-ucs(Vr*dr{XV<O`Yw^8TZWHk<!1N=&b1K8D{rvyG2Fq+jiUYx7pjelJu|kw&!SA
zoZCM7TlNjzp~8ydeW4<pyg<0k@W+=by52XHd-hWL^87va;;!(nBLg8O#oIf#BlBAo
zZP`HBw}x5Uf7VL1rcWwGz_}MtmyKVe{B<C`{>sPoBtK)UQTb(Gul#GN{rTr<Kg!oN
zn~L42HT!u?m>1CO1U;m*|3b=N2YO2>AF3UJcKiQ<@@ca!rTk4=`OlL%)di4XTtxUf
z?MGoL<^N7A-;VOF&%{~aFupou?fQ6_&9jwA*xL58EuZdp)Z@PkUsweFk2KhcZoZ@|
z#f#zTg~t->1iyd}a0Sj0@WFP$2iq!&Plh4A4Ek9aO-hq7Dg9(Da_<=RgL`=Lxf2G|
zPh!1^$%od=Sr~14KD`D5Y{x8>47zdQ#4Jn%xDN@Q#pE%XKeX@=#9!7S1M3p;y&h{U
zN+xq96?K`p*5as@E>1FZ^bEqPsJ;G!tcvz0a$8XleTz3_ZF`f7y8?&WC|vNuiO@oX
z_hiaIL8&J~vnlXnWJ56p`J4z9QQ)gGaFfjTDhiCsK!NBJA>1JKX3wJt#M9RHE)`4P
zGlWcay}7fk(~4^*t@Jq?TU%2Gnp~S<jBRae$-#Y$u`Rd}Y_MTHofocXf8aGydmx<7
zQ&<zEbNI^>$fjo)Hi1Rg?Y8FQVg??)bmObFZI`v{-OP&iX*XIo#6pGE;+I237_+g>
zg3mHryFM~`PEh*%;~2)ndyd?MaRCEhoc=00GH<lDWm{sl8)58sC+c8Xx9Y}W*y<s9
z1cNokY7EtZYcY<8#vXYSNJ$z$X?VX?mtQ<Za1;;jm<#&E|5jpnk4FzYd-MtO-o)#l
zkN)~Xkg+kP{h*PTL)V<E{WL%%+C0p#;d{UcwK)rIz7u|1`|r1*mG-HR<kHiyz5;>r
z`X^k%9yNS_G6$e{tpnGRY)XOaNK^xV5he7@Bv}HtYrufg+72n=>q}YBiA?HBa@w|9
zX}113>Q4*<_Q4fqATKC_J8YvAe2QjZS;t&9<DG1S-Fl80HVq}1_0CkZUuti^9oLOb
zyV8own$*zucmp#J!<en@V>0-pyL{F?i;KQHam#wwK(r^kbzN+Y;r{=Dp}wt`jDL}x
z4ejA?N2|L2wiOo`*3p?-<U@HNxwUv}xTo#pkzypK3#78`ZMSDxV$WNEru$A~fU;%0
zS~0ac8}~ucXw_lOun)XFha!=^Jw(0l<&+g4F?8$!w5;ujwe2};{$^y~aTg53H=ivz
zv2|d`Y%%<=-gWN{w06CRl}u-Xo)Itd%az$>gR}7%kvSN<mss6w4ZmkycWR*F`>UzM
z{4VCtff#l+V(C}18-J|Fpw=kq#2@QHtVv2>m~lFM_y~=sbTOD}ZyOqK3;JRGL;yOh
z7+ep5(N|GF*BXrQ8B}Y=%eLYd3>z+`q8QYHD>%?*YugBFU}5aJw%c;BAAH|2x)Z2v
zPuq5D+Y8qGt=5j)V%)e1XxvF^+!t@k?czlHq9>t}JUQZ~V->i33HYYA9kt~H_td<f
zQ;xdkHJq<~%{0mvgRbrcsh72F2BNVp{o*BC#<M3j4-6ICKz2Z}b>J%ceA#w<o6Nj7
ziJwuQ*~d_!i*TXZG5*x~p77BMEL?F@S)j-UE+7mI1ha_s0Z?rrP4N(5x*{9Ba~dAY
zF`&D%qPTO-K^`^RTXG0f*+(9y3l!ywa9zZjSO9{}wqOFfe1<jwjT{XU>=-p@El)ru
zJf1v&CSmpS&$*a=!YZwPXnk^PO_g=s)5O^dMsZ#r+_|OwrmA(FXV;xH8}9ihqUDZD
z9$14%UFKSA#<ti&YsWn#K;yiv^+|1RwfvPO=KjBziSc~`^0~SFfnBXAh$TWYNs74s
zb^`M)$)pEpRA!m<fW0m8eQR4=w-vWr+1e-O*fFO;9({@A(Jctafc~nr>!``oO-U{J
z-S*-w;TMkl9{DEd!`6|R19{Ze{(T<Ah*Uio$k(mP$;uGpJG?Z!*C4d4{X1n-uQY6!
z4Zp4Z&TQM%+rgE^;;j&Gg@z6Ey;Q(*@RnoPaFuwQjJL^#4Rjv>U@G2J!-fp;R)V(@
z!v<Pw17^gV(Xini@n*%F)v)1ZytUtjb;Mj=2Lu|{#jXfBk4z)fiyk==!DpbX?LT1b
zwYO4Rj*i9q58<)O_e+!E)`2(GTBQ;}b}J^7L@B}X`j^2xI}Jm@;a7CiQ6hR^4^?Vs
z<nX5_wqwDDdAL{1!+Wgxz0XkJSF~-H(pzQw=xka&RFm{v8<(D6PF_7g&k)`PqIFKy
zh}LlhTGKJA>TV`lBhEldsj271F(BH`t9oNMcMQY)F`{%GP7<O6=V16JKTTFM;kOlY
zKN~g_(MxjcnchD&eCL$qHe1K^y21<)e?{9%mU$IzpId4zZKvm1J52QmuCe8x!D=z9
zXZ*J@adn*;7`ocrEzh^v+CH(heN@)=o+UQ&O&o^Iuok}_exG}3k!7J}kzAw3<q0Zp
z3p$COlqbcc_kaBUfq{Qu;2#+H2L}Fufq!7&9~k%t2L6G8{~Iunp+L)_`(X_f;~)fM
zIioxh{p}4($!Li7;x}PJaHgWPvU`4~4bM#_&+pthU4y4S!$D`Gdy1MDRPA%i%59D+
zORZJCqpEtYy{dN6P{rzQYVxVgEvnbq>{dNNB`-Kdbq4~zz!Wv?b-J3|YRIQH`JDA?
zv#;LWG*oGJdc68r{(!I9A3`jO=517sdG&Z%>h=c5sm;!nYF>T*P-W;)BvGfR4Ngy!
zo8YQ@xf7YGceuUvs;@x}xmSiz9WEm{{>IYrjg8GsCVnk&p}??rsn@r{t1Nfdg;2LC
z0e8?HSniI$JM4~t+ljW+IXqr~LJ|)%uPe39yzJT=fuGiy-|c?o8HBe=e3mpc3@!n>
z*T}d}D+<(1u{CSpVv6FoAnM;0Wj&rJ;m(#KF7yRZ>A~Stb1WOrV|XH7?2+iTH~#&`
zA54Ou^6P|q2L5Lp{<cE+<?m5Wue~WH_m>EN_DqcmpakY0<xM}_p3A>W<r8~H{JTs(
zv8~3xk@7i8KB2DW-{taok$iqjKB11_Uygj@_%#2pqU0Yo_4s$Cd|oV{-<Ho|@(HyG
z|1OcwtK@UIe11nh31{@XT0E3v@|P`tWPeFNRl>#N*i7y9m-jCRraz9qYvJ0H`h|X}
zPZ4^qzpXQUm#&*a4vMLnE1q66bH&c<i>N;-Ov{t{hj35$ramJ45>Ba~sE??Bs86UL
zs1FG5globx;g;}8I3&Cgt_VMb6T&~?obXJzC43SN32%fet$#JT(E3&DPekFL)^3fC
zG&&JjP5<}gd43td=(+x-wkH)IsrX98M=HKj@eyxdhWHJTzk%|1_RN=GeEjA!YtlU_
zJ=KBmLpUV-5Dp1HghP#vHTu=)R-;#qPc;6}_(J0ct$&ae|FnKceb>TMpQY1#V+Y~>
zYWaHZlNr4+6ZQ+nmh{G!mG;ImXZ6P3fTi5maf8cxV`XqRl=sFq%<hf-5-!L3_1re3
zeGBd^+!5q41O~4E{59QFq?-rlhg)=WZ>$CG)Yo*IknS0{H{iNpM)f*e11!eBj=LD?
z^5Lv-Cpvp$k9YOPj>Fx*xi|LUmfl!5+>A(X>;<?LQRw2H?Tv-u+9=+B;r@qwf8r(7
z1#UCmn_ldV6>RN|4Su;d_A5NUzfI)-<L7!~pTS+Vqc>Iq=SP`O;rSE9e;RH$@+J2U
z!rsIC!kxf5+<4?OA8sw&@8R~rorZf8uxylB40k2sliT=wZ_JLmoy#3X92e5G!~F?)
z{P7uBgX|K|tp|EzAHuoue%UL?8?Fa_+iShC+ogLQ&x_y&$tU^G=P0h;`S4CoOGo)@
z{&Y`$dc9wZpZPlQ1gE|K4*n0|E`B2swn*pK;?Z-8&Uq2%$8g<mf}di&u}yHh;f}#Q
z)p08JF5I2>pNc&Qci>0Dy$b(#HlB(tgrl&#{^eAx70)N}JPkJk{%d|F+)wc&SFdvf
z+XgrFUr)te`Q@ot@W83qI=KG{_ZZwZxHsWWz+Ljnsn}I;Cb;QvcDO}w&2X#XI^g~t
z?g_XE+<v&Da06aF6}tki5RURE_np_1-E6?_gj)r7FWiH0kH9?zw-4?uxKH5*cAtt}
z26r8t1@0C&C)@*Yo8ivF4Sx-J!xg}l!YzWk3-0G|&%(VAcR9*e;ckR`9PawLXJRvM
zITJJ2o{3f7dM5Vx{4=q|3(mx@y8TS7VBwjV;ye@Ecqg9nr;Xp*ys61qnmnb&)1I0<
zrO8cWCBP`3wesne&yDi=sC<%zDE-Jhn||7Spp~!HM?eLf;;7PT;o7_Q<cug{iqpT!
zv=o%~rvbI6Rxj<T)i3p_#i#i{^+zyDugN4DoaR<dj_OZ7qIi^6%bR2}a^L#%p2;Mu
zWxX?v#Y;V0OG8ikX_Ak4gS?*dyITG<xD+$AES1@_?KX#{rl!2QmR3<y^6Kd^RH5}r
zL7n1Jr>PvJc0pCSW7gbxm9-3Vc*D(3EJZw_DQe|ByG3<2VSQ5HqEe8@+2py)T|ZPA
zA8ZMZuXiu^)VYJ?54oGid$G*oKY3v#uw0#?7OK}=qZ-5Bpr_I6uFpr<G&P9yz6PV;
z8CsIxZ!UHI@OU`nah@`?SXjo@anz;mfY;qLp6l;&HaWd@9&e*kh_#wYrA9k^uJ2(H
zHbo7!_}%JstnbEQ0oM?8hv?<5@HnoV=J#S<M}9vo0H>%TZ4g*c>wI2X^Z_(QVAJ6b
zc)TH*y;z8vOeU2Uh<FZFR&ar1ruV^jU1+6gMzg!w;R-i2xC1JpRL-;66r(YW4E>>i
zBb2WiQ)1=k=W)N;%k2(}tzveiI(@oYsD!X?^#x2byskS67g3PJG%}iuY|aHY-*BD?
z6N=6ifpOd5KX)z-Sc&Uu;qy}!xgB$>9P=tI)eC5yZCvhgugDj4=m|QSouPndrO_!v
z>YQGzmHlpK$mkrWy84E?#;N+a&EY1>LPj3L^+c|rfYTfF`-1NJq?~=>5IWL_z(n4N
zO}*`D4Ew^tc<kg*tsbB~ceC3YqIlFeE-n#>Xw*O~|Cc!Z;0DQTDT%!9!nA<vyl(S3
zfrwCa-Ux^^L2<711Pj#}pc`UQ@g|1@yJfzktioQyam-0^a95k?vZOdgNpXt$jWZ!B
z&V+vB6cYJURm`$_^0g@WRaD-rMHbNbtAG}1&<TulzdO(z4iV0&2NK=i51|n{G&&I)
z8>1_q0$rvVV%X4gam56Q?*y&l@<<x=oM1$^WW+k@&sRjKnc35<cpZLcFvOTqjHED9
zuTdv5>2R-%<$45sez(KrtXpc#R~vi*b=DGRXtvF+dVL|))9e={2gOv2J2Y1GeWETP
zIem#}0_tR5Bm;}M8X~Z0kw{A%rg6>Eaq2Q1i)q2GLT0k%x(3=6@g^Eprqt_T+M9q?
zDoYtA$|iCZ?-RH#M42L|1YEo);3Dq?T)ZdXqQT-Fgw$WJ-K6!JGvIavp*?gkOQ!IB
zuntp5z*v_UIU57eC4#Yeh;D2^LmV{a1!<;J$Y+`n4Ai01xrtg85&*LRh`OWu<0&LS
zOHt>?+@<heDkHRn=3h~Y3|!$0EOpd7Lrz^RXK-oo4$mU4OrVbHS*YZh%r^wpyvZv`
z`Pt~Hiw9r^U+xYBJwC6}<aRa$xiXl!MEs`kP=&Odc&Fqha}j0IkK^3a{28i{V4=*d
zQYv{ksVpy4D$C7qMQ{_8%2K?S;=L5_r4y7g7-lGy7JIo;2Sodqm}YRz&?M80<k_Nf
zZe@Rw1n%P<)HfH3Ajyy>S=qR+-a^8J*65*%rFynfS!=18Q=^pC)G9UAvy>Gs3^aO`
zF?iUkYO<Guo9A$^^!uRfO^(>$fy^Lga&TFoQA#O9D(oGq1YCksMJ5;l`esrGdI`CC
z7}I1GA;Tne*hEoWAu8VEMPGZMCMOb6AwId>fpKah2AI4;A{~7kNi{N-fRT<JBM@~L
zWU49ZwS=8($EkS}>eYHrGqmY>#dn2OS4+qZ@so0sDY#kZPoc3HNo*|4A>%3P9ctdB
zCP;9Jzl&tIV2nqMd=*k8A??P32@UFa`J~#cfVRCKz)%dxOmY&_kQ)0Ggd!%R4VX|G
z93fxG+2nxN;qyDf!8@qc7{=lS4OJTaO1+0WN}a9#6RNIy-76q`EJdtPAncw>LG)hl
z2_VQ;jer)~)>xr52HbA&rl)a9$kh~fD=U_GLT<z!o97>krUyeU*x#70Hh7w_D>0Ry
zb-pId4=6u4RpkK<{(!sA7XZA;>2fzs9}DCIH^KZ>3bs1#$Yb<1)h|+4snb@@$OHfD
zLV;O5h~hLN&e*9cc02qY@6!KrZBV8%Yi?ytE&5eg8wEufmAnG8TLH%crmPlJQxA$X
z)%vkhgJzg~_*+0Pm~JJ1(W{im1$)$ooBa+XQ+)oA3A5)4XP~}7d#6b?9uN#U1E6>H
zmN^q0g<|m32boiYCiRUq$<Zg)<i4>SbL7yRhVdlNH8zGDcoF7kz%G>o{p?2P>Bkgi
z;U(Z>cL0>;YzjhrxJDhTF*6A()ZcxQEOob-!0%zV>UVkqk|9$@1~CF*U}*3GpKh-^
z;A|?u7}4k@0he=?QLfDkUMmL!+6>dASx{LMIM5ShImM|$cnEnQtWX>mwz`&zr51MF
zu*KF-C1vh3#i4QL8<b_}7W!jWmI3wj2R@*e{=iq{n4>J4Kz}CEpGoxR2KqA@e>{^F
z53=xJ+Ao_|rC7?!$e9((tXWFUyi&zJ&!&`B+^W>vQcbQ_vCW+gfSsJ9dhTsCioM)Y
zNdrCPUZuLcs@zg5AC6MXtT{?eg>9bjRoiPEm2<1@isli)D8XSzqPg>G9oWDuQ)*^e
zY~@N#d95<5Y5_3;WmsdiV8Xn$qWm_cs=V4hueMyNt+rIwRL!j^S7w*jin?eXk#kvv
zWi~2rnT`Bw<~S-UZ_>TpUS2&{shN9IEr9dm9$C}5D61NO#48q(r4`C7o5fy5)w$JS
zE)&n9GNo$nTpPXf6Yn?KEHze#rM9+GdT+W(7bfz%t)dbew!kHM=2f!SZqZhH<SVV<
zMIU)*&8w-MYd0$if052VL77!Ox2DEXURjI!E^yfJ(oc{m*X#4Tm6g}GD726?t{mre
zjT1}!{OhN;m}ab0R+?IrmDj<Ihr6>#IsC)_xl(EJH5&2Z&RzeplAL2H{T=UBFacv9
zm>_c18TSc^9Iru7Q#k$-XbwUm3^a2J>5l{FU$+1SFIQSj3ulPR<)bnSXS67^!W4xF
zITYE-6$^#Nmn-zr=ngpoz7;{eKVCw;$H~p0C)T~er5HdOC}twyE_c93Awge52vN9g
znm?WoEjcJsV(EbwaKFn#p><8pW<Ps;-sKK+ecW5bUcb-RBtzLNLK~W#!6go7DCDK+
z;;n(}mVm^|U*VyZQGjtm{;*ej!@ho?g>yjHwxBNIV93Yl)_nTVaV_KSLY#mv7<6FR
zLUURiO?Z)M`-IC1P!@929=b0KR?ZOQWMNvMXv1rKFjRRgMR*V@98aAG1*y>J_yR3F
zuF-^xX1if%01*k+12RXqns-A%QG;qP#Q?|)J<icxk6xiNIWevA++JU}afurAJL_QO
z;8jCQ+-k5c;0)C*k>#j_XVvI+yGaUTJ(hDhtQFV<0<T&DkUbn?Ty@aGF|pti#4YGb
zr$NCwQm%nC4F^K}&U%zW9jqx@^coIzS|7CsqV>CUjxWLz>7h1(MxuYUNgX<mCJzWN
zKn-+ac-MU1xLzr)K*Oqu)<ZPavW}_J!4f)+gX*ePss^Too3>6MVKjJ0{7o+HOsZHh
z2*f2KOG#0u52=KaN)4~391vMj^XU_jilc}SaS9WQR8fq0k8A88%V3qOX&){i3i33Y
zW+ak{u14?abXlTYij&ZS=)hVjB7)SC=BZvO8+l!32uq2#qNKM8x;N-hNy`MJv(J1^
z!lnv9vwCgtYml~3(59$Pb-LB-NrmeG6JR+BrEeN7Dg;6=w8m3ZzgTuVpdEICv7!5J
zad`ctgkeLVgeo0bYI?u=Ak?*JZVCw#s{2JU$A{3y@%wfUhUp~YKy*nNoFJhM6|76W
z4iz>PMv@@MNzrf+aGXlTDqa_NrBo72?JNTO2~D396?q|r#cLjCNZxd~2~t^dJ-8@M
zm7G$WBp8*P;EBXO1`R7(x<qJA#qH&H5Cbg&(A-ZGsTLT!0Hg@QOBq3%Nud+s9JDO#
zggO{oAV$Kar~xKet!Jox2|zrR97h>lie#qiX~-6JDO6{IMW$eNXf8CF^RJ`Y2(VaD
zB*10?HcN1kV6J%TA^|Rv;0Y<<2?9*&S4O6nm*hsKhYPCJd8$=-o_ZCYr(%VvH6v6)
z7;*-j%|X))PiP4@NT54)pF@(cz<wGawCj`uI@!&b#XI>Kjqsz#-HbE%WtS(Ij5}O1
z4`WXU0tO}L#k)K)mO|50=k|vjP=z?m#&jcYaxBqsE+%PQNRGuF$C3Kwb6rvW!YTQ=
zM+!l(5^EA!5^-i7;><e4nI`~==xCrBNG1S!RlG1n_zR&h)w;kecRZ3BnnIMG=0eF8
zo@O(AL};GoBKU};Jk1l}1Mz4>ua2(!Ll3k8+A!D`#YTckmX53Ny>)UdP9-c(A0M#X
zV)#UxsohL#XeRj?YwVM#(PJhtyd<cG$jvNgjlR8yfu1|xOH>u)L5j*0Z+WU17Z}CY
zL)hVM2JQojLplS|3q?x3<OjbP6M0DEMvn(fKd2?fQXXNcak3b*A`%xWOGPGyieC&{
z%2Iw}KL-e)63e(o@Kh>QV%x*(S|$>SUEySGx@3m^!j`8`=N9RxwB=j__>#+sPj^Wr
z#6y$v)<Q)Syl$tCbe%dTaH2<D&h)i%CC{xXO-(Z^)d?b!vvf+Dx_)Oew=9R6FLSlC
zH%@{wQwpsUgh_E_Zo(UnD+q)F>p844KETO2jmfaDDjf@vrW9ws0ZL#fN&EiwL(y6U
zO_We?P9g2y4Ayj#_YUa6vE&t03;7?)O{x4OKPxt&=p3tQCTXRG)xX=V)MLr%hJfvZ
zZaZDpeUdYZF%rqvH9%1E>vt=m$gp5<^fpU%Wku!eDeCxhTOW`H%+1E=&{T((;BARB
zsJfurg*3``;JhfrdJMWamDI{mAHZ<ItxN}-%&xXnS;08d>Ft7knk|TUw$4o0;FI0G
zqAZ5%gi~Q9(h7GFt^@`yo1~G;Vi>w~!l^KlX@xroR|4bHO>k96N47xpnG!$TZn#3E
z+X#0It_tZMh08`7KiqD(LZsUWcMLA|r`f3@8$}kcAc;azHCtgyLooUgjDC8UvKj#t
zfsA*Mk&1`r&IKz$Jn^fVdHUhwd+n|&!cQv<wPQG;y@p_KKAMFKq3PGTrlH|CfVa-Y
zi-Uvk=d-`SHtWg0Zfrkc`gjx11J}pB5dTfc?}2yE#O{Nm=RMN@=eI=|Js(HdBXIQm
zwe<hzJ0gsp)Fx^>;}ZC!K4ID+K<barbO?G%0wEEl(|+XtDvr9597u@!I+0f^LK1WU
zYUFo{Cw`s$(C7m%oHRjlt^A&#8}XjVpH1E4bcXb7DxJt*m|E-dXH&66{%i`br%OKK
zP+1eYK`bK8W=Bm3TXL*-si>@ROm<Xwm*b>KprR%;DHMicyyE8DZAHaR(5*RXm5yz;
zP+VDqHF3b}Y^t7PzEMP=?K26jHk)~?Ko}5tJmnK|9fid_iqU$fYDrZ?b2<N%uU#KL
z%i_>hK2R!Z^+Ap|IvjN?SDMY{B9}AhsdEHju;Xp4sH+K?&2>whFklKeJ)vMlU3rZY
zf6B{iERL!<MY16&ISWCbn;`4V9V`E6{q^7VB-*2&_SE`c>-U{7xX*^M{ZzQQaCLBt
z;g-M!;nu*l!!^MWpWKu1{}S#Ga8bBDaPPpKf*TI_ShySE=D;n6n*w(OTru2qI15}k
z9Q~-x8gI~3bN$&Nl6}8+UPC+MfFk+a2GNR%iKw8??^niPiO}5agw+@}yXuw(3wZW4
zl44E2Z$$K_9iq}vpy<8i{i&tP9d*vSB_MTZOyb%Swp<Cin;J}{3&jE&`c#fKR1x`+
zZc-|L94oNSb8y~Vjit*~N+~vsXnXTUp*Kz4H`3@+Xl6vV$@In9IsN{o7LX|RpaV|T
zS?`2mHlR`?uy2jMERzaubOznnCH92Wf`S6AikuMo<A5`<B^T#~6om19BxNTm=>qy_
z2T$y<i!fM4hW)g851lQ;<j%T)2B@JB8Jtc2B~Ixlk%TG^2rwYxQO{5qm=C!_rP!2c
z%21_%YRgrV=i<19@l83#fKc>lb>gFH(f~o=11e`drbI+d3oa6?v`r_BDzD9}mu=R!
zPmAWk?jhNnplQG=VNyUuMJ?h{s;{t2DRvp54R7#;VIYAzB^oSh6xZ8@gM<%fQl(WQ
zDTHYxB8eH@(SjZFAoeCHOQ_KU6r~BOzHw?82z9F1dQ=hGFri3SqR7r<q86B_BBC*t
zbowZYSDFq;oaXV?;Uprq8Tim5+i{ZNgg_SU*h7?dqTRHoIu+uuTW#@$)nFK&5X`}Z
zV5yNZx|i(4SlDEWm^oA_wJ1bAZVlh178Y<J9S0mx=!1-5a#ZIQZ}?m1Z&B2GY^9-W
z;<n~spKV%Zha@cP8LvNPk*qFwvfl_R0v;M3XgQ0$DF>~3lz_L<5eycP))m7Egks+6
zR@f#M+jf{+3hJEL96_fExCq23<QI|LFlVVJCP|4G4;STN;G$us0265e6--4@7}dec
zd+qvKF+ZX8g8PE*Ad@xmvA_cLDxE8ZEZ-u}lxpJ$pD;zPspFg~emvHO(D}sYlLhj%
zq{7S?K(q$N8YqK*Fkz#u^^`<p@-nGj4S*35%c$i=iwsdC7<h30;>at92VpMa)wbm2
z@mMkhOOv-u`8Dzu73ZV0Y4ZO1^%i-vbxvyDu-3YDJ`hVLCgg>lgrAelI-ld#QhP-L
z;JCH2l3y#!=fcz)@vH5Qn_w3YFBI6b_*n(x=UX}GHcO4&QBz?rqwrGfl+zPN_)vnW
z%B?kAIy8s0wI}Vgf}#2;Q-rN4ma~CI>YpV}Z=3}%9gj(wp0%^YbavS3ZKT<jPYz?@
z3cF^Q=`VLXNcy5}7X&269ve3JEsojM<-+LSYN>{~Lah=XJri~y&0#OwX$ZkmP<1lZ
zGb1O+J)Va%7z`7Isy7yv$~lP!NgS5;s<Uqal4xqvNLq{rMD!_hDXoJ;+?9wftyH{@
z<=BmNQ78goq)05_bzm$IeXb7#d)I5p<1xt$rB5Jh7ARDooB|yXJ(2>OoGs{_l$ZhZ
zB>(iu8<#m#0|R6k;BfoFjfe^BCawAn4I-+psD^-h8Fr&F1&O$Fw&!3T$SD}6PSpMe
z-WZJ60GzH3){I*rHz7K3$a4Z~7%<@lVVZ*x4Ac|ARHBI@#-Q7+E(wMF!71a%V{}~-
zb`@Z9!l8C|pwa2|Egv739>@E`O-<v?6ALH7NP$!v-*nEC3MUIml$J9p^|Bb4bt0fp
z*XTt}1%r|!KDGx7R=CNAPJwa1KOEBDxFdrFN<my4WxEXxoCq#hv8=zqrTqmi?=R37
z4{T^4JH|kBA^;oq4&V%oQ(v&coF=>|P51;YoM-^soDgJ|g*^eV8PSH6kyYdM2|o?|
z7)L^G%;nb-;;vPVK1ddtNhb?QussyB17@hWG{nLOs0B*}zeX#pnr={`VC|bpKT*FI
z;;4RzEZ-F*)7VI1MXFJRac7IL397LP)udfYveYDnbF(iPYHCqw!%%u}A%n1l->mmz
zFUE%nR3Dp+#<bXEJEr-`giQ035t-(PEg6;qHDrvD5@vVTd&12rfi@bq<6*Et>yuc7
zr6h)(TAzp_EF~h@vPqj1x|;%*USQNAD{$q;QpG%>RGCm%s@yQCRKb|56o6Z?%nOG+
zO$AF}p#)(@JL$nM)6|MeVG>5KnEAw6AqA2k)*ZsWo%V$Uw~-w=LHeB}c#43HiHUYx
zRC$WY=xm@h75a6#2ghd6ZxoGY1jH0(!1+9H<OP-5039r&nULl7YFmZ90(G;^oi#@_
zUK?(LtOKJ@kSf%s0iQ3FkM)>t_N#X{I5DO>AWq`6m4YENu7j8|1+yc>8E7QJbs^X&
zf|LkrZ0js!QCb6I0Hu`~Ek1EQ2-pXvJCLPV2oH*h7?#>NHU?q2DO4adD-2C}HO(M{
zQabac9rPmG!1Bs6$E>P(Vx*2=ArdzN5|%<Nuw$~3OX8$8v?7+%OXYL9qOHBeh$d>!
zQvgk>cjQgHD?Fb5EP_rQrtPp{hFLOn&+^C{`2|T}l$cDG(e_fh<fK-tI;hKG-CnQK
z?I*m_Ae7a6mV1J{IB~Tgd|YCzp_Rr`6R)d+=oxSv7Dqs|%S{U?tZcQFh+N1)P6=UV
zMO5P73ynKZw^o)JF>SiBZVHn&%8jPe_{V$mL}<L87g)iOy&zf;By*^GMHdIgSlC`D
zDA41BEwX6$L5RloQV5Ee+Hhu2$(uyj;*Qi!tTfoN6isOCPw9{>)^PEHv652KATG;^
z&#zO|pmRAjMij@@#U2TIW>Q$s1406X&ZTbY6xHGIszw-#!#WJjChITaeVU%3<3Z<Q
zjg$;lG%V>C?9^?f<|ElaVvLR8AZ_{7dJWOSE-r3s1`1`dHV$`>2Un3MQ&T9&qOh4H
zEoQKX$kxET%%Y*EP~R{xW|Jn_>ae1%x&&Idr%AZc#gH_@T7%#cUl@}dkuiueFKmi0
ztZ}|2jFnxTuh9#{XvZ!Dr;u>zL)EeQw?Z>sX01_PQ&U*I4lD7du)@<}!W<|uKsGz;
z!7y%PLBTlHZYjIfaa)DWR#{$NmTWx~mzev+ueMZ{&9x`TPSZa0a<d1^PmSmqE9sCy
zn@=kObk`AEnod=<p$(iMPkaPvSo6Ah4J;Pa=9Ga>n<Bt*!nh08=NLkHBTE|M)9B7%
z4234MVjLq^ikMVLF3+cRi7^p2h69x`W3)v$WIb$DQ)qm*dWldD7vKaa?D?Uv!1)Wh
zs!3s44rJmdYdXyW7qfb@XorAMvwtc5S*FaX3kQ^2=nr0(E4Pww&Mk_0d~g}8PIU$6
zBUsavz(+Q<7(*yb+Rws&D>+9q#A60&UY)HSFhM#d#|j_S8^+S{TAFdAypRgxam*N<
zL0fEByb`BfMal^xrQ`Z|NKqmecttc4FEziI9#UpRIZe`N<Divd%2<H2W9J$vApK35
z;tZwa84SBH#ij(}z(CLqopbp4v$6ZX0r#7fJpLfx|BUC_qi19P_6@ivQ}Wm%-yg^G
zL%54R_^-RXlsv}E_wV3253czeaBV4h+%MnP;aPq1Y-~AP7u<fh_uz*0o{fDQt_-dX
zt^p2Rr#Q~@!_oG0{!){tM(*!;KaOlE<3m0_jxsg2jMq#sA$nkZ3o0i3H(`NVub4vc
z$%B#b300M1a>xB97MoIDO(e~0u|*71z^9lR5J*y>BGu$6;~5KxLYr?gc~KZv0Cije
zpWn1pG1VbC=_6@=g%@jo`~{Djh#yT@@zX0tMiZ^ySz(+MzzceW2wBhsT{B*Q1CyVg
z*pZ<Jj%ea<i{Aq^E_}jd+XSonkja~PhY;*+g82iQL*fwRqZU&l3>6O|$gK-%_P`1r
zYeJ;Zyr!mr{*BK2pm`0ACa=2@GyZZU=X>*HsI0vUuP!o5*RZhE?WT&?Ey1z{b#I1W
zOid=j4wURpY7wc%<wGLIgIC^Rbk<Q}c#ALQlOgD_KoE6<NXY;p=flfqjX*GNfUxFA
zNRSUon>^@h6Q7c!ShOa=v(X<08ll%gF5X7MtydowpQA~DAT2%5bi7oaT{RCD>0};M
zuPutlsmaN4Fpc-RG?^F&sIa<#uo};n)Gc%`p;}#TDYKUsGz*!TPF|4$j5bT~pG`<;
zil;JJsVFrzL#ZW5PE`hy8xtAF#RYW*CjF$EV`@<A!jw&cS|K$-A<`2H86ScsU5SB1
zTU0NU2x*-Nk*}8s`msF%_6vTaC22xEWzMCc57DlWX&D;TyzCn)i?(!-F{bg5W0@*$
z3!zmDk>Lu{a%_bvrl#fJP+^2ZWD5bBGRA5MmsxUIULTr-%?WfI={WGD3rp~j2ZBu|
z$ek|SP7&ub;zz8dL86H;@NfW1|CQMG0CWisjXHxXoQ-<FiH|@aL{Oc{?IT7(@jN16
ztTK5)?nLptWnywMOD+*vQ2ns5sP{F;s{wYRv9H0i(uC>(9)A~pJ^-^4ez~cf@Ffkd
z@uM|?#xR{G)Tx#eN0C|BbMwwDmLgED^o^P_j+3T{cr>BP1o7$y5?)2oEeS)F#Kod_
z2>uK%g0#<^8pMq*s%lz7j0~jurDm2NSV=%PMv6{`G-J~TeagBq>IY8fp@jNUDJ8Zt
zP?J8D;RO1~Vnj=yfQ#a7X|X=T;HCyhyQw-?I9o*dbayG$Kx#d>;_>oDAdM+B6ys(d
zFa!@}mL}|=f#GQbnY711LFzEdIg#FCMUZzNF-p*O9ZqyY(}QV<SxcPpq_G$43@DFj
zq=80+MlLt30tgG&F2n{g@(-xuCReI%GI2`)#p?gJ1aq7!PNnhyiG2bsFDyyI0dYf5
z0EeQ|V7>orjaa|Z8ZiiMkEg+d1&vSz(O!%X(<Uo3cygiLHZchXNgqYn5{IvoGv+X?
z168T20m`6<2mu>mPjHFMfjbo)K?`o`P+rd_8pl?Vrl4Ab)gz&bq#u4!)#}RG7^v}T
zlv=PUSZj%mYr(k!O0+UfEfl>PQY$GJDLJW?Tw!tIpH5e4;ixImMtzu!d3f>&Lr5UP
zNtDG$chPiz9QhhYp0@+#&LT}TeC#^WTtSYBLf9TfR2_REeSgRa7f=N?T)H8R_(6L?
zA^gB+T<9IAF2r>P3t^$Z5SJ1x91FKl(zvLIkObA6=<*3wD%5;J8K-5;dUGa#2BLLV
zqz7QQrNalEJnbi2k~yi~9$2twga!Syb2%!f>vr1SPoM#cN38<1V#nytWY)L+6p}pZ
zF+V4dK_CDhjtvTxPC&?2QZf<bC5QO9M9k$RR~S_#ZKy`~5<4ysY`qT<szcZhI}n^v
zq#2x8B5C^IELGXxTt_F}e1Q3dbuXeLWWxBoLi3DFC?+1YbZ!lyCV@B%cL3EvA_t*T
z{T5T(<vg_*<q-vo`ox3Pn%Zg@?^4bJPFczVbUB}c7O4m^-SwN71PL|xh2_{F$d8zl
zmm5rs>4!kcK^(OoHWG;R3TBV5SeV^)%+C4~*>+md=!lt=ye1fYV^9)j-vnjXi5#`X
zju<~!70ZK1f;z-GzN8FA4la>HVyQrR@SD(z(r6_chX?UCb!=il&_*5FVCA8VJ(5To
z1HLe>)`C9TjctXv0thA}boI7stSz@!jTtkB$F|AL%{H6UZowuLwoRZIjgO;RDYRS5
zrRZ04s4-qa@_}ZRA|z#NcAF;Yb56<?o4#^{pyl0Cfj(Jx6&*ogY|N3`=5aV4Ha)Jy
z*5%>vCl74VkwHjG9tGxxajLnXzkC|d$apP;2YVr8NruIVmKXSpj$yG(c9$Mi=g3$r
zSCcZws>|7g9U-}t?KgY<sEUp9boEYJ1It2aL9F7Cbeyzl1IU9bGPqAP!B<~>5V{^G
zlI%g@tdo|gmPN^>v?wc{nE=M;9G*u-`8tA1-wX^qNvm?oU1lk)|Fj6O(Baj+zV@7L
zVSl~aM*)C*ld_FR_-grTY66LHCKn8Gs$FUl;yub^PzqVe*IL}?G-sSz8Ln$`hcSZJ
z61Ck#)JD^wO_E5hBp+ek6bSNmlw3j6bc#}GBdPx)VDY5m=F~o>kxi@cI65sAe^ZMJ
zq^P7<h{PMKC8EoY_@FQ*v5;HO8#pekFB%;rX-GqyiEapx{1(p+Vi2x&fUL~76zKB{
zIm7+t*QD0Y#qOe`qRe>xcvF69g{6ijkHk{0FFZZ2Ao!ESD`xm6&r)}jXNk`zK5syz
ztdio`$-*E+YMN$Aq0r<ek#Jo4Y&`|;0-BI@l`>B%oH*fn^J1EI<i4+4?XSdZRsO)k
zwdcqlLwY}%n?&aIZr)Trcji>}_<EI&+t}@vY9^@C%9~77EmbaNslH+}nUo@WNsTRL
z1da0TpfW*lJjUyQ?0e13Ix)2;<V?0L)Jh@gUsS3ejhmFK)$|D0RCL-Dib@l2(xi$t
zM?D=ZD$r4rPP%3dpD-;-%mt8++}=i*&SG}g((97vrH;(nXqlwz6trILt18DlPnFR$
zllq(zY?4rvCThcp6X^&*vzs=pCm6j^)*zwrNQ@35jFzL+9BLZWGCWc6s6z$-9xrZ1
z()yOf4$>PETO^Ti5u`tjsgt!SqOf`&i^K8x0k-Y5A{J$d&P}L#6ZPYcf}OG4BBx3=
zU^|`qPFe}z=BDe#%}t51P21S!h9@%9`$#sYZO&$BTX1*B6iDo0?1ht35L>*s$5v)f
z{lW%BGzgOk4{S2R0lq~f&Z(e%><5#Dv}g#ar?tUME0e;o|AU1%R1Ar}Npx{Sb)_Xy
z<r)(LWL&MK(*CPZ8(=nuFpkB-3Z7i!s}NC#xbiEJTRb}&L!32GjN#4{vg_bH#kxZk
zH-zwdn+({QH%WVh9<LX&zX^7CG(f~l5GplpuuhKVn?@w7;hZ9=ADXe0ClaNhGPMpm
zobcr=R03(sQ}L)PAui5vJgze3EN^7ixLudACcZ<~k}M$<@yTWUZ}#2?K8|VqA3wXZ
zn{JcsCQZ|3tI=-Tv|Xf;5*2NuX`8kwK~NRc6+{I^5N__hy@XZ;MG&M~#8quj1W`c|
zRK;x(R0Ty48dt9yRBisw@BN(fOtLeXrPt^4eZ79K*Y`W^<~{R1&w0){&-pv&oSC7I
z-1LXS5=aP-yu)*Epu&r30zJ&+bc(`?c@p(HXlgNKbmgoP;u=8nO5G*hiOLPGn6NCy
zi{#FyDI_u(20}F2S2pKqQd0n2v$*!qW;CTcc!ip2l~eI_PpUTd7hZ|T6l0DK^Dk(c
zm#Di`c|##1&z6A&c(Z1IV07<5-v**tLENFrxew|cG1Y{^P)u|3a*LUKLAFY-yp`_P
z5PpWBX%5bq5tDO>NG<P<O2I8NZPo~@qhz~6w1l$!usJCq#%=d`Ff{G3QAO;6-w2x4
z7xr^ICw38BiI;g#Nk9BZl?8HsIy@aXS4ErFEJ`o9_{xiownI*-Dqm_E+w6@wW8h#+
zCZX3WE~2zfQmYkmQ=-yho&#6f<o0)?u%DxBV6@hkUZVW<E%k&&fLd0$Gc2&+1lgXR
za~`r+l{w_>ITuiSMg<g0*<AF|I#Merx1XUkpq~FyG?^+4TC_4|i;Q4Ul5tm)d+AAd
zUphZi6xMriVdetY_uNRB5;gsdo%X6k!zL7KO#YnBy(a1inNKZMk(8RE#D*-Q!vmI)
zytscpL23M*2?ZrYZ)Fd!^)wfJ0!@8hS&R~oXMFHnf>>B|o|r{D{KUyPqrC(JmQ7_v
zR4UcFsmxz&(zu{RJv?<j-iM(!bl5B4t?@<=B`1u+_1T81@<8Se=Rvf<`@ybUN`Eh)
zKTBv`TZpV<=@DN|WYM{N-c*P(x_GTypDFP1a;$$Zn@x-0Gx3Xs%or&vX1Bl{$LrVN
z+KCf-HjthxrRVUBXY_C?50~`-Yxp65X?(<54ufytLuu>(9)KZk6g>kCe};SuAI7!u
z9Ha64q#;lG=-kk2oE!cPL;~eMr>6e7-0EM{Aob5PSpAERQUAO{)IVRW`e($$!0z8U
zET20VaB0{l5*pzcNa;5Hj1MM6K;wH&k<gmG)@#r29LD#<BB7D5kud9bj<Vl5X8z7$
ze5WcB8s))2qg)thln(=qa$=xSUJNwKje$n_G0-SS1{&cQ7(dMX&j{DRgyH6YMm!80
zk!b#B#LGbcNb^4<o(3K;%KXoWw}D9qn*SN;FmUu3^FOMu)Lt0%^h&&);P-wl*XsIj
z{m6f~yP>Ze{^)!U(0iRiI=7QGbQl;e^LyvyhuUtsC$W38p|OKJ{AGsnnJ7%_srSbA
zZ8w#hTl?Ofh`}fxWn2o7FDRi^mrKvZS5C%``|Y5{(IW<z(i{5GF`b7u?c)OP4V-~}
zoB3zWJ83Rnk4k&dPnPf2rR7YzEu)z#B{+`Slu_FBrTF+xE;7r1Ur4XUdHKU3y%AUg
z%){FNnt`=t=qJ{P^m6ba;41J%kB9UY!df8s;L{=9kLxO603Y(H0j2{RfjM;jOh{i0
zY<@PR7X#aXD}WiBLi$=@Ij{~`1#AS?0CxdB&xQ05F!(}9FTh9sGJqw(d|)}S5m*as
z18xC^fX(FoVo2WutOC-90c(H>z<OX1*an;d3<2|ip3Nb>5a<V%0)xO+zzkqDupGD<
zSOshX)&N_9{+AFxU<S~Oj}I0B1Hg>ekWOG*BlG}0Z$S_7z&c<Kuo0LK+yyKGwgXFn
zE_~?7|4-Nx7zCyO>)(O>fX%=Xpyyra1!e%(0gHf}fo;H6!W|(!4Ieoy`Vi>@=6?b^
z151Imzz}ddux1zRgAXAVeUAJB`ddPJ5wQ77$OGGewZIT?3ozp=<To$}xEELojKfD1
zgI~ixunM>s7y?!Ti&|kXU>h(*JkXm2J-b7CJ}?Kk6qpYz1Ny%U=?%aja0j^q_Y&?2
z=~?(lV-c_bSPCoww*4FOzz}c;(DOq`Zv*DEBR=?2Q9dvSSOi=QZ2mE%*8tmqo5}qr
z#EW=fU<~~I3^`yUa0M`9FZ2WJfo*jC3(8Lb{((in(tXGWV9u|w6A+(v)VqN-od|a<
z($fV!z$)N&y6#4L=o%P6CG__oJYW@Y1zq<dJh}$9(X}4ZlgHy4m<#la4!xYNfwgq)
z!1vPV8kmrbYiEa^2dn~?(zVv1*U>d_7hSvXZNUk+24(^M?hbt!T?5wvGotX#G++(T
zI}!OXs6$T&HUo=+MV=153fKm00hUL1=qZ!n9)s|JZNOE)njszfc3?fQh1`JxAMC1$
z#dqm|jliYAHsC6tC$2-^2n+(Z19O0TfcZdi5afo!A20+g0G7w&dw)RxunxTum@&LV
z_Z$qpBOnLN_v5>Iz|hDJeFv~+6w-YN{3YQVeZZOnJM<c0{b+pO57;=SL(fQop7F3R
zFqn+*#{qLDbm&Qw;XbiLF9Oy~LOg-_LBtQ}KM3|og*&hWSbi|RHwO#>L%^Ct;C?9d
zq;%+Iz~E%0kFJ5Oz{b=L9q*JAZNMC02v`909145VHL#wpry!odHXvRDD5?(c&<lY@
zN1z-4%Yj>np9(u127gC(=nH}6N8wv|K>u{a7g%~U!aE#%7VHl!orU}a2Js!sW?&7_
zl@7ZA{lKDZgaa%EmI8xw5g%X;u$B0E(02s%&xd`0LEtiAc@FZ0uJQdy*HqXQ=m*vx
z1AV}%V^Q9KHNefl{NoVsBVm{05pQ4-umqTq*P%BMo`~|80r!*ey+B~b$;c;Q6>uFe
z1Z)BpEyQ;gfx$}=?ljn~81@3@Uxs`K76FS1uR^%M`oF?1gjXZ~rz1Vrz@4xRb_0fh
zyMQ$nu>TCCb2Yv%m<j)DQ2v1RHSiBCe+1?9XvF6k#1B|hhy14NXAv)8jtk$+n+bhU
zo%&K>O$@#@2+a3&>f3?MLp${rVB@e(JtYh7BRlmhV9_Xi(+^larc<v5mX7Vz_Yfv`
z>S7k$59!of36JX3(`Uoqv`&2m&@;VLZvgs%JAgr88!!Xd4a@<?%|ZCUWMC0616U5s
z1=auyfQ`TsU^B2B7y_;Z`e$_NwZIJE7GOTG8CV3|11txMZ0G?d02_fpU^8$AFa*p4
z`ZMuOMPLwE3d{hm0_Fg#f%(8XU=gqpSPI+)EC;p&tAMV#2nXl~)&o<3jle8mGZ3$%
z7j3{IU<kMj=sCJmuK@ai>wrOEJ+O2p@`J8tAwPimd9W9-2pBgH@jDUr0_GP$Kd=b6
zkodEqpRUh=eqbrEg09bneqeba^aHDa+krLbK|e5ONvFOSSbBb^?wODH7Io@Lz>Ev=
z%|~GA#hrRKu<BCiCA<>;a^U|t<kJG^{|Na341N#06aEbQ9s~DY<QFjL?b4fpMI*ZO
z<YS@V-=!A=%agkF&A_73UAiY1*W<eMd|*wmOJ4^pn%t$g19PTy=_$v-PDgd=%YdQm
zE`2+&D5pzLIv&>rUHVdBQ5p0Bb1J%Y?+LK?>Mp$mSii1I-v!Kn9D4Ggrw)36rS)BU
zBjKxEdcui_*FU@T0$^|(+<`^FDqz))F1-!t`JzkDI0<&qyYzBk&d6@P8R!|)tp`sA
zKdxIZ0hTW8*0%r~&*|0^79!m9yY(!fzoc7V1@x@z)^`C*pYPUFPC+=ky7ekx&ez>~
zD==esx1P2L@zA^VWkA2^(Hntnz=Tr~pCLVZ8L(}1kKP2#7~7+#=YyXB9#{jc1U3U}
zfNj9dz!0zr=$Y7~w*vjZ5U>;&I1Tm(rUR>hEf*kO2leP_7ee3U9(@JSe^`%R5BGZD
zUXsh`(X&oR`eyX#tAL)&9=#oMrE_}p{EHB;;}CD~rT8{yH?RiZ=iCUs4d3V#OCgVM
zbC#U}y}(LfK5!xUsxx}@b->1R5fAWf!1Tq4KQIT_yrf5uyBPW}fIs4wLJnAeNsnF#
zzNiH81Xfk{=rhg)56lNvt%f~-ZNP#{5TASDALw6;cmkV&!9O7$_rV^QB7QIT=q13S
z<{rJS81Vr%1N|RE?kwaFa0ajum=6pAyWyVGg7jSu|GRMwECSZUy&TvCYy*aXRlg!V
z1@H$fyaIBe9=#aXrNFhguIhr_aa{wf0yY9S0-J#iz&79xU<lX-^mHRUU`7w@b2jw!
z_UOsLa$p9q9+(Sk1Qq~m^d7w&Smfx{w*#wy?FhGYM6ce4>vG_#66hV%tG59CV|(>I
z=OTUMdiBby;BP{&zPJ!_z*1lha22o-SOaVZ7M~CIgM0PO!1|P4Jx~O@9oDNC0gI;g
z>NzM!jleQs@JL)EzpH>Xzz}dRU1#*_NyxWGpcna+F}+uB!gUd_mGTwnMZSfA-Ehx8
z8hUWu2y8_DRn6?xm*Kh%SON5BK_2;3KdV<S05;F*)oa1$WcTX3$RBVi@~3fLubu*X
z_;Y&o8AyN8G0;zT%Z0ylpbuCM%s9SR-vZ1(0rFQuKCf3V1QwlwXMtd^#znn)F0lF3
zUcC-{)#<%@39d`e=+#}YW6k1TeIc+BSORPVHo!f2Cc=dsOM#n#!LyNGxO>j&)jj7y
z4jBA1?0X^n<GLQW4A;#UBR_$jOQ8>Z1~8!%_9;fZfS${H^>x7fE0B+HF9p^Di%JmR
zCBQ2YF0lM6qz_mFY=eB;GWf%_|L<`B3;bP;bmKastXFRa2CqZ@gU??MI|9pB^y<l!
z@7KdFe?`1j!mhwzIqZP^ufGxD;yUMMgp2mZe+%pcth%jNucGVQkq&ByDp4*W7rLug
z&-xqm+>Lw$w%vnr0L)m8^euzjy+|LhZ4LAT8`mPgfI0U=@81#b14u8htr~jEaJ{})
zPrn*^fq`qG_n+W#UH^Ws9s=g?fPe5sAEG>6gZS)3c?M=Q!yeF64=jOu{>R84@(;{`
zzl={%FMz>M5gyRL3+W_(z})NL@3UTgEwJ%(*a=wk1?;pO@n}K#kZb$8SN9`+p>I%6
z!25TjJ^+i_kUsEL-@zWZuK5Z1umb6G=z1wI=+gBqzzkqBun4#p*c`9xS=U2ugs!gw
z<^!vN<-pBA|AD&hS&96dsOt-X<-j6fJ#YoEDpl8;fH_C#deRN>4@?7=19O4(3v|5-
z7&=kcTY;XFbUg&j049{f{R~|X0z+r%`V63_K-cqt`RC|*CApu6_yWs|p#Mgs>mpq*
z0QxW0^%}aq4Cw%RE=N3X!Zk1pSagN1F9ntY%Yfy;N^-v%@c>p~4ld(n_`{rB8IX<y
z1;X*ylZ2x*&N0e6$W!4^d=lJB?+odOl1%oXID%yS((x<433DkPocQAA`TWPk#{Ah+
zCT1Qm{ouos$AOi47X$0>z?=yYe2`x>9ir^Mkp3LP5wm@9vwi+qzQA(Vt=er)?L3Fi
zFFDUizQE<ixeRk7#61s$^z(sMf6kK)(Ye0BEaMzDgJ>&$d5Fg(@;A#Dx7>NQkD^ZL
zqr>snyj+?3L2KT7T!>MI4f5bd=`Dl}&W^(ROcxpv_(yza`vS9l!P&mFS-y<<zNV<<
zgH}b|;=a{&o3_&F%=Kl=@})_4NKtBL`8=~A1zj`nL(?J3R)=)m37w;GrECD5vwT7L
zT<%__Y_ZKr*Q{eU=}4nYsDs^#?+@v*IH!ZSMsQ2P4VR4Kc7ZDew=0Zm2bT>SPD0pp
zkY5+h3my#VaFPd=2|u{S;ARn%?F%e}#fYbO^j2DU(vbsxJ9szA;~;J^xJGc4ZaOG0
ziotCGhi*#RvdHJj_CdH1m#g6B`A0}k)Rfx*JD0h_?W|36+8eseb&j3GTwj?>#%nLa
zX;>H1V<-+dD6Sraxfz^MCMe&Nz%_zH7b^1z#fr+QcB|8Q4xOSPc*t)y<l-<VOZK6I
zxO{L);KK80DL6kkbm8(KxiWBt;4;FvN^lk6hJ<l7;8ub2g>jp~(cJLZFs=#QI*VK@
zxM~X*0=L$}d0l8DEL;FwB{(B4l!xiyD1IkUxG=++R5FnmN>3i#w!qEEbK(lYHGuPy
zJPzVY!EFY&L^6t71<sGT>;uENYH-QmCNq}{tOJ(?ZVS++K4$yM9Hw$|g3GQrkeDiy
zE)?2^M??Cf6ltVnxl^l&LV2fn2H>{vDRbVYgDU_RggYH1mjiAYIDkBeTa1fR3s(%T
z#KNrrR}7AlLkIb-0=E?0U6N7UMsT4pzbSBQ0Jj(1a`J3#OApeFh7dJawaN7E?1h-e
zPb9ra>nU8XJxKdv@Z*SwH1br|_W<}3@Ura$e-85*;8%gCI6yw_eDKLQ%>%y<Jcc{g
zI&-m6XL8_F3b}5`bw<j=0#g<~5g`i;sJ?E2u3D@aETb@yCPZ_&(|MU}b}6h@xW{2_
z;9vU}*8P#fqI^jkg!0!A(uYx4IH*2kgKGds{hBI2&dXsAk%r3xxFx@e^#E#I5w}0f
zy0S$6Jw5F2X<UcMPbK_RzZTM;W<P(EZkNfwN>2F<_tOYJDO*DN6E@-eN%~ZO&j<^r
z8-8lu2<e+lelC_0s!l6E;q^NW=}LbyqyywZHp!;6hH+$*d~hk?W{?!peGG?7@+EM~
zv&ff&%LO-<<(E6pA`^-Le(P{j05|l_QrCQ6ivt<_kj!9ERPGufvj;Ngkc{kc``YT$
zjdGU`r*_DtWBo_=KN0#}=W*t4m)hpwVAK!zyO8zw^-a`YN3C+-B71CSJ`D=`J8T|5
zz;6Zo+SG~5jJT0qRzZ%|rl^?Gk%M0~xK?mSq0b^a_Jym)9<zK|rh4+2!>*}=6zSgs
zUCmg#ipDt|#L?TycYrg(r2I+%*96We7ky!JTGcQ|%X5S!R*rDel+K0lpSQg~9IB&5
z;BvvmQ<y3psHO}0>S#H;ciX2k+m~lbr#nK|pGRTsfG&YGH@mPX&3oD1NHf{lgTX*I
z*745gy3m(qdpjTaTXtRuU4f?l^i!Bc;F7?Fhq(-#AKd>&-Zny)_kDAiyTEzCg@@S=
z&IRs&Bb_tQp$9+c50mOi9=K$17vh}zBQ;juk9x8Se)Hk?x9SOnQ3tu&KNv<k{I>2#
z7=8?RW?;SckECNE{1(ISZ>59k*)qsALJqgY18kLzirwF=oKH6zrj~{LZ-##l){-~a
z`p*mVUopUc0DijRzZCv&ApfY7=*H2k=JD-HnrwRm7;v^i<_(ifQJ)OeoovWd?CkG*
zey8r78maD(t_tYd`_cbg7nPG;(3SsjNWX~U-q+USzRX+}PcvH<>4CT(1NT+%i|Hk!
zy$ZkQv$VU;0=s568D0yZXAAWFR#^x_t{ienSpQ$vA0FQm<A$=txQWCZ0^I8$TMXID
zNH%i1&a$S9;?oYlb?}R673nwZp61U+UqoT}(P();{bOM)gx@^){k<@jL9XTxhEWf{
zd-fxY7RY5|KgA!hRnk!83;h0GI<g_x@(05xgWu%Oem{&F$Ss51Wt?w)b<A9rZS%Yp
zejD~93@<uN|L1=!j2!q~48Ol;t76F2{lPFc!f$9l!q@@1yf6M(I^r;}uY}*<N(Z%5
z8Iaopxw|Ppes^5vTxQokpb?Sx4=bQ63wwMDS=Vy9eSl7?uhkSExcyf9LG`tf;sd!S
zNk?CsV4hEMUTPlKqUgx6vkMa!1z(2ro7rzzo3Pwy6SCo!0=Y)Wg|$(?)g~^nOEcN7
z1bPJaC{^2!wX%KYd%sZ;`ilb6)c{=!v9IZb{nF+9*e*`9d<AkW?8W3+erx}DiTc<8
zxP{>U=vbKc$L;v0e~gqw=?B*Y4#m_cJEpO4CA+^GDNW|F@Z1RYkz?T&ivMo=uqZCH
z50BhOb6ifO4Ciq<xd(p7J&WBBVE@Ni{1^O&JDO%u#_lfikAw2G5?mhQbiIB0Eq$*x
zGeY|1Jx>dCRkroppX$n9a24Q|Q-FPSWx%-LGHd@Cgm@YzPFmsjwtm0dPv7QrUT;jd
zsCnMSkfptHG?yAV9POs?`zSTbi}nCs*1~`Ad$xlmo~H4k=64Jj;K53yqXjxjurKdc
zvgL2;7#mSXvLE#fd-?9M)e$z%)DE`Okw@_;f{vvBg!G4gSI5+dIyOQ_Ira?hg1+BO
zi}Nl!bI7vQ4jsuqV*bXyY-Rh3!|yYl|Fmx}ilwd@Bhep0S6n1rrgqHvroFBLnb+me
zwfQH^AO4<woG;s?El=daB(!(5{ActdnA^4KlLpL-*4kDaH7`o-eG)2l&AyQS4bE-!
zS?ZzlUAy#AeJ+HaJnY2`M6#2qJ##kM>6zurlQyS)ugTcY`R7Qw%r@_`*QIRU4PE&i
zA^kl{-+;OiR<4~l+1h*tyk(6-djcKu9xQwuFiXt~@siN_VQq;qRU+>*%HY2c`%*Vk
z9K!wE&O12Q+ogx%(Fi@epy!Xq!;^&hnXdm?JTedu>U6e1&wqR0Onz6vZwP+v?q|bp
zX738KC7MgsXEeg^((eEHc#O(x0u~?|pz8~q??;}zW|t>4wxeMV?TcR66Vk`9E}B=T
zFiOEM27gtga%`#-+92Diib_FYY=N$dUUL}4?*dPImT}7`({UWaCVnr=hx0SQdoc(o
zhy0f;Z@tIhvV2OI{rk!?<zFuRXXqh)kSR^Z7?gcK6gCEB-$D2-hhH!DPw%9#Q5VC;
zp!XYNP;|7it*nD=17u}i5?*Ie=V+o3^CleF9&^2w;i-0n_Lrw%KX$}1mZ=@lQX<&D
zfW|P{$lO}&<wo->k1YJn2e%nql*txqjA4%Bnuu{+3mxUy6JB7|q1pz^oT&3h`*Nt#
zOZ(*4Vn6wf_PQ)>xfT^6y;)Rt(*qb+U@!W$`=`qjNf+6v47!rBZ~ZK5pJmNUnr=2t
zf~m};a<>IKvarY9>%>K0c|f@%3vp4<PQxe*1xS`@wH}{G<xY%6eaAldb~=Y&lsg()
z(K7x;vfRmFvgEpb3S?=|{5q1g^cP|6(>Y=F!x+xdHhZxY{&RdCefO4#OaIvM%Kot<
zm8mMou7xa?>*PUp-w3V>+<<a#+gE4@MY22T4?$Nq_W5HvSRS(ej{^_xkESmd{)_Nz
zz%`^%*~d10l$KJ+Ziei3lI_bg)DPRVIIpuW+gWm7$QJ0z8j8K&q^mDqjCM!urEsna
z%YNhv*`OQ#YTz&2#&P4Z-URMs$Nt1Cay_APxfr_Y>6r+e_s46%ea21Z@}kyIs^B+i
zm^p46!O^o7rTZ5*Z>0REybeKEZbFCtF~u$1{^oJZ>ae(RYmNE>FB!?0Z-D>sK7{g>
zo^xph|EPT&Ep3)|c!W6Sk*>ASm4jzTzQwsIj`n4wI#L-)g7-FxFLXt^=k`oc?%Vb+
z-V1*#-V33t7|*)=w14s57%ARlgKFqX#`7`*`Vo5@Y>uQW1-kY?R~vMFNAa@O&l`>Q
zhk99*GyIY!qJA9Ep)bNY*?+mSP|iyQahU<PJ#agj-TK?ImCiGa>1wDV*+R$$@T`w9
z58vOnUlHE53m0r(0og*xK98}vO`gy9HJJByOwoLej@iB@{F?D=aV&s@w$BYV8*Db%
zW0-Z`&!iOj#fhskAtJB-7OrL%d$@pAlZ5jnTwte{495^y5-#wrjlkLA0zKgu_uFSw
z0~J5Aebpq`AI}ecigU7qv()FAi%mAvuGGP;1#SySI8uA}w7I=QwU%>S?eM!7&m%>V
zUK~`$T*#z$aKE^45#AQ4GG?u_ZJK@D^Qeqbr@9W$H7#IWE1h$OL><?EPpvI-Vu#@@
zZxk9}3cC#QTV{6Xx)vd9)BU)X7a{E8ba<y{)Jo@c=+ELjqFo*~+lAunItYL8EY&aU
z59Jf(44a5(r`H4hMHG~@e>&uwAYaDv)^RfL1-K-v?UeT^3COQv_)WlbTI;Ccw6nE2
zy+=8GrOs2aQ)#xZ9KQ<3EMFC{8owHp+*<tVWeHy3E5@d@2FE496MUtm@UnotEP$5<
z@Umclmqu_@0g9c)@LKFEzUKGAIa|qSM|tdu!CYUl?YuAevm?mMxr_3HF}}g`VnrTX
zyUh2so9q4G2ia+x1u@)jK}9q4Z9fCgT{<J^t2OT>JjCIvcOhvm7}e$Mv{NqQnRW=u
z>zR1o3~Q&h`DfXC*5$QRHB0V2qcKPs>AMim3F4d%ihCuvvWq%&M>H<_>bG(45VkI?
zjrN$w#3*JMy=pI7hRUGP{fRx$n{j1_{yNSn?y_#qH*S}l&!J+<G$x=@v_NhwwJ!t9
zD7N2RAKhQ0F;afdE{$_ZOzJCyzJ{w%=8un{uQoAUo%Y#rcKWh?wKSg!6?M=Tc%wsq
z%sy|e>j!t(?iGO^vcH%t#ELgDrl$4^2kB1$7ub$xO7~Cyoxi2Ofb_r9p<hYu8xGQ6
z0<Qf%^uzY{9WYPVWINbU@o#{x$`3m9kK9td)V~8<!N(o?EBmQ`{cq?Gq+<QzlMa1!
zn0@Hk+Lq6MBmU-l=Sb0)_Ai636)kv<lWdKn1;0vg#b0;m|DjGQy!@HQrdkMhbbXtx
zFgTs-OS^mkQ#ybdD5IS?P}83W_`7O9m@5XzoM~pT#RB7QB&+mLtml4*K4*v}jjGQb
z(BA!Jh+Umf?H%cBguawtJM@{9=f{Ta6JwikpVa8{yuSVIciuI#r-nrmZ3T@n{D@~#
zsKc^PZos|Ev1XsDOfH1qEcli63%Ur4z|8>1#Yl3?z-55rqAR%yaOoCq9k?_LN6%lU
zSoCcN7X*jpZh6GvZwolmhbgZ<j-J`3_!;vxB<Goed;v%2bPy(i%dzN71Gmt^WrHiQ
zaQWbtTDYa)O2LJPQ3kFAoV+jLa4W%;S^U<3TV>%kgR24;Zqp`kl@_^HaMczr1a2cZ
zV;o3v@*alqzJ&{bTW8_Y!I8djzd7L6TDZmFD1PB`#o#D@+#E_fuK-tJ;i|w<dc*y0
z1Xpg6YXG;x!tDS@_6rZA4csyd*A0&HBU~=-aFrk7Tr#*~$c4ux103aVxLhu{B5?Dm
z4&or27JysV(V=f(j>h&S;Ig_p^l>x>q<ac2F_z}BNxlMZ3!^(tbz&X3JaBKooet7b
z56+9{)ra=!i0ad^18#8#n02&~Jh<&rg9@t~TqC%NVeyT_K)Q95*>5tq7I6Ow^P2%~
zb5f^1Jj`z{xW)s`eha{D0T&%6R|2kKbf>;1OkX*;Ts%*o7AChATv4D?|5F%O3$6~&
z+vkUITfm9&o%)fS8Fbl<KPeMC^-<KXU<}oN&#7JR=pWJt{vLK-0G)A1p#Kkc>Ydam
zq3^j;-gcd|AAgUPGXL#gr8(T!{)B~{-?$kJ_Z5CVQ0G9`fijm!ooe^D+HoH4OP!w}
zWfsZ5*hvn%o+JOB=kv_*#hoqxULgN2kTV4;6cyX0k|}~&OQvGJJf&0DW-A-_-9H(9
zUyIW;9y}q+uJ6nCwK%ovNp?;cdQ!Gizu1DX7GBb+zk>O1(^$d?tHBhO^Fcd(*}eut
zU)qr<&$o2y&ku{Bugt7(wY|PFLth#6r9arI_YR7n4_n>Co-NakvmL9!7AU(kwrqmF
z)|yWJZ!r<{nbVtZAH4=A*)9nK0e@Ymes!cViaDR0d+n9Uyr8%(g}$6uJN3ipz9xKJ
zXXHh-$#xOqR_!ER_0TnATc>_08FHX5%bdk2_A$#-Hgg??cKU6+JA>K}J6+~`y0h$*
z(H$MdD-XK1G<E8qQu}15%e1bCTl)+Q#pJYiKvn}@nvi98Ep+9-k9U+%2XCj#953yo
z2$LgaTsHPVUvMYdZ0gAD^i`P3nr%9>F$Y2E%$SDx^G{K)BE`+DD?$cTIFOBE1#}g+
z;9VtF`<ur5Cxs0)G1|vSn#yPc{3f(^>TgA=KMh)V{c*P2865SSf+l@&(=k55dpmYS
zsy`K`c?0LS_WCNwETpdh`jUU`)R$P(rpi2-GdzA;z_xzRf|$}*3w_z0o%$_)C?A-I
z^L@4E^e(rLpDgMWKU&yY5#Obs6RCZ)=)3xN^yNX{I=r8xg~laz@hdYwE4#~HUl|qA
zLj0<sFMU>*PS3&E>Z>-TS9`~{4$SpcYtA?9f|>;v3dfU)_UV`|{q;y~lX?Eh`J|n)
zxtM;^p0b?|m<K1acQL}*v9L?O!xJGrP3H1YU|Sw&YFm|u2IxyZvrBKVuha5Tsqp?|
zMg)CE{}FdI%Ky1t`dN|cwRwIr(>Af#MMZ7IV(6+lzf0e1pH|CWEA5R4-MrLQ16_-Y
zyY$78%7M8rI>#=1Xv9F(W-oLV|D{XM#yOR#<K=ipjjQ8kqWr=wi{0?t8cn;)x|{;H
zM!1c&k3%-*SWR^qyYg&WcM?<jmO)?5bzS-ek;<H<?_V_7ULWNhl`VQ_RpY8IeNq4y
zc6G1TZHixWgt^IDjD1O8QWoO3u}gPGT8pVL>x+pnCtJb#mO@|A)-L^d9$Q<+eVE`g
z+b_@dLCjeYQ~ma4=xfG1i4Kj_wwmXmmPaV71(a6LEZon$)1@DYdqR8LnZ_pCwu9~S
zq88&3(w7f?8}at%T;Klbn_`<6FbVVpp`wQL`MP!4v8cS|we^))mE(5|U>7mfyB&~?
z8`iBicx}VBmeXtOQ&KJ4tH5mZ4|t!`_lMf*Li=Ja1D{1OatoJ%V(3dgp<BOrcm#db
zrv6p?erSZ)p?xZX3VNSfc44>vuW^y|c}!(~v2C41|KQP%8xtWYYQD&>DRWT2R^lC3
zk=ojJQ(qLJ4Qkg&S1EKYzNK4li&S>abs@s|saB4~=v{B(_HO;(m>;((yOy!Ac2$CX
zI^~*s92&)SHQjytVJvmjvQNml!cG<KeWNrkhOV4Py7zaF$I3{$ilD0oy0$~t2KzFN
zagVuv7DveETBmk&v|T`|s>!;WFc<Y_Q@4KXP+NP>_nGfgw2U}=h1E`s|Dd9f^u5`w
z&%w1#{#a~E`@wCxjk&(8Y&#yl$<~bsqr0(NKMD21CJf7(h%<t!0yzhjFc0y4ySwjw
zAT%!(1m^{ZB_w$eHv?P{9G%lam<KKyT$E%KR|qZuj*2=RBv%S9DU3@7w+fse+=(QF
zHjWCe6H%pmqPafxTq60cgIpfu93+Q>xJGcf;Lyy-gXVyCk-pD*)E))3j@Gvy=^A6Z
zgZXtaaz9bRd?B*_OMi49lM8?O&E4vq4fEw3d|!B&U3s0R-T5^){jaNmKgz#y_}vV@
zpHcz__T2;4gkEvm6$Yhi2Xy6qj`lxN+hQKWCD;y`F*>6(1rWI{Uv}#c;=F$?(z3T|
zxp_?;w{7y>2aDmi=%;QyoyvWm{pl%F^?I9ePSdM;q#Y{Yr(|!pK4JgrN~~>ngWzQT
zLs#>sJ^$7Ff~YJdEWrG6uel5d!7T%a=|Opr{u$ufp#NlI`qHNo7pZ)a{u20W8H9K0
zkv|-`#S!J;x}ocCB9t!mY?XPXCVV*yD>Y<kF8}bq5js8O9|w{yc7dw__rAR?t>f?K
zBiWL41&_h}zPCqzdjHDQYTGgeJyfPjpert}M}LYca9>$B$`o(ieKxF2q5e}!lOSFT
zza_BKIGmH;<xac{10!^Z$a(Q5xGy}iN4;}u9=Yej{dL){?}1xB+*<70NUXU{GL03E
zu}uWl+$N!oL+~?>MgC9k(dSTHkZ1jK%zcm2%C@Nxas_Blu&gN$@>>dSnZcR-E-_}L
zWqPV1x58+PkY49pIgKi9LGKE!Il4!`lC+|+bYkiiTf*njS^|}eU2yl{-HV0fZVCer
zd1LD-y}t;}U$${hWit+g*Mu*7^gQx|gKVA*E)U#zBBaeR_scd9g35+_;F})(Tyj@s
zx4)q~&1h&<nJk8EA!M)Vmqj@@@?P!L{99PoBZ~9r?J_}VR1;)i;FsdM8M@m3(xVR|
zT{!68t_fT#xciBaadoaW)>ihwt>~{k>V0N#rgANgLwP}bdQ4?ePROOrq9!)WR3yH#
z?VBN$09CorRdYj+dWRIo;-?~D%IiY#wct;p@DL|BEk|BanMJA!<O!lCRza=}avLJe
zom4D0-yp@t+I0e$EL!fMaCbpp&OJT)DdFKRHHS-fbRCcV2#|{+IUL0K!Fkv8nBLix
z0?q?&q?A%}S>O`Dxx%=G;NmP?5$lTylUv66!nq1?F8Do|J9ml;?HdW;U8K{(xO#B8
z_w~1TB)1)04!G2?Fj~MBf-}-acG(NA0Nf6eGuMl^jJl8or=$~Lhx>c<m#k$2jS&r~
z`kph;);Y}kaHus><*@)dDj(=k<0e@i@$MhEQGA!dt-P+M@7-Dj;3~isw)N;oMryN7
z^T*om2<0(vv0cf)AWEhmVPyTQN2gm|bNVkdk}uooq&&2jkjtU;t9m`4u4)<BJ7KAt
z)W+w-@6zw_eqO6z+Fj4@w{xCo_KIpG>trST7W{;F$Xe}g%$b^P5OEJNrL_gRN`6K;
zW?gd6Q?7a{z}ZDRo>V*F!Jsqg*B;BWVCHtoi6>velbY*GbK>O@;gq~rUkF`+jvgIL
zPlnwfqs9eG`<p}ACzU}q`6awp*gmaT9X8+BI3r|u6Uqy<)lJaXiudKtqOeikN{p5o
z=hQy7!`=02Pv1Uh**}7h!_SKb$^*ACIH!Z;1K<)O$j^YAA8v!Cw6bF!xH^RYxcxZ6
z+MZr(JLW(cp|*J~bcOH^-_J>Zg!1QHWX^zNkta*cl(b75bfvg^^@*fQ)<eA8iu*9K
zn+JoqRfFx_$qw{B>&-)Y`<~-kj-02mL+y4p+#2Av8|PTh=XW%j%g!~l@kZ^tF-pIs
z=$tIaa#VLJp|8N#t9Ot-#8usc8~S*Bgs0&83LcM-$YxFOn;hG#$J>`Di~a4FL#Q57
znv)jN^B@D$O!mrv+l&#twqpV+OZjl$nAoeoO!2@$aa#&5e`IgueQ)M^y(L0Xm-7_$
z(6#u$Uj0z}bXw;;qJK-53ms&^%jkEh6YH~g-x$EEYpU(AaJDb60N(SUYs+isgZ8ie
z)+5F17`yfx`m^z?Ll_lXdh{audqvZLrBfTQyg0{)LPq(~4L$X5_2@_2r`u9iw1928
zk5zRwXA$yuOs{^6eSTPVT@)eRvffuhR~Fu>{m%aBdgQlswLsUF@xA)rBISp9j(Cx6
zdxRH6q`_V(r=mXK9o|<^y!ytuG!jtvv)D!$z&(I;@xTw&g97-;p46)^p!N_4wHKv`
zQ|;z%y%BM89X@V<%G^~E;+2;Nee^E;v{$<I>BvV{tgXHQ&rQ5!o0+q)w_DAtCL!KQ
zuXpS8?oL<NSo=7d*Porc?aC}IjGtiVfY-4itU|=U9AU-1->u%U`)P!*Ol^bh)&Si4
zXdl{6i{$vEo!SwOA9F0vqr-ANU>gqY#xm8PD<Z5x(E1zMK06=d$#1&#4CJrt9$TFQ
z)_6S;Y+u1TtDv)`ty@13_1krpt<C}Q%(RVXPTxA=9_Z};H|8%8hHJj9PD?xI++mk%
z^c;Q>(mmre>?i-BTc6~CzL*I5D$KUrY`Z>`gSBnj=}5Ahv~dl>YUt?J7g2qAGhkm9
zEn&4sXcP<^cSC2cqerK;F4qQIodecRKee?nJq)DELf+}P|BLF8>u;_*Bj~iOS6^bg
z=t-;6O6NN0totYK;VJEZ4jS#wz;QShcKZhe^2R>RdAFIOITUKgLeN=u2)^Nfb61RN
zFXRLxo*Yp(rDYDh%fvX9ep1fB_%NkMKaa}KZk69wKN?=$VR)i&3g9OP-^!5Tyl(F&
zJRDd|(04EPZtu~Lq4Yk0i559l)3Tgu+C!FyddQ@u_UMnngX?cfuVtMAb2;X9q*q7U
zm*#e~Z&Z)Hi!uH?zDGY3Yn`r(luZWcthhDC)Hs|r%w8$vvY@j7Iz#XD$n_7`Zj3y`
z#}>$p{u@ZVAIh|qopOrvA39u3n9m@)KNqe8<3Xyi#-r|chUK*!59VQJtiMEn!rh<J
z8G_Ed_iztDHohiYXV{p>yq4{}*)FA6Hbb;T0KeQbG2T1BN6y{5wki89$BR_a55_qW
zILQ=4rVTQfh7+zA4H-;Vv5bJSO33Uj>d}vng3MzP@}c$yQ$EZ~6{n;M=gFz!j8t(>
zs+g-T@&U)E3cv}e;&l1<I8a5Yq99dVi8IiD1>~lROH(lm)^tNmp%1F(q>3};fB3mL
zRg|P+gFyWaG3Xm_h{4lAH^eMXjXTZEz{^Rg0ujhd700HE3sOa)ij(b=ELg$5A?BD=
zaklieQ2xawB8T#VU)7&5{=v6fUM3rSFxFgd`u571Z2-@R0Yb$G&#p-om!yglQ^nP(
zVlfg}-k0KYQ^onx-T9|faUmrdX)5$p8;OR0MCbBUu{1UAWM2&>8d1f=V>iT{k{Xw9
zW{_wkH$2f$N*Qs6U0FesCKC<YEHVDW0E<$Ek#v!Qv{#;mG~;_UD5AopA?+AVVN6Af
z26!3|lZmnpp~Rqqq(f#4<SReGcU(|*T;Gp3Y=v7RnrOhd438%*GKOWcUSvss-SD?+
zMUQ+Y)%9e+Xji2_UY~F-F#Mr8r2wg&NH4(pI{XEwzrO+g!t1n=Hx1@~uikd>2HPXf
zG!8F^&gxq*AC&-|CqQT4y)Nn_0+s`azL$Y&F+<IhO)_@WuhH0Y2XyU#J&vck{i|*L
zv8)AZzeMON3n;zGXQMto*rQ)g{rt&e?bB=7r?T0$Pg~HpN3|3>1CRFTZi?fj&}m7p
zT+HoH?{yBF^wPcvng7t0{7H}g0^;hL*_U3Kk9jKp3yo}4&7!hXH~ejczuDCPkLmM=
zel@HOG>tj6<7~Tsw83Qe+;cEq{;Egcgm@YKqox0_uD6_KTP$YzXo_5{g}$Yjmp;S+
zeQ&}>1IwB9p2`^^AG4C6Zx8h4evA1yveOMppYcv^a`&Da=Kg21JH;mj?i(NL(MMCc
z{{S;Y1LI>YSFhNn4^mmsw+#AbU~cgXoVyYco}~;>-^kA_y%5$na!pkI7=06{9&?Nj
zU;)6jJ75>5`MwJC{Mpa8Hwc(AN4`-y6L3S4j`_j_O6MWAcDIa^wT%(dNuSaah0y2P
zhxq^pkjBZDbW$tTpVVImlytbucx{9pnxA}@^n7ibR%^T}Ba{c)qA9weZx!b0Vo4tj
zr267z%ro(OGu{mwM6mU<;D5$>n18_>r)+oc3HPtsto}r4c>hcP;Ks;6YFo;nqrL~<
z;-Y@ezAd%xRiY>RZ8k@sb3r;N|8_v<4y{*TMCm#f@v<wY#imglUVLY(Qr5%dCFox~
zz4}3<FVR+?wI0Sq(3eH=D}lZe%ri=TAKBKU{nkhMNZ*@U5u@h&T=zg<q<s9)J|A7@
zql|j-9VSZO@^JtAmyZ;O9Ow|3C%l<-ToJCrT$Z%SC<$npWxGg=kLcUTCgt#7jCnzN
zN1^KkTbo$R_JRmDmThq}^wl2FtA7ODu2Hu7EH-vl+82Q=xi%jx!gy>{uc|A*j*e{O
z2z7<N?G~5BHvT%?|NhySY_t(Nk}%&X%gqPSF|b@)(xm-nIqZhMZs?<T$++&c)n{$b
zBfYn~fZ93wPFV`(|GprbpD{3v=625V&MR3IR6AD*JzJnh#%t#PJYKEPw>QA?8uZ8F
zMQv5eg=jDFy+7Ihet)2y{jFs*V*3kqa@@D<A|SqfcMTB{>h1ong*QUi9?UZ~k}e!^
z`0WB$j&JH=+DjfZUTO!o1Khufi4dmqGjla1f5D|l|Kk77Ul9HZ;LnBc0)9G3h1b7F
zLVI_RF`9=HmGM=Otv?yxf<u|4@-5fNolo$bLIC4&HM<bF$~aML=V(&SHkdQz3C4s_
zu?zbAi+c43aZYLN58Jug2%6e2*ToorL(cr3R<vDm*3|eI*Hr@=FB-r5;XfPx3t$_x
zG&0`C+E5NI3*eTF?-YKI@r6yDwU(J$+pz>D-ALC)=xRK-SC830U7tp3BS}{Xy1LK*
zpRb?D^k0JU{DgtuSEG7E-ycmt{hLDL>}{A_8Q5-NuExBF_zm0n(L9K$aUXrlv<dU@
zpNxmjXCmmVFijC?lLPiGd_{~}Q=#u@&cOF5Z%3S6XC;~ADeuALIE@$I7MjOtd|ydr
zX(9AfLQg5_nG~+a=nGlJoUl43{in$~K;K_ocv7!?-;?Y0QRaB4`%6nZ@7!*mGS!~8
zLSO4?+@3yUtIyibpB$mw6j1uoFGYT${gC6vi=Z!38)DNJQa?f8k8Oqjon)U=!~Lti
z(A-blVYV~(6It-P8G4qX&dG7h^nrSQZ~RGV5yfa9@y$?LD|H<buEWSH9s}EqohU6?
z@Lz=bDC3YA?tlOKb&3Mj#TA#K{W%ZcU>%GLyZX^$dS*)dx9yvE)ZV%#=qkLVSHI6^
z(S<ThHc_QdyTunaCP(p;@$g=b_GKCR%1G^tWsd!I+r~=uJ^9eJ_?q6n{kYUlECrVj
zZW7KZoy(ngZWyOzgL1eng<G*T-o~6tlevvJ(Pvlcm98DoRe^7(UVwWz8+#5|k2%h^
zy;121T!H!b3UfNr!PSDZOGh5uw!m$9lx^4p+PXs9xe}EQ`sT3*-*i3AzFb)AZI<oh
zL5K#81=^r13Ez1|6DkkNk8W^&a0e43^8;UF#3|*6zXW3*xV2EfZC55OZT73SYdcC;
zA#^otKtCMGCYJfK>mt~M(oqXt$xoQmu?1WJoLxHTJI(2EOHPQ84olr~+ddqj(jmWP
z{93O*jf`Yx6U$t~s0j6x(ouRP=3MZ7-SGTa1<nJ`nhuh$fm;&X%BU>hAZ|0b@(=LM
zTbHCH*95Naqkg?vaB2lt3oei9bzggJw700#eG?GcROf&e7wsIp1+PMQ_@3`sRHp{$
zC=OFG7dsBne#)6NUJHj8(y<gens@c;Z`#HK3l9D9csx=(YN4a>3w;0jcjMu-jR&>c
zd!Zxit6u$FTOFo&EU;Aq9aKkV{2A^4H`wz)=^0oaEd8`LD?-1TMR{2cT^o1z>ZjwJ
zj&%Iif(wDWmxw-_TlyxvJv985ItbAn@K=CuBR|#e4<c%lf^&nhK|z%v`WWqA$QtF4
z;zHlXZU;xjoDLeB2f^(DHwxn=o3sviKH=R6580HVug!ysQs~?KGmndrw*ImC@y5^`
zan^2i7+2V+9R98jyrPYl0-y{B<?D9n%-`3mpGO4te&2>W2LG)ZeBR=k>kF=OV-;a#
zRE{rW`Jnl}tV?7EVlAmM{eQvS6!aK2A$7X$xKyAXU3tt~Uh#@!$PY1MhvTq4G2&_G
z5g){e-Oj*!F=CD8`Dcu%ba`Hn5j$PkAI6A(xIM4Lh)3NqFG1ezei6(dH@?BPVbF^(
z#8;lbe-$IPMkm$9h_x}^pJGH!jQ7<TQ5W+#Tq(?<Vg(A<3THH)3%S+#BNbT52kV^f
zN8I9h=YJqr<8p6si^pB(2yvU+`<Pp-aeF~*aF3O;(?kt))HvPOdBo$+GMqo|a<B7<
zr(M_M{JE$pD?Q@wLD4;f#K(iY>pkL|LEaS}@#COlIIOwiQH1-b=Gh8A+G7atGne}{
zkNC><0?yY)x&P@A4@O~6X?s)xPInLT;$w_I5AtsD2z`*Z!6R<=V6k1rdy*hOFE~6~
zyy7*->S!T8jqyC~6<^295n{d93*$cRJyglg6K^Bjw>9^kc=5j0gSfsM<^C~V?1<U}
z9St6LOT5@R*t0ucyfgR*l-14g-cY<~jQ9QwnRqYwmUtO3?3gHMgJhHQ$cLiDBW^c(
zq9@!pK?~XV2tkv-70&36!N|whR|bo9F870j#Url!AhFf$zhkfn;j7q##frh++QFh?
zu-xP%{l*I1ykFzH&na$k)UdvBq89pUo$m5D@vL*7SBS@AJuBnH53$~x;zVccWVXd*
zaR=n?aJrv!iF=$E3URGA>}i*%)Q00_belBygD!ES%lo`b+~t}@UMQ|di8s*Lzv1%i
zixzLYHX+32QSQ!YQ66<NOc|Ac)7J*O{}U}XM0-Qg;+beKK6dy@GzNv_f6ZE#<5rjW
z7tH!EZB+YA?Rr;Y%S`QE-zm#8we7Kibu+Yg;u0R1q16wK-ZNcm9h&&B>DqO}qT8lx
zj}1%wcDlALA-Z+CrY9tRHC@{`+_Pi4wrj+M4KuZC6GuKdQ+wJUdsn9Rsvox;PmWBu
z_GoSA$mr!qYx>Ceo=olQ17h#Y)P6bu^ADemO6bVcRwPAtWoora@xNwjl?TS&o~b=_
zAdGO!=!9P~wYt&KKW1vXN5}ssQ`<Hs_O?vzlQHKC@lqh+-<eu#AbNMEcF)-O?=rQn
zv9Y&gYB!C8<G;rxd<|9Oqrc458pg+ek*Tdqj=eckdnOrb@0*bDX{NSiLiER(+P^2n
zf0U`cH!=3cOs!=iZY18Fl<;Auwr^7OyP4YhVElVfcu?#OncC`uFuSTBl&}r-!O?GJ
zYC8{(e=}2i;gHztGqtx4f#Z`Y39o}riGDRxTRA!Y6(lRwvwntld#bxSQ@cC08@YAI
z;U|7JT`Nmp2EIG}R3Sb%!h7%0+80N7^-S&iBfLLnYQG-g{XSD$KGpk8rgqy@?`N6X
znyKCoGPMm;y>BA{Q@w9wYOhT7LglupL(qv%9r4i&?Z>IXni<;aBjZuPjvVpnbZyI#
z!H=hFdydS&midek?@rfV$_TzaUE6rn#7)z+x28>eV7m67X^$XARWsfHm?`d?DQkdm
z{2gVZB~6yE!Fgj*=r5b2%8TxZS>YW0j#K=@Nes%L*?moe)A6kH#Kdh!Yeg$i7WX(3
zo;+GBbFRkuYAyQb>Doiu@SW4OCN1&h>Dn51^rmUrogNR2zkKiz<m`Qe-N^iVhc-G_
z&eT?q7}7OUt4bXC^-S%lL{zI<zx(l{wdzqZt7mFIjhYRSrlja6rfWZs@l;ILDguE=
zrfao<W#~7@4yl=`eK$6#7A1Pz;h01kHv|$*<2-9-YR%)WABrM(xcBR$weJpx@%9})
za($+DM|!Y)hW1Lj7jfT~?nSiQ(tXIvhmMF_H(h(~h+#0+J4eLbJ6-$gh+%h4*M2zy
zT2@XSy>}WiCAfT=_U_bqLfm!a=%1!(n~n@#J5Bri$l*f#eB|g4rfHQK!B39THfO{N
zu`MIEVVd?y#)ucEX<Ltq{qQL5!=rZ6m>OnZiiQq0y;l>|E0#G9*m9V7%yGcx!^8)U
z^cM~jZ#kp84iz6bQ>kf~@bjUf!4>n*DWcImWBXyEDJt&g!^ORW;@0APP-5d0G`zPV
zZ}-I{J~Tz#;=K+kZyS<u-4yZA5X_fU_~P$9Of>sO<ArJae4byYh+g0P_YTAN024!L
z;)Xa9zj0{Njw#~dp>fSq#8*QTpHCAVLvNs%4NKZEMXVhb_wW?)!?5_CG*OihU6&^6
z5~3^9#Lo%QYtqDH!xLAhiC0Is8`8w`#KfD@#J!2Qy|~4n^nRLn#2@!&n)t#W`*WJO
zdt~BmY2xvb7?(eIK+;=j;-v%P-bfRB4~X5CCjK#M-1}3+$D@w!KzK<-FCQjW9JtBx
z+F{~_(RG>-Ka4HK`M2XJoxe;>DoYc$PjdfosJMI5mvH^{;OGwy71tgz>h(j#+Czq~
zJ5+QY;#qyDcsOOow}*<|DZ@WKRNOb&-!nx#JK5hgMZ7)PyAy#;_C7L2+?DEWN)zv;
zdS6Zxw;bxdK25AXbPM(W`kJ8*kH;PB7?O=FL(xL(EjfqV<A_q+I^uN41d)P!CeJ3X
zxF-03S8Q;=?3K=O6<*Ql9M_E=UVB;%-lgn9%=5(FIz*gXD8!qNBkmX?wm3cOhlm#E
zkcuJV-_G~YHyYs&8k+3ztFSUuSY9zsh;ifcvA0x)ohVunxmL%JuifGnXJVUMtaEn5
zCS1NSB06v6+DYO|N5W$Vp<H>dog_YVo`E{1jr?h%xX+ca{2)>0_C7IDta2~Hz{EZB
zfr;pD6MhVerw4iOn<!o$1iI2Q^45vs+QA851V!s$@2ZKSeK6>)(Ia~%h-YFF-VTbp
zz22?~;vZhnA@4}Mar~Ph2`>gkqtE-V31Wv2^rNvOKb;`%iA#7SC_3Z3A5Rb~hl1WU
zbmTuLh_~Vs?g@&G!@P|XMEx+(w<e5ydV=UmNVq;IJ|FI_ogn@-9CX8ok@rpzk0&PJ
zS-c9r_pS+ItsnGH{*ijJ_+n(jr<25M2Y5S^#XARpUN>rFYqGdKDPij*u{X*4MY1S6
z5cJLiN4}LTULT$C>?E;%jCV`2czO)z^1#R^lEqJfgmo~|Snngr;@a_EnC+JFURY`M
zcrT3d@OUq5@YHxOlKj$mFH-o{crOz4;dn3N|K)fuqWkahUc|0rycdzVKG}<~Z%_6j
zsQZ$=(ELa;25K^YF5IHI-;NcVUG7b>;#JowWZT<=JdLs9!$DI822X=-jTOrWyYW<5
z`Cw$(*Mk$EiWQxM$83xhyJOs)v0_imw-`vg@AG!YiqC!C?_<TkeBNELsOMg2xh~cV
zueZih0g>SiMcIH&oUuP?;u+`rlBRN!_>p7SS8mbm@O|eNpSlONx={{jo=jge+!3AN
z-sN(P+~$(CfHV<5)DiUn#)2+~QbHVdv>!hN9l+T!s~lo!ft1r7BhZBi{6^fYi5Hy{
zTAV27N3&bHsE7A@#}GUq{;nhXc};xo7<s2AIvo=_o#Fu}`nRW?!+JFFrgOskn%Lu9
z4{?M8ojCc2Bk_5s*z6d8n^Sy2RgU6ygkv%~BkGXc54apZIvzpb_qyo(0T-!*sa@iE
zmwT(r@v;j;eu$Dk8gC<3M~}eU8c&_DOe~X8R6M<4_apa+C!@u`-5<gVB)?{|W1MJ)
z=VnLjjc&(o2R$9T+U>sIEgp2cpKv?YyZe+Y{V^XnqN|+ZGsp4L{}6}E?e-eh)7Mz8
zbw`|_k9|Dk7_;4h@iLa35XU<l;$e^vJBEDd6VE$_eBu+|Io)6Q#1GDAk<N$QpE+?+
z=2;FQl$&)vlpE6WSoF|md}4DnMj&^_#BcVAhhuP~xYj%Jx>!-~ozUtNpLvI1Wau6B
z0=x{FQ125@4vB5@iI<1C*ZailL%a|8#JfY#54H>$hWW*PLnf^8i8Vgz`Pmjv<Me5V
zZ-?8_;J~O>{OWdhyM^v{uZnW4pjW<=s63F4QGyoJo^`nIcH)izvAW$!r+3SD^vE(d
zD*AEFF`)uR?f19lAjcTGJ5C?^g~KtIO2BSM+^-Il7WWNK2U)C7SYI>6F>#9L0jJ}r
zFCllY(~Z*yoRE0Z>8^8%=bY{>PRC{<`t?&8cnTf#Q-g-A@uJ!5<h-Wx_6)?HaYS!+
zBC0VjIK^7$G)fHRo7*4#pp!Fl;B{Y|zd@X9q2(8cyVD_h9PXQ(j_blB@CV~OV!r!l
zr&#ZddCMtYa(Z$8mUFg>+en8`jK}*iqgQH<$zP&?^rscJ94)2`F+Jf?hvNi{%wKT0
zU!j8Le%Il6Gu)I&V_zEn&)9u{4bkR|if(s0CVVh3&e^!(ANI7vaVp}B*uLoy|8%%N
zayZ_##QA?s$G~!^JErb-1fKMW8=OgB4;HJm5s%}hSxdZOu&8#QS{9A6UyX?0<DvWD
zhdk~YkErq72!_f_o96gLJ8a0i^F)=FcE^0NJR$z)Z1G&ei08A#`r!%9^TfUp2|MPA
zn-b|A<eNtNA@k12=TH%@KX4Szzdtbk?RjF|=n*f@6L*do{`@>qGbV8Ze2p1?=R9$9
zAhCC@XbKGPm@B>sB;GYoJTZ3or*p;YV-s8EidEx=zcg1oI4=IRxnlRY;dl#H_qh0N
z+2XbF!|%)%AB~Uyd5(A@dH4-;#Ouk?c(qJ>a{Le3;<^d3Z)J;@Cd5B8SG+qR_U5_b
zrit+{%n@rR#=bL0d^9nB)g1BN#00#iWy7TS@3O`7lVab@7JDbf*UuFzgRwWx6|V;4
zAIcUV1Y_TtBW^t?{?TkveNe&!+2Zqq;#;#t`$4g9WQ)fRj{jkfc<JESJLZaxgNGoc
z4<8cW0LvVbaNQiSJ|+H@Iifx#_P%UUIXQmCT(N#~LQl5%VlqkIoEl%1E$&NAxIJ5Z
zoVvmR>b1k;pUf6J4j=wtws<i;K9nuqPEU9l(MlhE%Uto?5wWYX#nvMdKbRwKoLb`u
z9i@GdQ3!9ZAB8o=&yE@a;^(7AVqJQg`}R5F%NakqD(8z`v%KHV6K%8nZSzFgZ12u_
zV%2Q_C-cOsv!~T%i>}#U$o4|sQ&pncp`D8p@_xuM@dpQHj^;})`U3y84sk6NS9$F@
z(SL!s0QU>B-5X2g{*bq=<$nL$J*wAPLS&72#Nk+Uu0Sv0#)EjTI^6F#99u1Q<PWrw
z{cXYjk@k#i^8b$iXMz8lEzqu}Ohs+JvO)`EE%T9VllEg@_bx@6aO+1$EhekNvFs-$
zXGo{*Rc8g7?4oEWAmxks+E@$7aw+$MVT`Y`9AESDlnU&B@o!dw3a@y$!ltb%+%mp4
z;#WLExmP5z{5wjnny-!c)ymC2Sc@2`<jQuiz5~>?q0dM{i+qtdn$!c8yTKd!^6H`_
zmG+#%>a_~vhG8s;15LX;THa86-dhTbS-xPnSuW6|+{->zSi$lQezRQlm#nW%VS^z*
z$}HE=t=xrM@f9pzJK8MA(G-H?-FN-}vy)SAQ}MoFvBHZPFK4ty_s`5PW4xAeCF9ME
z6^wT=-plv^<2uGi7&kK3G8*CjWs!<k9rN{!n;Bnae1q{V#<v;YWBicu6UNUOzheBJ
zv7PZ(#$LwgGgUnNj0ubfG9JQsIO7b)*^KiTk6}E4@l?jejD?JsF#ehGdd8a=s~9&h
z)-t}#_&Vd;j2|$5%=jhacZ@$Wb}+jBq|!Tt@j%9jj7Km|VVur5kMShNKQUg&_;*Gl
zf30{IyVo(^Yr*^ZdK2Rdj4w03#<-Pn8{_+oI~jK|e#N+laUY}mES3HdjPZ=484tDK
z;e350;~d5Xj3+WqXI#v9K4S@E8RPAYs~I0<+|1a>_%Y*`j6X2`$|wp{I${_{FpgzR
zVVuS|kMT6d^BAvSoW;0|aXI5HjH?+pFuuZQ&1dU%^HOt9QQS<kdYEr#G~@$UsPq`u
z2Kp~H`>$nxO&0&P7XPZZ>c5#cuJ>XFhYsU<j?a8uV7YGKYphbLBToUV5&k1oym1>O
zkGS(xe6ib29)3pLG07vy?n!)Yr0$nq{ZgXZxcgLA!hRP_$I-#PlL+7*mku@ijn{Kw
zx`__jc0h*_9@bh+2YbVwJ{6+TOJnw39tkXM<U%4}7h(E?4uhxfTGC;}JB<S}`gtS#
zE?uR^5bfpbs$}UM_kGIUNM}1=FTy<`9r)s*w8s{8`G5W2G)p}&3^J9=<2=Sij7u1=
zVqD31H{*K7O^jO^cQSs<xQ{XFL>1x)#)*tm8Rs!BVqC&_72`_AyBXIrZerZZxRdc)
z#(j)YCvo_U6B(y6&SPA}xP<X4#+8hBGp=Xc#JH7lC*!w_`xt5e039P3Co)cDoX5C`
zaS7v9j4K)MW?av>iE%6APR4H;_c2CcJxqqrIFWHG<2=Sij7u1=VqD31H{*K7O^jO^
zcQSs<xQ{XF6b_$pBI8uXd5nt~moQ$%xRUX1#`TPw7`HO+Wc-$KA7d02o@DwNCo)cD
zoX5C`aS7v9j4K)MW?av>iE%6APR4H;_c2Cc{#}O8IFWHG<2=Sij7u1=VqD31H{*K7
zO^jO^cQSs<xQ{W4?{7yiPGp?QIFE4=;}XWJ7*{f?RN%g@#Av|{#*Fjl%$XJlo^sBW
zmy}!?I4U(gHEr^gE9KRcs}D;{O-oOma**Q1xl7N!<ov)@=Uslq1(#kDm@+jL&Sqtn
zD-FFbZMm`@ns8!~*UAc^;h)hjhP!IQ)z=@Id3RsGY38GZ(Z87aLBi;l%)Cc_$2pv%
zb4?5uM!y*DrctlV?WLKI5dmA?D=IAcqVx~xo7XVw@%4>2%slR%vy2|P-$lcYSI;wc
z{7_N)h8-U-s*fI6E;KPr7~`6L?PP}pVT@<Y{BV)R;~X<TLfG|Vnn)CRm)eEv?;Ed}
z^^6obHsx0n2Z&-@-WZMbYoYO<OZ=$!GwLi+!azG5#C*a`b3AEy<P`n+HA$%v1waSl
za~a3S7;hehliv!r_IR^?IS3N*VvmKtR)zyU&n$l@^W7HyVdk4Hd_D7?6U}<IF(0t-
zpEI9p;eTYl)WW-D!sEq83!lh*vxPqxJlWHj&$HSy?wte~lJt(VY@FEH^90tDWtk5=
zk@;c^U&y?5{_wBhDIbk;g-dzdDEWcqpB67E{>J<yi(>(v^bD*=^c^g__<tqk4-mE6
zjeL>oCL;Z6($inRLfl@qvwRtsw;;DGA<4&zvSU?!ui^*|f<Fgl-2%Ca&t;yb2c`ZZ
z6|Nyq?;9t+iFpc-j>DL5xKPO({4B{+{8fIb*b8wY^EI3<RwFLxljj>QSdosa`}jC{
zP1dD@_LNh;)NYSA@oV~c<Bm<qKgxUq$4Bj45ZG*?^oX}rJPn?<!%{w`(dIuK%Q*qx
zFfZ6JN8y?dtSC$Q0=XF-Ym%HWCvLn*Pu^N3*Uj>A%$Gf&`14tw-XB5oaStlq(4WG5
z@k3_$Y0S63tau~bW0-GgP&`jVi+tt_URV6}94>u-OUAQB@mDcV-(@Dg;TOfzMhQA@
zWj?J_U7yT+4f9#uil^t4>3Es>+Fo`2DDyj*FXp`g+nA?4xD;+#KpBjumj$-DsQ64)
zg2qP^5|KY7Up-FA%kNVns1)Xn`Fd;}mPZcr1#B2fCmr-YKhk5&pBvweE@Qr!=gUX4
zp0&&y^W(;Mn(M(+{i@*lWpw{<vb?blF`v`hZ0O;22kfws$9IN2uRn}KS5L?H%p2<w
z#(i`T^TxWw8%kUxK!DO^tWU`AqQY$i^Tv9Gp(n_EF|S)}XFW3wJ-ps9k>febkmvOV
zL(j>~8|x7NVLcZb^1L1~ob@ba-dLY7^jyunu}%@q@^>+BtXHgI{r50$tY4(D{3Fbl
z@jAwzm~UX-SjWg`{(a_+bq(Visy{Ga9aQB_eybB<55hzZ*}3*Y#mnz%f)6lXe3Rnk
z_c6gwV?OPE#rLY_Se$I|>lH7*R|)y0hW@7%FTXtr{(3|H6~)VMNP@qg`C@E*p+kNT
z5=hG`6wj7zikBa}1OFcL74Iruey0)qzYIO^DL#O5Mu&#@lAekW6)(S?2!1T{wVx{f
z6_%gQd~u87<u?r>e=75Nt%{f5D+Eu|dZa(FNAdEzgWxNf7u*3-`$Y#0cS*juOUcXc
z0D^zpkf#^d;E><*1AcGFAFlWou3z+^AL+@Pqj>p!J;)!ze0#3q<@fNwFJQjmRK?$+
zx<gUOd_jTY8(>E|{?2^he8r!|_F2Px*+q)q%X(fm_!7m-@4P`zGxIHfSA4eAE<_je
zX;&*gi}^8_sFU_#UVf_$Zkfzi-=O5>H`l-yFduh^;wJ@^qQ5eqMO*rC$ZwzlS2Lft
zUGef;X5i}#J=md0hx{fP@LfaChl-cq7X!bC`LaREFu6P|7>rviiht8^#mnz~L4G3h
z1qUcze%lND9OeU)6feKi1^!&-<4#q){N5J$Ynd-ws(AU`Eb!}@FQ&cKIOI33fLoZ)
zyIb+{dsX0Dna{dc@$%bK;C1HH@FWWz?eeb>qcMR?=?Xlfc%!{Nn)!C-<@cT7X9@Gw
z4N6{qiwXRlhWvYqpQEaQ*u;F+4#mswAVL0P@OI<RHt=$McsVKvjx6~XvT|4zFH$V`
zK_kFZ{A*j3e(Fc*xRvGeE%FaCUuEIzSWgSj?{8o|shDsl{S9BB;l&}}?F@W~`HIgJ
zFW)N-ev~XP@uJnDKNURbN&800%Xd3N{xz15%i;<Lj*e~eZ@gG+;jy(;>M0na^c=zc
zMc;jpMg9u3D->?UTS}hZ7e|M2e^h6YkCVe)Jg#No$1~q<;iod6Jj)#a1>j}A9H+vi
z_mk350G`Ir)jaM^Q?$62c@N)b8QRt{-)xb8gZV1v4f$5)i!AckJu2f9M;$8;ddD3d
ziQon9a|<o;KbZNo)WC)vnXIQhSLvx^w>;(pCz$#3S<i|&W<7rePj+as#Q#2)ue8Y5
zGQZBkzs`J}C7wIMll=obRD6taNGxt!=zgk}*T;e!|A&}wxA12SQ+o0~RC?$>fsU`h
zlilhp@%e@MP_{X}9(kLCa%G9<DCT!r_*CXY7Je4<RrAdH7c#%m!k^E4gN6Sq^Ls4(
zEzJApoAp;SpLUFye+E3+x&2;cANh?KSf-8TQ!IMAn4e+ceZy6GD6o_VdU%P-Nfz?~
zGCPhxfhYUtTl8GUe6A&)E0_=EnB#vB^F<arKgRqDi=J1QUu)q%V7|`6?`FQ$!gn#h
z*TTn+Q0a0VYmVo5=F2ScJQ6(F-&o%@?#j0_Z>;x@A+<P`V4_X7OWe*I@?-ppH`aGC
z43NiX;K|MjmiYY0e9*$X<$`RyXt3}jnQyo7lbMgpHK%tL^Fa%L3iAss`~}RHTll{-
zUv1%UXMVGVf0+607XEqWcUkzinfDxLj{ld;&#>^nFrRPX2g`-(cu`{E4`jaD!cSp-
zyM>?2ys*q$o(7)A6@?{gT*2LuC}H_Hi+nlrK?}c@`Gpp~mieU?ehc$e7QPugwZq0f
z1zDU>(w1V%SdM>qUqOW|h4_Pody+TyB^dJcl8+U}K7=}!FUB(|cH^tRftTg~VRb_w
z-$w&Kt0Ks+Hsq%$yUF)uK|a#_=LX0-#89!;W<K98KF>m)-gsbLFE|<#VosZNjYg?|
z;C%jj9M8NTRXdDfq&(WB{$T^w3p$v$t}l2;Q~Y7wVsm*O2R@$0QC&(ePpb)f6`s+C
zaF7yCiol=8da8XYo|8D3o0+c-DJ|PsSvB*;lUP2W9A0GJ*jL!1vPXO*dG;Iz7ZupY
z^B2gEL%IssZiVb;CK^miZ!Jv_;W&i(#o$T4xIhs)kE4oNzF~vnuV(qbFdukQ5zClg
zW$3?65%h!-9c#hMA&t81`&(=1`MaW&HSoY8c#2P2oGQ<=*vHp~{w77lvHo`E#U+YA
zgXOzfPgAVYBfDw19T!k`2y9h?yggN%4xYko`c=iGhxMN~mgSA{x$-7T!P}+l0Ym<2
zC3p?XKgE2Rd>T>otv7CCzWO;Ocp1xoD0#jJgXD24KE-T!zO4~cB_Afpr%$qc%gc%|
z<X>Pu?QX@N!}8mh&$?0(k1F$t4<*k+VbBSl?A9<t3DUbs=_nkp;@MK9u9+5BFyGGn
zEafc9n6Ld@5gaXXr{q~E3|?dT0Nc&5Grg;h%6Xt&g~-tt-y1yd+tyfDH}m2RMJ#7N
zDw$fbk$vkmM$kjRlYWt?%E|jq<#G`7P3zV5h3w%l$s4}=@e-D=;CLGGS#I#WzxZJG
za(4th&qUy(Ca8F(S?u6vzTznr|5VmLmH7gC(FqRRcFAKt^R<_%>m|&e&U}CyPE1G0
zV=43E9(8Tl;Y#MKx!o|z`PJa5{KQRAIhD%=rE;R;MY7^!nSXB*rEB0ZHmKy&5|sQG
zEV2kZ<ySG+JLQUTqvQwb!4t+Re@|DERG#SgGJ^cj(@~dEPOR(5`&dr}+tX+(JO?TL
zSxIiGm#4kNB!lOD<Hq}^<}hD&m@>dG(w7itNuDpl;F<{hU96|}JS8lL>p0y2ekjIK
z#=5_Yb!}q#ik}tH&iqy>&lh1Z90gU{xm?-pL3Yb!UTjpv6P%A{f~S11{z=&(kLAmy
ze3)WBeS+o10;T^W<_po0QMhSL1X%y&k{_t&K9(1^Dz|kk|9AxXUY2jN)YG^W6>i)F
z)luKeK?T55x{A+L#B;1ao#oSBSG*iX;xvc(>L?|6GV3`<@_Z2nWrqC!*WQ<aS$0+B
zo(zx>64D9qATf|i2uT1_+;eLNf$HHVx#{UrRf(OJaQfE0r>buHPJM^!u7IS%dy0xc
zLzqNgC?rfW*kFhfgx4zg;5{@A`tbx28z)c%3M8mF<o#<p_l!LyYVw}<`?^n^|E#^w
z9@buat+m(QNBtF>_1UWxZ+}AoR?G3{FB*Q2v@@#1zh6~+=!bGIPvM7DeEn5|+x-4{
z;1KnR-?x}P0XwH%Bki;LK6&<OB{@&=wU}JGRPiIfErd{G`1c&(%qQnbzkgowX{FzD
zg|wSI#aR!GD^cijm(s7lOhD5IK4AHjqy^qTB+q|K@uP2)`?KZZ+^_l5_awu=zqkC?
z1U*L|o(%y@{8{=V0ro3D{D|Ufe<irBt6r_R?I-$*=JP_})DF|HmYUjpcgX0K@HdtI
zbBa5VPxLEm1AU%x!NULlAOk=9xm5n-?{db$=dYG_v*vh#;<oSQ3T^0{fm8YSoGt~q
zPqX|vr9Yzc(H|=QOP0U-VJ=bp9>v#neldCV3C5G<{F2fieu9isJHzi=ikGlilzztl
zI0#1iOV^RD?&szNKOB<-&rp2z@8#Wh%f*=lPW4@<5A5J)^U)Js$!GmjLTG$)1>=H+
z|DRI&!`Dd()h+Ja1f238%1Qn=DPO%p@!V?#|CVO<yBYa^SMzWGj1c~fHt4@*(9h+i
z{L^m}dMm#VocZ%L(f7Vz@uMoYZ61GMgz`!LhDOuzQO04voGI-<-+`mwd<K1A>4*MO
zD)c5HcYa3kb+v<}v^cZ?_ywn5l!#}1Pe%TiY>{@beO|^tAJ_cXRNnsKljOyJkde=K
zlz#o=Lbs;<_|JylE%>ck{>O|-J3HT$H=3S&h2rh)QX`9(c)_@oGuM&~&eME$D!z7)
z0Q(fbS@Au(4-V=U|LTe#`k>tJRlKjb^Q%&x$1DCy#oIbA48Ke9Ttf&Af4|~;-YK}U
zp>vPX-z@d|jOPC@jO$J4@}m<{-_oxLX!^z^;~_%*v!wKgVtQIx@irZAg5Qg@U-l~Q
z#OPn9_|X{t2FCSeq09S#^Z1>Xdc92Th>Iq*za9``lk=mDhceJVXO!MKOA2WH;wygW
zun^ifJplZIWH!!6fKz!&FP8F{yjq>2@+5z28Tdyt@Xu!8Ujcp@#@#NBH?w*6A2R5#
zc%HP+>XjnTjsE$HmsBnp{}dEI`XS*@!*5|cS+Ac``qeK9rFOgXR>g;YUh<K*BmO~g
z=g%as^EJyOicfz`fU4qOW?XMVmuEjey`AS6N58yXh^;UD4EmqW!2dgN>fcpY`WMF(
z@bBI}-g|m79{79uUgtGZU!BIBvwtkToEHP9{GB&S{^nP_U-9+l3jPeO^EWc`nS6od
zbM)th{>PR6^}t!r{|5nVTzrV>l_+#MqV%?(m(PjD^S@R6h|c>PG|fLLKJ>HlUh@lG
z{X%(fTj_1=j48fB1@lvMKH8%AYD`bK2{@1UzmxX7OWVJv^oL^fKdbnjn4bL0iXVyb
z^T#y*Lot8Xy3((oCGVwgozm|yFOquMzUh+6pGy>P?~np+(Fr35oXS5<0ZaHnUF6>m
zr62mR+~2Oar}XU@{fg4B>HKTs>y3tgR5Gyn=q}A?UF!w0&%bl0)A{gH#-W$KUFdAQ
zUk#kwe~a|5eJA)g6yI>U;8#jA=Rx2$))eB)1sl`xZHz-dQTxTx-eUL_A~AMrW-Y@v
zpT>xW!#Z~Wr*a;AmXzmGrGLgI=`Tn97$)b(fD^s*9qE@Fv>&%A{XvzJe4a4!swzID
z^6kZ%|9;J<eVx4bwNfnSdz<CG8&nT}iPE35Mewy>k@w!H_{ECbK6pOE8_$LrSEA76
zh77!#fxiMc)pz{{DX001-ciu@8ItmRSj+!$;6&d(Tj(tw@9&hp{dy_Phm`(|YbBqf
zy6zA0$iJs5-qrxy%as0V#+4{^d9Bg61^m>Ra(!zC{YRAk;GYPciWTQDa4P@m6CTSC
zo}=%*=sGFq>T86)Cl}``#n)dczz_Akql_z2=yI*mHwAoxKD$xzq07%=gyN*>x2X8q
zy>f5kbid&rj^cM%KH5<xXK?&a>iBxM(jWYU<g-)jdynDY7T{^}*!i^LhyO)@FDd?2
z#i!}R!}vW<^Z7P#YR`jDllHt)=^uB!wC6enOycKCGUu6!=l)p;6?VoIKNM@Xg5rnO
zo|W7j=gAoOG=FWlS@MCI$-g<!Q+aZ}+%GEr3!2X!?cXOV{yxQz-YtOXb@wSgt?gsw
zykBuUzXB&R^Y70Y*Gxl~zt6zWDN6bG{BI$Ci?-**jE6GNKcB1g(+vSGQGBD}dLGAB
ziqB=_bDPnhAus%%(%-505zYTeivP0WYabE7`11q6sh!t!J~BVsH<W%z=bhU%pC4NO
zs_$8Uoqt0*-(JQz;$XG?FVK8k#Si|i0C#A+y-0EAO@c!m;@^ui@_((-tK3#~!+EFT
zhxEM{X+HNTepvbQGQ~fo`076j@Jot+jd8sRT^>;Sw#t(geRld*;fEo$D@|@ZNpU-8
zg}-BfG|y4oiLJj(1E+pzUoDJzp|;OUm3~94zCESibEVMR{P?(SQl6u#SD0RRjpEMl
zNx^O3!3z{$J0!UA$+e0ftO;)8Xa+cyb6x$_rpLZP>Dy|5ee7aM_in{Es9j9^k?8kZ
zis$Z=`(KrdbHC!#$`5BN{wIv<P3W?&^n2bVp!LgX+oc`soE-kf3`m}=_>kKB#^+ZW
z{mUgEtM8=ZPON_mif>SUu<^UgaP41{d$%e+9n1gaM*nB>-ml5Uc{Olu=Le-E*K51|
zveK_Ngz$W&|DfV)9}>XY;gi70pJV&2l-&7WOmBCC>s33_$Mr78vA(J20r59AkgTQn
z^k*aw+y8Pqa4P@wh_t|8Yx%#V_|YFoy(;?ZOLj_sIe#vIvXOI@;)lL0_<NN8d5r5#
z=&~aNF9WCYtm=6Ax~92P^B?+<0LzNsrSwPCzf{$E|JRg${k=l?Dy9Df<9ZXi+^_W0
zXG#O;@`m$u#rJ$i6q9#onjdE5bABn^Uoo>Ky}zyjJ=M2;m9)Q#Ip+Xy)5$}{bUv%}
zM_w=VU(|=*i=|)I{#AhU6>kHl{Ewa^^!IA}Tsk9o?r)@j7nQyWoaX(ZI|Z<PRXxMo
zf}1|`Q;N6Mk948tL+1_H-wN^NeM-Mh8*uP5xp9V$3p*!Ep5ogD&z17nxl@MUhkU4B
zNB&w6n;+@ig0%Nu2!b?x0yyQfru=g`m&$Q|g6Yu?G5=USBcESV`aORv8T_W^^Xr<w
z^9$1cHeY;vF1<YW1Gl+Yv*62_H%dDkykF2K>NvfC@i3+Qy=3$+5yI2X5qvp={%11q
zuWCNIn4b1s#n*pZ@_o7%fldieYv&z|gAeZzLTl%82L1b$euJ(rKU?$vWCr~gl>U(V
z87@-#Zvv<O-JlzEzN+}+;iw@#U)8wu^Az6#ocUps^jBBg`RrXnzu`RLH+hP4XB1!m
znq*+(?pcg$rlHFVlz#2k1PtbJ;Pl@1FG|kO)%UhD^7*ft&-521pTE?1ya9L`|GYaR
zpYLft?XSuUAJB$)?9Ec&w%X%1?w+Ff>WI*<>U*y-{B)tDeY5o24xH+BNax4Az|K!A
z{o0!axJB_dE54^Gz@*~uwS3giK;MR@-}$gpDDEX5&u|HF_P?pZXgv8$#oKR@^eb|4
z#uYF9jsR+gIxk{ei9(m2(jTS`X!!BBNpbx?1mgz8OFvcj5%5F&*YKr0cD|pr=fyuk
zc=Fc(PVIJBD{69M-%BK)L*J72w|2Xo@nk-4RQl4p<(bv@-5K=vDgElph5k*N<sW3w
z|Bcc+?+`+3x4+AvKMMt>dgY!k?e-d}hSOGjUB~I$&yxJ^1y23wd{gde|2h4BmvQ+!
z{QqBM;16m(cHZSx%AZeuspP*$&pk9bb2V@(=bFxs5aayYrg&~l%1_^2rr%3Fp+B-y
za9hv1bx(T#-VU7dpMG7G{=9i9&(OV;E50El7w1yNHz<EztoU;n*PGDgCZ#{}H37};
zSW^7Z?+gAFP19F=ZAmJ0H}pUHy(%OBcVyuA0l!?5M*jaY)5DLU=TGu?+CcGL#fQEv
z<Ys?8778cz`*dvneG21|toi2IO26SR<e7>^XAC%xql41U=WBoYRms2fr$Vn6XB9Zn
zADt3<^Vi*?<9GGDLVxb*a^0)Ao!`mdmc*mm72or6p|kORCvcJvaXtU-O7E!u++_GY
zj7uKj|Gy5L%6aGw(Q>XYr1RU8q2Q&-nX5AJRmL$NeMLs}^X1~)sW^SVn0_jzop&pK
z^e3e}e|(la|D56n)h>RFz|NNyKl~HYA)iwGdx{VJh5*L@XD(?!eqVC9P}_eMIPuAm
zR|^jFpMURAd~H+;g687ihc%z|H%mDmula0&{Y?3+s=)i8(%-1~^m_y_`MIF@9!JXm
zS4w|7<9ZXikbO@1AC0Y>+^zU}4F8zo(;De9s^$DbM*e@9f&YW%Go<>4wdXmvYJLBY
zl>a-@9M01fxAR~5TZpK~nBoV&A(b-wXp`b=+Q6p&%qc$I68fFm7Inqf|3U!sZ>%t`
zM4`(MET8uZsNLY4xmU`6NEZg_Jbd~+Me#M|x6dg4Y{QjdRn2kkoS=M?zbVku_!@ef
zG>plC|58t<e;(t|GuLIHSp3q9GwA<k2L4QptF-bwk8#YKmrH-Cnc);O=zXQP^Md(X
zv8d*X;%k2`b>1V{IX|uV(RT@sX6N6*jQl?goZ4qq<K91bhR~me_9c9Jo3xwFkC%cE
zF4sK6myN(_T$J7^=sEguhVd|^{9RG{^*74>*-Fw^d`;zss+G=vQ~cl;q@w2Ec#Gmk
zJ}Vi#R@>n{iqkio>35A>oR2Gh^i$D%e&6!Pm3}&moIh9G&T;1N7vuSZimxgCxtt1Z
zfcH^5OzXh9N%6~oQ+pnHoaB72mh){)uSB8CCo}LbYW`~)m-SRl^L5RCkFH-IQ2aSC
zOw#!9TE;OhRB^EJxI2UX14_THcG=@K|K9*k^*XHU8hSi|^P?TD?{`I>e@^LN0i4=r
zgSOif6kk-nwezqg70$^~`f12N`fZep^ETi^`gizpuhO?aE~rl1&Sw>0(|oLdzoK|f
z{q1*YX5UeKNbO~vHgTYXmR~czTInxfJX{j_`&q!LeM&EsH~zl%`-0-rrwRTf#aA`|
zlFnb3X_l{5`aR=9Z{zXpJ;}cv^Pl_)aBAnn8i$7Oi}LS#jBBQ$%b9(lFFi&IqGH9l
zjPVem{+U<$!>-VORPoK0&z<tZPijN&%E;#fO20wRp|yUeZy!_n*A^uI-_iU(ZS?zu
z@)lK~&sf&_UM9FMzd27=-1(|xV0=ERc<!j+mumhmWL$4Tmp5hLAI-o&qxo-mhLpc3
z7w1nEU(<19a^u^IA9_GuYVC0PeyP`yF9=>$`FuWb<}2MdWcmPoQ<(exu+U#97w27!
z>rLqLea+|KgmjeoE1nJkLiv<paUc_h|GHG@{W>q-ptv(C_}%BrOI`w;%6W81@*mcA
zxJ~i)zer6dG@lEhn9+L=>bn1omi1|hAH7KEZ&Lak<Dm-apC3^Ap+f>#`+P$2w#IKi
zpac8&6yNhIDW~a0SKX%V|0T)i^P0a4oa!}nhR|=)4r?>L5``|W%fQz(|3e><3_hXl
za}V$fLh0?#F94_aPHWr<ePWA#UwxVOyB1J~vGYyFg(mVp<kwIHkekn6ul=rev{uVG
zAN1TWpBB2;Y1+#apMIqPKdSg>M*cgLevJ-*!tb#1b6N3KUB5PcU>P{K^FK*Rc4)u<
zlhSWc`Mg2jd+tw4IS+kK8U~-c;@_pfDWAg{$NqYyuNwU`&tl}mioaZO`_99;kCO*)
zRebgPa(}ndzn^iv30?jNaC+~Nf0Xi@{`p1VY5jZtE2R8K)LyoE=W@jl>HKwzzGK?(
zj|l)a=HCo(dT(w<_|y1gF9rg)r}`^Q?mYn9WQAtIm$O!-JZ+tDidvwH84pv+-_Hl0
zR<D;bJ=*^cp?iv4oPCNPy-$EwEB-3QkLdcskm7Fwo|gaDlzvZazW$El(=j=A2FgMF
zy!r#F(8siYpQ?B}M*l3vwGiRU;$nZ{ntHw5C|}d9HQIH5xYBMrIDXdi4^;ei&+C=v
z>wYeQxQ=(-ww;?cZu7iuf8Hy4l#$07d7+HF-Lt)Ilb%;=)ta?lBA1Z_GL}Hb6No#S
zfbwn<$|s@Z`|>GyO(YXe#;aRtH5#oZH&&(7>-PE!3&Rzs>UaFbTDRwSyk5hr)LTuz
z>v&$Z<t^4*^X0l%?X@~xuiQW2AOo~^&#w+o=El*QX++)vTFonWI^`A5Z}vJX&O)c$
z@V#ok(O5x}(49x^8zEg>Y}8RC?mur~B=0S_<IbYj?Uj4JbKP~@w|R3nmkQo&ad$y(
zH*MRzHFV2QcWm6QPj{BQ$j$7{+c)jpW*K_(C}qv>IP_L7%W$oW;&-||uTrkpwHS+j
zz3=Vo*D8CtG*Y&14ImXxZv6x8&>g3shg@P5;`)uVUU5fpE&^Oz+_7;RRfJ#P^!Ix`
zzu9edJYs19%hhW1&JEvIDxo`K=t06p9~BC+sn(W5cU{!Y58wN}&|Rgy61wZO+F{a0
zx$Vu@n&pnrP!CB#=|QF6?X?;soI34Exa<S^eCTZhRC1BLmup>cNxR$uH+2PC?3CL}
z9&uy8=X<@Sj$f|I6T=o-o&Dub)!ElCH+!|)e6Q2m@9i7)a$d(@0K+#ce&U&%`E2YM
z>BhY5bd!;E*IjqxwITlHZeR8*7(~56dj$a4r|Kt)uT7$<xI+8zmK0JqLF#%SbrYnh
zT#S^eo0==?njp={l?oN9A!R7JMk_@{N+6-0;7nrO3SrWg+~W^0Vlp3S*%L`BmGmSS
zQt1ycy3!wDprt?P?o%lfe>ooX(v#F%i|G%#{YK^kVsjlM@i#U6ax*<$xwF_PA4q@J
z@!NhmqlEGdQna!+JuwT(?8nVkC!;2vM%O$3(__4u<X<DJ+^s&uNvD;;C@dZ`>f>j$
z#3Dp!cFWt-W0bdAD`)gJJwa8vdoe*Rq!(*Ia+}|2rPFj<3q53$(aZMq7$u~cAU$iI
zBeHwBT(8a7(o^Cq0F8`sTxm6zJvXC^_+dWdVY}6;XQt$bSt%Fl<?fPK?)5Upi9BD(
z=!HS*W3=M_T9an73^t;t{bu&_KFz%8O-WPdaURRWou0hXX?44-uC*}#*Ku>AbjPSH
zgtXTs)errZQR4yYiYj4xJ?p66?%l3i^((Ea@AZ2NlclYd!g#&a+6&<ySN5kIZ}+zR
zgy`~64;qbf$17Kt%b4g%eXQ)A**e=B$LtDawg_!3??!3;y5E3;F<UOq=Db3Iu6bOM
z$}5((0wZLS5CF3q`F1026D4s|><*~PuqoVe*c8x+TTt?PXyjJASaQdu7`@hBzv)%W
zy>flWl;<rr`(EY1ficf(wY%<^S0wG6B)V6_U&Y<|oLVgyKv3pr{9<qyMy~L3g+gI@
z&YPX^=5pSg>&cr(J!ZQaw0U}Y+IvsrweDo(wY^r`+Z%nUy!YX*?(yHNm5sB73mYiC
z<wMnVCp?~Q&^Xk(W#W)hW56v9bbSNOTBz>;ADn2P^j6w6XxicC*<^i^C?>P-NQ@;r
z{;gnFm?F(;4V>Y39|8-<xtO?_*q8Y`KN1}a!5fL{T4HhxuF>O_&V$bVFb_2ve8()@
zmb4L=pGpkMk)*{uLaedP9p7Hq?#&cl?9FbT**LfP`dkfgDqfh_3opTKWE}6l?xqwH
zcR5UQTmsG`Z?7-hv~gQJefGm`<7jLsIGe^GC0rMh@wQ~0lg%)i<2pz4O>G8uIg#J!
zcvi)8wL!(>tc}gg2`E{xF_Mtm@?*k>Qa3Y)yG!M^uN4iCfLp<xw=!?e-LwOvRm?!w
z^=8Riz-DD}$E-K$6`RYDu8{h@v0fkIvUmetH6^biRl<ifHq*Na0njYhXSTXJm(*M3
zsy>}@U3M7F!*AkuYBe6(Q*57Lu=}N@(gKT#hgzlvCcHX44r;k#%Fr^QmJn6k<8FSw
z+^tnG-a57BVzGkR#jPx1&g*r`wO+Sap&6_KnT1ha+G-PHEN5sM<wn;e6o}CYgQkfQ
zKg42;AGa7gw$k)fsTY#0H<omZjU`9)Skmh@mSnZ@<orCIbg_*mvmZ}#@<cMPiKJ6+
zVuXj7a-6?R^NjVWtD=`wB>zCKw6#{+x=^d(|4L%qOeB4L6Etx^keHBcgByBYcLm<O
z1{n;kg4@-hLw$Q9Wbx5Dm28QrWJ6CS%jV{iP2uK}P3`8AZ*X&I`6WBkP1Uwi-V67u
zxBHPLCAQ?i>5sHRjeFa@PQUrkO7`$dj*}2z54B1E!wNoV_W1B~Q&tJwe99=wCs!fd
zd~%?=`J@<gQ#2-bBq`h6k(B(PM<vQUnlb`MlQP^*Q4ZYElu-`#An}eAY447uNR6?S
z@}*1*F6ISZ_;IJ&HNWETuXWjQ#RS{0^pdZ~@<*ab<0(~4nOoeH$;BNX!|Uguusq@v
znb}8RnNGc({=<eiDOYo+-j@G?<=!j0Xcr4}jq+aK+qd@soc^SqikIjbtd+GYc*<AO
zy_?WrvDA0%3|8X2!j3H-+z{S%J8tv}*DHAamKn#}ax+%viktC_zX!^TZqT(m4J<KX
zy^p*lq4mLP6o69)el%Oq-FfY`vxPZtZsVqH1y6iek=3MTyIzJTh`dY-J-=R`p1y8d
zanoimKb#+?k=HCY6xy}R8yOyTB90CeuiLFHV670#m%i^U_F<6I5)v!_gBM}F5-*UX
zY&b}-Ep%E9tSi>zOHO0n*3d$S=lPvZvjxXcrQE}3QR79#Fm!YMW^G^J_u9Qq(VQak
z4tky60m*39UnuwMJr70%mTJYFHP-RFey>o1|3ZtW3MZ(9T5JKd#11#Nh$VEn-<*dg
zz?BSE9oWl=p5X@7XS9Nd#aj4lyS*yh>(Qcf`oa1py<^~93FIC58p!juC3xj>4Qrw!
z>gH|EufPK$X(zpzc^EprU!3qguKsLqv|X-sisYzk6sV!RSyaE$bGyCb?1%?wwC;Di
z{0e##pwOGpI%H%b_t{>(TWoA=jN}X38sqs<{68`^LI1n>FF!UuF*-Ri?vCNcC1f-|
zH99puk)I-j=<?$u<5T&`@zETS5IUO6=O;#T6JwO6JIVjK`O%S){Mbk##c+d%vFDD6
zdDp`9<Ch!7Id=rkyiwF++#?$}KhnUrbNyx$tDl&uFlOMqU92_z;+(e&zOceBxc=}T
zT$sD&P}f~^NV{upjQ$&^|0d`^xHEUnP0@d7FZ!R;xa9X#E)ln6uSIqyS|>a|jf<UK
zhzy(Iye<f%NbprElgNPJyl5$Qf33G9-KOk=wRCR!{FHSpl6J`~hwBR?1VIQi+U84W
zn4s@j$iVyD_0i{W%0|i|8Pmc`1BMWJZTJNfK<d;=Da|hSa;WXmtkU-6wJX4}IRm!M
zu)k+Vz1cBxw-(~<p^J{8RV8L?@i#~@Nu+S`(Eiuc5(Kk)Y3p(WGHLU!T^lzQJ$HD7
z<rIpLcdNB-rGv5U;SbpV^L-K*#ahA5@u>C48?iZGJf4elVK538Q2PaPaTXv%;WChd
z^=fe5a}10)Iq0^bD+AFbBMfP9B{+vV1a1iuc5pvTAf5v2MldH*dnbsiFh=3mRN8%S
zsf>V=Wm~OMV7Y4mttWUsdCQ}6TG+KwgNq$LAgKhcqq4+tN2P4F1?|`rCMaxiQ_C8Z
zsRKdYC`-<PhX%{=Te$<Lg2^jZVQHTX=1D1sFyjPy{1Nw1fHxr0fWt5f;%#N~PI$;h
zhZ<s?Xhg98!2&nXK{ON87W=Kf_AHjr5vmcE@4;$&cd@B$nPX1n;1|-xPPS$wk5yog
zk-&yThA=`#2)nBKy>hJ{mf)bCq)>>u*J~9q&L}lFWQqUY0d|E{*xrJqz(>V>;+-Y<
z#E{!YFQ4<a=mf_PtNhd+l{&=H9iLKo?N+^3S>Z=m5V#(Yg!ZUIZQDG={XB#^uJ}!n
zoy-XfjhU^D|7;a*mAs2)QPHiM3aNsHMx(^tBwWLin>t+DQXB7NS!f)SHspGtksKNr
zZSFX`YDo@3p(51?stktFOg<k{&qNA$u`u3_OnCDx)ih4FP{bDHHY8q^UKUj;Y0!GR
zQLsR}A|<Qq*Gi?zOjzlH64mLo`Y|1q>to}09*f+^5`ErsJ`q$C)-_ppM5Cy9R<868
z2rEz4T3w!8Nga#G9x5Fa)mBGQRiQb@g*LQ6XbTH%L@#;8olqO9w5Hv~mQf}Mt#Ct|
zm<vmeNs-Ppp*0t?CA3J!<Sgs$9?_*#(`-e9g~ATyd(h_a?1%JQDOEU-g(P*FX6@#%
zm<NA5EdUj;ItWFYR~Eq`Sa?8i**w+>*Z_((7tcKdu{ItOElgfSjj&7sc}Frw$A(GA
za08vF$G=3&6bNV(<~D&<!0I!2lPoo}(26%emCR!`d(xZT1kI%Y`(TT=rM+`b=YQfh
z4r}2$xFc{3vP!_+zuC>6mMEbb1f29_hmnretx(e=yqB2w=3$sgA9Bq)9?T44;BX%g
z3<{V)ULFyZ;kj+TU#nMX>5{?)5kK69%1H+4fwDJ`u-GPy(q3gLtcU5iW37!F8VpCV
zDqed@AF;9Jp{a<IrFa7yabytdM>Y`|jPg3f%_3IjHa|M9FcRv1%ogOgMxwS<Lqg1{
z#<EdWVUAH#=(_DPI(~l87Bcu-So~U|;K@_!ri5$=`iDk5sovTkVxpS9BrQTb!WtaE
zT37GXs7T4!><S8V4fmh%D8}C)3=REZK|UmkL?l5pftjVI?noc;Wfb{b_0UwOs=|=s
zpbAM$L4iU8Jtbsf4kBGBNGqsAqg;bZ#q1W0O$lOW55PeCcc90{W`wq^czU-YPCd#r
zGR0vcs#7perz}D+#RD-O^+06EzI?8Vpm0(id*EmlEM0g8<UP~{#_WQ0%gYE#fxV4N
zh~q<87QvbAR=0M5SOdJi7%O%>-gOdcbujHQuW&Y@u_UZy+WG!+myErJPBsHNSB4q+
z9Ui%o7=~Sn=h^EGmaS)IVXPNsw|d1JNLt6mHHU8N*nrq*;y#TnNb1-e88XF5ROtG<
zS?V2k6!H5AqpBk6n}VteHHwZvD~87pnz>kmEh#E_EuyUB4kN*4vl^M0wF7W)jLGZ?
z^;~C9$hbv#1$btOTRYOpB+*z{()&+#LSelO?6}yZh?mifS%t&xwE-W7YVNegdN_@x
z*z8Wti#VPLV~2cSk<Lh3PMfnVCpbp4EGK@HZaE>l<6BP0v=NX)L=V~+u@P!`xE*se
zOtMOkLb7INYcq4P&YZ$|K@BiV8tWq(k;3ILb;F6;tbn!QUux5IO<{si+spM{4PM6b
zJc8uXGO)-(p8Q4DA?^+)>vF9kyJosAn3&Bfe7%i+6GI~{m&7VPjx^yt3?AG%(9WW1
zMCOG9nkkxGW=)M}RQJsU)ugiwn@Ao9vJUJ|J2&7fJJp#^jdetEM72xeKj@C4zy>bY
zDBHo|Xwm`hOQub?*Rz?9o}}~K39uZSnJjiDRk4NF!;}%7D>~TZBXY(bc>6|p`hg~0
zp6}{Z*Yy0xd}w;H%`hS~2AUmS(CnL^h;{|Q@Eb$SIhX`twr~cv{q8Z`iJ2au0y2UJ
z5aFC+bkv&g=G0(#bQb#*zM3vNHHe>B2Ai7-Gc{!}UFry?Sn>>E;*dF4_1m!1Xi}c=
zd@oy-S_=ypYR+=E$vnLf<y<z9T-aef4m|BQj_Z=F4qTE|uvV33In*~t%AGkDf*4KK
zuqg18JeHUmrT=KwkE~IIRP4n*Dfv3OL%J0Yv4|xgdFfWEU`m*~2~*!#V5q_~k65rY
zXL8g&jg3YqqyzWTzl{ypIG)*e!pcT*plobX+fdY1X-kY7A7Uk9A0hk*`bCTvXhU7g
zTs7)5z+S`3-YCm#`SD?F&>BVr^nn8q$enJBjH2F(w=B!^1C2q0oLFOsGW4*k-chv)
zF&MG=rQ7f5dL219BjZ`rV>VK_t6HQkh#O00nO1U=5#djK31<(@D1JyA-n$I5v9g54
z84l~Nv|u<3NAYF}*(n#;0L*>7-GzmEpxd03M+D}0)bTKk;13>v2qza|0Sf}I`D3V9
zW!h0_%+59m)k`<k#_f?N=HW8dQe%D~(VIpg4q6@RH%zpBzi66em-Y|RBsv)9@TAZr
zLQ8)XUJXABJH06iyHFTux7vZ7QK7vF<k-i;2f6+$t!|9T<mE0Hlt>$}Zj0ix%|VK|
zlc*cE{+Rs0F5$%jrbwZMmg-LFk`D%TK$SYWGey#|RMRo4Cxv9<CsGR=S+|fPoK(Eo
zPJam8kXXhL=Oc<|TOf{A(#%*=^b=XNQ|&&jQi1miJ4uy~S!KcM0ZU#}Z5r6Bc>t0*
zIp%{Im)vNymJ(VN59TIp)%9rcA+KIO4h=Y-Z%-$$y}s_6p*a~Gj!oTMWV0Kaev8*N
z5X*}c*gDcC2bA2A(cG|ofaYc)pTe@P0$C<lt{TGv*mAR33xo^Ysd3rT8$tHLx;4f@
zbY0p^C5!}$C1yS_o0@f3)j*{Jwpnyv9*u?#mSiqNUjYra$6~{EoA#EnNDKO1RmH%2
zE-9M+;$Y;H&xl(>0-Xn>-foRyZMND`Jy73EN)&T#uu2=$G%y0WW*BTCD<yM<9%>@w
z73LS1y0AHe)nQsx;6$YBN6cKrd2$ob-c2q^_DcD`<0^lFqnf!q?zsASuvtQvmLL{V
zqRk71iK-86thFM-ad~Fe^-6Hfb`jSboirt54H+PY&C*H|el=|@nARrux%G77WIv{M
zEwdE1l*2HxF1D4re|vQ%%4`GOhWMlEYlBMI5j#b60CK1`T)JB3SgbXVt#EZY9~nHF
zflco$GNhIN)0Pjdf{Hy3xrEp%UQHOsY69jta#KNZ$K8%unDnFG4h-1AlnQlykjXqv
z%@Y(553yAz&lINkX0JHUW}B==hy>(_s)$!naOq~U%i<11xHsWk?Lrf)dXUrH!GobQ
zVUAZ!OtqM~Z?jg{ucr>S|73$f4<;7FBNLXWz>82Mt2q9HoOh;z*vdq`yn@Iz)dAr_
zqA@K&X}XlEyTr-vga?q;HRKHg9e`ciJeFHnU4`6&HLNca&kFWXZYSN5w)(KEg}5#p
zO$QH#MKs|_B8ZM6drH?XFGmf#Al!)s1e-JBC2xd$IPQc>S=)lmjc`(G0Mro_YMkBz
zOB!oL9j^{um9;Tm!^3~$s)v!Ebg&Pc?ywT)(7N4#^C{+$TZ<xvJj6blWkfa__RZ2o
z<$iGZiB<w}UHXJ)1UCE8($#t?YJLb&Xsf;IVx-53j{%Ab+3I2p1v;a4E%i08SWz#g
z%#Nxc=kr7ewGNAKD+oMd0fO$=pv6%BQXZ}^hl2?R33E2<!YdEKt38-ZXxDQ%x{D8A
zQKKr*X32i>)OivYT8IjD7QjnnRmXx<_eUcNq)w<V1us&hq~{$V`zV@@JDU5+Yt7S+
zQb`tQ-K0EP{Ua;92_r%h4jL7wv;-AHZ#H!gwJv!=iZ4;V2n;(ZX3v*<aO{pMJCkEB
zc#U)#b-9><+>`T_M$F3^G&4E0`sB4j>7u*$DG%`ywc02oMpvM+Li)fB%igpIOiCyF
zP_=m@fx=`o;83YH=^7mD5Mp;)wHS;xDzhFTFlb4ff+H=-CKE}sW-Fx^Y3YVWGv}tQ
zEzymP1m*xdmmzb&ER@A|pV!MHLU;*+cd1YqDEZRH4KmY072-63Kl$I+gO#r3ZL&z_
z=;CrPcS!h$G?fjJQFYzI+>r;Bl?uXwmHWff!N3$un+?cW5e}*5)))UFsf>K`(12H2
zj)m<&CXb$R!eWf%@Iu%3V?Hq^htJ=5B%$We0A`}_SdmmD9a~|6jxGBGc~AT#vMPwP
zHfHH|{SnEOix$aaaN-9E=Wu*Mc@`c49l1>o9pPfYfgTjz^X9g1#wq}B3edgzEzAgT
zrh<l|ep6j8p}<_FMyN9f=F-km1d7<=xD3l-+89ZqXRJapAAT2@kJ}LsS;a96G>^u;
zW#%&s#t``-1}3#}ki=9FPdrQ>u2J%((4r&<6hS+RXeDWpNLdUyn1Y}p*pVxPumd_i
zW?`X@w?zfZ0tM(sA!)gU5|oUoBfFOO6mT!g`6D5g;E_2f7N0;w9>wU}IBm)SBr<?>
z$&QYiXk+_H6{ms4(j$-zZuzND)CA9(DxRr0dTNCT3W)}EM~SVK_YwV5m`PDG+&q^}
ztw50yfp_)<!?7me!4y|nR6hzFawjQN!67GGgn_-B9s-8F02E6w84mplO-)j7Q`*gs
zb1LPCnen9ZlcmTiKgK_4ohAZqn4z5x6cmi(OpX`Gj$Al^O^ikKIEF|lzz20H)@rr{
zdCUNX8eTVGSW=$V;9y3;LIRc~$dO6&h|ch&W6%u00N-+KePb{Mkn0`0rMPH8OP=r{
zG+ZJ|TN_0)A+)p-6?VuMHOA>O;^(JG#7eo_i!z+NJbUwiRy$OyJ_T{oTgPZlT7(HQ
zx`6PVBDN`bjd{_)dIcTK!CE3zt%%SW2!8IDu;EO7A`8oi3?Xj7&J*<AU_^v`gn0WP
z#ESaR!cBHSRm_P09?f1*=0~N}_-|yF8pS}d-$DHMF{2oeLnMkJ&+d)_FUwcUaYKpZ
z7XlV(Rqa4|zP9Yr8dtYQd*ZNI%oYpp<d3YeF+cK75(K)TlQ7ertXK$k7c>yTo*5HE
z_B30!VBvh&6$0-Gb!hO`!Lle~#4;R^=wmE^kcHn}X;#3;w8JGNttEcYRA&khBQ;#q
z5E_gKvNVf>SiXaEF2MM)gD=ASr(%0}NTkF<N+t7<A{4Y|z=r9u2k>fbhsgualI;VC
zR|ba<aW@StE%HYp@C|Y<It7ZLXo|%f))djiSpn-R6zm`qJ5hqSc<8(a!Dk1=#C)H}
z69$z!*25kdu5(0;A;xEBt1J$(R?mYXx<$=L#6CB6Xn|GY=)*i~ZjWG{!ens`oC(>V
z?tEffDB5BCD!A{PT?j;<c0FQQHA)g49Ve(QK!1*ha0Fw5MxpS=z}v;VB%4yD`m*#h
z&~4P6qFNLMg7n}+LR6*%JE=!5qH`YK^cl>Mp%-zJ$SzH<+s2wb@tCdq1`&mm_7_y%
zu=F^=_Mx#I>oYh#*qg%6J!u52W6K}pi!p)0<-=N$u)f9{=W{qSrwV(G_xe}*9Xc6y
zC1D$otaB%569Kxsiv0u#<!Q9-^d*$(kug-LYNs#}jCBxtE*n;o;+WZ3r!(2^6`Hjr
zyz~G0S(|k|Y9o(Ul1a}4*F;A<a~B=v8t@xT+7!nWv?Cfb6f~$pntfQnI(6(9NORrM
zX|u?z*D|Fq*tCuTWP<?*rRjNCq(_mrI#sLYCnum#;}9+0A<cV1XdEZDQ{{7{idrQa
z*t18Yu#mmKid$%KNTv=$Oocd&I_YT)t=dE_Ms=o#cN%e6KW9IfX~~Jotef3Hrd4Ca
zMELtu#8RNhV)aV~#iMIG<0wmj8EkTvRtt2WhcR+gwt2$fd-Md0F0k7?ra>en;wf<q
zhXYYK4N`}iy9fsrrs$Q3y+dSX@ko-1ESqN{+ved%={66tOSO4u&#U}{VH}EzgTXN7
z?Et%A<ZL8;mKJzrCmUm-t!!f&!i@|lmQxHqPMTO27E;+2I$yvxw3)5&@L_mnE)NGQ
zgP}zN5ycR)tjSZTmKnyh8r52mZNhOiE2KR}GzPY@;PvBN2aAgyHkEUP9QNVU_P>a6
zCJW9hHNRes&Jwn&EU{2%wlq&~;DMy$K?WYkf#_s+5(*SD)oU=-At%eR_i>@+l;!_#
zx%QZrdKR-PUquc%2xj(VQ$C7SA{x|Yz|ux`W*OS1OJUoIm4+ZBo|PCJtsx;|7z^|8
zRhSB{jHwC^FE^06J_uV64@8`&Kw;~*=}1xceV)R7AsHX~%>3w-v*9k`h6pxfj}~DE
zCZc^wEA2*w4JQlSGbd<Zp}mP8IHHHP?Wb*GSGT4fvdzvR<{F+Xb&Z*Po$eTKyXGS)
z``z-Q>~xRrzm~jsSFxQxK|<C7K*Ec7{51iS&8S?eTR%6jNKe(mgg$6WqpH-gbpR=|
ztR|%HvGfp~F<<d2U0S&oLwJzqDJcuk=_&9wU8kofC!o_~Adsg&w7*-=*&VD0!p9d#
zIquTXGC=U4-P#}4h*TlBb;F9RG0V)O@v@aVtvmOkz6hL93TJw`M`8I3=(QO4QuILh
zHRie@JSYmkB~bb+OZ}#Pi6WRoX@l!v4vl;pAiOhJ#NGlsTXB8?=Sw$&>jrO;NWq9Y
z+3<XFczB6(d-GyxLDG$Q#a7qb3C9evR!kh)F+zxR87!*?HAm1ipDhtPEniAMU#FFP
zL>;6^)X>EBX!F3wCoa>i;3qCGz?`AY2hC7mNa9d8I%Nj_G(LNYW||W@9td*x*nTS*
z&U)_haXVDzx5S|e$Id_D?bg>N*ad;nwErPGB`{x;ubbV3s=?C!mE<7@@u(y^;)s(+
zKB2*%uSwgi6VZpcrY2MCC~+O4jNIpNWYSs4wCG<r-XdR}eHU1&uwbq5O42C_Ra<<5
zXiZs}e01%q4IA({9Z+_Y!drFh9i=v#B1SqaYgMGvRLZW^Ab=_ML7##A3BLAZaT2h-
zm|MuR%HIHy1NImjBnv)0rK)0Z97RCFTS6$DY4F)D&I3!af%{!wMGHi%FOZxYd_#i{
zC4)#uU>u*6ekz=THLE$VbZxB!YZNLGR;5dj*4P2UQJB1ssDTCO+eE;sYk8E8Wk`6E
zA*R`EOd5@%RggT=v+;P;*laV3Z>ue^>`6``Su&v{+kk6hbl~M;tSHhwgOyqi8O1A4
z!GA2gN2+4deJr)YDJpibkcokma0~KWQe;)CmctP;C#2`VV?ZYu2EhPndLja9PfgD1
zP9Vf|bXOYlIE0D0(eziZI7h+D_B}-eI?>k@=jdycLTEu>X2QcAEFowVSU_j%NqZ41
zcW}}M*4@LA2G|aXr5nkKNN@@u9ZutTH#T9hRlOOGdxprzmoP_$)q=`i@edyt2!5t5
zj}~8yJ$K**oT9@UqR3G(u}DL)*TNz;SDD_gmK{x=52rAGG@b_+mcL}T8k_TXSB~+m
zA>vfxdmPW1ve`?&>*G*#Y06BV44M+YqNVx}`IPjSEDb_H`++7-JT%x(84)-MMKwdp
zkQyeZ78GVu#(&5ZBArZ10t*l9HKM1Dq?u4)MeHn<4{>OGA=_IJGFX_f4R)xJeCMem
zUwx&`w#uCvYag`~Xl+<zw29S|sU@6t*m{~}6pL8b4dxYmF0=#PE40O&G^OwyYxYcJ
z*Ex$x5dpSND)k{dOwSq!CSwgnk_yG+ola$%iz-23e{(qYh73{S*pT+bc4ahfN=tQu
zIyH6cG4*B1&IXn4VdQzjJO$CMB4&-+rp)<<vR<HAz+}=7RLZyrZ4*m=`Guzo#I^DW
zu~OK_G>i))Urpk9jP{(!nUosX8dgidg0$xw(POqDS-&$zB7y9AG<1=ID|t4<83&u-
z3D4*3q%uoH+K0{>v=7QA5IN|CnFysum=2V=f9sQPGh>KwY-n^KN|VFMMB8rN(J045
zw-P`PDSAi~H7PjMZD~pX-a{YvK@0_}xwOA$p|o5;lb6CXA2c?0=dgDZE4PWmwpxqf
zfjyyJAl@)L7mLnLK4y?=WD#gP4R>sNeaNS^Ji6B0b_Sc6c+~~b`Zz-;xk>_i6Q8jb
zZ!=DoXLW>lCCX!=kS=7znpvt|rW%(G7<|!AJCMvDZ2k7DIKmRg_(jJ4$xQ26>MTw9
zC*Zcj$<jD9Cve+^7Jd4*6J#LN=mWixX`Jka+DY3DDl9d+h+5JP&<I||h8b-;j~Qk(
zQ_02}4vO86Lm*;K2$XNIwH}g{gg<|$aBl7dy)dfV1R<x|CQXe-&`2;-AtWM>HoD1v
zYhf6Yy#d2;lZ6&xAGYxqol<$;e6*aF#Dg)~FekR{=TuY=S@VVIm%i5r=Qd$}j)kA{
zxR;qkzjZ9EKiCP75>~=uIChe0oSk5lg@lK#{RXC+IZZ@xXwW8J&0nDl6w4D2hb=_j
zAB6jn_-T53?Ss<lN*9GDHp|PGCRGsH85KBvN_x*u^5uo2WRp0<uw)V0g)4%L__}0&
z@epb|!_8LDAI35C*O0O?jE|EJ)5(|4aCN1L#B#;CQTzlain5g&zVYx38~MtVK>upj
zd(JSgjtt|^;YFNNF-%`TbB5VY7{>RR>5P|Qe+fQX`iSgO6`9x_UWK!2WXlIMAhNGH
zGaQ7~DX&N#_BZ82Z%39~(sH9#LHbsYUJQg^gXK$S7;MDx#?Hg@`w?Vw8vdQG*}Yf)
zu<IqU*AQ*^_YB;eg@2a*Q~HNphr-18`^?zAi~Aou_+ZP@e?$MU>!{LOy0c^HpO5>;
z<DaD;IztHT`i&>Z{bO*UvRnG63hZpgm6BNcvoDkfcHMq~66z&h|E<6u43+;TY7qSF
z`a(@$<+u0Pbr)#=HI)9XS8#I2u4|f}%1{5S{6gh8i}-``w)`fqlqYu0Jz4JU{g&S9
zzfaR|5;A9r8Wcaf9z0j>DLti$r+*nRD&k}DkAJ1;mG~?CvvL@=j6cHZhcvxi+x9|D
zupxH6BbNRkH7I^|wR#FVD8FU<b9fMb{~=9pSN|(P%{U8tzvXu@mj0<@G#b!WdOeTt
z2LHwD|7JX*_gnut(~{n<kL?c9(qBW-Tjws^htr?^Ym(lsb>(ME7JvT-V(AZjTGHF~
z>J>{VS8F#z?~SEjd!bO+wf;n-lxzI`p8-rmREMpz!I$(KZj|fs`tJ{s_WV%#t2O=A
znqG1Nf&T2?u1A4W``P<*nm(uL1BTc1@&5l7aN>|4eM552-6Gc^?N}PS^t1O`K0gI4
z{Qgo)C`z7O-^-~6|5<vw{vej#u>rG3D6Wq+xTUk}gR%72_cUEqu2&CU&3{=sgCB#!
z6K3^qYx=gPfA7sIKs15X+tQzd^i=<x0eZFc(=nWA=@-5RV@DqG|A(ynHwpMOTrNT)
tE5Gs8<A7V)g;M_iqx*z%#~opG@Loe{dPqxu<&RO$j`Kt-DBmU1|6jrbdKUly

literal 0
HcmV?d00001

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
new file mode 100644
index 0000000000000..ad21109ac71e6
--- /dev/null
+++ b/gguf-llama.cpp
@@ -0,0 +1,4060 @@
+// Defines fileno on msys:
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#include <cstddef>
+#include <cstdint>
+#include <cstdio>
+#endif
+
+#include "gguf-util.h"
+#include "gguf-llama.h"
+
+#include "ggml.h"
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+#elif defined(GGML_USE_CLBLAST)
+#include "ggml-opencl.h"
+#endif
+
+#ifdef GGML_USE_METAL
+#include "ggml-metal.h"
+#endif
+#ifdef GGML_USE_MPI
+#include "ggml-mpi.h"
+#endif
+#ifdef GGML_USE_K_QUANTS
+#ifndef QK_K
+#ifdef GGML_QKK_64
+#define QK_K 64
+#else
+#define QK_K 256
+#endif
+#endif
+#endif
+
+#include <array>
+#include <ctime>
+#include <cinttypes>
+#include <fstream>
+#include <random>
+#include <map>
+#include <unordered_map>
+#include <queue>
+#include <cassert>
+#include <cstring>
+#include <climits>
+#include <memory>
+#include <algorithm>
+#include <initializer_list>
+#include <thread>
+#include <atomic>
+#include <mutex>
+#include <sstream>
+#include <numeric>
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+#define LLAMA_USE_SCRATCH
+#define LLAMA_MAX_SCRATCH_BUFFERS 16
+
+// available llama models
+enum e_model {
+    MODEL_UNKNOWN,
+    MODEL_3B,
+    MODEL_7B,
+    MODEL_13B,
+    MODEL_30B,
+    MODEL_65B,
+    MODEL_70B,
+};
+
+static const size_t kB = 1024;
+static const size_t MB = 1024*1024;
+
+// computed for n_ctx == 2048
+// TODO: dynamically determine these sizes
+//       needs modifications in ggml
+
+typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+
+void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
+    (void) tensor;
+}
+
+//
+// ggml helpers
+//
+
+static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
+    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
+
+    if (plan.work_size > 0) {
+        buf.resize(plan.work_size);
+        plan.work_data = buf.data();
+    }
+
+    ggml_graph_compute(graph, &plan);
+}
+
+//
+// memory sizes (calculated for n_batch == 512)
+//
+
+static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   ((size_t) n_ctx / 16ull +  92ull) * MB },
+        { MODEL_7B,   ((size_t) n_ctx / 16ull + 100ull) * MB },
+        { MODEL_13B,  ((size_t) n_ctx / 12ull + 120ull) * MB },
+        { MODEL_30B,  ((size_t) n_ctx /  9ull + 160ull) * MB },
+        { MODEL_65B,  ((size_t) n_ctx /  6ull + 256ull) * MB }, // guess
+        { MODEL_70B,  ((size_t) n_ctx /  7ull + 164ull) * MB },
+    };
+    return k_sizes;
+}
+
+static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,  128ull * MB },
+        { MODEL_7B,  160ull * MB },
+        { MODEL_13B, 192ull * MB },
+        { MODEL_30B, 256ull * MB },
+        { MODEL_65B, 384ull * MB }, // guess
+        { MODEL_70B, 304ull * MB },
+    };
+    return k_sizes;
+}
+
+// used to store the compute graph tensors + non-scratch data
+static const std::map<e_model, size_t> & MEM_REQ_EVAL()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   8ull * MB },
+        { MODEL_7B,  10ull * MB },
+        { MODEL_13B, 12ull * MB },
+        { MODEL_30B, 16ull * MB },
+        { MODEL_65B, 24ull * MB }, // guess
+        { MODEL_70B, 24ull * MB },
+    };
+    return k_sizes;
+}
+
+// amount of VRAM needed per batch size to hold temporary results
+// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   512ull * kB },
+        { MODEL_7B,   512ull * kB },
+        { MODEL_13B,  640ull * kB },
+        { MODEL_30B,  768ull * kB },
+        { MODEL_65B, 1536ull * kB },
+        { MODEL_70B, 1536ull * kB }, // TODO (likely can be reduced)
+    };
+    return k_sizes;
+}
+
+// amount of VRAM needed per batch size and context to hold temporary results
+// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,  128ull },
+        { MODEL_7B,  128ull },
+        { MODEL_13B, 160ull },
+        { MODEL_30B, 208ull },
+        { MODEL_65B, 416ull },
+        { MODEL_70B, 416ull }, // TODO (likely can be reduced)
+    };
+    return k_sizes;
+}
+
+// default hparams (LLaMA 7B)
+struct llama_hparams {
+    uint32_t n_vocab   = 32000;
+    uint32_t n_ctx     = 512;   // this is provided as user input?
+    uint32_t n_embd    = 4096;
+    uint32_t n_mult    = 256;
+    uint32_t n_head    = 32;
+    uint32_t n_head_kv = 32;
+    uint32_t n_layer   = 32;
+    uint32_t n_rot     = 64;
+
+    // LLaMAv2
+    // TODO: load from model data hparams
+    float f_ffn_mult = 1.0f;
+    float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
+
+    float rope_freq_base  = 10000.0f;
+    float rope_freq_scale = 1.0f;
+
+    enum llama_ftype ftype = LLAMA_FTYPE_MOSTLY_F16;
+
+    bool operator!=(const llama_hparams & other) const {
+        return static_cast<bool>(memcmp(this, &other, sizeof(llama_hparams))); // NOLINT
+    }
+
+    uint32_t n_gqa() const {
+        return n_head/n_head_kv;
+    }
+
+    uint32_t n_embd_head() const {
+        return n_embd/n_head;
+    }
+
+    uint32_t n_embd_gqa() const {
+        return n_embd/n_gqa();
+    }
+
+    size_t kv_size() const {
+        size_t result = 2ull;
+        result *= (size_t) n_embd_gqa();
+        result *= (size_t) n_ctx;
+        result *= (size_t) n_layer;
+        result *= sizeof(ggml_fp16_t);
+        return result;
+    }
+};
+
+struct llama_layer {
+    // normalization
+    struct ggml_tensor * attention_norm;
+
+    // attention
+    struct ggml_tensor * wq;
+    struct ggml_tensor * wk;
+    struct ggml_tensor * wv;
+    struct ggml_tensor * wo;
+
+    // normalization
+    struct ggml_tensor * ffn_norm;
+
+    // ff
+    struct ggml_tensor * w1;
+    struct ggml_tensor * w2;
+    struct ggml_tensor * w3;
+};
+
+struct llama_kv_cache {
+    struct ggml_tensor * k = NULL;
+    struct ggml_tensor * v = NULL;
+
+    struct ggml_context * ctx = NULL;
+
+    gguf_ctx_buffer buf;
+
+    int n; // number of tokens currently in the cache
+
+    ~llama_kv_cache() {
+        if (ctx) {
+            ggml_free(ctx);
+        }
+
+#ifdef GGML_USE_CUBLAS
+        ggml_cuda_free_data(k);
+        ggml_cuda_free_data(v);
+#endif // GGML_USE_CUBLAS
+    }
+};
+
+struct llama_vocab {
+    using id    = int32_t;
+    using token = std::string;
+
+    struct token_score {
+        token tok;
+        float score;
+    };
+
+    std::unordered_map<token, id> token_to_id;
+    std::vector<token_score> id_to_token;
+};
+
+struct llama_model {
+    e_model type = MODEL_UNKNOWN;
+
+    llama_hparams hparams;
+
+    struct ggml_tensor * tok_embeddings;
+
+    struct ggml_tensor * norm;
+    struct ggml_tensor * output;
+
+    std::vector<llama_layer> layers;
+    int n_gpu_layers;
+
+    // context
+    struct ggml_context * ctx = NULL;
+
+    // the model memory buffer
+    gguf_ctx_buffer buf;
+
+    // model memory mapped file
+    std::unique_ptr<gguf_mmap> mapping;
+
+    // objects representing data potentially being locked in memory
+    gguf_mlock mlock_buf;
+    gguf_mlock mlock_mmap;
+
+    // for quantize-stats only
+    std::vector<std::pair<std::string, struct ggml_tensor *>> tensors_by_name;
+
+    int64_t t_load_us = 0;
+    int64_t t_start_us = 0;
+
+    llama_vocab vocab;
+
+    ~llama_model() {
+        if (ctx) {
+            ggml_free(ctx);
+        }
+
+#ifdef GGML_USE_CUBLAS
+        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
+            ggml_cuda_free_data(tensors_by_name[i].second);
+        }
+        ggml_cuda_free_scratch();
+#elif defined(GGML_USE_CLBLAST)
+        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
+            ggml_cl_free_data(tensors_by_name[i].second);
+        }
+#endif
+    }
+};
+
+struct llama_context {
+    llama_context(const llama_model & model) : model(model), t_load_us(model.t_load_us), t_start_us(model.t_start_us) {}
+#ifdef GGML_USE_METAL
+    ~llama_context() {
+        if (ctx_metal) {
+            ggml_metal_free(ctx_metal);
+        }
+    }
+#endif
+    std::mt19937 rng;
+
+    bool has_evaluated_once = false;
+
+    int64_t t_sample_us = 0;
+    int64_t t_eval_us   = 0;
+    int64_t t_p_eval_us = 0;
+
+    int32_t n_sample = 0; // number of tokens sampled
+    int32_t n_eval   = 0; // number of eval calls
+    int32_t n_p_eval = 0; // number of tokens in eval calls for the prompt (with batch size > 1)
+
+    const llama_model & model;
+
+    bool model_owner = false;
+
+    int64_t t_load_us;
+    int64_t t_start_us;
+
+    // key + value cache for the self attention
+    struct llama_kv_cache kv_self;
+
+    size_t mem_per_token = 0;
+
+    // decode output (2-dimensional array: [n_tokens][n_vocab])
+    std::vector<float> logits;
+    bool logits_all = false;
+
+    // input embedding (1-dimensional array: [n_embd])
+    std::vector<float> embedding;
+
+    // reusable buffer for `struct ggml_graph_plan.work_data`
+    std::vector<uint8_t> work_buffer;
+
+    // memory buffers used to evaluate the model
+    // TODO: move in llama_state
+    gguf_ctx_buffer buf_compute;
+    gguf_ctx_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+
+#ifdef GGML_USE_METAL
+    ggml_metal_context * ctx_metal = NULL;
+#endif
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_context * ctx_mpi = NULL;
+#endif
+
+    int    buf_last = 0;
+    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
+
+    void use_buf(struct ggml_context * ctx, int i) {
+#if defined(LLAMA_USE_SCRATCH)
+        size_t last_size = 0;
+
+        if (i == -1) {
+            last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
+        } else {
+            auto & buf = buf_scratch[i];
+            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.addr, });
+        }
+
+        if (buf_last >= 0) {
+            buf_max_size[buf_last] = std::max(buf_max_size[buf_last], last_size);
+        }
+
+        buf_last = i;
+#else
+        (void) i;
+        (void) ctx;
+#endif
+    }
+
+    size_t get_buf_max_mem(int i) const {
+#if defined(LLAMA_USE_SCRATCH)
+        return buf_max_size[i];
+#else
+        (void) i;
+        return 0;
+#endif
+    }
+};
+
+template <typename T>
+static T checked_mul(T a, T b) {
+    T ret = a * b;
+    if (a != 0 && ret / a != b) {
+        throw std::runtime_error(format("overflow multiplying %llu * %llu",
+                     (unsigned long long) a, (unsigned long long) b));
+    }
+    return ret;
+}
+
+static size_t checked_div(size_t a, size_t b) {
+    if (b == 0 || a % b != 0) {
+        throw std::runtime_error(format("error dividing %zu / %zu", a, b));
+    }
+    return a / b;
+}
+
+static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
+    char buf[256];
+    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
+    for (size_t i = 1; i < ne.size(); i++) {
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
+    }
+    return buf;
+}
+
+static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
+    size_t size = ggml_type_size(type);
+    for (uint32_t dim : ne) {
+        size = checked_mul<size_t>(size, dim);
+    }
+    return size / ggml_blck_size(type);
+}
+
+struct llama_load_tensor {
+    std::string name;
+    enum ggml_type type = GGML_TYPE_F32;
+    std::vector<uint32_t> ne;
+    size_t file_off;
+    size_t size;
+    struct ggml_tensor * ggml_tensor = NULL;
+    uint8_t * data;
+};
+
+struct llama_load_tensors_map {
+    // tensors is kept in a separate vector to preserve file order
+    std::vector<llama_load_tensor> tensors;
+    std::unordered_map<std::string, size_t> name_to_idx;
+};
+
+enum gguf_file_version {
+    gguf_file_VERSION_GGML,
+    gguf_file_VERSION_GGMF_V1, // added version field and scores in vocab
+    gguf_file_VERSION_GGJT_V1, // added padding
+    gguf_file_VERSION_GGJT_V2, // changed quantization format
+    gguf_file_VERSION_GGJT_V3, // changed Q4 and Q8 quantization format
+};
+
+struct gguf_file_loader {
+    gguf_file file;
+    gguf_context * gguf_ctx;
+    gguf_file_version file_version;
+    llama_hparams hparams;
+    llama_vocab vocab;
+struct ggml_context * ctx_data = NULL;
+
+    gguf_file_loader(const char * fname, llama_load_tensors_map & tensors_map)
+        : file(fname, "rb") {
+        fprintf(stderr, "llama.cpp: loading model from %s\n", fname);
+        
+    struct gguf_init_params params = {
+        /*.no_alloc = */ true,
+        /*.ctx      = */ &ctx_data,
+    };
+
+    gguf_ctx = gguf_init_from_file(fname, params);
+
+        read_tensor_metadata(tensors_map);
+    }
+
+    uint32_t read_u32(const char * key) {
+        int i = gguf_find_key(gguf_ctx, key);
+        if (i == -1) {
+            throw std::runtime_error(format("cannot find param with key %s\n", key));
+        }
+
+        return gguf_get_val_u32(gguf_ctx, i);
+    }
+
+    int read_n_vocab() {
+        int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
+    if (i == -1) {
+        throw std::runtime_error("cannot find token list in GGUF file\n");
+    }
+
+    return gguf_get_arr_n(gguf_ctx, i);
+    }
+
+    void read_hparams() {
+
+        // TODO make keysconstants in header
+        // TODO: read all hparams from file
+        hparams.n_vocab = read_n_vocab();
+        hparams.n_embd  = read_u32("llama.embedding_length");
+        //hparams.n_mult  = file.read_u32();
+        hparams.n_head  = read_u32("llama.attention.head_count");
+        hparams.n_layer = read_u32("llama.layer_count");
+        //hparams.n_rot   = file.read_u32();
+        //hparams.ftype   = (enum llama_ftype) file.read_u32();
+
+        // LLaMAv2
+        hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
+    }
+
+    void read_vocab() {
+        vocab.id_to_token.resize(hparams.n_vocab);
+        int token_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
+        if (token_idx == -1) {
+            throw std::runtime_error("cannot find token list in GGUF file\n");
+        }
+
+        int score_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.scores");
+        if (score_idx == -1) {
+            throw std::runtime_error("cannot find token scores list in GGUF file\n");
+        }
+
+        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
+
+            std::string word = gguf_get_arr_str(gguf_ctx, token_idx, i);
+            
+            vocab.token_to_id[word] = i;
+
+            auto & tok_score = vocab.id_to_token[i];
+            tok_score.tok = std::move(word);
+            tok_score.score = gguf_get_arr_f32(gguf_ctx, score_idx, i);
+        }
+    }
+
+    void read_tensor_metadata(llama_load_tensors_map & tensors_map) {
+        const int n_tensors = gguf_get_n_tensors(gguf_ctx);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            llama_load_tensor tensor;
+            const char * name = gguf_get_tensor_name(gguf_ctx, i);
+
+            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
+            uint32_t n_dims = cur->n_dims;
+            tensor.type = cur->type;
+            tensor.ne.resize(n_dims);
+            memcpy(tensor.ne.data(), &cur->ne[0], sizeof(tensor.ne[0]) * n_dims);
+            if (n_dims < 1 || n_dims > 2) {
+                throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
+            }
+            switch (tensor.type) {
+                case GGML_TYPE_F32:
+                case GGML_TYPE_F16:
+                case GGML_TYPE_Q4_0:
+                case GGML_TYPE_Q4_1:
+                case GGML_TYPE_Q5_0:
+                case GGML_TYPE_Q5_1:
+                case GGML_TYPE_Q8_0:
+                case GGML_TYPE_Q2_K:
+                case GGML_TYPE_Q3_K:
+                case GGML_TYPE_Q4_K:
+                case GGML_TYPE_Q5_K:
+                case GGML_TYPE_Q6_K:
+                    break;
+                default: {
+                    throw std::runtime_error(format("unrecognized tensor type %u\n", tensor.type));
+                }
+            }
+
+
+            tensor.file_off = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, i);
+        
+            tensor.name = name;
+            tensor.size = llama_calc_tensor_size(tensor.ne, tensor.type);
+            
+            tensors_map.tensors.push_back(tensor);
+            tensors_map.name_to_idx[name] = tensors_map.tensors.size() - 1;
+        }
+    }
+};
+
+struct gguf_file_saver {
+    gguf_file file;
+    gguf_file_loader * any_file_loader;
+    gguf_file_saver(const char * fname, gguf_file_loader * any_file_loader, enum llama_ftype new_ftype)
+        : file(fname, "wb"), any_file_loader(any_file_loader) {
+        fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
+        write_magic();
+        write_hparams(new_ftype);
+        write_vocab();
+    }
+    void write_magic() {
+        }
+    void write_hparams(enum llama_ftype new_ftype) {
+        const llama_hparams & hparams = any_file_loader->hparams;
+        GGML_UNUSED(hparams);
+        GGML_UNUSED(new_ftype);
+    }
+    void write_vocab() {
+        uint32_t n_vocab = any_file_loader->hparams.n_vocab;
+        GGML_UNUSED(n_vocab);
+    }
+    void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
+        switch (new_type) {
+            case GGML_TYPE_F32:
+            case GGML_TYPE_F16:
+            case GGML_TYPE_Q4_0:
+            case GGML_TYPE_Q4_1:
+            case GGML_TYPE_Q5_0:
+            case GGML_TYPE_Q5_1:
+            case GGML_TYPE_Q8_0:
+            case GGML_TYPE_Q2_K:
+            case GGML_TYPE_Q3_K:
+            case GGML_TYPE_Q4_K:
+            case GGML_TYPE_Q5_K:
+            case GGML_TYPE_Q6_K:
+                break;
+            default: GGML_ASSERT(false);
+        }
+        
+    }
+};
+
+struct llama_model_loader {
+    std::unique_ptr<gguf_file_loader> file_loader;
+    llama_load_tensors_map tensors_map;
+    bool use_mmap;
+    size_t num_ggml_tensors_created = 0;
+    struct ggml_context * ggml_ctx = NULL;
+    std::unique_ptr<gguf_mmap> mapping;
+
+    llama_model_loader(const std::string & fname_base, bool use_mmap) {
+        file_loader = std::unique_ptr<gguf_file_loader>(new gguf_file_loader(fname_base.c_str(), tensors_map));
+        if (!gguf_mmap::SUPPORTED) {
+            use_mmap = false;
+        }
+        this->use_mmap = use_mmap;
+    }
+
+    void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
+        *ctx_size_p = *mmapped_size_p = 0;
+        for (const llama_load_tensor & lt : tensors_map.tensors) {
+            *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
+            *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
+        }
+    }
+
+    struct ggml_tensor * get_tensor(const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
+        auto it = tensors_map.name_to_idx.find(name);
+        if (it == tensors_map.name_to_idx.end()) {
+            throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
+        }
+        llama_load_tensor & lt = tensors_map.tensors.at(it->second);
+        if (lt.ne != ne) {
+            throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
+                         name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
+        }
+
+        return get_tensor_for(lt, backend);
+    }
+
+    struct ggml_tensor * get_tensor_for(llama_load_tensor & lt, ggml_backend backend) {
+        struct ggml_tensor * tensor;
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ggml_ctx, true);
+        }
+        if (lt.ne.size() == 2) {
+            tensor = ggml_new_tensor_2d(ggml_ctx, lt.type, lt.ne.at(0), lt.ne.at(1));
+        } else {
+            GGML_ASSERT(lt.ne.size() == 1);
+            tensor = ggml_new_tensor_1d(ggml_ctx, lt.type, lt.ne.at(0));
+        }
+        ggml_set_name(tensor, lt.name.c_str());
+        GGML_ASSERT(lt.ggml_tensor == NULL); // if this fails, we called get_tensor twice on the same tensor
+
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ggml_ctx, use_mmap);
+        }
+        tensor->backend = backend;
+        lt.ggml_tensor = tensor;
+        num_ggml_tensors_created++;
+        return tensor;
+    }
+
+    void done_getting_tensors() const {
+        if (num_ggml_tensors_created != tensors_map.tensors.size()) {
+            throw std::runtime_error(std::string("llama.cpp: file contained more tensors than expected"));
+        }
+    }
+
+    void load_all_data(llama_progress_callback progress_callback, void *  progress_callback_user_data, gguf_mlock * lmlock) {
+        size_t data_size = 0;
+        size_t prefetch_size = 0;
+        size_t lock_size = 0;
+        for (const llama_load_tensor & lt : tensors_map.tensors) {
+            data_size += lt.size;
+            if (lt.ggml_tensor->backend == GGML_BACKEND_CPU) {
+                prefetch_size += lt.size;
+            }
+        }
+
+        if (use_mmap) {
+            mapping.reset(new gguf_mmap(&file_loader->file, prefetch_size, ggml_is_numa()));
+            if (lmlock) {
+                lmlock->init(mapping->addr);
+            }
+        }
+
+        size_t done_size = 0;
+        for (llama_load_tensor & lt : tensors_map.tensors) {
+            if (progress_callback) {
+                progress_callback((float) done_size / data_size, progress_callback_user_data);
+            }
+            GGML_ASSERT(lt.ggml_tensor); // unused tensors should have been caught by load_data already
+            lt.data = (uint8_t *) lt.ggml_tensor->data;
+
+            // allocate temp buffer if not using mmap
+            if (!use_mmap && lt.data == NULL) {
+                GGML_ASSERT(lt.ggml_tensor->backend != GGML_BACKEND_CPU);
+                lt.data = (uint8_t*)malloc(ggml_nbytes(lt.ggml_tensor));
+            }
+
+            load_data_for(lt);
+
+            switch(lt.ggml_tensor->backend) {
+                case GGML_BACKEND_CPU:
+                    lt.ggml_tensor->data = lt.data;
+                    if (use_mmap && lmlock) {
+                        lock_size += lt.size;
+                        lmlock->grow_to(lock_size);
+                    }
+                    break;
+#if defined(GGML_USE_CUBLAS)
+                case GGML_BACKEND_GPU:
+                case GGML_BACKEND_GPU_SPLIT:
+                    ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+                    if (!use_mmap) {
+                        free(lt.data);
+                    }
+                    break;
+#elif defined(GGML_USE_CLBLAST)
+                case GGML_BACKEND_GPU:
+                    ggml_cl_transform_tensor(lt.data, lt.ggml_tensor);
+                    if (!use_mmap) {
+                        free(lt.data);
+                    }
+                    break;
+#endif
+                default:
+                    continue;
+            }
+
+            done_size += lt.size;
+        }
+    }
+
+    void load_data_for(llama_load_tensor & lt) {
+        if (use_mmap) {
+            lt.data = (uint8_t *) mapping->addr + lt.file_off;
+        } else {
+            gguf_file & file = file_loader->file;
+            file.seek(lt.file_off, SEEK_SET);
+            // TODO
+            //file.read_raw(lt.data, lt.size);
+        }
+
+        if (0) {
+            print_checksum(lt);
+        }
+    }
+
+    static void print_checksum(llama_load_tensor & lt) {
+        uint32_t sum = 0;
+        for (size_t i = 0; i < lt.size; i++) {
+            uint8_t byte = lt.data[i];
+            sum = byte + (sum << 6) + (sum << 16) - sum; // sdbm hash
+        }
+        fprintf(stderr, "%s checksum: %#08x (%s, size %zu)\n", lt.name.c_str(), sum,
+                llama_format_tensor_shape(lt.ne).c_str(), lt.size);
+    }
+
+};
+
+//
+// kv cache
+//
+
+static bool kv_cache_init(
+        const struct llama_hparams & hparams,
+             struct llama_kv_cache & cache,
+                         ggml_type   wtype,
+                               int   n_ctx,
+                               int   n_gpu_layers) {
+    const int n_embd  = hparams.n_embd_gqa();
+    const int n_layer = hparams.n_layer;
+
+    const int64_t n_mem      = n_layer*n_ctx;
+    const int64_t n_elements = n_embd*n_mem;
+
+    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
+    cache.n = 0;
+
+    struct ggml_init_params params;
+    params.mem_size   = cache.buf.size;
+    params.mem_buffer = cache.buf.addr;
+    params.no_alloc   = false;
+
+    cache.ctx = ggml_init(params);
+
+    if (!cache.ctx) {
+        fprintf(stderr, "%s: failed to allocate memory for kv cache\n", __func__);
+        return false;
+    }
+
+    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    ggml_set_name(cache.k, "cache_k");
+    ggml_set_name(cache.v, "cache_v");
+
+    (void) n_gpu_layers;
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer + 1) {
+        ggml_cuda_assign_buffers_no_scratch(cache.v);
+    }
+    if (n_gpu_layers > n_layer + 2) {
+        ggml_cuda_assign_buffers_no_scratch(cache.k);
+    }
+#endif // GGML_USE_CUBLAS
+
+    return true;
+}
+
+struct llama_context_params llama_context_default_params() {
+    struct llama_context_params result = {
+        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
+        /*.n_ctx                       =*/ 512,
+        /*.n_batch                     =*/ 512,
+        /*.n_gqa                       =*/ 1,
+        /*.rms_norm_eps                =*/ LLAMA_DEFAULT_RMS_EPS,
+        /*.gpu_layers                  =*/ 0,
+        /*.main_gpu                    =*/ 0,
+        /*.tensor_split                =*/ nullptr,
+        /*.rope_freq_base              =*/ 10000.0f,
+        /*.rope_freq_scale             =*/ 1.0f,
+        /*.progress_callback           =*/ nullptr,
+        /*.progress_callback_user_data =*/ nullptr,
+        /*.low_vram                    =*/ false,
+        /*.f16_kv                      =*/ true,
+        /*.logits_all                  =*/ false,
+        /*.vocab_only                  =*/ false,
+        /*.use_mmap                    =*/ true,
+        /*.use_mlock                   =*/ false,
+        /*.embedding                   =*/ false,
+    };
+
+    return result;
+}
+
+struct llama_model_quantize_params llama_model_quantize_default_params() {
+    struct llama_model_quantize_params result = {
+        /*.nthread                     =*/ 0,
+        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
+        /*.allow_requantize            =*/ false,
+        /*.quantize_output_tensor      =*/ true,
+    };
+
+    return result;
+}
+
+int llama_max_devices() {
+    return LLAMA_MAX_DEVICES;
+}
+
+bool llama_mmap_supported() {
+    return gguf_mmap::SUPPORTED;
+}
+
+bool llama_mlock_supported() {
+    return gguf_mlock::SUPPORTED;
+}
+
+void llama_backend_init(bool numa) {
+    ggml_time_init();
+
+    // needed to initialize f16 tables
+    {
+        struct ggml_init_params params = { 0, NULL, false };
+        struct ggml_context * ctx = ggml_init(params);
+        ggml_free(ctx);
+    }
+
+    if (numa) {
+        ggml_numa_init();
+    }
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_init();
+#endif
+}
+
+void llama_backend_free() {
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_free();
+#endif
+}
+
+int64_t llama_time_us() {
+    return ggml_time_us();
+}
+
+//
+// model loading
+//
+
+static const char *gguf_file_version_name(gguf_file_version version) {
+    switch (version) {
+        case gguf_file_VERSION_GGML: return "'ggml' (old version with low tokenizer quality and no mmap support)";
+        case gguf_file_VERSION_GGMF_V1: return "ggmf v1 (old version with no mmap support)";
+        case gguf_file_VERSION_GGJT_V1: return "ggjt v1 (pre #1405)";
+        case gguf_file_VERSION_GGJT_V2: return "ggjt v2 (pre #1508)";
+        case gguf_file_VERSION_GGJT_V3: return "ggjt v3 (latest)";
+    }
+
+    return "unknown";
+}
+
+static const char *llama_ftype_name(enum llama_ftype ftype) {
+    switch (ftype) {
+        case LLAMA_FTYPE_ALL_F32:     return "all F32";
+        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
+        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
+        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
+        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
+                                      return "mostly Q4_1, some F16";
+        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
+        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
+        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K: return "mostly Q2_K";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q6_K: return "mostly Q6_K";
+        default:                      return "unknown, may not work";
+    }
+}
+
+static const char *llama_model_type_name(e_model type) {
+    switch (type) {
+        case MODEL_3B: return "3B";
+        case MODEL_7B: return "7B";
+        case MODEL_13B: return "13B";
+        case MODEL_30B: return "30B";
+        case MODEL_65B: return "65B";
+        case MODEL_70B: return "70B";
+        default: GGML_ASSERT(false);
+    }
+}
+
+static void llama_model_load_internal(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gqa,
+        float rms_norm_eps,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void * progress_callback_user_data) {
+
+    model.t_start_us = ggml_time_us();
+
+    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
+
+    vocab = std::move(ml->file_loader->vocab);
+    model.hparams = ml->file_loader->hparams;
+    model.n_gpu_layers = n_gpu_layers;
+    gguf_file_version file_version = ml->file_loader->file_version;
+
+    auto & hparams = model.hparams;
+
+    // TODO: read from file
+    hparams.f_rms_norm_eps = rms_norm_eps;
+
+    {
+        switch (hparams.n_layer) {
+            case 26: model.type = e_model::MODEL_3B; break;
+            case 32: model.type = e_model::MODEL_7B; break;
+            case 40: model.type = e_model::MODEL_13B; break;
+            case 60: model.type = e_model::MODEL_30B; break;
+            case 80: model.type = e_model::MODEL_65B; break;
+            default:
+                {
+                    if (hparams.n_layer < 32) {
+                        model.type = e_model::MODEL_7B;
+                    }
+                } break;
+        }
+
+        hparams.n_ctx = n_ctx;
+
+        // LLaMAv2
+        hparams.n_head_kv = hparams.n_head / n_gqa;
+        if (model.type == e_model::MODEL_65B && n_gqa == 8) {
+            fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
+            model.type = e_model::MODEL_70B;
+            hparams.f_ffn_mult = 1.3f; // from the params.json of the 70B model
+        }
+
+        hparams.rope_freq_base  = rope_freq_base;
+        hparams.rope_freq_scale = rope_freq_scale;
+    }
+
+    // ref: https://github.com/facebookresearch/llama/blob/6c7fe276574e78057f917549435a2554000a876d/llama/model.py#L194-L199
+    const uint32_t n_ff_raw  = 2*(4*hparams.n_embd)/3;
+    const uint32_t n_ff_mult = hparams.f_ffn_mult*n_ff_raw;
+    const uint32_t n_ff      = ((n_ff_mult + hparams.n_mult - 1)/hparams.n_mult)*hparams.n_mult;
+    //const uint32_t n_ff = 28672;
+
+    {
+        fprintf(stderr, "%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
+        fprintf(stderr, "%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
+        fprintf(stderr, "%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
+        fprintf(stderr, "%s: n_embd     = %u\n",   __func__, hparams.n_embd);
+        fprintf(stderr, "%s: n_mult     = %u\n",   __func__, hparams.n_mult);
+        fprintf(stderr, "%s: n_head     = %u\n",   __func__, hparams.n_head);
+        fprintf(stderr, "%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
+        fprintf(stderr, "%s: n_layer    = %u\n",   __func__, hparams.n_layer);
+        fprintf(stderr, "%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        fprintf(stderr, "%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
+        fprintf(stderr, "%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
+        fprintf(stderr, "%s: n_ff       = %u\n",   __func__, n_ff);
+        fprintf(stderr, "%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
+        fprintf(stderr, "%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
+        fprintf(stderr, "%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
+        fprintf(stderr, "%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
+    }
+
+    if (file_version < gguf_file_VERSION_GGJT_V2) {
+        if (hparams.ftype != LLAMA_FTYPE_ALL_F32     &&
+            hparams.ftype != LLAMA_FTYPE_MOSTLY_F16  &&
+            hparams.ftype != LLAMA_FTYPE_MOSTLY_Q8_0) {
+            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1405)"));
+        }
+    }
+
+    if (file_version < gguf_file_VERSION_GGJT_V3) {
+        if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
+            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
+            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
+            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
+        }
+    }
+
+    if (vocab_only) {
+        return;
+    }
+
+    auto & ctx = model.ctx;
+
+    size_t ctx_size;
+    size_t mmapped_size;
+    ml->calc_sizes(&ctx_size, &mmapped_size);
+    fprintf(stderr, "%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
+
+    // create the ggml context
+    {
+        model.buf.resize(ctx_size);
+        if (use_mlock) {
+            model.mlock_buf.init   (model.buf.addr);
+            model.mlock_buf.grow_to(model.buf.size);
+        }
+
+        struct ggml_init_params params = {
+            /*.mem_size   =*/ model.buf.size,
+            /*.mem_buffer =*/ model.buf.addr,
+            /*.no_alloc   =*/ ml->use_mmap,
+        };
+
+        model.ctx = ggml_init(params);
+        if (!model.ctx) {
+            throw std::runtime_error(format("ggml_init() failed"));
+        }
+    }
+
+    (void) main_gpu;
+#if defined(GGML_USE_CUBLAS)
+    fprintf(stderr, "%s: using CUDA for GPU acceleration\n", __func__);
+    ggml_cuda_set_main_device(main_gpu);
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU_SPLIT
+#elif defined(GGML_USE_CLBLAST)
+    fprintf(stderr, "%s: using OpenCL for GPU acceleration\n", __func__);
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU
+#else
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_CPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_CPU
+#endif
+
+    // prepare memory for the weights
+    size_t vram_weights = 0;
+    size_t vram_scratch = 0;
+    {
+        const uint32_t n_embd     = hparams.n_embd;
+        const uint32_t n_embd_gqa = hparams.n_embd_gqa();
+        const uint32_t n_layer    = hparams.n_layer;
+        const uint32_t n_vocab    = hparams.n_vocab;
+
+        ml->ggml_ctx = ctx;
+
+        model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
+
+        // "output" tensor
+        {
+            ggml_backend backend_norm;
+            ggml_backend backend_output;
+            if (n_gpu_layers > int(n_layer)) { // NOLINT
+                // norm is not performance relevant on its own but keeping it in VRAM reduces data copying
+                // on Windows however this is detrimental unless everything is on the GPU
+#ifndef _WIN32
+                backend_norm = low_vram ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
+#else
+                backend_norm = low_vram || n_gpu_layers <= (int) n_layer + 2 ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
+#endif // _WIN32
+
+                backend_output = LLAMA_BACKEND_OFFLOAD_SPLIT;
+            } else {
+                backend_norm = GGML_BACKEND_CPU;
+                backend_output = GGML_BACKEND_CPU;
+            }
+
+            model.norm   = ml->get_tensor("norm.weight",   {n_embd},          backend_norm);
+            model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}, backend_output);
+            if (backend_norm == GGML_BACKEND_GPU) {
+                vram_weights += ggml_nbytes(model.norm);
+            }
+            if (backend_output == GGML_BACKEND_GPU_SPLIT) {
+                vram_weights += ggml_nbytes(model.output);
+            }
+        }
+
+        const int i_gpu_start = n_layer - n_gpu_layers;
+
+        model.layers.resize(n_layer);
+        for (uint32_t i = 0; i < n_layer; ++i) {
+            const ggml_backend backend = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD; // NOLINT
+            const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
+
+            auto & layer = model.layers[i];
+
+            std::string layers_i = "layers." + std::to_string(i);
+
+            layer.attention_norm = ml->get_tensor(layers_i + ".attention_norm.weight", {n_embd}, backend);
+
+            layer.wq = ml->get_tensor(layers_i + ".attention.wq.weight", {n_embd, n_embd},     backend_split);
+            layer.wk = ml->get_tensor(layers_i + ".attention.wk.weight", {n_embd, n_embd_gqa}, backend_split);
+            layer.wv = ml->get_tensor(layers_i + ".attention.wv.weight", {n_embd, n_embd_gqa}, backend_split);
+            layer.wo = ml->get_tensor(layers_i + ".attention.wo.weight", {n_embd, n_embd},     backend_split);
+
+            layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
+
+            layer.w1 = ml->get_tensor(layers_i + ".feed_forward.w1.weight", {n_embd,   n_ff}, backend_split);
+            layer.w2 = ml->get_tensor(layers_i + ".feed_forward.w2.weight", {  n_ff, n_embd}, backend_split);
+            layer.w3 = ml->get_tensor(layers_i + ".feed_forward.w3.weight", {n_embd,   n_ff}, backend_split);
+
+            if (backend == GGML_BACKEND_GPU) {
+                vram_weights +=
+                    ggml_nbytes(layer.attention_norm) + ggml_nbytes(layer.wq) + ggml_nbytes(layer.wk)             +
+                    ggml_nbytes(layer.wv)             + ggml_nbytes(layer.wo) + ggml_nbytes(layer.ffn_norm) +
+                    ggml_nbytes(layer.w1)             + ggml_nbytes(layer.w2) + ggml_nbytes(layer.w3);
+            }
+        }
+    }
+
+    ml->done_getting_tensors();
+
+    // print memory requirements
+    {
+        const size_t scale = memory_type == GGML_TYPE_F32 ? 2 : 1;
+
+        // this is the total memory required to run the inference
+        const size_t mem_required =
+            ctx_size +
+            mmapped_size - vram_weights + // weights in VRAM not in memory
+            MEM_REQ_SCRATCH0(hparams.n_ctx).at(model.type) +
+            MEM_REQ_SCRATCH1().at(model.type) +
+            MEM_REQ_EVAL().at(model.type);
+
+        // this is the memory required by one llama_state
+        const size_t mem_required_state =
+            scale*hparams.kv_size();
+
+        fprintf(stderr, "%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
+                mem_required / 1024.0 / 1024.0, mem_required_state / 1024.0 / 1024.0);
+
+        (void) vram_scratch;
+        (void) n_batch;
+#ifdef GGML_USE_CUBLAS
+        if (low_vram) {
+            fprintf(stderr, "%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
+            ggml_cuda_set_scratch_size(0); // disable scratch
+        } else {
+            const size_t vram_scratch_base = VRAM_REQ_SCRATCH_BASE().at(model.type);
+            const size_t vram_scratch_per_context = VRAM_REQ_SCRATCH_PER_CONTEXT().at(model.type);
+            vram_scratch = n_batch * (vram_scratch_base + n_ctx * vram_scratch_per_context);
+            ggml_cuda_set_scratch_size(vram_scratch);
+            if (n_gpu_layers > 0) {
+                fprintf(stderr, "%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
+                        __func__, vram_scratch_base / kB, vram_scratch_per_context,
+                        (vram_scratch + MB - 1) / MB); // round up
+            }
+        }
+#endif // GGML_USE_CUBLAS
+
+#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
+        const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
+
+        fprintf(stderr, "%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
+        if (n_gpu_layers > (int) hparams.n_layer) {
+            fprintf(stderr, "%s: offloading non-repeating layers to GPU\n", __func__);
+        }
+        size_t vram_kv_cache = 0;
+
+#ifdef GGML_USE_CUBLAS
+        const int max_backend_supported_layers = hparams.n_layer + 3;
+        const int max_offloadable_layers = low_vram ? hparams.n_layer + 1 : hparams.n_layer + 3;
+        if (n_gpu_layers > (int) hparams.n_layer + 1) {
+            if (low_vram) {
+                fprintf(stderr, "%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
+            } else {
+                fprintf(stderr, "%s: offloading v cache to GPU\n", __func__);
+                vram_kv_cache += hparams.kv_size() / 2;
+            }
+        }
+        if (n_gpu_layers > (int) hparams.n_layer + 2) {
+            if (low_vram) {
+                fprintf(stderr, "%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
+            } else {
+                fprintf(stderr, "%s: offloading k cache to GPU\n", __func__);
+                vram_kv_cache += hparams.kv_size() / 2;
+            }
+        }
+#elif defined(GGML_USE_CLBLAST)
+        const int max_backend_supported_layers = hparams.n_layer + 1;
+        const int max_offloadable_layers = hparams.n_layer + 1;
+#endif // GGML_USE_CUBLAS
+
+        fprintf(stderr, "%s: offloaded %d/%d layers to GPU\n",
+                __func__, std::min(n_gpu_layers, max_offloadable_layers), max_backend_supported_layers);
+        fprintf(stderr, "%s: total VRAM used: %zu MB\n",
+                __func__, (vram_weights + vram_scratch + vram_kv_cache + MB - 1) / MB); // round up
+#else
+        (void) n_gpu_layers;
+#endif // defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
+    }
+
+    // populate `tensors_by_name`
+    for (llama_load_tensor & lt : ml->tensors_map.tensors) {
+        model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
+    }
+
+    (void) tensor_split;
+#if defined(GGML_USE_CUBLAS)
+    {
+        ggml_cuda_set_tensor_split(tensor_split);
+    }
+#endif
+
+    ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
+
+    if (progress_callback) {
+        progress_callback(1.0f, progress_callback_user_data);
+    }
+
+    model.mapping = std::move(ml->mapping);
+
+    // loading time will be recalculate after the first eval, so
+    // we take page faults deferred by mmap() into consideration
+    model.t_load_us = ggml_time_us() - model.t_start_us;
+}
+
+static bool llama_model_load(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gqa,
+        float rms_norm_eps,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void *progress_callback_user_data) {
+    try {
+        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gqa, rms_norm_eps, n_gpu_layers, main_gpu, tensor_split, rope_freq_base, rope_freq_scale, low_vram, memory_type,
+                                  use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
+        return true;
+    } catch (const std::exception & err) {
+        fprintf(stderr, "error loading model: %s\n", err.what());
+        return false;
+    }
+}
+
+// evaluate the transformer
+//
+//   - lctx:      llama context
+//   - tokens:    new batch of tokens to process
+//   - embd       embeddings input
+//   - n_tokens   number of tokens
+//   - n_past:    the context size so far
+//   - n_threads: number of threads to use
+//
+static bool llama_eval_internal(
+         llama_context & lctx,
+     const llama_token * tokens,
+           const float * embd,
+                   int   n_tokens,
+                   int   n_past,
+                   int   n_threads,
+            const char * cgraph_fname) {
+
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
+#endif
+
+    const int64_t t_start_us = ggml_time_us();
+
+    const int N = n_tokens;
+
+    const auto & model   = lctx.model;
+    const auto & hparams = model.hparams;
+
+    const auto & kv_self = lctx.kv_self;
+
+    GGML_ASSERT(!!kv_self.ctx);
+
+    const int64_t n_embd      = hparams.n_embd;
+    const int64_t n_layer     = hparams.n_layer;
+    const int64_t n_ctx       = hparams.n_ctx;
+    const int64_t n_head      = hparams.n_head;
+    const int64_t n_head_kv   = hparams.n_head_kv;
+    const int64_t n_embd_head = hparams.n_embd_head();
+    const int64_t n_vocab     = hparams.n_vocab;
+    const int64_t n_embd_gqa  = hparams.n_embd_gqa();
+
+
+    GGML_ASSERT(n_embd_head == hparams.n_rot);
+
+    const float freq_base  = hparams.rope_freq_base;
+    const float freq_scale = hparams.rope_freq_scale;
+    const float rms_norm_eps = hparams.f_rms_norm_eps;
+
+    const int n_gpu_layers = model.n_gpu_layers;
+
+    auto & mem_per_token = lctx.mem_per_token;
+    auto & buf_compute   = lctx.buf_compute;
+
+    struct ggml_init_params params = {
+        /*.mem_size   =*/ buf_compute.size,
+        /*.mem_buffer =*/ buf_compute.addr,
+        /*.no_alloc   =*/ false,
+    };
+
+    struct ggml_context * ctx0 = ggml_init(params);
+
+    ggml_cgraph * gf = ggml_new_graph(ctx0);
+
+    // for big prompts, if BLAS is enabled, it is better to use only one thread
+    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
+    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
+
+    struct ggml_tensor * cur;
+    struct ggml_tensor * inpL;
+
+    if (tokens) {
+        struct ggml_tensor * inp_tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
+        memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
+        ggml_set_name(inp_tokens, "inp_tokens");
+
+        inpL = ggml_get_rows(ctx0, model.tok_embeddings, inp_tokens);
+    } else {
+#ifdef GGML_USE_MPI
+        GGML_ASSERT(false && "not implemented");
+#endif
+
+        inpL = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N);
+        memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
+    }
+
+    const int i_gpu_start = n_layer - n_gpu_layers;
+    (void) i_gpu_start;
+
+    // offload functions set the tensor output backend to GPU
+    // tensors are GPU-accelerated if any input or the output has been offloaded
+    //
+    // with the low VRAM option VRAM scratch is disabled in llama_load_model_internal
+    // in that case ggml_cuda_assign_buffers has no effect
+    offload_func_t offload_func_nr = llama_nop; // nr = non-repeating
+    offload_func_t offload_func_kq = llama_nop;
+    offload_func_t offload_func_v  = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer) {
+        offload_func_nr = ggml_cuda_assign_buffers;
+    }
+    if (n_gpu_layers > n_layer + 1) {
+        offload_func_v  = ggml_cuda_assign_buffers;
+    }
+    if (n_gpu_layers > n_layer + 2) {
+        offload_func_kq = ggml_cuda_assign_buffers;
+    }
+#endif // GGML_USE_CUBLAS
+
+    for (int il = 0; il < n_layer; ++il) {
+        ggml_format_name(inpL, "layer_inp_%d", il);
+
+        offload_func_t offload_func = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+        if (il >= i_gpu_start) {
+            offload_func = ggml_cuda_assign_buffers;
+        }
+#endif // GGML_USE_CUBLAS
+
+        struct ggml_tensor * inpSA = inpL;
+
+        lctx.use_buf(ctx0, 0);
+
+        // norm
+        {
+            cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+            offload_func(cur);
+            ggml_set_name(cur, "rms_norm_0");
+
+            // cur = cur*attention_norm(broadcasted)
+            cur = ggml_mul(ctx0, cur, model.layers[il].attention_norm);
+            offload_func(cur);
+            ggml_set_name(cur, "attention_norm_0");
+        }
+
+        // self-attention
+        {
+            // compute Q and K and RoPE them
+            struct ggml_tensor * tmpk = ggml_mul_mat(ctx0, model.layers[il].wk, cur);
+            offload_func_kq(tmpk);
+            ggml_set_name(tmpk, "tmpk");
+
+            struct ggml_tensor * tmpq = ggml_mul_mat(ctx0, model.layers[il].wq, cur);
+            offload_func_kq(tmpq);
+            ggml_set_name(tmpq, "tmpq");
+
+            struct ggml_tensor * Kcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpk, n_embd_head, n_head_kv, N), n_past, n_embd_head, 0, 0, freq_base, freq_scale);
+            offload_func_kq(Kcur);
+            ggml_set_name(Kcur, "Kcur");
+
+            struct ggml_tensor * Qcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpq, n_embd_head, n_head, N),    n_past, n_embd_head, 0, 0, freq_base, freq_scale);
+            offload_func_kq(Qcur);
+            ggml_set_name(Qcur, "Qcur");
+
+            // store key and value to memory
+            {
+                // compute the transposed [N, n_embd] V matrix
+
+                struct ggml_tensor * tmpv = ggml_mul_mat(ctx0, model.layers[il].wv, cur);
+                offload_func_v(tmpv);
+                ggml_set_name(tmpv, "tmpv");
+
+                struct ggml_tensor * Vcur = ggml_transpose(ctx0, ggml_reshape_2d(ctx0, tmpv, n_embd_gqa, N));
+                offload_func_v(Vcur);
+                ggml_set_name(Vcur, "Vcur");
+
+                struct ggml_tensor * k = ggml_view_1d(ctx0, kv_self.k, N*n_embd_gqa, (ggml_element_size(kv_self.k)*n_embd_gqa)*(il*n_ctx + n_past));
+                offload_func_kq(k);
+                ggml_set_name(k, "k");
+
+                struct ggml_tensor * v = ggml_view_2d(ctx0, kv_self.v, N, n_embd_gqa,
+                        (   n_ctx)*ggml_element_size(kv_self.v),
+                        (il*n_ctx)*ggml_element_size(kv_self.v)*n_embd_gqa + n_past*ggml_element_size(kv_self.v));
+                offload_func_v(v);
+                ggml_set_name(v, "v");
+
+                // important: storing RoPE-ed version of K in the KV cache!
+                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Kcur, k));
+                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Vcur, v));
+            }
+
+            struct ggml_tensor * Q =
+                ggml_permute(ctx0,
+                        Qcur,
+                        0, 2, 1, 3);
+            offload_func_kq(Q);
+            ggml_set_name(Q, "Q");
+
+            struct ggml_tensor * K =
+                ggml_permute(ctx0,
+                        ggml_reshape_3d(ctx0,
+                            ggml_view_1d(ctx0, kv_self.k, (n_past + N)*n_embd_gqa, il*n_ctx*ggml_element_size(kv_self.k)*n_embd_gqa),
+                            n_embd_head, n_head_kv, n_past + N),
+                        0, 2, 1, 3);
+            offload_func_kq(K);
+            ggml_set_name(K, "K");
+
+            // K * Q
+            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
+            offload_func_kq(KQ);
+            ggml_set_name(KQ, "KQ");
+
+            // KQ_scaled = KQ / sqrt(n_embd_head)
+            struct ggml_tensor * KQ_scale = ggml_new_f32(ctx0, 1.0f/sqrtf(float(n_embd)/n_head));
+            ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
+
+            // KQ_scaled shape [n_past + N, N, n_head, 1]
+            struct ggml_tensor * KQ_scaled = ggml_scale_inplace(ctx0, KQ, KQ_scale);
+            offload_func_kq(KQ_scaled);
+            ggml_set_name(KQ_scaled, "KQ_scaled");
+
+            // KQ_masked = mask_past(KQ_scaled)
+            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
+            offload_func_kq(KQ_masked);
+            ggml_set_name(KQ_masked, "KQ_masked");
+
+            // KQ = soft_max(KQ_masked)
+            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
+            offload_func_v(KQ_soft_max);
+            ggml_set_name(KQ_soft_max, "KQ_soft_max");
+
+            // split cached V into n_head heads
+            struct ggml_tensor * V =
+                ggml_view_3d(ctx0, kv_self.v,
+                        n_past + N, n_embd_head, n_head_kv,
+                        n_ctx*ggml_element_size(kv_self.v),
+                        n_ctx*ggml_element_size(kv_self.v)*n_embd_head,
+                        n_ctx*ggml_element_size(kv_self.v)*n_embd_gqa*il);
+            offload_func_v(V);
+            ggml_set_name(V, "V");
+
+#if 1
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
+            offload_func_v(KQV);
+            ggml_set_name(KQV, "KQV");
+#else
+            // make V contiguous in memory to speed up the matmul, however we waste time on the copy
+            // on M1 this is faster for the perplexity computation, but ~5% slower for the single-token generation
+            // is there a better way?
+            struct ggml_tensor * V_cont = ggml_cpy(ctx0, V, ggml_new_tensor_3d(ctx0, kv_self.v->type, n_past + N, n_embd_head, n_head));
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V_cont, KQ_soft_max);
+#endif
+
+            // KQV_merged = KQV.permute(0, 2, 1, 3)
+            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
+            offload_func_v(KQV_merged);
+            ggml_set_name(KQV_merged, "KQV_merged");
+
+            // cur = KQV_merged.contiguous().view(n_embd, N)
+            cur = ggml_cpy(ctx0,
+                    KQV_merged,
+                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+            offload_func_v(cur);
+            ggml_set_name(cur, "KQV_merged_contiguous");
+
+            // projection (no bias)
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].wo,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_wo");
+        }
+
+        lctx.use_buf(ctx0, 1);
+
+        struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpSA);
+        offload_func(inpFF);
+        ggml_set_name(inpFF, "inpFF");
+
+        // feed-forward network
+        {
+            // norm
+            {
+                cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps);
+                offload_func(cur);
+                ggml_set_name(cur, "rms_norm_1");
+
+                // cur = cur*ffn_norm(broadcasted)
+                cur = ggml_mul(ctx0, cur, model.layers[il].ffn_norm);
+                offload_func(cur);
+                ggml_set_name(cur, "ffn_norm");
+            }
+
+            struct ggml_tensor * tmp = ggml_mul_mat(ctx0,
+                    model.layers[il].w3,
+                    cur);
+            offload_func(tmp);
+            ggml_set_name(tmp, "result_w3");
+
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].w1,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_w1");
+
+            // SILU activation
+            cur = ggml_silu(ctx0, cur);
+            offload_func(cur);
+            ggml_set_name(cur, "silu");
+
+            cur = ggml_mul(ctx0, cur, tmp);
+            offload_func(cur);
+            ggml_set_name(cur, "silu_x_result_w3");
+
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].w2,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_w2");
+        }
+
+        cur = ggml_add(ctx0, cur, inpFF);
+        offload_func(cur);
+        ggml_set_name(cur, "inpFF_+_result_w2");
+
+        // input for next layer
+        inpL = cur;
+    }
+
+    lctx.use_buf(ctx0, 0);
+
+    // used at the end to optionally extract the embeddings
+    struct ggml_tensor * embeddings = NULL;
+
+    // norm
+    {
+        cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+        offload_func_nr(cur);
+        ggml_set_name(cur, "rms_norm_2");
+
+        // cur = cur*norm(broadcasted)
+        cur = ggml_mul(ctx0, cur, model.norm);
+        // offload_func_nr(cur); // TODO CPU + GPU mirrored backend
+        ggml_set_name(cur, "result_norm");
+
+        embeddings = cur;
+    }
+
+    // lm_head
+    cur = ggml_mul_mat(ctx0, model.output, cur);
+    ggml_set_name(cur, "result_output");
+
+    lctx.use_buf(ctx0, -1);
+
+    // logits -> probs
+    //cur = ggml_soft_max_inplace(ctx0, cur);
+
+    // run the computation
+    ggml_build_forward_expand(gf, cur);
+
+    // fprintf(stderr, "graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf.n_nodes, gf.n_leafs);
+
+#if GGML_USE_MPI
+    ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);
+#endif
+
+#ifdef GGML_USE_METAL
+    if (lctx.ctx_metal && N == 1) {
+        if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
+            ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
+        }
+        ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
+        ggml_metal_graph_compute(lctx.ctx_metal, gf);
+        ggml_metal_get_tensor   (lctx.ctx_metal, cur);
+    } else {
+        // IMPORTANT:
+        // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
+        // ggml_graph_compute(). It uses Apple's Accelerate CBLAS API which takes advantage of the ANE or the AMX
+        // coprocessor.
+        //
+        // When we implement Matrix x Matrix Metal multiplication, we can avoid this branch.
+        // But for now, we have focused only on Matrix x Vector Metal multiplication.
+        //
+        // TODO: avoid these syncs via shared memory (ref #1696)
+        //
+        if (lctx.ctx_metal) {
+            // We need to sync the GPU KV cache with the CPU KV cache
+            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.k);
+            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.v);
+        }
+
+        ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
+    }
+#else
+    ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
+#endif
+
+#if GGML_USE_MPI
+    ggml_mpi_graph_compute_post(lctx.ctx_mpi, gf, n_layer);
+#endif
+
+    // update kv token count
+    lctx.kv_self.n = n_past + N;
+
+    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
+
+    if (cgraph_fname) {
+        ggml_graph_export(gf, cgraph_fname);
+    }
+
+#ifdef GGML_PERF
+    // print timing information per ggml operation (for debugging purposes)
+    // requires GGML_PERF to be defined
+    ggml_graph_print(gf);
+#endif
+
+    // plot the computation graph in dot format (for debugging purposes)
+    //if (n_past%100 == 0) {
+    //    ggml_graph_dump_dot(gf, NULL, "llama.dot");
+    //}
+
+    // extract logits
+    {
+        auto & logits_out = lctx.logits;
+
+        if (lctx.logits_all) {
+            logits_out.resize(n_vocab * N);
+            memcpy(logits_out.data(), (float *) ggml_get_data(res), sizeof(float)*n_vocab*N);
+        } else {
+            // return result for just the last token
+            logits_out.resize(n_vocab);
+            memcpy(logits_out.data(), (float *) ggml_get_data(res) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
+        }
+    }
+
+    // extract embeddings
+    if (!lctx.embedding.empty()) {
+        auto & embedding_out = lctx.embedding;
+
+        embedding_out.resize(n_embd);
+        memcpy(embedding_out.data(), (float *) ggml_get_data(embeddings) + (n_embd*(N - 1)), sizeof(float)*n_embd);
+    }
+
+    if (mem_per_token == 0) {
+        mem_per_token = ggml_used_mem(ctx0)/N;
+    }
+
+#if 0
+    printf("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
+            ggml_used_mem(ctx0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(1)/1024.0/1024.0,
+            lctx.work_buffer.size()/1024.0/1024.0,
+            n_past, N);
+#endif
+
+    ggml_free(ctx0);
+
+    // measure the performance only for the single-token evals
+    if (N == 1) {
+        lctx.t_eval_us += ggml_time_us() - t_start_us;
+        lctx.n_eval++;
+    }
+    else if (N > 1) {
+        lctx.t_p_eval_us += ggml_time_us() - t_start_us;
+        lctx.n_p_eval += N;
+    }
+
+    return true;
+}
+
+//
+// tokenizer
+//
+
+static size_t utf8_len(char src) {
+    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
+    return lookup[highbits];
+}
+
+struct llama_sp_symbol {
+    using index = int;
+    index prev;
+    index next;
+    const char * text;
+    size_t n;
+};
+
+static_assert(std::is_trivially_copyable<llama_sp_symbol>::value, "llama_sp_symbol is not trivially copyable");
+
+struct llama_sp_bigram {
+    struct comparator {
+        bool operator()(llama_sp_bigram & l, llama_sp_bigram & r) {
+            return (l.score < r.score) || (l.score == r.score && l.left > r.left);
+        }
+    };
+    using queue_storage = std::vector<llama_sp_bigram>;
+    using queue = std::priority_queue<llama_sp_bigram, queue_storage, comparator>;
+    llama_sp_symbol::index left;
+    llama_sp_symbol::index right;
+    float score;
+    size_t size;
+};
+
+// original implementation:
+// https://github.com/ggerganov/llama.cpp/commit/074bea2eb1f1349a0118239c4152914aecaa1be4
+struct llama_tokenizer {
+    llama_tokenizer(const llama_vocab & vocab): vocab_(vocab) {}
+
+    void tokenize(const std::string & text, std::vector<llama_vocab::id> & output) {
+        // split string into utf8 chars
+        int index = 0;
+        size_t offs = 0;
+        while (offs < text.size()) {
+            llama_sp_symbol sym;
+            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
+            sym.text = text.c_str() + offs;
+            sym.n = char_len;
+            offs += char_len;
+            sym.prev = index - 1;
+            sym.next = offs == text.size() ? -1 : index + 1;
+            index++;
+            symbols_.emplace_back(sym);
+        }
+
+        // seed the work queue with all possible 2-character tokens.
+        for (size_t i = 1; i < symbols_.size(); ++i) {
+            try_add_bigram(i - 1, i);
+        }
+
+        // keep substituting the highest frequency pairs for as long as we can.
+        while (!work_queue_.empty()) {
+            auto bigram = work_queue_.top();
+            work_queue_.pop();
+
+            auto & left_sym = symbols_[bigram.left];
+            auto & right_sym = symbols_[bigram.right];
+
+            // if one of the symbols already got merged, skip it.
+            if (left_sym.n == 0 || right_sym.n == 0 ||
+                left_sym.n + right_sym.n != bigram.size) {
+                continue;
+            }
+
+            // merge the right sym into the left one
+            left_sym.n += right_sym.n;
+            right_sym.n = 0;
+
+            //printf("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
+
+            // remove the right sym from the chain
+            left_sym.next = right_sym.next;
+            if (right_sym.next >= 0) {
+                symbols_[right_sym.next].prev = bigram.left;
+            }
+
+            // find more substitutions
+            try_add_bigram(left_sym.prev, bigram.left);
+            try_add_bigram(bigram.left, left_sym.next);
+        }
+
+        for (int i = 0; i != -1; i = symbols_[i].next) {
+            auto & symbol = symbols_[i];
+            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
+
+            if (token == vocab_.token_to_id.end()) {
+                // output any symbols that did not form tokens as bytes.
+                for (int j = 0; j < (int) symbol.n; ++j) {
+                    llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
+                    output.push_back(token_id);
+                }
+            } else {
+                output.push_back((*token).second);
+            }
+        }
+    }
+
+private:
+    void try_add_bigram(int left, int right) {
+        if (left == -1 || right == -1) {
+            return;
+        }
+
+        const std::string text = std::string(symbols_[left].text, symbols_[left].n + symbols_[right].n);
+        auto token = vocab_.token_to_id.find(text);
+
+        if (token == vocab_.token_to_id.end()) {
+            return;
+        }
+
+        if (static_cast<size_t>((*token).second) >= vocab_.id_to_token.size()) {
+            return;
+        }
+
+        const auto &tok_score = vocab_.id_to_token[(*token).second];
+
+        llama_sp_bigram bigram;
+        bigram.left = left;
+        bigram.right = right;
+        bigram.score = tok_score.score;
+        bigram.size = text.size();
+        work_queue_.push(bigram);
+    }
+
+    const llama_vocab & vocab_;
+    std::vector<llama_sp_symbol> symbols_;
+    llama_sp_bigram::queue work_queue_;
+};
+
+static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
+    llama_tokenizer tokenizer(vocab);
+    std::vector<llama_vocab::id> output;
+
+    if (text.empty()) {
+        return output;
+    }
+
+    if (bos) {
+        output.push_back(llama_token_bos());
+    }
+
+    tokenizer.tokenize(text, output);
+    return output;
+}
+
+//
+// grammar - internal
+//
+
+struct llama_grammar {
+    const std::vector<std::vector<llama_grammar_element>>   rules;
+    std::vector<std::vector<const llama_grammar_element *>> stacks;
+};
+
+struct llama_grammar_candidate {
+    size_t           index;
+    const uint32_t * code_points;
+};
+
+// NOTE: assumes valid utf8 (but checks for overrun)
+// adds a terminating 0 for use as pointer
+std::vector<uint32_t> decode_utf8(const char * src) {
+    static const int      lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    const char          * pos      = src;
+    std::vector<uint32_t> code_points;
+    while (*pos != 0) {
+        uint8_t  first_byte = static_cast<uint8_t>(*pos);
+        uint8_t  highbits   = first_byte >> 4;
+        int      len        = lookup[highbits];
+        uint8_t  mask       = (1 << (8 - len)) - 1;
+        uint32_t value      = first_byte & mask;
+        const char * end    = pos + len; // may overrun!
+        ++pos;
+        for ( ; pos < end && *pos != 0; ++pos) {
+            value = (value << 6) + (static_cast<uint8_t>(*pos) & 0x3F);
+        }
+        code_points.push_back(value);
+    }
+    code_points.push_back(0);
+    return code_points;
+}
+
+// returns true iff pos points to the end of one of the definitions of a rule
+static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
+    switch (pos->type) {
+        case LLAMA_GRETYPE_END: return true;
+        case LLAMA_GRETYPE_ALT: return true;
+        default:                return false;
+    }
+}
+
+// returns true iff chr satisfies the char range at pos (regular or inverse range)
+// asserts that pos is pointing to a char range element
+static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
+        const llama_grammar_element * pos,
+        const uint32_t                chr) {
+
+    bool found            = false;
+    bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
+    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT);
+
+    do {
+        if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
+            // inclusive range, e.g. [a-z]
+            found = found || (pos->value <= chr && chr <= pos[1].value);
+            pos += 2;
+        } else {
+            // exact char match, e.g. [a] or "a"
+            found = found || pos->value == chr;
+            pos += 1;
+        }
+    } while (pos->type == LLAMA_GRETYPE_CHAR_ALT);
+
+    return std::make_pair(found == is_positive_char, pos);
+}
+
+// transforms a grammar pushdown stack into N possible stacks, all ending
+// at a character range (terminal element)
+static void llama_grammar_advance_stack(
+        const std::vector<std::vector<llama_grammar_element>>   & rules,
+        const std::vector<const llama_grammar_element *>        & stack,
+        std::vector<std::vector<const llama_grammar_element *>> & new_stacks) {
+
+    if (stack.empty()) {
+        new_stacks.push_back(stack);
+        return;
+    }
+
+    const llama_grammar_element * pos = stack.back();
+
+    switch (pos->type) {
+        case LLAMA_GRETYPE_RULE_REF: {
+            const size_t                  rule_id = static_cast<size_t>(pos->value);
+            const llama_grammar_element * subpos  = rules[rule_id].data();
+            do {
+                // init new stack without the top (pos)
+                std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
+                if (!llama_grammar_is_end_of_sequence(pos + 1)) {
+                    // if this rule ref is followed by another element, add that to stack
+                    new_stack.push_back(pos + 1);
+                }
+                if (!llama_grammar_is_end_of_sequence(subpos)) {
+                    // if alternate is nonempty, add to stack
+                    new_stack.push_back(subpos);
+                }
+                llama_grammar_advance_stack(rules, new_stack, new_stacks);
+                while (!llama_grammar_is_end_of_sequence(subpos)) {
+                    // scan to end of alternate def
+                    subpos++;
+                }
+                if (subpos->type == LLAMA_GRETYPE_ALT) {
+                    // there's another alternate def of this rule to process
+                    subpos++;
+                } else {
+                    break;
+                }
+            } while (true);
+            break;
+        }
+        case LLAMA_GRETYPE_CHAR:
+        case LLAMA_GRETYPE_CHAR_NOT:
+            new_stacks.push_back(stack);
+            break;
+        default:
+            // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range
+            // (LLAMA_GRETYPE_CHAR_ALT, LLAMA_GRETYPE_CHAR_RNG_UPPER); stack should never be left on
+            // those
+            GGML_ASSERT(false);
+    }
+}
+
+// takes a set of possible pushdown stacks on a grammar, which are required to
+// be positioned at a character range (see `llama_grammar_advance_stack`), and
+// produces the N possible stacks if the given char is accepted at those
+// positions
+static std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const uint32_t                                                  chr) {
+
+    std::vector<std::vector<const llama_grammar_element *>> new_stacks;
+
+    for (const auto & stack : stacks) {
+        if (stack.empty()) {
+            continue;
+        }
+
+        auto match = llama_grammar_match_char(stack.back(), chr);
+        if (match.first) {
+            const llama_grammar_element * pos = match.second;
+
+            // update top of stack to next element, if any
+            std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
+            if (!llama_grammar_is_end_of_sequence(pos)) {
+                new_stack.push_back(pos);
+            }
+            llama_grammar_advance_stack(rules, new_stack, new_stacks);
+        }
+    }
+
+    return new_stacks;
+}
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const std::vector<llama_grammar_candidate>                    & candidates);
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_stack(
+        const std::vector<std::vector<llama_grammar_element>> & rules,
+        const std::vector<const llama_grammar_element *>      & stack,
+        const std::vector<llama_grammar_candidate>            & candidates) {
+
+    std::vector<llama_grammar_candidate> rejects;
+
+    if (stack.empty()) {
+        // accept nothing; EOS is handled elsewhere
+        rejects.insert(rejects.end(), candidates.begin(), candidates.end());
+        return rejects;
+    }
+
+    const llama_grammar_element * stack_pos = stack.back();
+
+    std::vector<llama_grammar_candidate> next_candidates;
+    for (auto tok : candidates) {
+        if (llama_grammar_match_char(stack_pos, tok.code_points[0]).first) {
+            if (tok.code_points[1] != 0) {
+                next_candidates.push_back({ tok.index, tok.code_points + 1 });
+            }
+        } else {
+            rejects.push_back(tok);
+        }
+    }
+
+    auto stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
+
+    // update top of stack to next element, if any
+    std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
+    if (!llama_grammar_is_end_of_sequence(stack_pos_after)) {
+        stack_after.push_back(stack_pos_after);
+    }
+    std::vector<std::vector<const llama_grammar_element *>> next_stacks;
+    llama_grammar_advance_stack(rules, stack_after, next_stacks);
+
+    auto next_rejects = llama_grammar_reject_candidates(rules, next_stacks, next_candidates);
+    for (auto tok : next_rejects) {
+        rejects.push_back({ tok.index, tok.code_points - 1 });
+    }
+
+    return rejects;
+}
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const std::vector<llama_grammar_candidate>                    & candidates) {
+    GGML_ASSERT(!stacks.empty()); // REVIEW
+
+    if (candidates.empty()) {
+        return std::vector<llama_grammar_candidate>();
+    }
+
+    auto rejects = llama_grammar_reject_candidates_for_stack(rules, stacks.front(), candidates);
+
+    for (size_t i = 1, size = stacks.size(); i < size; ++i) {
+        rejects = llama_grammar_reject_candidates_for_stack(rules, stacks[i], rejects);
+    }
+    return rejects;
+}
+
+//
+// grammar - external
+//
+
+struct llama_grammar * llama_grammar_init(
+            const llama_grammar_element ** rules,
+                                 size_t    n_rules,
+                                 size_t    start_rule_index) {
+    const llama_grammar_element * pos;
+
+    // copy rule definitions into vectors
+    std::vector<std::vector<llama_grammar_element>> vec_rules(n_rules);
+    for (size_t i = 0; i < n_rules; i++) {
+        for (pos = rules[i]; pos->type != LLAMA_GRETYPE_END; pos++) {
+            vec_rules[i].push_back(*pos);
+        }
+        vec_rules[i].push_back({LLAMA_GRETYPE_END, 0});
+    }
+
+    // loop over alternates of start rule to build initial stacks
+    std::vector<std::vector<const llama_grammar_element *>> stacks;
+    pos = rules[start_rule_index];
+    do {
+        std::vector<const llama_grammar_element *> stack;
+        if (!llama_grammar_is_end_of_sequence(pos)) {
+            // if alternate is nonempty, add to stack
+            stack.push_back(pos);
+        }
+        llama_grammar_advance_stack(vec_rules, stack, stacks);
+        while (!llama_grammar_is_end_of_sequence(pos)) {
+            // scan to end of alternate def
+            pos++;
+        }
+        if (pos->type == LLAMA_GRETYPE_ALT) {
+            // there's another alternate def of this rule to process
+            pos++;
+        } else {
+            break;
+        }
+    } while (true);
+
+    return new llama_grammar{ std::move(vec_rules), std::move(stacks) };
+}
+
+void llama_grammar_free(struct llama_grammar * grammar) {
+    delete grammar;
+}
+
+//
+// sampling
+//
+
+void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
+    assert(candidates->size > 0);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Sort the logits in descending order
+    if (!candidates->sorted) {
+        std::sort(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
+            return a.logit > b.logit;
+        });
+        candidates->sorted = true;
+    }
+
+    float max_l = candidates->data[0].logit;
+    float cum_sum = 0.0f;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        float p = expf(candidates->data[i].logit - max_l);
+        candidates->data[i].p = p;
+        cum_sum += p;
+    }
+    for (size_t i = 0; i < candidates->size; ++i) {
+        candidates->data[i].p /= cum_sum;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    k = std::max(k, (int) min_keep);
+    k = std::min(k, (int) candidates->size);
+
+    // Sort scores in descending order
+    if (!candidates->sorted) {
+        auto comp = [](const llama_token_data & a, const llama_token_data & b) {
+            return a.logit > b.logit;
+        };
+        if (k == (int) candidates->size) {
+            std::sort(candidates->data, candidates->data + candidates->size, comp);
+        } else {
+            std::partial_sort(candidates->data, candidates->data + k, candidates->data + candidates->size, comp);
+        }
+        candidates->sorted = true;
+    }
+    candidates->size = k;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
+    if (p >= 1.0f) {
+        return;
+    }
+
+    llama_sample_softmax(ctx, candidates);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Compute the cumulative probabilities
+    float cum_sum = 0.0f;
+    size_t last_idx = candidates->size;
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        cum_sum += candidates->data[i].p;
+
+        // Check if the running sum is at least p or if we have kept at least min_keep tokens
+        // we set the last index to i+1 to indicate that the current iterate should be included in the set
+        if (cum_sum >= p && i + 1 >= min_keep) {
+            last_idx = i + 1;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the top-p tokens
+    candidates->size = last_idx;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep) {
+    if (z >= 1.0f || candidates->size <= 2) {
+        return;
+    }
+
+    llama_sample_softmax(nullptr, candidates);
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Compute the first and second derivatives
+    std::vector<float> first_derivatives(candidates->size - 1);
+    std::vector<float> second_derivatives(candidates->size - 2);
+
+    for (size_t i = 0; i < first_derivatives.size(); ++i) {
+        first_derivatives[i] = candidates->data[i].p - candidates->data[i + 1].p;
+    }
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        second_derivatives[i] = first_derivatives[i] - first_derivatives[i + 1];
+    }
+
+    // Calculate absolute value of second derivatives
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        second_derivatives[i] = abs(second_derivatives[i]);
+    }
+
+    // Normalize the second derivatives
+    {
+        const float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
+
+        if (second_derivatives_sum > 1e-6f) {
+            for (float & value : second_derivatives) {
+                value /= second_derivatives_sum;
+            }
+        } else {
+            for (float & value : second_derivatives) {
+                value = 1.0f / second_derivatives.size();
+            }
+        }
+    }
+
+    float cum_sum = 0.0f;
+    size_t last_idx = candidates->size;
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        cum_sum += second_derivatives[i];
+
+        // Check if the running sum is greater than z or if we have kept at least min_keep tokens
+        if (cum_sum > z && i >= min_keep) {
+            last_idx = i;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the tokens above the tail location
+    candidates->size = last_idx;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+
+void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
+    // Reference implementation:
+    // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
+    if (p >= 1.0f) {
+        return;
+    }
+
+    // Compute the softmax of logits and calculate entropy
+    llama_sample_softmax(nullptr, candidates);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    float entropy = 0.0f;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        entropy += -candidates->data[i].p * logf(candidates->data[i].p);
+    }
+
+    // Compute the absolute difference between negative log probability and entropy for each candidate
+    std::vector<float> shifted_scores;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        float shifted_score = fabsf(-logf(candidates->data[i].p) - entropy);
+        shifted_scores.push_back(shifted_score);
+    }
+
+    // Sort tokens based on the shifted_scores and their corresponding indices
+    std::vector<size_t> indices(candidates->size);
+    std::iota(indices.begin(), indices.end(), 0);
+
+    std::sort(indices.begin(), indices.end(), [&](size_t a, size_t b) {
+        return shifted_scores[a] < shifted_scores[b];
+    });
+
+    // Compute the cumulative probabilities
+    float cum_sum = 0.0f;
+    size_t last_idx = indices.size();
+
+    for (size_t i = 0; i < indices.size(); ++i) {
+        size_t idx = indices[i];
+        cum_sum += candidates->data[idx].p;
+
+        // Check if the running sum is greater than typical or if we have kept at least min_keep tokens
+        if (cum_sum > p && i >= min_keep - 1) {
+            last_idx = i + 1;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the locally typical tokens
+    std::vector<llama_token_data> new_candidates;
+    for (size_t i = 0; i < last_idx; ++i) {
+        size_t idx = indices[i];
+        new_candidates.push_back(candidates->data[idx]);
+    }
+
+    // Replace the data in candidates with the new_candidates data
+    std::copy(new_candidates.begin(), new_candidates.end(), candidates->data);
+    candidates->size = new_candidates.size();
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates_p, float temp) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    for (size_t i = 0; i < candidates_p->size; ++i) {
+        candidates_p->data[i].logit /= temp;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty) {
+    if (last_tokens_size == 0 || penalty == 1.0f) {
+        return;
+    }
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        const auto * token_iter = std::find(last_tokens, last_tokens + last_tokens_size, candidates->data[i].id);
+        if (token_iter == last_tokens + last_tokens_size) {
+            continue;
+        }
+
+        // The academic publication that described this technique actually just only divided, but that would cause tokens with negative logits to become more likely, which is obviously wrong.
+        // This is common fix for this problem, which is to multiply by the penalty instead of dividing.
+        if (candidates->data[i].logit <= 0) {
+            candidates->data[i].logit *= penalty;
+        } else {
+            candidates->data[i].logit /= penalty;
+        }
+    }
+
+    candidates->sorted = false;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens_p, size_t last_tokens_size, float alpha_frequency, float alpha_presence) {
+    if (last_tokens_size == 0 || (alpha_frequency == 0.0f && alpha_presence == 0.0f)) {
+        return;
+    }
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Create a frequency map to count occurrences of each token in last_tokens
+    std::unordered_map<llama_token, int> token_count;
+    for (size_t i = 0; i < last_tokens_size; ++i) {
+        token_count[last_tokens_p[i]]++;
+    }
+
+    // Apply frequency and presence penalties to the candidates
+    for (size_t i = 0; i < candidates->size; ++i) {
+        auto token_iter = token_count.find(candidates->data[i].id);
+        if (token_iter == token_count.end()) {
+            continue;
+        }
+
+        int count = token_iter->second;
+        candidates->data[i].logit -= float(count) * alpha_frequency + float(count > 0) * alpha_presence;
+    }
+
+    candidates->sorted = false;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
+    assert(ctx);
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    bool allow_eos = false;
+    for (const auto & stack : grammar->stacks) {
+        if (stack.empty()) {
+            allow_eos = true;
+            break;
+        }
+    }
+
+    const llama_token eos = llama_token_eos();
+
+    std::vector<std::vector<uint32_t>>   candidates_decoded;
+    std::vector<llama_grammar_candidate> candidates_grammar;
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        const llama_token id  = candidates->data[i].id;
+        const char *      str = llama_token_to_str(ctx, id);
+        if (id == eos) {
+            if (!allow_eos) {
+                candidates->data[i].logit = -INFINITY;
+            }
+        } else if (*str == 0) {
+            candidates->data[i].logit = -INFINITY;
+        } else {
+            candidates_decoded.push_back(decode_utf8(str));
+            candidates_grammar.push_back({ i, candidates_decoded.back().data() });
+        }
+    }
+
+    const auto rejects =
+        llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
+    for (auto & reject : rejects) {
+        candidates->data[reject.index].logit = -INFINITY;
+    }
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+}
+
+static void llama_log_softmax(float * array, size_t size) {
+    float max_l = *std::max_element(array, array + size);
+    float sum = 0.f;
+    for (size_t i = 0; i < size; ++i) {
+        float p = expf(array[i] - max_l);
+        sum += p;
+        array[i] = p;
+    }
+
+    for (size_t i = 0; i < size; ++i) {
+        array[i] = logf(array[i] / sum);
+    }
+}
+
+void llama_sample_classifier_free_guidance(
+          struct llama_context * ctx,
+        llama_token_data_array * candidates,
+          struct llama_context * guidance_ctx,
+                         float   scale) {
+    int64_t t_start_sample_us = ggml_time_us();
+
+    assert(ctx);
+    auto n_vocab = llama_n_vocab(ctx);
+    assert(n_vocab == (int)candidates->size);
+    assert(!candidates->sorted);
+
+    std::vector<float> logits_base;
+    logits_base.reserve(candidates->size);
+    for (size_t i = 0; i < candidates->size; ++i) {
+        logits_base.push_back(candidates->data[i].logit);
+    }
+    llama_log_softmax(logits_base.data(), candidates->size);
+
+    float* logits_guidance = llama_get_logits(guidance_ctx);
+    llama_log_softmax(logits_guidance, n_vocab);
+
+    for (int i = 0; i < n_vocab; ++i) {
+        float logit_guidance = logits_guidance[i];
+        float logit_base = logits_base[i];
+        candidates->data[i].logit = scale * (logit_base - logit_guidance) + logit_guidance;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu) {
+    assert(ctx);
+    auto N = float(llama_n_vocab(ctx));
+    int64_t t_start_sample_us;
+    t_start_sample_us = ggml_time_us();
+
+    llama_sample_softmax(nullptr, candidates);
+
+    // Estimate s_hat using the most probable m tokens
+    float s_hat = 0.0;
+    float sum_ti_bi = 0.0;
+    float sum_ti_sq = 0.0;
+    for (size_t i = 0; i < size_t(m - 1) && i < candidates->size - 1; ++i) {
+        float t_i = logf(float(i + 2) / float(i + 1));
+        float b_i = logf(candidates->data[i].p / candidates->data[i + 1].p);
+        sum_ti_bi += t_i * b_i;
+        sum_ti_sq += t_i * t_i;
+    }
+    s_hat = sum_ti_bi / sum_ti_sq;
+
+    // Compute k from the estimated s_hat and target surprise value
+    float epsilon_hat = s_hat - 1;
+    float k = powf((epsilon_hat * powf(2, *mu)) / (1 - powf(N, -epsilon_hat)), 1 / s_hat);
+
+    // Sample the next word X using top-k sampling
+    llama_sample_top_k(nullptr, candidates, int(k), 1);
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    llama_token X = llama_sample_token(ctx, candidates);
+    t_start_sample_us = ggml_time_us();
+
+    // Compute error as the difference between observed surprise and target surprise value
+    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return candidate.id == X;
+    }));
+    float observed_surprise = -log2f(candidates->data[X_idx].p);
+    float e = observed_surprise - tau;
+
+    // Update mu using the learning rate and error
+    *mu = *mu - eta * e;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    return X;
+}
+
+llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu) {
+    int64_t t_start_sample_us;
+    t_start_sample_us = ggml_time_us();
+
+    llama_sample_softmax(ctx, candidates);
+
+    // Truncate the words with surprise values greater than mu
+    candidates->size = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return -log2f(candidate.p) > *mu;
+    }));
+
+    if (candidates->size == 0) {
+        candidates->size = 1;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+
+    // Normalize the probabilities of the remaining words
+    llama_sample_softmax(ctx, candidates);
+
+    // Sample the next word X from the remaining words
+    llama_token X = llama_sample_token(ctx, candidates);
+    t_start_sample_us = ggml_time_us();
+
+    // Compute error as the difference between observed surprise and target surprise value
+    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return candidate.id == X;
+    }));
+    float observed_surprise = -log2f(candidates->data[X_idx].p);
+    float e = observed_surprise - tau;
+
+    // Update mu using the learning rate and error
+    *mu = *mu - eta * e;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    return X;
+}
+
+llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Find max element
+    auto * max_iter = std::max_element(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
+        return a.logit < b.logit;
+    });
+
+    llama_token result = max_iter->id;
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+        ctx->n_sample++;
+    }
+    return result;
+}
+
+llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
+    assert(ctx);
+    const int64_t t_start_sample_us = ggml_time_us();
+    llama_sample_softmax(nullptr, candidates);
+
+    std::vector<float> probs;
+    probs.reserve(candidates->size);
+    for (size_t i = 0; i < candidates->size; ++i) {
+        probs.push_back(candidates->data[i].p);
+    }
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    auto & rng = ctx->rng;
+    int idx = dist(rng);
+
+    llama_token result = candidates->data[idx].id;
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    ctx->n_sample++;
+    return result;
+}
+
+void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    if (token == llama_token_eos()) {
+        for (const auto & stack : grammar->stacks) {
+            if (stack.empty()) {
+                return;
+            }
+        }
+        GGML_ASSERT(false);
+    }
+
+    const char * str = llama_token_to_str(ctx, token);
+    // Note terminating 0 in decoded string
+    auto code_points = decode_utf8(str);
+    for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
+        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
+    }
+    GGML_ASSERT(!grammar->stacks.empty());
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+}
+
+//
+// quantization
+//
+
+static void llama_convert_tensor_internal(const llama_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
+    if (output.size < nelements * sizeof(float)) {
+        output.resize(nelements * sizeof(float));
+    }
+    float * f32_output = (float *) output.addr;
+
+    ggml_type_traits_t qtype;
+    if (ggml_is_quantized(tensor.type)) {
+        qtype = ggml_internal_get_type_traits(tensor.type);
+        if (qtype.to_float == NULL) {
+            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor.type)));
+        }
+    } else if (tensor.type != GGML_TYPE_F16) {
+        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor.type)));
+    }
+
+    if (nthread < 2) {
+        if (tensor.type == GGML_TYPE_F16) {
+            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor.data, f32_output, nelements);
+        } else if (ggml_is_quantized(tensor.type)) {
+            qtype.to_float(tensor.data, f32_output, nelements);
+        } else {
+            GGML_ASSERT(false); // unreachable
+        }
+        return;
+    }
+
+    auto block_size = tensor.type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor.type);
+    auto block_size_bytes = ggml_type_size(tensor.type);
+
+    GGML_ASSERT(nelements % block_size == 0);
+    auto nblocks = nelements / block_size;
+    auto blocks_per_thread = nblocks / nthread;
+    auto spare_blocks = nblocks - (blocks_per_thread * nthread); // if blocks aren't divisible by thread count
+
+    std::vector<std::thread> workers;
+    for (auto tnum = 0, in_buff_offs = 0, out_buff_offs = 0; tnum < nthread; tnum++) {
+        auto thr_blocks = blocks_per_thread + (tnum == nthread - 1 ? spare_blocks : 0); // num blocks for this thread
+        auto thr_elems = thr_blocks * block_size; // number of elements for this thread
+        auto thr_block_bytes = thr_blocks * block_size_bytes; // number of input bytes for this thread
+
+        auto compute = [qtype] (ggml_type typ, uint8_t * inbuf, float * outbuf, int nels) {
+            if (typ == GGML_TYPE_F16) {
+                ggml_fp16_to_fp32_row((ggml_fp16_t *)inbuf, outbuf, nels);
+            } else {
+                qtype.to_float(inbuf, outbuf, nels);
+            }
+        };
+        workers.push_back(std::thread(compute, tensor.type, tensor.data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
+        in_buff_offs += thr_block_bytes;
+        out_buff_offs += thr_elems;
+    }
+    for (auto & worker : workers) {
+        worker.join();
+    }
+
+}
+
+static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
+    ggml_type quantized_type;
+    llama_ftype ftype = params->ftype;
+    int nthread = params->nthread;
+
+    switch (params->ftype) {
+        case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
+        case LLAMA_FTYPE_MOSTLY_Q4_1: quantized_type = GGML_TYPE_Q4_1; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_0: quantized_type = GGML_TYPE_Q5_0; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_1: quantized_type = GGML_TYPE_Q5_1; break;
+        case LLAMA_FTYPE_MOSTLY_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;
+        case LLAMA_FTYPE_MOSTLY_F16:  quantized_type = GGML_TYPE_F16;  break;
+        case LLAMA_FTYPE_ALL_F32:     quantized_type = GGML_TYPE_F32;  break;
+
+#ifdef GGML_USE_K_QUANTS
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K:   quantized_type = GGML_TYPE_Q2_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M:
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: quantized_type = GGML_TYPE_Q3_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: quantized_type = GGML_TYPE_Q4_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q6_K:   quantized_type = GGML_TYPE_Q6_K; break;
+#endif
+        default: throw std::runtime_error(format("invalid output file type %d\n", ftype));
+    }
+
+    if (nthread <= 0) {
+        nthread = std::thread::hardware_concurrency();
+    }
+
+    std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
+    gguf_file_saver file_saver(fname_out.c_str(), model_loader->file_loader.get(), params->ftype);
+
+#ifdef GGML_USE_K_QUANTS
+    int n_attention_wv    = 0;
+    int n_feed_forward_w2 = 0;
+    for (auto& tensor : model_loader->tensors_map.tensors) {
+        if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+            ++n_attention_wv;
+        }
+        else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+            ++n_feed_forward_w2;
+        }
+    }
+
+    int i_attention_wv = 0;
+    int i_feed_forward_w2 = 0;
+#endif
+
+    size_t total_size_org = 0;
+    size_t total_size_new = 0;
+    std::vector<int64_t> hist_all(1 << 4, 0);
+
+    std::vector<std::thread> workers;
+    std::mutex mutex;
+
+    auto use_more_bits = [] (int i_layer, int num_layers) -> bool {
+        return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
+    };
+
+    size_t idx = 0;
+    for (llama_load_tensor & tensor : model_loader->tensors_map.tensors) {
+        gguf_buffer read_data;
+        read_data.resize(tensor.size);
+        tensor.data = read_data.addr;
+        model_loader->load_data_for(tensor);
+
+        printf("[%4zu/%4zu] %36s - %16s, type = %6s, ",
+               ++idx, model_loader->tensors_map.tensors.size(),
+               tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
+               ggml_type_name(tensor.type));
+
+        // This used to be a regex, but <regex> has an extreme cost to compile times.
+        bool quantize = tensor.name.rfind("weight") == tensor.name.size() - 6; // ends with 'weight'?
+
+        // quantize only 2D tensors
+        quantize &= (tensor.ne.size() == 2);
+        quantize &= params->quantize_output_tensor || tensor.name != "output.weight";
+        quantize &= quantized_type != tensor.type;
+
+        enum ggml_type new_type;
+        void * new_data;
+        size_t new_size;
+        gguf_buffer work;
+
+        if (!quantize) {
+            new_type = tensor.type;
+            new_data = tensor.data;
+            new_size = tensor.size;
+            printf("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
+        } else {
+            new_type = quantized_type;
+#ifdef GGML_USE_K_QUANTS
+            if (tensor.name == "output.weight") {
+                int nx = tensor.ne.at(0);
+                int ny = tensor.ne.at(1);
+                if (nx % QK_K == 0 && ny % QK_K == 0) {
+                    new_type = GGML_TYPE_Q6_K;
+                }
+            } else if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
+                        use_more_bits(i_attention_wv, n_attention_wv)) new_type = GGML_TYPE_Q6_K;
+                else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
+                        (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
+                ++i_attention_wv;
+            } else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
+                         use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
+                //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
+                ++i_feed_forward_w2;
+            } else if (tensor.name.find("attention.wo.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+            }
+            bool convert_incompatible_tensor = false;
+            if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
+                new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
+                int nx = tensor.ne.at(0);
+                int ny = tensor.ne.at(1);
+                if (nx % QK_K != 0 || ny % QK_K != 0) {
+                    fprintf(stderr, "\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
+                    convert_incompatible_tensor = true;
+                }
+            }
+            if (convert_incompatible_tensor) {
+                if (tensor.name == "output.weight") {
+                    new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
+                    fprintf(stderr, "F16 will be used for this tensor instead.\n");
+                } else if (tensor.name == "tok_embeddings.weight") {
+                    new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
+                    fprintf(stderr, "Q4_0 will be used for this tensor instead.\n");
+                } else {
+                    throw std::runtime_error("Unsupported tensor size encountered\n");
+                }
+            }
+#endif
+
+            float * f32_data;
+            size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
+            gguf_buffer f32_conv_buf;
+
+            if (tensor.type == GGML_TYPE_F32) {
+                f32_data = (float *) tensor.data;
+            } else if (ggml_is_quantized(tensor.type) && !params->allow_requantize) {
+                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
+            } else {
+                llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
+                f32_data = (float *) f32_conv_buf.addr;
+            }
+
+            printf("quantizing to %s .. ", ggml_type_name(new_type));
+            fflush(stdout);
+
+            work.resize(nelements * 4); // upper bound on size
+            new_data = work.addr;
+            std::vector<int64_t> hist_cur(1 << 4, 0);
+
+            int chunk_size = 32 * 512;
+            const int nchunk = (nelements + chunk_size - 1)/chunk_size;
+            const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
+            if (nthread_use < 2) {
+                new_size = ggml_quantize_chunk(new_type, f32_data, new_data, 0, nelements, hist_cur.data());
+            } else {
+                size_t counter = 0;
+                new_size = 0;
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size] () {
+                    std::vector<int64_t> local_hist;
+                    size_t local_size = 0;
+                    while (true) {
+                        std::unique_lock<std::mutex> lock(mutex);
+                        size_t first = counter; counter += chunk_size;
+                        if (first >= nelements) {
+                            if (!local_hist.empty()) {
+                                for (int j=0; j<int(local_hist.size()); ++j) {
+                                    hist_cur[j] += local_hist[j];
+                                }
+                                new_size += local_size;
+                            }
+                            break;
+                        }
+                        lock.unlock();
+                        size_t last = std::min(nelements, first + chunk_size);
+                        if (local_hist.empty()) {
+                            local_hist.resize(hist_cur.size(), 0);
+                        }
+                        local_size += ggml_quantize_chunk(new_type, f32_data, new_data, first, last - first, local_hist.data());
+                    }
+                };
+                if ((int) workers.size() < nthread_use - 1) {
+                    workers.resize(nthread_use - 1);
+                }
+                for (int it = 0; it < nthread_use - 1; ++it) {
+                    workers[it] = std::thread(compute);
+                }
+                compute();
+                for (int it = 0; it < nthread_use - 1; ++it) {
+                    workers[it].join();
+                }
+            }
+
+            printf("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
+            int64_t tot_count = 0;
+            for (size_t i = 0; i < hist_cur.size(); i++) {
+                hist_all[i] += hist_cur[i];
+                tot_count += hist_cur[i];
+            }
+
+            if (tot_count > 0) {
+                for (size_t i = 0; i < hist_cur.size(); i++) {
+                    printf("%5.3f ", hist_cur[i] / float(nelements));
+                }
+            }
+            printf("\n");
+        }
+        total_size_org += tensor.size;
+        total_size_new += new_size;
+        file_saver.write_tensor(tensor, new_type, new_data, new_size);
+    }
+
+    printf("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
+    printf("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
+
+    {
+        int64_t sum_all = 0;
+        for (size_t i = 0; i < hist_all.size(); i++) {
+            sum_all += hist_all[i];
+        }
+
+        if (sum_all > 0) {
+            printf("%s: hist: ", __func__);
+            for (size_t i = 0; i < hist_all.size(); i++) {
+                printf("%5.3f ", hist_all[i] / float(sum_all));
+            }
+            printf("\n");
+        }
+    }
+}
+
+
+
+//
+// interface implementation
+//
+
+struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+    ggml_time_init();
+
+    llama_model * model = new llama_model;
+
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+
+    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gqa, params.rms_norm_eps, params.n_gpu_layers,
+                params.main_gpu, params.tensor_split, params.rope_freq_base, params.rope_freq_scale,params.low_vram,
+                memory_type, params.use_mmap, params.use_mlock, params.vocab_only, params.progress_callback,
+                params.progress_callback_user_data)) {
+        delete model;
+        fprintf(stderr, "%s: failed to load model\n", __func__);
+        return nullptr;
+    }
+
+    return model;
+}
+
+void llama_free_model(struct llama_model * model) {
+    delete model;
+}
+
+struct llama_context * llama_new_context_with_model(
+                 struct llama_model * model,
+        struct llama_context_params   params) {
+
+    if (!model) {
+        return nullptr;
+    }
+
+    llama_context * ctx = new llama_context(*model);
+
+    if (params.seed == LLAMA_DEFAULT_SEED) {
+        params.seed = time(NULL);
+    }
+
+    unsigned cur_percentage = 0;
+    if (params.progress_callback == NULL) {
+        params.progress_callback_user_data = &cur_percentage;
+        params.progress_callback = [](float progress, void * ctx) {
+            unsigned * cur_percentage_p = (unsigned *) ctx;
+            unsigned percentage = (unsigned) (100 * progress);
+            while (percentage > *cur_percentage_p) {
+                *cur_percentage_p = percentage;
+                fprintf(stderr, ".");
+                fflush(stderr);
+                if (percentage >= 100) {
+                    fprintf(stderr, "\n");
+                }
+            }
+        };
+    }
+
+    ctx->rng = std::mt19937(params.seed);
+    ctx->logits_all = params.logits_all;
+
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+
+    // reserve memory for context buffers
+    if (!params.vocab_only) {
+        if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
+            fprintf(stderr, "%s: kv_cache_init() failed for self-attention cache\n", __func__);
+            llama_free(ctx);
+            return nullptr;
+        }
+
+        {
+            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
+            fprintf(stderr, "%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
+        }
+
+        const auto & hparams = ctx->model.hparams;
+
+        // resized during inference
+        if (params.logits_all) {
+            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
+        } else {
+            ctx->logits.reserve(hparams.n_vocab);
+        }
+
+        if (params.embedding){
+            ctx->embedding.resize(hparams.n_embd);
+        }
+
+        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
+
+        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
+        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
+    }
+
+#ifdef GGML_USE_METAL
+    if (params.n_gpu_layers > 0) {
+        // this allocates all Metal resources and memory buffers
+        ctx->ctx_metal = ggml_metal_init(1);
+
+        void * data_ptr  = NULL;
+        size_t data_size = 0;
+
+        if (params.use_mmap) {
+            data_ptr  = ctx->model.mapping->addr;
+            data_size = ctx->model.mapping->size;
+        } else {
+            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
+            data_size = ggml_get_mem_size  (ctx->model.ctx);
+        }
+
+        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
+
+        fprintf(stderr, "%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
+
+#define LLAMA_METAL_CHECK_BUF(result)                                          \
+    if (!(result)) {                                                           \
+        fprintf(stderr, "%s: failed to add buffer\n", __func__);               \
+        llama_free(ctx);                                                       \
+        return NULL;                                                           \
+    }
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.addr, ctx->buf_compute.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.addr, ctx->kv_self.buf.size, 0));
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].addr, ctx->buf_scratch[0].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].addr, ctx->buf_scratch[1].size, 0));
+#undef LLAMA_METAL_CHECK_BUF
+    }
+#endif
+
+#ifdef GGML_USE_MPI
+    ctx->ctx_mpi = ggml_mpi_init();
+
+    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
+        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
+        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
+        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
+        llama_backend_free();
+        exit(1);
+    }
+#endif
+
+    return ctx;
+}
+
+struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+
+    struct llama_model * model = llama_load_model_from_file(path_model, params);
+    if (!model) {
+        return nullptr;
+    }
+    struct llama_context * ctx = llama_new_context_with_model(model, params);
+    ctx->model_owner = true;
+    return ctx;
+}
+
+void llama_free(struct llama_context * ctx) {
+    if (ctx->model_owner) {
+        delete &ctx->model;
+    }
+    delete ctx;
+}
+
+int llama_model_quantize(
+        const char * fname_inp,
+        const char * fname_out,
+        const llama_model_quantize_params *params) {
+    try {
+        llama_model_quantize_internal(fname_inp, fname_out, params);
+        return 0;
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to quantize: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
+    fprintf(stderr, "%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
+
+    const int64_t t_start_lora_us = ggml_time_us();
+
+    auto fin = std::ifstream(path_lora, std::ios::binary);
+    if (!fin) {
+        fprintf(stderr, "%s: failed to open '%s'\n", __func__, path_lora);
+        return 1;
+    }
+
+    // verify magic and version
+    {
+        uint32_t magic;
+        fin.read((char *) &magic, sizeof(magic));
+        if (magic != LLAMA_FILE_MAGIC_GGLA) {
+            fprintf(stderr, "%s: bad file magic\n", __func__);
+            return 1;
+        }
+        uint32_t format_version;
+        fin.read((char *) &format_version, sizeof(format_version));
+
+        if (format_version != 1) {
+            fprintf(stderr, "%s: unsupported file version\n", __func__ );
+            return 1;
+        }
+    }
+
+    int32_t lora_r;
+    int32_t lora_alpha;
+    fin.read((char *) &lora_r, sizeof(lora_r));
+    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
+    float scaling = (float)lora_alpha / (float)lora_r;
+
+    fprintf(stderr, "%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
+
+
+    // create a temporary ggml context to store the lora tensors
+    // todo: calculate size from biggest possible tensor
+    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
+    struct ggml_init_params params;
+    params.mem_size   = lora_buf.size();
+    params.mem_buffer = lora_buf.data();
+    params.no_alloc   = false;
+
+    ggml_context * lora_ctx = ggml_init(params);
+    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+
+    // create a name -> tensor map of the model to accelerate lookups
+    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
+    for (const auto & kv: model.tensors_by_name) {
+        model_tensors.insert(kv);
+    }
+
+
+    // load base model
+    std::unique_ptr<llama_model_loader> model_loader;
+    ggml_context * base_ctx = NULL;
+    gguf_buffer base_buf;
+    if (path_base_model) {
+        fprintf(stderr, "%s: loading base model from '%s'\n", __func__, path_base_model);
+        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
+
+        size_t ctx_size;
+        size_t mmapped_size;
+        model_loader->calc_sizes(&ctx_size, &mmapped_size);
+        base_buf.resize(ctx_size);
+
+        ggml_init_params base_params;
+        base_params.mem_size   = base_buf.size;
+        base_params.mem_buffer = base_buf.addr;
+        base_params.no_alloc   = model_loader->use_mmap;
+
+        base_ctx = ggml_init(base_params);
+
+        model_loader->ggml_ctx = base_ctx;
+
+        // maybe this should in llama_model_loader
+        if (model_loader->use_mmap) {
+            model_loader->mapping.reset(new gguf_mmap(&model_loader->file_loader->file, /* prefetch */ 0, ggml_is_numa()));
+        }
+    }
+
+    // read tensors and apply
+    bool warned = false;
+    int n_tensors = 0;
+
+    std::vector<uint8_t> work_buffer;
+
+    while (true) {
+        int32_t n_dims;
+        int32_t length;
+        int32_t ftype;
+
+        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
+        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
+        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
+        if (fin.eof()) {
+            break;
+        }
+
+        int32_t ne[2] = { 1, 1 };
+        for (int i = 0; i < n_dims; ++i) {
+            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
+        }
+
+        std::string name;
+        {
+            char buf[1024];
+            fin.read(buf, length);
+            name = std::string(buf, length);
+        }
+
+        // check for lora suffix and get the type of tensor
+        const std::string lora_suffix = ".lora";
+        size_t pos = name.rfind(lora_suffix);
+        if (pos == std::string::npos) {
+            fprintf(stderr, "%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
+            return 1;
+        }
+
+        std::string lora_type = name.substr(pos + lora_suffix.length());
+        std::string base_name = name;
+        base_name.erase(pos);
+        // fprintf(stderr, "%s: %s => %s (lora type %s) ", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+
+        if (model_tensors.find(base_name) == model_tensors.end()) {
+            fprintf(stderr, "%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
+            return 1;
+        }
+
+        // create ggml tensor
+        ggml_type wtype;
+        switch (ftype) {
+            case 0: wtype = GGML_TYPE_F32;  break;
+            case 1: wtype = GGML_TYPE_F16;  break;
+            default:
+                    {
+                        fprintf(stderr, "%s: invalid tensor data type '%d'\n",
+                                __func__, ftype);
+                        return false;
+                    }
+        }
+        ggml_tensor * lora_tensor;
+        if (n_dims == 2) {
+            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
+        }
+        else {
+            fprintf(stderr, "%s: unsupported tensor dimension %d\n", __func__, n_dims);
+            return 1;
+        }
+        ggml_set_name(lora_tensor, "lora_tensor");
+
+        // load tensor data
+        size_t offset = fin.tellg();
+        size_t tensor_data_size = ggml_nbytes(lora_tensor);
+        offset = (offset + 31) & -32;
+        fin.seekg(offset);
+        fin.read((char*)lora_tensor->data, tensor_data_size);
+
+        lora_tensors[name] = lora_tensor;
+
+        // check if we have both A and B tensors and apply
+        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
+            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
+
+            ggml_tensor * dest_t = model_tensors[base_name];
+
+            offload_func_t offload_func = llama_nop;
+            offload_func_t offload_func_force_inplace = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
+                if (dest_t->type != GGML_TYPE_F16) {
+                    throw std::runtime_error(format(
+                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
+                }
+                offload_func = ggml_cuda_assign_buffers;
+                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
+            }
+#endif // GGML_USE_CUBLAS
+
+            ggml_tensor * base_t;
+            if (model_loader) {
+                // load from base model
+                if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
+                    fprintf(stderr, "%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
+                    return 1;
+                }
+                size_t idx = model_loader->tensors_map.name_to_idx[base_name];
+                llama_load_tensor & lt = model_loader->tensors_map.tensors[idx];
+                base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
+                lt.data = (uint8_t *) lt.ggml_tensor->data;
+                model_loader->load_data_for(lt);
+                lt.ggml_tensor->data = lt.data;
+            }
+            else {
+                base_t = dest_t;
+            }
+
+            if (ggml_is_quantized(base_t->type)) {
+                if (!warned) {
+                    fprintf(stderr, "%s: warning: using a lora adapter with a quantized model may result in poor quality, "
+                                    "use a f16 or f32 base model with --lora-base\n", __func__);
+                    warned = true;
+                }
+            }
+
+            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
+            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
+            ggml_set_name(loraA, "loraA");
+
+            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
+            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
+            ggml_set_name(loraB, "loraB");
+
+            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
+                fprintf(stderr, "%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
+                               " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
+                return 1;
+            }
+
+            // w = w + BA*s
+            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
+            offload_func(BA);
+            ggml_set_name(BA, "BA");
+
+            if (scaling != 1.0f) {
+                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
+                ggml_set_name(scale_tensor, "scale_tensor");
+
+                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
+                offload_func(BA);
+                ggml_set_name(BA, "BA_scaled");
+            }
+
+            ggml_tensor * r;
+            if (base_t == dest_t) {
+                r = ggml_add_inplace(lora_ctx, dest_t, BA);
+                offload_func_force_inplace(r);
+                ggml_set_name(r, "r_add_inplace");
+            }
+            else {
+                r = ggml_add(lora_ctx, base_t, BA);
+                offload_func(r);
+                ggml_set_name(r, "r_add");
+
+                r = ggml_cpy(lora_ctx, r, dest_t);
+                offload_func(r);
+                ggml_set_name(r, "r_cpy");
+            }
+
+            struct ggml_cgraph gf = ggml_build_forward(r);
+
+            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
+
+            // we won't need these tensors again, reset the context to save memory
+            ggml_free(lora_ctx);
+            lora_ctx = ggml_init(params);
+            lora_tensors.clear();
+
+            n_tensors++;
+            if (n_tensors % 4 == 0) {
+                fprintf(stderr, ".");
+            }
+        }
+    }
+
+    // TODO: this should be in a destructor, it will leak on failure
+    ggml_free(lora_ctx);
+    if (base_ctx) {
+        ggml_free(base_ctx);
+    }
+
+    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
+    fprintf(stderr, " done (%.2f ms)\n", t_lora_us / 1000.0);
+
+    return 0;
+}
+
+int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
+    try {
+        return llama_apply_lora_from_file_internal(ctx->model, path_lora, path_base_model, n_threads);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_model_apply_lora_from_file(const struct llama_model * model, const char * path_lora, const char * path_base_model, int n_threads) {
+    try {
+        return llama_apply_lora_from_file_internal(*model, path_lora, path_base_model, n_threads);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_get_kv_cache_token_count(const struct llama_context * ctx) {
+    return ctx->kv_self.n;
+}
+
+#define LLAMA_MAX_RNG_STATE (64*1024)
+
+void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed) {
+    if (seed == LLAMA_DEFAULT_SEED) {
+        seed = time(NULL);
+    }
+    ctx->rng.seed(seed);
+}
+
+// Returns the *maximum* size of the state
+size_t llama_get_state_size(const struct llama_context * ctx) {
+    // we don't know size of rng until we actually serialize it. so reserve more than enough memory for its serialized state.
+    // for reference, std::mt19937(1337) serializes to 6701 bytes.
+    const size_t s_rng_size        = sizeof(size_t);
+    const size_t s_rng             = LLAMA_MAX_RNG_STATE;
+    const size_t s_logits_capacity = sizeof(size_t);
+    const size_t s_logits_size     = sizeof(size_t);
+    const size_t s_logits          = ctx->logits.capacity() * sizeof(float);
+    const size_t s_embedding_size  = sizeof(size_t);
+    const size_t s_embedding       = ctx->embedding.size() * sizeof(float);
+    const size_t s_kv_size         = sizeof(size_t);
+    const size_t s_kv_ntok         = sizeof(int);
+    const size_t s_kv              = ctx->kv_self.buf.size;
+
+    const size_t s_total = (
+        + s_rng_size
+        + s_rng
+        + s_logits_capacity
+        + s_logits_size
+        + s_logits
+        + s_embedding_size
+        + s_embedding
+        + s_kv_size
+        + s_kv_ntok
+        + s_kv
+    );
+
+    return s_total;
+}
+
+// Copies the state to the specified destination address
+size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
+    uint8_t * out = dst;
+
+    // copy rng
+    {
+        std::stringstream rng_ss;
+        rng_ss << ctx->rng;
+
+        const size_t rng_size = rng_ss.str().size();
+        char rng_buf[LLAMA_MAX_RNG_STATE];
+
+        memset(&rng_buf[0], 0, LLAMA_MAX_RNG_STATE);
+        memcpy(&rng_buf[0], rng_ss.str().data(), rng_ss.str().size());
+
+        memcpy(out, &rng_size,   sizeof(rng_size));    out += sizeof(rng_size);
+        memcpy(out, &rng_buf[0], LLAMA_MAX_RNG_STATE); out += LLAMA_MAX_RNG_STATE;
+    }
+
+    // copy logits
+    {
+        const size_t logits_cap  = ctx->logits.capacity();
+        const size_t logits_size = ctx->logits.size();
+
+        memcpy(out, &logits_cap,  sizeof(logits_cap));  out += sizeof(logits_cap);
+        memcpy(out, &logits_size, sizeof(logits_size)); out += sizeof(logits_size);
+
+        if (logits_size) {
+            memcpy(out, ctx->logits.data(), logits_size * sizeof(float));
+        }
+
+        out += logits_cap * sizeof(float);
+    }
+
+    // copy embeddings
+    {
+        const size_t embedding_size = ctx->embedding.size();
+
+        memcpy(out, &embedding_size, sizeof(embedding_size)); out += sizeof(embedding_size);
+
+        if (embedding_size) {
+            memcpy(out, ctx->embedding.data(), embedding_size * sizeof(float));
+            out += embedding_size * sizeof(float);
+        }
+    }
+
+    // copy kv cache
+    {
+        const auto & kv_self = ctx->kv_self;
+        const auto & hparams = ctx->model.hparams;
+        const int    n_layer = hparams.n_layer;
+        const int    n_embd  = hparams.n_embd;
+        const int    n_ctx   = hparams.n_ctx;
+
+        const size_t kv_size = kv_self.buf.size;
+        const int    kv_ntok = llama_get_kv_cache_token_count(ctx);
+
+        memcpy(out, &kv_size, sizeof(kv_size)); out += sizeof(kv_size);
+        memcpy(out, &kv_ntok, sizeof(kv_ntok)); out += sizeof(kv_ntok);
+
+        if (kv_size) {
+            const size_t elt_size = ggml_element_size(kv_self.k);
+
+            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
+            ggml_cgraph gf{};
+
+            ggml_tensor * kout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
+            kout3d->data = out;
+            out += ggml_nbytes(kout3d);
+
+            ggml_tensor * vout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
+            vout3d->data = out;
+            out += ggml_nbytes(vout3d);
+
+            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
+                n_embd, kv_ntok, n_layer,
+                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
+
+            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
+                kv_ntok, n_embd, n_layer,
+                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
+
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, k3d, kout3d));
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, v3d, vout3d));
+            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
+
+            ggml_free(cpy_ctx);
+        }
+    }
+
+    const size_t written  = out - dst;
+    const size_t max_size = llama_get_state_size(ctx);
+
+    GGML_ASSERT(written <= max_size);
+
+    return written;
+}
+
+// Sets the state reading from the specified source address
+size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
+    uint8_t * inp = src;
+
+    // set rng
+    {
+        size_t rng_size;
+        char   rng_buf[LLAMA_MAX_RNG_STATE];
+
+        memcpy(&rng_size,   inp, sizeof(rng_size));    inp += sizeof(rng_size);
+        memcpy(&rng_buf[0], inp, LLAMA_MAX_RNG_STATE); inp += LLAMA_MAX_RNG_STATE;
+
+        std::stringstream rng_ss;
+        rng_ss.str(std::string(&rng_buf[0], rng_size));
+        rng_ss >> ctx->rng;
+
+        GGML_ASSERT(rng_ss.fail() == false);
+    }
+
+    // set logits
+    {
+        size_t logits_cap;
+        size_t logits_size;
+
+        memcpy(&logits_cap,  inp, sizeof(logits_cap));  inp += sizeof(logits_cap);
+        memcpy(&logits_size, inp, sizeof(logits_size)); inp += sizeof(logits_size);
+
+        GGML_ASSERT(ctx->logits.capacity() == logits_cap);
+
+        if (logits_size) {
+            ctx->logits.resize(logits_size);
+            memcpy(ctx->logits.data(), inp, logits_size * sizeof(float));
+        }
+
+        inp += logits_cap * sizeof(float);
+    }
+
+    // set embeddings
+    {
+        size_t embedding_size;
+
+        memcpy(&embedding_size, inp, sizeof(embedding_size)); inp += sizeof(embedding_size);
+
+        GGML_ASSERT(ctx->embedding.capacity() == embedding_size);
+
+        if (embedding_size) {
+            memcpy(ctx->embedding.data(), inp, embedding_size * sizeof(float));
+            inp += embedding_size * sizeof(float);
+        }
+    }
+
+    // set kv cache
+    {
+        const auto & kv_self = ctx->kv_self;
+        const auto & hparams = ctx->model.hparams;
+        const int    n_layer = hparams.n_layer;
+        const int    n_embd  = hparams.n_embd;
+        const int    n_ctx   = hparams.n_ctx;
+
+        size_t kv_size;
+        int kv_ntok;
+
+        memcpy(&kv_size, inp, sizeof(kv_size)); inp += sizeof(kv_size);
+        memcpy(&kv_ntok, inp, sizeof(kv_ntok)); inp += sizeof(kv_ntok);
+
+        if (kv_size) {
+            GGML_ASSERT(kv_self.buf.size == kv_size);
+
+            const size_t elt_size = ggml_element_size(kv_self.k);
+
+            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
+            ggml_cgraph gf{};
+
+            ggml_tensor * kin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
+            kin3d->data = (void *) inp;
+            inp += ggml_nbytes(kin3d);
+
+            ggml_tensor * vin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
+            vin3d->data = (void *) inp;
+            inp += ggml_nbytes(vin3d);
+
+            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
+                n_embd, kv_ntok, n_layer,
+                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
+
+            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
+                kv_ntok, n_embd, n_layer,
+                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
+
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, kin3d, k3d));
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, vin3d, v3d));
+            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
+
+            ggml_free(cpy_ctx);
+        }
+
+        ctx->kv_self.n = kv_ntok;
+    }
+
+    const size_t nread    = inp - src;
+    const size_t max_size = llama_get_state_size(ctx);
+
+    GGML_ASSERT(nread <= max_size);
+
+    return nread;
+}
+
+static bool llama_load_session_file_internal(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
+    gguf_file file(path_session, "rb");
+    GGML_UNUSED(ctx);
+    GGML_UNUSED(path_session);
+    GGML_UNUSED(tokens_out);
+    GGML_UNUSED(n_token_capacity);
+    GGML_UNUSED(n_token_count_out);
+
+
+// TODO: implement with GGUF format
+    return true;
+}
+
+bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
+    try {
+        return llama_load_session_file_internal(ctx, path_session, tokens_out, n_token_capacity, n_token_count_out);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "error loading session file: %s\n", err.what());
+        return false;
+    }
+}
+
+bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
+    gguf_file file(path_session, "wb");
+
+    // TODO: implement with GGUF format
+
+    return true;
+}
+
+int llama_eval(
+        struct llama_context * ctx,
+           const llama_token * tokens,
+                         int   n_tokens,
+                         int   n_past,
+                         int   n_threads) {
+    if (!llama_eval_internal(*ctx, tokens, nullptr, n_tokens, n_past, n_threads, nullptr)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    // get a more accurate load time, upon first eval
+    // TODO: fix this
+    if (!ctx->has_evaluated_once) {
+        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
+        ctx->has_evaluated_once = true;
+    }
+
+    return 0;
+}
+
+
+int llama_eval_embd(
+            struct llama_context * ctx,
+                     const float * embd,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads) {
+    if (!llama_eval_internal(*ctx, nullptr, embd, n_tokens, n_past, n_threads, nullptr)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    // get a more accurate load time, upon first eval
+    // TODO: fix this
+    if (!ctx->has_evaluated_once) {
+        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
+        ctx->has_evaluated_once = true;
+    }
+
+    return 0;
+}
+
+int llama_eval_export(struct llama_context * ctx, const char * fname) {
+    const int n_batch = 1;
+    const int n_ctx   = 512 - n_batch;
+
+    const std::vector<llama_token> tmp(n_batch, llama_token_bos());
+
+    if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    return 0;
+}
+
+int llama_tokenize_with_model(
+    const struct llama_model * model,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    auto res = llama_tokenize(model->vocab, text, add_bos);
+
+    if (n_max_tokens < (int) res.size()) {
+        fprintf(stderr, "%s: too many tokens\n", __func__);
+        return -((int) res.size());
+    }
+
+    for (size_t i = 0; i < res.size(); i++) {
+        tokens[i] = res[i];
+    }
+
+    return res.size();
+}
+
+int llama_tokenize(
+        struct llama_context * ctx,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
+}
+
+int llama_n_vocab_from_model(const struct llama_model * model) {
+    return model->vocab.id_to_token.size();
+}
+
+int llama_n_ctx_from_model(const struct llama_model * model) {
+    return model->hparams.n_ctx;
+}
+
+int llama_n_embd_from_model(const struct llama_model * model) {
+    return model->hparams.n_embd;
+}
+
+int llama_n_vocab(const struct llama_context * ctx) {
+    return ctx->model.vocab.id_to_token.size();
+}
+
+int llama_n_ctx(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_ctx;
+}
+
+int llama_n_embd(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_embd;
+}
+
+int llama_get_vocab_from_model(
+        const struct llama_model * model,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
+    for (int i = 0; i<n; ++i) {
+        strings[i] = model->vocab.id_to_token[i].tok.c_str();
+        scores[i]  = model->vocab.id_to_token[i].score;
+    }
+    return n;
+}
+
+int llama_get_vocab(
+        const struct llama_context * ctx,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
+}
+
+float * llama_get_logits(struct llama_context * ctx) {
+    return ctx->logits.data();
+}
+
+float * llama_get_embeddings(struct llama_context * ctx) {
+    return ctx->embedding.data();
+}
+
+const char * llama_token_to_str_with_model(const struct llama_model * model, llama_token token) {
+    if (token >= llama_n_vocab_from_model(model)) {
+        return nullptr;
+    }
+
+    return model->vocab.id_to_token[token].tok.c_str();
+}
+
+const char * llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    return llama_token_to_str_with_model(&ctx->model, token);
+}
+
+llama_token llama_token_bos() {
+    return 1;
+}
+
+llama_token llama_token_eos() {
+    return 2;
+}
+
+llama_token llama_token_nl() {
+    return 13;
+}
+
+struct llama_timings llama_get_timings(struct llama_context * ctx) {
+    struct llama_timings result = {
+        /*.t_start_ms  =*/ 1e-3 * ctx->t_start_us,
+        /*.t_end_ms    =*/ 1.00 * ggml_time_ms(),
+        /*.t_load_ms   =*/ 1e-3 * ctx->t_load_us,
+        /*.t_sample_ms =*/ 1e-3 * ctx->t_sample_us,
+        /*.t_p_eval_ms =*/ 1e-3 * ctx->t_p_eval_us,
+        /*.t_eval_ms   =*/ 1e-3 * ctx->t_eval_us,
+
+        /*.n_sample =*/ std::max(1, ctx->n_sample),
+        /*.n_p_eval =*/ std::max(1, ctx->n_p_eval),
+        /*.n_eval   =*/ std::max(1, ctx->n_eval),
+    };
+
+    return result;
+}
+
+void llama_print_timings(struct llama_context * ctx) {
+    const llama_timings timings = llama_get_timings(ctx);
+
+    fprintf(stderr, "\n");
+    fprintf(stderr, "%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
+    fprintf(stderr, "%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_sample_ms, timings.n_sample, timings.t_sample_ms / timings.n_sample, 1e3 / timings.t_sample_ms * timings.n_sample);
+    fprintf(stderr, "%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_p_eval_ms, timings.n_p_eval, timings.t_p_eval_ms / timings.n_p_eval, 1e3 / timings.t_p_eval_ms * timings.n_p_eval);
+    fprintf(stderr, "%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_eval_ms, timings.n_eval, timings.t_eval_ms / timings.n_eval, 1e3 / timings.t_eval_ms * timings.n_eval);
+    fprintf(stderr, "%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
+}
+
+void llama_reset_timings(struct llama_context * ctx) {
+    ctx->t_start_us = ggml_time_us();
+    ctx->t_sample_us = ctx->n_sample = 0;
+    ctx->t_eval_us   = ctx->n_eval   = 0;
+    ctx->t_p_eval_us = ctx->n_p_eval = 0;
+}
+
+const char * llama_print_system_info(void) {
+    static std::string s;
+
+    s  = "";
+    s += "AVX = "         + std::to_string(ggml_cpu_has_avx())         + " | ";
+    s += "AVX2 = "        + std::to_string(ggml_cpu_has_avx2())        + " | ";
+    s += "AVX512 = "      + std::to_string(ggml_cpu_has_avx512())      + " | ";
+    s += "AVX512_VBMI = " + std::to_string(ggml_cpu_has_avx512_vbmi()) + " | ";
+    s += "AVX512_VNNI = " + std::to_string(ggml_cpu_has_avx512_vnni()) + " | ";
+    s += "FMA = "         + std::to_string(ggml_cpu_has_fma())         + " | ";
+    s += "NEON = "        + std::to_string(ggml_cpu_has_neon())        + " | ";
+    s += "ARM_FMA = "     + std::to_string(ggml_cpu_has_arm_fma())     + " | ";
+    s += "F16C = "        + std::to_string(ggml_cpu_has_f16c())        + " | ";
+    s += "FP16_VA = "     + std::to_string(ggml_cpu_has_fp16_va())     + " | ";
+    s += "WASM_SIMD = "   + std::to_string(ggml_cpu_has_wasm_simd())   + " | ";
+    s += "BLAS = "        + std::to_string(ggml_cpu_has_blas())        + " | ";
+    s += "SSE3 = "        + std::to_string(ggml_cpu_has_sse3())        + " | ";
+    s += "VSX = "         + std::to_string(ggml_cpu_has_vsx())         + " | ";
+
+    return s.c_str();
+}
+
+// For internal test use
+const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
+    return ctx->model.tensors_by_name;
+}
diff --git a/gguf-llama.h b/gguf-llama.h
new file mode 100644
index 0000000000000..20dcc9f639510
--- /dev/null
+++ b/gguf-llama.h
@@ -0,0 +1,468 @@
+#ifndef LLAMA_H
+#define LLAMA_H
+
+#include "ggml.h"
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+#define LLAMA_MAX_DEVICES GGML_CUDA_MAX_DEVICES
+#else
+#define LLAMA_MAX_DEVICES 1
+#endif // GGML_USE_CUBLAS
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+#ifdef LLAMA_SHARED
+#    if defined(_WIN32) && !defined(__MINGW32__)
+#        ifdef LLAMA_BUILD
+#            define LLAMA_API __declspec(dllexport)
+#        else
+#            define LLAMA_API __declspec(dllimport)
+#        endif
+#    else
+#        define LLAMA_API __attribute__ ((visibility ("default")))
+#    endif
+#else
+#    define LLAMA_API
+#endif
+
+#ifdef __GNUC__
+#    define DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
+#elif defined(_MSC_VER)
+#    define DEPRECATED(func, hint) __declspec(deprecated(hint)) func
+#else
+#    define DEPRECATED(func, hint) func
+#endif
+
+#define LLAMA_FILE_MAGIC_GGJT        0x67676a74u // 'ggjt'
+#define LLAMA_FILE_MAGIC_GGLA        0x67676c61u // 'ggla'
+#define LLAMA_FILE_MAGIC_GGMF        0x67676d66u // 'ggmf'
+#define LLAMA_FILE_MAGIC_GGML        0x67676d6cu // 'ggml'
+#define LLAMA_FILE_MAGIC_GGSN        0x6767736eu // 'ggsn'
+
+#define LLAMA_FILE_VERSION           3
+#define LLAMA_FILE_MAGIC             LLAMA_FILE_MAGIC_GGJT
+#define LLAMA_FILE_MAGIC_UNVERSIONED LLAMA_FILE_MAGIC_GGML
+#define LLAMA_SESSION_MAGIC          LLAMA_FILE_MAGIC_GGSN
+#define LLAMA_SESSION_VERSION        1
+
+#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
+
+#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
+// Defined when llama.cpp is compiled with support for offloading model layers to GPU.
+#define LLAMA_SUPPORTS_GPU_OFFLOAD
+#endif
+
+#ifndef LLAMA_DEFAULT_RMS_EPS
+#define LLAMA_DEFAULT_RMS_EPS 5e-6f
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+    //
+    // C interface
+    //
+    // TODO: show sample usage
+    //
+
+    struct llama_model;
+    struct llama_context;
+
+    typedef int llama_token;
+
+    typedef struct llama_token_data {
+        llama_token id; // token id
+        float logit;    // log-odds of the token
+        float p;        // probability of the token
+    } llama_token_data;
+
+    typedef struct llama_token_data_array {
+        llama_token_data * data;
+        size_t size;
+        bool sorted;
+    } llama_token_data_array;
+
+    typedef void (*llama_progress_callback)(float progress, void *ctx);
+
+   struct llama_context_params {
+        uint32_t seed;         // RNG seed, -1 for random
+        int32_t  n_ctx;        // text context
+        int32_t  n_batch;      // prompt processing batch size
+        int32_t  n_gqa;        // grouped-query attention (TEMP - will be moved to model hparams)
+        float    rms_norm_eps; // rms norm epsilon (TEMP - will be moved to model hparams)
+        int32_t  n_gpu_layers; // number of layers to store in VRAM
+        int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
+
+        const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)
+
+        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
+        float    rope_freq_base;  // RoPE base frequency
+        float    rope_freq_scale; // RoPE frequency scaling factor
+
+        // called with a progress value between 0 and 1, pass NULL to disable
+        llama_progress_callback progress_callback;
+        // context pointer passed to the progress callback
+        void * progress_callback_user_data;
+
+        // Keep the booleans together to avoid misalignment during copy-by-value.
+        bool low_vram;   // if true, reduce VRAM usage at the cost of performance
+        bool f16_kv;     // use fp16 for KV cache
+        bool logits_all; // the llama_eval() call computes all logits, not just the last one
+        bool vocab_only; // only load the vocabulary, no weights
+        bool use_mmap;   // use mmap if possible
+        bool use_mlock;  // force system to keep model in RAM
+        bool embedding;  // embedding mode only
+    };
+    // model file types
+    enum llama_ftype {
+        LLAMA_FTYPE_ALL_F32              = 0,
+        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
+        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
+        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
+        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
+    };
+
+    // model quantization parameters
+    typedef struct llama_model_quantize_params {
+        int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
+        enum llama_ftype   ftype;    // quantize to this llama_ftype
+        bool allow_requantize;       // allow quantizing non-f32/f16 tensors
+        bool quantize_output_tensor; // quantize output.weight
+    } llama_model_quantize_params;
+
+    // grammar types
+    struct llama_grammar;
+
+    // grammar element type
+    enum llama_gretype {
+        // end of rule definition
+        LLAMA_GRETYPE_END            = 0,
+
+        // start of alternate definition for rule
+        LLAMA_GRETYPE_ALT            = 1,
+
+        // non-terminal element: reference to rule
+        LLAMA_GRETYPE_RULE_REF       = 2,
+
+        // terminal element: character (code point)
+        LLAMA_GRETYPE_CHAR           = 3,
+
+        // inverse char(s) ([^a], [^a-b] [^abc])
+        LLAMA_GRETYPE_CHAR_NOT       = 4,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
+        // be an inclusive range ([a-z])
+        LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or
+        // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
+        LLAMA_GRETYPE_CHAR_ALT       = 6,
+    };
+
+    typedef struct llama_grammar_element {
+        enum llama_gretype type;
+        uint32_t           value; // Unicode code point or rule ID
+    } llama_grammar_element;
+
+    // performance timing information
+    struct llama_timings {
+        double t_start_ms;
+        double t_end_ms;
+        double t_load_ms;
+        double t_sample_ms;
+        double t_p_eval_ms;
+        double t_eval_ms;
+
+        int32_t n_sample;
+        int32_t n_p_eval;
+        int32_t n_eval;
+    };
+
+    LLAMA_API int llama_max_devices();
+
+    LLAMA_API struct llama_context_params llama_context_default_params();
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
+
+    LLAMA_API bool llama_mmap_supported();
+    LLAMA_API bool llama_mlock_supported();
+
+    // TODO: not great API - very likely to change
+    // Initialize the llama + ggml backend
+    // If numa is true, use NUMA optimizations
+    // Call once at the start of the program
+    LLAMA_API void llama_backend_init(bool numa);
+    // Call once at the end of the program - currently only used for MPI
+    LLAMA_API void llama_backend_free();
+
+    LLAMA_API int64_t llama_time_us();
+
+    LLAMA_API struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params);
+
+    LLAMA_API void llama_free_model(struct llama_model * model);
+
+    LLAMA_API struct llama_context * llama_new_context_with_model(
+                     struct llama_model * model,
+            struct llama_context_params   params);
+
+    // Various functions for loading a ggml llama model.
+    // Allocate (almost) all memory needed for the model.
+    // Return NULL on failure
+    LLAMA_API DEPRECATED(struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params),
+            "please use llama_load_model_from_file combined with llama_new_context_with_model instead");
+
+    // Frees all allocated memory
+    LLAMA_API void llama_free(struct llama_context * ctx);
+
+    // Returns 0 on success
+    LLAMA_API int llama_model_quantize(
+            const char * fname_inp,
+            const char * fname_out,
+            const llama_model_quantize_params * params);
+
+    // Apply a LoRA adapter to a loaded model
+    // path_base_model is the path to a higher quality model to use as a base for
+    // the layers modified by the adapter. Can be NULL to use the current loaded model.
+    // The model needs to be reloaded before applying a new adapter, otherwise the adapter
+    // will be applied on top of the previous one
+    // Returns 0 on success
+    LLAMA_API DEPRECATED(int llama_apply_lora_from_file(
+            struct llama_context * ctx,
+                      const char * path_lora,
+                      const char * path_base_model,
+                             int   n_threads),
+            "please use llama_model_apply_lora_from_file instead");
+
+    LLAMA_API int llama_model_apply_lora_from_file(
+            const struct llama_model * model,
+                      const char * path_lora,
+                      const char * path_base_model,
+                             int   n_threads);
+
+    // Returns the number of tokens in the KV cache
+    LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
+
+    // Sets the current rng seed.
+    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
+
+    // Returns the maximum size in bytes of the state (rng, logits, embedding
+    // and kv_cache) - will often be smaller after compacting tokens
+    LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
+
+    // Copies the state to the specified destination address.
+    // Destination needs to have allocated enough memory.
+    // Returns the number of bytes copied
+    LLAMA_API size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst);
+
+    // Set the state reading from the specified address
+    // Returns the number of bytes read
+    LLAMA_API size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src);
+
+    // Save/load session file
+    LLAMA_API bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out);
+    LLAMA_API bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count);
+
+    // Run the llama inference to obtain the logits and probabilities for the next token.
+    // tokens + n_tokens is the provided batch of new tokens to process
+    // n_past is the number of tokens to use from previous eval calls
+    // Returns 0 on success
+    LLAMA_API int llama_eval(
+            struct llama_context * ctx,
+               const llama_token * tokens,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Same as llama_eval, but use float matrix input directly.
+    LLAMA_API int llama_eval_embd(
+            struct llama_context * ctx,
+                     const float * embd,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Export a static computation graph for context of 511 and batch size of 1
+    // NOTE: since this functionality is mostly for debugging and demonstration purposes, we hardcode these
+    //       parameters here to keep things simple
+    // IMPORTANT: do not use for anything else other than debugging and testing!
+    LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
+
+    // Convert the provided text into tokens.
+    // The tokens pointer must be large enough to hold the resulting tokens.
+    // Returns the number of tokens on success, no more than n_max_tokens
+    // Returns a negative number on failure - the number of tokens that would have been returned
+    // TODO: not sure if correct
+    LLAMA_API int llama_tokenize(
+            struct llama_context * ctx,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_tokenize_with_model(
+        const struct llama_model * model,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
+    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
+    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
+
+    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
+    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
+    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
+
+    // Get the vocabulary as output parameters.
+    // Returns number of results.
+    LLAMA_API int llama_get_vocab(
+            const struct llama_context * ctx,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    LLAMA_API int llama_get_vocab_from_model(
+              const struct llama_model * model,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    // Token logits obtained from the last call to llama_eval()
+    // The logits for the last token are stored in the last row
+    // Can be mutated in order to change the probabilities of the next token
+    // Rows: n_tokens
+    // Cols: n_vocab
+    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
+
+    // Get the embeddings for the input
+    // shape: [n_embd] (1-dimensional)
+    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
+
+    // Token Id -> String. Uses the vocabulary in the provided context
+    LLAMA_API const char * llama_token_to_str(
+            const struct llama_context * ctx,
+                           llama_token   token);
+
+    LLAMA_API const char * llama_token_to_str_with_model(
+              const struct llama_model * model,
+                           llama_token   token);
+
+    // Special tokens
+    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl();   // next-line
+
+    // Grammar
+    //
+    LLAMA_API struct llama_grammar * llama_grammar_init(
+            const llama_grammar_element ** rules,
+                                 size_t    n_rules,
+                                 size_t    start_rule_index);
+
+    LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
+
+    // Sampling functions
+
+    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
+    LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
+
+    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
+    LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
+
+    /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
+    /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
+    /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
+    LLAMA_API void llama_sample_classifier_free_guidance(
+              struct llama_context * ctx,
+            llama_token_data_array * candidates,
+              struct llama_context * guidance_ctx,
+                             float   scale);
+
+    /// @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
+    LLAMA_API void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep);
+
+    /// @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
+
+    /// @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
+    LLAMA_API void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep);
+
+    /// @details Locally Typical Sampling implementation described in the paper https://arxiv.org/abs/2202.00666.
+    LLAMA_API void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
+    LLAMA_API void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates, float temp);
+
+    /// @details Apply constraints from grammar
+    LLAMA_API void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar);
+
+    /// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param m The number of tokens considered in the estimation of `s_hat`. This is an arbitrary value that is used to calculate `s_hat`, which in turn helps to calculate the value of `k`. In the paper, they use `m = 100`, but you can experiment with different values to see how it affects the performance of the algorithm.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu);
+
+    /// @details Mirostat 2.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu);
+
+    /// @details Selects the token with the highest probability.
+    LLAMA_API llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Randomly selects a token from the candidates based on their probabilities.
+    LLAMA_API llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Accepts the sampled token into the grammar
+    LLAMA_API void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token);
+
+    // Performance information
+    LLAMA_API struct llama_timings llama_get_timings(struct llama_context * ctx);
+    LLAMA_API void llama_print_timings(struct llama_context * ctx);
+    LLAMA_API void llama_reset_timings(struct llama_context * ctx);
+
+    // Print system information
+    LLAMA_API const char * llama_print_system_info(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
+#include <vector>
+#include <string>
+struct ggml_tensor;
+
+const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
+
+#endif
+
+#endif // LLAMA_H

From 4f865181aa9b3f214ed772688de5ad24068a5f04 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 17:49:31 +0300
Subject: [PATCH 086/242] gguf : start implementing libllama in GGUF (WIP)

---
 gguf-llama-simple | Bin 607488 -> 607488 bytes
 gguf-llama.cpp    |   5 ++++-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/gguf-llama-simple b/gguf-llama-simple
index d7600282e3315a4bc805a4f6c01550d58bdb80a5..bfa61de57dbc77cfb6c180b20a23a46dfebaf99f 100644
GIT binary patch
delta 19466
zcmb_^34Bb~`}f?N%$XZRV#y{Ui-{~m7D9rM5KI{R7Hd%rT5UuVJCllSI%4P0s--_g
zRa<HqR0$G6t5jRHRZz=~*cC}I_x+xG@7$>V-v9e~KkqxA&*Zu1`#k44&-0x9oSVu0
zI5hX;(Cq@}_Qg%2W1lZ}*5U%gxiiaRBYZ5|?Fx7%Ly(!uU!5-KGX+2Y2EMqCL6#HW
zz=t1G<R89)pE;r6<!|5*m2#IAN7VAW>ZR!P9Q{ul)-QVm3w0M24MA^kJx?ll_#1e#
zpQ0_{4Ls<&BA@&QenwPq(;IlQQm#eU^9_b)fKtHlH}K?FilQ-Z;JJE5e*7D_XQ?7T
z`3*cKP?29O<JEOKXPxBktku}7sgQA*=uG+U4|X;k*pKH2HrY96@GDpDPF~({d){y^
zWE$^tnQp$srSL`vy>p36=DwqiFC!TgK|i?6CGoNM=6!vc!<Wa`#s2kW*Jc72WxsaA
zwRbS@*32GH^ieK+mpJ<<H@%yln`{4|rk;1_exh0a`nnvo&-K@@C)^QxScrbRD_2U-
zH_~_JezXUL=_%nmR@nDN>8IA^&fA-}*B=k#bwAi0WA*1a?y>#%_w@c8H;X<Rrw`@s
z*nQsDClhXkJ^e%dIah8Wjh?Lk+~EJaPG_?E()Xt8ThlsI^xtqV=#eS<XT0lKx__#^
zH+Pk~Ptyl;-_Yo3dVik(-EN+yuiy-$Sw5RD9X?a<O-IhqKj-`WW*<CL|AcVg*n7;;
zALF?t)O)_ZHJ5MiJzsyvjoVI>mg>LdGVS-4>J2Wu@8|YsEA(@CK5H?Zu}Z&(n@yXq
z)`xS8=+M<LcOA`R<Rbl&kv;a)tM#HgAO8{Ew^=`zJ4KVX=wrAwG;53g09Rsf`la5T
z;|9_;TOmA#j$|a;K7Xs;m**DPH*C|FaNJ${N8jjsg>a?zU(e|0)#tvm55BMO?!xzp
zr(ZqNujh79^J9Gy|6y~x?XkWO&pBwx-}*V+eEU03^+g<a*53HJex@gPoo+RdIPSLn
zPXjU5>}9e(NwZ2Oo5fQsxbQ-<rW(z23Op9BLQ^y4ckne?XB*F;1IIoyS%(;7%+~3~
zgt7)1O+5l<%iY>!+hyE~C+S%0Y@^v@_HaUbdy|G~6%PuEDlN&rrsCvqCTJ?2ZCuYD
z@JT~qCaaQ99EL*EY=_nZH=*!wll7Q*$5JVotVK3ms%-{6=uLvVmYeoT$|ukxZ!u)M
zSzo0^6q#(Zjq{N*M_e@>F6YgOqDS5=q|>aI(+=O|&DN`?*y9YH<qChgCEhh#MH=Zt
z8uIg8X%8O~X(@fjRj12XJh4gFeAk0%iG?}-Wx*bSx-R*3zlI*2*?K$nnfO7gN(uaJ
zEEu^!yUme2W`BW<DS49Thjn5L$}Ac3M+-};wLI_Y!!G$Xet@{nly|Kl_ADecevu<H
zwV!8^!;)$47lTCKem(JbNWbwT&&~{wclDNu*_)GTUvFNfL%dx~c}J2>iNE9&m=nvh
z{a;gPGZ%TLSZmBRNd;+%#W@dC@;krED|dLzl3a8;TdHw<UU^NAc_(4>zSA>wI*<Px
z_Q*TnstZRSNHr#yY?D2U$|s5T-(Mi`M>Wh@yJB1VjW6*@4mVq$i#rlfJNeU%Jz1N$
zXtq8|vlfVVlPV?K5aW1jiTF#uN-51+Y_ge+lbM2C7J<Y^Ilg5{l=zV*Uh82}RVW~e
zOc}JZA8BrhVIf|H4Bc!wLbujAp4q*-<!5+yvo<n8drz}%x-p%#K><~phTu;Kf<>nM
zY@-l+R!nWn3MQ;zA<f#|(`*}Oj6q#xn<2{OBzWwpK3^znx7cibV75Lp<@YuA6o(@s
ziW~|wx#o{E`cj7<srOE$7?%h?5gq92B&6AD9&_d7NTygj1(|FI6u-cqjQ*1rQCx&3
zUwC+`rBirXai;{hpfg1jXUCZKzET>l#gtg=kypU$bmDxbSuScuJ>qBaO*<^{Y1U!k
z_{Ma}cl{m(qDr!jlg(-|X%Ww}tDCGPW^1Lms!gS2=3X#aQ+}#(IND8eOyS{~mQKkm
zXlZ%JhIg|fc#U(=p~e{n6Ji!_Vl6fVWwTnI*n)MhqpZ6no2;V6l=lL$LdTvE#OhUG
z@k~ZMPopFz1Qli`X*T+r=3L5d40$fci-F-Dd)yFGzYJ|SVD+BOdhtc^YD*=4$eC*N
zlxu~BkcEzoC~~cdtqnShF72>GZ`$|OR^?;%`1Vnn^t0I-!%P=Ui4U^%qLDQ^BqzfY
zB)!yRPdNlEva=^tN|N{rWtNV?lNsa7c9haF_F32bHk+LN*<?LnEC7s7U7JN=&Ly+0
z`9@iCW=%9Dv-K7;y=Ci4soR_z*=T0B#jVh3wxTxJXu~q2inh>bUlK5a4Fq!rzmTLu
zp1@gb!{S*RJZ8#|4;Tku<#5{i8pBhq6|n`Pb+;rX@8)FJSy%cJ154~ZlPx~E50Aij
zUF*$b$ayYa173O>D(O9c;%g~r?o>*gO3@O`)@x8{d_!fbQ{^S-g39Tt%IUyO`AL&U
zbF!U3z)n$sUSM5-iQDk3dC@~D5QdM3CzkB@R98^m?p^<M;diuU*@baRdG2a??kKOU
zbJ^QyXk}x|&=ilR+1i>DZ|C@x4Psb7_NOQQXqDVS^h1@DwJEDr7R8v&*q^5OV=0T!
z<j~@MQQc?)_uxVD?0Y%-GW3k2ENU-gcQ7!CwGcXV7jYkq)7oYqll2AWIo21D{TpOA
z*;<?Oj(VCBOR~w}(ltN^AFUdg`lSp=>7UXsr7t~EpDg0)(t!;~nKz3WJV{G@=`q`k
zp3N%#?`CvRAUPQu-i-0ur6oG5cy)o5wu7MAW;80kqI`j6tB9lhf=G+bF+qs`Pcg8^
zS}dl5cAN7or>4o)#%QwM7vJV|x-vAx=i=%()LC6-xAmwPAa~*tF<Yj^;>*+YY!ESV
zy{UgNX~N}DQ!weypNpldf-&3k^v7Tl)^2}Or6Ji`o8?~BC6+bq=VBa`W-La4o|tSi
z)c`dG@1Fq*ps^t&hW{#xz8ym5^8bmUWg%obr>A2Y5-We{C;FrzN#wW3(56PRJF=p&
ztlMP0B7OiD#0#>6N5yMQFX|={-Y|y`|47|KNkXkK*|uUpvCqubJ7P=PKa?CA(=v*w
z`49nUJve1Z%FvWyDQ_JWKZ;Z`e+imqb48#!go_KJ8F*A&E|*jy-c>`IeNvJ<0+V&B
ze;D67scq<D6WNx*rN1<xGa8XJ-?;$FZX7k+cCuv!>VXXb&&5o7r4b3_7e!L35ecXp
zt7J3f6$;EISXA18Mu!ovKmnX?)>~}sSli9kD9tugNK;F@DwZ{-Z-<dct{q(%Mq+~3
z4n$cVCfl?zDYl70Hc#<=0GJKaY%N$^9*m<UVZ@vNqYr%^Mgn^}I<fwofd2T>$hxDK
zTQ^Q;&etc5;B@=;V`wi!^U|N!q|PZ)f0<<|9h2GvDRf~oLW=Nagp=v~$C30#IBCqy
zpq>#VxZ&V%1jJU?Xqt0&W_5X<bcpjH&Sp_WLJ}(@Xx|7DYM4F<ZT3nS&5a--)wf2d
zVcZZwzm6cuyp%>CM3CqF+;A#Jl7xV5Xbau*M?KZ0<rVru7n{sI6@|5Eo5m!On`EEZ
zm@MbGk+gagX~#{VU86`We<qa9j3QIen#-d|eQ!Ig>yh8wjycZNH0NHnV6r|CLulis
z<OSD>hDDPGKJ&s5!~B4C=+~}V^*wG#heVSMbA>02UJ=ZUy`YXLSWTVAXfVE^?17k$
z?9RDN9WR2E@uEAEuuT(K3WG&ES~G^s;IB2J`7tC4{p<S}vH)Gd6iY&?U6+d}5l@EN
zvt!9eJU59xYerhtUK0Z2@`@1w*pBFm)oJ_Y<O=z;I_zpm8^x1w{$M@YGoIw}`|8u*
z<4LZ|r2P^Nk0*`zg?<Xd8Xz;c&U9`|vY6{neG*70*OMlJghl$v6BnD!1va5!q2n-H
zuZz7KyjF)r^{6$0M7g!=iefkTR!UTANTT?M{`7tV`97?fj}nVdp-c1*V2Pt*gZc_M
zQKrSBV-G!(NIKRU03o*6bN1TNv{5T^sN-m8{dce3;l-2`ix)t(UVEUPQbM7>QbK@S
zVzIb!H;rsfR`Xx_(4yAlFn`RO&Tm7S7z|iLi2tcezi&gjaigeP5_z7sr!BAnb(kr~
zFaV>Da<JfN&1eYgzX8mZFjl$@W|x&!XOpy&xFWiP!^EEY+SA$+GZ*`Ztg6^-d)wD!
z>l0*-xUO~xh}{zXSbz?{a?w@@^Xt%K?TFcJwN9B_+f)B!a)e)0o0cS#ru-o<>XAac
zxNsVpLVn_=(Z?wyr2gKzl~U0QPb{yx=a0LGo>3&Q2Fq3<B1YQwWP~BYfpN2_CSBd0
zgoN*i;&hkV7{!4#E2T@*jh+|uVPG?)GegE1J+C~vpkEY4|7uUt*;cXx>0H0G38(vW
zh%p=mVYQYfozB@>^kUfIqKc;aXAZi)0~u9qF|-st8gK%GT+LJx8{m#Ch1X{cbg!X4
z)c1d-laWx~J~);9!3ETdWFk`ry8EdQUg0m8$drNgJ?z~(k**x~A^oH?Ce62Lq>1eJ
z9aD{k9&J(_hNsxGWhYrPIJ=p?G!b7eoqDB_7F;kjr;!*gna)Tfv2jfeN^@B0hN`Po
zgC&lNe(nmGCDUTDVv|%gzppfsCj6Od%IxDuwAtt26)X%Zm<7<l$(l;Ufif)?uYDoU
zKD9dw|7^zt2pteg4t5y;GydI>H@Grw#o`H2ts!rNrL2fua#1B>jp~Zf@{RKJL!;v{
zg?!_p^x`jx(u+sajBXfyr_rU|$UH=?UUxE$>q=L4$K)+;ptrh{ATEO%dXP3;1WoOM
z_?T!;4-y;Mf@9iC#I-Jn5A&849jpdzicE_|Zi6H(rsKPjVBTLxmvzIeUB`v`cPCT%
zvJm>bC+Wt|UndvB6P^k^NNfIto_6d(qIpqAC!)B>bB+qRoVF-W+=s$gIfrHcOGH1t
z;^Kz2teoT!672H*s|q?Qgv{Z0>XhYy7ftVj<w4mR`d1$^fU8Yg_eJ|mqs#k}*sjGd
zDkSM>-V02c_dRBRg!ySdTS=o2U{PGV9%j6J_pi&b(YBt)#5E9W&y>ABY-YATg-q<T
z*s~D$E&;tiihA}VQw&|%^OZ`vxE~qSXs-_Dv1VsZWJ|uZS8xT%aJFekP*#!gFgLlM
z-p9kAG4l@?!`WKp3c&uPFJir-KMBFGzPCSV>(>8eg;bh_?HnE77e@&DU(gkONldMT
z3Z@o&U+gAw>WL>_(RT)rR{XsRx?uqE<(Ivn-wz-W-NwMxZdM1INwG9&`_H98S1`*i
zU@h*`4EA?}Mlqe0_A7iX^FqkkP9j~b+e}&4)}XQJ#M?#xLN?yZ+a#7%GV__tFL8>@
z2?xzdCn<)xs6cT^CH(=3pzgBR0eQnC20wS^2~+a~(gp)DnGB_=1Ic!rLp>cxI`KR7
zG-(jITl?ru3=|cX*+x(7-?7yAwGM4Pm<;qye_DYqBFU?^6ziYZmaqeXrImEsVC-Ga
zmsdzQP?{a&#;O%I#{0%8qvQvFE1Zcqu2eW2IN|^meoKQN#R5)NVP8$ppA3#rVNQcL
zGbOcE*kq~ZRKAaizj&gQ8LPoxG5CfGU%}z2VhUyODHT4e!HXEYM}>_woYFQ1uT$aO
z8ay|avtYNc;sy<$z{Imu_@SoUn^`_cg+J1y$1&*+D*Ubnk6>^U74~#@n$nxWUMjq<
zhQ*1uWw_(9;{JS1OEU)FR^bRui$8<UsBmo!uFK$kDm+kwpTq$ET!q_fa2a5WXjAdI
z8h(?BXRGi@4Zh0Yp(<Rnrql3q4DO`Df(92eI9i39X|SEazAF57O&l&L^|^`RoQkj3
z#8)%;&Lbr%r!{yfgMU?F!NY0F0tO#Y;kFuVW$+g&oUOrg7;IUn;#)O*3d1L<@Lw7{
zp25Ra*bk$iV)z&ar>Src4Ia(lI2FF7!9y7Aufp8{TNEuOhP$eGOHI5jgYP|5-1oq)
zS&?qa;BzXRslh=EKBU5pHMj<Yx2SMk4Sp33c!@>DCuw*o!>6k7a1Fl5;I~zHq6VL2
zuvvw_)Zjx5j#uH2H25n9H&Ees8Elb<oX;6<Q1K<2IA!p|2a5Y^G<Z3KFQ{;~2G3&f
z5fvV)!4nz$l?sp1;4B6&e_&CRkJRuChR;yNhiLEs24|>ne+@P<xQ7al(BQTVPEg@4
z8r+1z!7AKLhAnKwEn;{z6`!ez?_=<j`-+95HCV2=%PKrygP$_#A{EZk;3QW2->C3^
zG&qjI^nQg!9{d+*_)Lc9s^Xt$@B{{rRpC4hKEUAKD%?kFw%;<iwF<A*;PnhPs<5Ca
zvB3Ss40l&?v!;9&gP+|~+^@25I)h78I9ijI`{8jF_SN8#OnQe3r)zLJgFgXmVRK(^
z4e!SAJXO5P6uEnjQ{k$R%iXh|3fHXT?4AisNm~`J3YL+<VJbXMEz*)7UytFQDjwqN
z)Pkj?POP}AxL;j^pRsDbuEN`DJEb2m_(v7qq`@MCcd77t4gQnCpD{QR?(fm?vkafB
zivOg+a{I&?%iE+G|LGH2XB6qcpDUy3qi`@Vm@XYfjQpO*bnhtgC2uRE%SK`Gb?XtG
z{|;%-J*HPONN9u0r4`uw723j!v0!30KVWC9zWXuF>)6I7gG}O!@6vf0q;XWMJ8Zo+
z-DrAa#WDdMq;$jI5>~jy*ygb_sFE^Tl0l3OPKf`5`XMxUH0jy;Qdxy$>)<-4z+`J;
z^jP%p33eTu5p`<^*V4a`Fj<e7@~-Ea^0S7=N=2_S_4`Ay4tka8em0bD8%>5+Tg=iG
zneMAXsrMN2Uj6q_C_5}KXQwvoNUUmcwf-JmJBAGAciyKj#*mTKN|;O73>4p@qsNjY
zKKwR4{4VjL-;X5?u&KB>7Dp6CwBTLhO&h#RI&kkP&y(LJojG^*Je2?TCJlU#4CMOK
zDesX%e8mm=`+MYFKK}+y8Ap=1?$k04J7bRi1j6t7i`IJ|+dS-E-X}dcAG+s#peL_W
z*AIZspgli;Yz+PE1JZ-PdX3%($}hV{TW6Ba{3kc*>`W5P`O#gOBpy4&JDH>lzv?P&
zF`hK%uV1I*#$z)WN&hpRbmW@T((yQQ+*v~XvygZHUvyL!Y0T&UMVDmZu(Jcbm_;@<
zTmF(=D^x$&AGN(7>v!xLl1|k}Q|oh+g|Pa;wAUEb8-US@?wmlP{o~|}!ap+lGybC+
z^wk7X-%|O4X%-F+dQCw$Ca~`Y8cTwRYGpS>$8sB^lDN`fY!0w8E>v?3eJxiW%azA+
zy?iMPty6`Dye9M`6B@>Z{*Z;}>JLdv{^ezQ_Cqw_yvwv}HqH-y=)`PdZc^<soWqV~
z3lu1=9i13QBiESE({W6WT~4^N%Xy35&n6N4?#t9KhlKD;N@x<JpIxLQa>&OXBmS(?
zByPM+D|5&P{H`nX-HC|xyepJWB;m0`ufVKrH*nX1g*nEbi>#Oh#_e$2WW`AwE`H*B
zm9ubpy$d4Nrw=C*U&C(~u>Ck)LOmvtPq{(#t4U-ZU;YQpm`nnz{Z)rU%PtG>2&9K5
zkuPvS5<dm|-AT0n6tdouagLR6K)&EL<tx{MiaTwfQ{Jm#BO;!by9rMHomWBF>G?xo
zJ7@Z+FkP?HX~)Jxao3<f{<mDCXd!F;INiZy_#!tMah@E6<;HNfyTQ48*U@+=8-Oe;
z<tzzZY<+Jc(D6r@)diyOrPqRH`<>aUAE!u~aB%{2;%lP>NZ!9l3#Jm^jJ~qjNilAS
z&g4H>?xYrMLH!eRG(RblJr}UB<|ox)GJO>pUxsf{arV4M4vhZZ6Xppn3YEep`pSi6
z(zeq`0KfkU9Wo77{LOiqJB=7Uw#d4a#v+!Tv+taS(^mfGS(-KjgKfd@bm0sVXj#YH
zV4GkT*f_cK{4=mKA;1nUGvEp?R#LGu@81WvIKP^0@z)=6xY}=PvU$Vxw%)Qe-=J8W
zXCIQs;S|`7dsMg=5AH$i*woM;hc?Er<R>*^oCJ;Yt;)$y>cki^8e^r#7{(agHO35$
zv79lqo-dBl7)6ZnmX@Qd#<0}z0Vh=B#AzHq#?Xcr(NALxV2qwx4xPqW#2CXg#_jV?
z-+yNelg9X2X0)~SaW&;z;(4VKr)rT(1#qywCcQyZlg1cT-p$t-QyHVTmLpSR>|%^6
zANp&I$BZ!$IV?&~NYXgXSQn|vV$>K}jL}ysq`JoVnlY;C@3CC}tjbUQQ`H5HQN12=
zbd^<Mvx**|aoRFYFRhTx8e<Y;_-TxV8e=kJBx;NtnSt2i$|rs$c<7A5Xu)Wuwmvn0
z{ia~<D)w$4Pk_9?ELicIDp(g-3kAEWU`3Ul9b$?q2Ni%b3!Lqo_@h&LGO%Ncw7p7t
zx+?upk^byATI(ZHw~c>*oH<t$%$p74RoMXw^KDi0vc+IQ`8vSx!%RnfM7%8b6rt!U
z4Xdg&G{GDQ+iF)amsc?}6z0=koelGA00*NXu%At#Vu02L3mIdS#&D4tZEefepv5qY
zs$tfGMcZUeZ4sQjja9|82NO0ds$#CMVrDDM37QQ%0UT_p6}4Kf7}k8wGDcNP%vBi}
zF+2mosp@E>wJiM^W3Z;HyT(|}7$Y=ByvDfB7;zdSKx4dQj3ABSqA@J4xVwFA^tn?N
z>f;bsYqBYX5QF@)a*JZEr@6{`OW}N2#Ys>(vlPxinZqKmhm|x&vq5Py7RDjL;Iz;<
zqh%Wy!^jxv8l$_$7{wTw8Y8~S>iGaw3%-RAtUmvXvw{4|a#fAA5^Bw&jOT74;8aEQ
zmQvX!+cHnKI>?Xj3O-uMLRl619@ZxDs}*dj6AJ)#Tfs7bi79;@((Hn4Kc$2Q4N*c>
zEzd>?qDNnc<u#GPOk}L))JUySZDWjX8Y4|(Tw{!>8Y4zybim!|!5$jJM`KK7jChUV
zuxQmo8K<Mhxv4P<8KbI6{{#kB3inE#%aF^A|F)KKo0c&Q`+<Y)G{!QG(Va1r{>a8n
zajM4ogmLO=Sw_eV%*e~m07HCNfRB1`JHx6}nU$4G>|v!Vey028kYuhUt(-%;)(S%v
zn5-`HTRTPNp8A^h{&TTA;D7j$?wCis_+uyN@p-tE@TL0sBsn4LZ`e`$1k_rc*J^t4
z@kzE=crIT28H)w&)i<_aNW<$_))(TA9(2Zh@&o_4J8fenm-*k;(0U)^6@t<OwBN_1
z1%Fsa^FJmbmUnK*E1fjkLe-BW=nJ?3zXk$#uAm%wp_cTf+$BYBpCY%1$%P^KUWyz7
z@<0(-r3kE$4ZbbDDsZq9ZSm>%4!qwnfHt*}1eZp~D(IBw`jGI$#T63XA$u<F#e;g;
zzxQ94{ht=ooi<$aJwM6<c{J4)q-;&qcYh@8E*G4jk8Pw?!xtD`L|YMCdL=zz&lUtS
zueC16CDq!HT{As6Nqa0Hf!r{fxd8Wwlj)oVBszF96LnFtjbqQxk1N@DYh#vcjQ#fo
zqz=c=E~0nx$$fr#A-%Yeq*tp~$Sf)uT#BprMOff}Q9#EoB8&LtM``6EvcHD!aCTqC
zCL*0qzEgX6l-^%Vc1I*0lIv4`qXQcq(N9){TPJ1%cAryZid!f8(-KmTpL>{!OGqo1
zE=3izQxDRV8%z5xB@xZuI;e;j7oZQfSHdH9QH&BlmuayWISe;@TAQJsp(MWcJY;zN
zu9+UB&zF*e9jR>Bo3DsmJHU!57Ttbeja_|3<nCbx9u@ySqS#=Ti!T;`8$v6Wk%#=!
zBlf$?NiN5E+TUME+&KQ(_x8yY=M4PP{q(>p(xT>x&1mWG?5#kEQkD3)Z2!K3K1d+@
z!=CI>v<}_}&(3|X6m><mqeN^0JD5EO`^z`yb$lT7x;J6=3L4&u{Lm!~%Kv@kn6+EU
zz8X|>XZn7njM;KoCE{h77K@|$DOZk*T9SD#6QS5Ifi!d(zq^9A!$Wu%=)mRmzupp<
z0Cn1%9yj+?(2o;v;r`Zbczk#l{cH{C)UuZc4k=#Nnr&Q<MFOsgef1rjhz^?u8Hsly
zS26m_NEog68EM94((a!TKffGLPB-$&tMJet?~Z&lBRnhHSc~8wBy@&7_cOB7)$`O^
zyr4SQ1qEQLjQVx0y~F3^eGacU?D!Aqg_8l7jU<%6yM{(?B&pm?nz@nq@f!>1#~V>k
zJ?W{9q<(F0tW9$GX$L**t&7SdT#aI|{pCh-i{rO$r8hR=j6zRiHj_s<KnU4F!uT>5
z+I0)*&GoRa-9lP&oCiJkB~DI~>C-RCGu&0)-AelKU$3yIe1)toTO=J_w2i!U`O=}I
z_qW0(cl*_^+3AQtqrN2$7q^!x{=PkTJHyA&Q#;6WZagjej`Zb!&A0pS#H7t{UqIXL
zBE9MiT>w|hBT|h)u?2-AXQX>=d`>^#MPh@~wpU0+m9<ihHBvn6)9d#Z6BsbnbCj`0
z(Ui>kpFFq!y^9>>Jv&-e1;cR(pRR8~i|r&a>A+JR;(RYWbhfc4f_Z1;qnz-pUB<<D
zc167FEBvZ@z#Efb$yIUqDNWi-zUMbSu~+WJ0Std`j=jnE>@B$b2dr9#{nat{MA6j&
zmtc%O!+!bzY0dH1XVIZYh(8TGNS^R#KT;Ac4<T_RS4ng(AX^dh#|2~zx0DV!LIUWd
z!{i3PWG3U^;@oKJQL^4$Hv?rn9#+6!l50aB9VH9+b?d36kfgTELSrm?nSP|`@AQ+b
zX?cxez~6mO7QO6y1VFmi004MML@Fnd$)tOw-=^=sBtiVF+ccnv+;?wvi@g|nkA*s^
zE%iA@!m3|cTOp0S_Daaqr_!`zq)S6T$P`{5+xfS`>mxc|F8pgO|7(q+(o0^ofRB8V
zBZPW++xHwJ^<6A;uET=zh}rU`Dzi!7gCVSwJj4(xBkwqo7-yFl;6&IG3AN^{_nM^Y
zS?|nM-6^x!nJd(ZjCUdrxNEe}Nu2Kuv43=utm66aCQ$!VaQ50&nsf?P_}3L$d5T0v
zG`|A%N0gwxTv;KdSc@{+TV0zKReGlyd!kD9wILy1TS>e9Og6`cGZBwHg+==#j~7)A
zJX}=XR9~KjUPU}?L-0By+M{v2>~=296)RTQLr;@%j$b#P_Bcc0>U_HV^`?vW2|iwK
zUw(#Ux>@?2eKTHx4EaaA;+;s0)6Xy`qQ&cG#+&iF^oy!Ti&upc(c*Q8`-SFRBq`Cp
zSSpTuG9$e0k7VQ{%a@r8Ts+}O(bF>bVYqb%6k$;z3{7Qc>642jjB8I9T_S<B)g>HL
zh0xQNh#$?qMCx%h8QzY+f0mxPMB@01XX)ZgqyY`NjEdNQmZn|?xaq8Y@?{*(^4=qq
z8tQmjsiDrN<QfWJ^%KfhYiRhK|5ig2At838VhLFt^Xcq=SI~RIl?uxIcLj~J@3@L{
zRZl$*KIE~cC+iwX?dkJt<ejDkv((x$wnxP_<Hc_oFcmKi88Oi4N4$$nj?C$J+q@z(
z{%!lm*U13V&mUV0xdp68;N&Lk=0q0D2)aV_QOLZ+2sU2s2S#U-mF}zWrzG)OH3l9a
z1NH4_i&By^A{Ga(7^0F&3t1%<g@+c6D|%I4T=X>S7rtmlrar3({27_<ZlEK7-5nl!
zd|ct-obnUEynct$k-uVCJ9!t$`nTjHM!_G;i(X|tVcA0;AGO!GLy}$iU487`?~wwo
z?yw)AuyWX0Hmsc=a09L<)uAB|NF{%~r@g^Lyyoil^M2XJ#9}tEoMwZE{^)-C?ISdW
zD0=1*7R6F``|U@Ba+s7XPq0wyX<zY#tl;>wcG~1A$>F-%KYxmoZ_eL-_8Hbg++7-5
zL2~HCa<ZR0KzmhUI6U~2c+#O27&F&&qP;80_2A?kDE9EwVJ8r3vk}J*p8C(Q&w#P$
zlas^?V`;w^WE0+=dietLU|A~lc}Y_0dTv*YeQgxf({V2`>txb}FG*J5jBgR8>odaB
zy(T%I>SsBhLf@keUXkJa+R-%k6=|N34hGsmQj9|5)pS&{k`5Jm;D2uNX9pIj)qWgN
ze3j`HYrp%7OmpFPr`X4GhI=kCn_4r&J>9)G%hvgNo!X>4)IW{~7I@Yuqh7E?rDyzG
z_PKh)F?W7_6zyKe5XEokMQ7A81oFAP=-N64AM923)G;*0diZ7?L%U|_pF&srj=H0)
zhTl^YKNbD(^wP-3@!`EM1CGjO@R5$Xn@GgZ`_w+5t|67<Qs^2l=;=*=^fCk?b=S+#
zD%_hw^zh8g(Yx5+%-=9OglD=Z<9KD{j+@~nkCaJ3T1mTm8-n?JD`}RuVK{GFMSt}+
zbZY!IlZ%@X1Z#eHdP2?{n(4#}M?Ovp?|oWEqo&YQA43KgXaCB_(2L_AhEa#FA&$S>
zmHAp<Z{fw0U1?uG!)GnZA`q6(F5N4f)z5WYQ`f@lLv``bR{(qZ^c3(*g|psq9n#du
z@F)p5dr0!8Vf73#p8A!r=W)02-aj(5zBe6R&(OZw6ICj-ntf+ILlVa|p^y9x-kcW|
z0t_2EHw%OE6_;+YU>TM|!LR6vx5|r7Wu0>^D$07|0_L;Ak^OY~h-E7GS$W}+p{^OO
zg-5!%PJ}7_bV*V4VSu4uo##*>%hrS}d+-XS>GcgCR{w=DAQ&YswV$hR@Z)MdU4nAi
zput87KmFW>D+3LY9UphBkkAn_dps<9H2fC?Psj{yx)9}l_r9RJET^Ni^i4S(uchzH
z>7=GKGtjV$-?z{n8f5U~xz6?$AqaMdZC%(7vo&^v?ep1A*~<!ot!0x~EU;yKX}FB%
z0mn<%jJ+=ptsOD~WW*;;=vPL=biOE>MuZvy@OEyeP{S<FNKb^Kb$s5K-VHU(;XjR{
zA2u@Jax>Dtw~?WvYm`r_yt75QULmr6GEdfj1;;G=5Pj%poqT>IiulkT+1PN0=f>E*
zni|HsG+vqH%)8uQ&O0Yp&iggUzVnK0)r`paG?0E7Yl!CR+AqW!2Jw;CS~)Y8_{tg2
zV7&#WwqRIM3l;kXBV#fh+T37n^cQYzQ~IY_&!zNBvp!Afn__+0&y*i^18^U6#4A(Y
zT|xY|p8a%l!;Wfuk_|?)<>@Jh<ni<W`?t{>I-=BcOHl5vL-POcU(5fF@n7-8H<WZx
z({Z5h<2&#F`fK@jrZ3)5@~xU)3;H|0_x?Y8-R`M++NsJsRKNxFY`H^vhHpxRL+Xp~
zXnYso`zgMK__|d(q+6h9H#&MP-w<e*@(o!HEl+bL>HqW{K1-7Rr!UL@I^U^pI`XmS
zai9zEWt4qilTG_a;f(R7ve$%P&-jmLmWK`-VF;jyzB3T}pzjQiJuJd|h{QWOT>|KI
zti97ea7aIa&d+p6*}(UL)}5@=P0n&iqd@0Ra7Y_L&wuEUmV;jox*b%^M%1tt6>}U?
z8&F}ALz)j74k|)^%nam1dKGA!Dabby5zoNU`QBWIbRLW?vrrXaoX>MemqCR&4#_wg
zhRt<IO>jJ!KF=YI1Kl#;A??BmaKgtZ80q6S6pU5us(c(*f%-0XNUK<S2^;|xmpi0~
zkej^HAtg_P9_o-5gN|9{kj_lAz{Az>a5@A(frp@D)}R1T(`OFpCFt?BP>iu*%Q|=r
zI%WeDV|3ZF$sskL3B{WoQqV_G{3CFr6Ml6_<)G((Lw?}m?+(d?a#o#lNaI;~=MfR;
zNw|OrTOe>A4+ofnix5P{y`V!tw_HMjpsOy!Q<f3bJPR4GIHWb8rV>O7H2f+eg<*K|
zH5dT7F*h8NFASV~%OQ;czUsC^+5#$ya1glf9Yh4_@Fxz*(gsId$2m!w47$Ztk_tih
zf}RHzLH`C#uY><~n*%v-v{2CVpzA>eA4xg@x~je;33E}WjU=frs1PPeV?Yx?XM_4i
zOVSz8<MEOdfT8_7Xbh;(Lb6C^JS23Hq*b7kdrQ&{(D|U{pnE~<&IccRt#HuE!zHN=
z=zP$gpyx+Q(sIVnfWe@`SQKD|XXDU)Kv!W4u!W^3z!A{-pu)#Ue+Y*_#SdZOI6RaC
z$ORR$p^yn!U<jx$x>|w_8SU>E8J_p%w$XQdg+RVWH9FT<XwaxlO@|by!zZPtXC6PW
zrhrE?zNvk5x@7QfGhS!vIA-vsSAB&T{2P8PKcO|Zk#_SFy7H4f=_h_dJ*;|n`w4-5
zO?(~FL5^iX-QqhASBvNU98xL$rPR=f)~P21>o3=LNYQfI{#HHV9Zr8Y&>`j1tY0AT
zz+VXF3TUkWs7^J~VF5xHesUwaH9&~t3hjRe2yeObJ_2nUDMa%735`7it8ycS8r&|L
zA1U;}zwZATDRjoa<A*jDdhj(W=)}fCTkbO5-B{?rCpc(jW1%<S*Ohi}B6R2Ly3w_u
z+~4$G6G*$(plzcdeS^-35;}0X^l+5Wj(bSCrb3VEG5!we4>;uNA4tEwU@%fkQ`yoD
zO(C<GUTcaRC^Z^-rZ=DyqoH;iEs7R8^Ql4j4<nS89ZZMC2%Y#EA#_uW(A(wsAP0R=
zOXx_GV+Gg9ap4Z>IExxSZ2NloJ=Ek7hcrM&bh_rC4}tX^LO+NVTJSZ7GTPKF9Qg2Y
zIvpI36&mvieQC8gl)Z`$j1$86paFDN99+6e5621qsIx!ggsytiD2G&?*76b>(LT*k
zA=Ae=D}=6VCM0po>7UJnKGl0Ra!3~&!sVUy=sLaNLwh$zEgYb8n!~HQ>GbF3(EJ;9
zkB85r2GL>hD0eyiI$mhUj~YVn#|xdfmo&Zw0x!Hxr?!BwNcXi6I_rZn98wJ$%nQD>
zeoG-50q@pQXstJ4SS?{@`O-fd2p)7-OCiYXGNxd?8UtU@aTCy>XlYBqpWFp&d`(*O
zxWR`8CkWAc&xH<Y?22nb7@eLV1jRMS==plq<{#>GF~Gu@I=QvKBe!<ePJ4ovx`h6m
zAav2s$LKncndePCelmE{4v9jr*P8YJRQ3`XR?#Vmg1>$nSXU|c%HT_1HG|y+i9(P*
zd5c5ZKvP`c%&SBp+H2!C|IA;Qt<$ZiDXmZ|yXeqXsFhVm98xR#h(%>!B-H%Y3hpmB
z`Oj<%a?p{fTWi6;!y2%1-Iy;*i#Y(6fbqJGYZZ0|SaXck&M+i{bpu!ou%5h*Ru2<`
z=)BfKeSI4Y*l*F;HbO8x*;;78r~gUsv=-W<Zlc<tX&2C}HmEuAChMuer_odYf~a2;
z49A0%;^Y>(2D9#DV4u*epD{-#CJ8=hd%co`#!=a&Rc5dT&L-sXzyvir8*{>ejRWRS
zd%Z%(K9nTH^KJg7&ys}R+<BVT7VS>=+`hD}u)2n28K?UK1I-tl`za6IKF<BDhpv>@
z|KXv#&;N(l>6YkAcqASg@*wm?(69B-l?z+&yuMnkuRL@+s`b|C@~YR~?V(#y{Wjp2
zHT-|{(0$?_dcs3@$la*Z{pIdg>7kR{Loa*iHrDjm?xEXRQ@_STx3^{u>Y*#FTeB`2
zV$C`$P+-l_Wgfa^9<w?7!L9;d-Ew9Zk)3k>!jAJ7b)3JT<Dt6j{N)_yFXgDejH7e@
zDvncb{TW4{^Y?I^zklQWy&I>T`a3thb<Q8QQGdur=ltCor=auqYMj4M<NQ4u=kL!r
z<y1R$x%MRk1Owq#*;fx0G6hh#4B@m7AG3*GoFO#jW8&=9W(q4gKKY{kbBnNt=aZA{
zgJxko;5{$Wt$D(Izr~@75BMim`Ln!{P8Xtc*X7b3vxP|*Y$N7?6wnQGghLo}N6Z!8
zVk7WeVG+03-fNza$%A>=Ds1LGOX<5G3k&&V4_f)LFp(eMgHEss!?^YI7aN+}c3Nta
zHH}yx`0&HK(5VZAb5PYQAF3W=L5%6)Dvel()ONaWq3{PXE?I=s85+D8>UClElZ%BG
zoS!aEvDi7C?Loh(&DTV)_ge}bxpc%*6fi!9PF;!+49}-qmtu7HJV_rT4W+@$ga?=z
zYApv@Pq`JaZ87y*A-u)Mgwm-ikjFEUZe79hgwe-H!@A&=XyVDQ=({V0Vf^^6_LD1x
z3=X+tR|$XA^K76PgviMsHw;|05yJ6(X~=5f93OO@R<0Hnas_tlCjx$E<|-Y#2BYgk
zx|)%z_7iJ_K#n_OFZ~P?C7<j^TdhN(xHDWQEP~s#a-DD=(xvN#XujuCT5kj5ww*rO
z0K;>s+vlj4T>FX71>CAErrbth5+BroPTeR>M4m?*<=}qw1vJG((%PF?{miAGZ$bfQ
z=r5a~F*%X)n`Mm`He>mLNxX8i@BuS&ix6FJd%WUXgxWlYy`?mdo;~SzTVyM*Z4n;u
zo&ogim#FjPCA7;{7WGzis!Wg3ty_V|ETbV`q3q!^>35*@g5nKwsNPXSwquN<c09EY
z+9ovM_?STZ)UO4?V*JrJ@DM}tx2*Np_kAlEICyq=yRexVxdZ+tH=x&dpuTh$Y2<fs
z<qYL^%B?AFCz{b>d+;v7f)<}ZKi`e&8y-ygJ;Ekr{&Ej=6wqFFwCOW+hh6w1E_s3C
zFV08g&nPUWSzX37QVhmXq5PSy;z6I${d<Lje%pI1`F5*udJk)5){WNNm+TWf=5aUa
UhGe&Bw^gd&tLzt&-NIb|7q%UA#Q*>R

delta 19318
zcmb_^2V4}__wUZY-W>}?1zZHAxJZ+xqM)K+iHk0Z*szz_MX`WgS49k)l}L<wG@4lc
zwwPugHHxSpmPCyjOY9{&2906`gq?Tp%<O3V{oZ?@_xZg2e3mo!d(J)W*10q6{LcgD
ze;&AlV`>EcMKtsixVsMH8_cA96&+IF9L3Av`B9w2l>cfp>(|Qo)sJx3dOAr?_z1sK
zEXzOt2v0dB<K-XWlVrPJm4wvs9O^FXbRGR~AN;;5Ll5U)lQsB##LZVU2Y-YY>Sb-Q
zAK~lo%6bw$!cznpH-3Z{%683~l#dwsUb2CqAK`^nvZ65`;To+hKj|YpUokx6BYa0g
zSw3IFYiTt0Hqq>^6XqVlJ-kjd#yqEt-4XqIv+T97-8loQ9hq_2*+cp4p-jMR7OunA
z>r5iMsT%&i&Ll8RA?yYg*9XJI8%#XA^T}sF-eB;{;@5%y@rGkFjtS-OK5$I+XDv<n
z7@`e#U{6Q$BWh@CXqjw&qO+EDVoWgEOY6qe=NEfv*AwOl9~7Y7;mACMtBtjtnUlO<
zkQNAAT+SZ|*UoZd{@|Op*PaYw=O5y2W3`tV<{AITc&!)1n4v6P8^}E3>rc=o5XPNP
zo2tF+$T&mf4DCkUnoAmu(V~a(bF{6Y-c0RJ%xpL|Q~QQp^gA4!rA=jqg41lRKNA6w
zv$bCA!b`kqwzh)NJ)-rjdKfxS>jA^(YTvRKFY*KCX<rg1g72QAJ;5?xfJd&jHS;;2
znydZ01``GG%d|f;>HO1WTAc&CY6Jgfg*Jy}%a+32RoZ>b0%*Qk8_X<$L95Z-wUAB8
z75I*l-Tb-LTEU6kJ0A{g)h=d|Az_;~it&cY+q8$6A$-JlS|^6-18ugW@JJX=$z(ov
zyH?LKwfPM{Xm2xUc)?HF9s$fl{^EJ<XFg0*e!xH4ZVv3l=J0))c0CgfrWe|H_GL5P
z`a;{2W#+-{SK1tg<43;M7BfsLANE!|&y~3i+jXQV^MJppBL?RlNtV|}%R8gh>?#yq
zdnZ~t8caEbE?@kv)o7CPI_Zs;MTYY@fTQ1-Ea`?=qh+9hD{VL`!o_!yG^~x*J%;^w
zC7x)VX)w7gTCIgt57H>9yl-J~d7T8ku_Pgw3K~lm8P?Ma{1~mXU5ysGo-i1V8m+&q
zr#KpqGFnauf14{sqovrY>1dq*1s=q|t6)4J${$~s?0l3>wmeG;DKT0X8FH~?3b|uE
zTF#o{%3ZQIVmZljBkAZ9)?~S3j6O-xg-bD><hUx6r5r-*lSZsr1KsPBP;==}7vGK=
zjmzR<tkyNp@u#G?BBK?9bXi2hqT$a+`uKbNGwOZVUGm%xBcd@`9!0+qy0xklvF;(e
zbm?pphPMOxouc1#&1<v8E^f3OG87_=bsAtDW6nx%;%;u&u2M|Sd63DWHepxmO3`G=
zG|Zr}QQ0nej{EJ3DRvUEq&&yP6r-_A-Ws!uUD3#oSVo`T8ZG|_*U&&>UZ?8p@@kjG
z*Bmq&Ye&PR>~d$9&)Pa_GzZSj#i;-1s7rQK02XPJ2;2U^rm!40q9A9q&TzH2tq=+8
z^rU`b6lxYW#-iVOa||ibZ;X~dO_s7GOQCQxzEZ@bO|sk;j`gk-lPuUWCK|#X!+i9}
z7+03%S=x3KywQ^;?x$ExrP_f|7Hjo^mY$@!8T;2F!z!$iY@N#`x1Q^onVOt8+BMk{
zM59bGSu+i3G&!!=vPs!C*C~D`3wnb<^aY`LTWVQHExRUJy1SaJ*kO}%-ej6kl*x*9
z*;gx<D{Yr=vOF_cY{tA^h7{pE=0%pXW^qQxymW&f{NqU)jC>&k#9^SA3e?q^OR~D0
zaAc$uCR#fC8Lfxpz>q&9$zn?iIZ=!~zUb(TI-P@)N;=142pVHZNoJIBf3@6i&BnMA
zm+aOUyD)=lmP}1(Ks?QUNk?sLl4VFRextCpccZ~rBZfM|q~tFt<V|KRqvf{A@<PaM
zgZ+bfNo9?B9SuW^t6USJijK~y(>Wm-V@S$AF*Mnd#U+PjVdqUZU?a}KL6sxAEc%Ag
zXo}0H3!_?~d2D7JZ3#w;U^ZsI!$i_Hb)~6Hz*H8Y5z5EfCMsky_!)DqV>H?Cn5+gA
zjB?pm1GCw16#b};X`~IGNeA(t!ughR{;+ySgI;PCnnId7I-y9dCVWpcW(~%#H6HkW
zyZkY^{9KkKo;6vbsOzrAxMxxi&(sPowCf=@1H-|hMk%I9d7$ZCG^kP(1py5|?ul2%
z1w~iU(tyN)iGvacCk`2O{D>=-h0kJWM>%24i}@{`(U_&kO_oZN<!*GLP`XzX6ZSHa
zd$hmT!togW)M$;_*po%)-8&$mcn8E~b#bb?2vjHM(b~%p)zR`8^{x3x-!C{UC4H%~
zzL(}yS>a1$l%^Z=;-8LYq;_hG?Ig@?jxp!OA3;p;M?<p75=1cqv@!{!crsowi{`>V
zGM8*O_W;eH`#FlW9BQp|>8D1E`FQErQtT_oldSDbagVa}rTwu`+PgHg^b=_|kQ*Vh
zqp%DeDvhU19H+-io6$cJ^iK%>ksAwpq#w2fw$4ZtPFVCbOIwOP_c+biRCJ$aCq<}1
z+aE`LG&+$V&{JN2g5_3?puN-D+L(PDoBnnt8SvP<1vG6)wlK@#T0>ImkxOH@woQtw
za#>`;u|>OdW^?${mz<8?){OGnCB@m^^Ohs_xE%_btOf)1lB0$=T5V=C_`{F1=)A)Z
zvulb*!<G`ED2mR;Z?c??);0z@k$#HfrWEJqw}L0C5|-02EXRel)KLuLl|)N~Ip-nO
zpBR}7u*RP>VP3;Yf6|TZ(-iapxZEy;t^p*dU8_jB>secy;+|oaPNPw(c^4`jm5-Tt
zX|&E$GVv32=wF#w3p)cy6dM)+_XEgcwox>UZ$#!WIUqD57IwgIFx5cf*r;gu$w2mo
z&TJy-Hd<~86EOs#pA_J6VOUey@dFYq5fXoeyg(9LXKMuYtpsyG7yG{j9xexxBV+h*
zs%9!CptU-Zjtfi}RZ=2^AWgD5Vxl?(3$-zIG&OFLrQ1TfL`#HMr$y1lH$kJs#~JfJ
z(aXK0KPty04tU%n9I6|WBt1c(^uckHbvIqx=<F?J{uvAoBEGCk7?^{Ix7$t(4XYWm
zi#X~Nj_t?a!p}j(-8UCG$(Dz7IsG=-+BnHNk4sXl-4Vt&f%`!ulsN*<!6eGRVSlvc
zf=l0+MC&v^tE=#|pOk?lYYQ4$#snA=Ogz}Wy<mDU@lBc6nU0`X9B^|2>40;$Br_UQ
zo_7B*MstA2iM<rNmo`n6>LW{iDof3$QoB(K2kEaEB&PRaFqy;F41*CNB#fzs{1D>b
z=z1_lW_2?db1uxQC9hP14jU$!_E}u)gju0*DTD;-s`_J}9UQzelmyfY4OOz}9}3~2
zB!M+|fs9b{mL(xDCXB>-H^HoQ%Ny}playVgM^$u2b(Qsf35UW+9P^fc6GoOZ%ssG%
zlXlDtI1^5y*_6g$3nw$NGtY=1J{~PYD#h-3&3Rm89E~|oGdZK>nXnPQk09@u6R<Us
zG_0>f|C92(*WvheR9kO~0d7Q+QKp%7&}a7m>g+XT%1AKAox^ePypa@va5bnBdy2fZ
zROSU-dFX6^mP%M>OM@3Dg(L7q6q(Bo3j(KT5{|>Gbu?LmBjIE;38*<-GI3kz(wKi0
zO%|}sTbS0Iw5sckI+C(WFazj2CReb+vF79!@v4n}@vtR^1hZ|u;Cu|pX5$({?-pdf
z!`p)*Y>Od{@qXi63|Yj~fL<-g0;U8=EXijs1H_U*<~$rm613e@+Q^ycBs0%hZatIb
zo^YYz2MyTC3u?rX@ES+DqEUaW0cHaUXD4~X#5nRx(C+$jCcGM8A6e;53&(|TePr;B
zL`#HuJfyTDo$6e{>K_i-N3f+8InwEo{=W^`Xb-BSMCgZ99kOk_WD6Z7OSgr!5-kx7
z`(S%(vYHLiL#H<6DEmo$;NnRW-DkLj3ys{Nbv#LC9)KmDyiID+4zVlss4FLM_Qm$f
z!i_^~N(1QRwidk<wz{Js3Ct`ltwon#xo}I!Lk<<o^5jpoB_<}iubUK=$@(d78+!Vg
zLhdP}0mU8)t7(FcRy(LWX3h;hNg$>go{YS19)mRr<QVHx4~8U?2)3OEEKMZt%r@AZ
zNPcDB!IbtSz$X@s7Qb`Foo=_h^rtv3iaFX{vC+_9*x8;8(|u>dNwbp+cy=HG!O;<n
z=6V~0aLu_=ypAj4HSJbpQ>26<xY4>*c1`ON0iSjtY0Nb^)PZ#N85_=MZVogAqal+a
zI7vL0wY~V=;G@MA5!z`Y_;n;BYPyQJDJ$!HN_YlebR^N<c}*Bi(Y?9-oR%mTK5MEq
z5)!zOzuu8tVZ6T%r6M!?IjvSM+_$}>A~XB>EalIBLb@`{OK4yuOPTwy-AE4Vg_<<=
z*e8Wscolu2^fc`TsoP*q645hPVMP*Y!EAtElSmX(2-Qg>y6I0k+AVGi9#}#6vNkOo
z7gjsTV3|Zqgqd4Kag=?u!h8d1!lu-c*PhixU3&^{;ht<}Em_3^N!4xPnnX*4VOylN
zr)~-N?~Z(hz?DE!(B%sH@t=#2zm{q%5t5LqJ-G>u61{9LnYt}3k!XqFwpm(y;O7`z
zKq4IF(H)|bMmO^a9wy`D`wr@MBcEZ~zV1e5GiSh~J1*U0Ho~ax#19WFKI=}}FyBE@
zcTCVpc-@^uhwfwKTr_mV1kq@z=?zNODkNGWEZ885`S3iM__J#?P_G-V*<U%pnr>tk
zJH8Q2??IATZoOoLUCcsmchZ_oB2e6&M6zQTc!TCb`-l~&`nFJj_JlYzM(rd@@!u9!
zD|zr=N9_~_5Pyf(whDL<Kyuh-jJ!Em0attC=3x9c@M$m7mstkC^ui8$3fy{==&qfs
zDn#*k_B&iM|8ZHg09U7jbcc+C05`nL8sK`D=M-L{(Ii@~;<9yZk0>VYuTR%y%WIU0
zeiMBG1)9d;@c$9=dy|>EGqHF*_#L?RAtM^c(j3u#XNsdU_^E0P1&hISzaA^E|9h(|
zAfdOmzRMcQ%sXTVraP6vl~CA+^uk=b_ay;1tH<^wZEIXcXCF_-+bMY73-91tRe^hN
z5>;n^1yzeTpLmCo)j&wH!C!qzD|SL9_@@y)TdxXQr;(6k;avqrX`|~XZVcM~dt-10
znWfiopAF5>|KCufaFtrSh_RNsW6J5iHBE49ChuF9z|J({;gIuAa^Bs;D2zu1vdr&I
zWtk)qUZ;^n9l-_^>b!&aek9TFY^59}JsJ=;ytUT}QtGUOZ~NhDaubUBksXWxS^dc;
zY&3zx{mGNM9UkILQDI(WaK(F8+-UT4gI@-ae)_AgD{zR2(w;5Raue@j=pA|8chF=Y
z(VNpS7Cg^c<cE81>0UWzRa1GI{O6U-na0^U3y>pRQQ#peJek6U3anT445ILN3e2c*
z5Y6B+1vZ*%+Lgyq@!1L<t-@gx9-+Xu@a$4{vOa~A75IV*yHL1=0vl@ErE5{xSAqAc
zFmAie8i7-Aor=GRM)9(ja`?|x<twR{KNWa^D*c2?A6MXUDtwp1KPa%PliigI6kegg
z>uQ_rcmc)dDR{1`WjBRKD{zRaWetT>6u7PmFQ;&v0{2tlsT2-S;Pxs!9$~XkOTibb
z_(&@L>V+KsX%!wy;hPHV>}+?u4~0)CFsH(uDg3hnH&fx36keskpE~2sz1*ISDZW6#
zSF7Tl6dtF*=Tx{Zg?lP6=VEuIHig?La9b6wLE%6J&QxJ7h0V?izFoyDq7Z*uCWn7t
zg`ZRSjskn)EGRoJQ23MrcUR#@6yBx44^{XEh1V!>H-ycymXj1;q~I-8@d64@P+*rj
zcAGy@xQ_y7sPK9Uw^QIS6<$K&U<Gzl;kgv9V^;9#Dn6FtmCxnyhpKRY3g1`YX)4@>
z!oMo;cPiYD!h02XfeMFF_*(@YM`1HPo{I^f_-6|Kr7B*Z!jlwutqQwQxSs-NsxV98
z_6j^mh2KOX9HziyRJe@7ZqLnfoWoW8A;qho$?!lGzCz)L3fxD9Pg3}t0uNK+0t)X}
z;4Uis1BEvza5D)u)9^b{&p%i2d8&9Eg{La;Xcd+kZlD6^s&E#S?xetzRroN4BNX_5
zRCpJK>p!b7OOyW+6}M4bqlka4!Y?TN_#Zhd*(%(I4xtMQ+*9qgkre(#fxlH@KMHSB
zU`|$I#_(M!o~Pg@RXI!HX$t(_!z${^5Cx7@rKNFbRA9Xd-=ose3Y@0GS1H^8VKZI(
zQdRsniaRLc?_H6G=aZ*$`0rCL4bMLm*twp4c<!f43KaN#vNln8s{*GhCe3*<UsF6^
z!2|SmEsH5UQ-N!#@H7e!SKuG&+NCon+*N_MsPGsHH&@{GDm;k7J`@hc@b{^BD#eK+
z{;LX0{S!}EMv!K#*DLsH1nI!`c>-5Q;JLtcs5_Dv*cLA#b|m?ZsU;Ujy#Q_$Y0pl1
z3PVN{L&Jf8SK!TGku|skH#XGgXY~A3e-M{>4ZX7&MW(ZzpMq{Q2@5}nvUD40{AkDW
z0(<4-4X^On)>4A^dGrKo$P-CDlLV7T5pU)OY#2>aS`T<qAzC{)<`f#OO$;tepTETW
ziy&-MO9#it_pxBK95ZI$n{UjUJTzJ?uFlZ5YK%Lc>I|pU#?WL88CuhomMb!x^o?QV
z7&6|cRG=OmGL+NP8G7{fejDZY4>TN02C&VZ!R)bQc+DZDD2m(h&X3^HSQ5`}djyH&
z3EmEmBMtHPq5n9%lkcQp-;N_4m{Ld>PduP}9O=yD(d$8M<U?3Dp7dibLB)8|pPl&^
z^iC(^Sf{_>XgY~!&VrbZcfkw5G=Z4d<`3ZO33y9~H!2fIcLt!vL`1vXheZ<+t%mbh
zWOjg02I<ZYy$2IBNLRMrJ@_SqbY>eo1jk7vl35MSCy^Mu8yq)@bYb;(Vc#UuoE?52
zo=n0UzwOXyGU>$Zfw7bEY_ZuLSTh;xwt4^$CX+DM=>gQ4f~T6tq5l-JrI}kbJ;AS4
z&<7j#Aa2$11|dGVA@-_^f6*M)DoFZ(VWo_N&}=G+^x9QLD{}v;xQ6m4{{?fW5+Cy{
zMWFu&3Vx#kaj5xmtbpxGpG%#XABL^OJod-?0Nez3RB8?Spq7I#)^eb=s;VR*Uqxu(
z2SQz`(5(k3bVU*Z&rH&iopTdXGqJBa+=Mfkcu)xNCX<+&e0~Fi!<&<Bs6g)6I9~A>
z<Q|QA4nEw#8=7GGhUPX*%pxId%uQIGMFQA5ci=FkKG)%H7Wv%e?#=g_1j9|3HH}PU
zo8N}Nr(v#jx1s)Y5*&T=7P|Gr15rGJdvKg;{-kD>7<ORbMhl*~;WL()-^*z}-Ty`+
zd;^oF6TPk%$_d@?z|!et4f7|2%^(NZ8CT)q4B}h!Q#S^?cB!|EFSMIMHsfK)-kEr7
z`xY+GB<sx&FH;ML<j0A5@)I~YoI}@Se}nO!L+JjPis1pi{ecVK)U82*9SlBcNy7)M
z>iO^>d>o)PzLuICdq~3ujHVy~W0AUypu_l#dCLt{DLr8CIvOuZ`(hOjsY+}YYp=gB
z(J}3**M-8W>#`Fx{X#9tSM8HmR5?B-!>4#Cxy)0NoOm4)W)uCWOOo61J8EEd%CFGc
z@%!j<W0J48)9VCDYy1)_b4kWmQ9M||>9se-t=b9yHugp@8G8l=vxzs`>J{9WjZGYJ
zZRH$da0$NpfwSYU@XhAnu_`<ABAl9w(`v$HaGpne&Av1W>`_u)cS&RC-+8tDLGS9G
z2@f$6JY(sI_uzecrWh?3P1cy<WsC-&fEcYF7<qw*<V*yb#dGT;(u{i)wI^AfG3vd@
zEx>y-CG$q?B9vjyi{C;yu_`B0;pD}ipo}P$;h{2aQARhFQH`F-M%?NnLml|S1C`N<
zGKQ!%&ZrFY63PiwIlENOYRXV&7h$!^xI!5zYK{3S!v$NTV6e&<tulI3hEZj7lNfES
zJspjC=9q<Y8&;~BDx%Ups<gkV<`iYTkB(Crm6VaH)_5kh08M&x?2I48a9L$cp^Sc5
z!z>Sk!zyR@zpHFg8D*5wOEqFu8R2wHy>GuMDkGUP-e<6%%CJ&KS4kD#mf`?VIR%u{
zLp2hlGTu^#r^;|v8Re7_r!rpSIKdui?TL?eVm!R09=QQ~u!Pk{te1@CDcHE4t_Uua
z!I^eJH^lbI*hm>Gu5|4XRb1J>5IL+jqrpRwu6F4eh<$<>P7p2MOV3fHC&|)2f54Xu
zh+CUAa#5oSX47X3itH7cdH<5^@K^cBLgl_blJZ~S?gD&CV*-}Y{LiuwI+nfay@nrT
z=Cb$9<?opfWoDMdr2Qitm$iaO6gWs%P+_gw2Rl;62$hj7G1^+cT8lk~dh|K+G2ABa
z&1`ya<`^>3L6`T;_3xRlWaf*D_A%QG*S~_6swq#^hg8aV-+eR+180nU%6UJ~9!a^R
z@m!{i0jjRED#H_>s}>AX8GBX6aLQ<^GS;e$Ih5h2GO|^Mc@gEjpMAzjoVK(bPNKNd
zl0Tsk=D<k`lQKIrS2(w2&dUo@8&giK!eRZTI@cr)4W|VH1!Ghn<UV7joW+#WLghSC
zTXYj;q^XRvD&qlVWT=e2?;9r<LB)f}0Lg<i)q|c0d=RS{YnkP#)<QY&GdfDvXS9Ch
zN;e02F=vpEJ;YfK)7lfaCNZ9kB&_mxWjuN#Hd@9WA|@Q|WfK<_W_rpNzD3Rlt?Y;v
zgr&V~<_|=!Q<1T1Q1{eM)x>~|WR-DBWelT?St?_P$~aCL-Bkvtj7rLgQ5o~hDyKd^
zRxRkHaz?6*j+F7fOPi2E*Y3aVyO4pD|EXHBiCS?hWwcWn^;E`L%8<t+oi~L_wS^ks
zEkr>BwaQ&N-`Ec45ySktBOlwra-QZ~UX|sC)q@|qpM$u?B!S^!)?(7N&Q@#!qs4)~
zAA|2CxN5)PFE7TM12+Cwh|VSM>?fzen2V1eRzXfKNr)|bg?=nMg=#JKCumw>${D&@
zcq{ZjhnofUWgNO;NV4EtTJMDD6sXQ6huJCJ;gE&gV0(Q7Uw@9T3XDAre|%0_u!#&f
zSxJETuLshuD#`kV5=VRNzxbqk7!nL;Fxrv!X@@_S8z9TY$#N~I9KN!pT_MY1LNa6l
zy)57^Is8bNd(=kHu7#{aHhhnxFZ^UBu?|~KRKU!)+JNB1lNBPqt@T!j#S7)3{{H`X
zsPA<Wnk~U6dee(&B9C{p`pGw;bfZq6nB#q9?b9%232D`6HckRUtrK+X6`w(`MNDLN
zYfY9zM@u96Kq=!4oLfSCnOpE|2|h9`2W=jS^e@NxO~@{msy(6C)0JwhC5+Y*_}+P>
z9>Y4GfN@`te^|F-=)aVt)%;rW2w!f%XY5OHe;;@h9xo+JS+^pXl}`@VUNw|HnW77k
zMk7Byn^Xi7za)D@4jhphRC;v-Z*+v!a%{o|>IB|?R!9^dFo9_qX~2?WFlHHP<#4*V
z0#0-%5zJ$_w2Xu_yImlQcRq@P_}Fn;I4+!)y%GWx9oq-vQ#`fPKz<;JsjI^ZA3j&E
ze*~s~MG89AFQk5b{4U6_!&GyLU_q)rj!Qo#o6D5U-WGn9Xo=8k5X@Rmp0jn2^XV(d
ze1^&ApROb|7<Sqrz8pw<hOPSxv{_ABI45nxuKw5l3d~VQB|#2RhbkZ=jvNe{%FCf$
zKY*e2IV9T}EID#p*rzzuwy*S<ydI#?hv97HD_~nIa=6P@)c&9Gl<k${@!a=8JXHr}
z8;+8#+rmJJmIx1e%a0dbV##L?Z&2~-SklPh`Q8dRf)~N3Q5Qa`{*PAzUZ6VlosN-M
zt!5lPUBCSZLr&ZaKHrc}TK?{W#}rj{78%y#HUXar>$S)2hz4(%BxJ80xr6h+glvVc
zzah<-XK?l#;_3O?mC+1;SsfgB=E?A~xxte&4Rr_}g#=db`<Co>bWL7IUww5z{kS5A
z_gu#x-$*7f_&P)MCei~>2C_GiKsNmw*uIH$WNh$k6Y*pXN1^6sY^U>(yqWmaU5T5M
zEOvH*3*Wl9JjBr;Y~bf?CJz}l^m`bw70)PgV8>Qch6e~6w~-)ryd#|1MpBt`e8cZZ
zONLnreYWG#Ng-ryCvWhvWcv4{CmX(!Kl(jZbqLll;PNA>atIMMF!6izDUToe6P|c5
zx$xu9#O7eBQt+qyc2Zb?<Xz+~^Bg+uCcW66U+`;o6Mu${%7cPEq({A*OEA*%5PV53
zy0B>Y+%$JXIRxw_(f+4)REWiubvhbqC%W)+d{Tw|6qw~Y!ce<-W`<9Laz1k}InKHk
zTNDLD@p*fib{}-!PvYX+ywPBmp9TjmGNfQG{~lhJ6+C&5As??DG21>e|9vaqyTQnk
z8Y1xx9NtfUVGXbNSqJd^h3&JL|KS&WcJJo&oVLo~`<<ds72oN59i7oW<hvgxtr>QB
zHrzZ$ykKhqdC8_OlneY3EVN%B7fK4rc1-z{qht(QcfO3hJxU(1b<DKzkg-6~akAcN
z{#-QjVlcj|JzY2ilZ(g_)^`JlMWkcPGVF`RRcXhHU!|R<J<I(81zz<!U0l`c7y@bT
zR}jF9I4otwQJFONtB>Gm74c(P0oE3if1D2DL(5LlPid~>3jj`#pjw01RfxmyR&yEJ
zA~<z|bZN92Ws2^N?R=@|-mp$LitdkPFV-%8eBJ#^#D~Ak;sV`Q@-0e;kAqozA3Z1!
zStPw$O?}dyyGNbGF@njG`~P2dB+5P_uGkT}N5W>)YgbHH^sKknvf5?x?X@=Bkx6zW
z17D81cp6_ey1_GN$SRg?Iu+LZhQSWI3x|KhHvIH9%sNX#L-yQ4^~dC-&VdRs(Na9B
zy~VLvaivE`LkhO3b{R?t!vKCeOSVRDTZtkr`-+MWhMp|0?02-dJVHBT3XUq`VjYO@
zEn<f>Op?N$kM0UHSMr<Bkzj`PoebyBlcx2&R(#lZBUz#53f}E^l2OC_$AypcwT>eH
z%GX{y5@nBbs~u7EHFEUF`5IuCQS&v^j;Q%+hp$39TqlW<t8lM4{N>!>wr9xjGV^yC
zOB`G=kmA>+PPgy@J3=951;UwW{6(00odhw*z~u(<g@e~I!5c3^$_=82SJz1c{I3jT
zv|}eqSW~9IEVuCn%C))(MK?%Ow(&*2{05%RvMYzlEmV9?ZlRLjq!tR+|AzLJ7P^!3
z-&*L+ubAp$7;}fLj`E@sA2m?vP`QCJ{<DFe@X>eiT-7yah?F&Kaym5PkHPeNWMo7_
zw$fUL_Smq^SRs-Ev+xBW1I{$syW_BuEn^P8*j<qkbDytypY$cY*WkT{)C1OI;-oHY
zu_O5sf<xiw<0w;wQ`jVFWNfk*sdc^f54ni{Qh+lLBK@>SVBcesHEbsyxZ)fYKem`Q
zQgLu#ae8rec}el>$-lG3b2GGMh$D5fQ2xce!GR~!i;iZMpF+&NcM-*KUOW8+i$0^I
zB2I&6%8RQfzogXzrxfu^{w4_yZ1Z0H*{7tCal3UG6;=+uK<BlqeIH=x@n6BlXQYxf
z_TayLj_<R&cl$;1F|LG8Ea&Ltq3!SsM83ci%O4@-1@4N?-TBck2r#%Pi7#=dcAj^C
zMOH9u>VEj)HOXSm@Bwe|<eOQ;r@qDg5SyL^joy$j{>O6Sz%XrKK{?5S#+5ipdc)2N
z;tfwLNr`{qPPBV;*5Ff^Ykb8W*F>$)Cz7`~D5nek#=#%&@I~YeFsBL^!ttE|sz{<+
z{tnsO52nE!cv6My&NFbXCX;=uf5tT3n;V?w{?>lwQ)a&kEQN2Y$xyc8n3XotJoYN$
z*bm}&$TUVNhZoA_z})WizdsiF7Annt60=yH;l7hkw~^TnY)pIpF{69x5Y@O1_1o2H
z<yOf#y?e59;bX)Lnt0kN!)Mb%rEAP>o)Fy$C)O_l&eqd~v;L`2T~FuBYEq%0o31|I
zth8{`Mc{^bq?@i?v#V=RY5Pu^<FtiKmBO#ZhhJYG{vswg^#;P>uPA)1ljbih#OT)Y
zSKM?R8RjT>d!U{R(A7idhoy86U8~@gfTBm|WsKfK|7Kjm)gd^;sSwXshVT3<_;#7R
z2$%uRdg%Pw2>@jtx}j{X)zGuP?vt?lRIce<KlJAC>r+zQz-M->X!wix;M8*x8eRcK
z^>w3|U3{2c*Mngv1;ac~T~jvQL}T^QnpvTX2`+i+zG*Q&6w~s}^`}J(dpo8_I2PR-
zq=}in60z5>e?$Cw(ZZ3A10xI+4>u!DFXC6i)&{yL*BmSle~}!VdWNDtE8$TCUHh6-
z6{*0_`DR|ac!v1_CVT5Vm=%!gt=rIfcMvLHas44p)@RF5!9^U256g>xn|#@^xOnnQ
z2V}k}8s1x@{q8G;`=-3;*dWJIjz!0k9jBoyy*2Sa!XzJE?|RcwM3VhNxp3bMSABF-
zYxSUI6bv8p72l_!&XcK=g|?(ggU%A3T4I3tzPiv(Q$DE>aUf)Le_mWR^moi2ml62W
z7if6b1Wt29Du=7(zoc@ETK-2W$NvP+e06)+xTXANKb<Sfl<@lkFxwrP;Crw1-Os$3
zV|nym*}VoT#+%C@qH%*Q;dMhK+yQZXy_&KQWMg+nnOu|+riMdUpl%M^DGI&|)Oq8p
zxF-U23z<!j)EK)*KodxBtjl4&BH(3X9X`km<70z#ogBlVqjYbJc2@*Q`WJsD=^u<E
z>Hwx6_0vJlg`tU;{Pr;2-z+2WD<X7b9l|`?+UvS`OLesiq`Kiq4qTOe``myPv-}{W
zsV<UP&i8Ao>(7P`Yi+MM1dmZM*pzuv#bBIP6r;i){#daPZZ_AM8h;wa;6Z<q<#J+g
z{Kve+UWt~f-p0J}5eWA*g;X1}pKwB?7vDWbx3lIl<NSRII)lla^_xvx`v2j(<s%(G
zDrFw&f>N8f=D&Z<|2xKi#SeU><hWAah4d+Ybx;0Zzvlm}5crXjNTu8mX>a^a`2X;=
z*mbm1E|w}Z8)<5}O`L|`qxijq-$(eZU11Zw@Y@l;7W|IFFLoUb+|ASZ@?xHDawBur
zd{O+reeWy~#sAZn*8h-i*H?x0==Bq%wHL~iUVk8)^e^L-@v*WGgg>nKuWMQd1`pGD
z!;xJ&!uQ{$d*Ncv9cL4JjnruNBb|f$@2eAR;;%@#44asVcq~%485&LbB%3$_DL2I?
zZbF(i)g~@S{tTo$kdDd3tc^mNWfR*V&7EcwbCGUCDxkbD2kT*3KNtBkvEDqJSTqXH
z&tvD?#H+{%UT71SAR{f?Cf-1ryT~RQMx$dnHn9nwQC?ka6Vs6f=i0<QcnrMXf`+kd
zw4!0$$?Eg)%nIqMr8bf7Z1rDaAV|k7w~5bDu6%_}Oqh*&R@%gTqypH)^Rvwu;cAR<
z4hnva5h4}VpaG;O*V@D?q{eSiG2V{_ufvFu3hPnvd@OIViOq4^C`amt6VBDsh+}#G
z1)EroH0=`BM|{j5HqnT7^p|boBx>&}W(4)@N1ACyfwXHj@eoyj)QA;h|Fnq%kp^E!
zgGlu^Fj86(sc9irylE5HB0YHvGlg{9ZOjzT#N~I;0hAN&+eAG&SpJty9D}(2kxkr&
zbc}!jBEIS|W(3RIUf4u)8$9PSGNL#GX|SUx79ovAdKKvyq_2=(t&9JAmxFR1qPPTU
z8q)PhbCDiGs`n8^ZZWoLV^QpdG#BX@r2COBLb@tK6wf0y#)zUf&hBYQqmbsth-T4*
z7yFY$aTU_?9-{aFDVHjW<w#?Zy5%B&fG7qdEgvF^ZIE(9MKJ|w+Hg@^PWdCzVWhcZ
z(0~Oa8!w7Ok?Qd_U>hw@#z2sAQ_#`Rv5YhU>6ocOqL_{sGp35-e5AQZms5dx=m^qP
zIMiaTSdpJltb3cvG=Y(N&X-+M0~YJKhK&z7*+gFreiEHsv)O*m9A296+p(ucQ;57r
zl-C(-Cv+ZgN6$s!KfTxS<XSTZNcQBqvgI!DwI|mA_q}^PIbY8o>f6KuhE~B;;Wr(h
zFQ$3g#K-XaV_jpY*MRfa4)n2!ky4o-(tsPuXw!XdVjfKX9R;3wasEsK)bU2uMU7yv
zH`j$N4}|UBTvMhaf76>A;>51gLEBI+6yJ%7cH-P&UMT0xG>5!Ut~=uhH$%D3xbY4Q
z<GQm;%3)d<*OnOwd&9U6?EY$~4C7MSOAgSj3D=EX&cU}xnM`=v1f>_%g0|r(Jp$&2
za~&8B91Z8%F_VCa;JVk^(ZD8N!FU|KeBtM7Is=#^Bu6(ypo}ZrjlddcH4^ny`M|VD
zRNDlKBe~9OkuNw$p|w|jFgS|)gk9nfTcWsB2V;Lc^s2*kf`n+!G4x50O*~1Hh97zd
zdt*E{<N%x4S3)$J=189-c4+`ijOJS4F9A%5<|1nZBRn)6-GZCZTo}8*C)8|;ruERT
zDHp`9?*j{)VnjpXXj9G$TkK|2uB-Osr#7(`)N$t;L(gW|995(3%>nD0aq)~B+-%16
ztaU!nCjQw7Rqt*9>$F^bNNtX7&<1jvV^GWc!rA7ixfeLaV8{>p!QdFQ>jpo@aP8O!
z1K^(+?h|GX#I(S?=MINiEl_w29B9FH)~+9k>lOI3oF05ya*>$t<d$4(?Ma+mZ&SDQ
zaI+!j0()9=e(nQt0oE!h&?8Nsg53!ox8%G?I<h7?gY!vUeejRvBDMKAd5&FimkWY9
zv7BGiJ^BBx+I*@;vjedps!r<bBc;Ahdua;tit^!XEZ0TL;hfr!y5|8dzv^6}LmcPt
z?v3;1hkjIH1wG7+<Gi#@kaY)`YMmaco1x!@ah#vFaH~z+0ErG5Om!R=>2CPx-}Q?!
zH5xxiY=y1T90s+*R?#1`iLIcFW~E;!s=3q(!>^6g@rP!YWZ^J|8m&364&KO`UxUUX
zcbG%S%Dsr|lH+^qJYsvUyjPKctOtngKrDsTK&>Fo4?b(n`DhQ}v^@l(+i?DHx;593
zy?O)wZq2pFwh3>8Jv#v=x51VhgOhO{_@BdZdLKnSf51t(zidwGplfl>El2EYn0yvj
z=(u>UKK8vH@myH=E1b9Ok<k84m*eM%aY}VM;{+r21Tin@QH_($k$5hKJ@gXZ#B-@k
z8YH#FemDOOzpO2{y0&>4quGqp%x1>v4`<B*#_3OI&0|)3+gbAuyO-5y{zrS0#ln0p
z8-=oKXg4`)a%=n#UVmS+&TePTp_(|ZzN%ICkh5lUEp&QqZO?@+n*Fr{A2@4n)^@^U
z*4kbNv5HgRerL_kPA=!2H3d%EEzX+bPPNxLYtA{*ccJe%)%(U-^TsJ~jk9Kr^CCvW
z7j)&=TIQ5JL<-9OYc}>@ud)AHjf)bp{a0%2ze=P03XR783p92)zw@#_`)|+Ke{076
z+cI`J<u_$gHTEBoQGPT=WB-j9yP*BIVeG#JWB=_J`)|G2<rF_P8ve_^oQ^Phe)S+O
zgF{+l6nCyZyJIu_IhTuIcSQ3w=W#0;w(uIi(ah~**}^t_|Ajaou=&?udp7rvr)!`b
zgGpH$mN(XD0yLOt*tv+Cj&p5D4w3}eki#9pDR<anZU~)#7jsJ)SH8z*Tn3BGqZV!}
zn=in)&$%zyLT9M_oSViz?*>z>++fBJez#(Gi-N~iNz<?;Tz&RV63kk{T}D+s@=(<z
z+z#XFFcd<*z)}<(_=3BF6~A1Hr4;bbNA>d?^QZH1e)XImEqiQV%=Y29smnU!sP|lk
zIy5kB85(#V3A2`A3hpd{?aOd-&p!n(u#8Imzv7<Z!cb>95<g&8pl`0=xq=(Q?l8ct
z6<8-f1h%iBbsED9ETeb+E3u0gR>8QH++g;37yk50ZWM#HqgQcP8sz)P4q@h`PYs;7
z>S79?_k@7e++}wCU8r2meZeH~majSdNtdB8b}i1XlVCL^L-|u{IbViJ;U9m4>m*yK
zhgR#*XabB~$1TOMp>iGf4@y5?&qcE5zk~)GFmF*%wgDa2K#h&qE*k#SMh>5=xB|0@
zo6fFp53@FL)38q2CMmfKHlwB;Ay9V<ZJ!+2xCITQ!0%g7V___?TP2Oxw&Lakm+;E1
z+(hc+HZHP3RE!*3h|)a<50Sga`R=f5o8;x)ZQL_9-wQ5$hizV%4_&s?tjEDDi5A25
z?TGJK1_9rr?K^W}7t#jno9m=hjZ{*$bBwI^`AWY34_rfr-O-Sr^&>}Uju-rd5#p5m
zGwnV6fuA`YgOMHG!EL2Z?!<Ttec;|sY_Iv(Aaoaok^;<bsW&C<#%|=w`|sh**yCHm
z#=Y2lcl?0e$8EvN-|a&k3DASbKAi$PdG1QnLaQ9_mx^N}a4SvQaz~)-FrF1kpE(M?
zek~l_&lPw^rONg8DtQ`@J7qeI{P-^qa4w(WulsCBs1aF1A0w;L^Vbq;1UddM%|cCC

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index ad21109ac71e6..d385ef79a5707 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -565,7 +565,10 @@ struct ggml_context * ctx_data = NULL;
             uint32_t n_dims = cur->n_dims;
             tensor.type = cur->type;
             tensor.ne.resize(n_dims);
-            memcpy(tensor.ne.data(), &cur->ne[0], sizeof(tensor.ne[0]) * n_dims);
+            for (uint32_t j = 0; j < n_dims; ++j) {
+            tensor.ne[j] = cur->ne[j];
+            }
+            
             if (n_dims < 1 || n_dims > 2) {
                 throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
             }

From 4c0f64e30271a21f1d779ba0506a58d27c9858b0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 18:07:41 +0300
Subject: [PATCH 087/242] rm binary commited by mistake

---
 gguf-llama-simple | Bin 607488 -> 0 bytes
 1 file changed, 0 insertions(+), 0 deletions(-)
 delete mode 100644 gguf-llama-simple

diff --git a/gguf-llama-simple b/gguf-llama-simple
deleted file mode 100644
index bfa61de57dbc77cfb6c180b20a23a46dfebaf99f..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 607488
zcmeFa3tUxI7C(N#i{h($eHE3}t*Edk_=L>5$is8>qNHS)7AS&H2m}{J^MPF`Pq#Z!
z*<?*M_L{NBjMP{`4+9lJd!aRDH5HoGy^)6IC|b$yyY@ck+<lK6wP}9;&*%UD4}N6t
z^*!tCv-etSuf6x$`y7_1xKhKz!X)jdhjgt(lx+q_9O?zfxZY|`k{nXJ)EB?clg^cT
z0=3}3L%nKD^TYWCEsa$9dLX6y)~eTy=}?tVOSKwOO81%lbWY$`v{bTjn(m9BY&xfn
z>9gNZOC(84#rxRv9KHCBhP2;6UNbEf@1qiJH}DqQazoQ>D=#jxpdZhu^(1l7eX%!l
z{@D4Usa=(qmh!7yCRESO>UCpk$2Iz~v)@4(rhns}OZP?ogSXF^77pY$XsIUq>v12K
zm|EaT6192KRP}jga_Q*93(`_9T}isCuwc&UkyjPYy`r$7sJ!Beiulo2j2?MqS@D&_
znLSYX#5-Z~4XgmNFI5ye{U_vzht<1sg|s3}swC_&{C^z(x97e;-&%FfxBj;euH5y=
zj@JR1{_HzVl2p-z@FYQu-QguB{2j1^?)ZJc2ZWT&iBfdK?(nftOn2o>hk<m5D=@q6
zaHmQAFYDVq|Gf}$cjf$m>UM`)P3p5Bis_Djt%-czHF<6V457PnENEAEa$9K9-qCOt
z-Ien?468f5-lUvaCiTDGr2J&)x4ZJsGqImVCUXA7g#Vxk?l6)63nq9L`e}FdEHP<s
zZxg+5FyVjO#GV^X^!}@f{JoHKcY6QCq&^uY_H%<tJ<l@1-!ZArlP31_vkAVjfA@0!
z$^;)~qTgr}e2s}8cmezubmM;>HEFNSq<n`-J$sv!lVgJ0P4qI_#2$8;l=Hm_KE=cy
zo;2ZKXW|cgn$%M<>DO17@IPQ8w<eSNEHuIIFzKhQCipfJJ#II_FEe@WrzZHL@F(5%
z$J<Q!dzi>=l?neY6Tk7ciGGKg^t-Jl@~J_;?ymm%CiY-4kz0pJJ#R7bxBE@%Q)?oJ
zSQC7y3H}=J?%MULiTock>34M|<64->bBCMQ|7j-p7?X0YGLf6zg#Skqyv@WP4lu!!
zOv+gey?5t7KQ!S#&!qfUOzi3x6T2FO$!&Mza=Z!u6qEWKGx0wMOzfe`q?{Ed&%MiJ
z-g2Eue~dD*hdWK`d9z79=i<5D>1BjTJ)bfu=Qfk^{U#HA9W&AId8m&a|BXM_o7Cr8
z6FCnt;eXGhKI2X5<1>+eACvNzn)tC<CU!X9M2}~gl>Zg{Om}|iTN8d1->p2yn&9u6
z$g{)bx#LWpTV!H~-<gaTEs#%l?fTioe;!6XyR%!5N&VZ^@vidvP|Kp#ad-Yf!uq<*
zJvZmF%V<qKTFT0rH@~<jtIR#S)SZ<jWjQm_vgYQN=FTf9bLW<3q$L#=7v*Nmo>Q31
zuXTH+#GPN7J9}<c4t~4An56-?<u1}d4Z5|sph(KPY4S98+{ihz%L;O`3i8U_xN*KS
z2UuKA{_N5$cj@c`cbPLMB_$~?W$Gj_kIBl)si=sHiyO{w!Yu_w^KeUC*0kBgn1Y)}
zWI2n=vZj>e7R@i7i+bENdD02)Jnmt<pfNec^Gjxz=BA`i%Hb6qU09qmyD&F7E@c6L
zY3|{%Sy^at@xrXa+@g8z{H)y4(&AFiKFwW<dUw;1sZ>9;!D@RmvNB?`Ql^d0nmTO+
zM1NDp^l9#d+=`ss5_dtdDkRf(o76rbD=lkwSy{onBFJspNGkS3(mPQpC6MD>lEQpd
z3|z}`qwolBFB~Ar$Ceir++Lmw#g{rKPs<veHO&o6perNi=H|^VFLcAI3Ul4LPADuT
zB|UY)f)pmp1X$Vpf+8@JY*E$GywKsXu!Fe^p@*!T;-Z}L($d_boJC9llcyDwmZeNB
z;Qa9gP&wq9n>?Id&2Soon2|xXI0aK<<rR!7%gw!Q9%?=^ucWxl85ft8H6J|&70hx|
zmC0~Nl$IB{3+Cr?#h=tfO@xfnVAHd+=9QzRS+jF)FE1!%Mmw#Tv{+D7mRss}&QD31
zI;}Y4KU8GWa8e*Fff<-+3Gre-GVg%APoRma=_h1OorCU~o2%9@Yj)AxtkT>$vkPYz
z<>byuhjij~mLl{*rkiQ8uyM5sr=$x!5cMWe8@n{R-$<?d6%;L)U049wmCh@lpIhYS
zT@dy$lH5^BdQKr%nBIv-iH)99p4VMZRlNWi`J$p66f$46|KruJJ<r@bnW$c~Rdo2c
zV4dItP3pu9??3Qmqe}A%)IO{>Lhmjkokaz1)uoAcVf3#iEr^SSfJ$>q3TNl!rp%uY
zNzen}ZRh2>5xLDZm#n!X9FsNGye9Bv(?(@wkoVwjEn~XK9F!u41rsT%0|n}tks;pu
zo4Nw2qG&dgCKM7kJc|q*qgY-+AsU@Brl2gV7><+r3+vr%pkShjgBdakcfe)~F{)6f
zLOUQ#8r^2kDOeCkgG*UKR!$+jc$Rxn33ar*V$D~~!N6RUl{4F&lh4Ww4n<{)FksK;
z9*JBd`9XulhWeX1#iefWiGnyq-H$HKEh|_IwbS@Gw=kW%hW{rt&)OU}T9jkX>@v3|
zYwk%)G0<Q<;B8Zf6LmmR``v%g=J=B066h`GHeInyfix@#BX-ujii)g~+|sh*A~+`Z
zqO1k6s73~cL~UHtMn+ycgMxzHQLXk!nk_Nc`yW(!_O##>M)!cf$=218E$g^R*YGS1
zC?->`X$8zT<W^u3m!pnf1yk7o28)F5iuq`x4R9x!EUDa1eX^8JgBdyYxwDz&(Hw77
z4)qjtKa3Y?Sqls1<`$v&0yh^O7~-=q|8UPPC`5-zFQuWoG&g^CS$<YYabZEuqPUT`
zttdB_hBAWT(KM%jDl-RNZv~^=xrK%E^vCLAKgFJ<&pqeQ*Jf`;3%m8)>WKUwm<7!(
z_(VNrDz%b%snXn_p9+qz#pUj-;=HWV*+uhmv-0NC__~-I?r&`vJ#>CKn-dCy1twQA
zDPhhccWznwqy?JL_0oTnwRST<7aO1s*8jmMI?_0ZY7=?#`6gpcO3>8=r}VsrOyHPW
z3vOJPGtJF~a@b@By``I}6y%Q!37QKmES`tKC%6D%h3C=yEa;Z7<eSe|d+ymq`FRrV
z8J;JV-CpX(Z;UPspkFLkDCHEF<w-^8z_ar(JC|}RO7f(V;)QvX!;L6zUQSL{85J2P
zW!+G;5Oa*IskvpCgV9BGrAL?H{itq5#aXn3VyhZ$9ax6$m<Pd<mn<rgG#EybTS45n
z8KCka2BQ2N-j=+)!t%0wDUU{5RIjX{B5Qu`e7N0tMbdng8}%<Nl;CfQa!M9SCFSli
z`bA4jRIDr~A00lYq?~zjDTkFWz$8Vw*If)TQBo+)pFg`q;`c2o$Ss^J<zc#0CSi4m
zn~4bvLbXR>WT+@`OQ;fBD&^6tLYmJWv7oG|1dCS^zI!e#RLY||5hqnL5B0#{#U5CW
zCu&7wQC>>zq8gCh6%|W)3rh>I+(tcT&jsV$1qHZ~H6shvq6G?y=1=0i9O8z27Gr44
zs-Pu}1h&$`g896j^NSbc>OwEeEzFbV<jhBN-KA>6tPJ;(4%8Lk{MlH|W9or5TAnl^
zDJg3NbUwl59G{dm{L0}8!R(k|_R3Lue#Dg{c|J1}caFSrxO%fn#;JE{WPB*|=nyh4
zE`&^wk}@+B$2+s)t{fp`wTEjD483XiaIp-YM?<uFARBB)+(?SJgiAeeE+HKMB9QAT
z^}=uZ*OUFDtKrCTI+EohrQYftg!W<g5?>D~lKrQ$!g21VkJOLRy}{oTcl1ZvLpqIJ
zjgYKJsci9oIPU4ou2Tv8$-HD<dQTi$p}Gvvd3qv7)KjN;=~Rz!R>MJR88*;Vhj3{y
zep4EWe-@P1m(>{|E@=qR)7e#OAGItT*QpMgl!zjnFp?Gmiueh`za+%kw0K_n#fuQh
z93sskfL-q)9YoBHNC*Ba_x}-(J6)O!fUvt6ZIg;sy8NsHL~?sei|HbM!tktYwftDn
z?lbB7VCk+9ev8207x(b88<3WCK8z*|k{(d`=P`Pq^rTAXGkSnjtI`V?-A{U3rDrp`
zuk^l3E6azWygt$wD*fTQ+Ykwil-gCgd^_zp_L2^%^zH0^NDV<mKvF5ABcy>U?H=|3
zB7i-l^VJ|@806ZZwioQu!%RL_=?azqjDiA4^E4@5rB|`*{iX3L&E*%a%R@WQgom)7
z4)wfPlzPrd!)r77b-Mv?Q}GxBUaOWP8}L{<AcCJv1Kyg;%g;98wRs%RH^EB`c;x|J
zex(7AJ;?Eu20Xh5x8NEBF1^nAR~zsWm4B@PcUX8iwFW$U5Xb9`cq1>r!GM>j{PsL9
zc}@O~!MvOp170$O<FN+3bqg;)-hfxC{IUTrIh~i2Zoq5K;CQA1FL|GrpKZX`sQe`c
zyf%uLQ(?f<&*XTe0k8asm%q}8pT+st81S~UIlk6_JI>*Fg8^?{PhL%uS`2vYYaDMi
z;P%HkzSDr$48q0$e*6YJdkg39FyO6QIo@f&JJjb&^FsCBHiVaNGvI4lc{x!AT>6aT
zb^~tP#_<>f?l_Z|A8)|zyE(tZfVcWNE*o%%`rLE_ZX3$W&o<zWUwJwC2D~zY^OqQK
zTY&Rd81UHhIDd@+uT=FCn;$CA8}C!wYrqR;aol0R=c~AEz)Mv;+kh`r@q7cmM8!)C
z_zD%TFyMEoc%=dNs`yF+UZdhQ2K*rvUv0o2Rq-_j{Ld=B)_^~w;<X0+ITf!n;4i9p
zg8{Eq@m2%AXgZhwP6NJP<!>|Kbt>*R;BTsUhXHR?@lFHYq~cORsGV$8@kj&yo{HNH
z_;wYKGT{GIak~NkOvPgi_?IdkYruD_c)S5`S8;~{?@)2sfFD-zbOU}&#WM|fpPRW}
zvJH5Yisu{f!75&2z|T?f3Il$bidP!&5h}jYfM28HH3odVimx`{sVct4fKOEMwFZ2O
ziq{(O8&$l{fX`C#1_M4<#aj$`k&3q(@P#VA(|}j2c$)#gN5%aHe6@;q81TQUc&7oc
zRdMOoQ2XDY;*kb?lZx96_<JfIWxzjFak~NkRK;Tq_%0QXHQ)gik2m1^Ror2~kEpn8
zz%93Mdrmjt{Z%~EfS;q{*#`V#70)-|qg1@afXAtLg#jO{;*|zGRmE2t@Tn?ZW591w
z@zn<WRux}kz?Z6cg8_d?#aj&cdKGUq;E$^KP7}P%fIp$~`wjT3D&AqhUsv%?1HM7U
zrQ1U7cB6_%8t^SDZZqKTsCbkC-=X4m1OBm!#~AQ`sd%gb->u^D2K)yVcNp+PDlQxF
zh%D7^4fyFQo@u}bs(7{mKUc-`4fs_mUSfh*81NA)f29E*qv9(Kc#?|O7;vYGuQuS<
ztN0oNK32uo8t_>vUTeVTsd$|McdK}V0k2f?76X32inki@M^t>L0e@b_+YI=dD(*Mn
zEh^q&z&};-P6NJ6#ihbf``@SHkp}#*irWmhl+Eob%7FJ(ak~LOUBzPz`1vXxYrrp8
z@puD%wTe3oc&dua27HE!ryKBG70)!_Wh$O+z!OzG-+(Vs@e%{>QSk}`?o;tf1OAAL
zuQcF)Rq+}F{<ezO8StGd-D1E?b|RdFpH>52qvCA_++N1@+iAd~)c4UMwUm`ANwsP@
zHUr+O;&uaGtCpYc4z)v#Zcv{~t9g1iOaD>JG2&Xg37?Mt6ceK#t-eOwFO<($h@3x0
zC||_Wi+Mc><C5h3nF2nDL+p39fYUo>+E2cKpQ(XJO9cE00k05ndQVOJsTA-S4Me(9
zz>@^LM!+Wv_-X;q67V$wzD&T^3V4-(*9!O_1iVhbYXm%7)jR1$#Pdz?2BDk>gmPL0
z{6PV474U}ye5ZgvEZ}VdzFNTj0{)1AcL?~S0-h<<f2n{+m2mr^UPW!ze(VCyBToEE
zjDX`vuKp7%;5<^!FU1Qu@3|az2sp0+$7KOW;9LJm7jV3Cp#NkFI0Di7Pqu*LNU{Et
zFW?9)>pvv|j&QX8Qz76u(yISd3b@v%>EcQOS3}(Fo*Drc&$+G^@WFcNlC(y^hY0vu
z0Y6>9YX$ra0k0GAC;@K}aM6#o2>4k7f2)9>E#Nx^Tze;)F1HEzIRd|5z|R%%4gsgL
z=h{!FfZH{Y(%XY}d%l223it&AZWHhe1w2Z?qXpb9;1>yajDTM(;IRUJiGaro_%H!?
z2>7J}E(`c&0-i45mkW5NfL|ft*#ds0faeSNRRUfj;IRT;A>eTWUMb+i1$?D|j}Y)0
z0cRVL>b2DZK1$$UBjBS2e64_w5%5|8j~DPd0Z$O{1_8fXz*_{I&gN@Btpa|H1|r=l
z;Nt|mO~9`eaKC_GC*U0d?hx=!0Z$ZgsWfQ+;{`lYz>@{sCg3Ro9wp$Z0&W-Z2?8D?
z;Ie?n3b<3i;{|-8fI9?yl7Pzs?h^2H0Z$X~OaY%F;MoG6F5vkBe!YN~2>4V1uMqHQ
z0$wTL83Mji!0DX^?Wab-Z`44fs|8%$dSqAE2)MfS$ndoSev?p6t$@!I@HzpXCEyJL
zezSnL2>2}m-YVeP0=`qgXA5|nfX@+dzkufmc!z+`74S|0&lPZ~ENK6E0v;*g^90-`
z;Q0a`CEx`DZWr)d1w2N;ZxirX0WTErcmba;;0^&V5^!0-iv>Jgz)J)?Q^0Q*@N5Au
z74UokFB9+*0e1^{g@Bg}c%^_Z5b%`(zEHqx1iV7PR}1(e0be8Fiv@hGfZrkDwF16G
z!0QBjxqvqac%^{12>1#CZx!%61$?J~djz~qz^et^FW`3xc!z-BE#RF3zEZ#?bw1w<
zm?Gej0)CHx+XUPz;86m8uYlVH+$Z2M0)C%>#|rrU0v<2ms|4I3;C~eG|7HItf&WS1
ze-ikg1pX(1|4HEgzY_S#dhSu#^FyTUjd<z2NJ*}4aEEub%AU=Uo0w3#M(qaNb<rOD
zww~uej%YTs543gR=c3Pu4g+(bmC-wh4kx;S(c6gbL3AynHxf;u%fK2&zfN=n(KU>I
zndqKGS2Fr}qG|dSC}H$dMAM=qkj>~viSA8wI-?&Tx)0F~M&CnpU!r3feJ9cVh_*9&
zG12{rwlTVtXbMdRBu3v#w3X<NV*oCiLo|gV18t1HiRb}Dw=#Mv(G+?NG%$K1(SwMt
zW%PKW2NS)9(N_~qp~pZCqlXhsA;&-^qc0_zLXCkEMqfZQg%|_bj6RF#Gl@=T^gyC1
zq!@58x)0G5N({s@I-F<<AqMP>K6)-_3LOS)j6Og#g$x4{qrWGbLWO~jU#b1OiMA8n
z#^}$8KA-4TM(-f{0-_rjy^ZJ#iLPbzMxrSM7+AyT*NMJ}=o&`9O!UP>S2Fr}qAwx3
zgwanC9Yb_BqaP)D7}4pBet_spiFPph9-=QJI+oFQ5`8(*c1ABI`U;|Lj4mabLU{p+
z(YF$P7115PF#Quvp}RmEqi-TQj_6iKPbHc{b%6#(Pb8W`bb(q%k0*L0(Q6oeHPIB3
z3)C=rIMEb}3sf@tQlcpY7bs!$1w>QmEs)LVvxugUTOgg$1Bs?kTfo8SK15T9EfCA-
zaH1)+7O*q==sBP%q!zF-`T)@sN()Gg{+?(Gp#?gAX8I@EL3A6VKO;Jk=vGGWAbLE}
z4UFDKbQ00EjNV8zh0FqL82viYDMZ&W`emY1iLPYy^F&V|x`feB5iJv)&FDvob`qV=
z=m&_NNVJ2|_Ygga=vYSINwkY-JEIp9okp~c(WOLBCR$?jtwc{Dy5lI*KhYHG3bZl$
zCZew=x|PvWiKft2pn=g7iKdWNpqA0&iOwK;4Wq9nnnGBC8b%K%dOFdSjJ}j;3S|XK
z7<~cJ6v7H*Gx{u|DRdP`XY@d#DP$FJFuD)X6siiuGCG`S3Q+~@j6OOPG=-)DHbx&H
znnF?miP7H^okeuV5vG5lvx#nF^k+oRCc2f;JBXe`bOWQe5uHPHEu%LQO`)g28b-fP
zbS}{~jDDHuJfbTZ{XEh0h%RCDQ$*7M6v$@uqeK@FozCb7h`yC*2cz#H`Zl6t8GR?w
zg+$vKy_o3vMB5l$N;HLv0urNdCAygCj!ve3qA4^KXk+wEMBh$yE2F0pO`)Jb1EVJr
zO(CE_Eu+U1?IwB+qpv2KLOy{SMh_=?0nwF=zLaPR@dQd3eF4!F+6iPc`YfU;q!UPI
z^gyC1loN0;x)0G5!U@DOI-F<<-3086K6*Cjr9|5peSqj?L`#hRo@ff`1Ue2g{S#eD
zbQ_~TBYFkVt&HA5^qoXEFnSx&9-?a*y^-iDqSr9`b)u_@uHpQrM@o4$T^H>LNpYS1
zOp|@*)Z-3W`O4o0|15jQ4Lbv*Z^)et{c>jDIhbvp^*nlnBoTj1`C!Ih%lVI>VDL}X
z_!o)%G#`rF>r}SiA}hOP&z_F-43{q=?_WJ}OaC7~r(|3U%Iz9biee7dfE-2z`64c$
zB%#wi1XHQW^axC)dUv&1&#R<*Y*JJ3A32NoM;)MK*fCk@kQ?`1CpUKXki)jf+mE>i
zqkt>a0wTNG&=hz}?S86X<+xXHuT*~74YFt4qeGCDm2cerAc|$c{q-2O{d1wJEfGG@
zVK?L6koyA*aZ#<$4T(1<PEWicF=Lv%;)kdnRMbz(ud?sWXg_N0Ps2pa^J|#9x9ly8
zti1M$@-eb^1ln-Xw>_n}2LHK-y1L#91EYS!(B6oX`Q;UxsjQiavl4Gkyd^Q~=1rbo
z!<Sr0Ty<dgUkuOc`L##s02FH@w3ATszq0Q!M#!Ewqw$IsT4{x>{)w+Mlg9T!Rn}ME
zj2?u-{j--@*mKuVMfc!CRI-;BFMF$^+W^W+a<rrA7h6>{u^@AAbOXx_a7Ejjei>ki
zZt7g<zx7><<W#mmUFFe`yk{`f;*G~cZ>NX)hIo;nsXb6RHbXr1(p}17dccey=mAYG
zWqoud9xwZrMCZ$Rpz>KgiOxUc@}82*3()1A8(kuMlcOuB=IJiwM}OM12r1d;_{F8{
zat%8oHwMCG?~t}Zk+`8xVnWN(3sBD=_po|?jl`vVfU0IfN=z$(N~X}N=#}ib<Nghn
z4J1_dl9Ef=>YtAxJ}u!p_oXi1s1Y%}P{7aL5%gfoOBN7^QBM>&<{%W~*<^7gd|loK
z8J&R&Kt}$n;2>N|VRUBuX(<W&Twdg?RrKX7l;Bd9M9coSu8IKXy{rNWEmq%${G$I^
z;Az8taW?)Co(ygMfI6{zjwLI*l}7)eaS>8=gSGl&l|A)!65(d5*xAS`O}Mx+fQ#0u
za>!85`9k(t9!DXN?llho8E_=B(g<v(>>Ij5*_RU6<y78tDbGZ&CLU;_25EbAUQM!+
z&~$V_qNO<z-H0l&SN1gab9Obm5{_A`hNAGzD;C=<Z^{XumVf9P)-KnPQ^e15IboBv
zY6j}yRK9dJ9t)T2UU;7WxUB0sTuO)2b1b|pg6v(I=o=g>Z%{Ayk(K>&!hr>&oo{N0
zE3Bg|A+36k`x;axIeP9ypXFJq4r`4QolI7mnvU9VrG2QZY!A7ywArb=i+jsA$L*cs
z8@V7=S@Mds`N_otl3%fmZ%%%NwM(x4eA#I-X=NYm&OaR^TUzyAcO*%4>%Y6YCi+~j
zNRxb)xw3CoHCnXS<yB?n<+2i_QvE#~OuHp+Z`@WH%FmaF1?0vfLi=Uwx)xb+MVD-W
zDMS5BqBCWmo9vrR;+{r}l&Ea+Kk#=-n-UYYSgY=%wh~{ma>%dX0$j{97xjXs$o%T|
zLjc~0CvDv7>QP_nCF#zDX7_*;WlWlHo<*+STYg<&0X#f?IT$Ty_75OFWmB>e3)69W
zHe2Wk>2h_KJHr1@ShP!-i!zk2aI3ZI8X6lQj5l#1a1|RUA}ZJeBk2>xyvLwc+jjMo
z{9{Ob$}jZHcLL#HCNWOk%RNvfxcw=9ixjulf9_(7lvdqkt@;<bq)X}Y-%Q1S00Qkf
zleY&wI2(n+)0~e4&3y#=4qStdiN`%gkBdLR@|Bdo9C_BY+Rti_YCpYwaQi?`_T`DC
zeL=gM+?fZmmfRP$%M1t7-miUT`;2yoW(!$4>V6JqD!+(4Fmv~}z+a%(`~Si^(2xGV
z|AZ&=l0GNt$^OR((b%t2*%yMHo~hwysCX9eyeqY#n2lp>xXGB7E-ODb_o9B}|Nf^4
zNlCIab;d+2|Ej0d)Hxt2Qm%s#+45UflA@YAZrloZlD+9?Tf}X|^YehD*yf~oNvga~
z_9c~o$sso<$<)<|F}*n{o32ql-{mWFG$&QWuxl|9W7TIYk?Y=j_ulIqfMvyVFS|O(
zyNXgg<sxu3dmaNtKX{&J4N{lyX{wk@d3r4px!F^TgnqVoScRKCEsUgEz0r!S?DMoy
zqIek9Y>AFy<hOA<{H3g0^9$a7(X!$t)}~Hd!~$MLNq!yQ$tt{ZPC=2Zq}#x-lWz91
zTeeux8GIf(ONP=raG{wI<wds@HG6shE-GGFgerF85=Bq8czTg|aF3neW22<m6GbR~
zwlJKoJNWfjem!2lPB>l9=GW8t^-TRb;dH%{UoYX;EA;Dx)Acp{dJVt6TE9*>+J9pM
z)aI$>R~yu;xNRpPil>ENXjLz4VK^1r$*=qQ^$z_y;b>hXJT<k~hLj$iho%#ROY~}J
zShME>MlvS#rBGtJm@a}ThSDvbD~PGtLoaL74+Y9-mV-IYCiooL`vzHt3nOZT|Mm{*
zs)zks@iglD*Q4*(ko`}C@yp(~Sxwb?!;?k$9k>A{uokNAWq1q~<oySU#N~aX4oMmY
zr|PRIQ+<e>_+h_=-%BmiM9l8};YU_x3CAlbVfaTJPv>~Hj=#zASdKe%{27khIBwVR
zdpS-9i04W=?&i4I3D|RIah&xZAL})|KFPq*Q`2Sd9n3|aq0UcT-s}7^bWOp4)m7es
z>kipFO_m#X_lP^>PuttoMJBVkv0^|^cFnV?L{9kGEzcyK1orNO2f~l;gq)r&Hb~Ds
z$mzY7@~(u#?soqcu(j_|Wek~GtJh+T3%r0|tej2ydP38m%eT-bdzag2I5}G=pXytI
z(t`CJi|R(I)qUcJkm`=bq$-wb+lSZoI^1Y@p3;eRN3VI25{7@vicK}-3pam#d%uRT
z&0jA%r$K5uvaDf~@cil>)<2NYl~&JDoZ*h!8hDhbM$fJ`*@tD2XLmzNT!YKk{~vgf
z80&(~Pc0o`=|#U?y=+atq_#BqZ`?6w$I@XB3|_%V2$zu-g)`YHnbj+nUEF!&9(FNW
zVI=Ct$bOG(esalx!Kw5+In_?d+*Ai8uGDO&r@`S$*jQdICp5ZaFzgLOO?3GtTVJzG
z!R*Cmxf4?%%OB*fMmeFS+>ePgmN5<~ZY#Ac(t0oRg+95{scdyN26{NV8dH>(L{D2U
z*>fbGHT)4Y-09iX8MjrC4^{z|`!;{Q^c>4%S`j4QXD1?gvW-&?PT4t?&8b+Ts_v^q
zc^q(XV2uu>bD%*7vZbbD#Z~vU=|BlY-{^398au7k6&Uu?usV!_{K2Nz2ON>ku9oul
zL~2P(kHm!5rI#tFy;%RFjDHc;AN-ae>ykYQ+pWGXJQ3KiZ7$4z4{w*l8lm!D?hCa+
zn2GjoZabTwUNpp#Dml9v<gRVy-#~D^taole0mAdT;U_5hZcDI?<kWDQZ~}}JL5m9Z
z6TzM;XcNIm6&x&reN-??1pBGrP!T*0YHo;hC1Amj2GNqK==OxZ5h6eAbwtr*H1ch@
zD#`!1@8KTnJ|Fq&CwG2%c9ML!Fxn;;xS}KdY~j=@H~O(MKRQqjYn8)V;OwJLQr^w1
zyblgf`})~GZMFx?bBwoQ@Y2fbWnNyOZ-6|O?yu^oCYyqEy64!SvP%#%@L4M18Rq|t
zNwPPi(Jm`D+R3&Xn|>M4OIA8(kQ6Y<Tpyt=XJ7uBO)@tjK}3U+`uDK9$AmKmUdcbm
zPcZ>0BpyH`k!M`RoJcIRs^c-_A(+y_?q46>0P(r7l#QkC6|CO|8UipNVU>ZB|BjNo
zTE>x4T0NO4fJxn@9C9f~{VUK?EV>7I`=5a)bTumdtw=pbB1>QQ_K!q%#gQ%)wlEh@
zU2S}-X3wF|yH<Q24e#Rf2H<(#{_AE(Vrf}!qvt)1{E8zKW+;8l8*x9oVy*fZwXNQ2
zt$H6RCg(TxrkU0`BaJmkiQCgYw8k586K(_B2(UT3HY*#Q2_IQKPmpvFKx*9G7i+(#
znyXHi?11o7y|tba_lGXui>FJN4#w?mKRqbFt|o8)ekd{Fy|QP7wk0dO5*0rrI;&lf
zDAiPK;Z@Kf+V)1Gw;G?0k$iGkqT-`mqEc8bwLje)@mw}2{It)+d}+aStT67x2OCiR
zn=k1l`HOKsg{`i|vPo9H+d#W0va-?t8d){)bL(RJNYV&PwDnb2ba-CxMg*OF<6a5J
zt#gS55t@6yL%L!!S&k6>fgxVjgA6ee{Ijr**)R-6vna%I%z;nQMBH#RT(jKHHfFiK
zaUU#q9+LWPSd&2-MNp}K#y5<Ao6g_D`9DPh{xQT)e<lI^*Ao9Q&i|;+znb$u$N75^
ze;wyPjrfo35#ouSaRpiU4wCi31`J1t%Xt?@$=)+$@03WFcLG9L<8Hw1aSZ{>*VLUH
z5@vf?>l}3vpKn|RE<)Ho#^Z0>P2<2J_Z2X&a{)1a?}8|&_jIRsa%7q}6_ft2Q0>=<
z(YwF~=JGF53|}vfN@d0HDd;VM2f&n9BYP)AVf~G&;Fq0##gNO={Ryr!lrhC@BK{cE
zUZyoW+Tj~@+ZU*!RsiE7e#^(W)IpbWM*a~dA@@EYH-2w-_lnygA3lU&z>dR*FrLfS
z2^eo0tg@%AgUd(G*}_WmB8)f0%FDKFL;h)2SigH%VcS??J*lw$n^4w%++0O=DzW*`
zLhyPq`+<LAbTfrxy_UvJc{PlF4fLG-$0F~rAm-EG3GSu?m-buHbo3UmuphnsjDO4#
z@|wZ>n+^0|nO9}ky}eN4jlt*Dy$fe1Tc7T<K0p>ArF*)pWoOpihX6UiGOL`^j2rIa
z6`nyJcsM#B4=BcdiE8_XBqD)bzKB&`eTj|P<zK#qo!t)w*3{9ITk!YLPpmw4LcH~1
zdYl6#O^78u_1{FX=&_dV<>^pgJlB|SR2tPZR7L%milChGi`#pvdMZB~<v4KDlqi~A
z^{3nIa>4}Jx~gGQ943^yyr}TnFgamLEX{XsbjS(Y7xt3J-YD}5QGIOk*i>7&9l0pv
zqRLN4&W@bD93g2~*Vxn;wb;<{#q)zqYZ<A-+8A-stikZg%g(Lo=<0ta^8U@}r25rf
z?CSEbE5{w2iGa4X>fgAJMsXYSe*W(uIrO16Tug~O<nmqg@yyet&AS(6M)>f#o2Dbn
zlcQr@zAGAL+Hltu^)m;8jkXBTe-o{*H8&k8Y5@13olTv?W+5_sbimn7yJtj<qkAbF
z*N><XMD3o@|0;&shFpep^X}QtZEHF@qtX8?s@rrlGhzt4I)m;pOw_SM;FyW&%O=dZ
zP*P`1GCgTV#CIPcs`D+o-PZqO5Gbax>A)>m1;<dB8l}^eeDl`@`ka5W2W4C=;R$G$
z|9SrC{*ThVYO%L$*B-x(trNCNbWgu4CI1rQ-v1-5A;Psa1gq1P7`>XYA_}KKuX6>g
zJPd>W7gzxaf?l>Oge+#{xOjNd*JCna$%p)Jeig!z297NeH{CQ?3d?Jbm_upUmWW#@
z-4c<d!kG-l^&8de)A{w8JXyt)BQv0S7eY*xu$j2eA<u@eNGWbd+}6r#!X>x6a!Gi$
z<X&3&W_SY_5E)kR8<d@F%v$vzrz$!1IH%e;^;b?gFcARfOPpF0q-bI;uh@vssY&vR
zqcmujdjt6Efv~2*PK@ypa@fWI4f(RNP4;XH13XoQv3?1VC*xPb#Vel1@b*HE_Xx#%
zv}bVK5{g^eui<z^C?3%sjVkt(eXGK$4eh6MF1W5>Jv@!Ql&xFBQMZ1m@-CDxD=gCK
zwWfLJMY?>qTb-V75wvbYAlYS2*<X&(QF1h#6SZ+kv~2@rUCIIfy<gB2!#&8;+0Q-1
z(|LONneb>Y{?Jou#~_G1+DCz2Pc)uBWHB;xwtp3Q$Cjp_XGBzBzKPN6d4L#A-bGRy
z=KuWjuCBNSJT}q#>J*zx8G&G4q;*x2ZRMl^D<=(XI)Zr6uLF_>(XSB*@};vnp$hi3
z*7EUi7vJBAWv#Lk!ywlD+J2Kd`+fB&Ir?}!ZT(Jq=#VBP@V!oab*esg0Br%lL)YkB
zf8bny=3HZmYdE;#&>0u{!(hw)AO6Jx3c7!!SdQ+evQQOxu^3^zMFpKe{Ri6we>R5k
zE&m!eZ0SM`X(!<*8J902L!<X|`WcOG=k!61{)W@ZG`i6JUvYYsMt{obJsSNXr$?gC
zF#h*AU8T{@oc>6o8#x_Kvm`424Nki?`V~%Z*64Md9!%3b;(w0Qb2a*DPOsPK$2o1G
zI41Ew%;^k`zMs?2XtbBp2Q|8y)5)~Hq5GF}dX+{m;`APkF5~pbo+|%*PFHF4JWhY4
z(X%-n-Am=4$!WJn-@xh38a;*6gCkXbC#UCXbP}i6YxFoyTWEDb^&i9O42_QC^fMZL
z8K)0w^hKOb?xWs+9;a7n^qHLAqtUcw3i}$_SLN@|=_-xx#p#bUI*ilN{Z#&+;aG6L
zTciKY>CGDbBc})VSNXr?^jwYpg463Y`Xf$TPE+~c;dF*ZZ{+kd8ePxngBo4S>13;V
z|6e)1N~53P^d60Vh|?o!(}m<sf4CIwsnTc<r$5r@#hi{Fpz`0&X}3nx5l7s=S);Q!
zJ$RtXKb_NaH9C#c>s7jXF6g@J;LHNFu*QlZjsO~u0+&z@|D^zZgs={ZkOK61$2tq;
zdQ&cfa<n?G!xCN!(0r(lqWpn&%F!Axu#0lojh6zSP!0<>DexZU`ciHa<@!;MHlgeK
zQ*IsQPNUqjl%s90z@I2b&OdNJ<=_mZz}=J^NV%ny8$`J>$_=Jm0p*5JE{k%fQ|<=J
zu``U5D2D-53M5kQOv;U++*y>nf^uh5?n259rQDg6JBM;s%AHHO2+EyDxu5Z?&Q7`g
zlslht-%;)Y%6&n(3n}*@<)SIKm2wwR?rq9lOu1JmcM0WQpj-^)o}?TtQv(lEj@H5f
zALTBioQHCk<5&BVx+^Ftue*|xqPnXnnO7G}Np@WvB{S-VQ!=$~1SON|MpBYoH;R&R
zb)zX6T{ni3tLow@iK$DV<ovp;DLJcdEG2{LuA!t~-8f1j>aInCFpxuTY{O{Y|K-kJ
z6733A-eHsIyv7}o{&hRMx-bB>qz{&6Qb;TrH+ZMm<?5|&9QaL)lzW7+*}gDd$-aze
zEYhdM$e!i#k~>27hP8*sHOSsltVM6PVS6eZ+cRPM`po+3gaKG%>Eoee<-!meD}YGx
zy6o*`SdT7@!ag!W?p1f9Js6w6f1ifv{vJ$c+b^n#+u=O?lQUtX(>h_Jwz6ak&D0hd
zD`MQ}PV+2_lFIjEb%Hhcxf8v~XT?nRW}GEs8@_d-bzSmVHugJ;{kD(y4nE6?1#N3u
z&d~d@_;mIDj255ee~jCac=)IGRkG3|))&F@_P#><So}Xx*`BEUL?PU>5li<jw9}SB
z)ibES(>p!V+2eE;bgx`KK*C>A0mFoN;$4Uu#<;vU#v=OXN@%s-#a-mMPY_wc4CNdg
z&cL)5HBz?o2}i>Y>_NPO?hZfm5#~;ly_TKO2c|K3kkRCv!H=@~@1XiytNuZZk2nt>
zMe7<8Jx9W=K7!gSDc-Nvv3eGwM_3O=(4R}BMY@aDlC#3?u7nS*chivS@-B}|^9=@i
zlijHdJ;SLapAk9UXBpdc<VWnXEM7j|w$^H`z67kYZ$ea>Zy7};T-M~yH02Bzf{>(e
zIcyUeTt3j{dyr<|r~t_#&6>PX8PS|XUQ9|>CPgCl9WkpJ8?8y?h!6^2o@ArPP$rgs
zrPHqx{PNWB>z=hpUDyOv$yPP(P*aJXhS=R7$dM!~??Vi>(kPeD67C$<f*^6Mb46z*
zc&#2<0H9uJN~81ec4ydrr*gnafpWVcUNG4^vT?HF-eg0R+=(dpk4_dPe-F3IzU1)y
ziQd7D6P1jHmWhbMMpCYoayH8Cq+Ar`+9+qIoS$+rl<S~eEaf^W7e7%M+`!}5Tgsnn
zFHTfCu$$50^4%Lwjd%G5{~HM+-MgXCLwbCB^OMVpEphbQ6=$a;BhEp|thnsWE0!%!
zj$1>CE3Sc(jJP&RhQ>KP`zq5C4wN@K6FRKbhuI=8;`VoXN)x^PcFM}6SlOF|VAhah
z1m%n*JJTaR@+0!$V`{|5x~>t^b&5VDMX^nIWbDYuT@walPR{nSXe#auYn7D=F@hm}
zi1Q7BQXHM2Zg)U8*?T_#o~&$3WFsNeH^6%5bFBGOv8DE*?D-lJNebJ%i<azPFAa~l
zV`ES0m(9DE&gg$ZBL&`?el6aP|DS5{qd5nDjd4^p(ug?cz`tW0_coGqpL1YujAK=!
z%(CCaI38=vM;89YhPgrcJB+ho&H-H6#`AbqLSq;Ti*z(L9Vuzq$Fr!%vIb|uXV$7S
zP;ZLQz{yw|VJo!5^-~}WC>9u3?bDVaSG>!XA=e^-s~duzU;mcAz*zGU78u@E+!D4C
z>!j5@GTVkkcL{%cpDjOv{uaT-!8kjxi^Wo5(Egb?r#2(F|H1Y<7z=hahz~A-4BtP3
z4%_J)8oK`u;dkK}8r=Ra`lILUrekz+h25md7yL223MilPyKwRc#;^<LwD?V1`Asn_
zens(6<tL|dj0TRC69#UXFaS{t+k_a5Du_tIRl-ox<>m=A4r0IF5{;gW_1&%wG~v)&
zS&lNXV4OTKapmLz*!;C&dl6f}sWAwg;!MwEl+vtxsf9*qW_0{{N7NiqNzaji*1Ky_
zS0@hic$eE@ic6xS*l!eu_T#Kl8cw#wr{Q#A3)RIpq!~sh=OA_&w!Lc?j`#dwCho_P
zv_DKk+H~~NElOvyvfs1oykzBoXLlRSGpfbuTbLc!)qcLgzsNXg7Pc=f>_=>aTdM})
zekdo=bIe{kijCK7)AX)2JZa!b8zuejg0q0%-dO%GCpMNaq^~FY!)dhvmb+nqX$f1(
z545kL+Pww!QSE5SMzzyoLt6ZY_K=k$tghGL0G8)lM8lRYYDiWNd%o+H!u(KYnr}*W
zO5Bd)>hJ8@m=+d5{hbNhtX0$S1XMn<EHWkT^FSq}1<Fy@C+_q1>3seHoFt&z8*@@K
zn3GDL!JJg`4CbUPGn}4H*eL$8{9khQ4)?=2m+Dl0!q#u;<stWDNU~%hW)<m7*jc_m
z4I3A9-YOU49Ft#Ol48knc5O;SK|cj1115>QfL=fnc^WAl1G6D4tXQJut`@Y|VSRO<
z?CHcV+LGu<t7jz)5YzJ{x&^nEsJGJmUsW@N1~tQ9ERDHiOSe)xTld$3yf#{^E<_>!
zWqEx@^4d!B`p~I-NV{pjRbC^Hm)9MT0m+MI`JuK=jb5_F>01u#XlNf)L#{9EM`jNV
z)~Y1%r^Iayv1bfQ_rabifjv_KdtSEOavwIwVb9yhp114voTQAr4}1G*3GbI5Y=24~
zc1<+SM(%}S!6zg~N6`rWi)aso+Fi<b%07x#UHJ~3#Ar_o`vxcUICM53XV@1y-Pb_>
zye;&4yMg}PK)<8YPUQ>!h&n(N46|cEiFEat3wM}+J+ta;SfA{j1wUN<Ii2;jxi5yF
zX>x`&2m1rgjqpOsPzbq&u6N4Y5toB2i#aa5^$GsEEDvw}vx$u>-y*@d^27m*D}RB*
zgrA>;5yL+kh4KD34G!DJqS_Wju1?VjCr3v@k??%*@o|TAC)|MpP1aW@4Me04E<DKr
z7oLR8q)8Z)CXoYfq7$nY)Z8{R0LRAmw}ky=!T+woW#)eu(q<F3%J2&%O(5s%@)pOt
zyl<<nclBGM>pf)sLpp9H)0B}a9R|0Xskz;=^zrJX{uZ+(^0$;Q`*hvK?9<fWE?O4i
zZ&9TXe`|=fGhe&^)O_s_?3W1s0{%7U{|){X5_m&(lC%I}{<Z1o<qbIXtl*qu*H+KB
zy{xbL>9~ok;3S6C^AM~HE;gO+!re>NyUUUC8%23BH{1Q-G&m3bM;wnvpoIUjfBo-|
zM`*DZ0w5#yAQf7S`2$AdHs)g+G#`6j$oMJv+m`>x-@Z@&c0UcVZ2o}ZGQ{7G{7wFL
zL(_5o_60zqzx|Ve_8RCr4D@XVdZte6{#FL`yZvns6ml|uJ8!q<Z*v=%zg>U?{`Shf
z@VB3FfBSh0{Oy*_n!ok-L_|RLO|~gpFhB<0!coQ${i$a`nRoLt@5#5~2O%QRM!pU!
zD%#7(`iiZs)UEj34dez-bt^vZ5EgL$^HC{Ul!e4!(y_tq>BJQHY3AVjS0SJ%=U7(Y
zxKkGfo#n9>YxPTX$$K}eW6t}qPs>W|Y~xzEwR%IyH5}fPX}|nhI<I)SWd&yovsS+q
z%C^B0sh7Z;u9a?m+amKzcrXR+f)Dnx$9f}nwP{-TilpEh(t!k8SlbRQ9D^2EU=Pc#
zI-T)V&e+Tu&&5iy{s=Mtkv$;dKAq9W8K2;ctx#|MK`^$Lptu5^YaZvS;9N3seM4Lm
z!Ih+QC33E*oNGLBeNSAp<#dkDbvEa^oO6vMt{=gLk^BBFu%sm$nZfAuWnEuPDZc)8
z{QVcBz9_DVdH%+r=rG`0R%U8CBZ<>><2oS7kJEKSEV9J!fcoAHQN+r@ks)OT@5@sL
z%ly)D+EWHo1-ua#ex+&ed@^s}kl{$6z3&5P`2~n2YuY>drN-F#CS&Z)8MEI8<2l5r
zYHz#F_yK2ZLxK_q5@Qq?nf6}Lxt`-(uXC<@Ao%*;#6{Yx*10^KYZd3JBCgYji?o-e
zbKT6jia6JD;u-`lXzxSxHvfhO-ra-qJ6(T&Ba!<0zxc}W`g<cp+L%QdbQhGj+TXsr
zT0I}EE37%B2rK#<6yBg~jU8>%do=aA;ALp;pCNbS603kWV%`^;N^;*|Dp`Pp{2%_f
zPkl1lt?BMso$(sZn8q2OfeP!dBSuwsLv+SLoH3d+PAA5zN#4}+e*BmA+`X@}=N?0X
z=e`fI*I!0lq`R#;S2O3@$+=pHE0(xOcTekFPjar8IoBrQ8U-%s?u<>)UHIG3U5&0U
z+9H6yWc>XnV}ILZ5SQ53ScKAe6hxQ2v7hmtd3l5Ey^nj_k5^g-=vrfCVImd0nx|N<
z=a-?iPf!h-JOnQ*miuwl8?o+lO@4o?XY$*C1oB&ew>azR+Z3AK?$Q~nIO9W{@ds$7
zK8F}pz1^%c&g6`Robd@_yoF><dK;s2jpAIXoa=PrnnGNpw*fkrm2=rSmyNhGh>P^r
z{+ZT>01Z*TAxDv*4Sk7g2DqTNtVZZ<;#<(0?l0B8qVHE2>2%NX^bzF|-U%5Pb<qFy
zr+D~NR`I9wp{Fe5PdS24?cehzkN3&GyGbp2`D{F2(M;Iu4*c@p2F5KO+Pb1s3T*Hd
z*Q@l}EiohhOmBrK8PT!y>Rdd9(i{{@ldBrsc#&xs-gm=qS9CUhXGE9a_hZqOxDlCW
z*xQWAycoTPBkQAUdFIXN24oPa#$psDBBomv-9`uqyk%uWHkOeHe(+jTsLcV$rR?x8
z$KXRpD^_f#D$YbB)$@!$c}LSr4ZcY>mzM?O{d;hNjlBZ)AhBZda*%YXG(#(AvSn`)
z-oi_zxN<zwc%%-b4x}<tyrkBl2J2{cBE*MizXdNRr(<6t3NL`T$2yfmvNtQT-OAUq
zX?O>N@Cy??%drd8k><T*qVLTMsS-G5?EiHGI}!V-E9`4*V_|oN&WS{7Z|qQe^w*zR
zRMO=ws?ghp6TTEF?ai9j-0Yy2W$?z$UiZ*M@0di|Zc>im+zk$`945Y9-;fA4rz!I)
zoXVdoSq(b%HM_e1#?ONuFTZEl&3J`Qlizx>KPJCV5haJ5PAI>`=2TWYXLD)=9t?AO
zf>%Hns~Z1|+uGhwly7i7o-TXGv!i_0YT6QJ@uNdd1&>}3<?>F9!i(yz-Yqy0Z+8!*
zm)5b>bcB`-Ljw1JUFo1eKCL6H)r&zpz02$_?=(A(qPu$2xp#I~C)=RncV*CBoh<y%
z?n(xYBd^#Yn;GL$8eF}1(p}c7vw?%AJAY=31UP-EwyLe}yhQKlL`<;LymyhmMDPT!
z{rWR;V3%I`^!-t_#e7VXQsVY<ao&TNywjTsapLqGGO)+w=y+sckmz&Bz#@~Q<wP8S
zq7^B&cAcJO6;gR$*>gt)&WF82PfB>-T1`92u7ppl)eXovUI|M}*lDd^kGxtdJd)Rl
ztYZ@n1fx>0A}Z5Swby-aqIYC;q8IzEc)Oe1Q5SJO!Z@4K@NR8In)=eXQ+X_UCGO*q
zAZEboi}lG$1KYVI(NujyJ2~EnxBsd6%D4W(eC7K{;444CW?TJd+>4EXU-F+@hYgs$
z0T=6`5i9lE?&7yS%x`-LBV>IWzwMVdaNBpUk@KrWeIzm!Dex*xl=sK(?F(NQ@GQ1r
zBAEtuSQe-#P}dRVxx<d^NS=*Bb{Mj(m#4;hN5j#7T8lP}wH&ZkkD@Ey={O(_Wm(rv
zkC(BHT74d&)^(|NXF{rMt)^GhV76RmSlmEo?ow!n>_j!2+e7UoX1Bp!;%QRG`rggh
zXTZ}=%~*Os6i{q=93r8qaeIRgEWa;k2XEjM8SH>rmwJQy1Kkpw2(iJRFgv)PapD!~
zZj3<nU^mm4wf#d)+aJ8dwB3dT+WrN>!1^9IlflN%7j&-YIM?f(%THVuaKZA<dmYN{
z^Qt(0a{Z2mU2*-U-zs&vqoe$|0jk?<!&Xy>zYh8n+DURLO?dB~yArq;TEoJ};1`AY
z#v{Tv-A3-k3D;s>w-I|ccBGAIP-YCh&T}8Vv`OMm@m`Duwf?=QG-<4*&05_I9m7}9
z#v^vhhTOQ0ZHGS<+zxkJ>EOTtX9A8Y1|9)7-)5v&gecZR281VVVebr)yFx+O3nlX@
zq@t_$SgT(~#xoXYU_YQ6u!66|`|0TZRcN{3b1+EY$T0?Lo4@TVQf+h4?{vJZ+0hVu
zHGtdE1l8wEP+iUh)#FT%Q<UP2M5Q>BLKz5rPMAfRbY!v<n<vagB_yO2?!=c;Ncz~G
zu!VZ2#{s7)Kcp$|)rP@^^`mOiYumQCL*!T-s2^VAz_tUmNNG-1vNM!?ymOx)`Y4p5
zKMIZRC%NsR10E}rO5FD)Gv@XlkOGZ+Y07JrxEaGAcUGM+ktlr#ym=F^92_g#@DI%5
ze#f?5GvL#pNa*uCB+&PD5MO;PEKQqre%Gooegjhg;~^wqOxOg*tH{!6*4e0Yz0JAE
zZou^ldUbsqanY>vPde8loa=ecHJHj8PFytWEZ4cpI9C<tno3+l`8qv-4&(pmWj-HF
zzcq~YkAoOxy6Yd^`$zizO1jb_dnaQAo@ldH(}yRtZZjFj>L*57t0@LgT|^!`$!@Ja
zN^<s2#_{?|G1lrInAE$-Ofg@~Ic`|=Cx5{-6|~I^cvXE&T!2p{oU)svvuEwoLi}qX
z^RH|u=7Sy?$M%1*M`FaIJ5Ds>{grW^>WGIIrXGZ9K6_s?hMg}mV+bIDn!m%KRX-OB
zL=`5#0@WnE#GDuPTW!3$OZR6d(8vF0=0h@$SclAqGGWBJ?ib^W4yUYpI&JvhnGgNr
zGtD}v%oACM!Z=UWI=Y(=t$SB9mA^gDOl1QSn9AD}ufbc<WGWZ_1EzA?I&nT!q2IQE
z-*z{@tr{a|{e%3rAJFgppZxtd&WHZ*_xFrX4LYBT-l6OKE__ktlyy$0Cr+im@4{(A
zKEtIuPNdP-80V>KRPFEIZ_||MU(1wu1PPRQ7=uOq>s%YRpsV|xe>*{cKZW()nXGoS
z-it&#%-4JSKVj|t&(?eE+0DXwZ!?(~n{x)RM{p|ZJ*wSF*L$bX-eIgN)ZX4FY;V}_
zY47HbNmxRAKYC7U?~#3{)?TXJf1|w#tSZ!AIumWK-vNy8KU~j$^iM;3_mI&t{q{b^
z_EPPBLwm8W&(6pk<1q|+e~0bbV`|1W?)7c^1XeX_Kb-_WLHik>q5U+c(I<3x|1xM_
zo*(SN=i^oMh!v0Hg*5ugTC#UpWTK}ld<ota!jT%;R~D7%yRnPDi9zqx)IX2+vPV!g
z@sTs-aI$g~vj@6RgY$5hrlO)WT{N#~KI1#SqIRL8+WtSCf3VAJJwFN6)ee~9{6cCJ
z&M#0@3^53HwuC0n$}T>2j=&>OkN&8K-0WZ>7Fk(cNu+19eI_l#CT`LEll%<xPcx9f
zKMmLp|5P!A4Kf>EgnxQ&t>B-u`q;zDbErO^WiespGuW!NyaHYTVHbqvG+nZnvS=kh
znJCIo7*0JLM{};5KV<c!L;Pm-e5y&S=bEQkJzqkCdZzsa_4J~iY~s5_=UU9Uyqqfu
zFF4lU$6M3!0$TI=U(uRiyfwJ~!zmD)K;jXkiHH+oYOAye5u_F&f<&<&r6m|aV$lF^
zSyUPp`z-FHN03<D=_rpNokn-DxYGd^7&=C+_TFJn^WFkeMBIr*kSJQk;!YGjLbPfp
zxOm*@Cw5m0Xcx9#)4X{+?$pNPPJbqx$DQz{FNC!Sa3VhFOxRj}rW2=U2RjpfF1My7
zY%cFZpVNv+Qx3M@L_twZzoD1lOt^VF8!2Ew^iBnMO=CZgGtN^T8P&PYL8@NWY`&P<
zv`JHM<C9FiJCH!Vm9WP8r@7+BpvU<y_zM(Q)4jeXLthhrqrRLhPB`s1>Wjw-XZ}Wg
z@i<}5sp-o_`f??FTt3*9@D=H+?PU5o17~pgOvZK+eeL~#IDe1625r>zHDC?X*9AzR
zuWz4%zTW2gx*HzDU-}H`E9h@c<JB`*otn}A(?_8C=6*AlF&g}4&}XvqlY)IH{|x$f
zV~qS8oHB!}R`N4W`qbAMtKq5eqGC`p=ugbz<!)ArFrTO*YiHx@bBNcUZ2p9Cl?msh
z<LeEm0BgA3Vr|Fcz<bnWp$6f44K6u_8YC)385FfA6&+s-oSdi6wmME_f1-d@q*-Qf
z7txsD#0hG1tzXU94e}3aorUvGF8_n?8d`55>t*ui-`6<teuY@s)b!)q&+XxsO{Ze-
z1*|G+`zg47lJ+z9-`##XHY#f0@@KvM*nT*f{qgqm%Kq;5v#O}=r;zeV+Rxa3cl&Xo
zG_?Is=<Ub;#i_NQSN3<epH)R|KZQq6(tgJNyW79j*nZ4*Rr|+w!l|{NSN3<epH)R|
zKZV*((tgJNyW5XrZlU^rOm9E-El#cdyt2Q${j4f#`ze@olJ+z9-`)OZWBdQ4x4-RF
z+s`ZeyW7vIqPCwxFDGd~WB=Xl-(qb4qk8*ko8ct;)!hH_%Kq;5v#O}=XF-t@kAIB)
zcefwMMMLfX5xxD`6gjo_^UD72_Oq&}?Po!f6StqS|L*qVcw1=uSL^M^R>P^apI7#T
z?HA^s!Sy6tzN=fiCBdx++WiPFx~8%k(uO-;usdNmGKO)U>LecF$oqoxBdp))Y{2E1
z8tXGeT8MvyBV>3qc8H7%aTY8%=T&XifA%WK%{Sz7B$)N@q4S?BPG`Y_`ya!sf5~cE
zUI(9NHXojg=Y{CQ{YOn7k0AViGJViYl<I0e6I1v5+4=*URoW{gcqt3trp*%%%{2Hw
zp>soc<N12jh`$5^b_XpDaEJg4l0ELT)g3Aqj-Im<G;GKF&$y4?Hpht|9H9w4!NR|>
zi#W^f^_u9~9$=!|j|8H7`>zn4x+ORLPY_+=!}Psfyl<%2Ro9R<hrbdJ43XNk4{1`%
z!LGx}q{iEZx1tB6vbj99?nDwymW1ys>hTf0c1VYg_;CSd!DK)WHd56n_=V6miy0hu
z>gYS-yN&U|c#FH;eHB+#w4Mm|C-&whePV9xcx*Ucf`vn{gLx;~u`h2&z((%fp&i~q
zzy@#rQV?Zupc8eXfDOL-Mxl`aRnc397bce>cGe;H?&L9)jf8{7jW7r>HUMB74`Wl~
zt#^M#RLI$yEkMwr!NLPyEC5D9Ioe>Sw?(27y{VB1uHfLulvcK%ftX<z57K4hU>;w#
zW7&Md9zU9Ko~Xyi=qsX|H&Kk~@L?8Yaqu9^EiY>}IID));Cv*o!3ht-2LFZb+kP32
z)r{uMaXRx=oY}#d<B0hS-q)W+H}}`9>gqa9KiK>okImmN&?0R^I2BvK<FN(2mBgoS
z0n@2r9FB=~j-3wCen^yd3PwA*bU#2QT^zfhcY?jskka=aa9992ytf;auFELh@V7PT
zqRJ<f?#>Owd8*Qdpl;)WihN0vYR~(aRL?*Hsh)=pUVn(BN?Y=~Yc;O6Kk)vK1YEoC
z2iIXP&gtm-{=^5-|GVGc{axpgocH3f|2NMgt$vG1I=ENAmh6VDt>2@V{HdNtq7SZ|
z^gPmK{9)`ol6r&5d89DLdGhl}iR&~4UFTy8N=E_(6(T-TKLF2QYCK!#I+Jr<%DFO$
zYaqDb+wOY^YFvCjMnL^M@9#U0gxEfvL4wm`=f#wdDaA3Qsr174bQ}Gm7r>{-(802M
zFJnIOC+n-p(V1+RXI`khFLPPesxFB4r7-jwOPbP*GdC3Z^^SmVs%l`b4!fsn!%Q;{
z&A~AR4|je-mdh0S3j+3hn89g8WcX=BWE|`?A~O6mBKvzT?Wf1>C7-`gEsxIfyog$`
zgF><wE;13-$w<M;pv)8;ADoqlj-8E<zedkx`i%lF4)j<g_c@^_&gei7;?)iQb5JrL
zfuUzQbXbX7==|AAr2O!qc%BEl5VVGs0!MM60(~>}Z<^|py-f8tAc5+4A~;e%PE);I
z=Q@{jUB$V6B`$RYUxm)=Uw$8({D|`(?R-r%J<xOf`LlEyv%Hh>#wbpBc_+)aO6mU2
zUv=J^<F1FL`*(`h>FB9&@0-DEEFPL}xc->odiVZ@?f2964r1gmIsDBd{IWZl{griZ
ze;-?`htU2;3wd#VtOtK`rF{V38tTPt(7i7>`3)Xv^k>#Hb4&C6T|56u9!qI<Dc?iI
z_s|x@DYy{X=cU^dD>qOgx$!qGMvXy`2QOFif)Y^>Gb{Rt6K(fjryEu1f!3Bf7*#ZG
zkL0)G>x8IBxZ(ElMw<AUnw>b2fu;THdCk&3Tgfb~9SJOL44PX1Aa@_>@ErcJKC&T$
z|Hy;&@biqP5Kr%o>SHDr?7frpP|vm4D4;`mXF&qy5!aU2xIWAdhKu%1^}#y?!Nr8;
zLp-dIEiUa3i;Q~k9#yGFA<S10%I9vx{ER&ma6f`H*kZ-kw{g!3+(Ywzs{HY@R{nPg
zLgKIRo;Z=~@(r0#-BWV15pWT6k=yY-4{W?)(+KZ<Th}$aeCKQiOInWQAM8lxM)rAM
zb|kaIiLd?-Vjusa?LH!{)gw>}w&0@3?cwleiwl1$s`^|8P9lGdKLL=0qnZ8aVcx`8
z?*Kd;lgSGh$MtyZqIhs^#21|&WY1$?bmB)q@$IN)Jnt~t$c{;a)TXP6IMzOcj)8Wj
zCH!oy{s=6x_j)>k`k5<X2A)T=S5VEUQX|r&7<vfWNB<LJaJJN1-9#+j+bX=*N3wMc
ze;WcK_J8<_!D==ubN~mN-1JUF1hy~8aL&g4=7w`0E5mfDj&AVJziQp!LkJw}$UR8V
z4J>$A{VsC2bk=d5&h<CWwSjXTM4zeOt+{1I=enD7t>#>Ni0f<KrG7<k^7|D!?o@@U
z!#(KlPq6a`+nJM74{qB%s1h?1`d*uEBkX%Y=uSh{;JPcP-^)yA{1ed3M)vb#yX1D*
zmN1hvTS~x2U}K~+eBT6y!(O7pTLu8Oxe_KpXMe&?WP+-i@J21!5`EQ;)I<X~J`%=W
zSz(7Fpec4P!XLffpspckwXUw|*fTh}y4wEHO+TkWN?%LRIe`oQqN&VY#Z(rD1S-4!
zZixAqri@>n)wqt56n#Vba;|%ctBWh6<X$La=E@NN7NQUOFboYksy=E~X!@X?b6p?2
z&*BlbJ4qjdxjxue#q?v;)Mr0c^?_5>5IwyGf&L18Oe1}?q~ZH7)vq#;>0^g0VH))D
z3?U)<P`05PwX&1@q&`xMi_SQRs$mB~*r$s8UL3Fs@xDv$5)|_mg83&`%mh}3p&uQm
z808sF<aa|5kjHA0hcBWI-Jt#q%$?Z0HAm-~&ACcBS3gv%J_=mu-Jje8#caF_iV^%L
z&YjSQO$2>Nm6|@z5cM&Q`A|H+64zZhL%BXeecDFoV>hIy`m~Lj`yzeJBz?4|;frq7
zZwK`OpEeWvU_It|eQeit^0KZI`kPexN1fC@!xcmOHvR*(2E}|03)Oo}G<F$Iu9&~G
zGLBb_aB>GdrsFA1G5bhjz9C&C7_1BN!1_2%G23*m7S8o4=c*yD;aoA-!gKkDRkOkW
zc>e|c>=5;ny-d>&KQOP3aWggjwBfod=Nhh`qroxmurpy3^s`UZ50TdDHc-63+(G);
znU?UC)`?huIqFKd1Nx~aw&V4)!-e+>tm}^8(oy`8QET;U;Nx9~KL4ew2{~8*_itQR
zG(Y#Jt~2Or44jIts|xH0oLpDotc;WCD)~uGSL2s6T}?v*U6rAi)Te2>I#=f!%DJxK
zToI^N{ba7I2kwThmU_s$3G+9De|s1DB1cA13_ABFt1p82EcnPy13pXGK(7!%JlxA)
z`*(vJCX-8KBXSwG1?WY>i8#-RVBUoEp%5Gm%?MR?^RnN;ziA7l30Nonxsud@V@5`2
zKWdF8juB)`z9GqEGk7<hVn6Cq=Q%hE|EfFr`vW2IAG`(nUK+hZqczi;iXn`J!!p!>
z4IKodZ!EGCVKFSjMKIYY0vq?IV^Q%Ce>#bJ3ll*Jb_PfUw_w8ajT3L9q$$Pm5QhW*
z@j4tA3H>?1NXbMhU%qG}m3Vy!WUnlp`)5roLzgnKT!92)`I%xks#qRCH}x-FVbFhw
zd}6TbPQ$vI+kh8;Vp3*<yx1l5lXCdOdF-f{&l0Pd#t0mpW8b#(UxbQ>NM#h+C6mfc
z*c~9L<kBiT8(p`JR@(VaWeSrIHrqnwV`?&uFPI2Nj5WgDg*~AOr|S+T9NQ8moG%dt
z(u7l91>s~@qTdShFFO|m{T<ePiph8PT|V77Vl*L)CLu5{VMhf+gb+(M%Y<+Pwikjz
zAOYYtrHtrwSV)Fbxe-x|DGo?Nw-PnR-%ZDP<9QP)g!=qn9@B(ybTJb`U&sO$@+m@s
znh;9h9Q-qvlZEK(yAb=3Jx488&nxH^iOTz)BgpSSJ{f;>)YA!KA6;N@|8QdYBst-L
zwdzvJC;VuwnuttF!Uxu>TN$#?T6H%=-p8k2kx5Q?FB$H(pS9{uP)_9_=2@(FIw+P8
zf5hJ9OT%BcC9|W5P`rP`b0py=C;rH>x-@?ex`l}OpG*3YV)?K#yl$9+*Tm?xR3!!P
zi6!H;R3*hh8AO9p@QIaC-;*Cve&qMPhWiq6(Uz<nQIhY)+gdmsif<+<$*Wl2uIy8i
z*RyO4zCYGa`ASKClaaAt$8kkG3A`-h04K}9(Y$HHhX5~vgW24;Y<Y@G*ccI;*qoAn
z4mAw^z#+NMcPRl{0VR6Nr9|%ykrYviM?{Iv@wTkhRGxx8if1cvSBkO}VXE@XL}fmL
zR5xe4@HeA9-F(3R-;Zi$@KXgd1Ir?224V0Q^+zyJ1ahE6beeZ{#%-MOb0n~rUm)N5
zqr`YMXM9>`e3CQ1%oz_8<Ilu6fHN-E87nxW!Wj<`<1fVMr{;SjuGbl-aK>!TxR)4@
z5hI_5Ma1Zg7jwojobf|q>>|cL@)Elq(OUiMLe}cjIAc9AN(i^G_uM|!xjx}sek5q}
zQ^XZUT(tK4htBm9=i11*yu=kwTohFI>0DmU^%&=J6ITy#VOARkcjN!z4(5=%>2HXv
z(Pkju%%Wq-!q)+|{Dtfu!HXAEv;P=s<ns?6OZqpq3{D|_5svDx_yq;B8u>j@*e%3c
zq)9fU6hT0rVKESNqSSbbZ_pRR>CEj=b+b|5dSrW!m$9m0T9a3C{FbBoj<uP8i9%GE
zUS9qLPoOuJ>F_q+mcdqy`i?Vg%c$=-*YYg!Jd3Pn>_BVvW9(M-ZDnnrrV=3E1w+Fo
z+1L?qn&S!kUHW%iZ+n(ROWWM%(OoBauRM`)o@iS}4DDiDW(nVxx&2|y8jH)BHC7^l
zomImz*Qaya_+}|=qjk}7`&V2(?9D~|O*k$e_U0l&pZpzg_U0mI+eNbAJRc*Q=(!dG
zdKWoac}MmvB1!Kd%5w*a@KfUUPQd<U680}?IqLK*r`Pn=@cdmsSa=?Pg=kV#aElAy
z0ic(x@wq{K$`GC#cG%#*=o#jiAw@e7{?aRA<lL93t_^`tVm<maOqC{O^Wn_isswwQ
z@(P8=oyy}iNVqE#$8=25=miFRB%ce^{zpxq=en6duR;QWjzp)ZzYPLqUGnHd8dql-
z<Lb@1t|zWSF7_g*$Umb3U9!7)j2!}}*X$KMn@<h!8RMu_dl=1>n|>ZR8GnSaykA2H
zdr1R(<4KX8&K^u8)->$??ZL!A_250xX3PvAp7T+bdBEz+r<gH@z7hq;9y+$$^_HE>
zTXq!^wEt;%x%#_!%T~e9_?Iu_v%ll|Uj{#u+uhlNnWzGtJxGJUVaJg(N%W}s1+3;~
zM-QTL+wn&aXx559Bg^Kjx31FkkzdO6u?PujJ%J)J>YVk<#i;ex1%f_;^O1D*{4|bB
z2;Ugd!n@BrNs<#z6R(H1^%dI3i0GfpJ*o$aGckJM?d?>xz3gKp>})~kQHzo6c_(&z
z!SezqkaznU6M0{FLV2&pZ6}uZ(FZhncb0Jbzn#hZ6AH(u@-BkM@y{s#59B=(&kvDz
zGHg(ncOF9izg6D6y=L;>{AaaioM4za#q-{aSW(m&qnU`!zDIMt_Q(@{k49fVu+h6)
z+l*ud3APyp%hzRfH8v(rDl6EdP}y$m@tG&Ok&nJ;41UjC=ug4pDyMWFY&?Gg#C8$H
zcA|kYigBK3pcFTwLeGPl*H0VVDZm!owxg{*Ve9Vk$C(&{TmOH9?XzK@b`3tD{eV;0
zx?{U?Cu`+!_NY^DWnf|#&Qqi}TS$@k(3mY^<ZSwDb+-O?r1WQMr5T%q_lfBHA2{-}
zvJV_mqW11cqVjGEeRu+!=Im4!K1(+MYbb5$WYbS*VF*r_+%DBU(uT^NI1Yz#H@FS=
zKr!3`Tb;g!;60h3jAawjc%<=29Y`HWWu$UD{yGKwOw-kBIkb7c3YAV)=4T`;x%dpm
z(pgvtS(KD)cCL%6mNJ)R`%yL@3!aO?PYWLWg_ScfIiPipFHQgc)1SCY!_%-1e7ur)
z*|J8QR>|Io^Z%eZw(|;^V~a(Cf#hZQxB9EG8^?zKpM4tFk=qzoAI`NC;Tp~T6}#d7
zZYqKM)92%)9~uEjKdZ91ekzgrbB@v91<>{9pBAKv-=D;3Zoe7Rs&`b(z(Uz^sL4{k
zAX#}QQTZTI`3RF-@t|6RfA6D^JVx6^^evHOr6LpG5>Ya<(aA2M=Buwtx|G^Vkm6`M
zRBzr{ym^-)fe^^v>#Ilzw0JpmuO@_pw=y9_aIOc5t6CFPtIqX4=lTi>m4z6NCaPr2
zz5G`dcR!z)mQIHH6xCSyiP<H-T8(3!{A*6idx^@&>|lqur1|L)>Z!u6aF6+#ImTeS
zfy@yB5%oyN1ZIvDDk5{FKuL<SkbUfh*_VZm5KfpyA+{J)!r45b0tqIRLR=9pN1KVX
zc{QngS-_<7BNDXtc65mPCpD?Ou5;CKuC1JF1rCPRKMO86lq*Xil{1Tk@l!o7$@?dE
zWw;pb<)YDhr#5TZx0-|?j10#}nXFG<_>v|B^Brk4-V)7^PX^_Xnn|;lVN4EYA2Wk#
ze14B6girIC5WYtOAv}c>ob?-d^QPR6=3PBs>>rwa^VfhE)0*OY);L<P(X!%SOz`-|
z92S#flW_8m?agp8d`q0{BVg5^Bb%(e2WJMC2A^z8^7Xq;&e<yWm}GP2Y{y>`U^|e9
z>F<U+a}dA9A5h_2oW1BY<dOCW+PXO?<8QAkmS_}U=Yd_U73c$2ES$?;@WmutSf?y~
znDsYn)nD23^*)CVM}MHo(BbfkGA&HB3AkoIYZOg%Pt0SYTZaUq^T2}Z+sJ-OAmGJ1
zR|V%%I9K8-aD78u)I&3LuBn_WhjTfptX;%K;e{)8uFE;sIL_6V!VBtL?kRLmzYj+N
zL*o6x@l-vps-AtE0wM7wH+56`BphFI6V)FSbd!t_R5q!`mvIHY*?B<jfpa%G83+;!
z%IT1E@LBTi{<_Wnpl+g)e1oA}Dj|wWC<!USnbW~bNW>X1`lHBwQV^47e<dd}ZU^!>
z=D^BA;2zb%0a@r6et>1LFeUw3pe9;eKK_uZ--nof^Wjso$*01#W>eHzk>?^QQRIAR
z{IU~iTwH)c125jK>HPUzrt>$DK<8z2wzBWpsy6S^xvDtVL!7Gy-MYRPagjE&b*@`D
zS25=rM*#s<n;)Uq`QI)$&flgejp$lfeK$B8{nt6e8l2nxfqB2^-%kncpUek8!bI^7
zg>AJ#$Y(iQUQNAiSc}{PrRB67-@)i_QvX}fVX%LpJXnj^tuDsNTN=i0JZ>1fyc-#-
z>shTkU4NG*+9`9HXtR+(v{AH~k*EpnGM#G}=enA6MbdI!722N*A+&q)kM~z}evIZ*
zsyp6~N?}aGV1+%8Y)o1*lM&WU4{IQyY?p_%$~}^7X~@x^H{%=m@H+U@VBG6ymhH2Q
zmRCD-TKNr8X*nMXHw;WszDQBtWiF22Ytu+F>M!ue(ln!nEi><hKotp<W`bnP+aKUU
zElgHQ_;<!CV8_|n_}&;9aVCryNwy*v!j`WP<otiAdmH#Ds;hrIfei!=PND(PqDEaU
zC?d4bk_WH}Y;aa)0W}KPXsH22MMd05!~jWlm#_{SQL$>P6<hm=l~(F&K+psd0>1F3
z^}RLX+pY@=iVzUV@B2M>W_C9T)b@G&Kc9a-pX|)sd*;5JbMCq4o^$S<#K{WnG^fvk
zHl{zhS}?s27hw9YbP7zfG2P%my`-SFDyZMXq-kS%DCA5u@0w(L)r7=(Pn5uV_%+1m
zd119O$K>e@h);-cO++{&;g64fjD~~L)XA4(2*fw9B`Vv2*Mijf?RV69>fXqB9Nq?d
zl^!18E<0XqRy<*H9t2pBVHA)XPvgwOiJ5x%I~Cqug>W`7sTO<$+3+IZ`$vt1;UYLI
zize{fgTeQBFlma(N)LZYN?eFHx!Bgt0dDvYI@qoLlCWE)tZb}q1;<14$GD#nu+cvq
z8b6$wb8rFrYaq;5RTKRK(Q177r%1sMbKrlb;4f70y>3kqVg*D*^uB5B#=nY}bj~9^
z9`x{rw|J?5ej{-!M<vNeK!!sSS`xix5lCXVA#v?#SPP0ZmQ%58On8!D(B-|PfDz7e
zm!0l~q7QrGNy@AU5|u0K;%y`m;c>blT^(3gC<3+I2{eV*gPZvcNOO=m0m}p3ULLGN
zE_5X$d<)`To~V@f|H$dt2){u5>)|Kv=fz|!OkkO$yp5^5Q;J-yF#KkfO_4XC<AWlf
z-~x(dO$Ubm2#SFC7eSsxhfd{(8SLib19Gax15;WYn}Sj~`*vb)qrPN>*BhXTfzQ2b
zA=;TR$$G<^z2UdS`-Q_BWteO{!XY1L<BjA)K6;a?5MVfYV<ITC+d3y!3+|G>J12cu
zr`zH4@;IRpHFw8*zWgyz%~+z5PxEh)lvT_x?Ociset8tfbZDUka`P1q?~t2udH7<1
z+}+v_dn$0J=kqYV19Eo@(t5!nEa6BI*C;h(V66(lhMgy=GxgUIYc%qEtoUTqYhj%m
zoCdM&&a!7&B_54%og(u}Mdtg3%*m5T(jg|nTt(SK3v9|ByizE8>{UY9A7R#5wH}m>
z(XEL=`MLwRO#$w}1uSqn0k;!yu>$<F1GroPu2O*A2>3byEmvBV1Gqo|E>VE-QUJa|
zK+Baj$pO4v0bZj3w-WI01hiaf=R1JsDZmR9;IjmLlYmj8&CKZL0J;=lF9rA@0pB8^
z<x2a<d>cLQl?Zxv;(}fW2>3PuEmztz4&YM?aJ>S&j)3m~P;}i|2Wqi`dPqSP6Y5<;
zQP-W}KuuRra}?ACgnExq)Q(0uPz4I=Vg+>?m-p60{~U&6^brg<WiR2}-Vbv-*B_UW
zTKLe}1yE#f;=NjL$QjFjH%hlFX9mWYIuhp?#=FbZ-c9ynePbZz4+=nH=08TV5egXZ
zPK~G#`Q|N{&Ft~!M-ghQ#k<K)rz01!%`O68%-qO8c&iaTt#fh?LcYb1kyB|{(FS~1
z)8Lsus;PidO~taAP~1KSsAi}S7+@~9AwfL&fg5{wA!DHK=Ll>!?rukHoy>y22g^{i
zu&fRkdxObD9ZqM0B5Z4Fk5$-Q^QS9>Yo5aeTob{(w5kgNeuNpS9H<2fYKej>BUD#H
zkr{sJKuuOq*D0toY5<i<C^ExP2kNH^%A=r;=ThBbhR3f4Gc3KzivJN_0QJkR1s7QQ
zUX^sU$wri3x=x9vQha@rYY8kGlh%s1{&g{e-Kf;gRHw51<OtXi=SJjq8#Cr)!OVlW
z05ern(9U!0K40TNO;b>}DyV^seLUapbD;zE3k5YnL7jSIqR-F%8hu`NWm5bKCU<mL
zJUU745o#}CM7Xjau$y4^oI)u(1bwk$vpTUQPr?hG`xeBX%OKo`=Gy)J-6ZMn3S7|N
zxm>ndbje{G50%K2=-Ag7Z*wU(Zgdi&--3v>6qC?y(p;<|z2I&D`HpH5^3fy&6A(Ms
z(FA10;{zcn-y=?}kD<TXn>o)B8S*;U$706d20}I4j=}PIG$uTCmMqqy!$)J%JIETT
zUmcPy)Hp%*>h>qdeirQ_y+iDZeX9EL=$+#Q81gXsE~aTluJBNG2Cnc-b;hlXk+2mX
zf<CCM4n=kGUn`Z@+QmI5VG+07toI-)PR4me8rNim%Sm~@OWXZv?KEBMtt;^syy4b?
z#M7YZK3gg8KcuE$Mwz52RJ-Dfzz((LHajf7T-f0uT)+;$!}POiDHt1U*yUGXhh0~Y
zju}@u@FfcV1_eKy>tu_4wqry`*8)C8pHd&TjTOP#D*K6?_^qSxf2afh{|tRP3jgDU
zf5rcE><<6#GR6P6kpE#hwL6f*4jh{NZ}2~Y(o*?<qQn1iHhf?HXXlQ{|4IBVVUc8V
zMx7+PfNn`J6ZzRXPJI+wM=>%POD9W={2Vornc6R#tT(dphQcG4O#hOR7K1U0Zng0`
zx>)d9j0^C325hf3UROb+L?4}`<h{@z`{^<Kls_wBU$@*s(J;c?z|)C=_W*fL#LA0s
zIjB*`ps^!=%fN@|GYmZ(C(bT=aU%xiP7ap0#|NfNIe!XY?>v&ck;uzetVMG8JA8NB
zKh27j<?7aFedQ+nWwg1PJ%H%AZ!4-t^x9kCHv4-F2jN0M90x=rB?pd07s-KR(M57x
zXi1J~A~}ZP_4p0ey1B2r?6?#mvVaR_2{~l_*4lS0>L%H*O^}_XV6og@gjg?}%ooB<
zAp9=f?2GmCQR6&Pgvon$$jk+Pn+-->B5dHp1#Ivmu9PkL@DxNt^r6e!%Lm&Zn~YuZ
z#6-jsI!KL{Rkg!ti0mdj@n<?F{tbM3n~N-FSX&9ibO>fiIAQ6ZJ}N(K;Ck>anZ~~h
z9BA`rhIVPShfrt{zSZGx??OTNV)QSS>_^hh*tM4?`8z*=dqGJJus3eN8;DfI4Vc5@
zF2D`A12K_`y}9AlhlQHj>za>h<5K}rPTew5x3Y)~MZGe+1ObKyaCU*2)d<g;ib;m=
z>VXk%IrzT)H#)?k^34LlKj)cU@}UlUuWZK;bU-ScKig)9^Ck#8T!0JN0d~_>9z;=!
zuGqtY>aL*rD5y`lK(m-;9>hj;MlodvTW+5xQGBTn(JmJ99n3TCC(&^g{u~`HEl<Mu
zSaPUv&MZ0?L`nJ2(Cx^TBxEG}{QBK&qv7oFf`*^t0yN~ZgG2*n*#{@0gEg13gO%)`
z?QibfzYLo&W&trk;1h@676w`^rt1$4NL<MBWd{FjOPxvbw(H>!qUX=IMv<-3{NiHB
zC441_&i1Q+!*Rq<e;iGtfm}ies}37%HUPnTMHBuD&5kb+$z(w8W-=lThCVGwFIwUA
zkWr$uvMp!?HQdMMS=@L(AS^irZtPHMQLwEOsU<~BWf9@^kXrb(Fe>$iue9eZM5!f|
zpJa<_(cre&O>LxkM&u@&G^S5T^B^uD&22E9u9|Oi=rs=1GzE35g7Of`;?Val0|7U{
zKBe~8Ce7Er@Mc9XndtT>h*e9xBm}oLaYXN`u&86t8jp}`gf<&sum4^$4z%%w%N(+_
zlXZyLOJF8hk^>{R-)LjxHbXFS2QI+KHCzP}BcR)t7>Ln#C!nv+`rzODJ_-zIkn>*J
z(Dr-7-ys;?;@^uD|4steaPbH1$Y*q+B(|FgC`4x<Rt~gJOt0nRl7#4Q?@_qY!&rV`
ze#I4<IXux}h*9%G9amjA4L8N%>(!?~a6h$q^n~H4J7$t8$-xGMYbG}plpPOeX;P>q
z!zc=$OP+kuEHd`yxfWwD0|c5#y~|?k2}xBH1^4;;DoGW7N1X>GuiuH+F&sq78n6^%
zi;o*-0&W<5p%Nt;VatH}w6d(xT$;+@`(m*1lfhTaw8{DBi-epn-~w{q4}<BdH8z7U
zbfAI?>Rts^N+^rL4_*SIe=^R}-;?-T5fibhN+n7V^S5YfC;a_41h1uNMGa&$;qOmj
zu>ZdNP1;de5+*0X+A(>D3Efultl(xK=g#YG<V3uJoZsUD<lGLMqaX*Q?8<I`;Zy2L
zg5Q(+pICpcMJtpxn^=E7&X&=YBJ0oJsV8gwsg`)DpKK~uO1ynQw-fp-8@b3^v&)G-
z20o8KVMmMS(gpJSM>@r8IR3A*adMlk#y>8=$s!1$ReP=RkD(uZ*GL}!VjoCgqLTfz
z-AHnjPuVOGHqSeJOkUY^-1Wm<zp@e);hde0Grt43_en3i85bN0ejP4254@DBpT$A!
z`kXm(xY$S(Zri(HWM*B6w4p?U2aO{_*$mK`zg}y1X2pflnYFl}GcQ0;tja=wyeJ@x
z9H@l~>OKYaB%yi&3RNE(k1l^U)|sC#-7v<Vu2x?sC)-5W%&iE7#V9YA6VJ|d3h%E9
z|1VYeR2F`^D*Ry<eg_Iq<}W!80Zt$D{OFdG($l=9-SVUJyLt2bZYaw@2gQ7~d4}D;
zSI0>I{(%en_sK6&c<2IY`6w2C9t%(QZ!q44c9QV7tCv@INgKQ`dJu7Ln3wtWz+a(0
zvumh+evdYsN#I|PsoeR3;8ipuUGCIdtm>Vn>b)VU-j^XNqR*nxcIPSM!;d5t#f5Tf
zwMh$%@Tbwo(8@y7<%YAbHl7QZ61NKrmMaFo7eAS;Px8;`ZR>S-zw=+2H^gW9-_I1a
z!yil1F)UF$8d0*?5rIt(!S^#ODUe+5cmr!VfMd(3sFcJ1RlvQZx&c)h%MD*YJ{8cA
z+qQwx6fPtDs2lI5#SJeU&gFXOJ^y({;qOG_z@k|Xe-VwphPJ~U!ZkGaC;bzq=;&|m
zWaT{DqiwS{a2NsTsvK=$+wo{hWXxfd*N9C+*rJb}8t<5+C$w!0w2qv!#W4TK3JvoK
z;p<4x6Mg{#+aIW=`v`CB^FM|vx5z0ru`_Y_J$FhWz21MP7)U+l^~HucuRIa5=Nx!5
zz;De^hk*smP;gGe%rpsAGG_ctT(Ip;3%;0TtJ2m4MIFmB6ELjbgFSby*rR~55u*jO
zG4rG%2fMYR@4*rndk^Lxr|rF+w(H@~qu1Bks4^i8Li@DfMgZfeu5t$~wltIVSY1zj
zeSrPC&xG|0-!Th@<@H}uus{h7J=mLO^Y5H}qxkZ{X!YfTe$f3vajdXp_m1I)KTq{i
zISbR#PvE2{aAHdx%TrJH7m!gH{u109`x?i39-(!$aGEJTFcBk3Uv~UEI3`OkEjCJj
zImsx!y~HRjodW+vxo%F#)=eFonG4Vi9c?S2?(sokF(mAzg_lITHUlegqR}$K!qDz)
zMRwfb95ZX#CY$h}=vtcgNCdob^G0N;y#a5&W23qmo&~7yh*`qoufp92n_Ml1QCr)r
zA`afnKHxB&c~C?CR0(LWc^T^efZYGuYVra`iV!ROLF~hBM?^-|Vh0%(rT>XIm{dwL
z;>?HgK*~1%0O49A`~~=A@TTa+h!xVFL!_RSj9t*{q5~3z-T(~mlE&bpbymqX8bK$N
zprM4=W$+iYs{%dYwyaIjaaa*{sA7|8wcBDbtBTW6g<;+f%0YO*mV%+L3hBXRC0XOJ
zK%$i>c_}n_g(LC|TmBWoMWDu%$KL)HH`V~wf-mEVJ6bW?P$Y;f;L?8-!IVBIrzodz
zQPJ^*8;Ww^Ml5j`fUCQ*Av(V4QG_IVn2NHY8PWCw_Vrksrm+(!4+SG?1q!!Bt^5e#
zv>-3bJRW7{=w=b#9i5F5i}11CoUV+ly<kYUwAdf<Mxvc(v~MSn$laNnqCK#X{NHcq
zEjN9ScFz4NronJqbRiUJ;tw*T4-WqkI-q4=u2e*@va(|n4ifcBvTMQdcuEyuMtBEa
z8%DP2BcMtKX+bxhgq)NZ6k{&%VXQ|=acLcjqddGFmf_ed2a@Z33s0#7*k%A~6?1Qt
zjzbb*W=xK<kR?Nj<X(ui?_-suPxwjB8^XiFK>v)Lo{_}w=OhiOXyso!;r937OtM5)
z@__m&+Nk7|l1tIAH;~Fu8-$sMqA11QAAk>HEr<~~(ipq$$xdpW)_JXKTUD*LU{>_d
zv0A{?rJDsx44OY}&){tidylw*-Dw<LFV^5=fg3BH9mgEBw{*$=$YbS&q+_`-GF{qp
zaQ{NyW8X3)#g154MopCdvRE%2JxK;BZlKd*Y(lnAu>xse29fBdp7^j4wU6Olwa`6U
zT3UV)rlqHG<Hd|Oyqgx<ihK2G&~cFeV<7*>LH-M4p*DnH7G8pHEavgwu#TuJyfpe%
zJ+OcYf#Thf9(C*>+JQNRhCm+VXH6CpQe?Z0P%_v_GZi}JdkW|n3K%^B!BdzMuqENd
z)OD)h=(Fqj1}m`QdcYcfR26<#jtVf)92`=GqdDZ`spWD$44O*MgbYPBauCW#Ie4A8
zu!wMbj%?n?;!%d(_C5(k#xnPyBsTpjGz8SuBllW{_GPpr%98RPcoBr=N%Py0zCe7U
z^k=;ndMEV774AwPfs-6+0e?rMPve74#pHsQg$v$dZy_YtMgI!VlgRK~G}+bxv|u-O
z0H!pvw;t(vOg@@4+_g<8iUUu2P`i^)xsJ)k*<vQLeWh7<qc}rFOl!esF-7rvt~?k@
zJ;kc=T0&C&>6`YFilU_e?TpPrR-z>pc*2QA#fYL9h7-pMaGI1(UNn32OTyWxN17G8
z02-DL^9SbLU0QGq-ZDo|GRKw#TDk^4>}p;*CBSc2O~m&0{+LRdJwZ8xqZxbe$uP~>
zcaNgdfTTQ-9Us8*xchl<hZ$Un3yhoS#V?_GxtMnn{rJaTHV>Q*yvoq}To1n!UET;l
z9!rqm>t_HcERo=ZWEpumqUtb-K}K>HP*NnC$jD)m%X3SftVFi;wN^()l9=`+5i3MR
z2}F#<6R>BFMZoFk5|23+<0{didZ_;KF$rj`BXnRVw$)^392}r_rhJMkO91nHfc{>W
z7G3iSI)<TP2~Q|Hu@}BZXPj6(h!B{c%z#*Y7?P(2ABT=*jxG+gcF_VSL8QS)5vniL
z?nR`rDPRhtpF$n+hNIE`?+|wzRFPWnHX-_$L4}}CCQgH3tQR6szSDy1f%3qCu3GS*
z+;q`G{Q&C)RiSBOG&92ArWOL4QEhbOYLMzXfTKNnptXlqa{vtj*B4_Zc9ty(6z%+~
z{ZwbGPaMv7!gxi8s`F)+*b^!2@>jfVHvj_@{nNv-=xtwG6AG9}#7pc}zzwdKHHv^I
z$tvKPs=GCU-;M4R#}cFu7H_d}EI)F}q@g(O&+PlU&+H2e-d5|oT59|TFLiPCxRt{b
z(HraJlw_vGN+9MF$G2_AtP2sqqmRb)k<TvIGlj>n>hj&?zX?FmAoDVj4%kL72|gaS
zRD@t=my!s{M+xe3l;9(PV?Dd^hzUg%Q-HU82nO7!g{qi>GEh&Eu7zgfQ8Cx&v8j8K
zWy2IaDM>6-&F9f6n1o|RNKMdd1YzunQq~s9^!fmCtPZ^>0I<_%LXzdr@-Ws3n4_am
z<4@9$!9{Zrxue<$<@k%(>fLOnt(C39Q)&~EM}CQF0!Drg>W}?H<gm@KKFFawA6T~^
z5V8)`0yVbz!ms1JpSNv3x-N~&s|nFRu9AV67yn6u;pP3LSM7Ja;Rd{RN%U8BYVbmK
zGg3^lMfN5Zh*5I^am3|0<&32g>_J;OLHIbf(5S6w5rR48H&1^16`DS0bo$T7u=HOa
zIDjF^I6ury7?Gei&t|li9~h9n@V4lR=OB|{a#H|4vk3i%T|>IR9yPt=3x9wLTJ6^g
zFDabJL6ci}>4fORsy})0^OFlmdf9#p2*8V%Mz4B41p$&xt;Ql|v{oEWSgp#0Y5q<L
zF|NylJH8;1Aysh97mp|4FVDUtu>>=h5gnIh^M@$8te-diWunR)OS}k540EbG(?4DC
zLT3{=3pxW1KcJure`cZNfoFk|>+L+uf>5x=c7hTq4X>r3WU=aDUL2aUDx3u*p+xFY
z*q8v^Q<jq*x(r6-;w^#*Ktx00t{Pn)i|^tWcWqTC)N4P6+*?TVP^4gwL7@^Uq2nOq
z5$b4v)2|oqMdaw%Pl7o76JxyrprVVvmLx6YLW%?_s?PxS!o4M>>iNn^u`B@Cp9?O;
zm;QJ_eJjL+#4I{5sd_+R^-`L=ghXi^6A_=xm$WBe2O3_%m3;LINDl4z^9dLhQ97G}
zhuDFg=<Ghl^PwzQK4LtYxCrv3D725GrF{NPFtx=`NZ3CugW0VUm66QTm|0owPxD80
zb5ePsdA{fg|BOE8qG7Z<45M3M7;W}%f^Bm3rM6A9ZZPJaNUu$}z$Us4CQ7l1#!e9g
zwdz%kTq8MtezB-lIB(BYZ=34v(R2yqz@i%afO6H$xYPl>SOH$G0LQ_+yXs<?=3~z(
zz;hhHehM&O0eT2{2>~MtFx>%c8zc?SR)FJ&0q_s3da43^`x3i{Zz{mga6u0rX4KZB
zaD~K%E5Iimz`rQK7Zu<TOv<a~!0;D4UI8w20D}tfUIkcM1VH+KVlhl@==GHj;4c;6
zOa<s>)tgy$y#gHK0RBV)j#hx*A}Dy(lkD|<3h-D5u!jOXSpjY*;L`-WQ339_*zR@9
z`GTH3xS-dM6Yy;U7Ae404&d_&aH9gOB;a#!9>n@8z$Fe~tpa>l0Zt^~Y65;OI#R|p
z4&XEec&h>&NWjxcnU@vd1rFc{1?W?N83gQ4z(*8dF9+}h1?X0Q@6&+3hIlMjfIBDH
zz5aZl^!i&|(CeoOSWm$53UIvx*q{KnE5JJlm`3!RtpI=P0N$?vA6J0CCSW>yO+5~@
zzS#l1NdZ<Wz+nW;AmAGcP<H^wD8PvdP$OUu0zRPtPjdiIQGf##;HQYFT-B3+a7D1{
zedFz3?>$d?ouL5N5-^8=lNI1=4&W;a@O@l>o}~orMZiG{@KFcw4+`*U1$eRtfPD$r
zT>;+i0N$nm?@)j_qXBp-@%T^ZqG;nK4&Vd@_-h5&m4K%a(Aufc-vRuw0vx6QzZ(U>
zJOWxf71A8Q!(1OkdT9!<8F9q&C0%=`!dv5P^!!}`ay@}wAJ3|1u|#X9!e1T06$)^z
z0zAm7ZziC%Qz7I41{B~@1^590XA>|g^J+$k19*i3yj}q|67UuRT00ef>HrQ_fTI-P
zK?EGH%HU9-1_L}E;{awVz>^f<HUfSK%1Chf$3DBuJI)o{e2ojb`~;ysAryntUv!|>
zD5%#IRESV}3B};_`yHs?D5%F2)UOED3Mec=vWCEWwf}sqC{C+HJJTlU=A1lwT>P-m
zWa?de>!zkhs<(n+p>5B(NjJx1)6=LtJ?zSX6)GolC3A3-&9`|Yk?Nj^E3<H4y;gRG
zp1w_wOzdMs5WI{@Q1d`9g#~eZp5FDsO!0`(ez9*->}oWEN<>n1x89{!t&<8OBd;qY
zI)*-l3z{7|4^yt0@tk2}>zQ)|TTQqCTMq(i)mmuC06piY=wUDEp$AI?#@`_x6TVCx
zep>NS{FyE1X0EQ#rcnJ;s<)f9nm<X-_)H@*7pABYN9%E6WQbO<Z-b_VL^D&p)+4jf
z>v~=De5HqvW3LVKn#}Q$AtNkbAa<5TuN{b081^BC{~6tUC@pQY=^CM2lHUB14Vh)r
z4Q$`AqM8}CY2nCe<icMv0>Z)5r~<*$9k`(LR}K(7)sD6Cw79?E=^<QzCqJMBPr@%V
z9H{9EYL0?hZ~>q!FYEh*fz4)M-{O}#;%a?DpGm|Ea6hZ^yoR=~#YB{*jsIK^G-t2w
zQvkZ`i_S(Nh30XgO<L`LLt+<(qhlj3=zy?;T@-OO;RSif^da)4;Um|$$cPZrjL3P}
z?rv#?YhG<bF;KVSKZ5Npe}k@#*&Y5oT>L_=-qx(=AJBp$P%11chcnFy6WU%592}|D
zEQ0Lwl^XY^`O;%zGA(TT&@3p)kA^?jjID?+(>#BVzhFW;|Kl54^~0zx{s-Nxti+9-
zcn1?BhPjA|y19g^RiT-#M_iRgr03dSbWbZXjfOl={xMpxl$7C)x=63RXLU;(t3By@
z89L-BD(o0S_~{4fciY+B(#Dl$-H~h&yA=%;i&*S*DPsw1Gmn>c1+pZmk~PC7%%w!L
z^??onvkp3li=0<bfpQ*%(8iQNId>tQ4BQyz)!82N7{v8zLE0S<(1M(404iLC$F$gk
z__aD}n9Ed0bD$3g$08$e5WpIN8Au6j9JfC6;y!fjVfk_~n{_c4INQ<B34QeR?J@P$
zZ>!x%YS@wIU4@F$(3mtM9mgjeJk`tPFaH)8rxRCLF9wEnp_vzp{Lx!u`*>qY8G&Kx
zT1}dGKMROak2^E=HS}!|r7{79kH5e+tM~CXe3<85jdZEam%h#0a3Iec`0$XCk7d^Q
zm(k`|?3xU1G~xj^d$lopV#`sYSx_9PKY+v2v{8SLJ-`GA3gHUkAKRos`4hmDUNcu6
zw2ft3e7pmM+;7fI$(@iLA0+%28;jB`46Z}lxEALMkt=%s<^f4v6tC;~_?pd_uVN?i
zja9&QT3{IP{w!`h_`pqJVBKkjfgLSH+LLeiO1EQ)a!H;iyrbds(>&=r0y_@t*jm;V
zK35XMNZ;PL<8S-dIm1TP5PK0WJ~8RUo}fPwvw#+kHnemL)N|)~`Ub4>_Jyy<dJhhW
z(V3_iBeZ=?7Nm0Cg|@Ug6<iMvKOYxJo6jzW9Qk`eV(#2O#s>Tb*aKiU1^6NX|3*M-
z?%eDEzN!HKfeT8!mw=ZQVnohDy~`b_|5i}XE2wUmXjeUi;$-fucAzR1)Ljbd04B#(
zzb6#u&JqXe3I%n&g4#)_hXDoQx)s7T`a(ZOoH+4;m+9e8WR9`6^iL8O|HB+^rDL*v
zjK^4q2OT@W{MZgYKNF^b50PMYhs-ZQFn;g$@gMj&#zy}roW{E)GcXqpid}(<+T64S
zpQ4e^7uooG7W@VLZNLR+S<FVx23pX_bJ@t=Xe3s`dbnt-9-g-m|CM^UY>~2X=M*jR
zgo~Ep|Iy~6<s5qu#06eBx*&$(szGC0c6<9{Z`NLVTUQ#98!C-(QGFo{vK5#5BJ=ya
z0C|9V&!UPh%}?ON+l*)d2#N5oAjof`M`Z9mc$AFrf#?BD0KiA+4Dd3-T)cs181$~(
zTJ|c=hY{xTg>S?Q`=hM@fY~iDax;z<@1j*df&$Hv8}ML6Zu*6{VOQ6}z$dNIDs+<0
z5vmg~l}2-ZjDgzgVSsUE4;v<p`#dnX9`6s{gn8Nja~Xv3;}h-G+Zu5AhyPSPyjgGf
zqN|br#VtD<ckJ1Z`Fo4r@MSk|k0FQR??#8BE>r~T9Xb<z{#@0QfxrW;ACH~W))tso
z-=$(CJUZAM+0s>y^ere{(+3r|!_QcA#Mzh(GtO2ej6n(BKqD-uAC-NKmp8UmF2Mxr
z!O`Q9X>TVj;R8Dk4c^49&S~+JQsRXI4K5v@Y6zYJI;u9K6JO3gkv{|Z<HgS{Z^x%w
z-^U4WEUau9(RlT${TsaDP2PrBxA8q(uM%c^p~@?)+?d{3{_~0aE}i6SfmvumcuPU$
zXCn$Kw{&gX(Xjup7vDQ-h;|#>V?<-@Vz$ui`PC@U(d-{?j?Qu;Au{yTVPIkg1FQr4
z&g>5>%5o9%85;qqo5KLIAy3%-EnL9vPxk}6FDVpu9|8F2kI!Ps8TUKzzftgyDfj^4
z|ES=z2)~~dWB;!MKV8AkQSdp0zgEG&55h;A75p#<{$~pQLIwZXj{$$3f?q-S`xJbp
z1K&--pQzwpCHxHvzJl=A0N&ECQuL`|2xEiRipeKaH)mubbcX8EAam$*zwe$FINU`(
z6rfd;zhO0G;OL-;XAngC{(aa!!2P)UxaYP(Z}{+7BjT=tE=4(Kj+~44jmQFwr7tm-
zK5LB*1qONWlNw70qfKYFwT(7|XQky~tYY8{8yi`CRvP#@&<Zu69s_jn_T}mR(@oc1
zQhp3fci@)eMUQCQabW)@d@mAXCu>9WTjGKv=dcm}Dmo9{r40BSC2*+UfH6y-YG&b!
z;0`z^X1cRsE=!EhB8VmC>t=|0B-?@|%Gx~Kw{4Z2Ac{Ky#W5&kcbo3kLiYn7^q6NO
z6}~$}{xHEr+z7Cjha0R%;J_;k9LCxB5=#PQuXHo)qc3Ojr0)sfxMv<ZA+2)6Pv{@^
zL~;R+I15I$M2CN^sw!*?x>urH9Cvk7&|QbSrUTy=;4tmE3)G|P76R>t57r|5V2-bJ
zUwjCw^O#fm73J^sf9)B5y0)078^y-qN8$-#E>!(3P50(+&?2v*;Bn^EJ(*+7if42A
zcGH0`>zfWPD1c83K}|gC8}04c_#oShgZxim$MF(}6x(ZlMU|}$RI5TFGf|bDc_TBe
z64s@ZK-5_!cmc$UzekU48fif+GeKC3P9RYb3<VySg4n@9H9DqU;3B-IM(<a|)7C*{
zAY|ScZ)qd!=?$5tCzNZq7wnDDMWKUEj}oE=Ku8?dhns%z6Gh&DmW1|3AB5{6){F=2
ztL@1OcDEJut7Dh3rW%$cP{SJnpB^5({Uv5$61Ml>=z(=zpm}M*Q&ENo?ClFTKuC7=
zfxUZD>TRJs8xPJlA}${41o6GmHO6$MCjqWb0gOKZsds23rCvUy-rUoGnaI$7DwIb_
zy<0IzVmqOOsqf2Ry-NK+tG$oy4u6aDJ&ef6M^2*jJL|tr?v@t+soh?z4l&4KRL|e0
zYZq=q^EExNV>hOlE_s*^yON(uH+u6o&l`7<x%BJIN)(9M5%%WQ8(FxCC5d^F4mEKj
z{zp+?w3aJeeLU%U%Ue2RDa-Hi70<#dyHkYAy7-;ROVp(CiuhXu!uBg*yWu}X4BKbX
zx8Y>dVHB5eaSY%d*5X^72!Rb~@Ta@I=5czY;wNzGZP99$qGkB>EDR4krYGw}xNuL*
z#E3r8HaeWxhx{2ClFt=**)To*EpO!VJ}Im33%ud^dETz$z#<eM4CDDAI*x%0RgG*O
zLX8n&lQDIKJ;z)Pyh5eA1sBXQ7r{ntTNK_t12TI3sdzR0D?KuDtJrgg_5lPYta#rP
zeK4U5O7}hHHLjx2$Mercyt|C=@#dN@GSgt)kG!-MU7!)5KZ>V_^|m$n@^_Z~qsTm)
zubhQfrY{+9R(yfbT?hCO)`tK-)&g!pC?vj9;_&tQ4)cepVL3AWM0ToI8!n+u^Jf$d
zeh(YH<7Z$za?uG-E55IFVY42n2Q8rcr8j!p)*Jcn_zN#Gi{FFnC6ml0leBiqB->Km
zxn+MA|80D^>?Y?nkMLCP_se+lzc>g6y>twE&Uoz8z@iumaWy-B$=NZyf7Q$zuDJ$?
zy&65mg!O^S{77X&0h2if`J<3;j&*@#Tbw*3pUdQ9k@`|{YA>XsroXAep0%?+#0v+}
z2&Y964o(C$V7PTVoj5z?AYmiD*f>3M-pBHE-aB$RYqKN{y<RT8{vem0%Oo%3LAmt&
z9WUYioF8b5f-Cm2LM#<fVgxybm*A$-zPUwiqSq2DfyN1e+g+%6<P|^RTz%fxxM;P{
zk!K?3oeh-<$OZi_GJ_wHpUg~;S%Bl{a<`cJ1o$0aOKUomU4U>aoso|EFx{M`>t>yd
z`3&K_z)l#&JnjPdk_Ce#20`10k5Wz#N0Sl1p+5KUj^Lzmf?ZXhd_tNQdLauf3Di}x
zV_N8S^|VMm#qs2AYedkP%bVY*EuIM3@1Xb`4kSTwuPe}DP$g3?y>z+s{N-sxc*bP8
z^t_B09LQ7;QuuLScqwkC%Z(s!B5#Ssi|m_`_RS#s=1lvhw|&#YzBzO%UytsSHZ@KO
z3<r+$xp<45wf1Ki-@D@Hq^{4LZj_#aZnQP&rCs&>CN1LPBG2^1Gf4;?gvi<gGZY-`
ze-}bzgBm)9z><yx&^d<{0P9J($HWZ|9)(sa5?rp@H$iH!?Hv%~rDhk8c@b-CHb=3x
zCe}8JwOxeTs-?E7bqRVGyHOp8-dS~T0T-&e&)7F`pj{ow%30%owYnKjb>;1<Tk=Cx
zcL3f|)w>?;suOkbf4g^gIMw|Y6GSpHm;DgE`;k-KxvAAvE|=<{fQ|YNg7JTkIUnz9
z<8#%IR0m@%P+*+#G5TbrMusi~m>pIU{oo{|L9Ypp%Y=XkZ@}5c6!wmiH`)c>%rm><
zw=Wcj7imS76+6zlkKRqPMQ~PgYVz$?a$aikR4ch}YO>Et9`qtwkSP0SNT#^eYWqR4
zkZJ1^gLTC(YPWtrg1+T7*KS0CvOs5JKLHhzBw%D4)XZ__3K7+XSc$NoxK3bCam~cl
zooFe$na5&&k781j`O%YTO^0OcqTCam#Y!c-ohQ&14dO`8PIOdalJ-O|;>$_BhyJi9
zI#j*dM-R)ZSa^2!s@B=72SI++21QL?#eCA)t8W0RL#;2VSNmD7ifHQC;0HQ;HON;Z
zYbn_T>8lHH$DUj-8S{!&YMzyPnN<^_ibU6iYK!dy_@VK&2&BN8A~N(B$6-E#()%Xv
zU*~=4kl!S9Ul`QCqrDW1>vge1xRu#rSOyseOWaT5FbbxDr7_th6OGHTJ<&HX9K3K;
zly>D8Ya?ComTV>;XU_Z<i_Y)O5sOhyCmd1MAz3)$D%SchIbs-JPH+VI9(Fsn`c(D)
zzhH}24C|!#)vF&N@n5sW+Rk2Gu3r5YY*F3Wt2gk~&e)<@GTO7n1yWO}5?q4O;WhuV
z2zNenEoUH~`6tc{J~OmTE`M2$3#69;{_dv*Kf)8#PpH$U4&I)ySLnlouTvlMkI>t=
zC@cHI{3*g$_i43fK|_G_818vJ(|?j=zFD>>rXe-io>#RlEgOMPxyjOiALxes>qFs*
z=w;KXITxO=;VaflePelgYkmXjJ3%yL_-$&Rf-oCgPYLGOtM%bC*Gv5XUWfYc1W0mz
zhE$KOgy$=JVF!L@Q@UGC36+7>J3wG4Sfb{G6V+Gdu2YEWHMo9>F?>zO_Hr@D_IBym
zUSoJgN_#(mAnmlhJ&({{t0~3Ol#l&4sivF^w8(nQ82)I-_JAhRp4q6Z9sZ2rQ&ZZ*
zbF2T8_B8g!>V|R53?0O``}X<ERLk<D!X$b<)3JS=7m2U8aBbH<eK@wYlD{naJ_C^d
zq<wC*Ph#cI{Z^J);Pc+73iPx1ZFmyDL71UE^T!H(Q}`?Xb5!10{jqls(qMycNF`fk
zVyEOdXvx?#z+Rf)kjr&KuVP$^!mZr8k@C{86EHH&j0YU}Tm|pO1&)Uf@R<t!S_l4`
zW2F3YT(I&u{sb)j`;M~QN2496LIrh+g32M(DTJc?=oAO)M+$0yg32P)ser;l^RDAz
zMyNPWHY_{zO<J#4?c^$v<_erq@^_pdVMR++xckAJ^HaQtY3Xhh2;m%EnoVf{RjQb~
zGc7NOeCsKik1Tp6PKVzZn`dx~<6*Z)&)^%8Pyra0Q*;3>q&g~0+W-q9*DkOla_Yu%
zkqd;@9G5$KDx68!|3R^*1&>2bI(8Bl@OxVYu<5=CNu_is2Dd*ojIcP=ux=xgW%@<D
zq!Iow8U<1H{HPY%B>{?`MSrd_2Kz6BLl|&c)d6aOTIf4u7Y63JV5{tnx{Ay)%#6OQ
zQ4!Z*y4&Gb>E`W<-Fq<6_Va{iBeH+7n?gB^6EE9$D82Iz`QW(`^P?V4mrm8APH&*8
zkB>WZXL7vn;;|T0cpP1kInTudfiaC``OLChsRa*xp+(?KU!2>{6dfr8yylfd_<*H%
z(Fp9L^zdO4+kj=*Ff8z92-1kBJAO{GG~k%a35^XE*+m9_eKWBltE4uA>!RJbGF3uL
zjEJ3HQ0%R0T7f(xyqvhh=|(8s!)6phW1C#Shki!**HEMuM>#9#W??^ghFhT26w<U;
zNWXbup4kJV>5>dEy+kNPdVR`!6_Q|m)}X8#6rDXhUtHh6hi7$6;eMAHSk8O33mX%b
zBG9{MUVjeMd_C-PgF_eOpsYx*o!Q;e!ZEetx)1cmTG^Zbg%<kbVNyWwlD44$+dBhm
zgotTca5m)H;9W)LG;l?kt4J_^S(ewF2h3xM46OI{nO6$tbwqv-@R{?5a5l!m*9WAR
zd3ZR`ByVYmIKplXpt-_5%LC(6g%;`x+Cs5M?n*0H3-Ze<k-D_qZEYht?Q!P{0yKgm
zaG`^Lo2hF?t8RAJ!$J2lyoHx3Pg%4U?DB8COhp1O%lMX;TafU5YhI#WUW1FO`Ni+~
za*XEVcJlXMR%)0(PR3E_eb5UMv=a=34m(-x`xk-<%X+JQ>qDcQtVOD0^|-KO2DYco
zpi(@gtOu<3{bhbmYOg`^crEgH+mWXdh4>=#3~%J2#kh@(`2^t`9<a@bIRn6yO=)Cs
z*FrsgcUz0ae#7V<c@^8}Vvpes%m1eG!q13r2SXQ}U*W-L#3}MPdS69!nQktY?65u|
zep5UA+GUHY{Sag*D&aR%-HZ6%v3mL<j@<Y#+kSr}_`Nj<>OX$^5<8!5owB&CjbfoY
zf5zw8k|H5Fvb*BGpqK&<GJ*Cc3|ZX3`e(SO;HCf~LpQ?9D$xe^Vi~XDMI;0tNk%-9
zWPFrsco9=AA;eSx4li0}0afxEUbI{uxw#?ii9xB+DR=fDBu0Bummf+VJx`Bhyh~DO
zPiAe!{f+4)%aA2J^ti`v_zd$fmu_>*ykC~9Jecv^Wc@+*3Bct+X*2@juo3mbQ3j#3
z!4^uol0*G_iX57nm5@W)u_A}`{1qZ{w9pbLH^gv1I0ZO39Xq<j%<HNlV4wBJN?CnS
zmM9d_rei=J$i57>1Je_kkb8Rp3O;$10A~Q8s{^1ZfY$+_3aJDPC+dEg;6bX=0Dt35
zTj|-a>U>m{R}5IogPyT61tyLy?HLa_Nek6$H>lT6;cHmL!SxnI|LpVSBv_?tXDdV)
z+)GUdu<`M`oS=Ivp5><AzFB79RN6OF?3)7nCeOY}vu|2Zkt)Z&S!Ul<svGR!1pxI7
z{^sGAbH~UB&*+9d`E#P5_JGp)QzZOGbz@;yS#85AyYojhgTIDS#3^=Oz05Y}-7n{~
zj4T6q{Kgc1F)z>7O||dXXJ#Gpn7?0%mm5EP{hWoQcw@t*=e*bSZT49&?9E7P`u3_u
z{J+B7txfT(-oXER(6&6`eJfpZO2v0wK&3!qwz$=>l;1F~=pxhqH5h(jJxYTZ>msW!
zUGr@;=TFenvkq0x_u~-8&$ZCaq%hAl^<rNi_RTWvLGx^jO-8zT3mh3Eb5w8dM<ujI
z%^zh%h8~MF*cWX2!3h7Yn%NQCg)ld>0JEQ~5t_6od>DLJf_KHvK+}731wQ+^Pmi4P
zA+*tW4$ph^1~z1Rs}Ao5sr+YOWV(hx3-W(H)*Ny+J6(_7XZ-E!3~<E<E7PMI3X8vn
zh8+)I2!;sxB{4)|1>oBx{WY)?r#HSKZa-?OuTc>WzZLyF1Kp<n_zArN{J7M}3S{vF
z7NtBw8Hzp&f+J%VGq5cqVVQeSWn9OM?Bt99G79M>rE{r3t33{DN4QDWfInchA7ta}
zMXpW6hKS;Cf<c|@*9%!F^pVo;V!MAG-SU}Jzsy8P)*0ATv?*!5i~zkYnMVHA<^IhK
z70TaOwyO{a5&EI7+=Y^%L;rN<O>=l1blbF<Gr6L{$Gm&6o*jAwsyt|Ne>XI>a*do&
z0@a3`-ZR*T<2<WD3;r3!cq8ik67~8{ypGp?&DXF`HnR)&k!5qU9?pIK6Trn~k9~hl
zsD<@4;)NIw<eI+3-UYiAPdv_NIL|xMYX6Q$Bfm7$f3Dyn$sQ2irsr=(Xj<grbR&Q2
zE16}xxbHIXC`&`HhTJH9yJq&ZX|dD5sH#FAD%8U-qmgSs-42Zm{t}SfmzXClMNn=N
zz&RJPwFB@nP9v$0R^nac$;m!i@DgYf;ASA)?$)Ua50e2F^C4H-GfmnvXU4Vu*qunz
z>t<ei%?$t5Gs~!P4<g90X=yP(vI@=dFwbbUrz_6*D3kK}l)Hxjxx3;XYrcP}KQRMe
z<UJ5E;GeaCO5vH|u&k43PUvp&%y`6Gy{1Jr3e!CLB?lu`P(!3yv#!1IdUR5SiP_|T
zz}1obT;P!W{?k%Pzo)y9emY)>_u+U|DEkMJ6}}DRB-p{C2HGqLJzp}xCNOGZB`3cO
z$>jtWXmx3Ko&mZfTMtK`jaS*bq2M)=qO&)Je}GT;qvKidI~!Qa)iw%(UG9j!8+$iz
z!K&jk5B9*%8*a16yJ)gRbo~~HjqCP!gn+!|3=6pag0{A%gICn6UM#~4kU7^A{G^x8
zw8(NOJ$1c&q((;cHX}c#)t0b_Ah<1BK49-AT8!37?_yXGDmkLnZlzH+au%mfa6(Z5
z86_6tKuC<2!SxWRp?$hm^d>$d?bST5Lbl=QZC^O%<5MI<`w<y!=yAMg9>}|)-6dtD
zTr>~H0L=Sbn`G1Nnwc0GcOmw`8-5!@V;UrHl7#vRT816NUuNoo7b?*pNT2udD-uza
z7U?yu8``@G@Ad!&b@Lv=iGYM%W|I@&0x7a3)0;o397EocXHaVg@3ZLx-rAq3=U-gz
zKUqiUNdE4!UGd|R`cGhlQf@+*w^;~4kSpe~^$7$VCy^9Gd-9?X&`nZC%H?PXSb`82
z^xzs09X+L{iJtDjZ%R*Z$7<D%0_aObnfDM>b#@%ZcY<F3U-b7>v>H@Cg%{P|-jXs>
zE=TL{vOdz^Icd^g>f4?5*L0!3<Y}O008wM!L(u=T{tiT|(cklUQT;tzQbx+<X#HJ@
z02FkhrtL84%Y{W}{rz26^p`eRfQ}$)%zFs>f7ag%&}#H|BrmGJKa-S^ayeRm>#>U&
zowyGxW%PF$I?-8wpXh@Ax{wD@gQzj@A?W{Ef5)NK=&z3#)!#9aGEy!_>+jZH(%;9v
zlm2o|i2e@2A}eM5zJOKVDo8Z;w}hxM?;&WRi9;f^+TX%3oGgz#lt++P$A~=IQ??5d
zh=&oY^$d<9XhZ(x<yvqWUQ~O5%3`t0@n9<eeKB?rG;=S+22NZ&g)4T){>nh~(hD)A
zuh$}b)s+4Lr*tpOQk7^_nt!ykosp?|2r$G7?+_NSB50SPn_hFNz!~8ImU{DJ-~h*V
z_`)q-+P|W|Jp?F}CeqA5rhRFDMQOG$8*ddz&_sc6(wd!Apw}u;#6`yfuVjIiv^Y#q
zw2+{Q0z>TzZ2eZXP{w1&0<&14H5~t96-dxTfkWCA_?T7TGC-wuEbxW442R!KtO5y|
zDDWd#xs$tmxmBRWQ5Qn_XphI~EYxDE?pC1$PZV0)uF!N8Dt=t749>^f3AKSb!<nE3
z6qBgg3I@f}3;-AtLwSK>TL5hgvF#vld@p&uQBSOk7*^Ob*IvdS`xRg8)fX4|78ZPm
zR_1zy^{WmZZi~*`cbEs5CipSjit{x6UYMiBP}$VlFJs0YEKOT;aDZj@RP<<ib3tx;
zdUHXa>>txy+%IF%XUK=)r>|vnDj3mRPy#?MrPo_&9{2lDjSs2Rei`5Fc8X+@DzfPi
zEdTGaB!pNS?p7F#hUlwcAWvhpctW|JT->rz>s1K_y1r^>`Cq1B=n!HOt4!`sA|5iZ
z6DO*xj|m_u^a81@0Kc$I;a80!)IBB>KjZr0=W?BcuAq6$43v$+FQ2txOnJIFCX+Gf
z!E)3H%`~(Qo&s#ubW702hoD&E>_1u(`ZAa<9H>MIgu*x9Qb3}IPLq5!;yTHHU*)R-
z)Jgstm9NG_C;1Pm{AEX~ze43JPVMCV5|zIa?=wyWD^x_rU<KK}A432KHxN$Ht^+6}
zYNcE{1o@gjflXoJF7RY(v@cc(E(RIDk1T}zKA#l!5Nrgh?YNiNT09&D-z!cPd~QYU
zo#K13%76Pv`7>31%aQU8l^;D){s5J~8~I1YmkV$TU#$T3l};OBgxwv{`VYnng#)x~
zb(+6c<@57P>@IB)(B~~4qY$Re_<jY%!22Tu-J*c{00`70{yGWK7V13^KO*%`#mOb9
z)br1>DHorqLRUIQiod-_=3tJ2PU>vOC@L)cP1JH<f=9kjEwfcEr2G$9OO~o-$q!r0
zfA1EIwfwNP6sTJ6|6yx+_e<#uN7N5U7r&}y*$-RGajF)M{vWU}%fAqeQD*<Awajxy
z<!whAm7`QG6i+;qD9F#G{)g3R!M~sssZxAhN2+qmF6r3H|3Sx&aXLl^G9u+VikoXy
zEp=AMLf5cku&Fy56m<x`3_~QgB{bN69T;A4B;~p4ZA#?-5F7sQ(Z;5of(gp?AF!5R
zsaiNO{D8F_h)ZA8%#zsdWB#O_WET~{Zbzio_D?X)ln!|+Ny`6g#HG+Nwg5GAwUwmb
zI{T%IFI9&{@MCYGwz7fP5|9VQc@1lYC6{3+84u$@1)|(bi+zBngEzz8=AW)dig3!p
zqwe~{v^Rc?m#IWT3(nT_@H3$R`eXa3dH{YpcsrHCYjFN2N}D24jyPuVtLRWLBhJN#
z#k`EMkI`>>!V1}~P?iMvdMP&JwvavL!=tJFNhr8gJNd-{J^Z@$2|EPF{dpNE8~!li
z(x5sJ?j@=Rzl34sqsCN_=>YMWkGeOaS%xXW9*C;5;(>6wb@Z&yK~wIT{b%$J3?%pz
z)Da?jK91cl`z5wJSqF}M7CndLw?gu;2XlukHlv=0+wAWXpp-A7E<g!7Fv#9T(2g%0
zg~Nv(a)#xVv0re=ni+$-+l!llq*kQYFL1%)=E8$m+*s#^Jc?l&y))L4{>k_%$61;N
zA`9v{0K%{6;cxWt8h0I^B4aYmaX1JsUB`*QG7RC9UnFo?D&VRGo6#EBq+v4%jr4IP
z)Q<j3-BX}}fe{wF3)D@}&Es5gUPd-Tj6T9Uxb1`6x797r%y>=R^305NxK-YhDX4Go
zCV1iCO_S}{N$WK+qH4j>Y?@j8N@nF;9Oo7*tZJi#(jpU4h_@7*X_oGtg*CJ}yN~wd
zY;2`$#w8E`{R-3f#aDmtwwpnB1-k&Gtwj0!s6zUjK{$dn=$^$swhqj=`mKz#!nItl
zYk{*2q#ev&?4GGIH)?_7k%Ur8tWr3MsZO4JN>V3>IQEr)(RO7-hLN^!XZ^3J2fL(a
z*zD4_L)azzBVm_eG;4MngZ-KnXcJHi1j9~35~cYwyrpl(PsER&f7z_E<2*R<(KCFh
zo8kF>1Y6C}&3UA_q9|@`8e<`8Q9X6|gU31YZ@^Z>@Rc(>u!*oJLEH$U!VFcSTHTB+
zlFPkHfqYi!UTc>j(J(_qmSILXKNf|yd%%&*_k@?q>GSh(?#R6ujPNDi&4J6$9Bns{
zNp(c1ptH)>{)oGfx+hzWJ_ru26qw0Dw@0$zjPD2fO8X4;R?uB3S?KcwJ7o!Oa54#N
zqtKkCn^o#QD568Rb^w7}-bQI~Qro%pe8c{rqjLOv?Q(+dR^)mUMJ>l8Pb5GwoJ2er
zql0iQ3boCzfHCg9_=(Bm0U^l>&i^aJT+v9bN1zrDyPU^^V&?I%%X$6q{e$_hBRWwp
zg}DUbD%_q97FYrk8@%l#)gRo5cBz`1<9Y_pc3=MfvX4FKU+WM{<_Zx6@uj+{L<9NP
zW3CZmU=ueA0|@(|)jo({4%(mD{H)ZIP817Px#i=-DO+ZL{cmrXg^B3kKfvVHCW6|3
z;=bJ`g1K#K6G7}&+!*FFBF(VdwH(CG_|xm0@#aMTbTXWwOWkgvuO$)zN<j&RnuSaB
z?nZI7VKOK`lYX?EP$N3aYdx-v)FI!&RpURwESmAaevB-<hHp&+#8p*^bC+SpL0p^g
zQa^B@)b)(TDQnzzDO>TD|G4H5S%-fc@OeoHkYNA^Lo7!w9fZ&I5grM&v^4IR*JmrF
z?oxLFNKdw$p_gv;mA-}k^=jIV{(W9jk5K@wLlK0o2XxTgf~TVJw@he|ggm?s*am@Z
zK|KlBm4HPSh6EZU`MXo{dFGFdonitWMCsXbgRwKfPRYd0MUlP_p@g)t5zj*i(s<L`
z|4jpkBg2o6xE|o#REc{`hQ9D8Md3q*fsao!FzhKE`0bpxnb{v(i<gRd*P9Nf;4^MY
zer&nV)9zlUpZkm&iB*SLP=R*$hJwsD3l=@&9swXv_@9AKTXekAf=qcrtl={fC}2Pv
z>3>-7@ZK=IXW+1s{+rw9`P~aV;dcW&Fi@AuE`I|9u@VJqcQ<(9)4zLN-ph-M-Sh1;
z_QytyhXkiVH|2=nya1(pv7hnodN0mezPll>VG*R2RJl|RIq>4hA<DEFNYeu!RdRZ!
zJZ#)?Tc3<Jcu<8JOOOkuqr@oZZO}akHQ<R6mnxcYCGL#K8ut{CN~J_mj0a3U2s!a9
zE&sgD&k6ox3OPOcw+|1u$@YDmEfIZY2bJ8Qn}s0u9aZrT2q1%M4@nBE)PlWxqve2P
z!#GL^7xAR0hl^&|rEvt7qQK}lRDq*hb+l?Dz!(QRL#r*vEffz3zT=kAA3l7A$VGqY
zICITT$~&)l6J5<W(F1=I9q>0H!haX!UFmwrxrGo&N&brk<q&w$XJEf`q@9ufP1&cE
zcFFa|m&*5Ad9W-rCNu%F-ZL`B=6eU%+w!fk!I<o-M1Wupe~rnW$~->Eg--?Q2^T(1
zQBPB_#L6@Bch7Es;JhD27v_IK;kpPYqF^OOAS@oWg~bX>Sga9Y!4b37LS~(6*BTkk
zy6qdmmHeVb?nX9ozlc}N^&CfevvGE@O2YmryZy#wgDp2E7qiL6<Pz0Z)iueg$dRmy
z9AvTmUhHSV3GqQ0ek`1WGtGZS>iVOx;R?YI=9Nlo2g6QKc>a|{p+}JFTG+U*Xt|9<
zVA-=Uk0}XMN?L(RL+gm%0!f>q(x$XS>z{$ol!*z*#3;v*&>|90+#V%bb&wdck`3T5
zjrc=62&!={#E-Es55L6=S#QZgsTVCqD~TIih#OqKAKLzm!;%shJ1cp;1}RE>cpoP|
zl92XUXcuaJ#!Vru)&3f-Lx2-yTADu#!MnHE{D_U)Xgwu_+HZrGT;1?vbd&aXllFI;
z(sXEYebe5_o0@h`-rn@_<o6(qT;1_wbO%^Yad&{_lynE3qm`udDi+#;LeX$<K8z#0
z=Rq|4xAFW*WNu{wHubVI9s1LB5Vc_ztRV-qy<CQK3J-LrEFk;ZT&q=cqm}gEfya|)
zq04&o)Zh-!_yU3y=c(bh5IhUBoM+tOIoo;09iDTYXWZerkMoQ>Joj~;`+~3G++3&0
zMk})Ec(vUpp>N4}sUp9*@QX2Bl_$){3!e(q6E1w3qMj1tT?V?Ev9w4s&m~6Zl{Op7
z_^|84_*7LCvkHt)RW=(@kt10cpQ?6baeU(NIgC%9ZD|k96J@+#GMO)dwW^ZFqwv;B
z43PO#Q};uWe`*@n`If7iTvbT$)PUuE%M)FfH1I#uYy?Q!6qVK<4IG~sk}KRTBBlN&
zT+mKRY1LIp$&UP%_Ta~;0i$LY*ML#8OKPMZv=A*IE^r|(aQS{{`g01JI*dp3|Ki8n
zAr0eEtE~mXFbt_8U<HL?IBf~0R`7LV{DC~#t~CA{X?eU9rx8QhC20yN8O+NuGMJZB
zoWZ=Dk_;Jm++8T8wGa`A*v{Xk1vwKt18?8D?*23F(Wk~cD4Xr7Bz-+0ocboiEv|-}
za``E|I`~Y1=YyxxrQYTRPnqh87d&ODCtmQx)0a>WFL=sTPrTr%k9y(-Pkq&sJ;2Hq
zA*ORdnm-3Pt4NU-3V+MlOH9Us$CP~H&XvwzBb_+QE>0)TvP;sZG(m=?;|Ce|MLK?w
zOc}<pfavkXBExJ{ipa8M7*L5V`#qI7k{ndeX0Y}odm6igK>%U4(8BI6TbNY>diMMy
z!wV*=TzJ>BP$9Pp0t^odw^fi{cu=UVg6P77!fX}f79JF0t01)SpzvA+*@XuStqo;&
zHLX5p><kR8MygbU-am|WN_Nls4Nw!m!Xki{iGE<#rmwErP33+wMcg<Jc*h^!*x;XB
z97+R9bL7G&;u%l4@QLWg6E1w3qMjTH_mUdgYVmA~cr8NFZp*cVsJq%pvS!@rOPu=s
zlS@>MQ&m+n?W)pPqvTcONM1z_@|46|SN4*gJhh&P@-CWeus?HiaIqxb=N6jgM0>uc
zQ1nofg`%&nswZYPiWd5yaKQXgz>qz^V$#eFG#$#p&)&(!0pT?K@}B(0U*1oVdj~6<
zMDC}~cTJ*BiU16Eik+zm>{v4f^%1Fm@8lAxbgEQ0Q>v3@G5JsqlMm$}*`5bBr3g4g
z<T&)k+?N#q6J-S~+<Za7c1DIQ;{)GiYJrc@I0_mVJD}a;0gJemT;pR;4cY@f>q8{k
zy|JfroA!;5V5)f<CIus3EwubUMpCxxY1+<4HD;vEUek2&;`$C?W$z5G4}6!W1+IgT
z1xI6h3~neC*ale{=unj@A!5%X$6mjnJ&WGu+T;k>W<{Q-`!X({Ey^YVf;eZPEzznK
z>od}{YVJ2nw2Zl)r(JAxgxP^^%tmEDLS?E<mx9S8%(Vg|MF3|1J2p(Fg2^IGk-!{A
z2UyPQ4osGUfkl2`#_0lc(1Ll=fyq`dxrF&{9ow_df-xPKTm^%VlMc*yM_|6PV6Jgs
z`Y0GgvJA|4mN3;Xz``Gk0Z`2+Y}zmj-1W4A7A({VG(u`1Vifs+9+fW`xJL51Wi64P
z-1`e9AJL>q`N=*1vE(C6Iw?Q7*WD!_+ijBalY9JG1ABu_yh-`Vy=|8K*S=Hvu~%_J
ze(sdPlE=%dlb}z(I*I!9tCNbMPro{;q#K7j)$ANmvvZh?xu+Y(ALpLtvQ1n=wg9R4
zhKe_TtHaLz<D0&kg!P3zoj}$8f=qvR{V^W*eE;!->ti(}MPj^m)JH3_J)%TpD<vXZ
zDG}LFBAO0eg+}eYijCTN6&v;ORcI9Dr+Vc`dPj5Fa*i!FSce@l|1Ers7wn#*cMNv)
zjt=u9G>@i(S78~`fT8ASA)nauWUA8td}{w>k`#HJZG<paCW^tNrRGhPsiw~mrcb1$
zXec4fqDV{8Q$m<bk;ci?(N;p3Q<0XUvxG3UA}vL831NmsT08wk3l)R0Fh@Y_Eq14M
zh29KL6Zwbpmn%pV^VdJw6U-BQ^TMYB^@Izbrl==po@h|BCrpV%x1|7Sktaxqj%(J+
zq+~SYQE3PXrxvfwPHSu^k@l2KHrRxc$;E6!$>b7Mw`!_nVNzGELzbF1nz?>xb=D8U
zwQQLG)C4~^Hq4gtA>@PB46>6m$YX%BI5{W=Ek919h(*vMQIMSZr!#(bfxOV~Txc)!
zI~Upu{mwN7mlEMye~GaA%o1VssU=`_<mJd~IqJ0>^;!;Iv-Q6}wb%b#?OA$Yf~B4I
zkspbqxh3FaGLY&f7O@UL6VF;TPl}WEl?9(|1ksJ8*MKRM!L%HYWMzATWL+dG4wfe*
zl6R4)SXiErNajVN;$eA0BDoidiizb3iDX|SDlV2MB$9uTsMt7Y>x){*Cv4m{znmu&
zw!yx}@1YNW(6X<=TK>MWuiZ%d+KsfY-AMb|jmo~}Ep7Hi+(90kXzNF?sb#xXAm$Id
zQ#f}Vf@4?qu<R=PxG4QgvC71Mo8X^hpMvdaKv>*aP+7HFejq?JjRgEr{Rs<RP4y>S
z{crmd&^Omxa4Hm`{!~XoS%@JL@nAdszP&@C_ys7E=y$Q~lSuU>K*^m2o{Q=$Y)`^K
zY>;g_SQ6o?u4#8MbR6VE&~0ow2*+BsXg&kCX*JuC&w;(0*AUK0RaRTup6ee^>5yz-
zO>YQqd;m~+C~rT`reBx8dBIsIG^+*+fsImwQyr_KaUMFrwmH`~VJ@nH$$*w)kl=8P
zEzFM8;|Ty8wVHR4Vv2X7tv)PQSbYhtE#8K#cz=5ewdH*!3krOVZz0C6{JO1eC0xwP
zo9fY?TOn>$950?;Ij=9yHT--a)VcKYh3pCUWO~`zwr4WWgX5JvsFChOe#lbkr1?dj
zvW)lyK*3iDS@`|^l)!2o8gKb4*S=x-D^&$>Q>wLKgvL$XythRz&qn34mgFdcFAV*m
z4S){GP%F}W%C$#_WIcTvG_25jNx5uTZoh=PZ;2G9&JF~ynjRg`VI|&8@*FM#a68{1
zTL|Bv?L3)ZhBF;-+UW6OKjM*zw!ORr%qsSBpE;lY!TIzL&ZmEHzVZ)xOB=nBzLm-2
zIWdx}vHKMIU5@JsISchmj^~}!YbsHH{65OgMMR`o&2H-E9hq{moTc!H0xZd2+9HqD
z*{MJrA1I+uYAV;PU`ci~ZT_vQdnYLU^ncnvYOJm&61_AK_;Q5)r8+;^7_IgJ{6a!u
z``RtgC$Z@v3u<Y6*!PrITbfhl)qiwkFiyPGOa1qPgPlWz?+{TH=HRFlOqQ2p3}@8g
zOLCM=bqKWwwT|HI<WP?O=`pG|ZCdR@=^*xiSp1aif4O0t`KaZ8nJGINahNy+C^l}$
zugno!@Llcu?Zuc889}ErM5r^rdQE(_j}DnLq3Vl6=6P8wepB{I_-#-6ZhL-+FQoiq
zZ{UF4kNt&^Q%dAf7?R@{Hbm#AtlX8$+Th?u2MmPvsBwfM@JtECfuz-v-|#K(z~MZt
z`gZ1o_Za;T=v4<47|xNIJ(C|cXwf5C_aZ+{4}4fT7+?38*C*rVXEV}ZU#OUvbs~R@
zPt4<`<irAArk=<(K~``46!*@<Z%J>i38wajKJq!%1lwqjLh;xPTy~7Cx74F;kHVqS
z-TnQ=%hv<-B<9zKxfJ5NCt`}o$+B%wF(qqp`h-^f5LzkkZF@%AmjVwp0H#xy^}eC(
zPzDU?sz5{8qKuj>Cqe8{MoqSpKotaK<T?prmojSlI0<5#GHUud2?>4U2Iw0g6c7qY
zR{-P}hGoaZ{E0ZZX7|KRP0@+ln_4Em=kyn{++Bm8nPTtt*8qvx6*W^DcU)Qbe&deI
z>i*ui!&kStaYs=dk5h(I?d{pGHhnw^j*Urha7?Oi{Ak`u4e3p-lNy@7t-yBsih2wR
zR_w{&twnxMir$a2=s9@yZ`bjS9JqP|pLX}3WZM%sXEyG*c@;ie0b}asvM=<&!S1tX
z*!m8}707WAmS`8C9^crw<Jwg(;~CBg5aIC#+ya}DC8^no)Lc8YxTX(YdwiP!^d-PL
z^aYNHhO%XXl-7X_bRw@rYiIXrntO3Q_?ms-UXxJUg0tfDIIQp5gpe|L5ee2*-Y*qs
z{uQYRw%ZS01%mkt4D$jM^YSlnyz%Hb_T=x^B99ZfrOlmTx35EAGE*voVNV5uZ?f=j
z<pU}O!%VhOtoTvkSfwi@t8|5B-rHzaFsXRH-^TOoZZucK&uoI0?w&=o{?|rWw@@3y
zToiIG>{GmQW~XnA(xeHd*30l2W+?i<!sxt;v|sF=0SHW}7_G{JB}!D_&{^;tAS8s5
zGnIH(3*HTSBGeM^D6!^u;T4pb6UuWaJ^!-yL_}qvU$xM$TG6js(XaZbW8{CAe$~kX
zq!ap~eL#P8(g+b6RI4Gvd!ssue$`3zt4^X{wHhYs-J{d*G*Yf|Lt;Q!G)tz{<4<Bb
zguRzI=A?)9ZN@xN-*ZCLI2d@Nm%imI-7G|rkUma2LJubbkSX-T3XC7bMsZm2$_+_r
zr@RdqFw9gt!-<ROgA^Dm2jPPA)m{adeW(e}G-=y}w5>=h3h(woG{LxyGnQr{Zbsj~
z)kxn+Rk|MhQy6Ha)rd;i5<KeRZASl?HH{&xB(q_W;a<R_4`+n;1U_uxNp3ADl@Ex3
zQ6~{03N<nJ*)Jh_C8qb^Mi{OUQi?=49V{Q7a6{ljiPG7P5)s^6gh(2ePw@UP>s`@a
zM#v13T(rxcUR4!~-H)-s5!bW0V0vw-WS^D@@J2kMtty%b$eThO1>JpFG9H)Xg|>}I
z*lH!r!drSopTg>QOKJ7ky`^GO_yVbF=TlFIl$HD$NcTlXO;K?%%cSnbDx?Q*M*#_?
zLaV(7)hNFe&dZYNI$oyyRf2Qk9<Qd_H#jnnY@a)d!)jd_zRIDfM2D5q72p7Tf$es*
z1k`yy6S!evH;(zIX2Zs$$XBt)B^*{e9c5wm&Cp*U+CpSnvDe|=g?o?XprVBh56VGB
z(;6N`A%G=99z+>{1wtM~5rE}E9z+R%#X%ly1)yvR@Haor@8EBSg+JoW@^{0(CkJjJ
z6oJS?ECMeGM!@ChHl7B?qjZ9@CzDY|SH`IvC&U~5P0;@_|0NlAlJHZWk}tTzJP%;o
zf`eIJo)m^K(aV!UQ4Ns;NQI-C6e|s8e92Q-s%dAJDGNiID49f^h|BK4rFjr7ctRZK
zbX!YmdVAu>%7EtR)5O&Pna}-dKCi%hUI-q9&b02bI!qPoe03$5D%KU%6=SMcH?q#C
zhEA$fr-r8YCm|k0)Td2UonB6F+CAyzrmrhD@n5EOdg}}KHNaguJ(CP)9b<;WOlDd3
zuUt6HuU(Z@EgI9NGT>81H+G+`@QHk2y#$zCck=_eXwk1%adJQ!G*P_zfPt9-PsP9|
zAc`?JNM5$e6MPl}pOPm`R|0(Ey%Nk20@)X3q7-2}gyI38tpj`h{c+foYI>grJbpEK
zsU^k{Hc9g>QWjfc9P@M@QksJjf`-m=L_7GE)rA^km{PVP*Fet^9<6LQT&_^iX|?C#
z7qJ19kizfrX~N&_?VU}CujnK`hrE7t8+}8Hd@vAi7+i^}lxQ&08}N15?*6kJMz{67
zrh|+xBJVkddExijx2o}%oR51Y2xxJa2gU7_3dm8+PN{$g#p{#`cu=fPselK?>68k1
zNVcPU^Xto&AXOM$oUY)L0-!Bz{xllXD{+@bekO~QEl>7JLoASmnMfg_oYPDMLo<;g
zL;0wgI7KCj0D;HaX7)!t@c*$sE%Fp<(#gSO)!j(Ka<JM-18uV*`W^X%_MlZ9M7T^v
zow&@jYe{2ebWh_#_cSiw%XQsQ_NaX9QHAm`j?StN`Z4-{+v6!-Y%=x7T%T8<m9X~U
zs32wU!6EoKV%dq0eCA`?Z;xc}(My}Wk&+PgQ=329+e1cRJHBlX@ucsv_0{-&7JcmX
zIb*`1%gTU=jGVrbI}Xlj!$qs5cbit+pU^*YRQv5q&!yT!F8R;wx7aJYG|3vWSgawY
z!e@J%$fJK!HjzhVYgh++tr$b@{|&wYv}?W}Xrf^)<$Hk1_K-_2GFON_#6bK#j$@pc
z&7VB}IQBIP3YsMF!9t0@$G!?Nf4IL<E@E+&Gy4A1qV&D-N0ua6J<<V!bscm8W!<oc
z4z9`v#g*e?F#|W158Q<0o~s|tkPzQAuylK4h)qgcCB~G<!C)_P7-gEx!BI02zZe!P
zaltx;=b6bow0-_nsq>ICK7PRY=PNCen16bsDgS?&e`sXqw1SC<>-?Stf6Pd`^=8~H
zwSs&HZq;fg^b!Dj_lPn$YBgRYLwnr_{KN3Q6}{GiBasICyGlh62Ns)Z6%)JS05;p%
zvaBBObVslP%r3EV{D{SXluNR;a7uFKHxX%}G!bbcTeJNlO|vUPM_o`hCmRH8q0rIz
z;>iaEKSZObLHH5sn4&#+DDR;Kpnk{V0{tQWU+w$z+<x?@Un@JR=3+;s)dJ^WvtjIB
z?l??A!t5Izd8#C7b}^iBj>(4g(h*0SjcP;Ef)sAIM0(PK6iznUZPdGo1u2E|eIRs3
z=_3#>bD>ad1kGl5qvZyc1*}XzeV!{*Yl&f6Rs}WwXt}Y`Vo~c??Wxs{UO$BW24tfB
zM#z*Wm`s|Fk_=*-a9&wPJ51k_(mt{80J}e!{f6kipXlEEFrtZ+{pK8?-5MJ}_Z-YC
zFyW|#_IU4L#+i#TNrGW;_6(s&$DSi*9LU-wPzF#Rl=7iCorU{jFEJwn@P_ia(G_I*
z+(^ikfL<^Fg-HZy$vSY$RqL+aQ_-h5;}<9$8H?St0Z558ADw`^xjDI2khru3q|Lto
z5Ml#rY$zYN?$%u7-kXP<<`BR4&A<4MOu45&j%g4Qd{vA528BUzaJ`I1_C)$Jrq`ar
zW|xKr;Q{(0jP;wSC(hzyB=-)exc1zj)oy^yQs(jXs7VjMY4rcX+6bdnD~0ndI)RoT
z*XZBOqq1QnXshHS_L;#ReK%7gL+JWX)4M(+3v+x#513hk3MptjBjsX@%;Qt>$O2wU
zMoz({#5D|&5+%kkWWnD*3|a8^51Z0-D93ZbGW;4B)Z@4Kf)@Pty+99qiFow`Wsh^(
zY?1CAmS!0FTgtLxCnE*U8n#Qv(U|++GSasQhYw{JxFD2f;R<FCvcYLGwx8vt#Pv7G
zSl3fH7R%eVvFy`$rVXqI+PcdsK&zIi9d$1kT#mSNN&ELb&ws3vxZWx8{LO_jJ<mr~
z8r5s0$)fxDohpr5HqvD6{rql~Mr9jmvIKvAuS%oNjWk(_Ki{g-sCpw!7Ua(#Qfbt{
zk;Zko>v<hfa9$X~LziK0pe*et?H*FANd6KK07M!6@exNMsaCxLNnY5Z`nNDM&0E!4
ziHk+*-{4k{jIc<(M)m~K1!~i}O4j!H<=P$>uI+K*+P(^;b~RRk)GlO!)GlO!)Cl`{
z1WTdr%Xl%mBTf*CefJh#`gZpsyvm2s|KAc`g{B+$@GYt9!@03pv13SXa2gW4;a2bd
zt-kb)Lhdu!qNMfzOT>u_afmst7jdGQw{26|XWl@oP1JSl7LYT;LQe8}kVthU-uJwy
z$orBa?;01$`=TQ6ONzW}guE{*^1h_VyGF?Sq9X50io9!tye}&9zNE;zM#%f3BJWFz
zylaHKFDmlBq{zEQ$orxmXa;#(wb11tFFHW|j)p81jjM(thGs2CORJ*YopOFvl6C<4
zrS$g^EO}~34LA%)4LA%)jqDLMjeoeZu4Lh|#y?zESG=&k@ejT_V_{3<AByT+3%#%}
z_~4ppNY)ap2E`FhI8k^c8_nxXWsn_Mw8CkTo%j_Jj!1Cph+2E}(!e<@)&2(Ine{xi
z31{HSd<8hiD#5p<^y*q6^ipf|=H5ZPD58(JGvN<1p8pu77Gu!vxa|SssgLRMF`kMd
z&4WR^<K_oEPwm8mfzc`eyCr9@%GoP9ttzJ#55*Xmc&GTOLwInZsn<S$rpmb7Q+5{`
zAVh1M4->;ui%7Y5Br61B1ai(3Z90@a^C<|%Z2wabjM)`WwKQSSJw;d9Cv=tlldiJ2
zX!e0|bp!lm|AfCR$>;-WfcQr+{m_sm*gubr>8FN+V2Oi?Hp*G(MDA~hk(c?&rM4|)
zUr;?jhJ)_eGZSMetCkqJi})<L3;3imwd2>TUdOX>7jgMGeQCva0h7u<X52+QTB*RI
zN-enyIQ;eF?*caKF6O_2@A884x0Nm94BUdoacoza&vKR~RhJ_ve}8z7W^8Gze*oI#
zhOp#?@8zT@*4|rj0|O4N)0V-q!aXBCiv~;RoXAxABs{^ruDOp|!pw<%VYr+*aVmrx
zE;AwAa4CjxYk;uo!r!S~kcD{lF35r$>jGPZ+MH0wLe+!?mp3ZX^ABsmuTixQYx@2-
z_4LC9`<lD^5u&E{T_?{^xiwesnb~Jzua@7i8BiXo#qzeLY=<5=jD1$H`fJtfN5Vdb
z-J_@c4d7V8y{=xtFI6-<!7|7d!!>My6gx(_L<*%B-pWp-P<A0DH<3cgg_J&t6v{26
z^i8B#_Tdd$i0281?<H%6Hd)LMV?LKY*lq(|1a^ep$?~XeL;hwh=tqtzPMfaE8^G<?
zq;Ys_q4wkkRiJ)94JJzVS)o*s5jALpc$_(mnzl~F3crKA^DB8*0eN>iEvYTMz-d|S
zs~b?=SY1>%6{eu-kua+kG*%CVS#?T6-n}vDJ!^Pvx4IM0PHsP|=V*I9y(v2B^`>~m
zcK(Z@l@eJ-+l3_hETF`QxN7XFi(E2G*{}Q7(S(}2+M2o&_Ey4X<Gou>kEsw+yyA>Q
zJ5#H^6BSViE(fSD;;Kw;A4yeGCzIOLx+smTeOEWGo4l%IHD|c&;<}wQwHB}D+y=ni
zVuoF<_@ZvF$`hU_scThv!VgpH4l!@)YT<=kbYL<H0UL0_I#3qYOSiMrZ=+=xxw-er
z1doTthU>1%T_Q<}=Xc(dLE{94r1XJ;eEuw26W>h|nAJSczY<w#kcm@QFKb*kZTZx?
zWilk|C@rV1u2*UGDvh#p>gpDi)}qo{kmg3`r!9BaF~wbn=0bvgF#FFIX@zXpA(GsY
zc)QUq*b;w-nme>fC}@Nm*tbD|6M3TAhu@j2*#`e=);eN!;|?@<)iPdYuBzw7zp90o
z5v!UG<sb*Y$iOes@jGG_+lINn?5_qw8-O9Y<I=2l;v`LQh779U3y0X;HI_7Y)mR|5
zd}%tE-FFEH_jupCLE)UfcSEdGuznV+#*6fFi{%zx2@h?yAP{y1RSIV1!Y9G0T=*o|
zl#73IZY2?!TWOnQ8w7)k(NbqgDHxoM*^ky)%sRwW%KFl(5Br5!k$>`3*5IE!K#I1S
zB3Tf>s`3F+AHT<k_~xw+a{Ow4<lM=yx1y`!zu?+(Cr`<(ri2l>1POFb!d{}J4(AX{
z;0TP8O5^a35<p7yaE(M%b6`VFuIWV8;&eq8POLv3Il8`8rSz>TrEgU!eXA-Vf09CB
zEXrDxkM~gg2HmSOoeS|Bsq%*+`3UkS3&R%**8f=k%s>^L$sa?>pJK|N+aAcm5QM^|
z<d28)2O5`>KL+IwbTTD>Jd{7o6ZvCM{xDC?N0dL%x|IBJp##?fu!H<*0V%-aE<`ul
z!YDiIK7`O>2x6012f=d<Idr#>H&<j%8b)t1<qf<NO5WfiY5$MCcaM*%xckSmWCIHg
zoTLVf7BJY2CM{4>Lz@U_7B+Gg&vMZy!J@_lEh-g+#0NBBAc?RptJI>^)>^2x4}EH@
zwqEesUD$;L1R-1m0Sy-=>k{rH;g)>g@0oM<l7!&5ujl*v{qy5R_RQrob3QX?=5w2w
z&w$=A4MP@!z-r7~z=7T{4da(-n8!&eXl&Uk&__;_&gsqLozR=-i9!@AlGNyQh5~Zf
z-Y4<H?^Ho=K=TkDEZ-`Ar+$z|6sCQU$H$BhiUcYj@RwP5FbfZ488ipy+dI%4rE05A
zX=dRwtKek%cqBSe_&|gMy_xsdL_u#7BlPAuQCylZuoVlY&lRX>y;wbhq;@mxBFlyG
zo6_5}1@*83zIy;psXtyQbX#dm4fU~oya;jl8vD2sPa)~e1Zgxj0WM<CE&NDqcBs&M
zQ9gA;BZafs)8I9@z8;SG(S~KfA_{-3=k-AsYJJk!wSDr~wS9^L7n04r@T2uYa&kd0
zBqvYpg$@g1FW_fos7T0sE!Yd_6K`cyU*$imy3TPspi=c?AWA_|A|%~8i0+GeKa>dU
zM~kfXW4{{D+CvAP{SfFg=((Pckew&$x!5nlLNDd%dZY_|lqc&E7J4X8)gvtQPoAhp
zSm>QRO^-13&8TyzFUjtljmtUCYMJ}>T_dxnl8R5)dv1By{$qx?P@beEax^6sByuz*
zO-(G~pfW4`)f8qW7bwh1o~r2mwgL2e1S*v77WM8D8ztbAfcq!=>FfJdq~~Gp19H7{
zzUY%>$dTtekk7DO*iS~!$Qb1s#DsVwk7ghy#2dM=kC-4Y@-PNsg1pF67>Eh+MjpRF
zOc!&p9?jkErPzD&6|KR$o(Bu|^t~I4HAun!)+=9ZYpa+)Ag(l>#|yj|VA!Bc$bU%4
zF@Qu0z0n={EjLuscO0F&MSRBr8jhY~he~)PD3xbGG=h_HIM4QR6{bP(G$xnr;|d0Y
zB8YclgBrKwPiEpH_<?J&zpU<#0hogW60bpd0M75=0Z(-M-h(H)0L=H|a?Xk{)B*|h
ztc3bAp*{c<cCn!`CtiWam%YD>^~W7^zfp$Gwb|ym<r*+x2UvBAm6xz`6OhAo!9Cat
zBh@wHMi1qgv%uHs3iw(Q?5-2MDl*Ll5@ENd!m(X414&tX1i$l)xu$xL;#dnv>`OLz
zu_f87nOYs5kuTffmv9Iw*W!6o2u5lu=8H{vcftwRhROqqqm4VQeL*t<n?(A_OQfH=
zMEdDVq@TG&debG+n=g^xa*6b_mq<T%iS+ZANWXB2^oy5BZ@ol%Tj%LYC&`Y=_1{<9
zYLseTFr>y$wWiS{S#F3*twmWn6se)x93@}kGjiWESF&UIiOrLLO@4KwO>uB=;FEjI
zUe0_A>|Qo;FV4w^927*g4%0rT>4GD-v3vur3bJ}^a@?Vq$;(nN!i8l<WQMI3<Eipo
znzHl-b6lM7+XU5jpa%>)sR_SaoWL=T^@IBn5*p^Yqyf1IQ*1~tVB*=CL(-tT;5j@4
z&)}bc1m$M{v<2)_;ElyTMeC6l7+L7|+ZO!7TYTGsj4ZFeI5#H{$S(YpX|*D))?b{J
zv&rwvdh~?3c+SU`^vy>C^LCE!hVu3vX#xL~@*YK<*5WU`5NQ2HTk%O$+~z-E%effv
zJ=$z8K7~43j|9%{T+j<9qVG!+#TslcfjEm*6kvuWx*q00PQjLdcQS5BoIA8dc^><#
zpQ|mNbdxK|TRdr~E2%CptYBC1)}fX)8%;>=ndB42EPmUy_g*3|v5UN<F7i^k$V=@a
zuYVVLX<g)5yU5GvB5!aPdBeKM%kCmCr;EIiUF2a?*d@P%y^B1xi#)B1yop`p6?BnT
z*hSvdF7k@H$eY$h-t;c=W^|D^tBbtSF7oDfkyn0+yc?(Z{9_(WtTFk=%o|V}=P#Mp
zw-0vw4_#l@U2%VS9_MI$#rL!X-$ICDK#Y9-|A4<<acTbQtM3NNToJ`!x$QZ^U>{ia
z|MS;~Jn{bxe@*${=C2h7e_i-p`74GA%#NKf*t|HwU#Elf==>E-H6gVEd=tzy1y8|d
z`}Sc9jtwC-z3xfLW`}%?0NLyNZt)!D!!!5=*le`E=^s5mbpwCASSLy8fuE^8KpO{7
z-Qu@pPpR|UtW$R}v?nmQa0i)gonX3k{^HD>odIj%KBnyvX?y&|gL1y~`vyIF3XJ&g
zmh`X4i0?6@&H^xE)(ol?U9*=pdr|YIcI<pJ7<PbZQEY#m-<SL7an$^dC4D_>z9kXt
zIVH@V$6#hfv*#P{-V(U+UZ3Cg)4S1(pWf>=_-{-cJU1qu)GiL*8k3*eE)E_Vlb_Zu
z4qh3P5B?Yv2TzQN4{H|(?~BRLX%`2Ni^<2gi;07m#l*qCV&dRgG4YA*;^0j&`GxJ`
z;6X9@_`)%9@S2!7_)AP2JS8SRt6dztBPM@tI9`MAIkg9Q#NBJiC~A>|quoEfXM=yt
zeTiEIbEu1Bc$Wz8iSyes@7orbx1b*a$1o=54&6YGaTW@n-*(qMTZ6|Ck{ly!ABy1@
zDpV|+tH$QZGgPtytA#q*^H9IwO<c&;P`;9I90eyrs6mN+&_XSP0*C9WZEz^>$?r~<
zk(O0w`-FO??W4Fj6Uxu9=J1xrVbSx`Gx{35z_CUb-V-aZadL@jTC1A&iA(BXN5#D;
zNA0+&z(ypfB47n4jKAt(kwg^nxwYQi;A}ni{HcFUZFB3O&-@2AHn;f>oBg95&3@aU
z8oOtFqJQ+in!Pxq-Yeg=!B%;oPoDB_o7VGlhcB35+uC5Otbfwkvj)oG$8oo|*#ppq
z7j)yAXJzX^)zv!tFI<DGT-#6M`a4|@b4?(Rr(rVG5d~Hp6{mLUW}9LA8Kx}#K43x<
z93{InPqwbhw1Gc*#zKE-gZ{FfT~Gqo_~v)<hnv-ps-|^NgxPdB^@c&<u&C2v`bsl3
zYa^h<mg77GRGlbI2EBpKF!|IHT=NifiI%=TBIxaggfP7o`pj=1H}sjmB4p?@4<an|
znHQM|eP-5ElHjWHIcg%G^JXSY=X3td-m}WZv+ol59I!$@(^umeyaiv6`VUBPV<2Tp
zb`SiN_s9z*7M>;{Y!W24$zR+v=S;v{c!6ncBCX9|Y|6P<ymfvYm6!b&<MymCKIpNe
zuWLcXvhyk4r{(wcd~~PTZ|k||I9`y|<R9I8><Pcm^yueEH0^0&Vw-<7Jn9$kPsxDZ
zWWw8LkX-i;H5Wfxp4Rb&ft11{M50k3(daKu&S?}#oCzaQ9YLaxK%(RO&ij4IkL~~x
z$$J`zM5BLn-?2xEZ%@cbKicXaGe7sJ33;KTe%riRXYgwT?+`)E<wv@K3G@py0r(mu
z6F~1ho>blgm?T;Xynb8K<VH^SqhU3zk1MIEc;$kHP}cUFEd@2jZ!L&hQwv3{PtT-g
zQPxtt^0BykFPWX#b#_wM*(qITr*@s)zw7L@uCuLOXJ>SsJ-F-aVO?iucb%Qnb@s@v
zv-7&nws)PacAc$tojtMZ?1HYd3%kyq+I4nO*V)s$&Ys?N_KdExXLX%j+I9BauCvQ8
zo!x87395jtg0r^?74RnT_qjvQkfXnc%Q*eE;(Lw;|BjI0=*D_(EI&6m16iqH<YhV|
zN0ezvm1kjL)yT&c<VXvv3y0*v#Uv*OSXZ2USuFqd)yIQ>=e1+rt_=<8QM@}wUj^^Z
zb8DN-f^m~S{6FT`so>XT*wPfktI0)Uc=dJU)tXweJDOJudtf_$O?zXW>M0G-WQ}%?
zl8jnZ7|p10s3Vq9qo-JsHpl3zE|&@ZoUGd$E7v8_=185$JjY<st+T;p;rhQjQm7LZ
z?_Xd^>irQ?$*hO`H?yG?y}Cd=NBxaYFzK~?f=O>I70<p0<#WJ!`Aol}3`}|=J`+Dr
zM85K`nV))CGVOA*opP|9Jije*>X|^|6mV|y)Qe!@XQEhmz&zywIQYdV4t`1IeG5z(
z-<5L<UM=3AR=mGFBgx()UcMghE1p!IHnp+cTLa6d9Kzn5uLG^WI3$dqhXTKT^t3sU
zyysZ)9PsX=El{$Cl0j2Sh-T2C!0t!4n*;Cf!QQ|5S%>^<?jC!ncw66$^lw`IYv<>F
zZ3_H4bSN-y){&z|;0_TuyWoM-VArW(cCAXCZ7|t&DNL&0zq=*y{d<FcTd%t@fqUKC
zDAFzM;)(6z(Ai?(Q`*I$v&H21Zx>H%7q_;HLuZRAKe$~SI$KPBcDp!qwwU~p?cx|O
zG5Pj(ap-I@`C7X;bheoMf_8D}Y%%#$+r^98#izB4LuZRAKcih7I$KPBX}dUdHa-80
zn2h!w*A!1$5O?=6{@&Y!T+EhU_nf4z7N+QH?@RpJ<UcrXz@fOn!ut*dFm{eZRy~Gs
zF?VP)sr*7{YsHfui@T>a_?V%%EsLS=pwYFq2#uHPXD>=Zj!oQD9^BbjMkA3gg-QZS
zZ&9B45deh72CuPPNL%ah<Ry^Ir_x2-b+PlGwmfX@ZKmo0orV>7eufG<&_OI_`}QZO
zp20SUCp+U-&&ZGLq=x`i<7Ze?dCfh~2P*Ft^uFlYndfcok>@>X&WmHvoL8~V0`+UB
zSc4-->*Yb(E)NpdHHcd2Gh@v(O{so>`WyARz_n`bE~Sd+sj_N=6Yy}534|A)ps%%C
zZ!+4A)h5$AVG!HUFqT|^eI<HLt~)t%&y=rex9$Y;ZxlD%I;=ibpcZK#I}9>H`&i}r
za`FQ#6LA%4{|r!an8{4P!jroku?F=H(m=Bc)8EzuLk=Sl^r9yHd=Gd;bKp~`URB@*
zf#Ny&8R`C(z(<8`*d=6X`yPm;Trm<?wF$u2E&~X+1aSh;e;L5<OTf||hXNO8UMxQ4
z%t*h`0{^Pbe039F-6ZJryH^7cboAOV9sLHwC~zJmv?@?ja2&J7+<GRk;ofF6xxB~H
zfVZF}99tF>TOJd8F(&qMOl(C=?6sKKs+id8F|jveVsFR9R>#CXiivq+V*Z#|O-$_5
zm>AaNqFTE>Cblakwl^kL9}_zi6KjZx9f^q@kBOa*iM7PUE=0vf_6SEd+;hTzP>8Z1
zb`WI`&g*+qT)=k>a;v0s4`Q!E-~gyx5d-b+yO#8>;C4C-Z0n;e&W7nib8U3EcEWS1
zr>_&zj#^?~q?)$0)!Sfac9yhw%<_QBiw@H|+Ce<oPhm5K<~r9f^9kHgeiFi!T2hOX
zP|sjjz<As#W_Q6uSc|1R4PmEitAFK!ndayGug)J}o{W3zvC+KJr(mPt?1TBPt+vYA
z`21Yod~3d|PWXde`M3a{rvsZKFy#mHan*$9A_g-O^IdE6Gq*W$&3SX3!?eaxd9X+6
z0PJ5`i51tU$p^e~{#WPqHBX3hwD}xu8_T|Sz)zC0h$pTbo`UgxRFC!ywrT?2?m84H
zU>|ofP9`=)C!W?5k3=Un>4}^{`uptTTJ*%z(TV5u#Fps93wq*(=)_i$Xm&-?%;V5&
zK6u~8vcB*?YVzW%1RsZyzsYxvQeA<7!L)R<pV{m1m~RnZW#lv15f7Ue+|D>$yNpXB
zgZ_POZKCwUos_OX>8H9X{mmi0^qn0`V|S`2Wf{f_ZkrOrWh916Q%4NyzSZj((XkG5
z9x%XQ!>GCh{?p(dQ{aDHRB4-MI*fl_82=i47ZVUtp1FmjUTkE*7CYa81l(^l`c{wm
zI=ya;(`%Hc`T@ws+bY>n4ZCq57#D6`mylKaz_P5mSpKcNTbgvy)tu*T1HU~6e#;>E
z?RxUtX0^?HiV&@rg=o7BM0Qe_064D!R*St9)ej3+%S&jl)pE}&)wd!p^o}arJNZ*I
z%!8{Qc{%`Ff_yC&u*iqK1;$Li#HYD(uQ`szc@VNgac=GK)tfI<g|$^y(FRV_p^68+
zZ*msu<m3`a{4+5@AaSts8tN`zLt3V+VSQ&cJnVhsG6-Zfnqg!YLt-6HyukT`E?q!=
z!B&_SV=j@zVz1W+&vn=Ldaa%XBI@Z|iwE|4eLf$%xNcD4r?mul0Pxpgz-J}k3JK^V
zU@ie~lz`L2fDcH(|B`@zBjC*hOqPH*hXF@P!0{6Bc>?AWumQb=SN978rbxha3HT%d
zZzZ5O!st$D{I$`T&`i;o3wWS04-)V;0*bSX?gW1raE%1qCIRyZcn<(W*oB_(N*L;8
z3H5ggbv2=;5o#Qu=7ymjlTeE#)M?ne-+zcuRzTerhMFj$?vqfP2~`LvTw%NJzpz#A
ztvRi2(enMmBda-PT-zbLXJbp=;kgNM4M)mw_W=k+)ui6cTfg3J)c}R>S8mnBYRd=U
zZx{!@1|KSg5UMRad&?bT$12r#C9=di6-A1SB1NAhA^1Taej2vUZf(Oc;wvg!5=;A5
zw)88z3WMUrLijtqQgu(Y7KMI-_*uq#An<PlRL`i3;xYW1MUX+yCnanuPKbNc)NtLS
zt!-I**2nY3>iRL<fCb<u%z?9$y@9jIW0E46Vh3r%mB3W3Y%wd<CD`N$KuaW_D@J()
zI{BCZbQ6Fy&%=okAl!0Ekk7&(KVUts0Phs6;JaN@u`J^mlSDPA(B{s^M!))a(AJ7h
zdINX*HU!Xs4t?=8R-xG!=)cf#bU`EI?2CUPP`tI-@`OKdeoDey_lcf4h(vs~1auI3
zBxM$Q1YB;$Pzy~#g{sSBtDH|+rresQRDBJukX5T;iELsl{pQ;WRd*^?wH#QZT;n_=
zPmNx0=FzdZaqI1;EDS{V#My=#j1i~nBJXxK1#WR$PuVayOai9iBQ$KB3&k9jZRWB`
zB2mSuTGs(Mf@^q&qstV~a_h@}HqKLiDkZw&V8cwA%}-@hmaiY}wx7cKbW5+teh|D8
zq?hs#<g{?Ive<-F)zkanbaX`tE~GNmwAJTVwV}Nr1bjiNd;wyZ?odV>62q@B@fAT{
zA&4FculI@4bUq(dS`2tmdNZ~W0)GR+;6eb36100pl>l1$`(6&0=qBDb!6?BCh((Ff
z=;Ty(GJ89Xfv8Tw`d%74?haIA^!Kz=sZkw%D?U6sJdGW$R39<QhdVv^A(DV(>Fn{5
z4q=mceKc-gMg995&VV-P4J#XmO_Sy*8~_H0Z4n4IWV~m_sZ@Rm7^7M&vorCdRQ(aP
zR(hM^e{jx2!O1=G+N`!%z&W4iyYDySlP!M`^iz4m3df{}8-5AO;Be(zh~-f6W{JK0
zWK-muF=+T=ar`and*PC2e(>%Xrv%Tz1;-kCZK9Zjxd?OoFY3$v%Jo3c=Dr)Y;ICAy
z)ixZ*SmkSRi+Je=bYS14x!@WHP@x|f_HMKQ{)AB(#EZ2_j@)%~mJ!Par~7^j#QXlX
z_2&L45a(EL@85o^7$5IN(L7ETo7*fL2J$42XRbbFT#pb96c<w_(tfwxpby1@Bj9#a
zBBoA@=!IVTdMogwvjrj@yq%&C=5wAqPJ6gM=Mto6+^-^}=SL*NLqTC40<#DW7Nm#0
zn4!9Csg?6Aab>qUTt|ZOVw&~6n?R&TKSaRqJM|+$j;_61kfQ>mfgC-C2gp(N!z4#B
zRh7Q88g3&%D(%*ZS+&B#@Wxb7^|P7M`-8L}5QPlQJxdZK_2Jz)=;eWdvl$Y|@0*uw
zADDt(ABeix>jT^OI_UKE_^@FDmG8C=jG8Y$K=aYt17&Ydj2dn1<wndV(aZH^!yTUT
zfia9E*sxh<bN(yKy*9TtFsZt>tY7HC@Odj7T#%`ycQ{#|D*+R6$w5EtEDmKquq=9B
z9hT#JUb{-MQ#=4QuF`LI<nEZ$XO*dJ1KekSEBmJkC*1`_ufY^1h8fW{o^2S+$^@(X
zZvwdBpsRe~FbTt+A{(<v7>iO}DPa5#Q=`*##F#$%5K=4gp>j0bL=WF^i>7h^GzN5r
zx;{TG4Hdcb6F=k=_t=-PWTEEHBGZF~t^#ev9Sl8T@n3eUyL6}pl&0YnMmacMUq@vi
zL(6R}`$n`wR2jNfUA+rq_j*)^%WNYVr9H>Erw+wXxEn;OUiCGAlMMvjM1+38PVvIX
zE7Z(w4%1!|GuJm+bq%2zaE;@-pLh;KeQ|Cl=69`9bsk~OeMhSL@Y$UFtUan{z{+V9
z;f9|p!O90*1V~OE%8liZxNs+{&e42Ab>ESudhAJUYV#NBpfxIVIak>*N9Eca9N<){
zcskm@X#Q~f)hRGG<-%k1+0(wj*;zG5i%(v(<iRD-*#+LQ?((ZM#+M|1h)T7dHEd5Q
zwz+uKmmJM!*_Iy$XM#@?fiORU?Eu?c@P6!Ksh)(~3AkaRHBte@F7%uM6{H^yJ9|fT
zj#B*<K3lL6KamzXs}Ctv4$NrCT-g-|P4o`xtM875riwNgGZI~}vvyZaC#v?SN`8%s
z)6*-^I2Ha3SK-NjOQgdg;rIqo&S;$OnTZ2chy~6)X|_TIXn1ZhCtASaGFu@XWmoMf
zD_7n5!w6jjTZ6R-LY0&b@kPNYybWo~F?K0?zD#-p$M5ddHOPWDG&z7l3FV9#i#aUz
z#mo_ayC{BP8%!mkRabtFZC!q{Netu4oVc=Ug5&W9P$RrKHcP3ZyBi9h8LE2>oH5NG
zo|h76eG(YDCnmXPL8Oy-0y{S8$=?7$+<^{Q)jbPG06yM=H%jR?etK)$x@xX+(#LCd
zTv-|s09Nt>13^^Xj#M??hAt4-Ug!vRQx^RLg+7$M(4=~1pliUFGeTEmy^{B-NRo~l
z$8yx%wMu0LpdCm{_5TS2ZkQ$!9_=m|rjB-(4r9NnuAq+cr_b>#tQYX(kPAb?zavPG
z|7B~NZQbH1te@=^>v8e+EcifK%nlAdh%tuOXWN$KCrgq*ip2g=9)$giPI91@d{|{@
zOixSIY9=1|$RofRNDmI8zrcuthfpQJmxjUrDZ%q3c%lRkIRP_03}cos{RxA+1scA2
zNK_j9FQ5j%o5O%M33v)B5y1WpFz%L0Yq+0pM<^qNBbNyfAAm#7t#0jVeAMB^vd|}@
z5ZkplTy6-;g>ZFM!rn@wd6g1+nS906FlaY`;!`x-<zhpRH7r6yvl@B)WTAgG!ciwd
zP~XyUeNPMVzVGkxfPy(4)xLkLLuhevsJkI_KQT|38iv11!apqGa|o|W_$~Mx4IctN
zik{&17o4QT>ur;5ciQf-72K}6&a3VxlSIZEWK=$p7^hVI3n4mPOm(=Qv^0E!aRM>O
zJth$nkW%>^GBx)TNfSKQcW-r<9|nbOD@|zd0&0xgemDifU|A|N(wMPsf@j9zl!iZX
ztU)uXZ0T9{uW;sHCwR&ar`jrSwv@eObB{HHMf}?Kkwv_{F?0p*>xQAO(ht#VlTtYo
z-`C-uU>WBgIfzy7Z_q@oW6hzRm?uvpHH<_YfmG(&++!@LZ@BFfSzlX7L4E64-?|(}
z?q_qLYu%cPy0QrUbweD=qDW{DD`+_R0KeN;;m;J@kA!96ZnL|eO#Be&v|)GMf*Tz!
zaFAan0a1%MLKLRo4lIztI^1v?bpw|^Xb*rgoVmFFBouFVg`^GgZreS!dvPKN!*Fbp
ziu(jWhu!v?)V1gmZom)Mhx{}oM8`AQ#dD(L?y-sO%NzA$rq@`Q*{=TR{4}*>1Nea%
zuog=@{8*m(SR^qoL0$zuT08vv<D-PzqrA%oUHW~O&42GQ<u8*j`ZEe2;2o|ox<8Hf
zUk)G9{`T#0j}=`SU0*7{=fkyN1$N7@_VB<bI={pBSWULDHuAmyx8qq}fl(OW_VM=b
zYoGtW9lzZB#KGwW57=B_$dkSKS+&uwjJUsNWqFeIY2jY#=``V9>ghZ_%AYQxf2sa1
z;a<v>CfrN8^5|a5?t;G+9~Pl~vHqo`W2uhL4X`0s-U9!kS_0ncg<~*>sm0+s>wq)-
z(m}EF34;voAm~48Ar^q?8LZI(&tS73GyI7#A3)-E12`vGNZ&`=LNf;^?>B(|0p9W}
zFF1hv0p9Y9oFtjUL)99uIA9J>RTuD%1NbH2Ex!PRPf7SCSg4;HUXS~I&n0&UM_U`#
zS7d)ae(C=F)BG!};~;5VG1%e9r78OuoHj0rj~Nc2aW}3|5lQG$CW%MsxM9rMByt=G
zJ6HRSKK^^O53jEJ7<;-QP2R<WlO6gwzI3GK={s|Gce$M?GC%Cfd;LfA61l|RF=+EI
zoING+cY%|4)dtQN`t(om1wKK%^r`|Do5@-XeIJDqW%(rx%JNHC9#L&{H&&g(EX{`S
zLsL?gu-?G8cWn^)a6!j>rU@9tADWUFDc?P+Ji^iC1%8!YGw%D8_s0j$-{tM_z0`$A
zD1?b^Ts;)}2k|u|EIh})n$rVY3T)UZ%nKD-G|yuC<?(%Mt~`KE-zjxi6HoZZ1Ss4W
zXZBu)C#2o0`c^xUZ~=}FRsa^W37vZNuI@7LQ6sMcd1BE&9w`hAY1Q=dcSx85YmumK
zpb^az(VLBEnuzAeXo^+Derjaai|CJys8>X-Ms$UU4ltq>{^I#p4ey621EcyPP<-HG
z##ISd7@^+DMyRi1ga-6BLh0W#LPL5Qq2WD@(5M84idi4qDnh+YMyPKank~`>Tr@)I
z=Z(;ivqormvk@9~1|iXa(|TaQDLpX!BmxgC8>_tcvgJ{T5wL7La^Ant<}bHYn&k6&
zj4;OkJ6=BD$7o`FV}g9{!w6#h7)}k(zX4Y$-%>MHdG}?@99Cz`fnyfiyhk<`-!?z<
zIr9hp%B-|<OSQ>gX~Xlk-Td*kG|lqY_&{LJ@dST-79t;-`7E$(N(j6&=L8Zn5&5dS
ze4gmR{pEFmA$#@)(pzzCsr(mv{I-!e*o4;>|8#!ltNowx`-aS}GY5vW?!eabtM=eB
zb!5$E)+dC=k&%&8>afLb8itE4`_8@LqA9Ng`rr8#3)RU&tS{vi(V9}xnt6|GlXyI3
zlFt{>0^;{S@$&f*kY{|ej(5F|_sIy}I(o6#mnY4zVgp(BonMEs>HiltA&_$C;izVa
z7XGDOGt$0$Gs<F{u~=`$OL{Z@rZ?kby&0eB&DgIu<J4uF@w%A9|5;0{*s}LszX#8v
zaA4`*V~O67-{}o`U2n)IdPBD84LP7U<n(15GQ(Q`pS5C|)%)F>K~*uT87{pUzt@}b
zcfA>K#3_c=R=pVq^=6#;4$WBcpEjf7yElVQxucr#wBC%D^=AA-Z-!TIMxEY_LwYls
zF58UPt%Cgj-%k*>;mh!x!VFKm)C`{&+XlDZhCk?SctdZ4PjACEy$xUMZD{Vg4aDPc
z7!Q4Jo4cIb#aY=IckT+8NqMC>%Z`N3>C@w}*e}(we?`asO&$9+I`-Rj?7z{mZ|NF)
zfy(|cDiVEx&i*cllPL8?xRl<2FFJ3)6R{2O=neRz-hj9C2CUT^utRUaVZ8xoze58K
zhEa)Xz`@HkAeujjChzUM0p+m`ct&r)3cUgU)EnT}8?aMvK!e_ZbKju>?U{o>=j+Qg
zAbR+V5?^-Sfcdcvcvf$~O1%O9(i;%a8}PZ_fS}%h^WUKX?Pq{Mrz`$eoZNm6Bwu<C
zEQoEuGQ9zR(i`x$-hg#_19s^R_*QSgh08R6sC1YEM5oI+z)~@DfF&+52cC>=z_0ZN
zys9_g9lZhT^#<(L8xYbPaPd1dpu-#>I$h2IHXvpWutcYGV5$WyBg~SY)3JX|$NpU%
z`x+hlFLdmW=-9V@2kbjc0OHc+1R(Y?6M!W;od8oU3!||AjgI}Fb?o2Mv9Hy!-=kyS
zsAJ#u9kA~(`iV<dqd$-xGjsy>&Iis^%c3aEm+P4SMaO)#j`;>1^SwIeM+N37mI&hQ
zB<k!tV+M=8vtbeuwGGL74Upa(CbZ^f0{%OB1oSFzL=HE9QvBALjO`nX-@2HweXV1m
zj|<US?gscr!p+)T=&SlCQlKt!|LSBl6H7z5;LoFFVaaBA+y<c;7ABm(N$P>Un-GY#
zcPQ`O2xTPewr+vOJ-Y&5x9$#HfS!O&;nq(Ad#NlWPgxOYq?XXwx)phKNUcNaBx-?K
zxAlrHW)$#tRv;N${@_~({_o&S-vygG)nZ}VAVOh#R`B=lwnf{&X*H^BgU_Yq0QVik
zo_Ak2)dcrx<~FRPdlLHBb{875KTl}NefP@efYW$#{|U7E#2cx^TWJJvwE7p`HO0Dw
z4LpP;uo*XY3)J#F?!h~9j+z2H_5=evE_@qkotXs(+Lnv3gIF5jK-=6H*l{L=)+2KF
zYHTO*VJkWuRpIE^AN>c#g|o*C;Z}2ALXEt7_FTW3z&V&14C_s#ebM%SChGp}>;X4N
z*#o#e!}o1`<$-#9+~;Tyc;#kk4|o?(X#Z%&RefiViZThn9?+{sw+Gy3WK^IYfEK2o
z8mViy<7Jj%2Ds9Q=7}0}jA)vOX33~-2Kcd&$pLUxU#k)Iis%3%x<W+z8Bt*d7@{C5
z>;ZlBz^L8`NPEEdj8N~MMyPKOBQzkv2&Kmxp&=$CG`tP%619xF$dIrHoHs(f&l;h=
z%|>Xz86zb6M3fkE!blr_%m|HYL`bych#nXaLLh3q3VXnP<VLUug!~I9`CqqCb%aS^
zl7EK91(N`ue}GAV&$nO_;4_dQ3YO>vPx%)*{BK*R<LCu<!&cz%ms;GYfX_eb6*TrB
z#gJ+LLS@Wbp8ynZxER=atWhTH(F?=SP(1&Jbtti=xDHl@!5;)_k2MuPH~&Y^WiJxS
z`u#cr=V@m6$U@0pw>WsZNz22^c=>$CET6Rr@(E*xuuQ<tP&_~DGnge#Bm{=<@l)g8
z;xE2q)lNa$XiZ4@B=9kf7QW%Lx0>5aR8O%0e45>{YBOD?Sc?MIJ3kLo#l#n2gSc}W
z3)My}4T&!Z<atVm<a{H|AWJP2>_O3K8F>a2ozFjkqVxGKC_0}TK+*a95)_?JX>6zk
z-rJ?6K;TV-M;;9vVKh=+(UAu-9T|GN7NmXm7W^`{1(fXd7CfuB;8nc^@98aoElFTb
zGY?xJ?F=1T@S3nL{HKlRZ~g9Vm>$~(3io;&mg#MHO>e_$y$zr0ZJ^1AZIC90j%}D`
zt^ZFOFxBe)?k(T~dDJ+fbg#GI*Ln;7thWFbJ%RBiy#=)Pum#e(&{+#s{O2vG`0g!u
zD7FO@@AVcur?=oQdJEu2Szt`F8!@Muhfir(=-2|e%jQ38L`sK=EbNROCo<0fM16oK
zBk2D|M;|6tfu=NhY8IN?h&IhTd`g=_d-P$!=yVEqlxm}<u(<=)pLb4vW<+DHSt|4%
zU#_FPN=KQ7PNGP|8lN=f@F@)m?NAnKUbH1c(w*Y0k)7EyB3O%Zkk53}ZsH4~#MaIc
z|5YsF3w6X{z7~l6RY#nrO(IHD8lSY^@F@)logp5ry+tA3QHN`X__p?lUxMJ%D{Xk3
zXqH7f;xFik)3i(kY0Bi2#xp)?y5UnA4mv|TT5F3!yem>KdVEKF#4j<xxep==aV!`J
zglN?#Vl*c6N#iA-w3zWplMSEJV9*)j(F$7>;$01Jfp|2B6+`6`V|-RD;#E50H0BdA
z8jty;xsp#B%lM?JhEHiIXpcCRxlYD-^5w=j%f*awmbk<i7fUpTY@l%+MIc1;JrSe1
zm`~a$`J|DIPnu}>lm>#%5btD+6OAs%I1!H-<1BHBF+MxC$GLn$glMkklLlfwX`bYh
z#xXu=n&DI020BB$lQB*-x*X#~JZ6lu#3jbKT*cC*G0nmS6Cy+dJ)g7+^GUNLpEQc`
zNs|np(k##!;+>3fqS56TC*m<<oFy(X#*GClAx3HzE|3r*n&kPUL6}dPBl)B;j89r)
zveFWzRiGW-v_FJN15D{7I~yNT9F9o8{yU*!>I-m*{|J0^XUJbN|EkSW>^%uNI{P2O
z8n8iF1J=MAumXymum(J4SOZ=~3Y0|`{(jre8gTM`-5St16dH`Ney;2P<IK>GM|TUH
z*|P<@L@L(*{sa@iQ&4&9c<ab^YWyn#XCS{L33})Kt3E}NgF0Z=xI`9<0rhqO?FscC
zXG5g=|8#rxKcXK8&;LC}{};OdZ_q00{=UQAp!omWxJN@60DgdGK{f8Xc9>9&2b{+<
zN<WL$|Fs*TPCcG97zThz@7w2@17A$B&J%V3bF3ZUTj>Ah8S}8AgNA}A<$n_t{}d|z
ze+gN{Ru=s}4P2bwP>6f9<|S;ESE!%sw>5AQSE)yt0cej98v(8|QLz{Pqp{7^(|a-;
zs8hYa`6s%a=C&cynlT=Y0Sbi!a*@D5`be2Gq(C-z_zcn9NdKX|J^1ZLE`(ZEIJTR=
zcvOB3Il~Dvv>fPJqw?X)a(>pm-HYd>XQXdz2@G%j)bC55y$zlwhh(JhZh>t}C_}Y@
z>?xlG%99hjlUr}lV_Nq-*0IT`Y#l3`+((o(D&wMZ@tjc^>04W{Fe@}dqcW{~8f)_g
zKAM8{@8t6CMd;74JG1JWK>UPFC}I!##al7sA#4xgIcUQkgfW3}u~;>=WVLa-#DAe5
zIVPUP6E>ojchS9HKSs0Gl=K_)4%~)Wfc;0|@sVxVhsI6G7WSbgjEnLcy9eH<JD9ck
zIW4BZyhrReb}PQEEW_gsj0&v@j4EAQeEY{4>0YIHP5^!r4#9Gi>mNPJQsZBfZ^3nC
z!_94hz?qAf8*sSoT)kem9QES>s{SsPBkBLg(r&z5&jw+<DEg@AeeI8Bg9oBfV!ZT}
z3F$+S0LUTvqP6xowikl}10;K4yP;x^8Y+=tVhN=0IaoYrct-mE77n9rW`4iD{^I;q
zUt)0d$1uXEscq3Ex_oN!$@(Dql+^^p07=3=ggv5~-J%+OfXH%dS=&04v&+%pi^MrR
z!tQgbZ#ITUWV{FW#M%o5z2JUXIW{by893r#elmt;9DZBUj0tiq4EQCV`yQirpD9DG
zM%OVg{3_Xpqy9QLieAd`V!|PPv8f331mQ#sn5_407XpvZ)NaJp5t)zD&M`a(L}A_|
z1vi?DZ~Ik-2U{^h+X6#Mx1%d|z^iLU`py=A@ep&*!0<Dl1#X;T)BA8VeAD{Z4Cj4)
zz6{ZP43;JT_tZw7KN0li8TJSLg{&*$1xUVn_+>HUP!Em<?fsY7k`<nBXdmw2I52;Z
zUFn8l>*Y|vhPTsjTYDk-mK8DwH=AGz>uFw@_WkIzdYN{0R9bhn@@Rz^|G2C3B~J2)
zKm8x-<GXTQH~KR{;najHGPngi0DmS~2nl~C7iSLGhi?*kB#OS%6_pRvq!E8jq$-$I
zmPTGHX=#c>9G(&2#ze&7Sm)v|8u8Z@*zxdDJe%=%D6r{a5cm>&`}==}`TOK`F^)!D
z=(zwc$br=j-bT{!TizoNztiAjzo5Vy&<bi`pt{!=up5D?dVs2z56&Ei-Rp(h?t~gR
zr^=ys-pBi)1pM>{ZoH?KF5PN{OE>A&dpw*c*s_knv+LZUKDzuFD1^b;_tXSa^r0W7
ze?!=3#}X>rOl5EY6Z8R5o2vz0N?UF3TM@xEte(EDZ;654`x~4jm*e3U<-MuB_KIC?
zzhk&3!r0aJJN9`UJzY%g^(7zHUi_WW6Zc2nZ#l*A$ey%`gNQoKdje<9U}=C01;1Lg
zOH>KNI_89|lHn<4Q6;2_yDfg0&@G%?W<LsIAYoXc;99lP+YP%6?T%b8PWqyOxbo0c
zIvHmPF>5(QVIv(0hvEr8M>BXmOX6xIm~cPy6kI_Cglhw>GYh(P!|zm6H&kgkKx`qg
z+-<H4G@j{uj3Y;gFlDRPU?AXw1iCiVOL}{2cszoBNdGz*kqgsN8^+^GKEYikp;LoA
zNEE$(VT{K^?Z@OHF$Rs1DMzIkf5w=+_=~o{9j*R=v-LB-ZB(ld0|@7mFdkaHP#tkH
z$#2U=DueI_DpKcS7Jh(5nsdqbTnOxYcpt{{;lS4=d(jm?+k>$@5Mvo6gjY_kDV>L5
zYvnkeA;xj37{|OTVsF7Nj%D1meD5v{Xgy%p18{J5)h<(D^}T!GcRpclyx&&@u*dtY
zg~LJknEv3N-F!W0*DzcoYK+^aR_ePPw?`0?lf4Q(6({ANAK_e-q(^BWaM3=D23Tbu
zkCM;6XMU(Nw9|SZ{d(jClkkD$c;+x@()|}{aO6s#L!SySMvz9{?%>6!jn`Z>NEiqx
zfYGKD^mq+(7^^j(PN6FzQ64;PBib~$uom&{#dDk)>DyWYuNOAqt603>qc1z%RW}`T
zgVWC+)W)n@k)>bMfvoXxtwLW$7#4%PVTVRdz~X0tk6Yn_cK#q<s5htyv8F(vwF%(4
z&89$YYjfc2tk$DO;0_UJ0qLrnaugl@$f2lDt5oeqf8e<N4ERIZQb4~i<}+skb@u`i
zRBI`HBq3DZW0^?vB4*?*7kNhP#W3W{9Uxc4K)x1+T-5>c^%%%E!;o)xfLt8|`B50s
z+X2!a16dP>{Imn)mKezGVaQz_Aos>V)`uYvb%1P$fjkn1l>HdppQmFWTf&eR+SLiq
ziTo(+__}+7{GfRM<R-X4I!d~K9gaT-zP;y6Z~z0*`@d~>-aZkBS9l0$^;;vx7N-vW
z_X5>*(tqH56Wp3KB&rX3C#69L#H~n~z7Mhr7hJ@7!6<%-JEGRYREb;cn@T5WxlN_x
zc(>X*Re9s2>N`5*!6uf@Jg$OCZ^yBvKaV$#O}z>!2=c<)cz`0$n1W+Vqizz%oUR&<
zV@u{N9`S2Zsy1PgYcBZ`hx^?MMENfG4gIR&X3%(~Es0M1u}tH_b{OtTnf79ITD(kK
z5uJ7n#0B-OicZ^uG+X6@iE(APE_}h<xYDuiE3n@>DK2jQ0$ar=TL2qZb_G_)XW7sa
zS2K>8t-^tI_Z4c%8qHO@*5YuFpNmtJ=%u*0Hm@sQ#TlVr*zQ^DzTK?X^FX+s6x7p(
zm1bE_5$n0#EGj|=V2?M-kHJB7SpmB9PN&hG1rTJ=o%iE`?p%vH-hWVb=Y|{5o$vfa
zoEp0>3~!b2!zKJfgfEitzaxA#;DtOj5l5qxDky~*0>|CsC$`mAo*ktu{?LpBPu~}=
z$6HGhpH@v<+xDyOH_=4Ettku!SU``vmx&ML@)ZU@8QW>{FMru-a?%r>CZEMgx=!jY
zhE&=~^2Xmsl8M>*KgY&@fv;^86ZwCOjc@#WY<$mKvGLG*vGL!-8GUqp4|`+dKdy<y
zqk87_wn#G3DEK@!zWrDv9#w1zj&2%ni%M?7Dd|p<pTs4coh0Yn&}p)7M5oCQjqZ@_
zUc%{yqcDZR*WriVcm><>@`Tv<k;2&c;3={3sZsH$29*`Zg6)_Q8$YwOUHmVx@jt&F
ziAPoY$s3&}AKo8HW+SdY9vlDf)22ALT2Ohmn^IYZBgi;wTU~2Yp7$Cp*I!}u3r2o6
zT$Oa1tGO#kmo)d16?mvFv$dgRShu*swZ;qV;sv$h1@Ol&=UB`O^eRNz!Eo?Zh*Niw
zso3km$eY=iFK}IDbp8i8M6Kdw(P=NZy8)xTa9UKCJ-oDATyqbw5s-y<h;ek$^@VNw
zjZ8sp4LqZ&kHD19Rb>yU5-j?SuJwhswt87t=|mkj_hd5;e`<ImK&L^j2aNu{QaG=W
zbVfGaJ;7|pSv<{M2v}8zJt*rd`+1~?qFQzzT4r>DQCDPOpL&tGA|6I};xru|ufpqq
z8txbbT5S^x_=be{#v0`L9mx2gm+SBVc^+E@a=Rl}aJwJl4!VYZKN945aTvZ@!oMKl
z$KM9{GdBwO^FsjtO{RdK9EQI`!apG4H;n-NixPe#;op_;*N5TLCH&7M{2vJas)YYN
z;hzG$l!v4I!7e<)JA75wHq||TnBCn6z7syVHX&|;yZ>g$1au$-pM1NO>et1n7?1Og
zw<JLhup`;iHyiTygc9@9;0{_(c%J(<tu&F26(AY-N)tm*(8&SAWsjm*9*T`mrOZ%*
z&@E|d?k#FrA&%V-bAQh<!o0am)7)cnkToXBS&|~_tP^$iV4WvJ39RlU^uV%fQ7R9P
zX2!rVx3?^XWl-H1wXCNt7hWL-4S`qkYB;Jb#Zi5D5lB_4j)S$qrL9S+zWrv-%O~K&
z1LytWJV13QZ>-0W;WX8KJ3ORlR-EKtZWZHPd1C{ruF=qB&3$Jo+5n(C0c5wPK^&sL
z1_YB(!Euew&o!7$@SKH0YxwfH2(07&yXro8sWF9i0TnuM%EQr7cxQ6YMW3p9bjcvZ
z8+V?nx~k-T!2rNpXARTbMLLjX#k3&={VkN}(;;^%9J9mU0UT-6SlBxFdKu%fo7!Ho
zHMh9dC~s^vZL#Op*c2zOTdA_H7M~JT+u>4%Wr@vn4L(iKOyM_6?h8?>kD@j>4@9pW
zLdcnWPO095fI6ZS9(}8~Fj$G(<KvWP;wd^HoQ7~S6t@w}#cLGLw}{|Nl>@_)lPa#6
z7H=={Wp0H)sy5dVp?V@@LYFNU*IZZbk7w6R$EZRF?&MAAD*&NE<9EGnrEG@eR(6=E
zt9duSS`sV))yTq7uCi9(IW8*hMI63_n?TY<&xh`g-QUPpq+uF3Am$=?9N@etndk3_
z7y7e^m-?oP_-DL9mBv=wv_fdAneb7i;nLYzO4T_Kb`2=7I-(+S@!Z8afxak+XN!pI
zb)x?Z*?RFpjG(z$bs;rI9$#*)VFBDEgExwCVzOg=0kes!iJrksRkBLcQu!*}CWGPd
zLr`>9EDEFZMd#syYIa;)=>z=Sg>*|XH){|5E2V0<K@;Oro;r%yPaW=$S%Zcxl^nC6
zk23m=hTBmmTnDn%wo!|pW}WUZ9aAx)lFBymQ;WA{?&e+U4JaPZS75-GO1JrW$|5V!
zqmPz)<T^4xRXy{JFHc<qSJ{@*hsWF2708$0yxgh+svD#d3@SJvt1@OP#se<QL7pip
z@6B+zkAyQN)FEcTDw)1nb-yfnOT)*NAB3)|uYn_uPuM`!^qFe<%He)TMB?8TzEW^)
zo$C6wVM;ojC+J}DdhIc=e76oeBpQ}j#sIz;0W`+~s?mVx3cb27t>hCV%y@AHN~`fN
zTfKNXD*YD}VgC+p058|vC1EO;2O{N<rg9)5pjhY-oy6S<P=mxhmSIdHr)vu-TN+BQ
zf%jYv;Cj9g(-MDM)ZDE~bun0s&Vu`5$&TzK>|Wt+s-?V^6=UWW&|*6pPfmdEMuZUt
z9#vv=kbq(QJq%*^1fa(N2(F3{0E8fRGliHydZkodEof+y&13J=_EPg%t)vDR26<Uj
zq0JPu<({xB&du)c(UZ2tnY)V^q;cja;pSo#h3lZ|39KUb8>RXd1f016rFs+s9II?0
z$k;H>9YiZ+bP%HUTue2^(~rYed1IQ@;((`M(9J>{$=O!58NWJx#cfI;P#jb?ot}ZO
za8!c`l3KYwLl{25<aBj3@1~#iJT;1FiNR~QB}TOzv>;-2AY!4Sj`WL!MW<g4faJ6}
z&L@eKPQN5Q1N~aA6D>haBvk|bl7v{P`Z<(i&MZl)swCONH<(LGBI((3Ik#OO=_E@!
zHRJFQ@5FNjv8@v`p1q$MnsjM;<`sBKDzB3?of(3vZ4sFo-VUBpfr!J^s8sdh#|aZQ
zd;xx`UC50qE#~#u%MjwMUV$h4LV@rk#k^lIR_Tby)U3~NOfPzId~%RZ&oEORxK9fs
zE|2t#FUkBuG=?wZr;XQu%c7yg1(akpQsB2r^*k=97{0d;-}p0fCr~kdbma|P+b&4h
z;}H@@PN!$2p=^YHfuG2Xj|?IQ7FlHwvaKZEbJ<qWLztx&2zOP$fgfL|U!Q|yJ6vbf
z2nkz)Y|R9pcn2C@NFxaolwFON?+_GTCwHaRSvugaq5;9<W1*(&P-A1E)L1AGFY(er
zfYMw?*`1(YZ(H9aK6pDvH=OYq^y>@IFY=m*tXY#vSrb+X%9?Or@-3wu6cUcS95`Mt
zV!^g1SuhP<Cxz5!MZstxRjKNM$)XFYAW1SF{2Q)Dy>&$PI;Hw!FmVhnjwcm2B2Yl|
z^c{<O;OA>mGy*s2^_&RiBBXi}7F`KlV8mcJvHl6p@oTt0nsgWY?NL}Q8A<Hbp64X?
z%dm@|9?zPCPm2&7g0UUdBhdtS-W3hVP;(`}SE|!R%XIFqys?%!$AO_z{XM<S5a=vu
zz-FcT>={mv<D!bZ2{F>NQZ;9gZ2ZaKBO)YZbaq4aUvV)Idwm$NhTtlRv{L<?NY+WV
z@`jYJm`y^XRL?=S-Hip*gsOh%q7j3!P;)QR^w#5}#q`-|{q-k;qx9FK*?!O@cDGW^
z3;XC7j-O9ebHM|YYA$0DuZbdc7*mUs>MNNFnbXousjj|~BEMIu8avn!SzjC~vnifV
zgSTr#JS}9H(-EFrd$QJ4-6a{PW~xR5Up?CI<y`XuB}!bJ+F&x2T_EK*X_=b=ahmjw
zL(pkj=gfpK0f`^UO{xi&SA<>bGIv5pP@BO+r7)zKYRJqAE5DI)j>82{nX)^bP&8QI
z3cj)VOI9#3!VTCJ9<u{~V_8840?#rq3{vj(Dyn!9#VFxsH3qN4)e1sruQ~{ZM*rZ|
z$9@{=iYYCTu5cn8melj~j3o3|Hp3-#l4?N>6bg+_i%azGU|;anOYmw(<~lIM6EKaE
z)Cmz$5Ldl>Ut}*iT$>%0-?XyM=_K7^c2WKSJe0}A9u8BjYC^a3i-5DR&sEb|xYy5W
zBiGvlzITJvU*T`x0cVPohL?&tfTEP)TlgA?;^<b$zmUdfG`GQL<QCv2=ZY4uBafGW
zok<4maMd_npJJgK@l~jSUyBG+@+l1EKpvzsPUgT7zta_jt9>Sr#lXR$)70Q}ZDJWw
zc7#T``xj&hco#`CecE|?#d9D@xmp9#0#<oLvN<gt?2R=DVyF)YdLj0*xNAQPu(v28
z<b-hHJ9V`jG@qIt$ZxHaZAOxYkGTYZO4W}*!8qWPpu2GsOn%m(CQwt8Q1D9s?R4!%
zpKNlt>hPd1OVb9{m4g^-jMq^SG7i>P69;LY^v{2T_0@&*9x*TFuvOmJ>@e*{Ik?CX
zrHYWZz?=+cEW(N36!d_jZLKz9s#R609qbQ$E_@Y`nI}1M+t#ZiB-7!ChDd2|$@20m
z*98XvngULEzXANAq5+})27iN3<Vt@xEgtjL2@$S}vRNlhBIa;^JY;~t{P85a2ZOM4
z)E`e4sH;%o^2#O}h?o7$K)P&L7YK*05#m?`UyRg5(HPwyP$#%yJv=cNxGjXUa`tK$
zxar+0a8s(g#dPdR$ZBC6)4IShxB-v?r=T|+dK+Ap`_0sN&=h<&4zwD642*?BPN)h*
zsGxcAod4tCY1}9z;3FUJ1LW}IofDsqpML3PKxvsR;ZFS4K(KT@fI@p7SZ2(GP==h(
zQFwP!-=~+UkK>#~{z>j{L^uuMIS7X-+XH$ADcfBPmteGutBp>AvO#8FF15cBX|xb7
z1esXDre%H(nhupu5+qVEY89Yh`kbUnNeVXJlL7%C1rE7Et}u{m+162d`yd?(;hTOR
zCB@Rf@aj>&gELQZQcy0{wOhrE0p<E9u72P<L@^QyvU(=)`;cigh$qOH22ATkL`vl(
zWV8t2MI=$et`iy_GoPqy719YMtOX=dG$G8z#Thh~6%qlEFcsrWgJMaV>dZZ_czzy3
z!obBa?FIYtinqXsz=nwGbb3W}NsdCdqA#N7GjU^vaO*5J8>b1Q?=?XIou&&$Np-|h
zYXtxW*;1+sK-F-;iHWrFO(vrsR^h2sZ@_O>t*vQ|pm5Vm|3F;$3h>qSr1~`kC)sqm
zpA+Hu<ziGjO>3O48Z-s>9<0S^m};G<nbta8yO6Ywrjb;qQ5yFPlRR2RW9lY5TrGfr
z90IRpiK(KKG2orK+mz~oVupeRLGD+oO^ET9C{UI(rMlc8eHgM~x`*Duc)|XhLztPI
zUs<)GYoq%M0N51O&0&I1vh9xC%}N#T1(saE{W~_t>G}-lz<c`|JOn{M=X8CAFh(ov
zI?i|@<~vR29B`>ib6gQx(f@}JjB)k9QlF?vWhyxvv4;0(9pkc0F*iHIHgpvRx70T{
z)W0WNA=VdTI@+}n-L?<zO2vC8@`6JbywJY!rNGP){|DlaRIXOD|BFDjQ@TBHiz{UW
zF~c`Y65V<Zons)VR0k0YBRTMrNCy3Ylb8kHj|86p;fvxg9OwKR80W^cnLi}B6LhBo
zeTw>M_+p$GOaB2lXXbewH?<7(PJHfRq<6KKp?96s_s7fB$E(pWTcJbg)|fV7hdeGu
zxGvJb{D3*ufF#^;4MV;LB@kBIS%#Qhb=y(W7g>Q$C@F!ou!K%s=D)zuDx{zlFh@rC
z8x=QxHNJ!M5cAluOOk~wL6QTnh4#+U;@v_l8ppX1|GJoX)S*`(7uh;7z>}4g-sf~}
zraiKp#=;ra=?^S(xZmUJph8N4(lZiL_<@EU$<QTDU&Hw5lzRHavK>UJ0<dc4Zgvsa
zK_Tpps7hIM3*H_|6Z}gDY7Zgpk*^CP=E&TndL~IZ9ph41_F-h6)Xj*%STo6FL@Wr@
z&<MZ=lm5~|6P90e{8iIt)%0ad6TXiojQ5nL!{jKMsK%GVQcfV<vZ*)q+ax$YH+>TY
zm8Vw)o2TrJ^#F;=Phvhs3S@L#b8(<y6|%|Ki%<>z^G<XXUk0ALp>tDS4Z@BKLA$vb
zN@!WBp@8NI6|4+vVut!9l(M3-+u&{wR}&;cbrr6--~T~e-0G~j9uNjztx*>3y$<Of
zGtL<F9(ZkckltxN3pLW7Ez;vRbw9>mQkyXj*9M0(PM6-?gdi4b9%2m`B=zV~G19zv
zD$l$W!Wcqw1!9Okji_#xds5`+{2yVhMBIV_)1Or0y;Lpc(qxBW{US<Btp#QYX)j_E
zVR`JJxTr#Ff#k~7o<c|iTL=l2#SE2YOGIVC5Ro=0Y%wQ2ughu_l&4O2KOQw6<a}u$
z4$ydDr`m#`YQh3Xiw2fgkC;;R2JVic$Uhx~Q_f`y=DU!`DYF7kHbj)flT*jwhjfcA
z*&kClbwYX3<B@M%!t#*gAOeUnizll925g4Ng?wItB&A9tx4|X#b9se=o#limpJ5QM
zEof&}qcitFdAc+6fX+|y%8sgS0d>R-t6I7l0HLqK{DPxfyn(_K*y{?sg&${Dn4kcF
zSW!$X3=M`32=t@-4-`)r(p#V*Y9rnjpHrzyfqXz6G)IVmP<^DsV=q*nGzbRlN>rFr
z`fqH-!GvWE&>?1_fAKxVrD{DiTqD7pB!{a(q=R9#Xf6tY7io?+p?f};*Om#mB60s(
z513xpp#4fUPx%`ipj3}R%;*HAy2upix__f}l$$?C+eGgLuLrEIF9?0d8yXPduXPS=
zgVlg6Ry@}!RX+x-)3p=A%^8R3OQ~K6tMV~{6hxTMG}twHEgd%8=1P3g987=zTHF_x
ztK(aiZ{S+?V*}5!>x|Fa0mo23$q#`u`8^Sr{65@2VN^g`@je(aoF*uWke@#nsN{6$
zx}S%;?y8usvqW`W85q1AaUD<!9u!=is04RKg0vq-#)}4FZyEX<Ux*{jiy-C08a(I!
zG<cMJK4yFwZ|g9=m8CwxoFii#%w8Yk8#u-xr*)uMhrfxTSi2C_Db^;DBPo`T;SyLJ
zGzjNu8iaA(=)wuF6D`ge5G{Q6QwGt(m4Gm*lwu@94G^?IEHOy>gq0>}B2ChM0)?b5
zbffrQ@cK>Rp#jx}4cDeq?2;C~!E{`6t%EwiiB*pTrOFEAG}kGh^s+4FGB};Fk|`}$
zBsc6dK`FCf1~cRY%sQ$~Xcr7)f*3{5XIkp{G2q8rU(J;CO>>>sK#Z8Bg^AJ47#RGZ
z>=#Zs&rL!xfrX!KAuQ0uny_)7fnF6U2?0bOLC={UvWMxBL=6;WIxzs9z#<fcB?tI5
zO1#J_O&g(`$%S%YvII#=)%_qh;DRP<D+FTWW=VSsoUXFDwkEGoNK#8z3awF4lws6I
zR<Tm|D*|0epCr9P?+~g8mfDQTvkvAfeD8oWelcNTYDDXxgt!y03acPvT#3UB`jaIl
zkr3na1PV&^$_UwkDdzVg2IC9-<iC#^NrY%*6C5kpj41?pivr^*NPR!l<zG@wv4}!7
zL^Z7i*K*`;RI0cKQ**y7Fq0aH>e>MWaaE6Iy6AL$4%9D-B@-=PEJ+|WtbvSmXLrwc
zW&cE?UUs7{K<5!&zAbEjQOmM)5kDi!Kp=4r4T3$vC=p7bd>5flxIP&C>OV#?-<>@$
zf}&D=3f<EQnm^GAhFUh{l4u5fC>y5lb-{lc#Hu#<-!d3{97IcgiQvr3(=SY<`RPH@
zqh+1E6ygUb$I_Jsy^=$M%88()YANUyloQcE;?oLajMV44sPDzg)F&;^t9eZnEzb!9
zFu)*$(?0Xne$qin?Y1Etw$H57GiaarfMHCXRO60~5?l`gTMhRri^Viuyf2ZvKR8bz
z;exL@GB;z+K$sIIXxcK}3&rfJY_%xA5zDj|Tqem2!q(Z+u(|1sTKRPgE^^5zU*kGe
zc@UC5<O9k+tHh*u<1-So6$oq8ps)z2HsgYsl;@EGVWj}KQ`7r$?_o`5anM7>6IRl<
z7Kj71QwGdBZN@lkv@Zc|4qeAJSAh+BU*|w7gt_CaOPZ>E^=LR~xPVgLST6HG=T?aK
zE!U+p$-Pw9hAeawbOkZTS>-Z-g?=jMt0v_BC?QYJCl->LZP2)4_JAOzuUjj`D|D%k
zYmxCE>96BrC6E$@Xklgz?4QukX!<SgL*l?f7v`w2W}_b&Ld%MjDmrAyuL3iGT>?9?
z#Jjc%DmI<4>UdA$DqL0sp$GkW0Dr;alRGl<ju)TYmyw6NWY6}4pAuZPbB@OzWbi!W
zI5ly$AFf8n?@9d5&utQ5O)#ydB-7K<xq#0<GS}o^Y4eY~svF^OXu{{u&WiW1bojGP
zIL_<OPEHV!tnQ5D3a86WJYgNcWy8WB(#W^{3(J#7zK;vkEP;L74+eH@4{5m<mFhc5
zYF$A!SFj%ATi9_mDwh-+FsVa54S5N-^wn~E0NR_xIgl4uI!bjlL#jk)LpvSs>Awn7
zU#Absdsh;F>vH@q_@Se1g9>4;a;;g*-BxzcnYq=OxdT^_!`XQyUcydK(lTqEo_X<@
z0j*BxsCe#lc;-<(ZAEx9Ws01LJTwM9^gW)H!W&jTo@`xZL&Ati!n9Rpf?N@uVqpqk
z1X?TSEts|{5xS#UgC2>qm%a;S{}Vg|qD9rM>KvA!xX0q>S8?`-cT{}2j|3pBx_ySI
zE(r0WUV}8OW;lidKq%r^Pm?rYgJA`A;O1Nmv()g+H`F)|J@-puB6$+uB#dB%LPX-=
zMUSi*=h3>yv!*Q^kCm-y?P!G0mR4(DB5IOT&fz-ks613J35Lye##Z??eCW+85}TAL
z@5~+N$;O5Nus%ogK3tf!0~M1p^zMVpqw(x}9iBMGGiAUX*p0|Q`jKNYFyvM}Wq2I|
z<i&L)8EiKvmA)znqe}yk1>LO2Q>l7pFS)NY1#@=MA`5WnX`0+kw}jxBv23HB^d^E1
z(=H-B%UEyZJ`D(rp<F}XFy!Shy@;Y(_z=@+$N&}<s0&6}K}00sa$uC}<N~;0X-}GV
zutdeL00B*3Sd#G~E_}_RiVGrxY9T-yYh$tYPq19*DYT6YqkR(lIq$+taU)YsT<PsX
zUc*G%$d(tWCQOzu#(T`m9JyPR%AZo&5tGieoB4U;;3SXRmMXdJ1+uYhd2+1$&vft`
zIbYK_Fek>9j?i3d@S2^f>Ew9N$Yt1$h@Ciceh1DzawDhe&5pJj6*g8}wTK->WerZq
z>Y3O$)q>fITe*h=X=0nEWuAawblhQTlI1ziZsJTFx$IU?3jDG?o;0#M<K@fD%o~}6
zuP5U;I2;pCY!^>z7f)#yPvs0AndHX|E??GPTvQI*S*5Q<seTOT1%C~SvJ3_|d_MLz
zx?Cm78R)lPJ52|0HOIx7=WyK3Cpu22SAqQP8v~=ZZ!4btaYo**;@N?Wy!ztVf6vI<
zQat;ejJz7;9Vwpud`4b_-)6x9COWYA{`}0}WMAjErA|H+j#~$!roF|p|BPDvwz$c{
z`@Mi1*+-lm3_~XNL!CSHcdRUr!ws>lbP*~w5A@s8CQA?Yd`bVIsI*bOzIbJ6+~nFY
zXyQ=6EopLNxYoh_0&|n6J^dzrOP^kWU+w8F_$`0hi{GhFA1~ftZb`O3-HO<Zr;|hs
z8&WELX@S7@Rv@&_k^3oj&7upPu5Yy5X<SFiyNF)>2Y&S3%YInfwo{v`2XT&UaTtp<
zQoYzM4p~T*+m^2v`9jQv=n=Au`{TrRk<+_ao7^rUt#wLuJ<=kzNsAp8Nsua}G9R;D
zUv${57rfyK?1#s2r?@tZKnSNHY(+RMu3f2TP+a>S!!g7fn+C$-njqCHKqb`n;#!1g
zN5r-Fb&3tT#9i1_$D#zHbE-$XNvK-8SYc&rQag(6wWZbAm(YYKL_3HN6zvR34Y`g~
z(Ba9xIYOfo!!&vWuE8daJ|Le1?m9v`jFj|Q2$5J1x%pcPbbF`+7KPFTnHAKTdL>Bn
zabhJ!xu%`$l#gT*=rZ?#LyFb)bJIDts)FE15{;2bUL<kTlp%tGBo^`_P>^OqE)ylN
zxi&PEOc(nsbRmr@tC9V-w5yqRQoqCKyF(D?mxT}sa}et$2f*wW+W;%XEAg2MIL+mX
zm<9@pcpi<7f?Mj^ExCnYD6W(8-a&pYUT2pFz1^?EqyP)ybZrEd%wp&>8$c5#Ob(EH
zAnf&Un$C{*WEVQIkX3mf$gqaxXB7*8!lnm+d!4y8O6AyS0*&2!QT9L3XQ%6gqj@W~
zAn4Tj7N=_sXzEUfsSeb+P;;#XbvF1J#G4zjPDk?WFyWq<_1#FGgMvJ-<+>-tBeD0r
z$RP;zVDy^9wcBZs>79Z-MC2#bchdu-T_X!`^^EK;J`mQd%-ykOWiRX|;`xX}62|5v
zk!bELQI>26DFk{;yTo3d6;Bk2KxJeOa49eB0}0UrlLZdTgP3#wMUeM<^7}ZkhlExG
zv9p7+yu%cNjDg{O8h__O&e!0t7R*TMe}Za*c1vo$3Dn$gJG-Esjl@~=d5<*s$DHj~
z51$0{a}SyPreyzM6Vd@8!`)=I<O4DZjt#=u@Osg{bV5{icRd@`T1`Eeu`Bk)yTpt4
zM89}f)QgdgqWw+cm0SF#K?sXiZb1ek3B<cjAgpJLSIX=ZywddelLdDs@nO9a95N}e
zY5N&a_=}+Mfxz(Xo3z{=N;Qp*TtCVUiSM9PHA6EF-VZ__CGj4qlTpqyP|JOY>Jo3A
z$dS}_9m7#XRHGS?1O^iCR)`#!kKn+Lk`s>1CVa<}hV74x8IrC`(#^-)Y_2VE?SQkW
z2e5$7((V~tw7}d(CEv0_ELv!H;OpZ`p#IMjC)LwGSv7@JPs(jVa{5*^VcARV&NrrF
zn(@+COxv{3Snf0>=@7O<eK`W+oE%NHoI$f_Bm)r<L;^$vZ-=C~DPl$Zaw|ENtcH8F
z1$ig5*jkPXxURtpA^zSLJ>NB~8A|d8x+I9MVZ?<{$!%kJJ0!o2v<ARl53Ah_bPQS&
z7E*7*bkM1qSQ0}%Ed>*)*cVX(g=`8jG<2=Bhj4wSfZKn!swP}S3z=_+ntK9UCs|Ep
zTl|~>Ft1d5xj(?7R6h-rLmQ*$6APzeL&!tewS=XMWf0wx?CJ3BFmHf2js|%9ZRc>Q
z8BFs(j840Noj^z%6rBcVA56P0l4iaQ+;5H9Y1-?^{k-fohiOL`lPaz^;9GB3P2YkZ
zhUatOToTk@@$y$;{hyAH7|={GVs9ur6jVf;!b)21mY5;?BSiHfdku4h`0BBG&}8s0
zVW(Lp#vj6kYRPfj{7t3PU}BDvvyRMG5a!dyptWVfnsOYQA_=mBwSe%$(!xrjw#tjM
zJ{)gCg~hlaqtv&_ce*hWdralVNX!;kUa>LqS_Y<zU6FhE?NdY9)TOqtQn_T_z#xV<
zj?W0Q?_9Z9u7>U2gFoZ2(aWH=xyHI0_lg$6k_i*qO%Fg9IK?v<a)?5u(bvPhsGg>6
zIY(gH)>pmsO)W|leQdJeD&hgfm~Dg=S(e}woK-qa=qMb{oXons9oew1dPZK&)!hvz
zFXG#vjhgA8qYW!A$Fy7wZ@C`MZiLF=aDB;_u%lt@LY2xQj+amx?59#M>jYvDAh_Fc
z1xkq}@khmy_%M8y<5b$PB>pIu#6<*4;*ZkiAR=N(oXuZ?qHIi2csv^W(M_9qq(yH+
z4>SQ{!4Hj0u7oq2;san(+026{E^-+>3M5YlNi#sa0Ex<n=G4ly?hLpVU=1rpHjoj$
zCD+0=@M)}tpOS0g_oBr?9NiScF9qT2H_0#s;U6#@MXJ@3b0q&s*m8<?i@pANYRTJj
z5d_o-`xM|~;RNPn4Z8sRxfMj5vxr-TVTEL%vc6s{PI6EcQ8a_;W|-zFECP?lmD~3`
z)E#FSLO*9S=5jv_js5&&$JwzMz8N^E#0pdrc8Z`N!W>H2|N3cIZr9apK$9a3oBtLC
zr=++Ae-&IrbFbvMaC!!hbecZJt>!<EroVi%g$rGHDF->XWq{6Z68obZrcb%Z3t_Ve
zJx9<y%>^M)h}o>XNGS6$^-20V9kdG@nm}1g{~)kQo6|R|Ey&p*1g`M@38yFf_gEbX
z+sFKC=HFnyV+$YJ9raY-==w=m>rxhrC>-vIsQFn5bexZ=cfipPV0TSEBr>hfis<pl
zwIZ7MN58K;x!{EreAr)ri<|P~X)o0Ck@x~gZW=5KWmVAoHS9l6QmTH&=HhcCshQ`1
zVY%i>d@)PNTM*w;76`iiTd3d3=J#n(&plUmST&swwqa<AjS`yjsuxqRdtBVD9&-f_
z37m=JMhW{Gi2F~X`?hF3aZ>}YM;}EY8HY=Gr8su526GU19p_#R>{*N0*A{HxKn>g1
zW+JMS?r8{OyzNuPmI%-RS|3Vk9j0|Kp1}tQ$D!bWoY!1}3a>B9!5%LT{Bgb)fu7>N
zW9lR*lZQ>+S>9!gPc(J_sip?*J1KJT#_9mMY%0e~`|~#-XvX;)&4pPGdBxVYodyT2
zTJFPsFh|>_vYmYWMC_*`XHVs}BQFMrila&uc$O_@HU_4b!z_;m^#(n=9#5sZfKz#T
zExN(AMJUWROj8W|S#ib;d1f4PnZ(I`6T$*jhbKP?9f6C}2QNlvZ06j_)2ICs&J&zv
z)x-%dAa0xv*do><oUT*O%xz$JSS)<mS^-#Cd4vq+%-o6n<ERcTrKZAtn_HM)`E`Bf
z7H~i>BFNBp(c_iqVtqBjW6m{rB<I|8@+3jkl{da|c+h3^ayfkX10)nQ?G=Sak*~0N
z5{S7Re?auOh*M!TVQ=GWlBk>f)D<H%Bc~>cW=I%*_(D>rx${>^vY|ZlXN*C}X`{q`
zD!i*g)U8W@q)+-u6>wbgKk>$C6shn!^ng-T!s&xPSi)SL36gLOW8+warcKdEv>++u
zh>@@O**)2**xO`xox>r+Hk$qf>EVD8=itaTIBM`_yp+dtRi)}SF=xd-A901d1Fits
zI!rf|o70c$ws>3cRlGl<@2Q@|VUXj(^a%EV8|;(l1oZ}OQoV3HN9$7^Zcj*gN;nZT
zqUL8L&~ZNcPa^3d&5`8i8Q~k$h$g-gxR6}%JV;cs{rNQfPJ2ENzuNOflJ@K<TcWvw
z0zF(b4SfT2L^IX1|3Pk4+WE{c1kNR6CpqK~+=N}$5ZneBeSW5L{DM>LVqbF%JFwHm
zwbrih?k2$vw@+*t6LT_2bSz5diB;eAwzf~p_RE=I>=6A^@Fgle?u1vmcgJYUfw!BN
z51FA^^!+;?aEx$&G8_>Ur(w8w8@}Iov8~}^16g#!BVqU%5`Mmf_nkN6cJzS){=gBy
zZ<6pgh2cj^c&CK_HR12NM!^4*@PCl-eZugG5`KV$-;o4-Zj<nf2>&Yy|7C)K|6ZJ&
z0sf760RKM|{#J?qZG;~s;r|hae_g_VEaA%t|39++YYE?5!Y>HJmrHn$gwH4ZKH1)r
zA;9lDEc$1B7+#a`cT4!bgx@afUqkr6OZb6d`2G@ph=l)I0sOyY{m&78KHy`%cZ@#>
z9{aSr`(dATA7$}<12|B2uKsjBy@I~FMRjGv7Y3OH1!|>oWdbS8ElG|Mw<KYC{A$kH
zds3Yv?nzauzZS6tY0eP~u=DwI#8mfV8IBQ;VOMl5V<T)R^KU|pm;(5BL}(sh|0Z%D
z1>8!8$Zeud7!$$vCydQjxJAHL$iAj}bKJNQ<~pVNI&4T3LY`Xr_%Jg$1m0#tqUL@y
zN6US5n6kJ;#B6y^?5SG(lz_LXhKNsxgNs)parLKdD1^o<m1_{S=O$zOF-wAGSHxkr
z;*DT6mi~fxpSR)Mx>h0`hm01-lWomsU2BlM#k3{&xZ+%wyGHRy(E(lQ9-qVREJ7fU
zfkFg;0gm<G>PdVKQ)zz-4ivf*1q*_SPTLHoAehkvOlX*wRPfHqb<zig$NUz_Sejb$
zEv{EaD5DzsHibrzZYbASV!;6j3%f-qZWHltv8Fo_?}kh$JxfTU>w>-2EKLMxreGu<
zb3ddlG#F9k;cG~~DQJ>UdHKu~w$0TDYMyCt72(jYzP~`<@Is3>sIrtIsceBetYAUY
zW=X{Jcr--t3%RW-gDaO1dV#-E6-S{Q+w$Zd($cFCrbyI=ZE50Pn0~n$$nBfaZ&qQP
zk*kL#yRd}^QV^u{2D-h*QJ|i?RbMbob;);@_F#4ArK6EkyrKGA73;gSoE7olCRukO
z^aup7A|$*6eC~*sW@`*Dh&eKh>Q=E)+y|56fTIwxmf~>jL*RA6(vT&`4gxRh29vei
zvb2ol@O2=J3I4>+W(V2vde<osV^woYp9Qwc^J(B$UJh}Lg0F&y0jBGRs_Uo|M^8QW
zA*jqadUzH~I1sHisJZa<BFoIVPQ&I#({Tz^WC@}^*BPfP5PS*Ih`xq>!XER_kgj3l
zkH_Al2!~AY>G+|^f8a0DzGMY{wIyEs7A&d9Z~rBboD1zsAx@VTYD*zb&nPTd3UPW`
zVgIERsznz305c16H?!a?cq!LY)ZEQw|3fnYq**Us0##-;n2cecw7IqlGlCCGbed_m
zxS}0=CwuV(k9{}<t_G(kr2>Z?gb5h?e}pK$2nLrz^HW?#Gd~3h0awG9eS`MMwYe#%
zUc>2|OrO(ZulmSo!p6!*G_@K_2KFBPvzns$BkU>mHxGg17RpC59wMzlrtyKa29b6|
zrtyKaBO>j%OydJ-^t)((^R!Ik18Jv48r&fB9eg0IMWkJjX?!5<0@BQHGD-X<o8JUE
zu+Uf~_9p}v!YY|xU8htHG}FFv4x3^<I4f&&cjsx@2_WCXzsp~dMmjENtt948^(Cn^
zi1}O*Gl=;^A~rt5IbuAB^i;$+Xf2YWJDDjc`WO)+Mb8x>2Y`Qqkl_mFC*oi+Mtfxr
z@Op<hRL&jNO?8u;bDg((tC6NuHzDZ$FSGke3xwqJ`IyI=+cpd?AadaXq74d->A3FX
z#!=aVBf(Fl#>IiIdp7XAFC=^mB>Ybp231hqlR*27MUzP&f4Us%o}@Bdfgwlo3K~0E
z5Xrd+01CP_0sz5E2PY904Myh(fg1s+y|95H09yjJ7X6@+z75I!Z_Y5aWV0mAn8%od
zcGpgv1$Mw-OQ}&1Z-*xj)>qPJ5N#av9V=o2i9FRaTExKU2B>#ISp>ycfm#idQ3(5F
zeJWPS1(gsvy7@_|TH8Y6Sz{2-UI<I#*~3*MsU?kfCqW-SXI3Uy{Ti+37t#S!EM0WE
z&VZ7^1OWX`5V=CF<bcz45Q)N?%DvSeSu22;q?)pCovwNSNUZ<?Fe89r7XrdKJ0`x6
zS#^!HvKL@aku-FU1#>v~6{_OG5%&wE!YDhzGx9nu9?Sm@(|V_AAGbR}vA_-@eLWsq
zRmyGWFf$TAWN_>`IN4A95EjrxI8ufk6poZD=k?X{{||NN0v=U$_5F+li8eZkjT&3j
z(MFrJXrt0f3T+b%I)i5vE!ChDBT%KSZ$E^jq9O&7DC0OZUaDxVMQbgsTE#on)^G^~
z5CufVdsNiKpaLQU<nsOgYo9a81Z->H_j$he%kyN;KKrb*&p!L?z3yx6?O6t?HRMC?
z_#DU`4sur)<gU+w+$G3C6#?L&(>@1If?pGE0t><oaS0_2q{aNc2pqa@Q4cTl`q0$m
zA&5p;&A`#H`sg$fa>d2>Dkze*Ef17FScQmQy^DWWaOq&_mvfUUmM^+<LRD@u<NeVC
z=bnZhm}uR_wa>LL0{z93B!O`$tXRqCz%BhSpME6ID2#&!Jk6c8P9L!oT*A4{il~r_
zD%y_`G(lWjUv<=@t3M79dWvpV`DPH-f=IP0FR_~%qo-u6fS#IGz?5SFkz|HAEn+P6
zoW8gtxgI8V<U7{TCZmgq_t7J8qRMg9U?^yoZ&n5i@pns%&=#zML((k7m%HQlH2;td
z|9}<W8`Bw0tff!raND~02I<@wr6c%!T0x(*%Y(fK79F3bAq=qSXgdqA=x{p=u;{6r
zb;F`3>!M-N<E$hs`ZX&FiylTPz@qJ&_0#~1?mmcSdSKD7e%Y@JH<VNSx3MVkJ>!8F
z7OmGqd@MT29{B$k7R?n--4l!M^P?|@MXyr$@c*Y+bm_m1MZ4k8Mbjo!dDm>tz#V(y
zP6vgrW={;8fj1r0jy<8e<4ghNUb81mH+=cy0ACLOm-uq(7sZ!`DyL>VJ_A**X@3(`
zS?OrzCqjM=$K`5?pA(5CM@6Kiby%|IAk)}Jw|)jt&B5(`fHF1YR(IFdFX^sRi{r^D
zXyls$jKp0{MwfNhA?>Sxs!TK)fwqS=QPB-c%vCD`V{l4;veH>#iQieN0xWSOr2tDT
z+#oC=ZMDSVaKrlts390Fa0TuODD1-xy$I$b$7b>}u~@5Mu~6fYo)GD7DszW2EReK`
zeg?wvaZqh(4AcV*f}Ki0png;-s!~-H21y0jp}}?yA$G7`Du;C|g>~D7#yQkdS=K~;
z!lqc=u54L>jRK~*o4^6t$w%ULX3!t9zCd<Jub6uOV;oN%hxZyq_nOIvjNwW`DwBV6
zMN}4Ea(YnCsUTJXAJSRsUt!(bN0P8c%wLn<Rs4?=ieSO2RhF;=b7ti>fuv5ga;wUS
zDu~w+>iP=yLF)YUkOR2K>ed=uXTa83-AaRld+8Wx3JRZnqnG{=AneT+4EpL>x}Wn#
zJNJj6e!a!<ws`xoV@F%0=0`qTN1S&PwA|azEC@BjMc1A6fXpdmxM>RJ+Yf{J7E8R;
zZSqph4Vt`(=R3}Hvwc%V_cm+qOY78ERdjE%_WoePR&;Mt)hOY&ev){;i<OWrBd=!v
zfS2fim>V;S`Ju_U!*UO&B?;j0)X!4DrU^ITxJLQ8Dl&4a21G1G6##Ms5Q8zUqHZUv
zaZpjX?#CfigXYGNQ&IA9EU`Yebq$GIVs-ELsj0$jxUbA-ss0bh_7I6C2Ne<L7Y)6H
ziv(5pP2=Pff-~__{Wyo8IDu_^vFHCi%cQO+C5L7I0}XquY<KOu(a=tr5K3cpJMj4Y
zItVJx154H`4;zjNsqDWh;8F7<+N&>5B3)SSy|P?rdc;HX<DpftRKcI4IR0yLoUp~@
zYvjX+)-`Xhs+`L&QXjw6AYW3$FEu7>K7@2h4ZqZwr1=oCB{lp~V{+z0NS4&_ON~jH
z4<T1l!!I=^V?G2&Obx%(n1uNd71?t+ck*SdY;(9l(F5&6z=o2PB8(vH_!jfNkdZO8
z#DWlC!XOeOIDXn^Q+b{6^oc^*zz>k8Tj(W)RF>@eJD0Lz({>3&Zf#nj;psyGh0_mt
zy<F~%INu=Iw<w+}`m)#SZZq}7La*`MePtMG<sKf6haUbG(<bLt{(iO&-tU`q82<ZK
z9s2&B->S;nxr9#Qb}pflxSdPrByQ(Ye&ropLML$tm(WSv!6kGOci>tT@ygz;y(gO3
z*`<?cn8VArpMr!fBsAI&KbD+%j>?8?_l|{p+0e&6E<31ja1Ja!&v1x`@}E?0;>{#+
z7NdzMz-Qe@MRhyySr@_#g>PCASun#*82OcAPZpF?hm2m06XfEGx^<-Ia%ikmt@)^H
zgS9M+Tk2F8B<D$vN7w=9FLdsZ6FFkJ7;*Sr^L>}~G72IHHiJ{A*<GzPBdFu*eW-7L
zX+bsuTfaxSkk#>hp|*&jHa~p%-^OkS5&~`P)_)hf5th~syPX|iH;~?cjNQ7oSo>EW
zxf#j>8=Igc>=xNOc5|sl%HtqI;jDjw-N0y<h?JTEM`6kM2{<el4Xs3lH`O_}@?kEK
zC5zQ1zb1s}zD42sL-x_4^7zO&ONKT|UzQs5RjCoR%woiShTc3g)fcUc%OdK9md8S`
z4o&s_uADcXFqn_M&}TZ+SC32Peol+A?B$B^4GTeg9I$V{9KK<Gw5+q7lOgq=g&WrJ
z;EJcYBda{J&3Cb+Lk%;Sm%+N$Y*bUkR;KzMp*$R#ld;fp3kEVrb7?R&mAlEeIN~kG
zsrz}3uW?-G9K7*5-(WtAo!x!UM^E^jd85mx+;;+h>+i*Dn%N3IQjWF~V_e`AzH;=P
zmt>v0n!2^4?;M+DKIE;V?;M$BM!UArci~N*JHQ;%DSzX}f>OKvuB*OzW8WO}Pp7)y
zjYWA>uN-~iRma^psBlq5F`fx0>duHkPq^yn8w=RM2^V#%?pr+ir2mQ(z7-mM(vKsB
z9r!kWyYM-j6Z=+;eR}qyQ;sQ=r(!|X$fw!hr+=0AG_IKastTT7RDpBCle2Xg`6QRX
zSx<6la(=;+?3R;uNRO6n3pcC)k;X7rlf5)Y7&Kh}GjLW%FCwlaeQ$lq9982D>uN6X
z60ggG9e#jFcaf$eqM>$_4D+O9AR2&JAM6rP!&S@=DsOk34)w)GDo=3nmQbfb$wl)A
zR}F+>ko0gjWbG4t<7?kTf=|x`Insrvze(0bp~@T&<3Z>RFZ9OHRPH|lTfU=+r>FY!
zFdxUKe$@*}oG_=;_MxAP<Eb-49G~TXmqk!jf{ajoOul*ZqB{py4H_VY@8s(_QC+9i
zDb}AlPE@zU>U0xzoDg4ebMpfK&Oz7X0h;ZF+Pt#WwSQ_qf!;D~>33*<(&$<5yV`rR
z$YlGRy~J#Xb2(r;eDzIilgAuow#h?QVtWfcW465sY54IDewu^-p5T8GG5GhM1Af(V
zgCCrRFL&@~Iru9D|FMIAOz?L)_=D2$c@F+42Y<TY=Q;SR1%Ii7f4eQ{f9qoF|3(h<
zf4SgKcKts`@D&dJp)~x14t~Cae?ssB9Q?6@&v)?C)9_Or{Phn04#5v}@LQe*{w*>W
zV;ep@4L{7mU+Ca37yKXxzgX~d9sJR0_=tl)!NCs~yycl}YP(tR(;WP!_k;J{&}{D;
za`4|0{106F7YY6>2frW<-{jy|abSE77ks&c|Bm2~a_~2#;Ts+NoeutjH-N8j@cRmW
z<5GLy(P{W<2Y;!9FCk`Z-me|}iz|Si@8C~N!<Rbv3J1S!Bk)%__<IC@y@TH;4WH%U
z^BsKN>%d>>=EE-qf1!h4{a!FWD;L@Lyv2d{{q!~9Cp-941%HBrzbg$t+riIu@Q17Y
zT@L;b!G|3Dgf#qk2S3fh=LkN__5VGVO53UyH~-V{<qrNV2mgus6LRp63I0w8e^44e
z&%qz%;8TKc@#nwbFLm&5zZ<-7D=ZIo<c%D7-${aB=HSl}e1(I5C=LIhgP-r<#|VD8
zgFjaA`3`=18h)ySzuv*ei1(Y<*Y$r3OQr3tC2szw;fFc+3myDnf`8AozgX~dfp5&P
z-<x=wc)l`uL`CY9?RA~|NixkbNh@3@S7k5xi#%`OPtjbiQ1YJSsfE~Wkp7?M3~g@&
zeF9p9xpsDHu{Thfgg*!pF*y>gyHYtN7wWw3%0h}yxMGCj{jOL{@h*zK>Y-sl2;gap
z)UiyvJssR7Tg~*JB1&M;By@B$?r=jIUL|ju(>>GE7(K0{IqkvV4LpmU*4o)G=;=@C
z>BV+7ik^<3rz1G?J2}M_)yavjs7{V^MRoGW6a)Q;rR;>mLO+5+B;4>;9_?^W(72te
zKhZIP{pkagBd?>t7%1Bl8U)|`v=$x4La9p+;Dv?|0y{HXx+;;kXmTiA8TJbb<;EFO
zh^l}kNe&gs%egMbE_pfECCw%;=em;A_%8m=pcG9$={_-1R%K#^Q(p{2aIR}TGMebn
zR~jEJS%|HuS}c7dqaICFSb2uZ=<RtZ?o$<}f|zP`Xgj1DN)4tCH)nHLlLIF`!G_&_
zSg&9(oR`IK|6yJhuY?<hni|RaUqMG@R+H9;*vo4!uuGQ`leQ^*&2*g%yfhlVW)7vm
zrDAQkZYiggWru|uHVA^LTCA>=)(u3za{%8*$^zTy;@liG?+i3NbtS%eY;cjM4KB8u
zbA|12VHza4SNs~ej0CYV`!num*tny2axNC-fr|x#J#M9`-cw%cP5P}|G&t84X*6O<
zZv)T5GdG&NtT2|kO5PRQ$QJw$Z_F!hxW^A^_)udsEIjk_J{p=jeUj$wd0EY!C2q`z
zBtEK4wDC>i<Qn2TXzXJpOOTTao0@mzJF|W?HS!<T8t1}}LEZRUIQ(`m^kFpgk=*H-
zJ%!FYVv@We$WdZxRmaPMNI?Bv@N%S!2{b@oR_Yj3J=;6-)Aswy#9C<(Nj=Yin7^`#
z?9nl8_&4x!%XCjki^g}qZSsha*T^`wV^emnj^2&(G?lTVI6BK4^8sFH_8Fqix+hBu
zuAUM~1~{LHsLS;cOFCakMzWfR=-Ww(3E>lp2e<!H&enQDQw1T|6Av{xrMa`%cMzew
z)g2R*FBI{mc}Je}M!cx8`L(Lby4n1Cbxr&ZuWRMEFPY5pRgKRa)%em;y(7PG9Xp8P
zy1S&77<>Y8o#Kq|)cqmh`UHGjdc4>?v?{ya6g%~Jm&P!Wt#7{=h`_&s@By;vG5E)b
zdrfTc_G>fs8&*|Pn#H#6?C2x^+={KAda|m@|9gcl*%*8Kt<o4fZ<Fo*wl>jO@x*4n
zyIdswREdOBusw=Xf5|O8u$V>|NJdYeu$U>=2SwKeHcFB+kkUTAk7mWbG0Dl|p$(*J
z@k*9Nr|s6%4%Zz3E+(69qNRcit%=WKTM2LJzyrNc&2YR^&rIch9VhZwsvrFJB}!^a
zW|`2ccq(_EY_EBQ14$o+KSfpUpT#hH_bZkpUe;FI;e|dXFS8~k)@NZ3eg?$qK0^)m
zj0qqe=z-#)_hrOO<vuMAUtN;>NoZv>mHmWN_3V5?zGak6UZXT!-u3)amX6E2o?pt+
zZh6;B)8#cvujO4Y9hcWAjh1)4v|C=IbXnf@(rbB*(qehnOQV(BNHO<%+$nQnWKy`b
zy+8fZf^0z|K14|`3|S)Irm}ODq(^Huu4Hb?UfOMaF1oX(YB>5O4kOVoaTtq!i9<i>
zm-2_tG9}W;S*Ao9JIj<v{bo5O(yO(2TfZq&8nFcpQirwQ54uYBl0QZ<3T^YpIB9=j
zl~ePCO+}s|fu;Kyd4>c^-G8*5odXG!%D=$Q0%E-{XFdu1rYnjBMqE)O@L*RI3EZDz
zKmx1eA|MiI?il6jf-@vS1<76(oTGPFF-;52Pyj6)kVy+2Hij04)GY5?h+^zh5^_hA
zNny+I<M|{9oL084@MiY)K9Rh_h+rg|yaKGE2;k%u;29%}_~m-Z(+2on7N?b~P)ZHT
z6DdY~2MeD_&eF(1R6(s3iM9$0i+gv6NJ|tdXI>QcfMID53pwM;(L1N(%L)%-ee*a@
z3E@_!waA?oDp=&(gGIh6D*Nzu7CGycMZOZm=<iYB7Wp=}$Tvo3#S+TLVMNZUfJ#~9
zQoRTIYp2<fmz)-P#KtXKUJ&crAJS?`^*dBb=<-;o8KuqlP3lHn!}V_SlXyw*C)-({
zRRxJf?_8RT3IVT&)#mjue9ZPNyDmw0uh)0D>$^Vh`Yv}}0`p$(FBpSk!y;0Zyc=$)
zf%&nl{P!}=J!zUgBhkqibsG1%PS!m-&T6gus51Pe{r7O77{+=nn3vJ{g_j>(@I!o?
z7bp(nwf1w>Ci|A*ZjfYTy01^stzA&+E`I~&kP>IrWY8+K&s%+e1Fk`0_=58CGb*(L
zLKW-vXU$aTRb}FX-Hud=B8#+tfQ!K-DYv1<Y!MPr;{qHEx>4hU)6^ITgY^3a^w^ve
z&5OYIP;A4^fDYhhx7LX9MkB_c)Z%#Jjc|kVlR}HPv`37q%s>y9;>smo4Tm1a*ja!c
zeq?6>dN_?UA3a>+ib4+;x}wm-Ij$)5a5}{RJ(Okm8L-=S26|W(oa3pb;=hd^-uAO|
z3q8E3vO^Cbg^=hz44jxst%al<wJ_~a;CssBpOImWJDdT(>~IEbsKXiH1k3dt@Xrn>
zn5}b2qE$G7j4t2=tV?ji0ZxFa3NQrvoC8|>gP=YaH5_C90RJPXA;^DR97`+&9gKwU
z7iw^s;AGnRT-4w~rU+I!R74w3{bqFUCkagqz&i?c6;43jRiB3suG*eu*FO&*T(!ep
z|2%wf)h>7abMV3NtKfSrAqshBCk6=NQ5G*~3$-9v?8;1pFeyL?>Gkq&;e*jMiRP&A
z#oka(Y2prf#3cEzbAZe5^9+*nNk^Mz?YgIsoG-(p6!^9m0{HhHHu$X^fIoi)@ckY9
zNWq`x;GamtKjz?<IQTy;2mZHClJ6(@Lmm8#H2l>L{uT%Snfi0PgWvHa@b5ov{i#aB
zpX1;!cJLK~kGcLlBlyQ1d~q87SO<TygRd04=iqM_{51}~b4~ERTOP9a?c?B&6nxyl
zUn2N(9ehg~ezAjpi38)aUGIB_gC8XLuRHizY51ER{GS~B3xXf!;By7P^)Y+jacTIA
z9Q<Vtexcyca`3PH4frJven1-jI}ZLd2j5%pXFK@&1b>Tz&q>4Y>);P{@LQJxA9C=M
z1%I)FU;BJ8J}=I-@p+#E?|Zu1-^amwf<M{8&q>4I<KQ26@VBV_=iK-0EBHPReo`9#
zmk$0K2j8&__>~TR{XF1bdep`znub5s!Jq5k!_p+);`;NH;Q!>{3)Aq2IQXwS_&23l
zt9S6%3H~w%-?ln<-}fGL^PdCnyQKyAKfC^n6#Qup{)sgFV-9|agMU@<e{t~r1b?W5
zpOJ>Y+QHxA;CpF&3S577uvFUKXEzGgV^tdd90z}~ga1hFAL8Jj5&Yv0zBmnktb;$<
z!M`l{Lmm9>g1^SWcdiQFcgq8A{yX@If`8lf{}REU>)>0`@QWS%OB@)VQG#FR;0Fo*
z>kfWa8vbSn|0f3@6a2dlK3DKtAKn?rlNtKEIdB|x-$ynWEX_h<xi)f7{xT2c2|_?`
z>f%P_Pe%yY3ZZ2u@<?>x=?DSsoZ+d79_)PdV6Sq<hN`97nCZBV`{?t)%4S-<P4|$m
zH?bqiMtcyN#b`T!em~d86iheA5Xa@p5y|#ljg~FJ5orv!`F<Uns0lZmO@+4UgJ;@?
z)sN5`d*IKbP%5I&*iWxiPpX?DW!f<+Y6rmi@9h&M3Z|jJD$dc}*bMQ3?L`H5u8NhN
z6Ad?fDh_1KDY&i3k7&#(RVtKI7^%Vl3f0j>c2V7xW3y`Ztw?4Ub>!+-BGmq&tgJsH
zQ0=bOZiE`lOr2wFj8~LB1i64c+|SL6CojOe^~^{t`K#hs^0ZPlpED$R?9gP*yTq%-
z@PdKSKqvSCaqWXw(nI1lN3tPeT}Q|Z4_<488UX~oUOdqft7}tj6SZ`gopEAy?W#~j
zttVnU-^1x6t~42^;FkKsDD`Ud6h)h!!uDOgbc_t;_GUiinaDaWdo9K|FLglkz1dl8
zw@U;L;)QLu%PYVVIo}}4a7E6iy=JiOZq`xmD+^<Oj@Ix49gO%={+ILL(q&o?*I$6>
z$+>;<_7R*&eX$`{M$WU-7N9~2N9$TST_}+%60QRq>WkvOc8<RI2Acb&^kly*)H|sV
zp{sYA*wmEPjksR<x{<v!9<IAp-H<PrrOaWD|CI`wY$k%L6Cv|^()sN)jVoSb%)9s!
zIbV=QUmjq2@R0jbj3<9B#$8I0yb2_<<jwB%VI3qizzRC?-b)q~A%x%v%!j9NkP4;5
zHW9940%)vc=g?H;LCiP#wV5In1t(R9Sk*9<0;Oq-CIW66(yXF|)qRw3BIc}6;XPA%
ze8-`UzAwZKbq0Bxd`V|9W9@6Wu{uy%TJ%Vr9bb~Z#W_W7@W!n560b0}OiJHFKf2!s
z=AnOpM4p`;`=i3D)Ti6}DXKa@Qi%{Gw<`z>2!?VEqd68GjUtas-yt{aS#*WZ=(3Fx
zSrne$rW(^1l5c^X5kqXm%gDq-t+7yNoSjG#8{;MOW7D=x1u#5ase}bkY+9$argd8E
ztp8Ex*b|<vhme5lyIu&-ZAO!z`O<i55Z*jWra=l09I%Gr$?p7ju9gxd&%0Vml)T|r
zONo-#_iBEtDzD)Z{yW!j3ICmIxP<@CHH+@Ve}}s`=q?Vri$mY5W2gX?^68#9no8!$
zxUAX(+keSKapNUJIIo&Fq=X$HcSX07@8u14`sq`10h*xi^2(>D`aYp9=a4!Pomip8
zAUK^E&iJtECw$ml{**%FUet$G*?%@A;v8JjV`U%JqWk*T){lHGrZO6&F+)A}p?&-}
zypmn9&~}pS;c)ghJ#7+UO!!Hyb#k)l$Hqr{4JvZk@Vm*AJg@Q=Y04@a6_u1<c>^a@
zNnuvm=%SHs;Djpa%_<vRFwza?B6f?^XO)dE7U>2~fRYxiveAVi-C#~)w@8^*+2|sX
zZZI#gTclH~Y;=K0H<+8)E!1Q;+B@GMHCwdo#c;zYka7Edyrb_%2yp<XGzT~ZY}WRr
zzwX@K-ru5udY`*2AkaUWV+8sl4iMZIpN4?`5rW<>pAtBQ<I{ls9N_mH;NeXG-YvkJ
z9boS?U@r$)-~gXl17M%8u_VXS+?z{-=f3_Yd+tXZc<#kGHO*T=a~(s0nwN%p+(ETC
zsFwuwte|i=?R{e!>UsxtmxH=hU0Vqhq0oIEgbCPmFQK@Z_B(~Q8vTD4Et`|Qjg+ZA
zlaxj4E*Eiq#}so$XLqTXJT($cP8lFCt+g5eHd`;$5906?sD<+tq71cgu6EJn&m=S3
z9Zfc55~Q8)kCUZ+f3M`_dBTb4#V*H1wSvdUX65H7FtgQDMYF8|S+4vvHg!Ak=yj@9
z;ugT>Q0bY<u;$}yWthQvg9HVBB&DSjPmAtoof9kMcnaOGdkeJukp%Yf%3EPxY*Lk(
zsSAYSeB5MzLeo!2i>Uws*B4DyryMOW1=`_j>FTgIA??d)7lKXjzfTop^@?^J5yTt%
zZn*=Lkw8eB#46y9e~GWtYQZji?YYfAYrxkET>VmjuZdQba<CJu?C(vISVY+wJg~X2
z?I5SlDL9tcPT3_~W$_r=UP-Uwi4N*rbSRbnH&cU8aFt8<6*W;XzgP^C#>W_C3Qzxx
zdLet<Hc82|*hkRp-LO`eUdVn^oJ6M6woe5ReLlZJP*8(5sDY2{pQIK$uX%BXr{Aj+
zqaOf9XjIspd|)r40sNK({_0qW-JPI*@lcC009Ju#&4CB|!kzeUl81$ZM;;aq!^y+K
zq3_M$|B;n93jdGvZWR6>Is8W9|B-!f<WhcRBbUJcja&l%H*yL5--r(n`2R)W|4)Sf
z#|!^!zncL6vqdoYzp~^F?M=1yd8I@X{zt249c-nKXQr}esh`CjSUDd0;LKFPZQ2y2
zRME*0hL=RfA~?4FUWH*$gtYzwgqWB80KX%%|6<5d1R)ARz#oeAalYLF#wr>$d&zZ1
z5IVe)SGB7;1N01rncJ0+ajHgDF}rQ>DznUDBG;@O3j)JY>oyh2En2k)eUSg~$RJ!*
zZp6-km?}pf<VQ>uB!MbNALK|(6(oTwM<3)#Ocf-7Dn}pWN=y|bfhtEI<V#EyB!MbN
zALLA&myH|oNIy=M{6mYO51;t-;a04eLJVJ)LCwbueJDH3;8NT<Cf3IcLJ4kRn*x>L
zNN5p{qcBs2YBBCO-&f}2GRy2Dv!ncFqjfude_6SY!$$eevKts!%NW{hNyvN$S!05H
z<9Vxw9@GN@sW7)!IVB2#pgeTmvz<W5d-m{I&U^OT9L;-n?9IOS>}S6bN$Gn(fX-v~
zZ2!u)U#8H7f7s}>J3k>kC|Pg_?euMC{*>v$N3SWcRdGrGjDL)dSxeT}4F1u_M|!Ve
zviAxp?Tfuv(ettnD^rlm#scRA3i0v#7?K)`r*z5SwN;4FObc)2B|b<aq?nJxd@AHS
z3knYjApz!mo5|xm$p*5}3^~dH6OnNp%gdO3faB#E^Fb%b2Op<#P9ZL6+zCyHCN`0E
z8-GMtRJOrcEZ^617V38eHI5AxwLue!ZZVsWhnXr+HF)RQxN<o6hah@lJV}@KhKlSJ
zT67q>Vh9fK%3i{$b%8e9xnaInwiTz=#qqL5@@=iCUlwk7j&EP_4}He+2#Hc^6j&nE
zfI>P)_@F%8utd{?RwJ;vIr31~WFnW~P~X<r$|>*T@^Gd-T{RCI|L6Q$M^Bqv3=Z9C
z$Kf~cvg62`l~@^XSKqT057^;Z=PC^{ey?7ZC1}R)wUPwPWK{mfjYWeDT*;OD^|uVp
zRDbiv{<-GRIvtnSK?TnJ^&p`8_ciE)<nmhN++WX@&ujLgkPx`;v8>OdPx?`$a7W0}
zI~VmEeO66m<j12=8WJho6&iikC6ST4<q7*N<*|{GOBbDTxH-ZWja<FxsLA=4+%j8-
z$+tA=;N8-yL%}UV^Fb<Rk0jp>4};=KQ6<k;m=>E-56bmbyBBoBkL1|v_$ohp{gB9C
z+Ahj@$>zlyme%|N`}N8Az@CvRm}?XOkJWvYWju05s)!n-D|U|pbCGzcNy*ls8c>Wj
z1#?X0`zmlGk8K6!4<0!{<Y3T47#zu35G&hQyN(!&9^G($vYjA9nJ)6;C7WZRtvF5N
zzVf)dVugp4-5tzN+QWi}6gwhR6iZE!K4B+@c}28ng5w|%Po1Of0bYQDDHm>NSv)lb
z9mNAX<Drk0bqf~)c9mj7Jr)H^LpwalKs)(1u>IhXgZ9OQKKTxz{K>b0?5XKvHNyCl
zZv)p;vyasX;Saj4d4Yo%bO*b;<45~uJoHLjzi-6KUaP&Q{V3u4Wbe`c2$=8c+l~3Y
zlmpCniL`|0ibWUmJtz(JT?ci#gBps8f8G&-67zjX8mhoS9pj+N1a%}(FyGhS4f9<)
zTN*gz<qZ8>F~e#6_ZlMh$eBz)KTc>$Z*wCR(c~`*W62SP3LM_gy!>lW8ML`rDMI!N
z+W!kE1rYmjexo>5bYMJrDOhc#x8DZShM-=vh^$>M&Tjxvev+~+7U%bMyKNNumCK9G
zp4b`OX5Kx-iz_@ZWVe}f&sMu_0}+3C?9tixC`AYU6b;I2Rwp-s)mD<1TnnXcvW4Dx
zud+oJ4~Y6YJQc}o6CEQ@EI-;r6Dj-v2=<%-suKmET&{AHD}$@uwz{$8<-`iQybBL%
zdYU=Uryzdbh1qtGgPG;Z)>fk{TU#?!uG@`J0c$2k9YWxwX=PLs{#2?_AzP(KXtkr0
zgviv2_<;^9JX5=iDHZxmW#fw`uXi0c$s^|EkRgfY%EWvm4p?uI#{sbta}OY5?2Kxk
z4>;2b2oOW9K=;V7Ep=gPC=!KbjoIeQA!k}eL|a(EGUU*RRg}y5Yh1^vnfORf7+Asl
zygMXk(OXDy06*%#BH35$72yO#UP&or5(z17%(+M`&!n1*MbHTa#N&jpROg~}vC2{w
zvH7<wja|XaKk>uxU?d2ZSsTpMvAzj=mu!)0$IW`GU4&rAg~l%s9oX$LwYOTZ-=`Ic
z)fL?zQu|v!!f$K$XVl*ATg10ibbp4bS85@tW8xO{>D{)rj%ALUGV5;BUV*jzErBEh
zH33PDKO@C*%yu=81gUx4hB}H8j3)XOg{KQ+3P<LVmdFde<%I}Y;X|1~^deCSQ35DR
zwg!Gw)0Y~!968uN0o@o_yJX|vsYC+WHkCto`W$}2(fhl7E5u^t7Qx5lE81e1w&zRb
zX4?ihJPWOtzHgP2M@;@PO_u4uB>$KuOZu-nCy$W+>&}`H9{R64FBviX8bqV9BYgeW
zG%i6Ar*R4SXBwA~f2K+Q<w^e~)_Q<bI@W%v{U+oL=@;3iD6zN%!TMv!oXCY_pP#CK
z8irIZ;fO{+v(a7SP_{D0ZZ%SS6`z^P-d{puDtnHI9BwLse^_4j9Q;E@pa#2>@W?+a
z;jrW#yc$iU3KU@S6dT8CMhn)5&Uo~I5KXaQJ7THq`y}90d%gM>!8Z{WN$eyvj!oww
zA}A97qLD3U8!MiXS2=T?9$0y;4*8WcbfB`$Rq8;c4pgppl{!$V1C<+Gr4Cf;K&8S)
zXj2C&Er_IYrs77Z)PYKiBB`7y^Es6|P^m*6$^WR!zE)M)Gsw%H<CQ%ZZa72wu)|<O
z3=>U3>m?tVnYE_=%EUw__n>gZFav{5;c6AYGp1(ta%=~i2*^8D*J(T0)a}*=SM&YN
zYpRr%w(hM~dA3S@WCG3-Lu<V}(3s2%#yx<*RJJt$0xL&%hlq1ikv2&YzQUY-j#cDQ
zfe(^Rbcs4ACY`7=l%t6{mnVEtC(kb&_kaAtx3^H}_<`oz=is;7i?nb)AI1I9gdA)i
ztEKaHC~mlq;>ezrMsY6MIY4pjQ4TJ1<deWZpL`0>v;-*bEFbfM;?{JWc~OARy0uw*
zUpIW#twja?Ir7`R!P*CWEC)VY<8US$bi)%G4bM+w&I5gZFpcK4FZS&|ntM6U?uUVr
zp3OcfJ4viNrgQ1Bx??&&p_-(TC|P5qLlPpQVTbW7Vz3pNtIEFYIEh=;Q6dF3v+{ho
zYoO#Vw$T6s;!Z+P))p|JQy7KVqla*QWs^hwLt2`JXQr~lQlGPh!;91)SRJk8qBbo<
z>q)UJwiW*OnRtk}>mA5ch*9XnUqixL<dtpV&=f6OODJJwJx*iKGQ~`?s)(%hafK1S
zyKM@~13DsBzBJs>3}l=c&a>g}K|iKt3K(L<GG2-(Hloxtu|0XoZDG=ytd+t4|If)=
zCK3H_j3+#*?})LF1@VMO6^t197@|}E5#D2nPW?s{JSG8Z?8CEl82K=lpdk-)390Jg
zSl!~ZR25g?x8e!$iNoL-FMG>ANVvX4V%5q1eAf0!&2gpTfko>$%I&g*!?JzH;QMux
zzesI|(v7La4$w8fjFW1|+$0GmCjoE)A91x8dL?S24aiAx`s36zAKfd;K{Xf0B9+R1
z(Bv##E!Gd&v)2Xk)->d;*~lg%+%8LZv0(|BxV^HsYS;O?&O$VAL?e!tl!Z>UWRpjZ
zB_lI^5vXoAOz7qqBq`4xn9BpND{>g|9BsWuihxm)t={xfzm!O&JXl+?En&|h&T!u5
z<^>}Lm5LZlesZ&`f_4l}*LAu&NC$O=4sM6)2F<goJXH;PQqEKskGP$9#D|>C<ZE*S
z@#&~voA`7h2gIk3?m^@k<3yfYZb9Ukexo^AeDYubzj3C)@8jVA#_rqm{^sB>7W~-`
zeqkDZzJp)Qf%aFa{Yeh~WWgWp;BQRBU+>`Wa`2A{{&EN3NAR0&*u(xIw$|jY+taSe
zX8%zAUS*Wqfb(UPM^*g7_75+4#TT-Fpy9cc**qfh@7O=cFpQ}3h?!dVuzx65fb1XW
z<X{g!$NoW{lIOY%APQmtQ;)M)0)_hcRuGII3iy%nBXyb;hV`_9%n%)bOhVx!&I%G(
zMm7X`Of!T`BD*6BHe?fmk2>;*zi?UQ288QtOv-2pgaFLu*r`bq(s59CeTocyh`;6)
z%^E;W=bNj&{Wh7o`an_bT^+drzCho&<)7I<Rx7fg%l_f$VvD^_m;J-Dt!M8Gj9z=%
zKb%ZjLCR-E#{0qYtme<Ne?*l@GPKt0A7u4jm`-Bx3{Ms^fEC3$64*bCDe~<fhU>!9
z?@<?%{VfY=p>G4Jb~X^)xJOF)Xf_ZKN{g`9Fl-<vyxQ2jBacKB{E~-)U(!(U+xHIg
zV~(o)EqJVEl=s^v@@bA5{@YgaX^!gqTg#_O*ui}8Hem-Gh7)$cq3<m8F67gEz4kA$
z#D`-4$I&A`ZtVf>$M7*t{{h=>!cJugc|$vh35kb3GZ~=P&@7#ze~@`&@-eFQ#VY+g
z5#S^)F>G%9xo~bF=8wP1{1JuJcC&wg#Ihd;eT~X~7=%U=5q6P*^4Gpco=t5m<K%ly
zIdrGwPEpXvHOK?7K=G0eY19}Y8A0w7%C&hcBZ%C`5k&l1>?g%5L<fdsfO<czZ)@I;
zH3E~zs6Fi;RQ0rfP}S4^K~+!t2UR`oA5`_Ue^AxK{xPb%{Ui3d_77nO**_d+_+0yk
zhC&ort}C}{_2YDseTNF;0{h38&Hho%pcY!bklpEgAyaiT3~^m6D8(rmEMLfy3T=lA
zk^xJ|)L{ly*ga?GV3y)uY?Iw~13cjPUpX1zRE{PC)LsQm5d!Hbaq_I>>S%`3OSexR
zu0K*l3>4vO7V^_b<p3W)1o)&aYSx84@QL~wt=Z2Zl|GJt-{OsFf#U{iID@Lo*Ppgu
zkow$s?KzA=Eb#$gPFLVrf<iwtf2RGyrS3H}^aOWtyKldEs%R?b45e5SkShv1tajq5
zaqN)<#p|P)?H_`b#+62=rNTbgpYYVq&S!mrW5nX83M>!Qc+|47&<8>ylaRH|Y=Jcc
z)9fd3*EwoHR*V>`2Z}OVl$gx?OtkC`*yuRXm$F)HaaM~pVzBF%v;QBRbViHH2)nUe
zSJbLrJ4T9Li(d{mG*KsHP>9*$M6KyklfB8cPdP&OPk0p`+P3NC<JJQ$xssiF*>I0t
zdpj3Z4DdH&JR1u@_Mqk_e;dcuoDS*_PRi?l$D)Np3Ohs1&y3pIyd%HARKw}5b;%=A
z`+%>r=S?++e+LtEe6t(B80hOW``(3b2&E&(lh8&^yy4ZF|3W|ChwLYOg1Pn<oF62W
z2B@MhHTdtu6P;;&eVebZZ^1MZ=wq(PE4p#f{PRXd#Df*xtS!soISC?LhtQMR4KhE$
zNXp+ua!(HGdU7G5famF`>)A*Fc2?*~cna2%a04moS?5H^JQtONP?w^fKD~sxgSns?
zks1{O>`fB1$UYrf&Q<IHng&b;)FA2aK_~OZqO_pHMYfOVSW{$B*V+8as3NvMz9N6D
zsWGSmKLNInt24bt(Y1o=W3M_BwFT_v$CwMePzPp#IB7h7-~J8QeN^rD-V^YVf37z^
z@^2jA3n$=nJa7LZUv1xzhN^Q=H#w*q(aX<U1vhAF`wP-gKXy>Ra8T0)wc7EF&tn;H
z`^OBIzP{^wchle3tuHj@u^0)~S8yK`ty@&sgL4#*nD&clXqxgGLBV-KBRLIAr;wuL
zi~;5l6U2|0V%Q8wL)c?2F~5RradGK2QTDRkM>I96+DlD9;=tOpIhJ^#<CqNjGRF2B
z%Bq8Z*G{&}iK7LozZkZ!0<Fyi7?Bn}D-av-?9dly55riOw?*AYtS}L11l*D~a8+#U
z>Nw^e#cUqw&(HbL*ZZ2ckAU){eY&3@spjbJ7W~cXw)dGZO%7R^*Huw@Uv}3Ww#{ti
zeK}n<ISwYbt0vdg<aO2LxtfBmngUl-*i}>LYWjB7^o`av6M)$kZWt>~*qd-6rvC2w
zo=RrRU)p~H?#BXq^DMA8Z+M;YLv_tNMl>~V9<j7}{fO1g>qfjPen$>~9K5R?6NK7S
zOwiRX26$oZqa?WT-3uE2!h_o{6hoWr{q`S%iQGEPCh|rOOyocRmC4tiiHyzYoNJkU
z6<25GTc}$fb^J8?kQ|$N7xYTcJbY#sMZ0DmzOziQXtF|cuR=aq6~)XwtS43-%VX}X
zNGx@{ej&6Mst8F*tYE9@y0?t&N87nu(MXo^K0Q9nFONN33Tu9?SrUmRhvBL;r-gZA
zvnHC1GM#7HISgQ4A3sHuks%QO#4rf8T`R_?xBiSIhkq{w0(tWi5<9tfU*X-Oum`pc
zOzyO97&G@c3Ia5nsEWe%??CP(vQ|E_uDSF0`Qhms=|uDPvv9{@Vm`jkU2C&zbz8XB
zwt-RA8##bB2e94&=;pTPG62>)fDH~{y#W520kFXVY<2+e3m}yNV4`>BBb^T5Z2?@8
z0bp`?<s+oz<^}3r7r^Nm049`YKVsLXZ_q4y9-waXwe71J@ELDCeSao>*N9}i@O16P
z*L5x9U8fghT+4XP=}J7@bzjCCPCp{!TE@#w|60bijQ1+!y@DpQyM5w@!?j{ae*mYB
zKsXILr%6_<NdoWrAc3hmFAdJwH%8N@xcho;`wSGG{^*ZtPf4%r({O!%8WZixepq8W
zVQujI$^fMHt@b>7VDA%V2MhLFQ*6POb6~-C{)GiQ$}QMGvR2y~r|;ao7y58C6SWS8
zHQexh<Iqkez+y0F)3^ANq-7<HaxkV-eOHJ90Tx3spEldIGYGI4iV5{`&U^}^b>5A_
zuqT(LSXC%OU_7PBhzQXy6@_8hFRw@}f$3$1%WNVfk{9q(IAjh0^22Dazi82V$%_U+
z7v3cj#OrmBYFt2idi@K<X5ZfS0wPoM_Wa7<f&WR%_9plrvkQB(fpcixOKhqAWBXZr
zauD9eJA(21c(RRO7EC+i*LoA<_jATCaGq^R11@%eFL40y*M9~uArdd=*}KwEvmMl2
z2X+1}K>bWmvfocmLtWvZk`C%rLDdLK_WN_vP{SS6PaM=BLDd2UE?!p8r+swlp7wk2
zadKaU(AU4skYiVAcrJ~P!}X6)Mn6=nZ2KDd!l*1SH(D)USqWRNwv#Db*-df}PAUn*
zH&gK?y258d{c{HR6)Si0RFFRyI_f;rVyP-JE0AsRy*>n&c6zBPef1!A$@)$vJ=5nP
zAtKpzB{8s&isk(d(@l;(BG2Is!}Xt<3SMy1v<qVw)6kbqUguUN%|km{wyfr%80q*E
zJ5OlRUd;7O4iN5VRr9rJT4+U_O;F1^Yo2Z1c<xft7&!OB8vx0muUd1%e`gl6k-uwe
z{@}dT&7V{)YkvE@=XG>f*Q;8^k-yu(l~)bold9$J>gM1kciNeGx})pftnMvq$Kp7e
z|As4>lyPOiA57uQ)X3juDWJV;eown1uiy_sn<#JNW$)E~(0(w$ZUu(xU(s6Y_GQcY
zvi7mUGp})%7)a2PeehbV@xSpnPSW4_d`|RvX{ucC2u1E2Q{%h?GQTl)`~$T~<8kxY
zr>Emv0sYi+^S>`QB52rNHzq&X_k~i5w%MZB_oZ(FlAPOdsPt`24W(NNwJ2#U6QZ={
zUhFYs2H_)i9E6wseB#bY!^kI&m3_wdg@Q6<`o0q`Z6trUckw1=G4WE`j=;MZS)l7=
zOZO8r7F}1icRxWe=(@72>m>JLzZ|Pu#OG_RZM07(hW3oq*q`9WsQnK;%l7>nP!Z8!
z&yqjoG_D)YCm}(TB_wF#i-G<8t!@^*bNRtwfP&`F&TI0ML0GD0FdA622Orh@gxdoA
z^{q(`|8o%jzZ)dc>X5`A!4GYXlY8L*D&x|g$PV!TOO(1v3RPYF8cBCR<>pxbDnLF~
z-{y+Sf^}$bn{<WIG7*b!*%@d=Vnnn|6l1WfD&`D|ah}e={<n$)&B>Z_J(F?Db(P6c
z-~h5@`{C<g6T;V><UuAZwN@-A6517Fn0c>F-ci82d`814fQH1(C1;8_#8YEHL$6oB
zHKNGrWKUKYx1}9F=vq&*n+%9)6HdF!jC^dGtlSvd{C$<j79z}iJ(Eh1(K=|p?1KZ)
zey|0@DldU61Lr)K-IR2HVk4Ys$5)*^B6)`!$??Gc$uEgD&8bPo&23$DNQyebY!b`V
zt)t;5;|Zm2+@LQnHZc1{^$m-UoW$nW%rzwXp)x8@xXSQf^Y%RNFYEXn{+G7qP3J8Y
zz5_cfv!kx9w_07_U|2)fKdV|`_iaw!A;`{*Gq(Srwe5KYf9c+&lUF)wXT%#mWX~Mt
zF|qc4g{ukht&pDgXjae^>`c7u)7tiS<!gvK%RC19P<i;K#pRJ_@w$oCz0+!{wx-0w
zx_A1p?`~}C8)u{jHS_t*h4_`9JDu7lj;2#P>Sw-P{fVDu3puLtxB@DhTh860lkeTa
ziA2w%E@}ir+Sl`v{qUUF=?xsK&uL%sE7D6O-bMDnDG)cGenot?BaA$C%@2}UD|-Me
zk-psA@_I1yTEkb{xDhikSwZ88mxF!d<*)iY8>7F<aUI9?96#l_bM)KnAdd$7)Rxhk
zt~%-u{X?VQezZr$g!_xjR~p1aSC#ysfBxtRBx=p9Ab)FS#cz66q<U5~WLD6hCG_Vx
zbQ!JA|3V38Aw2a5@r63}Ibmrmu{5@IRg4|U&=(!U-q}p_Rp>mFj;K7VVhO~UwYwX~
z6S9xISo=}5WEBqOXsK7RGH|fMIZW)G7it&0Tz|8&(C>r>FUhuY#ggni)#0c4pA`uu
zo!^PSq%W_Y_?rOzl}<GDSHS`Fx9#_!zb_vH`<8ELqW6Ajz&;MJ&;c$;0oYLlU>n_n
z>uS9zXz`5+*5bz;XmPOsHww@XyLvng_^1P1>;R_;aFYQ2u&bGA!0Q~~?;POu0^9<i
zagO8CP!~C<%N$gVptcH1oa5j$RJnsX%R!wds7^tNbL^Xj`ig`4rh_UM)Ha~t92>8I
zC|vnVsgTnAMTUIf=mXT`pHT%k4Y3EXj7?Ko;SoU2Viq`CO=Ojr$0_V2Dx^Lb+-O8@
zXq0_m(3ouh26_O?OX6d9=)QRWnRlfTyW>4T?2h*+9g;vlb+yComlUBkdZ$mk<h5vO
zydn1QG<M;0aX9diIBlTd6@F$$<L3fC2bGJ(01L2n@i&<E0X}CJU>~1X3ZF0dJbbP_
z9B3`HK={143!mpWeEuTO5k5b*8$SO8WV&foi*S0b!|B8_g44Iv+%Jao1aSJ!J`*l)
z6egb|d_Ho{9^9v0Gv?E-XYihOEzNq`S(@{-Gh;j{^_|n5?R5Bjz3};~pNr3jgU@|N
z*zox$K0X%<!%H>cwC3`!2Y7w4(y0Y_y#<t9xH9qC-jMs@JAY!x{g{I<ayb3C%CWBS
z?N`U7k^A>fR^`I04qF*oH2SI$2Ry;4(7WiR;dj2K!|yAE-;Gl9WrwBldwYQ2oA!d=
zn}pxl<%vn_<99;?IMWEf=cMub8gM-J$m7<$+kDKL4>@k(xNG!V_ZP2OGWxAQ7O#13
z^xI(YRcRbwkU#ouFu1Fj0MgF}rGvrWO5^x~qbv=#t2hy)pG^fAJQD&8o(Vy=?o5bJ
zg9_KiqKf0ztXfoY%o?@>6qb+L6bH+{8LrP2me1{4?_KzQoACYDdg6QJ>HK}Vustcg
zd~Cnau>H0!Z14CTSk#cPJ$yiZc1U+;=Yi?JNWtOv#{&F*_OAnMJnX*=8(+u)Y<$>l
zV6vg$cQD!FS}@t%OVj$u;XgFa_7=ocpP$VYFFUH@f*%_vi&jGjs)M|vrKmXa(pU%)
zc&gzZf<~Q`GbGUzPetJ(7oh3(h1pSkvHWb$>-|nOZhXgP$dB+Z(94a36a7w<<^kg|
zR<5#zHB;e{PWX^ii!F{=N*gEoC>7oChfOUu@?ki*3znhpSvDlGwK6<pXJvTkPO6qy
zq<#^idU-{7$j24op&z@O;Os6~HY7D=AKf8vSG~c%YZ-U-(xuG1;Q8#X?BFheyE|b3
zFVJ1R_q+>k(C*r|=UoL4`#1IJev{Oo+DW$IaHd;a+oFkWjwfDf6n|LonC=n$ki=#>
z{%%G1^z~qt*zV_Q-)#RIki?&eSDb#BzD3<O<cqLMBrdCauN9kH@8$~)ZyfMCN3nbd
z=kh63IQ$P6WIN~1w{&{k4o;hwe(d<dUjY<NY~_Qc>0NW}ngz{Qty#wLd5*7ge3#>g
zqbGb?T)sJ2sdx52arA_pJuBeo7Ii`FGkO9XU1kONAhQB|kXZpf$gF^)>t50P>6+)7
zcdl7NCVQV>Q__(7<E8p@p*W!>Y;Y5k350Z68<~8fVzI4jEr1k$je$eqwj0aYQmmo1
zA9*D&Ixpi7WVkBX0Gitv4Q*o2&=rF0!T<W~UTV<(OR|Zm&dOSnU8V!!Ghev03DZMo
z?fvbSI*Zxm$zWk$1|v`El^GmZ*yY6k&70&F_J^$LwzdBy<yO5On;D-ngLkuqegL^(
zoC>pql41{tREff4h=}_;Hy&i(IWhKWv`whwxFXk_3abb=JjFTo2nfj;a3SfWh1?a}
zx=mXMS(J(RIfVpn)CNG>@v6by7J;FOa9cd_ihM(Jg6$&n^-v@SVg%JdUPl$r#5}?_
zL7dmZ&s7wAYz1vkbPhrN%yzG<{p2}N=p(QwDO~ivnKHThq9~b~HSm1J*9HT6>^K|H
zlQ}S;Rdoz#Jp<Z)rBchdQ2Ar6XcE2B)qoA}KzA|aR(438a+5EWUx%`9$~DA!O_@%X
zUvz-1X5@9Hr~3s(3w0roo_B|#BmO=4Q_2@2vEM2N%}WqB;3C+Oi}M&TxW8%~eC1&w
zw*ivqs3^B<@&j88D@C?#K|PBIvDt+X5?3SRVuYNmvIwe-$`OU7L&--DNi2$M!^4}3
zq`r!myhJRTz6#1KREoI!SIZk=n3wu_wL%LRQXGA+D;i*ra&E-(?=IzBf%+2e6fI+Z
zIKUNK{e5oVXL>t>>*5Qmq~a>KhfHGcNU@q#Uh=-8sa%hR@E2BCCV3v=E2_iwKLDwz
z8<Tji4KjU?i{c138(3#C=TfP<HeS1ZPW7vY>5gLOG8!SEUGZ7kYj_UV36s;lCF-RY
zBK#og$K$ID4^oNkA*Sec4EkQ;6FC;m5@;FXYm86lwqizy?stODG;y8Oit3jh03zWs
z-%TEStyB;JtInx$c}68(SDXg^@S=E$S3ADp=)ZKWjC%wRR&_=5AdVS?z#V<@aQz@2
zPSY@5&SS)5NZYugo9HQgw;rO~6iUk=;z5G@jC6g2HQwm@v@MdpQZzIf?w*INwEOG*
z{ZiWW+)g#1w9)uqc>`IH;rh{f?eGH~+Er9f)K=Fyd|c{2Q^MO;t9z=boI3SEO{jMp
zIHDN%uK#{KTz;<T1?=q=cdXEW82;Zp*Liw>8XfqQP3v&Ir**=#vU}tQ%ZTf#8s?~u
zsFv)Nb*CPu#ykjJ@)EsuQ^z@mkkj-^&l?7aO*wzohF*{di%nF%bCp`fL5AK`3lOz8
ziYEXx<Kty5;ran?BwCbr*!sXUpTtEy-r81L9U><+5?UJL?V4+j&9v2Zdq`p<)+{sc
zchOVLOR^*rYdr9gqN&{I=350@G)*XSu8Ysgam}9h&Fie~AahRLn+!FAlQR2&mHGaf
zxuC!yCCkx^#kS&8{~mHhpPIw@7>*u@pUD0-{ENa$(68=ZUdwi7yQD|LuZeB!Vu|O&
z)3raB`+nWwRj&MXwr;MRk?qcM+*yu0%XMeD?kvxp$@!T33fx(NJ1cZ&@>S+qUw76Q
z{u@5%nc76V_#e$*UN3`RBq%AbGNnkRJTIk4rM%%OMJnZuOl5J%uT1C~aT2<gKRlsp
z`F#^KOcsadYMbI9SrU1VW2Cdf4GVZm`{n#V4ZczzP=mtJMh!|hKn>o)|9IY`Py^FB
zY_1ESHo;0kgnK!t`!O`ls}q!T4$r2co^ep(yJ+jR%YnL4P|`WfNkiS^pdNQn%LR24
zP;lyfe+EJCbJ5=Oy{+XfjGh+~lj-Pre@cYsA^)Qp9tu4ttZ%913=b$Xp7M(SoUdi-
zD6?w94MJ4hMozOUAn4p?>D&oXRb)T-Pq!V^eyxAo)(Wesy-FI4cye&WZkcaSf{2rU
z%2#r%uE`e>Cm~o+Q5)kPQl^~;6(?)_%XSY`-0gQ+`%Q}`@u~P}P;tFwPbzK^vNf#Z
z-eaZ*ALK%DtEq!0aA2tadL@H6pFwQD9zHDQ%&LPZsvD)Io3|Wa?4eAhVp6JwImgkr
z0Z|rqs~e7Tm782Ss#}M}{B0?7;>ii+vE<L970LbKxerpN8pF&0RVl7?Hg|~_2ssOR
z&az{TXcqd^8r+4#4Zl`hD*Nvo)b}Gs@d{&duQfJone%r(VRdxt#%LLepqc|ZM#B!d
zliG950s=N>_rS%qF)y)!GSM*2wcisbfwtZ4geYgl`Lb17WRBD1n7cSE)%>^jIbloN
ztT8(+?R!;wA<9a`+VIRU1q&_P*9zD(cWSQT0Rk_xiJ~5kO32dmi*Ja|Z{0`3Q(l?q
zgtt~P8c*yXOKzw2(@XVh)=P?`Af#Ocqmk2&fj#(VX_*F?5z8O2P`d*DW&`W{pAU$Z
zy%Mhfh57&<o@y7cIe_}VjeaAb>qb}I*n1&`b<oKwuMA@AzLF?_?ybQ~ZFi17@(j)<
zM*Q6#6_ixozMQH(`Ui>a4%G8{;!TCHY-&4Y^iBl0cM0nsNZw~x_bt^C#uS_%+2cIi
z<9zppneZR?IDa#J&Q6gcU>Z{Qs6||ovI@_;SlN2#FC{KD3V~T`%+)9H#4cvc$4dOm
z@Nh}^mOxH~$;2ydbu{3GK9!8w(SDDiNY+gQ1Lz{7#aK1*I4z=K(jtx*`-wO(RuMLr
zz{Jfgm)xS+bTS2=9bxYDqi7TBYZQ!0{)*Kw!}6_W9TC<P{HeCnV*F7&vPgcz)3Oq)
z;!XpqMKmeywl}Hd$uL7VIo6nyh)zVBhv1(}vB`>WtEeIJQUxhBrt>)*Ss`zt$RR2T
zR_jGb1dQ`?{PbdpXGG66Gh!uc;;Bj1nx2ygajaJT2I@~Bz5vqe$B9uX20{Gt6C67R
zG)9TK&ZB5k+J|Ba7E(pZ1bwp{E505b=(-=S-%syMOLm(@$oswIDaC3E^A?vQ6t#?W
zzv~v2Pv~v@KyLv($hi=cp2r63*qixWIUPSr^8*4dRSU8k^Ch*|2Iw4344pHa6>?l`
zl%v{Lr^V`E9#W@gQx&e4ixI>4K{WA_nzT`+b2F?dYRQ1EA`@en*gJcqCLgfwoeZW*
zFz}mtH=)dGj(SHuPJ^tTVdt*s#y`=(I9=VnK+&I36?uj690@c~S0_=7!}8J9VH6o(
zV=>)rd~aywO~#+ext4x5>Q=Zp_w1XBX6c+hs2zdY5CgZC_tv?FSof{8qkA;`@7Z1a
z?`+*E{x?`Lw-?P(719n>E!^jY9&ubbN42YdKM05fBUI*9ZhYzNRy}bBS3EQjtbkHU
zBvW+`0fTJgq^_2IK)s4o74)t}MPT(dVn|nV)LZE(8d-ISkv(2BsoDn9O<~Dp<8T|+
zk{&OY5Qeb{*l_*PHjF^it!3(#9^uIdtn~-W>K-A>FAnSDHJjY4am9v8#5VCJY8P@C
zbmP~Dlm1j6Ge9#>6;0KMXt6ML<DqSy+b1_lf^v`Ev~e|DUrBFRwcFMIQZK}@hJ_j;
z77zKGq$ffb<pAI@StDr+ZdX(R2X5;gA_dzYxktqkJLPuPH!S&4YBuCCq%-*QMtjr<
z9z{l|39v7VL7-dM0puk4X_swxd!$HQEmkVDvR&c29o%vPgn$=<O0si`%-6~i`K+m+
z4r|_?m;I!YA^sg-$9#UZ8?gLumESR*SQ(yv1&`|*pFQBIsC+WpTHu!!_@xGST-xN9
z8h)w49hVw@so|Fz+;OSlml}Sl!5x<xeyQP?8r*TI;g=eIslgqW8h)w4A9o4<3TzZ<
z{<zOUCVJ4rp83QU&4+(~RF1Zoz2#_|H~djK+UAXXG>ZdU2Xl=CopZTHg3h^IBSGg}
zF6G-LSbJT9wbvzBdnE&7kDkr7f7JF0!5%%~2JOnz{(Ylcy?-z%;1R!fj`4_RaDYeL
zSBjpjV?oqR&)$cnp$>IW#SUuqc%Yi#B#pKI<g%cxjX$!s_Hj_5UjVgIP-5*@q@k8O
zsMk2~#0}`;=dA(?*1lpCtbM=p%p*NL|My}~JQA%u2{cgAl_n30rVbzRL!LV@`+4#v
z!<>;xm#i}9l*i#)S&2=aQX<PCmwe6<rswl6EwXLp>Ksz~a}iY(r233<l=-@AL7Ysu
z4PWH~6%m$6i&u6=G~BSC)x=6Gllf;1%;kwkxe7SH5!`#0d<$23Wn9Ry3uozxuSCm6
zaPJmj7O4~mj=(wcKwU2*{Z(%YiZW@~A#{<HOJocOmybA$4IBmJ{s35B=a50-;?>Eu
z_`oxKC|@#=<7797n}r_vWs<+jVSlf+Y_Nt2pt8(uDx*QE3^i4jwXnn5{ED)stC|~L
zoie`yZ|+&*MM<b~+;_-)zUb}C3lI44G{zS1l!%W{(dV!&7p^XW{2WV;l7HpeSg$(_
zaVIa;7;Nx*&2^_!Kqn}nsO$9HURH8P>sr4Ex_Q}u!eQy~rmW^|&HS*Y8npKpyt9O1
zoYl7H4==*paPIuJQ9qVjX5PVP&@GnTg60oKHQ^0U8tUc`Mm8;a`=_^ZSMw%(**lBr
zq@?KVd(6r0-P+#>;}oO!kV-8_Cfta)T0R{w6>c~Qw9@{BTOUW(26Osw&8^hI$8%t|
zHfT=2^YtzR{+5~maP!#)*xLb47vMSp`Ud<JX~5+U@HGy!_$vXv3!tsIKc%7m=%5~R
zP&ea`GVeV>Ntm9NhWeF*YIIOlYU_PLX~mtNh8p3Ze&(Q###49Rp~bf1o@24K{p}o9
z+}_sPwDm^fAr#}CP+h1A$^>Q{$9n^-7_>JJ)ttr{#y5^=NXNVV50|+5NEA6@78pOc
z6bU!n=Pp4k>K#|)-^FacUY8(4<^Ggb)^T+qQ`ptx%l(}M@6{dPxH&u{YsyL$^|-7l
z&6Zw?Ugb$rj7)jd-Gw(l9{lAJlnQA$nk<j1A9+?r62<=ufKvfT3%mv(LQRV~de6ej
z#OrnMf^0wg264Hc#>Omc-bR?#XQLL&>55tTy}TjWSq%scj=gFAEZ<kvadQ=u8LAzg
z`Cn3>mu$dnE@iRSik>OnDLhlDsA8!rVR|PQjq}~lVu`INSIMN@S($j1n8JACHDt*h
zm5Db=ur0$D#JXIsQL<}jD*O9n8C81&s3Gr`UOta?&*og05{2M+E3vl<oOsEYQ63yi
zokfHqK2_rSn_{VQi&rc*=81bm>?I|NS&6~=pP8EcM75WAk@fh!BPy|;S+v>}c?Zlr
zh5U<6dVCg-mrOC6TiU1cb5Wz1QdPs!NHldL?{h$83!?q;x{7xMfa56QB_Tk40X%!n
zc~?5jYJ+N$F1N~{mPnEvX0<@A)MM@Oa73CN;ra*6obO;P^4r}hrKK7(-(fXBN#mg>
zOKpUs9dkup(wxz&vtuzsy@8Sju-rQ2c1<$=NPV@7sPt*s!8xsLG*s^kH@v5lS7Pqv
zT4^oD=5kDsNqV_rdktRBmqmc$-Zp#Yb-sY(w~b=lEiiXIT>o34FrIACyh*9`T}&XA
zYFJMhAY3?!><l^?rm>3HmElUaN2LdX<b88=FAQX!>lrnj^r2&9H~j7RW5wX<&8PZZ
zoS~yUT!d-)!KCAFpUDZnFZ4h^#6vESOD>3ba#3+O^|bX}40rOH7Bzv?I}Y86i&LKX
zW%FKK+TO^efc`{N1v#$T$rsE<@7i_z4!?G<{*2eA{TZ*_!=Lfm41Y!f`~rVSwVC#3
zyf)LHaX#@^&ze7D4Mr4P+M_A-RiXKEcY|N1C%M-L4IxLw%3i4L2#|%LMbA%Gpm*&x
zF>KyEOWS{I?^w?8%k_wtRDtlu{ysqP{`RK$?TlP0GC7kH0bGICxkdpm9{(Dp9@p^c
z@vl+JxK?njf2m+*_bX8%7`6F$LcSiS>UBwHRB){?U+C3m;%JSEI!#aLU%`x^EgJ25
zih`KSme;<~{7D*fE?G2x((z3bzNBE4jvv|hX>LBjSm1WimItNM%F($ng(jm-<YutW
zz#jxC2z)e?ljhIze7tMxH0fhm`dlT+bG90NXcSBXn-&T|A|-Vag}r&i@ENn4cjW6=
zWqd9vD>!cW3}WRdw^HVvcH|e#U_3|&&^}>lW!dZD)CTl4n&?e5EmHI#$H;{9PfukZ
zs@>eDa+>A0zj?$+Yjdo%)=#au$EtcqzN&ZR_w(C-wsxtS6*5<|<s89h(|MYtxdD<e
z(&N2o{uLu~ui-a8_ZoD5Fpl_(p9`oQv}PWfgQ-+q-DTN5$eIu>`6QM)1&qM}kblXi
zHWu3VZYR!i6?N;_+UuPi?7k}dqESPQzio?~eUdPZtyHNwL~X_Ar7r8M&r%3qxKhY;
zs^C-S=Ef3Ah}juSotDd`7OiV7<ygvlyjn^&mxqrT^Sx)k7ko^<?1h{d^>Fr5huX}3
z<>-f$2EuaKD@6_rX(<LI(4?g4M?k$;ogx-Oc@`}4h*1jp`6-BKBi1wF$~L;@fnMiu
z4A)<38y8;ZlQeg;UVf$Cjdv7g)luJE+G=Qob%Z?RyqCHcm$^&q%_K)SeC<a^ZN~SL
z(hz&F^;s;6QNl+Y9ifAe0eb?|p;b@fC2d;ZO2`{&X4o6KwIK}5UW-bnMWfOU%;KJK
zy>YodGG}jJ>BDrV_G{n`tUYc3318>G4wxq(X4+9T!E`jD;C58$_+hvCryk8v&oW7Z
zvaD#-;eFIWqXOMXLN$XQ(1LKP)5w6|P8TI8g%Z%t#9W^YBzR0?^#48nkCv_Q@jq$)
zer>3~Yd!82|BGpWz}6HE_&iEb+H2y$J|(!5*X(M~$bep2h3fwu^k8{yBHpmh%-J;+
z^lc{eKnNg*@tE;N=z&Pfzn}*&!J5s4on6+Jg@>j+Rf`6dy_=3A2FI5L_yc4zjgw4b
zK<auSjV8@g(F4%|>_uh|h)g)oNP@%KjsFS0UheR<;YIMZ&?T$xZICdEw;uYSz2fV~
z0j*5Dg2p|XSR(!5xM;~s(Ny-yT7_C&DB{7SV{uPOk^LW}D}Q7koyJp_JW`FBxOu@@
zThl!G;R6s)^{SF72EZ!RkdBB1g3wP-`bF4n$783BOV0m7eMjFXD+Ymei#)QXJA8QI
zCg=8Ct^^(USU6R9VCJ(J8Kd8he{t=%(wOw8Fo5RVF}17lY8qEcc88!8|70;SmQhcQ
z3sm$6%L+C(d?o|_$Ml(9+$A29_^r@StSeX5i6mi$c~D<MtObF}#fvAN%<mG-pQbq4
z1ZGRt$z8LBaS?iEi1M^XHDpq~s)aPS5KCnrVJRZS-~BiEg>t<n<KCo~OY;j=%)I0O
zTl~UvZvJVa7zfSddjXV+^TS2TUoax~YU2;CPV)zkbJf^_TuG&Fn!H4cR=ne}`Rt09
zyumAf$9RPF%(H2hIrDHg$(VF=zRjdd&$n?M#|tS)0~1U(Dbb_;^crTLEg-l1Y;gWg
z%&f5zP=I4jov8iGuPd6W7wei*W?i#Z?ski#Ybw>Gv?>JeCEP_dRV}kip7tauhfZDM
zL>J>kZ1O}2^Nn0}vp;#B$ecXi*5tuoN%Lpw>2Cfs_3u7^#`T;(F`z*(U0-^4WWtXF
z9P_On{v2N%-$?v`Aj7|z5B@iqRBR0%tdG6omp+1l>l>1ILlL&*0glzZ-<{jrQ)f^k
z8BW?V41(ekG^SD*#1bpOL~XIuk8`>7Y=DG7Kil&T@58_}uRV8NpyCi))ya(=O3v&2
zZ}dHdWc=@$2OnYrqptyI(xom9Kn8ZgOv_gn)|dlSxc=D>g*9Ar`pm@vHd24pke67Y
zabhM3xv<VbeH}kDgq-G|_3=v0IOPlNyfpaA{Hdzg4pyrJZpHYsR@zjLY@rFIU80Q=
zuzqf^m^?LvAD2UEvInYD<J2j;fi1<1%R-Y)J&0M7F`Od^GOs(mpQZ^ItLudOLEM^n
zpwImzsg->{w7^va^2PrP`S&^TKU<K0Qv}I61}ONxd-6Xo_K<)5EroeQ5@q4qI<YUg
zTa5Qv-ToV1vd8y<QA}*dd)PrV(*E$9)ABF7ZFQ3{?OIy?W!tTu*V6JY`)_4lE4Vq3
zS#Rln1xq*b@AJm*)O;tFs`(ACyD^#?t|R%iFH(OB8iRYU&$w5@|C$W`W+6W8tu_t4
z#7mcbU+I&&sZsdanrNDH42#XdK#8)%CjLlv54`SG)LQMQ@<UsTCO#!~m6JixWVOA)
z1<mM5T7`7&@Y+(;&MY~BNYkX%*o`-0K?KyUT^_ocTC}?)eD$RtFn{5hG@C0EpC}jy
zU8Dk4^DSkH)4}4b2V9_r&6!v(Jpm@obsafq;mK25gTY<6H&;P3WQGu?eU~S6PW(uS
zwe2`MY!5Z^+wrn)-7i+B;ONaY5AAvHSG(Q2aY*L9{~UVZ7lQuH7lOX{3qhBCA?PoC
zA?VjD_ckVZU85gP-2w5pQT)KZz`dDw$fYcAWMJhu@VjQ^IFbX#i`{2oN;taMH%okQ
zegM@5!;R5o8wa4~5}Gydd_l=9u_z6-z(GClp#Cx(sA8Zn1P&OCS>kKomychjzy2hs
zgP9^W9U;tnDz-Ew8rk(#al~NGQ(=Ogqc9f;BY&3*sPrQy*ZX{fr5}05E}RO52sQ>7
zjQ!y;)grBHj*Dh!k^Bo&gs*+Ao21?2rmI)*aVxnmYsv~0r&@7AH*19J-;@W$4%mZ|
z;|4gJ1fBt;8-mO_8Q)8my9>$jQI>9Wcy!FKa>!dQoX8S<zV02fL9C9BSr{L)so880
z`OcpQsd9lcM&x^}{Y_O(#=9wvgSa1Ft&KKBjc0QdF&wY=ZnYECkU}C&oP=k7r#FfV
zydh_<Kw(sgVKbF|fK(Zu3Tu_dN>;>DgCV+3v5jl~ki<(Bb?<{4-bU|%QW&+7*mooA
zKn>#q#!8ymO6mX}TW{_uq~Q`(mVIJnJat}Q*ocqi>yia*+|mZN#!__KdG8?(Il>%E
zO}>sgFvP1Jqr2#(`FFeb6KgS^MhKnn(zoN?;=d^V7u9x^cL|ILTa|5loH`lW><zr#
zzGS#Q%k&WtbrdPD1PqWcr#oG#0o`Dy?q(@}s?;=uiWkvS0u{UX6YEprS-HPsfe~-C
zmYX}y^yMX|r5A6`5JLf;C10J27^m)Y+I+lD(t4j|>anypBtw&z{8X9>*Ke^Ny9mSq
zdgdIeHFBn=3O}G^x16<5E5=y&n;fS%3G_b^^}W<y;}ZN#5^IdF!8&K$v%c;yO_~|B
zfx}8b?}?Pu1kTcZXn7zfW0jh9ktz%QN;3&Wbh$V>&<T#=R4>R)*P58Bu0`3PF#JG`
z_;p{Q#cSJ}C>g@DT3^z!^_ln;pC@{fn|jFCMqQ(6C6VZ}*JU3d;aQ;L|6s(jbXhRi
z3A$ZX&4xW{uov`z!Di5-2D?E<P42bbD^MWgFTzd+@)v=`#~XIn9D~~AI#Zi~|LAY-
zjEp;!1mwBe_w~K1Sm=%=@K?KUX>rtjwj0Bb(Y1DJWHm)Kl2#pRiDFF)(F}ofm$ki6
z2A#n2UIePC!ZZXy=-s<GJ8MGNbnyqKzi(R-yeuXdr$#q%rF<EtRW*b5Y2$#oUx(`_
ztT(M$$LX>auY9^N8xXM{O&o3Z2l|ySq;G$Df|GYN0rHptkqH+IgW$!)V|sl5|3scz
z;pFG2rYnya#*sC3wCsoE!BqOKE_`G1RlUj6^{iRB9<OiTYWjAtW*y$=hBc2kP%F%$
z5lY#MFu<Dn6c*6bUYU4-BqQDNrg@=hoLHzr5NO=7?0!|JdTA85-O%P0LYw3d@r@!D
zeB5$IYUHie-O#2c0Om`EYCy-xbbcLQmo0xhpDut7g#@$@BUKK$r17RcyRb+s^ob^u
zAqZGt1f7*MAFhACQ_(>?z~;D@BD)J2Mu4Y(sF$DeWgq`WDW6?)LXdirl+k)P%4nEN
zAMb%C%|#P^i`3luFqRiZW<(~QJ`ih1ofV2Pqbm4PJ<YOnHVMDUw+k%u=;>7JX^16<
zOEx`iRZk;&w3?=3n>4iL!iG6&AVJiaYfIfUGzR9=W64qJ*vvA2JYx?38Fc{lq1_-o
zzDbK3eR5RGh@D1y$M1>sye_0CwL6OoOc|g%&6hqCzvQF4`#2+3I^s@QqR!s<zZSn~
zlyNMaj7kkvo=G$fAJ!ZTy2D_K)lmAFz2Q44*xx7kcTa3rEWGw0!{8+i+p+0QfbE2h
z9P;mh>OLSfo}s$pE>!oGuYu}t@Wbr*WkVooeL&l9!Hhn>qtC)`tB!W~O$(`12x1P8
zitH7=DNzqZzG}infxf5je_P)pqp8^gg$49@AqN)Tbnp|-^gZz3Uh&hpfP$7#1EZu5
zV~+D#I^GBFOc`0Fswg)OE?)8i20B=KpNr5wI4*Hb>ZjFlLZp;NSDM*qYVtJn&IGof
z)kMoqQSFe;@Gg9*MQMvvad=&vn3h>oQIAjgtOJg{*EzN^v@zhs0t7|0!)7hrH+eSx
z8MqE;d8$<^+}<kct&I;@l4UJ4Aa+KN!3ZaAsv~CWMV&D0x*HXb&tz$(M#W@3W%4n~
z20dA8PdS|wm!_i7Xt-XxFXL>(gI8*}P~-URDJmLuMpQ^dx0*MJ-yl8FGn_8FUj{t^
z>8fQ^PB_^j=941CSrEc0OE<{NH0oBKHA!A9;^nHfw-Zajsn4THbG@_t$iC7bV}&S)
z5q`jT*ow09GB|<dC-}<)w~FI?ha$}N?RoBFf_TIt<g{J$9fLVw$&uYCeJqvT56zmC
zgx%=lf1)4nqVFK#^uqENp<^D*OsWs0!I*G1&mQFC52ug+0`~P7*u~b9mT^>>^CrPP
zv%!wdkG-!S02~R2!|f^ZXm0>USMp3-JoO`Ap25%GEta)p>Z|TXlxOf%pC}9ex$Rmu
z$=wp^)StOVI<aVInrjN15zKcVXL@O(yIEbWK|)Y~U<ULErUUm(tx%hrns-bZyu+BA
zGEM8PC8BALin;}qE6>rB9xOR~)o}MlEfpr5=DSau_)>i)9^@0IhiS`yKJ*EXQ8G6w
zrS|`N`ZOGb?wIgWke+mza%SN#O%GTBd*{dZBu+U2af-SG!uVtFZRwGx+-Jgvk(I%%
zurA6pwJ(%uf1^zMX@0bQ=_Ju3?BH4grGOt$0Z71aHxZ(ALepMpkw5-BMtBXgaW1i6
z%5pccU&?Yiv0utrR))Yd;o-7&0;uR2LrkC~*;)vln4uXn@NTERr$^P{>z;+M2&398
z6GjC`iY?Qm5-vl6)jX5-QWI4Q#W{BVVt`h|4MVLBf97}RO5H>!qqmw7&Ng1l(GZap
zW7XhF16H&_`_s0c@{!Rx>q<EFtv*^k5EoCR?m0*5jDVo5Aoib~jl(9|I?aQ#uwK`l
z&PB)?ZFJM8b6|juSl9dzPIC0b;#*qjrc4AjAc}9f+aHU~K{vaI+O60xNPNU{wHy#E
zSHCHL`LkTz-I5H{VD~joPC{E>8hs5hanwF8AeKev3jPV~V{3xZcB4|Se3HrrBEkSU
zog^Y=jToYEuAhMcEf7l_+k2FK1JM$3w%A>LJFrcM)c>JfLa1qXzDQKnUP#oWZX_zb
zPJAZLq*>=9<fv5FQaR?E>7@ddI{z@EQhbUuiCXt9N1}>!31L)k3bV$i-aA2=aJ~~)
zQ<aC4b<?Pr!&m<SJu&;MXt;>UtI!HB^lqRZk+(W)Ydh<VZ{*91vcCrOq<Bc;MH%JN
zq9F8yDS^^Fp8AOqlR$(!#q5{NH}}1}$!~Z2t7?r_$_&1~zLLKz0<MdF#ya+#Z~j+D
zQX~FRWI=c9)6d~Z(pym$rQl8DLubzyGiOVEEY1W>19bcmGf~`LG?g2}^*^#1_A)b}
zMe4>Jv*mMU87lymXi#qaP!Ilf5Bjr<yG$kIX+8~?7~!NNK5d_gKCNk=tEY<I{Qo2$
z8R^*~{ZE?oNME%Vdh=}U`+L%xQusj#Op}&cR7>qa1|irAi(RYbJ-w{0<Lmy)fp(at
zM@iT^k8io`B)L*xzg$n=i-76eD<KJAcQNy|eII_rBEE5OcGkp&Z4>+L+}-h19QUyB
z1Gs3gwOVyIE~k;M2yvQdpp~fU;`p8I+_2h24NbZTpVwv|+$~>}z%$`$$JYp+{YZds
z6@L_^0HY*vlKJNIE8WmdVGnd;Tib&b0~b2W>YtVv<qsX&z#w%YnWkgABbhA)K9VU-
zBbi@0EOfxry5wuTpui-MHVUo&vay2qAJW?uUN4{7?~YY_ev<HYir=(NBZ<xP$tq(a
z?M;TL(lDP7rUe}FV=r6zCR%dEcqMgV2d4uHH;mCY-uHdJu|Gk=*F3$C=9AMgvu6gM
z37pOs*T)pQOMcw``#HvvD0~k^)odOSC6W*h)FFrzCL!)pPytH*OVH}GD{hEfQ%xP`
z+IrRN3D=mHNU*IAeP%OtJ-ST|Vj(Yw-soMBlhAsRC&*VU@%iw$L=V0az;OMk?}-4p
z79}cAr8cA~6A=O`7b-$Q<)oWha>uW0|0cY&-m6g8CRwaf2LC5qsL8DRS#XoVKbm}9
zvnmk6oSSlEAdL}VnKV}tF9eY%Bhj*0Vi9}uw4Glheh!T0DKs1B)7G?t6cgz<&4Jl^
z9*j-d3zHwz%NOdaO|s|iwS6t~Y?}QSEn;o1R+~?WG0;23>K5&z&lFYrIp#=eY(8(w
zo-5MVL#;iTZBadk8#XdSwz%04vH)G02u#H4Ts3LN0Pzy<@U=(62+Ct^yi}+WS-Ks%
zkNFv%$Yq)Qdy85yopf8}*EB4Tq<y`cbV&%x>vgw3AO!Mno&0qBGUFF}|NIBNnHjjF
zfBUD0oB8ny4wxU$!wG5L*kS@u`VcqhShvhc>@?`ws#5cqzOD@BaVcG9LdgoT!Joec
z%wLH;%wGw5l#4}xm@Leu8r}>XB2NyN*MXD|i#i5tQL<+j?YjxBK}34;XEVdfUAmtV
zvvH*wjVi9#&Ee_quxx|58*ca=0|8bH23|&?C#b5q4fS$*oC#%8?xkXXQ#rZJ9qC3d
z!Txh>9nIK%9L%$D>U>dW`ZJ4$qRw4T`)g72pw;*jm!RQgZTBA49VxyQ2KoQd2`z2o
z!IvGgG*GsvX4}0mWg3ubP27l~8wC@Wk>StgTB*%tf38*8?Dpqcbu{E0I=do`<+r4T
zSEz;O7C;AcNy|j<^*-+v7?gsSyDIi?yclzIS=;JJ_ZQ32i{0(N*e%EKVz&g&N9r-J
z!>Y%k{ZAhrPARYczuSScGCFY0fplQD-+@Eafg3TOHw+49dEHlRmfysIS?&=KIq$d3
zax=1DkcRrPgZhPo8ltuEGoUcCKaV}5?H@<{ll`Bjt#1*+ckCMe-qA$=BD20%g*8pG
z*@|&op1yXDH*g$g_<X+_Yj&Yk>(;91z$ljZPqBaK79#E<wmux|HgGTzfiHJmL*-$-
zi`cq>GRgnlu8F+ZBmNqkioW_y-{#j1Q<L%tpvkNKI@lmrM^h81>t@Z@yY&n#q<_wm
ze~<1*Scqm#S)=2ytSQep8-Bgxwh1b%nldk%%;k;FLc|rvle~<=NKB>4QwGRZKP184
zhugoglMss5*qHV4F`u%JZ>$6?59#vTG1W|cQ8}Fvt`V)80R@WTwl7W&IV_@Yp!@}t
ztBp9lXy^#xaJIW-0es`QL=wEZR-CB7$!ZV5tFCp2PA73{7muvO*iNWM30RFu++~$E
z&F=!7VSAsOD`&=1n8GXa4u0=ToQ%G_@bGsjc*$5f`39qEaez9RJdLGY@60jD)2zPK
z>aV2!DQ7rJUP<&)_Kkv@ZN0~<32j}YDrQl^%7bFZ676IS_=fJ18aJAnk{e5n%VkT5
zYH4zX;Jnn8)zQ@H%tD{BO`Sm~qqA(wU!j=$hNirv-5}J%cGj^})J}!hD3jv3N#g4h
z))ANdPK#Qd*6=>aF$NZe?CAZZxoxz8ysv0HM`(xRGNd+}WfyKTJAd<}lG}?Oq(Uz>
zhdbM^=7CNg7eCXYcCqKfW`V7nxE2k)9t*uJrA3pjLEA}mp|{k;Pzy41t+h+=qmDr7
zXy_QwIo5{HUxjY>5jO#f+`{gBig1pG>kkoSOt>d%h~`)qn~@RP^c)+8P@6j4sYh96
zK{pB-jSXrGF!$N+D~jrF2i~Sja>V!F&B&wsp1QNxAUCMtR)cgM!=y(9keW0)qpQ3|
zz|Y#7>$Rv~t7BS#@n4^M5kx&eBco%U-$ffauIF$r9jOMj%VHP!4)~7Nz-_R0T;d~b
z2gd@m77=bOp;v}{izkZShi%|OzI~JdR|_7;PPIgRmhR`M9_T|^&p29-d6OI)5&|M`
zvX|3K=TW7gPg-eJYkYB~n~Ec;;TAV6TtDAF-(prfb`kfDjJqISRjrZNQc3pgGci-E
zR0o**HjfmI)47-t7OP#l$rh^SgT0S#2i$$2n~F*ee5JsLQM4R6cH4cfchsxHy2ZW@
z=i_b6{C5M@kJCl>iOjg*j{+Fom>nA$HLu%%v?1dnbexydIC3&xC+_or#eB#%y9gTN
z-0@ZA0&DTBr*drnzwEsWd{ouB_n(1@1RI@+28~Cov5jq_)P~l!q|!E#0cUhZu}y1Y
z(WZoYXiF6#v8aH7B+7Ic#CoY$immOzUex;93Tichj0O}1M8z8hP)rCaph7?(|L^Zv
zdnO6m)4u1t@BjV(KcANmnZ5Vg&t7}&b$RaVc^2{G^dX*~7XeomtHHfuK!BdC$`}z<
zq7`0a*VuCGsJ-8gzfCug?MHUU7(6q=Y8wU>=(EW&inU5|6U$2P4-*z1?!_UD3<BY`
z<v<6Le`6@MS{ABG<1)P*atTpU+n+!}jM-lzh`V31OoN7ir)O}SJ|eaDf_HGOM~mL2
zz8@fus8sj6hU!Hw-j_a+ok54`v+eVe`GfoFFa=9Se*=RNz4ilsE3cLB%FC9I=%cC?
zoJZ%n{jkqht;CP@mCT)O)s=JcrS&o!=}CFd4!lnZ_lXvNt&0|4{k#=yt{y(uvWK{Q
zNZZgZ<4@ZXXs#EZgzw8Z5$vhq<mq#b0maA_PWtB}n5XlHQfKZ$8n@Q!+f|)=i`6GD
zp69E&#AKJfRR0R@FgsGKE`BQYNL7nAezZ;-Vk-0r3)zj2Qj2Z{n?F=EpYD|V%ioSS
z-p?!eDCmLAD0X;Vr0`XLA6?A;Qnu*=JaPb1pK1eUDloJaQh96kK*K9F`(Tm0%sfQ+
za@LszM{&PGNh`mmX=Ag~L9?daT4s)_R#H<b5bLfmahK;D)MN`Zem$F&ur3sTk8MHt
zkMDU=39<3JtF!dj2+J+n*E1vL+ll>(4aofF<ytNz!k2AVC>?FZ80DHW_d4p72+H(+
zQi^@e@U*^C&qr^0lirHnvcp|;+68HlbZ5I=C|5HVJM2PPnz>+7<>lVu&U@^JWyXx&
zvfnN&DJG+*igWR+W*haI`WxDoS1j9+dj7)Isl}EX;8|&9*N$y|b9PKK16gaiMbPP+
zzuT)PJh&KJg?@G3V&^S(-eKn*oDYAH)8P*)Hcqr`>$H_)%ZLx>sCXcGyM6yn#zIKa
zD&+eMr~S~DGpF`Wt;#ActX{oo5^vgGJ>@~we!6OR%~0|FJQb6T*WMQ{+avp;=Qjc~
zSxRR^Y3e?wJ};!tS&@C;v#T24^L=^wx7cQJ@}pQ#xQBVQW=ml1JQ^$Era?!TCuLhM
zXPyY@gzfm4hx-pI)KwbONoiWHfLFrq=2pHjT(&FFI9*fBtESMarr4{d)Ie2@;(r+x
z@x3B5PT4x9AUD}%7K*<cDtT5l>{&;geoJ-!6@4EcMo0!Qdc?P(-}forOYXe3yQfv<
zUQc+Jyxi-x`K<TZ`mJFStg1)Sv$Bm3)R;3Xmq&dcdG<twgo!4|uSPztU+ho~?m%Iu
zEBwF}>OkRkSJ>kUb)b;v^tdwg^0+1HjV5@2M*Tu%_6r=~f>UT$;4H#xGy%8@72%aa
z3-p?saAk_~N?ESLwnodIb@_2pfAC!4G>gx2vp9KrFdZKN`@$`YnJoI%xu%bPb*@RI
zU-p~d!YOm2aXIsMvp0VoM<-;XnJI2Etm*1;w4xuPH-Tg)E?Js@#k!lB-!)`W28kFs
z=r{jdd$6Bd8z$M?N9gUYnJQ*~x;1TtwI@=x?$G%iXsXvM(L)zv?*Lze350zt-%nSy
z?j=W`+P9#EtZLHs$zb$h`XgBRFfR&Dc$hv27CoH$pem{bURBBZdb4T*>+7|uA}u==
z(>QP41D#i7%Pvy3v!0X8atkz{4-xLV)Rdq%85bg?%7g={JUJ3ybFC;cMg^mfP-C$2
zkruuG5o!z;J(4<*7k!-KyvoNZ&YSQ!#d$@K6Yi^!94z&O99}ixT|Xobd)F5&SY!6u
z5!noiG*izVZZdlZ2W0m4&moQfzSyMku}33~mj<Ns%Dy`T|7!>TfP<gRHpOL;5`!Nm
z_$&v1c?SM62jAe}FE|<aKe_tfItutTM-Ir}HfOt09l6oNnh39?L-|0h#ubVkgK-;Y
zGC9KaxKfp{I>(LD$z)E-^<a+8xp5}`WpN%&Ol0Z&q=5KbR!1QxRtIX!t4g`B$HScP
zxG`sE@+^n~`H}|}LH}}|`7!UV;FYj2v`n$JxV9HARZHQ!3hc#B9(_P%M7&Z(Rq04Q
zHE9_k@oDMQW`q*|7EWfLGQ^*C?ir(TiI`MWT3>A2<7GE#W6Q#Qx>uWjyM@zVXYpo_
z8M!X9XIFRrQ=y`KUP2#XzUm&M2)?^AWvB!AAa39aRR4Jq{4h196nm(={#ad?gQ&$7
zQ!+JhQwz<^kCYMcwvHW_v9eCB0Q45u+ZBl8eollv3!-Hyvd!x%dGgf5m_8Jaik#tp
zEztZrr37nb<H3H;(keo~=$g8S7-JZ_Nn1TQaP28FdxP$;JBt(P)6`#q>wdshBG9;Z
zP@rj{aD6iW4mxPriP>3+x;2H(%xWh**2gN!-VZcQF__%H;d#)W6G`|#ux~02Cy%&7
zEvwD_OJAJcK$AF67pphELJ>}{pG|?KM&=vVpkKGiB>q3UUH#hlQ~_@Z2SPpJ#1M_U
z$pn0^^562V<PpJ-y`^r5y`kH0VSst&RvRHHwt29MsAH?6@#lEHCtQX~`w?1@wM;+Z
zz<E#eN8r%q;h`?D@DYl*Os0nC331SUxGJoOKTl#=)kI{IP!6(m2KaCTCSG&*VS>ab
zE*Uz}0lZ?<#sEG-$KOp8l_jE$Bz-i51n2(4rRK!#dqX%7-9Q`n<F*$m!x1(1w&@(y
zzOq+ljfj=KJ?rz)I5<HwM<Ek2)<lQy(+dJkCM78rbyv3@nA4`!+^5m%`V82?+Gf4K
z;C_I}#1L5UdpjW;gEYy}#5DyNITQz#ozIdVSSUGK8k3zrm2`NcWb_bzh|0AIzS1))
znyUU_D84Hef1#q6z^~gB!WI2K%I?+mh?m;G&`Z<@?`iIoOHeA5wV;Li>hBf}$bmAr
zFu4E_Knza|)`bLuL%(T-Rvj`Ih)I1h9s^@^XivDcvrw9>CDf=D-^sD=2pJz;r0$mP
zPmw9Y9#z&2kwrnGkYU2u)T4hV!>e!L?`7!?*64m~2iU|(yV0a$3N55OCL?wrN}AIM
zsB0E;%;alqTn-X*WwEl&f#lctkVx|EfK!ITxLM`&DhP7EC&Pj`(l!`<NQde>&GYAI
zWcCF$0}8^VJlMc3eLu@*1Ka6D`vpB~UNpWb(%2S^lrhhiM*>lHYqV#C(-LO*X2MZd
zBAO5yYiPT=tS+B(brExZMfXu2{lgnwJ;mH-KdZNIMAHttM09AUJvUl53Z6S4(SyL~
z#MvL(Vt2|o_hOy13${J%#-<L0->&u2+kN^%SN(_a7yCb0{}iET^x7SY9h`oyzFkRN
z8SlXTX)1G*r}D9!Z<M=TLJc!P&Gp80&Jfq52@on9QvX@!(}Eh;#WVv8$KtOu$Wz3X
z9w4tocP>+x0vFg<mYu>V;dX_2SgSkvdzQNhfhzF5AHfPnr-<-G2YWe6WE`_tCvw<)
zff&-7nK=wz6VEVM$PM#2l9D7W17BA!3Q#TVVT8f)nfU-y2tx21pq-gSYIF2XYX`(S
zdQN)kDL4yjrKa=L90fsALs=+Q>6Mg5`l2J_)d!lJf#l<`JPIC*F=??1<{uCbw1gk{
zjIyFMU-Gp<#+QUSz?V$>G<?Gx_>8VvJ`})U2JlD+c%lRRt^nr>@aqon?a!ube9LDw
z?&d&^qXgI}KrfZz;~Bum9N=mPc)S3c1gJ6O)i-AVZ*+k7IKbH$G?rZ_KrfZzlnmfc
z9N^UsaGC(;3DAqgI3@#F;Q+tw0MEx!XjuZlZgf!DCuE?GcTgi7RGF%qFQ~JD+Vh$8
z_kNgV-+P#Y`m~^ug2FtIy*2~&oP+A%K(7oI)B>Qe#*7^*{xACw{%k3okT?w5**r6c
z3CMC&mY_2(gj0B%0!0gTaX^O-)$OMVG|H_JT*S)G!Q}C6&b+L<KNm#f)!L?bC#OnU
z7AqTr)#h<QCobjcr=Z;qt`b8Ud*BGVq4v!!1kVrL-wly!<u7H)yVr38>>5ct+D`kR
zi54{<WUc1*BS7c-#m<g1N)LWhMznF|LSA&R*hP>o!CH;0tl<PK;pjK#23Ez1fstYW
ziwf{zPV82Bxc0mpR;L#8)9fZ3+K)Cz65rt|R#~+6-~Dw%j8HoU0pJey2t~`<BSbz0
zRR)p@=P$-RhrN#bw5;RHwGpQyT)R44%kDkNvEb_BW<mxyM>_FDi}_Hsr!=tO`~Iw~
zvKM6fEtvHk<Kl#?iz6ki=0!i>@zo&p2A+wK{4Sa66J8cZ2`Bx{H9AsNP)3D?Ty)(_
zGy3E)44dJh8w|S^>0DVMBJ_D5ItI%*blhDC{SL)D9KX>Ir2cge>k46;_Y~XBIWhMM
z$f;D!b+FBf3M~N4&vX?%h_9u19&xVa$o+O}ssT@i-#MlR<_op~ckjOTMiQ1dIYRq{
zwr}N&BEFtjBB8m`_M2G+e&2b?XkJY;aji<wRGEWqM*l$!<6v~&;<n|pmPAQi(mIpW
zB|1>p;tEmdsZa+BJ6vIhE7U>i1)k20t0*^~4$<~|aT&fIu@^akX8EJBBVC>hrGe&O
zNhmyY{-PILp?H;(6pzb6O0-@=tA%wJbF5p{Z+M3F4bPJzrw;#T#)IHuX$mU#YDV3#
z%(xxZum3V5^%VU)6FOxbiTzvzM8h;5RUhfZ=v^3G&ao-lU`@8s3G>TfQGy~R;(L*p
zXe^Jf*F+QnJgRs>`G=1k`HoUwy7onSQ}hY}n_|Rl!tQ_s^K_u;FmYeN!=o6P6V=n?
z^cX*q`1Bjy9qhu5C4QS3;34NFL#Nk7lfws(>?Pg|qx7g7&!vlPl<YvE8_%U0$h@c>
zD0Jhw)QzVdg1L{{crLZ^Tq=<}T=w3y-&=AgNj>vu7>cnQIhYS8Z$4TYC>$pJqhqL!
z`{$=sdn#!UhY7TR!zA9xVFYifUs6fqI82~99465K4kPT<x<yNMi-T@)7-8*^3V`-F
zJ|QiL;ePaMQ#OD@q9t#{d>@>b^xt`2a`^30->dACar_+`qhRhGZF8@@H}_VBE6%-@
z9n56ftlw(+E6ZL8B%S%j<s<i2P2zv3w~Y<((9LqtO)UPr<c9`sMM|EJB~Ql|tT9YR
z=hO)2sH4On+ejuu*HQ@L>HcE>_4}XB_dwH&xK#I>^Q=RMk|YToA^VFr-w}uoneNAA
zrn`1nW!8IUzB?o|RawwA@zMH$nLKnAhP26r2yd_V`=KlOyD-`Lm7rmGOB;0Dc02o6
z*1KlCJao`t+IJ5|^{S{6|K|E%OI!25SEwK-(6rw?mbOxx7YD$%Q7UXdiVHMFNYjPW
zX4oua73q6Pr$oZt#X3w_(xStpB^^49Skm?z=F`24b(nB(iw=|S?a*Pwy#VCt76;wp
zpj#Y9+#6|36~@ZCnNKWu&+Uh)RLP2TZnY&JMw7V@hLa;6MiVaoP|~HhEoc4^6rTC>
z7d*csq)gx1b&|J;bK=k9PaJOC)YE;0@|lb9H60<{ln7tWw4;zr`}ozy<vsQL@LVRD
zP<$8hmU>s!f7HFQv1e%Dy7!?>%bpnwANkwX>2yf;p5rq7&z`}Mvtftz3>IhecoCe<
z6Q>ww^V2LioA2%4-`m(j8qNo(f(IYZJouP<a5V=WY}SKq?!hnV!4r58_yrmG`40Xr
z2fycN;42;cAi=-$k-=Y_fuG>uf9BxV3;w$feqHbW-p;=}_|r1*BOLrU9Q-wc|DJ>Y
zjo_0G{$N4+!}ssD4?n`ee_!wyIQVITzsSKqn}L7E!N1Od@7;-$&$0_0{F#FPf`ea_
zfxpwi|JlL+N$?jr_@e}WV4r>Or5X4s4t|zb|53nCbnu%G0>9e9e=`F=+QG*h{A$71
zI`{_#f0u(lG6SFM;7@SyFMJ00)ee5H;C}|ZG|HLv-7kLsobBQU%!4SL!!&Tw_kSZL
zRh-QUR|J(0^6aGluRwvPPy<-l`#Au>_d<%q7ApRN-B}Z8{&y*}&{V<*wuH;Zg(2&L
zH8(tThl#vs@`%yz4<?Sje{G<7qCrP^%jGbWbx=B(+GujvBS6Gz^M5G?%TTBD%QKyy
zbagp(!@5d%P{QdhD+n|nr<a(X@_ZWPxFW1qqzvno9xicyA;fWAwCsEu^bRGlGLq>b
zumze=)stl_0?nV{1bdgOE0#Fh3;O>lisjji_w?04R?7=K&1nIrQUP*W$mtXAw3yR}
zIPK@tr2g$=z{S9b2X<KiR*(gtJ(C5XR=SQYA>Woz8Cij2FISYkHtX?7qETc8lu{^J
zGZ?Pj28;S<Hykjv74i3k`nD6{qa#*}-z_pKf%+#EjZf{*L#9^T=x|_x8M$j$h`7){
zozaqZ3uGvDU&)%NuU$9b|KK$cDYnOw^)+%4X1s}V_<Z;0HGTT#5BN$H4sTTD9TAc)
z4>aW&X0xPSk;H>)Ge)+$GewL<5pvAc-JKVp+&4pm%)IaQ><5X`$khHI?Iqp~ZFi!!
z3&PCo`<j_VGV?dJzB;r}ihup;T<U-Q>Rd(u{jRU}FXR-jCbQ4%BOWjT2+;#+)RSdG
z-OfHRs&Lx(IPLd+Uj6H<qgH37)izwUx!0-~tAo92xYZvf{uK5^bOFNOlc5b8M6-PF
z1+?EG^-8t>SJ+$pzk=VzySM(`u4TrBc)c}Oy;V{6QXm-t6-VOTF>)i;z8S84RfAOa
zRG_IE1TQ^KW~|82`L2d17%6$tlpCh9OkT#Y;l<ls^bf|1IE-Xsdo9pJHnrY<!}RU{
zbeO!*dSeNpK-@ejO3ObsVi~>%8suD4<7CDDh{&Cc-xmo_dr1Eh$(jBou=4t9MQ>hT
zJzVjYLl)*)#q_VmTOMxlmNW76MyY?v@77ez0G^%9C0u5?_3s?@?;I%*NiO$pm*r1V
z&s_y`6sc_s*LH_WwvZu^gp(29yJ4dB8>3Y>MZH(R!s`*2mBUF#cn_@LJElBL^W(Vv
z3kqE*HpK3_FL^w7zaD&5idWKtGY2dVYQb(zYv!6w<?hF#CHSV2#By+8?%QU90;7aW
zKE$hTuu~)&(!z6~5@{k?)NBi!=W|#+Ci9j@lcR68QnG_PVFTAh<nJ^99vT-T#%6Ym
ziTknzU%-h_wCyhq@@aDRJ-**~;C&n;PgY}Uf1v4XYuusB`14`DKgfL64(3r&9~(ft
z?=1bW|95`8Urr9Aes<3l^nb4Bga7<Fd*f8^$#DNCAOFvv3^aW~n27e4dru>s^|iNQ
zz|&odH_>llbs^e?OVIey@sOhR;bh&hHDTYr?$3S<|G}my4ZLrkt*4{0Ey}8)89$zr
z0qalr1VbhB9YZDa9YZDa{c+8AvtO!MR!pI;;by-iT@m;s{uDNvx*HY>e>|f3{fKsn
zlpSDxD{C`6*x|u6&y(}6@jS<IfQPvUUzlYt!vmx>Y?}gUz&GA8z#SX_oQYX-*((5=
zhV9V|)G`OP%0Zom32xbIf|7>q<_y%04(c8U)uflcDky2#F3Uju%t6g|P+ygXE%#L8
zL$@$_I-lL?*poi`GJro>!kjn$<nz!~gbtE!gvFPfaXE^zcksE1CC+2_!#OC*-ry>d
zI1g>yIS}@>cJBv}@$aK{TcLXdi%N_sO$>%BUb>C%Kf_NRq6WLOKb*K)yCI(B-so=b
z)#lGdp%ERr#`HW#aqcl7Pm!23$}^t*0uJP<K=T<Cgi(YsY4(Q8dIHU-P!LW0RI3xM
z7WP88S(t@ur0hFVr-`#hg<H1{Qs(#O1AY};pr?tyPn)(=z>{kuWf#D(Y~+f}KPP{m
zaARX87~Trk{<Nm?>PcC3hgBr}McujjMJL%LUGBZ$_TG9WYEj<}<UJy_<mIVAY{FLP
zY{oOvs=cXc)!yWa#E`bu*_WY>ByV~w@jWG5T03#qfl=Se-ucGhtkj2BtR#JXYwE!C
z9fcEj9w?sHQ5a30a7owTETo;;xhql!E*p01mcdyQcfD17g-dk8)5SB}+qV2{(8`vH
zyFM%~Z>dPkF0KH#wQZSI*6K^`rM|sW+uzj*9f>Vzj1*oiiKFR<G1>t#MtE$D7)$^2
zZLND}v$wT7n*7G~yf#)l`d7v(ni-BvQ^9aNzo_mUjo6-Ub{m*hG?I8&gqDvX&cyvI
z`Fol_Sf6%plv4+<xwEJN5K>x0aJgYa-t|>3%v6P2TqJN-OSrLx%uXFUBCw6u{f(Ou
z$2PC!G#UCPDV+|Iwu!2J?M#K(Q{=5m%EzH`bP+1eNu0%;&LEr^T_hXAjcS+6^fVss
z4xeinrqtdSClrO-w(LCydh1ih(i4T<xmAp>IH_Qu^|C=GL#NmoI)%_FxL?E5Y2wD4
z>{EcVvQjaEHMUR2W9RzTo8r7;0>nmPesY*F0!^|~^z}>~X=gAOMfU9L9(u__^$V<b
zw3fGE*CIP1H%r#o2xRQG*s!`k<>Yxm?+o8c5=ur?WY8j%@Kxbj9(|sk;zPM52L}C0
zor!Be<H5pdUsRi5G+&PSx?=Lq_EagpbusOEUminyJ7=$rS>rI~KJkg7iK3#hB_uFC
z5Z?2GJVG1a98xzXmdrh#I?hX;zD&Dt6toXT=i2j<xm-KXEwcQ1`ky&hUh2Q0ZO)am
z{kLGLi4h@mdyE($jR%GV<|=<&IR0U@<UqvtYRuP-^5{{QqR-btyivHFrE;<<waOg-
z0P`MN_@*=gp&Q?{i%xfe*B*D+k)S(x?QtQ0J}&Uu<3j#?T*wtPv{3$hT;R6Hh5Y%r
z*q_>asULkFExl3xdSWhb=o&#2`_rpfa&%A(Od;G~ELn8C`n140^ch$I;!n?+eOc(?
z#X3YEZqcFg;SL>!Kg{nnp@($qn&?Bibxq|%x^>O)hcJ=h_q3<JHG(y~5p!e~xRf_{
z{h6ubO%X633nrCbFw79K{QI}>?@b-N%&oPqGn_}xw}z%E)HmKX3Uw|ADAeNz2WKri
z4tkL$WyfRy^Bv&H4luhHzybh`l<mk#SNHB#tFD&=)jce2z%)ULl&#7@wL7Sn9n`GB
zKvf7zr0kvy)NdTr|8P)$&>P19H4;bQpLY&>4{Daxy}!2~{c79}r%nWB92nkD9F`ov
zuW{edKx2aC?nFd31jyQvm@ectU1)B4p?-?lroP}KCUiyOJ3)9toE`iPkfy&tDwyEQ
zE8-tA)jj)g*vW`z*N$GwYP@<>WUz<tt{BDjp2*<83BQP^A%{ps?|LK*LUqMpBvM_!
z;vLEH2OIYd3e5d6bVEesYRXJAG!cZBZl{tNM-%6WRX#@>OV5!+eNORMxi5U%8agJ(
z!wbe4oMAMqKBFQ|EgHep<RQJW@s;t`#!V+;%_f$42V85ZgqECGaw?&Udhw}=PRh>e
z9^QwaXl%rHP$mna@-@B1x13iy;t^iI4e!h8Lq@*Z^&85}f#dAth*MGH+P{pvh=Ba%
z&_!!>h+YJfHLg9?dBM7g>t>bwGQY_#aoDy?|9l6uLx%|$W!X)_^xDm$`F6?HTf5A?
zQ_AMbMT>O^Eo#wW(xMI>au<agSKx&9YJC&3Q<yGl8b#ldDb?h`ionBDF3MsB2mOE2
zI`-chX0v|T2K0lHqCMroQQ^k6Q3I(rgwcqgkweXsF~!X-vtOd0A$SOb&Ff~bQm5-8
z$0I9|1hXJ<pHV(HPfYz<5now>U9BCe5>p0YC`qKGElpJlqRCk`Y0|-lZ#Gp<Avr-o
zT0e^%xd+r5Lltge1kSBJ{SS->colS5TA|FN1jSRb@020KXvv1~HyRkEK;uLFDrf;(
zZVCywz2~JU`-REJU3YxipHw|_M`~YQ=q_ABR)k44ihp$PkZIT221@YVRl(4rHywbx
zkpFY&E^}rMEt3Cp=q__+4lR=ZbLg&I2PXgL&|N`yAqVKtUHR@p4$z^y3UuLLWas1!
zU4aVe<@$wPpJG%@N|V!jFN%O*Xz5~pv5@dvxs*8-9Bu}Q&=WQtqEFa#sC>ev!|*4Z
zPvZXils4XiD+|Gudxb0iB5IPzKGUDZR^^)wTb<7VY&Db}kIOzk!kFd1Z3DgBy>)-@
z!h!gOt!o6si^O*m;m-`yP4b<#_4@~OR~5X@&+DF&9xGA7raf7hZx=J2xUwL8#?>&+
z^X&=`zcZ%8IM24L3gUo{fpMPduAJs>F6^Jv-2IW_XrfvOrUixF&<=57Bb>r+n{$S`
zQzH!r_M!NsCS7&a^HjP)a&)5!aPn8vpf6c{I2xWYh6WYcOVbTH#_mPn<j<i&Id;#f
z5{K>FXbnQoH^zW|Jc?EE6x`?xwX}8wo80gZvj){&Ex$Axld7BDH4JF6CDyj)jWFc$
zWMm4z5iQ#gXkM)X2?J#n@ZV3^Z3Ks--M9PfRp;2Z$7=oVqcOS518k1>(J+Ut?y74Y
zk7NJUn=$lzuYwEhjFxo<niB(FHOp?FgNzb6BNstHuSSf}S3QS9w6TvR3p8B`rsdjr
zRMp)ygMzLkPw3060nOjD`o`$Xp2Nl-XdbHq`?4{7+1FG(DQQKbUg^xA<jbnqqh0E~
zy@945nNV_W5Z<nKO}Okk4AKjLi`N|#Xx_o8Brm?>yr6z0B%J5S!KpkcvOoYo#G31c
zdf>^yLPm#7&~lgvQUDSn;Li|840=Ys8clvBb}{Jx-}>n%gadppM$0gQu7M&&Ye(P4
z_Ye;3DU3#qJvr0PJp|wXX4aSemkdPK*S5{6t0JlCoSAw4|A;iU5xjg$J!#ZmBa(96
zqJhtk3p_mH5=(-5AHVh#NculymMZJ0-_$jQ{^?_I0?kOYOrJ63nK2zoK{WAw34hpq
z&@E$vfs#u^t2@HPGt9XZbzkg>COr8vD@h1*Cl3*&Zbw@3Jc4r@VrFvpP@Thi!*-WF
z4`=l}Odq1i<{cw<cV0m<8@<D;|M&0}k;J(uj1eid<R@t~oS;IwQSO~p!0XNm>PJ%5
zS<sTeg3EYJWB2d&_^(VbftsMzb7Tb%V6$^Ng8sPfJ5mDUo8*ES2`S-Xflin_Mqr-S
zMNXt-ML0PQVZ~n*ld3W~&O{TL{$k3>wXgf%j6TX9=7YV4-ovZB6li)yGaZUidUoyb
z#q`evM4;q|PaB>2qPD8yDJ*A67~CggzJt-S4+D*#=NVZMDGo+_8>7hCjYm@kixF=R
zmGs~hhAuFskj0Q51(5CI7w(3>{5{LYeyO$(aG}_y(4jX`;UNTED1xBO0M=gFi$jmH
z|ABoBk9PKZjIue?szOU`vqFel2Dn`ip{2H2A@nGlA@uYPS7^HxbY+DfxWXV;d7(#b
z_eAK?9#@#}3T^j9=+XVIu)r1C?g{^+u57r<{Erg+g*i_a>~m>**KHCJ(*6cvW{m~>
z*zL(bch3xl9*`w87=1vN&|u{QvV;bQKfwB_3XvL{#iZ2Ob4Y(pYV0|as+1c0ZulU3
zEYCfb=N`)&{(xkKSlRpazav1T=pU{<%~y#2<%#}%0-uvz3Zv^B^iO-$DB`y`KoLL3
zQd)Kv6tQcgO^WOx8Nh4@c#H!)fc0<LWC8xw0dCx%uJN^3tj2dZP~-ao3=43+1N>_S
z@P9bKHV3#$fIk+X=Yuyt1NaLEc!vZ0yZ|Et^nCDYGk`yIfYTh{K`eI5ej-562k+|{
zz)=n`;sAY8z}+gqPA#!Sc3uWB-~bC9;HnR(ahw1>AH2?w(qFyxW&7$q9Qf+H1^Dj*
z^nCE1$N>J`0Y2*h-%%SY0W`7dwhYv-9MlpAwLwtd6_muP=^3cY9n@S0b+@3d6qLlO
zSO)4G2Q|S#eOXPr3MhQ=S~tP;{ONVm5N6hAKmKRVc5R?B{owQP6OCZDK=Urn`dQ!v
zaVQ^sDQw42K#DuWgq-mcSdKsrXN|LE*ZYD_z1<dYNN2(mjd4ZA!egUc7=-V+2c*!u
z+)l63>3lmSSyk5TCe8E$G(IzgEt>J}w%BWlOUrS*3r7=YbDYT08Osuus~a7|AjNq_
zQd0~ibT&bGCZSTU!?uK&6b{TyK{;&8Mly^ywC4|;%yl-wWd5JAc%Y$>=`Q(Q+??wU
z%aMagyCx31i*z2;L{HXb*$wR#0csp79xwe`a=IT6tq)Z98lJ*EZk*2LN*G41w1K0-
zGI9EV6|IqnONUMyHy;e|!GYgxz8;AmnFNR8?lB^;neoebF3AtwtbNWYO@4q_W2&@4
zZZXx#o`PPAh^}81nlN8|e5kz)m2forN^AtXbes_$j1f_H8+cUTve=He(5s~~!;-CP
zWO_exk8bMyaz%l(U@T}i`)AaLq@&js1e(5MJ!vLW?GrFnCT5n?2kAAcfx^`p_h8C)
zP$rcO!k0Et&=2#hWVS!37Io^8<jk-76+w#u>RkQ`H2Vn#fUY|LSGAloqR4b`AbgL8
z9FuG@Rv9(Ks>JC_RfdzpbL8=+7s@-jK}QQc>-nSmR-BtG`a7je6}PEg23X&ok-MYu
z&7EiSUi4d+ipt%M^5{`>e1?Zy(4n5@O;a^U?oA<|-&z|j*$ms|LMuy?B-g}V7>@%l
z@5g_c=hF0~;;|fiZ#vqUm;QsLQ>Zp;PRVPpg`<3=+c|c_!u6XUO<v&0Cl$i5G1Vi6
zP5o>W*MjqSF5gP0NxAKD!m&tFPt&6`C6cTxLZv@N*6#U5wks()x`f7#r?L9JQm7(Q
zoYSM7hPLoAmwQwW;P@C6&kjVA-}Ms!<1`*R-g~GJ@9$#QgE{GX6eFOJ5u_h4^)F5B
zogF%LKeTpGl+7*0fd!@B$d&WxU_H96bM*%LtH8=})PxH%Q)IDHcsA}N5*9nKclp=>
z6S!HaLSavOgmnrhu0w_1YQL-Fd)QbkPuSA%o;S`-`rn}ovruWS))76XH1JofLREj@
zvJ)0?M4eH6^55H$(c<rSe^%b-T4(IwwDlMBZIR>pD4TJCrj^*T40Q#|1HVj_=d7+f
z5;dU(*Xi0x{R`vI!mn;7Iin1R@dI3J(*y~ht8gJKsN#rkHPH=YW$)TU(@!K$DG><q
zu%Dh!O}5P9JG*{IUB*AD5rs6ho4;LlcOXALan}pQ<$EViTU}h<Dz4szLfJoV&BUW-
zmE4#+#4;*Q{PwI<ZuEc4V0#2bLCJ6Z-<-IM$NCC#znOYr{r1%A^&c>&dg|}dTqwjY
ztU)h`?+i4(!R)cu_svIwlSa@|3<jI4UrMv;*t4*m&)ypGy~cPEbc&02W4_mMtr+=^
z^s$$YD3Vt8XW$sRV!h%nmc3WMHB#~}fyTn|6nn&ABd7B&vp>g5$RwYsk>C<B-^RZ8
zODUoHuF5-2{fwq;M82Ka4Jk14Mz}kg)FT)JIPXYu#9>M)WT+^Z+Lt#1)}4nv5bmH8
z{(t>89dn9}uy$D(;tjH7)HE(*%n?$nG(OdXl+XI;`i_hciR_~H(>%;)FtMNo{W1r5
znC=5G3U9(Fr1`9mcBKLLA}9iQxC49$gWj^g3ee-T)@1<KIKWLDsPS~nr^}uYU?-E4
zSKpff{0|4X)BzqVz!m{|d{#pSaJB=y-U05z>ba~{fF7T9K?d-<4)78O_^JSx1898K
z$PCmM9aOo4Y8BK9L5a`$R0e9OgZeiIby_D-DM5+PdTVF8SGGK7z48GEKK6a|=F6T0
z3O=hEoYWawzaJc4o;4nRs{Y8-zF_u?D+Xs(bw$7DQ(xSbspk$?&!1d9llD_j@fWO~
zJk_)R`GNeJ)pxj}VpMfyxw_8J)ODV#tH#w;pt?r6x>7Gu*8|kmkKb$DKG~SPSD-MS
zv)V;EO<c@m!l0SVy^Ddv^$1l*h#QLVju8_twsPb~XIvdYx#yyWUgLYHLH<pUi8D(p
z5=Y^lb{FoHPSI0Lc`PvvJLrW{_B_eW!qrlA<=)WG*cFg?z7sjW?&z-F@RMflGO@pt
zE9%9DdNy=Orn2WFWw^bqvGQ+AgT5+Ob~Y_|Tzn|(9Tt`M1I;h$3jI;@Y5~e8t&J16
zwdCw#l49u!o~NLns(A7tEG`>wos(IsoXjHemOekTcHY|>^0kJ_R^n&2vZAbg*5hJ?
z3PbT#jhoqLu`gD;R(b=-P7npZ2u)64!4&w;v3Lc_MS_Ik=fIKSY&JGd#OAljSNU43
zWPd2WvT+kmC>RqXB~MqN<yReH+|9yJXHGFqzMzRUtCcATSb4N$HM#DNXk8AmLMb0k
zo^Wn7z|ze=3mdy!0;{@@CQg&0Co#fY2ATp!I!*OJmy81Swi{X=GW{4*T=o=i=)QO$
z{Wv#S7a+%jZaI0qb>a^uEWjYA1Ex3~Fv97u!yPcZ=?asZ4loH5e=vN3@&YLva*GlI
z6~Gn8!@j-d@4NX2!<~!oo_Zox!8%40KjuqusiSe7>2Q-qI`l~}bSs&pM8*pE&ra5z
z=1;wR!J7)Vd0bll2BBXYX>1u3ljiguV1>2o=Il?cx!@h*yFNK>c&hz^-4y24U2DSr
z>2p*(J=Oh9<tJJ-;r+HPKhE9lYuj?=3A?jWSEhDnq0~Bkx~e(-fz+xC52jwNdi&$g
zH}hz>9(~6ieMgUe*!Sp&2U6=U{E$btc2zM5X@1XZzLEZ@uS<h5KCA9qsP*lA*QHxM
z{c}|Rs;Yn0s(%$;IrV;)8pfQk%3r7QEo#?@IjQ$Ael4}-!dKy43btU`7;oO*>aQB@
z9_Ys}re3R_VvaQW)j3-EJpJk%-8|=nCa4}EkDC#mf6YtNR>JbYk1|L^8!<wI`9f_E
z7f$<G-@CMHA!p!x^4|Aab)CIqCbdtf_N#W+3>ELsQ!&a$?|ld4ABgXiUQmvMK{0R4
z^xx*{qQ5|)i`*pIg?|rQ&w7>OO$vli*{VSDB=m7^<+)YW8ZBAL^kdUdEdFe?<T>?H
z_Lm=L6%l>@RF2{+a=M2JpGVl2>Mfx_(^|}}gkiId>=7@_5n1h`@u!)qt>-4Ahhc>?
z3tEOBN9j{tc^gU}%w2Zi3f+&cbl?gVkgjy7&c96;`L~f%0TbBs(^_Nkk6_FTr+tSL
zxBfEqRZnm=OmdHout&XDxm$VeR^AA!Pn$5XH$P2g(?HY7OhGo=V86b5T6O*+&6@m0
z?9_;ry+Xbx#ig{K>kvE3Fc;WYH?*^w)RE$a<g~z`6gF-JPql_icCslU3dJDaTnSN(
zH-fNtBCnO!@EZCx`}xO=l`c496({1BbGnaZy%8-XDlfN<QN+d|nNWCROq{#>Xq0(%
z)s@#R)?vbREjmoPu0w|r*GXq)x7c)`TQl?Q7Ml*PWBo%7mg{@lm9q1;jJ}R9t<JyJ
zzV})NHU`l4uAl3BOW66_65g|BcdM_iR-XptvrQ_vOo@W+VBg7+H6wT1Y{}*$Y&=rL
zL;lO+)c!pC<y_O(emU3rvES<4zi`T426TDhvKObVRQNnR9r)l`dnIv|`>fjg9)-zZ
z^2u1(L}!aSn++gLFh;KpBWXh)fBRp2f4^#>?fNuTEByCg{2mZ4o(hAE)&!cz(V<<x
z?J*AYH=St~YVniCLj8pUEYxVsPs_%@t{BrcGXr&%gGxB4WATMr77~=0w(n)2Djn31
z9Mo9VHC9k!+DbD}r#Yx^IjB9mfV#mkNcTJogLM5W&wZ}1Kl|}ZbG9#G0E}fi@o+}M
z>Ho*Gqg0-b3NwOX#^^i_2F*5D@D+?Nwx?5oih54|E~}#5fg1^OC!DynkZwn=k(2*H
z<Dkmnpn@<---m5<98>_vSYj$1)Ob(~919ZwHtGv60}-oLir!E1@Gro_{{?C^TZc(z
zf#wv~vBZpmNTR$PQvt80wy=1kKiZud*)Z5g1E1G?8ztEDuvxlRJjk_+WOBZ>q8!lH
zLtmfotSs_wW<Nk(L6`RTx)yk4f#x5(ilOi6Z3ZxO7$48K^8T9)zNJ@|2^}<F$f>b|
z28GGISz9f{%|*|Mz|DQ#9+t>f_9Xj*xr&v&5onIsW7`AG<yH~~@~fPs{V{$n)-e8L
zwDtn=d3e%whfqfKF?rhSxh~{e2M&bETdQ<wV7yStOMwNWc{CK?(b&;3R&FK}F=vLI
zZ7$+`x5e<Gy*dz0+}qBah}O2pYPXIBqty}^)C;DKBHvIbE`0c5UuQg@w*SdN<Q1_3
znjNd6$^4Px^F~GXtc}Irj`=p}YBKRDF#1V9wvhVm>cnVr(#RTcw3K<p?8-eSnTxp-
zk2GQ;VdUvLJ2~Zq8s*FhlRw}%HF2smrB$#|UxXu&eyu@`^^|UGyps;KwEF_~i}K~<
zM~%f_v~cWR^zL3_VKJl0D&&PYem&-UKJ0r%rKkg;1kV%bO?x@v*>V)2MM}h2{M|s)
zKCGXTXYjt5XtCH~T~i$w;5WW55`SH$l1Vp%?`H<5+$^LXth*U}KQlP<=B&247cHG|
z)8e+d-(6aHQ%l?2*wW}t9c^>RF7@9OVWUE%Z22?=Q(8$_CE`t0wRC4+vQSlSnROgq
zl08+esgGv1=r2^ZL*8)ezd-U{H4J{CsR7HLzGl<_{%Bt?_x85smoM|*fhje@?y+~o
zhD;tZt<K?p-G{za-tK49+HJXav@QSfvgjRtibA*N@ZfEsJCtQDdV8+!NAJkBJ3)5`
z-2!*=-JN_cDsL~)oyt233~V^HQGw@9@kad(Q42o*Mumk=Q~PZgm^L^u9FVrML1oF)
zPb3e3f}G9cGFI|p#osw*MO?T&GwV&OroX}W@EgG>eab<7%3%V1%Ax2cdOa_CgN7on
z@&*k>-h>-86nRBAsJmljYwO`pR;xRTv+9R-E#(yp<sx75u0p5mF*ieY>eZdtkKe?u
z7R>#l=x^?Q?A47BDQ!Eu5Q*Q1t#R56{OS7c>pc2L7p?3*zq>NaU-Z5lcjW?<-IwdG
z%+?pZFX*n!$QQjY-(8t?FM3~ryE408^uFQfM=|Vm*2mSM(~qW0x@QmMS4EcFt*ab8
z=KF({)+xaW)+xbB)+xaeH>9>tSetrt!po`GCTvQrpU~wx&ic=R<~h*35t;t`OU&(~
zDbA|<bVhy&5=C&=gm&vA=R>)szEcehqi9`BCr>{dx2-LlFn$_;p^cZsHcZvX>!jw~
z<cMQUPl-ua1v3;Mr0iXGIbwMoA0@yC`vN{<7j>lUmFdUhI-u^?1EnrtY=(d3)9qfG
zAJO-R-1kS__gA{_AASS3szX}3b=6TV-6_?TTDp^}C$MyxHtsQ|6!$cxxHmIzc&0CY
z2OI6?lC=W|5V-AyW8=FjN?u_VNA|oxfD^LF45|A}zxl*6k<U@JpGBDGU*xUPE&`9D
zBm5RM*l_mIcMm=oUBnViKSlh_(Cu8~ie?LU5i6DweksxF4J~5XQo=7KwuBe4b}5l|
zFr;%}SiqF<ONkb9Xc4QJ5`HPM6}@P<mGDc6x2*A*xkUWZ4aRJPe;k4dG%lw3U58_m
z0#%+D#``=cTuT@Pg(lRdV8zEg9vwF9VBhx<dQbrxd&H2ltO9f8FBvXrPmdo=8v9|p
zXMh4{5n%?;f<0?|YgE;-)U<M8?@E&KvbP?rU4ml^hsK^E(;}xr1sd5&`%(8d2R!r{
zJ=7mxe)97}GtYmbXFWXIOdR!{&H5@>Yt)Yki3!Q<W7fm7-38B<Ru+CI19-jzoag}G
z!vMN$kpMkq;VBuwVh32}08<+P{Iz(+sl3;pfg0qXj&@M*KM&MmL5c6&_<Fi^uZcrX
z4ts|Kt-Db#y<1S?JC|ml{^p>ba!}VurMLqql!YVO;5(0Pk+QIXkvSwEJc6nSyJ)5Q
zzSQ}oJ(%nHq`gc+_B0RkI|>yTN3ap1ykh~|!tGdXxC%BeD2Ju_A1Q%5s+8gigqlF}
zbtlS`rmC_cQ9eNmmPl=RjaedPf@!@fSUyR=;mJ(S`7XyT^P`mT_gDeb@e;ytyz|W%
z7fnnpk0xe>BZ;YEEDK|an&MCbj_N#2qh2&_zF{&<Ybd@IAtM<&viG(~Vg@xlW@frr
zVsd#TF_|c3uwQFJe7Mv#A>SJOO=3^V@JPcH=v!8=w$`7VvXVtLY{7j9>f0_}**Rwg
z+n@+nrFa6d<~IWie)%apbzyo?qs3S4dF7;6Mz)}Ut6die5uhH^N;rA>U>r)rfe21Q
z4|!qIZ*t+i%Wksx;eiD)DzOEK@wvjzG1bzJgW=Z4>=DvDJY+43?`-Wl(YJH$UhO{c
zOoz3t$y{9XSMH{J{ehcEkpXeU|I%L(xU;1qXH96H7c{x|x$xTev?WP7)rtAQ8(7e!
zpmA~1A>S6^QWHW+L6iRT*d_l}b%t!toc|p5@GMMHVj0n=okJvy>5=5{F*50iF(6bM
zx3;Q-WZ@%z@|d)kg3pi)<K-NMhh#8+$n1Qw<S9t1V!|v@i#V>aql;eG^=m2~NFT*Z
zG7au2Qp{G{-0B|D)a=xIeVvNC)nTVy;<}~^(6>#@$_~{{x<F3#TA58<h+q=n++<OM
z3FIMj=Xf<r7PmVs7HrJ-kp@BGVX<>7XP@Rkpy>ulUGcY&P~KD1Tc}>&ucpx#_t>|n
zphxfJn^ggOq7))+GN%fOqc>>@R@p7Hn5#))_#8S%4HlEEZ_rriv!aZ>zLZ3=5v<5G
zyjTw~@a0`ns1NKfx}_Z8IAQn{<N@d~mliS3^*KsFb=Y`H*=C>)tFqH=>d_iIrAOU(
zDg6Fqib#^EA((1+KTwCYTPgYZ?z8~mF{SFmYKh%hiCC6$H_%jp1x-T2D4VRBHyda!
zlVniarbkX`A$bp*zL>+v6!=OomAzQ{CA-2bPr<Y+21aYox7s%M>{_mZHO!}*EW$oy
zuE}r+rcP!sl1Y=rYt>ZK7o#xyQ0EPti=s^S^#Gx+;liK#=$Gy6#GY?4+9&S3VrbA^
zW?8?RcL#$(Yc+H;cu2BMp~g=iu49AUbbX>Bmn^0^q3*J|VfS5+Y6Qh7Yh=~glJ;h3
zcr^$*inCQC!CYAP6|$YI29$MQu~2hXr`iWZ4|eDjBRkR8J<EMM3cFq)h)#p>V)p~o
z?p-(75re9p)Te1<5bw6i(Tz>E8HI%&?Pk0u88%BnIZ5z@GXwJ#Uyw9Wx6<N3^5-U%
zO5?zoHE>_ELoezJujazc$aFYE^F?1MJ`a*O->@7Cf7kn<T$eM!PzCwuig6iW7#hgs
z#mjIR`aOq?z6vOI$*JKFa8pEVyoAyMo@?z4Ji_J`rBV*^2o*EZooA%WMy1Pwoo_EQ
zIIJHA6zn_(An_#QB{vsHC$8auG?_{=@NjN0oczZ5U(Qygxz}@D%1Qq9XL44^+3@T2
zS!xg$k6AK|)bi@hTsVK8w#}D6y0vY~`MGbJZv9QH0M+X#T=tR*S6Si8ENb%?tKRE0
zj-tEX``dnd<uII<Fc=KWE#?)~tALus{<F<LeDoRBu+po+APThSdKLGUv8V6+TqILu
zKy1sH8(OkdmA_QY)GF9j-D16NOleUtb1in0UdA&JrP~VrQiJ3+p=K?lSy;?*8Bm}A
zAA5$12}ey}R^>m_LLXE=+Hu)|1?AG*OL`j|Xnc&`ANeZ&GxIpX5k!v6UFBG<4&k5G
z@5P1<99d3B8x6PsqRlPW9M#nc%c@A7VL8RV$^~Ix*3d2}>T0eeV;2gAHBcnnKhKWL
zM7caM=>7T0zL_NTh^7-%OL$KoDt)C0nCl;Z=lu_R^TzmbCf7z4Y`B02`euu!f*ZWJ
zmPFv;eEdkhkw4N+o?mH{ZT4JucG6$QV7@|UPbmsA9UeuMP@S>Aw5_;u*l@z;uX9sn
zjqWYCfo=gIcj>l2H$9Sw0r>Eek4zir*?S4eQjY}Yk7nJbC#Q!)au@*{LTH5No};Oc
zf3Bx%ft#V1%*gpVquMnCIYIrofd_@=3nkBaF-qyF9lR*BK4mDSchqT}4S(~x%S00j
z?zaL>EvDKeWrxq3#A0T=SZWqaLp|?0FR9L>v+^+<f91t5y575mg3zcz^jQA2nXW7F
zy6%U$C&Yd`IO|Bwbj(+M{r2PB?9<aXTC!4`eu#jGW7SE_-L}qG|9VIx4_6~?OuZKI
zLE%KEIdFMjIjetjivK}#`bM%}Q~HK5-Hgyl-t6Lg7()$74_~eUYHY*iwPL^GLNq>k
zFwneM1a(ipMf3)`eNFHsJfKLL3c}RMxsnnTYp~=J)?o!`ZYb~u=Iw{PZ%)tSW7G+U
z&SS!-wx|-h6DkvFnnm~IhlC3%iV~>Dp!Q)a^fA*<In9mL!wi+r`sTXKP<?!s|HHn)
z^~8T+*8z0=!VnpKf#Z8l+|zF>)||E((2G73yG`FxFEiDgqa0b(qzWY_S0QZ#BegGt
zG4;<+=pMwQ(72R}-6>?4yC}mTpO{>ksG1P=1Q>N)Tk9bLjJP3h-S-JCR9~*zcu#p)
zWV{flJ_%-OF>3PCP~yC_6f;AWfOVV{v-^{!m>Q&*8PxWeG0`T?w1=R8PMT>?OEW=w
zK~l_WmdCL6yFDr9QKqvPp(VUYrnv>rFpJ_Zr)8LJPKH?(2w9407TZ<GFzb+EOm0|8
zw+TgV(6Ma3@j`(GaZN;NDZ#k2I48?_DBgv%67qGW#g%q_ne$Vax_gCRtiLdlcqL6u
zn~5mkuk_<pxdP#`Ms^G897UZasvjIgYr?(ZwWKM5zcX$nOSLv`LXPn~AmMW9Ymc?G
z9piA1_iL-~RrubSS`vg#;sr5b572<_(Mq5IjuzSqv%Od&q#6qBs8n?%#e@wgmJUS8
zG2xN86==E<l12Cks(;KqWTJ^zU{6!PT+*Wwe~A&HX6ikW_$sl=){APf6zx;H&xgM7
zV)e1&UGu?Ja`l)huDyXm&Gi}L(1i;1d<#d6s0uBqJ$}2QXE^Q@1F18)>O0i&QQwBP
zO{-36Vc|6y+fH{kImwPlH6FX%qgr8QRV$c+^i(cdqZ%?f<m;fr)HF#e4R*viP7UH(
z4I24s_vaMvY=!i|!a90+hmOh#n+l&Xfmz)&9Qs3yyHK*m2m7cQJM$S%G*<&jiv%O9
zfjDCPsFAZY2$}`>T{;@9X)kte`Y@*TJ@yUgfV;{xBqx^JOXumO?3<0VQLqoh={r3V
z!+7Kvy5D~PA?W^ivYzu;nkI>ML6b~H7)e~`dLHU1EI5!hp+61gF41r05>U8D#~m@E
zlHtc?P#ag3$NBB`KRfjGcUnQ;>(lSl>x|~;1<@tu8d$j`S7QU&9PlzA@`j(3d7CaP
zPl9<%L0KJFs$BGhVs31Kh;*~IV8gVKaAS6w2C?y4l%nc30(|L0bt<$)eJ>^Oc#-LF
zpy^Xun~L#lLq6E0^ip>zN<bgAS;FJ`V!oa3?5w}f>vkF+6mM`31?ku7qe?wqS61tC
zHoP3dMahA@P`Pi_2)T6z?tf7|bLap}g3*PMdX3i(KO8>Txa>7tRUXcm+E=(w>KfxQ
zr=$FV`OBcA;l?H-k*0^yyocRI<DK+hgZhPj+@Suz=OCK&Un}%bh>qkIbR>9ZO<>(>
zlNd%Zv?Mp;nEWWNwCK>2Jo7RB)}>q1zBv!+NbnXhA-1a|^T~FK6Ua>_j%%^Qo{h4&
zT#lZoIpcvc2WwBj7t~`W-^!O9Tv27!tO$c59Qs->bGeo~H?-6Rpv6^gN3R-1#B`M_
z^&Et0l$LvqJ>r@sJgU~F`=l$bkIc|#zGWVFEAZN-l?w&vBv&`TnoRifzVue^Mtt_A
zlKAXP*&3i{#ETj_1RulyXxPEWbeR*&?HeZ$?Uf?%^9&vdG~I1f!y!O@g}vMd*#O;x
z!c;Bcg90%$34ai31S4vBi?Vn+hSt72ExO|n;SRo7-y|IXX3aajOl)3D=s_V;@x7pZ
zhe2iBtkF-ov_SSjbY-Es(rFl+;Y^~x`as79>QL8~+BF*#FaonD;z#Pm95sEQ{7+(G
zrYq0rIM0p}?dg46mM8^SL{5pIqqUG@mTCgX7Po}{K+`$uH5#ConAA_lQJ;sb1M{zT
zg1<x9_u2Da`aY0`)eExFYL|FlZsLIiezi&3fCf7fzgMqa$_4A+Z7bl!X(<>9FoLJA
z<^r35IHv~y9w6U8+4;pkcvvKCSR|+OVpq`QL~e&!4o{Z_X{^EdJHKjWLDk$R<?jp5
zL{wQ`jr)8*{6l}=sy8Q0dgJ+AG0FL*{0uJ^$`Pt^ts~9ZeD~%DtG981(1dZ-wi4!z
zC`&y8_Q9~bmQTt*j>kn}mRF-=mIebH{tRU|8Ge-@ze;#Jw<NW#`WYv?>+&QG{~DTO
zUizy+;^g&47+?<&*xioH{p@|j36){&dJdd+2)+@D*LZgQ49VD1V?FZ$0}h)}7*3uK
zbG1!z?@+TbH*GYvO^EXy*87BW%CJ^Fv84Z7!hBPS*Rl|Vhl86FJ%m_=0oUNW6-yEs
zGMOvP$LQfi%Ehk;MKe+%d_&N5O~LMJxEOs}0T&UUW&7HpsYNteURY?QhR`~#6h5j)
zShB-zv<nNYu^Z2!6crP+gsVW)uJ`5JwSktnI<%fzXh*tCZ&R73p)jA-!PK{E#a}u_
zE1nxM;(e<Vct8;t(w*Q~H)w%X<n)|E2oG@Gm7vy376RJCl~UHGX{rYg4qB?`SR8`a
zM@+VBPe}-Tm0dsPkUQ{UyS~SvXlicJbG<DQ-v^ee3`SDZ0k?d(1!-QJJWRNV$q*6R
zbI6eGQ=h;MHmGvG+PFawj<AqkK=+ZK%uBT)4CvcH$LOQfXKJ#Ji|y$?{ieff-Y0<c
zZngXFQ%vn|&!wphz7clceVQ1*d0e|(mBrmZ9U$+Co$j9t9L8UDk?dM^=o=l%bGTF3
zPVd%zwaJoV>snJ>RpD8Rs&j&}t&gXh;u}0wXARTH*wbW&Jzb1?O1J^FEPV+A4-efL
zZtW_pJ4(tZmPsqto@<7gzpb2+KD2_sg0SgaJUy23%~=DPPeO|Hx5p;as|_mP3)Gt;
zQr2Q7g7qHjqtwcsCR7VRUk=u|$EcYie>l92Q+3AvO5RlfmryZO)<hY3ax-(ow$NNE
zqcB^mjT;i%fKRkI38n&X%+VXmR);{{S)3AZ#^}SJ42ei=35%pT+$?qq#=4-Q@xb8e
z#~@~PP393pKpB1@y%A39|4mJtARt<Vqg1=;*WeG0H$MNKLwLjn3I^ilKe;z2bBm?+
z5T1@C+&*}Qhp%uv;SWWM@IS!0^QvHejmIwcXALi=6DNTDH&;&(D$dhynpsqKlJS(~
zUr4J~bUymqLt3?s0*!?}K#N!(M0FVfVnB<N(PDqPiOJKyZY=8$&rXI)EjeDMZBrcE
z`Z}*2esPAACqJOIBhuLB55s92A!tm&#6QWPz{Cq-`!dSPrYsw(Ze|z?P<3e}elQk)
ziaO}x&RxGb<U>xfFGG>+w;ld_B5Sv^XKFz6r2^s|nj;W!Zvqn>9E3vG<}N-v>0hi_
zqml3^rbKE(u%Z<7Oa4MbLk8efvdXdx*JlPZiZjxTaqh3TwIa>rQgNR|qfStu#YsiM
zuHlZvW#-pEN3$&TuItV9A^Yc#r_Jp3cbb{~S`L`mzqt|v`tC1+#BHS}vd_%`MjhZq
z4)D!~0BjJTx3%cx4B%%S;OP!<?=k?d6QJiUu>bGrSMR&SUi~Qtc%o|TJq?;T6$}51
z8L0IR>Mag5@DHl)Fsd_efd?{B_c^F19MmrbHAGNF)b)!D)I0}un}eDnsChu)EpU)2
z(Ao8Cyj0WpV}O6EjFh5-9U48B{RWQCWneQFphg8KUvylb%fJg1N^B8$MdBxwV-sgr
zjZIuNK{G^;%I#A`R`Zy1&}y3WU0%*Ln8g!>64S8wpNYkPVzFoKH?%|xgx3Zd{9}nT
zgTg1mt(=!e!APZ?Ah>{mBZ&*kxj<xWGSanYFS<9>L?F%v)-6E*Jy(D#uB20Ipc<y&
zDXmQ1BN`W=MB8<7JLlQQa^88V);`oq_}W|oHfQUxy`Q8_qcTGgXwB66z=FR4s|fY_
z1}^_Ym+_6xxIZ3<irwr?MNXI@S%RelSB!#ezxfC_BL+`huBSu^sN-YWu>xvIYsVS@
zh{kuR@2wfOMcV88<mfY_8JA+ud<uo5<D$vj3WVBb9#%9Ccs6xojU<<Lin_~<aq}cV
z$z^GI)p)pD^?rD`IJ>bu7Lr_~5LWe9py{vZ`V`Pr6mY6Q8I*<AOjjoAp)z6AJ{~kJ
zBN1J6zo?0Q4TPk}o}BNPB$%*rp6dP)T{D9d^3aJ0fV9l!=t0xkFgL}530fquz_G%V
zWZ9ib5RL>}lkpuS1BLP@_;#|3J$lBU+$c7EBqYmQG-6R-XWORbo-A3Yf_NvI2~w9p
z(ye*a1fqLa4iC3dkS5I`vqr0-X1uh0r{c)*z2Zx(B{)!oWAUwo1ODm&{dX^tv0FA~
zu#wvdsvX}dZDYj^R+B5xw3|kT;yt0{3~AjOc+boyIjQ0#klbQq!Vz%^spHd0LD0Fu
z$eWJ98vl?Xa<#pf2dB&B>>gvzN49j`uBuG`=GF|U?jrH^{UG|}xk*KYFYtI7CetV*
z_LJ`PsL^De2^Hmb-e?7o(Ohgp*u$9rQ;Kp$=7l=aYB9#Y!!AuXhIL5hn5Dx!>rmj~
z&_HVpignaM%Qnik(ciY_*<b%Yz^InC$e3X$)w%qRLt41g3cMCVi@Y{P`fNj`H0IQF
zV`Qo-9bg;Ev<CZf8>ZRJPAg{(R%_1BnE2Eh*+7od8Y`u@wLWp(YeB15DW%AZB7Itv
zOj45wUSJa(yumLRUU-879znt9?0^!>2_fnRQqNaOMsML^$%ji-2}f~tfhOB;#V#A^
z_-gcCIFH5LA&MDOxsFtVSeG=JHI8hTW}q11&Kqy{x@(j-pvx)HRMqfhlzX1)!M7p(
zew@cyF^E_mmFcr82@7cDr5mX5q@|vX>B*x#8&js)huWGZ6^+u!zt*+|v*1^+>oW^F
zHcx$xS@4APjTSQt(%+$BG?xD9x)FIfvxRd`n@C|l6Vst&t|D>a$H<(Gjgh$!dZWY-
z;lh86H(Q#vO-s|ZrRrqqhm5dgY(QzkwuAml6SgWx*c|rZyj*mx(*Ul*%`HRQghBdf
z+bFI=u<pef!uAKgJ5AWW$$96^tUg(Z+5t5ykyP`aY$akTFHwu3YifC5!A9W|boPau
zuj5?e-aPIIaV!+%tzlZ3LkD8$5%qk4LL+aZ0?pqsvyaoI(eum?Jt0<;V_?Bt)y)D>
zC*WR(zT#c-wS?p5U?*Hh0&N5_7f=K<lwhrO($rF$#^CJa=yBMBjHXG!s{UdoL$MH`
z(N{#G$>E;9V(-Xzh;m75OL@EWeUrPtGEkn8OC8UZ!@-j6izv0CFk$H{ag}LTo131M
zW|@6;B*+vlEz}2D_z(0fo|%c!h9!BLO%J9cLD#CqX$?;|kZ#VgMkss=Sh}EV4(*nH
zp@AdedRf})$i6;vUzYAWeW_KzY=#U{f!2!Ft(KmVU9PM23UKT^%GHD**GBr@My&v0
ziTbvR^w~f+Wtm*-hSEr%S(so`iXGhp8#rR4kylOha#xys2z?d$It8INFl5#iI!J9H
z#GR&}HAfC}u0_ZtSed>&kp7`rVLeH><qOBF2MAHAG)4>oa!;V?W#&x0GdWqLPYYq@
z)0~XrB#?Yc(}`iNk#)(C8~m8&7NpM$umh*F)Ee}d;X|=M<EKk~H>rqVo4s9%r#e7C
z;Yhv3c*H<<AQo@S0H@_!=%3@fL~iBc)YduaW)EGKrs<y*WT*>RRA^OW<q)VjBWS2P
zXq^|i&xEj2{Z^LxGjdd^5FZ;|kfTaGSrT-|Wo3yIv^#G)<lV;}@@~%=!+VQ}hqgWh
za|`WNg@?Y1ruKUiSG)*+AqbUQNOBfp%}fp}^a%frgIS|aHj9+T3xcK9Yy?XmsJZCA
z+Z!k5Y76C>-8MXQ!~+J$@ObK+O<H%7eW!|KH84(M0c)6q4>@kl^NLXa(?}7mjbM3B
zVYIzbID)Rx_FL+7FNh?Xg+r7}))umFLo~jZw>bS+gyy*0HyOdT4CzcS-iZ?-b>_<Z
zs?wM4F}Matm0B5`V?C7vgTrOzvGiB^<J`w~k&#%wO${^vTzCamDyF=y;_a=SC;Hmg
z_JnKY9Zad_H_h4^PUer(2b9tTa!@!$HRm8qaw+=9=^fVf+R`Dbt4!m`lk0P{*-s~?
zyNBwfrzxev?nw(p=u|{p8V%=Uy@uD7>!|oX4;e@88K<V&-J3iG9D!^ea@KD~oa|=t
zbrK^~ndMgnHqi7Nj8a<oT7z07x2gtwK`9E`se6@YKi5ZgUnqo}k^gDa3`#4cp+kwO
zc?PYOG6=j^sP<)XTN-Nk`U&FeF`t7hKyJh+e^SDUuw<a=Xb-swqX_etn+-k9+y^Pu
z_iK_?B>g^`s@F<fD^U9jbf4}lR2MF`k8y8KBRW~QLHTY|7>1qidi=3E(<YYeJ5m{<
zzg(#2rN`8fm)59quB~2<_A<F8a9?P?nEPnI4Y#@4Lh*LbmZN(iU>i8j{AET4NO7sE
z=jj#n-O#<J*fcLY;-R%%b8Oc{Is`8}PiG@nyYZB0Ozr(3CbdTkRqYgJQoEV%4)@E~
z&Z*i5jF%*V7QxeVlMNM%KxhpKH&%^&N6`Jom(__Fz>-aMj$C*Mhgo9<IM3xe-z7I`
z^;sLJ&*M2^!*fenV8Pt^c-~nMVFF99b0UV>(J^vrQEu!nAgGd#zDndaIULu^K1RLI
zk;_~vM}f;ygEWGra@Rr@qXJuJ1j%ChmmLVqA89kpeZFVTdqCd}dPR?qyz2VF9!_I5
zY@V9UlRc!lR83Owk-sw7T`fX*{bs$J30()^Tw%wG-Iy@Ys(BvAz=AJ-2nvCPkM9uv
z*dhFJkW>NT<j*sl)eg;A%^Js9Aw#EmCM~8VstyBzUR!l8fzHlL68IKWqruiZCW0B4
zFGdM`0Jz2#Ytb?1B!|y2tAW0ku;=L|b8hnVYlWn8coH^kE`c^iuIy?6Dy@&w<m_b|
zAa451R=}_lOMWkfoy7WU)6QPoZT|aAe{RGGzMy$(Fi@L^%zxjI6#809mA{bG+2A($
z(rXg{OQc=3$BET@ge+w7)oC|>Tl-{8Qam|kuUe;7ZeI8xO4S#JCeUr8?DWGhu6V);
zs+tGXA~u-GP^4C4TOu-q@Txjf>%Ws4#qxCM8mJ9+Wd3mMx|^H*)&srJMlp%m`xrp4
z2Jr^mGSbP>W}qF>D5xp)T8~}B@)#V3368?-vLeCjZLy$XFwPIT?H0n4#Lxp5>v<pC
z^-bPZiOoXcOL;obbRknwpIZ)$-YG^u>RZ#cX`QDSxm7QVuc7~pL9DV%_ip9_TX{Xb
zGE=HB<ZEF%=nK`w<5eYyv)t%`T|8uXSL|Dl^_rs->gIBL$bGANajLz?eY5(0vfX#z
zp0;2)ixbf&%;MEx5b)vD+&$I#4~dKvc=o69x{qCHv3Il@X|*}y>vclX)(n8c9*~-;
z@)>q5yS%eR5xf#|AyOR<G<}XfBxj9hiVBC)oa>wn=jumyrrzD5>O}0E$n%L7sbjhn
z9|7F0l`b1qV8K@m_h73Lan<(xsp{WbRhhJL_GNUuFu0a>xklPr%zLb(nOyco9QB35
z3D&{DsT(9~sart*j)<G)Z~}Z3BU8hT?#t8tsY2)cQb;*`y!n0f*@At@LNkT&5VX8a
zRN@#bvAa9-utTe6ty0ZOdj`E8A1-@Z6Ssk)K+{;8lvJacQ%aGhQAYVJ><Gs{v#4fz
zK+a@Js^HOMy?cg?RlB*NY1KlPXo@biqvNOgtw-Y?b31R2AHqnsQJ@vNfyW|rjxeFt
zx@$b+=M-7t#Lezb`b%G=2Xm_}Fh6Jc@Q_BMGK}NhFpgJnX97e|)+@;gI!~)LWWAa!
zW<P%#L~}*w{&|Nq>{M%*#}_hs_FXir#TrPLWLlfdrE(gB`PiBgP8cgpL;<+vkGJab
zOdFy;9IR=<R31aM2QZo|`J*XCK$VzteKQgQ+N#GGD473ZM!=^|(b^+zu2D7(s<cpx
zDdRJ(DpeSXZ;!@TBRpf=>OB6?4?LR!1V>^6MB+%FZ%Ov8iL8AmLI^+xLIcuI9Tn8v
z;PZu*H0f{EZL@Gi<6XSn@%icpA673xc31`G9U@y-gI(TaBuw38=B{~G=sYhn=B_D{
z2nP0B>s5IZj)3y{OER68?lYOZtmC>?LeDeynST}Q8)E6oa(1t4ew#Kv&6;Dzrv)4^
zD~)5T<FaQ;%@p)L^P+QOvlp-=!(aEa2Q_Y=kvOce=jgz76(A$UD2adA+~O@P)a-$#
z<3Z7lvI-poKzbu-zO?>{Sp|t$VZ5DK7*7T6A6tA|#hzXy>zq*HN2QU*{rLZmFa&o|
zkj<l&sK*uQKSB|gRSFpQGtR}ul}||RErC!(6ZNB_iJz6H-an<K^X*>{GIh(9pK2MJ
z`e;PUGKNw)x%RZO0sx)3b7sK>+25%foW&d2d2kBT85bNT@x`rdb@SH8a0g7EXD<Yl
zn@EM7U*NWJbvFMBnqLhp2o1(G(%c(Za0Vw}65?eqDCHC%d30>$ky^L}YS(^hzs@E{
zYT4*4Uya&lbR31fNXPMrDmgmIK^!2xMiZ^p9F&P2VF2%}@<7u|^r*B$k|#$AIGoxh
z+Z2m|T^QX#W*g$B@F}{F@pV}7btxgZ3KZ5IB7gCd@(F*|rq%GyKeISUaYDeGaPo+T
zgKRI6<+>35_g?=TiJ|@x(KF0V3QktLx2D$J6;1qE>y<QwETL%pL#m&m&HRg9Df<&%
z(=(H2-iXiIsdRBNl;xwD!WvMV&NALfGrF%!&xk>MR5bCU%#3LBX2gCsBW5updI{J3
z2f)IK^o;1xjChKo^o;1{{2$JU7kKACn-LY$K5<5TorPc>+;2wA2GpAo$8-CT8Sy11
z+*oZ8F5$$R5ufJtUzid2SlW!(^syP?#hW{1K0L)h{5R&qs6*$&UDxVkZ3-;E{9dNT
zC(nU<|K&OG>rb8oU*SEP12-;vrZ=E`*waXj&^Du=MjeawhDB*K_2NuJQ*UzsP5t^_
z(A16C$gx;Q!BCxFo1-zvUXp?Ty@P+)!GH5+;42;caKR68@G~;-S2+0V9Q;pi2mZSb
zetQ$}FEv>G;SBsZ2Y-QsKUeVIbMTJ|{y!XiK?eR92mg5o|GArhALiia3;yR0zVoj1
z_inw~zIP7?zW0^efIr5;PZ0cB4t{9{{%;QcDF^?!;E#3iBLsge@E$)i0Dl*I_`8m!
zgFxRrkz;yM)^Hwfv^eR%veVfLnf{ng6XzDP=84qFPjQ&wl_2D#L>VG=VtWUTkCl4h
z)08lm&aC9;?5c3>H96b|NuSN%Z-gtNUL17-Sc0{3;_NtJv;6(-=O*({0>eEQBV-ag
zV9p_Z1jnZ;;>4kO`$QtXnc6g3@{xk2tkutk^O7?^O%RevClL@2H>fGn!oW^<A5Sbx
z(_a$;`<a+;RaC)!*94MtwKB<Bho-vK1UdZ4Vfz(V?{#H}D<mXlWX?wov@&3tR<Ng*
z9zmIQvI#!2=lKYx;KogZ>rNf0pO8j$(qGwfdC{iSD?jKU{7YWZgNkI-yRv=@DM!gE
ztJGZ?`u&OUt`HZbzDg!NXdXh7{+cte)HVW4`YQ}yf7o34ghtDD*DvmpeYziiYSA<L
zU6TSUIO+PuI!w8~MTffUJ9HR*ed@J~HmBBKv@P|_MY~dM7wt>E_Jb{{^*`uNJ@bS2
zQ*A%^2&lYC^Yv)nl=*rzuWr5`%^N)*%VQz@cm3TF92&rYWC3peI}Pdj89&RO^_IM{
zW)=4>dF9(?6>q*hy$U*J*eck`fmLwt@0isa-K_o|3%IjlmaWPMGw=^M_*MtsvJm*c
zIrz^BKF`6=&A>M}_!}Jj?*#v_gWt`9>D)APVE(2#+k2P}jaN>9p*aD#NSv_yABAfx
zsw~MV`5s%}`<4uMIdZ;S+iS^21(>L)Z0s2nXgZBpUa&*55aqz&aL47D2{aWQX?KVp
zMq%UE>8%=_lyTC)N#t>+tNt}8+003bR!AGJ*Bv@np*a4;xwl$Zahe!xFU%*wYm8Y-
zP+RzeaDreP2sxwTds^Q+(YI%9FBenHhp!^};}aX->JSe2`LWarINSy5N1c`H3{C2p
z2#ffBWCTTgYNX|nz(B^CAi07H6f_-0Q0FaIJGD7ILl!iO3Sz<tF`R?K@Iy2@oMjY_
zm=`KTnuGOO!{Tp^p_UMGP}L9UNFIbpGIY#A-3<F$RKD2C{YR<hsd|nCpDLe6xn(bf
z?@?io3RPKr4T)oPv4abIT84&OH~B~Q62tt6j|i2^aoEv6bD`HyIP<4gaPpa#+VKk)
zy5m>YIlymh<JdNFR(1BWpD-__4u~<y>8BaeDlAmfkTI9o5uUAk+yMQh><BHoC%d7Z
zk%-E*DfHZ9suQgtmTv8MP-5)Q6ac`;ORom}V&J7E>(o)Y^oN`f!X68BX{we^@pkg~
ztYbC+DPhMt6wOkLuC2B@Q~=2l<Ag17m68U6CmXo!R3X9c(F5_DCT|dvt7Uf!H!`gT
zDeyhBo-7@8QmVT?2s9lf6+`2`oN1S-Ma4q#LEdSO(AJ*h=)(xbsSmObN&3mKM00Rn
zPFvG4#$HHA#L6%hyhsYA9kIZfYa(U)0*&8gLC6NPhL)n3$MxN^^EQ5ia#J;I4DWdn
z|4((d#ym_I;8AMFc(DBn3x#P`kx0^CK45&#LMeRlC80QKuoVcp;oj$z&rW%S2Jt<y
z<d`7}y(&^r7?J$fzeZC^L~*8yB(FY<_v}TF7EAur#}i!`x?;)NY)&`fyB|w_XAq}r
zF+D`c-~lY*OrMP;zdxAMm*v323(7~fT(WQsJ}nntQW4J|!dd0TgE+0s&a!jAiU-;0
zVVA`72isW=XNTF@;kxa&(_9MO^GB-Ctv6!6f~eY~L}0zW>tyh9;KYz-31ZH~Y03c1
z-~|JXFK90GTFD!!le^dQN7Oy1n#P_vaoX>S%R44c`*U%5XIsTpmz_Sltzy{vuC|I3
z)^BR7-Fr-jukD9>XK!sAGpmeX2Nio~ceIV!d(37#?G`|{PxrQ4!P~kASnZX=>gwU<
zbsU@JYx}Gn2iftvS%d9(QI_BCUzC+?*R!$?v*VAlhLFK8Cw21rS5seDPsqmuf#wS|
zbn)(J*|tE_8K@GvUgL*plA3kdOZ;u*?=AkeYc9^?d;8Hx`?4MOl|5}0N37pDan8Y<
z%lsGXt9ICV^kUz{Ewgflg~qDIdym;+AGAADXdSR44Km1qOq_OCae1b2a9c(8`n{^M
zCk@BP?LDT)s_M;Dl|2CDFb6Vm+RY?QNIyq{1sdsEr&dKXtuhEMz2*%7Nw>;-!rC*i
z5^K*6*Agnp)}Rry+Iqoz%G%QhGK2{hY^(Uf`ZaA8C$Hb28PTD+K`OnCf#!RdaN)9D
zfu=uNFo8Rr$3gs!-OC}33p7TfP;)r?=m=)hJ9s`C;ma%bh~!9}h)AB#$tM%Z8r|`T
z<SfoTB6$a=9+7<SGa`~AQQ+OeG?6Sngh-;6tw>Hz6UlW!YULRS;D3!s;!)IjdEF-x
z$-f^>D-R)(e^J^0K_dC$j~tPFmm~ak^$90Tazyfke|JRkEAIG}KRdv0qy#7;c~k8_
zC6Z1Fs%YqjfBp;f@mID<|9_4?V&#)24l?@RAdX)|-}0~0!_TDY;gtVv^pFYJpBR>>
ziJ=m=hDnMIdUM2ZcipK2Nmqs#evJMRFH`wk19L^s|F_V;?~DG4=s^la|3nJYs&Po*
zK>h=I_Ak;yM+HR>9cBE#^za|k!<YXrJsd#F{=Y~MzalffM8g~imQlT8PQpuB<oeeT
zPRoT~r`Sjy`kD&_;<PyRoO&d0|MCA+E}SY`*Fx$Yo4owQipH%b8(vCS@=qXpNO_VC
z)1G{u(Zs|<6CC@?hBcj6vGY{ure>gQc*&8}jb<GGCcW0g!*F211r!TaAs)(5xi*Hy
zrQL3OzER2in+-fURnA`Q$J>fP;>lmh)0<sYo^ccbsc|bpq2uzLq&Pq@EJ_-D>!$3H
zyCocMyx7sb-5ibXoqSP-oWY|@K9lBNBJl&(C<%OH!B`o4iEys9SAP0}M?Pu+R6Er%
zFX<^y-XrcUof0Lkc+<oq^!+dKJp|LmFTko!XNDsVi{~3{J%Whi3B(ky5u|TTf2y|B
z9`+icy$F(6!}wP87x`S92ppf5`o7v{e{l3oc0>Y8BJW|uNXvW9pry@(<J0mU_K$xk
z@kNvV!(tMe(h^_(B2VH&*1gkY*jRG-9cb1_3t=f8&f0~<_lh#BG)^BN@gbbB=~;PV
zzi&j2%t6c%q(l?@o|O4{2Oo~sq3%n%6^5=X_r$&^(sj~5$Jw8i3a;pvIC>W<)0(d%
zLgLtN34HphoWR`u@TsxnG)a8~Iy9-zoL%}V)F<HRPZF5qG)aAI5_vY3JX2C1sd3w5
z$*UywVHH6Zyiii#Yxq_~k{62Vy$02bC4VxQQuf;6TH&O=e)C7FK-!0UVs@6P1%K(6
zO1{XuP~!BwqQUNt5~ugUVb99kvAknNLmXIvZXBi$pY+S&%6wLI81c0w3o69+g07)_
zaz;O?&I{T~=@{9<lGw)I&a|8)`A_-YdMw{tOKL_icWx@Rem^IbbAy)HZ5oN)a-T^3
zokHb#(`Eh>_<81#+O3<^ZtsW7o~(Z)qK!#f<cw%jT-^Ui8e!P(?Grp~H$=f@KaTB=
zp*W50N(JJu-8r0k*zQO)TK|N>2irMkhq1}28EiL%?Q8!nY{#~Y&iRvt?Kta)?Z_3g
zLzwjA*lv^ZK3GTmpTKrIsvWk=xzNLQU#k+flW0V!Y>tOwJCAb@^_#E%623F9xHZ--
z(Xtl<P21%3|38oKD#gP8tH@3+un24Dw?2mK$_BLmzku)l0Z0G80N)u$--qlj{%6R}
zLTvZpCLysjvUB$2|3&yN`VaA4B>9tUFxz%K#r_e#yGr;@2(cfRpQbks-wi-y{{-Qs
z(b@of2a3xe%>Q5DyVIl(slD2tn4Uvkkka1U_TY%wPDk#o@a-L2Mq;=aG2T{{;P(ig
z3@j*9tUB}>3Xk>$&e>0dS?w=kH5z-^<^7f%YJ-96hvWd)Itf?Ac+Z|S6~5N^3yp6L
zAsajDaJJ|K8n?k~5Sr+Z;<GaE|B`!u3-9l#@b!!>Tf_U8hvLh<_lKZ)QvwSf%5#k<
zrx8myk0!395qEoyP|RmqaF`Z1jV_Op#rR;PaVsfTpY*%dv{b}jsId0*SbItXjp75f
z%k)6kut5JN{j)Af;F}Ow@ENZ~>Xu{vQHwqWmX#Ons0hsBCv^+W*l5ilV`|re0nPYF
zJ(IHTEFmWb!xO^Ah{oI;?CYL5=Ncn#S3F~6-6JDNap3xm9^N0=N2;q(gGM*KlkTQK
z<C!#w5gVe~Yk9Bv2FrlLbf}@LoufG4@4a4qak$Q1^x7znC{t`pE<K>!E`wo;_lJ`+
za<MOvlM1$nbS^{5P+M_sva*6OS(A#!vtZZb-uu@7!`{2VM_FC_-x(Mrw$YhrY-1bS
zsA*47Y@uQ$NBTCwfzIfRqNX*_w5AEQw51gxv1kzkNt9_E3?AF6)r!{k(2Dg^t9VZ!
zBNuDs;sx;*z)NC8xe4K#_xoG>naL#3)^pDPyr1{;e%^dOndf=dvz~ojd+oK>UVClG
z3-_eTvOyAFIIFMh2;S`E|8M-;u8w?8mMx3<2(h=NyNL5p6o0~t0KkApfXp8xtMge?
z6>(#R+E!HTpK*i^)tiPrg|%C5Gx<U2x}ErLc+qM@(5sk4=(-*cxcSQCLI?>CxIY8)
zI0y7!mjOFN(n700isuS~$DM>W3^Z2AG9z&>$_|FWeMqy!Q{_z7^Vua<?-oqc@;sa=
z=%3e@Gv-yibE3Yt!zHWgHgz4Pr2cWIkO{-)YK=8&tzmPy?o%Cui@fb0EB)W?meVrk
zyMNw4>!gUNr2k*)pGXZ$W~c<#<<d9nMb)WsgRI8T-N0!FMwoQb-*unrA_5Zq!(oFb
zWI{>dw*ckL;e5UaaCw-_zHN<<_@+qPCjWVf$_M;Bzs4wfAV;G}1BZXpE;M)^;4tvt
zD;bTTldOLfZNnQpej7ZCi|RoLG@hO+ruEOR?u%06hojX}^F2f5k(~^V+>p~CLXY&=
z(+O_GJfbrxa~V~V4E&F1%j*dY5!?rSFUo`9DH|eneM4lM4H3erXo%c^$Nv2ck&N*{
zD)UKYT*kO~uS@>MAE(``ExLgl8sE}=s@>UbR9ke&xWHK)c1ydPo=>AMpwSm(Hab3`
z|DAR}CDnfay#4<ozMUQ)|1WR<5G$}#L-7r}H#BVt%s)y+uSwz(+k6quGRvYh%<pgP
zexoc9-hr*6hVEJ;U(RAvgqa#7j1beXe9{PkQ5RdIgePFzSS(R~;fR=TrP=Qz@qMCh
zI{*MT2X;sfu-`8CKMv{IiCnSx3%Cyw3@n2mWuDNbT|Dl1%MAKjdiAkWD_3KTIkxsY
zT)w&Fopu9tI1ra_E_uU&wmQ&OJ&mc<i<bnM`W<XJ<CfrtuuM)W(-$XSt61xu<F!(U
zcdhazw^po5?wGaG{`Xe2$&JVVxZM6N)Hym!(?z1_UbgnIXN(Cf&}8=gUi^$!g#i?E
zMz~}QB6f5k+I7W|!#F$5wf$uhiCo*Cn;))uFBIQp4Zq84cyt=u3xHD;4ZofSU3YGx
z`~quuLUOop9M%CFTnB_MpaT$^^Z{aBabJ5W<PS~(Y~(OqfS^nL`vXZ4U+A-!`!DV5
zm1Qx-dVJ6N^JDeLw^yr)YcI*@k8gK4P)2`z+d88564xK!cKyM3!!POA9ru3B?)Wp^
zVSB5VsXNNm9mG42aTeTJlW~>3-s=zT^2?6lB0#4U%9|g8mR#DjN!_v?9Y6lw?hSUh
z-Xi)4CKZaa<0UmF;vejOBM|<`Iwn@bV3C;6HyCcMa)Y6*gPrWD(Qso`q-`V5<i&Y#
z8V_&Ae49lL{|n<mj$60M2`gB8o1CzMvu<OVbym=Sn;LlX?F-vC#g<gw-l{e(zn$**
z+LGw)FfB*>oV)wqp#2{Vr2Ve~z@tSDb|oCI($~lb)X0e;qW!&ID<6~5${zI}H0s~2
zMy8V@HHWAF1=iN0|D<R&Y;B$z01IgI&1|?6!_qKF^KYa7orK5uU!s4t|8HFT|Hifd
zZ(RFxG9T@K$HMldwErEg?MrF@I~WyTTat4}s`=YKX7l%_^(lwYy4U^#XdN=Q1_?AN
z{!7;peu*aULcP75%fcpux2+kOL5mwUv9aD%A|vMvejJVW%BgF6=dy3^?>#?J^yO&$
zT?cp&K+MO^3-9@^0naxMc%D1pdHR6o>hv?L(`9i!Ah+y|uS#?<WL2_(gFUH!(-vc{
zrXd9yhvRO~&PYj95#JlTU+-7`-SXVTRMhFS4BqKY*1gxYcM$3rhh|z^yc<sVx1`#l
z(AvV9LGf(*>R(QE`_&&nB(welo^KrRJa@qJ^a0P+1D-GBxnJcARr#q@p0J%7V{|S?
zE!+yJi(=v8x*GAlS*Ggu@&9!B2V3$VmQnrgu6@{%&~3fyLkfqlKI`%6@0VRPu>S${
z>;D1IHx787JK%ZxfamG~&lmE1`1<#+{+G2dV4bh(TkQ{9e@p&Ds{el1emdKY1hE_6
z^O<8IyWwrdLfZb_SN{uD9E{-*!O%3(-wQN4n1Pmt{N@ly{1C{`4uO3C5XeOt5Sy{w
zkQf6P?qQ=LyX5si)3^HwqPX^pI&M*i>oN|E5?a#bAIyg_-@DfT9E#UN6ZChz)fuVj
z154xUcdthhTo$f*K2ozSgy*$ySt$N`V%AU}XZMxGd&+jN?``cqN;^|Zm@W}?YX2CC
zuVk*n&3c=6isR_A_=nw}mB7)bA0SEPEMToAR<gaeP0|IPd8goum5vWpv9hf9*>NT7
z>vqQ1g@(OahV#MnmE8gFeSDn(mXyVpl?{9KQuah~{^~#ZFKbxVQ@5deUjN!{wc0V=
zP`j<6`0g^_GXtxZ<Ej>Dd>-EEYW6%w;_ArGf`Rq9fb|oOqGn_2)hoR!{hL*(6bJZ2
z>T_^leE|HA>J#tWy*?05y1t5k=t~Z$i!t`@bcGJGzS6%!`TwAQp8FX6!@-x+SE+lP
zz8PO9z@x5vEZ)g~r+<2^f0k>{6yg9q)LZ6TewbPn_tmO6T`P|OyN?t9mgJt3(LbyI
zb^rVyp*I}akDr=yWdHOz=Ez>m1xNPNrr>Bk<16gtUZr`fdC1lurw~7g7-BCM5OaiB
zD?|?O5aRMQV$vbL<Phgy0pe9c#Fbr$zfB|F<`D07h~F3DEh9lZ-yzne5ob8WMu#}I
z8pKydfcQm+7)>Ld;}E~?5NF}|_2>p69?&$@Jmlmw;)xFND-Q9eLhKUa3l4Gbk5aYn
z30tj?bcpwS55)I`c#lI|lSX{rA$D+~d0rM`w-Dz##6@YuKRCok9O7klAWl~EOmv9X
zq!F)ni1Qs{R4LxA=Fx5zD)X&0V!1>7zC*nBQV@SH#GwxHv^3(W4snb_oGHYIg!qQ0
zspcX6G~ytKc&tPGYZZup72;nV;;TPQHP0*Gu;zJ>3$@-T#D|4=y+eF3jd;I9Y;%a$
z2yv7UYaC)cjd-0yyxAc>im%?IX9)4%9b#1)@w*Q3GKYAN5JN)Db%<xC5zlal;~nCp
zOF$ea#7?FeYJGGXF~=dE;1Kgv#_>Y*;-|hnB~|Oq<E_>^xlro|m13_Dz4)n5q!Ax;
zh|fC2_i;>oG>g8r_^H22W8LJie($h03M(KiowhnHjrCK9b&bRN54=7e4GOCWtaH*>
zVTU!*VNFoca=~IZ+EeGTZ{&~X{G<4(nsl}0ip?9NjzZUDTY?;VCUS2MJy&uQZkVkN
z9XIJc$E@PRbDyB`ea%B{Pscl0#W;^1;cs?+xcP!$IG#u8F2Mf=8MiFnJB}rzQ1ed<
zxqhhq9PCzL6?+QN2}{AjTBfs8<sh<%!f;=z`>iTiSvju@o69P=R5h3B$kx*4vMOHR
z&+Ab0_a=v$FP~ytJy30N`bq8gz<k6qW9OX_C$No>VHHOgO#NEG$Ov(Iif7%^ip1ya
zeoPU!$8xKn1fSHE#qq&h<7?RqMh=nY`&W>=e0D@jG1=DTz*4g|R`Y!PrG|HW+Ivt@
zCb?kqLD>j*zX9%QaL-PRD$=UVglOVov~jmn``pBsBNgqzIem^PLRRVt`v~trv^Ktx
zJ#iX1WavVUDYs&Fj2y$fKb-kY{y64RgSQB`t@?)ZZY()MV5m_8jbmk7v-NCm_rYM#
zdB=+zfT*5w;c)?g6i`4(k;#|(+h-&Jv)_Wu*vGG&tThldMh6<FA^K^lsfy!l_l8b&
z0jgAB?X!R?xC_h6X9@*XgK#^79Qwx(<=xG^BMWsF2StofRx?wFxrfinD+)ATK~6Z(
zoC-xi4)wwZJDlDM#43E?K5aW;OGb;{g0+yVa-S(X_OU7X0DCG{f=J18wN2yVTOrcq
zIA71qu4oCAIeQZdNfL8vzOU{AqEB&GkyZN~#|4o$Dc6(Y%%|e5;gaO+N8HKul{h7(
z;!#?ZfnaL2C+u6NGEQKr<~V;vR`<0$sP@MXXkR2Z-_}UUy4see)|n>|r@!R&*`JE~
zmL>NF{dailJj+;Ny6_-2jLp7fGoK^A1USzeisQd9tNXraa~Wo)3yPb6?hiGen|*d7
zl-<kj_C3&<uP0Q}S{8`4TBpQB>iOAOS?ush1i7Zx8~k=}BV!Lc#=_0tqM_w|)qFvz
zZYFSJ5O}~v`Qzdr$UQGs^Lni288e5c0&g=YtD?=<ieSvV^;bASwB~t*U!?EGm3$Cx
z&MDHk(C`i<PQnx>BRigUTf@6o%aIs7g1NsvIGh-Z?UBwm1<2ckrF0oR#F}N<ilDqa
zGH^aClw#m*=Om(8eC0)wNaRe{TIV!u^=#?Nxm+)795rxY1(hl!2zu5LDZXSe$hJN0
z&+5jvM1N`hA>}e4>712L6+-8qmU@Svir?rQ6U8I3YBfHrYLS@+07R;t`0Vwp5@?=4
zIHN7HtJ}$z8r0$Df+C$~rG!M$c1Rd&wH7RF1ukAuvIz$orH`jvC81ue9TEgWDT^u2
z6c;qB<1`KAQi7`u*bk=k7pBs8ttb{#4~Xz4QuEmgMg$a`Z*V44rVM=9-<0V>u-u;6
zXTzM-W8*fF^9xY{t~yI)fyC*B=mNJD)q@&|ug4zW{pHN|OzxZ1YS!&X&6o`@WY-c-
z1_ectk~ag5-ytVcolw-0ofof7c3k|bO!Ddekd=4xB(wict-${Koh7e3klBByR$y@a
z&XN}$$PB+zE9^3Al3qNg7ey5^QMSt_7A@H^{a()$-<sT6(JFI1Czke9w53$j<lafG
zv8FAx<E1jPg8Int3a!cJEaq-vd)xGhIg69qC#_8Gs~}L+vPjA6b+1V&w$<_c!85+x
z^$DufSAN@htdW<`&fE`C^^Dy5-Ul9;`rW;$45m)n-<N~l@W_b*WIc7#LBc9b%v-E{
zlkZJhrMH0eRv-#DNeHfRJbkzC%R&9Ww30V0^G8Zvt(((TNs%&j*Kp07E`F*y)W)zT
zIS#=JBsMa>)jA4r!XVxg-n}O3d&P}k|5X~lGyBFbLxw?0H0xcOKSvO@%Z*%y$g*?T
zav4sHnyoj}!iNk{5jWg9aoU6@7@(uA!P4{~;?;Q3S_wpJ<C4Wo8<)&PYHmi&iBc+<
z^9)nrhZvOdkpm<?(ief00(1}0q*n#NTxjWA+)u{iGl~pE*uHw8we_O)h1^jOg^OWs
zu-T0kLLxz%6o;FMr@@JyJ%Qs>G;W9`WCmh(AFc_Xh1XT>uXJ2r;Jz^kS8UdaCW;=@
zfJTJcEk^}DclvH}$E1~}V>U*5Tcf_0ZT-PtRQsPsSi}%nA!0pUHzL?9f^FpKw2AUx
z>Ll1pINpal2xe_^N5v|SU|;nJcF*AH3&AhHWUu=`_Y?Q`ePZu)#$Dwvm2UaOg9-se
zawL`9G-;i^Kj_{I#&h08P`n)VcuB_BgZh7M1#)y4Xwv#`SCAjb0r?Iyn9TwbC*2gl
z@>2%q4~g?Ld0={FTeO0e(7^~QCKN`Si;B|rwHPNMiU1}=;5zYjhpZPy<2~fij?rx{
zG1EqFIfNOx4I*j!Gfq?L_|dowepKrC(YOqLRO<NAILD7l)BI@ixGaw$wZf3LvOf40
zHjMWx{sBW`Am{oQhb7m%9zJV$QJ_iwBydxO%hl4L>KkY}-GFUT&a*3ab4QcK&t=gA
z($CiB*}hq;Dao)utd&lab3+HUxH^A@<m?HwF20$xZ8%XL;Ahldp?>C$Jp?W$<L=@C
z<Fl8*f^;8eiOAc9^0_ftyC6bHJ7R*yH*}wz!QbT-&6u4K^S#0%nxeeblYH%x7%Rr{
zXBAWC1Z9wJe}y_w{P7(M0W-R`l@2AIPvP`4z46w5=d9f5;|txT)m=Jt$$6XrP_%bX
z-D{Ef@<>U0?L#U1jsJ04Ggh*{ZhQAasWk1XVDpeiC#7`7M@E~jxPlA1V!hPVEhuCu
zrLrZBINu@O;Sj%b0f^TM(NijakVgEzL%hNv{`y-WUMEG6X?#do8tdy0Yl6etJ^`%X
z3QJ1mXVX~6JFHV2)``mW4q-{D{AgmT4m-}YI`|#d6H-vOfrV1}!%~#W@-YhGQJVG5
zvTt3~a8(sr<96erYeLOqD?(?Dtqe5f`cu(6>axFG_TBF&o<gWOQXX!8YK67fc$PlD
zrDGXaH-9RcIN^I|kYnJ!@zNH?E11W4S&sedc=ZmiVl@T-!KRDoksHEo?-kY#9l18L
zdq?*W9c;8CQnHlo*h}%_CTvT`*4Ztli8igBeI1^KL^^m_W%KoTk|-nO6u9gaqsJ)-
zz*rw~4;uDx=J!(W#-aE%oyXW17jG*go(BAA$ZMwz&g%Y@^Uqf(g+$H*#ZqYaH=&JO
z3+WN+D9^nQVfn|H%x}rs+O=bs`xo@~Ueg<`*+Ig!UkSyZ6X)#?ruhx@+524W333+8
zdnDQOr8QS<k^f8H0`OLKZ)_2ISwoM1Mxgs{wYNNB#+B@-{dM0LKRZ?T<&O|N`Achl
zc7}W4G*KSD?jDFwO2X~k=U&<Zz)L*iqbr^r2)56!Xd4K&gaX(8jEeWY>eBk3Ye8*_
zuolcQBlYTt#~Oy*GK?lf>7p0<+V5RVq)4Fb>+U}u{c#PzP`uDxO5LR@($H2ID_K$Z
zT8u;0N|w6uNDIf+!pOr>Jl?i;|JHTfP?=jp>h-?-erEZ3{q?zYe<{$sYU7pxT+gjg
z&q1UpC_7}KRquV8-lP4dSL8gBLZ|QQ@PzEip8iOhpY**KZuAkwNg-|bpztW^hn&m@
zMX*l4$;*7m&wMCw50F7NWKh@Abry}U5>ks+3^=J4Uvaqlvt#RU?VV%o9d+#;jaz$1
z<A-l=W$%xBtquQ!XS97Y#WS8h-FU`oF7Un2TnNwj>gmQa&VXHX7P5sJ`~_+J8yx;Z
zhreB%=uC${MEG0JH2m+S@h@@s(;R-WjCFYq|HZSwZw0?!JfSHKS8)V%(-N6>;%geN
zn##kgJcODrfJa?W6=*tOPZKLcXHA?OX!<<vYw0r;8rrAyUyy3P64qtvtH~1KX7R7_
z;$N-eUq{_Zl(oa~uQj~-;z;o?*oOGmsy_a;&GD~ok&;&US8JMo%|&a+iiO66pc)gG
z`U%}^Oby*MTP|?(wb0hk3v;_g)nghrb6TmS9yB~B3AL)EDC%>J;c_1g#aTzYk8`_D
zX&t9%aYtY;9;f7XnBn-lqlp~Xc<vAkfW24+cWc^j9$%_t!XcTrhKAxVH@picRep(Z
zK{&o1CcR7u#M{mJrN&J6*uMGK5LFx;^x(N@tdd`Y;Y8jqq-F<AX3)$}0)!kzLJu^|
zAy=&tQ6U5h<N$aCA7xAb6@HZ|+tMjX7j?AN7L*jR1a-8q_IiC*6}F=T5&F@v*&Q%P
z{^z6Bo@k(LNX!!X;14$V+pfxuHlZRQFt3bhv#;8?WOeP0Wy?cNs!^6k7`|bIHghJM
zw0vPsKW~QRzPq9JLEM?j8uqeH>uSES^xNma1OD?Z31wvAo;v$Iz2EO?Yrm(R{hoUI
zJxSL~rJ3`e-b19JJv&<RZr!{vN2xbn0*RcL2>l+{&9_qgiSp#&9%7$HdfTHV@6>g7
z$zMYqe?h9_55g+c@hQ%d*5Mj9rer;JUc=HVDKH#z<B55RF;A+wOxfrj3BOktp~TQd
zbiFKjK~y^|0{WVeGV1lH>vbX-((CkAB)*o#N{@e8JoLR00w(M5gxdR2cESz&GUkgp
z$ZvBOw%Y~Efu1^a$(b8zki6U``E3P$bs6&89O*_S%Ot<0sC%^Nz40!LGqgif?>?0U
zD^NtojZ%f|NPJnOVRMH^ypyQ5>XR6iP<4s1lC)}jj-Mhx{DP*=MgAQ=H=3wMn7=xI
z$W13P32HZHhJ5M<5~FvtrPzOPMA>=h`zY%BNJl-{04eDSG*nYKSqKV!b;Cxs?qawT
z!U-2uC6HbF=WxRwwox3<OzaJtu8G`pjt$3B3y=1F1^WN2qkmWqK5BauaN4(*1FWNv
z{bHq_T{z=dHMT5p-4oOD3ZQ=FS5t@PSo??OSo??OSo{0usQuY0srh=nB}1T%+Rt0{
z`}hO%>~(JbrbnKs+iLxLuyDpum$94s>EAoaW^Svybm-!@{vDWppz%~lIir0^e@<rl
zU7zH=T51HWdPUT?apc;5^-1o}4J~N3-wuBL3+%QKN-xAHTh|dKXrkBy!#&XWb+Jbo
znn&1HPRVD>{qvNaE!MB9+tyW($w9}(U&1yzcwGEt0>|x7@la0R=;NX1B!(LgmE1YG
z5PBJJ4c5nRG@jHpVMU7&5%wn;fhKS!SSCLm$-xX7S&pA_h?fXM-8zt;Dk&@vXjqu_
zIQjZee#&-TSuF7;m-S(^SU8&X8e@u1-jYu$(S^D!2O_G}-oD)5q9%#PpO3`<hVk7b
zV5GBp4H@k3I@HnkW~}7(+NC3R!9M*n&gO`p+P7gUk3bGc)*1cD$c>deQ!9Zjn0Kwo
zweKhQDj2Pkm$eT!QnI`DNmyL4{GpvD2A|>{dhFqJ_psj{#^@pMAxo#d`y2A#@Y2fo
z<cjA8K<(Qr+WSFWB7tjv4t=Fxbh&BUO`s^=9!EiAaLmNoaw5z`O1QT>9G81zCO84v
z+@qcxab9BToGNHzPBig7$4tKQIf!F4^P%YBzr;UI+MVVfBmW`)P^i!p|M(&IDf`kP
zd<5k@vwsf9Pj37F`H8s7zvd@j{@46OidAY-j<WQWX?tNlZiAMzG~-1f;LHL?LvlLS
zh%D}Ab7^LO+59Xc5Y!1Zau~t`8re+FwwEC_Ie);B1K#AkN|SSbiXQ&Q^K+}_XO_^G
z{iFGL$oC>C{o;~;GyP&37xatA#v%V-@5%pPmi&LJ=?f#%_+NJTr4Ii$!cRE--wVIR
z;rr6~y<fESgATu1`12ipo$#w1{+e*AJkLA)4lb1EY~kPF@W%;%gu}lxjem#3zu)0c
zxd{AA9R6{_&jLR~f0^_CB4qLq!+*F03WRWDw(eYE9djCHVR=0~R7)?~h#qKM#;+~s
z-+8L6*0Y1Lk}Gj7f5M(F&5xCQFF(-qGEZU~Bf`yLri2yIVI5lMswbtuyv=&)mi&Fg
zQ+TZV;{xy*@3}PWo=;D`kCx0NlUV9CkL2<L9wQA`%3+{Mw>oI9p=WU5+J*MS(jjqA
zuX{otk+}<j&jJ;AtR;HO%DiZ$RN<1Pf$JU@rni@HcNUsF;_AaxR|bAHLb&-Hra~72
z(gjT)ON%vH+;iY0>Q3A8YCes*%N-pMZk`6>L+-qc=2An{kr+ZO<t2z{4k;`hni2M}
zf_Cq=H%4ow`J1O@N1K0Ch|l5s=vdDsLL;?pHJ-#ZFmT@+zSm2h3&gtG+xDKa+(vNE
zm)ZhzH&cb=-V0hwx&w)KsYchER;TlVk&+0F`a!>*JJQ;Ko{KhDpdPOGdQPhrfqBR2
zB|S%jr+Y>{#dFmA&U%jL@2AoUmxRgWn)K_EFg<q*k6zC`qT6)OJ#SB5&voh}+O$(`
z4|(Dl;d*Y70UC=sbXCv2CCq+3S9ct?WD$3N1^aMa=?--mYh8K2HI3`aMRspQd}voL
zw3n_c6?Cd!SI$gz<>(Boo9+I;+Laskp2F;#>P-3h7Y61ksA#H3kAqCR&I8a{blU!2
z_3>a6KRLr8asJ?jjan=@>Vi)Z<^bzW!hdA61_RF0AQbe{IDA@R?zZ{hrlN&3X1piG
z?91Z&@Gj{v<*8<c<JUF2b>QzEty)MHK}2h|g=<)b^$rA3)OQwWsU~cGcwrko23Pqg
zI*cG6ZtRB>6}||obR}<w1LwcVVJqQb`@#vo>2u5@EyyFkC*7>@FwX7T79NI(xtewP
zNDZO9R*Mp*SsQ98X7_3@l-U%4$TweCR7bI}Y1-k-2(JRh4)TUgL{MH_WZg9E#qi*%
ze9yCW2y17B_9coJ_l_1^X)OfE`s$5u7r$`YiciykZ@+Lt2Ukv(B7zQ<x^Ut<c+s%<
zra?;ST9nW*ymj1JLs4`>FCc5Z6@<?yR*2<?gx{7dZ~$nj(wBB{jAG1iIG813wVi1N
zTMot(cKi&@<9)^v3!rZ0{W*#9mZSC1LLu}pLJvBCMh|!waI%a&gjxs^M8$ZJruc<h
z^QaN0uqM~d+5(B3@e4h-UfkPWR)SOX&x1!jcN{ZS4Ts05LzwYLvui6g#$Zpdb#zv>
z{Ad+3;inXh?)V>XeLtKi!&3?Vs`n`hAULVDal#MNgDF!S(-^~vZy53ftCQap6d@G5
z#BpG1ESd?joCi8oqQu$M6pBy+^uY?lJKO3w1sHAh%PQf4GmopD3%utnr9;E4_TV`l
z{AQ(<@DtFuv8YP-6k2TzY^Gopg2flRgEA81EdL4a8{+c9rwsRN!-;{evy8TqZ0f8}
zo$$4Y*9tW&cGi7XpMILEk}RH_!L(3l-pQ7VI%=PJ@q;)fN18d)1}*O$V8ZNrWM{ao
zv#|D4qPSZkyASl&XP{TDV0@%~<}>B&VI5YpLM*ic%i44LY#5Ji(IYq78E8i-mHiq?
z;qFhi=woaHJfuBQNaTJs8lD#=iybdWaRCx7oc;4K(QVjfoKIZ}r+-)98HuO9^C97}
zd9sl-v|*i~+P^UUd-5+#|DOB{)4z*;;q39N@Mlo5Y3gXnO>U&aVhU^XOdm8edOiB^
z(n!fmb=#b;fToDjT}cd#w`*ZW`zNNtI_F<wE<eW9Z(ae88eh0-104fYtW2Qx3eHx+
z?D~*9x=N?-q@pI)>$7l;nnONqL3Ni7Ny&CMH`!@>a`^JC#q%eq5we<B=Opq@S34NN
zvf@rVRpBpmI#h&Bg*fg(H2j;2M7wJ12_otaB`FEaz7L&l)V+l4<y3Qey8X(ooJ1ro
z&+pE4GIeb)n|W+`6Vq(lph!tqAkj30iMFOqB&8uunLBN9>L?PI%1tc!ykWXwb8+6V
z_OjxO*S41xUi@l%Sw1I6mjy50(q5K*@vio=tc!cA@0^u&aclLRwOJRh^tP+^(R20E
ztBa2<uRKt_iEHJg1J!r^q;Scz2G!G7c<T0W4th8T4QH8$bI@=$dpNxwPOm*rp44k^
z_u21j2Ft57l!8eH+hH*8VV!faSr(XkDYIhr1+|40{LU(@pdNnx<W*3?f(oiyRI#@E
zXj^}E)l~h@zWVqbgz3!&iQ5;rEv|T>z3k*!Ym=wWdT|`4vP28o5=}j6EoayJ>P}X0
z>xCi^sDBzQn^iZv>)W7Y*cU_vu19Hax?H_1m_7^Ppm2_WD*)=g{lu$+1;1{uIq}4T
zUk5b`L`!;6eR)y-MBb1!H5gX0PC2(KsZ`E^#!bkuV%^iC!?vHFD6g}Ao^Vd0AXmPQ
zBBb*ZQ)@AW?#k5Hh&MU4?jTVyXH7rI(Sgw!2dxlKTOn=;x{U9$j6sdoWq~?oI+(*}
z2(bvn<4-KQ55$7}iA6V9;e`E~-Lb4RC|a_;?v}0)rBi$N4bDv>eyI)7=6N>CsILYm
z=32=0CQp8bYlR8Uq+^i^?E^P5qn_KW5H%K^b>5T65g+=ESmX)zbssSz>RT5{M31*Y
zSQVAI>w24J2sse;y@IXYMyF7iN-=wB^?#(Pw|(^9UUupvg0UHM)wuI<lIicUIz1C9
z`5@4EG)0Wnuoc-2J4x;~?A-dh4ZEDfjy9Hko&B_Z5F3pzSlxco)yK}daCQ6Gqn3Q=
zKphU$A<(-HwAF#OR?lVZ`7h9$Hw7AXVfBSq7cO}T(GP>^(&_gMr2If;eLT`%XY{^~
z`psf|@M8r3X$Dez*(s{>^(J9)x?1fdY3@1+aTu&=K>I)>1?MCo<2#^{f^hPri3NAG
zmz^ZCJ+)$;t@Ev{+aMo8jer#S;+yt6#!dI9Q|^KpFLb`U?Eld9vjl88=OxHEmK8;?
z(^;;f-`HMdc%N?1*{d9(S8}vh`*~I5UhPkQW?~VqC`WtQXC}R7?+@C01KRfoLCX9-
zEA{>Ga|V3wD%8-%ITX(ZnrbDPzX_2KXb*Ybxf!17&Sp||5(<&{My}O(NY+j)Z&{d}
zm8)Ofx7vNH-FMi12lv<`XJJT`*%(Xl^pz!3Wwpim3)&Yi2x|<8oVf-PT13#$!hrLH
zNy3HGe~^*0xJ1yimwNfvPOK#7+KH1}l-qFS=87dSGdAE^xx+30FhTSfDa53I!sOSi
z+7mPQU3=M86Pbv(<Z_Ys8CLlB>Sq-Vr<^yLI233+Q>Cq84Np9-UWH}Il3XEhG@dFh
z7La|lsuX5SMG|>AbT%=}U#%W_p!l-4)9VKc!hG6gpCF(c@wY!xB<?K&eQp1dgR`pJ
zZZCdexnN8y2Z|>ibv}62cLH2fedq6sUszea_V>kSF0EdBd-0j8aE1QLN`LaeIH12Y
zXrlPcwSuk{bZzz8CW9U*JadDfuL}BV_1cdLmBN<7GdBtPmY{D{KQL<;QqBRcr*J*U
z^;E9CT!-g6LC4Ci@~C*PtMG5fO8(BN8K%$OWU_L!1c`*UsJpOlye4NBza;qvrA&>;
zbspDf;*8(@$I)yxcub8+OD?3*hR*m3Aq~Orx(2A%-||~5dGG2aXwtKHupBAbyS}%(
zGbLY$OtM^y``y<@(CPJp<SCbw&wa~&JMH&<``v24JMFi}e)sd6Hy86QE1zw3cnPko
zEmU!fRb1c3FeZO^UW>3=OqyF$w}XO9)ubM7lMwndMPaNQPBvkUAdl~_${i!JWE8Fq
zByJ?}>W#CrgQ1oKPIh@<R`!>+2Ii`oo*cvLTwZ%<>Kia-oKn4Jc2*_*U6!-fM^B!x
zHY>R&xk`SAIp@Ix#@x`}HfYkz$<J17BFkAVJ>K_lM~O&YHnRt2(`#t{77w8zCN{B9
zRrKWHf0nKL_J-;QW@k-Wojkc>g8+w?J<#wYrFs@Wm$rQT^p34~&r5G$As4}M|5-X2
z&GE^K?Oj)b-Zvf(D@C4+>oj~9Y3kI97xnW}d)evV)jTt4ZTr}(^EViqoE^zvQeux<
zPruvv<9=gDy4PBoeswPkxPI>zf80NiKZ@VfLlFNpe(3Tmf)lnTizZGb^`eQRl)B@G
z0&^Y^;90q&Jbt+DpYg-CQ;Z+B_2Y-C=gm+-pK7g;HK$P6aKnM@8J`&0+VyuM5-?KZ
zF~_g_76yd)H}Pqf>ii6}`SLk$IfCW)xbm^vu@(XA%!KW9u=g`zyBut5CalN7EF9U|
zESKJ12eW8oDcF7ovtVQ?*g?Va=JX|(H<sG8X5&j2X!<6^iH0_`mM*^D>DjIlYfvl8
zqJ!@?z7R?H8wakRl4eUrnit6PI>Hq6jzn;`{VS%^34ga$VnU<o`SRb$Cb4UkS*O}L
z;h!fH#spz3Q8T-^OM(@;nMnHJ2z%pnM3yy8<&$XY>4<Ag^=k!y7d6Y7GL%($w!jt>
zVWk@8i)ERzC^qU_g)!n7mBSei3ef(h9y3CKY~Hgza<n!TRgkls7!Y=ytd68}4B<Ym
zHycjG_jf!lx+C~Mv>&4XnL3B^d8f#tFgf$syiY_36k}CaeF(-&<Xztiih_@6AaT`)
zc3gHoT=qk%lBxS0FWoO!y5Et9)BPBcS@J(FzhJegUFF}Dx>vkpRkf7A@{6SI6)$<u
zp>%jC()Nm%EcH;N?G-QC<WRPHD4x=Hk(9mSC424nRiWfv;9+DLY`Z0rcac4?ead7Z
zcivd_8kD`4owAqAso37$mM19>T`xHtRqxmg{$y2?&|7o$@!NgLd8mV^dnxg^?Z>El
zKc4lX1gt<}A*{{xJy{p(PBy_w)9Vx(7T+(;t6e`5`%BM{?mTcA(C~WZ>{o9$9RhO@
zb#IKRd+Kac_wHn^`04h|KRhu{>K=9g)AhDZ&7YX}>tqi;^TR3-yl-;H6{oXp>@i)+
z*;B@$xqa&(v((fJ4>DCv&=|bWl0<oXKq)<~lr)Wf3*0dmf;nMOhK?uIP0HmB$v;Xl
zo;hiq<;u3PFb0-|UCk!brt4>8?aVE1to`$m)MstU97&ULvx;wIXn7l_boMOXT74tQ
zOfELkypeRb9XZbVc4`U@8d-OUW%wSk&<fEWZ!%k5%!k49HZ$eP)~ut_34^X7=KBDH
zOQB!e^v1T>l+$NFVJ9f#a}k?}v=yna9kDfUC|+3>7L7$rIW6G9>1$_4AO0Edr-#l>
z|6u5N2RfwN%P75UHR9f}wPVreezN2P2kLMj^tqq75MSkGTODXCGvcvh%Jkw5#q-J=
z+g?3n3G^=aqxIA89YEX1NI>h$&rCUx_F7+Z5!Byp<h<6Aa}HeCX>-iCQ?hEmKC)Fr
zdA2z}JT3AV+kPjze4e%q8>d2ilHajm*R3~)loaU?o>AQO7sx)XV`w~}!`-2w&|h8@
zeTEc*4@%el_`Q8O*gMbOVDI*~pNRJMbn+fi;z^U%0UQ%@v?`e4fPWGtv}#xa9x)4d
z{DAM1sshn4(Ln-D(<LMwfSNlyk5$a-%A!Iope%hGXXC`#gBv<*@g$FJ{#rc2;3|_z
znB|bf1-5`9rI+1;8(BiZpoa#9>tkvGWt<(|s0EZM;hLUE%?`JA60Knqfo2^n7;eEt
z1|8Dz7EB`XU6El2!|?-lY@JQ-<QYwr`tU=F1<qd<^L50At%pKy2aZ*-->Qjpuwh#j
zQginj89efJj<XFUPI3k$ryWntrZ%#FYjt0y<a*sR(R5N8{8%@#zOjLI6S*{oBg0;b
z_&$&!vK6c-*z>r`4LDW4jHQw<+<uVeVyWb>T*t-Vz#ECQ)`t^QSS!I+O!`q3(OVZt
zFKKuyA^E(wX#3v!?6B2B4Fa`Qi?{X-(#^!vZrJZ`&iEoXoPJfMMG|R7aJ#QPC9Ja1
zZXII?-G}ArV%VlvP0T!`nXQs=U7;K$Mo}l4mE4`ph<m9B_FQZ@(w@^EGc*D~AypX?
zTY>eET4R{ty-ckwNb1zs7#;Rv*=Ye{y?`unaXZuD->BS27_N44$M5#UfW^*Ah)q~_
z_wKJ}j&H4>Owi$aXFXxoK`kCskc82*>`on)p8GV=_|_&yX}l-3XpqajTQoS@(+A@{
zZn>a_f-zySLhzz+)DyTCZ<1@V?J=+YRC<{p8@(&dZ_?`p+2YQUX`{^-Vw!OEjV1gy
zhDvU;p)|YJTPF}D7Eb?`a6|x6-`bIIlEp>!z$Iu;er%!rO46VFm}LB-$LRFJaLIu>
zh|n|mu@s?%YCqSX-<07UCXLqg+;w$Z-2y=s`NZU&oM1vw&me%A#HQxcOTrs5C80O#
zp6P`Zfo|rY+uS-L1hbyJ!p&|KA#mT_MfclBoqR;30p0sLqBWD5mh>|00TNROS4DjL
z&PmMrOqED>s}f<~G3I;G6FW$BKp+Pf@opLbR`EWA2Q&)<JV@fpXTG{Bx8P<u){0@p
zN?r>zuH<E;##TvslpxU)wMSyU_7|kj1N6OoIGq_+n*sc}{X`<_-LVkexHuTP<Dgxm
zcgW^n9L&3;ebA64wYM*<J_d8*)Y}hMA5&LI$i3d`V=(Ji-VRq-Mv<PK{#biiP6eW)
zXJ3D{`{)dLOREY9Sg&DUEUdsReDW-8f&7GUu13WAENodu#jmO8YJ_N_;L{_q6?^MN
zW!Uc?Z?8W7(Sju+`;tAJ*ECrB?U=%<IJW6k=A7<ax2))aR@l_E*e%fwI}l4$KKTSA
z@iAL+AqQf1rs)@5cJgCy@jLag&i1h<-ZJ^|x7x?%K05jFPGFUfz283e_**I;f8WrU
zn1!&_-t2U5cG{aB_oj!3sgLc~o2ie>@>uW~iQQ80xK8X7`7Y=h?xx6e{x%WMMlXS-
zM18BB_LDK6V~wjc`=EZEm2Ot9v(;6a5~|<Yi3PIQ6_`F6E!jJLWvHZ5+wJZ)h8Ino
z)-=bD9{e9hA>vD98!ntaJ(aD5&(Z3DAk8BRNte&dUzS1H2c`JdN*`7s7V>*IefVYc
zVTMerazkpBW|UR?8At+s9(tZwP;XRS?^y@mH|wB;xBEd1G|E*_I+<}41~cPiwbIc%
zgmqB+tIRscIJ(JZ0plRp1~;2+@csVfKkkK(k|(@?x#yR~FFaGd5#!(A%@}yx*j<A#
z2?5<;CP55*m;slW@$Xp_&a^QQvma25f$IcK83Qr<0mT^js-P)jAZ9?IZx^1qSy0ay
zxJ|~u#|~6)lQHnIgIqBNKGw?>W1w+183WHyd1MECM~4W05J6RI`ntvjmCT~b(UQBZ
z1?tS<yRUy{7Qe~H3-w*Auo7Rm^%@xhA5go|Xe|JT&cLr5L*Sd;SIX3e0r_ei*v<au
zm}PfYdc3ORCh!dw*Fo+hV;Ft;opebh_CNi0+VA`JyVZVo+Ha5j?&r6_v;QrZ{co4-
ze=O^$ls>js`{$bd4=fc*tGTlON$AS3|J55d#-l@*C9(f~3VWAKVtx@mQr-qu==i|w
zW>HLhZsAh4xjlioH<GYh#I$epUY5dWi!%Z4nenCSHCJR+(#s{8CcwQi0gjnf-6pHz
zlUbfIuFrUO&`Tw?OlrAJpV==bXJE^5R=_=hhHpp|{SD=jz3yM+8C9NFWKCKl)8pYw
z9%xXcl&-Jw^Iu61+ntx*!1DbkOzK}2tgkT!SM2So<rBA_dT39zmwmQk13bS1JNL;I
zZ|LVu+5a>TurP%EZ?j{A7>819Fvb6qw`JJ>bT6quzq*%fpx@j2+y7Gh4|_0etTa;$
z@A6azX@=LTUdZHkF24deU2C#nVlAl`OdQi^{}Y&lzyWg(xbfR?iTE8dQ<~L%J|n&M
zHnf#`oWFDP()>=`SLT-yyEMF=K`QDQq>S%LsOYq!qtHUo;W$22NGFScDOMsH^d<nT
zCJTv;<vtgTAnF7ZUfpJXA0~|4OagL=_nMZ~vy-dDE=JhU;?Ts~OddsPjuK)SgN26w
z`H9MtgOWhbu;Je~AK){7mvQf6WD;8kjrg{NeccgXx8f4#2O1}RWJ)p{<>t#ik|~Cy
zOiLrTr1^~^`j+3ju)Pd1<lffmHi;qk_S!XiFGp<tWJ&F!h0=egE^6g;CB%<K9M+2%
zQm`o8kVGLqF#U=4wj7z3kv@_|6&t#9GyBtJk3PJ?mt2_Ehc~F?g($^S0*O-?HfEX}
zPO)T9&AX?)`iGAe+{c!87Io{k$cGF9LO7&4LuZ`O)noESCNMDftrY#Yu}$Eftu#?^
z@;y7-m!5b_`JdiuUwZtb<$u!o!NJOVkQIJ-OXZ)mO$*HZ4#hqwxW~QHE-ha9-7Aje
zRa)74H~CNdT_QP>F$(@f6ChW7P#5o-Y-JkKRDe#?jNWlZ10i6;iQ{hKf-i9t7vx93
zat7wARq~K&9#WY`oa7LH>=0kh2l3KTAl~T^PfsI`bckae;v+)*fe>dp#80FVvmN5`
z4)NDQoFYUoCf~--rpox{9xLNkE|jrOh(8h{{>&=lk~HGO4sp3doFc>@3(<?ocS9O6
z;Shi85Wg+NsY3K(@?Dxnoa_+m9O8LGyiAB*OunzB5zlgnQHS_tAx;yb7n3h9jd+Yh
zJlP?hlndg|ggDVujv?KjN!7Zm$7;Qo3$-4442Uy@IKm+=O(V8C#5E4FODSF{L@y@a
zEosDGJH$l}@mV2WB}6YK-?TL1PaWbl4zU|Y(MKP`VLTO+?;B~vaSriYUW!8es}Q}I
ze4k4r7C6My9O6Afd{~HHOumnfPt|(IZmYH5A#M}m5+Qmq`TmwhT;&j7<wEnkA;d?8
z=*8swT^jLDhxnjFd{&5$3DI%RAy=glXF9~VL#!6!lS1@j@?DTdJl`Q!ImEMt_>>U6
z@Cn6f#4kF;Gacer10W_rwD1Xo(pXszE7xItE(q2$!czEz*YZ;hvjMq}f!fW5`VJP>
za$zZa!oz8-ha6VYVRb6U6<`rQ;gsHey`9JH+sE{aeBICAzY|X0Jg7nO_&&(hkh;yB
zXrj$O4Mt0TjHmXI#|Zk<{BX&Sanvs2iIdQXz6W1Dw=mE&Mz5Xo{v^A-N`Ct@?bbZ;
zf5ok$_yK*$ZhtQK@U`4VOMXQ9+jNJ|V)Hq;m*X=MH79icu}b#Je2R4vx8_X}Gtcoc
zJjP0<P?m4ooyvF>caesl3;Skn!wr_l!`w!j&mA3U*ux%xHTHB-DF^%ou6s`pSiTlf
zd->WIOQV(^Ap?*s6oeHcZX8a~hJe+AE8!Ia*S*A-ZohV*aZylT;|dsJD^+v2I9l@!
zf2jH5Y$n%_h)4H#0-xn7@Y&w@QoQk3`&NfaRuKEFjn%HiRDIigO};EA`C2@}S!VJv
zvLbfKsXI6>-YtLYm@^Mn#A@)$WJ88`zLnd~hnW$}%`9~OnpLXJ3adbyf*w(%QrqI}
zT+2B8ILhfM*br7zkx;yiFcY%;t>DSF;}EuNp^}<5F207NkLHI<p68MiNsO7t-sQlJ
z7{O)lldK&nIf#|fOkrIA^ZZ2Z!h%?0^5~-@iJZ|&RN}i<oC;aDJHIq=<8rScOMsW2
zgi2mIZ&neTExmAq%?&EIJ_tse7ubExT4$4)$ro}#$Lxq(d827^;bH9jzRk=+dPO=G
z*=lo-ph11GuVHgQA;EaASIWa)k;!wW^|zi$Xd&0gFjnVSVemc~tP+&UmHYGhmd5h-
z6AW9fc5}uZIy&4ulUh1E1ox%7pF$d`9U3N#A+ecOYPi%mg;<e+C>ctkog+7wXotaE
zqTvkOQoBd+l@TKw#b}9&*Q#3BvZV`OscPz|JGQwybEYV>0&J`EXfF-A#Q{bDh$Kp@
zguR-iMDOz~$+1|NoYB*EOM_&99w8-;tGrzP%m_v)`>nRJ%p<Y1GqVM7OzVSE>F!tN
z_>~YmN1bY?XL1cRPCkk!OrA$>HJ9aM&HVXt_?GbQgWUmV50F9h<&kT5?-HeEV{R9b
zEuk!ov%jnPEpb^SdD9j?vdesyhbaxL=CWGctm@p)7sNHVW_VXOvuK9=P0}A7iWivQ
zZj)`QCp`zm#m>>0ea#ZC)t_3!Qfb(13hW8D5+|0BT0pL;r3i8zo+fOqYKcZ7vb}#`
zaFGv>=Ok*g*$Vq|I8k0Q!j=!za@K7=ViDqixJrUNHJT7kL`%#8G0ck9=>z0%#j}%v
z6620pNbvVe?#+cNdXv$TWT5eVveQ}vOs9(c*H}=7U#mT}Rht3~3^3(f_|pB^ei!lX
z&R?y|)bqUfQU>a9pbi)B&Oloo$S$MKck!ioQFNY#mvZjhuk??1$4adIrSns94h@1g
zPAfB?M=)2sHyw7nf@;yfwNYpP_Pxw%p3QgTS=hXc0A<0PU+8uU5fLWl{2~=0PtiC_
zI^}PONmnBLdeUK`>S3hi?mRQ}W<a&ttI4P|HmnPq?N49AH~L3f#|>L<^|WFZnS_3{
z*D{=|`^&f|$GOVU-tE`BJok=!R@3IDV&>WVgYLax_C3#p%}agn|55sLKh?_oT3<vO
zlKvR}-7a3)Ru;2xFV%RfSGUu{(^8ABgs)rhlWcUbCf*QCr!a71>nB1*Lu-#_2lQ2a
z`kZ#;<Ff#g8;{BTL>T2KdR-yE<=2(+TX|hoa({*YS^^eEuPr1}SoyW3#0;yv7I3bx
za}7J!uyeuIQl-4U9+X=ytzJm-u!Ft)>Wsyu8p#UYN2s1nLu<9;w~rq7D*wI84YIyY
zCc@rTbzKA@z}EXH2@IegsY;=16%QjAy|&Xn<USWra(hMSTEBZxgbdKw*$vFj6z0Lf
zwi+zQ;VD`McsmUibTHG#Lf7^f%nlmn!QruDNP77WPw_K?DmI`3fp9-D_m{Rn)3+dH
z$X7kJP)#wvm*0`C5hDEr=03r_Iw43awYY@Mp{F&gU4dCc+p`?dJbF^VogUC!dQ!k$
z9?*PxQotS$m^x$7(%kC-Q)dhsaK8skoiS*@g9h|pi$PN-6aw>y5W>u@k4>6XjL;xD
zgxZr0aOYtTx(?f(jn5?gY5qHWyP0pze@C7L?|78kYVC0=>=-J|o)H+ivTG&}Df`dJ
zR=~l)+;SR6Ol{`C{@--<sL=R)mI+5i#|L%E8{fWh&+IigtH-jEdw%*#`{q5LT7yg9
z)g>&6Z)~rOW%&%j-4F6!^I`e;YzG^h@!o&v=R-1HAK_l}U3n~Ps9vn>m+{JUmgK+E
zGG3XHvAw38mSbCTTRZobW1AJY)#pH#qxaf!BxLBe_7FKH_g)hkALKW0Jbe<>{zyBI
zOQWuymSxm^-C7PAiNWYWA-X)EX0NmM%T~kXSi^Bb*q%?Vvg}=X|1QAuzV@~~vs>Gj
z?fF!yXaXelF6qPlwCIP~p0DJdpSFb)N9FOH+|%yAz{kT-@8NR1lH9s7!0l5U#+8KM
zJ?d>9IH0fHU*_lO<<wJT2u~|hPd_|@r^mg|emE2s_h%kNd9BojM)0c<<Zp^DBf_M*
z3)KjH6K0A%$~mNyH?>6r=qd1s|2)|GxU5U#{_}{s5h+>8dFkm{evvh%w;_s^cTM;U
z4dzVOvIEL`ce7wfY!l46g=KYHjYkS5T4rGWnI?RsqT$9(ESAI()>jfUN_8V)p%!5Q
zIbno7W9M5HPE0NFvmvM~<-eVh2i*Lg7;}U9DH#7LvA9N0xFuoVK{FH{h%$ZHMrX~u
z<C=PBUv2wMJH6Pq1N}E3jAJLycJU@>ZPy(Cx&yg*ld}d1^r8c~c$2e^(bK5edSUTz
zC(k;`i#NF(y|VXT#hV<bc+p?-=KgTgtF?0T*UI)18li<<1!K+;Kg@XyfBji8xD(Uw
zV|^1oVcVXqdnHN?<U;lTYCamQuhqyiPVD-dH^(3PGBEdk-jV*Sn6Yw~K@?ZAR0j6W
z1--S*qf8CV(2u3&cYAN_(!o@~=q?M3YX!9TM8icsaNjeHWPa~v_%>$v=K@VLXo0TJ
zQ6eMI^!&)#NLnjzp<38iiK2ilc|_s#uoN7Y-7O3f5nk~w6v;s468cSOlE4@aG@Jt*
zWeym7O7%|W<27M?&l_~o5FSDC_fj*oB1w<1IeJ_=sl+I}DkVxL#$2a`SPrvg4@kOv
z^7QfFz;A_rKEJvC8@$+V??}kCS$yRVnZ1?rFk)Q0)h(81=PI6rd@in)L!{IZb}Gpp
z8@m53exvtmb`L$seXbzP?hgXto?mcmZhug*TkIZ)&Fl~I3xr=FHm^U(FA#o#*sT5_
zzd-l}VsrX~`~s0jkpF(N^*_k7|9*nOz+JWlnzX;H%SSz;#O9^6(y&;~SAY{7A9zUH
zw-KL0^A<{=n%63UYF-Dw1@k0N&6hki-{h(JCQr?mJZ0FqhMjBJxdrox!+|`tO7hfd
z5p@SS;C^@^)Y0|x?X>2c_D=e)|L%815<S%!QZM6ZG-nn5$y;V@n#~1c(^q9Fd`?zj
z&*C~hjTmx>-*kwLgF&niqGxeEA&vN1hdA6J{y>Nmh3HvacMeHqy?wJ~J=h_hBg9EU
z^enE=rV&><#0^|1<7q;y6ryKwy(^8l&>=qH5DSI)9U*!a*DKSAKXZul9Acgjzbiz~
z;yNLXc%DPN#32q6;w3`#EUy2SM*M<9{HjCTH3-BiA$k_qgZ@;l_p{9n>)ufg@hu@%
z3(;+G81iBoah*eaiwm`WMu?Mz=viDBrxE|;5FdAli-h=nA$k_qhBV?Fhj^ny{DlyI
zAVkmNIw_4<;Shi55T^=piV!`E>&P_XmmOlMLp)E2KNO;7QmmEPX~ZKO;-?+r2qFGR
zh@Qpu&B3YWd3~cb&j(zn^%iXBkN!l6p2hXyG~z=JG3gN3vb6B%PeC+`YfBnyzQel1
zVLc+O%Y-G1>yOe{QykWn4(m2yO%s+ZuAwy6Sci3y!}{dW)L{)+SX`HMVR8N4+p@S0
zu(vYYLf9{^S9@?>#I@o7Vx-<%C?Vv$AkpEl2VR(8mN-Yz8bH}8)S6c!CD^fUGd>at
zoc~G`(IPHqLO3K+L{(tkO{CyO<9N_+uac#8wQf<5m5~>l;~hZJl5;Se4mq*!Wz-^Z
z6aaCC*0by#IYdfIv9^AbM~Os*ekKPO^W(>WBF*Dj_)?IS1|eW)QBOyj&npd=tO#6p
zlK~NGWMJ*!jch~}J}}NYFC4h;H@x-^b-DTTysWv#-_%-n4l+le={f)s+=Lw_HYK4b
zi;;ZFXrV_}kP`P76!Nn=+_q^DQMO@lv~3@Q>&km;ACt(p3Zu<eV2h!c6prv_BsRWO
z_vUR+G)>)W2bo$)MI=TIxSrv(ISlUg-fndl-eWYlDIt*vBvj2a#VKlu(jpJwmWS9!
z^ZD7F^PvWi2@1pT(lR>{Bi0<j#*9wnrKg6AHUE%ck@&4qY3@hszRKN~>RuaMQ{Uxt
zzk5$SSr%{S5RrF7zPCdqN$ks=?G_{?m<kdi^V}3QBBrtlT(yZHJpE_|3@Iyle^z1F
z37;cc$ga+PST=Yg!>rJF7f1}^mDZxQRJ-d0^5!g8-#)pOwsjPW{z}6&*kheXW;ii{
z#(*quDq@L3rmhX4_+*87nk-k?%?+D7Lh>c$ceUqcYJzdw-9@d`eGo}SYF?0ms{)Gl
zhikCIvLX<x*{fq53S;p-kzs(=7|GK=e1i+3a<A7d!b?>+G0hh#+0FnV(BoE4=Kdvb
zSgJ1itz?@i?15+!ATsQ4n6)MInN^S@S12s;P`T1rK{aw>)O=uD^{mZqO!G&5b`74a
z(}~=RI|A;s_zXmAIJ=pcK5AXg!x;9G8Ii+ndp<WVe(=1+@V_cE_1hAXU|Ickd#K`p
zEgr*4^`@Q&6-4quY}kSD?zL`P0F78`#EK((c=wA@-+^cx;~#RJKHxfnOd}0%cj#WC
zke!DaiXUird#KPx^KmHtV#C`Q&t<QQKNlIcMWoOIZX~`fqSz>iSXJh$+m<}=!cjZc
zQon=OTwherw*ZC=T4#X9qAE*FZotv_uJACduCxj;d-_06Qb$7ZgAMO|UY_%=YaS|E
zp)w959J1>NES$fs681E!gY!oUV@gKOdrxH313c2uUL{9YC!MPHiM)F$Gp>)LZq>&#
zLBb94R#IX}@t^^kO=-fr;gHumi>YTU;oF4!HhODRFBjRz*WbDjsHh7k-RP~AdXukC
zq(P=Q5jk+*DnYjb#rG=zgsYw(loe^y8OPYxNd(uuFI)<&FA@R=Bp9_}5;R|I6>lDx
zqL8o}BJnMjYgkiKkr4vbX%c8L$me}ZFQyps;-X^e5{Y-RA^kMUh(RN@xML<;lK+IT
zv#h)i71fgru2^RZwgi~k--cBp$H-5$C<YGg-(Cf-YF(@c=bD@{UbCb^wt_nsWmHd<
zGm&$z`-ojqmf`iz0G`OPc2(jXf|FSauaM{D%wcS2B0+p$IM>PD+*Y)Z8};o%%F$hz
z+HVTqtx#SV^3Fo6=kq{goCrp;0n$zyqT%>U;Um|`upg%Vg<YGEfuEp2tP>p9U;p7?
zj`+6S?nKUL&3ews(u!JsIQ}A0L5c4li@&Z`g}1(8ET8-oPngmEN`w=W61a+y$IX_j
z4t4Uv@tF1`Wq!2EN+ageZ`o?~X`}$&Vw+J&#!V-X=n(KIKd*sa?>6iTuUILNo|Xl!
zH=fSTL$)mpHexi+H&Et~DrJP*rE`j-l~3^&({2Xg+(dL32yjA6&pGSUgP({J3ls2a
zz%d^ShxoY~My<*m9bw<B@V-%R$$Qc{!FK>93egao)LQh4fn-ggl4>x7eefcSX|%{*
z!}|F~4Ja)q+uc8-7Hs97df!&mN>fHTh9nCB3!HC65o5M;A5N4}lt$<1OnEJzgdggL
zqmwv>XJ(iQYN^8RE7JOsp7C#oHoW<qqg6(SeV#>ytI8eq-R;(-xGF6@Tl&1Wf^${J
z$oxzbPp~7CQd6Jh7-+1)cph!O49TO>21g_@wTT9K$ar<A<fYklp7DAF2^CJiSa{P^
zX<w&P^LnT_`YDdE=y&>+V^}5n9getZjjcnUO;K|Y%SI^;^jCKQq2;gMs-5y%?bNJq
z=mS3NAdA#O6J0e$Rc&_WA`ut{bghj{0}MT8RjKqMov1=iKjC>=E)V;bE8IehUeFl%
zI*7_v{6;WdQQo_UF~TzS_gprGOKF$9z=FCqQw9T#f8qOB4Nte(A_`r-q3PAx+6M8N
z<em>(liNN_AE2a|zA5`H@qU7n>lN=OIJI8!euB02iuV&7Q{R5qca~HV>i4d1E-5F}
z?_IGaQ9}LRHExLu^&4TYQ(NZIB;7-ix0M&R+yRFsg-Uh=5@jgm5@Izu%H6c=yoCSr
zhwR5{!*o>NzUlHuu`~3xMrd>V$vq(h)w-0AS!wtP)w=uB>VtaqLEnqlc%c@lhzxs1
zQb&F);jfP+3K}#9)S@oKzV;-)XoIa!ls9y?FTL`XXgzkaA3hqbcMA&*@3$|#{FYF?
zEh$7BcqbIIK!qA`AiMk)f4xjb(S{!PN=q8S@&+8oF2AL`UW*w)e*-C!Mvhx>cm{v4
zQIu2mUBo+omh&$!Hqsqn?@TV9=qwKU)jd`UY!Iz>kFA3H4(|O}EOOaH+t7vyZ;$-T
z%aP$AGD$XShe8pbBzFnVny*DDk|Jy}gr12>q?QCVTQX;9;q=MQx5C${5y^q}^}NsW
z%CmT)EFg;~$^x=@qU<IA)G&hGMW;WJ?ue|~!c2NjJ~)w(>$2uBy4ByUb??_dVHz3}
z(U%}<<zA)lvDL0nPfLcv&s=J<iBGL!li3<RzOIpLlLs%m-b5yDZZcs?E!x-L>{mMB
z&E<$zTVB|l{OE;OF@{b&?WYd#5r8cMd@U)!PhMCKptpMaN5yBhR&U=@eC7(djeAij
zw!dxA*Ne|wEvV;szWqgmK39C^dO<zM^X*B4K3RNbhoGM0`Q<pCS1veEeL0Tjl?x7X
zJ(X)O*WtMZ3$%Y$WjSG`%JR_*&r;%btjq+OMl+C9=Pb1jF(}j^bzkz<b7l_f7^nFD
zHgVHh5_pErC?1J}`LK@enXv9?$(pOzV}zf*V|hq&1mPw6`<E{PNR*u+j1_&<ul<sH
zFA2@T#>X#qJ$|w1@r(VAUu<^#VyEL5+uSAoIga3$WyF`5M|m_wD!#-@*$*#)Y@Hon
z{^ioQQFnf69nWlG*ifL8i1;v*ktNq({MN{|oVPH3!qM0#PWbu+4g~u~|MIS?WL+*d
zXA3JPSDvs1{*?Sk3n8N}*T5Y9m*m<>A0=O{cn1o!@F{qy?+qlrNcPFS!P*5oEh^v4
zFTFGUisXumcY#&gBFkj$f}SMM8O6zulYV!_OME%CMHbB31^d%!R;SaHC3DmQB^hYA
zT*k`~KjIIh&uU?A&MqteE|q^*U-?J<CAn(SK~Bng{*aQ*@A@7|w0x}ZH&?uA-)EMh
zTgJ<$_)&BAjh}(dPu@dDCn3;%6-)VF#(#TqU+|~`!FGI?6WuxAZxe?_IY##l4P-Xb
zFh_gzx)UC<5cZ*8-HX5JSNGy?`rSD(PaGz%kH3lYOU(@@YMZi**#!~}P>z!C$G?=9
z@h`VH*<$G`C&>dlCr-B1hbwhg41w9!W?2un{AAcg@x`#4=3bfYv3FvQmcnz~Y}Hza
zLWZQr5S(2Sj(>=01o3eRRLp5eoxN-KvA@^iwh_TiO3a;0X{K30C!BNR#C2xJ!2Cgc
z<k+5s+A0kQ=E0dZFi;{ih9sZc0AXu2>t*%<EMFknM}p}g@`DKuH<_6vCdU?B(_V^U
zGH?735|(qJWX9FpN3=uRMf{32-{@uzw<Z^Beprv~bFa?#f6jGl&H0IfL5f_~!ubHf
z=wA>Tu3i*<Kw?F%E(Sz|5Kuw}z*YxDpb$_Z2fz*ibN(V&&I1DFJV1yt_OKnC{)Z60
zX8sVTLWE7?kzJ!OkWf4xZN2k4Jup|y&<Ku6|KBN+Hd#rh^X4yQP)_-{RQ|=vF!?W*
z>u_?dTU7oaMXy^_`5=X@TQvDWidVO2_=75nUUAVYE_%gf_=B!|JL~?SQg~BbMgbd}
z8A1`?_O!e;2l?jK_Peh9WAyecEx>RgrZyV~>)z_RBsJXn#+$V-dxM-p7nr};BXS7~
zqHE~8EA2AW#UkNG4#*?`p=V19)05_(!)KcLOYPa)eL%}B1m1sq?+;F1z<3N!UBEzq
zTQLTMV-{$!kPT(+n~()t7|_UqEgkKfkOf;@oQJ*BlbN!hsn(p$NSOYMTfNdq*+wGy
zI7K>~1c^X%_4mqlEyd{&6BqTU{f@Z6Od9}w{DhULmQu-&h!gm)v~oJYv-qbNU%*FA
z@&jKCRd40b7}fQ-Yk;(TVDPTr1NB%*CVpV9S&nH(O_AOR%vY28-b=|x<j?-Ay}|Oi
zo$Z?t73aLwz6ohkvt=+e_kBusi%puKZFLB?_BD5>d!xxEI=9EY(R@)pcfa10YeEhB
z=aSehZd#Q(vcRT#onzL}%lRF+4qapHx{rQrD5xq<x@4maYh3DIV{}1dbV2yv`;}{w
zZ&kjMe64b0@`cK-<Ui(zT5*3=5jq_|{Gw%;Za)<td2V~<3Ya2yrZ+9G>!f_si%nj4
z{lyu|frfjt)yyzA2aA*}3pCsUOpBar(~F#(2rD<F&!_xVkQAA_C;A)6!AXsBL3+n6
zUOsF+=L||-2f4m_&zo=53+kUJ`{VXbG+Bm|Mp&=p9tihdPuW{rubj&-U(rwC@C%OD
zX9l6mfT-g_bLlZ~_ywZg3(ck5fba{%bs+%%9G?BD-FI%iG9~qD^>${U@nkynABn!3
z;b$LD^ldQ-^1~;XWisX6*$&Vr@4hI&{}b|Vttao+dh%|qC-08xlXqWHSv+|+LeThr
z@@}-m`Gf|V-hgN>ZE^DtW{MPr51@ZfUvcWme^Gx5B<A+<0aJ;-m&(UP3X{*Ae)eC|
zho1NLqnSK$(tz>PUmwySe_unS!pOm5ub2s<UBVBHXrAB%L4haM+buiIurjFJij(VY
z*(QMwjWIr0!Ify1-CuCy!Oqw8Oa|_VP2ZbQj&~0S8W)hBw8Rb)i#7ko=KQ;^&d2(I
zVc~n{Cq{je4QMut|A9%9S^Q3$Aa$985by`4Pyuba)Wv{Ir~=yLsSCiIKiGVKr{>d~
zJ0lJ4g^`lirvHJN9-s8dMHPX>#Wu-1YpXc;dP;-__**W<-zdvq$|B=yD}|;-)BH#o
zlv6$~m6V9f<Ud%h!^xHT{w|8ne18{(Wxl_Q;xXUfrK0E+7ro-5S6qhQrSio%af|u>
zAs4EO0<9od+KR%a48WW%WfT9t0rc2Dhxs2-<H|qgJgn(I=V4-##!7b9z0-9$AF$O{
z`dJ`Rg!xN4+J{a@!(#ik@jrERD_s~)M1xhfuaH$i>-d*+;OcFj5ek!#1fPWzi&K9x
z+Jim%ME8sOaQE0`^K!K0<v`;Vq!4p^0o0&u9i@wZDo>@+F<+Yte(LS_rv;-6Xf%3}
z3r(Vn0ck=3t+{kDAPp*@HJ&a2IU5m}PDrG`7bpm7gpiqkz(%A3(cL<~R-{@_=`*_f
z4j!$$^=^?$Y(-N&t=Fx6ltH;GQKc)xWS8r3a;3k2MbYW+Ur|{4`&Sf?{{EGUqE}q>
zii=)x8U8C5r1b^$H`Bgay05<}ZK{=Be@Btf?*}K7{^;*{*5ALd{{96Lcnp=o>u*GT
zCETg1vJlnE)kKiM{2SkMZa6w&7mJ2+#!Tqs|A#yiZuy8RxlT{RfJ`i0&RAl;eI`t#
zC0}Q>gahHuQyO*xGHQWR@I4=MhXWGiDBnNYanEW3E~MqtHv0F;g|zdcsKxo<S9?IC
zPqsQ3QekfNu~rA`a4=*<U>y#oK3asw*Z$lb+dMCqvOk)NW_GARVj|tENiHq2&r5Zz
zK~0YCSDtdpLlG(|0+-28T1Hc!BqKVM;&G{@L|h0*qRVjQ&KrJt!7p%rhg*3fCHw0B
z7-h$<$uF+48+G2Xqn|eV>v6=dKjti9j?Pg~iwK7TzuTqpoXP$rd1G$cPb^C^Lh2-=
zpzV1M676CC#1LZi&WQ|!<ogrL5u@LlSYU^X2|L%YZEy5`FD!&&<~^7es{&1X*?ip9
zsiT_)uC41_T|6QFebZn_Mq5co!w~RciEzP5!?y*0iLqzmq~STc<fah=4zbW7{+1c^
z(S2ut=sCM|ZcL^4&eN9SZZ4$wmJkmJ(Q|frI*s_GL;Sl#TqDGTLiC(new9YN$szvU
zAwDI<ULktUE<Z^l{@5YTa){p);-IgB=sCND(uiXn;zbVet3n(sL^qU%d^U}EyhA+2
zA)faph<+hHZDVf8_KsAow?1XH_Bq5a3-L%H{>mY)Od~FLi0iq~JTJclVh)Jr?DD%b
z)}0RPL5KD9%U~TPEIGSeoyPjP!@AyK-7T!6g(YW~i_=)&bXb=<thlfOVBzfY>@zsK
zJiK(DhVa1sg~qN}ZDXOVwy4xRT^f%_ZK1P6S%lH&wc#iW3rcfog;sxNHnq-XE4Bg3
z1;__rw%lxXTonp3A`IwZ!$}3ptPtpT28uRMz@v8U>|g4slv8Z*AK?rl`?k?rE0)2g
zW@e={Muj)p&cfbbX>b<l{lC<eMMHf(0?Wh-+cO3XmWWTsRxXqYZ-_zRws(lf?|gkp
z<k?7kCHT)r+BULg{HiwB$$jO`aNA~3H-gH{7U4mTJX(wA87B#?u`87pTIJCy@_9PQ
zPuWvitmPnaM2xa01s=IYZX85}!Wf}N#u4CavlcDljO=QeX<E&hBCjDSAl*}3|D*yb
zu^&jaVn4`Yx9`iB)O0&VGM!Afwsv&4nRbb+C#ga*T?BVVZWkHC^O8Uo8kCbzr%Gu!
zMLY^e3N7Mw2+0nVaNZj6PS;wzTFooDDb(v4xNj<r!&9l%!X+tp70ZRIGS<4WEGudq
zD<1e_+q49w4t7sV%^#|tjX>2ewc@bYJXs??PQ7rgp<4DRIl{U#eP@v~ArMr{Y`fj9
zS_Zl0pg`3!-=MoxOClI7akcy?Joq82rCPmM$btJ_53|g(-U`ZITEP%~Rb8wlH}jr_
zQG3ZMv0hROD{Q(ABv!~}KpCnHtPJseIzdDwQ!}WrY6dHR+!lDZk|z5iZcx%1E#bjS
zRtU69M!!!yE#Dp*$)-QeHyE(Ajf^&!ptQ7@>d_F5R{<#Edr|4I^t{+j{34FKO|&D_
zp_*x$VWopZc;OW54f>LnQ%3|EB~NPUK5*YQwSrYRG3E%}mnvUYQ&{mI&8=#zZ$+##
z?b*{CsO}VlyK-nTTjf*rRrEmOD9z=f5kcvWVqLB1f&1EIuPiilx6G{gbyczW!C3r5
zHf*Wa?&odykNe4|gT2K1wIfF(t0|v;^6*WmZDg#X`d*LuUW@sjQJf5xrK_TlONFRV
z*#yiI7mY86JseOQECMJa8f~zU+pC*Y?$Y#fE=ykC@$nIzS6<A);?ej9KIY6KTub-@
zVM+q{u$RMgdh6-#K*O`@bQ~aF3h#aiyQ&1c77!I%`xLh87Q)g7o>zd4!2fH^8H$lZ
zl%3v{#ItBPh<OTbVpJ2D`#H6-j|h*^qN*0z6k5y%u|QUgkX9o&YtU>KT3_JatQ9S0
zaabTbg`bZAn}MRmZ}(=JXvwyF86`q3Io$iDgA)&UE$2b}5#qtwr6imfRopKg7xK<i
zCht6D^3GEoNIeBlA<~3AaS<@G7vS6iPu|*7$63k*w_O!zG?!?F1Y>r)@|D!Y`?E)E
z;+@BZ`Lq-d<VWifV3^&uFnBuGKQ32Jr+;MZpKo}7rcG&gq6%l6;yttfe8w5+@>FRa
zk5`4CNp(EFDDYiw3U$Li&1y#GbS9;-2FMTB{HPrFwXk=N&nI-R-K^b2PTcCYt9H8h
zd*LJBv!lH~;q7)^-F<}K$1$rjJ1tGU{<QZxgd-+0XMWZ-pXn;6PhDDRXkh-=0h;Mw
z$1^KJ<^0pQ@u)D?Y&<0z-wh4!54XKL2p2cmDYL?DMMZ;R@y$A7G!%b_5RcBcBSe(E
z#5bxuCkxJqEW@pUVKK5*9({OAVt?p9bAW!TK@gVF4gr0HsBl8W*CX$?(ul8D=XJw>
zf35xt+}EQN5G_)>ewBE*_H$ls%Q;Lr-o~-R_LtVjQ~P}j%Wwe<+$ZN$GJ#`?-H|}H
z%2sMoIdKw{q2lvOApO1U6~3;R@55NhcJv$CCrSr<eRYiPt*-(pdcoUKS%zSC;{SBG
zvMRJ&)FzH!+g~_+cr^YYt-4qHB^dR2SU>Jtt;!cOx})qtPYv)lY=Fn&TO;w8BD-IT
z`QFjyx-W&B3-k#?v{F;)?u_zk{QZZG?_X;RcJK3bTOy3_Sp1p5+!5;Cz<p39MXGZ0
zfel8vBjoyjZTH$5nH%~w<CamXJ&`;xaK#1O10z>#z&!-*LB0y0<=KN=6?n+!ft`MK
zU*N|13&guyOkZZUulYX`g)r5A(r<2EDE;PE26nLgRt9#k;8u<ITL9-qZvmWJehc8-
zf?IUJPo(5kjrR=}3o%CWwx@^j9)B@4-mm+MjrW_mFy1G0Fy5bV<NW|by|eR?e=y$j
z8?KsgIuR%GAo}>a=CS2CKOY&m&i^S%i96VQ_6Eyg;Tr7Yz7@S4WesQH+8yL&DrUr&
z`7qLab|HOzcD{a!*@Q9wc!SOeYSHZ|yFZti_aZ!J@6jwCX?|RnviMd=Holb*1svxq
zUJ@Y~LgJFDSfZ{9;SiNEf5`kl;FHMBQB8Jpb^dX)xQWE~Ys-Fi_h&9`i4Y@E5x)uI
z`eMZQw^+&ZfkuVRC~N2`oc=||D#Ze$hf{WN;dwdA?&j>su%(eXd+RAgpy73XM{bF4
z9o89%zZcoPHpUD@AcoqMKOY?&ELfcED1Sv=U3o9!MKavGqV8R0v7yuF*#1>}zutSV
z4=NY?`ND&k|4QZJ>^_M3uv9M2>4RuHE?))vAVCMo>w{=3E;#vpkbDOz=z~CrNhX}m
zK+|N#Lf4y6%K-nJ<oo574Ga05+<+ooKDD8P-%$-10&**_m02Kn^0hJx<W9X-W`SIG
zKQ**MgUjmhCVoBKFs29^curI}mmfzGU!;>$BvGKSD2I^55e9W6v6sj!5h9uVYm$JK
z_os;@hkCz!H`Z+Ym%pv6w~G%8N&Xj}0-&4I%D6ES8D7+#e>nP>NeM(BGmaU#rR$Rj
z6dC;GFcgyqWiB1LB}FoGUx8#Y$8+sH4BX(Xd+>(~hVH>1E|_yq^|r~yOW*L><y~L(
z8lG0W3tM1&79b2&Z;KW$?FJ)MymSYbF~v*waT!&-^jR*m?k(QQ@1KkJ<GpfUaXY_%
zEdHB9DHeaEK#IlY<x-qX_OvK1JYR~RQ+L(=+4^wSpIh6@X61(d+@VVjXWA#}!)~{K
z+4_+kZHRDh(1-o!lYt{)aBZmHd?At>wYYlQdD%->QO?=%;Md0ej=+-Jeww}X?*bed
z{95Z>^aP`2R(iCoWV8^5;IdS9ee=(u>Gh=38h>PPkC%?nY8U%6^F17%jkVgvG;Uw{
z4u@wWuy%3Izrw@8EXjf=<4FQW*L@g~9m!WhLMf6T7C0QqA8SxY@<Z+Z<B<Gex!}w@
zJuY~e-M|Ge<tEJqwWT*>K0lNn&O7@5JU_hnU+}|=58;O?NFP5;LHhV%3ev|9Q;<G>
zn1b~2!xW^CAEqE5haX;S{P1GqhZh?^y!d~}52w)GDdNaE>Ttxd3%B%0vm=i8?Eb$)
z9OkD0E1YnLV};^=C!Ebgnlz3-?EE;B96#>-cogjMaPwoaw?5CgIoV5d7CTn(g=c>2
zM*igV$bVDwWMFO`!^iAtefU?zgBvxv!8KrXv5!>xpneZJs2?=jgAVQo&GDfAe$b!?
z9nue)=RuF?2hI1OL;FGd=turzUH8Gi2F$0`+ol#TjW!(MGj7To4!WDlhF*75)W8+q
zy_w%X6?gM{Pw@_Z7Z>kyldbXVscz1#b#v~R^qhNV-<+#chq}JRk7GcmL02gnTl9&;
z(b$a!JvXRO+1GMk7Jr^|;*qn=R!}JqRP3AK<{ytnL?Y;;C!Bp-kFUq#Z$t*$5+z2u
zO#Z@}OVIzj&hI~n=hdCa!8!}DUF&!6U^IxHHM}T>;D}8oyYBa0LFH;f<02?7RocBr
z9<H=+8Pt__J@=WV)!wf#w!!S~PeIT9atSov1M+}M#F}eHBka9RC1QhrXO^Q^iQ#Vb
zP43MN-Hs5qEK>4j?Sq<D{I?}JN&IK;bmmeg+h(@6V#wvU4a(|X=L(;fQTVSNw($RE
zP&b@EYxn={;Y?N2qxZ9{KOIhfS(5$n$M46HZ}S!>-@9bB#$tJT+-AzR{obp|%8_r=
zUl>0xN4`x%jGvby-=-nP&&!c-(-7n5<%{$BAjZ$jk#EyD#?Q+a7xY2;#^XPfZ_%lc
zZ}G=LzTF?3jC{L4IJ_bG{w4V1T!KH&CHUi9!YT4BP&#?H(v7{zZtM+DkG;S4jlCx|
z_NJ%D-Y|_U7nX0zP-En!h`b!&POoF8APSYa6-VveMbeG8M@nXd15N)9B-}h>M5uW%
zPPPvqH=TRdD1vvK3q0C<HU|B3v75{o9gD9*m&cC~XN@o0yNe1TXd+pkO)4h<i8Y@c
zjGlEiyIx<@dv>Qbt+Si8fu>dV{-WaOSr-)tnx5b(HS}~oaJ1%pfAcljDCBJ3T+8Oo
zfA76zcW=06H94*d`BsHWI6Nn|woHrW=Yu~^D&2T}==8Q&vyG?_yoAVhwiuF_fXy<m
zsEiZ9qr+Mwn5Jj~BI@SlgyTElCa=U2wV%(4`q*u`DVq3N*U8zasbAQ1GS0=HVf8j5
z{cTs+w}TU#qL^B<1C1{Yp`p+<2G7WbTXo%q#BCv|O0%@yA0p7KMd!hJceR{%^Def5
zd=oE`d3QP-FVxTBae5q%7y0M#I2PH;f6qV8%KvA@r;18E{65nI0gnQW@}DGT@IB`z
z{%`W$1w5+i`u|SCM4|>~qM%VxqDGqlYEWp2fF_dQ8J!4LRHzgwXp6skL6`wlE`dpe
zaU7*;TWzh{+V-c_R%_J)N=YC?0^aa`s|E3P#sLLX2-m!y@7^<$38HO%{_pdDpXbd3
zbI#dkoqbt*?X~y1?6v#<N%6i_F+4mnnY63EL+6_MJ|kxIbC?m)z}JReOKi?i*d{tH
zMPXw=1xtnfQrJ-QE}O#o3UL_<8_6=!QrHmEzmCHCn0MI})`OHz3VT5BMPa`(H&EEm
z%=<Kjg?UO-*fE~}j>0yfYcv%00XP50C~U!Tg`%+f<DjszuJX<P>Fvjq{{LrsQx~V`
z?F>-yb@X<ud6!LZg+g40-qx}Fv-EZh>0d{0x#nFqy?urQPA9#M(|gg|h2{o&JC~cf
zXqw)J@|32xKeIgjZ#dd);0?V6_2K_1j&?+!gG6t>KL07bZG)lx|C8RFa7%Glg94h&
z<xOc#<}xS<*NYdppvkO(e=V7THZL8o<rf*iJ1yWJEa34=0UYoh08{Z=uFe4d)B?`3
zfPDmbngCPrT52+YXIsDvE#TTE0RFE4Q}J4k&Hx^10sC6O=LPtn0Ov!+G-m7abYnLE
z#xy3^0?rZORRSDq0as@LTP@(5Txj);0=!y)6&CRR4B)*M@KFnR+misEF2L><a7G3&
zWC3rofIk-CLjp{>u$-R(3|PP)TEJlfyhea27nTz<fG1eMffjJ%Y5>2GQ%uT*W%obQ
zt=`GfnJzro0<IKb-;)4Lxv;#F0er~<Zs0<ze<i@t0!+EE{3!$YpaooF0k0F_908_W
zSmtE_=UTu;7SIvkg93Cl+@kqoGJqFbz@J#aJ_3AMfGHQ2Q!{`kTfj3d;5M8+m+Thc
zFKmtnWB@%Du$KjVUVtZoN8`e>epwnf|N6DT&1YO_^8X6zM}m?I%hC+gVhgp}Ld_G@
z8-kJx%bgjh+bz@s7V1(#-2)UxzJlMQ^#A<6ed4u@GRD0}sJM(Rk?n!#k6A)a^o6E#
zT$?{SvY03)#i1+x(Q``tLq?Q@F7rn(Df17xxGZ#$<+i~yS;;o@e+B+1r&UMdIzPZ4
zovw}FU;4Y5O)Pw)$JhMbekr-DCh~E|E9P8*O`+Eq`URasUoKaz&!iC(|CK?^>s&z0
zQ;*Y_oWZFk|26}4pM`qNLQQ5L-;xrbXmYOyXmSos?zCsx_AaCrsp?<nk9=Xwl-@@%
zTJnRnV$QfgB<_#A;E%jfEuR8hOgw7YJ864Os(qhl+Q(jdzJ0}y(ONA8ae98%-|aR%
zV%YO^yk^A9cS9?`+Tz!qPPh1#yG@Hf;6jV<Sw&keux;`Gj<yWESKU|DHQr7;C753P
zAtGriI?mq@R)@Z#80R!@FgE5t1I0c@WQSk#&WVmkrntC@O0G2F5^o}rC^9nIyzW$9
z>~NO5vWO0^sE!^+RgU^43Zx+(K5ryl({i2SjuTRZaLK>nW!<Enw<G5S4t&{IYrO+r
z_HRePX$~HixR5V&LYf}EjkPx4Vy5mYq>DCLtQ#v#;)Nw7yj)j#A;lj{3gOXxx!%S~
zd#l*zr`X#PZcnhcW!x5XTgO(1#z9JBe?wz`Za7TGLJ@PQtZvA8W!}bKdQXfIqL>>q
zXtaAx^qhh~^e4sP_Pk&vfq8z4?U;bEjX#o>B37GARC7cc6M%=)(M${;;k%P!6N~QI
zrSNn5s6QV$*7MUWCx#>Sn@Hr{QLGhs!$WzNdTzkWi8b6RNPtUbh+Ll&-bTUHa}`Xz
zop9vXYzuPR($UnLy|J}?P&cVb8|(_G)N6rBM4s7H8%gLe#=1$b)kKoUG94%#&C!hh
z$Xlt{G8Hy7w7hftF%SDw6xRmVEbHJeQ*QZM8*7~tKR>N2-)kJSa-B$0{wGa*=9Cke
zoNxxJshqd^=;OuaMVtJx>UB$a1;uVj2bI8f&pT}6DnM>v=D8sizuh_yotrt(v9Lbv
z@kBs~db-92jncyeeO;o6B<m@LUHkRcydK}_8?@$TG@MPBbg;!Q|84G_n8#8%(D$oS
zv06k>S=w=lnJGIuCU5+B14QhZEZ?$XIgk006=wgGFk5BJ<cMKKo(A1k$VN{N74lyZ
z3cU?Zf%-UoG7T}!JI_!qbE4-ms%%xANMR`znkNUOH%(=FFC=W?9N6_u;t5Xle6a|8
zh~%!|bf0!KZm#WRtCEwnzkkams}q}8jHdLd0ir4Ow(pI;iW**4v*-hRaqUG`YZ5ox
zi!9YVwcCrV)0de|WY_^;4>fUqZSY830Dp#S4RHz<p|rcqGvD9QR9D#=+G;99QxDK;
zI2I`I$aCsJ>KurAO6y`heX1Tl1IK0TU1VRx@HfZh+JK7Q1|8i^x^Em?ab=NfKjrhO
z--)m196W%>-EQ{w<Eu;Zw5H^xUFEU(hH-4{t}cAX?8kj4Cw_f=B~>}5<*-9?&LKGI
zaPAi$<}q@A?{Iv>58h{o>~mkq*c+aN;7w-i9nU(^{`&X|Wi}%#v!A9i>lx<I*SE>+
z*+dzirun^Xuzb>+Xcc%6?Pz8Qo`(?}TErJQ)y8Q9DUaX`)rC!_@=e-I+ypK5=uUam
zoxs37b|31Ixw7XxN?So`WKq=f4tDhB{~!WQ?;78n+0jc;qw(z=EpjYzq7YSXn_^As
zFJs?H?YG56;6LBryIS!J*wSieEN|>)8Ybec*OiuPgqKXcBD-ACvR;4d{k>#=7vGD(
z2ImBi&d?8=wF<S-tT1e%6RZe*%24s%dc|(zx`B^Ocu$GWgEt{kjD9*B85+4`y5K`L
zkNnn~K;`Z+0d2$^W41hxQD72cPHH2vNj7_+abGtZk!@fHZA50wuT9nQ2&*4nI>u~-
z^QeAos?+{sdlaA0Q$|>%!Dym7`@`tNAC@!S<IE0HiF+QFj=FMF=mRz7RJ?<t;zSuj
z5<lD7X!)7kG8clT#FzY}^zkW<2F>9k*OrFBg@NhLWTgz2q%xRxqB?SsFl`COU)eIP
zNM)w!Ht{R-vJ;EiYK8MspmGAPb#fB!f={%i|7ZN~74!OW`4k%p=&X;op&QI5Ll4=}
zSJ{EBjfq|CnKYJYAI)q}Y=CNPudqAOwYw9|+>~xlZ&AktCpwUUli454_y1qppUCP(
z@l*}(?LO@kvrK4GPs}tEleRrg)<m}@oR^b$g5Jv>|Dr<LDCW@owO{_7FSGW~D!zno
zG`lg0*{_(pu%b2Q%C1e~O(<3GH=9$i^<J?D^zMty7kX;%^0Yhk9h@yIOB<hJRcK-L
zws`m8cT7duW%HP5&_t*p-@=9Ij7*%07eh(0z{F46Vc;y85-R{km)vgGPomB&{s$@e
z3h&hK=sP{>e9{e4a^TJaC0htg5S%KXYLb2HOMhe^dv<*+?)<iQ*LKN~b8V<zN4J&`
zq3q59+fw!<U*Y5FJhWM;a*J&bHooFZ@|VqHpyTnvrjvG>)w!6oTLbdrkEZ2;l`$g^
zyvl_|_+@`U9%vriDQVu50sNH(eAogGMm@1)1%O7<oSuOSTBt?~wd~J8trV0b%~2Vs
z^DWe67V64}fNBwxB+b4Vs4@#xX`!}AJ<+$)h+X&o98o3mGqYLZEB>VOHoEF|7GqWv
z+wIdr7-?&*6RoX?UZk1%&n)H;HeT^I9;LU@i|`$-ld!SEi5*iB1tpV)-eG({ccz66
zXZv;u8;2%b%<T9g>dDJl$6|pOu_&vRtg*%@5LcPR*|NZJl5Q=s#uG|mJy;Ql;FV}2
zf37BNzv9zl=EeEsd2tn0YBu82uTA2`5}#hS;#0)r$?)|@tB|35wmtWdS{I#y@H9c~
zSrxDi2^^Sdh+kE2L29}VP)SXC^fpd1AI`*_R8@r`<>Nr*dT-<TCUJsFZ6SmyEg4D*
zkwhz^MYg=}^Bjnd)k>2no|Af8ti9cdwa;<l0&9JN18aK1lw>mz#>N&0D&O=r9;_6y
zC~+#QDP=dVBXyNu%FDjO+t`zrb(K3TR9)o;Z)2XxK(UCTRVWLpki0(Q6B&ZNjUSqi
z-@-=4jh{F$jjwRy45<ZbweH~8qqgVaRD(nR&{A(hZaUGKG}RNBbZ!C7-9>HMkp{dh
z@R2#!m<Dy%;oLit`NA3}&&cNXx&)KTcE?*sS92!SlRZ>`4IYM+u^3V$x9v5NT|~3w
z+;$|X*Xt&I><}Z6R}kjMf$}%%BF%M?wf^mE>vGpi646q)Hu7OLVJQ8PH|xq>CEibB
zU8LO~UcWajt<*)@Y9sF<pCK_3Uo=quqJ$0~;Xw54_#^Msl|M@naKHS$vl-z?BLzTd
zZBb8l#_YVy*Xz4;#V&luYy-wPwQVvt*B750<i?2&eG*^@*lr?m$YlZz_^ZSyvj&Kr
z2bR+8w(v=eF#xlvS=UTZLk0U|uRvQW(MC=rs-|LBA~9L<N8(e3q`W4;%;si0zU6ew
zibJWOS)nl2a_emvA&mqxF@*3}<s67!fH72o2^l9yJDU{-_g0}=E6&Hbb|=0948Lr1
za#9w5M-n`|w@ufA^?m@m@c;Ol%YsWjFu$((-DrL{o8K?YZ@c;3&F{b^_8{t=p>%YG
zJ{$tJl&uE?R;TAjpl;KakxJyV5Pt2oRNEeZ&H&`zESwq*vbZnvcM5@H)ZS4})DMHO
zOY1<a@Mj2Ib8hhC{^4!ZHi>xiB^}z-GqkjNTWhYoz@6ab>}6nrFZ@eO^<Yi|YMyyY
z^Rz+n$H#4JdH=fXz(lL)v|djiU}TB2xqXZv<a|faismzZusObR%-)u2)|(G;$WW(z
zKzXK=cyj#q__IIQCNkPv+qi8GM+w!?^T-t2vhW?R8C&z%{lxIoukP_;<i5$=<IBkX
zdhUq|Kg$M$Cm!0isq9%(*V6lM(m{0birUJ(TH}>@W4Dr`qKtF?<dUgZr_0q62lj5&
z2pM_HmR~bAV9FeAO6#rCY_*h(sh;FnMZF&*AU%GJTQmFpF(j+d;4zu?WJ~k8V~5UT
zZZ8@-&qx?8p^fLo2Id*^OE4h9yGp!a$-6oU=&I6^#Le7y;*WEX4o3=RgN+&%8=7l7
z(23=rmVZc2(utooZdJ?h>m>3GHwVNtUyltM6Z9x%ZErq;YBfEysHOUhAGF2)Hiqy-
z^w6WY5#~9Sj~LB{x8YMb9{m`2l*3Ka1Tgiamg<U2RtubTQOBc}F#eIeC42ot`7fw^
z&MKk&)yUn~$6rd9f1AgaUr70<af0&0w)`#`dc#-3ZxaKJ%v(iIr35uZ4K_AseDe)%
zrg;#=GW~a6Y*wB;v%8E(Hg)@gi?!3+H*0ZI%W=5CE?%!o;bQIg4lXyFz-KMjtD>6a
zdR0`o+#hZ&sbk+Tht|ERbucQmhGq@*MP{Vvl)6@q`D3@5Km2ibx;I1LTD1*Sv=1DU
z32!I=e<$35huM5SR}FiMBC5`h4J|UHA^6~e5BPvz`{0@f=~8&Z<{sWiqr!vs0r6`e
zv=Jz{K-v}FXeSWA_CZ^Lf(wjVW8ebr2QJXE7x66gGbW->ps;BLAntPMZ_kU><W0!h
z4<7&L7|_Kp*ou!E1G4yKYxM3SCGRd$GKeA>JS~QOLGQcd|DVE>iT0O0em9I6c$4ZN
zbf^lSYV_-zyi=)%uqn1)GVKU9h-u6w-VvHR@##c9sf)y!^n|Smny_*|b1L^vJC;Kf
znZLMimz%Wtwl1=wF5)KcCI=N#qC`rXbq1Tz^k!siR^JIJ*v$Ce?F&9i!+f(IrD49B
zM`@U$?Z9}_TbgUe^P;7C=y=|=R1X|4vWIN+YF3eX1vc~wbZ8iXj%4UhDHmo9f2|v2
z2uktu>hjMB6P86gCyHD<+Z*5M8~9*+&G><T5z_u5+XOU(;Mu_*G4h%pH*{2!hcs!z
zbZfmDh?odaWBjzWH#-T~K1XxY>;K@R%tzyc4}KuB&@U}!ZX(4{ntn-Po+8DNn|?`Q
z&LYLooPJ4R{vyQ?o_<MTE+fTIpMFVUUL(bjpniP=A0>e%(V7Rr4`hg&w~;Ry*;@4g
zQ$bw3FEs}^W;RICq^&>v7s3<C8UCLlKg0hG5sDPy|BC_PphoC1P`Q2TYS_Iv#1wBV
zW~iZ)-CtvhV4ZJD5fb6~ihq)dcWKLm-&0b}*wH5U-pbupL&@-aQ*`eX{qGe08#PS<
zrViJPYg2mgFyirz9V+_AXHwBFTcl57-$`St7bmC!OnNVx$be{&EyoqBg_EIw6{{Kz
zlj1VYFe%^VhDrJQFK;>i<i!D4+wmtZu7S1v?wG~1U~RwawRqrS%fGf0ul0#B9r0(z
zY>T&yA@0P`(yyNPN0dWF-`buVZyUpeJZqc&gFbIeQzwgp?!i-4$qF-P8h;_?uMO2p
z?1+XqH)_T7Ai&+H9hrEb6Wysq@BG_LW52@sO#g~w&4Tk)<A$_=2wMKT^|YaK$t<3D
zC*HTA`tB49oCWhUAZZ?G-x(%23-)K;nFG3O;C~aV@79G}1}-4WyDcA7Cq#q`Gvz*~
z)6rpw-C>B+gl{cJG6YzaBc34o^O=tKnU42C4@1Yb#s`jjG2VaN8}ZU{>%Ngbb}4>V
znm<Z=)OEZ=A1lV~P1A^Yq*%0#o@G|eOnsHa_Cc}Y4etV%fn0VRLQKux@J1dvl5O7!
zsZ$>wX9V7)MMu|33H?->p{(K1BJDM;IxiN?>kWL6v-)=WW^sDonz2FYW{Q-uf_uX+
zbgCaP@C#B5ztA~;q`)soG5msirNA#pG5kVj_>ls?AjR+to#00b{DKt2FSu82Nk~`L
zPFt2=B#xt^nL(>D!wA7%@D1~OV62(nYsQ-Sedt&-zYiQ+xZF(rW>ugG%}o7fRlt;%
zp8Efo=`zE28}E{j<-Iy8iTE5~{VS@HU)`HlmE1MUs7fB<!lG*s7Stu*Lm!d0-d~pi
zyw(ENTfl?w1aPnbQ`UPY131D0USa`dlU-5?pwX!upMff|Py;N~_XKr@prljT{hM^{
zcHU@ecd&&Tq(X-XN;;L-Gf=Nus1Lc&hTl90)N?>#%09ONoywqknSX1t&$k|BPNQHx
zkDd)jomgBK*}%e^dHV+WhQ7i5_si^=D;Rvu&T4{JkC(>a6;>q|m+8dkMs1&VqSxbr
zIsk#<)a&zd&K)vztGCf1J)k2$q$!D={1h+FpLEreNRP<ec2};k^L|M{VKf#;0xsEG
zk7xr-rq$9`v;>r(gD)u;L0mcnjiEqv4rQLiGWux(44yw^C_0T-P5KDAYQIQfe?Tg3
z3v}*0ZIfNuly`XIAwX*)q@;_yiv-NwQe3PwMJ^zLA~A7~a6+?dE5ES$JJ_+d=Sag4
z92=PAaH4fp0mj!nV|;b);`27Pves0t#piWJO?lg#)5&Cq)6E>pqKXB&-D+gNNo#rW
zMj-M+SDSv~k-419@)gI1Pk)`JhK*nnOz+`BKVSVb`nmtHrk^Lyr=RQQ=vbHhe`MgF
zvGC7Z_#(mgv+$=0{um2?YX<%n3;!DnUoH5PEqtEfH_b*jm(345{Ue<4#<H5|#L=~E
zj9&QjZs0TWia)Z=AHf;e8qHB7U0V>oVN6aizdG7Ic`!e6aebQZen=<{d1-lB=RjS_
zNjb&UYy?MFhT(cd+1=X-HF%>A?zxowQwQx#=N89{76fkJ?0@*97yGI!U!DCVMr`9!
z9C^K_{3BE)W(oI3_#*<5H6@EO@*S2-k>E`>OvZtW{J9AND9&c>+4DH|Y-{`QP+Gxl
zvQX=9VuAWQUI>=eRlYKZcvv@zNZEh=Q&An;p7#Xoz6acawJ#vaDx;&@v|h;v)9Y77
z6T^wQ^7Y`)$zm>2%PX_1>SELK>MCEK`y&%;?DaWsC+0zZX1oS))g|lA6s00|)Dh;O
zo7wQHB0Tr;{j8>M2CBKwlPX}&egCe;1-1s>TeUy!-vJjUbj+Y5*YW73Sj69Ew4;li
zRsufC*whb&KTpz{IyU#yFK`iUItRudol_!3RxxkiAmQRN2lw#Cmm!2e^umhhVNPt=
z`#~stNY6LCjV43|%?EG=ZtnjRlFLZp&$85ahLvn_DqCSCTfv-F<8cmR;|FG353l0e
zU5FlN3`wB24kfQLDoa@hIHw6k>*0iIx|&j*As*a6{$$dp8%GtY^Mzha8v08q2U9B7
zLhN1Jw=>lT4^GaN;#k1|r<0d_V-!raa&zb>+@dA%i;G<9SpQJJjn7MOzDu<@NChs{
zrUSm2xob)bgB<Z-c-59<cgv!-P*K149Nd4J7$NI7QwO)<MmCpViz+QlJSJObG=I%6
z(>$UL48bGb;sTHGqh?>Sr+?=(_1g^KeHQRB3wRoH$CB*;nrZ6B4Ak`&>Sha7OL*2L
z-jh?))I}Mn3oX>;7HWu~e1g(6b!rCcWD9ksg_?g0P`fFSY3czui+k(!u$PQ%uJN`~
z+$L-Nr2|5vJubeR7Dh%^AqVF7<sL!zoDx0kxyw9smpR0YA4eif4jwQ$OLHKvm5ccy
z(@YtU<(d+{_W45CT2*A1`^0VF8*5U$y<8f+Ry;eDu~vDgH4#_PJn|ymUgRltkN72A
zIMV8_`?X=_RUE>3z@>@bF51;`>AvmR_c`$}1!c9j3x6%PH7M%tw4UQr@!qEnXn=#>
zOQUc`V|_(`p9=~GY{ICl5#~#Nrz`$yA|W)UaXe+N#Tn=+PpQ7I<$Lc_O_AoNgva~T
z7XPHe#~v>zThlUe@|T-AXs0xH`|DNPRuzWM4n)?Y@bgDr&G1k1C@jT>b!c_-M+b!8
zJFEJ@7udkXntkerfyzChMeM2X(tmtQIzD3JFK7}9>JYFOL&5r+>u1b1C{|4%vd3xO
zljnr@<e46Gyu;&8taq8S#6ouTaVnp6yyv_{Bt6G__FFpV^cwWpQV8d`M?gmYNW3-@
z{4>E_;A6qkf;oo}M1mukI4H`!WESsZSFt$>Oz5bK0>XB<%`-K3O1_MoD)iLFX6Z<(
zxjvBk<uCHsXkpRS_!3LYWKS^MrGUBjB_F0~!J}GhRf!Ci6HoC#*8u-D4gQ;;BOGD1
z%W2+4n|Ec}yfP3QSlroWf8`4S?>SotHWToky+sv2l(02Mhwq+1?fsFKqylZ$#roe<
z-RikV)u8=m|D4+Yu@m8Z8QKr?&{h*+t-y(_b63p}!GfMEXz<kq=fylvIFVYP23r^Z
zY62<@JMfCU9QO|)K8Ow5b2;~SbDzvh(gZk^bZ>Al&{*J3F}bdD$532){>!50c^jn@
z6_r*vi!Z*&{N=+#z2P>^?wqBSb`e7YZ&;8gCoYd18{lg4$NJbk)fB+sJ2%$51Qavo
z2q+%e;C@%tV~1afewdkprDe$z3_tK^dS`e6?t8poM_gUyc5g!^1(@M{ZnSqhy9@(V
zY_PXc1sYNMyjbB}NU#l`iM9el7q$6_2N*q}uKYu1F{WnyLBDZt!(K8Xgp$9xqZ@jA
z+n0Xt@Kz378=UD&Z>lc?L~pTuNdP1WpZ+vWbT6E)CdAKet=Xd{(~q0x9Af&2)yyaE
zosjtXu~~aq1+^9si$)~;vAfr_yI&x3V*w_vFF~&6W*VsoF@sYrZ9JT+CUlN1^0O1|
z=8x2p%ZP%4@Qs`X9lMnMQU?Q(X*4vL-RS52(eL`pzuc$Md&sUXveDnXyBkb`s4aQZ
z^3BNQ$fXizsj$PJ7r$7h`Wck@Vjy?Dv)Cqgc)>xtD3zWn2!=p{X|#L3X-{05e&W7#
zJ2z7Ep+Q71>Ex6sR7A|U<>CC;<sZ_LmIAHXpzH1Kr$G&hQ|cl=7Ct~$WCZvC6`7_s
z_M!@jqZ!dsG-gVedvEfu12$zN-+S9!Nt6ru$;<;j%>yOsolMdMB~4M1G~xB;iX|uz
zE#*}2aqyNHk^Ta)S)~(_MVjAWD)G8lX$w1kL)T|=sZuVZ_!!59p{03KPgq{Sy?YkT
z2}B_3W)6~}fr^KP&Cbye;@r9vdy^svG_;%i(W(^)CloMaYqj@Ikcl^=gBxFi`{J}+
z9f=!2z;aDy?s^_&gJRqlnCWanyxoU&k4ddcKMs8mAT(3NDMP5%;LTdA!wo*005^zU
zXE?+Ax}R(BU%rR?X}XWBOdg&*K1mC3*G^R<-*r^3P~!fpAU7wOZ&?WmKKJAUb0Ok&
zZZAf<8NXTlBfOD@zhe1E^Z-LT_@EcazTF`4tquZduZ|uPyI@avXP!5_jx?6EjNKc3
zoX5nIzq~~#$1qaV-JhCLv+26UyMUDooZw@(WvVC%<t>KWa|(#^Ru}n+PeTNp(7RI#
z{N>w%2b(EotGgYtt;_wgj!7hWM!Lvg%~#9X=lg@0z$07T2SHU`WJZA#9$ruk19q~}
ztX@V}I+}vJyh(f?AzJLND!e0)g)dA9(FxO&{6wP%qQI|mqP?3r4uL@vEFGM*{%O$P
zD!AH8$U}@`aOgJEs6ey?3p_<M8Mhg;>gl9*5Fxg=q?7GE?_J>UWP5Z$a0Iw;Z#A7x
zxI=g}CX{yZR64&k5HU<_`4rJJazJ(X%)DuR{gI7Cp9>G6uUhGyL_Ivk-~2HQg|9Sw
zrXZ)j?$@HBvvd{>Zy5fw*WbJsrn@)8bXO6Ct0J4}KF<W2ApRE&C!w?C8?=oCezS=J
zbKHMXAtq6wVDNbn&lfVj+IjO+&6{3c6U+JwA<=2vBp&mbd$Vc8<ytlbON&RwhHf<V
zRWs)QgG@ObBY22Ea;y_Z5GY7Itfcaf_o;sAx2?WP$W*@%RneRg8TX~;{m%Jfr+YbL
z?EF~pOBlnd&iUenYxTiZsk*}=H$g(Mw9fUhhG~TZ(}LKsgSoRLY=l?BTde&)`csHx
zrQt4Lz+F&#GHl^4#9KzBcdw>^#9iF6D#^yG_qIQZo0<R!o6`05-WCuuzK9>V$vv!N
z-C<XYH4OXtIPM47`-tvO<ldcf6)&UtFWsCLd|sSv1fPE*e6VJ_>*q)pC&M_AE@~k}
zw}Ow71<3=Fp7e4hRoP&^+j5OCvCe&jao$DVfKi@aeM$Aj7dnKQjLyJ$S3>*VkT-dp
z(FcTA6|094aL*J|{mi|AoRTiCmD7qdOUoaV7KolMR-kVW(_+NpkM1zvdo$mkVZH}C
z^L3f6+ojdxt1n9&&BNEp2RMdoXScumldu{Sudbg<P@YEQBY5EqdOfv_s*9d_Uaa?Y
ze=PW$9QIQ$!`0<wC%knIBUW6r#BD{0W)u~9Z+RE;5>vr(;VP}UE=QJ)U0o=Z4Qw%y
z>>bHPYOVIfcg*n@uuUZpd5`rg?6m<1c+RDDvCA7}GOG5;{9Nmqetxx>o*05Z>cN;O
z4w`c3$F01|tZ+VcpA;nvP>U0;zUoMAR{o*iCr+RM)-Tm$65nx+zbEeC)gFrX-Xwc8
zi$Fva^qo%Z+%KY2cSoker>B15L{0<l?2S%j&5REdRRGdV!}v<P4Ofw)KUP}^8j_{y
z{^`=b*l>2BuM1Re_BNcQ_C;3GvUAcc8yy{P+Ql-$HY@UWp!}1dR}m<~t*k;nb^pb1
z$h0QiJSUTF=+B9dL2NetMb9auc_Y<4Dbe2Y-t<2j9+9_G_OJh-eXMxsHW|38OxQ|q
z!+Z+y#~PcsXnr6@b<fy9WxM3zKiR*Q4WDmY?Opm_x2YqmqcH)tG~eiJY%w=8Pid4?
zJ-tNh$JmrFqBD0#W`f_D;CC<;tpdNBXMC6h4XYxD#M=)W;qlLlKgTD5i0g!R7J9?l
z<yXDx9AB<KJXgm0kDyV1^n0LlgvY2sid(A7E_r#tJ36!|8y_5ah7w7dk7EHG2TOGy
zVnlEPk914Qp{VBAf_UFCYtdw}nM&NLkhS*IRd`x@8|F|BuD(8RI7|(6hUN+#Dx8ZJ
z=0!)Y0VfBTD9?}8+{C%U>o}dMrTGRQZNFjs3Vw=E{}FnssW4FaV(7yF8&ym&nsQ~z
zN#mMg`Ep#B`tI~qJwkPz#U>M~SM^9s^VIP;#1>U8=0#OY=0%Gh*Jd8q>hYhM$A9Wk
zHa7dMtU6UPq}fNAYivEqm?kz&V2RissyF*7v-kzUPg7k23@&;||3Nr$w16!HRUUO-
zJ2;ts0$rCHKNo)yowGYKhyI;I|DFlR2>SQ+86QRtqHE9gaIke@)G5^WC_zrC@=X7A
zc6qU>Uqo-%9l3$;Z{Yhg0MYo{%J`cmlH=G(CpM>GX(8m{Y~SErgpRM~0Q?<@>h7IC
zj=D8qXBf>DwHVs3_V}J_^M5=RHP4=Fd(7dgNck^g1{`~gdkKNVw(cpL^^qP=zMDtd
zBJ$~+&vaN|)#vedU)?6is$KEjBif7&m!u9J)=r!{yZjE%Fs-+Z2My1C?<gE(o`(-A
zZpw`h1><}rzz00`#iZFsuF5ruTS**fzUGLO=Q$@j&fb>x&_d3N@HXDcL?~J9F!}ed
zK;h*OCNLL&d$uHZUCm|mS6$Y-gZ+``BCo;ZnTy={eD9B};w18F@5qgH(VrC5RDM_!
zS?>*4#_!zaZES%u;r6`H;-o@sscd#15h~2T5oZ$5<>+@B8JC~AkAKdRt3e;2z;8eo
z$eG*&FF!Aa)kc%VtJ#ypr|#%2nj&^HMSNn$wH63W5$Jf4x-~^q|APT$<_JYh$mrM7
z{B&oLU%F;E8XTDULi^{b4wKTjedY_xRUsePiKi45fYEnk%@vXV(Of~KGdoojXsYPU
zI`mMYn7q^cJlc(clA0|HJN;>zo!a^M-}BSmx!#5!l4E+R(DAQd)4tdUCKJ@cuD79|
zzKE=%@XUNs!}?qkMuA__i>H*O|Gx9ZU#>Lsg=tc{iOC1E%hUX@)0dhb0+B7=n<USX
z)qgNQi~^EAWMxO6X=)&LY?w~Vni_oHG&PL=`l;c>G`_#p{4gzG<^~BQ$S(4dMCOvR
z%In^811qWW%}#DOHE0HB2}hcb_T_KEC-&uS-+{Mf^p{TT|6Ts3rDOHz3rE5B^4l-S
z%b6BJ1!8ntyI|Bcr5^VT4w7)6M4dxFvRfGO0}}%B_a#kB#dPeBt<3mwD%*l*qJfF;
z9OF4$ex~6Sh2-5l^&_1t`<_2?ehN{O$T3ikx;Bj{vsaoG<Vw^BDX`F6$&W<6S@he5
z-yrm#l7F^peUO+=cuqmLX|mB}MO+{^tbFo@pe+(#_qtc&qTzmkQl>6?Ed!AV><P{$
zALeaboG*r8I0aO1q_4k{<magIWssu&`kWst{WGlw?qhe<BrRhpA*I!u@FV5<6R9)y
z`YShj7lim){It~H`~gP=eC!@;zChkgeQ^q3IC%Kvu1|UfY#PvnGU{Qc@+~ZIqAl3Y
z-5QqB@A=U)FX0xG@9^ifKA3s}gztA^qkmE8RKDU3$9YFeR8y71i9BKLqfsdFH;i5J
zMb0zuhu`ZKJcZ@M_qBZJIWku0Cp<D%IureO-VNhSoNvz^3eOW>(T$U#LmOlr^v>@|
z$>^@XkC^q3!Ta<FL@|Pez_<}i|3#iV^vK<~5pwOQxx=GpZpV#~(?-o51$yRo+yUm1
zaFZ)M__$V}a|=D0ctnketkDE|;`i+YI*couq}ml9clOv`e;j4}#K#`<LvGCcZ|2Jl
zX0H516P)pTVwyC&XLpZyb0QPo9l=xb%pA%?-#ccVl$V+w``uA<O1=(LH9%LLl;84x
zZShCBEh{f7c5^w&?nBHrrBBPH+8igg=T)rjJ3}#J3pA&>pyw)2cpJWp4#sKTfo6P1
z-ct1v)_|92X*ZT%@&wwJ_@imr>ke-CatLa|;d^&kj|up=b6tyJXx!5telLHzl;5kK
zuHm=l>C1Wd`NXZEvbkHid)3^%!rdNow>v(otZCv@-p1q4jeCa!_+E{-P~~UFxm0`H
zGTEF)FT?0mbosI>ey>`_imMf)-`-HvsoWa;uDrC&Mov#tmQ9Ol2mVI4jcZm9;Sdp<
zLjc#ZW(TDXjbCpji&`xx)iw{O23-nEQ^QOb73V)&mzV8t8Q(oRpt_7Ubs0TQbs2s6
zvI*44q^Ig=(y3{tj`&h+4fZubk*`;dX@0HxJ<0xo(QVX^JF^98EBoVED3^cYZTJ)T
z7jJSRpBnmr4;=u7aV+nu(C1<oGsw+*)1&H;@QU4l(U4i*`I=qXYLO>@e+EtX<MmMc
zGULA0w(oi8EJ)-gG_J8km&+Tgbh)Z=LVV+Rg077x=-POKu8mK~E6BnYS=b^ATLdht
zbUX`NLyzjb;D!e%-O7Ws?~wS`@oL>6>3Sy<n)T8xY~z07mZu7Cq=$SxY68&E*6~eQ
zjhTAhH#9_b;#fs{QT=&R{kdf7^bU0j9+s|8Q78V0b^&*Bg8@-`+U{ky<!MVDvqSQT
zefWdfYS0!uihkg*Xw3xEHwG%-nu@7sP12(b{gF!~`UepqXSj7zwf57RBljjVfuhAa
z7@1q|ntv!}$+=6}{GAtx2lXO0>|jhV?jx7(>KO12PkY!S>)y)~s~YLTj*u>F4v@(n
zq$DYp<dk{oCnU?lHWGHcPyMszn|T}Mg`;6nDk(C5^|??|c#KCg1dCpThg}0Y+=D-3
zcDV^HjT31ot>**3u{kNuz(Ss+qCl44ATlNP;nSKzB|~=7cM)4<q<R3Mo?=s)&mBR<
z41x<75zEkk8{rW&S?O`r)Hcz)(SnYWYJLK_<~VX*xt@Y#)~mqkQT>c9&Bhu(l<Hf(
zsTpO5r5~$0^`?ku$n<uvjWO8{cqdt55R{qksbDQ5iKxN$KpNI*=$gYVh3um8hU?&r
zK+qy53eB-opv1jlLb@%~8O@2|HSMFZ704j_s1teDU(OMNoSRA;ybUY6s{x$9gX$IS
zX<irYDR#9^?P=QNs?!rvCf6%X1E*vRua+KKEh>!(VTy56`MZ>dWe&U1ivkgP?o;>G
z@#$KMcx2det05-)<$1CGfA_~c_hAwBM_Lnm5dN_4b<Jt+Q^|P7N+RTSpcFPXw~pu4
zS*iV$ewp^}p_8<AI2u6z8D*m$bF&7_&8U=CL23}(^w){IH}xefM@AoNJJwmAQimJ_
z30V`aR0g68mNA2qB|lz-bYfIfX){{95YjPcyA@LCxltM6V!f+$u#vx8@6tND;JH9Q
ztfTuVfGZAQDHGY_m(hE1S_^|anTyIqtCOjx5{aeEiR^%POdqI_p1JO{?`iRK1|rcP
z3_Zq%R=K@@ppg+zo|H1arN5c+9PWPhv#CS`eU2|fs|IWu8N2H09NKVpq~^~Ov9Pci
zlX=QyOt+*=#--u`>GAx_u~H^uKbHWBF{=^mVrG(_39|v}WZC8hcVaK`0X?V{S3zC*
zmqGSC4Mo_&Q2UAd{5TlK5wd8c5#PAL^fn#3?W@SBQ!zTENJCxK>FM4m<!C?43Q~*R
zjN<gOF{*S*PrpGAD11jrO`Gq&qQ`kHVmVn7gxWDi!(oC1*-<)Mq}d(nZF~qWM7E}d
z+B9F`ZCHpfk4?iGq$3^f@04LAz4qMcK7KK)Rf_5pU~eBAA0QhB$|x-AJ7=;50y;`0
zw71D0`7BXQUQ8<8%<PEI*!s<EsE<p<k^gH4cMk2k8rx&h2)jD#wLEom%(m3ciFMrA
z`FPK(3A7Gd<9p{qdi-AXWeLB*FRMsRnrGGkt9(4Q$U*5vcHIS!?37=fNUk=1U|vL?
z2yMw$4^t&s8SxU>tLb8^E8e@eNQf%%-g1M+SoDWbJBr#^9om3EX;t+zq(U5saIDGk
zwKsz+S)Em0^9GE;7_@W(7*S**ad!zM%ytM#&JpsH*#U;f>aW$uEKJ5`Kh9Mj=A0i5
zerVno7}obV%%kvE4Cl7#<qVb&DVd^orL%T_WQNAN03I+&e|kRfN2Q%EoUhLiRHmQE
z5Q76}_^<F0t(JRq;e2V-1UKW8KxBaAQZ%a-#y^HWMVOF`Z^$1F)@$62(+@<?$DVst
z#lolbxpzT#BprXW)|7mcEtxZ>Xx8+T7^Rv3_E$cUWYgt4kt0*%?1wq#R0d%^fWoX{
z`%Wk+T1Ue_OCEzfOicsP5WO-SvmaYq(Ov`flAt2kC@?ZoZWjJ-R&}PAqhMyam)2c_
z^H1I58T1Bsf9q1eX!i{M(5Fnljh`;>_lv*86=1-&Al6t<KT)gT71p{uuX(0kMqbzk
z4T__f@fNz6-PkjoXv{!^`vO#-KF>n^77~SrO0xoi@r*KC4GI@o(<Pmiz?zK==;VuM
zBztydnRs@GxRSWJEF;T!a$-(FvP--OCx&J_as;_Se9b<Cy*mUMoq{78s(&i6$W-Dn
zRRSuP?zA<t-f$w{bD}eBogRaPEdf!dBd{A&TicAR9Hz%mMQ8W_e!ovl7LH=WW>vEa
zEB-#5CpK(~?hEbxWr#j*4)>C`3z9`zlAUgTlKHyy*Yd23v(Cy3$%GVN%6-S@{#7aP
zIv8e(h2BSs)^Q6iku_dDkz-9`;f=odj=5678P(*+lbOk9mT}NU#OmOsde@Q+QCrfr
zc5fu!=j)XA*RzlImu-SZ``ZMFw7;@8WNCjV54XZ=J=ad{?<X`rOZ)q49$C`=r}|#_
zdl!ALdETGC*Sy$Q->by0>U$eLGxgrb!Y~9Cfw3@ntc5|ty|ceXO8?H@pgAkc-k^Xh
z8GaGRD)4{P-k?eOP)!f-{}@`yR@C{7{gPL5dCb(ebNvsQK_z*Jz6_-N=g!Q_2_4{#
zZWYxip5CjLw#<Uv<2I5zO&u+LS)34q3}F`Ij{uCD))W5d5zI^;?=9_M>K|wF)3T!N
z%WY2cr^uhH{G3Ip9pRTWC68W&Y_|q|&3q4tFJd`}T#Pq?oC4^kxg(-I=c6ogBKUH=
zXXe-1%A(*EwUrlmg0=h>1V{2)7#!(`H#i;7`@<_Y<FlAM_f0>ZeE#sRZoy0W%?pm_
zH#c-YUx$9p?)T32^Pz`W@)>;aT!7pbZpLT@I;5S}%n_I`+f1`}zAWVNtz*|D&f!Lt
zc~d#)=Ec;Fw^7HFC7aOjm+}#<JY0<gap$57T2aVpLn#YjcyG60AFcO;mHg(0-iw}5
zhZIveIP^yHoYeS^p5c#dF{wjKgQrzjR)&rsI6<;3J%8$x_ec&VR6MWr`CVD<L0=i6
zr_N2)|GISjui{2pUbBqzx=&m#a&h-jSN>2Y?;k=i8T@q83y!&*WD1UUK{Bjp5fW38
zv{r-;xj{<NH6oyQrAvVcIv_woAHotc7t27mnj*EP!L}^{3bD>dw?loB^bkU^Wgs+>
zT$Ynlf0jwAU{&hGSc4{yOzYPp%^lTR^Vd~=<ZX~GWaOk-PcrYrNm=+g2e!+zJ6OiO
zbr;8D$xJy>&zsO}=%nQAF5|yOaND<cg$j;!q6IZWt}h7QAy*#UZ5?OQb<9hf)A>_Z
z&b~A5=x+xPPnKuqA9ExgCWyB~y{ap>hnf;Pk}V6L<^)yuqdHI_pPQVV;Zd|c5vQQF
zN*1}CYO97w%K_+)-2+e0%}IF3MpLO%r`8av+>S<md$WO7jIS<QSgNg+Oyc88eMq(=
ztUnqCi?{nWF<kF!itoC;&7Aqu<J~i)W*NJtWy6n_v>Bc0Up*3BCG*w%)g#XW;)^u=
zl6-MTy!5<D3R2M)IFOKpVD~8QwY!Ib`Z}WhRw1@Hxklw#5BkqD>A@!g(ZSV2M&jfs
zQe0FQJxh!9d;OE<g3tDJj^LAiP>!g{j!w)H(#-f*a<xgeyWChLKa=<lud7!<KU(x@
zzI-QG<Zh7exu0D0QYv^zqFAy=?`;0(MEkf-07w)sjqikzQ6(FKHlM@Fc)&LIr{}`y
zAW1CCQi~K-t+ya7nDX62^kukZOqK+bah@x1M{N-4uXEoV`8EExW+IXP{!4j{3^2;_
zn$pSa=)hgzJan0?XU3II2_Dxu{#f%kLq-*Nf4(BvLl$IULdj&_<GBk*mwMwnK5Jqf
zfa_vLNn#$eJjcSbW$qu!pnO`BrV-qAYO5A!I=-}?D>2qH<Gs~W%;-bO^r=p+z?L6?
z7Y~_DNX&N43@oCZNrB?seohfxEGugC7k0bIQ)Rq%%3p`)ce`@i^Upo^T;-P8XxIlv
zoan%<sMmsl`ZIHa_sj3F!u`cst>}nd<_zHo)L?ad$F=zN$9K-<G*<N>ZlefI5CtGe
zFInSfz&7#<JyAca)QhW)23JC6HDx;r^8tCEdm4Uu^lF8bJ1?a_mFXD#B{<xaz4XDO
z1CiNu)NDoy9%Og~!8y1|97IN1pZWsjacA*np7@(f5IJwd62tR0AtB@QViW`!cUvU>
ze^}6t?mAhsa_bBl@KVMHx4Se4!_dVzxK~RD6y)`+(wQ82D9vp1#~$|$zj1@(97R=p
zX7czhA5X<1vnK2NP!rBw{^k!o*o*s*j1296USw8(Cge?XdSa6aC|kt@aNmPik?Yhx
z_=OHuR%gp`xz(+uM5*p<%jG(&nh|f*BR66AFj9aelEfF1l|dB3G+T<)ePZH1_jWOQ
z>+Q%$k-fd4yCIQ_!0jrhoSk;7Haqk7c9606CT$jkHsCC)BHP0sd#e}wNC|jrTLa}(
zc)ReR17znlH<2zu{51UDjTk=fd%C&4xL3n>VOo2yy<2`nTz^#0&y`L54A44s&e&b%
z@+0gCJYCsgF4yHw-1=cz)ms^7fzd~i^6pnsUi(VQ^ZTSwHSK6otGtaNsvyaAk)w^6
z1E;svJkIc{hdHn=sca6pT9S>D4hN1j*`J&8Ip`$ImUBdw)^`ua@}zl`21KJGQ2BZ2
zRB<&}<U03mz59yhBt0L7SK?LjIg(>^i1mm2AHb7tnlXl!CAEp5Wgzm&gj(L5EDK+{
z;O`}`$bKwKUYz}SWb&NUqsCImL{#wU4hqnOz7}saRnTRKxu1{7YS6l5kL--ry07W2
z`RLcFkN(C-$jghNqU`-aPV_d*SLRv160v-x8H;AOD_2{-GN;g)^a46XSc|7^<$$KI
zU9S`qDcgx2N)T+OpBE{}s3&rrA=NzJ&$C(Mu5-6PE!H-oz~Aoxr@6hGe_6IdP})En
zo&?+KGPomOeUS8J@E_-l!smlm_}R~vi=u)7xEGn+eSkda6O^*7g4&C^8U2OB{s6n)
zDaRL2D<HCCMb;7gyZbo3<DjY2(?jm>)mOhut&O@lt-~#37h8csaM%1O?TmT$S;iT2
z<WGz<X2&%+W8PFCpDTAU6yZKF+?oz7ndG~}DL}GiKR7`af9<;7GwS9lVUazQOj&b`
zO6M(P04I7W#Hr-~hUR8;+J1P5?YTv!%la)c9oWcW^TewZgB^@aDxQHp3FkZ*aYl{(
zMva=zwh5LyL~^sH>ZgGfi>h#zr`T&BC)WS&>SxS6_6|ad^ub0Gbv(<=T!x<3Ez;Eo
zh@lZ)Bh1n%m5C63f$X11Z+e2E8>_gtVGl}Re?(jzYj#(zUPxWA5bl`K4b6y-#rB@P
z-RMjQmIe)3rS}0HP2EwN;{l?rKt_1*9B)+TkHV6T*;_8n;2KT$CvJeg$mmO#(<jJJ
zk3z?#{#gG71_uTKl$~jH;&|GRGD(D}C18bj{^4S@u@MrYd<Qs7(ctRuEPpJ}3kJ0&
z*Ew!Ybu6Thuw-@*p3phJvtY#6;gq)c*2|jM&xMCq6Igp8^f4+mqVOyOY3Idy{)NR?
zgGf^<GwIVtNu22XC^kjUYdP47Oh#-0fV~Rnayo%VFhHlsRR&MB%$Fm!0EB_W$B&dZ
z-VraDdc>ymRHB?NQ)LKpCe|XKQ=FB!m(P1Ev3FJ?euPadorrg?g@EyLka-a>xyrI(
zQ+6o@gs->lDJm=#+C2BN-sP?PHUk0R!qWKGF>Sml^4yz>|NhE=cgUfzi+(Eo>G)=M
z9z4x_*yW>tB{}kFn~ZWVNRd{ewPmr>ytA9LOx#O<>`^6}FOy~<y4crKBf@3oEpaBj
z3;o0Bj>Gre+i)m-tkSysV<VBERyoak@o8Vx9@*afSzZSOx6?RK?z7*J3LV&qoO52X
zqvA#N@7~_FfBmt6_ht9*fqZj8cK^QQ#IDco-<<&9Z5+EGyMNyRNd4>2?%$mNsek?H
z{(Zyt@Ac|m;{}%P-<^S2wffhe>ED2N1fdHqQ2+Wf{p<IRSW6S=-z~O(&B#UG2+jm9
zud{=}xz)$3!z9pn{+{aJoeVNyQ~f)S{{6c#8GRmlg^yAloLT>-Kc#<v%g8n#n$ev8
z=wFTOTSCA>-xe&b^4@x#Sq4er9`(_WI+nKkfYk{dRD*9D-kJlenD`#w4b~T+;t11{
zYVYmM{>lz6oCwxj<-MJ#qr14Y4p?2=z`pEizO5?R>uns&P2|(?M}0&W<y*8W`sb-u
z7xx8M>Ed3uo+QVNJH@V+PJEb~EPme7JWHW)_zf;E(Ph^1s+MYkwk)5ZKu%BdGy3Tg
ze#bps#qZ@$Patf|^DPsve(-s%pQ@4*r{chwG|Q*c@U;x|C2-m7BRK^(?UG3fQC?oq
zCog1y9od4??uAY#&0yan<G&xby=BAI4{kSLkCR)l?*O@NY|924(9+y{+~$@QR})l8
zWmq`>SK$0||3C@&oHeZw)zZiKH>zcp7&+z7!?>O8_}Zg*{jghbD6<U?MzfZBuohg6
zjhYC8HFkt~Zc{f$<6FnIwQM->p*H4}qCy1~M8$&NP+RD21`mEiH<2y<DL19ZQOFp6
zV)(&6-$PU9%8%GrbDM{p15pCqrn1&6*S->44d5%6{mKODsku$#DiAHqesz!;$l(>Z
z7PW<LmL1YKwQiEixxCY<l{Mp$5^cN4sH{a3qlYoY?whKW;h*}lG8{2P%d=s(>i!Jw
z)B82L=%<U61<So1Rje4&^MeyzU}yb>cGh2LW_>Nzn#Jx|qAU=LV{^EjPr9rx@UMLV
zEs}IZ411}7tkUt{WE%gZFX|7#d%Bx0Fp`sXgq!dJ1n|7KK@O|p6BTuAdz!RKi>PJP
zfj<90b3qsVkI`jzw6TGD+B82Re~P6{-11SujfAIbF_q&WQXZd+?}RHvX}u}D#BWS7
zHD&mJjrWwfQ_TyUM$~&a$G2>_a!H%{+;pcL0~9_$$!l;hz@cDcXr8jHi+rJFLPb&(
zk-u^b1KiGAoMZVAzV2ob#@J=5cZaEv_Od`}3ECTr1mDCh!QxAL941JW(9@lKreRz?
z+RI4VIu0=E!`Y4Fn@~(rKldC1(0{FQ!LzvYyrQuq1Mja!iKH@)jFELow72pzJWkvw
z?%u$A<21<@8q8Udt&aS}_j+bsj-PX<ukiBp8oFiT)7x78&2B!SKOC68(BHWpshsQv
z$5&4NB6wM4S@4R=$%Vmj{tm55D$t#-^`mu|`%h-iJmyTin3`M(TP*$m%5o()$Z?W8
z^KfS1aOzM`zCRI&LaxW8u54RvOQK0=a+8dorZ18xWSi6*RR!(?khzWV{ML7bIc;s=
z4V2ij;<`T{0J?D$0wG>Ef;y)YA>#2Z*-o%s+tY=<$KXYSvDH0?93)ixgvgKF#m~wd
zi4`}qfPP1_MdLdnT6}^})l+(J{nM5Dw%HGbSi4y<ZwYL9ciPnk^ObSJNuDJAo8c|T
z(4L}Pzjl*1<H!O7)SirOc1vbmw125=&96xdQQ$A{=v*0_Z~()Ve<~b+8oMXGvlz6O
zWfT*Aun?qn7C)z$0O_5@ec0%?UhGLb<&B$!yJN5jrb)}=rd0KWUC)Ck0<}acklupz
zjJEvMp_?clD!m2Ulkro*QC>;EAPa@;jTJt~hQIrbP0?V44--c)>dAV*>Lix!Hc6TI
z*097!t+2*@4yrXtnzT_U{fom-++Tht7kPyk?03mEKiYqu`v)mx5!HPu<zYwoL8INu
z?3wLJ4J0;~Nj%GU^Hm|4>Ey}Z{=HothAgu6g0Rz(C&EsXgZMBu>~!6)<36J=EOQqj
zk|w<f%E^OFszF93*WZz=*zL^_YUg<ALSOLD*Xij%=|Fx_pry=#Z=(2Ngc?Tya?seV
zT`iJh#9oG@7|TqYZg*j?VGDwhvC!*ye#1S4)xRpsA13li%fw%ATkmh)$mk_Q^gd4P
z%3^6_di<``uqIeh_IPFp)oyl2fF9T`YNVFtnWr4UwqA%zuXelr_I))!^{T-2?fdGw
z)T^HEmmnj>_u4OMPGx#Nul*g*!AvWNt^l)OtH5nHLJePaWC6w%)u3lJZSY)5i{~Lx
z<p$4?TooM7Z@174qx#}JE!DNaolB{fM?84~%PTd-^wErAV&F-0zD)_xCNX(Y`eXBG
za~PJ%VOlx|8T;R|TK@9qgHA$&1V0w@Q6_nSaaB<%HlHuQo32T}bUw2FpO*e0G*K@5
zqMe}o7Md0Kl|Q#p9)}R92%x4_GUR8;AIZX8lKw2XI$18%h&{#)dyE&Fige;C1A4Cs
z5R&ngf+4dz+mso)#ZD0GSFx+#OcmSkcVMa*Nlo@URgC@5rV8{!|JUY+Fk{vB{gUHS
z^8*TM=7)iu^Mmm1b+42zHG}W}WPZ4CRBEK2VidpF>=~KZR5{veT0Z<dIn9a<)rI)L
zKj~^&+c}+XSh=mE`Lk})&a;1R7>d(8<chg?#3#MzBbng*ksW?!yI?QsTZN8bCOb9p
zk(lv*fUN%&h*f>Tj5&JhA<03MC6>w4wDPFYIQDXX<)}I79Uqnvnntmq!B_`1{buRE
z*!hOdkuiCL-H@hjlPjkGL}g0~Jliwle{?UAj{9qLUuN&8>VBZTpRW5Vdw+xO1Ll6Z
zw!1l0d9>2Tm^66)jD@JqlVP*}MAB)hr0KuV<~Tju(x%_evu5>FiPLZ8xm3@#)af_#
ztX@r(JpCq~hw9muK4X5e!A$S_<-hYbKFH9OIfqTad)<j>b}&7ME~G07{y$RYoZ#VU
z`%S}xsIBN6W57ZB&yV6u>O|QJIw;c5DDHl7p;6p@pY`986$cr`UB~Gt?mjtH^8h}u
z{^+g$PBr?LT<MsTf~Aw^^l&CUYn7w}+gQS3N3|}f?va0&nd1?w@keWCtWx17e8)Kl
zz(iD8gPW@q6i_4;CV2|+Riv^jP9D1tJ_v{62@Jt#;cd8+uT=h++9(Hy@O2<MlkMA1
zNlN1>!wxFXQJQD!Dqr+AG#L$_4R4>+5ktshfv9~ZU-oGO1ag?G`<U^;0SF%lqLT?S
zQI}()OyqhOOjKP7n=qh5o4j-CaLjr<x2mBdSjZwE@<w>?0fY;4YAdhsc<)@1Ojbuj
z1>wE<p*IXrs6;QOdpIKNaPA7dci8QVo&62GkIBVnb7~In=;pn%xdI{mQbuK_VVwc0
z<bhNEFDENh!wcR8pPGh#2?OosRK6KJLweWuQ{7sEQqz^2UWE{i=U1!o0Nv^yY5=Y(
zEvpXC&dHm0V&b2gD4IV+AqlYciv>Il;GO@B-W$z+`JPU1S7W}qUP#2Gw{Nrt-1a*C
zjlCSGgi7!w>u-!}uCZ9vldHGk7Udlq`WlxUx6IVcS`p5V^#5F@Czjglq`ss^pSYKt
zVi!xqpcweM0wCB$(^@5KzoG~feMv)UJ=rl8VBli)qPnAH3Q94IZ=O5vY4{##C4B8|
zcpo#Adn-94-r{b+JE?-_E=c|=d;OvL9m2bwiGD1)H^|MnMZ;N&sv+fSSWn8ayptIH
zmFw`!rYg8(YxXNn9-5JF6)(VF`7b%qrs548L>-cDTAo!AQ^-8WX{|CUdRey_fgSFK
zlZ6E0gGjh@WTP3bP&i|fy}HylWxv5aEoHy?6W?X$PY#F%p`MAKp}<~bvfvq)5=Dk!
zeRu=#i4YH|@k9NogSSy$eF?3G3`%gSEXM9p5}fTWIh%_0gQ4ifWG8o?mI;|<XFpYY
zUW-^;mX#ZZ^|q5{SZ|(W7}kg98-}&)Vi;Be3jmge^Pvp4vL89bh6PA>H}y65wMTLP
zp6;Xhw`cNMWb^rr&F2qXmON`2-vP>Ji{M!tf2T|Q2vLLSBv#>b`zv?W=x(nSSxuO%
zWH2+TBKtg1XWquiEEq8XxtCP$>R3LQ2lt17qP=GqQvL=Uf8)E0a$2iOs%Si9lmFQ{
zB()b+m9|z51e$x}s#1lyy=nr_dRZ?_t*T$FM<-Ta44jk<sM?A~PSNje^LS%!nz-?|
zE?Cb+fBwh<SC`fWo4F`|rs|>v?c7wiE+|j}elX@sIm!T6VbU0nPc@Xj$Nj>Z(rW{;
z-<PUT8TIvzp^e`ksfMnT5pq|VT8J!UO}r*p)L!4zQ#BnJ_}CL>CK&LtJIajf2Ad2Y
zUPJ{Pwo91oV~>>HPxi819BsUpx_h5Gcws@3tpsU~HGvM(1i;u!*^?<$Rd>itWv%k$
z=F~>7@YF^_g<3Dvq7y-GnW<O@wT#Si%Mog$vwIRJ=js}MYHMlaHM$wTu)&;ci4Ov-
zvY}LX;(oC1g+2GZvqWD3NqO_u^K$u2e%u=^6;*kkdd;D!rpZW@HEpUEV<ix~nl;8U
z^@MT0aH6N9Wgq1Ze_a6G;=8q&V9#!aI-;(pert(kJ9~0qk}*(FeHu|)7Om|ab3O|1
zIDW?Q#DeSv<v5cHZ+3ED>Ud9lvA9)dR^%9sOLA<ueN^bsRC2zO^^-hX%H!{*lHu4n
z@;b;#_DiKO(d8`BPZDQjlOs!toF)26b~h=}Lu!b;6&=x2J~P`QBZ%{}7kTfC&^h59
z`JuBJe7$PZgLn?h4J!IDmn8*cF`!A*5@YNV$*vT`7(`f8itXc%XR#p4ufX9w*<GDR
zB4gv0(t1E3*uw&){S{)Quz7tq6V3v`fascu<IvKviE3;!4o3wuH=EOWb6@Cq!8k~5
zDSw?QU}hkCazjlwdT};!w>msC5H9ndC9cuAlQ=pdx0zPOdTi~bdJa0qtd0ilX8$Cn
zHe1>CbglW8PNxqC9U>ULbXR4)VM-ZvVCKyuS#MZf5AtN*+@AG@iVW&V7v;{(def6P
zJ2P*_XT3p&9`t4AO-<ID-n`kIdDA!R4Kv-KWaf=8^9C)Rx{yu=V4VesG}Ew6<+jPT
zBg*1Cf6xTP%NF9XT=)6ZdlFx@iQr}?A3o9L!~Urc<Czcd?DC#9vB~%u`@SynDS2O)
z0`X5m2*R!SP}5(h!ZrhwJPZA~P08*=0gA%*DgBpx33&zB%uMOuHJ_fT;4vL-;r0W(
z;i6s)ue*R4!uO1n`HU1kyTN|e%YN1Y?}9M$<{l*q8;E!`^}uj@cW>jc!#N11z5Db-
zlgIMDTZpK|=6&}Whb+-j#ruF<Hz}h28N)MJ!TgnEVgyYQ1MA6T&cV<H?!BerSliw2
zo&@V?9X<sZYJFpYd&S2Z{ikD1us#^IJ9uVF=>cHXlWb8)GtUOY9Tf^U-`#v7FB>)m
zd%<LxLMJoev!+n(mwo~Js;;YmWHc_QPLIZB@rzimSGl0224fzYp8PwWVFh@G;Zo8T
z!mNnnPe<~5-lr}cWA9n-`scM)QBDI#K~8ayjQrjjro`ZxK3r>71-vzTSsP103%wqI
zPig|8?Gs<xUT|Y85kz{Nd^R)OuCW{$IF@)SgId5l)%eB8yc{~b{yQQz*7kP!b^&eI
z!shzpG`+H?9oAQuIsglQtZQ;Y`FLF>&-8QLfu;`pv#vtzuE)FV;}7-7ykjR8uk;fA
zUsjEvQGaSVsMd+4kD{uYe8zBHcxUMZJpy8`S}4bN_$*L-heE*iWJ*`1Vl2nB0YQ$c
z_p+|3hR1uVX+)t&Bt5E#lk-ezJ(V)vpW7&IGY_@EqSHBVvJ-14y`Kct!_*Z{SOqnK
zn^fglrB*Rx9!oD}fM`V(waR~tz1*s~a-ecK&eCb8%(&vppG*!qkx}@?++roJ)Ev3W
zQvej}iXG+I0+J0t_>U8mztS-Pnh=C+2}+&76!AqSBWf!gVQLCD%Vwx_z-srR7SPIv
zgSCKW{V1K~1<DM<H7!cF&=BtT%o{upztZr<*IYaGYSmt?RD)U@O8=LOX$|-4SWfEB
zS=CUgKx_MLi6$ubi{rbr#8kYqC12N!TUXj<RAn{eX63ZsW=zqWbi>kJxA@vna7rk6
z3aY(Sr!M9p)g29`f1q+ssv$kNry+mlvfqYinsob7NC4_76G48Pu35@bRbAtx+Z2GP
z?k8)y=r0;tskQ|w)oJ&TLg?JmQ5q!t$dQrN9Nu-va?aAqTi+k7e@wb;sQhFa{(PUx
z3lpt$9t)Ucujz-K7aL8Kxme*#(nqivB~Uts#XcdR=(p-;swmjR$b_TFA5_f5;X0Tr
zv4oK~8aS$y8WXoY+~Vl$_68kGABbL^v(%@L@!g9K`pr-{S**w4qcsEDFJNNps6zE(
zL<2z$*oMP&tjAOAS9hO+;Dkg&ig>sh>W0g$3HC{QNjgz>y4eHdx}1-;!<WaIS%hR}
zFOL93fg13wRqCFcRaJ5d7R|#6Mp?c*rbJlfv>TgdIxTvSDLeqOm4lY8B}9+)O+8i(
zk`T)9K8xwW_A$sW5s{9Pcvr6D;X=La`d8k?S@x{-YIvvn{XQxc6}WPsv*8X}|0>CG
zmdGB-UlUczWvg9#FYisIoSwLpiOrxiHt6T<gm;(qM(>^G4xL(t<z`IXuM)_d62u+M
zn5xKd2UWMsY#&7RgFO<ia8?uEJ&~Fdn}i_=QV`e>Q_Zdj`;JNb76>~e!>_s^tjr*+
zr$yM&UxTmXkNuDEm5aKv{V?xMd5qSj!lLP~L37RnTels(e{7k7J5~92VhiF4ly+h(
zGRmNDAA}vU+#;+k3t<PDdUi$FwqsQHZ-KCP&0H!5-vwdcF$g=%B1{pzGyW-^^tBbz
z_z&nSjjw+(XtVG&iS6<3Wk-DjzRp&gz7}7Q*=CTxf3&^9GDmc_HVbX9nCf;#+j5ik
zEztHyjcDq&585<>b3=s|ZHF0sNGHBlKpg)CzLuhi$ikO{SmK^~<Tv2!P&MY;;p=}<
z-3ecBX5s4rQ{Ark5`BLwGMk>_6#osrZihJj3w-^8nEzS$I+<m?+i}D<;A?~O|5|+Q
zpRnfOR4YWS&qCA<rk-69^<$IvEeNYF#U%DY)PFJ_Cqf)wi?3$alZlFp4!F(RFcqeh
znLuNM4rS%<zIgcl@nyLE<I2AiU;m!lU$E37%gsVoovCD3WL289Z-K0m6o=RcS$)kY
zJlu}LW5n(E%fC)5{sw$?nIDNe)PA`4rT`0xrAK5q{+3!W_M2Y&$C%;x^(vtgWBYJ?
z{4!_mpUR4tSj3s#7^!Kpr>SjM#O0W@Z-KavHIAv+K8QQcAg+f+Tu*Bc>EvH|45qK8
zul?bx3?)t$zTT<<Usw1127HZD317jN(Bw<6D)u`|x_`I$YR@9H4<KbdX0OB6nzV0$
zufL~Q#eapbmm!V+3SS<Sa9Q|jX1B9D?67aZSBcv56@1-UTCa)G_x1F(_$iC8oHV}D
z5`cqMP780AWxqGI&%&Q9c3PbCZlQT+SI<ZZ-iAv!`OTe5YNC!i86Bn=1AZVnsB{1e
za}*aJAW2CKE^u{#xI%gu0Vdkx=-a&Lj5!e4Y6rXx2f}UYV)k%C!umS%Ry5HbdF{`1
za#vPI70xC9-CNmyC41d~ZIu_4RC(`N#lhXgR0wQggUD*m<BauK3W(u!i_uvQ8CBxF
zg#+nxoNi;vyieut;blXUw_yc^IniD(2)6NM>`Z`ENAb*w<v&i^j~9{ux#|2lxq9?g
z@_$+R_qyHWzt-C@M)^DBKe99blT-P3*!=g#Qu+7q%zp!E-<1CfeS=$hLpxenlmBPj
z4Cp}BacxeO_s&)|rIyuvfpnYusu$QFxqX{np#1B(j9Tb)v$cq6FY$k^bTC`c!K_3F
z)9axMqk~z24rYZ(T@(L&toe^d=0a_&kM;7fJ^Qgd_fymJrh6w9=pZNwwe`xJljj@q
zArDDaT9`R8E8rwA`3nC{om%ktp&B1)MSU&V)fol7+2a?fEz#H^8h(8nxv_w6nGhKA
z`_g(IybZVUTPLB^!bBewj5FlHwOoQfBsK>1s%1m!z|uOATw2nDz4@jdgGhpf6Y1(b
zGkj|4|Fo&0j*V)Nvra!1E42Ij(ET5>upn^(iL&>T;Y`9~K$(OOabv3qqI#Jzav7}H
z#t&#R`lx`nrXAXM*sIvC3c|#442pJ`(cvy?CK2tFUgiXkkeWN*Aqnjn1J`k9>a*cf
z(z?MqvRVkRAGGV*MVWYk&ngoWMHsI}OLx|<p>%x;`c1AW^`p0Ob`sOvE&O06n<EVz
zn)j7_O@Gfi*z|YgbOh;Ge(hki^KU}42-90(Mee&^K@Gc9tI?m?`;oeDwD-exf4#l`
zp6;)<_x*JLBYWRR_ZQjw!*uVr_uX|r%--+P>TQ6%AENX|3b$LApFKoO56|I2!-lD<
zmJLzA)1V5eM$^hDn9fLM8y)j8=p)7*!+e1*q%R;n(F#UvmNPS1PVAcHI^y;0ljUwI
zVzRu$W;rmG<v_Ab&XAEdHfZoc28VsC3=S{lvLr!a)bTISyn8pr*@8QIq}y@?J7lxk
z^5d@-98MQ}lr8w_ezq-7lTMHHzuAj?KqV9x0X;qe5q<eyob=<niyBHbFqH7GQ&I_!
zDPfN<|IbR;1H>W{$mYoiOec-O7%Bst)k{m3${ZsiIg?FIK>g%dT?Q>Qz`b^y+AYT|
zwVm|`lH39sXed1KTg9D;4eO)S+o`B3w}~{><<N9p(v^Gf<b5jlm=dxo_u#^BtlV!I
zixjrT)N#;_2DsNaRc?VgI5zAqlH8wyb6dH~sR#~?m|uA}OYhF&UE(mfY$o|NC6_6A
zHzPMW)<h08q?4>m-jIkZF>}H|>MJ-zmDO!rrgFckD&(*^zDt-Ng(Xkkker$R){?Um
zOX&)YdbF5q(CeF8hkS|MbcO{tDI(~@MM$w(rBmG9^ek34Vit?!5^`=#9-TZSS%{D4
zJ{3*enaRk0+m!iMD#*kLeS3DYDtS@z!sHpr!L<0i<Y~!@WNmVIvTw31IV?Fuhr1*P
zB~MQtl`Kx4oa~o8H0ev8m^^{f`j~?kj!YhrZdyv^)5CSBZSsK5<SMHdGi~k9JKL-@
z3exy9uX|(~dRV%$jrUSn*#+huBy9J1yi8*q6D8~I){YBgO54H4T?Q?zII@CKSO%@S
z7sBrnPkba2Xibi{l}R48|4RL`FjFbv_^wASj)RfpiK!3hgQ)<eTV+~cyJ4c49QJ3f
zsaCGkyPTS2{5?cChwAJJ_xRsK&XIMOCXNJdfBP5aZ8Yv!nz07KqIK|Q?2*#V7#)_k
zV=QoU`A(d&HRn1Wm*{7)76SXoLu&{`cC0gCjdSIiD<a*mVsmxD9`<`jMz3U7*uHP}
z+BZZ=+xJa}ebep?o5BDJvgPcPV+X9UdFB%Cv{1(xdXl#}t>JpAD4<p0h1^+H_nsmX
zx41R@n3DJlM4ndjR)w3CfXdZ(o8~YK8qqdV;0xh5xsqMG0Q8Ea0yJ{ER-~?@Qr9V|
z>!Q^4vD9^K>bf~~^?@Ywed;=jYioE4zv&WYrrz9@x-L##-{eYkf^d6Bh+}=kEH8i_
z1F^qVkTv>IY5Owh&(^1XE&f{!4!ib@#eYl5*WtgVEQS9g_r-spg&hraz^$@22+8UJ
zDVi&msu!R%)%lnMcXia@`fzs5!iUSKc>V0*n9+oI?^Cm@>n5GUVsj*G&2x+;Yeq4K
zEVI(|Pa4^q1!h}d633f8x&v)?sauWqnzYtH3`))^p+`S<m*tBl!y?i^<N@_uYdD8;
z^k+BQBF_vav_=(USOO}KGPU8JW$kQ_$Z{D8yQ(r#KQ2bZXfBlrUlVVDs3I&@uj9fn
zuRd7jL!}qv#7FuFE~J-}E~fc2_#2^3*+e|AMWmKxD>dHhtvERlw{WMs@VolJ#I!Rs
zM?=!DuI7~{kVtsFY7`sPjj}cR%hcL+em;y1+FmKEM2EYUM<vh1ghL5Rx?0wp-1V@n
zT_=cbRZiUU85@6O8a)(`{VdbyAK4p48|HJtG<t<-EBb({?Zlq;fx%h~zddAQ^l)@8
zLP3%<cX^1cau2*u{Wu~=uY;wEQ15>Ccc44=w+Vdbz76V~=-+Hs!)oR+;my$N=wSEM
zH>qeYTO+p1R64+ZG=-0>BsT#Bc7A!bz8nUO4#=X6JZ=f~?Vq=*Ud23+mknGeU|%Qj
zYrkc}xsH9r9K16M?pMFu4Etd3fUTJpy4aSsyM;7HS5J4_iT#y_*>EQJ;KZf_w3|Fj
z-=p(9ry%g%Q!jxpcmSJcrd+ChXHgQ|y`Yi3oJ6|z<<A$scjFE;N>8KHPZfGPDgDH3
zGVD0M)WG2$Uq63jeP(&or<uychB+&A8MI@r;Cj7%5-X5=>cyMGRcOM)y{(4UvBPqe
z*rG&;vNtu_U;TRm)LH+?uJm_9Z3|py(ko76t%)Sryq=mN9W>=*HkSJn2$EM4>V>KU
zNGhQxa`rj;B}LxYHrBVZ7r02k_1yukU}oWaS#lTOW^H}%w`|z8ttlHa=O7uVHMa=T
zal-HIcADQiR1qt3a(uiSMRukZW1Pizl=^SiKbuSW3g@LyoZJ-zCFh$Bd;{{Tl-~p`
zIwkgF4j=a6oc-b4{R{RH-oDYK%b+F$)a%L<LHM1*XKdKJB)K!Ybwzla5-(ID2!FUh
z2+!Cz()99AU^?+lPS5J27XVJv%j0Z>=%km1^pl~Ni_=e{mml%v{^{j2PV)kw*jB&N
z!1ekQ?96>lwTlhAScNXh>stM4CElq-s=twOkWDX;yVEIE(<_j#CFuZuqPy9vQ^K!8
zLd<X{yus?6=2VK1k)<!F61w#T`(b^40j9xOO6iL^+dDc|)oW-H?x4jX_Jz4*;V)}{
z(7cUEkM!W3rAGL~Pf#8y@)ML#1q&T-Lyw~bKxl`(?khPdPHJTl<0pu?Zu|rdd`c=o
za4$1szrkjc5#Dbx*~>oK+vqWM+%1B;aS%+Y9L^HbGAf6ZGY1(cf7wO@=oALa&zxOW
zc(~qZy%^oWrfOL(jD%{yg^|<fwI8dqUW+uy%{-B*6w$`M$z$;^x!DzBw8;!tX3ew>
z7HyNJ5FKvsVcu}GnDyIj9o=(kybo;lhG+4pwg|T@bz}KALvv?GJ&#d=LJjy)jjQJ#
z<UV7_915`Bh?`WPxXu$6dmNnbvb)&u)F)E5_j^j~$x=?kWlr>H7})zBE1R~P-rMFz
zaKKRb?E>{~H2>^lG;wJo!FGbY;5}qr;8v<&>Zi<{=ol8`Fe4{A*;kEmj-<+D@ERSF
z8yj>Df?yyvY$+GMP7dtCzn91$4WTmuQPWQu)dN<ej_>gPytz7;tye2?E=mSVmq3yh
z3@g9eYB+3rN3gk?DB2b#^v~t>z_>{|lkHA%8MXHmRZY-Om3~V2$%$|IY&9XQZn18o
ztdvS<wElS6Q0h~QNvz<iKlK)Esz!H(oB^S)7hzYfbt3%<igutT=SXvPthRe?<ECn_
zKTd2HDQ*BdJLcJ>9BRtl*r@J^#s0N~Xds6MUREP>^T`K%(qccUx1W4yKUv@PlUXX1
zxqeiHQ2Te5KE`858|CQCEz`^$<!AtSpyqN;(;v6eAL&n?{0*f|yeFt$vQF5~8qe`9
zS2YnG>lTwie8;O6JIYOh?P9L(5-_SI57tksq!Sjr4O@%I5n0#~<B0Ft3D$|wVv3Oy
zNs2V;XO77so?FzJLZmx?H#82$LnK+Aj&Bns6P%r(FgkWl$4==G8VjL{bPf=g&*?X&
zK<YllKx9N(A=<5&D>}9hTCtS4pBhus6Q;F9g5vn){LYCG#pqk`+-vAvCz68UNH>K;
zZ1)@Q$3ALBS{{vMi|C=fJ3E$W!eV{LNccz#&ej!(ejV<);vNtRLFfq&hM&YvXF#*R
z-)Jft-df-sxW;?u%E$2NyTuIU`s)wO2`P>ix$I%B=N?U&H2LRECFs7Hyz6)6GJwsF
zTW=%tM^5Fq(n4>;6PdDG_0@+v%~uVtkZi_=&9R$Sz%RT1D=wfx)UciRLSz;F5^mat
zPa_&e<H5LKBT35(^*+|)`}oy3lRWb{47_7~WOF}D#IsJXe8LdyoW8hHUfHHIofRsq
z`IED7z%9vriLmSCZ8W_w@*=*I-p2csC0-^uiJx<4^!M^8a>N^EPTaCQwel^-TfcMV
zJFFX8ES9{#+p&{vwjU%e!|_62u_HqGEMMVmxPS#Fo5(z9`kpB6MzpWdB-9Uv2$9rv
zoM2(3HFgfI^LTH%k&+X?<Hv9oi<YxM8E9D~)?m=`>;0ic1ID}HFts6tmP@JV|3llk
zz(-YG`#*sMf(9oj)KpQ@8kHz2v`~qNCJ^9^On@R*tf&;IVnyM~1hDdS5@C8Ap+$?W
zwrcCGTD?B%qxeoFApsQudHUi}l{1b!RR}2L|NZTKW-^Ji|J(oH<^yxi+3&sf+H0-7
z_S$Q=1dArB0gEQ({}wFEDcFKVM`-5G$nu-KRG>nKoNRhEpXS2Av=JR#WSYL7iJ_*4
zudr=P4rwD-bN6}RL|{^R8@VHmx#3xn#*Tri9mv|R0m+QSjrP&bf(iF%76HgNk$kr9
zX-?v18&B|$n}d5LuKZ?h!HHyNrIJnmQL}v2YUCyzZ$~b0`!wF%JlDyJHss(-XC3Au
zt<rMH#f);`_WgQeqhTvGl{4i>lP&P1*<W=;pK}1d{>-G8oN-w;bgCEnendL#-YwUh
zJ43Dsi$taA8y+_0pfmiv)XDcw_|_I?j#cznQ7@I7z?X3<w7{KF;B978gmQ-|S%EvH
zAVq&lq*70c0x?T2P`Z|IrJXdc6eeTLMjC3&MK2O8Oc^r@5?x+XA9UKLpfob4fW3$%
zbQ*scBQoc7la)yiEp4HUv~HB8v3pt9#>XRtEBZ(FJ~mTI;zE(&ADCtaD)&%Li@g3m
zL%cg#mGBm{$f5Cjl1)+2G=c<xJ*Lf@LLKBLzW`81H_2<}e)hz|`SDR27CPV<^MCE9
zRh62Nl2f6A5QycFKaA?YIq<CEQ{q>9=x3r6+f^-tRGmjuowYvmHPj650|e^N*eNSj
z28H%|PsYhRIfFm8CI52jwSY0r!f=tD{AHlYeZI}z$LK1aHJu8z<;bGip69f!G`7Et
zTTcqc35#wi{<0qeRiN^JBmhW|VcZnk?+(LdOGSrqCRj#ceq{w5J{;{%G$wPS75mx<
zegw;3G#+XSwSCrfPENVh1i>XZe!VkZ4*4LMVB5>q0y&oZf>{FvGjVv-<w%~C{)igp
zpxYF>;;<^7ksqZZ8f;LB+qywAi(is!daA)wB&&|rDg-{;<+>Bvnx-rihJ#{L5XQB?
zD@d|LcH1|Nz`J|`{sNWY=;b%>FWv{~;I<AJC?NeRb$1PdW#{~qp-J*UpOd)>_hrcg
zeV!^gK6RMyN;AvqI3dbLjj=8*`vBg_o1o@I6Zk0!RBqv&BbjWcCA2b_*K`j7f!m(o
z7n^UbMe#kB#br8gP343{C>zL_gvD<LZi^}}Czf$28sa{U$hJM)%>EXJ;8}0>T3fey
z_ebgSPHoMsysUM#HM1v&))nYu)VdOVjaes25jX!_sSsK(S#i|*65hwGA4~L{LS>mV
zL0R-r79L~G{RP~`^=Pp8?eMAck73l($NIi^QOM;8Be{yC;uF2daRrx(?+~7RpF`}j
zgk41QlPD|FCHT?BIszL+KQw51KA?0Eh|v(WLXIfLcQf}mbm<g-5WX;_Uq*ML!Yg2=
zIvB~(!ivwhnIaYMM2b=PJ*#Oed+*}S6+7h)K6z(Q>`P$`4m@J^C{xfqRRYmg>Q=FL
znoY$1q3#l)mZNLixmW44HwbT0DD5)YMQmdiR*{GMg-C!dtS&Xn)f>%np8B&w`;U}Q
zMyBz(n(;}}sYE8{HwAe?<Q}3$G1~<0s8F#?ERJ$N+H|q-VdruB9#K{tFAId?rFdcp
z+&NsSQmDtv3^l=t`I5hCHs1zK{e#Oy+UEG;V2oU9<0DGRaBbzu=TrKG@GRx2C2+OA
z!FNY#8~C8lt9fLUDEvON{*dgNPm|;gr3j}P=;IsPmNvariV^gkg^J*&$w)1kI3N=&
z;x2BxUI`VU!)a`*dy6CQ6U%nB+~VRgI&Et9YEb-b$7>=Nb2keVMYo>>hHyPRhjQ&v
zTqjdk|0HLwC|;oQ4cLST(8S~v*1+HWxrUJz5FUS=6wT*TEq0hg#oK$WkbheTX<C`~
zO#9j}*ctI20(|V8YjWFWOzVD0EGxla;|wO;b&<tKob4KXl;0TkaqOWNl#7Y8IYIm-
zT?9r=I8&q0vq7k#u`KQ-NTs6WY)Y07K7(5g)TX}_N#3|m=;f15B5>(?hZ!-^&z!m=
za~zh!EP;rE;xc50puSwFf|4}#gySqt?WeMQnwq4g!N`RlGOMmep3@j9TQHjSp>^{Y
zsIqrE_=}durYaHZ$8K|}j(#2R8o+)tS)1Of*e#@)A=E7xw|8UCt)(4g!tu#*Mq>6&
zI0$ZNHsR?1PNA{g7`ICe^RboQXm)U`K%jEHUTO|P$Wzzt4ZuSbmm;$I@7&-NuMb3G
z>UwO{TgD<K!nz81nIrwp*C93pBU{^szk|Jg?&?rxXg!8Dy;tk%!zTPi+iMRyQO_bg
zjIMSvXZrE(%+)#i%5mC;dE*lYI5XG7gu}e?nXk`$ePRm2-G<Q8SK^v3T4zMI@6qgM
z-1lC-qAq^<+PZDK-!tsLS>9wQC+J?B$>v_1$+{Qk$mE${j^+B_!aa5C*G<|}Q1{ll
z@Sc*ox^*-6jAg7%^0aGoj?7W=!|Ss4;Os3#@}28Ok^JLzWA;Gbmvd@hU&Zf;swQZq
zo_LBYuE}JrTvK;&O2aO_)NVUC8z*}l+g|}kP3Je9@InQ08M>R<BgXB$DCgF)!sE@M
zR6M8~e@fe4JS%GxZn_XUbSMsdc!~=Ir+bS(V>p{Q1pkc@HDg))#?7^_VM1)k@<v02
zK`@Q9vAVC@EoAU_C98P7&k!n%yApd2;YQ-P!2Cbupl^D67E}4pn*x!$_!8<B)E$~#
z;_$<fDY#Xj@;&Zn6UUMpaCT}?orp|$4^fQK8|cfp*UpPSMN+hVj@gsB2(KokOU1lk
z6ThW_b}81OE34Nt=^L9nr#bGiubxF8>?M$~{v*lb{S{JU#M=1#pPQ;#DZj${)Z3e|
z%7MRQrwPyb>oMXvIj^3bR=|ZS;CdB+AeRxpmjdv?7>r%GUwuPa+qsT#kZWe)L`zQI
zrzxT0ukgqpscUrR9hTg2b2rgqY06`vwy@Aufryw{Q1@6n#mlDNDniQD?XpiZ$4Kyp
zPCxk&ONHl)jC~^No<}g$>ALo9G399z>o+@xXEIeKE}Ll6Fy=e-Hj8aU|G*!^F32Lp
z7`<cx?{Ta;xm9W>w@Ps<)u);sL~n3nx$n>y<Q(A>qG8tOOo7nn$S`@MKG$>@^Eooi
ztWH_`CAEb1Bg4$<G-|(*VaDv&lqe*}oC)O6FFE*4q5c!14^>^zbi~i%XWZT+Iln9`
zJf-fd<2#Nx2C8(B!+i&@klF}cS?xb=$*#dfr~eIXLO8Kg=IRms&|-8=9Gr~x|CJ3h
zZWHoY_y1MFx@d|nxi{uGv9eQG7yTkH<c`Z{>kQwxA@`RBj(bUW@%#}0B9wX1ac@TM
z6_d)`%Sm3x9e}gbpOy%G0$V6^Q^+0Zgfdr$+)GA;G8<5AgxmpWrG7aUpAj7Hxf!R~
zAjKt9f|;v=xNvj_%-|8Y69h8oF_xBo-A%0zqdJfEdEUG;KeRfuLRpB)8X5Hv-+RpS
zE+goN5Cc5Ez17~KrrMep&sxQV7jT&B%jA&H9)3Ng#1y#&5In4e+ul8^u9*&?5+&kJ
zQ^HXqDon)ovua!C7-J$nJ*&QT#6VNWr)SyKPwUvz{n)a?iLIlKNo5pdTgQI3Lrqt)
z>T>JYRjF9#D0!bs#R`0_^Zq6kD<HRy4g0aI+*?Ook%}78I_jcSnIcTBV+&Fl3tC5I
z6J^M)cZ2A1a%wY6CBShEOeVGyc@4!T=ul)=yf|C`?MB`%Lv?87?HbA3Iu~`?r<B76
z&&hg1p6Aj@b|u1tzj2883I;%8!d^|3`hinO$lT%(alvD}q#Yv8c!h6@Y%nv5y^B!s
z4r+SAe7<Zp1Tg0%<Y7X<<9W5(z)9N{2j{=qkeT5vQE&68#_P|UwZ=P1DndHBh~P4F
z2>!wAkY-yN<h=@TjP=}Q*&@ze_Gf85yT#j2c)XXUR;>Td??R*d&?nxv>AgtrxM0Mq
zaR+pZZOldUL_PtlfOotI*@xV}q@v6=#0%Tya<9u{`NsS%ms#|}SQ-_lfyg<ya*Xw#
zs;v9TJWH8%O>M0IHH4X>3BJ|@o1{)M>B#1!z0;ERBiO>C*C&(}MN2!Uz^gL*cYW?B
zVG=0T_UMjlzX6m-DZv1x#B`y{wzd{1+X#>Uik8#64-{s4_a2Tn=$+TW^zJ$;cwrxW
zKHeYXGVPFX@RyHGQw+nFifX?$zG8irr;U+q&f%Ybqd6xfoAb1(v&%@oIqJXdsMD(S
zdu-`Fr498lz?OdhH%osRbT*y6)Rz858{6402v7XLmhXLJTOaGQl<{Zl(YswCZl7*8
zPqm+?+Dwh?8hskrytdWA=1q2{FkW5I5!l?upeG7};Jv9-zq6_SVpEM%s!xF+!yW+d
zy+f)C_c15Jya456e8yK%O42mUqPP?D*L%h+Z5l(5g<i!^6JrzqGGVz*Xf_%p<yVZ1
zZ=-B$Rw!G*W_z*$Ksi3GMbLQ!q4rAUjz6N_`FT>Ino!xwJYOZtoH1TRF>$RJ<K6Ug
zl74pbV{0=x<6}se{PwqMwH+FdrNUM49U6O<H_(pV;n8uoIWieIFg`6U@Eg#oI2J$!
zs^)^%iDSI}6eqxilV##`n%p9wqiG)m3TF!hAQ%~Qx+<)dp+DQzh>i^z8u%8Let&E)
z9PILp9y0fn&x&Gbls5q)vXl}t;usLIR5fOw^ymG5r1fV({H2ya(GbPQrG=#f=9*M^
zqQEbZW{du{M2;)KAZ(y~@()N&sk#I{!NsnoA!*&M>?Bbtm?}Hj^U8J#mvdA<k5Toc
z2Hm>I62iW_RZDRdn?#lPJOO*G4Kz6tXWEa%cl^Q_JShQ_3)WPz8i+;Iskn>0g#asY
zeyc3#QKT<SnLY|>)Wakx@LsszC~MP}8?cy4fFf-d#`!zE>1hyBG7XeBjzE=>r;iHC
z(4q2<8ZH9o(jzbLc43Oe#Xx0)9CYn8n9+<pmCF>mk<b(zUscF+%tIrex3E{(w7g9Q
z!de4`JMycyUpAHQIb)B2P56j`f=<ph_;o)IHU>ei_i3fK!{{)r(n_t%Fc`c4Mf4yU
z-iu-j2m%Up+!AkpWmW;N*@3FwQuzDHwIkn?^@7dwBf2^7WfSnTN!MK$s5-M%(j1Dm
zNslEdX9yI)CP8kibn9q=P4^E$Catp7shAbSl+i$i;Me7>YLo-YSGP|Nbgo8to`opH
zjzz5ktz(vs77j}=4L}UtNE==wvc!vR6Y|w@@-L$V!p(hF_imwHlv|q^MuqB-KLl8g
z4m<@3G<q>=b>mb+DXp>JYlq2;vZbdcRCkleiqtAl5*q)k0`MTs%i)Xgd<E(PH-#1O
zq09z{k;ljCn?w_NDxFC*g*U?HDj=D00Uzb=->V5JcOMo;aN%3{bct~)Bc(3|tx)27
zcLOj~G~G#rJ{8SjUvfxZlQoGGsM4m3-8U-H!>&R|CYnqGQN4o_Hs%pR?3>n_HdA++
z_W)n|vvGaNm^kLzMcAoSwMAQN23bd2Q1OrQ_Co`jL!K>sh<x6*R)KFS@J#}bf}L#E
zUY<wGma2WGjkNhG-l009GG<?(<30Lh_1tp5B~@$mLvM;G5xeZ;%L+%K-OdUYFQ2(Z
zqp^UT8rK;bin)B~4}?%S8>)Ai_L|PvN#UKLR*t{VhEqU3*e)mv>!nrQq<oRDl5WKm
z%6w8N+qwsZ`N_oKwn@x1H-#WkpGVHj3eJn7x+ufhOHSgHX8jRJoSd56k?S4)z=F$`
zfnHcuQKlMV&35d&dTGipt_g=6S6z+WW&2&cI{XVA1+qXujbe~q1sP<VtFEvQc~z>(
z?2)OcxJwNBv%74i?6{2)@teWMcV-1PySBmw?GuAq=$EhrI$8}ezyH9&0<j_SBXv|Y
zc_;e4G`$0IrcJ0IW+W|>OqOI3jG+?hkDY8o=J|Bk@yPbBQb`o+5h<ez!EO2Ate@4l
zJ9pCLn>GAf3CUJyW#LX9))ko(Q^$M*iVmV6W?TNR9aIX}7(ZmP1v6^}`?=%_M&Bwg
z`Z#dgN+B(<9|6t^fF(RbZ71OBEBd$@UED1DSfZ%p(1IoJ2*vyyDpjbER4B+fVWFHg
zSL9(-mRkslbV=H$$aDB7y4rmEdlofp#!9o)S3#=Bz%5pWIs&X8E!!%5k3D2suFU{t
z!F#lCQpPr&MSEJ+-?U2m4XGxf2@>O^Jzv}Q$jI$GDSf-zgF1h^H?l>0%r-WzSL_IC
zhp~8b;I?O^SH~mfSfMC!`9Q=HIMI>O-kp#&kdY&CIqc3?7K&uZk|VNM-=V@$sEh+u
zZ}Op8D}vyRzoYP3JweT$7#~l<pj-<?O`|@qHnDhIY5WxI_Ks&*jp_6j?>Z^Caa<ay
z=X%l|TbcRR+-vyz0_igiDo<hiyi*iH6B`vo6K8lQuUBNg2ED`#8MIYT4xt)AkCh@#
z_-btSj5(o+2gV8o#+yI1BX6@M=K}*rF*4rcJUN%8{$I|M@AI~Aw&QQ7Q)OXa(+Z17
z$$4YKjGYkeuSV$vUS|hxFJoiC7!ixCnr0xo#`@fQ1aS4+nAT&6>Q`&&9Q0wDdTdBc
zinOKocqtV$_t?%?u1qA99`DesOl7;n1Pj+yoO*no>G8?-Ao%x1;8XUoz21x8E!fNE
z{X>mNVw3H*idLE5q=<glH@kfWdrzruTd=9yj~%AlWgA*{JIi)Ea5oxnEA6EEZ&qMm
zm}Xd!DTNCr=3QbcvF6+^wQvjU-b$Gvw?L@cw><hWGA`yIq6(4v0*n7oyu_FAzeFgm
zbj@*XKs=1(g=l@z$>?J5$z?_bZtW!M?B$p~Izs7$Px`a2wZ^N#z{?J|?_J9x9qMvh
z#JGL4bNT?)J1jB@DVUYwB+7+Don13T`nU|u@3_$%#le`Nv95ERntx<tLUZZ}$6cT)
zR34joDmOxePKB6+3j5Bz1<SimjvJg}E$>b(2}T2n(;YVxgS%5pSW$y19cd-~ulfNN
zPw2N##RC*K{m~>-VkwfL_bwY=`$2aAmx*uUf`$t@e3l7GPYpw|m;d`IIastSmF=5T
z@}GGCy;@iUo%(;6gE<%+rI4|C4xacG$oThnS|a0hqB(AU3hYVRWAN=4v?0=u?LIlt
z)*>|4=bb|ovR2*=k}3VZi4#DjW)?!FELNK{%15J8KYaIJf#hEpJzui^{~0Dd>4AZX
zw)+JW)QoiJ7<K2*Ynx#r3HVZ2MMziNA8ii)gf^5$E&?e31q-*O)E%UhSFp(ny5%VI
z2Xc*MO`z%_<X}#K=Lh=@&JR@m)~uOkvH-Be=G^{i+~j>VV<rV~aN1v-aFKZA^~{Wl
zHn(CMMK<p+2K&bQtk};J$}p4+CeAPN-qN%pf2R~hq$n@`I6T-SlS8xUyWX?Lmxs^C
zex+n>S`@V1+cz++pqrFpD;8JWrK*M?BdC;DjUx%alpiV0%Q{gX^I-;gA2Z*6Cn>g|
zj>xC^>pk_53~epnf=mZ0PX|2B9>l(qTjK7aq68QE1dHCEcCBW@BV@!LAXL0H5RrAE
zjyt`8CP9C592o&@4ct}<bqf{ao_{O*|E6`BTbzJ%&>HR>gfPk;Nkw5w$2|A`!i0eC
z#E0@RG*;G!^Zy@m@k*nx_#TqQXN!!w*=OLgK&m;(jtsz=2aJC1XZtf*#dr^b6w<e3
zB-cUl$!55-_v~ts>qK(J*Q=}A<CkbNWDUxL%pJ+vOuysDnkqX!V=@=3Op1mG^~tcc
zCcez%93J%+CF|A5Kusl<aS43YS_+8khlbziHzC&NV;YM_>3_&g-ks~vt6grJ8BC?w
zX={kuG%J+mhc?aWsWiDZ&ChL`|0WGazmNgZ%{cu*v9~iAn|TU)X?1bJYE%}q4?UX&
zskw^=u}Ek%pB`wx5akK_N&}2tq0GnaU?c8m2qf@|7u(|uzO<9XH1FpRgJwcZ?`(EM
zC5JykJNPScT;C|=DFP76%?c1i`W>Ihp!e|*1A2jH^pdvGHbOF`vFrCnW}-^MF?Cj<
z*Th+GyMAW_BS(p}T>XG*L?#-|)A~g4FTcL&Z4ni_l=q!oOYD4rlugwDOD~VKt-xM9
z`;E_pBt2g{mmqIHFknf}T;V`p`lzx5c^62kIxFw{O9MlxtNoZw6rox@Jx$DCOpMkb
z1TV0`fyzZ($?UmA#c$`$)+e3l41z+@&&`xPjd>Mf&&9o!ol36NrnvXL%Sn(}PouwM
z8g1qSh}Wm#3xiuf{@mc!uc4?d=0l5~z=J6|t{-0)brIRX%oBK6UDOB)NSxuv789H}
z*8eE6NxglhqVA-7h0y?hr@WjH7~4vK5jIK<fx}l=TZxC<k!8MPDB%eA3K$(p(>Bq-
zyr4UbM3uSb8x%sJZvKR2sAe;~``<jIi;3pdO0kx{k7O|K={<H}dK$#gzSI4Lc)yq4
z&+VA_l!l${>(t!h+HDCgUHj_l>>tu{Kz9Q32+h+~418*;Q>uSG(kH7_hCPR`S%lfw
zLq>rFg)F3^b_|FS+>XAZ*8VMZ^niE2ib>y5TeVQ&X?Dx|#Djc;0n<!<_Z3Q$WN;i0
zP9*e~Qqb>-im(^F{|cbgVYVXFQhHYX%}EJcq@?Dy_X5Av)XZ7zmwJ?$(>twHrB9Zs
zq`~5^>GN9CsU~kF&v+ZmP|SQ$>zA4si@!sThQCWlkXWBezaTCBjYms=ok{QG-Z6U(
znoiooBf1$H`Y32*Pa|DJIxPWXKL3NRSjp`}@1d1|J|%TDy-1fbyub~BCWaKQ<EX#+
zg$1#Z)qtB>{h8V80Sk?O-(&$ISSr~2qY8i)`SP82?g}*kEdmO=!1E<o2;XPUhG<dV
zK$Q#Z!5PuG)fyI7?>dcKX%YcqXxz!#J(UQ~GJxxqC(gYUvAVb2$eKe#kUm0gKkr#K
zIs(r(fK_&BdzV@gSvCduOM1FYpHS?c)S|N;CGjoRJxe=t_^QuEJYWkz?+qbEtpBdI
zOpjQfeu|F1OZYQ3yoPY^0No<Zb9ztc-OavlhvqpD>GS(KGae7K|HCF(7e?oWU$YRS
zh&+n$zOs35QQjwQ-m$5?SKGWd*t{<@q%Xv5-gOM6_nOVyUwNn6yoIT}-?w=?BT!Rs
zMW(+kOTG6fZ^Y(ZFM0cXy+@a^e-Xby1PT2CbE)+}ALBKFB#E{}ix2MOBaW{aC6AZd
zC*Y13`a!bXh#OKx{2+al0qpoL{S4910R3niH7=fP?@3Em%hIr_1{CB}3qSe#`8}VB
z<KwwXdu*aZqJ2D5um7aAfvVe-eVcwZ>j!Z;BmOo&e(zebr$FVqN}!q&qm$QR(i$rl
zCY6iz;gh^W04s4qVx+n~O37!XC2tM03Emq@eidc@Br(dbx;6is5og((q1%~&dn+bb
zg|D@YxFkvAPNY*JV6Iv>PD$c_BG}Iad$V#?1(NtTm~X{E1-7cU!f9PwMS0>f?<66^
zH;V{!Wr82ms+{mr?~?}QY1#DfG2Yt*TYP~JI6AXdqEtT9c$(PMmefZoI|dE${%VMh
zCFNenN%SygWfw1l58b~vL$who<no5r@c~i2l(~<+>7|bQnvT|ci$88^;@~-(*_)NZ
z?Rd;bnvpGbFNcGSaP*t>h;FgFm$!n}-e2DEcQ3ExuK5=Fm`lA2QU%Yo1wZ|dZxp<r
z2}{8}(+mE-3NBA8*wA-(zT7b^OYLvS6-nfX@oC}BtykBUbE$Xq*rkwDjC^V5KlNYn
zANL;uFwlq3FQCDWyRgw5<XIwL2u|Diea&NuoNqWqtKqddJu)&gmMv<r-#25KsOOh-
zQNyS8jrX`<$j4HH${%|7!DY1i-CClteztHmPl?N$c<K)cSy42=mmg8hog=5!1}PDk
zw-OmDt_}r;RC5?kMXLxoqxO1gDs~Z*G92EI@`AO(51US_J5?1bJ^%|3EVqqQ2`_42
zhW)jEq?S=pcd?CpZ9Nra@P`YH?kyU1*1u{Lccz#Y{ZSKyvl+=25$?UXj0#+?g3zY0
zqvb?j)j|_YEzZAFOJ7q<-*42?*BeDGj?3xuB!t8j7aB;2-(zf$?O_0GW;L1u*)4Is
zI#QapWoIF4+o3T~@h%rkIKQ^F4Z4JOh^Zx#Z8g40xmNE%cd(;yGt_SON>XE-nWvy?
zlN4$TJ(Zj$GHh1{cbFz+_i3G&!D}Rw_c%jD!IF(J@cL$@!X4?!^{1|Or(|z>joSNX
zi^vg!!S3RGP&ZhFmym53^tN%L8yqbBYmRW)-gEg3y%UY)n8$N@dKA4LMltt{@=9{O
z%naH~W)`@{8;*{?6mbyiV4I=Q+%ty9CY+HIEJBGpsW30xg`Aox?8VF9Nj{m&GrxA+
zh;a(yx|&*CH`6sVeqydOz9h#P|Cp#mFq&IfN&z{+NNuh;!iEJ9Abx3+rlR)uy#Fz}
zb#Vgvy-1zb1`Y4`=$$|2&jUPgC-7N;=1?^s%}Yr^$}T0JHJd=S1ybht6XnH6Fz#Un
z6azils!EOL5jJBu(ZZo!uiIW?XvdUL^nnIbaJ}~xI|$}LZ#^0WvdM3Y$)<gf9F}p|
znc~^6oV|;!TA+}9tP~9k$RCO>O&0U4HX<0`xhfX%5##lz)B-V;iRIC4<$)*P3Pmw0
zg$c<Cj*;_hyY~DdV}w_DF$?fWE!bu@cWeP_3mF-Nt>nRocZ<YI7G`RZ6FpoW-9ypJ
z{_#aNyV*eDOn}tF%$_OUq+aNV*iROTfHy&|y+^~OHCrl5GS&+!iH*8bQGip^0Y#UU
zvV*@N5H*tmbpZ*d0A6w!Zttnzs;?2ftLqvZ?P3|ow6(ig$+6LVRDSR4-<rz3T5ZAt
zr#w)aGAyba#B$n3ESfGhh~IXxFfom0<63gMf1`=$b&F+WcHR5pbrNfr(Rf(gYK;F7
zFR$ByXgZl~6~xP8(Ag_|P2*+c8ADBu@m{a?g#m{#G9;xdcQq@CPkJG_$9VUpQu$!=
z-ZdrR?+Z$BZA*SKk@j)j5&R_N{)n>ykfoOkt4n-h?Nuu$A;NucARi$F3{-=h@eD<K
z<7%kU?AV+lxud{|egcc=FLDElFgJ1wjSng9&zx@dXYezAhbf)?zY_#-)r5edU75>N
zm77Ht_<%VrVz(fZS{IIw=p+tNR{z2@`qS&SbsARm`FnV@)iO4N8))ET5E&qCJz3JQ
zFMx{xS70qhD@=X<ktlQr?&Dc1r@qoW+T_VnfbLq=RQi6>C9&T9M6f2j0^N+%=Ww)d
zF0-IwhZB27FU{~Gfe3_vhr0kT`gpEt@GWYV@`^=GFDbm-K9k>VLHKPJtI*JUlJFY@
zceB@!I!X9#kv^0l{DuLTMdjjGRe$z{trIhz9%XvZ0FZRT-?~x}y$49BKXpnh#hd*S
z#K?WT&`@&M_;E^34s(X9jZSz~RiaBN2P;MB%YmWDztn%1kycy+LD4BXDAF$dV9}~+
zCx@aRhhl%W%wUI?|0nhN!a`26CXSz)cS$VwHQmM0-if+~$>;`G?HM2PP6f!r-3y<9
zhMQUFMoP?N)Oj-6(io^Zm9`9xO?rZ(icR4zoW8=QSS{G)ZGZVt(<QM<uP`KoGTd?w
zI@G}w*gG52M#P0Zc&FI-Y711IP9d~?sm6Ew2(G*t#o@CgZvkV+VziJO8o$kFCEj&r
zuS|MLfH+qbf>3%F>CqotWD$=P7cs*~L0kaKV*LT4EEl;+d$f@h7co=Q=OUlKsPS#V
zMT|rR|J6*%^VWV!0?I^5VExS^%0Hp~rcBRqUBgCtfyw1s;p4D^)jFC5(QVcmdBLdX
z{n{^Jgvw24NfT8HB~OG6TOJZFbjtBNk>nv0xO_G6B!xk@Kjl2GCUibUiW9uo7Xk|-
zG24L8_#EU5PM{dB!wfAE)z63hlL31+%2l?9QrVVMtYIKBVmOEU*Tg;q8J|b|;wG^R
zlZ2M<^Nvvx%RYt^m7Jg{_HoP|mVKBBkj_5-iy;ESN%o;p5p1S7Z70!nri3aL9aBzk
z*&<V|=fY^kRA<oV+^8$LBUJP;<@_5gM0b$R|CQ=9+DMV<mTYA9h9n!A`K)0h(f17-
z>5Z}V3ze{uIING>W?z;{c&SY|)+YSE5-ucRg5A{Y^HQnKv8gVwslKOF_bb&Ur0S4L
z)!wG+Y*P(Xs&|!2cFX@@=fiTiZO@pd?Rw8NEmNtUClws-a;Ba)U@83UJBiMY<#L8F
zALG6MLXyh?U<25b2+;!dZ_M>;7vXOW(JLR$Ioc>zXX2szk!vzCp2=jRC{mk`Q#G?_
zE&MyV+>V#gnvmPeoMQH!YyM>`>k#bRIB_ZGnr}eTy!|vN@v1rRYzx2tWwl*}52bK+
zJ}4TWmL;r@BcJ6B!b?uHA-V>}>2UcUpq7<FDUZ1?$vGh`2&*@c{K{qCt<rOy9a$~f
z5sftPSZ>@%W+qd#Tpzbs?~SI&GA7y>!%2~C0(Vu*O~KtYeH#K*Ujxjt=o`kol^d16
zHu^B#9wujl?0mJsSOyu9`EN-|5N;?ER_5N6>kjMe&dx83diXKeTN<s_p1p{Wyd!Gw
zOw{SXfoP4NBDpdM(U>V^{RS5Vs#a3FdsA2U>MXW12XnI?p6JT-%3XQ9PS7jY;_*7x
z9Bgt2BNtj~7=$g3)x3bN6a?Yb6Ae?qo&z4EL<Kw{ky>kjOZG(s$C5{-vgC0%yHZwS
zfVQ={U6Pw!9%1#@vtZ^)-kqbq#vCBr$$QScROGM_9BoQPaoV8L$B!xO32V?YafT#n
zCw!A;FRRlXuO6=iY+v>L*jJ0D=;-a*Ob$GW9#57(Ad+|2^Mu6utdOP=@!;-8Q2|cn
z1>Narf02;)c|W^Ad%Lw9PnVQjfGPdOvJWFwckQ7eY;X2<ex#~uL#`3rareN<;(G5y
zs^i@=1XW4B)2YsspDYdkML8<$4l4$8)IRW}7!Uy2WBNG?EP^lebTH2o8_Hab4;*4W
z#p()5Fkmb1<MvL|q0Upyle;QqQCoND$JLBy$Blo^#LD(+-Z53ZPZ=q33c3At1;uSr
zs4iWOmF+ZRxtaiL6gr~U+IS4$x6lXGSIJqah9n7gSi3PAB+T1&Py@__M&At}0B`{S
z_RhzC8KANwPOCsWubr{c(!?i8TP5jC687as=pt^2_J~W|QQbAD<nbOczUchaTk#v@
zeX?Jvhmb1V9ZpE9cfFgrts}J+zgelO(^4S@n^Yllr-xtGAC&6%r1C4;(qa={k+V{f
z!ia9^4VkR?tJ10v-l!q*^v;b_s`FE&;5I_+vaeSQ!U?$y1o|k^@gx!;t2H7VY>w19
z!txQJVpdy(1uCoCX$~T}_Bka1xdSnj7Mj>zSrJ#H-Pa%7YlfMkcKeslDgI{(DKtaP
zGc`k%&&*JR+@1qVbX|>e)ruKh+<ey?a!TtIzs~MGF2E|yXS3H@G1=~B0r-%0D-iv}
zALeeDBcz68+NX5#oG<ObNx~T6xSgRfZ<wt6z1QW(iq*G)$S5VwNF@&EvvQJwo<yI%
z-h57?mPfxfi&J6H`AWU^3l&|aV#E2AJBBxt#~V%ZRLOZt`xnxhNOK=HvJyu>t&@JU
zo2Rw?;<H4aw7^ZSLCd1s%c5~}GbMKG;IU{XmX2PIG4b$)dDHx-;#5+ati%ut%72QX
zQPF)<MgP+xpDVG%Q8Ud#WSolJrOkBuh$P1)&Ga*0O*9BKH67L$C~=Jv3j+JqgGZ<w
znGre7>#kcJMaHyxcYE8K8ihBe@okj3(yxhSoYgx(Bum&wNnt4cnX4?i@d5d|Z!r3{
zuJ;hBa$>n>AsxPH$du~wNrVWS=k`b&jCDf%?<T8q+<PQnN_bdrcDCw}Q0bmoo-S3C
z_lgOxS6E+(tcKD+Y8zXmT0$x+ufw;j-_H*PB<2&cHq4?t>XieBtu)0@E8)-#Bna=$
zeU=u;2nNV&&fi*qoG4CA<Bbrq+*>+cq9e*3|K0`W{AEfd5oNje`{r1UUaz$;v>|Kh
zQr<I4&HcvS!{?>J;CHAJye?Gici|q1=z_r-bm#Pfo?7xG&~K#>mSf%^BP2s^I5~`3
zCW*Zo0Wk<I2iBfxTZx~PqcowACZzg)Jh~0<Hz>1gMe258Iw}0}6XI{S@!-7WC0?fC
zeg1@9uO!1;6!Bm>9bBRK^HTBZi%`Yuaei7mo=~D=Qi;4Li__qAuOfGvJ$-`^DO`D!
zjRr^oSYXfXv_OSBOk^FAkNYa~G;Eqoq=m5XC?^!f@eIUlq7QFT1-zO{vAu}43T55`
z?kTWs`8zGTk`^_CZKsU3!~lg-Uw%w{zf}CuhT#wAq@`b>^vBxtIKN<EXQ%;=H(zTM
zF_#Y|y2R62*aOP55pN4(o<3C#=M<RK?}{=PcvsM-6#2Vd=^rM2OMX$cpas9En)|=w
z7gXb&v;l4rv)m%E;C(!}GPrNG<QV2NUah!qk4?$G>)&8g8d)e2^f1Yu=5k|%0K>u6
zkHk;wy*qaake8R3Wc$7I@Ny_79qaRypWsH5U^<k9a<+R%@LOgK4Eo&bCkT<iS{4;q
zeyk3Q+v=|s=k0D&7|u(9-fG0-WQRm@8p!fb`EQ7+zNi!xnDHKul>}h7+e%2dti!$I
zuK~4UY->|FL15EpZ`Dk~C|O}GRIh#JsdcHfb6>@?Qnq=Uyx-3IstJ-irmYc3KZqKi
z>cR2t5pc`h3+r`$f!VXmJK_qRvB;^DZfNe(?)+k{!0bcg5ON<;sN>G{l}e$axK%3c
zms06sc$N1S1~9sMH8J#`rY_z)b%+=WSHzOxzGQ~-srf@=q2~}&8VuTpk7EZz6(zFG
zH%;{}{*3Xg`2=m0_<Evb+W$REE?G3KdN!?a#;<_ota4(#7x;yUP!9bobJKmrs<W#4
zQX!6ogljXx6QI-~UvX9(pL&uL`-76n;!};^;_3buj5iX;Ps<w`%YD*Un#ELzX4axa
ztGeH^b$?6;pa-9NZ$C<c$SXOd&j_Dmg}0aB8d^?eR*dnUl-Bg@%EAYTcQUK)fw44I
zJ<f09XEyOgS{I?+a~U3|!(O@33%8g1y#0xGtTe+~epIP=oBWz&+2My~ck_PSvh)YU
zwm_1enN@QCk5!hI!!7k7)9RAShA>MLlmheKFv^BKTTnJ!k?kukMx;^=v#GAMsrD+>
zM5U7A;;dAv?lx5)n`*OCg-L~A{Kv;o7DOL5#(R|jA)~cEU|+gQV83@Rc6Z-Y2~3n~
z_Ft<6Qs);+FSi;G^kq8TX)m(f_=v!~tX}c}BP&%r-{Q*(C|%026+{EQdyg9*He0+(
zviKT@(yL3aS!M|M7Vk!awO+mKasPa&5i6E8pPy)<j}V8Is?&*SQ<cR8mI#>jK8Zjw
zNRl>3P`}U-R0@3dzeVcU64D|x(wKMiX;P<jcVb`$x{@x`Co3ePyBnS)4F|5LcZN?j
z=a!DK=a$Md<PKXTR_NIizzZWBW$USI#`QE;x=TjUz68E|FRxBIIhX3eC!MGcv<F;3
zGF0baGkY#<(}u?ns@xwgGC<W)<T+Gefhr_6Hv}suO+zkb3qoCR)CKy13Rqe0m}QZ9
zqD~xVd#9Kt;=|2;9_N2z6^r$&(trII^<gHZ(h+o`%Y#Kng6xu(_2a68z`}M68pd>i
z${+H=u8H+j7z_+tZrn&0+<XlyLZ|4EGw;w``f@W%5*oT~Yp=Eb`7RuxXD#6;v;=qP
zd83xdJNlR<W937AvHYViCb-k6#RPX6vv@2%stV+zYK6IYu3pf=KH|VyIk;y)BcHSn
zpW0GCRpw?bEQ@BjPE(E3w5nncr}^3kZi|uA+LyYW7e+g~?x=-X(NS&|&m5jPJac*G
z^33DOU4IKZ^X$wspJzVLt~|S1pCSzvFC?(S$-bv1IlGU4%gpWq9=)ms#x9xxKoCQG
zcKY!p7p&RUrt|>vOuY^p8BqTk6fDED(Y#vWZ8PB@KYkzk_mO9iSq}*hmtjMIkY$F&
z4C7D8Y6jc;Y@QuoMqD}wv$fJ!v5?yWkxOeX6JPhW3=~VCSo5W}a?YM#phu0@_WJ6#
z-hA^-l!tR^MTR>hM<y|;`RxwLZShTh>FC^&9CDPYZvQ*iert=?7VR!nMWG@&D{XuI
z3RS{^gisAz7<V>NQ>nCfD<a$4MjErG<_tpl@_<6Q>tIl?_xQBqHQde09~RBp;FnrE
z@0WRH`31EP{<2fnZ-%>N`Q3*_%Qob+h;<ZO_M0q4k5Dv&V47qzPqEQCik_k9{IuxG
zeCc#;irgR`(t`7rsafUnH8w>;N7agO8QV0!85Yas)JjIi@K|n?O}7nP(9nif)7~7~
z%RMU1Rp!|C9&gHv)G*oBRBV(<5b#40<}MUOL+(5?BJK+oZ1oN<Hm`3oV5<*=>IJqO
z0BXSYc4S90*dC=7X<%zPADZE3ivOqZON%`k{L-S22ESDF{{ntZ{|tURTEZ_?X4yC4
zw=o4j@J})Z$Y=Q--xiDp`;3J~3j6lX#@YgTCO7K1Sq)safW6aF%q=m`O5wZiOM#Wa
zCTq8Sb=K`ti{m;w`vW}L)P2>dz4!I=^1t-CY^OU7%n8PBm&QC8yK|9w{8mCiF!sD8
z4}8?!v|K+c<?~%HEA+FvE<R~R?LR8|R&y!D;08I!AGhyt(XVsi;L>g$>HbO0xV?w_
zUXgvh>O9<OU|y+v`y&*{K~cjq1LY0Tp*?GBemVPg$u@Ndr&Mnnw{KQ{|LkrH$#HPQ
zz}`BQ5~<Y<E+QuB^V#ytQQ9Wx$u$w`<oskO8`-g-TdWMdF3sH>!?z8st@->c(_dYu
z^M*-`Gcsk!tF1|#b-Q3-{frZoZ-NAmR1l8PNF~^$g1B(=RrQT7sxPIHrZ&c8A0MJ$
zmAX$nMOfr;n?NX`Q?lioLhk9Oy{x_H-i6jXR5>@|X)Bf`%PaVV!}Z@rJ)-+)BF)00
z-TjlCg57Y|79}&G`Ix=GAuo@8@1(zpt5?@0u6Tz6p0+v5q6eR9>y09t$?@J<ADCPP
z$y{w(=JM-{#>v#JIn#S4Q}<-1w_+`-I%2DeZYI;G&0QGS(A@SMv#HZlYyo-bs`xUJ
z8T&`BGOK@91m5h~_yR?%%f7DEup9a!`K<J^QZwLVxwq%>e0x!AJ9(sayHxQBubE~Z
z9Z$SNCU9u{ypO5PE9r+nQMb>pUxdxYI^Oc%;&EdNOUy0T@E5jav_Cb9f?^#XrAMVp
zQP@hAW31BtL=lo=@<LI@$3;w$$bx&vm0GHFCa9Cys7`cXQ@KrT&5Ta%gRiiWt-or=
z9f3-7lx%f!ocfX+1uieiqsh`+WO|5fZ34GHkf*x)^rmhny5mJ<_O85F)lQHd$OM^r
zq}Llzk%XCwdnS%aR-8p+Y<;K*Th$j<23;-?>p$YeK!*3#@5NM&p@xj+-u#0;FD$fX
z8?Tx#_J~6WVQ8{tWcXLYyzC1MC<mh(16b!ibW~8zz0`p6`sD&8r)yK7y!-)4!T(z*
zCj0WQVK;@{H(=LTmj=6T^zGXR+jqYM*lK^*0&I8Rmkzf1=1X8R5SziL_m92?Vx5;7
zh<*L1qe5)hNCUBp=|t&LHmgz)JN^D|A7YoVqX@*7*QP;i=RN-ki0$YM#C}83X^1u+
zylzIc%gk2_VuShgYR)$h+y1<P*vbcw3bC#~HV`|JPL!Ja5R%jNbI|nL#<4<}D-gSD
zSsKJ%TJoQO*mEZUv9T1L2C-gdrY5K6Y5u(RajY|+-u*u^5PRx51F?G^I4Z<G9d01@
z1?ON}LaYWf{k9=?<X8i-aa5L~Xwl#PCm<F%9*CVw(P<ECGANgX*j^t2eTZ%6)0@@D
zKy3E224dIVe^iJq9cCc*L5(1m&Jq@iHh;@>Y!hdy!LgoHmV(&jcmF3K_Tx@K?CZN)
zK<q7pa!H82<|CjFv8VX-uIz0f7J9}&toI*}3bEVE4aDxR7R1si|5b(}f4h|bZ`#$5
z^*Qod8pKYgTi-tA@0JI|RxWM<v3m{5B_S5`5zvR&Y(Bl>9~y|A^fv>sHh(xO#72b-
z#46}S=~7T2MfuN$(R|xDb_enY5PK7WJcVPQ|K>jdv7N^Ov3n>wZB||nHaAoLE6rDm
z@`w2JPW*v^nD>-{*y?+a3bF4GH4qy{CtBjzLD2NuhS;^nIY^)1%LXbbnbrLIKLN3q
zIs&n4DLM^e=bK5I#IYiO%KA8V5})3eJq^U3M>U8%_Q&5J6=HjW24aPDq9u;igQnj$
z#7ZEP;MhbNb|oRUaM6DPV$ob6)|;Z!AeL<=X%b=wd<66%=JDyhbDn`%<r4;ClYVzp
zh`s6<h#h7PSGqJeZC2iGDDt;U$4+9!48(p!WhsbVweUXyvCEGIVr?io4Px&YluJUa
z&PPBWV$bvGJ$<f$*vS7e5WC=>qeASCAqHZb*pY5IE5{p({Ov;Q3sy)#EQ`ug5Ig(M
z{{+NNKL&`cj<tZ;g9hc25L@gcpbxQ1KE2n@F%T<w+(0a6$x$Kp)4>K}PoN8F39(+_
z@V7lHSFv>f#MZLykb>C3JN^?8`#cAT{gI;6aBQ+cxg^AX<|CjFv5|awf9PQ#w(GA3
zVjJ#0D#W@EG7zhz6D{SjCeZZ%A7UZwjXAP}n&$S$q*IRd`7?$BDR8ZH{}bS<Z4bC6
zQ*;`*N<igiM)?Etm12~qna}Jy$cJidY)aiORr?YZHUKIo^ZHMAq2-NbPgTNb_82|5
zP(qLT?2$b5w><k&y*Y0G36Bc$T?RoGy(z-?SpOhV-Y7;gey=I2|16fL-ZN@ztp9)M
z{bBq5bG_ed->=a7Li=8>_lSKj)%$e&-be2@*!Lp6Ut`}-(fg(Ly_4RD*!OI`_p|Q@
zS@3!1+4r4#Kh3^x)BB0M$1^l`c9%gI7N|ObfVlbj*idq_0b!$LJQX)TD_ci>peXb6
zWb3Hc6=i<zY8~~kqV$uP*`f^#6|SEIo6D&NRNR`Fo!}&GbHFbWHzo$82V9XDljxNm
z;3S473ep44Ph5~dA=F&P8HpnQvw#DU73hS_%nMWv6=YEP_!DolAjkGfjV9om`tbEX
z@5R5@5{4D;BsQcK8K~?^y!Sgn`*@qu97s1w>@2@0{vn-B3z#2&t3^Pd@@94+;<C<a
zP8F!UN`ZfF73e6CE%z2Fe@LJYWZS*U?<wRL>7@k}GCDovF#atPgVRG23PI5f{QPn@
zE9Cp>A@vGD0oR=7IYPX9ekGV$cm}E-Cn&C;=5%Py!pSt+*BZgTR-lSw(@k;xr0cVn
zR;zz)%_bBY&x{uP!H*DUcUU{Mcj||rEhMcwfJ;T%I8lt}1C?vU5iu;Cs~Bu8XN24-
zjdhJhoX6FM><EI{nJISM>$#U?cc2pGGumag^vpre#JP*CLeia>6Y4iHFHouLp~|C|
zVUTte2X;zw@;HN)XL8;~&QiBs%sCu6m>HNq4RNpBJu&EBlj9ych<lu(`xD)0|KBnP
zRjvpZyR-9j7fvv`Clp;?8eJ1<YFAO_&d!N6WryDn#Rlh?bQRl+hh+usdOh)O>U=o?
z)3SQ4O}tt>EH|+D^~CdLJE(boBhuIwBmJJ-2swa3j@+7so$YAe<WE*s98RboHn@RG
zp-po~*wfC7_fj%9FIcqRS&H3&{^kTZ2~zn9f|zd1syU*om7BMOhsD?hPddkHA1r>8
zL)+UznS1bTwawghb2pVF?Y%>&gk{B5vm?t}vOqJeF62r}Yz^lUHwn@uPBg19SX9@l
z0sln<LeZU0&EYHuK4feJoWQ`kPz=3o>jrSXDAj;@rk5L%^>(A)^60+AWL!tpe3EsE
zJaMVUVbM{2GA@aZ>BA|lowGR+f;r7QoFLYw|4>5>-VM1u@E}^3$MFYq#?<Wla?&rl
z(R*~hdUJLtdP7dI=A*1&=4YYkc=p#h#uWbm=h>J&)O?&3iuJ}?t1}AdG2K;-<901n
zRrSWN+Nanmcw3nhAr2jZ7&yf-1xS^?g^A;pnMyXv@}NWr>e#9{<1_(9v^JsR|3zJ3
zV5ItRq!A31Fo@}k`*Pu2ppZDj?nee=)7W{aTpP}17cBaE<Y2pSZlY&tWHyL$S9Rjk
zQg>KR<Y0EiMkT7qNk$M1sxQy8Ie^A_4m-)50%S5_@t_G+pb6EJssL$SsA#A2LK2Cb
z*cD944+8U_qe;Q&?qE$5m#ul>OJwlm-Y!E8WB6JyEOud|%qV&S7FiVm{+Fu+6D&7y
zhu3E6594pDd=&g{Fo!dktImUc$owN%v!kuaDIf01<bm7%Nb2&U9l_!)fyy!}*UfM>
zhq<{c4T+psR|aJ>yVw_|a4Fyf25r_swZMSQ@=sJ5r4H_I6W`15Hqj==i!)<frn-|J
zanyh_9J7xzR&<z(9P+Z2B(j`=_FcADntouusKM(VjOo%wS&^i`H&)=AqVi?nQ6K}4
zvBI<&Vnt^hFPSCjEMGA(Zhn^iq|Q5gf4@jhw4OCqof>fMaR-Z!1S-qnLLs@<0wj3}
zHbI%k?Ss+D1Z;BLD;dCLknhn*ccMjG$YTl&ROXVgT&@6aB2C@bp?<gKRn+45%FfUF
z^!ZsIsMNNl?AL3)&NBDVf8Azkwwa`Cq!&1^T3XW7-oX<_@xEz2ofnfK+($GO?@%21
z;mBS1M<F-~SEPwVgewh<gV9Z5Icr*Ex%n+wiTJrHR>~0^^KNKt%;_1!qh+UO42@1Y
z9moTc!ou`HnX5TOk5hxPJ0eD<9j7QLz@6)u^yIVApB2jF__)Iy25j*yu%sk(b()ye
z*cX|sV>s*1+?yCG{=8xsh$4cp#k<GEv4v!faH9TaiT(VgKIhenccp~XtD5(>Bg;5j
z{kfbI^i%rX-X)|@cp8_n)KKxq#6ut+eQtq!frWFK{qZP27_MyjBSNb=<^91-tr)s|
z`j9b8d;M`t$>OMP*w?!hg!C?&M_Xe3e=-Fjao=3?eu>^cvF{h@{ayQh4tc#*_WcZn
zzs|d?xCB2{CU3C6r&IA)g?*SKPS(#@{ak6%ArPoXlZ<0-1rJLPzAQ04{c~_)V*2O#
zi7Rw(2etW=z?jLLcbl3#lOqdtCyFdskXLjw!UuIp#!I!XqDbTRnYw*I-Ijo&_Pp<T
z<};NrywTf0h*G~w>X!Bv&HJtMK9fXkkFN`V$I3)&qB)@!;%x?39AgfK)$Cm0!HYeQ
z!OLBR57-dv5}uHU3af31JsCp?4peN&%l7!xQpq+Vt|A)TyH;sTS@pqD#4?=>5dr3>
zL_KCsLt6BP93!(Ks6}2c-x=ge`qW>%JrH?=0!z8~A~rcw_PQDLS}@i<1Y&)8a|B*&
zMF9omSO&r|GMDLG^YdgK6#d$X4Wu$i?*vVvZS3!xTM)zA*xiDi-TyF+b&W6+i;$st
zg8+DMQe{@;V_}CUZaYBHfxH~6Bln3@4eh7??OwI&H#tRkhUP5K%7KC(@cIf1Az<@r
zH3F$ggcO?n@x4jR|Hmx`^VdIVF#mlXFPu-Cjj82psZ_7nR8QJeYnAE(QZYY%2yS_&
z&Q?tvvpe7|Q)8Iik1-w7?5A4Dqxm$)aoah?gAqL~cF5Ai&0z}mFsD(WC$B1k3jRg~
zLHAtFz~?wco*Y}>94bW>3B?Ab^Fp!H_y0}i0!^Mbld5Agvusw}nhbD5|3i%2n@v(F
zO?7Dnaq=9g?dIj7*sXb?Xg|%5Jnz3L!VImjT;`Ccip%(eltY)y*(1hD@zH^h$o=60
z<2B)!!1EwfDSq-nyRdfqnHij?5M=z+$+2TV1cB6mT&1;lUpgUs0oqyuex<=H@>zG^
z=QI2BI>|~s^d8s<X|dYs-|hEpG7?=h0aNxnRosO$1!f#_Pt%F|6A<lWk5>muu(+?9
zAwKO6#bB(+89%T@mITpE$GxF%uy|)6azCX*{7VCOEf1fn?BDgX51vJDNUxOS;g$xc
zSj^w!dd|WScUk9D^mbfOouQN^0ZlDXQ=OufxcjI%1bTAB64N7$G}WBJAd#}Vw-z|=
z)b5UZY0t8vot(BRje3edBNMewCY|iH6AREl)j`UNYn`^pab>K)&=^w*hbw2zn#1_A
z+0_<n*E?%euRD~sbcpW5{4wul46(T+wD@2k@{ZzS7p{>BXJk1ay_zuZ#81Y7)j<wi
z9>t!LRyKw*a}PRA^~Rnu7-?*O^Us>$?YO-goy^``5EyRc>e28)01$5E*3s}mS`cpJ
z!qM=-e7zOujqZjUyX%ckhZ}q9t*74l>P_CjkpvG_JPI($ByO8287WX%2$N+*0+lC=
z+uA{W>Q)xg$MRho-BTLb%3^+FsEAv7R;1xW?Sr!?XRXk0Xa#0?vnP*QQKH9~6=Q23
zoV6~iq4vRwb)g2LZd^C2fvBIa8`D74>^Uaa9Gh#7%{7NyL2Ot;#cx;qrCj)=3%%pR
z85pgC#Vw3w=ig}<N~Z43m@a!9&Rfj>CA>NtijnBXkBoayl(2EU6X`#438-nW@7YAE
zWO>=%FH{(WWh4aV+C}mo=C!}Y1YH>)MqEq&<8XVoI|tZJ$paB1GYT>*vg2nOMQlm;
zrmZ%A{R}_<ao^0(2*jrwW6`<bNBVl2ljp(cfnZS+K1GCNfjhbxc^wr8{=O#X2J7p-
z020BX!->MwdX%9OIe3l;tfa|Vt(<p$OlO0UU73N~ZiXTt5j#sC`>Xw{-Tu{nRzzNw
z6YE@pvEE`%V1doC!C&w)YH?Ow<GCw&8{9bU1YTDM=HEhzyfsYgkQlPayj<W8%z;FJ
zibf!H2jv(BG6?S2Rx~L(h$s=2K^X}HzGi$@k<-qCLcpR)bgUC)mO43b7js7cSh1U^
z!-w3*<RrScs;Si0gePKKQ7I`^5dnL2E%+i@J}?W<@g31Z)?keOW~=%!>*I{2)CQQJ
z*5;}^(-YjqUYKjwX9ScqI<b;3rhkbepFre3p+ELXIY<~eK`oeBM>F!J`qpY)pNfVm
znb^GeEoLl^0`ay^%;B*3N@vZdPHaY#vu3vwSwYKbPbiiZ92U!pgkrfPv3zxd(Y1-w
zWlvll{gPuSH*(Owt6LVG;AZJay~~2(Q^r<gYj~{p$YE~RXRxWwD|;rVoK<u2#@VCG
zqw7=v(LJUOj}=ZB;7)ohXFzn)pL2$}Jsukt?eS-M-7Y>d>)hcO^S2oeYsth<%8gu@
zloxsoVZO&SK*8JeS5Q49mL}~#A3<^m7H_CHp%k4=Y4rM>a<^xBw3nptyz=P#!J5ws
z%8NGy?koccwKccg)EfTfv70zDv-4tVJ+~>i=5VskQI84n)&y=pmW)oU$7A$s8eUlj
zL_^O2*>cnaxzEs&^5XS#UW&&V6+F*(oOLE-0G|(IO0P0je9IPr)FY!=i=#6^WFWN|
zZhJ6y_GZU5=sg%GVW8ZwFNA=@zD5P~g%(9cn;lM@)-<(oUQEssj`1BL>vw0~<isx5
zm5UdLVr3`N;zsCS;_T-6M}3$m2M4=j7Uv9(j&Xr+=f%UYfe%L4go<CE)pK}kREIW0
z`N<p}>pD`y6uK8}gkEQy%j@))Lk#xcef#iUYfJXmXIAtM#fES$z`OX8*v$4voHb3N
z+TCC+BUzXpp9#8*9PEZ7Ic3q%OVkMBM6+I!GNWd9fjmUq`2`TFJ9yLEq1XUaE;Ye5
z@!_$~LDg=jIFMdLl3}sViv_Yl?oTO;E|j~+qa|4!imrqcWlpcgZ2VEurTW`KerQdv
zJZOzM1#%<=y>cwQa$3_XooR1PubxoONCKQjO#2t~Dj3WCt)}oWL$LHV>bGeW>wDl3
z6l+P6Vx5?{tkwJ{LBq5QN)h@XV0gW2=ZKJ%Lc&U;gPGuiSW&cthUJwSB1T+LR4kmt
ze^sFT!$iQykA4V>1qCaO4g*g~eH<h#&yX-tIA~a3nvjy8TG#BBnu>Lcg{?O#k|x?}
zXd3=_Ob1f=)#bsUPb&h~BL3f@Z<83@Bz>bhqHj>MZxXmXL*PX7qIvP@nxicz9wNFU
zaVUK@Cfc;{H%7rogL@o2MFDe8jx^>6Vl0f2rW6C!h`a{F(@J|!F#2Aw=BqZ%T#F-r
zn}V^SO~Kf}BQURv%VTA2kCaDW53V^Bj5I*Hvakuwogbo-Mm|P{Vx8x6Pu6z>cPy4j
zS`#Nn_*mvI;@E-2xhZ_+HUJUV!uioj^9{wA5BO=0VGzX|1CbsyNkAVWG#ipN?Gy9C
zJUNql1cTA`sNqVZLvkPrhF+WU;Hp7))O>ixLQ2b;??kf}%12-Em$SGkV@wBpB2MbS
zwbEHv)<JFOLv3jU#q&D-)nN34#0#nRmF`~##g)>yv7%F{!IlA_DcI7cTq>}Sxj?%B
zb!OmBd0J=2zts|7g0ZWRZZ<n>;!N`()s0%>6WTV0&}J=xGywsqd0BMQ5}L~l&RW9#
zRJvtsdNqoU_$F{fK-tqg9JS+K+FuQQY8!;51*0pVFXge(fHz@S8dagzqv*~-^sY<F
zqbrm8&+;fF{~I8;?CcA)CuA9k-e?IeEn5ON+gM)w&g`F-N7pCLH1^iS&w+Q`nBme`
z_}P=Rxbu=rqMe_l<nrPzv(BYI=fKZqo&$h;j1+*M6g!(E_&#aiyRij)t1~OQlp}EF
zxtB7MKcT!6f>G|7d^-tTu;9)?K5!QWeQr=5n<Q>PXMkQhSWh-!?Yv}o?7}&0g$#(6
zy+)C3E{SHnhA$Gaub1P8H0c;vx$}t}Y+D-m&9c&hWlr?-#A)bHLEi~w(cbgRqBBz{
z8CjO=6u%RQNTEt?1AS!Np?4Nr`0q!%*A`qv5W}Na5H>KV^1blWtuUsldiLew%5RrP
zYe6y2$4K-6K_rue0GDA9PZ|uFcG@MetS2QqfEz&iVc`bdcsLkcN2@cZ*V0xqSK^n@
zXG)A~%R|TO_H+!snq;K2PETXc*Dm5*Y6-IezO~RyHzr8rQ0{EB?I{M$Is*nhyqQ7o
z<G>r-G>F4E95}2Z!*^$17mNjA%bM1sIxrB2n1F>w-svi#HaNf?bAL{B%tDxB*ZV`!
zuFSuh^5WIAeh{+rudD+!qpO*J_k$i?ON1WxXH932HYHwcrq>wjm+r5DAyXfOCw?&i
z>i|PO7%Xn6=%(HbkCkP>+H*?=bBc{g#LqA{m-=!S24TD$fz?Xp&bGitpl9*KUy5Ec
ze^~pk46gY!DIJWu-=CnCKX<)<IJB)4W$}RMnEBA1lZQpS&M%L$nq&-^pVje4K&K>s
zYUa-)g3&KoD+K2qo(Iwd?&u)eDtA#EIhW$Z*gc3MIoH=UYS7@(rx|=O>xn~hMf5kq
z&648qT;;!Wg;`SUhD9IY8yPMO@gKLbIYZsfyTHS&1^jf@0%5h|&Tix*cj+|FNS?cL
z+DY)q#LrFne8Yc(?sr5k!9RWTHSNS;^dcwvlSXqJU^;xF#|4>`(w(arO23lQ{Pe>b
zt$HF2p3(2%FP5!~y%NW-T@-XLQrnh7al07*&+Y{rdPIJ%V<mZR*3#*_La}j;&G|!a
zX(KC#Moom=xoX|?&)8WDME(Jg%LTD$apJG+66t846B~g%zedEhsli#}mF}<02~-`T
znNDOSg8kcxPUX=zhDWnLbRq)WT$o+Cn+3r<h=ySHr6G};uM!1m{T&_)o!kbOhO350
zFZ__q!Qz@(XM<fQ4|iv-;!Is=Ri4h6WxZHCZ^pS<znb>8GGr#+N{)Z9ctypr<!<hV
z@@VcVO&Zh^cZy$Am&tfrOZ+R3U4cBFKpr3SU8d1jRQ;hT`o+NQC1j%+tIDIBgT-IX
zS!XWuFxi{og={%A+L<{Sr+*L7b7TnW)q&F4pP)nMwN?MhI?%t)%ZHOof@yKJ`j^#V
zxI1dO&I~Tk6X>&+NIK4%ul_Mr)BD#j+tr9JvGl84^M`2(7;<R;L5f~wvMveSKGw9q
z^YZd&17M;3BMHEero*h$5L7t5m0C75_b13%NM=^f><^$wjYhs320|=HX4xAm+H2Pd
zY+xPiG;>1hl|e{5&1)0iOP}BHhvVCnyInU7k9OU_s!mISb6{e}m%C#&z{p2!u;cl=
z+8X-%>uGN*Lt1}(t=(UfQ*q|emIL^`zVP^<JM#!a=Y<>0y1<D3#q<)$3rO=9rOE$k
zkogWt%A=jxV6|fYX2g8S1Nhf7=gr-2A$LwrsQ9b!SrCiF3C-|f|AzgMuhRBM&iaM+
zM}qFD+JEVYCWTXRrPstyAlSU69sYwp|NG^yy#A==KSS_8Egy%rvFYf?=CG056h1j^
zeudnNy%}Uhjf%R9^UG}W!1(>UPX2hE^|rxabZV!l$g-Xey*<YE>Eh-(lk@H0@6Vx$
zNA1r)2>vJivFFQ>oNK)P6*KJVNilqPc%C_ciMMT>SX>rz*#Ix9yCinubIijG?`cYu
z9~<{H^)OVl9qgEWtkEBT;(c)gku`_gr1s**lNxn3!pD;iLIlsh^UO<466uovcp3Xn
zmt2cZEXBX1FEa`;;GFH9N+ALpAMLJ;=c!9AhQGv&Kd0qJ*lbgjGTXa6O5N@=1(3&(
z`%-sNNp?@Giu97)@LR^vgkmqrD2k+d5EZEENgBs}$>ZB`7w^%-UR(3d&zpx5P}vl*
zh}j}Ax7FON0^Mh-$wH+KJWQqRfizb-j5IhoeM-gBkj|8&eM4_K(35&n_C~~Ozc(4b
zNb!I1<NFa$9SaCkM>@leg6^~&?<X`)n;N@-rBfeKy#Ba_X@-SJ7Xh+<;0!->gy<D9
zfyzG-qA}4Fdy3rb9Ou*S@OTx?F0xK|CA{AfllVybBh|8iukz?8M$fxiY|K3)h^<R7
zHrT)K={hC1_vIwZ>)xWg!KOTahz#7Xr<}MJk3I-Z;0gu4$&cSc?FhXyPaZgXdXaHB
zqv`8~#?!G_pFIn;wZf87E;t>7HAyJ4x20n0_#yjUe)ck(y`#<kbb9s<WRL$=$<Y9J
zMapT$PTjY_POfB~*V!vzr+}%lkE^4KHpKsi_4%}p$cj(PPfC^(9nQ&o`4<H@KbMH)
zUTnuAO6R2@e)?ycPwj%ytC3mF;a}2^Ftd<D>;0u$cfW&vSR67nKYHyi{&7J`;J)Qf
zpd7y_p}kO~6%}z^m~?Ev5d|~bVZ+h=3QexI9K`U-LAd73&`UeNUK`s*7^b=M6{rKs
zD8M4eOCwToTWp*H{J4@-+%g-d06(s0D(+z$rvN{$AQk7@I0g7|d8xR`HckP4Tt+JH
zQX8iLKW>lwzL-9oZ{rl;$8Al;oowS2;K!{_#WhVZeOG`VSDlLchmBK!ANN!$uHMEe
zz>j+%75A8pQ-B}0C>6KR#woy$o12Q8YU32($4yDaU2fwP;Kz+k#r3ms3h?8cRNNMj
z91_zr6|l+%<fQ^$u>tb%YC+kx(q!3_3V2U5q&}hm{n;yAFoyH}bByqg8Oz4%&#&?Q
zy%>CVyE)PG#lFkAS8YlQO7%FYMME_IT3jtQ^)%6|=Bvdn?ZY>=Qg&v<J1c#1mZWAw
zQXe$x(&&2hFplBV2;#<ROM4b2%)KEPG7S+AK-QOOQ>P(4>xYz>i|%=M6o9SHkSQd{
zcumyU>_3{hSsQfEhX`mx<csf_=7plCi7+$&nZLa=GD;TQFxn?hSK$MPq&vLp$(v}a
z4cOSNQ@nnhsZ@s-f4^n_rjpyKhq@>l|NNk-@%K^BvPdxkF#v(oh%G}=de5CR8h3_e
z8(Po<XI{G!!!%|%A+!Q$ck`SmOOoob<f+)UjQe+zgTXBq$R<;ByzUH^<f`~<3`R?a
zYWI)iZJ>MfEtYmk(#4)ON-4*^$V-r?*NUp0p}?Sny;f+8@9vtKZ2@gOP<AhSQyE9i
zQEPkGfSXV_G;?s8$LNptXuB$1u=qeQFn|H8x2M3ffmu~`#pxxLJHuUI6-vXY2Uwr#
zBj!Q|pZz%<trr(~>sE1rzTT`ch6@b7iXL&O5PN!nX}Q1vFoxnPo=((~;XX2OWvv&o
z8M@dE=aE53CbrRBLA*U#?G0vEDIMl|U2L`wc~H;NyJU-iL8`53=?6_sCC{NA3M*ub
zw~6e%FbM23*;LvkPY=TLm`cXxMLDtZ5Vx%8s5YJ`4EY|>UJu4bK-eV=TCQ9dTy1Ae
z8={l=?mGyW=r6zV<aAgUGkZr$!5IXJ%KZj;rgSfNf6k&{P+q8LD=zG_TkJ=$#k?<Y
z`-#FHc4x3@<hJCQJz&&jJ$f4cb_nzjqBwC_rxdTO=piQ4!COgt7?4wqUF2?7^wD<#
z-Nck&cU64B512`L8oW~lujsOJZi4_^-Es5g-TNVHrdYT~_Yej<oYJ2u|AB<b$-Dz>
z4=na!%dEp0DD3=-_gjMZ7I4LYH}97k(=JcnV!<nZniy&;{_iVI#b^R3a(c2z3d?O(
zSdV03%T$WnWrAPW1j1pWm;72Dg1n!Agsn#|+|io@b|hfn%QgCRhxW@qzH#a^UYKLM
zh{aCg3_m+P_Wk~x@0BQf&(`HyjKGrNfES<<-btjg7?EvFrof3afU6ShqFw3i-O9t_
z@geto(cQu2?raEflSIXf@g1Gxoeg9}aF$U-uzs*1d%@W39B<z#T5SKwkT;dR7F$d`
zf3GpLgzBHIC#?J(ip{}u&ve<E=qj`Hb<l5r+Cq4?>mZ4stc7|f#m`<;v`95}=f%pG
z;iau)%dOtAL?$*T1;VuZN6=5nI0peO<o?WKfAY-y0drqIa~pV=5k7f9;Q7kaw(>g6
zefhLKJhfUIAAb()PYVfDYX6{g?wr0E;r19b4Y?x0Kr|V62i^9(zFI)urV%~a_ZzUl
zmN;N}<!P!VQ}wne?DAw`3(5-Hg}V(3JU{1(0rh34<)1<FkbAvH1J730jEbp=zN+XI
zMyF*(*mF^r>C(}277ueDI@|ULyNiz~taL8E$TCD!c6gIafmwyTC!0=tO{JNG7rcD-
zR??%Bmj{ma;y**(-YC1acz0caVc3G1h3yR$HcapnFQCFOPMUXECOZ}H1l{N7N>CvE
z1k$6wb6m%ZPC6E2pBd_0KhwFu^OFkORs3+6`)ohk&i&z%(z&x~;&(LMD?ko5&Uj60
zp~;y!V8P@4)WVd1oRo&eJB_v=&t7KT*%kPMFBxEiw}cW=U(0jXM4H}fKORw_MkNG8
zxiEF^=N0~hB!(~j&Xg~MwSLAA%1Jg=2Z&k$2yfWU)fFNR*xmIRbE(Kvmtj#jO)UFK
z)(tO=rGUgQE!{cia@D-d`}i`$m!429u|B^hnXxFsou6)Cp1;vjG+m0l#d{QtHuS8-
z)Y4^J*wV9Ygu}VE>(kV|b)>0#?E+KxNY(ul>K1(h7+xuL8)AZG3q%H?A)UKuzI&@i
zCuvENySz^(4Z=f2Yy6w_y^;ZfFG`!s-<%NG5Bb1UHB?j`DqbC^+zN6Bt=Lu5*w%R=
zsi0@48^<w?Z7kvh2Cilwt&Q{QW%e&SSiF~G1*`A~hgJV747fua50DLa;K_J-)e2cQ
z7s%)`Uq+XC;qS6(pc6F!{lbz^^d0YPx)6-Mn`qOkI_S=Vx#{r6*TJH7SSzDI#k9c}
z^RgN+>uSg|d9^3Y22z7lv@u+UJwssrxwPKfVnz#Ny9(5(H^RFq0K*Z~71X*_>WW~|
zm*MWlV<7JI-|&tD&gimrY28pA4rq(?h_U4OJf@I_a12J9(0tH;0-L#+J44YT9nwJa
z)f6sKyUdr?bV+>=E^Tt89V%yxoxEVQ91S=3;nw3jhd<QH`N(o;$Qmow>qATPVilCk
zPu{$MEdYn}6fg+iPIArTAD>Ys_8!T2pyC}X;;#@=$}Yx<qP@96su@&|SvBgf6@pT;
z3=Viq8OzAOi~@P~UaHR5d*9_lz0C3EP?C3u0TIf#F_UhK_fzw%_qKm7{a*`wLf~1+
z#hwOn_CTw(E64^bTBh~o(a&UV?zoegh(pjrp%8D(;&yf?+PiS-SZArlLyXSd6C6&o
z=;#Ozii+Z);%$N3a3+xvD*hOYRb0To#TH>vuy{FWw#UhggJyfAO{m-kKtXk5j;{F`
za?!rKC&Xnnt1HHeUf0zZ9EvHA?kF!hFtyAa$-qRUh-r$TV{B-14V}Y7X6AUBy+3?>
zsEGR$iQVVz6(HQ+X!H+YAy=3a@1l-!E=AY?lzVgfif!ptNWfD3?q>3^QRcl!Hmq}P
zYj9D9wLE^oL(yBo8V>8|DpEKD$BIbr#gv$K>kvSB=7D6Ys)tAwpT|$H>hfrPuxKNC
zVMpad2f`;XYTQtya}4FCjRcx1XfExPX>cFH)m^T=Q_<>1u;@S-{Meyv-yxbPE{D+4
zXSa8|K>^Vi*$M**dKt`I9*RyC2BjO&(CFrNhH>vs&p1)>Y0G5s=QI;z{Ws47*@{OZ
z)FkoS>Q9p5OiJPPbJB=x9TrYn;e--V!C(DP^n;c^poK;}X{rc9?6Ifq8Dw5&drK)q
z^Js7Zb8ldQu1t&#Lr+^1?#n7JQJk)S5tCj(df^C`fPZJ-K8R{N7#m`AUt7H!h)iG~
z1XXH}{!@xEhp`OCreO{-7=Y>C@SkAEeQCb<(`g}DT)QPk;_h$=!yw<o?STZx!VYkq
zQr1SzcTMWS-4WPwKq|^Uj;6L_sQQN7@g-{fu)c&0?ModFbvPz{U(!1yM-#at{JnBC
z?;Wf@?vwBdE%b5E1J1-pk`uv(ChuWB(^xKxuGD!3V=r&KlR=m+<}ljZ6wg$iJ0~xr
z;sT)j+NlJ1U4ASoUC+6gc1^(pTY}$1hyi2q{9JTo+6hiu?mXM?Wq?KSV@#02I3I0P
zm-p~6<*sFYcjItVPHE&qF2stRe#XU_85KV;3$533RL?24h;GKjg`;Xtv5+#BefT2r
z$kQ*qjYnX}R@mshH(EA|r4}6d<9S3x_i@_+7Yf2ayHL*Zfq9wnFm#60H~Oh#eQw}1
zDGdKAWr$@z!sCTq)J&djjI{SA4Q!UcMeodoLdb^@v9nn+Yt@y`8ylSB;bSW#UEnJY
z>B7nlU(#cUC@ssU$;QT8Cx&vo`-t$@mf+47Zzc^j!omrrg<YPVWQB!i3HMhL^k0Vu
zT5$h5qP^84T66yoX}zYA=)2wjYDKSg`vsn9Wk<^403kZdG)Wd=gRn)|w7vd7bN_0$
zzp;NWzGNTZfZx;}<|>MvPV`{>T;wmqmUVwm7R*v2>gbF>)d(Wuukz#6>}uyUvEaq#
zXZs$ss3SSIlkH?4tUEZl<AEFGCvkF@Cj0Euev<j_AJFrb$=N5`XUA?Pe0McrT~4!M
zr<~0*7+qbrZTEY1n|FUyw`TXw+S2dt{;0OJlioV)-d|gqxjVJrJ?_C-85Dr0$8ir<
zWb9tS+v;%--k7m_EpMB~J^1qs%K3Wm)^QJBl|gA=5BA1Acv;5ooxJTC_ou@>7Ikeu
z?!oem-TV1CXHj9mr2W(3a~JhE(M0#|HtxYe8M~WI`1Zp^i^>8WcyAcD{qR|fLML_@
zm)W2C`8s>isPh%Gn)e=yW?oE8AKEi++^p~YI_s1yP|#!>H|zVq&OKe9wr%6K6W{fm
zRJ>^>{wWLAQ7BP`eN5WDUly!;Z`|IU1?xT<_nV6{&TJU>n;&JIxn|touFW{}y>WlL
zI^)cb#_g>uShsUruRj#6+d8h--Dj<fkLwjXd)>Zqy{dYw<HS>a1}dsxq<T-FO6P`a
zYr3p^v$iIC-G;iI>u{Jkt9WH*-QIO8>kdv?vr8|vH3w(ERrm3_>bej3zkXe{($r+u
z9b8vKnuc9^QJQKjx58&8{mDA+e>skO%ZOm+p2(am^k0Ma1v9^7M;-lEO~tlxYo5rs
zRrd#hRu-qiQ_YN$5--C{s=9KRnb#XdIQP#tGx8I5YU6_I*Mj(+W{GU{PP146SLAr@
zLz))5z5C0Vm8dZboJc>Abql1|iQJmQ>pHP^rot3s--OjN$M(v_OoJ530peYO$_^xv
zW89sIu4(%FVB4!5Xz*B*<P(NXMEAsTd+Ek9u12C860Hydr<XQugn5iWX7h=xyRd=u
z!RWBjtW41}R906MN%>zWzyvBk6iU$RleEW#n0A{XqJvY3r2?=G4z?X!5{wQWo#-Ze
z1`g=D(@I!j^jfRzC(~zbMZK=N+`%Iv|A)P|fsd*>9{v+ZV3EKE1q=!r>>>%GCJIW5
zU={-0)eWLj2^IkZB33L28$b<^ki@cGSE<@ot!<_C*Q&MkrB($MldzirT6t6CMI;2l
zyDUfmfrPjGzjN;0ySoW$+o#Xdr~m(-581st=iE7ObLPy<nKNfT$SYj1MSL>~H186r
zggE8U4dqt=Hm7p=B|PW1{V}6O8pn$S*+)(6ymSzj8YdNyBt|^j^oVBDZlq$e=lxhQ
z*&C1ome<1+jM&T%&=7s<4YbOa{K?&sS2m4MuhiYAUb$3W`O<o2q`Z=Cyz)EimFJ9C
z9)C!^azt49WEul2?=JZhjr_B%{IiVwg+~4-_*xz^^6#aS^bIb9XG17+Z<NQK`oy@^
zpp`zwijN$6snCwHRk(e3+qC65kM15M|7FQA>~x|GRdH5iPLcL@?fpIFKlUZWI93ES
z9b^x6<ktR56EZLn{M|W+?2Bdmk@DT%LoWNMbvO9?=Wy*G;;`;cY?7ldBD@Vp&Obmt
z(WWZ(IvOl7)qQOdB9Ky0l3baAaBrl!wzx|p?|vx<#}4bCY9jt|sp1q{_+BU)jR6~8
zaWDZj>8OFOYo$7q${F65By+_VV&1cvr!~*W>939SnI&P2plG-EG7xc(2zs8Wf3P%j
z_7<rjh)kihkyu$00rAZT5fIZTAtE3CK=bY~_RLH7Z=otK77(mo-ij=!&h7ON;)|1e
z12}_MA^s<P`^OmHKNo)ve*b!w@qJebKkmcfD6{A}J{G*{LO=HpvukB_4X#Xt+pjf$
zMJjnU?}+u9e+29Ch}GKL>6DtqAHOPrR*H(jW!1RAajReRM)wrl|7nT{$_wh$fw~@=
zzo-s!;XK_P!eLN5efa=x6xr0%6G!w!PHL;;bP}!}iJ4J2p_kHZ?iMAZ7%SB2IuX+R
zcj?M5CudGu=<kD20pb*Q&C=&JQQ_T87`1I82X|1svyfKM<Ryl+{A;DjLp5mfR~!G@
zYB|RtuQL+c-9C&Vrq_wS@6KGMRh<+H=A<=kL!I|vgiSh47YW&w==abLq`Poi@%Quh
z{8hfc|7Q&OXGj@IpNVU}I>DeRg&f_P!`-Q~dn%a^o=l%*!kxv5TIsN&F$-m1Lbk-B
z6h%z-H#=}8FkOz&$uR;sc0OX1^qmkmZrCo$P}czi;N%TmkoQ@7KCZ8jY=)@nMR;Qh
zO^!+p=gV$OEdTcAHz>DkrLZ?re52;gZ9ric7lMQu+Gy(WXx<omwG2~#*SxDeJ@UQx
z@wSGTy1M9ZsQ!r}rsEF&2ALbgU7-9gAjz%_=SS$<<sX|&hs@5<xOLJy8~J{0dm1q|
zIg8uJ(em@!iyWNbtjTMi3cq%@Pj`9U$r@gT?WJUzbI5avAVHNGo18%;=Jt|?2zRLH
z_tc;rT0@)gi!?Zy%NIl&dTFF|l{zKb+s%Ot>9%wRu*Bt03TR%ZXn2F7*8Wwho#<3G
z;WbRq?N>8F=j91i*Xjp{%Pjv=9|uZ(qmVClf^CXz?$oamc1xxeyRj2kJ9?&mF-9h_
zTy^MMluN%9-6_UwR+Zy=M{uFKVQHSjbU9W;oQYUVbAO7C=u6lN;w;TGB9`Xb5yC)9
zq~jv~3HVXO_&fZ83{iiFr}$;sqaSZ&kA6t9ZlyNdHi7@hc=$tT0v{0zb+l>9KQSs_
zk8%6&w$+YO!xss5;`ZK+-Rtgj*|C(`iT_pkMyV#23E?Xz4NdzPQpLFnpTsMH>)Xp;
zR@o1DI^njH&iR9M=iv6md3fq@jAjpbJOJCX-g`gs-H@KWF=w^?*VPcyC(NFWMWLM4
z3%a@o-<yU#gXQcvo#@l#1-u1+ou9Mclg6`i@Cfld2!=by%m;Ln_l5iEPw<ZDq3t24
zT=Knnnj0U)uHoIKAI}w%x;H~>=7x@YGeV<;iI&BA%!%&80(v>9WR$xp7N3`pFh*lX
zLB9rY0S?Wz`XjqysYUrqklx)&pV1=SSQ{W7Heh6X09glCw^RRuDzG<#?s;5)8!|;T
zh4~iMDOMU|Lx2!Dpm#kQI-3=>o{Q0D;`o79mV+4^d6eX;qC+x;+7*PRfw5qfeiQBf
z=J8j9i*c91wYFNDlt>UG2*9Sx%4uCRl!Z9!X9kqD)4LWgEO*Wc`{I#oB<JUxVwqs^
z|9G^2z$LDnkI;M%xfAO#fMIUhp&52os1#RXhrPTjVhFxnYW>zf8OE&grEo=dfN8#8
zI3&wZs=cf&xxgBd$P@Y4Nm<FY%xzE+&R%Wvb50XOjM^b>DE<zqxiYbOVl-kGfw50<
z=-Zee99nQ_iPI4;c6hqvXA)55OCwB*{w}E)PnusF5~y!DihRBWUr76Gr`!j^5~_5;
zq)mkzPiGX@WFO-^-kC*7NBGD)lCC~;)#qdhzQcAV%PJQ?lwYixmU5E)0k<~`Hpt+t
zF8VsuUNr{e9+81&@*La5M(PwA4UknSt_v1REG2d>Ic?jJ=RHwYZ-*H+a0fs}aTP+)
z{>iAh(Lt7zsCl)|8DEvgSDiYZT8plPJ5^25XKswR=Vc#L^H=kxG>G0y-M`M~kh6CO
zp;kQItS%HO`YEQC430Z-&uD+~Y1lkmnVuvU9Q1QKT5`=cbGZ@^iZ)|sA`5ssG$=3U
zdoch#AjV!!F)AjGwJc)j6@i@1R&IWwiH#i1b@{LR(CYNq^)FDxN!zdGGw<tk;|ueX
z{4&SjP|;_uoOPb_O#bS|5d=pj{F#d=_W?{mU%E1Pditp0|GQLh1#`v44X^IR`*5SK
zZDdg9&*A=*E#WDMmsm-iw$Ny1D`ybex|$J&R5clc56lW~jICb~`q;!r-X=vKn-qO)
zqHfjD$A+Rh>Six*W<;&}Df+<B41IhFoHgtU4{O+8hsYXX_!p|}cJnqOT;01}KAa4T
z_aHr<%1FLpLSjOos<_L@p25b9<m-MYzp&L^=d2y+hMc&ue&HCG|KSoIQM$%pk}D8!
zWRV&<nT)f<9VOrYm*=bm@XZ&j4=(|P;`7<<5yj_^b5LI-R)BQg9Zp7<tv$ggc3l4&
zx$teIur+#E7teZYl;NKYa<l$4dpg1bUt`WI|6J35((PZwru@7NSyiWFu0u4sHQn@n
z{N%d0WtBAla83LHlJy6N3V~g%EYhsh<9OZkkd$vSc=0T}jXZ{ZN%*bCFLOP8s?~aW
z8HwSomi<?BW!d!TjV^2oC>_WltF<3%DmB=lH!@UVZk)MVYF-gjF8*}U^;R<!1m)2J
zLt2!oS8hGojL7Pi;}lO}%%DxI4dv=KTOtOa(`QNSp+Vt`qVrqx8CPvK*Kom-+g?6d
zy~oW^qCSn0KmrByc|r}4gh&7&6&2ycqTDp3L}OX!&b{X(8~?B#BcU2d^@-HT<$M1*
z;DRrMB~;(XIdUQ}`8ZRxzy@%ZoQ<6?SEiPb&F$Uo&O9P)lQ<D<6ItuT#rCqLq;rZ%
z;&1uN)peGJdPzRrUbdK2mpWm~b!~N8_G*DHPNF?5_!73tJ}RK%{p74g%-=w0AH@6(
zELs}z*H-hgz5FNGk`U>C&JW_>Ln6uonJ3pmKI<p}D82crd}po--9_eUOXv271FhsJ
zbkT+3-pnsb_%j)32fC*I<L{r~#8%SF@?Hs0mSlVR|L~&dO>@peFkU39Z;bq**j2*~
z{*?E~j$L9GCro40gJ)*u{UDW3_Ybo89Q%VYd`|emtip{8GX^GgS6}Y#iRyc7_crQ#
zLU$bUt8;~4u+`ZZ=kM)v2RX`_^03xi$vt{S4Nm_k!gHOmFK%W97W+zBV`uwM)>7r=
z>lV;Qzpt{Fy(`^B+91t0Uo1QNTb~%WYDQoYqSA+cN<!imX}%1eY$z7UaA(N0d2JAc
zfWOQbq0Mcg=X)#l{m?$SW-*^Pu&0I1U1WC#;0;HKU_i`I3lE8*Uv43jOsEBpJRbr+
zu_V5*FRd2Mb2nkcc7~p#A<_Kg1@$ygv)eYMf&5wraU4&V)u++8itpU`@sKOvWuG%k
zn(-8sKyI+*4=zge+>-B~hB$huw2L5Tn1E$|A=cN)N<2GVxP~}Uq2%SsB*%4(#-Q$u
zM%)LXRMJh>#G*m=H#?%^UB8r74)s`U<zw{8^s|wV0_9G*w2)5marA-=sg|A-`c*k1
zmnfA`4(fImjZg-VbGzx9?_Ea;uSz-&^3ATyTA9Qg)m)@DVFr-~QHvG)Z4MdAzeR|L
z_OeREIem#>mf~T!Ezc8WZ&xYUZ`$Dao55cetK2*AY^X!cYolbB?7qRJMe%=%ndMGo
zw{&8bnD!?_!MRzo=#I>lu2n43OBY{O#Do)#S@c+CH@t-E2*J(0MtBGdfCLm`6SC|i
z9SUcgjbg_DNT=u0>FX#+@_<2wBNMX#=FiLI=w=b_gKkHLp<DN`__Oq<5U3o&uDjOg
zQHqdF0vS3(&3;$0R5i5vV8T_g$r`gM!cp8qu4Y^TvXtpi%Q?uZ;t*n|iV^>H^XC}V
z0n*jwvriD3@-#(;rt{>6g$k!3)XR<j>;4d7E1LIRM<rl*%em4A72AbzN=7kFVJu^7
zcbk6gU@=XZk=<Sbr37_j4wb8S8J`y-3`t<Z`;ZVH7T6M=g-KQX@MN>$GAP~Mcc8%v
z{R4Zl`(Si2=Up?nst9AFm&582cR`ugFT(T;|BqW`jqkSW7JM6l?*blD+iRtPEtF#|
z<dIFI$^nsL+C_?iTu`_u#L=RVyhtJ2!A>E?W+BDE5qaj{%}xq6e~+5$Ht+(w0hs<W
z&z3>`CZj-_GK-xm54$m{Z4~#nH{#p$2~$9|X@*f;C$qRt6c-*9o+2~jo-pLPij`cV
zxAq9%E-=h_tHS79n1_WfA%)?34+buLUARm5a!I^eKCX}tgk>`ulS1vjChG^Nzc^GT
zhnYM`JID2tsxfd)og0)WaHqbUx>@vlD>@;<fGJx1h1U(E0iwlDIdk$XkXRSl7<od)
zS7B4^dKc`>CQC3U(U;d*>|nLM{1?nW(c4>Y0xAa?=i5qZvk$Mt0+wKT$RxUQj-t7J
z<w~qXbNkBJGnE-5^F#to3d{6hCcs0YE}7+*>Gp}<Rch|tCvGurufrknHpHV-OGM*7
zh&_^(S)WX8gtCzikGgZfVlV$l`b4%YWR<an&s{3agb0BmbJ@$)(rp+zkyAPg_vY*(
z5t6*9hdxv~yiHofa}uK#766jPWq1Vg{-v^MO)=Q0My4hZuOQt!B3&#6BQwZ20+ST1
z&pU$c>zG|`#bA+tQ%B0pk|S~>(jUIv^AvrG0GM+M6DHj)u&=LXvRuy(sXPbD?B$y%
zQX&S1d&SBFXc($}Dum=4(t4|X_I%EGp*Ba_*G@IByDT5k^)T8#_I6eKuEo{HY~Ltp
z->69YMn&2;Dz1H_*u@`ZwQp3UeFh1t_GQo(ao3a~h_tV3q<zvPV{)(<s_acy3DIIi
z$0CRV&jMfs4G`)`T#a!PUfg>7#XlsA^HKVTSwc6pdLG{*K6lo9xQ}sF@)U?OtSoL%
zgWsgDReDdglI+wy>bq7ywJzQsqwII4domRglZ)Fw6N<ai%|!Q21LIU#@pV_(srb-x
zw<wI5!dW&9Tlg1Z3qNJp!tJc~hAmueCKP2$n`Nbn8p9Q7iTlRFENF_Q6|b;>+I7aO
zBZI>R)~ekjZq=SMLEj8IBpoqHqi3E%Abu!=zbs_YKT-l#*SWY-Ps?&;el0#RtPk~i
z8)jU7B9@MaSc^?U9iB^1N6)xHn5*;^R~TFxj}x`B-9TBwn6Bb|6(8yd$D2+5bA@V_
z7%e0Grzena<3GyvTm0qc+E%o`$6l-d>2lQ+mydwP9ncDp)H<1x8mEvx4xfX>jD&W?
z>HXL(v?3u&Zt&slCT`>Jt2x7z{4wzSEHSJ<<T7HKFwtdy0W-D;eMj6m_4W#pQQbLB
zMHjmU-<Mj{L-tOEG|^}F>A#1d1Z6#WaIRyOJRpb4K<Zs>R;=iCL8fqd-VTp6?I9L^
zg4Te$qpNH|D!RrqF?cR)foE+L&xh~JDC@~W*0nsz1RsMSHiiaWgIf>%`jMO}okdN%
z08R@thYTyYnpODbKvQEN-Oh1*E26rYOj~@IIMrUI_ooE*(8LHX964^DDTC$q4`U}t
zjK0mu`Hz)f1(u~+sCHzA=2d&K)buvkdzJnwjd&r^1NW$dVAQ5kwJzRja!OuwwhpT!
z<69?#f<o}Aj4wZqA9500%{zse(3#O~5?3P*t}%RV)lY9)*y*xdl5hr{h`gp8X4O%?
zZ*fS{ygfGvQtyB!S2G9x#9(Jwa>aE=SV_}mionJGz<zFmzE#Eo!~ZTJZr0x%;aB9V
z`hfU&7QZlfg;sVZ(O$k%bt?2Jn#OhZibMP!?eDKlm13XntELo!ycyW?+Wuf<KY%%&
zN><F~Y_kf%;HHUdW@Su#Gt~#g!A+uf@Q{NVIX%B#%(nM&x`i{^K&ssx4l75jbF(+P
zjvtfE^j}dRoH}i3_C{wE@;YI5G?Wx?Hmvtkt!!w5=M`8+ZTJu3Tt;oORBc++tYVrh
zQ*2H6hi4fklxvx@_|7q=gDoL%m0sG<=%*L?Hdfo=8Tunv3UHPFr0Ue2`hN9Yt9#ZO
z_XQ)*2tPrjHx6Sa>)WJr<h0KQ9+rxG|ETCD%MrKlO_5>UzLheWWviuo_{x~|4tYR1
zohr{e>pbYgpbCGnD&{CEbw94p1Wd~(vl}v57L!7(NEq;MEikP}kI|CIGSB7=Y5uYj
ziZyy6+o)&^IfX6)6u<sTF*cy4{pdHX>_=g!FUEeP<h5JJh@Y~oMeKvH!aX!y8O-$S
z$4bw#RQPcB%jDu?PCgR_!*dazZR{0ZsT!G`zPah#F|8%(zZnm5=1~Nu-v@w&L(WY=
z(FbSnt8C$zgrdvfcAX@j{mnTA`lon`IlVJ+EXALrdH7_}qSd5f$|(QxGb1C`y;Bb5
z*3bcaJMLzK?!@OnoYz&D(=Ar6%|@=>7#PWwAi20jP&^sjkdl6FrnSEnl~>1@tFiBT
z*<&CWjXU$SmUDE@6hgvbwAToSiFpVk(40FhZ`(!0z|*p;+&Llp;<xDycMit=koefT
zydTN|AiGv)xY*W-{xlRTj{Z{y*$s8$XoqWphD;LfKX)chpzKe0?q<2Ph3{0>9<f(0
z$dE-;0<7D@H%g$;{_GQo$f_dN!dUeml0!moV0e<l=4kt8^q<g3u?5fy{EulYt_1ms
zIJnNiDb|SAZt^5Ej2qfa@$E=EpZA5Hti48A&sxuym%yjy%Q7Bf?T2Oir;7(h8cf2b
zy=S?*S5Rx#JnU33JGwSiIlgBz5t5r-+~BgD*$zUFV{$la+R_o;t_~-bMQ8y5Jf9U9
z(R(uS0|T)9$uQxOw&rT)gZ*LZt5%lG(@0Ku1EgH<N+iZ#sBG8#8+<8=j0x&N$enmF
z&wtAqG-B^2r#%nHQ^Y;@;K^x4KUaHJNPkEz0qt!<M9RNK?1r3@NQC9t_~=Q^)1^@v
z%qMh=w&)l~<L(4Nk>*Z-i1R-F+lDQ$!LkM7l7d6CKj{<51Acp5ZXVjm!wj0!i_3fv
zh2(&Q-W%y1do+QFGe?s~dy9{9<@SuDY~Ly`k*tLXb~3IaRlUsg9qpfbl)afi8e=S`
zqEus`y9g!VuMV-$PkG3G1hT;%5TqFxZlfcDDl=IE31!HNp3XST(<0*#`ju?vFi2u2
zN>zrb%&%oHjfU?**laU<kdrIwFx;8<Q#<|hpss^#{OEULUSb16nOxM7I#ws#CG4*R
zu>{WEj3AZn8ayI{Geo?f?_ZD}&g8r}rw%ZVdk;eF=^*gkN=WsC$k*oyHuDwF{k{zM
zlv%ib6(A`$>*Hi-ru>~`+L(f~vS&5hw8uOI^XNL4My=;pv!_Z+JdT!Hv>d|<>3#uw
zFVbo;uuzU?eUjI+I>|S7eG;3<OO#)$#4_x^J~SOVkXA6?{21cG*TwLAc8)WgB8QX(
zsOX)<f-?I_^|+IGDoLE0QSaHycfxZtO!#*(@JZOBX4D`%F)%p6u^k7M=ZZZF`2`sP
zCysQn7BFHS2fk){p<3R=Ace$F9yjK5xMqLl$RW|B^AnKyMS0_&HD_V7=1l^#`v_wY
zt13lmf+Sb0O|TYa&eE&J&?MeXQYOlLvc8@-5JT<dA{fdGX&kHKB9J>%?G;J%iRR06
zW$uqo6*W!Hp^e7mXwzb1^kibll;YpTTFwVdjIH8<9JylPMaYQx@M{?jy@;Y&C!}|W
zdz0RR^2PoHIm5K-^|zpGCdpeYwn@?l^x{GN64gGbq&LUyy&n(qq~D7BMMy#=SoRAw
zEO^gIf6In}F9Ci8?Wfrvb;@oQt>jJ5M~J((GnhX@9Qs0-QM%k-l9xcJx~tJPk~Ht_
zHdk{%n?hup5AFVT>ebF8?Qag~?JTR|c874$Cqx~SGGHg%=MiBi#)Tp@(d`Z6_d^cG
z`efw30H`-ji`))yu^dRq*)XrI#<7pYE!HA)F<quhv+t(6)MkJ_(xJL*gm=Ji88ER;
zz$BF3#pd-s{U@+-m%rz4rRK8Tq=b*d-xGSE<!<lyr8yW?UU7SWq*#UYi<l|XL+3ME
zD0#7%7yTjpI-r=bu|GDM4p1G<IvsPg6$kAhO{pv-KkqHPh%leZ0tJ1ogT7`-4G?e(
zs9uDBnD(V~nXt+FXjMnN-T8%a>3V*~DtGEUTg(OEgSb7%*r@k@QU*@tQ|eTLj5ei)
zgW-w=V#&q(V9F;Fcax(g8&chErxNoR19XOHc=U&Z<9~+2-Akq*Li=7O;I2!hx_Ppo
zNpM_@yRwFVqsrap66s$t`iMV}%pSwIiyZ55r0KW9&qT|(U7M0icb}*EZWyHbZpk&m
zWVF>*#(0$-)^cj*W)qXnLH&~89VG-8%BxCOe!Nhs6hJAMEmrY>iFT)+B&>$1%uy&Y
z4<}1y;{A}6H8Cbjy&9sO`da2u4FB*Uc+zk97GzdS<Tv{xzhR;VPb_rmn*V;Cintnl
zH=klnY}9-=6V8=5g=mQE+jO~ECs;W1?B+%Z-%UiYs;0JaC#pA~zGT}UDHZnOzd?nQ
zqjSz-)jlIyNPp#zu!i$+uAAVOqkUsnr*c(IB@wFTr*<W>_53bfD_!2zY|bngMwH$J
z;yQLEt6Sp1^)!O5uD3}zs|j1sTCdnP?=?O*75(MGmx;T&ezd>mmJz;lM+`o1(bM5)
zvez7|PAuZw^=!h-eRVxMBv%n;?u2Y=^_I|$9F>!hkpo;Dyl!CC$y-koQ@hgEE?w8-
zR=n`8VAH~C(FNpRl>buseBaHyK;^}X*F|1fjk+qT+uSSYLl|9yYlBXF*jz#nuU8=e
z*GqlZ&a2{cLs7>EJ2-Q?EP4ak0GF{8SFmYw(rU)uy_&I?u#n(R;h~XlLgX-7WmUQv
zPbDhPay~IlLee2OC^y=OK;|BNcd93owG$m48j?DWP+Rh%TwjFGk9ZlRA&T2V%CpKe
zZzv%ta@tXGeo<ayTy5ERvaU+1#o=X8tXTASxIeO3BdjbWtV}!}4+!TMiAbM!vCOnp
zw0Aq|;Uc&HI?UeqEqdLn%=*!ZNvo7R#BwY0&`$G$hG!vkRbN>akR0@m;Bt8{lKLb2
z+(!XPt7OMN<`72Jg9}IL5?1u42F#JC=ml2fQIIk{kO|^OQ*NokD}z?`C@x^At0Z7w
zuAs|xi~o@~Z<P?l`7*qy^$pCG#JQJaOra&4wGD*1OPY)O&?+^&tBm2TR>OPFgSH&|
zqBq49E!bpqac%G7Ud8AtMzbnKRwWwWl_}i9!S`1k;jY*AJiGoPgRQjmh_P$s1*&7!
zwKg5FhhyyO$qFRK5c1+b-wta>qbTneg~X^JNym?38i}SLx066bAtO}6Xf_$Y@Jthn
zRXCOTYTCQ7m!BwNi*IXDkNm;YQj0h~HCKj1R`2D##M0%NN|XaxjK(W%l38q2sOUXq
zoDu|{_rur4;=5AtwMSu0P4HZ5zKv-MT{Qaky6D>**xM45NLifYoamkNSYl0omgxLI
z-)q%hkM{V#pufC~?tiwwMj^Mg>M!9Z%5ZcVdWh)C_~%>Bc!YXtm~&r@*dn37MZC)V
zP_c=Fn2np1wv0Evh?w8yjX%VacW%2-CUq-M{EN}P{stluYT3aY!|1_3-VmF2_;%%o
zU;LhUP^dgg3nf=Saw$%Nei`e3f6GNNN^_8jgVHQZZ7TbluL??}$Uj5x_^P*>mDk_M
ztX!5B=P&f#_R9uZh5fRBoc(gU^lrTU@=|P>5&LC~y|Pcdy>eXCuo=x%wB7YBR?I0l
zZDQr?(E3>AcUv*9K3}HFPW@MOr`{cs6|37tOmzXyRo}JxnKxsNyTU*JC41&L%0N4?
z?3s5<*)6Pw|C&8>%1G&f|I_x&yP=fU_RQD1N$1t-Tlu!^neQ4LTn|7i(z88PPuJ?t
zvwto98N<&w$)Ncr(Vi`YAkdaa)@ock?d7+zR*O@rn913{`u2bXgu96eNPrED#rrGc
z7=vuL=4aNJrV!kD4=Ce?wUiT;Sbvex$;V~IJi@??oEsx#5Hl(t$-^CB^)xm#2?1X_
zLfNHpkIl!Q=o-G=UXJjif3d{=Gw1&$ROY{Q{-6EMqD=%279pZmujwuXhDWK_aPb#G
z0{t>zy7$L;hKk119<^HkeOFnNW8*EFTWi)$owDF&9=EpE*uTUfMoz)OOpV**Bv_v3
zTKqGv*QZJi*o^R})X8R_aL%d_8g?_PC~m(Sn>EZ}T!u5}eC0-dJp3aGw;-mU>|>#F
z;UV7jE)M3nGpla!52ys^!*Y?#(m-kP?MaWyr~6SNp%&jh_R)0tn($~=U|De&xj&{D
zK{H*x)crBVUAR9+E{Vy<MHowGwI>tbV@tnAA7~#uaeHO(Bmq3jYPqkn24~haVkVxq
zTN9hT|Dm+VCL<1h!!v|mL&>r!JxLzT<y!5IaDZtjeL3ztneR31@uCP8p`l8I%c4uf
zw%t?MO<$QQ3g`)kJ1VC-6$A5gy%Q;t?I<%$(NE0$P@W9eOXMswDt=j`ecrj6>>@wR
zC}>z-LmZRC+Rkx-HB>FwG_JCxKYQ?GL89T!eyHdO(q1DO7>D`x#fzmeRr}g9FlOH~
zc(sB01wB^N^AUZQOL|hfu9MFR>*(YKsZ-bG%GZo_lY<8)ZV2v~xHY(A;;!H)6GMQz
zsH|O#tX+((UC0_4kJn^8;^N;1Po{PV9+|kB@!$+KdM^%cVQqPnGQ)@y)@pm%AE~MA
z$>E{y_Rm4Ry^+6LGfYj?owIMw1WqHQYngwC9mzO-s>XxJrc)$idW+DeJ%k&Jw|m4x
zXbIV|W2Z4=aIF}g4hN9$3>h+dpA&O}lpt7Wti#i_<vj4O;d;%hnRfq(-?hxHrxOg<
zn_*pKKR7y`5`RSv6;^+m(@e_=!7>{oP9e04QY@xu@#gSc5~zBpOFN856g?Z8G8Wug
ziLhp`*iE^ic1m?twwrX$v1xBdI43bO?Byq*2se91a%Ky|jq*w1F4k4t#fs%%7qc?v
z8cl+}?ZR|};aM((WeiX)JZY!_d!&W>W9PEJb21g6Y5Bznu_HmU?O^|xlYSE%>E?B;
zpUb(z_n~GZKl`T`MYgN!QT2={8nNeS-hS9|Y*>Z&P}4%D3yqF6u{=X4`h8>go9*F#
zE%a6C8W=F>1&8(4VX8|~LHqoiJw^D0U6r5tb!^9AAKyh5InN+eB-WGD5u+AlpCNNC
zr%6m%AIP0#_F+4zz0n``@(SKWwA;Z~g8mT0YV}EA*+X48;7H%8ZcqpN>h-r_twxs7
z{uvy9q_Uz&VDN$;L|>2T3&J(+i1tOa`H}Wjq|+hc)a(P%0vA0NUZ%0rSr&WEmC#=%
zPU3p7v&m0Hev$Bhx!k*)rVSpJj#W-P-;Zrge(mzagzeWk^zr1~&W^{{<i5NrkWm6b
zasB+z*5t>?mAz)`n(d*jUv016vF-C+J23CspI~=w*Y(=fV(RrfiG+e!wpMMjc2=Ta
zxFdn*We-92t|fesVTO0xhc^nFqI)`xZ%#;9N*}wqlT9d(MdP?G_s13<`qIWBaC{m$
zIDIm8tRp>8A6hOf!T^w!=dq5g9at~7bPDxPOxQ8Ok)dA5a5b+hdMS8vB8SkgNIph`
zggGdN0fXQS=W5w_8l)H>&)&@~{GB9VLB+9!=c5sWAgN*QS7_x8HtqOkEpc;rmzF78
z^vtY4u5q%6LX($LO7PIQHTq2YE&Q{naYgh^-%nIXtsxR&%f7gQPgv~u^Rp}EE4dSO
zmxoR%=0mRwQC9@grwtG}A2V}Ct&LAlZ3dskwOZkvPcqt8x8=v!+UbRJ{+`kH?`;d`
z1T)%h=IcO4+nsHLKiIx2n7=&~9I^dS@V@O$g>&A^=u_8NIOlH}eRRJ5oYAL&um8#D
z)7TiO8Zzf&jx7$H^9i4q&)LjppE;i<28SFhubrMk!v-FFq;`osTz>G$+85-Z&%tMF
zSMyMKcymVI4+;-|l+kxn;o+JLA`Kk=Dx>f2;Cl!67j8J6Ia3e*?cmou^_zJp_~(PC
zc)D!n30@Br-h8_M%vHfNi~85{;i;9g9mTb|K{;!KJ6C0@Hlz+huvf@gows+sYI|CO
z3#?BJ>~88_lc<&MC9y>F{nEysfaZHFl`p4vy|c8A?2B9-#9ZaR`X|$gW#-*nwm0$k
z=cRiEi8~uJ-&+atQfa-Q@P1UfS6<ONhzUfy`mI^=%G$E}#N*pa>*W<KoAbv;Ii<ou
zq|`JJcsFxm$W~lmT5r6sF73Mdm-Bf)ra&!wGX+ZZmP?fb^$(0szM!8B`2*uS^)Mg0
zWt=pUdf9=r;CI0jGiR)8;B&yb!0wyvU)X{h7|<g-Y>ZuSs|?iEgU_#9&l13B*!r&J
z=SjjtbeMT6cw$lUIz6!4W#3~9GJVxcy1ev>@lu_7=_BK%uzG1fFU?rTqE?kee2%?Z
zY13>W+lslcAFZ}Zwf}pMNh3Ai`QFWu&ilydyvfjn>bl<5Mz={uF&b%jRXnDqfiDTI
z2-Xqm$Q1zDRo>No{#@|=L>wR6b6~++Rl4gDf-=Ba>?F!ut<ujWf#rAAS@nB-!t@7a
z!SKyziIR1Sd-n{#sW~>KgQ*}6?hb~FyWuz`E+E_&(@{B2;r-#{P&H0$siMPRUSjJ4
zH>Ut?b^FrQb`0BiAf8SJPSK)5aMYnFA|Fvn<y3+cVy}2iUNZcr;GSHB#>z&$c8dSh
zl#Yh~)UU~tkI$b-AB6@+?`pGxMXw1qssuLAo8jRRdlfYXQgpT8X3c|FCwK}XNP4wI
z@?i{;zZ?EdVn&C(vN?4^?I7uW=rk0b*AvC<sD*g|8v+*EN<45^8G&)9pCjTc<_CQu
zPuySt8?Dth{t;a{M!vy5b<A+qDv<o_q}0+>I>bL8!DBw*G_Z;Nm}OfoYG?QG5@ZUT
zJLw-}A2|o!8}5k<CW9$%k9HJj@iDuI%!x(kuz#53kXufTZ6n;Z_UXADq+YUvfsF(A
z1lhm9{Ucq;Zc#s`h|d6ZzM{Ppu~YvAZ|YrAR5kS;oebdp>KlKjKg8_EMO|yY+di_M
zcK;XpclBtebP)0HTKPFX_)W&}>4~a;cIxBRcdfqbfBsefu2IHZglde7i^oyozsDZ(
zpYiV+2N|`thb&7HI;z#z@NL;cUKSd~()5Pv)SY??Y(O$S`FtyT$XV>We`%}U{?FRO
zUu3lYv-aJ)e}8uSE<VA;u{s^Mh44@tQx}i(mu(QCG~zGI56fQ`9v$N^tAg|S@E@DW
z4vXb4+g@#6p2c7G5n*pAtY`C=os(?vd)bYM_{&ae;pZCXFFQ=>|7txk{Beu21ULH$
z_{)0kjNoYLEBle!P!RuJ&w4q1mCi|y8)5Hqgo<bE*`lVQbnVo0sfxKC$bJ+&6xpUq
z@4-<QQLoD)F0UDu(<07n(5k*nx?hYmydSfjjah87!SSaF$-~Yqrq6y%zQ2fFxgzvi
zqCQN9?@z_FE>7Z?DCJPjg|XoR-_KOsLHNEH-|^G+*2v=6OZkd-fbq9?E@>xxpW{8s
zr%01op;oCF#ve7+;II(mv#nZiO`^=Bo%(NhQx|?GQ?j?Y7(Af9YxP8$8MB^=zJ3<^
zx-jAY4*EKerih^OpP;YsCI1Y4ZQb}C^tI%--$h^T)TI6w>1z!$F($suchi^75dU-Z
zwfRhwqOYOQnZDd$+Ip2*4|eK3C|{=0vfs3%FZR1z`TM_fx{2E2ff~oZaPg1wFMe43
z3qBRYzy6K>{(q$t{&W8R-DC(1`SE|RzkknTEr>I2|C&(?KO4Ix6w4n<_rHyDr1_0g
zgR({x$a(R}h}LWasH-XR9Mgf!M{T^GefM%l2}#_|DlR(7=UymR9-v`NheK;HXtzVd
z#S<?A4xYRHZ>ST)+1taOIDZYd*EsGdvEOvzn40E&B<WS~YJTHrr0!D0__8_`Nqi}l
zYmAkPqw3=MEC1w;SE;RDdlGx=s<M-8gM4K#=SH1`6|%np$K}$4b1%S`+285UjiO_3
z!6kV-X9JyYpHBu%ljM;o+hlls5SH-Eh0%&bpja|9G#{<Ok)(=K8f&>7NgPwvF^#bt
z!tWH>FKI*9a4>ibngBK(&1>UqhB|;%L(pWIhxq07koqPhRB`|cV^W&^VV_h|pyB|P
zA#0u#|M`g~1T!?1_>wOW%)mGxg8ebvHOBvi{hBuH_H<z@%BTP}j<0}pk}8J<YlYNH
ze_QbfN|D%7$-0XfE9w!o5bs5GTfjgCd)GMi4TU@>7*xuIk^)Y#bLyWrH4)9sUM?HH
zM!g{ywY5aRfixOlZjNHE?)JTr^c?SCNwk+g&4Y@#?y<8zjvOAoD6&3_4NVgkVgz76
z2gF3Tx365U23u+&3$LCm<K&}DT>7y*TY7TQMbeK2vge$!9OOneks&hl@q_qD`=q({
z@`pGdt>1R4sX25%PyYZpu7J2S?>|G1gXk3zT4rH`ExJHoLvzX+lQB6~R0itj9It#Q
zi3-|BY;pUaH=-b*W>2SI<a~lO0>kQL^Mt|yP(}9b@*H(4opXka+7na@9kDbZlQ>4;
zX<>hn{+v!$Bc70l^9e+jir$S?r{72-@qztA&X-78`IM&@NbhW-j>Hv5B9e~PyIxHv
zB7vkdOK&JI!zjJv>&endzSi`j>0wCPK(T=x2tlz%&2QtVEJK1%SS;}s{q-z(irzL*
zJZhqN7!>-?1as`Go?vK>7<#07t*i20NaE_5bvH?Ww5psjLG<3oBQ@=77IQJhwCGRV
zFW4A><X@+M*k^)q+;J5a=3_#_d4QI5&EnQU385qVQX90KYP-7sjNRf<_BXrQoFykF
z6m50+Z%!m>&iWDdl_{ws{5L=EEIW}lugmBuDL->1u5mW+a@HheO$fi`ENi;Nb3LDZ
zi`omQZ}@qY0y?g5;MFDWtVBu+`vrVX_+k0#6e^~v^3)+T+kCo2xmgRZ3r#=PWXZ!Z
z`#Yuk3zB7oHfdg5l1>XZn5>w&tO^A**N7sb`I-F0WYAv|xd7hq6e*f{_Iw)bS$<!B
zxyI8{o}kWznsSGnej;kdX|{bPaM1UW<vn@HU$&IFDrL;)d6s%EcWmJmevUDJu2(;Y
z=SsF&z=h*WajT7Ny7^Ng9gw=({Anz0Wet=nnRO#|z4^1w`1ykQQ>@y&o2uT;V@dLt
zWtn&yNy6G$ki2RSt76llaz`p{XZhQ79?ObYdcpoGnVjkfj~p3^E|X*Exa%l`D)e`R
zyDEP+Uw?<Qxn=|rS^e){h!JZ_4L-6#{GP+&+Fl?hRtk>FD$uNK4Nd|ud3Q-<jmH(a
zKdCxZrE3016#LM(G&U(wC%uDkVa8-&8YC%uen0-&NSO=YH&PNw(Z`wKNf!7}HsD+n
zoMz$KA>d3CoM?gnC`isV!5Tc;z`a~@pFCm|nQejpB;YSh@aG*Pxqm6(znkEV7Wgp%
zziEOUSY!>{J^}yM1pmzfKP=!SCb+!?o-N?}O>l#SySvo;b`xA`<(?tAhnwIZS>QVb
z+}{MJc8b*G76ErN!N0X~4;Ap4!$#v}zrm>Y)dK#~1ZP@EE)?*36MU`(?keErCV0FB
zJ_0rBzcazZEO1!BJ`?<~1>PfIj|pCDf$Ic(hY9Z7Ia0yR0v>6CZ5Fszz*m~!t1a+b
z0!}l*Q!MZw1e|DspSHkH3wYllqfgdZ;Kv2L$pqW%k(xXr;CD^%Pz&r8@c)=#j|G;S
zG4)4H@Jkl>0Rhi5!MiQ+ECEk8!99>rjM14cV3!HL$pYUk;HymVUJE=yz!#d}(H2+}
zaFPkW)&dU|@PP)S@pghe7)4$o;LRp@rUgzH@cSnCG7Fp{;1^7AR||alO29ue!SgI|
zNWc%8;K>$vn}Da7;JFrft$^Jo_(co+zJLds;8F|xCjnn%f`4Rze<$GfCirI-_?H51
zIA}EfSquE6fVY}pj|E;V;A#_mhXtN1;6It*=@$5Y0Y7Pi3oY<80WUDY6D;sJ0Z%i*
zw^?9Kz&DxTJ1y`~0S_|4H(22Q0`6^sueQKJ0iR=n7h2%u0zUGU(fGS9uo|?_Oz;v5
zd{lA=P4N8|c!-SrUrg}tE$~$W{-p_i!~!oAaES?i#sbe4@I5BD%mTkF;9E@atro?;
zBH+O$_%{~#IRQINu+2iURKT4~u-gJJ7VxnHM&qLuEEe!i6MUtWTTQ^#COFLk7fSBe
zOmKk(E)eiDCioT$oG;)q6CACHnwK+7aJ0YGy!?R)?#vFKF)s%RlA$Iz+Eoq#_cg&Y
ztRgQEaH<KuBrSrZoq(J68;$R1fse@$*PGy%&yD0hDBurF@E<I&F5ovz@N*V;hk$==
zg5R{j8wLEZ3I5OmE7F{4fGY;0>>_k{w>Wb?*APJL{+EbQnm=Xl4L;8vB%gO3A_*DF
z;jdWPL~m2FkuO(2R-Uw)#AJI#23QcRla?YHtwgG1CrzC<lbiD9>wmpcWGDlZzB!Br
zVP_Fs8j*E!C@>=HX#SU#4li1j6oa+YUVal9MELn3bylWW6i7ltsjarm2x3wG(lEKc
z$x<VNmQhKL{$L(ZDAjSgW8josH+#kBJfUEwIF{H#bv$$ak;)b*0e&Nv6EQ%H(Y##m
z-k54%{40^{hLn)cUdGi|xPFB8@D*pjh_vUPZr{<K-;+Vw<D)CSvpsuRJ-)d;Idr(u
zp58n~+Vef0X;0(=8~rdHC#|C|B3kz&s)N)iM4!>s`ez}jPQaI8Zf<%Hsy<yIincy;
zAA)BNN3`taA95wDla0)ZHAtYI9JCN+-q!q2s(!YYzeNTwQQ(ac*K$5B8Wg(tD7Gf`
z&pMyuMDmS$GwvOG`Tnm_$7`v2sGg@t?{);aMJ==mBm4_#v{yRx`&3dx?bBMw$Ft$X
zp=BV6cF9iq{=2(mjs7iNG64o>c1Z|c!GT2L>e?$d@GRX!zp&RXKfF7*EDWXJR+J=d
z_73Ottl2ln>2K4$awyNHZ>LVa;Yo5Tl8N|O1GiG?e}s|nLTEQX#1pBkzJzm2le3$H
zr)MUwU^9g8P8<22wzTljf((y)=xOymcIX-PJz;3G`gRO0JhUitkxE|Fk8hF}UB)-b
zi~2808LE{XEYUbQvRs8*Ur%u_%(9oumOkzh$xEdqt~Ys#Z#q#c`?7%pS#sS_)Fp^=
zY@U=Z;Onp`rHc{9^iy@;v%S0oeBtr0CZeca7doFGk>R`!w$ySsF?XrqyqDgWo)Sf?
z#fUyfw}jRmrcV;tvB!a_Dx9KX(|*JuM9`NEom6zNrrqV?eKN?lnv#pakr?_lZ$wJ2
zhGU<d&;6BwJsY1>V`#6K#Eaj=>@>ZmpQIWJqt7RlSyG(Xe2(USR?WDga}<H<JB3uS
zc+~EtA?>6gMm;Z9^*nkgTF(n8jvm65FbEelu<)iBu+sF5FBKZD4U}0bRwE%NJ@FE$
zgq&P*VhA&6jnHNG@@GM;&tam69*-&FEf(}|EFwq^DX(P_zv*oiK`48xh@xeoK{1%`
z_%_V<o*i>u6!ZKT%=U`mEG(gI4biS?Mu`8$dUrZI_Pyy=y*aRF)!SaN2XCiPaSZll
z--i8_J!Y?1eSW#BHpIpWwcM;Y)ISx}y<<?1{5I5MTcO^l>TIGO5Qq9oL0xk&+9zRj
zsc-C)rY~C7dR{E*L2;-*#t0g^7t|(cVl2l+_x6~w7R8h${UB>DeApmOKV|WQV?O7_
zmQfsPWG#(M!XYsjqAR&j_Gqc9DKvm%-NiC&Mi-PUk&3)#pox)O8c=AXO*e+ts7qoz
zvc<8;N(9-|R>*$;jmX~E9apn~vB+d6Idpz2WVd}Ivb*Du8EftEiy|aix>WGh((!ay
zOiSB+Bfj(F@aaz}RA#q5BFG+Wg={&(*x9>nZGBuzZ;VCuq#(Pz6|x!Mh-^VTvNOG6
z2J|UGRtK+)Z|23{h%74(nbB=23ZpQ+(q{#q2&3`%)?w~BdrRwf#kKU#*vdUG$Oxm|
z0@>nkMD|2HvaYeno)csug;Fz1<h~YENC{TeU`t|m)SrP9ghF98xnoA(_KhX?h%4FX
zmZO*w&9wkmvdFB6${n5bDMB+cx7eZ^B39Yu@?p=fio6ysjPJ*vu-I6YHRq@?(VnV{
z>&G5q%8kt0)wDWV0b{Su*f^?}(G=^acSq>NzLw4H@r|YRi!04&Zo6o6)54--SiQUf
zBV~(T7GtkI0Ok-XYgO#;$bR-l{&i<;BaLDoyTIz@o}unySdS@rT<em>>dbv2l-$86
z*{Wv0)|tg(t@|URR8x|{?}X6$kj0vN0Z0rjXDx^=w$9e7!Igq(xD@+%^zC%@_CvgF
za{4|B@S`Z8w*rih0wh{*Vn~YuzNY}!Mgct);L<1{Qvtd~0Y)p50K``MGNaO(uh)~O
zkTrcBSKdZrRiGkvqW<0smG)LyAW3k=By_xpgzw9IS*}~{=%<SO9qU+Z32WM`ZoyE0
zl`6p&SHgvY=#Y@aED*)M_cQCg6DOr+ww1TXR3Km8TQ2VnidNt<Re|1AfYJiyTX)2~
zRVHth$y=ugQ$nWSnwciVWcoldO_WTZs!Yt(nv(tsbqxUO7{eUGL`6S4*`E4Ly6s(a
zcjOJVTI<KLM1-bA(F#XT49$!J`YOQ0D8Qxw`K=0`{fz~eNx{}qZp@cSVljBiQ2)H8
zso)IN;jE>f|J0dgOM5StT^+Ncf1#qHn3gn$HjpvC(Z|@GiZ)uD5Ri^@^>bnhHgs8#
za6VADBx)kL^4Qu|d)JdbmpaN%R1^-!5!ZvPvaOY2*)9Y{uKqXGjnVyOd&RF<8zSwK
zJm>2_6Alrx2W_vI*CJ1v{_AZm^V~`vxi0Vgvc}6j9m*OnvsdJ**QhEY(1+d4ZKed|
z3%RKLd}#zdB(+*Z%W_LjUv7Vrz!e#fI3!Q7VVtoi6+HR)n!-bKJEThJ3Y!FUZqfQX
zcS!4vwLyQL1`?RqWXSf4>D(R>x<V>qs3?~*SP3CFk*rMs-X72lfvF<^hXV8i5IcWv
zW^cO%|6NZ;{SCT4)RsbGGdo&m-XfWQBAGXHB~EPS<05dyuAc<s*6)y<FO$=_es!qo
z?nl&yocKXVFz|slH_Alr_9MRZUMwb|*%XUOh5!<T4CPtESbV&d9p!b5=&Shp8#{Os
zFFJ;P>=i5EVj=m6Z0m(PA2YIosDSDz-Ld2yQ9~0veM{aEj?L+VU3`_iBkYwQAFCh2
zeEA_l-!ZghVoY*<D)jYEsv3eaQ7M&`N<2D^ZRXp#Act+{zVY^z|LlHq2V=ka9BaS1
z(^>YL^~O(y@ZK3?G_4@gv{$XB?LJR6ttaON{u|npA?@*%&6kFhu9_8T&u>+GD)9)I
zXS64c_T()|xmF^3n*IRmWz+v54rUU{Dy6-%q-jFY4wrw7@&))f&%QG0{1JW)Ux22x
zd8wnPBzawltDVhtHA&Y_2*2wrYrMoWfX}`~?SS+Rza(&Q<#D#){JbT3*D6PVCxLbd
zKg#Dh;U#=_3NPmKlF$QUu_flDT=O?=v$Wt{wSBDl%`JSzBClf}2$N7d{Ogo$hFImo
zG$)(>KV)Ezlm^6EAja>6Wo`A>YQ<y)|4om?gq6AUhg#D0pD51rq(UmOobP6?gvas5
zFVq`@Zt#s2GxWe}>9<NfZM#v27eJu<K9Q;>&oFX|Zw$2=crFEy^DCL}m?WjPg19r^
zDqNmER<&bD7U#Y7`HKFATqB#SA4<#pL;A^crFwfvisTt<<Vh3w3npBipH)4TeBg*Q
z4b|BwelGKq;%CMd@8ee2N14Q-hv;Z(ydv^Se@Ey-0Y&O58Sf%vcC-GMEwCdZ%p`<l
z!40+LLu|#dYzJ4L7gXI$#vP_Yp&jzdC-SjgKGx7tp@4k6&qqrP_KM$0CW-$RTE>S;
zH?WDhsG?jFeDYDoM|e@FSe|7q3(pDPAD$VSA&=8TcL`X$14BQQkJA(tz9n>%B;FWy
zg-5l-%lW^A(ADyBnS97KqaiV1hkDA#`Fw=W3kkndJrHgmP7bx<R}2~<&9^BA%ClGO
zkoJ8cAD_v`r}D9pkC^<bIpT{JdRIQ)I7{j;CE+*n@rzceMGMqP{_vuBltt6RvsysL
zh9`!{3oNpn?%9YJ)eYehz`{QY59N9Q?j8@yHQ@=K&f(N>|Hv=HI(#+_DMhb@)WgZ)
zNQ;(G!T8DfTbT=y$$6C$kmD!kLNdr|HVx`0WI!G(=54pK!LiBh8(7~Ur;7+FOf0B@
zL%S;i{tJ{vtl0O(uBbbqEFzgOI|h=v5)H+izN^7zBwQ*9gRF#Ctb~!0kYy$O!b+GZ
z38Srq1y;h7l3)q<`cx|+C<&9SH*T~NQXrX?eXNA5tb`ttU?~uK4=Z7uB#g1%NU#zf
zlZ1&@!k0`#qk*4Df@URrq!NY_B$GaCV@&!7t!~N#uoAuB?51TF7MCPM+qT3?cu*2<
zvEG<zCA=XC(I(txB^;K7@zxtdtc0tD#YW$7SPAz_!dUB#j#k2-B_TR^ht=52aQ{6X
z)mAH^;|08NqlK!{N*F2$H(LpRv=Zh?g1EIBjQ>$9VZJ2fSP4ZcfsVD&vA-Y<ZrU5-
zg+P6`rU3icfOU$%Qe^-xRQZ}dHuH4_c8vksW59@xF+5AcW010#lvfk=4<fm7F&kK8
z<bErfyV%U#VC4R_l~PthWzED3;aH$5i6)ju6UP~eQ_OOn5NjSq{T48da!210%bbbu
zfbq_i(Zr{siTzQ3Dec^7qL|^#H>VkiM>j<%_Rj!TviEBCcu$Ku1ClV+N@%MRhWeg(
z77+D22Yr+Jy{r%orQ^z;kCrlwM5^#uH1WA;qQ^*_ZB^l~09Ia024m2lR%0fFwNVnH
zG+JUMq+U!ybpG9KCEO+nKeSMdwi2F}ggdQ-fmTAjBwS@B^t2LANkY1n(AG*wN+%)8
zlJ-UWc{RDse!hZC^nf3}4owPO4KPzC87Xt3DTB<E#YW0_l_KqbQ%ag{Rl%SiNodD9
zxbhk+;cit0NpMI)ft4`YN|-7MGp&Sy(MroQVOHQ3GEwO*8zRK9N#2S!wb>f74xE6A
z>CU~zKx@7yL`jD_ZX_M*cvN9xN}@#%eA<9L7=aOGdEl1@Y#K1#IrfZPZOyd{(vpkV
zpN<*ga!SzcaXEe{IqtEVRA^Q1Wl6}l5?ofoE=hRMO31PjhC4{O$x1liN_bEb23iSc
ztfBd(B#f{UWc@bk9gu`5F@Hz`!rXySM6~@((x+H2zHGhNmtE79*INlsSP7#gAvzHs
zv=W|?1WbzNytz{)us}bt9vGebbJ7`v4(p|E#v;wpFoGdX4E_8gapBENAYORcetb|*
zYWR7NS&5Vxy)D~Y9jHHYY>mV*;vPEgiA?bl{-La~!@Nt-QLg3kQEf_6DJK$|>ons=
z@M+vvaO=#_0aZL_lD%yOGInNYdlVv86I-hKjU&FHPV{B?$*c@L97ss+OvNRN5(gvX
z8#7AVnj{;K#?dxh$`x{OkDGqik3-Kzu}g8&?@9>08V7Pt^V~&Ey3S7El$;H{qZKNt
zr)yf6@S+jS_uMyYrh3OC?_jn{s2Se<tnX=eNe<|q9Ot&mVbtgW$&#e9bUj;^TdA~U
zLH~#*w}=<U{Tfd1NA9+S%pq!Gl|IB?QAu`!56d~mlP$wnNywD}+%h5Zte7Y?Z&!!@
z-UrZvL}qKtow~1L-!Kj&*Q^mL?aHO9N5h@yb-6c&=!}Hq<Crf8=kMj1)c{iW=Vx*!
zuLRL)gVM-(G;W<Zp{HF6-Arkai)LIoU(e~}&g4Y<PW@)w8p8Iuod_%R7Gae=M8e{t
z_txjq{X_2m8u9U5>JSYLm8>6LE3)WueH)}2^&b#<lv880SAXrMGtl+O6woGdS58;&
z@DvFWI+6vv3!`ZqgzxI|%S9x^({X!wIE_JCGCWrvd#FdJ)7Q@1R({LA<&xpU<WWq{
zONNh-M=?h)89q`T#l*Q}_;7g?Q}~kMqlmYYI&uJ!ScsGqY1PD4%DfU&UpX&2_Es+I
zloi^jx#pMayAF?zR6ox#R@FbxF~Mp8`m)i2JjYa{338Pmg{w9guJQ6ywWHkeGmuML
zI65Uz)tYk0;{xTUYEil4=jKn<s&dCK%%7@d<&G!JpQ?4`4pDTh7UH1g@~u>C1=p^n
zB#8x-tDcCBG=@*NV;0w1Ng$h3U6LH)IsWmpXSkh?a}GZtvrybW@E(qI@@nW^^{R)9
z3IndPm7<|0c>1|~Z;Efe%l8ZP5Nul;J?(iG?|r9lG@;#_^{Rws+}MlmaeKcfw=&3>
zlm?{N#mb<<mHCqI`a<d9F5jc7F_;WkVz~i#7hkZ-&!IVGSQyeW-*zM4awFfHlCLj(
zy_=B_W_!@c@RX6^=Zbj0)K6EQ5m$Wu=sRcBIUJc=p+dCz$a$q`d<E{6T0|KCdFmE;
zO~u|qx~-~ji&gj1n&>eePHM&70zbfPUZ|?Jsd=f)n74T_?-O~u)qNtrA~C!*(k143
zJ_wX$KwT>0_Fb<ZUnTr_Rh}<hjOp|?)?=7XGchL~QJjMPQ|`jxBE>0UkSe;;cV5|v
zOFYEeKIY->ozC3A-Zszsy~sIa<5m#=kdsJMx-STB|B!?7ya;4vgA<I*oHaa1uSKVh
zTA$SU5;f(eViS#uMZC`logtk&f3<M0B!@e*(Ub1$C?{U}m{n<4v8JdEaY2n`Pu)(!
z<$zLlTcaK`<oP%Vq#lx!YD}jZk?|4U>YHcN2G2|7vUIs`$=lgY;CTNqu2!q`baBg_
zp3vIp)Ntp}PIO)goyggVE||}65IBErwAx`d>BpL39Ox?>WlLXW#p8>NKPRo>Y3?Z;
zW?H29dk3N#6aKdcyk5?}a?bx_mO<jP)G@8N19f_yBpg4^y;o7V=N2lHH~4;=y?g=X
zN&FOF2ZCRFzOPNWpPDpiCR&{Q4774?zI*sWt3C!&pd1&<IaX8|9;9wd)XSy23~F%t
z>|V`Qo6EjClkaVrp0NAqdHN960OQyrteNAA(c6@$&wopMpA7uN?QLSzzp=fo?_;?@
zFw6Z#U3iE+Pb=~9yhK^yZbL?!ix?5*B7vd*r&1Qjpt|iaHl&Z7hEf-z=_(2jHxkH&
zS^OY~juF;}01rfniG=l8A%o>h%a;h_z8h75{0CgbA%&aU!rXD5V8+hTB%F_*%v{S-
zbg_Hz6s|*4kr**8LZ0>dlgvb6?=L8X7fP(VI|U;n9*H(YTwfzL5Xd}lg^l<QQT#>O
zg1fy|_7LG6rfNQgqgM+Y>tk`OldUA&69&nhIk(CxpbVW7)`zQEs>F{tUB66`<Eq!$
zh7oBoHkJdCd&pO-r%gT^|NZXxHVw47ReYO2QI8h!e+jc>#s6(_e;J`Q{>A&t=-GcU
z@^8-jN{~|bm$AvL_fP~#{0EuR$Y~F#Z{j~pij8k03xdo0_4`@$r~Yu2WQ|5NnngsT
zSwu9t&XMn%Dg?oYifz#cB5e`g;s((FQ{e~hf!1ZDZHR4*wh`CJ7|rNaget7Xq?JYs
zYI#E+W~>eR4aWS^f2Y1{^?nm$53pPDJ4{jRAG)9ICu;T~HJjq{FQwdnOFUyP>tbC0
zPscO<rJ{(I_p|MKS9-Hn@9-W2CQO46p!$Ae<n^zsjjnd9Uar+Yo*;85Hl8sA162^a
zf3w7CNBC>zM1P9sg0j=iMcs(kpXwSsDs9eH=nWz#6rB?u>GU3uWqN2?Q*%)c(k?kh
zrrFEHA%l~iTta*sPvHUZCsz69F`^!(IK9E_N=bFw^BM(DHE-LWR3Q74)mhc{auHxP
z{~Y8fnj`T<yGs?iM5<sf>p})<!LsT0-OIfycO|aC)skI;26PnPtt{*x&+Q$vuzi)i
zF?)^vxby%gty85MygO`3{|t4{7su(NLd(b{;vI0WsI0?k@-oC^Znk`oT(R!uM+7jO
zu1J>-xK}!0TC@X}Q5k31sb<gJp|Og)811J6%+ez$X1{0kjnf<89-|^r+>H8>Rl4W-
z7Wzn}w@%S%)q~dC%MVkS+&xD5W<}dcy~2~uhT5t@IM+B}&!|{1%C}eKA=^sbIin9^
zbVzVhiKsvSyJV0XYs}tv`R{Ad`&!N?_J{w*E_tgFNEKR)^a1aO9wV1k4{5flLZ!I!
zFyx*KC_a`D>=j2y3W+n3!A@dhU0BVvN)Spj2`#jq7o+m1)4ULt_$5S$xYmr~tz|Z}
zxGwo489gouBC0^Dx-+?~vCVQ|2?V5FP1J7IsC`7@{JzG02kIZUZ;jZ8pE?0p_i^H0
z$k*n|Jf`j!Cl(y>8xCq!Unc9v-VvS}ko#S~a8>P1=5mz*sW*C4m@7XZ@bw7dO;mgA
zH~I$<n6$f!1<^&_3o4-|lwXI0ce35+bq^rmkc#K6<Wpx2alIv&H*p>A$>+iytmdOq
z!_FA`c6sv~kTLmU_0<~rZq&TF4TP*{q+JO$G`K97uKBfeIo#~dA$7H<N51zy-qIQf
zXI%6i79N6Z&L@m*_+2tLsN0Q|c3c+BBI2tQXweJRYhY+pEq$XXZ$x9hcTodZ4y*E<
zHSJY_F8`=>d6lr>D4Rxs5Of1EyNlkXKoQHOz;~oTxveN!e|&{d69R?Xo9^}!goNv*
z-HCfN@AQ@z2B;Q6#vg{)p}DO2BMahPZ=3eBh&-z#x9__Dlu`Syy+7c{o5lk0!{w~C
zn`sI1fd08!GIr|Muw=;6vGTUqwN}=Le`&ovpE8t{%3N>zz9mET@3;^3KXZRTU#O*Z
zyuADWW^~%qd|T^nu}~#FTdsO|r~aAxuGOdh;B4#df9igZxeN%Bp5ozCkkb&>xNj)j
z;(m`=G>ll{=mFn$zsC*a;shVJe#PJKF`J%{g$e1hS>FTo{S)_lO!|^JTJ=V}epR~P
zaPxT5@f@70^h5%j*#5{yI3Kd0NGzDBB&ABHMaCn6h|`j9wsWlM6w!;G;{^p79)XAh
zoq`OEK<bb!6r^tiQVAq`JHZV#ui>Q_3Q|bKz98qW?)NzCZ8)OUZ6@9fg+~r#^vQE{
z-dA|2NK#UhNa<CiQch>|`Cz~KMwfQL`z%oKqZxey`w5VvyfuZVX@M$23Aal%^Pbnn
z#CmQ7YuBX0BeB@JwJSWLuoy39xk_sjFjtF1T7lNg#keyZV_<`0wZYG-KH@s!_PENV
z(vp>8vzEZB{zn{@^pIP8eZ-C@uCRC)t_e0~VlyNqcw$l&wpk2>+>S<`69m03aQVmO
zM)3%e^(K<3`b~mC-YzlTmNz4>PgSpvaru8)LizC+>P!sN^-d8Cvsz=At}tj|sBDFy
z!Nf2_{{Yj!s>9^g7-lF8d0<E&F}@B747+s2EZrNyFs3zzSqeig7;;--kgYVq;L%4#
zFi0kPF}5E(3d42M53MkWBrF)_>&X!eS*<b5R~QCKKeWP-Wnw7SSFvrP`XRkF1~$PN
z$^p_3<d5%%T!n!lq>uH55hSTCkr){io-FB&R(Qskc*+IOH4!`st?@_(g{QytNh>^)
zO*~5k&sWg6YE*-IDW+qMpk@kBhV)D;JhMzZvMaCuA%dr_H6CNzp8j!2|1eqN`$uZ%
z_8UfZePRTUSf}F#+Q^{llP>+!3eQp#kL>m9-6MD^TjMdD0jN(;>7Q13o-^@?K&F30
zvg)5#TjP-osy^Hs@yp6qcwRN}h`WIPu;4i{DTS?pD?0rh_bL0sZPm!7UcQN6!Hp9)
zX!oufD~xj^p2^QTWU2&q-`V?FM=npP#-q?~y74GPm_^R#6=MnP9^d<U#U#inkF$)N
z@;F&?26o?(wmnhhm!35;Rq|x!m*#<aeD6vFsgS3UDS;<5Q$r-v^Nu<m71GMY;QsNI
z{B2?v1lCCI;Qj)B6s#&dyf7oXQhja=hGtZ<wX-OxUGVV4V7O;gzLjQ20(6loVe?L~
zol?Lju7Z3{aA<r^b>#6mgDsa6sUF95d=@(f1s{Zs4&alp+<cC8^yhN|7Xds-eg~g3
z*pPXU{5|=^6FGQzQl~M$riS|_tqIoiw~fDz{H^7$n!gqNz0TiD{QZu<XZU-Zzn}8A
zn7<RLPda2g6@n98WH{&P{m{xtabPcCv2fqJd+fB~-n>i482gb|)XkktYPbEo7u+{7
zAVXXkJT?9iM~NUhHC`41tLx~t;MWC&)Tis#7#$dFn!!~D_cusD_D%~iHx@o|)Yd2I
z(tUwa`t#D@wu#)*H&1^a6U~kO376iY3cp1xQral{TT?F`BX@?5A+TTefpB-uyO#x1
zT{zR0o|cidmlqQR-f4g9l;9Wx4uWD`fdSQFx4&6gcxYilKfK|`%QcV5-DyEln}=4~
z-x|!6NOI`C$?mK?ki-sKFq&00Fi=%o;C>}tzLH)MNLNSj6czn?;#wMu03vkqbpbyL
zsue)^Lh#gBa*SQe%?3|8>X<%(`l6h*yq#06K*ECn4Owe+4+%r6WBX`PdH+iAhI*h%
zvhyXvPSn4a_7OrpVy_u{E0!9D<pS=2EWXm|ZOr#;T+3K_CeORNtd5YojqT@{77!-Q
zCfUfVc!lv&QQZU*TFWn2*{(+J0hQm@{()Sdn724dzTAs#!4r3EWZtg|9_IV-q)G#v
z68|$b{%6<tpHOY=yFKH7rkg*vP<dOvoU9k`Syyu<RrX|#_9rctTa;q-SEwd6X^B~{
z?XG5SNLpRAkMZ4B)c3(YBCb?4+aD#)Qvwo=cO83gFTDhPD=}tVWhlQB<0O-GTfcRZ
z@mv{5Y2>zkdxP~8wn2{Dy=bYGu4ZoW!ZDa52Q|*Jz1)|&o&N5VoSZ1`_#-ey_gD6W
z;EH~y1NAq*$VC+2@5k*!t`}=Z*k5TLk$A#YwyI5j&gX<upsS_h`acugIALF~e*Dqk
z+Jdh+Kzy`x4|g{tlm-b5*_?B<xQ)bJJLs%xzBo{yaQFZ&;@9=a%Q@|7kEyDM^R+}0
z^X#vj7F?cZl7ed|90u8$V0FQ%K-B{aPdQ6D2eBx@SsD<0&YZ7{+d92pQ<D+iFBQ^Z
z35QRCbZU>p5pqo2Nl%YlQrsgE)S&_Vkd6sm#<o!K#6+0I9b0K}^6=G=O$tDIz-T!t
z@HnZ1<C);L2^*=xwqSihuP_S&ZQ2pvCVT!@PVZOFsxvgnrkXT>CLOh!MEC6o)Zepw
zmHm|i&cuDvi4fex>cT_CY2&)x#2V%FRy)h;+m|)l=A;Dc#!-juig=B2X*59={SLFg
z*}n%^_XK`-&x?*{_&K9TBEJt%$Ik>d<C!D{Y$KTf?hxSAXKV`CHa=&H0w%@hOjW?-
z_?%r8uw8u4GzDxQpR=cw32i{_pW7?)>Gpi~!mq}iqX!q}K7(aVEPj^#WH{{^Drf_X
zY0o>-3c2&07!FjP`>VE$`y#mQj&06kRh%KYi=8{htec(9f0%V8_p8~<_d<SgM0|N}
z8;gkh2CwQD3>3Ov3<RpK>i1=!+WBH#OZxqbNxuW0f#i<BPKD%7GuEf$$NHb4-(u*u
zdcw)zgX3$1-!BLToDVDv{4djQD+*OJj(IyCURjXrN>mgYjIZ@~Qt0b+Q2m|D>mn3-
zRljPu1d9t4+Uw8wWW-S@;~^Bvq%k|aQHBVRdH)9_0e!yS??B*pcfQy^LZ7eqI~v?p
z0L|X{A~Rc&O&iGsaIXMoSo9f}GeVznIV1ELmoq}2aXBOO8JDxCl=;unr?h7=?RiI9
zVbSL{?p8DCbGt#GLDA+SborIlMJ98-{Do$9)=8M_l@a~=|3mco;u(6UY2m+>K0DzR
z^#9YRns5IAefANx^qc9^cKTcCbN2ts`Dcey|9<+sZ|VOF=<{$a_x|6>KP%86Ec)E1
z_@@X*@K4e0x=lYJ^jR6sj*%Y%RkQn@6an>A@GwFI^68gqAtSI_72@ey!vjQ#r$0qJ
z4YOE^czReQ$|KIw-G)TD%8)3Jh(viL&%2}Q^u>W;V-9bZ)vSXvC+tZYxuj<Y5}fu|
z!b(2<smP}hVc~>eJL4?fYhbG}upJX@$DE0Jp7*FAtR8dtE5uVzheRYqd-*U#qhU*C
zbVv*h;z@~~m581Ng1+ldgZn0IqOR4lu+M5jz@~(1XKB4rLbXxC;XHf(<~;9a5kIT7
z!-v7%wL@OcW>3<vB?CIpAvGd<va(Z)1`&YJ3(S?h+a|ncyftNemE_q`#LVnwb<nP?
zP6?ga#(pAutw}lZY8|LvbgbZaUJl!1v-{~v;siKjMUsLNr+s$6Iwf)fow~xNplw^^
zO;OOK7I{+@G`U6It_s?&Mcy<8ZQmlVk~_IE)~4k&+ABuTrEETU`e}(xnm=hp_C`}a
zLFec~$9$<;vs9WO7vJV=E<!y*DbUAl4DP#QYvG}(Y2!M*ifIfBQb*Gn@eUN!nER3O
zCnzv>PH|k0SnFScg2i(K_XK68$LWuKgVcQg-HCzv2hx)75R{FXI0A9OVBvy-ZqFab
ziv1p>p1_+4!G((mu`kx_RI|Racaf&uyJaj!fZ|>lE#~R2SH+-#fh<MAZqWw`Kx`^0
z!3SpcdIpTag)_n9^1FMbqV!N&pn5?zKjdp{uQd6Z&`SZQ_R0uuICvnt5~nHAzwK|%
zZuTB8`_dkKyx;NQGvkj{eU-%H8ECONxO~nh=>cca#?=Am!0l@T&f*`g5#zxz`DhLl
z=D)BGSeMQL=Y4TmA_=3@?QdT7O~?bzyW(DIjdyfzz<I~U%7Amo_G&4c4$Oa{R!ST>
zSxaQXe^nA(cIE&g##h8`Yg2Y6$zE|ebU^L0UFq_9a|XNI>S=tI{JwLLv^`EA!B~>g
zUTzqa4mhkjAS4~o^zZ2aL7((Z=>ONcK-$lqb{n$~LS;>6`(?u=O`JQ*0%O^OW{)Jv
zVXyc-9s~HF6~vohmWJ*{XOH2pw|X~~HX*krETmWT(IooI<Y&cpyK7ZdqwU_xvD#BV
zjQWel#an+0nwBdaQX|tNr*m{Oi=AyXMfleI^OA5PpR9;o&65#cbDm7*XF>8c^0R3X
z&fEcK);P=)Ux=q+t~x0z!xz;j)sE5|%D}3&tlm9}MNYP?%9YL5*k9uA4lEyP<H|^n
zZZ*B(c$FlzcU7fcHFlj!*{<rGZX$0Whp}T-t~P@rNK<KU+`iQtjx0dd)|En%Yi?Z;
z72T2UY7U4yho+U*sO_vWM(W4z!Bd(3<()9)xV??C<1~f><#_eY@=e1vA}0!`!s-*g
zFTVY7v1+kIK7T~kjQdY<+4%#DY?#2fZHhIN_>HiR@^{BjuVCX?qJtJ5nr5qBlZf@}
zgTg~U?vT1(YbG*cO|A-kC^v(YRr1_Sg{tr)<z;-U;b5S1i6N=URkf>KAQ5Yc-BneW
zB<3r(J*jY4v5gYj6~5NfHnzHO!$lO4Qh5JDZ)zj&Vh%1`=CO@kTX^%_gh@N)ojkiB
zEq7ExT7}Ds<)xb!CQSOkNJ~*^b&{4%8jJTilbVbq_#3Z@>02f3!ybIlNa?Cl#EPm?
zHexm2VWe<>IPauKQmU~lZ!%JPs+25Ac})r`ymPc|(g`CiU8Utl(&`HnC)y%qVQ2<p
zY9wuKVd7j{q_hl`wq9x|II9a27ux9S(uKCv3HFZCQ<)PS9SeI;O`yEe&Iyy6%(N6U
z?HrX>U%2d|gh>a@v`%JP2bH$5aM=|JllGWtoy|0xN~<nhHZ)<<XDZEOx2rUdqkZ8r
zx0JVRbi$-f0=uHy=8A-b(f;xzA-1A+cHi$%77AZ$r--BRXKoqHOKsKb6AL$7W2^q0
zuaUOuJ$&6}tKP>KRMe0N1=*_4BnojB9%9hfrZD#2)XxeJwM(e}kPQy-BteW**$`-4
zDmV{K>mbS96NCmOnS=~BYNfjlO*<!*rTEeWN(EZ3wl#ouOx;rWvuhHlgo3KUKQt{x
zpd%Bizc8R`;15mfB+%OusG`cN2K~^q&H|m7K#dht4fvsHcA&3KOOOUgX)@HRmPx9X
zTNF^nSOK$QfNF51wnm;;jmlF^^~!U`PwJ<BE;W8qG4->=_(_fU$#yJgsRvUv*?&;x
z+r-@>{UV;R!Y5Ayq&Q&sHd8)eS<?9n2C9bjJ1pYws^G@)U&+J15q^~8N=%?V&j}tL
zzd_~TVVfjw3GN%eSsp$SIoVa##Qc^sIwE}LJ$fp*ZT#xsiV15(>Rf9{ohRBR<R%oi
zbFPqYXX%O4&DSR^;O8m6<;9doM=lwAPBSluc=>?59JWO4ljh6s7%x{V3AXg4;IR9z
zVVQQmBX~lW@fgF;4U??g?N?6qSUsAnjN0xtYdc^`aP9cL!|W@2b_l-I?<7A0bmTgg
zjX!@7sG5Jpnz<hq9-5Fg?(+J@bmzwL2Ltua7x$43HDKbQ;EIA{QqQO=pz66G&1qDv
z_#&tOPg2#K9iH;=rPG!=&HAf~FQ_*vZRN3-<96CiJur8T-z)vQN9wT4{;>ERQl}RW
zQF0BPUx{7$14iRRO#4;#hx?E$Du%O}-J*=o`TIE31=!LN6&vm4eF;Kjc0^Cb?xXA7
zzuq4kOaE*8<NT)Y?ho0&6{7ry>7%;jzp+1LPXiLmgekG>1B)0QHd#cdJzIwc{d{@+
zyZfvDpPf&SFZwU+FEyRE5E-566h_C2@{?5nqx3RT_(`Cu-&q+7f|4+lpDp~B4mcdo
zH5Z(3^2_30PX9B)Q1WuNdlrZL#52>~1~ZLjv6uf)7|M!%M_Muzi6i_i48@@sO4kk+
zH#rXg?Ujw^bl1I~{W#z6xX4+XchPjlJ;BApWO-0j)Of{u5di)yzgdSmgyef-HI|pF
zmac@TniFd(cO?z8xi6huwJ*tAwINip>r9~j{^M)C)n)Z<Nw{P>`i2CZ_e^4OdZb*5
zn=1;|pqs6+uUu2sbn$}jWlg(^FGLlpPA6mgd4G@#rg9UOUrebaE!eZXj*_D350`N1
zII?oul9YDgCDwiz73`F-fC?U-SSiZ!b;}4f&cR;<-Dys5-bFJwBg`dzJtFG|&SLn9
zA=RE_#lr<Ifv5KKI6H*ojj1EeVaNL+X8n0vwRa1KpuF_Sm3oqQYuT>0{Brl6?%lRw
zUrGI$K;gZ|4>>FMCOJ#?W;ty&&XSrs`&*}_GQWL}9eDf7nt(HnZLDXEy?V8N%c@W%
zojij^21<)lQ(lqJvFypC_*SwJzdE>oVx|1(cK??&Pd4J&kXL}I0+5Y(Hsk>mbQ=eN
zha7$xJMsnH3KWRVc;Jz3cw=8acxXmtu<2CtV#%@W0e0m@;tTD@>m2$p_(?HdY}u8s
zC$1!gOAt}Du)DeOV7G-YkJmSvE6W<wiPz_}=T&a3M{#JQm7ZZE3fG1E2?%a~^N|5J
z@vEdMskn<arO|2YdR<A?sU-$9{$<H4Rgt_e#N`!e_~&w3woOZ)ff#)gdrU;n9D;9w
zcaz~ZSrxX4rDfpv*(ih<<ax+Yp5v@89^}r2i^NiM@17qI$JEkyh7926ziDbN+;!SE
zGdYhpu&R)QPjcVr%j=82n&cSH!*q-^M9iL<;t1@%|B}^h3J))^6|b+-+o+#y@qX{i
zxIQqxr-Mv=dVmV^)2NTxb{p|oVD~+3Hl+kZ+q-oEUU;O))~8!{o*L12yE;%k7N+!-
z&#`@N@-?Avs(elDn-*BSAbSRQ`SSG5kgwvtS%EeSvh(^53OKV5u6Z@l`8NH<H4kz=
zHjyclm%u$-wS-|C$pM7QI#Iv+t=AIw1^bO(AE@d*XG40x`GbwS3uiB|J*Wqqy|(ZD
zf7p8$_@=6CfBY0uEQq8iXmxNrFs-E8l%{}Eo=s^Xm88_RC@2_`HZ2+2B+Y|TE=aL3
zPHwGo9cSkLe&aaIT<78ZoZqM)=g#Oj+Cn04go29J2Ow%IKEMZ11oHc?z4u8@(gx7$
z{rx}x-)925XP>?HbM3YEUVH7e_px`aNV{>j{r2m#k9|lwm!qK~m_UV$Ig5^x;^oGW
z(bDyvi0Mj*iAV=1(j78N91<&#m1Zo7rmIaz7gs`UA@n_6-63OE_dE8kdsx)0>|^gz
zWe}RPXg@1rDQQ3KhRJE?p_^*HcRRjB2~sR>pF0S5%|Q>6N^Evyjlqe2SYv3^(ck7g
z(1O7@#Z(Qm`KrtmQ_4PuHzM%p{<QYR+as-;57`@ms;>WTJZWDX87(VB7~-g6{qP-Q
zDGb4$kkmP-^$hSc8u%%Vgi8zF)^6y-t*HGqLt?$r`@K)&<nws_(D_gL>xW#)^+QCS
zq#w#t_|`mx0`MJhcJd)l=!f!5?Q@tB_Y^QuM(PHLDr`+n>nwC+M(TVDAq(i^N{B-V
zAq5PJbEguL&O(sS=q8cRCMDJ7N}P>Ksw^eMqlA2RSd>+c5IVpEt7$Sd?0p)4`1dOs
zq4&XvMf#;H7Gs^lOqaQO0AF%=ltrfrXKDES+Ty|72%4*`pgn@VNvNtA6w6E)U~{dM
z0^NwrKVr~wl}1W|?u|k{QB~8VOlDJt&!5uj(>x6Wn0EcUFv5IE#`}wOx=dI|rz16)
z<U+5~g!})<Xa<^%7DOOxOguvGHID_D^e(tzA(o!>RMcupJgR3hPl@#`H%)>vs4(1P
z-AAd-nWNIf$E{fCWNVnsDN@DfQ!et!==tOgRs)p8Q)W;~m=}wrE&Y;uMA9)m&nQWk
zqUNMs=Q)f~<q%)$Mv+wRY2rz_!hzZyO)V!){gg<Z*;AmT-pjE@wsX8}U^QC%__Ftk
z)ViLp4+@-j@YG}h%Tw#3si6mrPkvk^&*^yq$<4t@ne*Ex;aJuO8sZppW)3Ez;oDMB
zI*c~$C+gESU-67rl14x}#AUK5$bP~fgzevH-5H>k3_H*2Mjt+el8-c84Fh@IBbhCT
zg}G{MRi+2_rt~b7`Ew~OO&c<x<xL5Hxu`3AHvMtNhp;m}fR`tk4mA96)vGWPd^VGF
zU&FN5m@v|HTYK?D+`EGQMIn!7qC^xzX0?PnltbPQcu|xM#;>9G)?^ezbBBUDAQFWj
z0l|1>&9s+UZk^#1jVDMko0fIOd+$in#}IE`Gf&hYRY7M;_=NWacm(rnaptg(?a$ns
zW+<LGm;~HOh#TI_<;Sx>gUA*P*A@je6T@e&2BsnUre28{J%<4zwe!+i)YpR;#S;hp
zJp+})C~ypSp|-$_^~B4j)-5>ebsl#deD*a))z<Tao3e@Br;u%0-dWmH%rmweomJTK
z{;Xs3m^c-iZ&*Y`sb>*2dBIM|y~g%CaNHhJ7<RrguQqmkJP7F_Ywm(TTG;W)Amo%+
zm=?x%5EDCf)CTQ;hOlhd2JyQSDu&e4ja;io`U!|P;7f(Cd-{IbM{{Gw1DVu<LRgC;
z;a(&p;W7iO%he03ezghW=stAkqe|bBK{2eG!5Qe=1L)f)Gx-?VP{4bd#z&CP4BpdK
zV{YIX!5A4Ckqt7E0bcN`I+DyI9lBtou@p)3D_E>{f(W*N2tuf70a3yXP6@?h#`SvX
z7-iz;gE6YH8*rlYdz}2JQ|6&OsJd11Bla8E%=0S`n=vr_l?KlpyN1>)(BPM@3vZ;J
zFL+D4{#2SI&1=s&io<%ayZ!6*2|d4fM7>)#VNlP%vVipZ%;BZ&#S=6TxRKv;*o}#>
z50t&w0UxC8-b`dwyAj{iHK(THZ*eO2pTp>BtueS0lr$#8*l`SW@E+*km=T0v?C8bo
z;k=TxPD^|70tuGFU7^k&-t;Cs`^f{t!$sYANjJ6}(Q@crHyy#c?TPnb`#thT1eK@6
zV7IAYwHg8<dnI8UFpYqNlK_Vk5No3WQ>GIzH3@JO0f!_3W)SeQB)~BQ#FvN%EK?@n
z&?LZ20uD<8)DiIVB)}X34o?EiBj6QDfRhL~A_;H`0kuhh(+Q~R&Q<CzD<t5^SjsqH
zF#)ek0yGiusw6-&0Y@bPmJ@Jv5?~bpuTBD7NWg2702dQ5BMERR0k2I0tRdiaNr1};
z_`M{+dIF9~0`w5@`XoR<0l%LFxRQW3Bmu4l5IXl&H@yKx^$xtGfvJi0Sm0ss?|x#}
zBgC%1qrn;=Q;Isy8w7uo;E$2MoZy3!z<(^@gOk8vsld|Hz!I&4Dt{XDDF4(Xa5~3B
z@F7XyBs3HJvLtXi$U*Qp9H`3QB;Z4nq<>t%hb4jkLBKCh0)JA#hbMtQCE!;ifj=wY
zBa*=XCg9p6@D~I;Jqf%+z(*#5hXnk}B=BtlepM3qE&(5v1in|mM<;<F5OByV1Jchy
z0ly{*{4D{;JU3waLjrzn68KR8zb*;9N5H?A1pcvrk4XYQF5uTEfqyCB-%kQR&EbVN
z(But5hoH6Urh_s5g*Bp73cCP^_Owi={dn4zk8H)h`?~WTfNERsg<>CvVFov4z>a=(
zss?gGRccyc%ZIZt!pu5Wz~qN27H&>WD+%s|yr6Awj<kMw9)?g;3Ohc7?D%0`aoW37
zNn^)HG{r-L*a0ap0yC*sFb-`mZC{xJ3xL7e``suGMJ#3li8&RM=zA#W@T}f80H88Z
zUuLA}3ogeO<5c>TczwaA_zqjVJqc_G+BF?QWl$`s?XA{*+)rQd#ATPDFIdpUq&EK1
zP!efVZ=jGkeZgcUB#%k0h%;UZNoOI@7rb99^4SF3vwXVuiF`IHsrr-<j}r2d5>k#3
zwZ4G9qCwwl<@x~9yK<et*^$r*K%QRD><Q>_b%v942y1jRcVfSM?Gx??2B)J2VO)iI
z&wg6E;ggsm2q&(Oq2AC6WS<Qo556SFq9>a65uWywW21|dv?A!MMp_=h5@y%O&@Gq;
z=dn7JgZu_Adz_M<m#r$(!_o{?<_INCk}}I#nH!Uq`6+8$w7se_pGsQhn@XA_WqMed
zdy|&=ypmqvn6P?WDQ!RX$RFK2Yf&BG_IEtZO=|}%^MSz|(SHLdXP!c^P|m!fQIsOI
zAu^D1rfb`!DQ75symIE>dCG67oKf;AN83rslhISI<WqPN<&2WlBa)8k8LcF}m~uu*
zx=|#Rd*0kC+I<KOyNGf|N&S>ao!RrKl6o&seIezHl6tR5t?Ri*N!=mJzL0W8Nq$@;
z&*`}V$w@h5ah{~EPO6)^+IBJBOvx*3asB^K-OSwGm#&-ncZea1Zf5C*fpjw~5TmCA
zphVq_@x=@v8JmCD(2K73+5X?lGIsr-pIHmL3it-|{l6*DuGD0lg?a`{4WXVXVEP!S
zV<=pc!L%{$mES}e^NaQUl`+ega%GIWzcQvC(P?f;ri@t!Wz1jyE6SK>aCEt9(hb;c
zL`KP!F+7sYBPCPD@JM4RlIE{Tlra>ic+$Au4at--D*em4kAIB*B9m$;`&CAx4r~gr
zi>#n0iCT<8pO$8BUv?GDsA*3i)yn2(sOx~dg)@ri^*r|TmLAi^?^m+^BxBfhVPZkk
zG-Rq>KZxx+Kt*SE9%sSyp=;7j;yP7~bY<-|6Jbp~r<~q&MbPU^B=va5eL}U9iJ;XJ
z2lf1!0_eDVCJeDtZ9k++V9RhRcKs$-<3zI1c9Q-YN}uwlQ=S<FI|KtL7Y1>b;31q0
z>!q*CsOXhwp^N_DxH)#^-n6to--3EWR=igZztMO=6Mm!n1HcVO!)NZzwSRs`+KoHI
zZy)=N<|6mU;j?$+Ohn)hSTU|1IgS!7O^c>N0#_PKdckCN=;V+tJ(^A_&6R%PbcHNk
zM*nmf(RA;!YQBYPc2hO6a9us}F6`xXfI+to>Nq^8<H#VU_-vn3NZZdQm^*RV7l*#I
zt@PbsOK=a}Fh>1~OYQu)-L5?+kHl3oXsEF122<OlOt9`uFn85VO(qwNi=*}f1O{jn
z!57celmy=b?`!{tEwD(+>DEIKlQly+dIn`eMm_|En@UE0la+=ut+$v{aG~J&($+mG
zr0g{ZQO_n^zC0-tYP!r@Xe>eI{#&8v`e4>Eb0R%b^%esdd;_Ai*7yPyUD94NJ%v)=
zs3?0&pzJ-SLC~aIj2)jsOeb4|cg%x9r&v0`;%pl<zMJvf8-8XHPRlInvhe*H^R(5O
zc}DHSI7p5!5^EpcQd;nt_Tdgo!QMjc!(rq4H?$jQB~mE&Xg9#RkO{Ubf_sBHQ4T-X
zeUzG;0h-CZILV9`l!jDMVo8g3`arZ<rx2Ms$%W~wPN6S#k_(fTb?M(|QtpL`6fQVw
z!CzdeZ4RDO@f#Lm!*RLcVJ&#kxZa$4&ZzzAW@G0OP07$5WDf-ENugjFb3fvmr4&`<
z^GPEwzMqzOW(nKT(X64PPm=ZG`$?s0mIhO7LpRj}Bj88cpKj@VPg6W}C%PH@BI>dD
zekhXb?<YfT4J>J7?S?6!pi(RsO_6P^(~#FLs{3|$_K;^^j7A=~e?72Zo`y|W(1y8S
z4^f4s;7vv%ps9zq6s|uO7+NeJ)UNknJHyZsPrt|#`0B(8o$6V|@U_K=YL*Oz`Q>e@
zdS1txl%k4D=!(T%rHXr(f~{D_gsmiz6=cOC<5u}(k%u$PL~%78xUvLa@BJ~_hTQ_1
zf#Od@B<#=hF2{%ARrC_So-G3~em(2#VDkw-B9XYB<@bGoEtP$TPp+b=j$ihLc}L84
z_&5ojkN8msD#vCZLkB%%HuHUNA)EP>`@k?44pl-nB0l&pBXz<S#KF2U)5Bgfl~<Z9
zGg24uR~X-8>Osh&)LDzz^vQ1&>wS!7B*lJ;EtX^~UEz5kvygip$y~}kk7l-z=k3PN
zHGn_M?h&znv0ImQ=tyYYy=yaX1n3K_G7)l^h1A?AhtD<crLA(k;VvXS=bk}ddb58K
z)@6n3!{<(K4S&_RZ%}xRJGC(k_!{>WL-Dll|BEhbTYacAaC$q!M!$+S*!N@cy5nQK
zUjG4m#i@^-r;+^^lsxNm78X6uvHr#upocYQ6R$k1&bf!*ZUWGw{0pSi991i7ZY=EI
zjHRp@iPctDQ&i8sKUOXKepPigp*qoP&CHpyDbh*Y8$BDc4x$H^m7suaCQLE1X_^`1
z`TFNa((;fU6OmDUIwnG2;uOb3n6X*7@3W3$OS%0Ix9=i*Q93}fQbz{&bT!W?oa}hy
zg+u7-a8DJ#%*lQtH*NDaob_8}-E42XWoY`2q(?(uh6Mp^`>^vLUYU(7<Bx`n8sKQi
zF#FRvI1Is03ZNXE6i|+b@bfOJ6CxkN;wgGUL`D@=Cq#B}7(XGBN3elUh!iTIekVjc
z3gk`PDfBAOE`#rmDiAs$@+ODy6CxW45<4OCHHXP8)lvm^p%Wtf6bL(yV%E|L5q#{H
zrDMlp==4b92@&lDPKb<9oe()?=O;^)6CzirPKbOeV9E)R;i?lNZ!=hpI3ZGl6Dhx5
zEL9z`|N5Tj36W<j3H9~Ton3TJgs6=y82^Z^OLk6VUyA){ocTDI!cK^MfCo;9?8aFN
zoa(reofFx~U()USa7+Z|2m~2{0}pAN_wW~X0}7+Noj5s?8%)PBlGw?SVXzT+oE?V3
zIg?@O2iUO@afE}B#R=Et>3i9MkzVA7BOHt%PPm36ln#u1z(a`^!gT!PN`B&r9UFOz
z!-yIfEM0+>Nt?SktVMyL&U8|W9W0?{pd_LQq+6Xi958MAUKY4i2^^*hOh+JIM~FBQ
zf4xGy9tVd5iW;*V4)Iqsi@omUufxP^2Y;oi(7BaeDP%)hucdWc`Xnt&L9-9?Llm-f
zxaS|J<(YfBu<BqpSFB2(aTS7&Esl-vXBM4?^oV1XxCbP3q~Hly5iG)rwBU)a#d#8A
z%NALjDS3m%!YPLLD8eDcgjiDW#JjOGCG99)IaBfx&XnBO?@Y-J5<gQyPuNHw7H3Kh
z;aSptm#myCS&VZfA1&%b1`#^1@pbH6iOAsZJcG^RTnXi)ZO_H*Zu~pt)%yc@$-ec?
zd-jL#qI1>hsoTmKHqMAMg>>st$SB|0eQZHYUtnb4e8W91d&7TR)DbdXdDDBL&DTD$
zC1kt_){RBiKCv(Cg}i+w?%K&;^hU^j?Gp!qltsG;v?GRxzo_uAneea>TYAm+?o8H<
zuLT}h-8l}7?Z1#=H;$7o^}t0D4)hprTKIYh){Y0FH_MzA$asOC1{}UQEAn}Pa9~`A
zDG)a&wLh83u@s^VVd|bUti<rXaVTv6t^k%Gr*-uXW9UMcoZO6ycTNrl6=7h_-9a4R
zgSSxUxJ6rVoe|o8I(M74^#FxIn%ElNh2|X~^Bl}7`05U~TV`+NYNO>Dd!A=03UH}Z
zpx_lU%T3)vL7M|Jw&8&C41P5o|GQl=TGs!aif&#@<-?5RH@KS#9T~wV;KHl!`6y)6
z-AS~~Iua(=$L4Jm{dZ>3Ay9QFWURR9i_qqwpz42x1uHy9(0%*D6*#bS)1t#6`_Ly2
zQO~|cz5}uDBQ8?N_~GxO`yO6&7&LH{XyE8~*Lzx}_lCp6dv4Mo(Q{qYbM)2oQ0H|x
zBK1d{K0?P~y50%*mbUdO0&QNC5x&F@RQm0L-F5GC(eAqLo-UlGX5Ce8QM!wr<&E*3
z8Rx4wza{q97Nx%qGj^l?V(d1Qv)dl(uU-Ey^j9@BQ87+ifH%fzSg<MmwIBU;)K~YN
z_f^+lqkToJB?tqSNHkff1jS?5Vo4O$r{XF*Y(PSVrkuB$-V)xt=rcM$gO%|X*qGyv
z!Yv32BgamxSpKw&1gG2`D{v=aRmPo@=gk8#JiiEc2oyZ=1>)iIt$Q%cFJ!~~nb;{B
z8qwB_6*I)@@#-ToD?#nYZszi>@p(QE@QLIN<MV9biypbruAhg<>^8)&mBWCavI>2q
z{ZSr3sNoZRsK!>D&%<pQIG^|VqTS#WoYAB6c*d5^nEXbippvf|Bb}IHKZJX`_bYaK
z0_(ey4coQrpZb!SZ~<}bFv)j{ukYRVe;;4r`*03Q9G#3}EC0gSqVq<{uyu-Is|AB9
z$5+c2-x^;b<Dh}b>t#+}ADg}by2dHwrBjnvGg?BoA+q|$)7Qa)>Fd86U*$Mi!x=L<
zeUbS9O<4*2crkpvF!%opd_DZ#(-&jJ@zBQrQMg!D(O*QztKk5};b#=atIe4Guf+X#
z0}NM!o*tcZ0sQnC*B#~;^z^ae3iQ-R0@!L@Njp9_(h(^s*son5Pfs}TMuIPkx8|S`
zr`|~Ljm4(%D%5%900=T%m;tY34A_a`YB(FNqGMGbxDSIdn@4)bU<*t=K9M+8_%zJ~
zt8g6-yQ_hMep!NMpMkwCwmsf((h)nevnEp}eFQGFU4J7jOi27IhB|>R&y^rezyk|<
zS^`}_6oi)L5k>JN&ad%fYV2<Z-L8wp_1g6}cAx~h7)OT<z3<279~fKXzTB{B-Pg2`
z!ue~QKoMrpV^L3>zAKSJM(NH7crt1XeGCINuF8e{ii3B*qQ6zYlBxR9gLfCIG9oS<
zoeq%@9`%vCR2FohBX@4Zz_G6NV`%N%OZSsP<=FooTy=lM?T;M<4rqTYkM_sX6s6&L
z7{RbRhp+=8G=^!o*!jpB-NGx@{2Af|`9?P>zUvzmU)FN?=pK?h`{DDtwc;cIc^_fl
z&}5;HW_nO6^J47{FKVZn|49*PGC2<tCD2tz*lPIe3peb=8<EOmb##92yw1&&GH)zH
z^!!GVH*|?lF)y99TZn3518qhge@+(f`a=Gml{Oy@ZeM>x2U1Y4ZG(V<uM-$RqwHE3
z+D@Ay#_N!A)xDbG;oaUYTsgJs-WP_qP$C4Hhs)u^ji)r>f9vkX7S*Xi@IW5VhBDIr
zbU3ZM$G-nhX<47)-rMZPeVTA*<LeY7656t+;ENQTf}7lJ-~W%athY{M<udu@kgf0D
z&7s+SS9hj_X0IveNP*@rE$is%kWqwpq=c&bu3n6@3oPHr>2ULFi!(DQ^1&4BKc(9Z
z7Tx*n^u8m7atl)#UGxWmeOJH2(nAH9LRs#icu>v38zN{pCEV=VOv(1s>wZ>=eJP<I
z6v&!1z+N+F(zThB@>nS>*Ebn3eGnyjlM=nj5_R)L(qPK_Acf<Wjb_(7RD0Cn2V>>o
zQ}8gz!)3NJml3K)8F?BOq#I22>SE~WBITvbp{bPF>y+66RtRELBZmx4w8r~@gGuQu
zwhkAK02N;lUSFd$uk!LTST1NED*@Gc6<-Rf%G9w$>6cMfU8w16p&w++qtXHMXCgxs
zlQ)C~NrTuU4<FUSNeTv;a#=Wdwj1s!+$V6f^***6NFR&!EA}Mep&PEo9%V_VXhJx3
zND%%mrxNo_300%Oj3ETu&f=wAMhT%ySbaC`q>cR~Jbh6o9YQ&og4f9>Lx=nBJsEnh
z@9I-2;Yo|O3rJ_*y~jga`mR1f8-JrR!cQ(bfyKS6i)D#!r&7`fg<n|IjS751uNvT+
zV?X>XR<q=tlFGVO8bUj%V;;m>?(NWj6v#QK5><pkAAgWo0C~KRQ#7<cH}*4@AwZ8l
zNP}qjEhJ)ZEqH_NVVeEv2Hp7srq(`eF8wd;PhzWyHjXnoo?}_znA=~Bw4;V$m6&Yx
z@Q(1S-hGVW{KIA9w&LN3!o7_riQBrsb|;CKy2DR5z7A&Foz{JbvCD^GHUn)2Y;Syx
zBAg^<`-oGMeSbJD>%eKk-#NnH8A4TG3gIpSmg1ZU6teKsj9G9+5og0w#BA7Q301Et
zcwI2t0^RveA|>!Eay!TQ49wO+6t@-m_FcVNFk6_?Aqcsjr&7<Rg!^1)h~9Qm<eiM9
z!HoZ5l&3RhgGm7~o6c@PDLW|ZeGJH%t&0+Y15h~rQX<Z5dpWH$60T--03mCL*@(de
zv%QDt#B2zX8MAF?=-JMhjX@ckV1Qu!ib4>BD1$Q_tMPWuY!sWA4V64b%(k1-K!-?~
zb7n(qp1$wu-GbRD5ofj?sOcWY_EBb|ih|j|bHr?T2m)n%)}`V#x-kw6f<F<^9*sw1
zJLs*OrOMMV76pW$JPd@*h787L9V{AWv+bPkkoYJBjtCLNCydQLp^p8OI`Xp=Fkpv(
zpci|n7kdSpJ-Y}eots@dSe9~%;AqqxycVM4Y+&~H48YXj5Jx_7AoHS1|3SUScx}{!
zP<kDpfK1M7m`jME1h2uwfpnv5#(uyu1n3dYYe$iYy%De3pAP8G?;?&ft?SzaF4b;$
z66{7+z#4EOOdL$ibS1?0fnd6lhWq4CGbw^ZgM^Js#*e*ixZ+YZU&rrPx&TbvuVi^g
z_bU}4!-Er1ogYE2V4IG=n}%TmH!ocMdp-k?-u*l7xqcfwq*_K>7wW&Jmow>;trWfE
z&Peb0v_$zKWc1>igCb1g7z(d}v&#5WSBO9ErQw`~UE$1yZ-&P#JQ5zY@S|`})&7uO
zU+}8Qxh<3M<%N0CBO`ztDV!-lB5XK?jfoE%MPZ}j!^Tk9q(wh(;m==f;?LiA@aLZo
z(KA^!L|K^>C!zct3QH(|5``s{Kb^u*J{x!^E&7#=X5sz!h5Y&FrS!Z=`51Qlm2V!P
z{Hg)UUo6UR;pP2mlPJGKlz-?F%C8xq{Q3dP_YYA1YEeF0Fi`tx;Xv)@&p$7v=SAAT
zet_~H7@+(|1}OhgQGN?A?^l~d`5mJCLzhtg;{%lc<N)PAJ3#p_i1Nt}7480&%%8t6
z<j+4ZrRPQ39~z+iT?3SVV1V-966Lpu@;8a{JNWaVLy6^2!dF2QHWc)Kbb#_d9-#a$
zSy@#(L-xr9yII-?Mw~?$CLh^6LPL`cWM8A<9Ih5^_U!vK=3T(Oo!Q=f>D*i0Jo<qV
z8W5*QJ?fjM9uiBP`OQ<O#ZpiD=BbCpQWt;o)Wc(`tG;>a5wX-Y-#j&KUkaT2zj<o<
z9tlhRzzE{nZ=S)ZSO&z--#mk>V;K<Ne)9}6Vi^!afAb8ki)BEZ`^_^L6U%^D`kQC)
z{a6OXyWczmIhFx2^*7I89LvCX8xGyC8mSp!5})(KrBMZY`Ch_v7<jNy#|}}*IOWa`
zMSn-eDi3{$zl$F_9oAJHVzvgmQtY?S9GZ?>fv>|?pyaB(EOZACrL9_oQo6H9agf1w
zMpKMquu%^U41z4Cfh<Ow8-exVmASjL8*al{C+)u$(}^c8g@n(Jd||{P)RuWV6wfBb
z)51NgM=V8z)guboUm1U=kJ!3?ghZIYFCGOO&XJN1ht(DND@ViRFMRLza>MrC9{@Rg
zzXNg>j_Chp%~;&JN{=e-aC$7(j-p46b`0dr?QkT>(!!C1WW1ALOAE<Gcpo_e?hdNZ
zJyf^ar98V#o?Q;lZW7OK`Xyv1WZscK+k|Ez!^b-K^LK~v<arnKyv;oCDxUY^OUawi
zg*<;o6i9g&^5^fC;>q)_;d$5dy!|}y)t8cYi^zME$h$-2eMsc}fOfrT!6RCdFDUCr
zwQLxI?{O`YB`%Kc2PKjzs15x2SRsG@ZYiFW=L6cuqvens!OLNT4J(Jq43|(&izsK4
zD5pb|b0}I)C|VB53%ndQq_A?BY;XzXNF*ard-(IQLjL^SQapKkjz-HNxqz3$1`bvZ
z$p9DWpAhnYutk)!NtDyUpFcW;r#bjC8O){9vcOcZPrLqWWNLP9r*~5GYqVwUk7!9?
zFt`7RR#WDJ*R-uKVs%y4UZvGo3SdtD6yC5~5U1b7My6Tn3w2i9L`KzcBS)|xT=^dh
zDuZD*qJ13~md3*H=_*yc;VkT2Tv$2_J0BM|iiJV%uS%PNu;I|JsDj4ek}F&_(veoW
z?zDua@$W75ej&yVx}TzV8up4*>%Zvtj&XhcQe3q}26*ctEb#*oWKjxp`rzxGpQYiZ
z@Cvi@L&6{3nns5#HtEq%?DG1U{vG`SGkdas)8sN(IuyWb&y(`^*S9M-UtsG2IuUm0
zL1vLiOFXNmg+;&@_VS=La?k(Y&=5h>+ULIhR&0NSPjv6Zt!Oy#z9tX0GWWWzFT(0_
z|Df=zjr%ne*cIO0xPJ&4f3ExmqW`J8J9M8zvp81+FP7|qDbNnxw?tEun~4B?zGs1^
zKDRjZ-Z^RerwBM1y3eBV=hl$5=t86q6_rR^$V_y(c-a;H$5-9<&nN3D-a@e6Pr;i@
zr1$YM`F`<oK4iStJ@*~^=Sy@IP?;@h+>37@EPw_pggfeYga%F4JdnE{bZGy)L|5@n
zsMFH8cL=K;>h*Z;<HQQ<8oqiYm&j77<deA}Vh2&fXLFBItjz)GYkL<md4s%XS$=Pk
z_kHk9mna{vd+eVFP!iVM_}DyluA%nj6U32*KhPaYm@H^A08#^g*h_c8;}{I}u%+N*
z?FIwL21=4^>9Xa8n?vt4r;=^kLVObUYV4fQbL3zg@?9ELGtm5`rl&%pC)`P{(9;O;
zrJA1VWAqgM`pQ%zu*gY9MW3K2*tQCKQa*Api^!>BKDKk(=6+e)HvcqKJFe|8_l6>F
zZ@AC%#sa!m&-in~^pdoDchI$a+o7+qjYC`dJYDGk*AOb4bKk<Bk^=w@>Ke|d>Jhqn
z?`kBax4LKPuloS*{L{IwDExO)$-J+7_XNL|FY+^7%V+=MeO*?!7FNkHD-Jgv7!-D#
z+Y<f@oy6%qJNgqu3ID0_0Io#8*9}^v2T|Zp8=;ej-wzVRV|8wg{pqiC-K1QWyU>xE
zop!@Nblshyq8XY{XtPCg(r);vuDjR%UW0DRX6X1vYd&L@8r{Y!^|r42NNDpC&3o9Z
z&~={yWopib-|pLPfBG*1@&R5yus{7LUH9jd->Fb%Ngwp9mcFl8KG#=MKKW~8Q2jr?
z+RnnV+)7x;{`m}D#VagS-%O#;Am+L7KZpVU()j8iv~mC7@NXMm)r5Zueq;o>3;g-&
z5ZHOUUtm;E4SkYRIk>Lt)A0L^`%@|KSw`aIdx4QRD`s*&74sVu^RhsNeTHVM-GBz{
z3l-@#ui6d2)O8=mDJZ#OPiXV?z!18o8w}ylyla2@RbBT-p(6`4pMe4UdKelQRxs9<
z^qsLAeyhVrBY@nO!~f_*N(Fx(+VdN96266s=I*vX1OLas8~A<!HJ{$!XWX@i`Eiv|
zgRWvHCIWnXgE1DLw$h<$#_Qj3K@vto8JmZoVX&7=&0gbRCrj+bnJOuJ6~?jGMm2wV
zz+Vb)Eyf%k<t@Qi%%l_JxiVfKA1@e8R*l!EvukkK7S*#nVbo?^_VK(64%%x6_3U6V
zXw(KbYK(33(~NC*4#!7mKGm%IK#O1SmCnxwt@|(!zjcRa;WzZux+4tma_6Um)_p*L
zED8trZX5a<G3y>!uC8+PfsirpAbbum<&j)k?a})ygMnU515oZl-Hx&SFx2s2)b#M-
zuSkvDydA^^TEzH$EE|pfr83}SrdT3PKJ`<Cef8CDgo{Zclfv~8gp+5+)nHpTmgmvP
zy^BU}b8d0?G*r`_GsJiv3Jsc}DbB44*_TM8Q<1>_xdp^P>91$$QK6=C@WgB2iSCp3
zE|Ac6nq{8k+|ZfT{UyCV%UFVq=G*L_>p^#{E?<bcWDTRbAg{F^`{xc_#U8Zmb;|J+
zPt0Zqw1bi89-8tV<xE7A5NyPB)V=hfhvBebT1<mpGZqL(oPRsS_O#iN4_ukWzTEP+
zeK>GHTd-<6eG_zgER7uxAU!Uoflx~L;6l=$U%L-i$l(%NgmE9HhJE4Bp%FTBHSCLF
z>KU6RJaKd!wZBl)GtCzYJ#R^%XW9*Bb-;WAdXA>&^8=<LY=A>*NFM?(m=RaDPM<)t
zL{2HuFYzazTIeoUT)H(H!L!?N-PY_j^Mth7ZB-MJP%LPb(QH489r{|7W=UH48O_;(
zbT>Owex^|Bc6R5hLa*yE2~5B>c54c-uL82|rJaf9?KDREC*)!J_yXrQzv?0ZV<Iu@
zPrE-L`QoyNL0$WC0_KC$5OP3Ih&UL)lOg*8DK+~i0($Z^O#!tU8sArdq8CUr^nlU8
zYMxMzA^M9!^r4=&?UjGhWxaD69DrkBOGw6YoZE@)aE0r&pTP>&Ip_d8sPe(-P*Ikq
zteE9m!^Xrtl<VYEs1GZ81Z9spjLsqtC8WX@Pw>$Ftnb?^F*PES&McY|NL<;Dnq%Sv
zaJRh@?b(AGU$1#_Jvx_h&#uts$(r9iP9hFa^6UhnMa(alWWaji(YW%p66bv0!5;ep
z>56?+3sgtn0}sOspQUJCVc4<nNBM-5EX`xYCXBmwF&S<r;pnlWED(GthH%Cw*vWMM
z10P%%iKCPVAi9E>MDi+I`hb(z*!~764_-00uX~xqCa?|BDU+V6Iop6-iEN2kk`DAA
zE$k4oRQzC1^17NoXuMa;Q&+~xQ|q?wMcFZt3bnpA^{amaYUrdF@%go>Kl*`q&5P<!
zNWP*ggh^hHldnE6!O3%y!;gngEZv7WInAAw`yjCROhcOV7JJu6-`{8NlcVpSviBL$
z_t$9zYkCUZe^O(&!2o1EvmLSj8>E%f5F<{4GVgUi#4Il1`FS~PVeBIz0d>#-%7DN3
zkbM(??v3`5n7$9Pd~l-)b>dGz79`G01~uD9Lgv{{gy_%J*slVucLJhsgvAHKR_A8g
z*HHy>iz(q%yK!#pG-E)>U^a+(qo1QjZr011J?!u6_TRvN(0<i#by@p9wtMmN4#`*N
zNqpF9&!4Q*YzrOHll)Pl>+1ua<?itNU%gA7qx1)hB?xu)r)TK;&V)}a+iuSfq=9Sb
z@2l+ZN&7SKzrqBIcL`B-dx5AgH3B;u8owNbtBiGhJMAMk09`mEEkn+<uRAAqci4aO
zOKQ3Os-Ln4O)@y5y8z;+{of!1^%6>6WetCvFFk(pwl1p&@z2Rw+drmui-iYJ!<QN)
z6MjLM67vXQB4j^;zq=SdUS&cLE?-8-enP~TWS`M@Jai;ggTZ5g<`etVOLTpQX(PGs
zBVy>Uv7aY*Fy;j2*2tfS!+lt!V7A1XWsUrSSg!y}IMg1{Rlq1*El}-b6BQ;dMW6yF
zLZC{=T-AjxRL@npmnKss%tcY-Wt<0z*;kd54OQ@KQ}9&JT!=U(C=}Y}rJIIsGj;aJ
z{$Zxzaa^Q*>cp`j<T-VMj^SnRH3c`DhVC$Rek9{QU5p4UiYd6ylyahNb_PCA{NURN
zW!K`jp4MnvX#^_W+6l^zyhwixq;N3d1IcqsgJ(){%-GUibsfG^hV#a-bGxE;Bd&Xb
zg`2tHO@G$>w&~Yd3Qk+JbGDWiyu#1vzGBq=JY-D!IPkh9*cUv1Y;WuNh<1HF$~0A<
z=U->il(iLKE|<2CO1mvsb-8?7w)t{hz!2FMoSjkd@){j3p}aOvEgz;!IOZ2VJ+kn)
zsd`=pOg+*~1+T3B&=UO7HdNl!)|z+4Ajx#|RnXL$v~z)hpRj<|P+D}xU$TXNll82*
z{qBYRnj3!~N9$G%T3Bc<cwM`m6c?7@E=$|Iy_U9rIb<o=hQ@Z57HsE@J&eYNjcK1)
zv?aT=#ixLarok3`gsLAhhO{@O03XQX_yr>kVv(jLjicSL1bx&_rS$byP<Nr@Rh|3L
z4^Uz1z_#n3ye$Q|nPzIAT!b@EruNZ6JPaK@IR=mt3x8P0!dnq;ZciofbFDxq;-n!?
z&lfp;bUtb8T(-YpYQx74n$n5Lh8|Z4a2rwANf01Br$DR66$1R62+)5S2WuEGWuGZH
zUpBSPO=FSf>gW?~7y;S*QdJ3JKY-Z<*A~1~i0<qOPw0!32Dh4nd8XjPG;?q%zR9*S
z<9Qum_{98N)XgYrW)2=V4?V=I1izuH^wgKf(u&DvJoTkHc!-tPX&!pm)Y;1_MG;aN
zv^3>4ORz~cW#ejotBy{s3aYusXjRp(l5SIY3Am9+FU2dEi3*y6TTEMVEie!mT!~70
zcqL_C$=}@E7kRmL3SLsSwie()zpbs)72j-z*ykyE>ALguXd3o6dQjC)qA5znPsA#I
zl2`nssr^2RX$sDkOS4}$2meK93Ld=M$f_N0583+*eV<z2rv1<RsY>{C<a$=2*mwg4
zoT)VUC59Yx`%5wmBc4xB9fX!efbiFDB3*|jlLnAShoOqwUo<Sd_uR4T=e1AQnu6EN
zj^Kl7tUpQVS<s1(-D}XSw$;1R+U_hbZZCS@h~dV2W$T%1arZrt7?qz*!86Ccr_#3|
z-nD@r#`wRut*E@D-Guag8sC+s){k1^&wH@=*O}IB{xO<kYP(V$-@EzPQ--d%kf)A4
z7LT9Ozdb!|u012?rF%m~e?Qq#+SxPQ;@oz-jM@kHTC%rUf}L^tS4(gQ4U6;J#$03B
z(CyDFE!gkR$H@zXo3(#_t+ewaeTlP+(wAiKvlMg&_Vf;hm=taA{C3SXrNML{?3&@G
zwx739t?83WA#m;`)MewhD8Ch?H&FSwp5*5YYgB&mw<~`<wLkkJ?ML~lW>c{nm1gh8
zgn0?gj?qIx`!A*ZR?u5Q`B3d}v|Ii!luw&=3FU86%HK`qRF^=4Q4!(Wv>$~fl>Y~%
zd<)7qy%c4EBlzl&sq?epX6IHSVN>vRb2i=YsK$S1Y3E_k|0u16=;mv>QoI13K6s3g
z4)6>309WAjf)BQHKG;@Ta5@0#CH0zOnv|wsQW`xDxwnrVazC4V-obgv=dj+y<U?!b
zA7ixX@!g04wtcot2Hm)DVm2lM+=m3uQu62wpBUK?#J;RU2G)iAdp)K|m`vshORF`R
zrh>4EE>6<6_oQM~)K>FxT4~#}nJp-YzQyY|1>dFOt{p-{)w7wL7f$*YBfLi=0y(9g
z^v|Wh&yWqp<m7YGpHG2rioi`G+v_PXECM;APx^6#)Vn=TArMb<@B=EAzGn!T>SjYn
zu)~CFCQbA?8*{Ke1x>C>(M7ffn=^19V`K|%1RJ7VPv?b8+a7s~*B%I`^Ax5S=^Xj`
zB(mulj!j_Rb$d<OxR`;BUaIlc6x?I#{6JIMHvLx9hKN7URPee#A7eJQS@2nAQ|G68
z=SfPReFDRHU|;VXj0+e5qx4tWuDR6|OgBc%x5C)(9@N3uy871P*y<s91cNokY7Euh
z@fgSb<9go(QsTx>8s6_v<rhui8_foH%mw}8UlSYNqtU%DAA8o&6?^^bvEN<_GBzf(
zA2jm1|Hccop9Y9nn}=&R{16zSHm9M@_rPy%`{OpW(lYJIOnPe9mm*MH|Ab4}!`l0%
zumJR~Y4AFdO)0RIL^bdiQNprWBun6S4H!_G+8{-Idns!N&!oCABe>N>v-M-BKQRp0
z2UnPZyr2kfH;<O^DVo8>?em!#?-Voa)-w#SX(+_3cdoSUN=w^axNdCvbtYWaB>R8J
zHZZd=jG2R<u@Cp+E}zzyanW}NZdvafjP?Y!wno-!ANn5{>Vth`{EO_&Xb<~#w5;lH
zb3v}QmCn?%4|2Vnj^w6-t%084>D~e)rVFIfEy26ejFH_&py|Q07@*84Z<bE0NXLCp
zG+MQrQY?e-%AiPOZx2!LhZ)5Mz1sGDfEEXPO~D<e?9Ir&eH9GDH=i#&xplDLVATF+
zU+ag1O`RWNCDRe3XT%HrdS!a?kaRpoYKCC<604i7fe%fsX9jEUTSFygcQXDQjA3UZ
zmVSkM@yB!=YK_7U{4pKEnxqhh8D|4WdTBhRi@{_|a9Ff0=!f-_0BAR1a6Jr0Urqg7
zrPaZwl}#zHn+smiZn%<)VxV?h!GSiLf+MMcd66B#J2S8!{NQoA6DYVZxZM<d#gx6(
z)P84#HEtprcbXb^>Rqv29BW_rG*ptON6tB3ip!VKXVkW1=IjpMyvHa<Rr6{u*1iTB
z<qJSp4}jE*gPVb9tV>V5W=?te<mSQt0yD@CC^ijVO`k8@j&GA0y5jg5^_hDd6}k);
zsvT#aI^P#KR*HoyZYuNUo52Nyp}}AlzCHk|jif0a3QU)#qj%22V>}LYca#=%tUbg=
z&9>$Y!c=<iGjxHXSP`~{OtA$Z*lY_Xpr6f9CZI>Y2NG-_y`hCoK!3vHxr1mDRzJfs
zG5dUV1Fe2&eR4--xvBL<;%o_{I9ngwv!!iLd27e{*3$;<{r^I=+<wKVwRlu()|pba
zMGl$T?<WBo=WR{TDRZmwc_uLr{Jl(!?-P;FZEcV2K~1=5LMBNe``=Guz9pITAdSjQ
zCOv2g#=dWD4tASxyOp_ZQicU{8syQ}NFLpSa17{gnmUi^o!ykwnB8qD*b;c9_m9Xo
zMjz&OO$OvqbK8Av5F=7`Vjyo_ospKp$9H&Xc&|igaohKcr(LJrFc*Gv+db*#X?KAu
z3;0_e-tx2?=zFPvW#BDCyWx8NHU)1}v>WI?0KjCt$=VGm{H+jgh1w0Y)&@+6H=TCF
z`~1y>H<Nb5>v(Hhg>}Sywhr*twnnb?+j^%H>Uod!hVU6EQ`?Ui_S#ygEyu>;{m1Z_
z#QT-;aMR#9RVJZC5Z#IiC02@Wg8F3$o1KQE;J_QI=_nT6yN@chD|F<GliRUi!#v!_
z=iz;(?7o+%?@NQ*h4fa|HYS}`4;3UmS4E}g*W*_Y&@%+~fM^d-RfzW8478?w^xid$
zX#Y?Wm701Mj0MpaUf&nFt$jG=kC8>KI7x^OoQL6`{4`n3g5O-g`dPanpI+iy&*=RN
z?f1_Z?=-j1cyB@qh`%)WnsGsC@GE1LF?e=?sa<~*!IkFhb672=_5ArdOkADk2K#R?
zbc^$C=HTb%;HSmG4~>yg@8U3IimBl3z(=f?mKYZsmxwiLRGy&nwxE;fNqJICdjIF|
zpBVTj2L6eGe`4UD82Bd!{)vHqV&I<`_%1MzB0<Zb`e6+cVL=GSazc7C{QFy^!ZCA{
zC=)08W=T>DbGvt$@!V9nd)KZR3Ow-{@Y(7dQ{^n5Y?)tNVz!nWt4#9U<rVWS<yA|D
zNhU{qy<2W*mR+_6hwSu8S-z>V!|Qc>r^*4B&0g=2{cgG5ZL5(R+%=B+VN!$5=~Bn?
zc-;*iKVnfdSDmcOs=>>0hs&2EH`rFnSvA?iq+!F5M4l?w+MM+cg3FE-He@E>?Qqq|
z?poRJSm{S~SQ)+vw-!yPt81v&v)5cZ3JkcGyWCAKX@#TOkGf6uI(!cA3P<$aYO#79
zHngSM>U06*7kHR;Q;~VrRpW03ep+Vzp!<!N5Z)s2Sy)>;q!8#{E8@N=Nl^1d)~<z%
zNRr2hsQ-|p^?06!J70vjGpSFsa5z;QNyqazo`@HDGCcm)f86?`$?#Kt9dIwf|B}@c
z%!6P2o?_{hH>vQULhetWrBDHsu<S>9(+{`Tvfq{BiM<>4yGlH<t;K$$#B;QGLS4*$
zSBvLm;(3jDLLI<<Si!L$jxV#{wc?3QHTJtsJTDhds43WQxOk2c&nv|9dhr|~p5GTw
z!WsQ;;1B7z_@#><*+0@x7I6MJK1+H1^+PLw>Cd9?O1Sc*exYCDlZRgDZ_6zAm910B
zK`}I|>BY6Pns(isPyI<@N}kj|gnPm_^%3Eha7z6|eMJ33eM0>}eL#38ToaxNw}el^
zA>oa1Mff3{5dI10glEDn;gfJkcq3dX{j1Q0(yvN?A`1H{?N;bWp%aeP<bO|J<d*`B
zUg%$9dlK=Hh_6I^B;qR(AJO)u@ZTWu8!Ue3&wYLBvv*%wo9s#HsSbo6!Xe>@a7g$e
z94d6I(62(b3cV_PqVR{p7YaWp{e!I7Pw9umcO^XWSu~?BatQ8u<F|94&+Lonp|2lT
z*cWLm>WgS*_eI{B^X=S_;=V{R+%2$l+c38;@@u#Z)3<ZmkoG;e^KiY$V<-$v|MgqC
zX-KyK&I7mPw!TO++?j9bHX+?haPPo%!i?!{xLWhKbC)AsHk=9WWJh1*na;k*3Al$g
z_eCDv(iiE5n;GhhyaLx0hA!;ozDNKrNbwGE_dmq@v#+5paGUX7|7u?(cWYl{$m@NP
z-{5)QHlF`acJxKQgu8xcU!)SwgEC*h^QVaaBHRe%OYVJyeTetPyMS}J3CL$5+&Z{F
z!tIAU3->Ny=_s=R?mEOLw{dr0#DcnA$Q?r*JJPhl{RMga`6XCm?BUO?2m2zQz&Y@K
z)f>ngt_ObeTYZtcgnJv$%ivPQll&KR6j$xscqgZ%qx=<rvZp$|+ONddybV0TDer%P
z|6{nz--(4SQTdg4^qi`4F2wl>T=%=+r$}F96Wm_7<8UvupNV_`ch5s-B9Foye3-j8
z;s5@|Gm*t`6t?Q$&O}=9d=AgEa5Le*@#ozA3{P@3Do3zwaMS+%OyrGUpNaSmo{6-=
z{ZF{3;kLoO3wILkiZ{+gu7}gZ&49DOErDx*TLae)_e;2E;X-f+;Eur!dh<-=TDUwo
z%Aef#-;Q^40lNooHQWPmkHS3(_X6C0xcA_`fE(O>CUO<rO>joI+u>|*kHBq)I}bPF
zE#wWC3s(fU1a1}FV{k9SeFS$k%9r78g?k3>=K1F$GjBf^F;tz4l-+SI^31|>k)?~y
zMXtZ=TqJk#xrk&t7uk3Zp5mvB-^#qH$XSX!rNmR7iae#rP2&VWC!ST}=@QS4;`x+#
zl7%My$UK^U%6y=duhfS_Ih^9i!YSd(yYgfiQN#qNeidmcDCtiDDo>?e%2TOd;!}xF
z^MB%xV3b~wNfbEEt%@8qkbFe(D6Nt=$z<fN`Rl$ZB&(&pKb?t}YPgbyp7c{BAN~e;
zJ>hqQ_$hE9W++)obLU#jR%2ylNktW{qNZlm&|{cH>yzAS$th2lS(K_p<t5hH^B0s=
zF_6_2Xs}@^;`C3I%NAISvaKHLlbU9kf}FN`=PF0dFlmCX**BravBFvH@R8r|Xqe!_
zGK>Am3P|1+@=Q5TzVSv`7jXHUbuLFuHo~UMKBRZo>O408vg`qKsrCdWz#)%|l%d4J
zGOn6MUGDI@9Q6}e{q44To2%OCs+01tR@2MWXsg@)LoC9k%Km1LL!N>4T@Ds-wLXWR
zURDKiSmhMI3+p=aduRbTRpx1Zz=B-ucG02_ps5_2R*%=|@{8>GLR7EU%d|kmbC}e`
z3LHD5AHJ*oEA=xQ91T``ptjcGl@X<Ef!QqSbOB`O@q4ZQY+08OD?2-j^_!){Vl|pe
z=a$JcX2^MxAL~}PS3lEbzdLUU1qn<eqbbPdLU6+^7l|-2|3VQMZfiXk&ZQPBaaAqc
z9;za1$NX~Zf-+;pB3ftbRyZ9^*_;lYK5K)`?{%)!*+fXS&4snH$6@p9Y&o*Mf2ci2
zR>y4!)KeBB@>o_+<m&g@Tt1K6=ctLx*&XnsBi#s$<&D_X+s?XxJK&4Pjt^Dp0orpk
zI9z^;M~!2}#R3tH8i?ipGMfk7Aeb#7k;{>n6mXNvVYna=5%Mn@0g)z3wv|p_o;(wD
zLoCYQ#Bg9SF0>YxS}Iu_LtGr(mBza)E>3=2ocsafOpJ>&alklvME+D2gQ%W(&5wWO
zl{YAnIW+ppq4^4QBEz}I;cW=`3Fp)UvF;y$&<Gt0op6nf&Yn$ycKu8~Y^b@obfUm_
zj8<`ZBMo{sFrq^+VzuyROFY!T*wY}ntR9=s&oCqDNMR)3C{JRf!+K>LtB2R^aairP
z>gBp@xz_EKXD_q)=bA0D%k7t)4IWN%P)s#D{Ns4v$La!-Q<sP)piYJ*A~2s-g9qj>
z;c1D(6s}pGBR8s8ObNE<F(zA~YM@2pZ@gi}QjH3xym45Ww4A|s*+j1VeIlz1Q6|qR
z2IucFIL|u<=kGB%Z!mucAq~`PbCh1Qc^y_Cw1-y4k_mk8tHu=K)m6tv&N?r2iC}Ct
zMAy}#Ay%65d^FQZ<kQded8^Ustcgk$VgLgN@VcY>qbUSHNm1>=+$FKUREB>U&A+@9
z5!mGRF1Oa${5DlAn{T=AZs!tK8E-Y!Gf&FW8*cH*SyNV$^0Ura9Sy(?zQW=4Io&R)
z-eIftvC3fP;_>SP!z9vj;+>M~4f&MG0FLFJ<j*jP1Pf_?xm3o6ld_ULsjS2Rmk&2l
zDl5W!5#EdNUNliEhGB$MX0((@)j+gonSLg#8JeV@89!T;%`Y1$632bCgZk$}5d<01
zBr6*C&09#AP#Qf<GFHr$%BqZ&^D3p{$||X{Vz$&|$3UZ28H0yivLbs~S@W!pl^!>g
zz3~xiosb#$Ob#yd)(I(vNQHTaNnShWRGtY&fc}|OgI@d&HjF8<3YTGGI&7jSRv{|h
z=|W#Sp(ZC1ks&_W9o`(d4g*Y99+8eZj-VP5i^E9AjuD8u3o_MIc|2ife2$zou|}?O
zHb9%6Rj?``+nfCkh@X_3NWq$Q@e~T1k;KM?Ib=LlzFW?kTn`B@_IHWs7L4(Tku5`t
zB&6M%JF!-tAf8m4CTQCS01U%`%t%gP8d77wf>6X1v;h-Jt<~@L+v=_GTHPLNz;`#b
z8pBw$pkY$2N2+nMj*{of|AMN^E=LoDkL8Hv_XZr(D2U!`oL&T(D-h63+Zs($o!8+2
zZ#wIi`R(-qht#yp>31OZxGc{&G~MTK#{R|(xz<^aU5RPzS?#XJ{DAU((_}WF!S8ic
zyS;$d+w6||8RLL_;KrA|TEbSx-B}EM^)*Z6)$;U}Gqb?|s!(874Wc-8h%;`QjNJ~8
z)3yA+TpN@r&7NOYS%rR8)kZ>5Iw>pH;E=$vfJtjO)zp9@^;I71)Swx9H~!|*3#MDa
zU-T*@a?Tz#fd-Ei$t1VOugC1!Wb@YKD(^I@Mgx3)n-}yh-#%}WHIEOT>L5d6(B%HH
zCR_W(n$ka(b)FcSlQ5p(xw^VQEn9?HYq3jZML#>xdFnBRvG6kRvBL{Wv(@_`KHMme
zQ<#~A73%MPNtQdB_2Bn_L-yF5UcrzFBLg3SFfi1*flr6a;kDK0VvMMBk$}r`6;Z}#
z`NoUEfHuPvX%<vg4-Rztn4DshAw2k<5LPIT9a~+?`BDozZrEZQppqhYn&Qwn!!1%H
zx`qB2q(-2g{=f(H(jWNpt@EVDiS%a@{h3UEZlOO@@W(ktav}>Srv2gt<&v?un4Cc}
z&Ymq*E+~>L3(Qh+=^aw#?G@yzB=h{a09eRbE9T!>DOpO4Wi-%3?v*M^%1ew@;$bZ^
z&YmY#mYNrEUxlU8S~kDJA}Jmoj1sICB$~gV%8CulVySYr(Oe=`mQ+cz%NG$7P==Kz
zBPPr{N=xpP%1bIN3#v+_stRLSW%>Nd5@~Kp6|alp;W-zV8t0<&#<|G9a-OxcY>w*f
zu9AxRQsw+PRRAuGdPGg<qpS-2;jdUo7L`h~%|=T(Rp$<?p_o7Oi>31U^Ud_mo_L>Q
zHddOf#;U3^;hi%_6~^<sv$PBww!kHM7L+ls#i*?G$X8U#7JcNMy`ZvczQrKL{P`;X
zL}_-#{K`seNm&)@yU1$B%K$;VT$kJBkXGK@EYU(zw=&0N&*4k_?3-sa>u0W%R_dFj
zl{djnfV(GOI`ZTHxl*ck*Xi)#%sv0HlALup{hi>FFacv9m>^=+8TE0A9IrlSeIWYc
zZSX-N^fs^*!XE`LylD{%ULiH>7tiFC%SL4u&uo@xg~<!yawxJDD;5fkE?4NK&f&Lu
z-Az8VKUzYK)5e-XPpo@=%Q1k|Qp{MuDu>riAwGAlA5mD_6n``yT5?dN*wO<p;C{Q4
zLaXa-4IbukyH;2YHBoOq^LpIwdJ)RJJhZmn=38d9`TZ`6&fjWT-C~ev`Atq*8F?8_
z$RBViZ`jxOHnSYiwautYz~^@}bSpk}Xijs^D#Y=+eLgFOEi|XuT8|fzwqLlY0A(RI
z?V<a^U}f_|PUfZsk}|wT2Sb^Sr3m*yh2yMtq97R>9k;icjcYXFqS+1@8bCyX^?=CH
zA!pr^n_nwiiZB4Og&xb%QG;HgGFf6;<2hXJK;1Ig=do48%E2Z3mpNo#wb$maUM9+s
z3D2_5<#3P`#`IXW6|h!dCJ=a43V`fk<>RWA7LKt6A17{3S1Jt(){$Zjq-Z!0;<44B
z6zX6_(W2IH$kY3&JrHfcrE_!<7E2Gc2{aM|t4*rWc{F)IcmZml6T`dWb4B$^Q3V=S
zO|%}Osg~)O%B)O6r*TkTy;@elv~W|_2_%gA?vB2RrJY{pD+Z3ZSY#n7s`MdcFjA>x
zt0^l)mc)GeMWo^=B1D|L*dmn|!{4JCJIFFv<to~TONfFz4f>g}WW1}<dn#QPFPGxP
zv>+<5QVNeCw4_<GOUOnpy8*&d>@7d;ZKCQ8I#kj!0qM+VxFBIu1)y0S@B0>{%@wo>
zs#BG2^=4AxTEPTZPD1IMM2iZ6&;_mWRN2Fq-BxIaZD4HZzMHKs4=G`oAy7<}jw}_u
zUyTpyS~NF-gbCFHBALa9(8l8T?;Z@(am0b>f-+cwm^PHNF7-N8*i;xvf-FvghJ%1P
zG8HSi?5rz=l2~YG5ja3-@|>v1ODQZ`^C&~IO@}#<%Hr$6ic(a`3AKrXQOPl$i0xz0
zu%e`kg;ti{RdP2m&=LR*12mCpfw2ofiom^;5ww{UIx)&YjR70f!Po-P5iWTRFu^K4
zL+y(JqN&6<%FrbVX1bY%Y+jc<c@|h?DprSvJiQ_NCaMhw^A$x5Y~Wyn0Oxb&il)xz
z;Cum|m;j#0!K8j=$aJwKxsK7pCDrO8)ylg_z49(nvAo2Z5h@`J*}S#}pMIv(zl=4A
zqdRe*Ly$1Xei9(G>y!gJ*}*W2ck(kd!jB$zFr2|Jx;(~YtiuKKFzl&7z@Xr~XqU&v
zQfPXr9Ui|Gst~I|mu$q1kHs6#iisN+;$yLnW03~rb5&6T!YTQMM+#1`0&4<U0&xZv
z;tVRp872aW=xCrBNG1S!mA^0u_vb-js&s)t?0CdAG=V6c4S9kqoDBx}h|ru3`S1}*
zIU6Rz2jWqNUKL#rgdS)Elwq(xigg4PEFD$hyQ;-loJd$KeRRNb@Zl3}rgk%0Lo><G
zu*Q4>HEPVnhL;%C5V;w|tkJ*sFwnEkcM(<j*dRq^^0zFRj|&XNH$&KA+YGD^C=Tfi
zcrWBjHG&^Jd`x6R8f)|fz|@0UY%FCXEHzFPV~}{_JZU-4Bv112VM|)hp4iU;0;t4d
zRwH=Iq;kIP;j%aKM0{5`9-DTN;efCe$<sM_Ix1}is{wrR<wU2uxDukFad|7Dya_Ic
zO+~sk6%*Ldqjp>JTG`|0)`X@enU$(J5s6tkAx-swGnqq_L(LbtT9`LVf+AB2t>%PD
zaYb(28;#2egaT_=SXp#{6LT6P!~UvtEJW%PoZ%KIfuSVr|JMUWYd$nlT)jD!w0kp|
zrjxvPLl2H6FQ;0_|4?oU<tO=>ViStaaf)V=R$5s7I~-CCmYfa<*ly^ylV#m6IYTi<
zBGI~92udFHZY2~MM(mB=X{;zKEt@-4o^WC71JZywm@zsu)&6C8TW0ggcIb8?jWRoM
zwkX7U47xa()XGpFz;MAK%>bLstuU6Gz&JDL?UH_qEeL<M%!=9Ilij@}ErsiVlVK&&
z0(S_m5C$%rgptcq7`k-8$uN>>fjb0O2;<XDaOFrxwm|fm5D(m5xICoW2zMN=9O<5d
zOGg?H++Mgmq}vF094_&v*r_5LNfa+3i9}HqTVYB=F!~XUerlMs1_2a-jCYWcjECYb
z1S>>5{wtq#_K6c+7JE72rv--E5uA`-OR#sJO2cL4^y^q#Tl+h}Tjt}%%7XCc%fG=k
z>$(1J+yG(fc#|#ySI4~+{~YA^$Oq>l55m#&e&PS?Q65ImXAt%z96f(4{QvuX9!5`U
z6SbY;68NM(VYEYl#2-7;!RaXughZ51`H}ydIO;}nAR*S*vAkLk5~Bl9BYUTK{MW%A
z3Vq;(C5@3>3ww{z4S$d2&rIE;bcXcIR63SFH?>yf&rHQ)`7={^HC?hJ4rP`8TlgZ<
zV6ayDu_ed!E~RCa)+yFf*9x38@s?KlC;J0XjF;YampQ+n9=bIft<tg0=8r0CuqO7p
zZ1olM47c(Kw0$O^6$S&_D&Pi0PG`x)Olw{N8^vh7Q@*UcwxNXmq;Fjxc9zAetbCwU
zRO*8qZ?#&hSFSV|4Ec7O&slBt!C=Q#S6W@^HyEmy*<irrwK@I1(&~~*8~&7(R2r@2
z^YTSQ5_0B(K5K%gGwWFKN9nKst|!qR{gkKD|4P5_g28<{jP0kv&4;UoTMD-f&Ih*^
zt_`jphWO;3ga6lXe}W6c?Sp$C?hM=rz{kPe3O5gKDcn@JTi^=dX22QYO5o^6ZB}@L
zo{AgD4w3BpmGc_P83z=}&TSB_=ot~^R(m|sSS%458f>r{!)90Ya$hc+J$0m5Q|}uQ
zy(x#NR20a2FMfY&`3h^bt$G<q9U7CUwuD)(_#E}MdeVhr0S$dBi#AN+`H^l?D1WR?
zSm#+;-mDtSSICr-Zy3?`=B-?Bnz(PI&?nc-@NDDhi)Cl?c<P%$qS%A>+GJae4T@Q>
zOpU<4HTJUfGPu#^b6}U)>6deJbCoKxgwP**ZLux6C@&-+jO|BKcD#~yppSO&_zpV{
zgH>d}L!0-|*)o{eS?5qMH54L)t=_ZDCj2ClP{jcO26#N`843gQA$O=0>tjtBCgoCX
zS=Ge3IM%}GrX0fnSM(`$Vn@}40RqPdRL*Kl@ra5RoF|rPn~ob*j?bzQZC1BWiRQ%a
zA=#XuX}~IBl0$e!jr>uhuCPeScNw7#uXP7tAb~o?8q8}H)!PMpgb!O>rDY;1gefE<
zh#B3{j2&_x_9iJysL{O?r5>uj9611l+GK1!$_TBUn6E04XQ$Uw3-nYG-WX#teH6tj
zN(LlObGoW=5)s=B?9d{!<0Qigjx5@-hbV1ByJ=5#8pL6T-0Tj>z5qOan1gY_QXyq@
zFWHGPVUtnBtYK1-Q6lPbDEKZiG65ITvDXTPKFAm*N7WAghQHOGW=XEWRvOACZfo%M
z+oomg5Qk+o;|;_tg4H=s4j7@y>!jg<mb2KKveK$Y^1AA*K3^_rT``<MC}vyT60?cL
zwjJh{+-e&(N6={;&I2(Dd3YoT%voxPNfP2k!+ALvxM-Nk#YCD*1yfNJMs;wpy>@l2
z7(b!)ocnwZA0unxV~z#tRXSG+S-x4EDOJW1cES|7CXRD5|FN;wkIpAPpTd!^B;{tt
z0HQT8)<7BT2NO2hT2DwsCNGm~WG@&Iv2;paw8-E!f`JFipC5T;;Xas)xRfn<aXgj`
z!II=HQhtlPdBxdL+9Y{@`+D=dnRQNL-muoXV<8YrCMM*Co`gLo8&p2)9YvPX7{Gc*
zSs8mRE13^dYs9axSm(el9$qN0XR~KHjGymdL3bJ}E!N6XOEHBPVW*s)Fv5otOqTAb
zWTiuMNLzctPRr-7nL3r*nqoQYt)u=~W^+YZ0Mqf<gy~s1OH5~nZLT_+ZQ03TEL>sN
z3^V-|4l7AtwC#d`*w|yn2EWldx1xj_{hN#xFjuIOqN8Wb4x}OAVs;u_u;f%7PxXwE
zW8@yq!{+k^h(hIC^NPfrM1v#_OS|N`w*yHuwJ9XcM*|}Igt?T~!G6}2h%T&DT-FuX
zjkQxK0%4>`Ea0+YEZ}{v4g`DGD9NKS$qc1mAk!>Rs5&_XS|NJG1=ib|(K!h*z355y
zQzx%$v?T_5$uhv|@PHc;6V^>i^=oT+R8>*6UPmK#qcH{XxMH?v!E7L>V3;~l`)k?8
zV6+C{bY-w+xaD#aq63F)PGA}aOn5$+=3oQ^^>{IrD58kY=Wxi&{C<yb>VydxU6%#y
zxtN@AsNLbMv$@<WCPbyj37$ZG{RG3LyvZ<9Al1fqo%7_pDO?hz<%~?d%m-$b2*}kn
zYEe_dpk$4X?Y`V52ied`Fz)vR{K^~aNMEj$8x=>H-G%~=1?M(34ivb2puiOa1-hew
zwY6l&=xvAvV8h-DoPlv_a+?fE!t;}aPgKH*2C&TuL8dX_^n%TZHiV2U>u!$uY2e2=
z;&)&!A5VxIFYDZpEEJPYCM3c3kk1a7p`y|d6FxvKSSolFT4Ac`S{Vw~{+SFA^@n~I
z)dP{`2b^T;>L@H<*6}db**t8btgA;gX_t~LHA&&z;P(0Jn`PQC6yDp(AS~uLsQuWB
zabp5i$0nmOB{tcPDSk2`Q~YE^rubn?hNVCy8Dk`bSsXRaKtn>Hna1sC7;Mn`CFWrX
ziD9SKFCq_1h)A|<!X}08rog2a7<I@BT)MSLGE6LzCgv4Mw@faQFy>0R;8ra20)A(G
z?lM>?L6}iada#%2a%mYi38PoceEh7C1W6F<4sPF0`$C-C$c~&K15OfbihzxYo_1Vh
zaf(T2tEDv+`gMg9$7axP6pdyC#N=kc*=*j(@<~-*I#@<CA(PuH%%zr6)XhA9_B>fP
zK2Q%?2S%SDRj5t9Znr-h>oL{rSL3L)VNAC|oWyA>2}5R72QhUjW=DuK&`1QT{jgC4
zDG}C~tuvQJX$_14lvZZ6_{8-fU>}(7K$c=cxQ|c7u++w}F$l}`{#>qEVbFwE{Y)|_
zr88g3K`*imEGa9t&MsfTN9yPmB2gnCZYjhBc1$*6NgTI^miTgdxp=OSl(iQh(Rl6I
z6hM>e-C2`X1t!p+CD5tEv>i6gFiVE+SsZyIKPL%{5>v=B+EPT9oK*2u2X#5D+iPUH
z{e-PF2xT?S6;2;poY<QYo)a5uXr-}S&(>8w^b9x-iz6V~?w|z}R<_DYL@eYWr}#0m
zA}aCkrN&*LS}Ti;7;QSRZVHe#%7LcS_{a9<iO|@3o?`_^_Izl8kIbQJBvl+3V_|zC
zH&=}hW|2j^4_q|16hTnL)P^&IQr2X`7VAjW#7cuLlcEWY11TNi#Tr&TXRNrCG>D6G
zqVwxi*=Jipjp4<y>SB)sJu^9=<^e7NLg!Mwe5!1<x?~-U#bF(WW|Q?7@jgw@(D9&i
zu||pqOA40M3wG)@QuC2)ATY*^;2>?e<r)Ri+%7I^YX%BsVr>-em;kOKO{Sty5Jh1#
zNm$Ha50O~|vt<?yJ$dSefide9$yS9GW!1&e!g`v78&wQJBTQ@HTjma6k|Qz(QDz17
z(S<e6w*;`Vt9IA9fEeZ21?LnJF5Rd)HvgIw<7K8b%Brsq$TwjnULTOybQm)SiVctr
zwi+;uLzkPIBU_BccUbQ%HJi&yN{ZvHhoTa5zxWl#vf}xc_}EF>hhA=QV)>~MJ;O>e
zWYFeQlb7x~VwR@UWMyarCx{atJ{s0s4z>p73u;5cK&MO*;5csF1?zJRp=={d7~<3D
z&cGN7^#(~dR;&~;sgPWrP3sa}ENm<bl*W!#7U7Wfuu)B+(cS80TsfSJ6QHo?hr$Bq
zFX*Z!iOF&x6AxL_DHgbx)#F7w1o#^~%jr*}G_N|~m2Rg$cwHggLB4snONI%)Mp&Jy
z3eHBbqA7uoY-%xvP?)fvh5rt6)&_{j45WBfwsyb->6oldZmKtorK7bp;6`^Y6-MJ2
zFgk;_m|gKooOb0YC-Rimo1-E5v0UKg(MY^h{Cs*ym=VP^Nu!OGR*DH@0nUzHXrzGj
zH+8DbpOB|7V8;}j5QqZ<J_mHpU+u<6i}w82-Hi!(=*9c3c&2=OKJvZqz!fCqF<ZRn
z<Jkar|99Yin~=w!#rtD;z6$rrci@J85}%J&ybs2+>-71^NjPKQ`A8L99oz$OzkmzD
z4Lx%{@&TUcI>~yGAC9(P@Ryi8G4epi`*CDTn&5YPaFnU8d4ghs3DE=Nn@`f?za9(J
z8cFYmPaKSdkE^OAy(8+^GqEY*RZr5q5}U`+d)<=07J(!MN<vMZFrG01k!$nydKU`A
z3ZR-*!0pj5m-N+0PWniiU*W~tAAiB)dg4btR{Zpekx@_Uccw6o3t$U+gm76<4_z}}
zfCIgUp4gG02aagsZ?ne<H7<PIWLpobdcWQkdxsEgtB3gmnnU6c<fCSNEDRO*A;_T$
zYH-2|9&19RP`vtjulkM7`=EKXb$XYh4m17=Bxm>LiBM5{J6`Q%l&)Z5xx+yfuU>{_
z3+mngKcAZPgdHf^ZPX%Cjf;msj1#YHhtXC|h2br_n2(2`$Gkq&4I(82a5*1dcGd_4
z;{XV2euVhgL211cU9D%Q<R})cN${-m1b{~9b&!jzj&SQzhehXT5+F!R&t^KdRGwSD
z02b+F9#o?&igV=n<T#kd_PP|A7ze1Zx`416&6d<HbT6S?QDQ8%l;k#WnVC*rkphf1
zOYonWkkAxQWiq9r(AW&4mLNG*8Az^+WgHb3<R(n|aW%)(AXf({n_Rh6XoCEtC*(3d
z1WmdU1BbS#UMLaLIuRnfUc%?W_6XR|_nnrc2=$aXD-C^!cKP*<XjDVvcT^T_=^$fF
z<9=(S%x?>!RSS_}lYRxZLM46uitnf}LLst+08JTVHH6D7Sy?VOnuN^>bR6k8@T3b%
z@DK-r^?JyicHB-8<ul?(tffJsu`uwk7fSz?*!BQ)84iuwd`-4GwO`MUKp=!uo!;Rl
zMnUmxM8H_3cY)lA;@Oso-p*KZ8P9_1hlNFryCGT)uoI1awa%4#R1fgzJNL5#Fe~8~
zo60d?+~68LTH~z?&}l-IYB_cknF)Iiwlj;R2$U=Rqb7{wq$wgEjj1w0ysCkOt)l3b
zgke(bV$r)ffBKd{+Gm{V!;LPotY1cq45WI5W|lp$l7McE6rBvxfK4CtDbtNnKd^*O
zN~kWCQerCuHR)FwmOveuk7&sgu%g(uG+&=#aFe~H-IQ%jwq{;F-CasG5Lypb@o4!x
zkj4}mig7a!7=jID#(M0af#GQbnY711LE<pVaw5G&laK8_Vw9lmI-KZ)rU%myV=aEh
zlg3`GGoU=Ckp>zO8o3;>3Lq?uUyKc6<nNXFO|Dekc;XfSiZ$?W8Rj^dpGsu|B=!lE
zys#t*c=-)IUL1-_g7ty3)nWZgYeXNkJ<eJu7BpNHM0+uAOq)!Z!6p~lZR3-mkMvQ5
zEq?eqK4TV!b)YO%H9#5k5Po1I;PfpMIj~MeN6><sI+U$vV~t~0k$Rt8iPa;ailiU*
zBFhzJb1_ikRVTDyQ?b_K8`qq3Ih1H+x}3**)i0M(E<$pW%UFf^iGMm>rG%rVL>n~$
zGUj2EM*u<s5l*Zuc61j__e7DeaO6chQ0y$yM8l3<$C}H@F;)n(N8we+UP%8Ra>4~v
zJ_VN!NFyH5o{tMZ@aY!2a^%Ih&R{Vt^cUk&g2m(D77H5Z72%SgJcllykcC3c&6ROV
z#!PR{2%wf|ohj12Fx*n%gHE0fkS)oaRBtCNSQNs7e%iJIl~Z*)ZSTj>0L7zH0a~$R
zbZ2DNzx@;vKk6}ljvs?S0B#%`<SHF6m#c(i!pTbv@llDGl@niKRF$-$3f&9r*g>#0
zZa}CG0T1jza7K}4aAJwL>4Rk{i~eSHw9(B6m|vLgg?EHV7@b#Wo)HQ8#3L8YuO!sO
z5QpIoplV3uAXKW~Qfj-Hr<S5TqF`R1Xs}#aRRQB&%9+C{OHqI-=L^sx6~U*w0rL_d
zt_Huf9BVoG5p%NT1|!DgL!jUw7IgqN;)rx{W{<8|7`v;Oo#{_R+i69kB4$$Z>S6GW
zL5ZJz<CI;^b5s^PeEeXlST=aXsDq#5i_4JbVCOl+mI{;yd*fPB8m&a*@F3o%j*Sfn
z%BVvdtZXP_9zi5^UUvXjYe66Fz_vnE0R)o~x_VpIRh3xE$BrG##<nSpo6TmM#fVKP
zY@0wa8XZTKQfRl9m7-qFp~iRx$$J}QiV&Br!J(h5&N(huZ2F23f|hs7Ir>E1Wpo6E
zF=LLzHs|1Y*o>$WTa|}rfIP58M+PAYdE^>ub7VvAK>5_7k<nUk5B5UHk_?LzB`@$9
z9m8TW*(x=vE|4)_uEu4KRhO+EJ3?Y9J7D(eQ575I$?6@q1{Q_Tf>_2O=_qN_29Oh1
zWUxL_1YdRaLFj6n2(kx-Go7?RwJ1s~rFmJ=%s4PQ=dgK{m#-qI<jugq6Spd-+(nk$
z`cH`f3mvw)SJ$3pn>SFe_EP{L-?(g}5x!Zzikd(moRJHLIN2gJ3DF*9V^9KFiPtLJ
z=QJ-zE(=uGI|3NNtBBg>5Vg@XXci<=CCG=HH+g+*J4&peX*xxzw2?G$5wLhtadToH
z)5xY&IEPM4Mc?H7Tp=o{6(Z5bDv9W_BX&?2lbGM3<_#PdRu_#95;UYBPER)k2!4xZ
z2Qdg&J3v-uw-l)J^V<Rg=2tIQ&ByMdwX|4w^8|f%QK_+#CXd)su0K3Et|0i6#LH*+
zdgpRSy>pq{%|CBIq%4!-*v5oGh}1OAl0u>0K_X#J@@$>Cj$E3MRh2SK&YLvxX2VjN
zcErA~LmsHa@iP0s!}tqik0E`4%=J9;8VB1{zHsJL_2_z)j@wu)#tKGIMP+mJR4rL7
zW~sh>GntekYDtY*%y1fIw}XlVz6ltwy`t|GGwayY9+NYfZJ|;MN&mc318CfYTrDR@
zxT2y{rchLxc#|eotT}4vU{S7$npDy?)9`WAqS#yj>B!-#gXt`0cO|_lc~;`ctc;d%
zx=v2(74Gs9%=1(kMKh_+DaIx-MQN-yEO9Iy0cdtpru7)3*NGa$G#;_hfrrs@l$t|L
zgIb176l~NX0|2KBw<0NhOJWD<4T&w{NH`zTpTgAf+7w<`jhl(X(fI+k?X)81W%169
zsd{7eV;u!MW3fd}m8`{fI`y5f62Q$(H}jjDVq=@KvCSGD%S`Pf*_^gH8=!5$-5paQ
zu?MghPD(*+@!}p^kv;VbGa#Ztm`r$JlM(Q;TSTIq3fjkhFj+|RhLCz%8O)S2DGd8R
zScpT#5bK**7spgrN)lPDF(E)k)mk#`zjCz!W@8BBSS&QL$u+tP;dO{Azhb#Xv!gM@
zRtd!z?o1)O4wffhcgXyPkpIKp`@qLBt^ea^cXrclvfZR<+N2uowoThb8Yyi>+i23J
zZAuVS1$6~cK@kMG_ZFd5K@miyTEtatPy|sy5mbd1K~+!$p+T=3RBisw@BN(fOtLeX
zrPt^4eZ79K*Y`W^<~{R1&w0){&-pv&oSBjL+q8jQ)=io{LXH=xvHwKuyQ2o8Hv~1M
zCKDVoj+Sp4;i`t`6lwb5BHYST5v8Wm8ix>@$d|KF2sB%6wPQ#Ky~7zPC!<W7`Hjp*
zZ4Xgw4Bs)<;wqsg;>}^uM{fE;VF@IJN8X{i*I(g9G=Uysayms}#XO069W=F=GP-hB
z332tKd8O`>?nLDVS4>zI;YD)i(G(Jy3<Dt=?JJw}G^r^7u321rXfvA99k^W0w92V?
zx+hf|`wOl>WQs6Hhxr#Y&5PAts=T3)k!Q=m0=!wXFEF}ypl<`wtRU`C<=hALj+kmf
zVJM<GdAY?*z93trSKdnZX$U<-&@=~U%!tW3M5LB?M<wHynKo;L)KRis0a`*?e%PFp
z5aYJ{To{^m*r+0Q{vQNQ>kIq2ofEqVuE5K@Crv*52$cnLemXQAI9ElR)+|cSzvzmK
zjJ88gsVZM;8r$rRIb+~pOeUe%D=MV4PEe~Aa#NzxVx9w6+T`|kqp+W?Y+$t3mt3s;
z^)B^<M1WdWxiic^|9IJ+o_#K|SCu*B?Ahm2dqxEmOW9oX(K=EqD7T-XHK3mV5;U1A
z3|h1@W{ZqqP?B+1lXJ<5cwag{QxwvBaAD>G*ZbT^h!QpZjGgwXM8hT&YfS!}#l0r#
z2$@eURgsjMqQr(QqC*3gk-VsHK0#^xlL-YSL~mset@Sh)d^}BkUQvV+k7s=FT!L6o
zc&^B#9e&~@oY7u_e#@q^A}W<?-Bji;Hffw+tR9{^5AVZJ8#?S2@YZ;vhmsRUq55n?
zRe2zDhw~s(;Qe4%ETO;W)1Sq(t}R5yG4zP9CNk(;K5r^SDP6qQrOy=jcsSO-m(HR^
z@R|5ULuQne6|<V*j^p)faqYwjJsU{RmC|!~#xr_2m50lEfHnM(zf?Y=Er-F^^P#l$
zeGkA8H;SHthCf5TnGfUIc#hF{e$tR9eROW<HO>wH1|os-pHox+TyFI*Vu1SR8L0k6
zMyY?^LF%6`TKzNPVPMyv9G1@=^t&|d6Aq1V45W0Me#QqA!l3cJrf_J@UhB2zPY&bz
zVd2oo*KnBeCr9a@95esqFuqe24vq3)piwRiG|Gp8MmaIiC@%&Y<;Fmx{1|AIBLj_a
z42&IO{%3@1VBApiKO-In4vRPcGvZ}n!f^9HBc28xFv9%Lh_``>2b%vG=`e8QDDyw6
zuhd=`_4GHqo#4-Y`Pb|EAN|OGsJo$WDg4p-65vYw2<c}hYv?dAROZjl$q%*NbWdXU
zW<z5KdFaaw<ug&3)>H3|?cHuFH@Eh_I}w9XJj%EPAYV{It1g$EgRh*79`naRjiE;j
zE}=K{qhmT3Z`#KN-Wxar`!@5=oO|LNydIVIqMsz+txL<9bX!I<RZ4ISwJ9UC=}YkO
zog8FV!hJ!#3g=}H2lWPEH82-%184%)l%k(lAJogh2Z1ZW7d{@;n+a=x-~&$w^#oj3
z0{!@qPc?8dumPA&*Utp?MZl(KgL)CL6}TLjwkfEu1(pG8ftA1pU^Q?T(DPhS4*~-(
z1oeD;<Sz|a49o+T0ULldz*gWEU=Y|u?k@)QJ-|vJeHgGB7zeBa27s-=8NeVg7wFj>
z)C+(Kz!G2pxDuEKtOAw+Hv=nyjlgPP3ozj&#1EJT^y1@#g+M<r?KPwm*xCR+K+jvy
zLp-n+m<?<I<^gvB3xRFG5}*qoI!bsO_5=oi$-ugIVLxCKuo&oh4|;)Vz;(bv;AUVe
zu!V3(P*2524hug*`ha<#!p^`FU=1(`+zzbX1^eJbh=pGuzkmtNLA?;z^cCcRt-u;!
z5V!@H_BHYwm<`+uECI&gBZ`4<;2&5CTm%dPD}aS9uotiu7$hF(O@yA^K|K$c4O{}u
z1C|04z7Oj4zyNRuxdZnS?g{D{_()?RFdtX~EC#mz8}h&)a0k%yQ&4XOX15_e_)<|G
zFdJA1Tm)?TIjC0yTY;O&{TIZGc%Xk2{QU|!U;}VDFl{gN1M7gTbp0F3k01Vlg}{=1
z$OmBd@30dPpLW!{fYluccQn$|2|d6{;C8z1LVD;L=tm_?=tg+JO5k$3?m>8T4Q!=r
zJ*X#*#WgSom>}BqGP(xV(6s~KOQUOGToSIG?RqY-5?DglTDx9L*T7wL?ZUSO$Ke{7
z0ZeeW>r3exxDJ>Wfp4Y(tAXC}$cF*#`ea}eun1V_Y1b=(t-xkrS!BDOJOS=e2oKl_
zTnVfm)UIy_)&ZNz9Vqa@uIgxfmk!tfTmozbt^|5w+VzdV0B}1n8@LCU2NVZEZZP}-
zgTQ=XSuDQy2TT~!t~UVFhPLaTgQ0gA<bZhz_^uu>IJ{lo0jwT@bRPnLiTFkzu=>Du
zy&70I65sa&HjHZ5(~_ZQEbI#mB;osU!0d7Ddg4U5k8jrtfz=ZbPhegE@dG9t1pB1G
z9as!3I~d=a0|tRXVD%wzKNNbB+x1dlU?S2-*T5EFLrS}jcgl%YU^Xxa%m;c7g+1vS
zSVz~B5Kmw$5U&9gm4~<M1;D~7C<nkY;1=Si!VZVQ-x2Nl0$|yZ_!b^8VLIXqEIA6{
z9S%MN_6L?^B0qrve8;j0SPgVdhFyROz``tq11tfS00VOnA7C}Gh4{J9HwF6V!9KtM
za4E1X8~H-l_<p2oD(nhO0M;E1eZb0NP~L#mz|FwCV-fEoV3*?%Z(t#?7?_sZuGbTu
zfby6I_Y?8GKw#QQ$R}VWa2+rRYy=iAz;_mbflCnXG}x^O_5$WzihKta0*eT*M7Y4Z
zf50w;*C79=BR$u`ov;*k0|tS+fYs%&{|uyaHNG#H4*zRV{(yDW@DD6|1m*N7#OE2r
z4_H`>{HE(?5iel23*XF}34IYA`VwGu6uvbG%=2~V+ks7kJM?B?!;lU=IRoy)JM;`-
z;Rt-w4_G#;L$3msjPB6)5GHl#A`|Y1bm%RFM|S9wXTjgJ4t+V$GrdEv2POb_00Y2Q
zU>dLsm<^1Xjqrg<z(Qaeund?3tOn)-8-T^YCSVyb2wV$Hn9-rv0Mme5fO)_sU?Ff1
zunZ`&pa&QSYybv;O~4tzATSr0kdALE0t3JjU>a~GFdJ9}%mdZ}3xN&565uXi8L$ml
z33SatIKTv89WWW#0L%b30r5I|(F!aC27ya~o})VSa$o{*9WVf_1D4E0e$aI$@&lNc
z3wr?zfiZItzY}0DU|v4-0}Fu*h(8PZ>H2Kw2bKWK>G~Y#2bL8;Kd=(G9aw!X^aHaO
zcj$Y8CFgbMo_UCGVTYawOuGQzd<2$U)S+hqD=&dw!Ykk}8~&d|KFx>zkC7k1z>lyy
z;jggo(QxlUegOmCPQ3|OIIL4oItKa^I`tx8Sz@QY8CW>7Q}^WHdQ7LD2doZs>g#}o
z6Fc=bVD_X=J^5JJ>BvrfDKMDTsc#1sW_Rj|$Kg7^Q(posEQLN`c6q1nJs$R6-KiG?
z>(+JZyMTF*Lr*UB)Itxiq^?tMAbhn`k2?YJdb?B42L`sm9aspg1Xk|o)LVg`FFW<L
z6JZy<Q!fK%5AV{OfSyrZdf+7RW4iQWV9A0meG9PR>@GcS0m416OV0o%6nE(>fu40;
z`YvF}^IdxK$p~jxmtG0X{-#TB0jBNl(o+{A9(tF)6qq2o^#))oFzyt@XHd6Z3Tz$O
ztv3SGMtAFz^T3Y-53B}O0Goi-z*gX9U=Y{{^o;M;TYw3`Ag~1JKNa=|P6k#2o6kqQ
z4(irZFMz&@-THE1!eQNd9o*}Hdr2;>ThBNR>6_85uLOG1yY)86mCWwe^DacZjzzq|
zm*Cr+UBGI5pK~MlR(zvVEP*_}%~^Un^a3k@dB6qWD^Ks%*8v;OK|H{>0w*s*{DIlP
zrp4WQ%tg?DKKv2C1aiQ-i@Ws#@P);QC$O@jTc2?Tcwio|ay9G$Yz5|DjQHFO|G<Q`
zh$pZK82B6FaUbk)3F7y1w_Xe^Z0go)ix3}R6ENWu$eoG&0nPw60P}!BU>Dr8n~}cD
z;D0x+frY?2xR(JNfvvzGu=01LCm;TR1(!oE*sT}gx&*iu*Oi^HJFcsNmB0qzMqm@L
z9@q-p0Sp3Lfu1ge2TbdRea?cOo^Cw}SO!c3)&X;X4ZwV0wcf3l0Sg^H`gUL?unpms
z4C~Qbaa{&nSq#0Sdh}*s!ss4-&pAlnm>#|2O86Vsqc19e9Iymf4O|Iq09FH=fJNuQ
z{oo#bGq5hXNB0-PZin^gg}}n8J$g3EQ3J3P7&rpg$nQ#EH82R=OV?>VdLr_z0q8|O
zrA_bA8*yC-Y@vJwdXaBIU>Dr;j)ERsHvk)uf0Z+P^rg6N1(pL7G9Zt9s>|%r^MOsX
zd-NLc*;zgMF7gLlg8XTi+oLDL9tqh!`V6GM@M!2KyXC;&+0X|p1Ew8^=YN2C$3y-K
z$mjOx1;E0S@hlMR)v&Nf&jB`_(xcacuRN_sFUEDr={>p&cC226=ZSy~z+zx4upaJ#
zGY~H9SOVM(44j4Z!rgOrkM21aa=^e<xV{knah(QSitD_Kk)Ob_OP~*Y9Wbs0dM-u0
zfMu8U=<9$@mm?qH9t75qzGB38G4Kk63-nxt^Z^6FR>&7#4S%?<1kU(7{9S`|<GQZ2
zM{fdFUyu9;-?R*N1bS}h(UXwxX)9ote;{5r!mhyTGS~t6pLR3C#dX6i2p8>5<*l$2
zFrmChucYfckPcvL1<EDlN>=sg8UKWydytR7!b+3_VBKn@Zz<&NMf!k+YoQ;QeIN1*
z*l<7eUX5@cKzf0N4?=G#uGjbIldpl^4Jen_LGL#3xK8`9M-Kv<cECUQ){ju0u0?!y
zqC5lZnqUv;N&5u$g?kflGx`4%a`0CNTm-EC4B-JQcOjkR514a3{C&}*uLWj*2|ED;
z&9Kul#N#W354pmY9z6l^EBOxf1bpRg)CXW|E7Au(;Ro0Q*MVP=56h80hpv|Zt6jRj
z1y~1c0=5G80`rFGdd3Zq8;1EQU=y$k=!w_$&A`e7b=|W9`8i(K7XUpIbiELm23!tI
zI8@ggfell1J@H2P2c`l&Q*}KDn0B<TR{~2;)b$o%888T}1ICrXeUYvQfF=35J_A?=
z%mp@`t?L!!z8LWZdd`Rbn~<(0m^%YjUZU&ObbT4p0W1T0ZpQWHNDr_TxC9skmI6J+
z&`0jqA|Aj5%)zDI0)N0#ARTcEgySD43P(wdV}y5rr`)0VM7WjQ71R$UnXCaZ1WEW!
z#;@pR%%ym6;)|K<OE@|@>MBpEn0dhTgAY#{16Jx?1gyIgb0$RaL4MJ6h|>Fl`f~_J
z%<{#|@+D;Y{L5UoX}3GIa~-|}$$3un`7blhWtbZw?s*`np9i%1bDm^~&hhy(jdP?K
zqAmF4A|4aSU#2f+ne!|kMV-<|hvOf)IWqMF*1Z3?5F-p5<id^8TL2rJ6@m4cPBbF$
zkND2=`Dgh8vwW$UzO;G1#)xGDRz}?FzRh*Jw!-Pm@ug+@Ql&ejC^eZrPbQ?GYX*L3
zIz;K}psqWia|EuG4WKjA7l6-Y?v=_G+njXGI%biMRLX=}*sbXPpdO8LI*4lkw*=f!
z$tZ3YxDs%?Lbx_?S+L;*giQzeb>Te!!JrN&c~F^10JjKSCNWt)|58|tczQ=~g@q>_
z+2FT>cauC0;ue8x07vPjgYu#X+!k=?rlc(keV!~IgbQ%F5^kP<1@(AMx%IPisVmgZ
z+BB!Vp-Ww7+d0hfmAYiS_9C46bwNFf;(&wV>Oq*B!5L+O@;wn;12}Y{GLKNKsGMrI
zIh|+IDGGvz{ANKe26M7xA3BK31D6OcG>?{mO8|#1TplD>3a$WLS_oGGt{mK;5Uv{B
zN^rgqZZkNV8y+3PHG*4bk!t~0W#NM0)>=5P3vG;r^Mk7ZXT*i_a56ZG-|-YK%y0&k
zOeBWVlMA;ka5M6pxB_tX;JhS{gSZlKo53xXjN(>;OTb+AfgxNKxFm2Bnacszg3AE6
z1!z+rvwWovQ#m=_Wmg<XOqEF&3T^$PLH$vRG*Yt6sZ~dyyi+{=a9i+{Id3O}%Lf;L
zI~^pK4Q?qofINs>go_djR|Kxu!Yv0^1dftJ2l=f8w*=hXl2P17aKRA2$#AO&w-?+p
z@@#EO57LZ=5H(n}$@K2*1(?T=C%s7P$y~2JNc$r2V~B?|@>JG$Klozsvh4(aHuGuV
zSAwTFKtA<6@JTq$1-}kFhC9|ebCFSJvf)$$xh}|cgv-NxQx-lECJXbazHWi88mt*C
zr7)2uM01(bd8ur6DXbQ_$6#&Xhy4rd{%~PYzN8L7`Ku4=Lntg9R3Eay)q|scO_d+#
zWiW?G#brL+l3vAn05z_N+f}lzES7&y3;BC0*CFy#0Y6o*1@))d&p)NxrSh+mQ$9oe
zG{8^tmZ1KGO*ns(KGomTL&E8TpXxV)`X-Z~i=>3A)5=e1{Z2)?Cchcf0rDW5WKmi}
zII>9|xMXlMNDApbn!_deVz}j6<jcV2fE&&7%baJD356fObvVg~8~SFcYo4#!fed~~
zW-ur!cMXu)1DUf)M)tV9ZS`qJxtk29HporJ`j6~?!t}e&W6j+zwatNns2}im0qgJW
zo2b8zSn0l1_Snum8Wi+(*gSrK-*Wi1sS}qPaU;8|gdD9+Q8A?>8^0=WE#QtspG9`;
z4OfjlGJP4Qdh(dVuBn3*>E8ogO<22%#5o<r(c8&)fHT6R{E7qD2+k-My<u`%)i6iP
zb%Z2Vj&M?y&IRzFyS*<Qs-uPAa=^t>m?|Bprt^F2Xc@bA*{3thmupI=J51M~OJVJR
zE`c>SyRazDd)eJcGuhdL!9W+*@y_GA(3@s^J0JK*c3uEo{>HxaQ<#O|62XOrxfEOi
zxc`m3ZGbNC2j(z$f%AY14YLiL3*7%kI%l9m4}91cCe@Q%a7o}Uz&ZCvYOK5;^<*Xd
z=E3hD)e{P%7IHOzF^o3&ZP|}75-{MIf%V$Il8y!NTLiy<ln$zAOCi?)IouNWvsD%<
zc3-n{9^Gh|S{CxZ8U8(3OWt7XKR3jGc|ZSt`00ZG68OK7{G(2y8%ML6%eO13vhDR_
zz}W(sH%u~xy)smHvLI8wv#;;@le%+CxVl5S%AsrT$NzI(R8Dq5SKcQ<{X&X+Z(EQ1
zGILox)ofX$2jU4BxUYm?OfMPjRp>pRrQLPr+cmpM@R|=jTcGEU%0d8gWspn6`v20t
z@c5n>H<ZQ3O(f<9;9d*aBFJ7!vf<NprZrs@pEmfdg<njoNWUTXG*=mY5rvU}M$7xz
zUkhUa{N}>%pM|j$a@Bt^j5_$;vmaqJLoN&ZDgKJB5(gt+;P=nckp;QtzZga-{3d<=
z=V4SsZYkt0<$UX{W9G7Ko98X?TfZM+c+p`deDT-9$cEoV@cU=BDuP_?Ukqa-{08?U
zj2)26{qnD+BL)Ne3i$n_bWl5$2Dv?uTSfWtr{gl`QoHs6jflK|SPoqo*yB^cx|Y%H
z19Vb-t)lqA?T^|Is;><cAILpPI(pj#^L&!?67#qgMMsXEU6{DY|0<~8!hS>Agk?sX
zkOjA7$TdJNq>cKcHgU0Cn#p#>&?B%%smgw=mE|+v`;7?GU*wandgxk!eND&jmoDcg
zc5%w|<;$_K7n5gsEq&u9>SO)j7J&P!V`17Kx8vKsF;XI>A6z3i6jP(@n8v~t?EY%F
zG?~Z3bHdn1j)j{k{=4nNqPWmLJaQk&aXEo9oX6$l?*9|_40b<&{U2-bpZ^E$XqrVS
zySvCg4$9LCaJh)n4fg4`^u5~5FzJ`~Jk8Km(b{K!sw;cJm4jPG0ru9Fe&d2mt^H>J
z;;EQ8X@TF{`}}f0eY?|ngE8Hr=6M%EmiEfgTx$4mw3|ckqtq-f+5>o53;%&1*$$R?
zn#PA(f}`I64^|=_&CpSdeR;Q$E&ou*=&(AH5>U^um+u~19U<dP?O;0{xfG8==t%re
zP=Dx8bxaMbV<U8wVb9<$==;O8IPbPIhb&ud(2?{r=5OrFR+g_Q^gh%1wtah1Bz4Ug
zj{XR`V#4V%wPVgV?RDkLye@;T&A(v&@XzeyeAy;#xgrN9p}nJJzoH+(+^$WZ)Nfw2
z#<t?9c~NTb6H%$F_XYKDac-l}QV*T)*`<%_a{=_^VlSpYoSjVVnX}POPo^(d+MM>i
zCSgD4RpE4*ZQg0GOWC{&y7JnC`umi=esv?HTsv>JwfPKq%NT+71UlqBSm-z)Q_TzU
zlF)e}ZHX~eBJVRw;lBX;Qa4i^LjBv$J2=<drHA6t06n{)=dZ@YlZg45&i`3F(hv{o
zbhbdxe|z6depkY85Pt3MXG3mg?+&pgnoHGZG{EnYuK)RXjLK^q79i@O>r0&PN1nW9
zmnSr~qhSv1i(b$j)JL%{npdYVO297ye`UCGY^oF50NbjHN<m?4fv)l%a~Q<$0#AFE
zamy#uaV)|nelN?1@-x7DF$gGw{8ub*y~p6Pd~%5Wd&@HAUk?1I=|O#fDNV*0lzl%G
zG6rSe0r)M0UoZAg@1(F%7emIN_ZwqSbhNUqtc7enWMy9xT4zw_Xrd7FCLGuvbG?<}
zsdj|+mnUOCcGxkNsU6Xh!`MHc#xPmP+#2lVM)NC=4E)Ulw;5c7$rfsiVUFXPuyI@q
z9c9=Po^REm+6K#<sPkw0a;VZv`{dVRKlx4ex-4zE77-@B8B}&B`!TM-Ui9nsPnRd0
zF0xZ8bR}Wm`kB@~%bJ%o-E5i!Q<+KSZVPl|V2``kiHqLyfO1C`;-a9PicuB{kSx<`
zJwBJpofwV!j(zZLbPm5LcQmx3W&8_exs$<U$aVW<$kLwqbtG%)FGAX<vqS2KF`T1q
z_F@VAXZzZF?=2CRzOmyKePc%|Q<aci3t24J$%E{^5nLs>e&yb_uh0$(XLr&cgsv{^
z^T%|sJY@YJ10LL8O<xZD7vkA~Ye}QBk8S!WEhUiM4B71@+nZ;oAGT?6UT<HvGvvOI
zEzp%Q7<<1-S8u);?T*?@;anM#{m2!vK^Od0!(XV4W5!~=3EWAJ{fSrjdP3!L5p>nj
zGZ8rNi&ww<jGN8nMXjM!!f)abbKEw9qh~8h_AhSUaQRPp9fYo&xOV*$id(4t&EuBU
zA#vl@8ubNU(vmRW0RN$V2<0n1=h6cHQTsSr+AQtxFmcQ!U2CB$8_$k>hjUXL?aN42
zxH6Im@2wPH=n8kw?HQ-sx9wlN7yMDY7eH4Lo^|<U|KhzdT)fE!RnV1$=Vki!Blb4f
z98Om<bnStzR_OYk;$^L$HyP~@^|B~u_$7`<{WzdqUx;(E|1xKRoR<pVG7WBf;C2$b
z^|fUyoTnSp)lfyU1(5aQSs!B_zOQe;JhW>UF4(>tvIUTR9%FNxJfG*QH}CD3r1=^g
zvwV&CHR0Fnm=6hUpBroz*etL|GwZycNh$K16IW+IL|*+fR81y(IG<IMg!3g_V5gT1
z#}HT?D)63-z*(UJ-Juuv+h<fg6+g0l<pkIt&kucubFzc8#OIlVO*Yi7)WWS9Zu3bv
zTzmJlxxGWRmUCQf@Vgh!BSny298|_!$fP!KzqxP`+7_rXX05ZWntj}Jsf<ylx(?4Z
z&1YRJoU;c-9NTwKtu1t7hhe5S0u3;QT?+XvGuw4t3lp~Keq7596L!&Lc&BI7N@lm~
z&*D6+T^=^uh2rcw2!HS_)o<(%<rC!$n}}$q*8_b;6qL08WXLx{zLe#y<7D0oaB)c6
zDeqI_kY7dc8;9q#)=|T0XKQnMk97D-oTp%?(kx#Ye&vo#UnQ^#ziO1+8vN>H37+pO
z!ltx($Hl<oeI+IEGM~N7hnM;AGQXdf25?jXikwC8TI4Ib_RqmNOUY<QdhCk99AA;`
zyf65(!pO_Hi?V|;zQOZig&teG%=5LG>;2UO?6hS<4EI}5(FA?lPsekY&M^9F%zFtB
zaro+7NSX^qbvZlil*@Rg9)j|E2A(&=+No{+S@xcFdhJwY%Drba1}P<d7vMQToYO&Z
zuK-tiVY}{##6@rYHtrol)`hi^9`l$O#SEiY?M2H_88o^-u?Kq7u4va^$2rAa*3Egw
z?UM63R7{!1I8=&e$c?7<rGFX4_M00b`)V{s%J18yaSn+|eFe~0e<jNNabfh;#D}WW
zK0nq@UzV?i=2M}f7W({ewCj)A=dE@9;7;4UBG5zj7ZZh8{wBuM)L!8r{c+&@+wn~4
z{^`H#kM!r0{&(B;E2w?LLHdiqwY`si*xtVV=II)32OBE>_0U!EVY~jZTdJ4(cYw?P
zq+NeyKlQKw1O5IKtY3WEu8$0{4?SDk{KX%{-+b>JF8b2`rO>s!8P9Q&t#LHtR{^f*
zn|A#_)JcVwKhxM$3*wHhce51+r*nL%m-S;x`Z4`wv=jPkx~iYQEBAxByr0Y&W(Hd<
zFz!aON)E+(?)T_(23gXm`rLl)-QNe<)d|(!k-i4#Oa8rGpGkRsOvpYlwi)+H4L;B7
z+uwQjJu`bsNF>o#&=?~D@k|W1TlUHIyLUOp>{FG=1@M~zzp{Qo7hxf|8Q{1WNp2~)
zG;myWC07n^vV~g*F4e-(^Vi81ecQnWz+t&t9x?da436|+%Bz>7XSOMR#(WLQc_txW
zz|lD!go)s?E&5WyEwFG|;PNe89=IhIZV9*&aG_z8f-43m?@Kt`3UH+szt!MYTDZ;N
zD#3-?v=Ll|MXm*0m4yp}+X&7W2U488hhe;L;r!s%S-8pINMESmY;bEW+#+xkzfid%
za1=jo4yB!!gDbaimEb77p?)`lE3?SegIjLlc7P-Mg@(}zZmEUq0!R4~Di?FO%8yVk
z30x86LgSJKj`BBDE(cs8xOr3uaga^(!L4g=*EcXnWBX!o8J+F=7#ai8J%tt@P4n0!
zUk<kgksYQwu?}1=xHsTV2kEE-=f(5tgL`#E^y=6Fx0nOWI$B8{+;*u!h1CVF0o?eI
z_{Lx$-7>=LHwj!bxc`LsO#`<%u|ppk;x`9e!+~bM`QWyIiwuz~23J3_Lthi3uMAuc
zo~KU@ky{I{(BGl|ErhE9SBvNE^Fp{S;KbMt{Rqwsx@^Lq<nbN)2<lfbhU&ZL)Gl-M
z4e9+?hn(j_XUr7z|A7v@gBm6DJy*!vt`qm;?=e#5zx}H;hx*!|u(0zRH-n+RLeKl_
z?C;uN=3=Q+?fzCf&O?2v^W&w=Lirav$zj*C<==CCp4q;bv*h3N<=^>ora*<FV!K2#
zg)nRJRLqwrcj($IW#iuaC!_CccACb6$4A)peObO{r&cw=&Iv<L%694(n-SK6i#zmJ
zF#l~DOBi9*o5FHFXs0jBS8wP`Jp$$V)(-voAz}2Dn)R)=*H>!jD}}zv4|eE11H$OT
zR`-x+%d}%{$7-+z$}Wv98=<eIx<mhGR2Y5c^yb+|uii<vOT<7Rp|(T6CfpdsoKMcZ
z_R3^lP~4V4U-qjV`eAfm6FROl@}kOQyD)L9a+0n(=$f&uL%)O!*<Y7s&f;YInB^*)
zxsF6T{SMxpLG6c~F7rLznRd$Pj*jA$3td|pJM_<}eX`SKTGzv^eHw;ha@yN3s{t=f
z$g;Z@y7E52J4&d7x6@^gm-ca($&pen8+)KHuoG=Ib!2w>%1vd>Hl10RgP?S#O~d^8
zXQ)@<;%3$rCIiYH$VRaox{8|dt`e*LP2>F&Lx!3d?PDZOWwaiC<61iOx5Cw*dM&j6
zINR(Dj`~eOlfIbg7@y$19XrC+pK{Z@f%7|iedS~p(w7f?NxygKi>+x>WuD9#8b8f%
zTfZ|Qru5Z7Usgwlerp1h_s_$5z8Z6Sm)XZp7IlgrEo?21?bOc>*FIYGUGpdUa-nY>
z-cQm@;}X00m71TG-DR(@l!|BpepS#nIkQuz=U{C0RhiPOy=z+s=J={K=NooGWx|ER
z@uZ`DI=WMTJzU#lo_}&aX{T%srk}K@Y^MX}!inr%gm88&=+y7@gh@}Mxjf|CmIs>J
zR^_1{`jXD*)a&i*w0u-5wEvh9MxW7t#2kh4e@>@<X1IE7o}Wy&O)Pd%Q5&%cx~k9X
z)c4w_)w0)0dm~IYFLhNz*P^0MeNnh_VD5{~w#yzGF;KPH3tdHj@6@w!PG#yiIi6AD
z>X?}*zi`W7H$1mS)9$h^C&R4)ZlmqvkcBx`Q(eZcJe$^?#FW0J&{utZr+$98GH2=g
z7Y?-7M|nqOi{4q)u(DI1;Kzks-K%k%;@1>rZn6erU(%PDf%t9g)ScnhV#>|>qQcC{
zmb1Pk&{w#%Q-7Yv)|PP}Ciu+u%e8$FGZSK}-`)&;O?W5Kq2bzA^E}kDFl9BL((1{?
z{mi?a`VqJ%w6~pUY@%&D*gh|6FdiX&dC<2JZ-375?Vr9$ws`@QKwkhVs!5-(OP3vs
z%3EGrUx8IQe#ZcI5mUX}0oj-#U3$IOHf(D-z1BV@RkFSE&qDuz_c{G|sI4xvFXl4v
zc^D%%a~UXtzNF*3^oxdu(N|^aU$q|xhlw59r$VTp_o-zSbm>2g38&9vD)WnM>m>RI
zk9O>+FhNoCMRrY|jrz3$@3;!r*0!1YqA+bxn?|}ypli{sU3zP{vTLpjVa882ax6yg
zdJ}ha>Ho(3xJ}u$jD@u;<LuKZ*W6>!D6XsS>fH}xsiT&CLeAxOs%Y;UrEw8-Wk1rj
zzk57Zgws_BUDeRF9lAEymuZZ9%=NP<Og`5*wWA{K0#a2?*4?-{s6U&!^kW9w+H;=I
ze4nDF#n>yXa$@`k6$Pa4%`Sa5u5I$iVpG}=Zqsed@nvM$@$gNyZa^4a4PE+)s24V2
zSk^?GVN~VIIjFd~i0?aHz3&5|d8q(6FE}hA$%D8V-~!<2oDRZVa7o}IB%`<jaDH%9
z)af9(5^#wjToSmI;1a-{KtgEasNgygRk|md<5SNilHXd$<wDLuayW=<0G9&}&5S%~
z4rmwY`@CE2QBdn>z59`_HMTpLUl$|y6UEIFBICdGNB1!~@R!%rrQX>vPtL*jhKJde
z*J;|FUvty<x*GVS{40as&G7pfC7^%b-EU3k6}MesP`Y+NSMC>R|HHK{<}qBH?T{Ix
zGfI;mk=ycBm;NBm`_>{Yd#je2*VJ*_Cf|Lq2!0EH>Cz`tx$m_<J!Pt1Z!^YedR33K
zLpl5u@9omZ?O$Dqw(V{ZoXmgdYWl4Ezj|K~m8H1(m_P0@m*D`orQk3<C=b#<16(Wg
zpF~V=`c&e=l`qm?41di7@J>DQhXc1bq6}OYbgd#n=~B;DnOAB;m$R@^Lzd?95C0pX
z(?kAoAn9TkxN2}8*xS-N{(e53ElF44Xw2_>yY;8{uS~7BEmP1#WvUpuVq&`Wr>Fw=
zmUW{{@z&jEL&_BDKczGg;x+JF3_Fd%Ir&}Y#JeysLWhW)7jK07f+M=sJE!K7dk)-R
zm+kr<xaGmE*}jd$n%e}^Sm9{fL}1Nr0@^qPKjRqW|MYHsHpK;b);Gu8`zWn!n+hP8
zkM;!1n(`pOCE%7CoXPKEV@6u0rwVe*jkXBsb<UB~sL~enuF&eEy7enaD;i5Dre3in
zd@ijeP`TIzcMsmZSU~QkFz}E!ww}`ai_rXK8Rt|sV=#D)`>I>dB|kXG=1Jgk!Hp$C
z+8lGgZ1Vu9EV%o>?bgpBcU5-#8oE=BhE|oyBFGj%_S!yKlyf8R)n3hihGadWIG5fo
z6M#lFK?VkXDXyELtM%{Q`T){}gYNAb!L@+9p9mRO=UQWJWe?m6|Iw}9X9i~~*Wy@|
z7sRLAR2JogTxupYu^FZ!@wIK=45>J%%7L!x8@tszq%amg1p!lD7l5w;e<Fp4IKgQd
z@`}nVQk5@H5H+z9a;=ct5N__Ie3|(MDLUG&6ToEAG6#jb3;MF}>DEsU4R?t-T(YC<
zIP6D&Tm;GCAT9x%cTKnHolVK$Jm7{)DJ7QyE)JY4gj)bE#=;e{zNiqnrK~TMD+lL-
z--Eexr?}9*5kKBVIxU2&1DA7OUwcP#+rednO9=_18C(H4BW+}tz2Ne}?I1aGy?D#0
z3mI@qJRWwqzgvIFS~k!a(SWM=IRkB-!@LiNS|e2+^P!{Sfo?T!lI0Qa{(&3CcPZS;
zYP);it(6b199%(bw|+#pHrq6RtnCg{9&;Djl?)7`Wcm?C#t+>(-Rhdte}R#F*-j_s
zqP>J%Hl<(H>wa}rOT*p?OWmY4J`aAE{D}ASTK&@QdVas1^8~Y3R3lj@E8sW(7raB(
zYIkGK)NF&Wdx$Bm&Cpf+E6OqJl6#(V)KdY@PTKLL+5ry+or%A9Tb>0ow@Xeu`4XDc
z9AByvFOLYN<h}X==<>IB>sWd+>;@S%E?Cmn9Lhea6tYP#;l0B4X~pWW`M$;(Cc_(1
zUZ|~ZguWKMFE^9IMtLhXT56nA``89|*Q?#V`=Dk22tEcsFB&Kh+(zM?4wCnSiwh$^
z18xa$8z`lf9dp6eBK*hg#|hT<^g7!y2g(Sw&1<16h<EsYLHfg#Kj%Vo1{{MtS!|}H
zU0R`Qqq|2RPr771#JjDy4<oyIFo^R;+Pjk-=zZ2Rd_BF-aV<m6Q`w<*I}2_ZaNCV@
ztmpGPn#^VATH1J{_T3nz-%@l=l4CilI~CB^41MjS4{=rZ;D$aPAK@vu-h#*DBeGc|
z{H}}c(PQn)lg0k_%OO+`Db0xsFiwf-pJuXG8r(Jy>#-dZP+7`@dvZdL{xZb_2gPj(
zxTfJf;rG3n>-Cl}MP1HQ)Iry-1AFvC?bB(U^N9Q-T`qKx`7fj2rB1BZ-o0Z0tFEcG
z!@^m<+<bV?g|02Hp%2=>_FE4ZucPhSZ|Kj$uNGmHZ|T+x?e7&$1C|c0-}2&Y9|{@e
zM;G+ez16KBWuI<KS<(Eq={`o))$E1XUplHsztuiJthz1?lWtk>E1;_$@6>*G|8za_
zN4lD!D=Vo-|7W=TFwYS$v~7>@f{0YuEBO@Ehw(l7<rJ^paW0Jn)cq{B5&CftAYDB0
zL-imZezr{L(dScph=baT62z%ybC=$LIJpiVvp;3-$}sWD&4oUC7k=t1UHWw7qbu50
zU%%%jUb4;1OziDe^Qwu6cjD_^I=#Eo)j8Tej^_1e=We?)OAF)2+d1HMtO%<B@h?MI
zF&}iPckF%^CM;9iV7oN{w?5iOw$mcnK53^mgp+_dmgmu7xgM|$hjwF`>d)n2)*xv8
zjclKlhw<dMU3wbw*L9Dr&VFmWo-npAXPuSM+1%QtABg(xI@4BXzj&tG#xuKjop29y
zcKsXk7YM^O&sL|UopbK6%QboqzYyu3aVqNjPhI*15A;Qa(N}J^<!0OUp=_*e+fGN4
z-K34H5mtSBm%fnd%bR}tvS<mbEli_e*tiQia~$0|t#!FJ*y`-JcKVsEjp<<^RTgqj
z!~I`Gw_Ja7-4#ZsWxe`h+eJ@Wl~y{}L1*pTxQC~-UllOgo&Mu+EbR6T2;_}@s&kc@
zqB#_5$AZvVdI-MZfOA)rYA@sjBc2>lH>ITxyvxKmm41>>$M`6@TR)e|&u*39RzDhE
z-C=m5aPr|N8{f*1;k<6|Co~*bOwe~P_HOUikEZlKfQc44R?{+^YT84VhdRilrgZC%
z!Gr6cO0Q*|0&_X$b);8^+n44xv~N_8y^C=FeO$MG2G%-V7b=_d(^-C7l&Ntzb%?!E
z$YnrhK6D1(?Uw5wuH6`UhK?<e8GSd9ct4bBD?8;B=Rb6~8Zn<ic7HBZ2gZX`V~t1M
z?+VFlIUdZ#%vfKE0EN3hr85Yfx$om1fNXqisLqfvk9jTId5c|2v22ED2|s>0XJCAN
zUbmdPcWqPlTZR{@q92TLB5;x^f=nx9FbyYMFB&qKu3{MhWfhRwTiC517Xg{a!sJ8E
zji!8<n<7q55zdoR#OW#G>=ZFaUE~3dOA&zMQ^aZV@3Ei?Q$&7>xB_RO{{hHJ5tpQ3
z7Oe5cr~)5U&rT6%$p7$jQHm%|!3Kf48>7%S-WY|agKmskloE5QnSqxRQv@QAn<9=$
z5$C6f0u?9QCt0w9ePh(oDdH^YYk~ZWOGFOk1;5I_;eG<&Zh4t(@Zo54x#`_2XSM-6
zBL)Z+A3VD@MO>UBPDl~gq=-dGU|Da9&q)#INq6VpQp5$6Xr!sYS7js`{t=zaQpA#!
zn3H_flxRd150Bj#b#h8fo|!?Sk=)QkLn&p%>2_rWO`1$JY_r(-4+AVr5k}HQGSXgg
zCen=W)u4z9mxi=sG=(u0EgIlyJWM9aI)oB~3NjfoTOeQYA->~+vg7)3tYItM8qq`p
z#$|XsX`wMJll3A)`s;$fmCL*3GpVj8{YJYg{qg#QbH3pZ%_#*)?Zo7KtgpkLpZfb7
z;V-mK8+lW2?)U0!2XC-F;!NZ4GU%+j74uPX(0M#`_TKBFJ|bW_farZ0s1`NYJlP~;
zNBtU&Eq6fI4%p*3s@uQY)*s7Sp!Qpst}>s}n{*cH<AdG$Wz^4~G}=DBmVGLlZTqzO
zy?azkpws_ox9+AmUILw#^vcEDzVu%2ut_iNi;(#bT}hvI>n|X#u9>~*mHC*f^1r~y
zM%64TJ9WX|M);dW{r{+5f9O|3+CbBoQ#;nS`$rp0cF#E*<K?fr^-YME(LY-H59@l%
zskX%;(??U}VlDJ7!MyY#4(NLmHtJu_toKyTF!`8~2z`5?FXubV$B~_GRQil}a+ACF
zoDlb`%<dGQWVmm9tXm&R<^Ds=5cQ9bwOqYon?6WoK;Kg6n}NB-FLCaQM|hSpKz$=W
zv-Cnp-^evl^<(r+pgPPk##7nc?YE25JYTta{_F+YvWY2k<Qt_k4mTvVm@kZ@bRJ@B
zcgr|g+ZZOD^eH`20Da5%VLreCeKbzCq?1~yzNG%%ucX6W#%m+=ti}A~v!v%6+q7Ea
zRS~8<&=yV61$|!3QAU$K8c6lV%a~{4_h!5oGKgU7XTblAb20w{|FYe^C)B@cv-%RH
z;r%atgBv3Qsck8Rjx>CGi@qfgrY*JZRiY>RZ8k@sb3r;N|8_v<0#}c|kkWMw;$>G(
zi%g?9y!g&mrL2cZi%}*9_UH$ZzIa=G)_NEfMqdWSuNeCF8u~uAtw;N<kMfbeH|6rG
z`99Y@&=)QrKeEq9*Lj#PhJTrl%R>F{Up`VCvZ13CbB?!=jV=$>VJ=JBM3e-y%(7ji
z#YgmQWRo)ZZ^OJGy`#`|ysb^FWqW=Y8_Tx13Hnk;^ynW$w`+v0K8uZ=74}6SL$1vS
z3UU7leX_3nJ~F(G!_*b}wp)23+xVMM|NCcSve8E9Sc~~iSq?vhj{fD^k|ymB%V8Jv
zEggyP)RE2avejp8&%?d9JD=J)`cBzK%>R8!Hb1?88qMvT<(*fuD5!R>0(!Fi9Iu)G
z^LVvDU-4*;*MPqkFKVljFF<>VIZ4_6{&=9B{jFs*Z2Jp!a@@D{Ld@Oa+jrLz5vJbm
z?^<{RbS(+s`+lSgM+|<uz<Cz+=$Q7B2aT87z%2mxZ(_oP>HOSWP03$i3DSSYf9Ed%
zfBEpY9Nz`}Y=8={Z;yob-T-4X4<{<)D<PY<0N;W`nWXY9*U6nv@SK7l<8d{+;J?y1
zQETUDQqD4%Gvo=zgix^y`YQ3Q!3S|pY3&Q!x!MSt+Ar5d7?0vxgy#3OqV1BirpCv(
zuI$%%(fB<9{<Gk}88$^rBjatX4Q1mpA8zaLox&e6zObpY)-qFLJC?wt8|m5zUD*Xa
zder{u`Yc=<NxFj2wY2bmzJ4Ope=+PguK)MdsNT@`M=MaDC($^28zxu!w_BL2G4COM
z!*+f&7h-DMN8d8d#XS6{W1;hzFgnXkQv}*XzkLf|9)<ivJn1`{n^6Ytz`5(pL~}gl
zJ(wJ)@#5P8^Ei#~E2%6kfS$NKv|prWLZ}|2FJu+7L+Y6HpDODBeSdYwi9PatPp;QT
znB$@DFD>o7bGv=YRD0S2eFbP^WIO+mtv+ize^QuolTYcJd<pUs-|3Wn;f2r_t_`v2
z3(3y({n!H3m7QdtQ$qc#zR=uH+-bHm_Y)cLyBT`Aphu2druWzLXX8&wizq_-cn*)9
z4hhv^<Q0#BZN^TNmJIlBg?|}`_)!1**RN9)s4gzQ6!m8@zQH;W7k2fd+4Rhm_HWxa
z@2I_XjnLJC@15S~v*<z@CYz|zr`_rc8Iz;<$#{4#Lwv4AUm31_vCOf*ZrfO?z9$d5
zc3s=kyC0X@i6!8gz)iq8rE{4R&kf_0Y)}TbJ#Z_s#@m=vX*9PHC;042z0$P<x?*nY
z(a*;{oQ*yEt;ZZ|+uo>j_%BD9$9GrJ^vHwKF&SJcIJ<P@!YvDK%WzL)6L!D0uE6$w
zRHcKyd0d8XyPj%aF0A!7!}jqYM1#fxt<bf$%AAfaaFyT=hP%oSe2o#OlphJj=x^ZG
zO#QZ9nY6UouiCEdC|w26l~Ij;IGjx^^JUkEu?eN42D;WgVNS;ua8=;!(m~&8u7z7t
zT$prN>XzH~;Ruxu`j&C!Yd!ijGLoH5EOQMb!qihrM+tPTeA}FkmEg+2S<^xC)o@!2
zw^Aw#IEdQ}&Vz6A&TvUea*g07<9obDdCh=R3%FEpxm2%v+iRn}MWybYfY7Eo`?a{p
zJ9rCRiF<y0$M;OCQ~h)lg{YW=9S3MX<xCo{g~JQ!SOOh+pZDl*+QtJ54t?=>JX}0#
zprhqWeE<4S<KeW82esRKp`-rm9{n6!9j189w^afiR7Yl9g?jZZ_B>E}`j-bwKdohk
z=~pu-FUz1S8Q(KL4d--B#&0dS5^(nt(Q9){--Ne^hTc*KAi4won&I!MK7SBVn-rWI
zj13B^4AIAEi@)zHFBBL0Hg*v>D&}<1*gODk3Ahm$FWIEE-}4FYg?Y%P6l3OGs3?KH
z8GCtLjI{NQ&5tvN=7_U)o5Q%mM&;0Vb>J0kyc7UsI4EDYLub>z9{pS*u=o3R+%Y8F
zrorc}t~tKIN;g&!Rzzg`(v}UF=gYWQb|BW0D%1aW`2W4nCZtZ+9hV5yqbrVn%PU@S
z4EiZb>~I{mCrUi+obq9m*zNScA0^gkp0}e!h0F7Ll-TLY`Y1~L%k6n3N<8Y0dI|Dw
z_lsZ#xbY3P4Fg_;A-?up{dJVs8ktxdCDulHe~A*+QQlXhL~YdLaHTK@i{&U>%bk&U
zF61`n&s1O~AE<S@A90K4o&SMgwadN1EgpBBEyV3^?_+MU#_a{M!97~aP7~G8QSEeJ
z?-7qXOL6|V%e~Gco_5`U^XDQat?-Cd10uTzh))K1*L%db1H8*U;^zTJb69i4qX_p=
z&9fDLw8s$O=Pvha9`Uv71)Q&qaKG&l4@O{5X<I}bPInLR;$w`z4)AXA2z`LJ-Xm`D
zV6k1rdx9W8FE~6~yy7*->PR6zi}F0}72ib77Gk~E3*$cRJyglg74IP2cQp5&Sn+|@
zjkvxS;r=;R?1<O{9rYe}bFA1p(6c*MygTqGl-13#-e9a~i1q#onOHCQ=2#gp>=-X-
zgJh%gh=(G?BW^c(q9@!pLkrn>ilE8ga%W`wK;&ceD+9$km;1qi;t|(<kl5-@xO1Qg
z;;YyL#qxpPnt`Hxpxopn{YDGiykG0P&na$oRI|P@q6YeEobIw1@vL*7SBS@=Ju70w
zPto3+V?;;vM7G65aVO;Nbh@8&iF=$E2yvY@<Y||v(1zk=belBygD!EC%lo`b-0jLF
zFBI1!#T#hs-*9>MMT&P^n-Jo%2zN)MD2q4=ri_Tg>1zYs|A`bEBE7*#@l2!_A3J;{
z5`#kWzh<q=ahps00JHv}jcA*x-QbFEo~ga(J9$~UwmsUvZie=5Oxy!Aw7S8Od!}nG
zgX4dgu3bMQvUR%l*pT?|rfb{cB3q_wdR+Y1)3tp=Jv*jryM~S1FjKoOe)y9!wWkxJ
z?@rfVO~5V3lf&b#J4)L*JaXAlnm#<XJ6-$cfatr@wO<av{KMxX;@Z=-<%yA<=~_)<
z?C<GX#evaxq-zfy2qWA&GVZr@t#)MO&*|Fkk+J_t*S3v{zCB(0bksRQyyTDjce>W%
zkKCQE-7`A&`*f{ybo8z1+RbC&`0p`s-$2#a$gk41`mwQJrfchxqHjspo=Jk*`^Lq6
zmac6X7x_uL_V01AAE#^YkB`17U27hX8;SQO#C?>m?VAw!Ub?nE5c@t99u$3Jy0-ct
z%&zJO#ccz9aO7L*+RlSx-%QtDI3)UpbnUG};P_;6-0Pr|BVSF|R!oe21<6YBte>IX
zk>ajO*H)!;A-C>4{DjY^Yo(Kyg72DqiVz=8@!ors_T?0>p052k#rtcz_WKm?kLlX7
zsornXwcDq9KTp@zO!a=4u5Fm=eFq7c>U|?!du6H@Dz{A?ghp=au#abGKTi!*&(KyM
z5sLzL#IVn%Yg>*8d@^0zb3_`3NNK~~o36c-7I<g6w(-dEo2F}TO&kBfbnQRW9zl#M
zXS)A2Q`|RG)&SwS8fBw7RhF-Txua3&FP*K*i|&Y8?i~59Q~b+G49cI`eNDa7@vQTN
z_-#jNh09SE_c-F7JW4BduEzOlE%Mjt+C$pVozt~OE&k={+8TG{rfJ$;9uJJaY~Uc|
z?0o~>$ozW;H#k?!)K(80)HzeDj355ZOzo+7RI8c<_v1%tRU@KS&(wYyF$*G%iIGoC
z*M1%4DW9&D`~8nh*J}JrF-{#lsCuUM{piFRl;|;sV^L_#AV@Th@vNDtHI2DpFpAva
z-fxc5zCRqs+jscz_37H3lLKWlv{xp35%+DAy@+<}WFNBfp(!!zrfbhl83JRyJ0<4c
z>Dt#*hTJ_}`)vxete84-?=)meVA(Y7y{U7Bxci8azf99M9TB*0n)byJLxuSDh>;&o
z(<;&epB|}gPKy>|TUvDeH0{&0VJ}Y8wjLS%(UICmNA9FCHO#&Q4IONHuO_HhEOi{P
z<uLJ><ABYFi4Pr<UpP#><&5k+RD9@6p{8NnuZN0ySJc~+M1y<A_QOPDM9i;;i+cye
zti}0&_=ZVncyCAE?u&|lXp*?qdp%U%J}B<`N#dbFm@g^!#ol|EX!4E33)A-bJikv8
zJ-&JGABOJ%#s^cyjWHyC)8ND%lf=V=W11$3uLs9JpDNl1-$*eVlDJ`#SUV)<;Ys4B
zA+g=5qB1VBHdWNdMOLJWU*jUzq>9Ie#;;BluMTt9r;26q@i(W6d*g9?ace^22dUzb
zgqSx|#g_@uzov>+!{cvH6^{?cxctEb65mP{FC7r`Myl9*K=ig$@vjkMKA0pv8F5rQ
z!b>cC`7p8kz)g<V4ihhotks10X><wBzZ*m8{B3+<X{xwmg8QdK#i|Kk!S(lpBR@P;
zTzANb*AEqI4;i}dP|<OSXZ4}t;p7?L9V&Jw5B=;=ao@y*?n&a=i3y#P#5)tcI}zAK
z?<14M-6`J2RPla__vKV^>!IEoQpM^+x6l|`Uo+U@@wlTMgR+oiC|YQ}CFgK^91)6J
zN1W~$Cz3IS^K9~pYXc8>#Rdn=Uf~>5?iC%*F<t24wWrnKUCJ)NTv7CGgTy%nLcHmi
za_1ni#pzi;NHjYKl@AjCcD|3k(FlLQ;3P+a3M);8<rQOu7&9hMWZ@<h`r<_kBG=*=
z^o?8G>Wpu7i*?Q}*o4a$MnvZhUpqm3<%oOiAe1ZbbrZx#&eKuHwBf&u7x%g1mK`KY
z-QFk0i<Rz$7?`++KQJEMZQRcR@$>-iedEQ;13*`JhTk?`TsJW8%YbMZ=v_Hpv<(Ek
zHF9|OIPpwW+&clW%IobMC;sIH9rO;z8^^yL6!&64H2Azfj1xP2pdXDM{@FNjPfXk+
z0nriT{bZb2F&Ol&!NcDkC*F#UyC)zv4)HdO6LmvC-xfFg>2abnF7Aea_+qHHW}Nt8
zDCqiO!|xp@9*>X1vv}nR-n++%wF#hqNf@pti7$u8eKtY7c7V4dNxXXi=yfB8w<L)>
z663Z`5PK87UnYst13~XRaQItE;`Nbn&rT5QM|rm-iKj<_F7pq6B1!z>k6Q;LjrKm0
zB(59lh1qT$>xGq8kM+VR50CZ222YLkBFQg}^&*9DjrAfyAC2`Q{$GvtBD((`>qYF^
z$9fT&8<M;T`;H_pg1Rrs3(b!tVW1}S=YlPo`<-a9+2!68Enao4M7F&%z|#;dJ{mAp
zVDL2HwrH_zpc_wxl?_CeeKRorsc6wLaMZ?Vu{+A$5iRyaeTRX>2R?6CwD`j3{V`hn
z;PdW^Mm_gJ%k|M-c)cx}3WyAEFv<pO;*9=96VEt5khCmAA3KJ8?G{}Q-}i3unR`Hs
z8|8rJ$@Dct9g%VFT`tG)Z7x|0NE7jc9T5*;Ea-A5CB$Jz`*DNN0h|@J(jk`QOF7*!
z3|)x8Z`duGc+oko*@<#~6uV6pb?{#27=#DJ-*ZGhuZb@l!|&2WhhtoaQ#{~A|Mrw~
zNVg{5bdLK#6MLNNA&zjM6DR+2#6Ry8n;m0scZx5m%2B+gI3}Vqq7KRZfXnf-;}HaY
zuZzwfaFIHg+9jTMxwpC;FS{`0hbZ}@@it<0)G)lQ@sx2(#ZnnX#nTIRKXwm$GE)58
z{SmA{@@pnK#)u|(ZgNE5<aX?K(9^N2-R}F{;z76j3AbatyH~l=AN8RlveGF&cN{1E
z4|2HNZm(fIeU0T>cf<(#*vCVTQQI9DFJsvWalFeR9tQcaV-QNs^Nv9uM2qj8?vJCz
zPtJ#s&O6*+I&sl9U>Ssd^|<ek7C{e<O{*dYKNc;XiaZg6#;Dk5qs84(yFqt(M{J80
z_YNBOR<w9!P!bB*py4P~`k-+rLhF3dHPPZ3pL=bzsPlR6i56RYm`Q2!4Y?cP`NrK9
zE$)h@u3y>WX`DUn@a=Fr>Kz!>ir?MtF1OI#?v)Xa<@CyTxT^!{7$Ime?OBI=l@oUa
z7#H2)q|;UM9X+zljfi|)bBrs8QTzO@Imj`J?v5u9{?g%?LnUChBj$GpN{jnOr-Llk
zE3B`X<QPB6^MKQF<X4cp*XhRT15QXh>2%jR#dA*g7N=t~5q<ip3_OJn`l$hf)_BqE
zb#Pu&d3y$8&p0ACI}z2W7o1|Pa~h?F^39zP`Jj_Cvj25&oWDh!YoO&fhr7cex*hJD
zosR26Bk&jFJZzr(SEpF-jC#u{UUGVI{+4r=ira99PmIO;F(X%Kj)`BPfb^vmwj3#@
z3o$+JQHSGrjLctfxL=`y=6=uNcr(<LNMmmr{?FLGpAFIGjEHP=I>vq2Kh9aW;UDs}
z!*L4YjM%>E5N|u&A3GfHS>pV^rlWs3)E!fIJN!?2#Es6xZw88$+OWrQ)2zkcI8ao%
zPbrN=*{?#x@A1%m@IxMVwMSHYZURH)rB!o$svS1yy}6=NOTBZRSQZ!iYnFH}ZrJl#
zV*Sv#rnzF@u(%y_#m({b4)RUI6Cm^M@aIqwZa8oR&VM{G_MN$6-N<1t&J}l!8v6WP
zQ9UYt1AL7dde>ZWi$A_+j%f4`ZJ#5)^~c{mS3EI#=x1}p>!ahF=ZKYKhQ2gMJUAxy
zwK-z<n4x$JR@a!=ZCT>Au|x055+9F^{dKl@B5CN2v&HL4k$AOCTT<*#S>pO}(Qjpm
zm&V0DGe^8PF8Y=^;^y(OFU%He$49?ATYNk|cI9mG{rEV%re(u~*zdE%^An=q%o2Mi
z#MaFbD+19s%@MB#Vjs#99|oe|nk{ZSDE84TQFTz<16ksWgJN5<MB72pZ)Azb4vzh4
zws`5_=sV|#_Jao@r4JtxTMx?|5_kP<u|7HWmD!>$Ir_dVQ86)g`5du+VqABY_;Mmi
z-jWhqnI-N^iMu09e3G)<0qV8GW1q|tI}RWEV3v4sa%?b5yfZoOWkhT8$Xn-#=cYuj
z%o1Ct#D6$j+%&b?5j;}+GOYmKUOy6Rik}}j48*TT4#zz0H1{2|#aA<ac2&$1yE47s
z%@wVg39WNQ=`8Qgxnkw4giq&+S7%MD%@UonzLf2Syr-%_wL?1>Bjo*%WBgAJ%pA>=
zT;%x)*Ez&>R9xk?=Y)jw#re2jknP@RD))!HV=edl-|kVp&JZGF*dq?d!gB<A2{#_Z
zd)487*WuV|sUv@(jqGa+{*Sb0WRw4Q{67o)-)w<4HDxMl@{|>t8EcpiXPeZYd%Jfj
z(ui9>I%+Uk6^f<5C^<tqb+0<h*JKw(LjfsY#Mj1JK!!`X=MQ0gmF4)Fm#0)<|BHV!
z;#7D=Llrh|RpFNMwGqFfVamNcp5@<Fa#eh7#IHtf_Q6`ja3xo|gY_Mtt_^)g5}M_U
z#L=W4sN4<S(AQWSA*q_@6#DN|SUv<}NgQa}<&nQt@r`dOY-9Q6p=P<NM&&L(RT%TO
z!i?c&Isey+Z)|0ELq5?gm(i`<OWlf(`CeV8jxx(}G%@F+c;ov2Z+Ob>Dt_lLQg{*L
zWsKJNUd8-U#_JeYFy6vg&UiQDy^Iequ48<JaU)|5qY>`k7pi#GGGE8InekP|HyGby
ze24LU#*Y|3W&DEiYsMcL+Zcam>|u;NL&YP3F^=&-#zPnnXPm(}i*YXF(TvA4p2E0@
zv4HVn#;X`_V7!^Jl5qoL4dcs<uQR^G_#xvbj9)Q+&-gQAJEQAwD!qdk4`dwAIE8T%
z<8;Qkj3+Ywjqw7;s~L^_wc_3EUdwo|1@GtUO^h!vzRdU<<5tFPj2|%WWZcF0HRB$}
zeT?oiRr-f9#xjm%Jk)}R^Ysyovl-_zp1?SraS`KrjKz$ljCU}uW_*-!Gh+keCyZY)
z{>1n@qsUk3h+-VZIGQn;aT?=X##0&3WxSj*lW{5IGR9jOS2J#4e1*}P&(`avCFY)@
zsEKCvFyGE-$onr>=`pSiOt{GGzlQxaTKv~o{Hxxo?`Gb(-aANXHm+y;%-8vr>w3P%
zDy2Gd6|fpnIYq@Aw?XoVIakFOyWQlGz=%60c?8%!k*|%^{nn#jLR2evpUO(uzk=yF
zI#h2Ygdg{~bg0>Hyq*ioE_Beg13HZGu-0Na*c&!3s1S`_8nf^6z@(Ktj9iH4>q1O_
z&|&cOT}wKQc&BneMn7+a->Iwg7@|FVU6~}k<GxS18|iH0>xH-{qyt|(l=j%7F8{Cp
zn`)^ChC!xsd7R6*ka02Nm5eJGS23<<+{Cz*aVO(<jQbcPPEa8ZV;s*om2ocPLdM06
zS2C_(T*bJaaTDWK#+{7cG45lGz^p7C!x+aiPGy|SxR7x%<CTmn7*{c_XWYcNm2oHI
zcZ~ZOY5#x>pK(0nRK~fC3mF$PUdgzEaTViw#!ZY{8Fw;%$GDF%0_$Nie8%yNQyJ$n
zE@WKHcqQWs##M~#88<O*W!%a59pgU6h?6;d#_^0(8Rs%CWL(U6CF2UlRgCKyH!*Hy
z+{ySI<37d+EIi5dGmdAR$~c#CA>(4kD;ZZXu3}uzxQTHq<4(r!822$oVE$c(&p4iO
zD&t(ng^Y_CuVh@oxQcN-<0i(fj5`^>W8B9W!S}br7{@bCWt_{nka02Nm5eJGRVr{_
zS8TN424lu~vu9892Tnfwii?Y{@E@5nIVE-Cq$}jrq-zdKO-Y@cGU*`2i*uHob@6%r
zE6=^`^7Ajb*gt7%3Y^W#ELR$)*lEj^_0WV9i@a7=2o3*?elgTl6RzI=(9FAg`%N<+
zA&maT%nuMozhve;@;lC<9Gz=opfLKyP&bWwWo|Fce3bCp@?KGH$rq)6Q185kS&y%G
zykX{X_ncw$(0wi%cD#C?vEv7ek~i%5SW$IU|8k*;A;K8f^l2wM#0g_OW9Em7R37J;
z`C-DYAJas<$i2ia+=SloidoNak!@3cHF1C_vgM7@Sf3Ue|GC7^dLN_C;wALA!$Hi)
z%{0f8hDT1(mtPZ<8j%llAU>CJe2nqtkvRFIaBGe;>z9Kd5i9ms`0Hdi;B(FLcQM~(
z;U8wc$->t$?>WJ&XB+c=3;zZ4ITrqB=1VNROC~&4Y_#z4%r{y1gTa$MjrlyQJ!9UD
zlOaj(ILpF`ojs3dJsFnyz!R7+vhW4WTjvk|0iN>FC|9_Y$4!#&U;b(FlHza7PqH``
z;7L#adPLvBvWx%MQvLu@v)#xSxo#pRUqgEO>Q|84%QluT<?<Hbb|onJXi<8Ms_&H?
zp#kvcz^t1;NAWq#)AXR!U#P-0<mr9m#5Xce;n8sz^Ys@fd4tcCJjGw-mx{d*Coo^l
z>0&kF{9bv!;er+ExTcqnk=JBhI%rQh<x9=>SQEddmpAU%l>DR2*K>T-&IN(Z7D|tJ
zN5#|NX*(?CV=8U_!?BDL@GbL#4Ra)}>A;Gzl+Tx&(Xl4U33KAci}d8KRdQV{AH#g<
z1ByS7<>~zqBp>sj;tl=D%ojammY>Fa+sleK!abV#=6c2RG_=TLKL2&a-@xJ0_qSv`
zn-za0^YmS2;_H7?JZ-n2<2L3~JJj__%vUp?(WQ8Lu8xkEnXl<l*N-y4gZU!f8?cRe
z+Jj5smim>!czRi2n~RFiL?vi^G$9`OL-JK)l)U^tC4x$3-k7h))?s;MGoR0fp>)zg
z@AD%)#{9YQ-RM&0i+H|#B<oqryfHs+e5bh%Jk_squ3tv?|0c^D>k#uey-kK5UU%5Q
z{P%`DuRn}IS5L=}%p2<w#(i`*^TxWw8%kWnL4eX_tWU`AqQY$$^Tv9Gp(nt65wBZp
zXFW3wJ-ps9p5vKm$n$!Gq30y#jdh6su$~JHd0vkg%6gVCZ>&!kdahyKSf_|&`Ma4n
z)+^Sq{(G1=)-O_7{t@O&c^%_#%-1t-tYhRc{{i#Hx`y!$)t{KJ3aIiXztxGb2VkOx
z>|Aq!;^lWW!TXsnx>@n^`<URTF`s(B;(JtcEKV}`^@^9@tAzX#L;q8Xm*1WQe}f_a
zisI!rB*EX$d=WOj&>_DE38ZBeif8jS#mf)gfq$R*^7j-kztaf*2Sd;Miua?O(V-!}
zq^JBN#mjFef*;L%&1Z^#h2^I+U(~F4`AtK}pTc}@i{j<?3c=H~9_jb*QM~-_AovRA
z1$Thde$he0U6L>ARPypWfZ*RT<mtsVIOO;IfIk}Yhbz9B>lZ!fM|yH+D_(wI5AugF
z-<G3z`8_=F^O>(dMe%p4?obpkpP#Szdf1VUtC{zor}#73K5LjSy-@LcS<lM`U#xif
zoj2%dV!rum#b-(FLUb~pdX3^Um>-3SI%yx~<+s}4md<?DjY?jAa}9hx^D%cSeu7^q
z`UmqFw51P+{016uHS@XK6)(SK2ENwNgB^-=$ZwJX-!t@lq<Hy#G4OksFCCx^lf%P;
zfw;w@_%{w!y!_r5<i|6ge}LlUx4pp6X5K$R@$x%e;Ll+`<`l)t?`?s<j``9hikIKb
z0>7U5BHCMxLw@55xP|%LRf?D2s{-G`e8#<sm*1WOuQQ*DCt2udlYfO6i3wavm;V{X
z8}02;%(pQwzwZP;i<z&gSMu^(OyKV_<lk5PY*h`!CgwACC|-UC3G$zSw;O-9f|ujN
z%TPgZWXQjemBXr7k!-mS8U~)?U(>AgQ$I?_Z7iQ>k$;f+N(*1hdYXBDe*^1D!Gt^M
zum2JaFAn)`XW&E3mw&E!`Ce)8BV>7r6)hJ1Dd0&@>bFW>zS|k{ud#eg23I(6bZnD<
zW5ps1{}JoSAEfk5;r^ocKFA_}IocHpxBM+7Pw$JP!?-`HwaCZF;VvH6vhZV>Z?o`I
znNP|z$A3O}nJ>qxaOwS|bmW7l@pBcAds7uHu4CTA_gRLvb<8(e<lkVvl6gbEh515@
zJa&)D_{31hii6&9M@Kw(f&1J7OZ*RJel0byAxApvY0FW1YS}H9dH?Zd{yf&Re70H7
zKfse68ZGg^kL4>Y@-@t_v+%DoUu%iyPVi)Z{|*%&V;mBV+ZMW?s^Rsq0LT9!=G!d%
z=|hyB+>ewVx=*0vYw%>ZT1$L>V?LN=POnGa=Ac|z;yHr(T^2rt`Jjc*WWI8)S^om&
zH(L1fn6J0+|6qQPg};^ggn4HDRm`UzZRVc=Pj+s*SJ_8?BL<dfW%*=_o=)axSa{!1
zRUYy!<$)evqH>bKyr0aD<8R=}{&^NXmolGYiRW_WgW2Zz-@|;N#m<j0zucnd73SAk
z_z#({weY){Z?W*5%<r}E(Zf`_T*sK>IhOfSOFWMNPxd#~ca6L9?aUkNy`xAij>VX0
zlkF0>^M?GW1jQTcyBG$@<8$z2=QvAzer7&k;oWjUHdfSI_~FdAS@?;}$K;sPo5_5@
z!k^6i0t<gW^JNzPYUZmf{2k11w(t)#zum$=&-^Y6{|@t>W6kmZiuoB9{x|0HEc`&Z
zP#r6ZE&PGZS6TQ;%x|~wbC?&FdCOD5)3~CbSdA;VI}*h#A7ha(V?JQv*D}Ar!q+gr
z#KLc3zS6=sfv0xZ*ry<i6H3|=Oc~4Z5AQ1|m!%MYuy9ZE#=ZnYzE1Md!q|sU%ko8d
zCdF=i^-u6r{!`bh8v^-08u(cmMt-#+f0(kHd|wvi!_9wgfV@Kt7He(h^X=mEEad5p
z2iEn1qc9=nv{~0^kox=2=fBVK%>7xl!x%=&qgCo3(r>+>oq6l}f_EgvAJ#20m*+9y
zV_~=QPNkQp)dal?&uBw9NC_u~;m>6~RX!EZ2^`EV%+~~!mhG&pius}mEbmtiFEXFO
z`wE*?_K1%q&z?iz!hHL9{s#FmNLN1Ft$_W^M1x7`t)U4b9EUK!2t3Ib<tswxaa0k@
z*Kbh#H7x&k=KW79Vkz@04gI$(f}T*KV=Z_&q*0f>e`^dqS1U?c0}mX6r}(7CsPdf2
zKE5&ZH!32A^|vuEE>`^MEZ@a?8l#mS*-gXkSiiD^f2$JY?Wy84@Dy(2?<ywUtpD86
zEN_g@l{ZlW-Y#7a81hdm!E0ImDdtP%(}<#Xy>T1!RnIBGOIiLS$@4`BB#lw=DPqI(
zZH<^J`4B-qeUjyyUsi-6{{r)=s}z4W%Wq>o;|fJQs>~-ok~|BAKnHlTTSmGPq<53j
zQ7~4;v$;@RGc7J>zK!`z<t$2>ulYg|94&E|<XI>LUSoMb+s&{uy{nGOxxY<?$k7)+
z8a(gY)>u~;^WqIfEMq<*iCVGYz3Vkb(DdLQ=@;>;oP6L^E(bB+xL#dfz#a~hyy3eK
zFJ}32j;9fyWz5HLJ8&?2SrtalGhz6MaVnmv7CR&`U;dPee+uiL%6vY(=mZCDyW}yC
z`I<}A^<w5vW8TjVC#EChv4nYXkGeMOa0T;K+-?}<{2K67e#*zGoXX*XQZZifrAdm9
zX8!#Nl&=2A=zx+>jZ^YpvdBX4lwU<$@02UXjgs%L2TvGF`O?#rB$X#Rz6v8h_%zgI
zloRVZ@;=s6&h|9g3eQ1Ge@3EP>g8!KF@gDd-ZyT%e`+@KrH3g4{3d+~ai-+?A_T4t
z!{5z%YR*-{a=4Du4d4f39A&KgyI9vImM{NR5pB$GmGXQM0z*+y$<A$M%5D#`TMqML
zqavQ*d^`g@<$KjH$_}|KUnb>46!Ym5EMGcb>HnDd0(4{)ZYmRg)_<Ag`|G)n<;88v
zZ5_)$9!CEEYwug&B)iHwZ(byXz;psEBu3H+A&Eems_N<&2=qKENo8I|cPCCp!ky}>
z+ufC^*VLnDI<TbUuAm|~5D-K+G#Yux!v;f?And5%!Y-Nt{jtI#4l05o&>%rYCHwzg
zx9)RZiJI(Mzu(O5`@i2g_dLGyo$ve3ch3ER(jSeC)1l|edk;TLChE89tA>G7z1kH4
zepB<$D}Cno1ZOo8f8L<@+T(<9R`Z!>TyFxGRijsb#a4ax8pYdR6M)rn{P~N9-z)8m
z>hSN^6(9b-+{;t=Ar;?vwcs|te+D>2eeCxQrjNnS8P`httiD^GeL_i|sQ6k$E?uVh
z(ccw9s4@I|4shm^CrZD6M)7H--*=_7n>@u?4~#2O;BvRpZ@gSU(+A#b`IMvu-ZL!E
ze@F3SZ<PD<<l;P_`P26#gTKGG{MQ6MUmu<a0ZaT@`V#>TDnI;?;%k2;xUH*RqqyxS
z`l9CZBH+{xM_wZ}wfSz{=#}udmHyL;JCINGD{BLNhH=4y|9>w9Kks=|{`l_-#=+;W
zk#@7@c%kC9@8wEu=v#nO`S+bE1-W0d{28S`s`JqwEB;HCzxrV=Rs3GXH*|h6dG%4o
z<K_Ik(ob)aacXDyeM9jQHjC2F_#X$sNPp=%veo^9jNnHia^RVYul~Kf`yRPClfYRH
z?3Vi4eDvh3<g@W{Av8X@l5tHOxJ)Vik?W;|>K1oy2F~LmBl+K~eDzAj*Iy_2H#D!`
zPs#t=nt%JJgz#^)L4S}!KbMp8Prp^@t^6Kv=FiuL-}@fL_uVMC&EpS_Qa<tDa6TE&
zGY<RZY-tDj4jldFQ|S9jKl~M`(3^zZ`6<OW)DDtT;m`u$7oK@>ES~Y5DfwTzP1?cs
zc^UtFMDt%$dHcstkr)3_N<QCG`i+kW-J15}KO26J;J0b{&l!_;cD^oeG(GuB#W!|J
zjVxZ`h2v7rOiMC&qUN((@wIyeIH33~iXYQ`b5OVVS66)feR98F@xJ2DuS<EJp!iQH
z-qvwp_}z+U8bWCJdlcXID}pN<I`=BRev8!WlbZj(Fs?U&%X25BzNKFi(DaQ-#sh@<
zXG!VTBYIj{@irZAg5Qg^U-m2RMCf0x_^}B72FCSefy=vr^Z1>Xdc9ojh>It+zaA7~
zlk<7T0~zR_GfMBAD+RQE@f2S_B7`<h4*|b0o{jTC;8dQYFOl+?yjq>2^2C2@Dfov{
z@K2@SUj%+R#@!x`H?w*6A5!SAe7>}gbG68Gqkn<oC6!CYKLy2)y<hm#@LL&=*Xt*h
ze)aQ0son0pMe*UEm3-vwh<{Mr`E$u@lV*8T@#zl>P*wa3jO$I{^6VERxAPq1&>!9=
z#MT#H3jM26@c#y!`q#-x|KgYe{@vHdd(Vu=1Aj-~>%3O#OW%p3-+4cpT+T~?Q~s-O
z6hiYWKA?D8=c#8(HqO^l@|k?0<g-ubksnj~*8^ug{~rXfaq)hpSE9h>sM6bhUOp!p
z&;M5OqdM;oX_|jheC?;D{N@+B=0)<}w$j_!8B=^n1@qH%KH8@EYD7=C8924i@ZU*$
z-mUH5Q~LD?{ZA{tFQO;^vf@W0{QP0fe?8*w+EDt9bLGACtyB7)^J1x&?VB#C{JB)|
z_AV*lHk~jsz^VMx6tILJ)J6X7Qu^T!$o(CPyGq}V(61={n$EvAzTRl~ha>}=kM7od
zHnd(4`~16LI++hIV;p+v+l0=>`!&F+{A=5!f9*TLzoq!$D+Iqvk~t3nx3Q)WUoPC7
zjPGC^`ia^vmiAV~GioR9)y!H(zx518G#u8s6F8M~=-E=9%as0^Tcp1n^<$Wv9|KPG
z&bOprZq$C<q4Yy4C;2>K<W*ICSmoPGH2;H|&&Kug-q%U7obPOv_YSEZ{!*nsf1BWI
zzb5a!N%2b*w|($@hBuy#Fs?*_%Z({`H3feqaH?;6yOh)XMQ<-?`wUBYKA`3Q2ymu9
zPv|Wk@9&hp{dy_P`<4Ez>m;9Jy6zA0$iJs6-qrxy%a#5b#+4{=d7aU>1^oEga(zn*
z{Rfr)@Sh5uiWTPwa4P@mlOD$po}=%*_<AYl>T89*Cl}{x#W#LJfbZ*j^NcG|;BuYO
zHwAo>KD$ZrwJXkLgyN*>w`ll%a&P1Gpy3|~<98}PtQ}=?2FL#-j<0tp{o#*FKD)KP
z_Zt390e)B>JD*Ve$iE2idBwk^_%wZZ7{BLhKHmgR?K$)eY0s;a{_!_Rd$#8U&q^}q
zS&C==SqK$&#uZ<Wv|B;(BWlk|ZjSR*41AiuhHjC3U}o}f4)j!>j3@VtivNP<vrqf?
z$%?;Q@niQ0V0zvCicf3%SUDe1+|IAyZv!CdUoftj1}=Y}f}dZM^6&d!Li}cJ&r28&
zWT1aOPwA%{0$i&2X2p-_c^p?OK9`cu?Uv73^1|;Z{auP5)%>5L_%AEI_CW!RKi><S
z+IdaqBlEL;P3ebq-nm2b`M%|^`kwXIrW=#__HxD%2dnLWq2`lS{P5ojaHqE0ixqd?
zBskO|{=Fn6|JND4%57CQoL^CVUEh1L=5w#&N0dJ=SN!9Oul}O|zoht=8P}V@<w2!y
zt2|lJXJ>8~ei&A}(&WZd6t{C$_&Wwj^Bl#U$ok7PaO#(hYh<Kcr0w%Er5}pax2N>S
zt`d5iA0NL%%5zNh3e)SZRowXlDY)%Bc%kBJ>w+7fT&MWqn&38$W`J`!)n9FT>>HH6
zt@hW4FOhWbP<%-3V%m>Hzu!?jbHCjGx?G$G6rWapI8X6EWn6Cpmkp)g_p1V0znrmC
z+QH7r;cv`<<f)1etG#c0ewE^-S4ciq-$})tNdFcTA5wm>@w>-x?O&67w<$gy$^R8b
z|L5}FFU!Sw4RCJfhomGoXuJKg(r+|`aFf!%Pw}<)3t;W=G2rCSvHey`?)*=tx4Xgh
z>Rrj>dJp5!d-XgZ{-y?!wG98H<YD_??f_2ZKQbyU@Yh=Y&ntfHds458zWUPL(qGPB
z2%v1_T&?)}7X*K&(m$VZy$M`)rQl`YR37JQDbH6m&0U)R@cRW=R{U<IKdSzvs?Phr
zq4XQ?62ez2{YM$so51A(rJp`q8bFsfoUbar?^~jnyj|0LKP8_{rDT7_%(mqIx)$`z
zKU!ZEbIu{)rjrMX>3mA*OK*_&`J6s<Un2do_5%SnDc%N7`5!x9=<n0^xok%8%-=}=
zE-HN$IL-URcL`wos(OaE1vh=>Clzn2AL$~^ht3<YzZK%kyOn-}HsIiAa^ox=7j{mT
zJjJ&Qo+st8bEgcyANf$dj{dbEHb2t21xfF{2n0#^1aQh{P5I{vE|uf_IMbsYBL1;@
zN<P1&^!xsgWboUX&u?n}tG^)aZ}Y`R=90_v0C1a&H4DC+eUr4q;RghLvX0XW84ps*
z-%CdSQXxF!e8HDf=zl5&|B~jDiRfwHR(#`kCEsUg5$Kfgq;}rLIQZ~RA+&Zbr_jGg
z=?|ZCE+=@l=Krx2`p+u;y80O|R{F04r~Vz%4LV;^{0VT>5TCDVTskBU|F!`q`r)n8
zUtMkI^Y#e+&=ZB<<SEXbQGDaelHSJMk1(#81}-mD`nBH>(4WVF(|b35QF4BszPFu{
z&wtT;rave7d_~{!2H;8j^Ny5!zN7iHza%ewP#fZLw@7_Au9Eh*arZRES4V|@Ro{E9
z;b#gZ?VF|FPT*9pb)6q`0z0o#`n8`I;8w+dUh#cR0VWlHm*$gEKLdRmntq#Lr%>EW
zG@jv7;Ou`>h0%EOS&FycEa_L|;*2X^`aJ>E40T@2xDo{}J*A(1l@x@(O^WB=g<u>)
zy!6v`AAu!lcv2oa-_P3fk{>5L{%ZiIcAK7({7r5gc&X&G{tan=YqvWXkLUA7r7yih
zo>_h0kwSmJ(yzWk=-;GS{!t43-zdHFb|JKO`@0nSb5UUChZjh_UMtmb+KO-JIHhm5
z)9*gu)Q`^B<(~GR)9?2gm%oGm|78mPkmh6OU0$vH`P7$5{>Sv3M3Xbu0H<=U>HG*W
z&c7Xsua8Oj>ATDHdzmZrrIO&bo^{*4<o>+_IOTuj=fd<)oR{(p-$%LP8$xn%E>nC+
z`STLRpToG`1THr#{n0NAXm-bv;_H7X_!l)zU-7jisn9*p|LFJXl>FbGg5MAP3P~FJ
z{|ihHIee+e6aG#cD88-u@Hd6r?61c`;iP__j?BMLV?2~K-#lCChyGHYsaSNzfb%$d
zy|nWt?Jut?`Ir7o==I{P0w?-?&lh_0*WIe)clFysf5Dk@-LJTv-^t&W#G^Y5e}&N5
zc)trc$%m+(|2Cy})PHU={9eW-kKq4b1y1G6+!Ze8#zHc`JrxRGlAO6Z1z%+xe#<Y)
zh<<@woVygK?-$cg#kBJd!(S=o;qN-*`KJ{>takA^oXT;&p!ks=mk#;3;@?qx__qWw
z{y%$3>U-q7lEX#X{;R-=PmaDuaG3x6d%NPpdQKghi+>-`eA;i8ay~)x*#`TW@>x}Z
z_dcb+N%84-3SjbcLGgW#l>e`k{tm|VCU7D9obo>wSvR>y@r?-nVa2C4(j%|s{A^18
zUrE9LLGu|_eZ$)G{M)p?|69ucEolzt8H(Hauly}U)MHHX!(Wq1nSHcH@ilE=(|_g^
zpKb~LZf%RY;v0V{fcZC87+0de<$IRTy9Cs3aL(Q@<zLr@K{^kgeos?;P5JGUia*<M
zWmr{noVz9{pZISI^fbQK-YN}aa^OGJlj)z&ION-g3>1rBdPxfX|46}~g>jWsp64@;
z@qUH$mzo()F@@e!dOI(ezZHvWt{DE;Qs;e=o%1TikNv9PXm<V`PRaiRz^Q#!hopVp
zcb3qff%YZ*$PU46e!L8PaE0a>yle(e<D&E{f}XDrXBZDs%HI{G-*}_kpQj{!#n)7B
zs9Nd#SH%y1Rw`=#jW;WP^iz_->$Dx-sW^SZnSR&G#rcTh$37m;=MOD^T<NF7$oUJ!
z?Hp(Releaur1+ZBU%;u*26!K}!?X^pn-#wtIJM`|$4kx^XgS}?^hy-Cd@Ke3oaVo#
zaam8-G+)*HkNvFVe@OA?z%WVT!|ND_eWZ$mjmNzy^zT*r4YkXjp!xq6aH`j|u50M=
z1kQ6iTHkMrJpZ)PzY;jL&ycp;lN4W6zP0nPB^A!eQTl1fKl*K!i}P0C!}@pda-Y(-
zKO(43+s>yHU(<Z7f4``BM*Zz~Yi8e4d|2&eoi=fxgO*=2zDDUUWIR|B`TIwJbHBex
z-uQ>w?+c1gpCR~D6kpZ+OFDmDu35fL>5q*Iy^Y7W^(6mx#DDUqz^R>&XdD{8FUr5~
zFs_*fE@$_JzI2WhM8%48IpYCB{WGWZN3ufyA;q^^K6l9rKc)@6CncZvD*a)d->u*2
z+s9P?wFSxl_cZ@cC_Z#RC~s8-`mANG@8yE)@|*Jv#houn2FB-k#WTkQzfALg5#xFj
zxV$L^|4<74NzH%gnNt3uT%12sd`-uZ$&GI+zW$)R)Y{?9gHo@fpB22S@_7?*;;WMG
z8!~->zA4Q8enjXml8f`JjO$I{@?FhmNDXoGS3Cm(gz_mx;y@-0|4pgTdvspDQE?|P
z_&uBCB`*a|<=l6<Fv5to!|jT<|3zv#q4`_{#f;v2Sl9h$w5&g@__2$H{${1mFdnFo
z{`p>|A6^&0+UKK+w>5tIK^@qCVEMmV%4vGh)wgT=e_rzWjOL#OPW2i-OX#;~hqakr
zi2|3OOTpJP|Md?_1|QY-xfl3_f%Nw0XMxjuk7(QoePWA#UwXOryB1J~vGaAtg(mbr
z<kxTrkekokp!HQdTC3%30zLQ3Cxq_jH0|YzPyd7f&s991lK(EHU!wz{@H?XXTvmKl
z*RM?<SO(7R{7+JnUE1&er1V27pNI6l7yM@_=lZ9mVeq*t{#^!~@|o88@%2hyRr-wj
z$3CF=D-^fyJY4X2dGHp+SHCOw_bB~)7}uM?<-Y@`_a6O6DZlBTp97xMznfmE?XUK-
z%{x~pzOM7vt@@5>!#^wl*qDDa!0El~Gs2(7C;KrF2wziwg~`1Kft#$*EckNns+6a#
z6HZYJbP3}@O8NT*z?16rGNuRL-YIlXlZ$ge@niQ3@EXNmt@u%0Ul>;WO~8}#|FY8W
zi_F*GQhYih$Ie1Ih@V%#Cl&gz_V3dbZ%62VgmEoI@UpnrU%0kjFE`59c597x-5aU2
z+YXMOb-hCsuibNd<@vgoi6L3Xy?)2;Et_|^Znr=07G287<&4}wM(*C(-i}Gvt+i^+
zS}&H%Xbc&PA>%P5n~y=cY#hqPq4@i933*M#6HZ2}TWK{KttK~CrPJ&7`U?vq6{qTT
zyv16#=XKm(!>!a?O|R>?Znfnu)?4%Cx?AnFI$gKiKja_-w06&{j!b68(V9s_?gCoP
zEq6NQ71wL_IxEgXr`+(|YQND~L6X3oOYIvXU0iI`Q6%m^cVRT=E@a1@MYr24_dMtN
z>v!&O=WZz#+}YyZg4}M|v2}akmY?q0yi=d<F1ewb*;{sQ*}cOubmvjZn%8mYtz4Fo
zS{KFdbbD^4T(4^}7QK4kJ<zXJ_H${ZY~30_DxBPUhuVQVPC*a3#3;lKn`hnPuHsw>
zxURTs^A4&AzrN`mbbDU2+v>Q)(gK#N)$m<5cw4Ci?uelW2^)P>$j_!)TMpcHQ8zDm
z@AU$AmG(;DuG4A<NgL(1J6~&-J3>P}Bn71hm43I^YK(H~q$k0$59o7&w+&FqMe<&*
zb-^X=atGYh6=<<jZZEmSjs2eI_Le$cxhhW#TWEC-mOEAFK)>AV)o%CPPV1n1An#_}
zj<*1YZ&tk6vux_KvE!s0bJNpJhSFVs{Y}>e_?NqV*{fg>^#<+b17M%37c0ItiKgNT
z9l%=>NV750tP9d?j1-lNkWzIMb46Waq$#;lp+YsJ3?<iarKm_TB+wI_Nu*mLOxlur
z^Z`ap>H{r%EJ>x3oCHHE`2j{(@&gRC<OkgYDrKTC$AeySl6q?~`9ZheNPR$Tu45$n
zriNE;CZ{WR78~V5$<I1o+bgG(P@X}GR`w?+W+9pWxY_EY)TGnsx+i{moHvvFYowLC
z)rUCgv{D#_#bZi+yp)z$ga}P<d3$=C@>Xl*l-{N%s7iM~Ca8tvVhu=c_d2a)nr>^M
zhip=M*`6M!gftT*XU%g&dM}sjwfS0dN_+*Nkur`et>&_uP3a<jm`i!sZnf&EDfwYq
z%7uElyX2O8y_9hx&lgg9VUYSbt$483q?s&*jp%8=nf|;_GjDQJ($sl^$1-szC$Dr`
z-7c$ZEzJLQ+?*`kaViTT?RH7^Lw}{zc)+@%N|;{HI;y*OZ#G-?Dy^#L_InGHrR|l%
zc)iuy58)qG_NN?o?~dGr=<-kx8jW(tEmxPznCMA;tn8oJKHD3|><VSJ2yHBv4bysc
zuK@*Pwp^UexP<~;bGRavTP$q{M#v-~0A@Gy?PlD@N|H^mJD@7VrpS)NrhrD=f|A!m
zBe&YcQg&R5(QEDZnr^k+E7x~Tx$a`K?^X^S8gt!NyPF+zi=>^CM0acWtGG9pQL6<5
z2+ACdUkvWT=#_4!P$(?Vxw8}QT*jTty7K0{%WPMJHcu~4dhf}+Hai)5ZLigK_lI99
z?|r1Jd*b(MWut77g$<P4@`37RCtRLw&^Xk(W#W)hW56v9bbSNOTBz>=ADnET^j6w6
zXxhQ%*<yVXD<-q=XoMv@-fdu4m?F(;4V>Y19|j9YxtO?_*q8Y`HyR!b{u_zvI%0AR
zuKbBg=R)UxgohdpzT*~dTha*3O~nS~Xxw5RCDz!H9p72l>CO~h;?8cJ**v%PhD;4`
zB3_u-4==&(WE}6k{^kUd>~fIggan*N&R$=*dGn5F`t*l8#?jbNaJGy=N@TN;jJL<@
z9B+nvhU*;8H?bMA%d!0O<7pMo)CLugvNkp|$Dnw@#z;c$$c+gfO5MyH?k<(vo>nwC
z0&W9$-p0H+ck?ccRxty!S$CGa1#DIpcg?z!Zn3!x=?bae8|(EUE{iwfRa5dRQYAb{
zV>7**5CF|`eP(-B=aPD>T-B#D*(^JZ=HWN-I<*=P?J2fTFxb7)QfYz3#KSF90~1~y
z9tX8tF=c3(P)mp^?(uAHzTB-<Fy1<~=3=pe*(F<9!kpLZlxw|iu|hLg1u_exytLgW
z#z@Z4G|G*xNhlDb69!EaA%2*}7(Zb#c5SEWt5Pq-S#K=v78{F?=&`uhZ7j}e<MH`<
zJnmu}k7qv~=j4faUK4Sr-oz*mG37XKndTYm)2xbKQjxqvz0&qtY5PL0hW{(EaWfJ3
z@lDXg0YPFyvJGzNx!o0b^BQC@v<lg-4jt;-lOc<a)~R?)OvM{|Dqgm1Cf*d;OuVVH
znfM#BnWX&UotaJ4wi4b8_p7`2(Iq9eWx(l=wnB}&JKavd`S42i$V!e=5MU3tN&npn
zK4|vn@XIEw5@d4;qbL_&g~;aO12vnAi?M8i#*`h6%eL%jLVnPrVr9-JjKF+chG!F$
zgKR!wltVp;y(2-|XU7tx##lo65~hVL<^^8($xgLvUd20D>$2gB3ASJ9#b1x*k64k$
z6RManw`3D0m+bf$UOxwg<x!`|%svXsbo%Y|?>59qxtcruw)}T2_kPhuyI7cOl=pk?
zf&GWz^e6RHyhPVxt*lkSQ@)b!-Gl~<rM~NCuoCAMc5QRvhH$Unb(34TLBSig%{cD1
zTd+D;+=^%XJy2eBgRb3aV2KIqedH|(tPfVh0GvATquGM)?(43bEzG%do44#JxZ=AC
zttK_w^)fs`<YijudG-4A^z}Q6TeiBnk=zK4yk@zf(4IZ+=t$lPIXY0hZnw68wL&ak
zdY-%3he1wDNUZ!1UWD~Zyg-t&;UK-X&}lWWu2_#QIgPp70}CCl>vcNK792yBau1(H
zjTRBZFq`Q&YX|zC+wOIW<`j{4(CfSoNb*&0q1><cTo@5psug$ESjX#ny+R583oV{1
zoS+hFu?5f)yRw-@ETPN&)*LheuB5-}z+Oi53^%AgqZLFf*1})g?N#Ak4;P)&57sy7
z9RufzFYn0LK%Td4{wtSjSQ8ypH*a%(1s)JdJL%5M!_e`(;)LgN^=EtecDdFmlB2Fs
zpoVg1QT<9U+wB!+M_oYqy4UUUE9gyt0&ha=kdcYpXM6Q-v9Y5unk(#RjOX(Be{^bs
z{?FpS+}QX;esXj?JBFJqA^F@?erkLoH$@21<;F+Hr*f0y`3#W|n$P5N6Qh}lG0HMK
z$^T_@`O(qb*k~caaD#`jmmL-Ju7&BxD>sUB*-<$2@~FqSOEz$Bw1IEudd((QKQU8b
z%)oiOSZjL4Id>0yVTC<#{oy~jF!#)%u6yQ?cF)`x{WnhkP0)XEXYQGsqW{od^gpM`
zlHXIgMBI|S7TK9-o#6a5E_QYyG;Bt4x*&`q!B?qFA_Ic+qNUh_wce6+o3anq(z)q#
zQ`WIa+9k6bt}l!b1R>C9n=heZ{Jv))1MhR!L!ZMb8!CrnObag!7((Q=;TKE*sZ%Q@
zG`rZ#fwo7pO52myt^mj84A?fs{+`Xdvt#6LEkxTx7aaquO3c>cZ;)b=Na5n4{coTp
z2xj%t_T>g-($+nDHg74q*^yC}Qz$|%Tdj2~9gJ-kf585q?~}MF)(Y7Sk6M?!5nFS`
z6S+7S2BUBRwO=3?X8}SKE(0l8uLkEm$H0h_gKitTG7w%e!jJ}6f^(=t;Fcg^2lvAS
z;wi9h1al&>cl@{tV-$W(rQLUz$_O}Fw$&;Hmb(VfdV=ebw>&JTg<TspxY*$Xl1k7z
zDodPfUdmQm(2h-Dg2J|JVp;t%bs)$aWy#t1&|n#UJ9preKY2weEbWuRJPG9xW}G08
zKjIz=@CJk$a0Es{w5@F32@cutP(!Q}jR^KXSl|Xah-QM?V!zebp2ZS6LN$W&-Cu3*
zE;h9-Gt8+R{6f0esn(3-u?p-l64;Q)5Jt!dVOLeJSFY8A5**Z%6beyyd#xhI8Knk?
zEb-sFz^;%AJ6n(x_^7x?yt4$K7;@X}<}&U!o#6Okm7m(9QinLY<53E?-Ky6rEBpux
z0yhAX&>nTD?U;wSpNCM#6~8I6lR06bF|)n#AFaZzlFOo5RCKGRLaJb)(I{~@3D>aX
zrVf|3)W$ni78=K-4Y^)u#D@k(TXvjXwIqk2P?2f`RR+UoCYKASXCj5WSQu}ICcOEU
zY8t0nC}N9p8xpTdFAJ-bG-%zuD44HZk&@N*YNb+TCa82liR$!P{fLgr^|A3gk45g|
zi9UBZ7YnKh>Y6M(!ckN_D_44lgq0_2tu9Zlq>hDT50wszYOABDs?eO{LK|8jw1tH>
zqL<v_Zm11aTGQ@g%P12BR=A-}%mpRKq)2C)(3*?s5?Ul<e3o_hj_OjXX||%lLScvU
zJ!o@y_5=E@lqwj=LXtX3v(9F)m<NA5EdUj;ItWFYR~Eq`Sa?8i**w+>*Z_(&7tcKd
zu{ItMElgfSjj&7sc}Frw$A(GAU;~}3$G=3&6bNV(<~D&<!0I!2lPoo}(26%gmCRu^
zd(xfV0?nlW`(T^9t-X6r=YQfh4r}2$WJlo`WR-xse`_{<TB3w%5OC5{9Y#7<w?j=2
z@m_4+n}=a0eaJQIxG*z_frEWKFeqRGxj95s2Isc<eyv`mrArDIMEr0YDkmAFhsy3e
z!eW~+N_&;1pdO~<j<q&!XfPbbs<`bXeZ<C=i>4w@mf{U;#L+>lAK64?Fv{x?H;Y)A
z+r99(!bqrl5nGVo8j9Ld4GA%)8p}pig*iq|q3gEG==k|TTgc#VVexB;f+tU_n_{xT
z?;jfRq<U+Eh>2?YlC%i%2y1ZsYF)ikqaq<=vnwdfHQ0Z~qZogKFf{as1^JLD5|ISa
z1ZI|+W=H#oFQdrks*9#NT@{8D2USQS3JMe&=qUjca}eo5L0Wzt8s!>PDrPr-Y)TM2
zdjJO7zXLrsG9$EQ#nZhFaq3~Fktq%nQJsQ$I%N@p2_A^?um>VV_T@5F1cj6G*aJta
zVClj$An&0rFlHB|TV6&`3hZrELL48$vIx#>x4N}M#2VoB#YnNE@vf6ltNm$@d4;nH
zj3r?u)6Nf;yJYM&bg~)HWo4L=-{H|KiDB3!c%I$fVA*<R7RGvEcDq}=k)(B0TyyBQ
zjtz*7Chn8if~1bkkpWYjM1`)uo2A~#<`KV-Fsdq|zA31xP^0Jwv|@1lpqYy`*pi}>
z*FwrV?l2N;Hmi||Svv#=$C%8XP|tPtgp6B+SAb`hsI?=VOcITSCB6StCluDpz>bSe
zig?MV%qkpiuMPMJRCBjA)+1>wMP_$uUc~W)7(3wm3Ux-@a@v|^Il(cSW;yYrWXlQJ
zo!D|Zu8n{kLVD2Vkd08o!=0F;VUksP6p}SFTbr4Sbmo);;R0q!Vtqs-Qn(zZZa7h!
z6|gq^OKqC2DNGP*d%51L!OK{lN03}n1{PV!lfTG1#NEMUU9MGR*G#ts6SG-`ueZ@}
zVrV4g5?iImktV!{!G&7~+F3Y_$h=@cGewijtf}#g>b{wvnsk<76Uk#=)`9(L=LURb
zr#sWBv5qK?uy#rO2i;K=Sl{IuW;-|>O*+7R$+XG#dOFk5lVqMd36^6slf}-YDz@-?
zm@=YsMF*RFM9$a)_rNGmKhUJh^Ie_lny%NF4@@t%8AgQ0K(oUOntk&V(XId(eq)F^
z2a_Po7S6!7*FBCqG1VhfKt}KYBAio<j#?AmoEi*|&SIa!SCd7j2JsWiU~^Mprlt(0
zOCA0cOP(Q295Uysej9WeP0ACV?`5k}YheLH%~|d?nWq=RoXZAM7Is*V15f*n6S^d;
z1D9kKtW~914)x8Da%YYOKSq-^EDHQ2k0qu?=|7tFLu(WP6??HyO1_TnfNsS@EMy5t
zUb<DvpAzP7#?&|F8>;ZkBNi;lnH;uHBcl-t>A=18uVVu?PGt6-w6akgC>vYUHWYPL
z+7jc&gIEdKM+iTHei7jX+ECXrSB?4%u-CA%Kg=@QeryCAv_=pCedrJba;MuOqo}vy
zF3a-#Kx5D#C)OCk3_X%n@2J{@7>wBb((QM2y^b85q46y0F&in|RV`8%#Em7hOe;Rg
zi0~)AgtLcc6fd9+?_Y-5SXsj242N}BS}>f2qj<A~?3D9u0Omg4?!rRd*KJP8BLZ_g
z?06VK@COe-gp-S~fCT~9{BcyQGVLfdW@np(>ZO}%qxMJ>^Kcn!sS!Vr=uLTugI0(7
z4HIqOE1G86rTv36i4Mj&JSjAZ(9$1;SHsW3PH&3BE)+)Ft+sDxRA_GkIrg#eL9YKw
zs~cf5d3hELN~8@~w?*;UW*|kflc*cE{+Rs0F5$%jrbwZMmYSW?B_9lGpDJ;5r;4QG
zsitF8PYTJzPox$$vTh?qIIeiJo&GSmA-0Sm&PNo_wm=-Oq^Yr_=qIvjr`mm5r2_94
zc9SX}vC90_1D3p|+BC3L^AIF+e9ZeZF1gWYEhVrh?$1rws_W9?Lr%SX92#&U-=0oh
z`#s$?Lvu1X9Gkki$YwV-y%w))AeI*?uyv$O4k)=Jqq$-G0L{%pK80mn1+q-ATs4LT
zu;pg6<_j0NQ=_t_H;U~2b!&`+@Vc~_N*D<gOU!&=HZ|+6s)0%cY_sTr92yN9EXiDk
zz5*I-kHv=THtj8Ck>>Zis*1k%Tv9Z>#lgrYpAol&1Ue5$-Mt#a+HAGMdZ50Slqlv}
zf0fp+sc!^w%`n&kR!Zs$J=8?VE6gu2bzyV*tHZRYz==rL51F}$^W-L=y_;N;^p)~|
z!&UwQM>TVKvg7LK!Db0vT7p<eh&C@2CaNB^vDS(R$K{#XtXqO>wu`vl@T4gjYsdgG
zY?@XQ^Q&oN!L&BH&#k8uC;KtAYni36r5uKlb+N6~{p+hUVP+fXHpCxQPa9Oij@T)p
z1CT?l;nLMIM`EpcYz3>!`N-hW3~YL5ks+=8AGUm86;$kT$R)&9@oK_2RueGKk(&yN
zJMMPP!lci;yD(q}Q!3Q;K_>GwHBV4LJj7O=JX4tBo4w*Zn{BchArg=yszP2x!KItY
zE{i%4;ogLEwF^zG>OxL)2M>nMggIU@G1VgGzRg-)ubw#A{(}t$Js4XI4^3F20xv?5
ztm60&a^9H=Vk;B%@(Lo?R0o6yiN>@9rRh?t?h+@v6COZX*N`_1bO3g3^H^?Sbro_8
z*08=zJS*5kxs!B9+Umou7UH^KG#xw`7SV(!u^>8%>?vKhyc{;{{BS205Nytfm%I`1
z;bbRN%GwreZiG`(1E7weP~-F#SkhP{>bQ02s;rIi8Xo=|S3QjUq=S9nbcdBVht}-|
zoKG>2+*%YV<RbRbEF-efuy2+wD);@vPqY$<>(VDYBe2<zmaf)IVe><X0$c4>7b87R
zd<;-j$W|9)$k!RQYpJh!#fo|{VRlpnIhP|ssC8I;TS4Fv3lMa_1}z5im-29ZIUGzl
zNSL!(7hZW7UhTnTLc5-W(OrD_iW*hEHcR%4tIm_C&_YzGvjAQqt2z>-dN3SOAaz1@
zDR_}0C0+Lr*+=1Y+|k@mZfl-)lu9yR>n7#V>K|I+jTsS=aL}kYrNyrpdb6o}sCCH`
zQhbT>MPS$|F?+7ugJU<Z>`ab1|25KS)a7CZa!<}z8WAt6-^}FD>XX+Br3>%gr#!?<
z)M}%U7+t=~3h4tkEPK;JFe#nvL)GRD1qzeVfJ3F)q-$`nLxA0B)nYK(sLXnVz@Q~@
z3XZfSn@l9lnyrLhq@^1g&77OIwnR5F;+q5TTn5Ymvrrb>eO@mQ3E?FO-lalepyW#%
zH^@v2REX2~{^WmK4_304x5Xlv!;8!Q+#%r~(o{A?^6I*Uxg!rMD;0zVEB6PdgMlfS
zG#ik!A{bK5tuOvVQW^Q=p#iV591GfkOddYtgvA)i;f1c}MSNmR4xhjANKDP40n9|f
zu_CERI<|rW9b5JX@}BrfWR)LhZOqc`dZUsl7cG>>;KUCS&cXPC@+>?8I&zyFI>N<(
z1KlsY>(1@mid6vK6rg+aTbL2xOa%=^{U*9x0)e?ojZkL}%%z>B2o$l!aT%5)v@w!I
z&sc?IKKw2)A9o@ivWjCCXdaDv%gkr!k0J6y3`}a{Ac?6Uo_K^jTzT@P(4r&<6hS)*
zX(dUKNNEf?n1Y}p*pVxPumd_iW?`X@w}l1E0tM)XA!)go5)_ZABfFOO6mT!g`6B_A
z;E_2f7N0;w9>wt6IBmiKBr<?>$&QYiXk+_H6{mqk(j$-zZuzM|)CA9(DxRr0dTNCT
z3W)~H=83J9_YwV5m`PAFvN<lBT7e=ZeDCZ@hGR{_gDI}Euzut_<W5njf<sQV2m^aL
zIRp%Q0VtMWG8p<5n3|;CrnH+M=XA;uG2=<)Crgo4evE(8I!y%JI72%fC@2`mnVcw)
z9l3A-n;47eaSV|_fDh_YtkrA@^0)yCHN0-Xu%tYz!NH7xg#;`~kRy}k5uM>l$DkR0
z0lwwP`o>@iAlEx~OL5WsmOSZ0Xt+d_ww6aTA+)p-7Iw%NHOA>O;^(JG#7eo_3p1R&
zJbm+kRy$CuJ_T{oTgPZlT7(IbUqJXy5!)2p#=K}?y@C#Ae=QNJR!Ha!1V8sn&~T<c
zk%eVMh7dPk=L!05Fd_mzLcDztVnuyu;U>GFDyGDL=hGLI`B5n`{u|jPMln$Aw;%s~
z+$aX*5Q<{RvAd(d%ktH7)KDV%g@8p`RXbFkuPtY3jjLOuJ#knpW{ZV)@`u*gm>+p3
z2?E{FNto$QRV?_s3mS-EPmPHodz!6Vuy8Ku3W4{8I@EvbU|AF~Vkr(t^f4Ad$inZg
zG%Mg^+TjwA))GHxsxt+Mks2;)2n|LAS(?Q`EZ@O77hwFz!56{(Q;|J9BvK+FrIL9-
z5%Sy9XT#*!19-J|!sLNx$@T$6D}%#_xSIx+7WtzP_y#!_o&rTsG{xc#YKmy$w19OL
z3U-i*ohZRuJak@z;Iji_BEHWP34=-<>tT-$*Eu4_5aTnmT^0vftLH%x-lFCqVxJp3
zu)r#D^g*6Aw}-z@VY0Xe&V=kwc0RE!6z(v772NmDE(D@SyB@Ku8YT&kj+4|Dpg%`L
zIQ%g|qfmHb;O$~wl1-^neOdY$=r-z3Q7wuBL3;2ZAu3aXozx>2(m79T`V3~sz>By^
zWS6GfZDY-zc+A#){fNRz`wJ>>P<otX`_R~q^ckEU>`h_ko-_j1vE>i)#hAe0@?ot=
zSYP9fa~YhOQ-!_8d;Kf@4xJ3U60;3R)@3JX69Kxsiv0u#<!Q9-^d*$((J@r0YNs&~
zjCBxtE*n<j;+WZ3r!(2^6_~Xpyz~G0S(|k|Y9p6cl1a}4*Mvtqa~B=v>hl{++62cG
zv?Cfb6f~$pntfQnI(6(9NOIlLX|u?z*D|Hg-?WYaWP<?*rRjNCq(_mrI#H|UCnup$
z;}9+0A<cV1XdK73Q{^(GidrQa*t18Yu#mpLid$%KNTv=$Oocd&I__x<tlESvMs=nK
zcN%e6KW9IfX~~Jotef6Ird4Cag!ub(#FDSbV)aV~#bawbW1c0z3^qAStNFUmBN#a<
z+dN_LJ$8ac7uanc(IDay@w7OG!+|KA2B`zhU4(-QQ}jy6-XSuxcqGY0n$0tjZu9V?
zWSa-sCE7f+=T-i}Fb+h;!C)A3cY$3nayFAbOAEZRlZ~;!R<<z>;YJ1&%P9sQCygx&
z3#sf1oiAV;+RS!%_%J+Emxuk8!N4Mch++s?*5oNv%M4>$jcP5(HsPq670@0-8Ux!{
z@cMDCgT+M`o60#t4*T$F`(Ma7lLhCMnpdxeX9-(XmRKk>Tau^O_dwF|AOjENKy<P@
z2?Yw6>NS|^kdx)e`?%0@%JToWTzgzgJ&jqFuR@0$_%nOFDIdcs5e;fHU}+;evkYz1
zrLgVTN`oH~&q@rA){qb}jD>mlDoh1e##9A|mmA1jAB3$32O`c>ps@AZb)=~KK1boc
zkc<z1W`1nS*<hD&L--rA^F`Q!v1nh?O1oiU!^wR2%t;zpXm8>Nj_9Fn`$?PF)vc+A
zY_oHSxrQf8U1KI+XLgLYUGtHY{cd?tcDjf6UrS!RtJuz;AR%i3Ai+gE{+fWvW>hZG
zt)CfKq^E0PLLW4xQB~^5I)Ic}RudBUSh|SLn6J2%F0EXPAw0<Qw3G$t^b~lTtkctz
zlhElg5XjRX*x#+^><-og;p6kA9CvA88Nh$gZXFD2M5>V6x?x4um}TbCc-cyw)}4D{
zUj$Alg)_a}qp<u1^jd^_DS9CI8gtzc?iYpM;w$}?rG8VtMB&e&w83>Shlail5ZoCo
zVsC++tvJ7c^Q9ZXb%VD^B!5JmYIr_5Jlxp1y*aV8An8WkVyo-!hGT|UD<Tf<7$HQu
z43<^>nj>hM&z6XsmM^8BuhU9Cq7D)yYGC4eta;$06PM{$@DrC8V9wCygJvKwBzCA9
zoiYP|8lSyHGtJ2y4+Ob;e7_Y8XFYfMgdHmLTkKGU<L4jqcI)e6?1I2(+W!!q5}2>a
z*UfH1)nIA=O8k(6XjBp%am2|(pU~jX*CcJ$3F*UJQ<JH6l(-I2M(%SsGU+U2TJ$fR
zXpyhZz6&f>Sg=-bCF!(;sx3Z2v?i=fKDKt%1`T+e4k)`x;jKFUj#8UV5hERxwJOqS
zDrMKIAHWp(pwB@5_+NXnI0@KZ%q`?u<!^w<0eg%MlKG#WQdQAEj>0G5Eg=-nH27>6
z=YgeI-~Fzyq6H$>7f8+xzM(;fl0l>+Fpf`3KOIiNn$?_Jx~^7&H42potI{P%YwQ5w
zC`{f*)W8Dt?IK{+wVbD88DgGfh-o$(<3^)s6(o=JY&;${HrwR!ZM6lKJ@F|dO(v9N
z8*pum4!m596h*pcuu{t*qj=?M_>YD6NL7rxkEK>PMa2#lGBJ=4Zb6<)imXc1ayUZf
zr1Ttk4Cn+yKNuiMPeefN>B(8$351xA?n+}GhcGcWn*Qn*=O}pDzNd&lC;FP=9DR*a
z2rcN#OnA71B?OHE3+QY;X)j{s4o=#@x_dCv0NWw4bR#(t2~Gi|!)YA%rY0=5syoAR
z&k*_e66WZLT2SdL{=wq{!Oyhi(c+7-=MKDpQ*?Mk6ges;7HKH<T3E#9D%1PbvZKlK
zkrc*{#q;2T@|WyZV{`uQ%5lCmM4U=|kK;KLHhbxJeH@A|O_<5!K~ut4v{WA;pOPMv
zr9lX2&)39>hX(s8LjotJsHR96Qp4obg2GI~_z##uq?1WWVBvwiM)kCjBohj(h@GYK
zAr6f%WP1xj1`89m!45T&?>trHtFN@#R=HDS?ZcJ=tqqHeHnDmtwS?0STTj!BViD`Q
z!MuXcg?6BO1-6)zrWBlG&7KMEI%hE{BEZ&3r9Nbb=~)B*WUQe`QlV(P)9FleVI?T&
zZw|)ZkReJO8_=HEt_;UbX{k<9r>1T_uD&eU*`U%rj66@6ry#sl#H>-<lsVr}*7Fq$
zm`wVCN*Nb{ZDR2+zwmT{xK<t^Rto!=hH+u&t4TbM(Vi1IlTrg)gK7y_koJ5dddxN?
z>vzUTB#=FihAvWYCC_F!<6sLs;kk^RRAz}t`_Nf~_CeVMA_tu?6QR@y(}6PguYK}u
zY77yM4Gj-OX>vH3XxnWzALe-IRs!fDMGtGDCIyGOElCN$d+6gnh@oILm-hE8l$Hx<
z@=|c-gT}`09QJNv<u-QMR%<ahuqU(&#2aSkV$s>j#|=^qEduSN;f`&u5Bs#1$JUzL
zZhsRKueu;wA7|*qS4m)R;xpFbZN|y+td0<`gn29w(uIszGfULVRO7M%gD=`?2a@@N
zt>0c1M_A$*ztGq}m1#Xqouw)NB;0m5SsI7t_-?zvqEFv;f((QjeV{iqjg#F_J7v2;
zg{4LpQA^qZ8vd);Fr#hf5yOmTD%n`WL9qvM2t>pQf$|Nu)<d$A@aOLo&dr^q7e;j(
zKjc)~q^Z#e8VP19gha&AMmO1SEeu1l*Jl`RvcMwjBR2lRQ!3A!kCoGscrb<==H#~h
zoQ~=tYrZi3lK1-H+$PM=k?>O<_cD{{w~mGN2Ri{0!b(^SM@}-0vJ;H5fbg)jU*B{y
zr-=v-4cf%3`73n3VtMl6u!Ye3{ct}LKTUVPeNb9m>7vj?W_kJ2qzXbiqXMT-N$=T7
zzPxZ0ZxUw&mMkK>a7B<2UzhAJ9ztzrq}l3uBRGcsT2eMf@Nv=+I{DHWsjf7USgtrX
zil5*_QMOWpH!hxGBVU;k=wI!6&l%y>krDhkvWRmkM(7J@&IsEHBltcuo$)f_Ex|`i
zACX<EA``pAt8iA0Z25o&MD{gjhJ&y=<rT@p{-%8B?Z}c#T5i-TNZ;zwi-GWKuzcx^
zfQ>lb*m-1rKZI<~z`rv!yLagycD*$68lnyUo{5`t@z2tKT>r4^aF7^(pB=f+;{JON
zJ=C)FU(-MAnpb*DcU~m@3vmAg{Im4KX9<B_zxE`#KL-~oyQP1+z|K}&DT$>&?;?3%
z*PRzCp<bf(-wyntK>2T`2Eos+FVX~7etVBy_ki{Xf%I>=l9M}jUDNbbe)?zS7b?eD
z#2=Km<u`eiJh5x$sd8`cxAa#31Dbw|kU2}#p!nJK@C9;D=_yS#{mX$-5g&(t{40g8
z#9zUmmBX-Q{1Hq)tm*CAwijxGp~&^lNczLnp!nI<>M7`;{Fd#{;6d>H>zdxKU-_b6
zGtR=^Z}}aLq<_m8jRwcD>l68I@L#n4KaWTBe(OJHTGHF~aos^$`fE6R>)eg|VEXfZ
zL(<!|uKaAtqVIoiB>kaJNP4?ovtlXbYVBs|eUbEQFA@s7)}L&Ya*e+KlYnW6>acZ&
zJV`%vlUz^Ke}9a$n*!;t(e&47ddUR@`m=kx9s^G8XYbEw`i!Rc8D7&z`~Mrji9`JK
z4aqrkt6Yb*V`=Qt&)#eK{3Nj8`%5jMD7kWd7pEHhXX)+wy-0e;2FyO8xFOQumd>sZ
zMbh8U({xq2UNd+#|7Gb6J_m&-%<A9P^leT5u3J=qXacLZr9U6(ss0%Q^lIs+BRJ90
zFL?FGjy&T3*RA|F3;4shT#Q6ke&egh1GlmZrTqUV_Y38&JA>%py@t~Ckd*$aAElff
M=gC%3zKf^-zm*fAAOHXW


From 22de6c5c4cf466694ce755b8ff1c6b67aa27deac Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 18:09:49 +0300
Subject: [PATCH 088/242] upd .gitignore

---
 .gitignore | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.gitignore b/.gitignore
index fd41f26cc0d20..93ce82ccee4c9 100644
--- a/.gitignore
+++ b/.gitignore
@@ -47,6 +47,7 @@ models-mnt
 /Pipfile
 /embd-input-test
 /gguf
+/gguf-llama-simple
 /libllama.so
 build-info.h
 arm_neon.h

From 42cc04d11d24b7b41e19ecaa38b46350faca141b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 18:49:08 +0300
Subject: [PATCH 089/242] gguf : calculate n_mult

---
 gguf-llama.cpp | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index d385ef79a5707..b88a2d8bf9189 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -514,16 +514,30 @@ struct ggml_context * ctx_data = NULL;
     return gguf_get_arr_n(gguf_ctx, i);
     }
 
+    int find_n_mult(const int n_ff, const int n_embd) {
+        int n_mults[3] = {8192, 1, -1};
+        for (int i = 0; i < 3; ++i) {
+            int calc_ff = (((8 * n_embd) / 3 + n_mults[i] - 1) / n_mults[i]) * n_mults[i];
+            if (calc_ff == n_ff) {
+                return n_mults[i];
+            }
+        }
+
+        throw std::runtime_error(format("failed to find n_mult for n_ff = %d and n_emb = %d\n", n_ff, n_embd));
+    }
+
     void read_hparams() {
 
         // TODO make keysconstants in header
         // TODO: read all hparams from file
         hparams.n_vocab = read_n_vocab();
+        hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
-        //hparams.n_mult  = file.read_u32();
+        uint32_t n_ff    = read_u32("llama.feed_forward_length");
+        hparams.n_mult  = find_n_mult(n_ff, hparams.n_embd);
         hparams.n_head  = read_u32("llama.attention.head_count");
         hparams.n_layer = read_u32("llama.layer_count");
-        //hparams.n_rot   = file.read_u32();
+        hparams.n_rot   = hparams.n_embd / hparams.n_head;
         //hparams.ftype   = (enum llama_ftype) file.read_u32();
 
         // LLaMAv2
@@ -568,7 +582,7 @@ struct ggml_context * ctx_data = NULL;
             for (uint32_t j = 0; j < n_dims; ++j) {
             tensor.ne[j] = cur->ne[j];
             }
-            
+
             if (n_dims < 1 || n_dims > 2) {
                 throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
             }

From cfb8e35b73bd8daca6fa5179d30c48a5db43bc31 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 19:56:56 +0300
Subject: [PATCH 090/242] gguf :  inference with 7B model working (WIP)

---
 gguf-llama.cpp | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index b88a2d8bf9189..0c40957142b41 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -493,6 +493,8 @@ struct ggml_context * ctx_data = NULL;
 
     gguf_ctx = gguf_init_from_file(fname, params);
 
+    read_hparams();
+    read_vocab();
         read_tensor_metadata(tensors_map);
     }
 
@@ -523,7 +525,7 @@ struct ggml_context * ctx_data = NULL;
             }
         }
 
-        throw std::runtime_error(format("failed to find n_mult for n_ff = %d and n_emb = %d\n", n_ff, n_embd));
+        throw std::runtime_error(format("failed to find n_mult for n_ff = %d and n_embd = %d\n", n_ff, n_embd));
     }
 
     void read_hparams() {
@@ -534,14 +536,14 @@ struct ggml_context * ctx_data = NULL;
         hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
         uint32_t n_ff    = read_u32("llama.feed_forward_length");
-        hparams.n_mult  = find_n_mult(n_ff, hparams.n_embd);
+        //hparams.n_mult  = find_n_mult(n_ff, hparams.n_embd);
         hparams.n_head  = read_u32("llama.attention.head_count");
         hparams.n_layer = read_u32("llama.layer_count");
         hparams.n_rot   = hparams.n_embd / hparams.n_head;
         //hparams.ftype   = (enum llama_ftype) file.read_u32();
 
         // LLaMAv2
-        hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
+        // hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
     }
 
     void read_vocab() {

From f316b94c7c9a4f5a8cc5bc84f2fe7048a38b6ae9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 10 Aug 2023 20:20:22 +0300
Subject: [PATCH 091/242] gguf : rm deprecated function

---
 gguf-llama.h | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/gguf-llama.h b/gguf-llama.h
index 20dcc9f639510..6062f375ea92c 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -222,13 +222,6 @@ extern "C" {
                      struct llama_model * model,
             struct llama_context_params   params);
 
-    // Various functions for loading a ggml llama model.
-    // Allocate (almost) all memory needed for the model.
-    // Return NULL on failure
-    LLAMA_API DEPRECATED(struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params),
-            "please use llama_load_model_from_file combined with llama_new_context_with_model instead");
 
     // Frees all allocated memory
     LLAMA_API void llama_free(struct llama_context * ctx);

From e7d346c37c3ccd10baf1bc6895887a6ab17ee378 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 09:52:01 +0300
Subject: [PATCH 092/242] gguf : start implementing gguf_file_saver (WIP)

---
 gguf-llama.cpp | 12 ++++++++++++
 gguf-util.h    | 45 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 57 insertions(+)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 0c40957142b41..9c0b5651a0b13 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -626,20 +626,32 @@ struct gguf_file_saver {
         : file(fname, "wb"), any_file_loader(any_file_loader) {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
         write_magic();
+        write_version();
         write_hparams(new_ftype);
         write_vocab();
     }
+
     void write_magic() {
+        const int32_t magic = GGUF_MAGIC;
+        file.write_i32(magic);
+        }
+
+        void write_version() {
+            const int32_t version = GGUF_VERSION;
+            file.write_i32(version);
         }
+
     void write_hparams(enum llama_ftype new_ftype) {
         const llama_hparams & hparams = any_file_loader->hparams;
         GGML_UNUSED(hparams);
         GGML_UNUSED(new_ftype);
     }
+
     void write_vocab() {
         uint32_t n_vocab = any_file_loader->hparams.n_vocab;
         GGML_UNUSED(n_vocab);
     }
+    
     void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
         switch (new_type) {
             case GGML_TYPE_F32:
diff --git a/gguf-util.h b/gguf-util.h
index 74d6e61f792d0..94ea9006adc84 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -15,6 +15,7 @@
 #include <climits>
 
 #include <string>
+#include <sstream>
 #include <vector>
 #include <stdexcept>
 
@@ -61,6 +62,14 @@ static std::string format(const char * fmt, ...) {
     return std::string(buf.data(), size);
 }
 
+
+template<typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
+
 // TODO: can we merge this one and gguf_context?
 struct gguf_file {
     // use FILE * so we don't have to re-open the file to mmap
@@ -95,6 +104,42 @@ struct gguf_file {
 #endif
         GGML_ASSERT(ret == 0); // same
     }
+
+    
+    void write_str(const std::string & val) {
+        const int32_t n = val.size();
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        fwrite(val.c_str(), n, 1, fp);
+    }
+
+    void write_i32(int32_t val) {
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+    }
+
+    void write_u64(size_t val) {
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+    }
+
+    template<typename T>
+    void write_val(const std::string & key, enum gguf_type type, const T & val) {
+        write_str(key);
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+    }
+
+    template<typename T>
+    void write_arr(const std::string & key, enum gguf_type type, const std::vector<T> & val) {
+        write_str(key);
+        {
+            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
+        }
+
+        const int32_t n = val.size();
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        fwrite(val.data(), sizeof(T), n, fp);
+    }
 };
 
 #if defined(_WIN32)

From a356b0e228b640a31cd9d1718d097a1aabcd7675 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 10:50:02 +0300
Subject: [PATCH 093/242] gguf : start implementing gguf_file_saver (WIP)

---
 gguf-llama.cpp | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 9c0b5651a0b13..e88dc6a08510c 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -625,20 +625,25 @@ struct gguf_file_saver {
     gguf_file_saver(const char * fname, gguf_file_loader * any_file_loader, enum llama_ftype new_ftype)
         : file(fname, "wb"), any_file_loader(any_file_loader) {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
-        write_magic();
-        write_version();
+        write_header();
         write_hparams(new_ftype);
         write_vocab();
     }
 
-    void write_magic() {
+    // TODO: probably it's better to move these to gguf_file
+
+    void write_header() {
         const int32_t magic = GGUF_MAGIC;
         file.write_i32(magic);
-        }
-
-        void write_version() {
+    
             const int32_t version = GGUF_VERSION;
             file.write_i32(version);
+
+            const int32_t n_tensors = gguf_get_n_tensors(any_file_loader->gguf_ctx);
+            file.write_i32(n_tensors);
+
+            const int32_t n_kv = gguf_get_n_kv(any_file_loader->gguf_ctx);
+            file.write_i32(n_kv);
         }
 
     void write_hparams(enum llama_ftype new_ftype) {
@@ -651,7 +656,7 @@ struct gguf_file_saver {
         uint32_t n_vocab = any_file_loader->hparams.n_vocab;
         GGML_UNUSED(n_vocab);
     }
-    
+
     void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
         switch (new_type) {
             case GGML_TYPE_F32:

From b2440f1943f88243938e438acef18ad072b9004e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 11:29:50 +0300
Subject: [PATCH 094/242] gguf : start implementing gguf_file_saver (WIP)

---
 gguf-llama.cpp | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index e88dc6a08510c..08b7004350ed7 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -647,9 +647,14 @@ struct gguf_file_saver {
         }
 
     void write_hparams(enum llama_ftype new_ftype) {
-        const llama_hparams & hparams = any_file_loader->hparams;
-        GGML_UNUSED(hparams);
-        GGML_UNUSED(new_ftype);
+        const int32_t n_kv = gguf_get_n_kv(any_file_loader->gguf_ctx);
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(any_file_loader->gguf_ctx, i);
+            if (strcmp(key, "general.quantization_version") == 0) {
+                file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
+            }
+        }
+
     }
 
     void write_vocab() {
@@ -658,6 +663,10 @@ struct gguf_file_saver {
     }
 
     void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
+        GGML_UNUSED(tensor);
+        GGML_UNUSED(new_data);
+        GGML_UNUSED(new_size);
+
         switch (new_type) {
             case GGML_TYPE_F32:
             case GGML_TYPE_F16:

From eb8ca6996f5e3559505b3269d98e5d3ed860200e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 12:24:08 +0300
Subject: [PATCH 095/242] gguf : add gguf_get_kv_type

---
 ggml.c         | 4 ++++
 ggml.h         | 9 +++++----
 gguf-llama.cpp | 3 +++
 3 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/ggml.c b/ggml.c
index df8dce7e89915..ac45f12b08b13 100644
--- a/ggml.c
+++ b/ggml.c
@@ -19031,6 +19031,10 @@ const char * gguf_get_key(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].key.data;
 }
 
+const enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
+    return ctx->header.kv[i].type;
+}
+
 const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
     struct gguf_kv * kv = &ctx->header.kv[key_id];
     struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];
diff --git a/ggml.h b/ggml.h
index f1673eed52d20..4490e076cffcf 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1744,10 +1744,11 @@ extern "C" {
     GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
     GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
-    GGML_API int          gguf_get_n_kv(struct gguf_context * ctx);
-    GGML_API int          gguf_find_key(struct gguf_context * ctx, const char * key);
-    GGML_API const char * gguf_get_key (struct gguf_context * ctx, int i);
-    GGML_API void         gguf_get_val (struct gguf_context * ctx, int i, void * val);
+    GGML_API int                  gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API int                  gguf_find_key(struct gguf_context * ctx, const char * key);
+    GGML_API const char *         gguf_get_key (struct gguf_context * ctx, int i);
+    GGML_API const enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
+    GGML_API void                 gguf_get_val (struct gguf_context * ctx, int i, void * val);
 
     GGML_API const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i);
     GGML_API float        gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i);
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 08b7004350ed7..8b928c364c766 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -652,6 +652,9 @@ struct gguf_file_saver {
             const char * key = gguf_get_key(any_file_loader->gguf_ctx, i);
             if (strcmp(key, "general.quantization_version") == 0) {
                 file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
+            } else {
+                const gguf_type vtype = gguf_get_kv_type(any_file_loader->gguf_ctx, i);
+                GGML_UNUSED(vtype);
             }
         }
 

From e3a49609533646cbd1990f2d99668aef7a0cc420 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 13:03:23 +0300
Subject: [PATCH 096/242] gguf : add gguf_get_kv_type

---
 ggml.c         |  2 +-
 ggml.h         | 10 +++++-----
 gguf-llama.cpp | 20 +++++++++++++++++++-
 3 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/ggml.c b/ggml.c
index ac45f12b08b13..e00f09fa458f7 100644
--- a/ggml.c
+++ b/ggml.c
@@ -19031,7 +19031,7 @@ const char * gguf_get_key(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].key.data;
 }
 
-const enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
+enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].type;
 }
 
diff --git a/ggml.h b/ggml.h
index 4490e076cffcf..9a266e17573ce 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1744,11 +1744,11 @@ extern "C" {
     GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
     GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
-    GGML_API int                  gguf_get_n_kv(struct gguf_context * ctx);
-    GGML_API int                  gguf_find_key(struct gguf_context * ctx, const char * key);
-    GGML_API const char *         gguf_get_key (struct gguf_context * ctx, int i);
-    GGML_API const enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
-    GGML_API void                 gguf_get_val (struct gguf_context * ctx, int i, void * val);
+    GGML_API int            gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API int            gguf_find_key(struct gguf_context * ctx, const char * key);
+    GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);
+    GGML_API enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
+    GGML_API void           gguf_get_val (struct gguf_context * ctx, int i, void * val);
 
     GGML_API const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i);
     GGML_API float        gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i);
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 8b928c364c766..9ab7702774972 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -536,6 +536,7 @@ struct ggml_context * ctx_data = NULL;
         hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
         uint32_t n_ff    = read_u32("llama.feed_forward_length");
+        GGML_UNUSED(n_ff);
         //hparams.n_mult  = find_n_mult(n_ff, hparams.n_embd);
         hparams.n_head  = read_u32("llama.attention.head_count");
         hparams.n_layer = read_u32("llama.layer_count");
@@ -654,7 +655,21 @@ struct gguf_file_saver {
                 file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
             } else {
                 const gguf_type vtype = gguf_get_kv_type(any_file_loader->gguf_ctx, i);
-                GGML_UNUSED(vtype);
+                switch(vtype) {
+                    case GGUF_TYPE_BOOL:
+                    case GGUF_TYPE_FLOAT32:
+                    case GGUF_TYPE_INT16:
+                    case GGUF_TYPE_INT32:
+                    case GGUF_TYPE_INT8:
+                    case GGUF_TYPE_STRING:
+                    case GGUF_TYPE_UINT16:
+                    case GGUF_TYPE_UINT32:
+                    case GGUF_TYPE_UINT8:
+                    case GGUF_TYPE_ARRAY:
+                    break;
+                    default:
+                    throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
+                }
             }
         }
 
@@ -3873,6 +3888,9 @@ bool llama_load_session_file(struct llama_context * ctx, const char * path_sessi
 
 bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
     gguf_file file(path_session, "wb");
+    GGML_UNUSED(ctx);
+    GGML_UNUSED(tokens);
+    GGML_UNUSED(n_token_count);
 
     // TODO: implement with GGUF format
 

From 28abfc90fa9154be5ed9fded0870587aa6850820 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 13:27:58 +0300
Subject: [PATCH 097/242] gguf : write metadata in gguf_file_saver (WIP)

---
 gguf-llama.cpp | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 9ab7702774972..11dcb1cc57ded 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -655,11 +655,29 @@ struct gguf_file_saver {
                 file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
             } else {
                 const gguf_type vtype = gguf_get_kv_type(any_file_loader->gguf_ctx, i);
+
+                bool bool_val;
+                float f32_val;
+                int16_t i16_val;
+                int32_t i32_val;
+
                 switch(vtype) {
                     case GGUF_TYPE_BOOL:
+                    bool_val = gguf_get_val_bool(any_file_loader->gguf_ctx, i);
+                    file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
+                    break;
                     case GGUF_TYPE_FLOAT32:
+                    f32_val = gguf_get_val_f32(any_file_loader->gguf_ctx, i);
+                    file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
+                    break;
                     case GGUF_TYPE_INT16:
+                    i16_val = gguf_get_val_i16(any_file_loader->gguf_ctx, i);
+                    file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
+                    break;
                     case GGUF_TYPE_INT32:
+                    i32_val = gguf_get_val_i32(any_file_loader->gguf_ctx, i);
+                    file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
+                    break;
                     case GGUF_TYPE_INT8:
                     case GGUF_TYPE_STRING:
                     case GGUF_TYPE_UINT16:

From 781b9ec3f504e1d306bcb0027dac65aa8e4dfc70 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 18:01:26 +0300
Subject: [PATCH 098/242] gguf : write metadata in gguf_file_saver (WIP)

---
 gguf-llama.cpp | 21 +++++++++++++++++++++
 gguf-util.h    | 27 +++++++++++++++++++++++++++
 2 files changed, 48 insertions(+)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 11dcb1cc57ded..e70cae44c90a9 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -660,6 +660,12 @@ struct gguf_file_saver {
                 float f32_val;
                 int16_t i16_val;
                 int32_t i32_val;
+                int8_t i8_val;
+                std::string str_val;
+                uint16_t u16_val;
+                uint32_t u32_val;
+                uint8_t u8_val;
+
 
                 switch(vtype) {
                     case GGUF_TYPE_BOOL:
@@ -679,10 +685,25 @@ struct gguf_file_saver {
                     file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
                     break;
                     case GGUF_TYPE_INT8:
+                    i8_val = gguf_get_val_i8(any_file_loader->gguf_ctx, i);
+                    file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
+                    break;
                     case GGUF_TYPE_STRING:
+                    str_val = gguf_get_val_str(any_file_loader->gguf_ctx, i);
+                    file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
+                    break;
                     case GGUF_TYPE_UINT16:
+                    u16_val = gguf_get_val_u16(any_file_loader->gguf_ctx, i);
+                    file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
+                    break;
                     case GGUF_TYPE_UINT32:
+                    u32_val = gguf_get_val_u32(any_file_loader->gguf_ctx, i);
+                    file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
+                    break;
                     case GGUF_TYPE_UINT8:
+                    u8_val = gguf_get_val_u8(any_file_loader->gguf_ctx, i);
+                    file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
+                    break;
                     case GGUF_TYPE_ARRAY:
                     break;
                     default:
diff --git a/gguf-util.h b/gguf-util.h
index 94ea9006adc84..9134019a45066 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -140,6 +140,33 @@ struct gguf_file {
         fwrite((const char *) &n, sizeof(n), 1, fp);
         fwrite(val.data(), sizeof(T), n, fp);
     }
+    template<>
+    void write_val<std::string>(const std::string & key, enum gguf_type type, const std::string & val) {
+        write_str(key);
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+
+        const int32_t n = val.size();
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        fwrite(val.c_str(), n, 1, fp);
+    }
+
+    template<>
+    void write_arr<std::string>(const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
+        write_str(key);
+        {
+            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
+        }
+
+        const int32_t n = val.size();
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        for (int i = 0; i < n; ++i) {
+            const int32_t nstr = val[i].size();
+            fwrite((const char *) &nstr, sizeof(nstr), 1, fp);
+            fwrite(val[i].c_str(), nstr, 1, fp);
+        }
+    }
 };
 
 #if defined(_WIN32)

From d09fd107138d311151546f42a6881548ffd766c3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 20:07:43 +0300
Subject: [PATCH 099/242] gguf : write metadata in gguf_file_saver

---
 ggml.c         |  4 ++++
 ggml.h         |  1 +
 gguf-llama.cpp | 34 +++++++++++++++++++++++++++++++++-
 3 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/ggml.c b/ggml.c
index e00f09fa458f7..c8fa60328e2d7 100644
--- a/ggml.c
+++ b/ggml.c
@@ -19035,6 +19035,10 @@ enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
     return ctx->header.kv[i].type;
 }
 
+enum gguf_type gguf_get_arr_type(struct gguf_context * ctx, int i) {
+    return ctx->header.kv[i].value.arr.type;
+}
+
 const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
     struct gguf_kv * kv = &ctx->header.kv[key_id];
     struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];
diff --git a/ggml.h b/ggml.h
index 9a266e17573ce..fb3db10e2cedb 100644
--- a/ggml.h
+++ b/ggml.h
@@ -1748,6 +1748,7 @@ extern "C" {
     GGML_API int            gguf_find_key(struct gguf_context * ctx, const char * key);
     GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);
     GGML_API enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
+    GGML_API enum gguf_type gguf_get_arr_type (struct gguf_context * ctx, int i);
     GGML_API void           gguf_get_val (struct gguf_context * ctx, int i, void * val);
 
     GGML_API const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i);
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index e70cae44c90a9..27e0b5d43d875 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -647,6 +647,28 @@ struct gguf_file_saver {
             file.write_i32(n_kv);
         }
 
+        void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
+            std::vector<std::string> data(n_arr);
+
+            for (int j = 0; j < n_arr; ++j) {
+                std::string val = gguf_get_arr_str(any_file_loader->gguf_ctx, i, j);
+                data[j] = val;
+                }
+                            
+                            file.write_arr<std::string>(key, type, data);
+        }
+
+        void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
+            std::vector<float> data(n_arr);
+
+            for (int j = 0; j < n_arr; ++j) {
+                float val = gguf_get_arr_f32(any_file_loader->gguf_ctx, i, j);
+                data[j] = val;
+                }
+                            
+                            file.write_arr<float>(key, type, data);
+        }
+
     void write_hparams(enum llama_ftype new_ftype) {
         const int32_t n_kv = gguf_get_n_kv(any_file_loader->gguf_ctx);
         for (int i = 0; i < n_kv; ++i) {
@@ -665,7 +687,8 @@ struct gguf_file_saver {
                 uint16_t u16_val;
                 uint32_t u32_val;
                 uint8_t u8_val;
-
+                gguf_type arr_type;
+                int n_arr;
 
                 switch(vtype) {
                     case GGUF_TYPE_BOOL:
@@ -705,6 +728,15 @@ struct gguf_file_saver {
                     file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
                     break;
                     case GGUF_TYPE_ARRAY:
+                    arr_type = gguf_get_arr_type(any_file_loader->gguf_ctx, i);
+                    n_arr    = gguf_get_arr_n(any_file_loader->gguf_ctx, i);
+                    if (arr_type == GGUF_TYPE_FLOAT32) {
+                        write_hparam_arr_f32(key, arr_type, i, n_arr);
+                        } else if (arr_type == GGUF_TYPE_STRING) {
+                            write_hparam_arr_str(key, GGUF_TYPE_STRING, i, n_arr);
+                        } else {
+                            throw std::runtime_error("not implemented");
+                        }
                     break;
                     default:
                     throw std::runtime_error(format("cannot recognize value type for key %s\n", key));

From 61919c1a8f12cb14c41e73019180c5950c355c83 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 20:36:11 +0300
Subject: [PATCH 100/242] gguf : rm references to old file formats

---
 gguf-llama.cpp | 36 ++++++++++--------------------------
 1 file changed, 10 insertions(+), 26 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 27e0b5d43d875..7bd8cef6aee3e 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -467,13 +467,11 @@ struct llama_load_tensors_map {
 };
 
 enum gguf_file_version {
-    gguf_file_VERSION_GGML,
-    gguf_file_VERSION_GGMF_V1, // added version field and scores in vocab
-    gguf_file_VERSION_GGJT_V1, // added padding
-    gguf_file_VERSION_GGJT_V2, // changed quantization format
-    gguf_file_VERSION_GGJT_V3, // changed Q4 and Q8 quantization format
+    GGUF_FILE_VERSION_V1
+    
 };
 
+
 struct gguf_file_loader {
     gguf_file file;
     gguf_context * gguf_ctx;
@@ -1069,12 +1067,8 @@ int64_t llama_time_us() {
 
 static const char *gguf_file_version_name(gguf_file_version version) {
     switch (version) {
-        case gguf_file_VERSION_GGML: return "'ggml' (old version with low tokenizer quality and no mmap support)";
-        case gguf_file_VERSION_GGMF_V1: return "ggmf v1 (old version with no mmap support)";
-        case gguf_file_VERSION_GGJT_V1: return "ggjt v1 (pre #1405)";
-        case gguf_file_VERSION_GGJT_V2: return "ggjt v2 (pre #1508)";
-        case gguf_file_VERSION_GGJT_V3: return "ggjt v3 (latest)";
-    }
+        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
+        }
 
     return "unknown";
 }
@@ -1205,22 +1199,12 @@ static void llama_model_load_internal(
         fprintf(stderr, "%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
     }
 
-    if (file_version < gguf_file_VERSION_GGJT_V2) {
-        if (hparams.ftype != LLAMA_FTYPE_ALL_F32     &&
-            hparams.ftype != LLAMA_FTYPE_MOSTLY_F16  &&
-            hparams.ftype != LLAMA_FTYPE_MOSTLY_Q8_0) {
-            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1405)"));
-        }
-    }
-
-    if (file_version < gguf_file_VERSION_GGJT_V3) {
-        if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
-            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
-            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
-            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
-        }
+    if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
+        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
+        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
+        throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
     }
-
+    
     if (vocab_only) {
         return;
     }

From 7009cf581cbe958fcdeb94e58386646d0ebfa3dd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 20:43:02 +0300
Subject: [PATCH 101/242] gguf : shorter name for member variable

---
 gguf-llama.cpp | 44 ++++++++++++++++++++++----------------------
 1 file changed, 22 insertions(+), 22 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 7bd8cef6aee3e..77ba7d8806698 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -620,9 +620,9 @@ struct ggml_context * ctx_data = NULL;
 
 struct gguf_file_saver {
     gguf_file file;
-    gguf_file_loader * any_file_loader;
-    gguf_file_saver(const char * fname, gguf_file_loader * any_file_loader, enum llama_ftype new_ftype)
-        : file(fname, "wb"), any_file_loader(any_file_loader) {
+    gguf_file_loader * fl;
+    gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
+        : file(fname, "wb"), fl(fl) {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
         write_header();
         write_hparams(new_ftype);
@@ -638,10 +638,10 @@ struct gguf_file_saver {
             const int32_t version = GGUF_VERSION;
             file.write_i32(version);
 
-            const int32_t n_tensors = gguf_get_n_tensors(any_file_loader->gguf_ctx);
+            const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
             file.write_i32(n_tensors);
 
-            const int32_t n_kv = gguf_get_n_kv(any_file_loader->gguf_ctx);
+            const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
             file.write_i32(n_kv);
         }
 
@@ -649,7 +649,7 @@ struct gguf_file_saver {
             std::vector<std::string> data(n_arr);
 
             for (int j = 0; j < n_arr; ++j) {
-                std::string val = gguf_get_arr_str(any_file_loader->gguf_ctx, i, j);
+                std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
                 data[j] = val;
                 }
                             
@@ -660,7 +660,7 @@ struct gguf_file_saver {
             std::vector<float> data(n_arr);
 
             for (int j = 0; j < n_arr; ++j) {
-                float val = gguf_get_arr_f32(any_file_loader->gguf_ctx, i, j);
+                float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
                 data[j] = val;
                 }
                             
@@ -668,13 +668,13 @@ struct gguf_file_saver {
         }
 
     void write_hparams(enum llama_ftype new_ftype) {
-        const int32_t n_kv = gguf_get_n_kv(any_file_loader->gguf_ctx);
+        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
         for (int i = 0; i < n_kv; ++i) {
-            const char * key = gguf_get_key(any_file_loader->gguf_ctx, i);
+            const char * key = gguf_get_key(fl->gguf_ctx, i);
             if (strcmp(key, "general.quantization_version") == 0) {
                 file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
             } else {
-                const gguf_type vtype = gguf_get_kv_type(any_file_loader->gguf_ctx, i);
+                const gguf_type vtype = gguf_get_kv_type(fl->gguf_ctx, i);
 
                 bool bool_val;
                 float f32_val;
@@ -690,44 +690,44 @@ struct gguf_file_saver {
 
                 switch(vtype) {
                     case GGUF_TYPE_BOOL:
-                    bool_val = gguf_get_val_bool(any_file_loader->gguf_ctx, i);
+                    bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
                     file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
                     break;
                     case GGUF_TYPE_FLOAT32:
-                    f32_val = gguf_get_val_f32(any_file_loader->gguf_ctx, i);
+                    f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
                     file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
                     break;
                     case GGUF_TYPE_INT16:
-                    i16_val = gguf_get_val_i16(any_file_loader->gguf_ctx, i);
+                    i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
                     file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
                     break;
                     case GGUF_TYPE_INT32:
-                    i32_val = gguf_get_val_i32(any_file_loader->gguf_ctx, i);
+                    i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
                     file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
                     break;
                     case GGUF_TYPE_INT8:
-                    i8_val = gguf_get_val_i8(any_file_loader->gguf_ctx, i);
+                    i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
                     file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
                     break;
                     case GGUF_TYPE_STRING:
-                    str_val = gguf_get_val_str(any_file_loader->gguf_ctx, i);
+                    str_val = gguf_get_val_str(fl->gguf_ctx, i);
                     file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
                     break;
                     case GGUF_TYPE_UINT16:
-                    u16_val = gguf_get_val_u16(any_file_loader->gguf_ctx, i);
+                    u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
                     file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
                     break;
                     case GGUF_TYPE_UINT32:
-                    u32_val = gguf_get_val_u32(any_file_loader->gguf_ctx, i);
+                    u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
                     file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
                     break;
                     case GGUF_TYPE_UINT8:
-                    u8_val = gguf_get_val_u8(any_file_loader->gguf_ctx, i);
+                    u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
                     file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
                     break;
                     case GGUF_TYPE_ARRAY:
-                    arr_type = gguf_get_arr_type(any_file_loader->gguf_ctx, i);
-                    n_arr    = gguf_get_arr_n(any_file_loader->gguf_ctx, i);
+                    arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
+                    n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
                     if (arr_type == GGUF_TYPE_FLOAT32) {
                         write_hparam_arr_f32(key, arr_type, i, n_arr);
                         } else if (arr_type == GGUF_TYPE_STRING) {
@@ -745,7 +745,7 @@ struct gguf_file_saver {
     }
 
     void write_vocab() {
-        uint32_t n_vocab = any_file_loader->hparams.n_vocab;
+        uint32_t n_vocab = fl->hparams.n_vocab;
         GGML_UNUSED(n_vocab);
     }
 

From f44bbd3d88e2cdb68e59c965e28961afe4a0ad26 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 21:00:51 +0300
Subject: [PATCH 102/242] gguf : rm redundant method

---
 gguf-llama.cpp | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 77ba7d8806698..a8bd242b020e5 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -626,7 +626,6 @@ struct gguf_file_saver {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
         write_header();
         write_hparams(new_ftype);
-        write_vocab();
     }
 
     // TODO: probably it's better to move these to gguf_file
@@ -744,10 +743,6 @@ struct gguf_file_saver {
 
     }
 
-    void write_vocab() {
-        uint32_t n_vocab = fl->hparams.n_vocab;
-        GGML_UNUSED(n_vocab);
-    }
 
     void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
         GGML_UNUSED(tensor);

From e732423280ff72ece7bc8dfe8ec1fa7e3153714c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Fri, 11 Aug 2023 23:50:38 +0300
Subject: [PATCH 103/242] gguf : get rid of n_mult, read n_ff from file

---
 gguf-llama.cpp | 55 ++++++++++++++++++--------------------------------
 1 file changed, 20 insertions(+), 35 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index a8bd242b020e5..40d5ffd1469b2 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -177,15 +177,12 @@ struct llama_hparams {
     uint32_t n_vocab   = 32000;
     uint32_t n_ctx     = 512;   // this is provided as user input?
     uint32_t n_embd    = 4096;
-    uint32_t n_mult    = 256;
     uint32_t n_head    = 32;
     uint32_t n_head_kv = 32;
     uint32_t n_layer   = 32;
     uint32_t n_rot     = 64;
+    uint32_t n_ff       = 11008;
 
-    // LLaMAv2
-    // TODO: load from model data hparams
-    float f_ffn_mult = 1.0f;
     float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
 
     float rope_freq_base  = 10000.0f;
@@ -467,7 +464,7 @@ struct llama_load_tensors_map {
 };
 
 enum gguf_file_version {
-    GGUF_FILE_VERSION_V1
+    GGUF_FILE_VERSION_V1 = 1,
     
 };
 
@@ -490,6 +487,7 @@ struct ggml_context * ctx_data = NULL;
     };
 
     gguf_ctx = gguf_init_from_file(fname, params);
+    file_version = (enum gguf_file_version) gguf_get_version(gguf_ctx);
 
     read_hparams();
     read_vocab();
@@ -505,6 +503,15 @@ struct ggml_context * ctx_data = NULL;
         return gguf_get_val_u32(gguf_ctx, i);
     }
 
+    float read_f32(const char * key) {
+        int i = gguf_find_key(gguf_ctx, key);
+        if (i == -1) {
+            throw std::runtime_error(format("cannot find param with key %s\n", key));
+        }
+
+        return gguf_get_val_f32(gguf_ctx, i);
+    }
+
     int read_n_vocab() {
         int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
     if (i == -1) {
@@ -514,18 +521,6 @@ struct ggml_context * ctx_data = NULL;
     return gguf_get_arr_n(gguf_ctx, i);
     }
 
-    int find_n_mult(const int n_ff, const int n_embd) {
-        int n_mults[3] = {8192, 1, -1};
-        for (int i = 0; i < 3; ++i) {
-            int calc_ff = (((8 * n_embd) / 3 + n_mults[i] - 1) / n_mults[i]) * n_mults[i];
-            if (calc_ff == n_ff) {
-                return n_mults[i];
-            }
-        }
-
-        throw std::runtime_error(format("failed to find n_mult for n_ff = %d and n_embd = %d\n", n_ff, n_embd));
-    }
-
     void read_hparams() {
 
         // TODO make keysconstants in header
@@ -533,14 +528,12 @@ struct ggml_context * ctx_data = NULL;
         hparams.n_vocab = read_n_vocab();
         hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
-        uint32_t n_ff    = read_u32("llama.feed_forward_length");
-        GGML_UNUSED(n_ff);
-        //hparams.n_mult  = find_n_mult(n_ff, hparams.n_embd);
+        hparams.n_ff    = read_u32("llama.feed_forward_length");
         hparams.n_head  = read_u32("llama.attention.head_count");
         hparams.n_layer = read_u32("llama.layer_count");
-        hparams.n_rot   = hparams.n_embd / hparams.n_head;
-        //hparams.ftype   = (enum llama_ftype) file.read_u32();
-
+        hparams.n_rot   = read_u32("llama.rope.dimension_count");
+        hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
+        
         // LLaMAv2
         // hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
     }
@@ -1125,6 +1118,7 @@ static void llama_model_load_internal(
         bool vocab_only,
         llama_progress_callback progress_callback,
         void * progress_callback_user_data) {
+            GGML_UNUSED(rms_norm_eps); // TODO: update function signature to remove this
 
     model.t_start_us = ggml_time_us();
 
@@ -1137,9 +1131,6 @@ static void llama_model_load_internal(
 
     auto & hparams = model.hparams;
 
-    // TODO: read from file
-    hparams.f_rms_norm_eps = rms_norm_eps;
-
     {
         switch (hparams.n_layer) {
             case 26: model.type = e_model::MODEL_3B; break;
@@ -1162,25 +1153,19 @@ static void llama_model_load_internal(
         if (model.type == e_model::MODEL_65B && n_gqa == 8) {
             fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
             model.type = e_model::MODEL_70B;
-            hparams.f_ffn_mult = 1.3f; // from the params.json of the 70B model
-        }
+            }
 
         hparams.rope_freq_base  = rope_freq_base;
         hparams.rope_freq_scale = rope_freq_scale;
     }
 
-    // ref: https://github.com/facebookresearch/llama/blob/6c7fe276574e78057f917549435a2554000a876d/llama/model.py#L194-L199
-    const uint32_t n_ff_raw  = 2*(4*hparams.n_embd)/3;
-    const uint32_t n_ff_mult = hparams.f_ffn_mult*n_ff_raw;
-    const uint32_t n_ff      = ((n_ff_mult + hparams.n_mult - 1)/hparams.n_mult)*hparams.n_mult;
-    //const uint32_t n_ff = 28672;
-
+    const uint32_t n_ff = hparams.n_ff;
+    
     {
         fprintf(stderr, "%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
         fprintf(stderr, "%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
         fprintf(stderr, "%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
         fprintf(stderr, "%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        fprintf(stderr, "%s: n_mult     = %u\n",   __func__, hparams.n_mult);
         fprintf(stderr, "%s: n_head     = %u\n",   __func__, hparams.n_head);
         fprintf(stderr, "%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
         fprintf(stderr, "%s: n_layer    = %u\n",   __func__, hparams.n_layer);

From 2a5ac7af44c258a725ae10180dc5993db972e527 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 11 Aug 2023 23:08:48 +0200
Subject: [PATCH 104/242] Update gguf_tensor_map.py

---
 gguf_tensor_map.py | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/gguf_tensor_map.py b/gguf_tensor_map.py
index 4fba633b2bf12..644c5914de9b2 100644
--- a/gguf_tensor_map.py
+++ b/gguf_tensor_map.py
@@ -26,15 +26,15 @@ def get_tensor_map( n_blocks : int):
     tensor_map["output"] = mapped_to    # llama-pth
     # Attention and fee-forward layer blocks
     for i in range(0,n_blocks):
-        # Attention norm 1
-        mapped_to = "transformer.blocks."+str(i)+".attn_norm_1"
+        # Attention norm
+        mapped_to = "transformer.blocks."+str(i)+".attn_norm"
         tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to             # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to      # mpt
-        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to  # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to          # falcon40b
-        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to   # llama-hf
-        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to          # llama-pth
+        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to              # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to       # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to   # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to           # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to    # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to           # llama-pth
         # Attention norm 2
         mapped_to = "transformer.blocks."+str(i)+".attn_norm_2"
         tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b

From e76c59d5245825ac1e1bf06bfc7f78ed1f6bcf16 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 11 Aug 2023 23:09:49 +0200
Subject: [PATCH 105/242] Update gptneox-main.cpp

---
 gptneox-main.cpp | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 1667c4d545114..f2be93e4b1ae6 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -565,8 +565,8 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
             std::string blocknamestart = "transformer.blocks." + std::to_string(i) + ".";
 
-            layer.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm_1.weight" );
-            layer.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm_1.bias" );
+            layer.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm.weight" );
+            layer.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm.bias" );
 
             layer.c_attn_attn_w   = get_tensor_ex(ctx, blocknamestart + "attn_qkv.weight" );
             layer.c_attn_attn_b   = get_tensor_ex(ctx ,blocknamestart + "attn_qkv.bias" );
@@ -584,8 +584,8 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
             layer.c_mlp_proj_b    = get_tensor_ex(ctx, blocknamestart + "ffn_down.bias" );
 
             // map by name
-            model.tensors[blocknamestart + "attn_norm_1.weight"] = layer.ln_1_g;
-            model.tensors[blocknamestart + "attn_norm_1.bias"]   = layer.ln_1_b;
+            model.tensors[blocknamestart + "attn_norm.weight"] = layer.ln_1_g;
+            model.tensors[blocknamestart + "attn_norm.bias"]   = layer.ln_1_b;
 
             model.tensors[blocknamestart + "attn_qkv.weight"] = layer.c_attn_attn_w;
             model.tensors[blocknamestart + "attn_qkv.bias"]   = layer.c_attn_attn_b;

From 2f52008b203fc97e72d850e02b08c0cd49f56969 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 07:24:46 +0300
Subject: [PATCH 106/242] gguf : rm references to old file magics

---
 gguf-llama.cpp |  4 ----
 gguf-llama.h   | 12 ------------
 2 files changed, 16 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 40d5ffd1469b2..f80f823a0d004 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -3367,10 +3367,6 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
     {
         uint32_t magic;
         fin.read((char *) &magic, sizeof(magic));
-        if (magic != LLAMA_FILE_MAGIC_GGLA) {
-            fprintf(stderr, "%s: bad file magic\n", __func__);
-            return 1;
-        }
         uint32_t format_version;
         fin.read((char *) &format_version, sizeof(format_version));
 
diff --git a/gguf-llama.h b/gguf-llama.h
index 6062f375ea92c..d3c0d6b87f437 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -34,18 +34,6 @@
 #    define DEPRECATED(func, hint) func
 #endif
 
-#define LLAMA_FILE_MAGIC_GGJT        0x67676a74u // 'ggjt'
-#define LLAMA_FILE_MAGIC_GGLA        0x67676c61u // 'ggla'
-#define LLAMA_FILE_MAGIC_GGMF        0x67676d66u // 'ggmf'
-#define LLAMA_FILE_MAGIC_GGML        0x67676d6cu // 'ggml'
-#define LLAMA_FILE_MAGIC_GGSN        0x6767736eu // 'ggsn'
-
-#define LLAMA_FILE_VERSION           3
-#define LLAMA_FILE_MAGIC             LLAMA_FILE_MAGIC_GGJT
-#define LLAMA_FILE_MAGIC_UNVERSIONED LLAMA_FILE_MAGIC_GGML
-#define LLAMA_SESSION_MAGIC          LLAMA_FILE_MAGIC_GGSN
-#define LLAMA_SESSION_VERSION        1
-
 #define LLAMA_DEFAULT_SEED           0xFFFFFFFF
 
 #if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)

From 4fa017a1f9ed89c872ca007e0701f5d7ced4f8fa Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 10:40:56 +0300
Subject: [PATCH 107/242] gguf : start implementing quantization (WIP)

---
 gguf-llama.cpp |  4 ++++
 gguf-util.h    | 23 ++++++++++++++++-------
 2 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index f80f823a0d004..0581755b14f64 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -614,6 +614,7 @@ struct ggml_context * ctx_data = NULL;
 struct gguf_file_saver {
     gguf_file file;
     gguf_file_loader * fl;
+    size_t info_offset;
     gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
         : file(fname, "wb"), fl(fl) {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
@@ -734,6 +735,9 @@ struct gguf_file_saver {
             }
         }
 
+        info_offset  = file.tell();
+        size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
+        file.write_zeros(count);
     }
 
 
diff --git a/gguf-util.h b/gguf-util.h
index 9134019a45066..0964e6d02cc4f 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -106,18 +106,21 @@ struct gguf_file {
     }
 
     
-    void write_str(const std::string & val) {
+    size_t write_str(const std::string & val) {
+        size_t total_written = 0;
         const int32_t n = val.size();
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        fwrite(val.c_str(), n, 1, fp);
+        total_written += fwrite((const char *) &n, sizeof(n), 1, fp);
+        total_written += fwrite(val.c_str(), n, 1, fp);
+
+        return total_written;
     }
 
-    void write_i32(int32_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
+    size_t write_i32(int32_t val) {
+        return fwrite((const char *) &val, sizeof(val), 1, fp);
     }
 
-    void write_u64(size_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
+    size_t write_u64(size_t val) {
+        return fwrite((const char *) &val, sizeof(val), 1, fp);
     }
 
     template<typename T>
@@ -167,6 +170,12 @@ struct gguf_file {
             fwrite(val[i].c_str(), nstr, 1, fp);
         }
     }
+
+    void write_zeros(size_t count) {
+        for (size_t i = 0; i < count; ++i) {
+            fputc(0, fp);
+        }
+    }
 };
 
 #if defined(_WIN32)

From 0e1a3c7e7dea7877840ed7a203d589bd841cadff Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 11:32:34 +0300
Subject: [PATCH 108/242] gguf : start implementing quantization (WIP)

---
 gguf-llama.cpp | 33 +++++++++++++++++++++++++++------
 1 file changed, 27 insertions(+), 6 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 0581755b14f64..cf2c56955e6b4 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -615,6 +615,8 @@ struct gguf_file_saver {
     gguf_file file;
     gguf_file_loader * fl;
     size_t info_offset;
+    size_t tensor_offset = 0;
+
     gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
         : file(fname, "wb"), fl(fl) {
         fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
@@ -622,8 +624,6 @@ struct gguf_file_saver {
         write_hparams(new_ftype);
     }
 
-    // TODO: probably it's better to move these to gguf_file
-
     void write_header() {
         const int32_t magic = GGUF_MAGIC;
         file.write_i32(magic);
@@ -740,12 +740,26 @@ struct gguf_file_saver {
         file.write_zeros(count);
     }
 
+    size_t write_tensor_info(llama_load_tensor & tensor) {
+        size_t total_written = 0;
+        file.seek(0, info_offset);
+        total_written += file.write_str(tensor.name);
 
-    void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
-        GGML_UNUSED(tensor);
-        GGML_UNUSED(new_data);
-        GGML_UNUSED(new_size);
+        int32_t n_dims = tensor.ne.size();
+        file.write_i32(n_dims);
+        for (int32_t i = 0; i < n_dims; ++i) {
+            total_written += file.write_i32(i);
+        }
+
+        total_written += file.write_u64(tensor_offset);
+        info_offset   += total_written;
 
+        file.seek(0, SEEK_END);
+
+        return total_written;
+    }
+
+    void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
         switch (new_type) {
             case GGML_TYPE_F32:
             case GGML_TYPE_F16:
@@ -763,6 +777,13 @@ struct gguf_file_saver {
             default: GGML_ASSERT(false);
         }
         
+        write_tensor_info(tensor);
+        // file.write_raw(new_data);
+        GGML_UNUSED(new_data);
+        size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
+        size_t pad = padded_size - new_size;
+        file.write_zeros(pad);
+        tensor_offset += padded_size;
     }
 };
 

From c4f02b4f74c5fdb6493090c238891163709631e4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 12:01:17 +0300
Subject: [PATCH 109/242] gguf : start implementing quantization (WIP)

---
 gguf-llama.cpp | 3 +--
 gguf-util.h    | 4 ++++
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index cf2c56955e6b4..defe26fe04ab8 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -778,8 +778,7 @@ struct gguf_file_saver {
         }
         
         write_tensor_info(tensor);
-        // file.write_raw(new_data);
-        GGML_UNUSED(new_data);
+        file.write_raw(new_data, new_size);
         size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
         size_t pad = padded_size - new_size;
         file.write_zeros(pad);
diff --git a/gguf-util.h b/gguf-util.h
index 0964e6d02cc4f..17f9dc9680dc9 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -123,6 +123,10 @@ struct gguf_file {
         return fwrite((const char *) &val, sizeof(val), 1, fp);
     }
 
+    void write_raw(const void * data, size_t size) {
+        fwrite(data, size, 1, fp);
+    }
+
     template<typename T>
     void write_val(const std::string & key, enum gguf_type type, const T & val) {
         write_str(key);

From b2571af255e6a1ed42dffeb74ef51390f2cf5144 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 14:28:17 +0300
Subject: [PATCH 110/242] gguf : start implementing quantization (WIP)

---
 Makefile               |  2 +-
 examples/gguf/gguf.cpp | 10 +++++++---
 gguf-llama.cpp         |  8 ++++++--
 3 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/Makefile b/Makefile
index f5922c95d2f98..304b3035d64d0 100644
--- a/Makefile
+++ b/Makefile
@@ -393,7 +393,7 @@ $(LIB_PRE)embdinput$(DSO_EXT): examples/embd-input/embd-input.h examples/embd-in
 embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-test.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %$(DSO_EXT),$(filter-out %.h,$(filter-out %.hpp,$^))) -o $@ $(LDFLAGS) -L. -lembdinput
 
-gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o $(OBJS)
+gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o gguf-llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
 gguf-llama-simple: examples/gguf/gguf-llama-simple.cpp                            build-info.h ggml.o gguf-llama.o common.o $(OBJS)
diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index a1b8edc71f373..6f454a2047ddd 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,5 +1,6 @@
 #include "ggml.h"
 #include "gguf-util.h"
+#include "gguf-llama.h"
 
 #include <cstdio>
 #include <cinttypes>
@@ -7,14 +8,14 @@
 #include <sstream>
 #include <fstream>
 #include <vector>
-
+/*
 template<typename T>
 static std::string to_string(const T & val) {
     std::stringstream ss;
     ss << val;
     return ss.str();
 }
-
+*/
 void gguf_ex_write_str(std::ofstream & fout, const std::string & val) {
     const int32_t n = val.size();
     fout.write((const char *) &n, sizeof(n));
@@ -414,7 +415,7 @@ int main(int argc, char ** argv) {
     const std::string fname(argv[1]);
     const std::string mode (argv[2]);
 
-    GGML_ASSERT((mode == "r" || mode == "w") && "mode must be r or w");
+    GGML_ASSERT((mode == "r" || mode == "w" || mode == "q") && "mode must be r, w or q");
 
     if (mode == "w") {
         GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
@@ -422,6 +423,9 @@ int main(int argc, char ** argv) {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_2(fname) && "failed to read gguf file");
+    } else if (mode == "q") {
+        llama_model_quantize_params params = llama_model_quantize_default_params();
+        llama_model_quantize(fname.c_str(), "quant.gguf", &params);
     }
 
     return 0;
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index defe26fe04ab8..f1755fef551bf 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -738,15 +738,19 @@ struct gguf_file_saver {
         info_offset  = file.tell();
         size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
         file.write_zeros(count);
+        printf("info_offset = %zu\n", info_offset);
+        file.seek(info_offset, SEEK_SET);
+        GGML_ASSERT(info_offset == file.tell());
     }
 
     size_t write_tensor_info(llama_load_tensor & tensor) {
         size_t total_written = 0;
-        file.seek(0, info_offset);
+        file.seek(info_offset, SEEK_SET);
+        GGML_ASSERT(info_offset == file.tell());
         total_written += file.write_str(tensor.name);
 
         int32_t n_dims = tensor.ne.size();
-        file.write_i32(n_dims);
+        total_written += file.write_i32(n_dims);
         for (int32_t i = 0; i < n_dims; ++i) {
             total_written += file.write_i32(i);
         }

From fa7c39540cff0ce6f6d245baeb15ac1ebe6cdd69 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 15:55:58 +0300
Subject: [PATCH 111/242] gguf : start implementing quantization (WIP)

---
 gguf-llama.cpp | 15 +++++++++++----
 gguf-util.h    | 12 ++++++++----
 2 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index f1755fef551bf..eecefc0f6b1bd 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -525,6 +525,11 @@ struct ggml_context * ctx_data = NULL;
 
         // TODO make keysconstants in header
         // TODO: read all hparams from file
+        int q_ver_idx = gguf_find_key (gguf_ctx, "general.quantization_version");
+        if (q_ver_idx != -1) {
+            hparams.ftype = gguf_get_val_u32(gguf_ctx, q_ver_idx);
+        }
+
         hparams.n_vocab = read_n_vocab();
         hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
@@ -738,27 +743,29 @@ struct gguf_file_saver {
         info_offset  = file.tell();
         size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
         file.write_zeros(count);
-        printf("info_offset = %zu\n", info_offset);
         file.seek(info_offset, SEEK_SET);
         GGML_ASSERT(info_offset == file.tell());
     }
 
-    size_t write_tensor_info(llama_load_tensor & tensor) {
+    size_t write_tensor_info(llama_load_tensor & tensor, enum ggml_type type) {
         size_t total_written = 0;
         file.seek(info_offset, SEEK_SET);
         GGML_ASSERT(info_offset == file.tell());
         total_written += file.write_str(tensor.name);
+printf("total_written = %zu, name = %s\n", total_written, tensor.name.c_str());
 
         int32_t n_dims = tensor.ne.size();
         total_written += file.write_i32(n_dims);
         for (int32_t i = 0; i < n_dims; ++i) {
-            total_written += file.write_i32(i);
+            total_written += file.write_i32(tensor.ne[i]);
         }
 
+        total_written += file.write_i32(type);
         total_written += file.write_u64(tensor_offset);
         info_offset   += total_written;
 
         file.seek(0, SEEK_END);
+        printf("total_written = %zu\n", total_written);
 
         return total_written;
     }
@@ -781,7 +788,7 @@ struct gguf_file_saver {
             default: GGML_ASSERT(false);
         }
         
-        write_tensor_info(tensor);
+        write_tensor_info(tensor, new_type);
         file.write_raw(new_data, new_size);
         size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
         size_t pad = padded_size - new_size;
diff --git a/gguf-util.h b/gguf-util.h
index 17f9dc9680dc9..ed7d53f691891 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -109,18 +109,22 @@ struct gguf_file {
     size_t write_str(const std::string & val) {
         size_t total_written = 0;
         const int32_t n = val.size();
-        total_written += fwrite((const char *) &n, sizeof(n), 1, fp);
-        total_written += fwrite(val.c_str(), n, 1, fp);
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        total_written += sizeof(n);
+        fwrite(val.c_str(), n, 1, fp);
+        total_written += n;
 
         return total_written;
     }
 
     size_t write_i32(int32_t val) {
-        return fwrite((const char *) &val, sizeof(val), 1, fp);
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+        return sizeof(val);
     }
 
     size_t write_u64(size_t val) {
-        return fwrite((const char *) &val, sizeof(val), 1, fp);
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+        return sizeof(val);
     }
 
     void write_raw(const void * data, size_t size) {

From 1fc3d30b71a707187eb1f995c4776db7aaa6265a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 16:09:47 +0300
Subject: [PATCH 112/242] gguf : start implementing quantization (WIP)

---
 examples/gguf/gguf.cpp | 2 +-
 gguf-llama.cpp         | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 6f454a2047ddd..08f2b6322041a 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -421,7 +421,7 @@ int main(int argc, char ** argv) {
         GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
     } else if (mode == "r") {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
-        GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
+        //GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_2(fname) && "failed to read gguf file");
     } else if (mode == "q") {
         llama_model_quantize_params params = llama_model_quantize_default_params();
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index eecefc0f6b1bd..ea721a0c72858 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -527,7 +527,7 @@ struct ggml_context * ctx_data = NULL;
         // TODO: read all hparams from file
         int q_ver_idx = gguf_find_key (gguf_ctx, "general.quantization_version");
         if (q_ver_idx != -1) {
-            hparams.ftype = gguf_get_val_u32(gguf_ctx, q_ver_idx);
+            hparams.ftype = (enum llama_ftype) gguf_get_val_u32(gguf_ctx, q_ver_idx);
         }
 
         hparams.n_vocab = read_n_vocab();

From 202eab04d3f5a06304b9fd43b4b6b079d3f76dfa Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 16:39:05 +0300
Subject: [PATCH 113/242] gguf : quantization is working

---
 examples/gguf/gguf.cpp | 2 +-
 gguf-llama.cpp         | 7 ++-----
 gguf-util.h            | 4 ++++
 3 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 08f2b6322041a..6f454a2047ddd 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -421,7 +421,7 @@ int main(int argc, char ** argv) {
         GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
     } else if (mode == "r") {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
-        //GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
+        GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_2(fname) && "failed to read gguf file");
     } else if (mode == "q") {
         llama_model_quantize_params params = llama_model_quantize_default_params();
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index ea721a0c72858..700d6009bcddb 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -752,7 +752,6 @@ struct gguf_file_saver {
         file.seek(info_offset, SEEK_SET);
         GGML_ASSERT(info_offset == file.tell());
         total_written += file.write_str(tensor.name);
-printf("total_written = %zu, name = %s\n", total_written, tensor.name.c_str());
 
         int32_t n_dims = tensor.ne.size();
         total_written += file.write_i32(n_dims);
@@ -765,8 +764,7 @@ printf("total_written = %zu, name = %s\n", total_written, tensor.name.c_str());
         info_offset   += total_written;
 
         file.seek(0, SEEK_END);
-        printf("total_written = %zu\n", total_written);
-
+        
         return total_written;
     }
 
@@ -936,8 +934,7 @@ struct llama_model_loader {
         } else {
             gguf_file & file = file_loader->file;
             file.seek(lt.file_off, SEEK_SET);
-            // TODO
-            //file.read_raw(lt.data, lt.size);
+            file.read_raw(lt.data, lt.size);
         }
 
         if (0) {
diff --git a/gguf-util.h b/gguf-util.h
index ed7d53f691891..6395cf304e510 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -131,6 +131,10 @@ struct gguf_file {
         fwrite(data, size, 1, fp);
     }
 
+    void read_raw(void * data, size_t size) {
+        fread(data, size, 1, fp);
+    }
+
     template<typename T>
     void write_val(const std::string & key, enum gguf_type type, const T & val) {
         write_str(key);

From 60d540831b626443caddad12e04792bce91d3b4f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sat, 12 Aug 2023 21:42:31 +0300
Subject: [PATCH 114/242] gguf : roper closing of file

---
 gguf-util.h | 41 ++++++++++++++++++++++++++++++++---------
 1 file changed, 32 insertions(+), 9 deletions(-)

diff --git a/gguf-util.h b/gguf-util.h
index 6395cf304e510..70673404e6f0f 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -105,7 +105,6 @@ struct gguf_file {
         GGML_ASSERT(ret == 0); // same
     }
 
-    
     size_t write_str(const std::string & val) {
         size_t total_written = 0;
         const int32_t n = val.size();
@@ -127,14 +126,6 @@ struct gguf_file {
         return sizeof(val);
     }
 
-    void write_raw(const void * data, size_t size) {
-        fwrite(data, size, 1, fp);
-    }
-
-    void read_raw(void * data, size_t size) {
-        fread(data, size, 1, fp);
-    }
-
     template<typename T>
     void write_val(const std::string & key, enum gguf_type type, const T & val) {
         write_str(key);
@@ -155,6 +146,7 @@ struct gguf_file {
         fwrite((const char *) &n, sizeof(n), 1, fp);
         fwrite(val.data(), sizeof(T), n, fp);
     }
+    
     template<>
     void write_val<std::string>(const std::string & key, enum gguf_type type, const std::string & val) {
         write_str(key);
@@ -188,6 +180,37 @@ struct gguf_file {
             fputc(0, fp);
         }
     }
+    
+    void read_raw(void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        std::size_t ret = std::fread(ptr, len, 1, fp);
+        if (ferror(fp)) {
+            throw std::runtime_error(format("read error: %s", strerror(errno)));
+        }
+        if (ret != 1) {
+            throw std::runtime_error(std::string("unexpectedly reached end of file"));
+        }
+    }
+
+    void write_raw(const void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        size_t ret = std::fwrite(ptr, len, 1, fp);
+        if (ret != 1) {
+            throw std::runtime_error(format("write error: %s", strerror(errno)));
+        }
+    }
+
+    ~gguf_file() {
+        if (fp) {
+            std::fclose(fp);
+        }
+    }
 };
 
 #if defined(_WIN32)

From 5d81a715d4a8e1e6d9dad2d978510d997d20595c Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 21:45:45 +0200
Subject: [PATCH 115/242] gguf.py : no need to convert tensors twice

---
 gguf.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/gguf.py b/gguf.py
index 5eb21ee05fb6f..0854418a6d842 100644
--- a/gguf.py
+++ b/gguf.py
@@ -179,20 +179,20 @@ def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool =
     def ggml_pad(x: int, n: int) -> int:
         return ((x + n - 1) // n) * n
 
-    def add_tensor_info(self, name: str, tensor: np.ndarray):
+    def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int):
         encoded_name = name.encode("utf8")
         self.ti_data += struct.pack("<I", len(encoded_name))
         self.ti_data += encoded_name
-        n_dims = len(tensor.shape)
+        n_dims = len(tensor_shape)
         self.ti_data += struct.pack("<I", n_dims)
         for i in range(n_dims):
-            self.ti_data += struct.pack("<I", tensor.shape[n_dims - 1 - i])
+            self.ti_data += struct.pack("<I", tensor_shape[n_dims - 1 - i])
 
-        assert tensor.dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
-        dtype = GGMLQuantizationType.F32 if tensor.dtype == np.float32 else GGMLQuantizationType.F16
+        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
+        dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
         self.ti_data += struct.pack("<I", dtype)
         self.ti_data += struct.pack("<Q", self.offset_tensor)
-        self.offset_tensor += GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment)
+        self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
         self.ti_data_count += 1
 
     def write_tensor_to_file(self, tensor: np.ndarray):

From 8f09157ec96b7b0f9b8f2c861009fc47a13c7262 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 21:48:58 +0200
Subject: [PATCH 116/242] convert-gptneox-h5-to-gguf.py : no need to convert
 tensors twice

---
 convert-gptneox-h5-to-gguf.py | 65 ++++++++++++++++++-----------------
 1 file changed, 33 insertions(+), 32 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 22508bd3dae56..0c0cd6bb13c73 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -43,7 +43,6 @@ def bytes_to_unicode():
 
 # output in the same directory as the model
 dir_model = sys.argv[1]
-fname_out = sys.argv[1] + "/ggml-model.bin"
 last_dir = os.path.basename(os.path.normpath(dir_model))
 
 # possible tensor data types
@@ -59,7 +58,8 @@ def bytes_to_unicode():
     if ftype < 0 or ftype > 1:
         print("Invalid ftype: " + str(ftype))
         sys.exit(1)
-    fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
 print("gguf: loading model "+last_dir)
 
@@ -82,7 +82,6 @@ def bytes_to_unicode():
 block_count = hparams["num_hidden_layers"]
 
 gguf_writer.add_name(last_dir)
-gguf_writer.add_description("gguf test model")
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
@@ -201,22 +200,30 @@ def bytes_to_unicode():
         sys.exit()
 
     n_dims = len(data.shape)
+    data_dtype = data.dtype 
 
-    # ftype == 0 -> float32, ftype == 1 -> float16
-    ftype_cur = 0
-    if ftype != 0:
-        if name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-            ftype_cur = 1
-        else:
-            data = data.astype(np.float32)
-            ftype_cur = 0
-    else:
-        if data.dtype != np.float32:
-            data = data.astype(np.float32)
-            ftype_cur = 0
+#    print( name + " dims " + str(n_dims) + " dtype " + str(data.dtype) )
+
+    if data.dtype != np.float16 and data.dtype != np.float32:
+        # convert any unsupported data types to float32
+        data_dtype = np.float32
+    elif ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        data_dtype = np.float16
+
+    nelements = 1
+
+    for i in range(n_dims):
+        nelements *= data.shape[n_dims - 1 - i]
 
-    gguf_writer.add_tensor_info(name, data)
+    data_nbytes = 0
+    if data_dtype == np.float16:
+        data_nbytes = nelements * 2
+    elif data_dtype == np.float32:
+        data_nbytes = nelements * 4
+
+
+    gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
@@ -226,7 +233,7 @@ def bytes_to_unicode():
 gguf_writer.write_ti_data_to_file()
 
 # tensor data
-print("gguf: write tensor data")
+print("gguf: convert and write tensor data")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -236,20 +243,14 @@ def bytes_to_unicode():
         continue
 
     n_dims = len(data.shape)
-
-    # ftype == 0 -> float32, ftype == 1 -> float16
-    ftype_cur = 0
-    if ftype != 0:
-        if name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-            ftype_cur = 1
-        else:
-            data = data.astype(np.float32)
-            ftype_cur = 0
-    else:
-        if data.dtype != np.float32:
-            data = data.astype(np.float32)
-            ftype_cur = 0
+    data_dtype = data.dtype 
+
+    if data_dtype != np.float16 and data_dtype != np.float32:
+        # convert any unsupported data types to float32
+        data = data.astype(np.float32)
+    elif ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        data = data.astype(np.float16)
 
     gguf_writer.write_tensor_to_file(data)
 

From 4cef57c81a2bc7afff72869e6b9659177a11d334 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 21:50:24 +0200
Subject: [PATCH 117/242] convert-llama-h5-to-gguf.py : no need to convert
 tensors twice

---
 convert-llama-h5-to-gguf.py | 67 +++++++++++++++++++------------------
 1 file changed, 34 insertions(+), 33 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 0b477a1336570..519b73966fa3a 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -32,7 +32,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 # output in the same directory as the model
 dir_model = sys.argv[1]
-fname_out = sys.argv[1] + "/ggml-model.bin"
 last_dir = os.path.basename(os.path.normpath(dir_model))
 
 
@@ -49,7 +48,8 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     if ftype < 0 or ftype > 1:
         print("Invalid ftype: " + str(ftype))
         sys.exit(1)
-    fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
 print("gguf: loading model "+last_dir)
     
@@ -72,8 +72,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 head_count  = hparams["num_attention_heads"]
 block_count = hparams["num_hidden_layers"]
 
-gguf_writer.add_name("llama2-7b")
-gguf_writer.add_description("gguf test model")
+gguf_writer.add_name(last_dir)
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
@@ -186,22 +185,30 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         sys.exit()
 
     n_dims = len(data.shape)
+    data_dtype = data.dtype 
 
-    # ftype == 0 -> float32, ftype == 1 -> float16
-    ftype_cur = 0
-    if ftype != 0:
-        if name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-            ftype_cur = 1
-        else:
-            data = data.astype(np.float32)
-            ftype_cur = 0
-    else:
-        if data.dtype != np.float32:
-            data = data.astype(np.float32)
-            ftype_cur = 0
+#    print( name + " dims " + str(n_dims) + " dtype " + str(data.dtype) )
+
+    if data.dtype != np.float16 and data.dtype != np.float32:
+        # convert any unsupported data types to float32
+        data_dtype = np.float32
+    elif ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        data_dtype = np.float16
+
+    nelements = 1
+
+    for i in range(n_dims):
+        nelements *= data.shape[n_dims - 1 - i]
 
-    gguf_writer.add_tensor_info(name, data)
+    data_nbytes = 0
+    if data_dtype == np.float16:
+        data_nbytes = nelements * 2
+    elif data_dtype == np.float32:
+        data_nbytes = nelements * 4
+
+
+    gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 
 
 print("gguf: write header")
@@ -212,7 +219,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.write_ti_data_to_file()
 
 # tensor data
-print("gguf: write tensor data")
+print("gguf: convert and write tensor data")
 
 for name in list_vars.keys():
     data = list_vars[name].squeeze().numpy()
@@ -226,20 +233,14 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         data = permute(data, head_count)
 
     n_dims = len(data.shape)
-
-    # ftype == 0 -> float32, ftype == 1 -> float16
-    ftype_cur = 0
-    if ftype != 0:
-        if name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-            ftype_cur = 1
-        else:
-            data = data.astype(np.float32)
-            ftype_cur = 0
-    else:
-        if data.dtype != np.float32:
-            data = data.astype(np.float32)
-            ftype_cur = 0
+    data_dtype = data.dtype 
+
+    if data_dtype != np.float16 and data_dtype != np.float32:
+        # convert any unsupported data types to float32
+        data = data.astype(np.float32)
+    elif ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        data = data.astype(np.float16)
 
     gguf_writer.write_tensor_to_file(data)
 

From f8218477b37d14b522e6422bdbb93302d8b8f421 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 22:29:35 +0200
Subject: [PATCH 118/242] convert-gptneox-h5-to-gguf.py : simplify nbytes

---
 convert-gptneox-h5-to-gguf.py | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 0c0cd6bb13c73..770a5d9caa313 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -211,17 +211,7 @@ def bytes_to_unicode():
         # if f16 desired, convert any float32 2-dim weight tensors to float16
         data_dtype = np.float16
 
-    nelements = 1
-
-    for i in range(n_dims):
-        nelements *= data.shape[n_dims - 1 - i]
-
-    data_nbytes = 0
-    if data_dtype == np.float16:
-        data_nbytes = nelements * 2
-    elif data_dtype == np.float32:
-        data_nbytes = nelements * 4
-
+    data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
 
     gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 

From e606ffeaeed3bca9d5a1d97774eaf1a3d602088f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 22:30:35 +0200
Subject: [PATCH 119/242] convert-llama-h5-to-gguf.py : simplify nbytes

---
 convert-llama-h5-to-gguf.py | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 519b73966fa3a..bf6ff6aa78edd 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -196,17 +196,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         # if f16 desired, convert any float32 2-dim weight tensors to float16
         data_dtype = np.float16
 
-    nelements = 1
-
-    for i in range(n_dims):
-        nelements *= data.shape[n_dims - 1 - i]
-
-    data_nbytes = 0
-    if data_dtype == np.float16:
-        data_nbytes = nelements * 2
-    elif data_dtype == np.float32:
-        data_nbytes = nelements * 4
-
+    data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
 
     gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 

From 5e58ffa1ed8a97297429367947206dbece94e30d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 12 Aug 2023 23:50:58 +0200
Subject: [PATCH 120/242] gptneox-main.cpp : n_layer --> n_block

---
 gptneox-main.cpp | 135 +++++++++++++++++++----------------------------
 1 file changed, 53 insertions(+), 82 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index f2be93e4b1ae6..7420daf967399 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -24,13 +24,13 @@ struct gpt_neox_hparams {
     uint32_t n_ctx    = 0;
     uint32_t n_embd   = 0;
     uint32_t n_head   = 0;
-    uint32_t n_layer  = 0;
+    uint32_t n_block  = 0;
     uint32_t n_rot    = 0; // rotary_pct * (n_embd / n_head)
     bool par_res = true;
     float norm_eps = 1e-5;
 };
 
-struct gpt_neox_layer {
+struct gpt_neox_block {
     // pre normalization
     struct ggml_tensor * ln_1_g;
     struct ggml_tensor * ln_1_b;
@@ -65,7 +65,7 @@ struct gpt_neox_model {
 
     struct ggml_tensor * lmh_g; // language model head
 
-    std::vector<gpt_neox_layer> layers;
+    std::vector<gpt_neox_block> blocks;
 
     // key + value memory
     struct ggml_tensor * memory_k;
@@ -415,7 +415,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
                   if (keyidx != -1) { hparams.n_head = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
 
         if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.layer_count");
-                  if (keyidx != -1) { hparams.n_layer = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+                  if (keyidx != -1) { hparams.n_block = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
 
         if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.rope.dimension_count");
                   if (keyidx != -1) { hparams.n_rot = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
@@ -434,7 +434,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         printf("%s: n_ctx    = %d\n", __func__, hparams.n_ctx);
         printf("%s: n_embd   = %d\n", __func__, hparams.n_embd);
         printf("%s: n_head   = %d\n", __func__, hparams.n_head);
-        printf("%s: n_layer  = %d\n", __func__, hparams.n_layer);
+        printf("%s: n_block  = %d\n", __func__, hparams.n_block);
         printf("%s: n_rot    = %d\n", __func__, hparams.n_rot);
         printf("%s: par_res  = %d\n", __func__, hparams.par_res);
         printf("%s: norm_eps = %g\n", __func__, hparams.norm_eps);
@@ -545,9 +545,9 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
     // prepare memory for the weights
     {
-        const int n_layer = model.hparams.n_layer;
+        const int n_block = model.hparams.n_block;
 
-        model.layers.resize(n_layer);
+        model.blocks.resize(n_block);
 
         model.wte    = ggml_get_tensor(ctx, "transformer.token_embd.weight");
         model.ln_f_g = ggml_get_tensor(ctx, "transformer.output_norm.weight");
@@ -560,47 +560,47 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         model.tensors["transformer.output_norm.bias"]   = model.ln_f_b;
         model.tensors["transformer.output.weight"] = model.lmh_g;
 
-        for (int i = 0; i < n_layer; ++i) {
-            auto & layer = model.layers[i];
+        for (int i = 0; i < n_block; ++i) {
+            auto & block = model.blocks[i];
 
             std::string blocknamestart = "transformer.blocks." + std::to_string(i) + ".";
 
-            layer.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm.weight" );
-            layer.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm.bias" );
+            block.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm.weight" );
+            block.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm.bias" );
 
-            layer.c_attn_attn_w   = get_tensor_ex(ctx, blocknamestart + "attn_qkv.weight" );
-            layer.c_attn_attn_b   = get_tensor_ex(ctx ,blocknamestart + "attn_qkv.bias" );
+            block.c_attn_attn_w   = get_tensor_ex(ctx, blocknamestart + "attn_qkv.weight" );
+            block.c_attn_attn_b   = get_tensor_ex(ctx ,blocknamestart + "attn_qkv.bias" );
 
-            layer.c_attn_proj_w   = get_tensor_ex(ctx, blocknamestart + "attn_output.weight" );
-            layer.c_attn_proj_b   = get_tensor_ex(ctx, blocknamestart + "attn_output.bias" );
+            block.c_attn_proj_w   = get_tensor_ex(ctx, blocknamestart + "attn_output.weight" );
+            block.c_attn_proj_b   = get_tensor_ex(ctx, blocknamestart + "attn_output.bias" );
 
-            layer.ln_2_g          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.weight" );
-            layer.ln_2_b          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.bias");
+            block.ln_2_g          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.weight" );
+            block.ln_2_b          = get_tensor_ex(ctx, blocknamestart + "ffn_norm.bias");
 
-            layer.c_mlp_fc_w      = get_tensor_ex(ctx, blocknamestart + "ffn_up.weight" );
-            layer.c_mlp_fc_b      = get_tensor_ex(ctx, blocknamestart + "ffn_up.bias" );
+            block.c_mlp_fc_w      = get_tensor_ex(ctx, blocknamestart + "ffn_up.weight" );
+            block.c_mlp_fc_b      = get_tensor_ex(ctx, blocknamestart + "ffn_up.bias" );
 
-            layer.c_mlp_proj_w    = get_tensor_ex(ctx, blocknamestart + "ffn_down.weight" );
-            layer.c_mlp_proj_b    = get_tensor_ex(ctx, blocknamestart + "ffn_down.bias" );
+            block.c_mlp_proj_w    = get_tensor_ex(ctx, blocknamestart + "ffn_down.weight" );
+            block.c_mlp_proj_b    = get_tensor_ex(ctx, blocknamestart + "ffn_down.bias" );
 
             // map by name
-            model.tensors[blocknamestart + "attn_norm.weight"] = layer.ln_1_g;
-            model.tensors[blocknamestart + "attn_norm.bias"]   = layer.ln_1_b;
+            model.tensors[blocknamestart + "attn_norm.weight"] = block.ln_1_g;
+            model.tensors[blocknamestart + "attn_norm.bias"]   = block.ln_1_b;
 
-            model.tensors[blocknamestart + "attn_qkv.weight"] = layer.c_attn_attn_w;
-            model.tensors[blocknamestart + "attn_qkv.bias"]   = layer.c_attn_attn_b;
+            model.tensors[blocknamestart + "attn_qkv.weight"] = block.c_attn_attn_w;
+            model.tensors[blocknamestart + "attn_qkv.bias"]   = block.c_attn_attn_b;
 
-            model.tensors[blocknamestart + "attn_output.weight"] = layer.c_attn_proj_w;
-            model.tensors[blocknamestart + "attn_output.bias"]   = layer.c_attn_proj_b;
+            model.tensors[blocknamestart + "attn_output.weight"] = block.c_attn_proj_w;
+            model.tensors[blocknamestart + "attn_output.bias"]   = block.c_attn_proj_b;
 
-            model.tensors[blocknamestart + "ffn_norm.weight"] = layer.ln_2_g;
-            model.tensors[blocknamestart + "ffn_norm.bias"]   = layer.ln_2_b;
+            model.tensors[blocknamestart + "ffn_norm.weight"] = block.ln_2_g;
+            model.tensors[blocknamestart + "ffn_norm.bias"]   = block.ln_2_b;
 
-            model.tensors[blocknamestart + "ffn_up.weight"] = layer.c_mlp_fc_w;
-            model.tensors[blocknamestart + "ffn_up.bias"]   = layer.c_mlp_fc_b;
+            model.tensors[blocknamestart + "ffn_up.weight"] = block.c_mlp_fc_w;
+            model.tensors[blocknamestart + "ffn_up.bias"]   = block.c_mlp_fc_b;
 
-            model.tensors[blocknamestart + "ffn_down.weight"] = layer.c_mlp_proj_w;
-            model.tensors[blocknamestart + "ffn_down.bias"]   = layer.c_mlp_proj_b;
+            model.tensors[blocknamestart + "ffn_down.weight"] = block.c_mlp_proj_w;
+            model.tensors[blocknamestart + "ffn_down.bias"]   = block.c_mlp_proj_b;
         }
     }
 
@@ -610,10 +610,10 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         const auto & hparams = model.hparams;
 
         const int n_embd  = hparams.n_embd;
-        const int n_layer = hparams.n_layer;
+        const int n_block = hparams.n_block;
         const int n_ctx   = hparams.n_ctx;
 
-        const int64_t n_mem      = n_layer*n_ctx;
+        const int64_t n_mem      = n_block*n_ctx;
         const int64_t n_elements = n_embd*n_mem;
 
         // create the ggml context
@@ -647,37 +647,23 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
 // feed-forward network
 ggml_tensor * gpt_neox_ff(
-        const gpt_neox_layer &layer,
+        const gpt_neox_block &block,
         ggml_context * ctx0,
         ggml_tensor * inp) {
     ggml_tensor * cur = ggml_norm(ctx0, inp);
 
-    cur = ggml_add(ctx0,
-        ggml_mul(ctx0,
-            ggml_repeat(ctx0, layer.ln_2_g, cur),
-            cur),
-        ggml_repeat(ctx0, layer.ln_2_b, cur));
-
-    cur = ggml_mul_mat(ctx0,
-            layer.c_mlp_fc_w,
-            cur);
-
-    cur = ggml_add(ctx0,
-            ggml_repeat(ctx0, layer.c_mlp_fc_b, cur),
-            cur);
+    cur = ggml_add(ctx0, ggml_mul(ctx0, ggml_repeat(ctx0, block.ln_2_g, cur), cur), ggml_repeat(ctx0, block.ln_2_b, cur));
+    cur = ggml_mul_mat(ctx0, block.c_mlp_fc_w, cur);
+    cur = ggml_add(ctx0, ggml_repeat(ctx0, block.c_mlp_fc_b, cur), cur);
 
     // GELU activation
     cur = ggml_gelu(ctx0, cur);
 
     // projection
     // cur = proj_w*cur + proj_b
-    cur = ggml_mul_mat(ctx0,
-            layer.c_mlp_proj_w,
-            cur);
+    cur = ggml_mul_mat(ctx0, block.c_mlp_proj_w, cur);
 
-    cur = ggml_add(ctx0,
-            ggml_repeat(ctx0, layer.c_mlp_proj_b, cur),
-            cur);
+    cur = ggml_add(ctx0, ggml_repeat(ctx0, block.c_mlp_proj_b, cur), cur);
     return cur;
 }
 
@@ -701,7 +687,7 @@ bool gpt_neox_eval(
     const auto & hparams = model.hparams;
 
     const int n_embd  = hparams.n_embd;
-    const int n_layer = hparams.n_layer;
+    const int n_block = hparams.n_block;
     const int n_ctx   = hparams.n_ctx;
     const int n_head  = hparams.n_head;
     const int n_vocab = hparams.n_vocab;
@@ -747,7 +733,7 @@ bool gpt_neox_eval(
     // wte
     struct ggml_tensor * inpL = ggml_get_rows(ctx0, model.wte, embd);
 
-    for (int il = 0; il < n_layer; ++il) {
+    for (int il = 0; il < n_block; ++il) {
         struct ggml_tensor * cur;
 
         ggml_set_scratch(ctx0, { 0, scr0_size, scr0, });
@@ -758,22 +744,15 @@ bool gpt_neox_eval(
                 cur = ggml_norm(ctx0, inpL);
 
                 cur = ggml_add(ctx0,
-                        ggml_mul(ctx0,
-                            ggml_repeat(ctx0, model.layers[il].ln_1_g, cur),
-                            cur),
-                        ggml_repeat(ctx0, model.layers[il].ln_1_b, cur));
+                        ggml_mul(ctx0, ggml_repeat(ctx0, model.blocks[il].ln_1_g, cur), cur),
+                        ggml_repeat(ctx0, model.blocks[il].ln_1_b, cur));
             }
 
             // compute QKV
             {
 
-                cur = ggml_mul_mat(ctx0,
-                        model.layers[il].c_attn_attn_w,
-                        cur);
-
-                cur = ggml_add(ctx0,
-                        ggml_repeat(ctx0, model.layers[il].c_attn_attn_b, cur),
-                        cur);
+                cur = ggml_mul_mat(ctx0, model.blocks[il].c_attn_attn_w, cur);
+                cur = ggml_add(ctx0, ggml_repeat(ctx0, model.blocks[il].c_attn_attn_b, cur), cur);
             }
 
             struct ggml_tensor * Qcur = ggml_cont(ctx0, ggml_view_3d(ctx0, cur, n_embd/n_head, n_head, N, cur->nb[1]/n_head, cur->nb[1], 0*sizeof(float)*n_embd/n_head));
@@ -798,10 +777,7 @@ bool gpt_neox_eval(
             }
 
             // Q = Qcur.contiguous().view(n_embd/n_head, n_head, N).permute(0, 2, 1, 3)
-            struct ggml_tensor * Q =
-                ggml_permute(ctx0,
-                        Qcur,
-                        0, 2, 1, 3);
+            struct ggml_tensor * Q = ggml_permute(ctx0, Qcur, 0, 2, 1, 3);
 
             // K = Kmem.view(n_embd/n_head, n_head, n_past + N).permute(0, 2, 1, 3)
             struct ggml_tensor * K =
@@ -842,17 +818,12 @@ bool gpt_neox_eval(
             struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
 
             // cur = KQV_merged.contiguous().view(n_embd, N)
-            cur = ggml_cpy(ctx0,
-                    KQV_merged,
-                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+            cur = ggml_cpy(ctx0, KQV_merged, ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
 
             // projection
             {
-                cur = ggml_mul_mat(ctx0,
-                        model.layers[il].c_attn_proj_w,
-                        cur);
-
-                cur = ggml_add(ctx0, ggml_repeat(ctx0, model.layers[il].c_attn_proj_b, cur), cur);
+                cur = ggml_mul_mat(ctx0, model.blocks[il].c_attn_proj_w, cur);
+                cur = ggml_add(ctx0, ggml_repeat(ctx0, model.blocks[il].c_attn_proj_b, cur), cur);
             }
         }
 
@@ -861,7 +832,7 @@ bool gpt_neox_eval(
         if (hparams.par_res == 0) {
             struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpL);
 
-            cur = gpt_neox_ff(model.layers[il], ctx0, inpFF);
+            cur = gpt_neox_ff(model.blocks[il], ctx0, inpFF);
 
             // input for next layer
             inpL = ggml_add(ctx0, cur, inpFF);
@@ -870,7 +841,7 @@ bool gpt_neox_eval(
 
             // this is independent of the self-attention result, so it could be done in parallel to the self-attention
             // note here we pass inpL instead of cur
-            cur = gpt_neox_ff(model.layers[il], ctx0, inpL);
+            cur = gpt_neox_ff(model.blocks[il], ctx0, inpL);
 
             // layer input + FF
             cur  = ggml_add(ctx0, cur, inpFF);

From 8b5f0c506708b06fde1362fc2cbb464222cf44aa Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 00:00:32 +0200
Subject: [PATCH 121/242] constants.py : n_layer --> n_block

---
 constants.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/constants.py b/constants.py
index ae6e719fb555e..7fa238a73278c 100644
--- a/constants.py
+++ b/constants.py
@@ -18,7 +18,7 @@
 # LLM
 KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
 KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
-KEY_LLM_LAYER_COUNT              = "{llm}.layer_count"
+KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
 KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
 KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
 KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"

From d2ce9cfe8d0c75ef4166bac28e93854319219390 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 00:01:20 +0200
Subject: [PATCH 122/242] gguf.py : n_layer --> n_block

---
 gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index 0854418a6d842..75ba2870afc9a 100644
--- a/gguf.py
+++ b/gguf.py
@@ -253,9 +253,9 @@ def add_embedding_length(self, llm: str, length: int):
         self.add_uint32(
             constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
 
-    def add_layer_count(self, llm: str, length: int):
+    def add_block_count(self, llm: str, length: int):
         self.add_uint32(
-            constants.KEY_LLM_LAYER_COUNT.format(llm=llm), length)
+            constants.KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
 
     def add_feed_forward_length(self, llm: str, length: int):
         self.add_uint32(

From 489616e12652235ecb5c7f3839a2f5e6b617bb82 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 00:02:04 +0200
Subject: [PATCH 123/242] convert-gptneox-h5-to-gguf.py : n_layer --> n_block

---
 convert-gptneox-h5-to-gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index 770a5d9caa313..ac4b2ff8860fd 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -85,7 +85,7 @@ def bytes_to_unicode():
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_layer_count(llm_arch, block_count)
+gguf_writer.add_block_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.add_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
 gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
@@ -116,7 +116,7 @@ def bytes_to_unicode():
 
     vocab_size = len( tokenizer_json["model"]["vocab"] )
 
-    # from ggllm.cpp falcon_convert.py
+    # ref: https://github.com/cmp-nct/ggllm.cpp/blob/master/falcon_convert.py
     tokenizer = AutoTokenizer.from_pretrained(dir_model)
 
     reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()}

From e91a2224e49aeed6e1382643dcb053ec6c168440 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 00:02:44 +0200
Subject: [PATCH 124/242] convert-llama-h5-to-gguf.py : n_layer --> n_block

---
 convert-llama-h5-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index bf6ff6aa78edd..055b6b78da9c2 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -76,7 +76,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_layer_count(llm_arch, block_count)
+gguf_writer.add_block_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
 gguf_writer.add_head_count(llm_arch, head_count)

From c7bd8c147ccd7411d1cd9a2a394d2d9b743d7294 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 00:03:32 +0200
Subject: [PATCH 125/242] gptneox-main.cpp : n_layer --> n_block

---
 gptneox-main.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 7420daf967399..4773dfd690461 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -414,7 +414,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.attention.head_count");
                   if (keyidx != -1) { hparams.n_head = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
 
-        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.layer_count");
+        if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.block_count");
                   if (keyidx != -1) { hparams.n_block = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
 
         if (ok) { keyidx = gguf_find_key(ggufctx, "gptneox.rope.dimension_count");

From 9bf5a7efcb55fb96c08e1fab44e2ebe61964dba6 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 01:27:38 +0200
Subject: [PATCH 126/242] Update gguf_tensor_map.py

---
 gguf_tensor_map.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gguf_tensor_map.py b/gguf_tensor_map.py
index 644c5914de9b2..d73788bb46877 100644
--- a/gguf_tensor_map.py
+++ b/gguf_tensor_map.py
@@ -68,7 +68,7 @@ def get_tensor_map( n_blocks : int):
         mapped_to = "transformer.blocks."+str(i)+".ffn_norm"
         tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to                       # gpt2
-        tensor_map[" transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
+        tensor_map["transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
         tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to    # llama-hf
         tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to                          # llama-pth
         # Feed-forward up

From e3d1f07eb1d18e2545a5f1a2bafc86b1fd1db053 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 12:18:34 +0200
Subject: [PATCH 127/242] convert-gptneox-h5-to-gguf.py : load model in parts
 to save memory

---
 convert-gptneox-h5-to-gguf.py | 145 ++++++++++++++++++++++++----------
 1 file changed, 105 insertions(+), 40 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index ac4b2ff8860fd..a8a0c7e2df357 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -1,4 +1,4 @@
-# Quick and dirty HF gptneox--> gguf conversion
+# HF gptneox--> gguf conversion
 
 import gguf
 import gguf_tensor_map as tmap
@@ -9,7 +9,8 @@
 import numpy as np
 from typing import Any, List
 from pathlib import Path
-from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+from transformers import AutoTokenizer
 
 # ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
 def bytes_to_unicode():
@@ -33,6 +34,15 @@ def bytes_to_unicode():
     cs = [chr(n) for n in cs]
     return dict(zip(bs, cs))
 
+def count_model_parts(dir_model: str) -> int:
+    num_parts = 0
+    for filename in os.listdir(dir_model):
+        if filename.startswith("pytorch_model-"):
+            num_parts += 1
+
+    if num_parts > 0:
+        print("gguf: found " + str(num_parts) + " model parts")
+    return num_parts
 
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
@@ -70,9 +80,8 @@ def bytes_to_unicode():
     print("Model architecture not supported: " + hparams["architectures"][0] )
     sys.exit()
 
-
-model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
-list_vars = model.state_dict()
+# get number of model parts
+num_parts = count_model_parts(dir_model)
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
@@ -183,37 +192,58 @@ def bytes_to_unicode():
 # tensor info
 print("gguf: get tensor metadata")
 
-for name in list_vars.keys():
-    data = list_vars[name].squeeze().numpy()
-
-    # we don't need these
-    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-        continue
+if num_parts == 0:
+    part_names = ("pytorch_model.bin",)
+else:
+    part_names = (
+        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
+    )
 
-    # map tensor names
-    if name.endswith(".weight") and name[:-7] in tensor_map:
-        name = tensor_map[name[:-7]] + ".weight"
-    elif name.endswith(".bias") and name[:-5] in tensor_map:
-        name = tensor_map[name[:-5]] + ".bias"
-    else:
-        print( "Can not map tensor '" + name + "'" )
-        sys.exit()
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
-    n_dims = len(data.shape)
-    data_dtype = data.dtype 
+    for name in model_part.keys():
+        data = model_part[name]
 
-#    print( name + " dims " + str(n_dims) + " dtype " + str(data.dtype) )
+        # we don't need these
+        if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
+            continue
 
-    if data.dtype != np.float16 and data.dtype != np.float32:
         # convert any unsupported data types to float32
-        data_dtype = np.float32
-    elif ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data_dtype = np.float32
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data.dtype == np.float16 and n_dims == 1:
+            data_dtype = np.float32
+
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        data_dtype = np.float16
+        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data_dtype = np.float16
+
+        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
 
-    data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 
-    gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
@@ -225,24 +255,59 @@ def bytes_to_unicode():
 # tensor data
 print("gguf: convert and write tensor data")
 
-for name in list_vars.keys():
-    data = list_vars[name].squeeze().numpy()
+if num_parts == 0:
+    part_names = ("pytorch_model.bin",)
+else:
+    part_names = (
+        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
+    )
 
-    # we don't need these
-    if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-        continue
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
-    n_dims = len(data.shape)
-    data_dtype = data.dtype 
+    for name in model_part.keys():
+        data = model_part[name]
+
+        old_dtype = data.dtype
+
+        # we don't need these
+        if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
+            continue
 
-    if data_dtype != np.float16 and data_dtype != np.float32:
         # convert any unsupported data types to float32
-        data = data.astype(np.float32)
-    elif ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data = data.astype(np.float32)
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
+
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        data = data.astype(np.float16)
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+
+        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-    gguf_writer.write_tensor_to_file(data)
+        gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 

From 17800cd80fec468411481dc34a51d42a936442f1 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 12:20:02 +0200
Subject: [PATCH 128/242] convert-llama-h5-to-gguf.py : load model in parts to
 save memory

---
 convert-llama-h5-to-gguf.py | 155 +++++++++++++++++++++++++-----------
 1 file changed, 110 insertions(+), 45 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 055b6b78da9c2..98a14db9cf3c4 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -1,4 +1,4 @@
-# Quick and dirty HF llama --> gguf conversion, GQA/70b wont work
+# HF llama --> gguf conversion, GQA/70b not supported
 
 import gguf
 import gguf_tensor_map as tmap
@@ -9,7 +9,7 @@
 import numpy as np
 from typing import Any, List
 from pathlib import Path
-from transformers import AutoModelForCausalLM
+import torch
 from sentencepiece import SentencePieceProcessor
 
 
@@ -22,6 +22,15 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
                    .swapaxes(1, 2)
                    .reshape(weights.shape))
 
+def count_model_parts(dir_model: str) -> int:
+    num_parts = 0
+    for filename in os.listdir(dir_model):
+        if filename.startswith("pytorch_model-"):
+            num_parts += 1
+
+    if num_parts > 0:
+        print("gguf: found " + str(num_parts) + " model parts")
+    return num_parts
 
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
@@ -60,8 +69,8 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     print("Model architecture not supported: " + hparams["architectures"][0] )
     sys.exit()
 
-model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
-list_vars = model.state_dict()
+# get number of model parts
+num_parts = count_model_parts(dir_model)
 
 gguf_writer = gguf.GGUFWriter.open(fname_out)
 
@@ -164,41 +173,62 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 # tensor info
 print("gguf: get tensor metadata")
 
-for name in list_vars.keys():
-    data = list_vars[name].squeeze().numpy()
-
-    # we don't need these
-    if name.endswith(".rotary_emb.inv_freq"):
-        continue
 
-    # permute these
-    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute(data,head_count)
+if num_parts == 0:
+    part_names = ("pytorch_model.bin",)
+else:
+    part_names = (
+        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
+    )
 
-    # map tensor names
-    if name.endswith(".weight") and name[:-7] in tensor_map:
-        name = tensor_map[name[:-7]] + ".weight"
-    elif name.endswith(".bias") and name[:-5] in tensor_map:
-        name = tensor_map[name[:-5]] + ".bias"
-    else:
-        print( "Can not map tensor '" + name + "'" )
-        sys.exit()
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
-    n_dims = len(data.shape)
-    data_dtype = data.dtype 
+    for name in model_part.keys():
+        data = model_part[name]
 
-#    print( name + " dims " + str(n_dims) + " dtype " + str(data.dtype) )
+        # we don't need these
+        if name.endswith(".rotary_emb.inv_freq"):
+            continue
 
-    if data.dtype != np.float16 and data.dtype != np.float32:
         # convert any unsupported data types to float32
-        data_dtype = np.float32
-    elif ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # permute these
+        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+            data = permute(data,head_count)
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data_dtype = np.float32
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data_dtype = np.float32
+
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        data_dtype = np.float16
+        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data_dtype = np.float16
 
-    data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
 
-    gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
+        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
 
 
 print("gguf: write header")
@@ -211,28 +241,63 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 # tensor data
 print("gguf: convert and write tensor data")
 
-for name in list_vars.keys():
-    data = list_vars[name].squeeze().numpy()
+if num_parts == 0:
+    part_names = ("pytorch_model.bin",)
+else:
+    part_names = (
+        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
+    )
 
-    # we don't need these
-    if name.endswith(".rotary_emb.inv_freq"):
-        continue
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
-    # permute these
-    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute(data, head_count)
+    for name in model_part.keys():
+        data = model_part[name]
 
-    n_dims = len(data.shape)
-    data_dtype = data.dtype 
+        old_dtype = data.dtype
+
+        # we don't need these
+        if name.endswith(".rotary_emb.inv_freq"):
+            continue
 
-    if data_dtype != np.float16 and data_dtype != np.float32:
         # convert any unsupported data types to float32
-        data = data.astype(np.float32)
-    elif ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # permute these
+        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+            data = permute(data, head_count)
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data = data.astype(np.float32)
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
+
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        data = data.astype(np.float16)
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+
+        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-    gguf_writer.write_tensor_to_file(data)
+        gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 

From 91d4bfd536cee322c26a3d70ac41d486a3fbb7b8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sun, 13 Aug 2023 13:29:46 +0300
Subject: [PATCH 129/242] convert : write more metadata for LLaMA

---
 convert-llama-h5-to-gguf.py | 24 +++++++++++++++---------
 gguf.py                     | 23 +++++------------------
 2 files changed, 20 insertions(+), 27 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 055b6b78da9c2..9526bde842fa5 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -17,6 +17,7 @@
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
+
 def permute(weights: NDArray, n_head: int) -> NDArray:
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
                    .swapaxes(1, 2)
@@ -52,12 +53,12 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
 
 print("gguf: loading model "+last_dir)
-    
+
 with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
     hparams = json.load(f)
 
 if hparams["architectures"][0] != "LlamaForCausalLM":
-    print("Model architecture not supported: " + hparams["architectures"][0] )
+    print("Model architecture not supported: " + hparams["architectures"][0])
     sys.exit()
 
 model = AutoModelForCausalLM.from_pretrained(dir_model, low_cpu_mem_usage=True, trust_remote_code=True)
@@ -68,18 +69,23 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
 print("gguf: get model metadata")
 
-llm_arch    = "llama"
-head_count  = hparams["num_attention_heads"]
+llm_arch = "llama"
+hf_repo = hparams["_name_or_path"]
+head_count = hparams["num_attention_heads"]
+head_count_kv = hparams["num_key_value_heads"]
 block_count = hparams["num_hidden_layers"]
 
 gguf_writer.add_name(last_dir)
 gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_quantization_version(ftype)
+guff_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
 gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
 gguf_writer.add_head_count(llm_arch, head_count)
+gguf_writer.add_head_count_kv(llm_arch, head_count_kv)
 gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
 
 
@@ -173,7 +179,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 
     # permute these
     if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute(data,head_count)
+        data = permute(data, head_count)
 
     # map tensor names
     if name.endswith(".weight") and name[:-7] in tensor_map:
@@ -181,11 +187,11 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
     elif name.endswith(".bias") and name[:-5] in tensor_map:
         name = tensor_map[name[:-5]] + ".bias"
     else:
-        print( "Can not map tensor '" + name + "'" )
+        print("Can not map tensor '" + name + "'")
         sys.exit()
 
     n_dims = len(data.shape)
-    data_dtype = data.dtype 
+    data_dtype = data.dtype
 
 #    print( name + " dims " + str(n_dims) + " dtype " + str(data.dtype) )
 
@@ -223,7 +229,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
         data = permute(data, head_count)
 
     n_dims = len(data.shape)
-    data_dtype = data.dtype 
+    data_dtype = data.dtype
 
     if data_dtype != np.float16 and data_dtype != np.float32:
         # convert any unsupported data types to float32
@@ -237,5 +243,5 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
 gguf_writer.close()
 
 
-print("gguf: model successfully exported to '" + fname_out + "'" )
+print("gguf: model successfully exported to '" + fname_out + "'")
 print("")
diff --git a/gguf.py b/gguf.py
index 75ba2870afc9a..de3e5bbfbe883 100644
--- a/gguf.py
+++ b/gguf.py
@@ -12,23 +12,10 @@
 import numpy as np
 import sys
 
+
 class GGMLQuantizationType(IntEnum):
     F32 = 0
     F16 = 1
-    Q4_0 = 2
-    Q4_1 = 3
-    # Q4_2 = 4 # support has been removed
-    # Q4_3 = 5 # support has been removed
-    Q5_0 = 6
-    Q5_1 = 7
-    Q8_0 = 8
-    Q8_1 = 9
-    Q2_K = 10
-    Q3_K = 11
-    Q4_K = 12
-    Q5_K = 13
-    Q6_K = 14
-    Q8_K = 15
 
 
 class GGUFValueType(IntEnum):
@@ -143,7 +130,7 @@ def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool =
 
         if add_vtype:
             self.kv_data += struct.pack("<I", vtype)
-            self.kv_data_count += 1;
+            self.kv_data_count += 1
 
         if vtype == GGUFValueType.UINT8:
             self.kv_data += struct.pack("<B", val)
@@ -201,7 +188,7 @@ def write_tensor_to_file(self, tensor: np.ndarray):
             self.fout.write(bytes([0] * pad))
 
         tensor.tofile(self.fout)
-        
+
         pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
         if pad != 0:
             self.fout.write(bytes([0] * pad))
@@ -214,7 +201,7 @@ def close(self):
 
     def add_architecture(self, architecture: str):
         self.add_string(constants.KEY_GENERAL_ARCHITECTURE,
-                          architecture)
+                        architecture)
 
     def add_author(self, author: str):
         self.add_string(constants.KEY_GENERAL_AUTHOR, author)
@@ -311,7 +298,7 @@ def add_token_merges(self, merges: List):
 
     def add_token_scores(self, scores: List[float]):
         self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
-    
+
     def add_bos_token_id(self, id: int):
         self.add_uint32(constants.KEY_TOKENIZER_BOS_ID, id)
 

From bf2dad3100ea7ce7095c566d69391c3985a3f6a1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sun, 13 Aug 2023 14:38:53 +0300
Subject: [PATCH 130/242] convert : rm quantization version

---
 convert-llama-h5-to-gguf.py | 85 +++++++++++++++++++------------------
 1 file changed, 43 insertions(+), 42 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 53378e47c3863..574788ee0e29e 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -23,6 +23,7 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
                    .swapaxes(1, 2)
                    .reshape(weights.shape))
 
+
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
     for filename in os.listdir(dir_model):
@@ -33,6 +34,7 @@ def count_model_parts(dir_model: str) -> int:
         print("gguf: found " + str(num_parts) + " model parts")
     return num_parts
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
@@ -86,7 +88,6 @@ def count_model_parts(dir_model: str) -> int:
 
 gguf_writer.add_name(last_dir)
 gguf_writer.add_architecture(llm_arch)
-gguf_writer.add_quantization_version(ftype)
 guff_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
@@ -187,7 +188,7 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
+    print("gguf: loading model part '" + part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
@@ -205,7 +206,7 @@ def count_model_parts(dir_model: str) -> int:
 
         # permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data,head_count)
+            data = permute(data, head_count)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:
@@ -213,11 +214,11 @@ def count_model_parts(dir_model: str) -> int:
         elif name.endswith(".bias") and name[:-5] in tensor_map:
             name = tensor_map[name[:-5]] + ".bias"
         else:
-            print( "Can not map tensor '" + name + "'" )
+            print("Can not map tensor '" + name + "'")
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -254,60 +255,60 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
+    print("gguf: loading model part '" + part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
         data = model_part[name]
 
-<<<<<<< HEAD
-    n_dims = len(data.shape)
+<< << << < HEAD
+   n_dims = len(data.shape)
     data_dtype = data.dtype
-=======
-        old_dtype = data.dtype
+== == == =
+   old_dtype = data.dtype
 
-        # we don't need these
-        if name.endswith(".rotary_emb.inv_freq"):
-            continue
->>>>>>> 17800cd80fec468411481dc34a51d42a936442f1
+    # we don't need these
+    if name.endswith(".rotary_emb.inv_freq"):
+        continue
+>>>>>> > 17800cd80fec468411481dc34a51d42a936442f1
 
-        # convert any unsupported data types to float32
-        if data.dtype != torch.float16 and data.dtype != torch.float32:
-            data = data.to(torch.float32)
+   # convert any unsupported data types to float32
+   if data.dtype != torch.float16 and data.dtype != torch.float32:
+        data = data.to(torch.float32)
 
-        data = data.squeeze().numpy()
+    data = data.squeeze().numpy()
 
-        # permute these
-        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data, head_count)
+    # permute these
+    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+        data = permute(data, head_count)
 
-        # map tensor names
-        if name.endswith(".weight") and name[:-7] in tensor_map:
-            name = tensor_map[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tensor_map:
-            name = tensor_map[name[:-5]] + ".bias"
-        else:
-            print( "Can not map tensor '" + name + "'" )
-            sys.exit()
+    # map tensor names
+    if name.endswith(".weight") and name[:-7] in tensor_map:
+        name = tensor_map[name[:-7]] + ".weight"
+    elif name.endswith(".bias") and name[:-5] in tensor_map:
+        name = tensor_map[name[:-5]] + ".bias"
+    else:
+        print("Can not map tensor '" + name + "'" )
+        sys.exit()
 
-        n_dims = len(data.shape)
-        data_dtype = data.dtype 
+    n_dims = len(data.shape)
+    data_dtype = data.dtype
 
-        # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data = data.astype(np.float32)
+    # if f32 desired, convert any float16 to float32
+    if ftype == 0 and data.dtype == np.float16:
+        data = data.astype(np.float32)
 
-        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data = data.astype(np.float32)
+    # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+    if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+        data = data.astype(np.float32)
 
-        # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
+    # if f16 desired, convert any float32 2-dim weight tensors to float16
+    if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+        data = data.astype(np.float16)
 
-        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
+    print(name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-        gguf_writer.write_tensor_to_file(data)
+    gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 

From 2827b840e4235e02c32f93ca80af19cb0790d3c7 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 13:54:10 +0200
Subject: [PATCH 131/242] convert-gptneox-h5-to-gguf.py : add file_type key

---
 convert-gptneox-h5-to-gguf.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index a8a0c7e2df357..ba6e90e42ed40 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -90,8 +90,9 @@ def count_model_parts(dir_model: str) -> int:
 llm_arch    = "gptneox"
 block_count = hparams["num_hidden_layers"]
 
-gguf_writer.add_name(last_dir)
 gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_name(last_dir)
+gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)

From 6beebf3fd92aa2ab5c8b060b6d25c5e1d3d12e95 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 13 Aug 2023 14:11:01 +0200
Subject: [PATCH 132/242] gptneox-main.cpp : add file_type key

---
 gptneox-main.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 4773dfd690461..63ee5e61c22ef 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -379,6 +379,8 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         if (keyidx != -1) { fprintf(stdout, "%s: model license      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.architecture");
         if (keyidx != -1) { fprintf(stdout, "%s: model architecture = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.file_type");
+        if (keyidx != -1) { fprintf(stdout, "%s: model file type    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
     }
 
     // check required metadata

From 24f48833ab5a1fcacb15c1908e13a80b35fa0f31 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Sun, 13 Aug 2023 16:55:42 +0300
Subject: [PATCH 133/242] fix conflicts

---
 convert-llama-h5-to-gguf.py | 82 +++++++++++++++++--------------------
 1 file changed, 38 insertions(+), 44 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 574788ee0e29e..cf9f6f80246dc 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -23,7 +23,6 @@ def permute(weights: NDArray, n_head: int) -> NDArray:
                    .swapaxes(1, 2)
                    .reshape(weights.shape))
 
-
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
     for filename in os.listdir(dir_model):
@@ -34,7 +33,6 @@ def count_model_parts(dir_model: str) -> int:
         print("gguf: found " + str(num_parts) + " model parts")
     return num_parts
 
-
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
@@ -188,7 +186,7 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '" + part_name + "'")
+    print("gguf: loading model part '"+ part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
@@ -206,7 +204,7 @@ def count_model_parts(dir_model: str) -> int:
 
         # permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data, head_count)
+            data = permute(data,head_count)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:
@@ -214,11 +212,11 @@ def count_model_parts(dir_model: str) -> int:
         elif name.endswith(".bias") and name[:-5] in tensor_map:
             name = tensor_map[name[:-5]] + ".bias"
         else:
-            print("Can not map tensor '" + name + "'")
+            print( "Can not map tensor '" + name + "'" )
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype
+        data_dtype = data.dtype 
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -255,60 +253,56 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '" + part_name + "'")
+    print("gguf: loading model part '"+ part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
         data = model_part[name]
 
-<< << << < HEAD
-   n_dims = len(data.shape)
-    data_dtype = data.dtype
-== == == =
-   old_dtype = data.dtype
+    
+        old_dtype = data.dtype
 
-    # we don't need these
-    if name.endswith(".rotary_emb.inv_freq"):
-        continue
->>>>>> > 17800cd80fec468411481dc34a51d42a936442f1
+        # we don't need these
+        if name.endswith(".rotary_emb.inv_freq"):
+            continue
 
-   # convert any unsupported data types to float32
-   if data.dtype != torch.float16 and data.dtype != torch.float32:
-        data = data.to(torch.float32)
+        # convert any unsupported data types to float32
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
 
-    data = data.squeeze().numpy()
+        data = data.squeeze().numpy()
 
-    # permute these
-    if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-        data = permute(data, head_count)
+        # permute these
+        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+            data = permute(data, head_count)
 
-    # map tensor names
-    if name.endswith(".weight") and name[:-7] in tensor_map:
-        name = tensor_map[name[:-7]] + ".weight"
-    elif name.endswith(".bias") and name[:-5] in tensor_map:
-        name = tensor_map[name[:-5]] + ".bias"
-    else:
-        print("Can not map tensor '" + name + "'" )
-        sys.exit()
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
 
-    n_dims = len(data.shape)
-    data_dtype = data.dtype
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
 
-    # if f32 desired, convert any float16 to float32
-    if ftype == 0 and data.dtype == np.float16:
-        data = data.astype(np.float32)
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data = data.astype(np.float32)
 
-    # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-    if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-        data = data.astype(np.float32)
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
 
-    # if f16 desired, convert any float32 2-dim weight tensors to float16
-    if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-        data = data.astype(np.float16)
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
 
-    print(name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
+        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-    gguf_writer.write_tensor_to_file(data)
+        gguf_writer.write_tensor_to_file(data)
 
 gguf_writer.close()
 

From 196b50fee74625528fae91e7125dc1e9cc07cecd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Mon, 14 Aug 2023 08:50:47 +0300
Subject: [PATCH 134/242] gguf : add todos and comments

---
 gguf-llama.cpp | 64 ++++++++++++++++++++++++++++----------------------
 1 file changed, 36 insertions(+), 28 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 700d6009bcddb..826af549b5459 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -258,6 +258,11 @@ struct llama_kv_cache {
 };
 
 struct llama_vocab {
+    // TODO: convert to this gguf_vocab
+    // add a vector of merges
+    // add members for bos/eos/pad/sep tokens
+    // so that we can pass it to different types of tokenizers with a common interface
+
     using id    = int32_t;
     using token = std::string;
 
@@ -447,7 +452,7 @@ static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml
     return size / ggml_blck_size(type);
 }
 
-struct llama_load_tensor {
+struct gguf_load_tensor {
     std::string name;
     enum ggml_type type = GGML_TYPE_F32;
     std::vector<uint32_t> ne;
@@ -457,9 +462,9 @@ struct llama_load_tensor {
     uint8_t * data;
 };
 
-struct llama_load_tensors_map {
+struct gguf_load_tensors_map {
     // tensors is kept in a separate vector to preserve file order
-    std::vector<llama_load_tensor> tensors;
+    std::vector<gguf_load_tensor> tensors;
     std::unordered_map<std::string, size_t> name_to_idx;
 };
 
@@ -477,7 +482,7 @@ struct gguf_file_loader {
     llama_vocab vocab;
 struct ggml_context * ctx_data = NULL;
 
-    gguf_file_loader(const char * fname, llama_load_tensors_map & tensors_map)
+    gguf_file_loader(const char * fname, gguf_load_tensors_map & tensors_map)
         : file(fname, "rb") {
         fprintf(stderr, "llama.cpp: loading model from %s\n", fname);
         
@@ -523,13 +528,9 @@ struct ggml_context * ctx_data = NULL;
 
     void read_hparams() {
 
-        // TODO make keysconstants in header
+        // TODO define keys as constants in header
         // TODO: read all hparams from file
-        int q_ver_idx = gguf_find_key (gguf_ctx, "general.quantization_version");
-        if (q_ver_idx != -1) {
-            hparams.ftype = (enum llama_ftype) gguf_get_val_u32(gguf_ctx, q_ver_idx);
-        }
-
+            
         hparams.n_vocab = read_n_vocab();
         hparams.n_ctx   = read_u32("llama.context_length");
         hparams.n_embd  = read_u32("llama.embedding_length");
@@ -567,11 +568,11 @@ struct ggml_context * ctx_data = NULL;
         }
     }
 
-    void read_tensor_metadata(llama_load_tensors_map & tensors_map) {
+    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) {
         const int n_tensors = gguf_get_n_tensors(gguf_ctx);
 
         for (int i = 0; i < n_tensors; ++i) {
-            llama_load_tensor tensor;
+            gguf_load_tensor tensor;
             const char * name = gguf_get_tensor_name(gguf_ctx, i);
 
             struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
@@ -617,6 +618,13 @@ struct ggml_context * ctx_data = NULL;
 };
 
 struct gguf_file_saver {
+    // TODO
+    // this implementation now assumes that the data section is of the same length as the unquantized model.
+    // this is needed to write tensor metadata and weights in a single pass by seeking to appropriate positions in the file.
+    // this may not be true when we add quantization version and change ftype description (currently it's string according to the specs,
+    // but better to have it as uint32).
+    // we need to calculate the delta in number of bytes written with a counter as a struct member.
+    
     gguf_file file;
     gguf_file_loader * fl;
     size_t info_offset;
@@ -747,7 +755,7 @@ struct gguf_file_saver {
         GGML_ASSERT(info_offset == file.tell());
     }
 
-    size_t write_tensor_info(llama_load_tensor & tensor, enum ggml_type type) {
+    size_t write_tensor_info(gguf_load_tensor & tensor, enum ggml_type type) {
         size_t total_written = 0;
         file.seek(info_offset, SEEK_SET);
         GGML_ASSERT(info_offset == file.tell());
@@ -761,14 +769,14 @@ struct gguf_file_saver {
 
         total_written += file.write_i32(type);
         total_written += file.write_u64(tensor_offset);
-        info_offset   += total_written;
+        info_offset   += total_written; // position to write info of the next tensor
 
         file.seek(0, SEEK_END);
         
         return total_written;
     }
 
-    void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
+    void write_tensor(gguf_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
         switch (new_type) {
             case GGML_TYPE_F32:
             case GGML_TYPE_F16:
@@ -791,13 +799,13 @@ struct gguf_file_saver {
         size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
         size_t pad = padded_size - new_size;
         file.write_zeros(pad);
-        tensor_offset += padded_size;
+        tensor_offset += padded_size; // offset of the next tensor
     }
 };
 
 struct llama_model_loader {
     std::unique_ptr<gguf_file_loader> file_loader;
-    llama_load_tensors_map tensors_map;
+    gguf_load_tensors_map tensors_map;
     bool use_mmap;
     size_t num_ggml_tensors_created = 0;
     struct ggml_context * ggml_ctx = NULL;
@@ -813,7 +821,7 @@ struct llama_model_loader {
 
     void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
         *ctx_size_p = *mmapped_size_p = 0;
-        for (const llama_load_tensor & lt : tensors_map.tensors) {
+        for (const gguf_load_tensor & lt : tensors_map.tensors) {
             *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
             *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
         }
@@ -824,7 +832,7 @@ struct llama_model_loader {
         if (it == tensors_map.name_to_idx.end()) {
             throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
         }
-        llama_load_tensor & lt = tensors_map.tensors.at(it->second);
+        gguf_load_tensor & lt = tensors_map.tensors.at(it->second);
         if (lt.ne != ne) {
             throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
                          name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
@@ -833,7 +841,7 @@ struct llama_model_loader {
         return get_tensor_for(lt, backend);
     }
 
-    struct ggml_tensor * get_tensor_for(llama_load_tensor & lt, ggml_backend backend) {
+    struct ggml_tensor * get_tensor_for(gguf_load_tensor & lt, ggml_backend backend) {
         struct ggml_tensor * tensor;
         if (backend != GGML_BACKEND_CPU) {
             ggml_set_no_alloc(ggml_ctx, true);
@@ -866,7 +874,7 @@ struct llama_model_loader {
         size_t data_size = 0;
         size_t prefetch_size = 0;
         size_t lock_size = 0;
-        for (const llama_load_tensor & lt : tensors_map.tensors) {
+        for (const gguf_load_tensor & lt : tensors_map.tensors) {
             data_size += lt.size;
             if (lt.ggml_tensor->backend == GGML_BACKEND_CPU) {
                 prefetch_size += lt.size;
@@ -881,7 +889,7 @@ struct llama_model_loader {
         }
 
         size_t done_size = 0;
-        for (llama_load_tensor & lt : tensors_map.tensors) {
+        for (gguf_load_tensor & lt : tensors_map.tensors) {
             if (progress_callback) {
                 progress_callback((float) done_size / data_size, progress_callback_user_data);
             }
@@ -928,7 +936,7 @@ struct llama_model_loader {
         }
     }
 
-    void load_data_for(llama_load_tensor & lt) {
+    void load_data_for(gguf_load_tensor & lt) {
         if (use_mmap) {
             lt.data = (uint8_t *) mapping->addr + lt.file_off;
         } else {
@@ -942,7 +950,7 @@ struct llama_model_loader {
         }
     }
 
-    static void print_checksum(llama_load_tensor & lt) {
+    static void print_checksum(gguf_load_tensor & lt) {
         uint32_t sum = 0;
         for (size_t i = 0; i < lt.size; i++) {
             uint8_t byte = lt.data[i];
@@ -1421,7 +1429,7 @@ static void llama_model_load_internal(
     }
 
     // populate `tensors_by_name`
-    for (llama_load_tensor & lt : ml->tensors_map.tensors) {
+    for (gguf_load_tensor & lt : ml->tensors_map.tensors) {
         model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
     }
 
@@ -2896,7 +2904,7 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
 // quantization
 //
 
-static void llama_convert_tensor_internal(const llama_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
+static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
     if (output.size < nelements * sizeof(float)) {
         output.resize(nelements * sizeof(float));
     }
@@ -3018,7 +3026,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     };
 
     size_t idx = 0;
-    for (llama_load_tensor & tensor : model_loader->tensors_map.tensors) {
+    for (gguf_load_tensor & tensor : model_loader->tensors_map.tensors) {
         gguf_buffer read_data;
         read_data.resize(tensor.size);
         tensor.data = read_data.addr;
@@ -3570,7 +3578,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
                     return 1;
                 }
                 size_t idx = model_loader->tensors_map.name_to_idx[base_name];
-                llama_load_tensor & lt = model_loader->tensors_map.tensors[idx];
+                gguf_load_tensor & lt = model_loader->tensors_map.tensors[idx];
                 base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
                 lt.data = (uint8_t *) lt.ggml_tensor->data;
                 model_loader->load_data_for(lt);

From 5d22a9db133d0b5737f8889d488070a55dbe7682 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 10:55:44 +0200
Subject: [PATCH 135/242] convert-gptneox-h5-to-gguf.py : tensor name map
 changes

---
 convert-gptneox-h5-to-gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index ba6e90e42ed40..a3d78c9ef2aa0 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -1,7 +1,7 @@
 # HF gptneox--> gguf conversion
 
 import gguf
-import gguf_tensor_map as tmap
+import gguf_namemap as tmap
 import os
 import sys
 import struct
@@ -188,7 +188,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = tmap.get_tensor_map(block_count)
+tensor_map = tmap.get_tensor_namemap(block_count)
 
 # tensor info
 print("gguf: get tensor metadata")

From 51939d7d1bd6b752d66faf6050a9ab2d80915ba3 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 10:56:59 +0200
Subject: [PATCH 136/242] Create gguf_namemap.py : tensor name map changes

---
 gguf_namemap.py | 95 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 95 insertions(+)
 create mode 100644 gguf_namemap.py

diff --git a/gguf_namemap.py b/gguf_namemap.py
new file mode 100644
index 0000000000000..7546630ed26c8
--- /dev/null
+++ b/gguf_namemap.py
@@ -0,0 +1,95 @@
+# Recommended mapping of model tensor names for storage in gguf
+
+def get_tensor_namemap( n_blocks : int):
+    tensor_map = {}
+    # Token embeddings
+    mapped_to = "token_embd"
+    tensor_map["gpt_neox.embed_in"] = mapped_to           # gptneox
+    tensor_map["transformer.wte"] = mapped_to             # gpt2 mpt
+    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
+    tensor_map["model.embed_tokens"] = mapped_to          # llama-hf
+    tensor_map["tok_embeddings"] = mapped_to              # llama-pth
+    # Position embeddings
+    mapped_to = "pos_embd"
+    tensor_map["transformer.wpe"] = mapped_to # gpt2
+    # Output norm
+    mapped_to = "output_norm"
+    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
+    tensor_map["transformer.ln_f"] = mapped_to          # gpt2 falcon
+    tensor_map["transformer.norm_f"] = mapped_to        # mpt
+    tensor_map["model.norm"] = mapped_to                # llama-hf
+    tensor_map["norm"] = mapped_to                      # llama-pth
+    # Output
+    mapped_to = "output"
+    tensor_map["embed_out"] = mapped_to # gptneox
+    tensor_map["lm_head"] = mapped_to   # gpt2 mpt falcon llama-hf
+    tensor_map["output"] = mapped_to    # llama-pth
+    # Attention and fee-forward layer blocks
+    for i in range(0,n_blocks):
+        # Attention norm
+        mapped_to = "blk."+str(i)+".attn_norm"
+        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to              # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to       # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to   # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to           # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to    # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to           # llama-pth
+        # Attention norm 2
+        mapped_to = "blk."+str(i)+".attn_norm_2"
+        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
+        # Attention query-key-value
+        mapped_to = "blk."+str(i)+".attn_qkv"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"] = mapped_to     # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_attn"] = mapped_to                     # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"] = mapped_to                  # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to  # falcon
+        # Attention query
+        mapped_to = "blk."+str(i)+".attn_q"
+        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wq"] = mapped_to           # llama-pth
+        # Attention key
+        mapped_to = "blk."+str(i)+".attn_k"
+        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wk"] = mapped_to           # llama-pth
+        # Attention value
+        mapped_to = "blk."+str(i)+".attn_v"
+        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wv"] = mapped_to           # llama-pth
+        # Attention output
+        mapped_to = "blk."+str(i)+".attn_output"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"] = mapped_to    # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_proj"] = mapped_to          # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"] = mapped_to   # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".self_attn.o_proj"] = mapped_to      # llama-hf
+        tensor_map["layers."+str(i)+".attention.wo"] = mapped_to                # llama-pth
+        # Feed-forward norm
+        mapped_to = "blk."+str(i)+".ffn_norm"
+        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to                       # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
+        tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to    # llama-hf
+        tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to                          # llama-pth
+        # Feed-forward up
+        mapped_to = "blk."+str(i)+".ffn_up"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_fc"] = mapped_to            # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"] = mapped_to    # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"] = mapped_to   # falcon
+        tensor_map["model.layers."+str(i)+".mlp.up_proj"] = mapped_to          # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w3"] = mapped_to            # llama-pth
+        # Feed-forward gate
+        mapped_to = "blk."+str(i)+".ffn_gate"
+        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w1"] = mapped_to     # llama-pth
+        # Feed-forward down
+        mapped_to = "blk."+str(i)+".ffn_down"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_proj"] = mapped_to          # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"] = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"] = mapped_to   # falcon
+        tensor_map["model.layers."+str(i)+".mlp.down_proj"] = mapped_to        # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w2"] = mapped_to            # llama-pth
+
+    return tensor_map

From 806a15749d6ba4793c38295f1bd1128158c8e244 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 10:57:19 +0200
Subject: [PATCH 137/242] Delete gguf_tensor_map.py

---
 gguf_tensor_map.py | 96 ----------------------------------------------
 1 file changed, 96 deletions(-)
 delete mode 100644 gguf_tensor_map.py

diff --git a/gguf_tensor_map.py b/gguf_tensor_map.py
deleted file mode 100644
index d73788bb46877..0000000000000
--- a/gguf_tensor_map.py
+++ /dev/null
@@ -1,96 +0,0 @@
-# Recommended mapping of model tensor names for storage in gguf
-
-def get_tensor_map( n_blocks : int):
-    tensor_map = {}
-    # Token embeddings
-    mapped_to = "transformer.token_embd"
-    tensor_map["gpt_neox.embed_in"] = mapped_to           # gptneox
-    tensor_map["transformer.wte"] = mapped_to             # gpt2 mpt
-    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
-    tensor_map["model.embed_tokens"] = mapped_to          # llama-hf
-    tensor_map["tok_embeddings"] = mapped_to              # llama-pth
-    # Position embeddings
-    mapped_to = "transformer.pos_embd"
-    tensor_map["transformer.wpe"] = mapped_to # gpt2
-    # Output norm
-    mapped_to = "transformer.output_norm"
-    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
-    tensor_map["transformer.ln_f"] = mapped_to          # gpt2 falcon
-    tensor_map["transformer.norm_f"] = mapped_to        # mpt
-    tensor_map["model.norm"] = mapped_to                # llama-hf
-    tensor_map["norm"] = mapped_to                      # llama-pth
-    # Output
-    mapped_to = "transformer.output"
-    tensor_map["embed_out"] = mapped_to # gptneox
-    tensor_map["lm_head"] = mapped_to   # gpt2 mpt falcon llama-hf
-    tensor_map["output"] = mapped_to    # llama-pth
-    # Attention and fee-forward layer blocks
-    for i in range(0,n_blocks):
-        # Attention norm
-        mapped_to = "transformer.blocks."+str(i)+".attn_norm"
-        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to              # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to       # mpt
-        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to   # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to           # falcon40b
-        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to    # llama-hf
-        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to           # llama-pth
-        # Attention norm 2
-        mapped_to = "transformer.blocks."+str(i)+".attn_norm_2"
-        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
-        # Attention query-key-value
-        mapped_to = "transformer.blocks."+str(i)+".attn_qkv"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"] = mapped_to     # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_attn"] = mapped_to                     # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"] = mapped_to                  # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to  # falcon
-        # Attention query
-        mapped_to = "transformer.blocks."+str(i)+".attn_q"
-        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wq"] = mapped_to           # llama-pth
-        # Attention key
-        mapped_to = "transformer.blocks."+str(i)+".attn_k"
-        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wk"] = mapped_to           # llama-pth
-        # Attention value
-        mapped_to = "transformer.blocks."+str(i)+".attn_v"
-        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wv"] = mapped_to           # llama-pth
-        # Attention output
-        mapped_to = "transformer.blocks."+str(i)+".attn_output"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"] = mapped_to    # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_proj"] = mapped_to          # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"] = mapped_to   # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".self_attn.o_proj"] = mapped_to      # llama-hf
-        tensor_map["layers."+str(i)+".attention.wo"] = mapped_to                # llama-pth
-        # Feed-forward norm
-        mapped_to = "transformer.blocks."+str(i)+".ffn_norm"
-        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to                       # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
-        tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to    # llama-hf
-        tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to                          # llama-pth
-        # Feed-forward up
-        mapped_to = "transformer.blocks."+str(i)+".ffn_up"
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_fc"] = mapped_to            # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"] = mapped_to    # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"] = mapped_to   # falcon
-        tensor_map["model.layers."+str(i)+".mlp.up_proj"] = mapped_to          # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w3"] = mapped_to            # llama-pth
-        # Feed-forward gate
-        mapped_to = "transformer.blocks."+str(i)+".ffn_gate"
-        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w1"] = mapped_to     # llama-pth
-        # Feed-forward down
-        mapped_to = "transformer.blocks."+str(i)+".ffn_down"
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_proj"] = mapped_to          # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"] = mapped_to  # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"] = mapped_to   # falcon
-        tensor_map["model.layers."+str(i)+".mlp.down_proj"] = mapped_to        # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w2"] = mapped_to            # llama-pth
-
-    return tensor_map
-

From d753dfbcc89cc0c7fdc54ef2f34e1aeccbbe7170 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 10:59:18 +0200
Subject: [PATCH 138/242] gptneox-main.cpp : tensor name map changes

---
 gptneox-main.cpp | 32 +++++++++++++++++---------------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index 63ee5e61c22ef..f336ee88bfaf9 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -370,17 +370,19 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         int keyidx;
 
         keyidx = gguf_find_key(ggufctx, "general.name");
-        if (keyidx != -1) { fprintf(stdout, "%s: model name         = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model name           = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.description");
-        if (keyidx != -1) { fprintf(stdout, "%s: model description  = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model description    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.author");
-        if (keyidx != -1) { fprintf(stdout, "%s: model author       = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model author         = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.license");
-        if (keyidx != -1) { fprintf(stdout, "%s: model license      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model license        = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.architecture");
-        if (keyidx != -1) { fprintf(stdout, "%s: model architecture = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model architecture   = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.file_type");
-        if (keyidx != -1) { fprintf(stdout, "%s: model file type    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        if (keyidx != -1) { fprintf(stdout, "%s: model file type      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.source.hugginface.repository");
+        if (keyidx != -1) { fprintf(stdout, "%s: model source HF repo = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
     }
 
     // check required metadata
@@ -551,21 +553,21 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
 
         model.blocks.resize(n_block);
 
-        model.wte    = ggml_get_tensor(ctx, "transformer.token_embd.weight");
-        model.ln_f_g = ggml_get_tensor(ctx, "transformer.output_norm.weight");
-        model.ln_f_b = ggml_get_tensor(ctx, "transformer.output_norm.bias");
-        model.lmh_g  = ggml_get_tensor(ctx, "transformer.output.weight");
+        model.wte    = ggml_get_tensor(ctx, "token_embd.weight");
+        model.ln_f_g = ggml_get_tensor(ctx, "output_norm.weight");
+        model.ln_f_b = ggml_get_tensor(ctx, "output_norm.bias");
+        model.lmh_g  = ggml_get_tensor(ctx, "output.weight");
 
         // map by name
-        model.tensors["transformer.token_embd.weight"] = model.wte;
-        model.tensors["transformer.output_norm.weight"] = model.ln_f_g;
-        model.tensors["transformer.output_norm.bias"]   = model.ln_f_b;
-        model.tensors["transformer.output.weight"] = model.lmh_g;
+        model.tensors["token_embd.weight"] = model.wte;
+        model.tensors["output_norm.weight"] = model.ln_f_g;
+        model.tensors["output_norm.bias"]   = model.ln_f_b;
+        model.tensors["output.weight"] = model.lmh_g;
 
         for (int i = 0; i < n_block; ++i) {
             auto & block = model.blocks[i];
 
-            std::string blocknamestart = "transformer.blocks." + std::to_string(i) + ".";
+            std::string blocknamestart = "blk." + std::to_string(i) + ".";
 
             block.ln_1_g          = get_tensor_ex(ctx, blocknamestart + "attn_norm.weight" );
             block.ln_1_b          = get_tensor_ex(ctx, blocknamestart + "attn_norm.bias" );

From a7d226f87186c6ceed2cbbc469dd24f86471e53a Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 11:14:24 +0200
Subject: [PATCH 139/242] convert-llama-h5-to-gguf.py : fixes

---
 convert-llama-h5-to-gguf.py | 23 ++++++++++++++++-------
 1 file changed, 16 insertions(+), 7 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index cf9f6f80246dc..402fdb68b1e77 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -1,7 +1,7 @@
 # HF llama --> gguf conversion, GQA/70b not supported
 
 import gguf
-import gguf_tensor_map as tmap
+import gguf_namemap as tmap
 import os
 import sys
 import struct
@@ -79,14 +79,23 @@ def count_model_parts(dir_model: str) -> int:
 print("gguf: get model metadata")
 
 llm_arch = "llama"
-hf_repo = hparams["_name_or_path"]
-head_count = hparams["num_attention_heads"]
-head_count_kv = hparams["num_key_value_heads"]
 block_count = hparams["num_hidden_layers"]
+head_count = hparams["num_attention_heads"]
+
+if "num_key_value_heads" in hparams:
+    head_count_kv = hparams["num_key_value_heads"]
+else:
+    head_count_kv = head_count
+
+if "_name_or_path" in hparams:
+    hf_repo = hparams["_name_or_path"]
+else:
+    hf_repo=""
 
-gguf_writer.add_name(last_dir)
 gguf_writer.add_architecture(llm_arch)
-guff_writer.add_source_hf_repo(hf_repo)
+gguf_writer.add_name(last_dir)
+gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
+gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)
@@ -173,7 +182,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = tmap.get_tensor_map(block_count)
+tensor_map = tmap.get_tensor_namemap(block_count)
 
 # tensor info
 print("gguf: get tensor metadata")

From 5c5a95ba2dab6cdece5c5640056e3ed35f14074b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 11:22:06 +0200
Subject: [PATCH 140/242] gguf.py : dont add empty strings

---
 gguf.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gguf.py b/gguf.py
index de3e5bbfbe883..18f42267aa78d 100644
--- a/gguf.py
+++ b/gguf.py
@@ -114,6 +114,7 @@ def add_bool(self, key: str, val: bool):
         self.add_val(val, GGUFValueType.BOOL)
 
     def add_string(self, key: str, val: str):
+        if len(val) == 0: return
         self.add_key(key)
         self.add_val(val, GGUFValueType.STRING)
 

From 0c19ae70d5a10e563b43a4cbd7112a2df74f2246 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 12:56:48 +0300
Subject: [PATCH 141/242] simple : minor style changes

---
 convert-llama-h5-to-gguf.py         |  10 +-
 examples/gguf/gguf-llama-simple.cpp | 138 ++++++++-------------------
 examples/simple/simple.cpp          | 141 +++++++++-------------------
 gguf-util.h                         |   3 +-
 4 files changed, 91 insertions(+), 201 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 402fdb68b1e77..0bce659e697df 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -2,17 +2,18 @@
 
 import gguf
 import gguf_namemap as tmap
+
 import os
 import sys
 import struct
 import json
 import numpy as np
+import torch
+
 from typing import Any, List
 from pathlib import Path
-import torch
 from sentencepiece import SentencePieceProcessor
 
-
 #NDArray = np.ndarray[Any, Any]
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
@@ -225,7 +226,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -268,7 +269,6 @@ def count_model_parts(dir_model: str) -> int:
     for name in model_part.keys():
         data = model_part[name]
 
-    
         old_dtype = data.dtype
 
         # we don't need these
@@ -295,7 +295,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
diff --git a/examples/gguf/gguf-llama-simple.cpp b/examples/gguf/gguf-llama-simple.cpp
index 35c3c818349be..fa8fc6fef9f41 100644
--- a/examples/gguf/gguf-llama-simple.cpp
+++ b/examples/gguf/gguf-llama-simple.cpp
@@ -6,177 +6,121 @@
 #include "gguf-llama.h"
 #include "build-info.h"
 
-#include <cassert>
-#include <cinttypes>
 #include <cmath>
 #include <cstdio>
-#include <cstring>
-#include <ctime>
-#include <fstream>
-#include <iostream>
 #include <string>
 #include <vector>
 
-#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
-#include <signal.h>
-#include <unistd.h>
-#elif defined (_WIN32)
-#define WIN32_LEAN_AND_MEAN
-#define NOMINMAX
-#include <windows.h>
-#include <signal.h>
-#endif
-
-
-
-int main(int argc, char ** argv)
-{
+int main(int argc, char ** argv) {
     gpt_params params;
 
-    //---------------------------------
-    // Print help :
-    //---------------------------------
-
-    if ( argc == 1 || argv[1][0] == '-' )
-    {
-        printf( "usage: %s MODEL_PATH [PROMPT]\n" , argv[0] );
+    if (argc == 1 || argv[1][0] == '-') {
+        printf("usage: %s MODEL_PATH [PROMPT]\n" , argv[0]);
         return 1 ;
     }
 
-    //---------------------------------
-    // Load parameters :
-    //---------------------------------
-
-    if ( argc >= 2 )
-    {
+    if (argc >= 2) {
         params.model = argv[1];
     }
 
-    if ( argc >= 3 )
-    {
+    if (argc >= 3) {
         params.prompt = argv[2];
     }
 
-    if ( params.prompt.empty() )
-    {
+    if (params.prompt.empty()) {
         params.prompt = "Hello my name is";
     }
 
-    //---------------------------------
-    // Init LLM :
-    //---------------------------------
+    // init LLM
 
     llama_backend_init(params.numa);
 
     llama_context_params ctx_params = llama_context_default_params();
 
     llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
-    
-    if ( model == NULL )
-    {
-        fprintf( stderr , "%s: error: unable to load model\n" , __func__ );
+
+    if (model == NULL) {
+        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
         return 1;
     }
 
     llama_context * ctx = llama_new_context_with_model(model, ctx_params);
 
-    //---------------------------------
-    // Tokenize the prompt :
-    //---------------------------------
+    // tokenize the prompt
 
     std::vector<llama_token> tokens_list;
-    tokens_list = ::llama_tokenize( ctx , params.prompt , true );
+    tokens_list = ::llama_tokenize(ctx, params.prompt, true);
 
-    const int max_context_size     = llama_n_ctx( ctx );
-    const int max_tokens_list_size = max_context_size - 4 ;
+    const int max_context_size     = llama_n_ctx(ctx);
+    const int max_tokens_list_size = max_context_size - 4;
 
-    if ( (int)tokens_list.size() > max_tokens_list_size )
-    {
-        fprintf( stderr , "%s: error: prompt too long (%d tokens, max %d)\n" ,
-             __func__ , (int)tokens_list.size() , max_tokens_list_size );
+    if ((int)tokens_list.size() > max_tokens_list_size) {
+        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
         return 1;
     }
 
-    fprintf( stderr, "\n\n" );
-
-    // Print the tokens from the prompt :
+    fprintf(stderr, "\n\n");
 
-    for( auto id : tokens_list )
-    {
-        printf( "%s" , llama_token_to_str( ctx , id ) );
+    for (auto id : tokens_list) {
+        fprintf(stderr, "%s", llama_token_to_str(ctx, id));
     }
 
-    fflush(stdout);
-
+    fflush(stderr);
 
-    //---------------------------------
-    // Main prediction loop :
-    //---------------------------------
+    // main loop
 
     // The LLM keeps a contextual cache memory of previous token evaluation.
     // Usually, once this cache is full, it is required to recompute a compressed context based on previous
     // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
     // example, we will just stop the loop once this cache is full or once an end of stream is detected.
 
-    while ( llama_get_kv_cache_token_count( ctx ) < max_context_size )
-    {
-        //---------------------------------
-        // Evaluate the tokens :
-        //---------------------------------
+    while (llama_get_kv_cache_token_count(ctx) < max_context_size) {
+        // evaluate the transformer
 
-        if ( llama_eval( ctx , tokens_list.data() , int(tokens_list.size()) , llama_get_kv_cache_token_count( ctx ) , params.n_threads ) )
-        {
-            fprintf( stderr,  "%s : failed to eval\n" , __func__ );
+        if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
+            fprintf(stderr, "%s : failed to eval\n", __func__);
             return 1;
         }
 
         tokens_list.clear();
 
-        //---------------------------------
-        // Select the best prediction :
-        //---------------------------------
+        // sample the next token
 
         llama_token new_token_id = 0;
 
-        auto logits  = llama_get_logits( ctx );
-        auto n_vocab = llama_n_vocab( ctx ); // the size of the LLM vocabulary (in tokens)
+        auto logits  = llama_get_logits(ctx);
+        auto n_vocab = llama_n_vocab(ctx);
 
         std::vector<llama_token_data> candidates;
-        candidates.reserve( n_vocab );
+        candidates.reserve(n_vocab);
 
-        for( llama_token token_id = 0 ; token_id < n_vocab ; token_id++ )
-        {
-            candidates.emplace_back( llama_token_data{ token_id , logits[ token_id ] , 0.0f } );
+        for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
+            candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
         }
 
         llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
 
-        // Select it using the "Greedy sampling" method :
-        new_token_id = llama_sample_token_greedy( ctx , &candidates_p );
-
+        new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
 
         // is it an end of stream ?
-        if ( new_token_id == llama_token_eos() )
-        {
+        if (new_token_id == llama_token_eos()) {
             fprintf(stderr, " [end of text]\n");
             break;
         }
 
-        // Print the new token :
-        printf( "%s" , llama_token_to_str( ctx , new_token_id ) );
-        fflush( stdout );
+        // print the new token :
+        printf("%s", llama_token_to_str(ctx, new_token_id));
+        fflush(stdout);
 
-        // Push this new token for next evaluation :
-        tokens_list.push_back( new_token_id );
+        // push this new token for next evaluation
+        tokens_list.push_back(new_token_id);
 
-    } // wend of main loop
+    }
 
-    llama_free( ctx );
-    llama_free_model( model );
+    llama_free(ctx);
+    llama_free_model(model);
 
     llama_backend_free();
 
     return 0;
 }
-
-// EOF
diff --git a/examples/simple/simple.cpp b/examples/simple/simple.cpp
index 97137a6584aa3..55cce10442d58 100644
--- a/examples/simple/simple.cpp
+++ b/examples/simple/simple.cpp
@@ -2,180 +2,125 @@
 #define _GNU_SOURCE
 #endif
 
+#include "build-info.h"
+
 #include "common.h"
 #include "llama.h"
-#include "build-info.h"
 
-#include <cassert>
-#include <cinttypes>
 #include <cmath>
 #include <cstdio>
-#include <cstring>
-#include <ctime>
-#include <fstream>
-#include <iostream>
 #include <string>
 #include <vector>
 
-#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
-#include <signal.h>
-#include <unistd.h>
-#elif defined (_WIN32)
-#define WIN32_LEAN_AND_MEAN
-#define NOMINMAX
-#include <windows.h>
-#include <signal.h>
-#endif
-
-
-
-int main(int argc, char ** argv)
-{
+int main(int argc, char ** argv) {
     gpt_params params;
 
-    //---------------------------------
-    // Print help :
-    //---------------------------------
-
-    if ( argc == 1 || argv[1][0] == '-' )
-    {
-        printf( "usage: %s MODEL_PATH [PROMPT]\n" , argv[0] );
+    if (argc == 1 || argv[1][0] == '-') {
+        printf("usage: %s MODEL_PATH [PROMPT]\n" , argv[0]);
         return 1 ;
     }
 
-    //---------------------------------
-    // Load parameters :
-    //---------------------------------
-
-    if ( argc >= 2 )
-    {
+    if (argc >= 2) {
         params.model = argv[1];
     }
 
-    if ( argc >= 3 )
-    {
+    if (argc >= 3) {
         params.prompt = argv[2];
     }
 
-    if ( params.prompt.empty() )
-    {
+    if (params.prompt.empty()) {
         params.prompt = "Hello my name is";
     }
 
-    //---------------------------------
-    // Init LLM :
-    //---------------------------------
+    // init LLM
 
     llama_backend_init(params.numa);
 
     llama_model * model;
     llama_context * ctx;
 
-    std::tie(model, ctx) = llama_init_from_gpt_params( params );
+    std::tie(model, ctx) = llama_init_from_gpt_params(params);
 
-    if ( model == NULL )
-    {
-        fprintf( stderr , "%s: error: unable to load model\n" , __func__ );
+    if (model == NULL) {
+        fprintf(stderr, "%s: error: unable to load model\n", __func__);
         return 1;
     }
 
-    //---------------------------------
-    // Tokenize the prompt :
-    //---------------------------------
+    // tokenize the prompt
 
     std::vector<llama_token> tokens_list;
-    tokens_list = ::llama_tokenize( ctx , params.prompt , true );
+    tokens_list = ::llama_tokenize(ctx, params.prompt, true);
 
-    const int max_context_size     = llama_n_ctx( ctx );
-    const int max_tokens_list_size = max_context_size - 4 ;
+    const int max_context_size     = llama_n_ctx(ctx);
+    const int max_tokens_list_size = max_context_size - 4;
 
-    if ( (int)tokens_list.size() > max_tokens_list_size )
-    {
-        fprintf( stderr , "%s: error: prompt too long (%d tokens, max %d)\n" ,
-             __func__ , (int)tokens_list.size() , max_tokens_list_size );
+    if ((int)tokens_list.size() > max_tokens_list_size) {
+        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
         return 1;
     }
 
-    fprintf( stderr, "\n\n" );
+    fprintf(stderr, "\n\n");
 
-    // Print the tokens from the prompt :
-
-    for( auto id : tokens_list )
-    {
-        printf( "%s" , llama_token_to_str( ctx , id ) );
+    for (auto id : tokens_list) {
+        fprintf(stderr, "%s", llama_token_to_str(ctx, id));
     }
 
-    fflush(stdout);
-
+    fflush(stderr);
 
-    //---------------------------------
-    // Main prediction loop :
-    //---------------------------------
+    // main loop
 
     // The LLM keeps a contextual cache memory of previous token evaluation.
     // Usually, once this cache is full, it is required to recompute a compressed context based on previous
     // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
     // example, we will just stop the loop once this cache is full or once an end of stream is detected.
 
-    while ( llama_get_kv_cache_token_count( ctx ) < max_context_size )
-    {
-        //---------------------------------
-        // Evaluate the tokens :
-        //---------------------------------
+    while (llama_get_kv_cache_token_count( ctx ) < max_context_size) {
+        // evaluate the transformer
 
-        if ( llama_eval( ctx , tokens_list.data() , int(tokens_list.size()) , llama_get_kv_cache_token_count( ctx ) , params.n_threads ) )
-        {
-            fprintf( stderr,  "%s : failed to eval\n" , __func__ );
+        if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
+            fprintf(stderr, "%s : failed to eval\n", __func__);
             return 1;
         }
 
         tokens_list.clear();
 
-        //---------------------------------
-        // Select the best prediction :
-        //---------------------------------
+        // sample the next token
 
         llama_token new_token_id = 0;
 
-        auto logits  = llama_get_logits( ctx );
-        auto n_vocab = llama_n_vocab( ctx ); // the size of the LLM vocabulary (in tokens)
+        auto logits  = llama_get_logits(ctx);
+        auto n_vocab = llama_n_vocab(ctx);
 
         std::vector<llama_token_data> candidates;
-        candidates.reserve( n_vocab );
+        candidates.reserve(n_vocab);
 
-        for( llama_token token_id = 0 ; token_id < n_vocab ; token_id++ )
-        {
-            candidates.emplace_back( llama_token_data{ token_id , logits[ token_id ] , 0.0f } );
+        for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
+            candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
         }
 
         llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
 
-        // Select it using the "Greedy sampling" method :
-        new_token_id = llama_sample_token_greedy( ctx , &candidates_p );
-
+        new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
 
         // is it an end of stream ?
-        if ( new_token_id == llama_token_eos() )
-        {
+        if (new_token_id == llama_token_eos()) {
             fprintf(stderr, " [end of text]\n");
             break;
         }
 
-        // Print the new token :
-        printf( "%s" , llama_token_to_str( ctx , new_token_id ) );
-        fflush( stdout );
+        // print the new token :
+        printf("%s", llama_token_to_str(ctx, new_token_id));
+        fflush(stdout);
 
-        // Push this new token for next evaluation :
-        tokens_list.push_back( new_token_id );
+        // push this new token for next evaluation
+        tokens_list.push_back(new_token_id);
 
-    } // wend of main loop
+    }
 
-    llama_free( ctx );
-    llama_free_model( model );
+    llama_free(ctx);
+    llama_free_model(model);
 
     llama_backend_free();
 
     return 0;
 }
-
-// EOF
diff --git a/gguf-util.h b/gguf-util.h
index db0233cfca098..6bbabf667911e 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -5,7 +5,9 @@
 
 #ifndef GGUF_UTIL_H
 #define GGUF_UTIL_H
+
 #include "ggml.h"
+
 #include <cstdio>
 #include <cstdint>
 #include <cerrno>
@@ -62,7 +64,6 @@ static std::string format(const char * fmt, ...) {
     return std::string(buf.data(), size);
 }
 
-
 template<typename T>
 static std::string to_string(const T & val) {
     std::stringstream ss;

From 62490f1380a9f4b5036b7122b01449781233e5fa Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 13:04:35 +0300
Subject: [PATCH 142/242] gguf : use UNIX line ending

---
 constants.py   |  108 +-
 gguf-llama.cpp | 8394 ++++++++++++++++++++++++------------------------
 gguf-llama.h   |  898 +++---
 gguf-util.h    | 1134 +++----
 gguf.py        |  678 ++--
 5 files changed, 5606 insertions(+), 5606 deletions(-)

diff --git a/constants.py b/constants.py
index 7fa238a73278c..87d6b7a0687aa 100644
--- a/constants.py
+++ b/constants.py
@@ -1,54 +1,54 @@
-GGUF_MAGIC             = 0x47475546
-GGUF_VERSION           = 1
-GGUF_DEFAULT_ALIGNMENT = 32
-
-# general
-KEY_GENERAL_ARCHITECTURE         = "general.architecture"
-KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
-KEY_GENERAL_ALIGNMENT            = "general.alignment"
-KEY_GENERAL_NAME                 = "general.name"
-KEY_GENERAL_AUTHOR               = "general.author"
-KEY_GENERAL_URL                  = "general.url"
-KEY_GENERAL_DESCRIPTION          = "general.description"
-KEY_GENERAL_FILE_TYPE            = "general.file_type"
-KEY_GENERAL_LICENSE              = "general.license"
-KEY_GENERAL_SOURCE_URL           = "general.source.url"
-KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
-
-# LLM
-KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
-KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
-KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
-KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
-KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
-KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
-
-# attention
-KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
-KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
-KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
-KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
-KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
-KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
-
-# RoPE
-KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
-KEY_ROPE_SCALE                   = "{llm}.rope.scale"
-
-# tokenization
-KEY_TOKENIZER_MODEL   = "tokenizer.ggml.model"
-KEY_TOKENIZER_LIST    = "tokenizer.ggml.tokens"
-KEY_TOKENIZER_SCORES  = "tokenizer.ggml.scores"
-KEY_TOKENIZER_MERGES  = "tokenizer.ggml.merges"
-KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
-KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
-KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
-KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.seperator_token_id"
-KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
-KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json"
-KEY_TOKENIZER_RWKV    = "tokenizer.rwkv.world"
-KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
-KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
-KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
-KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.separator_token_id"
-KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
+GGUF_MAGIC             = 0x47475546
+GGUF_VERSION           = 1
+GGUF_DEFAULT_ALIGNMENT = 32
+
+# general
+KEY_GENERAL_ARCHITECTURE         = "general.architecture"
+KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
+KEY_GENERAL_ALIGNMENT            = "general.alignment"
+KEY_GENERAL_NAME                 = "general.name"
+KEY_GENERAL_AUTHOR               = "general.author"
+KEY_GENERAL_URL                  = "general.url"
+KEY_GENERAL_DESCRIPTION          = "general.description"
+KEY_GENERAL_FILE_TYPE            = "general.file_type"
+KEY_GENERAL_LICENSE              = "general.license"
+KEY_GENERAL_SOURCE_URL           = "general.source.url"
+KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
+
+# LLM
+KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
+KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
+KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
+KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
+
+# attention
+KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
+KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
+KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
+
+# RoPE
+KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
+KEY_ROPE_SCALE                   = "{llm}.rope.scale"
+
+# tokenization
+KEY_TOKENIZER_MODEL   = "tokenizer.ggml.model"
+KEY_TOKENIZER_LIST    = "tokenizer.ggml.tokens"
+KEY_TOKENIZER_SCORES  = "tokenizer.ggml.scores"
+KEY_TOKENIZER_MERGES  = "tokenizer.ggml.merges"
+KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.seperator_token_id"
+KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
+KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json"
+KEY_TOKENIZER_RWKV    = "tokenizer.rwkv.world"
+KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.separator_token_id"
+KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index dd6c59df3503d..99da3c56dc92c 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -1,4197 +1,4197 @@
-// Defines fileno on msys:
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#include <cstddef>
-#include <cstdint>
-#include <cstdio>
-#endif
-
-#include "gguf-util.h"
-#include "gguf-llama.h"
-
-#include "ggml.h"
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-#elif defined(GGML_USE_CLBLAST)
-#include "ggml-opencl.h"
-#endif
-
-#ifdef GGML_USE_METAL
-#include "ggml-metal.h"
-#endif
-#ifdef GGML_USE_MPI
-#include "ggml-mpi.h"
-#endif
-#ifdef GGML_USE_K_QUANTS
-#ifndef QK_K
-#ifdef GGML_QKK_64
-#define QK_K 64
-#else
-#define QK_K 256
-#endif
-#endif
-#endif
-
-#include <array>
-#include <ctime>
-#include <cinttypes>
-#include <fstream>
-#include <random>
-#include <map>
-#include <unordered_map>
-#include <queue>
-#include <cassert>
-#include <cstring>
-#include <climits>
-#include <memory>
-#include <algorithm>
-#include <initializer_list>
-#include <thread>
-#include <atomic>
-#include <mutex>
-#include <sstream>
-#include <numeric>
-
-#if defined(_MSC_VER)
-#pragma warning(disable: 4244 4267) // possible loss of data
-#endif
-
-#define LLAMA_USE_SCRATCH
-#define LLAMA_MAX_SCRATCH_BUFFERS 16
-
-// available llama models
-enum e_model {
-    MODEL_UNKNOWN,
-    MODEL_3B,
-    MODEL_7B,
-    MODEL_13B,
-    MODEL_30B,
-    MODEL_65B,
-    MODEL_70B,
-};
-
-static const size_t kB = 1024;
-static const size_t MB = 1024*1024;
-
-// computed for n_ctx == 2048
-// TODO: dynamically determine these sizes
-//       needs modifications in ggml
-
-typedef void (*offload_func_t)(struct ggml_tensor * tensor);
-
-void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
-    (void) tensor;
-}
-
-//
-// ggml helpers
-//
-
-static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
-    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
-
-    if (plan.work_size > 0) {
-        buf.resize(plan.work_size);
-        plan.work_data = buf.data();
-    }
-
-    ggml_graph_compute(graph, &plan);
-}
-
-//
-// memory sizes (calculated for n_batch == 512)
-//
-
-static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   ((size_t) n_ctx / 16ull +  92ull) * MB },
-        { MODEL_7B,   ((size_t) n_ctx / 16ull + 100ull) * MB },
-        { MODEL_13B,  ((size_t) n_ctx / 12ull + 120ull) * MB },
-        { MODEL_30B,  ((size_t) n_ctx /  9ull + 160ull) * MB },
-        { MODEL_65B,  ((size_t) n_ctx /  6ull + 256ull) * MB }, // guess
-        { MODEL_70B,  ((size_t) n_ctx /  7ull + 164ull) * MB },
-    };
-    return k_sizes;
-}
-
-static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,  128ull * MB },
-        { MODEL_7B,  160ull * MB },
-        { MODEL_13B, 192ull * MB },
-        { MODEL_30B, 256ull * MB },
-        { MODEL_65B, 384ull * MB }, // guess
-        { MODEL_70B, 304ull * MB },
-    };
-    return k_sizes;
-}
-
-// used to store the compute graph tensors + non-scratch data
-static const std::map<e_model, size_t> & MEM_REQ_EVAL()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   8ull * MB },
-        { MODEL_7B,  10ull * MB },
-        { MODEL_13B, 12ull * MB },
-        { MODEL_30B, 16ull * MB },
-        { MODEL_65B, 24ull * MB }, // guess
-        { MODEL_70B, 24ull * MB },
-    };
-    return k_sizes;
-}
-
-// amount of VRAM needed per batch size to hold temporary results
-// the values for 3b and 65b are not derived from testing but instead chosen conservatively
-static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   512ull * kB },
-        { MODEL_7B,   512ull * kB },
-        { MODEL_13B,  640ull * kB },
-        { MODEL_30B,  768ull * kB },
-        { MODEL_65B, 1536ull * kB },
-        { MODEL_70B, 1536ull * kB }, // TODO (likely can be reduced)
-    };
-    return k_sizes;
-}
-
-// amount of VRAM needed per batch size and context to hold temporary results
-// the values for 3b and 65b are not derived from testing but instead chosen conservatively
-static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,  128ull },
-        { MODEL_7B,  128ull },
-        { MODEL_13B, 160ull },
-        { MODEL_30B, 208ull },
-        { MODEL_65B, 416ull },
-        { MODEL_70B, 416ull }, // TODO (likely can be reduced)
-    };
-    return k_sizes;
-}
-
-// default hparams (LLaMA 7B)
-struct llama_hparams {
-    uint32_t n_vocab   = 32000;
-    uint32_t n_ctx     = 512;   // this is provided as user input?
-    uint32_t n_embd    = 4096;
-    uint32_t n_head    = 32;
-    uint32_t n_head_kv = 32;
-    uint32_t n_layer   = 32;
-    uint32_t n_rot     = 64;
-    uint32_t n_ff       = 11008;
-
-    float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
-
-    float rope_freq_base  = 10000.0f;
-    float rope_freq_scale = 1.0f;
-
-    enum llama_ftype ftype = LLAMA_FTYPE_MOSTLY_F16;
-
-    bool operator!=(const llama_hparams & other) const {
-        return static_cast<bool>(memcmp(this, &other, sizeof(llama_hparams))); // NOLINT
-    }
-
-    uint32_t n_gqa() const {
-        return n_head/n_head_kv;
-    }
-
-    uint32_t n_embd_head() const {
-        return n_embd/n_head;
-    }
-
-    uint32_t n_embd_gqa() const {
-        return n_embd/n_gqa();
-    }
-
-    size_t kv_size() const {
-        size_t result = 2ull;
-        result *= (size_t) n_embd_gqa();
-        result *= (size_t) n_ctx;
-        result *= (size_t) n_layer;
-        result *= sizeof(ggml_fp16_t);
-        return result;
-    }
-};
-
-struct llama_layer {
-    // normalization
-    struct ggml_tensor * attention_norm;
-
-    // attention
-    struct ggml_tensor * wq;
-    struct ggml_tensor * wk;
-    struct ggml_tensor * wv;
-    struct ggml_tensor * wo;
-
-    // normalization
-    struct ggml_tensor * ffn_norm;
-
-    // ff
-    struct ggml_tensor * w1;
-    struct ggml_tensor * w2;
-    struct ggml_tensor * w3;
-};
-
-struct llama_kv_cache {
-    struct ggml_tensor * k = NULL;
-    struct ggml_tensor * v = NULL;
-
-    struct ggml_context * ctx = NULL;
-
-    gguf_ctx_buffer buf;
-
-    int n; // number of tokens currently in the cache
-
-    ~llama_kv_cache() {
-        if (ctx) {
-            ggml_free(ctx);
-        }
-
-#ifdef GGML_USE_CUBLAS
-        ggml_cuda_free_data(k);
-        ggml_cuda_free_data(v);
-#endif // GGML_USE_CUBLAS
-    }
-};
-
-struct llama_vocab {
-    // TODO: convert to this gguf_vocab
-    // add a vector of merges
-    // add members for bos/eos/pad/sep tokens
-    // so that we can pass it to different types of tokenizers with a common interface
-
-    using id    = int32_t;
-    using token = std::string;
-
-    struct token_score {
-        token tok;
-        float score;
-    };
-
-    std::unordered_map<token, id> token_to_id;
-    std::vector<token_score> id_to_token;
-};
-
-struct llama_model {
-    e_model type = MODEL_UNKNOWN;
-
-    llama_hparams hparams;
-
-    struct ggml_tensor * tok_embeddings;
-
-    struct ggml_tensor * norm;
-    struct ggml_tensor * output;
-
-    std::vector<llama_layer> layers;
-    int n_gpu_layers;
-
-    // context
-    struct ggml_context * ctx = NULL;
-
-    // the model memory buffer
-    gguf_ctx_buffer buf;
-
-    // model memory mapped file
-    std::unique_ptr<gguf_mmap> mapping;
-
-    // objects representing data potentially being locked in memory
-    gguf_mlock mlock_buf;
-    gguf_mlock mlock_mmap;
-
-    // for quantize-stats only
-    std::vector<std::pair<std::string, struct ggml_tensor *>> tensors_by_name;
-
-    int64_t t_load_us = 0;
-    int64_t t_start_us = 0;
-
-    llama_vocab vocab;
-
-    ~llama_model() {
-        if (ctx) {
-            ggml_free(ctx);
-        }
-
-#ifdef GGML_USE_CUBLAS
-        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
-            ggml_cuda_free_data(tensors_by_name[i].second);
-        }
-        ggml_cuda_free_scratch();
-#elif defined(GGML_USE_CLBLAST)
-        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
-            ggml_cl_free_data(tensors_by_name[i].second);
-        }
-#endif
-    }
-};
-
-struct llama_context {
-    llama_context(const llama_model & model) : model(model), t_load_us(model.t_load_us), t_start_us(model.t_start_us) {}
-#ifdef GGML_USE_METAL
-    ~llama_context() {
-        if (ctx_metal) {
-            ggml_metal_free(ctx_metal);
-        }
-    }
-#endif
-    std::mt19937 rng;
-
-    bool has_evaluated_once = false;
-
-    int64_t t_sample_us = 0;
-    int64_t t_eval_us   = 0;
-    int64_t t_p_eval_us = 0;
-
-    int32_t n_sample = 0; // number of tokens sampled
-    int32_t n_eval   = 0; // number of eval calls
-    int32_t n_p_eval = 0; // number of tokens in eval calls for the prompt (with batch size > 1)
-
-    const llama_model & model;
-
-    bool model_owner = false;
-
-    int64_t t_load_us;
-    int64_t t_start_us;
-
-    // key + value cache for the self attention
-    struct llama_kv_cache kv_self;
-
-    size_t mem_per_token = 0;
-
-    // decode output (2-dimensional array: [n_tokens][n_vocab])
-    std::vector<float> logits;
-    bool logits_all = false;
-
-    // input embedding (1-dimensional array: [n_embd])
-    std::vector<float> embedding;
-
-    // reusable buffer for `struct ggml_graph_plan.work_data`
-    std::vector<uint8_t> work_buffer;
-
-    // memory buffers used to evaluate the model
-    // TODO: move in llama_state
-    gguf_ctx_buffer buf_compute;
-    gguf_ctx_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
-
-#ifdef GGML_USE_METAL
-    ggml_metal_context * ctx_metal = NULL;
-#endif
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_context * ctx_mpi = NULL;
-#endif
-
-    int    buf_last = 0;
-    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
-
-    void use_buf(struct ggml_context * ctx, int i) {
-#if defined(LLAMA_USE_SCRATCH)
-        size_t last_size = 0;
-
-        if (i == -1) {
-            last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
-        } else {
-            auto & buf = buf_scratch[i];
-            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.addr, });
-        }
-
-        if (buf_last >= 0) {
-            buf_max_size[buf_last] = std::max(buf_max_size[buf_last], last_size);
-        }
-
-        buf_last = i;
-#else
-        (void) i;
-        (void) ctx;
-#endif
-    }
-
-    size_t get_buf_max_mem(int i) const {
-#if defined(LLAMA_USE_SCRATCH)
-        return buf_max_size[i];
-#else
-        (void) i;
-        return 0;
-#endif
-    }
-};
-
-template <typename T>
-static T checked_mul(T a, T b) {
-    T ret = a * b;
-    if (a != 0 && ret / a != b) {
-        throw std::runtime_error(format("overflow multiplying %llu * %llu",
-                     (unsigned long long) a, (unsigned long long) b));
-    }
-    return ret;
-}
-
-static size_t checked_div(size_t a, size_t b) {
-    if (b == 0 || a % b != 0) {
-        throw std::runtime_error(format("error dividing %zu / %zu", a, b));
-    }
-    return a / b;
-}
-
-static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
-    char buf[256];
-    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
-    for (size_t i = 1; i < ne.size(); i++) {
-        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
-    }
-    return buf;
-}
-
-static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
-    size_t size = ggml_type_size(type);
-    for (uint32_t dim : ne) {
-        size = checked_mul<size_t>(size, dim);
-    }
-    return size / ggml_blck_size(type);
-}
-
-struct gguf_load_tensor {
-    std::string name;
-    enum ggml_type type = GGML_TYPE_F32;
-    std::vector<uint32_t> ne;
-    size_t file_off;
-    size_t size;
-    struct ggml_tensor * ggml_tensor = NULL;
-    uint8_t * data;
-};
-
-struct gguf_load_tensors_map {
-    // tensors is kept in a separate vector to preserve file order
-    std::vector<gguf_load_tensor> tensors;
-    std::unordered_map<std::string, size_t> name_to_idx;
-};
-
-enum gguf_file_version {
-    GGUF_FILE_VERSION_V1 = 1,
-
-};
-
-
-struct gguf_file_loader {
-    gguf_file file;
-    gguf_context * gguf_ctx;
-    gguf_file_version file_version;
-    llama_hparams hparams;
-    llama_vocab vocab;
-struct ggml_context * ctx_data = NULL;
-
-    gguf_file_loader(const char * fname, gguf_load_tensors_map & tensors_map)
-        : file(fname, "rb") {
-        fprintf(stderr, "llama.cpp: loading model from %s\n", fname);
-
-    struct gguf_init_params params = {
-        /*.no_alloc = */ true,
-        /*.ctx      = */ &ctx_data,
-    };
-
-    gguf_ctx = gguf_init_from_file(fname, params);
-    file_version = (enum gguf_file_version) gguf_get_version(gguf_ctx);
-
-    read_hparams();
-    read_vocab();
-        read_tensor_metadata(tensors_map);
-    }
-
-    uint32_t read_u32(const char * key) {
-        int i = gguf_find_key(gguf_ctx, key);
-        if (i == -1) {
-            throw std::runtime_error(format("cannot find param with key %s\n", key));
-        }
-
-        return gguf_get_val_u32(gguf_ctx, i);
-    }
-
-    float read_f32(const char * key) {
-        int i = gguf_find_key(gguf_ctx, key);
-        if (i == -1) {
-            throw std::runtime_error(format("cannot find param with key %s\n", key));
-        }
-
-        return gguf_get_val_f32(gguf_ctx, i);
-    }
-
-    int read_n_vocab() {
-        int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
-    if (i == -1) {
-        throw std::runtime_error("cannot find token list in GGUF file\n");
-    }
-
-    return gguf_get_arr_n(gguf_ctx, i);
-    }
-
-    void read_hparams() {
-
-        // TODO define keys as constants in header
-        // TODO: read all hparams from file
-
-        hparams.n_vocab = read_n_vocab();
-        hparams.n_ctx   = read_u32("llama.context_length");
-        hparams.n_embd  = read_u32("llama.embedding_length");
-        hparams.n_ff    = read_u32("llama.feed_forward_length");
-        hparams.n_head  = read_u32("llama.attention.head_count");
-        hparams.n_layer = read_u32("llama.layer_count");
-        hparams.n_rot   = read_u32("llama.rope.dimension_count");
-        hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
-
-        // LLaMAv2
-        // hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
-    }
-
-    void read_vocab() {
-        vocab.id_to_token.resize(hparams.n_vocab);
-        int token_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
-        if (token_idx == -1) {
-            throw std::runtime_error("cannot find token list in GGUF file\n");
-        }
-
-        int score_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.scores");
-        if (score_idx == -1) {
-            throw std::runtime_error("cannot find token scores list in GGUF file\n");
-        }
-
-        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
-
-            std::string word = gguf_get_arr_str(gguf_ctx, token_idx, i);
-
-            vocab.token_to_id[word] = i;
-
-            auto & tok_score = vocab.id_to_token[i];
-            tok_score.tok = std::move(word);
-            tok_score.score = gguf_get_arr_f32(gguf_ctx, score_idx, i);
-        }
-    }
-
-    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) {
-        const int n_tensors = gguf_get_n_tensors(gguf_ctx);
-
-        for (int i = 0; i < n_tensors; ++i) {
-            gguf_load_tensor tensor;
-            const char * name = gguf_get_tensor_name(gguf_ctx, i);
-
-            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
-            uint32_t n_dims = cur->n_dims;
-            tensor.type = cur->type;
-            tensor.ne.resize(n_dims);
-            for (uint32_t j = 0; j < n_dims; ++j) {
-            tensor.ne[j] = cur->ne[j];
-            }
-
-            if (n_dims < 1 || n_dims > 2) {
-                throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
-            }
-            switch (tensor.type) {
-                case GGML_TYPE_F32:
-                case GGML_TYPE_F16:
-                case GGML_TYPE_Q4_0:
-                case GGML_TYPE_Q4_1:
-                case GGML_TYPE_Q5_0:
-                case GGML_TYPE_Q5_1:
-                case GGML_TYPE_Q8_0:
-                case GGML_TYPE_Q2_K:
-                case GGML_TYPE_Q3_K:
-                case GGML_TYPE_Q4_K:
-                case GGML_TYPE_Q5_K:
-                case GGML_TYPE_Q6_K:
-                    break;
-                default: {
-                    throw std::runtime_error(format("unrecognized tensor type %u\n", tensor.type));
-                }
-            }
-
-
-            tensor.file_off = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, i);
-
-            tensor.name = name;
-            tensor.size = llama_calc_tensor_size(tensor.ne, tensor.type);
-
-            tensors_map.tensors.push_back(tensor);
-            tensors_map.name_to_idx[name] = tensors_map.tensors.size() - 1;
-        }
-    }
-};
-
-struct gguf_file_saver {
-    // TODO
-    // this implementation now assumes that the data section is of the same length as the unquantized model.
-    // this is needed to write tensor metadata and weights in a single pass by seeking to appropriate positions in the file.
-    // this may not be true when we add quantization version and change ftype description (currently it's string according to the specs,
-    // but better to have it as uint32).
-    // we need to calculate the delta in number of bytes written with a counter as a struct member.
-
-    gguf_file file;
-    gguf_file_loader * fl;
-    size_t info_offset;
-    size_t tensor_offset = 0;
-
-    gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
-        : file(fname, "wb"), fl(fl) {
-        fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
-        write_header();
-        write_hparams(new_ftype);
-    }
-
-    void write_header() {
-        const int32_t magic = GGUF_MAGIC;
-        file.write_i32(magic);
-
-            const int32_t version = GGUF_VERSION;
-            file.write_i32(version);
-
-            const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
-            file.write_i32(n_tensors);
-
-            const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
-            file.write_i32(n_kv);
-        }
-
-        void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
-            std::vector<std::string> data(n_arr);
-
-            for (int j = 0; j < n_arr; ++j) {
-                std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
-                data[j] = val;
-                }
-
-                            file.write_arr<std::string>(key, type, data);
-        }
-
-        void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
-            std::vector<float> data(n_arr);
-
-            for (int j = 0; j < n_arr; ++j) {
-                float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
-                data[j] = val;
-                }
-
-                            file.write_arr<float>(key, type, data);
-        }
-
-    void write_hparams(enum llama_ftype new_ftype) {
-        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
-        for (int i = 0; i < n_kv; ++i) {
-            const char * key = gguf_get_key(fl->gguf_ctx, i);
-            if (strcmp(key, "general.quantization_version") == 0) {
-                file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
-            } else {
-                const gguf_type vtype = gguf_get_kv_type(fl->gguf_ctx, i);
-
-                bool bool_val;
-                float f32_val;
-                int16_t i16_val;
-                int32_t i32_val;
-                int8_t i8_val;
-                std::string str_val;
-                uint16_t u16_val;
-                uint32_t u32_val;
-                uint8_t u8_val;
-                gguf_type arr_type;
-                int n_arr;
-
-                switch(vtype) {
-                    case GGUF_TYPE_BOOL:
-                    bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
-                    file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
-                    break;
-                    case GGUF_TYPE_FLOAT32:
-                    f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
-                    file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
-                    break;
-                    case GGUF_TYPE_INT16:
-                    i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
-                    file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
-                    break;
-                    case GGUF_TYPE_INT32:
-                    i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
-                    file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
-                    break;
-                    case GGUF_TYPE_INT8:
-                    i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
-                    file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
-                    break;
-                    case GGUF_TYPE_STRING:
-                    str_val = gguf_get_val_str(fl->gguf_ctx, i);
-                    file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
-                    break;
-                    case GGUF_TYPE_UINT16:
-                    u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
-                    file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
-                    break;
-                    case GGUF_TYPE_UINT32:
-                    u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
-                    file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
-                    break;
-                    case GGUF_TYPE_UINT8:
-                    u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
-                    file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
-                    break;
-                    case GGUF_TYPE_ARRAY:
-                    arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
-                    n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
-                    if (arr_type == GGUF_TYPE_FLOAT32) {
-                        write_hparam_arr_f32(key, arr_type, i, n_arr);
-                        } else if (arr_type == GGUF_TYPE_STRING) {
-                            write_hparam_arr_str(key, GGUF_TYPE_STRING, i, n_arr);
-                        } else {
-                            throw std::runtime_error("not implemented");
-                        }
-                    break;
-                    default:
-                    throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
-                }
-            }
-        }
-
-        info_offset  = file.tell();
-        size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
-        file.write_zeros(count);
-        file.seek(info_offset, SEEK_SET);
-        GGML_ASSERT(info_offset == file.tell());
-    }
-
-    size_t write_tensor_info(gguf_load_tensor & tensor, enum ggml_type type) {
-        size_t total_written = 0;
-        file.seek(info_offset, SEEK_SET);
-        GGML_ASSERT(info_offset == file.tell());
-        total_written += file.write_str(tensor.name);
-
-        int32_t n_dims = tensor.ne.size();
-        total_written += file.write_i32(n_dims);
-        for (int32_t i = 0; i < n_dims; ++i) {
-            total_written += file.write_i32(tensor.ne[i]);
-        }
-
-        total_written += file.write_i32(type);
-        total_written += file.write_u64(tensor_offset);
-        info_offset   += total_written; // position to write info of the next tensor
-
-        file.seek(0, SEEK_END);
-
-        return total_written;
-    }
-
-    void write_tensor(gguf_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
-        switch (new_type) {
-            case GGML_TYPE_F32:
-            case GGML_TYPE_F16:
-            case GGML_TYPE_Q4_0:
-            case GGML_TYPE_Q4_1:
-            case GGML_TYPE_Q5_0:
-            case GGML_TYPE_Q5_1:
-            case GGML_TYPE_Q8_0:
-            case GGML_TYPE_Q2_K:
-            case GGML_TYPE_Q3_K:
-            case GGML_TYPE_Q4_K:
-            case GGML_TYPE_Q5_K:
-            case GGML_TYPE_Q6_K:
-                break;
-            default: GGML_ASSERT(false);
-        }
-
-        write_tensor_info(tensor, new_type);
-        file.write_raw(new_data, new_size);
-        size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
-        size_t pad = padded_size - new_size;
-        file.write_zeros(pad);
-        tensor_offset += padded_size; // offset of the next tensor
-    }
-};
-
-struct llama_model_loader {
-    std::unique_ptr<gguf_file_loader> file_loader;
-    gguf_load_tensors_map tensors_map;
-    bool use_mmap;
-    size_t num_ggml_tensors_created = 0;
-    struct ggml_context * ggml_ctx = NULL;
-    std::unique_ptr<gguf_mmap> mapping;
-
-    llama_model_loader(const std::string & fname_base, bool use_mmap) {
-        file_loader = std::unique_ptr<gguf_file_loader>(new gguf_file_loader(fname_base.c_str(), tensors_map));
-        if (!gguf_mmap::SUPPORTED) {
-            use_mmap = false;
-        }
-        this->use_mmap = use_mmap;
-    }
-
-    void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
-        *ctx_size_p = *mmapped_size_p = 0;
-        for (const gguf_load_tensor & lt : tensors_map.tensors) {
-            *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
-            *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
-        }
-    }
-
-    struct ggml_tensor * get_tensor(const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
-        auto it = tensors_map.name_to_idx.find(name);
-        if (it == tensors_map.name_to_idx.end()) {
-            throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
-        }
-        gguf_load_tensor & lt = tensors_map.tensors.at(it->second);
-        if (lt.ne != ne) {
-            throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
-                         name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
-        }
-
-        return get_tensor_for(lt, backend);
-    }
-
-    struct ggml_tensor * get_tensor_for(gguf_load_tensor & lt, ggml_backend backend) {
-        struct ggml_tensor * tensor;
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, true);
-        }
-        if (lt.ne.size() == 2) {
-            tensor = ggml_new_tensor_2d(ggml_ctx, lt.type, lt.ne.at(0), lt.ne.at(1));
-        } else {
-            GGML_ASSERT(lt.ne.size() == 1);
-            tensor = ggml_new_tensor_1d(ggml_ctx, lt.type, lt.ne.at(0));
-        }
-        ggml_set_name(tensor, lt.name.c_str());
-        GGML_ASSERT(lt.ggml_tensor == NULL); // if this fails, we called get_tensor twice on the same tensor
-
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, use_mmap);
-        }
-        tensor->backend = backend;
-        lt.ggml_tensor = tensor;
-        num_ggml_tensors_created++;
-        return tensor;
-    }
-
-    void done_getting_tensors() const {
-        if (num_ggml_tensors_created != tensors_map.tensors.size()) {
-            throw std::runtime_error(std::string("llama.cpp: file contained more tensors than expected"));
-        }
-    }
-
-    void load_all_data(llama_progress_callback progress_callback, void *  progress_callback_user_data, gguf_mlock * lmlock) {
-        size_t data_size = 0;
-        size_t prefetch_size = 0;
-        size_t lock_size = 0;
-        for (const gguf_load_tensor & lt : tensors_map.tensors) {
-            data_size += lt.size;
-            if (lt.ggml_tensor->backend == GGML_BACKEND_CPU) {
-                prefetch_size += lt.size;
-            }
-        }
-
-        if (use_mmap) {
-            mapping.reset(new gguf_mmap(&file_loader->file, prefetch_size, ggml_is_numa()));
-            if (lmlock) {
-                lmlock->init(mapping->addr);
-            }
-        }
-
-        size_t done_size = 0;
-        for (gguf_load_tensor & lt : tensors_map.tensors) {
-            if (progress_callback) {
-                progress_callback((float) done_size / data_size, progress_callback_user_data);
-            }
-            GGML_ASSERT(lt.ggml_tensor); // unused tensors should have been caught by load_data already
-            lt.data = (uint8_t *) lt.ggml_tensor->data;
-
-            // allocate temp buffer if not using mmap
-            if (!use_mmap && lt.data == NULL) {
-                GGML_ASSERT(lt.ggml_tensor->backend != GGML_BACKEND_CPU);
-                lt.data = (uint8_t*)malloc(ggml_nbytes(lt.ggml_tensor));
-            }
-
-            load_data_for(lt);
-
-            switch(lt.ggml_tensor->backend) {
-                case GGML_BACKEND_CPU:
-                    lt.ggml_tensor->data = lt.data;
-                    if (use_mmap && lmlock) {
-                        lock_size += lt.size;
-                        lmlock->grow_to(lock_size);
-                    }
-                    break;
-#if defined(GGML_USE_CUBLAS)
-                case GGML_BACKEND_GPU:
-                case GGML_BACKEND_GPU_SPLIT:
-                    ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
-                    if (!use_mmap) {
-                        free(lt.data);
-                    }
-                    break;
-#elif defined(GGML_USE_CLBLAST)
-                case GGML_BACKEND_GPU:
-                    ggml_cl_transform_tensor(lt.data, lt.ggml_tensor);
-                    if (!use_mmap) {
-                        free(lt.data);
-                    }
-                    break;
-#endif
-                default:
-                    continue;
-            }
-
-            done_size += lt.size;
-        }
-    }
-
-    void load_data_for(gguf_load_tensor & lt) {
-        if (use_mmap) {
-            lt.data = (uint8_t *) mapping->addr + lt.file_off;
-        } else {
-            gguf_file & file = file_loader->file;
-            file.seek(lt.file_off, SEEK_SET);
-            file.read_raw(lt.data, lt.size);
-        }
-
-        if (0) {
-            print_checksum(lt);
-        }
-    }
-
-    static void print_checksum(gguf_load_tensor & lt) {
-        uint32_t sum = 0;
-        for (size_t i = 0; i < lt.size; i++) {
-            uint8_t byte = lt.data[i];
-            sum = byte + (sum << 6) + (sum << 16) - sum; // sdbm hash
-        }
-        fprintf(stderr, "%s checksum: %#08x (%s, size %zu)\n", lt.name.c_str(), sum,
-                llama_format_tensor_shape(lt.ne).c_str(), lt.size);
-    }
-
-};
-
-//
-// kv cache
-//
-
-static bool kv_cache_init(
-        const struct llama_hparams & hparams,
-             struct llama_kv_cache & cache,
-                         ggml_type   wtype,
-                               int   n_ctx,
-                               int   n_gpu_layers) {
-    const int n_embd  = hparams.n_embd_gqa();
-    const int n_layer = hparams.n_layer;
-
-    const int64_t n_mem      = n_layer*n_ctx;
-    const int64_t n_elements = n_embd*n_mem;
-
-    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
-    cache.n = 0;
-
-    struct ggml_init_params params;
-    params.mem_size   = cache.buf.size;
-    params.mem_buffer = cache.buf.addr;
-    params.no_alloc   = false;
-
-    cache.ctx = ggml_init(params);
-
-    if (!cache.ctx) {
-        fprintf(stderr, "%s: failed to allocate memory for kv cache\n", __func__);
-        return false;
-    }
-
-    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    ggml_set_name(cache.k, "cache_k");
-    ggml_set_name(cache.v, "cache_v");
-
-    (void) n_gpu_layers;
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer + 1) {
-        ggml_cuda_assign_buffers_no_scratch(cache.v);
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        ggml_cuda_assign_buffers_no_scratch(cache.k);
-    }
-#endif // GGML_USE_CUBLAS
-
-    return true;
-}
-
-struct llama_context_params llama_context_default_params() {
-    struct llama_context_params result = {
-        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
-        /*.n_ctx                       =*/ 512,
-        /*.n_batch                     =*/ 512,
-        /*.n_gqa                       =*/ 1,
-        /*.rms_norm_eps                =*/ LLAMA_DEFAULT_RMS_EPS,
-        /*.gpu_layers                  =*/ 0,
-        /*.main_gpu                    =*/ 0,
-        /*.tensor_split                =*/ nullptr,
-        /*.rope_freq_base              =*/ 10000.0f,
-        /*.rope_freq_scale             =*/ 1.0f,
-        /*.progress_callback           =*/ nullptr,
-        /*.progress_callback_user_data =*/ nullptr,
-        /*.low_vram                    =*/ false,
-        /*.f16_kv                      =*/ true,
-        /*.logits_all                  =*/ false,
-        /*.vocab_only                  =*/ false,
-        /*.use_mmap                    =*/ true,
-        /*.use_mlock                   =*/ false,
-        /*.embedding                   =*/ false,
-    };
-
-    return result;
-}
-
-struct llama_model_quantize_params llama_model_quantize_default_params() {
-    struct llama_model_quantize_params result = {
-        /*.nthread                     =*/ 0,
-        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
-        /*.allow_requantize            =*/ false,
-        /*.quantize_output_tensor      =*/ true,
-    };
-
-    return result;
-}
-
-int llama_max_devices() {
-    return LLAMA_MAX_DEVICES;
-}
-
-bool llama_mmap_supported() {
-    return gguf_mmap::SUPPORTED;
-}
-
-bool llama_mlock_supported() {
-    return gguf_mlock::SUPPORTED;
-}
-
-void llama_backend_init(bool numa) {
-    ggml_time_init();
-
-    // needed to initialize f16 tables
-    {
-        struct ggml_init_params params = { 0, NULL, false };
-        struct ggml_context * ctx = ggml_init(params);
-        ggml_free(ctx);
-    }
-
-    if (numa) {
-        ggml_numa_init();
-    }
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_init();
-#endif
-}
-
-void llama_backend_free() {
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_free();
-#endif
-}
-
-int64_t llama_time_us() {
-    return ggml_time_us();
-}
-
-//
-// model loading
-//
-
-static const char *gguf_file_version_name(gguf_file_version version) {
-    switch (version) {
-        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
-        }
-
-    return "unknown";
-}
-
-static const char *llama_ftype_name(enum llama_ftype ftype) {
-    switch (ftype) {
-        case LLAMA_FTYPE_ALL_F32:     return "all F32";
-        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
-        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
-        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
-        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
-                                      return "mostly Q4_1, some F16";
-        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
-        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
-        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K: return "mostly Q2_K";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q6_K: return "mostly Q6_K";
-        default:                      return "unknown, may not work";
-    }
-}
-
-static const char *llama_model_type_name(e_model type) {
-    switch (type) {
-        case MODEL_3B: return "3B";
-        case MODEL_7B: return "7B";
-        case MODEL_13B: return "13B";
-        case MODEL_30B: return "30B";
-        case MODEL_65B: return "65B";
-        case MODEL_70B: return "70B";
-        default: GGML_ASSERT(false);
-    }
-}
-
-static void llama_model_load_internal(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void * progress_callback_user_data) {
-            GGML_UNUSED(rms_norm_eps); // TODO: update function signature to remove this
-
-    model.t_start_us = ggml_time_us();
-
-    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
-
-    vocab = std::move(ml->file_loader->vocab);
-    model.hparams = ml->file_loader->hparams;
-    model.n_gpu_layers = n_gpu_layers;
-    gguf_file_version file_version = ml->file_loader->file_version;
-
-    auto & hparams = model.hparams;
-
-    {
-        switch (hparams.n_layer) {
-            case 26: model.type = e_model::MODEL_3B; break;
-            case 32: model.type = e_model::MODEL_7B; break;
-            case 40: model.type = e_model::MODEL_13B; break;
-            case 60: model.type = e_model::MODEL_30B; break;
-            case 80: model.type = e_model::MODEL_65B; break;
-            default:
-                {
-                    if (hparams.n_layer < 32) {
-                        model.type = e_model::MODEL_7B;
-                    }
-                } break;
-        }
-
-        hparams.n_ctx = n_ctx;
-
-        // LLaMAv2
-        hparams.n_head_kv = hparams.n_head / n_gqa;
-        if (model.type == e_model::MODEL_65B && n_gqa == 8) {
-            fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
-            model.type = e_model::MODEL_70B;
-            }
-
-        hparams.rope_freq_base  = rope_freq_base;
-        hparams.rope_freq_scale = rope_freq_scale;
-    }
-
-    const uint32_t n_ff = hparams.n_ff;
-
-    {
-        fprintf(stderr, "%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
-        fprintf(stderr, "%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
-        fprintf(stderr, "%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
-        fprintf(stderr, "%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        fprintf(stderr, "%s: n_head     = %u\n",   __func__, hparams.n_head);
-        fprintf(stderr, "%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
-        fprintf(stderr, "%s: n_layer    = %u\n",   __func__, hparams.n_layer);
-        fprintf(stderr, "%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        fprintf(stderr, "%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        fprintf(stderr, "%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
-        fprintf(stderr, "%s: n_ff       = %u\n",   __func__, n_ff);
-        fprintf(stderr, "%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
-        fprintf(stderr, "%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
-        fprintf(stderr, "%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
-        fprintf(stderr, "%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
-    }
-
-    if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
-        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
-        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
-        throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
-    }
-
-    if (vocab_only) {
-        return;
-    }
-
-    auto & ctx = model.ctx;
-
-    size_t ctx_size;
-    size_t mmapped_size;
-    ml->calc_sizes(&ctx_size, &mmapped_size);
-    fprintf(stderr, "%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
-
-    // create the ggml context
-    {
-        model.buf.resize(ctx_size);
-        if (use_mlock) {
-            model.mlock_buf.init   (model.buf.addr);
-            model.mlock_buf.grow_to(model.buf.size);
-        }
-
-        struct ggml_init_params params = {
-            /*.mem_size   =*/ model.buf.size,
-            /*.mem_buffer =*/ model.buf.addr,
-            /*.no_alloc   =*/ ml->use_mmap,
-        };
-
-        model.ctx = ggml_init(params);
-        if (!model.ctx) {
-            throw std::runtime_error(format("ggml_init() failed"));
-        }
-    }
-
-    (void) main_gpu;
-#if defined(GGML_USE_CUBLAS)
-    fprintf(stderr, "%s: using CUDA for GPU acceleration\n", __func__);
-    ggml_cuda_set_main_device(main_gpu);
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU_SPLIT
-#elif defined(GGML_USE_CLBLAST)
-    fprintf(stderr, "%s: using OpenCL for GPU acceleration\n", __func__);
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU
-#else
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_CPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_CPU
-#endif
-
-    // prepare memory for the weights
-    size_t vram_weights = 0;
-    size_t vram_scratch = 0;
-    {
-        const uint32_t n_embd     = hparams.n_embd;
-        const uint32_t n_embd_gqa = hparams.n_embd_gqa();
-        const uint32_t n_layer    = hparams.n_layer;
-        const uint32_t n_vocab    = hparams.n_vocab;
-
-        ml->ggml_ctx = ctx;
-
-        model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
-
-        // "output" tensor
-        {
-            ggml_backend backend_norm;
-            ggml_backend backend_output;
-            if (n_gpu_layers > int(n_layer)) { // NOLINT
-                // norm is not performance relevant on its own but keeping it in VRAM reduces data copying
-                // on Windows however this is detrimental unless everything is on the GPU
-#ifndef _WIN32
-                backend_norm = low_vram ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
-#else
-                backend_norm = low_vram || n_gpu_layers <= (int) n_layer + 2 ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
-#endif // _WIN32
-
-                backend_output = LLAMA_BACKEND_OFFLOAD_SPLIT;
-            } else {
-                backend_norm = GGML_BACKEND_CPU;
-                backend_output = GGML_BACKEND_CPU;
-            }
-
-            model.norm   = ml->get_tensor("norm.weight",   {n_embd},          backend_norm);
-            model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}, backend_output);
-            if (backend_norm == GGML_BACKEND_GPU) {
-                vram_weights += ggml_nbytes(model.norm);
-            }
-            if (backend_output == GGML_BACKEND_GPU_SPLIT) {
-                vram_weights += ggml_nbytes(model.output);
-            }
-        }
-
-        const int i_gpu_start = n_layer - n_gpu_layers;
-
-        model.layers.resize(n_layer);
-        for (uint32_t i = 0; i < n_layer; ++i) {
-            const ggml_backend backend = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD; // NOLINT
-            const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
-
-            auto & layer = model.layers[i];
-
-            std::string layers_i = "layers." + std::to_string(i);
-
-            layer.attention_norm = ml->get_tensor(layers_i + ".attention_norm.weight", {n_embd}, backend);
-
-            layer.wq = ml->get_tensor(layers_i + ".attention.wq.weight", {n_embd, n_embd},     backend_split);
-            layer.wk = ml->get_tensor(layers_i + ".attention.wk.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wv = ml->get_tensor(layers_i + ".attention.wv.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wo = ml->get_tensor(layers_i + ".attention.wo.weight", {n_embd, n_embd},     backend_split);
-
-            layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
-
-            layer.w1 = ml->get_tensor(layers_i + ".feed_forward.w1.weight", {n_embd,   n_ff}, backend_split);
-            layer.w2 = ml->get_tensor(layers_i + ".feed_forward.w2.weight", {  n_ff, n_embd}, backend_split);
-            layer.w3 = ml->get_tensor(layers_i + ".feed_forward.w3.weight", {n_embd,   n_ff}, backend_split);
-
-            if (backend == GGML_BACKEND_GPU) {
-                vram_weights +=
-                    ggml_nbytes(layer.attention_norm) + ggml_nbytes(layer.wq) + ggml_nbytes(layer.wk)             +
-                    ggml_nbytes(layer.wv)             + ggml_nbytes(layer.wo) + ggml_nbytes(layer.ffn_norm) +
-                    ggml_nbytes(layer.w1)             + ggml_nbytes(layer.w2) + ggml_nbytes(layer.w3);
-            }
-        }
-    }
-
-    ml->done_getting_tensors();
-
-    // print memory requirements
-    {
-        const size_t scale = memory_type == GGML_TYPE_F32 ? 2 : 1;
-
-        // this is the total memory required to run the inference
-        const size_t mem_required =
-            ctx_size +
-            mmapped_size - vram_weights + // weights in VRAM not in memory
-            MEM_REQ_SCRATCH0(hparams.n_ctx).at(model.type) +
-            MEM_REQ_SCRATCH1().at(model.type) +
-            MEM_REQ_EVAL().at(model.type);
-
-        // this is the memory required by one llama_state
-        const size_t mem_required_state =
-            scale*hparams.kv_size();
-
-        fprintf(stderr, "%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
-                mem_required / 1024.0 / 1024.0, mem_required_state / 1024.0 / 1024.0);
-
-        (void) vram_scratch;
-        (void) n_batch;
-#ifdef GGML_USE_CUBLAS
-        if (low_vram) {
-            fprintf(stderr, "%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
-            ggml_cuda_set_scratch_size(0); // disable scratch
-        } else {
-            const size_t vram_scratch_base = VRAM_REQ_SCRATCH_BASE().at(model.type);
-            const size_t vram_scratch_per_context = VRAM_REQ_SCRATCH_PER_CONTEXT().at(model.type);
-            vram_scratch = n_batch * (vram_scratch_base + n_ctx * vram_scratch_per_context);
-            ggml_cuda_set_scratch_size(vram_scratch);
-            if (n_gpu_layers > 0) {
-                fprintf(stderr, "%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
-                        __func__, vram_scratch_base / kB, vram_scratch_per_context,
-                        (vram_scratch + MB - 1) / MB); // round up
-            }
-        }
-#endif // GGML_USE_CUBLAS
-
-#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
-        const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
-
-        fprintf(stderr, "%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
-        if (n_gpu_layers > (int) hparams.n_layer) {
-            fprintf(stderr, "%s: offloading non-repeating layers to GPU\n", __func__);
-        }
-        size_t vram_kv_cache = 0;
-
-#ifdef GGML_USE_CUBLAS
-        const int max_backend_supported_layers = hparams.n_layer + 3;
-        const int max_offloadable_layers = low_vram ? hparams.n_layer + 1 : hparams.n_layer + 3;
-        if (n_gpu_layers > (int) hparams.n_layer + 1) {
-            if (low_vram) {
-                fprintf(stderr, "%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
-            } else {
-                fprintf(stderr, "%s: offloading v cache to GPU\n", __func__);
-                vram_kv_cache += hparams.kv_size() / 2;
-            }
-        }
-        if (n_gpu_layers > (int) hparams.n_layer + 2) {
-            if (low_vram) {
-                fprintf(stderr, "%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
-            } else {
-                fprintf(stderr, "%s: offloading k cache to GPU\n", __func__);
-                vram_kv_cache += hparams.kv_size() / 2;
-            }
-        }
-#elif defined(GGML_USE_CLBLAST)
-        const int max_backend_supported_layers = hparams.n_layer + 1;
-        const int max_offloadable_layers = hparams.n_layer + 1;
-#endif // GGML_USE_CUBLAS
-
-        fprintf(stderr, "%s: offloaded %d/%d layers to GPU\n",
-                __func__, std::min(n_gpu_layers, max_offloadable_layers), max_backend_supported_layers);
-        fprintf(stderr, "%s: total VRAM used: %zu MB\n",
-                __func__, (vram_weights + vram_scratch + vram_kv_cache + MB - 1) / MB); // round up
-#else
-        (void) n_gpu_layers;
-#endif // defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
-    }
-
-    // populate `tensors_by_name`
-    for (gguf_load_tensor & lt : ml->tensors_map.tensors) {
-        model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
-    }
-
-    (void) tensor_split;
-#if defined(GGML_USE_CUBLAS)
-    {
-        ggml_cuda_set_tensor_split(tensor_split);
-    }
-#endif
-
-    ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
-
-    if (progress_callback) {
-        progress_callback(1.0f, progress_callback_user_data);
-    }
-
-    model.mapping = std::move(ml->mapping);
-
-    // loading time will be recalculate after the first eval, so
-    // we take page faults deferred by mmap() into consideration
-    model.t_load_us = ggml_time_us() - model.t_start_us;
-}
-
-static bool llama_model_load(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void *progress_callback_user_data) {
-    try {
-        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gqa, rms_norm_eps, n_gpu_layers, main_gpu, tensor_split, rope_freq_base, rope_freq_scale, low_vram, memory_type,
-                                  use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
-        return true;
-    } catch (const std::exception & err) {
-        fprintf(stderr, "error loading model: %s\n", err.what());
-        return false;
-    }
-}
-
-// evaluate the transformer
-//
-//   - lctx:      llama context
-//   - tokens:    new batch of tokens to process
-//   - embd       embeddings input
-//   - n_tokens   number of tokens
-//   - n_past:    the context size so far
-//   - n_threads: number of threads to use
-//
-static bool llama_eval_internal(
-         llama_context & lctx,
-     const llama_token * tokens,
-           const float * embd,
-                   int   n_tokens,
-                   int   n_past,
-                   int   n_threads,
-            const char * cgraph_fname) {
-
-    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
-#endif
-
-    const int64_t t_start_us = ggml_time_us();
-
-    const int N = n_tokens;
-
-    const auto & model   = lctx.model;
-    const auto & hparams = model.hparams;
-
-    const auto & kv_self = lctx.kv_self;
-
-    GGML_ASSERT(!!kv_self.ctx);
-
-    const int64_t n_embd      = hparams.n_embd;
-    const int64_t n_layer     = hparams.n_layer;
-    const int64_t n_ctx       = hparams.n_ctx;
-    const int64_t n_head      = hparams.n_head;
-    const int64_t n_head_kv   = hparams.n_head_kv;
-    const int64_t n_embd_head = hparams.n_embd_head();
-    const int64_t n_vocab     = hparams.n_vocab;
-    const int64_t n_embd_gqa  = hparams.n_embd_gqa();
-
-
-    GGML_ASSERT(n_embd_head == hparams.n_rot);
-
-    const float freq_base  = hparams.rope_freq_base;
-    const float freq_scale = hparams.rope_freq_scale;
-    const float rms_norm_eps = hparams.f_rms_norm_eps;
-
-    const int n_gpu_layers = model.n_gpu_layers;
-
-    auto & mem_per_token = lctx.mem_per_token;
-    auto & buf_compute   = lctx.buf_compute;
-
-    struct ggml_init_params params = {
-        /*.mem_size   =*/ buf_compute.size,
-        /*.mem_buffer =*/ buf_compute.addr,
-        /*.no_alloc   =*/ false,
-    };
-
-    struct ggml_context * ctx0 = ggml_init(params);
-
-    ggml_cgraph * gf = ggml_new_graph(ctx0);
-
-    // for big prompts, if BLAS is enabled, it is better to use only one thread
-    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
-    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
-
-    struct ggml_tensor * cur;
-    struct ggml_tensor * inpL;
-
-    if (tokens) {
-        struct ggml_tensor * inp_tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
-        memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
-        ggml_set_name(inp_tokens, "inp_tokens");
-
-        inpL = ggml_get_rows(ctx0, model.tok_embeddings, inp_tokens);
-    } else {
-#ifdef GGML_USE_MPI
-        GGML_ASSERT(false && "not implemented");
-#endif
-
-        inpL = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N);
-        memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
-    }
-
-    const int i_gpu_start = n_layer - n_gpu_layers;
-    (void) i_gpu_start;
-
-    // offload functions set the tensor output backend to GPU
-    // tensors are GPU-accelerated if any input or the output has been offloaded
-    //
-    // with the low VRAM option VRAM scratch is disabled in llama_load_model_internal
-    // in that case ggml_cuda_assign_buffers has no effect
-    offload_func_t offload_func_nr = llama_nop; // nr = non-repeating
-    offload_func_t offload_func_kq = llama_nop;
-    offload_func_t offload_func_v  = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer) {
-        offload_func_nr = ggml_cuda_assign_buffers;
-    }
-    if (n_gpu_layers > n_layer + 1) {
-        offload_func_v  = ggml_cuda_assign_buffers;
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        offload_func_kq = ggml_cuda_assign_buffers;
-    }
-#endif // GGML_USE_CUBLAS
-
-    for (int il = 0; il < n_layer; ++il) {
-        ggml_format_name(inpL, "layer_inp_%d", il);
-
-        offload_func_t offload_func = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-        if (il >= i_gpu_start) {
-            offload_func = ggml_cuda_assign_buffers;
-        }
-#endif // GGML_USE_CUBLAS
-
-        struct ggml_tensor * inpSA = inpL;
-
-        lctx.use_buf(ctx0, 0);
-
-        // norm
-        {
-            cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
-            offload_func(cur);
-            ggml_set_name(cur, "rms_norm_0");
-
-            // cur = cur*attention_norm(broadcasted)
-            cur = ggml_mul(ctx0, cur, model.layers[il].attention_norm);
-            offload_func(cur);
-            ggml_set_name(cur, "attention_norm_0");
-        }
-
-        // self-attention
-        {
-            // compute Q and K and RoPE them
-            struct ggml_tensor * tmpk = ggml_mul_mat(ctx0, model.layers[il].wk, cur);
-            offload_func_kq(tmpk);
-            ggml_set_name(tmpk, "tmpk");
-
-            struct ggml_tensor * tmpq = ggml_mul_mat(ctx0, model.layers[il].wq, cur);
-            offload_func_kq(tmpq);
-            ggml_set_name(tmpq, "tmpq");
-
-            struct ggml_tensor * Kcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpk, n_embd_head, n_head_kv, N), n_past, n_embd_head, 0, 0, freq_base, freq_scale);
-            offload_func_kq(Kcur);
-            ggml_set_name(Kcur, "Kcur");
-
-            struct ggml_tensor * Qcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpq, n_embd_head, n_head, N),    n_past, n_embd_head, 0, 0, freq_base, freq_scale);
-            offload_func_kq(Qcur);
-            ggml_set_name(Qcur, "Qcur");
-
-            // store key and value to memory
-            {
-                // compute the transposed [N, n_embd] V matrix
-
-                struct ggml_tensor * tmpv = ggml_mul_mat(ctx0, model.layers[il].wv, cur);
-                offload_func_v(tmpv);
-                ggml_set_name(tmpv, "tmpv");
-
-                struct ggml_tensor * Vcur = ggml_transpose(ctx0, ggml_reshape_2d(ctx0, tmpv, n_embd_gqa, N));
-                offload_func_v(Vcur);
-                ggml_set_name(Vcur, "Vcur");
-
-                struct ggml_tensor * k = ggml_view_1d(ctx0, kv_self.k, N*n_embd_gqa, (ggml_element_size(kv_self.k)*n_embd_gqa)*(il*n_ctx + n_past));
-                offload_func_kq(k);
-                ggml_set_name(k, "k");
-
-                struct ggml_tensor * v = ggml_view_2d(ctx0, kv_self.v, N, n_embd_gqa,
-                        (   n_ctx)*ggml_element_size(kv_self.v),
-                        (il*n_ctx)*ggml_element_size(kv_self.v)*n_embd_gqa + n_past*ggml_element_size(kv_self.v));
-                offload_func_v(v);
-                ggml_set_name(v, "v");
-
-                // important: storing RoPE-ed version of K in the KV cache!
-                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Kcur, k));
-                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Vcur, v));
-            }
-
-            struct ggml_tensor * Q =
-                ggml_permute(ctx0,
-                        Qcur,
-                        0, 2, 1, 3);
-            offload_func_kq(Q);
-            ggml_set_name(Q, "Q");
-
-            struct ggml_tensor * K =
-                ggml_permute(ctx0,
-                        ggml_reshape_3d(ctx0,
-                            ggml_view_1d(ctx0, kv_self.k, (n_past + N)*n_embd_gqa, il*n_ctx*ggml_element_size(kv_self.k)*n_embd_gqa),
-                            n_embd_head, n_head_kv, n_past + N),
-                        0, 2, 1, 3);
-            offload_func_kq(K);
-            ggml_set_name(K, "K");
-
-            // K * Q
-            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
-            offload_func_kq(KQ);
-            ggml_set_name(KQ, "KQ");
-
-            // KQ_scaled = KQ / sqrt(n_embd_head)
-            struct ggml_tensor * KQ_scale = ggml_new_f32(ctx0, 1.0f/sqrtf(float(n_embd)/n_head));
-            ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
-
-            // KQ_scaled shape [n_past + N, N, n_head, 1]
-            struct ggml_tensor * KQ_scaled = ggml_scale_inplace(ctx0, KQ, KQ_scale);
-            offload_func_kq(KQ_scaled);
-            ggml_set_name(KQ_scaled, "KQ_scaled");
-
-            // KQ_masked = mask_past(KQ_scaled)
-            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
-            offload_func_kq(KQ_masked);
-            ggml_set_name(KQ_masked, "KQ_masked");
-
-            // KQ = soft_max(KQ_masked)
-            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
-            offload_func_v(KQ_soft_max);
-            ggml_set_name(KQ_soft_max, "KQ_soft_max");
-
-            // split cached V into n_head heads
-            struct ggml_tensor * V =
-                ggml_view_3d(ctx0, kv_self.v,
-                        n_past + N, n_embd_head, n_head_kv,
-                        n_ctx*ggml_element_size(kv_self.v),
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_head,
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_gqa*il);
-            offload_func_v(V);
-            ggml_set_name(V, "V");
-
-#if 1
-            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
-            offload_func_v(KQV);
-            ggml_set_name(KQV, "KQV");
-#else
-            // make V contiguous in memory to speed up the matmul, however we waste time on the copy
-            // on M1 this is faster for the perplexity computation, but ~5% slower for the single-token generation
-            // is there a better way?
-            struct ggml_tensor * V_cont = ggml_cpy(ctx0, V, ggml_new_tensor_3d(ctx0, kv_self.v->type, n_past + N, n_embd_head, n_head));
-            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V_cont, KQ_soft_max);
-#endif
-
-            // KQV_merged = KQV.permute(0, 2, 1, 3)
-            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
-            offload_func_v(KQV_merged);
-            ggml_set_name(KQV_merged, "KQV_merged");
-
-            // cur = KQV_merged.contiguous().view(n_embd, N)
-            cur = ggml_cpy(ctx0,
-                    KQV_merged,
-                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
-            offload_func_v(cur);
-            ggml_set_name(cur, "KQV_merged_contiguous");
-
-            // projection (no bias)
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].wo,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_wo");
-        }
-
-        lctx.use_buf(ctx0, 1);
-
-        struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpSA);
-        offload_func(inpFF);
-        ggml_set_name(inpFF, "inpFF");
-
-        // feed-forward network
-        {
-            // norm
-            {
-                cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps);
-                offload_func(cur);
-                ggml_set_name(cur, "rms_norm_1");
-
-                // cur = cur*ffn_norm(broadcasted)
-                cur = ggml_mul(ctx0, cur, model.layers[il].ffn_norm);
-                offload_func(cur);
-                ggml_set_name(cur, "ffn_norm");
-            }
-
-            struct ggml_tensor * tmp = ggml_mul_mat(ctx0,
-                    model.layers[il].w3,
-                    cur);
-            offload_func(tmp);
-            ggml_set_name(tmp, "result_w3");
-
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].w1,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_w1");
-
-            // SILU activation
-            cur = ggml_silu(ctx0, cur);
-            offload_func(cur);
-            ggml_set_name(cur, "silu");
-
-            cur = ggml_mul(ctx0, cur, tmp);
-            offload_func(cur);
-            ggml_set_name(cur, "silu_x_result_w3");
-
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].w2,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_w2");
-        }
-
-        cur = ggml_add(ctx0, cur, inpFF);
-        offload_func(cur);
-        ggml_set_name(cur, "inpFF_+_result_w2");
-
-        // input for next layer
-        inpL = cur;
-    }
-
-    lctx.use_buf(ctx0, 0);
-
-    // used at the end to optionally extract the embeddings
-    struct ggml_tensor * embeddings = NULL;
-
-    // norm
-    {
-        cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
-        offload_func_nr(cur);
-        ggml_set_name(cur, "rms_norm_2");
-
-        // cur = cur*norm(broadcasted)
-        cur = ggml_mul(ctx0, cur, model.norm);
-        // offload_func_nr(cur); // TODO CPU + GPU mirrored backend
-        ggml_set_name(cur, "result_norm");
-
-        embeddings = cur;
-    }
-
-    // lm_head
-    cur = ggml_mul_mat(ctx0, model.output, cur);
-    ggml_set_name(cur, "result_output");
-
-    lctx.use_buf(ctx0, -1);
-
-    // logits -> probs
-    //cur = ggml_soft_max_inplace(ctx0, cur);
-
-    // run the computation
-    ggml_build_forward_expand(gf, cur);
-
-    // fprintf(stderr, "graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf.n_nodes, gf.n_leafs);
-
-#if GGML_USE_MPI
-    ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);
-#endif
-
-#ifdef GGML_USE_METAL
-    if (lctx.ctx_metal && N == 1) {
-        if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
-            ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
-        }
-        ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
-        ggml_metal_graph_compute(lctx.ctx_metal, gf);
-        ggml_metal_get_tensor   (lctx.ctx_metal, cur);
-    } else {
-        // IMPORTANT:
-        // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
-        // ggml_graph_compute(). It uses Apple's Accelerate CBLAS API which takes advantage of the ANE or the AMX
-        // coprocessor.
-        //
-        // When we implement Matrix x Matrix Metal multiplication, we can avoid this branch.
-        // But for now, we have focused only on Matrix x Vector Metal multiplication.
-        //
-        // TODO: avoid these syncs via shared memory (ref #1696)
-        //
-        if (lctx.ctx_metal) {
-            // We need to sync the GPU KV cache with the CPU KV cache
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.k);
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.v);
-        }
-
-        ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
-    }
-#else
-    ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
-#endif
-
-#if GGML_USE_MPI
-    ggml_mpi_graph_compute_post(lctx.ctx_mpi, gf, n_layer);
-#endif
-
-    // update kv token count
-    lctx.kv_self.n = n_past + N;
-
-    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
-
-    if (cgraph_fname) {
-        ggml_graph_export(gf, cgraph_fname);
-    }
-
-#ifdef GGML_PERF
-    // print timing information per ggml operation (for debugging purposes)
-    // requires GGML_PERF to be defined
-    ggml_graph_print(gf);
-#endif
-
-    // plot the computation graph in dot format (for debugging purposes)
-    //if (n_past%100 == 0) {
-    //    ggml_graph_dump_dot(gf, NULL, "llama.dot");
-    //}
-
-    // extract logits
-    {
-        auto & logits_out = lctx.logits;
-
-        if (lctx.logits_all) {
-            logits_out.resize(n_vocab * N);
-            memcpy(logits_out.data(), (float *) ggml_get_data(res), sizeof(float)*n_vocab*N);
-        } else {
-            // return result for just the last token
-            logits_out.resize(n_vocab);
-            memcpy(logits_out.data(), (float *) ggml_get_data(res) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
-        }
-    }
-
-    // extract embeddings
-    if (!lctx.embedding.empty()) {
-        auto & embedding_out = lctx.embedding;
-
-        embedding_out.resize(n_embd);
-        memcpy(embedding_out.data(), (float *) ggml_get_data(embeddings) + (n_embd*(N - 1)), sizeof(float)*n_embd);
-    }
-
-    if (mem_per_token == 0) {
-        mem_per_token = ggml_used_mem(ctx0)/N;
-    }
-
-#if 0
-    printf("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
-            ggml_used_mem(ctx0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(1)/1024.0/1024.0,
-            lctx.work_buffer.size()/1024.0/1024.0,
-            n_past, N);
-#endif
-
-    ggml_free(ctx0);
-
-    // measure the performance only for the single-token evals
-    if (N == 1) {
-        lctx.t_eval_us += ggml_time_us() - t_start_us;
-        lctx.n_eval++;
-    }
-    else if (N > 1) {
-        lctx.t_p_eval_us += ggml_time_us() - t_start_us;
-        lctx.n_p_eval += N;
-    }
-
-    return true;
-}
-
-//
-// tokenizer
-//
-
-static size_t utf8_len(char src) {
-    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
-    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
-    return lookup[highbits];
-}
-
-struct llama_sp_symbol {
-    using index = int;
-    index prev;
-    index next;
-    const char * text;
-    size_t n;
-};
-
-static_assert(std::is_trivially_copyable<llama_sp_symbol>::value, "llama_sp_symbol is not trivially copyable");
-
-struct llama_sp_bigram {
-    struct comparator {
-        bool operator()(llama_sp_bigram & l, llama_sp_bigram & r) {
-            return (l.score < r.score) || (l.score == r.score && l.left > r.left);
-        }
-    };
-    using queue_storage = std::vector<llama_sp_bigram>;
-    using queue = std::priority_queue<llama_sp_bigram, queue_storage, comparator>;
-    llama_sp_symbol::index left;
-    llama_sp_symbol::index right;
-    float score;
-    size_t size;
-};
-
-// original implementation:
-// https://github.com/ggerganov/llama.cpp/commit/074bea2eb1f1349a0118239c4152914aecaa1be4
-struct llama_tokenizer {
-    llama_tokenizer(const llama_vocab & vocab): vocab_(vocab) {}
-
-    void tokenize(const std::string & text, std::vector<llama_vocab::id> & output) {
-        // split string into utf8 chars
-        int index = 0;
-        size_t offs = 0;
-        while (offs < text.size()) {
-            llama_sp_symbol sym;
-            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
-            sym.text = text.c_str() + offs;
-            sym.n = char_len;
-            offs += char_len;
-            sym.prev = index - 1;
-            sym.next = offs == text.size() ? -1 : index + 1;
-            index++;
-            symbols_.emplace_back(sym);
-        }
-
-        // seed the work queue with all possible 2-character tokens.
-        for (size_t i = 1; i < symbols_.size(); ++i) {
-            try_add_bigram(i - 1, i);
-        }
-
-        // keep substituting the highest frequency pairs for as long as we can.
-        while (!work_queue_.empty()) {
-            auto bigram = work_queue_.top();
-            work_queue_.pop();
-
-            auto & left_sym = symbols_[bigram.left];
-            auto & right_sym = symbols_[bigram.right];
-
-            // if one of the symbols already got merged, skip it.
-            if (left_sym.n == 0 || right_sym.n == 0 ||
-                left_sym.n + right_sym.n != bigram.size) {
-                continue;
-            }
-
-            // merge the right sym into the left one
-            left_sym.n += right_sym.n;
-            right_sym.n = 0;
-
-            //printf("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
-
-            // remove the right sym from the chain
-            left_sym.next = right_sym.next;
-            if (right_sym.next >= 0) {
-                symbols_[right_sym.next].prev = bigram.left;
-            }
-
-            // find more substitutions
-            try_add_bigram(left_sym.prev, bigram.left);
-            try_add_bigram(bigram.left, left_sym.next);
-        }
-
-        for (int i = 0; i != -1; i = symbols_[i].next) {
-            auto & symbol = symbols_[i];
-            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
-
-            if (token == vocab_.token_to_id.end()) {
-                // output any symbols that did not form tokens as bytes.
-                for (int j = 0; j < (int) symbol.n; ++j) {
-                    llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
-                    output.push_back(token_id);
-                }
-            } else {
-                output.push_back((*token).second);
-            }
-        }
-    }
-
-private:
-    void try_add_bigram(int left, int right) {
-        if (left == -1 || right == -1) {
-            return;
-        }
-
-        const std::string text = std::string(symbols_[left].text, symbols_[left].n + symbols_[right].n);
-        auto token = vocab_.token_to_id.find(text);
-
-        if (token == vocab_.token_to_id.end()) {
-            return;
-        }
-
-        if (static_cast<size_t>((*token).second) >= vocab_.id_to_token.size()) {
-            return;
-        }
-
-        const auto &tok_score = vocab_.id_to_token[(*token).second];
-
-        llama_sp_bigram bigram;
-        bigram.left = left;
-        bigram.right = right;
-        bigram.score = tok_score.score;
-        bigram.size = text.size();
-        work_queue_.push(bigram);
-    }
-
-    const llama_vocab & vocab_;
-    std::vector<llama_sp_symbol> symbols_;
-    llama_sp_bigram::queue work_queue_;
-};
-
-static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
-    llama_tokenizer tokenizer(vocab);
-    std::vector<llama_vocab::id> output;
-
-    if (text.empty()) {
-        return output;
-    }
-
-    if (bos) {
-        output.push_back(llama_token_bos());
-    }
-
-    tokenizer.tokenize(text, output);
-    return output;
-}
-
-//
-// grammar - internal
-//
-
-struct llama_grammar {
-    const std::vector<std::vector<llama_grammar_element>>   rules;
-    std::vector<std::vector<const llama_grammar_element *>> stacks;
-};
-
-struct llama_grammar_candidate {
-    size_t           index;
-    const uint32_t * code_points;
-};
-
-// NOTE: assumes valid utf8 (but checks for overrun)
-// adds a terminating 0 for use as pointer
-std::vector<uint32_t> decode_utf8(const char * src) {
-    static const int      lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
-    const char          * pos      = src;
-    std::vector<uint32_t> code_points;
-    while (*pos != 0) {
-        uint8_t  first_byte = static_cast<uint8_t>(*pos);
-        uint8_t  highbits   = first_byte >> 4;
-        int      len        = lookup[highbits];
-        uint8_t  mask       = (1 << (8 - len)) - 1;
-        uint32_t value      = first_byte & mask;
-        const char * end    = pos + len; // may overrun!
-        ++pos;
-        for ( ; pos < end && *pos != 0; ++pos) {
-            value = (value << 6) + (static_cast<uint8_t>(*pos) & 0x3F);
-        }
-        code_points.push_back(value);
-    }
-    code_points.push_back(0);
-    return code_points;
-}
-
-// returns true iff pos points to the end of one of the definitions of a rule
-static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
-    switch (pos->type) {
-        case LLAMA_GRETYPE_END: return true;
-        case LLAMA_GRETYPE_ALT: return true;
-        default:                return false;
-    }
-}
-
-// returns true iff chr satisfies the char range at pos (regular or inverse range)
-// asserts that pos is pointing to a char range element
-static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
-        const llama_grammar_element * pos,
-        const uint32_t                chr) {
-
-    bool found            = false;
-    bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
-    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT);
-
-    do {
-        if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
-            // inclusive range, e.g. [a-z]
-            found = found || (pos->value <= chr && chr <= pos[1].value);
-            pos += 2;
-        } else {
-            // exact char match, e.g. [a] or "a"
-            found = found || pos->value == chr;
-            pos += 1;
-        }
-    } while (pos->type == LLAMA_GRETYPE_CHAR_ALT);
-
-    return std::make_pair(found == is_positive_char, pos);
-}
-
-// transforms a grammar pushdown stack into N possible stacks, all ending
-// at a character range (terminal element)
-static void llama_grammar_advance_stack(
-        const std::vector<std::vector<llama_grammar_element>>   & rules,
-        const std::vector<const llama_grammar_element *>        & stack,
-        std::vector<std::vector<const llama_grammar_element *>> & new_stacks) {
-
-    if (stack.empty()) {
-        new_stacks.push_back(stack);
-        return;
-    }
-
-    const llama_grammar_element * pos = stack.back();
-
-    switch (pos->type) {
-        case LLAMA_GRETYPE_RULE_REF: {
-            const size_t                  rule_id = static_cast<size_t>(pos->value);
-            const llama_grammar_element * subpos  = rules[rule_id].data();
-            do {
-                // init new stack without the top (pos)
-                std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
-                if (!llama_grammar_is_end_of_sequence(pos + 1)) {
-                    // if this rule ref is followed by another element, add that to stack
-                    new_stack.push_back(pos + 1);
-                }
-                if (!llama_grammar_is_end_of_sequence(subpos)) {
-                    // if alternate is nonempty, add to stack
-                    new_stack.push_back(subpos);
-                }
-                llama_grammar_advance_stack(rules, new_stack, new_stacks);
-                while (!llama_grammar_is_end_of_sequence(subpos)) {
-                    // scan to end of alternate def
-                    subpos++;
-                }
-                if (subpos->type == LLAMA_GRETYPE_ALT) {
-                    // there's another alternate def of this rule to process
-                    subpos++;
-                } else {
-                    break;
-                }
-            } while (true);
-            break;
-        }
-        case LLAMA_GRETYPE_CHAR:
-        case LLAMA_GRETYPE_CHAR_NOT:
-            new_stacks.push_back(stack);
-            break;
-        default:
-            // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range
-            // (LLAMA_GRETYPE_CHAR_ALT, LLAMA_GRETYPE_CHAR_RNG_UPPER); stack should never be left on
-            // those
-            GGML_ASSERT(false);
-    }
-}
-
-// takes a set of possible pushdown stacks on a grammar, which are required to
-// be positioned at a character range (see `llama_grammar_advance_stack`), and
-// produces the N possible stacks if the given char is accepted at those
-// positions
-static std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const uint32_t                                                  chr) {
-
-    std::vector<std::vector<const llama_grammar_element *>> new_stacks;
-
-    for (const auto & stack : stacks) {
-        if (stack.empty()) {
-            continue;
-        }
-
-        auto match = llama_grammar_match_char(stack.back(), chr);
-        if (match.first) {
-            const llama_grammar_element * pos = match.second;
-
-            // update top of stack to next element, if any
-            std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
-            if (!llama_grammar_is_end_of_sequence(pos)) {
-                new_stack.push_back(pos);
-            }
-            llama_grammar_advance_stack(rules, new_stack, new_stacks);
-        }
-    }
-
-    return new_stacks;
-}
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const std::vector<llama_grammar_candidate>                    & candidates);
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_stack(
-        const std::vector<std::vector<llama_grammar_element>> & rules,
-        const std::vector<const llama_grammar_element *>      & stack,
-        const std::vector<llama_grammar_candidate>            & candidates) {
-
-    std::vector<llama_grammar_candidate> rejects;
-
-    if (stack.empty()) {
-        // accept nothing; EOS is handled elsewhere
-        rejects.insert(rejects.end(), candidates.begin(), candidates.end());
-        return rejects;
-    }
-
-    const llama_grammar_element * stack_pos = stack.back();
-
-    std::vector<llama_grammar_candidate> next_candidates;
-    for (auto tok : candidates) {
-        if (llama_grammar_match_char(stack_pos, tok.code_points[0]).first) {
-            if (tok.code_points[1] != 0) {
-                next_candidates.push_back({ tok.index, tok.code_points + 1 });
-            }
-        } else {
-            rejects.push_back(tok);
-        }
-    }
-
-    auto stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
-
-    // update top of stack to next element, if any
-    std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
-    if (!llama_grammar_is_end_of_sequence(stack_pos_after)) {
-        stack_after.push_back(stack_pos_after);
-    }
-    std::vector<std::vector<const llama_grammar_element *>> next_stacks;
-    llama_grammar_advance_stack(rules, stack_after, next_stacks);
-
-    auto next_rejects = llama_grammar_reject_candidates(rules, next_stacks, next_candidates);
-    for (auto tok : next_rejects) {
-        rejects.push_back({ tok.index, tok.code_points - 1 });
-    }
-
-    return rejects;
-}
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const std::vector<llama_grammar_candidate>                    & candidates) {
-    GGML_ASSERT(!stacks.empty()); // REVIEW
-
-    if (candidates.empty()) {
-        return std::vector<llama_grammar_candidate>();
-    }
-
-    auto rejects = llama_grammar_reject_candidates_for_stack(rules, stacks.front(), candidates);
-
-    for (size_t i = 1, size = stacks.size(); i < size; ++i) {
-        rejects = llama_grammar_reject_candidates_for_stack(rules, stacks[i], rejects);
-    }
-    return rejects;
-}
-
-//
-// grammar - external
-//
-
-struct llama_grammar * llama_grammar_init(
-            const llama_grammar_element ** rules,
-                                 size_t    n_rules,
-                                 size_t    start_rule_index) {
-    const llama_grammar_element * pos;
-
-    // copy rule definitions into vectors
-    std::vector<std::vector<llama_grammar_element>> vec_rules(n_rules);
-    for (size_t i = 0; i < n_rules; i++) {
-        for (pos = rules[i]; pos->type != LLAMA_GRETYPE_END; pos++) {
-            vec_rules[i].push_back(*pos);
-        }
-        vec_rules[i].push_back({LLAMA_GRETYPE_END, 0});
-    }
-
-    // loop over alternates of start rule to build initial stacks
-    std::vector<std::vector<const llama_grammar_element *>> stacks;
-    pos = rules[start_rule_index];
-    do {
-        std::vector<const llama_grammar_element *> stack;
-        if (!llama_grammar_is_end_of_sequence(pos)) {
-            // if alternate is nonempty, add to stack
-            stack.push_back(pos);
-        }
-        llama_grammar_advance_stack(vec_rules, stack, stacks);
-        while (!llama_grammar_is_end_of_sequence(pos)) {
-            // scan to end of alternate def
-            pos++;
-        }
-        if (pos->type == LLAMA_GRETYPE_ALT) {
-            // there's another alternate def of this rule to process
-            pos++;
-        } else {
-            break;
-        }
-    } while (true);
-
-    return new llama_grammar{ std::move(vec_rules), std::move(stacks) };
-}
-
-void llama_grammar_free(struct llama_grammar * grammar) {
-    delete grammar;
-}
-
-//
-// sampling
-//
-
-void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(candidates->size > 0);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Sort the logits in descending order
-    if (!candidates->sorted) {
-        std::sort(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
-            return a.logit > b.logit;
-        });
-        candidates->sorted = true;
-    }
-
-    float max_l = candidates->data[0].logit;
-    float cum_sum = 0.0f;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        float p = expf(candidates->data[i].logit - max_l);
-        candidates->data[i].p = p;
-        cum_sum += p;
-    }
-    for (size_t i = 0; i < candidates->size; ++i) {
-        candidates->data[i].p /= cum_sum;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    k = std::max(k, (int) min_keep);
-    k = std::min(k, (int) candidates->size);
-
-    // Sort scores in descending order
-    if (!candidates->sorted) {
-        auto comp = [](const llama_token_data & a, const llama_token_data & b) {
-            return a.logit > b.logit;
-        };
-        if (k == (int) candidates->size) {
-            std::sort(candidates->data, candidates->data + candidates->size, comp);
-        } else {
-            std::partial_sort(candidates->data, candidates->data + k, candidates->data + candidates->size, comp);
-        }
-        candidates->sorted = true;
-    }
-    candidates->size = k;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
-    if (p >= 1.0f) {
-        return;
-    }
-
-    llama_sample_softmax(ctx, candidates);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Compute the cumulative probabilities
-    float cum_sum = 0.0f;
-    size_t last_idx = candidates->size;
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        cum_sum += candidates->data[i].p;
-
-        // Check if the running sum is at least p or if we have kept at least min_keep tokens
-        // we set the last index to i+1 to indicate that the current iterate should be included in the set
-        if (cum_sum >= p && i + 1 >= min_keep) {
-            last_idx = i + 1;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the top-p tokens
-    candidates->size = last_idx;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep) {
-    if (z >= 1.0f || candidates->size <= 2) {
-        return;
-    }
-
-    llama_sample_softmax(nullptr, candidates);
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Compute the first and second derivatives
-    std::vector<float> first_derivatives(candidates->size - 1);
-    std::vector<float> second_derivatives(candidates->size - 2);
-
-    for (size_t i = 0; i < first_derivatives.size(); ++i) {
-        first_derivatives[i] = candidates->data[i].p - candidates->data[i + 1].p;
-    }
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        second_derivatives[i] = first_derivatives[i] - first_derivatives[i + 1];
-    }
-
-    // Calculate absolute value of second derivatives
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        second_derivatives[i] = abs(second_derivatives[i]);
-    }
-
-    // Normalize the second derivatives
-    {
-        const float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
-
-        if (second_derivatives_sum > 1e-6f) {
-            for (float & value : second_derivatives) {
-                value /= second_derivatives_sum;
-            }
-        } else {
-            for (float & value : second_derivatives) {
-                value = 1.0f / second_derivatives.size();
-            }
-        }
-    }
-
-    float cum_sum = 0.0f;
-    size_t last_idx = candidates->size;
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        cum_sum += second_derivatives[i];
-
-        // Check if the running sum is greater than z or if we have kept at least min_keep tokens
-        if (cum_sum > z && i >= min_keep) {
-            last_idx = i;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the tokens above the tail location
-    candidates->size = last_idx;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-
-void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
-    // Reference implementation:
-    // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
-    if (p >= 1.0f) {
-        return;
-    }
-
-    // Compute the softmax of logits and calculate entropy
-    llama_sample_softmax(nullptr, candidates);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    float entropy = 0.0f;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        entropy += -candidates->data[i].p * logf(candidates->data[i].p);
-    }
-
-    // Compute the absolute difference between negative log probability and entropy for each candidate
-    std::vector<float> shifted_scores;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        float shifted_score = fabsf(-logf(candidates->data[i].p) - entropy);
-        shifted_scores.push_back(shifted_score);
-    }
-
-    // Sort tokens based on the shifted_scores and their corresponding indices
-    std::vector<size_t> indices(candidates->size);
-    std::iota(indices.begin(), indices.end(), 0);
-
-    std::sort(indices.begin(), indices.end(), [&](size_t a, size_t b) {
-        return shifted_scores[a] < shifted_scores[b];
-    });
-
-    // Compute the cumulative probabilities
-    float cum_sum = 0.0f;
-    size_t last_idx = indices.size();
-
-    for (size_t i = 0; i < indices.size(); ++i) {
-        size_t idx = indices[i];
-        cum_sum += candidates->data[idx].p;
-
-        // Check if the running sum is greater than typical or if we have kept at least min_keep tokens
-        if (cum_sum > p && i >= min_keep - 1) {
-            last_idx = i + 1;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the locally typical tokens
-    std::vector<llama_token_data> new_candidates;
-    for (size_t i = 0; i < last_idx; ++i) {
-        size_t idx = indices[i];
-        new_candidates.push_back(candidates->data[idx]);
-    }
-
-    // Replace the data in candidates with the new_candidates data
-    std::copy(new_candidates.begin(), new_candidates.end(), candidates->data);
-    candidates->size = new_candidates.size();
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates_p, float temp) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    for (size_t i = 0; i < candidates_p->size; ++i) {
-        candidates_p->data[i].logit /= temp;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty) {
-    if (last_tokens_size == 0 || penalty == 1.0f) {
-        return;
-    }
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        const auto * token_iter = std::find(last_tokens, last_tokens + last_tokens_size, candidates->data[i].id);
-        if (token_iter == last_tokens + last_tokens_size) {
-            continue;
-        }
-
-        // The academic publication that described this technique actually just only divided, but that would cause tokens with negative logits to become more likely, which is obviously wrong.
-        // This is common fix for this problem, which is to multiply by the penalty instead of dividing.
-        if (candidates->data[i].logit <= 0) {
-            candidates->data[i].logit *= penalty;
-        } else {
-            candidates->data[i].logit /= penalty;
-        }
-    }
-
-    candidates->sorted = false;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens_p, size_t last_tokens_size, float alpha_frequency, float alpha_presence) {
-    if (last_tokens_size == 0 || (alpha_frequency == 0.0f && alpha_presence == 0.0f)) {
-        return;
-    }
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Create a frequency map to count occurrences of each token in last_tokens
-    std::unordered_map<llama_token, int> token_count;
-    for (size_t i = 0; i < last_tokens_size; ++i) {
-        token_count[last_tokens_p[i]]++;
-    }
-
-    // Apply frequency and presence penalties to the candidates
-    for (size_t i = 0; i < candidates->size; ++i) {
-        auto token_iter = token_count.find(candidates->data[i].id);
-        if (token_iter == token_count.end()) {
-            continue;
-        }
-
-        int count = token_iter->second;
-        candidates->data[i].logit -= float(count) * alpha_frequency + float(count > 0) * alpha_presence;
-    }
-
-    candidates->sorted = false;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
-    assert(ctx);
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    bool allow_eos = false;
-    for (const auto & stack : grammar->stacks) {
-        if (stack.empty()) {
-            allow_eos = true;
-            break;
-        }
-    }
-
-    const llama_token eos = llama_token_eos();
-
-    std::vector<std::vector<uint32_t>>   candidates_decoded;
-    std::vector<llama_grammar_candidate> candidates_grammar;
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        const llama_token id  = candidates->data[i].id;
-        const char *      str = llama_token_to_str(ctx, id);
-        if (id == eos) {
-            if (!allow_eos) {
-                candidates->data[i].logit = -INFINITY;
-            }
-        } else if (*str == 0) {
-            candidates->data[i].logit = -INFINITY;
-        } else {
-            candidates_decoded.push_back(decode_utf8(str));
-            candidates_grammar.push_back({ i, candidates_decoded.back().data() });
-        }
-    }
-
-    const auto rejects =
-        llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
-    for (auto & reject : rejects) {
-        candidates->data[reject.index].logit = -INFINITY;
-    }
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-}
-
-static void llama_log_softmax(float * array, size_t size) {
-    float max_l = *std::max_element(array, array + size);
-    float sum = 0.f;
-    for (size_t i = 0; i < size; ++i) {
-        float p = expf(array[i] - max_l);
-        sum += p;
-        array[i] = p;
-    }
-
-    for (size_t i = 0; i < size; ++i) {
-        array[i] = logf(array[i] / sum);
-    }
-}
-
-void llama_sample_classifier_free_guidance(
-          struct llama_context * ctx,
-        llama_token_data_array * candidates,
-          struct llama_context * guidance_ctx,
-                         float   scale) {
-    int64_t t_start_sample_us = ggml_time_us();
-
-    assert(ctx);
-    auto n_vocab = llama_n_vocab(ctx);
-    assert(n_vocab == (int)candidates->size);
-    assert(!candidates->sorted);
-
-    std::vector<float> logits_base;
-    logits_base.reserve(candidates->size);
-    for (size_t i = 0; i < candidates->size; ++i) {
-        logits_base.push_back(candidates->data[i].logit);
-    }
-    llama_log_softmax(logits_base.data(), candidates->size);
-
-    float* logits_guidance = llama_get_logits(guidance_ctx);
-    llama_log_softmax(logits_guidance, n_vocab);
-
-    for (int i = 0; i < n_vocab; ++i) {
-        float logit_guidance = logits_guidance[i];
-        float logit_base = logits_base[i];
-        candidates->data[i].logit = scale * (logit_base - logit_guidance) + logit_guidance;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu) {
-    assert(ctx);
-    auto N = float(llama_n_vocab(ctx));
-    int64_t t_start_sample_us;
-    t_start_sample_us = ggml_time_us();
-
-    llama_sample_softmax(nullptr, candidates);
-
-    // Estimate s_hat using the most probable m tokens
-    float s_hat = 0.0;
-    float sum_ti_bi = 0.0;
-    float sum_ti_sq = 0.0;
-    for (size_t i = 0; i < size_t(m - 1) && i < candidates->size - 1; ++i) {
-        float t_i = logf(float(i + 2) / float(i + 1));
-        float b_i = logf(candidates->data[i].p / candidates->data[i + 1].p);
-        sum_ti_bi += t_i * b_i;
-        sum_ti_sq += t_i * t_i;
-    }
-    s_hat = sum_ti_bi / sum_ti_sq;
-
-    // Compute k from the estimated s_hat and target surprise value
-    float epsilon_hat = s_hat - 1;
-    float k = powf((epsilon_hat * powf(2, *mu)) / (1 - powf(N, -epsilon_hat)), 1 / s_hat);
-
-    // Sample the next word X using top-k sampling
-    llama_sample_top_k(nullptr, candidates, int(k), 1);
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    llama_token X = llama_sample_token(ctx, candidates);
-    t_start_sample_us = ggml_time_us();
-
-    // Compute error as the difference between observed surprise and target surprise value
-    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return candidate.id == X;
-    }));
-    float observed_surprise = -log2f(candidates->data[X_idx].p);
-    float e = observed_surprise - tau;
-
-    // Update mu using the learning rate and error
-    *mu = *mu - eta * e;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    return X;
-}
-
-llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu) {
-    int64_t t_start_sample_us;
-    t_start_sample_us = ggml_time_us();
-
-    llama_sample_softmax(ctx, candidates);
-
-    // Truncate the words with surprise values greater than mu
-    candidates->size = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return -log2f(candidate.p) > *mu;
-    }));
-
-    if (candidates->size == 0) {
-        candidates->size = 1;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-
-    // Normalize the probabilities of the remaining words
-    llama_sample_softmax(ctx, candidates);
-
-    // Sample the next word X from the remaining words
-    llama_token X = llama_sample_token(ctx, candidates);
-    t_start_sample_us = ggml_time_us();
-
-    // Compute error as the difference between observed surprise and target surprise value
-    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return candidate.id == X;
-    }));
-    float observed_surprise = -log2f(candidates->data[X_idx].p);
-    float e = observed_surprise - tau;
-
-    // Update mu using the learning rate and error
-    *mu = *mu - eta * e;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    return X;
-}
-
-llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Find max element
-    auto * max_iter = std::max_element(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
-        return a.logit < b.logit;
-    });
-
-    llama_token result = max_iter->id;
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-        ctx->n_sample++;
-    }
-    return result;
-}
-
-llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(ctx);
-    const int64_t t_start_sample_us = ggml_time_us();
-    llama_sample_softmax(nullptr, candidates);
-
-    std::vector<float> probs;
-    probs.reserve(candidates->size);
-    for (size_t i = 0; i < candidates->size; ++i) {
-        probs.push_back(candidates->data[i].p);
-    }
-
-    std::discrete_distribution<> dist(probs.begin(), probs.end());
-    auto & rng = ctx->rng;
-    int idx = dist(rng);
-
-    llama_token result = candidates->data[idx].id;
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    ctx->n_sample++;
-    return result;
-}
-
-void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    if (token == llama_token_eos()) {
-        for (const auto & stack : grammar->stacks) {
-            if (stack.empty()) {
-                return;
-            }
-        }
-        GGML_ASSERT(false);
-    }
-
-    const char * str = llama_token_to_str(ctx, token);
-    // Note terminating 0 in decoded string
-    auto code_points = decode_utf8(str);
-    for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
-        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
-    }
-    GGML_ASSERT(!grammar->stacks.empty());
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-}
-
-//
-// quantization
-//
-
-static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
-    if (output.size < nelements * sizeof(float)) {
-        output.resize(nelements * sizeof(float));
-    }
-    float * f32_output = (float *) output.addr;
-
-    ggml_type_traits_t qtype;
-    if (ggml_is_quantized(tensor.type)) {
-        qtype = ggml_internal_get_type_traits(tensor.type);
-        if (qtype.to_float == NULL) {
-            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor.type)));
-        }
-    } else if (tensor.type != GGML_TYPE_F16) {
-        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor.type)));
-    }
-
-    if (nthread < 2) {
-        if (tensor.type == GGML_TYPE_F16) {
-            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor.data, f32_output, nelements);
-        } else if (ggml_is_quantized(tensor.type)) {
-            qtype.to_float(tensor.data, f32_output, nelements);
-        } else {
-            GGML_ASSERT(false); // unreachable
-        }
-        return;
-    }
-
-    auto block_size = tensor.type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor.type);
-    auto block_size_bytes = ggml_type_size(tensor.type);
-
-    GGML_ASSERT(nelements % block_size == 0);
-    auto nblocks = nelements / block_size;
-    auto blocks_per_thread = nblocks / nthread;
-    auto spare_blocks = nblocks - (blocks_per_thread * nthread); // if blocks aren't divisible by thread count
-
-    std::vector<std::thread> workers;
-    for (auto tnum = 0, in_buff_offs = 0, out_buff_offs = 0; tnum < nthread; tnum++) {
-        auto thr_blocks = blocks_per_thread + (tnum == nthread - 1 ? spare_blocks : 0); // num blocks for this thread
-        auto thr_elems = thr_blocks * block_size; // number of elements for this thread
-        auto thr_block_bytes = thr_blocks * block_size_bytes; // number of input bytes for this thread
-
-        auto compute = [qtype] (ggml_type typ, uint8_t * inbuf, float * outbuf, int nels) {
-            if (typ == GGML_TYPE_F16) {
-                ggml_fp16_to_fp32_row((ggml_fp16_t *)inbuf, outbuf, nels);
-            } else {
-                qtype.to_float(inbuf, outbuf, nels);
-            }
-        };
-        workers.push_back(std::thread(compute, tensor.type, tensor.data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
-        in_buff_offs += thr_block_bytes;
-        out_buff_offs += thr_elems;
-    }
-    for (auto & worker : workers) {
-        worker.join();
-    }
-
-}
-
-static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
-    ggml_type quantized_type;
-    llama_ftype ftype = params->ftype;
-    int nthread = params->nthread;
-
-    switch (params->ftype) {
-        case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
-        case LLAMA_FTYPE_MOSTLY_Q4_1: quantized_type = GGML_TYPE_Q4_1; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_0: quantized_type = GGML_TYPE_Q5_0; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_1: quantized_type = GGML_TYPE_Q5_1; break;
-        case LLAMA_FTYPE_MOSTLY_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;
-        case LLAMA_FTYPE_MOSTLY_F16:  quantized_type = GGML_TYPE_F16;  break;
-        case LLAMA_FTYPE_ALL_F32:     quantized_type = GGML_TYPE_F32;  break;
-
-#ifdef GGML_USE_K_QUANTS
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K:   quantized_type = GGML_TYPE_Q2_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M:
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: quantized_type = GGML_TYPE_Q3_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: quantized_type = GGML_TYPE_Q4_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q6_K:   quantized_type = GGML_TYPE_Q6_K; break;
-#endif
-        default: throw std::runtime_error(format("invalid output file type %d\n", ftype));
-    }
-
-    if (nthread <= 0) {
-        nthread = std::thread::hardware_concurrency();
-    }
-
-    std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
-    gguf_file_saver file_saver(fname_out.c_str(), model_loader->file_loader.get(), params->ftype);
-
-#ifdef GGML_USE_K_QUANTS
-    int n_attention_wv    = 0;
-    int n_feed_forward_w2 = 0;
-    for (auto& tensor : model_loader->tensors_map.tensors) {
-        if (tensor.name.find("attention.wv.weight") != std::string::npos) {
-            ++n_attention_wv;
-        }
-        else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
-            ++n_feed_forward_w2;
-        }
-    }
-
-    int i_attention_wv = 0;
-    int i_feed_forward_w2 = 0;
-#endif
-
-    size_t total_size_org = 0;
-    size_t total_size_new = 0;
-    std::vector<int64_t> hist_all(1 << 4, 0);
-
-    std::vector<std::thread> workers;
-    std::mutex mutex;
-
-    auto use_more_bits = [] (int i_layer, int num_layers) -> bool {
-        return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
-    };
-
-    size_t idx = 0;
-    for (gguf_load_tensor & tensor : model_loader->tensors_map.tensors) {
-        gguf_buffer read_data;
-        read_data.resize(tensor.size);
-        tensor.data = read_data.addr;
-        model_loader->load_data_for(tensor);
-
-        printf("[%4zu/%4zu] %36s - %16s, type = %6s, ",
-               ++idx, model_loader->tensors_map.tensors.size(),
-               tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
-               ggml_type_name(tensor.type));
-
-        // This used to be a regex, but <regex> has an extreme cost to compile times.
-        bool quantize = tensor.name.rfind("weight") == tensor.name.size() - 6; // ends with 'weight'?
-
-        // quantize only 2D tensors
-        quantize &= (tensor.ne.size() == 2);
-        quantize &= params->quantize_output_tensor || tensor.name != "output.weight";
-        quantize &= quantized_type != tensor.type;
-
-        enum ggml_type new_type;
-        void * new_data;
-        size_t new_size;
-        gguf_buffer work;
-
-        if (!quantize) {
-            new_type = tensor.type;
-            new_data = tensor.data;
-            new_size = tensor.size;
-            printf("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
-        } else {
-            new_type = quantized_type;
-#ifdef GGML_USE_K_QUANTS
-            if (tensor.name == "output.weight") {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
-                if (nx % QK_K == 0 && ny % QK_K == 0) {
-                    new_type = GGML_TYPE_Q6_K;
-                }
-            } else if (tensor.name.find("attention.wv.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
-                        use_more_bits(i_attention_wv, n_attention_wv)) new_type = GGML_TYPE_Q6_K;
-                else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
-                        (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
-                ++i_attention_wv;
-            } else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
-                         use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
-                //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
-                ++i_feed_forward_w2;
-            } else if (tensor.name.find("attention.wo.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-            }
-            bool convert_incompatible_tensor = false;
-            if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
-                new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
-                if (nx % QK_K != 0 || ny % QK_K != 0) {
-                    fprintf(stderr, "\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
-                    convert_incompatible_tensor = true;
-                }
-            }
-            if (convert_incompatible_tensor) {
-                if (tensor.name == "output.weight") {
-                    new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
-                    fprintf(stderr, "F16 will be used for this tensor instead.\n");
-                } else if (tensor.name == "tok_embeddings.weight") {
-                    new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
-                    fprintf(stderr, "Q4_0 will be used for this tensor instead.\n");
-                } else {
-                    throw std::runtime_error("Unsupported tensor size encountered\n");
-                }
-            }
-#endif
-
-            float * f32_data;
-            size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
-            gguf_buffer f32_conv_buf;
-
-            if (tensor.type == GGML_TYPE_F32) {
-                f32_data = (float *) tensor.data;
-            } else if (ggml_is_quantized(tensor.type) && !params->allow_requantize) {
-                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
-            } else {
-                llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
-                f32_data = (float *) f32_conv_buf.addr;
-            }
-
-            printf("quantizing to %s .. ", ggml_type_name(new_type));
-            fflush(stdout);
-
-            work.resize(nelements * 4); // upper bound on size
-            new_data = work.addr;
-            std::vector<int64_t> hist_cur(1 << 4, 0);
-
-            int chunk_size = 32 * 512;
-            const int nchunk = (nelements + chunk_size - 1)/chunk_size;
-            const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
-            if (nthread_use < 2) {
-                new_size = ggml_quantize_chunk(new_type, f32_data, new_data, 0, nelements, hist_cur.data());
-            } else {
-                size_t counter = 0;
-                new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size] () {
-                    std::vector<int64_t> local_hist;
-                    size_t local_size = 0;
-                    while (true) {
-                        std::unique_lock<std::mutex> lock(mutex);
-                        size_t first = counter; counter += chunk_size;
-                        if (first >= nelements) {
-                            if (!local_hist.empty()) {
-                                for (int j=0; j<int(local_hist.size()); ++j) {
-                                    hist_cur[j] += local_hist[j];
-                                }
-                                new_size += local_size;
-                            }
-                            break;
-                        }
-                        lock.unlock();
-                        size_t last = std::min(nelements, first + chunk_size);
-                        if (local_hist.empty()) {
-                            local_hist.resize(hist_cur.size(), 0);
-                        }
-                        local_size += ggml_quantize_chunk(new_type, f32_data, new_data, first, last - first, local_hist.data());
-                    }
-                };
-                if ((int) workers.size() < nthread_use - 1) {
-                    workers.resize(nthread_use - 1);
-                }
-                for (int it = 0; it < nthread_use - 1; ++it) {
-                    workers[it] = std::thread(compute);
-                }
-                compute();
-                for (int it = 0; it < nthread_use - 1; ++it) {
-                    workers[it].join();
-                }
-            }
-
-            printf("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
-            int64_t tot_count = 0;
-            for (size_t i = 0; i < hist_cur.size(); i++) {
-                hist_all[i] += hist_cur[i];
-                tot_count += hist_cur[i];
-            }
-
-            if (tot_count > 0) {
-                for (size_t i = 0; i < hist_cur.size(); i++) {
-                    printf("%5.3f ", hist_cur[i] / float(nelements));
-                }
-            }
-            printf("\n");
-        }
-        total_size_org += tensor.size;
-        total_size_new += new_size;
-        file_saver.write_tensor(tensor, new_type, new_data, new_size);
-    }
-
-    printf("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
-    printf("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
-
-    {
-        int64_t sum_all = 0;
-        for (size_t i = 0; i < hist_all.size(); i++) {
-            sum_all += hist_all[i];
-        }
-
-        if (sum_all > 0) {
-            printf("%s: hist: ", __func__);
-            for (size_t i = 0; i < hist_all.size(); i++) {
-                printf("%5.3f ", hist_all[i] / float(sum_all));
-            }
-            printf("\n");
-        }
-    }
-}
-
-
-
-//
-// interface implementation
-//
-
-struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-    ggml_time_init();
-
-    llama_model * model = new llama_model;
-
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
-
-    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gqa, params.rms_norm_eps, params.n_gpu_layers,
-                params.main_gpu, params.tensor_split, params.rope_freq_base, params.rope_freq_scale,params.low_vram,
-                memory_type, params.use_mmap, params.use_mlock, params.vocab_only, params.progress_callback,
-                params.progress_callback_user_data)) {
-        delete model;
-        fprintf(stderr, "%s: failed to load model\n", __func__);
-        return nullptr;
-    }
-
-    return model;
-}
-
-void llama_free_model(struct llama_model * model) {
-    delete model;
-}
-
-struct llama_context * llama_new_context_with_model(
-                 struct llama_model * model,
-        struct llama_context_params   params) {
-
-    if (!model) {
-        return nullptr;
-    }
-
-    llama_context * ctx = new llama_context(*model);
-
-    if (params.seed == LLAMA_DEFAULT_SEED) {
-        params.seed = time(NULL);
-    }
-
-    unsigned cur_percentage = 0;
-    if (params.progress_callback == NULL) {
-        params.progress_callback_user_data = &cur_percentage;
-        params.progress_callback = [](float progress, void * ctx) {
-            unsigned * cur_percentage_p = (unsigned *) ctx;
-            unsigned percentage = (unsigned) (100 * progress);
-            while (percentage > *cur_percentage_p) {
-                *cur_percentage_p = percentage;
-                fprintf(stderr, ".");
-                fflush(stderr);
-                if (percentage >= 100) {
-                    fprintf(stderr, "\n");
-                }
-            }
-        };
-    }
-
-    ctx->rng = std::mt19937(params.seed);
-    ctx->logits_all = params.logits_all;
-
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
-
-    // reserve memory for context buffers
-    if (!params.vocab_only) {
-        if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
-            fprintf(stderr, "%s: kv_cache_init() failed for self-attention cache\n", __func__);
-            llama_free(ctx);
-            return nullptr;
-        }
-
-        {
-            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
-            fprintf(stderr, "%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
-        }
-
-        const auto & hparams = ctx->model.hparams;
-
-        // resized during inference
-        if (params.logits_all) {
-            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
-        } else {
-            ctx->logits.reserve(hparams.n_vocab);
-        }
-
-        if (params.embedding){
-            ctx->embedding.resize(hparams.n_embd);
-        }
-
-        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
-
-        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
-        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
-    }
-
-#ifdef GGML_USE_METAL
-    if (params.n_gpu_layers > 0) {
-        // this allocates all Metal resources and memory buffers
-        ctx->ctx_metal = ggml_metal_init(1);
-
-        void * data_ptr  = NULL;
-        size_t data_size = 0;
-
-        if (params.use_mmap) {
-            data_ptr  = ctx->model.mapping->addr;
-            data_size = ctx->model.mapping->size;
-        } else {
-            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
-            data_size = ggml_get_mem_size  (ctx->model.ctx);
-        }
-
-        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
-
-        fprintf(stderr, "%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
-
-#define LLAMA_METAL_CHECK_BUF(result)                                          \
-    if (!(result)) {                                                           \
-        fprintf(stderr, "%s: failed to add buffer\n", __func__);               \
-        llama_free(ctx);                                                       \
-        return NULL;                                                           \
-    }
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.addr, ctx->buf_compute.size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.addr, ctx->kv_self.buf.size, 0));
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].addr, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].addr, ctx->buf_scratch[1].size, 0));
-#undef LLAMA_METAL_CHECK_BUF
-    }
-#endif
-
-#ifdef GGML_USE_MPI
-    ctx->ctx_mpi = ggml_mpi_init();
-
-    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
-        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
-        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
-        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
-        llama_backend_free();
-        exit(1);
-    }
-#endif
-
-    return ctx;
-}
-
-struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-
-    struct llama_model * model = llama_load_model_from_file(path_model, params);
-    if (!model) {
-        return nullptr;
-    }
-    struct llama_context * ctx = llama_new_context_with_model(model, params);
-    ctx->model_owner = true;
-    return ctx;
-}
-
-void llama_free(struct llama_context * ctx) {
-    if (ctx->model_owner) {
-        delete &ctx->model;
-    }
-    delete ctx;
-}
-
-int llama_model_quantize(
-        const char * fname_inp,
-        const char * fname_out,
-        const llama_model_quantize_params *params) {
-    try {
-        llama_model_quantize_internal(fname_inp, fname_out, params);
-        return 0;
-    } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to quantize: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
-    fprintf(stderr, "%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
-
-    const int64_t t_start_lora_us = ggml_time_us();
-
-    auto fin = std::ifstream(path_lora, std::ios::binary);
-    if (!fin) {
-        fprintf(stderr, "%s: failed to open '%s'\n", __func__, path_lora);
-        return 1;
-    }
-
-    // verify magic and version
-    {
-        uint32_t magic;
-        fin.read((char *) &magic, sizeof(magic));
-        uint32_t format_version;
-        fin.read((char *) &format_version, sizeof(format_version));
-
-        if (format_version != 1) {
-            fprintf(stderr, "%s: unsupported file version\n", __func__ );
-            return 1;
-        }
-    }
-
-    int32_t lora_r;
-    int32_t lora_alpha;
-    fin.read((char *) &lora_r, sizeof(lora_r));
-    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
-    float scaling = (float)lora_alpha / (float)lora_r;
-
-    fprintf(stderr, "%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
-
-
-    // create a temporary ggml context to store the lora tensors
-    // todo: calculate size from biggest possible tensor
-    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
-    struct ggml_init_params params;
-    params.mem_size   = lora_buf.size();
-    params.mem_buffer = lora_buf.data();
-    params.no_alloc   = false;
-
-    ggml_context * lora_ctx = ggml_init(params);
-    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
-
-    // create a name -> tensor map of the model to accelerate lookups
-    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
-    for (const auto & kv: model.tensors_by_name) {
-        model_tensors.insert(kv);
-    }
-
-
-    // load base model
-    std::unique_ptr<llama_model_loader> model_loader;
-    ggml_context * base_ctx = NULL;
-    gguf_buffer base_buf;
-    if (path_base_model) {
-        fprintf(stderr, "%s: loading base model from '%s'\n", __func__, path_base_model);
-        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
-
-        size_t ctx_size;
-        size_t mmapped_size;
-        model_loader->calc_sizes(&ctx_size, &mmapped_size);
-        base_buf.resize(ctx_size);
-
-        ggml_init_params base_params;
-        base_params.mem_size   = base_buf.size;
-        base_params.mem_buffer = base_buf.addr;
-        base_params.no_alloc   = model_loader->use_mmap;
-
-        base_ctx = ggml_init(base_params);
-
-        model_loader->ggml_ctx = base_ctx;
-
-        // maybe this should in llama_model_loader
-        if (model_loader->use_mmap) {
-            model_loader->mapping.reset(new gguf_mmap(&model_loader->file_loader->file, /* prefetch */ 0, ggml_is_numa()));
-        }
-    }
-
-    // read tensors and apply
-    bool warned = false;
-    int n_tensors = 0;
-
-    std::vector<uint8_t> work_buffer;
-
-    while (true) {
-        int32_t n_dims;
-        int32_t length;
-        int32_t ftype;
-
-        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
-        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
-        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
-        if (fin.eof()) {
-            break;
-        }
-
-        int32_t ne[2] = { 1, 1 };
-        for (int i = 0; i < n_dims; ++i) {
-            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
-        }
-
-        std::string name;
-        {
-            char buf[1024];
-            fin.read(buf, length);
-            name = std::string(buf, length);
-        }
-
-        // check for lora suffix and get the type of tensor
-        const std::string lora_suffix = ".lora";
-        size_t pos = name.rfind(lora_suffix);
-        if (pos == std::string::npos) {
-            fprintf(stderr, "%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
-            return 1;
-        }
-
-        std::string lora_type = name.substr(pos + lora_suffix.length());
-        std::string base_name = name;
-        base_name.erase(pos);
-        // fprintf(stderr, "%s: %s => %s (lora type %s) ", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
-
-        if (model_tensors.find(base_name) == model_tensors.end()) {
-            fprintf(stderr, "%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
-            return 1;
-        }
-
-        // create ggml tensor
-        ggml_type wtype;
-        switch (ftype) {
-            case 0: wtype = GGML_TYPE_F32;  break;
-            case 1: wtype = GGML_TYPE_F16;  break;
-            default:
-                    {
-                        fprintf(stderr, "%s: invalid tensor data type '%d'\n",
-                                __func__, ftype);
-                        return false;
-                    }
-        }
-        ggml_tensor * lora_tensor;
-        if (n_dims == 2) {
-            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
-        }
-        else {
-            fprintf(stderr, "%s: unsupported tensor dimension %d\n", __func__, n_dims);
-            return 1;
-        }
-        ggml_set_name(lora_tensor, "lora_tensor");
-
-        // load tensor data
-        size_t offset = fin.tellg();
-        size_t tensor_data_size = ggml_nbytes(lora_tensor);
-        offset = (offset + 31) & -32;
-        fin.seekg(offset);
-        fin.read((char*)lora_tensor->data, tensor_data_size);
-
-        lora_tensors[name] = lora_tensor;
-
-        // check if we have both A and B tensors and apply
-        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
-            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
-
-            ggml_tensor * dest_t = model_tensors[base_name];
-
-            offload_func_t offload_func = llama_nop;
-            offload_func_t offload_func_force_inplace = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
-                if (dest_t->type != GGML_TYPE_F16) {
-                    throw std::runtime_error(format(
-                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
-                }
-                offload_func = ggml_cuda_assign_buffers;
-                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
-            }
-#endif // GGML_USE_CUBLAS
-
-            ggml_tensor * base_t;
-            if (model_loader) {
-                // load from base model
-                if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
-                    fprintf(stderr, "%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
-                    return 1;
-                }
-                size_t idx = model_loader->tensors_map.name_to_idx[base_name];
-                gguf_load_tensor & lt = model_loader->tensors_map.tensors[idx];
-                base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
-                lt.data = (uint8_t *) lt.ggml_tensor->data;
-                model_loader->load_data_for(lt);
-                lt.ggml_tensor->data = lt.data;
-            }
-            else {
-                base_t = dest_t;
-            }
-
-            if (ggml_is_quantized(base_t->type)) {
-                if (!warned) {
-                    fprintf(stderr, "%s: warning: using a lora adapter with a quantized model may result in poor quality, "
-                                    "use a f16 or f32 base model with --lora-base\n", __func__);
-                    warned = true;
-                }
-            }
-
-            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
-            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
-            ggml_set_name(loraA, "loraA");
-
-            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
-            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
-            ggml_set_name(loraB, "loraB");
-
-            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
-                fprintf(stderr, "%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
-                               " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
-                return 1;
-            }
-
-            // w = w + BA*s
-            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
-            offload_func(BA);
-            ggml_set_name(BA, "BA");
-
-            if (scaling != 1.0f) {
-                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
-                ggml_set_name(scale_tensor, "scale_tensor");
-
-                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
-                offload_func(BA);
-                ggml_set_name(BA, "BA_scaled");
-            }
-
-            ggml_tensor * r;
-            if (base_t == dest_t) {
-                r = ggml_add_inplace(lora_ctx, dest_t, BA);
-                offload_func_force_inplace(r);
-                ggml_set_name(r, "r_add_inplace");
-            }
-            else {
-                r = ggml_add(lora_ctx, base_t, BA);
-                offload_func(r);
-                ggml_set_name(r, "r_add");
-
-                r = ggml_cpy(lora_ctx, r, dest_t);
-                offload_func(r);
-                ggml_set_name(r, "r_cpy");
-            }
-
-            struct ggml_cgraph gf = ggml_build_forward(r);
-
-            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
-
-            // we won't need these tensors again, reset the context to save memory
-            ggml_free(lora_ctx);
-            lora_ctx = ggml_init(params);
-            lora_tensors.clear();
-
-            n_tensors++;
-            if (n_tensors % 4 == 0) {
-                fprintf(stderr, ".");
-            }
-        }
-    }
-
-    // TODO: this should be in a destructor, it will leak on failure
-    ggml_free(lora_ctx);
-    if (base_ctx) {
-        ggml_free(base_ctx);
-    }
-
-    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
-    fprintf(stderr, " done (%.2f ms)\n", t_lora_us / 1000.0);
-
-    return 0;
-}
-
-int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
-    try {
-        return llama_apply_lora_from_file_internal(ctx->model, path_lora, path_base_model, n_threads);
-    } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_model_apply_lora_from_file(const struct llama_model * model, const char * path_lora, const char * path_base_model, int n_threads) {
-    try {
-        return llama_apply_lora_from_file_internal(*model, path_lora, path_base_model, n_threads);
-    } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_get_kv_cache_token_count(const struct llama_context * ctx) {
-    return ctx->kv_self.n;
-}
-
-#define LLAMA_MAX_RNG_STATE (64*1024)
-
-void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed) {
-    if (seed == LLAMA_DEFAULT_SEED) {
-        seed = time(NULL);
-    }
-    ctx->rng.seed(seed);
-}
-
-// Returns the *maximum* size of the state
-size_t llama_get_state_size(const struct llama_context * ctx) {
-    // we don't know size of rng until we actually serialize it. so reserve more than enough memory for its serialized state.
-    // for reference, std::mt19937(1337) serializes to 6701 bytes.
-    const size_t s_rng_size        = sizeof(size_t);
-    const size_t s_rng             = LLAMA_MAX_RNG_STATE;
-    const size_t s_logits_capacity = sizeof(size_t);
-    const size_t s_logits_size     = sizeof(size_t);
-    const size_t s_logits          = ctx->logits.capacity() * sizeof(float);
-    const size_t s_embedding_size  = sizeof(size_t);
-    const size_t s_embedding       = ctx->embedding.size() * sizeof(float);
-    const size_t s_kv_size         = sizeof(size_t);
-    const size_t s_kv_ntok         = sizeof(int);
-    const size_t s_kv              = ctx->kv_self.buf.size;
-
-    const size_t s_total = (
-        + s_rng_size
-        + s_rng
-        + s_logits_capacity
-        + s_logits_size
-        + s_logits
-        + s_embedding_size
-        + s_embedding
-        + s_kv_size
-        + s_kv_ntok
-        + s_kv
-    );
-
-    return s_total;
-}
-
-// Copies the state to the specified destination address
-size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
-    uint8_t * out = dst;
-
-    // copy rng
-    {
-        std::stringstream rng_ss;
-        rng_ss << ctx->rng;
-
-        const size_t rng_size = rng_ss.str().size();
-        char rng_buf[LLAMA_MAX_RNG_STATE];
-
-        memset(&rng_buf[0], 0, LLAMA_MAX_RNG_STATE);
-        memcpy(&rng_buf[0], rng_ss.str().data(), rng_ss.str().size());
-
-        memcpy(out, &rng_size,   sizeof(rng_size));    out += sizeof(rng_size);
-        memcpy(out, &rng_buf[0], LLAMA_MAX_RNG_STATE); out += LLAMA_MAX_RNG_STATE;
-    }
-
-    // copy logits
-    {
-        const size_t logits_cap  = ctx->logits.capacity();
-        const size_t logits_size = ctx->logits.size();
-
-        memcpy(out, &logits_cap,  sizeof(logits_cap));  out += sizeof(logits_cap);
-        memcpy(out, &logits_size, sizeof(logits_size)); out += sizeof(logits_size);
-
-        if (logits_size) {
-            memcpy(out, ctx->logits.data(), logits_size * sizeof(float));
-        }
-
-        out += logits_cap * sizeof(float);
-    }
-
-    // copy embeddings
-    {
-        const size_t embedding_size = ctx->embedding.size();
-
-        memcpy(out, &embedding_size, sizeof(embedding_size)); out += sizeof(embedding_size);
-
-        if (embedding_size) {
-            memcpy(out, ctx->embedding.data(), embedding_size * sizeof(float));
-            out += embedding_size * sizeof(float);
-        }
-    }
-
-    // copy kv cache
-    {
-        const auto & kv_self = ctx->kv_self;
-        const auto & hparams = ctx->model.hparams;
-        const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd;
-        const int    n_ctx   = hparams.n_ctx;
-
-        const size_t kv_size = kv_self.buf.size;
-        const int    kv_ntok = llama_get_kv_cache_token_count(ctx);
-
-        memcpy(out, &kv_size, sizeof(kv_size)); out += sizeof(kv_size);
-        memcpy(out, &kv_ntok, sizeof(kv_ntok)); out += sizeof(kv_ntok);
-
-        if (kv_size) {
-            const size_t elt_size = ggml_element_size(kv_self.k);
-
-            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
-            ggml_cgraph gf{};
-
-            ggml_tensor * kout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
-            kout3d->data = out;
-            out += ggml_nbytes(kout3d);
-
-            ggml_tensor * vout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
-            vout3d->data = out;
-            out += ggml_nbytes(vout3d);
-
-            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
-                n_embd, kv_ntok, n_layer,
-                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
-
-            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
-                kv_ntok, n_embd, n_layer,
-                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
-
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, k3d, kout3d));
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, v3d, vout3d));
-            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
-
-            ggml_free(cpy_ctx);
-        }
-    }
-
-    const size_t written  = out - dst;
-    const size_t max_size = llama_get_state_size(ctx);
-
-    GGML_ASSERT(written <= max_size);
-
-    return written;
-}
-
-// Sets the state reading from the specified source address
-size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
-    uint8_t * inp = src;
-
-    // set rng
-    {
-        size_t rng_size;
-        char   rng_buf[LLAMA_MAX_RNG_STATE];
-
-        memcpy(&rng_size,   inp, sizeof(rng_size));    inp += sizeof(rng_size);
-        memcpy(&rng_buf[0], inp, LLAMA_MAX_RNG_STATE); inp += LLAMA_MAX_RNG_STATE;
-
-        std::stringstream rng_ss;
-        rng_ss.str(std::string(&rng_buf[0], rng_size));
-        rng_ss >> ctx->rng;
-
-        GGML_ASSERT(rng_ss.fail() == false);
-    }
-
-    // set logits
-    {
-        size_t logits_cap;
-        size_t logits_size;
-
-        memcpy(&logits_cap,  inp, sizeof(logits_cap));  inp += sizeof(logits_cap);
-        memcpy(&logits_size, inp, sizeof(logits_size)); inp += sizeof(logits_size);
-
-        GGML_ASSERT(ctx->logits.capacity() == logits_cap);
-
-        if (logits_size) {
-            ctx->logits.resize(logits_size);
-            memcpy(ctx->logits.data(), inp, logits_size * sizeof(float));
-        }
-
-        inp += logits_cap * sizeof(float);
-    }
-
-    // set embeddings
-    {
-        size_t embedding_size;
-
-        memcpy(&embedding_size, inp, sizeof(embedding_size)); inp += sizeof(embedding_size);
-
-        GGML_ASSERT(ctx->embedding.capacity() == embedding_size);
-
-        if (embedding_size) {
-            memcpy(ctx->embedding.data(), inp, embedding_size * sizeof(float));
-            inp += embedding_size * sizeof(float);
-        }
-    }
-
-    // set kv cache
-    {
-        const auto & kv_self = ctx->kv_self;
-        const auto & hparams = ctx->model.hparams;
-        const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd;
-        const int    n_ctx   = hparams.n_ctx;
-
-        size_t kv_size;
-        int kv_ntok;
-
-        memcpy(&kv_size, inp, sizeof(kv_size)); inp += sizeof(kv_size);
-        memcpy(&kv_ntok, inp, sizeof(kv_ntok)); inp += sizeof(kv_ntok);
-
-        if (kv_size) {
-            GGML_ASSERT(kv_self.buf.size == kv_size);
-
-            const size_t elt_size = ggml_element_size(kv_self.k);
-
-            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
-            ggml_cgraph gf{};
-
-            ggml_tensor * kin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
-            kin3d->data = (void *) inp;
-            inp += ggml_nbytes(kin3d);
-
-            ggml_tensor * vin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
-            vin3d->data = (void *) inp;
-            inp += ggml_nbytes(vin3d);
-
-            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
-                n_embd, kv_ntok, n_layer,
-                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
-
-            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
-                kv_ntok, n_embd, n_layer,
-                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
-
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, kin3d, k3d));
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, vin3d, v3d));
-            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
-
-            ggml_free(cpy_ctx);
-        }
-
-        ctx->kv_self.n = kv_ntok;
-    }
-
-    const size_t nread    = inp - src;
-    const size_t max_size = llama_get_state_size(ctx);
-
-    GGML_ASSERT(nread <= max_size);
-
-    return nread;
-}
-
-static bool llama_load_session_file_internal(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
-    gguf_file file(path_session, "rb");
-    GGML_UNUSED(ctx);
-    GGML_UNUSED(path_session);
-    GGML_UNUSED(tokens_out);
-    GGML_UNUSED(n_token_capacity);
-    GGML_UNUSED(n_token_count_out);
-
-
-// TODO: implement with GGUF format
-    return true;
-}
-
-bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
-    try {
-        return llama_load_session_file_internal(ctx, path_session, tokens_out, n_token_capacity, n_token_count_out);
-    } catch (const std::exception & err) {
-        fprintf(stderr, "error loading session file: %s\n", err.what());
-        return false;
-    }
-}
-
-bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
-    gguf_file file(path_session, "wb");
-    GGML_UNUSED(ctx);
-    GGML_UNUSED(tokens);
-    GGML_UNUSED(n_token_count);
-
-    // TODO: implement with GGUF format
-
-    return true;
-}
-
-int llama_eval(
-        struct llama_context * ctx,
-           const llama_token * tokens,
-                         int   n_tokens,
-                         int   n_past,
-                         int   n_threads) {
-    if (!llama_eval_internal(*ctx, tokens, nullptr, n_tokens, n_past, n_threads, nullptr)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    // get a more accurate load time, upon first eval
-    // TODO: fix this
-    if (!ctx->has_evaluated_once) {
-        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
-        ctx->has_evaluated_once = true;
-    }
-
-    return 0;
-}
-
-
-int llama_eval_embd(
-            struct llama_context * ctx,
-                     const float * embd,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads) {
-    if (!llama_eval_internal(*ctx, nullptr, embd, n_tokens, n_past, n_threads, nullptr)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    // get a more accurate load time, upon first eval
-    // TODO: fix this
-    if (!ctx->has_evaluated_once) {
-        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
-        ctx->has_evaluated_once = true;
-    }
-
-    return 0;
-}
-
-int llama_eval_export(struct llama_context * ctx, const char * fname) {
-    const int n_batch = 1;
-    const int n_ctx   = 512 - n_batch;
-
-    const std::vector<llama_token> tmp(n_batch, llama_token_bos());
-
-    if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    return 0;
-}
-
-int llama_tokenize_with_model(
-    const struct llama_model * model,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    auto res = llama_tokenize(model->vocab, text, add_bos);
-
-    if (n_max_tokens < (int) res.size()) {
-        fprintf(stderr, "%s: too many tokens\n", __func__);
-        return -((int) res.size());
-    }
-
-    for (size_t i = 0; i < res.size(); i++) {
-        tokens[i] = res[i];
-    }
-
-    return res.size();
-}
-
-int llama_tokenize(
-        struct llama_context * ctx,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
-}
-
-int llama_n_vocab_from_model(const struct llama_model * model) {
-    return model->vocab.id_to_token.size();
-}
-
-int llama_n_ctx_from_model(const struct llama_model * model) {
-    return model->hparams.n_ctx;
-}
-
-int llama_n_embd_from_model(const struct llama_model * model) {
-    return model->hparams.n_embd;
-}
-
-int llama_n_vocab(const struct llama_context * ctx) {
-    return ctx->model.vocab.id_to_token.size();
-}
-
-int llama_n_ctx(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_ctx;
-}
-
-int llama_n_embd(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_embd;
-}
-
-int llama_get_vocab_from_model(
-        const struct llama_model * model,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
-    for (int i = 0; i<n; ++i) {
-        strings[i] = model->vocab.id_to_token[i].tok.c_str();
-        scores[i]  = model->vocab.id_to_token[i].score;
-    }
-    return n;
-}
-
-int llama_get_vocab(
-        const struct llama_context * ctx,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
-}
-
-float * llama_get_logits(struct llama_context * ctx) {
-    return ctx->logits.data();
-}
-
-float * llama_get_embeddings(struct llama_context * ctx) {
-    return ctx->embedding.data();
-}
-
-const char * llama_token_to_str_with_model(const struct llama_model * model, llama_token token) {
-    if (token >= llama_n_vocab_from_model(model)) {
-        return nullptr;
-    }
-
-    return model->vocab.id_to_token[token].tok.c_str();
-}
-
-const char * llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    return llama_token_to_str_with_model(&ctx->model, token);
-}
-
-llama_token llama_token_bos() {
-    return 1;
-}
-
-llama_token llama_token_eos() {
-    return 2;
-}
-
-llama_token llama_token_nl() {
-    return 13;
-}
-
-struct llama_timings llama_get_timings(struct llama_context * ctx) {
-    struct llama_timings result = {
-        /*.t_start_ms  =*/ 1e-3 * ctx->t_start_us,
-        /*.t_end_ms    =*/ 1.00 * ggml_time_ms(),
-        /*.t_load_ms   =*/ 1e-3 * ctx->t_load_us,
-        /*.t_sample_ms =*/ 1e-3 * ctx->t_sample_us,
-        /*.t_p_eval_ms =*/ 1e-3 * ctx->t_p_eval_us,
-        /*.t_eval_ms   =*/ 1e-3 * ctx->t_eval_us,
-
-        /*.n_sample =*/ std::max(1, ctx->n_sample),
-        /*.n_p_eval =*/ std::max(1, ctx->n_p_eval),
-        /*.n_eval   =*/ std::max(1, ctx->n_eval),
-    };
-
-    return result;
-}
-
-void llama_print_timings(struct llama_context * ctx) {
-    const llama_timings timings = llama_get_timings(ctx);
-
-    fprintf(stderr, "\n");
-    fprintf(stderr, "%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
-    fprintf(stderr, "%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_sample_ms, timings.n_sample, timings.t_sample_ms / timings.n_sample, 1e3 / timings.t_sample_ms * timings.n_sample);
-    fprintf(stderr, "%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_p_eval_ms, timings.n_p_eval, timings.t_p_eval_ms / timings.n_p_eval, 1e3 / timings.t_p_eval_ms * timings.n_p_eval);
-    fprintf(stderr, "%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_eval_ms, timings.n_eval, timings.t_eval_ms / timings.n_eval, 1e3 / timings.t_eval_ms * timings.n_eval);
-    fprintf(stderr, "%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
-}
-
-void llama_reset_timings(struct llama_context * ctx) {
-    ctx->t_start_us = ggml_time_us();
-    ctx->t_sample_us = ctx->n_sample = 0;
-    ctx->t_eval_us   = ctx->n_eval   = 0;
-    ctx->t_p_eval_us = ctx->n_p_eval = 0;
-}
-
-const char * llama_print_system_info(void) {
-    static std::string s;
-
-    s  = "";
-    s += "AVX = "         + std::to_string(ggml_cpu_has_avx())         + " | ";
-    s += "AVX2 = "        + std::to_string(ggml_cpu_has_avx2())        + " | ";
-    s += "AVX512 = "      + std::to_string(ggml_cpu_has_avx512())      + " | ";
-    s += "AVX512_VBMI = " + std::to_string(ggml_cpu_has_avx512_vbmi()) + " | ";
-    s += "AVX512_VNNI = " + std::to_string(ggml_cpu_has_avx512_vnni()) + " | ";
-    s += "FMA = "         + std::to_string(ggml_cpu_has_fma())         + " | ";
-    s += "NEON = "        + std::to_string(ggml_cpu_has_neon())        + " | ";
-    s += "ARM_FMA = "     + std::to_string(ggml_cpu_has_arm_fma())     + " | ";
-    s += "F16C = "        + std::to_string(ggml_cpu_has_f16c())        + " | ";
-    s += "FP16_VA = "     + std::to_string(ggml_cpu_has_fp16_va())     + " | ";
-    s += "WASM_SIMD = "   + std::to_string(ggml_cpu_has_wasm_simd())   + " | ";
-    s += "BLAS = "        + std::to_string(ggml_cpu_has_blas())        + " | ";
-    s += "SSE3 = "        + std::to_string(ggml_cpu_has_sse3())        + " | ";
-    s += "VSX = "         + std::to_string(ggml_cpu_has_vsx())         + " | ";
-
-    return s.c_str();
-}
-
-// For internal test use
-const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
-    return ctx->model.tensors_by_name;
-}
+// Defines fileno on msys:
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#include <cstddef>
+#include <cstdint>
+#include <cstdio>
+#endif
+
+#include "gguf-util.h"
+#include "gguf-llama.h"
+
+#include "ggml.h"
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+#elif defined(GGML_USE_CLBLAST)
+#include "ggml-opencl.h"
+#endif
+
+#ifdef GGML_USE_METAL
+#include "ggml-metal.h"
+#endif
+#ifdef GGML_USE_MPI
+#include "ggml-mpi.h"
+#endif
+#ifdef GGML_USE_K_QUANTS
+#ifndef QK_K
+#ifdef GGML_QKK_64
+#define QK_K 64
+#else
+#define QK_K 256
+#endif
+#endif
+#endif
+
+#include <array>
+#include <ctime>
+#include <cinttypes>
+#include <fstream>
+#include <random>
+#include <map>
+#include <unordered_map>
+#include <queue>
+#include <cassert>
+#include <cstring>
+#include <climits>
+#include <memory>
+#include <algorithm>
+#include <initializer_list>
+#include <thread>
+#include <atomic>
+#include <mutex>
+#include <sstream>
+#include <numeric>
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+#define LLAMA_USE_SCRATCH
+#define LLAMA_MAX_SCRATCH_BUFFERS 16
+
+// available llama models
+enum e_model {
+    MODEL_UNKNOWN,
+    MODEL_3B,
+    MODEL_7B,
+    MODEL_13B,
+    MODEL_30B,
+    MODEL_65B,
+    MODEL_70B,
+};
+
+static const size_t kB = 1024;
+static const size_t MB = 1024*1024;
+
+// computed for n_ctx == 2048
+// TODO: dynamically determine these sizes
+//       needs modifications in ggml
+
+typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+
+void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
+    (void) tensor;
+}
+
+//
+// ggml helpers
+//
+
+static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
+    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
+
+    if (plan.work_size > 0) {
+        buf.resize(plan.work_size);
+        plan.work_data = buf.data();
+    }
+
+    ggml_graph_compute(graph, &plan);
+}
+
+//
+// memory sizes (calculated for n_batch == 512)
+//
+
+static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   ((size_t) n_ctx / 16ull +  92ull) * MB },
+        { MODEL_7B,   ((size_t) n_ctx / 16ull + 100ull) * MB },
+        { MODEL_13B,  ((size_t) n_ctx / 12ull + 120ull) * MB },
+        { MODEL_30B,  ((size_t) n_ctx /  9ull + 160ull) * MB },
+        { MODEL_65B,  ((size_t) n_ctx /  6ull + 256ull) * MB }, // guess
+        { MODEL_70B,  ((size_t) n_ctx /  7ull + 164ull) * MB },
+    };
+    return k_sizes;
+}
+
+static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,  128ull * MB },
+        { MODEL_7B,  160ull * MB },
+        { MODEL_13B, 192ull * MB },
+        { MODEL_30B, 256ull * MB },
+        { MODEL_65B, 384ull * MB }, // guess
+        { MODEL_70B, 304ull * MB },
+    };
+    return k_sizes;
+}
+
+// used to store the compute graph tensors + non-scratch data
+static const std::map<e_model, size_t> & MEM_REQ_EVAL()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   8ull * MB },
+        { MODEL_7B,  10ull * MB },
+        { MODEL_13B, 12ull * MB },
+        { MODEL_30B, 16ull * MB },
+        { MODEL_65B, 24ull * MB }, // guess
+        { MODEL_70B, 24ull * MB },
+    };
+    return k_sizes;
+}
+
+// amount of VRAM needed per batch size to hold temporary results
+// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,   512ull * kB },
+        { MODEL_7B,   512ull * kB },
+        { MODEL_13B,  640ull * kB },
+        { MODEL_30B,  768ull * kB },
+        { MODEL_65B, 1536ull * kB },
+        { MODEL_70B, 1536ull * kB }, // TODO (likely can be reduced)
+    };
+    return k_sizes;
+}
+
+// amount of VRAM needed per batch size and context to hold temporary results
+// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
+{
+    static std::map<e_model, size_t> k_sizes = {
+        { MODEL_3B,  128ull },
+        { MODEL_7B,  128ull },
+        { MODEL_13B, 160ull },
+        { MODEL_30B, 208ull },
+        { MODEL_65B, 416ull },
+        { MODEL_70B, 416ull }, // TODO (likely can be reduced)
+    };
+    return k_sizes;
+}
+
+// default hparams (LLaMA 7B)
+struct llama_hparams {
+    uint32_t n_vocab   = 32000;
+    uint32_t n_ctx     = 512;   // this is provided as user input?
+    uint32_t n_embd    = 4096;
+    uint32_t n_head    = 32;
+    uint32_t n_head_kv = 32;
+    uint32_t n_layer   = 32;
+    uint32_t n_rot     = 64;
+    uint32_t n_ff       = 11008;
+
+    float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
+
+    float rope_freq_base  = 10000.0f;
+    float rope_freq_scale = 1.0f;
+
+    enum llama_ftype ftype = LLAMA_FTYPE_MOSTLY_F16;
+
+    bool operator!=(const llama_hparams & other) const {
+        return static_cast<bool>(memcmp(this, &other, sizeof(llama_hparams))); // NOLINT
+    }
+
+    uint32_t n_gqa() const {
+        return n_head/n_head_kv;
+    }
+
+    uint32_t n_embd_head() const {
+        return n_embd/n_head;
+    }
+
+    uint32_t n_embd_gqa() const {
+        return n_embd/n_gqa();
+    }
+
+    size_t kv_size() const {
+        size_t result = 2ull;
+        result *= (size_t) n_embd_gqa();
+        result *= (size_t) n_ctx;
+        result *= (size_t) n_layer;
+        result *= sizeof(ggml_fp16_t);
+        return result;
+    }
+};
+
+struct llama_layer {
+    // normalization
+    struct ggml_tensor * attention_norm;
+
+    // attention
+    struct ggml_tensor * wq;
+    struct ggml_tensor * wk;
+    struct ggml_tensor * wv;
+    struct ggml_tensor * wo;
+
+    // normalization
+    struct ggml_tensor * ffn_norm;
+
+    // ff
+    struct ggml_tensor * w1;
+    struct ggml_tensor * w2;
+    struct ggml_tensor * w3;
+};
+
+struct llama_kv_cache {
+    struct ggml_tensor * k = NULL;
+    struct ggml_tensor * v = NULL;
+
+    struct ggml_context * ctx = NULL;
+
+    gguf_ctx_buffer buf;
+
+    int n; // number of tokens currently in the cache
+
+    ~llama_kv_cache() {
+        if (ctx) {
+            ggml_free(ctx);
+        }
+
+#ifdef GGML_USE_CUBLAS
+        ggml_cuda_free_data(k);
+        ggml_cuda_free_data(v);
+#endif // GGML_USE_CUBLAS
+    }
+};
+
+struct llama_vocab {
+    // TODO: convert to this gguf_vocab
+    // add a vector of merges
+    // add members for bos/eos/pad/sep tokens
+    // so that we can pass it to different types of tokenizers with a common interface
+
+    using id    = int32_t;
+    using token = std::string;
+
+    struct token_score {
+        token tok;
+        float score;
+    };
+
+    std::unordered_map<token, id> token_to_id;
+    std::vector<token_score> id_to_token;
+};
+
+struct llama_model {
+    e_model type = MODEL_UNKNOWN;
+
+    llama_hparams hparams;
+
+    struct ggml_tensor * tok_embeddings;
+
+    struct ggml_tensor * norm;
+    struct ggml_tensor * output;
+
+    std::vector<llama_layer> layers;
+    int n_gpu_layers;
+
+    // context
+    struct ggml_context * ctx = NULL;
+
+    // the model memory buffer
+    gguf_ctx_buffer buf;
+
+    // model memory mapped file
+    std::unique_ptr<gguf_mmap> mapping;
+
+    // objects representing data potentially being locked in memory
+    gguf_mlock mlock_buf;
+    gguf_mlock mlock_mmap;
+
+    // for quantize-stats only
+    std::vector<std::pair<std::string, struct ggml_tensor *>> tensors_by_name;
+
+    int64_t t_load_us = 0;
+    int64_t t_start_us = 0;
+
+    llama_vocab vocab;
+
+    ~llama_model() {
+        if (ctx) {
+            ggml_free(ctx);
+        }
+
+#ifdef GGML_USE_CUBLAS
+        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
+            ggml_cuda_free_data(tensors_by_name[i].second);
+        }
+        ggml_cuda_free_scratch();
+#elif defined(GGML_USE_CLBLAST)
+        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
+            ggml_cl_free_data(tensors_by_name[i].second);
+        }
+#endif
+    }
+};
+
+struct llama_context {
+    llama_context(const llama_model & model) : model(model), t_load_us(model.t_load_us), t_start_us(model.t_start_us) {}
+#ifdef GGML_USE_METAL
+    ~llama_context() {
+        if (ctx_metal) {
+            ggml_metal_free(ctx_metal);
+        }
+    }
+#endif
+    std::mt19937 rng;
+
+    bool has_evaluated_once = false;
+
+    int64_t t_sample_us = 0;
+    int64_t t_eval_us   = 0;
+    int64_t t_p_eval_us = 0;
+
+    int32_t n_sample = 0; // number of tokens sampled
+    int32_t n_eval   = 0; // number of eval calls
+    int32_t n_p_eval = 0; // number of tokens in eval calls for the prompt (with batch size > 1)
+
+    const llama_model & model;
+
+    bool model_owner = false;
+
+    int64_t t_load_us;
+    int64_t t_start_us;
+
+    // key + value cache for the self attention
+    struct llama_kv_cache kv_self;
+
+    size_t mem_per_token = 0;
+
+    // decode output (2-dimensional array: [n_tokens][n_vocab])
+    std::vector<float> logits;
+    bool logits_all = false;
+
+    // input embedding (1-dimensional array: [n_embd])
+    std::vector<float> embedding;
+
+    // reusable buffer for `struct ggml_graph_plan.work_data`
+    std::vector<uint8_t> work_buffer;
+
+    // memory buffers used to evaluate the model
+    // TODO: move in llama_state
+    gguf_ctx_buffer buf_compute;
+    gguf_ctx_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+
+#ifdef GGML_USE_METAL
+    ggml_metal_context * ctx_metal = NULL;
+#endif
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_context * ctx_mpi = NULL;
+#endif
+
+    int    buf_last = 0;
+    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
+
+    void use_buf(struct ggml_context * ctx, int i) {
+#if defined(LLAMA_USE_SCRATCH)
+        size_t last_size = 0;
+
+        if (i == -1) {
+            last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
+        } else {
+            auto & buf = buf_scratch[i];
+            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.addr, });
+        }
+
+        if (buf_last >= 0) {
+            buf_max_size[buf_last] = std::max(buf_max_size[buf_last], last_size);
+        }
+
+        buf_last = i;
+#else
+        (void) i;
+        (void) ctx;
+#endif
+    }
+
+    size_t get_buf_max_mem(int i) const {
+#if defined(LLAMA_USE_SCRATCH)
+        return buf_max_size[i];
+#else
+        (void) i;
+        return 0;
+#endif
+    }
+};
+
+template <typename T>
+static T checked_mul(T a, T b) {
+    T ret = a * b;
+    if (a != 0 && ret / a != b) {
+        throw std::runtime_error(format("overflow multiplying %llu * %llu",
+                     (unsigned long long) a, (unsigned long long) b));
+    }
+    return ret;
+}
+
+static size_t checked_div(size_t a, size_t b) {
+    if (b == 0 || a % b != 0) {
+        throw std::runtime_error(format("error dividing %zu / %zu", a, b));
+    }
+    return a / b;
+}
+
+static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
+    char buf[256];
+    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
+    for (size_t i = 1; i < ne.size(); i++) {
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
+    }
+    return buf;
+}
+
+static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
+    size_t size = ggml_type_size(type);
+    for (uint32_t dim : ne) {
+        size = checked_mul<size_t>(size, dim);
+    }
+    return size / ggml_blck_size(type);
+}
+
+struct gguf_load_tensor {
+    std::string name;
+    enum ggml_type type = GGML_TYPE_F32;
+    std::vector<uint32_t> ne;
+    size_t file_off;
+    size_t size;
+    struct ggml_tensor * ggml_tensor = NULL;
+    uint8_t * data;
+};
+
+struct gguf_load_tensors_map {
+    // tensors is kept in a separate vector to preserve file order
+    std::vector<gguf_load_tensor> tensors;
+    std::unordered_map<std::string, size_t> name_to_idx;
+};
+
+enum gguf_file_version {
+    GGUF_FILE_VERSION_V1 = 1,
+
+};
+
+
+struct gguf_file_loader {
+    gguf_file file;
+    gguf_context * gguf_ctx;
+    gguf_file_version file_version;
+    llama_hparams hparams;
+    llama_vocab vocab;
+struct ggml_context * ctx_data = NULL;
+
+    gguf_file_loader(const char * fname, gguf_load_tensors_map & tensors_map)
+        : file(fname, "rb") {
+        fprintf(stderr, "llama.cpp: loading model from %s\n", fname);
+
+    struct gguf_init_params params = {
+        /*.no_alloc = */ true,
+        /*.ctx      = */ &ctx_data,
+    };
+
+    gguf_ctx = gguf_init_from_file(fname, params);
+    file_version = (enum gguf_file_version) gguf_get_version(gguf_ctx);
+
+    read_hparams();
+    read_vocab();
+        read_tensor_metadata(tensors_map);
+    }
+
+    uint32_t read_u32(const char * key) {
+        int i = gguf_find_key(gguf_ctx, key);
+        if (i == -1) {
+            throw std::runtime_error(format("cannot find param with key %s\n", key));
+        }
+
+        return gguf_get_val_u32(gguf_ctx, i);
+    }
+
+    float read_f32(const char * key) {
+        int i = gguf_find_key(gguf_ctx, key);
+        if (i == -1) {
+            throw std::runtime_error(format("cannot find param with key %s\n", key));
+        }
+
+        return gguf_get_val_f32(gguf_ctx, i);
+    }
+
+    int read_n_vocab() {
+        int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
+    if (i == -1) {
+        throw std::runtime_error("cannot find token list in GGUF file\n");
+    }
+
+    return gguf_get_arr_n(gguf_ctx, i);
+    }
+
+    void read_hparams() {
+
+        // TODO define keys as constants in header
+        // TODO: read all hparams from file
+
+        hparams.n_vocab = read_n_vocab();
+        hparams.n_ctx   = read_u32("llama.context_length");
+        hparams.n_embd  = read_u32("llama.embedding_length");
+        hparams.n_ff    = read_u32("llama.feed_forward_length");
+        hparams.n_head  = read_u32("llama.attention.head_count");
+        hparams.n_layer = read_u32("llama.layer_count");
+        hparams.n_rot   = read_u32("llama.rope.dimension_count");
+        hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
+
+        // LLaMAv2
+        // hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
+    }
+
+    void read_vocab() {
+        vocab.id_to_token.resize(hparams.n_vocab);
+        int token_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
+        if (token_idx == -1) {
+            throw std::runtime_error("cannot find token list in GGUF file\n");
+        }
+
+        int score_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.scores");
+        if (score_idx == -1) {
+            throw std::runtime_error("cannot find token scores list in GGUF file\n");
+        }
+
+        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
+
+            std::string word = gguf_get_arr_str(gguf_ctx, token_idx, i);
+
+            vocab.token_to_id[word] = i;
+
+            auto & tok_score = vocab.id_to_token[i];
+            tok_score.tok = std::move(word);
+            tok_score.score = gguf_get_arr_f32(gguf_ctx, score_idx, i);
+        }
+    }
+
+    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) {
+        const int n_tensors = gguf_get_n_tensors(gguf_ctx);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            gguf_load_tensor tensor;
+            const char * name = gguf_get_tensor_name(gguf_ctx, i);
+
+            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
+            uint32_t n_dims = cur->n_dims;
+            tensor.type = cur->type;
+            tensor.ne.resize(n_dims);
+            for (uint32_t j = 0; j < n_dims; ++j) {
+            tensor.ne[j] = cur->ne[j];
+            }
+
+            if (n_dims < 1 || n_dims > 2) {
+                throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
+            }
+            switch (tensor.type) {
+                case GGML_TYPE_F32:
+                case GGML_TYPE_F16:
+                case GGML_TYPE_Q4_0:
+                case GGML_TYPE_Q4_1:
+                case GGML_TYPE_Q5_0:
+                case GGML_TYPE_Q5_1:
+                case GGML_TYPE_Q8_0:
+                case GGML_TYPE_Q2_K:
+                case GGML_TYPE_Q3_K:
+                case GGML_TYPE_Q4_K:
+                case GGML_TYPE_Q5_K:
+                case GGML_TYPE_Q6_K:
+                    break;
+                default: {
+                    throw std::runtime_error(format("unrecognized tensor type %u\n", tensor.type));
+                }
+            }
+
+
+            tensor.file_off = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, i);
+
+            tensor.name = name;
+            tensor.size = llama_calc_tensor_size(tensor.ne, tensor.type);
+
+            tensors_map.tensors.push_back(tensor);
+            tensors_map.name_to_idx[name] = tensors_map.tensors.size() - 1;
+        }
+    }
+};
+
+struct gguf_file_saver {
+    // TODO
+    // this implementation now assumes that the data section is of the same length as the unquantized model.
+    // this is needed to write tensor metadata and weights in a single pass by seeking to appropriate positions in the file.
+    // this may not be true when we add quantization version and change ftype description (currently it's string according to the specs,
+    // but better to have it as uint32).
+    // we need to calculate the delta in number of bytes written with a counter as a struct member.
+
+    gguf_file file;
+    gguf_file_loader * fl;
+    size_t info_offset;
+    size_t tensor_offset = 0;
+
+    gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
+        : file(fname, "wb"), fl(fl) {
+        fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
+        write_header();
+        write_hparams(new_ftype);
+    }
+
+    void write_header() {
+        const int32_t magic = GGUF_MAGIC;
+        file.write_i32(magic);
+
+            const int32_t version = GGUF_VERSION;
+            file.write_i32(version);
+
+            const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
+            file.write_i32(n_tensors);
+
+            const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
+            file.write_i32(n_kv);
+        }
+
+        void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
+            std::vector<std::string> data(n_arr);
+
+            for (int j = 0; j < n_arr; ++j) {
+                std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
+                data[j] = val;
+                }
+
+                            file.write_arr<std::string>(key, type, data);
+        }
+
+        void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
+            std::vector<float> data(n_arr);
+
+            for (int j = 0; j < n_arr; ++j) {
+                float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
+                data[j] = val;
+                }
+
+                            file.write_arr<float>(key, type, data);
+        }
+
+    void write_hparams(enum llama_ftype new_ftype) {
+        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(fl->gguf_ctx, i);
+            if (strcmp(key, "general.quantization_version") == 0) {
+                file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
+            } else {
+                const gguf_type vtype = gguf_get_kv_type(fl->gguf_ctx, i);
+
+                bool bool_val;
+                float f32_val;
+                int16_t i16_val;
+                int32_t i32_val;
+                int8_t i8_val;
+                std::string str_val;
+                uint16_t u16_val;
+                uint32_t u32_val;
+                uint8_t u8_val;
+                gguf_type arr_type;
+                int n_arr;
+
+                switch(vtype) {
+                    case GGUF_TYPE_BOOL:
+                    bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
+                    file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
+                    break;
+                    case GGUF_TYPE_FLOAT32:
+                    f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
+                    file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
+                    break;
+                    case GGUF_TYPE_INT16:
+                    i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
+                    file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
+                    break;
+                    case GGUF_TYPE_INT32:
+                    i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
+                    file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
+                    break;
+                    case GGUF_TYPE_INT8:
+                    i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
+                    file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
+                    break;
+                    case GGUF_TYPE_STRING:
+                    str_val = gguf_get_val_str(fl->gguf_ctx, i);
+                    file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
+                    break;
+                    case GGUF_TYPE_UINT16:
+                    u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
+                    file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
+                    break;
+                    case GGUF_TYPE_UINT32:
+                    u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
+                    file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
+                    break;
+                    case GGUF_TYPE_UINT8:
+                    u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
+                    file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
+                    break;
+                    case GGUF_TYPE_ARRAY:
+                    arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
+                    n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
+                    if (arr_type == GGUF_TYPE_FLOAT32) {
+                        write_hparam_arr_f32(key, arr_type, i, n_arr);
+                        } else if (arr_type == GGUF_TYPE_STRING) {
+                            write_hparam_arr_str(key, GGUF_TYPE_STRING, i, n_arr);
+                        } else {
+                            throw std::runtime_error("not implemented");
+                        }
+                    break;
+                    default:
+                    throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
+                }
+            }
+        }
+
+        info_offset  = file.tell();
+        size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
+        file.write_zeros(count);
+        file.seek(info_offset, SEEK_SET);
+        GGML_ASSERT(info_offset == file.tell());
+    }
+
+    size_t write_tensor_info(gguf_load_tensor & tensor, enum ggml_type type) {
+        size_t total_written = 0;
+        file.seek(info_offset, SEEK_SET);
+        GGML_ASSERT(info_offset == file.tell());
+        total_written += file.write_str(tensor.name);
+
+        int32_t n_dims = tensor.ne.size();
+        total_written += file.write_i32(n_dims);
+        for (int32_t i = 0; i < n_dims; ++i) {
+            total_written += file.write_i32(tensor.ne[i]);
+        }
+
+        total_written += file.write_i32(type);
+        total_written += file.write_u64(tensor_offset);
+        info_offset   += total_written; // position to write info of the next tensor
+
+        file.seek(0, SEEK_END);
+
+        return total_written;
+    }
+
+    void write_tensor(gguf_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
+        switch (new_type) {
+            case GGML_TYPE_F32:
+            case GGML_TYPE_F16:
+            case GGML_TYPE_Q4_0:
+            case GGML_TYPE_Q4_1:
+            case GGML_TYPE_Q5_0:
+            case GGML_TYPE_Q5_1:
+            case GGML_TYPE_Q8_0:
+            case GGML_TYPE_Q2_K:
+            case GGML_TYPE_Q3_K:
+            case GGML_TYPE_Q4_K:
+            case GGML_TYPE_Q5_K:
+            case GGML_TYPE_Q6_K:
+                break;
+            default: GGML_ASSERT(false);
+        }
+
+        write_tensor_info(tensor, new_type);
+        file.write_raw(new_data, new_size);
+        size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
+        size_t pad = padded_size - new_size;
+        file.write_zeros(pad);
+        tensor_offset += padded_size; // offset of the next tensor
+    }
+};
+
+struct llama_model_loader {
+    std::unique_ptr<gguf_file_loader> file_loader;
+    gguf_load_tensors_map tensors_map;
+    bool use_mmap;
+    size_t num_ggml_tensors_created = 0;
+    struct ggml_context * ggml_ctx = NULL;
+    std::unique_ptr<gguf_mmap> mapping;
+
+    llama_model_loader(const std::string & fname_base, bool use_mmap) {
+        file_loader = std::unique_ptr<gguf_file_loader>(new gguf_file_loader(fname_base.c_str(), tensors_map));
+        if (!gguf_mmap::SUPPORTED) {
+            use_mmap = false;
+        }
+        this->use_mmap = use_mmap;
+    }
+
+    void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
+        *ctx_size_p = *mmapped_size_p = 0;
+        for (const gguf_load_tensor & lt : tensors_map.tensors) {
+            *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
+            *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
+        }
+    }
+
+    struct ggml_tensor * get_tensor(const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
+        auto it = tensors_map.name_to_idx.find(name);
+        if (it == tensors_map.name_to_idx.end()) {
+            throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
+        }
+        gguf_load_tensor & lt = tensors_map.tensors.at(it->second);
+        if (lt.ne != ne) {
+            throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
+                         name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
+        }
+
+        return get_tensor_for(lt, backend);
+    }
+
+    struct ggml_tensor * get_tensor_for(gguf_load_tensor & lt, ggml_backend backend) {
+        struct ggml_tensor * tensor;
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ggml_ctx, true);
+        }
+        if (lt.ne.size() == 2) {
+            tensor = ggml_new_tensor_2d(ggml_ctx, lt.type, lt.ne.at(0), lt.ne.at(1));
+        } else {
+            GGML_ASSERT(lt.ne.size() == 1);
+            tensor = ggml_new_tensor_1d(ggml_ctx, lt.type, lt.ne.at(0));
+        }
+        ggml_set_name(tensor, lt.name.c_str());
+        GGML_ASSERT(lt.ggml_tensor == NULL); // if this fails, we called get_tensor twice on the same tensor
+
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ggml_ctx, use_mmap);
+        }
+        tensor->backend = backend;
+        lt.ggml_tensor = tensor;
+        num_ggml_tensors_created++;
+        return tensor;
+    }
+
+    void done_getting_tensors() const {
+        if (num_ggml_tensors_created != tensors_map.tensors.size()) {
+            throw std::runtime_error(std::string("llama.cpp: file contained more tensors than expected"));
+        }
+    }
+
+    void load_all_data(llama_progress_callback progress_callback, void *  progress_callback_user_data, gguf_mlock * lmlock) {
+        size_t data_size = 0;
+        size_t prefetch_size = 0;
+        size_t lock_size = 0;
+        for (const gguf_load_tensor & lt : tensors_map.tensors) {
+            data_size += lt.size;
+            if (lt.ggml_tensor->backend == GGML_BACKEND_CPU) {
+                prefetch_size += lt.size;
+            }
+        }
+
+        if (use_mmap) {
+            mapping.reset(new gguf_mmap(&file_loader->file, prefetch_size, ggml_is_numa()));
+            if (lmlock) {
+                lmlock->init(mapping->addr);
+            }
+        }
+
+        size_t done_size = 0;
+        for (gguf_load_tensor & lt : tensors_map.tensors) {
+            if (progress_callback) {
+                progress_callback((float) done_size / data_size, progress_callback_user_data);
+            }
+            GGML_ASSERT(lt.ggml_tensor); // unused tensors should have been caught by load_data already
+            lt.data = (uint8_t *) lt.ggml_tensor->data;
+
+            // allocate temp buffer if not using mmap
+            if (!use_mmap && lt.data == NULL) {
+                GGML_ASSERT(lt.ggml_tensor->backend != GGML_BACKEND_CPU);
+                lt.data = (uint8_t*)malloc(ggml_nbytes(lt.ggml_tensor));
+            }
+
+            load_data_for(lt);
+
+            switch(lt.ggml_tensor->backend) {
+                case GGML_BACKEND_CPU:
+                    lt.ggml_tensor->data = lt.data;
+                    if (use_mmap && lmlock) {
+                        lock_size += lt.size;
+                        lmlock->grow_to(lock_size);
+                    }
+                    break;
+#if defined(GGML_USE_CUBLAS)
+                case GGML_BACKEND_GPU:
+                case GGML_BACKEND_GPU_SPLIT:
+                    ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+                    if (!use_mmap) {
+                        free(lt.data);
+                    }
+                    break;
+#elif defined(GGML_USE_CLBLAST)
+                case GGML_BACKEND_GPU:
+                    ggml_cl_transform_tensor(lt.data, lt.ggml_tensor);
+                    if (!use_mmap) {
+                        free(lt.data);
+                    }
+                    break;
+#endif
+                default:
+                    continue;
+            }
+
+            done_size += lt.size;
+        }
+    }
+
+    void load_data_for(gguf_load_tensor & lt) {
+        if (use_mmap) {
+            lt.data = (uint8_t *) mapping->addr + lt.file_off;
+        } else {
+            gguf_file & file = file_loader->file;
+            file.seek(lt.file_off, SEEK_SET);
+            file.read_raw(lt.data, lt.size);
+        }
+
+        if (0) {
+            print_checksum(lt);
+        }
+    }
+
+    static void print_checksum(gguf_load_tensor & lt) {
+        uint32_t sum = 0;
+        for (size_t i = 0; i < lt.size; i++) {
+            uint8_t byte = lt.data[i];
+            sum = byte + (sum << 6) + (sum << 16) - sum; // sdbm hash
+        }
+        fprintf(stderr, "%s checksum: %#08x (%s, size %zu)\n", lt.name.c_str(), sum,
+                llama_format_tensor_shape(lt.ne).c_str(), lt.size);
+    }
+
+};
+
+//
+// kv cache
+//
+
+static bool kv_cache_init(
+        const struct llama_hparams & hparams,
+             struct llama_kv_cache & cache,
+                         ggml_type   wtype,
+                               int   n_ctx,
+                               int   n_gpu_layers) {
+    const int n_embd  = hparams.n_embd_gqa();
+    const int n_layer = hparams.n_layer;
+
+    const int64_t n_mem      = n_layer*n_ctx;
+    const int64_t n_elements = n_embd*n_mem;
+
+    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
+    cache.n = 0;
+
+    struct ggml_init_params params;
+    params.mem_size   = cache.buf.size;
+    params.mem_buffer = cache.buf.addr;
+    params.no_alloc   = false;
+
+    cache.ctx = ggml_init(params);
+
+    if (!cache.ctx) {
+        fprintf(stderr, "%s: failed to allocate memory for kv cache\n", __func__);
+        return false;
+    }
+
+    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    ggml_set_name(cache.k, "cache_k");
+    ggml_set_name(cache.v, "cache_v");
+
+    (void) n_gpu_layers;
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer + 1) {
+        ggml_cuda_assign_buffers_no_scratch(cache.v);
+    }
+    if (n_gpu_layers > n_layer + 2) {
+        ggml_cuda_assign_buffers_no_scratch(cache.k);
+    }
+#endif // GGML_USE_CUBLAS
+
+    return true;
+}
+
+struct llama_context_params llama_context_default_params() {
+    struct llama_context_params result = {
+        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
+        /*.n_ctx                       =*/ 512,
+        /*.n_batch                     =*/ 512,
+        /*.n_gqa                       =*/ 1,
+        /*.rms_norm_eps                =*/ LLAMA_DEFAULT_RMS_EPS,
+        /*.gpu_layers                  =*/ 0,
+        /*.main_gpu                    =*/ 0,
+        /*.tensor_split                =*/ nullptr,
+        /*.rope_freq_base              =*/ 10000.0f,
+        /*.rope_freq_scale             =*/ 1.0f,
+        /*.progress_callback           =*/ nullptr,
+        /*.progress_callback_user_data =*/ nullptr,
+        /*.low_vram                    =*/ false,
+        /*.f16_kv                      =*/ true,
+        /*.logits_all                  =*/ false,
+        /*.vocab_only                  =*/ false,
+        /*.use_mmap                    =*/ true,
+        /*.use_mlock                   =*/ false,
+        /*.embedding                   =*/ false,
+    };
+
+    return result;
+}
+
+struct llama_model_quantize_params llama_model_quantize_default_params() {
+    struct llama_model_quantize_params result = {
+        /*.nthread                     =*/ 0,
+        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
+        /*.allow_requantize            =*/ false,
+        /*.quantize_output_tensor      =*/ true,
+    };
+
+    return result;
+}
+
+int llama_max_devices() {
+    return LLAMA_MAX_DEVICES;
+}
+
+bool llama_mmap_supported() {
+    return gguf_mmap::SUPPORTED;
+}
+
+bool llama_mlock_supported() {
+    return gguf_mlock::SUPPORTED;
+}
+
+void llama_backend_init(bool numa) {
+    ggml_time_init();
+
+    // needed to initialize f16 tables
+    {
+        struct ggml_init_params params = { 0, NULL, false };
+        struct ggml_context * ctx = ggml_init(params);
+        ggml_free(ctx);
+    }
+
+    if (numa) {
+        ggml_numa_init();
+    }
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_init();
+#endif
+}
+
+void llama_backend_free() {
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_free();
+#endif
+}
+
+int64_t llama_time_us() {
+    return ggml_time_us();
+}
+
+//
+// model loading
+//
+
+static const char *gguf_file_version_name(gguf_file_version version) {
+    switch (version) {
+        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
+        }
+
+    return "unknown";
+}
+
+static const char *llama_ftype_name(enum llama_ftype ftype) {
+    switch (ftype) {
+        case LLAMA_FTYPE_ALL_F32:     return "all F32";
+        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
+        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
+        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
+        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
+                                      return "mostly Q4_1, some F16";
+        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
+        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
+        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K: return "mostly Q2_K";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q6_K: return "mostly Q6_K";
+        default:                      return "unknown, may not work";
+    }
+}
+
+static const char *llama_model_type_name(e_model type) {
+    switch (type) {
+        case MODEL_3B: return "3B";
+        case MODEL_7B: return "7B";
+        case MODEL_13B: return "13B";
+        case MODEL_30B: return "30B";
+        case MODEL_65B: return "65B";
+        case MODEL_70B: return "70B";
+        default: GGML_ASSERT(false);
+    }
+}
+
+static void llama_model_load_internal(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gqa,
+        float rms_norm_eps,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void * progress_callback_user_data) {
+            GGML_UNUSED(rms_norm_eps); // TODO: update function signature to remove this
+
+    model.t_start_us = ggml_time_us();
+
+    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
+
+    vocab = std::move(ml->file_loader->vocab);
+    model.hparams = ml->file_loader->hparams;
+    model.n_gpu_layers = n_gpu_layers;
+    gguf_file_version file_version = ml->file_loader->file_version;
+
+    auto & hparams = model.hparams;
+
+    {
+        switch (hparams.n_layer) {
+            case 26: model.type = e_model::MODEL_3B; break;
+            case 32: model.type = e_model::MODEL_7B; break;
+            case 40: model.type = e_model::MODEL_13B; break;
+            case 60: model.type = e_model::MODEL_30B; break;
+            case 80: model.type = e_model::MODEL_65B; break;
+            default:
+                {
+                    if (hparams.n_layer < 32) {
+                        model.type = e_model::MODEL_7B;
+                    }
+                } break;
+        }
+
+        hparams.n_ctx = n_ctx;
+
+        // LLaMAv2
+        hparams.n_head_kv = hparams.n_head / n_gqa;
+        if (model.type == e_model::MODEL_65B && n_gqa == 8) {
+            fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
+            model.type = e_model::MODEL_70B;
+            }
+
+        hparams.rope_freq_base  = rope_freq_base;
+        hparams.rope_freq_scale = rope_freq_scale;
+    }
+
+    const uint32_t n_ff = hparams.n_ff;
+
+    {
+        fprintf(stderr, "%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
+        fprintf(stderr, "%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
+        fprintf(stderr, "%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
+        fprintf(stderr, "%s: n_embd     = %u\n",   __func__, hparams.n_embd);
+        fprintf(stderr, "%s: n_head     = %u\n",   __func__, hparams.n_head);
+        fprintf(stderr, "%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
+        fprintf(stderr, "%s: n_layer    = %u\n",   __func__, hparams.n_layer);
+        fprintf(stderr, "%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        fprintf(stderr, "%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
+        fprintf(stderr, "%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
+        fprintf(stderr, "%s: n_ff       = %u\n",   __func__, n_ff);
+        fprintf(stderr, "%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
+        fprintf(stderr, "%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
+        fprintf(stderr, "%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
+        fprintf(stderr, "%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
+    }
+
+    if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
+        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
+        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
+        throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
+    }
+
+    if (vocab_only) {
+        return;
+    }
+
+    auto & ctx = model.ctx;
+
+    size_t ctx_size;
+    size_t mmapped_size;
+    ml->calc_sizes(&ctx_size, &mmapped_size);
+    fprintf(stderr, "%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
+
+    // create the ggml context
+    {
+        model.buf.resize(ctx_size);
+        if (use_mlock) {
+            model.mlock_buf.init   (model.buf.addr);
+            model.mlock_buf.grow_to(model.buf.size);
+        }
+
+        struct ggml_init_params params = {
+            /*.mem_size   =*/ model.buf.size,
+            /*.mem_buffer =*/ model.buf.addr,
+            /*.no_alloc   =*/ ml->use_mmap,
+        };
+
+        model.ctx = ggml_init(params);
+        if (!model.ctx) {
+            throw std::runtime_error(format("ggml_init() failed"));
+        }
+    }
+
+    (void) main_gpu;
+#if defined(GGML_USE_CUBLAS)
+    fprintf(stderr, "%s: using CUDA for GPU acceleration\n", __func__);
+    ggml_cuda_set_main_device(main_gpu);
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU_SPLIT
+#elif defined(GGML_USE_CLBLAST)
+    fprintf(stderr, "%s: using OpenCL for GPU acceleration\n", __func__);
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU
+#else
+#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_CPU
+#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_CPU
+#endif
+
+    // prepare memory for the weights
+    size_t vram_weights = 0;
+    size_t vram_scratch = 0;
+    {
+        const uint32_t n_embd     = hparams.n_embd;
+        const uint32_t n_embd_gqa = hparams.n_embd_gqa();
+        const uint32_t n_layer    = hparams.n_layer;
+        const uint32_t n_vocab    = hparams.n_vocab;
+
+        ml->ggml_ctx = ctx;
+
+        model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
+
+        // "output" tensor
+        {
+            ggml_backend backend_norm;
+            ggml_backend backend_output;
+            if (n_gpu_layers > int(n_layer)) { // NOLINT
+                // norm is not performance relevant on its own but keeping it in VRAM reduces data copying
+                // on Windows however this is detrimental unless everything is on the GPU
+#ifndef _WIN32
+                backend_norm = low_vram ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
+#else
+                backend_norm = low_vram || n_gpu_layers <= (int) n_layer + 2 ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
+#endif // _WIN32
+
+                backend_output = LLAMA_BACKEND_OFFLOAD_SPLIT;
+            } else {
+                backend_norm = GGML_BACKEND_CPU;
+                backend_output = GGML_BACKEND_CPU;
+            }
+
+            model.norm   = ml->get_tensor("norm.weight",   {n_embd},          backend_norm);
+            model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}, backend_output);
+            if (backend_norm == GGML_BACKEND_GPU) {
+                vram_weights += ggml_nbytes(model.norm);
+            }
+            if (backend_output == GGML_BACKEND_GPU_SPLIT) {
+                vram_weights += ggml_nbytes(model.output);
+            }
+        }
+
+        const int i_gpu_start = n_layer - n_gpu_layers;
+
+        model.layers.resize(n_layer);
+        for (uint32_t i = 0; i < n_layer; ++i) {
+            const ggml_backend backend = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD; // NOLINT
+            const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
+
+            auto & layer = model.layers[i];
+
+            std::string layers_i = "layers." + std::to_string(i);
+
+            layer.attention_norm = ml->get_tensor(layers_i + ".attention_norm.weight", {n_embd}, backend);
+
+            layer.wq = ml->get_tensor(layers_i + ".attention.wq.weight", {n_embd, n_embd},     backend_split);
+            layer.wk = ml->get_tensor(layers_i + ".attention.wk.weight", {n_embd, n_embd_gqa}, backend_split);
+            layer.wv = ml->get_tensor(layers_i + ".attention.wv.weight", {n_embd, n_embd_gqa}, backend_split);
+            layer.wo = ml->get_tensor(layers_i + ".attention.wo.weight", {n_embd, n_embd},     backend_split);
+
+            layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
+
+            layer.w1 = ml->get_tensor(layers_i + ".feed_forward.w1.weight", {n_embd,   n_ff}, backend_split);
+            layer.w2 = ml->get_tensor(layers_i + ".feed_forward.w2.weight", {  n_ff, n_embd}, backend_split);
+            layer.w3 = ml->get_tensor(layers_i + ".feed_forward.w3.weight", {n_embd,   n_ff}, backend_split);
+
+            if (backend == GGML_BACKEND_GPU) {
+                vram_weights +=
+                    ggml_nbytes(layer.attention_norm) + ggml_nbytes(layer.wq) + ggml_nbytes(layer.wk)             +
+                    ggml_nbytes(layer.wv)             + ggml_nbytes(layer.wo) + ggml_nbytes(layer.ffn_norm) +
+                    ggml_nbytes(layer.w1)             + ggml_nbytes(layer.w2) + ggml_nbytes(layer.w3);
+            }
+        }
+    }
+
+    ml->done_getting_tensors();
+
+    // print memory requirements
+    {
+        const size_t scale = memory_type == GGML_TYPE_F32 ? 2 : 1;
+
+        // this is the total memory required to run the inference
+        const size_t mem_required =
+            ctx_size +
+            mmapped_size - vram_weights + // weights in VRAM not in memory
+            MEM_REQ_SCRATCH0(hparams.n_ctx).at(model.type) +
+            MEM_REQ_SCRATCH1().at(model.type) +
+            MEM_REQ_EVAL().at(model.type);
+
+        // this is the memory required by one llama_state
+        const size_t mem_required_state =
+            scale*hparams.kv_size();
+
+        fprintf(stderr, "%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
+                mem_required / 1024.0 / 1024.0, mem_required_state / 1024.0 / 1024.0);
+
+        (void) vram_scratch;
+        (void) n_batch;
+#ifdef GGML_USE_CUBLAS
+        if (low_vram) {
+            fprintf(stderr, "%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
+            ggml_cuda_set_scratch_size(0); // disable scratch
+        } else {
+            const size_t vram_scratch_base = VRAM_REQ_SCRATCH_BASE().at(model.type);
+            const size_t vram_scratch_per_context = VRAM_REQ_SCRATCH_PER_CONTEXT().at(model.type);
+            vram_scratch = n_batch * (vram_scratch_base + n_ctx * vram_scratch_per_context);
+            ggml_cuda_set_scratch_size(vram_scratch);
+            if (n_gpu_layers > 0) {
+                fprintf(stderr, "%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
+                        __func__, vram_scratch_base / kB, vram_scratch_per_context,
+                        (vram_scratch + MB - 1) / MB); // round up
+            }
+        }
+#endif // GGML_USE_CUBLAS
+
+#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
+        const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
+
+        fprintf(stderr, "%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
+        if (n_gpu_layers > (int) hparams.n_layer) {
+            fprintf(stderr, "%s: offloading non-repeating layers to GPU\n", __func__);
+        }
+        size_t vram_kv_cache = 0;
+
+#ifdef GGML_USE_CUBLAS
+        const int max_backend_supported_layers = hparams.n_layer + 3;
+        const int max_offloadable_layers = low_vram ? hparams.n_layer + 1 : hparams.n_layer + 3;
+        if (n_gpu_layers > (int) hparams.n_layer + 1) {
+            if (low_vram) {
+                fprintf(stderr, "%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
+            } else {
+                fprintf(stderr, "%s: offloading v cache to GPU\n", __func__);
+                vram_kv_cache += hparams.kv_size() / 2;
+            }
+        }
+        if (n_gpu_layers > (int) hparams.n_layer + 2) {
+            if (low_vram) {
+                fprintf(stderr, "%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
+            } else {
+                fprintf(stderr, "%s: offloading k cache to GPU\n", __func__);
+                vram_kv_cache += hparams.kv_size() / 2;
+            }
+        }
+#elif defined(GGML_USE_CLBLAST)
+        const int max_backend_supported_layers = hparams.n_layer + 1;
+        const int max_offloadable_layers = hparams.n_layer + 1;
+#endif // GGML_USE_CUBLAS
+
+        fprintf(stderr, "%s: offloaded %d/%d layers to GPU\n",
+                __func__, std::min(n_gpu_layers, max_offloadable_layers), max_backend_supported_layers);
+        fprintf(stderr, "%s: total VRAM used: %zu MB\n",
+                __func__, (vram_weights + vram_scratch + vram_kv_cache + MB - 1) / MB); // round up
+#else
+        (void) n_gpu_layers;
+#endif // defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
+    }
+
+    // populate `tensors_by_name`
+    for (gguf_load_tensor & lt : ml->tensors_map.tensors) {
+        model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
+    }
+
+    (void) tensor_split;
+#if defined(GGML_USE_CUBLAS)
+    {
+        ggml_cuda_set_tensor_split(tensor_split);
+    }
+#endif
+
+    ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
+
+    if (progress_callback) {
+        progress_callback(1.0f, progress_callback_user_data);
+    }
+
+    model.mapping = std::move(ml->mapping);
+
+    // loading time will be recalculate after the first eval, so
+    // we take page faults deferred by mmap() into consideration
+    model.t_load_us = ggml_time_us() - model.t_start_us;
+}
+
+static bool llama_model_load(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gqa,
+        float rms_norm_eps,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void *progress_callback_user_data) {
+    try {
+        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gqa, rms_norm_eps, n_gpu_layers, main_gpu, tensor_split, rope_freq_base, rope_freq_scale, low_vram, memory_type,
+                                  use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
+        return true;
+    } catch (const std::exception & err) {
+        fprintf(stderr, "error loading model: %s\n", err.what());
+        return false;
+    }
+}
+
+// evaluate the transformer
+//
+//   - lctx:      llama context
+//   - tokens:    new batch of tokens to process
+//   - embd       embeddings input
+//   - n_tokens   number of tokens
+//   - n_past:    the context size so far
+//   - n_threads: number of threads to use
+//
+static bool llama_eval_internal(
+         llama_context & lctx,
+     const llama_token * tokens,
+           const float * embd,
+                   int   n_tokens,
+                   int   n_past,
+                   int   n_threads,
+            const char * cgraph_fname) {
+
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
+#endif
+
+    const int64_t t_start_us = ggml_time_us();
+
+    const int N = n_tokens;
+
+    const auto & model   = lctx.model;
+    const auto & hparams = model.hparams;
+
+    const auto & kv_self = lctx.kv_self;
+
+    GGML_ASSERT(!!kv_self.ctx);
+
+    const int64_t n_embd      = hparams.n_embd;
+    const int64_t n_layer     = hparams.n_layer;
+    const int64_t n_ctx       = hparams.n_ctx;
+    const int64_t n_head      = hparams.n_head;
+    const int64_t n_head_kv   = hparams.n_head_kv;
+    const int64_t n_embd_head = hparams.n_embd_head();
+    const int64_t n_vocab     = hparams.n_vocab;
+    const int64_t n_embd_gqa  = hparams.n_embd_gqa();
+
+
+    GGML_ASSERT(n_embd_head == hparams.n_rot);
+
+    const float freq_base  = hparams.rope_freq_base;
+    const float freq_scale = hparams.rope_freq_scale;
+    const float rms_norm_eps = hparams.f_rms_norm_eps;
+
+    const int n_gpu_layers = model.n_gpu_layers;
+
+    auto & mem_per_token = lctx.mem_per_token;
+    auto & buf_compute   = lctx.buf_compute;
+
+    struct ggml_init_params params = {
+        /*.mem_size   =*/ buf_compute.size,
+        /*.mem_buffer =*/ buf_compute.addr,
+        /*.no_alloc   =*/ false,
+    };
+
+    struct ggml_context * ctx0 = ggml_init(params);
+
+    ggml_cgraph * gf = ggml_new_graph(ctx0);
+
+    // for big prompts, if BLAS is enabled, it is better to use only one thread
+    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
+    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
+
+    struct ggml_tensor * cur;
+    struct ggml_tensor * inpL;
+
+    if (tokens) {
+        struct ggml_tensor * inp_tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
+        memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
+        ggml_set_name(inp_tokens, "inp_tokens");
+
+        inpL = ggml_get_rows(ctx0, model.tok_embeddings, inp_tokens);
+    } else {
+#ifdef GGML_USE_MPI
+        GGML_ASSERT(false && "not implemented");
+#endif
+
+        inpL = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N);
+        memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
+    }
+
+    const int i_gpu_start = n_layer - n_gpu_layers;
+    (void) i_gpu_start;
+
+    // offload functions set the tensor output backend to GPU
+    // tensors are GPU-accelerated if any input or the output has been offloaded
+    //
+    // with the low VRAM option VRAM scratch is disabled in llama_load_model_internal
+    // in that case ggml_cuda_assign_buffers has no effect
+    offload_func_t offload_func_nr = llama_nop; // nr = non-repeating
+    offload_func_t offload_func_kq = llama_nop;
+    offload_func_t offload_func_v  = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer) {
+        offload_func_nr = ggml_cuda_assign_buffers;
+    }
+    if (n_gpu_layers > n_layer + 1) {
+        offload_func_v  = ggml_cuda_assign_buffers;
+    }
+    if (n_gpu_layers > n_layer + 2) {
+        offload_func_kq = ggml_cuda_assign_buffers;
+    }
+#endif // GGML_USE_CUBLAS
+
+    for (int il = 0; il < n_layer; ++il) {
+        ggml_format_name(inpL, "layer_inp_%d", il);
+
+        offload_func_t offload_func = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+        if (il >= i_gpu_start) {
+            offload_func = ggml_cuda_assign_buffers;
+        }
+#endif // GGML_USE_CUBLAS
+
+        struct ggml_tensor * inpSA = inpL;
+
+        lctx.use_buf(ctx0, 0);
+
+        // norm
+        {
+            cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+            offload_func(cur);
+            ggml_set_name(cur, "rms_norm_0");
+
+            // cur = cur*attention_norm(broadcasted)
+            cur = ggml_mul(ctx0, cur, model.layers[il].attention_norm);
+            offload_func(cur);
+            ggml_set_name(cur, "attention_norm_0");
+        }
+
+        // self-attention
+        {
+            // compute Q and K and RoPE them
+            struct ggml_tensor * tmpk = ggml_mul_mat(ctx0, model.layers[il].wk, cur);
+            offload_func_kq(tmpk);
+            ggml_set_name(tmpk, "tmpk");
+
+            struct ggml_tensor * tmpq = ggml_mul_mat(ctx0, model.layers[il].wq, cur);
+            offload_func_kq(tmpq);
+            ggml_set_name(tmpq, "tmpq");
+
+            struct ggml_tensor * Kcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpk, n_embd_head, n_head_kv, N), n_past, n_embd_head, 0, 0, freq_base, freq_scale);
+            offload_func_kq(Kcur);
+            ggml_set_name(Kcur, "Kcur");
+
+            struct ggml_tensor * Qcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpq, n_embd_head, n_head, N),    n_past, n_embd_head, 0, 0, freq_base, freq_scale);
+            offload_func_kq(Qcur);
+            ggml_set_name(Qcur, "Qcur");
+
+            // store key and value to memory
+            {
+                // compute the transposed [N, n_embd] V matrix
+
+                struct ggml_tensor * tmpv = ggml_mul_mat(ctx0, model.layers[il].wv, cur);
+                offload_func_v(tmpv);
+                ggml_set_name(tmpv, "tmpv");
+
+                struct ggml_tensor * Vcur = ggml_transpose(ctx0, ggml_reshape_2d(ctx0, tmpv, n_embd_gqa, N));
+                offload_func_v(Vcur);
+                ggml_set_name(Vcur, "Vcur");
+
+                struct ggml_tensor * k = ggml_view_1d(ctx0, kv_self.k, N*n_embd_gqa, (ggml_element_size(kv_self.k)*n_embd_gqa)*(il*n_ctx + n_past));
+                offload_func_kq(k);
+                ggml_set_name(k, "k");
+
+                struct ggml_tensor * v = ggml_view_2d(ctx0, kv_self.v, N, n_embd_gqa,
+                        (   n_ctx)*ggml_element_size(kv_self.v),
+                        (il*n_ctx)*ggml_element_size(kv_self.v)*n_embd_gqa + n_past*ggml_element_size(kv_self.v));
+                offload_func_v(v);
+                ggml_set_name(v, "v");
+
+                // important: storing RoPE-ed version of K in the KV cache!
+                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Kcur, k));
+                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Vcur, v));
+            }
+
+            struct ggml_tensor * Q =
+                ggml_permute(ctx0,
+                        Qcur,
+                        0, 2, 1, 3);
+            offload_func_kq(Q);
+            ggml_set_name(Q, "Q");
+
+            struct ggml_tensor * K =
+                ggml_permute(ctx0,
+                        ggml_reshape_3d(ctx0,
+                            ggml_view_1d(ctx0, kv_self.k, (n_past + N)*n_embd_gqa, il*n_ctx*ggml_element_size(kv_self.k)*n_embd_gqa),
+                            n_embd_head, n_head_kv, n_past + N),
+                        0, 2, 1, 3);
+            offload_func_kq(K);
+            ggml_set_name(K, "K");
+
+            // K * Q
+            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
+            offload_func_kq(KQ);
+            ggml_set_name(KQ, "KQ");
+
+            // KQ_scaled = KQ / sqrt(n_embd_head)
+            struct ggml_tensor * KQ_scale = ggml_new_f32(ctx0, 1.0f/sqrtf(float(n_embd)/n_head));
+            ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
+
+            // KQ_scaled shape [n_past + N, N, n_head, 1]
+            struct ggml_tensor * KQ_scaled = ggml_scale_inplace(ctx0, KQ, KQ_scale);
+            offload_func_kq(KQ_scaled);
+            ggml_set_name(KQ_scaled, "KQ_scaled");
+
+            // KQ_masked = mask_past(KQ_scaled)
+            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
+            offload_func_kq(KQ_masked);
+            ggml_set_name(KQ_masked, "KQ_masked");
+
+            // KQ = soft_max(KQ_masked)
+            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
+            offload_func_v(KQ_soft_max);
+            ggml_set_name(KQ_soft_max, "KQ_soft_max");
+
+            // split cached V into n_head heads
+            struct ggml_tensor * V =
+                ggml_view_3d(ctx0, kv_self.v,
+                        n_past + N, n_embd_head, n_head_kv,
+                        n_ctx*ggml_element_size(kv_self.v),
+                        n_ctx*ggml_element_size(kv_self.v)*n_embd_head,
+                        n_ctx*ggml_element_size(kv_self.v)*n_embd_gqa*il);
+            offload_func_v(V);
+            ggml_set_name(V, "V");
+
+#if 1
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
+            offload_func_v(KQV);
+            ggml_set_name(KQV, "KQV");
+#else
+            // make V contiguous in memory to speed up the matmul, however we waste time on the copy
+            // on M1 this is faster for the perplexity computation, but ~5% slower for the single-token generation
+            // is there a better way?
+            struct ggml_tensor * V_cont = ggml_cpy(ctx0, V, ggml_new_tensor_3d(ctx0, kv_self.v->type, n_past + N, n_embd_head, n_head));
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V_cont, KQ_soft_max);
+#endif
+
+            // KQV_merged = KQV.permute(0, 2, 1, 3)
+            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
+            offload_func_v(KQV_merged);
+            ggml_set_name(KQV_merged, "KQV_merged");
+
+            // cur = KQV_merged.contiguous().view(n_embd, N)
+            cur = ggml_cpy(ctx0,
+                    KQV_merged,
+                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+            offload_func_v(cur);
+            ggml_set_name(cur, "KQV_merged_contiguous");
+
+            // projection (no bias)
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].wo,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_wo");
+        }
+
+        lctx.use_buf(ctx0, 1);
+
+        struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpSA);
+        offload_func(inpFF);
+        ggml_set_name(inpFF, "inpFF");
+
+        // feed-forward network
+        {
+            // norm
+            {
+                cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps);
+                offload_func(cur);
+                ggml_set_name(cur, "rms_norm_1");
+
+                // cur = cur*ffn_norm(broadcasted)
+                cur = ggml_mul(ctx0, cur, model.layers[il].ffn_norm);
+                offload_func(cur);
+                ggml_set_name(cur, "ffn_norm");
+            }
+
+            struct ggml_tensor * tmp = ggml_mul_mat(ctx0,
+                    model.layers[il].w3,
+                    cur);
+            offload_func(tmp);
+            ggml_set_name(tmp, "result_w3");
+
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].w1,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_w1");
+
+            // SILU activation
+            cur = ggml_silu(ctx0, cur);
+            offload_func(cur);
+            ggml_set_name(cur, "silu");
+
+            cur = ggml_mul(ctx0, cur, tmp);
+            offload_func(cur);
+            ggml_set_name(cur, "silu_x_result_w3");
+
+            cur = ggml_mul_mat(ctx0,
+                    model.layers[il].w2,
+                    cur);
+            offload_func(cur);
+            ggml_set_name(cur, "result_w2");
+        }
+
+        cur = ggml_add(ctx0, cur, inpFF);
+        offload_func(cur);
+        ggml_set_name(cur, "inpFF_+_result_w2");
+
+        // input for next layer
+        inpL = cur;
+    }
+
+    lctx.use_buf(ctx0, 0);
+
+    // used at the end to optionally extract the embeddings
+    struct ggml_tensor * embeddings = NULL;
+
+    // norm
+    {
+        cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+        offload_func_nr(cur);
+        ggml_set_name(cur, "rms_norm_2");
+
+        // cur = cur*norm(broadcasted)
+        cur = ggml_mul(ctx0, cur, model.norm);
+        // offload_func_nr(cur); // TODO CPU + GPU mirrored backend
+        ggml_set_name(cur, "result_norm");
+
+        embeddings = cur;
+    }
+
+    // lm_head
+    cur = ggml_mul_mat(ctx0, model.output, cur);
+    ggml_set_name(cur, "result_output");
+
+    lctx.use_buf(ctx0, -1);
+
+    // logits -> probs
+    //cur = ggml_soft_max_inplace(ctx0, cur);
+
+    // run the computation
+    ggml_build_forward_expand(gf, cur);
+
+    // fprintf(stderr, "graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf.n_nodes, gf.n_leafs);
+
+#if GGML_USE_MPI
+    ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);
+#endif
+
+#ifdef GGML_USE_METAL
+    if (lctx.ctx_metal && N == 1) {
+        if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
+            ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
+        }
+        ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
+        ggml_metal_graph_compute(lctx.ctx_metal, gf);
+        ggml_metal_get_tensor   (lctx.ctx_metal, cur);
+    } else {
+        // IMPORTANT:
+        // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
+        // ggml_graph_compute(). It uses Apple's Accelerate CBLAS API which takes advantage of the ANE or the AMX
+        // coprocessor.
+        //
+        // When we implement Matrix x Matrix Metal multiplication, we can avoid this branch.
+        // But for now, we have focused only on Matrix x Vector Metal multiplication.
+        //
+        // TODO: avoid these syncs via shared memory (ref #1696)
+        //
+        if (lctx.ctx_metal) {
+            // We need to sync the GPU KV cache with the CPU KV cache
+            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.k);
+            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.v);
+        }
+
+        ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
+    }
+#else
+    ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
+#endif
+
+#if GGML_USE_MPI
+    ggml_mpi_graph_compute_post(lctx.ctx_mpi, gf, n_layer);
+#endif
+
+    // update kv token count
+    lctx.kv_self.n = n_past + N;
+
+    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
+
+    if (cgraph_fname) {
+        ggml_graph_export(gf, cgraph_fname);
+    }
+
+#ifdef GGML_PERF
+    // print timing information per ggml operation (for debugging purposes)
+    // requires GGML_PERF to be defined
+    ggml_graph_print(gf);
+#endif
+
+    // plot the computation graph in dot format (for debugging purposes)
+    //if (n_past%100 == 0) {
+    //    ggml_graph_dump_dot(gf, NULL, "llama.dot");
+    //}
+
+    // extract logits
+    {
+        auto & logits_out = lctx.logits;
+
+        if (lctx.logits_all) {
+            logits_out.resize(n_vocab * N);
+            memcpy(logits_out.data(), (float *) ggml_get_data(res), sizeof(float)*n_vocab*N);
+        } else {
+            // return result for just the last token
+            logits_out.resize(n_vocab);
+            memcpy(logits_out.data(), (float *) ggml_get_data(res) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
+        }
+    }
+
+    // extract embeddings
+    if (!lctx.embedding.empty()) {
+        auto & embedding_out = lctx.embedding;
+
+        embedding_out.resize(n_embd);
+        memcpy(embedding_out.data(), (float *) ggml_get_data(embeddings) + (n_embd*(N - 1)), sizeof(float)*n_embd);
+    }
+
+    if (mem_per_token == 0) {
+        mem_per_token = ggml_used_mem(ctx0)/N;
+    }
+
+#if 0
+    printf("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
+            ggml_used_mem(ctx0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(1)/1024.0/1024.0,
+            lctx.work_buffer.size()/1024.0/1024.0,
+            n_past, N);
+#endif
+
+    ggml_free(ctx0);
+
+    // measure the performance only for the single-token evals
+    if (N == 1) {
+        lctx.t_eval_us += ggml_time_us() - t_start_us;
+        lctx.n_eval++;
+    }
+    else if (N > 1) {
+        lctx.t_p_eval_us += ggml_time_us() - t_start_us;
+        lctx.n_p_eval += N;
+    }
+
+    return true;
+}
+
+//
+// tokenizer
+//
+
+static size_t utf8_len(char src) {
+    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
+    return lookup[highbits];
+}
+
+struct llama_sp_symbol {
+    using index = int;
+    index prev;
+    index next;
+    const char * text;
+    size_t n;
+};
+
+static_assert(std::is_trivially_copyable<llama_sp_symbol>::value, "llama_sp_symbol is not trivially copyable");
+
+struct llama_sp_bigram {
+    struct comparator {
+        bool operator()(llama_sp_bigram & l, llama_sp_bigram & r) {
+            return (l.score < r.score) || (l.score == r.score && l.left > r.left);
+        }
+    };
+    using queue_storage = std::vector<llama_sp_bigram>;
+    using queue = std::priority_queue<llama_sp_bigram, queue_storage, comparator>;
+    llama_sp_symbol::index left;
+    llama_sp_symbol::index right;
+    float score;
+    size_t size;
+};
+
+// original implementation:
+// https://github.com/ggerganov/llama.cpp/commit/074bea2eb1f1349a0118239c4152914aecaa1be4
+struct llama_tokenizer {
+    llama_tokenizer(const llama_vocab & vocab): vocab_(vocab) {}
+
+    void tokenize(const std::string & text, std::vector<llama_vocab::id> & output) {
+        // split string into utf8 chars
+        int index = 0;
+        size_t offs = 0;
+        while (offs < text.size()) {
+            llama_sp_symbol sym;
+            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
+            sym.text = text.c_str() + offs;
+            sym.n = char_len;
+            offs += char_len;
+            sym.prev = index - 1;
+            sym.next = offs == text.size() ? -1 : index + 1;
+            index++;
+            symbols_.emplace_back(sym);
+        }
+
+        // seed the work queue with all possible 2-character tokens.
+        for (size_t i = 1; i < symbols_.size(); ++i) {
+            try_add_bigram(i - 1, i);
+        }
+
+        // keep substituting the highest frequency pairs for as long as we can.
+        while (!work_queue_.empty()) {
+            auto bigram = work_queue_.top();
+            work_queue_.pop();
+
+            auto & left_sym = symbols_[bigram.left];
+            auto & right_sym = symbols_[bigram.right];
+
+            // if one of the symbols already got merged, skip it.
+            if (left_sym.n == 0 || right_sym.n == 0 ||
+                left_sym.n + right_sym.n != bigram.size) {
+                continue;
+            }
+
+            // merge the right sym into the left one
+            left_sym.n += right_sym.n;
+            right_sym.n = 0;
+
+            //printf("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
+
+            // remove the right sym from the chain
+            left_sym.next = right_sym.next;
+            if (right_sym.next >= 0) {
+                symbols_[right_sym.next].prev = bigram.left;
+            }
+
+            // find more substitutions
+            try_add_bigram(left_sym.prev, bigram.left);
+            try_add_bigram(bigram.left, left_sym.next);
+        }
+
+        for (int i = 0; i != -1; i = symbols_[i].next) {
+            auto & symbol = symbols_[i];
+            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
+
+            if (token == vocab_.token_to_id.end()) {
+                // output any symbols that did not form tokens as bytes.
+                for (int j = 0; j < (int) symbol.n; ++j) {
+                    llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
+                    output.push_back(token_id);
+                }
+            } else {
+                output.push_back((*token).second);
+            }
+        }
+    }
+
+private:
+    void try_add_bigram(int left, int right) {
+        if (left == -1 || right == -1) {
+            return;
+        }
+
+        const std::string text = std::string(symbols_[left].text, symbols_[left].n + symbols_[right].n);
+        auto token = vocab_.token_to_id.find(text);
+
+        if (token == vocab_.token_to_id.end()) {
+            return;
+        }
+
+        if (static_cast<size_t>((*token).second) >= vocab_.id_to_token.size()) {
+            return;
+        }
+
+        const auto &tok_score = vocab_.id_to_token[(*token).second];
+
+        llama_sp_bigram bigram;
+        bigram.left = left;
+        bigram.right = right;
+        bigram.score = tok_score.score;
+        bigram.size = text.size();
+        work_queue_.push(bigram);
+    }
+
+    const llama_vocab & vocab_;
+    std::vector<llama_sp_symbol> symbols_;
+    llama_sp_bigram::queue work_queue_;
+};
+
+static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
+    llama_tokenizer tokenizer(vocab);
+    std::vector<llama_vocab::id> output;
+
+    if (text.empty()) {
+        return output;
+    }
+
+    if (bos) {
+        output.push_back(llama_token_bos());
+    }
+
+    tokenizer.tokenize(text, output);
+    return output;
+}
+
+//
+// grammar - internal
+//
+
+struct llama_grammar {
+    const std::vector<std::vector<llama_grammar_element>>   rules;
+    std::vector<std::vector<const llama_grammar_element *>> stacks;
+};
+
+struct llama_grammar_candidate {
+    size_t           index;
+    const uint32_t * code_points;
+};
+
+// NOTE: assumes valid utf8 (but checks for overrun)
+// adds a terminating 0 for use as pointer
+std::vector<uint32_t> decode_utf8(const char * src) {
+    static const int      lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    const char          * pos      = src;
+    std::vector<uint32_t> code_points;
+    while (*pos != 0) {
+        uint8_t  first_byte = static_cast<uint8_t>(*pos);
+        uint8_t  highbits   = first_byte >> 4;
+        int      len        = lookup[highbits];
+        uint8_t  mask       = (1 << (8 - len)) - 1;
+        uint32_t value      = first_byte & mask;
+        const char * end    = pos + len; // may overrun!
+        ++pos;
+        for ( ; pos < end && *pos != 0; ++pos) {
+            value = (value << 6) + (static_cast<uint8_t>(*pos) & 0x3F);
+        }
+        code_points.push_back(value);
+    }
+    code_points.push_back(0);
+    return code_points;
+}
+
+// returns true iff pos points to the end of one of the definitions of a rule
+static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
+    switch (pos->type) {
+        case LLAMA_GRETYPE_END: return true;
+        case LLAMA_GRETYPE_ALT: return true;
+        default:                return false;
+    }
+}
+
+// returns true iff chr satisfies the char range at pos (regular or inverse range)
+// asserts that pos is pointing to a char range element
+static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
+        const llama_grammar_element * pos,
+        const uint32_t                chr) {
+
+    bool found            = false;
+    bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
+    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT);
+
+    do {
+        if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
+            // inclusive range, e.g. [a-z]
+            found = found || (pos->value <= chr && chr <= pos[1].value);
+            pos += 2;
+        } else {
+            // exact char match, e.g. [a] or "a"
+            found = found || pos->value == chr;
+            pos += 1;
+        }
+    } while (pos->type == LLAMA_GRETYPE_CHAR_ALT);
+
+    return std::make_pair(found == is_positive_char, pos);
+}
+
+// transforms a grammar pushdown stack into N possible stacks, all ending
+// at a character range (terminal element)
+static void llama_grammar_advance_stack(
+        const std::vector<std::vector<llama_grammar_element>>   & rules,
+        const std::vector<const llama_grammar_element *>        & stack,
+        std::vector<std::vector<const llama_grammar_element *>> & new_stacks) {
+
+    if (stack.empty()) {
+        new_stacks.push_back(stack);
+        return;
+    }
+
+    const llama_grammar_element * pos = stack.back();
+
+    switch (pos->type) {
+        case LLAMA_GRETYPE_RULE_REF: {
+            const size_t                  rule_id = static_cast<size_t>(pos->value);
+            const llama_grammar_element * subpos  = rules[rule_id].data();
+            do {
+                // init new stack without the top (pos)
+                std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
+                if (!llama_grammar_is_end_of_sequence(pos + 1)) {
+                    // if this rule ref is followed by another element, add that to stack
+                    new_stack.push_back(pos + 1);
+                }
+                if (!llama_grammar_is_end_of_sequence(subpos)) {
+                    // if alternate is nonempty, add to stack
+                    new_stack.push_back(subpos);
+                }
+                llama_grammar_advance_stack(rules, new_stack, new_stacks);
+                while (!llama_grammar_is_end_of_sequence(subpos)) {
+                    // scan to end of alternate def
+                    subpos++;
+                }
+                if (subpos->type == LLAMA_GRETYPE_ALT) {
+                    // there's another alternate def of this rule to process
+                    subpos++;
+                } else {
+                    break;
+                }
+            } while (true);
+            break;
+        }
+        case LLAMA_GRETYPE_CHAR:
+        case LLAMA_GRETYPE_CHAR_NOT:
+            new_stacks.push_back(stack);
+            break;
+        default:
+            // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range
+            // (LLAMA_GRETYPE_CHAR_ALT, LLAMA_GRETYPE_CHAR_RNG_UPPER); stack should never be left on
+            // those
+            GGML_ASSERT(false);
+    }
+}
+
+// takes a set of possible pushdown stacks on a grammar, which are required to
+// be positioned at a character range (see `llama_grammar_advance_stack`), and
+// produces the N possible stacks if the given char is accepted at those
+// positions
+static std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const uint32_t                                                  chr) {
+
+    std::vector<std::vector<const llama_grammar_element *>> new_stacks;
+
+    for (const auto & stack : stacks) {
+        if (stack.empty()) {
+            continue;
+        }
+
+        auto match = llama_grammar_match_char(stack.back(), chr);
+        if (match.first) {
+            const llama_grammar_element * pos = match.second;
+
+            // update top of stack to next element, if any
+            std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
+            if (!llama_grammar_is_end_of_sequence(pos)) {
+                new_stack.push_back(pos);
+            }
+            llama_grammar_advance_stack(rules, new_stack, new_stacks);
+        }
+    }
+
+    return new_stacks;
+}
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const std::vector<llama_grammar_candidate>                    & candidates);
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_stack(
+        const std::vector<std::vector<llama_grammar_element>> & rules,
+        const std::vector<const llama_grammar_element *>      & stack,
+        const std::vector<llama_grammar_candidate>            & candidates) {
+
+    std::vector<llama_grammar_candidate> rejects;
+
+    if (stack.empty()) {
+        // accept nothing; EOS is handled elsewhere
+        rejects.insert(rejects.end(), candidates.begin(), candidates.end());
+        return rejects;
+    }
+
+    const llama_grammar_element * stack_pos = stack.back();
+
+    std::vector<llama_grammar_candidate> next_candidates;
+    for (auto tok : candidates) {
+        if (llama_grammar_match_char(stack_pos, tok.code_points[0]).first) {
+            if (tok.code_points[1] != 0) {
+                next_candidates.push_back({ tok.index, tok.code_points + 1 });
+            }
+        } else {
+            rejects.push_back(tok);
+        }
+    }
+
+    auto stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
+
+    // update top of stack to next element, if any
+    std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
+    if (!llama_grammar_is_end_of_sequence(stack_pos_after)) {
+        stack_after.push_back(stack_pos_after);
+    }
+    std::vector<std::vector<const llama_grammar_element *>> next_stacks;
+    llama_grammar_advance_stack(rules, stack_after, next_stacks);
+
+    auto next_rejects = llama_grammar_reject_candidates(rules, next_stacks, next_candidates);
+    for (auto tok : next_rejects) {
+        rejects.push_back({ tok.index, tok.code_points - 1 });
+    }
+
+    return rejects;
+}
+
+static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
+        const std::vector<std::vector<llama_grammar_element>>         & rules,
+        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
+        const std::vector<llama_grammar_candidate>                    & candidates) {
+    GGML_ASSERT(!stacks.empty()); // REVIEW
+
+    if (candidates.empty()) {
+        return std::vector<llama_grammar_candidate>();
+    }
+
+    auto rejects = llama_grammar_reject_candidates_for_stack(rules, stacks.front(), candidates);
+
+    for (size_t i = 1, size = stacks.size(); i < size; ++i) {
+        rejects = llama_grammar_reject_candidates_for_stack(rules, stacks[i], rejects);
+    }
+    return rejects;
+}
+
+//
+// grammar - external
+//
+
+struct llama_grammar * llama_grammar_init(
+            const llama_grammar_element ** rules,
+                                 size_t    n_rules,
+                                 size_t    start_rule_index) {
+    const llama_grammar_element * pos;
+
+    // copy rule definitions into vectors
+    std::vector<std::vector<llama_grammar_element>> vec_rules(n_rules);
+    for (size_t i = 0; i < n_rules; i++) {
+        for (pos = rules[i]; pos->type != LLAMA_GRETYPE_END; pos++) {
+            vec_rules[i].push_back(*pos);
+        }
+        vec_rules[i].push_back({LLAMA_GRETYPE_END, 0});
+    }
+
+    // loop over alternates of start rule to build initial stacks
+    std::vector<std::vector<const llama_grammar_element *>> stacks;
+    pos = rules[start_rule_index];
+    do {
+        std::vector<const llama_grammar_element *> stack;
+        if (!llama_grammar_is_end_of_sequence(pos)) {
+            // if alternate is nonempty, add to stack
+            stack.push_back(pos);
+        }
+        llama_grammar_advance_stack(vec_rules, stack, stacks);
+        while (!llama_grammar_is_end_of_sequence(pos)) {
+            // scan to end of alternate def
+            pos++;
+        }
+        if (pos->type == LLAMA_GRETYPE_ALT) {
+            // there's another alternate def of this rule to process
+            pos++;
+        } else {
+            break;
+        }
+    } while (true);
+
+    return new llama_grammar{ std::move(vec_rules), std::move(stacks) };
+}
+
+void llama_grammar_free(struct llama_grammar * grammar) {
+    delete grammar;
+}
+
+//
+// sampling
+//
+
+void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
+    assert(candidates->size > 0);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Sort the logits in descending order
+    if (!candidates->sorted) {
+        std::sort(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
+            return a.logit > b.logit;
+        });
+        candidates->sorted = true;
+    }
+
+    float max_l = candidates->data[0].logit;
+    float cum_sum = 0.0f;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        float p = expf(candidates->data[i].logit - max_l);
+        candidates->data[i].p = p;
+        cum_sum += p;
+    }
+    for (size_t i = 0; i < candidates->size; ++i) {
+        candidates->data[i].p /= cum_sum;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    k = std::max(k, (int) min_keep);
+    k = std::min(k, (int) candidates->size);
+
+    // Sort scores in descending order
+    if (!candidates->sorted) {
+        auto comp = [](const llama_token_data & a, const llama_token_data & b) {
+            return a.logit > b.logit;
+        };
+        if (k == (int) candidates->size) {
+            std::sort(candidates->data, candidates->data + candidates->size, comp);
+        } else {
+            std::partial_sort(candidates->data, candidates->data + k, candidates->data + candidates->size, comp);
+        }
+        candidates->sorted = true;
+    }
+    candidates->size = k;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
+    if (p >= 1.0f) {
+        return;
+    }
+
+    llama_sample_softmax(ctx, candidates);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Compute the cumulative probabilities
+    float cum_sum = 0.0f;
+    size_t last_idx = candidates->size;
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        cum_sum += candidates->data[i].p;
+
+        // Check if the running sum is at least p or if we have kept at least min_keep tokens
+        // we set the last index to i+1 to indicate that the current iterate should be included in the set
+        if (cum_sum >= p && i + 1 >= min_keep) {
+            last_idx = i + 1;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the top-p tokens
+    candidates->size = last_idx;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep) {
+    if (z >= 1.0f || candidates->size <= 2) {
+        return;
+    }
+
+    llama_sample_softmax(nullptr, candidates);
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Compute the first and second derivatives
+    std::vector<float> first_derivatives(candidates->size - 1);
+    std::vector<float> second_derivatives(candidates->size - 2);
+
+    for (size_t i = 0; i < first_derivatives.size(); ++i) {
+        first_derivatives[i] = candidates->data[i].p - candidates->data[i + 1].p;
+    }
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        second_derivatives[i] = first_derivatives[i] - first_derivatives[i + 1];
+    }
+
+    // Calculate absolute value of second derivatives
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        second_derivatives[i] = abs(second_derivatives[i]);
+    }
+
+    // Normalize the second derivatives
+    {
+        const float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
+
+        if (second_derivatives_sum > 1e-6f) {
+            for (float & value : second_derivatives) {
+                value /= second_derivatives_sum;
+            }
+        } else {
+            for (float & value : second_derivatives) {
+                value = 1.0f / second_derivatives.size();
+            }
+        }
+    }
+
+    float cum_sum = 0.0f;
+    size_t last_idx = candidates->size;
+    for (size_t i = 0; i < second_derivatives.size(); ++i) {
+        cum_sum += second_derivatives[i];
+
+        // Check if the running sum is greater than z or if we have kept at least min_keep tokens
+        if (cum_sum > z && i >= min_keep) {
+            last_idx = i;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the tokens above the tail location
+    candidates->size = last_idx;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+
+void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
+    // Reference implementation:
+    // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
+    if (p >= 1.0f) {
+        return;
+    }
+
+    // Compute the softmax of logits and calculate entropy
+    llama_sample_softmax(nullptr, candidates);
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    float entropy = 0.0f;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        entropy += -candidates->data[i].p * logf(candidates->data[i].p);
+    }
+
+    // Compute the absolute difference between negative log probability and entropy for each candidate
+    std::vector<float> shifted_scores;
+    for (size_t i = 0; i < candidates->size; ++i) {
+        float shifted_score = fabsf(-logf(candidates->data[i].p) - entropy);
+        shifted_scores.push_back(shifted_score);
+    }
+
+    // Sort tokens based on the shifted_scores and their corresponding indices
+    std::vector<size_t> indices(candidates->size);
+    std::iota(indices.begin(), indices.end(), 0);
+
+    std::sort(indices.begin(), indices.end(), [&](size_t a, size_t b) {
+        return shifted_scores[a] < shifted_scores[b];
+    });
+
+    // Compute the cumulative probabilities
+    float cum_sum = 0.0f;
+    size_t last_idx = indices.size();
+
+    for (size_t i = 0; i < indices.size(); ++i) {
+        size_t idx = indices[i];
+        cum_sum += candidates->data[idx].p;
+
+        // Check if the running sum is greater than typical or if we have kept at least min_keep tokens
+        if (cum_sum > p && i >= min_keep - 1) {
+            last_idx = i + 1;
+            break;
+        }
+    }
+
+    // Resize the output vector to keep only the locally typical tokens
+    std::vector<llama_token_data> new_candidates;
+    for (size_t i = 0; i < last_idx; ++i) {
+        size_t idx = indices[i];
+        new_candidates.push_back(candidates->data[idx]);
+    }
+
+    // Replace the data in candidates with the new_candidates data
+    std::copy(new_candidates.begin(), new_candidates.end(), candidates->data);
+    candidates->size = new_candidates.size();
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates_p, float temp) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    for (size_t i = 0; i < candidates_p->size; ++i) {
+        candidates_p->data[i].logit /= temp;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty) {
+    if (last_tokens_size == 0 || penalty == 1.0f) {
+        return;
+    }
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        const auto * token_iter = std::find(last_tokens, last_tokens + last_tokens_size, candidates->data[i].id);
+        if (token_iter == last_tokens + last_tokens_size) {
+            continue;
+        }
+
+        // The academic publication that described this technique actually just only divided, but that would cause tokens with negative logits to become more likely, which is obviously wrong.
+        // This is common fix for this problem, which is to multiply by the penalty instead of dividing.
+        if (candidates->data[i].logit <= 0) {
+            candidates->data[i].logit *= penalty;
+        } else {
+            candidates->data[i].logit /= penalty;
+        }
+    }
+
+    candidates->sorted = false;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens_p, size_t last_tokens_size, float alpha_frequency, float alpha_presence) {
+    if (last_tokens_size == 0 || (alpha_frequency == 0.0f && alpha_presence == 0.0f)) {
+        return;
+    }
+
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Create a frequency map to count occurrences of each token in last_tokens
+    std::unordered_map<llama_token, int> token_count;
+    for (size_t i = 0; i < last_tokens_size; ++i) {
+        token_count[last_tokens_p[i]]++;
+    }
+
+    // Apply frequency and presence penalties to the candidates
+    for (size_t i = 0; i < candidates->size; ++i) {
+        auto token_iter = token_count.find(candidates->data[i].id);
+        if (token_iter == token_count.end()) {
+            continue;
+        }
+
+        int count = token_iter->second;
+        candidates->data[i].logit -= float(count) * alpha_frequency + float(count > 0) * alpha_presence;
+    }
+
+    candidates->sorted = false;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
+    assert(ctx);
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    bool allow_eos = false;
+    for (const auto & stack : grammar->stacks) {
+        if (stack.empty()) {
+            allow_eos = true;
+            break;
+        }
+    }
+
+    const llama_token eos = llama_token_eos();
+
+    std::vector<std::vector<uint32_t>>   candidates_decoded;
+    std::vector<llama_grammar_candidate> candidates_grammar;
+
+    for (size_t i = 0; i < candidates->size; ++i) {
+        const llama_token id  = candidates->data[i].id;
+        const char *      str = llama_token_to_str(ctx, id);
+        if (id == eos) {
+            if (!allow_eos) {
+                candidates->data[i].logit = -INFINITY;
+            }
+        } else if (*str == 0) {
+            candidates->data[i].logit = -INFINITY;
+        } else {
+            candidates_decoded.push_back(decode_utf8(str));
+            candidates_grammar.push_back({ i, candidates_decoded.back().data() });
+        }
+    }
+
+    const auto rejects =
+        llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
+    for (auto & reject : rejects) {
+        candidates->data[reject.index].logit = -INFINITY;
+    }
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+}
+
+static void llama_log_softmax(float * array, size_t size) {
+    float max_l = *std::max_element(array, array + size);
+    float sum = 0.f;
+    for (size_t i = 0; i < size; ++i) {
+        float p = expf(array[i] - max_l);
+        sum += p;
+        array[i] = p;
+    }
+
+    for (size_t i = 0; i < size; ++i) {
+        array[i] = logf(array[i] / sum);
+    }
+}
+
+void llama_sample_classifier_free_guidance(
+          struct llama_context * ctx,
+        llama_token_data_array * candidates,
+          struct llama_context * guidance_ctx,
+                         float   scale) {
+    int64_t t_start_sample_us = ggml_time_us();
+
+    assert(ctx);
+    auto n_vocab = llama_n_vocab(ctx);
+    assert(n_vocab == (int)candidates->size);
+    assert(!candidates->sorted);
+
+    std::vector<float> logits_base;
+    logits_base.reserve(candidates->size);
+    for (size_t i = 0; i < candidates->size; ++i) {
+        logits_base.push_back(candidates->data[i].logit);
+    }
+    llama_log_softmax(logits_base.data(), candidates->size);
+
+    float* logits_guidance = llama_get_logits(guidance_ctx);
+    llama_log_softmax(logits_guidance, n_vocab);
+
+    for (int i = 0; i < n_vocab; ++i) {
+        float logit_guidance = logits_guidance[i];
+        float logit_base = logits_base[i];
+        candidates->data[i].logit = scale * (logit_base - logit_guidance) + logit_guidance;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+}
+
+llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu) {
+    assert(ctx);
+    auto N = float(llama_n_vocab(ctx));
+    int64_t t_start_sample_us;
+    t_start_sample_us = ggml_time_us();
+
+    llama_sample_softmax(nullptr, candidates);
+
+    // Estimate s_hat using the most probable m tokens
+    float s_hat = 0.0;
+    float sum_ti_bi = 0.0;
+    float sum_ti_sq = 0.0;
+    for (size_t i = 0; i < size_t(m - 1) && i < candidates->size - 1; ++i) {
+        float t_i = logf(float(i + 2) / float(i + 1));
+        float b_i = logf(candidates->data[i].p / candidates->data[i + 1].p);
+        sum_ti_bi += t_i * b_i;
+        sum_ti_sq += t_i * t_i;
+    }
+    s_hat = sum_ti_bi / sum_ti_sq;
+
+    // Compute k from the estimated s_hat and target surprise value
+    float epsilon_hat = s_hat - 1;
+    float k = powf((epsilon_hat * powf(2, *mu)) / (1 - powf(N, -epsilon_hat)), 1 / s_hat);
+
+    // Sample the next word X using top-k sampling
+    llama_sample_top_k(nullptr, candidates, int(k), 1);
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    llama_token X = llama_sample_token(ctx, candidates);
+    t_start_sample_us = ggml_time_us();
+
+    // Compute error as the difference between observed surprise and target surprise value
+    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return candidate.id == X;
+    }));
+    float observed_surprise = -log2f(candidates->data[X_idx].p);
+    float e = observed_surprise - tau;
+
+    // Update mu using the learning rate and error
+    *mu = *mu - eta * e;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    return X;
+}
+
+llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu) {
+    int64_t t_start_sample_us;
+    t_start_sample_us = ggml_time_us();
+
+    llama_sample_softmax(ctx, candidates);
+
+    // Truncate the words with surprise values greater than mu
+    candidates->size = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return -log2f(candidate.p) > *mu;
+    }));
+
+    if (candidates->size == 0) {
+        candidates->size = 1;
+    }
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+
+    // Normalize the probabilities of the remaining words
+    llama_sample_softmax(ctx, candidates);
+
+    // Sample the next word X from the remaining words
+    llama_token X = llama_sample_token(ctx, candidates);
+    t_start_sample_us = ggml_time_us();
+
+    // Compute error as the difference between observed surprise and target surprise value
+    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
+        return candidate.id == X;
+    }));
+    float observed_surprise = -log2f(candidates->data[X_idx].p);
+    float e = observed_surprise - tau;
+
+    // Update mu using the learning rate and error
+    *mu = *mu - eta * e;
+
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    }
+    return X;
+}
+
+llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    // Find max element
+    auto * max_iter = std::max_element(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
+        return a.logit < b.logit;
+    });
+
+    llama_token result = max_iter->id;
+    if (ctx) {
+        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+        ctx->n_sample++;
+    }
+    return result;
+}
+
+llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
+    assert(ctx);
+    const int64_t t_start_sample_us = ggml_time_us();
+    llama_sample_softmax(nullptr, candidates);
+
+    std::vector<float> probs;
+    probs.reserve(candidates->size);
+    for (size_t i = 0; i < candidates->size; ++i) {
+        probs.push_back(candidates->data[i].p);
+    }
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    auto & rng = ctx->rng;
+    int idx = dist(rng);
+
+    llama_token result = candidates->data[idx].id;
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+    ctx->n_sample++;
+    return result;
+}
+
+void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
+    const int64_t t_start_sample_us = ggml_time_us();
+
+    if (token == llama_token_eos()) {
+        for (const auto & stack : grammar->stacks) {
+            if (stack.empty()) {
+                return;
+            }
+        }
+        GGML_ASSERT(false);
+    }
+
+    const char * str = llama_token_to_str(ctx, token);
+    // Note terminating 0 in decoded string
+    auto code_points = decode_utf8(str);
+    for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
+        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
+    }
+    GGML_ASSERT(!grammar->stacks.empty());
+
+    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
+}
+
+//
+// quantization
+//
+
+static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
+    if (output.size < nelements * sizeof(float)) {
+        output.resize(nelements * sizeof(float));
+    }
+    float * f32_output = (float *) output.addr;
+
+    ggml_type_traits_t qtype;
+    if (ggml_is_quantized(tensor.type)) {
+        qtype = ggml_internal_get_type_traits(tensor.type);
+        if (qtype.to_float == NULL) {
+            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor.type)));
+        }
+    } else if (tensor.type != GGML_TYPE_F16) {
+        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor.type)));
+    }
+
+    if (nthread < 2) {
+        if (tensor.type == GGML_TYPE_F16) {
+            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor.data, f32_output, nelements);
+        } else if (ggml_is_quantized(tensor.type)) {
+            qtype.to_float(tensor.data, f32_output, nelements);
+        } else {
+            GGML_ASSERT(false); // unreachable
+        }
+        return;
+    }
+
+    auto block_size = tensor.type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor.type);
+    auto block_size_bytes = ggml_type_size(tensor.type);
+
+    GGML_ASSERT(nelements % block_size == 0);
+    auto nblocks = nelements / block_size;
+    auto blocks_per_thread = nblocks / nthread;
+    auto spare_blocks = nblocks - (blocks_per_thread * nthread); // if blocks aren't divisible by thread count
+
+    std::vector<std::thread> workers;
+    for (auto tnum = 0, in_buff_offs = 0, out_buff_offs = 0; tnum < nthread; tnum++) {
+        auto thr_blocks = blocks_per_thread + (tnum == nthread - 1 ? spare_blocks : 0); // num blocks for this thread
+        auto thr_elems = thr_blocks * block_size; // number of elements for this thread
+        auto thr_block_bytes = thr_blocks * block_size_bytes; // number of input bytes for this thread
+
+        auto compute = [qtype] (ggml_type typ, uint8_t * inbuf, float * outbuf, int nels) {
+            if (typ == GGML_TYPE_F16) {
+                ggml_fp16_to_fp32_row((ggml_fp16_t *)inbuf, outbuf, nels);
+            } else {
+                qtype.to_float(inbuf, outbuf, nels);
+            }
+        };
+        workers.push_back(std::thread(compute, tensor.type, tensor.data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
+        in_buff_offs += thr_block_bytes;
+        out_buff_offs += thr_elems;
+    }
+    for (auto & worker : workers) {
+        worker.join();
+    }
+
+}
+
+static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
+    ggml_type quantized_type;
+    llama_ftype ftype = params->ftype;
+    int nthread = params->nthread;
+
+    switch (params->ftype) {
+        case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
+        case LLAMA_FTYPE_MOSTLY_Q4_1: quantized_type = GGML_TYPE_Q4_1; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_0: quantized_type = GGML_TYPE_Q5_0; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_1: quantized_type = GGML_TYPE_Q5_1; break;
+        case LLAMA_FTYPE_MOSTLY_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;
+        case LLAMA_FTYPE_MOSTLY_F16:  quantized_type = GGML_TYPE_F16;  break;
+        case LLAMA_FTYPE_ALL_F32:     quantized_type = GGML_TYPE_F32;  break;
+
+#ifdef GGML_USE_K_QUANTS
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K:   quantized_type = GGML_TYPE_Q2_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M:
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: quantized_type = GGML_TYPE_Q3_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: quantized_type = GGML_TYPE_Q4_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S:
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break;
+        case LLAMA_FTYPE_MOSTLY_Q6_K:   quantized_type = GGML_TYPE_Q6_K; break;
+#endif
+        default: throw std::runtime_error(format("invalid output file type %d\n", ftype));
+    }
+
+    if (nthread <= 0) {
+        nthread = std::thread::hardware_concurrency();
+    }
+
+    std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
+    gguf_file_saver file_saver(fname_out.c_str(), model_loader->file_loader.get(), params->ftype);
+
+#ifdef GGML_USE_K_QUANTS
+    int n_attention_wv    = 0;
+    int n_feed_forward_w2 = 0;
+    for (auto& tensor : model_loader->tensors_map.tensors) {
+        if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+            ++n_attention_wv;
+        }
+        else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+            ++n_feed_forward_w2;
+        }
+    }
+
+    int i_attention_wv = 0;
+    int i_feed_forward_w2 = 0;
+#endif
+
+    size_t total_size_org = 0;
+    size_t total_size_new = 0;
+    std::vector<int64_t> hist_all(1 << 4, 0);
+
+    std::vector<std::thread> workers;
+    std::mutex mutex;
+
+    auto use_more_bits = [] (int i_layer, int num_layers) -> bool {
+        return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
+    };
+
+    size_t idx = 0;
+    for (gguf_load_tensor & tensor : model_loader->tensors_map.tensors) {
+        gguf_buffer read_data;
+        read_data.resize(tensor.size);
+        tensor.data = read_data.addr;
+        model_loader->load_data_for(tensor);
+
+        printf("[%4zu/%4zu] %36s - %16s, type = %6s, ",
+               ++idx, model_loader->tensors_map.tensors.size(),
+               tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
+               ggml_type_name(tensor.type));
+
+        // This used to be a regex, but <regex> has an extreme cost to compile times.
+        bool quantize = tensor.name.rfind("weight") == tensor.name.size() - 6; // ends with 'weight'?
+
+        // quantize only 2D tensors
+        quantize &= (tensor.ne.size() == 2);
+        quantize &= params->quantize_output_tensor || tensor.name != "output.weight";
+        quantize &= quantized_type != tensor.type;
+
+        enum ggml_type new_type;
+        void * new_data;
+        size_t new_size;
+        gguf_buffer work;
+
+        if (!quantize) {
+            new_type = tensor.type;
+            new_data = tensor.data;
+            new_size = tensor.size;
+            printf("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
+        } else {
+            new_type = quantized_type;
+#ifdef GGML_USE_K_QUANTS
+            if (tensor.name == "output.weight") {
+                int nx = tensor.ne.at(0);
+                int ny = tensor.ne.at(1);
+                if (nx % QK_K == 0 && ny % QK_K == 0) {
+                    new_type = GGML_TYPE_Q6_K;
+                }
+            } else if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
+                        use_more_bits(i_attention_wv, n_attention_wv)) new_type = GGML_TYPE_Q6_K;
+                else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
+                        (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
+                ++i_attention_wv;
+            } else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
+                         use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
+                //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
+                ++i_feed_forward_w2;
+            } else if (tensor.name.find("attention.wo.weight") != std::string::npos) {
+                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
+                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
+            }
+            bool convert_incompatible_tensor = false;
+            if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
+                new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
+                int nx = tensor.ne.at(0);
+                int ny = tensor.ne.at(1);
+                if (nx % QK_K != 0 || ny % QK_K != 0) {
+                    fprintf(stderr, "\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
+                    convert_incompatible_tensor = true;
+                }
+            }
+            if (convert_incompatible_tensor) {
+                if (tensor.name == "output.weight") {
+                    new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
+                    fprintf(stderr, "F16 will be used for this tensor instead.\n");
+                } else if (tensor.name == "tok_embeddings.weight") {
+                    new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
+                    fprintf(stderr, "Q4_0 will be used for this tensor instead.\n");
+                } else {
+                    throw std::runtime_error("Unsupported tensor size encountered\n");
+                }
+            }
+#endif
+
+            float * f32_data;
+            size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
+            gguf_buffer f32_conv_buf;
+
+            if (tensor.type == GGML_TYPE_F32) {
+                f32_data = (float *) tensor.data;
+            } else if (ggml_is_quantized(tensor.type) && !params->allow_requantize) {
+                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
+            } else {
+                llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
+                f32_data = (float *) f32_conv_buf.addr;
+            }
+
+            printf("quantizing to %s .. ", ggml_type_name(new_type));
+            fflush(stdout);
+
+            work.resize(nelements * 4); // upper bound on size
+            new_data = work.addr;
+            std::vector<int64_t> hist_cur(1 << 4, 0);
+
+            int chunk_size = 32 * 512;
+            const int nchunk = (nelements + chunk_size - 1)/chunk_size;
+            const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
+            if (nthread_use < 2) {
+                new_size = ggml_quantize_chunk(new_type, f32_data, new_data, 0, nelements, hist_cur.data());
+            } else {
+                size_t counter = 0;
+                new_size = 0;
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size] () {
+                    std::vector<int64_t> local_hist;
+                    size_t local_size = 0;
+                    while (true) {
+                        std::unique_lock<std::mutex> lock(mutex);
+                        size_t first = counter; counter += chunk_size;
+                        if (first >= nelements) {
+                            if (!local_hist.empty()) {
+                                for (int j=0; j<int(local_hist.size()); ++j) {
+                                    hist_cur[j] += local_hist[j];
+                                }
+                                new_size += local_size;
+                            }
+                            break;
+                        }
+                        lock.unlock();
+                        size_t last = std::min(nelements, first + chunk_size);
+                        if (local_hist.empty()) {
+                            local_hist.resize(hist_cur.size(), 0);
+                        }
+                        local_size += ggml_quantize_chunk(new_type, f32_data, new_data, first, last - first, local_hist.data());
+                    }
+                };
+                if ((int) workers.size() < nthread_use - 1) {
+                    workers.resize(nthread_use - 1);
+                }
+                for (int it = 0; it < nthread_use - 1; ++it) {
+                    workers[it] = std::thread(compute);
+                }
+                compute();
+                for (int it = 0; it < nthread_use - 1; ++it) {
+                    workers[it].join();
+                }
+            }
+
+            printf("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
+            int64_t tot_count = 0;
+            for (size_t i = 0; i < hist_cur.size(); i++) {
+                hist_all[i] += hist_cur[i];
+                tot_count += hist_cur[i];
+            }
+
+            if (tot_count > 0) {
+                for (size_t i = 0; i < hist_cur.size(); i++) {
+                    printf("%5.3f ", hist_cur[i] / float(nelements));
+                }
+            }
+            printf("\n");
+        }
+        total_size_org += tensor.size;
+        total_size_new += new_size;
+        file_saver.write_tensor(tensor, new_type, new_data, new_size);
+    }
+
+    printf("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
+    printf("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
+
+    {
+        int64_t sum_all = 0;
+        for (size_t i = 0; i < hist_all.size(); i++) {
+            sum_all += hist_all[i];
+        }
+
+        if (sum_all > 0) {
+            printf("%s: hist: ", __func__);
+            for (size_t i = 0; i < hist_all.size(); i++) {
+                printf("%5.3f ", hist_all[i] / float(sum_all));
+            }
+            printf("\n");
+        }
+    }
+}
+
+
+
+//
+// interface implementation
+//
+
+struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+    ggml_time_init();
+
+    llama_model * model = new llama_model;
+
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+
+    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gqa, params.rms_norm_eps, params.n_gpu_layers,
+                params.main_gpu, params.tensor_split, params.rope_freq_base, params.rope_freq_scale,params.low_vram,
+                memory_type, params.use_mmap, params.use_mlock, params.vocab_only, params.progress_callback,
+                params.progress_callback_user_data)) {
+        delete model;
+        fprintf(stderr, "%s: failed to load model\n", __func__);
+        return nullptr;
+    }
+
+    return model;
+}
+
+void llama_free_model(struct llama_model * model) {
+    delete model;
+}
+
+struct llama_context * llama_new_context_with_model(
+                 struct llama_model * model,
+        struct llama_context_params   params) {
+
+    if (!model) {
+        return nullptr;
+    }
+
+    llama_context * ctx = new llama_context(*model);
+
+    if (params.seed == LLAMA_DEFAULT_SEED) {
+        params.seed = time(NULL);
+    }
+
+    unsigned cur_percentage = 0;
+    if (params.progress_callback == NULL) {
+        params.progress_callback_user_data = &cur_percentage;
+        params.progress_callback = [](float progress, void * ctx) {
+            unsigned * cur_percentage_p = (unsigned *) ctx;
+            unsigned percentage = (unsigned) (100 * progress);
+            while (percentage > *cur_percentage_p) {
+                *cur_percentage_p = percentage;
+                fprintf(stderr, ".");
+                fflush(stderr);
+                if (percentage >= 100) {
+                    fprintf(stderr, "\n");
+                }
+            }
+        };
+    }
+
+    ctx->rng = std::mt19937(params.seed);
+    ctx->logits_all = params.logits_all;
+
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+
+    // reserve memory for context buffers
+    if (!params.vocab_only) {
+        if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
+            fprintf(stderr, "%s: kv_cache_init() failed for self-attention cache\n", __func__);
+            llama_free(ctx);
+            return nullptr;
+        }
+
+        {
+            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
+            fprintf(stderr, "%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
+        }
+
+        const auto & hparams = ctx->model.hparams;
+
+        // resized during inference
+        if (params.logits_all) {
+            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
+        } else {
+            ctx->logits.reserve(hparams.n_vocab);
+        }
+
+        if (params.embedding){
+            ctx->embedding.resize(hparams.n_embd);
+        }
+
+        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
+
+        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
+        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
+    }
+
+#ifdef GGML_USE_METAL
+    if (params.n_gpu_layers > 0) {
+        // this allocates all Metal resources and memory buffers
+        ctx->ctx_metal = ggml_metal_init(1);
+
+        void * data_ptr  = NULL;
+        size_t data_size = 0;
+
+        if (params.use_mmap) {
+            data_ptr  = ctx->model.mapping->addr;
+            data_size = ctx->model.mapping->size;
+        } else {
+            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
+            data_size = ggml_get_mem_size  (ctx->model.ctx);
+        }
+
+        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
+
+        fprintf(stderr, "%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
+
+#define LLAMA_METAL_CHECK_BUF(result)                                          \
+    if (!(result)) {                                                           \
+        fprintf(stderr, "%s: failed to add buffer\n", __func__);               \
+        llama_free(ctx);                                                       \
+        return NULL;                                                           \
+    }
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.addr, ctx->buf_compute.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.addr, ctx->kv_self.buf.size, 0));
+
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].addr, ctx->buf_scratch[0].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].addr, ctx->buf_scratch[1].size, 0));
+#undef LLAMA_METAL_CHECK_BUF
+    }
+#endif
+
+#ifdef GGML_USE_MPI
+    ctx->ctx_mpi = ggml_mpi_init();
+
+    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
+        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
+        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
+        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
+        llama_backend_free();
+        exit(1);
+    }
+#endif
+
+    return ctx;
+}
+
+struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+
+    struct llama_model * model = llama_load_model_from_file(path_model, params);
+    if (!model) {
+        return nullptr;
+    }
+    struct llama_context * ctx = llama_new_context_with_model(model, params);
+    ctx->model_owner = true;
+    return ctx;
+}
+
+void llama_free(struct llama_context * ctx) {
+    if (ctx->model_owner) {
+        delete &ctx->model;
+    }
+    delete ctx;
+}
+
+int llama_model_quantize(
+        const char * fname_inp,
+        const char * fname_out,
+        const llama_model_quantize_params *params) {
+    try {
+        llama_model_quantize_internal(fname_inp, fname_out, params);
+        return 0;
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to quantize: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
+    fprintf(stderr, "%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
+
+    const int64_t t_start_lora_us = ggml_time_us();
+
+    auto fin = std::ifstream(path_lora, std::ios::binary);
+    if (!fin) {
+        fprintf(stderr, "%s: failed to open '%s'\n", __func__, path_lora);
+        return 1;
+    }
+
+    // verify magic and version
+    {
+        uint32_t magic;
+        fin.read((char *) &magic, sizeof(magic));
+        uint32_t format_version;
+        fin.read((char *) &format_version, sizeof(format_version));
+
+        if (format_version != 1) {
+            fprintf(stderr, "%s: unsupported file version\n", __func__ );
+            return 1;
+        }
+    }
+
+    int32_t lora_r;
+    int32_t lora_alpha;
+    fin.read((char *) &lora_r, sizeof(lora_r));
+    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
+    float scaling = (float)lora_alpha / (float)lora_r;
+
+    fprintf(stderr, "%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
+
+
+    // create a temporary ggml context to store the lora tensors
+    // todo: calculate size from biggest possible tensor
+    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
+    struct ggml_init_params params;
+    params.mem_size   = lora_buf.size();
+    params.mem_buffer = lora_buf.data();
+    params.no_alloc   = false;
+
+    ggml_context * lora_ctx = ggml_init(params);
+    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+
+    // create a name -> tensor map of the model to accelerate lookups
+    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
+    for (const auto & kv: model.tensors_by_name) {
+        model_tensors.insert(kv);
+    }
+
+
+    // load base model
+    std::unique_ptr<llama_model_loader> model_loader;
+    ggml_context * base_ctx = NULL;
+    gguf_buffer base_buf;
+    if (path_base_model) {
+        fprintf(stderr, "%s: loading base model from '%s'\n", __func__, path_base_model);
+        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
+
+        size_t ctx_size;
+        size_t mmapped_size;
+        model_loader->calc_sizes(&ctx_size, &mmapped_size);
+        base_buf.resize(ctx_size);
+
+        ggml_init_params base_params;
+        base_params.mem_size   = base_buf.size;
+        base_params.mem_buffer = base_buf.addr;
+        base_params.no_alloc   = model_loader->use_mmap;
+
+        base_ctx = ggml_init(base_params);
+
+        model_loader->ggml_ctx = base_ctx;
+
+        // maybe this should in llama_model_loader
+        if (model_loader->use_mmap) {
+            model_loader->mapping.reset(new gguf_mmap(&model_loader->file_loader->file, /* prefetch */ 0, ggml_is_numa()));
+        }
+    }
+
+    // read tensors and apply
+    bool warned = false;
+    int n_tensors = 0;
+
+    std::vector<uint8_t> work_buffer;
+
+    while (true) {
+        int32_t n_dims;
+        int32_t length;
+        int32_t ftype;
+
+        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
+        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
+        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
+        if (fin.eof()) {
+            break;
+        }
+
+        int32_t ne[2] = { 1, 1 };
+        for (int i = 0; i < n_dims; ++i) {
+            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
+        }
+
+        std::string name;
+        {
+            char buf[1024];
+            fin.read(buf, length);
+            name = std::string(buf, length);
+        }
+
+        // check for lora suffix and get the type of tensor
+        const std::string lora_suffix = ".lora";
+        size_t pos = name.rfind(lora_suffix);
+        if (pos == std::string::npos) {
+            fprintf(stderr, "%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
+            return 1;
+        }
+
+        std::string lora_type = name.substr(pos + lora_suffix.length());
+        std::string base_name = name;
+        base_name.erase(pos);
+        // fprintf(stderr, "%s: %s => %s (lora type %s) ", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+
+        if (model_tensors.find(base_name) == model_tensors.end()) {
+            fprintf(stderr, "%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
+            return 1;
+        }
+
+        // create ggml tensor
+        ggml_type wtype;
+        switch (ftype) {
+            case 0: wtype = GGML_TYPE_F32;  break;
+            case 1: wtype = GGML_TYPE_F16;  break;
+            default:
+                    {
+                        fprintf(stderr, "%s: invalid tensor data type '%d'\n",
+                                __func__, ftype);
+                        return false;
+                    }
+        }
+        ggml_tensor * lora_tensor;
+        if (n_dims == 2) {
+            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
+        }
+        else {
+            fprintf(stderr, "%s: unsupported tensor dimension %d\n", __func__, n_dims);
+            return 1;
+        }
+        ggml_set_name(lora_tensor, "lora_tensor");
+
+        // load tensor data
+        size_t offset = fin.tellg();
+        size_t tensor_data_size = ggml_nbytes(lora_tensor);
+        offset = (offset + 31) & -32;
+        fin.seekg(offset);
+        fin.read((char*)lora_tensor->data, tensor_data_size);
+
+        lora_tensors[name] = lora_tensor;
+
+        // check if we have both A and B tensors and apply
+        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
+            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
+
+            ggml_tensor * dest_t = model_tensors[base_name];
+
+            offload_func_t offload_func = llama_nop;
+            offload_func_t offload_func_force_inplace = llama_nop;
+
+#ifdef GGML_USE_CUBLAS
+            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
+                if (dest_t->type != GGML_TYPE_F16) {
+                    throw std::runtime_error(format(
+                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
+                }
+                offload_func = ggml_cuda_assign_buffers;
+                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
+            }
+#endif // GGML_USE_CUBLAS
+
+            ggml_tensor * base_t;
+            if (model_loader) {
+                // load from base model
+                if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
+                    fprintf(stderr, "%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
+                    return 1;
+                }
+                size_t idx = model_loader->tensors_map.name_to_idx[base_name];
+                gguf_load_tensor & lt = model_loader->tensors_map.tensors[idx];
+                base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
+                lt.data = (uint8_t *) lt.ggml_tensor->data;
+                model_loader->load_data_for(lt);
+                lt.ggml_tensor->data = lt.data;
+            }
+            else {
+                base_t = dest_t;
+            }
+
+            if (ggml_is_quantized(base_t->type)) {
+                if (!warned) {
+                    fprintf(stderr, "%s: warning: using a lora adapter with a quantized model may result in poor quality, "
+                                    "use a f16 or f32 base model with --lora-base\n", __func__);
+                    warned = true;
+                }
+            }
+
+            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
+            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
+            ggml_set_name(loraA, "loraA");
+
+            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
+            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
+            ggml_set_name(loraB, "loraB");
+
+            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
+                fprintf(stderr, "%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
+                               " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
+                return 1;
+            }
+
+            // w = w + BA*s
+            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
+            offload_func(BA);
+            ggml_set_name(BA, "BA");
+
+            if (scaling != 1.0f) {
+                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
+                ggml_set_name(scale_tensor, "scale_tensor");
+
+                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
+                offload_func(BA);
+                ggml_set_name(BA, "BA_scaled");
+            }
+
+            ggml_tensor * r;
+            if (base_t == dest_t) {
+                r = ggml_add_inplace(lora_ctx, dest_t, BA);
+                offload_func_force_inplace(r);
+                ggml_set_name(r, "r_add_inplace");
+            }
+            else {
+                r = ggml_add(lora_ctx, base_t, BA);
+                offload_func(r);
+                ggml_set_name(r, "r_add");
+
+                r = ggml_cpy(lora_ctx, r, dest_t);
+                offload_func(r);
+                ggml_set_name(r, "r_cpy");
+            }
+
+            struct ggml_cgraph gf = ggml_build_forward(r);
+
+            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
+
+            // we won't need these tensors again, reset the context to save memory
+            ggml_free(lora_ctx);
+            lora_ctx = ggml_init(params);
+            lora_tensors.clear();
+
+            n_tensors++;
+            if (n_tensors % 4 == 0) {
+                fprintf(stderr, ".");
+            }
+        }
+    }
+
+    // TODO: this should be in a destructor, it will leak on failure
+    ggml_free(lora_ctx);
+    if (base_ctx) {
+        ggml_free(base_ctx);
+    }
+
+    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
+    fprintf(stderr, " done (%.2f ms)\n", t_lora_us / 1000.0);
+
+    return 0;
+}
+
+int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
+    try {
+        return llama_apply_lora_from_file_internal(ctx->model, path_lora, path_base_model, n_threads);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_model_apply_lora_from_file(const struct llama_model * model, const char * path_lora, const char * path_base_model, int n_threads) {
+    try {
+        return llama_apply_lora_from_file_internal(*model, path_lora, path_base_model, n_threads);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        return 1;
+    }
+}
+
+int llama_get_kv_cache_token_count(const struct llama_context * ctx) {
+    return ctx->kv_self.n;
+}
+
+#define LLAMA_MAX_RNG_STATE (64*1024)
+
+void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed) {
+    if (seed == LLAMA_DEFAULT_SEED) {
+        seed = time(NULL);
+    }
+    ctx->rng.seed(seed);
+}
+
+// Returns the *maximum* size of the state
+size_t llama_get_state_size(const struct llama_context * ctx) {
+    // we don't know size of rng until we actually serialize it. so reserve more than enough memory for its serialized state.
+    // for reference, std::mt19937(1337) serializes to 6701 bytes.
+    const size_t s_rng_size        = sizeof(size_t);
+    const size_t s_rng             = LLAMA_MAX_RNG_STATE;
+    const size_t s_logits_capacity = sizeof(size_t);
+    const size_t s_logits_size     = sizeof(size_t);
+    const size_t s_logits          = ctx->logits.capacity() * sizeof(float);
+    const size_t s_embedding_size  = sizeof(size_t);
+    const size_t s_embedding       = ctx->embedding.size() * sizeof(float);
+    const size_t s_kv_size         = sizeof(size_t);
+    const size_t s_kv_ntok         = sizeof(int);
+    const size_t s_kv              = ctx->kv_self.buf.size;
+
+    const size_t s_total = (
+        + s_rng_size
+        + s_rng
+        + s_logits_capacity
+        + s_logits_size
+        + s_logits
+        + s_embedding_size
+        + s_embedding
+        + s_kv_size
+        + s_kv_ntok
+        + s_kv
+    );
+
+    return s_total;
+}
+
+// Copies the state to the specified destination address
+size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
+    uint8_t * out = dst;
+
+    // copy rng
+    {
+        std::stringstream rng_ss;
+        rng_ss << ctx->rng;
+
+        const size_t rng_size = rng_ss.str().size();
+        char rng_buf[LLAMA_MAX_RNG_STATE];
+
+        memset(&rng_buf[0], 0, LLAMA_MAX_RNG_STATE);
+        memcpy(&rng_buf[0], rng_ss.str().data(), rng_ss.str().size());
+
+        memcpy(out, &rng_size,   sizeof(rng_size));    out += sizeof(rng_size);
+        memcpy(out, &rng_buf[0], LLAMA_MAX_RNG_STATE); out += LLAMA_MAX_RNG_STATE;
+    }
+
+    // copy logits
+    {
+        const size_t logits_cap  = ctx->logits.capacity();
+        const size_t logits_size = ctx->logits.size();
+
+        memcpy(out, &logits_cap,  sizeof(logits_cap));  out += sizeof(logits_cap);
+        memcpy(out, &logits_size, sizeof(logits_size)); out += sizeof(logits_size);
+
+        if (logits_size) {
+            memcpy(out, ctx->logits.data(), logits_size * sizeof(float));
+        }
+
+        out += logits_cap * sizeof(float);
+    }
+
+    // copy embeddings
+    {
+        const size_t embedding_size = ctx->embedding.size();
+
+        memcpy(out, &embedding_size, sizeof(embedding_size)); out += sizeof(embedding_size);
+
+        if (embedding_size) {
+            memcpy(out, ctx->embedding.data(), embedding_size * sizeof(float));
+            out += embedding_size * sizeof(float);
+        }
+    }
+
+    // copy kv cache
+    {
+        const auto & kv_self = ctx->kv_self;
+        const auto & hparams = ctx->model.hparams;
+        const int    n_layer = hparams.n_layer;
+        const int    n_embd  = hparams.n_embd;
+        const int    n_ctx   = hparams.n_ctx;
+
+        const size_t kv_size = kv_self.buf.size;
+        const int    kv_ntok = llama_get_kv_cache_token_count(ctx);
+
+        memcpy(out, &kv_size, sizeof(kv_size)); out += sizeof(kv_size);
+        memcpy(out, &kv_ntok, sizeof(kv_ntok)); out += sizeof(kv_ntok);
+
+        if (kv_size) {
+            const size_t elt_size = ggml_element_size(kv_self.k);
+
+            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
+            ggml_cgraph gf{};
+
+            ggml_tensor * kout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
+            kout3d->data = out;
+            out += ggml_nbytes(kout3d);
+
+            ggml_tensor * vout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
+            vout3d->data = out;
+            out += ggml_nbytes(vout3d);
+
+            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
+                n_embd, kv_ntok, n_layer,
+                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
+
+            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
+                kv_ntok, n_embd, n_layer,
+                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
+
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, k3d, kout3d));
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, v3d, vout3d));
+            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
+
+            ggml_free(cpy_ctx);
+        }
+    }
+
+    const size_t written  = out - dst;
+    const size_t max_size = llama_get_state_size(ctx);
+
+    GGML_ASSERT(written <= max_size);
+
+    return written;
+}
+
+// Sets the state reading from the specified source address
+size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
+    uint8_t * inp = src;
+
+    // set rng
+    {
+        size_t rng_size;
+        char   rng_buf[LLAMA_MAX_RNG_STATE];
+
+        memcpy(&rng_size,   inp, sizeof(rng_size));    inp += sizeof(rng_size);
+        memcpy(&rng_buf[0], inp, LLAMA_MAX_RNG_STATE); inp += LLAMA_MAX_RNG_STATE;
+
+        std::stringstream rng_ss;
+        rng_ss.str(std::string(&rng_buf[0], rng_size));
+        rng_ss >> ctx->rng;
+
+        GGML_ASSERT(rng_ss.fail() == false);
+    }
+
+    // set logits
+    {
+        size_t logits_cap;
+        size_t logits_size;
+
+        memcpy(&logits_cap,  inp, sizeof(logits_cap));  inp += sizeof(logits_cap);
+        memcpy(&logits_size, inp, sizeof(logits_size)); inp += sizeof(logits_size);
+
+        GGML_ASSERT(ctx->logits.capacity() == logits_cap);
+
+        if (logits_size) {
+            ctx->logits.resize(logits_size);
+            memcpy(ctx->logits.data(), inp, logits_size * sizeof(float));
+        }
+
+        inp += logits_cap * sizeof(float);
+    }
+
+    // set embeddings
+    {
+        size_t embedding_size;
+
+        memcpy(&embedding_size, inp, sizeof(embedding_size)); inp += sizeof(embedding_size);
+
+        GGML_ASSERT(ctx->embedding.capacity() == embedding_size);
+
+        if (embedding_size) {
+            memcpy(ctx->embedding.data(), inp, embedding_size * sizeof(float));
+            inp += embedding_size * sizeof(float);
+        }
+    }
+
+    // set kv cache
+    {
+        const auto & kv_self = ctx->kv_self;
+        const auto & hparams = ctx->model.hparams;
+        const int    n_layer = hparams.n_layer;
+        const int    n_embd  = hparams.n_embd;
+        const int    n_ctx   = hparams.n_ctx;
+
+        size_t kv_size;
+        int kv_ntok;
+
+        memcpy(&kv_size, inp, sizeof(kv_size)); inp += sizeof(kv_size);
+        memcpy(&kv_ntok, inp, sizeof(kv_ntok)); inp += sizeof(kv_ntok);
+
+        if (kv_size) {
+            GGML_ASSERT(kv_self.buf.size == kv_size);
+
+            const size_t elt_size = ggml_element_size(kv_self.k);
+
+            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
+            ggml_cgraph gf{};
+
+            ggml_tensor * kin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
+            kin3d->data = (void *) inp;
+            inp += ggml_nbytes(kin3d);
+
+            ggml_tensor * vin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
+            vin3d->data = (void *) inp;
+            inp += ggml_nbytes(vin3d);
+
+            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
+                n_embd, kv_ntok, n_layer,
+                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
+
+            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
+                kv_ntok, n_embd, n_layer,
+                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
+
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, kin3d, k3d));
+            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, vin3d, v3d));
+            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
+
+            ggml_free(cpy_ctx);
+        }
+
+        ctx->kv_self.n = kv_ntok;
+    }
+
+    const size_t nread    = inp - src;
+    const size_t max_size = llama_get_state_size(ctx);
+
+    GGML_ASSERT(nread <= max_size);
+
+    return nread;
+}
+
+static bool llama_load_session_file_internal(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
+    gguf_file file(path_session, "rb");
+    GGML_UNUSED(ctx);
+    GGML_UNUSED(path_session);
+    GGML_UNUSED(tokens_out);
+    GGML_UNUSED(n_token_capacity);
+    GGML_UNUSED(n_token_count_out);
+
+
+// TODO: implement with GGUF format
+    return true;
+}
+
+bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
+    try {
+        return llama_load_session_file_internal(ctx, path_session, tokens_out, n_token_capacity, n_token_count_out);
+    } catch (const std::exception & err) {
+        fprintf(stderr, "error loading session file: %s\n", err.what());
+        return false;
+    }
+}
+
+bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
+    gguf_file file(path_session, "wb");
+    GGML_UNUSED(ctx);
+    GGML_UNUSED(tokens);
+    GGML_UNUSED(n_token_count);
+
+    // TODO: implement with GGUF format
+
+    return true;
+}
+
+int llama_eval(
+        struct llama_context * ctx,
+           const llama_token * tokens,
+                         int   n_tokens,
+                         int   n_past,
+                         int   n_threads) {
+    if (!llama_eval_internal(*ctx, tokens, nullptr, n_tokens, n_past, n_threads, nullptr)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    // get a more accurate load time, upon first eval
+    // TODO: fix this
+    if (!ctx->has_evaluated_once) {
+        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
+        ctx->has_evaluated_once = true;
+    }
+
+    return 0;
+}
+
+
+int llama_eval_embd(
+            struct llama_context * ctx,
+                     const float * embd,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads) {
+    if (!llama_eval_internal(*ctx, nullptr, embd, n_tokens, n_past, n_threads, nullptr)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    // get a more accurate load time, upon first eval
+    // TODO: fix this
+    if (!ctx->has_evaluated_once) {
+        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
+        ctx->has_evaluated_once = true;
+    }
+
+    return 0;
+}
+
+int llama_eval_export(struct llama_context * ctx, const char * fname) {
+    const int n_batch = 1;
+    const int n_ctx   = 512 - n_batch;
+
+    const std::vector<llama_token> tmp(n_batch, llama_token_bos());
+
+    if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
+        fprintf(stderr, "%s: failed to eval\n", __func__);
+        return 1;
+    }
+
+    return 0;
+}
+
+int llama_tokenize_with_model(
+    const struct llama_model * model,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    auto res = llama_tokenize(model->vocab, text, add_bos);
+
+    if (n_max_tokens < (int) res.size()) {
+        fprintf(stderr, "%s: too many tokens\n", __func__);
+        return -((int) res.size());
+    }
+
+    for (size_t i = 0; i < res.size(); i++) {
+        tokens[i] = res[i];
+    }
+
+    return res.size();
+}
+
+int llama_tokenize(
+        struct llama_context * ctx,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
+}
+
+int llama_n_vocab_from_model(const struct llama_model * model) {
+    return model->vocab.id_to_token.size();
+}
+
+int llama_n_ctx_from_model(const struct llama_model * model) {
+    return model->hparams.n_ctx;
+}
+
+int llama_n_embd_from_model(const struct llama_model * model) {
+    return model->hparams.n_embd;
+}
+
+int llama_n_vocab(const struct llama_context * ctx) {
+    return ctx->model.vocab.id_to_token.size();
+}
+
+int llama_n_ctx(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_ctx;
+}
+
+int llama_n_embd(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_embd;
+}
+
+int llama_get_vocab_from_model(
+        const struct llama_model * model,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
+    for (int i = 0; i<n; ++i) {
+        strings[i] = model->vocab.id_to_token[i].tok.c_str();
+        scores[i]  = model->vocab.id_to_token[i].score;
+    }
+    return n;
+}
+
+int llama_get_vocab(
+        const struct llama_context * ctx,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
+}
+
+float * llama_get_logits(struct llama_context * ctx) {
+    return ctx->logits.data();
+}
+
+float * llama_get_embeddings(struct llama_context * ctx) {
+    return ctx->embedding.data();
+}
+
+const char * llama_token_to_str_with_model(const struct llama_model * model, llama_token token) {
+    if (token >= llama_n_vocab_from_model(model)) {
+        return nullptr;
+    }
+
+    return model->vocab.id_to_token[token].tok.c_str();
+}
+
+const char * llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    return llama_token_to_str_with_model(&ctx->model, token);
+}
+
+llama_token llama_token_bos() {
+    return 1;
+}
+
+llama_token llama_token_eos() {
+    return 2;
+}
+
+llama_token llama_token_nl() {
+    return 13;
+}
+
+struct llama_timings llama_get_timings(struct llama_context * ctx) {
+    struct llama_timings result = {
+        /*.t_start_ms  =*/ 1e-3 * ctx->t_start_us,
+        /*.t_end_ms    =*/ 1.00 * ggml_time_ms(),
+        /*.t_load_ms   =*/ 1e-3 * ctx->t_load_us,
+        /*.t_sample_ms =*/ 1e-3 * ctx->t_sample_us,
+        /*.t_p_eval_ms =*/ 1e-3 * ctx->t_p_eval_us,
+        /*.t_eval_ms   =*/ 1e-3 * ctx->t_eval_us,
+
+        /*.n_sample =*/ std::max(1, ctx->n_sample),
+        /*.n_p_eval =*/ std::max(1, ctx->n_p_eval),
+        /*.n_eval   =*/ std::max(1, ctx->n_eval),
+    };
+
+    return result;
+}
+
+void llama_print_timings(struct llama_context * ctx) {
+    const llama_timings timings = llama_get_timings(ctx);
+
+    fprintf(stderr, "\n");
+    fprintf(stderr, "%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
+    fprintf(stderr, "%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_sample_ms, timings.n_sample, timings.t_sample_ms / timings.n_sample, 1e3 / timings.t_sample_ms * timings.n_sample);
+    fprintf(stderr, "%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_p_eval_ms, timings.n_p_eval, timings.t_p_eval_ms / timings.n_p_eval, 1e3 / timings.t_p_eval_ms * timings.n_p_eval);
+    fprintf(stderr, "%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+            __func__, timings.t_eval_ms, timings.n_eval, timings.t_eval_ms / timings.n_eval, 1e3 / timings.t_eval_ms * timings.n_eval);
+    fprintf(stderr, "%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
+}
+
+void llama_reset_timings(struct llama_context * ctx) {
+    ctx->t_start_us = ggml_time_us();
+    ctx->t_sample_us = ctx->n_sample = 0;
+    ctx->t_eval_us   = ctx->n_eval   = 0;
+    ctx->t_p_eval_us = ctx->n_p_eval = 0;
+}
+
+const char * llama_print_system_info(void) {
+    static std::string s;
+
+    s  = "";
+    s += "AVX = "         + std::to_string(ggml_cpu_has_avx())         + " | ";
+    s += "AVX2 = "        + std::to_string(ggml_cpu_has_avx2())        + " | ";
+    s += "AVX512 = "      + std::to_string(ggml_cpu_has_avx512())      + " | ";
+    s += "AVX512_VBMI = " + std::to_string(ggml_cpu_has_avx512_vbmi()) + " | ";
+    s += "AVX512_VNNI = " + std::to_string(ggml_cpu_has_avx512_vnni()) + " | ";
+    s += "FMA = "         + std::to_string(ggml_cpu_has_fma())         + " | ";
+    s += "NEON = "        + std::to_string(ggml_cpu_has_neon())        + " | ";
+    s += "ARM_FMA = "     + std::to_string(ggml_cpu_has_arm_fma())     + " | ";
+    s += "F16C = "        + std::to_string(ggml_cpu_has_f16c())        + " | ";
+    s += "FP16_VA = "     + std::to_string(ggml_cpu_has_fp16_va())     + " | ";
+    s += "WASM_SIMD = "   + std::to_string(ggml_cpu_has_wasm_simd())   + " | ";
+    s += "BLAS = "        + std::to_string(ggml_cpu_has_blas())        + " | ";
+    s += "SSE3 = "        + std::to_string(ggml_cpu_has_sse3())        + " | ";
+    s += "VSX = "         + std::to_string(ggml_cpu_has_vsx())         + " | ";
+
+    return s.c_str();
+}
+
+// For internal test use
+const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
+    return ctx->model.tensors_by_name;
+}
diff --git a/gguf-llama.h b/gguf-llama.h
index d3c0d6b87f437..540167bd196ba 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -1,449 +1,449 @@
-#ifndef LLAMA_H
-#define LLAMA_H
-
-#include "ggml.h"
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-#define LLAMA_MAX_DEVICES GGML_CUDA_MAX_DEVICES
-#else
-#define LLAMA_MAX_DEVICES 1
-#endif // GGML_USE_CUBLAS
-#include <stddef.h>
-#include <stdint.h>
-#include <stdbool.h>
-
-#ifdef LLAMA_SHARED
-#    if defined(_WIN32) && !defined(__MINGW32__)
-#        ifdef LLAMA_BUILD
-#            define LLAMA_API __declspec(dllexport)
-#        else
-#            define LLAMA_API __declspec(dllimport)
-#        endif
-#    else
-#        define LLAMA_API __attribute__ ((visibility ("default")))
-#    endif
-#else
-#    define LLAMA_API
-#endif
-
-#ifdef __GNUC__
-#    define DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
-#elif defined(_MSC_VER)
-#    define DEPRECATED(func, hint) __declspec(deprecated(hint)) func
-#else
-#    define DEPRECATED(func, hint) func
-#endif
-
-#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
-
-#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
-// Defined when llama.cpp is compiled with support for offloading model layers to GPU.
-#define LLAMA_SUPPORTS_GPU_OFFLOAD
-#endif
-
-#ifndef LLAMA_DEFAULT_RMS_EPS
-#define LLAMA_DEFAULT_RMS_EPS 5e-6f
-#endif
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-    //
-    // C interface
-    //
-    // TODO: show sample usage
-    //
-
-    struct llama_model;
-    struct llama_context;
-
-    typedef int llama_token;
-
-    typedef struct llama_token_data {
-        llama_token id; // token id
-        float logit;    // log-odds of the token
-        float p;        // probability of the token
-    } llama_token_data;
-
-    typedef struct llama_token_data_array {
-        llama_token_data * data;
-        size_t size;
-        bool sorted;
-    } llama_token_data_array;
-
-    typedef void (*llama_progress_callback)(float progress, void *ctx);
-
-   struct llama_context_params {
-        uint32_t seed;         // RNG seed, -1 for random
-        int32_t  n_ctx;        // text context
-        int32_t  n_batch;      // prompt processing batch size
-        int32_t  n_gqa;        // grouped-query attention (TEMP - will be moved to model hparams)
-        float    rms_norm_eps; // rms norm epsilon (TEMP - will be moved to model hparams)
-        int32_t  n_gpu_layers; // number of layers to store in VRAM
-        int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
-
-        const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)
-
-        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
-        float    rope_freq_base;  // RoPE base frequency
-        float    rope_freq_scale; // RoPE frequency scaling factor
-
-        // called with a progress value between 0 and 1, pass NULL to disable
-        llama_progress_callback progress_callback;
-        // context pointer passed to the progress callback
-        void * progress_callback_user_data;
-
-        // Keep the booleans together to avoid misalignment during copy-by-value.
-        bool low_vram;   // if true, reduce VRAM usage at the cost of performance
-        bool f16_kv;     // use fp16 for KV cache
-        bool logits_all; // the llama_eval() call computes all logits, not just the last one
-        bool vocab_only; // only load the vocabulary, no weights
-        bool use_mmap;   // use mmap if possible
-        bool use_mlock;  // force system to keep model in RAM
-        bool embedding;  // embedding mode only
-    };
-    // model file types
-    enum llama_ftype {
-        LLAMA_FTYPE_ALL_F32              = 0,
-        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
-        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
-        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
-        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
-    };
-
-    // model quantization parameters
-    typedef struct llama_model_quantize_params {
-        int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
-        enum llama_ftype   ftype;    // quantize to this llama_ftype
-        bool allow_requantize;       // allow quantizing non-f32/f16 tensors
-        bool quantize_output_tensor; // quantize output.weight
-    } llama_model_quantize_params;
-
-    // grammar types
-    struct llama_grammar;
-
-    // grammar element type
-    enum llama_gretype {
-        // end of rule definition
-        LLAMA_GRETYPE_END            = 0,
-
-        // start of alternate definition for rule
-        LLAMA_GRETYPE_ALT            = 1,
-
-        // non-terminal element: reference to rule
-        LLAMA_GRETYPE_RULE_REF       = 2,
-
-        // terminal element: character (code point)
-        LLAMA_GRETYPE_CHAR           = 3,
-
-        // inverse char(s) ([^a], [^a-b] [^abc])
-        LLAMA_GRETYPE_CHAR_NOT       = 4,
-
-        // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
-        // be an inclusive range ([a-z])
-        LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,
-
-        // modifies a preceding LLAMA_GRETYPE_CHAR or
-        // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
-        LLAMA_GRETYPE_CHAR_ALT       = 6,
-    };
-
-    typedef struct llama_grammar_element {
-        enum llama_gretype type;
-        uint32_t           value; // Unicode code point or rule ID
-    } llama_grammar_element;
-
-    // performance timing information
-    struct llama_timings {
-        double t_start_ms;
-        double t_end_ms;
-        double t_load_ms;
-        double t_sample_ms;
-        double t_p_eval_ms;
-        double t_eval_ms;
-
-        int32_t n_sample;
-        int32_t n_p_eval;
-        int32_t n_eval;
-    };
-
-    LLAMA_API int llama_max_devices();
-
-    LLAMA_API struct llama_context_params llama_context_default_params();
-    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
-
-    LLAMA_API bool llama_mmap_supported();
-    LLAMA_API bool llama_mlock_supported();
-
-    // TODO: not great API - very likely to change
-    // Initialize the llama + ggml backend
-    // If numa is true, use NUMA optimizations
-    // Call once at the start of the program
-    LLAMA_API void llama_backend_init(bool numa);
-    // Call once at the end of the program - currently only used for MPI
-    LLAMA_API void llama_backend_free();
-
-    LLAMA_API int64_t llama_time_us();
-
-    LLAMA_API struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params);
-
-    LLAMA_API void llama_free_model(struct llama_model * model);
-
-    LLAMA_API struct llama_context * llama_new_context_with_model(
-                     struct llama_model * model,
-            struct llama_context_params   params);
-
-
-    // Frees all allocated memory
-    LLAMA_API void llama_free(struct llama_context * ctx);
-
-    // Returns 0 on success
-    LLAMA_API int llama_model_quantize(
-            const char * fname_inp,
-            const char * fname_out,
-            const llama_model_quantize_params * params);
-
-    // Apply a LoRA adapter to a loaded model
-    // path_base_model is the path to a higher quality model to use as a base for
-    // the layers modified by the adapter. Can be NULL to use the current loaded model.
-    // The model needs to be reloaded before applying a new adapter, otherwise the adapter
-    // will be applied on top of the previous one
-    // Returns 0 on success
-    LLAMA_API DEPRECATED(int llama_apply_lora_from_file(
-            struct llama_context * ctx,
-                      const char * path_lora,
-                      const char * path_base_model,
-                             int   n_threads),
-            "please use llama_model_apply_lora_from_file instead");
-
-    LLAMA_API int llama_model_apply_lora_from_file(
-            const struct llama_model * model,
-                      const char * path_lora,
-                      const char * path_base_model,
-                             int   n_threads);
-
-    // Returns the number of tokens in the KV cache
-    LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
-
-    // Sets the current rng seed.
-    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
-
-    // Returns the maximum size in bytes of the state (rng, logits, embedding
-    // and kv_cache) - will often be smaller after compacting tokens
-    LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
-
-    // Copies the state to the specified destination address.
-    // Destination needs to have allocated enough memory.
-    // Returns the number of bytes copied
-    LLAMA_API size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst);
-
-    // Set the state reading from the specified address
-    // Returns the number of bytes read
-    LLAMA_API size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src);
-
-    // Save/load session file
-    LLAMA_API bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out);
-    LLAMA_API bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count);
-
-    // Run the llama inference to obtain the logits and probabilities for the next token.
-    // tokens + n_tokens is the provided batch of new tokens to process
-    // n_past is the number of tokens to use from previous eval calls
-    // Returns 0 on success
-    LLAMA_API int llama_eval(
-            struct llama_context * ctx,
-               const llama_token * tokens,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads);
-
-    // Same as llama_eval, but use float matrix input directly.
-    LLAMA_API int llama_eval_embd(
-            struct llama_context * ctx,
-                     const float * embd,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads);
-
-    // Export a static computation graph for context of 511 and batch size of 1
-    // NOTE: since this functionality is mostly for debugging and demonstration purposes, we hardcode these
-    //       parameters here to keep things simple
-    // IMPORTANT: do not use for anything else other than debugging and testing!
-    LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
-
-    // Convert the provided text into tokens.
-    // The tokens pointer must be large enough to hold the resulting tokens.
-    // Returns the number of tokens on success, no more than n_max_tokens
-    // Returns a negative number on failure - the number of tokens that would have been returned
-    // TODO: not sure if correct
-    LLAMA_API int llama_tokenize(
-            struct llama_context * ctx,
-                      const char * text,
-                     llama_token * tokens,
-                             int   n_max_tokens,
-                            bool   add_bos);
-
-    LLAMA_API int llama_tokenize_with_model(
-        const struct llama_model * model,
-                      const char * text,
-                     llama_token * tokens,
-                             int   n_max_tokens,
-                            bool   add_bos);
-
-    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
-    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
-    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
-
-    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
-    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
-    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
-
-    // Get the vocabulary as output parameters.
-    // Returns number of results.
-    LLAMA_API int llama_get_vocab(
-            const struct llama_context * ctx,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    LLAMA_API int llama_get_vocab_from_model(
-              const struct llama_model * model,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    // Token logits obtained from the last call to llama_eval()
-    // The logits for the last token are stored in the last row
-    // Can be mutated in order to change the probabilities of the next token
-    // Rows: n_tokens
-    // Cols: n_vocab
-    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
-
-    // Get the embeddings for the input
-    // shape: [n_embd] (1-dimensional)
-    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
-
-    // Token Id -> String. Uses the vocabulary in the provided context
-    LLAMA_API const char * llama_token_to_str(
-            const struct llama_context * ctx,
-                           llama_token   token);
-
-    LLAMA_API const char * llama_token_to_str_with_model(
-              const struct llama_model * model,
-                           llama_token   token);
-
-    // Special tokens
-    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl();   // next-line
-
-    // Grammar
-    //
-    LLAMA_API struct llama_grammar * llama_grammar_init(
-            const llama_grammar_element ** rules,
-                                 size_t    n_rules,
-                                 size_t    start_rule_index);
-
-    LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
-
-    // Sampling functions
-
-    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
-    LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
-
-    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
-    LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
-
-    /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
-    /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
-    /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
-    LLAMA_API void llama_sample_classifier_free_guidance(
-              struct llama_context * ctx,
-            llama_token_data_array * candidates,
-              struct llama_context * guidance_ctx,
-                             float   scale);
-
-    /// @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
-    LLAMA_API void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
-    LLAMA_API void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep);
-
-    /// @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
-    LLAMA_API void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
-
-    /// @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
-    LLAMA_API void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep);
-
-    /// @details Locally Typical Sampling implementation described in the paper https://arxiv.org/abs/2202.00666.
-    LLAMA_API void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
-    LLAMA_API void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates, float temp);
-
-    /// @details Apply constraints from grammar
-    LLAMA_API void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar);
-
-    /// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
-    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
-    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
-    /// @param m The number of tokens considered in the estimation of `s_hat`. This is an arbitrary value that is used to calculate `s_hat`, which in turn helps to calculate the value of `k`. In the paper, they use `m = 100`, but you can experiment with different values to see how it affects the performance of the algorithm.
-    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
-    LLAMA_API llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu);
-
-    /// @details Mirostat 2.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
-    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
-    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
-    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
-    LLAMA_API llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu);
-
-    /// @details Selects the token with the highest probability.
-    LLAMA_API llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Randomly selects a token from the candidates based on their probabilities.
-    LLAMA_API llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Accepts the sampled token into the grammar
-    LLAMA_API void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token);
-
-    // Performance information
-    LLAMA_API struct llama_timings llama_get_timings(struct llama_context * ctx);
-    LLAMA_API void llama_print_timings(struct llama_context * ctx);
-    LLAMA_API void llama_reset_timings(struct llama_context * ctx);
-
-    // Print system information
-    LLAMA_API const char * llama_print_system_info(void);
-
-#ifdef __cplusplus
-}
-#endif
-
-// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
-#ifdef LLAMA_API_INTERNAL
-
-#include <vector>
-#include <string>
-struct ggml_tensor;
-
-const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
-
-#endif
-
-#endif // LLAMA_H
+#ifndef LLAMA_H
+#define LLAMA_H
+
+#include "ggml.h"
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+#define LLAMA_MAX_DEVICES GGML_CUDA_MAX_DEVICES
+#else
+#define LLAMA_MAX_DEVICES 1
+#endif // GGML_USE_CUBLAS
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+#ifdef LLAMA_SHARED
+#    if defined(_WIN32) && !defined(__MINGW32__)
+#        ifdef LLAMA_BUILD
+#            define LLAMA_API __declspec(dllexport)
+#        else
+#            define LLAMA_API __declspec(dllimport)
+#        endif
+#    else
+#        define LLAMA_API __attribute__ ((visibility ("default")))
+#    endif
+#else
+#    define LLAMA_API
+#endif
+
+#ifdef __GNUC__
+#    define DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
+#elif defined(_MSC_VER)
+#    define DEPRECATED(func, hint) __declspec(deprecated(hint)) func
+#else
+#    define DEPRECATED(func, hint) func
+#endif
+
+#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
+
+#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
+// Defined when llama.cpp is compiled with support for offloading model layers to GPU.
+#define LLAMA_SUPPORTS_GPU_OFFLOAD
+#endif
+
+#ifndef LLAMA_DEFAULT_RMS_EPS
+#define LLAMA_DEFAULT_RMS_EPS 5e-6f
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+    //
+    // C interface
+    //
+    // TODO: show sample usage
+    //
+
+    struct llama_model;
+    struct llama_context;
+
+    typedef int llama_token;
+
+    typedef struct llama_token_data {
+        llama_token id; // token id
+        float logit;    // log-odds of the token
+        float p;        // probability of the token
+    } llama_token_data;
+
+    typedef struct llama_token_data_array {
+        llama_token_data * data;
+        size_t size;
+        bool sorted;
+    } llama_token_data_array;
+
+    typedef void (*llama_progress_callback)(float progress, void *ctx);
+
+   struct llama_context_params {
+        uint32_t seed;         // RNG seed, -1 for random
+        int32_t  n_ctx;        // text context
+        int32_t  n_batch;      // prompt processing batch size
+        int32_t  n_gqa;        // grouped-query attention (TEMP - will be moved to model hparams)
+        float    rms_norm_eps; // rms norm epsilon (TEMP - will be moved to model hparams)
+        int32_t  n_gpu_layers; // number of layers to store in VRAM
+        int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
+
+        const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)
+
+        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
+        float    rope_freq_base;  // RoPE base frequency
+        float    rope_freq_scale; // RoPE frequency scaling factor
+
+        // called with a progress value between 0 and 1, pass NULL to disable
+        llama_progress_callback progress_callback;
+        // context pointer passed to the progress callback
+        void * progress_callback_user_data;
+
+        // Keep the booleans together to avoid misalignment during copy-by-value.
+        bool low_vram;   // if true, reduce VRAM usage at the cost of performance
+        bool f16_kv;     // use fp16 for KV cache
+        bool logits_all; // the llama_eval() call computes all logits, not just the last one
+        bool vocab_only; // only load the vocabulary, no weights
+        bool use_mmap;   // use mmap if possible
+        bool use_mlock;  // force system to keep model in RAM
+        bool embedding;  // embedding mode only
+    };
+    // model file types
+    enum llama_ftype {
+        LLAMA_FTYPE_ALL_F32              = 0,
+        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
+        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
+        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
+        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
+    };
+
+    // model quantization parameters
+    typedef struct llama_model_quantize_params {
+        int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
+        enum llama_ftype   ftype;    // quantize to this llama_ftype
+        bool allow_requantize;       // allow quantizing non-f32/f16 tensors
+        bool quantize_output_tensor; // quantize output.weight
+    } llama_model_quantize_params;
+
+    // grammar types
+    struct llama_grammar;
+
+    // grammar element type
+    enum llama_gretype {
+        // end of rule definition
+        LLAMA_GRETYPE_END            = 0,
+
+        // start of alternate definition for rule
+        LLAMA_GRETYPE_ALT            = 1,
+
+        // non-terminal element: reference to rule
+        LLAMA_GRETYPE_RULE_REF       = 2,
+
+        // terminal element: character (code point)
+        LLAMA_GRETYPE_CHAR           = 3,
+
+        // inverse char(s) ([^a], [^a-b] [^abc])
+        LLAMA_GRETYPE_CHAR_NOT       = 4,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
+        // be an inclusive range ([a-z])
+        LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,
+
+        // modifies a preceding LLAMA_GRETYPE_CHAR or
+        // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
+        LLAMA_GRETYPE_CHAR_ALT       = 6,
+    };
+
+    typedef struct llama_grammar_element {
+        enum llama_gretype type;
+        uint32_t           value; // Unicode code point or rule ID
+    } llama_grammar_element;
+
+    // performance timing information
+    struct llama_timings {
+        double t_start_ms;
+        double t_end_ms;
+        double t_load_ms;
+        double t_sample_ms;
+        double t_p_eval_ms;
+        double t_eval_ms;
+
+        int32_t n_sample;
+        int32_t n_p_eval;
+        int32_t n_eval;
+    };
+
+    LLAMA_API int llama_max_devices();
+
+    LLAMA_API struct llama_context_params llama_context_default_params();
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
+
+    LLAMA_API bool llama_mmap_supported();
+    LLAMA_API bool llama_mlock_supported();
+
+    // TODO: not great API - very likely to change
+    // Initialize the llama + ggml backend
+    // If numa is true, use NUMA optimizations
+    // Call once at the start of the program
+    LLAMA_API void llama_backend_init(bool numa);
+    // Call once at the end of the program - currently only used for MPI
+    LLAMA_API void llama_backend_free();
+
+    LLAMA_API int64_t llama_time_us();
+
+    LLAMA_API struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params);
+
+    LLAMA_API void llama_free_model(struct llama_model * model);
+
+    LLAMA_API struct llama_context * llama_new_context_with_model(
+                     struct llama_model * model,
+            struct llama_context_params   params);
+
+
+    // Frees all allocated memory
+    LLAMA_API void llama_free(struct llama_context * ctx);
+
+    // Returns 0 on success
+    LLAMA_API int llama_model_quantize(
+            const char * fname_inp,
+            const char * fname_out,
+            const llama_model_quantize_params * params);
+
+    // Apply a LoRA adapter to a loaded model
+    // path_base_model is the path to a higher quality model to use as a base for
+    // the layers modified by the adapter. Can be NULL to use the current loaded model.
+    // The model needs to be reloaded before applying a new adapter, otherwise the adapter
+    // will be applied on top of the previous one
+    // Returns 0 on success
+    LLAMA_API DEPRECATED(int llama_apply_lora_from_file(
+            struct llama_context * ctx,
+                      const char * path_lora,
+                      const char * path_base_model,
+                             int   n_threads),
+            "please use llama_model_apply_lora_from_file instead");
+
+    LLAMA_API int llama_model_apply_lora_from_file(
+            const struct llama_model * model,
+                      const char * path_lora,
+                      const char * path_base_model,
+                             int   n_threads);
+
+    // Returns the number of tokens in the KV cache
+    LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
+
+    // Sets the current rng seed.
+    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
+
+    // Returns the maximum size in bytes of the state (rng, logits, embedding
+    // and kv_cache) - will often be smaller after compacting tokens
+    LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
+
+    // Copies the state to the specified destination address.
+    // Destination needs to have allocated enough memory.
+    // Returns the number of bytes copied
+    LLAMA_API size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst);
+
+    // Set the state reading from the specified address
+    // Returns the number of bytes read
+    LLAMA_API size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src);
+
+    // Save/load session file
+    LLAMA_API bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out);
+    LLAMA_API bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count);
+
+    // Run the llama inference to obtain the logits and probabilities for the next token.
+    // tokens + n_tokens is the provided batch of new tokens to process
+    // n_past is the number of tokens to use from previous eval calls
+    // Returns 0 on success
+    LLAMA_API int llama_eval(
+            struct llama_context * ctx,
+               const llama_token * tokens,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Same as llama_eval, but use float matrix input directly.
+    LLAMA_API int llama_eval_embd(
+            struct llama_context * ctx,
+                     const float * embd,
+                             int   n_tokens,
+                             int   n_past,
+                             int   n_threads);
+
+    // Export a static computation graph for context of 511 and batch size of 1
+    // NOTE: since this functionality is mostly for debugging and demonstration purposes, we hardcode these
+    //       parameters here to keep things simple
+    // IMPORTANT: do not use for anything else other than debugging and testing!
+    LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
+
+    // Convert the provided text into tokens.
+    // The tokens pointer must be large enough to hold the resulting tokens.
+    // Returns the number of tokens on success, no more than n_max_tokens
+    // Returns a negative number on failure - the number of tokens that would have been returned
+    // TODO: not sure if correct
+    LLAMA_API int llama_tokenize(
+            struct llama_context * ctx,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_tokenize_with_model(
+        const struct llama_model * model,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
+    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
+    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
+    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
+
+    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
+    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
+    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
+
+    // Get the vocabulary as output parameters.
+    // Returns number of results.
+    LLAMA_API int llama_get_vocab(
+            const struct llama_context * ctx,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    LLAMA_API int llama_get_vocab_from_model(
+              const struct llama_model * model,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    // Token logits obtained from the last call to llama_eval()
+    // The logits for the last token are stored in the last row
+    // Can be mutated in order to change the probabilities of the next token
+    // Rows: n_tokens
+    // Cols: n_vocab
+    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
+
+    // Get the embeddings for the input
+    // shape: [n_embd] (1-dimensional)
+    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
+
+    // Token Id -> String. Uses the vocabulary in the provided context
+    LLAMA_API const char * llama_token_to_str(
+            const struct llama_context * ctx,
+                           llama_token   token);
+
+    LLAMA_API const char * llama_token_to_str_with_model(
+              const struct llama_model * model,
+                           llama_token   token);
+
+    // Special tokens
+    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl();   // next-line
+
+    // Grammar
+    //
+    LLAMA_API struct llama_grammar * llama_grammar_init(
+            const llama_grammar_element ** rules,
+                                 size_t    n_rules,
+                                 size_t    start_rule_index);
+
+    LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
+
+    // Sampling functions
+
+    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
+    LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
+
+    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
+    LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
+
+    /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
+    /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
+    /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
+    LLAMA_API void llama_sample_classifier_free_guidance(
+              struct llama_context * ctx,
+            llama_token_data_array * candidates,
+              struct llama_context * guidance_ctx,
+                             float   scale);
+
+    /// @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
+    LLAMA_API void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep);
+
+    /// @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
+    LLAMA_API void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
+
+    /// @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
+    LLAMA_API void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep);
+
+    /// @details Locally Typical Sampling implementation described in the paper https://arxiv.org/abs/2202.00666.
+    LLAMA_API void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
+    LLAMA_API void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates, float temp);
+
+    /// @details Apply constraints from grammar
+    LLAMA_API void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar);
+
+    /// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param m The number of tokens considered in the estimation of `s_hat`. This is an arbitrary value that is used to calculate `s_hat`, which in turn helps to calculate the value of `k`. In the paper, they use `m = 100`, but you can experiment with different values to see how it affects the performance of the algorithm.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu);
+
+    /// @details Mirostat 2.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
+    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
+    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
+    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
+    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
+    LLAMA_API llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu);
+
+    /// @details Selects the token with the highest probability.
+    LLAMA_API llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Randomly selects a token from the candidates based on their probabilities.
+    LLAMA_API llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates);
+
+    /// @details Accepts the sampled token into the grammar
+    LLAMA_API void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token);
+
+    // Performance information
+    LLAMA_API struct llama_timings llama_get_timings(struct llama_context * ctx);
+    LLAMA_API void llama_print_timings(struct llama_context * ctx);
+    LLAMA_API void llama_reset_timings(struct llama_context * ctx);
+
+    // Print system information
+    LLAMA_API const char * llama_print_system_info(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
+#include <vector>
+#include <string>
+struct ggml_tensor;
+
+const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
+
+#endif
+
+#endif // LLAMA_H
diff --git a/gguf-util.h b/gguf-util.h
index 6bbabf667911e..774ae57eef3db 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -1,567 +1,567 @@
-// GGUF counterpart of llama-util.h.
-// we may consider making it a part of ggml.c once GGUF work is complete.
-// this will require extra work to migrate this to pure C.
-// Contains wrappers around OS interfaces.
-
-#ifndef GGUF_UTIL_H
-#define GGUF_UTIL_H
-
-#include "ggml.h"
-
-#include <cstdio>
-#include <cstdint>
-#include <cerrno>
-#include <cstring>
-#include <cstdarg>
-#include <cstdlib>
-#include <climits>
-
-#include <string>
-#include <sstream>
-#include <vector>
-#include <stdexcept>
-
-#ifdef __has_include
-    #if __has_include(<unistd.h>)
-        #include <unistd.h>
-        #if defined(_POSIX_MAPPED_FILES)
-            #include <sys/mman.h>
-        #endif
-        #if defined(_POSIX_MEMLOCK_RANGE)
-            #include <sys/resource.h>
-        #endif
-    #endif
-#endif
-
-#if defined(_WIN32)
-    #define WIN32_LEAN_AND_MEAN
-    #ifndef NOMINMAX
-        #define NOMINMAX
-    #endif
-    #include <windows.h>
-    #include <io.h>
-    #include <stdio.h> // for _fseeki64
-#endif
-
-#ifdef __GNUC__
-#ifdef __MINGW32__
-__attribute__((format(gnu_printf, 1, 2)))
-#else
-__attribute__((format(printf, 1, 2)))
-#endif
-#endif
-static std::string format(const char * fmt, ...) {
-    va_list ap, ap2;
-    va_start(ap, fmt);
-    va_copy(ap2, ap);
-    int size = vsnprintf(NULL, 0, fmt, ap);
-    GGML_ASSERT(size >= 0 && size < INT_MAX);
-    std::vector<char> buf(size + 1);
-    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
-    GGML_ASSERT(size2 == size);
-    va_end(ap2);
-    va_end(ap);
-    return std::string(buf.data(), size);
-}
-
-template<typename T>
-static std::string to_string(const T & val) {
-    std::stringstream ss;
-    ss << val;
-    return ss.str();
-}
-
-// TODO: can we merge this one and gguf_context?
-struct gguf_file {
-    // use FILE * so we don't have to re-open the file to mmap
-    FILE * fp;
-    size_t size;
-
-    gguf_file(const char * fname, const char * mode) {
-        fp = std::fopen(fname, mode);
-        if (fp == NULL) {
-            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
-        }
-        seek(0, SEEK_END);
-        size = tell();
-        seek(0, SEEK_SET);
-    }
-
-    size_t tell() const {
-#ifdef _WIN32
-        __int64 ret = _ftelli64(fp);
-#else
-        long ret = std::ftell(fp);
-#endif
-        GGML_ASSERT(ret != -1); // this really shouldn't fail
-        return (size_t) ret;
-    }
-
-    void seek(size_t offset, int whence) {
-#ifdef _WIN32
-        int ret = _fseeki64(fp, (__int64) offset, whence);
-#else
-        int ret = std::fseek(fp, (long) offset, whence);
-#endif
-        GGML_ASSERT(ret == 0); // same
-    }
-
-    size_t write_str(const std::string & val) {
-        size_t total_written = 0;
-        const int32_t n = val.size();
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        total_written += sizeof(n);
-        fwrite(val.c_str(), n, 1, fp);
-        total_written += n;
-
-        return total_written;
-    }
-
-    size_t write_i32(int32_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-        return sizeof(val);
-    }
-
-    size_t write_u64(size_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-        return sizeof(val);
-    }
-
-    template<typename T>
-    void write_val(const std::string & key, enum gguf_type type, const T & val) {
-        write_str(key);
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-    }
-
-    template<typename T>
-    void write_arr(const std::string & key, enum gguf_type type, const std::vector<T> & val) {
-        write_str(key);
-        {
-            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
-        }
-
-        const int32_t n = val.size();
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        fwrite(val.data(), sizeof(T), n, fp);
-    }
-
-    void write_str(const std::string & key, enum gguf_type type, const std::string & val) {
-        write_str(key);
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-
-        const int32_t n = val.size();
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        fwrite(val.c_str(), n, 1, fp);
-    }
-
-    void write_str(const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
-        write_str(key);
-        {
-            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
-        }
-
-        const int32_t n = val.size();
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        for (int i = 0; i < n; ++i) {
-            const int32_t nstr = val[i].size();
-            fwrite((const char *) &nstr, sizeof(nstr), 1, fp);
-            fwrite(val[i].c_str(), nstr, 1, fp);
-        }
-    }
-
-    void write_zeros(size_t count) {
-        for (size_t i = 0; i < count; ++i) {
-            fputc(0, fp);
-        }
-    }
-
-    void read_raw(void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        std::size_t ret = std::fread(ptr, len, 1, fp);
-        if (ferror(fp)) {
-            throw std::runtime_error(format("read error: %s", strerror(errno)));
-        }
-        if (ret != 1) {
-            throw std::runtime_error(std::string("unexpectedly reached end of file"));
-        }
-    }
-
-    void write_raw(const void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        size_t ret = std::fwrite(ptr, len, 1, fp);
-        if (ret != 1) {
-            throw std::runtime_error(format("write error: %s", strerror(errno)));
-        }
-    }
-
-    ~gguf_file() {
-        if (fp) {
-            std::fclose(fp);
-        }
-    }
-};
-
-#if defined(_WIN32)
-static std::string gguf_format_win_err(DWORD err) {
-    LPSTR buf;
-    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
-                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
-    if (!size) {
-        return "FormatMessageA failed";
-    }
-    std::string ret(buf, size);
-    LocalFree(buf);
-    return ret;
-}
-#endif
-
-struct gguf_mmap {
-    void * addr;
-    size_t size;
-
-    gguf_mmap(const gguf_mmap &) = delete;
-
-#ifdef _POSIX_MAPPED_FILES
-    static constexpr bool SUPPORTED = true;
-
-    gguf_mmap(struct gguf_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
-        size = file->size;
-        int fd = fileno(file->fp);
-        int flags = MAP_SHARED;
-        // prefetch/readahead impairs performance on NUMA systems
-        if (numa) { prefetch = 0; }
-#ifdef __linux__
-        if (prefetch) { flags |= MAP_POPULATE; }
-#endif
-        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
-        if (addr == MAP_FAILED) {
-            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
-        }
-
-        if (prefetch > 0) {
-            // Advise the kernel to preload the mapped memory
-            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
-                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-        if (numa) {
-            // advise the kernel not to use readahead
-            // (because the next page might not belong on the same node)
-            if (madvise(addr, file->size, MADV_RANDOM)) {
-                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-    }
-
-    ~gguf_mmap() {
-        munmap(addr, size);
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    gguf_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
-        (void) numa;
-
-        size = file->size;
-
-        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
-
-        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
-        DWORD error = GetLastError();
-
-        if (hMapping == NULL) {
-            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
-        error = GetLastError();
-        CloseHandle(hMapping);
-
-        if (addr == NULL) {
-            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
-        if (prefetch) {
-            // Advise the kernel to preload the mapped memory
-            WIN32_MEMORY_RANGE_ENTRY range;
-            range.VirtualAddress = addr;
-            range.NumberOfBytes = (SIZE_T)size;
-            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
-                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-            }
-        }
-        #else
-        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
-        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
-    }
-
-    ~gguf_mmap() {
-        if (!UnmapViewOfFile(addr)) {
-            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    gguf_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
-        (void) prefetch;
-        (void) numa;
-
-        throw std::runtime_error(std::string("mmap not supported"));
-    }
-#endif
-};
-
-// Represents some region of memory being locked using mlock or VirtualLock;
-// will automatically unlock on destruction.
-struct gguf_mlock {
-    void * addr = NULL;
-    size_t size = 0;
-    bool failed_already = false;
-
-    gguf_mlock() {}
-    gguf_mlock(const gguf_mlock &) = delete;
-
-    ~gguf_mlock() {
-        if (size) {
-            raw_unlock(addr, size);
-        }
-    }
-
-    void init(void * ptr) {
-        GGML_ASSERT(addr == NULL && size == 0);
-        addr = ptr;
-    }
-
-    void grow_to(size_t target_size) {
-        GGML_ASSERT(addr);
-        if (failed_already) {
-            return;
-        }
-        size_t granularity = lock_granularity();
-        target_size = (target_size + granularity - 1) & ~(granularity - 1);
-        if (target_size > size) {
-            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
-                size = target_size;
-            } else {
-                failed_already = true;
-            }
-        }
-    }
-
-#ifdef _POSIX_MEMLOCK_RANGE
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        return (size_t) sysconf(_SC_PAGESIZE);
-    }
-
-    #ifdef __APPLE__
-        #define MLOCK_SUGGESTION \
-            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
-            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
-    #else
-        #define MLOCK_SUGGESTION \
-            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
-    #endif
-
-    bool raw_lock(const void * addr, size_t size) {
-        if (!mlock(addr, size)) {
-            return true;
-        } else {
-            char* errmsg = std::strerror(errno);
-            bool suggest = (errno == ENOMEM);
-
-            // Check if the resource limit is fine after all
-            struct rlimit lock_limit;
-            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
-                suggest = false;
-            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
-                suggest = false;
-
-            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
-                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
-            return false;
-        }
-    }
-
-    #undef MLOCK_SUGGESTION
-
-    void raw_unlock(void * addr, size_t size) {
-        if (munlock(addr, size)) {
-            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
-        }
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        SYSTEM_INFO si;
-        GetSystemInfo(&si);
-        return (size_t) si.dwPageSize;
-    }
-
-    bool raw_lock(void * ptr, size_t len) {
-        for (int tries = 1; ; tries++) {
-            if (VirtualLock(ptr, len)) {
-                return true;
-            }
-            if (tries == 2) {
-                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
-                    len, size, llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-
-            // It failed but this was only the first try; increase the working
-            // set size and try again.
-            SIZE_T min_ws_size, max_ws_size;
-            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
-                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-            // Per MSDN: "The maximum number of pages that a process can lock
-            // is equal to the number of pages in its minimum working set minus
-            // a small overhead."
-            // Hopefully a megabyte is enough overhead:
-            size_t increment = len + 1048576;
-            // The minimum must be <= the maximum, so we need to increase both:
-            min_ws_size += increment;
-            max_ws_size += increment;
-            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
-                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-        }
-    }
-
-    void raw_unlock(void * ptr, size_t len) {
-        if (!VirtualUnlock(ptr, len)) {
-            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
-                    gguf_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    size_t lock_granularity() {
-        return (size_t) 65536;
-    }
-
-    bool raw_lock(const void * addr, size_t len) {
-        fprintf(stderr, "warning: mlock not supported on this system\n");
-        return false;
-    }
-
-    void raw_unlock(const void * addr, size_t len) {}
-#endif
-};
-
-// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
-struct gguf_buffer {
-    uint8_t * addr = NULL;
-    size_t size = 0;
-
-    gguf_buffer() = default;
-
-    void resize(size_t len) {
-#ifdef GGML_USE_METAL
-        free(addr);
-        int result = posix_memalign((void **) &addr, getpagesize(), len);
-        if (result == 0) {
-            memset(addr, 0, len);
-        }
-        else {
-            addr = NULL;
-        }
-#else
-        delete[] addr;
-        addr = new uint8_t[len];
-#endif
-        size = len;
-    }
-
-    ~gguf_buffer() {
-#ifdef GGML_USE_METAL
-        free(addr);
-#else
-        delete[] addr;
-#endif
-        addr = NULL;
-    }
-
-    // disable copy and move
-    gguf_buffer(const gguf_buffer&) = delete;
-    gguf_buffer(gguf_buffer&&) = delete;
-    gguf_buffer& operator=(const gguf_buffer&) = delete;
-    gguf_buffer& operator=(gguf_buffer&&) = delete;
-};
-
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-struct gguf_ctx_buffer {
-    uint8_t * addr = NULL;
-    bool is_cuda;
-    size_t size = 0;
-
-    gguf_ctx_buffer() = default;
-
-    void resize(size_t size) {
-        free();
-
-        addr = (uint8_t *) ggml_cuda_host_malloc(size);
-        if (addr) {
-            is_cuda = true;
-        }
-        else {
-            // fall back to pageable memory
-            addr = new uint8_t[size];
-            is_cuda = false;
-        }
-        this->size = size;
-    }
-
-    void free() {
-        if (addr) {
-            if (is_cuda) {
-                ggml_cuda_host_free(addr);
-            }
-            else {
-                delete[] addr;
-            }
-        }
-        addr = NULL;
-    }
-
-    ~gguf_ctx_buffer() {
-        free();
-    }
-
-    // disable copy and move
-    gguf_ctx_buffer(const gguf_ctx_buffer&) = delete;
-    gguf_ctx_buffer(gguf_ctx_buffer&&) = delete;
-    gguf_ctx_buffer& operator=(const gguf_ctx_buffer&) = delete;
-    gguf_ctx_buffer& operator=(gguf_ctx_buffer&&) = delete;
-};
-#else
-typedef gguf_buffer gguf_ctx_buffer;
-#endif
-
-#endif
+// GGUF counterpart of llama-util.h.
+// we may consider making it a part of ggml.c once GGUF work is complete.
+// this will require extra work to migrate this to pure C.
+// Contains wrappers around OS interfaces.
+
+#ifndef GGUF_UTIL_H
+#define GGUF_UTIL_H
+
+#include "ggml.h"
+
+#include <cstdio>
+#include <cstdint>
+#include <cerrno>
+#include <cstring>
+#include <cstdarg>
+#include <cstdlib>
+#include <climits>
+
+#include <string>
+#include <sstream>
+#include <vector>
+#include <stdexcept>
+
+#ifdef __has_include
+    #if __has_include(<unistd.h>)
+        #include <unistd.h>
+        #if defined(_POSIX_MAPPED_FILES)
+            #include <sys/mman.h>
+        #endif
+        #if defined(_POSIX_MEMLOCK_RANGE)
+            #include <sys/resource.h>
+        #endif
+    #endif
+#endif
+
+#if defined(_WIN32)
+    #define WIN32_LEAN_AND_MEAN
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <windows.h>
+    #include <io.h>
+    #include <stdio.h> // for _fseeki64
+#endif
+
+#ifdef __GNUC__
+#ifdef __MINGW32__
+__attribute__((format(gnu_printf, 1, 2)))
+#else
+__attribute__((format(printf, 1, 2)))
+#endif
+#endif
+static std::string format(const char * fmt, ...) {
+    va_list ap, ap2;
+    va_start(ap, fmt);
+    va_copy(ap2, ap);
+    int size = vsnprintf(NULL, 0, fmt, ap);
+    GGML_ASSERT(size >= 0 && size < INT_MAX);
+    std::vector<char> buf(size + 1);
+    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
+    GGML_ASSERT(size2 == size);
+    va_end(ap2);
+    va_end(ap);
+    return std::string(buf.data(), size);
+}
+
+template<typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
+
+// TODO: can we merge this one and gguf_context?
+struct gguf_file {
+    // use FILE * so we don't have to re-open the file to mmap
+    FILE * fp;
+    size_t size;
+
+    gguf_file(const char * fname, const char * mode) {
+        fp = std::fopen(fname, mode);
+        if (fp == NULL) {
+            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
+        }
+        seek(0, SEEK_END);
+        size = tell();
+        seek(0, SEEK_SET);
+    }
+
+    size_t tell() const {
+#ifdef _WIN32
+        __int64 ret = _ftelli64(fp);
+#else
+        long ret = std::ftell(fp);
+#endif
+        GGML_ASSERT(ret != -1); // this really shouldn't fail
+        return (size_t) ret;
+    }
+
+    void seek(size_t offset, int whence) {
+#ifdef _WIN32
+        int ret = _fseeki64(fp, (__int64) offset, whence);
+#else
+        int ret = std::fseek(fp, (long) offset, whence);
+#endif
+        GGML_ASSERT(ret == 0); // same
+    }
+
+    size_t write_str(const std::string & val) {
+        size_t total_written = 0;
+        const int32_t n = val.size();
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        total_written += sizeof(n);
+        fwrite(val.c_str(), n, 1, fp);
+        total_written += n;
+
+        return total_written;
+    }
+
+    size_t write_i32(int32_t val) {
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+        return sizeof(val);
+    }
+
+    size_t write_u64(size_t val) {
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+        return sizeof(val);
+    }
+
+    template<typename T>
+    void write_val(const std::string & key, enum gguf_type type, const T & val) {
+        write_str(key);
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &val, sizeof(val), 1, fp);
+    }
+
+    template<typename T>
+    void write_arr(const std::string & key, enum gguf_type type, const std::vector<T> & val) {
+        write_str(key);
+        {
+            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
+        }
+
+        const int32_t n = val.size();
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        fwrite(val.data(), sizeof(T), n, fp);
+    }
+
+    void write_str(const std::string & key, enum gguf_type type, const std::string & val) {
+        write_str(key);
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+
+        const int32_t n = val.size();
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        fwrite(val.c_str(), n, 1, fp);
+    }
+
+    void write_str(const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
+        write_str(key);
+        {
+            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
+            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
+        }
+
+        const int32_t n = val.size();
+        fwrite((const char *) &type, sizeof(type), 1, fp);
+        fwrite((const char *) &n, sizeof(n), 1, fp);
+        for (int i = 0; i < n; ++i) {
+            const int32_t nstr = val[i].size();
+            fwrite((const char *) &nstr, sizeof(nstr), 1, fp);
+            fwrite(val[i].c_str(), nstr, 1, fp);
+        }
+    }
+
+    void write_zeros(size_t count) {
+        for (size_t i = 0; i < count; ++i) {
+            fputc(0, fp);
+        }
+    }
+
+    void read_raw(void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        std::size_t ret = std::fread(ptr, len, 1, fp);
+        if (ferror(fp)) {
+            throw std::runtime_error(format("read error: %s", strerror(errno)));
+        }
+        if (ret != 1) {
+            throw std::runtime_error(std::string("unexpectedly reached end of file"));
+        }
+    }
+
+    void write_raw(const void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        size_t ret = std::fwrite(ptr, len, 1, fp);
+        if (ret != 1) {
+            throw std::runtime_error(format("write error: %s", strerror(errno)));
+        }
+    }
+
+    ~gguf_file() {
+        if (fp) {
+            std::fclose(fp);
+        }
+    }
+};
+
+#if defined(_WIN32)
+static std::string gguf_format_win_err(DWORD err) {
+    LPSTR buf;
+    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
+                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
+    if (!size) {
+        return "FormatMessageA failed";
+    }
+    std::string ret(buf, size);
+    LocalFree(buf);
+    return ret;
+}
+#endif
+
+struct gguf_mmap {
+    void * addr;
+    size_t size;
+
+    gguf_mmap(const gguf_mmap &) = delete;
+
+#ifdef _POSIX_MAPPED_FILES
+    static constexpr bool SUPPORTED = true;
+
+    gguf_mmap(struct gguf_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
+        size = file->size;
+        int fd = fileno(file->fp);
+        int flags = MAP_SHARED;
+        // prefetch/readahead impairs performance on NUMA systems
+        if (numa) { prefetch = 0; }
+#ifdef __linux__
+        if (prefetch) { flags |= MAP_POPULATE; }
+#endif
+        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
+        if (addr == MAP_FAILED) {
+            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
+        }
+
+        if (prefetch > 0) {
+            // Advise the kernel to preload the mapped memory
+            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
+                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+        if (numa) {
+            // advise the kernel not to use readahead
+            // (because the next page might not belong on the same node)
+            if (madvise(addr, file->size, MADV_RANDOM)) {
+                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+    }
+
+    ~gguf_mmap() {
+        munmap(addr, size);
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    gguf_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) numa;
+
+        size = file->size;
+
+        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
+
+        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
+        DWORD error = GetLastError();
+
+        if (hMapping == NULL) {
+            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
+        error = GetLastError();
+        CloseHandle(hMapping);
+
+        if (addr == NULL) {
+            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+        if (prefetch) {
+            // Advise the kernel to preload the mapped memory
+            WIN32_MEMORY_RANGE_ENTRY range;
+            range.VirtualAddress = addr;
+            range.NumberOfBytes = (SIZE_T)size;
+            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
+                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+            }
+        }
+        #else
+        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
+        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
+    }
+
+    ~gguf_mmap() {
+        if (!UnmapViewOfFile(addr)) {
+            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    gguf_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
+        (void) prefetch;
+        (void) numa;
+
+        throw std::runtime_error(std::string("mmap not supported"));
+    }
+#endif
+};
+
+// Represents some region of memory being locked using mlock or VirtualLock;
+// will automatically unlock on destruction.
+struct gguf_mlock {
+    void * addr = NULL;
+    size_t size = 0;
+    bool failed_already = false;
+
+    gguf_mlock() {}
+    gguf_mlock(const gguf_mlock &) = delete;
+
+    ~gguf_mlock() {
+        if (size) {
+            raw_unlock(addr, size);
+        }
+    }
+
+    void init(void * ptr) {
+        GGML_ASSERT(addr == NULL && size == 0);
+        addr = ptr;
+    }
+
+    void grow_to(size_t target_size) {
+        GGML_ASSERT(addr);
+        if (failed_already) {
+            return;
+        }
+        size_t granularity = lock_granularity();
+        target_size = (target_size + granularity - 1) & ~(granularity - 1);
+        if (target_size > size) {
+            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
+                size = target_size;
+            } else {
+                failed_already = true;
+            }
+        }
+    }
+
+#ifdef _POSIX_MEMLOCK_RANGE
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        return (size_t) sysconf(_SC_PAGESIZE);
+    }
+
+    #ifdef __APPLE__
+        #define MLOCK_SUGGESTION \
+            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
+            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
+    #else
+        #define MLOCK_SUGGESTION \
+            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
+    #endif
+
+    bool raw_lock(const void * addr, size_t size) {
+        if (!mlock(addr, size)) {
+            return true;
+        } else {
+            char* errmsg = std::strerror(errno);
+            bool suggest = (errno == ENOMEM);
+
+            // Check if the resource limit is fine after all
+            struct rlimit lock_limit;
+            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
+                suggest = false;
+            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
+                suggest = false;
+
+            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
+                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
+            return false;
+        }
+    }
+
+    #undef MLOCK_SUGGESTION
+
+    void raw_unlock(void * addr, size_t size) {
+        if (munlock(addr, size)) {
+            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
+        }
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    size_t lock_granularity() {
+        SYSTEM_INFO si;
+        GetSystemInfo(&si);
+        return (size_t) si.dwPageSize;
+    }
+
+    bool raw_lock(void * ptr, size_t len) {
+        for (int tries = 1; ; tries++) {
+            if (VirtualLock(ptr, len)) {
+                return true;
+            }
+            if (tries == 2) {
+                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
+                    len, size, llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+
+            // It failed but this was only the first try; increase the working
+            // set size and try again.
+            SIZE_T min_ws_size, max_ws_size;
+            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
+                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+            // Per MSDN: "The maximum number of pages that a process can lock
+            // is equal to the number of pages in its minimum working set minus
+            // a small overhead."
+            // Hopefully a megabyte is enough overhead:
+            size_t increment = len + 1048576;
+            // The minimum must be <= the maximum, so we need to increase both:
+            min_ws_size += increment;
+            max_ws_size += increment;
+            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
+                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
+                        gguf_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+        }
+    }
+
+    void raw_unlock(void * ptr, size_t len) {
+        if (!VirtualUnlock(ptr, len)) {
+            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
+                    gguf_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    size_t lock_granularity() {
+        return (size_t) 65536;
+    }
+
+    bool raw_lock(const void * addr, size_t len) {
+        fprintf(stderr, "warning: mlock not supported on this system\n");
+        return false;
+    }
+
+    void raw_unlock(const void * addr, size_t len) {}
+#endif
+};
+
+// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
+struct gguf_buffer {
+    uint8_t * addr = NULL;
+    size_t size = 0;
+
+    gguf_buffer() = default;
+
+    void resize(size_t len) {
+#ifdef GGML_USE_METAL
+        free(addr);
+        int result = posix_memalign((void **) &addr, getpagesize(), len);
+        if (result == 0) {
+            memset(addr, 0, len);
+        }
+        else {
+            addr = NULL;
+        }
+#else
+        delete[] addr;
+        addr = new uint8_t[len];
+#endif
+        size = len;
+    }
+
+    ~gguf_buffer() {
+#ifdef GGML_USE_METAL
+        free(addr);
+#else
+        delete[] addr;
+#endif
+        addr = NULL;
+    }
+
+    // disable copy and move
+    gguf_buffer(const gguf_buffer&) = delete;
+    gguf_buffer(gguf_buffer&&) = delete;
+    gguf_buffer& operator=(const gguf_buffer&) = delete;
+    gguf_buffer& operator=(gguf_buffer&&) = delete;
+};
+
+#ifdef GGML_USE_CUBLAS
+#include "ggml-cuda.h"
+struct gguf_ctx_buffer {
+    uint8_t * addr = NULL;
+    bool is_cuda;
+    size_t size = 0;
+
+    gguf_ctx_buffer() = default;
+
+    void resize(size_t size) {
+        free();
+
+        addr = (uint8_t *) ggml_cuda_host_malloc(size);
+        if (addr) {
+            is_cuda = true;
+        }
+        else {
+            // fall back to pageable memory
+            addr = new uint8_t[size];
+            is_cuda = false;
+        }
+        this->size = size;
+    }
+
+    void free() {
+        if (addr) {
+            if (is_cuda) {
+                ggml_cuda_host_free(addr);
+            }
+            else {
+                delete[] addr;
+            }
+        }
+        addr = NULL;
+    }
+
+    ~gguf_ctx_buffer() {
+        free();
+    }
+
+    // disable copy and move
+    gguf_ctx_buffer(const gguf_ctx_buffer&) = delete;
+    gguf_ctx_buffer(gguf_ctx_buffer&&) = delete;
+    gguf_ctx_buffer& operator=(const gguf_ctx_buffer&) = delete;
+    gguf_ctx_buffer& operator=(gguf_ctx_buffer&&) = delete;
+};
+#else
+typedef gguf_buffer gguf_ctx_buffer;
+#endif
+
+#endif
diff --git a/gguf.py b/gguf.py
index 18f42267aa78d..93591d5c76fa9 100644
--- a/gguf.py
+++ b/gguf.py
@@ -1,339 +1,339 @@
-"""TODOs
-1. Implement writers for known architectures, LLaMA in particular.
-2. Add docstrings from the format specs.
-3. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
-"""
-
-import struct
-import constants
-from enum import IntEnum
-from typing import Any, IO, List
-
-import numpy as np
-import sys
-
-
-class GGMLQuantizationType(IntEnum):
-    F32 = 0
-    F16 = 1
-
-
-class GGUFValueType(IntEnum):
-    UINT8 = 0
-    INT8 = 1
-    UINT16 = 2
-    INT16 = 3
-    UINT32 = 4
-    INT32 = 5
-    FLOAT32 = 6
-    BOOL = 7
-    STRING = 8
-    ARRAY = 9
-
-    @staticmethod
-    def get_type(val):
-        if isinstance(val, str) or isinstance(val, bytes) or isinstance(val, bytearray):
-            return GGUFValueType.STRING
-        elif isinstance(val, list):
-            return GGUFValueType.ARRAY
-        elif isinstance(val, float):
-            return GGUFValueType.FLOAT32
-        elif isinstance(val, bool):
-            return GGUFValueType.BOOL
-        elif isinstance(val, int):
-            return GGUFValueType.INT32
-        else:
-            print("Unknown type: "+str(type(val)))
-            sys.exit()
-
-
-class GGUFWriter:
-    def __init__(self, fout: IO):
-        self.fout = fout
-        self.offset_tensor = 0
-        self.data_alignment = constants.GGUF_DEFAULT_ALIGNMENT
-        self.kv_data = b""
-        self.kv_data_count = 0
-        self.ti_data = b""
-        self.ti_data_count = 0
-
-    def write_header_to_file(self):
-        self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
-        self.fout.write(struct.pack("<I", constants.GGUF_VERSION))
-        self.fout.write(struct.pack("<I", self.ti_data_count))
-        self.fout.write(struct.pack("<I", self.kv_data_count))
-        self.flush()
-#        print("tensors " + str(self.ti_data_count) + " kv " + str(self.kv_data_count))
-
-    def write_kv_data_to_file(self):
-        self.fout.write(self.kv_data)
-        self.flush()
-
-    def write_ti_data_to_file(self):
-        self.fout.write(self.ti_data)
-        self.flush()
-
-    @classmethod
-    def open(cls, path: str) -> "GGUFWriter":
-        f = open(path, "wb")
-        return cls(f)
-
-    def add_key(self, key: str):
-        self.add_val(key, GGUFValueType.STRING, add_vtype=False)
-
-    def add_uint8(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.UINT8)
-
-    def add_int8(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.INT8)
-
-    def add_uint16(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.UINT16)
-
-    def add_int16(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.INT16)
-
-    def add_uint32(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.UINT32)
-
-    def add_int32(self, key: str, val: int):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.INT32)
-
-    def add_float32(self, key: str, val: float):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.FLOAT32)
-
-    def add_bool(self, key: str, val: bool):
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.BOOL)
-
-    def add_string(self, key: str, val: str):
-        if len(val) == 0: return
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.STRING)
-
-    def add_array(self, key: str, val: list):
-        if not isinstance(val, list):
-            raise ValueError("Value must be a list for array type")
-
-        self.add_key(key)
-        self.add_val(val, GGUFValueType.ARRAY)
-
-    def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool = True):
-        if vtype is None:
-            vtype = GGUFValueType.get_type(val)
-
-        if add_vtype:
-            self.kv_data += struct.pack("<I", vtype)
-            self.kv_data_count += 1
-
-        if vtype == GGUFValueType.UINT8:
-            self.kv_data += struct.pack("<B", val)
-        elif vtype == GGUFValueType.INT8:
-            self.kv_data += struct.pack("<b", val)
-        elif vtype == GGUFValueType.UINT16:
-            self.kv_data += struct.pack("<H", val)
-        elif vtype == GGUFValueType.INT16:
-            self.kv_data += struct.pack("<h", val)
-        elif vtype == GGUFValueType.UINT32:
-            self.kv_data += struct.pack("<I", val)
-        elif vtype == GGUFValueType.INT32:
-            self.kv_data += struct.pack("<i", val)
-        elif vtype == GGUFValueType.FLOAT32:
-            self.kv_data += struct.pack("<f", val)
-        elif vtype == GGUFValueType.BOOL:
-            self.kv_data += struct.pack("?", val)
-        elif vtype == GGUFValueType.STRING:
-            encoded_val = val.encode("utf8") if isinstance(val, str) else val
-            self.kv_data += struct.pack("<I", len(encoded_val))
-            self.kv_data += encoded_val
-        elif vtype == GGUFValueType.ARRAY:
-            ltype = set([GGUFValueType.get_type(item) for item in val])
-            assert len(ltype) == 1, "All items in a GGUF array should be of the same type"
-            self.kv_data += struct.pack("<I", list(ltype)[0])
-            self.kv_data += struct.pack("<I", len(val))
-            for item in val:
-                self.add_val(item, add_vtype=False)
-        else:
-            raise ValueError("Invalid GGUF metadata value type")
-
-    @staticmethod
-    def ggml_pad(x: int, n: int) -> int:
-        return ((x + n - 1) // n) * n
-
-    def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int):
-        encoded_name = name.encode("utf8")
-        self.ti_data += struct.pack("<I", len(encoded_name))
-        self.ti_data += encoded_name
-        n_dims = len(tensor_shape)
-        self.ti_data += struct.pack("<I", n_dims)
-        for i in range(n_dims):
-            self.ti_data += struct.pack("<I", tensor_shape[n_dims - 1 - i])
-
-        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
-        dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
-        self.ti_data += struct.pack("<I", dtype)
-        self.ti_data += struct.pack("<Q", self.offset_tensor)
-        self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
-        self.ti_data_count += 1
-
-    def write_tensor_to_file(self, tensor: np.ndarray):
-        pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
-        if pad != 0:
-            self.fout.write(bytes([0] * pad))
-
-        tensor.tofile(self.fout)
-
-        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
-        if pad != 0:
-            self.fout.write(bytes([0] * pad))
-
-    def flush(self):
-        self.fout.flush()
-
-    def close(self):
-        self.fout.close()
-
-    def add_architecture(self, architecture: str):
-        self.add_string(constants.KEY_GENERAL_ARCHITECTURE,
-                        architecture)
-
-    def add_author(self, author: str):
-        self.add_string(constants.KEY_GENERAL_AUTHOR, author)
-
-    def add_url(self, url: str):
-        self.add_string(constants.KEY_GENERAL_URL, url)
-
-    def add_description(self, description: str):
-        self.add_string(constants.KEY_GENERAL_DESCRIPTION, description)
-
-    def add_file_type(self, file_type: str):
-        self.add_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
-
-    def add_source_url(self, url: str):
-        self.add_string(constants.KEY_GENERAL_SOURCE_URL, url)
-
-    def add_source_hf_repo(self, repo: str):
-        self.add_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
-
-    def add_name(self, name: str):
-        self.add_string(constants.KEY_GENERAL_NAME, name)
-
-    def add_quantization_version(self, quantization_version: GGMLQuantizationType):
-        self.add_uint32(
-            constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
-
-    def add_custom_alignment(self, alignment: int):
-        self.data_alignment = alignment
-        self.add_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
-
-    def add_context_length(self, llm: str, length: int):
-        self.add_uint32(
-            constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
-
-    def add_embedding_length(self, llm: str, length: int):
-        self.add_uint32(
-            constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
-
-    def add_block_count(self, llm: str, length: int):
-        self.add_uint32(
-            constants.KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
-
-    def add_feed_forward_length(self, llm: str, length: int):
-        self.add_uint32(
-            constants.KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
-
-    def add_parallel_residual(self, llm: str, use: bool):
-        self.add_bool(
-            constants.KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
-
-    def add_tensor_data_layout(self, llm: str, layout: str):
-        self.add_string(
-            constants.KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
-
-    def add_head_count(self, llm: str, count: int):
-        self.add_uint32(
-            constants.KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
-
-    def add_head_count_kv(self, llm: str, count: int):
-        self.add_uint32(
-            constants.KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
-
-    def add_max_alibi_bias(self, llm: str, bias: float):
-        self.add_float32(
-            constants.KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
-
-    def add_clamp_kqv(self, llm: str, value: float):
-        self.add_float32(
-            constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
-
-    def add_layer_norm_eps(self, llm: str, value: float):
-        self.add_float32(
-            constants.KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
-
-    def add_layer_norm_rms_eps(self, llm: str, value: float):
-        self.add_float32(
-            constants.KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
-
-    def add_rope_dimension_count(self, llm: str, count: int):
-        self.add_uint32(
-            constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
-
-    def add_rope_scale(self, llm: str, value:  float):
-        self.add_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
-
-    def add_tokenizer_model(self, model: str):
-        self.add_string(constants.KEY_TOKENIZER_MODEL, model)
-
-    def add_token_list(self, tokens: List):
-        self.add_array(constants.KEY_TOKENIZER_LIST, tokens)
-
-    def add_token_merges(self, merges: List):
-        self.add_array(constants.KEY_TOKENIZER_MERGES, merges)
-
-    def add_token_scores(self, scores: List[float]):
-        self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
-
-    def add_bos_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_BOS_ID, id)
-
-    def add_eos_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_EOS_ID, id)
-
-    def add_unk_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_UNK_ID, id)
-
-    def add_sep_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_SEP_ID, id)
-
-    def add_pad_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_PAD_ID, id)
-
-
-# Example usage:
-if __name__ == "__main__":
-    # Example usage with a file
-    gguf_writer = GGUFWriter.open("example.gguf")
-
-    gguf_writer.add_architecture("llama")
-    gguf_writer.add_uint32("answer", 42)  # Write a 32-bit integer
-    gguf_writer.add_float32("answer_in_float", 42.0)  # Write a 32-bit float
-    gguf_writer.add_custom_alignment(64)
-    tensor1 = np.ones((32,), dtype=np.float32) * 100.0
-    tensor2 = np.ones((32,), dtype=np.float32) * 101.0
-    gguf_writer.add_tensor_info("tensor0", tensor1)
-    gguf_writer.add_tensor_info("tensor1", tensor2)
-
-    gguf_writer.write_header_to_file()
-    gguf_writer.write_kv_data_to_file()
-    gguf_writer.write_ti_data_to_file()
-    gguf_writer.write_tensor_to_file(tensor1)
-    gguf_writer.write_tensor_to_file(tensor2)
-
-    gguf_writer.close()
+"""TODOs
+1. Implement writers for known architectures, LLaMA in particular.
+2. Add docstrings from the format specs.
+3. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
+"""
+
+import struct
+import constants
+from enum import IntEnum
+from typing import Any, IO, List
+
+import numpy as np
+import sys
+
+
+class GGMLQuantizationType(IntEnum):
+    F32 = 0
+    F16 = 1
+
+
+class GGUFValueType(IntEnum):
+    UINT8 = 0
+    INT8 = 1
+    UINT16 = 2
+    INT16 = 3
+    UINT32 = 4
+    INT32 = 5
+    FLOAT32 = 6
+    BOOL = 7
+    STRING = 8
+    ARRAY = 9
+
+    @staticmethod
+    def get_type(val):
+        if isinstance(val, str) or isinstance(val, bytes) or isinstance(val, bytearray):
+            return GGUFValueType.STRING
+        elif isinstance(val, list):
+            return GGUFValueType.ARRAY
+        elif isinstance(val, float):
+            return GGUFValueType.FLOAT32
+        elif isinstance(val, bool):
+            return GGUFValueType.BOOL
+        elif isinstance(val, int):
+            return GGUFValueType.INT32
+        else:
+            print("Unknown type: "+str(type(val)))
+            sys.exit()
+
+
+class GGUFWriter:
+    def __init__(self, fout: IO):
+        self.fout = fout
+        self.offset_tensor = 0
+        self.data_alignment = constants.GGUF_DEFAULT_ALIGNMENT
+        self.kv_data = b""
+        self.kv_data_count = 0
+        self.ti_data = b""
+        self.ti_data_count = 0
+
+    def write_header_to_file(self):
+        self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
+        self.fout.write(struct.pack("<I", constants.GGUF_VERSION))
+        self.fout.write(struct.pack("<I", self.ti_data_count))
+        self.fout.write(struct.pack("<I", self.kv_data_count))
+        self.flush()
+#        print("tensors " + str(self.ti_data_count) + " kv " + str(self.kv_data_count))
+
+    def write_kv_data_to_file(self):
+        self.fout.write(self.kv_data)
+        self.flush()
+
+    def write_ti_data_to_file(self):
+        self.fout.write(self.ti_data)
+        self.flush()
+
+    @classmethod
+    def open(cls, path: str) -> "GGUFWriter":
+        f = open(path, "wb")
+        return cls(f)
+
+    def add_key(self, key: str):
+        self.add_val(key, GGUFValueType.STRING, add_vtype=False)
+
+    def add_uint8(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT8)
+
+    def add_int8(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT8)
+
+    def add_uint16(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT16)
+
+    def add_int16(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT16)
+
+    def add_uint32(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.UINT32)
+
+    def add_int32(self, key: str, val: int):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.INT32)
+
+    def add_float32(self, key: str, val: float):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.FLOAT32)
+
+    def add_bool(self, key: str, val: bool):
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.BOOL)
+
+    def add_string(self, key: str, val: str):
+        if len(val) == 0: return
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.STRING)
+
+    def add_array(self, key: str, val: list):
+        if not isinstance(val, list):
+            raise ValueError("Value must be a list for array type")
+
+        self.add_key(key)
+        self.add_val(val, GGUFValueType.ARRAY)
+
+    def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool = True):
+        if vtype is None:
+            vtype = GGUFValueType.get_type(val)
+
+        if add_vtype:
+            self.kv_data += struct.pack("<I", vtype)
+            self.kv_data_count += 1
+
+        if vtype == GGUFValueType.UINT8:
+            self.kv_data += struct.pack("<B", val)
+        elif vtype == GGUFValueType.INT8:
+            self.kv_data += struct.pack("<b", val)
+        elif vtype == GGUFValueType.UINT16:
+            self.kv_data += struct.pack("<H", val)
+        elif vtype == GGUFValueType.INT16:
+            self.kv_data += struct.pack("<h", val)
+        elif vtype == GGUFValueType.UINT32:
+            self.kv_data += struct.pack("<I", val)
+        elif vtype == GGUFValueType.INT32:
+            self.kv_data += struct.pack("<i", val)
+        elif vtype == GGUFValueType.FLOAT32:
+            self.kv_data += struct.pack("<f", val)
+        elif vtype == GGUFValueType.BOOL:
+            self.kv_data += struct.pack("?", val)
+        elif vtype == GGUFValueType.STRING:
+            encoded_val = val.encode("utf8") if isinstance(val, str) else val
+            self.kv_data += struct.pack("<I", len(encoded_val))
+            self.kv_data += encoded_val
+        elif vtype == GGUFValueType.ARRAY:
+            ltype = set([GGUFValueType.get_type(item) for item in val])
+            assert len(ltype) == 1, "All items in a GGUF array should be of the same type"
+            self.kv_data += struct.pack("<I", list(ltype)[0])
+            self.kv_data += struct.pack("<I", len(val))
+            for item in val:
+                self.add_val(item, add_vtype=False)
+        else:
+            raise ValueError("Invalid GGUF metadata value type")
+
+    @staticmethod
+    def ggml_pad(x: int, n: int) -> int:
+        return ((x + n - 1) // n) * n
+
+    def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int):
+        encoded_name = name.encode("utf8")
+        self.ti_data += struct.pack("<I", len(encoded_name))
+        self.ti_data += encoded_name
+        n_dims = len(tensor_shape)
+        self.ti_data += struct.pack("<I", n_dims)
+        for i in range(n_dims):
+            self.ti_data += struct.pack("<I", tensor_shape[n_dims - 1 - i])
+
+        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
+        dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
+        self.ti_data += struct.pack("<I", dtype)
+        self.ti_data += struct.pack("<Q", self.offset_tensor)
+        self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
+        self.ti_data_count += 1
+
+    def write_tensor_to_file(self, tensor: np.ndarray):
+        pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
+        if pad != 0:
+            self.fout.write(bytes([0] * pad))
+
+        tensor.tofile(self.fout)
+
+        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
+        if pad != 0:
+            self.fout.write(bytes([0] * pad))
+
+    def flush(self):
+        self.fout.flush()
+
+    def close(self):
+        self.fout.close()
+
+    def add_architecture(self, architecture: str):
+        self.add_string(constants.KEY_GENERAL_ARCHITECTURE,
+                        architecture)
+
+    def add_author(self, author: str):
+        self.add_string(constants.KEY_GENERAL_AUTHOR, author)
+
+    def add_url(self, url: str):
+        self.add_string(constants.KEY_GENERAL_URL, url)
+
+    def add_description(self, description: str):
+        self.add_string(constants.KEY_GENERAL_DESCRIPTION, description)
+
+    def add_file_type(self, file_type: str):
+        self.add_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
+
+    def add_source_url(self, url: str):
+        self.add_string(constants.KEY_GENERAL_SOURCE_URL, url)
+
+    def add_source_hf_repo(self, repo: str):
+        self.add_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
+
+    def add_name(self, name: str):
+        self.add_string(constants.KEY_GENERAL_NAME, name)
+
+    def add_quantization_version(self, quantization_version: GGMLQuantizationType):
+        self.add_uint32(
+            constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
+
+    def add_custom_alignment(self, alignment: int):
+        self.data_alignment = alignment
+        self.add_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
+
+    def add_context_length(self, llm: str, length: int):
+        self.add_uint32(
+            constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
+
+    def add_embedding_length(self, llm: str, length: int):
+        self.add_uint32(
+            constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
+
+    def add_block_count(self, llm: str, length: int):
+        self.add_uint32(
+            constants.KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
+
+    def add_feed_forward_length(self, llm: str, length: int):
+        self.add_uint32(
+            constants.KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
+
+    def add_parallel_residual(self, llm: str, use: bool):
+        self.add_bool(
+            constants.KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
+
+    def add_tensor_data_layout(self, llm: str, layout: str):
+        self.add_string(
+            constants.KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
+
+    def add_head_count(self, llm: str, count: int):
+        self.add_uint32(
+            constants.KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
+
+    def add_head_count_kv(self, llm: str, count: int):
+        self.add_uint32(
+            constants.KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
+
+    def add_max_alibi_bias(self, llm: str, bias: float):
+        self.add_float32(
+            constants.KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
+
+    def add_clamp_kqv(self, llm: str, value: float):
+        self.add_float32(
+            constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
+
+    def add_layer_norm_eps(self, llm: str, value: float):
+        self.add_float32(
+            constants.KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
+
+    def add_layer_norm_rms_eps(self, llm: str, value: float):
+        self.add_float32(
+            constants.KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
+
+    def add_rope_dimension_count(self, llm: str, count: int):
+        self.add_uint32(
+            constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
+
+    def add_rope_scale(self, llm: str, value:  float):
+        self.add_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
+
+    def add_tokenizer_model(self, model: str):
+        self.add_string(constants.KEY_TOKENIZER_MODEL, model)
+
+    def add_token_list(self, tokens: List):
+        self.add_array(constants.KEY_TOKENIZER_LIST, tokens)
+
+    def add_token_merges(self, merges: List):
+        self.add_array(constants.KEY_TOKENIZER_MERGES, merges)
+
+    def add_token_scores(self, scores: List[float]):
+        self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
+
+    def add_bos_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_BOS_ID, id)
+
+    def add_eos_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_EOS_ID, id)
+
+    def add_unk_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_UNK_ID, id)
+
+    def add_sep_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_SEP_ID, id)
+
+    def add_pad_token_id(self, id: int):
+        self.add_uint32(constants.KEY_TOKENIZER_PAD_ID, id)
+
+
+# Example usage:
+if __name__ == "__main__":
+    # Example usage with a file
+    gguf_writer = GGUFWriter.open("example.gguf")
+
+    gguf_writer.add_architecture("llama")
+    gguf_writer.add_uint32("answer", 42)  # Write a 32-bit integer
+    gguf_writer.add_float32("answer_in_float", 42.0)  # Write a 32-bit float
+    gguf_writer.add_custom_alignment(64)
+    tensor1 = np.ones((32,), dtype=np.float32) * 100.0
+    tensor2 = np.ones((32,), dtype=np.float32) * 101.0
+    gguf_writer.add_tensor_info("tensor0", tensor1)
+    gguf_writer.add_tensor_info("tensor1", tensor2)
+
+    gguf_writer.write_header_to_file()
+    gguf_writer.write_kv_data_to_file()
+    gguf_writer.write_ti_data_to_file()
+    gguf_writer.write_tensor_to_file(tensor1)
+    gguf_writer.write_tensor_to_file(tensor2)
+
+    gguf_writer.close()

From 6f64b6c0f85b4abfb8d919a1a542899dbfda5487 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 13:51:09 +0200
Subject: [PATCH 143/242] Create convert-llama-7b-pth-to-gguf.py

---
 convert-llama-7b-pth-to-gguf.py | 302 ++++++++++++++++++++++++++++++++
 1 file changed, 302 insertions(+)
 create mode 100644 convert-llama-7b-pth-to-gguf.py

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
new file mode 100644
index 0000000000000..d763503564c1d
--- /dev/null
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -0,0 +1,302 @@
+# 7b pth llama --> gguf conversion, GQA/70b not supported
+# Only models with a single datafile are supported, like 7B
+# HF files required in the model dir: config.json tokenizer_config.json tokenizer.json tokenizer.model
+
+import gguf
+import gguf_namemap as tmap
+import os
+import sys
+import struct
+import json
+import numpy as np
+import torch
+from typing import Any, List
+from pathlib import Path
+from sentencepiece import SentencePieceProcessor
+
+
+#NDArray = np.ndarray[Any, Any]
+# compatible with python < 3.9
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+
+def count_model_parts(dir_model: str) -> int:
+    num_parts = 0
+    for filename in os.listdir(dir_model):
+        if filename.startswith("consolidated."):
+            num_parts += 1
+
+    if num_parts > 0:
+        print("gguf: found " + str(num_parts) + " model parts")
+    return num_parts
+
+if len(sys.argv) < 3:
+    print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
+    print("  ftype == 0 -> float32")
+    print("  ftype == 1 -> float16")
+    sys.exit(1)
+
+
+# output in the same directory as the model
+dir_model = sys.argv[1]
+last_dir = os.path.basename(os.path.normpath(dir_model))
+
+
+# possible tensor data types
+#   ftype == 0 -> float32
+#   ftype == 1 -> float16
+#
+# map from ftype to string
+ftype_str = ["f32", "f16"]
+
+ftype = 1
+if len(sys.argv) > 2:
+    ftype = int(sys.argv[2])
+    if ftype < 0 or ftype > 1:
+        print("Invalid ftype: " + str(ftype))
+        sys.exit(1)
+
+fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+print("gguf: loading model "+last_dir)
+
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+if hparams["architectures"][0] != "LlamaForCausalLM":
+    print("Model architecture not supported: " + hparams["architectures"][0])
+    sys.exit()
+
+# get number of model parts
+num_parts = count_model_parts(dir_model)
+
+if num_parts > 1:
+    print("gguf: Only models with a single datafile are supported.")
+    sys.exit()
+
+gguf_writer = gguf.GGUFWriter.open(fname_out)
+
+
+print("gguf: get model metadata")
+
+llm_arch = "llama"
+block_count = hparams["num_hidden_layers"]
+head_count = hparams["num_attention_heads"]
+
+if "num_key_value_heads" in hparams:
+    head_count_kv = hparams["num_key_value_heads"]
+else:
+    head_count_kv = head_count
+
+if "_name_or_path" in hparams:
+    hf_repo = hparams["_name_or_path"]
+else:
+    hf_repo=""
+
+gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_name(last_dir)
+gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
+gguf_writer.add_source_hf_repo(hf_repo)
+gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
+gguf_writer.add_block_count(llm_arch, block_count)
+gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
+gguf_writer.add_head_count(llm_arch, head_count)
+gguf_writer.add_head_count_kv(llm_arch, head_count_kv)
+gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
+
+
+# TOKENIZATION
+
+print("gguf: get tokenizer metadata")
+
+tokens: List[str] = []
+scores: List[float] = []
+
+if Path(dir_model + "/tokenizer.model").is_file():
+    # vocab type sentencepiece
+    print("gguf: get sentencepiece tokenizer vocab and scores")
+
+    tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
+
+    for i in range(tokenizer.vocab_size()):
+        text: bytes
+        if tokenizer.is_unknown(i):
+            text = " \u2047 ".encode("utf-8")
+        elif tokenizer.is_control(i):
+            text = b""
+        if tokenizer.is_byte(i):
+            piece = tokenizer.id_to_piece(i)
+            if len(piece) != 6:
+                raise Exception(f"Invalid token: {piece}")
+            byte_value = int(piece[3:-1], 16)
+            text = struct.pack("B", byte_value)
+        else:
+            text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
+        score: float = tokenizer.get_score(i)
+
+        tokens.append(text)
+        scores.append(score)
+
+    gguf_writer.add_tokenizer_model("llama")
+    gguf_writer.add_token_list(tokens)
+    gguf_writer.add_token_scores(scores)
+
+if Path(dir_model + "/tokenizer.json").is_file():
+    with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
+        tokenizer = json.load(f)
+
+    if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
+        print("gguf: get special token ids")
+
+        with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
+            tokenizer_config = json.load(f)
+
+        # find special token ids
+
+        if "bos_token" in tokenizer_config and tokenizer_config["bos_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["bos_token"]["content"]:
+                    gguf_writer.add_bos_token_id(key["id"])
+
+        if "eos_token" in tokenizer_config and tokenizer_config["eos_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["eos_token"]["content"]:
+                    gguf_writer.add_eos_token_id(key["id"])
+
+        if "unk_token" in tokenizer_config and tokenizer_config["unk_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["unk_token"]["content"]:
+                    gguf_writer.add_unk_token_id(key["id"])
+
+        if "sep_token" in tokenizer_config and tokenizer_config["sep_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["sep_token"]["content"]:
+                    gguf_writer.add_sep_token_id(key["id"])
+
+        if "pad_token" in tokenizer_config and tokenizer_config["pad_token"] != None:
+            for key in tokenizer["added_tokens"]:
+                if key["content"] == tokenizer_config["pad_token"]["content"]:
+                    gguf_writer.add_pad_token_id(key["id"])
+
+
+# TENSORS
+
+tensor_map = tmap.get_tensor_namemap(block_count)
+
+# tensor info
+print("gguf: get tensor metadata")
+
+part_names = ( f"consolidated.{n:02}.pth" for n in range(0, num_parts) )
+
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
+
+    for name in model_part.keys():
+        data = model_part[name]
+
+        # we don't need these
+        if name == "rope.freqs":
+            continue
+
+        # convert any unsupported data types to float32
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data_dtype = np.float32
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data_dtype = np.float32
+
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data_dtype = np.float16
+
+        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+
+        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
+
+
+print("gguf: write header")
+gguf_writer.write_header_to_file()
+print("gguf: write metadata")
+gguf_writer.write_kv_data_to_file()
+print("gguf: write tensor metadata")
+gguf_writer.write_ti_data_to_file()
+
+# tensor data
+print("gguf: convert and write tensor data")
+
+part_names = ( f"consolidated.{n:02}.pth" for n in range(0, num_parts) )
+
+for part_name in part_names:
+    print("gguf: loading model part '"+ part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
+
+    for name in model_part.keys():
+        data = model_part[name]
+
+    
+        old_dtype = data.dtype
+
+        # we don't need these
+        if name == "rope.freqs":
+            continue
+
+        # convert any unsupported data types to float32
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        data = data.squeeze().numpy()
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print( "Can not map tensor '" + name + "'" )
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data.dtype == np.float16:
+            data = data.astype(np.float32)
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
+
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+
+        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
+
+        gguf_writer.write_tensor_to_file(data)
+
+gguf_writer.close()
+
+
+print("gguf: model successfully exported to '" + fname_out + "'")
+print("")

From f00780b2ee99955fec8c02cd9f90aa230c9e026d Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 16:28:44 +0300
Subject: [PATCH 144/242] llama : sync gguf-llama.cpp with latest llama.cpp
 (#2608)

* llama : sync gguf-llama.cpp with latest llama.cpp

* minor : indentation + assert

* llama : refactor gguf_buffer and gguf_ctx_buffer

* llama : minor
---
 examples/gguf/gguf.cpp |  23 +-
 ggml-metal.h           |   3 +
 ggml-metal.m           |  15 +
 gguf-llama.cpp         | 981 ++++++++++++++++++++++++++---------------
 gguf-llama.h           |  28 +-
 gguf-util.h            |  97 ----
 6 files changed, 688 insertions(+), 459 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index 6f454a2047ddd..b32367f301ee9 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -8,14 +8,19 @@
 #include <sstream>
 #include <fstream>
 #include <vector>
-/*
+
+#undef MIN
+#undef MAX
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+
 template<typename T>
 static std::string to_string(const T & val) {
     std::stringstream ss;
     ss << val;
     return ss.str();
 }
-*/
+
 void gguf_ex_write_str(std::ofstream & fout, const std::string & val) {
     const int32_t n = val.size();
     fout.write((const char *) &n, sizeof(n));
@@ -377,28 +382,28 @@ bool gguf_ex_read_2(const std::string & fname) {
 
     struct gguf_file file(fname.c_str(), "rb");
     gguf_mmap data_mmap(&file, 0, false);
+
     const int n_tensors = gguf_get_n_tensors(ctx);
 
     for (int i = 0; i < n_tensors; ++i) {
-        const char * name             = gguf_get_tensor_name(ctx, i);
-        const size_t offset      = gguf_get_data_offset(ctx) + gguf_get_tensor_offset(ctx, i);
+        const char * name   = gguf_get_tensor_name(ctx, i);
+        const size_t offset = gguf_get_data_offset(ctx) + gguf_get_tensor_offset(ctx, i);
+
         struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
 
         cur->data = static_cast<char *>(data_mmap.addr) + offset;
 
         // print first 10 elements
-    const float * data = (const float *) cur->data;
+        const float * data = (const float *) cur->data;
 
         printf("%s data[:10] : ", name);
-
-        for (int j = 0; j < 10; ++j) {
+        for (int j = 0; j < MIN(10, ggml_nelements(cur)); ++j) {
             printf("%f ", data[j]);
         }
-
         printf("\n\n");
     }
 
-fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
+    fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
 
     ggml_free(ctx_data);
     gguf_free(ctx);
diff --git a/ggml-metal.h b/ggml-metal.h
index 16f1a0caacfac..1a5d96c339913 100644
--- a/ggml-metal.h
+++ b/ggml-metal.h
@@ -38,6 +38,9 @@ struct ggml_metal_context;
 struct ggml_metal_context * ggml_metal_init(int n_cb);
 void ggml_metal_free(struct ggml_metal_context * ctx);
 
+void * ggml_metal_host_malloc(size_t n);
+void   ggml_metal_host_free  (void * data);
+
 // set the number of command buffers to use
 void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb);
 
diff --git a/ggml-metal.m b/ggml-metal.m
index b47a98e214b61..4e4491414ef20 100644
--- a/ggml-metal.m
+++ b/ggml-metal.m
@@ -224,6 +224,21 @@ void ggml_metal_free(struct ggml_metal_context * ctx) {
     free(ctx);
 }
 
+void * ggml_metal_host_malloc(size_t n) {
+    void * data = NULL;
+    const int result = posix_memalign((void **) &data, getpagesize(), n);
+    if (result != 0) {
+        fprintf(stderr, "%s: error: posix_memalign failed\n", __func__);
+        return NULL;
+    }
+
+    return data;
+}
+
+void ggml_metal_host_free(void * data) {
+    free(data);
+}
+
 void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb) {
     ctx->n_cb = n_cb;
 }
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 99da3c56dc92c..0f8eb3c902b1f 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -47,7 +47,6 @@
 #include <algorithm>
 #include <initializer_list>
 #include <thread>
-#include <atomic>
 #include <mutex>
 #include <sstream>
 #include <numeric>
@@ -56,28 +55,75 @@
 #pragma warning(disable: 4244 4267) // possible loss of data
 #endif
 
+static void llama_log_internal(llama_log_level level, const char* format, ...);
+static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data);
+#define LLAMA_LOG_INFO(...)  llama_log_internal(LLAMA_LOG_LEVEL_INFO , __VA_ARGS__)
+#define LLAMA_LOG_WARN(...)  llama_log_internal(LLAMA_LOG_LEVEL_WARN , __VA_ARGS__)
+#define LLAMA_LOG_ERROR(...) llama_log_internal(LLAMA_LOG_LEVEL_ERROR, __VA_ARGS__)
+
+template<typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
+
+#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
+#include "ggml-alloc.h"
+#define LLAMA_USE_ALLOCATOR
+#else
 #define LLAMA_USE_SCRATCH
 #define LLAMA_MAX_SCRATCH_BUFFERS 16
+#endif
 
-// available llama models
-enum e_model {
-    MODEL_UNKNOWN,
-    MODEL_3B,
-    MODEL_7B,
-    MODEL_13B,
-    MODEL_30B,
-    MODEL_65B,
-    MODEL_70B,
-};
+typedef void (*offload_func_t)(struct ggml_tensor * tensor);
 
-static const size_t kB = 1024;
-static const size_t MB = 1024*1024;
+#ifdef GGML_USE_CUBLAS
+#define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
+#define llama_host_free(data) ggml_cuda_host_free(data)
+#elif GGML_USE_METAL
+#define llama_host_malloc(n)  ggml_metal_host_malloc(n)
+#define llama_host_free(data) ggml_metal_host_free(data)
+#else
+#define llama_host_malloc(n)  malloc(n)
+#define llama_host_free(data) free(data)
+#endif
 
-// computed for n_ctx == 2048
-// TODO: dynamically determine these sizes
-//       needs modifications in ggml
+struct llama_buffer {
+    void * data = NULL;
+    size_t size = 0;
 
-typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+    // fallback to malloc / free
+    // useful in cases where CUDA can try to allocate PINNED memory
+    bool fallback = false;
+
+    void resize(size_t n) {
+        llama_host_free(data);
+
+        data = llama_host_malloc(n);
+        if (!data) {
+            fallback = true;
+            data = malloc(n);
+        } else {
+            fallback = false;
+        }
+
+        GGML_ASSERT(data);
+        size = n;
+    }
+
+    ~llama_buffer() {
+        if (data) {
+            if (fallback) { // NOLINT
+                free(data);
+            } else {
+                llama_host_free(data);
+            }
+        }
+
+        data = NULL;
+    }
+};
 
 void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
     (void) tensor;
@@ -102,6 +148,24 @@ static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph *
 // memory sizes (calculated for n_batch == 512)
 //
 
+// computed for n_ctx == 2048
+// TODO: dynamically determine these sizes
+//       needs modifications in ggml
+
+// available llama models
+enum e_model {
+    MODEL_UNKNOWN,
+    MODEL_3B,
+    MODEL_7B,
+    MODEL_13B,
+    MODEL_30B,
+    MODEL_65B,
+    MODEL_70B,
+};
+
+static const size_t kB = 1024;
+static const size_t MB = 1024*1024;
+
 static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
 {
     static std::map<e_model, size_t> k_sizes = {
@@ -143,7 +207,7 @@ static const std::map<e_model, size_t> & MEM_REQ_EVAL()
 }
 
 // amount of VRAM needed per batch size to hold temporary results
-// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+// the values for 3b are not derived from testing but instead chosen conservatively
 static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
 {
     static std::map<e_model, size_t> k_sizes = {
@@ -151,14 +215,14 @@ static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
         { MODEL_7B,   512ull * kB },
         { MODEL_13B,  640ull * kB },
         { MODEL_30B,  768ull * kB },
-        { MODEL_65B, 1536ull * kB },
-        { MODEL_70B, 1536ull * kB }, // TODO (likely can be reduced)
+        { MODEL_65B, 1280ull * kB },
+        { MODEL_70B, 1280ull * kB },
     };
     return k_sizes;
 }
 
 // amount of VRAM needed per batch size and context to hold temporary results
-// the values for 3b and 65b are not derived from testing but instead chosen conservatively
+// the values for 3b are not derived from testing but instead chosen conservatively
 static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
 {
     static std::map<e_model, size_t> k_sizes = {
@@ -166,8 +230,8 @@ static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
         { MODEL_7B,  128ull },
         { MODEL_13B, 160ull },
         { MODEL_30B, 208ull },
-        { MODEL_65B, 416ull },
-        { MODEL_70B, 416ull }, // TODO (likely can be reduced)
+        { MODEL_65B, 256ull },
+        { MODEL_70B, 256ull },
     };
     return k_sizes;
 }
@@ -175,15 +239,15 @@ static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
 // default hparams (LLaMA 7B)
 struct llama_hparams {
     uint32_t n_vocab   = 32000;
-    uint32_t n_ctx     = 512;   // this is provided as user input?
+    uint32_t n_ctx     = 512;
     uint32_t n_embd    = 4096;
     uint32_t n_head    = 32;
     uint32_t n_head_kv = 32;
     uint32_t n_layer   = 32;
     uint32_t n_rot     = 64;
-    uint32_t n_ff       = 11008;
+    uint32_t n_ff      = 11008;
 
-    float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
+    float f_rms_norm_eps = 1e-5;
 
     float rope_freq_base  = 10000.0f;
     float rope_freq_scale = 1.0f;
@@ -241,7 +305,7 @@ struct llama_kv_cache {
 
     struct ggml_context * ctx = NULL;
 
-    gguf_ctx_buffer buf;
+    llama_buffer buf;
 
     int n; // number of tokens currently in the cache
 
@@ -292,7 +356,7 @@ struct llama_model {
     struct ggml_context * ctx = NULL;
 
     // the model memory buffer
-    gguf_ctx_buffer buf;
+    llama_buffer buf;
 
     // model memory mapped file
     std::unique_ptr<gguf_mmap> mapping;
@@ -329,13 +393,22 @@ struct llama_model {
 
 struct llama_context {
     llama_context(const llama_model & model) : model(model), t_load_us(model.t_load_us), t_start_us(model.t_start_us) {}
-#ifdef GGML_USE_METAL
     ~llama_context() {
+        if (model_owner) {
+            delete &model;
+        }
+#ifdef GGML_USE_METAL
         if (ctx_metal) {
             ggml_metal_free(ctx_metal);
         }
-    }
 #endif
+#ifdef LLAMA_USE_ALLOCATOR
+        if (alloc) {
+            ggml_allocr_free(alloc);
+        }
+#endif
+    }
+
     std::mt19937 rng;
 
     bool has_evaluated_once = false;
@@ -372,8 +445,19 @@ struct llama_context {
 
     // memory buffers used to evaluate the model
     // TODO: move in llama_state
-    gguf_ctx_buffer buf_compute;
-    gguf_ctx_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+    llama_buffer buf_compute;
+
+#ifdef LLAMA_USE_ALLOCATOR
+    llama_buffer buf_alloc;
+    ggml_allocr * alloc = NULL;
+#endif
+
+#ifdef LLAMA_USE_SCRATCH
+    llama_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+
+    int    buf_last = 0;
+    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
+#endif
 
 #ifdef GGML_USE_METAL
     ggml_metal_context * ctx_metal = NULL;
@@ -383,10 +467,7 @@ struct llama_context {
     ggml_mpi_context * ctx_mpi = NULL;
 #endif
 
-    int    buf_last = 0;
-    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
-
-    void use_buf(struct ggml_context * ctx, int i) {
+    static void use_buf(struct ggml_context * ctx, int i) {
 #if defined(LLAMA_USE_SCRATCH)
         size_t last_size = 0;
 
@@ -394,7 +475,7 @@ struct llama_context {
             last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
         } else {
             auto & buf = buf_scratch[i];
-            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.addr, });
+            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.data, });
         }
 
         if (buf_last >= 0) {
@@ -408,7 +489,7 @@ struct llama_context {
 #endif
     }
 
-    size_t get_buf_max_mem(int i) const {
+    static size_t get_buf_max_mem(int i) {
 #if defined(LLAMA_USE_SCRATCH)
         return buf_max_size[i];
 #else
@@ -418,6 +499,14 @@ struct llama_context {
     }
 };
 
+struct llama_state {
+    // We save the log callback globally
+    llama_log_callback log_callback = llama_log_callback_default;
+    void * log_callback_user_data = nullptr;
+};
+// global state
+static llama_state g_state;
+
 template <typename T>
 static T checked_mul(T a, T b) {
     T ret = a * b;
@@ -470,17 +559,16 @@ struct gguf_load_tensors_map {
 
 enum gguf_file_version {
     GGUF_FILE_VERSION_V1 = 1,
-
 };
 
-
 struct gguf_file_loader {
     gguf_file file;
     gguf_context * gguf_ctx;
     gguf_file_version file_version;
     llama_hparams hparams;
     llama_vocab vocab;
-struct ggml_context * ctx_data = NULL;
+
+    struct ggml_context * ctx_data = NULL;
 
     gguf_file_loader(const char * fname, gguf_load_tensors_map & tensors_map)
         : file(fname, "rb") {
@@ -499,7 +587,7 @@ struct ggml_context * ctx_data = NULL;
         read_tensor_metadata(tensors_map);
     }
 
-    uint32_t read_u32(const char * key) {
+    uint32_t read_u32(const char * key) const {
         int i = gguf_find_key(gguf_ctx, key);
         if (i == -1) {
             throw std::runtime_error(format("cannot find param with key %s\n", key));
@@ -508,7 +596,7 @@ struct ggml_context * ctx_data = NULL;
         return gguf_get_val_u32(gguf_ctx, i);
     }
 
-    float read_f32(const char * key) {
+    float read_f32(const char * key) const {
         int i = gguf_find_key(gguf_ctx, key);
         if (i == -1) {
             throw std::runtime_error(format("cannot find param with key %s\n", key));
@@ -517,27 +605,26 @@ struct ggml_context * ctx_data = NULL;
         return gguf_get_val_f32(gguf_ctx, i);
     }
 
-    int read_n_vocab() {
+    int read_n_vocab() const {
         int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
-    if (i == -1) {
-        throw std::runtime_error("cannot find token list in GGUF file\n");
-    }
+        if (i == -1) {
+            throw std::runtime_error("cannot find token list in GGUF file\n");
+        }
 
-    return gguf_get_arr_n(gguf_ctx, i);
+        return gguf_get_arr_n(gguf_ctx, i);
     }
 
     void read_hparams() {
-
         // TODO define keys as constants in header
         // TODO: read all hparams from file
 
-        hparams.n_vocab = read_n_vocab();
-        hparams.n_ctx   = read_u32("llama.context_length");
-        hparams.n_embd  = read_u32("llama.embedding_length");
-        hparams.n_ff    = read_u32("llama.feed_forward_length");
-        hparams.n_head  = read_u32("llama.attention.head_count");
-        hparams.n_layer = read_u32("llama.layer_count");
-        hparams.n_rot   = read_u32("llama.rope.dimension_count");
+        hparams.n_vocab        = read_n_vocab();
+        hparams.n_ctx          = read_u32("llama.context_length");
+        hparams.n_embd         = read_u32("llama.embedding_length");
+        hparams.n_ff           = read_u32("llama.feed_forward_length");
+        hparams.n_head         = read_u32("llama.attention.head_count");
+        hparams.n_layer        = read_u32("llama.layer_count");
+        hparams.n_rot          = read_u32("llama.rope.dimension_count");
         hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
 
         // LLaMAv2
@@ -568,7 +655,7 @@ struct ggml_context * ctx_data = NULL;
         }
     }
 
-    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) {
+    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) const {
         const int n_tensors = gguf_get_n_tensors(gguf_ctx);
 
         for (int i = 0; i < n_tensors; ++i) {
@@ -576,16 +663,19 @@ struct ggml_context * ctx_data = NULL;
             const char * name = gguf_get_tensor_name(gguf_ctx, i);
 
             struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
-            uint32_t n_dims = cur->n_dims;
+
+            const uint32_t n_dims = cur->n_dims;
             tensor.type = cur->type;
             tensor.ne.resize(n_dims);
+
             for (uint32_t j = 0; j < n_dims; ++j) {
-            tensor.ne[j] = cur->ne[j];
+                tensor.ne[j] = cur->ne[j];
             }
 
             if (n_dims < 1 || n_dims > 2) {
                 throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
             }
+
             switch (tensor.type) {
                 case GGML_TYPE_F32:
                 case GGML_TYPE_F16:
@@ -605,7 +695,6 @@ struct ggml_context * ctx_data = NULL;
                 }
             }
 
-
             tensor.file_off = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, i);
 
             tensor.name = name;
@@ -632,47 +721,47 @@ struct gguf_file_saver {
 
     gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
         : file(fname, "wb"), fl(fl) {
-        fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
-        write_header();
-        write_hparams(new_ftype);
-    }
+            fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
+            write_header();
+            write_hparams(new_ftype);
+        }
 
     void write_header() {
         const int32_t magic = GGUF_MAGIC;
         file.write_i32(magic);
 
-            const int32_t version = GGUF_VERSION;
-            file.write_i32(version);
+        const int32_t version = GGUF_VERSION;
+        file.write_i32(version);
 
-            const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
-            file.write_i32(n_tensors);
+        const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
+        file.write_i32(n_tensors);
 
-            const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
-            file.write_i32(n_kv);
-        }
-
-        void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
-            std::vector<std::string> data(n_arr);
+        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
+        file.write_i32(n_kv);
+    }
 
-            for (int j = 0; j < n_arr; ++j) {
-                std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
-                data[j] = val;
-                }
+    void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
+        std::vector<std::string> data(n_arr);
 
-                            file.write_arr<std::string>(key, type, data);
+        for (int j = 0; j < n_arr; ++j) {
+            std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
+            data[j] = val;
         }
 
-        void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
-            std::vector<float> data(n_arr);
+        file.write_arr<std::string>(key, type, data);
+    }
 
-            for (int j = 0; j < n_arr; ++j) {
-                float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
-                data[j] = val;
-                }
+    void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
+        std::vector<float> data(n_arr);
 
-                            file.write_arr<float>(key, type, data);
+        for (int j = 0; j < n_arr; ++j) {
+            float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
+            data[j] = val;
         }
 
+        file.write_arr<float>(key, type, data);
+    }
+
     void write_hparams(enum llama_ftype new_ftype) {
         const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
         for (int i = 0; i < n_kv; ++i) {
@@ -696,59 +785,62 @@ struct gguf_file_saver {
 
                 switch(vtype) {
                     case GGUF_TYPE_BOOL:
-                    bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
-                    file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
-                    break;
+                        bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
+                        file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
+                        break;
                     case GGUF_TYPE_FLOAT32:
-                    f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
-                    file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
-                    break;
+                        f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
+                        file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
+                        break;
                     case GGUF_TYPE_INT16:
-                    i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
-                    file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
-                    break;
+                        i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
+                        file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
+                        break;
                     case GGUF_TYPE_INT32:
-                    i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
-                    file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
-                    break;
+                        i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
+                        file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
+                        break;
                     case GGUF_TYPE_INT8:
-                    i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
-                    file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
-                    break;
+                        i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
+                        file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
+                        break;
                     case GGUF_TYPE_STRING:
-                    str_val = gguf_get_val_str(fl->gguf_ctx, i);
-                    file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
-                    break;
+                        str_val = gguf_get_val_str(fl->gguf_ctx, i);
+                        file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
+                        break;
                     case GGUF_TYPE_UINT16:
-                    u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
-                    file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
-                    break;
+                        u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
+                        file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
+                        break;
                     case GGUF_TYPE_UINT32:
-                    u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
-                    file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
-                    break;
+                        u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
+                        file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
+                        break;
                     case GGUF_TYPE_UINT8:
-                    u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
-                    file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
-                    break;
+                        u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
+                        file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
+                        break;
                     case GGUF_TYPE_ARRAY:
-                    arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
-                    n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
-                    if (arr_type == GGUF_TYPE_FLOAT32) {
-                        write_hparam_arr_f32(key, arr_type, i, n_arr);
+                        arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
+                        n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
+                        if (arr_type == GGUF_TYPE_FLOAT32) {
+                            write_hparam_arr_f32(key, arr_type, i, n_arr);
                         } else if (arr_type == GGUF_TYPE_STRING) {
                             write_hparam_arr_str(key, GGUF_TYPE_STRING, i, n_arr);
                         } else {
                             throw std::runtime_error("not implemented");
                         }
-                    break;
+                        break;
                     default:
-                    throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
+                        throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
                 }
             }
         }
 
-        info_offset  = file.tell();
+        info_offset = file.tell();
+
+        GGML_ASSERT(gguf_get_data_offset(fl->gguf_ctx) >= info_offset);
+
         size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
         file.write_zeros(count);
         file.seek(info_offset, SEEK_SET);
@@ -983,7 +1075,7 @@ static bool kv_cache_init(
 
     struct ggml_init_params params;
     params.mem_size   = cache.buf.size;
-    params.mem_buffer = cache.buf.addr;
+    params.mem_buffer = cache.buf.data;
     params.no_alloc   = false;
 
     cache.ctx = ggml_init(params);
@@ -1016,8 +1108,6 @@ struct llama_context_params llama_context_default_params() {
         /*.seed                        =*/ LLAMA_DEFAULT_SEED,
         /*.n_ctx                       =*/ 512,
         /*.n_batch                     =*/ 512,
-        /*.n_gqa                       =*/ 1,
-        /*.rms_norm_eps                =*/ LLAMA_DEFAULT_RMS_EPS,
         /*.gpu_layers                  =*/ 0,
         /*.main_gpu                    =*/ 0,
         /*.tensor_split                =*/ nullptr,
@@ -1026,6 +1116,7 @@ struct llama_context_params llama_context_default_params() {
         /*.progress_callback           =*/ nullptr,
         /*.progress_callback_user_data =*/ nullptr,
         /*.low_vram                    =*/ false,
+        /*.mul_mat_q                   =*/ false,
         /*.f16_kv                      =*/ true,
         /*.logits_all                  =*/ false,
         /*.vocab_only                  =*/ false,
@@ -1144,11 +1235,10 @@ static void llama_model_load_internal(
         llama_vocab & vocab,
         int n_ctx,
         int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
         int n_gpu_layers,
         int main_gpu,
         const float * tensor_split,
+        const bool mul_mat_q,
         float rope_freq_base,
         float rope_freq_scale,
         bool low_vram,
@@ -1158,8 +1248,6 @@ static void llama_model_load_internal(
         bool vocab_only,
         llama_progress_callback progress_callback,
         void * progress_callback_user_data) {
-            GGML_UNUSED(rms_norm_eps); // TODO: update function signature to remove this
-
     model.t_start_us = ggml_time_us();
 
     std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
@@ -1189,11 +1277,15 @@ static void llama_model_load_internal(
         hparams.n_ctx = n_ctx;
 
         // LLaMAv2
-        hparams.n_head_kv = hparams.n_head / n_gqa;
-        if (model.type == e_model::MODEL_65B && n_gqa == 8) {
-            fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
-            model.type = e_model::MODEL_70B;
+        // TODO: probably not needed
+        {
+            const auto n_gqa = hparams.n_gqa();
+
+            if (model.type == e_model::MODEL_65B && n_gqa == 8) {
+                fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
+                model.type = e_model::MODEL_70B;
             }
+        }
 
         hparams.rope_freq_base  = rope_freq_base;
         hparams.rope_freq_scale = rope_freq_scale;
@@ -1202,27 +1294,21 @@ static void llama_model_load_internal(
     const uint32_t n_ff = hparams.n_ff;
 
     {
-        fprintf(stderr, "%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
-        fprintf(stderr, "%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
-        fprintf(stderr, "%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
-        fprintf(stderr, "%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        fprintf(stderr, "%s: n_head     = %u\n",   __func__, hparams.n_head);
-        fprintf(stderr, "%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
-        fprintf(stderr, "%s: n_layer    = %u\n",   __func__, hparams.n_layer);
-        fprintf(stderr, "%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        fprintf(stderr, "%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        fprintf(stderr, "%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
-        fprintf(stderr, "%s: n_ff       = %u\n",   __func__, n_ff);
-        fprintf(stderr, "%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
-        fprintf(stderr, "%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
-        fprintf(stderr, "%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
-        fprintf(stderr, "%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
-    }
-
-    if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
-        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
-        hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
-        throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
+        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
+        LLAMA_LOG_INFO("%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
+        LLAMA_LOG_INFO("%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
+        LLAMA_LOG_INFO("%s: n_embd     = %u\n",   __func__, hparams.n_embd);
+        LLAMA_LOG_INFO("%s: n_head     = %u\n",   __func__, hparams.n_head);
+        LLAMA_LOG_INFO("%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
+        LLAMA_LOG_INFO("%s: n_layer    = %u\n",   __func__, hparams.n_layer);
+        LLAMA_LOG_INFO("%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        LLAMA_LOG_INFO("%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
+        LLAMA_LOG_INFO("%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
+        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, n_ff);
+        LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
+        LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
+        LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
+        LLAMA_LOG_INFO("%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
     }
 
     if (vocab_only) {
@@ -1234,19 +1320,19 @@ static void llama_model_load_internal(
     size_t ctx_size;
     size_t mmapped_size;
     ml->calc_sizes(&ctx_size, &mmapped_size);
-    fprintf(stderr, "%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
+    LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
 
     // create the ggml context
     {
         model.buf.resize(ctx_size);
         if (use_mlock) {
-            model.mlock_buf.init   (model.buf.addr);
+            model.mlock_buf.init   (model.buf.data);
             model.mlock_buf.grow_to(model.buf.size);
         }
 
         struct ggml_init_params params = {
             /*.mem_size   =*/ model.buf.size,
-            /*.mem_buffer =*/ model.buf.addr,
+            /*.mem_buffer =*/ model.buf.data,
             /*.no_alloc   =*/ ml->use_mmap,
         };
 
@@ -1257,13 +1343,15 @@ static void llama_model_load_internal(
     }
 
     (void) main_gpu;
+    (void) mul_mat_q;
 #if defined(GGML_USE_CUBLAS)
-    fprintf(stderr, "%s: using CUDA for GPU acceleration\n", __func__);
+    LLAMA_LOG_INFO("%s: using CUDA for GPU acceleration\n", __func__);
     ggml_cuda_set_main_device(main_gpu);
+    ggml_cuda_set_mul_mat_q(mul_mat_q);
 #define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
 #define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU_SPLIT
 #elif defined(GGML_USE_CLBLAST)
-    fprintf(stderr, "%s: using OpenCL for GPU acceleration\n", __func__);
+    LLAMA_LOG_INFO("%s: using OpenCL for GPU acceleration\n", __func__);
 #define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
 #define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU
 #else
@@ -1353,25 +1441,29 @@ static void llama_model_load_internal(
         const size_t scale = memory_type == GGML_TYPE_F32 ? 2 : 1;
 
         // this is the total memory required to run the inference
-        const size_t mem_required =
+        size_t mem_required =
             ctx_size +
-            mmapped_size - vram_weights + // weights in VRAM not in memory
+            mmapped_size - vram_weights; // weights in VRAM not in memory
+
+#ifndef LLAMA_USE_ALLOCATOR
+        mem_required +=
             MEM_REQ_SCRATCH0(hparams.n_ctx).at(model.type) +
             MEM_REQ_SCRATCH1().at(model.type) +
             MEM_REQ_EVAL().at(model.type);
+#endif
 
         // this is the memory required by one llama_state
         const size_t mem_required_state =
             scale*hparams.kv_size();
 
-        fprintf(stderr, "%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
+        LLAMA_LOG_INFO("%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
                 mem_required / 1024.0 / 1024.0, mem_required_state / 1024.0 / 1024.0);
 
         (void) vram_scratch;
         (void) n_batch;
 #ifdef GGML_USE_CUBLAS
         if (low_vram) {
-            fprintf(stderr, "%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
+            LLAMA_LOG_INFO("%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
             ggml_cuda_set_scratch_size(0); // disable scratch
         } else {
             const size_t vram_scratch_base = VRAM_REQ_SCRATCH_BASE().at(model.type);
@@ -1379,7 +1471,7 @@ static void llama_model_load_internal(
             vram_scratch = n_batch * (vram_scratch_base + n_ctx * vram_scratch_per_context);
             ggml_cuda_set_scratch_size(vram_scratch);
             if (n_gpu_layers > 0) {
-                fprintf(stderr, "%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
+                LLAMA_LOG_INFO("%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
                         __func__, vram_scratch_base / kB, vram_scratch_per_context,
                         (vram_scratch + MB - 1) / MB); // round up
             }
@@ -1389,9 +1481,9 @@ static void llama_model_load_internal(
 #if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
         const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
 
-        fprintf(stderr, "%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
+        LLAMA_LOG_INFO("%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
         if (n_gpu_layers > (int) hparams.n_layer) {
-            fprintf(stderr, "%s: offloading non-repeating layers to GPU\n", __func__);
+            LLAMA_LOG_INFO("%s: offloading non-repeating layers to GPU\n", __func__);
         }
         size_t vram_kv_cache = 0;
 
@@ -1400,17 +1492,17 @@ static void llama_model_load_internal(
         const int max_offloadable_layers = low_vram ? hparams.n_layer + 1 : hparams.n_layer + 3;
         if (n_gpu_layers > (int) hparams.n_layer + 1) {
             if (low_vram) {
-                fprintf(stderr, "%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
+                LLAMA_LOG_INFO("%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
             } else {
-                fprintf(stderr, "%s: offloading v cache to GPU\n", __func__);
+                LLAMA_LOG_INFO("%s: offloading v cache to GPU\n", __func__);
                 vram_kv_cache += hparams.kv_size() / 2;
             }
         }
         if (n_gpu_layers > (int) hparams.n_layer + 2) {
             if (low_vram) {
-                fprintf(stderr, "%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
+                LLAMA_LOG_WARN("%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
             } else {
-                fprintf(stderr, "%s: offloading k cache to GPU\n", __func__);
+                LLAMA_LOG_INFO("%s: offloading k cache to GPU\n", __func__);
                 vram_kv_cache += hparams.kv_size() / 2;
             }
         }
@@ -1419,9 +1511,9 @@ static void llama_model_load_internal(
         const int max_offloadable_layers = hparams.n_layer + 1;
 #endif // GGML_USE_CUBLAS
 
-        fprintf(stderr, "%s: offloaded %d/%d layers to GPU\n",
+        LLAMA_LOG_INFO("%s: offloaded %d/%d layers to GPU\n",
                 __func__, std::min(n_gpu_layers, max_offloadable_layers), max_backend_supported_layers);
-        fprintf(stderr, "%s: total VRAM used: %zu MB\n",
+        LLAMA_LOG_INFO("%s: total VRAM used: %zu MB\n",
                 __func__, (vram_weights + vram_scratch + vram_kv_cache + MB - 1) / MB); // round up
 #else
         (void) n_gpu_layers;
@@ -1459,11 +1551,10 @@ static bool llama_model_load(
         llama_vocab & vocab,
         int n_ctx,
         int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
         int n_gpu_layers,
         int main_gpu,
         const float * tensor_split,
+        const bool mul_mat_q,
         float rope_freq_base,
         float rope_freq_scale,
         bool low_vram,
@@ -1474,41 +1565,25 @@ static bool llama_model_load(
         llama_progress_callback progress_callback,
         void *progress_callback_user_data) {
     try {
-        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gqa, rms_norm_eps, n_gpu_layers, main_gpu, tensor_split, rope_freq_base, rope_freq_scale, low_vram, memory_type,
+        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gpu_layers,
+                                  main_gpu, tensor_split, mul_mat_q, rope_freq_base, rope_freq_scale, low_vram, memory_type,
                                   use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
         return true;
     } catch (const std::exception & err) {
-        fprintf(stderr, "error loading model: %s\n", err.what());
+        LLAMA_LOG_ERROR("error loading model: %s\n", err.what());
         return false;
     }
 }
 
-// evaluate the transformer
-//
-//   - lctx:      llama context
-//   - tokens:    new batch of tokens to process
-//   - embd       embeddings input
-//   - n_tokens   number of tokens
-//   - n_past:    the context size so far
-//   - n_threads: number of threads to use
-//
-static bool llama_eval_internal(
+static struct ggml_cgraph * llama_build_graph(
          llama_context & lctx,
      const llama_token * tokens,
            const float * embd,
                    int   n_tokens,
-                   int   n_past,
-                   int   n_threads,
-            const char * cgraph_fname) {
+                   int   n_past) {
 
     GGML_ASSERT((!tokens && embd) || (tokens && !embd));
 
-#ifdef GGML_USE_MPI
-    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
-#endif
-
-    const int64_t t_start_us = ggml_time_us();
-
     const int N = n_tokens;
 
     const auto & model   = lctx.model;
@@ -1524,7 +1599,6 @@ static bool llama_eval_internal(
     const int64_t n_head      = hparams.n_head;
     const int64_t n_head_kv   = hparams.n_head_kv;
     const int64_t n_embd_head = hparams.n_embd_head();
-    const int64_t n_vocab     = hparams.n_vocab;
     const int64_t n_embd_gqa  = hparams.n_embd_gqa();
 
 
@@ -1539,26 +1613,35 @@ static bool llama_eval_internal(
     auto & mem_per_token = lctx.mem_per_token;
     auto & buf_compute   = lctx.buf_compute;
 
+
     struct ggml_init_params params = {
         /*.mem_size   =*/ buf_compute.size,
-        /*.mem_buffer =*/ buf_compute.addr,
+        /*.mem_buffer =*/ buf_compute.data,
         /*.no_alloc   =*/ false,
     };
 
+#ifdef LLAMA_USE_ALLOCATOR
+    params.no_alloc = true;
+#endif
+
     struct ggml_context * ctx0 = ggml_init(params);
 
     ggml_cgraph * gf = ggml_new_graph(ctx0);
 
-    // for big prompts, if BLAS is enabled, it is better to use only one thread
-    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
-    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
-
     struct ggml_tensor * cur;
     struct ggml_tensor * inpL;
 
     if (tokens) {
         struct ggml_tensor * inp_tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
+
+#ifdef LLAMA_USE_ALLOCATOR
+        ggml_allocr_alloc(lctx.alloc, inp_tokens);
+        if (!ggml_allocr_is_measure(lctx.alloc)) {
+            memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
+        }
+#else
         memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
+#endif
         ggml_set_name(inp_tokens, "inp_tokens");
 
         inpL = ggml_get_rows(ctx0, model.tok_embeddings, inp_tokens);
@@ -1568,7 +1651,15 @@ static bool llama_eval_internal(
 #endif
 
         inpL = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N);
+
+#ifdef LLAMA_USE_ALLOCATOR
+        ggml_allocr_alloc(lctx.alloc, inpL);
+        if (!ggml_allocr_is_measure(lctx.alloc)) {
+            memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
+        }
+#else
         memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
+#endif
     }
 
     const int i_gpu_start = n_layer - n_gpu_layers;
@@ -1595,6 +1686,17 @@ static bool llama_eval_internal(
     }
 #endif // GGML_USE_CUBLAS
 
+    struct ggml_tensor * KQ_scale = ggml_new_tensor_1d(ctx0, GGML_TYPE_F32, 1);
+#ifdef LLAMA_USE_ALLOCATOR
+    ggml_allocr_alloc(lctx.alloc, KQ_scale);
+    if (!ggml_allocr_is_measure(lctx.alloc)) {
+        ggml_set_f32(KQ_scale, 1.0f/sqrtf(float(n_embd)/n_head));
+    }
+#else
+    ggml_set_f32(KQ_scale, 1.0f/sqrtf(float(n_embd)/n_head));
+#endif
+    ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
+
     for (int il = 0; il < n_layer; ++il) {
         ggml_format_name(inpL, "layer_inp_%d", il);
 
@@ -1690,9 +1792,6 @@ static bool llama_eval_internal(
             ggml_set_name(KQ, "KQ");
 
             // KQ_scaled = KQ / sqrt(n_embd_head)
-            struct ggml_tensor * KQ_scale = ggml_new_f32(ctx0, 1.0f/sqrtf(float(n_embd)/n_head));
-            ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
-
             // KQ_scaled shape [n_past + N, N, n_head, 1]
             struct ggml_tensor * KQ_scaled = ggml_scale_inplace(ctx0, KQ, KQ_scale);
             offload_func_kq(KQ_scaled);
@@ -1808,9 +1907,6 @@ static bool llama_eval_internal(
 
     lctx.use_buf(ctx0, 0);
 
-    // used at the end to optionally extract the embeddings
-    struct ggml_tensor * embeddings = NULL;
-
     // norm
     {
         cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
@@ -1821,8 +1917,6 @@ static bool llama_eval_internal(
         cur = ggml_mul(ctx0, cur, model.norm);
         // offload_func_nr(cur); // TODO CPU + GPU mirrored backend
         ggml_set_name(cur, "result_norm");
-
-        embeddings = cur;
     }
 
     // lm_head
@@ -1834,23 +1928,103 @@ static bool llama_eval_internal(
     // logits -> probs
     //cur = ggml_soft_max_inplace(ctx0, cur);
 
-    // run the computation
     ggml_build_forward_expand(gf, cur);
 
-    // fprintf(stderr, "graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf.n_nodes, gf.n_leafs);
+    if (mem_per_token == 0) {
+        mem_per_token = ggml_used_mem(ctx0)/N;
+    }
+
+#if 0
+    LLAMA_LOG_INFO("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
+            ggml_used_mem(ctx0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(0)/1024.0/1024.0,
+            lctx.get_buf_max_mem(1)/1024.0/1024.0,
+            lctx.work_buffer.size()/1024.0/1024.0,
+            n_past, N);
+#endif
+
+    ggml_free(ctx0);
+
+    return gf;
+}
+
+// evaluate the transformer
+//
+//   - lctx:      llama context
+//   - tokens:    new batch of tokens to process
+//   - embd       embeddings input
+//   - n_tokens   number of tokens
+//   - n_past:    the context size so far
+//   - n_threads: number of threads to use
+//
+static bool llama_eval_internal(
+         llama_context & lctx,
+     const llama_token * tokens,
+           const float * embd,
+                   int   n_tokens,
+                   int   n_past,
+                   int   n_threads,
+            const char * cgraph_fname) {
+
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
+
+    const int64_t t_start_us = ggml_time_us();
+
+#ifdef GGML_USE_MPI
+    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
+#endif
+
+    const int N = n_tokens;
+
+    const auto & model   = lctx.model;
+    const auto & hparams = model.hparams;
+
+    const auto & kv_self = lctx.kv_self;
+
+    GGML_ASSERT(!!kv_self.ctx);
+
+    const int64_t n_embd      = hparams.n_embd;
+    const int64_t n_vocab     = hparams.n_vocab;
+
+#ifdef LLAMA_USE_ALLOCATOR
+    ggml_allocr_reset(lctx.alloc);
+#endif
+
+    ggml_cgraph * gf = llama_build_graph(lctx, tokens, embd, n_tokens, n_past);
+
+#ifdef LLAMA_USE_ALLOCATOR
+    ggml_allocr_alloc_graph(lctx.alloc, gf);
+#endif
+
+    // LLAMA_LOG_INFO("graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf->n_nodes, gf->n_leafs);
+
+    // for big prompts, if BLAS is enabled, it is better to use only one thread
+    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
+    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
+
+    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
+    struct ggml_tensor * embeddings = gf->nodes[gf->n_nodes - 2];
+
+    GGML_ASSERT(strcmp(res->name, "result_output") == 0);
+    GGML_ASSERT(strcmp(embeddings->name, "result_norm") == 0);
 
 #if GGML_USE_MPI
+    const int64_t n_layer = hparams.n_layer;
     ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);
 #endif
 
 #ifdef GGML_USE_METAL
     if (lctx.ctx_metal && N == 1) {
-        if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
-            ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
-        }
+        // TODO: disabled until #2413 is resolved
+        //if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
+        //    ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
+        //}
         ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
         ggml_metal_graph_compute(lctx.ctx_metal, gf);
-        ggml_metal_get_tensor   (lctx.ctx_metal, cur);
+        ggml_metal_get_tensor   (lctx.ctx_metal, res);
+        if (!lctx.embedding.empty()) {
+            ggml_metal_get_tensor(lctx.ctx_metal, embeddings);
+        }
     } else {
         // IMPORTANT:
         // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
@@ -1881,8 +2055,6 @@ static bool llama_eval_internal(
     // update kv token count
     lctx.kv_self.n = n_past + N;
 
-    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
-
     if (cgraph_fname) {
         ggml_graph_export(gf, cgraph_fname);
     }
@@ -1920,21 +2092,6 @@ static bool llama_eval_internal(
         memcpy(embedding_out.data(), (float *) ggml_get_data(embeddings) + (n_embd*(N - 1)), sizeof(float)*n_embd);
     }
 
-    if (mem_per_token == 0) {
-        mem_per_token = ggml_used_mem(ctx0)/N;
-    }
-
-#if 0
-    printf("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
-            ggml_used_mem(ctx0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(1)/1024.0/1024.0,
-            lctx.work_buffer.size()/1024.0/1024.0,
-            n_past, N);
-#endif
-
-    ggml_free(ctx0);
-
     // measure the performance only for the single-token evals
     if (N == 1) {
         lctx.t_eval_us += ggml_time_us() - t_start_us;
@@ -2026,7 +2183,7 @@ struct llama_tokenizer {
             left_sym.n += right_sym.n;
             right_sym.n = 0;
 
-            //printf("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
+            //LLAMA_LOG_INFO("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
 
             // remove the right sym from the chain
             left_sym.next = right_sym.next;
@@ -2046,7 +2203,9 @@ struct llama_tokenizer {
             if (token == vocab_.token_to_id.end()) {
                 // output any symbols that did not form tokens as bytes.
                 for (int j = 0; j < (int) symbol.n; ++j) {
-                    llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
+                    // NOTE: old version, before #2420 - not sure what are the implications of this
+                    //llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
+                    llama_vocab::id token_id = vocab_.token_to_id.at(std::string(1, symbol.text[j]));
                     output.push_back(token_id);
                 }
             } else {
@@ -2904,11 +3063,11 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
 // quantization
 //
 
-static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, gguf_buffer & output, const int nelements, const int nthread) {
-    if (output.size < nelements * sizeof(float)) {
-        output.resize(nelements * sizeof(float));
+static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, std::vector<float> & output, const size_t nelements, const int nthread) {
+    if (output.size() < nelements) {
+        output.resize(nelements);
     }
-    float * f32_output = (float *) output.addr;
+    float * f32_output = (float *) output.data();
 
     ggml_type_traits_t qtype;
     if (ggml_is_quantized(tensor.type)) {
@@ -3026,13 +3185,16 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     };
 
     size_t idx = 0;
+
+    std::vector<uint8_t> read_data;
+    std::vector<uint8_t> work;
+
     for (gguf_load_tensor & tensor : model_loader->tensors_map.tensors) {
-        gguf_buffer read_data;
         read_data.resize(tensor.size);
-        tensor.data = read_data.addr;
+        tensor.data = read_data.data();
         model_loader->load_data_for(tensor);
 
-        printf("[%4zu/%4zu] %36s - %16s, type = %6s, ",
+        LLAMA_LOG_INFO("[%4zu/%4zu] %36s - %16s, type = %6s, ",
                ++idx, model_loader->tensors_map.tensors.size(),
                tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
                ggml_type_name(tensor.type));
@@ -3048,13 +3210,12 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         enum ggml_type new_type;
         void * new_data;
         size_t new_size;
-        gguf_buffer work;
 
         if (!quantize) {
             new_type = tensor.type;
             new_data = tensor.data;
             new_size = tensor.size;
-            printf("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
+            LLAMA_LOG_INFO("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
         } else {
             new_type = quantized_type;
 #ifdef GGML_USE_K_QUANTS
@@ -3089,26 +3250,27 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 int nx = tensor.ne.at(0);
                 int ny = tensor.ne.at(1);
                 if (nx % QK_K != 0 || ny % QK_K != 0) {
-                    fprintf(stderr, "\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
+                    LLAMA_LOG_INFO("\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
                     convert_incompatible_tensor = true;
                 }
             }
             if (convert_incompatible_tensor) {
                 if (tensor.name == "output.weight") {
                     new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
-                    fprintf(stderr, "F16 will be used for this tensor instead.\n");
+                    LLAMA_LOG_WARN("F16 will be used for this tensor instead.\n");
                 } else if (tensor.name == "tok_embeddings.weight") {
                     new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
-                    fprintf(stderr, "Q4_0 will be used for this tensor instead.\n");
+                    LLAMA_LOG_WARN("Q4_0 will be used for this tensor instead.\n");
                 } else {
                     throw std::runtime_error("Unsupported tensor size encountered\n");
                 }
             }
 #endif
 
+            const size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
+
             float * f32_data;
-            size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
-            gguf_buffer f32_conv_buf;
+            std::vector<float> f32_conv_buf;
 
             if (tensor.type == GGML_TYPE_F32) {
                 f32_data = (float *) tensor.data;
@@ -3116,17 +3278,17 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
             } else {
                 llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
-                f32_data = (float *) f32_conv_buf.addr;
+                f32_data = (float *) f32_conv_buf.data();
             }
 
-            printf("quantizing to %s .. ", ggml_type_name(new_type));
+            LLAMA_LOG_INFO("quantizing to %s .. ", ggml_type_name(new_type));
             fflush(stdout);
 
             work.resize(nelements * 4); // upper bound on size
-            new_data = work.addr;
+            new_data = work.data();
             std::vector<int64_t> hist_cur(1 << 4, 0);
 
-            int chunk_size = 32 * 512;
+            const int chunk_size = 32 * 512;
             const int nchunk = (nelements + chunk_size - 1)/chunk_size;
             const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
             if (nthread_use < 2) {
@@ -3134,7 +3296,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             } else {
                 size_t counter = 0;
                 new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size] () {
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements] () {
                     std::vector<int64_t> local_hist;
                     size_t local_size = 0;
                     while (true) {
@@ -3169,7 +3331,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 }
             }
 
-            printf("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
+            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
             int64_t tot_count = 0;
             for (size_t i = 0; i < hist_cur.size(); i++) {
                 hist_all[i] += hist_cur[i];
@@ -3178,18 +3340,18 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
 
             if (tot_count > 0) {
                 for (size_t i = 0; i < hist_cur.size(); i++) {
-                    printf("%5.3f ", hist_cur[i] / float(nelements));
+                    LLAMA_LOG_INFO("%5.3f ", hist_cur[i] / float(nelements));
                 }
             }
-            printf("\n");
+            LLAMA_LOG_INFO("\n");
         }
         total_size_org += tensor.size;
         total_size_new += new_size;
         file_saver.write_tensor(tensor, new_type, new_data, new_size);
     }
 
-    printf("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
-    printf("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
+    LLAMA_LOG_INFO("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
+    LLAMA_LOG_INFO("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
 
     {
         int64_t sum_all = 0;
@@ -3198,17 +3360,15 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         }
 
         if (sum_all > 0) {
-            printf("%s: hist: ", __func__);
+            LLAMA_LOG_INFO("%s: hist: ", __func__);
             for (size_t i = 0; i < hist_all.size(); i++) {
-                printf("%5.3f ", hist_all[i] / float(sum_all));
+                LLAMA_LOG_INFO("%5.3f ", hist_all[i] / float(sum_all));
             }
-            printf("\n");
+            LLAMA_LOG_INFO("\n");
         }
     }
 }
 
-
-
 //
 // interface implementation
 //
@@ -3222,12 +3382,12 @@ struct llama_model * llama_load_model_from_file(
 
     ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
 
-    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gqa, params.rms_norm_eps, params.n_gpu_layers,
-                params.main_gpu, params.tensor_split, params.rope_freq_base, params.rope_freq_scale,params.low_vram,
-                memory_type, params.use_mmap, params.use_mlock, params.vocab_only, params.progress_callback,
-                params.progress_callback_user_data)) {
+    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gpu_layers,
+                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,
+                params.low_vram, memory_type, params.use_mmap, params.use_mlock, params.vocab_only,
+                params.progress_callback, params.progress_callback_user_data)) {
+        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
         delete model;
-        fprintf(stderr, "%s: failed to load model\n", __func__);
         return nullptr;
     }
 
@@ -3260,10 +3420,9 @@ struct llama_context * llama_new_context_with_model(
             unsigned percentage = (unsigned) (100 * progress);
             while (percentage > *cur_percentage_p) {
                 *cur_percentage_p = percentage;
-                fprintf(stderr, ".");
-                fflush(stderr);
+                LLAMA_LOG_INFO(".");
                 if (percentage >= 100) {
-                    fprintf(stderr, "\n");
+                    LLAMA_LOG_INFO("\n");
                 }
             }
         };
@@ -3277,14 +3436,14 @@ struct llama_context * llama_new_context_with_model(
     // reserve memory for context buffers
     if (!params.vocab_only) {
         if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
-            fprintf(stderr, "%s: kv_cache_init() failed for self-attention cache\n", __func__);
+            LLAMA_LOG_ERROR("%s: kv_cache_init() failed for self-attention cache\n", __func__);
             llama_free(ctx);
             return nullptr;
         }
 
         {
             const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
-            fprintf(stderr, "%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
+            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
         }
 
         const auto & hparams = ctx->model.hparams;
@@ -3300,10 +3459,47 @@ struct llama_context * llama_new_context_with_model(
             ctx->embedding.resize(hparams.n_embd);
         }
 
+#ifdef LLAMA_USE_ALLOCATOR
+        {
+            static const size_t tensor_alignment = 32;
+            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
+            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
+
+            // create measure allocator
+            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
+
+            // build worst-case graph
+            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
+            int n_past = hparams.n_ctx - n_tokens;
+            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
+            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
+
+            // measure memory requirements for the graph
+            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
+
+            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
+
+            // debug - for comparison with scratch buffer
+            //size_t prev_req =
+            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
+            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
+            //    MEM_REQ_EVAL().at(ctx->model.type);
+            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
+
+            // recreate allocator with exact memory requirements
+            ggml_allocr_free(ctx->alloc);
+
+            ctx->buf_alloc.resize(alloc_size);
+            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
+        }
+#else
         ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
+#endif
 
+#ifdef LLAMA_USE_SCRATCH
         ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
         ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
+#endif
     }
 
 #ifdef GGML_USE_METAL
@@ -3324,22 +3520,22 @@ struct llama_context * llama_new_context_with_model(
 
         const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
 
-        fprintf(stderr, "%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
+        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
 
-#define LLAMA_METAL_CHECK_BUF(result)                                          \
-    if (!(result)) {                                                           \
-        fprintf(stderr, "%s: failed to add buffer\n", __func__);               \
-        llama_free(ctx);                                                       \
-        return NULL;                                                           \
+#define LLAMA_METAL_CHECK_BUF(result)                            \
+    if (!(result)) {                                             \
+        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
+        llama_free(ctx);                                         \
+        return NULL;                                             \
     }
 
         LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.addr, ctx->buf_compute.size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.addr, ctx->kv_self.buf.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].addr, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].addr, ctx->buf_scratch[1].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
 #undef LLAMA_METAL_CHECK_BUF
     }
 #endif
@@ -3373,9 +3569,6 @@ struct llama_context * llama_init_from_file(
 }
 
 void llama_free(struct llama_context * ctx) {
-    if (ctx->model_owner) {
-        delete &ctx->model;
-    }
     delete ctx;
 }
 
@@ -3387,19 +3580,19 @@ int llama_model_quantize(
         llama_model_quantize_internal(fname_inp, fname_out, params);
         return 0;
     } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to quantize: %s\n", __func__, err.what());
+        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
         return 1;
     }
 }
 
 int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
-    fprintf(stderr, "%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
+    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
 
     const int64_t t_start_lora_us = ggml_time_us();
 
     auto fin = std::ifstream(path_lora, std::ios::binary);
     if (!fin) {
-        fprintf(stderr, "%s: failed to open '%s'\n", __func__, path_lora);
+        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
         return 1;
     }
 
@@ -3411,7 +3604,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
         fin.read((char *) &format_version, sizeof(format_version));
 
         if (format_version != 1) {
-            fprintf(stderr, "%s: unsupported file version\n", __func__ );
+            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
             return 1;
         }
     }
@@ -3422,8 +3615,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
     fin.read((char *) &lora_alpha, sizeof(lora_alpha));
     float scaling = (float)lora_alpha / (float)lora_r;
 
-    fprintf(stderr, "%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
-
+    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
 
     // create a temporary ggml context to store the lora tensors
     // todo: calculate size from biggest possible tensor
@@ -3442,13 +3634,12 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
         model_tensors.insert(kv);
     }
 
-
     // load base model
     std::unique_ptr<llama_model_loader> model_loader;
     ggml_context * base_ctx = NULL;
-    gguf_buffer base_buf;
+    std::vector<uint8_t> base_buf;
     if (path_base_model) {
-        fprintf(stderr, "%s: loading base model from '%s'\n", __func__, path_base_model);
+        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
         model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
 
         size_t ctx_size;
@@ -3457,8 +3648,8 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
         base_buf.resize(ctx_size);
 
         ggml_init_params base_params;
-        base_params.mem_size   = base_buf.size;
-        base_params.mem_buffer = base_buf.addr;
+        base_params.mem_size   = base_buf.size();
+        base_params.mem_buffer = base_buf.data();
         base_params.no_alloc   = model_loader->use_mmap;
 
         base_ctx = ggml_init(base_params);
@@ -3505,17 +3696,17 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
         const std::string lora_suffix = ".lora";
         size_t pos = name.rfind(lora_suffix);
         if (pos == std::string::npos) {
-            fprintf(stderr, "%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
+            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
             return 1;
         }
 
         std::string lora_type = name.substr(pos + lora_suffix.length());
         std::string base_name = name;
         base_name.erase(pos);
-        // fprintf(stderr, "%s: %s => %s (lora type %s) ", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
 
         if (model_tensors.find(base_name) == model_tensors.end()) {
-            fprintf(stderr, "%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
+            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
             return 1;
         }
 
@@ -3526,7 +3717,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
             case 1: wtype = GGML_TYPE_F16;  break;
             default:
                     {
-                        fprintf(stderr, "%s: invalid tensor data type '%d'\n",
+                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
                                 __func__, ftype);
                         return false;
                     }
@@ -3536,7 +3727,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
             lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
         }
         else {
-            fprintf(stderr, "%s: unsupported tensor dimension %d\n", __func__, n_dims);
+            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
             return 1;
         }
         ggml_set_name(lora_tensor, "lora_tensor");
@@ -3574,7 +3765,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
             if (model_loader) {
                 // load from base model
                 if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
-                    fprintf(stderr, "%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
+                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
                     return 1;
                 }
                 size_t idx = model_loader->tensors_map.name_to_idx[base_name];
@@ -3590,8 +3781,8 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
 
             if (ggml_is_quantized(base_t->type)) {
                 if (!warned) {
-                    fprintf(stderr, "%s: warning: using a lora adapter with a quantized model may result in poor quality, "
-                                    "use a f16 or f32 base model with --lora-base\n", __func__);
+                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
+                                   "use a f16 or f32 base model with --lora-base\n", __func__);
                     warned = true;
                 }
             }
@@ -3605,8 +3796,8 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
             ggml_set_name(loraB, "loraB");
 
             if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
-                fprintf(stderr, "%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
-                               " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
+                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
+                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
                 return 1;
             }
 
@@ -3651,7 +3842,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
 
             n_tensors++;
             if (n_tensors % 4 == 0) {
-                fprintf(stderr, ".");
+                LLAMA_LOG_INFO(".");
             }
         }
     }
@@ -3663,7 +3854,7 @@ int llama_apply_lora_from_file_internal(const struct llama_model & model, const
     }
 
     const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
-    fprintf(stderr, " done (%.2f ms)\n", t_lora_us / 1000.0);
+    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
 
     return 0;
 }
@@ -3672,7 +3863,7 @@ int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lor
     try {
         return llama_apply_lora_from_file_internal(ctx->model, path_lora, path_base_model, n_threads);
     } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        LLAMA_LOG_ERROR("%s: failed to apply lora adapter: %s\n", __func__, err.what());
         return 1;
     }
 }
@@ -3681,7 +3872,7 @@ int llama_model_apply_lora_from_file(const struct llama_model * model, const cha
     try {
         return llama_apply_lora_from_file_internal(*model, path_lora, path_base_model, n_threads);
     } catch (const std::exception & err) {
-        fprintf(stderr, "%s: failed to apply lora adapter: %s\n", __func__, err.what());
+        LLAMA_LOG_ERROR("%s: failed to apply lora adapter: %s\n", __func__, err.what());
         return 1;
     }
 }
@@ -3730,10 +3921,60 @@ size_t llama_get_state_size(const struct llama_context * ctx) {
     return s_total;
 }
 
-// Copies the state to the specified destination address
-size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
-    uint8_t * out = dst;
+// llama_context_data
+struct llama_data_context {
+    virtual void write(const void * src, size_t size) = 0;
+    virtual size_t get_size_written() = 0;
+    virtual ~llama_data_context() = default;
+};
+
+struct llama_data_buffer_context : llama_data_context {
+    uint8_t * ptr;
+    size_t size_written = 0;
 
+    llama_data_buffer_context(uint8_t * p) : ptr(p) {}
+
+    void write(const void * src, size_t size) override {
+        memcpy(ptr, src, size);
+        ptr += size;
+        size_written += size;
+    }
+
+    size_t get_size_written() override {
+        return size_written;
+    }
+};
+
+struct llama_data_file_context : llama_data_context {
+    FILE * file;
+    size_t size_written = 0;
+
+    llama_data_file_context(FILE * f) : file(f) {}
+
+    void write(const void * src, size_t size) override {
+        fwrite(src, size, 1, file);
+        size_written += size;
+    }
+
+    size_t get_size_written() override {
+        return size_written;
+    }
+};
+
+/** copy state data into either a buffer or file depending on the passed in context
+ *
+ * file context:
+ * llama_file file("/path", "wb");
+ * llama_data_file_context data_ctx(&file);
+ * llama_copy_state_data(ctx, &data_ctx);
+ *
+ * buffer context:
+ * std::vector<uint8_t> buf(max_size, 0);
+ * llama_data_buffer_context data_ctx(&buf.data());
+ * llama_copy_state_data(ctx, &data_ctx);
+ *
+*/
+void llama_copy_state_data_internal(struct llama_context * ctx, llama_data_context * data_ctx) {
     // copy rng
     {
         std::stringstream rng_ss;
@@ -3745,8 +3986,8 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
         memset(&rng_buf[0], 0, LLAMA_MAX_RNG_STATE);
         memcpy(&rng_buf[0], rng_ss.str().data(), rng_ss.str().size());
 
-        memcpy(out, &rng_size,   sizeof(rng_size));    out += sizeof(rng_size);
-        memcpy(out, &rng_buf[0], LLAMA_MAX_RNG_STATE); out += LLAMA_MAX_RNG_STATE;
+        data_ctx->write(&rng_size,   sizeof(rng_size));
+        data_ctx->write(&rng_buf[0], LLAMA_MAX_RNG_STATE);
     }
 
     // copy logits
@@ -3754,25 +3995,29 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
         const size_t logits_cap  = ctx->logits.capacity();
         const size_t logits_size = ctx->logits.size();
 
-        memcpy(out, &logits_cap,  sizeof(logits_cap));  out += sizeof(logits_cap);
-        memcpy(out, &logits_size, sizeof(logits_size)); out += sizeof(logits_size);
+        data_ctx->write(&logits_cap,  sizeof(logits_cap));
+        data_ctx->write(&logits_size, sizeof(logits_size));
 
         if (logits_size) {
-            memcpy(out, ctx->logits.data(), logits_size * sizeof(float));
+            data_ctx->write(ctx->logits.data(), logits_size * sizeof(float));
         }
 
-        out += logits_cap * sizeof(float);
+        // If there is a gap between the size and the capacity, write padding
+        size_t padding_size = (logits_cap - logits_size) * sizeof(float);
+        if (padding_size > 0) {
+            std::vector<uint8_t> padding(padding_size, 0); // Create a buffer filled with zeros
+            data_ctx->write(padding.data(), padding_size);
+        }
     }
 
     // copy embeddings
     {
         const size_t embedding_size = ctx->embedding.size();
 
-        memcpy(out, &embedding_size, sizeof(embedding_size)); out += sizeof(embedding_size);
+        data_ctx->write(&embedding_size, sizeof(embedding_size));
 
         if (embedding_size) {
-            memcpy(out, ctx->embedding.data(), embedding_size * sizeof(float));
-            out += embedding_size * sizeof(float);
+            data_ctx->write(ctx->embedding.data(), embedding_size * sizeof(float));
         }
     }
 
@@ -3781,14 +4026,14 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
         const auto & kv_self = ctx->kv_self;
         const auto & hparams = ctx->model.hparams;
         const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd;
+        const int    n_embd  = hparams.n_embd_gqa();
         const int    n_ctx   = hparams.n_ctx;
 
         const size_t kv_size = kv_self.buf.size;
         const int    kv_ntok = llama_get_kv_cache_token_count(ctx);
 
-        memcpy(out, &kv_size, sizeof(kv_size)); out += sizeof(kv_size);
-        memcpy(out, &kv_ntok, sizeof(kv_ntok)); out += sizeof(kv_ntok);
+        data_ctx->write(&kv_size, sizeof(kv_size));
+        data_ctx->write(&kv_ntok, sizeof(kv_ntok));
 
         if (kv_size) {
             const size_t elt_size = ggml_element_size(kv_self.k);
@@ -3797,12 +4042,12 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
             ggml_cgraph gf{};
 
             ggml_tensor * kout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
-            kout3d->data = out;
-            out += ggml_nbytes(kout3d);
+            std::vector<uint8_t> kout3d_data(ggml_nbytes(kout3d), 0);
+            kout3d->data = kout3d_data.data();
 
             ggml_tensor * vout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
-            vout3d->data = out;
-            out += ggml_nbytes(vout3d);
+            std::vector<uint8_t> vout3d_data(ggml_nbytes(vout3d), 0);
+            vout3d->data = vout3d_data.data();
 
             ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
                 n_embd, kv_ntok, n_layer,
@@ -3817,15 +4062,20 @@ size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
             ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
 
             ggml_free(cpy_ctx);
+
+            // our data is now in the kout3d_data and vout3d_data buffers
+            // write them to file
+            data_ctx->write(kout3d_data.data(), kout3d_data.size());
+            data_ctx->write(vout3d_data.data(), vout3d_data.size());
         }
     }
+}
 
-    const size_t written  = out - dst;
-    const size_t max_size = llama_get_state_size(ctx);
-
-    GGML_ASSERT(written <= max_size);
+size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
+    llama_data_buffer_context data_ctx(dst);
+    llama_copy_state_data_internal(ctx, &data_ctx);
 
-    return written;
+    return data_ctx.get_size_written();
 }
 
 // Sets the state reading from the specified source address
@@ -3884,7 +4134,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
         const auto & kv_self = ctx->kv_self;
         const auto & hparams = ctx->model.hparams;
         const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd;
+        const int    n_embd  = hparams.n_embd_gqa();
         const int    n_ctx   = hparams.n_ctx;
 
         size_t kv_size;
@@ -3952,7 +4202,7 @@ bool llama_load_session_file(struct llama_context * ctx, const char * path_sessi
     try {
         return llama_load_session_file_internal(ctx, path_session, tokens_out, n_token_capacity, n_token_count_out);
     } catch (const std::exception & err) {
-        fprintf(stderr, "error loading session file: %s\n", err.what());
+        LLAMA_LOG_ERROR("error loading session file: %s\n", err.what());
         return false;
     }
 }
@@ -3975,7 +4225,7 @@ int llama_eval(
                          int   n_past,
                          int   n_threads) {
     if (!llama_eval_internal(*ctx, tokens, nullptr, n_tokens, n_past, n_threads, nullptr)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
+        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
         return 1;
     }
 
@@ -3997,7 +4247,7 @@ int llama_eval_embd(
                              int   n_past,
                              int   n_threads) {
     if (!llama_eval_internal(*ctx, nullptr, embd, n_tokens, n_past, n_threads, nullptr)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
+        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
         return 1;
     }
 
@@ -4018,7 +4268,7 @@ int llama_eval_export(struct llama_context * ctx, const char * fname) {
     const std::vector<llama_token> tmp(n_batch, llama_token_bos());
 
     if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
-        fprintf(stderr, "%s: failed to eval\n", __func__);
+        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
         return 1;
     }
 
@@ -4034,7 +4284,7 @@ int llama_tokenize_with_model(
     auto res = llama_tokenize(model->vocab, text, add_bos);
 
     if (n_max_tokens < (int) res.size()) {
-        fprintf(stderr, "%s: too many tokens\n", __func__);
+        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
         return -((int) res.size());
     }
 
@@ -4151,15 +4401,15 @@ struct llama_timings llama_get_timings(struct llama_context * ctx) {
 void llama_print_timings(struct llama_context * ctx) {
     const llama_timings timings = llama_get_timings(ctx);
 
-    fprintf(stderr, "\n");
-    fprintf(stderr, "%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
-    fprintf(stderr, "%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+    LLAMA_LOG_INFO("\n");
+    LLAMA_LOG_INFO("%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
+    LLAMA_LOG_INFO("%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
             __func__, timings.t_sample_ms, timings.n_sample, timings.t_sample_ms / timings.n_sample, 1e3 / timings.t_sample_ms * timings.n_sample);
-    fprintf(stderr, "%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
+    LLAMA_LOG_INFO("%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
             __func__, timings.t_p_eval_ms, timings.n_p_eval, timings.t_p_eval_ms / timings.n_p_eval, 1e3 / timings.t_p_eval_ms * timings.n_p_eval);
-    fprintf(stderr, "%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
+    LLAMA_LOG_INFO("%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
             __func__, timings.t_eval_ms, timings.n_eval, timings.t_eval_ms / timings.n_eval, 1e3 / timings.t_eval_ms * timings.n_eval);
-    fprintf(stderr, "%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
+    LLAMA_LOG_INFO("%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
 }
 
 void llama_reset_timings(struct llama_context * ctx) {
@@ -4195,3 +4445,44 @@ const char * llama_print_system_info(void) {
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
     return ctx->model.tensors_by_name;
 }
+
+
+void llama_log_set(llama_log_callback log_callback, void * user_data) {
+    g_state.log_callback = log_callback ? log_callback : llama_log_callback_default;
+    g_state.log_callback_user_data = user_data;
+}
+
+#if defined(_MSC_VER) && !defined(vsnprintf)
+#define vsnprintf _vsnprintf
+#endif
+
+static void llama_log_internal_v(llama_log_level level, const char * format, va_list args) {
+    va_list args_copy;
+    va_copy(args_copy, args);
+    char buffer[128];
+    int len = vsnprintf(buffer, 128, format, args);
+    if (len < 128) {
+        g_state.log_callback(level, buffer, g_state.log_callback_user_data);
+    } else {
+        char* buffer2 = new char[len+1];
+        vsnprintf(buffer2, len+1, format, args_copy);
+        buffer2[len] = 0;
+        g_state.log_callback(level, buffer2, g_state.log_callback_user_data);
+        delete[] buffer2;
+    }
+    va_end(args_copy);
+}
+
+static void llama_log_internal(llama_log_level level, const char * format, ...) {
+    va_list args;
+    va_start(args, format);
+    llama_log_internal_v(level, format, args);
+    va_end(args);
+}
+
+static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data) {
+    (void) level;
+    (void) user_data;
+    fputs(text, stderr);
+    fflush(stderr);
+}
diff --git a/gguf-llama.h b/gguf-llama.h
index 540167bd196ba..a8ed69d918048 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -41,10 +41,6 @@
 #define LLAMA_SUPPORTS_GPU_OFFLOAD
 #endif
 
-#ifndef LLAMA_DEFAULT_RMS_EPS
-#define LLAMA_DEFAULT_RMS_EPS 5e-6f
-#endif
-
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -74,12 +70,23 @@ extern "C" {
 
     typedef void (*llama_progress_callback)(float progress, void *ctx);
 
-   struct llama_context_params {
+    enum llama_log_level {
+        LLAMA_LOG_LEVEL_ERROR = 2,
+        LLAMA_LOG_LEVEL_WARN  = 3,
+        LLAMA_LOG_LEVEL_INFO  = 4
+    };
+
+    // Signature for logging events
+    // Note that text includes the new line character at the end for most events.
+    // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
+    // if it exists.
+    // It might not exist for progress report where '.' is output repeatedly.
+    typedef void (*llama_log_callback)(enum llama_log_level level, const char * text, void * user_data);
+
+    struct llama_context_params {
         uint32_t seed;         // RNG seed, -1 for random
         int32_t  n_ctx;        // text context
         int32_t  n_batch;      // prompt processing batch size
-        int32_t  n_gqa;        // grouped-query attention (TEMP - will be moved to model hparams)
-        float    rms_norm_eps; // rms norm epsilon (TEMP - will be moved to model hparams)
         int32_t  n_gpu_layers; // number of layers to store in VRAM
         int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
 
@@ -96,6 +103,7 @@ extern "C" {
 
         // Keep the booleans together to avoid misalignment during copy-by-value.
         bool low_vram;   // if true, reduce VRAM usage at the cost of performance
+        bool mul_mat_q;  // if true, use experimental mul_mat_q kernels
         bool f16_kv;     // use fp16 for KV cache
         bool logits_all; // the llama_eval() call computes all logits, not just the last one
         bool vocab_only; // only load the vocabulary, no weights
@@ -129,7 +137,7 @@ extern "C" {
     // model quantization parameters
     typedef struct llama_model_quantize_params {
         int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
-        enum llama_ftype   ftype;    // quantize to this llama_ftype
+        enum llama_ftype ftype;      // quantize to this llama_ftype
         bool allow_requantize;       // allow quantizing non-f32/f16 tensors
         bool quantize_output_tensor; // quantize output.weight
     } llama_model_quantize_params;
@@ -182,6 +190,10 @@ extern "C" {
         int32_t n_eval;
     };
 
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
+
     LLAMA_API int llama_max_devices();
 
     LLAMA_API struct llama_context_params llama_context_default_params();
diff --git a/gguf-util.h b/gguf-util.h
index 774ae57eef3db..d8557d94f114d 100644
--- a/gguf-util.h
+++ b/gguf-util.h
@@ -64,13 +64,6 @@ static std::string format(const char * fmt, ...) {
     return std::string(buf.data(), size);
 }
 
-template<typename T>
-static std::string to_string(const T & val) {
-    std::stringstream ss;
-    ss << val;
-    return ss.str();
-}
-
 // TODO: can we merge this one and gguf_context?
 struct gguf_file {
     // use FILE * so we don't have to re-open the file to mmap
@@ -474,94 +467,4 @@ struct gguf_mlock {
 #endif
 };
 
-// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
-struct gguf_buffer {
-    uint8_t * addr = NULL;
-    size_t size = 0;
-
-    gguf_buffer() = default;
-
-    void resize(size_t len) {
-#ifdef GGML_USE_METAL
-        free(addr);
-        int result = posix_memalign((void **) &addr, getpagesize(), len);
-        if (result == 0) {
-            memset(addr, 0, len);
-        }
-        else {
-            addr = NULL;
-        }
-#else
-        delete[] addr;
-        addr = new uint8_t[len];
-#endif
-        size = len;
-    }
-
-    ~gguf_buffer() {
-#ifdef GGML_USE_METAL
-        free(addr);
-#else
-        delete[] addr;
-#endif
-        addr = NULL;
-    }
-
-    // disable copy and move
-    gguf_buffer(const gguf_buffer&) = delete;
-    gguf_buffer(gguf_buffer&&) = delete;
-    gguf_buffer& operator=(const gguf_buffer&) = delete;
-    gguf_buffer& operator=(gguf_buffer&&) = delete;
-};
-
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-struct gguf_ctx_buffer {
-    uint8_t * addr = NULL;
-    bool is_cuda;
-    size_t size = 0;
-
-    gguf_ctx_buffer() = default;
-
-    void resize(size_t size) {
-        free();
-
-        addr = (uint8_t *) ggml_cuda_host_malloc(size);
-        if (addr) {
-            is_cuda = true;
-        }
-        else {
-            // fall back to pageable memory
-            addr = new uint8_t[size];
-            is_cuda = false;
-        }
-        this->size = size;
-    }
-
-    void free() {
-        if (addr) {
-            if (is_cuda) {
-                ggml_cuda_host_free(addr);
-            }
-            else {
-                delete[] addr;
-            }
-        }
-        addr = NULL;
-    }
-
-    ~gguf_ctx_buffer() {
-        free();
-    }
-
-    // disable copy and move
-    gguf_ctx_buffer(const gguf_ctx_buffer&) = delete;
-    gguf_ctx_buffer(gguf_ctx_buffer&&) = delete;
-    gguf_ctx_buffer& operator=(const gguf_ctx_buffer&) = delete;
-    gguf_ctx_buffer& operator=(gguf_ctx_buffer&&) = delete;
-};
-#else
-typedef gguf_buffer gguf_ctx_buffer;
-#endif
-
 #endif

From 6f1485488096faa2ddc1aa4f10e6f1879622d61e Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 16:39:02 +0300
Subject: [PATCH 145/242] gitignore : add gptneox-main

---
 .gitignore | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.gitignore b/.gitignore
index 13f728ed0bbf6..107caa228b89e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -50,6 +50,7 @@ models-mnt
 /embd-input-test
 /gguf
 /gguf-llama-simple
+/gptneox-main
 /libllama.so
 build-info.h
 arm_neon.h
@@ -67,7 +68,6 @@ perf-*.txt
 
 examples/jeopardy/results.txt
 
-
 pyproject.toml
 poetry.lock
 poetry.toml

From ec1b100720a66bd0bdcc6fa65d49145c07c25eec Mon Sep 17 00:00:00 2001
From: goerch <jhr.walter@t-online.de>
Date: Mon, 14 Aug 2023 18:30:28 +0200
Subject: [PATCH 146/242] llama : tokenizer fixes (#2549)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies
---
 convert.py                                    | 103 +++---
 examples/common.cpp                           |  11 -
 examples/common.h                             |   7 +-
 examples/embedding/embedding.cpp              |   2 +-
 examples/main/main.cpp                        |  12 +-
 examples/quantize-stats/quantize-stats.cpp    |   1 +
 examples/save-load-state/save-load-state.cpp  |   9 +-
 examples/simple/simple.cpp                    |   4 +-
 .../train-text-from-scratch.cpp               |  20 +-
 llama.cpp                                     | 324 ++++++++++++++++--
 llama.h                                       |  60 +++-
 models/ggml-vocab-aquila.bin                  | Bin 0 -> 1963875 bytes
 models/ggml-vocab-llama.bin                   | Bin 0 -> 466955 bytes
 models/ggml-vocab.bin                         | Bin 432610 -> 0 bytes
 tests/CMakeLists.txt                          |  37 +-
 tests/test-tokenizer-0.cpp                    |  45 ++-
 tests/test-tokenizer-1.cpp                    | 122 +++++++
 17 files changed, 611 insertions(+), 146 deletions(-)
 create mode 100644 models/ggml-vocab-aquila.bin
 create mode 100644 models/ggml-vocab-llama.bin
 delete mode 100644 models/ggml-vocab.bin
 create mode 100644 tests/test-tokenizer-1.cpp

diff --git a/convert.py b/convert.py
index f3bf1798089cc..d9d0c0b38215e 100644
--- a/convert.py
+++ b/convert.py
@@ -238,22 +238,58 @@ def load(model_plus: 'ModelPlus') -> 'Params':
         return params
 
 
-class SentencePieceVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vocabtype: Optional[str]) -> None:
-        self.vocabtype = vocabtype
-        if self.vocabtype == "bpe":
-          self.sentencepiece_tokenizer = json.loads(open(str(fname_tokenizer)).read())
-        else:
-          self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+class BpeVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
         added_tokens: Dict[str, int]
         if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens))
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
         else:
             added_tokens = {}
-        if self.vocabtype == "bpe":
-          vocab_size: int = len(self.sentencepiece_tokenizer)
+        vocab_size: int = len(self.bpe_tokenizer)
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
+
+    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.bpe_tokenizer
+        from transformers.models.gpt2 import tokenization_gpt2
+        byte_encoder = tokenization_gpt2.bytes_to_unicode()
+        byte_decoder = {v: k for k, v in byte_encoder.items()}
+        for i, item in enumerate(tokenizer):
+            text: bytes = item.encode("utf-8")
+            score: float = -i
+            yield text, score
+
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.bpe_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+
+class SentencePieceVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
         else:
-          vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
+            added_tokens = {}
+        vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
         expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
         actual_ids = sorted(added_tokens.values())
         if expected_ids != actual_ids:
@@ -267,32 +303,11 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vo
 
     def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
         tokenizer = self.sentencepiece_tokenizer
-        if self.vocabtype == "bpe":
-          from transformers.models.gpt2 import tokenization_gpt2
-          byte_encoder = tokenization_gpt2.bytes_to_unicode()
-          byte_decoder = {v: k for k, v in byte_encoder.items()}
-          for i, item in enumerate(tokenizer):
-            text: bytes
-            text = b''.join([x.to_bytes(1, byteorder='big') for x in [byte_decoder[y] for y in item]])
-            score: float = -i
+        for i in range(tokenizer.vocab_size()):
+            piece = tokenizer.id_to_piece(i)
+            text: bytes = piece.encode("utf-8")
+            score: float = tokenizer.get_score(i)
             yield text, score
-        else:
-          for i in range(tokenizer.vocab_size()):
-              text: bytes
-              if tokenizer.is_unknown(i):
-                  text = " \u2047 ".encode("utf-8")
-              elif tokenizer.is_control(i):
-                  text = b""
-              elif tokenizer.is_byte(i):
-                  piece = tokenizer.id_to_piece(i)
-                  if len(piece) != 6:
-                      raise Exception(f"Invalid token: {piece}")
-                  byte_value = int(piece[3:-1], 16)
-                  text = struct.pack("B", byte_value)
-              else:
-                  text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
-              score: float = tokenizer.get_score(i)
-              yield text, score
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
@@ -319,7 +334,7 @@ def __repr__(self) -> str:
         return f"<GGMLVocab with {self.vocab_size} tokens>"
 
 
-Vocab = Union[SentencePieceVocab, GGMLVocab]
+Vocab = Union[BpeVocab, SentencePieceVocab, GGMLVocab]
 
 
 def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
@@ -1044,7 +1059,7 @@ def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], conc
 def check_vocab_size(params: Params, vocab: Vocab) -> None:
     if params.n_vocab != vocab.vocab_size:
         # GGMLVocab comes from the same file as the model so shouldn't mismatch:
-        assert isinstance(vocab, SentencePieceVocab)
+        assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
         if params.n_vocab == vocab.vocab_size_base:
             print("Ignoring added_tokens.json since model matches vocab size without it.")
             vocab.added_tokens_list = []
@@ -1093,7 +1108,7 @@ def write_vocab(self, vocab: Vocab) -> None:
     @staticmethod
     def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
         of = OutputFile(fname_out)
-        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0)
+        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0, n_kv_head=None)
         of = OutputFile(fname_out)
         of.write_file_header(params, file_type=GGMLFileType.AllF32)
         of.write_vocab(vocab)
@@ -1228,7 +1243,7 @@ def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
     return {name: model[name] for name in TENSORS_LIST if name in model}
 
 
-def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
+def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
     print(f"vocabtype: {vocabtype}")
     # Be extra-friendly and accept either a file or a directory.  Also, if it's
     # a directory, it might be the model directory, and tokenizer.model might
@@ -1250,8 +1265,12 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
                 "if it's in another directory, pass the directory as --vocab-dir")
     added_tokens_path = path.parent / "added_tokens.json"
     print(f"Loading vocab file {path}")
-    return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None,
-                              vocabtype)
+    if vocabtype == "bpe":
+        return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    elif vocabtype == "spm":
+        return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    else:
+        raise ValueError(f"Unsupported vocabulary type {vocabtype}")
 
 
 def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
diff --git a/examples/common.cpp b/examples/common.cpp
index 9f8aab9a25681..18de8ef43d403 100644
--- a/examples/common.cpp
+++ b/examples/common.cpp
@@ -633,17 +633,6 @@ std::string gpt_random_prompt(std::mt19937 & rng) {
     return "The";
 }
 
-// TODO: not great allocating this every time
-std::vector<llama_token> llama_tokenize(struct llama_context * ctx, const std::string & text, bool add_bos) {
-    // initialize to prompt numer of chars, since n_tokens <= n_prompt_chars
-    std::vector<llama_token> res(text.size() + (int) add_bos);
-    const int n = llama_tokenize(ctx, text.c_str(), res.data(), res.size(), add_bos);
-    assert(n >= 0);
-    res.resize(n);
-
-    return res;
-}
-
 struct llama_context_params llama_context_params_from_gpt_params(const gpt_params & params) {
     auto lparams = llama_context_default_params();
 
diff --git a/examples/common.h b/examples/common.h
index 375bc0a3db416..0e520ed4e7775 100644
--- a/examples/common.h
+++ b/examples/common.h
@@ -2,6 +2,7 @@
 
 #pragma once
 
+#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include <string>
@@ -100,12 +101,6 @@ void gpt_print_usage(int argc, char ** argv, const gpt_params & params);
 
 std::string gpt_random_prompt(std::mt19937 & rng);
 
-//
-// Vocab utils
-//
-
-std::vector<llama_token> llama_tokenize(struct llama_context * ctx, const std::string & text, bool add_bos);
-
 //
 // Model utils
 //
diff --git a/examples/embedding/embedding.cpp b/examples/embedding/embedding.cpp
index 5192d6df5c2f8..8788571cbf9d4 100644
--- a/examples/embedding/embedding.cpp
+++ b/examples/embedding/embedding.cpp
@@ -67,7 +67,7 @@ int main(int argc, char ** argv) {
         fprintf(stderr, "%s: prompt: '%s'\n", __func__, params.prompt.c_str());
         fprintf(stderr, "%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
         for (int i = 0; i < (int) embd_inp.size(); i++) {
-            fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], llama_token_to_str(ctx, embd_inp[i]));
+            fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], llama_token_to_str(ctx, embd_inp[i]).c_str());
         }
         fprintf(stderr, "\n");
     }
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index a632bea1cf2b9..2424435607b0f 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -191,10 +191,6 @@ int main(int argc, char ** argv) {
 
     // tokenize the prompt
     std::vector<llama_token> embd_inp;
-
-    // Add a space in front of the first character to match OG llama tokenizer behavior
-    params.prompt.insert(0, 1, ' ');
-
     if (params.interactive_first || params.instruct || !params.prompt.empty() || session_tokens.empty()) {
         embd_inp = ::llama_tokenize(ctx, params.prompt, true);
     } else {
@@ -278,7 +274,7 @@ int main(int argc, char ** argv) {
         fprintf(stderr, "%s: prompt: '%s'\n", __func__, params.prompt.c_str());
         fprintf(stderr, "%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
         for (int i = 0; i < (int) embd_inp.size(); i++) {
-            fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], llama_token_to_str(ctx, embd_inp[i]));
+            fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], llama_token_to_str(ctx, embd_inp[i]).c_str());
         }
 
         if (ctx_guidance) {
@@ -286,14 +282,14 @@ int main(int argc, char ** argv) {
             fprintf(stderr, "%s: negative prompt: '%s'\n", __func__, params.cfg_negative_prompt.c_str());
             fprintf(stderr, "%s: number of tokens in negative prompt = %zu\n", __func__, guidance_inp.size());
             for (int i = 0; i < (int) guidance_inp.size(); i++) {
-                fprintf(stderr, "%6d -> '%s'\n", guidance_inp[i], llama_token_to_str(ctx, guidance_inp[i]));
+                fprintf(stderr, "%6d -> '%s'\n", guidance_inp[i], llama_token_to_str(ctx, guidance_inp[i]).c_str());
             }
         }
 
         if (params.n_keep > 0) {
         fprintf(stderr, "%s: static prompt based on n_keep: '", __func__);
             for (int i = 0; i < params.n_keep; i++) {
-                fprintf(stderr, "%s", llama_token_to_str(ctx, embd_inp[i]));
+                fprintf(stderr, "%s", llama_token_to_str(ctx, embd_inp[i]).c_str());
             }
             fprintf(stderr, "'\n");
         }
@@ -662,7 +658,7 @@ int main(int argc, char ** argv) {
         // display text
         if (input_echo) {
             for (auto id : embd) {
-                printf("%s", llama_token_to_str(ctx, id));
+                printf("%s", llama_token_to_str(ctx, id).c_str());
             }
             fflush(stdout);
         }
diff --git a/examples/quantize-stats/quantize-stats.cpp b/examples/quantize-stats/quantize-stats.cpp
index 6aa06ec8fa115..a330b20df2963 100644
--- a/examples/quantize-stats/quantize-stats.cpp
+++ b/examples/quantize-stats/quantize-stats.cpp
@@ -1,6 +1,7 @@
 #include "ggml.h"
 #include "build-info.h"
 
+#define LLAMA_API_CPP // TODO: eliminate me
 #define LLAMA_API_INTERNAL
 #include "llama.h"
 
diff --git a/examples/save-load-state/save-load-state.cpp b/examples/save-load-state/save-load-state.cpp
index 61c71c3589fdf..d5a81978e97af 100644
--- a/examples/save-load-state/save-load-state.cpp
+++ b/examples/save-load-state/save-load-state.cpp
@@ -45,9 +45,8 @@ int main(int argc, char ** argv) {
         llama_free_model(model);
         return 1;
     }
-    auto tokens = std::vector<llama_token>(params.n_ctx);
-    auto n_prompt_tokens = llama_tokenize(ctx, params.prompt.c_str(), tokens.data(), int(tokens.size()), true);
-
+    auto tokens = llama_tokenize(ctx, params.prompt.c_str(), true);
+    auto n_prompt_tokens = tokens.size();
     if (n_prompt_tokens < 1) {
         fprintf(stderr, "%s : failed to tokenize prompt\n", __func__);
         llama_free(ctx);
@@ -92,7 +91,7 @@ int main(int argc, char ** argv) {
         auto next_token_str = llama_token_to_str(ctx, next_token);
         last_n_tokens_data.push_back(next_token);
 
-        printf("%s", next_token_str);
+        printf("%s", next_token_str.c_str());
         if (llama_eval(ctx, &next_token, 1, n_past, params.n_threads)) {
             fprintf(stderr, "\n%s : failed to evaluate\n", __func__);
             llama_free(ctx);
@@ -152,7 +151,7 @@ int main(int argc, char ** argv) {
         auto next_token_str = llama_token_to_str(ctx2, next_token);
         last_n_tokens_data.push_back(next_token);
 
-        printf("%s", next_token_str);
+        printf("%s", next_token_str.c_str());
         if (llama_eval(ctx2, &next_token, 1, n_past, params.n_threads)) {
             fprintf(stderr, "\n%s : failed to evaluate\n", __func__);
             llama_free(ctx2);
diff --git a/examples/simple/simple.cpp b/examples/simple/simple.cpp
index 55cce10442d58..0dfcd7a448826 100644
--- a/examples/simple/simple.cpp
+++ b/examples/simple/simple.cpp
@@ -62,7 +62,7 @@ int main(int argc, char ** argv) {
     fprintf(stderr, "\n\n");
 
     for (auto id : tokens_list) {
-        fprintf(stderr, "%s", llama_token_to_str(ctx, id));
+        fprintf(stderr, "%s", llama_token_to_str(ctx, id).c_str());
     }
 
     fflush(stderr);
@@ -109,7 +109,7 @@ int main(int argc, char ** argv) {
         }
 
         // print the new token :
-        printf("%s", llama_token_to_str(ctx, new_token_id));
+        printf("%s", llama_token_to_str(ctx, new_token_id).c_str());
         fflush(stdout);
 
         // push this new token for next evaluation
diff --git a/examples/train-text-from-scratch/train-text-from-scratch.cpp b/examples/train-text-from-scratch/train-text-from-scratch.cpp
index 54dc2beed0080..d446c6ea5cfb2 100644
--- a/examples/train-text-from-scratch/train-text-from-scratch.cpp
+++ b/examples/train-text-from-scratch/train-text-from-scratch.cpp
@@ -1,4 +1,5 @@
 #include "ggml.h"
+#include "common.h"
 #include "llama.h"
 #include <unordered_map>
 #include <vector>
@@ -1961,7 +1962,7 @@ void print_matrix(struct ggml_tensor * probs) {
 
 
 void print_token(struct llama_context * ctx, llama_token token) {
-    printf("%s", llama_token_to_str(ctx, token));
+    printf("%s", llama_token_to_str(ctx, token).c_str());
 }
 
 void print_tokens(struct llama_context* ctx, struct ggml_tensor * tokens) {
@@ -2188,11 +2189,10 @@ int tokenize_file(struct llama_context * lctx, const char * filename, std::vecto
     f.read_raw(buf.data(), f.size);
     buf[f.size] = '\0';
 
-    out.resize(buf.size());
-
-    int n_tokens = llama_tokenize(lctx, buf.data(), out.data(), buf.size(), false);
-    if (n_tokens >= 0) {
-        out.resize(n_tokens);
+    int n_tokens = llama_tokenize(lctx, buf.data(), out.data(), out.size(), false);
+    if (n_tokens < 0) {
+        out.resize(-n_tokens);
+        llama_tokenize(lctx, buf.data(), out.data(), out.size(), false);
     }
 
     bool verify = false;
@@ -2200,17 +2200,17 @@ int tokenize_file(struct llama_context * lctx, const char * filename, std::vecto
         const char * in  = buf.data();
         const char * end = buf.data() + buf.size();
         for (int i = 0; i < (int) out.size(); ++i) {
-            const char * s = llama_token_to_str(lctx, out[i]);
-            int len = strlen(s);
+            std::string s = llama_token_to_str(lctx, out[i]);
+            int len = s.length();
             if (in >= end) {
                 printf("%s: unexpected end of original text.\n", __func__);
                 break;
             }
-            const bool matches = (strncmp(in, s, len) == 0);
+            const bool matches = (strncmp(in, s.c_str(), len) == 0);
             if (matches) {
                 in += len;
             } else {
-                printf("%s: mismatch: expected '%s', but got '%s'\n", __func__, std::string(in, len).c_str(), s);
+                printf("%s: mismatch: expected '%s', but got '%s'\n", __func__, std::string(in, len).c_str(), s.c_str());
             }
         }
     }
diff --git a/llama.cpp b/llama.cpp
index c8ab313d9c116..5504f14834c35 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -7,6 +7,7 @@
 #endif
 
 #include "llama-util.h"
+#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include "ggml.h"
@@ -575,6 +576,7 @@ struct llama_file_loader {
             float score = 0.0f;
             file.read_raw(&score, sizeof(score));
 
+            GGML_ASSERT(vocab.token_to_id.find(word) == vocab.token_to_id.end());
             vocab.token_to_id[word] = i;
 
             auto & tok_score = vocab.id_to_token[i];
@@ -1060,6 +1062,11 @@ static void llama_model_load_internal(
     std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
 
     vocab = std::move(ml->file_loader->vocab);
+
+    if (vocab_only) {
+        return;
+    }
+
     model.hparams = ml->file_loader->hparams;
     model.n_gpu_layers = n_gpu_layers;
     llama_file_version file_version = ml->file_loader->file_version;
@@ -1141,10 +1148,6 @@ static void llama_model_load_internal(
         }
     }
 
-    if (vocab_only) {
-        return;
-    }
-
     auto & ctx = model.ctx;
 
     size_t ctx_size;
@@ -1940,6 +1943,105 @@ static bool llama_eval_internal(
 // tokenizer
 //
 
+static std::string llama_vocab_type(const llama_vocab& vocab) {
+    return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
+}
+
+static bool llama_is_normal_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token >= 259;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return token >= 95;
+    else
+        return false;
+}
+
+static bool llama_is_unknown_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 0;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_control_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 1 || token == 2;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_bos_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 1;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_eos_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 2;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_user_defined_token(const llama_vocab& vocab, llama_token token) {
+    // TODO: improve?
+    return false;
+}
+
+static bool llama_is_unused_token(const llama_vocab& vocab, llama_token token) {
+    // TODO: improve?
+    return false;
+}
+
+static bool llama_is_byte_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return 3 <= token && token < 259;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return 1 <= token && token < 95;
+    else
+        return false;
+}
+
+static uint8_t llama_byte_to_char(const llama_vocab& vocab, uint8_t byte) {
+    if(llama_vocab_type(vocab) == "spm")
+        return byte + 3;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return byte + 32;
+    else
+        return false;
+}
+
+static std::string llama_escape_whitespace(const std::string& text) {
+    std::string result;
+    bool escaping = false;
+    result += "\xe2\x96\x81";
+    for (size_t offs = 0; offs < text.length(); ++offs) {
+        if (text[offs] == ' ') {
+            if (!escaping) {
+                result += "\xe2\x96\x81";
+                escaping = true;
+            }
+        }
+        else {
+            escaping = false;
+            result += text[offs];
+        }
+    }
+    return result;
+}
+
+static std::string llama_unescape_whitespace(const std::string& word) {
+    if (word.length() >= 3 && word.substr(0, 3) == "\xe2\x96\x81") {
+        return std::string(" ") + word.substr(3);
+    }
+    return word;
+}
+
 static size_t utf8_len(char src) {
     const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
     uint8_t highbits = static_cast<uint8_t>(src) >> 4;
@@ -1981,10 +2083,11 @@ struct llama_tokenizer {
         size_t offs = 0;
         while (offs < text.size()) {
             llama_sp_symbol sym;
-            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
+            size_t len = utf8_len(text[offs]);
+            GGML_ASSERT(offs + len <= text.size());
             sym.text = text.c_str() + offs;
-            sym.n = char_len;
-            offs += char_len;
+            sym.n = len;
+            offs += len;
             sym.prev = index - 1;
             sym.next = offs == text.size() ? -1 : index + 1;
             index++;
@@ -2029,23 +2132,36 @@ struct llama_tokenizer {
 
         for (int i = 0; i != -1; i = symbols_[i].next) {
             auto & symbol = symbols_[i];
-            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
-
-            if (token == vocab_.token_to_id.end()) {
-                // output any symbols that did not form tokens as bytes.
-                for (int j = 0; j < (int) symbol.n; ++j) {
-                    // NOTE: old version, before #2420 - not sure what are the implications of this
-                    //llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
-                    llama_vocab::id token_id = vocab_.token_to_id.at(std::string(1, symbol.text[j]));
-                    output.push_back(token_id);
-                }
-            } else {
-                output.push_back((*token).second);
-            }
+            resegment(symbol, output);
         }
     }
 
 private:
+    void resegment(llama_sp_symbol &symbol, std::vector<llama_vocab::id> &output) {
+        auto text = std::string(symbol.text, symbol.n);
+        auto token = vocab_.token_to_id.find(text);
+
+        // Do we need to support is_unused?
+        if (token != vocab_.token_to_id.end()) {
+            output.push_back((*token).second);
+            return;
+        }
+
+        const auto p = rev_merge.find(text);
+
+        if (p == rev_merge.end()) {
+            // output any symbols that did not form tokens as bytes.
+            for (int j = 0; j < (int)symbol.n; ++j) {
+                llama_vocab::id token_id = llama_byte_to_char(vocab_, symbol.text[j]);
+                output.push_back(token_id);
+            }
+            return;
+        }
+
+        resegment(symbols_[p->second.first], output);
+        resegment(symbols_[p->second.second], output);
+    }
+
     void try_add_bigram(int left, int right) {
         if (left == -1 || right == -1) {
             return;
@@ -2070,18 +2186,22 @@ struct llama_tokenizer {
         bigram.score = tok_score.score;
         bigram.size = text.size();
         work_queue_.push(bigram);
+
+        // Do we need to support is_unused?
+        rev_merge[text] = std::make_pair(left, right);
     }
 
     const llama_vocab & vocab_;
     std::vector<llama_sp_symbol> symbols_;
     llama_sp_bigram::queue work_queue_;
+    std::map<std::string, std::pair<int, int> > rev_merge;
 };
 
-static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
+static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & raw_text, bool bos, bool escape) {
     llama_tokenizer tokenizer(vocab);
     std::vector<llama_vocab::id> output;
 
-    if (text.empty()) {
+    if (raw_text.empty()) {
         return output;
     }
 
@@ -2089,6 +2209,13 @@ static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, co
         output.push_back(llama_token_bos());
     }
 
+    std::string text;
+    if (escape) {
+        text = llama_escape_whitespace(raw_text);
+    } else {
+        text = raw_text;
+    }
+
     tokenizer.tokenize(text, output);
     return output;
 }
@@ -2670,15 +2797,15 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
 
     for (size_t i = 0; i < candidates->size; ++i) {
         const llama_token id  = candidates->data[i].id;
-        const char *      str = llama_token_to_str(ctx, id);
+        std::string       str = llama_token_to_str(ctx, id);
         if (id == eos) {
             if (!allow_eos) {
                 candidates->data[i].logit = -INFINITY;
             }
-        } else if (*str == 0) {
+        } else if (str.empty()) {
             candidates->data[i].logit = -INFINITY;
         } else {
-            candidates_decoded.push_back(decode_utf8(str));
+            candidates_decoded.push_back(decode_utf8(str.c_str()));
             candidates_grammar.push_back({ i, candidates_decoded.back().data() });
         }
     }
@@ -2879,9 +3006,9 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
         LLAMA_ASSERT(false);
     }
 
-    const char * str = llama_token_to_str(ctx, token);
+    std::string str = llama_token_to_str(ctx, token);
     // Note terminating 0 in decoded string
-    auto code_points = decode_utf8(str);
+    auto code_points = decode_utf8(str.c_str());
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
     }
@@ -4132,7 +4259,8 @@ int llama_tokenize_with_model(
                  llama_token * tokens,
                          int   n_max_tokens,
                         bool   add_bos) {
-    auto res = llama_tokenize(model->vocab, text, add_bos);
+    auto escape = llama_vocab_type(model->vocab) == "spm";
+    auto res = llama_tokenize(model->vocab, text, add_bos, escape);
 
     if (n_max_tokens < (int) res.size()) {
         LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
@@ -4155,6 +4283,62 @@ int llama_tokenize(
     return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
 }
 
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int length = text.length() + add_bos;
+    std::vector<llama_token> result(length);
+    length = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
+}
+
+int llama_tokenize_bpe(
+        struct llama_context * ctx,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    auto res = llama_tokenize(ctx->model.vocab, text, add_bos, false);
+
+    if (n_max_tokens < (int) res.size()) {
+        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
+        return -((int) res.size());
+    }
+
+    for (size_t i = 0; i < res.size(); i++) {
+        tokens[i] = res[i];
+    }
+
+    return res.size();
+}
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int length = text.length() + add_bos;
+    std::vector<llama_token> result(length);
+    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
+}
+
 int llama_n_vocab_from_model(const struct llama_model * model) {
     return model->vocab.id_to_token.size();
 }
@@ -4208,16 +4392,88 @@ float * llama_get_embeddings(struct llama_context * ctx) {
     return ctx->embedding.data();
 }
 
-const char * llama_token_to_str_with_model(const struct llama_model * model, llama_token token) {
-    if (token >= llama_n_vocab_from_model(model)) {
-        return nullptr;
+int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * str, int length) {
+    if (0 <= token && token < llama_n_vocab_from_model(model)) {
+        if (llama_is_normal_token(model->vocab, token)) {
+            std::string result = model->vocab.id_to_token[token].tok;
+            if(llama_vocab_type(model->vocab) == "spm") {
+                result = llama_unescape_whitespace(result);
+            }
+            if(result.length() > length) {
+                return - result.length();
+            }
+            strcpy(str, result.c_str());
+            return result.length();
+        } else if (llama_is_unknown_token(model->vocab, token)) {
+            if(3 > length) {
+                return -3;
+            }
+            strcpy(str, "\xe2\x96\x85");
+            return 3;
+        } else if (llama_is_control_token(model->vocab, token)) {
+            ;
+        } else if (llama_is_byte_token(model->vocab, token)) {
+            if(1 > length) {
+                return -1;
+            }
+            str[0] = llama_byte_to_char(model->vocab, token);
+            str[1] = 0x00;
+            return 1;
+        }
+    }
+    return 0;
+}
+
+int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * str, int length) {
+    return llama_token_to_str_with_model(&ctx->model, token, str, length);
+}
+
+std::string llama_token_to_str(
+        const struct llama_context * ctx,
+                       llama_token   token) {
+    std::string result;
+    int length = 8;
+    result.resize(length);
+    length = llama_token_to_str(ctx, token, (char *)result.data(), result.length());
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_token_to_str(ctx, token, (char *)result.data(), result.length());
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
     }
+    return result;
+}
 
-    return model->vocab.id_to_token[token].tok.c_str();
+int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * str, int length) {
+    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
+        std::string result = ctx->model.vocab.id_to_token[token].tok;
+        if (result.length() > length) {
+            return - result.length();
+        }
+        strcpy(str, result.c_str());
+        return result.length();
+    }
+    return 0;
 }
 
-const char * llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    return llama_token_to_str_with_model(&ctx->model, token);
+std::string llama_token_to_str_bpe(
+    const struct llama_context * ctx,
+                   llama_token   token) {
+    std::string result;
+    int length = 8;
+    result.resize(length);
+    length = llama_token_to_str_bpe(ctx, token, (char*)result.data(), result.length());
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_token_to_str_bpe(ctx, token, (char*)result.data(), result.length());
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
 }
 
 llama_token llama_token_bos() {
diff --git a/llama.h b/llama.h
index 92b474891493e..6ed1c4b916c68 100644
--- a/llama.h
+++ b/llama.h
@@ -336,6 +336,13 @@ extern "C" {
                              int   n_max_tokens,
                             bool   add_bos);
 
+    LLAMA_API int llama_tokenize_bpe(
+            struct llama_context * ctx,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
     LLAMA_API int llama_tokenize_with_model(
         const struct llama_model * model,
                       const char * text,
@@ -377,14 +384,23 @@ extern "C" {
     LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
 
     // Token Id -> String. Uses the vocabulary in the provided context
-    LLAMA_API const char * llama_token_to_str(
+    LLAMA_API int llama_token_to_str(
             const struct llama_context * ctx,
-                           llama_token   token);
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
 
-    LLAMA_API const char * llama_token_to_str_with_model(
-              const struct llama_model * model,
-                           llama_token   token);
+    LLAMA_API int llama_token_to_str_bpe(
+            const struct llama_context * ctx,
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
 
+    LLAMA_API int llama_token_to_str_with_model(
+              const struct llama_model * model,
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
     // Special tokens
     LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
     LLAMA_API llama_token llama_token_eos();  // end-of-sentence
@@ -472,15 +488,43 @@ extern "C" {
 }
 #endif
 
-// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
-#ifdef LLAMA_API_INTERNAL
+// C++ API, will be moving to common.h soon (TM)
+#ifdef LLAMA_API_CPP
 
 #include <vector>
 #include <string>
+
+//
+// Vocab utils
+//
+
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::string llama_token_to_str(
+        const struct llama_context * ctx,
+                       llama_token   token);
+
+std::string llama_token_to_str_bpe(
+    const struct llama_context * ctx,
+                   llama_token   token);
+
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
 struct ggml_tensor;
 
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
 
-#endif
+#endif // LLAMA_API_CPP
+
+#endif // LLAMA_API_INTERNAL
 
 #endif // LLAMA_H
diff --git a/models/ggml-vocab-aquila.bin b/models/ggml-vocab-aquila.bin
new file mode 100644
index 0000000000000000000000000000000000000000..e06b39b5a31c1ced47e4891fc892bc956e97bcb5
GIT binary patch
literal 1963875
zcmc${d7Pa^mH2%G1X0m;$NkD4WVwW}gFNgY+ZDnh>%HmjBn_SPolZhPWaa_fN1b*V
zmvMVY(n}yDHwh#_2zl(>cA3$cc3)?_qmDW{YLDBD>%8By)DwJv^T+S~>kS{0Q%_Z$
zI(6#QsZ*y;)m?eZk|j?Ok*T-n-~MM0<mm(W=kzxYE*}_KymIjNm2Z5Fe%3;-63KTd
za1j5O3LGqw8E^<$1{^9<7;qRx3^-h*XTT9u_@vgFC(>uYks>t%UM*5L;3yCpaI{Fn
zfMZ0O1{^EWHlR*h1{^2yHUs91j2Upe$cGGA02v0HKwk|wQKVzQNg|IJa55AcaEi!h
z3^-L})_~JQzHY$j5O2U4B0n+!Zp;~QrpSLAkkO@S?blf%`xtPx$kPlsN936XoJ*Gt
zI8WsH2AnT)paB<%yv%?L;f4Vhi5z0UYekMQ;9`-Z47dbF8E~n{0s}4+IoW{A88rhM
z@X&zQiJWV|6(ScJaHYs41}ucV23#ew(15E&t~KBqksA!S7S0>+dXWJGt`k{m!1W@-
z2He1)7;vM=N&{{Zx!r)9Mc!yYlkqa3C324ei$vaJK)=Xa3>aW24Hy)8hXIR4-f6%R
zk#`%glu<TdNaVc++#>RR18x=hfC0k@fC0-y{?35qA|EwiMC9WJw2=)1{z7Dp0iz=8
z3|JvDVZchn#(-5K_Zx7V$OZ#ei)=LDb|lJxJ4Ch^@CK2G47gKdn*nb`*bKN!<S_&O
zQsnXJy;NlG6M4d+HAtdEYx(ZbI+3nJ<07AOXaYfXXj0_w9hwsPoI~qH{?VcPkz<D*
z5c#4*8$`bB&_<E3I<yHfcc>%sO@}s%eA}TdBLCvhR%U@i4~cx&p@&7j@6a}pA2_s~
zN#f8WB0qNMQIY>}=rNI>Ikba0<j~_Hzi?=$$gdoFLgY6N?P7`<v^y91okMqv{K27n
zME>N^y|8n)@^tr`&{Ph+S!6GV-XgNMLvLk*I`lS?r#kd@k$oL{hsb^o{WbH~p?8Yx
z@6fwMp6<}QMV{f%-!Q!$8WTCdq4$V9%c1v*Jlmo7G5Z~QzsPeO`dg9bI`jdN=Q;F2
zl!HSb5_y3`9~OC`Lw_goB8NVL?r`X%A}@C6V<Inc=;I<Ub?6hQ7KaLvmpg>iy}}`+
z?v)N9bv1{Ox>q@b)E(pyQg^UJNZlb0A$5m3gw!485K?!zLrC2b4k2~(9C}dXNQaQW
zS389K9pw=6ceF#u-!Tp$f5$q6{M8*o{*H49`J3+$@^`#L$ln5okiQcgLjF#42>Cn7
zA>{95hmgNh976t1bqM)8%^~FPbc4FN$QceHf3I-}`8(4g<S%mw`8&%Y<nL^UkiT;r
zLjKNm2>CnDA>{9ThmgMu976ssbO`yo$RXtKwGJVF7dwRfUE&b(cd0|j-(?OVf0sLi
z{52dx{$A%0@^^(p$lsL?A%6=ULjJCD2>H9(A>{8GhmgN(9YX$I?-25LokPgq^$sC_
zH#mg+-RKbVcauZN-^~soe@%ywzm`MD-y(;QzkY|1zX6Akzd?tPzr_wAe@h%f{+2p~
z{0%vT{N3UZ@^`C4$ltI-$lo%DkiX>)A%7zdA%AU$kiWlh2>Baz2>Dy#5c0RuA>?nB
zL&)E44k3T59YX$YcL@2r!y)AF4GtlHcRGapz0sf<<nJzrkiWll2>JUfhmgN~_dddk
zQFIH*9p1fe8Y=GIy|;Z|Df!)2WcQ|N8og)tzI;}=yHWCcqEJiZUc<O_P;!Tcma|rS
z#kJivk_SB51<9Ko32=8p@;6(}(Lt7UgGQE1{uZyfZkoQob@x-95=H*j-A@a0jbwS7
zRT&(W+@aP=k+%m`D<prrXG6<7;^T<q@9^d!tBctDwKp}<eM0hp$GZ&XJEOvC$={i%
zf&cG{&;63W%Ri5I5wCY!Lqjwf5On;FcL&-W8XUk7ae^o}7PY6L;XU@XMaSQhz9MYz
z^{>N{zc(6fi@Yy2dcWiWPfau4?~fvjC4avcSvQSn{B3+*B>CUk=WanW@&Wriw0w!=
zA4muh+YdTC+1)02z!P1D;6qV$1neJjszyaV9A6pg{KNKjWEIxa9`3G{{O=s&INX(w
z1T6!SCvX`?Rw}VU@_^SMF!C`cQKjKyUSx6_{(syW8tRw)<My>%tQ-;fgqN7;!j(^0
ziDBfaFzD!`C3mQ`Lgc=Xrq*)ExX-Jtn`W}Dad>KaljLi>4@?B5WUc3n&)h3{z!P1j
z$T}|thqTT*Yitx^#&`EUNk&ypeR_=to@tcEL~3oL<P*V{r6QA2q`On{fG4^TH)Y@H
z>r%-bW|lIO)?3xV<;>Fc2|LB_H(L7ndjFoDptK%H6hF~@K=Md2Hq)rZ4XOTi$pfC~
zB3KU^u0e1GG>jb^<S=RTjX^z9wJA_^VMl^hmvjup3Q#yS!mw}lvJ*<k=0xOUI^VZ=
z&id|p$+rY@+T5BdKwrS)%7Tad`;;n~hblx}di1b0JH-5XIM5&y+r0QVE#(yqYqkd(
ztm{1BaYgb-Dvx?dFqFxo_PsmKB+DQ5hM3-udFk~^Ou*wPO4*SJdqjn3hnJd|K@&gj
z@T87-z*EyGl%0li1fDrG1kaxEvQyI!NgnWcmz8Cgx6fqW<?SG<)2x1FbeoLQirfYR
zJ{gU7nQx!8udC2i)9EXOPglh1aO@70-G@};M+yxw!-CR{1}>63feS@u9M|~txa1B`
zPjp$TKV{#$g=)!9dG{FCPdol8E0cfPzPIS?XQJKdb+9A<jN_c_GIu^3tsWuyXYKRR
z{9#|?MW&`7mi+Gx&F~!}|KJGwC3k2QI^c7GOc8(1^g(Oo5Q;`V?-}FMn<f9eQ@EPV
zk$vqJi;(tzbaV{$KN++N?fy@pQWlS(6lWd9gsRZl#K;}6LcWlY?~?or(S<A7X8Gs!
zO!NGUUV$k;jdAcL&zaJR@TG+6k!j48FI!600qb4vsoiKz)v{knIPULmmHaD)qg!b2
zzFOhXvHz;m&-bqxRIf|^HE(f3HPqJ=U6A{Auc<Ad5x(I$Ci($SGL7Y%3I9%AOTQUi
zJWJ$T2`vN#Jgx)zZLcupv*_EYLA3loN6z%F?heT#g~n6s;a{v=w@`iJ@a_p_m3${P
z{|Ke>?*x%BL3&2YZE7Vr*e%Wx`ByJ7+1(;}!0U7+{jLx39g=@HO6WBBo?}PwQI`4l
z4C4^$SibKW*bK_=?>kqK>wojg6Ld8Hw<t5cNmrU5SebsL=?Cfib`_8xMva+V@*f8G
zj$>!+pVv<_H3Od1{{Fk68DdHP_tX?fer#V?fc(dau={m5e-a4k)=xZRYWhA^P5+T-
zcvRKOe*~&&tbw0ePFlTEYPI$HXWlp*{h8O8RN4Ev!QDbFwVyj)mSh+#bJ0q7rz*@y
zvA$@(|1-78vQfd9$-jts)vq?dFA{>yEVlAXXCbTMFa0~CsM`8hK>#a#{;TNmK_b6S
zmDj7<|FyR_rPkkXcGv!wwK}cY>0ilzW5`w_Lh@fhuByZTa>Qy5Bboo5awa4XcxoE9
z{MM))!Ql99G_9uD?;IZQqOX4EcrYteKz?sIC`u&!_nxzEddKtzJQ$urC7=fKNR3Z-
zP+0Ou&l`u~Na!D<_m|@ANMt=I`JaNPZ8cQ>C)L=XGM!S>>$ES@-NzWy8kS7DdjYpt
zMgm$SrF3m0fbGf3B~P&ugUcj)iq+da&ia!(+$~lQN}0avvay%fRl}nrS){P_b*k)T
zm{u*5axcd;IgOC+ZC~4M$@cE<ZFJox`QCwMdL!Dh+&c(f&79oFijNLSwvQEWtwbRA
zsfshf%1E)G=<Imvp5CmN?5W+SS{tiXj8ftbjN9_5hIoD*-%rrdkZfNkbp14@Ou*x;
zcjdmRn_E=b?dK^<VNqxM1)J3Ti5#p@m{aZ-h!4S+WUc+hD#@M}og1JZ4)0zMn(}F>
zi48gm`#aJJCUUmF=cozGu<vh3M+a9(xxXPDVHtdSDzXz1%${!F+ce<NXiLhcCxnRC
zGrTlLIJ=|l87YUA<Qay1DRZNAuv?ssgDXgz)^#F#reQzh49TDAH7A*UWt5rdZb!|_
z0bXN0T49Z32gD#ORAq8NR6<V61FX3r)W19{wXqSA&7Ng#EM?L>%ZdzAq>LgH(_6dj
zvY%~uy2aq_l07>rcPCZ)J!kg>ohzSh)mvEA(iat2XS2RN73j6+V1_>-Wx$h2fIKJY
z?#If_p5t_{?~Y6U9K+fzE|jwRycTDkV`YM~r^q6O0jJB-b5pl>O6f2wIV$LRUWa7>
z`N<;11X6?MdB%m|0V(4<>OuFj&yVIuC3}AA$QIRM&+k4BX;&Lc|413R6WujxhrS^7
zV7p{5Fx0vSc|pRk5u(c~wT1ncywLMh@|fK(3|<aM{z6AJ!Iq$Om_>XL=VVZJgJdrX
zyy#O_JI@##MQUFZWT^A|Krf`qk_HYmqK4212U;0$GEoooX4MPv;#B+zG-&o>%NSh2
z%yzI_oGayvW6)aahImQPbG~FRG2|otlE1{Uu+4yHWu(UGBa6sOqv1t&NcK{zJTyEk
zxqse0HN9O*|2(QYpO;yQ5n9S#=19k<pHS;BQZSB<Zuv6r@wje!Uv6aFN{yF$4FpPs
z*1wOmrF^-Uo>cYvilF}*$s8P7$*TW~)YdM9u}mq19|P%?UVBRU`O4J7dfmoVDV@b-
zl$%gxQ}b@DS}s{FP<J=5qsdbWah#B{N=+!IUKM0wMrH|~UMIQ37!D|%SH)N?XY(Tm
zCDc=RS+awoXY(ab;1yCH6i9V}JlM<Q&0(d_4z`?Swm)`w_c(%89vpZs5jn&w=(y5A
zq)<+*!}%dbNx$t4%0s-BDYeWG4Hzw-9U2s}d|~pZ6nte(I@H=3(rwpa268wm*<q>1
zW>iuhDWn-Yzl_urYZ`{b;i)$dKy_B7Hmhzp++^a&Bc(jtsOc8VY37K)-Mydovm-2L
zXo>2?BcjjD@F=7!?t}Q*BjYm3<|WE@sGZ~)Bco{1dEVg(L_&^?@@GhPWNPgJm6b@b
zU^7e0Na0UmIg(d<N!Uu0S)?$1;9wr9_1*25-({q94>XRO-|Mld>rP1Kscx}YCsKZt
zai9%(QEH-VzPO`}z|lcieYDd(2{ZGf9Tgib2r46mpuk!=#;UaFYIclQ8An9(W2}Y4
z(Lcwe%Ij5^9&5!1ndI5ADPxW9IwGZ35v%jDj%xihYQ7#EWP)aOtE1aD-3ogO!Q7y1
zixkegnR__1@LE%<<^$FVn;+MGCOR7{i;V+IkZ!oj<D3#!Gac3Wp3-dzT5rB}WYvJ=
z^9>_Pa;cOKV@Y7|9q%oT>(U$W1R|6>+%1^r<?(?Nb%mZ^5SZ9eXA2C!T3HKHnRU~M
zbEMc;>BjMdkeu@*OW*~PpO7fk9+xLL9TRF0oai}|)02o-7Afs5!j|&bP@s<^#l}^I
z?Ife7&FGxu7}z^89{EYu{`~rJQaU&~$og=ym%(a6)n+GKnISBtJTk_+TXir`cEYA$
zr<@X9?_VmJ!y~I$adU@ZD7x|#BjT8YI37wJG@W#+H-KH8pXw+l5$!Sx>Hd<f&uLag
z^&z?=JIzPi+>z>?SKI$Iue%<lQ+jC@8l8A@di1BOwoDc&CI-_fKi!C0j6GhJvavkF
zuwf<-NOp!1*wPj33`eE=epV#U7{TQsuZb=(7MPB&vD)1NpKI>$?n#y9C^gyLrE}~|
zufs-BRY0Vef6R#d%;1E|%$eR>-8gQgx3=%eyoGhlq%2E#c4A26p0j(B1+>hJ@>RNq
zpXCV0F=4W^yu!pxu6~KLoc}1xGBR*p(5{>ved=!1r6W?hvtz@0wxe6G^DJ^yl`#v>
z3FOEjR`)r^-w}37=Xi|?<-<8nms+gn2JzEdRoEh>wjW!cNU6PsnR~7ym{fs2FJQG0
z&x;19H|cs4DHV+eq>L0hd#nRF--ua)0G%K8nPB<(MohQ3Rm$@n)%xyQMo}*C)cEv9
z)ixJ+l__0nB8OS5y{XEXlj=fer_O{6t-T?H;KGFJ5w-Oqg`ZsaK^H|HjbgJ$iYcJV
zCQ@j7wV5t*63ln_+JteP^5C_h{<=5=Ohgdv*BT*h-Gw;V8Xe%kv->RD%5}pFum4JR
zal*D;J$)Bj%~n4gsWKi_v*;2dV5ndBz?XOxwkjA&d8Bj&WvX9d9l!Gi&QiUUt}z=W
zyENFYrxF!R$1V->)k}-ynk)4(uZizhW&E-jq{}6bjLGQ@*f-^6Uez1`rv6!tE)RpL
zpOc;Ja>t`u5(Dvar(iv{O&K{;(>v8S-0*U^&{az`tOxUsk-QNtV^<`Q-R|pB=||{G
z7AclnU5g@xS6X*xuX7=I1E<6lB`oV%q}0x0u&yw=y2UbPVdPBELA2MEUQ4%`XqGF3
zgE;#0$iZu_!f|ELc?(C+*5s<TWDC8rst#1cLfhDM-{E5~<w754+LWuJ;Pe*S%&xMG
z5&YnJfJ)a@scqa>R|kw5%&ztxE|UCe$AXT~jeDe6gVg7EO|-xmGoIIY)6B*En&^59
zC(1Q3a;w?n%C&~~xVmHxt)lR?R&PLs`dTMpg3in99XW%iswz_0{;J|5g&=iVR$iYv
zft+0DwbajWk7U<*r`3xZ8JxwSnafDwM^deLy&-5VLbqqvN3%4<YU(*yF%ai4vW&@g
zgCXt~$4hoYq7xQL9w{t4+AnW#%5eU2#P8)e@_}{Pjh?ffqm4XrCb|>K;wP2EDRh(7
zQumVXrEW6z^|NaE_ucsS@M4ycLTvQd=4Qj)Ez~%;Ik>NGk^E+(det!2q=Q5FTBI2r
z=x%|TSu-kMgg-LDG+0I{tV~sKEw94~fXaQ#@R}3D!EW(dDIIL-im)g((a{wlQo4V^
z$cfY>lX<I@iz=<h!pIqgIyCy7SL480e`<I`ca7wcGsRGq4zE-fjSLt;t>sv415p7P
zT_<^@)PIOh^HjIMjnO7%B3PwXK{n_Z)W<yN7<2_?{f*QFq_Njq?8P{R%oay6CN*xs
zNU2qU@v_)^Fh!0mNxebGJKQY>MkHV2_4ULa0bSyV)%m_O)S-GIv47N|oh=Po&A49b
zHPtl2w1|`*Bw&&cnZ&eim2AjMVJxdjF=SM5P(C1K<fw;=#r~Ga;V>?{#nnJt@>_x*
zY!ke|ZuQQ}tyXwsm2UZNwZg}sUvKrM)TU)fBc&a~tQ<~cY*Bl6*w7A9Js&nyXYfbL
zVW*lwS!Ts`i11Wpo?<n8Or1K*tn$d{GASbiK~VS9a?e<`M6v*}gmMRmmSE(k0yqdq
ztf>(gG-3o`)uCWUoJuu09+on4Y|qh7^lwsej1)GLietOGpZipo4q+>lk*VetBk>nr
zaXog+qmo5RO+WU>e-XpE2=!d1f=q3yje0e9rMlXVTF<)0Rgy=Bdc@XAS*3Ppr4`<l
zDc$I-aPcvDfOB?mw8D{U>)5=hGy@_ljUB5NNw(7K*-jUwr$-O@N@pw~73E4JdC1Qc
zS0y5!&_P{gt+OV*L-NSsC`@%*q$WA<Waocds<i=@XH}}B18`e#WQbueZ!@CW%!SpC
zx?oR}ISgrRuUA`%{(dP}ceDS)7<#f{^nV8vgXAn6pUEW)cuH-Y+hYW-m-2R}UzdA$
zeurfYapaZV5yY>?kD1@$rN-ePlOa+Q-Hlr34Uxxj1giH9(YY42QVz5Aeku2)*rmyx
z(JqTHe2Wyugl;lY3dM|*_fDrtt${Zh*c!!s_r|1SvA5Wgn?87B!nqxtTt-UG5Vp;C
zB`SBSJVi=XKh|u@>*}&t-sMbNucr22dM%YPyfl9qIhbYnU)o>|;?b+hJc10%U)gv3
zc-dbC16!Pi<bM?$zf;QeeW^yd<levpn}p3$%dPUf(<IBiduE@arF2)^%md+oIwPuz
z>X|8Wr)-LOQ$~(%B)3a1QmR*>?(QJt)l$3LTUPsx<?L=lzyiBMnvtU(34CjJcl(e<
z_c`zRTZ$-!+ekM-cL!~1oAoNxkLfY(J=Up#RZ_brI?IUMC)qvT#R)a-aOZ2JOY@#U
zvlNw{-{YN{U|}pAiS7j*ns=1+0-bKdu;cEv(u2#9-I~MQV#$bP_geGEog&S9y?OP;
zuvd!I`tA+{qVQ6S7fX4sv1kza?FGmREN`-}I+1H{a=Ot%Gxs8SdQO_X$xvHx0t|eU
zH96Fh+`rSiM<`XK6a=%LzA1HYhxEM08TcUH>_n<#YKzp~>}AH)G>sfK>?qM@Rc@2I
z%-$U1x<HD^(A^>amPirjf+^SAlY-gumKcui8Wd<PQYvv%lD)+m?iQSBH6sU%>T7+A
zb*^=r6b_Bz#P`y>Cm^%;79(Xu&kNpaU`t2jt=;FUMhg>vkN%SEt=<wj(tNAozWGkV
z2NN6Cuof1MXb97`d}|UkWTEHZ_1Ib7<`r;@q3mj&!s1l_einI?cJ9)AoAdM@7?Quu
zC|rcORU`yh(`?XF74v%ZB87iY?Z&qU3A%-?y*-$S6f=$9Zn#@2gw40R;LRtvF?D8>
z&Q&jis~a;UatJ-5cJJ-ZI@FuIBf5h*GbOcm#AvKQo+C$9*Q3((9873TiGU{%kK!FM
zxv0M*8Q4|LjTqRyDz#0QvA=f0hl#|g{dIJyyFopYDTO0sN}7=xX9K`e8Y#S<%&5X~
z;DJObdX9Qd2k($x!eJX9d1v%y;Zmu+Gx~{T%Xx03R9SVTnNrZPj@V9P&L9L9DFZVe
zeO;v<f+4*sHAy9TmyH2ztG&w`!u_Pj2a&>_Knj}gGJ-}{Em=yeXzKiS{5gf^pn$PS
z-(>_0v4!a+WHa|Lk>%adpP|K4dv`R0alnS}-BzWAsnLuKw6ZS1RcbT)-y)@0L21X)
zVI*&sUdmy8W2OC@U=}B}Tc!3lhP+#xAX&<o$uYZ|p3%a3%Torfgw0Y!iV!yXR{l*A
z5!^#~Le<&ci>m6F8#C4+i?y-D5<NV~{5yND=9nX}7&eSD-7Mx&mK2L*F&3N|yiLjk
zuV_inaolwm0kzSIi>3CSU>yhjG!rRZ#db*ZJ<*f~P@qjMj?=3S;ma=G6E-W-HpC|P
zJ=QEn@PzaN&jj`jYteht`l3_!z1G8HkCp7b-n6b|+ok#5pn4gHab5-`QutRawMi-c
z>v6|PFKSLsZ-z{HpVcHnZH3g{m$<f8yAvrjUDr#q%6kAKBllX&(j8K~&-gL28vEW0
z;d;>%$lm*m*u&X}zJE_=$*8^G+QdQ5*5duq528-CO4D=LfU?@9oarZ+?M0+G$v|0_
zUaqa@M?HrLkmLOGZ=(my&2g!DiVe9PzN9=t0Pd4!r0_wi*_%?#Vpi@*aRSS5l%DDq
z1Zl9qMjk6Xi=BKR_&z)&HHV1=!;tzwH9BB#Mrsm;%fc5aJ(PocA8_vBPb?!tJ+^D4
z_kkGqWg1icLB}`VRdrmeQaja3{-Dvogh%K;xJSO%?~a51585!?zEX;)H-&qYb@oZA
zZPH6A_KvuBKNNgKk8PLQhrE|>@4j8KfD0manhvvu;(+;(w`|1g%tB=pA4;nZgI-oB
zJ1FH|q&Usc<~|(FO>d)G?ZeIz-5WaC;&i$hWz>3siIGw>b4rR2$1p4(k@CYKYa`~8
z=mj#)`Q`5t6E%#p<|&Q5)J;v~O_+DM`FDm7oy9p_sx@;r3uzH4-PPgyh!jyJ3`8%L
z>`tkJ_aojKK6aE$EmGqsSzP&%Vk^%;G(Y03Q+sk#9M$C?F%$OxZG1@a5gW|KY|cL7
z!#>UqruPx!*dXo@`KW;{B9m(l-ZqFi@lhv+xE^L!Gg5?<pnLO=M)P_`WzECTD4}zo
zQ;CnJ{k5G$^*$QhUV_KxW6?`Y5H<smnljF1AB%D;+S2@3>LdOmwVY8fd6ecDLxrWP
z`w;6!87X?l81x(yiVR1_$Gt^We7mIfamS0X*nOP!XBVm}ixPxbqx_pGg?eW+Qwklu
zS&B&M=?xRT%2O9p@8gNTESI0~HjzEtSG7;r7!Y`teIk0Orh0-`5YX`nr^@!*Oa=|$
zD?VWZGdjpr@S<qXj`SSat%zq~XzxG|7v3E1ScuG03Kpa33q#W_u9m!r>9u<4h!jPT
zawVeSF!Pd$=H*w7qD1bqs>9rGsNI)Zco-V8`<z^hI6(sRD6#2<TO7dVsW1fWlEO2(
z#pzO385`KB_3ktJk2`_u4vAeG*c#U&rJ4cTEK&sOuqriENlt$VR!eyt<~%G#l{b!&
zR7Ofq7&b_6O?17r9K&<1H^)vAl(n@!Zj*ZS5IJ}~7`JAm^jHGlPnFt5tzxaSK#6>W
z5z~XgGD=`V(~aI*<DN!d$vQ7=p=GLNStV-gVve*H>3%X>XEb5WFGagFJ;k!N4vEYo
zrN<~}lyye$=pa#GW#p)#yh(bVgJK`&-iJ474lm?f&#h|X(MD@njqE5vaAbEKLCbm`
z+4K@v73<Y!kjk$d!ZA^dJD)W6r5v}u%_mH|7gf{+LA2<Ei_FMssZGS7vPr}H5h+}a
zP}+<XF}LL9k<!_?Rf-8?%x=38b%c}?-KYPVpGbNMF$N<$mC2CfLvE1Tq)F8EnPihj
z$KdE~(wuZo*wGdut_SP+q|?S`j+L|UYO7fb;yZzux_$8PIM`*%aS;DB{TSTTc$nIh
z^Fu?4vZ?56cMVHhbIKZTt!!~W#_7O4TvTQ5XBjWOHiz6Rq~|#nK~G!j6RypSeJxUi
zMbKupKKg>xVIvZU!WJUF*-<Hz(E<o798-*Hu|8Ntm~H7fMDA$>dT+fk?D#sjbW*E&
zDXvzfHZt0orwG#+K@OUcr)N4i^ioOI)p02zg^y|GUMcUlc83<DKzjE(<vPYYm@js7
z;DPiF>7+NRMQaZ@9V|S|q3nU6d}NU{BZrtN_?JgY=frj?BBgT6D*HeT{xa!Rl{T|v
z$Oh}~4L3?{gB9r(hfB60dXBzlsoM}O-pa8`<fvWEo;FgfBic7F$J-KcaYHDVA-22n
zpkW?WrTJju>3UZ5EK)dQ;A!*0sI$l}GdUwHWQZYDbEAmj8lk}Q_Mnlqtc4);B8T8F
zUEVfYdo5iC9Mo$V*+%QBvTb8Dv_eByH>SRBREJQN+RQ8{Q)=d3ES26yYmfs2OoUBo
zOf>YRwkbHYnDwm6S;J^GH>IXDa5u6vl4!jYRZ1^sl#!amFSS#86%xH=B^_@}6%s0|
z<D6k3ow>)vh8<_e2*4=7{_j}t`}NQ)Dyqr*s1%;jEe?~?K?GiU2{|(u&*x@GZdX4+
zt{1*)o4t88WblSYiiJ_P0G=WgemU3t^3B$C8;4Ed;OO9DDK{H&L!+Fg^`f>~9-Jp_
zam<sL@sCO^QpEG9D{4#dl)FaFE#4(e;5>cS{oj^gZVUHi8MRpyIV0+&R97!_$yTf1
zEl!l$*5DL_oo)5z$MH|#Q;8I3C>Wx-m%Byn!6NF6tCHFpZ4RiGN(k5na(??z%2JQO
zL($~xC40!5VP~%#ixl@xG#dCJ=gsmZE2M}V<ug$kj$nw0pI$<|M8lyT4x;hG!?TC2
zK01@I;fGzo(eDq37SW44YL2pU=MNh}qeF|N2o$QWki1B#`G*DKDf8g4N^i4Dbr!YS
zwiu5UY|XaCxZ$_ISDKN+8Ko<imm4<5rE+LsR>*|mBIID3(R$Y*T<mjPy4}I^xjjvL
zE?VH7ut@VLq2~_xiX)}5iP!;AUL|#so|;!YN^mNRORTKQ-_Ln<Z@YD92!WADs!p(-
zs6|SH$*`~<vCb~mPBc9SRo@yK;EtgoSjw;zQAJm5PDdVb%JqzVQ&<;0M}y5xz~7@`
z+%Qwr3-PG6z>32;@1w>HwgSyZV{j1xSQ#mfv2^-Iim8JQTR!T;iB-{y3?xb~8$Oo!
z`MAnQm3kE3Ws%a=nmtpbCUF+xMtsbOIPOF#s_$DMqKuTDwQ#QRm^C{xJiwhvZ%94q
zEK)nHKmEh21|*9dWem$lD#voj0m_b;BqNI#OR>Y-JYGt#j+J%4^iq-KgIw-=-0<#R
zk20-29%5|^rdn5=V#YVSl7|53LfG^=-Qp}3tjIT2M!tp$7LNy)QHwa<Qi|E|p!6Ph
z0a1s+&QxqO8MU33(Jk(hEHX66=OJnCjDgp~QHENDF5elwT)}B%vD3Q1#R9f1DWk=C
zU(ZqYBTMpxH@o~+sXbxM60JAHmOV?U)w+f?pKykq%>7a)PNe}4NTf7IhrMQ`beoS8
z<cXkV3CpVNa$?omz(MRb@UGO)wOGknq}1Dn^&2U|7+DYVD(`+}W2E#b2{!JEW=03;
zS!A(IV40UPXl31FgIlH6p$@Z1p<C6@obsl(Y2;GrOj(HvjtsRFh#83tVq|+|k_ik@
zGehs=3T=?GlsJG|O02|~%|2=TTh`*~DT8j}%;A%XB|E2gF@Ab~u=u1=tk*JeJeQx0
z0b~vDdA&CJV>%`@0R%dsr=!u922@Pjxd*#b+!JhB*n@&uUsGsK`v@-J8n0nxcgRXM
z9qq9-A|hZqQG}O`HC;W4z4%7VXm<<S8NH+=yPdBZYQhYxQtQ-AjFg7zVGvh&^p8uz
zj$fxVs!A6LuHqYA8e_NHxZWDQm3x`dr)49EXxB%BBDHROS8ra{^)Rktw+4(LGF`_$
zIfIuxZJWAP|L~C(sro2xmD01!pTXQuS<6~%N@uJ^d{9zza0tDV%_Qo%9YZyZa%|27
zCx|paZoKd+!UBsK>(&W%DQAq`LnG|QdkKXGO52W-lj2XsATPq}^C?HiWJa@RDd(J}
z(u^F`i8At2MxaitA~FaTKv|TZ3L5otRg8SELb?Hp<<r3t&e>t8r;LL{)qOhHyNIdO
zj4YOFW?BAeN317{kFr0|OH)OZm|!P%pOlf(89pw(Pg}264Rfw3pD}99OZ}O|uMM~(
zvd<)ra;OPA^txWt3%3R^xf~qCPh5N^C|j)I=79t+0aidSQo0;qvVJx-u!;Rx?XxaA
z9JFMgO&#FMggFWD?Z!7Xd^YV4H~(ji@ZD3mAc{byE(mxVIcvLXrT1B*hXYtFS^4`I
zHS`gOXMgVvnLi6L)Bu?5?}Pj{*Xx>IoOOLAYgQh`Rhbd<X)3Kq#WlvKi~<nJ`L9pr
zUbM(0!@`n(2nphvW$holbuI&-XF;b^GQQbCv^B9=nqH@^hDQDmCaX8!EQLeF0G0pX
zebSB1r1U(68(q(-<#RD?ddjog?V;M|ygYkj_1%0fRMcWKzMGM$;T%{Ok%Fu0lKNb*
zW^|BNQ@wt7Rwkq@7rYknd2iMtSYxzmpSP}$>VqYj!{o5`f8GXuaCkBI^;0$8?9iPB
z94kH_jSlEG;qx(kY_>4qy=J#KiTnL73X@nVHHS5(O0s`Uyt%KtMw<T^xNg(P{Zy>Y
z3GY7!6IPCDP+b|dbZ+AW^HjIEWtGPI{gY!<57wmA{wc<j;GHaT)aZCbnw~=!iGYk8
z^-3eE|76NgBf<2Kl%9)_t>Rp+SCzF(Ewj;KY*(CLk<ydGNy%oNTj3sKV%#sy+2~<w
zV1Ty^d@`erizw?Rj!R}4*e1FUG8=odCRxY9%P;s$#ImY=A<zvizfH0)I4#->TKa->
ze2S$o|3Z>7y~^VAs7R#=MP%jo(rh@<tPZzUs7>?5Cy%)LV(_AT(c0mLkRCN<o<bUS
zBCX9xsSgL2`WJ%+He<yXjsAlVmNLpf8Drjy7Tk}e{pUjDmr_qQs&D&CRul2czO;v$
zhiOq`$(moP<d5n1CFdwhT~Xn9Oik^`VB6mdB+4eX8>T7c%Ly^pgVTKYa>yL6OE#_9
zmz^e@)!=JJif~j6=e)|>#4=x`iY$8A1?88WCUtRmIS%4999F(!8G1{+_LUeQT=h=V
zqmq3k*t}4KESq03R$@`&ee)cajxFqYi&T#0Q0QMVcA!m}#=R)0cOTikeAQ5lSR~9>
zqjN2D^<-a-&T&X{w=}ENMh(@B6b?ZSB#M;Bc_9(3Uk&E#eV<;!IWV*sCyIP6`lJss
zn6h9IiErivwUv^U>ucUNN4|P;7b(utbaDP#;O`&74oHcmY?#U@r-vJaUwkdNa}Di&
z-NXe!u6^CUGM-$5__|Z0r;VHlvJFKAd_A@NfV$qi^pKv$7ExMnu5%P+PQdc(K}!o;
zwii|Ptcq3R8<91N6;b<!wLhP;{cm`ix+{MGDs&n*zmeLU)H@E}FcIh$Crc3->L9{8
z{)UmhVkPTP&tVYY+xez<NpGgszUioRNrjD`vszE_nvtW1472~6PKKIxyW&-`BEGo|
z;ujRkZ@QqeQ1((YtTFPf=-dEX=i0Zdj7nPeE$io*2TAi=MhFU0gIn`&8Hppq+|Vh!
z_y`tA>4gx4U2Lv;3F+YNZQdw}PR(#Y81Mv=oqgLosJ9+i96YC65WSyQIU6xsif<cH
z!_2Aj+um1ua)4phSC@KGSdW2m<^OYZr_Elh_RmpVqg&W{Xgpk&3XH=0=07JAo`9g-
zQ{3xWjtCZ!r`{?wXr$Dohb0v$ja%Yc*1tFw4ZUIW<SEqxBWUe^aT4tKg!2lGuWSB`
zvG(+1d3?pF9U9i_#f5*qk<){O5cR``K2quu;4J<-RvELi_8sTzgf2z_k1P7`boaNj
z2b{DfFL`86LTm9Irzd%|?eficoPIqI!m$}CT(}yY*$Wox-5Jx#YoylmhS#fazZd*o
z!Q5!}43S==WpT+p#V|d>uBu1{+syT{XPw}vp%?hwuvD{WBhSiM?huOeU%ho*eV)+P
z@pWd^H#J3lYt;N#6Okb`(f&0CV|u$TOaE%5j;O!0j6x{wURCG`q*cBfebnt=S{}Y@
z{X)~}Uh%tzW#t|0E}P$V;%?_{8*8*%U=4mZNr0ZrmQ_lld3u3eIsRB+m+$$w52!Y(
zeJ@qkNYm_lPbNm$!rZ7zbo6ldd(QrGlztI88upFnvnsa>o!Ik6R(TBd_q`F~r8x7e
zecv0I(7*>KhTROVzUlTg&Jw?GO<@aQumuJUsM#vT_oMTtNcnx^$Xz!`&!G|ah5wdD
zLGM%6A~lKD=UT$QdE<85X}3D#Ff6<TVG4SOERPZvaDaYRcX5$Q^y{7BGD^5$xJRHD
z-+bkFBTIhZbkG*lwiYQAhZ_4;ie0uee-MJKoBjL;(Yqlje&E!qgMK?~(VMg753I+$
zfYFOwojT)4y51C)AEp|-8bpo9bgT2jJ^WZFS)`^|3^-Pd6uvke-5(kqBg4b0gMJwN
zzu|f*1Fh;1_LUZ}-}Azw79b}-GOCU`NNPVa@<)06Ci_t$hBGGiMEX=q^GAl2;5T~e
zmF6#y;z!{mz<9AcJU<GF!JmV3!^_RPlSh-RnPZNU+P@o~<+i2x_h^S9Uc->7qptby
zjzPC-MAk)09bo98NU3tb9hOq89uG?|aM=qDFjXJpksmvLdkZNvLG8!EomRhO4)+go
zm%aI8Yga9L4e<7q_FX4T%43)Zr2MflYUmB9gdZE<NA0G+{3IyHvGah`eq!wm*#)l1
zVK$>)nyDO?Grl)3huw_3=qFCEIvXFM4|?Ui{E4@QOQrV{m!Ks)9_3AGTpkvbryyOe
zpe*v#v4Au9KdjDT!uR95Mt5S6{ztIkD(u4qufVW~Quv-ARel=sy@Jq;+E2YvuB)kY
z;iraRw+4DPBS%koIFb0NQM`DFwZ5v4L#h0!_fAJ>XXvzEg>wf)$<Mq69X)*9k<z^q
zi`CD(J&k#3{>)n7Y@cVaLI8)Bn~%EiVz}bzS0K>Y^nmmt#T-}l{c~>$ZNX-*_VeiH
zavng+er_z|+^hNX?t%Ynm~*$#|9ET5ktj>?o2C?UOgv_!boau3^yfyg3WLn0PTCRU
zoHa&(U^W*cvU^OLsSK18ju9!1@4zP<DeV=_R3$et+Pz5ORY!E>Kcl6YyU|>=l)_5q
zB;-GXZ^(>Zp>O`D(a}$gq?h9Ym_GGWRJC++mQ`xAwpNwY^MCn;@lYTAz|8(d>f~1D
zZT5>jBR_@m){yjOlwdq{7DP&wGM=wrxSQ3~F_NMYt$qYhR(SPwi=H>BUIW}Fzl<@w
z4IfJFm)<_n^<d9_Y24Q{>*g<=6g?Y6wOMwazNyl%@DQS(9RYI+AC8oPj?0i3;=Nyn
zt+aA!DA8YeYn<yVqkm<sEz_er2fM`^rRilf{u8$5zcO|c8d>}*O);F><*!TzaKEyh
z@FFJ@u<+}|%xy?+?bn8Wl}5t<ItH;lXvRu26+|?M1N^nmm~m!k5tY;&t_rP1cl11?
zTO7liY}O3xwz|4&o`U;pGz`j9_9P2T{7tmG*jJqVH_lX*b*z<Xy}(iY#(H?rQBwZK
z=peF}O+=KmJF`fFU4Z{D@1EK^#F|7(N1b@J|1vD|=S$O}1sM7cEyLkl)x2+dT*^pk
zln)-~Dz#mw`+vu<v)FEp6|MH)M%U;nHP-$+SUjz|yBVcfHEm;P!SVl{cE?LvylzxP
z1>BFghssnP2_huV5f3d_XUT6}^fZLnL3J2qzx{J^^d@>UzNz*02qfC!U3@cxl?75y
z0E*u_**Zw&Z;d@C5}WkfBqR24!S7-ySF7Dq`<>Mo)G(N=%F$=Gnx0}FJTODF_wS;M
zt5)dhTtudtQX8fGT_Tx#NJxd=@98;W4wW(9es83-ZkO8cos<cJznFU|g^uJ9;rAx(
z9)Frg76v?yvcj`gEz%&o^7lqEPfFl>`n|WTm#Vg5cj_$x`Gd2yAGZF%P_A+(<sXcS
z^RAHQADjve>*VO*57qXj-%n=#5Pj_*wkU%r&4!%4d+!h4qN>;_TGXcj<d25=pu42@
zN5`zvzD2S>rn2h))a#`fXYox}@NL)~`odWL$C&4X?4^r8ChdENZRg8BT6c$*<KyfF
zdID^9Yx*Z|OuZvprS_+w+AWUkPe#-H<E8m0qr%svv^xFCkTM^~rEoMF&w=(S1F15e
zQA~NmBZL*p|9FEMu)=JN6e~6hQT9KEk*I_v()^!pjkR&l6|(;w;(0^okKKz$gw<PE
z1XlHH5WbjFoPA(3{*R5Udh$6a+Lym}Yu~p>pUgZ{Q=pj1(-e+xcHNO2b*I*4W-rQ(
z+%B0z_(U6liQAH_ea#9J;f<WR=`N`4CA!7vAa|s4nb`*%LxZf0W2v;Q7#g0tL5lds
zT-8fs(;C{>N#!vAu<8*GT#V5!@#pMjXPura(d{zxRPgg29+Pk`D%ue<>zrQ8kf(Up
z)fKr_`kvxl_x-oaSnVk@ez6yi!N4l1rzEi?I?bPA4R95f3DkIs^=5?o`qF%gac4!p
zwEZib#2w_t2kphKv1nWEyh$l4N;=qfok*#AV(S=m*_C_kGWvTqN^swj<ECCJkDq0m
z%td)saT{n@4^CyTnZ04f;#I@ax0iJquH1$%skWC<-7T)qq+aFdHJB`=`j@iQHyptT
z*OZ!jdFR^F-pe_*$EnDA7}eg<)nNUxD4~8X-FxV*{$j5*EZhj_qyiR?rOIrbl&OIF
zdS(+DOELeY=j5(#aa&=|VRi^C0DD{ihB$BT+uH|~lQ1<CJVURO)+0lyz|u`AX6GZ)
z@M6QDYVK{_TgBMz?ee*d(aracd7;|{6p%h}GZrWayXfMdt=iSs0y-(rY{YW!nP-?J
znuFG^oy-P8)A49}y@B8Jrn%xZ7u7X#fGaZlSnrPEhJD{YNk|D&SEUCXwS7{L^<lGm
zlu#cvhi4WXmc=*hv2KT9(0BbB`&d8c9U{#q)!!)XD68V5=ZP8umZuUd4x41m$r&8w
zcxWFNA4Xd`0T)*eO1V#r@oG+@XFY=(hXc}c7^7sJ%<XB3bAB^xc&pU=qK@-P>3eES
zKkA}UYftsg`y4jALF(}hgvJoP_K{Uc9z*jrJf#oaARf)924!5yk#-cMuc+ZXQ0dzs
z+f27i$k<bjOE=yrh37u`LWFdpzAk9Qf~98hjPI1$z^0D?JRYa}z0@DI67G<>sGz%W
zR@;43*Eb=!efvgVHNM%*Wr7#I1V<4gSPu0lf$ZSSh-U)!joxV_bz@%>8_q(cxvw$O
z?%J~YwNtTW7+HC=tGfb(df%C+xzcn8Tj5BSpw&A28ZVBxfyZAB^DvIZ*`UHaRjh(~
zDXEuwf@5Zb%vE&qZVg^ReTq=_o2h9l?|AZB`hUXGw_i*<@Tyg^iB-0?pYaJBUvK5q
z_lq76aG^De;uapL?g^A*W53XF6mBMT`j#UjtPjkz9Uby#q2v2mABl|_i?a6a2Dy6V
zRU};a2|D{34>SOw3>1tb*7R)E2v3dZ1kw#al+bI}Xz0{~0ll*(PxEnCAwuMz7CqM6
ztxOZWXkB|+bYWmf>Rwu3q1Lmplo}Y(C8(;aEn954Gy}VyJ$9u1G$$HIZytqp-c3gT
z#-cD6M$p4WLdUZ57~E8sey1v-*VN0@yovLyBX8?%_9({98mzjxNa19J;j+K?S1q;&
zrEh<uXp|>TYx^7TId!h@pN5_L*sxF^+0OQlf$rDkXn*4f^0rQzfsbdP*s8ZZ#ctK!
zc#AysnQWD@s^l)+twl;*IQL7(>1e4p+QFg4gEH%7Fd*~={oejDHx><=CuuHWU@t@j
z>(diW6VmtekkJ(_BHU|F_wEpJGR@Z4p02BV>DEd%)mhY>!uV`>O-BA9wmi**hp_&Y
z(tf(}s>OB9JW#S*(TEFsq+=|qsyZP^|LMlfi!PInqsN2Gt@Eg-8t*aHx!5JMj`ldc
ze39Dg=MY-vqOjTK2!P!xdxrJWPC5IYVf5U1#KBT~h7T$SEhv%tGpy|!kK!ri)OFr2
zL?rd)u||}|*P(V|q&Q?_ZD~7t{7i#9y&QE-wY%aNtJ-;(qxK?7=<HmBYEiqh6D9Ow
z9un%5%s*LXt9JA<bI;q^J%-+&iyZZOqne(X6bdehzGpfKsv9>OyKn?P(>TC=nE|Ok
z(>gl;B;J5F=!kjJc&4+0Eh6Zeo?^eRaWHMK%R{^aycHd|+0F3c5vC8zP1NPJP);if
zFTHz;CAbqg<1E2E9Z}WgLMP_PIBvw*M0mg65=}&~IE>5OGedAWCbf%L2c*&3#MY!Q
zQfi72s(FBu6{qw@Rz1F{GE>E(II{z+!vrEU4lps$rM?*jb+9&?QrRYLN00ciA?2~_
zm>C+$D`U~LM%*B>2N-8tqnMwa17hjdXdV11soIK_oF2|rH0mxz!x|Y;r(zV<HWygY
zs0f5S%R8ipbnBERbWmS>v)h|=NMG=(Ju5L~Gb2%dmeZnsw(MEKz6ddEL~T75AgaSG
z?B=sfZ1fO&g|wfgZe4aB|Ns8SQ{lnHl}q()i7}^<kzoBQyn~}dxaB*kgL+R{9Rr?+
zZ)JtdI$QL$PPN|>#>MvZ(_HXOS3XW8pY5~ZMCp6BYbpZEYR``8z~RV4Qh#>Lu4T-C
z%uDG}K;zlQZem8+U_3iXpk5@?bDizd_M#*GLp)|464cU-{8*r8j;MR1LXVVm94C9v
z$9eTUCM{}oHc`aNhtXYCLBr2QmP!fH;C)tU@hGPleSO{y4a-Icq!dQ>R;l+{Z`GpA
z`hw$ai}q+lISoWuE6qOR&mGI8?cepKHfBv8r8J~;i;Q`zttBwOh#W#KbYJN8xdBl2
zC7F)fPxxPF6Bal@a9?VH+uK;0juTc8g7%ykf))#P-*bXqG%of;?KwVf%a=?2IYul|
z;MsG0Droph<2k8eUL%Xi)O=2wI2&L_+e;tGA6|v`(jEX|+e8`j<_8f;Tr^Qtt=Bcu
ziQ=Y+SZ_3PymX3Ip)6*jFm6p0V$a**y_!`rm%2oh8gI@!IcnIT+@5RQIHfMN=UO*V
z`5mc0*Lc9qWaKT23Or?w;ob0J{pvJmJ~#E&4~W$r*M4qrY|)_l@A4>u?`B%>6|1Q-
z76{d^z{W6A`ijPS>3E8fx0m0ewl841{T-KCr+7YbJx(RI3_-ke{+U-H<asejw@Tmh
zVlFJxSkv0`s_8yNEPVZWG0-d2SK%eq`oM5&JkL93`%c?k;5Br#H}leG`*|~dN0LZY
ze;_=sDrn@XZCYh4zS)DUEJ83`isu=-M;1#b8r0n;j?F}i+A*_H(O&+eqM9qc_@-Oh
zaa%dqIe4Ykt;%S5zV*6Wyg~Y&p9F3bMtRLMZsZnT-81wJBZMn^e)NBsC+-_5Yw&ix
z6ZL!-=@rbe_Va^lM8v?TJdktDz?2(%eqxkHlPhuCq=*uj)W~~hPYDj^Ss~SBlUiOz
zB+G17K%e~8YX$f|<C{Jvad$*Y%%yVEOccgM-@B9-L{F!;pg;N|#cWa6Mx-V&&(T3I
zFvcJK8h*$iRx6I#^pT#%3#_mDqt(*%Qm33EZHEqe4L^4fGK3qrEn}(0nS0ox7cWQ?
zfH!ODoc^Lsr~0Os2Gjs0t|DrwP5+q8MhU%@^dLRfUkK@WdzW7(bCGLrzQR&{(NA6&
zV}PTEsJy-xCa!DX)C5PFcx7Mctk9)qQtFY?v)TLj9R`z)k%2p;;Td|chCXY)(C7Rl
z5nSzrbecCF*h}imH+lM|o^~kJv*0nW(^|oLP`uDYsD;%uC>?L5TX5RqWpJV)%|VI=
zG2!QZ;mkq*7dzW-A#*RZ5j&pi>Mt_r4xDo@GJ5SUZS6(YDUIH#zbGVmrT*&6i<~*C
zGtynD#*1PMm+8SrGZjCFyW;IXEACGL<nc{SqWkE!UhW#J%Befv6p|E<RvDuvXhOS~
zw;f7P;LE~NpLJ68DdC=j>R6bI+N%9|IqN{{&H(FO-+|r(4F;TKc-2a)9q1!9!R)A4
z`TXPtr04_iS@q4m{pyx##5bLmNGApcQK<(erN_&yZEv$%oXYQK#CP3tj~(c;q|x%@
zQg|u#=yp;o%)UoW9+=F3_6?TN@@ybb+2;KuM=*$%noG@a1i+7T*tqh!@5SDrx~LK5
zNR4;bGI#1;>c(UE`3~bjx7dA;G+vzgfkE5!BHE+$`Fwq0FeR2^&y9IymUzxpqX9j}
zd009x4tDD^Uvx@;$D(wML~YJi#B5<+^&F)}t>80HS<}RV@XH}KfGvC@eJ`<&oy4O0
z5+AlPf<x*K<K1uZ+afWcS1;8U$Qmz6{eJ@W*mO)oL_)M*;<T{|H-_swV!30VH&4bA
z#<Q?jUt$d5!m^H&W4Q6oyZE&dM@<wE{hIX*4P);)*iZ2}$9UF4e$B+`RS#X?OQR;;
zBM{=HNho<47^7ZqGS#E>q=sU=GzMgp_X`_QSl6zHZT*~(=1Yw!i*c_7VonmUHF8HX
zxLhsp$WaFskv3jq5vt{-!7j8PmQLvzyg)D;4YA_tq3?`76VgkR=)IJ=ssJa*%oTgK
z{be>%-6E5|m(6&@A&4!;y!JBhrdmgn+D%rHdVI6BDYjEteDl)~zgE(CnX!QDBI-zu
z0?q=R2=`0d3vhRc0(lhha|>Oi9+9z^CH8Yo>}AG&Jq%(e-SLz$o_HMZ!4fVK&Zc5F
zT+bOr?`6j1yY7@ZZ;NAM_5Z&-34xxb^hHXoF`_<SZrm6oj-~$ciW>xoW8h^kH+1t)
zkw(HW{TTDR87W<z9+&pZLp1DA%r?RNI*qreJB5ynrP|oB_el{cPG`DnrITvwOQ=jg
z`yq%Zp+~~(+?|R=Y};S%9A2z9m*<@D5u$(P6-H-^srL%&jm0~^!i0updxd(Z>aQ@~
zvg~2{cov3`J?qhUg?GTO*Ye04OJDO9#tXelU~XY)zruTjHv!SlBefpi7G7b;&3=$Q
zN#y7Y)a#@ZDLvug=;jqMNqE^oW}{ULB*Rsw!DYSb8*ffy`sqRLTtdP2XB|J+<07Kr
z0DZ5FVMUMO{*IJhKH;3h%Uxry6TC8-J&P@B!;4vH4L87EX`EZKMA{A=e>}hY6SIxb
z5<;%V0xyp=qnwLK;eb=OU@EHjoVd*R%8&$BE(@(^<(iFe%G{(lda2qlo_-t>DT()j
zpc5|S>EY>u6Qr-^_|&L*9NExGRI4Rf-lRK~nvJ)f)n!%2PBI!%)gl{NMC~SDeDej7
zU>+94c0z#1DG+Ex$a)!j5;_d`Dz%o?wo?m+F2OHb*217%F=%Hvvo#-~aR$5>)l{L~
zFLPCD2h&Dg<(<<APmEj)E=<_?ruKz)SYO4jy~?`S;wkR>s|@uDz4Mg4YUa6ciHbz-
zA}R!ds9D17Y<Q{RL0+M1dgX3$th8U11Q>Smt4lr|^eKn2$k8wb?Dv$yLx}wztq<eJ
z50DM>Y@%H^-Y6yg0f%0E)8nUy;J7|lJeS(Evj#aR*mem<Ti-!xY<W0c=RKB9?VvQa
z8g@{R5_+b~WlvjjvV)8ni`8(CY~_{x5CmSXre}AH+ofG)^Yb&1qfa2^2Ze$gP`P7o
zI~Ilf=Bq}^;#PDV{Tv2&f*uXI;<@XTVj|urvxz#@Vo;}nj=d<sxg^mla|aoVTcc=u
zIXJq0wt5u~j*(!)z`Q-!rke&djBx5zk4%kHAZENO+QBE^IM{i{Z*ZBrSbwasd9b%K
zjQ6m8uyJPulOp%es5opN|GZKy_Q0S_!zFC5sdb_au0tZ*<-x%t{jE=ZDrvTAR5xhu
z&g)fZbPMATJ<fR-hfLebAxYSF;AZYS#Aw{Dm+EVWq;c0bb^Lk1y5|qGSV<N&)vnzk
z4bS1aJ*RY0Qzc$so|r_y+Flxe!3erHKg9a4&a<(q8Fe3PxHtwypw@9_qZTDRK*l0E
zq&y^sslhgDHWgP}KzgY-oYO-Wy<amI{TgR-%b{sdA64V&Q0p-lg7G^XTDh0)LB#r@
z(TGNQWKoR=DCkbZQ$+mPQ%a5mha{Hmp{5kixq_b=Gw8f4W$aLAr8>s$H<BKZBJfRd
zl!kszCD;c(6k}J$H>_XXpfWQ%GMmc5)Tx+^y+fUu7#VYamhCTR$YIf|0grm>J1lxN
zfB{%LEGT37-v~Z^lcs)H5{V5Q@?}*@BY_)-8M`<JAUdSV(V(gJVX+utv#vs{Q<B@G
zEEQu>7T**Wb`h<reqW1l>kr{{suHST%1Cj!5=A+iNM43M>K$g?U4<CURe1CjoCpy)
z+^N^wW~{s^g-ePEgTsS*UNEj7ZliPAS^Q+1^T6-vyN+xeZjB6~zM5X}`1#U~g8tf-
zT7txiMFToK*cVX(KPd#c7p?I6haM%Jsv&Lw;&IU56D&RVLN+jtiTJod4kLOM;rGzW
z+~F~1%TWz-gwyM8dp%5kNcxVjp7C=SgtI!_EiPIt^&^t7Z$kxTM_8{0N0&(Bh`?x(
zeMcBC&bmU{UU#uRO_ir2E7<Ff1tNkA(VvB9wd__=C$d;u*t3=1MvI8c*(zfb8}pu{
zrgsdS*XOpE8ol$9403#dqS5y>`sNv>E%Z!no(a~_8{E9B&kMDuq4)Z1Y34;&u&Zo%
zQD)^zefzl?h{LgGHxt_PVm2U3>bcG5C1KYqVq;Ye9xbuyfPFtNIHh3%9d8Wu{M??e
z>wfRbb|4{wvsH&UCIpK<$JTQ!t1M`0F43tsP1s)_nFNrx!k7;F5N+R)P8ShgOuO2V
zF}X%Yb%g6j8h1|B=WCC&fgf#gKe=&aum=C1+QZTe#P(h^+Fy^0+toK*yxK)JPWh2G
z{v$*SRL$BIoItKByW?eAZ{SA*om^j<)E;J$t+4VuFIY83vlrh~Z~MKLhh;8(!(*dY
zO!R$1d3DTK?|7tCt_c}<bqrVALp|%MEH>ddKT+tE?Ag9PA&n@Zr+aK>qeQYB^cl4F
zt9`Dl;Djr$@;S8)7tNSs81?|SBH?2b_PBIjy~mi_sb-wHIMv-<MmdcT#B~^Wy@Xj>
z(u;5Y^q@cMHW#&YgGLPgQK6rXJ6ZaUvN1X6RH+@6=7$BJX;8A%y(IIUP0~>z94nS-
ztUx0nX}6^5NRB^I+6f7R!7^(Pdgn(4<FQhnkg-VdxGrw4B2sK05Ri`J<o*VGqsU?3
zM7dc{4fyY9^rDr-7_}qjQhC$^k+DZxx9(UjeMft@^aqX7ExTIvjaQzSO>C^Besr1%
z8(6)wqm4y~*kjT-+GggE2Sj->!fEspe>)J_lkiBs{%B?HRq^zp&W?_g0WNX+S!a=I
zz~6N?FYb{}G&ZGoCxQ%}o||O$=+IC7(u)i(Ng}Xwi41&MTueR2dV2@~Z^szj!+Q8w
zJ0=DcXNX2T>CcJR<C}V(`ArKBA&6%`CN$551Q0d?)lt%PRNZ2JUE2N?HJBfhX3;||
zD`Wos@MTg2Vtc9$#QG#e$1&VCM6_$^@b34p3z|J9B#A{EuR<>+^{|?gn^a3rF^;v?
zSgbU7sqa`LRA1p+B(-Bx@3<mJ@AP*<>c<+pMsSd1#~MqHx<ML_Lk~_Eou+?pt(JDe
zO|Uinic)GBJ}hI7==J=}cTr{RLV7!1eSjnOs=8XKvq6pBvuDEUdsw~rX73p>)V2rk
zF0k8Xv6Fi6fv|Ztj(sV)gc}^#O!W{*j>@o>>Z!Ykr^*dRWmdO75;D_>vg~W&doxmc
zjJieIks_v6b#UOZUtR={T6<&h&4bc(W6r+5?i$Ff9kb{+;w#nP2<g;)YG?p@Sz*%)
zr>b2Zlv$@rFP~RE&|ejutEkc?P>%E7sqxN+AW{=JQ#hhJE(EfljZ6JFZ*DDulO30s
z!0!bjWSk;OBYiV(lN)}`3i(b}wf1qL+=rIgN}fjzl^L$sTQtL1e6t8)H5JwR2rO7&
zI`XL<n@w0JsNeE9lebaEayAO{61on3s*1;k$y$-K3xZ|7cS5%e1o_%=PhWhqsBILI
z{)kX5zS+xpkf~-|eSXXb9%kT5c7=&`#}@P{Y0P($FyNY1ww~6w+1hpl4CtgjnUK)&
zz+k_O&5w1xrMF3#6Gc^cEi<Q+=)oPLZ!Kv|Z&{UijNSI^d>^VYQ1=pQI~ksfavITu
zr{j1l$59v&^v4^$L!<pt^Ut8(B=zH?qsuiSEIU3}riYAlRDZOv5#O+gkOej8wo3DO
zn-HyAr0ux61q)a1UoXEx#u6Sozw-CJ3U5iL6cJ>dpkWH{Yw7h4KT2ku$kwt&(mOs`
z3Zo;75YIU;#7kPTAm%Hq<sf}Q8fBga4)s)v5^6`##RbL%UIQRxW<jhI`Z9q%*xK-1
zmJOV8UQd5h70uQTEIc!eQz2hqI)q1;Eaq-(fsuDk@*WfwQJb`PURY1a^>wQ<Xw-vu
zJB();on7E6l;7L#MO{^ikII~93@t`)o#1`57*K|T?YH!uV5~ps1gV|ie9(icT~a^6
z7ta;?iecsneAqnc)Tqj7q+~N?^>cLCKEa6SUrM;`3D&>XB7a(7EEOBjm>e&r0Z$z-
zrava5N9JY1jg%7V;j|6jDm{meUm$Y{k?MFkF?EP1<aD7xQTCmfy3s)w)J_bQw3y@Y
zdKA~W%Wf9+z>iXSaRlymeon9vmGvoVF4snR`)wDdo`$`)<C{IJWwZN1e#hAQdQ3bx
zG8RZwtrEOf(Ww3_H4pHWJ5?+nSnH)1$5yj5#)Mxzjkfr6j2_>xl)ICv`_VFUCmO5m
z_tj5wF4)6`khC3hvoF43X+<;`9@mphaG97K5kx*-Zw_4i4TJ2YG)35hp@$k#Mh`Z(
zS>Jf_>m*;w2lb)ewkIs;j~`}^X~Z=Fj#*QdI)aOnOva8sSvpPvZp>r(`Q|k5I%->Z
zN+YA-Nw3PgACqz}rRW;rN+)}F?bj8|#(rG-PENS=LET!U^!ENeQco!?Zw^>fZRSpg
zj%%29<7A`n%7xN&h|BKnlZ{#3LgRZ9%$w_DUSjB0DI7ZDXz8RDh|uH5%ucqEQQzq7
z$)TBXD6iCzy`E!2sjrv0sHK;qxcqR6_dt)fINCWS`e-ND{<dc=z9Ci?)2dIW*P|A%
znV|Txs)V~r)wF7y5)+_JxMVYm<5XgLwIjvl2^f<{Wm~Y4f#D^(1=nDQq=*t~k#=HT
z>YQS%J8=QOm2J?er^;+13J(ar)KiQdssZP`$g7API5kaNKit<xEBa1#4rsu{gw#&8
zYUgwDyB-<3Qy!P>)acM<dKb$}wYcuuJT;86C4{7L3{qixfJ(WS)1Pr1a}a$ag@fJV
zb<#Q2s6X#Ae#ASGmLha8jpCa*e`4wHB{~KOewj;+V+zxToMys<;YAGQX=x~l(?VSJ
z$F^&ySywf>9k$d@vmUX}AjTj&Elt8vn+c7mZVxBWwEbp)S6A7`E@?*zJt5Kfb$xj#
zcLJEC_|~GdMnT}NIL$h`dbM<_db;IvS9h7dQP%E;%-~&fTpZUEjTX2jI4)ye2I2m)
z-s#>?^J<yV&XArSV@24V)K2&D)tx)5sAuUbKD-qiSu8y)u#MA$1ug7pxY<lt@b*jl
zbff>Gh5QbBFp<f%SjJK%4Tz!Fg;yF{xst`c6ZqGg=~4!zX0D-;)gqjYZ}ui0vaR>X
zdx>Bc#0i=66y8ev1=ce{JuD^8;|vqk!!@uleO`?h<_zzzKFY(@)iaFlQ%;aZ^&O*E
znrArMxH4g(*ZLC{?K6UHS1ujo_uIXInfJs<>w_I*slW(Pt%bMQTB+g0-b(!he&{__
z<1B~`y*}JM8!hV!KQ28dd+AVH=Ay9f%O7S1(Z?O-HI6`coDb@F(r({ttfwt}N43`&
z16Z|KOkNWsF|3yfvXsJ=Pb8(MkS~@IX$DdcBlO*7+cBA^Jde_PNJM~PDs6jiC9Wey
zRYIR9?o@5r#aqXOe{Wc3qpnJuhK}pgj=c(>O1v)59hrMg@OtIQh{&0)tML5vojG%W
z#WoKu-*a|cJ97_D^jfvvIjo=Q!>n)U6Fw0olwX^f|N6T#jWdl^cdXE7)*@dwe8dDX
zfgh3fnWnk8uxqAZUKLqmp$po`n4=&(i(t|-O_Gj3Njm<urE$)slRm_o6SH1m5zCDi
zVCueG=FYT^;%-OF@Ib2GhQ%8)$S+OyW!CK>Vr^<!uwxO@T=!DSDR`Y_)(v|tu91bZ
z9ke&Ro0*%z{35D{f=fFUSZ2?e<c=2WmNT@m%(-ZChG3}krEn6s_bZ*0t;fWsY>BJ%
z>=8Dnvytrz(T)hD=cHjD6Dblzt5+9)(pf$_`ZUK5={w6?b{(a`CPZY#H`B}rif)hU
z@y+iU`<vz2S;nLhWTWBWAP-eE&kBC&YvJu^Om8FN`i~Ud#3~r`l-W1SrKn1-WnOkt
zNzSvdAEI%4T#f!~u=K3A*y8!KUPZ4zG&ji+*xXqr5(7j9pB)2BKm>7iXQ!PZcA~1S
z+Sy5f)k5&=2=%k03rqFAk|==us_@y{?TxczDx^)0oe6kDC=2wPw4<6*t5&Q12wRlU
zZQn*2ixL|A%P}j1!$$pVV`+P7OFEIQ+IXE=$=v=pJA{nqZ`t+Crmop5SUpeSrPtMd
zF51F&P(SH8X-e*71JoBO;vU#a)y}atnbT5tP#?_Csxmv#6AjPM$2ywl1PA?1BF6X}
zn+4tC8ZLP_ylNCyk5w2n^sR`L-Y~-}d`=QOJMQgTWZF5wb^Y~&*(i?mv3U2QxL&E;
z7*{IiyaZCFF)(tjjrTG}_+0O*ZVLDnhI6fR-Qqf_dj?a2TR4(MF=ArSc?~b7WH--s
zcahErRf!mtq1W2y#w^tneZ)|Iza~#em+AXVV}Vo^^SBh}+Gx#-2LL;%Is>XVc-f#y
zd2aA-;i$}daUKz{C&+r9HMA0%<^r$A3vjRHJnx|fR<J#f6gT2nk80;7appIo(WV-Q
zR6j4Q2er8g&&*O<&%sGI0)yW3+9J(_kKkK8PJf0^mn?p8?K~rz$K~&pv4E$r;fnKO
z0QG`!Cn~03-$!tLm*<7}uC(B!S+8sN8G9*<H?uWv<~$qDQ%)8+-x^alj`W?MI>-J1
z4(k1o+WFP=W^JuU30*Dq!kzwZRTkgki3EoAA!(c+V!;o#Is(RX6Oz{U=7@B|+Y&ka
zmKVg21sc1z&wMn$+WFRF0?JV0ovQd+1h4!vTGb`Pna}yws}oO<UX<179lUDi#{dls
z+EJ!l5Ddna?YqGG$1#@9>e>a)0rzFvO)aTkV0{_kNv7<AxNhz*$7qd4)JSe}{o%^y
z1xCk$lcnt~4d^2+x#MYpXUrQl=_eLEP^j6)5xT{Fcf7E@Kpfc`@Su0ZxS1uh7sN>O
z)Cr4gFHo~s&0P>eGK}L|E=&We$3i-5xEFe1ObN`W+J!FZ8iPV~u;&cwEX^W^z%7h{
zMx^u@jrTz=v?drs+$n9(>J|(2w^j30%{(1rUQmBC0Jpzm=UG)Y<rfC|9F%a9P<j^k
z+&H4W@X6j}-C=NGY0n$<qBC|p!PL{AmYciKIDOV7A{QA-*85S8U!?CM8?<5GtE*k)
z7}fi<!-AQ&CqwEYm;GB|MRt+1ifbGeQw2+;agnn}pKjqt#v-Ls!EKc&zvqXh@UUGJ
zyw^8Q#~e8uejc=q!i<1TEdAAsPD0SNPys@@P*)dOcNXC!cIb$Mq*oEfPx>+CbUQW|
zq^QvG+Qw_GSznj>JfT<7mP+llA^4m{s|eO#8y#Z2i4c0NciHZrs$ije8eSchud4dB
z#vgO6wxh5;7G`8@lHBuKi`2w&)SSQQJ;)-V#*54E0-)-Mr%05w1_*si2|VpG8>MX(
zH!)=^6r^HBb51cADLuI*VESU~#;9AbeHVu?we5+J+QmLyIUlNDZ1me(_L;+6QsN-s
zVq^31$4T?zG~e-~qjdD7qJ43U;s`&6k$Y2SwlZ%$MfZ2H87(5E>%>FSaikb)Wi+Li
zD`qbacdh-tiaH#7(WI_#PZ%bmDJo3cIpUB@k|^?f@aRpAR_ePXM)E8k=E4zHyTnHl
zOOc&Zy-IP~Az9Q{rKqR<8(@CJOX%Hx?$tLh2_0}Q9@I;MearRj1j#Q6&3&zmUE-7L
zGASJ5MdXg-K_lxdD-)g-x5{kLpzb813lc?4I<`C&MLL~5U`Z~u9;kAFD}9$n{|MmF
z{d4V7qmZ&3XkF_1#)4$COO0GT+0$DQ4aY*cow~M0O2f){6vuINiz_jl90%e_pxoG{
zp~f}p($9&Ds^%I!j*OI!EYW+Z?rM#_z0?QCOmvuQ&z*X43^omYo{JLhHM7HHk;{Cz
zbdnLPd70x-g{*He>v>J>GVi1r&GpM{5LW5$%{q7+Dyk7k^fZD%f=DTnUDB=^P-k9#
zS&Z*xBQh3^*)Og!6?K6sqO|&kx!CO~HnF;(l~La9Vh;asZ_Zv8Qp_9gajqVvv_l%E
zrO(~YRW&<Ajp*wcxK1zkj_VIz5<_yi(Zvrpalzs8B<1{SEK^U92I`l`)N3J7Sygp2
zLeRL}STeMDu{19?xtzy!x+tO-)Ol^wQyK!N`q$gyapn0^ILuA9W26)9T7)>tMjux!
zJ)aeg1t`(57;j%M8sNulnM8Au(&I$j6^%5CW(r#5NMFOnb`5J&tr1=4$~o~7jp*8K
z?8&o~MFaw?Yap_{IegRD`eq{}U_{!HZ9AAy(|IE#Sbylx75`YnxN^=yDH^eZ&>MOB
z3UJ4p)0@WJ2riw*ZgCDiiy%aw&mca=b9g|H)2dYJa*Y4ic~4mriIMRr_rCaMKLKIG
zw_a+mvmWck12EKI=UVm@$^1LCKZ^Elyv}(v$r99z9KBC7CGC_lZ&4mCaHwSO%<3;-
zjlJ&AtOGrGU<maZQxT<gi5*AG`ME3Uyv}&uzjOpgcp1nw?w+R^Qi_cuPu@jJ?SB@V
zL<X~eyT}#Z`FMxLt|9haVO`)7AU0Iw+kquBL2rT9uSim_@xGatwx6nQTwzQ(?{;ZA
zG{7;PLkkv2?j>*LkBnWBmbGCWkHS&v$sWPkohy=?f7PhHdQ_%t{57j&Hrk(@xd(OA
zbNnZtEOY5Q{&X&mU6}+_13vq%OoPmf$4t_@&b9Q7S*{1r8V6R7Z=~=pvk^c)WPwR_
z26td9yxYOij1-<JrcnFJm<EgR0_9hlkn8X5j0ILV(scOPtXJRkO0?Z}kxqQGD30(J
zlvlb?JjqxL18=q}t-nj7hj3e^7vI!&^5d9|G8f;p#QoIOLv>l`+))kpg!Dy<*C7$%
z+Ct}*hKp&ujQ%E7eWA%jdzh!7vxR&1@b)7rji|2S9vnU|jKxk}CfH7G&(*QTr;{&?
zf#<G@KH!#;JQs#~E*83C&}kh=_1Xq!&_R<<0DX!|A5@)PXx#QAf}YpJ$1iiKCZ;Th
z>{oej)Y8yjY|;0~`>u-KV=1na+Evyl*A%3Fl?m_({1(7fKD+FpgGS&`3E=Ht_Atbk
z$2Z+9Zbe1t55%?O8&(46r+NVuGtv>YhSe=I=1nbSD^nyI@DXg2PE}LiMfYR0(rfCq
zWZZOD*#Hn_%A<=FIqHiYXBu#J5a{%J4Htl~_AcSu=({?GLx1Lq<L27cF}m;&epuvF
z{c7v>T}McEbx5%O+Favm*JbPR^E9KHilMSVSM|2%bDT9O`PIf!wahVi#ymwi`$go0
zbP>M;)k*D}bh+cIbaGn+%Faej-69h|l4`cOlrUG-)Hi3<w}L<Nnl!rlJgi60^<5LB
zF!ivtJOwF_s5c{0`p!QGeAk39a$g0nRpXlIyuL-$yvBHP4z~xY&x}KUO^m~x2g_J$
z2-Xft;VFKyN#7vtRD|&wI*seIt7U3#*qpet39s%FdqEg?^$>@-Yhndt1I(}cspTTq
zMo%ri$35YF*Us<)EpF%kZ~sZ{+K`Djg|1)g^MN&*P?RiMQd`&V&UDvEBfe>9<;KuI
z&8pTzaJC&OgoRm`M@rwO<n;Pl6URaQO@wQWe@jM*ALs;9yA#9I+{}oVsUdUttHR`I
ztoVkVqAmnTU<VZq=#<0<a;=TrkiA?kuXj$xD=jnnz;fT~U6kJgrM1_`F!#5(I^-Ec
z`gnL275&l>va9AxeA7Uu^e$dAzNvA#LEU6J^<QsYwA&W15ADL-!ciM7uT`I3<Y?$N
zTSl+Kn~+kbDkv##9i%p~C{)n(Ct!Q&n{o*rXuR57d~+W#0<OEq>t^~83I8diJ%3Bz
zbw=M(J(IZ3`J<XuFRtiin|gf1rKFk*zbUS->ud(hJ5?HyYsLtTt6#8rUC0b(6AtT0
z>FY)4ysErDVy8m{AFi_zP#={P4if~)>mAqG9Gth?-}xxL(GyRU*+9Z2lI}xh2l0^n
zbs@O?ev;~$IY-1RH5?LO@7+@6d$)0o1Geic!G^5b^)cP?=IC)uRN#6SXGf~RcSx2h
zbax>&jq8nRoFqvzWpJ84Dedb+gm_+3^2pLyed2nrcS*Bfyp9%hjd)mFU@z10e0HOR
z=Oltw4r+|g?Df$*4&U_2nqF0%SJ7D%^y1`P^-WO{bHjv{8$vaNKS|xkeK$C}mlL8{
zyCJAuh61angl;gi8)7LZ<kdFFjT@}T_PZE?mD!*hZ43oYQg3*&g=%&v-I_og@*9%S
zcc5}CQtGc~LsrqJKUomZL3j!MWf2~jC{v32#*1XuiD+>c+e;ZsRmIL#Xm{d$kQ=Rw
z-Qo&DVsA8AT}C|YjcLf&Fn#McI&1tku{~}nnU~;qDOyVi8I2OeaOl3y-p^>pHx2C4
zV;u`r@>&Gv=npXFHwH`3VTm1!;^SSvs;fV&UZjQ@bUpdk^Zicxrsp|KBaQbegIw<4
zBRsP=8vFJ4;CfMbk_`d}&iKxG5ih=}m2ba6fMqK;c@J%miOr}7vwiW6Qg#PYACRua
zH(iSO&33!^S-&X_+BzD~5+?fr%_-C$PQshg3Tcl^G`)ge5N6G4-(*Z*gjE++?e;6Z
z<rk!6%yUmUNs60%O5DqKp%YE2+PP24gwO8d&fa9bzlo=0dR}GdHf}M`B^r1R8ds~_
zoI1?US(_Qa*8Ap|sQ8X<4yjl)BK4b%z@_|dT~tatdiW%b_@=ImT?~hw{57j`>{M;o
zxSoEqb7>N#ODKrr9J)ou;=8VsoVZ469Pil{=|oBo$*^8;4ry7588aJAXxys?u;>r=
z^c?Ax=gFM+MBn-37dHJ(&5=cRnBUh7Iymg(oV6J^)oQ9omfmvUCx@HXZKg#dzO%M6
zn9Zhh&i((cj@t1}ZFh}!(ch-bn=$BYV^FMPe^$-pM~e97w{diapvjxT`6CHNtI+am
zAgBledSx~d1Cbk~ml|2dxe`y;rbaYc2|Dyjv$VX!>YvyqeJvZ~Rk&1Zk--$ASiNPO
z;2jO{WUZM4S^p4l<*quD^wtgMNc_E-p=dZ3yt>k?uxti>J5v0F;tWsMwv2^$V9Hjd
zwnKH%N`{%mQ0vKUr)Aw9>gQ@?=`d~?)-P{K6+|z6K8y2o$Gm{w1zePL6Ook=th?#H
z>e~}`@T)K9)Z&}Pm9bu{53Ih(S)+P^U>HwfciN%NB4ZSHl%!dqV-tsRK-#3eC<KQ`
zCM91KGDhT--jp44OvD`!c_76iSNBhHUF*V%H>y*W!}gRisL@bd%$G$WK0IBBGq>mD
z=*`i&iX43bPn|2ANA@Q!=u;IchJ+vX2Z?sSuHU<=Iuv(l6xUc6`=KLjuPnZK2!sFR
zd!wq=(al=Fk)**DdfwZ1{QNFEhJNmO>`$3VT<h=mJ{xt6${tFKKsnjf9|!J4Ipv4a
zy~EC+^c{_|KSghZT`sdhpqAA&Mqixi{aL@&pu_BsnuWemXYnmE5S-wUUit={6KYUl
z%?vo>I9M|R^i9tCKny3p6sCq&Heg~r?+9rOm{91?dpDz^>N5>x(hD~2_@?T}ZGLR{
z)b<K~c4*A&xA+m}B7I-PkaoO<lZlW@pNCn!W`k5NU*Wm#Rix?*vz!C79?cCTB{MiY
zL<G&C^@a<}_LiXicy8ZdvWMDj{o#Syp!Gq2KBk_sMi$*-Kj<}Rtie0FiZ>e?o}zEt
zP=`%N%~)<TKM7%d(jE-MN6V`vnWsXwUaLEdSG*EZtdhl2L<tKB)@guTby6*LB-1f<
zKbNUo*Dhz6r9K-3>Ef{Y{~_<qg5$i7e9xSh2|Uf<VPXO?F)@ML*Byp?QI}=ehNo|L
zK$c}gH*BlMvb!<S?rwl2z=A~*C4vx3w&gBMUSQdl;T_&fT#JjyB1MWtQY2vswBa4z
z;WhAxM?B(je*eroAL)B<--wBM94dsU@B7Y@=bSuw^5klpr35^c@(e14{C3rnd>C}3
zAi|pQ@#3sugELNKuzB5Q!!rrHycNJ<tIhZ-zv&G&Hsk0bM8;HXCJ`h8_;?iE$!Ky4
zMNTtnz5PLJ&lJ4s!DKAOdy>6&qEko?@eUFIr7K*i%98`q*7Hdha^fP+__7j+z|AkW
z^COf2s%m@vWAOTD3gYpv($0H$Z!WjXg5b~h23T_O!!bIaBvRn)-s@8{OF;L@`~Y(Y
z-__h(b_L0yp!xC-YUjJS&_)pdrF<8LWpQ`v?9BtLj^*E8pHt#|tW;njK!JFyXH9Xx
zCByh;zH>`JR+O{Hw)gs4?DqIM*2sPQDx&iYmIq8VY`Bq=PZ*Jxtk!T4&?6HYwh=?o
z>8gO1KJJE(dx4y4lcg1WT&>y2x!Wr~4R?X88+y1v)}`v*MdW&6ZP$rL$a}#--)p3(
z@s8v<i>)M(O7wHd*2@!uFtE8Az<Dl`+Xbq*T|_LI%?W+&eKt57qY0igB0LYz=B!|}
zVznsDv4CJ^)`$6O`B#=U_F!+FZRl7J_fcX~=R1v2Of~7%tvwraiegp>J)4wfY8R7Z
zovh#<t6l-@E=3MciEOT9eNamJp{x)xa@K(h;AVNpgm97p4|>i3TIA2(mji!fw++58
z2PxCQ;r9h(ng!gL@=58SPuf^1c)uc>mWTc1oSZDp?V7b_{Ek!@?{g%5*MqhcZ=Fnb
zvQ_w5pzpmeF~b!p=x^UwHb{i4*R^gvYjCI{>f6i`+~tu4NK1%vs6bHPXZENsj#4=d
zK3WzPbr>k>#@P9;q!QC!sO399-f2=f1jiorIhs{Cs`{fz(iB=lF~lb0W3vfbHme4)
zY}DprRP9IOVEY?edbG@_;?Q>jwpX*xy(sd~TWsagL;(DTAoc56Lzh!@xlyA59o^0k
zM1GZ@^uB1ASkJF4*<e2DI>y<%&tYfbQBTJg^iNXtj_vcwIV5pn{G5Y)PiAf59BP87
z&0%%e+WUOu@^wnjrKp9qx^?y?+-PzU67{kX1suXJxHR&*nyp8jk=OjKZB~VjA?V&N
zu;F*xFxy|Y#W~pq?e{5sH4d`-9p+iolC}N05F}A{W$jpb!5Qd=KK1>v<RNAWQa2e5
zBM6p@YjeL(PE+|^+u0?i+3b*|=<<zR3G+QT3dhuo_YXZ_D^Yj1;+wA*aB)7rW}EpW
zFDz_C@`;S4sP^nY!D*#q7Jt>Z!2>?L``L5VS1i*JnMfX3I}lTWD+m}rb|5vtm{^Af
zxSrq9y<(6wd7!A&Noo>aM$IgsQ{b$%vkO-kK~t1!Z>j3`Eb;R~q}DlbxBLVJ0pQ9&
z1)J3hpc|Z`D+iKhJce7vdVG?CljCmlK$4cXa+iqTC|bA{&y{^}MDYybNuqga@#f0>
z$ec6;HnlmQA4FSYbH4l@1Zyt)y<hPhvR0VBxkIxxqwH$uKUzCKOL{A*tPC-h<{Y@`
zeVFog3J5b(gm36&X@W@QW&6CX<hz))5OJ)RVIWn7o5&|6IVIwH=@CxjIfjAa#jKg+
zO<`Cu0vc@Q!ouwJY`9sF70Toiw2M}2-dXJVNCDY&s7V&WjX8e*NB-Gl)|Pr-sv{4q
z)66}I1QNY#JGxsXZs;BaL0Q4Ao~fWriJbyqW9o6;&Uy(T!XhZ|O8JUX87n<okM5B?
z5EC@In%kXGEUK<#NoJ4v^1bR+Hu#uBKed1Vlnuvkj6x88JeK2xddzdcavON8pn;Sn
zx^f`kc`WB|YM)!EO_ml=;x@B<XvSq5b<x_7#r$IU#o50UPx4qp)zT^7;GyA-OK*Nt
zlT3)=K9=M~ay}wH*m_nW@&LPXGpop5P%-Z$XWlOFxQTfVFUYSQEXyjPg~GudvB7+&
z`;5Y0h&>-Zn2UvbJvdk^kMl0;96MNcOpha~9#3*Xhsm4F*2EYpgRt)5%`6Yo3ta7k
zK{O%sOYuP_Ch#$xswy7PekO8Rtg60*uzV#;#0eQ!-t~jI^X54GHxK5bu(RbkxLrDs
zp2}?A7aSqOYi)4e#~`baeKR&ZU*;uDZ=`gpMF|Ly^6-t7@-pYK`h50-EZCD6m9I0?
zXsw&`3DRLeXlK)*!WJiu4SoUS-P{$Tu)+ee*-M|y8zRnplKt<Cwq9B#aD;8n=K?7~
zH{&e*>qcO9C`Ka^s}%4k9&PYY!gL5x_HY)E=w6Z4<bP5t1++WI4*41>vqAlkPueR_
z-+tOA55>{&h&2x-wIHA(OhrCPGKTyzpTrcgjiX&$pZ8#0AB`Q9qHiUB!?C}~)}spC
zw08Pt)!cKiJKF(^qNm$JwM^)gf^rTnIDE5O?%{>pxzmV0{Tfxv8lpj6c=C&gD#!9&
zSkAbp+Ug65(lBipjAaQK-6&PC99f&u&nq3I^zHJU@#@Yg!Ol{ACsh`hyx{XR`UdM|
zyHcByIW6Abhm0%VA1CbyiQ5-)adx2<*)D_N2qQia!?MQ<>D<rux53AKapsy&*zn^q
zA(+B)RO+U~LRRk-<xYCU%QhBO7MR?6`Hc0yqSGSbv&qr{o)<*tYCfKOgWiyIWd$**
z=b;(yL=h!uU0g)x@jT@T)tc6By;6V^9qJX`E3Eo@w&r$VIdz+*`&|q2D&MX;jEjjh
zc6K<{=Qy%~3pQ9i$v$ZKaM{*xwA$gEs;S3yc#a+Rg=o%NJ*&vogaxjiE^|CK59cgu
zF*qI-(5HRamt+b%YT_9l4(>z5QEAm+{gWB!9QL)4eLydM%0_J^n@6GseCydf4c~s6
zdKf9#X23kVm$PR(eox7j!CGaUm!T7bEpHw3IvZ{k7_lT`7354>s|xp!Y>u@OiT1$y
z)LXf0Iq4^}6~f6hvlZP<WU?h~YCGR~o*ckKN@q*0oW{9>iV53^)-X823?Vy(lF{}`
z1p=0jw%)q^zy7;wb=%BV_CY|mqXVfah+b><L}Hwxu-V`f2|UDP_z9naSwdjdq7XPM
z)5fuUl94)L{YnE<o))DlDfh`{KFONw1#3SMH0%Ledg5+sA}I{lpy-;wfiN6Va%#a=
z;ukKK>j92}E^a;%NFiBS#y5vik{kDS)v#>1)iBc9_xp(<H&bOsv%&Y5oht?_g1HU9
zKNkNKI}LU#pOh+|9sK^>WA4P^(Yfl`g!I49r?<;wzEh_%w)EpS8<MXhaXPbvT&hCb
zOE|G8vD$fbnpk_CKWRf6g`V}YjT0aaAsw>>)I(uZa0b9}wW_5(64rp5@ArlDq#N4>
z3JxWVnLiN2*Qt0h5pnPX1@~nVG@K<ubslU<?E}7c`-yUxUnQAW%t5)%*5fN@>STVE
zz!Y`}+Izy9A8=SWZr)(+=n3xYpyJVfAhx>aTJ01d(Do4@Au7!tgbJ^eE>kD16qD<v
zp?Oc7o^`vj;kNB8FZ;NofT&;z_Q4po@<5z)dX3JpdZ$2V;e|My??j$2u_qOgs`f!2
z{6mk}*avfW60-yDv|cqJMS!ZX=RfG9UG3n<*o@x<%Op7V2a`w~#NT_VJj~;8*U45w
z<t|*o;JLJdVsRyYPrXkU%nv3vI+I?=n|DCa;v>_?3FCvYKg5M(+<X?=vtvG4W`Q7@
z>=`<<!6zNCeN*xw9ey$gsNC<!tmB*R3dF2=(nkhwj-d8QCk_YkX_<_lQ%KyKPv$ma
z87^5no<criOHu5}e<8tOR)(4GMH!d8BIt=HEV7h3c0Kx=RveYh1iU2D5arp9@3RUu
zWFIOEGv|B!L%9fV!*=9DiQI6Lh;=?x%>dg}TkO1z<-1Tao_bln81-oI&(p?ZG8%MG
z!{&#4!ls)_z1z;ZWv+#rOCL%em53us)Y(Z0<Vl6y+=qPXb|FJr$qzF7L6aVR%!(b|
z%w`qgLJ@l~d$H{@5N_X`%l{U`(t$!6wttHy*NsOO>?~|9`S8ERfIWvhh<{ZCF6t_M
zFRM28Z;sv%zQyYQmVnFfoLQZWFK@cfn(^zQ8Ecm^@1d?#t^ejzjiWOhXZ)VtD{Hds
zSCTWiRVYruN`V>1m@_M%IMsD^EKeSlbGs_(@s-;#29Ioz;~FUMFaSN~dk=kUJD2_b
zJvP)XH0?ALYPjuVtleuPS<emWv#naxBSdb)M%!`haq^F4QJu_m0Tc0I8!zu%it1tj
ztH(PJAOjcX{+SceJ^l?!nLSya_KW{+BxonttuI)!d?RlAw6)3*<h&K2a8I(u>^3xK
zv6+{m;v9x)gmzqR7skOub9UN}q};~#JYn6^5N?Xuw~pkK)^8WV0zs)}32|#H*%6`*
z8IJ+sWm~K2TtZ5<o=-C5xndhx-giw-=VmEE*s`Z=E1ytxO2tsxmpwDfCWDfle5XtM
zDQK=3bNg`FwIH(%d^on;ex-RC{IIWba}ELHP!=Qb6q?rI4+oyOR5xAkN2+&n^@=!X
zwP;SK(5#I{bE}<qAa2OQahrE|Y&^b`llC5~XLs60*KOj%zFm|6O8;y!zIu?mUDo)p
zBhqdzb<Gbu_F>=OieIB|wDyN%#|F`H<CfIMEe3RCP3P=eidJT^F<-8xq=#ujr;OnM
zdb4hPlGC5kfn`B*U~!A<M+*xxGqw_SP;z4mvsUWhipK5W!-+nJW)&Q0qkNnrHMnh-
zC#*(y*;Y<cszv+`yIr7}fvoLh|2SQ64A@h#k_5S7(+@nA<S1k#eku)S35gZY+R#%;
zgr@P<Wx0n-eVB>!Hu6*+fG!Y$Qr7ab<ZUn7=u?FPK-Zpfq;VH38_!}0Fu}=sLJ>?*
z__j_w6(r(5n|vzg0;4;&WFx9-XG^pn#h(0EYh{x%qC0NwY*KCqy4j2WTa0(2vmBto
za4$U->n^@Z${)p{iv7G?0VRqV*S${kfsR>mAG+Bif+sMbSHA652)6!Hj&5$6mc@jn
z7i_Ie!#*Mwtw$+P)HbS_oaK~a**44Ey?LK)-8Gu9=OlYUd}kh-wiB~3{-D{@v7kPQ
zj7Kunfv4}t%-n2t@aez-oq(+H(>~L0c%u#Hdmx4B8Hx9M5d=P+%lMv0Y&1Ivy9N(_
zXk%GVW;Um7ymaTu%E5_xR-swvv}K?%5sfRD%+o$oN*UC6I#Hxa9KdEgfd4Q!t*VDB
z9L?>h{CDrQ#cW!v_gQF_drB8*|I^>tax^}NZDgkm+QX%GqlqU8@?5~>k}qkA^`p)d
z;;@wh&_e*aRyA-{wn)(c3Yd-3-{TluR<m;*gUj$&-^zCi=X*rw7Dw}T)Ikui$9A$^
zB!!5U>?1zcv(jG-e8g8~YSsonl6!#gAb6q=eWYw#*)xN?R2=?DA<;eb2_yN0i3TiC
zEuM%ypa?Ld<(pq|$MQ+$D$m<^J`smix5)s8yTa>j;v;bq6WV7oUQa!W=(3R&&^^n4
z{Uf=84?t&IrDb>sRH`3ILV}YOsWojep50A)7Q61HD81UjvTE9vv#wG`z{$jTS35)>
zLLK2Dow@<UYUk}Jwo2W3<;&{F7Yqo9%$M!ROX`NL<&!u#l$NE|KQ<W1MtMSCI&GU}
z!lyjv^;Y?Kc$(mM+d2KxU1>#+p?lr=NS@ueC@6m=0``xV<4>d|lnnzPjYWb;p%ohZ
zXsk7gE*tu2uJ8jMe|`9)KHE?J>j!Kk+mU7(!!}greXC_Vnx6AE8trfb<6D!ZbpsDe
zp}vhrX)QnaUH@p_pj|ZNvpw<AT(+rQM8wIqbqJocMyb0?S<agIO``w4wW43blfxZF
z6T6(-Tr42Fsi>Ry*;2lf13({~mG$Ow)zvkQfliEhL8P_xx&U8cy#gLWMe9c?Ibg5E
zdn}!KSo^4NmXTN6dbTQZ1WS}pVr0(SW<JSh{(0Lf-EcEUcwDBAeSs?oJKzP|$#>E%
zN!)Z@PMaS~OnyXRcIR#2W8U+JK4gO*%Xxg?0xWO5ZsB78v9dKUVoWmfvA_$tzJ|v-
z!ME~VM8Lw<5ry99#~eF$VP<S&@q4v%@9S+mT7z9$kaa`0jYb!KZX(KoLabw{^GMSg
z1u&fbtobp=Eks(@%AcGe?FvYlu`eb-+*}aLgt6~ZzSCJLyHiPfmp@i?L<)q-#nC!t
zcBXNz#M!G`N^r_W@PXdP64klYV#D6gW=}yT6qrQjtSkABMMG4m^QdQQWo%fGt=|EK
z@7jUqY$LyG+i&LgST^4awiRRD$0pg1pQy!br<%WWEVq3;*2Y~#px@WHV+>U9t^~1#
zbPVRZI0Ym7{7_X6wV1}Ci%!1tY#e?qx^5#McWjWxg)|klk0;8!1Kb|H(*$O2nrGbA
zt9MdON34MH(kyp1IkPDeNj(~3iV@jO!~^yOuxyiATI~tp#0xb(o;W-o?w`%li(J!^
z18RLdkNwxa&e{dc!wCNttL_Ov>`caUwv_L5wd^_TeBG8mo(nyN+0$O@lz}%J=pVcB
z1UyG|4ga{~<UYXd$4}H2wh|B=R933BQhV<d%eG!Xa8q;XCtk3PsOXWf<ELa1Tk%u0
zal16fA-EH-WshZxu{~ubcG(6NeK+4<x50cbn~Re+6i*Vq%%CVMPedczXCsR_n<-tX
zm6}iu$aA#FxjmNeMT?Hxcs}VE!7a43fk%iK?TN)gCRi_oLpW^NWO;{yLW$DR$amcA
zZ$Ir<M07NZg)Jw0)GQ>bm4ze}lb4%k+A4*}ROcRNi;KSA>`Bac%lHn-k8imYo!S>y
z*v?{*5F9MJi;lo#r;(^QAFv(WWBtWk{_wzFDNXLy0<IN^9N}aWPBe%zJxtq0+bAU<
z<HD$y5V&sURL<E}mL-Y=v|&5?WlzCM@8l;rqn<STL=s=w4cNdZa-|4rDp~sAC(4G`
z<_?AN&?kIwPk=nbpGfjHk8Oa56CNoQxUuW(=daaDqn=7--bSNK>@75TW1sK|!aJ1)
z$Kw;(E_9u&M^6&xO;qjPf_k{X)+RsUsQAx+YmHC%(qO)U*<}t$Nu{n^D_Td|!SuJY
z5iy>U!pH`GG5UD&Rj;z80_XGr8A;i4v<jido3y~}K^nSIN%GKM1sF$+J3b$?e!h!{
zuWrw3D+!jA>Cf9*G|+6!+Indjjo_4Rl&3us7;Kh?FqcJ34qCrtTiKBE72LAzfbo@Q
zxsmLAA`xo8OJD67U#GW<Q+~$h$JOV9&m=|N#c4G3OssEMh{VlqIN#~05j{i%Z6o>4
z?LCVeDjMRd$kdKj1y0?TE<=akSoAu(hYd6SjIR~G1@(CQ%F~dwi2!jwa=pm{)_&b|
zqKpR=Z9a47dgi9ct*R>-*a;P$vi389#k#T`wZ&(0B6j0nfm$wlnZk8>xvEQw2*z;+
zIZP)&o8D#JQjc6LkeFW8=oN0J{j4fA%2BtdMZr;khRoSo76=Y5E;8#m*b{8djcA=D
zGoB`9v($YTAZ``7++w-=@=>XWqArfJ?Ad2>U#9tleX{J!@VvnF+6F%9>pji+Kln*s
z?>$qP2@QQRrv@@j%Ail?>c5sC8Y9t)9H3>?>Q~8HDex#K!ssV`)d+t!_Q{;mxfz+>
zjej!X!9<$jSUo<^?OSk7^h7-F&+u=P2?D|~aK9075=zXP1vbpKk60^ut8tv>@PvHw
zsPC~%ivbu0hAhoeRMcsM%wst!zG2ilQHYcoU{SYB5`@mx@30HKd?#6oY#w^dNk1nJ
zxx`9#;#08}&EALIL3Ahg1_<{?yx$EUReC_gaSc>dLW8YnO#z6vqu~5wwi64&NzKLQ
zQ#p5pRnoaX@TsyuM<CUM@q~^IF^mo6x8qQai#A+J5;OOhjeIJXvPoC9QjLfGHu|Yt
zEsXs!T^svU!4~Zp*Ob|Ksc=ZfZS`y!qQY*PDAh3n;p#A1bqfz_jciO@qr7S)=W2ec
zh>2ceZ#ZLCn4|*aKKXWMRaYoeOlAm+pK?^)ubcPMrySkxyVsU0h?S30oCmkuC`rnB
zxT69uW{>BrSK4)5Wmv!V^PL+|I!H%sr2+;4%37HKO#HD~twwMi&p+42<_I>*WK6eY
zf4W)Pknj6eSxO0{vCP~qZF>HPo$^G&DVF`yzCCsI82EHq4|z36tjxMs?;_9b4Wtx9
zpY|12P+@Egvp}Tv3enZbryZ3fLs!-pt9`mKHVfglF`AVm8kd<BszV!#R*^JvGK_!P
zr~I!;U-D^3LkS?colnFQjMZgMnaL>i&_mWJrIb>|nxFQ?TkSl2&{|moCIRNH9bXWQ
z+-Hl`&=9w9(^ssXr2@nQAlPyYLl<zSXWBZY6QW*DTQ`~_PBm5qpDu_kGg0eTeLnv(
z7Q6YRYv^fPtLEz>45H^ovGoe7du^k1LLjhd+pJo85mLGp<E;_%Bx=ZG!gi{0;Pl|T
z{es!Ev6Og)LSx)?!v>zs4Ng{3`H2iZ8!NwSpA9|h)BUgSwc%%TsZd?IwcN<FIsb4N
zgecB-<%Xq{2`_Shj6Rzi{s0#6V^slIJPHtd*T$cXZ77KX#O*e|%ZbGGXMLewIf+<h
zqJTFQt`n2dF}$&2>NK)l8QySrq9t_V&sMW8$xSN;F*RfDEPRr#<msfKdW!+21ScH5
zQS=ROvgP<iNU2WMgIwXYTd!HS%*Ome0M$#Npfaeb`}xFXM)|f<2H;s-=WVUb%hW77
z=WJcV&Z8nv(3p*UCyN{$QT5k0O9u!OuGm%?Awjx0-nQc-`Y@?>cXC=qu+Xxv-cfSe
zXL8M;x{upHc_M-oaX~&4#O!r8^qIgBaYgcEL$zbWQ5FE7=MFs*W#9F%)jpF`O-gxo
z#%LCUkrqQ$J@sG&$3Np}HMM}TR@9%JwTXDC3`>)rNpwMAZjI94%a=HUn+Z|z*i!th
zR+b>CHt@ASQz-f!Q>I`K7pu`+Mc$_6Be74_-c{+T8DZa+vpxBe5XQY~T{gSj&p0~H
z%KEw&zuD>Y)-R(cMk&N;r93w02<&ESpUEx5p}$-C64w(*u19V^jBARM?&G$R0EzBl
zs^#Lu%?e0mN`b_K?OUY>?AfNUsS7As_bKwu?qmgB7r6Ah=ejNVq|d?g48+gX&fnYM
zQX%MW|Lv9vLrdjYKL;rvUh)w`4_{{^`31K^X1bP7AS_#LG@oR&imtzO@gHnF8X%Yj
zabv2cUx1WOEIFd>!k%Txk!Ln@m^2b9C;JRo{!PsaQp~^o-3@EyyRZrfd2M?skw?PU
z#nK<TBcf;pBvabo%K_j5tJ5jtz>ejjb@Kz%ycgwNoYRlrpa_0^QGs}0b1-XbF^V^i
z+Is1F53e`k^@H!W&1|1k6ezH^s@|lv+pb0`{1cTpYdcGMH;qshC<g7bzIL;mBLkoH
z;c<6Cv_AOR+;ay!Y|PMSV=?6jti;iT2e9GK#$w7W@I@Pm+PYNkwpxBVftU#zos~UU
z86*{FLmT^So(C+RWWnQELj?Lot4Cu~E+CjFeM^<iKolfnGT#+WM$6dvtfLb?wa6>u
zJDLHHYQbr?$6DD1?jM{B?E(M+VPyugSl}mCn=O?GWYe}>Km+}bTPL2Bn38qu7C7Xz
zrF)my>8c+rtgK%qa=xjcV=Gxf%qteAZvM4ay(7pB)Wb6x$6!cV+q8{n5LRGH=F6Y;
zHPn}kCb1Q3xG;}Ld+CkiQ&*%iC2z*S=5w)x@?{}2DN-G@@}G+%aEU!P_&MJ)4|4(j
zTvBjsQZXosCqLVJHuAaL16+=*mY=xIb9#(^E|H6B2!xpNvGSdWqo-{Ab51`0>##vq
zFO4cT#zerh+WEV`v&n=x%mISdHu9Zpvxzd!lq0-~I(`#rWv7ZXgBxay>dAkx#XI25
zqak4NmP$)-Uy8*d3;X4$jk>l0bKWU)CRr1xCv9xEY9D%Jy?9y(KdOFy1Je<&uH=&p
zcTU;b=X|H>3b`IYz3Fwfkqgk}B-<>Vi6dp3AnLcRXabH&^1bacghO-rW+%H)vd8Dl
zK3@*iFpTVR3=QNvlH+-5Zv~tf{Jbx9?Tt3{`Pjo!6hV+b?;sxJ^c;x~gt}KW<XSwj
z=31fAsNUof6~FF(hmB`9lE*0Nr>IRm-#NnsU3Dey=N*3zxejSEd)KGy)`(A2s4<aP
zWy^1|R{VPV+pQhN_k}(DVpb+?9h+|H^Nz^`IcwT-3}Y`C*!g^tF<kt#T@#`bc~DUH
zstH0-1LVq0-j6<?e2uLHP#Co7u#E>SDj@UgSx@>j#o4*Soo}PSgszJP{k+pS_xP9S
zZY#d7b`EQAZU>aR@B!Qj;IUZ8iunr(F#)<*@_{ep`3__12^$B$;1h@qA1A@k7YbRH
zf=ph@kmV@wxEo%MWC6wR4?Y0j!7l_GBeDBcwza#sxEaT2mX>H0FJs-0Z0rj@v9r=4
zkAESF_Wmhu;;>j%dDPo<FQw*0z`t-HeOkVdv}E5LD$+(79NJt(>HC7?;r-}+vn3t3
z#hRj>J&V#TmYzL5BRq683-njfV=w2E%q6Z_C!chwI%(YkqnvbEUR}+5Spo$#XKp|G
zA@><EfdZn1maTok_rhxDHMSnbSvTE^H?lzmeo#)+v)unTOP3riFWOc#@}_%jJIi|3
zJa9xN_dEIS1OhE?F-S3=H~V5vELsogA_l(b6SFX5gI{zo<_<hzLtl)&s&&D}H;No?
z*hqO#*lU~&zv%15OdPe*FD6CC>IbQZC~}sSAo!Ave=*g%%{c^E9)GwV&A$GPHc=`|
zyyiT?WPW>tHNIE~jnqT#8EwrkW`+qe{7mngYlacV7af~-Pd8l#*3NpcOk8u~3|h>0
zQl28#05Foq#2{u7j@xp9L*B>a1!7a@Amoz@J&q?B4<SnLi+MiH9(bSi?{pX5f>IM^
z(+Y{a7Qb<JV>@S~I+9P@MrmehX2v#44<5#Ks}$sBhe>C4A)a6U54yPR#611^Ys|jn
zi?Q1I;0JBsOG#=$OEOA*DQ2*2HCaLoRqxnH+D_25;jASU9JW*WbP~k`@iF5`B7y^t
z+vt}H4iOZFi@;d6MtB|^7l|>29pHG?+7<jX>hT2LE5P_fmY1Z8TtQ9{Ccl)#Xun6A
zX?!Uthks-*iGGBSX)_y!l7oY-s+;HOv|SoONJ+-<#j2kWrGo{wl<zP>@>#se=$60a
zd*D5&hqJkpj?~>R<qG2tV7>A%b>M*YvlZc$^6Mfcu$4an89!WZn>a7=Wi3mGE2%Oj
zT(tFAuXjq@v=P72!7SKjK)^*`=ik<sk|rw*y6E6`0;pJlCu}DhF8TvZBF(-Wi_fi#
zb9&&*Iju89dBdOK%LNxAm$<?Y4t?22zjw-pzw8uXeqoo5eEH4=6n7$BzSuAvKURzO
zNfoApX7sKGWdbkXg|U2hXYG16{^i?a_?(ci=l|0`tAE+i96x4+WI56kY-a-Y<&&7P
zi`FQ!!1;37nlZ3FBot_UIoDKHKJ6$4TFa|v@yj{cx&3V^CJ!^*L$hqo<<cMCQs^zd
zoSWdlw1>dxR{fLh%z8P0E^cF96a+V}fc<J_6@JoVs^k_}iLtGAM*hjxsyXx^hHQ2+
zWE9Z%kjru-3&<Z-Lgky#_RX@`SZ-mKu@$4924A<!GQlmJazJ{vQ@xXmncH@lE$8gK
z3SWYO<=g*@{DhwQZ}8EB%Smn?n6sf}N1*#&Ys1TlZEkyFBg=tp)zt}#Rf`@xqP~rm
z@_VM`z&{oZjlRjom-9G@i1|qW>dTJS8R2LmS`h>HgiTi6oPj+GjZ`Cwb7*4AW6dlP
zn^M*_$E;O`Dmvee;<Le9FD_S}a*=NpE+EFcUpB?dQS%8@Avk0#`<{6{`kZcn|Ipj4
zm%WRPziRyoKFEc59T>Ki(mTwPtrhS9Oj*8M+Z+Y3Dn$mnkp<9QD&dav>zkz#<V2!R
z!u#1)d5648_$Di+?R=*L!<SyiMnHPctIfVr2*OK{tH=g8@Rh{uT{vw+o(8|-YpnnY
zHdM+>YA!^A4;%i9FO6awA)z0M(sNjuS*?_oSs^g$x;gq4$Cll&^j}G28~LX(8~;in
zI5NZ3C7}qD>e-0461hkH?le&vjyw!LW0R%1>DF#*WPva>*Hz&;YX)Tdr}xZSE1;<1
zW%QNY^%08}aZ3!x`6W_>S5H@HVz~_UprT21qCw<Uhpk(_$bis#@q{Q@M6#c4h>Jje
z8c$Okg-p>}0_wMc+=c7aJ7um0c&PNoR~*k4q?XRlAqovW`c_V#bDgY<dv|g1#}YSy
z#L8jbcCw}}wDSBA6K`MjhX46B_t?N!eL)nGmsKwh(nkz^H4c@N2&D1$`>JnqcV@dW
zVk7Z!wez24$y)oWV+;H6jE#P^5CAtO0|Rl9`)Xkx@Su<iB%79s;H1@~*WlqMvbxw%
zg(6a9@5!q84T?7c5`<VBp3~MW71;HPB72I{s1@TQ3YN9w=Leu|i&Zmc;7*tFNmr{E
zY`H)vUKSU^PE>p7;dfg1tB$<;NoZ7h#s!C-`!VW2dy;8fi9!#*$JP=aNo(X3qWNAg
zbD|)$%#uP2Zj^T}_1DdWFg0Z?K6r&w=-O7iTkYUa0}Dzp56p(5H{xdRWQ|af>n{$C
zU&|F>;c%gVQ9)Ck?~*l!N{}e)Ye~8HcskUfuf=*})TQ<QnlGtmA03I`NK#*GwP?t>
z!bmNFbxCOm4_n;=zLvX7bGeUe&-m9I#}p@*q!{tQ6rux~sKCM~*hveq%?kKZut<=M
znQS9r)2VXRnpr~r7h(}5?QMN6_6k_s{+cgX?M=2=07EN|+<Pg?DB#%g*AgSiFeP{7
zPC~^cdAD_=+`oK)OoOFk+_;~zem-gcv8gLSpGP^Roq5gHq8&8!xDT&a1?2t*%YwjU
zBj%Fg#LJ6*Ry&inRn3`##)G%p1@t|0)AK08qGwcv(@dy+J)y@99Z`o3d_AeEoW=&f
z9_y%t^IaSIdY)?;S&j7>{<_Z(!K_BU?u*PCA%yvQo(hmxPOP}Ge?5=nv_xlZqo)NC
zG8@k-iVClI@se=Wt9P>52nWxJ=w|iPtIybE{_fm+pEbUoh%!Ah!y%7YroiAfo}r~}
zl|u3>K(J=*9G|53*K9GX;EU^;9uMJ>?>y)cBthBVm&;Hh@~G~4Uysu#4#|)rIa6)j
zn4s5__bOWwaTU+shl%Ka-7yVYWm_p7;cXkf8Yn>%Jnettie(A6NCK3Tvj2?|tDOfI
zh#9b1E!PG1=~e<2d~|r&y6)b#vm8QiOiOo4yYiHayOP;A@=QJ?Cy#+|#CnEDq^Lg*
zt);|%w{*x7kwHr(*!JH@njrb_K^ysok9GDSel*{3teJkyM!(@$139SKShi8<isCVh
zM=zlG5n8O57V+rCx^$xYh&PW-MpH1G&sd`Z_&RC7#5fbDHolWtuWhaT9@=<T1VY~~
z-HV`6N6w3j-^k6en_C-F)}`zbt{I}Y94)?E;h;MKS|AoC-rvYw^aNY1S0F<|-v36i
zPe)1?dFNKjOcK73n|?f@N|zu0dKodMoQPvL3OtAAuolOH<r|eCd!7pkDS5X_`>-Pl
zj{<UTSA)e<0mM^Cvz;<nuuUd9O5|bR%w5ZMUXg>PdmZ@ZpU~fy4Sq9c1iiZiy+hwj
z3bA_XHf|c(1m`?P)c)5-zUfmjg--CW)xLT6y<C0BMoky%(Qi5?;OB|CQU!@i>*W4o
z;{_7DAY8gpj|$k69I}Z5)qWy{!|hLgGjMI!L2Hz@7-_NnZI-sM6Wc8bXDiARn917Z
zn@ge)W-Z3I2A+{i)nHG;i7dy!VBer$oicJMoY<TSTGh>wd#ur376_EnKEx>Po4I^t
zv1}!KP`H6(wpPKe)Fk3U@#`q?DVMG&=SCLNc_y1popGDli~^EewXFi^gOUK`CyB2v
z+fG*27EmTt1q`-t`I5S8;fxJ@E3iUVy;nGR&LZ#_{Fd)uCT{3kcdsX1yC(NCeB<!9
zat9y`wvqDi4or7y-}044$F#>rqb~^lWhmBIKIyo*ZsXrdRP$uAoc7V{0}t3l!sJw)
z$(H%q<hLqJJ+Mdl?izPqAgJV?*UZ+H*05`>Z`~gGcW$Tu^?h}<W9<Sdcm!bxSRTSB
z`CH{&^T@2Ew_c9HVg;*{p#oL@EzVe5w*Z1F<hb?9FlCB|jHO>DT-*v1evR%{@*RGS
zIF{;^!DAp=S?wqcwXK&{gtpunHp=M9+v1Xz&48HLom<&G+$~(jcB}>wai?sjG})Zx
z{%@Vwk@yJNKqpTCM9H$rS02j2POcBM=}-eEzvEc(_y5U;v$Sv~%wKNVNWK&OIi5U0
zE#JxRQ=to=7afRR56%(Vd@O$No@OG)vn+A4xZn{02|ZrDYE!wOTvcKdojkc_6bx!I
zU?SdEqm!o*X7ijz%_v5vIx1Torx$ZGGuF-zIxlb7VkrhoK$_A{;+O)C&!TNwj^^L;
zpmoYnV3ftG%a5sBx;%`jG?%*QWR?gm{iyfGH`q#cplP{kYn|LQ4*=qN=?#wQ&@{f>
z8(H=wfiTe7<P%~wHu6^Ki;Efu!*=v_5UaMGDCbE!aR+!V7Sf$wwBaHA1JC)wHK%Ow
zxw52FPPB%ea}>bZmpI$c1*KPTkmEM;+}#^L1gvpAs6}h<NBb7F7v!igmbK+ub;ZWx
zi~N7B{#>4jk*885Z6ZF;D#!d}0t~%k-KBp!Z;fb7F{#mqH=pyxBSe<9s?kZ{tGJ<r
zYtMI{NEG1*hmI|lv2bC+B%&%IciM=VxEvs$3P-A2X5~P{@9Ji4-EOW~uj*5G3_?9&
z(%8=uCtp5_i|WdAzN%=NF}hnTu+L+QYwH2~YUj^wBilJiLXV@ggMEo@#$0mpZ&iJs
zhN5o=bgP{@9?;RmyWke=NTSc%uwWWEa_2bb;PcD;;E~)Ms2DlBvxFogBu9m`KWD>7
z5(DO)1F?}TBg!S_;TZT*JL33q|J!V|boelGg|RHEhGZO|<d5S=e8;$o&($pTs5^R}
zO+;OoAV8Un*ROw_HSX$A@=bJp){Itg3PZyqoonkz+01jKK(qD{M^jYI2W+u?I^-Ul
zOW9not{TNzTRsxp!9m1trSz_Sb67NY%jgO0flSwWWqj`=??ykW5eVK^0^IlhtF4t8
z-AyzhThHEfc|P`v{^IEoCD?0NT-jzR|9~tox3auj9#VpQ6t*3MS&)IwPSzz(IA82&
z>_D!J2ptBF`Y387<{Ug)kqvAjvXnAyatt30Ebx<u^-~D$ky3m&(uvwppS#u0oAJ=f
z!of;JaFVg|4h{jUL>P(_Hh$C*;7zZ!dVUS69zl>M0<bwO*(a;YFSuup67x0U^Xvg^
zUrOK5`c||#zhLe7HTpVREY;o4fZR`*=qw#|1a*^evFOWL#+CLWXxyedQ3i(@nP;-B
z(a&s)-@WQvcvkk0-c87uCO=?x{aiVkC%e1kirro-@Iu1ZkKW-oq>{9a3J@Hg0OUMZ
ztIcWvJ+_3h{~WQcSRV|2_JX?GF&hVGTeg#s#8L?oH}QfD*s<KUGs=ZE5KlS-X>RaX
zIho-Z{hoBrhK}W$zMfQ@@tYHuh0mH@G!TRuI+j?oN0BMA8M?jwG}Pxjn|JJ3nHtGP
zJRm0Hl3IHFm@k5drP0{yrKz`Ks5KG2a4(d_^<?Q2*%+2j$CB8?d1D@oEY+IXv&T*n
zhs1zeQAsLv2z<M0;xZ@Q;vGn#))I<A!J~v?z~b1xT!zmH-3drx(s2ZW&~&qQ2zCRV
zdr^K)!ukF&-&ML^t{ihLfB4<DmX+L#oQY5tko5u-w--b)`J^L<JWS==%a`GkwqhF5
zgZvko5ZlgjO1yK+c8(=RantnRKX%*~Xx6n$1INq3t)w;JIS1EZzJpOja}kl-hmJe0
zh<k^=4IfV&o;vVEG&&OBHE|;4)%V?p2+@Y6!)Vne#3K2$v3$qT=}O59T$ji5on#Xi
zIXsjat$y6`?%)(B+eGPygX}q*EUiMiWd1>R3Ts5^`NwdRZI;s5g5I=N^+ebjYajO|
zIehrAEyj%CID*&1QdR0axV@Y`h?i8XFvSz>l+o=+KQ}8|$ZiD((i7k~23hS@?;?kp
z?!x`!dB6~;iY>QNig<Rph~cpoV9m;5w|ov`0k<yMNa%nE{Mal{uzA=@709^-B4ICs
zdW0}#JEh~t7L+~3PUOx<Ry%D2@x<AB6#6uHB4NaQ8A15aiNc)9n~#gc@QJ_}Nyc?)
zI4`yQiP+L|?K|z`e~B$8XWm+tCWigXM0?}=RU6Iz<h*>&#!mQtxaT!Ce!^EoYVCRf
zV$Za^O(zNvgg^HfKa)}E^+@O{;CrA-vvdbpkOSx$?fMGfqwd|KC2W`W@pqe^vBeXP
zVJIlksa1VRluhT$0ha8aI@uJuRF|B~<-8kBDH*y-_fGikn({Ce{b(1X4Z;!?&|KLG
zRprE6OW>tuxoqnv%C2g{o6~}AL}N;}MSR+9Kw4E!0?!t=6$SP>6SN(@5%B@T$1;j}
zR|vwyoh-|SeF>NTry=qKCo6O$PO=T2%%zqXO!AMR@_iriRfzn0(pT#BM4&oZSO)nJ
z*Ge&+i^$sJ<7($0iGE{aCzJC-jLn_V#!m*SNX1OF50vBzlB&pV^^?8`<O{%&H=%XG
zVt${@?q6$->|PSpOV*6{_xzK!vR1KwAkl}l^PRGJd$?NAbusI695V|`CmrqP6oD#0
zz(o!cpM?<7LM^)4j6$j_rhzC(FG@(AhQ^~`28yqu4zHCm1fu##wBAF@vKH-Qs>403
z%-a1p&mFXl3eI$~V54lFbX<brcmk#_>Uwyj?VLovuR23moDX-R87V(tVI__xM_)I0
z`|f>nSUGon70Glr*v(xV<{~IJ`#IG^-CQ(sqKQKpjuw0r1Y?LNMYralt#y6b7&K8*
zy5%%c9w@SnAwe0-cigbKfW!@ZJlj)vN&?koiNc`}D%K_n%*6D<*{t->f}OHPRUTPl
zvJTCB=d37nM*L(e-$k6XkS?^Vmil1sV%L}cp?AK`ma?#W)j;SK)oZz%8$u3+5{GoE
zLJBPwOiMRg6ul$UT0S8)z!0wAEgKA>DWP3s>fd;;t=$D~?ByG_9#yfi5gpmc#i7}Z
zpm4;w7;fm%%KZmzt6PLb9zh?$#C8m58mawG1*-&jx;MRGb}H93Zag+{s;t-9%93XA
zRP60jxNKdsp;I{p`=@1PfXCS>$4Qt63E#6UHa&18wNpNuvyXB|uNw9+_1q5#x|)SC
z^<`g<ml7TURXg##)$^SPo`*Il9M(hud1{}#15QRKtDVQJ5unI9jj-vb0+|*4;=>I0
z(#swfL-fDu_cFV^eJadHGE3CmLOEMJ<)}Qn9~bVW_>Fu6c}%ul)Yx~jPIU@&ho{|5
zpc{QTQ}&GYs-ABlS?LG3GMgvHP7c*Qg25yN3$~W;q?IW8t@Ug|=O^C6r+n{@-fNqu
zV(Fh*{Y3m{Tc;A&X55@_I~xlt%?MPzQ|5P77Q1pEA{Y7TL<wx@2x?#hr*px%xw>uF
z;OW2z5zt#UR6cug$lUO0$A@N9c3>l=F#>f>!}Zlp2S#wjj-C!wi%lL|XRH)&qTWTh
zHhwyXpvVER|53#ixgJZOd7H>)N)*8}NR}t_of3G<eODrmMu3HRnMAau5}Tk130tRq
zDj#WBJAiyr`u8lT>w*+K@V~Zn+K~eLr@ye}`2BX>Ksp6@DMUCgx@Dda&ft9-B|*d!
z){ieT(1d?l$vza4(Jva}0}6m#?_OX8N~g1)EgV6rypi8!<*tYeikG|@leK`xMD${-
zfU|$<;G>Y@?E((0#Sz=7I=I1Mq-!-fy3hE^5gG(V@W7dzdMuk^RtC?+qU$nv!iLVo
z+CfBIb}U1s;WIveZ@_aYzsiK@mh)VM=gfDhvt+0**yx#@$afHNDQX`=09&<v5|)bp
z*$DMBfs={cHW8(vK9gr0p{8bLtdSs|z?BSJx)=;cC0E|onWT%PL&8iBv}qT3Tn8Iw
zNsG|{;vdOqmP&I>anF{cJWED`3#{GQ8rG>GB}XU7aUG3FH@ZFS_E5b5^WjIVA44T-
z?gd-PYR+sh^~X5f*0M8QGs)#u0e6es1?W22XB$-srrb8m>O8^Cb+2tzgT-!lyOg~B
zG}?`w@&tEIj@q+vHcJ68aMl;>HGg4)@pj+b6u#tV%L#BqXX$NXq}cFTUwmbW6^%ce
zm?7=~C4BYhMz*8n5i<-fTw_sH7dMO%#<Ov@BxhJ;9j{mK6gDtAoG9-YLUB%=EDg`?
zRwjW)77@qLXICj2ck`?x@M`C6tY52?-?dM{pW3B7R{-v{=z)j;5x^l#kTiIybn3jk
z$kOw+obNz%86V;~{<L*s-v08v)-Ch$e`EG)y|a#ru*TL8KxRDw!AfZt6@c5@GZAY!
zEsBVUk!?IZf;T6Nxsl~v#>DEMfoIw*BZCvQt@5A*_S>cPg!(#Sik(sbQegIcP6~E_
zHt@W|sLTL^&nKP7I$b)3q35eT&75mOFWd0*#S&vy-s7@99m!&1LzJTMI!AJ?H09jf
zH5+}tV2CK8jTH#?(Cv6B%a(PT{=8%1>)&J(*|Nuv@dF!Wd#PHdokl5ufUekMH)BBZ
z18qf1Z+w%rOaFVOp~beC^&+-<+S3;>(Aa6xm!D4x;rB~a1lGx^;_lRaK8W8EI;kb=
zmF8A4AqQb)M&6IYN*T6Neb)UV^rUO~E`ok&>M*m&cMt&N`C&}8ksTdJWk&emc$z(Q
zNDyvS?aCP{{N%Q)@-i~t$r57iSV2S}$IZ_9f-9ivz`1g;N8pRWbL9YzoU`X-a395S
zO}624PAVQyXzY>ry<2e9&IMl;<{oZA!*<dMFnTUI1%w&1_#W8UIbYj9<0gNuAR)HR
z_3T)8PJ-dMK42o>$)rQ)k;DR%(d)f6j8+@x5~WCA0QqiK#TEI1xF+1bTBTWu4%s)H
z{nkF0lxvS%xE8ZSupANup0cIrWS0W-EywRyK8uZNr_9z&7!Gw~W?ly@z4D1$9XEOF
z$KV#+>UO0XDJEKLF#&%&V(Wi0YoSmMCT}C>mb8YpnE<+>1j8cpn{Abmhlkj9RU{mo
z#F6YI++B`JJMSX~gSA)#@x*K>ul3;hSn;$kmTjyJ<vV#gyQM5tZ}@!KzqnUU+ekJ7
z4~;UkcHS}SA2>lq&nHGb(7@1?jeb5?8J)C^pO2&FAkmxZ=ksET*i9joP9j1Mj&Gt=
zIlu;=j4IqC(W915VE{-Jnoy#raUMzPh1U6U*&$0C(YFYnSe1$(cnUAHSj~r&^-D1_
zxiH8aa{0U?=mX={DH8=M9kXscaix*dX0O0FgG{7f3IYi#w`gGDQCrLRvK~Hd>($dO
z)?_2bbx2Nfn+anV`+pHJF%4VibK}56<1qkHFFo#;<fd97u^D0WWWB^|tU6J_^~#-W
z_J9XI8SLdkkgwf_qAZ*0g4C<MT-zDK%CNJ0ff91mkTs*wsn*M7mG%2x)HPWKfiy;g
zag`qH#fpX$MNCXFdKu5s+GNt?$xftRT9nTfy-sAAu!hub$1x(ol+cm*G5I7x*ad5r
z0V=tlZ0TF2FsiH**1iKE+^}TUs9t-HrcEv#OVww5T9V!@R}B!!z&bt0)hEQ>bW2|c
zr*rwUy2z>sHCZ66pVdVK$t2ZtG<xHGwidts<-N8ZLw@Bc+`9PRHhSp{hT%0?bc09V
zjAqa$OtVY2qNBS#Pi#9Nd*w3>bEiPonBDJ*rtCtTOLCni=H7)u&u_*uAzvD--d%H>
zJ1JF$@|}kQPx;R91)r@c82^zAF__m`Endz_j+1qigY7yvh_jlFMep3kB<&i%;8VB@
z;l9=5_iE>@`)nd%l)M~6*$N&Zhf9pD5gqO&ruc>2(P`rj6Bo6ShlyitUvRXZA{IZy
z{6b<WA}QT>mdZ2U0P+M~zL1FCB%~L|dDK<VhL&|>WH7AOi}xOM&iZATM3j;^M>*tI
z$^ba0F4|ho&TDNwDk^>`26I(Kh#M8*3T!i{52k7B!fpE>a8dbh$Qb_r^v<>m%1l22
zDcp&<y#Ic)i-Ed#rFyV|i;jnHxYq{rtJ^%l7)q``bTLq$h&pfz!}%m$;;4;W^pz*c
zhF^qh7YnHh!!@y6Az&UY4MU_}fcQy2F;=}(w2&|m9>196^nQE{>K7f+aU+?tiHniu
zUe+ErX#jt2{)VVPsiP+^`nEz^XpI07b?gDcT{p8A*RLf>YUR6NtYn*Z+}ag1+!7rj
zkJ)0hG>iQe7p$cmIuw8tez}x~Yz8^MX`KKV;cvsbWh9({){A-KitIGCAHW_Yxb4ct
z+*`>;tX<5FIbUM8truY3isU5M#7nl3@7x2JDVEXiW_13~e}(VNR_Tuo3WL2}L9W<=
zJ7s=2{in=c@JaS4ZUZme&2R?qrb#BF)eE=(C)w6@oK!;u<{Q|>4ZV<O;~}M`8IG1X
z?f2No3xQ@a>s{8H<SF?M2T8d|atz+E(bB>LI;_T`1<Xo`Xcu3gTrl8P&nLJmq@5Gl
z5=w2yCP~L8Ur5ZE*|%Vgs`yKsaLv-mj7VmyfF|!}(2neh){e^X|FSo!Dj$VITY4cE
z2(K3fTUxHZzQlIuyx?2sRex>Wd@q)Z&<uYvlfh)l%!=T41<psfB(7Ajkhk3#j<dE_
zpgpXZH|yo8xv*=WZImgL)+SH<%?hAPOw?9P;qC@0=C`ZT7vVVE4&Qc4dq@<}OkMKD
zZyef>km6FoL`hwcLtV-xgu9Rrnj#$xT`IKT5(1CmctUqO&5<#3DNg|mZ;%4ivZxqv
z1W=cJdrsqnJC^OboL{6sMn#u$@#aYhRF8*!ep{TlRQTBl=ZV1xb;!?Uli7}RawI<2
zrQ8~23>dpf6w!>Puqor1YTW^MHz!;*6dJI_Y)qU7CYoHAj@wdHo<5*6aJdvfKzt6Y
zK~QX+(gb~Xt$V53A}&M4SdrtZ7Xw#(`F;*1sSfu9+2&v6R=-jWMcVja4cB66nD|q+
ze(CoA;k)m@cZAtSnZ!E0)n-&*?fl)}+t#IITDj3dn~+;&e{o56hf&)p6_EJYWyk8#
z*V@44vX3Rp@e8yK<~vCwlmzJvQnjH{4z5L+W3}eP`A*JS#~BsT#cbqqf<ojQmk!pd
zLKLH-5<rYzcC7qIg}EQg?_o<4*U$0GNh8T2JIyq6>nYts36=Z;?=YK)Ry<VB<mJFy
zlqIlgjTqj;@33Z+!Bh%H_j2O@K0?!?J#A;5ldE#>S-hOstsFd@1=(THNX^SBTP{Fg
z`$6&>cTVd>SNFf!x}|xkv!txP>^tHSax3<$iipa9o|D#A@}0~cxVf(?Wx`rC_Ru?Q
zJ-^CSL#lnn_u43}xUeM5B{rkR|2bk?mwktEPJ_(b)zDt#B-<%#LR=UBY%e-!v$zQl
zyjX|=halD#gD(a?M1-$!zZ!b6Om2@;eK?;mq||yxQ&W4UtoEX>>T2iyciL$Dma4+W
zqR1ig^^KR-B`;uDvPrJL=zvQMjDvSP%)Vcvn5;_6zeZ}OtJa9d7T9^s7afUPxOTpn
zYlqJ~v8LJqAt`(2B(f?`q}stBpo|yy5UFIA%h$(cz1+FeqdWQc;TyUE5|-iIa(mSn
zkxWMaN-J4aBPE|$3y8~FPLos-Tw+VN3IRxT<BhCw$LjE`ZDx&=D6EJ&TmaK0$F_6D
z_RZQ(G!+&jOoLtV)j?B-o0|<>$-5DZx~#P{N*jzK9JSXGK@a5<c*wCje8q8Lm(KJn
zxBq9SLNaUj{Ot!0(LwEsudA#sY&7c%G5R)kC6LA0J)LfFjN|#vV-rFYB-E@&3rL?3
zO=b(qzKwb*ONSIkgb{1Qfasok{>GZ|bB^o<t*FPPqk&+)eZ|-QZ~np-uOz|2<j><m
zEoGCQ*PhL$6fnyLC~V<nb&idJb*|*Wb0^=bFj69m(iJ*}lh(U(2e}?4VqQG5SHcn@
zhDMuQXsTXZnh9TX($@2x=bBDlc_ZJEMk@6JI0HsIhZS9StMn+dcRcxX#w1okKvjSv
zYI)PFpIeg+h0RI7S`R-l-Q+(@C}aXb<L{J^S(rN!4fk_^v-_1Ub);WOKq4r3E$b*s
zdn5rMgnYgec8gC@EP}DVW6~b<<Ky`mt1<7!O;7?5ZrhJ*qxTCphah{h6i3jAsTqg!
z32XEn>zEH519J?rYq@BxzOVgX{FSw%5&6Zk&$AJ!J)Q#@Z$MC&qLDeS(zYDK1WA11
zI%QsPuz$(AWey1Cget6;ZJpvaBSRXJcUwOyV5}oS)=H_c%k3T4%A_G3R^Sd>FGFH#
z+-eI<+NiozUe3)@0x_+)WdwP)RlOrjGSe#A+;+YT^Z&SQ?ZmiNI}g3n?5a-``H%ZH
za5V_~Jh6SQ24;wCK4(K$12eQ<@^z4L)o>JE*w32hXBm;|rgo0&MlC+y578XGni%4T
zx{XDTo*{kwYMvkRX2hwkUbS=zX^mX+dNz@u=!hq$@zrvNJgvA+lAKabQ3V=VHw0)A
zJGYtdq`{GGtz>+ytBE_~Tn5`$lUG1NLCk*+-o-m0NFVD|Y3Zut%zyYNTMppmb7P$t
z;v9!mw^Y(mD++-at(S10;4+HmS3Yqy1wB^^42Q61Sj$fe{)fu3dOCx4cq95FWL3j9
zOQoqj@3XC{7%Yu#M=>G<V^COqd4coMuKCizgrhjZDo6x<Al|Wo#TrM_tHEozNQdV3
z;IcJztx$%()Q<Ud0+hOGBk`oPV;r-!Yrcf=QE|y1y_VDa4)A;|dy04_fqmnpCmi=A
zqPi9fs<Z6*G?>!wvo>+f(Q$4c%H?dwBXQ&{y%8Pj(u{DJZC;)aV@FBftx^ZqycgUx
zv2X2YavzSUi`SA+=?<*m#7hCVjyzIC#29fnnLR+<h8({FTi>#7KAnUn60j?~@r(|R
zV#lq2&GBUq_xI@V$tT}yYgtWUm9Qo*(=<IZV;lKN;czb5W(*`(i(vEBB;Km53%BDP
zLX17(NDqnHDRANA^OV{3T=?c(Q{r>s_Q3TVv*O>%GGZ_a0XBDJC&X86=z1>T<FivX
zT-9*@Aolw8WYgY3D9!7MByOELdfk^xKJa7LV@*X0p5qY3(+R#t!r<|!{tt@2UC+*u
zS$Mq33*1j9vPmL9GD(wF2TBdyxSl9F?}*d9p0wit-eOws0@-`tYwhdF$4no%<B76(
zJ$0z;UXCisN|&O})B%p><pf7ETy_P=<!S4b;bTJAn1ea&mdV+LM57nA;cPG?j}LqB
zD4nvEG72f0ZLI)`b^Bp%4(rvV_c^cRu}M&`Xv&$e&AYxUiW^B+ZL0$AiqgAt=RRiJ
z1-K@wEBVMy7E%nG=aSmlOOA?f;I8&kA>xYb5;<lDU-A{i&|-EU`iXd=<~2oed&yTC
zwrkWzUP_!G{fgCIiemzc`O$3CjS65874K>6rNR&#Xb?bfJc}Ghk}Crl`5e`=BbiwB
zZK5>(4mt8h<9|=KyGAsGqQTkFW>&(N0m(|Mj0QeQ_vGtLQaejPebJ{a#uKsqWDB_z
zqw(Co%N0oE_4EZeX`Ps-`h(WZuMu8XszLH(XUme4AGLn8g%3XIqhCraoIT{4fVHZ&
zTgkhmVEwNCE-<?aHn|bwR8rW@(%ikTwXFo>5!=o#Stw}2PBpX_IsENLE*ms>!3N@q
zV+VDE4c^G<N5cb#58d$ff9qRq_=c}1^8v4ZBX9vW8)Li(1YV2pXbdqFzTxYQ!#CJG
z7ESykSwW)6VObB?v$(d-MCnuKA~vDrk)U`;SsK}~awj*-CnXwMuvUUDBe<*9uAsUO
z{(9niTg-PSF+h<GtYQQ#<vZQ35$(Wrz@ukN!i27M@=1<Uh+e98#h(OU)Qb}6Ok_pd
zFXfp+6b(0AnEdZaQ*12;^vcun&c2cBzu%pfZ6jNfr0A$^-bjMW^_YB&Fy|F0oN=&*
zr!g1XuHGrYi}S>`6V1qgnOb%;mMeH@myFp!zEe06_cMly4BiZaj2uKlmz$0W5C<1E
z4c`o+T;xkVY$n9#atptiyZfC)JgD8w0m;e`R2eOOD|0nlTuQgGn@LF(5n$S*{*Pxd
zF^p0+Dy~~S0N{k4mSk@tx{>!Xc`|R7nLz@rKv9kIVjg>CVjozu)ZXPu_*ywOISqHM
zeKXkBXE~1-OF{0Kkf){iL@pJ$_NYX7Os@9U$r;cU0r`QOG>IoXYu$t%Coatj@*1+F
zmk}XFCZe6Jm8@TNMNF9ewsO-q7glp?rK6^DFt3+6!*!ghOa}SfZU#5QL~K@{72h;O
z%v)s*_7K*2JIAWr3}_qkN$$>sU%FM+_$EBkz^&Zv@G0;>gST=eACP~@Er)Jy&W5uN
z<cnn)$)_$1HM#n31#XEfDO#5DBah}gT`SJA<^(>qv0Fal2NgnQJX;Vm+qZgw2xzei
zzLofc^i5&cCrjTXDm`kAY*~XA*1iOBZC3CcWkjtQ0&%Z7RonULIC%li*kV<XtEVmH
zlMc{hwp_a1HH|A3Yn>gPWC+qSDRWRa-^m4BhC7l>_HM<p?B)os8a$2x&PwSFONlGC
z7NbWfgy;ROyfd(+cM(AcrQ@x#wQwbZoFd3)Y?8@s-byqV`=rEFTNMmvxwioc$KiH0
zDWt&W-8Eg*qeFszA%<V^sbB59YQzTOH-r}MZ6%L^eIgP=E56)MG;O%NZzxO4$cp0x
zzJ~~SYVm-S4UTuTl*7yt?MivttB5V*E4hprW~9F2h~^=@UK+890OQ`;y)~PR-()Dl
zIw%@p!j(0oxsvn3CcDHt%}&_TwDOC@zS?k7@r4~qyVMCChHmrBrG&$a(GLMIInJ{m
zIqMLZql~JK9KBOOfP*7bMLvmWawn^vx)^@-qrq7@-ma7eB}A~b6~~GvxOJ_s6xy3c
z!5mZ@QSop1XPZ$m1lG0{Pv|5SvUfWs4U2dp?sGGXYFI@=(!U+e>NHaLZdsWReB0-E
zmgEVpVfl9KbR8ux+R(Rsc^mNLRaI<-sVa~Dk$mT=sF<%?x`Kb(2SMy68;zgSoYiCB
zc4T^7&Sv9T%Oz*jlCGs*dgUM`183<AE<~;}9(W^KaQ#c;+qoBr+CPVA27|Ep>IAp-
z?TW~wK!js%e><@XUu|6;7pvw)Xd+PcQZ*C>Tn&Hl<qFW3kPDnxC5@(a0#NB3!TxT7
zjr@k&cCTue*x{g;!makJfe~-qR${KjBzd~k(ipRalYRj+{ZThcbKq&oB(oXgd}SHY
z!`8P0&+m$y&uog#+y1t1K5~dP;LpDuO6rK`q)C^uN;;h|BD9waDV9wJ*CuGgz{`OI
znBQ<c8q5+AF<rXX8#a_BFsad1{bfg!hu&o)`JUShC|j%Axdq7`y{jk1>yV@R*vmNx
zc8WCCiU=rQ8G<)tX<5L^IYF~oFOYDg5MrD_g<RfV_Q8raab?c}Qo>$f8L!Jd=w-(}
z()My)e>rgjojmz!QLrTBZ?|^U--~#pisfMQzwF2?)7hoyM%M!J*2e=~?wztu6ykPk
z-7=aq>Lf)Zv#jjVz5L?jCO9CCmdj~B=18e%IYCxlPSjSGdJl%X77&s)eEsD-c6Ik8
zp?w0Z^Gx!wbEvmAOJk27;I6w>8q=A9@E0Xp%msT2m3uCZtg+9Sy^@U|ZnzKVD?Y35
z_OSGWulVxLb6I;Ow-SljreWEKUnz7dLI=uzY~&T6Z$iSs*S!*mgp3JNGn!Ax^B{y{
zQCCz8RxG?F$Fqc{scZFYT?`}U*RW2U^ie_r7<?cn1G3jVX^mG3B1wfAeg?s2HDf6B
z!%eSM0esz0V{)BYyL3wY;0v~xB|L-&I7S>9TPlD!-BeI!f^23jl2B8kQ`+D*dda$3
zUL~cCgdeb8nc(S`1V8<lB@(2t=viE}kYE&4!ucn@V}QBG)=O7x5}bbStbxl{HsdRa
zc;z#*m2Jsy?L|&Mr3l-O64Nj@lGs!wpsZ{|#SFJqNAw1vVh2`ZFaLuL#;*g!Q(aBu
zh+Q_k8mk^xexxR>)yQgfZpf@6OZen40fKup-$^Vg-?bay>)2{C8E-)VpWivLahi4#
zW%bqE4pV#5(SM?Vg53lX^vSH`>|)aGMhvEfC4RGfW==)$h%0QZRYz^2z#=qi-vJ7m
z&4~!dwphK3#Dbc_r2zDmXIYx%@*S#=jL+lTcH;XJ2k;p0=Jzg`AsrxTIjvVs&kZK4
zznaK;htXiEq;Xiua&qTAV{2JXQ+5XE0v?rHuO<#x(d^F^)wH8K!)D9@41sNBlTZi-
zw;hEeI($0o?_>!%Q`|DUovRDCtOzluC8WRYNHGQbciUI&o_p^3GaI^{tCN9>pa-`d
z3$)kJJdWIUtl}K9TKO>Dn%-@r*@cuar&%?nr5!7c$;HMKmyAap#;r4^o=?bGIEN>0
zJ0bel{AZID&=41pM|Loc=>Ex9{gpLu`<gRq*wI#&75$YDKY^vKo$n$rqy~*Fr-JeV
zhl$Q6;)2{Rhb}e}{?2_mkOM78Zv>#TPV_drXOCh5WGiuv4vVs0jztC{U_;e6z8=0=
zuq?iv2Ne$baAw@~$^;U9hTA~Zz%40hvVq`hql?Yk#dl*`>3no*ZIv11s&K}(V<F)S
z7i_0$z!Q0Kxj%3Aoj?Qe4G>gj^q38N$Cno&v7}pr@uaH<GdNVri`wxHX^IU;3G@JS
zTn$IQ<EZl%B^<3)r7!yJ17c;P-^nq9wR?!%Ze!mmlU>FC*Ct?T#=n#2kYqntmejwK
zb}pRjn7;4I45JC4a%i2f$?x3$-~PJ>E7m9j;;M!pd4S7l2+AZRDaP=q`I0T4wRTnK
zC_<dY(jy`}%(TAa_)pd_TaFK_od;xg-N|BIg!#`~_d7w5JV$OX>L?>;KT1t2%h9y0
zl%}ae%+8#@;K_n^tpDkBIc*7b=|<@fj{mf6W?f<QSrJ-ra4Wi<rlalr9Epi_H$Q4S
zRRbrPefw_i^0$$|zy`kSqx`ei{H+asH_7?E_u5btf<9hkGKasLq~e{f6yO*d`EH&m
zvj-HuxK`Q(Q;8Gv-CSRIgWaIoSTrnV-^NQ*ZbzuIwEo>dRd<$x;3}B@L}~tj2L_wW
zB4vx5gQzvCYAmGY0dB0BQ0rJcZLRP6$|!&sTD$TIf_~B#v$D&@=u@Etm%dx@Rm4Hz
zRwPhVsG#LCo*6k0A}Gj`LD;1w*NJ$QF1uB`kRZ05gyOvdsGJ3p!1w>81wYw*DSNh(
za1wzO<XX$RF7Qd&-}-l*nm<BR$1>|P@?_b}n#Wg_PJXK_=Hqj=T|sqGf;P8^+R4(G
z5g@67K>d%tyWTBZhk><R5#8M=FD=X9n)gZq-Jvx{6ORWmyq0LN%RN>`vSMW65rw{F
zwP>12z5=H;nokN(cgDu@3CC2n-+0yz5mKB2_2>*BuG&PF7EOE3ChsauFinof<JPE3
zoP#ZDM%~>oyVlCm9_zvB;Sp=ccYJ~W+7{Q6;6|ne`57-oW3DX3y(pVzB|yH;T4MMt
z(pH-9#_as%Us`W1&#HO2v3~X%(j(j^*-I;HxtD_2u>sZs4uZKNI#{c&zuf5AXdBst
z@Jlv{iW#_B3X!1JO$)ZF(ZRB@XJxFm9UwjU4%;apy#=1zzE^hJ6_>AI$7uuK%Z-0u
z%?7`hm@o}jZ$nv{rMykVClMtZj?zltZX@w~O1HCGmW7}616P5!Mx!-9P`ERcw%u8E
z{CkNz%7%&_|9d`hGqclpQ%qz%-HXuaeJ^KQ)_D-%MifOvi{P|b1_Z;SaQ;fT(<%e{
zzlao;-!Y2ftn$CM_&rDF2i|T=70~CL;TCtcT$-0rT_+2O+-mP|WwvfKuZWA*%kRkG
zQHzL?J!}0k&N>1#Tgegx4FLernwz#3V~3;nOTc>ch=K!NVWW&zni_6ln|C148bBWj
z+g#?A<PS$}`+KRnjQCX^_1|_Xz%a*P+_JE+@B88crns<y?<cA@5rGeW-*JE|6ZTr)
zPv!wpql}l%!&eS}-_HqIy^ef87mhn3TzV}V)rLK5qgmGtw5fX&mV4~`xqdSd4|qJw
zKtCnjbJn?D`b^=Lj94eW@5}U_ciUus)hdvF!|zy)Qp01x2<fj_Ge)u6dDW||Rfe|*
zAlp^pqY{95xShp(Cl$KvVQ^lwrRp82I(;V*IHZghA%Twnfa;%LW8ECNBEt8r_x(6O
zg3A!3isq*D2>om{qBdf`#MoByo%Xftw58)*3+U%(iHo_O-{rKU{DzQJ+o)z5-+bFF
zt%BH+ghDoLE8dAyuFUne6RvVv2`-?6mDq`ONmV~W;~&HtV<bE9gCqlor-&6a7$0`c
zwQT4Iz9#n*D<X<W%z-?6<Ohi}9{q~(Aj_@gJI#eebuLD#-YM=;%*9x=<ZK^tBjSk>
zJ&(ko{)5CFBFyeZyg!lk!_ZiotaCP5wHa~g2+?JYJMWOM#5y$NU5NwWq_dSn;{a`E
z$6ajp&>ZBG#89VgsQ^Vlfg85`gS(enL_rCz&VlNk?A8GT&q$nO-I%g@LPh={3GMsl
zte+(#nVPQJnCJ#D_aPNuD-D5L@=KAO;(7w1u8<`BAdf7!qxHdTmP+U_b>kHCvh{;J
zQKryp>t45AUd$~V^22NA2YDhXJSE<3`^<jmi~Mli27Z{|;Oe2%GVT0fEVPV9)rZh-
z=!ZTockLSfVbB<%*i{?J(t!dJy?3pa@1(Z81i?82#U9OCCkRnN7UV1uH_UKD9RFcb
zt9Oq-X!Re)0z@o*H%wrTq?VhAhH#44p)&cy9K3G_h;bwPImt18&YEQ`Kq7k|j7qFk
z`uWcYoZ8j!vG|*l8{6WY;X!$q;%T+x!6BC8$!{>z)~SX`el9RjqS<cNR-~+}*2@yH
zGsAGd|HHyrD1`*4M|nnKEeX+D)^WWY`v<CP>jlagJobr(fOIiSMF1tKyH46>X=z>v
zL$uhc_j8^N^fcseJKk~SQ}PUiHgNAd70k-(Y(Gl)c5$;op)~NL!cWL?Tg!aj27i<X
z1P0JjD*h-iLLxx?le2<QT4W2GJR8XpOUQ+ntd=Eow|~l;<XHJpEcjjI5TYnG#@LUX
zXerLvcqs(+kyaeh4d^8+)OrQNhoMUo<p~}i>QTDW3Q<ez)QA#XM`w2<ch2%kU3A1+
zSso)eg5ZVLs$EL3Te)3i32_6bXi=(%rK(n!GY`)a-7B(|!VtV@orL~RMg7tUcC(x$
zcBifPqui28x9nT8pEbq&D7vuQ{D(lQ^q}NmwpMC#FMAxem7s+AU2%o3+eTS8Qlx|Y
zn?Ld$N6aqn1Y22Bfz)r<cGd1>g(=CgNu;kFwt~`roCnD6DO8vP@x*N1;FdJ_<HBE@
zmmw-A{v{jAcM1UGet7T_!#~a~kG#r8%EK<QS=Fl23UnFF&FGJv2+i$PB<cJvc28M-
z#MX`fI8aFTnX-`z+hS>4>RH<DD44&~jA^2@0IwkdiY%Kat5HaO+$aFOg@8!SA3MeT
zvwN)d<GU!P2j6d(H+zWX<aBzmGz1wSwR+VR*K^@JH`(PfeF#LRtdk{lp0H;;=tB3$
z<?4kR1Lu!yQua}|tRHi++Ichfgi){<Tdq}oUt~M4=aW2qiBS6E+}s=uy4G)IzsD8D
zM(jT<{Z{FB|H3>X*zKyhBS?gI;t9@NcAEAR-<k*EzXpEdi^Ca&#lcU?@{1RX2v0*l
zx&3BcK>v&Ar>mX+xBuC2X&pv63=xQEdDlB4GC9K40;1K<y|1&;pCoEPz7*m>>2m^U
z#!9LBl#Tx+XJ>i_`xOKiIXnUjOk@M1<e6U~8<?!RzDcNFkKbjDpOjOH#l=hCn$dEV
z?W0wCn#bfGbZ!?Y?j5zopX9nX8wz5#RJD(kEc;ykN$$Nva?t2x2`9A-UWr+{KZyf=
zm-V8#gGgdz?%9tIvQmp*57lk0G`C+7>es6d<#+3tVH-d3jXHa1fxGZ#Dd3({@VDF<
zx5_)+UJ}afXcly4)@&!Av}1{5w(jV1aH{!)4Xg*cNX{bpiMZ#%_)c^z%+=Q8c%c;(
zK*L#Dbic%m8(DYo4$tnv2(6aIyC^C+SJrb)J;LHX8(Vjjeb+nhxAEwMoY6!*DZimT
z6H(;JSKVWi*_k`wuyZ|cdj+msiR0LBuIHNVnN`S#R_WBE8)LGvUKlXh!YMh@Vi^<4
z)N{6Ur%BOEjMs1t`;&17N_W<MRp0T5bt}Nma3aKYr?*~^{HlVt#+G)cg?^0a!FRsf
zR?0x{-)n2p+&%wl>ji%7D6sI|i1~br8?J7~0~$~S9$Nu7GLt!wX}gSbo@*{10z1**
zd;h}hr#T^q31mT}lb`x3>^?XfN!bQ}>I=p<D9q4L9T!$R5AU+!pO&?jmD5OkS<v}Z
z`)ST0;r=5&$mmZCXQDW9E@QK?pT-=B$ST||VZ1ntJz8~uu-bt+a;aqWM05u^8BAoq
zoPH8Lm2b^C_^3vVg^YIY4A3kfl78#FwGs$j`#t44wutQtFe3ZXy_{?pW5f}JbLpqP
zjt>zBp(^T0)K}phtP@3volEE40?NEX5cjfp1g_Q=z_7ePpm^SnquevfV3jHt7H_Sz
zf+HbO%YW*7U~12tZRC8#oi@(QxZ7>U+|LmX!?p^P>8g1eUSd1niOsuAw>oEc3Md|s
zOPOkkarCoTCDOm*CaYWBO&j=`W5?Wlb5<64gFo{{PpiS9tiX}X=@C?VINyo=!~%hs
zftPILXNfjkd1|E&mR&5nNP~?=tAKW$r?D)dx;i109&<dvP)Ip409pNKzAC$BY~p9h
zmngU3<j<0rI|V_U*Z31)bXR-Pn$elt`0Tb;X_z!bm;$s5bk2avo2YM#RU`6<4W3}B
zddKA*ydjl@EypO)Ga;7#S<<pS9&_EgIYNcPxM{rt!UG=P_GfoUjttv(+e-O9kI$y9
z{mgNItAedZ7gMviDQ#4sD(RA@`=)JXH&PW`v#pq=Kl@wT{#op`Fq%-=qHNpARsv1=
z0*`Ay_u0h|e;|I%z$pAY&oK`IH}vzuHpqcTHy>h0loBvnkK0H*F*P38ET41_=-X&M
zVHOB-jQu>PiXE)@Y2)Q9J_Hx6{`1@a`+ry0wA!(Ws4<NcRF0z+Y_hBs1JScazLVlt
z@kn}bzRjQKogMQPWJ$9#GuFytsbLgbYVDuL)QM_@Ls3aox|pSPOiF|4>BX~jY|`j1
zM77Hm2+9~3+N93UeP7JVxwM<Z=IRMk(EE83W!()BP{caC=Uujv%}6jWGauQBto_{g
zfIHIIdKm-!>`nHUZuA@3w)WSHwi#`|;x0S*4oBN{{7g&RY-bteBbBzD3TzViK&7uh
zNbMK7Vh5+?FKq+ANZgoFLaxDIBo(1EIi-*a)`ouJ81?2SZ1@*G<U<X@RE=awPh6`h
zkw7Ba;+(t!29K7WkS_0msE(DM_QJi|czN1A2X9i(){YY<wD-y>G>e!tWo!6c1I1+Y
zz1n&5udMNloM6~Kg%Y?6XwTxnU7+=gL=fzWoFjug{35B1Yz`NHk<{k!!3Zs}RN$G~
zjXUvjfd_Ff?tCt<?-U5InC>QQWQ>9;^|Ey2s*js*zm(oTYb&Ma-LP3}r6Z|mZM_Ud
zGSS_(ku4P;T+b9;!6MhfV2-w8O&)%`ZRc0_zXwu@zjjI&gr9ue?3YQv>qr@X>8tI8
znQZgFEW5~sQu5pyio%U4BJB_VG6{v-amCRy@=HgCfB6fmWmDKKp!yh%dPI%Go%ENv
zJg9s*TE|QOQ%$R9r7pKq+zX?zSHH$4vn$OVwvKLq(kS&Fc#AcEnPhebS8z`=()wjN
zJFvQKXS15N<F*)22w&u9wN#!6@It<@>`3`d<o<e{J7Om~+J}_Fx&b@qrL*z9EMCGy
zNz2pEcOn`hDw5Ey6lm}wvbCxb*JTRG)_FY&aEC|sx=|(wH(|EIW_g5bnSI>0ewjNI
zt-fgD_AmVifk(HUs%xFk;+CH?`;|`w?Xr9WzseQ$Q1uE1V}rjcmX8E}2O5zcr(jO?
zkU9R!n<oIK4gab-Cz!aAXc!)p+e+<MNf;yopr*uAqop`rSY(L)RoRrJXhiy4D)_x$
z>w3)U(IOE$!LEr?UZxz_EN1zjClUCkQFWlPHGh?O3_sV3%5m9ZYnOtDr^vaw7*CA#
z3e4?NHq7yf&vwL`TP{OWP?=5^a60o!B);hu=-qDm6b;?7-mj8m99%fyfk*mT2rdWV
z$Z;gN#I^lZB5Jeo1lM}FuC?e3jAHYz#~5Z&+S^7<JoL#nV>lu3BnDb(WKID`rM=w#
zRpLYh9JifnP#_nGAkN!<9f&ZykIUo0uYIzhy9aFW*Rgh51S!)!z3i`ZsbSI|wc&W8
zp$p>ps`4^@ii>A08doBnIU6kxnE%>Xmd@<}0Vj<4uY<*#vih&f72R*(Ec>s0B$Wtf
zvJ{Ab(R-{>UcfDS-pyw8#+0!?TTva$W101`^1n`mN0xxSl#KBgb1-Z?Ou&|YT^1zl
zD{(gbwIlqaTKY1OS<t4N13ATQMj5*L%xSLxLu5u`N)st2=9SQ%2W_RACoG9r{~ItX
zYt_5+t`?TSY&|+8fP`)QIyVVNB0)em3#c!n)3&VwO#_#J?NXAUY#c~C<=ZX=f!PQu
zu2}FFY+%Ee8kb&j2yf(4!_~<RQGDf4zEgmvr%_o-lQtZ66eQS2^1Gz0V%aYsp{s2;
zE_noU8;!!)Y|PkLsrc|)Z9EE*7>&uOmk;;vvWfV>=?wju+$e_(EFxlHCb5z5qv_?!
z(A>xsnwQu^viVj(#2qs7s)`D>OQ}8BXzW&Wi^UCJM{(Ip8+j#nk)#wDu5JaFmje_s
z%92vFv*BpjB=%iaV1|54oQG$MD^RbD<DEo*2{;)brs}p5{pmu!wsB`WDiY&*H2lH$
z+D1;MLeo5r)WQ`8`A)P~6RenKTV>KAk67QaGmqGIzSE7Q#97~oPTz({gZ(B3Nh5sb
z0#@mK2YwT)E`FV$2dhU};lbbd?7&sq&~I|DA0SMh4gV&|=!5to#upIf7I<3wO(Lfv
zvg2SH1#v=sqmBJ02TY_1L>D$*iXNVpV?#Xw@N<FIB)^%6>V!)M2Lc!!z>WAhwHI|j
z_O4T#9HFHPutwVPb`(U{W{c(1JVb(RyHt9IgW_OV{!P*tj7YfWvd#&PB+znW5XN@{
zoPYiR{9`XW?t(5Ct$!ErAr(4pE5AwZ;vd;mYgs@Cgy{SYZWilhPWKR4-ZlyZaB(h6
z+blyv_j<{;N`G93pSSIRWH(nH+bPYw88+1Hw>itAb@;dq{MMm>OF%$5_}f@y{1sg6
zgvFo@{Wb`Ju4XUU@Ne^?2sT}YmdpG~fj6M&QP+*k6(v?KF16Ag+?}jXufcMUmUmpd
z6q@`R$KF^~cM5a7Y4{8d_uo2l5$xCMzx6%!DCvlPo9M$5Ad8(WZPC7BjVsejqZAR>
zjho?b9TlhCB>lHZ1)R>yP_td$Oow$2OnVGhSvJJjMbH2IFSeZR>?K!xC(20khLujY
zbO`YUr+QK4Z8)682;_CSl7Pk~G1%O-d?&Dp_R5kItGcwjA7@9~$f;*jL&G-9x3`Fi
z+zNOailDz;M!~r7w$z&cHVNXRWYLV^ME1MHV6?+hPujrmVzJ$<n*Ir<XM?}<nSSIE
z8~R-=y8^g(ZTNS-7_&GcjO2$d;`5tU%SIvtM3{_@MhSu|$UklDcLigl57F)|CJp0R
z7tOKMHi$}5@BA(om=voZ<U|&UB+JtI5bLoNA)_S*evRM7F(S;dW&(E{Q}J`wDo<D*
za8dtVV!i@Ha+)muF6XGypW0GXm?gX+$H8*6tQ=5}Tc;Xm<g=G!U$?yDE*Xcx5$hG`
z6qX;)<1A959>VR|O1_hUScI{}%CxMGmQX3$*GoGD@|{AMQ-%WXL$JqK4)jyD6^(I@
z>0Bc&6U1u!uEx-c;+a%l;3WuJbGrY&kn3skMh*PlSN+d1D~<P@Q+*iCq2DKk+6x53
zzt7FHYXR<0^E2{$M~XQxwH763$qb7(+V7K!C=N0r0jV>`N{92(#*AkPnY_vxqtC3=
ze_!xK?v-45{0<@BW+M8TC5Y@~Ksq;Vjcfweku%EKV{86CSJ)**xExi0+(5@Hqai|s
zWH+%OfsUc1Tg=k3PSTZ%6y3J;``~+!7u$05{qA>JCrid%+~Z$Zx9Urlqk`%w>y>wY
zA&RID{py{pPJ<s<`F(E9xOsBB%R(}+Wd^za9kcb{=g#=YIYs*32*~F4A(Y(A;vv74
z^c<Vgs)_S{&`YzkovSPM;JocrB`}Q#k{is9{lOs#VyMf+z#qzqfMKcD5)sVc9~@ow
zP1#U>b=MY71LbfY{zIbHt~t1-k$8g6D_>=;l=qvlOhK?FDeoAi)Nxmkl#czur=6Sr
z1sg9Ah^OHUk+X9B54q69WKeE{i7cZG*eATq3pSZ;Jc|QqBcTxdI3)iegUR|m)2(7y
zZq7HSPS0EG4~h2g;)rSgAvq#1>{1%$N+|^wv(vbg1nFC<fR!g<D3q3~cap<0KzQFy
z1ss`7s_#A4A<l+ge8&WcW8@EoBu*)R@=AG}@}P8Uv0?;xlO;cl@A@B-tl_dLilAFy
z_8^t?6`(o;^qXaw9#E{Ktyt2(dbMq5StSj7DU*inL|M0A#q#WrzU`mjU>=B{3-G;z
ze+)!%A5!1G(2sbhsBE~zMQE?#KgPL&Oc+`|^2Z$8jLv<l{jsp=<WlBhMo76o=1Nbu
zJT%@|d6qDfO*vkkp_?;!ztj^9Ha5I*XjLZu=<BJti%^g|1<=p^Kb*a9aNl=z=X;Vk
zL?lk4fI~!LKuLYKg6mQNF4cD3u1uU%um&{Py=bQ{m2AnDiEL>tIgzR9sIBje=8pHS
zcg7ENXZ+xYW5;r$Bub(rO5!9MqImSc5A?tf^uV(SFd)Ey00ROH+}Hba&iUHe&fM+X
zw0QLUd_RBA=im99|3~-#Y3ub)QjBK<)m%LG=@jT`xPp5M(t0Fo$ZU)AAo=#_|0stR
z-|Jm=G-^M=)AoXmRb!HmqO1)eETV?g7t=PL_0g;#{{J7<%S31gbXvS4-Z=rDcH2y>
zIu~&6f0V0;^lCoFa0;Vne{8`UZR;i9^rHX0M>$cKd_nMbhjL#|u#uOnL??u*V)fi!
z%)SvDTLMdhk>Dcgz7YtEc)(6LexvMTVUIa*4c~|xw3+9i@f*J5XJ!=uw<&%pnhD-(
z*^x{wbW~fuQ5Nxla8GND>~6PyG1`2f*&1Oc5QDX6`;sS-b?Wf%{ku^|_a3y)1Y`%_
zdcAK>#hW&eA3lv}WY|Ao8HS;Gz=x|1G{fjO%KEmGoH7R<96E@sYAmbGDuH$_!6-53
z8EQO04@na|0KNs@7dqwXYDvQm3mcr7Z^Y4uqs6k1ZJw)YdHzHNs+sp;AxREkEwJzn
z$4!BOI7wpt^q4N@^tPfDTB?ANB2|>y<F@P-(7$pXR=(lX-OetvMSs(km};@i{UJF1
z#ki$gu{G(nhQ+|}1fGJ|TlDiG+e_qyXvDr|P2Th?U$N%JxD=oMw6)~l523sFKETx{
zG1F>`KFa$79e)-&zC8!^oWUK7iHAKM$E(DY^4GI78Y`@aEUF|c?e(GFIx0t#_^!Jy
zn*?bQh8N?25>?(t7Gvqj1Wts%-Mu!t80WidhrIRiFG$co{z)72CE7E%(`r5ooK_@K
z7>|KYqoFwsYLS=AWR#V+4&daWJRNhFRuuhi{L-vX+FblnT#ILH-haVU$w_M=S7fk9
ziJ2DjC)qO4Z7w;8F3wGGiKo1heeA*y)9jmm89`XQqjlfR*S#bPDYi=W-;5pGiW72t
zFIk)daW#H3F!HA%R0%-Y^vzhxt)c_Hv0^e=i+_Lxm0~7a6(JX@nQxYJwjImt_6i7@
z82O%deABn&zowCqbvoc+SJ~P3MjK%f>;<&*Ht@}SONuxCaKa^-)yOx)YpnzXC@B;r
z-O1@e8It{qUN9<;GENIZ6DMBBd{+A<C9P$*kM<DdJw6=3Y;Lleh8*H0ZF#|_6X?BC
zT+SqbT?d9>6z5`&wf+GGm?yaAd<755U4?$>VxfZPH3B60p<0YhJcUi<v@LnJk;=iA
zD>%U7OiHP3R;o_U!q;o>aV`5+;`MM6rmXH;v4r?m;<8`=t#TAlT5##V6<0#e{3RXb
zw_@Knk?^O<J1aW#VuQ%)Hh(M71o6R%GvvSI7lo9*HGbh_!wlMdoB^eeXb-SF(#07(
zp3&iaD_Db~ot4dEm8&eF+IoORgkkKBt_tg)5PpgiFzi3vEnwu@+za#Yh73Woj2BL5
zL~a5{zg1A+CgeZgdb=Jq^GE)t|Ew0ErS<c*oCo8%mY-vDdu=j-=^|Y1bb^h};efpE
zW^#3k8OU6KB4Ss;mL%$$&((aaX9t+yr%6yv;)_LFLNw?!_&E5Df)keWhrDA}a)M|Y
zi;ho<oo~k`ij%JUw&Mme9KbBpe>-u5XMU2Du_0=RJA2+5za6J#+x`R4kWJq%SGpHA
zyxHp^O2bIVTB@ENhaW83A#441kT_V1fGlLJ-j>~w>Ks(k{_S%3_B={@rw;EAfy;g{
zT4(kLLRVTV*?;tYJC8_}3@lK|K-QM8E%!lgmBZhT!z^xF8tRc8;4{K5qq!<_qBt6J
zK7&Q&OE#7*(PowuzqM?+zX#i+@oxumRCFZrOigA_P22z0rUMdTQwjf{gL0UOGQlP2
zh^}RG-pKFZe6C6tP8GNj`XVF0v_)Sa<Y0K7``^TtCGtc7Eb<Loy=wXrx0Os_W|zvY
zBcNcN>n@dT`)bp6tG|?64i@Iz@m5+xRrUf7<g$sJv_@~T2l~6I`dorRz@rh1<WejG
zOCYgZi&s|!kJhTX0>UUP%7nE=ncipog%BnZk0Q$Ca3PZAO)QN%9BOohpyAG_AWIB4
zb09SIB|pDHi)U;gex1g)iye0ul&^Lq`<785#NczR(x_Kf(hvxyqgCg!&UTxIvFJ`J
zzGtkKLjl<nHh!rbRV*$E4jgSEgs^rNr~Qge$9IK6n2GN}t-uK5IqKh;8E-)KT~2P!
z?k-#^Xy~(>Y%$^7ssK?-{;LGGW#4nz2^S^$aph8wH9WbH)-0NRCut}tSPOT`j!NHL
z_npAi(9O7;e<v=d)TU^pK;{kElv^NA68uE~j`(KQ_?<F&BE>-qd?)VJfkB+<^#)|W
z-O@U@cvYnRpS0HRlsRqY-g})c7rM3kFIEmsgLQl-@YClZcsjlLY>mL)z2EVp>X9E@
zSTj&nhFBW@PRu^uMbvotS-H4JqrJ#w5fINog&zHm6AACIYqqgm4A`hJYpZ2Xdoec{
zuX<91h2kD!lhq{N<n^7dYMc!*)C)EfWs&q6+K-^{QC1B6*%&dJ^WQ1AQz^a|z7y99
zZdwYV#i;Jyd%8Unw&c}!3->3Go3W2w@gE44IimL6yyj9}xZt8L$^_3V9C<ko)PL8H
zi++i*K+0dP;wfwRZh`8{{n+b#H@4jUt-%OY@A7NX0e;S!zZ;wJ<pbTji&cF~mJrXa
z3yNXCwPr^TAK<-&5@?H#N`A${sUseaGJ+m)cV#hwtMPZ^jBLgbrSrRv14JkhCj7gx
zWJSr1GT^)A(EXSHV#8jq5G>esB+SJj&V_fQ(G&PZzJ}k8E1^Uz&_lf0W6?;AsovAG
z$nR_4jVE!(e&u)__wlxFw#o07TQXbwWV22C>J1>hok=j-L^01*V7MKoZ0+U)4A?fP
z(*<vew8R!;cmB2fYAL>G9=NYu=N!MBbHZ{G0ibvGB{Bxt_X2Kwub5ih_ws}g&;c}E
zoXWGxPK8{xhVS|FJV>mg#_#z-;S3@+D9S7LMxxuM=*SHt_?IWIIa(r}vX=NRPwh9L
z4NtJtt=}t~ggbEC_sY)p7tf6L@0Fe9<&)D$hj;SU|BXvk=l9|W?kkxgaU4tVAq^nK
z%gPNzSq8;cq_g2DJ5wGJPZ2xvy>hhmZjJh%A0r0dQ6CN~*IpY7_(6_V%aOP??ooBZ
z#=n=+o1#UQRAVyNj?l*J^K`V5_*=>lkybM~34*dfTOb4W$%{H7E|E`Oe%xU8g;;No
z!cb67toNeV!w#OT;I@==Ml7mzI1PQc?A<D?8YE#<c!E!|_2FM5c*#vrx;p^7ZYi!0
z3P*xWS^bh<E>Y!3_m+a>f}T{~68YZvt~W0E@sZFM%Mry=u%@MQUZB7GtU12Rs#xB3
z$~NB;Wx7qW&N>C+Y>n}-li(F?iyEh97233z@mu>+taY(%ESROE0upJkb><Yh_8k=c
z)VmZ1y>Kx$kbMH9Uc}c?K@G>einEK46Z0C0+C`D6^+bX->cin#E7P{4KFitK{eNj=
z`S-{F+-ghZOn-i}jeF@^teH(#qfZxq5S_{C1dtg)*v&*0m#|AO`-08+4jmwJ`Mm!*
zTl){TP%YMT5&(HOT{|P_Ktf)kyn^!4Z7k>LusWwf`H#Z{t>khcXT+h<?EBY|a=bxx
z-!F$_w)P*b{`>I)!|9&k5G&Gb!}kLf6MoPd^M?X&HGMxeN^bT=Rb|cJ_tS)3KLlFK
z_sjVh)cJ0Wu0<osV@zHoZBbUpeZ2;3ZhQ8Ci}*9vk*#p3mHtMaVVw#8gXjv8<MjF{
zv$ZW<HgK(ekqZf54g2>l34uoZZ(d9|f*idlyrp3w28oYOvTnr|82i5O@Pk$hsHEC{
z(Z+KWmPzhPxF6eOlu5dvm>tsCPhTq&oi0N%o2f=6XN+dVI?a7Q4)7Oy32bTeRrk^*
zx*}`g``7C?JO?e;)Ag<p35!*?(%wq|O+uM0gZt%`hb$%QfN!8|SCJ8ZKalLEy;u-R
zQuu?|Yo+w4`$0hU*}_xQ|DbGbgr#rzLA)k~f{^9R<JS0ta*!THGTrn8$0Nx?%h%8G
zNb?W!^kIR`JJaGrB7tDRTccMkvz*2WL27L|(ya1JxT8(`56Z#nKLD@Okzk^#7eN`-
zi@=ffRpO0a2fbirn23ZxR)E-j4OcBraK>ChGZHNn7xqaTtw6w9w)FCp9gQ-HH|4|f
z2KaC+d+y%{PVceW4<asvthOKxq|LSQ=$a>$eVg=Ov&6K9Tc3_^lDiV3pc>-M;$*4A
zJeOTw%MF1SX;j1mpcn<c4y&|~9h09NN5ZP(6L1Dg@k`Mm5F1BpiI9|(TolSmlu1sZ
zY{MtyDe^<#@?vEhw#{`>mb!eYh3l(&I7onl6rH#s%7pB66XZDD7-i+~a5}`ZH2ts~
zGLdjtSTt8XDOO4$W3A<fac0GfYt34#uE{=FkjRtP7G*BAm5OZ2+OxCb!zt4GgmwI|
zoTIKsxD1`~eH<ToQO42%^!~7%=zavs0}dHbJP*w_>>m#7N16D;xF%cSO0Xi#VTu-s
znM(ecvl^0F<yiD|8gBBW)uMh$Im0SQl0E*zxFb*vLtwbcAI7!(98%ZmAI8-Z1;+uG
zPThx@sX%-hIoDhT$b>ZZMF=(j!-CKrLe=bD65<$yv*@!SCxI<_IR_Qqa`}fr;tv$}
zL}s^=(9)gI%o`j@yBw$hiwUK=6PByHoIuKSO}_HVut$IxtG^ukM^LCiy@n0ZTw;uL
z25ZbNwS3Hu$Zpnjx$MP<ar<k&9H&GXA*|(cnbd(j)*8*6yYL{SZ8@Hz#$w7~?OwfU
zoppFO;*8kq&dX(yNzIKlP;UYN4J%p2Q^+a?5>S4<RL=ENHtd~!_%Cgw`YMo5ae>Ee
z)CVA?lU~51Q9~MiPZmCDV^Jm{k&Gq?m20)jai9pe3xPJClP%oGWHpRTR@||0I?6b+
zMZ_b^t(nW$zZc^6cyyJ9E-v_74x)fzJ8Zt{LO%HT!*kykVtHf`Hg1cT%T>Yhv(J{Y
z@eb*=Y}r3`i_2JvjTb`1wzWhkEO43q$d5_6c|aj2mbxf&6}%ozED@@D#Y`UC0eI7@
zY`RD)WVYP+qXJMd2y6OL94c5eo_R3GkDLxUg`fy0gX$Nj(^|8#5~j)0pe=gWErJ1s
zz$H7WH^Kc7`G9r!Sdi*H*7>7++C<<Lk~N;TUav*wi(VUuAx~o;2MLlReElWl2`+~o
ziLzp5$Vw@~*65GQQEn$LQharU>4s!uSq7=j&F8+HvD%LUH4Kv4V>|{W-3D^t$((}1
zP1tlamQ9Eb4w+JIChAKnIBs(lKv=EEl;(dFXR&`5Op7fze4G1iv4X|PIlWCw*&P=I
zQZylh4YrkBbMj4Ng@>`ykIQZT*q=39-H&~*PQ$>SvictfHj$74v~^P3|Km7&Ten%`
zkINE$<R7i+$C13QIQ|vMt*%+~j{|8HjhYmFEzzYjr0yir>VqOhlSh1;4*&v0M(tk^
zo{LHA@Zok56#vJ8l+h?kF4kN1N-!8s%~LGsKuksP)K1y3xBbdN8_BkVT_ptfXtdSS
zr+AyvDjxlDEcI@zCa_HUacnGcRIK)6KXccK_hN&_y&BG_w3&?I&+vwmzAQV$K8zDN
z6Jsi@k4*nx5H;ZtZfmpo3ZiK#?0^drB3tONgw$s-R~M}~nhHFTmK*}f`+IHKxApT7
zKP%B%k_O>$*#+58a&tWahJxzU{Uq;G6;++_pMxi=|4E?vUBWC4(H1gV>y1w(a>}f+
zLzF;M_Fjk<cS38f%9NK5a)#U3@)Lgrw^-{>V$Hit=-IZPlr@KsdtsloCtSs*RTxwP
zA%$(~$WcFYfO&TQByf*N7NL{gYDlst#5N85q@3WM{h~Lsfs9#oVn?z8%%_yu#PzY!
zgzRD5J~32xkaz5`F^5rp(TFOdrCS&)iyDIKtkA|aoAhxxumr=4${yQTblEF5<Nbd6
zVVm<w_3tEOJ%;H&iAz+rS+Rf@qBT^aUI><-c-dKNA9{d7OE2Zp3@BvjvX8z~O6rwr
zl1~;gjX>%~6xACMOKidIWI48at8!)5FUPigrkMCPEXS!qaUj-e0_);NZ~CbRttq}1
zZh;*tVf3TSZS6#8A&l@LXnz9a*t(puzlwKlTaF8kh18QG7=4Jp`J8nmbRTd2*e9)X
zxttP&K2Y<$*$AX2vx|Wkg+MH5(}tJ*Y<v7cr8O3XD%Iaeb}6SO8_izGpv&`NM+1Jn
zm7Iw39FIj=(J_G-pGF2-b5IZsVU{|cKYE6C+GO^Rxdh=z0;)5t;F~t%6YJ}@xoRA}
zwZ&P@=5z7eaO+1o6-|k117EP1fZ~KGv!$iwa;_1>V4+%#zpvL;a{LFSs<)q(?d=*;
zzy+)OX{<e(I{G1axz+zPu${>M2CvN_dyQ>q{HgE0JHNZ^xhcz7QISm;wpsH}3n)9g
z>7m7YAht;J$3AYYIi3=FxJgo5bRgc{T_)bJcJDxzu-4&!&({9kzq8JtmeW{iqZ8Ck
z7`+bsG%i1^9m(`CPyK0J9BCo<?1PjXNmxipN%*R4K;A#63V)i<3dzXnaV#b;g5V{q
zIUM2xY}^MfQYo9vYOqbXBE+vVvhlK+n4F76U8pwa9Sx#VxA|)F61q9#wcwLX>%zsX
zf`kKye&~afwv-cLQ8(CfRRdq&&_S;JwA|;<v*ABW)Ga=^&VO-em8GTrXJw<_`fogu
z4f%zGPVA7M`Bp2E_B0n&Hm^<D(_>u+aXyF|qR3yNhTNZ8eilo<7qhhkdbC@81nj3t
zit@8UXdKv&%RvlAbY<S^j;gDt%Nde)O{Z5Do7MZXaw5qgcM#?;S|Y$?z~P_yG!Z9x
zIT=JtXtL<>)U1qN{GK2ES!}NYXM!Kb@^^PXN{1e+Mb9`iVH-5=<D+P{$>_{|^xYw7
zIx5S-8%a_Q$J$AYEBn<shnky<qHaE+Mi@I~3sqy@UR(4z?BM<O#oDzW^OfbFB@Kc|
z9-RHNAUlek>~&0p+0V-!_Q=$#?&oENd%Ak9{^!1d(ocvteHNOw;pe`PMIzGZ?PP6Y
zO;K4UQ^j|e_o(^jac+h%LbMkDh2~4y%Ub;x*Znj|tSws0vU{A9C3EU_9}34eP*X?M
zJt<_%Q8{6qKabt$-%(5idSe<ABMx|D)WyUgAI^#J5-4V<0v?aV_#z`obl|f-00(<C
zD!b@UG4`Ifu_!CF9iyYDx7yF+po((BumzNzU2-vSGx&KqSGy0|boMAhp2=y)X0kQ*
znf}<$d23Y6_=wtk^aT!>ecTpurb5(c9v0)PT)if3$>&;h8Md6=$h*N-et!L)OTVBX
z{dkU<T`A!G^}|+o#kUzT6%uxx3y8HS3!^whj%X9Elv6TW``eFL<CSts6pyayO5lIY
zB0UqPHD8H?hi8K}K04FzGAU4I-`4Ex)7{o~CD1M?vqVR0zixC1Z!~(xm3UK#lV_b#
z;abX40*FPK)DtqzdK|8Dpepk`bqdQi>?1=Kp0*MHTj30CG|Cl){%B4>0i~ecuat8P
zC83bRwdhp#*U!A=nNr8SMtP$C%qDY5ctj4cyU=#i(S-t6ylOMmucr~Q%|(Y=dbt@X
z4flN1kluX47NSPM?m`YM$YR#uX(2A9|KhywgmqkwZ^}G#(pKV^4*prQU-)EXm`5^$
zUnISIV1F^ss{chCYmW{r!-8MrlMNfB=NZh@_>0($2MOWS6cwZl(EF{By3H{PssZ3{
zi5e10N&ur+maS1k!3s`V+b`nCJc_LZ!N9!#1G^MFs3ZGA;_<9?#xGt%&QWi|kLNlT
ze7F<-qMXbR6M=pB7v+e58N-5V3ZxR#3N!Q3U*y%emNLwR+DCs82Ypk!jpan~V6<8c
zEen`eY&=26{uBf_>8-lRy>8P{Q4H@HoAF<`=3>ihbKVRMZ3%ERpA+C_7X{UVk0^sf
zE_wpa*2!u64N&k>lquxG^DO2`NqwsYR?I@t^QUaZSK_N*U1#>o*e>K@R`*Lk0Fa*a
zR{zUD@@V>$pa*rbHT*IVYcF=1oPfq?AxRuk-T0*^k&X6Wdxmh|NB$qZ<yBW4-qv5n
ziBgE}l1Q}emt|wNK4|U!*CQbK3JzIZJG<BU%QzIh2c^R3&7TPV&|VGvvg~yKUg)&^
zLI5lj4<pf+^p?t~hl`GlCKQ{}g!t$${QwknUhv*)g)M%@#<IJC5=Fe0lPac89)^Cr
zDk~y35im9x9hOrhib1zQ;UjJOm*qrZ5rjcO?!_+I6J$3qi;u_4G@q-#yRRfZUx+#q
zvMEJ}c4#s7K<PN2w56(syY4B;_Oh?+{rB5Se9ib=vg)^A`K}jhF%ZfGv{_x0iA_;b
zPsJy#k1~mKaqTIVp$)%^>+$h_Z;kmK0io2>vVd>;RXJaMa1^}r&A$o+1HD{AcC`3s
z$hkps#kKw_P9~Xotu1;J3M36*2`k<HtFp@jc)xeVXW>7xU==g+&Z>#hBFUx~|I%N@
z@!fim#7MXtI>0F7$WiRw5xc=geif%{M-S${HtKNkJbsPP*GKas0XMsL+SsoWSs)r?
zYw@UOx3C?vS^@>-qSRd6N*j+d7jF2ioy>^}S4{sZt^|Ib`-|3lCcD8pUk)g5a<%4i
zZX93*?!h(4=5zJ-AK+zOsA_nY`Ztj^Eavw7#b4S|1&9(LmAve>>^<y3+KY+OO3oTI
zf^*r59}7)Tfx%_$QMVG?jdZAfB~S+&ebO4DwkRdpnkn~iqu0iO?9Z&J`u;3(UVnX>
z{d@mT%phB$H;z89V5|QsPv|GCEl1sYz}o!_T3$)&I#$YoAb5s#dZi*R>W#s)+KTKC
z6>S46<#6@mB7mSK`yV6~ppC3tzoCLn8>(QnqZL%oLuVh2dSw%JDL}){3yiJ2-Hf;+
z&EIGLjj=VYRso9klfFv!(`w9VsLjc0-8_=I?3|~4);oKNK<cwD`PSzAZ@4c=hLt$>
zdk>aG$O}0T#=>&#mRsIp)*)P$7ISGO4kGdn52m{81Icb~#g`BD9casbo!D|pQ4)CS
zl{!yx=<0r5u=!WItp3;KctFD9^k@ygE(hnW!(iV={{rzVTCVBW@#aK?J58`T%0%MG
ziBNV(El~#dECvL}T9RC385rqt$|M48%QBDEE1MQ;|8*dit=zYc1ne^hc+NTl7)eO*
zJd9ro_`#9+bs~j5vI!slbvbOGBZ1tA|M^+GxkqC>w}AKTvPYw=LQcGtV>#=$N5I*&
zSV0#)7OEDEDdop_w3VRw&EnCAhnWxeAP@GmPsRP+XM9zO+r{gV^Zq*VbUB=2OXg#g
z>&#KW?-H*rR8Z?!$(;E$Tl{sL8O05OI9d92l7svEcj-Z2&XpVJl{HdK2&pqe*>Cb;
z$6#c)A|KZMrffNZ{4hoOjpGXG_+PVz_$6|Mq!+)5Vad@T8B)`4;yGed{T`mM=HJ8y
zV$EPJzlm+xRLqlFqc_p8=!bn&UjA-tuUd6MQ&b?FWtBS<1S0D_X}vxqXZ1h{uOso_
zKu!pj&aM83y$#ywBHHy|a&5!1XEb4Y5MT5zJDMNAD07stoFP&-vDLM#Tm+Oh?jXa2
zkPb260u^8nGo9VE6FSIdVqpqB#A#Q)lDViaq5m11|4lieAO43wvW1*`Ka!Hg><2y&
zMR5KqIVP5)SIATDvWQfegamNco?+I}ep_}A5x4we>ikzx<PTW=Zygo5v#Q<-IR!~z
z(eT^2*l;;|#gsd%F={(iB=JDRpR%UkmeYcw9TneigNwwMGSYao<e*5XnyvM>aU2l5
zSzC;wv84l*47S~?bERH~%jx)SoXrvqM5_Ew?|{H=h!c9>0hFw*6v$FS-htm<y8%0;
zsTlrkSzv;{*oXt%zaJaJQ6CQLE%|F6eMf)7iwc><!q`}VDm!zfL6)Ud?Y9MEKDf!o
zs|Azmw6u8`k$9O-+0I8I|LOE^<NWWHm6Xj~w={olb3U^@C27!nwKT=+zTkv~oD+0&
zi3_m!+qi!+%2UYEXe5^Q@seQ}mLePVhVpDB;X*IcyN_7(h&JqM5C>Zyv%0Ik`65(G
z`asyT`m1?~q2C&=mX-V@`4}3bZdwW|)nHSW6*f|9hy*Fi!Z4zQ8jUiUMPo;TaOyQ{
zz3LoAyS2qv$FefFLKb{Ar`tp57#Paz0m2HGd5{hn(XtNCQpJBQ>J&K2%{yxYQ6~CS
z$qXS}Y}jYmrKIXM5*4t!=llz~!+2Z`JSYasj^;4Cx(Yoy=3j}5N<>hrMO`tNq`#`b
zRH}MsyeBKLw7u>`Je^<_V>}r-owAuI6M-)$2hqiIzW#soM}K7V2}XAh0Vg0N6G-Mn
z4ASgSLF+h`MIC%KE+}kE36y`(mam3E6bO&psBUw;a`k$`s@O4xv6}rZkUVcTYID(P
zR`<JdB$SuC-v8YGcOSKesHuQN_~Xywlw^%j24x3tQ2w>1D3e7KTrZ?Eii6+za?2X~
zYu56+xHpKB2e4p|j}kh*W^K`2#-cFN<sh|J!#pQX01hE`MZa@A^%cYmoxcn8Kt7A&
z0^M76sZ|=NAQHhXYUgDe_8JmJqEj9DUAYmcSda>hdUfo@@Wa9QRX(amqYI)P6j_B#
z3da&EF@J$;Yrpe_76n;i+H&c9tHymL59}bH((lT>`0N&&PACZ3K!B56m)%8UK9~JN
zXr)eTJ|?DI&m*=FHFO;^p}=AVsmE0n`LUE8_w1AJ!LrY`hgTjQa@1F7+ouY8B*EtI
z{ZL3?pb!^t09@Vg%X)89kd^x1#|FwRbkG|76H)}E64A6gJu;wxS!w!xnQu2a4Sruv
z*1#YM^nUM00<FH}PSuD+aAJq`q_#!3iUHmJ`#@QnVI4c7GMq0jWzn@-XM%)Z7jIGT
z@8d)eq=q|<1S$*u+I@3uIK~(4Lz2Yb$ICQ$V29)(aGq63iIu^dbM*IliE!p}!Y=l9
z?Dt8qDG_3~)f{rdd4Ivieabs=o|*I^kumk|8L;VuxV)@3lPe&>YLJa{+3#bVOdhoP
zfLSJ&kdD81`y<ITBG@J9PYFQ?eTv3v(MR6LkzR6u`rz<vIR`?PfNY8P5UEtcGDs*4
z%$NUA_U7|DzAR;X-5>HaK%&6}kVx6;|Bx@ahtEME8MTyeTS>$uVQKtB+3v0(BH8_+
z93PGxWW^ui_$Zz@n#5=+(V46~PC%cu{vlSB=x{u)*5-BmPAYu7&cq*LkCCbODd=BE
z0?G>_shN&`XO2}O8(8ljd=J6B5a$xEpDt&9!Ul3gQnWuJ3gZvu#6b)`-fttmAXGhz
z2w*hE5iOzg3v%Z=n#~co6MVwPa$F7sba$;9SJK#`VNJY{6$cSS;SL>ryNn-215%@D
z?^}A+;vF^<ojHZX9m7ZpSl<2@o6F9KaRPNU?~4xe@`^3^-#Y!a7#)=EStvc61!LKg
z&w94j-Lu1%ugw}teI+2{eN_lE9yIh7cJ5ZO+clKxqG1MYm7Tk#l=KvD@;W=`Nf-Z<
z2)xyHZZ#j-<2NwRlPws=icwZ3z{<$7ZcUU)K`SNYgcKci?)K8LMk0DMj&g&xHY#eh
zq+TdSYj*BE8pQn&T$bj<=(?)*D>BzCN!**FjHjpAzbeSXdM_h33{i4-yC*<%)GLN|
zES8Ad4R-EkMnZMQ+*+#6u~K4&udy8)?c6FVJfiQdz7is6xp159+-+qxJ4HGc%c3?r
zCrgd5{-tg4W^CK5)*js!?OTu*+j1y&9!Fl5j68A`d_`9bwfB;3&oU@jA=Q@&;L#ai
z=+vCFZlA$y?K5~+_eOOw%UVoqTCFe2*qzFXY9PDCbxZFa@)sMjbMG$mDSM;q6T6^0
z;sJs(M!_LJA$UU-B7G!!lg*Swb-eI)I2Pc)BODox?;f&1aJ*_qe5pSCCwA1%$?N7n
zLj)YlCG8UNRD4>-@^}`zA(#$39=#RA)d<O4t00gwlJ3N~os2SRCA1bVL9mYJS%&rb
zm`x;H;6p`W6SA3%#U*45OygAci+)9?Sen}Dgh)B0+4g;Qw(5t7;>=%$<uln5O`>qA
zT-avw{9sUi#^$0%;pWOS`-+{<tEWg~L_)K3b!AJ2u&;#Myihd}xdM&P{ax@TJv_*`
zN!vwV%^p^BF{i={e%da@h8%?#Tk?K{B5+Q1HR<I%M_tGbmSdpe0L)ELq{1tCKBS{B
zIo@Na;Q_l^0VYkPvs-q<J1_B5Qol>QWvgzu9*2mAe#PpdOp-u3zD&ZntiGY#J%n&4
ztp0{_O3-52n&_BeM6tT=wb0THUg;|w-nH2uBVr&~X!ph}bA2iDlNW5=4F$V#KAwk$
zXu6?X;H^7teZt5h6EYnk-wW1!LqUxFJ6-j-;fA;{>`meRTC$Do0_jn%MjLM^2!?<|
zY))%b5bp7?e~;Vdtbn@O+M<HO<BAwp+|n&sfzW-{?n@%J#<t~$?tVlv4*Bi_w*7`c
z=#UByXy<jqDuO$7t#x~?66Y6HzDzl?yFNnR_4!Kf>Bd>;hI5_e5dK>wvH$<-*^sZU
zLS`vb%J2<w_7TnMAv~17F<%imIg;S^B6U9Ov;W9PZPWoLQnek4X-AIf@(v&M?VhbY
zEJKcC{);#*A{WQJ$G`5vW5SMSW6-da_qxBeD04T@V!Lf8<1j{2E~GGtB*(owg`wPM
z6M0s;dv<k6<vDr7xdv9X2qj<o*Z<m-w?;(3V?P~-^fZPw({?t&ka2i<$FNb()fp7t
z?6%pg!%>%FSZ1YjzBDLa?0kIo=r(RGHeZ!Zi0R|TJZ~3bV4dY>Y$10DHzd~UVtf=K
z_RqE$&Gqy=&xle*U-C(OzKGP9e0YLY+U2M#uOPkRQxFWxH^g%)+eUv^u6T1qiCtvn
z$-(#X>~!1J{E6I>(_uF{O7S$6&aSPxF|OVolu~s!#y!VK<4Ie6V~{xe=!^(%(Ha?2
zA&)UPwlz1#?Zz90aF(!0H&*X`@RYXpMn_Qp9c0y*T|#^!;9i$M_lb(KrWhWUm{)y$
zj0ds=T?!k#WX%afp@aM}+HhlBWyL*Zo)FqT;%sgEM%(Cr!vPW=-uqWvGkD?63A?Mt
zu>^_X6}0<#+fohxTnUFlWE^YH;R`7$74}oM?Z$$0KZcaBBR=y4f)2@?j0|9VHi15n
zZd@iaoemQ$NS}2l^bslz@+1cqwr1Y)WbE@6dv<WA2XfJN^ecksP_==?6B&*vNWT7r
z9lFu452qXZcO7G~bH&xCkRKyCVbK38b~wfzXKklI=3;+Fvvd6G>_{${Ld+<D&e3Qu
z58f#|<}fNPu#MdqcnRM}oxJ0|NFQFe-fG$97dW6NZ}jI+!rT)471<0s4H_BGwS4=Z
zfSTy{oSuoOEkuJ@iA^Z~a%(63%WUmW{-^(JDq%#&J!#YaYqrFVva{J=A#Fs#WgPpN
zgn#Fr{p9(x*<5qY8{cc6&Q2uV=Wi^J;itCRyw9J^WIcAlOaHfz+Ja9P3J9UW#Q;X$
zGbd~@fT;}<cEK*?y8P`&K4MEbUSEGdy2{JdA$nOnGYdqp&IEBsPcfOjk_#p(ifw)p
zb=6ANA!N?}J;Qc2#)rm4s$w@K3xkqlx2?J<?vX@nm~#>L^`;;}xDc##VRctWSuu|0
zJQEJW>Z>x@2)I+knwt`p2v0QlFN}+oXJPG4i6jo}m*q2ux-qLmji0r3Hx-nzzn|x)
zDLU7t%Rv?%&eq=)Xh)8FbYRUl6)dLs75i;Nv?x(OIQB0<X}4T!fjAz4-^Qrqc2q_8
zWv$+e7`j2*T=k-OUE<+hwYFTb0W9WwY|Bk$@3(EU_M6Iicl7~gY1^_fd^A9dP_YiL
z{YaN>&#$~Y?gM(-I(=;3+@2lQ9Tf^a&YVv}i}c0{onaf^u)eAWIado^jPl^7a}PKT
z@{R~4|C-XUAz$Bqa=O~E|4oYf9d@V!Oh`hS^=E9P8cC)x{+=Jsb->~p!NzF9t`y=&
z0!~GOMB?n9`U~lI%!fiYLUfGyP|_G5ujWS7d8_4KeSYUqzw{d?bGThdt?^>Z;lM*S
zaZ^FR3QIHT<$Lz)ZJ&z%i)HNeO^#*?YGFNb<PE9B**uC7R)Ll?G2yV5g)ex@W_|A-
z>ae+-9ytDMc0PWIogsIx|8f*$7xF&AUArmS!gb3lD<j1{xae?7(=uR-InljPKYQ$w
z_lL0x@fB<-M+R5EV3*_96ZqvnWy`(_3a5D``{2z-OS=;FPR%}!H27+M+zC000%cWM
zJi;N{D*uz9eO9;1ue~T9rM!IxzGiil6%!Lc=Si{pRe|KCMP=dEtO^;(&VD*<Se4X6
z@j+U<s@#=7{VQwCuKW=#{=4h^Yxgc|T2)<bE-FR>>vI&D1afhkSNUE#W22`{);LnC
zZitTER0$_o5Pr+5Kx&e4ACUFSMxQA3tF>M?Q9Mf#$h~QstMS<aEWoX8Rl!BDfnByG
zKX&(!p@}KCM-P#9{%u8T+E(?746T4hgt{n`h)2N?<kz}A+w%H6&c%9@W$yIBptXCf
zJKKOg6t4j1T<c8$uo)_P?!K&4p7l^7Mp+RPl^c7)hP-UHwz)^L?1D6hqn<ll3!#6_
z4&@#`wr5w5jrcG2`whf%hkfO?_F%4B4Lea{ZHS3G5|y=%gvMJHbmfi%2kn?|?}HE7
zSe}ZW;hz0F?RfSBo>Z1+g%zqrnd}G=Nk~g(Cu7s4w~_V+W|)b+ipkG-u834x+OG*e
zFGM;QeTBnw4gpz%cXJ5^F%@O5rqg+O)~2&wkxM(CJDboB3?7u*n=e3f^T%zrsx2{;
zz&dSnKFjTYZ0EBbFDXL$d0&|jK3wo#_QRwvTsxg5&fK6~tojviTlghg^nRO~>{8wZ
zoRv|7*-|wRf^z`HNr<G&Ui&W|uw~x@C(N(pP_o@90;!c~tm#Y6iJH%@dgFneW;e$@
z+|>hf#UxhUT<*aal`^94=IT~S%E1$``sRGnOH4O;5Y~HxxBeeAQ){vTLQP%_P*20n
zL0ib~fW1VnX=|fQ`|AnuUbV)X;~W9Bt@9teCiLHwwGlYE3hIol_b~@BinZpO109t;
z*pR<uGG#3{I}!=2MA#iUzT;YD8>3e-a}r=9s<&3}wZzljoc)wM3T^&HxgTxI%>~0L
zlU&vztm1ClR?YN{B3hKQa!0~m4qKr-ZTroEFndW}MGS$eiqg^PuuodIcMMoBK>_wA
zsPgIJJ(Ehauc`stWCJ$`hF0=m&dbov1>ZL}f5L{nu@nvE=t*>YC?UWkz3UMh@zvh6
zOK9_ObSyC_d_i{H)lDduAv;n5`$kb4Dm}^3SYwx2OQ0h2dx;N^MPKr{!QR!^kenrt
z*zxG62-7(J$`)A5llTykofK;HWEKqW<%R>>c#ba7W8ohrY$72n@kk*KqK*_t$cTA+
z9=EAHTk>R@&dH<rgZn<~-xTQui-!cu9iEYPowiw@M2Yvc)8=wGAnGS&!ue}cg#4Y4
zUs{SOyHNE8#?`(d7+CN{D4x+5eU}JxW{U|F*2_El>{2yRMN{+xyOgUTb7>j>T#icu
zABF{$_q#_{Kv%q9Oi*m4T8)BZ%Nu;vt3Ij-RQ9fNmfOEztKQ`|1uWr&CU23O+tgKG
zJQ3WrkbkS+m3RA9PJ{OHWvhQzk^@-!?YA}W3iQ%5jFCNPx*^)YxS(|}7QeRkUFGg8
z+j);QzN_FLshMnDRCax^AF3w*{;;eq*T2h|4*`9~nsclo>*Lkl5T8#Wil4HU=!{U)
z3L1eP|7F`4bMq8Rg$L(cy#~?K8g1n_^@eS(08!{{=zNqRZB>8LAA!qw8@I$vWN>Tk
z?+WbM-4hW|x4kRw>0n>cCUwN{q4<I^l~a~f{M)1R;&V;sucI;QtOgXp;cAlZ7_d*0
zvEJ++B}`xUKI@A)h*KjL#k&epz~az`a$zLTvtjQ8^$@Xe4pnU+pKT<Dl%%8h$e<-U
zT$M@9hPIGnJz9aQu&Y`}lp0Z1Tu}>s^sF8Ap-RR;JC^%}6JGIa9IM8D8hDTAXC%rc
zWPdFSU|V-S!dW_*BS2NWs7I8$KAx3|mK%Hi3GY-0u>TR8^hM~}*Kbq#VJo@7Y&zS?
z?0-L<pUr7vtnw18{ggavsxl%>Fr}ASj@c|D(epo3HV$bs7iEf|@{E=Py5oFRuc)bE
zH=eio95mF8cA=_N49#T-YztY2^hS?ALB;K&uOsgVUTBNiivsdV)^y2d7O{$!szWB7
zqT}cC9@^zRFL-SC@boXoNCNmoVC5^(x?Da@<?VlAlEV8d?+Q{t4x)(pxNukV@+-_I
z*yWZu?j%T;;nJ#G;-y9jea7l;iR(cEP;7r`zWNsD?)MXO%<6Bcq!v(ow&s?CskUyi
zhFg+tDF!tKNw2-7APoZM4_RY;7w&(8jgwM#o%gH-B08y8CW-JPwmvI!6Qp8)63VLC
zE6>)pqNv~CFtZwpe$kSjHly0!7@rg;`DJVMUT15c_?T_>zu`lWHs+*v_k*Xm<gd4+
zdTNi+OT1r(kTKD=D3g9r9FJrI9k&D)EgOaUeS5AQerf~78K^Vi<j9?{?rK`kLEVV*
zd(L{J45EVFbuo8e&SxLOjDcGUf_N0Vf9RG_P8VN(I7~&_8cx{v_9OZ_<oiR!jN*DW
zl0AKa=7)W=y7$;<3`%$y+=#>wy>3T*b?|sVw3w?#K;C0fSK;?W_A4pkn1APC*lWlA
zdm&}^#jv&PA2eOO7eG8GvkZZ<++AK|^5eOJo{D(TCStu#q1v3b$y_}!*ON9C9Vj=j
z<Y7AP(@|E8wYY`yeL9=xdn-~m=&G6g-PQeN$xCK^8NT$t+uSXAS7a0czYgV)tNEbP
zk<3T;9<9T@94=%TRG<j==R`kbM;?+_2kc^0)A4zp=?V)jX6Ml02kcV9B{3Q|WXTs5
zm)A$^vcvfKUACMvM1rb-GtWbaUdeN|8K&I1iC8JU_RrhZXd{~ut<(#wl-0$>9Xf!+
zx~-}!r|NIEY_U53bEv1sR@Vi&zw;pGSJ`44F2txb^C!t*$*b0oRd;h8*Vg&vmqYCo
zlYJIezOgRQ49^~z#@0n0y^vt(f@z!T3WAVw%huNg8rs#vEpGN-)EhyPx}oYH_gy9=
zDP&8XlSXGy(7eL?{CXAOa$}C%KTs|~YqW+#E2x@(ZFBT5k6c`k6E;X*`3Op$v865$
zVOL+cEw|UjeIM-YC!;mIMGmBpZ*o$~Ze$@MIt}t*d#;n52Rf^HD6}Y6oe<#N30)z*
z3M#kWT<cB5?cHg8QANCc!JU{}4`jc+T{~<j0pW2G%R5~4Il=2ED};ZtLv?{X+$KOW
z2%IBx(FX@GK|LHxT8JH%K|uwCd|$KCx^s;PH~;c~@t+-uft6%e{%Sm`b~J%OnN6fI
z=Ru+j^AFIeW5ThV1k$WMNd1r3mAB;~Vm9Vhz#8luRQRHkv3>Y6V>LZqZRS&`%O||u
zXWMNuml1Ir0j{U&;<`&>=4$8ZfWu><xSr!|_Ki07q|L+;mHXywtdF2R;cr+FhI2W`
z5*3d}0=fM8=tsdXp0;_1>f!x%A?n6%mKSWH8uN8%;)@A)kK7Y%vFiLeMbyXidD1TB
zl*Pi@QZCTvcuNaHxE!FCyTSs?PSlrvIeNR!6~(Uj(tkAmvlZ_>TsE)1<5)ZqnT8Zz
z)!yy*z9?_?j<8^>-tE-?IOgteSlzqJ4KJ$f)$v6JmAIM5uP30aU$8arF8HXc2hrub
zU2i7Piw3T&^)>!xZ8YSXvdWF`4m9&<e;>S!t$Vkl1yDBaWDWxlREYPccL#F&EckEz
zyUXF~>au3<j@-WT6xa|m60zk8s7_eRyHnnxFh9t(Y@@e9t0YlKtN$X9kL<WNM-O`8
zble|Du-KM0;Gdv2sv6R%cubgfhv)A<YTLYS5xwF1?setcO;{d=r~UQ{o_;_^1Z16A
zA~j|a33SI8#cU8Ll!3fAmd0b)xZ#@K{&DN8##FGofp^DKLB4dj(INlH*m@F&b0{)E
z4v^Bi&kn^<rzNW|{N`yJ@ok^2eR``Mj_+Dh1(K8L-DtKcx-ES<Ye%xpCq8aR-yNuS
zpr15GxL+Le_4yRaP2a}HdZk4^UiBkwEz(N*u6aM$g}{qcXNi<OEJGpk!`SK>NVsvY
zi;2{Jn}|h#S%A}=j9z3`IptqwYel;_9b<|VRG-L{?X1_v^>KsE#CMOMtoI*w#J|Jb
zqrQxBh?0hC?0nSL6)jRX;fg%dI|+slEpZ_sV?@G#M2nQhN<w=v`^4+{Wn1*iMBWd(
zRCO=wT^1TjY{^%!q~o^Bxt~LLG8GfTW&iAnovuWC7YX@ATJgyguZOF#z&b}IA|bXV
ze(8kNNlAXIV*|0OFWRbG%PrsBv&-tL4^ILA>RaQAiOnlqXT4YLMyIvrR>v}mrzOjk
zSD;oKqD+RxQt|QF+FEb274o6+*1!UitrU62<ADX%#ke@cf}Wag4eUX7I*G5>$5#Z^
zfZH6uWGsdqX7pJ^^`K#bOfXgXFSZ0Q{>90}jo$X7AN^DQ-&$`C+{2Z!&9|1r?I{(8
zu(@ihCdkyVzI+<-;ufFG4tTEiTMG^)v%+85Ht!a1Ea93C?{<IjM%eDvNEbg~ozWx6
zT3Qf9)6>@Nl@*8pGlyzOxm*^Vob}xr?_+TgeRL0wj{`C1j0*cINdx>440&(FJ|al)
zu+K<=%^gb|$^z6KPH0q1iHWHlj?r~Fq%D`?VbmwwHHZ>3Ps0~_Lcn*_KYkY1|6>7?
zTNTT5<7;|~d5y-RS7Is`Qw!Nc9ly2Uc}!7izC8VMbv$`%&_bTf8a4{i!gx-<P#NN#
zZNi~eTne^nGUlLMGbe58*1(c7ts~G&3@%qv`L8Sb!P(p_;x0d8Gtr!PO*%~n?BOjo
zck8)3i{cnru-h#bs`dPJOIXnUd`!I9M6+lL7t4i4=5-+<SH2=AY!-4YcmN;Q%M7J?
zF(Lk<BGubsZUT-aMFPRbTuNAqQJ3ETqB>qmSQNNkOL5tEe-D%tQA2Z@yA*cjim&rm
zzw{5=ZN<Mh)o51}%yPa;BrLnF+zXiw9kf-q73{FHcaPQG7Wkhoiym;bSLEFDc-Q+6
zx4>C*n`0Pp0MA>)ZH_D4<kEu)t@VDpkfSx;R*u6*{>0XKv)S5i*%CD6cyi1@ZDi}C
z%?`4p_@6t#rW<Z6H*v2-A1${9$^v!|cd+rcWU+c>tk{~MASM*GDF3z1Sw+UfT>JzA
zxx?$WC1xTleZty(Ea_`(TXZFftL_V|pmjuT3`3arcJJ6JHe8kbBR}Qh*B#A+t}oc2
zH^ImL3Nb$1OkXvKEOCUsAq5gZjFDSx$U(y4WS0$Rd!976`#O~Kg?E?bsQB}d0P+d=
zUsRaadTVa6(c4lE)kVOxZab1|gLH%Icr?4?VSJV4lX>8=JO$#H$9yk$_1p1;Wv}u9
zDzrhZnhP=(@E<Ys?PTt8c=+0QHbdyX2PZbPm)_-9|AS55mSm(9WXU<olWr<uK*nOz
zF%HBYjH44%XS2dqc6cTy#a8UvWwQwbcYU9{UgmsB;qo4_^SL11UCQWS^SNBR`ysO~
z#8A>$@T3(dgoW%*$>XJ*xtLSJ)sawUtm@)ziR$F_&t8dfy_9PUDneLXyx*5%tC$KB
zjmtUqTmQnZ<y;LB(!#cdmQVnIE4L+2RpK2g2j5BpN+4CGik8mls&|Jcj`a4nx?J>+
zu3KlTR_8^hy%ffEtD{3+4onxM{93)b+`8G?-~YYUuZ}~r6`Nb^9M)v}a@4_04XYgi
zpgR-?3QO{Ytz8{N2=bw4tTAf%Gf_A{TNh;+-cgK9)9Sd+vU};a^;sn%qCF!Q)zyAV
zoXZ-Q`EKmYhSdc-NSzak=a!tsql325|JKE`RtIjjwxbUn;pWvrC_wgFTdoEBCX*C)
zZj0AOkt;n{yDwxPai6#4!Vor(q;=Ne9f6?vu{}R-ao@vE|0o}%LF<l5aC#As6zx*H
z@x8fHILJKB#FXcwFTsV@{r5JIi=x2ZthQnl4aI7Mq`33LIiX$HjvvZ3z%y?onu_?B
zSX(UfVPCb`+Cc*Gl|UAw-a;W8?MQU2x`-8g?)^kWKIVUJ>c+6C0_geT<%OKZj#p*R
zU{NFI&{I~6zK8<|p?5NVNew?`<59n4Kb15Kl6oS_^myxe#|YOZqpUo_x)nU@Q#suS
zF)EzS!ztq#iQdmTppU?r%=nNTkOAJi**t4F@FA9&%h5p9qNh>&+WD-4GLd^SA61^1
zB_@|J9FmU<IVqO(v@Jx9(`eQtAd>>?VnWovqa=c~MTba<XfMS%NQ+;OCnU-1Qht$=
zn*|ZBlf}#_yPOAWpQ~q<qh^|RloUq=Rb2_#&U{^w(O0rv8yXHwk*`)G2*F%usBU(9
zx%`jG@L|>M`M65~TMmxZ-5$>{Se-|euobJLmQ&-dH>}T=5<9Y5#1yeLw<nuR*lD8I
z-|k34ipbY&t^ZQ`5Txpj(Y*ML#Zs=jz1&o?mBQ0CRgIyPZT;;9zX{bf#|On5Ic*z!
zGJB-7Z@JxZ2lCLj|CuR^5_YV<(QD1twn0Qht$4aoo}3BU>E`HpoJ}Eynh*J5tXH?(
z9w<^M5fIwFJ4lQv+ve5w4&sdw?ZvIVVcWe@uY$L9`j^?-Tgu<meS08eL}s|+$RDmZ
zNAPT;)^~fMsy6MgfdKbZd9*bt1$frU8@fFZJ#oWbkdXau+R6GCGV_ox>1?eH%sx_W
zqhd_3+3d>UfLXtuw^5(Wwuio8N4zCyv4@<Hx!H&s<U%gn(PLRwnE#SZ!p3qMczWuc
zIquW=!l$j~wbP(<9}kh8%n9&x7ZPzi#+2IBCh{;Lfg#cPWDF-e?z3fcpS7u~1BoNq
znCYmbWs-_OQuDJ}#wiqEBZbCH?#EEq;3GC$of`?6rRW{uK+D8+&esFuLOUM|rHK+a
zDwmp&D?Boib6vRo+*&am{~xxhwvezuAHY6;UAby6`WU=wwwTBC6Mtrx{O^za-9LQQ
zmU5nhEUVtE%K?mXwUM_bFRX_lf?&9kYqx`C!HXid1zQ6&gLXB4cPUmQNs{k~d)qVI
z%fq+o4o3<?2Qur#)7t9ps5sef1a`UwtM7>Ow*`xMHqh$5o5I5E>9RG^4VL?|2jrri
zu!bm;+(>Ace{Jm@1x?J>9+k?zF<TYFs5qS3x;x5k`1}{GDcj55pL?L}oa?<ky-o65
zG$&N>`Q-|1xZ|9BIk|byTT4!ql#v{<jduiTaF8FZ(KWaVIi;(GHs9eG1+fz6zE~n~
zJ|u2yyCYD7oGsY>E#6vwOeM_%iNme^j)JsUWZM=n65|Lus3R9MXjT?;do{61@$bBl
z)7F_|?A_Dd4PDcn3#IfR*6Y963T<hh&k_>i6E+YtuJN1<<#PCqMB123Af3QYJLF67
zP)VOV;_KYrZig#yUM|^5<bgVxb3jgmb!4?%#fe>tAUm4#;^iQEq$IP)s%pd@XI?T9
z9!r?^qL<QfKAr`|sg~K)a$_)vvXcpma!JvJF7SAi6{pX@hZ9jIF}XYcJ!O+oCRSD<
zbYy2RRrMz>vKZXjbhd@FoB(l%(>j|g#ccX_*-X_{QN$o0d(mcNNhFKIUNUPy_Ptpi
z(+N8tWg-Q!(T2{#ddg?V%$e$is)8JKU6#Ax`|#l}*+u{BZ?@QC)RUA=LU6Li*d;$;
zU;2k8Te>42Z8=I6YuL-V85}ZJ`2dG(Ilz_4J+7V6ql85~fymuX*h-XXRZ8q{5Y*LN
z2kBACrg>-e?8rT#+g9D_=z<3seX1kax;y>p7uAB})t!EgN_0)6aVOYEf5Ymdo<#O%
zY)$-<<LoQeaA#m81?R)bB8Jh;6sdxA)Ec8q_+E>LAD*qdv)uQu{yox?CjWbz49C|e
z+{GRatk-;J!77g&v<(5DlLW<0pgfPST1z%k7Is;UZp;EqQn2R$ZDd>caoh>s9ucma
zC4!1=_Swis1QBIht`hn+IeAW^#MqK$(rXqBNu1_8U034=#<RZ_mk$unYhciAKBd{(
z0}oh7Om9DI&h|S4m+gd+?@TC3(9$Oi-yI{##Y+UcT*Q0tOlbjLkpzbC%Q~)3E((qT
z?-d~c*L0|A(1%F_uHwEyCFU|e59R0Xo|1fO#G&irrycgS{^)iaP54QyaU^$`2zMxZ
zj|KoDhvnGyJYwu)Q6|-%d>qh8+n9IJzZZAU<2fkjUNF|GNhmnAQ-3F8i=3N8%!f$W
zkKg#61w*$GLun!h580i}Qzw0(HsvavinI9yP9!I7I*0A`a0zFtVV|aRZ0?|0XR-yS
z`J~O>dG3xPlawq*FK8jz=5jSixpY1^4KJ3mY4iTE1Rc3hb@mFh*Me^g9WmvLcgEF6
zZ0a^Divf*9`54lI&A=5Egfbw}Z9i>G-p%09L0*^3*+^n0DoEVAoEy}&L*YfQWEHO4
z<DB6Vfh5a#%VqbhNDf!;3}PY0Kwypfaw}235Ne{ntonaK%w6ZjdxkJzTV3yX#N*eC
zcq1u$eb5GjC{k=q)IbF&zbvOH>f_w+@84l-z3185m#i^ce)#~>t9A8(w00;@IWRPN
z%`f&IwDsPr%x%iLOB&nkttE&gK=y|0Yj+Q<WJ^E+4_Np_1<l*&^VlzKYimNuBQEnE
zCBNC6WsuI396sf9Yx`Gar)*1o;5_tVFmdfZFuO5{&}CbK(TB0Tb<{gj9cL@0Z}10B
z=@z%=WXb-$XD>q5PHz}7lDl&Z5#i`9Vhm(f$N>6$B5*2WHjw>@DJck`KKUTDv@l%5
z0fOUA&#;FX7p{1Ra&Cp)b{o`@99Kq8gLc?Q!q}d~a9N!#eS*0tG2xLYi#&ChX+vQ6
zW=rWeWXG~XahWOXo{dGBo5UbMz)a(Klobz9)?QiUYI$n59@t?gV+CY{@Vbr1q{QaR
za^(aXGO^DmZKA6894dmz?BY|L-KqS6GQ_46u%f<oUy8F8V9KcAEc8r$b#WhsX=g8H
zvzEjSaZKiX*+i&I;Bh`zQ6?AIQ@wKw;E5%e;Aq-~`trPO-ee2e<lvqjn9++dh}cwR
zy;mI2#h98bn_jd_{%f}OpzQsZ@)$hYgN9g`B0smu$$9x55J;LMdAgK%S8{%X^3j1%
zwUs<Td-j!Uc{Num$PH=E?XH3;@b?E(+~p`i+}IPvp^4P(ca;mzckb!xy8=HvrpzSu
z{t3Y!!Lruel~<l9*1b?#4cS?F0AvsMjIGTUO|bcm`9nd+*WDG@3kn7?!gWfqCLdE8
zDX63M(W~zHvuvzHxXn3IInMsnY>2X8s)`J^<*qmY$Q8)I$69ZUz9mV0+FJA1?30YN
zxf<z7yo#kie8Jl83VflQthj9Cau7^mzqRLl2tsMwDyUzE&g+OpMx<@qv)U#A=yZ_2
z(4#!eFg6G?{WHiBJyCDg-367=cUPc;?jG!V$u?eri#?(ZC1?b2!kc6`KVlGe#t!92
zbmYkCM-n)e<wZLjOYb%?yhNZ`U&PV7;t`e4m(;R|eF*vB3!@azp8Y|(j=P+UlB^M-
zH4k#J^O5*mn_|a&#M#<`Av8~8)zUtPas#3jHsN@lpHKDdRCF+_Ws^tbws-QbbN{Zq
z=d=HYP{lSLQ#uVGFYzLi0a};{>@<;!C}UGpf0MZsME!Z%rmCeV=)DA#llTSC?*Li1
zSZ}eI(z8)psyeZ-JmG;%XL1xG)ajhf-c_F9ehAY!|Fa8?>-jttytVUp74*^mpk2s?
zD8A$D`$8UqK@1fyCL9<IKE-I{46?<nAn(gdxom?Zd9|ek1ky#i&CAuwO9SG&yj*qp
z43xf$0ItN;bbV0$$*Cb~uT)E<tcurP)T>oL3I|&9#+$t-5XM7F$G*ycL0Z3#P^a!a
z1@mxR@&Af1BIw_=`u7Ab5hCfrBWlfi96Jb{X#gQmhByZFy$xAa8AG(xYv1EIEbWFw
z)iy?%8v%H%#B~W-yHwjv?<pI!A4InPJ-!p0tT~!VQv5Z>lY`WT_rxXKCFBlu@}8iF
zK1(>2ja3628aX5iInnCfZC`Jj{jb3;YfGS!LZPL{peh<l3NCK*Gy?l}ul|Kiw(UKE
zb$GjBfjWFhMVf^V$?-RRZnJgfoQ9-X>h|>|Xcb9BvkT}5aC2W)#1pRD1}d;8F|rd6
z`GO7k=);&s<ksLqHg1O!(h|&3YNJUT@quu)-ff596SSyXIx0T$Xw^ifY_1D9;_Kay
zQ{K^Pbg2auY7RAN!ndyn!OmFJkf=zvnJhbY+~L~`3*svP+T(pb83)C+97UbO#tHE8
zY8p>Lo=@ZeL&DsX@lCAq%Qod*e-`z}dy>oV`mz%L;JtATlwt!;f_a<CBQXGOWjSZ7
z3CT26qIx_xsxp!@v&2e<em=Ku0A2fhRB$h%lG)xaRHq(|gXUc!=7n74^3*=I2XuKc
zI&j`x<UHouIcJ#QC+t$rd9VxL+@<USaW-*4bagKK#2*>5W&i7M|JbhN5{Mr!&P6LZ
z5M+;Xb4t^A)$4y@o7sEIMS_8`RqwrCaT~l+-FwTW?;<3%t$uGvEsC6|{=LDE6k-e;
z5j1gY-dpbC=eAmd_uKU0kJ?)AVz$;oC|<A7jc=z{K-FnY(UXWt5l3RJ*1tFItwJh;
zRGafRyt6|6ZSao%?k}w6y^d>6BV&sw@qzj_R{biYy{tEh=$>=HSQ6`q_r~d$5NQ9N
z9UR^^hpQju;g;wbS007FL%x8m6mfY7|D3%5>kGCm#&O16_wjY>@JVyZ5nlS==tIt0
zXS9R}g-Gs>5sK}C&s;Z(L)!b^K&iMyA>||Zly?GGPco8h1_iAJLpvE$9&+eE%(4&r
zNVBz`ZdCh+{1-8bc0;0$1jNG0XLR9Unh)m+cQXdNJsOZI<r%h;RSOR-$-j2=y@BU2
z%tKasETB?%95m4HB(O=7IF?;QB^}ti&yMFJaM#d`R$VK;Hcp@ro+o3r3NAAq-(?Yq
z-WD4Ed7G%p5RK7Bn>qR3g4z$1i$4{>6d5=(B*{F$l;j52z$|e>-NPKUr=5*XC5Oue
zoryB-s`z)sB%BTKPjTkY*j&^IlMU|189SdVGk8$ZhUa58kMO2ms0I<q;yT;~?-0*B
z*??X2)#A9k$p2oq#az7|{X;`UNV(*_LEquRkpn56HGyM%3D6ZFuv$*QL=bVT6wdQX
zlsTPHY%Fc%x-toOp$P42LWItQlzeOQeedck7MrWq6s)nk=f7Cp8b<@dVA7f?s=(?s
z1!?qy?W}%{BMqH|Cu~jpI?YzRVGV1_j!7f3*Va}YO|hj4n|Rt9qxuuPu#>jVyY0e&
z%9^U$<R1oT8HcZrvf_{zQl{o8Q-4$CEqc>7tf}5-!tq;6&Tf#mMI`;kstwqk2SGLl
ztqv-IW-<2M?0{fV^@z3k-!OiV1u>tZ|7myF_u%A4O4w~{0-yB^V|xZskYm6Rc^cnm
zZ2OvczK99q7M7iB61P=SQ7*bX|GKM{d6xIL*MT79<^k)Ao+HpS80w@AL|Hn}N`NpF
zeG@;E)9$C&3C!UfHH~ki<|<hEkdKO8axVvQB<FycELR;;@M$|7aFItCCTG;o$ydMH
zWJkQR#7O%xRpe0zn;_pkcFg~tt^NCE8_T+1yk@d`+}C!t24`$F{{eFX%YQPL_EBU2
zFk#~XmW*ggKv#rG6W7Yht(yY*PUd7vXuT;1l|(f*9hI+J89N*ELd?qQ&#Wo$+?Vmt
za7YjVq2-#3g%fA?f*;B_KIgsnJxZaO!+xa)$;gFjESd70LX7jAEkv0|q!BqZX&0-q
zGw?1>OE2a*-rIF>2N9*Rt72n?kR8f=X-&N8i1pC;xvA!5hgr%_lpfiN=TkJW5aE^o
z=>SyDC$^GnT_!weS94c(U_E4a=T(0e<<F|SLpXu%;uX-g@*ULOU2Z`t_HFguL5&m3
z&FZ5$kn_Z@@v3sw6fa~AcNd)UAn_(_Z8qyUh^a(2<HUW<*4<q&lcXa})rYgtlk4w}
ztK^DPNy_8RYz{yr`jV%mym>eH=p1x+p>DZ5t|qP=J%Y(LMyo=(fv%o~N^QNndbCL)
zhRA+%RMLJaDbZ<+ecF5?+a#mgQVos7lkYMfY7am`rC6q1?*zH+y50yV%5!+yI_?f!
zMq&%j{`Q!&KE7_9RadU529cGc^Q?7ySBS94cnSL5heEpd?l>#`#gA7n%0cVPh3Exs
zF~C4VjA~uayJ)qcYN205Ejye`g(!I+sZ3C!`gUQ2!W|j$>F<KzvcuW=;K6-*=|}zB
zZ0(`_cElS!NZ?Jv)W@iDgIA6<H0E}!8kGqDHs<@jv!~CFS8Jg7(6#)9F&Xi1eS`wm
zcy`Lm^;~({PuWDS%dP_?3PFU1PHQrHAjLcc+*I{T=C#xQi*up~8fP8mgL`+t&duB%
zv?Uwu2@7mC+F~V)8eq;r{$r8{o{tZD7GJmdY`Fu=m3_rI;lkYo;Xns21k^fG&jPg!
zi!N4B5Tn}`vonn7UbjmQQx^tqkPF#!(Li_00pW7B3e#Bn$Tw=*mZQv#N&HQ}5@lj`
zA%xH+PTPvlpbLT0)!e$K?Pl*QHy>2r)rYlGc@0;287Ia=jlCgPt@k;;5Tbs?R!0rJ
z9A5)pKyzEY*C5zsF9gk+sDaTOmsg?(w4m?vYwa0NJid#+Vr$<QFRuHQTjTrUa_&Vh
zD~z%(x)W(9l@FGw))Zx8UFG*FyTA3>5mDd<NygF~b&8-VaL9%PjTWMz#3Zx?Xobe3
zFD1qjGJoARzGLKKd}ggxuV-K$^&FqE&F_mNgc*(G+t592xgxk@AkciDb8ld~_V)!M
z*og?M&$eaLVx3u<cEkv7s0ssu6ny*p0s(J@t?2X$n7q4vf_TjC>=p-NsJUKWi?>MY
z!2eEKUqUSdhcdfK8%Pl0TX5!jU%}hp0*ouF+EV@CAO!+Dlo0$4KK&#9L#EN)cG!RH
z0qqRhXf`e34t<A@M43X5m&`)YfJfs@!B|K=>c2=rqC6yHQ6DOZCD!4dvf~Nt=B}YG
ztND;J*FiRsQ^zP6G%%j$M<O6?#Y8r3CwQ$*#%>qT;+VS#bIMnMK-B0y0t6AX#YCSK
zIP07Ez*pDXOa-Bwn;g5&<^uKh57=C8>)?UiyFn~0)A<Awi&fsG`MeO&J&GcthsBrc
zLUb;de+ar_YqAh!1OerAmK2tDF(;~fg7u6oI)GpO>fhTX-~Z<L&z7=J?h6!8;mt04
zndHqp!=8k5IqF4fDQ5qYU5PTkYvMbmY^AEFBy7HZSF5s<5}~4pBS8N>1$DsON;SXg
zp19gFkQw9&sJkaAeBeiLaZivU5?pXf<tJ8u&o$P+n>_NY<eF^N*R_LA8lsinQee@8
zXKS;{=ZnsyF~QhDf+AZN4LDh#NO|kMU`<h`U#C#fD16BJ1cx%koO)crChjTMq<<gp
z<%R@pFS(E4|8Y=i@qt+37x2E_=#{bABHl!FCd>kZ16?z1o9`*N8YKV+qAf=&-qo_H
ze$BR2Ww>9)@o3LQBs2mPPPc7~N>bgt$YDIgx_3mGD>D_)Y23CuG_$o$NHp&$2uV>t
zyZujuRo0u!^i~aKr0<@frJx~)3hOsejpQ*W<*hrED=qYmR5QCMrn86c39Qd_CG&Zu
zsK1m<c_h0+1pYPF7Q-|<9E+zYxo2!NJJz+{VMiSLoeE5H)Ir{~?IAmMPkC^6l8-ri
zEL;vzb)hzndquX#Y5_`mB;)T`@F28CG?ef`l(kZ9j7M37k3b~#tW7us=#vOr6`(43
z?-`rQErEKMRfJ9FDhwXLYVK@y#x_VT^Ew+eb5Fc_q<H8fr`v4QDN<s9P<Z#b+yGX#
zc+1)OEFrB74-K0?pMXHBpe&7Eil-sxMf&-LT*t2NE@T`RtAS1<YSH6Nrj`o+F8AEM
zR2`LVwv-?HOF9;W7x#qr3X2$C3*?sqWZ8$25Rf=3b|vS`gYz7RQqlfbd^h&&gAA#F
zDw(FNA<#q#aJH(U+%e3`u&l^0m@p8Aq9Kl?0{9i=Uf<v+%%c{G$j^JSCceP0l}G~a
zhAIhKLqnn=ZtXrW$l8X2cM8udR4Ai>#)d#mo4d&NVe1;oIU698j6*kD+qluzH^j5t
z!<*aO5L-Hk2OHwV4f%01k5Eg)xd;OLztFSU#vB+84Fk72Y#bX%Ck_Ug@@#XIIg#eC
zMw^3$<>bd~i;s-QH<N3R%8)ljYAd7XZ8^q{634hB7la_nMSr_J){Yy?3w-^V?98EM
zq^27UINecO4B=}aU`4g|v)AjZwzJFnDk$Wy?;sB}6eRSOgL`erx4F>CXKc7393R6H
zSUb(jekfP|u`U~l26Dwf)f&Hq{BRhJv7}jgQvx-#&PSrmPlzlip+Ty8QpR!BsdwzU
zGU*<5ZpW(mNy$)N`<TJ;sO==CNd2d+mdD^h*9o1>I)(IlqL?9%S7*}$k4QUJ+*WO(
z8ea+PWu!|K?d(Bm9vHt^f9Abl!~fi-tJQcFt@zofuYD_0W3a%?wLZ%GhWglM^EN$n
zU_Tb^P?mGPUk?u1c|Rc~&Yq|h->N>Wh%V&FkKin73)yeEWlj!XjM?cahC1R|_Gk?g
zwwUK`7Xh^Q+a>Q+!Ge~eAB9%d8Y`dm<*IK*A(iRexGj4V+zsqXOj=$)gtv;y$g$u4
z7q;g~-ruX&bt;FmH<6jx`{VM<sH+FgdDZ(1wt1|dFf4WNFL<ah9;6F;e;_&$#n9DG
z+1J0noLX`5YrJW{$j*lM7aYRz=kd~{c}3Q~KX3@zFrJ{ssG>Q(6#KaD{ROXp5}T?H
zCeX$e>rz|){=jZwMP6x6toi+cY#x?<L$7Uke_$dkuB;`;C59R?oj3Xz?VGLDVgCxW
z*5>R$S}T^JtpX0QC0AZCrnf|yF3roVsW_4Lsvgls;u>tr@xDkt3p~txS%~ql#kS|H
ziieQzYiCx+Bybl-oiWLA4$4!Y6j$F!&(`bP!8LdTYN_x2f&66=FWY?^i1k;@<e?Y{
zs<9XbDzwpXl<9iPhlXr6b|?o|xK*-uR@0H68Ut~E4p+@yfS`1*{ZTIyv8?EJ#VPcN
z58p*5X*(KiuHCswJC;M?p1_MT=Cdh5?*FSDk6F88;WZH=MqSnLG8D(7OnzwDJw)hB
ztorMiT#Q%s6nxIzt|nqd6d7YOcNnh{39+W~<jehsSI(yM%=|U7^0NV5?K318wHfaR
z&hmiG#yC!2ORjLr=DcaiPH5+2Q0X4$vm1=f$@x=(NYYnMypokKWDUGP2&|hxxgL|m
zKmM5;^f)BYrJQ$3wt>MUWa3hGiVj1HJ`@Xf*+GD|{j4n~yl?*#n81zau0*fW5JQOM
z%HR>x19rRWtsm<}nP~UAutNb7p0ZW<`mL4~FYafcr6;YfDwFwUN!nqnqs)c*#g@_P
z?=6T01pRSab8lQvq<GeFZ*o1*%ssX?Mns~Ckgl=n{j?%l6=D6lEGt>#B>x<@ri2sz
z@U*Ru8b~w>_InBfy4h<$)ev^jHbi~+S*Z3FhevXF^eFL3@7I&on$U?DAl`oVg0uq1
z(|a8u=@f{pyXMfnC7X50sL9%W@?Y+Hlw3dA1;jgfRufe0*9dZ>*TJ?&w}md?#+{G}
z4ACymSm(V-R6Vw52t3yvtp)3ef-FE2KGwt1PW8oiH?83*fV>^Z@dwEFXG7Ve{Dfg9
zhOhO=(v|dqcF23|SG2Q{dkemwt^K1NzPBLjFZbK%I{=mmF$ySQ89Q=sU@1Jc_u&_^
z(~jO794z7*NJ&X+j(PtGQ1{wc!b30#h3Gnd@40%nu<qtUaZgKr#(h=`I3xZ&X)B+#
zld*HsHIbVm4;e`^ZNg#QvDYSj0-wVBZ0g>4{FF-KQ3jjNrGe({wzJV4Cm-@)#(~_g
z5E!$$y90ZAq(I4L#39^k=W{lH@>e#WBWw*p*oEAX;>}2;7Ea$nuF(JsRLJ;?QOnV%
z<FdsZge>G}Q?G3ksHC`>+ENaT70~7C`>P5jgp?5ZXpSvrkTz4iUstL!@=s7tE;1`#
zCi<W7j6?v3V88044>+4;_m%s5U{Jfa>b~GscFQTU?!H73vR0E){iLnF@A`MUyec4a
zf^=DZ)Yc__jgu~A?;7uq8cNO3a9=rZ%4f1RD;FVtAsehQ2QM)e*WFi6<mdknYx2Jy
z+-mEimE-}4{)IEF`My9o+sezeA%7yT1x>eHHw9FT#eR6B_xHD7vex*@mD4MV-Qh{w
zd|$zBc%Na3;cGKnlehVnm^n8Os!z92p_Jw4*?wO+dk+mOU4Z0r+wODQM%TjW-4}6t
zEf-`*&U91XV83mzU__$LtRzI}jIu%+(pJ$gx??d3s|o4co2_7|MPlO<X??!HifcCz
zmG#)1w4wXr>cVeh_11&6=Y4^FMC<bVP&FX(Aq|KqY$VD`Tnx%2iazW^Z0$i;5<B#H
zJ94ceOqHA19<9nWdtazyzP56ywK4B#KRyP>6BarCKrz>H*^!gm$#)#Mq6q+j5{Ssg
zV~VGVmq_^S_@%@?SYo`RmE_+NW<CcKGnEr%O<e|g#-<&X2MA$*Ha<H8>Qvo~cddja
zLKq3%ChqQ>|L7kgg1Vj0vxoE!8hYMm%gepfF5DNG5_%<c{0qK(f7)ahv)%5ZYFf;n
z!m21$6C#;~aXe%kwiJ_9u$MROvJdpuf8)7aPUv<btG$v<(FQ9@0~DF9<YKpzs018-
z)$5>#Q%KIiVzg~*%N_W<w2Z6P#`}(SsMYxoR6=}Nz4jV<7^ins@KzsX<s~mD<5gRe
z^`N|8P(A3B1|MOzw&Blgt^fULF-whW<2HA76IaRBc~i76g-JnMgi`e#LpVZRzt%~c
z@vol%y*I}%RH9I28`j3dUhZuX=*C%YSsQPL3|)2KH+n;?U_?Z=t}W<l``=-RnLUDT
z(Vr#+ND9hJ&9U0DHg4Ds0tjP!jXET13GoXbD*5HM=tt58GK)qHxi!2jjDo;zyDvrw
zn%C)ngHdoT&211z)|*{pc8(`>F{$j!GD)#PxXM8^5M_lwq~17m=THs`N|v-%2yD1&
z>yduNq8`e6CD;OY8Z<JJ{n1uO@l8YxhjZd2u<ODg=%9_pf~VH$S^|<IIf(=NcG=M!
zfD0?-r@VBI`HWE%67(*QLXigBac|q-C&6%SZFw9%(qt#Ao%tF9{c+ziV*A)cc8|HC
z8<(81$sB(tv1q5N3LcFbUm=^0-ViE7BA<;KqR1qEmR;4%J6aX;$Y!&Q)4{_F*@|O&
zmMM)wEc5v&)8(LEq2T5#_z6m87qV5^97SY1WeaQ1-BqY2B=P&<CBUADvJeYjjLLew
z*zsZ*wiso+bes%H<e#=n)uDS6^XH}brTB3q_3;JA=H<_9IokSxRj()QO7<t6M}WLi
zEz)b8v8#E#7z^f}(<e)AIhol996#tic$Fm(kG<*x$=rMJxw;P|Pcp$-T3xkpeJBDq
ztN%dR>OcMqTjO<=cM!DL5H01g`?{^oZ{+00AfxKXtq#t@`mXyx!DGbKfLd&d3J^(9
z#`0!GC*lM3l1sh%It^>2MVzz^*LC@#L?L1|qAU)2F<jZ`!$3^&RJ0~kp9Qhn<_{FC
zN?h(wTH6N#&wUz>pEN(!r$4v$YINd31DjZ<ZMkfp9xNVg9bW(MgqyZUeMh;TGp#eq
zw3WK3;zYVX5E!Z(J!1)E*z2$k6raDoYBCdGX&v^LY#_>@F3{|pB9Hy+HWcfAMtqVD
z=M2#rVdCjvAp(2MMiMNHZ@~Jv=Md|rYWot*%V@TRk8HLhxgw%e_uJ7A6y!_PVLO)f
z`nrh!GZy_ipH`OMj(cTz3aj~?NS=hT!O3b)(uw<1F<!wSllcjU7_~aNKclwD9bI>M
z>P=;Lo9J%3s-R<~YR}o(1Sh0~^75F;GRPEJ$0Pi)*$)H(AlGkU^lqEWq8&XXS1JiJ
z&PPj6@GHDCVe`KC2sS*u(1iqP0OyW{s(}&|7Cok2^f|$+_S#~C5{cWT^}LkpNkC`(
zJ)#D_g%GsPm0yms)1W9FLFEuxjx`?oie1T*R4fU_H>_lrAgCU@nx~Z<A`qNrAB=;K
zoyjWyTb?Rbm*wrBwbdUC#IOtBWR&*xA57#?B-n)kUGu>pMWC;&;e-CHOPuJguF^H@
zHCy|^SZegJ!KyWSOT45ZAlBtjo6+*@+;2@WlyYC_AtFM&t^Z&!LW4xkux9@S3q1+t
zy&*;}qcP^B#Jb2HPFagj2tw?nZHzXLBiDMuTB|XL=@xKq_Sy$zPu=#xa@@ONS+;yI
zsFFdlLCD#nJ%I1-29#~lk^8{A%S?wiM&)DM<8#sV#W<bb9%Fn&b=?&pa=B4>5~K_t
z49q~7?EXQS<n&dw1g;k5NgK!|dFx-WdKvoQxpn_W?<yO1_^_I_L$NAPz@5HgBe|hF
zNkw<K8k3aZ<+8qEqq)fv+H$-V7RNXFNJ4`_H;(DiEt!H~1&&3lLXxs+qMM%pi$8B;
zx%CC@Vg{bQ>>zTS)$;33&IxOEGIs(g`7Y7Q<I%DBF#tgwB=fk}AloJa0Ny|Nu*qsF
zZejf_tngG01-AR7O((#-tAy{*Z=XgQZ!_5%uAVoy@T{{58oQ{#fphsALlJG*`Dmjh
zl2r#@rZ)e<cw=zF>mit4RYl=bivO)G_|81=R~zkO&MhQy%CD`bZ`>B6VafLtWtXC?
zFi4;{$<&u}H!)nZ%TXaEUh=VC&S?^}5MTAG3q|$xU0um8kg?d+sF2UA^bYo+KmqNL
z!K*&xC_s}IKECD`Q1_u^O%&`;DB>AgU6r9M;p8h6Mg50ji7)^rl;E1Eq7G%}p^zyJ
zA1cTTRC&PG=IC3x9<#;|1;*-Dcv(^eeW+aFKW+LGYx+=JVZ}C;cl-M2cpQ<p8w8)S
z=B$MH4{L=%+Ypt~c2_2bEeW8EJ@(qhY7TPZ)g_v;)~q692BPWZPu$xcMfq`E3mF8n
z7PjR>L3AN{K4b0vYtOFYLB1_#h1yJyREI<9%5TXwwnt-yp?}#reKbxX->|swkFwu_
znd?nZ_V?_@KB+HSiupNZ1KAIR|5G-U4S6!K((<Jw#ef}(Jr>*OF;J%Ctc=9WxPbC<
zg1dvj$Pqt<HS{Akn)~%R)Bs05bnf<ov2dZ2t#paOax}Up6&-TUl9aXl`-UCMw!KJJ
z#{yW=(L)p*&r!Mwib~*3KOt}ZpAc#%y_J$*lQz_U6yIT+@E@4{A(@@bU6HRXSNwJ8
z&#4a;LibafY})^Nu-(o^J%!&baknABW}?hby||6DHtQ|^>aT3hm#&LMLgy3A-9+oc
z5j~f9CuutELhiGozv5uA5DOuTb48?Z{idYm7rh-1xnBB3pN6apy6sYcBKf0YWIhQ+
zwUlKn)tgLBk=8Cdcn|%xEyvg54$BRDl{0iD%An6gjDP)QFz8B@N$B`$!8uRc)$AEs
zM)4Gsp3)lQ0wYWACa+JU-+Rv(u6zHA)is7}V3t+1)s22@|1ZLT@qbp|=r|-vTq27D
zf~dGPjb%S}lVUu2EmS|JPtiKoMmtx^IqmH@tg*2mj3NzRmo19`N(!CFt*J3k*kkBq
z)_VoXXG;vAehyo6HY(x;X<cDId=j&@|19srmd1Fnxn{DTK@r**7-v@x!PwZR)*Pj~
zf2VEE9(l-LV>PABZ>yH&Md*_)u`Go6LKnQ(+T*JPyl>dH{OX2T){$+w$xqt$=-maQ
zB8#83PG1C6+Y{F93;F3y*6To#DHD9zms{7>r#Qm{6^!V&*#ZTT8p@vGW^rdMy20TX
zn@61se<-%0EIl>!YIz#;5r;qqMs~O{C_=<^K}C-?7Hlm8mLm?;!}z#ifEC?8F*^nO
zdn^EOO5zL*&>J?EE89oV@#EPu#_!Ufa_egeAkofnBg-CyE}52Io$=hhFMtcW_M`im
zh-qk2dcH+<`Qtm8tMr+Y)OP9}^DS1sh?X$EvT3jQ7U`wm{--88o8!W|$%E)kZo-Uj
z#KW6x)+-Znw_w3JuME?qI03G=IPdF+s20_yH+*!DUC7DrCdukTjG_2(Qkh9Ua?zX3
z*8aTN7JVgVYhO^_f=mA86Jqw3qGd7q0Xnhpm$U0OOfi<TLc|uBgvPq!K=0Y@_E9VE
zSc=ncSAP3;H5#9R3=-R_HMjd63y>oNEU@Z+#}h7zk#)~$tGhoAi83D^DB+n`-(Rlk
zRt1%``e-9D+Vjj-#)xa8><QLak;)V<u;Ko4|B1XslA!qRA{7{B+B8O4(U6qOIAQDV
zFSrJcVpDuotghFA{S*w_`uhvc`qTK&n(t4l1#u8`gHH`D1vV|#lG7j_5L~5R<&EAu
zF;`ptug*@}>{A2nATMdlX%uS$$E;f_&`*@G0ZKa4?o;D&!#4B&z&PBf60W=>p(3UL
z*L%A|CDIvzOXvNLW_3tp2~lFL6>hCN#wj-*QcW`I-Y65Ma#iR{)>oC4pgz*3*g$S7
zR}INTjFGrN{xvRC9j@jAO~uwpk#Q(jjR4I%Y~=oe4)!3$KOB{%WYV4jtc~993LUXk
zy2V^F3B!2wI6<(BM-mX?MoigJU)!D%L-SaUjpGP;dwCU!uNT#&2!6qiR|6I&wjzwG
z-5<ot!zASt>8Eh5C;c2C0K@!^-{aAZk}ADu6S@A~{Reg+&G5Y%?8gpesv1uQ=hGFi
zCy?Nr^<VNc!c1r;VNtYSn5Ji`Kbx&3$+0=_x4jGL!uj|jHu99sM_XO07s0|xrFy{+
z+<*LITgVwfX%BEJ_bB-IMPGKzjBN28yO$?Ki+jmOL<4{idO}Mg7DDppFY;_YV#|3t
ziGg?J9lc3bBy$&q<XiF0*|U%2+;-J};2`-DKBf<piwg^kC;9{B=A(gu?5}$ujP`g;
z_V*L933LG6xB3A`BAilmutjQQ_1?$+19--+@xRf!_FKaP<$Nhu%Gw9~D!|sztO%tz
zMG{;zJ`nd>X(&;CM8$%<sR)uOtjuX^dLX0|%4cHhz3)Ab_FHp+bd4=LVH=`MNyE_W
zL2^B3Ez$M$0oOd<&c>*n@tG8duGNR*F+T$-viX7Fdv+9Iw6#TFr@#JGh@7RQR-mRW
zUelQa__g+G!CoyQMsnxcwg-Zg*@M)t7=3rdXtJLmmsl()F~O`eXU7+pv*Siql}+*n
zN&Gxzz1~9s54Uyi0U;-D>%LU~_>cc+L%F(;b3(K-RP=2a&Mpg=U&8v@p`0|nU`n|(
z^8XU|{;*wMS-<ZpC@3h<6$QFNSKN$~7~&)*aS|tS5(5TIV!#2<JG>gtxfpD)CkFfV
zpTx`%$!A`U#F5ZBz7o>ql_(r=f(cG=5>H|hhoFfQQBY7&P*Av_prD|jprE*)?_O&?
zkSUJzHuuhG82_GUKYOpe_u6Z(|0{sRjS}~S<NgH)bM%1HYO40_2_I^4@=tcsk!jfL
z3}){#696!JaXaN3|L4$Foz5oWSXoPzGCPy2QoNxjpDOpojx9D7ZCNd`NLm@2j%VT{
zBqb_``An`exFwCju+4gpq9#b=bKYY&Wr=;g4h|`tEP9d@e-;#1*{ZAYEOF;+bjypm
zn36_VOcaK3Ag%@6ioQH+*K-*+4i66;=(8KXGNhcw(d?#!a1>$0*KQXl^!8IJN-Jm=
zm!73+a`JqW3)=;|lVCFZ*Pto8X?H7-vaWfS!?~A0P)$<)5ci{{-!@Q+qILP4MSJkL
zrTA=nn1lcw^<knDqwvThUonzb|BKo0g0ny+A{<rju4TV--a~`}KI3<Vv`}LhHho1j
zFE%{r?PY?}R#wjxD<U7zE4J!)!JPD{=2vktSRKuDQWZ9*OloW5S#bxTJBjFPe^<!b
z&&QEHYJL~@SQUvO=~Kb^G7+A&+8i2>4*3kqW?ARZOU{5O)kh078AznUH&T56F6b7;
zRw?Az_`AI8#kF%~7P3RUl{Hl#>mK&V!RDNqJN`?>&=SpDijx}VD{GBs+9bIXD8XCX
z?+Q^RuF%?jQOOoH2y5aq=Eq^%>ur=pt}`DKG`;V4<pBMD(7N+!?Nip91LM3R*;Pz)
zf6T_YiYLT98}u&;N<CsDKFs>{>ufC6*srk256aa;htS?DcFddZhp!%wrha1<gCL_4
z{mZ`!YdxN8Go+-5u<G4*B5+ALBQMy=09gA#TE4>RPW&$Dvyu?vGJDFGreyB5)79R{
z8-=<6?1L02|88fhxqmOC1q6J`CgYhr0i`mO9&_q<aUf9^_bW@sbaV-gV7py*aY8!d
zi+o_vW_`}}^{?1mwLD(H_Y&sOCzi_Q9efhocTpwafGq?J+W0GUkCVM!^^RHW#e_z6
zB@V%sUW?|E;iYJY7})h(k!=G5$~|%;PGAX6x>?OprZQI*-pWpj+$sN^PwjR*!?szH
zL+H3ICDygDnRjAm(8?+7Jyu1@b;2;iG;xKod(jc%tWrJ|sp)=pbck#6LGHS|8(Cpk
zjfeSuKYE=<-T|4cRT|QzHLD4ZU_aF^Y*~#HA+f0EVlS8HGc+<1)>Oi6MNK&cFVxz~
znsSunV!f)S&?m1Bv)Wti*=}p%duUX#38@yy+M48;;8tpaw7?}|?S=fEg?1?!zV_@#
ztIh9`(NQoJa#L61l)}%J3%|v3Vf8*A9Kb~-8)_<jhwy@|O^sD=x|nowD50Kfum{&R
zYxZA^F5jsZhjg8^v#tKOa-CRPO&peO@E(`|+G~Q84IGe7q9Xt-25t85an0$yF<mc{
z>p_mLI};;Fae){fTglIVvwbzyEygj%^wpiaf#@usdc{!pCj7*{eqjB+4U3aIaDp1F
zz{0_dn3DKfnpA}$Ou;;kWs{Fe%DrPbF_e4CkUH)giMi{zPk*Sd8y7Ag1pfB|S^9i@
zwAotoiR_?Y83jiY9^xDAbO3NZh`Z|Xok;+&o&f)(x5o>T6F24UqokwD@6*wT)E6Qy
zSbc1!TJN)kETCQzs>-C2C{Vmcdk!77xmaP$v`E$S)t96}5J4%~3<w0Xg@lVd1Eh_-
znm>0BkbcY-6K!!0n<~d&?hDd26A2F+ajlxE+{IjizwU=+sOv9wBVPMXgv{fwNBa-j
zt$<NEx~N1FjXg@DI5gXm@5V?km&l!5azYZNKfYTNmwwi(B33@TKnLnUG+1V15CsOJ
zC5c}`#w+$P5ii`5?2+TaThE$3T@HV1hb?<LhzDuGdTsgBm3&fsmA!)PS)8_3JRM}E
z=!nJo;`RA-92oSk$T_Q?4n#+ZB5yKVoqdyZf5g^A4|$$MSl2%N>w5I^)#>`D3xWOr
zXNT;XrwiHH)Atvum_410kzYq#0d)xyfw{Qh)jI_2tu*BgPygy!o*$P4hU7N-T5KYH
z_|wTDDT%oRVB%TuDjSCzQY2?{y!R01C!+nq8achRJY5KBKjx3trwbwN$EZtUo`6D3
z#W&XO;~}wi_;Xx(B=YTz^>K#4ZI+!rTcv}recq+2ksY*de?c%xi}h9uD;<I>CHmtT
zxQX<uWAeQXKArRqodTEEh_6)_8ZQM_j8$tTmziS@<<<@xkFKy57dxy}yvM8VuW)h7
z3`))91c@wrC!;R{F$pc0$W5WzWj9p^0?8_0Bd9o%5O@3O=vE@f6pJq1<{9r6YYXbt
zc#ma4Y^xX_n~G<uIHSuJ{|%dt5&Uiucz;J5G!AK=@o5bowArWQUcm?=;fu1TPlu>q
zlvtjv2K<=!b?+l8l#G742^&W+)>bn<Y>T;i#Y^y7extf(*K^%TCa0K%8`1Pr812vP
zrXvlM+HPg%qer^zb}k_@$v3tXGa~gYH>}hLnTF?%x8tsmAIsgGmGg|u>Rv8BwsL6R
zeTP{UVC+FGowQwYdSOxRp~FBB87TQEAs_A;lM}x^Qw|>}*lWw4iPL_F;(xj{mOoRm
zAzY>_o=HtXPf0zn@)_qW4ya_>s_cgZrmLSxl2b)rvx{Pu);!}Rr|_e?C&7lb*)2y{
ziEC?~iCyd&<7!oA!`k@DE!Hju)g@?c^6jjD#*g_EoF^zu?FYtPfU{}v9^kQvus#AK
z4LNSCpu|IjrTVJq^Axh!l<Uw<R$diaX-*K?Z22p;1VEBygrLO$lK?3fm9;%nsPVIJ
zTDz|j8Ps}6qLDKJ&lSP_6A@E;tIkyG2SWwiduPBV{UT&bzJLrPm&ScQ4Df!zukL4(
zT%zE{;>1SuMt3r^OU75Z-2V6or$6rA!E8}ngiu=2(a1AF_q#qK7o?3P(A~`TnE$|A
zk%^7R$Ayb#26B!(?hOX0Y4VYs$aSD%C|248u&Y}|opp95s^Lo!E>hloVW$E>MMKWm
z=^ULf96J+Bp_!2H00P3jUG!^{)vR!sw5jNpN)KG6USfPYp1F68^j+WDj3a6(3OMUM
zj1qRn)aG*AC7umI)#l@KQ5z|v6qT{yupJuE`MBzR!Fu)EVvY~<iah~V`J!EmXA;cj
zfl11x>(9iEg?{WH$|WRZcI`&Y1+5*V;bsoZ9!j8-RDR0`CYXZq61Q`(BJ8`GY$<>_
zLOIDhiLjBN>#lDv!UYAmkJ>$NN}<{=yYC|(RQl-=dyq?ikPwN72^8|;tUbzCstjPy
zmU|Ivmo0m?kO6Gu%wzepalj$Xilb8q-il`nSs^i$=kBqU&jwjR3GsogicdTUTA3H;
zZS}K(j>=<8cuD?bp(#AKHolN>Dklgj32UAWs*x41tTsOa>6lraf3!G>+mzJ62f*xl
z2fetUT0`{eHVsFy#(c&Ymo*ZiU{f{vWf0Q+3Y&d7KCbJu7M~m^?*|sG)n8E3)7lb3
z^mSgDy4{CY+HbTR32yJmh;sZUXzx<PxHI~cIfY;n?~`)x)*Y~v%r|JtK7mK<tziBN
z%WH|oxBh3HBFF8?je|~r*-#$!*<V$y3D13sqgG!k*js~rV`B+5w-5Q=L=JNI+I2gP
z|E6b6`nwWLvtY-w=ik3)C-NK8`I3n%=XMxV(nL-HeKUR^sKvjvQ;soSj9F-PIw31N
zE^*)sJL4mlDAnJjDzM3D9cH{mcGRXE0(35-V$;#W-I>ImP1{U7(;cEDD&kaT9nzv=
zwYda)_fDKTs@4kK7lkJ8PCP5pFqk2w`PJ%~+rg2FZ84sCs$Y$SV3OP4jxu-HuIGaF
zjzThT_^+RFK7^{=^!<O~PqlWdg2FSMDn{(~v*nsbYIDJCDLz(w?j^h9A481oZnd?&
zJ!Dm5neXMOC6V_1XH#DBIvwo&gXl{)_viL7kQ>K`M?NSzx+CPRr2I&29C6&^Ua@7h
zg<AXtp=i1P**8o*vuuN?K4L3topOjuW>uqA@l5wDmS3*uZ)|mKkOrtO*TWj`4|c`Y
z)~2w6NlM|}HMQjuZrN?Mk2^x4?UzHHf4OB(ht<~>61F&bkf7IwYzu8w9E@_uG}Z=n
zK@KAC1QfU_TR|D<5iWyF-?il|4JyH0i}(K0EA`fz5Rc%}`GK|hV9K_7)Y|hMYDc8(
zj_8$hguv_`&U<~CIImZ5dF+f9MJA9Cga_-seD;qLkk7gP&bn*;#4APZWr)%DaF-t5
z(Ce%6`yJLFZ3<^0RsoVNerdr59s0rEK^w_kI@pKD+E@&wutpR$fP~XVZG|V;uZmmh
zcr<g~R&M!!v*R%j&Kz0qPSpA}D7T&J^xDb%t|+v@F;37#ZLpntlv#x`B+-w;Y!>Wv
z&RvFE$cGZiobeg<kJzODRrkD2`CprN*>sFNsakyau7LQc<Yp2DhVgBf%^x42K%2|G
zmktR|fvSfUlpfDck_21GXYzCbj&pE!SL2zy(Z1mDp2g<An2U9I0B`0ZrCsw0VutK3
zR+{TMC017=kx0gkd{$gy#M?m!s%KDzGUOGzRpFC3RlA)>0u=;>laAO@d_$ke2Z&Z|
z%$@2Pd5S?g^?f%`Stgo$*@^Nsii5kK4^(|uEgnP<$}ovu91iGvdzc6aKj!g38M5~%
zkV|=h*9DnC?>TDA)+OgJM{gJkTfXked&{|duU4~ST^x3-a32V$mFvpkBV7k9#X6@Y
z4v(q@5TdrO5Q9Jc(bmLUUG&Osp;XUn9gsp}zO|a)42wTWA@Ne`)mGmsPgGHHT3tMI
z8>{rdGge=<L#skNmECTro_&c)%Hy{i*A>dBO2|#?!bhOHXBeM84?S(pZwi@_IPi_N
zRO2Z_jl|<i*6RJ#uB)>)UkHqFUDh5=iz^G0lWMKQn{Ify-uC*Mtb58j{i6<fx$N@>
zJ#fI?@x5CNit^ifeW?zzMVy8HSSo3SMX_8`Avv}UR&S&s7Ipa|DB4KXE7|j*Ys#TN
zRz3TvI7><CdMxol-gHA^1jik;&D>tc69GQLpF%gmiReKZ=d+CT1zLlXKCRKB0rkIs
z{u}busl+>&aMYjcg8Xx1+L@}G!UVyN{%Diyf&?kk0F$2bb=%Ww(+LF~DUF}aWK-<Q
zsD^DeR-07ybL%QPi4)C`YC+BCUcLOXEhGve=)gBr5F&TkVt!i5uXZiqgl#Uyt&%0&
zKO#%ECVV4*QH^idZbo}9G};#?$8N=vOiG)XfFt`b6I=3kB}(B=0@I`9<9B0YU5_Sx
z?%&iOg!kUM(6$w)b5(LU&qjcDpdL4d-Oq*o5GDA7Tsl0PDWHs1q*{3|A3`9AJz7^K
zSW*agI*Zx*LO0;j2+6T!>*M6m4_bfN9!THv^<`7>8elRj*2e~7Pb|(~>qAT_5tq}&
z?{QT;<35K+aB97JeQX(?l3Z+B)ivuw@)@+X>w^{H)<cV4vp&dI$&~gNwzoDvZ!gzk
zA6mY;fLu;;pIH6+pkipEl$O#OvP;Zg<lt<~_o`~F(VKkqEqnG@bHFQGvxH0KbZCiZ
z_({p7NoHSbO&}0_dWH@vqNmN*W^wWl*y-E-$CjeE=*TzRM7B3am$$rRl+Ns`tN)P7
zVf0e!EC3qA^!gZ}!of*@+MCZf{;Huw=``#2B_OASvTO|cjHz~XNYOkaxrhge_utRe
zY-8&S;XaJyk&DwkG>RYZc=nA6Oh!j(I*!M)lB0?RP!#HkSaOWiFdiq@2V+G0&zv9y
z_D=ZRi86wJJCzFoKW?WJeTV~>#_f!cBKbr`Sc+HPWI|K4%c?xFU{enI<{dU2Z$$=)
zH*|rc`r^kTs3|?n?D}6_9NS6-#C$Jb0Oq2||M_@UZs{`T&ut;Msl<CrEl~|diiffY
zXKgV*1c1w&o3lta*YZO-Zz#3Wbsrrk!Y;d!9~38J-C$F8GuMIGZqbNaKEEwn?6xl#
z1_zdBDL!>q1QkC$XLoX9n_KO!|BW^R=Hf9jz<AwvY}&Z9!ye=YZ0zo~hd$7e{q`vP
z`T0N0o(tko3MCzu&RuqvWzWS9WB0*a`HvYc85if5KNn=cYhc)l=bVhlmN0KCpGzGW
z#TUt$<uCE1Q&x}F&&A;$#_ZHP0>`+<-}em`pWC(h@qwc%*;bPu!$A_`{#?p7^57Az
zq_#jjllns305w{zKEK-e`?fY~h!!Qsq9P*rsxfCknG?BXHsu$a6@k~BK#-4x8MI^v
z-22nqvkF{n^$);`d8CQ9=gQe?@38jg3b~ZK@KNi?ez*|XmA&4oZKHKoJ;FE^pue<z
z&lQy2txSU5{vQAL3~r`wa;{w0!Q>edXo$=q8}yM$ScRL_NH$iABe_M|SO82)BkWVh
z0sxUasmsaJY2)4jR?jKSq0t;x7S{Ge_I(6v)ycpM7j!(=iD&+vc&^a&&AWEmspo<s
zqwj!eCK*;fPy~%g@R<MsUoYjQEQZd>3aoP3$?e;ws%LWl_8^4m=klg14pT0m&qTwt
zH!JhwtS{=~<WqmNxqussR1pflhx2rNo`1Yg{cuMM)lX_-AGWJG7*oLu(me|nV=zgW
zN(`dh0N3J~C+u+#rR(ubx=(Ni(fdogQ9b*#_zXzfa5J8joMqsc{NHZnGf3<Ov{I^b
zdE0TQWsfc8@kFQ>d!X!Wp6&ilwLv9WabY5~Swy)f#*h_#(C+0t(d!Y7bw9RT{Kl8|
zAomMOgex4y*aMo(myLI)JxaXTgI9pn`RNpQ4Jg0JsI;suF8-c_NQlL9vOJnardr8)
zwxX__8ocTyfz$>4kdh1AslQY9i7{JUmy}0oM7wQGHD<9>mt|X97pOO^0lB7YvJqYd
z&~ZE)uUKt7ljQ#i2LOGW)x|UYLL<qgch>6TnSwokfOz0eZw(0qy_dL?1eok%-F?<n
zec~lEk>#wpt{m<jt#eCGsN*BbD_6k6=XUY7tu6@UK}tK7z})ue&y|jBB#d`QJiCCN
z8>$RV`@;5C<BBu&a_^mWPMKkDIL7;YWR=w*bw-_&_t0gsDvD{w&(C_JBmKe!L52Y9
zuUdP~7|+)%W_2*$KtzgUBKPpJf;N&96+sw_PTf`tw+q{LEIxIKub?O(T!Pma&m|_f
zSHZr=y(?TY`Eeq<La6wRv+^Yj(#afa5dGUkt|2Po4@=ssQ@P9o!b8a6Y2OgC4m8-A
zx}YWqAN!TuPqcf|S8j1~ClxAfD%NqD1ZYz>{dlEiBY{c$#AdQRo`|@a*A+^L^PbH`
z<IKz~ZRUJ}vxS5RGS)3bBPj$gLY3rWakZ{oDj)ROV)a()bNS1CYuDnLOvQ4Qk-_(R
zUCAVdX!f@!|7SOHHHSxw9C9;Zr5MJb-AeTR^Jq86?{?L(Do|=CzqO@!7NxeZCx2~s
z^0RGd;Ro$*Ho@EOGK5H6>%Hn3#S7TkuQ+7)bK&<KwFmK@Bzoe^O8MM}*)3Ne%E(9F
z{O4(mYxaCukj`Va?0HA)^F#_P*z)HqwxEb)I?F4b4_S0GUOp?IckRIN0b3PqWfK>e
z3UOH-y~#uNYg<$G=68@H*w^M)#hF{Sl^?C<c|X!r<+9p*FNTzf;5uq`&j+bOTiFdO
zQy*<u(V~t*onsBp2j(Fxl6KASu||iMyh_&ee2_BG7U`Vi!1(~oKBRQXEe<=kzI39k
zF%x-reP?aeycp5i6GnDaUYx2E(cweYZ$O7py`GVZa}k<io$)O5Wqd`a`?4KV#dQGU
z@U3;nvq)`;^5RSDjp_gQ#qwY4f8NQBa}B};MIcKjH|Sf1+c6&GBhi(|y*RNOtA@nA
zwW!6uwqwsHAL;q_5o2Ni2?dI?Kn!3S7unzA)%THL27jOvu{HpLQtM>?;x3)H2?q$p
zI6Apg6+lXSUz}0xbUfqabM{J!cRS<AIof5D`L^&G#kOrz`K$!5tCZTOHl5o;no{Ca
z5?xd;hJ~1|)~lq6VLHiu&qZH8QTZEvVe`-DJ&SO!KnzHVH(Q*<Y!}j%T}|}i#xMA^
zm~fSuxGC$J&*YDtcK!Ku2bhPqcAviD{akLBX}gi#5e$suZnxdc2i<*#u+@<*IJc?7
z?%Uq`z+Y_156~-$e7h5~byE(uXCyhNA$J49Vj^c;`Ci}i4I-9I@hy(S2L^iWL9~^L
z8Ltfwne|bC9@DF@xLicBnAr<XMC2g$nJs%E5Q^B*p0a?;UvRP#?A#(gKDHvBITq{R
z8C&^6keWS^npOE@cMmCoUx?#N701Gfta+gjqFos8)<#cAgujeKX62ky6VH5Sm1_Q+
z)yA{Wet|C+LZhXud!djXane@re=knrwA~Ohl5;s}H*!XsDR5D&VNEXtxlvef;bYmN
zW{0VrIJg%7^Qjl#w$>L46UJwt3VvEXtx@DsGH>`hvX~T)`;Hfqk&t}t?1b;#-dqd>
zOY3ytFcMRS$MyxNM8ZPByAxD*8DNaw7YeOk_m1^@Cvupi!fK2qla8}(8J;qdYt5-e
zZ-wtgE@pAYbT6u7)s{~|5yx{2P^v-lSm1MZ{E26>iHp*o@MSqb0M*Id4NjesOUXNP
z!gqj>Rj7pnd!!WjJROZCsLHlA$L(?^!8|gGe2Z_{q&HfejQC1<`cLIjbtwJXbgaFu
z0Gsh29rAIu+3ZYtLl5KDKNriBd0JJ{N7}3JY`$8!Qf(M+_H&R)_tb*#X3^A5J!My`
z1yEgZ<U=mrMIRK4FY@}eTmqqI(SR)2_3D{>QGdfBz2Q4}w99U0mq4Nz=F7ly>xH--
z_e<%FCq`mGDa<&4FY}Ur-%i$W|Led(yBqr_VgK)T&->{6OQ??SSBOa17~?!9A`cS2
zl9N<QTMR4tjFRQJ#6w^&*`r+epC`Bn?8Ojh<Vj2m1-LJJ(TN68bFit)vjw+^vfV7;
z*|H*9Alu0iG+JOeKuS|cA{Er4g#tOgwAC*Lkw8R3P{tJd;;*i|s;#xPFNQ3%X+Zu&
zH7}M`1G+2x?`x~A#`=bS>*AMek6&ATjI5hb&|gFVXh>+}_`^{J=dJO@LX&p)pcVDk
zrOuH2V}y&%RX5ITtAJ9Af3o4Fm#y{1Lj1(ULM0sXzJ81W?GE{WA#a58By~hLg<$~m
z<T$(c#fl{;u`rTj-Ne)B9gQl%^uAmI6`UQiZXa9DGrhhVB};%*9R2<ZKW!U)Vz?|<
zGm;Cf3_@~c8FSbbBSEc!oM#aT-NZW{&vf@_xfG3f+(Bbn7wtrDB%JGvoveCLo@1dX
zw82{}PX7Eqh}JokOMSpjM`OQ8#3WF>B%V<Rs+jr|^E(;OBqCkln$U$a<#2ycg7~K$
zo(-?R{<_Unok|o_<1cJBQ3RwQn?}__A;fZYxoq>*EJ%K53%N@B`${1ryBeE1giy2?
zpUZ5m;HVi~f3Lk*DD_C6UC*Jwd=hk3yQ3=f7*6sxP6g<^?N;o*dxhz;%Y=ZptA4%$
zyrmdZ29ohYrK!N|FmX5G4*p_XYVs*4rGg>;_k4-C2&CA^{!nxMXb&nlignN)=7bL@
z%_x9A@?GkqqL|ee?Ea{?4+ef&eaO;eV!mL@>*LsND=x#tz1EjwyGP91%6i9ax7&Nv
zhwt60`d~A<v(m!q`k*qL6KNOGhS{3<KCOX|>pZTlTFYiGE_vE&95AjB989d%;edNR
zY<1bX=!i-<l-2utJe6LuhWwI{zaDFh35qUJ3$6NAsZZRWp}D@Y<CJJ+R;#7{S5!YZ
zKK+0DUu(@JE=mf4r!aC!m~HhgH<PZzr~30GguA0Y`~@+BMwwHXpS|9qb{#p`I%C42
z{y~HyVMi>9G%P<_cYRRft(3EZN6FPgu}wg2zC1FZ#T<;@p7gXFsg_i(=a4#R$XI^%
zXIvpi>{#{*egEFZV=vK2*ztUWu5Hjx<Yzb&5TJgtz7X+^c+^Z(xTBl?NkQvWe$#VA
zC0I_^#}&-AgUDcK^7S@UYe#I-$3}Lb;+mfWFpT3g?d!XHkIi@!Jmx8olyifrK=3M-
z`nhUFuSizJ$Ny8C_iyk@=(U9a)p<s|OZl=SE3vD2x;CpKAj`QJU8TB1K?~P%NZIG$
zL==c{-65sUBlppb1bZX==grtkgihG%TX_y;%{wCZ+}qKcYOTRk%8kyJ5@`m84q@wX
zM#JvpPLS;b`V-%|7p0$*d-X0j{U!{`-6V>l=6?OJ^&Fb#|C9f*2f1aWgw_rAFekVJ
zTz(XTx=pL7@?2ix-^<+EwxJNH?Q(2gwxN)rvP@nTbNL1*Ll^#W4#Fb4!1uNyo=L{c
z409{<8TZ@eqIsORRT~PK#25JAyKMD_I3z=OeGl0he@CGZNa)%PK^NtM##YPM$W7ig
z8wvt%c&X89{R@>J=&`ySuEe9eF0wv)Nj*C(2~(7xVZ(+X(Z!#u(fj=P<G)&yzwnAy
z)|?<>td$PMU$%7}-rwD2t<k&mG}2@$*viG3wheJ$WT%EvYp-CFp43xfbY$10<j0v}
zZ#0oz>KuoWhSr&j@Q;!%_z6wOzG&_7W2!fg1Da1_{rdIR>l-MHwf>mt1>pP22CEO@
z8901Duz4fVLMqEEHkPn+N1`#YWARpjH#0WAAq9PjRWk5i;@`)&&hZU}q;AGLB>NHH
zPOgiS-jB)xfPxdbB9LR;+L-P)cFHF(fST;IPXG%O%#L@I&SN&26T)wbItM=BcHGST
zn|>;9+e|L*2S?=TGwZJwC*Q-!>qxVw?-0J&^Zr6r4s9W(rmI*t1nh=05srR#H9*$&
zjE05ddEOR%9OUK^4w_xt@azAsEABTwc28b+c0D?gF8XWk@Sl*YZalFTguF8_iHvgs
z2ZoL!^xTRj&^&2<$%8}5*lH`}MhF9d$}9A0rAFbBckz_VYj(%yeLyQ>cQ?fKO!C}A
zM=(`k7mguZ#jbiL-`IU$9%W~@2mZd)gv=27hw)AElmwe*VS69N_MB&Kqz!thoHH1%
z0nX1d{{@(HaQe$%Dr5$h_y4gKFU5K1s{4}nj}thp^j5INWI}(*NsrTZSz5lb)%i>j
z*L;bOxUGq2u0hp?E!f(Z5_91>fz|7B=sfD?o41<y2K7mS;73VKF9n4{{sx8X928Wr
zC}{&=53`aUX1Wk4QjNV7TAuwdJ&oA|cQG`s>7^hH2YM+~-)qe;6=K%TtCp97O+?uL
zP-f581c=l9rGVu4wPhnR1(ZhZFa28kpJ1x2!>4fcFg)wt3Q#bj^QFXy9#6#8ZTk}T
zgI#!@Sa-mKvo%Lw!s12i%`RFHjQXPi{s%~%xD~S#gFZpkkRFNGTztPM&ix{n@K`+4
z><XDjSU;Ai0CACVO}R>3q8g9ZVPb>qerm_7UUdr857~xZDn#fIr6Kc+qJr==n-jT)
z2Nbf*eSXTZl1ijq1YMmD90g1fpX>l$Kpy%fsz2NBQR%L!><$#CI7GKi`)jy++3;o@
zmXDve*_>Co4@<!yX7?S^-dr_^TbkvOKcD+Sra*W|IRh=c6cRy|miKBx@&~0LS<KBi
zg16o^-#na;RLIY+`#y{g5&2Se18acu7Kuar!>xTiq=LQWz?85JmZ!Mi-FB=Y$?G9o
z%2A-|b|-&yK(z11G<2&ddTN1FbuSTq7+w^e(EaQk<!P!6A0$A7T}0FkA*nr#o^%<0
z%_4~%e3Z|M7l5w3DYKXJ7Qe{hSoU%puEPVE=5R=RIe7@LMMUmqprR_Yt%x_0D5dcr
z18tz#%IpIREvW!(RgCZ%J~h~XUUvH9R{#~XLDw}e$GQE8Q=)UbHX4Z>>WUTcH7^Gt
z874O~?kV1@%6me2>*51>P?AXpsy1!)-ePf*ne<!36XSy9(8w#+_;QFTL#o(=yIfJf
zHYNPU4m51d+0OxminY94C@wl)$cz6?q4P0o^H!t#M@hG8?Jt+()3(Pt{5cxQb5PK|
z0fkbliPOO$#5!|4hz5#c+gCxlfZVTJ?i1^NIS3*?b=I59#k@wPrs;pVkia35aSv9j
zMBY~{JkggNcgey)QXGUJhwaZ@cFeyUz~Nxr2OuYL&zKz#n9;(a+d7f(VKM`^>}2#Q
zMUBV>6_#xx;rjzVk*8jcLnd{0QN^FGrlfmcmjz_u@iTPo@_72hCIhJ)Irz6^r;<P1
zR1R=pfTUvy5k7u$OqhwT#Ei?qU-HUqe$Ym0mbnC3zQ`c3R3`H=$!U(k6<erU$j@0<
z+vj%m@iSRe6>)BhzU7OPo4f3qFXXyE+Vvc42lCsE_)M}7?$$^m7wl#ZhEDQxyOo=?
z=df;t+x~KKa+A!8OR)w@^bIScXoYesPTQT|td84;6mBqWcVkE_8^zqH6zlgKx@Z4n
z_p|>}OyxoDj_#;(+8$P4D<xj>LbOM}8B51yLF9;qwT3`N@<!UShC)3SCkdHpu;u;>
zIxvqxdPV$F_VLTMvcb>2Tn@gXEBQ*V%C9#MbW=^C`g(yWt!XGYt%`1I8_JQR^7db?
zCb~g5DbY1r+}eal$(ri2PYwnwMl_e3&3E8Gq{JE;oTj-T4EF;6QV`A>qxE^OH1Q8~
zgVt0HNVH5jjbB@H^dLn3%)iyYAazT&miW!y%m1L&!6#F|khN7KklU#ogmc#J?+0aa
z>By<H5Hh&8A@~V-aC_W$XZD}dXQlSVGj3PiuS)N1-T4f<d=WZB>L2U<&Ai0`oCj(}
zrS&%iX+^RC9|s%a5*R*=KWndzL}x4qOo>}y$8rEp6wD^(oKNjoJkxMW8m{b=<4-(O
z-AxHl$Fu*DUQiGh6W{7SM7Wa;&SA=^C@M`cXFmGcv!f9PJP~s!ax8O2e^*Wz<sftA
zK7fRtPUzTh1r6e!Z)bAU5?V<RfBxRz(+js~Qw@b=KhtT``L+926|2llwdE>R_@rHS
zHfJd7JYz{`KbL(E<5I}B&Sx9Quvqgc$nZin3hr#8IG6{n<`Llf|Hc-5N`vI(=&@_r
zyku+f;vd(exvIfq1IiYv8bOD054~9p467)7<0Rg4e1n8-wA(%evIS5sRULSU&<uNe
zCuc(V0wu2PZq*@gaTFo{q}@yO`TY?YG459<kju#)_-NFUM}rdZP6HwJV}FB>QiaH%
zDRr;C60`!c^C`<E_m*X^l<jHXWy@a)ZegT{QnDPX6|WRhv2kaMt$fALHUtaB3iTvG
zPN~PHY*oHtY<b!!Ued#8R=-ln3)0aKw&oS*hR(4?U)WmzwK&->r^*_K;Dz;8`$~{N
ze0QxbfK{Y+@#Wyd`fMOwV#6zOoVN9#*+TQ1@DCH*M8aQd${wUt16+9*HOI5_h*BsP
z@l>Lqy#vrqM)mBHjD;!+(3U+&YUM7)h0)p*hJ&yS9nnA@on&w<bZ~FNLDr2!vVC?&
zqe2v7EB3`R0$$2i%p$ROSM8Ld-i>*^Ioc*Rq(8yiisclaz(M~&HSAy@x=O0nksMti
z<zh9*s*#k{&22)*V$)Lj^hx0!<byFDt>sjtbkf*h?D#8D+Yzm^vfihc`-$8}LTd1+
zjV*?(U=W;mVlMLexJW9$Q#qH7WZF3G*nyUouzr6bq=SCElE#IBkB;;YK*y${uM)xl
zjZpQ$>Fi5v?MyY8d>%^>jLil(aK|t}a|vp<if7K}Bq1u|B;@zH5Xio}!>-2bVrt-+
z{3PqK<H#TXWY?l&g%(}5>#qcfCxWw&oI(CGV<c|o;V9;N8OYAF7P7$IO8D7HyZy>z
z`JZSd@$L-kwf~f^d@1++Fji*p;7+Wa>agC;d2-)Ud^&;S&O%2LC-4*4{09MwBnnp^
z;c&+x_}U)&H<GG`2knusL+xKX&0h6GOG*rYSmwW!mJDLFJl?6K;f$?#)d`E6dc=xQ
zE-s!G8=_CgR=rwC!T@=NZFRgCK&rZ=q+@GdEyQU1CR_Vzkb*t9nbbs6b?geri&mS!
zK~YGfT>bhUdv*P*K^eNbOQ1|c1<Dl;n|P#IYs@dmX7-IW`EV>K7}y-Iq{aW*T3!uq
z0qj@GHVAsFH&A+gsBha7(3Rp7pl5ru;H+TH=<pV-up%W&WP<H|H4%rBP5suHqYzBd
zKX6p1Vqbs(1xInvoeT0&50Z+%!VP7RsxtlA)nW25!Da+Jiii3;2cA^4Hj+(#4`+`b
zz3&4$%&+WNG%u_g*r~L<<FCdU3{TTxJN|093x+A_c)|xKXL*rYysLM0Cr$X@f7aip
z{BJ}l>OuO)|FUJbok_@%V0&ybCpFZw-==cVQsbJsCN^Dx`Mu5f-vkPMXtVxz$;&b4
zT^t-@cWgd4UOFh|x$x?*{~IbO2s4Q`ouSbc1#@>bM;;#Ownay~_uK4ReBenoA(r?P
z`+>dgbJ$2M#tk2SP=4Mw1H}KJ&s#aSm+CjXZntyE!2yi<ORvW5x&xz87khXoVMXGF
zPq4fByN8SrVBtLns-KJB`z^&QzPATCO3xqy`a}QK$6bR8<58j__tlpGuClfEnqTnP
zeqZy|a0Xrb{dZPF3drTJ#c3lAxvhAu9MaJe8~0ksA;Z1M0ITxXM+o({`ZYiFo-S1X
zlG7)isoP6*PE~eW8_$pkg43V1n%4?C^>F9bdZ+BpC6qFC(O9=W8VZc_C>-Kh=IT<;
z+=l8aglTJhEf!b0Tt$vG`J@k1=`lWbpAFy3mRN`|XmEGW2`!+Xt*-^*%Ym_kKDOoH
zL<jzj%SXDM_H2c=f>mPkI(*GIQ1iCezgSdVUZOrrMx@TyoWgUhX8159QHeuUJ?#hE
z=ij%=2-zJ|L!-ykdVOn%j=?0~|60%@ii-6r=<cWu`bv_I+(r^C1wR$SH&y}fb$E~;
z>{#wKk;OKipOi2~ISw56mD_`LHa=z-;UJ|lnzoZwhj=YB1DQD{yhAQB&cmr_AVZR~
zgHv|;wN!q=JV5Kt#PaI~mGG$4KfW=OJ~&Q7Q~3?mf|T}hI(Jz9W}?9}3Awa)T<REa
zE7(Agf;ty$&R1s-v|zs4f-1;_2^6g+u)+Yu=Y@&mO-V)?&r0F2z{bUZzOc_aKS&u)
zQ?6A#V0@y784kqt1j;!8nNRIT)vH&h<fz!qT$9u`7pA>%3V{yi{~<T;+u0x17Nz2p
zwPh*$Lu!yR3Rfw+ljo!O_1}#avITG@fW-KpI?DCwnBRXbE+cv44B3PDSX21S9%fg3
zMzl=Tlu+Gnv&QOn<l^E+AVX=RlNr$fG47X&O+$Rl^2T5&(9g};ifE@;7U>orOIIu7
zjZVvKp(5z?n8<gwsxcT1e(A2o$H-PUhPF{QMcudXCpe%*OF@B`AX|LV*82Fxd-Jf>
zG#0|RW0%!>TcUHw2!@G#7*B}0MnAFAMTnA#vy?$)fdwX5!RnvrU)nnHkPZLzOvI}(
zo;g~%>u*zIp?<wE56z8**pgp*m$fwh`hV!Q5?iKlEcqZft(NWk%g_+C`WK^=g0Z&f
zTn12-(fY#L^BHX9w_u?(109c_mAD33()W5lN2DU^^uJLolRmjI`R|f?U_U`N)w1<c
z7NxiPHA)C`|Krwfo3X*hJnbY_LjD->-#`D~n80xPv#|;m1ZiMa;MlQ1j8EBC;%0J$
z(*Ag~SKcheh^XSniN;@*SMo{o<^TF`I~gM=YtFPy_%CduDqwLcMsO~n82IdTLiCTX
z(I1=%WGLz8fI;Cy{<ojDNk^q^Z8jCC>-#1IhG8v|=X7H^V<l|}tjbKa3_mJo3`cj$
zW^*wIN4a}MjpqUkNdu@>^RN}Rj0A63hP?sajw%&*HIG7Mf_KIHqNCdf5W#D?Ie#9-
zTdLcxS0fTc(<2t*Mz!2CC0@8B8H#7;L8eN8sUZ7~b9p<@a=A;DAwt5VZq%j5U@bP`
zj&sL<>@PJP?k4D)qzt>4<No6-NS{m6_wyOn2Z^q7FMklvSX9`Ta#ZYLz=Ad&+wG%l
z!7nZrprBsu>p|d@v;br8>*dt#SFWMuuP0FwW~_KUu-}9HA!jS|$>6|8gv=9{y6W|E
zeBZ#EVYPpZo%nx3Dc8L2)JaiNI<9D5Y^@K4PdJ(m|8T>r^;Vm`4#~k}h;iy(kA>PO
z{c8403KLx7vtM+J=<Q0Ox5fZlim}hFDOwcQdJdH`+&4!Hf(`JRZFxOdq|Gwf5$fOi
zdYn_0mV<?9d%cjuzFupO&K?&|`_?)Vy4`r>?TrSdNFDP;Ibxj&0gT%m5Rh-n_BjN^
zOv=%_`*pt;+z~(--bgK4gyG)!*7c53%}^62o{4^Q?9o*crRP9w7oE2eUs%*_yKF3%
zi=vvipXF5H9F(Nd#{D%}L;Csfcs4#HzWjv$VuXvRf=&jEg`NYQYEC&>oyc~WNWZsJ
zIoonOC{y8Ab~*vt@^5w~@eNi?I*okQb<n1AB@g4zHT`;CpU6y@WobK;`?7y@MCs>e
zy}e8_LpGP~OL^8}1DcQb#lX7|*rrSUYQPAWR!$VY+qPKE;uH4lTD8VX8{+!F>yD#G
zaKyinla!3l@e>-|%x5SV<epAJD!Y{f6-}*c$!@>y*XNVYV}&5FoZm3?QZ7)3l5pA`
zhijNfRP<VR^KkTx={|<7ycgZ+?k!c$Fkn4-Dc*Oe`{05f<od}_A?0W_N#espd-N!l
zL8vxo_J$wZ*z|v#XHLuBD5rxPXqNx|Xv^Oyr(@&4*otUfqFv~Z9F(^5ji4Io3umj|
zh>a~w3FnJ+$!|EJ(J=G2CI<~`pOUk#^%kTalg+#)f9>qSPpI~dAP@UV#!>f19A+$s
zMZ-{E_4XxJAYDpR*6@a(H=UqMJd^s*d6&lgyzdBsp5#Yq%Fh3lJE!@LLL7_js3l;=
z`&an_ViYag#lrN?T3a=%DP+C&H<D}OWcyZwBudxsoKHuzma*f4?T!A#+aYHc6;Nlw
zhF6L#$F?u0<_u!tak}3KcWG`(sttLtYp-w7voBbG%-rF85>EybTH+<BK^I+Xg`^&k
z3Tn(-Y~(DWq&ij&{tY~;8$8CVXA-KETo)B-j5xjc4J^3CfiIY`d^t~KYeh}47E(f=
z%;nyJpTb0Zny#0m#-$lyA9m`Ea?cEL?oVeg^1uSpukf!v<GmmqVbhw71~O}lWmf&_
zDQ}<<Y#d&_vp)1iGqDJqR0!>C0;9x0bH4XDk)uqh##BL4Pj9vG#$!Q2{6^K&ITm3t
zJ}6=##8ZhBujO|Bfir*ug<W?zNFHW4^3)Wk5fnIfGlzu8%B;eo+pX+DA%GOhw%geV
zo=GG~>9m#{5KfNrbKHp*;@TA<Bys0%LQY^DB<5budgrbVyPwk_Wci{!@HN0X(P9tt
zZ+u7|#YU+_*yZ22-`Sg?C4qlFV$0qPVsL~kYrSNNvE^@;v%9r#h-F=ouT`tFPd+Xy
z-wdH<Q95@Wjh}4Qo3YS`2R`gW*Ah)!11fwyYioQ|7WxYa^V&D_I>$$goveAY5UibK
zOn$SVQP*If)nzwj>7@Q3&qzEgR_M5p8?v2PF0e@{yc!)MX+aU=t*L4vR}x|TtTpG<
z;C6ZjmBOecS}Bp^94A8Blh%Nn0!Scfo5Q(-tRvo5dXpm*ZYyf$j_P}*HzQF1R{p)d
znq3FQ1$0IO-7UncXkNB2o)rv)PsgHd-To0=FKg3V^(yO~j?9$xzgegrmG!Wzcw@4!
zLxxA*bTV2z0zf7&a;N-4W4Y}Z>0h>ExgCY=9e)C@?;#Q}F?QTH0wunk$n6~*ke%sd
zv@WT5nC2Dq&J!_Ba0C%08hFf#R(sk`=Z-@s<)nNj+Q_Zv`!cIpn~Z1YNz^_=$CARQ
z@|hy13V-~QO*`=G*Z=9SAJ^GTw1{+K(B&&Oo67+1DTGrMnuLclHkT8}j7_adpQQ4q
zFT@Lp=`(gUr;JsYQ?i)QZ!Yzg?An{bFo+`rQl)nG6#NKLa4y%YJ=wn9Zuk$3Q5@x)
zzH<XdIZC&35Zw9kaJU^1XyV`5QvBk&L6EppEtg!voLJv=1b(afkKKz_7=rOE^s+(s
z<5@BzO2z%aA%e!iQ$DPQmBYv*hp83uK{{z`3V8!OIcv+BeA<<gEn|k6yiyr2Z;E5R
zA8DJfS2Q^`x_l`*PX_D~)VH#!9P=0c*leqs3NEA1z>~cx7?ELov9w;OMw(Jg`J}jL
zC^^E~Y=-^^A}dd@nrK$24acYi@-_wIA>S|Af$DrVI4Jg0yrjPRMCuY3Y8!Giav^>L
zxxU0|HRh~_|7}z{ho&4oa%h3^%|3^p{~IcUmYB3d;tE60huYdysNCjl))vbuV)eDP
z$Dle;7a^$%5$k9Qx=d6N>Cq7b_eQ_+5tcjWWAj{R^ro?<Y+qAFj7!oH2aTe{L)IGu
zi7<ax4#ZjOkB`wcFxSCcp*@ruwUKHy9L=YotYguhayOilW4Sgh5Gxj9-1l`49R2Z}
zK{xvMlFRNybf@s|??6wL-8&i2NO&ofEqUc9d<L=)=@LHWGeCPyI%QwN9oS|94juc?
zY_h3bTBHSs2u}G2-9r>dpY|6Uhlb!gW)c9o)H2u=Zj{-aqiUkYjWQQY?G6Mgrc!JV
z=vL-^enSI?Y$2EGL!~RZ+VrcJ^MJ)YQYwxJ;TC;#(2v!>mgv~u2V0Cw>-9vWqR7M|
zj@wwaVe%y@!Oa|nr1TWJjs6G<AGX`M_U%0bXftgoW>Od~9A>yt?2bPJ<MD0$5T8jd
z!@Jq#ASuf2UbQAy<ZL5v7rP%gbEJE~9#rV8<bqF#n)c9vStLZM%N|8DKk71*epOE4
ztl3*Buy~}hyfT-?GqDfK8Sxd;?DDsq)Nl%rI9B*CZZ&RjTbX^JO;8=FPl+d570>*f
z$vNV(t*+WlF+MkjBKzKo^Pm_K0%EcWI3R|Y%so~ULo44ZiHhDT1PNXUit$#E5Q^+6
zXv^x~auVp$zmh|G^_x_9HslBJ1rX2?YmD(^7@cLjA5ic#B}gS{gEjk4jDQ4Q5#N_n
zKiJ<hX07qPlxp%#kWX`4J_{LEMQ7XJ3ff!h$zbHkHgcRPxDGRt<9<jTj04sgO}O%k
z^AQxq_W85I*C=zT{L#9jom=CfS?pHtTS1@j-1sxcn`pcQ04;(_st(4Yvcp_pBX9Xt
zApYe7JBU{fK*?!)ET;=&{h=L85RtA?w2k}Al2R6PkM9gAgc|Kc1(b3PyZY*+w<bna
z`o)O=fV3p2BTwbVK*Vv*B{U{y(0dfSv<}o6AG???e|o_teKdKB*;I~(KZisn`65qe
z2THIhW4p~H@c2CA^E3NaC02JUr#VB+ML%MS@Bu6SuJhS9+b~yHoNsL5t#Utd&h2Wh
z>wZ#b^p4nKG?%R$gH(yTx|VCGkVd;66PNn<yxpkU>&7UFf9z&HE1Z$|qglJ<L&A#K
zZU6I6<m6q-N$%&)geB;<J8u<I%Ob)H-t{$-0o(4qRj#*9+wFdAw#<bJZYp_PIBB!?
zAa`M?2L~W|s3+tH(d^lygbO{x1(xR%IB0u21P)TTNV5A9fn(X*PAhPa#ffhD+vS9>
zQ^|4k?JN8xjz**0NGtPmWrv1O`4*C}>h0K=&Alj=DL!(<R{KX#2vT*~n)paY>C3=T
z%2&Q!=1pxI^fWcmKn`~DO{bcQ)xQ1f3;!uSzpMaOmp#Ems;sUY#Ore?DtlT(%vH4J
zk~O~Vi`-^S{#=KTU1@$hjz1CAN|)V|UA&0mxApDd!;1Gx@evA2-<F?YpH&opyDz26
z&8WIvN40ofX@<x)+xvDQk94qLo$=m%?0^rtyf5HIZ^VV!9WA(w!p=Euz41&;qg<-x
zsofvX6ty)|m@)D&2iPKgnTJ_i10Et(>#q=`T;v-|(0A?cf|We>b`Yo%xT$c{@%#u;
z4)#MXly>~>a?Zq=pUBtfNb({)8P7`EBxWp=)kGrCcCsUo{G03PR6fEfiQ*RO?eyF3
zhW1YpY<A}D;1g2$rvJCe7^g&Af&^dVbUBsqppnuYQyj?>9@uet^vb1m#^+51=3+{m
ztv-{e>o(Up-;$C7mf|o~i|<f%N&kT#YCQXl`(LRIzrZW*YCPkD#X7Dqq|a?J@ddI!
z+;hOL`TBjluEwtCBA^FFo)-Iaqv}E0Og~09vj;VUS-X|nCJh6*&;q+88ZY_Zyx4Dd
z0vDc0ei_W#T|W^!w%fh8Qz{d#xj9Q(ab17PDRw`HLeI_x@W3JIBch`eCx2M2%a<&;
zKL5@h`7u(Z8_dV=1m(j~(w4mwa*3Q(p)2I-u;uTRqw%jjUA?yAokF5Y7U!e3GFnrZ
zM4IRlnz$;SDb0eMHKdze?O)N~95VZws>N)XkOD>4<}=i$3hCirtBGfXP@~EdjU(Ut
zI|Z|ckl*Xx@pbWVuX1BBJy~bOA$?}`KCj{)Vh!&U0#o~pHF~e`j*8x=<Q9s#rWmP6
zpgOu-FwO6zeiz>cZbEDEp;cb3^_^h84wclrZN5N@lY3gM{T(MtQYBJ<n&lSa$^O;x
zPLO9Yt)8TI?>j->dVBf@Im?~ZIw2L>zIQ^yS_MPVq&xa|j?`^U#hPQi36$>SL)Py=
z^-+?{24ks;gAm~S1mW?&olK47>MP=c(?wk>8_W5l{>O)wY|jqt&}dlq$E!sncd_OO
zi+#N6=n`zTdpw_TK(MUf10V0TPeR785U(eye!edyWo7L=<=?<@;koAf_z|u)J$5D+
zkz*>YE7HTHFB%Mc&;AjciZ5{zwCQR&@iDcTgbhN8ItV5OjY>RIY%_QRJX1Co&y=I^
z3dkUB@qB&@ii@+Kg}q+L^+vbL)UNu#KmR*NaxvGg7a2&oMXvcAa4|rrxb8o;AC~jl
zjdy~elUCU?!rn}zI7~eskn>h{hL(ncY)9>O4y-gky|$Ed;IJ-m;O5{y@5D2Rli-D+
z{><*?v)FzjjO|{IHQa;J<dEGD>_Vx*;njMC--SJh=_&5_VXiowJv!NHZKkz<_@$pS
zYYy&4N;giyMUMTl=1{tz2RmoWn*;TcUsU#LMRTD-m=SP(S=n4Dj$A=)RdYG<8(w<F
zR#&}Vc0VwP#2Wuj$%c>E+U9b~@#Cmzj%8I`IFah4R?GqSW3YjduWK%67%gXU)2eUw
za|`VdKc+ho{8iEJhUTO*iokkOD^|{+HG0QK2C&y*&9)|==i=l>2|CT$ue?*}QPxcV
zt|i;;SL|JD0vuWPNDA*WQ<Av>YmaVRM<wA<G2Ml%_*&ujLzV!a6SeL7hT!))ed#C$
zZ~J^Ozz2D)I~Gz-0!n#s4pF8zhda{sq4h_D^YGeYFMs3=3`Q4<#rf7oaxcIzg}uqo
zW-MofGIoO<^A5*~OZ<33ro{P(JULsE8^~t~ZEUV`!C0KVpeYh3u;CQV(a9KH(F@Rw
zYIH>|74h9pMGNwy(c*NpC{jRC3eDJ=Sj42A=aDpu3DG8f58adna!fS`t#%g#X_=;T
z2B>=BL}wh;HgC1rd_lSsnEkn&2(}lg%YWh?mG)I7FqGwOKEQ;KKxG#ypudN1TumH>
z-}(?Y-^FSY5FC2Z!e48SdrA3WMM17t9e$1saU(w|nzZ6^X*XluQo72Hf1zk6l(+Mi
z?~#fSpm|9QAhE42Ia(D9+#R1QN#O6sSdzgAj`8PP`|ZAeN(o?ujt9{QtpL2nL+=QO
zuYa>gfslx?w$Z5p`Ok<AShleo@{fA<+wzS;2(;)cHvU?ihn1g|%oxxSQiT+Mg_Yh6
zi#=IuR&7jS>MKcR#oDks29QaS;1?g@1;mE6*?3XhKW>5uIW-##;n>|~wHt$QkOhi_
zax6=oe>yz+q18v9(<KP7!Cw&8ZH*g)njIb;R=WA7YzT)oZ_U{!_P3H$(^>)sF`?<D
z)0(ZIPbg{F{x+Xu7wN2#VPYwyYS6VLQPP*9BRUhAUtmIVHs8B3nT;__@P~W2M>?a~
z1wNId+AM0qeH(wRDbCp?S(zk`AJu{MwQQbRJ<fp8ox>DK!c{)K8v_})%8JGy{n2Qe
z4U;Ysv$G8b<R$K<oC4wU_voZXa@i!@!D5W%Z+K|s={Q#Hp)R%Yjlp#t>>s5L)bWkw
zlKA;=9HSGyQgC2sQ&0N7>>xAw#BX4hYV@p~dZN=G?R2zo({IUlXlLS?`x+J%mrYhZ
zDJi+jEmIr)W?&bw;%NCrvlh_hi9j#dbWB{oq~!Xsc;!ginyE&>4UkoqEqOLtYXfxy
z$+GQ~cdi;6a|8=0vOgcQ&`G>x3qEB@#aE*-`d!f5udqc&?`~3%<U(!g>9^~-ds0G6
z8GoZ1NW8nJ^}FfIj8HRRw{j;PEl^r0eoIA7Zs*>RpNFgmOFrPvEq2Et>!E_hUGItf
zydT=V+-dG-yie?Yf&y1p3Y$GhP$WmQ8ec;mAI3AC8ftnW>YlSlKGIPr%BCa&=rte$
z^66d{&!m}?N0ee#mv3^KQpiL>QKgoUo9~Kft^Oq%DoAkUrsP`wh9P{_rb4Hls(Z>-
zZ~D#V7YAA=P-1srOg0s4C5wTsJX^b|U@LfyRVSfB69itMhnct9guc{ml7@ynjNXOH
z;Ps=nsE;O6EQ<`Pv|2+pQJq-g?>40hs4DK+R80y$A*Neo8wUoy74z_>Pyp>8CF(zB
zOEw@l+_v~7U%F4NJvst5_&nj~H`d`D!F`>zy_<ptmbz?k9|?<Mbg&=5z3<Dh{sq6Q
zZvPhD6xQH;(LVt5kw4xAyg_e8T0-h!`rnI_wg0D$J+T-{lBoQu$8tm*^GAnECdu*G
zY^R#~_mkM{;|cDNRDcKVL`+ud+p~5ux-UgKST$)zCn{9<uNciwZOV&+@6m}9PM1%_
z)RZ8lGubC`6=!X7Q@Lb5e$J+<J)!alz$;7pw7)$(U^6-Y-KgYdtES?}{p`;L>V=&F
zeG@0-2@jw7_{J%@WBo!bg!A;S`9T?gpnKH^9z0}=K5!QW1fUeriO80Ggh@<l*K>vk
z$rD>r_}=hOHd8K$%xHc(Ho*Md^5>)g7~OBTbEyvyK!E_Xl)qK1oISqd-wl8WcYTXT
z<+yOqAJ^1$+Wi2gvJygFAH?D+;`XdPj9;X@;#fS2*~SX}@?Q%n)ET$UDVeAs_xWG`
zh6K87bHUt!!K1E=UcT8iBGeY5%C^5;oRr4=06LH80IzAnaODs?XRG3w2t8qgVqtCd
zW+xoVD)X(a@h<Vt<`}Gv=F${OqEMRinyOnB!E<_4>s>5PwxQ#)x@e&}iDJ#l1AcQM
zGNi~ZHMSZacMJ(s!|$!pM_!zKm*s8pA0^+CHT%a3IwU!twIslWBWsOcpAqu#jkRt5
zwNXmoCu&3fubsVr`_|e6FnOH*1QyF%wPSPe4pfyY%Ei6jNP@g9IfD{NJ0H&o0`9JJ
z`{J4Ar}Qim(7SUko0WXidZP)ruTum4x@`S1BT0Ba*`Tk)#*XdOlo-i1IzYnNF>gou
zijkfkJ5~X}ZNsV~fZ2Ej!1u+tiqh4N$1|b8Sw@t0;Y0-mG+z6R=J;ej<Ip3yaCx4$
z2_K$R+RXS=E<B6$jh(KVgK0D;f2^H}XHrn0v;YIYv&qdtjo$4-N1#N9Q#mJ?KbwwE
z#MLRQgWL^fvg^f5WsNR@rnC7BQA0s0Uvuc^e4o6Yg<N;$H~(6ga{1xNYvU}RYuXF8
zkYnzFa$SwS-0#Sl^2K;YbAhf*tai;OjGT)>*{(<D<plUG&Z=Xtup1R@L_sm2^i4PY
zTgCXJp}V!Y+|Dh;<lW9wD@w(YTgu<k&w?O#a!~UB9<sasVH@`Edx=P>y||?A$DqZl
z3B*-I(t`w~%-`|vVIuRNai5h#*`sQva*9$towH^wDf65sGte^swK&-->;CeVxP}P^
zu@x<4BAc<3t!#06g=$j%kzC)9o0h^$P#Y4DR{y(en55!syhjd`Cbc%cWTgr}sfd}H
zc;@;IX9jA0uyrriT3t&VNrJULloVL+?GfG-XDjCwV*VY{fwFRqEy0r?>KjtHfAmlM
zG?KeGg)7$V{Y#W;i3N}z7$>xBjMnjMZ752jHgCLZH@;}m_#B3|1?z~$@+Sr93TfNh
z5<&5mg47eqcX}sB$zZTAK6AZEvD{gA^-R`E?caIpZ3)KdAPHHyrI=@bj884$U;bu;
z@#|BJ2zagLLaA7-7JJNJlm1TZ@39!6Bx_>R<Q_SmpHR@Y6j49klE(-;C1>Hwq94*7
zb)o`L(Ocyfx05lxECm;B!uLQW3du=Sb;O~eIc23em8x%Wx`JvHR5=sv@@UQZ5v1K@
z1+UVqpk)9(r{b9h7&xj-JF*p5>KX4FR$+*GO0!jKbR5iB{L)-JE3g2OTyf`f6_A6m
z%;9h1dquplUCnox+KT~d(Pw$2sDZD=0$|_b#=RcDM35<jm4|1TqH8xD$fw?JdeLrq
z?@z6J!EWa&AvR25ileF&Yb8-?Dd3a?hBE2NSL9B%E{Z@W>LOa7+ufXL@ltU0(Y@G?
zE{Z1FeIK8KKYQ#!G?tHnJnFOr4;`j0|2Arm{7)Pr&9?YC7hCb~1-#3)Y>U$iad1D`
z^6J+-x95tg{ohLle*U$UTM7*!;WXFzs%TK4W>)x2+UhNFOORZ|&E9LgQ&L|lYvEe&
zfV>+Ak6O)^$M`9IIrl^G*|qWJJhwTROU8{KK^kFhoezp}CCYTcFx6L(3eR=)5nHwd
zp}{{_?oigaC2So<SdbMS8rqcLx~~-qg61tDW1#vz!0(pmTW91leN(#6S_3+9x6-Z1
zNGBE>Klz;O4yn=Gw-hSM%s#h{s-G{p<^9C%_0}KBgO~&&IScL`)&Z62_qNaBD>bV<
z;;}pXf$5vpE^_dDeNl^L)cSJ*%5I0Id$3wcrG$9G|J+7=c%JFbA9G+;T-lDrn`wM+
zzO(UMq^+6E@A#H-dkkPZIT7$k?<~z98c#bJ&-_?AFEFvCknynI+No%wW%~)V6@z-(
zM+NRt>QbF?corvjZ?VZOK|4{O783?}o^zTbA3;j-e6Kp36Ly1)KiNz?lgEe1@C@wE
zY&^qMTb$43TruYZ7vCMy4MhvN46(29Gq{yhy`!fVD*_f=^%=po69I{PN&YNpTkM)I
z8upV>%F{|acKtClV3JnlQ@atH<2R_DNe3xBznQIBQMuPCVZ*K1mQOKPp0nH0LbxQK
zpzqnLr7cAO{Oe!sPKBc~#eK%xasfhk?#6endvYDbz35r_{C}|f-pl6g_P{rqq~rF`
z|Ln&%@KLlr|BrL*wrwpm1e1VlWy`k4`RN-T9_9Qj-<niMb=Znw&{kB>6p`U^RV(A!
z1x~zJ&ccOc&-=UfAGOu_3*G-qwq|RgL<2Hr*joRwI7x*btJzvekUKMRdGuOs_Fam1
z1s&O1R*H0#_1W*AM-LN)7j4Um1Fy2VHg0uvhtxEAV=xx|qcwXcn+L2VA!y^8Y28|I
zeYe;d#oWa=D90fL?G7HuCUYZ-SL@hX$OJKN9k$np>V-CSdgt2)Y@c`DRpjCB_;H9_
z0oI$V&5cjKK<nR{N?CjsmD_<jG}VNzKzLlkKa$TNl`<ZmWs+k)Tzjh>i?>qapB3AH
z@@PEffJYL_^y5_vS#v$N-wB6YnR-vgYegt|zUx_=aIj01pq;83C<(C)py%y$PG$H5
zxO8|4|A0~lxExRB9+47Zz^1CtFUctY67q>nZ_T^JvusLV>N`DC0ZU~QTEk!2tOMJP
zGv8di_8Zu_kokCq(--i$V6l+!9qA%<$47QGLHc<DhI!GUg<1{lw`-5#!~XLb64;bo
zk7q^Y31pJ;Zsf+dY_XgEHv!b=?N-&6@|Lu{3}1b_0!guP(^#)A+R_u_l-w5RCGI#N
zlyyf!yBiH8%19JcV8Fe2_9-A}@6htw{lrPdvB7R0vIo&hNi5;`im>_67fpGc9%U!+
z%8L5BfZL00^OKJES)n*8=DjSQ$(}7~Oio(M{TXyqH7te;C%j@?In6j%g7_=9#nr6}
zx1;z2uG$s{q;{RH-d4`y?w&zgv#k&wj)i<?*5*){5>a`gJp9&byju#3aS_%gd}7JC
zuch0q%LW}JAgbTyC?rhCeAdOdXARqat@}@Us9B?Tt>oh$Skty(K8RQt#THg9mdy@7
z%knL<N=twjga==gOV+wA%N>^(c3h;2dr(_0&;hEx%GIqsr-o`ORXZK|OsBh?@$<HK
zTabrdIF3@)*g69UO_~fP+k$#;9qLoSS$8Z+SuMxpI90vTz4QxHV1T@=`lBy7Ly_ry
zTL^B4$+1EnME?NYOW$xNJ|4@7?1}oP$GiuQk<w0#`@XB1jve3T%J3&Vq`A@2_HdKR
zdGW-yLeCHP*~#3~9T>+a5=^4ZgzY*3r}9}zbLTsJIw666;c`m)62~+7&QDg0KnRgo
zK+!xHj9HtC311+0*57S9ekt4AjLqcq51_duz&8giYJA=I=WVWfMi$rt9BG%UXBSX<
zczU>nc*bocJqkSjtX)l@ibrs<MTk@*;4VgMc?KzeAI@rat?J=xN$T(~pV{?o!K(KW
z4ppv-8&#_rlzW=!MZ1}kV$?aiRkcvwGTG@b+HIexJVg|Bu#_k?LQpR%gFA_HUE~|*
z#v?-hZuF9IZkLfg?)}p<lH=KZ-_sBJ?Llr&H<|JkNB=O<fwNOoHAMm8(HTU`*dyP*
z9pJFF`mugO%nTe})>>!@+Cf|1>cmTvh951$|BCp~C3~j|MXhZsqqPSKdZO`Ft#MX+
zj^G<Ha71*Cj#szFi9#q>pzE4yAYHxgzqGcs;O_6YSWRm=GZC3&wf^G0Emqf>9PkiP
z6UB42-rFiKpdyYMT4Os&BINb38e0odRsJv}ooqc)gpKBEZl2l3&5M$$TPgq*`z@Yj
z!CG5`0KeK`ZSiG^UWFvMr?ONA?X9VNItMr35$#Zbizby)B?=A1j8m<i)NIZYb|&=U
zAaT_8MJvS?aCQ75hr4cv1A{F3lz5-iS~ivboFx8l=m!ULAb99$KoH9|lHI~<pJ(ur
z1U^5bZZo)btTjmu>GdK6RMrW_HK{<|c+5<Ye^MCvZzU4Y`SIvnT0DuZDsFM2TCZYe
z6a@<OJ6W}sl@kv~+yo|C<L-lM^!F&$QZ7u1k(WN>v@bJCbp_L&N!W!2-8ShMvvspg
zIpU$|@GMc&F<K;qMe8tYGyZIGvQ!E3%Vjp&c^)q3hvstsH}AIj3TkCbmdJ9>7X1BD
zdHi2ZgvX?)Ac)2M_V+vOTC4@D1v?OzN>MjmPh8!^iM^3C9HFF+-Skr`=e}Dpw)98S
zjI994+pXnlE$Q7kU>rK6kR=}!FJt`fs|iu_j%Aib>TWy}gD3G*{L8(Zn7q~tf&bR-
z`*w6wXp_tgsO%o(>^6J-golCipMl?>*dw1sDZyj5y-=57rm$>#s#(5;+gcvKC{Tey
ze8u)Sx5PbvG%y6kT<ISU4ENe9Z;G&R^Z>MEdvX!@jS#6(Sf@47n_S7KZ0&Xzi*N%u
zU^O}HFaM78Si3#$V+^-cAg$Z(6wj@1T%w%w`grEzs0RZ!L?;yLW#JmD7EU+Fkz-BU
z3v~d`x~w@`E2QqCwN&5g5^D8d7@J3X`mHU#`T!hhcc=*w=sIK_-t0h6LFm2JfS2Jk
zJ?BAZ&g9)v26|sKK@i3WCvlkiSGlT0^$>dt6WQzI;WW3$`r`{(d#=za|Jq={ihzJJ
zcEnr%{NGs9F^7(73asg|?S-Dh^#2`R{j7~g!{b~!$17M>2v<h>OLiijRr+v*sgv6a
zz2CCSCMuwI+o|en*QQf5z)t5(Nhn05$eCOxr3{&j_F?^i1;O_{p2><SldAZEX&>zn
zDFJ-6-{Zz@vyNm3dnia#4Gaw{l<cg{=SUw@ZfwDSphP4`+tnvpi;XTUN3J%y>Vc0$
zq^@oEOIHdZ(Rc-RmpE3t?w!ePwcBpwFe4aoZ&tvUO7RMXvs>QaAX$*`agPS_vXKt$
zTUh+1c;=4PTt{%Rcj8&bT}kbK*E>3d(@&S(OL!w%aM<n#yhUUzI3$Wp=}~*&i-x6+
zkYC??LT>wxD1PCQFQWV!?A^F*yU+zKd)LoaxfOi&w)|aR5n<c+kQ}*sx)8GB-Ew~Z
z6}yhDj4#FHNmxaeu~qM;JQR1r>URtNyx3`L-Yqod<BvbKweQC99{EUim74rXW)<AU
zYQ5ohJnZUX2oGj0es5O)u3vfL=+2Aw1&e!ut4awT8sZy(mnK%1xixy1-J=JrDIrAc
zM9X{Fn*H4eT*56e@1hzkIDk>SsJL1yK&VNHp<D>uRt<EeSR+xET6^|RU}y1l?szve
zPQ<JH#rFDGUBft_<qJv#_uIaAgJ)8r0<Tun?F~v$8CCIOFgQb)KIdlpeQlq8)&>J`
z-43x1apjD7FTy}5z}UON;kEU29U8@+8@&eOBK5wF=WhI->)?<bPh=UP_Qi>4Uzj5h
zLmK>(KEl9(qY8eTsCv}3r)m_J>{P%&9eAwgr=x|$`30b)5}2LIL3Xeqli7glPp)5|
z6Dm5DfJ^G<ZkrAy5ci9O6HXr<4TEvPX5NjP3eR{~M$h_U?P#$%$Kjv<2cXTzSi*i(
zesFG6wvgar-1^?GKAupa*tp`_Vmwo6du4f-IC{;ub|YCyufJO?asSsp{l{+PEMWlb
zrnmnyxBji%d?Af$AY!p~7|5cg_|l~quW;{WMX<^EEOZaq-P|%6G7j3kXs+Y{3S0Sv
z>;8Ur*w*(!k3EPbn2-e~T1%XeRYoOPA7u|k>~t%Iz2}5NS1s0Xq{7c^*?WQAWHIZt
z<<X*8s96?~q!r$xK=GZe%s$xE&&zT{GUA!Gpty14e!V)L>2CNDEc$}6*1Q+E%Mp={
zcx|*1yGvh%HLHoyJ|m8fAm8|<xbv^9E`CAIu(Q`!qy8ix44AwN*6?1CCFm}<N23o|
z9Drn5Y<jN{FtiQUoa3-&=dI;ECzn10)iBN=^VVpmTBYZ$?Y*G8TX*fY_WbcL%G#bk
z_TZbpH$Kt<`OZ4u`_+TaNsT0r*+WJA-B(S%RAnfY3%V<y=HYXDt0k1@<1`ce-1>6~
z<@L|SIv9ZaO(G{)&JymIU)zY!6|ZU=doM`(VU&q|T!hEoi?dZ!BXYVP&wk<c`snv~
zeo~^*_~e9lx1pilPQI6FUlm&{IaMaA$$0d*(zTz;F0t{b2;1rR%EhBFTsz}`_jEJ#
zWK3Fne8Hx2VBFYPr(v7+Eq%JwbvtJ>zRs$qM4`J11%!AZJ;bxm&?REIEiRJt6?jVH
z>iWfn1dpV=au>Rq-?=meCxM<do)t!rdqn)_HHSx;@ujo69y5^=`G)_!IQiNucJsY*
zYaTdax2nPA2kV#7ZEyMWX|Q@JXN)5$o(_X{Cy^LmQjqv=zL)+|y2pF|@%r_(&)a=p
z4kh42T*z%zKSh9enBV>Z@kp7>9#wsV3nON2aU%D~O#pjJn^OqvfjZqK25Na*a`MB&
zG+EK+M<s&8(R>Qyq_tleAEz+`F9=&zJ;S?`4kZq)ZYzg~1In_mi3XwpasvGkW!Bmp
z5m!^}gEeghaiI3Z%lY43a!RiYNZiUO-f)V`NLh61eFT)ghxc1UTVQ`z_lLzVt1;i=
zCuvR5iL8h48e$8epb8*qj-ix)sU@H+-50jvJ8O++q;NnlsNpYJTRfAbe*uJ)r$u|4
zb2|vHFk$7kEQ${DNEXJVqs=8ZJOuappY`>xT4!6(<!uTG*w<FDY3p|D_7RjEMS(y?
zA=~S1;go(rG1%{LbFB>6V8A0CoUXZxa8e`rep8=~`GB(A9LuJKhj!iNcr_gv{T$rm
zxts^^%u)t{6W&J`sh>|K(A}dLNhh+U3$(*_sv5#w4XH_Gryn=MsdnI~ory+LD1Iu3
zMx@rsL<rbjaAqnukl0?E&i2Lp<NH6;mb`;j=n&@6S#PyC37#t_sraHf?}kEN{Fu)n
zNR35myoDU11?TRoF@%zvod-LVacVJ#h~)a+b}e>+dzIVidUl}EZ)9WSkRi(Ob=%Df
z?MOdWxSv!O+^Rlg-*5X~{ro?<(U;;AWi?jhvpfNHU(MK^oLtFjf~9W|t!K_>chT3~
z%b)B8Hp=_HXMG%x2f1hH@d)(K?NAaOt@Wb>hY)<T9YG2Pl}o{v?I<`3LA?x{wR}f8
zrlfs%-Bx5nnWsOnl{?Cb?<<<bRXdUkAsUT&NUOU#+e5jK&DQKFN20#|HCyYSZW|ha
z<Eq(Fj`GF>(Co5nwGM%lj=1>NWiS1u;A6c5v{TuL8+HUnz<Q$6WltKtFFwUFQW((W
zBf)^--rnq8ZRvNlM@xR*-9rkAem1amM^Mhe(V}>2i&?ocqx@^_IRs@5x+p1Z9SQD-
zT?Y=>-dw)IrFQOcx|%onjP869CEJ%n^&LJuiv77ee;-7KV>5eWXxCHeUx+|F15FAO
zF9Y77Pmm;ZP_&U9g`ynnvatlJRsM>{eChYYB8_{Kk&o<n_J;98awHKub|T=_iTfNx
zlp4V)#L0w@K)??eY9f~b^pqQoor+H-6Gl`mN^m$nI~_}+s7o9}6S9LiXWL{h7hWh7
z!mavHCP7^R?r|_}(-p81CZx)}U^DTI46;msR2T4avk4}?hJ_sX@^EE<_VY0qDm!km
zg&3^FY>UreZ+xy-tHCZ|{%{s{G1qwqR;6oIgRg!Wr-3fO>)GI7*M7UP<JWPlec`I|
z|J(oB%>=ZBL%E6VmLtPcALDM8E2uk!c}P@V+7YVja%;M&N{Ap$x0L&Le4dzLIB$28
zJPaPSd%4t#YQ7&ma|0Kp_jgdw2k{K=Nbc5fNp)cUL?VVq)qZ%6a_2#m1J-s1g&Kwd
zBK&`6AtX3t+wz@$zC<!TafHMd*&9~ujH?!f@`$bUU&YpN(N^sYUbjfCkP19Qb_PEo
z(<0drY|YL>&=i4R+#rg5IG>gHA9yh-N&J~7t_gOwHsL)4jAK@peQUTw;9-5fAJ{KL
zc|(G;k=(vR3Qx4gfS1S)wxnrip|m@odHvQL1L9i2UEy8;E#7>%un*Sycu|6Rilc08
zkKYuT?y-kMd-SSe+;U-`!?&O#o_U<LE1LH1be^T%I&)TvaUVF0wZ(zJ&0h>eclEI}
zg2d|M0@dp+O5H<hR74B8sio82pl|I~R4*d|tK2lXpOKbB-EU*w4RPwqT5)V=xuGy=
z4cmD9{Q-X;uV(bMoIR<1dd^Pxm)!Pqn3qoG3_QXEB}wj!Dz?DcB!G-9HdplJbhTH9
z6?Lj~)n}qp@D530a%Ws2+_RKtv8nu^yXQckIDMF#oEh=_UvdsS+(&^@Gd>~)ZW1-Z
zyQ}W{Y|eIn3ET~s9N{IxoU{3OCO@o;%uPul7h())8L(TGh;9*e6ekJbfv(Y#)2RAd
z1}Bf@yOs-qG7kUp>yCdozoT2f;cNW!f54;MbZq#5pu6a$<OmY}V2{0>pAB~%uqB@c
zm4JH?s_ytc%i;HK!lOB3*}WG%N#ijESyoQT`;U7sI1e9g4{{ZYvxeUlPQD*jGg9sv
zfA*;AL7A8o<Vkj;U2*1RFqGuEY*#QlJ2u+#U9peDlHd;6iu{d;C40llUC!{xX-;Nr
zspwYia@I+U_>--UUxaItCu7a7a)vi!N%+9lRv-VwjyZ%i3C+QQ10)aZwpwqbV&#WM
z5t~JU>bw`k08U7~4}gQ1q`Zb*$;M%d$5%CBLFF=SP5#|S&`(N9_>VvIQX1J>@-rpD
zCTYm7AdUDnjmkJQY;6gL^zB`qVy8X7Is`{a)y#xltT1?Pdvmp9J0UK((;JlWOR|rB
z`GpdkXnnhLwV)I;*6Ta*GogppAB*K$Z}*NL%yk&Rk$oh8$WAtvKiYZ@*|A;arZ_CV
zCwh@jwS3qVVR$^DgmF{w(ur(?_r*8jemKu3cNKznsLv+itsgy4W`8P2!<EiX`wt>R
zaW*;Qt<e+?4IJp{#?(f3cb^qh3%an&Z!R}5JeibjI^Os=xSZlHDu3_KZN{fF$^n|q
z4PbeXlH1IOK3r~yd0$Gb`Z)C2LV%st4XP&C)mUtTb2wd#)vxntc&_b=cm<U<bZclq
zz3b7B?uJyq-w05PhX@qX`883Qo6-7Hcq36?%KqA|$IoE3F2Hll+3k3y$s#vEp+2*v
zJcg8;y_5LpDT{QM-}O}>fsEbD9-W4wl_Y^;Jkzx@Q#Kk}k<U;_XKeDrgdI38N#^2y
zktzBIdlc(24L?LIUV9J%-14OKUDoagUmJ(`{0~{^mbaI~O_a=b^1inhvJx3eF}ba*
zzBmt8i>`n<u4+%lMGgS?kpY9#fz{Dk!nz0xP6M{4y`0K!FsG>Ia0*%59>*PnFS}gR
z?xaVN)}LE#{Nmb$Qr1;tiAnO67L)Xy)qA@Ecdu&jSrDhhiEZ@$J+a%5*3@23(gEbP
z=JrA*TTtX#i~rp{0G((}z%h16d-OTl_cmWV;_hAB-`-wM)6SjNk&7aA3Cs8i7xLcr
zkmo+^BfGM=tIn8+bfDkazT66&Rfb6&(w%E?h*0|;>&;zw9}`M{w(VEk6+WyLbfu6w
zi~Eg@WE;H6bUw#oG&c2@f45`p!3alesfWiAu>$I<F3z~OMjwT&XvgE*Y1}Gj?1ZmN
zDd~06@c{$8iiLCZClYMP4tnKN-ga^Fxwq_e{@u`EXZ&xt=^>kR1i-(SB*-4~H|0CB
zg%dHIb3@rG2H-5QCo|boOP|eF?<MJpu2CU3kfI<RyQQK^LW+5tPw*&5C;Ab~AfXfY
zl~4^9Z!=CcsmgpyJO65n?UhQLX)1A*uB>Z0ef$<-tKcT8mHI&rthwh}4BU>6+RX~G
zALVji@WW$ZZpA$CG=L2tF$M~~+rH^tAE+e8QgqEJ0J1v{*|tq~H$E&msLGj|w|n_a
zi8ypB*z)_{8K8qf53+$w7!VRNli9-x0##luMoxPa3*<tQj+asaJofdm_k&XqkKo78
zmc5_c1-Uk;6tn#OP`V(eU9c7J7joh{iItC=f~2Goc|Yx}r4TX)haB^G=zU7ps_DJj
zyTU~RcRE}1emSi5%5}3g0CD+C5vG@|CI+TdGzPTVXd!AS`5X6MtIP4F<45GHk2bh8
z7q*>bCf4wNQZuE2Pi|9a*7$xo@Eg&@RDDaSpk*QteD>YZJD_amEzw5zg}kzLNw-Gd
zeo+=DE=Eu+L)-hozd2^%1LEt3z{Z8&k?XKcnGN>7ALMd)U<jL0XLOqN+T_+PJpd^b
zt2b3yRd$opVch|oxVs<AVv~>P{h(NIJ*WWbxxZ>f2oZp)vf^MhEu}XsNN6KfyGyzt
zu_?{lSWdOzyp~9+w9mnaW6|2P%!%9l-o|}tw{5fIv5CUbIXe-*@C`yyeKNn~V)*;d
zN~J$#6S;6?))gtWQwbo<(G)<F1H?|pGvfBSt(9{6ik)$6B5JJ5CIi?~rX?`YZ?LIc
zYho8zk?H72rXZY2(M11XGmoF44+e4JmuI6xg<mU**!gOuES`(jGI7q*gOXRz$1^;N
zxVVLkgxKVkgZhW;YSo{V7jD0B+PCN{hXUl9_fqV+s#q+m+4bldItrB$OYoiDct6-;
zvdjfrd^4I&vq9o1RN?SeEd4a*Z=!2*nPCo*{HgDqpJ1MxDY$Xj?!+^B`pL)f2guOf
z_rpnxq*w#oA%%|Ki-yH&3Rw6u+5Pv+tvq%Z_txLwPVrp_D1q_NPX){g+Ji^Y3;J2C
zp%==++wK^S@DIgCeJF*<vfZf=@yrpne7B!*H_ViBY)pa6+g4;#Pr`@y+3q-O=ywm{
zfwsHQfNfN#$QNDvl^=c0Zl{cL!;$j@W=2~Z!+Z*LXRkQ}LP!l$6Rpp~Ur4MY4#jGt
zwWnP0D%IY)Zg(Mk0}_qu{qM!eKaW~N0_(Z?hOM#cmRtp_tGL&uc&2PQ=XBk(Db4v|
z&k?+m;Z0hqM)JqhIsM66cL!aE>qn}zw){wP5=da}@sV5x@Wa6VjGL=}gg@9U{$NwK
zH&^LMQEGJV4jL1tE(}MzlifrS+r8Vb4ry-44V~XoGmyKsKPN}X%1>aIyd(zwYgBGj
z02qk|x;y?ZxgCl><~=S>?$R`mC3c|Zk^aqd&1b)HJD1Wn`B|@Wi=K!U>EBEROFLN&
zrn9Z3{@f<A9i|V@{6b>IoZo)*4X2}9Wpn-Sb|&YC`;>xu@@uXNndX!Wcq+gCL(hJj
z_N|BSpkS8wgi1qtkJ$uQ!SFU09m$(hPP(9-&1VZ4#BCve!7GwCpc=Sv>EOw<E$;T4
zxg^8}A<#I{Q%OH0^P64s9w1iVprE{-{SlFYl{tUOrfWAHYBVK1b}LZ4pfx*G1rXTn
zfI&JQ(pbTnU9hDJu1_UB;0Qw^`dqU^1F8ggH^2a6;koZ6hC(ypM7r$0_pms*O&416
zLAM9Kr1&{N#U92GxOt&oL;=a+c;qNdnOnDJ+nz$0n3^ryQ}B}jntt58mhTB<mRg_)
z{wwyxk@={rzbIDx)0e4zx61$AzNf=h?+H?YI4pTd@e6C>S;?#iazos0?Vdtwe&1p>
zdwe(0k&!lB8Jv~NQ%;1Q+q!m75UFm$YYtnTcLp~vYaXG;$OHA&0Mpz^x;wtGhCSu%
zf#Mvm#yzP)B1Z5dC{0s#N=bU&H)p2@hdeI<{8P&wr=!kt>&nh-ajo%25raxu@z~9~
zZBHR)IG(myd#)#LCuE5Z5XF2dDnmj1dlPIp+pky`RjcZ(MwR$eZmWVZpV&U15DwgE
zPr9@7Vd@oFZ*~ONAdfGxgZ<tD@cxc5V9<AHL;Wi@l5;J_SWkI379EsGWKczZfX8yp
zSd*S^$|a3g)2_VqX)20;+<PZ!l%4Q}WovL29I=zJxlZwQQ$stz#~BXDcL?MY)jp6B
zV#rSU91$v!Y){7w$a{dd;F%abkJntuF)ua1WB|`CT2#*n6*d(e6*+D?IwCqxj_}c3
zXrdJsNt~@(d=06aOQ71Z!XCBxe7Pt-ZDCI;-io8RRAN`N{U(A1kJw^B`b3(rf|Ij$
zErynl&?Q!27P;v9<7XvQcmebFjXY{f7&-KT-OS%Ryf)Y^he%0IP*UEmCUlvLNw@R7
zEmbgIKp%Cd8so>J<Y9cf>nnt6{Lp~iiv!&8uXaDTefKCNf^B#ZAD{n4!5yM&)ArEY
zkMxY$qa+k$E>y%e$p<?8;Hcs^iXBYM*cIS#NJ-yO4*ibJwmjOnpFs({rlsPnR&>M>
z+Cy|;hZ7TR)mQYPN;9iEf-sfD#PTa!onO5Aimiz+ih`;fABqI~Gi=Rdf+&hIlBPvF
zUGEgp<HUSowH-kuMvus43MH<03_H+e^&N4Q_Z-3y^#8K={;^%3S$^-)3awCuR%nG*
zXoXgE%iZfqnOSFs%&=Q_mjowqrY7SHPU0j^;vZid8;tGPUxOhg<7BuMcg5Y(3RSd1
zl|xK0iIX@9Ax`1|nm7@vP=zW~p$b)~LKUh|g(_5`3RU;@e$F``oV{1NYX7=_+!+ny
z@AG+noafhh&hL{yp6ASNj;U~0b+fTgDDDcPPI8dqW#Ct3Yg@sx#K-T)!l0?Dv493u
z77m!s=XAC`V=W1iEE^S)t+lNnLqzahazt<QpC!2#0<3ml-wEU~(Xn`wk3l!Q<JRdN
z<GKa@c-S%F*y*X?;aZx0LhssOG)G86xGr@vXCrOpsTdyXvU=6Kd}fPG)FyH&3b@WO
z5rHyU)pM=iyRarxQHHxPxIoO}X*(QC=y%(*fXL9}tR0EkdY;9tE8F(bgdLeRg43EE
zYYTTQgyy}Y5FAd3A8$)u0D1Tz&sg0~B#;oTthfHPlfEq9qPMc?07OKhSR0&r%GI5<
znQB)E-kRJ0EPA84{Lr)4<`dc?K`5R<cFvb=aA1!uWYxzWv&HNKQHouv+Lege1*a}$
z84@r+MfeH+Nbu=jH%op!SA63NvGTQDjb93u^QB$OO&y+q(Y~I)$O>Jz8x9x2FzjZ2
z_?6@ax7tEFG%S?7F9d?OW9YM76;Vq{XLu(D!^;|zguA&dGINxb&Aq&-5E~pZO73R~
zBx(;d%a}dLp^0CCLELkG-_vhNu~Sw&SMWi%h`*K3IcDKNe)k&z5hez(HqQldB2Rma
z99Bn9j%h>d<2@}+Z+OGj`riY)iG;W=>cTK7<x=7B*GHKiWqBBj(cbV}h*s;?@LXVn
zqU-S=g&wiS08t2Uz6k2l(ri-&k|>F@phYrOL~T8ZpXj7X=f#&r+wxqXxo6-=F_A<Y
zlwfkT6x>6$3{B4!Y_iKE<264Q2yDOX4SASao(qPtSkse;LEyGVKSjRHXdqu}tNJF+
zC4I9L*8W`3UE^cKwYQF_qPRACWlB2WT=kKGUd8;aIXGa1IW%eg{%e5fi5#mzZ!axb
z&+x9HfrFff5ns_~+M2B%-8=o~A%7-v!+J<n;hhzC3=C$YwpI+?BSO3n>~NI1t6s!R
z^H$@dMe}wv!6lK06cop@PmIQ)Jt_!3MK~+I{R2CZ3s@2c(FhHZoy^Vu@sIz~rk^WM
zT1%JB<c>l6)Q9ouvstg$TT@WVIiKr5ZP?}$4rQXD!*kxpvrpMV1<@dBJr@%qc)m<}
zb1BNsFlULVING-4U5~>AUal5Fq-lAQ?23aijL9<TCbGTch@3)sHC*%d-F&|8pzM6k
zZshjM8xE03o>Z;}4+?$>^Q@5~zuiihAw^O8cn58yx`~k#XG8)Mr2)9>of6i#n91GC
zDoQv_r}L0v_oGaYj-+azK>Iz2GG`Gmb)hSJz8udI0h!~q;`y=<fAaTi<@1h2B%4gF
z@+z<F#vdDPbyRr|(1_k7GTzpB6;%4zN!4xb^U2Kjgf;KFXjBZaKod$^zy5jW>GkVt
z+wgo~g-3r%KJe!qSD}X}S6eP^Y!(-5qc`HI9Uk6ejrj{E3rjig0*52!w7CKkoB)_5
zE!Hw9v^=;$+t%m5Ur|C9Jkw@P@jb}oB3NpEz92{B7V=}8u@;}I!bcW_kPr}w1S>hO
zta~bNI%}`yQHid(-^w%E5km=MDrBfwnoi$__I9f!P*3i#{;0{G7DE?#Z4c%|Ffx(|
zS+w0qF5u%s*vQ4e%VHpF0U;|Uo{xvUIJ~0Q9<#|T!qTHgI0MMgsqCz)w$~2l?vCwO
z3h5)y7gXK~MRnBS#7rK$l3YAk)J!;Wyqcw~a9x{xB1Z!)U|FI?JXuw9IRhDGZMrH`
z;3LT&v7D=ha9poITC-919xM@LuKu;Tn7!|#QzCBqH=idHB{IRHgg<i2`XF<#^<T)}
z{o_d7k%;6<JPjVd6oaX=cR&DV;RTlB80(0gv&#uK@1c+@47pwLF0jT%@pCmRfFot2
zAy1ZTQFeyQDP@o%EMKqcy@Q&*h-vMHul&>CnwvRAse?f#3o-#Ow9hTyEoOtDW7KX(
zKdBWGn{X#U)~e3h-CPs0CiEiA9J71TMy!LBoS(v5-}gzlwEIDxh^Gd6(L2}+53z>S
z_^1wSMV84G9FmskMYi&Vf_A|1CGz8{7vjoeS!JtV2%IrSB$HxXwdRF_A2^Pfi`iPQ
zyc089WUAgduHE=7!D-1BYkdwF(O^ZjZ}0(0cv5^)8(s)Pr?`d?Pd=Iy1qN%Z;98AQ
z<^p;*N!^ryDi-eM7ZPERU?MwUdLVWr2%p+kuPk?&9&38R<uj5+NR=k_s@6i6RyeRZ
zA(BlTqKTGj^rD;MdY`h^C==>FM<3e3HU|O+|I^m)>x6iJ99z?ls<zxGMdr;~=L<pG
z>>wnfOj&ACNnlC6FSS^Il<5~jTZQSi!57NIuzSEpVj>Q_bjNQOP|n8@uKTP(Al&DP
z=&F!|9Lq0||4zQ(C|&lIBFS-Eivv5AJC9YO%$djSa0Ln(Oc+Mi0Y_daSf?Ka+R^NO
zNm+{&sp&&I=8M4MDmDA@7|apCav#zYIT%58#^H8$*~#b)^Mn#8^L-G=bk-UfCm090
z2W+OQ1+E38i>%mejCqQ{Lm-_PQzYp7z;O{K$ZzC)1+O$?<%^y3=^TIuUhvNdB(QAU
zSIp;Z(f9YYH{R&5OJ231<7Hc_j*ImDqjovB9+F_puH;u_=rZECTKy&rD&Y{X<uTTE
zxn2RKGwHF#Zd7|DA(iKGDzWM8W)9S?l%BWpJiRc$e&cg++dqp^rt3Sodi#_GQhJuV
zeh;CBiYxv-|G7jo<#FJ?AL}U1*C+Pig`m4hp<ALU%bL*IgPzu{+G8u);}Hv92t2;B
zJ;fbgJ3>TYyQ<wWh|6GCxBFex0ViMwxbIW$yi6U$v0~U3Jt&y3woX<<Yuf{3xIo9&
zwfiZS&A0r<eXA8AX}ym{*{2w>Z%6=<hz1kb5EW)Q&X}ttFz}im0E~!~h@vH_iZ%Mc
ze=q)Ho1(RZ!tdGU02NZdfGQc^miB^8J?j6~3K~V=YibXawdrTp98L9HiELCH{+8?x
zAEpx87yw$m2PCG7C*GE8gNKNW5|sC}J<9a>C=wkyRO@(HRtQjq-|mbue3M{2<)L)e
zYV8HF{OOOazv|gz6yXtRgY9tymbo3;NPEhzi17q*MfTvT<3%KM?)o$l8?<Z_p*zTo
zYLii0?v@LM<NwU2+6z*7_3yuGhkcERZvP8Al5>~8qa95MW%R;Uk=oaed2Oh|FYI`p
z2?)Eza=oRb5e;)9S~?yq5_~)98(a7qsqLn#I#`9#PVxL^y!S5dmdz&EjE+wLXyr?r
z%Qg`BdQ}uHY~ELNJ6fJxsQq1aTtYx`3t7T-ErM5Dj2=A7ng?#Xl(X&|=ZY;=3n|m3
z_r9z0)!5~{i%%j7>9;F+)Cp@CTHLGcAqCk@(&n)mdnGFGn)lz;ji8ATCH@Q2<pP4?
z8vzfZuYz82zP6jbIWRkRD?5~WuH-~ihqry11_r?}cYXk_^FJbAzOlR2RL>)!aHEQQ
zd95)3)ti1lfA7UL61@7r?^4$;ney0+L8fentz7Y9kUEIN*^Ux8Vr7&mmWQ}|(o5N@
z7vr?bK!_w)*wVh}?2d@8k8I70DPOBe$;b0=YhNt46sN8(TbIKWPSsD-djGyO{U?86
z8}jeRN30>?8$l?#u^Kl58(QOwfx@~K1cac`FQ&>!DIQ%V$`O)s4Px{Q+x%h>3wwHd
zOW^k{FP2N9aQRy+u#|cQL<*8^$`L(G9W!lyv0U!I`YUUBG0-+LVa`-*Ru$SN+gGo(
zIqWPF?b>r9KO;J2M}mQ3q$rg;9SnH~)N)EtbyD{A`)V9I#A`Vi)001@-ZiDuU2c6o
zlKno_%T7aB)?aj#?z93YN7AuE!cJsW1+_*cW0Ns`aaD>QAcKyns!TkIyFDGQmWi9m
zvJw&3jyU)~`il-bn&8V)7*_RIuF^<9!Vfzh=U|A-b;4V<^5~v?@%ul5QH?lDZt3AL
zySWfT7cSDYpEBj9Wx;2>tL=&?Je#Xq??>=jLG8ljPmod1=XCKr92bdju9`*(OOM#%
zBU{LWx9JzQ7$;ij<CI;hYRhb{Jg>Ht&|n<0&o1Y@p-&Wg<4UxUq$R1)tGO2L!*|V_
zJc+w@iRFGh2Y~_m5|l3n>_$~~3O}X7P}|MeRDYX_&ky^h_$IgFi%TM$k&`BzaMIZi
z7ZksLMz~;id;&#yMevgZj(y|g+{=>#-LE%XYF8X*0qMRs8zpTXAsHU{nqjsKtMpR2
zjRSps*rg*zTk%r4^tf@2*~*s!XJAO}QutLbCC>N=(Zg!5%VrYl@g+aC&Qd_a@dP`y
zxz^i|*qm!)>+)BhnA!C&B^fn>f17QH?iJhQBWs8*W!WaRM{!~;Eak?R9O+?zQt5FU
zUn*E(2Pw5T#pmLA%iV#>waw8r8dMB6wp0Tt2*=i}kIYIuvaFw)5;BZt@EmW>257<;
zt;K&qA?q<{jpi~MpJP`S;FjB>Oz2F8W<^7s^|;KKWbuW7200A82EpDsy~@&bH@3LL
zyGlUA+Dn1%Yj_{@`*vX^BKJQGZv7*K>_DH5#DaNhM^BXJ1XVr0$Vycw<3sE=3NvjJ
zF|qg1EJKszj`xwKrsP|kdMU84to5O$?Xa)&lZ4?vl5KG#BrncO1<Q8)T{~7yUQ$2z
z^*<hE;@8ANiD^6$P?cP=P(TWG>}Td=)P8TtgF-7i-gHhKX9wOAn~Bz9$Fyti#4Gd1
zSzi<k$2s70IVD1$!`CCKm=Ca}OedD3@4z^23{vecyyUMDTgK&;Ui)((dl(OYu^RQE
zVXjNiF6E(MGsbKw>PaOn>tBLK1EU09mvco+gj^uJQcd<#P<=6Uw5w63msW;$Owq1Y
zGr_6}O9)a4$4q2#LZYo|iDj)KDal86^9Ox<V7ptsQDekI2p}DK3m?g3PeuTDa;$Cy
z{Dm#OTTQ$Q));l~C9FlzA)m1ecHhS!G2VmxS{z<I<gmS*M2ljxJGZmq<&*=C;LB1(
zVJlxQxM*B?x>vpIm`BXgXSUjZElm&g@~W(PIk1SRu^w9+y(`)g=2x)Br)=HJ<r0yN
z>K(AovcAv+#V;Y<FRUScNpSa(ZH!+m4U|P=b}YYOHsusH51U>t_ronDL<DZG=2n6b
z61x!Hag+%+2$Ny;Wm}_6zMfyO1|LJ}H2H7}>o-a?>X!>f`_Zd!Sj)@hZZ*7Nt=YZg
z>ktx)=vCW%orb0GYxjvRO%vwMI&w{kJlRjYS#C||%Z~4P=yl+&7X6{2m0Mgf@$HW?
z=^a0Sg!mA0VKB-v7Kl#rh>yE8{p~-o`pcoV9aXRq7J}fuBN+GRjCP}dw8;dniEB9Z
zvcDL9LyG-G9H_%N7Kjpzz_Gk&M-on+0mvwXmG9b7AHPrc_L%=IBM&=X^^E;nxqEU+
zk>Ke>^y1PGbQCb7CzqAEiS)-d?SKy9Cxz3{%MWE{(w4{;D#$XppVKy*b5!6&oAX~}
zwUc0QJ~|fxCp4g)Kj$?-!|?jHkb5Lmn!DmJMkA-DBol+0i837_tXS#jGh3?aiIWf#
z_{=URFiLV}SMnG8KWtZhjWDz#YC`Va_MLXUS}zQ|082t&Ss7sGAE4^pn6sNv&qI5-
zld`8pyyfeKjUM>ywx3%vkl39ZWz(Z}H~%In=DmQt3XBW6bAQ<kr3?{c`5+HWzZA+H
z1sy!0L<cK6f+Mf>?iC9LebQ?yI|`1#;<RfYX@fh;ooRi>R(Ax)QRKK|w#GZ!hUn4O
zb|eLYD;}2n9R*K3^2jT15Sh(?EKTo0nX|zg;s3hB8af<*u^C`zJ!vzxF$Vz&;mTl*
z9f{9KV2N!!Z11M*ua10qbMzr|8Kq|V5S7W6DAPlsjkw6c-<ti(l@$)9DdE`ptTk8l
zF);^@kGT{r`Tfv8u-5!uY#&@^-xg)Mc6bdS_5IS?y^X?ySx2^!3<?;&ka@l@EQjJ0
zhoY8CwzCH{r1fWq%56piC0`oMDkSaWM2z^)osl}I-r*NY<TB!C%-^N_l);Vs1j%R-
zJt~MYA|a-Q^+~SXqmMP(R928}8x?d54@a5g<R6CU`$&gl&_^2VXf7jJ9mGo<^C@A<
zfRmsdkH*4+_zx65@<g7dQQ7L8%<-Q9uTE!YGT#&_(vie3d`mp`<)YfGuk5#nr3K0k
zUm#kj&1WA_GqRICmtT?i{FyB{<p0Z$Hris2EQ0w`HOsG&_ql`4Qm%#&r7#oX=gS>I
zx<j@Rri4AXk~c<%Qxa_JHMr`7a@(L#>{@j69$HSBd+PXIugVm##Z{CyqD&I5FR7>S
zVmEU-D3yqyWVfP@c2E8XsNwClm!bA2n0jc9?^F<6#8~fcM^KvzN!{Ch!0!1v$OZa-
zEOJpkmuc!L#ka~H<QI`HuGQm<6j}C4K^a_)8C&s6!2;i&fXMmhoU)a#I7Pv=0RC04
z_?0H56AErccK^gyzfz9RIA$elUMXl|J5P<g71q8IH2E$PaIbsi;rk!~hWk{?#K~!?
z_0hGgGn7~e!xY=#UAOHYwg!i>MH=Icuf&`9P`nkLG*+D`NJq&XZJS~S*j^#r+Z;9A
zI<tg#w=GcvjsX0<)d&6KLqD>nSK@pqIs$e<gVvk_An_}511?=lfTO%6A`Wn?VSeIT
zTeD-zu%*v>=#zv3<kRkRZ}(nN#P?$IRxi@_j#mORLMjaGl0}Ji#^8u-;SSvGrS?h+
zUH2mG*yqt>`g7rXi1#rVK)^1d=NidwAy|vIUd`$OmxQnKhjoK21QRxqosSF?btWX@
zli33jiO9`56@Y11b=)LHJ?z6WCwMkH5;Y3l%f?7&cQm1FLzZ{U_jjn9%E$e~K*?)<
z!ar=2qW<J7<+0k(VABZ%vgYBzeR?-$mI0xVE3yWgt;*!eDhs_YY_8grb8uJlxy56>
zB`OanCfeXURJQ2?cXA=iBotzaBr9C>!wwDEI|vPTDXU2D#qdJ#OSuNZ>2}$#78*Yy
z{@{RdCBKdh<7Ri&CykLL#MHHbQBjZIv+KD~+xc|Ee~{}CIl)a|oTcd(n(fvrffb0)
zUYGgO?d+c&*kgBc^qsx3g0Q;@Z(-s59Nvqv60Dm!d=7HBAE3x>XU-nva$%fC5C}|B
z?bU)3Ff|=V?pjoVw&K--P!x^FR=ygf!`@-scWhPu+{w1#f9$QLwL(MT%78Vm#;xn?
z+K1T^B+J@YA0i>Rlx%|}?>#mZfvBy2HHeqT`xN@ZHbe_#NW9DqRf|(a@=o{`+Zbg+
z*2*_6#dBkh%q1Kk;@g;Qs#<ESV)t&Ft6KHCjPYx>B}T(X9X-I-oMH*uZB6k}90EBn
zUX4ev*nkx2ALmKIkXvFXnbAoeAn&|ZZz~@#LTt5FAWPxn&suv<TmdKvIMCrs|Ls4f
z$IhtlNrzyMA)aap90nsOW&2+(H)`8Iw7~$ZoCK#9078*JMxq<&H{?j7pWd_jt1d_S
z)pnckKet0BPF9`E*}|q0wlPdJ;JOay+{+aaCz-}dM{)@GR%v0irAK2Xr%}uu%XRn_
z5en^i4uh-$?%_l%usmyFR2&>=U7L<xrTCJ817^xL6P580;eyU)zhk|<du+}Zqp8K_
z6VhPEWelMsg1_IZ<xN9L!(CkPv7R8z?P9go%Bec55$%$Hei9aADLy-fRPL?wcKOv1
zqVOJw_3|+JZc&YB(5~c;K;V5=36LdsxavS)u%x)szD_N&thye<NFgQ7q{0f`@L6qa
zu$vVSU-LS;x!|p;thmSGkn*wJe)W5gA6kg?(J_~5-dR>tW^~GRaM#Di>GD^0&u4q+
z(5rU;)j&mU`{fXhW~OQ@C9NXxUNn2nZ-So6Gld{nq}R`(GFH5njDrOB{+z6gGCd&p
zib8drvQ<?*r85VGK;W*fdKH0@dJ;v8Tobi(ZJ=Ro?Q4!$#KV7N>-^VFOd~l~>*Je*
zF_0p!m3<$B9wglKYvtDLY_W~6Ii@?n^=z!bQKo6P7v2<Qa!8Urn_h*@Q6`qPIBFxD
z*z#KPL2zHl6w!@s^{$tucRp!Nua)ZtC1A}C7I^@q7HX;bQq~{HQ~Gc9%7@Tdw7nL{
zw;x1GEa}(EzDU78;zWi|WN_R%eSTQ8uwAtr`pM_4Kf559gM5f%KsWl=q5aB6e869z
z|Eb4kZLw&HQ^<iQq6`ryT0-HEMVovra6E{@mwU<qfKNq%a=04hj3k2PHXU)`We)3(
zTt~eHVODzg+A&|2LN(Xycy$WCDeRdR$K{GAqGtsNmVd<pYVDIz<{RQ6L8o7f2L-NK
zTI-qD9^MH`y2p!hbT-Cuq*;hQ`jjMnF8b7k*FwlIcs>B;N?^g3YmEAD!M9-_c569o
z$+`zs;Z2ZrJ@d3J<*Fz`1pJq*)Go&<m7~x5b|qIq_QrO#npe?r$OFf&c?}YcaU|@z
z@7xQBJ#P3&gS`X>vzyDH=LQxZ^INY4KkAmN+*&<|XNzqNo(*=p8tsEZMv6wd^RORx
zE1>MIA1D%D6ei(bR7rO!#lpHDlTs|Kv-TirlpOU14S#6X8Rti2Y!T_L=!~P@Tpop$
zorztlsQ<+P;<T;mOzcvwX^}Rs?hH<4M{l34=`2WLX?la~c-D62L6E~RvJd~tq9}*Y
z`sfUiAjF~sK$x#}7M#$nsLu_ZfeA<8UJ=3A#%LfDmTy23a^`HzA><p*rQQ@nNOnq?
zg3faHcamm%OOE|pa-7=M&Vq%WBQ`;FTC!pDYU_1r&N8=N<0&|AE#B1#W3<NS;t(S2
z%js!zz)OaMcK;hLW5GJ2uFkS>Dp=V%vwK-?apvbRuc&2N5zNSb>#X(X+=eGg{**!Q
z7Zn=<kJuwtqG;tm7xPeb9`)P;yua!;k&D=aagjt=lUdM3#9~Deoyz(U&%<^&;7A_t
zGwA#yF|LG8N3%Hzaf)E$San#y&rHw*IqM`I&#uZrcg^LAC}YydF@yqf)k^@dw5r>5
z9;ZGYi<zuW3<rr}Fd((rtRU1i7o)p~x9FX9xxwV~*_5;PWjXmD*g0?7J+7#s3sqA$
zGtsNB6+CN;aa5$mx?q?57xa?M8<+eTJW8I)%YKHQLaiHPF&{Qm`1^gkns=(C>#=M3
z6W2(T4cBvPi^&%O=<{=ZbR25$rjLnrqqwtMKIRykb-NnrD`ur9A4HK`f;C-{0^SYy
z<xBsu-OKgT(Y+t-{W!X%+XFAdv=kj5ZgLvZy5jDYj4<fpS9Fz|q#Va9yUHC!$}?)K
zx*Q=CF-f_@i?+He-eX80IwOvmt;uRZu1RCOw#%g@$G~{UkPWS?>U@q&-rueDU4c%X
z8tp|u+7)POTh}NcHFSNy9`k@l{$KuM8y()I=}muTjnyEZN==C{@Jrj2gKWp1!#3w2
zm;`WLwsa*fOnPb&YDt!&Te|}9^$hn}lTQXW<Dnt(kImjEXJG*mS4)iQkDN;rTYW*F
zU=iDLvE?eQ)o=H%M+PNF?+6IwRfGYa1pMVV&s%3#L1Jy&tX56&=hh#C$u>vUKN3a^
zR?|KOFE6Ta(MEEjT{TShte#!8jN61)FHY0gofUEBWLLf#L_0#uIcif5r8A?4a|BT}
zTtR6Mj^w;rM~8Py@_RHZaSXmLh>xHpu~P7V+}h(&L&_S+Qu%n>iKu}Y5lq<0s*g_#
zVIoV}>4g2;Ih%>^BC@6Rf(Wu%|Bg`ZjLrGFA3y>yA62D0Rniz4x17sW8AGU!vbNfO
zavHGga?9{rvsl4_e?H^Ls_;^j5mttJ5}BN~rRYH_UJ)}q?S;%aWtX!y(q)&$pSCMi
zuablmY-(4dj5~`AUUce*cCE_^3jNZvLRJrYExrW2qhH(gTr)@;^6(`L1?v_EOb&tk
z25v^*sPd8QiirEhZh6_#G<Lu!frk_`^S19baZNd%ccKD;8}h$)Hvp0g=qF|BdfCpi
z&C;_h+P%CWgI(j|weMGhxN-$4p*?V*C79)xb(g!}igP#C$Sb-X(TGft*i$a&E4vFm
zKn*oI485?*t3-yp@_)9vyIfMGz1wGNx&x{84nZxzrmXc2{>mMFY+ZLjY{RmVTAy_Z
zTV7Yj$qfO3l!Xcc4hYuZAR@$4#_)~)i&cXXu}1&JA}j*xrf6Lhf}9{lCEr~2Sn~aG
zKF-*dZr|63mL8{KUtNB(HT&V3$>vgiL`^<NZY%K#o1+6Ec;N{}t<qzwZZ7Z|v_uE`
zBFjBe;#zBzk<_~QH+_h(s;%4cUQ*|7G$n<9dv~CBvI*|(RbDdd@VP4J5Tr?GcR@=}
zv|7!}k#FImZ~Y194wiG!EB#f2jdX_;c7(7fJQ$q016J=2Yy+u1UKEWJIdf#J#U}u=
zd9u4aO22xp&8EC_sr88wo10fVFoc;&j~(%fh=}cIP8NKlBPwUEW7&6+=Xu%jC?nib
z(ZGESo;l(3VcBG6c`{*?8Q|V-pq}mydKdI4Gq;(nfm=Rydz^~dSXLPhT`b#kfx}GB
zDx1r;@N_conD+^da6Pa~i)Om+BAXNy#l?;ber9otIB$y&r@^kf%;l1IyHimg2l@`!
zQf^6c?YjK%w8$QfV<{B%P>6k{>JcN|81ZT@E(<$4j5Wly+*kI=u6KXmD4RV;RvAv{
z0|~$INtUkNjh^+IQir7#@&ZLwBjEs8YAxE$YO!Q_suP=wbt_?j1+I6YFT0%u(gEl)
z-iiLi2AvY2#1h~2wUTwN-OF{ry#~!s)W&j1D*zeynLVgB`%|notfw664L`CKJq1e~
z5H+~6CpMf|>54SEswXf^t(TZ7oW#{V59y<V_qkeYdJ1A=n2*5>YkT54jY=$CIOBEM
z%P)ufhv0l{eYB8i#8<W<e_<bue6}YfVJweqawG$j3Ckx4sbp}qF>1>v<r8ZRcwEH=
zk%d}gQ%{gpin|N}<PAy$G3cM1Vc_Ad{u_#IzEEvV*<ny8Gl<xlqfAcAazrIiv$X^m
zLY#}%+T+;8pP~{ji8tdL<bG`q%qY=7q&97@x{@Lg_6dW}j)Zpz`$9B32_ag+QAp%k
z^j%E$S!CJ1`(5de&NzYP@)bR@@9X1xZN&Q-mJ*?hxaakr@^oNp%??aNQ>nR%I1{>i
zGRoW((PP(4Rq%?r*FL;whXbG@)dC;|YjhW$BR$Ko91Jp!J~^41M{qk9%QlM`;i4UP
zkh%^^lXt?;A-5XcvHuMvSF>qvvNVmFx6x)2aCiZ!R%F33>$S0$eBS2#?{5!7Cd@}&
z=_Ws7u@xchTrTzqmQAN^A?E^5$1KDyan8=Oaf?-BNo1u$nz2h!R(x%!EibtxU-19!
zIlJsb6CrTWu2f*o$?yr5c+sv#nc^-@aXb`L>snq7-E)cCu2&6(W5PT(%5<V+Ji)(q
zGs>hV)_EWfncb?&6v0vN<DA`&l^g-<-igWPW2^u$1^3$BT;o0kA-tF4At*qJ3DG)l
z_kANvs13<39`ux_N}?#Ml|SqKdu>H64uD6=l!ZynZ-&d$g!4q+>A6`|bCJc<wz^g@
zP{+#z`dU*94ykJbzoNCZKsO?LnUR9ktn=CfQaG)zCE4g<h2*2_ZZU4gy&>U|H*c4`
z*Bfeqo~j_Vw$X=VKR-o8-ss(vsA7k0N?=+@<YJqBp5q4(+LoA=EGdgQ9@xX*r>(Vk
z{UwYdZed;TcT<AVlRllzF)^7{K~feiX^W3c#G5f|tp%MY%CW5$SeX#tiZjq&t%_{;
zC6W8sI-)G!DFP~3r+1>fGX%QNMjRihl%zZ8ugYMcfI>FZg9+=%FszcGAIWYcr>W=X
zotcDcq88{D1joUcj6Ox!>PAWQHx*?%lBjgJwBOj_T-&}uOcjn~w>X;-8-ad~#?~%N
zyIiiY9$47{AFKKl-Q$t+k0+$hfO<|;72HktBeVo3t9^TDi|VY-bQ}U+*6)5}Gu5dr
zCO_bh&$)Nx%eL7(9@|=N&Szfh?E`<+Y~By_kN&=$i=kywfn@-^-}|-@Wy*uCb@_&M
zS**qLN>I@;5TsqI;FH_IGG@q@a=FM~jMtz;0Y?0}UCH&A9pQvs&BNmM19mNcV-f&v
zwd>hY@sqkyP4p8uJm=>8Y~0Ltl8dmMw-VNot^-A1aQkn;>Ni`m!#lYRBrvzTerx*3
zOM5RN-7k{R?)xM*5%kp_I8?<t2v(^#h;;~;cQ~s`-{p$Y74bz&vrx_h<RhmF=Bc-!
zldfS~)m!ck4huDu%DqmUC?Y9(pWfiQe*3=5-$+5QHo7DjCc>_Q?dVwH71#9!uGkB?
z-!m>9@OtkXB`l`p8=^PKrR0wIE=aziH!vV!0H^g|+nAGyQ-pjRkQIunvA5jo$Dytf
zN(|Yi-Z&0y6%JN)+w8z&Rv><3OMdPf94~R8w<a)8Kv^^eFp7*q#7y3^Ppr8r`vOy>
zV%}~oQFa>S&RbVf|F!x!C^AGFv}Fwryo8j*mbCWXf}+Rwax**f`_kF(-exMYq3iVT
zMUGO-QGfeIQ7vk!{R#U%EI!3J3}!)JSC4dDHWC}f8H6*b=Q0*+8`xc&h+g&lY0nj@
zeX<(q<05#L`^=O>0{%wHTtTW-pz>QjZAW~Z;gLd`9<6G8%xZXegJZqrJ$wA=C+&EA
zmR%icSB~S1orp5Ww4pT$;7P_tB&})RSM*Qrv_6x)loQ~xq*(_O<UoR;tly(Z_#`aO
zwE+cl*YF4I4HS#9F0bN3oG9mP3ZcfimCf&Bbv-H{D9*CXk1hoa#Y2}-Az^Ik;Y>I(
z`tq(_jxx~|TINr<HCL)KkwUs5AKTTKu{?*+BxMa4cp+mRHM^cbk03zUYd7++l9SnP
z=8Q3*6j^*Lx0@(c1Kq<lGWKMR;!!nbcXC@F`?=lq4cbfSsC!;4;}E-_1A4}Y2UVZl
zu<Me2T3=ENRkqf0kQ4?_Z1ak~xPNk(C5ncv?2Ct20u^Y&Rej|a6e4MLpW`UKX7BK(
zeOmJBevHTQnrzjF0E69In>GEd=jxx1$7Effqa$u6+E9cQ5OM4K0)Y(o^pFmd`?$eJ
zU77|@HuNR01*>5j`wGr$-f4|qUhLorqtX}W1q*51Z8rCn11){lmV~W@n;){R-UBN8
zroQk05JorAmLsp_|5|fj!K+wtk;$PYK$if>jbUc3H75EVh{O3?(Rzw@(iSUUo;}?o
zUN`q)(dj&6?a^|Xn;%PBqO9b^cx8L7Gr?DkGv3QuUtnlN^Ti#JOvecr#@+G(SK7?M
zgd7t}8;K<(%R2;k-PdNUUbBhpyjw1$LpB*bDS{8&r#MlY$|HhpG0zL4ne4i$7qN`*
zRL3hn`iUKlVPyLKfgMYz*!_38LYU;+@tj(pIO`Mn>uF8wq=Pkt&)sx174rov3=a9)
zW~%N!g)Q^fYSx<on9ap!Sv*ONrgWn7c_w~II5I5Ta^*6Za<45^onm1f6yjpl(-*}W
zUo7S&KQnEQ+fsZ504M8Nq02rGJ*DU(uH>kY!W`wRvCo27LPtd~^SgH~r@vEbnCrff
zkSLGYjocPwwF<m?Gx{hR6|DzX#|h|L-Uq)F#`bph$TjM>I}b+&RN1?{!^jFEb~iyJ
zo-_)>d%45pxGhm}?q@&5V%cR6@)sv=x7n`5C{h!aebB#UF+>Ti*p>W?#NQ%`-?f!d
zrr-=9f+dv4wrZCn6SqgvJ99oZVD+-v+Wi?eV9l;_?MM4<ZL}-)U&3XD4#AYf)<rvs
zW|dZpbXm53SKv1{j03YoC4nsK6TRdOQ6_Q2X*EQmvT;{IAf!aH#$AqpBy)JzHbrH<
zKIsUxIm&{WhAR5lw(JU|HiE3G<e1sID^Qfotp}~iyCt54*#GA27PeOL21LuWWEY5O
zE+XtBs!9$d$0MLuETOhtDZo;=FC3B`^2SFWwT@kZzP3XpcUEl+=To#QHLv@VEq`kL
zG0WMd&)`%CbKT$}Wd?2|2`vY7v(>A<T$QE7Koe0WB!=_eZ%H*`li6yg98IR8hU0#x
zZV%@iiPOngJXuGwO%dWBt=gQ|0V%Fb$D)h}-3d!!mK+_Gnd(Urf5Lkr4G(%iJDH1t
zQ>jh+C(a^<oi-C^+HaP7@^L`e2xkmtqw`{S#pcK*XU@x{a>YibM4q?#UH(`~ihs`k
z-aCv%??SXkPReP8Pvw1B%+mm~fv?J?ob?WfZd>x>{`BK^c~|EAkqtIpjWWy!tz4<P
zlvw{g&I>_gqpVyRK>w2W@>*UqUTcC`u{Sd@sq4FfZSRvQtN0s)I5(Dc3(k<T$!=CR
zPLz)5D69$fp4qLat>X$QiSs%*$S2Z9BzBnkJ5_&QabN)qP3GMwBZ&`NCBERi-K+NF
zBNTl1{aB86y>1U;@H5CcxvwSgm`Kx)&H598D3-FL1J8uFBECp@C4mxou4B$C`^yF$
zWEE|d|J|)1GpqaK4&jj{$B#ArNu^-e!Ia6=W^4O{4uRCWXzThN4Z1W#Xz+*HyY>AA
zVf}*++u-9soj-03(QlC|G8>|q*v2SRPYB1ceE85B`zur-DsGc^Hco)Js4P~-^#dQ;
zmSr6%aR6yYY-^OsiJZwJvH!}N`eVcKhm`fv=m7eS=9rZny2!U)b)^{iA{VfybNE62
zeb&|=2nCI?wR_V6$W??b9oZDIp$-~!vvqoPg3gf*u$KEJ-j!`<ru-{;V!t=lRZ_zC
z!fFrZ1in)Y4|X2`^hlJ6M3;L5p{1?vKwt}s?A<1^6<C1%09$-ulQFI^zO3`*gD*?O
z&uuElE#eGMn|XLy?GLGUk^Ofha3uQH-U)IURUVDn1tkCx$%mr$u{=>&KH<q?$9*Y?
zwca&iC!*QI!!uJ}()aCTe|Shr1qlxbn|u%gU^>SpdgE6hEtz`G<Xpjml4FRKne}s1
z4D=wd=CT<e<HiM9_L<G+f_2Md>0H#1q*%%<EURsy>KKn|T{#ZMqL(S6vb^YaNl137
zKY4e8TFRzxDH?LFxvQX2F;~mh6B*Dl$@!JM7mqz^SM%?kgud>tL=dB(B-w&q&ykQB
zfoaLp?2{-2HU7(aTZG9F+f-oaaT3RP(QapzF=^54j*s{4VUW$;>>P)uF<H9YO8_L0
za3|yY(IpE6|L@Yg2RRI`O`*^;8z}d7d*oAEF;K3H66w`#<v_tbI0c|&8VF7bhl<mz
ziqq|qPp<YR!{k>cJ;y+-@L2Dzax2yjI1ZA8@U*Rq?g<u)w%dmW%_t<YZ5T)lCihm?
zPc~F#UxE?Of=i)us<L82#eKYBjWNnu;JRp=qK1TlB@_*~aPvSCfpS^b&61mO37lI*
z=UbwUKcZreFt(}6WP#x+eVTj*NPBpDng<HW^~|$RT1%D_hQ?Z>VTsAhFyi{#a<;@e
z8ymOw{E5j2Mv5Ii0TL}K#&Bo;+PmY~$F1huFggJH-0wZhd4o{qXdMxG5Bks^gJ{H?
zA{3SbQaw8o>tz!IDONAhiL}L&0V&~@iXA`sgKVl=Dwj`uEJG2v0BLd*f$eba@wR>>
zVnk*?;-heIKVXlKMwdn6$4bix|CrYxkBFRQjt`^)0)gr-No7yuYRI~nK$Sk@nh1;O
zT-a@(p_vL?od;Qyd}^~%R^s24xAT<E4aCFIO8^j=VMsAP@4MB7^5k4hQBQfX!Y*&-
z!+d@V1M#|&e;Ki#EmmX5B+ykHmvR|KBo}5MmZFl#Nf#b~Onj<cdq=`|+^)o8VJ^s1
zbv3S4(R;Cgm|q7vzBW*v<EQXremH5aGFK0&(2c6>U5=(SY-jAIxA^l8yA@yM)GfrX
zyvMhF9l<v&>YZHFCrBB5H()?w#KTo&Wt>?e586Fnu}2^Msol@X@#YhKfFu2$J&1#;
z&~&!juL4B#irsO{l<XAI2IH;V?ROi5UaX05?y^-?S)r4P8R_cXj%6Z=4`)ZpcVU?|
zQM)(*M%miku}jd)@-<zzyWko!Ta4mHw?10LD_V>!H$<7j=15d1zqf{L)plS6SH;~0
z=aodyjX4q2M3FaO)9!*FM~3%vjyG2@DX0b_J~n2HS6G_<B^hP5?oM8g>^nO7O}oP=
z-AV8gu#04ezIH_4v6gJ%?z05bk1-UkT7j@+N8GkM?#~!nOa$I_YmZum&SZB)ZFWQ%
z^M|q!Bm0hZ?shC#NG?{<Bep0)oyXy+RWlNP7uE8i^(RPeksNj~m%aEeVioL5wlw`@
zyVdjWE)4p7T8}nvw#n>c1nITz%2W<i2%~aN?QoXq{T6VX+9SR>;2+6sj|TYV;xo-+
za8QT|P4X){R!vzs{Z8YHcg~JSnV5?+pd*QzPE_@zlNTZRg`G?gQK+HX0`cTd%BXxM
zmz?}AgN4AE&5~{`#gKaR+FaFZy3Zrf$<g)5XhEgj_B9$FI<VIka*!U}pBDW?iM4fU
zce(y8+iWTBM42Kcs&5rri+A9yFMFLIH`tZjlII6T6;|YG{(go?bl}%(0fy_23e{)V
zbLdfUz>TPoB)tM*-}JQ~7%Ab&Zh7^d{!tM0?W#Z0LozMdJKo6!d`gaONWkyrS$OQH
zb}x2L{hwv$U>WoXSN!Vz1ZEr9;=%6VCB{njC)*RZh(`=T7{Q$OIFV9f%B|d!WW!el
zHIt^<R_$>#A_bHPorlS9R!42(U2`cUOhHn!r=YGETc5PGdkO~Gc>tL(Y1^an8N?$}
zI(@|5Tfe8k2B%+`+6~@-F=V&UuqQCzaJNiPB_rIpr|jcv8?7+`Wd}t;$!KL$G#6v}
zfo;x-Dmtl{x-HR!1?EL@M7TBP=8Q4dRCIDdkxlXC3=D__3TNR}i;225`!K^`H6F8;
zgc9DohNy@0(dq+1R-jBqx=#=$YfET^tCX0c-IrZzwF&D8(8&5$XnD5OIv;|otq{pz
zwd}dLFS{R5|DM2Kz3CY=SdAr%F(}%RtWr>~6p?jb_yMHl3MDv^E$}aZdPJq|b6lE+
zJ>G9q**?5hkO3$3V`Ppxvd8gYpb*(CIsPtaXU>jRvm@v%OZh2?<XG&LtX602cr3Pv
zbsXeMdOsIco^U`}OQwPlCx7BoPrjCEKWslb1Wwsgo|>iUU;NZ&9Xchls@q&PsiRRu
z)8yZt#L)48oy#=_NxJxc(H5di7A!Lb)6dwVH^LH8d{0gX%H#tk`XOedOHrn?BT{Ms
zp5t<^(d**_cEuqb$5r}jmNyZG>{`qm>RQYehpn)Jh+VGdp6nzXZ7*EVjhx6cV}qo{
zy16H1785<fO1E;8!eSDO-z<@6ZTLVhluFJQZc)6jyngRyEqOU~b#of-1xzwvAxehc
zEzrx?F!cAUU7MF@QeF{V7~VXYibY6;K}RFPvYO8Y2&xrP2HDGV;$-~F!Jsh++E}+$
z{;TLDKeyF`iFTlQh2ECZjAl*rgdzqZh>y7T;fSC?L5`>NV60HYCte?Q^inH`81J=h
z7<?$l;qfZAVQ$P;UI=SQs1cuwG?aDq#)P_i7`0I2V8Nd}*xaj4{-d{q3EP}-7U4Fu
z)^}63Ww0F6;gNn^9Amyxrhi~fQQH$H7M{?Wa|T_b-H7U~Ww0REeYh^SdiAAg@VI;A
z5ty<9mMN4iyJ+n`hzwkbdepIOx-#E)x5<an`_3GYyv>SGTJ!mkwwc?~KbW#ebVUWN
zIockKPDL6k&DY08O!yH`v@5hma=lQBVwC4yA~N9Vo5&?9s(EY*5MxdbmKO?RwKM$h
z-C>*Z0>VSeEF0eAurH#5wb_yU81_zfG*(rfA&Yh_e(C%mCH96()5!cz45p}shmfE-
z@dlk740MVFW@v~!e>NS94ka*?;1;DXYHd!?Y_;g;i-eLW7&ezcVb)G6A`E4635e&0
zayyzR`I@|bB{yA2(3-|=aj-mdLj!gxp(#eO_~hG?*ZcOLLmpqwVTu-=g}D-UMmd8~
zzCfNy$t}nCtFcEe;w~yk?x09MrQSJh*AhS?x)0j*=qJzOqNt?uEZ)e|E>E(X)fH9X
z_u?msLyj_4-|}+=Mj0;tbhi^y*@BB?z2l4eXyec9Zmfo1aW@jYmn#HXRQ_N7wfloX
z$UHtGF6cp3<$M@apE29(IHSPI@vcHu*^0dd(`c93j+J`@GwAX_o2=R!L=kQ-7|{^e
ze{YPX#73jG##`&cz_PFPUuw&lSeJb%(g%8m^#OxtnN&7c3BJX3ZSYPX5B8;DuOmrO
zan2qr*v2S>h8L1t`UE1Wv8qS@JkZ?*kvdT()~?9V5k_uyK)Qx{Sz`{vmc0el@CFhC
zH-YRXg#>R{)82wex_c43#N4EzB@Dw}$A@AHAd(7cDN9f}Yf@sZxiC07aW~rXCpt{f
zUwia*7OLz$>&QXT8bfJVrw`iP3tn;9U}<qG*c%8G*53wW{37Y#Y!U*bj7+w~vW6o+
z++lFddA%>NA$1>nX&P-AUe~ceNU-%FL-M_Elese(a40G@@(#ZPHkEB3Cvv_W&RHO>
z#C0;i{Sr$RJGwWBQppD8jY(8FpVqD#p<Aj^Ngq{SM2u(lmdm_zhn@7V1!eciLk9nq
z>Fj;%b<p!nt{7rU<!GCY{-idR6i=sUE@q+YA~RU>6vUT#2V{7t%g$xPB5c>4e8(32
z^P`QMY%xC5M|mv?KWUd@)(SuIl`TaLNZukFA_Rn8_L)4j!>;_Gy$FB2CHIDu=oz`@
zuryLrUGu$_hP!Up6DZd83+|F^+iy5fO-SZ$27D+kZ#gXTA;xg(cHYiyJUjC2xs!`l
z7qYsWXRyX|an|ltqj-{<Zz#K;M@&x14`P`J|7Jt6n-XAin}K-6P`Ne49cZ?dL&*(&
zBy=iz&sGhUn==4htA`xh=zd73C!K?vI;|Nh*rrfEAKTjag%}70^17jdS}?DZYrU->
zN;0lSFgyge)tE9*EFOT?FcddrC)3?njVg()URw#J6e_hb$01JKZ){W4bIGCf6~%`Z
zDsFQ%1Zak~IAz!thiYkh8-|Lc9*Zh$EjH&O;4WH|cfT~fzn3?xIcpRE#3d0{&O1@`
z6LD*;xw^`w#T!deTY#uERQMtJTBF*Yw~p)&a#P;@(y?~>TK_Qt&T3v287B5F{x_PH
zx(#}j-l1_Dsd{i#XECbt&p#zS(?l+`)Ky4%CUeH5e}a{m%GQvqx>oW|J3Qp?@LoIO
zf9>3EN28uUp)$&ww`18BXI_|5{cFd4IFF{xKA))CNt#kr;C6B-k!|tXf?&D8G5WOF
zX~?2sI?meOE}O}<9Fbb6*hJ6f1mxvPbnN`OUp81aALB{Ntc_Q|w{yO5B_YznP&|2<
zZ1wBlFXkkmZ9W1F<(axP^!-m|Ayv#e|406{rR=RdWS-&ZvTuH0o#^Pjc4a6Ct$r9%
zcxM<RU*yS2yB5bp913Q2RbL-UqzW!PZ#Qx`VY_=c6F2=gxPO7;G-J1-jckt<;9BbH
z+e6>m@BGiq=6_`7|9{>8udjgJ37nyn+J4ZMrVmPoeJ>7;$X>{f@8nu_KWmUf6CaQ`
zm>9Rf2-<Mq<(}bwTQMBG?AQS)p_Rks83CE@vQ@+Wj1-?y2;|=&@mM_^ZwF~Uq!(3!
z9(_M$YqHZil9#p7RN`<%?1&`J+PdNLVo3f3U4?<R56s0tykZ-CVB|{8)-W7YERql?
z)s0b6**h;<V}2a#8AepUDL#sHJ7b%3+(KZ<z5|P<C@XSTw8_HwTeA!nx3K53GZ-!}
z<oGzziQ}_dZ#ezZ;^UTpdMr+>*FFdaYm3?-E7*ymYqj>A9B=Gt>+oN@hKF~-0d)FQ
zJ09tH#A-P=0D0f~z0+n^XRv}$SydI-;XNCP#V-JXi;056+Df5c&usz!qnP0Yr03a$
z55Vi!HHx8R^jZ?D&;wMsP5HjylZAYSBXBrClAvMHj%52H3_UuWN{5l&t`acdSk#s{
zdBKin?LEpIbYi%?LoL6ulUXja0-Mg|6=BHSn2F7J-KU?l+59S_DQW*93g#-fWil`I
z&ncVt7UTVPE@uFrElv1>A4T{ah_1y1d3#Cy!G2xJqxQ!Qwv=B73iSs@z8pg#rIDpC
z&l<Lo!yzeL7Oj7#=)j7!&8|8q&o<e$Xsw4!2G~;iUQh5O#^eRLF&q@9MA_qe-Msx~
z4(p}{T(<Y@RyO3_mx<Z!Y88qLHstG<cE^uQ<D-w+-MqW_fZX#Xfb)jtx$mn#fMW!|
z9z-vm{soUDWY#uP?j{EQ(4wC~O^!Grcm|pPJIIlT9G@OCxsb0KiIYImLPZNB=Zd#M
z<SDV<nvuXU&te$6HoBAk>OET*zhtJhVC$>S#U^@)uni*x3t<#M2+{a1Pd@4K6=8Ja
zh+`p_1eDyMTm+>yX!KzffVJB;B@D37wmE<F5sYNZNMM+;5(#i?_E!!h#$JWg0lQgK
zjPBAu><U0K%_9X-48b(FjD%<j+EM<}0}%zR)dwSlFwv0Pq7Oe&l1QGnb_bwjMj$a!
z$4F2$za+$kavyXCL_(lC@c4T%LwV2Cd~F8$*ya9eT|b9D9;{$>zY>w6Bhin}qr5`p
zrBlRbR<GJBAIU_nP75NYNgo2>>Nb^~77zWxmfB%ofRS3C9f{8p28is(<HL?-8H~qi
zuD4R~9LohC7#bShXUC(;8LYlU$_s~`h%yhQqkpk03Lv3Z5kgL94f3mt78R{IQ`I;J
z{+X?YzJNfQkhHJ}bFl<r#8c>WMg-%}kHkeBB@79iez8?K=N*#fSv>VZHg<-zXpCK^
zyy*3I5?UL<%OzjGp4u?mvgA-GECLT55(YN(a;zCo$CUt5mPREkDah<<u0K?zUGpE<
zS8`OZ2L#A8U}*|P$kb&HeZwJ8@HE8pH?!B`p&+BF({?Ls{8B-~Z~L}l)-_~z{K7vz
zs3-@!`t5Fv;-XdhM;9XZ`1kEz^vmhh(*j1r(9w4jZ8nN$sEerL!VlRtBhq@kV2~Q#
z)NqF@UN6{SM=^6*`FgO9`xPdXNIw2?X?h<4Rad`W-i)WW*_zkmP9S&UO4Mxa>yCz8
zi7WX&j$g0mMebL+)b;rj^Iv-2@OrtakN))M*6@0wjE7+vWmI?GHaeuz-hv$)vtB_3
zC9@8bR)+UYF}#)q%ML_5pV{Ud4@Mjuv*q=Ix8#IoTYVyrzi3T90J!ZO2evt<B_~XY
z6kEIzo{6a7d@0Z_69}#?JC~gw`E6r-*Y-JQ)R9ZVwn*S4cgxOd1`DWWe6gH_sl^v@
ze;*Zzrx2=yGySjUyFOk+NRAY4u<994UJ*4I9*!jFU3<lCV_D>&kM)k@laU{H2MU|?
z#ed|HjW(4#j^<VdOosyyH_Aot$$~l(IO1!?Ss2(iY)8EdTG_E`R-ze-3MjbU<F5xd
zst2HOj2@BlJm^S{+b61Zlie=6hfL|DcRbX)%cfrsB+xrSzz%t<z*Nro#N-uhv(dhU
z0HDtcti)XQi4mPb{YcI`?`6Dy*j3AR9^&Ji_tS;xku6k{{s?AI)ab`W4Dzlmz8>P7
zS{EU2N><oQxtiNBsFXY7QtnV2$uCNh@XNlap*<xqaliyB#t=o1;=6X$=ZPDlUGqO5
z--#`epE90E&gl&=AK1IgZsvj8UW6&PDzKC?O~DiGc7oM6u&X!++=)RYOZu2eUF3bg
zTLFR_uBOQ1+>1LPtCSDyzAwvAFUBqpa>tn${-mQxYJ6Uf-B)0s6{CK2-Cn0)Hm0W_
zsVhfg??}L*NZ)1eR*e=6vbV)nM`sdg>vo6%Uo+~bMGqFPGcu9^og!!jc%)XgHrsGc
zls;A`XkE04$5^zPYO4(x<V#y0ZAvm>+FXQ0*${0cZ}G58EDuOo2p@c64G!h2Kemnj
zSKE`;IGSWnvuzp;BOx@^GJEbDS8$rm)$+)KLfIAHu`RJYexmd^$x1+SpsijTo=SXF
zojh1b(0#O@{MeeaFHY~f;4>WOtR-8NAjtk8wnqO>LeM|+b`qj(wBUF*ue0{if^5NK
z)=_o+A+kOs);!*wqvZ*}43?C8wW`95j)>ANT7Q(uQbRTiIH22Lu1F7Zve7{Gy?eWP
z4eD8)ZJRHY%_lbD05M@SI5z3O3Jt$tQ`ISYm~Yh?iNm9T{E+61_Xs5)@xEo*T_>DM
z(Y+k4+7qmSO_Z>%AW}Q#6K}-C>3CF1*H`EoVI*jc6WJy9QpIKyEORnv_e`rz$8^Hz
z4MO=Hn~5?BOELb2t3z$)`}Y?OHs{mhZp!juel*mA#0Vv>1utfaW^^uB<lDc&N^-#=
zedd{GY|;Ph>h88n{^yghFH2sXXxwtixg6sw=+Fn{tY5GzzKu)MPw?Dc^$9$S542tL
zUUu~Ewd-Eo&uectvi{h}sNM8BkN)g2yXAGhJ;5&D9u54rkIZ0xiSJZX*EN+r`#W|w
znj9lw!aeWl@BaA5c0aaR=kqIjkf#K@RBSwL%<+jXoutmtzqTUET#c!?n<Pvai>uds
zz*da~eIg46kofAcU}d(0#n<FVg+Bl<udN2qjr6@<Hx?q5;XTmy>&FT{YCdQivMy#D
z9|dk{7z?_j&>QJhhPRz<{J|G@t>?jnA8X8hTu52OBAcR2-t!-FFy9BqY>qPUgzhZr
zqjR<;d#w_>$reOW_qXOW$B6&l6xE8ilWs6bMMS^_YtBv!E$n1^OVk!gu2&$k2r1l7
zF2$Gx0&V__3;zWs#_eN4;3<r^u3g8N<L2dfwwODJR>3K#GuIyLo*uk(YGZ*BkZ~c0
z?a!aTl%(7SV=(!2T(ps~IAgL`^t{jYn28RKoQBTB?@c7U*eW1vf4HciNY^Az<pTPK
zfd}LoC0onGQ5!C|cq=Kk){(J-6q~p#N2?`OW@?!XDKZrVwH*scrS6-x<FN|<LP-av
zus;uN9Zn>OU1Qh|o~(MUJid7xr9+tZBR#MSH;$QHnx0`|-`i}S_FoR$od3;@-)HmH
zid-xQRDn3o<p3^UXXi2;*FtW`b}YJyhrAfA-C6YXcT<S&FO4N(Nx*^q%5=2k*I{gz
zUG|YNLm#y(u>^W>aam9J9@}Ksd<7xs?RpNNP$jG&v4ZSIt{v~}g5C6nL&IR#TRCT_
zJfgea_I?J&y6jGV=_AY~ITd1tg`te4)<r}omr!BpLSUWT-$FX*q5Z}lR4XQVPjPfL
ztH<?6s{e_ts5=J22M{AYe3Mwyl~JZ`<tY4?q%^jwUQWtzHwwCXIoHKP28X3JRS$%-
z;>aij=-McAk7rkOt&1{oMj|j3XnK8B_KkFXXcgvdLp`oS6OP*W(4jYLsH&Bhrr?8(
z)ev$I(i=2qja9u*$`R+_+omY<$rM~;n|%gLN{vF!j4kzoL5p*pZH@X`ZbZo319WtC
z$AZuxD8`rlYt6na&vJoU>M3#>#~46Nb!)w#)}|+{tzPi=HcpwfM`z3T({E_ULr{y2
zoZQwsAC?ImsMRm6mLT$UpJUZzJ=I^2SD~cH#o;1%xm*1INVF*I6us#5)T7J;fr+Pn
z*CqnCg2;W3IC1ihm|O<O9Ahdp^HeS)VG+<m9<Dl6J`-p0kL032k@pv?>Z875dq9E5
z97fK}Iqu@|df;0EGNU!dk=++<Y5K8W*vV=-^5B{Fg*`Z6GdU*o*gOEs(PlQvB*2q?
z8LvK@i=m{r`qbvD8j?|~?5v%OGR}!oII=`%Y{AQxrrp%92w)cdu#g>v_~w@$HsYof
zb5~o6GIwiLe6kC6IhR2)RY)S(p{ivez(J$L^@`Q^UGkullIeYJ*L=}hcl^q(=Q5%G
zCAtJ0QtrYI2rI0I^7?EW9yx&7G>$vBVl1KTZ(=`hSEtYetE1_TFL$E0%M=xGo0JSc
zgskKrm!RPTyXOllb`l2QzHb@c%$w{%0*Ye0I2PD=ARa#d+*bH6v9uh8mE(?v)T#20
z3AL{pj{`A`gN&@*S9=rT39(Ai#A#jm4xF~N<3T5V`@h?|@wgF?`~x`dug`D9Jl`<x
z?92bTJW7Q%WRJyjOinCB#;D3>{8>aK(>Px6MafoUn-WU7$%#zZ?Bk3>oZFW1f;CCM
zU|Yw78bBaN6q=^|NqVJG%!OiVNPj?qE&fY^u^{kTb0S^!;`iJZAf*|P%yill3QR(X
z_D*b_tOQ$$X9ei-qIG(=*z(BaqE^l9qosEtzdmEh`=iWrWeeLZ+Mq*P(B8!#*~ob2
zp02n>e^|uTp^|WkO}&DrU!IzSHc{;zc8oTe&_PllmxWe2Wm5?cOYm8Ab2wLNfV@&L
zJ~@FQr9L`39%_$baHps<V1VB$@$*^8fQxo4HhEeOTrw9glFsAV2RV^nw-Y&lJVP;r
z9<M~MXhXQ~C2pW^Hz85r>t?b!fki*CS>IpM-q>8$lFjXWH5aj1P8OewG6he+2&z&%
zvjuM?3!7eBbXdw|VWdk@--#=iC@ekfgVR{@f7s<b9)mcLjt$zCs*W5@-J0@hZbwg-
z8wy?<|NiIyQzi&fL)&!+yPwYLb|dfAV?VQ-xjKU=0PL3k63qhfb9>w$Q3a|%yoXly
zV}0T9-tiuWkRjOJoDJEe2YUD0z44$1Ug)hsHrL=-?#BdWAcbUrAdAE#%}JO1c|~Mr
zwy!+DEjw(*zQ82ANQEwYvz7ZC-MECW*yGu%effTqZn?-0S7$9~6-W%=MvuI&+-L+4
zwsxOS=!cSfsd?7zE0|7dF&rG%`!66R!i{b4UqZ|fmJR+3pGu*QjZsrl(R0?AP!IKv
z<IqChsC@;6bODMSr8YY#ue|yb+u~DwebBZB%*q!jK?`<l)>M_vREpQ;eSs;Ve_G<Z
zGiSwG+gf9C{$><GQ{2h6eFgjNFCmTFqiLb(xp(IvDLbNRk;>9&zP!v*4t4HJB1JK8
z_F63g6O|=G29w%;pUU>_Ht5rRf&e+=?AeHac;-o~=faTLqD$T)6V(Jng6dq5Pus!7
zyqOA9KHKMI+jO|93=dFvBV4$08=jTH!qEg1<L53S&mHsD1472f_r-ZE9w&IV;TRh8
zU9JrCkO?Op<eqMu_Ad6zL~14g5K=BO*6)5}vk4S*jB=xtY%dN5#<Q}?!Z$Qd$7yug
z-`F|deP|_HNa!DzCHP`a{0VrHOa8;X<(G2ZU@}RZhSn!W72%>dN|GjCNl4IreZ=x8
zQs&hd`7E-pdAsI){;E9c*JEuE4BL$g!1tk2Z)QjEpxDwCJJ(xLChau3H-x~a?6#Mc
znEF^S4b`C|?&J=T08G}vclUk2>3<Sk*}d#*v{x3Cc0XX4l^=FFEF`=2DQ#<e2I9c>
zm)pyykGRV#_B*O5!Z<+w8tavnQRW(@BKWmc`%?@<SjE*|fjJ@9>~`beHTyj-)(>co
z<yJkmHoMq~<;c4I<${uPwb9lm46sf@o!XNP2_-0W)*AN5%^DjaIw=`pqJ`j<G#I;Q
zYusPS7>cnQr+HIUD-rGDes9h)IAk|*TeK}%CN4qU@w{%`Uohdoo&jsxAE*<QCM5;*
z%znp>aQ%oQwNAjTSV`6rUtEqVW>Am4f~vIojK<vot1X6<-1vQZP=1*9{ej1ctU~NR
zs2qo(n{Z2=-Wu6r|5%^Za&>rYzq0-qGbttl{4)t*q{026oe^UNUj4rrM(=kPZkyE;
z?uYDanb*8c_*hTqc#%dX;1-v7@mmNen6s&z3pOK?3~2cdM}K+(#kmk~p8g0%X-D=4
zrARU*@ZQn<DO?28<U0-X{tf@TV8^|0c-aBW3J3~<Y~h5T9ik!cvXi;%P{9#HemcOB
zL%lpBMTyMhC}VpEP`k|Ljv@~4CiDSOFMM3NJQY4@KKrfVSqcpcB0T4d_2)0!LX1$<
zQCu4FmWxqFOdSx5<P+!ZQfw3Mt;pb)vK6L7Qlc&Lm|U(7_BcwEx?S<9DFC=#^}io~
zqUA}uR?SJ)5n9gA?0Oyo+zh|48`+cmrkUc+oT8-Y1bcv9@%qT;(P7!`I2vx1-~!HN
z5lQS$j3{Ns=X$(Y#k*0K+xZQ4miK&Bo*cCM)mMd)7T+>&4-zU}Nw&)-3i9cy6<%V+
zgd-oh!zmM#Oj}k?6vQCyw5{^mFr&jrxK>AP1*R!MUlCrfnFu09$DJ{W48y+&_H1o@
zah>`y^^C3a5n3c;TpxY9Z?`gcU<@}A2eeKU;xTK8Dq>Y7t(C>j#vHb|^k_LUgiE7$
z{<HwzG*M6p`efUjos<Nrr}1Wd*S1viImey~M}2BrvmSHA7XVK5Gi#a%)Qh!)HT#nA
z=^JoLOTtm~VG=Yb@5U!AV{0{i>2pdF2@a)nA#K?MGP_T$JqFY_wN*NkjwmDX3fm&F
z{a4nRThZTD8?u^%M4%b;$^9{u$MA||0?1@Px50@DCFAx`LI;gxKV#$Fb*txek(f(<
z{}otpqAJs7I=?mPTLO<$5Wv(#AmrX5{H<i2a5xtN8yF!!Pm86=9EnY+57^O(%3cUd
zF!?dl9g>}66Nz`_%*fq1UM<R31-;AG{X{IABIuI?B5L^U;JpX+Pe&OKW>HHM!O&)6
zbl8R2_$=#uoZMK|i#9h=F=TG?Qj|%)xKJ_BFy|%;hJNZ-wooneXFOcGUSg4CKd|U^
z2mx`a8snqlrh@;mE#>hU+jGD!`|(u#IJ@G_4i4MZTmW_n-*wbUF<GIuZ|wR+aH8nI
zLSMm&BAaB4f1?^z%(%24Gj`Jl>Q(;3TL}S%6?Qv!3bB+pBn3CqV$9o}+|>xIg8RC6
z!b*(Ty*xM|@I7`v*P39&dSI1v^?}#kt?(swATY~!Z^41<5jtfn4#d5F8t-_FZ4boZ
z)XO3<1-hpLajrZ=GS9;5><5ZY;!<V7U30(*C_Sg|+S&tUgC2R+)@3<oaNgE?%O$y5
z_JDwe{We71KedJfK|mBeh8)Een+hj++Zeqlpxigs7~>1GsRm=(rYvLa#C;2IY(5Z2
z7(1$ACEea~ASos?oHLQF2MR7a^rH@IiuPI+$-ggxEt@NNU^nR#B&X}h#cw%K@G<uB
z?^|n*0)0|m-?j=A(HCKSdmzrV95|%P6v5UJ73E{deit!iXExT`Qub)hY6*7t0GW}&
zEdf_xcFxIpc0s9s2jh!ofy9<eg((Ps&)7)zp>spb^m^_p@k49fikmm#`|x~|O$H2N
zRzygG@N6n}Kr;OEb~xrg&1-Wc=Z{4OI(>q#`>H<L_zOFhGeldVx9lsJz2nskKZPGC
z<g}eg=)gZ?V9%4@|I#$hDARsIi<X|dKa*7ve3m3K(VZ?$iZka9#N*Hcr#oMb2YtYc
zBBO$Hc`CWE=WQWscrtnq@wS-5z@6aHYL}u8?lf?aE+<@aD8b=fcDY)4$#$2E#jaGn
zOXcZ|>Q!$E%}>Ud9EsFvY=&J=;Jfg!y5Vd2v&S0krti<w<95qiV;T+Sh~5d?Q}$W`
zLGR@3c)rfsUH|oj$LGG6qw(}8BFFt@7ap6813ViKVk)lflXO;=D!71|9bA4vMb636
zh}b~3;$XS{SQ-uEShdoBab*@H(>>^D2f|R9#pvmrt&SQfu`thE6E!%*Y?1T2YY&zy
z*3*l8dtLSbayh3P#VfJ?V7a^a@Ysg<EODT8qC_CDhJ)n@jF%vU8?!<YV*C5h7!YtM
zIFy?X1{!SX-HVUV=9s8FpS{NnP`btY<*i_=wkF(M!3Z^PO%6BGondQU2DkVbCDm=o
zGA`Hq#T4~RYt2qux9zaD01U<(9MtaBQAdQE8)&Sf>f)o~Q3zXQoz;?~6tS8E(K-w@
z*B^6n?=^yq7uG}30tXKUvX@%V6ZVcA43xQDAvhST?)|c{;{>psh><;MuTy7}-UK^|
z;a#Q4lsDOl!86H>4@VP6Ngh)9NOrnSnw_J*laD_7XLjrd2dvn#ga-3?E)mHO?SxMc
zGm{ZJ8U2@RE}S?}o=tm?n5o&!!*yYsXMfPY-%??8GT}W0vzYVaT4wX!wzYRyUef0r
z0uU7Py@fn43W(Zoi(Yf@a0v}|>EJ^l_2(RC+2s_#M9-qJSaR@2$ozFVz&iua`iWhs
zfZ(Krd&||FALr#0oqd$i*Blfiv0$d_0T=ed`|L*aCmQ9V-K>7ifT(YIPck9u$11!K
zjklwB7($WTg<HQ<!8DJ!`>y|j*X$kJWB00G^6kE#u<>+>*Iz_M^PnnIC@Xiwu{V;X
znJ->_I-)DyaDqW9r{CDhsz#)|!H-~7lvP%aaK5YG2x8;~QUMHf?IoM$8}YL8u9hUO
zB3{>e=kQ3VsBB&S8W<wC<N6#b7F6%~hA1Qaj);3{D6F9x{sK(B%iTA=QRtc<O<H5s
zLWHZxyLsF6MnPF){kR)#e#6ldNi(**;aD^ADNJyA_CB<&)ewc~K+5o?HAR^p8`o1d
zzY(`m?i$i0wM3T%mC>0@F+5ypYczKnTpIpaYje1krvL2Eto;o?o68Maim?sbM>Z$x
z$a%pn$P7-Z)6O?s{UePv=sAoHCB%_{><iBJ87Pcez$D$Y%zYH=1e^nf)Sv7Bl*9~!
z`3n=!2^+~@`v@Fi_57v8h&E9zPf-vSUU2e_z*z%ByA{d^vUVzqBCZ=)>+l;v=%KHZ
zHuOlqR^A4nn&*x?<#E)<P`U+$sVE5S*c&N~A%%L~j^~mV)G1c(8#|GJZX<-Aos8lB
z+Et4-o%@9~DdfYL%|sW_4aNUokLm0W_OLjHamsjZZ7yoN)I_p~Pi#JGFB7h@?_BG1
zQCo7C4|GnU0~h@CJ-K72EheN`Rlp17T5+4@kOCFDlxsyg1sTR)_S5q1G>}}0zJ=+f
zSNa;Ha@EV0rg!v?jFSUX$w#hbz4BfPuCMHR1sBOpz<dNqu^TZ9$y5}bNahwUHoqC8
z%1r21E?-xTOdC)nx1*~w96eWWeQJ07D*nk|*xjtYXP}Pn%#huS>UifB131N<zh9Nf
z$XT}o-YCjM$PicmJGtT6WLz~-vzR{Hipe-iEEvhAZRMojW?eejOeqha@FsN5WYR!G
zAb{1Aj`P&H(7ys7uZc4KdWY#NH2K;nlXvll)RVi)x*W5AK(_C;ezG7I)Q^POnJkCX
z31tK^55Zv=s9Qq4xZV23$%27MMq!N=EDF``4ilSlT-hK~CNsp%lR-D{<63Q*jGKmK
zTahrpFHHva9hKTqk57}&`Pp#{V3EW4FP2UzuUfLVC&9X{%RrDViQd6BuMU=(fuU}X
z8iilc95_QAK8Ie?YIep%9pyR75{@L4S9GJvzhnLWi|m1l2{0I)YQ;T>&bu}eW!{tP
zjq4RCGEthCENK2OH`?S6>O#Fh>rO@4dk_xi+57XL$io30vS0M!E(g}X7wm{np$p@{
zqtVa}wG|T<hxgoSh=0jCkNct_(`5xuR1JzrUs3Fx%rY*S7+r!E*mSNmsw|s{3hoW)
z8#^0iT6YB}M!sNkxxb{KMb{*^*7?bJOCk2K`FGofb3TC`z4!~{I6KjhE&8EEYf9>=
zJQ&LRg`qE4rCLx#GIrVfc#=6?N%+M#a1j-k<!Y31zl&9`(Bn04^5`!b?RvD{vBPfo
z>BkMP2!d`#L-FQZL$+tmZdG&rO2h`-z*)QP&`2=|#=A2a<_yK*L4m!MRpaaLe#5D&
zUNyU$I|H7SbMU<%96$&xHrgX*-B0j25K?JM?)4yBBY778Xouoz5o^{~_z#6=++`~d
zIYLp&WkvmZ7*%`Kp(LAx+4Zh|W2+C9E4U4cos?V(!?q@B>0$a>1Q!Ql?I9<wz^0IJ
z>!OB4vRDLiWIng`QA3uY=WIju3*#tjI21ULfI~F}QQdebaG)3k^e&D0yH9R6D5(z>
zRMXEYZBA&)S}1$KDcf?Wpe)gAh!<^ZwkD#HbfZnnrYYf<xTNMo1>-4sm9=<tgaN(#
zNdeG$$SD}TGN+5n&?0N2fcj3OwPkgb6s&c-Ppu2Osh`;D{>#nHtTTay%+W(6fm<z`
zkB#rb`KsU7TN0~r;^whmgWd%Mu`D4*V&%?Ypg3c7pPFPWMSVNr3nM;gGAn}NF?`86
z^P;`WrXe@9!%>5DK0K924iy~qt7bdu5NXS`e8=Lvl$iGgXsN#?{W<Q(YFmG=Je^|%
z4itJgCw5L&WeRoT(XyuVU_vU}O#T?-)tyZk2}UL{5SPQ|a_=zk!couW4?UD?A&+oL
z^to*K+w+)Yf~wJoEF4O@4HFyXcE$E3s)-CZ%OmhH*XEM<kG+%!uU~S2P;&7MUQY0l
z_?0&tj_ykC#$E-mxmukKr4;eFQPN*%vU+T<`F<Qaw9&3dXD*u8Ykt;l_-OLG!+SA$
zmhSfduv`AA{{V5w>~>BNd4jwU(1F;UC=<!7m*zvc9aV=13(W&MJoj=6D7qAj`5m6Y
z``KH0i6NHQ{Ra+c53Y>%W|9?X!)(Qy@dT3-)osRCzFDrynCwhfz3I5bO`;-fGm)p}
zZS|W4|Bx;Z-SV0@<AN)ZnJl%|M!RxMI4oh>jmml5n~sD0u>^x)$+bR5am^CfMg)Yo
z=ndI1fl>B=_^ctCYZKTqM7kGkV@_xUPjhR`8m@;Gb77m_416OC8BpBjsKbj}1lJ<y
zwq#jREO;i6t-cgw!i6$PAdA_7tMc3!x;cSFZ(h<Pw^UszEi86$NC;cA5++HC(PwRM
z792M~sEhWfU$m&4vNMPrI^HbT_30hf>0jX}hS9!4ezEPfXyK}HwfMmLvy5M465-Nc
zw80#;grO69@wMWlAP%y+Z{pH)H#GM|v`?D`q7vDpF94_gJ)5d34-DJkH$%%Kn>RQ}
zWY8rU&5_)T9m7Z?jz*UPQ05zoX>crSv`9a2+;`^Nzu}?#$B3s+RI99POw!nku~NRE
zlfEE35#Ux6#k`f}6^k>Iy><^^u`(MqLfppa=WQ;_3X=eJ%ricp@F_?w3opU+xrBgo
zfEb)(s4=7gT6i-b3dIT115faTWf3YCeX*9NJKnHM@%LXZ9krzx#jk<bs#&}Iw?>v7
z!IeCvT|)=<W21Jp8gKc-^Z|&7dk%WO<{Lt$cm%uGV`F7Lb-`|U_wt)YGc8dZad=`#
z9MR;l2G0B1ZoL`kmQX#z7&y8NC%V$ZK4W)$6tuLwUUz*Ho*)vbS0XgOv}yNqIv`z9
z7Lv<9NX&w!$#t0a>w;POg;`eo+A)g@F|-}hIh+EfmA`fjA@Qza(k|liv?|I<1O)0y
zPrmxsg$lqdw@?9VeqHX<qm4hewZAU731KDV>$(`Ixbd)`(1Nj_>m3{z&o<le>%b5-
z85+X|{W>Ki{avt(qqgza<v#B3vPK`MiACS^>p-PD<;=V}0f%H-1V~a^+LkEugh)E8
z3$``NWRCfzUQ!TUlTT)8`d58&gl<l7`Y`J}Pjt~}BAfYd*#1v=+FG;8zFw?#t?k!-
zI$fONqDcRn`+@MW-Df+%qiP-5z2flFwzKM9o>eZmsre!@{vuS1PZT_W7TpG`B~`S(
z*un2`;;|9$2<wN@;XcJ0V*~2ZxhT<hSQaTRC%p4kr6;k;tRN#2y8B$nCd=3_-g1%S
zcKFxvRPlN{VQ|Ff!Fj6p3oRZE!2CD`8(!}2F&_}=9jx?mUkbTY*$H1o-WJTP#_VJ@
zu0q}G`TEeNy)BbMvpVB#$*Q-}X0tMg<D#Bza=*;^_mVJr-v5>fr=5!x<UF&23x1lO
z*lvr_NAYjrw4dR$Uy8Djt;5%SW=lC1yq4r-vWC6vuoGYf<;wxP5>>?Wf2~UhS$;Lj
zTtHGHnAtT4YiW90i(PlH<VgvNxZz(w<z03&YmK5Hie4nJK`6kRipiZ{ga4K6f_L(>
zL|M9G(h^0Pc3t90q8Zw~+(6{2&I{g;LtDOcHe`&T3-J|)P*JrXI1qiqcsJRU^B_|3
zoU`BYUrW>5CGuY}71x(@N3<FHUH^);rTo8FPC5JCV1H*Sg}XA#jJUF^rsC$XjRIe3
z0=T_Jvdp_>zc&@EmRu)R2Nc+hAe{dFDaTduz!e$8nyHZXb&Ws{*&jp|C>Guer_9$T
zxHSba9kM^1a@<9Pn|0Z`d0d_tf0WbYbv$M36OyssUG~QT$r+-LoVE=WB=d!Y{D?F2
zC)xPP8nM6|a_+J^9jK2NT>r?FqhTk<^g=1Bk_>(~X7!q~)?ksR3-%KS^zm)~&>C|<
zwnIpOe8u)t|NN^RO|~gM6Robql>Y>He(s;U4))sS0A4bU5ADyrGNz7^9ToG<EmJ|@
zqiYwaP?dj~^Vv0uRrl8DLQ!(h*c0COPd7DMQ&uC7(wIG!3yORRlB78%?Eap`YCE2T
zU-~I&+HNfYdz?hc!0g#-J|4pGE9TScV<I`Q9XSewmb^f{uMKb5^HWJ*_bR|Jm#n>l
zz+?A<5d(skrsC1rgTPPA-QnHr+-tA;UmIVuPXFuBpIUd$vt*{MIWVo@n!bb;$Q1wp
zLdE){tlS<Bxzwb)r^<U=?7|1D?;eL|24ZZeI&R~AHsbw1@<@Y?X8HCWY#I{^lK0hZ
zU!EEzb1KgD%y=U9hHSY9W0;W4@Q`C=ojs6ve2=}EJs=xE@JE;ldYH;8c%qHk+qq7?
z6Syez`EcH+E+>Zmkyqzbz6))5#D^ayYU^KTMJ(g+|FEOgN-67z1b(yjH@+$a3wz3r
z<+yz?%l1!mji7Q|NArapPvlYjhy6S)<D~7MC9EWyEFQoo@?2vbMW|Vwh<}lbg4kDN
zsFG)#%(1}t7eO0}2mdd<jBA2-!6*+F{(;lgX;NT3m-@Q|{8W4q5&f<(a3M;>*U$JW
zRCyiW%dwygC0d4MwApCwu_%gIkW(sB`~9db%g+zW6G_TN@9|rVN%`NReU$6WF&6bS
zA8iDMgxuKJ*r%~c5*y$;=c`WOo-RA@N4%-k76J;zygUzNpBLXVy(M4e&OD{~#fy3U
z`Z4Dkv<vz3SrU9<YmxJlCZAbc;069|HToAI)}`F@-Cg#5wW307Y1t?O{c^RUUqdG8
zF_d^oo{j%12Sm?AM1U**4~Js8*x+9$@Pmq_bTw)(i!VyIU!%%kxlwoho2l<NV7RVd
z@EszzL;#2k(18_y^w7NPT28eH9tmI1QvKW9$`UeqIY)Cn52qru4{#Fy-NSuCdRZ>1
z-N-$F_LPX>Vf6Ta@3WQ=Phn1O24<SH|B(G6)ltkayXEW4n=@$tF~NZmQ`nv2RV)pf
z-Hy=}3Q($E#XtE^d0nt2EQ~YaR=>|<7>)nu#6*3(7<YY;H~EkKm*@&1r^Kqf>~nU{
zzZVI>f6Wcs-gN+X<oi(_TTq=4tPj7o{}ymcb5P`<l_}nXY`h=i>bm`Rho^OH!0fF;
zX}q$*e&?;A0_q?&#4K+G4S~wie)p}U8pL91Me&!;GUy&bKw3ciJ8yme``Y-kFzXk}
zzv7XU<Z9Jh->={c8TUr*_ug{yWx0(pHgu4$PROs$q1(vR^!xsU7nSE_&0B@+cyh;+
z_6O04PZr@LoQcHWUqS6nL65KXs)SCmKdk!eC3Q@IEjt1hiH*Bxe-vP|wW8-;4g-4p
z#MXN+kY)DAZw2cF>50WnG!eCUih(^2o|x`Gc`M#T!u1Z+*yV<03#EG?;`k9?4&K2=
z+Zf-SqLGmM1N%u1RvU+mX!P2o8rWk$je%2B@GuC(rW~gT^_}DXc~vICx8_F#3h(XP
zzu`gsb03F=Cqj*Fc`Jz012E699e){L&k$I2&bDS>ASH41L^QG|{P?`kY)$zUlK2As
zy#pS2D*Nlh#1kcSbJQu0)Ig)mJip9+EL5y!rHfRoNQtZQv9;u}#NC0=%Fn)45m~+X
zd=B*-u-2RdVyjPVN9-7x8QXH-hdBh#zvX0&G}Pt4)}FJ2m7lkl0yc>wuwZ3d-?0u~
zDMSWi_G-c*!NZVs`Y;kPbVom&M9_5Yt?&PUgYJ6prN0&I*!rr&;n}LJzgjlA^@$TX
zW4m)Gi32bfsE*ph$j;?ivLRo6uJK5oiefk#yZ~&;7dBcA@QL^X*7sAZ=OPyTR5bqk
zsss5mo2VxAF&xc7|Ftyz=$~zFw8?4<i9gOp$u|GZoK)A~uHJ60%2Yz59W3_|^-|$#
z->#s+XRe%pIXfI>e)}}DQ}&OdEJG=v8MY&_gVW9AK^{;T(Z8;G6T&O%K0E50{`$D$
zLjO%(oR?p*V^x<5FD3J!ckQ1hpk>*X<zBPnIVbUF130AovuIwlos6Z}YbSF2Kmr6>
zv47#?W8n%Ze=_=a&DL4m$};!Gzs#mQh!?<-pK=+eqs*yz>9Nk(DKF#d<JB%FBM{1r
z5An>9QbfF$dkIHPXhLDh+2}{!3(^vN0JZXd?kOV8&ulJGz%+{6kNj_pA+Z##PTu$^
zMns=x-(7vsL?wX7xvT-*N`SfZxxhnYOsl4Z;<~(Gv-VYDfRdMQ%oejp0vqE=C#~Xz
zTxoG^62)JNMN`sBkt@m#@8$}>jS)Ql3@j{|aw$i^wQ{&;)V|N-S~P2ytG+6=CXk7w
z{8#Zs8VGr5xWA^Hg093UIkMs;{Of9Nl4h3JWOlXMT=Z4oL!i;D!6pADYP&hDA~Kw^
zYq8|V!2bU>H-nevyj_nkvtW;Zm$M;qsj$wC9Jfey9rAylGbF$Z;hS<ph$6$&{vY0Y
zn2E;Y!@-F{&SXERLnW}#wf&EY9`|A5b2|s#N!n^uQ~xO&K#NN;_zg;lJK10iw)8*e
zw(akd%}LGfIw1SODGDj`UmQU``smN?UiMw|UPJa@tA#7sh9L%J(kd&o`+1z68?*nG
zt+tba1VZ3J&dmK^4~9j?f6vYc3e4-O_`B?;!u;6V!N_8cHa=K`@c5m#%M&iaKi6c%
z+fLLd<dn!18S(#alnEDkJdCrp^6f$ik&lh#{yY8yQ53$=tKN2U$Jr-;yEp}$m*0Ea
zNt}Wpi~LEct7P9IdG3O(emjXRSy$Wd$FSw`=Xgmjw<fE=pmWpyAb+9}suAMq?LxOb
zJ!pUUcA=~Y7_!;cML)VA(xfZB%OClqQJE;1<@&ede2`0s6Q*E_G6?wNoYvz#&^FZ%
z41xc|2O>#r!`ngFVE&2OcHJI%+Z95?6Snc~pd1)PlD(hA%KU(p14|M{BEG<Q7O^0~
z^w}>FPCxbT4gk(3hZ!<L?l3>k85FVk=A2_OMsv%%KY#oC-_xUuJ0KS1a;H^{ur0na
zdgSacy$=#dio4qyeH0;3oV_P<)*}NW)>I9m7fjUntUcx3VUk5+&|C|Wjg6sgvtQ<6
zL$ix;moZx62)G$zp=^C;neI1`Q_tpjPhz0i>f=ca-HV?_38b?lV4$f21arPWu(n(Z
z!XAv-^9cnpUrQ>;_8fDW1KR(^+S`D7S#A5jgNTTT=yoF{BD!9W$L%q%x;<{U$1C}g
z+dOKbZg;!w_O@PK%a&`cvtE4LZt0krnVFfHc_UW7EMGEPnVBuk!qygvkcbG8h=_;?
ziHL}Zi2M1@F~*~vd%5=y=eF6he$O+XImaAx%sI#V&=q==k@pgna{py02{O=@Sd^HP
z@`8-J9+$B=hZ!@9Y8MaWFGb{At^X?CmTy^oZUG#TdKrbJWzN1?V>Le{t)^Uc&d1}-
zMKR#M>Px-5;>uBPEskS?xEriB29ox6zqNaJN{Lc$9XSuOXM4JkuG}87)pXkOz@2lF
zaAsVkE903Y(PMCU#OT^;9}j<}leQ)SZe*5gqwxXh7RrH1=ARscG<FS*wr*XpqNvF*
z9LQIE{kqWBlJJki+4t6lT*s0E!;PyO<C&ypIf<%F=cXK=8X~sYH$wwEa7#X4K^*_q
zY7$B}<g)xWM-+6UMC;qv6+`F9|MH*h@NT>y@6HOQmr&N|cz?@w`Sjd(YnVNs!*;PB
zUPy%3)RnBvl#$OT=qpNML))F`Y3%6cM%f$7w;#pyaqIN~ac5c8u8O<5`43dIQ=&9k
z=3X6=T!V*vY)m;Xw%TF;p}_kFJCgH9$dUF)d3yRBSASnL!j4w+I#i0=V|cS;ISh(j
z-3^#5in!Wq$Me)ucodi3iR>IUyhMCJcm27f7>E2^fdf0`Ahi)O+hC`?7uj^0OHD_g
zI7RQY!OkRBFe<_t5BN)Id7ziG4*BE-lkHqUse~IIvYLTi?<MM#&s-kAaAxlZq@@M{
z`$^K)`CPEcbMLVa;_Oo_(TCAkMkyU;@+<r(m#>T3%Bn?aA3H2Eyj}=c<W!^zl=$?q
zFiKi-F;5&%%u607|7z+z3+&UxvrfxG(@M$tsxrViFatf)?zPVo$H?jjO0PJqIZ}ui
zA^oy_nft01MuO~<WES|!M`<FX(ANn;2?qrImvRxix|gBdux|nk7pM23Q0;qHg(zOL
zZxcZcc=~-8a1?n1IF(85`wBKfM1hSAoAyH@%55(!+K;)SrLuwjlpk@iqYKr)_H%^_
zv7OFb`d<AdyTkLM37a|IEe|^zqGR6Cw$HA5_-xUm_}vkVC`5VMu70@Oxqto3U3SgG
zg_}WNp^)lpqZ`TnoUrh-`|P@Cu?J`4*X;U-FI$L<v+51;iF*mbEg&x5_^^8h23ojB
zxSp?2!|C~9yD7U;Zi%9pdAKkkOIFE#^l<P3b<G{BGB_xPDSoEX8?YyD_4arkOtRY^
z4o%Yx&b`~c@jwp_w$OXA$KK)780cA|{5pfP`MsnP9rAGSQLvzJze67mmT!JTT{p)s
z%9p@oiVRBg=oioM#BtGIAs83836eS4h#cW=x*VB<kCxxisE6as?{sS%n`KN)PwvY~
z%!3brjrHl_RoZFe5*_sowc@PDS6IiCfI$AAj5xvDBWHHl#Kb@&l^^6w%9%bXfe>qB
zcLn4JVCMG}GL=?RsUoK2Sc+pNS=6*@_Apowxyai6z@}GF94>6Ll2pvdXO0mBnxi%|
zdvr&Bo0ThHQXn(^IS;2YPoWV-)?pyoCB+Q%+1y`&(EAT%&U7uryr528B#=fHBxa{J
z$ukGbiY<(;B^&H561$Ee4BT}61{eB$xlUtQnfn8HH9i_6EJ|P-R(jT^#fiuIHrS^J
za=tnd0s3wHhia>5MISQ&Q*ZT-0|l-Y+fHLlNglB4fJIV_@9rkwogCA)G#ZtXnev8{
zlC;PrE%~Oz;<=&1CA-$dm6WNj#|&a^ufQr-rEW6oaH6ScB^_l~yeBkODK9yMPTBID
zeYrKbdsgNHQY)3(cdH$NurAQ$8t?Ai^{_%~6Hkav?e>swGTcWC1HLZS4W~LZfa_zC
zt~5_xsYu!Ia1m5yPqB^uUFgI&<<6ap4=i@An1GxB*cD>YY)cHZ2M*%2ZFSsI0Nl1k
zV;zhK$^uZ1)yG*d-*#lrbxoz<jrZdDmu**)8TKeNc}lzUyic6qpmFv>tbmsVksv14
zS?-%Xxe>VLC1~%?6_Hc}@w^40?5!pNv(0zX<o3ohxyS10c^MC-1NkkICmF-hKkl<b
z4h7bKvW{GSufExi$L&b0Ri}#9_FZmJoK22^pQE|DU2W*bXml*M2jYYs4_tB=aX+1?
zzK0!R09;3`Kc=n?)DJyN6)uJPVLMgrLwQA#{sC38e7t2_Q6BlzInb0g5?4pp;whc8
z?6ou1vwlqLMGa;H(W#_wdFRU5R}3hCql@UQ<AaEkPCFOC3sbM!d)^xk$&l?5xNI=?
zezZ{flwLdU0LcW3v8tMvW~6z!+P~jEOj?kyat+dbr+pNVD)2}vsF0nH6Lo0rPuPWM
zA=eK@;wZ!IMQ<^$p;Lu6Kgrp02fsldd!d3)^DFvzGtl-~JWEDJd-ikhsjZH}f|TJ7
z&|NK{BPN_p>`UJacwE?5)%>JhD8$473IhE)0GD?0q+N<n@X`1tK9O8>0(|!tevQy!
z`?i8pY9C-FAnv>BnG8L$#=T+R$1^3b-S4YsKlqAa(->nv#`~h<FY?}T`ziOON;<UJ
z&$*!xmX+1zm+F(Z7{<?2Tc36(@lpQlDt|W61Fq2*Vm)1DSFcZg$yv`s?$gKOU9;YG
z6j*+)U0-;VcX#smy7j@MU_y{{-}N~VZU*uV5jS|l`odqa+0lL4jnRVI5+uHH>&i{}
zRS9s_6@1ce_ED($d9U3P??v6hW<Rt!Xno;WRG~(W0Jm0<!(|mjraQ}R{;|j$*CUcG
zLgu$ehrY@sp4skjFcs8F!1dthuK=k2pn2zahO94a77PUXnW2G}=4KmKO+v=A!g1Pg
z2XZ0&iyY!c<PZy$oN%Q7D!`J8BHxitqRYnkuc>8rHg>)923I)TL7-y6kaCbX$E^>w
z&14e)xXTQSIo<&s=$T3BnF;HIX~lOy336>>jE-+EEXyRH$6tT*Z+2JAgT=!V>>U3T
zpNIQ(P-h`~u0z6~o`4fnw#R0y55}&oRw9RL9?#6yD^{$qSuvGd$`eIaE$1Nw@ruo<
z8gtRkP3j`gvKD#X`f>+Oqbywj>D@?rd(#$1*Brw|gDgV0!Zd34`5gb{Uw&ivSJWal
zprl~a)dMOQ<!ta@Xt2dGj0Vstu3R7w#4s2ziq8Z&jaUY<EeCznCSp1CUbT83bkyjv
z)|doRgaNnOHC2G%7*V<@mjNfp(xm@sQ?13(pNjp~gbK?BELzzP?dwBh5|Na}$Ezdh
zLJl7~noC)euJvc-Rq+38S^Kr6&6dYx@F&3dyfOwBclrjnRS~Y$Il(cuW_`Mb7e$V8
zrWP>P<}xBnENAyU6q9z}j#6Mgj`DTc**IHYja7<;%B#UPRL@>3*n%gtjRB!{LceW_
z7W+AI_kd5Tf7s@DCKI!`>9@J$xA;KKRF2x3(9LdG*<9CV+mZqr#GY(-dc<>%{ky~I
zuYRTNj6N0gw3mfP-D0~EIwBPC!bA)`WzSdb-r`arLj?E03(;jA9GvxJ15rO89U))y
zZU>IATHBkbg148D^DxS)a@$yMe3%KL#ONLHsWeKtA#25<Xro{>+<VGdao9Vb(a>T?
zs@FKCGxkXif<7m;fu03TM9SDvr$SYdvt#}@Di_!wp92R8Jm-o02E74teSa+CsNdVE
z7@>r<a?riW{yFX4Q22^yKgaddlC#^+B<AN*#=r)ut$~?|Ws^p<Xn*Z&wp#*!X6O7D
zfGAtmd%2qT)+(yo-p{tR%~Y#9@4rCBY3BMtKvx`(L8}URei+Z>haoMR^2~jdXBtVh
zq8{MK6>zSYQ0`*CMlR%(S+U|N05@_mo~6c=dRF$y`n0N$4x=bNKh2>B_Hgh@-S?7x
z7UMW?qnlAzi!uOyzTOoKE~z9NSxGH%7=2NVtYE;BY{R~cXK{AIH@;$DRqar77jEoT
z`?{K=>fT4<;bK_76s>h1X-6H0E&3+U8u@*ZQ9(Wh4fd_?m?_ijJ71V^6r*TTE=(s%
zcl#kvJIaRsm{*lN1mwl_lM~c$M*hWqUVrv~=IjE(lhQqV&S?B=*VWlC)fAMcmZJe{
ztv%wZ9n>~}c-14ptdQtfAt+ZrQWzP0X*%tiN1XE!C$4&3CBCsJ-L8Ek7D}-U?I=&}
zx<^7;f}bur-RpgjHfaDm?S>c<Z9DwxjgN#*W-jdFO+H>5=>qNMN1P)QXCcW!nPzSY
z7({!D)97PYI<Gb;T6<<~h4t;TTYaE`o+-+?blW3wzfXl(#|U7z2OL!0K<c?8Iv}}P
z$*&<Eb#MiO(o%TFjv)!}p9u~anlqo@7WJWr<?CfgD8qd^EvtyV8j<}=27@AqgRqef
zfwcFWoui@!&hk)`IKxvm##^*{%=g#|9#OfR1NYmwM0JgPz-)ZgM1`Of$+F)jI2awx
z)V7^iy;lk+HwR3ro{{h*w#8lXOxh|P-!CJbO$qFmZ480ERXAm7>82$f2xcZT1iI4c
zkHk%<(Ed&-@BF5l@kqIa=XbY<53Fo>Gjk4&H4<xP<-+`99p>{n2?;_D9=nLTFWcP2
z@;ox5%*%x(KTL68Ti{;|^wcUDx3uYKc%m0k?=ctu?7mo@hK6>#zZ&IWIZID+aW9Hz
zMdQUcN_xKdku=5N_2E^y9;iO$3_|AIgO0S;r!asA2|R!OBdNJMiGJHvt%U2ARVHB4
z8iVZXF}F4)E@m{ef|{ij4B`OGfn+W5tdwJq^=kDl271PgvGzxbBy!KxyR9Q(gA-A3
z3`xeUD>u2DFm3rGVW=%jRWb>Q99IU=t}SM)qE?c2&xO_1K1FH&*zs$8{CA(ie7QCk
z1d&<{l|7V%Q)C`ZLf#fqBd$wyF6&k<L|w`2y{9>Ho7|B5y?D(yQ{_Q08+{TSmap^w
zleVcEvje>ickE_ILM`suTf9fispOhYO3kEuw#`w5y$EF>GQO;9+XHO{VVpgx!@eUI
zMyxi~GrUvOqz$$!!9<_1x6Dk|A6c58ukgPY-x5!-_d+~V@-sQ)C`P#_o^hBJfl~UM
z-3e-OB9($+Z%&deg(|DldSj=EyMDzE<ia+$5(ZRnhZ167Bysg&hj^f8&fRt-#*=s_
zH%qxS_dQZ%hbi;y=oN2eC_n_TV-<wB7?=BRza7s3i+!TrP9)XWs<2LfV8zGq44|%k
zJXHavOVCdHQYq8;JNc$$)3q}(sH=JRp(lRX267N(z_YW_LKz-Xfp*R(`tBAOzV{M|
z3!9aLjqGvnCoyt1%6*Vn8dezRMmz7{HMf#0=Yznk>KOIfhY8k#CZ)IkD8b?e^W=_)
z?c;bxVRSlFUgA@BAv!@efcw3eUzJoTD3)3GPhyifr=hD(xyz&$bNcx-dXSFmpnc{&
z$ac>XH*<n^+UL208dhS7C?>lvVj!tZh`o}(t9_X#=3LcrbcBzF|NOcF3cs+BF)sP*
zQKQG(H#wrRA$2s6@r9(A+0`-@?7NBtmS9C8=sH)YzJtt%KO`gt%i8j9+K-8UaPI0j
zEQLjl{8LVay;*z)s4eT)MFeYNJ`~IJOY|y_j6G~B1@zm7;6aEu@tfwV4TZD8mLn<U
z>J8;ggdetRHWWIaTVvO5$aDF!_n;&J?P0P;+jSe91u3=DSP>Et+x6KkDQa5mh7G~!
z;4@0yv>W3CS5YHw;U;=hJabk|^hRx!4Izul)J&!u)~(KN$&b;M$-`Jx*#>P0hD=d?
zcoN+jZIs3m4nY6fZ8-o}_33iP@3GrA1XtD8S!Z`_2+jwq5JJEz8ysKs4iG&I0#D<d
zH^dPnFN~p%pvlv1nE%z(WW#f!4fV?AIl{3}EC+Ac$PI-}q|Rx*jY|B?)>S?xVS`tZ
zT1~rRtdDeWW224R;5=n72d8$S+<oQPcF@Mhhv^j|Xa5QDOc$seu$07qq7R4nst(9W
z)x_kL<j2}wNg2~JihqiOi>G6YO{)OZ&E}W)bbrs<tg;!7NMbLz<YqeLgfZEyfV^;@
zg<M3%ip!f5L?wfCNwEPRR=5@*meb$dT#cF7n-#h>FEKW&1`=A3EBEJn|7Z&X&4?U0
z2;7&jqL5OmWOR>qe~hN1<D}1+H#ng{K;qdJB}U+-WEO(CwZ*xXb%dvr?I)W^1<%JA
zhpjdj2De-Jsf&AXsk%O=#}1LiT#EPl!a;w%)|kCXyZ{KIFs#YRh{DrL9ktDkV6&xa
z$(acXDLQAZi5&b~3NP56u#&_YZ-kEgiOCW&>`K)A>7E5MY<YATaTvwbWTg+iyv|k!
zE*0ZVg&beXu9}s$HplF4m+q5@s%R-g%8RxxesRYl>8|(w=Kt9?RN$il<wAT7K6qoo
zBZH4^3RL~ysGG?u(e&o%LARSkPr2r7N!t8ZGBjRZ+rluZ^s3x=w6-!AY)fQxwGlyO
z+x@I4RxdfF?Wm^l8g}rVv7%yTlpI|t;Qfx-6>n4uQ}!in{QQe%$eypJejLT$3l)0h
z$EwQ(N8$vbv7v!f<-4o->zdq~LxOP;*5wrZy@_)67)kxfO<naZW2h??4rQZd@X*@b
zhkb(Zdwu-B-;M-$iU5OsNs`5V4JS64#Dr(s9jzu<_Bg~Wrluq`h+!4hxu1iOG97k2
zp6LqEo_?Dn_C#(<Jn*bPmuUVU?Nl|zVlCAU;7nWx*_?Jl`ORPLod8n^c+<{A_gq^g
z4CNSS??62BfY36qoxKc3QI;z#>|A1-f^dWbyqA{)I&e7vy`R{hi|}se^W38`1@qt*
zKZq{HE$#zM$;cmG?gQm%y?qqGD!(i4!|_Whls$H#ns|*}+>mlmRR-;oid-)T)yblX
zUk|6m3H!9V^omI6@qM2ql+Mz!`#ujS_n~$<#fc{S(iat)ixp987sUHz5{bg~<x%ui
zG!+)ZzyUG5&35@Z!JoX`F8LsEQi|Vo=Z|lyf$^(_bW~j6+eA-0w=^jHu4*F?WMO#O
zzK>@rx1!hp1-<=Hp%LP=AEO201Z!Rx0sAQztE-zJtDmEVcxB;B9QQBrj8OMN&=9HF
zqlJwjdfKjfG?)`MqVh_Rxx%h~)bCTd21^%sgnj3m{F-><vaHj#U&q(xHz<r!IMuPe
z->$2^AsZGufjCR{$m=MFuYa_#7o9}Lb=eJ%h6)KR@3k8rO`jYu@^n*l!iiFO;X00P
z_U{z#K{>ly9t{R=c>~6DakhgVO%E9M|BB+1cdJi-Hsa}RF@4t<YP%n@+aEov(61V*
zx7r=qAIU3m(4oJ<@kL1y!F1%gJtUr?cm+BISG{dReR9fa&}zf{2Uj(;H9X%;p7UoL
zQ30TI2;4dUc-TfhdUkyA4*s=~;D7X&je0cLNz@i3@*9&Fnu;K8V{=C0;)vwO;U_S^
zpCj*(jZc6e1w}UPhg+TSXyJALu)roJHdkQpZRkLCVUv8cHd16rX?It(Qu|e*{vQ-@
z4dR&+J-a!=FRy9d^RU0zbnm7^KD}_-Gae0YbE)_nSg4u#bvcruaNI6uB_?Mz*V>!}
z4PPoQ18Icj=A&9UH^TXL+Pp*tmqE8Ja7s`T=Y_uN<SJ>i`(i8!?t55cxf|SHk+E{q
zpvWki=%V29);M!Fs2)SHKrRlPO1rk(9*F+RCSq@s)6r@ZHqxKq(I~dOzMA5zvNy4N
zUbM!XVkhikQ}n8YizjSpPKEFVqONjXTcstFKSirQ0sAM}+Enkc_834A;20qtrFFzJ
z-vO@b>#EjJ1t&AsM5oE}Y+d3qin-2KIuciq_-(aAi|Pc9aE<?{Sw#U%TN@u%z`%7r
zlqetnueL71Cx-xjFkoYSz8yEtHpJH~2pqX{z#DzozyHf3+f+eS>LD|)UfUdVaDhqu
z=vzUDTM}=rDj>5p2cC}#s#)ydHXj&g-nVRf^o-I-3^<Iy3ES~#Sb<v`h!U)A>u6Gy
zl%26Q+C_V9SA6FlB7S&2?=46_tvIL`Dwt(<5!3&w^<*C$y#2P@;h5Wm)8*b=5ky?f
zR~Sg8PCbQEq1X3~`*Pr;Ts71=dxh4`4&|wU)|y!!e$?$63VkhXRdL2pH6*A5AwWi?
zm+VM2e|+Ph7-fm+bGXPRx5|!I?~j!0<1qKsv3MqI^_JfH?Rd;g?uM`0iQL>oFPCgv
z{dwHi<3&lZ&M6;)olm?~E>QRvuCjN00A=`-4j#o}r=7{|#BqVkk<6b16>NfCdiT?I
z){%hbW4SiXxx|TFq#Eo!KaJ|^?EM^+&Dm1&z@2~8rF>yWZaG!-I9QS1>w_2#&s@$2
z-MxMGp-%%-0iHVkH@e&f_Hkl}R`Mcx1J3$Fj@YJ%5N&{q(Mk?X?%hqkyF3o+xtu<Y
z1`d+vbNS30u*Pjm>~r5DqelPMzVHQ5)?}1wU&fd6UK0bN^lV=xP(>pzmpEd>y@Pgc
ztxHK*;wqT)+i)pf5c!+z3{{NGDawicZ7zp2OvJOGZ~4xfk(Zv-D50hO-us`{P-j2*
zpKY~E?Z?a1P^_0f1@ZLa8~wWd9Dw*CR<x>`0j$|CIVib`>K>HJ-fUyxbX3Ohs*Q1@
zbT*)J;_knCV_`qaMP}D*tn5S_jbfno*|qUZhBx>i*z!GgT|5gHkzz-_eq-`WZ|TZI
z^r0kQ*_=02y;2z<p53^yuoonvnr}C4ypqjXWjAk3W`!%vFNIq+rfCgE%m(?#<f*H%
zTdQF?2LM?fV8ubgZp*34xS;rx+pC$pENKMi@e_83Pao47Wp@WhUoO{qS|p#sknBq;
zdXeQ&2VwBv$<H(_`bDSeIzkwq;ni45b;T;d(Tvy_Y+0>mva7L?J}neUDi%6wV_`6+
zF0?Tli%zS)tqyZxYgY+Xv9TM&wm@(zHB04TA=|<@?_;0`kC@kOyw3;b0WM^M|8V2z
z#BA51(iD>t$1;18k&RqAcLk26p*vzza#E;X<b6SkSexdP=9D1c&Bn9q{s;X!2L4Mt
z_5Z)WZAPM+I|o$`jQUG9Goj~H+Uvw<v*KCNJ#yakqfME!F_=VhreH9d>k}jcBJMwq
z4lfU33xYJ1Uc-fI*BTwSg^pxwa&=91U(!iK>k_KG)Z6_@Qn=mzC*&#rODCyC71guC
z7<Lv{AXE2-md8Q(0}eyUcwTR{P9OK*ZS@H@E>SD$tua4Epg6>*yOD01s)601foN9V
zoaTU7nyuk~l!irl&{}-r13iEFi?#YNe;sV?{<K>&>+ru<>DKIu9#Gaudx4CO-?D=`
z*>l^kbfj?RW71ijGwocptQI}tn)qC2mtwl{Y_hc*Lw#JVQ+}#>1v}R+Jam~}3u_E&
z*ZGKaaEz$#_0g%sHEH}+0ANE7Tua_>Zl8_0i10x#a!MSPnv!_rW46g>hhR#&>+Ff5
z+QlCp;eAVjhngI9sp}^2#z~xRjc+6gp?)eV_-#2l3E=7wf3{cC)^(%!oHuPpeg`ju
zz=$7gzwL}?)Zj(6;N`G(`3X)EOncrzz4t**j2EiUbsgzCK4Crno+F~(c31DEKUDzq
zY1`}X$vd>fdgFcXz|ns^#PD+OJ5YV2ENOf}q{E>=T#;opj$C9B3rKhA#M0SnN2>26
zGP$my&&OKD!rReA2Qf5Iyd8@^eXZR;*>}eActdhjZRmeo7M@PT7+zz`Pu2e13#E>W
zok~I}p%0x>be@j>%Cay^1rnB>Xzw_gqC??CJ5x=7l$GxNWCQU`7oB3Nl<47XK0^hq
zaNSohte>mCJ;XW|Z%KQvdWPeG5D72zeqI3>_g(FJK3btJmn?;9d9QttQ);Br0g+&N
zsK68!lj=u7`Mqa|RJM<O3I3OVwhK98k-#M&yl5BmV>UoH8G`K-|G3;2C+*YdN?ss+
z_E}=S>_W&NCaxIQ?ehea!$rEl6R?qARKvY0mWSO;c)AnnsL{W(uW}%Ya@N?_xd_}R
z5ds)I<E89|ohYS=;{3mfXY$)UKs!8*>|3Abzr#dDU%`HG{t%ra6_<Tq_0@;Zpw?lx
z{g4!)GU?68`{dz?S??<ePJePLSlQeFFaLA)z)n9v5BUGvFFq;hFzm5%VQ{~~JGa<X
zk2%961H~RtCClQ~@yu6LQp!=g#y=yps)MWycI{(++Lc_6zQr{Gd6{t0zG2r@-ICC=
zoRBZu^^X-Mu4_e`-SAjp6l&}2#>aw<8uwef$;W`T<d|r;n;&zQNyg$=>=yszZ-Z@+
ze}dNcO}q86!u)YrL1wo-<_w%~NG-AWZAoOe`_N<l^<V7{hpZ_3GK_4n_dV?%8{&W6
zJH>|j=mR}dTe@p)ST!G|-jvd#&xXe{eicK%%x@#AXD6{hlU5CpV&r4NvDT11xE_XQ
zluuyfVxsBB_}>)OS!HA6V=q7|+Sp^`5{wQ-501|kVq%$_j=Kr*%*{%u@wm$-1|~e7
zKr&!f5}R}xM`a?FcGm9tFMTW1$dt!iMkW#W<^S0<hnZ9|a8bShq*}waIK#Uj+keqP
zcQ)bfH8Tb*CnLwhe-_6&cB{>bZqPv4oX3K*oPn2qi_+1~^|$Y?$1iPOfF|9(u=+L!
z(SpQk9jWhR;aHf|gLbf?#qRsnMBQgu=l)^4KgW{9S-eaZRsAbRi&`DG#ZEGR|J$AR
zz+=JAqxF@uC`(nFGhNb+T;zGn>mLg#V6qonZ}b^-C?{uALaUH@soR#uklHT2W$z!f
zmc->W>WsGr{LXQS804~}bjIzEg}qp=gUH#1iSF>h>PQr8UGb@q9l<ogvMrA-)gx~c
zTbZ;qTMYGT?*W3JZ)>s#_K+eq;5b2x?6qqX%d=Id%pQtqDCdhTii%@gSABfB`l%~<
z*C!&;E0MXe1s-UFqj@yP?Z(6@<~|aXmuYQ^{!4;3&XG4!UTw}##@d!dc|+@JDka<2
zm>+4_adq34hZNdR%%L3oVtn65H=_I)2W@+FuKl5Cc)7>zh-XDiWae<+wlkmcm0aRz
zZ@ZF+rjjR^<O{9#e6F6%nJ)w&VuAKsPjsOB2<<R&F>l#!KWWz7X?wGKP8G3BvNQJj
zV%#m~;RDszCt&xK=Kd8s6dRN13q(x3{aTkmm4~aAgf<3acqDt5EUXnm#_G#<MGYlU
z|1~?B6Cy<psrEtUv7|G(d#K!y%%R6~mM!RaPWT)iTw(puqv#YRO$h@&b;Zbq%|xs@
z?O)8DZSUmLspveK;xkoil`m62{WehbN^u?*NN}D_0>r}{Mew;KDKhhsSVWTTd-0VN
zo>G{;CblJ>=~CE7?|snG`Rs{+S45`|@@H+Es`q`E@X%In5v7WOH|gkh;`}(8lpCK(
z1GrttJ%#hzZo8P+nhd}ANzlY@8lJLGD=4LHkYLkqpZTgccTn2n^CVssIbi%!^`t2C
zOYJZFBEFZelk^m-tMp~!6akw`bM}=Zhii_ED=Z%*@^!+D%7fwog&Zz9luEJNVc#Uk
zMCaSL*&2}FE;96f`!1%jpX>f@`~LEfgrP079~?z9xanHx?MMIlt{kEE{>6UE&L?-Z
z*w4O?+bC)Ci%*=B8wO=l9HXe8CDLBC$vG5h@ly|ZwYMoAEL!qwHYKZYObVa)Lb^in
ztsEY9-KN4Q;9EoU>-`6zJ`}OMVN+-&a97umQMdiZO@%S(RKoI`ynAUXm$li=`KyDY
z-)@P)q+md7g9|-qgZx|kx>wn)n+mJ5c7)xwsqjA-bL{p6fbQ*%%M&avfC{T09M8mC
z;KE)+ts#k+1=LF;RpQY6-AJu^MjGa$QX#X=h9^{7V4xBoGQ!~wUG2zC&cR(72<f%y
zs0xMXIck$LQ?mUkJmXcEd$AaMY)tg+9+;BzPutkU$vkEC9p}@N+a6BH(cHxrr58JD
z6B2Yv8>2e5iT*jmD4umvb_R?0hTWCFrYd}4%BI3hEkpTD5*CErX#uCl#V83Qeta@6
z*>rD8aVH{CmyrB=Mm|u!8QtPDeQo}Eh|Ss*SN!Z|G!mUSaNC@Oo2v~z8ouf!o4YBr
zE%%lzHuD^`Qe6jUo&11v{0(-j!hII{nv`02_iYLuhv;U{D|vrDxkrH^vX?E&0pxv3
zHsky$b$Ww6kZ8lXj+`G>>%G7GFt=`f{QVRXbfZs{+NCQ|%+RnY2YPoWJK{O?B})_I
z1ff%qv;tkh0HxfZwN}rLlqd?N(`?Uv>RL(-%uegbR>chOl`FcEx~X$#%X11-QABmv
z${;%_FZXd9%OgPl5}Q}_@2e9o5)`bnHQD=_-CUf`UaXA;*H=V0BDd&3e}#OS4@GOw
zf2OUt&(`^-8$ITCw!Z3&yM;kzn%|J0a7!G7Zpf5p8~vGfduy<Ww#mEcZYOtseB#D)
zxfdO@E!nkv8-Pg?;MREN4i@F~vTXso`%uNa<xgeXefe-?>aZP&vJx9&JFCtmUTRvs
zwkyyWT(!L4fL``Sb^5YB@9;JgaQH$JL&?;Kjitwj`upD(|IKz+K&p6-Jj>9p<ak&R
zP@I;Vr8l05Ey83ay@nmgZxL~&jaG&EL%H;XqS)bBOAa4EI#M-%8*Z|&Le}R5%|4go
z_9;8+pv`K=_wAT}4L{n3FnBzHVmBTHD3w@0;giJ5V*NQuiH<trPbH>BRRRv@hvIu|
zr#BUg+5dYypS=@ETKu<Rb|$GF14x_f&jbVd;rQ!AXD=K6Y{Cf#iZEPw%5(87a(u%6
z;fkC9LuB~mG}!w-;Jx?GUts5x0A|%-m$VNY6C9`VGW{@mbLtTpAF_{f*A&*`<NTZ=
zk?4Uh1hDymYAzPVr(Mj6A#=49N$8UV3;TVsK7X1`!lF-kCI@t&XH|oJp3AMg8DCUW
zSN3R~!Ef1@@vOubv21<Zj$h?j+uq$lRU8>MzRpjBkX5Acl7m9Q+{i}qO*VqU#oa0b
z<J$zUR;#Y1`7T>=i=m{%DZAgkPcZ9hmfH^rQVC+{#n0_+`!Sx$=}XxZU*!<}DLaAz
zls$}GpZ2q3mXc-<ss#EkNe0E3R;*3j)v7pQkH<w(5-QqNj~8ZQz6xAi{dnO-rc=ka
z!LE5cbOS4DI*Rkawf^C>29JKZ?(xDium$Y;$DI%Iib`7dm+gjlCYu3LjcmYn<Ku;Y
zLUQmBikqTMT<*m+<>tqOev8)D3-L+p-DkHX;CGj#I)l73IHiUaGBDlxcww~gqZa>v
zTXdai3kV;u+p}u`R6-zm-j!3z?sz=z*m^RNE2|;C$JyOfrA;03_}Rh5Vdl>NkN<3F
zb_JtbkND5J4+~garO%NH*FQX=S4tp+?hzHb;A6zoAl!}2HaJJK3v5(^1k>DPWBd!h
z3rNK%d(~J+(a4gq?4*tJ51<etwQPLi?;he#R@sEW276C@9H*xoEhQyS%#|ly<`QYI
zCRM<lfT5RS?WEmx`SWtXL#v>agWje*o(j0j$=9{POIQ&8KhqM9PAp@nX`*x8a8$<T
z#NKB!qI1bw2LazBsxU&@%xGPtFl`4U&RL0<nGN`A&&dU-sjru_410KPe3Vy`%1_RV
zA>!FT{<9=mk(zZuypdK}`q9hkjfIH^PWl%CRbD&yRp2YEwIqhJ`>P%l_$MGU`eIS6
zh$g$o7RN8lG<bA8kUhaD%63939jo=-+0oh9Y4!d$_Wfzrm{Xprpn)bQ6wG8zrR7q`
zFr~2d$Xa}-j>bse`gk&W-6bt#y|w$xyQeO&j$C)5cw43D>T-xgVx)#?DN^=wKO+Wu
zR^Sf1(m$^4TxF}1SeGy)G_xjgP&V6@x7^y8XNfw8X5`EEP;|suy)HXytXYn$g{{v4
z8(Z3JLwxHOhOYeGwlSXR98|QS;-@z`u2yK<Yz{1AKt|!QC0ZB#HEZ}PYq&LFEk=Ge
zU*(u<+p4~`lD@jzy>F@nkOMh^n~S)%(_g&%EGEKT@u3)2sdnVw`@B!GZ6(H~7*QtA
zm!J@igo24Ttj9ZP6y&?36EElHby4?vC!<FH-g+Oev~6BxLr1`YT;XNXf+_XEp_~I-
z3(K5z?{?U6qM}Q9799x;DKY&U))(y+`GYa%%rPkSXgn*HUEn2cmL2oC*DO(v#pBf=
zI$OQLi3A1OX)aZK`g7|q!>5Ma$AQUZ_)(mW4&=XsnDmczT*vQJkYH=@Bg)SBH*<B#
z4OCDaFU}9jD|5E$$Bzz)tmmrrt*nZGC{FmjB(~yTOOSchqZ_i6Gwys&15X`%f#7pK
z$ldqulko8$R=rmqOhG9h#YqU(5K+aI%Tc>TD!`kcwvTi4DGew>8tg)hzMo1wqK9M9
z=EZo%btM~C;o3fl5#_1vClq|AQqtL{3E3Q7;h$Ah#+l9GPF-F8eTmci^9p7+nyZY1
zhpc|#^mXsNyY0&)3k1yl_LaXN<jKDF7qa<4ahI|gHnBJDo9gV5McTftULND<|IYis
z2g|;HyeJ=5w%ZT>@b8Nk*^kv^U22f5{L|xSC;T^jRR?<PXGb%~q31vP#Ye%IbGd+&
zZH|MnrhX}s4mz=`HW$WdmQKd2H~W>QeJuVC|9c((g&_m+cFpGCJ-Be{?Apy`YX5Yn
zUFXkPw%2j{xPG&<Hv8fHUIN|j%R@wl-4JiYlkA3Bx||jL#?6H#Beyu`{7st+pEb}^
za^-li!Oi}R-62!1?wU93md%AVBiDlsir2lUnhx<AJ+0kZJu9W+5DX96ZJUDwiW-8L
zc5inC4D?L+jolFl2++y|u2jhk_7PXETxmmME?2}+Fjw8(LwzpjvAC0m1rP@*eI+Y{
z6g$In5meqvWrJ*lBao%j9Wye}$z(wKsOUjHgYFgfnvIEPFj+7YB~7G_&3Qby%*IvU
z;I3aBuJ+mZ&4nReQPWv$=@WbqSs@|ri3t*cexAsAQetXZjjAnn*W=ye3r>EcDLH;o
zrx(V>ruo|`Q*3&^g#{GEtd`B#96Cj6Impd$W<rQ$s`wWvUuWerzB_`Noqlaj0@mCr
z!;iR}K)kb=SNhb>^C{sF#xZ9L9G<@}vV}hLfu70i<0W=q(q~wxOXvV(b=2-pz)_9p
zHao$wx5z(6?|>uOgSI&Ps%dF&s<8*+3%@Wp&|tZ0ec;J;Rv&Nu8h6)>MyC$UbuABC
zQ_!34M`=>TZ(W%IXzAwgkd<b%Bx#n?tfd-M_7nx4^;v5?Lj=NmM6M8h)*jD_K*UCd
z?X`}eEt%T;t*iRVjYn^zF<)L$UoXeu%4~rpiy#$DAE#|~wxGwiY)v&vk+(5uoUpaN
zPv*fpJ(T-I9IulqAnQ_Ew>c!Q$=HsiNL*iyMZ^w`l&*F|(v5Tw$Y2|@J5GQ@Wr?J>
z;tsS;)ymW_$HjZIgD~~p*|x>u?NpfV65CqAtMEpb;<n}98U7pFo+H*tUDjzk;(%0@
z*qshCT4z)_2)nkc>P!-`G|>C)`FIwqF0S5Q*j#uR>X9fdM10&+HB@R#9ki!ycg#=b
zjw2kwr5;K-R`v$gxnH?2dvmr5Ho{vue^MwM3OTg7q<i?^{>y)MIEimED&HgdjkaXH
z^*Py2Tx3VH)q`Z9IaVQ@(@hBHa+MRy5j!5>OQFosR#?mvd1}B}7s0T>`W>HWi(MK%
zRn1yO+l%tsPAAS$A`sAO@AzDG1E9b><4ZXk(S0EM7pEn)#40=M>x1~(W#^)A_j@Xq
zU3<^Ju0{HMKPM(};e7s3rKb;~lTJGP(Eo0eN$Mm2yQF*mIDsNS=C^jiNuiRH?P4J6
z3V~bH1r2`^P^I&b^8J09kd`={;^$$X`Cu^Ca4r>=h;e{BRhrf>982iwRJ!an`*QQy
zY3@I{Y2d%<A8B7XmPQj~{`Kaw|5M>auE?ohE{RLIVcV7VoLl3YoHBb`2farPOSt^j
zr-FyrJo_%E$|0vKP*?Z&@k~zs1W6Mc`>OpA&s1myiwOZAC+x>;=VvvRw%6>ZiX!C?
z=^jSzxBPP;u!n2&7oSiUX)ElB!dwv9L>-%}{0EoN64D1g;Z`dJ-@R?uJW*Jofu8A;
z?OJb!c8pzqUB0+ys$Ks?VLF;P4Q_bC86A1J$j`^q!J=lc8=okw(X459lYh#M)M__B
z5q2~fD<z1Mj_sC&fUghPplIPWP`y&+txv?w+1jv#EJb!(&SqKjJi9$UND$<`eTR36
zch<w~(!o^=)vHpDp8YoDiA=?`UrsDef(VoxINV^F>|m((BbzR3JuG`Ac(K!lCs1C|
z)kZu~m<w#-rSZtXjI)5HYDyXAbsH7n=x}^F4vjI<+OxV!-Ti7gm-}pN1w1Srr{O{D
zyW^gSYXi}-xTcS<mccXQmpIXK0ZwpKc{QbpSA5r?!~>IjB^NX-v%7qimZ_k_lvt5H
z<nKIU)1q_V)q)x%a#EkN=~v7|@(%f`Y({)5j%c^d%n`L;B*@KjEd8b!V_&v8xsofy
z*5r=`NJ#S>o;H~;7vu{p$hOdX5ci<4_xl0{m*{osQR*R{c{&DJJ@(k5cqZH)1&;@~
zw--O*%)axtt{8Jgl*?ET<hWeUR_i}VvOtF8dI!QIv^uTviIgcQ2HvDx{Y@1JO7G=q
zUzaB2oh2_pi_cJ5GOX4Ap4@`W)$Rjvt1ZK{mETaJ6OOJcXG;YU@dqkCzdUAfh5m{+
zmsz@%Nw%|P(O8{5z*}Y*!J2G=;itjY=C6{?^r6HXq9F>p;>oryMl5kXj3u7!N*Q6j
z_b{u5{70nBNi24EwAe<U=qfUg+NLCq2ccu}C7UbQPRTzQz8I@kem%1d*BYnSyYE|I
z+j9Hi=7nCuw#T<pE6Sgge{Dy#EsFMBS|OI=MVOJD`L#lAA-!EWN@GKVJ?}6M^o$#C
zFGRPXn*&Q>ZoSr%t4+NUl7jO2?(7a>Mo8Y9znfO+I=9}ODFtj$v%;+Wit>uFTlyC{
z>VnM!cIb%^(lUKjy&ZO>7HRSoJK~sFiY>6On)zO4PHY~OeKcksc@oa-E!Tvi?vD9P
z{{zzsjx>Hq|F0A5M0TQ#5Zp#`{O(Vr)UnS_`4m*}T)9Y5<HQ#n3w)t%^}4;22uB9p
z&xFWMSdOs+b|&#i|5{qf!UNg=LJDG>&FO*Yx9wcg;B?MjdoMQusey^&e?NbAHK5h9
z^NCz(yX6G+fn%RAHkGq%un&D~N-&f%S@^2e*+)5FPZWU`ez$$>{ZjC-oZuI#$-t_Y
z6Hii_#OjMqAk$^o`XncYY^nqw3J(7?J7==;eJ2m3edY@`8k+jN+CG&VSi#6){#D6{
zFJfR<SQlR_hL?f%%W5!%OzFTqZePW-;_p;?*kfPEGj1~0q5Nx?9B6LDWA;tZr@Z+D
z@V8C}g9qDp@uhR_h3f75T#VXg*8PWAr~`;ONA1UGFHMQ!Eo6Q9$w$Ia*lIt=`<P4F
zKfmPen1YWJ_o{7ira)tg6H;pRtF|~>BID)>yLwBx@DxT}vVqw(@d<`#qTYm}e!F%{
zn7jT!28Shfo%bMZHp_H<egK_Hk*0lo!xm>=Bu>CZ6-sdZ#f{Ng{G3uxC_Da54g+2f
z-L4<M*$2Q3r7eC-eo`!Ta!WaEgW{QdX=U(`m&2|8Rnai6wA-SErVfuQ?V;TsU};Mg
zo=TBJMgF;COQ?a-$H||nXj%v7z~PtG2Z0W8Sc`vga~n+j&}^jho7?XVPS#-we>n<<
zZwVfY98Rr>03-a1Vsuh+LOD{6j9$bENX3LdmW|4xNWNQ)hhw%B?(jE%|C^1i+Dog;
z{lq>PmyiI1Oruz1Y`jAv)(cC>goKOjOd+f4mrjC7TjHXv$2kzaO`Y8p-|8+pW>dBl
z{&n`<a~9aN3XdMsCf7rmVvpMN>_uD&Wnn2azNK)hfBmb?%=t_q38FaPS=qR06)uU&
z4mKyIKtML`YwoE#cS|tXlQB%qiw4rbvc*ak^aYOJnq?Sj#7r(sYDLpbiL?7+8VWl)
zZTCk9`V{?R0eX?+75{BpoH*o0le0TapFQ9JFdIsP)J6l5)nUj=iJR)LcqRr;cE?6X
z7eS0Q)|4}tPk7AI00%A!2D;@k-67L9*LF(@)tV5)io9y=@k{koc3VdR2u;4s!uPQm
zy7Cz(8BTv^j-v5hp2(g|+`2ruS7syBPs%i|h=tX$j#s!(gLL@;uZh-Jhl(_z+!t#D
z5H!?S+8*)=a+Uwi)@_M9m17%?qZVO(bnSaz>>|OBVy$e0FW0^J-V_hJjWHDNxh0Q<
za+7X~X9smgfl);#Zr<W@1&e|#p%g;SrX4OIE2oZRzb(1=nA-qjt5e?}@3C#U2&$K~
zJ*J{`ps(7F>Qz&R?aZO&fS{8FWbUdaBaLV|isRgP{)+X>LTC0u&PX=^sD6bmdg7Uz
zspP=C&vyH`*yK<??9Gq3*$y#={1$q1gVbO;EIGdq1SDmZfUa`EwnMq#&N>k~UM((i
z8y(8;R-?aVN8(Mnp?GsS=C7~%M$f!tv6NVSb~NFqCbR;`_}7kYaV8(FI6kiauWl6%
zF#T)C9n0j@Z^h&FMBw}Ybo`R_$4DN;THp$lp30G=pbm08ox}!1|4#nBY_^^8)rNmB
z1#Jc@@HJ&OF}VKjY`iI4cB~+&8_<QEOCYDnLGQg7yrfnxt54<5AUw9b)Kfws|KcPC
zaH!<;H#_Eh_2mh<nsNFbw-2f*zrmD0%!zT2m8^5ztRGc(LpPa!>+R#JmsbW<O^&>w
zcENkWd6hCAb}`p^MoD40%0BVaOEpeEP1FosWS{w0a%QHQ@#i^C9qFSGvA^(^pnSJ|
z8Gj!@XZzJ<JTbpg2n_<p$I8C0-l*Q0Bsyf4a(rG2n7Jpo>|L+$O*N0>sAXVRk*&XV
zh^YLn*x&CG8z{#W7p*P+y|+-B9&D~ZWFHt;lt<wuB<3IE*?we=BV}tHwx9efQoP~d
zYd`0XoJMNymbP|Tc<h(>tV9^Fto<D8_N4Ph#Vf8X;wyI5lVO9x{RIV1{+iYX%C33R
zwGafx>wvq~UsCFl<cxORlfgew>9SZk<gb5yjNm7mdyL-TpUG8-J#nM=38O=DicY)9
zKTz=fGO<D$=jP}{R!fyx5+8HRljSb{Q(MP_Ht5OVE;^gZr_f=yMtfaQ9{qXPZu9nC
z1jpL#Ii&Ovq|V9Dmv^J^=&-@wR4GzyNY#}BD5QzqZ$qC91<CTZmTrttHtfl>rKs4i
z?+_@#hWk`GMqWm7G{QR`=s_AC`DChy<QEc<LQ(RlQ+91PW{isAWt$R(u+MBv^x1oc
zaxpeG`@D0Jje9cq7p0+Y)LtGR-<R5makxyVpcUU#e4<1}Fzyq*XMD~{FJhDY*Feu)
z$&`1+SG4CkDtg!GZ%U%0uB{vPuHL5Q19%T2pXzOTeCwVSh2aI2N~|;^$9Q)OC&5gI
zmf$Km1*&e$Dx39WVSPGkR-;9p6Vq`T#(aV>IyV=ysVM5_JsAfwaO&<VDHahX6!9rZ
zX+gl^B7+C_i7nb=3muc#;t$$=6^?X@`s%nqbbnx90T$?eVl>C1+P%#B9c2CZ8$<Qt
z=<{L3*9R&LdZIOMuE1bzOiNm(!bQ?^ed2;#JruGni76Tru{iZVsPY3%i9WdMPFot?
z;QCM)H)UsRNs>?v%W`<OI!K5u4c49=E-6ceLMUL=jiI<I&EC(xF8^$xXX+|jp0J76
zky3(lcV#ZaRMJki*y_uZEf<E$h^zR*E4IcLRH>r7Y;A%^FPIM5d=CXiy>>8yJRcVE
zznpY$+q&GW@^pa0*FRYVf+cM04f(3RLE&k(F<>bE#7su&V;6Nc1*Q~O{t9oz!EVk8
z$YT7UZAoO}%c63i)XKK4)n|my(UpYTZMg&}587>e>}qvM7MhYrvX`TlRJ}M9OCXkL
zxFh;eiXok6if-R|8DYh{gT>P>gYDRr2%C*`@O;iGY^#@Z489PZN@F5fUoqi5)g-(S
zv)p%gS0r0f)kPvOcRG6&y~*Aui?A|hs`WZ*7fzmK2NF5|4%TbzkiX%wvBNn8@+YdD
zq76Ob5Tgt@*lK;zK$}A;p%pW5G?#r^_(|Haoa6#hy&bQnLV;73N6FbvI9S7mTmO?`
zv(*93&E|K+DPM<9Qe_~JpN<~(4eU|keDrlk?45Y#_FpM9_S>02C#NlEVM(}LuztV>
ze!U*Bck<%bdYpYSIQ_^=#sPOOyB>fYfG`kV?LEiQXrg}KPiSPj1xL!}aNg&$0^jGP
zAOd$VjzwGNZk?0(D3LP{z2?W&SaP;;@$5q4Q||NxQnc8`sv8w~7wuyv%SEh3Y4M<a
znjMuiIy#ki+h;j9@CrjI3H5WI#|qRfU&I$)VIq`94!x+D&HYuh2o4c^9k8#X1%60)
zg<VQekxX$;<BI(zx9wc=A*ez`9!l7UyO(jUeOCb`8z)xJ2Kzq0sl_F%q0xTGUgaXT
zL_EWf{-iP=-f2H2o<UHp{hW|;L&!B$k^jHMGu?60sPq#bY+K8Az*q9Bt>vU?f@isU
zYq^xV%MEqS*1Y|-=N0uqCNg?GZP#uMoe>G1$qs#8{+!y>zzK8xR#!qv%e2RC*m|Yy
z3PBDMy3wD(3yH|3kbpnhTKFE0s7||i>)E^9KKb9WPTDO#$GHn_P{6bgZSg_7)qf52
zOi`V#+XBK;GlcUD-!i-1ThxHk_9nZdnvvqV2S<CDONBI}IeyiK<SaQSln;PPZ#L9N
zA3D^AZ4LfK!9Q)v6+GN=j-J{^RG-Qr&utzf{dFB#8Amx@$oBqc8{?1v)KF_<^DBf&
z&whT~#`)@zy$!?5_-LTLeS~$Riil0vnwFarl5;pc6js_-R#$#B6CL7z`X`%|T@_`K
zuWyR~Q-Usn(_nw33U->qI%o3iyKVZ`I88WL!+<}d0!jK6Pf$EFmx3A(Fhz+d%+zo3
zf82QDn_dK!;`YeZG&i1!n3U{M32f(W4c>Jc`fJ6mEr>>Lg5aJ*=3eOVQ6sp)?yG>&
zK_V<2u=}?Lzl{#<L5{ygj*q+N+G77}&L8ao{|?&q5*f0<)NTz4p<NNN$P9P_Tm7%l
zBqr?`!zeVrF^2YpQTvKIpecuLqLdhi$I@uyX?$sk$l2oHqu=YaR$n0E-SA^<|J6{k
zWhl<(FiPc)+?us=_pq*t0OZ?W0{S?%tE@t{+(*DaqSjVMr{R3SWiA$831C;}($*{N
z6q#w2t?>_M*KpFWO>A72bf|<CTWBAb1Rgwz#K~f_-2SjWSCH)YCDe06&|oJHakep_
zkw(LvbmciBbLyt-nH%_Z+Z;1c%)pik&$6{R?zdK>w3MRawk<jqU#mt{`fb~uq&cgt
z?Lpg-3p9CI&1%l!o%t3H|8?7ygo}=ux)RXN^NIQPHtNRM3%Q;o_EEIjZkXtv>{1`0
zCfZ%iQLBTvgaGAy-QEhwk`<-6RF%Yr)?4-K+Iar81Aa{Wao#LDlwB5|ZJqy;jFg1^
zup^3y$vQi-HREVKF`ITeE>nGpD?He0<Ue*adXsDB5%)n30FPB`PKCu3Z;waw!f0zg
zhz~CQod~F1(<fK+e(TTCYU)K*4R$I%cP}*8>pMD5SCf{GrWfvGzrEwsGSKs<$@A<?
z)xQK~E^tJjfuIcXlG<7Sdz=b1pUZ_*J_vg+X`5^<QB(8%goC?O(F)3QRlKb1d<^58
z8><ys@((K5IN(@lovR=I3LGgj`|P6}c9!%ZAN%>zw9GD4eUe`l?8+{AF}5wIROMbP
zNk$1ZPII7ql6^ghUDiHLcsbKNPF$)sad@94Z0H+gGLqQxd9>DH@JiVveWmpQ`yx?s
zH}v4l=$AQ=GWgk7-W~RmHr3`So;_b*247l6nCCvbl+Q{evKR^6pmVv#{Oc(Dwt`Ey
zr+DpG?7Mgt>~*Qe^L;|pDYl`pnQiui_k*7)HuxW%@Mn_@{-^ldFDdCSU$vioOE!o0
z<(Hfc*FlHbQ-$@xeH+f~s;A;W54#nYHq6meg+m$W8FHsx^HlKac$c;m>x5+bYoBsn
zM}gO3VRbxR_f+sov*GVr?E0sYIXZ@t{)VUG)|%h$SpaW*%6Ts7AT#fUd#~ISA1eCd
zwA~yd!HpG{r^Bq|E%8jcfl`r_XM>(9yxsgRlt)jM3;8nRc&FVKUmWn@Ccf*p+vAyh
zIbLB7DCF&qfJ$p2Jpi>%d80Uz!8vB^4V7{j63=wn^pzcO$c83D;g0JXh`dk&h+)yK
zlr$_mB8@B@vJnh-uu0xzBfPhP9yygYbF6dQjC5RJ4Mc-rqn-+`2D^uLW~t^a-9C`Y
z4|>c~sYOG}iU>b8aa{PHA^~fyuzQSqD)_WcJlqLJZL{&YdXtw(W|$D4I#mfBsNsr>
z-9#r20<;g>B!|8ms=6y+LpRNuXy<E>sic>498;12knRxtm)W$KT1S_ZQ`4O&iQ(hM
znc>}%F~nwGCO^36E>a_iKxaKwSo$IVht2V|7L$^>Oy|a`kZz58PDk#`HZO^h^WdZ{
zsMb>Xq_9$Ol`PEp6SD-PabLBxC5i=t*>Cp;PK1qDK)){nZppT&LbQTKeA@23wmAB!
z@Tl;vs&z4w;y9E3WVP8Ey_k;FlaOS60#?&q&pI{c&lPNDif`q_&LFqiQvXFvJ$5*2
ziFa7cIXYW?-Ei6&ZS7CF>av&X&GUx5!eOK!WQVh;jjY4hggo~IV|4{+zJnAWbl8^X
znvw;#IOwd5nYkKOm!}ks@G-VJ!JgN)3aMa?1A{k4o2^Z%RSeRq7JDdrB}H-(&(<YY
zipE47#YtNqBN3`nZ1c)!USPKk-gS)(XB)FC1cf^Bxtn}SlWXO8v^n3-MUYhn*DXE}
zI{0=Dde@L|&1dZA<;|XoX<G$QDcT~J726)qWcz*zWXj)qhff-T*3&jSSudDzx9!TI
zJ7Ej8-=5D6DfX2#nM>@2#3dX2W#&dMIBg{kuAba#MJjNyM*7qvGLXYx>`q*l{MNQN
z$7)z96B9MKa?dENyG|8-?Lf{5h6@85<bKQ!B{8+sz%Cqq>Z~fva%?$iM*LL8T<Zi%
z>F7ut5nCY1zUVpqHKnCK8qdNDO=>+mmQ*072sZe59tg|e=1(Lj^&|$d{^<TiaB(Wn
z2U6>}GD)~Qo!BH=W=XM~yc0u`>P|aLYI-}9Lkt<V$OZ!4MPI}D1$_s)&sNZ8!Ku@B
z&Y^WGLB@J72b|mm$ts8Z`wrwl&+JY+?+ZV9>fQE1)mh;Y#pBzDxj^LH=x(WRwU6Tb
zLm>Y+J656Kl5X}w;$lh*ZfzI6>F;6hK8db-fn9zfN_q0BzwZ2x7W>Spf@`H)>ZQ;9
z2<={KUu1{HU+K#nu(MsN8d`o8n33sTMp4Q^aTO>V{n!3w;g9#*rTBG#IG)$-n}9}#
zB?%#lf6BM<4E-I~i5zVY*msGi`BUwC|C_6>&3>p5<<(tY=HUF%VeW$I@s_o#5I=wR
zwWw)b(v1gcr~Tq{9y5BZ*|xBR>1eU5wmH)x*Ha7&#kJ(>cqVhW3{9xZ?V4>V<l@f(
zQ5AOY+HHRKUJ<odd<9iEPd*UWZ42&aIsRdAKidim#i2pDiEU>m6>stXjtjcoxGgUC
zy0$jCdfv3n87SXaekZxx!_C=JKIkaeZ`l?K3M@I1A#2dK!YwJ$->r!igbNAN%F%S&
zwqOqkc}Jz!T4T3Yt)ylx<;Cod#O}RKTug&~##kw^xkc+t?FaW@L;cx651eLA2Z@(B
zaECc$nDDA1j>BVGNE*1Xx#yz?!;GjpEj|rp`9@Y?5pKb$tBBT7+k#{5YN%^!Et!R(
ziZRizj!-$Gp&8^~Wv(2X)2M4>;l@ReI>9}g6m=@Rg=io5y9v?Le(c62Vpq~jv2joI
zc0`X+?!qQTPn?UKjFLs)w!5Myq#e$|DOC%(_!n!iP1}|dYHAEiii_zv(Q@D6QNu)M
zRH!OabTN@>d%k8fw*|Ld(y4Id&+^5o=?3X@;xj2+2=oVMaKHN7<|g4_Rc&ruW%FXt
zl7yhJtw(G@wrQ^^x#|}tsmKzG#Shjg<|wb5gLZ%RAsYbWs07%eYAs}i5r6nHYaz?%
z;@m}~i-L|GsAgS?W5_ILwcCpLvaE9nw^DtQHc3a|m^FN3&K=u^Dt%g0wqe&`R@LRQ
zH2Rfbs6^&{=-OIxV#@!@mDn0>oc9%i$ZD9&M@rZO*0HT{q1~&k%h8KIpv#s!d;>k&
zdMmG(sgz1O%ls@{?Oibq^g_0#!jb|pi@ws<R+D#YB#Cb9A;%BL%2G~eTj!k&^gP%=
z;fD3TssH|)f3pqwEm|Js`cZ7mMhADsJlm9?p+4-e&Hk%cdOK`OK%9)A!cdD1*0y>(
zez8`zE&CuhcBgIkU(KzBb=lz~O}W>0W^YJHuX1d>uG|We!E9H&QQt1hkWjB=8G8O#
zh<BIK7ZR!UZR7?@lB>Z@fxu(@tK$>Q8n7>g|F*qVH*b1A3%JJKm;!!3C>suVi*h2B
zyA=lUPzBZ-gaOFz2-xLTIP7bTh8Hzb^dqDEQR_<rlb2cxmYFU)>hRWK4tdayB}I~=
zS`~Qhc#_(DkGDG!-QlOoj_D7;C4DON<Sjrwm0fTnqea0y7L62^vrnR+6hrx8h<m5F
zGS5`kmN;Y?HA}MX;-mH=%CP|_sJ4<U5HIU$oKog&nVpM1IlfSy!`Gu+e$VNQ<eK(=
zyxvdTiDDm5AYGjg6l%2vf5}W%EPx;QhlJeOhmOt3{AV9!Cz$X=tP;^a_V+mEjkXI3
zl<qxk@ry|fwcKByR1iwn&JD;J@u?#ek8T3@>@#mMpv-u<H>{Kz_w)D)^n#ai1lbok
z&SWK*<aGJ+3hI^bhRo064Dm3^CC5*Om`efmb$-o$#_zn-F6CLqZmn&wZ*rCdI1_$)
z=5@T7zpYRuPxz8Fh3!SMRr@aANUa8gCHEpfIKKC})q0|U9~`A5q_rP&06IXOTDtiu
z@4~t^GNPnw{?Ca&g?|2$KWY&}?CD@Krpb6yZ&y7XTn#rGI-RSZ4!)qJrn?oHs@|@7
zI(RC)Hs!sj5Q1x;jw`G_Dl523?YeAQ^7){L00GxOolJ}Z>7_)|T_g<O5Ur)<$nw`W
z=J3R`H`J3^&u)6!xiTHnit;YJX4w+btsStNpN@-gMMDd#PbS0|&@Czo@l6HApy<kF
zSDhhJ{Yd(k)`r^bPX{jvPefLsrwao$8=T!9kczY5hU0WRXm?b@=ysLx7HV!A{B$sv
zrFL~F=!WFPyl@~XDu*WElxbbTJ!ZqIl|g^TQVf5(T!RBW|BkRRB44&5AJw8Zdb-?9
zcTcua0UREpC_;6lykTP!IOZkicqLeO+8P_Jb?zm-jeGj6N|bTx;Ea^OlM=BFN39}S
zO~{W1_LSAb$IT}CGdWFGmXk@*u3V)cRcV6n@>$?X&22x$XMr{B^t(z-IxRkc353N`
zu=_rn?lU9d@r*<Od0=g30#>l<W;mPWkPr0SJ<aC$>ai_OLw0j3kkbOc-{$4|VhL@v
z1<}G&@hbg1c@lF1B3&>>(LK8_*J&!XwRD@_?}RZHW$&WgJoBjC)MbkuoYB7@Ki(e5
zS2gubFsHTu1s)1sT79mmvMkc3F{j9p!*4=Y@U?47phy4yccW}+1)zMU%EDSpqG?Lo
z3Tv&#@dO~s#n2wl+<Yi++CJ+D3SmQWyY&8p(i1@;T|S8pIr}aT${-)9t@OXym&<H*
z;-siiI3(7@5UOk?Aqx@Q)>gy1*NJ@pAId31USiYMRf|TR2V3t$u$-N?A-<OHu=o>2
z7H-T@;M^sm5G`e!5<hq+x7z0X)!u?RdW)|Cxf>dEr*DnVb%?wIkO#RPw|SGP-C!Je
zw)@(W3(I!ojx9V0I&ItOv-bR3stC6$km#ILDI^9Gu|eDOx%v&uYH9dFj9hg09Aa3$
ztmo-U?e0<H#f@^eQ-4E~?Y#n8;s+5Y4p?usYF;Gy6ue7u8?*yS1dXi5p&T9M2t4HB
zTrId0azw-2B%W7yTVFNJOsao0arkGcVvb!NlJnb7KtxFNsmFcDRjue#?L<K946qjt
z>i1!A!RfG5{`b4<Q5>9(E@kbO#_Mvd>N^#l6r$;7-!uN@Y-K|p@a4e@X=igl4sc~_
z!Z*gwCFHY|LFK*t#Zhz6-jDWPdrh7&hwXd-0*w_}QLg9ZTCfkQ-o?AhKJc1-n3y6$
zP`Xe8<MRnA1UtLM6@KiPS9T9&oxc#>6f+P6OZj2D=+9&X;Ba+y<0lnx(owiG->30R
zC$+@XQ(T#!#WOhqc~+W3_Ia+Zl&Q@GjDFz=YwK!TX<s@HrmKn{HP&<YwV?Y%iS~7L
zS4ufUVtZJeOVxVIOWG^K7OSOwlRzpbDwC0C>$g7Eco`qR%lBw~R927bdG@`(ALv=M
z#(t=JzI>Cr5&p*{-g}lUv!9Z9$(Y;SZa@1Brco~wh<~XNAp50nB-<V=OQ;7id2LUY
zrNnzG(T-ic-4)^@D(-7e6|F12+K}0avhGs6UK3w>qIbVCuifq(jI33!F)vj@_Y}m}
z<v7e$G<XCV?E30iseZ!M3#Sv$q;XJur|hUVMz6c^Ykm7&u4y3A->2=S?V*RFofvYt
z{gpCCz}wIJo3|InWF8y+mhHj3)exN3WrJdzqCf{m>YLlGi5*nIT%8<7w{3T`9h43^
z-d@2=r8UNRg`IuJ_QDVh{`+5t+29=F-i{@lfkU<zri8@7Fy=$6Z;zH}Fl7oG=Fru1
zf)3vvj1WV&X6B0#F>gtQQbwu1vx?0fw~^VK>{rZHIaEgZ=t|Pm4d`R4Ch|ZjsnTrh
z_QKG@pv%XV0;JJKM-_Z^v1pC2p2_1vTfEOE_)Hq8MzX{v<{HcN02U4x+pT;6Cpj(#
zdM5qvc9(yGw{^{fHYH&}X&P1)o3=glaLRbh(aRD|4{RenVNaP6AITOjU5PG&nK8D9
zu{cJY@@D1O^P6o>;_b2)zEIKun_In+#aT8$ao9>OGq0Mqte54gw*}h^<2YloE%cRY
zt*x>9a$*Zv;``$(&lv2~x+tD0%{w#+wNjcQDdP_?fUWjGfWa{TQ25DF7~*0ov05MK
z4|gxHdjFHy4r|<A7<X)3kr%ruXZ!a>w$v9DGuJXQ40T#d{w@htQQahrux;}E)!uCn
zOq2Uh3{Hg=%&<UA1$n4V>`+&HD=vjIpyan(UbT*MrQ}ePqI6}nPI3~Zu69T%6i3na
z_%e)ZSGcpRb)>r6-9t$k^P9<XTcbqj>l^{N9FX!JFH2kR-L}hXe1lJhbWNQ^=55UB
zV*^pe#Z58j6={;#Fp6N_?2A%^z^Wa(#k-jgj$s7028QG&^|o!RI&H&Y4c;eDiq0Ap
zT3}b~@YVVw`Qdgt+1&f@w#%obgr0Z?Ra-o?a~kvoZ&5?#`Wox0@K=~Fw%&`jJD$m+
z^eTXpCC~OcG!!#+$a|xWD`+I^9<l@RtOyElzX%O>C`ni5@Wa($@-0G%#kGAT8+5h7
z?NlSn(&KA?G?3|xi1v?rjr7^Eq>A7{o!t@*9gluAg!1vdg?Ge>3V*Wq34i^LZ=n<R
z{#4TU{MmN8!rNu3-QLOJBx3`;)Z?y7JLAKNXMmZ76M~x^$nCd+lnt$RHX%}2CG`FV
z#BeT0K|cW*Qnd-7-iwAptXzN&p^JIHx^mbe7(Mn#iN|~GVAan%AK$o4T@EVyz$e#2
zwgZUyL;r<euUtYuiuO^ui8J`9eVjB)k2vlYrIEYZU#LI_XppJ6(2I%8$@Oj*vrn7?
z2mjSR^%cd41b_NjMbob^<IiId*W~Qyet%h~1ESp*Nd_x`@5}5R#9-m$$ybScxq|4Z
z<P0kZpnaWlm)@N_=LiOZOV!p<DX=B>O|%gS$?T;Z<lpA{{29F~38C{RHd>OKe($sS
z<6Qe8`!7sHadNjG6C?0>(*3De!B15?Z5mg8{G88%fmNN+U*ee#EM?<5$@<ze<<S4r
ze7owIP?tcF#nIyGXPj5^Bt)*7y5^Z+EB>Tg)vkS}FdToJwZN`>CR8NYu*n7eOz4t8
zBdcULJmZ{_h)O=W`yKn@cCsa^vo_LW-1JQFY~-M+p<oz(($9oABsULNV?kE81au&l
z%UtKhaT}CVsPCvD<a~+U>hRrj_k6o8X9Lftuw&WiZ-1t+Ay+sH-Fc_o5e=2{R8{FI
zXK#a_DQwR`&)`Kiq?(HMqPrLCXuuUZ-UuYjSC@S%Y*_S(&7qi$$joHJ^BH?et8of3
zWkmF#L+v%5VV#I8Y-F^Sz#*;FG0f|u;#pxifkc_5##HT;@I&^&12#5$Ej|<nVI`61
zbsgEbXbr2yg(pqp9vhzvNdn9+n~*J#GUd@J)9=LU*~{fLB4e{n%4I96#u@@P?|P>2
zgIr;FIZmm-QtYL;s5fm|K670+7Na+9dQ!u3^1z|)pW%pq_y2@vndyr;ni^@d;#19n
zlL({29-EV&5<X14xXn#s=v>xTEEDr`5$Dgf1<^rK8Eea6kz1IwT`c={pTj@UQ_Hn_
ze|ApE-wssBiyTQ(|H6DOPLvfR{R0(&9Ac@8e_Wh2wbhJIz}b7`P<{5SveS@kV?qJm
z2oh*|rgAW^YiLIsgwwS5Qj&oxJ*IpDTe4@}nBZ5YiB^Y-u_07zPjn#ep{XHQEIui$
z4<{Na4y`NaJoVl`+VY$g1)Nsc%4b5ciy5sOSQ%E}t`4N`MdVu(3p8eot<6~>c`YS{
z!9x`Z%4NrgRPf6oTbHY{=uTUoU(s_1Jj8|o5!u6IWm%<-33fR{bP4XUP1(oqerKDj
zZ(N_Oj5%ALDf#|n1R$koc#=UPTXTj@I)}H#m)gF5Joei5Tnz=Kt+E{nAl7xM`FAE#
znI>NG4Yn(LESZU)uPEWL;)>af>FWGKeuxUGR8H=xz?S~T*^k}T=U%i!qLgbM_vSFu
zWPRw3ZzDhx#2mK+ey%s7^6|fxFSEnGK5b;wIpXsoQ+OMtU#u^BC`NR)gF@Pl=G5=S
zv%B4nRS=W5Ujf6%6RE`k9z9?K&O{u{Cmc2i{}^!eJ4Mx@SHRc#nbg?30H$kGK%dk_
zhAZ9a0OKH6lVYxss_mTwqpo9BdzYQ@CbHX6#!PXQoi>oznA^e{o%LBZQU>>2)gLS{
zBT4dkFTj&c{)D}spBC4+^Lf*VPB?&wG=lm5AbKet1HOv(Va!CAz9PGeBO0`*`cHT?
z+eZ$>GCa*b&K(SmENQK2<!HQ+bGlnzS#~i7z#Lx`D@x<`NupcT)4G`Cr$G!FQMtA}
zGFC}`KJ(t1xf?|dpXZe3AZpqdj(>DcR1Equ`Y3!4%M8P!p5RcwdgkoJ|E8OYeVy2&
z7;h8SWj~oNRoHWXIre~k<NI|g22=Ys-b+1LqUh1)epfwv30*e(k<0G;ykKO=D@jIv
za2zjaSZP1za-gU>q7S&?bje8Jtb)uxMTg>iq}IVV(SA<A;6at_>@Du(U#?iII_@pV
z*><>2WFL8YUb3rp1Yc7FD}p2I)jOO=V$0x6;4XOA>~JdAHp23w3@lKMa$<;Cc-5}m
z5j<3P$$EEPbm4)fTG(TD{f@$wjr=#eA%@@ok99a5-{`*vdKSpGdDD*I*f^YEHXH2b
z9l@r|sIgmel%}>;q8tXrC?1O-H?5cK)_7L9QeA?e`L^nra(%ftfBTNOgKF{0#n}R^
z+~FgQ9)osoaDs79<Wm|FeTdT_{-*dP4z0#7I$b#~+przM6qb!sRL88HDqs$e*5b1X
z!GLKzY9kUdEE2DS$``TajEo-CLe^u&)9S+AXQSdx@ro*>POcT`;*gDT(C_)fj0H9}
z`Xsm!<%L_H$K?`}%0@iy_~^+mEk6WYdoiK<TB;%UznYl1B5088m>r}wob;;>_CcZ$
z=NaYLzsr}i1;L-oWXg`hOgD3IO-mTMs4sx|Xu7{zK;+{L2Oq|`%Vq`uLUEovvywz8
z>q!|5ZxQBiTyBZboEu+3QbnLqQr>w9c8!9&ifM0w4+O{2U<-EyQ_hU)sRV<g&yFxb
zcDJsu`@QKjrfiEG)H(Mqu*I<vih~quaIBJDi))uX5bs5Q=`b!{R?hcI&|I6N3Z{}4
zyFS)VX&V|7M^ilkeN%<z;~-cY4&_eq-dI|#(;ls|BqD2xXU+f;42w9>>b<{v7*4!B
z-b=3afT@1#sE~v@eK}_>p=RE!xRJ;ON%0ZOt507kg!%?I+sZ^21(R%bjPoSI=bGH6
zvx=Q?tz+*ue;;BGRS+t7g_x3lTUSxFPB=-*B*ZSNtM&25;brHE&EkN6)i&e^T{SDY
zM>gheReg>INMut$)5E#5If;k30E}b-flg8}l`p|+ZS@vtA0C27*yi6Bh0=BheKtnG
z9o53ad-}%UUr%Uv3PEXSH8tWDc_SU&t{j@Yi7obg3`1ou27Mt3fvBb>4Z?d*wb;tv
zm3biKMYh{9{f7qS0@<6LHZ}9Bx0(;`u|Z=8;+fYwFCqPQ$e*EpYr<6GTDrsD?bN1b
zJR^<-bPx}>SznSxCwgY*fsQ&ZY96eyW7Wc_Zuu3F>UP`@`8o6KM1ncH1_`)7`@@{y
z?27qQ)uVR2buiJ8Vle=h+mk$PUK6hr&v0^O`TdYO^D9hAX(I#Y5d^(~gbe9Nn$pXa
zme1yJQE`$(<6L$FYle;i9j>w~sfg`+`KCyh$y#t7+WWafXTtp9&wSpeDV-+F_Xm!n
z(PPHihe?wp1SsXpKB@=^!wD2lF*N{`Db6EjKCULOL>Fbcb|EJ(lN=1C$`f65`W)!_
z{doH%kvt#5w@>rUOcX`cm-6^<w0-XDjY%8n?~9nC7z^q2PD=@<P*MAGN9ZQf2eD{Q
zzKRYDWpnWLqYeH#TPn!EV;SDKV%9GCFssm+5K8YqsE>&T<=aG_9A?qteOIBmGJvuG
zC<^TR7_DeRxmhS>YClvE6txo&?6)6xgwdrzK{$2p!0}Tw^tdF~i~bzXgu`QCU&7=s
zj=g$zC)8li7N$tCJZ#3Ro;_O<=)~Gz!RlQ7Y?|9t%vDCJUc2Vm;96>0y&n9v(Wvlw
zLO+byb=9*t{IHN*AI~yDin2o7@ND6D(ABN78=no{Y<6=!F2;6KwD(k_<zKrwpJCZ6
zbsb@OZm9rjN4d=rImkQ3B27xh?35^ABm%q5XU^Hs(!IoPufTLPb9dPtIRuA;>U$i=
z6=CqR!SXGja&DIm$xiQ9Nw@|Z8lUL!R%)^%?B!w4hOO_4_$_hlN*Q;Eqj&hTg+*L2
z%|>|V#ZtIun2n4N-QUmkQ^ohE?4Va{R1QwP%Wi73jLE?z2++C8VvUVwMNL))JYeH|
znBr5-9L7fzZA&SOr9qjHn4I6<P)eRpe73MOqklWzCOumi29$yL*WTqAE?KFjL`Rp4
zV)ofIpWLFwHa%!du3+p<5WpEwZoC=368w4tZhqz!5K9*;XYg4*dd=$9HYZ+tsEnW`
zz&@c>ba@jVGV=m+SH$Q+jxw_9*nZU(<SI5StHtqjVFC`ylo|y**nR%&o=ERsF2?&~
z3R>O6fP1;%%cAOys}ki}B$Y0{(9w6f_O&NqiXVt)uDdRPS#5$ud?wZJ>T^mY!)2f1
zI27OP#<B8stjPRpP4P^maTHYT=5$(`WKG88Mr-l48tC~0nFd?^mGU!jZnpc%&6c-w
zhj%11m4XY(WV;+7tV)H6T%J7<-eN0bx;kNHh!poy(oxvz1PRiXmQnovntWD7-CokA
z?X$Jd1|K?=5(UaPVh=^DUP=oRF^`JnxUCDyAUz)5l<Th;Rl!-8h29Mg9NDT6(>Eq2
z>qu6$$~NWP&?d4^%G@?PJt-sc7AMfvO|~_<51ke#@-YVCPPT2{BhKtajj+9fRAIBy
z(vwQaySaP1?R1zsJb&`8m;s7Bq@3sDOYH$kK5}?^;n_+i!<bxhIQ2v`?MT_qj$oeI
zoy|x+-;E-Z(|WJ3Am<m+VDX7alNG^^(Q5}{7ZjZ~XGSCpDAo&|4TrMJnr>J~DTC~=
zPvb9x?TB}Xya7FSqLN_1z8C~ovJzw>E7{Q`RF839Wyc&UOi@%3^VzGs1%|m3{^s2+
z?DBs9jLEbR5Q)z2)UU8#L|9M@V`8UsJ-a(<F<rkCC()SS+Zl(pzQqQr=_?6=ms>m=
z&ve(z3G^sh>2v-uCCcr+oW0~bDYfu}&iP5$&>@Td=X{RN*!7fd_a?*zlrlYqN%%0E
zkbw(F@sE6(co1~j$5oGqw30C1`|U!I1R^yX;bQ!fXIY<pl1F-N1B<fCK6S!IMxl<$
zXEFAs%LQfa^JsuY4zHlVp{{R<UY@aC(J^2C3cCnS#e+|q^Q&B<84?t}&hb|?)Zm+F
zmmFXB%$a82__PSYA~jHAL2wmYF4=(+3$hm0+xM|0<zQ?-_>X~}mCIV}$ACpdDvz3z
zQret3$^HFPc7YieUFOe$JneqvP9krW{qn1cXz!^Qsp2JVXL2Z>+(|l+tKyk-5pwxZ
zIm@ee#>vAeNKRx}w`+E~V(0+3(H^^Yr}IEX;)HC-Wm4|jH|)BdAt?`8^f$YHXYeSH
zc3XR=-4LJYf>uc-;)M8E0mnD)EG$DugCC?f1x(0d#pnMx%X9P2!U8ll+burETy#AR
zHmI7?9+X|)+pX2JeXzKy!`f%J<&?rtm&}SggEhk)kdvsx?#L!Jt*zK9;rs^2thHI*
zvLOy6d>~4JdK>Bhwbj_Lot4?*Kf0TXrC_*!S=&aG<T4wP0Kr{~rS9jj8tE;F=ava^
zRG@~{V#ZfubWB1eHEs7Y;2gWtUATlm<$08)_VR%&OPb>C$3>5Mrd3L48=n&=Bv)n@
zo8X|dmdfciu>zI?ZtM`%hMMG4*CB<W%<jr2a8XJ~1tVuu5(y<?PSKl9^RK4L;|e{`
z^!Q2^Ea@ucVl|_hw_L(>49RZHy=XIc7GdL8=E^&5R`d~J7{y^~PSV`y(W7i`{%CJ#
zv3Uvhlv3AtK~9|D+G4L>nDdaOnsFso-j|KYd$UXx3h&PaK)>xi8H;v0w<=Y^VXomy
z+ZoK-;tKW3)1(kW`8X-f&I5^pVnpZk;liuUT|&_!u1~A4+Q=k}vog19qpuM569p4C
z?K~?-brpNHrT#a**5j-tzH}~Cer+eLHKzmluy_<d&FM_$JX|H=x+76G9YXB#{_wVM
zwdK)Y${=J+x0UhCrEWzqke9l8Ch~?HP(rVXX9u}cbTBDb$=bkM5~~zpZYPqb-hO*1
zL2Fp4D~}U$U4=<0QZ*yIy=7{)_3=h(%o1(KIVnBk5!;Z^l>7`Z_277;&#JSV@_?Ii
zx^uDkZjQ#jmprxF7H`Z6fU8gpQy5;+XYuKhiE&%?rDrb^1F}7VD&7-Usy}w@EV982
zw(rjBL%)G=hPGXPmi%G9J?|^s?J4A6*co=n63`=)N4ZUVVw_;FL!}`U@Ah#<|8|V+
z&6S+;XX~v1E0&Z{L5nvzkl2Ekle3LnCk`detU&LobZUq5tWrpk9m!uaR8OEU7FE)d
z^k$0C^nI+8=a3!E0m`7|M7LuV3f#F?W|HIC>xw35@<alBg|@U9gRDP+qhLvwG)SkS
zX$jTk=#)altzW0BPNao)zw{!u+dBaO&P6CmbveFmXA;n&dJ$XxnhhjHc6L(;ReSHO
zgNfd{jha#Ca_^GKuL-U8d(nsdg$d*UC9m20@k~*)%Ag5PW9PHqWn^|kD28eBg=;|J
z_F*pEA7<G{`7!SFsIs`rkK4yFEg9eBnykP@-xL=TH<;0oUpnn#9)K7RF>8O4q>nw@
zK8>#B6DSRjQjLF>?Yr<!A)lg8mQC6R7yd;xS<2tBIls&mn<o#1ul!Sv8GK;s?Cb2Q
zZWVpmrEJ?Iy}~y>v=sHk8@Md>ZOqCK2agcrSYTnk3qYk{QYBJ}jNd2l14^k>m^J&M
z8beuEeoPdX+z>*>Px%d|R(TVl@3Eg_hNxZ<Xn#qd|AdfCjM#Ib$moDAS!P!~7u=If
ztI)Vu->c)9=)RO7grmCVxo{4ejj`LV^<Th9f>66I+Lt&6ZfZG)U;kV=&IfvaH{Ncj
z-m82r<taE-ZuIxa1V`+q=Yqewn=t>IpG&QY{2)YMP=3*%$eHG4yXCpUe-UPMlK-Eu
zK>^Bs0`oA4<*-B6*lpD`M7Q#&IAFKu?CYASeQ0-N6XIr0=%|&a&~wfQ`kFZhSy*Zt
z5<`(nk=V?k33W?btqn``bYRrB;nA%dOiD*KqU!b~)DfiwPu#2{6XthcvQdGS9&Vd4
z(Thk#k&9v=#`>TQC7s^5Y8Cd<e0&AnYp5C~JQs|h(z>Y}(?svNyPe3$N&YuE)nI#a
z8q;_&C4ejnPf>&PwMADmEhk7xmj;_2pkeLi?kNA-jCdx_NkPy$|7JcHOr*+Sxd=Wh
z)>r3<n4|u3eq2s}GAFyi`5+6}%#CNBkU;$KL7VqnxqSY4#6R1D{KSj37W?bM9C=wY
z>i;7A-xtkv<m&e3P`CR7X(*^U))zT-pt3KOKo%#lAOhmZ^FVglO5Sv`L|{a%b>zX~
zlh&xdT47l<g6-3#X^l>^13fan@mW)>#WLiFr2!SLTT+iHRMc8>LVV5X3Wv~dt^Um{
zg0!u@daoD=@Bw9XJQv~(<(kQtffcLEfh*<@TOJct?ZFpqWxV1Xqh^?6a{hvOwkA6+
zUaHcf9kI3ftY~pyk<iFI<e#Xr@4D!!T{`*o&lP#*k5g?!d?n%)E3^B|Dccy&qy>SX
zvGN|WP0wYtDK}p!K2n&0&51GPF=<|ATPh$#HT${Sj@Z_~w>pX(TCA#gu5C%IxC(XH
z_UhZcMQMe8$9DMA{<RdU+iN?kCc719&2e+mc2yvFLh9$EjZOp!dM8mpzK{bEKC#4l
zsum^W2u|Uk?T%-WzEq;G?ak4U7L+<quIRmi#*TVa9S5>O;i#2Y?NvMEsBLSu!_~am
zWaU0S3OK5mawOlt_b8ZJSBtyh_Br*Ct+&OFCZ$a2CQ?ZE|FQT^7C!G$8S&*=d;GZ~
zuoSh2=Z|A~>_l#iMtoA5tUu2hR~b=FEa<z7`{s1Mq{JP%`FeXN1}h1mb*#y*AbPVi
zIc@kO7#Qh`ZNM>%lZ`y*&nD<-C*6E|F20d%lN5{Kk_h+QsFS(y>WdcheTN8%;`eqw
zp<j+6lJw8nCn+)1w$naLkSRliw>Pn1`9*tY1Fmo%S8!tEF59oyE>zUws=SL;8~J<{
zwZDCmu(GYCBv7RIr~V9okW6>7%07#B&Y3I4$IJG)PoSZd#HC+kKkUz=B6%DUzD!E|
zg9^rf6)lRd87&mp|8?Nq?={@+irr0kzZ9)y5R=!9aQIF2OahIM@ommgRRsu1`p(gU
zGeM!E@ADIR1wubW!PpND7WW;?Y(FNDWcsPMp8_DIVv~S;9Le0d)}MW{XuFTwFF9zb
z>x1EHmn&`LQTU==6~AzjMex3QS7ADu>k$dB*;S6ByBFHEyRN*@rByA4ESQ?>{A(OU
zyQx)u{jSjLv?~({nw?#RQ|crQi7aL}?kX3~LP+MOU2)dVNAJwJM}>@C!NQFG?I^p&
zTaNziI2*Lf*(+`vcBF@M-Wty&nxhQXzjoXIMcw;{cwKM%{)>o+h=_=Yi0FDfUGwYq
zsJY$db^Fbq+dO6-b-Ozrx5xSEjEysM=ZC4g!t3rvH$uD-T`99^9LI4S$8j=$WfnF`
zL_|bHL_|bHL_|bHM8x;;e66+Kqvw0?<@@h<pVQ#y{eHb)uk~85^;)m>T0g6H6mbW4
zuOeXFme1bWRlK|uaesUDE$+r(RSbGUj@?wV#U{pNtJ{jsV%B=p?#N~}%wSUVa~QW7
zPnSzFL&sQCr%mw*+<E7nHr2n{-m(?GG%fK(LbL~M#&aR56>cNf9VOeS3<EYZVdW|(
zSO+;_XT^FcMswI^`>*@EyYL6IInmxFgHm?M=Ek$)>JF|HaW^lfsCK%UQ+_(Q)jiLJ
z(6fb<i&8Ex@F}%-+oI@5c#%t&M2ICmx3;2tU+T28=3iDWwq?<{a9&1`mScG=>fO_9
zg@avNau(W3{}9&lds}rmsi6vy5uZ>bTN7`J*g=;HTyn{3t$*LGycQ1xdL)jVvvp2@
zSounX&-%m#I95tI-E=o3B|g;L)g+&J+vp!r<5w19tBsEehb`ltw0fUPH~PNDcwNkG
zyjGN7Q+~qfqw->>5GOb1*qpVfxsc5cSW6NV=_;+YYCuc^_~puK?Nvvz%0twMF&(+d
zN^T=mGl3V0x8)+qiReeq)=S<<U3{EnmKx@iZHskN$g}N;LDc<e1}1-{J)DEAEjbA7
zQ740?3vFk<na{@M0<g=G-=TByu>==N6^Y0-;_(1Nd`Y1KqGZ2^KjBLS?`XpW`(%Dw
z6apf$bM{m`a~C+L-QE6-^0r-Ds8<2+pzNu`K$O^adpg&%9*t}9d4DF+P`E0)rtnH7
zf_OIGXf#E=sv41Gp>wu3zmXD?)l%HPb_=yEa)*?lF_`eaB$u|jI%4tm$CtX4A7>$?
zy7Vy1p2VCCMv!W6%&wm{)Llhmb;A1MYd>PNjK%E<i2CD=--vu)IRy_S<~Zgg6soA#
zVAc0260UNRu%<%^8HB<*vw@tXPTO#_2HL{QMv&`As%H{E-Q{{Tr^@Y|3(*PR1<h0q
zIq56VP~XsKr(+zSyOdEU*aJ!|ch;$B8}g0!u8^&`>Eq4geLU3A@SwesoDrOZe<e=m
z?>J|?5M$`7q}#2hF_8o}mNSQG=sGUHs*BZJilUB<aNOQ@;JJ6$J3cTTGJmjleTwKS
z6q9T3RdZ1s{UxQ=?ER!(d~dt#gZ#A+m9l*p;|a%7DM+{Tk*_{C?$?O8KWZOWZKO7E
z3&kf6$h|n|+owq_Ej15q=e+qW=f(9yci9p8_&lD4!(h2Ee&O>{ZtO3;JNW7!?5kWl
z?sCYhU;8U*W$c@nqn{AY1$2H6*|z~RQ7LaKGUdCRWoKJmvwdGdB>I&te#Cyr7Vv}Z
z_G7ltJ769_HS$vhGx67aHN-jnbIuiqAQWi!i$k|Xdc0r#-_4uB!Ee>x+t$>KUzpkR
zMd0vsARzt<|4q_KQXKHXmCuK&hV_J%U-i7p5;CXA)GP0qtDi4I2}y12ns_aBr`*fM
z2(NwqV)5%KiZGT^s%AgVuFJ7HxsxL7T>pF#tBCbixc?2$$3aR#uTuHw#^>W6u<(-j
z3%e<X)P7d&2I*&RPB2g&D0jEi0Jr2sO5n8JnoS6s7U~I&x7(f%O@W-N{Jgu}n_)~U
zHHo<&cG`sa6kl%cHgftrY7?I?Qr}q5z4zH2&ljoh-Un<_v`lyDS2)8b=cFG*X2X;+
z#V1Yavxe4go9g{E@;{rFecr##W+XCJG;~pb^Ff<gVYN7Y6lJ|LvsvEZn|on8du?_O
zCnF09B<4IHLSh|xa_3g?oh#I(3-Y|@L$1U=>0I`n7|J=Sk`jA}`$=1nP*Jw2O|s6S
zL^A$v5YLhvyq0h$WV)pe8u3nYAy}4;J5`<AmgjW;Fw<7#O5i)m2_=8nl@-=qNhll4
zs_cQ%k+{;Vi6#(%ymmPT58B#zCSPHZf~>F)IKeJzXt#9^5tj#CL2P}6)Z(^Y3PQRd
zupyADI=iE`(dQ*DQ+%OoeGqtYg90r|?u+__t)&o6W59u19S4&B9<Zir#<Og5yf0B`
zI&NTiEeSiDt;CGtH_;l;oH6njhTZP{bjh2)!vP)ZSuJ0rF8@W19WBS!1gVz&)MVRy
z7E&`Q=*D(DAJ#VzN5X+}mLEcFm&242y*-?iLH+qwqT@R3(E#QU(KlQ$OBLFk&;Rdg
z+JSmbS1;PKD;Ky`mRE%X*kegd*aH+bVUK6;2+GLr*t>@9i6q|z@UAC)-gi!(^gDa%
zaw7Fe#4*xI*>2yC+qc`E{P<6>s;B+qRr9Ii`%I36epUul<@<iNq6KyZyIk=X<vg~%
zxdg~LI#~&$w&$FrsFPv)avr#yP@AK{_E+s?P%a5<tS9K{knUC1n}FfeYJD-VJJ}YA
zmEnic*Z1ce*sdHp29ZWXfXP)SLZ{0H6S>W;g~=mh4&?}onEG%6MBL+9`cWpzkqQNh
zo|bdit2UbL5PnVu8$ID@BXSQ9o1`bn(U_K=56@7pqu6(G>0wpRCN{QU`PZ&LSK(ER
zD>bmlcy_)5LE33o(Z7)(Ks%?wzp5l%s3zOp(r9Bovjr_}xW8P?E)Y=iZu54aTDFXS
zdnXSO8UE0ix?1+#m{mItpzme(guxTGH9kkh*nbevE1${%_m%X&_F)XII7!K(uj8-(
zQT2@EIxONjLTf+H8KDDkiFnjL$<_G_k!PP)A1Ui3B!`mVGbgV{)>F?k@uzA7EI*l&
zLbm)OdoCGdTps!|_iPg;#bTQO%BQ!$FOFYVgSihqjt}-tj3CLc#7z|&pYEUECNdgZ
zq;~x-LDAhFIvP*k{e4ad8<G7GW5~r)Tw<6z<B!=>9p^@!{p5Re>2muy=|$_KyllV3
z*8_MY%avddY3JAMUS(T;i}jTX=bYI-mo~&f<%UF&h1erk>?>kQGl%Du`&^3g+kx-M
ztKymGoGdm^yE;0P;#ATO#fl3LIBM5;Xa8{LRJ(RxsB9=z)`t1-x)_IzcKtXotdh7#
zB+0iM_7&lVswt`>!#=rjU#gTS5okC0FBH16uid;aHWo6N96riA-V*Q<a$DSx&Y~{9
zbzeE$OT>`fmT(aQth4!c@AJ>?q|LwsXI}^`By*h@U<!k;vLz+jIbXTMTWvyDW|Q_^
zEa}@_dmuSZZl6^>q|4TSvVXUw0fo(!=w07Q#ij(d&Pb=moCg^96`Qs%q%ZOn*3@$m
z?6Mh-%EsnxHq-w({jR%gR?J+*(ND3{PC~P@_l20crDhXG(m7QVpOM1!=T`V=C~ijc
za={@;$*SiRN)@n*-2-d+>ld#D@g{{JPvnVi#TH#YKMWz6>2Ars2pF!391}7EkxqE2
zkF0PX1PIJH$YskM7bK*?;%&<VHmqMIUVadB!HR@eL0{SkaOjneXYv*|*s6G6^h$7W
zsj1dDnQ_&p)|O<~wTUM@K{zQM2-rj2<sx)Z>a25kwvfaV;cI;bk4V@xu^YUF<Q=r)
z8~<GkmF=!|=<e6Wp}vAn2dQ*3uUTU};}As1(AGX_P5X-Q{I^NAIcP}K+7d1M`A&zi
zaul@YdM@y~u<f~a#j`>9QM8BRojUw$S;eUx<>9zpiEnAPY-^&Wt~l9lOTf_Qa7Mw{
zUbP**w^|x7KH9_4Ae1rq=%2AiD>@kAV7t^QVrR4#)yo-F=D1xsVoS+^{aAvCk)c^C
z8Cb&O(V;%m-Oumo$05@vE*V}@iDcm?bNYn<<=?3Al@naRp7NpE<aULax7+c$qoJeC
z_T--9xbE&KIc1)%fPJ0wrHG^UjL*Gmdpk$qvk58|KAfHrZ@1Xq_=r8Ov^RM54cT+P
zl8tq?uj<mhd{ppnzvCZ1Gm1dn6Tn~UGw+Ir-uRZpQ1TE$opZ+eD#W0!V3NIW$+Rwq
z(*9_j@d_&FIN+RvGT1g4gHlEgP0>(dOuklPdnjeyJ7str;35ycju<%--%8gsWTQ!Z
z;n02}2S(%+DYII3vU=v`7iIf8oy*?d-GYdF))A}ACjK_QF!=7SvGcick+nM-NPY4~
z0Gs?rMl$V~!*(G*yKj+=W%mf^9h>Z85;?krW_vqlfSKx;y|XXwd39tRCHVs$$i85I
zT50cjm$DSw`!Neq!mH4lEOH-Q28{d&&Vo6GNPn1k;R-F?!$<kM)ib(%>=RzltpK=B
z5<KL`P4;QhV$xNeeHIYQ31HMd_ix8~Rv{&Qu`d<>2c&RAR1r@c-y&c76bOiFu&-hY
za``JFCsgrucDA`~yM2=@C4~i)puDo*#xv#fJHxZ%cy@f39dUh=5<{5)zE7~X7oV<r
zVEZ9plg&$BsYjsCaHm7)BWYnGWi1AL6s|uxg|6mK_H#nQr|9c|=Y;)|Xj;>Napl**
z33?8Epni*Az0{50W%hy#KVs+P$ANEUYT_$iDB=a0;s)+*SH9r#PtnwZJOvzG^+I91
zwe5EG3vrfiBGR|ZuBjTIWt$WpX4mF3$2ce&u<PO(X&$vxd2`0De<99v=e!a_Zg5DT
z7w(L9;|sowF2|OPtGpQxmbvMLBBrb;9?3B~fuPEKIKlZ}yCuN^y~+zDqvowSG(rgF
zCpGkKFNCBfhg^l9zfdH?`D>Qigq-t298#!0;cseciK==b)KHBC@#zXMDN)&sIA)V$
z0>unguyYQcCZSwj4wQTM37{FBX;WhYzK2166{s^UhsU3r(1jT<gpy{pB%hd^h=FrO
zcQ%`q&&J5$1Fm(}&yHuFF~~jI=J-6i@O+<}fXuqv<^>={<H+aZg=FrjFo{npm{eZU
z1=TZ&Hf~d0blFr&hJD`XcD7`^wY1)>#i%W<o)xEr!m5|~l+%NMc{CwSZ?U`MjAJVt
z`sp)&Z!4?UBgL+J7JbC3c!m{H*I2X!1Xsl~pmd2F#H{t1V^M9j2eKciN5Ml%q_r-Q
zzgRJqwmw$$Qgh3pF=ab;8|H>woar;j!kpjMw<)PktwX#L)p322$?`TtzQ!cIcCM+o
z$2J8VEFH{&`vn5(=G+;$(H*ju!0@FaHd=fH&er(WU06t_#97u}JtK`TSXaWPjvSLJ
zU!7<gyP}O7<h5F-Z0lv~=qg$W6G3cSApfpE*p2{<a`)UtAATVYj>ZmR7_h|HqtzCH
zpJ}7*%qBJUn~7B9Y}plKqsd{uxLvS*k2!qH5lnSl+2euuxG1omx<~7p;XC>XU&k8k
z9rom9a7&1gW$bkoH4V7`@O#@Gjfx~u$WyWjVam3r0!N~iOBzr6x=I^{zfLsJbwxRQ
zspV(S=E5&!-|wveAcr?1qSxb7Le%FH%^aAD4_CC)z9bWB+dvKbtKN^7d>#^pdvbr0
z+q>82FatY&U$EtF>#tx|sAQ4YZQz9}CDE2G1Yo2e^I)`<*-vGQ*$y^TeXkIDOk7Cl
z!xbQYr;sc*l8m!tY3#yX+(rZQ()pc;Pq^$CyO9pGlTIL8vG<^BIvp((+@Vl81;d>6
z9#q%S&Q*gt-%(uB`Ae26bq;<7eIwo!Pj@I=7Dj(_A^RzwnNSUTW6Z}P;ZD6>bfi<0
zs`$yh?aS9xQgy%MlcWH0RW|6|+<pjer|i8X;Sy-w%(ULGh8rDI$v>r?`XDJ^aa_&q
zBqoeb<o5cCedN%n?xlSkZ<Jbj#6HQr$r;kwP-ma|GOwRtpXD%O>X1+I9`?5@+w6-9
z8*&Gi%yky(l-2IbT(%|LC()nQwb@rmu#+cGvaj<!&O}9P-D%&%P{k#Lt#%kn{k9q&
z_xTWozDryXa)VzndgbqZVkEC=upj&jl;0FMr|KODvmdMFl(C?AmoXJN%lwo(VKrG8
zy6oo|PCHB?o1FaiOMF~nAVfdR)SZHVbsW~z6|3}b6{DskZuwT(jK|aU*#08NWLo_z
z_J;`6`4G;0SMGNasNC*@psw;3><y(szS=+Mmj3475wov}m)LTk=WDCak+On;T(|#X
z7`l{$NU>P58<Q>9uHPRlv!<3TI*1W=L-d9V2G{r-qc_elU1*^3Gj>z-=E6d5WaUM_
zIRPWbJzd?hzeq7NW=*qO_ZPW=6mS^FZ;K9`o%&Jl8|QZaVm|nv5OC;}8McZ3qNaPB
z-LXH!9@5^I7|BTiqaTH`boSfic&4nq9*=Xvro=P1^y<PkWK$h3BzH`c)3Rsf8YNQG
zW+dK9xSPCBG1SQ9VKeuqkk2j<wHHbl&oD8|`&FUE?KXRV5f!o3&a^qzhw|v~xqCp2
zyO+%MX;T21;^j84LO*8h5W4S)XYvq)gW?!)%og}pt1;aZpRp+LPHfbY{UPQqYS_|7
z<S&Yrr8(seN^va90WjBS*D9c7dC~;ATP2`2*ovx^SCf?#yE2ib9KOnjxGF#ECbt-M
z2e5j16Y96%6D-fjwb=x%9h?A>^?|Bqon}R?W$W@8-VVA-o@PDQ$1_>ExZiM|5<?yo
zBHv)*tP-B)*3IS8SnI?nj7TM1@2%TV-J-T>bQCpj?zSfX=jwl6ZkrP|%ZZ$536i4%
zRBMbLw9d|X+1e{8dr?igR=uN|Uk9vCUVL3aLE<7MNQP6@8T3{sibtkSoM_u}N1?7(
zN`>*t?>nN^Bab{{4?959XU?=oD~g(CJNK9C(O*{CuAnUjE3&spQ`;emQ`uvQBJ`}d
zGIZJFF^9<S#@+0oJ>gxC^{kk0PgZk88iev+XZJr9&om>+sU*9$-SJFrrs9>SkX`pA
zFTiCNE!)#o6Gd<M9pIT9>JN9>v$@rj7#!xLsBBkx?5$Af1$R}v*tx}X{yl&lw0#w*
za(6DWy6ul=S~lImrE}?VuojY2Nuf-=x#IVso9)X!Ai4AO(Fd9LSI<1ry_m~DK7%Qq
z2E&T&9L#4$yQh%mQ5%Y9kyTh*%!Zu|mp8O-!mK;uztI0FZN_M{mm#~T^I<e6;+cC$
zy8_^3^-Qkxemb5GEcvOS#pw-peYWcLcro$~ab%oJLS;9UT9|g;A;Et{?jLXD;&pa)
zW4yZHFALiHZ0vurRIcf}=nHZ;|FgIK-}j?7w08o8lD3{+Uf~dXH-{kT4%YBq41p1b
z3&Q*UG09k3?Sr6d&Ak`ODK0XYn3O9L<A;4%fm6Z)MZ`fLA60!~;IWS@JiG!6@l4L2
z#50-3NY2L%_Mm+l&x)B3Dmcrt&*E9p#nZcd(m#)J{TTGKOj_}m?Tdtn{-mH8W;d3@
z$-Ydy*KzeKH~+8l2^x1*^sBe8D`daSW#^j;>QX!pZweGZi8^I|CA2C&P?EQE_I-9J
z555vG{zD+pmDy6i%QwuurL^OJ+#f*(_z;t1oj<Kz=pcRy7KxJmXCIK_pCtj<FNqBW
zv$7&G!T%bq$wl&81#$`0DqfTJVhS}fUzGFGu6Qxd&>|&E9y!k&rJ57ddd04MF~p+c
zmwwfY<q##OV2fS-;w3SX9Ss=$J7vAN=EV?mid9&yS?9#GxWHZ;Z?sp$$cWrzC~U`e
zop+1*s?Dx{F%C=jIa0ve4G9Cj0OTKhF*TE=(k$oKO*tf*S|>>9!=?1N*}r7r4{$o&
z@?z*sR^bhDYj(Q&A%?ikf5{1o98$MOht6JoH|tm&v<cN$b=dMJI^4vqZ?`+*n-s_t
z>n0W3t2XJyqE~uz50>*iaFfaZ3zQcjz2Icnro=L(+wf`B_fvD+dt0%~OpCGPiXbDg
z{6Cx<&v-GEb2z-SMl=00(0mj&IxAl;Y{KYgvtKNd;Um9WZ*%-jaiJ#P|6E5DY2?rg
z%<~^ylG^I*o`eIsCYfM`C%2}K7oS|@eQ!bCxg@EM05%-}1NiSQeR14mNlXn949G;w
zvLSkF8Y1W&wdK_WPT^X<A}1g>nk}|6K9HtE!3sF`*s2%HLEO}0YXUegU#rN_owT7^
zrLB$cMD0||BsUG==dmchCw&nmot_Lvvo3){`vN<`LeOXH6P(rUiq7BQ>x>xNAQvXv
z7_eiSX3$!1++2$N5{Xq&F;gjKzs5vuXC0cTrW~)_;ar~HoGmEfP;V_UKM`3(5QAMA
z$y&W#TVtc-i}w5q54;xZsOC=bM%qim)m1%{4`$Ih+1A`EolPCKEgA^XuNI?EukFam
z-wQ1@*~1BMtTHzjd(;u)6^>yIJ9E2m!P-@gsXWMjkA2MFkM$s;Kkn<eb-O*`@GD>p
z@!-h>h1&qeXuNBV*i-QguM4z#<*)62F)k=X#gbvO-u6TzF$Ur&naO#3+Nl&Hl|AEQ
z--oM;JsVA3Z&|3o_9j~*Mo4wqQDZ&lU-Q8B#TU}lk#<M=f7|apmI9GI)%(E`fgqpE
z-f9%-8l`o^bd^*AQ7AxpllAAgZK{7$ie3!FhfbbuO&d&Nz!NBPnSxaIqYdRS>?gTx
z$+$CIfpZ>vsrd1*ja1Jh&-evvwA$awVxjGlv6d&Ib&2NX^kHk;$t0PsCK5E~T2iv(
zY?8`?`|n@9#?CoeM0PqVjV6S}D^y57Xm7;Go(9kLX9tl5iNLB_E5D*R^ido0!Af0<
zi#gaPvYL`o&|ej{`A&Q)W_y+?p5*TNZekp*hX?t;H{J@i>Zj!UPEBoFYdTc~xWhhh
zqAcnRg0FIm)wglU`6xOF^NCc1_HhNZdpR5eU!Nq3FsXf-VBi_)=>R`V0>;fAF{RT!
zcXFG4*WLC-g&mG)KzzL@vt@ewGS^@^C2qd*E=njAidSN;63!(u=zWu<rF?sSH~iMe
zxPRe4*>?eP$)XCL<>vjpLyV<^RDo4r^lv{_^HX{vcP{!Vo)yxD@^nZ3Iah~kFu9_C
z@lHzCiB9{qIw(gtC`kuck)O2n<+{SEokv=-o;208)k{LSqUWMAb6{QTv@8AJIB|5@
zRXriSENp9Sw5xku{*eGFR}(J(c1_R!TmJFu8renI7T3l%VYDfMGS^l8a0dq>2{+fr
zGvyP1JvPJ*J>`IHs@ZBc<|l9;u{)*4xvA<}J5Cm{VY|5}gc~y2qb_A?w<LhYZ$WDL
zaCk3=`eC~@dXXaRl=dxB_HB-W1}T$ocObgD+HHbkgVglYL$`@N!6B7&uoDgD9sXL^
zXe5qFj$#(#HU7gfxhJ#V<h}NMLE+_bFH?Goh~*4&YLAP8+N83`$sCVuor~JE7)lXW
z$8CmBVXWt&@oG#neG1KWBqpAfQ>3-dUY&8XeOl9I++}lm+%#ElbE|GuO><uL>of~}
zPlauH3MdZ%xpQ*hnpzN0DiqtIo{*qgnrpc=E%9+4`M349)IlO&z;s&{y@dqEdY-W5
zu~JM|?$WG82|LiF<)>pSa&0Lp)zY?|&R6Edlv+h!0Mn{Vdd@rc8lU0nzu4O9t;X<O
z`9L+U-_4y@``ukB{nq){QzzQ`o{+60850uX2A|)WwszYX?f%hf9S22Y#2Hjy4f;B?
z?Sf`wcBM*}Wc5MA)^rI$!LS-^vlGKu&-59;x0YB9jE3^g@mGK8AcecNCt~OR(K@_Y
zw|Ci<`@CG$I&EwAxT0;VZSx<~OE|!eTx>Lz6p09krA?QwsQg?W&Av!5E*sm<T#kyV
z?~1S7M5^2Lh&|R5l4?uCrsmG5ru(=fjbJu=q9;r&Qt@|?V>`5LPgZ@U3tXQ)l@P3{
z*{;+MyB%h!#yK|j_;gUwvFo39l&rq@{uTC2)fLy6K>xF`wd4hizZASf=d(_Lz0q3k
z)N&`r*k{k>z@_+4r|ru&7e<hU++@4IdR8*k!)DRbB)aR+QubCG159*TU$l3%Rmq5B
z{n-RZcYHGk0!Xjz10a&Xi&w;8Hf(I5e3T7EU${`}F3X%w+HgEWn3A_XV(=?A5*>==
zD%p&TqwX&?8m;B@r_^HyP=uez1?Gm)y=fCZa3`yAq%(4L;pqyKazB?^Ucs%x<k`g9
zcFG~wQ$VX~iGr5%Or|Dpc|J!fUd-HP?2T$9S@ESqJ8Ksln;h^QpJNqUm-mm~yTu{p
zLql;6yXXYa-lhr@6)IpPuy|@S*gJ_^)GxAYNcZhK^WDTPYJMo`z4%y{dMRx&hTHqq
z$5kHm`1pgIbB8KlAlX!4j%QEr$0_{`LsBlqK8l`%ht~l8)xzk{a8v#`-guszlJ+ph
z_$1y)N>I6iagWGPb3;&(c#D0OkYg!T$iLifKKB+J0_PCkzNqHm9u*YiWd#*4{$-+!
zZ&mO?HoLFlnGDz{fJ<`xucJSa=o$MaN2GV1(g#qHeVfc_RgHaDL5?l}$h91<(fmFk
zXKg9Al}muFm?QYd1crM7Y`(<m{^ZPd#Y+1*JLQ1SoE$DJ{gR8Zf)u)JlV6?fOrJ5`
ze#?)FmPH3_r1Ua-$%PzDHYLrU3I<;BQaP2;(7<)Bd@16jwzSoIIS*L{u8Pmf{aj=X
znOq&uyn9_I4%#&@g+$fTOlFX4tFKB<F%)>`?YfsP7CV>dROt!*AX0SV^)H2RQ>p{$
zc5}l^m4A<1uQ*1denpVVjnThiN+G$jWH%)QT{Z3GA*^6jCJB+@3A^Q`kZed8TTC9e
zz7%IVVjyDhZSjq60P@q4rs($ZehS-^f_B17MXID8^je!3vvSQq!H>wDv*wQRFS#Tg
zf~e2iq?i|%nZly5l$2YcOhvL=Ope}M$*fy@Im}@?Q*w4Vi`U!KXjha^Kw4#PvuV{E
zPt<^V1cEHZxw3~w@s61<g``}gv~$Q(`MZ-49Za#a6E!P~Wp++XMkk86u2PlE%|+7G
zn6$*WOtk+hXqxBLMgB0(@_TZPOZrV29Twz>wGB-*L}V;V$lb4~_;I-Cvm_y><Yuca
z{db^n(Alz=im11>!<PG7calaiAXc{`=klixg}APKsfgeAE?;h|yk}UxGX1Yf2%u07
z3nhDswOpGJV1C9MsDeP%gFL~~x<H1ng^T0TDk)>5Xl8x%9llfCArzX$#o0Dw-)K_r
zvW?Ns>8a>iQLHGlPHl_|gDo!fui&{?AI}s6e;{rgjfvNyvL-uW%#lKG2w0o*EgtWb
zAGa3A4bC(ro}e|i2!_KFcheptl=!OxF*Qtc{)N$LZymKR-xJjPMH<T)Vb|}pZ7&sl
z@a2gke`h-qkiRsy$P@hG023D^(Dmp`A(4h_0IRsuN1y~+gYEK<yK8YHZsOfzzNCLf
zxO}|YV)C_jHX}x*i-z*wJ&_Ya(}NNZhIYuFOjx#Jeq;hq<phiT<j3c3Z^NBOm(X*z
zr(y!__H+_h4YjlD?U|fEJ`>&cY@!t52Ngks?M*V{3?hO6v!dh+olegs99x?4m*}v4
z6&(0J3w~S(_UAL*3V<KKH|ufoU$EGE9V+f|NMeqZzy58i^;aLgR-^(cxd!qX+(U=j
zAue-+{z)rJf}xy~JKYX@$CMrFw2|t|VwdJ{f7wRknIfV+#n5YZBD=scq0GwFno@ba
znF2SIXLs5l(13J)ZQGhT0Ceb_bujM4`g*P!OF0Am{5c=bBIc%ex4jX|vl~6yZaBh)
zga^fz9DU_HKbGr4#e*ihn7z2z0u_(i+wsg*T5j}zCplM59clKIOYhxi74G8fy_5D{
zknaF4f4l5`$5vxEK>{BHL5e{NGb*0Kxm6ro@ISK;olUMJ!eEPi6kQh?lnt%m#g8jM
zrKu~fZT3lmfQh%HOQNGslOpeLvCrb)s$}!I|C=lwo%Ti5zkA#2{g+i=#~_Zv;q9vc
zS`xLpZ!5!v<eFmju&<MbFLAzb8vZ62O35PFY2POB5V&%)D(lO4)iWuD++g>8?vhmv
zo!D>4&76=jf@E^WwjYzAn#l-E_*bsyl2!NLXFtaZ9MGw0zf^qmvKTBj(67~W2TB|e
z6SUuQ<K0I{F%p&ay0FB-CPxsiLnrKtcvgxP0+8;GSN0aEhl_>0$gb**GnG1x%DQ%S
z)eF)`nL7ywdxNoQi*|6^ueWP^UD%N{DzO6>n8L~>(o43zuIffytcV77eQ!C#+8(O0
z8{##xzO-TB`H0=v>#~+gW|Enx7~7nU-4qZjlXsupoZrezLtSG`xurLxq)Acu{nngT
zT^rV&F1sy%t0XGOy1h59!epaCbnq=aA$w7ta*1>kvlU@x;8?XMHrX9MjR(0mPD;dV
z!TRdStS09`Tc8Mv{!Qr(IT3$=re>9WpPHj^0p|c~uxT+NH)iW-I%hMyo67TLW*|rR
zTTk*qb>9-bG|R!cYsNI2U15+*C$C?_iQWBe=j50Y;uRS)xA&ryjIvkRyg;v9ReSB8
z=tL!ga7yK7nD>wc0jKU9FR{+U+?f_tu#=rXo-L``ajOtjaDpxM@!tGzXs?#VduN4y
zt6%Q3_~V5CXe%mkUPVS<nUG^bZIcdgRfP#9Mn>3>z3{DFV~9dNMqzhry}!DyZT3J8
z$MjU-m86)qE<w~07KhaO7<T}^J8B#9D~@*L_@emU=q(8D<;b+!T>K~`&90~q*vbh<
zx3N};<c;IgsP9&tiKbjD`lqM^eD-)O=;q$I(-3(|Y%%GzC3&_~5EBR!En9P`I0xnK
ztGp&o6z$%c5IZU)Q+!h9_YQ}CtY;xgdF#qi@EP3F&H+X^b-QhiQ8HVzF4)_0<l>yd
zk`nB&qiQ$8n69XOINQMtNR!5ujX*J9lChpe_u0-w0Y;`$24GhOzODlJU~&vP_>c9*
z9cNJu>Rfv~za#>QT349SsxR@t5rx6<obNoDU#{-rPGC>vZfqi+3pTJjVdM^W4&EVy
z{ho@I$uG0qXu+kjp^C&YtW|w@c+xXIC(izQdp6&7V#b63_QtzztccGgkVO$xlogT%
zIP~`QmMd&?s|t(m?=9DW?VGG8fy1xoD8OM#v)%yCV`@ruV%~T%aH)>_d_aPpsA?MD
zm&1vPD7n-?JS&!{G9l_w#IusBkR7D>_aO)AzMAc|4K|#VvIsq(jpQ2MeV2_ou=D@P
zPE?(|QkLKpz@Cg9#HNHNpUxKLCakEUK|32e$OCiFRTxrT6}MFQfqXvRNc=s^p2b}k
z(vWNS8;Nl-dg9J6_(tEjaifjp+7oG9({302%_@9&-*!++;t}EvNma1P-U(pw^hZ1R
zt~b8(&MEd@?&~ge+g`5t{c2)h-ag22<P%(PACALXv1S+BKl0Dols&f7J`Q{oCp4C4
z1lz$U-U4?m?Y>W=MbRrlgHoY?Ry||6wWx^Pp9fKh@Jlor`m(rHO95qH1VrJNddZFa
z%UnxtZC^#to|ar3ZqUAtXOj3{5w)`azlmqsV7*8L@)r2k>EPey*>_Gjc$zcC@1v<-
z0cFkYwI2d^GC>`;AIAqMBCe_o{8Y7|0-^ov9gg*MQ&r%XXs*lqK?wg8Qv0v*tfbsw
z-V(BYOAhf*EG=my%?`LQ;f3xT^di986$c7$X~y?NV*8cR&~0LxvBVi|ovRL{JSFpl
zHr0q-9nUavNVB5*zFp%Ta(i!ULk7GyI+WpDb`UCPf=DS>u8WVAf)Gnn@eR9vd=SY=
z66@`TY=<Bx*;!?hZj5JgURJprB?Z6fKw%>?N!(>O9|$>aIW7V)gIf-i1HPt<%iFE~
z?^P@2ciL^<t{IC^M>Cr6+Yc0>Y{vBIHX(5ad@^p0B3_|n;JQCC=Y*@5BATfc75wRr
z11?O;i|(*ZsvsA8^$k1uK!|4DWz&=GFdLPNs^_MfdLS&9agxcVDbHNkIslxtX*s*v
z?h->hBl`%A3uP*8rlX^_1)q7F<#17Kf?UEjJ2ADqn}GTJ77r_QTt$eQ>ll(Zgv~op
z81f%~XZPgF;uC`&d_jdD`6$Ua`Vv4Yt?{DhQa&Zp<&#g<mgF!vqak(_1I*HR7I~Z@
zw%2W0PMy_G4RYCd;9jJ?Jg1IsgTfiMBKpu4mFL-csAFY923-_C&$s>_6I*p4<lG{>
zHn@x68t;M|Y&~fvyYQVRTHY~Gt3rbhM0f5icB!p;-GPv1J;5~GZ+)~acyt05QXuyY
zx$qt<Lqca?lP#WYtJcxHc#GBh&;$dow#Jy9d$`I1GGI;l{;yMPbFL&3xwXXDLWMks
zUGu0YgjQcU*W2MX6gXDHUNEC1bFS|4n@NX{&_JDc>k0^Tr%_!rNx55-5EhrrklQMx
zBFmMl14!GE-yzB=4Y0m@*iq2hna*gB=3?R(iP_Y4UNQrjVJfAh?TR;YypTVb{)TW=
z5zHxgqFY<U8H=|d*89g3MN8VY*%JYkqew^3s681STymjjXgINsV1(nRF8M~PT%=gr
z?HIaK&gFZ2033Z0U!KmzS-jkyNi>m+hdb7@-p-2x?~OqvbRetBA&wQ3fy&ic&mD+6
zW+_$GDb~EtSyS`oR@<L!hAnZ%dXfs`PJq{Sfb8Kt=yOnZ^*Mp9Y_|UFo3O}D-G$r-
z64d!6LB?RVXdqO*#G%`ePmGHl#ox1atpbj0B$*2tottbl=^Qs#9JE^P#3hT`3;q<m
z>l%oY3Fce>l^xnI%Z#0lXQ=K=7^(`Sob~bUEXlJ*?Oa6*QlGi};77{&_?X)=%C>Tw
zwl@;GE-}Ij<1LU9I6lxOjQP~i47J$Bc<%(Fzenxu#Kt0JY-UOg_Kr6z#d}&i6<^wA
z@8&byts#XXRlVm(U(janN5{G?Vp-zv8T%k`s@w`6=KEsR(z$lVK8j}o!!I<i+sE<D
z^I`Pk;CReFsW7YzdqTwlrbfiOed^;+{i}WE)U$H&D*HTF2p2-m#(Mk0x8eNyu&I0*
z2vLIgubj9_)kd<_aNGFW-`2_m@l6HIs4Te`Ug6&Pt&g;x62#wCBX!q;kS_avd`6@|
zlE84>en|2~on5m-*!?kDNn%2+slWD9vI;o2a?<?lGuz%~zvTCwnGX5az=t#8a-_1k
zB*grdJ<CBz&k)S5uZRjoEBrr@(5~n!f{v<42|Xo&_m%#P&53Rs!jgUTs(7XlsN>B2
zRbt$(?u(mReFMr-yQa^@frDeOAUR%}9YY{cN4Lc8>-vgZLKqV%p8G<8`Qsn|Xg9>C
zM5M98H#&@7EDE*MO?{~xQowvMo-rFGCAc}DQW_9t3X@CDEqz6(sYf|^YhQ>TYf+Mm
ztK617*LUk`Y`4ev<^ITgl~i#;0`^aMcTUU!;8qfTR0;Twcy@@u4k@(t(WGow@{#yW
zZ?ey}xIs+ma~a2X8rQqR0xCnjO^x@LNb(?F>Bvau&lPQ2qPn4_9_~8B=Qv~bG@F@k
z>TtAduvz24p(I!E&ncT7&(Oq5ajvVp&B@Oi@%Y8z$YEVgsFlqNbmPk^vbk`ooIZ%7
zc29KVekC$1pR)yh!Mnu>l;p0{7Wrzb(j_=wlK7l75!WLJeEzC`v1Jay{l#0U)0W4m
zVg>_<KZPNdMP5-&Nwq*ZrjWZ=Mr$d5WE4>pf}fD9l9V7@)Iz@CD#au<hpo+NQ(=>|
z+7DC{(nYp#9$OdBI9%V_1!qx!-g=)yOO2eJHdG(U*~zuR8xsq-8%Rt5rL{TKgAEO^
zmin9pX)<=Qw#08*W0GTQgM5#yDcb04lrs#rV%zL1Lt=#))>46{8=C8;TOG7DrQS~a
zB@N_rMKr&4RL{=QY4P*2u3RBZc(ZM5q8?I{C&+0~TCSJ2ExtYUmO{z=O1Q(9kW>@)
zaDE7q`n57#4SW?IjW<Q;;29a?<IT>T^ZeHO4g%wLCFYbrvZ18rc`UbWLn~QJK;`2;
z<P_wOC!%BBt>hmpb+;!zeKOuSAJi69WfUcHe=26-UO7c$WV;hUBpjtU7Q%}u$oAyA
zLA~sTr>l>>{)dW6Jd@ALXB_XaqG$725#`H1x4lXF${qZi_q(K<6roVjK7WT#Iq&xS
zJEf?Ug0?4s6R0v~6g4`TrC+k%;3@53r7uP-=Q`Mb4L04MBTk<--3I)J!g@)Tm=i7@
zHh#%);vPzLlP<HQ2OUlf5Vs1G9Lbq!zZb$dY@<n)Bn&>0D~fK9qy=`;S8MvL-`i=Y
zGdztIFn>0ITL4=)=Yz;`L)VV;Ns$z#>$EpwNY@e<vC%H{g=!Qb5g$jCGumZiF)i0v
z7qqjQ=oIaugE4K!40}6itdtfkSO0ga?@mFw`XcivzMBk$h3vQY0ut}p5%#|KOuVpt
zP_4lsRN%VSAGQy3nv1u!+egtr!5Oku$<*_4U);-NSJ+hU5LDVq1hlGB(5HRwr0Uu)
z=eJ4NifnSO^I0_slDe@nic9gc=kpl$Fm6()@Y%kj_C-AN@Ev!C{nBB@O$k-OR|x{@
zwO8%yTvnnRRWjfke@n7!l`;7?-y%mMDB5>UL<Am6#m4%6AEPObtPg&!5}ZE-ygm~h
zMBohF(}4Y0wU(Z~9LM%kJQMWNbm1m%KljBQl+&h>(*-S2U%9b2)!VO$_l0C9{4G94
zK153No4OqgHOalL2=-le#X(o{z-U2|Cz!tS;6>R+#(J<#Uv+TY^hLB9$rkKte~sg1
zja_rl<sGfXX}k7d5j*CS;pw`AA$H(ONy@2OyZ&H6ExmEE0NilUMH(3|l}TU8;+%8d
z7(nTQAnAhjw3`lIyd9w4+{J<PfB4UCc8H4LwAF4om}WXT7U<kO%pIMeMDTkI?dh#P
z5}JUPO<2OZnlO3a=CdGx)NW5qOr2^I5_6cnmA-#s0*PW*=0^FP-BC?Rx+57ik{Bn&
zo69rJsHDPVAC9y~HYK_&^edi+#&~LcSaQ$$c_(=UN6=}}8h1EF`oQeZ*o=cAQ7$Hs
zW#++<O{t*<51SS3%f4~x!e&=ZaIdPdIk}uTxty}OF|_Y{{gni-g#Yt$?l4k4)1B>}
zq!gtK-d@{Qzugx2)HZR=TI64#=tTrUCnbZ-lEfC%N%gl=>C&8<r~5(aJ8H`g79wA=
z$d*@c4~k>S71dV6v)ooF&}?O{-b$Q?lKd1Via%?Et%=c-rO4A-(G7=fZ6XCHU{2=;
zd={Mz+iYF-LM&Y~^X#<s*>+8v(t&LV1fny7{cg-n&`x?asuWsnHDl?z+&#xhzTO8x
z7O}<z`(C7>n3fy}jzMH9@wC~8!~}=xs>NE8;Fl_eTWjJ59Z&H-Ve7Q#GbzVIg6T-O
z@8uHKU|l)ppC7cX-eE&0RR-|{st_yRR7s|XY)7IFMW|er&v1Nr4&R4k8o!@qkH&|h
z8<q9Yt!8H;WMxA~Qw<W~uINc3S=KRyC!&OTEVrAo3&WTn_wJ}^Mxn&6CTejs*b^~?
zh~^mRgb?k?cqY!OWIZIsv8QrZtVX@LyZXq}_YEN;?@1ggADum&Lu1vJR{xm*(p?NC
zkfc}fv(cCD5N*p*+nar%dER2r<<ML4eb|?;IvePAe?nRU^!<G5NqnG=(zO%E?>I1f
zgOiC{N-c+XCU2CGK&k}{Vcs|EaE@R7zFGeGHyiL#l;ppal`fat!33&~{G&th;omO|
zfYRZ_j@&28a+5DQfRd>p{0M_lj~)zpM7tWxI>#PH#}m2AgsqnBJSPvvVZShh79EbK
z9SAbqtE5-b->N1%7a)l5OARj-!}$PGq0yuEMlMh(7tm!F4qkkNvtg{~|KI;-W7W(L
z<HhJPZ5NZRDyFTwM2o+jWF*78y^||e+*AE-^KLwI>kCIK5o+({Gdxz%UX=a#emslv
zkSND?*$43qLMlh*A<+L}JUhhgSw2OtKw=*y5iLdn`Pf$kn+k>6q3uuN1NWIl@<DF@
zG?^zlHC)M%BtFYkUr=M8#~0d4L|r1u_uCi#ekqYe_N7m{)D)*|YNDX7t)|Pq&YorM
zFBR0jNeWT<P^>)PI$u(B_;<<q?yK31K>mH=SFV#C9WuK8kS&_plmzg{fJ{L^3c|nx
z$$rZIJKMQ*cG}OeMdY<}1mqsAd=!3(H=eglYH?+HlH%sq>=&fU@F5${ZwDg|zewq%
znQ}qW`dvyZ_cj=vA|kHnzc@ie|0MtSe<XanvcJf4k8Jqc@9e7RsTfy5GnLZq>i!VT
zx?5YCA5^x)Yob+u=q3*k0()&fi<FJ8AwFH#pCX(W2X+I5O#IjPr?RJn`-$2l&>$`F
z7qRAs><yloUJo}WY_K(jL0~7eo3dTGvK5ACH}@BzQz9Ps##{P}T=4J)yR{lq<`9>9
zb*m}=W}deNw8EmiH!&p#AG^K3h)vW^U@a#&y3j|U`|_d6t)|!R=nvs}rLsff{!{^c
ztlSD!g?h4oP2BU{HYMBFqy7M`Q=`4iWXg1Ro?S7`+r0Td(4*|(u%6-5>Dp>DqqXid
zx;Dw#bXI?nu>TWg`q}Y1&K_M%ZBBpM8e6+N3A?hn(Fk7wG+6q3(&oi8KRpU8c29qZ
z`ARgTa-+C<`$SgCB)F(QBuR>Ua5Y_$Siw8_v@IP6gj<dbN-x>6c!oLwt%ro(v$niH
zqN~V}Mp+$Ixhw;%$mPTnm@qp3r8EL;)>S!hd$Vj)YkZx`CbzW##s5Ue{-5?h_EM+f
z0PDQrSP$h!Y<>Qq+>mVuB5-vM2F9Y6#{0-N<{Mc=As^i_{q(8LzPqHTsIQ2=6jdvR
zBfr!(=0Lh1G$rV@6g{Qxs-u%yfrxAYxk&VkwI-D;QMho2wdZg0y2-zyBXH?Oa$Q{P
z@{w?)Cf0AOcfCn<QMUQN$9k47wjKWO`&%Kshoci$sP@B#Ua?1W(J5=)*-ijPhwb!!
zar__#a91_3oSbB}f7u=j(7Eb!LwMYY^&cL!C!(!(H#w4oLVGe1Dyv%Tet+|nLq=wl
z8ryxzC^CuRWN&-o8A2TH2t`q2Py07@#H!dcN#`XW3)6cxKJcAI+8U&Uz12ENVv-tI
zmARCC^SNB7H7M}*^=B-sZU8<a=VrOTaaCe;rQF*e^Oht3aqCHfSO(^-H*r3jg7E=}
z8>!ra-JioPTztO`Bw4U@XcPu>ChZL<f$MSM_8s2JZDTmGfKz~CPzejMk>GJsG?Km;
zABBV5fW{q@N0XQ11VZ>CNArp3MVHDf@pUqRSVM@@gLXQBmN^F;or2w;4eV(_Jzhfc
zs6t52`E*eIz#z|8NLJzs_ki{_ej^|iE+e2qN*F0gl{42a<e<eGtpsF8Y|Kf&*q)Ta
z{$kaJLh{hYaTx5|{ULJ_xZaKCUOa?^Nc|yVMyAq|3+>$i4(AL`WqpoOt*D>gOESjT
zGGy;3I$BVYH`xb4oeB&1@ZX&&MGc9U$|+_c8vm%mu9pLp%hJcyGbKPR=Zt+4NULo?
z+Wj;d_;Kzg_|M{*yxt{b<1}HP2b78eIbvT_Ewm}5AEKOm%&<7qfPS1+v|TuWzlzo+
z4uxm0BC>rQ{T;k8$X5O)e{m2Vv2T-z@1Abo`7$gClYxEjE!|tfe#j-~ilDr~%B$jn
z#E;$wH?}VODaP0FBo%?Ao1d#0;QIx%ieC67dUMhdwJH4#wR!B<<b&9$NK9zI#n+-w
zC=x})pxL1~n6`8I?X)ZMr>-MicI6=#o65bJg^<*B)uA}UMLzK2s}IG=2m2$WMP<6K
zITS`b_SJ}8d+1^bffI_AP%p~4vi8?S_c+KF%K)x=cKx9^fpKUf`){k=aHt4hWU;Kb
z8>1_kx}~wv4b}a(ZaP#15Evp^pl`1FMF{pqwOituOD(SLzcpv8niIHqM-%Xh_#-XP
zD|UN=11D%kvu+c7Zb-JWp-s%0!gUYY9S#JdC)(FZ)vM!}Q77k6jp%D}VLB8NqrAjR
z4vnda#!f77_0*`Sw`l=S*(lm~P~{9Cg4-xc@|pP!sE@=hvm9G^Ahg)*YCekTD*8d2
z<8Ult+pAF6+-M@j=<9A3wRzcVO)beRN~*kjqM2B^qWTmLy`WkP!Zp|*<D%)Joa_8f
zlA_xZ2kFh-h)zobDKcH1uw~v5Nf1s;Q8RByN`bp=MFK%_xdw`wVdGwzbX3F(%%8>e
zgM9I}D!@|(8yx!I+R5Ryrh2BryYfOmZ)*>w&b$r{Ry!GF!14nX3QD4axBe@0TNlq<
zRpFRhpLCD!w(QUws?I#KVo?li%u(5f@M+GNSFJYJx{VUI#ei6Us0g8dCjCoeHPQ)c
zn2IO$T2nj|PK)A)DQ@;pAUkVG5>TG%)*MiAo6rb3Zmm6^@%=DI99j9TBcH*;b(km(
zOIJLLLx-@NBepe=EIrkM!u3iZ483lP*6txw{3&fm;$#hH@h0*fIg~Z6JM2;a=k!1P
z!FE<-ACqq+44}_;#W<QHo>xUX?-KfB)i_5>><n(7_P9@(6xxu<6FL7?-PCc7)^fsC
z3SQ;Pe#(jLFK`T8cXxa9?qccLlK`VrZ*L~EKzxqd6!9kl?PYuB(8W@_$0Y<=k1lhp
z0(s+=6P~R$!FcXeKiBpKr5}Q~k>~o5-{y!_;G!e{N6AG}`{_bEV*9Fn@j8pJKbg)$
z$_RlmyeGj%ou(^zZ-9-m0*oHw95`fsxpF)!KD}Jq3bVET{2I$aBYbWkTDpjJNu03_
z=5$KB{jwgMq*6Q%c^}eG6wAzTg?9uGXzUC^-H0zana*uA<{P^d+X!~nw{}UPh!SAH
zPUIR+pE=D=dS4z7M>#tcu{oVQg!*<J-F7zF0oIT$WDNrc=dwk6vnt7*Pv)^=p}mn?
zlk*p+eQrVs;}<T$0>QpE79gtR#7-9DtRx(+^KRZgY8P`6YHQH8!0X@kHNvW++sQlm
zdVUFz+ivexSbbSG5?GP^|D}a_FCpzH2Kx5{b*>E*H667Nyzezlq`E?8%DJ>R+eeA1
z)tFO0&YqCrq25n&O1QCMC9+Qwrbe_p_2B3;?`SnSZtU|I79I}SNj^e9l%3B<h5cm(
zm&7Ca03EWgl6S3;mQjV%zm8U}13+>{i25d5$xpo)F2232FOezbTz#-wltkO@dxz{l
z{OeyQ*$+7@R9aXCe~fRv7KppX{1nf0;X4glaa*^ay^A-WfXDvgeWMFSi1{@^<cdX%
z1|su8`z^<)BddC+*+2+3D4G<;F$(!!F;JuyDwB0@!L55B#2hY&V)s}1+dt#=4Pvj3
z@1?HlgKU(3_?md8)O(zzyg6Xk`ZHN=Fw@p->SpZgyr;1qZ0hKF{1^B})owRrADqYN
z7-5UUcH=;5f-2t&T912x-xS|CEA>dpo1<5Z3P5qoKp3_D3A(>^AVd#>$CO!<bdLie
z1EN=e&Tk*MDC;HV3Q*+?xa{VBEE0kh5GJ{O;y}?FwQaULDwqc)j<MFGHYvVy(WTr!
zB{snLC&wG;KJ`jEW;k@Gc#n_#t4(zX#(L(>vuOh%UUI^8Hc0HU8JBhL4jFDenHiwy
zS|?{U8OmqHGcA?m=+_{{+3^g!2NU8i*Qq(t>tUiq4)AQ;Z)$F|Mv==IEi3dPo0sH)
zHLQ!P<~<Jlq<^yo6$~Vo6B)pp7deKvHZun<+e@;E=oqPM2-#$54DTij`OIbtlVv_s
zQLh#;DHjl%0roYR<%)qKPqdS1Ze=VyLJ)V7RWXpqgLpO3H3K1F19`nt?W}cdqulDZ
z2P)L6tfUZ_P^`-}L36e~DY&b<wbeHG8#eD2+n8_g<?FKA07gh3vikU?OIDvX256eE
z6u!!A+*E<95NipW$8B@AgZ5r3Y<j?25|-JstTo;vCKQ9u5S*bs-@o~ob;NrYP>ZNz
zU9qF2?9eqz#gJutQb2~TjazdTYdX3cY+Jys9evt%<YxKjZt1cf&cz^gfb8Gf?a{>V
z@}kVgMLTfni*V@;ewP!?fBw(*SU?Hk7HQ)gX3EDCia&8>uqSd{B?xV?Cw<A40`{q#
z?WF%?yQ3coc+%RcGVUI4Ivppir~P#?B69{RaMhlPPjq5QeWduHXA>OH2~zH!djn~T
z3Y9kKb?p-BOe*L^Ns6DV2>4V{KI8BitYTk~C#UrO<V5#zeDzc#Yp)gdWxa{L4rSy3
z`+d=^U-e6R5`a{q&i?!ySzD&v^T=)kj?p!1Z7@5+g}+363=LeouGqMuR<oPhh6h}E
z^}i8R&|51obfhBP*I>w_xiBkL+x$dD0UVJ<yg%ua#3vH{)ajh@U$C>B^+hbn1>m&j
z5=N@{%KB&LtH~;Too=85!JXcS$+D!V0wpw3%q&*ql32q+z*1<*pL8sF6v4=yF!PJi
zhda7T;=xnf=R)ms9dhsF7i3+PCjVVW=iL9>-m6wuNp@W_)ZY<%zk-Y!IkDm&B(1>M
z{Tifux+i}^y=fnMzfBL>M@iP?-2)>ZJGLYTHP|Q7Gy40tcCk%$pnjSVHu?qpv&6pY
zI~4ca&vTZX0WMzm*%txNC6`8Ubgnq<9mg8=Wgaqx$x<i+FK$N+>G(1G+Sf6Covt2@
zC4cb|EI8kc1JK!C!uh_<Wr6ie)~|{ovy?dN+IRUyqaw%aIS;<~4T84Oen?1*S*L_H
z(?!Y9h*S3|`!PFba0Cw|e=1HF<B;E;eoFkj`FA{jcG=H~U*u>!ntrL4@ltz|0$zSi
zsJJyT4bFw%D)bL492a>Ib<yD>8?01B{1u0b1h9CkU3s{iKdTq7v8xWdU?pDy#o!;}
znt%1-xOUf3_lf!#$oHLg&EX<+lmLak-`TZ?Lqu7GMg6+NE_FziCudz5@m=b;-n$nA
zBlE1+YQ5obk=q(6>Hscp3;;a&EUXEP-*mXha%EbwoZlS%NA)kXkrMjlp?-_ckI4J&
zcB{{-8%}jw4okI3LNxfV)o%A;F`~&kYeEbQ7sD%Cv6-*h#CYZ*ghhXDcN`8bRI?Q}
zUQ<gL9YalumO6?`q#*Lh<U~RdltlgMH#z073xw&>)K_d_RV<H9tzbEWo==s=$uW62
zPkoRk33i4fTnX7bZDut-8b#>sAiB<3hePm^a;)TOw%O6lBPnzcoV7XeO#E3&MKrcH
zHwUYYd&s<h*GTDS?s5D<BMLxU;K-hS*GyX!jdfQp2NLSDC5J<3r6i2sLzX)Ia3?9z
z?aL}`p`$EnaySm+^QbM4amZl7MPWsPR!?$TW$j)W$h=&oIF8I!73insN>Y4**TgeY
zWN;J~-I=ZRXZWiZHE4<Zc)+P?tY`UlTjw-bY|Yr4ZGAPa{02P?;L<!&8xs8;4Mjl+
zX>W`!MEtd!CbbC~XLfFo`kXELo5R+aU$nP9Bz0v|d=XTN{`9nMPHN&}eq5(8w)7C#
zi=f|<J?chlt<}_z;WSMc1>C$naL)z5Y$VCl^dqg4FWc5;yey?3>$EQK4fN=|*_x9?
zlr4fZ_P1OhdBng_ZAZZ8C!g~=#5nHk#mHn2dyfd=T^(EOQAhckyMSwF!s@a-Gdpd&
zDx3{gSDD9hNUGH~Zz9F&<2j*XGZ#ww95MEUQ(ar+`*_l+mSo)aRCMer1XsZC4j!ih
z=);q2Pt`<ZBN6SOJ)I;nf1y3&%^@MO;!r!*JMScm@7~L)yBPIcxAUBD<JDc;aR=EK
zOC>Lp1GYa|>U63d^~BhRkmV#(9D~F3CNJ_V-jRQ#uOdvO2zDp6*@1Oj<<*}UESHyJ
zy|96t#q8-e7%R|&&T#0`8P(hHC3sb*;~DZtd<o`PIg&;l_&=8{DkpqiV?ArAyK5)&
z*Am%lPbX5bE6N97>Fdr`=o<GTMX-gBoO9qPHPd0|9l!UlSZr?u;uJbWYD$M%$)|B(
zw6P=)@*q*51vx5)bl;qEk+rw85rR{l?0WC`=~%zT-p%GHSY-$VmG)jdlVM-k7yG#X
zzMt4(Bg*SlDU%A0>;s3ihP*c)dbi8}#Xd?j;FN}DD1T8UYP)@sLn4`bpz#^|G@i-Q
zB=;pbhkWL}t@x9Do?n0gS)-s6`ywcTtUmymx-ssvaTtFY-$+q90I0H0d=--@nf=OS
zm25`A>t82^8ZpXC*89en*mGZh>kIRcZY1AjOJX^26|T4MtK~U~2H}SQ>rxf93v-HP
ze%XEuuw;prO0OS-^-lqouqtitxV`u13SimTC0LV8MOIe(B}fn*3E3udu6PH+Yr)bP
z`z=0YCt}{eR5@aUE(K{g!i6OT+ZFLF^{pfD$1CF*Tn(_wU%P5Bq@^v=Tk@Y6<<)~B
z9xX?sea&FF?=)=MM7n^%BHXMZyG?fod0ws?OmqCd3vC2Kd1T0=^N?LXSbRy8E%?&^
z*$snn&_?cx7Q1n<h<r=ojCRvt5!&R{g<jfj9t;ryy<1EDL$bc#5;GEE%jQ(7%Q}Re
zL^PtU_%C+bV8|lUOUl{p_QBL}p$7`e_ag`*<<Oasvj-P8aqyy>6H7c0ii1i~az_PD
zb62a%cv62hky+os#-PZQ#J|bW!=(gNWHg%+(94jdD<#>>ZK^*b6AnbX-9vk)MY~vE
zbg6?jBc92$MZpeK9>;BF_J=f2sDHU9&x+A6b!1mRM{V|Hy@poGpW(HOc3}sx3K&4<
zCOS5;)#o{KiG4?`X7k+Rzj1|9#=cXwAciZ+;2<&iO-s3T&=$oT2FGBazmh|w+gmai
zLh8zf*74Bur8z-tX}bC=lxSHlCEiHTDG_*1Eo9jt17!>#%>wpR&7Si_=^$46XA2SH
zvFri$Dko`54Xf(Snw;QfY|v}Rv4UjH4URkK1G&-*YqpSuq8{z_y1{Y>QKriE@rA<l
zCC#2fu-{MtPgWs%LDyj$vzLZ#<lHaydTPCw^;6bceeQ(L?f5h(2(dBu-NNFY<W1+-
z)!F8xGHK-06QYZ?Borz#QMAwptTmxv8Kih3*@Ly`I1BlTE!1Hhj>B4|uIq{~q{>lB
zALSd`>hH0a?67SP#q^m}o~=4pHK}kXuhv+RY{ff2*7(t+8{A?Q9=FpcC#M=zth=Hw
zWrkPEt8>_H9~&(9y?ImZ@#-z9u$;-;6FH4~_P9d6?MY`AV?CsR#831oAAt2Zh>C6Z
zW%Cm6JZO93S;UOOgb&!$@eH3)h7ush?U{;Qs9+}Xli>H#^aIa2PL|xi%J$}xJ){Eb
z&*d+Cptjh)=-L%uMGI*At7p`sW;j)8=t(k>`3$zgTzac!)F;)x=AO|P^Fz7G+0dWp
zMX!RiVFR%calmj^W|5-7L>#2=*)@iutt-32wYY4|;p)u@*wAK9_3TJ)M64TlAa>el
z0?=6K(N1Io%4y;Mzuive@@VU}+v(_9$vQ%?FP!+iopp4sMyNSgt;Vt9J<q<j^D%Eu
zuebhgZ&VOUy)QX^7jQ}kej&asE?rQ8Vj#yH23#an-@z^>NNW(VTkP%V8GjWh<{by4
z7A@ktK>$*JNbo35{794i{97ui_p0|VV?dY1Q5t4>K)ERLegG+X0ZSJGkt8jhG#>=}
z!r8+<%w8_Z_IPP8A328Mvgi2^uccf_=mhr1M*gHC6GFaq5r<=bnwxjgD*Md;-LW0p
z-RJo=x|fEc3;V*q9_zu8Q-*IT3cif({`=YXRZPP1>GvYn0eziIy^`Xlghtvo(bTOm
zxJ%)^D^=L5_H8Z-%vo2GS6I;R@)>!(%DI3J`TN{CYqXDkNLWk0T*)Ut#y4&v!Ht*2
z{VBQPqDu+~{~W8x+5Ag<Be5i%41V?3?Hp{s#cSn@Vv(_09ydGUa+Q`;v5HC*dBu^i
zUNvHj*lt%INq3}<HU$~N+O9g{5<|p!DIP-Pz}bFvyeUDuJd+cQZZzlZnj_^nX@c6W
zJyHaqJMa9*Np_w8yRL=Rz5YmuG05c|vZLJ)y(uqoA8ZyrP{{7I-FPI7iOWz9G;Ft<
zj)XYZ*wA7(C&bce5vv0_Z;5%z`QR`o=id=)Z#@!H-xlI`?6zu7sqg5AxXG_=w|l$4
z{_PR|-zG#)#TEuMvUE+%o=_rm*&S62Z6*)*p5)UkxtS-&`@<zADa;a5pOW)g+}hBw
z3BT-w2Jty_NV_O8E#G)7eo>f6^Nc4`g+B6W)^XI$%t?bd8Rdqt63seN&YU~{&1P3%
z%7KLiBSFiYTzY~CV0m-%=X^YJkc?tz-?NuyTe>H~#|xysxy=^%V%I#>YKxA9{M&{e
zr?`wPiGkym!>#wAEzMaWw@C}D%F>?6e%TQhWsA&TB>%&rUFpn={&u<d?@FB_jjo8k
zB#kO(m1t0Vnxss=3OHSpv@$^^K12ZqYvL;%(CAs^A1+3Rz_-@n729ZPMuw1*>w#Qx
z48X0nu0o{(vC1i8>+_lBS9FHRO!3UsQo7|Fv5oOeJ5b7Pg~r!bh>kpGMJP}jjd~v!
z+)xJG8gs?Ba3pb0wkBWQ&aJlDm+n6l5Bt}xC9o*gfLjiGRa$Np_mDo~u(c*OxS|s|
zZ0!M+G*CEz!50o%M};A2PF?zBhj6j$a=ec9O#D~d8ed|^;&{;CQQH<@DxhCUWRw=f
zC9fTcNAyc0_OSQ!$fUomw@1AnssG5@f&d7E*cr2M)o60)s+g?pN~rOkQJMkDPuOFL
zBjS(IY@&;DRzZL}eBp`Y8~@io{?4BCAIJ<a)29-$tx5>HJMb=lsPnex$i>N6zFhDF
zgUW>%(cr~CqYX0ZG$&lMH=oXNRNeZS3XunjeSbVg<=F)GFY{Nen!jR=?e&d>8qJ=I
z$sgook;|Y`|M?<_bVy<8D_-(70P+67#Q=6|c`qqtRL_yPh7_v|t_elc+na08v7$^P
zeetCnC?bcK^#{oCAa;iiUh!l6*KMGBBOlmu@8Ylc(O|rBJqK@!9v{)dhN^G49F;wN
zoE@inSi{xWj0Dt@b4Fs~nH90bM<A%kbNV<M$Wi}B>dg|tp&RK5pA$7D{f2t7>Q}y%
zhX{~<)lTQU>$Y?JblO>OQxc?f!9W!_c+S@yHzP%dollx*k#O}!(1hGm4%!7@sX7we
zj8%hqsum>;zZlPa5=s(s&fd;;2=L05Dx3a0$#RLa!twjvXrjHUJI_vTEAM5K)m@~S
zw)dk6E^9@E)iL}*AXv*mf=D*PU?}fDtU5eUWENTAKT5KwsiP8)(n1J@A78deew!_9
z(LTvOE+`^X$NMYxX|#6Im##tS1xOcTpG9l23YlVbAO1XHM&On@OHAYo=UuB?>`P|@
zc=MdGuk!dHkPclTLc!OuB{q@zDYwL!sKu?ex}8dQgl5LRjbWs?I;>})<~#p<3x1#9
zS2I+eTV11+wM5FCA999l-f;c19|NlY#yR;@62bx<B|k^&;+)1XQq=#FLpx|}r9<{>
z5*)gontB8>@3FbFq19}tQgA8j0S6p)RfdXy!ZtX{Mc~S4A%=1o+>$lOt{RHNmB3yv
zigvYsqbi}uz}E~Fp=&x>YOWn}nN3z3ov>m(uC};tsEA{KhMMep|MyrAwa9O%KvEr3
zkMzDVL1L1x+f73uDU!qu7qpv)iV)`B0k;f=ga%7d8jD+_H#Z}?CF!=I(9F$mSK=q9
zyW6XI9u;>)y?(?d#52UVl1xJwo|pp^2g**nV<_ZQm>P;Z#gI=LN;Q~rh-zV!Grtry
zd`q=3CTACD(&Vo`C5K_RdA9DeHZ>8trDn?(n^pk_R|hy1nV8`?<D%bXGZQ~!B*wQ{
zf%8K2V4KT8Q08p!M+v@Uo4_Y~yUp=$(SOKkZ*Gpn=5Sw*5u2B=Ourio#63PPiC3Cz
z!B7$FR{qHr#hj&@NQa|5DHX4`B;LEJ&BJx%7$o0h@MBA3Tu%kyr`NK0rWWp7e9D%`
zvk(S#bG8*p56j`ZEw(bjy$_|Ut#WV?foIs7YLLT4GCj)2Ykdk6CfNf{wKcVMwr<FE
zuR6<NVZ}~^R9#MYRY&yAw%)r&*VSzsqHFh913DzQjSdKzUtl3t>)o%wxL)rRJJz!R
za2pdSSU|xzE7;^;!nTOivCTPWG${%{6W?zcic3ppcgI#88La_wD9dQ1;P?~p=2VZ$
zB+{O^S*OcKhl50vk#z;lD5bQn;_^awrnxmnXYUs3B!<Ygc{ginIyND-?Z{3yk)_@q
z&dGIBhvU%-E#h*{3U}uAi+fTz@peT+VOBizEEN8j_jIqShTG%0&vZ*8Hxab@M1`d@
zu&pO^S3Oiys;WGdaJ1EKw%z&TKJr2DiSG^~Rh+Y@lX|w)P{PBW2|#@FxE|u!{2JxC
z?Y(?=sd^I>BAM^G>}j#rKZ%}P;ALXqe}DEg=^yU2o@h{1e9T#LO|R3`@&@XN_T_sN
zwaRNIYkPk@6JT;(IfmRmke$|HR2cLIY;hP{Dxz}p!NWs$0w-6-k|I9ZaJ<Js2f>Jo
zgoGRxBCbR>;y@9ad&EWq05-b#Bp%^s?L?xUxM7L6C-XP{P}0zy&fjEwl{WuuA`0S#
zU!L<V_?N|YK5*+Ub|HOmfjnSuRBz-dQ|>DODv$OH0gx`5;wHl4SW+bXV2fRJ@W*=Q
z%(1ttH@L9aJBhi~nBm{8Q0pNW+Gg+Nv%&(Tv_50+2Pl525fhWvNiM1%#2aUP+KCF$
zkaF$A3TvwA<|6DzNrq(+KaLiD$v7mXgME_ctO_^PE33(;-b(5Gl>@j)OrOQ-93sr=
z^Gnc#5&)>9wv~O6GyolPj$i0!n-@Q(((0^znPWYKdrk4*{>uBUZNYo*>s-N3%B!J3
z{3h`tW~6<M#Pw}FbAw31i+z{qZfv1m?Dq*hJD-D3MgsdGu)u{7dG1Fahnj9&zMzz!
zlFcowX_pPney%2Xthlraokh8AzxY~~cxn4JV3F$`+Y>f=)P75t5y%cBaml~gj=KEA
zg_E5j${(;Rju!EvVH@QTkiM@xT7;Qql002?v>bPf7TMKDi||s9Ba~g^4QknZ*G6}e
z2Bnasf{9&sG(CmzziHJyNwLh=`$wp3?S`W+BDulI_n@=h8@;=+9{F%<t?^*H>8ML*
zn%QByIR=yrsItyS5H)XcDDSJYTl0H-53mCw=+xP5M??5QSxfZ=&wYM-bR!M8tj4dg
zuO>KzEe%_3;?X#48u4|~2EXHI5xG#UFqKL9G43le49UH3^3fuOiNlwSU{fkMQS*Zs
zC8wY2I4l9G)8c)(!a+|&2iT0GAz2YY*iui5#(JCSOM|ag@z9&)xWIrSVPy7E7rS&o
z$^pg`xPwX7UpnDAISbs^$}Hx_pvn|OggZR^jLoY6ajUDC_B}qCV!IRjTHu%gcebc%
ze!6%%%L8vo^-OWp&ZL&c8W+I<9LX8O*LYd9mcP9W@^BJco{(}ULVML>EAl}d+d#LZ
zmGPAy9daNWv{n90#p7xS;N<FvdMPo4qT{5ktss*?CH3YJd%#zrQ8L6jr<}F-+4_Lj
z)e9;(t81|s@rHO4Ih*K1`$h>YH&(s78bk`V+FY>3r7mB6VsaCE8PaJyS}qdkYtXIf
zuQfRcO|7;$29mxii@!U{S{!8DvF2N=_W%iyW>*~1+I`A+_kla>h+aJF2;LHiMczwS
zeu*ENYX4CAEheBs7)OitY0$PM#7c<7wReY)y?Djy#rAOQ5KNViI@pi=)pq*U80%@o
zO><ZD=u(wuw|gu{A%BXF**<QJkNf8vrq~nJ7xK>#xsb^zVQ6eA9?(xEQmHvskCG@m
z)OrRF*yErnDZ<kZ&_m^7|4g7m9ZShl#ctTM-i){~v^aY&X?FB7a(ylaW0Rp*kx(;e
z`^KNij>X1%-S#^G_aPqkIPt=MP;tp6w>PPoATW{%;u~H~x>UgBX5Vl9)w3Z~hlfhK
zt$}!Um=v!=+Tp0v2YtE8cdPu#L)o=BkGdXCT%o6d!bcL7e`;x~Wn+!{ax&`{JCW}?
zbuntOlQ}I+7AQeD%}4FDw_n}{6FghJcS#8^VLO-HRcUKlF%qbn$N6X|*#x>O`l~nM
znfurHaeu+5qQkMf)yA@W99oZ*%{pKg<CzN>!v0x%J9oqCR(r>Lkj}WS#oo=iH*wqR
zviE#RrV{w^evV#j`Mw`M2=-aRdqA#&l(5z^bzjjBbHB&|qeL8hR6#gg%x_BU{c%Dl
zRRYuviP<NKSoWxn{7<8e>XRKUJK(f^mJ5dq3Y(EIx2kg-y9=dK88@B{5UwE0xU9Ng
z=6;(r>38;3&Y$bvAMESw6FX@+slKTOF4WHz_q2VR3<5t0(Ude-9%}ZTZ_4^s`#!s+
zn-2RSN9=5Awjcc!vp5Vh{V4`Ggg8M47{uuT``MeLPOigN=8M%`_uzK>)n9aWbBx+=
zNjNoi|D{{5z2fpVoZ1;rlOuiW6~itql;aA#NQ>(WcxAk)h=ZBPPI}U=8osDVFe2<;
zv#X;|0Vi=;e!kaK<9Ny`6}-JRp1EyS3i+4pI)5frrJ_0tjMonr@#yZm@3tG_6Std5
zkU4HQ4u|MK;=ht-^QPfY(9|^QwqQ38`>9>d5H`no2u^3W3^6*jZ;4q+#*9>_G2dHb
zI4nY#I;7q|X}9HY-KdphXuZAq+!>zgQac}>;DD2P-6lGI2*zJ!cMKQNhT@k9OOxVj
zDZEv6fuxA+335`~<ai^`s6#wc)Uu1bQ@jJESM0Q@IopngVi})SVHIa-U{>y^8QEsF
zn*e5d2TS2WZ8j@DP_BoJ!<p?L{Bgq6ssG6T+nhuMEJ~SrB*f0Go~gv5E)@rC-f*bx
z?v>ZBN>1IA-~j(XA(n%-AfCBJs_0p4ks}CMgq%n59;a?yZ?I{O+R}VRtuJk&GuU{S
z<!Go-k>@#ImpjJiuUcd){GV&Kx7*6<TRHB@y!WcD8ZIJuO*j7PoUUuUbzMW7tsQpp
zSJvcL?1AdrQD$B&d$!J7Pz>j8!phc1dx@oIY=gJ=umc2BZL*Eop6e|F%dl&H)ECww
zXjmDiDzDifz4|1WKh;>H!;0fphl;^9`KxAP$ZWH(B#9bPina`w%L5FgHEE|DzuH<E
zw$Pq9Yo}sE@dxcl=%~@qYF+WAL^15iQfjkhbPL|<LlIQYg?d|lRNvf=$dW|Nf(?Nr
z2{+C$9(K@C=a=jdj|MiO7CBr>;cGi{c55hIi$9Ui4vX}9dn}rJsv$hz(U?9yJZ`n;
zwj*x;ghNImCJwbH<C`Mki%>9QKb1ptl-yvutL?zefbaJN_~o=?%M2Ims`P75M`v(b
zE<cB`osQTu4q#_ZV}m_g4d#1aXVKoouqs8DIKSt-73-&rd0owojh(hHzLe!qtVO1j
z{ljr_TGi05(!F&xD7$(R(+ibBzBfNZs8e+F3C@?kc&2bv#Z??fJ<*?2+=4d<rmunE
zfpW76cX&>>%DiWTfdWrnCgWybF|eMuAxAEEJGAJ-F_aqwWJkk_<;W$GHlb;F6U9h6
z=cI@)H572mk<+T;W=`b#;p?Sq1iSWRE{_tLpv%eSoG@X6oejYJ(khTsx;__g6mB$3
z=g0BzDwjJu?<=$ziOt?{^l9gUste;QtzCz<6DwUwku(<V6l8kNE{>zQT$!Y*e>)d~
zYj}&jlW>8mVUVtD02-b0Zq*pF-(u-wB*|ri_FgVj@sDuD*!#Iqcq^%v)(4IYN)EMd
zLPq(p0;KrQA@Urwj}nkISOe_iYDr+gOp}S)Covi2t~qX>X8WaU?pto3MSHo5AWE0?
zKc71g+qMNEei3YR;Trog*eKaAzlxRBA(aaFl5B`cYT{peQ^lKnlS6hs1e5wUKD<QX
zCnk}(6zscfy<C}lzK_<SDrDyo%>)+lHIOfl{cy<yB>#~U$$rdd(8s~T;m_DlxexK6
zaIc%6^GVcdLRfN2QBpShHTsAQMkNK<Z?Ts0`IXS1UGAzVI~Hmk>Qx*C3i+K~5zk!2
zE8YNh<*|xgk*}hJ9bT1>s1HPq9}?Z!)zz1>;T&Wm57;%wLX24mkHXG$?Xi$^79zb%
z-GAM&B1dpZ?LymgtO#JW(w*FJEc}0JiRQE$j|ErZE<~bjcR9W3nA^{OKhthbaB4_-
z>KdF|j=2Qr?7t-DgHn!Q#=h0Rn}6R*yDc#Qw8~^At?2F9S30(l7a*rmTe2M6#OgD-
zBzl?PJ9288sqV>>0)P}W$E`S%eQw1;VA!S{3p-sa6>qz2sv}|XwuU;J7USq>RZPz7
znDu5vxA*~SYqI@a|2$wbb3jN;>Z$XEDI-1Etb|P_ca`Xw9S~q|<Tf%Vag9$|oz2Z9
z*+e|=ym1JM#3>`~J^73nWCcCSj=Uf{L`%EN7G;Nv$dhABj+G<5v(}d8b28Dk*s|!S
zFs~vaow4OUcypVSa4Qn+t!=f+55BVcK$};dJZEf`KO5^=)X=E>fQq_WQ?--cljMb{
z$<`)06qZ4?RUI=W`ul-+BZ~JMq|D|LIoBoH$b`iySZC`i+{+D3f@~V*HpGy5Vk-yW
z#++?g%UWO88gAHDe=OXlim0rCFE1C~=veqmb3NJLni2}agiG3?%`uidK~=<xGs;?W
z?Ju)kXzg2_ygZrb1Zz(`6sagsPkN2ltRpd5&zavMhcNuHbHZpmM_B4pZp;9mTl0;w
zr1|*+ropOh^NEi2{ONw%k+92<rv3hK1$#;2U*eY@^$)l$HQLS!j~PmG+;$xcfgIG;
zk@m3D9t-?k%8cDo(~soPXOAc0)JfRH1XlH`+fi|$wkI9O_b$J8&A-@FPNEH@mvrV(
zv!llLcrUt<AOTalt?K1mNmM2Bd&WPUI>(;Psn#}-(-;o6_t?cHB?5o!|Cny~|Jz%8
zuEKYvj1m`TJMK#|gEuzW{saRxAHm0)tS4VfPO;t!|75aZLFIVYS3Of9V&4?~z9ySE
z!fn8N&;iEs4rXhtCS;W4N_5bMl7*o_X`=kwcoPU)XU(fN;&`QMhK*)x60RZLoJfcp
zTiPBH(VfiIEBT;0?R2)6nh3L)_jK1?)9hRh<UYP{k#Wh+$1}{(#c$*oBGVi3Eb3rg
zkn8q^#0$O{?KYN7Uk6h+S#P%6#hA8=Dju`9EBu$jU`S%}K*Zs}-f?hcX4<On-i?-S
zdzWfK`b#Bxe9w{E)r83Neopefwvxs5gTw;YpyCw!VZw(<O5%Wouak6aEj1nR@Q<?x
zbOy!!IzNC@Q4J+`Ve^CbS^oRaYwhzug;ph1B3}f}Z`x#E=3YbJg4YMq+gJW_QA4eL
z9W&!wj-HI{A|bPV6TN!=Ue_Fao6E)~q@V8+h`GPB?-SJfnsLzl!AF=r`}g)^{B!0k
z`{`IX7Zgu-WWu6}mDjtjf|+{Z=R|^1AwuoH<VsTvRi=5|-t5;r&{ng*e~Sk0TOr%&
zNwX1`gOuJ$ftormu83#p+n|yaS61yxP7rh&y>`_|h#7duVY0czBjD<hBF&&K>9%X)
z%K^M2WKKQ?0bW}@<5(}U<Sx5zB#bj03B%=tQ@;L^ph~{RcKwLo>vbhLY&VRAT-FK8
zk!E+3-8fR@DQXIV@|#9NvLFT#Yvj!Vs+)}q^AO{P1Kbjzcok7;m(JR)F_aSO4cl!w
z)Dki~wLPSvZXb!<<DNwVv8Qm!2_wO+Nw1`$!3{RC8h4y^c|gP+*_Hxe+e(V}Ne(};
zuLuZ}11vFhZrYsO_+nSj$lr&7_EnT$Q*#aCrZ1mjT@6m#w2>krl6k;pjHD|Fv1PJ(
z%#5D26QtGAC41IL$g2oS4}!tj(X-P!0vPu*kDZ&7xWy5!ufyg#wlKL#=RGfHQPbUM
z_xKN%hz=Jd9&u{g%)!E=Mcz|!L*VC<{B|iu)1?)RXG#k4*PxqaJ``mR#VnU63O6;#
zmV=|$iueYhQGshnNyltug7;u6lKQHg<q}UKu_nLZ2G7kD|I3$cZ9L;9Swf;zSoVPg
zzF}MOwFcUCj^I0qAX}dZ-a^41xomFmnE)mYH|De!P{y@3Ixjap#D+dVtoLWd!LzP`
z@-pDDF}tjTx-{FST<M~BDqg9Ou$&e*I~Hpj>QIL@SWAo`+3F}GD0;6oM<`kAn~SNm
zJ*XUqE0(__Cy4M<!cyw3D>{--H(?}PO<%LE)ib3fD>o(E7SB}4MOq;$wc3uP9yo<0
z%kxOBp;B_*P>*rh9?ePW+#!X(?W|xrkI`#aPPrH%1jaE)P=Th8#TzN@WYJO)0rvrY
zJbKVlzGhD(i4gdQ#vHsp>8mz><-PV)j=s2*8E&=RNh-D7Sa<g1>(x>Ou_vF7dCKM?
z)vW5JJTp=R{Dlkb*~9{-+t&+bhHP(ATf4ipJQu%Jh~U4E{de1!)5UY@W!oRWl*4$)
zdi+;g%T|HiTj7%=lI1=v+n<uD^;J*}7r!jDDb}AmbZw1%GX|=UUr~s^%-DQfT%i-g
zAreECEE9e^Hk8xBwO4`7XTj#MgH&I$3BtCKXrYrvdUveTHd+CrW!F{WxShyJVv{>=
zC!>XAEkbZgtp4fCpeRJex3GRY8?6QHuut30`3g>-Ior-V_BT$kH=?0pS9H5k0`?2u
zu%WTh#{8e84zi00K#5@SL*VUrhPJ1K#;V5HJHBhB$H7pRv_OiWmhC$l>^*<W_C`VS
zzW09b+Qs%k1+av8X-Eg`!=!>mB`=MA<Omt-Ax49GAl2A%>2<f#Pm*l6G?Xx*PWv=k
z0dI-y*7o?!S9yMD?miDJdRlC;z>(N;U&IO==3wZzFa1|@7wr46eU-qi#O0xm3bxT!
z%C%SQn<S(^5wJ@cw}iaZ8Tl@MZ2=4ReRfnwPMzkh_JgktH=!SWHnnK1H*t;QD)dv#
z#%&`yg^t<J{!Et1Mu?W{PQCpSGJz;vlB7aZB~_v$VwWAXUq`}oW`(vW!entv`Yom)
z3&3f!<8k;=damUBxZ?Olui4htZC4(5q2uykO<kQ8N-%0y9e42v=L5t(sg_UL)yG4$
zm`~ChT$!`G``hi>Xg+`(MQ*MV&3IilU(g{9@b%dv%GXkc8F}M|Y(iZsPO%&FXM6KD
zXz!-@!WE^S9{lFxsT5L{A|+bl#Jr_yD0Qv)@BzE^xXWn5n!Euie5|d{ptl_lEfZN)
zTkQ5|SF~PCRFN>wuP0PJaN{aG=X6;E9G$CgR4mIyQ<os`sJ<yzSx_nN&tIFA*!s&F
zo1F8+Z-TnLQxbgDhbu#zvZ>WPaf^ZvV)$&bX~#oSSzKQ#a?gm-{MLaVcbUCxTr=a1
zE5c;4i5z7%D+epdrA2|S*z9;#?l!=y%<6OU83&CNKB(VqZZ%$UgT$E#+GM+(wt4YJ
zzXpr^q4Q9X+!KW6b{P*um2O$-L-9DV1;>lDwvs#Nq8L(cE$U<lfBGf9SS-wI{92kV
zU@ugVkiEv1T{czcp^oz9-ZjarZN>3WIW8pA?#il{V)88K>Z-&ywg6J{h^%lC@`i+W
zH+N<ptF^w4+oi&O;F6Behe*q0>yo;#vHjlGXNy8tI8SnfZ#W(je2I8%)IDiqezaV<
zachs4qjmiwRv*aItmLbOT<$~+KBU@jjPBhX6l)XmOjAyE6;`p$*?{Yl(uOOGVM{!N
zOXxDMJfhZ`AFU*+uss<E$vj)csXOukB2q~O)-~R(Y$Ym0W?Pe}rLXF;ZI1lu7{7No
z%3xA0_OK5$88g`wdo&@D0<kC+hiqp&lZ{oenv(N&jknWPP^wOQEFoU~&&%!cssZ7$
zjv;#@fmYhQ7IHN_>AQz}5n5<&%TL8e+5<=yMX9md7ebkqY8!1&PU%l5yvd|peV|k(
zE&@JNZ3zViI`tKnW6$PhrLdojMSHWeM%?u}?K#KKjA=7%U-VHnBUJrzIf^y${XzW$
zY^&qeb3ATYGN70A_r2A&Q}6>iY52G2)9G`Z<FV9b{Q<Y~9jo|`;*kcbZCMH>6oXPR
z-<M@LZiDenNd>EUaS0iUZ)B2^g$&QSBAIX~iM(vMnujcCct@0@K=T-h-bAO$z(GKe
zjYey!S`<)=R*G@tlX1dPbKm?2{vX!f2gK`o-S>Tnh=_<zHzFdU<MFm#ZkJbGEtjj^
zk@8>WHkVgjF0b3;ZvN?v&A&Z=OkE{zcPApc-cCfv8xfO##>q^laU92S98JebA|fIp
zA|fIpA|fIpBI3T@pXYhLWBboJ`*K!`CG-1!Ki|)v=kt7?&-4GWq3T0L8%QEMZo~0R
zntMe^pM|=P#53OjnO#rVXn;WJHpb#TRl;?}p|d<`=W>Hunzo74Jn!vi^Pi0;cHK)_
z>52TYl6aC!-s;gur_QuD{cqTbSL`is<Z%{n$8dd+WO?WuhTnL{R|U6k@A{wJD30Ds
ztf2TKgy8)EN{Iui4!k_wK1c$E&w%WeMehFaGKeD~0C8P)N7;*ylA`=y4K<DSabgZ;
z8;ZW5NctqlT`kM5eVT`*76;G<Xz^zWFjO3c^?=fT?#mtTS=en~<lpx-*_S>RA(4tP
z`6|J>wWh8c4aC>g9O1i}-65`H=fS^;F+7Gy^!}iIoAai0EP1kX-e`f9$r-Qb=uK`L
z;Xw*51Uc-7c;hBbx8FMX_%Y{7u=r0&eDP#!>V)`q+Ry%44y9%TaZlX0;*WO4KoIJ#
z63ihSxpJVeP!BbbvFxe==P)!CmBsa(a0;Bex_U$DG{m!)3KyWd)7K0H@6oxP;;YvV
zq)EVCl>2F2L<hD~X$r0zm?&xb!P<W<&8AaFMwXYu^#jSUsgw?+1e44S1HmtlF+-kI
zH~Q#p)W+&;v76!xi9PcC(LudA=hxoUPNt)T26fKt)>sr&gd-$cm5G*abD-cZM9N72
zq_#oOnv&y_kzBv0X7AjMciJ5}<5Ju1W$vcqHZ4IeCq<H9vOUxDqra7wWyU}-wemvJ
zJu@?Uk#~&zki@&p@(GRiZ1@|S6VrB1NIBvAgT&oA5Il54>t)&d=K63zRYh#)Memq>
zibh1aS?<aaRW=@S4Kmx^6>z9J0dQ2+!XY|8TK5smD6x-Jh;~oHbrG?t3kHI#UM?N5
zEgC5N5SN54@n&1gZM8J|K`|%x8SfidpZt_ev7DCW)x;<P4I=fkp(_UB&LF&!^r#r>
zSNZ^(DcQCv@b4E{HrWSxHy(xHugTp&ya;>9+T2o{ILQ@d>wJ!wNFiqHt2q{(Sqbmh
z5RmZ$sXPI%1!OkH8|g?&8QP-kRj|l|37^^Aj<q==rlM{<c<OVN9_xxFvd=bE;Fppb
z@E0FBv7e2Jn^^L;#Fsj-lG}yw&DEE>K}%f~HVN~VwdS}Bp*?I&djcDUKV+_>S}SQ9
z3h8~tx~dgHg4uw{Qo36QLTsTv5<HY=b=@{l_&idp?#QuL38x7Z^F1PCvF(g;Bwxsz
zZa=r(u4*A<Lx{7!JEw;!3Ioey)z*>+i>atq86q?%*9~@$=Q(<~p}w5WClcn%VV9qb
zg_L`ruaz*t;q}yjYv}()0ublM0fh9(dmQR5cxKttaiF047}}rlmA008?sy$Po5TJI
zpS$N0rsR`yuZgp^w>poQcI^51g~uA({(}D+?^!_|C3~?t&v@0<dHPD+ki1XDbrLGl
zT5=5Ri*baL3iDQul>>?EB&QL7#bw+Z&59ty$`m2qmta<%14^^`LUD7z!8)4DX9o2n
z;R=WBXo8slSjKL){%GWGk;O;V2I3jk%wmgEe#gOh=47i76dUs8$|sGJH|#^VqkOiJ
zoE1Vb8A3;ME#cb)6&(wFa1R1MA#hvxVu|NEV&}Z?rN|##6z5}f@sdiF@hX?vcmkW~
zBY2|x)zZfJm!iFjsENargTb4=v)D!3$g-Us7IlJWlX}~qDS1{EZSzjHQ}7%hm!H<V
z)iWJYkM(=chc7|a4fekO64ioz{2==+B|;#oOz^`5&}y6oYV4z5bSclX1NO1Q6ODa`
zeG>0IFEdUu2*01^H*gy&@v53PpH)!B@s;5z7q3%BogKOSoVCyMy_<8S^W)3PWzxQ=
z#!)JHsdmoVmx(e`+fYfwzKTY|s{yu4E=ynMu~~-a;5St-6s@9vsQ|x?6D(7}E24-f
z<y<5FU9|ROH${(Z-}~Ah5a;(p(jtn&C<yw;+-bZC4%<%&NL=0Fmwt}Uq;tVLqu566
zc;Q6Gdx&p@R2KMOaeSg=G`!4njY-L2;>zRB1F<@YQ>kZM0HO0ggf{r9<8J*DTlbP(
zeLSu+Sm365l}ftixM%(lG8fJ8W!j#KMWILM&ZPgj*5{`!EH|C%*IZXUdyOOL$L;zY
z3ogXj?Hj6AxW7Oam9Oc>U(8=$9*ZO(o(J%z<CPy3`l^!a^=4lZK~rMAZ%J@<x5^#3
zn1^l+a1|oLL{ZEH{k<(GwS}}k)Dc2~6_ZM@ATrL{lpLWJcY2$8yfCd~@EWo^qP2W3
zg>jh8Y})Z)6Yp=W?}n&N_x3Q2bvEO;t7p*n&8#L>s)&UekFXS*l}nI5G?E6)$xd;N
ztb;kZ^LVg1MCC(|=2r6{YXS_J)JyX!P-^3nM!8Rj;u-GBVaST1El~oHyM2CZ)~vMo
z#|!ViN{VZ{$Dhyn-7H%WLpkmhPfS~s&(7~FZl&@8TaxR8vB1g!A2h<mRDW=3w65Gs
zu>mj3^?5;LTOMshpWL2&$W}NYDRoE^Zl4{(6K;m9yfL9!tc2)Itz=C$m_7Tqwl;cm
z*#ZkQ-2A%Bo{5a;qFwJFt!Y$E`V9fUOd%khQ89MYSP{D$a|=jzfnDvvJd^aKfY%~@
zK%7olZ8jvED0@|3HI%+gvOb<mg?rj`+yxY6Pm$2!BNFje4jd9`(JLr31|{EoM{4lg
z-x4d)m{MB|!Lv8}>n2h<w8m@k9SRi{Bi`=Mh#e#08{rEgJ{>_CNIw90x^k9W0vaDT
zXz?U+0I&KRL1Ew_JKN$}Q56(m_S=r=!DA4Gf0@AW5ue=jozrb+G}2-rwx!~O>|!Tu
zS01|6YwlfRyM6D+d)i>ea%z67UHnHVo!@Pb=cZ#cw<jFzO*?8z1<)raM=G_yC$p$L
z6$|ghhx54Y$>o)-Z%+Q7&VH3u<(cClGr`()bTiPi)&7#k4`*;+-Te<Nw!P7FA4bGe
z_I&g#X(J*z?1fw<7k|CIm@C8jq>EpdjgalQ_4xY9wMj0?=oK2yrFbBF=#pnoZ~lTj
z^;=(_Dk27?qdQ!!O*mR$ZbxEFsf?63K@``ET^`MGkUWO0zxqXpkPXDILrBZ7*kHn^
z9)0VOPnFEQ)H3mA^=<gEH`quL?@s8V$7POsgYllaZcg=Bf_76_ug}GV`^NEb=R%Y%
z_Ix~(`9W@Tk~_u|J?>p#6S+tNrDR@}L~|+5vaIFDid>@D;K?LwZ~7LN)Ie<ETiykR
zbOM6gI@?qy>Fw;6Gpz_5;AHO{FCxcMbWriulVNgDf!KTTO!_hHw{FJwvmJ1g(1eY_
zKFD1kKN7ys@+3aMvOcWFJW^DIsQc`rSc<Gz1Dtj#`kY{t%s}>WETwP?o%V@u4OVhQ
zwoj}6s5TLx`z)~+-!YKu^Slw2>8=EUe&IbpCfeKW%WAGAiy+Lde7GpqXI~vJG8y9C
zrl!rLrzwoe*FNBQj~Hpx#yjA3zX{aAGRDq)>%Dva&hH$abvx|)9EX!2iyb7XkeXBW
zLvHoL2kb|mJK7D@&|UUZqWis`ui)qWy%2nBYcU%P=7GQpI3DTZuNZVrN@r~!(3}i~
zxUzcVxz1D()sgM0!MNZ%8*5ZA^6J6DQ%TH0YQJVMHAkb;Q*rwpvTNg+b7dtFlU+BM
z+A5xM%JIVil4lD`tLx=@?-j8~iTZCC3{9F&)s2G{+t%)mI5))ry=;imxAU*voUL%e
zhE)Rcw+trb;!>9e>DF9H7k;t$zTf6!vQDwiw+|LZ=V3)7Pf2KS1YhC+{i8x<<$7yV
z<4rMYvQsz)2AuAQH<wcu*RD<TafsBBfo8hT0*a-ysAp_OPTaXtuFRRa`acmCHY+C$
z2aRBBbMociLW}RrEtZX`9$inD&CLxdj%7mDmu+57U=<!ecGqAqQ^*Jno9*twh|DUQ
zJS9Jso%nDm77#f>@YtB|&{1_>yC(rlFm5S}v|w;zuHqp7-E1^lkChF;fef}NcN3!x
z+6JV?ep`~OfN@t@7rDPJjc1<AzN8JZW%(_aOwkO3e#_&T6Zb>Ku(l$$T4i?*+RCc;
zqilj?nLb-JSa{VrbAD@U5-)H9>a(>u1lm7&{6p<*UF<bOz<}ysw_6^z_1QJ|7;zfn
zj%-8p$Bl~_a^qz!#5u_s`$30px%+E#!*PD8K~r3x?VW&Q%pp@`Jd-!uarz*p*c#)R
zG_sgCmD$g>Bz`>HrY!wB1kKgdl}jGy!oqvWEx0we^-oP*+`L5mS54(_D4{f!?;Vp6
zLX+5)GpwA`U$w2iqQzLxY@7dGs)6l@FAu^99<WFJOZXg8)hVANxAx9KH(<`UUHKMI
zw~i(%=5aQ5dtb`&XOG3Wa^#jtX^=qi$NfD%lq>CtL1!?fq);*y=~WB7#o9q(-6X1$
ze9E58#mhI5s7iav&si<*7oD~z;aE=?o9ByuI)LN8k)F@NeaWAxz;UDuat2oW*=Vg?
z+OoA&evS5Ac3nc&O6r!q4x{m&CCa7qyszlO|HBRZLQqj<KuM0tOY@d<YcB>8&-$(P
zB=Cs->kdJBU(C!2ubgfV+JTta%i||N*rLt~g4SDYzoMkcS&YHguY|s&K~1d>+2Lxr
z10|+i8to%K=Njm>9gS|nG<japsO$x&s6V@zJ!_5)BvryKny&_PKVU|P7{<ptyAcNj
z1jRfQ?`#|PMU%<dMshBwuf+_>hkCSn<{DMS_K*4es25OU=lt_dBFE1UR{kNS<aW8#
z$CEzmVrfHfZ4+@q|L^ceyA)HKRF}x%jv6-hW{%BtrCL_j&bQ*3;(T@2OY8J@(BPt<
zp`Aon4?wycN8cHom?biM`R*5fnJ?U^{I7rauCI18&Pev2uXgrtX50ITOhu>UPA-M|
z#x8$Q?Y}hQ8bVqs-6tP90A#K#Ao?f|HM+3<_HiuGbwtvS9kEYxJBuw2HNmI3_tU3O
zx6fi>ZcO?&t{~Wzf1b+s^TCM)0?7Z*@BSsY?2DXVds}f%wlAwijKHbDRO6lT)!@YX
zxaR(6^n@Y)of;)TI(Zd-9ixg?6|Zj?n{VRTK`enqK0@2@ZET$%*P@64E8wA_Y-bR7
zzl#k4lejIucUZ#|lNITQT<LveXW86TV?QPVYVC%~mi_$c7iUeOfk-2;1h3i8-uUta
zk9H#YLq)c>*cB%nO8?87A~MPC_Ae3h$`hdvEXG7s3Q&J{W4-ExbD0NGO%b0eR)UG+
z&n$u=4d98e+F=0fCQpJ#LSJ(td00@D_NpOqZ9c=nKyo61TXx-v!j<Br&ZTz!iQr0=
z9j9<XH=Ia*6E?4u+KYMm{UZf!8bWT0-X}9hAGVuMxULf0Aw0t^xkh*fh%eam8j#qB
z^m(hd8}F%AS?}9?PW+<S`|bXUd<QV+Hl-R~9?h<ofGnYQoSGkXAcfy?BE0hP_>`fJ
z1Qd#Bo)%M-nFFIIZo9)aJ-~aJ-~ebublf3y#$_W4<Vwz4NR!Qs#gWt%m$6y71froI
zvN?%BE%-vbgtD(18REwGB;>QXxgA9pE?gY3c`=%dw=n5NeNd>B-BtCV#Gg7I$L#K^
zhcoDL++aLEo{5c8Nc0KZL+{CUMX@EVJgFP9KbRJjXEJiyq7%VIuij4f9a~cE7g@kL
zi(@n_S@UgaE)Tt#ax*Q9mXd~bPbkLMjfTq;EL_-~w&FyQM}WymTUqt)zMPY}msjP~
z;La5Jagr1xYqFo(?I^m|`ex%`0r$5qI|720#S&rz;~LG#V9{V3ve{+6qlJI6jrm4e
zMK))W^L;P@SGLGEqBdu8VGoq7{zP1_WLHu`fFf^i$|fjnwpwGJA^fgN8OAL!9@{Iu
z*Jb9D<`acszkkJQYxSYWdy21?-%0J+6W2BtA{2f(zoOBQYNX$~a&O2a+CT=St<~zp
zpe1!x*6MA(-NdmV19#dEpEdmR0oeCPaubS!KykRRok?mWd;5Z+8oN$R{0FWO<q|`T
zhuia#OYpVP;UOtV?XjFZk}F!84txB>#EpFSpZ}Hr?1?y)&eg@FFDpF!WOS|j_>eu7
zU_dh}rhX9R;~qyfOkN%Kbbz36dTg*fcp*Rd20oKt!YShNq4+9$HmP7kOD(ow(EGWh
zEemj#+?#-b><kK6h#AkvbY%5;#a{4TeDu*@*^9mzXwq=arRZi)^)dCo*?PH8?2C`7
z#f?_@K+Z^vo>MsM^$ur~qol9G9o6Km2n77!;p#)rpD%6rkrPGkz!q@S2PvsOaoBg@
zEM2_tZ*3r#yAmV=hpBE{fvbuzW~9@|UPCd4&W&(MjMK0K2R|wxHInNikHCqblcUj{
z=*}P=Ia?>Ups}3Pz4ctO=Mu{I5hH*8L@0W|G&oubQR5B;sX->P=~l8CTuQ3dPH;@O
zz3K1J|6>z=E0K=iF#74Vw|z*0;D{Bocf4(#0-fyL3V;eVL*a`A{oZBI<iMZ|Nbl#Y
z>uT^+_@DyGD0hm2=g7qw&!mKvN_>D^N*@Ij9CRhw0b_^~f1C%l6OR;mqE+B8QbEB3
zsh?l6Pkr0SYD-m^{OX~m`c}$|e4aBc76#8G`b9pY4=f+zdF;z-H{_({#16s+vV7&p
z(zwIEPN@9}GfE@Z?l*bxk>A@evWcVWv~S~(N~z0bRn9|jY+RFl_MHO?Bt8SD@qJ9j
zeQR{oj@u7ai~SsgAFCE<r$7|cA@)-oV`c06xmwNv>HozknGHFgqsv?h;e(1=94cG~
zE)cpNu8bBkP)nl5ByCs4GsUt=0j=3yo$WZ3o=l+MuBqC&(QOR*{@S774L6hL1CQeC
zhRU5b-ZSkUyMCz9?=4Wb8+;O^ic!+-8~qn`cjcXMQ*;Y{z)c7XPT0*u&gS`%E*^X=
zN0{jH=`G#^T+3mY;ai8CT`YD&+KHQWrQ~fxp=QJd#BQ%T9S}nUVZ;*0`c47|vK85s
zm<Xr=TjF}4sX6gAOYdK8cT_EuE5N<Urp4e+PGz&gD1|6JWYeRy9AXvotR(0&s>L#O
z+Mx%xnb|9)wbf2`?^!->)20%fHD@STkrl<M?aufBB`ZuLHhQ*xZani7RT6XByqKDE
zcyb+->5=#%yQ>=V6-17rWU{++sY?hIwfO;IX)=!5Jwxg1NFFy;RM~C|h6*($x?@q|
zKtns%V2druj<zVb&eFt)Ki+T4avdr^R%^?19Uz{VDJ95WkvrDXU1uw!for@ZaShq3
zoQiU=c!cYkYX1)?%?N_I%<lv)+S*)&qP1+@P_QT5*z!JT6;WKD{mG}0m{U|x8v;>^
zJ09~Imh-}8LdM1ldddooLZ6L&(3eT{WS!OK8fK%Wsm~uP3GCaHgDHtTJ`(NZa`*-Q
zh^6jjL+}Z;E%{|5W$v22|JlF)-8^fJRS9DK)}FsWoFVHN3aJH^S`#LZ2J6ahIta7e
z8jX8l>R+~P-YMS7Gi*o9LmZ|sc(U~LBSRtfP_v*&k2}4S@t)0G9JVW<Au(4hfC2_~
z=f;Jph_qPtSbnueC3qgs!>F`n6^u((B_#&8CwyvcwZQYq3f|<MFGqHtJylH?d8Zt%
zH1qArC64chKs_DbdKk8wp`OY9V8!Hrjb_=Njs7xkE8bkZk-On@*&6XqDyLI;1MRJP
zRY0vM)1W;+6k^sNTV=iJR+V=yuoq%F|D6lwzuSvW6T7i`^yI!Q;!fG;%U$`0Rcq{k
zca9BF6;jwlc>LZRP&7U8qc5JtCCXwB+u<SC^~=C5e^QFU0i2T17xd8aSMl8=i6ooI
zEYK}ywxc=u_Rj5fjpVi8T)egW6JNRA@r@&ojtx{ZIZ&h<)jt}%Y$h6mAX6L4!-k_s
zRpw<l8iKdQ5%w(n!HC0wD6p1J8_jJklD}>#uTgWv#^TdrQeX@@VV;X;%0dDIi6LOX
z&ga_j74WPL;|}xLbLZMbfXD9=H}SYR$1C%=Q_iI%Q>#i;-ka5=rIUpT2PxjFW+SQP
z5Fwl+_I5R!eMOe1wv@dSyDx{D6ZUSNq?VdGgqB8oFFsMy9bvIzPv1|NH?|RLgfnTU
zec&L1&wx687^ng-XdfjG{h__fKF-dN6y>3!*!WKZo_;&|O(Ys~*glN|Sj;Eo04bbR
z!6kzFDf=vXa}LFCrO)G;V_acg?TZ*eY62aPqPM=d3qkwK1mgqvlQh~_eqpSB*uJjl
zYvt<(6;+t0$9{a{8`gyn{I{`pcmBq{%Z;V@Pfd|^zmJ!{n`=MB%(OzW)k>lKqXS3j
zb4i1hkXnT9`RBZ`(4ruvnw<=yS-g?NCtYzeHAYHM#{I)RedWox1e@SZT?W0XYSv$_
z_z|w|t4{{cv4(Sf&B>shggbD6NOu)qV5h>cSQJ=x-O0G#Yg$`jkFL+34>u@t;0^Hw
zKo_Ot7*kH~q;oulR~56#X>!i!rpsm|Hz3vgzWHQ2(MeE&8_>w`=2D;yW0t?gm!Iq`
z<lxq-ms5=Cp?J6Xx{BMR-JXlYci*N|YmzKk)O9vBW~kdyL5}h*hUG$HxZ`Bp!XPmE
zNH=Xys|HkcX=hTVpA24d!xWowvhbNQgqMseGfx)oqh+hjsup*uD2bE^SsFH}rx?)Y
zWH0~m@9a)rsMtn%7C|rP#<<S=mWo6+@1*PL`Y=Qsu)Cs#QpAfZR0P4@-sb`ZuFjA5
zU|u=(N?mwQEMC-B03QJ?ez+Gz-y%IHY*Dppr9CYALtApP(qM9b$g^I4j7v`zUV0rG
zlx05ng}tEU@?6}~Hd_&$d46#nN0K~L1Bm$JN)M;FD$(mcWoffD)%;H(Gel;rc*fc1
z{wO@Ht@9aBsl?XD>p~RR)zf5D*>Ey$KeREb5euyn<=Pm#lWCbsOem<QYy34J`JnG>
zTj7N`FSWiP7_@q;&zWITi=3Ud$>G<;{R&hXs}`>oB;#M(QcX@aLB)g(TC<OWOQjr`
z>#a4BlfxwI?xf>Qd*by2#6@*Dq}u9q++n1mXG9d{@20@5K3{SiZMSXJ$cXbm4WClm
zQDK5=Ao~#p78n1?9*NdUJtY|PukB1~(a^eMy8_F1<pY^4<*B?oz5;rcvF<US^TVZX
z@#EDbqy=?V_Y;>5H3Z;A9G~=wlimJ=J(Z&rk7D4uC*h;mAsybhocir)|7^VH-X->o
zL$#@-9RxnlCY0M7Nq39mSPbjQHxaraj(c;c&Te$`aHr4v<Ow=LA@o8uc~7S-&-)kS
znNpj{NTrD6o<!m1Zn${c=S{|Y<g9~lzVsspPC6r`s|w7A9g^Z-`eyf+)$>=aHwk}7
zgD5(Lzb`hlxwFj<Co;^QHP?<f#aX@9jyeu9j-289bJbgwe40d|eomGzUSxxLVMw$f
zuQJ^a`L~oIp{|4tS8GGMkBu3ro}Iw}3PXapHd;NCV(uVp|5+Q0J*sYYQ6HG^gLW=w
zL`Lw3nBn<c<Vqq&a8zrx@v1*1LntYJY$Cs^QSe!pUGl{P72I{+jKK<j1RYUwb3yQk
zz2ysHXas3{JD0~ckpCf(zi3>_QPg3B$9Xr=XgvnYybr|g5`-iJg}oo2O64i*oV>!s
zm-N{OIYcRL^kJ?|$rW_d+eg)xO0`iY8CT=S)vCRkCVo)%NkyN;w`BS}X%jti>(lBR
zueU4jz|Z0t)?T3JLc-@c4|z>dkDbCV(aQNJS*l0vOUK*JZZ8@3Re~I6MyZg#p6s~T
zb5tn)n_MnA)9dWpSX?hs2s#deess5vA-{Z=xLQjr^Y@cu!wreqec679XX0ffBuixZ
zF*Z+LU1#m5Xd&;rp+eTj?B`q;F)V&SY&bMN(uGkI>WblHMM^FTc*{($apiDuRifs~
z#$Z<shkmN~r8yzGI-dCr;Y`dm(WjI+$AvPSe1CIov{o(xMP-Y}zHYd%PjzG=R*>xV
z!-fBnN^`s2P>uKskP!0q+l|A8uek7ExwdZd)nL6i!T(R$&ECGG98xkS(cD|SJy+CL
zyESK4oUfd?@GqRiZPhxYs&?vs`>=Boa=jR^DcJ&Iq|G}CHczek6Dur37QS-`(zxPB
zYtKL?x8z=%md}cdgeMuD9?$w{QSco}WJWwIv*3L1W2BkG&IA@cb+}$TTjoAYvvOa_
zGa|lVPRy)$xCvh*5Eq|^JEQj^VQ>!~gVC5f9DHs~dpmg@J8hmX9Y;L5#NCw>I!|sf
zmcAeH;O^nV(zN4oIN!I7B8UV7-4j6TL*5>=1^H_BoZs1^Sc5b{x-(^Oz05dhOLCZT
z&sQT?+NF*iiX$R2%9d3Cm&>qNM5PgspUiS!!YY!6`W3PwnuePL8WH(`DVlbrPmu(q
zt*Gi(`9d1;$XpY1r&qKfYomjr-VKnuxBEJuC?5WBN}F4YQ_%WqX2%NBE3JqRw;_T1
z2U)bC=o@1gX)n*%gSj6i%0?kOqgI<>TMYu&`vjXhshL*+R~Z-^E1W3h%Ve2ikB_=X
zBsa;H9Jr>Q@;RKX=A0wa%WgF-3QJM+)|f;w)3erH+^+3@a8}>HezA3US19#W#|gLd
zE`P}tidD{Mi*^aZxh<!@WWoKm!*?&OWRLhtRxQ5IcE*53s0iw?D=AJXVj$<--4$#~
zX;*@u`s}gckmG8>&4)HA#r)$w8D*!oC-NY)OZxA$Cv#87kHbAb6(Dolcj2!fD+*5X
zz1ou?Tj|x;?dbrSXq6X}mreKpT|MKAS$y9Td$yWpx!~D?A$!hW-@p1^+nZoVo>1Id
zKi9)SY*2mnd`zxa8)7f`GHR9j|Hb%5w-i1WAi8MiX*jAL2QYT$16{Vy0o>8xX2=71
zS_t-O*k--eYIW0y5QqB&SF%#O_r<*72qaB%D;=`K@l25*CxMFC;3F~39<H5!J6fTg
z@^QFxi(m02t%vo;)Sb2Qq(Qox2W`NoSIShs`YRiB5dF=R`8JeruW8+0tIWW|i3?n<
zT^LNr66%w{lR{$kXf^di$UP`8aY7iYmRpDeY)(|BW$EJC&s7uiGPn|_{r*4iz0o($
z6>>X)iyqGz+$%;AgPTnxbUTR$=(I~!cdB>it1T)!dow@f>fDUMugl)@ttFdax4k`F
zgx{v^_D(JW25H#YBzrgUcQqAe$?x%Aj3NXaa2AnC>+*R&zSh1<u|+8k`yj^w?LLez
zU51xQOWd<kVSZH2{Io2$V$5_3IP)LJQ6`?X!9MYcuDRbnP1yfQ9*>{pujMuD)aN;2
z?5&MR0bj%}$jXX0K1(|Tl9b9)Ak&C_=~E>Bu*SZs7AHqv-ECxJEV{w39U<6#QjzHz
z|0Y&j)=pCi>fh!>+B>i!ewV)}7Dg_K+^l+>3p(uwf7wau(%J_5u>zg^gGCIE*-yC+
zlF~Lb*W1t4Iwq5-P&`mGgl@lPb}G3c`F!XIowX}Y72f30M=29=<*CBgEQM#6!R4w`
zg_$4^rf!<6<6~|xt|e#juBo2MxuKVffNC_S3a<Rs-`jPk3ST7m511tHOi2^~%?<e>
z^YW;Hm+i(JZjqv=Z}MM}QS0VYg$KOwG)N~8_gnn!{R>NVuUmZ$YseL1w|UQ8bm`$+
zZ@0&2;+~X<QHIzl-UD1GxdCkIsnCJQ8@H(>PGbk|$XS;B(#5D{(_&r(MIok4Kjplr
zmM*<F)d|=e1!m7Um7Xb7J|+yYnD%EzhaOg1f>dpm!)xQOY|g1v_fh{!TY%%zDd(XM
z;gKc|`DJ&gg@u6l!&%PF4oNjD6O_un&C6Ls`Y@z3_c6QcRAGHqlSJcg-<6u$TD<+|
zM_UQ1sy(A?;GSwtrREbiKo%ic5VO&ZEwNsa3NCGUlD0)T8)T16Xt2ch2LFWZwlv<0
zW5DmdsBxB^DtsI+#c1aMk}dZSxVKwvg>O5XiHL)cw$hg|`?tTbRsJ1rFEt*;x+Xd=
zgp<kXwpbg_g#JP^l{i?JAms?T_bJyD{#PF0-)HON8>c?fbPm~uK)?USwEr6iX`}Di
zlEwC*Llft40xN3sQ%IJANC%OE>T{lyP~s$zPc(PAy}P*uofQYI@l?nyq*Wt(JalY}
zV_D%(NhF$sG~>MA>hKYwT6?~jJ?r<@5g3hSfV;jcc2Rt^9MQB`H&Ji(Q%)QV)9JKr
zF=_FPy5uQ!XFH-R(1NZWi54P^Y<n^M@ARGM#%F$4ZqnAa=597{xA#BZ!+n3W)gG(*
zS7Ku^2fg;V!|y--%ASa~s96esC->?nW9mAtsY!8>;i()7BD)o@i#_pms68+K6Ibff
z)ie3UtB+UgnS2J~dPbrFdp7q$`PHy_aeqD++l*XxVGneFZ|=fcU2D(BwEWuif~-aL
zX)jbK#-+!}Tl&S^S)yska>MrZ<dcOel?mT(`+m_ca;EID$g9WgKu()8SYn%CWqK0?
z^{|}Ks6I!5lD6Cq#|PqshS_dG^T;otS<Xc;uSb0s9@=XC(MFXm#Vn#IWeNmr*}$nZ
z@BOpzuHYV=2OV^rB%>P9i`ArcM=P01pAE+|cB5$b2VreS;u+e6;vI7wk#^Ld$(N06
z%q|R%`2xGz9s*Y9s);(kBswHm9kug$6c&+Zb=+6>*HdgFk6&{QY2dhrF8QlPf3P=m
zcRNYMVQ(ebA@9fA`H}MnP4<qX^y-+>yWYowKiYfMe4I~)RSDp}AItNypCzY^eUN*P
z-be}3F|FlDa4#5{I_#rU6Eoz^_vHA$J7K?loS2MLBBD+c_NGr_bi@UYm5%m)`?Q)X
zx)<oSGE~bEYRo>1)}A$?Fvj+Iwk~qGv(~N=`6BPiCfpYyyuQ=EjP_1x3U>>skj?a~
zoC~*;ZW5(K|2m#2rc<SsDB@w?_<nRYwCHC2w!(9+1yp<ARdeL_;10+8&%V!R<)+h7
zKVd&iKI8b4Dny=5@yzu)E=c^8?TSE7lrrwWKj)as;2Du<Y{dB`ul%U1;J95eQkl+_
zd-uwbP(xsRhvy>;*GOFAu7Z`YdG$!}QA?FE>KgA<b)j|Tl?)8m#y~FLN}@e$*Nr%9
z;r;=-nibt9d&>3E+66qxJBRIt5!XR=>YBW9q}=wunfqJ2$;ZY$KzGm0@ujOU3OMYR
zToCCM(5~G&;v7fm7A#ThrV|Zsew%m5GdV)to*%c6th&*rjAY=SeCM2g>&zc6hg2?$
zm>q1YPwx+Zq=a9Bret%UmcQ^4Ako%H=>4Sb=j6`F$wS<Vb2b;%%m7lcFf*V#j!O$O
zD?!T5=h4QL@2dcZO&=Um&Z;}JSMF@te1_1Q&Gk*K!$JT!ajBRWiw15o(B;G+=n87P
zJ62uLox@gD@@7H`bHg`^LpDDrBxL7kkTKNmiGKT#ZBN^RT>e7Nu`Q}5_8P36n57Y0
z63?8D>P{T7r6a*SG_|516$iB}N5XXYGIKbMyZ7=06nCU{2Ga=tN{_N4=5PoTWv{LD
zy;-z)m927=!Uei%v#qK6WK32>Rb*S6a45mAt+sBYh!WVH+ibn}GyC`C9L;4>hON~$
z27U=STzjb9sdD1lgWf6Wyqchh@Kn{nvK;8EBUYb>4mJSoRF`d<e9%_B4jV^;`>x;K
zse(d99@`Q_yW+P%FQn5*`4IE(Ky+`-VcAU;k`$`6$C`foJL|}<pdPXwD=a}a<l<q+
z#&!j`rA39x674HTREp2q);uw7b;@qD&Br7#Z@KNLJ|g~@=Hgj>j~>b8kT$Q5s5DZj
zZ?~O39kRvQuJ}?ytG@BX%*qI~J9f7RH=dX6u~?N1WJ9GI4+V{L7m1t@_B@dU%mXMz
zHlEBTR3;Wddn(tkS|RSXr-D_<lUuy0o{ne2)Dg&t=+-liP)(|ZY|kb_a^jp#!vZ2L
zvptuai4fE&7y7-?S7C~1r21LUS7R1uRCbBt#yQQ_2D}z)?ZwG9#dd*flJ$&C{BH`l
z;yfV&yE?ik^lEM2ND*9@blL%*Xam7ab=I4`l&zK-V4w9Rj69T3+ONYo>rJG%5zvk#
z+BY|JKV(O9Z&@zJ5Ku)ToX&5*x07O*(FUr~#Jwt(_K*!`PbANPzls*p1dgx*)QmRl
zQ{VY38*zNXT-|7+)py(ZH8wc|bcZ75I2Q;#?OfGkA@)U+dcJxlgS^Dg{Wcywl%OHb
z#4)GwtYN}OVm2gYuuIWShm}pmEl9?9JU{Hss#hiEkRpH3-pXCOcT=hK^LBjgjy+Pm
zXX?Ut676ccJ0Otn`l5e5-QIK9Z|Wc%Q6@-xKU;2Uu0h24pz4x1EGDiT0VSxxKCGrX
zdQpxf9+fYp;ztR*<w_XvaZKF9l-z6MlN>}QhllOck%>hm4&&d?x&2!$?X#GciyD_{
z^X>B-l){F1D}50SW$WHw4)obL)?dbqrIQ$8waT6GRl)(pQ|7YjqOz}}buoc}c4yH8
ze3PW<zE$>ZehKX4<AC(TW(!W=#g|U5gpMQjeG<yw+Yf$}maCkX{a8(&i>k$b$|IxP
z5=s*I^GIl6pzEC|p&&<GzINIfAqR$-BPIHLg?svn)4_|?%eU69Jnih7dtSO{-BsD-
zUefNi+SRASCMCOdxlNRk$o*^pvt4t#uxg7-+?ri`x-eC8$LO@{P6s<e$_iaf*Pjl~
z?~ko82~fowVtDjwML93>d*kVHC9bJ?sKIWEbx2nf*~TFR;*f9lamX8p-|a2YjZTkD
zP6{2o)u%Pyvw(tHcH1ui;ywnVXNT?f_|`pgdU5LAZ&RxNWWrLf$8759;2B7M)!r^@
zb%&1tVE4llOgmjj;X~av-52oaqtiCnjMIg!T)g@Lo9U02EWh_Yn-v29NAy8brMzr&
zaw2F{6;-~UMcf(B4#7m70{f052h5FOl`rFz&C6*O&j#tg?mAs@(^NDpO7**|uFKV0
z^fU7lJ`bSvvwQqk%?=py23wHtmbBsOwkYv|g|mrE0+hn&5*I3h1*`p%Ee%W&0b(2A
zIZqXWeWZZ!h%L)6A!5qmplVF+v$#Cr0U?zZtT>rF=u}&AW#SEHxHdF{wkoGo+a_P|
zH31TCXxx?7=IsB}(D0D0OGNqguYYapV;l?>%Fcj{cg!|az$x(p?8R~0m|OVhRC}=c
zMwN?XNqE_6y;F$l=0>Zp8pt2Hu>ZCxC)iojjc%$jCqbT0+Y;@a=a6-C$eJB{7Ob(>
z3gwmVPF$2?9i^pd_f9zV=-)e{lj6$FtfVsN@=j(<wXM;KA8Xdh>2i9uEhh^SNc*Y!
zSUcjGSRr{Gpv$*Ms!2*SQ_>FD&Ul6-z|NL`ZP)2=DI-r{8&1is^~7N|+wSa=#GE32
zc+T2m`MLP=0>~kIJWwV@;F2;j_-({+d*XDegqx8pKxY{^pL9Ufm7MHPRV`F%2=P!Y
z_W1rORKTA0zc;D+!ZZG5SDVtmKI;hf_kW&d&sD1(k$XHu;Gpdda6?l`{}31^7eo?3
zL~{{Vd6o}jnSL=q9`{cKO3LxN$H&2U$@V2dL~BX3mmQcqF;&h(nF$WXx#&$)Zj{$j
zUo}gKB2qb>w8P$cTXU-&$y3sRWVxA1AFbMJ50Ui1?EZM>qsU-)#s+d$Kt=5AU@WB&
zKE|OC4%?+{)<ctTpj$#q=}B=LwPA-A1z|alBZ>WW_?W>|RsAa=6dAXYLbV{JZt12h
zjh*vlba&zj<@>eq&%fSa;{o<wq=geU;d7_T3W=rNebFxYxQkVY!QQOEiIi4|j!4s_
zyYyQ*aNENa231)?d)r6DSMEN0r-Ii=VZUJb>|KA)b=z+5`B<tx&}r`nD7c!$6%wB<
z!lmo|gVPhs{z6=hb3687PV#|2*hdM>`i2I=2kqmShiqj^409Se$c2JW93M!R(LqWv
zvW0(ItxqC0nk)H91%ze5?YGZ!<JcIS1HZ`2r4}z#+^$hsCHjzx!JQ8K?5hgunczh_
z@t0u^okV<{y)-v%q9PFKFTU~Jp7G}m_HFK|Y)@QP(#3ptx+Ihs?}23hFa2lV$JE5X
zo)$tNZ3?TUal{yIKg8-K>K(HmeOmZ++D{dz#E*+GzG^@F_SZMmQXJPt3!g)cDg?7D
zM$0X~YVm5ja`bY}OJhkLFaOr7Mw8Vd2Lh)`1r)9x4K)|p={xP3(csNGNb+58*G5;G
z_K2pfBUcXWbyZhh$V@hx>%A*{xJYtmH$)43T)2t}az~QAvHGgSm=y=6n|v9lWI&~x
zvjLhy<vBQpg5{Rc2`@PGz+$^K-xF?zDH@(^zuh+K>MNHJe7(KU1tqMWbY8#RS4PZA
zaV42jbtLymsZdGj70)DJ0dZo%_S+q!g{PXo!KMW)%0<V1Q1aTh)25FG1B?+K{x@>K
z&4}I{Y9&>j%^VH1`nP0=fA4df<@5UAr`ep*;BZ$HwarC;r^6GkQ#8tRW70+8U+_*^
zSi-1no+C^{SDoDzGsS_MtH<TLyAwVjiA+#(1DqevT!s{s4%t1U!Ro0-uT+t+Zi#<f
zvV<L?ag>%t9P=Wdn$+DeasER-O%1j*v5@l4Ew(JNunVO_oh`2*PuhDhsl*5rrNfFG
zUAbB|V-KRzvv-Tx1}<`y|BxlJ&er&f=isEfHdhDJBG*jP0@}Lli`=4E*yLigKIid3
z6LC1Up@NfFqu2U3#xvI$DgoE9JvbWT3apK)n#f8iSv*X^RvX_Gjj>?MH}wfLtt&Rl
z^>v%PbqVuol_>+hztO?f-he+eX2&@DxZF4AB4lbKW`Z-^8f}Rr<j!c%2Dm9I^D&ZI
zN3{g$_6lEWUGYo@<7Dt{Tk{zT$RQ4Z>Yr|lXG-p^Y;p2>+u;bQ;<U|T#2$&I!GfW`
zh37_*S`LNnj5i_plmsIRrW~<d(J$8fq70S!aktZflAHXooC|!Q@<3vJv&S6_^=(b|
zM08x72ACEZYhuVJ<HM|13h!%ARh^YM&7uk0Q%$<mUMV&I{7!q?Pu^c`uxEV0`|n#}
z&qi~&Qutg2!)BP_a}_cuxwkYCQVYxfq%fSe*9VvJtfiraOx<kR^P@#bMYlx74T|sB
z3wb<NZI{cI1L6URr#;o&aT%a{YU<cNe@{u|dOJ`}%qjS!%C<MY6m;bHC6>EyG$hQ0
z7=>H#53|GGS6y5Cb~{piaaJ}$%$rB;Xgtd$Nfp?i6bU5U!bLTZ-=Xe!4f-;uP(be&
zre}u@#d|MqQK+vC2iRo9ki)AYJ|(9dDIh#*BiUz|%3elWKAOi>Wwqe(vI$b)XdJV1
z4#?GumoHwp#?I$qAQiAmRB{-`<0C0xl#W!c6%#&JTo!P54|o<!Ksl#mCFB}9WN*eB
z=e}K6$iL|M-b#2-wYr}3`gV+=cy$ReXHbs6<6DmVX>q+rPd2%6<pdP5d7Nl-doMnZ
z^oJ0Z{r0{iWn;H}kR8=2Y#u%Dhdz5HEwhh&`I<4i{qf`wu9M_oA~JImep0PP#o(0P
zm;(<iPS~eDxTsMHIuf(=S*{P4U<j0b?vvTtWnbiuc&(T(^J92tyl3&n*=}EDlZ9>a
zu>3mvg4e@@Xy4@TQYnG-2{E&xVF52nSY^2jz6-F3X^<5iM}7O=XYyZuZ9gQw{rS)K
zqu&aMX;eA*DF?XMo$Y_lqgK}jziMX+cSNmeU0GM0N&X3SHXA2ozw%5dl<HZM+y}2Z
zQ&=b&S#=*>eI}R#FBK&I5{=e1XX4V{q~!dacI{-(ptzT$yDpx&U$LxY)aUW<@TMBW
zrR9L#aK?GPq7z_7a+?lTE8Y}SgcCo}cDDVdGZRbyg2I8_93x$xR)LMdf#a4lg+nEX
z9YEZArf{fpX3w$PVrAtH(X1c{w>zY9T$FR|l!Pk!jS|B-)n7->#XHUfom!GsqG{2k
z*uPQmLX6?`c;;qp=WAx1Da<4Z{~By&0uPB=MlBUa@J*c+Z;JC6Z%SZviT|GCt5~&o
zvE7-I*i2Ygq1$u4^9RKK&im)^7ZYEK3E!2&B2f~Kd6_=<?t}qYL289+Nv9EX9HRGB
zy<#<|zhu7qZGpoYlT_0tTa@6Vd^NdUDGgnLLdHcojJ#}1eYA!0W^S-$XF|;Z|JvG#
zDn-1*^7yQH4@uZ26lO(ElIn&nw$gt=QI5e4t%~-&<4Q)NYw(n<Ipf?AmP%<VN<hSA
z8T@SRncyJtgjPf#T<f}O<$n9i2YtOi8}F&DX?>`pyS<BIs~e)7QZI0^z>@E`jg#LN
z34j=9d+@ST^E7Ip+T1yKPl=V31M;04?}0_y6g?=<xDt)*N5|2a&|PuwYTJ?!SctC4
znxnZy4iD!$X025hvU8~(5W2H?hEoT|MF=^^x5K+_0)vPfbi%<Ez)jO?TYalNwWP`&
z=l^V5Ac2dN#a~D`itxOnLJx()NM@9TEHd>-j4oW3YL_(qwlilAqb>)_A(*UP)y#29
zESKhMwmWvvC3_K~lOiY}H2<Q%e$2lS3qw!`cmr8~JTXw=FrtG`<Tk?;Y=XH)$@And
zW?P&&zz?YrpQ>hdN`_c*>9!|Y%TZH$j^Zp^T0fnDQ+4i6d*)0L+5UQ?JsaI*)huQD
zdM>-cD^oE~r^FkSNV&c71}?S;tf%oOF4WVWkA7JC#ocidU+}F11-P|dj8pLP#a$Qw
z$$Fx>{M$~^g_82_tA>?oEnP9ia~<&3g#jdua&b`YjUJqrC~<Vw7d>2_pS4gsI~@BX
z3Rn2tqGAvRl$-^2q?(Uc09Tf`qt%{PF$uA!u9EJLZ>kVz!qNufnLCeSr=Xv~TsbP+
z5gW=)MmMHC=41`~=Afh~j&CD*kQS1raI~73iuvmH>bJ3Yrlq~g=|hCFbJ?yK*L7<g
zv-AFp(}D*KzSa_p#uE{02~)t0-Vb%j!GFpB-mc5`%~-qk=Qw+T5|e>Y`GW}5DYUn8
zIGA;h{W)lF`#K9>+-mQ{UNFg`5m9uIn-SiPZ?T_>V@0-m+1~SM-G|G+>Y%?LP26Wt
zxu9RO50W3?bWpI|K8!ZS2E(+9GkqSZhHbL4)!WB8&jJ?5g-;UC$|)%d_i}zl?9=Ma
zW!244kQ}tnqMy<O>kzs9JX%+3t8xWNy!#@l-_|xr^_PCw)-JTKa>~$zQq2_x;A`J)
z+~k?=H@WrAI4pgeKjzMwW#2j45EGW3Y4&~KmDK*YhxzWo3{XiZF%D*U)PBlQg%m~B
zCO4_KzJCp|wKg^(<q<Bsuy13*s-dP(9R|B{tneqLE}dO9=FFX-fX)`mLXL*9VAHs7
zByC+2Z3sDl%XpdH9k6T1oCR|YfOI4Ki;QyIuJhI^$qlPXkOlbyudhZ?E@e?;>Q|2y
zCX=s*pvpIn#eG@y9BOf>;FdJ2@HTeSSg7d!Pt?|3cJr8fjm?{7w^Xx}w~w3zkJzo=
zx7aE;^V{+l#mYC^?fwh1XeShVig$#C+NMTF3QyyJaA6&^J7SrLU`YRRY?GN&K|9kb
zuy(ZJ`axoQpCP`IC1v4^>dTkpLIui=#-%iKEd71(f82^I9=;<st6D|Bv~b0TXHE_Q
zlTBO#9`?})!$;Bo>x!G}&qy9t(}Cil1k24!V5vNNm)$iMd}D2g)DzHxyQ|%E1BH_H
zx-2|DVb!Ufzh^At#LpB~!_($S9)(_ldBw}c7C1toZ~BLBTa=UTz>Ka~$0h!jB5pW8
zEsbsNg(okuw+R)Xs$;fnY+}`4kV~IbEZTBE5-8+rY(*^J!$U4tt*xxKLARmbgRA_@
zf9S>}yQTu10uWvD@O&z39g$XH^IqpLWYdT7rCaaoM@@&OcSB-DQFw(d2uA5u+n7LH
z%8oslC-4v3FnUnc+NaR1l4kY(H^pZ0C5!1Ef~Je-c@4{5V-7*?D6SVm-*yG!ZHYG?
z!t8E{E*v%Irjg-`81n{eO*oX~vW^<<iB(%k1CAjydQ<%gxj4F;g;269QL+wNiQ!dj
z-d3OWcuzM8@qIL*s;u7iwj*D*HrXRN3JFo6v^xV~bkB&lQ)H#P1It%u*mfmum)xgf
zxppT8FXkxPW7S>^()w{9Xf<NCJrPUfXi66jRqM4U<C*B$DUQ2>cc1c&so5;%ZBI0C
z20{BbXirxiDNl*4cKhs^c-D(gue6J*r0{IEgF#iq1v)W%E~Y0X^lP>^;jns@J)f<)
zDG|`5PPG>t#~P_J`C?+o=9=c3ZQHHKe`C@k?gF>)ebod?!Lbrya-e#4MA<Kj$3<^q
zOCY3z0*YnL`VuGHk8isjuC_u%10k_60TnZ0N2(dC-n5_Rqt&xBDC%H3ka_$4nd<E*
zw|!AS4fx7Bx3}7$uOy1fkoi96!3K0INziaUAOldT_&E}TIHSrP#&H^rXW(Vw{P)4(
zj8(Xxb8;?Pcm!opx!U=$kjki8g!iJ@jq$|C<qZ!5*NK2vI#<cw>CTdusu2rxxg<vH
z%{<iiVvT*P`V<@~mzzX(X$;=3hNbX?TWa1(ET*b`hrJtpPbLf5Z|@}-O1zOctUQpg
zbIO^mm_Sjme)}L6j_RGHZy$~oNuj;gK8kML2|z6Har-!Wk(!QdW#wOI@;;edpkfXS
zMt&O4BsCK0N1>8k_F2++o)yNNez?SwJ-5%Rc|!5^M&qCrfXNyQt&!x5QgT#2RbN(T
zWFOo1RW*wuY)dO|Uq_Epf}Nx<OwX<*uy1^=^Y5^4bFD=SO^gHN;Ja$Am8CsM<@XL;
zEDx{=3VZq?ck)kgP(N04kl%(@@T&ckb3h5><=hm0B(d)2#Gs`p4{;GY8@F+NTOE-V
zAoLa0lLk^hiXdKj)_Ehj`V81rXD7tX{fVF~yV~D~yK2~K*PJbE&<aY0UwhUWxx&Ih
zUC0Sy*PRWf2^~AM^!l@*HWSd;Kz9?OdZyiQHh4rh#;!&tdE?o_Bi3{%A<#|v9cB^f
zGC(c18i~$sjzK5+Wz=8QMEd283^<P2t=SI#d=!8w+0AY7Oj}<x2pojl{aF!ja5*YY
zCsSf{*G=QXT-n*D=DeV8;wn!UL?f)x?x^}MmaamQr=1P9i|Wk{9VM#(e*0%DwiC;%
zlo^Uw_3YQEENq5DP(E|=t)ChFa!?T6SO=2qtlW{7ZZx-Z5|X5+uX_mK-RWCY*WPV&
zW3WPNL?B@?ihE{ZEHO@cQ}}(mE0&g`u!58Q7P;GZX1r(ioI7p)*>aSZwb?z6BZRS`
z6pL9h0{SSoST3p{GEicn$@p$d0?XI|OruMyugKWR?Bo)-%zH)5L@O4p$vlm;zoP2z
zIIX3KeZ^K*Eu=hJm5{(Qh?t12wx$A96Zibue2rUQhmQHW_<BFatpm2+SGJ|wHU!kX
zu$pKsiMWs%V803nu(6tnoA;@EXb)!R*cGHTvs&NBOgm7Y2cy2QsGF*{_1(w|jnSiQ
z6gVuXSdMKuTV$e@ZKN@hD!bWd+t9p;NNa0#IH4P)*0r~U8%7a;mZYQVO%-Nj{XA@4
z@l3&`5wc=itBsxY8{6hTl=-X0cAU*yX*)U76^&(&`0!N!#o}!{69hON$o|j2wkw{A
zJ3U^wjia`^dgg@zUS-=KbEKOyd#*j6NQV_cE@}!e*KNArp78Zi6#hTild&I#_rK5w
z^QtiaV>&+N&U?zIR_ekv*q)pk8c=bGV)CC(2o)!2R97)BZjD77!2vAvWze4W$<6%L
zOnc6^w5rlzd*dsSUkPQC*-@X*87fgMD=7H`ido)`k1yozceX?B?8OQyUd2=O*m`n`
z4`cq~6z|K;S-D`f9f%bc1-SLbXwo21Ce6K`3hX2<j#*#K(=+e7yN0|&UbVvsRj~g^
zOu~H=WqTU1qdo~Gsx37)`g0OXWnQdiT{cTvbsLOll2(=SP}%r~;+eC#cu_O(a6FUS
zu>wX!=|&RC&|i_-L&C~vehNX9+o*W^F<(29wb<cvIUp9;Gtm0ikay1Ks#=!YxbNTW
zdB3xXyyT#zgv%g6T=IrgoO{UL3?O*~B8ViyRR(-3y8{Q_4p94AF#C=x`MT0%IBk5#
z0pyPK?<TbB5R~4lmQq3$O4aiBlSn|gOZd(QzBY=ZC{E|Y7)<(PF)&q}O8NVUedPVO
zk@t1GeH{JEkSMdV%y6H?GhIwNBH{%;jc0EA_weM;g6ucc+2;w}O_G_u2%ubMAuene
zb*>XV(OWQG`znXWfoPbsupe&s>-bt!Qdz7?bEM!NTn4|1)_@URgS)4G>m#!^n0Naw
zVfVK%pWi0}YXw@o2!0VLe(;YLHsZ9_Wj{K}et>kM)c~WA5Kx}W;pcd!<0@!y1kBE*
zW${?K3J)LyU2(2(9#Y-3+m+`kAEPQ8Cbq_N&c`V8hO>JrA);)=SD!2Q;Wk{ls;=dF
zP<(vs+H*liS5pU2{%|<TM>8dV+{M08ir22Mx-TvQc(6(E6ydrdI|Zb0cl<kGH=YYm
zvPOAa-gK@oWL>q*c5^f?pF)6DrTG?rO@0%*wHn3Ujg_CuG5$7h^e`vz_H$`4MM`RD
zwJFtf&!B^Gq@S7-K=Zua?l@Oz`|u5d>6@0Xmnnp6dUm*=0c}yM&4><Nli|K`GjkAJ
z4dp|J)Mm4)*}BJzLSlTjbF#;-hZ~wPi{BYv7kme9P$!$~+tS=(^UgWHb@}BX8AuqR
z3PpFF3y!roR@vQEXP%v1fv@uuL1xdMZTBRC)U{DR&-2c9+5%s*C=HNabS@a~!UZc1
zeMzh)j)r(;Hs}->xwINkv2tu02WeS6lcC`ZtCp+j@<b(k8yY$u7A;)ii(J0aR{HRz
z%Mn{uEvZd)Sk_d%4Tyz7(RtR^=BBOgYG`j{pVvhr4vQF4lrY1#-U0oqlHTXQG}};h
zGz<?>?iJgZ+oX_Vo|VK^aQg%HVAYx&5xjYctD!cY$xTuMyRM7+a}(}%_}71Jo8rAn
z^$>2VOT`*<1UN7WI9!&22wUQfZ@hT?)ZI468?8nf+G5=*@<XeSM%q#MrWg%tl6+9g
z;p&KI`tmfWs05B()eJ<+-I}2p_S@EkbZxV0y=}{%^=&)oe@7x}CxvO-HrXT5w@eYZ
zg%)GO&UhvfptnHjsO_rS4dBY!Tk3l3&UV;;6lx*j@L1Ijtr28W8TKFd_05_^g}*0q
z{Dn#u*kDg)`@*34VR|Z_2`#jzP;%Q7M<xhoAxE;<oCT7~Rr<7J8-<Ihxogkl{vmHX
z8@=gE`TGrE>3q(KOFJ^gUWbpEdzq1*_cwE9&9fJBYHY<A!^oNb#dub%hCD-2WIZ|9
zf`zs(q5V)R-o|zyX6?e1-=5;c27#8Wu)S4xvd&2seaZTKB}EJ)S$j?bELId^KQiMH
zAMnCIaRiTgYqFV0F7Ef)FYf5Dfg}(FAlaZlUyV^@$Ui{D#xP^Uz8$WhM8F^M;flkd
zvTW2~X&=oAaZ#SMu^bTZ!x}qRp`mUGE$~Pn!uiRGqm6i(2%<3?kJe?IwG~2LQ2dll
zBp`}8%c;<%>X}@n6vlbd-b{dENsva-jc+3|VPh0NWN!s%R9E!v_+9`X=6hW|?^J!t
z6zk)>>)#OA0*7=il4r;fP<vhk^~!OM70PJye!_P4oH_PEbm?BfD$8)zKFnd@1!akq
zlJlcPjrP_CG!-A`XYfkmKC!9TC)MhdRjjbbe2l)QK=!148sFgaST4R-Wx(^Tt=V2<
zpT|at<=t;z<Vou!dq2GOmpSRRkb$rKtIa(B+W+kCYO`;uY0Jk+ZXNsW+r(}vYkuc{
zQ()rv_I<VUiqI4!<$|aLBtPU$)7dSf9yiU8i8HkdhyAHK%8LD$YD8hJo@xeX1p7H}
zl{RAfh$`edwbz~3k!lMH$iHK3xQgap@p@rANK)0>waKo0y|UwRW4QnSRj(IT3Lw;U
z+ttyL40f*CK3lpQv}@v1QtAM2F;yq*+SdyogE7&rdp)jC3Yc`ZVRyFcy<eTYF1x{}
zfu=}A<i_Yr>U5=0z>n5$dc9mJ4P@Z3o4wyf4b65-_EJwY$gR~67|z>XcP2qpU!KHr
zWc17uo>uzyYQmIA=P0PWz?9bu=eMZIrpDL05uw+j%^*>+ZX)O$vpakWP3>(qEjof*
zFDfXR0H#+fQ1<IyTm}!>jDVdxjYel8W;Qe4RPHETa77TbS-$F~W;o+HuLt9)gqF%N
zOzG;^gHJ2@p?NU3+8^=9QVe+Jw8J*fo0jl1;(G}+^sZqL$&eM(l@F`}`U&p0`M-n@
z2aH5nzD?sjn-tKr;C1)blVde1i$w|j1z2a6c<*RV>TIe1J>D~QrY(CtdC@kq(~$cr
z0mU&^fG+kxTj3o&@CU_RuJj+oTj6N6Dq0tr7_jR4SmQf^2F2D!Tac*1*7@uxi`=2S
zU+ewnuOe7@L*iqpRVt+07@%|F>Xs;>9xhO?PptThn-i@zAwgzh#pcxKd^b^%XuEC7
zE#dke7V6<F)L8BAh~z`I<fU@*$;h6FunAMOGTAl9?kcfdYrbsnCTNPKw&ymY46e71
z*F&WVUUid}n!W6_t_sp#Lk;HIw#M)d8);HYItwi2+g_h2MRg~$EO@Sk|H;3$BS#tE
z51kiXI%$vOv%;gOEcHIynG;4*?Xq2oXo>*aop8d%P;?LryFKQhg&}M%mfFW1MOrAW
z@`S^MtLcb6>3^@O>3qnZ^1n$aN_L0WC$49||I<JF&z_F;i(ABY&BZ$;`xlb#GtujR
z1xnAxgw$#T998+R=W@n>%k=m9Y-*a}mU3N0{g9g7e^Jp%oAjc8qZ%d{=6rzJzn?eT
z_Qe1SfQEfTp6<5;NgG@(t?zsNdt#BG3w=IjTYbG9uIA_2s+0i!NS=nyhRrglcG}Tu
z5RwG~*0UFPTYvTCK4_`)ngc#D(ze}egYn);n-b)`WJ3X0aLa}(___Yxz2HXh@*asd
zx(1IHErTE@et5JRQ?^vEE;^Ra#Qg^V&b{u8M==g^T%;9uiA3V4LieS2I-g&%H#$Ir
zkmd0_9W6E6NN?V36V)=#(1{lFs$I$roIU6F_U7vo*A;rexuslF$<e*t-tzO?Ou-6)
z|LtnA1jsN+h2MB$;&&V*Qwfo@cVjGcJxo$4bKKtZy_!CKroEr@Wj$KG*6@LkgTjV*
z&u05Drw>Uk@hu-!1HYDOfWT6T#UCfOkr|-TKJg`C4@DFFX%eH=UiS907`S*?aaeSx
zeeNGDUbX5EtL%#;8t_I<HMO8DH0#S4+AmGfESQLLU49k)sWP9)yz1J0?bGpUp!Q9)
z;9b$;j6n*%%~_PoTUs%)FXSRNH@DT=_tB!%5n;nsVfcqc<_EA4pzy)%DOw?Kz?&4V
z)PBmoi~0v%5zg!9{F1xXvx?dIiXlRaAzUDT==snt;IK(q`zy~+to)1L4&yy`RSfOy
zg0yTycJ+DZ;~;Y|uzt?OC>*nE&KLHGe3C>SR$nUBxmUBf?tI+4#j6?at6G?3M2#_O
zH)NYy)F7k`zwvx<7Y|UBv8}$-ZaVJ<MRmE)ZcaGVH*G~lbj$g|JGIo=t=0H0sWA(?
z?R;D^gfG;4L7GmxJsKW_KYFR$XGd*H4tjql@;9hC)mMhOt<~;`4>Dc2j?1+36`u#o
z<9RIw88MEq;WoV*L=NB5Q&J)@XH)p5z}}hX(@m*30OLI=`|p1CT^8wojMF<Sx^pL5
z*G$dvDbD#V{&;sfj7WAyPF$Ott&w~70p?T4YV#7z^+Y{zAKZ05bd`(vad&*6BP1<{
zyj<r8*d+uWvwOS+1GH|2Cn~O53jzk7wvCIS2nI^{;d!1GMXxxLh!-yL{7d2+vFe)I
z7#yxN0vP2LOlC{t8>%RQArci&+cKZY=8m>*viB~J9#FcKW2Hy{$>oAF31@6YHL851
zJx+9GE_ODt46CXkC`DVQb<oyKHu$}*t$L6%h=j+Zwl4P$(>`RUwxbDYaD6n?AX>EY
zBW{SpqRI5ryE579V(7DtxdTe$1<)UimW9qiW|cYAiEVAp5I(&RY)}rq`ow$E*}?X1
znw+E5)bONdtT6{50_`CZX>N%D@+?>7$gMdiUIcd^wDo)un|}A(S=L^?-Y;`Gn{dK9
zDm>K%TIBjzU2$4r=CaOoc_)|}ZEJ$G9k;S=zJ*?%)^;Q?b)9r{``!OYwW8u)$Dk)|
zXEtn?GoX@h!zAwVBj40*yM46T^L}rS<=XFW0X;kI@jTG%6ge*OX7{$JJMGDAi*;8{
zrB78Ot5O;EE9?w=;+aeahuFX|Fze|!5#l!w*)#F$5K+gc?b)P}x-9`u;a^K!;&V9@
z6otk4Y_9{R4xxxDIo^x%+IDQV7owNq1<pB~+_4v9K1#?VF-;1Mo_Hqvf%KVpo!h>A
z#^osnp4b69kf_trR*#U>8#`IHy$m7@l~iB4r6Z}T`y5DB-pRw!aS?y`N~Y=~=O?Ce
z2?4TFAr5Ta2Tn-8F8h`rU#a?ELj`;^7oe&0Octenr(l#ir@GGhkTYu=8i_wpQkucU
zU*sX_eaDJ{m~s_16mN7tNmHw&Qtmb~T)mMu7-}lCzpjrRiDe6&OZNJZBlfYFNe9=_
zoF*K!+h=fak7rKb!38eubMXvd6^AGs%ItjN6)F_GG;BQDs0yn}_~0^c6M;37V@}wm
zn5~Ef^JLGG!+kSP4^f@=R&*qZUTOhSh1uIVfpUbYi-u3oI}Xp<rp=q}-GpbuHu(Wb
zAM;*Cd#l>8+9=p1bhX^`?<c0#JWMrew9p^qf+A1ThdC)SZQDoZL$|WJj-=Z%ihi6^
z0w#nyLGLjn#+0N&`ZV`Oo6oM1Vi4jc1>tAjibNpy+vnK}jGSB&m0$XcoEgagn(9gD
z{$)0jV%Rs(zKRWY`=r?HjNUCAF0GXN<m;FO!<U0}x*QWUp!Q83#f4h?w=se)w^IlQ
zFVo_?YV%Ih)AtF0l^EoTF5-s(up^2Giz)@VEWN5{^J9!122F?r%ArC`emWn_eK6Q1
z$*KJubBNGq86P-ph2dFPV^>^=8@!%0BUfGsuBo9_@hBJ?FNE@quqD};C#&~deIb~#
zCTcdLvbg3#Xit#?23YAUVgWcfk)<W>q-!rE!=r0emwl?HwAxqgx|~2grmXFluCCAN
zwKg@_4Hr@ohZ}p7^3L9PA-FmS+DmrR1?PB+1`n7J!fiL_H^9WLVYgf;S5y-f+-|*)
z#x~LbZnxW_w>~J*8M{4t6JeAIisNllq6K6U(RpgDrdSO@MW=pBgW|TlJG^`1AWItk
zX$i7>5nPbfr~5?ONHH+O|0MsN&8!xoOaCC+g;AT8U*fN!#L9DGTS_5dKr8<ar(Ac&
z8@R_3hcZc%=3a0mWb2L^n-^nf$`TWmbM&r+MJ>^7ILJp6$vd!Bvi#ZncqZml=@=!_
z-IF^;23-s(q=CpCTgnMr6w@gRTNa~?%bpWrN!6DU6M9XfrM_4AK9+<BFeuBai56d5
z8OCgR_3SKiXRvGW4C9oDyskG}nG>0fL1&c%cq^r=*Z7~ae>=z4=HwMMP!GDT%Wvuk
zQ(B*>BW{hlMfJ8J_d&>npVY<}wj6AxLftcaFrJB<R}h6rYOTY*y%7<;KF6fkpr@FR
z55)iJvVhQAW8j$>*dg1Jr;zg>$_3i#?rY9XThIU_V6D+GNrxxel8vmr>hLA@y2Ix{
ziUHD+<9}>jxv&+b;ID0UI5c(bXy~wQ)wif*ZHKSr!d`}d#HS8FPa$H`WJ{I0GhqTr
z5<9B`b-NOh7bpk8|4-O%?{l1*eB1($#X-_$YJs}^wT6BCD;>+@)w&eJRAeJi2Gfu|
zaUnxVny?rsp)Oq7lerq0C1s*`Di+X3lD1LX6P@)!o%Vw(@<nzt&C@weqOpiLX2YM!
z%`E!L!XnwT@l4a)Cw&0^#m_~bT)wDvTzTWU<n~5uy*V5d{P_!!xh#|2DoT!0(7Bmp
zl&bHr7qidejp+tNQMM!26K?=f<$>-7;C(R~t16rrw8#!rYoKl~z&#8C^j6O#+xbcE
z%Ws24lF!QF3UcV<?Z}0=UrMa7hqR+^i|>js4?HAn^(T^-Y*_$uAR4*h&pFS*c-E_!
zeGnRPz=o=4B&&&u4d=ITHnRN?gJmO*g?O>l*l5D7v$1AVmyJbp4I<Y<dEB211S=|8
zb^%renkJ#zrQq|9F{&?EvdN4mjuh7?|H`2>k&|)35uvlsE>+Kt70;kC=;E6RD@<@(
z8!*t8c;mOKmWq7T8Ky4ug^B-AlrZh8=!`GucK712cP=<%?59}>0J>sh)T9;DAg<>y
z04JHDy_+55<)m~|Duel+6S_qWE%tsK+W(4B@<D=&_>$H-`!EknDR)K9=scN4Vf%6R
z3m+zXyTUy_apHgx7$g3t(Lj@QLHUS%7FeUKkz@9Gz#f7Kwt4~ww;AddJ!D^0Lx{T&
zpWkm^`qby&b(eh=uR|jWkvwi+=O-vSTw*_N-^8E?Inamf+h9k;V&MIR1wG0RY5<fY
z&c=Ng->|{>DSls#B$D91ML$%CP<dAdR{nU|9OSL3r1(Ew2=SvhSSvQees;*$bP{!C
z<H5hw6Yxto8=;Bg6YgxXaEV>%e{aW-XjhFFzDC)+Ydh@fap%H(+nq2<8FbXHi8ta-
z(k2c@v&+|J4|x2x*mdL1cR3L0K|D&~&JZC`k6%BYoRd@_5?5cf8^&WXUi?drL^npW
z(g0k9IkuakYgZF1Bwra@Nbm-nl%DVA@nVs{olOY_+HrOEd*YUy;D!GeB)BztDZUYf
z_RFj<4#<z)ZL!eee56|ew%gy_J{}D0>^XC7%6KraTt*1S^)@xXV@b+gGZ_nZhmS8?
z2n2E3c;N(LVv15~dVIz~l8+LuK7xw$FEhr2?|qmQpWHPwa~W%tRb<w9;VozXZl2A_
zhV`r$rTU%YSpiu=KW)uzq^ss`nj0e(k4Yw|TjFJe%VC=rZ;F(hG^-?-y(_!!BnLQL
z`rX;GsKA6DqXl?pZNB3a8JXJc9^bt2o>~>IQ<i?rps}H8!yhFCcu@dGIf75vlKhH9
z@rp>2JHgV#jLA$z%c3K{!ToMtp2+=&mA1nFuECMlRz_QC2zA5D-*VM>5lHTBv^BXe
zcnFW$+Ny28*ie@LvaQQaM%7)j8J1)Hc*-6go~e)rp@;(R$>I(=iq39`uB0v$2_>Sy
zHYVgCB+Ac38ZW*f2c13WJGq^bIkhpP{qW)^tv)^xhb`v~35bq~n_>XhWEQKvHC6*~
z@zS;a_10S)<LVUx3!*kV0M_1btw|-6YDXc|)}BBp{Bt>C)={;hv@H$}_15Ln6eEk;
zV{1;PgcEDcBeu<-5qC!YhlXM;wH?t;p&XKO#4JAIo3)}F8_mvGfFyv+)F+VGUAYbS
zb~KS!rN(yqipG2J+2<a9EN5N93rpalJ)XT4%Z0n=mBVaLRIU5DLq(w|LLJXY9mU<!
zheF^rd&-Y{Lo4<X+v5#Xj+Izjf+;y?PdgZ}o!2AAJd=}D`4hPvJez%@a8tl29JW2@
z>&20z#rEdsE0y&6`J5G(MsaC_F1`@0Ymg;g^uOgJZ#}vE<xQ>K+iYKaN>N2HSHy!7
zK^3hc*r3<H?NnOIJ|A^4OMnqP9K?`R_u#Pqt1wk$F38wPJCbu;q8&P#m_QmnpaKZ`
zb0*ZvL2Wt^Gx3X~*dJ`ro3z!!#lhkad5;D6PuXzxzY+(38}WI9V>rYo<ylFRdQRS0
ze5X<y19mRIYa#N-&c|NM5mI@$C;-iIk)mcizEVnyBQ}vc!dZ21pi_3qPv1gB{5SJ{
z`Nb8dEFfOS?5#NbMIb7kbD%M@7x#mFJcUZI%y0W5Nc7Td@8p0m@Mpt4?cI3h{Hc5=
zj@f(FGg%hO0kQYJAEcjC_CdVY2(I=#Vjoubm{Qodz^fnw<pWXPz>l)eqPB59?c;cs
z8g_UZ`y`$@ebL(b?NbL^TSsdHrKvu15Y3wNTl+kBWfRd)qaF5zqZ#hlkoYg7Cn-~l
z076e+`6}8ehG}2t$4b`(asMXT>j=t&mf#ut)}M{{l&DZIHUC|OWEg@dhZuVbUyG#h
zLtgrIa*Jx)TGQz{0e<vJqk^}eVx06(LWO2OCn}R~qm4v`ZDto;!GnqgF5!xc!9XGC
zLFFKNS6(c)zmobl*i{#ui_s=33WzMAcJ;;Noe1;O`N0}iyie?!i{&OSNg5QUbZvYs
z*-7a?l%DN6Z-dCHyj%P2djCz*7QfVQxEMSVal07P5vJ`%pAq3{v+uN<E;`?JS+*y7
zR=A&MfxS6rECppb-G}X#>RI9G3Prxv!LU%d-CFE6Z=v*-U8TMVe8lbEqM;cUropCU
zA0*3A=?2uxHq~2r%1pZ>W+uWU7Wz<8fxE~pP298uwcMbT+S;Z^pVH-IqTd<0eu!(A
zXD6O{vE1Zctv2goXxBPhsI6>s{Kd`<cIU<5(&}quXQ{Wju}Q)?iA6`Mi2Zr7|B5O)
zW_RVJI=7PvMHbe(a|{y1W7U}-bCi2`NyCD;a!)l!9Lw9fwkuU>_M@sP?J8HaC>zz3
zVgjO~ODY(a5K560Tk6vp@4*4~AzKzLQnERVs$+QtrBg6BMY~`tVi9Drh9O>AVWb-d
zCBuucIby5gTM-l~wb5nS8vmALT}mOiHhV91LW`2m);UhLG@+lDPIP@P#kCmH$!v%r
z<wGpgehtCcxsZ*ygko{x$e?a|@M2}}Cj2hc{<XRC7C>+HzL$&1niU<C0uCa-n5`TA
z?V5GAB>}yvq{!t4Y0jw>jF%LXxYU|!gV>So5kYpu+T$7JNQ%5n?FZ|)IN>tWg5}ng
z=aIXw885o6`TK8CD-rFnEn(4CD<hq@b%%f6BG0Ku;t=R^mA2uPLbHp1jUobUXLd&x
zMYsdo71PlMA>Sa8587@&?@AR|Z;w@LP@)(aFAmz{x!)J%?JEvT?l1~9c*3WGO1HzF
z{8!H7WjR9ZsR|ORkcEH>FSMtcju!)0!0OZSOgC+b%;w}j6Ei^6DN-U*x2Pjm*t1oy
zuc54zQR<LAm+-BxwY~XXJTX*a2%7kOt_LAcao&|C-(E<>DWaap0U&uXx`2De7$6?H
z*Lr*in7A6-7w?PM3epQO2OQbDyF4pa?m``90bz&x{Ob~Kbh!Fj!Tv=%Vn=e*aqpqh
zh(ZmH`X^X@D4&?KgbSv+t?S+fq94d|u@@<Y#b7)`N`W2~lZFi?y&*v>rE7-$=R@Fv
z!dOOfHh_y{4<-5WsN=_)rY_Fr*u^3rPI>fi>|EXxD<~#)KD#ZU5jLLvlq{zT%7RVu
zPwU#-?UI8B<yYf&QqSi!sfUA-Lz+r^%Ud8x4xzGrJI^kYCW)1Iw1A>*?_Qi({ulQD
zvG*>9l2H5v#`bpLUvKXx$w%v5{G>nd2hb=TOb)<@F=g`E68ZU2LT5EQ_HiOWa|4!>
zdix|+ETF@~6C8BPK203wD67<U@gC-G8?w*hO-3A^K=<dWi_c@oLxt9Skx;s@hp)bj
zF*xebXE*16mDEh3c*Rrh>x-e>xVIimlgE}TH@@+{+{4bkb?j4YlYMv5m3kPEp%8A9
zBhXm4bP4IcJRJUe2ZAJof3P2X+whRJA7e%0tmS$-DT4l#xZp}Iq^vXcb3RKBoC9bR
z1rJv(USd~FgsN#p&DQO9W&WyZA|>Kg6O%Uq<tp7idG$oO)+u_88iSOk6S3dgN?hFG
z+i+XEHv2(~ILEH@;r?#=bi01S`7x1DN$Hb8Qa4N#Rt~RUu;IpNCativb(40ioAOQ3
z8mk+r9iiWLbJexXv10NLFaQQxyJf-|L$WWjnJ5K_1G;r0RbsMJh~t+%-xG%@oAPb(
z4LVQAc2O4F?YXK>JX>u_H4|=4Pz1-FxK*b{->z#cdNAF(Y?j-D?(n|H$)?-F`6$=+
zwD=|r&zyPL^iIP>NR#lAP4}^diYQfQ<m^M0{hHJqMWN?fx0yKt5}<@iDSKu`4~H<L
z4cZ*v6Ut)Pov~B_6GtkD*BP5TQP{YL9<q58g-`A%=}0hJ-WAIZg=}$Ab%)oxqaRr*
zlwFYv(&ooARDzgpitmBl<2#KLD1Pw^Ch|c1t9P&nT9M+E3Ve~HUKZk95`BxS|0j5@
z!(7=*eOYtn%(i8*UTK&P+VX%{@zV#U<mN2c;l^4jD^3Jcx=^WCI=0%%SZ86BSdCYv
z#Kvo@93q&#@S-A6dX29~6-c()+G@c?&s<EF>n4I_S-8;F`^ShBDvmyC8xo5Emo!O5
zbSYNDjn(o-N)<W<xIgGUHOnWt*8eW~SnIu4R2-YT1=URxm84W^^LAQe^d`Tfvlnn0
zfEC@6ON4}!j3%5LzsZ^djI(}kt^OejT!bu$x%ixRUvbUD%Kgxh^Hlb&a#qh;S3C=%
zi+XL;w)*akQx{7$__ig~X8-25wj<g{U_j9vKN8P+VN2xBsC!|jueq%o-OjGu2Ii#n
zvAgqhQmeVy9*f<VgXUp-JjY&EQ-|;E6Zz><<qqn!Cw;kpAW_CsfhR7F%j%+FZyc2%
z)-Qq3d%UNaB?|S3J?+ED{Te>w8K2&xQ>SjQXZ-~I=lS+rP6e)g$oA&>QOJN37S4sT
z6g@vN@mKPM`Zw6o%Z~y%6YT|`blc_{doh<MufP`T@o6BYHrhUKRfi(zKmb(dT-BCj
z8S70Tb9&A?J+?l7hJ4u$<0}s{T<`OkUZz^nW|d$pnM(ka?jj|8IO01ybNYYz^P_fj
za?s)`Lw<GZp9ozzayKj<^^uC!eW03)?scV$=(E9G*~(^R#2>198?F5KhXcGXqdb4j
zMxuq&Ud6qS*l2*<37j0J<4&=$Xzi5t;CWmX?OalV`#U>o;34gNV(OX|OKd#8kjstI
z1QTOr6Y<O~*Tr05m#Sw9k~@U!581FLLS2C2P2t+O5WMAEh2mzw-u7`w7vI%w@8sDm
zl36iRGTI@O5bt_3F&k6{sxTSugG2CsPOs<=HN64*Ai9$IPLYq4gJnVrSNbqo>zbAA
zjeqT<3RsG*6v)z@>|-BTexVetu0DN@W~g`3&wQFxr=8@hRD|7bpCv3YL)9y_%;(-m
z9U6g7`y$So29cYqAVqjy<<{|KjN$uMDp}fB6Q!0)nV3|v{{yRNf!6_ODn>4PKU}Dr
z<ODYed_9q7gMVgC<to{avb$t5v2P|ON_>!JcTgaL66AdUN8f$xScz_0c>7%dRXR(d
zw)h#pG{4Ucl>HSSjvr#pY#p=iBiqwU_TwZ?k`n>4sgj$>>xY!yXexh-ZzMGL(z~+H
z{_N)-0%!I{DA9;Bq9*nmp_b_;*Ecn>^Cub=Qte#zhVy6K)Lawpr>W&0v8&^a_-(~R
zpHWDB@d3K#4OjVWvTNT64IBn&`Lb{b?YcMO@@wg~>*M{Q^D3I_*(Pqt_Y3jo$}tMk
ziz(|ZTTBe)8-+U@?`cJDxjBa^y8++B8NcO?FwEhr+svW8HQK;0LD)(fLc1-xFKTF>
zO}bUz{zh=ItD|ObNy;}R+PjiWVAvP9Gqvie@+S>o+)<7DN|6DT7iwD0txJkal5V{b
zd^J=+40Caunc*Kob=eoivZ1V)8J#J@Sy$iy7tkz+cFp}Z#~}yfz18k~BS_2Q=B6EF
ze@AgYH$GNISeb4ViaPHX0QMD$aXMmmd1s3{Y96+`t1m^=#VhsN{CMU;$!;XO=Z)Z*
zh~3}@#-5lD+G@!JjVWW%8_s3RwN%OvIV)byr!I+JlgT=c+`TlO`Rrsr8M9^8PLw@>
z)7?*RBk+XF9l6m>qcgS@Zxq31!{7g%t@Li-(%PHQ&6b)ptNisGGKa0HdaRr`%luB+
z+8AG6xng4npt<Yb2!3i)8wEDBHS2R+(i3*thN`z>nZ7JX+s53OdR0|LP1#`&Rtr3Z
zADuh#)>hACIPh~-?^{wUAAn7c|1~8~PouxMpxR(t0>W8|MF9Z4MLj5!%WKx`ZO412
z{2$hut0)Q@lKoqIZkR&)<?+-J^YBwHIr<gra;Vj{)Z5m0Eh$~De2QY(mN<nN+Y8k^
zt(`8qj~&_j3PQ~Uv`2h0ZKOZ4o%vZ@a3%3(adzL8Q(pZi+nwY|B|uTIIT)H+T1fQ$
zxc^X^Ib8Ge1iK?*Ia4rp+0~N!?Wx=em>Q)OLr=aZfXCrLf^aWIowe|eu%~@=LN`(G
zky|2(&_nRA&nDpA!peOs#~ZvU3e`Mj&*h<S$6+7K<K7CgXN%7B07BUF)iW-3fc7d(
z`3u#N5d|(vMkr;tq#)0~=!jR}RQQJ;UtqWV0QN<1I$%l|A=kqLiKeQ|XuaP00TPK?
zpD(7dyM@IZ_D@|RZT85VBfjk$|NPf>)Z0Q7p}zgz$!2Np2Xc{6vaW`0HkdoU2Des<
z_}Wlz3#xXUV5Dz@Ef~&OZEo)FvXPiKsU9!xV#P|cGMby!)S3a}V{v#97tSRjx1&5N
zl=8gyC@(9H<9IFy8v)s=OBthyXaQ#eCSD5c<0@uM5qOnCK#aCGb8gF4|6zr_l?!af
zr-zxpoo~AuxUAoCP?Y)u?^bi_z^p1w{ChEmOmDJ%B1Uksl}_&cY%9ApnHB5pgG8-0
z4WxnhZlR%F_EBysJc%+#Dwz7?>Y2*nVeLeC_#``6%65L5KmJ5?$Y(JPai@xtBA&7=
zxZgg{%^@?o2SQ3u@P%J6Ev@#Y|Fv_veN`Qll2HeZgfbE+8zv|Z75+MABtR;Pgg5q$
zgNh~NUT@#VEX1)L<SP||EIrtFzOR4mXtwWjVX!bFFUP>(AEKL=OHQ#L6E8?dso;Gg
z5`K!0R91Vmh-CxzvqO-)Gru)haTyG#vW~;mrDSiEw8Uiucb&ZQlCv)8{fmE?B!R0g
z6{Z940;H#_FU1u=TnW}L_UxKVg|CwB3$R^#DflWc(|Dc#tZQyl5)QlGC$U(ObvMK$
z+z;_!*;o&kyD=u=<x;%(<4u?1X4gHB;K0B)U-Da3H>9{pB}wtWfhabKTXL0R7=XWg
zviwT9wHg4o0^l!m{B4)w7H{q5cA>)W?eV$5Q&ea$=kk5+w<*2~R3w{hYIG&W+GBA>
z?2dRQIglV`XkxETyHvQtNB_#E$6HNR-fF5!G~-gaupeO9nU{Sm$4%9%n3X_9P$V7u
z9EZ3pk{pveeS7X-b3X@lZhTz;Df~v>ns+I9LR`-j&vn;iZ%vyv+1;010JuDuN510%
zspzePY<_|Vfs3+<jdqU@OWA9-VnI&2h=XDl`fZUzX${%RZOP=<(sS3_QvaHC^%#_v
z#n*=}Qq7A|<zBhm*Vu-G7PrQVXi?(#pioL>wKBU{Q?t!hd2<M#T%Ffsb9u>f5s^mF
z*5(wr_f+H1*5!i+rCNrlt@jS=8g!gC1ai4BTe>JHCWD47OdG2qBv2K!vj=^}K)KCo
z<2}e#sJZ;3>*IY3A$^;2Q^*xcbuVj-_EN8aM<O7$#j#>Dhow22cM{OW<Xe5c<2{SI
zL5TK5Na>-nWj*H=?R8gl3~&psW8${0xpbr*E;F3l1J|{V5-|@S;F!6?yPhf~#v{>{
zv(klx*-js>=tzojFyN+Bqjl&|h6PW)!nV8Wqf{qFZ+g%k^D!R4c59EvdvW<rpkK8o
zd}AJc^zSKH`=lSFO||w^BEan5{KocF#9#!4gz)4LD^R2bc8&8`1QRSotIN>xOwNPd
z?DLlPY#xl2o!i0i=b{Ohm9#9%pl^HgvqGBP()N5l!xV^kBKh53sGcdFP>F<&*^AkI
zO)XVot;bir2#tm9tD2Lz1#BAS1RRKGQly+@cxC18jeU}mXMa(N^QK6jeMzm#O|HB4
zP`o*O+1~2nQ{iA2x{t)0VnNq2kvX^YcGNNNH?wA2e@sKBZ6!EhJvNY+zyk^eSG>z$
zE&#xnGzhHs|D)|~V7xA~e9seN3~P)rt}(_KW3$AI49x)2H0{u?AR^L;T%c*kVQs6Z
zDyp#Fre1g(+c1o4T;p;XV~jDzlu$zPA{2-S2xzdHC?X;vA|e`Nj4{R-V~jDz?C1NO
zbADC!+<W(CZ?-4bOa1=O>v_&|p7Vb8WGhHEkr8Nf_9hXpUjZ{%EMWT`3M2-uw*xVV
zun-ZG$MHEk6z4irVeJ%I4icnq&x=rUG99i)P@#(A3}plH&7;RKBcK*IQY}{=CmtC0
zdO~oZuPAVt9cU=a-_dH+D$dQ0#kagxq${@L4z+22q?C2;w4sL3`oagD^k1jrFn6jt
z66K0@ZHfc_w1We_eVLuft^J{uSpKus@>G`xuP!<Al*2mbOTkULyIDnc#B`mnj#gpE
zCATMiwxXr`BySDY5c@O<L4!y1+h^4*at|#eK=v1a@OjQ6SD1LtzX%{GcfPJ$aaxaW
zpcqa@<-hh-Vgh}&ygK{ZS0SF1GKaaOJGd6#=0%0EZ{o-Aaz#buD*KX5^vWFmy#sW@
zxEt(;oM3TN$6`vmlX0tBMsEpzm(Rpc@l6!GkEN-S($6uxE4uaELOr+%2hHAa_M$ko
z0|X@ymAw+Ipor%h8}>$;)Am8>gem*&f;WOYs>4@Ed3@}`H-aCMXGV(+k4{oaBH(h0
z+w7t@Trowe0yMW54;fZh7)jPHdE-bpd!Rgu-=(IM;#>jkm%R}NSi}ScAyH8c8KcuK
zug2<ytUI<`@rLt$LVEG2%7o)I>&iC@6S4{x;i_1Ye1UOGk72x^Y)J!Y*Hp_;K{st@
zzg_!Auuxb>TEs(-@OA%5j-ewH5SWNj{#+L`D7I^hjq<6-p^+Hv|C~Uo#4&G#wu|@L
z#(IZ=o;p$-*|_|xr3q}GkQ2)RM$h9!ALNdj#Wu<RIk&6BCi`;ctHq7Er$Y7gPHnO&
zapH2n7dne{dm?Aq*8bIQdSlQKT<SuIQMlR1udnO2X%$SA*H4Z#RDI5Q{wh;j;jwJS
z8-pB|VvuES{f`{i%z%*i#{)LY*FJ;tIX2s;{4?oi=H&1LyLqzpNbIn=)i<4RA?r3T
zzCrvU@&twA)V?6TNpo1DxWy&Kw^Bm}>=I{uS2dnz+9XM@-JS5SMe4If{+AeMn2CGx
zG-^>d)i-q7y*a1|5O6_QvfB72R(&J*gXC{@3762D?kv)e-&NwY*`C}>8uBvy-6CrY
zWR`(LjJ$F{E0F4tEsgODXU3dzsA+b%-9B};wFYp7;0LTd8cSy^wpUC^M|@*2Y+s%w
zsw<|Gg5zIlM-mnOOTs)`mZ$-RLlXCVYAa%hU?61Fv{ls*dgF@Hkt0%~r2F3}Y<e3d
z8pwII*dE9WLuPyk+Un}wmYhk%yn~`e_qWCYvXEc~doZsI`V4!hqF|L(3JjEqzeK6o
z!#*PqM_bVzJn}}Uf7{TzwxB>Q8resE7jk<gsf-fc*4f%>XM3fQMZ*n;^+rf&Q*oDb
z-N|FV8mXo_>~ZfynOxGsU?)_PvnOJl;@Zo0dQq`HnKTs90+T&=^{HspM~)@sOGKk?
zPX{<4YCME8D9FTdfd0?qQNZXcl&x%{r<#aT?7^<DufEB((Sr(m6C|D8wG{EQjnT&Q
zTInious%nl=K6Zul6yutR)8~$-4;kJ9RMf^*CZlBbXs{r$XNv;*kRj!`2#&vxa=AF
zf7YL?a@RmvIc!&aQ#7sO&vx=o?5+^WaWS;pVD6qcc8ODNb8|W`Yl-sX?oAwN$8xsc
zu`FG74*1@F^{Zdmp<D<YqjY+T`+3z-jbNqc>~LHmW!Twa0||+)wq|$`1av!+bf8^<
zV5q@g_ue!3pB=46mqJG}-zGcezas|J*zv>yu^UQ6ZzmE6XaSW)K<~!M#DZUJl_R!O
zd64rwJ^$&PkA6xVwgGy7CVGno;rOUz{DDJftJa>?Q;8GK#Wy$RyH?_Sej6ZGiM|Js
zOFzlJ;G{BkNdNe0-sQTL!jaFSNie!-*k7>Eee(l7)8^S1Ll!30_+UkoFQcb>(`+Is
z*%tfCx4IaEKQdpdef>sJHQdr+-^4gd0#-ac#MtsI_1d@IroGO-b8H~YhDY}ML=8Of
z?1%hg!OC{~F@|EvGKR3L&72y_Ci^LV5&Nf<O*}0@(SG(V)e+lcSbZ-pR()Ptd2W&s
z-i94@%~6T?v<v(%!dIo@zVK-962w;)A}~A#;Gy8Q$}GO$E{bpDGbvF>(zjiFG*~bc
zy4^+1dPz>rL#x{(dVT59!s{%<1^+Uim(x?M@8$l(_}}8Tezb61WE8y5uJnHsnA&Am
z9SxR_EJ{eMSNrf-tm>BHnO<!8*Tj@O6y?POQg*F(yFOedMpR8oqE9%%LpJhgvYfk$
zE{nWig@VY!F3NLV0tJ^FY)$nx%2!(lr5f!kT*3cr%#gA1oZ`Z~U}OFFHf72gmpefI
z=b$G91Y{oEY!ed#H`t^Y$KSvbMrM=an~qxy%}zLk8}nN+t*bWA3pS<trYqz+wwrt+
zvXh~4v77y`nWcP}P0P7(ua#sYN;TbEuvOif8AqLS6=Su9T}ePgL4^TfrZ2f^NsG;j
zVKjiGT|zKzw*NeJ&TTfw|M?G7{@&)sa9-G<ut`3s%{y8dCUHzFX)8Ky!O_BuioNW#
zg+2_$sqC)YCliX9+sZlK?dvWlO&TkQ{6NqA`F2mVcH66j3UVdnsj7&JHd;HR`wpv1
zz~~qOq_Uqh91WG;tu5rJMtf8(LY$`z;IG)y><4=%6Zsw<jpm#lk68J>yo@m4nqwiS
zsup&8E=R%yT6yaTKo!pe7O({lvXBAm$|+QQ8g<0K@%k*w`&UG<qFg!}i~(+|JPneH
zR>hD(0LxP+?rxLa?_kCRLwcnLjuuMxzh7^w{j;1d>TFG1lV2M9inB2zy<iXKj;0b0
zhi>hmXrD@V#T69OEypF!>*2f(dZ)<99?d<_Skmtn(O`B6al<nE=+TTwcOAa_UT%iA
z3M#@(MM)P26I&ZciShyNdR>Kba_t!shHsC>FGK4a3wfWl$Ne}|=GS8tpKt(A0~Si-
zh9~fo334T~EP~%taWLZZ6flY{&z`QnDehTUdcQqW9c@)vicN?m?n#(X$+D@<)>kL(
zGQ~@1rF#?13d`VFHpXDV7t1Od&uO1SL0%xl+HZ-mHt}eRhwX=H+vax!Yf_tS5165G
zW0w9n4Iw@r3?(i3CagDdc|OSNw<|`I2!Ry{0e!3Ou2v$S)S>RRd;H>#9`$S6JGdo|
zZeyuXv46<T$F{1BU8tp^p<C=F7Lvy*@AQg#do{t2%Yq~5;iE3#g#$K>{}$-Dsw+sg
z$XuoavC6`>@{3Y_xu?sKgcspfJM8snSE2%dC@ToK9*vddqb$Yrv0T|d{U<w~J&O$E
z>*wIX#1qj|q@qx9dY*LH)V8g(Qvr0j=!ka6DbwF~*ZQ80rWl0IRJ)LxvRE8qb!T%I
zw<x6aoFkMXwG@GRKJkkH30T8Vs+G!{i2YOw=ucy%t}$^{-DkPSxp&&<@w<oS5+Z|h
z$QOY@REmI6{W8}<-CGjBM|)~37bOs3T=sPmo$=!*+c(hwJ9N=tlR479^$F|R7I)fr
zzW4gNdi&ltF}<Ej0U*i`K1@^H5;6w<n8Qew4|);CW<OPH-BZqR!(ac+ey)g9N#<5w
z(4;nGxiEV(NIhW$%0xKq&0vgFNf-BXyWmY%A4x;nYZty5I+NO(S{wdm;b745H`zsR
z1{J2-R-_2N_|4$S29z$ty&Nuy(V&c2ATEs-8<DIe=LvN#i*H3D04wCIa`~H~R4S-0
z`%fw-uXxj08Zj_^MFc8kcv8P&S5`fxa2#?;?5dc@zj>X>t8?=uM+r3Nnry_q3Vo$r
z>syd5NIIx$3q_7E!AK($NHam;>)s4E%sX%=#`lci8f8}>^=9D^JMpj>9cwJF1v|!K
z$8}YZ*qHc5?jCTxh;|!+l5)k5t@Z`EpwD3T;BkH^3ufLr&nEaJi0(Eq041iHpgrDF
zyhw^ug^RV5DzLj#861R7elvL8rq-pX8-d4-F|FU0p(4$cY^$)1#fovg>CM8&D%%4D
z+-`QDkgkqwGShsKZHwC$+w=-0_QC7T@Fh<e_gkBppk?1Yju1Gi<m7=M{V42MDOhGF
zxEngUZO)tJO<yqE=4N;JIR$A+z?|pv4)pxxdRyS#N?|}-m>a{4UZ9Uv%}McH6{HK5
z;+c?xBLB9#V}cT8suSZ~TNK|=pF;KZknFL0d`Z*LRQXmLn(CYDa4pX(qa>y9fnZZz
zb}!Qu_dsi?_MsAc&c-z6mZziAUOJ?cc+#DuY-;w6wu|3ujo-x$$;<R5<kxlw^DNvL
zIx3`J*+TsYOnR|ne>em<$*m+}uw^+D&me|ROn%SWiiD4H_OU$L2e4Q9Zh4kvgSx-k
zMq%bKmv6QQyajIEG90gVghA3NxVFZ(KpsJIS4HQ%2x~~)@lZCORzf=-P6T}Izi>;|
zf=Z9%{I@A?`BDE1_(;pQwpz6+fjDGbmR>gqt)lXjW%aQ@S17MNp3_o1jhE$#3M4zA
zxn8yY$-HYwFIz;`oTD8O!pgJcspunxBJvZ;g%|DVL<V`Nx*yjwgDat#i$XosF7`qN
z)>o5nLqK(%L9efM0ffdj=I=}^A5VFh_W3uUyrhad-Qw7A3$Cf#{GZof|M#}N>Rnk`
ztv}}9jBDCP+Z9mEEa38F@KoZv9p`IXZI2JQl(%JX9tbZqDGy?JE-_a{izA!619$4Q
z+h^at;2-T!qWS##7JIdNsX9>jciQ3TujF9cAP>E~fCD*0tLzg;{I4dsRvy;ZeIer~
zlAwKXs3v^!>{w#_5a&8i^>^G`65xa96+TZV^7JtTpbI%!fk?XAKCb&tJC#6z10X_*
z$8QsSmpn#JSHDQR>gr@&m_v3Z_aYS|M=b}3v(Z|#7E?E3mOFs-*}0q*)mbT=1d%x(
z-{1oE(CIm}PvTqg8|T{|Wc*L7b{pgkUhF%cy_o_jdbGkp!EJp0W`+#u<<#*=8C^oe
ziYBWlhXbd+NO&SfkWa<F^mA)b1{C|s5t5qhwf422K`prXjeoBBXC$a^y**jM2ow1(
z52L_hGdf9G<o2QZ`rdorx?raL5Zgd!D+3dacG{0|OWQkF+E4!PrEMe){yDD}*>w~O
zMm`220*05E%!ZCqwQJb1;Cg=T|Fa8@Imc0~c-*Y*g^iWw@4{o@!h|*s?`y8`3pPAj
zdr-Yz2WjT|(E(l*tpRj#*^(;m;%Mz0jzd;%($Cr@#|n4VMuj^vut!&gWf#c<&hxTk
zh2x@zK)YRj%o#Ju`Wx(u_)AGb4%(Gf-}SO2NS9()RZAh1v*6`EyZTsYgjlkx?WJ5}
z1N0FGa7~T_mTq=_!LE(v53R-Eo~R>o4KR1xY~-=>63?YtV2;9rfouUI*>9tc75?Y8
zHdIS7puaaBq3K@OV|=@~=+xWT7$jO1)6ux>S*5$*DvX50PRL&h61fdbGU+y(7~?4a
zlaiJxu60r_FpQQ80-Icopvy|~JPGK!%ywgJlS~7MmQ(!92)QLl?WSX)yQ^QpRkEAC
z<v`CI!nAE#%pzNy%C7YzJWM|pR}W{~Cj4cKHDN|}du<))GBX!ao>s3TFNSZw&5F|l
zk>u2)WY7#tW_A9Hsmf+M__`E@KPSheJorkR8%vQ+Mb2kD@ixylGSD-%jw;I0z@c4a
zGh0|eX{QjRaHx0XKqxStWeEs9cK5Lm1!Sy}svghn*n{Yj6hkk`X^D4ngWZ#ZmQ;UI
zaqqQz9XD!FJJ#mKY-%MxC?=h{06roG$J~(kTliVo%_M3M#+Nqw=&&H{ni8t=-ddVN
zqE}cV{-`;ZAngv_m0_Z_+LuxrOZeYn?Y^w(x6iYVJZEV{8muc?%T6R){mTG<SweyI
zVdRbycef%T+>FM0RW2Q3FRs2)>d;?yru(Z^Nu%wRzaH=dmBGSR=U{BZ*7y%-a2DHx
z0qcSz46GW~a;Z9G59Ko1j(j}C2tAyw3+js)$(jSnw?_if8>A7iM;&TSROWI|&9%ou
zIJk{`0VLjQwRH}k@soaQk0rJda`l`&?tjUx0~z9p*tjrNiXty83}IyBGNV431Cs+C
z3oi%sR6@J8rWL~Xbagh{K_-XRGd}kX<0e{9tb5a6SBphTFpWo|=uNUu@$)8@vN1*y
zcPl?KRT1sW>7`9vqGQ`qT{EfD!;%h{EmN+f*liVZYeR)^0EF@g+3sUa$HpWWp&sbZ
z#i1dEf6Bk~#M|x|XULsnHWW9uC#i(qFA_EsDzVoWPqwA{mRfid6+7CW8*Hy9moBNK
zvSVjAp5&O0#TFgP3$Uz>l;v#j)!Z^B&EouOhmS?NOze&X0)xTf@-g6WzGLcKJCe(V
zGua9#lo8#D@9VjW#WgIq(~jnWf#h-ydmc{hSgx$T6$D=F2HoS)P-y6J>j&&a4z{4X
zLm>z!17@<rZ?sc>yp3eaA_1qrIDdD&oyp55^PbSI)6Q0i?gnnD?i|t2#jLUt${Z#x
z`Fz3%Z&Wnkq(g{CP*xsy{dcQ<>XR3fl6~guf_)O7_<5o%4kUP|e38Ae*UAA%u4i9X
z_n^2HlzTcv<!oO$sHf5W>p{2_#Z^Cgl5g_R&}ou2fNSz?H3@7ed&3pOzDq#V)FOnF
zL%hzu_i^j-^!OoqxZ1NQ1?<Owli(m}a$Dj4q&xp9T1!c`D;|xX6N4cV_%yv0dYVE>
z^=cGl8}J<V))BcFK?qyzg0~8nGj;9)R1_D!6;1)poGDC|H4cBvnJg(gp5sa>-PJ{J
z#j}FdjT9AiJYyF}Gp$wJsl4efd8@ER6r7??mR;)e-X|Bg%ifB|nT*BQ9WH+>cs`8N
zQg7H5xq7kR9Fi!qzkF9#^T_BfP1XUs%I9HL89cA{!Ei-vbxp}Nxr~L1lfL$?;5Xq5
zh&r<oRp(v=)sZp0-=Q+GUFSVyn<^m@qjJ1*g-1sZdE+09igjah6cWBk1ed8|Y|OJ6
zrs$xJ^S=sVTWJ&C3IfjEgV2FM6Z1>xY&vaH!f1ZU1=VSjbCPmblqMMSTz0F)`@c~+
z|EFZD@CH%p^P3U`R7O=rshhJ6<U{o*#9L2`6L67PrjbmcG(B6xHAxXi_IR5S-*g$}
zy|Wc{=gjyP{&cKzlg-M$oGXfQ9zO!w94}yY7TTP|3+O1CvboV5-<`r$iFcm&R^e9v
zGqTEpSR*(mZxJzC@dho7j>@FSYl67dYj@??(}?k|v%9P1DJeh2ICy6lz2!W(lYNiR
z9h$oLp6F8&2hpe4^b+LD>+vH|U_@<BbVr+#eb?n$V3Qy?$VbmvgKuI!7r)W}dBeC#
zwlrF5C-N$jA*tDKsH&rPSZj4~_=cnLXwR8=ao}+~5=BVrMYcxNN3kE|hr-xMFDhV_
zz%5A9vh2_aY2?D#is&FyjrcG5Vo6@ozxVMXuSyajM_-ag`15!FTSdIM?GAe&b|C3g
zx;Zf~t9^sxCr+?6Z-oMuASUYVqOI<h+3>+Q$9Qdv-$r4i?V)NtRZ$mi`Jgk_!wwyo
zbtu&%)#-ShEMzBp)FC9}nXc8^yzIrDVx6z7gLm;UM?#>`XpiS6;9ljhqG$h!>S)M$
zuLvY~{wMR~DGueiwx03<yX$J~=^O;Ho)<vA5cW)bb7imG6AoEVe3KSO27Gb6>yyA>
z(`>QcJY>if9>p80fnO?$B&D$G8v-V&(7Z5%EGy2EZHYd9QOZMN+Y+(vn0LEvj|O@U
zr99aQ1?bO-3PbN;*cIP0w~&JTcUOH66|<LW-0aE0Td)W1%?(h93f0s8Xzz(aeSrt!
zT*!fj)N;sMASz&(i(Cn><{*Dkreix?JrLLyx%ztw&njkpAYg$DYmuGtVFdhz(a=Wh
z$Xg-iVS-mdz!E3>dR_#|E)_&M8jDQ(4C1C8%U2JR1iXmS{v3A{m`W!xbci%3V)~(;
zMp7%{Lj$?!vy(Y}2}#B0KtV63Vm1L;R}B|OJ00JUp^zFp5avt*LH=j(QGV~H&$M&d
zm@Q_0x${ZE6$q_sUzkGW-S{NdD6s>jXsMefaZgDiWCQy&d&*|ca@Ok&Mqxn7nOtbm
ztFTAJ#?Nz6SR0^zUnD5;2_@i#gq+c*k5^nYUy}M)ao#$l9rksu2Tonu)q@aw!AvT4
zh5UZwBMkJYa5;uraqr*ej?247=}vK26;f;|DiShU#s&C3=YxmeYCq%>aiZ#!9Q0%S
zEQdqfoT{N!5V$-m_H#Z`?JY*$3U_cSY74#R8*JF|FwFt17wm#)0noXF7>h8RU3k2(
zUHFE~w&BNv{UJX%>eGv&xpW#EP+U`($u9N*71_=UeMz>!_@$RYh<<4fqUWsBF7pmt
zntG&-%a8j@CJ9_RJ28;*DHI2{0Uh@h*_uOZu`7=|v*-l4Xd*!caV=NH#Nvc>kNOaf
zu8wa=t_8r`0qHfz3nSUYJy>biMg#P2#s8q6vl($bSe(|TF2a#f*p5729`Zlj`j2*9
z%s<3WA|8)X)le!H2So=RM*Cc7E8zjgI9Q7sdu%ig+hPe|8FxIddo$c1_Ao7Qf_IuB
zW5mShL@=zP7+_rY+9Y2ZciJY$&!wt6lu>Hs8+`%fKU&#pQ*u%S$IXQ&*l9P#H|4Cr
zSx{1$-JIXB{%_|*bP=Z=PYW+Xa2wCi^k_u3h+_4U@_9yf<c6w5<Q4>$nE^1!Q;A7a
z(jNEbn3ZdrU(;Z-qnXTSh*(^AX|%=O&B^n}=cC=`9uIyVsZ({Y@Y|W^ARu1X7F3Jc
zP&~>7f`#5+><p-QSK^PZD+ELAy8DLR9ew>KxFUbi;Nlh60Ab05EB7QyFGsmS-hf=>
z_}@`quQqo^B)HW%<XXE$bW{Yl;drq1q(xZ9m2Ny<c#<(=Y-s|ayu?L{2Nb>Ht$aj$
z)F3FW(M)TU*>TXzwmt5Wm|N-dsh!};0AQ(va~;(L9^)e)W51$ZzIzB)i)8mwPi$Gz
zf!ZauVh}ndMnr*MtKu7KE@d1*wBBO(J6!$*{cBs^X%7tPCg&Kcd)Vqc!or!glDOE`
zL^GLzsL%)E?6U_Gq)_KBd#E}I7usZx7Ofyv=in4jFzNiCw@0dP?w_-pX&+4th99K_
zXsfNwdoB%XeFLst>xQljH6(}gnD6J-)+G`$?QuV=(Z9DR{1oMkvqWCwPdXO1@r*nb
zvrDn5#}uC-dphyOvrRtZO=Pu{cBYKo6I+yOUZ!V+ui`bg*Vb3tkz1FT{cYBplW>_u
zZjiWbV~$E*!Y0^N5s<#5Dz_4%jIT*?1>KUcB~lyj@Cqv^Kn7iT0qM}OV*3!NE2EF1
zxQm!&{k|VGJtg_mu4?TvJ9&(;ILUps+acDDO)n?Kvbx0fCiIJ{h(av3KX!zy#pU0K
z5650Rkh=twLTE*_9m;XA*S=t{`d^SUj@}L@1nNizqOEd-12Kl|>dHKfR^E=}T;&Ss
zH6ycmeDepS@QrpfzUifxBNPEGcFezVcHLxJC&NB-9FP6`Bj!1Ah0N>iM2-pvqEs&m
zUOAcHipR3l+Z*guA_&+Ba-Mc{8vnaVb|$Gw*YbKhJLE)OWYhb&=YCe_62G&4#rc>e
zo>b(VefCLWc-smxu9g`6PotSMtJLg8gd#C#k|+!ms*wA6)#i}sWpM`hBHLhiSAv-@
zE7~K4CMm4IeLz{L#a7sauLhR`RZ`}^=k4q2n_L1thTxm{R=L2UvH!Me=YC6ako+#b
zdEF8Hce8z;sL@v2x!8Wl7QAYDQ;4InAH78*`Bt$q{uC{Wn+W@o3jXIj2m(+zS$n&%
z3ls=L`!nqAu%4CI{erjS!LIVD!XRJxcKQ}M7zln67yw(*m)Y>Q;~}lBuj^>T5#yq_
zok#P4obvkE#W9F;ITCSqmfI(3SYf!ARO_G;qg@&!6m>uQ+<{<rS$r#ktxlC*msj89
z7wXx2uJG+OU>xkQE2{=Xm~n)JWZ6|Q6`E5h%+)zDE_#RTnzxU52q4MVu1#Qwo|PAb
zmuiHM&VA>Pk#7g<7S)FE+k$SX&d_x}&T?f<8x?E7b<S(kij8_ZZ9~*HF1jrnlY_yC
zZw4xQ&-@;YjRA^X0MN_-a9n&-(p@|Yi_^`7{HD^6@jg!U?G2Fq6P4qnXi<WG8C_+4
zCI<jE{dF~@;Kn@6uD0%yi*L%?sVyX6q~30NyD*Noqd&hn21COGYvQ3>(;Pggwx~)m
zJsQ9_@{Wml+GsQ4n@EI;fytd_=G$q7ZdMqt&GJF+Ebi;G6H>%F7TZCm%?U`!7g6H9
zJl`Esb7OQ7h;1whqm|8z)-pv1u6l61^#b3^oeQVh!V36uJn+iKqWSynu4>3v6afVr
zx7F^hFi2Hv7FFBa02$a!q|_n1$G5PcO=`e<qk$*jmF8-*)h5&ln?GcA34vA&MH||z
zAy19+!_C%M^<Q6nDaAf5%}G&YJTFQB9Gml7G4@LHxzSqVo1RlqFG?<1d!BI-=@1(`
za{8uLPOjd%;<8HBqO@?^IX+p9Gd<TbU!D>RHlw$CJERL`zbP8{Rd0u?iQlQCo=L~<
zuhysXB3^O(fw)g?%W;KSoeM*OR+62rsm^Xw@i2CVTiG@q4EQOF;y!yQMwTK6N{#rs
z#~vQsMA1SN?*YuSnK<&1>d3b`cfj$Ag1}cL%Ha%-0NB*oVrv~$mSE&u=i4RqWzCY-
zA|O7Nr$vO>BldU&j^;>2^+aBw=C+j_=acy-dJM{OJ{4=130~1_k`SLxKr}RUmD*j;
zyj?izzaMKoNtKifg4pe%eOn)6l&bBFfwzP8<`_ySCFAqPY=phu%d2g+K3{sfS6JDS
zcu=Rv!aCcQV1-Ljh86rxZF?Sxq6p)rJO|jQjT}cJXXj0}tD4OV0AN|P-4)1n%UruH
zX7UBwlZ$O@z;l_!?v25uir8!WW3YUE#j77kFhc3!%MazBEj1nJs9ud8o?JzqM1<1i
zIpI(6q7CFJw<<D*I#nHZBs#fd=fc|Sj-Zs-xM89l&2<Fjk=y053LwtGS3;p3_eX#l
z8Fh9d_9InKznx4#<D`uy)=uTPSie-0VgnT7bbQN_JVfN%8ApR<b#^vp9pa@zKS`J8
z691HtzQlB$_mN~Ftx;J52?n1eEV!ag_Gxai2BXYpu~d@x!jya-%kbm|a=+STUu5$+
zq__RDTDwvJ=r!-Pul%}9oIK9HPVn$9is6>9<Gyj<1b9^I=i5BJPGzmL?{b2f+?Vh3
zA~S;6Ecf0-KhM8qkNY7OP;6ALXt5vTn*=#A8nR{nlt8~lK<}`hD-cT4?v@gN7|4fO
zluS|-D6m07*|7K~2uM)V(O+;PO-DG!m9m-_Mk^h&j8CYbZ1{;v;ncxva8WintmHpF
z;l=UI%_vd?_uC~W3Y#XjmW)m(3SWZSrNb`E@!(0|t1mxsWVC#^{wBf|TJ+!eKfB^Y
z;rXz|tH8^ZxiK!H(ztMfc2#byi|LEj=IRsiMq^1PJG)&Iy#$CIyg#_Y*tI8|9g?z@
zWCdWcG8#%sACYquzHc*}^2;_dNBje&`pEQlU9>3=C?m=hU{s7K0|?GXde6&o9I(*|
zgsE6TTWw6$x$vl5qZe)Li9x<qv6Mxh8ZwKLbdPhG4)pwfkxi%;s0?k=t-fLt<C|C&
z<*`8BVv`(nEQM{`WdDnpZ8;g-m>XJ54I+ZXr<^D>`^MkfP1QDBAt;TJ3!OJ7prxcJ
z`5LEHZONO#Jo1+-=5U+t<NalX&2ZE}or&N0%oD*%PAAP&sY+|JVvK@Z9D$rl{9&0L
zdy><U*dSSo=OieVET+@uR-lk0bCHm3UiD4>Z_h!Kdu@R)XTc0x7(gM0n%q5iMO!ED
z!jG5j?jikkj+l{bQBF(XDg^SLXd;%-8N>B<Z+7DrD-Kr1pxXE*<A9i4S-I*G1YR;q
zQBn=rwOAiK!$M;=Lv7IIL90sxM?^d1qAVOK_g()*Ol!_AN<=|v1lZoz7}<qtIlpbN
z_WXuEO{wH$Z0PVqQaB0q^5R6o7zkZSTko<29@lS+t#F*|mN)LI1kikBavaA8(Tg~+
z7JDG(bYE~#7E-gi+RmWPQJzm8C%4(wB+Add-5!j#+J_`nIe;rVUxu(l5dMd9K5Aph
zU+iH&@9BTEM|>2@qPN<kIpUpNw$}eW4YtSDRcI{3YjG8bYxP*Hdl1Zr?D0fch$}uc
z#TCr6Y&;RIQN0xQ`yeZO(&r(?RGmHL|AuYow5PKzpwwPeeE6B_+pcmlB)hPlymU)S
z=~N_+^)Y|p6FB<8Nbuh3c=%=;<1b2*^9=SSgqF0C+l;)HF^aMP6N`lSATLsdsciFA
zc6N2w+4gAV8qG+MU5pC--fEy{#7NtfYg|r+^iq9n_lYW<33N5DHbx-UyDu0zjDN9T
zsb%EYqbRz)KCURY?RTua6&vtMI}nZGz<5phFJ26X;u~lI*YEnNS93AYsJd>worh0^
z%wO_I!><h_vXbK-ACn^|ikNsSafq+yfIOpeAeHX@Xs(hsLT(JZQ2!lEqBfJq;J72(
ztoe2#KTqSFP6q6B`K39K>#-ZFPvwH->MU#XY410OYU`bLrkW4$cs9BdAL6t5O)6aQ
zUkTvP<%u>oHAq!*K6kZTJfeM)a3GSWrnriHnw!%z#~=rIS3ZkxkO7nu<$<)%t8e(j
zaUCeNVPE8Eb8DLAlST^698EekK=nLg%vaU^Ik2w-|40=9{(X~!mXH(C+k^J4?|_ud
zC_)LF|1P^Xko)xeL_M^Fl4bQ)ipM!<KP2kiG24DjP_|R?QK0-OD1}I3de-{w^;qPe
z6G|;@ig2-a!t<!Mo4kj^-U-H`xaT}?7rc{h9wl-dS%Is{fjnEx886y}-W#OYjnBjI
zcY?!0;exvY{?8Y^<JUmedi4|~)gJmeR#{O+4iDMIIjj8eImk=ian{7IjR&!AD*Dw+
z-$^r2WZSyzox-kZxrcZC+Ah!COjix2D`Ffe!d1zAJtyoe&Xw;Rk>&_7qRFm$$N5K*
zQ)S&*4{5&I$C=$!5({4Qi}eVYg;m1${o1Op;tkN_A{~v0Z+>FDgXld+<_=4RUD&4>
zaVR-wTxK2>5O>9y`#-3_h0b2TkM<42#Vp4fV`HMZG>FPtD#cmxwqYYOc8q;z&|cL=
z<QkDu<9vk9?)D<JPskRn5Cd1qPK+JM8&B?$ihZ8s5UH2<alK8hTI`f+6EBE8c4MA?
z9T}7vh!TAMWoeD7JStS%Zi?k<%u-~fK(drW(Y3If{ao<vB4c=kTk`&xo+mFg$TFO)
zXC#=pcoZOmHtn&Q?-VBPmN_=d2kLHy{LQX_EhI$yR{Xe?OJPnPENCd*lyj`w?%Wt%
z1VMo$FGC0C<&0cKam@HC+XA0wf$XCT^XSm;>0R{DI}g=TahSW_DKf%9&*X9A?CzM_
z56j&yT)bEm(<tZgW@pUpo`fw*1f}96Dlc21Dw6$W3o=u!e^XYXa<+nEoiA`^S690=
zI2JaMc~C-?HO3fH%1Lq}bl;ZxOp%sA{sYZE!hBV}X?>@>6w_<S5AM4`i_eYi@HUNd
zMeE9KHltNsmKTz016{TvCR8RZ*~$dzRn@n>NTkZkvDNO6Z|+xC{%a4!H(fskOh~V^
zI!Ev9CLhU~#ANx+myA*NU_d$4(6CD48$knmD5t<7M$;<m!^3faB9t4TO;tKUewWJs
zk-<|arOE2-(Rb2jzG87R+24!T<JugWVH5+ih4*V+wR+qNS-k|3$11#(x4UfL;zp$v
zcsxhQ!v$BLh;J_X0tNQ}WHqWvGz%ur+Eck@4pn4|2>Fm(pRQW(D;^38)_tbBki~Hp
z%$ARk-dIP*o*b>P-2Mxz$DXp@@7jVptk;oZ{BOqFMxR3A7?qz-UtWv25LGNHP?hic
zmc*B-knC-Fq6>HtZF`Ua?QFgEdqe0CUbMRsc%j^ps8bBmwmW_)_{<zTd2`&FwkH7K
z#V+;s?6ker21UVzT<Zl$UfKOgag?-!=h+VA+U6<6&7o>A`CDwle|4k1>MfR|2ZpOY
z?CY&>rk0=$M01%!6!Wb*?nfMiciv{N=dT0Lb6i@*^L8{=7R+NITjjm5W7V3gdX0fN
z$Ne5+f@`xA`R2@!dot`;;wrw^VqxleZS9nI80fjP(@w`CHi7JV2&Ftf6H_@b{gn}T
zi1f?WJWgmvFq-bLbJaFVDkLO>7wmjB!d~#(!;Z_7Zl6SJZ8rot`?Nxq4TV<kFRYbf
zV?3DTvz!5MiWeb{1SQ$$zK7+Q3hj%)7!DWl^UHURTv8(D5W#T6R6;M#@v9i41Tz6R
zxk7)PSDRAyIGRz+?3?PLZO8l4zKu>P93AAize@-g-EO;mpMUfGDYN$v-bdb8xWUzx
zh~FP$9HBI>2Jm*IKp6a19-F-=<$iYL;ow*?g?)yxkGiEc?4(-&RV{O;U2xJ_7bQrR
zg|Q!b>B5uF+Te6*!%v3&XPI4e(s`=lFAs((=m@~WQCxhouuwIep%iqNc%OD&>q}3@
zgN5^fT^4iEiG|CCb;vG1nSuq3o0!ild@OM%`t!;h;JV*mXII4lGOS9}+KM^r>gwAD
z31NjETyrv&SLDcI^Vgn?L!3)8uNtCj$OKqpBOD|vYHXy>sWMx7uC9wAB^{Dm1bKR=
zjdG|B^vqDc(a`}40l$|cP!dz8c4MM%AErl{+ZB&CHpfGSF8WMXR~r|nAPdcYn~=M_
z4Fkc%7~F4dF`JY{Y_~~+1C<mnSm|wYe2Z9yLi}&^187^^Y*S8#Z`zP-?Mht3Blg$T
z!2hNk(gm90wRv-VlbngI3C7?RoA!%uiajQeR2~)9V@!oJ66zFgQ^3VcU&W~F|HWqc
zZb{OsfP>jFmdnDfVxHrTCrtc}&2{v<wHsK>JDIj7m3)vF=mH;VNi%g2nr-3WPztJI
z`Rnbj`1TwKF6R}>iP+sI2boyKu7O#La_G+HdaAha<(}LRrRRt~xz}g96IBDAw6%^c
zq+71@@Ay=AvCD=)cA|V61J)&@7Zb~XCaUL7Fr<!308MR%bix+JLS=YF1y<ZITYYd`
zqp1#N?fw^z8Zgb);eUbT&$G?0YMcF>5~2dMc3FHAJyf)gLM>PP0-c;_6hAOql^5Dy
zc=XeHh2-vdf1JjI3HCrv!(g)E$VYT_bj*uf3hxwBz2;<cGuR4AFIY4g4_2#EN(HSH
z3tsN5<kx&SAy>pKJyHDy?+%uS?0`KIJI5sua`R|4o}OBHMQpUSd4p=Ym)N=(F5kV1
zfew20*pLCFQ``aP`*`k?zxrt>=I{weh<4~gN7D*>GV$XMB|o&M^3S?@g468j96+{N
z;3!YcXB?FWddT?Qlc$gFxE+LApU~HxV?rmVa>?$^UkWt5_?;_0H&%e(F9%|lDqj}G
zllE18;QoA@{H`JhTVh@AsOhZhwxkO5MVk+g_$=pArINAjG1^9$4Cx2;U&oC87!7RY
z&6XN<gYEKNDg&|Y&Xa_?NT`tio+yaoE~SOplTZSYc9h3!kL``qeXcks?XPxH<xq~f
z5ByhFT|8&YI>;WCI53Ym(N|+oo;g^}!}-%KbN^@q`Nw=*N{>{iwE@h2-46-8WjD~5
zqj_j5go!+QtXiBa;*=F%%*-A;9vjou+W|0DDfxs022P3iz3Ld`@>p(;or=*m<G_I@
z0sr+<TGVjKdqn0dVz^Lgkb>h(!k$XA{|e_KH|(=MMU#phob!F*Qn0zh&gVf03DBD;
zuB7m*{QE_S_a{!zsActO?n(M~4qKK2$>Y!4XVruoifU69k<X)bVc}^l4Z0M18|{l|
zEwofNZ>1~z(xEF;WXEFrs_HAaNnB9?rCzbGqp#=-k;Dah-$ZMKH#iLy8j-ucq^NIw
z)9sBbQFv8jDs7xtu{5%MpAg`&B8?uhFj+cE*sYky4faE#f#OhpjA2|aF6qimkC+i1
zP}3IL&v_^wgW^%8_HN;DsF%&-H0)jHrt+?l0mUwOH`HUZ>)|afd^aAV4jh3f56ZJM
zJUU_ImLD<>4qNS_cY}jzYK68?|K;6y#zUM_K0+65OURU662A{!{&U#x-Ez6ZF7-JE
zdTQ$%31`AXquDNtd3EAS;UXi=<vy<r=KR0?c15)zrSotibY*;d4mLt|GO69Kif_dw
zOROD&;nluz?!rNs!fO%?_&~JVweKETP}u7KmeKP6`*#~rt(ls=;0=lb8~JW(5lf-P
zqNe$enz8HNjn@lb3)IS<baYfMd<jIsM!y?OF#dB2uu#?k*ZGg}b|irk<32V4!uyY=
z2juIsaqos<xTOZy8JpluI1>^T)OW$oG~2}Jrh6cgy9aC6q-<A;5}}cYqn-Tjkr9I5
z|NZ}F2DTexM4nFY---B?_=d(Bt)K}1b9U3a!HCZ(Ofc?V(MWPB368J<?zL&T6#0xt
zK>YNC586bSK)((HJ-6IB*Jf7p$UNoT*{pZNSLpV_9Y_zp1YvykVAF!M(oM|C74R6l
z9eA6~&HIDEgkPsPuz3mA=|sN5Am$<B7~Ir}4jB=^7RFG*MS0<h1-Yx5Ql1L_Fy5Wt
zij$e+`=Z>^k|ub>+%1JMCfmIUG?FQoyIdQ1rdL3$iSov{$^-tKnO#1qi`!aiR7wpf
z7Lor-s>dDH8md)*0gSQ}xAaC|1<}l0-KFmq39TN+#G2nN5?TvR=&jk02U~m%Ne!(%
zp;HnpSw}R{BOvcvm}Kk94K8TA&zAXcH;lW%R^)JnLvcQRRl>hy>c;_WvE83TGMEfL
zXcO&$*rA*NBN1YGC!goP(vz)@Yb3N9Or}jD4_lLy5||}Van);kFu}aIjpKx5J`~-{
z&%O&#TmG{TSBsHCQqS;SdnCTeMuyp-aJY~9o^gFf!?`ww)7qrAQkg^N0M|uVKTl_2
z9&?BmUop<&@roiSQL}t3ceCt%fbc{vdnVx~6eO}It7Q||&H3!*k#QgEr}79?GW}^k
zg5Qt0*`BGE4go1Icc1mVTSWP(%WS<PBMD|z`KH&A0>w<DZTtlo{7m}Z4OvRQ2Tjr_
zZK+-rXSn-#XET=vr7dimk2rO~t<!CLZU)V7$+q2}Pfkg~rzu>27#Ue?S1duSgd!po
ztF^m=l{DMB81nsA?vp)vB}(+HfKQbUu>pP@w%3<3muJ)V=SJo)ZtJ2dd7T}|=Ta^W
zXe|y6IX3Kn#qjfLuBWg<MXPCtb5mY^9J6NJvN{UV+mUKbGMuG-S9s-*?(2R|^>sCN
zGzKR%CR~V~g<W<muS!#EV^b~B$~f6aGf!zK2IvJlkx0SLl?!(;u=-@Kn2fcapkard
zs=n!p3%`5qbbf=;QFuTvjs<5DM$kyS!gw;yIt1JCQnYi~W8h$Me^F4t`TPcaWFN(I
z(mtsW<awDp%Y!eYrwR@#wD8mD<Ih#-*K%fbnD-TD#?KvLJ66~iNsI`g1{_}|)-Q!&
z|0?@U#qYSK7E1ng0<ctkE*i*ha*=YwV3GZ(Cce!@9Hw#+NTVF+-zB1zw2#S?D{|rc
zXsvJ9MigLVKSXP({G{+MY9Dcp_G6B%3td3F+kVP56q|>*r>*vLJ`3}e2%N&%{|m1(
z97KIbgS_;IoeIWcCUHB>kOJk6x!{zu5@H{eMqY)(1=fRh;i*F2|48_r4L?<w29m;H
zi6VT_hlZh!oWvJ<Ta5Hwc8UL+@?|x4>8ao#DGG(E!Y=dQW;ZpE0^*djsZOkTUvT5O
z;#8q3Q>NIJ*|Vd*SYWU66%6z&A)|bkT^-8?fJMLI>TbEmT@$|)3k1MYsu9PDYfq&X
zYD0Mz34Db{N<UKUjW#0TQ!jT=1(|XDBTu=aQfi`S?YdZVv4%oUdbwev;+wN!;%4^S
z=&GF}*L6DkY)pQWc8rUxP=c{l-)9TWmt*?4Q=zR?{K|yvRBn~fg(O-I3Jg&cAg{Mc
zvEfp=h_jOb?dF%sr-IhjHk0RGPu7hwT;k#m9*8M^Ciwd-x0`&ob@eLqVK<)&E{SvZ
z%ycLwF=!6L9^@@gRCz$2wdr|i+;-u~j6AgY)b6RbnYo#gZVKZJQq-(GE+n;`tbVY9
z=<EtMie~ooM02VU^lG_p*4+38ly#uoo@Z@dE`(@Xgl1b1ZB(S*?GoE;VUCaDX|JAd
zc%HkmoqYMY#Y)n2x5H$hr?IQ0*%swrfWOo3$-mmQzkB`UaJ}U@uZ^MH|K8vGI&V-*
z&X_K1$OgDo6rVfrp)sK0$|n(cg#h7UU`xF<8flo1&W7^%HOC0Zy5))53hU9DE28*f
zi?t^%O`Sd4I{Y+9%8XiJiFM_52+4LI#5E67?q$n{9CJw`A+rY%!Cx%qEmS9r7pron
zX>f!0`yc24`iNZ|vdZ0HrufL!)o#Q+lgyX(tVv`+B84nJm`$9qDB)obWfLAVQXoHE
zjV%8bUFB`|NPH_k41T@s(NiJrPHnDV!As5Cy!KS#S0&Z1i+z?O0t5FH&BHcM{;{BH
z%BuKyQZ;g3JdxN%&RdKz_M|VXJQ%F}sRYHX;Kb9hp7ec0#b(b`>lS9q=&{dw@~+mm
zaz2U-USG8;eg#Eg-<z%QiGl^fGi770g=%Z^2krCoXCM1`8n+~&md&B5z1_CO&p2~%
z`L<V$x1*!$uaHk!!Q_sbF>$;821L6{^qR!N-44fkTsZdlKgUn{o$ZaK7xN&8AaAt&
zF-8$*7=6$U=zw>W>f%s>Zt8-$_G+GX5wh%X?nPA_gfIg>%RrAbA)?n<yN<-n!2ye#
zvDb5)TS^_2qlpb%ge7fMqHnch(M$Xz*0}N<A3qi58o9{g3JK|h&puJLd>KC5GrFIw
z;5MYRSmEza<;>!DFihr<=qT$|5P~J6P|G2!s+fItwgQ&BRF=%2Wyo_rWOwW0Mlx5}
z`8YuWI(hp(sdg`Yzg+7mPZQUimk=(i)cY%(BmR+a!~5+sA3^D6h`Igz7bh*9tZYy6
zrv4&+k&amJX36$focF$rS>+*s#$3Ly_Eo;6gAN#-_H~{hAB!UMo9L<M2t@;BvG>@w
z2@Wt+Of1*YzVpl1-AeY*?|->M*q;|fRPZeL!SQHG{R;arX&XW#TDQ;-Mygxh=br*s
zCyuwDtJTX0E&tCK%-(ZOK~^#;ul7rjEk3=-pC}oJz2{5;{3#H=;5}z#N<a#&C0bti
zp0iaB6UTO0=(gMN>K6n#PU01kT9vR^yXd{p%n+c8YX0JA=yXYXi}2;JOJYLNU&;3}
zx?CEqU1$+N#X}y1rM@g$BZn7-t%CV4kJe({l}AY;4t;2S#V=-+j&gTU_$%KlyxD@~
zc2#aTOo{S_zPg%U)ky(Dk;0`Xz2?2(XBN~n*|h;VS^x0R&m9+6D;pVL7bVAco*=dU
zvNpZ+y}WJL#T=eSy(I5tCOO}XdM_ya?bJ0v)<CQr9g7qwT$TKSjmd@c2+2QK*__7u
ziUxX=;?BlJ3wO7W%yz&gWDCq;?KW``H+Y<jQ>8$M9>W!mlimxas}Z+E^f8NVa_$)w
z{Z_j%H-U_<01yKAU~XjZn&JR%USc;T$jDb?H@{c-aLOr9vuUwBA=C?kGKP=o?-d?<
zC25@r&4B5u+N8>CgUzhAwGES;PG_&p%5TN*Q&(t%&3>;ui}R6$=Hy}Mj<Yu@<i%Xd
z-3>ojo9ny6&!8la$iZNfbzEddvjus&DlAW3odjb8&c$|Dg)bg*iW?jFD&G0-TtB`{
zN~Wsh&5QDUL;DC9dXF!4<n?3i-UQdQwiQ;Jv;K28fyY*tv!W{djWxtkY$8F+W^44n
zrk40~Tk3y7?umc1X8(&~Omc&7eJ>sZtSA&Y#SujhP`j^`1T5G9c!3Bhv5uTxVhZ`V
z{P5&vWy@k?!WJoBlyYf>?_&I<iMA>r;`hDqwgvAoghofuk=t0z0`@?nQIRb@d4;3V
z>S$fE>vDIICvA;S(~U3ogT5ksC|m5I_go?EQ?jn2h=nm{b)-2g1Z0gpoZxM&Z?{Ks
z^uH>JhXkj7do+eAQOIam6^<Yz0d08trF&%Mw$_Kajp$@N^>a($l@4Bo)E-L!6msMW
zFsU7T?eX`HEd2jTQ9saQPXwlC9F)A+oCLXC_@qPZwI?wpJ>~0GPFoc)TqYN;r(-X1
zWC-rCc<+U*Ef>ZN<L+^s8t9pZCEeC1SSS4EH`ePv@&3T{+s1@MsS)8M-9@^-n2VG_
z#cyZ}Lh}}1-=beF`ju_-gPw51c-!vdcNb>AKjDp@ON2u?hj-;(h`R1*#buqa``t;s
ziVjO0)t>ChE2u<aWD&K!@lAq<UN1cW`vZy|JrDfAA0dHaBIKxeAoir)$g7k>3H-Oz
z+i?Om&;y(nV?lg10W*I5Z|!h3m7eE5MdR5(HkjJ6#E!(g!d~S%f~IqA4k`s3>KQ8R
zRO{bWC=*<*9nDL+w41n3I~GkuYlO23ggowmy#CMsVkaEFe;G4!q@9eWind_qn#WG%
zSu2Z2W~(~w4F`I7G!!hh1i|1;Aku%4+nJ9~Vq`lPV=ASFJIaz#K6d5GAW;kjZ^;}3
z>F3IQ;zP;(%|1=E!hhsG`>YyG*$!k1J7}NBW=8+ozNmgi^UILWK?lD~aF8<taC}u^
zUkR8!%)KZSeW!h$fWN)9!M+KU6%wHTdT-Ej(=Ef^chy9`4jB}`e=igYgcO#1xUIxo
z<6q_@kt9RzhCfyvmBLMl7hkcT{;OBCNPUF`{W+E_CXIk|*^*c?!PjP|gE^BzQrC+Y
zX4vUi>vEDviI87#x;%SxYL?lBr=7J^a?h<c{IoNIj$EQ_2S72|p=DLsVi(2e(pNtV
zG;tTSi%%Cm=|BC&F3Gu}lIUc#HoNq6Xv8QhB!S|xXp<RF$e(4G`x<!hRdV+V@3FYP
zlnCpzD^CZnbBC(kmx?Y|#iBPtJvQ3aF{BK4nHuhzY8uBOJ*wAM%ahGOnMB~oVk8wv
zdfrAx3*9QXY^iAMx@e&+Xd7+R>ELsT<PxJe7<N7Sbg+HP;Yr3sGk<855ku(%aL=@{
z6+9YYdhzPd0fCvA?!8#?MXvg6f-eTri|nh~uY~4IthRu2GSffLh?5-t|8Sj6_9dVy
zAQ|nARa0eaa+)<I0o_^Ow90NeT^Ox-w@#gIH~Szhci6OC;Qa0u6gxKE|CUA-J<feL
zBasR397;|$+sqsk`^I9MRe@09S^k{Ou9mWsZ^RygHFL5THayaEMT_t{02g1edA_8^
zc3a^8n%Zs)6UgI#JI?N^hE*h*1MTiWV3gEQ#Q!J=E%H5sYPZ@w)l&DAtONuT+r54>
z<|1m=#?K}D6gznZ_N6YqaX<d5w0C4Zw1(=emG-QB^NFpoYE8a7_N|IjOMM4zB!(dw
ze6`xFlccy$rKok*ruB44NVO#;WPAPz;Uzk_BYR6mWB=~UROD31+z#ss%%!+2n7%9)
zBR@9T#-)~Dkyi^Pn+j_|TvsK8=eKv*{n5vrOW+PLZ9b3(x4Z_E*XqQUnK(E=S=Pjy
zGC}UJ2Xju8Gh&hC-t~~TAb}MZ+#dGxAe5KvrU^(X03lt37|}->;79Wp$qnH04!w48
zO)uEmn6LNnD{uzuV!l#wlfL5HAFKB0sy8_E#}kg7XzZ{UMlXq0F8$)6pS$dAMFx7x
zSJ@`<<Z1sm9B#8clXroHyGSD)))PnK0<a@tTB60+t*@q%zf$o@wq8d$j=(m?&tf1q
z7Q*|A^~E>oCX`S@$w#-uuF>4U9&F3C&XCc4d;F%vJG-slug)JA+O9m<+gn?0cP_k6
z0k>S@J-(}2au?WMpOL4$rd^~&>A?2;IE%*Gft>A6_@COLYQ+S5v*MR{yk3pd@_d6s
zRosVj4dch(U;{Z&a}8zJkDLw$GCUZS8_8a;_WlZQyWjAm`3-uEbxgO#j>Vq20~nT$
zS8I~aO@iJQJCS##xEMHD9_q?-G66!sQ@sL^PbFeEc$M(ezI1{Z;2O^)0(I5bAYz=&
z4gXcO!$NTalh|+Ps-^VHA)du<xAT79H{5W8eG)%AiBYDV-S(;P1xIXUf&I*fE^Jlt
zeI;*LaTT%8qpzxLX10|t;+w=H@q?;o`(-T76WR_jn%Gx)8XY8D`8t<^%}pGIqEx?0
z7~q#9%k{T8Jica1+$Vpd?*jCS#5rW&I~q;-)vteTKb#K5fHEkc`lF0sg6$SL_1jPW
zCm=1X_|Mho@)^_1@sip5scOQ)SZ3U3!{Qs)6|z^vCcD7D0VZX|L-^!nz3~0=yv?|M
z_N_L&>QihF#h>k>_k*}j?-ryle!nn6c>PMneaZX&&X>>>z~M>Ql+fT(AGU=8dv=-s
zd)zquHZFfZ7@a!^VMLvEh5tr6J^bl?>;|&4!xmqao08F#Q<A^v)$bSXp^0a<%dV+b
zTV5H(2-~%}!@=STHo`kmfu2!E#)@6vEJ;qzC_C}QxXznFphzn@%9p}V`yq~_-w)2P
zlPED+amK_b$S{Qik}Hg>oW}Yn%01a(<8reIb~1cT$jQsI>r2pe?YD{5X7@{tTC9MR
z;+s>(Vkxl6KKwvWD=*uP?-!PGRhLbvy34Lvd}i&YY&v)5T)Vj%YkPUt`r%Tiy&vjK
z9wFBsyvU2JcstL}Fx`j8UOS^2UU_yT&{6{D{lXkCCF{zp_mBLq$l-d6;P!w8TPQcE
z#%3oRVG?)PoW!O1O$4mi-0DY8!R9$84D{UdFSZ~72h^msTbK)wnJ6HASM^P2s8sW7
z>~0@%pl7Kr;-Wk@f>R1Ze@}EN3^7o~<H_#LZ?MIvsEQ)DHmC0*%>tr)owph2K{Jm~
zOl3TL>m54to2)SbT|@Zk(rP$mYj*ym`TbzcG2l?Og8)Tqjb^SkFM1kl&&AIr*3LSj
zMRC1HDNh<T>k9lSp)4Hp7DQ&BVp&e1te=EGa<Nw=usfTfE4C`Zi?jayF}6fu6#074
z?SULYbf^qrtFswg{4S=GJio>_3oV3;eXv@iglMJAJ!lWb8p|`pQF%^HiTV#a#x%8b
z+9T1?9nU=b)}w!Or&Q3w&97^7+Z@qLpy;3z@H#()skho=KKKpeeq)cnA6hxg+hoFt
zhv11s254+ClRg=12;D%jW;_+&v~1<-R4MVNt8Xut<eONz?3wr`Rr(qrLq;&`@fA#%
zJlWP)Ka1to*}ZJNgR3H5g5(QD&f8eERDv2ez4SRAEUvGUv*s56>orN9TWni^Lf$1}
z+~x37^lP>~e?if--SIeZqrU<yHNxqm)Z&gcyQ<lde*v~K0qk}>X_M3To}dr{2M&3%
zNYY;-o!0`}>&P&@zLvm-{Rv>Spe13!0l!7lBUh*$%6oDfJmIU}x)ZLd83u~df!Q7B
z5qQK-8_4VM+8WUFh%c+2YOVHqqH^0ZJDP(ooI2Z%<-KjKrxr=Q9k13emuYOj=!(Ro
zzXV<?MD;{={^R^<b~4D_3_F!ep?m^)JeS$&3Qh;Xa}*z3gfrf^y{R7E$k}Kh&qY$%
zpaj@w=W>)Tyj*mW=kr6!Ow2?7N%X2x{==hvnp;5IEU&E3vXjz-G{a<n?whA#$x8de
zL8R=NNYh{D4X#(sj1IlLU&WsOWM5b4;ZnV8;ADvT=KUiRiWW8M?EMe_fA*~}1xx&W
z_FY0?2zq@Va}M#96G7ma{=x5xG{-+y8)h61_osZ7L~b~<zb&Etb9JWf;!~wZiD$vi
zg#Xat{gNgTk6yH4XHvT&)TDlLwDFhKZ3!2iaaKmORd3v&ZFP9|E;;gWbBf;1?m&gn
zLR?&ZJ5*8>$W!)`7!m4TeCMUN@_Towk2qJQo#b|K*_k8C*U(Pzja_~w7%X0zPP^hv
zVR%R?cVClTc_vuJ8B&uiLms*6j6dO?jq^VO>uMj2>-qv_oNIE6As@=At8CP+tpMV(
z051GC;+L&qV`bw}O8Surkk0OoWt2_3E?SCA$YrJERv~l(16<C}jGJwA1rXJxf=hu<
zC=SLpCgztxFAA^O*!boKO7Up@HZI$hz!BtEv{4hzR0$qiI~OZ^dU5QV7)?dBRR8TI
z1j0$V@D6!0b=hQZ12d~K?u~`NzA^fR<_-ROy-mq);>)0bGUeXnXVxu`_nYH)0aKyw
zl*hDbesVZ@bJM4LV*<y~3{}ul?ggJlGWa%|>8-j{J#JRC8X^`0F-6bMjx*m_ym<w&
z%)sNCgE<wHUnzz%)s=IZaIP;?xp$gtC?2#D5;-qgyN?6^E&6v^_7|K9t_8)KJM`JY
zGes6C1;|K~aaU~Hbt?XN-8}>pJVQK2{kEu@Lr+ifj?n2d$33}LPFH>tgZ`!W<~4u;
zBfv3$-O>UxmcNj`q~00=8j|=GAt?7~;8Z+#ckrAw#_#T2Ul?0K1AkywD=dxm%GFt6
zPI9_w&KYiRTV7+W(L%}!1>x@@hgf@_=iEP7N8ZYkh(|A1*O|hX*DkST4lq)pcgU-9
zMFlXq5W8J-RW6`N#i-(-SNHqyw<Dz417||^z&nI=N0ekvBvyhkn6NduzVa|D+-a*l
zc&0GtB=NV0hRi8>aX<3n!|_e#id|gTogBp@XM$@go}p+`i@E$!pS{FG6mQYBabks#
z0Z9)fkP3TUt^*^gm+g4b9`kifm@wHM_xqsSm6Pm=>eQ4kSEEP@Bh--v^2MI4@Is7<
zOBBN4etRlr@fsE-<D@;EGsA}|=1^?tGkyk(yRGNUkxR?_)=^)W<Bo3M+Il}5n0om6
z4(qK3A=L`U+>6e5V}8S+Bcr|?1N!`I;BfF!&nrS{V<oFaEx#>s2LiLww&&qZ4X@1p
z>`@ZC7lN}Z_F8ydpitrwi{*cJQa0322<*kQh*+~HR#oCLL_knq?2T^{NW<J^`zu0G
z<%p>GLRFMH;D_^<k-xS>iNh!`<X21mLI3gg>GSPyZekwG=%|A;^WSO~x7m@H!;i>~
ze2{8{V%a`++3VhP{ABW#R?9w6v@Sa!lE)lcG$v-$j^}=*g6xo>Q~-1$LCWH~>}0N{
zWhLxA`p{GPGXW=!b~=9sTf3X#k<K{Y)ncDtX=nZLnzV~^dH%QD`Nw$+?0hV56LvoN
z;mRrF6L0?7^W2$F^L`A8dW1dm$x-m*iD;kakWH{(b@qir6|7SMI?1tL#(@7Pcl4_`
zcd^n~(s&J5>z!t*uYDNo1f|nA2{^D&DrqIk{<gXx@++4BlB)ghvUM?eM&kDPhN+Md
zJOJc}_?B7-o^Jawz|#fi@snSmhIacoL5dcSb7?jEz*#Km5v5v^m&&jY3dLDS(v1r~
zC>%_ED<H483qJ^^g=Yt&Cu*MI@#T?$|7)P61+|O32m4SG`HOw1+h)wQOFjtV=ZTpp
z!0f;M`Okm0%RV?VNm_^U>Ir^0sus`xa(I$g<xoAu9f+37Y=m!(ayxvGHUqRx@L3-?
z_veF^RN%3(D}BcClW(xA{GZ~|t_~1*ngINAc@(bsAoVGKlTe<C^}Js@1QIpHlX^sq
zA``BnkEFyO8Q<U~#5JOi!N1(U!KV~+K*?@2D%SP88*KCk@lq3MgH$Q8VN5l$n>!%W
zHa6$ySkQ=Ih8Xuj7~{HNh9(3E{)StZa_WP^KEJk(7jhk}&!mLie4+)&mKAe}bCy9u
zv9&kGiAu-PXH#P6!VYi%uT&x1H$`86l8Y(ZZvLQf!&0-@v_yagQd{V4ot{gU6>Y``
zp$!MO2#bXNvh|tXUF9QeR&KeiSV`fSXJ;4pK-9IZvpEjSTjwpbxv^Bw?<(#rvuDin
zEs^X*x$G8HC`+h@E%d*LEL3LuyDET`7$LxdyLfz)u#xtfMGn`2o<B8p;W=dYRAVX#
z`rhbKlz||J1lHP|2ep5r)kO<;rvZJi8#UzmOKk`e;mbX++|WjEEkmiodyA9lQXlV*
z62M{2(bqL1!Fl4Uv&mX>-r4dcXwP-erw$!XQ61GGfqn8_?MmF7Xv_Sc6Mi?&R`@@O
zFR@j5E;v#(6Pj6P_xq1i=l{VT$RlnfZ^i0_0#^j{dT|9?<85nT1|R$&xE6>kD!y*G
zf5`kpjsnwG-d<-9XN!hr9{%oP6+i*y3m{`8io!yt9`!y=Qmn2G3?z||t;?;G@`dt%
zdDU1rPgE1BMIQItKu$)8W2ZfltDlXf>PcVei0iMnr}EPLp&k>UJstaU4obYuc6%ni
z3I04mC3vUDmx2XuhpqR2cPrk$H`+=Oq-eGEyeJ#1^P?Cl-^8zK<UZ?*)>2(4QK5oN
zwnXdFq7WaYhLT^%w!!mMGLTN&9+b@^)FJyNMVXj3N3H%G9%OOfydAbH;Xr_t?T)$9
z-Cbc6vflOCo)3;(C+_TjC}Xtkt#++!n^J)8vHgili?L`FuciahN(&WPRo34_*;5z<
z1n@JmSMx}2!9V9dJM7@aF{i->^3?GUoMcCGFT(cMV<g1}7Vig0>(Lykv%dbmPCHh;
z1SQcwnhY_qZmw1Ch2yyx4iEhu9w1xoL^Z6cSj(mIMLQYa3M!P1ykw`U%h%Y}-CSp<
zebq7!*%?PSu-eWJb}jqeg<NwkSBZ)jf`2~$#32<0+$WCOXcoG<#dkm~KlSmJ(fG3p
zw_Zj<7jC5z@pA`Nb3OX14*R0QD(M0+Pw|Ke!}{&Z#3~}Un1qZA6;cd(1h|O%b+!1H
zAYX-_v~Ti6=1zrM`!)wtxPzXu?_w}~2H7tz&9Xrhy7zt6ln3HImiR;No9a>o#Qj(S
zUP&ChWX4ZE84(6e^~>$&>NxczYc~Y{f<a2zAPt(GE$?F+f^t_qcYD~`;4P>^-BNEC
z_@DfWBUOuCSpA8itKNp6O=jwV&<XByvt4wyFjI?3K6`Qe3<}7~fd-sx>{S=;HZaED
z%r5nBL_M@ES1Pc2X1XleDUX<3rsc_V`Pon};hEZKS40ctH&crj@ea81tg|s(c3>#r
z?<+U0B%a~MUUfEjGAfDJqh++K&zARE>8I?P9MG{HCyJeRZG2M{p<J5}VrU+5HaL&l
zm)vF}eE?|OcDwFuu!|V$yq?6UvqhSoH_t}rexPECLFornV{$^V9mww(aBOU@a9S~`
z;yvUCR5FzzNZEvJuozbz#h*^hSyi#U<Xf6_Hh8MG+NHs~PtK{3Z1yod0-N2K;H>MQ
zLhF>=DM}d>Vd#O#U66AvB`Bz=VWr)C)>+Nsqfj&s#P1ebKh2MwFczDh<1T0;xM4<&
zi*#2^Yw~TMnX5r$kgGVnv(54WI#;&X?8L#>)*e2v#O6fnl9&bLeHL57+&s5iZ?SpV
zoO}<!V}Zkb{(ZJEuwkgDkh^Wj?d>JIE8#?zVziP{Dczkb)Aa*AaoVv(xr0dcrCjg#
z#Lx=L*$av(xvRT$xKi@o=%e5iIijgzb#3%<t)3nL#mOqsMNuAEokKv`6&n(S(@6Pc
zjRAFkZ;M|KQ_8JwX(9)~H~4B<b4<%U=f)MML2Hd~eN3oAsXO4s+I{?55)(roJE8@G
z7&HPd57(@&_*S+9;**KZmiYp&|H~*_;qb4mC&Fn}G*zBumBx|B;QhHUVr08pYV3it
z?v@>u3yZ*0NSw$W7%2$8Y^y8ua#Qf2fqP8?1Z700Js9mIPZkrn*NBkok+*wIN%(s0
zVPEhBT+tuN0j4P($fL1jK~$U>&JMQLuQi43S?{_SM|z5b_E^j(O)Tyv#Tlz;4easi
z@?mc6mhJP23Yc;(E~jiyCWij0JbO<i-YNItVthVuxXw@Yo~;yl`<Vo5*k!QbSx+oP
zK;Ftp$*;Xob6cMa!ED%Ei=1!0)fWhAG2(2@ZQ!J$-h-&umls|7RdG#Qa>X^Rt!*o9
zTOQ|Bilg!FY_HZPTZ||>WozP_;+y2ztJh^$puF&LKUky?8ow-D7!G`Q)u#mIDA3pT
z`2D^0pKY%r8SF6@x@ak*h?37h#_T{s1Nx~;MB+z&kwkweS}WRqk9La6=+%UW$V5pB
zb2wTRZ7P@a75J)w#1jgrblH(=eR88#gwM<Ndcp!;fRaphbV!@MXoIMcXvcizawW3k
zj)WcctL#Jt4o{S;TEdEp-3MrpEA3<qm1c;S;I~fYp~=sec7sK_PWwV2lKc37JMB#L
z-o){0(}z%Op7n(+*MoU3es?vU0u(mc`TPc>sf&Om*FK3E!*?7Gk3++Rvd2D+^%qqS
z-#mauj%-f#K1<+X!&_pX$2b!1v`=M2^5gl!K{D<31@@(PLCY-B|Eq*;6jJ&+Cj`Uf
z8MmM3!gY;BfoR{vH3jOhq2ESp@lo;;-dkRON-A+j2)wfI9I45z(kA8e_f>PRCgMr-
z2E&1VID2HIlr%1)mYS6eRArm-y#%8F=tEy$lC$rzpYlG52NR3;b0Q!0&)wo_9~MUI
z@9BXHiw*m*a0(p_cEN|i<xtR-qJbYeizNfF$|@fMyNAa&cM2D~z1uGOu&`Z}3A2my
z=c1V@$^g5>zbUX3P8Ugohw{=7{V{hsEK?Q!uDr#Y>@pv$X-SP;K4gBS786@=g|Dz?
zNl|fL87*YxLLdP7cG*?&tz_#1fV*LvuKutxWY)AZ{WU&Hai`s9*T!PF*RToU)yr;S
zw-O$RGrzTwF>-qLDS!WU-a<um(T$FBz~0hE+#C*uqpRg9vxTc7#^fM}DgVmyAOJSD
z`nCs^rfVw4IUEQqvI)8R4$4DwpcAX+LI_9yNgoDxc{_>Fscc}AbLohPkJybL7QPWy
z3QV7p<1KF^@l>nb<UfvG^iThEv)$}l80bL`hv}bI!A9PzL#$>tJvUeA91e6CU}nU(
zO%xl`<`Cd)W_**Ki<D&8x%+I^hr#;ZzsP3$?8<&YS-@!QtWqi0X>&jH_gfEU6nbSC
zfX@Vjuk^g^NjA6_xGQo%UJ%Dju^X=7!VjH64_C$V5czv5mh4%(%SRaKQIaX1RQ7Lo
z`{H1+dr4Lhun;-nd3E9hDcfuJB=nnF8;TRry&qOa!1|^5tf1Vfb-*?)aqWCv>}u#>
zu`3=ds2afQw}w2bb^-(OENM)@-LlY@Cg9MLTXSsR)zZby#ai<lZxpe=5ZLY3?gP9=
zJeqYRiXdNgSy#3$mJ}J(`fXY4xCDRrd4k6BL|Bmvqi=g-r>%<KI;8dGspzx&2S+UK
zHp-~{K-Eswpd=Ia+3M<G6on<#E#9R=7@akFj03w^^@C2W{``0LP&NL}%u{3!$2Vt=
zWw+dHk5qGp)eSxNquEz%cC2IVFWSk0j12AB4n9qZXGtV#k0soQ)?8+f=h@4Xmk@Y+
zBA1IP|0R1e0oC4!(w8#K*?CqCI^d`CS7cVI8QC+ABwe`Y>h|^I5znb#V(Y7`D6gEN
z(u;-n#v<hqlWNzE)i>P~0lMG%{P2<9m)n;39n})Pul_b}i*HoMQmg@HidSrV%<2rE
zvn#rto2@@v7ZszMg?ITk{48<UU*fLByM5NC`zVyY=fe<FTa`&frk%YpimsD5cyc}3
z{`lrxfS%j!b|9dNbO@ht$p31oZ$S*OSA7O};1)aV=<(~3Hc*YczE}k0^l&7pTT9JK
z_@~z${IC5x`ha!#vL5w;u;rnQ3ao_0h@~k`P$f4Lz?8=Lcpl^}jnL2&F<B{#7vMZu
zEx=t^l%VC5ll2MXe`BZpaBJ%1%5o;!dJP2kbU#~xUU7ltM#h1ki(jN?#}FMl4soC7
z159EZHrpq$2PdaR@&s`zPbT}+Ta&s0uQ~fHv8@(L{JH-`rW?f(eUW(D)GD#wzVxQB
zp#&d)6>E1t8h4rgI-jo^`{u(VGyjKGm2TPk@BP=l_31D`x73nJr!WHF<r28g(gN}T
z+4sIiWp1ytAL69f!L6@@u>KfulEHet{S@6PCal1-BBB3Woj=bI-{gER6QD9HQ1{mC
zT;Y|Vb+5n!4Les}w2n43ofr5osMwkiXD&Q<<TAIn$?-~bNi|A!R>RLZf0aHMdx?&}
zsQT8+nbQu`zxZ4zeLBgFdda!CLEY^bx5W-z8eMqz#0{f(+H05PFeT$2ECLFu%g=>+
z4dxEqzv5gxQ8P)%vQn&oUFj_#KIllh>?+^WK+mo7?P?!zpr?&^?YP+2#Lk8a);Ps-
zH?GYUPbEeCh;vo!h1entNW6w4&xO$sc&O--gl^zoXsjx)T^AdW=3SmC_*~hj>KnI_
zLsn_u(Qyj8lUr@fx#06kw0fP5^>zHItpknkxT-rM97p*A7i2<w^JhyO_IjHb-}Ex8
z<oqjeXp`cbGeBy+*(Uop1=@BIcUkc8#%ertO)Oj0R;L7*oY$2&Dt|V465NzC6{V;i
zzRh;?xnSfXL7GjAMxsL=Yusnkt7+VYpG+w>!}qGnS>%SwQ5H9<$ZeO+%EO;u13O}~
z|7J~k7@a?zlOUAA4k2LfkV95Fl3jucNuAf&ygcOFnrdx9tO(bb>iu6BT?I&0>?(SH
z(DQk_%Q36D34(pM_k`_&=?ctRl&eJ{fF}5!#2}X+{NcDaW(*A*{6eqQX1{4l(Peef
z#NC_Ze}<tirz2w`_p_W@W3+SimI4xX+R|9OcuY_eHz!4;O0m#<u82E#LIztM1f%)i
z7;CR!&PG=8CUiJ%ilKyrc2xt)%>yk;<$AEJTBa0##k0*;MC<a_fJlSRx2tm0Y4vRJ
z{`_lkH@1ied}+V=%{W^fgT{-EY_;Cj<cT%l_WEEn2!S8q^w~oRzgq|_xz8T<_Cyon
zdHjgaByS+RM;`T-?VN?MWNq$U)zaIcx9h6W726`mfPMCu4~Gy+kH?ezc>Rqf2*pn%
z>Y?aS!Nn(YZ?`pdG}}|THXL+O96p^h{$pGBg6`USdnN~DeJoOfPLHEmr%HgV_sOTj
zm(*Er&Qt@-faPJM&s2oUPV4ib2hF`(a_T=2#)w9$+M8?{E|+dk(AJU&qCW}e)Z1-W
zG>#zpBBJgN;-_bMKL@Vsfu6S9!F`CGd#VZ0*Ty5T*N=AoKhC%P`7@dc&knNQ4g|i)
zeMt5rx#4<b&7o+G3L8C=gv*2Ws_$3!Bp#5%{#PdjaB;UD@WoS7xYds22{b5(?)3_T
z<zhUPLqD1T;@<I&9P@6s+&O2)?YG%+AMmvvh}4N}I32xUqn*q@yV{n+1f9wwbo;W_
zb~?ca-H>K)A0+ZjK7sCG$w8mZJ(hH1&g`Dcwb#|OwAy(;pHhs;KFP<2n=V04MN)@k
z*|ty59q|G&l#IdyF(M&8i|s>*(fs<0Aqr1@WgmP#&an5{7k*IVCj8dE^qu4G(QaSG
zzm@;wYj3ik1Li2FYFsJ9II)3Ziof->qjTAeto2>uR%87N`@TAS91H-6^0)txU_=p5
z$jXnobCzFkKV@&+!0hL2`6tyYH2Wyt%UTaG9rls)DnkmC2>swyE{I=5A|tg7R7rM#
z3qNu$#@{E`;1B<(kjTILyMM5YJ}U27snS`9?ZqD*k<<2T=z;B$=u0&M@|L)x;H4iG
zHbv#gWE8wCmk2v0M(*-hh`aG_E~0pmP3($~f-|ZjC$9V`)LCSd!Rb3kny(}@*xzDu
z41uUNR3(#)=03S{+r!%0wI79#G){z_<a^Z<1=vR9vXu}GFrXc^k-n{_74^+_UA4wN
zMetQ3FBf%3eH5%*Ee>C!{g0YDk)ScYR*w8-l(=JatqpjJa1F-AjD&KcUKG}Cu?Zig
z!I8uqj9+3kN@#D%2w)Sd{mHKi2T}GnDZZ6F5J4a&`&_`6Jk*f(8?#?A`Ku~cuTAmY
z{RvZ*-Q>SZVbg(7OI0b1bJ0h}(qh!W3O}uaf^?iibv4s-hTw&SU^5&To~>|ZG$^)8
zy2^ihmgCgag}2)5K%q4--g9y+XCx7O_S;-<inF%O3m_?pFMK9;C^0d`MNB^PwjjF8
zgZm`_f~T7;^x@H>HQQaW3He0G<8~k06Fc#o-5ote2yk9U3%1u5CHl;->8NS6d#Y3H
zkeiF$>+5Q5y3cAq3ho<zt;-Fzb-5IYA1qQ_NP{nB>_7b<|6+|fc&+Lk*;2n)#V@Hv
zj8t<%0=Zpbm6G^d<6C)vp%xSgvi58zNul7<3)T@}a&Ct${q^7Y1uZ>&o>v52;_XbH
zw`Cs%Lr(^|dU<uJhR=%JD|ttByH<G*OgdcJ`#&lg=H+$vKsIZ^ksgx2I^iI_nfo`a
z@t(LOan&BoiCa2J5uOk#F5AWEVh<-&YU-B3)jSf{NJh9F_NdR+CZ{#UnXir0bXGj$
zZk3*QU97Ol9&?Zn^xQGc9{1VFO3x0TsNe>vD+Xw^I|*OJl&ugj)o*($zLCCzCyN~Y
z*hikuLD7LcQ{4-_`)<y53fAM(@m}x_uXkh_b^VRj8$F6V!HMX63-4eXW7%ES7aQ+G
zUg@_j31$pfXsK*lg4rYPMZ~rzWJwol{ex>NR?R_nb5~N;#T}fO44AtU4hUe_^Gk`h
zJwsNAn+ud9?JawwwUiWmD@y$RIRe*FrY$m++ktF{jUDL>aeJQ~if`zsieet@v{&Ps
z<ESp*i*`8a{M;r`YM|PS`*6zqr`h-0k@&^QimX)9qbX_9>j|$r331^8J{rVk{!BYo
zEp~f}k>I~}JO)yh7sWd&r<;fNoydbm@;@0pQR||glumZPo$`qmV2hE%xSftRih$!O
zk^haI@i7P}XtT5a@7s9T?OblNy?fOvJD&$bOt)vw`otGK{<jnC)0`TmyL19C;34~2
z{yg43&smG=PVUWH?TgrxtH%qeh`0Ys$CMf_B+8!XD#)uOl5#A6UBOiXjO18_&hQ@D
zH+ee=i~y&<jggA0$alE`m&6$R`)o~)aWM3Un98+R&w<$c>_^|k4dW);Px14=1Gs+&
zXV&32|8v{|Y0{rJJAXv1MolgCe21NPb&cri7Q5hlNZHLb%R9T#*<5%&jRS{dW=9z*
zYG0M3wc+RE;hs%mm5a`YCIz~J+5O`4&NaC?RhLvgUds1!$@!r2m7}d)ngcD6r&07Q
zq2WSOJfx(+dEPERAI!{L@}<HUuCyzB(k_~9K-9f5*KkWCVys<tK6FV4dZ>VlfPQt3
zUOc*0IV>~3Tob!RuKcUq;FZHC+>6U~zBXqsl&_9=bVQ7=O7ytR0GStUWPFoKLkbn#
z?CrV$GkhY}u~E@N1c7`;hYz8K8l5nq;3SoYbqU7g;(57byOHgAY<`OnCpkTi^T8E`
z#XX<ke^It%3G(2?*i!L@W-~I;xX#)c*h#r%L?b0kQw}LT`X!$@QvKxIvW}`eO*dwb
zVh};M4h@o&P;_jlqo!29SCxvHaWLQUO|b)s`YQM;{qxOvB)rau#YhqwP2AkH^TCSO
zx8kDRXwwr+ov2|DLQu)hNRX7+HCNZnjBkA`Op4D#$l0^b7w#LIrOi%k>uhhcIsUzY
zw2pJ*m}H^twRzR5iY1;?D5^)XNBIl)93?2+)yqu6<7x|i1VZ!VHhEW!N6si7)4QXW
zkU*(}T;nA!Z&9}6+YY-Y+Qr2Y#{)FP?A_~2h2OSXzr+JQBmOt5i`|y6ewL%~_=fl<
zQUvSCATL>Cd=nE?Jc6OBORKF^iEBlqP-HbnYd5Lrh__p7EL2ZA1y`!^MD(<wSFEGz
zrBX98RqVE|T<gjjTjpC>imG)*G{tC+_Ds=ZFW9QQzcnqj)S$UP2~uHWpoI@aA6J`6
z`TwG=PMqLXZ@~8!Hfv4n3P*?{Y7jRF2$gZ*!CcGq+wGwkOjZmPjhEN+!?{D8J9gV6
z(Lxq8xkO3n`Ka$eQs>%g8d3P-bOZPmLA7phADF=HF`pGZ9v<G_zlll(Pb7BNAd|K&
zu_w<LhOld~Jrxr_$HWKVs!%J~(|NPP|3jggil5IKe$O~o&m%1aN}gXV&CBoTaXMh@
zfANd-YSO<`IM90YS}a5Dv@y31AQYvp8m@i5^ttnHwJixdWCI1WZQvlb`Oo8j`#aly
zK6C}Qw6(M}Athqv>#+VjPU%pU0B2W>qLg`hSrnD9+eabUz3r)Xyoo2k9gRHfM($P{
zY;QuT@b>QXv_B!=ify>w4)}Qv^h}p`{h<VSXS<5U+N<6e6;DaHbhug$a}|W{wSnrJ
z(vjfs0vwL`67QJ0z+Mk%ifi6Um&&omj^;v&D$kBZLyX2yH<|s#4XAeU@dW#LN`jt<
z>5!>J1i7jE>|}g%!tMHnQ{D+dvyPnKNk5=Ouh^L!?~b-cI~#pV83`T{RcSq!P{7em
z$>h$TKXNk+!SDYk|7V{htSQm4%s!37E47i?iI=jjDizyj-kM8GqILT`Q3G%C!svgI
zySNjz=$Ekz&t6$br+wuOs8vXXv9I%;YHw<<SMa2L<9BIAgMI6#b?2Ph?7KXKA@&sr
z4|MQ*zoA%R8tn%knL;)I=f{LCW|)`krz9T>rrFQ2sEtYi%1xBH!9Grg!)t3&LBfW8
z9B<gNHas547w~a;h^C<ux$xsC)`oDqk0Ewo1#>Y$^fryw!(&kA_=Gh5c2Q1Lo+UBz
zTkYbHg8^$0gJPFN6Cow;*tnSNwM%{O+u`6Zi=Q`Pyx7kLm(SAW9~T1jFMqKsK2GjU
z;#e^@7o^>9SH>@*DM}Zj5;Cq+z3Ss&A5oy<df8-G`*3yl*)`dzI7n(E2bogHc<slb
zjUpXImyL*aZV>buhXTbe8(D3b_=a*f_uF;31=ueUcx56VHMr}->!>jI=<Lg}i`elf
z*qG{@<W{F0V{^RX*{4h>2W{NP&Mnu~t+Wa8y9ZIhE1~F{nBPKG!mDnR^2FPv_@ahg
zwt+{$P>N(nithNv7Z_vZW5Fiou;^|s>du>d=7FA>tL)}zO<F;C!Ez03+Q)_Y2K|X3
zo1P7LeiQ_viqkVb9yBp33lGS>(q>i--Qc|shIH1)!B)$=hb+AWw)xyGC@5{t$HA^E
zdv9IQR?f|nZbr8&P29W~RYqbZa;5yUEpS|$ddID{Fn-p-<1d1INb%z8+YZz!fRw!a
zcl(wGdTv1pyeJw3f|jhj_ry0{0`aIM8nt^rK5`f2>%V6m|DXBIYJDFQCjG|h0zHcd
z4G1T_nKRN2dE|B2jfvcDjD^b3q*_-hzO*#H$=5)(E)^JVj&FKkB@4WSrmfY{;I%%0
zx!&4iPF-VSQHquhQ62HTBcU~(tm9qPga>$8Txq;4&lN3FQ`-uJ<rVp5s>+EeZe~^9
zxWCE;nb);g^yHLre=Zkf!qE}t=Vf~!mxcrHzuD^OyAksWn`EnUCN7DcK6b$|NtT>`
z#dfOBo%9tCM$g<1?#le4;J}BXXLxoqw~9k-RQfBGZMKJ#$hm^}5r;u3i(-$)irr)`
zqPWf0CVin6MwhLtfGHJ)vxjcKe9XsPPEeMD@gI+6ltI{<!nj3<6c{ZL{)q&ehb6#8
z9b^Mi**qCNpF`=KOp7>2F>7L4p31v3oxsUfn3AXSn)39uleV69cG@!ue2&Gb^Cs)b
z?Lo0+ImQFS*2gzp2t7hnIksNEyd5|t*v7oPdS{hyuFm=r2T1MPR;<}ua$wL(^=!q7
zZ_8~UcuR0s?))K7lWorhVx$l<D?LVkBINBe=Gd+T{8SN?-PIJz2|~tb$lac5?Q$v5
z3$)4hR^N(OnglG{pCjRLc)$)si-M4WmWT585sc7?V(-=5A~pn~_z0#y9KA%;RmN8~
znt>#+Gv?V5|GnsqJ2?Rnwb!GME5`huAB_v3WM`Z0SdLm4P*jk}O~>P#)+dHh`HW8_
zU>Zqnbu$0B50%=fs<pyW1mG9!be<WdAkYaZK=@2d`)`L|go8L+-8<yP|A)5s5AnL*
z_J0=<5fRciA|fKXex>8?>-OkmZnwwNH^(2_+-8Hi-JXueZ!^w}Gd9l5nIER^ws<^!
zqZ{G%MD*=O#AK%9WNPX*GqbQsA|fIpA|lMp%*@P8A|meN`C4neXFQh+_m6wo>~P+n
z_xtr)uk~85^;*BzMJ~rrV>o4g6n!2R^H~y{Dq+cabkqkUzDb75V@~)hpnEO)c%Cox
zYkc`9LPAkeh`5h@yBuJsS>iZK5C7d}C-XR?7h(i18K;h~Qvs0PQ$au#yE&ae`gK_1
zhR)=gvG3vAA)n07=4%w&EO!b5%DJJ}8dU`yHS+w>NiCAfUNMXkzvv9*LLM)Qik1=x
zTIEkkAViQ-e2!D`b1tELDoaR~a{R^N80?#0`n8o>b-%h~(|MoS(YQO6s6ox>M!V!_
zkr>8L95>D`JsR?$)Mr?S>LOeg&pa(d$l7I>`!~!n)2=w`hKh+bcIDBKsQ!HCf?0R}
zX@QN1Z>2Niol{payXt5d;l|%O$*#^9;8@9QukjbuUudaado%^R(&vxUoq?X$#iuIg
zF6~G41Fk<BBISIQR<j#&syYtYjq$!42nuhon~oL%^0&XSkui_wVD-cLl$IiE$Ec&}
zUZt)WLh#K;-3BQy-XR<9f2rECV#O1?DxQW<Xtj-Th}7>u41KJ>P6QS`S!|r6PQJGq
z|7Q;lrwIXy^GI2oOT-YS?}`4B`iM@-mG*St^|i^q()!XRYf80JFVLqL<lAFVg>JtF
z!BHZ(ct2jXJF2$_ih)}Wwc147)D;ajJzmgF;%>{GHZ!NEYCRFnvyO)7rp)lA5fnN*
zzLQN++JZU{%t;EFPY(!TY;HbKRZ|7e)>Rv>zNLzU&<R%0s(ICT`xwt%y$j;mR$Mo#
zpCfg5p%W)HR(ovGQNP&3`9OS;T)i?<xXa+49FtQmp03iMi}M-AKG~h5C@ra;5r$r*
zzdd$u?q$0Ys2k!9DK648UV|JPV=d{c^r99LIGO{lk`WXvu?RsB9lEr&#Aw@4K=zl7
z`kO7UP(@RDdC}XBhNRoEl*nUj7L?XZn0h2-cUDj<?aUp@T@{>UyntVfm)4V5g6q8w
z78U90Su$V=lkLj*Mzkow2OY~+1;KW7*qQ_uo3=hD9>X3;3L1aw?`&;;59SZCLS;d%
z3yMX6S#9h6585UZ>+oP=t>^|teu<;_kgtxv$V9Y<{a`92S?RL&h|j{=D!nRU0ITiM
zq|-)Jg+|*@4e%US^Tr{BQqhR>_?WM$j!2Kk17rzF5C)q<5%B6+X>`VAzuTS+V22kD
zk*DT>JvFpwd4j>s%l34Rz_uRZ86G-&rh2a=Zx_pHUyfSDHIbm_dA(+voVZ(QS8AIb
z0Sbl1ck-NV8QMlwLN2OqacF@sN`qN{%%uJ|yC7n95+1l{CYP98<J^Aq$bTt%iVQ;X
zd~*Tcuh4=WJ`A!)6lDUjyQ)R(E~Z7bXx;7Dtf5HD-X!;8CeUp~FA5H$?8*{C_W7z7
zKR|1d{S}m3iV!Nb<3N%GN`(rm5o1|6{)=|dzigmeiyewDMKUrXORIb1qV-82P-25}
z<qrFoluGHaL8l8^*WjnIBQcI#f`szPqHS*{R4@uXqhOl76JwW1WsXG&<SJVDyK!ek
z7kGOw*6;U<$~mh=OlhuG?wI!-Rimi>`$1kexTVx=^+P`za<NeFz<<QODgBs$^MJ|)
z_M#2hC-IB`Tm(8duYa2OtIQv6FFatMCBvAdl^^wQC03HnkNAP?aqMV>&y;?(gbh;X
z1V{SuYDg7fLT@aN74hE_eypaEeMS{4pgb9;jzm*V%&EYIik-a<e6%=*K8K28oX#~t
zE6-|RMBg)hPuKpFolW%ADG#mN&L#YRAUeR#$G1{U6|Ew#@P&AmwS-mV?x#3KZEYR)
zb8-P9ze+O0FY!Sk69#z=V)-=(K?~l-_U|Ax#y&5?L-!iH<ns^_kOn1%Ui!I<QK{Iy
z&f&W(o~Z&83NI1XuiE9Ghqyv)GnF&0h*!Ay%cZBH^jB7&VX_98t#FYMpBIViw(&J~
zRnCO?vlV2@2Yhux)7e4D^lH22a~D{)Aio~4Yx4`<m2E@ewAWRWyjG$MrTt%@uuX5p
znQk}uoZU?wcB8+of&!)2ykgyM`aHx+LeACC{yrO7&Ep-a_rP#ReSSo)+`+!uyX<Bk
zu;$hYHaY-QqrrZ=<#WHX<7#AK16<7VL4i&N8509bJSvSTDV{d=^CBe<E2I)7Rczen
zMcTysitJqzLu06`9AT35Ww{iCoKOudUnpz_Pv^wy8{K`fg(z1;9wdd<?6*k?JWT&(
zo9xpmR;AmfBvQD1NEt>=zTFo+j*@_PILOQ_XN)#>4|GK?pw$Kxaxv}Y07$)^9<xXi
zdc|hO8^uosBUI0fSrys{8RD*gR<_v^NG9vrL7V+~M!_K$(U?s^zntzl)sj6<UY?t|
zJ{iTn6^T(7Zy-V;qPwMLUIjquP>eRc(-!#JT1b9hSbd_J)z2SRpQA<jNpZxw1@@kJ
z=9M=UYO})@$2#OV@rLb)tL3VINnrV25_l^S)@47uJi_yKuj8Ho2lzrmg$|LQ3?fno
zX+s9O<hE&yVcfPGH&Wpj%?T(*Fy&7vo1o?MG!>KiC$ue(&*kTrfwKgkme1Qh521`(
zllw$3x93*BC7YGV>feA!g-bg9K)k6U&t26V_|#>S>#0_%($&x@)Zi+~$~@tv<^<Kw
zeZvO(usH0qHI4`D)96nRR47*Q1~IckwzdL3Lx3S>VB2l$^3Zel;BQ(VAA=-u_p-hR
z6HI(i%i6h<9{T*q1Rd)CS4E62wrtwNLDo_ccG)AjVBM||?`v`?5kvH%J(_#s=tiM`
z!8XJWOBj^LSt%&uoVwYl(PSGFDGh4k`dE&@ZI1mM+uTljJf6u~w6m}PWW0cgmL%vw
zPktVL!Q%F5q~ePTDL&=9*3p1*;^~0dEr&9a?u0$aFaAsgpUSd&<#D%HqJ43F^i>e;
z#$H+C!E94J(_e~dmf?AG%z#(5NNop8D84k)Ee;3OzE@j+;*O&t11>UL5X*p*g9;Ef
z+4h{EL@Fu4cEEPzr--_*h~O0mRhF<T7AlX|?us-}@fvNfkBYIUiBjqLTQiKV&GzTY
z@NldXdxEtd$XDu@=|~)mIaSz1;Cf9P`Ku2lpSq|Nx4n@-mi0-tlkBMI$ip$4f|_5B
zH4o-8QQ?Xg6bbG~67Eu!X~R13wl9Gs=^4}QoeE}R!J(#?U|H{0&)lYniPYXp`hblQ
z312NknL<itlG&9=5&OVVhcdF%KCGaTXUh%gA0;#mt)%vlrtxv|4JFvS<n>8>CyvF+
z*wdZ%X*`ohR@Vk?sqC|W!mpC8P>sT4#b1tAD^R42AfpCDJS*&!?LNS{KOQ5Xk#G%p
zT(OhDiHpYevPf<dKbhxCRfFtQj2fp_4AD;K?8Sruk43pYldMa2V`$~9kAi1c6~E6_
z`1FtuRfbfA!s~Y4Z<kvqjI#^=&wBi*KP3z$qWkCE(D4%{*e{7svbhnSC8Yf70^2`~
zG&>do8iIfqc-m~2919WV4<*bUapBTqMXCTFo^q^o4CmXI#rNa}bGH35yZl((5rw|E
zg06^{0-ZvFU3sj?jB`3{L=29*sKjEuTBHj}6jvP!16ghDG`qTbjdCZt1PUWobj>x#
zLU>e4GLdEZ=hC*`20G#$3qNHqt#j;p|FIEAD;kd7aI6TB^QTcnp#o9{d5qdzu{Rwn
zVp?O<(l#6E{~kXH>)@z>MW+w<0MeKkg}fj)`&&?ZkB#=3%Z+l#Zt-8{+&OoKjmfoy
z$fFir3ai{^V-ug49w;wB()Ixx7vl(~bykzRgNd9D!q^NZIKE49;L>qv;<1o$^r+W`
zc_?g>qkQszwaGDy`+u~n+ii-&SG-NPJ6bwg2!;25jzS-;&6!I9S23^Y2`veY((L6|
zof&Jxbyno`orEzJn}E&o8D`P@)@D~<mPB-kH8v-n=?2M&;JMX%UVu#H!s`<I@>uCq
zAb7YqFfZSBqc?{9J+>gB?o^LN)oEYoWQF0Nk?Q4+lz;rMc29h)<F0lc&%sg`SI^w-
zB)+^Pw|H)=-J28Yf&vspa5cnsZAFlH-5M|2>cTL1PiZdA$3j<XY-*4_yv4yn)WmnY
z{FuxBy2o)Nl?deFp#^Pu#(Fy`S-?8leI56AT4&Nf8BknFT`|(|61Z~NWUJ<$TtdNS
z(Zy|Lg<*7Oc26R-)AuT09I+ntR8U(}p<=c@5aXy=%u59o$XEVskUpoX9^c(T_TxGq
zVX$wtOepJfc0_aC7~<6rI>}9^3fDvaG65aRvVAySF6A(Ic{fJ%M*`Hsq=zHQ9?hW#
zUxZaqaE{A&Lp;+JsjJ~t2!CUR>Q}igGNx=OnEJ8gLOnD&dOS`EHK*-~YOvR#ek>Z?
zluyPp-BePN)Dq^Y1O#1N^?&vt;XPepct~t!uhBW4N&GBTl3HJU>j@ntB*!-S8_c*F
ziEwjncOxAp8*NK9aLGg-W?JcwUp(vI?ZAF)269SbEN`{#F{Rh}-Gfo>knO0zP{^d)
zLU#=<t7w7}^>+J6GZdKBZF{RXU^FaTY1{h}+@6xiyMO2qP<5@z4mf_*?$ZwX$yL2X
zQZ}i^WrvdL=^WK6&EO4xz0|Wg9JA@-EeUPtf`k4$O~|Q0<z%I%eQ#IGp<!|Gj&}m9
z#OumI7=LY|j69KE{+Fn}q0Qc}2u49;Md{?fL;m*<oM1?k1>_I?4|M}2-Ye!+1j_4t
zBd6NOesD+rcl#vKg$gd>R9E{?D@b1~mdrhzkI(WM8o?n>nyBWeFLbbPWxX962H$38
z`OmenjwjBM<<ZUTgd=AB#PN1AXrv@Ia0${6Sp&ccPdN$(`~HNh+fK(6Wlw!)p^BnJ
z>A!Zy-|pyAB}F@%H0j<}u1lnA1S|@eCv$FZ>g_6_l;<lXs1HB^-Y(?abDEla?5F&T
z&eXKY%lF~7D(a=`dcWiXyLbbJ|Eqriocl2v*%xV(D(FK%d(JNTB19G{8xzIdZkNVO
z3d4Q2fboi5_C-h%T}{nhc6q!}4g-r<`u-L9tlU2(#?!9+BBYb<4(vfT;)@Kh@Y-(N
z7~om|X2TI6u8Q9YoZv<Yty<S3WGb!y>KyJ)$&Gf+7ezu5W!kmzX>pP>oZ_Bk`_7Qz
z>%MR)jvyF$Cy+vAFpy4ieWHYW;SflCj(QU}d{KliB3yy`#+-FoZ!3*BZu0kK_-gN1
z=`!WW7(-ho*`8cM8<qHg4=UYN?vk4w;$G72Z1m7K#ZHZ(!*2N^qn3KO6%pv@vlp-F
zWb2YpGoP_v6!~#ki;at4ix;0?pM$;D1S;H<<nfo{d%Fuwzit!a8^y1_&NCH&ndsZ=
z&0}m*AWMnN3e;8qaL<XET%p;;kD)?_yv0$R5@XA$A=?;D?(Oy=7%{x4{?H-2<BKAD
zj>9cf>o^(g`{$81J=f3C#GS59p7}+Q6iVKgOc}FsH%P6*;WoP(d^=clt!j=hhC70U
zb(<S+D4a=^GDWL(UxbujdWcGzn-_1n4I0Y;?0P{?1rw%u=fYUKD_DhMFS1E3a-zYm
zN#lcie1SE$O|r$l0+K9yi*IvDp0|c&95*Q8-axlU4r&dXL@y}i4T)}`mdv8=z-|$h
zFKV|nCcacX2b6K%n{&2WI{M&8Yw^EZSwm5Omd8Q~-a|4_Am46li)S1Q%<z5|+Y>10
zgZA=XBeID3ox`YO6F`yX&Q8lf)>UChfj6$z7yD>jGd+n=byM%QmB~{mK#ut;0VngB
zBW-IOon%W<>-&M&Qc>g{Tbr8&RI=KtBEh<NCXskQVw-B0tgj%2=74c$gbf_D2c76h
zs_wFf;!6pJ5~;UCvkwn_St3YSU)hoa_DDdYEHC;E$G5UmzK-0y8`|Cw-$wKZ7nM|W
zDQg=m7}W*H<9Qy-w~JY@tc4ZH`}_D8N3Ox7c98`n{;((VtD*{usGnuu)Trsn3L+0S
zlw<9wB$K79d#Ls+GI=@&C|VeY7k=os=`$6O#cEQnv_3~yy$aWF%Cke+QPPaD0&R{j
zl&JZNYNo=fw*)C@WVQnh*&|7u6K=A8AMrPT{mWt-i0{LDiu6b%SM|&bb&CD(s7^zv
zB{_641bOYMAoq)fKA*NbNs<fZkWLUft6$tov-z9tO{yb+yA9{>zQBu63du+giOt)e
z>uzrAaI5(NU&CM@@lgkpmMP2FV~1iCv0!D^DN@kI%{Kxbk(mm)ZNj|mQk46P59b<0
z=*&$1oed_89bL;{i+054Mev(wZ&!HQ0oN$U)ZX!vvw$)L)V@ltG_T`+d$0O(oV_1k
zK93DaX>!U0`XGp#XXxV}=8C&}8<rtXblOJ=o#G*Z4!ehaT&>9KU()Bpb3=8Pa#?*6
z-$*MEMzOrwr}2$09|c^f=~eouDYt2#1^#4#Q-GPGRUCguzo;q*R`REFOs}k|@P=J>
zEEl~%KFVBZF+>&<v7HD2JmyN<O!_pSiSgpVwUc?uN!Fn=a*j}9BK#^pooA|~Ll)M0
z#(6#&33fI)T|=o9cP@8ut~25Dc?GO&?JBv^7h(oau^_7l3(!wrL^usemB{c!zGC71
z*>~{G|C8PN<%=VuU=S7k{O9&B!=e%FSD%3XDUG-U911(0J~oBR|KIiOOOB_%1EcW@
zXz9}983cu+or8bbaTlumM)cTvugH6Oe%jcICcq|Kaoh#Ztk6J;n_W3Hza%8={m=5>
zy+z3q-H$k4WKK#oXoRbB$bao+Oi*$4ahD-##@jUkh$K;Hxo`&Als4J5F|V@Y<g!)U
zhwF}ql+~?>|Kd`=-a+|)N+z)zVi*xI?#AM_xiOxJ&yx<#fB$AT9e0s*8CB;-R#Ve?
zfT{8L+Nf$Aac2pjFWSxdo+_e1f3M=F8+|;ifAc7te2e2@u#c7Z*ce}SP0d6bTS2g?
z_zktVamR~nco(%s{a4hYR^?`v$e;;1ZavXs6XSDL5?8-3mGhhwd!q2s7Yl#l@<&~H
zuKatGO+H>EdUY6ACCn)f8r80q;eUG!D765#$>H8%clZZwD6NWJsP#WlP3V(CSHS8E
zj}C@BY-YkZ2PDqQU01g-32L)_#-i=KU~}?kS^F@V+uR%w{SsfF{6SXdn?mukyKG*r
z8DvXrkq>Es!$kfgiFi%6aA?fZ=-3t|+0(m8Ek*CiIm$Bq)?6IVBss1JSKFzyvBbY4
z=np3<1K%6(xRK2_x*_llVIhqBO(CWv9Yy+XV^B6_RqzG*e@l|DHTy=_&a{@}Ve)LH
zC>v=ks=NiJmM593raOHW5Tn{0wF_E^`Ly;JT5h7v?AA`?K~eD*>r4#L3XKEYWnBT~
zuwZgvQSE|KN!og<l}Tpsw1|~H7p5|{eU)#)GPM9)lW^f4$CoB0^?_t0@BpvJ^8%x&
z@6+1c8%1_dV<HA^Z$x)r@3X3E8-)?PozH{*0zt#*3J>|LC~@ufaBkvE{^W{&B>z^C
zD{>DXjU|XA6ep}++8Z3jcQ3My0brQ6O1P~~FBL@NyAs00iB_Eb<JB`Yh><1X1$)Ae
zZ5IVYo~)*l-@w)Or{Y;Ly|VW&p@lx};FDQHNi1KNf+p^?zSxeu_>`Pi?5}Z2Zb}Rj
zkQ&EkvoB@w-`kelTors2o%H)Or62h~fWgJAfH#;X4VV4IEu_E}TeTzc)<SJAl$Tw}
zoCwlxvfVzGOklP*x2~~ON$>k&Xb+q4+kAiZ%ulQ2i37(&H$x)xMid8q?X~cjMmu!;
z$UP-`<YFiC|95}c8^`0`T~SY(5+%--vFyVEsr%K6#AbuO*x!xAD{`bl+}{-V3Hn~J
zw-eK#ie~Rk_Ktr*h*6ilTOI5dh_iGN2yqV+|9cg!y^2&KElNVOER63v4NWJB{{zQ~
zBCCiFi^-8-3e(dlA0_r@5yRDHAN$lzYMJ{<yza>e*tU@+K25%_1dUR&L)kO3l6X%0
zXUPHFrUFSTH%6V!Ri9y}9SyduHUe}C%soVOfv0k79UlUbi%%-%UOQ1OR1wOOxHoG|
z6t|NOEgZYfPWhKaqY=?j62f}ybP^qQf=)XVW6Me}gA&FwJDVf4wA#4}cGbT3tdQZ!
zf#>6Q&5fbsE#iwnI03(`UG0@F#Q4fFQLLJLT|fEwZz>k)=W2d=Dv_h${l6rHb5#N0
z*LXu1D|vui@%E*Q3hmVUxy1iXgExe%FN>r!*r(WhEg6&JWnYHiMXfLO@45WT5D!$_
z{R;npg{cm6KVlDAl|Eo&6JD$l@w>Q%g2yCUUG=3$gFVNk|EgV`!xP!Aw1R7fzK{~c
zg8J>+Y8I+tFqbsc>*AS51pC%q|78e`^C-Ke8~=tcUCMkO%Np^X;86ZT8IIhBapRXR
zM8bGEIT}QD38;)=H~Au(70EL)SKNtbWK=ce)}n@}U#OzC8NRIU<}btSII~L#8tsd%
zqv#la-BJOq5N&xb$gA|xYn2&jWBm`V1?C->L&4H$N>lS?8tA&aJIF&u;+PP?$TN?|
z=341n>lL4YfsIyyF`vu}J2@>hCbLPgW^a2w%uryH6Me1iEmSG%u_^fgd8^d#z5UDJ
zk3?rSm$dmizKj^%(v?^l1_%I10#@q?oObv088$s3$L7>Ve;uf5W?};O5aCsrN1n4;
z@hmE<%dlXxzYO^Tij-v0+uGg2#?FaXv6m}&vshf_<~B{MZ?L)?AN5LFw@keA@)^4=
zACYSLEbx!WxF?BVVUDn(x6u~G+d4Pm8n{yIo_MBnD0hrZjEnslb%bfAN4Tl3wI#W8
zSiW$s-0L(Bg~J0OgN8T(y0Q0IV{AxjGRl9HrE<WU<2~x=!GT-yq?8T~bk0d=r{bk=
zOG+;3j@)yV*4iGwh{wshK`5w56*k>@5k4pujWM<>Cz*+q%@y0@APn{`YeR=$nN!yB
z$ExacodGmc%p_~RjQfjLnI%wn8OqWFiJ^H-D?8dscQacXAIl1%<?1A?i)T8AN=c&L
zoUPAE5O|c-@DhIc2fr-J*L1a4u7379$Etn&a5eO6x+02H`G`~FZ<hS!xAthF3hS8q
z(+ySU*$`iD!5$`iw=CKlb1T(vg($8#an38GGOH@o5J!2u0!Y`PXKX)_IHa({3VSl%
zki4y!8o6VhiWMQDK0P!+xd3H^80I*zXA+)9<b!VO%iSI3sV;^E+Z6N2Fo*`lzHhb7
zKF`$oGi*x&)m-1b)cPy16>?alavR7u?ovYC_Iv~9K5k}M&8xN}Cf^Fa6jn$#0RF35
zD7*Yy#TDUD+?^Qb;3-y7ag{EF?M-@_+PsW<uAFgDurHT5N6{j-KmY75D$;>K5HwS8
zuaGgXZi3)UTF1edOkxYE9-Mqn!+XgNRZ9?{V%l5njimP;mC_>Z(I*r6_Ht4O;}e%j
z302}89`Qra+_98+p|^8PygoQN#DVRdMBJNCa^1b_OtfT#l!Ws4a>{n)G<4hh$z=;i
z<%azrRv;SF6-W>;WGa8rhxut|TQ5;;_L1{r`4fZld>rs>)qZosaK7vlCw{n3mwlS7
z*h}syjO8<*pDT)db;a}_%~|GCgAWPfSOpJmd$vy%_BoxTX&%qn%JuJYQYWhY@p_TE
zbBD!1*va_a*_~^0aMEH2)R?CnS{$BTcDh18rDm9O2Z6|E5?$2XqJC4io%IhoC<$Zd
z;sa$4WBg>lUa|AOocac(kq{eyA>S#TpS`2}PXScvKo4z}>gu|H__yM+e@>J%cQrM^
z(th!SRCC(|`?Vr&Wm9<4k`uR`aOuKV>fRl!iLCLG6GevlgW6izr6-Ej<nKn?WhY!R
zQV<+<-A}mLN;W@TK=LYF;U8iyK#jaIUXU9^@u%YHBVq^zZNFq!1qA7hl-BI|=2xGH
zxKee;BG1$7j$d;kjcjOC#H?QHpf&Z{btkIOCVbb5PwQ4Cl<Q*xwz2p(B$WH&Z-`$+
z#2x_Y%Jp`;F~BC;TB0{B{gB<{VAG!d_mrI*8E?p-x(SqSL#Q6*Z<M;wU3PPh!5$Ug
zB6ic!@l3HkDwy~J#B$4tBIv5VhK;#sJKSp`Q}|{Z8(+$GE5o<^CFA@{Vr8&q#daas
zwA+MunS~*Gc<9l@6X|PiT}3YBq!S@_>X7gXT*H-%IL4DJ=+(y)85fQ@C1Jp@!<F*Z
zzu_CZy`qPgU?g`qM9B0sW<C+JAO$Ixwa~+hf)+^{^ATQV##-I5Iy_WsR&FttO*IIg
z9rJq5w%1*l<L}^<pou&B;#_~HG;r)~a06~#!dvVS5;%6*yc6YOt|6^<L4eLdWN{1Q
z<&55x7PBaxxz!=cStJ?2>N{<*W7V_#yKRa8S=T|13tZ}6r_I_{Ylx8qykwgJl#s@1
zv*iwSuZt6Ab5gQO0}nM@ZK>Ad_IMbQEg!azp&k}$RG6Zagtl0uF7cPG-B&*g%VDSQ
zMGsyO>bF-5_1>+U+1Z}lzmk?)stl|=5wf+Cxy!|{DyEl=4jJL}lYd{PGV$P-Y)x)u
zcYU*%(gP<#_-bxzTIFSiSKC^rn!&!hR$F%>s^=8(&<$!548$oZxh5KY>)&m?e}S2Q
zvptwoNan;X{ZODasym`;$VDX`?cr(@UN4Pe)Fktfc-tdXq<&ypqY&&id(__^?1Lxw
zE?ZWT6gE`YR@DfHNa0uRkltw<lfY?^Ymeo0AwNpKe>|tdyOsmEh3g65nr=$WJ(>8p
zZ<#%n<W|?)&}L6NeYtc!)}Bdh(xnu2%&ET52dbSh&o%`FX_g~j+cw8D@ms|=k*Z}|
z;@PIbeJrK?*ZOn&xd2dz6k9M5w4|DJ2W`90qXrv09ALuusINL~PD19e^t^1lb3u1a
zv%T?qWCFq*U$cEN@Fvve?Y2KT6;^Zj_kj~dW3FFWPgs}O!NJ@FL^b8}(@&oN?&S96
z{zAY%6f<l^qxR(8m+Xze`*1IgJD3mq@Fn~J#i!c_eO;(Qb|gkpIvOk-Y4|mJJ9kj}
zpvW;S7W?d-ScS-Er@iZIR?T30FTU|xwrF}lth@02n4#FLIWweB*$45qN3R#9B`&NF
z6L(M#YsQwe**?m#i5<+?XScz8>~G`KT82cR8b4H-h!Gra-3+hWrvW$lksP1T@;FH^
zM8Q59*me!V^>_W7Tw9crLv}2ALJx7Ar3>HjYE9(k6+!(uJCS=z0TaTXPYy$(L@(Vw
z&iYQ}i!@`0vQFoU>2;<R;}5nodB~Q+V$UW6!XPyiQ_+trRZBc*=X`PVnh1oYu0#Sm
zlY#)HJLQF#K!;GlPz3$iPd)**9weBb9UpK0AA~8W|Kl$?MANEbfgps;38oqLW?z-t
zrFB`4UGi1%N>c1s%a&}Heib(`O_4$DW&ZB(C*5Y3f8|n`yI57fTnS;ysJ-GV7qA@f
zC2q^E{K~Ix*C@1189?PPFK60D#P6~Ph;0_TOY!`<&+n=jQzrR;<4a{bT^-NBGL$YI
zcDr5kRS`^<sUYvQIcLcND*JHJnY-?*B9)cf1QFe?&$ZGrmm_<F4~04KfZh025iAw`
z*KIfD%2ChckJ9bvj)Re36_InnA8gcD5nCty$8Pq&7Lxt}eT@FfuM1YrSud_<ZG5>;
zce0^QRJVK;MmIS(NXW7=iPn<7=8pQY336+jx@(QAFtWYiR~gnd2}x-SKEc;K*mrk(
zsZKXB2gN(F#bL2YzF0~1_t|9soDh)bA>%2zO>CW9ZLkAA68b%~|Mo-;&&tgn5y)HZ
zj`&gwJy7UezEXK!eYv&qyDXYY4LQW=8Xi6{JpffDPbG)$RA_z$yr+njY)J<m2NDq+
zvjdzI6;*2*&daBBzDlDXHbLsQVqBUVFUzt)%Vy+U>hYFJy>)Ry8e46i|C55RwxF6_
zRVT{2Y+*dp$`FrAQQ1Ygk+gMb(M0#;9u(7Tw=MQDYie$_C3yg+u{$M1_}+_Vk#fGn
z8gfg!xn8U>-r&#_oXKO{?9cEMV;DlLlRnqt)BI-f7+db7jk?ofZT|Syalf;6rzW*$
zku9+^-<mVcx&k>mxzZR*4O&>>R_jS%meo_<Xk{)y<@s03xw9%U-9Uy)d&v=7<5Nx;
zcbh$sxM=TP+Sb%+Ykm80;Bf-iIp(B1lN`cW>tm^6e3>KhV4S8jIH|Umpj4fNp*rag
zC0R_HJJ%leWBSMW_J}`Lag#24bl4i@BOxQ3l`3j^L(Hoj5CxU&=MLK#kfP@yHsv{-
zM*dm1Js$6?R`N@dC%AK-@Rd>)7Th8tJn2Yi!ioG;^}2WCQ3v>^{paa37TPn3LGq6}
zoGSWq(`sx}VsLJg>MGghT<JnIj4c(yXwnB-@IG0;f7`KUwGG72rDZ}K#O(<TJXQ*(
z)Z-n2<KdD>$OTwt5LMxXyKPs_yqqH3yJHwms?<5<9@^dnm7>_~xUTmF!i6B2`*fog
zc4+$(pEHTOCyG8^^f}{JXBP(SVEm$kAyb)KprvRY3WlN<az*BWVGXm2*c%Ql5eztJ
z4m(a->(|&|wSMtX4l;Vy5hs{G%(Ay*KCiapQB2SX65KnvxH<xR;63m9>7hMMm%SHb
z%A%l~O3j?!_c3R-+6T!XDDs1KpvOLpNhInhSX)KwTzme=2VJU^qmT1=-dAs*__3=e
zO?WH+f3Cwm&4Y+uiI78Ih|j7WQKej&t?58mp-+7_y+Uf?%62S%(NQ7@4dt)c86bw^
zISz<@9+d7!qdSqX{9&4%bcWFDOypF&&0X=<ddP_gd^=sC!<~fezMZMw8z_}(<fJ$|
z?3wf-Ind8l5X+((X}A3HvYpRkLJR^?vIH{P1qTaf?gaZO1}g;`pmo(q`MH`vCDs*A
zzr}uWy694Lvi<67!QLvfpq(tj1=fmgyX2$`G@@PA9VXC*K~?O+O;MMgEW$-6!6o;r
zBiCgo!@c)Mw&L=WMdWIv5%3lMPogsH%99~CVO^AHiEwBmPKNwNfFiSAm7~nU@N#v$
zug-$f&y}cf&B-D)M0Yw^=CuwpdMd`I_&|Fqr>1%<UhgAPwF%yF!$mK6WYvwnkW#o<
z>4-Pw-|#)%W+P9A%qS-me)V=6b<!onY~XPSv$Tr`H&+lxTp@cfIuI%2p!%K2ls!EA
z<Q4~z?jeXyV|;Q1+vgB$U$?Q2y=i!i$K^u*GSX@S0_PXSv(L!VT_&6yvbV|DR@BdK
zo0uOjpfz&<=>Tbe_t@kBa)?LCayS&iIK|OE*e7AKv8mZ^zX$@C^mO`PusdR?%rGro
zlS^Yum!50mOL=hQEFq|ytx^&2={ZTUp|iI$9UGa#ZnK<Nx|#?oBgtWQ?t_d-98?()
z{J73ZGU_A%bZ)FSLj#_}WmI=EOdfO)>hXl7Zky*w>T2!2&lXf;E6xyt!p|?MeWxvq
z;lz<89d9KfcTv)e${<PtI_VNHnRrEOkj}dmqeu8xb&%kwi(?KIpH}=usr^%^e96g>
z-b(Z=-k)x}H(nMfOgrpaYDj<pMd}9Enl(BCdfSwn-0VcO<S(~aOT4QSBGDZFYs;&p
zD#F_{i98Iptr}UK<cq*y?YT)@agA+kQs>E{36R>Vxw_&LH^4~9JYYS;CN0&E2+&#S
zU&;H~LxuX}0_Xvpf;I7h->u?{&)WluV%&F9GS*h_?Bc9=bmO{&?JlIVWt{i*@sg|#
zN)=Sys|WKLIX`0W3W0pcF*JTsjXmr?V_~As8zw}Fc#kBoqec<>{b+7ZGgf#ijj-Pv
z^3h#ntKCn#!Y138<NSH*ym?dSF1T~Xe0wa{fKyuiNFEO|(HSfyAY`**X9bq4d}O59
z6S0@2YS5GZhaZ5aV&}FNKbi-L^dtU1Yfn2uRku;icAiNd1`}+hY3)+$^Kq1>*=3u2
z`FGFzC)-?+nF4vDyn%dSwk7da_!w3m?vnm|hTx@(VHaBKK*dU`QZp<^VIGQS-EZ6T
z8}>xjA=QJ}5zjmkLNe(=+vN~)+izQ8yZwbyH5Bj_Dqws4g}FHUx@=!=JzJ*gtG3_w
zjHz`a9|)jydLmS#+-Ch8ZAHW#JXsXmxeM)31)c)nWfb0JZ^Se93HP|{k-P11LPVf>
zDLfCF2a`^EP`T_#d?f|{WxkSQ<?RZ+Y7T_Q0g2Edcqib+N`kcTZjMC2#8P`N_+U%B
zw>x`3Pb@4Y?5IF0q40zJrP4AL_aYR3=zILvzy70r6o82dY%7N89rkhcp&VdX-yuSe
zuKgsh4l>q0tq`GnWVK=0Z=c1p0)RLrrr@LTOx01PKr6TGST#eqHIM=P*dEUrW=*vd
z@s@<b=QSD^xjQ;e`gXULfZ|jBN5}ornoh@T#D;-X{IekXOa*~@H|eT7XlG+=;^9iv
z){t_3&XGe)Ry*&&s8O(8$X&p@gn*T&mr|xOoBv!bTvKqJ;Wzpvb`he59)9(|S~-P)
zX<xheBi)`8{uTwi?Gk^c5cftg8!?khzb*`Q1?j|SwU>Pz;zt8TJc(=Vw#)tZ!M=rx
zqr0LS^<|)NasJA$i){AhMx?ckcrr(P?E*^)mk~`PPB~Rv^>v~9?k>Cf>kz~Uo5#6$
zO$_Tof^>{m?b>*TCW_{y;t<c-b@?8;@)juKb&kvRUmx-Ql_g)bwSmYDmUY9|M^r*f
ztsd=0UqaKWrR4bAO%5(KSqWHzEsp%UNRLzJ&9_kj`8KLYDZE%U@QZS6H-8=GqlTt#
z>H>7zXkW^or`jzE6Bf}OyiNyt8}oG$PkXx>*om=;)4SxX#{+KT5`<dzw!YVDa&!t$
z5R2Go6C6DX#UZ5ACi-G(I~rHpq?k@cVkrwz@*pPXGmOVFw<>{eN<J%=7YQnR?DoX^
zEVZq(I}(gm!d!7V*ZL=jUCJ%=*tF?6j9f`&=xsK$!gvd@%jn-FjdxZK-rb27XtS#i
zMJeUZu{pWUKQ_@!kHyUOoza<toa??Wa>%e01DofFtdrbi3u4F=8<n`V&|hd+tsv<|
z@q!l_@IbPAa?bmk+M2cBi*rs^wAYq=?TUvKWrR?+yrgsTZ!zxyyEjNezsriD3=C&}
z4L<W=UoD=@#u%oA(-$Jx%9{5gf&(m9ezWi3VBg$%){@JB-E8G#4B%2&?&S2_G1eBp
zs$@;-@9hCH*PsHTs#TWq>ctk&70)#1-<V&u(0Y=ZRjWcUVOsEWxH31eigZ%?$f{~w
z6?-YUezxXo7cGgkfd#9CM;MA2Pd1PeT4WEzm)e+qR;>JtwZ55yee@$EoK68{>wFmq
z!OE{&Uw!mC#!aR5ZL<esp^DMiV-NXXgDTf`z#jJZ2m7YAAT!fO5(VxNN86Hrwnt+$
z@o?>v%CT-p`X*jS&FMGBI}-7Dr+8W(%Uzs?J^t~;Gt<9fPxziSN=<t*J{49}=UjK(
zQ`NJbaEl^6+SAn(WSRqpMEGZXoX({Ut=8xNCWeTfhY8;DUJ^LFZF4SnSw~y5ZK+_P
zyr=aind0J+oIp?}GyV;SglxOpZ6IOs0IMR_Z}(?Kia`Ku6))Nm^RGaN*_As)_5=B)
zV)na}j8Vpv&WQ(H6O>y~joS7GP|^YWZC^}UZeh*~z69H!i<l2<vjchfD2}0EB|8|O
zh;Ee3Av_*R{@Busl|ivSZzN`5*(872;oPM8#7qwQAN7qL_mk)sG%f+rBRAxSRkrKf
zIW@I7id6beKA6{p91Zcjo3qR$>lk3)%US+J4*UDQj_FM_O=_|alDD<M{;?<}ng(U7
zE~{YqLv0W~s;25_pq(9c-UCcodOa#!f|#7op9Dtbso7$mItbAF^Ii5?;7Y`-^GNM4
zJNk7x&k2}UW|l0=$NUsfWBhn-xr%BNdn>{9L;@RC2OyV1KwCKh^t7~-2`q6`B_H}!
z;$bB*0;dBAmz6~eFW8xQrX{P3)iwz5Y;J_M)K7D;kaM|06vj<<J~2tTDXi@m9Q2hd
z?WY`|l#`LVD=&b;$$kzVB!O!kKqJ<}eYQ!UE0>Z}(Jx7O6beHY%D;&0TxxczT=5IA
z6<u<wh&vF;4p`}>r!v%FC^OS-k;4_!%T5(hYOoIpqd96yU+&KsXQ%7{kl!J@BA&_Q
z$00@Vdd03h<-(9hN63ZI57;UwFd}|Ypy5`q_$;bIx!LTh7>7tf@hN5)u&d*3ed|@+
z-A*uI*TmaWn29GSO7^w!w&Yqx0PEIrG+pPYn0VVbyFO5%_JRBDhH61x7n3?$cB8*s
zGhve5<o~>D>RcOnD#EWMKf*^wohqW%oBxYa0yq1ZFllw;9$k%js8BGzCcDMoZlFc2
zjX4z&NHAoXlS`6iS)k<Amu>8+A}`MWqmA>gD8=7nHK#%rt8MLBN|=pJ@E-^J=FPN;
zr@}DUKxodSn13VcxSD4E2Rh5-c!ofUvQSzl+LYW1X*Iwx%mWv1%jD$dwCC-PQ*j1|
z<(*q?wLi*=mIzkrf}re8cg&3=8*yf0XJ%{L3Y+Dtn_F+Q0}oPAWScI5pM;z%O6^so
zgE>CL^gHL<+yF?yud>7IfYc>I=yF~P1<XsliUb)?X#+(9<G*UowBS@o-*>K9(c5kd
z^ZUEeZPgE8QA|=ymhv{RZ}-GAm#vB+%@+GJrI%%k6I+tcR&+EbgSyw@U)9}a4Zc4e
zoeGX>ti~*sUJl1*Yd&@4s*=5wcDS%9wj_{K2?436c<S<e(9qRUJo|02>R}18?B!4j
zUVF@@<EcObRrKwQXSzjo)F?}7T@Ed+<g6!tmZwU|UeZBUR?q06jt;T|{clw@^!}1!
z!O68X@yrb+3S-)B59Bj1X2%w9x3x~aO^TdZmuu3|k{Yw#e<XIH%^q~>@$1VT@?R(<
z*<}yMu<po1dqDwyBnVPjwW6tgXbXPjJet3RAyXJ){aoG~VjLL<an^~U*v5E9Vgc%|
zjILYlu^a*Gk)%TT9V9Lvu*ZFSTbB#?Cz9OfK$rGpu8krB%X(V6?Wq_=BAX1BB|4jO
zqWkUX+(~Xe%tBZ|xU32l!_fHsnWW!lw5cBJ3lK%(5~#)5H~BMEU5}9#Je!?yp`;$-
z|Ja!P0Ny>*`eOu-xADUJ1My7uQh6>FX|uh0CT-q<*^%t<{?_hBZkSy@>b+xa_s|02
z!o?93TzK!PP_rp>)3XfYq3uh))q3}Q+n;|dx@(Rd2xPcqiz4^dzXJ_nZv0o?#Dk}b
z)|5J%1cF0}`02{)cq72jKJk%kOaLRsaX6N<72}vn!@Nwm$)G<&YOL=<Y%hJmj`(<d
z`x54sxBagUDkGCS^^TvGKh2nL@5XRCOkTxez}|~b<xbyk@8`nClal^{BcW7S*-Tu>
z%l4tubXPBT$47B2x={Q+_Bm?C-)f&E=I{bxU1V23t@ecM49+O0(mtyo*^bp5)kUI{
z57N+sz?46ivI?bsxgF18a1^2Yld<z}cA}y)IBScY^jDxM+8~q~|J`;fsk#e8hKiG)
z&ao9o#MFe|*qI?hC{hdhIoQT9WoHv1-6TJrOI#5>w%^XjK&lz?Do9c_w+r!xlG0wV
zpW+RfcuJjm2}?uNQs|^aw{uJ-lgSJ4OMv3uB$wbgH)y)Q#%Me%d41}JV&9}&PO66g
zf(nE}udJd2cF8y4WAprl!G1LIOTP*6Kndp<t}csrqTZ@tvCH!v^pIBM<14-?B1B28
zyYibNto_};jIj~^c(8Bg0=p_dA-qJH^jF8Q@HDuO-$&QvJKYHVXgk*qO|3*Dcn75o
zz6pV>2&%(<wAaTk*>6{^mgKWIWH&fSw~ilYH&!b~A1YGYKD()U2AhQfsi|)xzX>6M
ztFfn^EO~|Ajru0;-5H#%)poOgN8BO4zo?TnI>#!~x?fVaR3IY^xL+ao$e3?J+PQ0L
zosD&DwX0_DxNls{RoL>YR^xvWpwf<ybV6+Ea2Xm^4}DA);ypGoK=6D)d0{Df;1ji~
zJ%~+C7>oMjTxd!>6OAZTiE_nud%&GC>w)5SP*~9&@e8H`84;Mg2>bj}8^4sy4YpNI
zK~ZVHP0v};+of9XCV*;YEJaEI-aX{pZJIlif^C+6BfpPqe==3s><TJbv^=WLFUC2E
z&H1ZW*j(T9sl*=C#msV+Y%Y8Lip~2bqRw2z8aiR@e^s%?7UYQIfA>3E_)WTY)3&<k
zn~-FqLx9~A>v5GbLXh5C2b@J0x5Wtw>XGt_)#YG`16oT4)4lPtf36#|-x{1M#*Zfj
zyD@&xz!RmjH9Lr8Toa$zQjPyI<4bg)olcOF3SqrqZMmWuT<Pucx=L;rFCnJSPJf-6
z-!l1h<#vW-J(PmhlV`OK!vlBL%A`@Ys>N1S1R}GXWQ$CBU6W@>RU4^`5HHGpq_Tlx
zy7B|OU~7HhB-1Unb-8dPxNZEdbc*%yrQfU4%n4IN3wzccjJH9cL?;B3-S$vU`v;Yq
zw1)#JWY;3rJ>pxioEV8m6N8A7#hwVC-cV7gJg?eGdGJ-2T8(=LY@-v#j8(-0s0;kD
zY7}`dyoL4S2?4?s>FrPCE7<>NJj|6#lO11j{W9*)rxNp$sDjMhmi@H9#f|%%J(Ihz
z61ECg?W-sf?}S=zZ%Wu@QYy5KmMY~eIMC*5qm*kZD}loNRI7MPh2Mka4py~o*FpQ8
zhDfo6nGECreDo|6>|-1Jz;FL1+|r8rKxRZ?*b$8QcjIiA|8ouwtKESLDRVC{`(AFL
zy-BA_>zi#~qR!3f`{R9i{-73Cv)>NHvwqCA3Ripy4gFw#HV=WM+YSY!!wT-lg5L1S
zD7Q!4$zew?HLn_NFrgq%)s7_ax7pkNytjmyy_0;QRJl|z+q*e2wdfh~z5G+rub8G{
zOFShd?7Bta^enIs5`dZ0>_eY(?8uS!QMKwgQpgh@SI?9JRO&<7C$UWTu*vu&pPCY9
zK211ga=7U)9UqIJwe`yWJ(~1`kK{h8Q`oV1sVK6A>ig|@JoA`(I6QWt6Atv9QVUPU
z?-Kl@CYqf}LY<C+D!Wwv9qQ1TWFL6H+U;x{Pf12E*ts09m=-*;=e#4P=GKXJp;~a|
zS>q~EY2KgWLj};t-uf*1(9by$;(n9;;$TxU3Y(K}Dn1)hRqb@SfRPm`ZETmEj*I^;
zav+=R($mGR@jq^{%TBvAq7ZELFsZCscDWA#DQ?C?d&TLDAW`sUX}sQTSLUd*>hG^_
zwGsIz#cbG~t4>!AKt(U3aZ<)0rc2W?!Dd0i*Z6dIDru?Ru8sHnBJngD=`XwNy3-f=
zOX7a^n&{l8c6|at@kwl|H~6qB5<(cjjs6!7rutQ#c9UbF?vFO|bP@TQx>wt%({V}C
zHKV0?@^1EzATVSC8||-lG&I;P{?D3QYi!KvA_sLKpS0Q71fB{to|5{4ZV)t=aS7aN
z+-ew-P8X?D1;}hdV5>v}FxMgY@<g9t0^x3x;%6~9c|d5TVUw%9^IC;Ufc41CDe;T0
zmQ6(m(4{1K|8@s-HF9sa-I1GGJeMU~akbUvOo%zs+qV;!HQk?4@B-4J6pz%NnZ6bo
zdgUIT<pbAI;j+zU#~7Mly52zw#?3h$V!49K8tZLtg(anJ^<&?X@lf{oy4d0rjXZXs
z>Um-Fau>O<+H670rEj-!z-0_xm}3Kdx$qER1j~!IC^r*>%7EPy-%6O<!dlV1Y;inu
zlgsm{LX?TheN%D?!Abd9sUo18d$%>jGs&M)t}x<RV?1*z#uUW8&>R2|M@2?^OT3}@
zAz8(Vmb2yY3<s-j0`6(rImd!Z5*Zt+is#}ZyR5x>n`)yxqcE>^CJlEI@PdOEG3a!;
z%80|}By^Rc7(Iz)CY5DIN*6dPikt%5Ru22UAI?B4HkP$3F5%Z~RZddhNVNtP3ti)r
z4EFt*cpoHPETaz$L5Gs$?RnQGGTS&rt8HB@tprbr@TIjDI#}<=V;=GA_F#^!#=>2#
z%}B)$Rh(cb@}^n_o{fk7^ST++7T6<!^*#btciN*#if#4O&g!-eiCu_7&NZxJLI~Vu
zW56$&UI_p)&pqbPl;&BoZ#r;YKJF*A^}Y^!B54nH#Yt7NeXQrn_{DP~z%>f8uk<O$
zC7aWzI9qHcPdo6lm9z0o00H((=`KY)^~K_K!B7$zBEih?Ruyt*n{tVFq1qwPZO+}g
zOL-#-dDxPtsE!W3)*lGs(uGS5_+O+W$S<`$<}Y4dkf2kt!{KAEXV@-3`ZG(Vlidj{
zrvilzacYb0O^gj0tZiSshf#?8UFlaeY0gb1V~6;F;=uE<X_tyHf6)J!R<GRNLotBv
zCwcM|0Q^Q^MQ2?Ou2*HZ!$-5Z#F=fj!#==_k{LVbOPGofWJf9@q$->!fb->B^R|C7
zpHf=bnj9}MfgW)BZiTa*@QC-S-E{}o5I26`-=o}_YO;Kg3qSm9F`=nH{fE_BByhN)
z<Rjlf42Apo|Eu<Kwc{#MBcVkyPgR-9K8Z25fOwTdPz=VWrz13l4KKT1a!@{VVy$Is
zkNOq(mr-`ifgJqb^*=ja4Mx`)nDaL5Gbc`mvPe`Nl>)o%WYU;#tDW+3??nAS9l$2?
zRmz<EyYK+9_%js}b}J|VTf-hZTOFJsS1Pw5lKi>&1@m9I{FG+ryK_FtnoPmQUfBjN
z#K;Jzpcl;^o6}E2WKr0gY^QRh>r%i!Wj{M5a86q77vJyt7V7`oug<DqIb`v7n|<rT
zQn5XOd==;@{N674w%p{n1*pv)Zxsmjd)R4SvCF;<YZdEIG#KKtT^`TG6a>}ZkVUVk
z-g}O#(dC{itM^{fMag-2-A4E`(t@ZQsM`PVva1}D!M;D%+tvB+1vD3o?_HKs-6L8I
ze(SQ7u%J*{GVb`V*Laed`?~mD*W*Be1Hvkn2Kn08+w~PpYDpqf<BN7f&bhGFZVXKL
zaU|)W^tn_k`J3WP98M6tN8XK$w{;g%e+p(MTZ*hTqrMHw>n0sk7x2yB7U7ngL9zIw
zzYUSCySHm4%(KyM`PRdqO2YJ*m`iy+3aV1Q$+5YUCBzYXJMe3#`fc2|A*R%NS?d~~
zYOt>XDzFJTRY|U=?3GQdMn|BCy_$5<TdIaIxmsJqS0k2bUw7J+Z{vt628s?Dl;uO3
zx;<xIu^Msej&F;2I&IzztE~X{^7)=1qa{uEPo^syiCFS#Ym2!)?szxdZ$|K#m0%aO
zx+K-u?3@&3qTA+td!&@E9bQmU_l!H1RV+$oFxfP>*j(Rqm3Fea+${2!NN{SidAZlc
zE>%<+TaeRpeW{eAsCi+vv>oh|&SxeQF3Mk^L)@c9`*(5no^L}fsdCSezZWM2Xo-yk
z&)br3Ly-r-sHk{v;Afb|!+}O<V_PKsGz5TRBPF1cquTgwC{fcWV70=Ub6rJCbZ>i0
z3|r7rkh3jLj5GD+2^vv?Tp``omSd8wPndUmj47k1xB;F6>#T646d816wy!InmD5<9
zjMkG_f@gYc1>(D`gSOI-0}+S!5oHyxu@S}Mx<f`tQGu;V;Qn+UjfBxj9*9|_RJ_1l
zu2jUeIWt}^RTRRiV(Xl6;K|6Wwm#qBB2tG3#Tq<VaS{=13X?-Y-w%CT6w-gHvxog(
z|NPJPh*N!2dxt$5Ln7fYWGU`o8)6-x3fV_efb<9TGu{|)r<GAos>gh=TNY2T$DLS-
zaoKK9RIqws=>hDSPx|lFUtMKSCFOKWfh-jWo{q&zq1|H7B)hn)<StmBgTMB-w#kPf
zmi9K=9N)-kzSXwmp3fm9s>}NG&r%7@1_C6Pwq)DbYTNyL(k-aUYdd_7rZu+9KP9)j
z2zk4c@01R}vYzct#Bp$SXR6Y|zNCiw_Eok&Ke?}Ibwh_*9vuk0N(IfV7(Z+WV?$^)
zf5^wc5(u|_BM{cX6?wQqXz6n+9nuE<cM@rc5|7_qlhDPo$KI}<sc46cF0a@-iIf)V
z9WJ$Z9Rb8T))O1`ULw6fOlhbXz?~`5ptsluKB>eW`_O-v#7;^-)up&6KMJ7K1m|O4
zSZ{ZOeNqi2Bf6Jc`LsGQuNJngZVH77*=G*O49p!x_&AzSD=dG79rHN``zQ!|JYlB5
zQb$*VorsBK3+C=Ud;mG+WIiiyLAU#!%I{`%l&0RNeF`tXF~-hRpQwWM)<OJ+b~c|C
zTY*c0=i-^XuaX<3o}SP1PO5HkS6xUlB^+j({S=?bJCX*BpQ~Zi{>G)VUy_9C8x@^~
zVEk(iypS^B?e#0n&bXjaLLOb5<YOh%x)^uKnJ`x2!lcsE85fhp)+AO?bHy%;XDWdu
z4j@~`<!8z*-^P{SWmlXza<wY07af`@T#lDgaOIhh9@=}U#)Rc!gwNK9p^VdXReUG`
zOa^nAnH{RDt6!v?4|5{fHPtVbKY=e*DDB!a<*L4qV{u))<ceOAJ0Pe{FpcZ0#i%b`
zKiMKL*$we*n`Ggl(Cx7st7)jg2+0k#sko{7MF6-==BA&K4#4<H<89QLkc+TgdpFOU
zV;=wsD)8vF(o&7xP$=H$0K$76N>n6XssdJ+afjfTGjS6x<DTfYv1ih>(?eTrTHfR*
z<#sB8x>j?hNXWE6>9z@H;y!Ncplw#`8jv;dOt}DN5nrB5Lsi#p>}4gT22A#))ztjX
zro;^Kv;2T|yKgHppRj*N^+xF@QS2hMNeT5!J)>y4&-&j++svG`R7CUWKc|9OF_6=T
zS2UV^CejW&$=nUm1EqaV45B^}A{dHFvj9obb7K&$sZuxmby!?oLQ_|dMFTy5UZ6<f
z7z=!@Q|HXLg?Tup{BOVgtu1np+uQ7(7}E<J$We^(DmJ);SqXX2O_9|(GROk9B)-up
zQJ}PvA@B9^p-qmYHN+c)-W2gb_N2yKuHP&8cT2si=3GGiN(`Uiyrmk6^JmKw?uITZ
z5w%t0Vr>T|$=dA|(%s9X$U<kVo$(9NgoWjb2(+33>&g}2fs{0)eEyzTHO7|8;ELf=
zc7<X{S{dJnT3+C6sk8j5Y8?9Aad=gubd4juzL$2)4m{;#$gi@sxzalk(iQi;F1|*G
zgY8O_+h^;2XQnQ)2lLk=@xp+ifQJG%ipC<AJk}>2K+5{x5W*bRN2+h1ML$(E$*cBg
zJkv>%cMI!;ZSZGG+#?~&4R9OdJ*~538&Rj=F=vuXSHtA(@jSO`Bi3$D#QTyd2f*b4
z_|TKpGu6ZNeR|42tM6G_3SK;2ZJUy(O5_7eP(Ic(zDQ1m_-tPcrKm!+jo3`|!zQQk
zQbuL7BdlkIZAnsTZUx)@F}_?s5GibLcK|XFNK_G48qmaF5Gtg-qgq>u`q))bkm7;4
z7PU{<YyaJL=L%h$Km*-sdt>KUblAS+nv}t!tRrN-KcTx{C-Z<ayV99mF%<`6IG1R2
zSw3fne7H%ICfOVDv-V^=%TSE^;kdX+LvOOdY8AAywj)W(B3zL?gkSn~fMKGr?d_z*
z7OZwc@;mupeot>{x%RG4NXUUpC3@*SpK!2G5uESmAIqEWw+|}tWM!95@QQtyFWgD3
zn~wsIian$PF{e(HDlz+GM|cUJdU1?>;_Fb*H#2@3<H~=4+g7G*`^>R2uIBf4G=6q#
z1eucT!LeAnTOgRp&0`1&e%&9hKov(X4l6qm&+>AH?(JlfkH=|n2NfL*1@2U!8dAh^
ze%j$DPQ4WHIFr*4eE>d6tDM3_&c;*(Toz_@2*!9Wo;{B|qT(7aqadA+XG+B4P?cS~
zkRTQ5LYf2T0VEPaRsEvaI5F`wMEo4%l&CmVrhP2nms|kz9kgE)BH&|*kTU^O-^Eq@
z2l_%_c(~-dkgS&0L#vm5=Ry)}hifSHBjhl<>^qmaygQ{g^1M99LX)RXiCvLnwJ06^
z${5RC?jE!*6U$~B@m)v<>UP#+SH;lmsItlnGOzwF;)a$XCSQ{;sGUYx$4<L8pOsB4
z_HDZ^W|SyJv<LeTxes>zcSn?LhvlQ&Zip}W28z~wO&obV^LS!pJPzSaxf(P8lJMdU
z?~qdhf^1a2(b++4%gx_aGAvNaZlI>fRjWc9?J$sh)^4}>m#n29%kUUqtP1y#j8`GG
z5yx4hjZ4sqx&>o{tZEWF;|YA95HINB$o#2^`H@s|SEp@KJae9*OYU`>9M2qD=h0If
zHnP`MPw)2Jt8O|VG}#^12a@@vyKl1Egt)V-o=UPdJrE(mTsNm;`{lOr>HvFeW+IAh
zqO!1b&LC{>VLmHYcPlvnvmI-gD_d+%MNwX=rj#US;9WNNyCPNJJ!iVr`9IeXT|Y0d
zETrwT1u-pIi2{hqpq_`O<OsQ*7y5@a)H+*K5t6D?sBHKSAe2gRPplsj<2b7S*iKtq
zA=bl;#le@vGbvn(hE-?Cdxzmp7Z$AB8j^&@|9;{mYm7ILj>Of<h1mREXeVA^fIiC=
z`ebx15fz2jF~G?GW6LXWQZn9aZGj2#@0YCI5zJ+UkGC^k--3*xaL@yU(sxzsSN%Y*
z{v=lSob|-ps$C&vNby}OeT$a%bhX*43OKeUfGc*)kV@9XL6k@Pf!yad<f*mQlpg&%
zB+#wP*XB@{YW;Ub4fyi{doUp(`$SYmV?cXo*dXd_qXX`R(;oJDXDuLJ`H@_)?5q6#
zI<n}axev9yjg7V;-tYj!QYgbVR?l`IpDM=m-|aC!9OZ`FGuj@{fr)Qd-rI}3@`O)Z
z-_S~cd9I`utLIbkvT&%}@>ejhJ^fuMWGJkyh&qk-Or9<2rdTEVe29AKRkkTXf;Gs(
zg`><iCk)WE^!<IDjxF)bPrmYUsR~t1Pn8`=O?H@pH~|=>wkOreo7&O2+IIN+NDcCh
zDIu@Pb_G@GUVFiIe-}!dIuugEpBn9Za}roF+@di5eX&O+lrsbv59{1t?cPopw@hKO
zmn#SDfKQ}cUPOn3F{)p*oJ*9@LwUFhmz3Q>0r}gDX;$U6-iWb13J+%>$ng_>_`9$W
zm-zH@*;BdTyCb(0V?j|+!D1Q61x1A_yUL`k+Y6xkI2cEA_3GUq3sR%K9dK$TL}eBq
z);kqOc5z7*ACtWsuxoonkIMb@Y&LJm@LsZDmB=Kz>boPgf~|x+@n28a<8TU@Vm5MM
zh^-dI!>ig2HBRK6;^LAG`va%VCOT}`hZXXb(xsR!RgBhU^HG2|TzS&Ah@C6;<Kx^8
z<O?~8)DGg4YMC!rj?z#4S*d<Z>~m`aZj8_3z3hlCDm<FYZEjUBm1EUV+EWi8J066`
zUy97iQ`3wzQ6h^@`oaeLMo+O*@vf{WFWc#yFic;hb=jFbmg6VhW@nQeXuT}8sf4ne
z%QvR?QlB%A1_hByt=9{II%)DR!sqCv!_MwR)ceV~)$;r7=NzGyN`k-S9F47HFUTT>
zjq%qCGKqK|>1Jn-NNnk7Z=?X#Sr;!b_@Z|!f>CBJ1-*D3p-ayO@1T+%HEHohU3S(*
zHm57W^^#qFHpIVD=%?MT@OQK+-U%5i#g*|L<3jdsWgKxfZXpE$;U%%Fs=-zHu=on>
zY9AaUDiIkRfNRdWG^u1WRTrli0&hyOzxHfMo#>e@c3r&ZA=PP`(0kXPE$qITatZmx
zpNJ>28v_|)DY}|HQ%*tJ67;v(O#xjgs8N1@!A9oNaRrxH_gp4)i&R;HXm5^Zo?+*~
zucOaq>lC$7=;&0!3Hx=6zp6Bp2>zteb1X#SC^>c<LA3EA&yLNvX$?SR7ky~xOA~LG
zHvk;(NZ9pi0t~O7>UYhA3Q6VW<!xxKj6Lye5q^F%>bExOZ1CVrwjmb2S`zFERB+4N
zl#6C&OOVZqfNr;+4bi-gsy#Rs<6|*Yk(=rx)m8&4EnX~)(02!c@D!WrG)tgclg;wc
zAjoozX8RtJ`andR&B^_k)k^s`s-s`DXc-$Yfh9OJZ*J|fc}ay++k42-Uu_F=4E5LK
zMp}3_3r#`pg+F;k8@MzkK$V+Pr<g@aOEZ*ZevkhJgI)(}7W+!-SFN@sLC#Qox5^LP
zo5(0i8dm`7R70LROsIIo7_KpKTLNC6XP=Z$ty`=)CL&V?sj=m3+&fT2OGl$uvRQsM
z)SmgI-?!z7FNScJ6WoX(B7e==W2Vxwn9V6^v4tW!eGlsHp1;7l5^A<knle?9ttZh&
z?P{%TWx~=zOk;^_<n*peEG-~2-`1QR%6cp*#*y{-fKSyRv+i2|_nZF{9Itacw!-_?
z`{TjB+B$nM7dV{;+Yk949R!2jZx1IXDS0ew6=kMkGsO#EEYgGI0YyusPJuz*9<5k`
z`{-pa3!jOsthOO%tfw5;#(3F>5i%quc+iNFj~)wT4YADE?eTb4%J%p%wI`AgJC)xD
ztWU;X_*p7WIR=(p@2U8usPmAa@<o83{d+ozdj;1zMX9P6>0VoydTRQ8Lt7{TL4;+S
zatr^|PE-fA8S-zUer1!|lFJ@{>+h|<TEcc_cMIx3l08DqF56x$i2hkk4IQ>4$D3Bl
zw%ab>#GWShdAINF7_7~E{W~natwe27h+to=0(OQ(GOQkl?GGSyb;-7%+>Qf2s>0Xp
zVC)U@6oxRjbojrE5y%czSU^hPcU8fDBS^&eOBRkDoN?9QhkulmG6fSI4wy<QU>!z<
z6qJK9=wlKD%@w>FeHX@lB*}H&-S&2FQEf*DsdMinQV7Ibg@^y$07(~|44tZ7|DNMj
zxn)?j-;b%?ZB%UiG4%)Wt<#0Xvd#9PPd8%~k}eqes3PrQt>fXh!P)F%2ZvsvEnLct
z_DP=DH#ZRP@#$HQ&+z+H)cZPXm-OdNg-cLPC?22Xy+N!d*P8<Cj^<hk+v_3%?pO}K
zR4JRstIcpdkYF}DaW*1Vi2rJ)`h(k?Pv)Cp5TOBroyxUynx%ddJz%E;Q#wI9l(Klc
zLE=olO{guB5ib6}+1YAKU&SY&<Ltbl-_AJ>mD#Pp%k#-rw6d-quH6fMGL(_sZ9i4e
z$ri!MKzR5$kwIZ(&&jf1s@J5}deFqL6}=86)Vs$_4HP$<eIIu*B|`|Wli%}_?_D5r
za@CE$MF#<@xio%pq24*VljvpNm)m_llzDlMNG5g^(YhkO^Rc#+x|06gmHBc#fg{Wh
zeT?`%uAEZ0zC>zcefvHnLOK=JL*rNbe67nm?3(Y7Nc<!$WT9RAeUaGM618N%&VQj^
z!FIdeKd6~7(Qf#@NQqSG;L1=W!i@<YLaXF7iM9^jO+Lc?t@pLsNFQ4Tzff^UeeYr`
zDt*zeXjx=8e;>jvMO8|coQ=-qsdgYa3o@$3hq@ir9*gJ_8{<P0hN^P*V*?z|rE=BD
z!5LQ#CxN_2LBBQMm%CWLZz?z>me719G)xS1$hWr7CMBY#_1NTUCx<F_2%-Zv#Rnw0
z05R$I9B@8fX5yFcs6bORs!S%zZ&%b!Z7j<#cpZ-&Ha(ve_mI-?sE+O+p&?WB%4X&G
z6eHjO+w2Of3a6Euag)u-IpLdFkC1<4F_7tgZoUoRTY-k`607^(6_AwAv*h+m{4Vik
zUXJF5`C;z01>dJoi5n9kEe4kaRqk@`!bJ{z3-NnIIo%UO5fLaGKN|5aCc!^qi@y&Z
zBmRmPNymzoHnAJeqvQR}?ydep0=I^kO^I-X`4&S*qtmJSn)<<PjtRX=kBmv~+9Cko
z2b)!@d)o3G8msY6YfCHvpaLT?30k|8g6fl5XRe!2!KP;Giq{ofA;UlBJL~bobKAJz
z+sf)^rKri9ywg_YE-q+k5;t2@EvAHjNtmz);#mZ4va_$)+I*&BZ;T~-_PUFnNkw_x
z*88(T6^?0CWI5)Q2fshkP)_||-~Ye=vxgEtUOnw$KOZxv{%H;Y`2nqhMrFY!rQ06O
zXWV5{mDB-fLq3z!la=m-t~Y+~T7c-XSVf9Es6=mg0KW{=_E-{GsaW36GdyWUDtJ7x
zF1tQhe<I;rh!?xVo{TB|PFDu-EB2J{W5a49^J82$aQThk<31A~ij-wMRcuqA)62A3
zQ)z|c)HT?L4Ax_tt7)+1K)76$w#5mTHj-WS4GpmL{+LWMvS@0*q+e?NY_fs)+$~AY
zO18(_vgOKDQ{s6`1mKQ5SJLnAZ)s{Hlx&&pN`SZt@3Y<c7oE<C91C{f_NtI;OSGok
zMtf~vTno3}YWov^l<Y%xIFRV20E_%T2V=XW;piC2U!VvHnKbF8g2+u>e>;?4Ev1Mr
z_K-J{V=qvtK5h=GaOEp2sL4fpjKKiWO(wn{rIIg~!yydUM=A`L3XpKtP4>3$2X$hX
z+B?<nUejzV_S|=?F^7EH9v7-B_&IwoP^|N%tk(l2=|;Zq_mlnq%U}L#ALK^LdZ2uc
z4;}W7r7f5TnENA#1CMO2eH>$!YXsEnfG1;JCOq+zfVPc{-%oRUrmGdkXYoxj`|u6L
zErPm_CLC0XwPU$8%7$BQ$Mc?TY*IB}2<(K<peE3EGH1Yhf1jNS?6{=w2CvgY1O|nR
z!4HIC;f#;dpk$-7@s=MTtxDd>UEE>k5-sQw5J#h(&q+j6q6!$^1%JjNpgbm~oF?A;
zDYqUyM0w&r`*eR$pO;^fQp?4ptChH6S;IIf^a8VA9U+7o;m)>mA+;gmFLyK6CFfiq
z!NCTgu8drIE<`C9ClYgO(S0t97ljdJi>X1)<>y@F+cI=hGsLUPm^^nx($)A|$$P)@
zT-=adt;^_SGoqT2qL{i~apqo?gL5WS;F)|QyZT&+MNkdKQsk6te4x2h71aU0Ha|jZ
zP@I7H-gW1~ly|@Kalqa6=Zb{6qP5;`@Xt%>5`MWcfk!BKfpe&uQF2)CwVQGfq$6n`
zHqvK8xWP}@(Tf>@R7qdq%w8->H=hgdXsN}cokxx))=(kHI|xG=9gr5T$TPaf*)8!*
zWerrV5(AlyaR~1!MiU~-hK4PAbxDx*vV(<Tj>{oDRFGq^)oK!U6_->tz=RwGwx}$r
ze+TIj#1+nfQNboAZ18QFW@V+G6sRF^MD8bk+01&R@J~KhB;{x=I>rAQ?4v6{yWO5-
z14HfVZS5(=YwpOQq69%DH(G5@4E3uE?p~mu?gIiKx#2c52lQ?r>fzLFv*H!07Kd!M
z|0RbK!q=R0Dem_+V0N>)v3A))i1<T#@K~A`i@<fz>du8lYBoaUJg1buj{1!)@KOJ@
zw@H`5!hCB!a`>X8%0_i^De~O8V#Xcw&$c+$S}sIZq#)X&u-g(J5K%<h=e_Y>KPR4~
zd#*}V)7=Al8+@vcUM?zYOw?(gfY^;cctot(r~LDs){?N@g+9K#!YC~yf+4oWGkJ5B
zS&z(V?M}Lbea*er87T5lf7eXu*q|gUX{pwgke6tNp8RV%R!>`*w1$<BE2Z03#bn6Y
zD)i4S@uICsp2O0qvZD1rkT@ilx`)cg-L^Jgnz4#@k|hUlU9M~v0{;5^Lgwf$vF-;0
zIj(K`ZSR=?4+Rj2RdRyk5GNZlSx`fLBWL%K3eHzijGgoz^;bZAX;8JH8f71(<08(+
zc*cP!;uy4SkNLB~zV;?M6uOz{@qDi|FXhgEYrQ?;%c1PU^ONn#L}RVE$5R)PO&HlY
zhkNYl_{GgAAqL6)XV3U>zxn&oqkn6Cxep%H+i07Ta$sk&5la;*a+z)RnK3l1BNTUw
zpO=zl_aB%p`g5StZx9wckSDOQX(`53*C)1Ds43<?SkXS)Q7yUn$s~!|u6X8!1SG2M
zvE8{5Gn8O%d#in=^>Nn<_H|!w`)sOg?9ZP&Iy-Ty<Muh=KmX&%v19+<4*FpCwRyhU
zA)l|v`*<mfQR0n&L0Bzb8elkFk&c{Mm?=t@v;rUo;}?$&a*e)3&DfD#3$m4*QYxD8
zb~UynUc@`bekUPDhE>u%bxs}Q4Xs4HthV>6xji6mC=%fPYR|kHqk=+Ru>HVCXr!<J
zT;aoNRyB`vYt2XTtVHevJ%8-%>{gVoPpY52dk~Qv_GvYYoSm{U?Xu4*_PH7_&?-Bc
zdodMWdaS~Yu4t!HKNiQU;a#sM#@<fE+X}fLpkSD@=wvV)6jsh&(XPQ=X^|&?w4KV)
z6vM+sAbxi`o|RNdp3(fn&Nx;mW7>4*bURyt-c+inp37U0q#Vk7qDv}zTUUI!=<R&|
z&V49fucEt$*Hpd`GwN+A9qRb>Q=X2y=$uOXD*L&5NyeNKt!uv|sPG-N?Unuc*LYUK
z;drL(L;E4*t|GjsBE&ws<cE-W?xMy7c8ylMG~bfpf-hN&UG_to-g=<&;(NOMhmd+^
zco7@B;)l2hWj@C?cjXT*sKEp<k-FMe^duV*Gsq>SXnY0nT@}wVK2Z_3SN{-K<*3nr
zXV>Hlg9i)$lvncFoN|uaWUl)mEGp$Pk;k4uD8(XysQqw<>wh>R39%?&Hw1XaBwFZK
zyFrYZ-B<xWlzrk<cT@br-WG${%Ozk^5qKjV)IY26wT(&)D(92+5wUDH=bHJe=w#f9
zqjP<;(5-HX7l=ufHNO<K8x!!VTh-Vc05hC6nTnw{?uU?a+d9-<sV31uCFYfsu%1wT
za<B+fd&v2n=ySIAD0O#IyilrJ`1!KQKNR7$=2l|<bBigk%M2t*`UDiIcdmR#E?^$}
zT^oaO2jUkiN|jBI4JnM62^8EZ@-D3np~#tmP?ug*8%+3;5oVSzeJVk7Hrr{G0<*I0
z$?iKR>9MP+lcG>EJkPE6Q!2Z!$ak$SfXJQRg#~9AC$M?GDjWoo9=mP94@KTwQQt|b
zfh|lNK}Q;wx{0<Zp2>128^<fq;ysCca{XJ0id^h?Y$d>XN&a+qgL;S9z45v7Zp8>_
zH*O8_4Cab?=6Rf}+pICqCOuSCEOfOs$5Q0?kQq^KddVTUjiuRJON9hiuoc$lNLZdQ
zEbm=Ghb3!s(rKlUw&cxrUo(R-y>;fx^=%Ekgp71m!-{9=R+sKe^8@SgZNhloW-D_E
zC0+*W0E}Z*epxgD=WKkB-oNGt_ohhBATox`mvSXQ05U0}FNptfB8i)^2QI?52QRO6
z{N8=n3|kkI=tgnyvLyKWT*yjqD*m7^6QgJsi+;%e!rt9N;O@gQ38%I2J)+?3kqX~_
zX{w&xaF+dO%$$y}{YZa8&q3ReN8{Fs<7{IL<9;HKFf3T$eo!FeV==XkPD#7KlK8j-
zhh#B*?o@jsMsTStGN+12DD%!Xd(yX4MqH&5KNYXbNx#D(;tEsCYCm01SKFd0P<|nn
zq3xN(n7Ryc_WFDsGU8JjEI!);cLK!(V9DquTg+yMR{g56wQO;!09C}rC5mJTWGXM#
z@-kP{Kt(Si-~xng_vxpWmN(e!c2t0t*i`v}pbQLJwkvsG4^3LAWA6Z{Y9<7;Hvuhr
zcJa}(ekFG9OK=!O*#QT@k1D;=f%=EAK#=elZMD1YfKypVW4#?rZb+*+CB@sJ7^=kT
zFuVuqD}eWnd=;BVvkEsHPTDF+@H=%dpK+EYba0mKNVQy5doAH&95lt-y`8)OM_|)Z
z)iQb~pK$l#hl4A=n<PUD?09=G#+Rz0P<avB`#F9QK8D+DKFDuzpETQt@vUfCx7AAo
zIDS--po@tl<4OQ6h8bDjKlV+BU|a2zc$Xj`en%O2OL3z^!F`_w9O2#rbuX|^*%Cg>
za{~$3(Evj%ZVxD;?2H}rX)sT=+wlsd0l0@36)0{1rbKhK6ETgGJ}5@M)slAENuLI+
zL9VBUtyX4awPZS-*llg6D83AeXPgvPtnRk6@qvOzlrO1d`g2J>_|MXjV&~&UT@e^u
zSjG;!P%Wwi)fW@TPkHJjJ0WoY{6mD@#Gx6IpbG62vGNuB#gD*zzR~vo>RVU}B~Uu!
zybF7hY4)&cdPLbJ@yu&Kc_8nl{%o+Xqe135R!UEx^X{AMLtcK}E<f*rTm*p**=OvE
z7;lJ7Nd7$UGMID|l}1$ySzZ6Sb5$g8M0`^$K?SaTcGdYJ=+W(+^qi~DhaB0})T~wz
z*90)iVc%!h=I3(fh$N7$u8U{7oJB`aon3!EL?b$J%MN!wUFn5G74fCpZp@dKl6KZ^
zH^oenOLyDI7<CKhQ(iSmPovHk;i#u$mED|JfvqT^4BrE*kxDb#7tme&kX*I5oDWHP
zI<`nfx{b*-$Ws8k5+B9}3Wyq{TTr|pDG(7{c^6g_^9`@s<mQqI@r&OLC3==Y6{j&V
z_oBqxdX@ANUj}QDq&w++mNa#*<WRV*Y;t>Il*#AQC`##ZdJ^~8l=DT7p4ZD&bGxIY
z=JylqjtXo-*kIFK5>{J*Kp|-l3Jh`pqZn<c2hP>WaIeku!B%vvv{{aj)@Cf{HakX6
zj!36Vo0DH6FTQMZlWKa~DG}afb-A_Nhe*2fVgQdhLcm0~THv2xiP**e_u9hTr)7L1
zcg3PWtjiTehXQt7?siWFPiVJT4i`J@i~nV`Ejb_RQXv7A!n1qN7mMBec54_0isD*T
zp1(2H(vNVz#hR;mRRc>lz0&`I$l8|p){O<)%Kf(7pJ6DLj}+?`{@ym<pt*mVGsoKF
z8&z}N{3g{un6EQ-#m6r3ONB8?ZqoX?Di{Z3)kmP)Z9NVnRW31d+sgcFX&Z?ls{&%(
zC>ZKdtusb%O^{foIVqG{t__$%X>3!R@b*BpE*c5NU9~oz!Ca&cbCYbcbvaMmqw;CF
zNqN1m7=vH8J?Nx^REMSOp%{kir%*JdjO^k6G>ocqi|xN)kL26XzY9hrgZglTeGR?M
zwjnU*0j*9G!zIFv@w@I!VFI&zU{>oGu*a&E(*LUbvfUmJaAG%<7gNs29(y8xkxm4N
zO6_ww=1Hfgy1Dk0|EqD0JzW9+GIL8ve9@lCb3r>G#2V|1cSHxzl{|~xwkc;Rl;+e>
zLWga091&GP=Kr=OzE-5a{D?A&^gDX~`d5|R9Z1Tj@EzEflDXaCoBUhb5nsz>q{_0d
z;c(mK`!<t^6Wi^-5B4bp1>PbdYi~6pGzdNqqATr7p7K^zeod^mY@&F|{^}dm-F2pX
zAlMYCkOvb+q-Ba07f~+HA2P`E_J+eWX7QNCb~wh;6<34@wAR7IDOH{85eJS`^XZz?
zseInvjuF+toZ=dRVL^!is|*YNOR)uex1yKQXbaPjTHTap$lzzUy%%Fk$WXsb1qZ*M
zmzn0LO0KJ|AH*D`mqDSdjA#BZd0|(_YObG;V#rLn+J#{FarMlzg;e<S6Gy{f-^eNU
zY4!U~>{*_z_nB{eLn-QeG+t0L82f~zv)_*Sv&fikrQi;6Za!!$0rhrb2(G0BN}Y6+
zq3Eht#HrlmVyANnE+r7z>`o^G`eRd5r=6)5Uvf$~%gWqSvyHPp31LPwwQjd_LpuSd
z@xt#~&UQPWOC&f)y+#SpwhOt_&2lag-4?s545l6Svs3n{f3RN?RZahDzs3sNSGSq{
z7@=1P2CslNN;=vlKc=3c+K@y&>X2L-FX|4LMM2d@FZ<CYDfc9~DD4`w{`;(59^a@4
zvJ(Da4|c_mML?-vx!SJ$v0TzrC~mhAF_f~h)g3`*i(w8byDH&C{YTzEOaQ-K{bLcz
zxV#|oYvKj!R3NHLI-r=DPp}5v(5_4P`2`PT(Fw2ru}G1Febc0p-jIJ`q$Z9n-j{z&
zRu%Oax#`Cc+1}i6_}~f~nRrsE^w#hLjQTNdR#=4YX!hjh1PxPx3t|H{`o|Et6i(dS
zOL9TNSF(7-mUr41e+GQT<y=&#u|HNuOGO5wK5Iy2M2`D0q@Z5RLX`MiVKq6iTZS;k
z?6L`oXrlVE4byPeG|`bdgR0IH-iQG_!KhT|waM{J79P1lRU~9ePDGfFoQbbcq2TtI
z$c+guOb<UEcT|ALWHa3U=AvP3BD<OVi8h<=dpX#*pw(t3j*B&+U8=>b+`Q!-_tEIl
ziLB<f8k_TD5!S(=nq-o5@M95k{>ML9osT$f94!>)<z$>eqI?(lPj&Td%)&spf)gdX
zD>2NCK8t)#dUcbiXZHjST#BJ!5#sA&pPX>YR@}~Aw!|l|oon~{w`|?>h|UewrpkTc
zoSd29q>TXyZVqeqtr_gQ`%eq3CDBK|ZyHGavR78{E?XYpz?Wd?xH!4m+v1sgW~h-R
zT(aFKpvXd#b>{ETpe}fI@#{)lH`AaF4#{gh`C1Xd{q|bv!_2CuC&{YB7#vD9BxDXo
zw3pk{_vN-*C)fjtnz|0Tux+jHK-%ip<@uoWej8N`Y<(_%I^CfjtQN1zA+Ed$YgrNG
zp@6W2s5oue!|_avnj5U-<~{ObC|&p|DT*3O*Q3?f3RCfvxD7e|tlkFO=!;!2gJg@x
zs+V`lsH|`(d%W7t(y5#)J53Z%_?K{tCVMizlv1dZti;-<;+YmBL$pNir{kIY1X9MN
zH$4;Nql@@i>vObDxNV|sO1$;7zygR2*qng3!4S5@XAFfGyMJi1V((T+=YS72b=oxB
z{$pf85oOZbOne%9kL<c5&er(f+peSy7~L{bi%}(a=L|?b1GYCeS^etaC;NQ)Z8DAQ
z_kW}NwA+CM^6zh-Vh3{xxV%*f$PVQ+WUzI2w&*r`BS6>ns9Lw%P^b?(1tCUuO2ZiR
zvo`tf?1;le-j)I~-u8KaGiKB%dnce(9Ke9R8-u!ogQBTO2=C<zNV;f+v%Fu?wp#un
z=@v_@f<Qj-H7r#)oqg!A;Q3bfGW#e--b$?}X)ChxeVnu>bBt0jN@U0<Ip19bLjmol
z7m?C%r@&{vjyL}&si;T&-wOJ+W4YR8D&2PcqCJ+4px;jT9P@DsB3dHWII#EiHc*eC
z*G?s{+}iSx%T9GV@0N0_Va~BLxhbRu$|{2-d^SETfu_0{pr>>3Ol;_H5INyLcHZaX
zax8@{FT`Qi!Nl2CLV4sYQ0k{>{ihh>KTr<UCQCl~pM7&@-EF`4(y1iZXutZvx-WYv
zcxe~hNCbz%6?MslG&jjTC;9IcyYxb&I1N=H*1zmB|B_e`)n&Up#*y5*&#s7ZIHHnN
z;L<PImGOo;A`RGx3w}k*N}~icv>fK!iFM|x_*Og-Ih?5Ae!KcYSd-*Q!bf(^1sAH+
ziQpy1SG6O*9m_3j*XHP4{nA9GcZmo0+jTLzWN<MY)i1jKLU6doE-KNMbQeNBZn#h`
z!QZU48!v<hbAD5s-E_ePG990hF^xFUt^lC8$81yre3z;h6X#XE2?HsJm0d=;Rik}8
z<nzC?TLQQ$!4YmZ=0cGLr_+sKtiyr4+0ks{swJxTp$i`pH1=6dwM6BYI(<xtx4rj-
zl*7#RnHX<-tVeNy*ra&dyBm}+lo)4nj!lYjmWi5jp-6mmyeKZ^+v61_IZ1~YK-an>
zpB05%($4c%8_$a2mG@LwetKfMX2Lj|>6lqsPw$Lb75-#I_1<{Cve`)rh+2E#a)&Up
zpuQj?c+CEsoI1*C=(Nm@XOh5`&wT*7wa%ZB$-io;{HhX|=EZy5SBzJbZCj9dFH0Go
zEV0n>fTGkQWoA(|@$N!t{MYUYU`s`Gz}&%k_aS6`*lUZEY{c(q5V$1I*dN5F?u`*u
zg;>6Z-7xfqd}E$Up*1E5t5&oj&o?{mbRhb-<j6!JDj06L|5P(!g0(r)Y2ksBu-%_`
zbu?IKQf7OnbybT|S%%_}vYvQ`3W95*{MT0IG<3G;VwzPK!Zul=DMWYp9RG8<F<X;3
zEdhP*Cn&`T?15@t5A{2Mb!TlZl4}t>ugi~%ke70f&ieZ5W3`5J8Rfz1ZPiICVqlQ=
zLw?@>+i&gRe5XVndQm@?jj#Wa;9;thc*q{D5b%O__%MM4wjmKf3WWj*H(m%;2mMEh
zR#4Srj#u)E=v;jvt+LUQ-JYlheX+QOUxY_I832hHs)Uo6C3z{F^{0m3F4Cy#dp(`t
zzqPL1P425x-o!I`yrv_+_xT?^9jZUNshUBp%7#ZaZO#P|swdHAOF}#q+Ymt{tE@lY
z=;|f-zRL#kjRh0~?^F=Uc3<(Gb7$UdJMwMrbF~xS^?#A~=HYo(Rlo4+C?X;vA|fJI
zgo34(wv<v4={(TMCbSG?c#=FB+9XfNleA44go22O2#AP?n9k`ur)`?HSke{|5fKp)
z5fOP25fPEj?fL!Iu#>*uxxVi@-?`3Thf8zyx$k}NVePfoUh}fu91VwMRtj1huaQcZ
zVNdG=eudPF(oH~`ea)0&*L#HeCZ9izRLq5{inbjK0J|1%y#%o-^yJh6LH>+a40GTk
zYDBgLR*kHO3?z^HNj2w5Bb%DU5tlC<dt`?#33k#tX^RPg#8A7l3b?n+-am0W+)AGG
z$(dS!dI~zl(%SZvwZe6JlRO=h!uArIT5)GS6VD8{!Q9T3XYCtgM=g}+3?J*Oki)!0
zp7%K;_aH?Va1-RCEW;3aiWfd`c^Y2A8q@ir9rj9W(Myimbc)$7F9)3hf5Pw)YQ%W4
zxiBoR_`F!Li(ymrT6i_)3W`r3QWS8I*J3xx06~Cx-P@Qs3wHy!pEqJGb%8dH!R~Uq
zO5cocAdOHRv<z^2D~-hjV(e|Dyq%!T^rtGAp*6|<J24CRUt~)f<Xt=IRnz4?|7P({
z>L9tZ^gJMs|Gtl)YXdCTKyUg%;Ns|Nq)cZVr|eCGvciX{c8$^4pWaXrOb&s6+4VtR
zFg=@K%EzD#<6>qpvYuPzux~2Ofv^DXQI?FUDC(HVgGn(u$20oy%(A<x>mFs1igS6`
zcCuB{-P)i}hK%ZsFURaLri(~Tbway}`dDwx5-H08wyn?%9hZ8*O!17T<oG>d7i{4s
zL!C)(6cx_#Z>(ERuvv`PIbo(J?s1fZ{t610dO2y2vNLCW^(;AgkF#BxRgpFi`e~lB
zM;THCYf~sVb&v8E6SKADaO9EG_C$O^MJ;?@Y~`n?RjK7mQI?WolbqqRjTrUykPQI|
zjjLFKL)yc}8mtCWIdhNG;;LK`ger()XE_AS0mDpPJ$<w&#Hprr)e@ZRZxrdUqYTL>
z3O7qH>lrXW)*!e)A)VPC<sxxuBkgOz87fPr*<}1y-=q4X*@PBEh*Q%4=r<kx@%;_g
z-i(o%^@b#8nGB3wK<31HN+E;x_^ni%ol%ixbso9cV1cq%;1A@IXpqll8S?~w_|o*K
zzPK2{C0NA}TLrR^A-!3G)dJO}AjyY}h#3L9=}e}nv@|)|s6Eb7k1HN2V*;s&Vt$CI
zfu@aVE0n}zy=x}&;(^&fOJRJ&dl|^bxD`wIh<}6ZPxPT3$-#Iqv(YN}BTh=aX+{|h
zPb{=&ZH;4Ky)IDcSSEWrb2x+67q|Nr-YYg2a*fg_%$MP)SH*%R)YQmS!`imSC!hlz
zC)e7cfNifHF<F&zz5ULWG$6?{ER;pwt04{#+D4)ej#G)Rl3_q352_>qf>U|B1|nXm
z&(U%|!IA@|R(Y%(LL#BL8Z3j+!D-Q3d)`v;pqtd%!OR(z2{M+khqYy@4J{CGl`X21
znm{Dlq3EMr8?Q%?FPtEC1_WfYhFxHWy#mymVH2Je4dJp?F0=g|RmRKYCZh@c&c8rz
z383Q?(B%XHd2*}0&XiZnoLCz@2GBGAnm+yxLq&#{$enSJAt@14^KzH3ya_T1g#;~5
z00<+mB8FHmckfYIZ`_q~PwW+vc#_$;H&FcQGv&U3Htrzmyl4k_AXWlqYmq$YEof7t
z+{*ydhuUXD7Y`h@hXWSSf)vRkX-#y0p?eHVziTE8YlrHRrC_rj^S<pWQTvcICUCK4
z01i9Gn-_D?yZ}Z37FbX9Ow0-vE?NKrz$9=YtEDORL{%y0zfP708YA)%yz25G05Er;
zo-A3}l}5tVsUCjy_CHx2SW*sM2^YN9hR<96hV}NVt#K?EKxvMFVYDDn;aM-6V!JR6
z=4DtMP9kKBpfYapZy-ADgi8`}MOyut69K*Ww>fA{#cAetD%sb_<1vSEnd)lUVGBj1
zeSIZt5p6LgWf|@4>@w6wr4fX>K50l%k<Y<_D|$jz6QsUKGXp0M1egz2@flwYZe7gQ
z-1I`mp%(ht^vwz=UCCrN%X6_BAQ>q?GW^T)ftp3gVGCCI1z#OuLQVA=XohUEFIr2}
ztU(j_QnUnXMm>dUwsqTIPUDaxfRi(nH?MeWn=kEBB)TjT_-dLdMOrf)zviu}(4wb2
zq+};vuLt_T9~~S0h8^QRknG=#QI&X;fX<h<{27TH{Slcj?d>2~;M$q{ggM<hLhhY3
zQ=|#8Q5^#Bc6g(vGKAe&Hm>(#R<MhJ_P!se1{2m?G@;wESzwG=`@ld6rP?Ma)#l4y
z+iuw07eX7l-`CXd0y${^%vH~p!#=p)444o$*%u^xg8`$*q6u<#$948M41z@V3o-$C
zk2pu^x>p%MJX_6JW*muy(%qlg-DI)W$L#ehb;RgGId-oSw6YpGZm)4{7Lf<}2+Jg{
zAXb*92|0eRvjVgeF@Ir+oZ!71r6KdQO-}Ur`H0}$!9arMt(B8vE?@u=Fadv|)pBxl
z3ffhSq;{YIr|dOFBRg0jr+Rn9AxdQ|JHS(ln%&-=rDxajqjLIQ=OWn~G}m&*Q8U!=
zJn6C5b%X^o!JJgfnen>_#o(~2vm6lQ6`*w?y;hh~R$%NxPlF#URZxqbYw(*kO?vrI
zK2t{VNbkMMl@2a}Y6=IjPlABOTF}tgHy{8?nJLOpl<l9I;e=yQ66*mr&bW*Wj1N#N
zwnEjuGRQwL^As7^HF9w@v-CEE>?QF?YZk%)feMvJ@0M{4$`TI=sDK&8&d4y|7Qzso
zTedJ%HiuIJic{`&j@Wg|3uH`S6QmYR6{#!+bu(sB8yg*rs{m^w<Kri+EYO!uv;~zT
z?I-wayFw<VMY8^z;Xt83Gj}7U2r@aAgvDr(zDiZ(53cYzQIcY!WCpv^=Ie@IxyrvZ
z_6l&<ap;0E)u3NjBiF`CiyKe~fi}?V?cMm%Qsm#$jzsqz*r2$3U|dQJWz?=Cr9NpY
zQt3a!u-tP0)J_sb0(#6W2DzK&t$BHIqJ)0sUYBmLg}7&McR~WH-n)y@E;+io*BsRy
z?e|oEQ^c*c?kFXKF*C^4`PME{oAwO<mQZ4Wu=b*Ivuy1!UPUqc*#UnD_VoCi4_Mt4
z&vtA8!%azdiQMAP5OfQeGF#<Jxz(SUI~D7WY>_$f40i<CFZ`7|(pt=Un+y!)$;EP~
z&AY8}6g|K0vi0}w-Mg3EZU4UcM))Y*vsVSlLZIorwhh2p_@CStJ2@Gu8RRR-$OAz=
zCS{>Of^hSob<@^}2)BoPYHo)BtwSz<pBmLen=mPOiDARxC@+uLmk>wu@@TXIC|LT0
zVM?|ckHvt4i$P5w(>B_QO0wk*P%X_1LOl~`wjdTe0(Kr*<g)?eOI85a-co6@<^?eH
zRmxItpedH^#*^hfkwq|OY%#W!JQJc4SH?mI50=$2&`h{)$lA1QszG@snq>XnUD8S9
zXfx0MeZQnRVP<&JqHYm&C9)}CNS7aJaTuzm8W0B;Capn>q0^wY37fyoyCX$_FClJD
zC@s~s=5eEr>c2O?r|gLS)K|*fC))fOg(#}@z$449z#p2fpG*zNaOzDVPwkDM9Ogu|
z1RuP`@o#7lp~C|U>|x>Zv>l_WTwb1u)s^O8yL{H@-=It7IUhm=1&}X_iqFSAK+Mr|
zOYxKr!=&Kz<ORb#FxOdmu>(9eD2v4iZ{GDU#aV@slf{i|<>k0AaJRx5@k*=?VML7Z
zU<uV%;~A7?Je>%(k=JY$6G1Ouw~^~B!6v-nP3h=GpsxjWdDEYf4ze~l>wUhJo~=^V
zA%&l}t;^m$P%O$9OQKIF1ZHBP3T*>MUx$#@cVp5sGnJq~@BL4jcLe$SHbX>eqU@%_
zhAK>E`+&n8Pc^QNz5d4hRq#f&gnYBV9iFO$phZllBL`y#5o1~>hkaGtH*&vQz`n{p
zV<xF!1%bg0(j=YtxrY~WRMbn?eeS<CBunO^yMMqNjP7Zy^mfd?(85uvryhZMId)&-
z(QtXyBVX>mu&TBvh|SCK-k;jL&>w@yPVi?`B(4SlV=lZC_bGWFFkpb3v`=q9Xj7KT
z$@`SaQ+E75XqZlk^=O<K@8;EVYU+-)UP0KQLOLzA+W@T`%CbUXe7X%Wdi-T_Mtq`-
zs0D8Kh-cI{fkj1CB01Ba88pZ<&6TqPVD-3q8>HtxWy6vD12@yT`<%H>WK?>^BNXdp
zrY`Vcy<>8~EQ<u(Bz<hrJ<d7jE7CUrF!6HfZ~rc;MG=w#28_1G+DtL`YT!O&^-a(<
zBaXQWSem3kJ`S!1>?O@`bL3*%HX3u_=nJg5Bsx*foJVdAwxnGKymX(5FeV64L?PbN
zxnsya=N3`xr%aPK5A(jjV1vsy-$#pIP))gv2ozx+9i^6B#wZ`b_%$m|M@`UH851~`
z@dO)V(;ktjAsf;p<D(J4tJ;T&9X3sUh_I29qETrMKFZ*j3k?~vj1`RyEe4;8yd4H{
zNHKMCRlGv7I|R8*^`5Dj-3Ux$D(!32Xw?52Lva1R|966H@G&u_C^mcOa4Cu5@-wM`
zDNVyM8J_7K%l#Rp8fuT|jy5e8hnzDNv!SM`jGbW!DoTlgt4y<`Ll}+pTcpOHb%@-Q
z+Q8OQ+B;Cn5uR<nspfTSsSeB=h=i+EW~5n?QMX9NSw><&Zs$S0Gdt!CwKl*8Z|Svi
zQxGBST2^j}_H9&r+!|m|`;E*=^EB%ngNN?$4~SiB<W7SjEedj1d_0;rvfORQlitjg
za*y{!trKe#-iC?X>&qk6qy`f1{@BIUkTz6GmHVQ<YUCd&kq7pLftxA{Rao*PTJW$s
z-uEQ24+czd5d~5_WHTwiCe_Kq_5%rufQyerYw(|{$j24+XdDXcGl?2&z3N|i%v&?X
z0aKZhg>H-<aBZ*}7SpNBiykO2D8ITHJut1|g4B8y=r5KD5H1QhRU+|HlQ-Z7(@h6)
ze6rNL^WZ-X1K};UN|wi`6Ef3frGM&<CkXprFRKF`p`gmuHNcU0ZOi~LgStyqQ|p7e
zY3km*GzW$Zt*xIen+(TYXu(Fb_<W4rX@dD+uAjcv*fAzntCekmzs^#ziV2~{;1R|h
zJRalVl4+J5-mPAhfDNt=K#eNIwlp9ka(VJdT*yf8N^7=oXdE$l(tv`5Sdd7b3c9C4
zIE+GY*DH~yy~7THD27V%j7>@1<{ISLnBce!#E|D~u(rmOq&hFp2MrlESYAleq+D#*
zu2H2I1HdSXxdqCsmx63CqJk0oFZ(*66vU@ALT!;(Vr}?@G>GZRUQIA!yemliPAFv|
zjI_#Yw(4m)7%$~@Z$^`<TC+4YALEYSux14{@bHs2eFQgpX)d9+Y(Xl{z3uM-Yt=#B
zhP`;l-k)<#&o9Wk0iK~Oiuqow0sLBo!o8nBLIWt7YT_I;TJnLnHYF{c?Uu{lm^N5e
z>WWwqmNGuHf7Tk>k_v<#BHt=7`C!ZrghRP~FeHc5989-2AMQr7-xx7EZqaqV8F1{J
zo}pwJw;n8V*Zt0%L5GBpPWy$bqtF1$G5dE>HIQMNyJPp;wX9+FnB^9nWZUNBq6aOU
ze0+?7)ocbP<I<87{9E|M!DL7kVSIa{jZj((^gL<5Gk8N`#-w0#@_xVXq2yzFWW*2b
zSEg<B(97i1{q7ix$XlANp{~jSTThG03`Y80IX!x)MTnj1$j<QTF;fc<6@#C8#K0NY
z9hnmUOmAc%+ZIi8)_!MO;U@wayXSu8!k7&O8F2Tz3KVpiT36;<=;$KSYriT>$4-{s
z_OH6>(x-iDjK#zH`+6^0174M<^cbMXL$bF<`o}bfq9hT`d;_ervoDc>KH<f<>R8&6
zX><m8cc`WSlBYElz>+b~j0OY9#jzlifrQ2E5(8~T$t=0ldxH)D$5K@VGGxCiKg*zY
z0}nMUM#S|)ip@&rBMkPK?s~X*j7mV!bR8Jk(2!mXRA7$T?{1&@`n=|QcANOvm{(Pn
zWy8iBpy7>De3?v)ca=HHrjvXU^rU90&J8eM7y7U&Mgo3t-o}71)?-?<r%m=r(k{1A
zu1GzQ2Pd1gNUn+zOOW1cYBT_3Db20I{c~+>9jGg!{BeI>@2x0Bsh$~btwsBtHD_Wl
z^Pee6AW8Buum!Wj48ccBX$%I-8lsNWJH9-29UdXb)&j>mKTO8GQs#tBUlF7-eZGZB
zz%(;Zb({y}B`lGe*hQ+W>)~~`-<S)PmcdZW0o&J+mj;P|_*9qrWr{qWnRzn9pV4-P
zWGsNoJeg(B5ZVvbZ+&eo&+zORF`t8yyiRV~ADn7?-aD`vAnVjPL2V9%2m3*lm)zoQ
z)g=ex!Z3j?+NzQ^^4bv3VN6jkWx3Tst9mI_z60lRc~D<KeZrgoBLefGQg4tu9E|N8
z6<pp_33F%c|3v7q?(%*~EJ2)M=7(-UeN%PTa<{`l&z!^N9wTbkoOku5a<84yVTE#E
zU=4IT$Swyq>;c;w5H1>u9!xzkZ;IZd@=zcQIU=GbAYfOAvViiiod?K*>TdOjcL9t-
zLz=S|$)l;;Eo$OI<g7eqXMMu>LTQXo03~X@X%?OrM+6XGCkw1`zrGj9qSTm{7%Umu
zlBPf~q-xH~QfoXEjPUY!F@>17@+fe#XeHghnL=T8bPB4_0?C=~*lP`I;6~N}y-ilu
z$GJxV6R4@14XCh~Z`$8cQ&5Xk%CG{o_*lrE1`dEE6c>?R^Q}R1VCTtIL3`$A=51*X
zys9a3%$3LO88qG~s|(j6Mm+3D*wF6Zv=B4Kse!p?o3te?q`QdaPtZj37#~S?WS4<w
z+=OBBWE^8?<|EmdiMdZjC(1{2qbWeG0s8TnZI`F_yWV3cmXNE4<W`9~0nhjnEFlh2
zWvM(HV`xt}U|PWlnF{1N-(p0P<A6V(el=r~FZhNJ93U_Lx2c$?C*4{|G1kjV)-;cV
zcrW|-HedWjzptcCr&GEq#0+s?O|!KG0*tbHEf8iLqBrFA_yx|VM&7W2Dv&z&&A^3%
zx=d-QycIvGGHsBz<FgXb8+pf$^5uEz)87q(%&kYX6E6Dq?4-6eP981q+Yq_R>GFZk
zZXtvY@?vzV+iP8+78nHO{V`}63lYn~KsfMtEx2Au-Ki{JEQe!EUIONwD+iQ^nF<p_
z=L2DRwG6L}D(-r~c?ld27^A6N!ryd-O7{aU4?<Mpdo^4GFE(PHObxHRv>bClxjIeE
zhiX$k4qMP4qY0>bTzW=S=9^V=`~l~25Di(IkrR9>p!hiMhFBd1nG<agRWC1<lMXoh
zR#Gt|R|`J-<O9wzGJ+YS!iN%2*@B%u#otiuW$HDH;z14pIn`%nQa`!Y4RV@2YimRf
z0pwxPaFx>!C`UDMgq(4}ui3no$f_u-mL3O`H=D><UODrC@~}fPNS%C^FUU--=Jugx
zMc^jHIq4Zo7@w`s?7HVh=Wv$?%|_WC>E$!Dgnx9vSS`H|D07sLSe`yMCIVoPZOzBD
zaa4Z8u5D!C1SD;6b+I~7Ns<|sKp2p|p(Je{W@N)WwX+O7;8YI_MD3?~Rn85v-Dqn>
zNz04m;+Q#(1>A4Tz+YbCGv_&+i(Z+&@KWDGtRHgv5dVc}beQ~x1vWr;SDTX&zG>kH
zLZ<`D>rFB$z&5TvUs(aI`WPR{94bt$F*bS|lOfAIK6-=ZjvQfoq3wzOF;J4>IU0R8
zseNKQ)=8o5Z(AeqOeQB>vOp=1KLlHU#Q_%|$fni9F;}kguE`{X1Txi*W8Vu0$hGkS
zT<NLW0bl1IsLB{~VL6E+`=AgW?j^DCVkEJq3reXUPESt=ZUKP)e;m$oLl3aD6*A4I
zPhKWGW{u;n^s$-hjhrV-=2aV_Sqg42z%`C2mVcFk2ah)k8Jb$GHu_N4XS<{7qL0fh
z(riWoijo>A#OBB>hd0DGI9trN7FvT6ne@QdAyKaFK|wi0MY$!K!`p6~+?tj|Ifz(J
z^T9Sy!#Ob>+<8>$GH2u+{*0P`CNx|RuKdoJ68dWfMMF+oxyyhyY;fUVx!cB@X}K!y
z@z$haP`%@#ZIXKf+u#)ruJgV$2(wEohbs^Gq)B9!$%Ebk{`(9ng4$|@JQSV4jSiRp
z!#0D-<Q^Zfe<R&@gFI@WK*~PcD31knnf?TqPGc+u@&ijiAs3p=ODN0ZjpGQRm|$8c
z3#_XQ2whedB{*ux0Z#2EqXp7rD2NQjvef!#{%@#WGo^GFS>C={fL?7=vNCWRMr<U3
z1R+~($V;s&<Y#=g*2e>yD?3iH(dyi<k432AkeZCG((KQWdI<ar{=y8uDF&wg3i}Js
zP-zK5lxM(-hNrfsfoQO_;yK3pYzv5^nkUfvac@YJ)`$hAfFwCP3<CYm?<;L~71W`E
zR7{hpov??r%RK4NC?QdSPX*jl{><F4Kzpz!Pshm6LBkLH89!J&mK$)Gn4X*Uh_?V6
z!4Jr@vCVn7je}4;7hP)YQ#vKAlIQ&yOc<ax=C|~MZ7>pL0xMti<uN^4B$=a9YN5Pj
z%}OC$=Hz9=HR#6c<dxK?avI<ZmdLB#4bU1c4mM-c%)DmZ;O2z3;B`k?<R6ze4ENg_
zkrNIIG4GU$+09t&8XV%c;+gs0s}IQAeqt7ZDlCw9yag>dEe`E>11zPLIXWM_=WQU4
zfQ^Br1((eG-V10dj`aue1+I1IUSzL-u>^`Ex~(8YQ})M5YG>rL^>Q$lMnUTE0q2N8
z>8mw684YD}(6}qYA6hsZy%y<w(6wFTs1)ydaMy{@Z*j?S=rJUtZ@R}D5MMo_s=cKt
z_ZWXeMksQ$7_^VIzHyY{%OS@dbZ@z#1qE_^{6O4-oN!Qim+9qlqW5K{F69Ss99SFS
zB!3h6gXmUiqnsRFF$3Ev2bEz%>I#<PkW<r&;fD@l!(2S4`7;ylRnV8y?K2cSL6Nz<
zoN>@OH)Qq&)hdx52VL1yo|(<Kz%v7C(EY&AEt(roo0p!^9mKxubUF8+b0>6$K`B?w
zRNpJQ7z@pf^tNA_Dui>PAZG$L<2+0@G|r;$LBGPM)mOuN2z*2TgUUT45=;i9St2kL
z$Tl#h0Xyjp^)krDsK}$@ZB{NmXlylsY|eQ|76ox1Iw@R6Epmy!r`?*NNDM3Hg+kG5
zEe`Xg9k$)xDEEUehXgnZRHn$wFq>k1cD9T-xQhzZNjdoOj`EE#pEF)`i)2h}W@V;K
z#>S6=>?|1{D~HaPqGw(v+H2_Q@MBV74zQZSEtZApk}?*?enWO)O6k?OPbWvuIRukk
zanKp&8bpd9x*tHjD&9}ZOp~dxQxhq8T^pM@+zV||`FFjKv6zlu0OJ<8t`ym-`}V&`
zO47zaRsuSe#{Q~t!_3ASnkwbq5rQ$GkT$dUv{;(vzn~zCBF-Rf+odvQhM+eD4#X%$
z$W#w%S`(9Uv6;ZAHTbtRo_)5|#k!%}hGvTEp5a50*r0$R=sQ?2Kn&3Y(HTxT19;)y
zA+!A%k%LM;I%?h&3xLLsPH)IqF1Oe-WI8~BXPhW{LEake$N&)WaK`5ZN@Ag0{7Ti&
zv}*(Jh~-YnA=l=e-UB1qc?;1g@2&s=69V8$zdL@H!O{*1?4Gnf+KF^!N>bO!y)nmJ
zadCZ(x%1s;UqA6ZD1z<<);wT4ikuwsU?2!Hk31CL4hO^2U<FSej*-EJQ~^I93G}VZ
zpe6}%=Fw=8$s_YbjXY*sgBmWgq%pRs;7Xb2>xAwdd%~SrEeq^h=wPbz@Pd~`0U1bj
zFdhbafjeQ`3KIP5-=(tD`vNnJHNn*vrl944(1;z)(Se<LjaPa@3*<K=+iLGtL5+{s
zf<GsUtdX@bJU*_L_3<rkPcXG0x{IVap5dkpd*-Gz8LP?1^64!x8AvCPY>AY4X$@%f
z87SLqDBMsiTkv=c3=o&gj-YpS^+f=An{C(V;lpHCz#8}A3=oqigG4dJueLNRPx)N@
zYDC!5TzNWyky?M}jc^$@%QLYWT$6eh;Uo5}VFdD410$YG4Pd7Lb|b~DJa5yPfwX@w
z_&>97)$&114hA??jZ2I2OK}>M$E0P<GA_VNT~%u@2S$}32Pc#H$}6!d#+cEq{MDF-
zix&(#c`ZF7b3~0?le}(VL^c$fUEYXp;Q0-{yG-7Uk>JcQL-nd}1*Cz1@V|OHPQ&^A
zFO+xe^NLa+rdA_*H!YEQ6FS&h<-HgQ_D~e6zFOY*$x%e4`xvX5eBjTPfGS|Z;0D&n
zUWX0Bv*|1YYdfNV(FH9V%-FyG;I2>8P6yD_TSgsk@VAv7SlF|Taxl$_j08Ai&9+L9
z!^7#`TT739gxj)Hs|>XgI(<20_om&2%wV(}YoedBDa$(_^2=Tirv|p=rRyQ%yeuTZ
zXjS(^rnl-)v<m}iYKf%XS}DgI+QkdAh}xQC4<%j=$Lu)!#VG<?j*k&kX{-~L6Z{$P
zK^{8fD3cTYS@K<g1I<Z?jO&ArNyUWr^!z9L8*@1@>z`rsDc+B(!su=!r^fsW$V;B)
zEi{e~qc6dcI^BV1!Uwhj8$zQ5-8~_Q${GHR6`^3J%dOJGo`Gl2)z;Qq(akf{jF-_T
z5NZYT`MluHI;8AS1ro^gOjFb-7F=U%<lIBf$YqLZ$#?bgO>kwD+Fk4rn%xDvyz_SH
z?L848S>6?uE2YmN=Mi)DbttflGMzpr{RHb}#!m9i&C)-b(qUjGRKo*&R04ojj1)%%
zOoBfLkbDskf8Zfym@_3MVAlqD@8;t|vkQVs)Ti}ghbe7!MAaa=N_yd5bBVuK2xBm_
zw6%L*T$<)hdkD!K#X=s1Apv=0=4UK1leY~!q`dkyH_HgaOytb&C!-R87y-|SnawiB
z$F*z@)OJ#xvkW?1vU<y8Y+y@n2Ha4`dk>H-sgsAE0|`Q+g%*;Ybja=7(6?woCzxG9
ztqT3a#W?<M5t7LPEIm$i#Dx^6R<A4k(6H(eY7fMqZWM(RfJ}|enu(-gq<+`N1LGl}
z1M`J-aP(7#2=!@T09q+UHb!|?3$&K_7<3e7o(GTsDGe+n`NSH^z0E?9GcKO2%(O$U
zdWOonv`#8hH!H#Z84;`YXVl)&FPHHTHMTzmL!{QGF~LquU91A~`E)L8hJQ>^ehIe(
z^P1%=Guno~IG7#j^=yYLv(!*d&=6pq?M*i9EM!beYk@00&I(;|ZcWd$d=?5f+%9tx
z`t+AO?B6{ZWpHOq6!=Dal-%VXyY3XxPK)L4XjFpofXdjV5tacw2Elv%8D(u|Z31e`
zeF4?M&;VD;1L-qMh6e9qdC(>^u@VRMAz!fR$AQB;8BWN2v=2KXvC`!{c_il36E_*l
z(mfh@wqp^9^J7MoE-x4^jd4AI6M=st_#3dP^L!d$mlfjkpmAJaUDBc5xY0$ioSn>^
zpxmG|1tPJ`HQei#2I7P1a<49I?FNvm<u)M{7@v_XmX)z?DB(6cxK^jlrH_uG05>1T
z$3!Hlp6fuOfQQgn;=D84;(CJ{n22g=j(Nf5mian1#cz5DfnRNj@#*-0URZ#%)yN$<
z(BcN!79ZrxRd4dR{f3dUT5ZES5~hYB4SHn|MIM8dIK^!-fH~dZ&Yv&4{24cJ3rP|9
z%_q|{I6q)%^ksa?_YS%-?A_De!rWKPLFpNrYOaO{=~-L%%{SlNQ=aq2JRq&$NaiEB
z^Z7$AvLina&O7ulFT^gXRhkr;eoLr9Brn>~!w`m)E0dSJH(Iah14qBcm;KvNkx?Kw
z^Xw~W|JGrbz>p{<rq%ORL%)6l`^#(5CGIeUFVho7UO(jWFKXtMWe|gn$Wc2c)F*R^
zZ~9k&92f*%B%9@}L+vUi)&nyneLEJ5f~v4QVXNOUSWPL1<F&jSLqWv~6F=Qq-m~{@
zjT1o#-#1v6Qu_bEUX;M_RxEqlXFzWV(}Km9Gb$?OV1g;{8JeI8G5nRoc92FuxX)?c
z1v#uNLM7Y+J0CXA$GC6WfmmXWNn*Pm4qcf#QiFKUmF|ZfP^^k#`6`Y%tamgVmS9{x
zHhr_5ksn|h*UE9$09l}ypq%DmWBHivgRXImH16Pmgb6dnzzK(y&ExI>eNH@V{3Z?=
zE$ATf8bVL8!W3gMDT*>M)4)=Ea?Aq_j0&uiQ~X&IWCtYC#0|VvPCdM9DtODb|AXa~
z<+Q`fze22>EvNe&xO_-4s#iE8&6N(nru6O+OQQ1`oTq^}XZl8g_W}u~hII`{6wT0Q
z*>3jj-%onlSgcDU=f--?f0Tln^g0{{S$e?&-FkbkCeZ>3DCdzRNgwN~t#Rs?r0-$p
z8K-E46X|d7G`-FMABJ+WrX`0f!N4@M-mD;z^JGvAJv39JwG1wfPRlDShg<4s4VGpK
z<x73*p=ZY_UkEma_Tpw*!J*$mJ~vzXgh>T5Easbsj?GXGCn|#yoYtnIv|7d-c0x3v
zvH=mV$R#m04L}128AwK`P?oV9bi7ZBthJS0ZIOxA`%GkYtd>dMK=n8Hh!vR%4=aaQ
z1xsX1qz+f@6|t%GG@xr@gIpDBrio3ZI0T4Bv$45)Xeu++8M)Ta1KcE$5ajwaC%Syo
zk9@5Z*^pDeASKZyb>h6lz-g5FvtTW$^e8{<%w!1*M5!|i3Ei|fdU|109uBK=RfZ-a
zRgqPDZ_ZUxCnq)b3@+|qTB^&S@~M_uZ%5c^ED$JdU7TgA-m#U8i|y#{HpAa{qzPue
zpl49C%nIP-aWE{q<?N$S7=vlYEckL$Kz{_phg)Lrk>idWMWx(oORa|qKU*bpf)FB3
z=V+739fm+vs10$aLx*z5Jh>}A!7|9%C-*hEJ8&E2qUsQrbdSv>rv{sQV+~bEZHpqK
zIEMEnP}%iDKamG)OD7MO2M;?V%xq}Y@=(AZBy0=l;ap7RVV@Z8Z(ztHM|DS<x<DR{
zsqcW@SRS)Aopa9Dq%jT->`L?Ekg`ArBViW=)DSxjr87`=ku9Y{S?H!%X91LUu)Hsg
znO4-|;*{lSrW<H!Q*yU5daQ%@ZA}hh{OYuUEII}yDs>dp@yc48GtxBIK|Q$Mdvi4h
zZi7~7wsn*v0nR4-CrVjLON<K@8OVm_9B8$^K)~ouvCS3@@K?*@?a<QmgwroOQV*=t
zLyyYk(iV%u)s1U?mp4)46PmE<*OM{fA_DIzM`p_MjB!|@emhkEQ7N-oo{k>CFf&e)
zH|#Thkk654ZHj;p9PXpv46C$1A5e$a5_5~a5d8qaTkSZ>i)jI@WPsd{z^Io3@o3U1
zRoU@nzmGd8f{pT_n>Dziti<*!ffmIXvktr(y{nh4B~W?I;6D@Qtk+{g>XwiD8r;$w
zL1|U7Zs~AH&*pH#pdfD=UcveZJNH}On54t?Hdw=Qy_{%AeLK-5P_E!a-Z6m9&f@ZW
z*Jogn`AQVzy_glw4Y9EW*7f%t_l#_q-a|PI`M_Eu(9H@3vJ&s!IPiM=Qh~G(^mjhE
z=>2J{k@uN8MTW`%B`71V$CYp};dr|Mq54!gY_n$V0+A!G-a-M}taLsStQmza4Hh7O
z4k@+k5x>SO^3dwTD&E~boW)eoc{%2Yv1D1M^E~#5vL9`Y!wcoOBgWEM`9E4%7Rd4T
zX$HQ;`1FJ$yUqkdgwaUSOz1PGs18(GMWYpRV(N_=MirOjq$6QT!VLiI#N~BzOaL*D
zryNPDL$J}tbdbCQOt{sX<W!rC4vYXwA3!ylI>0J9-Jh|xQY&!+U!CF4n5q(_0}>r-
zFdcE4S)HS;?@Wg!rF`=5>R1AO$znO{h;#9oGI-1)7Z0-RRZ7n^%qn+SX_9keA(z4B
z%NIi0LiIKsji}4A>}$)Vx3wL@<J!mjQV3yb>w)ckqkk}?bmv0QwDeDNB!^4hE(Fs7
zv6$Ly8GMWe29PuGOn{;r*Y+R-9Fmiw1cqGP4rV9^Ic_V)XfNV-$q{8x@<_1eV}@dr
zIqAWI(uNshj)3x|T!#6C$%MCJw_y5^5eW$G&e|p<n$z&8BhGGRn6!j^`550{-k<Ab
zto;Ichc}vxKjKX0xGZhJGVzErDAk$j955NVzfj{$CfVFaUnPZc3P1+6cH!i-n1oTf
zc=&xqQ%IlLY9QrRu^jU<gD2G%nHm$I?HK!Zt+#1{x&eUzifY#%G36@EphdAdgq}iH
z#VubFojOv`qF3rGhI$`lpgeYt!6&$TaLT6zRt>&PD!l_Jdi2E4hlR5`Jwr+yJs4Z1
zCce3>L23ht@bSA)>S9uA)~SUEJ;R6r)E+aZ%PgM_WAa;Nb|4(AwPbi%T3Qdj+~k9z
zRtY$6V=->AZ%5{GWjIi`21N4+K!;c<a{^yxA`&GdcX$uVp5hATWqqfQj_?3-Pu`*l
z3@lEv<{rH}7B;G0?uofV6T4XM_0}Y+>emVl)P2@Mmx>g(0oy#(qZt(`9yDB;c%?iP
z(?y(23Cnss9IYu_o007iAA*+-?RwNfKI+d@9Sh*FcJ5>GX(1hdsh*O?^eOGY(DUQM
zn`dhojQnS^Ak7|yWpF*<xXB`iKDBzZ|1M9LNt1n9Tu}pkR!kfyNp~_?9uJV$L+gXB
z^uEb|E(X3M*^;b|O&<$Y)LO&YJglngeM}ekRM4TNWQ{bZv$KUFxtcbu%-^Qi6NpPK
zu_hQ7THEIjLPax>nUOZjHrtmrRAYkc?c-_fRP|c~%MRaL){9vnZ8pL4FFId##i!#k
z`1HvHRu)0T(%^yjlz&RyYpXmRXBO$=ve2bGlLmu$kNtvkvpj3v_q(W%Ja^Rk!SsN*
zn;ZV~2C-Zjk^2SjMbl(qTok;L7mxZ>Wn{eB3TEV`!2B9|q{z$uu_=eiwcIMNq-Run
zP`|QDUX9PF89{jfo8^$P!{v2v9X1a7M809mc5j?FqlMuP2>e=k%K<~byY2F}x3Kih
zO=y9L#k7R*i>0AuF&6i(Q*zjm=v|3i-%J+8OoyhKe9t)wWEjcH``)#wLCvS`gV?u`
z^>9qsYriT=^c3z7j2&+3UODKa(Xqw~OfvmzWGriv!*L}~s4dG>SIjoCQ)KstR9Ayj
z2gjyPvimr4D!kq3cf<I$&azvbw@7r<dBaV~Ez(7HAFChdXI4~cv|3mHo;Ub>@Kp>1
z>*mkszCr~v9Q~xb?Cypyr)Y|<52Zm-=E}GK7=Naa1k^S{<Mv_MeF`UMxr(um>S#zd
z$rO&qTBmJ|L*YgJku;M9s?%Yl$#F5?2sW0F+9yya<2F8Cc7NEGLxmSE&yUIOkJ^KX
zoC6x4;NT!*!m=?fda2Ml{PAec7)oV(z$u=XR<BvWvHFeji5O_8az?1BX}fSkJ!Dsm
z`$^f|1>^Qb%8Zj^8K5U%H^A6zlTY~!X!HTwMF9{PrT+c_0eFnar(?CXIjZ{6@2N2t
zjn@G`gsb#3@eH0nmbGi6oaWDn1++e)sH}X}pOMJH1c<}dET_xv?%WW_!XUk$i@_&N
zD3CMK95v!tg=$l*ecn5wpB&|T=A`Rk3myXf#usFFXY@I}NY0eqG*RbfH<Za2W%n`o
zH$0<j;p>x?vtrgzazjJUzjAhTj6_`IoSA2$r$3`#xtemJt~)2@IdsBgIoBI3hI1-e
zMtCsFmwZ`3Rov&j>=#_YnQrMkn-p+tr1X~EAJIu6io<;pz8v!`&eb7GSNeGOfF!oc
z46EnM?rZh4w*Qj~(!S}`ZpEp!MB*kj^@}54!W&)s8=4@mH$)XJTzi477)liSZVd=D
zf}LC|YRHA&bx;~2iCK0B`WxOrjC%$l$6qMa<e~)Ut<cA6oevolkY$n)`HG)z=HDWP
zMEOTdn|0IGlR+*H(2fPWp0AXz#@FMK3_vci{%Os(s`Nnjuf-RZ3QjKdIk0Bce02@k
zE`tLY@F4?tIwU?BS(`x~e;FEEjMrt=$npZO!=twp*Vq~v?$7AM5cxbtq-U$uBm(P+
zjI@IfUoAL;*8upV(k?MGDr8s1Tsj<#_BX~<(5eIl2Tt7v8Dl4;c!peNAb>rX#;>tH
zATLMA|Ed=n7r5RR)sn`i25T7rs$~Zz*t53AX{hT3n<0IlCwjXjI1}3deUi?v%kJZe
zwBK#JTmP@9E0gRrR3VM26xdKGJ&*HQn1-UY*4W?6{h4Ym@QiLFlYP^PD%@xd<5$}n
z`Nh>TCFVc8vO=y1vN<eQ52JZru5>!t|9rVB4aoZdv>6K8tJ8qgA(C@|IFj1IV-wHT
zA>GY2Hb^~8x7Bj3!=3st3k0P7j#clk^SRQbMzlu8LZT_=4HO6Yzdm(MA4^K4+a;3@
z1Ot&QZp3)tqCnru92|d(?OSSk_0&-(SdzwT&tGEfgWaoH?^{R59JWI{!obD(QW`VC
zFKjdf_%%`%&vf=&!+a^XIiOY++3gBD3h)(H@2qKYGKy>Kp=X~SWFS*nkE_1EN-Ayh
z@o9xrdGAcCfyiZbh_3c$hQStVlZ!P`ouIdfg~_oPH-qJB0xF{l3n$7AaiK!MtSG9$
z<;X<U(Jcm1IsutCAB%(Xg-MQD(N$fb4%7YAr_-di81=`2egGoSj5zTqamq|%Gh>}2
z=tKcX$Sgy<ahe(;2cL-sn<S{%CMbPpCxl5EMF4V!Zr>RA%J{T8?C4Et$7o$%xFaRm
zkej0^m@ue};O-EF$I627_Wp)1(h{&xd%9U}^{Fg`v+6Rr&3+-f0-}Xvj<u%+6sf&H
z%0VM;56)ts+>t;sKZS!ENWSR<k#dt((-rg1m>>`o;eKWEtpK^_C7>PQF5g}Ci2*J$
zB>+2tDOuI9-!@o-yRxh0?$k%K`a3pZE5DPTLG_FCMXUQg-Y*P%>WXBg#>wWYJCuCa
zFmcM@NyFt{n<&hv&f$J90i6{bZ8|(MkaT^h`6^}UU%B6(87VRq=L7yM7DR#R`w4nf
zJzAM3QkZS>U~mZI%H#*OA(J#WzdR(nT@9#}=2hoo&ORK6PJyW;nEWu%5_U@JA|Fn3
zXr^Zb@N1R)D6Ne9WGmwVTrZFK8zVUO#uUas_I|A1mI_?*Xsj7qrCEv~0{+Cn2*VRX
zz3Ghnm@TWKrlvxEYFGpvg*5s|G>5>;3-&_>88S)5+`wSmO|{7Pf{SHdYD`-oW8TO~
zG7a+lKrgT$>Ugujx10ozXaLp)D<qV-npRn8(?Ld!I$0D)mU-SYR7@@o=#2z%1#i(5
zw5+7I9uAL75)hUEhnV$Xt1R`Cgo;!^_hkVRFe5Z|Go{ROpM6KC9gC4!5v_p$)K~%$
zWTijjF9Yg4S!I*1$`qH#>X<ZK%8M(RDr-%Ap>%<AAUvQG#7HZm+&XnA>*9_2qS+SX
zFVK2@^bGcnnP9WBA@!^|;J{_9mF57^IOe;VDI4Q=K8q@ZC9)~?&(mfuO4O|{k<A7v
zrV?Qe0B^S#sumCnu*^#VQO<Bn0R4t~=I?2ZNw~)zqhh!EGwNwj;Q*fgb7fmPKjUP(
zcWx$4jc?@@2Z`bFc%zbqsu3|srXc#64U1shGJ2fv@F6I&>s^C5rzh-cc;c?MEfvxh
z_r#euO_iPAEsi=6*~DjC%`WS<C|fs^^&EHmahN?jBTuGlit)QtAA>&r+_<!<=gCtx
z1BeXHngStz5uY$W0`0^&W%6{IDp?O|_gdta2`yLRTt5?M3K4$PuwkEmmG-S30W8S&
z8M~i_M|vd_8U5N8sC6FXIqLxu3p~4w3jfB3HO+(mmFNALaTO$Qo8`CZ87<ESl@}6a
zUIFQmArmyiF<GhdPx3qe#t4H2u)KKGGmAL+eR_t%f#`v5qA&R~I&E*|Z32$v4*`Je
zET+dHpkKD-A%hm~w?D==YBr|YZjrp=&&V5DBBurNrx;IdPPCP+lUHT;>0ohzMijLH
zew;YtT{NT(tD{Z$&jIg(hLY-RgS^(>6Il<qjBl5}#4|Xgkj}BH!Rx+}6pdQtul@_>
zOZeivVZWf-gD;o-SNfXy^BK^)LEiLdESy3@xDCAJ-{PxL;9TV`--!;C03)~iT>L#+
zQ^_E2dw=9o$sw|Q??2MGR0cCi{X%)io*^+-10t_$sg#$08ucGJVv@X@4nJ=}^>6~1
zcxC@9I&%OsDaU*MjKX*;Knu;P{5!yzDT91HL*BQ}2lTl>{^OlvMO1U>MW7e0eBf`4
z6DP(lL3cED?TPN--O31py=fW@#iuf<N%jSSDI6~Q9coTmfCp4vT5ts*EeGrre{GN)
z3~ENBNmdR;Ct!XL`yh)^?-aUN5d2MYB;8-?=fz+ZY)8sQB+|*0YP?M77rjvpUSvo+
zbuuj%CX1^_^&|MpBV}mXxs$5S+8Pm)BV9U~z7B>v3d&kgr>l2uxZA-eOuBV)EebeH
z-pAcLxdMYR&2S)sA?QP$Tn7gFpxF+NNxgLpp5jD5+$pqU4KR<vFOS#8u|6S&K5Pao
zSkLPt{tVY5z-t<x<6;#P;AHpF_z20Fwg7<;IVi_>QXLpV4CP~;R1rkSXF0*Aqq#M3
z5X>fpjNS5aTU>@~VN~~t_WOzZz(3pvR{0aLc$C#8;XKKDtw-Tl`DDxx#^>$MVVvAa
zm0Kj+RYmftfXf7Uxzxg`<CHY#HJX=cJLIlU`!jl5;8g8Ed2l&3paeRff-67c4GpzH
zWO47E<||7@ewA&2-;I1WZ2?ai$qAxT=&&cJ+ZvGI;7a*itZ#Bfz8=T)j5L|KdX=@p
zh5Yl;hKc(TThPOCjeD{kh`$P02&ZOnftEq>3$_UruxXx|08bmXfuF93>*R}_cAE;h
zfTeNeEFYKKNOxFfay>iEg_-xvlG8Ju!GnV`8w(zuW9zA*AbzgDCqbj?1DVI<OE#YZ
zXl8oFe8@lI^iznjJLx>zxE$ySQe^b@C9<0hHZ*`|$;xQv{&H+=aYJ3M4#~j#1ZG|a
zAf9i1O{pj;L*!%MSZNuow|VIo(5%dsW#C-gKYka~)t3N8F0j=M9t;2M0r9RjQv}e<
zh0#?J6d83gu#;=j&9NEic~LZgr!qo^2RStC0w6J5C|~hs<bKGcF2|*Cu|cR3x}~qi
zC&-4%4JC7QNlarTB8Xvi_*#rRVR~f-fqEIa)K&~Mpn+s?w4Z{S*X7WUC8%4mbCx`y
zrlv`Tw!>LNxaq*X$ULW&8yps3odRXTaGT}e`Vtvo&ta}vH&aIXlzBtU*K@dBM#YC^
z4OKGQ8=!X}EaRh^WQ^~(`C42StA?)yec|AfAY<d1hPRmIX<R%5_fOUnky<i7R!jd6
z?7@U+L+>u2;9T5@6XO}aTLCTybv)(k{){+ARVWQnlYCal#SF>6AbQSQyndm70lo<O
zF}u4iw}lQHKIHO|GT8wQk6KkG&DFbrrF5qFr{;tMl*hTg!dsK+0w!P@p>(@41_DXq
zZGyb#a+PnD-NSYRS^ST>4$0Lqw&yuuxQ0o>Q`0i2<;%-8_WOzNLBziofWFp7Yirb`
z9A!urS|-<}-YBrAS%W3I-sc7oGF}S|N3&!M2$!QqLSBjthR|S;k1b9Z)~+j5<%DA3
zYAH!eWz-$-y%xy^e&bp|>@ucYng->?j0<Gv94Yf>%3}g+Z-W?L?&koeRj3jxQr}vs
zgTcx&EpE?g5JD?TD}XT5(^_flF?k+`0A7=ocAR^u+Ifpq`C3#niVbUb_0TIps=Xst
zVX_z_3p?K`xik}2Jm&W;Qsb>7mVnu5%w%>$9G>eC0#j>qh^&_R=n3NbJg1#?{tT6r
z>1vMivRUfw8Eh}rB@E=HR5c^!2a`5}!$5<n?%x7`zM0+w)}hK{=KY)%XfqtTILM3*
zF<<Djp)OZN2-~;F?DU4!ggR<FZ%ni0h*n?Zrg#SBL#*KDPHI28_FB0m_J<|zz!O01
z`i6HOXk$hhYAbH_H<rpy{l!6E=!AA#AYrB+J_mD*rnfcr=_|L#@MyONQ)0fiJ7NSV
zIGa@*K*|mc`~7BIG(CU=%+h;jK&1puE4Tx`mAcbtCS1eo<gSFj2m#M9xzxA46=@AC
zwa~rsZXfh0Ejk8;HqBd;CaF;#X7hUtLZwymU4M&I){x#PqivLX4M}i>D<NXa_u9KA
zLbThK<-UN+5K8y=$3VEJv2j3!CGtSP0-ON23q@rU?Y{4=5lKPb1VYm)dC(vWbxl@&
zVDD+o0lNim4ub7N=^GwzE`NpmFb*ktSS}CSt#ZNn{p3dmTf|VJ;2mK5h@UsIpbVf{
zhP%Ode$^?;k8N=1p6cXL`}Y%bfXzRNP2&|`31`K};y{gt(dnnYK-0@`O7v!D3WLVr
z`s#AX{*Wuv=;pZrBd|^LNVI_P^J0Tb;Jc*uw)x&FB!kU><br^Q#!Et107O{mg9i!O
zrmE3ZxQ-UZ8<o9O9>dO2q+aZAOuI==lGisgAUCCNEDZz<_X^|XvLxQ9S&1G+Ov}M5
ze`%U2*(vl1H=boNQ~EDipuH>)U}!Q(Wa4Laf))NTImvLCSQ#C{GEEHw62Hi*IK`Qw
z5`<`JLM7jI2D&VgHL)hF1_~%_ErmfUpw=2bjDQY*o$YCkdbag}xCmLSl?{G{-3Q6z
zen`R1hX0d?PbwHaVZ3bgDe27#lvxERZi*4%CqfdO%t5j_b&Yg!(qgzW0#QeK+2S9O
z&e7w{bmIhLs}EuLvsk_6wgy_10(~{F$~HT-myMF`333cB1dYV?^0++%jYeKLh7<rn
ze%3ysq8qPqYVa1z4)4KaLt<=`JdqZq0;{nQ*yd(wGcacI8QB>_1GDJ-NNf2nKU!2J
zcm`!q&dtC^Dnp<iOy7X)IZi#9p3%43cI)T<3}Aq`uhc@G@@JN<ombjA`9*+cW|k}|
z7W%YpZYeXQ{nEcsQJ6dhwPr!mVA83|%C7>WX29HukP-CothE?ckw-1_2KlwEcr4`2
z=MoUHRoFIOOWeYx@|z&vNO;YF;pgqQS3}FoZ`;=c@T-PWUP!&E@ldq_xQlRZerH2w
zZ>W|RV}3}YKAiy~W%7IPOCd$36B>q>yf129EEO^+sp9<~4DSFPOzUND2F(WeT+a!k
z#NdwsNGL5?kQzmLd<iJWjN7JC{uI;?EJJN&MOI#o&fxpP1G)`ZfHA>={yFWGCEZ|A
zwbu+~$5zN+{5wTfQVLjNY_6}z-pqveL7n_H=9EFsyb{uhHw-Y~JmkOPbPWc7E^m6D
z=2b^1Sh}F)^0zcg&0p^n3l2^4R&+n4zM@jysQzx<&&rp`+r9)^);1}Ju}=OGn|yf<
zF-6|-Hldawck|DfR$(STUEU31N1G*_4Hdus6@V;<_c$KQdxo;eoLMFRj!A($GC3E*
z6>S5HaU133KRzT4s0>l0s`3M0VFxx1Z<uF$kFBRXQz3i3krFL{yPYuZvl-*rR#3eC
z_DQaxL=M<L`@-k-pdaf+@a3g(Nfj+{$58zY5cZHkCWk5rdTkw!B@Qkv%~asNJ`&(G
z-ItnBq_e53aEt7mgVVGgr?XRM)$Sl{8F7>L-tK**e`kzum(HpP>fg7IbnWc=t86`-
zG_%sJvn#8rbMREe>F?gzbz=pHZ26E6M!HH5a=LIH)7g}3u%H8&W)omgEp5@5l6=^|
zSw!2g0(+|*>(4M<c)rnlajSf!v#IaYivjJdRw_N)j`KHmR3Y6_a%Adp`KXW0`H)*>
zTn=4^;lBj2Y?&PI<CxV!tL4bYI;#@O`Q{TkyS@%a11}E$@%BZ63fSdwVnAVRu2??d
z4d`;n6jIb`o)rB6<7(@*a?vM!ayk_>gZ)}*HBCQYxaDM<2)v+jutt8$7pas`r%!p4
z3WHOk>pc9hi5PkLbZ6HOO$HBz+-bJo^;OX9eJ1r<i}W?8B|x*9G5saC%V!gynZcxE
z)sEBC!tCVJY4me3dKE}}nerKD#OQfMRcGb%{uOz9W-nr*(H=G<U{FrJVAwVk8R^dq
zYy*_)fzV%!-*xp>aF&s?VgkS{I(D+e#M$=wI1taC={1ErLUEOxW9x;XrXq*Zq`o#}
z($u>J83?fIl`ubk$-0?%d7<?3ZX(OfGTh(ib#`qLEa@!CTPD4&V_@e>kcTe^&>-~l
z&Sy1xcx0mEKGtqxMs;K78=_3gF*2a9K>?CYKj{|(psZEBQn3E^7U5`lxxlBy!RTuP
z1E&m#qc|#4i@X~pa$!JgI?{Th`+@Nr?07l!Ocy!Abzq-JWgujjHe`^2pzlQ&$X8-l
z;V7%sH1l$?y&VSf`PDR6svnW^T`uwY(QstmAkFeMoB!ZSWNErIK+4+^yb7IE2Rq2@
z*pj@bXxKoYnUKHoGSpG_?zV5YHOesmjEWQHa%xXEJUrH?*N+`5Fab~t*u^)(y2W*X
zyS7?J`f~eSB%=~|x2w8*6?S~I_ZF8OO637sEu>MJ%xFwPC0a0SG;^}MtaDhk%xi)N
z8}GpjMk}()V-rY_2pWB&MxBgHAh8ez(hu94@!p#4$a*k!#)QthFNO8&-2cu0WMUAn
znTQPidi*GY<K3hLYiij%Bxc6~+(iYkMM&cbb9Z6Fh?y022rDd+%N=~Y1mGzSjAbHC
zn$(T^n%r5H2E8wqDTXIS6=eu5PRm45G<>xHI@c@XpyAh6xyq+xIYh|yt_MER=5e+4
zQ^<(Lsg5#@C=gXG*F>*+Y0#RzMXpWTtKf(3fLTx%fQP11rCF{^<IKaBcjT_G_uiP!
zp8=k9%gqGpVVz!$1Do9*cvuui5h-d(6N_V`DSBYUR&aHc@SsxjH`r+C?wX(y4!JcO
z=e-u0QV{wpr9KhdoM=Rvvq8#YU#he9b+u9+`+@*A^D(Q4C&<adV!b6Y&EZQ#Uo6uT
zte9(#OvoCk^pk>=U{&6jCA^nNb^DZH)4?s3OR!n1<tvtfG}hR@)MilN<A&5bgd$+V
zG6+(&-X)XOwo2Y1X$1yI>8O-q;!1M~sf+nnR)AT{WB&ElXS<Iih{O!<ljT~r%1i?a
zQlhCCKPx7w$U}~-MH*rfJK%jHv+bAW8}BbS#+0ztxbQW%0H;SgGcFpr$={R9pf1`}
zTu3C$f8$i&Y`wubf}->-KHbKNlmYa@mT$y(71O69Bq@I7>l!NMwhpMJUM&#SjGs|f
zK<3!yl*2vnc84@?j5`7mMk7+JQob2KhvsTw+qly~09z4=s8;2XZ^gub7T^NaLGiBm
z45e{!gM2%#1tc*mnORW+5B|FozG~5E_(HV+L?m$ENn@;ll7=|GN$&C1h_ZoEo!W2$
zl&!yO%a~CyL+*{kG8lCNA&-2|J2IWmLa-4_<UU8q^9RcP=`78qKE@plse5@KXxYdS
z6XpAXIN&58C*THuFjg~x224~+gBHvn2#-62r`t3f&+_uav=In51#P0o$YRww(oBIo
zLN=v77xa-I*}$;*))&hoG0aRv&4UjC%Kq4P7%0u~wMYHeVuTv3mY*1596VYc3p|*C
zYNGPf1kUBKelpr)ku)aA%u`PrhD6NuXD}zK)sPv^<~fM`&cq$!cahAGv7r{+EDLPj
z85Fdvm4!Bu`g&R9pR0l!Dr!VTU;?=IWO3)+AES?QV+KVip^$g=Q<H(BzN}1^1c@lE
zK%TZD1cJr<R8>Y|yk3??BP|A2R3Xa)1u7A6i5SHdF$3ld1Xa{5IvRhm(!1kfFhB6s
zvdU(D_Ia|}_7fEks$`9|L^^T!_vNA7#eT2#udy=vCPP42DeIzLo|QErP$JLI`VL2&
zE=}fw46j5xin5`z8|Jm|Qlm4?fiNzP88<*(4==X0vM~liEI+jTL_*nQkUR67Ua~p*
zM4SLLw7^+Vj~0VeA!RT43uI-BZ{Q+O@A=Z|a5C(q`wh)9JYQRVrS$k>5GbSkw{_mF
z9_GyE3QUFL8rQnrTUsB8@`p<BaqriX{}cJZt3k1~On&Cfj(K84WDIX_^4~PH(NkcD
z_vyQ&cjgn`C#7uOanOn4FXH5-&0E8bS~cLz!MoF2>uQX!B0~CBwBBVPtAs1d?(`uo
zk?`e#t@25KMte1RawKJupL=IaxJV;Bi58vi7s^xKnimfl5R~DRUqla<o|Yb0@^l($
z8R)U*J(gb@vb8lL^z9iVoV5*>v+Y-@Ig`s6(m&fi6jpA+%&&u1lxAi?;9;Vr=X|cl
zwS#ruImfs3H>uO5IM2pZKW`_P0s!=33BuH*b+jbB5aS?e3P?Gr;qT%Ms$5^xATL_?
zjDE~Bgp+*B@9kvGtdy5h-(<t>hUV_J$R84L+bcesZ^O%Ro6OA67v_&Xf2#OsDQN<w
zdL=#vU7$sc@xyTXh7$5mj_70)n&ef#>N1%5pA7+Sy6#4KEhzr5$;0F?-an5u-8Z&j
zFJJdFNq(HMR}_0yPQzHdGX84)!ZJ`IZx}@ENW*te{wwBRn48{^lQ$jpjjJcCLCKHS
zL(^s0!TX!_crD8-zU4DgEX0aglKbzSclRa<y}RvyAvQ_gb~qXAz;uEX*?9kmLrsr#
zIxM1$EwpF`qx!m1`Df5pkijx6^IZdM8S^vc<zGHU<|UwK6`V`uy#OTKhZ*p&9=rT+
zpMaf5dY)KSgeH04H;TIGmGYk$4SF_uJAGhp5wBMPui-thU!zg4Yp>0*@A((XzL*81
zLb6l{=a>D~cBp#p9!UGhl9G#HHQk72q_8Tb$iXyaT2;*x<B%cm<l%DI_Jk#O<cPPV
zX@;s7rk9c~sy`_P385Qir!J~5Lz&Q8>D<MA#lVJ`t34bSmB3rbN|!DXp@t%HW%Y1H
zkgnESTO-N|0|MQ;IMs!ntVA7kX>04=#ofx3*Ifc+-69`~XVj_dO|x8%>7wc>Y&Oy$
z$cMYQQcTUU^ay80ZzdE!*1x?ErjPsKhj2f9mp;-(^?JG5B00`K41!LU#w?p3wMK)B
zP;WdYJ_4nCO5|hqW(M-)pWwaY!~<xH;WxKMKHg>b89?Og{y+Yc6Qft=uguCPtk-`1
z2FOXawzkG$+2Tt1q>n*T?14&havGPRH}v9KDW3`$U`HrLaYneI!FZq2MRk+7-58tc
zpVA|irZWUX%Bfvky9q5B%}1Y!Hd?maJXB8e?hVnY0YD`;#N{rQ&qix<`Ff`#Ev~Jm
zxHu!9i`5jvDMQY%xgh^3#M95)xTQD()wGZFIBE=Xg-oUkg@}B?-^8-oZ>lqGbfjXh
zhh%37PQU0gqOG3N2}?enbyVLZ(cxNhc1#e72kNC~7uRsX;}kB}I2q?Ss+;E-X8=@Z
z%n;^AFSK*x3*1$f`s_<FKIX0$4iK@Ik54K|DHRDO@jWwypO@Y&(=)mf$~c9u^bQos
zD39~yIDfsSN*|jvr4yj!`EeTExMx#Lsd%NYPsltHw?1(nWRQOT%rX;`m)k7;ySQ2#
zGO^lWF7QTN^(H@h0}?za+gaf<z9;iS^VEegYY<-KFdyj687fQKNS=tBN*~Sb;E68k
z04%yg!BbTR`7<L%)b_2EulO@M6wse&8929#4b8@8%jB#6nkp8QVp0FBOX3r_$I^Xq
zGmhui3<293OF`u?wW*@)mqwQiPAgju_CyV5_=`|I9BtSiWJrKVU8w-~q0t$Xnlj8s
zQ%x1VV;Se+{>)O=+s!n>_d85_Er6ot+m@03hHzviFyu@IfnyNVC_5>2xdMcD<E|fV
z3!gCQG8to7ab*@H_OdRjO{znsiC8+W1l7*KIzKKD^)g^heev`L8Se|`fNopICboz3
zP4Ee+^D}^eC?mgKC;A&8GUz|$2AC9Rx{$A@H&jmZ8>;od1IwfsKU<B8Wn?W1IxLv-
zavYb9QW*1xn>raF+IuhWvioFE*026g|C7nK7jWlmu9hh=3G5#IOV%h5m~QlnI7ByI
zDpz_dqiM#i@D6R2s~j*L7tW<p=m8m)s{<2CVQkC5Np))4=x|J2C)d~}oilZ+Tx%P1
z^?7n#S{U3burNAVT;C1`5bZEhGKxH7#~fEEorsET?sesmfQnN;@EQZIzzIl7j#`hM
zNmDas+oy*^NN=;bQtHphpfEZC!laZL5T`N3LCS4#bxT_(72X^g3^1eWCn?j?Bq(T@
z95UUXx!Dgg>ME)9XMyt6ELQn5<}_u=0gy$hj&0LtCSe;RlZ)06Ul1G%i3lN^!DAr%
z4orcpmUDx5z=0^sfR_bAGplB;bpStwSrBEJh`CPk-l;=?-e6J}1JMHkG=n%(@6Q-r
zN`v}(sPkq7S;I5u8NfNTSGJuQq+t>gsLk>YO!%THe6=+AGnyWV2=tGf9mrWxAvXpr
zK$%G(kZ%Y>Q?+rEbv+zO%q!()L$VCwQ*Q~RM1oNeiM)IxSS+u4ZT8YzkHQk#31LHm
zh`dT}i*>dUr-<(@GAF2cjWW=;8zSF4Nbcxj>V0ZMU>#Q8k6?!njI!ZCKe#V`vwfAu
zkKmTMGd*J(7|Pcx<y*G$F$M5^xy$}N4NeI1Z3nXkg%A_5``lL0A;{e=a<{epJpPmK
zc)y&*B5XUdzso%x#-f&jo=@NPXAn}<I}GYfxi^@A5@dLk?*;5JhP$pf-McTqfto3(
z&33>296D<7#KPf|<N=#aEgcoVAG5)pu@5pK587t+z2JQLfiH|odx*ji;lZxq7B%Hf
zi#!zg1q@_T6(-K3BEzzlw#pCFB3hN7K-Lp^*q<%I(pj;Y3;R(_I5U%Zk{$`PECiqT
z<5)Dj8nW`Jy&3~fEH6K?mB1Bki9F_0v6E`2lEg(x73lI$(^Pn5^w3FTdWPzmn95R^
zndf$K?GWo8n94UV^BmMptyH%Fi#XKn^Suj`D=ito0&m?xh{u2od0yymxN0H?<8=wO
zCRybDC>`+WE%s*`w2EiQKq5^B4P*=B4ZFmNA0^p^veZGacxT6iI!y6Ik$$->OLNuo
z4Lj!Jj#!?aEq4>b23ZliQ3pkJk*tgj7>uH3vMNoq1#SUmhF|SxeHK-bYwQQExJ{_#
z7du&5Um@$FqiWW!U7ygGMP(O5IZO9vs=qeG_+yYZr$L&1L~1cf1Q^)8F{lXBuu(YO
zlzLN4Q-h32lAELHuq?y|#1^!q{<Pu-IJFkp;-8~x0+5&6Oo@dKyRCr_%=iVdShd_+
zy^AQ&LFbumvAV)hg9~N5{X$-E@cZ((w_k)?48#Nr*)BgzLux*M9Qxq(cLXXsUjVlS
z>>w{M!=fi*B)D|dNt^d=WE+SLi{H**VhEl}c!<ibv}mT%GbU)ak;zdL<;m0s;E%CZ
zgOB`qn%`oLTtx;!$JM92HNnKv6;v+&BH#}$CajNoCgkZvf;%Q&;y8Ni+kGUzj9sG8
z1wDv76CHxU!;A}>N>5=XiT+huN((&)K>ZfWvmqRoLjMKE<JW156am$pP@W60mE=&h
z?KklS?qx=oJ|DfRDxOPQCcm{WpctTQ{0qJ<cn3A2MsXiS41X6r;BbdP_hKvy7gH5g
z2fq*8M(`;P@JseSkF1qe!0Hb%EqLEmB7XAa*vYXOxPUPe)*pQs@ZB)9EdochQeKIr
z6;;62_$O<Dis8sQg)kqwNxkae<ibp{qkW1O&7bXC^aQ<>ycSE=vHlVi1(zkm%;8O~
zxI;%P=1hOx=Yp)Hs#)Dgty~&u3xueSE-7!smf|)CF5?2Et)1-ne;Jv8QVX}~o8F8!
zB-{Zy+|z$E;N>wo!sq0z)E{H6xUz7^`Fp^~bOV^(ZSuCQtO86bF1~*RTyr%w$XkQ0
zcqi6^dkHFD+;#KipFV=k)y%v`uirgtY#O7P&x@w*e|bM<pf<hsd(k)Sm#mfo%$I)$
zYOu&PC-8ot2Ao=fDgTL|C@Te{;0LiJ=;-OzKu!Oi;4P7?vbq7bGDLLRSI85f%;Uc3
zWI{3AFlB$NdPc^rKL^rm)g=`;wLuQ1XXY(Im%2kilX%>S3D_Ez5HhFFt#a6|;wmV!
z<w%DOq_aD*#wfdVb#_b>7g0aK`>(y0gofrwr>-jCSJchU)-u#cI(IdF8_zNA+{~EN
zrK{`iP$#tpuGCdE(zUCq{)Xmi8l+oSQ?^Cbd#1C&cpx9$yCy?QHm`-ED7Ad3tEuW>
zS%c>J7@sHY&AgRZbIrn24IOQsANFbSEI?j?U<hxxuBLLdb4G18O{WG&>a0J~)!7Rk
z_y$S@82mV&Y`Cv11~M7IKI%i@m?#%Ww}InhSaeQxMk}<BA4|`W*%AYC8&B|F)mnkg
zLS=C*g2ejcF$u;G)K(PLlf0eS)gwxC@`>n;{2-|327v7(e};uoipD*;Kt9>w8NKV^
za3d%CGc~7UR0f!R%78oE+!9YoqchIdD>{7IC(AR&d$3(obE<d9(?t;k!BFxUe-<Ib
z8X%gMo8o)W8u@JGv%V+hkkx@AIgn4c>9sW?i)e{_F17w|F*e@%jIOG_sIHOE8@|CM
z>;mc0Rn?d9LKx6rzF;pv!83AZnkv0ts4LngU+n6tOl0H2!IiV34ODe&F$@T!oE^(i
znT2VSY5ZV<@}AzB6`lwx2D_gVOPE3<%elS;yI@J?kG?^_<UN?^Y1XV>wgB88bvZfD
zPGC+`A?6t-+I6=DvF4w+1Lxz*hAUH`o$8b3PIYG!G-Ipe{FpoDi!eu6BB+ey-Q71<
zo}HBiXCnQgCCDxF%URJ=GQ|D6x{8xA9q??clM8$#MgXg0KSmnhkgh-w;f3B86Y@nF
zV`#e}G7j`F=^?{|3?-XfWJ^R!DR`!^EW)7l07qgv4#Zc|S8-*|m5XBxB?)?YIa+)*
zp$%hrd0v@KhjvmBOE{if;$M<jP?Nw25<BBxi}L`BdqtU21StJ%VFk=BGdMuJ18Hf{
z9qS1W@n?(N5`ue1hT1pCuTdhy0(>wTR-k@!i43<r>(~DR8DanId*MJC>HX+b&^L3<
zj*9Vi-VT$r*}q5oGm;jQ0T;>`>jpfBT;{7=pu7>+%?ju!swQLoLpyfdkNGmrun7zi
z+@mXHy!A6-%6OUJP((k_$Y~@rV2Ql-fF+7y$i$c$s2L+Up?sIG8|E@xB`cG>0d-c)
z4ndMske-n;7zr#)9bt3)+2wJipx~lMEKc@jOF;FQVtoPTFeN^<s}TI@X1T(jkp^;K
z=L0vd^k*gp!W9YZ)m8p1E<HL_TpgriEcEzO4GY_|o?H{7fzh+%IjiMbhqj%UFo{C%
zAlJqDq9;*Vtz2(k!x{$WAX^SR$V(>83$p}tIoG0ScSAkm3~(nw?^*2a2t<k=P`*h?
zdd3J1-VEy{YiL&kUri~EzE#O&_+4hZo%74CJTS9f)96EXtB5B|uZNN~aC5URZ@&4Q
zGiAD?gz30h{G8?XjN_vdUS$jk?vw^KhSpUXMj?t{s{J)Bp;U{3agm%qLyiqSt8_1~
zN!>7o2(%#(j&BHrD9e&TN6u2I4FKosi%Q`Cly66ToC}i&jiA=q97bF|Uh0E7mr&A}
zVK4gh>mxILP;>M!{&AMAeH2EN2KzVClNHPCARsQ=lH3}AZ%j+1RzYDSHyKhvfeu;f
z=7cXi6jWSL1OPtCEwO5<*6@tkvA)rDcOf*`q+UB2g#hyr-q15N?sV$RDIIGJdXj~x
zOJz>xWzY*y@3jqW$lBp7tKZ~iK>5~mzLDOTZ}8si(dt7ew>fw{kD|;m$c=z)9D%X7
z`#OwR@oP(+woUHvb(s6WLZ~)Y%Qp>pP~()#oe2Udzpw-ETS4|wTSP5scX?BGkK=)5
zQs!Zfz;FAGW{c$R1UKHxAU1gYo#=^qw%|c2_r$>-jnu*4^)F1#uXmW-8<amg%Zo@M
zgz|eGrfLb<H~>mp$$j31mWRVW!-6jN+aiIQd3hiW!2(n!^nTyokI&A`%Y*hFS6)3F
z1gR-4ksrjxj({c%{!m~l;^d0thp9tL*|-vucsS+^_c&5-?d3-?Gx$1_&t3<K;Sqmk
z2t@L|Reo%<X=}W$mpp3Ffu3kdnf%1@!t%YbOa^^OsrVrEStgJ9dr~XnF3tZ`bZay}
z?tq^<o@li(X|!Qb3pX!w9h^ZjwkR$#&G5YR#(ba1<eNA;->0rNYv2j+SQgj_?vA)H
zEtqBrGM4jXQR<J%YKW*b=h2cVi~Wt^t#J*cM@`b?Z!~62&jZp#sS<2S0^dBiy#R=S
z!qN~+c5G2|3X}GjPFoQa-}8(u3p^YLDG*NivfP2!vA&UND)9j?r4>FDv`cF5h1pY9
z+U^wA)<dvb6-2KvQvn`XRvXe(!b_0Z8`cEUU@zy&+UN}ngd9zG2w7(zK%c7766?F}
zHc!NSu7=f#=!ytj+z{B2pCQfO3t2W`6v)YX*%;?({J2T7DM6c&FMjTUG3#V=G#i>j
z^7tU(WFw&o0gi8p_e1c$)yFZWQk~^wYitTgpM_*9%HdwwmU^qwbikPOcJBaeg(|<H
zvyjIPAewi+S$-Br3~mdVnX<!Hi2|;D<q7-uP}+IgqHk&>=t7v6odHTp1d8ulWS4Jd
zh&~KF3yO|{w#d7rf^1vk<#m~Ic``cH%72(ZgZ$j&msv&fR5Z;L<tpVDam4<+Hy7?y
z=mzBJ0637H!5<7BQ35Z&j0GZ5AH0m8Nn6L_juij5$*=rl29r?=q7I&(bqnR$*z`&D
z^6Lak24lkc1E#L$V!cQ^K}4YOfj*@?TP*qco3vS5RaV`GQ*0;bdFy!+11EnQ7>qDV
z)(3bYMuPhWPaGw#-}$Z~dK5OH7t?ZKT*HXl^lGK2%I{-F!T%{Z+M}Fbid}QZlyx{1
ze@Nr41pOfwyiQ(@c2qS1l$3h@XrL@ZLP)y7zS6!Y3T=&12HA7nsr<<&xj?T%UdgJ}
z!<l~7J2wpkcWa6KIV}n9rx=hd9KT#Hui4F34Da2)7!*cU=8C|YzaDc0e-O6wza}b#
z?7-L&IK}~jw30EK5iTCcK8xiIM{NWy%70lej9bLHL6|zcdEX4o#DiG~%7!!R-+~X+
z@)(qS-ijHcOR6>KIj!ud{5`Zd)zFUB!W|brl&FXvJ&XfEsikc3kJ$Kv2K=s;ckJ*Z
z`zeC3{@Gy;V9ltXeb<qo5u4j2|4QRqR;Tvj((gH7c)PKv;#v@%e<zqTs&O%B>uPyF
zW~52bS+49q0a8#~vPkqs`5--8p^hhTCy+gHI#KZudcBzHVB~Bdb~MPo*t8PaAFC{C
z(4sU4tUq-Jbp?+cOtV{|N`@7<HV?&jHTi6PNlp&iruFU9_k1~G-G8;mmqog{znkB?
zbAY{_x|za^X(CogXK%rTeRLV3TS=F0s*Y=GJbS8i?dHC3OtWN*_Koc3Z#1rf*99m8
zfPxd)y_>33kR2FSjt_M+y&9QCU_bt<IhukARR?lRH`QL<d}B}fu+PQZ<!ClvC&$L;
zpg$BJ=zK`f^K+QX<s<Fysqh1dn+tS<OR^HbCONK~Ya0i{3r8;!`KS*Ag&IXw;-(xQ
zLxXft7rh#w`IvzLAyc(-g3Yp`wn9GcP1T--^V1BI%ZUj}C~t_FvD&zK2!6s_!w^pT
zhQOE2a+2);15co>_+<3gw!+Qgt#Y!pXlooiapG|Kl>K4>ASh{1iQT!XT0Y%P6|QZK
zLp*)wsoh-pGAYX_Yh0S2>84t*k;CM)ZmtRf{|8&`XJd>!i;10Xdj_}ai3RdGUk>d+
z^T@)eE|nI(GrH}zdIcq@_Ki#Q^DzM`#29qZBN`OfHc&VIg@8Y*_dz$qBA91J6SzDh
zy`FqA=BySEesq>Wy9S;wXGi104B}Mkq-V@}Ob*^(=d=&6weIEIv=NNHh`d46-+sx!
zj3A1VEc`)X5^RuO(Hh9h1aWj5CtJ2e&a+_<m%tmow<DRw#?U1}BRhbEbb)M^FZ&iQ
zWPAsEKpxsY2KVdElk<I)ql6#G1#_iuLSB|LF~O{Z04DunvH9r;BkrG235f}?D`Z12
zNF&oD+OE$5_7z0oWpE0((0@_sxK#$)tVSVD)xRX4M%k0`K7*n+tdIw21FTG6vE9kF
z^C1`etZ{zSzR5Uxvn$y6UroL60=7)Tmvq}*0@1YXf5H3u|8HCQS{#s}m&v7p);)Uk
zlEFSPlRv1PS|URXJka1Voz74L&(L}qmfD)rGsU>!=~!4m18W%B?LQ)p796&aFmGh+
z+Z0|_qXNHCyri}P9HNXi{LWS7WK04Ua=rkJl<hBb0Gs}aXhCCbdkF4WThv6>wUBW!
zQLt-tT!eq3j1Rop+0mbXDq3}~kO{GaXw965v<VqkVTz%N@g65%-Jx*)DV0J&Mgl8e
zk3)^*J~$#YUrma0igflow*>)_x*V<ig|SR!y%s^*Y?aGnZ%VQiGC97dUk=XslmHqe
zjB+glaYaA{91VF;1RCg(jlkn8qlfX#r=OLpVz=BgbTu^=DxqE7O-(sZuwKRjtC!JJ
zrUo2&tteZxRj#oa7BJnCTx)0r%~o%@&gX(d2%2p=;SR?^uJ_K!q*2qc21hUB&rG*Z
zjpQaNGDH{$+i!7#sYXyTy09bwR0N$w)_%ZD2d9A2K(Lw|iqVz@!jGLS<<Sii?gKqY
zO)G4Wa}n({Eq0|yEtS*l!|}rl3Zyas4p$K7P|i!0PndBxJmD0GtD}SRh8pCQ%y|>F
zY669}<MyvfpKSzkF&pkmxuIK_4^8+;?$Ll9;6}a!mvF63bb1cjIa@7@1mvYIAV<aw
z6JZETeV}Luw*?C^(liq?Bh6wB?_H3x`7$$Llh43>h&Z~e%nA^XtFM-Z7zf-icp;p-
z^)frYLBdvo1-HhH_D!bt3aAlpig}=QIK!Be5dY4Vo4rqBf|enI7X20jVgVUoaJt_x
z#D%*@9Y>H`6XF`8Z=f+<5%)F&&<rRQ=h#148Q^wrZ&5MG=m_S++TRgJ0<1dRU*wxU
z8MQ><@Gx`2orVJ$*0%jtELQWYgI&tYU9sUqkjgtN-;U*iP?-Xdu5idkjRUz7l;`e%
zCn%kKC-AZaDz$(uv@2B3tp^;wo8Hfod*dc&0Tf`ae6QQ?bE(Xv?<tf;umCeqCZ%7p
z75WWt%HjL0yC)i<aJfHVqB55$M9qQpfq=<ODErDA<okA%s!(hPsP>>u6aM~M=ueHj
zKM1I3(l!lle#n+N8Ce^0^27EmR*rz4t`FN>AU0*>N8Q|R4aYQS7~M|vh_$F;4v8P9
z9=51RM%(A3Hbjnv`{XCK+kFQPkjLWaBX?1a{M24tc)<nI=>J4MnYl6I6ZeDLMo<b|
z^0&!6YcO)s1eqU)9WDx}8V*bLg0w&Dus_VVvQ8EnsGj&9825XTgJ+SqVVyEML$SE*
zVq3=W@k578lT8g`I|Z^OX;HK(nFG|)*kx#-ApAARGMiHs{3MnKrsa`|Ix8z;2f^&)
zVkB4QoYhM2SToz$?Io*hXyC(k5R%o=)s#`gaX{8MV%!b}c5QsD{P;pRv9!v%^o%JO
z8E3pw*84M7DnxuGuy(s_h}naBYHV|}bqs%<V%cc_KBxCJJ!O-VyKpr!y$b~>HKyV(
zUDr0-+EL$LT6{ES_u_%t1Y7f#^lS;(EZB(ftTlk5(%3?@3y03Ff!1&@B+=g%Xo5nY
z7s&SbHJL>3apQ6ZT_->DF*qViR=~SlcBC;D@4Ow)X#F-1fhYV8j}shtL3zmrleuoQ
zrA#QACOZT7V94@-SD@D}|5TYSbhm8h9H&vcZ3=CT)ySelH)~MXCu6#-bs#^tlK^yw
zd&pDKoMFT2XPuW{baSnw9#BmaN0ffr7e}Uw7M}IE0)FYwmVig6GwC)={u!rfq+_%g
zP}-qkfClhQkhOnhhedPCkP|^I0`EmN8t9qEz+VU0U^-#wFT-Pu3O|>?M!gNP?Sp&Q
zFbMgLF(QyjYLPqHD1$s7i!XrIQ-141v5XM4HCu5BzmO)}Zj0fq$P5|_Ar$=1Nyki9
z#(B~3^5&awzD|DcotgO&^Tu`kQtS*$Mq%E#Vk5@!52-a{;c1s$EH59mmX5(ae~iNc
zx{5fd8hOPIA^k4NApXfvgcZ?V4eEo!)fUq9XM5Rqz(w*}zytx=5Gj>f%3l(|8GlYG
zdX>B$hz9p_mh`PwdGoK<Q$G;IHw+BCW1-0kk^qhs@xE{RI&_Dy8?9I&m3#B#Z|y??
z{hLrxaIw4<Ynxc!kf*`s@4f@%TU7W_#8F32RcgF#dpI;#g-YG>5ATwy2D$;zrS~0Q
z@kImVpWSwUTxBDTfdw0Ym=C>p-c7((?*}mJv`#P?#=oq8T03!%zSrSbll(hYnnw<^
zoV;%@3t&eE82%G4S$(uzK1f^4;1e~`${qvT)N>h2yEip(k1>$^eL?lk{<7@%p_z+q
znH=y5(*Q(VYOnfs(BCv6R*F_Jy26>A>rfhDE#M8}i&J{opV6p9n#ts!M{HE2GOrft
zt}3jGQdXvyPWB5jJqcgu?yjLIq@PMwx^!3l$JIThYj<}>gWv~#1qWuEbnEVSGqM99
zU6B$Xgo%1fYO6^6QQ%ZQ)ZO%0AeFe~`4@UksHno0dyLHjkzZ6>f4IA;%}8N9mn9A_
zW37(u?*2xw@W@Bv2l=?;V&>&JoS0hqsLz5|CmaZ1rr0dU`!mKy67i_k{+N#yl6<>;
z`2>Gsnb#C`ICGr(kGHRe{m{7#y8>G&{<6}?iQUa3Z|vp6<P&L3&3^&6iB>tOyOUfs
zCZ?1WpNuI(7$r4o0T*#{cUL>r!B~nEYu-e66(hHTTRJ5@TaJr`^mLniI-pUMfq@q_
zKI1p+`5AC(pNXNMNugqGJ@D_e?xFq4AsZ|7TJqU=GJF;ywGoLer~7wWrh_PtGygf;
z24oQ^$;ufq8C(a@__xu^;`6ajXdyk)$nz)>s|n%@vB*haw2@*h0FP^i=L-HbXow3&
zzL<umBQ+#PoaVE-yK7+{3Af?%t)-sMjv*(3&Fh)IvNTqpG;&T13F-%rgL8dHNx?O!
zMEP^|TE$(oLcSF6D}r+c`0HNLKdQmWdEHeB+xMb@(mS5_>C;EPoW|vB6`7@vAvzB+
z<goL*yM7hQynL>522_B3tp{R8o%D<Dh9T`$|L(4?M0skeC*^`Tesr6GTFz9K1AJ7w
zUdfxRk_+uqvpLe>J<uBggfMVGB!n~NMX}p(+Q#l<BL)R%3rJJdU|A_&aZp9}BPzRi
zM0gO8Z$>UoE8K!3_f;Q=&LNB{-3G1ECDvk8wvvGpUrUGGwA1vR;LWg1E=@<Xg=7Fo
z%8(A&S!8fv2n9PMg*=)#T0`vUjbPr9p*}Xk4piI0Wd6g_!Z(8ufrx>O4G+LF*(guT
z2=AUXtr1C#;|NPA{LdE3$h4dlm<+>R7sx1Ytr<3W4+8Oxwv3ML8xO6+7(=03{6D!Y
zFmN2GGU{{&1$^QTDDKA{wLpvGrHr^z#z#9yT$M5*e$~MRKqjV{Y@>;ae6j)zfu9DJ
zK|Rj7Q6>dg==Fy8l(-74A4q?jfC+`(g6cV3xzIW-mCL;ai?42y$w$qPpeKY_yK{<v
zOYVk-OQ<pA3L70kAq{e6^aGg$v7B+Ph69I@r0?qZ4Lfrwil|Rbb6pDs03G9+<(i|$
zRvj`nA9`51Ha4;d1>VSNUgv|sP6z67eVjI=BCLSZ3(j#S#^V75+f@{FKqX%@MOYA>
z`2!Zmd^9i+PRHoBq{A^R)l7u6z-6uHq3|y$qI-YoQL`p7Ap6gJE@f%g)V#|Z0_&k!
zwA|JMT;_p?>`~$CxqZh1)K;BlLuBfi@EEI|ZgeP9RVkG|Px7kdK1uSb{F&ZiIEqZc
zP#x{Sc+sf@&Kr^om{jIw@HEs|xBdDJlpE5;$Et}M9f*&Z=2U~Vq}G~_9#JUy=vnil
z*bM3n4}fJGq&`MLk~jJv&+zuViz4D?W>6#~N2p~g3Yq1wvRj?K@*byWXhUo+ye%qF
zv<UQZwlOc`^K#=+ODBW1Ty9EWr@I5RUl6K@>PzJ2pm3lYaO8(Q$}<%4a>ez2i><N-
zk?r4zS^HH5PT+sA_uk=kR>!;OKmhkf2?PiU9aG)FnC>DMaKp0cZA*Kr*pjwTu`OEY
zp|{YhTqXD3Tfj1=21p>(5K4ehLP>=rz<u8-OL9)mbDn$8ea>I!cTW7EZ-3t^vu4ej
zHS^AlT<M=9Is`I8LFq~!o+}8mm#d5l$Yh+Ct9RPE8?@K|9yNW9O^GS+;0(=jO4r)A
zIJAo7x}Y$$03s7maY(MWq4hlMJo$MLxRSZDvT}nlQ%boaH^vJv;?9E{yD8dZ<^zay
zT!-Z5m>*PSCPrP2K5mIMP}8ub4zRo0t^Nx&KP9Q^dgSEWVrLfT>!DY|K;;dSBtY}D
zY`{r=hjTZC&=X7L&Ugb7pPB@*3U|ej!S9TM{zWu9uuc_!x!dSdhzpI}W3(7mFiP$<
zK}_#>mE0GNR3jrX2y?%6+tPSu8H@=JSPQ2X3?45J#(eacVJ$<_8G5xmw3C@+DLvgV
zj$cq^7ysgdrHs@h5Bv89=ymvpYJiWl*+_IxfbFnLQ>2(YY6BkBzz~JUVwSklAuniF
z8sh;f1Zvf)X|W_cH}xbWW|mnarpFfIEf2ng8krHxjOziG89DDzNgz#>I59Ow@>GTB
zo@L}X>5!hs%WVIW;-2kqfTG#tYetj63~5el#)5r#4`OGIEyF;hLzo-mu18d1HF7`B
z^Y`aKg=PgP1_;iN<v;;qWL1m5G^${s_lCGQgc2>X6BLr0&RtoYXlfBrgr;zMI0G4{
zD@)SgX7E(UePW|54e~_RcUcxsYH47mJ!N^CHgETOYmpUc;PX_zGKb<yf9B!P&>hyv
zsyO6vnFa2Db^JLc&xFZqY?V*1r>$abtRXCDjBg;Q#kSSD7&xXuOb*b1vfiKB6}$-t
z(FWO&-kYgc5ujdWWAum}0Xy5KASV*ORLC!FY0ien=y7|`ygG-<6W$c;77&ZPQ!J9r
zK0tG(;XXujADC>ZwD__xCD3}=5))~YBOLc4*gu}M`O>ISi^79;$x|*-ZesKxGnGD-
z<^~?dGbBZmr$ZHm3kN(8%LX>aGtOFrhs(1uvEBpv^^)h(@E6iDhXZ1fJRjS>v>u0a
zR(@rJIW<FB@j@Ci@-<_=j(~&S^I|lm(hfTndd5rs3?>O^P*8rM)c)FAuy08*0*8py
z{!RP>9}pE-LcwfdU$#NlG~nV=A+OkMaPgMkdIuymQGUGpyc#Q64_AIxUP~k9mZ#K#
z9*^2`Gvsyq7;!xSQ)Pj@L)&k7g9^AngM?kYX#_yB!OSFiE4|PR-XWZPrTos$&v9qV
z+xF_@a^h)!@B54@Ju5=4;8hI5f*y1xCw#|$=V^iL3vgk1HbE>UqbJ|>u2>0=@rz7Q
z|6VK(z~xNh_P)PkbmXmUjr<|ng`8!c3{;}Tyjv6LeBcx@7h&XkfZbR=^wxm;)GR)b
z4DwOv3#aBFL^H64${$VEq2fxNd>s5{t_-H&Z;(%d$qH!&`!uy`#pTIlpZOS#Dzw)!
zHEe52gg?dHY4}w^y;wf?XFN-g;2bO2BwvK2EJmF-p59;D3XIfxYhU?H!*xpZg8sMs
z*`|Qh0CVLpG1}e(fVuyxvClXnd2!+WTWhC~GgPOh%iq)B8G)(>CM+4slwwdo3ClnH
zcglZ?R?lUs_cd<suf0tg2M8c#UfyqFBIE08CIPV`|BQ~YLpbWSI8Fa$lLL+trQiN-
zUyQ-s&<uIsre=BB(@+!nE~GTB4JBZhtb88}26zgLlxm9-X{XkcLIr*|$ad}2nsvx2
zvVA+V(9l=%gLWZ^XqXR|Dh=~Hv@@F*1&CiWWk-LTS6;>g69#C-J$Le74Clf#y$M91
z_S3GNTh(w*0mUZmQ@e;?LL1C9)ge8b3z22=p<{Zc_Hb%d(#hJ&SHQc=Jo=se9lL#U
zPp*|NHoKxc5Ds12xz%7GBHXHA((D$Eax-EBu|8n;cBamlEH=vlF=;gACmXHwNL>*p
z#>5PZWoLT^a6Q;cBgV6`i@&EJXe<~#xw|@Xc#R@pMom@pUdwLnl4%fJ<)9-71;iae
zK)?d#zeA}6LGdpQovlJec5i2hJX(gA${x{=3ScwZnb!UzAJrTM>vLh2$)5f#iVA_W
zd&RgB^P`b;&|da7a^rdk7&VlPAA9@U11y|?40ogvm}MVt3s^u1w4VfBfY4+}G6Ca1
z4K^rY6num1Yut%4x<HW6lKs4Co|TjumhKfiFZ;Js1JkGxqve2Tqc#Vbm6ZdxnJ-60
z$QfNK2ibTk=ore$!9F)l!r-o%L)vZK5tk2`doZz{sUfHrJWt>dp$ge7hx(^3pKwQ<
zDTi6(EsaGru<0Fc6M>GS(bq@Bys7~`&C8MgJ(-IjU^)YjvaazxG}2z}+zN(Be_Y#n
z=j<KitIk(5K_P*kebN+QR^%ixw!SuwdKRrNmVQA&{HAwEhUBR4|7hD0WsK84#`;FB
z5Wv6tr$%QmH!rl<rE+W{1(RuOM<4|ck^w>efmP)wCLzZ;Z+dVo_7Z?!FnDj&e^<)!
zsU@a7;_57duviIRJHfeyRUVeeiMGHafigJBMhXeONlx~`F-^O|$01}-3CaxzYI$J%
zF}5r(gZynMOk6?S=$FgjSPG=f0D*?Yc&S5TW|Z^BGemXGl7e`BGD6EMWT-Wd`y&p&
zVctA3&vZg8g#B`O^j`{S=Lm1y95wV9GJz1%!*f&PcsVte8O04yKmt}wIW0XZX5o0Q
z{77$w#|M*>X$B<PFYLc%RLC===d2@GdUT9Sui)qi)E$#ZL0pNsMaCM9`}FH0r#tZx
zpbLFzT)gZlFi^)_3S(l0xIK=KQJPbYqnr@G56u+G8NSM@-e7&#0V+oMvRKaarcEhq
zdSX6qEoa#RBe7zoobA8Qh64z)7!rBDoRbE}N+;*Wbb+GE=5cnM*QQAZ3sHQZZ){m!
zs|RMST;T7}ZfF4qi{!%S3YRR}^bn3K7uir+8b^<kO!|fuB%#-%-J*8x5%9|+f+J(9
zn)X<?i|?>(`W~eV&^~>?Ie(iwABy~z6v`kaMn8Z^P<f-&K4tnZjR)hgR@QE-#mW@c
zgZq*QEl<4=Fs-3?GSQz|c3Zj}R?DOqC+u_-h^?re3V)j+87y50!b_!(rj*`?svt^z
zndDnno<X}r)>Ow9L6TQ!D;EbN499({#y$nowpMDrA6kLndd9!Yq%P=*-QC3hNqu}?
zg!1c?qiLGEYGjK2hI<<rZmKs2Qf{H6Zw1bnhIn879&(8@P7_L4TqTzpyZ1Y$mt5vs
z3a2f$6gktgZTZq&9#RbeDx}VmD}qE@fOrRsEx}#jXHHPNZ|aFk$p~lIl|dsWqsSo2
zq*ktK)4&{RN6?I~Hb!P33Xjq{xyD~2MuJO8_ujQez)8%few|IHrLhoooaFlSHegp^
zFQ$I`dCZ?D9KcpaFX(@0{YHPEwhX3#TMb9_O+E)cqs@u={Y|dl9B<=1$F&WjbEDki
z<fMdNB)7(Bs`8Wba+~!GV-{c<h+T{yffQONcWgr{xBuZHhp5FM6>bwOgNXC{uK2Ws
zr}-~@9E4HOcn)B-+}*CNgF7;JKsw(OpF-U&lY7%5W;D9|zO(~)g`}sDU>Bkg?+-47
zjosj<4_LQ=uOlQ1S~v3oJQ%HSz7BShhpcr6Gi{QGW0Mq&mPZm<7pb8XmsRwHJZJK#
zPY1br)Z)U7Dv#Nh<x}BbY>a-joCEts;9|_9u8hVlnV!yr6t}};TH(HAf*CeWEiGIt
zGyQE^AR}Ia78@d|VWyT@vEK%Rr)JxyP)U&7RhoRT^f1wQ$yknNe@3SV*Mo`K=lFCf
z)OgEbl=v*QD9`n`;m+49`6gV#=C#v|!$<uu{!ixnx4aNBj+jc)g4nkh5tK1-EeX32
z|IMWE^JQTy5zO5+vM4PB-3XQd#>_5@V-{*1)QgEM@wc0BO`(H;&TF&iEsc>F+0gd}
zSr#k<w<HoGW0{u66T~%Pzd^TH(M~T{6V8&A-ZrU1gQ>ZhSW<aa>a{gjg%(_u)!wT)
zwhdjrCgxXLUtBC}O`!EW@;q7Rgu%5{W%vS+L=j3@4j3?HeQJz%G>9x%B{rmIbXe-L
z$wq$$T?1Aw%m1}0gbE-Y06_fGUSUWPG<j5#!b~3z76sN4|6iVH=N@SIXaGI1;*6WU
zZNqFSB1&4KZE93G*%D_vMIwp%WMatm(6P6A3swk>EHPW2G9vT^@a1W14AI-b3{8?}
z?3Zc?!e_l18X?xobH?&4L{lZt+gnE-yr29khzS3NS>SoEGE)Y{#0xQ47(tQh;zer|
zS{nliU-B*(>oZ$^?IT5aAG1ne*;mVNtXtd<05W;mx-CJfgjb?lnjCZ;SY!vqF8wh2
zt$oL|V3Xw47y~jE!IA%35FYvnAgA(r>KE5He7^=7?;AccmOff8ZzdY9XY`lq#BZ4d
z!hw#gVj&NPVF$hKFY`o-i^K0@jmoRa;QFtVcVY}#_@kf}znebdMlws)d;ZL#0+_^R
zro8XZcqz7Q%>ZE;S#UX&5B_6k=sgan(M<W!Kb(Wp8}*-%foO(&WRt|LyBw;<AA?HJ
zlYp50*gt4)nV~2>)`IxNsSb^h$~ytX5D}LAlCjXA`tOk4+)GfRiL&E=^WO#XS+oh=
z;u@?lrTd@Kj2L%s_pHyOALL=BshVnS6Y`i}<09<~n{l6``^uL#>QUwJNCCk5m5Gn4
zGWl~%PD{7`B~F^r!{x6*dqjq?y42s2pjo0p9!&6H<M_L+>*m{@SYIjsh&4nk1P<LA
z`8q}pw=m0bGt%dqXbb9*2kBy-JOut}BdExOd;b+Dcufwd@GL>y|Mte{y&%3G#E@@o
zGD8Q+cm6AHS%e`W7qfgH-9w9Gs5p(L(q8Z4lpQ(QuDxTeVD}^y5#lS`xA#jjU@p}R
z8>x~X*f)q21U7Al_G$(Jnzjn{9e0esN;4>>v{PyiE*)(0=@??W_GWgW8btIG^TXRQ
zpch_gZ!HYZs1LY<w*X5RD6Q(BbWG3Y!dzlT_D<;;y)X25&6LjV-2yd~<w8LIF6{%}
zm8y^7bAqwD+UK+;A!G_MMUb}L;#n~b|7ITOo?4)#v59fTWK@u;hYvvW&R{}xPwd=Y
zt$;0!#fXAKny>bHP?b!TUA-BV{8$_X>#DtAxAtmrIb*~S`JoX4_kHSXySI-kQ9W4@
zyKfI$1+rSH{K&iKawC)*3iJ}$GggiO+w!V<RKeQIU$a_-w24iVy?sI2n$<1;BN0%`
z<i{~HC_glu0@Wz{*f1}K(*Be9S#uWhG+ZM;ZLe0&md0NFWZ&o#iZ;~<M&9k`&uFQq
z`;DSy|MUzQhVdTm$Z|mN$`;<>NN5}wIB6}9EfRbAH{9<m0)Y&Oa!~uNR=5Y3W-J<x
zMqXnM_BZ{+#g^G1hZq6jhJhI98C1XosD;s@LQ^Yu7C{V$#_-HvuLi2ad>Jt?9OKB?
zqo#ElUJv)ys15?0hx3qHT<gG(@YZlYAV3}()=W7vwZ<)hwp0|dwWFeMh)=p}7vPrA
z%ip$pQD}q6(j>k87e%R|9bAa<FO0&{CwR!qz&A+W7zf>CI@86{&%bB7K{dsxF0~E<
z>1cl&R|}M7q+&2vj`3%aA)d?MzrCBCOUgC$=2&BgBGg=w0qwUQ!vK*cjOf2)tic$l
zbAo~Ouwq67Mf>}fKF;ZC79{f&gxix6?)Wr9f*v)Wh@4={4E3R`PEL%Wjk02cC)pMq
zIbv|Z=)wXy+4#C9H%U&hUj{4wFe?L7%WKulFb%kYK|xQ1WwC$@A`$&93O5Km?3gAQ
zV*QtChV}v@MO_{yz@biPn+A8TChysDjSNejHR)B6IFsQf4$frZqZ%1u1FfzomQ%eM
z!l5I_4=){S<g^%HL{7;_`>ejYPDX`vgaHADPUQWM2lYke7o!ZXjIkbZDkH>M#(Fn;
zHrZa|zqk&Zp1J`Y5O9jH!uGD1D##le6GWI4+SCW5YMG4p3EJ9$69D;`kmg~<OINF#
zcOz~AXV_9OBQOmmXNFYJtw=oqCkWzX=^7w>@2uz@S>0*e)6#-x$L!$4&1DcLcTOyo
z`b)rcxcHpw&$t%0Emz8U{*0I@3IpJD<@`49AuI~pVUb*5)T%7cW}#wVn7*<Udvudu
zWQ@SuZlz@W1s-3_?1dcZQWVcX53VGGjf?#m<SwndkVP9M8w7wW+$z4&T1>a3?petN
zlcVq?B0Z%f2to>KgcKxVX=;Q?xoGv&=Ap?ntE0ax1_SFP?RHG@RGyZ|aCh_=sozQ_
z8e^e8mP`uakJ3sNMik^jBNVMNnuIpOkP4b-s{9%4FbwOTgM%yY-L#cpnq5GugM_s(
ze_kBZtgnDyzE*1Dyvt+JD>YG4txuA2IPT0`T&Qc>dtDk4on}q2lP;6`7;{k$H;Bpc
z*VuB*U`irdis1q27$a!v-&A`WEzQN2X)vNptS^;I66KhMUAf3Vxip&81%_o&MLh6%
zrYFl~F@Vii<05-`{8g6$66*>dKG}z9Y2Y%IpLtJo4q;fTqUb$GuJpGxiwwrVW9_Q;
zTYITEM!$*+1QXF*M*|hF&&C&;Vg$^^o0ILat5aKQhQLjuNv=sVAzzpaj1D)#+kl3V
zYh%5jMCakpzs?vXUsTi}*IRpx;L9Oh7O>O}^7EJ-?h_2aV7ZVRLe`8$;Uc7|L~LVh
z?8%Js0i@z4lNLqb<CPfI&E5(P7z_`k0rM7PBl<FPa;p!U*13)N$b5{<TV}N@Z;N3-
zrqVrzM8k4>YLsP&VLJd)<c=5!!v3|8$(=Ue3gCG(t?pemU)1$RA%LL8;4!0zju<9)
zC;nj2h-xzY7t-|}TX9f~dd$5(SnATWB5Upgs{WuciWka#-VX1=jJ#y_K4Z=Mt#{lU
zvhsirzX`IFUjlU^4|>aBXH3&T)?{r`9vOw@A)`F#4z)@#tPdM|4}(eYk!Vn_Tz1=d
z)Sk68s$mp?yXg2a?*u!WET)RD`BbS2&Xq=MGKbsPG;cEek?SD%ru#FiFw4;`kQp{F
zU^|z|%=jH#PXq`Lk6G5_Byh%TBV{do;W&PqOdUV`P-%`$HH1l<=d|B?0Jnr4bPS;4
zmY^18<^ltFpmQA5bB&Q$9x^BMtQpXb!hC*=1tv3s@MpjZxFF5R;d0m&3;h{XI5Ur9
zPK&%H@<FpK4!K4%4@}*#{49xW0EA{mz5$e7YSSzPoMc%D$<c!b%5rZWK@S*){wpi2
z`9e$Cv(j0{G|9FwgjMz}HYKv^<4VMYKC6Rq%j&C2p*5`W_D$olz>Kwt5Gw(nS!aL2
z=F5A`dVilcB$_QYKu_7Q%?L0KFfH#UGi9SKE^^G*$R=xtXGsz4V~Noz9Bk}LLU}xG
zkyZeZ!(u?<rSgPPsvly_Hpk#`zc5VCd})aZ^00(YVN3j-DJcPNQ=UwNWs<P4We{Jx
z)dmZjaz>u=mbr(+M)7p}t$n~)|L>|9wjlS0JmUnNrPT(WjZxD?N`VDNd@hc7Vs4>l
zLo(LN^Tr5FRGyPx1y>9kGge-(p|&)R$;yk~9L?1Pn>67v_ENk&cF0iqbs{xmlqgvo
zmMp)CU!Z(2UJ4a6T-v;Bo2#N)UWsKxo|sDcZBP>iOQboFSMAsGy3N<%bbHN~rf<Jv
z<@Gel8EPS3ij(<`*a$FzAd5WXiQcr=Xno+`0E8+hZ^dh{qZ%pXcWI6@)Zm8<xAJ!4
z6T-<jEmec)G=tb|mfzc8@)y_3J4UVxF1_?Tc{j<)HhI)>C@>?(T*%S)V&G86d0G-_
z|Gv|SHCT|71J>g(^Zg+P26+yZi1tYNAU#u)JTBa8<U>2u02|WuKp(Xd9$qO5LQjS&
zfAo)e!w0SrAJ3PM?HTtmSemtRYvmK4hUI4B(3wjVcRCDWCaxQwT8E?E%KKR&R>BHL
zw>X0~$e(;bs<f)KBR-zch0lFJvjE!y$_yZdwem%<7u;C#Wymskrg4HpMtl{sgG$5{
zvna9QZ1v|v8pdX+gqFVqgPq1(#$Wwgp5rXnuu4sdtepF|_FJ#q4%_elKr);E+sFT$
z0`hm;R~Z~8|A?il%H#6<bsHYz(pd~<GD65V_BKxXnNY$0nK+J)>1J4TSj!aY{^Vb=
z|0hGiGX&JXjp4u`r490}GZAf7#*wS!J8#XB>S7D1t&05ny$!d|fB_;M%sfh13?UE%
zUT)XH&7*MJ<SXIGu9NLMn1z#<86xV4HnjuF!tQ_I&A@QWk)yGVvO@<q6*8GE>Y&;y
z@czID0;fDxzn%Ou^IPLn_*mL?a1$pkKnmAv-$Bi}z;(^6ln(v{eR7(N0jFQb4q*e;
zD@JRY4(a5>Zfluo3v=!4tpVDuN%N4mS-SW$+N4my@d+sV($#wio76JB@Q{?<(hR1<
zUBkUOSGxN%hy~TIY54A8&nVPcE?hGe?c6~PqkWF<FT12q=xYVW0y@&J_H!YO^SgC$
zgEXwT)$+stn6Ja5abe%xn=z<9-6Ku1hc$B=tRFuzO13oiM?~eG9o(Wdm^BpYWG`>k
zE;((MSoRKnh|pv}_hON1P@8QEk^=1$t*MHE5(pjCFnrvh@>6d&*o7g)+z4xBUmtD2
zJAtT>{nFdKJM-kGk)3bvALGMcyiVwKT?0T4h<3nsG@~*d*?}G0wp>a82tlm}#Vn97
z7Bz*DHdYRfL1#*9a!@V~>7X`V;5PEo)3_Ib*ziJD$f2<Us7Y8ShuL4_5z1a#CWpsw
zytK;^9o$gF6Fp<e_~l3+B5$?4a53-lQL#UY@_Fo!UNH--8vVyCpwv4(18Cb0C+U-(
zEk>v`9a9UXuPp~MQDCbPveC~OpZ#qEOY=kLX#btKae_lNr6K%L^f)r(;4#KD#~$5F
z`UletnlN6Djm~knvNZHG+|LHsJdk>xeo;BjU*TSW)fr~{nR0v(8aH8zlM~WA7W&D0
z&nLz+Sc+);ViA>-(wY-AY{qgxf^ii&+1_r^e4s3<1!;1M4P<;7>e69Z1_qBJp9kH|
zGRS+NW<Vqxug!yP$v58zfi=W`r-6Xs=X3FUL5zG9l7OQO*HHT#Y8bA(_<NWUDpPlf
z47Z<vTUm_%ZIThj7N976e{yR23m{%O&AT#x9<8O*Wn^?kjs+n{`5IZOTilMJ%#o-B
z5-)*acyy2eT0CMjHCB6!6O3=u<PY-?{zX+#8Jl{<`l5x62);Yr-wutRARbPBg^Ww>
zEW~Y^Zxc;V{A^*ccX<^GQPvTpS}WsYZ)F=$;261>D`bNA$0LR5<d|pX41Y#^J&pd1
z^f@!$(~C2&JBV@hXDmw293a#n&$h*75`DV&&xy@5Qf<fQrj`+j3H53^z=7xa(q4<N
z&kx#W^Ox`jbU}PRD$@WU;Dx~|Q!<Q&E0v49y=LG)$;5=RW@4nU9WcQsVk@O6CO#Cw
z@^H&w;>EFF#zW|2W1o49!aC^BwC)1I2OH#c$=MjoP_jTu?7t7`dBjOl>hw3ENna3a
z{js`MSu|f<nU(U8JR@r0!<?7~J&mR|WvWRrXjN6YVP{E2TK1*11As~B_gjv0uF~HI
z><t~7D5whBjHso_r&jW*NksF5)+j?F)$tN8@7Nc0a&gFCpgfT60OsSG7%R|OQ>51R
zO`oF&NL`XuNOwdbr@pWRypj5#G5EYZ3m+=beL;kwaC1%Z>8OJMTt#i3@S6=XHAV{L
zEYnD@#}pfE<zc(g4Rc8_MjfoV80DpD(^-7je~3@HEZ&E7FT#;=d5okEp+;ADe-^c%
z70u*l+pHwyE!N&VidO~?mO!;xDOVXqD7mh-Mr)C#D<{{aNs!!TK)W`vRjMzIDYu62
zT<5<N#=+!4mU)@QPMN)Fg<Nla)dFYv^Pp~FDj;`5%xN$Z>)jZC^g%kWn|v&R(MFGQ
zbM%D`L1Y!6<lv^(YRI?PH%KjnYZ`Jg(hq%Wq9QJ{a9rHx?HaoqHthBw&)^Iry6)Jv
zYuNK=%AHQKY+mm2pTW&4sy0BA{6(}-+)xMm<lP<Ii@NzXaM*1)WN(9^<{lf0M#vBX
zeQz*sRjwXsqo5_+=MxMpwv~R>s$Q7pM!CPk*6y@{;3$F{8`)(UUJNVl1OA;E-Qwh@
z5!Uw1gEl-bf*=q10YSUYVtF{Is&F3aokXgcP-mvFN7A5hRdoSvIuSR}TzND$g&hHU
zaaRtJ$DC4hFEPvv?@MDy1jVwqUA<}cHucs>S1Z$FCj&?{MP|f-z%&E6>#CV%rgt(2
z%VDuu^JSLFw+SN$j+5EZENY-kq&^H=W0TRNI+NAHq|NclC`S6=sDYboPV}q=JLg)1
zFnA;S)jPDizt%#K`S!ckFIf=FnyZ?a*ZfQi{R{K*sz+89C91Jng=sLf+AWpE{x+YH
z%NN2rzGRy<qWPLfCEaWmthhA#R38<{Bg<_3kchNpEl;yy(Ef@v1ELZ&v{6=i4|Y4F
zPcf7^Tjf>$4R{8s8#!l*tWI;Fo=goI$FJ(iYwT?n?<lI(OGuThjh$%OM`c}_r&VZy
zK}gn{E<d6|HpJV=C&6QKqy5bzn21x^)M4u`>Ux2Rkb#*BMdQf=<7S#3ig-aJEkvx^
z<33cn-I+{FKhlHfiFg|WX8<qVhMTRk&9`HN-v;U4;$N%Ag0E4z9%SRCY)hO*0Lhh?
z*5~BO7&kUs4oBG5)I3w95z8<`o{GU}fI4>^&Fu^)d)h?ch|w&4{7lG8cnykkS&EEj
zW1Vp(RKgrpE6=4hSVHvyn&ccz0trCm`P4J3VZ%O$yUMR(xPx-#6G_$={3W39IVWfz
zE9FHS1b`hdGQ1S6K!sWbal+{FUpuexJZH*eRtGbDh9P%;^B?nL{xajDm%TMRI_+}u
zO3Vp0?13tSD)Zae=SZ+kby_3lUX9nX5S7&&`)fW7$~$;-G?$zGx3Z$V9z&-D2Z5Pk
z@(tSv2$-IM1LaMV>XTp}e=BxKuCk~`ei!n5B<%<CwtvMu`!nVDiDnyA-$asKdB;AO
z2&_~Uj*xfbC>w=!d@q(redXNHOlh_jcIWqFIciagl2iIayb3I`l{;V@^jt`Q5BwK{
ze1*Y=iH|=t*)wX)kOKKASO$PkJ!g>h&4w}z*Ul>WIOavgzfL|$6oS5@mesZLsdwZD
zB()nX@jBcqK8w*JXAz)5C`;;}b5j8}^+YpHgU{n7cvv-^#~0pSvkH6@FLO|ndaZmJ
z%M8po5q=sm@s+<0u2LV@Qrcm01D`2>PGp~s!4mDiM*b2L&?FAVV}DH(KpY~Z65;K<
z&&`&<#n?t5iANPC@ONKVGU9UiM~oEW)nnxSCh~Q31KSx<pv3uplQzM8M#%v5iGc#L
zqL)W#fB*E>&^44Hs>4)Pur=jh-Wm)(s(&tmA?M%z3{!wrYYu}Q<=YS?V_`;sG3GlP
zZ4r4aC*Q|p3Z^201Nx+NRQnp@adaPPOy71L)y!}yP4C;=@1Q0pF$<>)svh_=v*MT@
zze7hirXf@d0&U0i3YZO_0nZ>ibqq7uN`ST)RA$hb1=6mgTF{6csgU*^)s~b&=6&q2
z4*roT&!(i<igirC&>mrd*9OV$)KTqL69DMz9Mb~?1%=5V4YSgvqnlpH&J1&xt~TM3
zdB6?2r3o+Q6+=Os-R<|ndL#nu5x=)A0z*+~Hp|Y@OHHnnak1Q#yBI+b;;xR{UE>u%
z--^k&&@y-PrVL8QGjO5&u(f$u8937ovb&L|0LYRC*~9*5X&eQW@kbra?5T0LU^~`#
zYI~yYnOX~*)B@QnErTrpbIww0pCx;zx2dBlTgs0+x)rW4mtmT?ef%9H7zWE3K;S1q
zP=Fo{An;Rv3+bk4D+1VXjqK}V#he)Irvi;kKv_ix#eP06{?3zu08Mk{>>pzo4YENI
zJ0Q(K(S3A=_?A1+`=BQR?P6o;zZ#=Q$NoWW#y}Q~z<geMsg)jVoKp%M?IDR6hJpo2
zq-PqU#t>lbqr}gl>1|Y@flvd)P!2N=VBkO&y2s(h*H|AwwdDx=vDkAsA88$zWQye|
z@5DqOz$Dx(kXtjPm(hVG+%!c&Z~Lg{5$8#tj&76&UKnV#TIuVpfHoQjMgd_4l76u^
z(65>BjJ4*E_GcV8Mb#RZ6OOUgM->c_{_(S8AE2nqvC(4$+z*IVf|O)%&v7=xfmJwq
z5jT8%q7Y2@=oT@3a)R%BW$Y7UTv#>u#h~-dmXoYas8q9|Qmv4atxZ(>7(M2UfpSW8
zIWC8i*Htpm_Nmr@801S$kHBI~ovL0yOy^;B2FG|X5<`D1mLV}uupm|3Iau?8j_z#0
z)d}Jv4|nv?_;d@3ZtEUcB*WqvVXk<#2_zccaqB+RHnG2<#nIPGS1#f{5uBh&b(#G|
zM#P*DWB}9cslIR)2?DlE3wK(~7GY`3z&+C6v`aMelQ9?3LSQmR#hXJ=1|C9wv=0vV
zdbDO3^D&7XnelC`QEo7+Fr03`v@{MvX+;?qjHF47<2)wjuh721)oZehPd#kZ`y)^O
z3HCec6qU&t@$Q)oa;7oXkqGWNtD{=s&m1CWN2hR7SIaqxnj0C=2JT-V=lU~pCk=$~
zP|A7M)FBsMDCZj)3+=up7o--D1(v%T>ehwEZV0ObnBk&WSJ;D?{~rfKrsGzNL&a4n
zixN(-n6WNQQBVqZ#br|L4fA5kELltho%Lt5LjvXowzx)eiEb=(Lx=~JUglvdv0>o4
ziWC{8@tP$sg1lsg4j_Z2%t$tX*(J;4tu0sS-T?dnoW!fd#GqRiTChxtS(}axj@u*^
zLHfaiq|&-Bg^pe)RUx(ui;-%I`8x9Px$2zMK4wU@&z>H@RdTU4fYS#F(Q4v#^=sHs
zSR3QXRb$iF#ke8op*^UgV^I_JK{9B8j2oK=U2U?z?WPDQI>d%CAYzJji$Edp=hXNV
zE5KMkY&dD~XWZ?CXR@H;CH{<}eJyCe0Y<D#<69cnA%#}8Oqa_#x<e$JtIPv20sl*}
zT%JCrl4~c_6*daQD*%M^v)E&`fO|qKyfXcblqAr!u8OJSAmfsn`;rIU)oBbsu;3S5
zh~yf78vy}o@<ifeGA;P%+SVgWvj`e!oa8idew}{_OFC@nWb^=9UhiLqrj-tQ^f?2X
z>(A2^HHQ|)xlC^GXPB@~8y5<>G57>|1q{A)Q|z6ZoEk?UdT$OX20g2)SZ;}Spg%3e
zdJsZ;t37LJL_7^`u%zN`K_x1dATv;JwpZl#;I#3>hRPlOnh8bP9wPYX&TS^4scPje
zpLiR^8rKw)%dCz2iy+LFtFfD~X+gNQVCK7HDF?#bj@$S>-Uoe-dN+XZyEocUo0nbj
zUGsv0ce&hW^q|7Rd=~fHY#>yS6Z-+5jV%`U4Xq{sb;^UORXQ{{GA7$@wN*Y8hb|NS
zLPW{K@c_xNaUIOcBOSMn{r?1b`k$i*_=`N+hR!hcBZuEr@|Zs}>kf6nh0++$)Y^pO
z2$=n8+su~;eK?BvuS_>88H{ME%&_BN_;8u&glLOx;oe}1;EgbM%}RVjj-g3`nf2CV
zH_tXo<8A}YUXzc8pTdBH*P5i+pV7-kd@?OabK*URw6BDyn484@d{`jowfTw|Iwb6t
z`RS`A)B~{fX37HlN?i)IvM{}|6zfUhv`7~DGvuMcwGY-iaMR*o_H08=mYBfKR3aDr
z()eNQgmJRW{s6!qIj@&{Gh{vPoFz~tR#@jS3NvDEW$aRe-}6$kDw>0qD65V5rSL}5
zZL+2f??L~DoiNhjnxI*m=7Msk*e_IZ*Tt>}-V|}E>(e&nX0v>9YGIKL@v-KZ!!;dj
zvC*G-_z8V@vMKnv2DPYhOOjv41HcT_g!i~nr|;1?7N3Y8Cm{}Q5>D{V)+8|;m3dfy
zdCO`^-OPn(B&)2FExt;XwLnzLlZn~v=3$1PHEN&2fi7FU1p}%m8bF>h&11~aq4IR_
zqF!lqBfD&V#v4}FK-b*PylUa(f#_K;&srn-qH=l8xs7yPD$g5T;Wx%XT`Rvz?WAxf
zuv>~DkQdShqt(E$=F~;z$%`Soi}Pi)eb>oLM!(H>GM3ldZF9-M*NjS_sB4zr7`v2W
zakKKWjj*M$2<#%S#L5pI!@#-U29ec}uYyZnbv{Oo19{D6T8T8juiN96#v}KWH_`yi
zI7&oZBy&SFn7kRoM@AH$s#$r<8k&p?)$iimcKaCut83+L?+E#0fA7=<Z@&|L9&ot4
z>smRbuV?77<--1C1QX-!=g51B<W}&GdMF9&(LVBiM|IX9w-eIgAj(Pc_40?<U!%(>
z$p?uhtp3I{d(-8^P%)ur;9vlQYNX6Z{wl{roreh|{urka3|&kC`*F~JP<=^>d}5R9
z+qb`b8vS6!9rYxir3QKGn`P)vv0I>#BRY19d~Ty<a*8VXBAOpPuyCM!8Gj9}K=OpI
z;>!_;lSeWi`Lq2@crp(0zc~Au@sC~rOSZC5{_4Z9lg!kzzooZZEp2M(lE3>e&B*RW
zGzl4z*)e9xKb&dr15bRNn4JdPSuHbEVWNfn8(&%Map04g^fMH1sOprp*r1E$pNU+H
zXq(`DO#YR|IS-mO)5R>6f2U^*Co<&Ow=s6y-&nDFk;VDp1o_S=lBM^*Qogs{F%b1O
z2)C3@VfhM!1GP7n``WIPn%Xk3jmq|&4C>ar<w}(ktD(IAz`u+W&WcaaD8EA|HIo&?
zyaW-pV<)wswKNWa9eJntV|00`wDT4;Un8c$#Ons)Z14SrMUh2{I>e{gZPL;HfcJNO
zm2`?<O3*`R`xGw4dQAjJt4x<pZh{zs+&LNGL%POGa64pSeB(>qI{9t78tzL}{3y>z
zcbf?IJrhRu@IfKJJnn6(y2{Q$KI{udcJg0NVX^GuZzE$cR0GA|%C0fcQJIOdTPHJ+
zG0;}yXlP6UfXM<OkZ{iEpUp#q-#xmfA_MtK#@oXNhr~1}zbijVtHByuG}sXa&t%1W
z`nOsV39QSC!+UjdlW2Wa1^keE$J8~ml$MhEv5{;nObePzT=wat_u&a;@{`mRF=J}i
zRK8p;KlNYipaK6Tu=~DFNumW;$$rs2?z_0h?;kTMsxPgT1A?R)3*M$$m>ifWmXZN(
zL$QNmrk?8??nXH{mL9&Py7H0+Im8>`?jjs$A@*q7gQaKC0vRuJCH2g=d8n@h^@zhd
zZDsA$@=IKkIXsp$S2bCVNb~1ut-djn-*RMHKw3M^U2s$yF1)1JvADKNuk;=xju8kC
z#_w&<P+X}V*eQSjAm#c5$;Rh!x9;0Xt;+e*Qt9UnP&GF&0RnY9nXyL|3^YmwK1we#
zFR>|6$0WkEI>9Wm##H70wrD7Ap;qH#6Je(?ZAexI1SNX+?IXwedeaTJOpcGy4$0-J
z<%Bjvq_vf?sV7F~Y7^8Qgn?>W1pK>PPV)Aci*ciz>@AR)nJxggBsnGeD9Kk$l7Y5l
zd4dTB*^*f_|D;+O+{umXBcTe{XHZULh_Blm*f2DLA5ztbQjlnml(W?txS)Vz^WIgd
z*OPa>4D)9UjWa*jd>L-5Ql3Gr&=?#u3(BeSS8-W+rJR<Ui#ok4Wu$$2_<4uRC?8u`
zC|c{4j&_>aZll>!OJ4B?!>hWC@rDgj!v%*i&ANs$mp<pyQ`gL$Zy7Sj#WS!8lcf>9
zRoKbR5QWI)dr7WF<JZOqSw}%hge7l6kkx7^K)pS~IASP%t(7y|?2X8;N%eY>oMmfO
zUJdDewoSZWuVdvL9|!>n6s?eqa&BT1RejWGgQHFlMt}$8yx6=MI4v74;qh~RjGZAN
zR=O`I7Z~%Zj-z$1F7(k^%8}Mc|7N);rU~0MQo2ee4a=~mye<?an$4sE8P3J^Qf%KL
zjWQBQ0PUFd9;m}{g}4c^a&Xb){A+SEt^FHu6D*0%RF|J5rQ48>rzPE8ywaA%;LcQ>
zgp|h`Xnh?9`%W~mU2wWgYSWEXE0`-4F>R<Ha7REDVSL|gSOP24lz4xonRuR5h1ecW
zGrHvM&w4L(Rc(FaM2(2Ri_?_bMv0(ggVb2-w6|bUh*?3VI-7EFInBa#ZKg~oU8~c!
z-lupg0;g`||C7na)0E<n15<q9Mke+0$ka|-HHMFMUl62aj$42pRKmh2XWF6qT%;jL
zs=j)J>UMJI6Oz&)2cG3NluHwNG_qe&5SRHglL4yE$mMO`quxjB=@r4G;{x;}KMP8M
z&Yn*YtFao2<CTfyxgGVI?h`Npb(KxNph`2gT<z^Nfr)U-MQe0Fv0}_MM%_`st7qie
z5J1(L1_a*vNAwIZwA)k%0#>gNcFLj(N)?k$|J>G`e-N;FLmUE(bf|}E@y2LNbtIEK
z#^pCzp8%wBeQ!>5;VRJHF<)*8_ClTr2#j0(6%vg=5R|i&+u|84iu8gsZ@F4-H<2;m
z*j{pn_m10=T0LmRfU*RG>z&b`nvcm9tK=?!n|qN!8D_@enKNI0;cpucQNAvfyVKh#
zjWz~Aqv<^{DCI8R#Ma8a@l2(%K`QS{OG&%GLJ+J4{7W?8{fTyJ4^(fBE%F0FW|oKH
zP5D7T(Xms((5A}R0pY+Q#T!2qLL^^QD-WkR(cQ!oVuL(lk{3R!dU@3T8z%i@<gs{O
zc$zfEs1T+`I2yy;%vmtaTSS}~l~Vc#WV$Uz&l8T48UC|raOzf)nKlu~h&sSvpft=%
zt!u~}ZC8Hs&bF4$zu<6bis4*#qBO^x#vm8loH#9d%iLfdSQ}xSniurP9SrhA4e|5S
zfC#K3+H9RHNYB<M<NQKvidMQ>oX?9oZMDjNM{N6lve-oH&{49)B@iI#*s@D~PLM}i
zXrE8kDp}TgVykD{!l_%ar^`+D&}FEYy%n9d{?II@H~o(`P+1xLgbl)d%Qy$aAsNzj
zl~Jb<NY&M`;fv~lwOtc5908qmty91hMuv?qlyyb{2N=rwAST2dVKMkuHl!9Q7B*T}
z2*Q_5MwRnVlwUekAjlx~9=D|*1^><zz&~>GMEZFZlz`14YFJWC!~R+l(-R$zlYC2z
z6r4z$ATL2r#xTaRsMA)byq)6B0DT)|e1SY=Q@;E{0!N+>l^~nPB@zA6Qt*tmGy-{F
z#*LR}Z9+YJ9w5(oqjqGsJu05}UkuvA<3kU2sO-P8?odP>C;tonnrjsb-ogq&;=LGz
zwWxCJqgnD&5CVY^8}Yv-@@s#KDGM}Z8)VLJtONMkv58&|-bdzo$lzC;u+5NY^X0cO
z*-<Pl#p+Vk<??E3lIeAzVZkN*S{zGIpkY~|ru%wu7;WX;F+?-y#X>FA@<uGtIF@}w
z%-)-}Q!^}>`BtJ^QsY%p&6nSK?=U8*O_0NU+n)i8$k<_Ey@5-!XVek^e8(*vOz@7s
z2SJ3*Wgv`qy(ha4Sj>l63E%Sxv<+h81Ux6FydUgSoT)9AKRDrlp2DT&1Djk)jh3`4
zZjcX+Cnn*x{gKTRu~0~RXa8M~v|#eFkEAW0-bDB(Y0g&iemxfU(_joW*)UlPb8Zk;
z^I3YEwS26FKiQHao*bL!bAJIjL*XefdHjWqug`$~@@1#3C;V@^TI8!(TwH<?i(4Uo
zcJfe7oGyO}ro+m^M)y|}U9jCw$Ny%@-+X{fgw5<Bxc185Z3f7-ggfg$;ssUW%?$W;
zEXO#adcTPQ3@oSKUMK%F(K2PK{41?E6lguVaSCc}&42sJ2wuR=7Rn4v<q(RrYvYh*
zLTqM%_|`gt>9)QoC*S!{Xf$K+2i>~g$MB$<%)v8yZKbo?weorom+d;6eT=5x?K?Z1
z0tx{xO>PECg5Ov02c6B<M$L|RJVKJ~`n!X*j9Mu)XYbhAtZhs}%20F~5O?x#!&|>b
z)h4>x+r_to^4SJyAKzjOk^Z}bKXdc-6Vq^I>DYN|J90>4reab<g>;Hn6n8{IESJvy
ztO;i*O;gJNB<|8Vj8<)<CsOJ@>FUFTRa3PL8WmM=O1JbSpW*sK-3*LR+zFYZdm6ZT
zLz(2Nhqu<|R6}bhjUzkz+jEc|f|l6n0E+C=+3)q>R(NDkU1Zm2Q?rN}XkfR_YQRCc
z8N^>Aoau)_zTN{6rLlYbHHc9`dqj)$0zxMRs?BFXQNx^P%=o8a#?<GqEA~v|nhuCA
z-r91F?A2K<XGk2mcOp7$D3$W#&Tes}a*b;!#M?fd-B?;%UL-&9#?0zUP%3ii#gwoK
zSF!_`L-tLbnqoum%6^G-kY?aA8X--N>>q<DuA#mBfS^ZlK2JZ@ft}q}nwc_5x$z+5
zbGZEK<lq?I86!r>A@;*SR*ml&tqf$s3B-KH&;esW%9$-!gP#uTywx+QU?4UaP;jGS
zsn<hyQQrcdFz4{${yh_=&{{xG<Pkxx5miXtt@$I4?5t^xkNjW!pBxqA9)hHly|&>|
zy-2ImYnAl&US~jPEmKtgO6e0b0i*+Ao6zv4OW*%!fqEXj#xtc~5Vsa7({gfjkYyP5
zHKgP*&OWZO!>mlX#JCm?W0~|1;*&X9CGFVGYS}yTh$Ccx4@6fHE00d)IDf_z-K)WZ
zD`BiT-um2(M3DI35;-B5crr7)mEvY~Vww`SnsLQR{)`;OMCAau$jPZk)`~Ku#3?aC
zSh4`6%gMlaiB+sI4W?urWSr5`xZi;?IDW@2r;Uj~r6K7VB5>I-6v~2FqHKMU4D}aC
z6^7X{Dd(`(XKImvSd!sJZs@T<wq#{QVtTqWXCo4iwgYAnKGj;!)Ye1QJ1yqGYc0GK
zGBQ>TNtNJ|8x=j_^xz&_kBh?S_y+b0)TFJIG2T~zTTre6OUYo1v9ZWR&7yyrQ`=%V
zPWLY<WmkivmN~{^p^P(0F@BH~MjNo9^VY?r<GKJ&jPc$E3B(Xl28LLm=!DLSve@>n
z^1uI=oDuU?^Ak-A1eh{Y-<e4!vKl8D2x>6U3VLaWCG4z_scIoOJGMkkSwme}rJNHp
z0hx)MSq$ACGzFKO8~Z(zRR=Llx##&Z!mL{%=eN=b(nFa9f{qSN9+^nN7ppq9456uV
zVGLsw7FjL|79H0h8DDRr=4rN{kL!MsKO@+aph<?&7st>G>KUIdS$_x4cRk4ovdBfJ
zbu^1hNi+@VK+6+@l}lsX2GVVrfiuu-xU3ztR?5;q3F<D7PS1b_46EBjYmqnttHue<
zbW-#JR2|j%`PlXq{*3k<S{mVzluD;H6E71W$o$gSu^3ZTdYjnhW{9M@lJ{q-N>fZ>
z8&n6`hL_95v13OA?p+hZEY0T`vRWHI9(lk4QrFq-$CCjN0d|qDWvTb(!7$MHP3xK*
zF9JnOkiDw>Q-Y)=8XGt@7zX+reHTo2plYx*1mVUqz2GIC-Eaq^39@12X(7$Zr8a?6
zX{@>|7_1Ub6xjBFf4V&AfL%+$vjOM#73mrE9PDS<(SPR8=mo^_s)7Gk`ZJ{PAYCDn
z5t*9an@40hGQ&-itNmFMd>zytmg07NO?tLgu^otLkZWUjg!gLDkX#p}zzrHO+v{UJ
zu{MNu0~CpVp?k^CZJI3HDL2?s5H(sVH>N=v+c8i5P0o;Qq54qTC_({moh>(8uLWAu
zJS(@v0zf>{;=nSk&5*V8<W_Hd4(vG05k4E%?c4lqlVN&Fs+^o9x5u2fTn$;m{X0|c
z2r&y2k!F0pGx!RY;)dc1xywJIp^l~^PWBhAWPliC*bZ=45GN>FkvcaV;Y=u{cHZ6N
zzZm$=u$lJDy{R=MBvA?wb3$|Iecm^>6yq<L<?#OCvT9(7^A-70jo^R41o>fy%Y(KM
zC5WAqhrDG);eupL&_U?+GI==oK2uf20IgbDBp!*TQQ4rL9vcl2dDJw5zDFM`kEQVt
z?N7nGRvMEqZp}y4R=7-yM(eF33jvvK<E=zUs_h|&fEkcxGh>NN(S^WRC$rKDY=Y{o
zxOJIrvqUYKY0_l>4Ut_d&E6W!qyQey@%gvG2a$aMD4CnuWVLQWB9_R!V1*(Sw<qh)
zkC(W!up~5VFYsqPOL^V^YAy?dWkzKxXy0BG4GqbaWRPb>7W+qPzQsqpUoDBTwK;H&
zqDw=6LdiV9vmx7-8MW$)WO*7A_|3UwMRZ@*fbfE;vNHar)e~}WRs0QSE@Sal$M67G
zK#H0QS(6yhZc6lhC=gB7#=tRKqELv-T<6c|S*L_)#^JR-J!8IU8ay`223wYvMg%8p
z^xx5fO0T7eU)bb~gD72vASjL%=R2*Z*yr*~oAkKB@^}ygSE5QSM)X7o5RLzA<G00l
zctzcun3A`qC_}6TJ8}9flNKA>BzlLpn6&EMr*FTb<;f6%1(CZ~wx&&VjbgBQ!RI`h
z<thKg?w+_AV($2F;PaohJyV{^%QHd!iP!)*PM-~lN{zcto=cQbs2T33E9H5c`7nUp
zQV&#;?J7=7ShKtk&DK}zrRl}k(CU}N-X+51r64ix=DgR#Zis&o#p#^$Yu_!rF2Tg-
zL(zkTVjZf~%h6439+tiuc_o$t(X>{`=ePb9qzf<D>R5O+III>)cX$ZCmS$<FLgI{G
zPh2-$wKyi)P~j+Vq=Bq(eRh_-8DfNoYj!fAfb^rk<!?qbER)^+&L?e~UK4RZ^rhUY
z^0uwF`qt(5{<6ist3}6OHoMq6MxpX5#j(Ge1_s)wJe?u$87mEfw^rV_|1Qjc_J4?X
zaj?T+2C2JIKJaH`A4aeu)4Y5bQ^27_yAMlCeH1)0czD6!)8vm%Dbt_K9tujqsPb_r
zJtg_PeA3z@?l3fh%#}~mGmGSAJ>$>(nQA~-vBfYW{3*Rh?75!B^0_~w;hVW!nabyj
z|FUT6qLD9+QX@32k$e@)Rh!C~fero~^TOGRSnR)e2UsIQh@j`MmA_gu)Hb0s|IPl&
zWGB)X@OOWowrpB?)<BQ>NAUgNGNk-LVtE_`^0jd-jM;1Pzs2%R8nxydc_M-&7s)?k
z-f-9;PmHQn{|fFc&Sx~8_`f@AEX;ottCw%na1@-U0ZiW|=HpE_a`h0TWheUg{tW?M
z%w5f52GT_hM2gyGri|^nsMQbUF*7)Xwzpqf8fk<8kNu#FUeJdW43!<as6mi9TVzLj
zTo0@APF>8J*Muto&r9BA+jVj48;h0IO8YKm7*tCcWLA`zS|A<#W0TZe+O^WLi<^xq
ziN#SPRwsWA)<-J#5L-*7bBwzdMo#I{#f@VqAVF7gR=Rd^Lk|M|YKV;L7E@E(MOM1o
z9D4RVPkMAQv!9=s8u_Jsrn_^Y>>NWXE=Q7a6fVxmE?r#bhbmuQ<TZ46P1D)P6lD<6
z(`7ds$Jh$arX2XuA38N~`~uj)h#Rj0EW5{)h}@Aqys@@^o*((MSu}-EMS~)-XBV}p
z<&grXSoZ3o%mce;5ikrWo@olu-ZA${0HlF9KTa#sYOOSBxKAuyW-2}4Ke0dS8c-b>
z`N4l`eF1$4l<&SV80a&M=U)QL{(eCzXahxd5!&CrsjsANae#f((wHxXPr@L$2iijP
zJ?2<B$QQymg(ow&$vmjSOpMbk2OAFpSX3p4#7Ijk7|+tvM+2v}-c}dFBzLI2KAynP
z!-52wQE`<V-eqfvT8RHg*W>^HPvnS1D`?fKa-al;F9M8&R6`U5f{g&NbX4$e4!M9?
zm%3M)+$uuEpFlMm>FxV5i)8S9VqFTx4wSyZMgzfc($D@v`v`)Rql0KDVgzOF7$ZG0
zm`VTC1@26En-+m%eK~MW!1%?h5aDQ>z<PFF9+1Al{Q<MWks-%<e@zIsT_wlI994Bv
z6E(gCzEMuF_W)a>@_XVo@2Sq}OMa67LPL{hEW_xicAXr&2KjDfV&s&xkz27IynwTg
zHcm1bXajC(ESRbWZ(h^~MKid$1L?(>p~11aAxTO!DI<>9Awgp%UM&bZvLYTZbG8-$
ztB#OiM$<n1`^s>q4`c~Us6+~^lM(5eR_i2SLryi`#OXno=qfqQ-?42#qv9eN89gFx
z7HZp!3bH7U!sf^Y*dU|*7d|6fA(1>}E)ESDV<YR;ub+(VqDK2MwQ{=mMZ+|6AuuRs
zT<V`G8LdEcVek=-&sj1)eYGB19+lB`G9d^w3><lez24GT!c+)n#$UOL`eHdNjh#7G
z%vfwX?%CeSHBZcif8!kM1P;>jY&q;DNTMv~#x@>>B}aD2^NjNKRX{$T@9iS5jYc+V
zwtC#An&kpp&~46CxzIjkQYN{`7wwq7l8LE-a<ptBAX$q18Mi0X(X537wb-BW4aQFJ
zLZ0>UPlwjLMsofJ*ML4DJ}XIM<d(r2VY6{{R!FH0p(i}}WvM&H5Ks|YF6F7QW{Lm^
z#U`0(pJd81GAVur51`l<HVfcy|Ftg)m`eK)xh<v2s8BFu#CXZa!eIAPFJm%z@E{;V
z)zLi6LzG_}@Wrv-$Xy~e=>=>}yuc(YwXrMqJ6!7QfF2CUDm=+*AoI>>mP9H)tvF68
zzeXm<FjR5nRUNrIA-<Yqim`3^#WL0Y>x{9|5VM3@K;W;ZibMaG_%Egy8CNKFLN2wD
zoyxolm&G(j4;(L-`!tX`9y&7}CG+J98#=DniVMRhKl6sUaOR(N$fsQC-_vFaGN`6<
zmF-+@ZP{^hbqH0=i;*pgl7aGswE4IgO_OVEY=tQ1BG=k~!?cx?>%7zO7E*lJadCan
z9xh_FxIq>Fxpg`eNsfp%yCIDWH!To*Aryxj6M3u--D=F(>=ie~!X0v<+#FjsTaMfQ
zEk1a|VZkv;|Gnl{;sRbJxBBnU)WAV{XoFW*<El#M!A7~w`?O7OkP($~a(l?PY5<&a
za)%F?LX)A!v>4u*Ccz}bChhL>2F==NcmdKBC^s886d<T*a<@$hfmj*2CmpZI9n=8p
z?!CTkiHuk;_XRl>;7{PuY`NdqEn6lJ_`5Km!mjdQ4D*Q@fRMn&2cykHL9s$WE9Bv5
z4FZCJkhHWt5=Rb#_gq7JG`+T3FScl29!tcEkYL(BDGVA@w?N}U_RtnPE%<m)rc|cK
z9}@}-WQP5LqD`<_&GZ%08z-7$ESs&9S_x&A5$e>;<c!SrZYTqoa-FxurqHyDfxlJ5
zRkOcERhB5*nc%HC@g*3WZZjsPo9jJO5JoM$Hl9M67cC=MEzG0f+WAfZ^#-ElO^`h_
zuK?KQ?VJ0_oT>}k^oc6GV42ynC{`BpM6hSf9X=4cE3Eu4i{Vh!Tq#RTd#<f!3hEN%
zAIxdd&!y2Y6c<{<QEye2rDv<uq(|fAa+9jZ9MfM`7!?7?Izm>)Sh;jO25>`K6+AL7
zk67Z>A;)1IOD|8?4TMR{+Vl)bsC1~)WSy<)XaGmbWPP*-D^HPZh_xSxpEmlpGznN7
z)<)T6-HjP@n*7qnN$0pc9`w(%kQ@yuPb3CZ*8nbV5lyk#n<d<vA!js&w4|-buxUJl
z<;QqtOYH5jAZ(pHY5l`mBU@vvI2Yt8Z`3pw-n&=J)AmB1QN{9%&65SpYpUc~`wJ3|
ziotU+BG~kyRDcVY%k%M!)(MPgBY5Cf{*3kz&Dw#3;)Qqy*HTo1e9>CRSH#P_WOPC)
zl}YmJ;36&OYgK7DWIf`sw!iV#B7^TjKn`CHwPj#M5i@zc;;qxg2oRf1vIZ8rT6s03
zQyrrl0sDK+$Tb9}hSyCDO)Qo-VrsbBQ>-vq+MD)bK|z7M6{E+2x?X<gBX+5cXRPD8
zm_QY<#VmO{{mv4Ayr5ItPX$Hf_r9`rkkYcxZ2%jOyki3zSugJ<8mh}y)uH!-qsD3S
zONjON?a2t(TW}d>naV#HWe7t9E;=h8_*fX56DeppA}WfMA|Lu#7-UEnEffZvK!AzK
zM_ml1#6yvmN6_C6NEMJc7_mk(<d0ppT29UkK;2sL9?A&%{Wy3OY1ond>JuM^Ei8nk
z?mSC_&AjiQreU<&6SW$qd=?D>)R(K1KgG&VJRPBi4f1&`ugZSdCBCp1Hs6NjAfVw(
zTRPP3$)U{bSN0bHVj$L^jZtuE!fyCW>>Wg{$X{cL6#Wl#J+`W01ODa<L$XjagPJdY
zk3qr4Lx|-+63gP+>}t!`#`2>E0ww)TYQwY+`cMDqJyI-dOb{q!NBO_}7tju)AO$1K
z)4vmwF!hgG(&by5Mzxx5zw;M}7Nq}c14O|0#@2m~?I+Szja8FjI>3!=yRHs611nM=
z;3lWoUoYEt-Rcn$Gt_&Bs9t3STD7Np6o5%`<W#dZ5t>_r80j6l8k*@e*|Dn`wlo_!
z^;38ecIxU@pRpOpBe-u_B&eOg$gWv@5>Rhx-_>DQ74=hz$><Ot!jud>3dxEg2RqtJ
z$T|%r1@q~Y8d{|;zD3y6ox8dztpug)E2K;OS;*7@(zR<VUJWb{eVJwL@c+0ma9DRs
zb3k1YU)S!@L}f0E7JJx)_dQQ`?z(kHwTMi?jcphI601kf$7Y--yLNRWTc!xP+Ka$9
zyZKvfBO!ih^wf$srLuc!hj75zP=8j-9x;Ln0F(hb^H;FHabg(IXwR-{Yy?PYmF(5k
z4T9L274^BQOER)|%v+NtE1xT$Q4I%<Uw&+3XldO10CeGfx?23;<h=YO4R{5>DGa2z
zT7GIj_c^w|?Az6ihFDw#-%gYLtRZL}v~Ta9zJ~6Huh-zNen5;4p-?PN1qps&Jc8+_
zoNQGr2gR4*5rPt!3gKX%gk36wF{zK)?}s>hYa)LvE`Krov8T62paX(Bu%)NVp=mQ%
zl2|MlSsuK2SPTaeqy}euZKWLU4VW*9N8EBb!k<~(AMt)LQ3V}O%~e&&QSk;8I4F8N
z`lVND3HM4oV-fD&@s+0a>*H@2&cdeF*9WWD4LiMjYW-3lGxVAZ_=6m6EPB9ka!gE8
z_j(wl5V0ct(=W`rhKm=(hh4vp^<S`qsg^)J2#5E8SZy3nNHCU@<Dzq1q6xX7mFf8S
zTrF6$AYLhC@UNT@Z{wE7x0hf|PxQCx0`g-Szwo3yDI`J;xs(9KI5};CdWMIcVsk7w
zTLyZw%@Bc<QLMW%C_Q5Y372(&436zu+fXS(Y`=^v7(8AITF1ppCG)JxP=BU|AdH`e
zqG8rR4a^6XGCaMoiKhg15%QyT-D;sBIl|MC3L#Tk#T0mQs?X96M`}_G{-O$Un)5X0
zOihZ}>1S<0p$FZb^JS!WqsRt4+knZ93Q9nGg?)T<h)QGx=NYKiw=v$MS?EoVvAr-h
z)(0A2n-aUws!oq@5cUeJ2Szj>dzLxr$HiJyTJng(v=c3h2X#3jscC!R#dExm+L9ny
zg1iaQ#$XiChGqMVSdPMi!E&a3peQB?1<mbwmMu|BW3T>lwl^5lA^!86_@*4W!r?pT
z32|<4Xb~Vkw3D7^Y%*bxoFA_vj0J9LHK|+>&1&gnGs0YGUyK|f7x}N$-w1+&9Z51q
z`zerjRH=(>5+z7!R3^pNIj(sYNGe$;Xd7q;mlfwsE{HKs10JQsnl*?(gOtX&)W(g1
z%CH`l#p<|!tTkY?++{9ISh!V8jQ8x2$96EI$#j{Njw6;=BVSgeesD{~(6~)1tu^M)
zg#uOOZ39RFhh{##k4Q5d@dR&H`!m+=q82$97rTr7D~7(VmYSGs4%r22N~Jbh!LDM6
zGSlPL`7_-Wc*bC!`WSCf4d60d&B;b#z%J@7Y4Vid{p#8VI63U>m}=)!UlapsaJnLX
zfbiK%Y+PmKS)y+*jd7_3B`cRX<zlzcjY7dVTP}Bghu*A^DxNI#tr6nM1WZ@NNRht?
zt|;{MGyj76HY4(w`>EAKHchUykqjMw=D=|Ua#f7DpImMKS&rEKYy6z;-(Rjx<Y6sw
z-t-pBbxCwEM20R0Ue8)D=hw$KfGjc>7(*6*9;8=Ti`4_ZA>P9&MRg2u<8q@vV;bo-
zOePH?+!V*6ZZ^g2-Q4C=`k>d~=5~vX8fnwy)<jUt_KLzKernxj3yI)GSLJTEug)4Z
zQtpUlK{5)axxLf&*w`Th#}~+5wmZ)|P<|0K=T=vc#obO2Q?A;=PE7LMV*=$g!b<M-
zk=qSz4tRH?+?Qr!MLjpc{Ct1xCIH$fpP{^CHCJGtJ<!$hgw<uh@n|wWtowtlgafC5
zaNzBSY^?`D0fpRq*rwiEBjOSNgqA9g*q28`#1|nxu2vq4{RH)GlQh~qdJpK|Tc)Kk
zYc(4z^<tSGlC^i=-ZH}oi-2&M>F=6~%Cl&r%nBybONpiBRh|bU(84o2sEmZUb=26T
zDJ?1u@@A220)MZR=GOL=A>pW%IsVK-@Q7wuCv$B!a4He+G|%6#6OIhLT;_Wpng+*A
z*`PYZLKYbHki3C;yl|;oXpK!|&h7?TWV7vg`Q@@W*a#P8mNZ=A{mg;imB8~Q7|Bwj
zRIyr6m!+Z3!u<@x0V$WKXTZ>cBy>=(u#ZqZ5-YPZ=!e@gBaj*X%`ykH6yTJXRly}d
zn!`$lqL!twP^|W56^>+I$(op1F}!L}8)U6FZAyV5Whic}=mj#Gc_WO(fw2b1!g5(3
zXXc18vLUgq<rSsmP|+qEW6-+j7AZ>AVugt)*%YLkV)@p7={;FWo+c#yStpOjGqs_s
zGfbZFXI40v%(PTC2Qv;U7%wfhJef&lvc*OUuVS@4nTS-adK0R(R3m!R$4^FO-e<cN
z!_*>AB}t@V?z}=g9eg?%p*dQ{;+Zy|(?&y&qS@S^Z9`aENi>HJ<vax|d^gYeZkvu1
zf0;b*lQLi9y3Kb&$Ng3G1_KNNbxP{w1)I+1>tI8D(R((~JFWTDb{W9*QtaGgddaW7
zT_hR7h0*H#{l-Ve#o%?HV1sa9aAYsXH@N3(5fey*SNs`M*YJ+D943+9`m<TE^YBoZ
z4YcK}K{P1u$goa>$ZNJXxLjv*^16);wS24Pjp&D8;(PANH=`eDr@+Q&hWoed8yxC+
z`CTv#ONhZ0^R}%uFpcv2G#ug-X?$8B@0cLJ@DO>|$rWnWJfNjbV87SZ!5Q^cmQVnm
zWbB^zyN2CLYf)p2Ga#8Y|2*UiUAUAKfACeDgV2N3@<9v%7js(XH$h$hFji5+xqw`h
zk8DumhR7ddH}XJd6{e59fhI5t<^UqhkWYNM7(LI+3r{&-PI<2*a{p5sLOt%efPZ`z
zTn?6RO*8eUG_XZr6U(RexqSyeF5p1&MZ5uk0G&r`F`X}i#n3aazx0<X3FIqV5-93*
z@@IdGS1p!GBB7M2<S&WQTI2G_CJ6Px{AS8uV+4~J%=tHKvz9S*e~;0qYD%fJSpMPt
zaPDD9v=#h1&apA$h7^pGZ+uM{&$~+gX&h3MLjY$;iLsCp{|-*NxL!+ud~548ArQ(9
z@?Gp;jjxOyqE#}u6!N{b+^cV2k#1&{qWmya+Dh5ZpOG%qjF1Rgw)bbebuU&`X1V;J
zo7<9@=%-G0h|e^cg@P7k$9UyTAPjS|Q#Utb6=x8GR4eVexru0SeVw%LrpCCD4Cb$r
z4&B@s=iprijTzB8+IzqxW%Je2DVjJNrnAmI1d0|`<YjTtF7`UcybcDKuHD>RTg+U1
z$U~``3D8{rC6#>oyNR>yZqHg8S#1$H^JyXL5uF2WpOc+!J^*9YVFy;qF5MjXh0}&*
zGnm5ID68BhyZWTUPGV>&ZpYmcW$J5S43i)Fc$#r!qP`7I))lh5y;G>yXIRVj=;ns3
zYzDQwa2<i|D-nQs&uKp0Bzwj*pbxNQ6mJT9#isyO;X_vQ*xTvExTuBlV{b?=xM+xx
zTKo9yI3rJcr+dpIs^ll#+}Z_W9x@_U$WP+|Jh03UA^Uo((4uj<Rz$Gu7YnU3H5CCp
z;F`35H?_0%J@!~Rz&`}x^uzVALmg-<pDE_(5Aw0K%~{hH3wLnz%!5Wn_8d9HpP2@s
zx`XtzXDy8t_3)B2xiJ!&#hQ<$L*XzVuQDU%Z|RN+ia*@j3)Pw_!jA}!&>IC0P|BmZ
zYK&%L&Lh+8SqF+rB{OO%Xv<q8wKW|yL~^gxHwC$JDKs1Q-Md?;+YGEIMU_A1Qt0Dt
zsL9ABU|%B}jv<h(UpF_w!W_c{^R;qxJlSli`ZvijvBOGX*XeJ6V2?sqKQ<<8$#&H;
zG{9cKu@5lFafz#;LV_WVH*rvlC1a|w6P!39p>T%lcA+noB{NSn2DsqR1LUL_7F3}6
z3M8wPlVeq|F{|Yi=U(7ufTbB2VrM*%eUoHRH?{s|fclZaHfEL?%Q0^!QG`RbAvNR3
zXGwuI$fF61J~SGvEFU*|pbU#Y2W2KfM<4FrH^GF-8t$wjIU=>GwIv{Tm&mE<Jtmc*
zsA-ndte@d58Biu8eSr-uul6jLdZS_$#*Y~yqhsz6ym=YpWQ2ym>o;N=h}xLX92B}F
zV`CK%16L1zK0Rnz0;DNeyjI4=lg)R)j7`W`p%27BJIr;&I77x7kM~J1BM&b)Oqw_$
zHhNJ7YN~-EXZTC55+lrTp`6(*Ag*}@1FM7PtZrKmwbTp$C1=Zj15V1>sc8b27)#bH
z=hy;kLd0`}HY17<<9c4~7`W(zxu9XeOn-j2t-a7QX0?r>Pw`W0R4qh+BHCwQ7j3;v
zs>bW%1x{^~>pTysiZ0;whW7=z(3rNSx>7C*vH%UP@-O3~VK5YHv#(R=BePr;`7dl0
z@(Ef-*qe*v8(g87Zj2S7?I6syA?W0C4SB1TTsL(%op-#H1jit|o0JCC^d7^5VXl<f
z3#fR9Bv<9$4V5)mZ)j7)v{_(kqD^Qx6qHH9_rQ_Dv|5o^!OBlB#34}`bGE96RZ^82
zZdKQ9{CWSJFc~$MVcS#(Cyr;n0UlKs+k|=^EH&2qnYjvRq==b}g$B~RPU>uw1ILe-
z`k46O5d&p%nz)4?F2X@N#fQXYB{miDR!-8X)@B|6Wyp$#ATW^h<<QjC?s|!95S~xX
zvZ^7k+NEhWyh*Agm&-!>YeE6KhnQWQUSG>Wa=Cp2R3jl0yepx!r?+GLMZ>^S`B~7t
z0A^fNDDYRzD`2j<s}h?~Q^Lnovtf@}BUh)lSqckEERLOPVwHwqS`|>Eu8jv3juN;o
z6eZSHLZVYB4{8&>KB!(;ehCtsdE0alGmx0U<2NMTq8xe=>h)E~jkZ5G-w)gFO}08T
zuuhVj<9i&BMB}c6c6y7?galxo66P*6*Y2(Uw%z>n7r8B+Ayx!hfp2Zh0JnQ<R0hm>
zOxpt^749&`y5dN=vvo`3t{KSb))diq1qBiN21W%IZ=L+YmV_yVV9}JjgDivcQ!3=1
zwC*Wb15-$8{p8-(X)DZwSpe=c2JQ=U=KX0c!7@Ax721iyobrIZG_kl+9*h+$tjoc^
zS0@kId<K@t!`?lutE^Q*HI(+2a8W<v?ZYC@D-&RX#1SF=tUMai$ijLgkEL-Cug&u+
zg7zEZy|DmZ$h6pSm8c~$JrN7xI(nS1ml-y+(Pzs{<Aj#R!Dq=VYXn+&8E%lXZEBZa
zaHuo|dDL8sw61`{HXFT<RM_KEnPV+hm&jaeaa5*ulFUm3Zq1Ko$L##rLD1^pBVG_Z
zT9bnpIFIYTEc8z8o=T#xltuoGsUiRs10fg6Vka)bQL&pTLI`SItfW~20k|Z6GZUw=
z-JX`FwxLJ+nYk>UVUd6{g$!IO%k3G0XBy~0DJKwVg}+DDcQM){d}yVQ!9X=;kTl$@
zmAhQ!znctkv<Wl<$|WE<WVN>$H)BW{Srb~qXp|%ZD0OX{eyQQZ4J_+I%Y?`+uhxij
zyt&@i5!f*Vh{%R$8yk+WAnFGj{TcY2xB=S4WK;BqIAEqG|D~^;u9xo?J0~7bt4b`e
zTIDew=FfW~^#zb1+C|MP+3as~zcHmxD_W!_b_@!=An!_6w)p3osvlQ-7A$_!Kd1kk
za)&VrHjJ&&nZ_V$!Ap7SzYv<o84hpqt~_m|!7v!3@Jw1ZW(dI>IMAD*J421Zvd)ob
zjY*kG<%!ux%5$m1O)$P`+>Si&&zR$dAzCw`$^0rk<BlV0V~xDv&j=^uy`DP8i~h`R
zA?ERTDZa&OF>VNzmF3s|OrvN)X{K8FjqT1}ef!DFX*rmI+%AYlp;tl<6d=p}Z+)^R
zPN{JaP2DQWUJVgaPm}L!_9@s3!3eKA)v)cLf4t$tvE)V8)0?rKvw4=Wek=7wz#Vj7
z2ZTle<liO2QwQNu0F*1l-YQ?(w`0cOOgL%zz5L!r#6a~1dB<Ovjaq?M;sqwMd^bJ<
zqY+y)4-tm%g<OG~84*hid3rx2-pC>HhtO`uL6@ima!)>pT~M1rP+?X+G`Uuok&mo@
zSRWCT{YPsgn}-$d<J2#qdd)Z=7sx06jO<P_Q{Vp7*GVlgcn0lEWq7lEmfqf=JdQjb
z@~0RG-B+-`d>)&nEKgukseBR25R>H=!4AsuvtN2kBq@Oe8=xS6l{O$yoM1<$b<xcz
zfA&SL%c1PaU(z&a*|XTQzovE4!yJ=g5?44t(e!$3puhPyl)9Vn3^ptIyEWR<I59I>
z{$X^g&7pwA*G9^LQ{@}~jN*p~1{TWur;n1!O_>b<NlTp>{*~s01i5&IL_G5ESXX#3
z^EmGj&-rbzAnHJ(x9?(^CsfGyX==D?aG1EdOZQeo8*R=^54K%*y_dE$j+rFeclVn(
z4M|!Q@CV)90>-#bXKIrwJ9OX5Y-B|U*ylBRM{nAksTOjxlYeZEB~E!77J;d*+HL!b
z8qv)p+&&Evj&5|w#Ks-Et5L3{aXg&N($U|bFOJ}Ts1(x4pZR$RaKtR>+}-VZwfP3=
z;xF)k!kt17gsyF#LC0eSPU&W|LbKSUs3s88hs}gVzgYE^kjuH!BS^&@)taVbXMY#g
zXRtr#2|0+1%rx1>J0h<H(q-460GO8d4hAZzEk$;V*&+c|gZ$9H#TL_(oTudO_6AyC
zg|lXlGz}Idu+ja<pIKHc{yS6lw5h=@mqA(z*{i!7?pVvT6!~AWvbTSw#sjh=Hqwvd
z8AeDVHA5=xv&}Q6e_Sm;iTS9xmp4+Jg#@X{PrIuHtr7;seS@5|{_|%wgE0rse#V8+
zHyULB7${I?h_pQ*eFTRfMnG4P3E%@`3J?Pxt$L7=hK40bhJ&po8hU7-JEXf?ezQ3S
z&`QrV=GLecEq^G7rapPv)0fXIo^<aVX0sVHcDx)OpJNFb%!Bh2<VU^|Il>48s7yH^
zuW-G9xJTLyafd0Eqq?h+5}`_^7EjvCo1)2uh(f!ia;Nl;F%!KG<7yw9QZ-cLDx?wW
z8$TCS<e5vbpD}%o7brQ}TUBc5_}vL3#{>sKFm90k>AkjPEg95ttPNDtU@%@*23YrF
zYiP5sl;dKl7=K(V$2()-=1wOAHo_`7!R9@741zXJ>>j3RqB9^Q;F8faM_~PR&^=DF
zcPpR|%gH_{WoRrohi1!uinqcA!CewZ=z6q510{SGxK|ELzZ-r%F^myb26b2K!~qw`
zU|Y<Baie93%@nHxgbG8s3hcKE<i{OqMBRMNmTPKdn9mUi1KiV0u^(<Njyp?6_^3G^
zMV%|-(3E;=@C23-@&cFAtej?jj77}fNM8|VK$|Y3q7&?Th9R$nUOqaOx^O&D7KJj#
zJ}w7SjJ3ftFfosOXr*#`clR)igC7e(K*XFY8h4z{q#CIxYNgOS(WJ9Dn;GXbJ{A;r
zlF}-f;IAl*2;)EoF*zeTt1ZJ30j2rO7+4XiY+)*AwV}V6yY%Exv%^dPT+UAHrsxF5
zSn2vrmvdUD4V?gaZXx{7jn)fMc?3So^K9BDoph3%Z~vLk=H-GIHwy+%k_){d?p%7z
z3CX+2-sTCNNh6@G)YL-bt)etxCP*N#bGa0Um;t_`$})Ooqq{<wl&d6XWG%~8Aunf%
zH$h+(Lq;e?OA|Be{#Jx9Jbgg*mIXE8!&Ed}dAw@KVP^Fsmw|&O26dPr1P!mkU3ro{
zYiYzGR7AiSIJd%B3F$Iv{;u@Z&>JjPc@Z3-)byES6_IXI<*l0hQd6Sj<C#h}MAku>
zuTJeSI0!nZLMk<tKH#TI<zi<oCgqZvL|~>_<0>$$VFnB<we}iweWIYJ_XCF}#4OxQ
zz^~PNKap{Wp@dW?qEy!8G>&x`0dGi1Zca}(-Xd&oig&Glp-mR|;H+24)Syf60mmF8
z4QY<kaBOnFZjwt<+st2YVpHRQ*aA4=(pVd{2@pd<ZI&3F%hH#I`TGyaBbWP^CffN8
z4^zgYUlIJxO^U7iv;UaCm93%y*shyb8XF_Wv0RnNhm>FVc>@4aSNk(u4`^`FbR5^D
z&NcY84fkE^Z?g>IYV2>u++7!w<ef{cz_hT_op5~&4d*)H63~G4_plQGJai((TTl+&
z5KnM|Ohp<CSmA6DFj+##Z;B-uM!|8jw{Ipz5DAlRlK@0+@zL@ZRNsMlc%|HG<g3VC
zTnk0CPHqdyhZ-KxVT*G>Au`C|_LvnUDoeWZ_C;vVT)D%4!6g#cDqNv(aaN<1+?i;}
zgg+1yI9%?Eb_k^Aaj*UdpbF#{{<eAEXv$h7cia3xD_qs(p4Rc>uxi4Qw^8o(_E@-;
z!ji}Y6Giuh*s<I#B<KCM_NC=Hc_1i<+G{y^FwFrqAx)3O*p^^BJ><V|#;jPxNMdXM
zVcUraUxWJoNE#Z!25OU*N9~1{#?hH-c`VI)365@VC}dXEP0?tFHZZ`)$TV*T>C!+&
z+?s%NpWfZQf&leGK-DQMRA!`kQz19a+Qv0A^+Y9+vV%rhLUU%>oHMBEgDdE4?~9yH
zL<4i*G=(%RoB}i|@V?D42l}(;c*_Rev20Uz6`1R7G3<bM3R)`WrDp(}gPIE=Am&G(
zFr{z<!}uTz>?|0UqoFw~3!RBgGTEJ!!)N|-<~&&BO{kdz+RK$NAuW!jrgn&=B;eyE
zsTm8$Ci08R5+0zXF*>y{fxlp$To(J27&y#g0luLx+XGW&xqrj`!-Pr%KCVc^A*`L0
zStl!<&P;@rl~w+Rg+Hq%Agi5{<bz0}VXDv?W6?@PfUfobMCk6iScbA(gRGB5LbeWC
z7-d82xKxQ_(v^+D1-f7LJ_&wf#==c8OJD>bqzQ25K4icvJXW*im$5Sta+{OKy=~qT
z;g=+SQ=UluF%<#lyiqp$GwxeL>X}xk#ab!^!Uqo8Ej}U|>R8JFQ9r(uo{Sy~DQULa
zv_~Qv^i%N(Fi&*3)4gs?{&XS;<lZy3gXrLpXZ;f%5wsE!55(EcmFME1D5@`_!}xg{
zF|%URDFh=FvYpvTSTr1L^+KBaD$I#T;S6~(=!yfJ>7}Vw8wTQ~pa8gskUN@}e(lff
zerk93-*^k#czWUF*&r|52qz;qk-D^BiIsxc2lZ+ZmN^&ewM=7GX5rHJTW{Uwszw@l
zHO8krY;kx-ve)7lkdek5reL;TUXPK3<7w8Qb@>gW#7JOGpc1{=rY9()P*`TlTmFnm
zeqfh?F$9^<q9vS!R5*w@c{}>`U4?39>;wDl_wlv@OJU!@UFscMSl$(?<lRI7)w<k9
zXSJX}2zk%$4TC4h`@V3{t1z@bc$b8OXewc7H#XG|yhS?}Z6jM|8QT69%7<x4$Z&#%
zC4}gs^vyb4E-ht5TVu>0LqIb|y&PtDbuoTyEwwb_=q?2ag|P^q__$2pQZWg(;?jTW
z{c<T;l85lY&*JbeE(4Yk9$xuVuyA3fBnJ?{=SHl7wGD`r{vtKOG?*IsC|`!2T{{V7
z{p2g(TDWxlIZYI$it#oF@fT|@1MpIn{55*td>dxXAk@G4=!4zp)1%N@EPqc+Oz@)}
zZmhl7EdQ`3YjMcd$=9||2g6q&-*k5%0RUOK8moFN|MWpA)*4%g#X<gMD?0=xO@Pn-
z9iM5<a%k&3*1xq*c=xQ9?`)2z=8NQeZ_cz8Wd@NRYQj^6A9Zs8Aj!&hJ<Lew6H^O`
zIjpvCeY+~_MKgcUL+vpG;V|66KP1{jZ)yZ7t6P4@9%^z!F&&D&oqD*9a9EkNi~k-y
zUfQ>5z8JYZSZu9B4>htNCJ#Bw@OMZ18^$|k%deGA_6?}RoK^<v>fFOkb!QGABVGIj
zT)iOEfjYx(?%E@C2<XUg3ufW>%}Tc(Zo=Y0RZ{{Nv2^dDR!JPjS~RprbcPFt>Z-V?
z?Cj6f1nWBQE-^@WWtjVD*B)l5Nd6mWWuWdBcg)Ie)(r}z()K4mbUKDw$$(;V(IVO1
zJLA$&VIZn!kD$7v3e7o$vhDIC|HX<}X`KnMHn2OgXZmKE-u;l+na3e^*<L-I5i3}N
z2s#LIqsZQ|NatN4KMvv!0?GD?$>XR599A`wpV(SsbTn}Nw1)%eZ0%vRv+s+`kdjRH
z3xYzH0dfMtzC!koXHYSD$}z(50DIQb2)k7AB-p?}r2~BqZKCioJ4P@Z6tkn!Y!U_S
zP~AQ_2t7H2aB(@r1_xY5Rjq;@dm1zM=|4aYjX9q=QVz5KE*M=fP7b$MvQV0ju*bOn
z9V<ursONw+pz<k5>8K!8pWeNsm;J#Y%3A5|Z*v`aGsN0RpB`>42auF$wrZtsVjJE9
zRbqp~X{JUTFchnP{<}F#3BW`Vf;Kq`j`rVa6Qhz%cbKXtXp5GuV}csB$R*t%{e4jM
z?}m}!SQ|A0eVZ~eAXXNuJzb8A@iHt!jfux6x+{Q?>J!ck7R5NhC^ZCzQaLdwg_s&r
zYAH^xll&P$l+2aI=(m%-HJk_>J>7!-%PIZ~s0XgIqg#D;GBBn9io&ssE9F`l6bm_Q
zybSg)x3OM;GZxDbTh<BVhe|;YHRw&Mk)f$MXsci@6jYI6F)WaiyO^hoX)?p(ZK8|u
zHq%Coux4S~l2eU(NS{VwbDGm~Ho#BIWu$!yV>+;a@USs){HPv=dFS<&b75k4Rf@5a
zO$v%*mXy)KW0fc<JjNO=E!V`^V{PJjR7Q}~<26Jn%eWA9wM@>b_t`>wcf{yng)-h}
zs7!(`*Fn_S-F8Bdk5EQ1A-!D8=rBXhuzA+xb8=>U4yBR%mq2yXHqLSasZku#q8ak$
z?2tFO$AL0tEIlWhuz+HEZK#by+Mesb;C4%9r`rn+5xjDqH;#h<zv#_Z&JQ`9fiVk2
zzaajGTaw7<3yoe{bf;D>vcHP*I1p+jv+WqvV?$reGAa6xUlywIi}eJH)9705{5JJ$
z{DK?9a_Fu|nB?Lv=bMxlID%O=V6iEz_%EX8q$I}35DqB~af-q<nhd<m7OsTTDNiI=
z3I(04xke_&S|g$|Gbx*&QYDj|)khXcg%68zhyjJIi8d?KfS0Q3+M1BG%6s8;ocnh=
zWLG|hHI@mXaR;mRmvkMlZJ9*i;uszwT-YD5oJftW$mTnc4D=2Ng4!4^ip{WqJM7AJ
zX(Z|w(cPf?K<ZP2EMq{)xI`ugxdy_LAyeACv0N##UZ#3`CQsY=N~IxsqIm%A(bsFs
zcS&rY;ka;KYGSj##xlKMX5yl7v|R4rw=ptt<KS9+g()0ol;@@(y&L?ka3~>S)YcB!
zHdo4(K|v<^XO5ezVj`nh3G3=4%C&wyj~6T#tuNH$Hpw;iHpB#DMXrramCdQ^RqeRf
z*(+n=_JKZoeNaZF^eh-t5h@y=F**h^T^08m(xjLItu5l~M&qNx@*26xTQeIbO<VKj
z<~W^rCS$pon8_3%3zfMgmfEsHP;<RC2Bjux4r;O7=FfO=(f-Z49=Er7j|w&%^>W8H
z%S!tf(W@{_%bh{V!3^oR%a#&e6NtrMgg_|J%oK9BtqA;4m2yv<j<{>&Cds{R`lLXk
z)Vt50S#LO0uoCyjGu-)TNmz(y4+LX*ybbd!5XG`g9*kBY5tyxZ6>V(6UzV8mVW%T+
zL7~l{dzuijb~EIW;G9g=R7F=k+GA@sO;Q4dxf<6k+`?c=lE=I&(<%+y(HL{g<|^tk
zGR?<HabTDdS|;E<l<EGlUDYfUY=$3De4Abot7VyTX+kaoA|x~YONx6`cTxR&R+4%D
zfm%T;bWO~TZ;(?PfpQ4+S|v>`DQ<^2;`OW9)^Q-LKBY3p)@>Ako&OJS?;Txdb+n5Q
z2^~UlZ$Rjn>TXPpdoNhF=|$yOI<}-EbR^p{z4s0Q0wHqmy~_=-KtKWsBqSsuKqvtM
zgc=Bha-V0)mb~uo_s4hdS~uVNSk`;qckkJ=XV0EKve-XX`!+sC3Y9F0eMSXo8ljfj
z$JydsNk*2%k8n59wYuC#2HwhLg^k=;fB50D(ta2+sy}YYRXz<JCk%X=2WGoEJ<|*q
zK!1&_vC*|VBHbx#{T*YurXH<}Nrv{9^)^Z2%p5eO4J|eV$YZYy=SUkJ`NqYlnJ4p(
zUVu<{z;ZVwTv)4P80@9Nj&1*O<7Bh_vq#?^vc=y-J8qi3pldFXt<AGz8zRS=H7vFT
z%0OO}XSXL(purg;1MZ*TuQUm1-#K|aD0MglZk7dlA|}N-Wi%N29LSSFy<|ct!Am~|
zy!bo(RhpuZz=gnpRnlnS_w3V8p0a<Qg08bq8!~~u5^0JV3(++f)s^KL!){~!0GPL)
zP1CP}_F?(K&&3PF@WQX{g~s~QY<0QCGs|z{nQ4h|3FPJZW<-I;;ds*Pmz4a1PeETN
z(aI>S-x`3ZX_ArO1rvlbFb`c3jQ_=8SEFjDX5=Lw+A5{tzchMLERx^Hdkof*KLp&R
zg)`;l_z4MQrSeDn3!)Y6e18h~v2`#G*D{v*m4Ff<6k)IabI|0Naq<@%frW_i_FrSF
z5vU9KYAgV@W<<SGTl`xf9jykmO8;v<JY+%WW2Sq&?$3-ZK;QvYy<t<bT8RAJZq<>M
z@}}`9ZN|3k7<nt!#^z)4b~7wx@{W%Q$p`KyGdcdlUdQg@|K#0-&6G9j4DLOnXxw^P
zHd>MQV-dBe$43JEXBu2pZ$=Br2eB|J)g*fue?E+(fi__%$}1oFJ2c?a+D@O?zigp9
zuEYBuC)!~n%Mk6U^nS9_FXR_c`I*!#pW63BYGLR4%ooTs;05w|Km--abhribh2aR7
z8g7Aq`=F|_U^vEWs<~Uf3_x+hG}9QvePt6(M;$u(j{|)z^p&rDQkbm~mG_MuK2+kC
z|N0<g0Zc<!gA@I29MeAXo&U_E4#yHJpvw0NWOV@25+Fa=p2LoZl*UStR%*~GWMdrJ
zrIned%&kYeh^FA;s~@#eV-^k|t`YF7c^Li2t=y<H5Dq+SEC$#$9wAgIQ-j+b{We>v
zoyVe7FO=PVAnq2K&0^Tg_h{vY49%{D^>R;#Bf@J``&PYwuT~ylHjGu(_ikmzAFY+4
zW^1@1TS(gb*y0eNh-9gKTQ%*eclU(p_#1&{HY+Z1Mkj)0?&qM<WXw2v_UPF^fs6Ee
zkk~r-mJje*l_;?U6W-YW)GqG>(>QbCeo%QrTD4NUToJ6c(z=yfBbaxGL^>F;Y*oPP
z7U*ql)0`+mfHOoFS37^EVYsfawRbq&1_h0r$$&{<kPiM0M`cSNT=>#4=EjALR1WM<
zr&elYoD2kXwyi?6HZJRG>0-YW;Fp5}?WjmvAqV^R*ts#!gfY%Tnt#EyrfdQegCWRI
z9T@ARV;QQ4+V=1csFA}G?U}<DRyLd-mYY31U>*qfP7wme<Ol<l{Z+GI<VOa=Xv_w}
z5SlGV`7=t1)QGl2wmmuqf<Ckz|G!O+vA1BO1uob}0SM`8%kB-s(y=j!dIh1iKt(++
z=FX$VXN?^1VAHOx=*QUvr@=u!A?OVkE#k`L#8ypS`mq-J6AJNJVIw&yjbOe&dMyl_
z-JC3FVP7iUV`7jEu7Vz|%!YwY&?)JDf}#tFfa<R1U*MALX><pbqeOc7YGT6}DA3z!
z$Y`E^FD3&vC!F?)jqlzM9fSJXXpQwH)8*tAxY3FbsL#>QpE2fyh7FdNI3@NKas{2g
zq*64|!`V7Dw!dJ)csVVGgS5vBhFPko`!iGeba8TqfiRR63(riL;~il(6~o-m5@VR7
z4eT1!#<Uz4yqORC`^cnV#=GN^lmUSX^}|w3(+bxx4^rK4kbwz8Mwn7fhG{F%Q&kJf
z8EkKWENg3W6^Q4COfV#UWu^6&<Bl1cP(s#WFqm^@SnyAkAmCp{+s#-cFh|L6D;=-O
z6GjC7#)D;!bm&`Xt{{ZvdRa5RVN?(T?#<D$je~1sO#FM;I2juhTZQr}fPI|(maRb}
zhqGF#<!;h=8SgWZ@C*-38qzRA;B18y#3(pV>;XI>QSmBZY+_mt4ag)B27SYsBa;IE
zT56D3Vg4d|1meH6)|~B_q=^gJ@?g&^<s5&;vdn~K8#Kam<CITEmsc%6ex9A2D)wCm
zM$h+;0`ZFh^aXJRjV-UMU^3r@4mI;fHpoTB6Kcxxa<P9+eVg2QITYke;u#|I7~Zo2
zXQ41gWLPQl!i$_%n7MkZ6bF_dXi#@938;(fYH|oK%Sgr{fVzN#I@za~2kC>kEA*n3
zIxe+>BKCk?k!ch}uBO2|1AH|b6H#P@Yivrek?FXLYGi7RtA6?UPhATZSMJXY(;CYl
z6+V}hD54cD9G#VEM`@W;-!<T4<aI7kFc8HMkkc<!23B$TRJqiM7o9%AnX1#)v$zfu
zT`48O9DyyR%Re9ZRadedm6`w&p+V|+2hM9_ZOEoTJn}Sug;CYZcC4U1S*H6g)FT&q
zvdl1ma+Nu$vzL+S4c&;m8I&?J7BHbKgKB6Qxy;u}(m|{X{>w@wR2eQ$zqmOM_%Z$d
zidHF+nWi5L@4Pa=C<2knRSqc8%7he1?s9b+ka0ewDXdZ@xNG9vS0mcz+E_9UEXtx>
z7g&MmY6Dpdf|;&Q2No=g&mi5H$;UvyAz;~Y1DxnL2H7FoZ>`+q<5}`Cb>UTVvqMI!
z3B+c#a@k0?1p6v1wTS7A-0Ck|yfk<_?K3QZF1Iyrx?We1zJXB-H|D0_p1#}+x{6DS
zxW2=7ZPeJIa;L**)-i+Pu7pk0*~5_AU_Q7zaMm3Gh4;jw;74G@90j9$19V2=(*B#5
z`y2pX=Q!0E2aO_hv+uVv1rdVj@0;X-IAiF~#8z!TH^zxH;3|30CtCmmH!rD;IA9MY
z03e=VS9#ezY~Nhb^$7H}w0~z%&s82tt653f(1Z1806PF>0J4bJM_8EQq)*?I<uU)7
z{SbfUyNMiDn7`_MESiF7)y0ZrmSG0W8a!uqS~7dcX>B2p3JVX+iDiKM!q*Fazh370
z+Z2Qu^vJ|$dwZV$Vo4kzUfA4ex`h2HmQGS~L3Tj`OK)Qp^cVUwW>Zi%W4Xvh{tWhQ
zd~3Oliw%6%9LviRpH_DXAJI2QP}j-Q*rEQ`EpS<|+e~3KJ^LtqmOx&kK|||cHZh~Z
z3dbjc!DVH7i|K-fAft&@L1Ci`Yh`tud9`FrgKZkzWKEp2(QNRx)<*5otw#@8mlj5i
z!CDNjk7tm{Xn#U>k!*+&V5ZwH8`GLlcON<VoBRtqs}z^6lLkLznxKmH@R|<!klkrF
z2dw~mTbx#Koxtg{)xV@82thZX70pdj-oXr|Z3#Zh_T@L*{TZsafvz>hM;^ClkQuG!
z02TfxY}P^)w3jD?ma~=A)7VUFha;KnOK)`ujvB@+jsA<KjsyKnQqv4vc`E(VoO_~7
zPM!`XKDOGb%r&Jw+z7Iy)yu*p(d5$MH@%a5c)dKEp4pYh0Mh5uv(*|RwG8gFUk4+e
z3^z@s{KjA58rTkksWZ>}GZsi>3dAf7|ALWCL2<4Ib<g|-2Ao=3w%PJK!&UcQ-MYz(
zF#xy_vrQSh^HQrOEjDg2TWBn%17~V!xilL6y?;;7KMoZv4Cs34&&+nsqqR_8PS4g6
zL*N#(<&Tav@YEEW1PcN5vp)rbVF|3Q03p2+sDQr@xh8)OszulCO8HCLD4Me&a3Cjy
zLd#0|Ypjl<ovv<?yc%QYbLFf!BY!hFbmT~REf%919)VmVuLsd8xtg*;j<gK&;w&)g
zH`4Fidi7!OkmFGBT3jZ74^RuBAIqEZ*XY9P8S<6`&g^dLc$UIQ-uSkUYGEKP{5YUZ
z-dFF$K?RAkuGK$mOLyE0x%}N&pf(DGnE0MC@<D?O<bA^<#58KA^3PZ>VlUt_r>r0!
z1ga+?W~o{}OdE&zOGV*2`N${?DhZ5w|MC|!8P6B}ae#+jzzkCQ#FlNoBl$FD0QTW6
z`z*~sKQ*FoLHeH?$VYeWDqlE~h)#=#v$|4C<lnY&XN{LH?e(rFoFQMe@@S#T92+vY
zM*N?cl5wFW^0iNC3ZAMA@=c&z?Vj914M0{kj{$%GjdjxL0Wew4@!NQY;6r*W0ik?n
zYh@-86w2>oGmwmnblW=l!FH;#enhQEYc=7F$BiMov^H}X)n-`NA+VzAzx_TWmqX*q
z)@r80Cas6Df7f^$WR64#2om#UH-DzVfS?naP<L+~0X=01#lgz=XzeDeVl~#wp8gSG
zr^9D9B%8fraFfGODrSZ3-CAury?b?&ef(#wOmA1B?Atn$yC`n6N*;TRr}$v`i9wG#
zWfVa63rH=dS<T7&r)P}mwdy1X1R&TF+9#=p9B6=4qvTR6|Cvrz$U%BsTieeQ#*UFT
z_A^>DqNPl(G{=Cnwt+=ptU=nP@sa+C@fo7g-UjG}g?6x)3oD>&)JVtHX33;;fa+3n
zRgg~hUSs_*l5^)YsD>zO>`s^TjQO|}Ugpa|HqkMM9WDpA_7FI@0OgRFM`MR*hlEZD
zhO|z8+FBLn^G}gO?OXOKJFK-aYYJdWmcX8xNq~QrT(szDjeL>A)8<2g1RFh~wHi_p
zMO>Cajm#tCXQh=oIcjI9sm{sKt<}=V5F`BmI<V$r{4EvJY&z+h*1%jEO1lirrPO?E
zdV3SCn_64?IDZD;A5I(o3t~P#aN55TnT;p-cp5Qg#DPUp`8v@FFXS`Z6|Hm*DQB6S
z<X>u51HPfzUfHR1v%`ge<r?W8D<5USI6Vx%J$v_)p7GNlt6kO0fJIF{#(vhI{8pv(
zj{O^hw8%d30qoNYq3`r<K7yzhI6W^XJILhsTE<<Glj^iczqBMI(epafykWD@Go$zv
z16<?HPj%Gs+GIUD8aGc%&o)5KTZDM$WpcWqvjmy*P!7-VH^M~6tMW{LMne#`4CeT)
z^0Qb9TAQ-mTL1WtCk13o>3cxh8W@H35VBb>dCx$@3iETTWl+3^U7@wf;^J7?VQ_1=
z`NM@s{~?&v5F;|2T(}UJ)+j>*I@E6f`^TDA%djAk$+S8Qj|sq-nfnC8Bc`D@e1w1N
zfy<gEIx^m-`itYWM)g5t|HvpiWEY(%qm2kqRa3^aZt6orKMgJa<NuSfHW|u=TGybg
zcI9MJ@;KYd<M2N@D;ABmI+n{f-T>&<^JFOqc+~|&W{tcn*v62wnf63`GJ;X8mx-<Y
z4n!L)gv%n|l)mgq!K3jV1ti*57{NrNIKMjERzBuzIVV7wfI@!Y4b^gP+N46%Rjgw?
z_<3pKIlA#o893k8lcAO9g4SV8X2}`Yl3|=PAzLe#fGa>DlnZSW5mkdINx8`R+Kh}`
zY}7N37B;!W=TZd?WR5acQW(suKx?TM`75A2nyqnT%#-3+2mCSUSpZ?B#HV9m?H0)x
zxH;DUnA}Y3T%nYvQ);Gfp1&4>h*Fl8uNdT2DcPXNk;BIfk}3WMt!Qj_j;tw}n!Z|v
z6U1jLq&)4dcFF^rr6@svbcMgo{bn8%Gma|#8Lfyclg7M`oa3rGmzykA#&2jBLZJ*G
z?^53gs~^j;kOjefL9B=77Dj$FNOg<3nD5T{^L~OW*>1eX{y>!@nEYyE9k}Tk3&0Ak
z(|jGlidijP>-NcX`xNIHOk_s-l-^7`qIK~MEJf`}n0#h@G!BmI%Y1C=U-Y{g=j7?W
zJZ5EFA%0OAQrov(VLM7DjRy2}a%JrB03?T@;@DL|K;z1oQg(H0-Vip9xF%i(`LP$l
zwP}-*ISI(8?drOKvsdqKa(#M@HABc>X)YjNxglN~i3qG4eZ5u~$)aIziWw0!P1*cr
zABqJfkwAmM6;}SD@PA9(C>gYSL`Z|&8tB8`vKG`v;N`aX927#7BHIzzp>lP5EVL>M
zBgGvtZGnaR-|1tSK*$xWlDpFQC~#nxsNC)6!y;-Zf}v+DXp%_;8*z8s6WA?8g~yEC
z8%GgNT$GxFWPG0wO2-|q!3`+ybiY3%L2;YmEtUt;LRhni1i)ZcY_$B`K!?3-BdUHc
zk_Xe9sn8k5WhcLf0%Kt2uxggb!|{NLqLXXO<re{8z7}R&a7=k5*d<s6!z`jZ8$&^n
zmo1Ol0t=u7(Lnc0pPU+Eq?J>mTrZFLFMclYJG~opcmK+ltIC9)xdl>hbW#cHaGA^s
z9BBj@qaRnv>|pK_GR0+;x$+z`Lh~}mp~beTdxK1=Wpw#mXM8jiY?paK0tI8n<z>DD
zsN?}KsAu6y0E1kRKt~ZRP#-GM==QIag+3k(`bg|v<Z!AB1-rw*nZ>P}EU?Yo5jcZ2
z1ayh556zE36-)g)3aPYGp?{?;3xXVznNEA-^44zbM)6P&bIi+%v|X7QNIb#mUTNdj
zpvz^ktg@?Y*l<}LD29f{i*6gJY)#CnMnp*DaG<RX9B2s>8jUGLFtB5tU48J)q5?s_
zrVcD@y)T}jD-E(CkXcP5+{QE^<tR0B$)+TuCs#UtVZ|0&Fsam1cDF#w2D_8WAgRgb
z0MsH8wG7>scm_Wf3qesQ+G=ZOv4<MjmR{h!tT7|9Jyr+C$%5J&VE$s$_096Q|H565
zdJPL2SHaHrMEV9+Po)RS@H9F-8OSdN^;L3Hci33~qttrEjs9tpFKvxD%uhL)(n1cm
zGL&pei<H5z4Y)&|cEP)QA8GP+*s&q6T_ew=wV|D?L-VYmzuF}C=X~loY{>lwdRYSE
z*GYuYmd(uc`HjC!aTI(LPHt+MQ2gb2pUAq5+70<aJcDexhERjc{Wbx#2DhD-Qj^~W
z<3Nm$GBJ%fZ>|XRVh}i+lp-<Zr5HmQ4L1TUHv?E!|Kb_1lRw0;;FoS7(3P<S?Hq8w
zm;Lva@ysoqBY*V8+D_B^LlKfH@1FtyTvWs=#er9h49jW5_;d4qLp-lQ9ge^FShV{x
z#*rmx?QZ(3oxBm_<keuY2$q|IFlqUlPXjDL<DscxrM%`R34t+SORw8sgR~N~yph1N
z&1Qzi-~AZ_CkQID+~16w25EL>S$QjcW+K*F2xf2FD~O@1W6`j8{FNYj^gx$?_%mA!
zSRszUyRn2)R@sE{=Dpx)yv1@%|4<{(`~Iz6(quB|J&?AmSzsU`LQ7LW^9RA!Qf54^
zr4JpiE}Y}EQhcC&_@iJl18_0OzXC9nhO5rO|NgOYfU^q`Atj$UI7|XpYXhj`(;#&S
zs%%m)OT9$QrTHvIE@6+X&;47*ZBSlk0QVPx;2k$NZqCZT15=3j-hi8I0b<y{j6rsw
z{*`=X&=5i8@}C5aZkh7buhS7g)HH$~zp)P*>&NHH<-ajL$Qt@Da&s5Tw+SpBXcF%R
z`7V~k&;>kONr?-*WV3u9S0vKdP&yxOD27D+kdV6qT!*Pn45gw#C2idPatwOa@6smX
z(iTD9uor&h&loF#@L~AfR?Cmus5NHdNZHjshnAI}EW5QaGm+6&3s;Bio_=A@3lI#h
zXnOV@ZTyNG4x2(1j7U&}WltL)mD%?451pq`)u34brhvU2E?ff;Iw&vAk$u{zp^!D*
zkY2v8zd@OW^-36G@)LviME+0qYoqFS26kiw?`aaj{%zDOgwA+M-3PQW+n&mGICG?D
z{u}ZhavETf^WU*z^{u1~BuqzxXsh_1Z{wL6XH@T2J-&4tH%X4DEkR*BX%l}}A=(2%
zLfbYT+=klWW>jrw05;Ym5CY0U``Etj{d!1;gpyT=`Z1JrY~xnY>I}t$8tD``g5Eh_
zIy-<=9a&q47c*51NKnA_Jn7QL)#>42e`ub9XekKY2PK-AM^`1{%2vz4X>yalHo@X`
zh(C*nY#>g4>O0E0px9xSD5G+yzl{PyH2#3Uu9Cz2ZNd~~BUM(D!-H^OSwem9V$cn&
zdI1=u3N8fU(sE=xgKJ6)2ty+|Dgn_fE*m`_9dBc=;5GpetK}F&GDLTnQM<+l<PmII
zBggvtOkHG#7mIcuXDCMRFP1_)-u{BRlAl^AC#3NqK~V+*m7Ew$Db7}-z&<t_eBq=Z
zg-QKKO@#ZRn}3M#S&$NKdW)ockOwxmwv?6MdbClqdSg8rtM>H0#kI~ur)97`CJl($
zAiaWiigG2=JBF&s)xy&b0Q$61vv*nHbm<#ZfD(E*=pZ9H7EiY0gX&w^awyyUre7Nu
zd?CXj7?UBH4AwctphEs>KRMO@jh6J~a+*O<Tf#)2(_=V#c@et`J@5=)GfzB^98BSI
zrh(Pt<bLuq-$kB2>R#l9#<cqThbDztt&;)v;T2s^mw~Z|$i~N|G050W`7{~qBO1Hm
zxXhSnK)m1(2UR@@Lw$I}NicW{BGf_|w$rDcexlZz;r1!C)xwO7NRTZzH3hf$$T*S^
zcTf?{5Hy6gztPPrgla>J2mMgXK*3||8z6-Ntz+$*VVUwO8Rs)3mWhLSR?O;A1x&ab
zZ}>RogbSn~Xl*PlUI;P^${2wfCKLTzt($@~#ca4q_HAjVLeBP|EjYpIx}4L@7Q<Kv
z<`<cHOXb|yY9xJSN^_)~^L9cXx#w)$6YQ6IeqvFxc>m$%#l@}{feE)4G-CqExEIFI
z16fKHrIC<K83=*b-vR`;$i@DQ6Y=(#h1kkV(tE4H)xdGEHHCqh2^!%mMX_3t2KW{(
zgymA~&lpBVpjY6YDRH1JP=~W1p8yQVtNTkPFg$AT1eu&(W^&8|O`d5PyHn~DkfW1l
zuq1a`kZ^GwOW9_P02=Gh&dL;j-CQB&E12rfOaP_owM5Dt<>u=)81-tFiXg~h^ofwl
z_zRWJGm=Zd(6fPB?yyHz`7`y^r~z`TT<XuPeTDf~tBq_XP8cS6U$cqqR-D6vzxCfz
z69_^!snn+N5$XgCv+~%qosI^B8d2H@H^*$59`KGVL$LLXfCE7<jBI0CLtPuQ&)N~O
z3ti1U+;yC24f~Xt!4hF%nS=kIE0+a(gI=Gj0P$K2lgnc-%E9vUa)qrDW(>J<C#uHT
zp^b`UAXmjVs00K~c(v`{WE8WMYhtx(cQK30wf@YyJX$NR>wIv#8~w;=jQ90vggG#H
znnCl1KxheK`EHCq3Lq!l6f|B~Qo(ALHwVVlT%(eK+>%CQ0VumFZ*B1&>vU6zy)CXX
z<}u~-l&SJ^d)n(25Wp8A6plI&Zc3E|H^?0U%@APXP9IejN@aOl<gP&0bkt9&mb-)4
zOKRkv00OETy&fyU4e#~sZE@Mzy@25qd{!5d`+`$|1@Y2FnT7ie6P7>)R{Veu4%~wM
zQi5D4Kex$u%mR0KFaRG^I8`1pva3c>Sz6gb?m{^(EeiCDok#&k7(PJNMCFm@S0`h6
zd3n@^rI!oFH5(QP+YYw>nEhq-vgB7mp(Ehots%A7`z&^*Vad|cZcG`N<-ZUqjCV)@
z!!kRuAJ%>(Q)RpeBeKXRmdc!jV#|+Zq4&A5OF%eHtSD9{^AfgLHdf_3nV-Ie2Nb6d
zsj;#kJ)_az{GbaH4OXHvSh*}p>rnFq*gelK4YG^<>xD>Sg^&xXTP;h{h#P2NgnEsV
zW$~U$j&?^c3p#>np*kze<4?HwAb_k$7@~WKW@BoQbUUsLjskr?BdgjpS?WP`rVOEC
zOm|Ty(`uiQ0mQ`JQn=aI1gTCKEo&1*Rw^DLw6ZQFE7)Yz$_8&0I(Npt7L4^Syc--g
z<5=`6GYo1wEWsNCa|O6YK`)z(cqV6OfC?L8O8QCw(k7ga&GEr>bbf%yxWx$!-clF?
zxB3^FEW$$#EoWPd0vkFtP=?BEk7sJfLNX?F-^T-=$lx!pkSF4F6reN3lZgkQdLf8;
z4ed=zo(!;CAUo_^loUY#cU~HIS}*Af(W3b83VF&fT8q0~p0<A@o3BEeY#9jcpyTHm
z{~jk76InSi{<}e*HSXNAPhWX102~Kh{nvpO?J@-enT>~nBeZ@z@5`mY&b-FeFvz^%
zh+BxeXPx{uz8_*O6@C|p#BGHrLA<kGUQE<bD=!86LG~`#E{#pUPY00=fRxFT=^)}w
z{b=%s^gHE*U`njzron0QvJKDrwDQLwp**M)w|HLuv=e!2c^xc-<yBsBP<e174*k!z
zxW@WPgNF{0zr@ZC%puuE{u+PQ)Kd6()!v0c2=apbEf7%xZ4p-C*AgVfkZ)0>N?y15
z7{r)IEvGule{UG<$dS&;-)&#|m$G`!n`zN3Q)#hy7TBR@LEefJR9k5=R;tIpZM)o~
zZ!dYrS7*`%FUv*p58t}^Fb6M`cl`~+C~>`B-b>hGKLd4l*&QP9$9NDLwcHCFBLB4U
zw1sX)J_v@+idjtB_%MC40frAEeU*IV&v*wg_<b&9nt%CP`36cD($<1)PCjJ$*nd~k
zH-4c>&BF5ZK*=YuX`|6JNj{BTq9j}`p9NG!2##d8xzBwxEE{(^V~AM1q`5Zv7mlO_
z@JONjzhA0$bD4hA(%x#j@nujolcI4FVa)l;7wT6ES-JctKpW4N<nncTWhq_qT7vHz
z-vHxyb_}fD-hcfU!p)+eny5hovV0qxrhyRlyLnl^3(UduPG?;M4j8_KUaVlr_YQ&j
zbg)AkA=m$4G?atUMB1uaU08vf<z3pEkxVZ~rHnQ5qqb^zQWmvTe%#h=F35AxOzB<Y
z0j@R{36$M@3<mtH(;nsvW%ssbJ!8d&=qarKmOX5ADD~MXm9l4B)ysP#Hhr(QW}YG?
zfMGy!U{+<bDSz*_Zrm$^aI=rUOcn+q8qAP3B1R0@*6r&sX^JHVL$yKqi9ciT0(+)0
z_+USqzuU>(WPd|TWBoBF$^kxn3&q9xE(#4oc0MprgrPP_D_^Xg<d)~Tb-+9X#@;qQ
zD5XVYmceGmTg+^pbWpW5So-wtE$w1?CM)og!u-wBeEYT@YlD1DI2C9W?O<O|MIqUa
zu|ue80t0tmI{C_th|DF}Ibnm@Y(`jJ+PVb~T}4>b;UHg_3FSPT%(rHN=7ZDz(uu)P
zI`e+5mO~tL`-U?T*JxO;q6|6-1hO<#RfTWAg%5RF=+{pUbKtN^w4SM%`0xN2#w<!-
z5Rv5we@6Qrkg596k!kPPrwT$eEWPTc1wliWk)xdaVdIgb15*exg<xAJ#{>q(A@dhz
z^sZ^r6<jAIhz2><-m%hn_~y8_Zr7~HA%Kaw_{ZB6-MV!<SxyKL8?8{-d`K-P1|>ka
zVn!nqA62ozzuC|Dq#&Cr_&~cQ5Gb*#FIc+AAsT@EQR(4t0NGlZi?QE54JgQXQTL`-
zfQdX=b{>J5%(ud<8So;)vW$w82$t>>^hMd+`pEW;DJPf1h<0)yPKUz9*spC9JBc*b
zBOn^Suu}|2J05}+>tWnVr?z!lS_vo%1!I}(1S$2jSR3R^SZQ_tPPfG)>K7DvhGU9!
zgiD0+O#o3_P^)|E%tTGJL({DZZ<5M%KQn3?IChluZ`(9QVf&xQeEpw&Ap_F77O1X;
zLm&fpIx@y3Eb?TKzfF;a{x=xrWpI!{5qlRnyCJt&Aw&K3`M748{=s^>!xBT<2=Q&W
zt-W??Rz}3GS7l(cV}!)WgfeFI+U-pC-yox6U@%qsh~ejv(eXW8!Wq=^$jF#DUmAW!
zIhF^58kQ07SRaKx7UuKP-8U}2fslMH=45$p+{Dn$xTP@Z?Tj0YMjt<tX=S_}r5vbG
z3gXycP8M}Q6HPasDPhP2pP5IA(2%K4bV9rbZ0K%ij*|?pFr`7vK0C;KU~M&8an#^2
zpR*Gr^VMOG#72l&=LX&oQ9+BoIj|UVd7i;IE(fRA`N2RS4nh}~3ldHj(_sH3>}GPI
zuR}9Lu}<htcCPVnCP!Z6-yk*yQX-fR^wx{x_!VW*G31hXL!%s6?dn=7OrtjEANk4F
z*qNeOY&9e(WXD5JE%r~0OOt(A)DI0H^rw~hG`ulL5UZfuWC9SR7*@&Tv^++!f#<`J
zP#P0?vqxaHEFLjz2hsdsK3PXvf7s8av{e_}G*n5mr${%PC*_XtMGyt7aRAL_i(xje
z@V9;eBSNKZs4_aq`GOo^wfVFR_Ny}Bp;R#2_tLbZjAh00aaJ{uQmXxzxWt-+uJXQM
zCcEMihitY$YJwS7Lyx1TS{sK3%upL_$TT0&1V9`*_&6;#e!4MEC^hOSnUQ9OS%=F_
zC!x+@&sr<#nT7&58EfRSod&nk)=-Jy5xYD#P)+5Gx?Ct%__MI;Z^fowX$)iB;GuGr
zO;e2?$B-(n4x(z2>O~h!bOgR8*!~IppIn>Bfnjr|SzhNS4jNUZT<<TC)~NIY6LN#k
zp@vE<nF$*VDxIavUn@5nHcJ^qY>v&Ff(8dvXQs=|hKgg4?RuizVhgE6aP+MKOLN?v
z+}73|v~bB(mQ>_4CieEW?iOqr7A$wz9D2}~$(=r!$+JK)kam}&orjj$)l332rgL{p
zje|#n(iUvRJ&82gi_9pXrMh#k5s?=D&<x}IY+1O0?vVR^Kw7#~&O!L!12H{9C>S4v
z5GFsiKliC@5^_UztE-iTn)`#XsSszB11pzd4k_t<t32c*)7owa&wBH$l!tA*S>6s-
zM)^e$J^~#nhsq-X6L{JtES$D8V<nHK6|9EHV>kUTZL<cC87PmXpVu(yOBJAB`6*w4
zh?08$9ZRC66<5c6ndQ%*_o}s&hO*gdoF(98JU+8!jse|OYZ+~qx$)UknqQ_e$Glhw
zeUI!fK0l4aL}BvBWwOBFfmGL578+4AXs=Qh`Km29fh>yokPN(CY@Z<3tXh`D&A^)m
zQ!H&U#R9y)4V+|I?D?p1vON9XK*t@@g%-&Qn-+;42xeayrvR?Vp0dhcC#)<6!P0`O
z<Mk2fp0*}I*{pt$XjjVGL=P#i9=DDK2B74jtc!*B=qBs^KD`s$bSFfO4FM(Y!)P(G
zF{nU8Id})Lb}o3VYziDqE=22gv<8hgRp;{_-+J|Ij$iN{Bi))?UTld)v5)lD;6H`M
z#b8(492zZn2Qs0mvfZC)12v}l;|6{__8U@-CUid$7ZhwbENX$v@kyJfu!floI}C7-
z!p}%!EL-ER;hTqH^{IG9!3hf{YM*x0lHW036=hGj7#uY_7EOj)%$Jd8VuN5qM2aj9
z4-O>a>#mk(<31TY4b{pRwD4R!$-&AF$qZG~3==GJN$7uLD?}75N-lE+&j)nWh0Poz
zFZeSC7;RIYyIp=87=^73{UU!Cn1cBS_TqAR(Z5+h)fcN+4p#J1d;_~LDEs$`o~#90
zb6A2c`VTQCtc7roSIf)3a@Xc@Y9mDYj{!AKJcBb>?dDG=Efkl^D}e>ZUcusAB!4!R
z(zAO{`HSseQ6Wm2%3p)<%ChR)e>IKEo)ZkDTrPjJ50DPEDkrZsgM^42O%cUn*$wmc
zSS<T&=H-nv66$7Bz{9TpJ#h4t_4GjxVR^IpGhT#m1rD@87#_Us6Y~68*e<g=HiFN+
zlR#-co`%SO_%ARrL9C{cpUw_O)4$u+oPs=dhBsF1To2A*1y=Rn@?M&g^^~D(tHMl=
z{QG|Uv8~v4`Dd(A#Rv)*8|8yoqx!zod?z2qeoTkwUF~!qCD=F6$^d-O$j2T>|B83V
zqvW~f^nDzwftW%`uQ~Vj6NiSUXrX)>(7^YID-Zu&BcBE198n;j``XFER;ulR(V$ws
z2&>VSbqI<6A`vMAitVawDRunYp;-u+89=A8iyCJ6rOlP8L1gb&4vV=mVXneG@t@e~
ze0Kbp{_?d?)WThdO-N@#zVV6F><vJ$jlPNhjfJY=g?7@F@~w}idK2GMwU`<N?U!ri
zy95woumLv1ioXw91830oWIx1dn2zeiFo8?E2$f?TF9TFI$S&>7f(Mf>XiIbKd8vVt
zf7H(H9gR$(LMoN~*uN(QvJeIv$?n?D?S?#?G(zp>FNNmI#I)VpsVQ>cSo9&-!~RXd
zO!oB2c^Fuc#43IvCs6ikrv}4XoS41anZXcdUZ4znNAX>w+Kd|aY3IV8G8rVAk`3-_
zvmj!wtNg?u((V{#Nao`2`?Yh+8t#Q@*=bCS-rs>CN4Bn0iyRXNv{QQ-irJw{FBso}
z?bLGGSg*oxEC0@v2n(LD)7jeIgU*7icxe-uRzojEyoe`kyYs%_Mp-BA{25JN2p{Hu
zrF~$+RB~RRzyxbiS#nkfL)vikyX+X7S6q$G;DzNd4#xBto)u;+k!&v7fp+$($QUe+
zV3Tx-C1zL(>7ev2`#`vxjq;`(Z0~^>mVt#G5<rtVVU7)8<3IIh)YVkJguHa9KcoGg
z?_p*+%#e)ISw+~B5;@%GR2B%rv?fkG8%H=A%-TockLzJ35+f=a<VgR0C%IS&hMw03
zIm%(?yz}MgSm^{N&K(nUfp%nwD5D##tCM0yg&f<i$*f=fOW6_aI0Lyl<9F}z_U1$+
zlbsNlz#)SblPpY53`}S$fR!jd$v&y6E|+fU=Z%_Iza9qY?lxg|ChU74ydHsVMD3um
zn}5Jt3w;hS$2{rf&nQ4so?-=%-f7@vDzM<c(kJKx0u^E&a3cDqw;E^!054iAC;Kxv
zVfEr+cxXSzu4Su4P!n}(yR1&}edCVNE(_la{!5|#)MhGDO{BR3_-XCb=1=7~FQ*4b
z0oN*LFYPm8zqNB9ENvK9qh^$a&5|>lp`_Umb1j#j`Lji^?4S+?@U>F<8&5bA|C0ee
zrfE&&KfD;3tVYk#Tp8%Fs$C0sw7dk$nS*S{dY#-`2B$%qT^Zy8G)gfmVMzL&1)K~a
zlx^7U0LgSR)Bs}Gi405VU!_)JNcS?lT~jw0H1XvUrj&s%G1Gpt8h=rJ2i%F)QWILR
zOqk2bLYK&hz#xJiA-m*dWYG8kbb66dJ{8$Geg3qTkG5yfI8e4ui*VM-7y}Q%0);Zx
zMnI-0w`g2o7Y2Vz^j{%owR2IVnKc_l0Awaq8hws8f<8NoGzs`!3mna+LSmCt!=q#~
zA@Biw@uFn>*u-EFQ1Rx%ST-qrolH8I3?>OV+rK7*p;sE$t(=ocKA*=bK=S9NIkZqK
zWd}LWpK-UaPZ~OYzCUC7gvrzmazUUJofT^2!dSf8On7+c9$k)|qAWwrlBISoa(V#s
z#Q`88v+Yx}&*b7@F9TGVy(9o0on;|aRJ|;;Ck#1<+nQ~Wiwq^GeS{>qii{LDqr_}o
zxL0Y=n=2(TID&d%anJbXA}2^D2Oe;*FxZbl%%!nrWY4g5=XM0CmDx9q^{8&0P2X7J
z0aL@L7%)mCR8FSau|{=I*ptg+KyPmW;a@78pF#P7#9e8_qaIRk$@$ynCA1WcDjSZ1
zeMT<zg;FM_IS*bwkQ>t8q`F<xu{<>@f8yB^gmaoSthGw?j8KB7*_wAfip#ZAyX4>j
zQe#uF2X(pB`V_o^xrGfdRhf-=THtc*1Zt<#gBB*SB<+lZ8AO->`7EbXXH4V5BTkf=
zHpLZ3TzJ_Pa+!Sz=?u9Sa(P-4#5usvS|L`g@`^wnBva;vg2%{}HYSSr%2ffW-jZf<
zH*S1&!sABe#vp3BCIQHhdu@&<*E(tHMmxVVYd=r~Md3PwS`9#Qy$_&<KMcSVcSD@b
z$+>bCt-LXCGK2-EZwgYbz}2oL7;iQnRgz`+7_FPP#2SEY_MKtH0b9eZaVN8+aAhe=
zFy7`6fVJRang@^cBDp;+i+D4Q`i{hbRohaGvM4q(s<EB(<xc-*A<Vq&QVO|P?n<kp
zd<A^~n)%&n9GDgGt#&X}zl>d!_q5nK=2z2<!Tp;r_Zs2`PZ&0KklYtY8rOgPgy93_
z{`8qeZ>4lYc_2-=QEh8bTx~%77m}O&JdWaF7t4cb;fpnLc?qtLhYY=$$tbCVmg+U~
za9S19au)&zv*i~pCUIF5!>57|@<<@Q3|&VG(H5>w9<^=t8iv12gA!F5u_TWLG>RmZ
z?B!PmZQ+!{8B*_SCd;SHPn$c1;UzLF5lOZZ3>u=>Y#$3IbiBI=J&h3AX9xvTfw(r;
zU*Nq*;~C1<%De>BR-ldmES33bATv@yFO~)VjQd3Y8X`Ppp`EFoef!9w7)HYlG-^Z^
z8>2YpP+4NIQstmQpLRa`bP)8cWdY(57hNXHgI7$2#bLdyXxF60x=jAJPM0`h%EL@|
zMVf;X{>nHK&26W2Ju79EZ3Dg6a1*U==ay}_ajm{21M=3SH8b*xeH53;S_5)Gbxzi$
z<ukU@+>7g-AQk`_^h&UtUkiOzHpE)S(aR_s9a}y7$tDLC7ay)ZBAv1yTQW8T5f$R3
z!#MzV@a8xMNY7zi^DRcKY|dgPf~^i4;t}C)qOi0=w%Oc~$gw@ZS26>uf-d`b@Eojf
zx;&94h6Z1(3h|Q;C8av74^T#u{f+<$r2r8EB#kjVG#OU;TZ-%Lso2({>RNf)hoa0v
zuK;5vsDvz(rt~RoG8Dwu$uj{9YJBDMsQXYM&+fFj3!b`GLk!_%mFMiVpt~a?{9;M4
zZ`H_eLa;-)hZ5cMzG0SrWP+C0O@ZPuOJ0a&c~hL<2Gpum;_!ic5*Z8RcL6oztp(UX
znzdgvKz2NYWA{=_jd1T8CeZ)hKj2=IdCY~-_J^3Rumnz=yu54#k}a*4KN|AN88;|@
zijzJFVFJZqCi05CISL}pp94lZb^%FqDA`|v-g?Pj{U!6I1O3=ZyDeS~K!>sm4bH*e
z?5kn5Me>?|WiV^1BB+$2<?Ah0!z3(>CvW&O^<wZM215THG|(O4J8#+_$SkU;leglR
z0t9osov_JrP?Y|*$~*B4cTLzm|KV~Yj}41W^4`~Dqu;~@_^zQfRYiHvM>0DWN8cdt
zJ8pXQm4EswJ!`au-u*$~23Ax}J^e6XQb&&6SRdI)y}R|6fB8se+5pz*Y=J<tN<Q`v
zSqhI{Cv=08Pwc~SqXx>S!I8j_s&ggmW{GqES)2l-A|UXau_>P?j<cB#Aw_`eTriKX
zkuQRxaBe8F|J&y^WlH&seCf|9%rT;jmBqhG3uA~q%gbz)|0I}^apBbRb&LnHXMhtO
ze&3{TwU!roko-5^!*Qm@gHRs%Hekn%tR+~qrX1rNdFU6)cWE4!bysdG-=}BGp%=sR
z2GQvUf2QIqwx6wwq<xs>XeQF}@HVqScWLiNJeW^;=h9aFBY&oOJa~_6=*Rv{>jq%g
z80E2R`zAI(;2nVw;d0rnz3SbK^~3XpvU_{AY5{<4K-L~DrXfRT@vc4X4Kx{tuR-?m
zHxMdEnMh49d$)IMAo}7VX0THB@s|{K=<AT}WM7*#2Ukv&{3O0W#qoSCc*B0aa#rJ7
zr!~3!zp{Vegvv8OLcLMn#RFok5cXLNi~Va3Y_Ar?{sa0;tM*~+Q&|YIg|u$(#?8VK
zG`nR8Qkxh9wgRgpxIx;sH`Al_Trgckd;b9MYk}i->6;}`=<Fc2cPNz`;ThSNVZVc6
zpLR&d4U?r~ptFjR=bhr$VW?gQ(WbL+#{%d^Yov?Mq#X@09tzdSLGe9$*X8Bl_G)9z
z<|}e?NQ_lnR}2Y3e%jtmq;#LrrOTsqsA0ajb;n^gLSYf^ox^=TyG&Hzl_Tu;=~;xn
z9hu<Sz#LaC*n5;iJRi-<ua%=6cUEUoON@a?%<4$*^f9sVMR2__w6|+})xu_0%CWK0
zdIRvNvjP&!!yOk0r#Ap-n2k+1KBmQc>*a(rtr`qzcvFh@mPvA=kDWqJ)hht;RZeQ}
z#=SwPO_7mq2?(ohOAED0G~FGD6rm64q=)0K4v6g;U$Tu20Jpjxw1f8w@NvC*^s@A}
zFX4X7OP@q<C}#i~11H3sEFQ)~xm5b5nOS<&xX{V|jG25i0I?%pzw`{bevsf$u2fD*
z&s2E<r(G_m8tNdzGDqaJpc(`?mq0=}-Pgf1VV2)rs)7!qt<Fd*wEiaZldg7VAQ62|
zil@rYVn{>+A|z1yI}p^q*<)`(;~W`apA@rdBj|phZztEYQ3eI&YSvo5Do4q7u)PB7
z7Of0J{Gg91kf8~GYk^-1F}Ry7s0kb!`K|%Cix$nidU(vrG6OQgIMgVuemT-%w%i6l
z3_)j<J%hQdx{$>(m|8sA;6SrT8Do3eSU+H_jP*J3IoPC09{lf88D}S;JR1hLv(gr^
zYymG_1jZR)$UDfv@K)dp<W-E+LCt+gl#B&V@DW%xh}CS^u4H2DK|x^!GEiiaEe_`G
z%w#z`_+SONqB@Dr2_S$m92!-V&kY8OEkd3QxyN~Ne8=ImFz@;NL@ksD$>}L!Gk59&
z9}x~Gm|W-_+$0wUos`w0c)45@lY+6)VAX(EF7{{Cj$kcg&j-0AhJsg`AuG7>r7(TF
z9AwI>X^W-E@kG^uL3a#tFw=$!iN%4FVl+!f;97}~Vbu=ow#(Q%v_a=;OUq<(nkTG<
z4N@BLKnt*9SL>uKHh*ePva$J~8jZe&v(6xx5?@kIckVXT-`+`y0wNJ3XSCI)PdBNE
zm(=~kj>we1Siz~%X_$d@=o)I^XVC}&rNPm8A%VXI5)lldh{wFtSnm19O114rt{C2b
zyqx!!7pRRAVg)43njn?3Tm|*Eywv(zTACb~--2nU1$ek1S@@Unf@YVS9($sJOs4tT
zahMNB%wn10U-B7y_|QE-u>saITRoBmW#&#NoMv{`Dx{XE;TCdPtWtvuIMfokJU}vx
zS#4>ceT!USs~(O>WwuVZ(sr(nWtpyOK51$lpgV=e3R;fl;<mWjC!!mK4h~}InmDWQ
zT2#ojfgSutC)W~)rPtX?N+5w;Z*$jHAR7ti=?42#tNg;=eq#&|YdVW8Gv((dUx!tj
zF(VThL4({JZ|kX3a+h0z#M#UiChA+`9UPivimiEan}IVNuC3dHYD-XKF3xsA4Z{N~
z<j(dHX28yv=4*pAq-qQ#U2YcSk-K6>RK!HEirgK4k7Ze+YPrYZH`(1hLqzEkxi{W~
zD8T%w`~1xap<%uzZNQu4egiXClA9q981&<YkCC4n_Q1ka(Rk2U95m>Xj67t2G}aFR
z3=g+=Q%Nm4C>5fFgXOsfZgAH!R+?wykw5}GC}@&cBaa3XL`)M6@VI(@>3A_Xt+n_6
zo?rBs0oPbxlqsyp%dg@}L!F7zGGw`Mq<SYemabyx1pJs-F#>V1_h^$Qv*LH9VY80X
zb-<j5MYxe?+Z+hU&{)SgJCQt(0l6y0uDQNc)&WAtS7vm`Jbz|Z44%2=GT*Q^a9Dr%
z>=*2`bGVo32EnOYBnxd1dUo$Fi-P*mv$94OJ2RpwX`3vGHAkmTu6e0XOSzhBW+|X$
z%{zfTryG5qEca(xB$V%A-W3L3Inu&1veM=nh=R1Zd{^0Dqcc2NvN}Fio2<p+u5n}n
zBe+?3ue0|p1zB}x%G&l#W&aNo31nSD1g}mj^tRqnZdE`nqQ(rY8yxzGSTSEgvK`9C
z*ax`LC(EWd*Avc>h9JaZ1S(FI%|5Nx!T~9;Q2CbhjFp+;5rt5(H8u~bJTe2&h!QHr
zHrx1axRJKU0qb90B9A+Cx^<?E1Gi)wO0M!myj+AZt|#LU#Hq`Um>SAD4N7Z44voeb
zil<iQrn3&@Q^6=IL1Xy)=>RPcab{+&stgVSX^JymnW=@{3$dFF+IYqQ9yNS`Ji8M)
zQ*EPr2NJ0~XV@B<MR3Zm1Gq8JV}IkHs3ie3h&Tkl$e$0SPA+F@y%&OQ<PjLfrh~tY
zFPXrhhtQLN<ahpyX^KXSc9Xwouojk9$V&;IRzr9>Y}y*eK&;}v{k>rg&FyQ|lTX#<
z4*>%lg?SoYUJe)sqC^vVmfE#nUMzodoa2V3xJK`#yb@m{?yZEi75*GZFJ)8HYWa&H
z3Fhq0@>d@~J^UbD$lda4FiPY`AX<;TF#cvU4$8}GzG#{%SdEOaXe!6b>oE=5B{L-P
zji7UA3ABx|*_Ot+=zn2~Gdf54Pl`8hnsfqYjW!W_3A`2P)fjUv%cefpCGxh-gxY7D
zc3cZ8d#4!-#wUS=fB5Vw3+jxF)m*JTap)0x*S4rP=>I*R$fCVy*jG^xoAbV5YBUnE
z<)1zY6<PHrLY@7<mVm9CgIoN=7E>~$2NniyGFXDjIr32&VLA9OZ_ZWnFC)^y<xoQ8
z<3uZomjT0D2(w;kpR{kX5}z$PS3~&yG&Ves#%Hi)%4ca)ms4^<spn<#xlIJ;F}Cvy
ze}UnJRL$x9{dYhIy^RG!DCK_{TZ(qARfyaDDp(T^WnTW1ra^Rwp;*3-FA8dSe&m~=
zTwH4~6aUvgp*EnZ9ZVzN#xT4nfiRkLzq8#LGjX(h-@d7nG4ubQYT)vN4?ukpo=4oe
zi$pq@m5mo9vwImDuuBKEus~F-kRSQ)EG+~poVI`XaR+lFSIMp&JU|T9%cjb19Xx=o
zMP!6pOqCmKlHKiX%(5PKlRa#fDYZ~3_w3NL8(^=k<T2c<gP8=4U04NQ0@fgV`!AN-
zXa6-VV;`TI5&+YZHmjkMQmksTvae0XCcN?!Uy`Z&RGYWRem*jdzO=Vc;INVR_qU;|
zfwx<r{{bDEtiS5`3`?8IfgRk)rjdCtUrVbFZetmT=xg*rjXzOB0cW^P2NN@?`ECV)
ztI37>4`!Z?n_80}q;2z>)EvD*^#qC#6pQ9bJ4YwD2F`YShZEW#MnIQBkYxvd14e5Y
zPdj#Sn<cCuDvNdUPb@@$*2Hbnxr1BKC=H^5uyU9#{vzRJVI@=omxGK5PCr5pwpkFB
z!Fn=>_`5Wf@fu(+`k%(WK{ud5nxeE}_fUVEjt%NUC=4lwb+8=jetqTeKx}3POeR?<
z&_~2ZAykO%EY&K90L{g6r2k^Ie(5M8KVh8QQMUQR5uYeWJ5p&wr_&pi+VsB=jBt|Y
z%Q3N34F@q1rE8*Vk_DAOi;{@^lw*y0p$*h#<hZm}#z)bpMKd~$Nypo3xNUI2PcWpx
zMp_1_PV`w>a~c^Ygi20IsIia_TH3oAY9@{!F5NqXFBFBuYpY>g<OSFxz-JX5%uDx9
z&!AS=aM)+Om%l^0qsbjced!(F;htj7`C>fl<J3dbZiDm<#H%bx)r*u%t@Y#}8P;G0
z$)MqVo%D;H!}35$cyfw?F`ULP9)eSSKplic1p|g2KRL~hhYG1A7wTd-9`kt@JjHxV
znlVpru?#BE@Y%>2hGLj%Kq_Z;@LK>kXNyRppT${(4-TxSzYnAC9_-h4WakWs4cFLJ
z%1x{+X<>ed`<8(=O=JB?NKAu#9SRlp4vj(a-&K$q2B)Dn=ncp4ks-0~xuQ&+3{7Z9
zbc!;9Wil+@K)v;>43DXXk1voBM#zozhn*lJ6Z%*ZsAZ_?C?7P+tkPsMI)So^ZXz(~
z6*9(0&=_8woS_G-|2@{<Mr&=Rri1rxlyR}4<Z%c|$jezVreew5Ipcln%$aV6V?#_!
zHMpRI`zaVtia45VPA2$Z_Oh{<iAL?y5son_NCFXMAOM7WIvHT)Td^W52zz#{2$vb7
z2EA4TsP~+}+^A7<u5qq0qvbrGH2B(f1ZSL|o~alMyU=R6z)u1SP$Qb;LjMbC>2UH~
zWPg=Wqpy^UZJq))+_@wOZQx8)M3+J*Dll)=b~iF_tOb`Uiq~1yrc{b;Bu2DLNf5!P
z;iDzvi^P&~E!gE@)IT{k#I7n_5(}i%ft`=a^h>2It^jB_2*g5Zx<u)S;Q?N0Nal@8
z`(gvGgDG}s855(*#Z;eyGC9-h=o~7KL0Q;@?OZGTB|^peKFbkzS?N$4NoXPnV5gri
zm)@pS!XXz(RoZTgM@5}%W1?KIu9|N9OKl^k*GaXlvax;wqX+Y8&8V9yHGy{}3fL|f
z8w{Ad(umX=qDRz%M@@@yN(!e}%5;YrSChwj&`ZJG<QcJ7g?V(Bm%4-%(_L0!!87Ay
z1oKh~F}oTUd|8|%CV-Zs*0x;k@7hW9G-_g4u85Ht+KByuTrOAo@*qaxud57{!eX$0
zF!!td1F(L0R;YZ;k!xauQ2ZcUg_o|igH?ru=j#GY5IQ4SnzQ7tH@wtV;aa&N?GaT-
z)_C0}H>PVPR}E{2+!R=ckg!&64ouF>0ztQ=WiVTk;>b3+)%TclGNK*Q2j*uT_?Dsm
zx22igBnj%cJv}2cuuJHU*pWQs=W4k#aEwx0b~D~(NKmuBRw24OUKw3jS_mjJa*yMJ
zDi|~p$OEkK!?Nf11_oJE47NA9&p{yP;Qfx0-g3Voj2W-;fGr<uhYeYNp7xh%GQr|#
zN_;SJLe^8qr+IlOK|#L+hh}~$^A;brZ5udhr2Hb#TYv;QdBjQFu1X+8PjPdRxJP|R
zbJDu}MFX7t(qKlAS*1Ld*2G>V%w=9Czw&3y17itCNC#3Mh=X>8$aPpJW~B|W9xkkm
zJ=>p=l`n$k3#KP?Vl>lgAhs=(xei<S0$`H^owN19yg)jthG?<g`7Nel><UeK3j&xa
zwNRtjhJ0a&Bd|K6dURnq+7`(oCuZLGivxGV&;(GH1mQr|#Zs2qpLF$5o>~^e=aBSM
zP2=wJ;Kc*_7sv{S$RcM@>KiCqnHIT4wO6LvuCk8`CiI^$SXKvs+R|fMRxKB6VzF##
zMWf-`=EVZ@R$g~qIx-Y{t)!3#kpLOAGO)zvdLxtmSlou#-XqVDjS1GZs$#5wG_uK1
z0V>}FTs9=i(|}x!OtpsPo9)}idPGsvsgAG;+0qOkxU-U$Yz<_hKrWo;+v3lH=~=i@
zw%Zyl2$<zcA2;xJJcv8?i2x6GEFx6FpqI;&{){5GMyNt6-4VMt6n%uGF}=a+ohD~L
z70ZC!MxVHq@?b*E)4p7$NvxA5yFSL}<r)7Xd@+=ZDfKLsX9FrQgE^pL*2{S=##c6h
zC~}CWdZy*qX&kaNvJMLK4G>I!<Ks{wr&hETrrqab%g4jZ#7;ji*v8eC*UE4GeG8RY
z2v-gw*ZdhQ9WYr^V4g28#u(!!jvqI+V6eOt07LW9w0s=J-=}8;ORsnNgM&(5?y4?+
zVM4;o{x*a6uT_q`TK?z^CT?^p7r3dhaNgH*<WEkR#(KDL<UyzfpJ*gUXHDx@Y-IQc
zkeyLhgX;=5x44~Kgnm-ihnlIP*I(?Gfo;tW<X;_bLPFaS<192@MP5y-T&6L^D2gS2
zOV5l5up)Ua&cvjQydM7^S|@K9ag4z{FfIPw;UfPfbyEc}bzS{Y@}`eW{E_gM;9hzw
zaEf?R;@+YKn2P(hzda9XIfYgHSKf*H6DEW_Y)1cxtt`prW{|7D>mRuS1W)irdCw`B
z;@kpx-zUU1uUkb8#RB*LX$!mH3IyVR5TL4VXV>M2aWWyaqAKo3>6K+XXy~qhgw|LM
zt~f-7s?tuM#7g<t=yWu6fKOr^D8F!qeCpr8N=l~X*>MzPKC>5RmSwVNcqX4Ggs`pw
z&&WpkBCUq9C5Z^)7TW{Jztgacb)iGfZ&CTu$Kk4Ipkuh&S7~f!hO_h5Z26CYeDSfz
z%GbfbAuGx^{w|isa5e~sE9JkoF+1*tJ@Ia6gWvi)<g}LYv_`%Q9-g_h7Iy{u#Uf1j
zdqaAz5Ltac#4N)G4ixEVMmCy_DPJ$ak=doAS-B`lv&c1#pj`7(`BBHF4lst{Ju6#e
zc~syT*aRxRKkn#8Hx$q=m0deFwdUKnpY<S9Syuh1M0V@wYSw77Y#5L8vb)d9Bg6iJ
zso}Nk5u;vOTbP$U9b)EOA(LK16!-FHbc><Nms)oAPOH#z8xS{1Dui;M^ft;+>&~o~
zef^oP6Kkdx`jeP<KrPyv?Pq@>riMM*_wU$rFxQ6nRqQ0&(Zh~8)K#i7@&O&2n5+nc
z75kpZfj-H6a1nAT^yZdU2^c1|^R`<ptsTwaO?tMO%77}=N;2eWZ92M9Z$PGUN+FH=
zZ9A%AYBW6P?K-LnbV`Y|cMR}=u%Z{WQ47uNkbVI^fC2WV)Eh=R+A8UFfSsa~17dq(
z?h&oK>h#dr-bNZ3<$^8=Cl;*4_>ky$+e{+|`GP_HYS!iLcW_5F>mpXTP7X<Qz)X1}
ziMJ9R&p++xRzhTA6;ee!v}2PQJIk~1GuFsqHgscswsH#MTUAUtJPoHYF&Yd(&c8{H
zi0y@RRw76Gdn`QHoQOgGa8%GLBKI^vszi>CiD7YII^Qv|iS{v|T`XPwnI_$lCXv^2
zY)lD^<>k2a6-vJoqfcEY$ERm(Pr~S$m2yICH&kPI2OtWb7-Pfn4A}}i3bAt?)liMd
zP$czr>lo37rKPBiWfxR;Ba*`6Lg^8xL^zjL-RNn*_3Yh8dIbo$#@Qg1zU|)W86(_y
z+cZd@M5WtdHLwg=vN;Oqb73XyYeV+x-Ahh(<kF(O1fN4@0M!w>{nFZ8%aBw2-Gz{0
zv|24C9^L;_0|-xwU`X9*4n#PnsS8-qGv!-ydisWmIh5U2${Dr`46=k5^~|)r<)s<=
zR*L0k@ez)UGIfMF^-s?>>OI7if&q!PHtD@iLoqS1QU>}j%wz>~#5STMG5ts~C}G~Z
zqsibju*EbXR8fZbGmG7!D6vq62Bx65;hq9hA7-bfTet4LWw`wt^=bObh@direP?8(
zW5}3}aY$|;C0Q^Q8D(D<5f`KF-&IV*9%KJ*tRGY$V*^GU9Edi!C05J0z*|Wz3b|<i
z#j|`WG92osn@j-(Zd7}h@o{We>7}w(3KCgR9EMc@@)#NL7MWnQGivY{nHW1v)m}*p
z$;2qez*-_JV9J~na7=^O5e7hjezt**z>XPmPD14xusUi(tK{5(0~!vZ&*Z!~I>osn
zIY0gw0;|gf{<U3Un&KfB8f-%cOpuFw0c0A8?odN8LhK+HTpU=}v~AKIgGqSbke6<d
zOJbIb&_AirW@)U4F9CYF@~xtn8B_!DUK2u#18!9PMy*LH@pmjjn3CWI$=K%O5WsIH
zlM@}bsN<(5|595;MNy%YrENpAL_oP(vwb(XVQGtG(+(hr8lW&ZyJfQfl%Tf(nQ~;D
z+MS2?E)W8yO1Xbb21r{tokT6A)e2t&)emciyjUvz8O_A>q%6js<pN>TQS-7Y;oXWC
zu&O6QY@wH3n$|*T!-Rh1A>t64Tff>!gUNJ~4|LMj3$?MvegoX-*6KWh5%hVPmNrB0
zOXcs=(@d6nU_s`VmB=Wc8TJi|DVR@}5gT<Kn@)i|(6ahj<Nv5Rvdj$j0oVT)2G?BX
zpz++01g}Mi+vWbuGz9uj*UJ^@y-j-Av!~LP{*1OQ#!@gv=Bl6=)sb*+!8TP8l&k%1
zvLjl5C~sfmRFs9~E=|e{)XaO=1%_enV2!)$<L`Vq5}@;PLo-k+8!pGSe4_(kC91eA
z-U77I+!SB4<V{hX+-&<*ltZM#EkPbb5Po!P5bRFc&KzYQ%OQr}7B7!R#n0Ol5-9E3
z>AfSc%p46i*t#<wproS8I(ON^(OygL4#r%907le(fweB4KuBM`;~rR9?o9-ns?h*N
zbb;L$iyB!;%~S3-*px%c1NH|pb<i^E=LV>jKWB**8i5{+uZwE4<t4Dt#Zjz6H7a@7
zK8Nobul~Xr18i$g&XY%c)hy0H8xCS;<<W%sRqD`Ef`Z(<RDS7i)BMONes<@5%%35s
z8>h><ll;nL&%=+G`kly#G9&d9_I90Rn5%~KV0J)Xh;%;mNSxEL$rCeEjsosZBnIPC
zi!!C|MvZ;8%(bb87RtPkbu!Fsb7g#f8X9@5xf)pzYvb89YY8_1t6u2qAdRtd@@m)-
z7a3(W){ht~iw)jE5c8HKG;CHDNZ0AoKm#Boq0;TREG?ead6*2<fy{<W%l&OCY?PzW
z3sP42GhQMP!*P9Vl9k3pvy|K*S+7bHn=}kABdg=yLn%zyX!zQipi<NZt1JOWTWeTA
zbA%jhkkzuzX@IT6m&y7d78DtrDjU)q@Es|_*T}}e8fCR~_p%Pbrg+;EEZ~ik2Ai#b
zb&T1*YjYg4{!lcx_)rvS8BxjoB(_${Rv*fgSk>lbTPzMYUA7bnp{26jwF@AKmTE()
zDpHnYXS&_vvF}4sGJQs!JYl<s^~;mKc!WAbcH7ah$pUe-oGD^#?b>aO+lxM^3VF&t
zFRYy`Pp8l4K#Xfjx=wT9Muh;FYc?xA<4dC)N1b*Zu>P!JB{!9h%I5;Xnk0jTfQQ>c
ze(gLAp)@!;C~F1@k_KxFxmhp2ac~$Lu~eQ<Z<AlEks4f^;oDkpL0+)aGIEssHi5*m
zM8Ra1{LUFELWbqVz)MZ)mCVq+mtrpm7mSnN$L1qg4gPQJ;s*JHjWGp@Q1Wt`gFbQI
zKWpWW31Q6&b68=_SHdXjKcz1hYvL}|^;i5EW9kv}iqYoFpVND0H{65+@fXJ>RYq;U
zrgAzQt@76(^r_jBN_jPXyA1h`Dq_puY}=t=)nuSfzZU0z{6Kl#Unje!?gGn)yb<J?
znNe1V7RFG?{_d|~b`9fO0+ZOAzDE$40P|bv1dy>ZxWUrpSIXOg+-!BVyc37GDu+hu
z82KLo3_9Ar8$<}pz&3d=h+Q*tYUF+2$3k3=|4i_p{tzgF&hroK_c|mP$cOfAQ7t$J
zt(_kkrT6I5UH+9;yc*ozEW95Fp(=I54+zq-oPtjR2=sC;1Gu01*A(7p7g_=)`I$ds
z`2fhlNT*vOpQn>n3?Ht1VH7%kocueca%yK3teuH3W2$k5)i^%#m7%Htj;tEk?hwuv
z7cb(FkzomFzD`#PIT!^`2FPrLWsm>nQDG$tJb~{0P3&rM4#BW`zy3D{M0w#-3|J%I
z8V(VJSt;KIJkVxXaN~PhCVdgG3uwDT`5^|)Rzhsi!e`P+jWxYd<8v4LX9@iCKkB5m
zqQ?4(==xnQKknqFv>n&tvfZ_l8c3kr50~9KsiL1LqJMUG`)VMIQ^CTtM<+G@Tz0wa
zY5%(74B4xbb2`QjTjL={$?o0BEPv+Gq`bCL_UUAnKjUObA_Sq@&Ax9Zv;4KNZs91J
z{SSJp{G^lHW(Y>MI4}DJ{9~&TfxEvC+%htsw&pEzfUOKE`E_z&nna8BL%pLXoLo$O
z0n*CG#3c=Lr?mD3Xvi7ZF1AzJ_&nTVIzL%H*=VP&A<F6-<fUDDk#(aOO~P<`inQ$w
zz2k=u9X3Ha7#7dTO2>pQGi9yBu5^lLdX~v|=1ON<5A%DaOCY5hiYW+G4)O({Q3#fP
zu>CbPg9f(Ma!7)Roz1DJ@|Ht|9QmpLqVZ75P8mVMtMkxKO=bYgLxTYU9u@$BJ**;C
z<7znE*3(#DFnEF-5idYmp;4F8%8~IxkDlG+D5uSC-Q{RoNinbTV-g5dfJ_CXF1JFu
z`iKncTjx_AYlB>T@iB6o|IF%ntj7cmSdNbg`y(gjgidOynpnxUc(st;GElouv;k@n
zl75oUJm2m1OBetGhr<Ht78C{!RwCUKYUjZ$)FZ7ESvz>22DzU8jE&T<0~(pN-7Aew
z;UD;*9YT%7z5Q)c62u{@_R^=5+FjL}3!60n?He#;P%c(ZHjJU)bfxt3kts4;o+ue6
zgiJZ5#mF=Yz@jFn`rFjuX&GkOT4w6bmec$fJCZE5#PA|k$v@r4fqty!brxPa!?8fg
za+RFvZ}9$MEo(T><!64z%;65RwZ<Z<Kc9(z(m%dI3IO9FxM2hQo95^T-owqZ$iP@D
zo>}Qp%!bI6H8Lo{h=35#%_3;>gVShQ>K;NkMWbd`GsM==zqS-Ns|*bcBa*#ThIMK>
zkP0-DJ}dJqj;*H6ybL$g@3;X*j}gI1NAU)gkpY7>Q3s!xEu-uij9ah()uI>~Zp6_x
z1ab&w$e1`-h)c;;6v@~i*@6AX$v9tt8uYMDrkX%dBb&m3I?D!v8M;Wu`-ULK42JuH
zKrYQtK+y*9n+g7mzEjj9V^+0HOwY({c#hFd!Olc>YMGpE6VP}i=fsdyuDC0pjB|}n
zX$J?M7{h&@!CzQi0WNWVd_v(5pD@b!g7^fv=vfx+kPDl)2~5;7bZE##N(#B9TohlZ
zF$(1+<YFJ!NR?18lS_<&o^+TL?o7y>6)R)vSy6n4`-D|w7lG#$+q1@cT<r)uz%Ca{
zNv9@N4|yaR-p-=p7PO_Uf<UZvZknY%6C|l=#)R8LYGtzj(sK7$CUc>b8r`78s7_ZF
z+<SVq5>+Vkk_|X{jIr;$Ev9rbKako|b*KKn+{;saP8>Q-<F?FMGT(9^6G4_Z#K1Bz
zTM@*^zM$BzO5X`Hp4pM*rDv7v)=DnHwHb$uVV-NGDt&_%i+WfY9j&4i19|0A|HfFJ
zT?ExJ@AO<XjS2ZUdgE&$P}cYtBr#QVA#&FS^f+2(ci$w_93w^rQ0Jf`SbnsHaItX0
zDw%F1+`V6~eljDSr%l+`<io?ib;VM*6Xnov#5O31UY41$%PuOgdj-|V1#($XZ?iTA
z%vdocgi6chJB^9xA+S}e&vb>Y8gA}N)OEVjaKc(C8fAEuox&a`_mrz+3N`C8CTp7#
z06hi}Fy$Km9gYljBy)0_96WMLt~C^#J$C#cxh}nn9xlp8ka5~#i?6qL8Luoiq>Z3u
zSOr+QF}AL%wwxu(Z;E$}{2@clmz#r!72<T?;xEv@jlu%p1ncD1*zZbur*89o<T*ip
z5C-DeGLg(WX|d}1?ZInH%Mms#cO;x_VsQw{WcY}dU)?sjGvGwB7>=3TW$P-$|K#pC
zZw0kQOz4+;{Krz5)b6!^!@tp6?u*6w6*L#x#Qg#LfE<K@2ND>~MrY$)41_{6R@+4=
z3G!eJ2DStHDWr@o@{m769ckF_c+)MChn-+pt5Y>`WnYZ-{DlqH4P}2Gv48eL+k!_8
z;A%3CUj~v`w<9l)naHEXkhxG?K1fORIjJo}-yy@2`c6&z(aDV9Ayr;hJURQ)OJ$ZV
zb;nZd>ROp?yNa$_7<rEUL6<)Yvju}5hXUbQHZGgzbA&8`O2RVVMnko(O0b#*wt%t7
zB3zgj95!|nNw&%&|HbU;mU3&{aIsHdN-$fku979Oc(iAMJ8EgbK=})j46=P?neF@_
zV8`k>EVq|{p<0CZnsIAIkOzx$K}?jDJ`&CeEJ}6`R{6%e%bRQ+`)9PsK*-g$D~<I7
z2FV&j77Xc&U@7x8@#wO90C+RG4`M9V1s-%`lo(t@T<^b7#AB~V@(@L>YzSEEvKcMK
zxY0=&rJ!Y#odGYkmY0SgR<J5HykVFYDQ~hl^fy##KuSg=Q*H?ysi_zmBs_v!n^8h$
zL+hAq3o;m}d4RCxY!A4RJ;9T0NgOn_JZ|qI)jG=}Z^KLE30ny&YB-%f=~Lh{oB*nT
zvt>stHwPuTGCNZm4Yf!oL3H_3ojlgV0u^b?d)f!F&ep&gMDZqv%4+Fq0{{{}?!#x&
zl#6Hz#{IKVo{f#1R4C8I3K29}t77@DJ2jQ7nD{^QuKeG;Ex)lfL)^k1KOcNf%f463
z3wEyhqM+Ar4f6hj291~B#eRXdnA<W3R)`ma-ayuj*Pu*eF_tg6{G+9e@b+p|ZRY5m
z*}xJ7&4Ft2`v72ES+*Q5&OaC{QPV^TYk^IpTK95bPFo1D8_OT<wC0Lg8Vb!{oi^?2
zw*OiY5B;ya;tOz{3LtHhKc{ER9KkQN&6yZDxmNyanCRKNhrH@9GUyt{aHKUMEH2<i
zqCTuD@>*aBv5d$*12#6x>j^_-#gsEWU_;)pY0e%ie^2l5_G)>RzUjYMX+`^*lEDUf
ztHn2ApfmH}+b+d+m3M;Ft8y0k`wv?W98&Ub9OM?RPDdOJyNl($!1naQYMQk3@_t|d
zav@bgs;U1p`sm)TcQ^UK1|K-8zkC=MTyY`dJ3g`>h7TGn|MD^PTwshj&>uebl`vb1
zdIrP}=S-i(+Xx=i6b)QppV~9zFT>Q1uD`saKJ&GhTpM{8)E5{dwn;v>Lp`Jx)*Tqi
z8B+U2@TtL-)3Vh$?H>h-`nO+KP(pECeHqMv#UIGVOXMqCaAgK=y#K^1anbjXuUjCL
z6(H4zt>TCYtlz}jn(FXhJAlZvC@ht4?GFHl=+*CnigC;~D8W#1pi7=&<RbY#P~NAP
z{9qr#mjo$EI;#;5Jd$Z|yL2{-87-pBcxV6~e$?44W{!Ob2Q1%@(oyo`&K8mdTi&kr
zPed)rZvMVq!AJ<C0Ube&b7Xh_r6vCZbz-DFI-BjS`BI?kq`4;Jo@q?V+Gq`!yxn^_
z{LIx|Dtj9kxv7P+Pkhp%R;a21vTtWKL&DUFptrpIq_c;bO{RB}!pwe-1r034Y;0$^
ze_#fOovA~!agq;+FB$s=<5#5|7+*5v7GJJ}oZiZ&Ypfq!Q3y=6?yUC9Q4^$%Lug8m
zdWEoP8ok<f_WK0j(BFsH!FG<JmW9q}j$>~IYPanT56F*T=g|%hhniym?J}6gJ0={r
zEUD6*a?&ZmPO}HXXYoyEdxl!vT1t50sQ%-nOJ_CImS)jS;2{5oG6mh1ENpmi%nkpQ
zRXkWBhxjv%oCQ2ghx%zO5~k$h48nd<EG|$rKpS@)7Wf^7tP=H*9d7$u3oG>z2_dY@
z!)}pmN7JC{A+RV%#>e@o@E$Vw;V658O)wL2ogAIOw=}-ZP}PsI<srDEQo5$!EgTj>
zo^otwwN2v;jv6n=8P*ZjG7a@OkN4M+Oam^0YXQRN2{GKT0y)v3&t_`nq|R#AL@F@y
zcCqqqMinK~rF+n=GB!IzMz=j|=0eRK>lrf_mRFTYufVUWXzD+d-u}!RSHnNIUHbSl
zgmE!CidD^}Z#qHSa6_CNUm=a{I&uXV==udJ5s`yN9YEzNiFDa>08$uvJ3Xj!YJd+`
z#vtUBb0eo2Ch{nUBd6QHk+obdX9VnEU2I9a45<-k8ahf5sFr4G_CgxZ=<nY#=bBO}
zja~yfyNy0yp4Hkp1B1xGzl-n)62YJ#vcag^Q6+<Y0u{t~MJLuW#0DD=;sZIN812yX
zD*L!`)iB<}xVc4!87x}IrvHH9qlQm7SB3{BA@tJZikMj$kzlscz!WV<re~CMcpEQ}
zQJtIGSQO3wR!%@h`}_2@(&V`YiujnIvZ5*(n=odX6fh(ct2lAvjQ)Cc>mg_PODsuF
zmmjrQ<2~c;r6Ui!LJHCfPz(pBn-nx5o<Zf<j!$UanwX9vxrtgobVzISvRo$lWYK$*
z#kft2I6Jln%^*r@^{zZ82pj5JUe1jTEy<$b1`f)3fh}f(z)r{Bedqf#Q;BG%m?Ib1
z0Sh;kTo?ll1DBs(SSJ_7IUWgmyg03jmLJMe^W~BNq%4;!k-~UG%^Mo3EJf+rD#)hR
zt*h9+f^!5ZOHyJ`AgNT1xRSBI87d8imzT*2a`H;9X0epoJ77T7nas?LlsWcL{flP}
zs(LsJvyOMt79&V$SONJm!VUNB3YlU{MDt^KtEMJkSyjdkYq<|ihrTwEz%QhNg;HUk
zR%!j-$~ZCTN0XT*IUfn!7XA>HRhuhS>6sRLrm!!U8Y<xTVs)?TSQh?fTkE`kK@wuB
zJ_~BpBtn6SROJ_v7zKn{|Hb&Bc@I?RlW8`KDiF(MdVo4X6a8oSXP89g21<l_#HB7?
zEWy>CmzmCxiqO;evcQz08coY^z%KV$z}gWBaD~6Y3z$Li%u2n|pCM2WiAQQHldJrz
zdaQA=Ty0ZofpSptHNJb=wg(8XOa$0Iu@tt_xYj5(2hW)i&vn6uS;nG{=HlyP5&7&4
zaJC!bg~6qD2%ds;cVk)#y9m<4y+&?|lYqKwmF38Tzd1q9;x}r1m0OGp6xX2jX(AGf
zkV*YX<+k_*oPjd9T2W{}liLHZNyx{XDR%^efX_yKmg@U~QUr2k$L2c&@bc``46@ZS
za+l4Iv;?_3cw5P2xhElhJ9vl&S<1a}kSCyE`hBrr&m3aJ#r?59Fl+?cLxWo`52S$@
zkD{zpe(uj0rNnkFYY{~DU|J$WMk%SWXafxi5BV<^mJKSUCdf!GdDsA|Km>f5{KEcV
zW0jmdV%OTRN%E-w3)S6x9D4t#!X#s|JQgP=4^0)!<X3@3T&kKIt7aUHv;ZBhkb1*e
z!Qg>1E5L@&n_xGnXA(Onv+bp6xnh}<)~jb0JYc!Z^$+L3Yc_ue{tE#UssP*^^E$ha
z9WG7QkRSttq&YvZKLGJZ3j!y&Pc#TY&FF}Rl7(?f!JL=KBKxQiqEB8H8$`Xj^^+w5
z5rj_$(=q94X&e(Z%<3IQo`7BBzqiV=SSpMrI~C!S<wiJl<={Rmf*=w43NQ6apNN_W
z?`FyltNa;s^t8#VV|6er$(oo6n4&rk^Mq5$TK|;v!4M3bAz2q|LDftKlr!skeVT#+
zRETOqH0lP&3~CpnSj)zk!+8&A(CKZHy@&XS446lQj|3x<a&zP<%jPsX4=REh5J9jd
z){O=(xaYRUXSivY4}+Ivn*%Tp?7Ts?2X}yIigvVlc|0D#Du)N9@`QsJtk?>KZ^NbW
zq#+!NZ6Dc@MqLip*BFDq&!x2<pGvTj3m^yp=CP*(*)?b|+vG2Bt0~xTg8cGKyfAL?
zczHIBp!XSGn=8*bXch!xUAgAt)GwD`$1m7iR<&w~vnIbu;}8-Pzn=GJaI`S&8YM?J
z%M1RDwXrD-u?){|ZNWRPM`+;h5}lIOdLb|^l|fFjP+p9krI@U@72Qc_rzJ1N&O#hT
z#{)wZ_@=Y{KLr09Q7A75U|^f<+C!VmA05*?tc&GOu^ecHER9RFyy9;lfB``NIW2_Y
zan{M|Fa8Yrw-%>b&XBS+l4Whs7xJnv!tPTVNO5&zj@9xvUp`D2uvlpt0S*HFSTsYd
zkk@=P`mUH4$h?f#JEs&L<eZ^=DCL4Tj58j7qWnDpWz`3^Loj|bk<|ty@V{lOqbItg
zy&cd33-sSISNI)!Sqq^f#WQAubzSYuC}J+}rq48_Q3LGXGcwbHJ@S59v02Tj=c-3Z
z{^_8AWN`g|;P26Pw6l}!LkF@?Px;7SGWulfdem_HS3(cVxH4LOwR{}p&OoqF3@uk4
zBcJ*p#J=9TWB?%O`SMwUfj6;-RVpSvx1q8XG$nlDud(U__!xL~2<rbfZUKa8u<e)W
zMOy#EOu@_Vs~~nn$vGKgAcXpYq-i(!Pn`E5!zYXxTu>li`$@S0c}CwD;u`A_nTKL)
z=`V=8@~z{6rEYOA^QM#U0uO`G6#V<NVT{3N(eY*SLp%ee#bqvC%xV=SPc%e@)stkG
zE^6^=tlx1rxZjVun8k|~beIMJ5lDXQ&$wa}1lG!~T_P`(^;s5xsn80vQa2FQm}Ivu
z;U%Y?o4z7#*1*%`Vlwh)WcM`NQmqEcT+}@RQa>5II<=6I?Ab+44MU)PPL;j7xS47k
zqKJ@pQX+f%hsKD>D7op_dfCT+F`ah7Q`Yrm5kk}VO#rM{RSj85@)LhH4~LSa?HI+q
zpH1CZKN>gm{_)57!Xi1q{y;IS>LNKX4ae*tr4VW5&os&bm|)ah>%ch*$5u+41Rf8a
zi98nY(>8tq_hPh~2F~b`+S#Z~Akyxt3}<QYZ>z><XC9-GK0Cyc7>)yMZUNbK^tThb
zo<fNh5dS>s)Wt&&#|@Fr{w38Ca!1%mq)Qhy7#3P2jvVBqV7VuZXQJA!G0b|{4vxvy
z`J~Z(a!49hqj~YoEct1A&pLO|%y?*e#_%?3X)EP0e>NW=Sv3-bymfI4A!^a0_ZE6P
z$PsA@WIteKn;;S#>6=wf-{?_Y)Lh%Edp9}SabdP^6JD*o$}zT61=)%!l=J8sOKs8N
z09H-arC3P!SRaR?m}OAYs{q`|aSqJ_HD=T0yH$>l(bRUpvMOxz+Q61X$~8`~eJDfw
zH94`1TXH5*d4}=jq`(Z0s0Fw-NVhb7OOqzTMx?vF&Hi0_k+St}k03(qV@7(~roxWb
zf5520(kqS^j6Lui0fOH4dY0vWGt$SWhb2X0q*?R6Zvttyig?SQFq~{i&cj*KFXlt9
z*%~>;U)Gd992H2ga%w=SwSi~~%*$zMq6Rw0KtmhkbQ|=r3y+sG{AUgTF$7o0N;$K|
zClEhjvs)rR>(X=y9qX!`DwDU$iag*EP|g_3&4?$#SUd}A*=ObpJypnm#JORaXtT-V
z9dLS{3i_H@O4DkN^|w7k#gP&j5LD$xG;K)%PKadD1odWUA`GtFDuetP!bQLZ=+#*%
zgAHYahEEtic8m=1H&`H;^;IZi4oz#_jwCNetfOdKJVW#kEQqv=$#8q-VRTH+DwYvl
znj*R?t(T%;#G*2G8?9YcHW?Y;v(64EkXth<PSe!7av5z4uVuK|7y~=UK6n`!8{nv<
zw+NRAQc1g*y%&M>ki#=_mVGfEovBdoV!Vx2fO-Z};7GJMNTQuiEJk5W@NpSF#A<x3
zr!g@dDf6r|0*9o*DtL7GO$sU>T8$ova<&g{x_XQKm2=WB^d_2s!rfjh=NhPt|3<aV
z^8!40-lHTSCtNA#$1kw%(;7xlv^0-gU?VVx7Oi(K?4k~k(L?2;SP)CQD%jiPVnan~
zd2O*=5{QHghj;2mxLyl`FBBmPhR3wXHU-Vd!LN#aQy8jV;;5mV1YZ_84Cs0}ZDb5^
zk&KUyHa@T)LybqP(8;z<Gax8wlx}H^p?V^3b&CGHn@lt*vx8Zlt&*%SUxU-|^HN-h
zb{S2vmug|2nCf#<&SrI*EeN|Q@6rS}90ET7U;dv|B;1>1gp(tczJ!(`x+V_fYzL9_
z#EYcLu#5|$MlN*>*-c_zR4teV8iIF9s%`OSqbyiH2&sl<_!<WQ&VSWbuLjqyO#p1u
zTD|LV-|=c$Dbwt}gUPA}Ekx7@G2Mm(77-*SGi>MJsX`6|=(Ns%g>e#`dZzuALpW(w
zom>`&0%rn5bGZ@Zupx4VFNR{l0zXG~%Uf;CD;+*M$Czo&R>=%p*d$j4I_Zo9t73rX
z)&9&Xyiy~aC)fB0TFn@w)Er@at-oyw61D)VzRsROQqJZoG$8~DriOgPMs1QC?B9*`
zM_ep725f3b<dL*k^(}H!jI6c6X))L&H~ZUk`?5U7QXBz`47<f~U~YYf6Vgr85xUhm
znTN8=Z9cl;Nki%6_P`_VYF_coeQ%K+Eq54E^gOw*+!_0Z{0=x^<t`r%a)Y*NqecAg
zW~AZvfL5uIAV`{!dtxUMLklJ8-UJH;N4sY3v#p1?e<uW@`{M&teunSq0Uw#@VrU#o
z&uL95`FRYbTZC#~V73n?*sR(JO_vY3-0xedT+!k#m>0zAI;`~ea6$;oJU$OM8a7$2
z?(z#88PQ<*x?&B0s*^|3Y>;S`q*lqJw&6&Pua#dK=8%B94*$PF9`m)}J_VMzBJgpr
zp}F!apS=ZFC+UGuDJ8CzdOs@+6<8*-61EYU1V+{zg*)5FF^Szf0>Pk1j=Bx2oa5uH
zfY7p4=K9-!1d@a4Vctj)ro1gO&(@NusmaQGUn}0l%}hHrip!<-Alg?Wy-27u2`)?!
zD%t7|sRSyEn#WPA5hNX1+{LX7RT?LR^8KKcB|#Mf(MVX922~72nr>E>*_F~*KM9F)
z%Y8^zA!iK`8WvXA*N7-&lk1hSxe#(0K&WLnV1|HL3+LQsS(U&gdC??8@th&Ts~usy
zC)UXtn-(HMRbJMnk5JDKyTINCcAc!VkF;uHUe^12l=BdU+#+;jgWZ(HNZ_u_kZo`D
zS5+7%oVd(P?b>AL@rtfjNP{nzmQfzvEf735#{!#+l3^y+Al7&+*<$F+mYAPns}I48
zk^#<a>bI?VV<1v8iC~UwxAoSdE=Q$29_!T;)N;K~IFjhMStCydEEt@|d^_mP-OYCx
zRdwx(VpEMF(+@^jB~5mF%6_HSU7ijwF$M#c7XYLthk`aM#+h3HCpF4vLhjHK@vu|<
zFV5aOyw2iiA0BGxp@a|!?hTvfZqsp*3%Fq!Oflt1I+ldh(2;D*bQ1ze=%IxY<SM!M
zCO2ZsIP@9<2_cjKp$7;x1it&8vJ&~de|@>G*Drd`^X$&f&dyGsM;$D>MH}R?v_2|A
zmhtj<;2Vd7b{?&QrU&kcm>QXsqy_vdPsTI6O&c-_;L1}zg5CyrkHPJnZ-qRaexd!B
z*Dia&ahzuiF7Ug1aHT&RV8iEDoQ2N(Tmn?R!#F!D<oTEyTg9jq)+~O(hSByCXtC;V
z@Vekqn2h?O?*tu2^xM;Y!K#StI@+{tuKYPp(XgR|<fVXP7_2;O`}T6|qiU8+cxSv(
zOAf<dd`3#DtckJ(2I9ZE{7$8aC1+sJ0JK(afECDSfzp0mICQTh1W~_KK8G)0_LaZ+
z%*Jc<OewdFp!QeOmm48|@Y?dgEctuDw1asWhNsv3ZK`$JP62`oH)^fCZl@xP-iTH3
zXuV-)tO~NpoBoY9L%}%gg>7g1AN~uiCRF~s1p{n*Z^b=XRf=%l4f1x-Hrtiwa*%i5
zi8rR<0)c1dpFyCwXEw>Z2~LZ{BUT9WJbBN)1DF>7|9-+xsuqCVB~h9x^MQ#=MN{QN
z9~1TM_4Z`wnVBp;a!jem5%bVk`*F}i{+My{uK+)9dWM#UoP6S+@Zy5OA3l^V@~Pph
z_rPBAS&(%>7I_>0_Lp67(Q0U)`+DbZo4eyW`6AvY-OHDD?hyqvMZRh^5iN*}!T37f
zu7Gh(zVSD#m_2uMll;e@F~y8*lM#a7wi;=TQrA}bZ{S&Pce^Rg+5TOtx7EZ8+<*Ur
zXL`Uz+W38?31BSTg3aBfja&b!vy<VN!40-+8#7#*u%R7tWw$nFq*UFIn+6t1-a*o)
zDz|$ZH>w~a3SrDOvPT;=385C6>=`q_@?&%U$VWq8FttUISDCtOX0yE<0`3@vCzSJ&
zz5N-yE_&%BPhIwjXAG6mqqao$ZR0_Odi>2)Xh&ebHg29BgBwovZ{w!9qmGsX(u#RF
z?D+lI7SJ1^ln2HFEJPc|{eyfGvtaO?(b~v3oP%T7GfU+Ve~T&`3s%g6U4gP0iqP5M
zMF_S$G_95CO)SrgzN|J#3Cl$u7O)wQ)^Nw4#WP$j)W_)?KD>?E0h;S;3y?5dD@UYp
zA@5^cdX;|e&mvrgHEWKHv2^?e5R<NzHvUZAm%uCS0c{O?WWji&U3`W6fqrWiHEACZ
zP?v&K(4o~%aQRCifZ0xT^!<-ObuD_+iXC>c^>@GW;>)G8kJ2K1(h$EGLSdKKmLdf7
zAS}2@y0%esDpWO^Ris-Rx1583=HWat$Ke01sdtnE<GE2#3h4sh1iKpu0dYy)0~xR;
znLQEK$yh2Ts>6{v+TLld2ggKVZEeVVOd6NXJY9cnb)6pT<C+}ISoKwMTr8jp1J%g!
zHcbxJ8?8jyGqA3LDR1S~(kngNfHOc^*d)E%xD7W`ROGce`uJ<O;ULZFztT502*(GG
z8qhC^AALroUmRx*(*Y0dpGX-p30~5CNlh>q-~+P=CeA8kh)psu_KYTT>^TD3@Gm%p
zoRFYpu}PZsSICLM4H!=hL_sTuP=2jaEuD-f`8VVStPzLHKu(TvU`A(nD9wJS81j&S
zH-+Y#Q-eI1^g+#DPD>LpEP?S%EH_P0BAHKvobH3to5`4#t+*%#wNabWnPX*eun%+r
zsF5M@*I+mvWT>6z3|eO9*&xlRW5DZ2OVA8+d|(AihzoFP$p{)`xKGKtP^i}nFt*Bw
zVAD`sD;UoOL0d+~UNOi7NV3+vE5Lfsu)P`r`-9fsl{0-z6F}^W80F81Ll&6Q<`MqP
zQX<UgIwrvbMM<e?iH!ZhZW>Rvq_=SiC=+y9)11FAm$TA1^Z{x+R~c{5G_w{c)Xc4X
zL+|u5nGhsbP+f^qP<3*4>>tYk6qTZv*g4J?F1-9onHVQ!2+Qruxovi`(-4{@qRR4l
zu`t*RSqFfXvCfa-k<FzUP#4%>s8~`glMITQ8FHb2YUv<6VH@S5m}*cNi-uhsXFpe4
zUJk3{B{3T|m!3OZjgpDkCL<k)T48|^NIAl^3;lhbcp6~YVzwyOT7ZJlTE;#bvrIrJ
zIR==V0Aoo91~+Yy;yC8hO10`TZu1h~5vs!s;8qb;ce^w|!Yxd>PA>J+U}WM(87hxh
z11%s^3@P1Dm9pSu7znK#)CO;nftr&m<$)mdokL<qMSwmOb=m=XrTvYL>Oci*epMOf
zQ930fm)SAOhuHy1i`DVvpt4L3k@mSbWT@?ul^O%6%PCT8yPZYb8`PJxk6@`-v;B75
zt<xOZR%RhEAL?84cg%#(ZMvffX9!$?Az9=C%drb}LFf={tZ@)q7trlFBM2Qz7QNC;
z{HGOWfm|NHXjm96!t`P+!F6~=&;_Ie8mVBAlPhEIA;hS049>c%0()gj$yX<2lRUiI
zG^7GIRLC{0hNZAaYu&Xm1a3W0<V+ue?8eJ4=&__I#j+vS86wAL<QMT_A&ezixjyCw
z>tlWgo&DB$ik^Y>@=K%QLHQHphFA&0D-|@jB5w50o$=_cLhAvsjO<tWmF*P5yvqvM
zSnStE2sFCNZw!*`bQHh>Bscl+<R&yWkyqYquatp3@X+33X9iVa7(~so5x4qYaNPvI
zC8PN^hs=+)V!su|%kBPU2u6$}HpY2}ExGr=-tyZxq=@oIiVHxv(-3*$@p4xnlGVd<
za(9BnqC;CErSGvdvuuLg>u(vNDdzGfqEX{_|AWaCdF!nEKL3*D4%V=>3M_1jb$=Xl
zWF2Zf{|6Giz)B2EA__zv3}^?VPf5A_KAoWDI7KVup_pCI0S&dxUS_XRYUvG9??dub
zk>PKGcsMI<*e1xj4Z!dQneF3Hla3ru@)l;`Y=lvMj{nY6Lq%c}D9P0TghDmR+yHZG
zrWl<_<^^|{jJSF5y7{&>Ou%Re_J3(+nlvV~TKXDVJK-F$>c@f@5Y>F(DqH9;syWxm
zo!%Ve-P$fe?5dH)_E9C1v|&zK61VeY^wHIdc_4(PF+Uq|Ela4NzNrQXSstkPLB2a{
zciFwWB4`Vhj!MysBqu9l$fLR+C#zx;)i#C5s%<M}by_B_DGY;%J1mnHr8PbotscyR
zr1aam%lcX$jg6;aX=C{iSr-7(4w`JD3`f*UMAipgA&a(#IoBHkVl=8nL?1SxDKG<W
zjU#Gyv5kT9V%%&w*%W_{tf+!iv)R``D+&v>A<xa%u*H9&<t3zND_*!YE`{y)B7FMc
zK+{P0NB<B%jc1p~Z2>Yv^;xERdptnx6NuM2X-<n?hFpbEsZ&y@?1(k8SPGi@)X5_;
z9lOQsOuNDca-MJN&*)2ty5NucG*p;ry4?sv9(`Eg&B%7URvxoKue|ao)U$m&4a#x>
z{G>^qFswnv(Z1zRCfqEA#lTJsPt;<0D)0$=D*Fwhma60PX`|@M$&inpiA|WGHC~>L
zRhmN@u~N^aVR-a$Z9=m!%j)w%S}NJo4jZOYbLYJf&?AhAPFh4|%b(KnP4;Bx1-z|(
zIJN_MF~k_kB@hgO=06*RmENx9CEM59Vw4lDk(UE}mW4t)SNl;fi6;4L5Fs`l1}xA4
zIK?Xtm02^XslZqlkYWw2xHBJ;_p5=4(Xd;fvKl%`_~2BFu9epuFLuH4ZX(;F7X7-9
zrb+`wV};l^(hN)^QU)q-#tf4)r8Om!krVKbcuNgoymXr6Eq{h_G!z4M!R76s?K3F!
zyc0MmhD{<X|BN5Uq27zU8;sI!Q+Y3j15Gma3FWKg{rCcnX>#&GS|(hk@?mU0P9^QH
z3_yaf3AWph0(|ti1%hx<e;g=L-w>5E+DRdH%D-Y(wWDfJ0{O(>M%)F?fXcp3pP#0;
z86t0d>@$C+IVT`MmSFmKf(<r%5E$$JFh#FPKKEZB*O4DmU;)xu?(~Zodk5q{tcPyj
zFZ~&$SzL?kCca+23Y?6|<Vxi0_-fl~++GZvTqNHFW^+}gY?kw%U?3m`o@THRtO_|k
zfR@$LzYTn$GBvB)7Ri6@tXH#gf_#^zU`124Y{~a&8I-<w{&~gna7$a&aYk$A%r0$J
zFP(Ih?Aq21KWK?mUQsH$wROYRIpbybwq^_?sk+k^D`6Pk9&OE_$G;dANQu|(3za5h
z&$ezm!8$B@k#{p=Y>ez}H-6Mst$dYL8QIG}Fk=`Iui3)T{4M-PnUU=6(|Up_NQ3o2
z=*3mXi0tD4TTLrvk32cp-hJD;O#_+FXgW{Y+s|gNnp!CPw{@GPawLMdMGo+1mM9E0
z5W@D4<2?osg3MUO<G>aq(N2L)gu$PxjvUl>CtFZdRv@ew2Cai_z7x<h`4At6rBo4a
zp<y3CX}fc8ygd}*FsMBV8^NJIa4Q60G0bB5X~KSsh|)Ud^RTvVsvJs&_p|gi<EymN
z4v^XirqY13yjBi(VB#h~P_rD-*3G)56_Zh>68@Q=#}mj-S}a743<Rl0r*3v>V>9DK
ztVINA+js*W5k^<7fwjGzKZDf4%>}iumiDnyx{!Fnl*KPb-gJlfg<J>lu2ubsno~y~
zhaLd<-WeAroeYJ|^(<YB+dw)S(2&=t>rv+JVgMrUstC-#YuY<Dyzu%mj5&~Q_Fg3|
zKRG$d4h2Gca?(8(K*80hQ+gy|Hh|oXxQ`Cd=^_HzQ*dQZ0u<p3RE}ZO18S<r#+YCe
zbl_|T7e6lEn#vFYss+c}2_B*uWIfX-?9piP^u3J2`t=+rz0=Qgz_iRv&?lb3c}kIP
z6Fhl+?UUnr93%bwXC-wW8_-67`*{GW>J0FoDS|_+!wtPs1_r`Q>Wa{79t~doCCXj6
z`cVsUmYnF%7QpI+D$o2^PHOd@%KPAsC;Kxs^<l5)0K8NDnWa8i@xfDL(j0BoP*_iM
zOrs%(-FF&50Ai8T{TEussDz;WKL#fG3`&!2HYEgPI5>b(BbjBv3`vJ)3&>3GS1gMO
z!PsS-{-L%fNE1NAq`ZK3+X_h2U}De(hNS_J`KTMg$VJBX*}XH|zoDi}KRJELBm5ce
zS|u@-?i=Y0(gL_h^|T?-K-!9B${GHpMY);T{!B-LN1UrBX5NNTM%EcLAC}Sn7MY(J
z{G;6Q7^8tc1NzF?#6=+=RhIK;jPn!U0Fi8yoMj_!zaFa?Z>T}2_Q(m6ADcF{G>giB
zIhhcUvOWUV4L7u$9e2@)%(P5tshkrZL1Bh<qev$DWRz@H%DM46Epi~f^L$9_$zbR|
z-vFvDM}+tV{(FmcZk<euXDX%A-?LOMv;`w6TQ0KyEUmswE=~xdM??{*^w21m7*xoH
zqaKs-4Yy-t`8M1xZa9mu7X;K`PzchL!h{HR(KM)w{24ccI^bN~_E|$GqK<KwOtvWz
zQ^suG;=uIyynHFK71tFaf~?eGCPy(fh!z>vBfd0#vDV^FXbg#+;=eGTh{Q$j8u7A1
z%4`~@L+7O2F1O}-REsHs)h_MKRs_OSI{dH=YZf<L@ls`9Z@*<*(_p#GUWdW86lT-v
zfWKGY{*v?g8*rKt>Hx-W5pSEM#x|oGZm(MVf<>b%WvU+yXE3l`dL>a@3pn#w1MSt*
z3?@vO1?xBUa(duiLw#8^YrWLPfyHHHRiYVM$zo>ojx3QGX>S<eV19|q6VBGD0hG$L
z2})P^XxK?m0Oc|Fc>=0c+h)s^KAQ0q#$v3Js{%Q6w93_S=1Ni9C0AO2ZN0``7zFn^
zyu#NefJ~u))lO#mw5_d}t&wqE?B01Wh9Gv&{w%JptgS-jsq1|?l*Cz~8P*o#HS)`}
zIpjQ4h&F&Ywt_rw@L%*M!Z#YZtyv5=I%$yb*2}N_FJ|>EmtXr=RD3Loo0c<L8GqAu
zCkwFgF4q=QcEoQAO352DRBnzN61o;l2vB5h35Ed!u=R>aE1~kJvOE8U_4_u*8s!#N
zbEE`+dwRwWP=slL+z~^TXNonv^S6OICY?`~J7crZ5Ug14@)2lqR6h$uiMwMj)G6hc
z6a-vq8K^WY_r#jU!O3*54Tr@ru={sD9Fw=mjv1bIpAQ!f08RXYYC=T@j%2w%zEO_G
z6qF6}fIo95CIZQ4%Y()zX3PLV{oa0u*9m2^xk_-;ha7d5IM#}gq(1PTDMGp~Md?{S
zKKT{UjKv|L*Prq3r7A@;-<+Uj<c?#1DCN!#{L;)q32mOiMaPBAxBo_Egq$=486mE9
zt~C1SywllW4gX&j_%ob7@+>Wty3lyW2?GYoqC{&f@vB9&NT-l87yE~FqtS-S;>t^G
zYLsow$kH?!!(v&{r$LtaXc`*<0GM>E|H|@U@Qm=l3Rk2zQ9TWtWX&lVIKR@gu!8Bb
z%Kocjx~vX*0-MXq8k-qzdCIFWQ^Q-hwuPmr@_@Rfu6r@VV7aWbCHL&xPu2%U;GBi{
zjCih%vLWbqOcr%Uq{&BXZQCXxxsgRS`tKSt$az$<X<>(({CCQW-aMO|4MEGq;LW~z
zBaO&<+kz`-ON<OIf-R&#3EVRQmKr<4@cc`Ya_mx{52BtW4+j~+b`MJ|m8(A(0(+b&
z+X8_|_w6Oy9RoBy(-_IEYqmG4QHuaF#g2G?2pZeTBMDrZ7+T-*J7ZFshH6p$s963O
zbOC&^U^B!+c{FzW2boD|)*z1sT@>Jap)bngag1OS%~hdQygXqvhuYhP@}$#WL1B?R
z6<?wtV=376(`j3oj7TosD9;!dK^`;s#<TtcRq|*t^PH`7dp%jRJRd_AK?gydi5C*G
zwZ<m&TdM7}CjBW01ot5p38rXiq!ec9Y<SVeMU_@2H~cw~1KONm1mqC|i!JW6mmD(Q
zgfwvSzI!<s-~<S1f3ZK*{VjhDWU492h{w3gE9sXO8zQ|-^W|@`pV>O(I=$-e@v35l
z2kKXxwMqV-$WJ@PXng}CLUNN;@|qo^n%V++J?%3SQ<*uoMBWGh$aL|{>VL7L;+wVt
zm}q9fKIfk@Y?a}D<cx2nPv@#mp(b-^C3EHNfW4>|m9O715`Z$BlYcscsFrvAYZv}i
zGNF{VPTp&~(<*e2&Kv)Oia+l=5Q=WXiqrQ&LJTctb}{hxmGWV%RwJ<~`Yn@>(xg<E
zjOm({?c>;R^@`z`fg*uM@KO2)=E=VdjibkpluvwMQ-45RQ2IXgXS^^&&iu?LQ^E$7
zg`IS*{M%_5GAY*kx&Ji}Cbs4Bg@2{4Qh;bx7wCTdGB{p7qekT`BYiaPHX-oqG@XWv
zX>bfLy$$k>-A(N6Zbc>j6N_ce1P(m=;u3P!gjfuk`POy}jmJvlzpc<{Y=5<U7ndE5
zDe@Gm<a>JwQVxQU=$gb4l6Gnpn~eY5rJdUBFdi=HUF|v4F4--fm!SgC?(NiY)LefS
zDsVD`Wsi6O*{-Gpsat!tb9+z;V!Px=?abI^Q6Sv;Yh|x^rUfo+XZQAJy!zpB06*Ix
z``E;Jqw~g0kbTn&bYLj!mi^j!kRS|4=&_C~9V*fO{$gmLysQssr*=koe~`y(E6W=-
zR1OS4P@tW`r;w-RpmrXUXKo2-QB+WAo;Wx`V1^?G9vl+5ITM}0f6~san`0qL<32q!
zL4v9T*fd0V$WP;gA!s>sSUa`Z71dRsxoe^P%>Hby2YbeT9BzPCV|$NCBO&AtBW;FE
z`Ey&-m6MK{BuBP03n?ujbaHG4&_?)_IOj;4KoqNlmR40k8p=uAv{FmfUO@F<O=Ru-
z7Yz#q7luZKUuX<$?-Ma-m%_7g6Lt_#e}#02<>loHrK7)W-NG$EuXVOvr$DI2zSxb{
z*{5z5#f+7(5>=M~mvXEgRIN^q+o)?$-0;eZ8PY9g1rv<x02>G&6I~sf<*0-U#@s`q
zq`|Sf;kw5~J*0=>6}dDea<oH8ix4juxL@R$_#TnMAUG>wb!;pJ;v5S<kT#ABT;bzj
zQ`RJw{;5Ve-WO!Nl^kj|<Va>#^h^tab5eTwd-eteMo=V}XQT8^vs$$c`i=X<vu4_c
zDKM^;zW&VXaqym$e*R2b!h!aY^eO!f3;j{ydVu|B_m0QNz&PPmh*Zqt!au=2)vQw3
zEi^dlM5BGQS}j7>V2Pa6PHp~!M##xNk$F(8_WCJ7CPVY`<y1p>A^s<)InE3xl!z!t
zkz0c^Ly|b%alt4pH}Zlft_R;8WH<w6RwL$auun=Gnzb5ODnsIU)>g65nhb5{R{D}G
zLI<%9*wpL<4jFa!(2-+?%P@x~6h>yDs-ouF`EZ|w=0k+%qYOloj0kKELR|+aiZZgD
zTeso-2aCeNJ|iZk#9qc2DmgR3+n`2i77H0=D`FiAoYv8{A$<o9lraI8M`ckdv$zFP
z_;MK=Ck257O91L7^@>ou8rN=T59qnnM{piA6c!6&(Qyh-%%tXlW{VfAm((C<Ia+CY
zfR)YvLiaq}>iM>D2$m|939X2fQAK8yfT$p6r(fv33Du6m+&R8zdJ~u~vKr(&(VtN}
zf=;Vh%;)+Dd`3-KYp-(ca-O{nRg!(6ti8wiMq99v0<RaO$$17<3eY>WK_)p8S&Y}4
zJoxWaP8Q0AMhYljue}*AiVcDoio?vhkaDp<GaIrB5OPVp=hw$7U@hZVW2)pjDQLG-
z^RJmoMO!3T1=hs>HZj6TU%9sw+L!r||8XZ5`8c%YZjfv&4LuncWru?fd@;Rpau7Nq
z+93W*u`i_-1)@RZB?&}?Wa@{FQtH$-pr2eCXY*2!>J)!hJ?eI)NSQx#U7ZV+as!Qh
z4AD-!!iInn$&kAA`f@nn?Y6IKx6_;QQ8h=txN+gjd`ZC(my#sa4pHsD@H~@LR1fFb
zmYfk`bN!i>Qsa9{BT{tNQOlT5TN_8Q2<A}r&`b^BAVl(tMhdn}Gw6Hw#Q$V^LI6*U
z3H^0(#9<Me%HZu8{*rB=7Rr~)?HNfErnCx#-xSLgX?;u8H?S0v49geIlPeAXMF<_b
zDn>`J3Su~Oa<%=|r`JHaCJjJ~7*!$~A8|j*wT6Z|uxObXBQSP^mD#UzywHcpTbns=
ztqpX)@L>@H5$<DzMO}*<;rbvYn8+C)`AcK#r*)7UY{wBsjzCGdF=4*Nc^d)<9p1l+
zanLdn?hE;KI(;p{G+Km77qv`&V{5)NGhJ>9<Z2+dIaF?L*%h#Nt3!E9Y%V*=APEmm
zC*n&8FeDmgs@|4RfiwVKDypKPVK>l%{yuWM;c*x|dU8jgg~|hdr+|bnz;}Kdi!aHb
zpxvE*qV<Xbzok{d7H{3XEE8O{6}ih{!k%dg1wE2?2XfF#4$<FoPuk_RfKH?B!66Xl
zDEB(_5XjVPX|bKEB|#}$CBHMckX}}SZND$zhq$pq*Gd5cQjy#rk|qRVi0lv8qTzqc
z$b<2dnY(|Va5$UvucrBjf<v$kV7<LSp%w1MoCXBX3iMQzf;MKyh`F*Xo4@*qCU_W!
zpBo>HU?S?gw8|9Cp{zOfXnwq1oh`{qgJBQRfUEI;%cU{sU^HrJm9l=z0s~-FDIB1v
z*}pKBgv*Y(9!#@c6jw5GB#UUTFO<dpnwdVe52-BiXO@>ug<+{I4QOggN~>jA8r-5w
zHe*+o$FpsylM5VTH)MtHO>0(VH^|DgYb)WzrLxVqaHi0%@^N@yavVAuR{Jy4sT)A)
zD`ZWK0KwBM=41ES<6><bI|dUNEd#TxSF09ES?6B{173(cMm1B&dY^|UhB3%YY}jBc
z#|TTHV>J2LJRw>}6~dC80^|7`19~XSu=l~Xwoo<&+7V7QRmCIO9Ly`1EuFzqXj}X{
zD>KT>q(<55bNyd;DD}OE4YEvi85)4gAA%zCYhem5#Rav^$5iVoFu4N${Oy4oHF8i|
zMT#yYA=YU_Mx`{zDJ?FmE37P&9q~SP1KFN_>RASfI9e=^#3>mE65Qz%5}8O3#5OII
zKl%VJm8esKXrb4bp^=aJH){F9Ga6DJ^JgmP0X!yXJ)WL1J(}o86rOJ+#R3X<j2ojl
z^*VVn@QeOusPvwbr|j@gEuC5^Pp2K7hajG3Vt2B*h>)Fw8|>L2_Q9o4rJu8znMzb#
z3%@Aq3q2n*LoPtjht~eH^}OJ4+8quOV!T~!4Kn)Jp|n_D^tbpsZ8R|P$e#n%mP_PD
zla^XjFZrAdHf2d37AFGMaIRmr#i2U~g70E3B&t-aB=px%<up(fN<mdgoxBnYaGjAM
zxhh1+-~1QrW&$x$4G~25tAW()bXGOa$=_qGuw$}NG578@e`a}pq=XIfy4`cV2lSFR
z47KCYUI~7-H|>Z11AEFp?1y4Yws<Rz!jN&(*WXUFZ9-_5iuLkNJcDvg+r}E0i~nhi
zu(qtM6mpBa8x)P|zeq@y_xuy&j>2DL=l=b8M$H0vU?qwV{26!-gCF$f`OpwugdUgj
zkpVm{TQCJ)?T;O6BuwflY}KheNB(6e1f66M;QWbwr2d?(@@aw|v5t&31Ec*c$PsE&
zWldcbl%jvfN6-eg%I8id&<s>aq-f!u2As%+@<r^dS`5v4Xr%dN97XUo=2TnMD;!t&
zZ-!!hWxG*2BO_l2Lco&AgmLENn>gyZ%MfShFTiZ5Iu?)ew{ZndLlvc*{I{LE8_Rh`
zQGcRe>pS}pt?=0JucC-qCf_^Aru(-lk}T5ROg;cs)37!0omK&0>lzqVcWG~SEL|WG
zP8&Oz=N#G9e=#0nAzFrp-P)V6P92n3`fP-h?A~6@c_<i>L0{rM{96woQK?x&sANxv
zN9{yp{IF7x5z=Kqk?BW1HqUifNcIXCvA<|g)MBuaz0*7!!4I`23;t!3pncl!WGhr$
z^-cJ{^|G(O$t)sD@Ma(<()aTjDQeNN+=Q?9PtO?tV8{A^G#7(nu;GvjY)wD*x5Jf7
z;y=)zQQ;;zLm!cY(tGp3^Pu8FBRaUf+V&8T0;f@(9O9Gkz|u5Nw#<{WR(|4$Ys3vg
zwnf!OW99HWyjpT-dpA=c<eBy==o>$6zjGgTSC^yDHMzJP*4}NmWwjOZvjigRQfh9A
z9G*bjh%>tYw0WHz5$}OjQUyk>D*3r>Hz0t!K#u&si42mnv9vb!c60r3wsvkC<256V
z5^H3yxpw}n0gRlHscZw*K2TE#6Y&)35R`+@7J}oWm?bC`7d`Gb)XtEO?ag?L10kJ!
zIK4dp1ao;i8$|%qwdk8wCtU(z<(YDn`eX(0uI<%ojcCH!LJex@=GzgcgXRj-&IUQk
zXMl+V0)Rz@cTby_%EjPq2v^d>-=?5LW&<;!9Bq@q%#H{LZq6}@WLcwTr5qbrgP2|<
z$F<+N8#eQljveLrSRL3bg^w-p()6^CM?;m<oU2{|xt<x+&Si-XbrSb(!6WV*LPVv7
zNjK_hmOeovV^9jZZ|p3uCIG$=$INKYT-p6v4HnVQ8be0%hH<mM4F*|xT7?X-9Xg{D
znG6F1`=|q#W6{}SIl-4jIg24d$UBo0(=)51NLBu%^lX`0MtIy#_Ge@z@P^UFD5s?N
z7UDfp)<!us(4tXlG)<rmht{w=EwEikANT1_-}IhukU{>HWz|r^1j=X-lEJO!V6rRo
zGKU1Ajj60=^Ms)>@{qBkM@ydV;|MqchWW=3)EP63Jf<}=JO+gHVpf+GGQyCkj#e4z
zFHrGjv^Ypl&hTesp@2g5jWgS;WwoSOMj6`C4Xsi}x9k`31hU}f!xIcLqC`B#*F7JB
z3bSRbj|&F6{dR~9w}Ul~i`5M0Wpq};r*<DhXcfHX%7DiQoJ{2hoJ?=ePk5ve+g+#B
znq`8GIzG#C3Uao;#{g<t8y70aV|2nfacICuN?>Y`i3vm=n#}}L&b61+SfpVi=lM%y
z#-!pUus5D>&rn4QF``JiKrWA5U>t`BC<l#nQhbK{3-S_d?n~rC+rGRpgXAK6TMf7{
zpIsd6ig|Rka!Gr4I^m8kgaKE>yEF05V1&%oNkPJgWeHiOEM&$>q3v0(o~S=s<a^S9
zJy|c=z(7@H5fUA0WO6(}I6eh%DK>IpC!abg@il-G>S2L>Q%Y?O-LE`GF14}Uur|f1
z!|oUwsOT9n8xpZ!mUbKM&XsdpRm{aw-rhx(AtkUwvz-E+eii;c?=fUlI^$-sh03^^
zF4&Glxoz_=l&V0s7FuRR>0-IepEc~5Nr9EkBBa`YAByvn3v3RB{i<AQ;!mcG=A_n8
z*jztqtW0%4$a%xQOR`!k(^_o-FCLn3C<8Rgbem`pEDd$()M%AX2KHbzsCNq-u)*>S
zn|T6~chJW`F1J4iqYpXx+7&*9(YxjsgZ5d}=}KRp1=<my1#(qdMJ<Cpua4t}7>P!?
z#)slTXF6z$ZgZ_|I;=6Mhc(kczq0#9a-Fe(Q!bES*uyb-<K_Aw3h-FA?uI%bzie;W
z5e$N$joI9%8tRG5!Jo(t?cE}a+Aq8`XP8CyMu*FY#i}D%eCDqb@oq$j^{?$O282K?
z|BZj8dM>8K^#a^8FXW~`833FEF$9fkfN(e+<mO<mHI*>j=j4_j%&fi@s3g~d#BL4Z
z0a{rEf+p|V{2A|XD(oO|xjjg+5T$HY47kJJGU=5FSuVegFI6PtHM3mqG>jG&mdafY
zM)<FZJrfS^_Fp_!h)x5XDZ1e9vCqfl=gYmZ>7yCYC%^l_qCDV-dZyeLM1AGua({rN
zw;1od`SO5&sWKx_N`IyapAXvGj9r{+iHztx8>|DtJ~gGVKRo1wP8r)`gL%x&m4WOR
zsSl<G0|}af(=Sygvm6ef6Y?DHCCH4k9S-Wi)QjiP!^Dk{IkD72hygj7>mNf<ge+tF
z#k^R)$}YAP?!tPR??f5_oeWub<0%9b)|zR8{@UQ*>p2ERnRdYdy2hA%0@CQpWPwjk
zRfvMrT9DpCe?}<;Jv4FFH_4)ShEr>Ho;9-Ab^^z8fh_Ug8!-;8k0jKkb|9PU2jP}l
z7Saxr0onYnMwSO5BP5};C?_jo=P-sUj1btUh^>{CX*o2SQ&eNLOA{oGRW>g>HKH&F
z93<fWs~r}z{xzUp>vgimV6A2I!?ivd<7&u47z_or2kySk=y3uJU+ev)IM1ofv1|x1
z5jC`qgba1xRDvesG#xLIjW$qbN{wtv&@iHhT)jy)r(32a57zQz87z$$yuc>2<8b)r
zfbO@(GuRMW>kH)t<zb^<B#*&b$6Eb=a2^N&E+^XpP~1AK-N@)b&1RGBF&3m1idf9)
zXpY?&4D&Sd_;&=Bu;XAJY6$?Az*ZhfTZzzm)Iz~jJ7Yk|*DHV%Hc0uS4*;X!j_W{!
z8-ORfR~}76YitDMJe9(6<2+_~>(jfBJZ}G4HKkad2tpbMXY`YC{&94kvj0X^*6H%J
zgTVWU$#q+h5BH4yj?SS2<yrsTs&qBrU~QG>;xLdSl*#iBmz!&F!!TNdu}UxahOtov
zyR+E!$9g$xPL@9f@*qY~B6%^m1OzLX%nIws?Ef4Ps>hS<Mi$CT@gYLyDNVi{h*Trz
zJWXKzORRX}c=>DG!@3`2&ohDUmO#E&{IX#~o(1x^Sl{3*d!)V^cq~I(uD_>EfHw#D
zVyYFT&86~Mz{L6;$e(%LVS;hV?bTc4jd%vH>@3J7?ELv=&_z*JYvdqW<R3AKmgRsU
z2{Bp5;ofpU!J9zbZ~Jh(`_}p!@A$W-43URvs5o}xpN5dq8JOYSK+e!J<h{7ZP!kge
z0)of;hJeakb~>W4<byOMgB>^|6=)<B$cGL9cpce3{#_v-*)zz;Os+%89t`+#tP!GR
zA^IDF`Tr|sMQjz^CSX9H7@lEKug=J)2?gt52B92BtHx3UOMez0jzq8nS|pXpzhg0H
zkCV>>!+?@8>D2MQ@Z-_9uY76$nV&yOzA{uc*Kdc0C|{>FF6P;$@52%LO}vMeik0#o
zAD_Kj8~lxL{aao|t^8@&p#S>Y+#ptqLG=OoE}$*Q$oD}r(D)z~iF7cdT5KMz{pbV$
z{d1QNW{#t<gKigk?~uFTn#Gx1DZBcY<jxwvM^lqZ3%mK35e31Hf?CG_zte2Idj~bZ
zk)z}6@6kc61*M3~+p~k(d|(a5Dg04~ooW`l9gK>q(q0|ZXflRjEPHp@xgSktl-9Ls
z7H*h*Y*e-Zg$-g~8!8_)f=XmRe?3H=)!3r_V+bw=pv;C*^niE|Oq&VCj5htTftR^7
zBL{YH1K0#IaXH8*tSClUu^jB*dhH*T*bi|;kbolrm{ts4X0)~ZB));+omPIt;>w{N
z+!B_nsjblFs6XwX>hV>_%VGXLx$`C*=2>8kKl5im1+AtK0_Mr#0Vt$bc<_!$uka#t
ztC#_gz8iFr9_f2&a=+R$01oTuw25y`AB8h#mbCR}%sOMf>3Uc&+W83lmZ)RtBQr*}
z_h0OSo(HLt))#oGrGxD#lFMqOqmN084Sz?txpcDM>ma^K=LCl}*I|fp7yF$xrKIZ*
zR$xp>S0mjLw5%wqW@kAnji#ni(DoeZo;HhqGKCQTa<eScut)mF>nHD+DMuTP4nYhe
z0x&A%7~85-yUVfhwr8y{G48lT_lyfV-YB%WzE{5i(lfRLrh*FT6*MrUvK;z8YHRe4
z;Xsp!1A!@Xmh_2l5S3afedC)!@P%Zc$4fu^-Ku9u|Co;ZJ)nc@z{r4uXBd4(m@YH0
z!%n79BXASWq{_)Pa>5Ugs!<?72D98w40zP-YA*Pb?4(T1WaZ=^?749WWbVMDaf<%}
z{6g4Z)jK*5X`ebZO~&#^zJqG%IV~1CsmG)qa=QJMDXt_Z9@N2Y7dd707H2%zp$nN1
zL1Qw+hqv$~$OR0v27BKiLle7dDP6$qR91pVIYP;cDGD(K9PS!kIqU~3BwysVFl~Lf
z4`(-sSF?qSfR1_Q^Nw)fFa<Q8Mp#M~$w*rh{+bIZ`V4==;t5PeJTs1TH9~%iWK^sW
zQG;9Y*(`WLMjP>s86{%^VFf5IFJt}PVDS(HWZVzHXD9~mdot&<j0j*E@s*Ab!m6z(
z%S>a@jC}vt+@7KIFqm<IecW6>9$AgNhR^o7xt&nVRd_!q;hS-4X7}d5EcXSUj!d*e
z1>x2LG0zQHu}aMZN0^Dpr{|^7t&<pAjhx@IS|~sZfbvapfj{F4(Vq{y-XxP8<#UjB
zvOzA4O#?nnPh&n9m>K@1igZyd*<I-rrYsZmVxxK#*pW+o8neAxcyY!D);s{97?L1>
zN^!hWVEc+vMw201NMS$<vS7<yh{#f8kme(B56M*71PN2%=E1;A!xn=NC#RKLL>aSQ
zi|t$Z=ooiZ5<H+HTP~$Oka>MAkk-(8X^;<V?8y|{4ur#?#R=}^GUK|L0?7E~J|%i>
zs8-4{rxiY~onTUjS!7m8Wh(%e<N9LZ)T;Dsl{)>_Lf*P8o<S9%WWjcj)#(r-E)u$o
zdggNh=m;1Eq{aa2bpnFiY8?n*CkaS~rKt(Lja2k`(O_HXU7HpQh36g;p-c~AMs+r1
zqs)~$L*FoVZkHJjWK?OT&VRW-LrEuGg8X-mToH?*FUBquxiX!AWEBJRfSUre2_J`C
zWkdmUYz+>9Tx|<ueLuOz{u>VOtXvyB0bU_U*fTr0TMggg^wvVaz0MFc8mbP6>lY4+
z?F_YqIdZ+dq0K<&$S(t-$i}RY8xjJhTU)Rj(*V5Q8$f#~D`)hdpDmVO8*Fg)*2r(t
z2GAx;1{2NbZi+d|ses)ae<2;TTyBYDsLDEJSk|*i9m3_-K&GmEZYYZpGQ|jqxZS|%
z+p~|{VcS()PW#Dk<4e#ZY@up)yEBf%7^FnsWq;JbzLb@_{cWm|Bo8{}?uijpZ>1+<
zrrhh#_=XnQh5Pn*!7q8Wvi^nK7xUy570LYpZcSxTr95B*uqx+N&6B}CKA2br3qH`-
zLVwqCm|8S5Q+{ttys+~z@{s-KbhH4fcTl*wvjH-r-)GsxxU(P;&kky#hbt#@{C&!+
z8a4=?I5%K}T8N;vmJw5$-2Ak-3hkv+Bn|!+4>}F!OwMSGXSmo|-HBX%L41kYWQa>C
zM}t@=5xYjHW;550k?)JrFIwpqt}14oZGp*aaiFpYGU1Y#rrB)W^x`dzY2b*gg4wP{
zmIdd)tQAF*5z3FFu-rd231I=az$#f`$gD1%AuH3!+5sHG`chdH+roqfFkUDJ_?Kn|
z+9+1Z>Huvjiu$eb0YJP^{OJLnEo%+CTCuH8*4b=5P+xAnog0K3mdgfTI!;Vjd@u{k
z`Zn1IJ^S>Rje#Gv@bYjm`I~HVQ}6)fx)&lM2%e42@iGNBpwVjA&W6msC9M({B7mfP
zwzUN@BtC4%e0ez76)a3F+(FwZt|s^+jdA|LXCT8{vK>j*xOn;Bi)EYP1Wl`9S=^qc
zTcol8Ya%oUQ6gXl-n1RTg5i%~u*f5Z8CWLkWM}N`)C|nn)DZt@2qhB-(LCB==YbkR
zK%tr4^HfmF$zw4s&Mg%rDs_(=XG6?%Wi7<(oIK%6Ft5uzB`90mlK~-vaI54g`!k38
zhtAih9R%`igqNkMAX5O<2sV^w<4lbhbWWZ;ml#~?2W^H80JZ@t&GRuHsv9hj7yK`T
zu0crrQ-YBO3hDsVTWR%rF;FuaUIC=Wp@f@5qGt~?u%YZr0Ug}2AX*rN=E%!NRC(aU
zf3cMnKpU$roi2Y3az<ZDc_s7()Qrg?$sHRCM)fxb$h1aFLA1U%ug0P;Wd)_b`@|$s
zMpre0P+zl&`wr+Yulvu9_<?ryxeyE92tvp^cU<0(3G${9LUpznwyGeCv-9LF|CT41
zWqepN=4}Vd&b9}(GIUAai8VpXl7ISZ$SuPW1>2<HrykyS(=QhB$|9&#8)nISaU&uO
zK;z%v|G`|A&H)|>)A0xX3(?Y`mlzdEI2PnX|3!r@POMVq3NWva3@J5m=zbiV6&^g?
zHvjTxwBo@vsP_I(oS--<*e?d7{l*GHufhrZG?qLVNq3*6LCKXEQm|P59rQg8j4vyn
zJ8-mr<Jy5{%vQ2chU5!Fd#?cyuf9xZ2Z3QeR2Y@7{29y`YC~c%g|9o9r;c$8jL4m%
zUIE6-aq&#%nv0RYaWW_KZ<7B6kXQxHwu~qL*5+oJB>8UuT%|>WzKhqvBuPC8AeZlL
zKy+WM1ieT{GZt|fVPe~*qne<~Sv|8(cJ1gkm%<V>JY?Pf-8!oAsTt*2O4ti%_l~YG
z!(__ZSIB+<*Vw~Gfl~w7k+NsUoeX=KT#g*kjyqWpyB0cBne65Bo1@Y?!0er1qUJ@T
z$sF0opSj^1y2={ax1*}$&Gkdd5by~j=6)T+US}zejCwZO+~0p`Db>FSoN*E8=75en
zcUR6hXqx=EqgnMx@6o*}I)Ad!z=7#^syOh5X;gkoHo-xG2nv~q^)qew;EpY#r$@aW
zVv}k!-k$^p)SyoNYpWb;xT~xIx^wc=^ahLdQs-)r!wiWzbVx%LHhZC&$tM(U8IpRq
zZ4%w`&0BGg9ub=c?vJ<R=RT+g3n?Wv%8?!2hF4ULxCUuse^yo-M{65|fMy)3eM&nY
z1NSD53_H#?N&Aj!kQ+Q)Iv9w|Nr4H=H-P329!2v-vXNycd#$;CEJBMq2b3(k1o@<k
ze+J7J?WAD*+);>n($#<Q5_*(E{l@L)zl74jC?B)79hKl^A19MiyC*zsRb_DlI9ZP%
zEHq~;p>lAvzhoCXTXQzbG5(Bt^BOFkAw$Q;dpMpjljE>2l;a#6G9l|c&FTkR!O@O)
za%w<&9ZshQkfLlJ!3E9t^xyxd7YR&&ji|OUM|v5R4AJ5y(z|1@6P1a<PN;<-k$@}+
z+rt3ZxzZ=LxEg1*TKdMH`5Lg(FW`sN#Qwg7vcEs$Ib^WUYM7-5#AlhJOKXc{pzV2w
zqvV8GlIkcH`m8x0n4TidiGd82kW}eZCcjut3U*OYJ7b2N?32>;5yc=*iAnQNtG-T7
zO>ogZY|QesbU=B4Ex?TuG~Fch8=vmK)AY|qUzR3}D*-qH2b4j!@m3^}X6(U1k2RI>
ztE0C<AP9ppq7(J;WM~lZNQm8ev1izAs^_$rgk=T$BEuX3t@qTdhQovCSV@+d5F>(b
zAQ&WPapoiA9fYT-h;@d;L&qfg<xx$LGvk~>mWRLtL2Oh&hyY|6olpxJ!nUjhrH%1t
zT7MRMybcF!tgl36FRTOwi+&)Djf<6FtC`Y>WE8wb`{!BlC2n*ohzw3MO2nza$m0V*
zjdY^oo}b2o9s{~VaR-dZtk?<8PHN@sKs~N;x_cpF%Q>;IUVTrHi3t}J8X93inJwq~
zGp4pP0*&=$&-2MBTr387@T_A>xP0aO*llD~v4rCVfo@oUOAwJGlZ;5Q8pOuPg+5T2
zv{v}ly<9G8MQ0{*kQ(Xx(<U!+v4O=Pjv9F2GAz7xiQNf9sFq|}E)v+Uv(;7L@;NDp
zt7_aiQs@g~@mkvb)&m<w2?6Za!<^|YlC?v_(@m*#a(cyZ1jC{f`!n_wWe@<2zfuyX
zvAh=bW=oy5w7wUkPA)Cgs-()LhP|P19<x`3O!09k0kMx3GttT%Wg1%u+%wwHm|1yj
z<rIW8R|EyQC7P`tY;q7fr2R@CdOkFH#0_it-Kw+<I+D#+beTyV{d@M3>O^9#d)dzh
z+HygTWTZ5#N=?8r4%Xz_1oCX0Ws(pb%?zWPY70g`<17O9s%4tZ*7pSDkvcFDfWvM*
zbv_VT5!1&}hlq>Xah=QvasgRc{cYY@nu$WsqFkPqt0pn6s&Pd?&EoGG9eJg{MHbEA
zI95ox%2raInF9E(wzG8F#d3{7N;k%}=_{)>!G=~d6ZDi|>2GS1>(YC(a5&K~6rAQ4
z2{p{oremDKJhP#$Prq0vNA|4-w}-y`%Q$0sNT5TF#~XsEN9M_mX%1`hKtI;6{8<F$
z!=@m=wpAgJ79zrL`~~tJHgQJ2f!q}2n_q)dd$UbhsbUP6`YmZA=sLGzU8XDE>c6P@
z)%Ti8@oc#*(2WQk4O5ZZ6R?`5P79^nVR#)dV1WEKK*&{P3gyoDys#1(x@38GrO#=0
zXEe|Xx!V9`o-ulA=j0w6u1~MNa&HWfLq`=DyHUT*c5`U0+!s@gM-M!?KZtkam?82&
zEXakA)f83Kp!=YG1eT7<@A7+_IUj8^i`cH}A%_mv7eboxFW7c{LNxaj7JP)d%<^Xt
zULb3mh%q}I-j$$mQ#4y!R^|i+z}5yomCW@qW<p5u|C4#adEwqfa&K1V8!E=3Ykfm3
z7x!C%G{#?uZ>W<6_7}QRF_36sg4psODFQ6=XSgY;Z-K$gmBleej`62Uj3S}MpgNoV
z0?nTNbe5&nS`RK3<yan4$*2s=ov*N6nK(*T#u0$-1vfmsX*iTk?0FhH1Z%vk4s1bW
zM4XxjuBzHg)1It}jfW8wf&omAYYhrSbd;pbsC5n*856~h6%ee~`?Cd*4d_7M2my72
zKeOtJb{{r{oKXbRzJ<DOj7NjYW|U;Lmi;DMCnD4kLr^W7eISZJJlZfu%N9cxwtOy_
z!&YCv#~I;<vI6f5<l$Hn%7|vrKm*+o5)zm)z|mL%sijr$HXjR%01uAHCFTchPYd6o
zA?qMSX>N60v6>Hq{U{&Oz_mH42|Hrr=mXBkBZ&tmYdio!Pfwf>R*c&juUD|fto$)B
zgeob>WaI_@sF5g=kma!;C<JR%%j0QvEHsTupioE<pFr1-`p=&5-}N%UvsA>-kLr`Y
zMn>sDp2C4$Cr<@9xoOO@e!AmMtukTSDiKH*9eZqY;f!TmrW}~PO`b_8V}TQkS9vyw
z)#O^uO0txf=UU#zU0}T9`8dJZaul1-$_xIEDhI$9a|Hh6BhvmzyA~7HR3MiZgZu}h
zB-@|U&N7w2YP#AaFBv-f_3SS%J7F;%Y_9wzmOcTFs5<#;40kEW@|84POYL}CQyS%O
zanD?GiM*QL(yHa)z4W+khE3-0u`-nXEU26=uld&4CBap!!kdW8!&>N8eBFSoMQ^D$
z(k{`BNzz#=Zzkkfu~Qaz`$vF}SOW-kZ}}^XBw~;*TII>x0Yz1&49+EaCq6-!mm<__
zGageoUEYl!kj0J&i}!*kM`tQfR0_;o-jA~eIz=I4s4yP{BD$R{AKK)&Qc3`go&Aqm
zU{;lzrC&bwXRKmw9(5F7^Jm83Em`D~^sU+LP|{95wZ+h!T*f-fpT%i#?JHDqq=_Oi
z0Bsa}4C<TYo0}1C^?ApgRzL*unUx??iRTN$etGG5`O^Ngu&PwP3SgncQV(EnhOd3c
zsZ^*d9rB8i**E@+NxyWYAZH1C(J223<RU;0m&vy_!whs#|8JtlxfuI9pASU|8Ho!6
zGJ2f09BH-cQfTyu1?=!Dkxn~VzMl#Jot3+EGQ$;m*Wt)ggu*%3fp+a=CMmlZ{@)gn
zyLB@Ao%xAKpi5=<PHxf~Hd^-Rw3EiRVW|IBBYVcns(mvKsZoB^NzHr3*_u49O;jU$
z8EjKat7}<_Xm1-KSAvu@*~h`5TLq25DTykdp-G2<<@@>!FzmnsL{2bU_VdNDnIF9(
zC?@8oXMdZ!xqfhUW#yEzx(YeKK!;T;D?bjTz}mV7Y%{6sa$qO7s3GHyrCwnTILKdC
z@e`LFEEkxH=j~vFzqx({)TToM13;A4N`Q3z6Mu#@SXa<i%AuXyv{;B**A&-&>f;-=
zv-mt+wkYW;hxzA32~AL7sPeN;YC%;(S|NuAxS-yO9E=M&Il^8*JJ9ON$?|jitFpWf
zc5^whlbcs3WVBmk8=C-$!iab*EYrxIwmvC2s^;s#I@vA&(a;PFz-aH!R9)7cS}Pq4
z$0%XiM>+;_hC{sR6dR?;Vc5=mtOvdr>71tHxn=`~rP9S#HL^nQPt`8FrUAk?z;s}C
zT<I3*!{vpjINYdal36211=R5AQG#Rz|L%dE3g(&iu)pvXqs5Q*_o;$2Vh!#dIVM)g
zKt)jEMyQC#c2YYedNUr^Nlnk0l1e!~mfNduAL*G!V&zTqZuE+gAkT6$SHUsTJ7{4_
zwhkRUi=>Z@q6J5!Z_w}%M4-U0Rwn&ok`AYwBK>1ChpMJ80}QOpWil{WKou18teoI8
zQ&Luw0oHJ0K&WY-j2LZ@lN^)8E7P1=(Z=@r<W6P)v{n<Ar()!4J}4`0dWub(DJy2U
zmYnKSaG<D}dRm&6mQD6$TPLU6Cge&os3s|cVuKI_las;zKGX!YdTxNCKg6biY7FxQ
zawq*2@&FnZ+0rO4@YH)iPZ<_3z*yk9Dl$CafMEhwA;@u?WrRO7(;>7yD4QeGK)rg)
z8FAQ9w4*G8CgeqOrfpI3PSnY$cnKZiQ3-<wd~_NZiQ|}z`66QyuGMsd`jyscW}b|-
zDaTGmF)JBoe+(Hva@+(t%g4bz07_@n+g2GL=*muqzLTGp#d^mo=E?+n138SPh3GFD
z?@wUw?sJ0v)T@B1{u6CIaOJ`-REjugIkyE+m}9tOBxra~<UE57YE<cDIX}(Jz&N^K
zSIY(QOszBxh+$nVlVZjqG>pL|AQw8@H&YH}8)ny$7x3RDa*^#e9Dm^4b#if#HbTjf
z@qI~P8%})JVKR0m`=Q=NLA=Z!Uo}$b@GyFvS?a8MQ{>B|i;yPt)%eA3@2tOVG;dKm
zll_^U0G^XZDNgULgVB>KUneCtF%BkTU~*DwY!PxwMlSXBa&=S^*{sa7^x%hGEmI8q
z!pX%_W}hL6s!qx~xtLN}ptZut6u}TG9Q?4ZZUy$tiBTDA9}c@yR;v62QwroAX~_bN
zBc|sAlBcGmTxM&5W33jwo(fA@qPp6ruuCYeKL&2)eC70eQmJBaY)!((65Jlt4c1F-
z@G;FnqGx)ltqgtLYGj&!L712`%{YPSt&pVQqToc@#sI%J;bN>yGqa0~Sy*SJXKO&H
z5cr@rUT!QBW)+Y!&~-&Xf~|lh1R}_lvHzNc1Dh0d8QcG>d_3cGj0IUCR|oZuhOEfb
zaZM|sg{GDTJ7xkzOYNhF<JvSlBe&?NAnRwm=gb(26{vv-xz2C`VPc5<!j2Yh@ac7O
zeG4+w<3deOolcmKdg?DbxsMb%$t@M|ZgB8}q?t~5qoJd`+?V((o2I#bY8Hjmd>ZcN
zI#39!N3`bdzcJ2%PnY8!y(umq6kDnVBfL4V4~A+t^AfqGlg9|d5M9E+5GFL=8V}L)
zF@v6u8UcNq5q+P2z2x?QX?W#Sxg+*#Q1ukZH;DE8t<8&bRiQa@ccx{cx(eV$jEvmn
zqxI<}cgHC}eGAaqJ^nUn4R%aeKsU*~zV6oAWS9ZSA$}KIsP$%u9bOpZ;rH3#5Gc_`
zHYfMn;Pgtv$zl%L2mE6rBAO`~ytEWz`hx*k6;g4KXBjI+6dxN`<>aB*4|ix<cA6R~
zsA))je1WtYXsg(RSw5LTuOcc)gl?1Bfz-Y|`^ubzgsp1Lo&_43YqG;dN69?@Ri!gX
znW$YM^W)MOO`)kFC>Pd8l_DGclX;MvtUStcSrGIw3|a`T^@V{&7*QEu%hjm0S{B7R
zawuL5J1Xj>`MYMUVU#)p>&S7I#MD}G3igksJ~eM}vNA4qnL&%xBM!FQ-|({s_JN2q
zBapEO3|bj*!uU#m_bS5_it#~kS{>h#>0osFl-I;D%~Z--|J~TB1#+zmT%pTsv8<2#
zsS5E<CE5&hLu~xSOdaNE@-afoS_50}#+YM5zHIW}X>?~)^fHL!n-dz~;nB)w1~szu
zph9l_0nO2-!<0}MN#x;xZ6f+G${&1W3u{4z6c})}+027+re%8&Aj0fm_yf_gC-Gc}
zFb#-)nk&sgRK0Q4?ub8n_3JN>#5{=4$jQ#M0{Z9e()^=8Q=H-0&PEjPqe;jl69!Jf
z#?4|y&pzVhwYtJ%{v}ON^napcp*)^uU<#>TNAiS^W*)TAaT)7N;$sdb3F=9s`-!9F
zDWi!g2qJtsog>YC2jV?XOrGiFv2K)JwA~#%xnOF~#)5~Gp_<2YLD5zv0$l6)MA4B&
zsLjo=l`q&<A|%)O<DY_Ja796>V<hE^K{22x7^42{`+~-FxXQIOiJICNKTq@6OKCDq
zDua<OjB79ZFTkBHiGdBI-6irDn*tHoOxu&c29kj)lpSH<n!J)eZotOiGYmnrE|k9+
z>iYHTC$9#mXQ%<~@Ak`}p+n_0LlJgyDJ(ay+h4eZ8PD^EPe)j2R%w77^rlb8$Uw?T
zjOsAj`G?Uc`s-xmEt_u;vUPwzdD~H=!9IX)QPghXr&z-~ad5_!RAws>ZTC+H5D}p7
z1_e(5j^A^Ls0A>Ln$FDk{h5|V1pTZ9ulXR(2;QT(qZ$UT2$o%eD?>gEq?uVn_1^iK
z%=J;ASc_S}t_NNC<9Hi<j2`i|swMO0{#RgQ+d8mVHUK3~mZKcYC+QoSLCx){aS8Hi
z{DRAhxWiTfLK>Vu^HoDP`L}}+3Km<qejd~e!(O3$kpOLgjYSJ+$(OOsu$ci{U->Hx
zTcUi!B>u01u`;|ClF~Ou*C1-Rq)T%pxUc>b7|%ya!?&?Drt_n;nS+m800gX^RF&Qq
z$#+gl2poW{{e7C8u6+x+l+J2-f<Fu$*w8<>T6XDd#w<lN-a<sJ?AlpPDUi6bWmU3U
zXEkl%h*ik$_TR|dUnhHXb^{s`R_KFdW}0Oj?P+gyxS&T5`H_t^Jgc2)_e!%-&enTO
z_O>^Go-JUs`}i9?L1gO;P~5k(8q^T3G;Emcm!?fs3-B7{z5V@DivUrPRSt+xN5B^M
zWB;^uUw7QK;5G;Pr(k}#`WrOl^Pu>29K!Vv?i@D3mMSQ=WaXoBNM|zy;G|L;1Z6RF
zip9ZN-!nfk^f%X^UBOzv(DDxTi7Ab;_vsvDTl_R8o(}u+Vetv%!BQlm=HzFF-JvYF
zcz9>Evvfb{C^^C*<VbXKv-%oT6B=C+m9SQR?lU3i5i`-!yGV}oZ(3KBvuH7XzgXJ%
zH}jiULq5a#hF!RAK&<&8FbUzbx9jYtyXtH~Em9Aqy}{UA4+*D(!Ppm0g^vC$%Pb*A
z3ioG|bP8evQ7x6uw)Pxzi=<0u7vNEQoN0%4r*yTk5vrDvZuT+j;lt^CRLc?>?hIzI
zLb?b3m}LN?Lr!|w6zB(mWagv&6T44Q86G_NnDmTofXJ2B$+7;7oR)4LI5*|EM2R#X
zfnXs38#_Hd{bG@KR2kGAD?MXxt8pLpGR#y#6Yd=tgb4*9;5q3NaKkOlHVAdn*9Rrb
zXO1G`+MV$GIdC*TFlb{VSXqC2#zYS5?~2=afWOy(fDVQsu@lTdf2JBKMx_=01cR%v
zAR{OG;iOSup`2vUtJ)$br&-yu2g?VeJ|z(p(ZuV4k@?KmrcAp|P7Mfn!f?)LxIN9D
zF-I>~g`$j%K|I~|xS$lKzd@bVHfxr0=&FMQRS0=kA#X?mgvl|iX|h&^+D8S5gj7*H
z&z5uoyuZV2NjX;JtdZgVF%B*+ka-a6H^_)s1;l}HjY7;rsF4=!kdgj7(;OS^8NRdx
zPSV^No!wSm1Ur{0Eoa(-N-IzcdQ?z+Nfyf)?b}KO(X4E3Ls{5sjICipWj+GnaV3rQ
zH(d$aHkYO>NS_O2TpEMIN9%OHvl4DdoeUP3E#ng*C1nFx+Xy2?eh^(Xg$U;CPKeDJ
z2{M{mTEt=qXZx6pW7~+?(4SGxaXM+(#?I&yV+QzI%`eBGFUpnYCUnC@glD+R<vf38
z5g;6ItDNuS)k8j5CKq&eg&CO^aHz2!SdmN$U}0orW=T#i^nFqN38-WQJ^l@GSuXO4
zOmo67Fl1hAgwen6K)J-BZb?0)(ngnKCO)L^7>5py3@PxpX$s&<Hscp7@mv^ht8Bw*
z7D-WH#Q6tJ33L;q52(M*0C&dW(cnBeMrfu;Nj;=Niv1ZRbr2m5ZEK_?J!8%kugpbK
z8sDlyNhxu*Tx#cb1TxzzWQxCF^&2P)E|oGP&q1|F+%69mH4JiKMQ6MBs5@|#Y=poY
z%^eC->FY8}9c3xrH!LAomA>S`u*&nQZ2_kU#*)ht7>t4kmcR<CW9nXgdrQuCrt2w_
zq$bvcwhU}fu?)9lZESz0i0(<m#80&k3SbqS76T1OW47sOUqOL@gjCBhmPwtRFhtd(
zUJ01N44(&FoK7~Zra>+@)F9<&x?B-(Kvt(ot_e58l{UuqTX56f0{QN$7?YJ5L6le9
z_oHA7k!w1ugQTQDt_|q%jwY_jOnYza+33V{oy~({2q*%R$srG-NPf}T-IuWKqlCou
z3Go`fxr4pzA^3oM|1vFtA&GjZ<OY96fj|ilfZiB@A`gw_q<$3>Y@fMp4U#8=i4|1V
zA^MzkG=CFs<c%7cKS*veh=+|IJ6dkGm0j3FZV4Pg&LFKIz)^1XXV&(FwKNyVZ86x8
zv12BTm)p}3uR+CpgxRo%!5xMrNCxW>VuOF>w@y`73QprKLHCOxoiswaUMP1uT6hU+
z7@cxGFlV_d<|~9#H79o)u}r#hlHB9G25te&aBo5%uP~meh4Q=jHXlNJR_;s7fe?iy
zHsFZeAJ{|`18wRLw2=RfnIvyY$MD2xP$~d@FfGKm6xpGhyP3}?zi+uDAPqcZsIEaw
zq0}b;c~wwlT_m#%{S{MXw*SiDVR)&ap*F~zI1Lwco+NW)U^vgpWL|)c;)9j&bIkY8
z)dNLw9)!|hgLmsFjfv>g*Nf|qQ9cZrT9BqiPzQj5<SPqfEcMcwfUqbqSqZaGRu=m!
z#&V5WE%9d-u5B*)rT&aj0@Qh^&M)iiHV0M&hZlqbA#D!J({Z0e<_R)noz9I=I#<ME
z5!}rbgO#yIxVvd2V@Gd@cGxd1sxGVIeY9181xZ%N-w;NNaU|CSr0~3h4%Aby)}#~2
zGM>q|&Q?4TJv^A`x!&jwEu3US?8F(hV9QPR18VchM%(SuA{2n%<S-iTXvie!#0*9h
z(KkB`^l7nJIewEZX`A?-x;;R9$jOkcHgaJln9;-jCH>UAJeU;whj^w&Bq-95X11jx
zWjY%zHO!e?CfkFFO=LSYHZN&T7%?@7jURRdBY`7#3N`1fJQ7<3px_<YaUJw|wcN|j
zKrya#ZqXl&_C{9AqX{(JFSNI9TPctEGe|@#PHx7*c|3rE<p_;XAU8eX=bH*<qdaMc
z0kRjudY<y1SXiS$o=#vPOG(v}D8%zjizWg&h(0C`{H(uk5n$#4vY^)IY}2a|Ne;r4
z=L2?kakClW3;qleAFLwQu=q~_0A(d?nbsF3amtHnDkec_;1vGdh<o(UJ`Up}U_*z5
zvQb`&XR4E^yeKb+*o;nGI4U{$i@k)HMr-SY$MV;BsTfxRm-vd$3kIfcKv*(-^Z({^
zSXeof5~$Iy29z3q35^G6v<RcWr=w=Ym*z5A7p2yy7;J*b{aOG6RS!|6=!9J)ucsMU
zVVSbSDtW`7u@3<)pD@qJn?5W1H$pCfzD5G2Uiyz9J><w`PzN4-|E)N5u(VXm+eR#T
zpj}+5T<|+-QVdOH0>0UQ8d2ghuBl^j+joPXPN*!9_tFRSKN*zo+cAT<RV^Rbf8&nL
z$%g??0px2`j>CK(8G+Xo$;b9$Wey_tzy1f2NUBwlWwm^grn8>D%**;T)-<vPTJvXa
zico=@H%~LfFR*InJ3`~bIM_t_cN#|pJ0>m3=l)CuIDE+%^e^H-PAGvcRU}^qWZ*#%
z{?$rjBYx#uYz3!mgDB3@8p^JKUk3@`#v%_#;2C~F^+esQRq{<hL^B{x-!(9H{wF<S
zim(<PkZ%q9?EHqXFb9Wvxs_?6MQ;O`;CKEh7$QJ0P2+oin+`5=9JQXSULsx84A!i@
zq_mlJvFy@CjVp+!kX^fYmZP#NDrtJBLx0$<i-jFAGj*Qq-o-6d6DliZ5Bq5tEDgAp
z_OuzmqIupK0=`Oq)W!80#7<X(-4(EY;$B_cn8GFu<sjL;4GxrulYJbHxB<*t&6xdt
zeR#zAsPn1`d%T|k&|JU$dQi#!0o(9#`Er2$kwda$k^I>HfWriC?E|~G@>)e3J4iGq
z2gNv(8R30!j5B7$C^^Id3jCAUH1J7&66@8Xn<Rk^a%c>+{a!ese;R+ysDyLjus|`i
zB2-s^KEV7rB<RY|d?fQ!(ImJ?4v)9V)0KeSH7iHhs7fr$<mZ7FPyws-tOv(DvP<YR
zmZHoeWEPxF{S76?{2SW9I4YBX+a|2Qp_oOYKu+4F1u{aFav19FJ6H@jrbs#$q(|4#
zR_V}1tr*SqXP{{zBF;M6-`j5mL%jv4>Eu9Bkz~|?)&vLYY+1{svyV(fq8}I37sOOj
z>X$BE)IJNJc$IVw+JORJNA}XqHVZ+lK=D!btJ#Bahe>y%tbBA7>5;}Kuc1)K&~fkv
zIoj5#M+1=8(iU}0pufdmsn^kbIo5xnl4rq9yhEX*$Z<jJh^f(>A3|`vBN0|h2(y&7
z7fDZl#<KG)ElU}+SK2eWnbqS3@MvE(n?~<IFK#!)j!K^((q8@gN?%*%DIJfMez9o~
zdRR9e+a~?fvvoK+W`7va#kH-;)mfa7f$0_Ngbry-PVi?mrSei_jlvTh%+~!_E$l%H
zG0RC^TxP|A$KgtC`A!bp&@+Y*DzuL+mQ!L&z;qZu!zNN{W|C85wG2%Jhd9kh0GS;q
zzn?&5trJuK2E}SJ<%leh!GRn^JIWA$pC_1tGE?V=2DM|WkdI4GYMvh}%HwlnSd7rS
ze@_{n799=zXbI7>dm!V5GNOw*9h&PAwr{z;BLhaYr*Ozsa)wWUG$-8LXZjn~e#$(^
zqq?Z}d}>D-9n%jPG)Bgx!FXd^e#O{$1`&!mn$)Pq1-m)36h+9+GL$vfpEFv<2Y@xW
z;t-yemHgO`aoKA1yQnS91{kjOo8|1-9f%0gN8N<KbNm;(fZ=0;;SgX!KAISC_d-C&
zxd}!Kb%HxX&P&f&N75<~pP!yt3tFb(Utj~F3RF{tOftM2f3#ef4xg3HY=mRR>~0su
z65(<}@D&5bj3%N5;w3(dMNGpPi9?r(XD}jBZeTz}L7Kv?YdC^xi<ZK`?+^s;NRj<j
zXli9P)&P!1Ql?TpIi9Id+^7XciUaS6A_A{L$~)Aq5~nY6ZzkC>?H1=&*(4)nOZ|6x
zP#F-R9q(NNWi^sZ1F+HXA;=V;){^iUkiACAd<3dBaOGj~2zJxaC<HB$ayxJ1ASi1z
zYQ_KTFfL#~!~n}m+Zg0?fw@)rRMwe=rV!NSZ?%{omf)PH&Yx=khItnas9G}z;(X4g
zFD*m|FxFhCiOsBH?O)bFtBobH@3c$}+CH;Zrule89xX)6Af-&V-z$-UAazDOB~xUE
zzZqd#MBU}-85j?~UI}Z$75<EtLukFYgHVm%pHXt;<}jc2Dn}Bvdb&TBz|6;(x=nI*
zjKDkr&=pI`UlY%$*x|i3;IP;F2=i1Vz`s-)R9u^BZ^OGmcLA$+SIc#V1NI#Q4u0X|
znKO|8Zj|f&84W#9xp1+g!kFP!-Bx6}H`oR;MW8}%jO%nHGJ@$X{Z(K-S5jRI|7nr@
zI?Y2VK;<&|jg8S<KW6X{xyfIE_Ng}<ttB@br~rY~cZ>hUtADNBYMV0|iD7bEd<#7X
z*~Q?gE9CZoiM`wyA9qK3X_FrJwKS92jn}95K)KVA7zTpW4OH%mk>G2qtgbGRyW?6Z
zKnws1mE4oowgieAMWn5AZ_LJEJqSlM<^HbA&eM_b5SfJkfB(PS=VMcVr|F$_A?}Z>
z3r*YLFMS|LLd`S`KBbt2a1?nkZ4EoERbXp=Z<uVZA2LQBN<&jDf&2jCu1{m(&cpIF
zF$rS)5}9R)*nT^#-M1s`a&~Menup39|4fAyK$+TS$=n1h>wU<)guXc%-AMn({Pc{m
z5M1bTX-Ke{#gU1hjR70(Muz0l=Aoo43v5+`s*%aHF!8MxM^)<*ri*+O8h>C<z-`Mg
zclyc~`<#f@U;+&?+Be7&8yU6x%4DfwU?@c0a^_htGlCk0NtdTp(w{0T(u%CtF(Pec
zrHw?TMpnh|dVJ@pfoiosgC2%R9r}S+${ITbYR5wanf=vVUxdnnvd-V9t-!AMIkG<1
zjuG9xh&@d0IAw!>V}`SNG;&E(fDZW+IrbX^J>W`+k%X|q<cme%xSQf=XeAXCoZ1`|
zkjq{sThfx&V`{x{WNT~~j&yiFDW$KLhtn+313-E$Nr-=lU*NK5k|7T{#jxeFEv94f
zp6!7P)OjqD=9pYnQOdwgvLjv}pP43)BxE(I_P{WvopG;X;f$YGpXeX$MHTD|k<mz+
zeKdj8QuNlOn#U5pt?E9LAlwQMIemdV9x#l9cJoBbEk|P=Ee<>>IC@;)5_!_6v5XQk
z078O>_u{GGL6xxC;I4Z*7N1`U@Scgka4hG`vo<9}Y1kB>3%t;)ri{2ro{zy0mW(n=
z@`8h9iD%U8=y?L0!TVdGkN?TfJ^Ds&p9y#MiwTuhEp{`^&VP=1podxmtl9EX%X>Q*
z{lRU2*_X)JOZpL7Yz==gu#0L7aj^bs2aAPxUJ0bA^-t}gAWHf!{$|Sti6GryUX77m
z`QlF4ss6jahaP!={xw?-x?9xLvZm4N|2w{t<3_;xhCzftOdQxZ9Z3eV>TZ~rQN*g1
z%0B`D`pn$FZ*{To7Y#Jk6G82fhf&pMXsXviB6!<a^w6x8Y{Rj7#~8)1T1brlwEu=_
zV70s(#0Ggqlk=cQt2G*kM4Z07XD}e77$&wFG)lmY|9*mtax$HFOXPz<2)1{bd>G__
zQ&5pBl#haBd-d%lAN%Jm<pSr*zv9@i(G@gkSe$b5N$`f@a5pf~=F=EvTy24T7Ju~W
z*GK;CV^a9CB#L?Rc><GNB8*nPu%%yk`IX1Xm;SpNl(sjL|I1gwx5lv+o_y`E;8Y>}
zlbYp5`NoDo9on4yCr(Kbb=PlWLRGflJ_h^$Z%l$cgtP^6nJeGf2<S{yA>XGNG&qi0
zlt@=Kkzoh;%yK+;@n`eEUa>>U$an1;p-_wnV>kx$cay0X)WM5zx2|ejLz7$Cy{j45
zOkZL|1I&LoFRNvbt{#XsdhGZ-*|Vz~*M<VmKT0!Oa1EnR_UfuidsznNRV;jOAJ-y?
zX+45=4-3z1*{7?U*chIHFcR4})&_Zo0fwtF`F>r^Oaunu&R|}|2&7gX)+qb?HyR?K
z)Y&Kp#5j<SEE{}-{MesyVP;-vk^|#C*gB~!H$Y}P$e&RUU@;X$p~=Da4DEj@>4QPT
z<atPZ3!*|kD0u5e`H4TXIgH018sq6jVTAV6u4*gFAkIb(Gvp$y9?fm!XMqw_^QM|1
zhx_uO4MBgUbfjg#pq3J}IdX&#?-v??nIk_>V64F<NQU1eNBT3~&2&&P6}Qd*?0Ab?
zv~5?nKH`8P=br*byELe|0Tux-?bGDeA&*JS9gL#-_8%x6?LS9Agzsc4rbj_K`&V#Q
zq(J8`4hp@E6tSSD`~G$fP$uKvMP&+r(#>(ghzzqP(iRo}y+n>m6lY}s7}DQ>L$O}E
z$J7ut=;)=4*CS94gCg{poE+_+o5*4WYnSga_BqPns+&`<>SF^w)T*k{ZoP7x&qAg<
z*J=Iuc<mf4qOz7=x}Na}qrs8M+q0KptQ_GIz2lhWYBSO&!Oh}_8Yd`yySgcp=|f16
z>1XdW*CYMii=<0`|Bh~eHGl;F%7FL|Wi`-&Oa?kOans=@nva{#YN11-XVkEqkf?+;
z7&Nrhna7C+Z4GRiCnYjfi;3%>QTbqtoZNyE8V@u-<&;*>$d((R2cBxHE@hyhoaQeu
zb<!%?E|=5oxKN4)M;;V=Hw5};EiQq<fwf-!ddrXis#O4)S+6a}X{f=UDJzh?AQ=^p
z=i_*=P7#H=VO`y9hFyb>3;GzI-sM{LCd9ia1|lPTLY#cn)Sy=K=7lh#*`G4f7t{zZ
zFJ)YE;4}Oga?}{KgP@EiInyWu9=Jl7xuNu;vVVFFryi5iLOVJDN!EAR;K2dr%9y~M
zR$+tf3+9HgvG3Iy^eW@h*J^*qwbZ~Phb5mC{JseFL1et4lx31@B;R<u_UDD8a)L82
zSilxk$k{$C^(jPw<L*F2ntg!S9ynEUGBNfiH>Fh0_4gTb#R@S?!0yirD2uc3y41+|
zX-xPGKsYREa6y8c1V+?r9k3nO$)tc>wWkYX>qeDM$xbT;>$xb+xD*!*OF1o+i;W5{
zJZ6$y;se`liJevJA9pEZaM!{&R3Qa1%T)Mgq%i)<L#~+=1x2#{2}%n{)~9T#g1lKK
z`!kvvn61T7$zlVePoLgWVk0ut7BH6v%7)k0$)$##LH<9P($!;g;lY@{{T8spTOpd3
z#lllXvT_>|NS=eJmkOV&e_yF|P&BboxoCq_rDq$Ux{!J{%4M-G#&RNrLaL3UW6mv-
z^AGe41Ff6ztR^<T3g-s3!~#pbPC$&`RR4ft5V;Lan$rxZ3og7|rY8U_CSn~-(RD!>
zldC~(^JIph<(xsIQ1AEh0Km8x?1zix3InjYepsGd88}2s7P-n_rZ~vhc3MHNju)m@
z*2y*T$GAb`<=Qka<6Kcr-@q&^uafKHr-Ir#CdvIGa5NDi7QCpgHxQ2N*!?8=WnxCH
z>Iy^Qr_#g2d_x+n5+3QS+!#2pq)o7}dGf3DY#9y=gec_b|2p9rO+C<&(zmv6f4Rx2
znYIzVQ*|!d6K}Q!0vQz<xy8O14=4JqK7Ir?P~N;PjY8p>24AK%@l4$AzfkvpcNq#F
zV~No$Mec~34X#U2G0XJ))?c=~J@Of7ZMbn7<W2*o0R5S!$X)&#eReE1$3t(k-5msk
zD;o8|{5|51P_fDgulPlBudjin8sJ=pOuayU7aN0O#+s&mUxEe;0sXNw3WR68dVf&O
zghBE^SM_upbDTUF$4pfgWa8nru9M%#uxhr%h4ql{@efK|LUn+#WE=B_>unABY8sPS
zHr1dBLq^DKhgu!dSP!d8&xsZ1Dp6s4ZtU}9B&_6QUi^t`sklVu$GfnX)<}c>g>fM4
zNn`9P_J%hM*t0C~XT&hIG*(ky7+fQtiYjAM78!)K@bSyySR1{Fpjl?K5OMH=>NYNM
zlyXEkupsSH%p?UZ4QkNJLRzeOSzr!DdN*!c*in|p`qe>f2|_D;6>vL2Yh0P$piIx|
z7n@~OY$JA=C0~(X0h(c`@M<IKe#j?X6ML7dg`@;m8Z`bISsOqNC!^MgwRLd<$7H7>
zZ%Ed6^{{}+C<lX$S8;Dcy5hAguL|`>)2iP{Q#$5Wh*=9jwhH*zm=I3`75#_{WK&F>
zDS{@HmCdmcdD^575tv)-4OS{bXJX_~#uw=115da-Y?L~qO8yXpkSl|wWn27(#J_6U
z?sGFUi&2?S*`zt1sTUVbRp4;ik)F-bJ9i`4%p?9xGgh$i=sO`h{TXF%x*O<a{G&f(
zsx!RC^e4-s0qxLi0oA<6jI0I^%FmO><F|<@2?cfNi8z1xgXPKi4UtIM@&b7(@ZxS;
zn5BXMz(`qi2VElY?-uAb&o~^$Y1FhO&-&ibRLIa5wk<GOg+#kVp7U=g_^GuQ&z?67
z<}-}ASYC+v!ZrknndCoD{*>lp{apn6!AJ078gCiclhw@mv%_Fz8%)vBTs*LICW5@=
z-{_?PoX&x<|K-?4oR+onmpD5hNw_{x2Fz%E1jO)H(jkmEoXD;Go4r(vsv55bTEWLE
zQG)63aov>`mC9=^>wsti`UH)<?xWK5gP3=$m#rpI^GDtYE`%LIde)l(=$UYk$v^xv
zR<F?xx@uizK-gRH*{BRM(dBJ}t+2RM-ia+PQWXn)=bsJ`6<k`gkisbM24bN{u-Fj;
z`QA$h5BG)L8PLjS17P!uzHhK;W_C_KNN+Q6h1>7-vi!FBFxUe+M}lN|<~~Y0!&0QA
z`IYi<0?z9F@>=|toyDU&%O?p;?gp9cdigXa#`&Nz&1&<}^=#ht&tl>tc%IDh`tLXr
z9gmaGV@Sq1ka&U3F9I;wOGvh?u>WP6#+#(j>4i>_(7kXU&6BVE8|_1@c-kso$JjU+
z3@Bz2+&4asRTKxFl(^|bRHi5YiJcsm0dtjagS5aeOVP&rzkYfuHLm}=AhfY#^5pw~
z8ut+iZ7Wn$>86IJsg-55i1ym0TUffN9+>ZN*KTH{Qhgj>wG<xS&EF;)ih3ctcMIE5
z(!uZzwKii2dvsIlSD#)3WY09&5(pF~a{S1jfz~wwiMr8V-8>>CGmTN;dv{Y~7zKbD
z*{7SUimbO<&bA5rc5@5Y&};<+k14WWKnYoeoq@K3a`#VDLk=`Uh8*DQ#BSo$Sgx7b
z`hM(dGMhd9zG2p8INcUGu-nc<;RQVQ|N5UC<R4Q@M5!MLPpIXk8HAd3E(ZrFxYsB$
zLOPd23`wwU$WLN!WuSD20n?->tc0e5Lk&4NF%`_I`f1FD7Z-uZ82m`9748an^*703
z>3913A*zupLl<Op0gBPd&jMMxWfp;Sc;Hye%uq-<B49><7nN$}GW|S#ITsoX?H;S-
zNLxqq{TE1^Zf-hZm@6!`ZEc$nfQ86S`q|obQv)4>atoz>8XqnpK!z+s>0pEktu|XR
zLpu6qk&S5CNyxTFV4Z?mknMou(Aho%%Gcw1=#mCr3-OaCi8azSPy*+Lrfgx#$dhFw
zpKi8@{>b7ys+-&BAkr2?kcai7dr*yjZ5`ZrW*15i|And+Lj%onu@#rg(f$h=8hIdp
z=R&s1G5!lphJ2=_N$Eq9V-1S4aYr9#sALx$RIAO%@wVc#E2dOb!kN<35IJld#E)LF
zDcD?A+eQ+(^!D|d=Ezd?^xiYX$^w@A1foaKQ`pz$&7oa=zu3ZD9c)E4(%%PXsKyd)
zTee6B*yN}(UM2(mclA<pF|Y=n5KGtFTx}mf4ckSXnD&k7;c}7<G-!BUzMO3TJsJy^
zQv$0L0gNM(|AKT+jkUtYX=NZ*%W1w=2wpp8VwH^WBc@rfidxy}wj|cNqMRp#Vhp&r
zc;Pb1e{cYX0tN^&9uionHV!+~0kSDHO%ebx-agMKZ-AskVl%@7loBu;Jdt7krN@!0
zT+0(VONKibluw|m<FZ{TBb<<VTG6JmWfM^{S%rQqsih9#8O9`$yH$XaWVVN+k@Yg_
z2Xk7DSF=@(_Ft5Eu>#hGWjxSY8RO7Uv7_;AE7;ChpTAYxE&YWdk8DCa&POAwf`t<y
z&q$Po47^&-^3lk?&<h?j%#!g5?JOKiS$UJ>`v@kLVY`$QGy!iDS|G(Uv(jrha5+2h
z1{sR=N#G8Xa{?1NoQH}2scsM!Nz<+o=DEHgD;wiTz?L9#2=@j`o2AV}svo{&pDtz&
zGy8?1Sm(#7LHu)>T;TIqaHl1v+XXktL9~QIc57TJ7si!^hHQvZE0T+h;L)+=8o4;u
zM7b1<i5}-mf+LSY<?f0cw2@5g2MAX)vLWd(-;c%?)=w<(QPRSYc~BUH4ZjzLDDqcm
z2jI!t1noI%lyfQE36uS2?hXU==wvL85vG-t=B7wV0HLG|Nr`6;9)7cgG)ie4ohr12
z0dKi94apcQN>=k_ip|*jgx*pXTfzh#3^WVsZG)X9?IBej!=aEi?X?v?Tm#J7)JfJM
zkg760gPIRqFjTe5H^;6~YlUQ+b6N2BiYx+*3OQ(XEO7|r>q^PRR<W8iu~rl0)vMn?
zsf~SMpl^0MZpx{)lXRELv;+%F=2K`~FVp=Qx?td>gCF8-)!C4_GV<RUjs&YOM@phP
zfbJ%`L-XbG00>$B=v!JOR~P~-Yp2VVi3;7#q<0GMrmKQlN3$d2)d7TBfYcHz*Tfs!
zZ^Hq-wguH7x17whNvffrUFRR0R%cq+F9K=sN1}b_^$v$|Uiznb&*E;AUk3IjlorSh
z2|tUNkdDJOOK$W}jYlF10I|z7gv^y+`7iSk(njsi+>Q+x^w&NPWf1C$^T5M?V-q1D
z02^^r%sOOnzTBJ^&tPU&Dp@DD*fTgEv$R~m`<Rnk0|gm&w<wg`VvUngMRAJU?!%CQ
z(@r=K7r-4c3_@pWVcnrh^IOAQpWXxH&TeY!I`2HWD`3#Z@|47y<Zhd<OLuMHH%IR2
zwzDlm&CEr{X}~>QCk7(}z96j1y*7O3V>`?50>JzXgXQk)=4NYjZd3b<+E(uOTZP%?
z8{~n6(^eHmDA>q@akSOD(Mo=GGW@--fN>g3C`5-<c_>h#H81Rh)%z;UCq!cw{a-9V
z2r&qx-2gb6WY!N(KA5ehKDIED*$$^23gUtx8q@^mBsjJI5x5~kDre(&E7UqS$hsVH
zjxsOy6p55<HafrCPHSA)s+becR6|gqdW7u|lI|L$(av{a2Da%1{+5wBuNCeJC4q&u
z6&aLIq#t{cBcatLi-l`_PF{xql}#WP2TGw`G3|X85VOReS@@C3b4v|`BJ8p(3(6ip
zXtXT%mq8cMIn1@TB37ISfd}5CmG(g$Y?xVD6%<np!*G$TPSn8jsQLkKTb#W$@!|+{
z?#jtpe^C_yOvEDn6jK(<x^8aIW1SN8Ovb^<$@+i`awF|=P!wfDEEZ)bSW}D=Pm{q`
z4b3Je8)KvjG$YbHL?CjLy}@=l(`0ijjdmkAcDMMD#1h+Mtp)4f8Vo**QiJB%emEY%
zp2EII*zP|B8)qTL9Fv2$C9<i7Y`Q)Es+@+mn+^Y{;xS%!#Ft+2HYbk+e;$k)?-^KL
zcKV5HKm;IccY!(x87w8zqKQ4~@6Ct52Kk<b@rl4UWCVHKj(GvP<vwB4;|kBplTQ7u
zf|iv3ararc@F{y2#h81`(}5mJTi69U#GdhI7EP%A59C=xYe^~cCm1L9Tzm&EOfwEO
z^ylLl{vucU|5*FZ@H(qw-J!RHkkCRxNvQ6Y+)bAowsFHUm=0UgwzO?YTWCwRWx6S$
zcR~kOx%b`^TaMBz1VZm2p(GGWAR*lMow6k7ocrfK_uMDG=-c17%B)$lX5J|aS5L~*
zwl>J(2gROAFKp4yc+31B{n>o=`PoK!)+b;IjZhKc$JMIKZrnYW+EEaPF;em;f2L8o
zc$-S!^XWapJcM0W$P2M*1IaBf+VD_f7;@mHBsEY*09%o;mohN^<wfx?8}(_rG7J6{
z`xbyb!t`DZ`XfO0wf|a8f(ww%u!LXt9u#o|g&=M17w9Rh3igJJ1WSLST+{*{Nxy}D
z0O|f_>^pb_!Fg|`S+0ZBurdH|+o5vineu1<m7mhrPw?;^8+jv&-`2}tVt<1?EK`rU
z(&T{#@>gRoT$&^AMjs5G#n1B)G4P&0)3RR9I`5}vyy1!>aIt*g&v0>)dW?9i{ms}3
z`I<oY0ZjirwZDbQA|U-D`7ovlM@xB?eB`f?*CGu5AF-`a{tmk!D<Atyb|etwYnD%9
zS=ILD(G;nVh{lJb{L>g8)C&3papjzR7UO{`hJy@l^>y;k5D}>;g?#Q!b4;4%MU4f}
zD0dpKL;1p{0M&r{x%|sX9XIyci?GS%-)XKpZUp0gnYw2j4atb&W-$!~raW2%gGIS6
z0zvXsd}C^%Ugj<x6xKt@ca#B=-Qq8Jf_usC9Sr274$oppyli+6dq#~1+B7Q^)XSb7
z9Cs@P4qqjEb#Q=Y%vrK`2Z!AtX&coYvQNBRPS1_(+rePI9zA3~?~Z0ZKwJfj_HP@)
z&g+eEazF<|4(Fl9flXbDr(pj!V>_&ouXS)>4}~)kxK$xv@6gKI23lqe%uSX9;}6*J
z@(pha^n)qG@LI}Nts1E55C?To&FefQc-m)mg(fiCw_;XEr9;^doFNDMI5krP4bew<
zh!GPW@i{mo4~==FVjQfb!(tdiky&_nhgQ8Oh1;3Y(oBP(@<4PE7foQtw%s!DZKEy%
zQ5)olXrxC`5BW}z4|0T+ESX^P-SliN^c$+uv*pMR0n_Ro7sOQ1C3T{sVu)k)a?GQn
z^>LuoG0~ERF=C@2m9v9uinxU+UycY<L=b1>`~IdD&Br)cSM^wbW|8;Mv;l@5*P(S^
zT@{E*0ETkBuZ3CgAzRaZg>r%|tGcZbebXQ(#`xfCk&|N4M+_S{codvh9lT3FD)kbN
zj`js_>&Qzde*xGAv>tN67U}G?QJ_faMXq30eV1T5?j)0bRK=LK185gXoyNzFH96_l
z!4z-`^td~TwhUbghB%Sf-P8E!<-q0dAW2d1RY}EUMr>C|u{Dc}Po+nUO{E9U1qG_$
zK#`vIQt99k(#v^{*Bj23-rfhMfSp3_S}A>k>;(uMWIFbZzsulCftl7X$ifgtjF`Yq
zf7=F4ljP(y2+oIM2ltY&T#%gN-|-pQk23t!m=E#wYy}R%)6$3*VEESh=}s>*QtaTH
zukrBQiY!^lU44d664!B@nT81L82gp?&zJZ!<2nvvy$lF`E`cF~TQY6u?6@Dm$RMAG
z&48{vF7!Mcu!EiAnF%r^NHaE%%q9ev1;dp$OqHQ&zE&!AE5ap)1rwG-b70!|@V1v|
znJ@xI&h}@9h&6AAJtHHm)3JjG$;jx0DbS;Qn1+m+MTa3n9~>R?MNJELsmYia@0dp9
z=jY{|=n04-F0p{kxLn2>5f}g^rO{O}0z$_*gOds@MP?vmn<;f}%xxem4T7BKwaHE%
z+R#gxh^<lF&{|f?_+5X;E~4X#sAe++_e?q8=SF~)_srsi#1gOw%>~g_1LP1dmAlZ$
zGi<~lxhN=t4aMLH4QW*Ab+Pw1U#$h`QWTt;KBsu6Tw=4WMVN4<T-w0_dM`5x8{snR
zqyl8W+zDq_7SJKHP%`mFDeWPpK$(xl+*;gT5pYqS<_6nOMLD-Pw|#{-tze?jOS1k<
zuk->92)<AFueDnM-+!e)qtk)g4g-@a+whFBgbd9^OW1z!b`YDDiLncY<|<LCV3NN=
zmp5esuVbh-rWjKNF;ipzJ){~`s!bg#hOWge^Zv|W980RGOT5!oF2HgdXc{b&E7CWc
z6j9#_2CffgEkWEdaw-ZgxoCvWkp>$@$=HEIrP05JxCICRse~3M`!h&)us>zSDwz^3
zmgnno*nCs1jWh6n(qygYCg38?f=o-}VLXf`(aV+53)<&35(L<;O7F3vk5zuXI_3&D
znJMkW*-W`3+;mM^`nKbh3$azMjV~1%(H`(6ZkACW??$=K<)9*7${GZ>C@0P9VuAVB
z+pOwp8!BYF$>oW)azo60Y#xVjrQDe4!&G05d6Jvbvt@u4Y3nSLn|(|?$^hBx4LI$M
z%O6;~D8vFI271md-UjV17K&pu;jQ)znWi;0jkS3oybW?&NPff@!W)7oxBL1+cZLwW
zBepooMk}a%XWC2148pW6cX>H<hr7H>K4Y*J;pV%8r$^;kKK~wTdqh6N`*H8JF%N=|
z3-?P^$bB(ps5*4@Gw$jB=pQv9$`CRx4;Wd84lWra5Bdl9K##fyQsiDh??3cs^fDMG
zyFnhZZ(3$_>LL&O_$dVzU|Yd*1BqF;=aJ}Y0xETr5|7&8K~bQLP4bxiIX1`Kq6&FD
zxTIYW0*!}=?0X`um|vo;7%2@NW(3ccC*ym_@zof}T7uyaDYQ|bP0CY79o&~=)R|$l
zLnULBL6w<44d|e>zj6MC!+n<jVq+oq@RXmAQ*U-G1{|sNd6|<2(02VQMJOvTQ7M>f
zR6~kWgEXgAhjav+2SU3{=J_w-aGVKs!zAc@UkMnPiUV$fHMAhuEjzUw1}`@M!Wb0t
zl7Kud@{gF^P1sQrJnb1x8I+?l&|*m%JZOmy7wMtJ@{IFw(9V>lv6v$eL_i2=SsD#h
zJM&yEx043?6=0ARMqgHKg$l9K+s8qv)-O`$F>R2=)yP%$cBTyFNGfHuHH2sd;L2-a
z3UHc`mhg(n+IXghvk9$rG4o;dnJaR#-U(?c1OkLqo)9Y9;2)Ev&FMBzHg<3vu8|jL
zA*Pnt-zIAjk&)^?*=z@8$-q+C;*Far1~CUcgQx9G+3FKzx0chgnnE?P@wOl&@KtC!
zmGYxRc@~(T@A@A@kSPxr%a23C;c_Py1>bI?C<B_Z!~V0wu_s82x6;1AJuNBd;aU-I
zr%fe~a7SKh_7mepTz$|Wt)UH-)Yr;Soi<dqKx{4D^)u_HPf=g_xxHJc%HibtMSNH~
zs8oI#9Ieqq^C1V^>ngvp*T&)I(t7!|wUwEem*03-;BF>0u`G`KHg&&I&v6u^m)|)f
zl|xNOE@x0)L0_&s?+m7e{s*IMc@BvH8hP+^(y#R_fNP72Q~P>;G|`?(4MI%;=iB~g
zA)!AyO##1ycjmcW0NH@Ikgp}r8WG{B%gJ;8-!+)ypKQfgMOyjhc_&)7p|MV0u-QOq
z;o`sO^QH1hK0&^wyku{*%(!6mxnt$!;1^tHhAUGl<Q4yfrxg<v82a&Qnkna?cMs_9
zqMn1;2=D+SeP1&k7ze)_3z@tgjAt?m(^IS&@h)@u-||Kpo5hRLVe@8M<+l33^H4%%
zlf30`Gw_YYB#5TI9b+HNTbh~E8K2`wV=9jw7SuT2v9UoWA=tDae=%~l%ovY*1wzI-
z{#=lEV@piP<KmZUc`x<N!UO6pk@t;?nEE=LVjrZrqW(Ylh3-M97z$v~qW@bO7{frg
z4n(Vn<p13mb_7s|4?`11UTGB}_JVvAYgNe$9sd#YMc4_VcIxG0`?y!nBKgF}$c@Fo
zbjWr2)StO$2P)$h`7Gv!6a_}n{xf~Ys-@KX=ga3towM>Nr~ZX?(Y5R8@-KhYwKDAf
zh4OEIM&}puL1D2jkuQVRc4gjwQV%$_!P<1zf8~2t@4=Fe4kO`8BH68@f?Q2CupD;p
z=*9$1Y70f)qoZL|VdTwJmvo!Ea?g$it&zl3Cr~L)_Uf3RI0)UnI~q)*Rw=}SUOz%8
zUG_=8EQ9%rlNegDq2l{S3#IwmCfU!1ScRK_W&bpG*nJ1YA3gi@manDOxa4-++ClsE
zj;#jJ6j1mVQ$Db5%s9Jv03mZTzE*K(v!O8BA_sLeU=LRseluRu)Im;&#x{k{w>mm<
zSdOYfK+$6SsIpuohxli44U5I(&>MWHj|3VGBkW-6&630HTM9i`Ifr+2+!0C@qsEZE
zY2)bwr0j^8|F8ig<vSf6Eu>vb|Lu2u>ByltmSJl}*;%dKEJsGu10xvdsE)0659~-r
zq{5L6hd&^*ql13s`2y}>lw*QT%Illz0GKt_%lB;1J8sx{13vhEd;sg57mOY2(=<eq
zr{^3wE)gXe4!+Zm@7Ss!;=H!V2QH^(`ZFfRSdseaRyo0Xy5b6mbEF;1i9wN(Bge=|
z-YsD|F4PJjNgdJ>LYK770lCE_{*FPv0f=Yev^#Y)_)D2cWeAifurJR+!rAm3G)rf1
zP{XS@3XV|i)h=!HS2nYIr%^i-Z^88`U8BR23S2LYe3NcIUOTfWO;=0zm;(ewWMUWj
zGiGA*_V}4Vwu+-M1W~ZUrSwRREmF*6CQj6zY4WV?O(tjbJfeoAmoZf@faSeY>)UAZ
z<6iSs(kB*_KzSxBeeG|g>!C_K_IN*U*lck-uRw&I)1|)?p&2f1=%JW6cU-cr<>W-Z
zZ5Rx<6!Uu+Ykf+ZId2R{n*e=IwQa@{@OjMfv{>!o$SXSCTZ%|Vie**Q<P0A}JJ<;f
zwGPf|RX5~JlP^f9!VLu0WC88h$bcY)hCMMaaJ38!x&rS)B`81%b7hcsZJa{O&Md1{
zGT6H|*i`iz<8p{@ro#;J*2q2dNzMUydX~4QqL7@gBq8Q9G_}_5w1yf)c%AtH!~FLJ
zhyviVxsd;4ZeFL&tI3DQj)yh44*x^#wOvOJm$8Kl&XN%!!bZ|xI5H%|E^z`fDkK4_
zJY=aK;pPIH5&SSvZwh6N^>Cz&F#=;@m&rLHPW#K)MBWu-Q7G8+rPMba1KBN%4r>oN
zjLaJhPTjBIgK;rz7ZKcaChO025)g*-Tg1V2wp0uUJ}-z1_}d(k43eT1H9m>(DpZJ<
z^L=n+8kQrah`1nVQk|LBBp3P{sO+Q&-ByT~i{gy}^U*H$!8Bu-bhoK@6cU0R^VGP+
z2cyZE=mUlsHwS*XG$!bi=HW^?j!e=RCY=5#EPZ*98}S%X{38<|t6z!d8w?j-FCb;{
zHZFLGbnx<!90Rh|mAS?msYo-&{ma~pAWqg+Ei<W6Ciu^&_o^nra)^&g8x*`56*=TR
zRoN!uO)a&OGm@ZWWmQu_CU#UuRLxYG<n0(vj%h<$RLA6yHHhGj8v7InBfP{?>(eq7
zm>?Q_qZBBT5ALkZ<|?boplDU)^A%DTGeZD1(9lY`BHo+G!knohW4*B_QUfzmXt%{6
z=)omCI+^lAUs$s=gvjY%R3wc)1#?Rn15OU=f$TWa;ZxZrQ>+6d@0ZI|n-e@!EMVSb
zzu>x~p)$>yMr;*)LRWe-rim~wVWV7?#%8eq41BauA?y0;m~S-#bvfT_e6RH>mTTi%
zh%(j`C96dSA)-a(x{huy=gW{*VIg30y^U@x6L_WvQxW~evB1<x=s!^7Zb+-N2@dxg
zQ%Bf#*wm=OfeqYtW#diuHb4_ve>eNv++{pbk)w-$k@p44(bM|_qe*>^0rR&c2Edic
z+$hXzW+;{18Z)oU=21AIAh+281<NQ~&j6&`y&u&*RMe=nCD;TMqFL@Rw!G?ixzjt*
z`UiG^xM;a6ez9a(2&mQm%iaD9j}wY0?pAa5+!F_9b*4fgy)*@)Yp_U=+ok`?eMal6
zufF<Hxj%Y>u?H^$Q!X%3h8I2H^JvDXd7LtCgPKs(QFt&o5-~3Ll3J7eFr@xiFH|oN
z#Tu!Qr5XV`VAS+z(0Y0}`qzWntQnn~Jd)aH1RG`5GI=z3_@WN-nDwO|Iou%wRPb@*
z!5;mK<%!tc**fm-C%qTSMIP03KbeQ)sa-U$_Kq}ow&^}s)nZ00y2dzSk!Skg;>cVC
zlWdkV=WHf!$n1`-2dk_0e<ZHRoY>f9xrs72$c$r|k@K@~zBKPl0f^EZGZ{8&+~Bb?
zFa8{YdZ9U)@8dwhdj@<%daRKJ_N-;b0F)@Gtk-n)g-!rd8(?-L9)x6GB#WX29P2pQ
zuoYIxVt>Y?2(8l1Buo5RGmZvIC`bfZYJ+LYH#RhuAwVvugukZ{mtCI7z8srP!+o5M
zSHu`AfC(~Q8RqIr?|~{U&r?RZtg<EvQCcOj)rntNmX#+)V4Q1g_2EX8wMO6!V&by0
z&cD#GQS3*^5&TQkXT5hwwU)6ZTQJQHwmXN6myQ0NU2X^xpg)7vHW?SPl8kJ2{=smm
zZh}<2=uK~1$4LU{T+i4oA@WAH3)R~i<T3!ZAlqUGj~G0TYyBf*s1|1BAaE|ogFbNU
zA0pQ_$oBMXD-KBR)VZ?5p5d04+J>seLM~INk(NZYR8G?8veV^uzy9(Q8xJIUrn(CE
zBuJ}&1*X8<njwuDXhH4pr!mq@S)L31v)JWKXQY(T+CGd!(2ybdxxLR^X}E=J<rmiL
zpj=jd>74`t+f<fc`Q&(u0>Xg!ulza$z}S)U8*8LrQ9t=@$JW!i*=}o@@&Eq+<af~_
z){p7eI4R`!u}BC3*UFrKh|N0z_su@-U#J;@c{9%M85>qD^05BsKXa$iF9pwlJR3`~
zi@;@K+H=trayl6sDSt9v>~OL??+w}YE0o2H<pt+Fc*tR15S*mmW+g6pD%eWBn1-Nf
z?2zAU<t1M&Xf(-z)h?=0UbcQKM#w8|ivum6W+L;V3B0_TMzjpr62Lf^QLly0gE}wy
z8ldrp>%49RVT#Bb!O>%w5cH;xqrI`DrYO#Q#;CsKeegW8h`nX<wuzj&>PF;<739x8
zz%X~(idnxCM-gtl$GMC+)4#;Z)go8tujxZ-4JvG$V(-Rl6S2MD^Vdi!E!m?&hg;x%
zADmU;#gT@i{e##RQxF&imFsVbNQ~O07O_zN9waD-D3A|h%~6>c0eSWEQT#lCcLx8%
z8&FY=9Y{?3WBUZ6trF2EP(VHjGC@RfDySSPpV~gbbtDb)Su8iqH;#g_Gg?vBMZiCO
zj!fQ03>5?<9TE({{ydQaVX<FWH*i@pp9`U~(E9!r<N-K{lrE@_tXRyb?0=_5lL1aD
zL&ARPzgQta>y-1GFaNPY4;(jkgnZSGH)s<X6m3<mn2Ro*S}i_pC>)WybuwfS4u6=o
ztc<@%cJJh9Rtd9?sUGan$+5x-c}!<d|3>j`Y<6BKwO1#_-%dI4IN7_CVwGc%BPaWG
za;y@9x;leWq7|}le2qO#rIRQAe*O&fQ(=r;1=+uogLSxV8_Ruw{aK1wXyz+?&A*^$
zkcDa3%GXoNwoc4xJg}1?qY4Ar3gcgT6E$Q*zR}5WPkTm%c_m23{JMku7uvWC)`#Wm
zZ2ryog^HE7Nev%Jm3+&8G4a4SI#rf%-XHA0K#c;YFfEL2(w}*VQLbnI-g2l>4oF|6
z9Ojdm#~W2JL52?>9%KTyFqwnw`fYz^@G}MhZaBi9S*>=BT9NO>(9|-}vjrsJJ$B!X
z7KkxnI^?G#joPEj8&Ku9UXI%3W3&1htGZZ@wjt+XQOYsFSi>Q&zUQq%*}+(?O2qd&
zIrv%1+r%bJkYoKNvhsA$V>xX9<Ge2&2n8opXyWLr<oMu-`fLt0AWld<tW|e2t6iNK
z%MG2Hxy4piE>$a&lhOz>HHAiyxkFkPa-;2FwOOR2H}8=rM4Y8ld|rWS%-k5A(@T~P
zPnPSlE3tqf(8I$pfUdC{#x>RXCEDF~c{!54kXa$!jj|b*nJV&b$i;*uSOKWmzjUk~
z2dL)GgH3w)+v*<1=}()zXRsD-v+Cu)a~&D2fm#64+kS7Ean=Co<G%yp1SVlsRQe`5
zF%`h*WXEJbf13vZcLOSAN&o195Gy8+*cCZBo?)bxuDM)J3AwW45$xYngFKUSWnN?D
zH1F0xc}|>l=O8BJ^tR@-3+|6II;nZ=sY#lkv{=q`masSzUtz_SV3GlrC`k>?Rqr@U
zl?@0^V@!Un4D?s9$JDN&lpbWyaG`G5L{?`YmOt3vW0W0Y*5N)`Awz6TsMo=odMy1l
zr-PgoeZrke*})v^q45lr08m8|s>y5_W;+Tg({NV|j~<x?n5)bokVMW-9}#ObGkZil
z19-|fFaUBg(i*C1g!fxU+3a|oR|UlRXn!BaAM_EU?wC%k=8VbKqA*CcoD)k^Xe^^$
zgbenvsT=x*?C2^@&(^4MfofASE;Vo0Sx`=%>(9V@2zr6d1|9A^Une_dVQPSc<R;2V
zn4U7;_XCs{Ip5iWJV!55s{AAu7)ODAjO)5E#KHDih_W&hT#r(G!#wMY5^pffhGzgB
z7bGR)!kXpcI0<<Pa4pZBOKjvQ2Y?!nfJf--xzva|8P|np<+4Nq^6ESs8*}CI5F3Ll
zXxtSf6Z2PxiK)wFJ|~qvm@^OT`C5s;+~1xbX?WO875=tarI3|4Bov^OY|M!u0a-$K
zo82{lgf*GqUz#h#vsggKg;JUNp~|p;6;q_j7{OpTXiT|ajG<X1drXmuc6#+fmg%JU
z387p0imX%{=PRk|n4HKhQj=Dh^P7vbDy>5eigPWONUe{7e=!P%E1&l@Wt!|<sq^t0
z;Gi*KATyQdD{L(Nagx@1GX$k+NzOy;V!@xOhk!x}o~VZ4YrPf&N_VX^+G-+is<A>Q
z$2Y7-jlF@uc$t!h$kL9i3cO6FruPhrVB~F+KjYcQ+JpeYWLl@zp|G|%B4~ZKc<K|w
z^yvq-q!^yd%s=Fv+@_M4U8vbMSmw4c*((!i)?y1?Wq;wy31&oHZR9Az4a?VbQatyH
za=F&2r&1kVun4y92)QoiJp|cc*C!@PnP0j<i)DH&Ae1BqLNH`T@gR(y==%+}O>t%E
zEV(gS!D6DS2Rt1SJA6~@h_YP12A6oq&1qbAkaFtOq5v*B83FZC$A%%bNN$NipnLU6
z?3}K{Q*#9H`>jqE5_rAb7A-*9a(~mAe|vhy3z01e@eY4Rsl(mG8n$<)_m=P(IB=HS
z<!345Nbk1)ra2+^#Hax4@q{xEvKo{a^)k6P+JdgV!aaib1p^|*0BuZ$f_%RX9;LLa
z@>zKxO$8{VJZKA9KUsblAEJB$v2?(+55)$=rOeD&c-U!bt}f=G5ul`X@<{yBR{R#@
zSpb*DqtQ94F6HGhpAiGDXuevJO+V9N4nH1)C~rdPSInSEo-nQ(JoW;4vTaSV@X+zp
zal#Czr-Bkd-Kf1Uhpe4p{D%`4>@d?`P@YoG+A6dB**pdY>;7FUvt!D7vX#r6^v-6D
zhe51~%#Fi*Oe6Pqvkh=$p;G3>WY2;RUFIjEEaPxl^oXY?e83B=ms~>?<_JZ1q4#2#
zDEBhJz_w>BybTuEB8!6|;AMyw70Cv(G8}e^_h{ZLN-0m^kfpxr&?M0RGCPJ!P|Y`2
z*2oal<u-|)J$lKCm>$FBYGq|?<C1~nWL1y{7nZ~GC#&P{I#$kHAZvWkbaZKSF(mL>
zZ@Oo1S(hlnBs&wyRLi%>`j{;ozH|uF8@9pZJ#S_KYCv(1^KzrV$UVgyM0l6NChMVp
zzrM0L283W#WQ53;M6ou5npbx*Vt*DWw$+;>hDWmiF}kuX+CWGJl6&Mw@pm0UGl9kZ
zI93@k#Rb{k$+L(GfKO2h4*^A>=MLLf!?V(2gBv$`tn4&W<|-I0@)M^8m0}H)=79^p
z$y{j-S^ybk)B&pz|1>>A@Hw7g4)QZ^!H+hOocF&|m;JddAxctCChYnPlPTkdjT|h$
zOx@C<LuHN^tg&SJulyG>A$1$pQc(0;e%;AJGssZk+@W%(Ay!lq;pn7471B|D;{!#-
z4{R77aP#H2Hk?UWcmQA`{mx%8XRTGm`F(t5NG(OP{2>jM!F-tI&KvB6lc)V{cF9bF
zO*na<u}%o;a!dWuUV+d9ry~4^9X}fj2ia04&)I5BKzw(t{K@&m3d%9$5@#oLYod9q
z3-G*siqh%eMR~zT#t6G5daSSr9X&NK#@sL_*8N0Ot-R!IQ(dI@j0Xgr6;X)j<q&$v
z%fRiq+>WnUcW|CE=kwJV_ekCz`I<LphrO*9k9H1$viNmpxRSHvjWilwImk`9R^Ifn
zx6d=trsOT_2D%oB^|t*r3G)R;_h(}vgjo-fcR~+??ylv8c!k$r>;)AWe@)B7T-OBv
zjMmG$@l4M$l460p=Vx7QMPA<b-}wtQbUMmEh(jF!MguY*<ZtmnE4t!D`nyktTY#Yt
zDW^d`44M_1uw_4T8PubX{KNjUq%J2Pd!LL!U={{1U-~47P=i2C`831_U|vmI2e%VC
z;R5n9u+MA;5H=0+Pj8JeFC_LN0MMVufKhm%I*$gwhz7AeSRIz0PwU=tT>ljY0=5cH
zOfd4ltqa_!Q6*o7B!WhqfywkAUuuArUxk=PIZlzz4!Bk2an(=0vPpL9tZ>@!T#fAB
z*$^roDIQgi(L(ge9{vmWJ|puic9;Pxdv<obt^#MW>=n%;#yZPumSykGhWgRUiR9{i
zIxB3q<00rz`*wEo4;v8`QDr}W-$KKP(=3twqYaFO$RVD}0i9b7TD)LzseH{}Kp|kt
zMZiV4qM@_Fagj{BQR%rcCDR^g1IOXC0#1N$bXHl7+Sw?7${eTxO~WeWAa9pUWD<0~
zeAAy9kx)Mzi~X&bHm|vcE>$N7$MD7?A@h)!V*$|!P`eJb`H#Sbk8+s(H{^bU9PUj=
zDHp=)-?pY}GPT(6nw)inPb`ji#x5(qM=pmz`i`{-tFcVJYyaJ?TenN(NNZi8Vchbd
zvFS&3wve`61rVe5qZ%`>nLgS`RXVO@+_|N4jJ1i}f(H4X{eogfDDNWQcP`=DanEVk
z69u6h>%W-3t3Ux*9N{E#TxZ8#wVbsaACzRuRyC_Rp3vC?5hehtK}rj9O$u^ikoPPU
z$37{o(q^@T87$TzF)V%I)EXk)qoeWr7{r5gvYC`-t0zcj8wc-srbW{wRv9=s!y}O6
zCtdv+FFt044i>KJmipHGKx~h#(%qk##e#4fY?C4za8|=Giv0!T#uBhYg5gT(;m@d{
zva%8RzGu9rz^sYlUY#B0E(czYN*K^HdI#Ax$w|XkL40DAeQd5JW6zSlL0Hte#M%Qs
z=odS8$gmNErGHR#WNxaQ?0bVVZI)BK{r0&KbWTr~Q`52%90_=E9k8g=Vwkv66Q?Lc
z9ZvVzFm;g?Dd;&qBW==6nw3MXpE8D7@&uF4jAllcP0r;T3vkzzbXEiN*sEkfkj&Uk
zHN%1FnccXDvs4E8D5=9ScZp*)Vh;AVn}MD17hX>_BnZ`5huVTX@6L)rOh(<H28CXR
z8nqAxju56{PA%Rgut|m+Q^BY&$k{>EfmIn9k%nd}I)W@@q|cwp5v;a{3XR;wtOG&D
zlTpT%12h=DN=8Q$cAVkg*)k^0VX>l<fcWGbTlaHIOJr>N1h6H}CIsF{X=eq;OUhX;
zk+<cH^Or2pl&ii~&W%}DWaCoy^K3N3Gqt$gtx3iQ1Jxp#RnE70gD1=7g6LFNW<KgM
z%#sT`w{{{P(LNB52Z<Jx)szc5zUUaEwoUlxB5QHSGU%-rhX@@E(+JFTNl<zGNVzmo
zfVrVG<(uWQp!|7IQ7-ov7-Gpl1mc~USZ1xIz-2BoqCp>&^7tBI8B!6=0}*G`A^nP3
zd$)gau}nzZ1-KE5XO{<1-KzBQF9M@i1+7p42O)5YKC_?;5^7&06Wg|ef)CUPp}c`6
z&5$5VWm2#Tggn2o-&wIlLouY<hSU#X;5DiFc^X8@h^X3V9ymEG8mojh%v^m#dH;rJ
z4RsVs?Yf{*396-_fLm6sNVJ%V4a2aT8>BuYcrDJ-xl-^~2%7PtO|C{m@RD9<f_N<4
zS{f~3HJH%50vmO5v{Zu}5mbw_8FjhnOquF$$Fvc&AWi;^9viz&b*)UZPfohz6uHvm
z#ObHVRsNzum(<#~$<^s&+M6bBukn@Sm|4u%5`_p0Uh8i+!<Gi^LA(WW-7d0^M=xT$
zpq#Fg>rI!yoj4VUnpfi#a%;LDwW`2^Pk2FJ05_Wy66VeIHw2#`=)6I0jI)n*ne@!Y
z(R5P`Mr$9i<F#_L4~8ZLeTSA0McnxZ{!23qL)3nO#?60u<>M_z?J=+%XiD4~6oIP-
zI2(?kRdSm@n+HH8$|2nDj6{g08T%FJ0#>=L<n<k~cP0a;y3>0nKM{9jNS8_eyE-d8
z`9I>W{vRL6-A-N;Da7son5vhYXb^qr+>-Ys+F7_7U6}W#jTIOEXweKVw1&dpmuQ>f
zv0Mt=@2#;P-sH6bi~m3frqSc%LF2|=eT(FW@hZ|*sX){7`H-zOLZ#{Vc-Y>6dIs=-
zNt-bC9*M5ilWN$Fxf34s_Mi$wT9B~o<gv8WIDxUH)X_;7gFNoP*bWHmnkCkrh}9|?
zJ5HYT*Icz#c}HA)t~_PD0;OK-+Rm^^Hlfgs%#2=Pw-WN?&2}I*MC`1XGqyMu04Zv-
zWp>E%^Q_$VoFLbL5}E7oYlM`?Nj0Z7SSUu1NtqYJWR)$<oqLi@wOQu-QqqQ|Gti>q
zH!JYZ-xqjmh>(HMfp7%QxiEwd{QWoz7ulMj%$O{WvDTpIHk>a@5*4>-zvk%B5Ig&5
zV96L+mdHU&mO45@tyIHX?wzY@#yyJdzQW%ApG+?+(<Ci94_6vsN$JI36~h{usb>+Z
z)wYC_^Qg2WYl4Q5QM`D05tN9v(J>G-tk6!@*Inn$>v5*5#zSE-Oz-vnHn|NEZ16_R
zk_|S$PMt23jj02*5-_txvdQ-z;WO&4_yk>ZhiOuS@0QK+C6<NJy3F|6vdb{Ih{(NG
ziMBNa;Siel+iU>&%Dnt2nrT5qFn1CZnjiZ!6MP2TZ?{h}lXJ2oD2BkkdZ@Q8(KdXN
zb+XexX74N*J_l#ZPy89rB`SuSaD=zUS37P5JowYLuK+%2_@VqP-Y5aIBtQ4L5r<*)
zNn1gWU&N;CQPf+08Ha35W~$Y1|CJ551fXV90Xc}jj^RV|;orXr2@LUMqDlb{^^x#S
z>EA~4BLH9hE^UdFaSYnG6(RV&u|LYS78uq22iuQO=$ho|Gy(Dw$z!nUe0e6wLpUA#
z2tFkFV|ts89%x~Ve^h1ZS^tIGfOSgeVBIv<M4pSiSI9Ram;F!HIn>R`@_Zr-Qm;_h
zJ}WQ8w8r5Od@*Q3m<fFUQnbYA#f-dci(k~Uzr13La_*QxCF2Imt3L9$hl43euf=R|
zNYSoaj5Mp)gT@h7Ktrt^i1bGEk1`L))si>84Hi~`{)#hny}adZkaCPAz`fJ>C8a0J
z+nrl2yDn3q;kCT&g*mh~_Wu3*$~!UTF$0ImUt;KR1F{Pq(0}!3w3euB(U1CWC>k#Q
zC>~>*n?=3?-lq4gm(nqV<$Z6&;(8Wf#_Qw*TZ%$CycK`5S4$8K_;({FCc`_`KMW#a
zg46~n5inU*>3N06N7e|gm_(3VOl9QIY4UMQ1@JO5To@H8pTy=VWv<<)iES4XSc9&?
zuxkq(gpiTXthoZ*She!cXbzO2i-20`=kW}>4%2aIlzw3ospd7Q{xkEi7&i}XTK?^P
zYQ!?-fm-Ye`O@De9yc4CeKk~x{}_XUv>VlXvq8S{w}{<^qffdR{6!3yFyRtN*xkA)
zWQGDjxKvhl_cuT|2ol>%d-yX;_QyGfb862n2F@}1pUdA?aBDAr8xk9W6@^hTm%aUM
z^Ipt@%-1RfV2^#eDEL;#E0y=P|HftYgJi!hiUYzGAp3iRX3QWj9w7Pw@r<yCAH-ja
zA;I%cvTTJ2vMvAhF6~n#+Xw8EgxpF577p|_Sw4TAmd{@*->@NN8HjpNm)3nWI==*W
z#?;C;(>z%x*DkmEmeI0h#>C9DX>zc?f$(?Y^DVr`=8zaTtWeTz6Ar9HyEy2GSeb>W
zw0@YiLLCM5#t)A+i~INPBj1iTajfC=L>2Z$a)i;UXW#zvooF76!W+I(I8DB5Kaav@
zKGLSpgxFRVem%<m<kho`;5ga|s0nAv<&>s$;2yKfFD^l-JIeREDDVl87h>o8G1Rko
z#r3h)dMS!qA_zi`i`BijgB+h`VdXiff1Y41F-I3Q=YrOy4Ov*gCv{PXK*2)jd<E&S
z%Zj=w1*T?_4&qlpI!1nX^lpgkL)5FC(Wy%SvG@v{qZiR5DvCw1h;eidR#hbzB3RuY
zOL5|M3EIM_hEty^jdYC}PX-($-O{v}P(-6@jdZt};DY+v9L)M6d!=Ot#4E~SXhdwW
zf0p(xfao6ni~?*f6c^sK(=&#yL*qeCe;le?QYAoicduv%f5B*0HLtfnYlg#{834Rf
zq)&Q}CIq(t?um$L<6F9{34QS*y9LtE*r!+DzS7?ZfUZ^lTL`t4HZUE6OyxP*8&*o9
zk!_gBDfaofIPFgL&%*>_ntn2h;9ET{jg<zY1|`Yq(WELY<h(_4hL6=O5R6q_9za3m
z%rrAroS%<SsAegN7BEj1>LJWKz@HKPTZ8vF_<>zot=2To2cT-mFCo5NkU?GCM^T$)
zrQ_PX3=U2nGDwER-W`qF1=vGpbx}NU?71>D&9*I?*{aFHF@uzwVgBY28J>0*--K1c
za0SeFt(=|avzeX*tmsA=VJw$Jt~-E>k=B1Xj0Lb)K}N-Bc^8q^-571}V>RjC8k1(u
zbHXgObL<W1Jamwa4VFgPq;e_s_X(BbwgU)C1h&QItj!eQBRSV!aj^l7b-kRIo&l!C
zWELZ?c)W2!r%or!`R!V@!aR%R0&5jXBpIre7si%n4NB&+T;x=u#Yq){YQ2f~i(?H&
zB6I^t;3dHySn;@$#NNC#4Qr)(P;eiET-HTR|4#p({!cEqX=WxhNhYztVs1#a{H;>B
zy7r~aUS+w|avw8<668Ah+`{tKNJW=+k>ri)^2{2k;MLN*peOjOv~n@_^8yH~${^oJ
za0y)qRcY#|GXN%A0Ieh!A3}06XN$nz#5554%`L!51-?v53&8SGs*3mmRHwErXKOYv
zpc;S1ATXBZo40+o)W)F7kPE;QKOZYsR^P}QYU=C^Xqi!kEVnD-ZBuHiq<)tH=-~#U
zGWE33#l12)*1l~l*W;`qZH0P-TG7|uXbrT?sD#_5KF<=T%=1LYlf7wEwk%?n2Kzjj
z63otm^eA>`d%wOBd#%ZL7N9KbifJJg#+J&J_6u$?B_?r|H%_rgO@Tr97O{G@w?MhP
zL9R*Sm4)MI>{$g=+?k*eX4l40E<_#lO1UmkopDL5K(<P*w^v3ZeHtW|>4{=Aw=|Ys
z7jCKCkS5IfIWCs~4>Ze-)<nJ@2}PjUP5$S!rb-@hH`|C=VhtF?4}8RWhG0(6Cll{7
z8~hd<VxUM9h$7wU!v-Wqd5i$YnR1)S!jfrndyGr#N&=mP?rMAY4*$guc<7K3Q+H<<
zk8D8-H=M|*oX^F+E2yc64An+3EfgXi^u+V;HeN##IDBvS*y45Uh+As!O+;&l88Dqq
z?u-45L)ifeOt%ftx!+nDfr8EQfVW~&QDf?`0KBB-!L&0hPna&-9|q$WBN_UkpaDP`
z#uzSEqo2sa!_lZ3QwVa=vmZK?JQCk%Qaz;z21rFk9*u1@fYC`vynD?12Mg|;4wB4-
z?euuO)QF-_^}vUp@Rzvt35Wx|lP7KMTV{;N6*BUazp+3glI&!dk)}qS7~f)KGgC_&
zm~qY!9GR8gW56v=dXQkYKMT8-iTiVWRu)7>twh5ZVLZ)EZ?A`aZ%B8u@1~ntw%~TG
zc`@2CyrFu2^hh4(`>?7P1c%TK1x_S>p>PlWT>)roVGv~qWCYFnMQQz)LWW}HK|hcN
zTBoHsERN*?GNu^;3JJm7mL;)uOSN*xQXdT>S}}hMYg!f~gm$1Mf8l3ZZU;oK9=&Bn
zOb2E_e_0s~0M}9>0Ya}zqf=}LtZvu(GMbvVxUcqa>~Ntnu|d{EI|`I)fT67QXLQ}7
z1QhLPS?A9v%~9V=!@t&t(A8@iGV}=4%Lbb$g6{OB+89Eyv|cv(^1!UY`LNkrr09r^
zyd~(TK1W`L#pPyL)7CgBDjRc%ZIf+5&oL+i)FeN$?K67ppuuD1$1zrnAAyXTlDR$2
z$ugv<EbfRl3K_&!NsG0SN6b}|><nrTV1nIGVt4kyLbTdXIozI6ivo6W$mD8a`?Fx$
z&%CQ<XiOv@N&j<ezz~&RB&OCAmX-=;iCQ`G%Q)&0y+YI-C*-fvqO!;c&jE(9wdJ(^
zIwpadO1S6hH{Ldl14~-xEtT{U{?=a1*3x5KA-@Y!>oo#2plG48u;uS#{LrxIw4gWe
z4@UlcV;N=$>F~6Ft>+THW2Tld#WU$^EF8871>PV18TE0N&|;C+XZ;zHFrn$NP}ov=
zE{Hy~5#Y|BjCFWVk31i2l*!2p@fL3xkrz`tjIguFv6s@^7zW7xf(S4BFO-keJ6UAM
z$n#3<b;xct)LspaLs<eOkH~BB_aG$oVqsp7g-1*kXpaNujnp^+HX3HkOJcq1H+_)&
zi_vckl7GwA1Njxa*h=1x<}7#xH2!nU0%Ij~jsK^Ou9tVBIlyf276NyjFMsi8w3;GD
z_^-jX#kgMNUF)yRt4F<O6vgD~<b7XV#PmTODwhx1O@tT&L(PeHEtkIql?Z_%V_*L6
z-I5WsTP)*;!Ls-ejurq!tOS?JXvs(Z4IMu;HJ5@y|LD?sD&6s9L>|!pWy#}-I!5tD
zRLyCpM}x|A9(|n1g3FH3BQg!;6Jr)^nOQKoKJ~wlx&V@W7Ara?Q=Nfy|EEzXS8G*)
zJ`X+{2sBu}h^<);vkC_aQsDmO^E5w0co^t&FeKsMv7~xK4%zt2Xp)yu$bbBM;`vne
z;c(rBPG8x0T4qdvpKFpxSI6#X8EO|BWjB9@T1h)tMfYAr+3em`5t%NCvg#sx_zS*1
zsx0r>RRO<oh`GxYWG{b%fk6a6X?E`2)lFVl!4O0cj{Dd*faWp-t7YG=YUGX|FZ*?M
zOl&}w=}G%XV^}C^l?)p?AR2?*wt8z{i^ed-Zu0eLLUl?xu&YBoh||gPI-hSun~rz^
zRg{Ce8i+%#49}YR*NAIxly63#h%Cmz4LkW;-iiS_CY291N&%dLd*P6-4#HI-y$R_5
zp*9X2V;<0bSd4RIhV^C-k1hr?x0xW-x1$TaIf^RAMmeIZ!K8NPYC(3jhG)um{B5|6
zaQ+!s^W7j!VG3@$KQjJ;%c)V0^7qjarc7w?dO5nQL-ue{z=5d)#c6m<R}W|ediXt?
z24W%R@P8<N=Og0j&cGupS<muo$J(lnIaiL0sgA`S&&u&#4L(&bJQ$UbIv{>1H<pn@
z6?n6pU<53wuf)a0VD=Mzz*d>o{KO~4Gk|n7S%}4SuxBkZ6oy2=4z^ZDn|cmKX<)^4
zis1}IO{uJOw&|f*#YE{6tB|WjeQSQ&HReaX0b+}Wj~-Lft&M^(z^D^#lkTzVL$iS6
zkZ)=o<2y`Kiv0_OykN4V)ne(9TG;{}kYah2^o*4n*jSd4UOoaEsOk`gmS_N7@92$E
zAIegn_<J(zXB4DwjJ;h}y-Ggm*Hz*0{{8z&|1_wL&;_74z$0*SbO{Za>K=@3U8jw5
zioK2S@<sqcATf5-sn+d|yP+W71?B2AUtcqRO$1YXQX|mojcJR?=`qQnEIfXO_ekNT
zY!qQUa;7a0?pCXnl9(Xs02BQj5Su$whk#IkO#^KU;#$<Y;W8*KY^4m2-k{*s$dI%w
zs6HWQB?{2zYRBnNACE=-(E&x*FBP9*X{7X!`}xQK5E&lrfTLMkXbmLG*)c*~L{(ph
zV|YZn&v_W?4Hh`1*2zd4ZetljV@A0^vV9KI7QBc5VR(G9Oh%`!pqatuQYg+_rk;!I
zG=V33jcPf^-_~#O3*L~i{<ev^xlmoq=`c%5Z5-oi@Q+JvGHZf52%UUTAmrS(Hh0oP
z3@|{>3+mJFHa-o}5-?GTM$Y$V%+_FcpsQREyAc=z(nBun>hM$sSc;`ITogl9yu&j4
zIF-e6v6G3uA)=4}ttegMzgtj(njYk41@z&^l1qI&#74Ln$WV~WY-^MtYV`78H{9z&
z5tOmlTV`;-!O6iq%(58%S=k1Ni*j#~QSH=2>0hfbg4Qr{J=?XFC7BwOmVx164ROd!
z_+Jho?s+;*SZb`&n6@EbiMvas%85cffsv0)s8$?Za=~)tDEpSLK@qwbB2XU03r?~>
zfh9FG0;;Nx*T=xq%URaMR;kL=RZDG<2!ery09bf2?|mC7snA7fu)25~oWd}xRoF0B
z1QmIxV2{?v$bm4SdIY54dMq7XHU+1ECi8Gks`49jN%*J1mZlHNb~dIVu#%d^)Y$%<
z91Mgjyff2Kb*&&%j4!JoA?jsnbPqiqN&D18eP=f%;a^#6m+DN5cQsgu)_`6yfyKHq
ze$iH#cutIcF+9CjISKq)Hpso@a&>yPjrxV6AJ@d7aNCt!8#|}uiblE4_@y3kIdZ-C
zL*7PWnHCzbc~18>Epvxv?<~0?ksqNn;68+A$c@%?22gKaZnCxK^(@u!-`yPVtA1rc
zWIym8wWb9YrY)%Lmgo^IM*)M>Qn@v_X9B7@QzYNkwe<*U-dqy8o>{WZ4gy`zjNBf5
z0@MgQ>JERl0Fmg7KxFZhJA-SH&W1zgu2|MOTm%CM8kHOGj=m>0)@Xv&Jw`Y%a)sO*
zT!$jtfS}|)BSRjJvr4%?QI*(fWCl~|e;_`aNI1734_cEbGEpx-j3G4E78=V{4SXmD
ziP|}n3i7begR4dBiUD|!7}J&lNq97-j-MG(KuGE_8yBKW0l7SG?_S-pi#%cfIWbcu
zPg=hv<+!MUG{!0NRP518oP#nW2muWj)T))4zL8uaLJ3F2CoDM%9Nu0x%LhW7%j&i-
zmf30DS&_*QTs^yGj&+8c!We@!x2t30aQPy|l$W-YOS8QKcG&_wa$X|C24z|X>CN|N
z3?b2KhQMqV*c_|juUP0j4~0S&ITfzDSQf{=7>8}V#G9a0wQ@0=WND11v<lk=$Pw>v
zT;@|@C<BX*r}zz7?oFUd9qlR{8(U<BzfH%V#_A?{v@#Zacy<CH9jfW8($p9)q<chG
z`!g-t%Tov|xhD8%c)qF@MZMRCYTZ~12NbRdgQNwQu8VanuSSN+`t%orz&99g;TL7#
zjsFYJ^5CB@oBS8xehjTw=#6^ZX4xDgBSwN>SVn(KI%e83(h2VnV&MHwTayq@S5Dx|
zwz!^cwo*`VE96Jc84xD58Tqk)NyL>#H`AFw4QTf?Mi<HUGy~=+7&2;A@OH!!C<ICq
zMyq}xE~F*Rz{&tFhF^N8KLhp)_R_cz`AIy3sn2jUvRrF=hGcU{L*A_Z)6^clw}26_
zRzI_~Lb`sXS{gr((HKaemI-&lJo!bM0V1m)?^B&y`DL2BMP)K92{TlGI5c1Rm2VJo
z%v$+%8U+fLfI2+<e&f$H8wbxeBXZ-n-k+Hrs`Tp)UMRn__NKs0Mm*8){XHlOsCS^1
z%lu6HL(l+c6C>_<-NMtsW&`s$|DUnlHn!vf`J=zg0WphtHdyFcd)9(Vj<u8O8|xa%
zo8&n^QlN(*tj%X`=1IFO86Uy(-YDe{4X7pHgcrJohlM&vJ(6{aO}?0B3&$beTnxoV
z3;D=Pv3rmVn45;|P#E(FE_&Hm5y20W<rNpy5XACoT8yP|Tab8!^kIBpPH^|gYd$2}
ztj?nHdK~_!6`a$P?2SYX9=%i)HprXN<G3O6R_di~$xSq=zAkU4=`3fAmO=}5bpP4U
zBF&|HCyo#lphTwDUt(<Bz#t*Ej{MbIG6`iMDT5Uj$h)ZzR$bBq9~%Nr5P8pkX`dj-
zfY}Z5ek?z#9n{JPadyBbQO?@-e@g>kQYl@Ro8|9`6fEpQ-drvp#-KFV1qVF_U>d|n
zsf7s0n<f8<PW5KbkNpipPfUt_;?Fb?M$ezQ@~QWsxG!3t1C>IvljXB$LE}p(1D4A_
zV{vLn%I9%#jHwzVU)T>rMvodN|MCy5oEt^vboqCp6*W#8d%PoQF)+O^?MoElSSJ6m
zDPUW#01JN=9gGJ|F5L{4p+e02g?ZA>hkmnFcI##+8-tq}L4wR`+1=k}AUd@{LzQ{s
z$R7SnJ5vvf%4-_;?4}S|*KV>`H-+1<%#5AbJKA@tzZHh$KHc1e3ZSqc`*u@%zfW;L
z*{_@8c>{+I9(cCwZ_S=tA_v5v)X3sFKHWF)*Sre=@>C!w+~w;wh<q7$uN-J^ASI-h
z*YxRT{)TmvLxx;d4(jHx9<CL`{`{to2G<vUD}EV}q2uLX?+am?YFy*+kV9gM(3f~M
zAo>2#^b8&(2;y~kc38Jar9^TGBLe7AKHR5efd|y!wN5ak{<p2?BKto%!p2lO7*|Vp
z<G65H&(qDYG{4);Fe{eGX7nSZ9YHqe!3woGsg6wR$BH(*mJKm^T5kQQ^b6y!n$_FP
zgy}_cbhp4B+p5$sMV65BHk??;7^z!k06If8xNZ)mR^_WvPXYOP-#0?zk^odLIo5}%
z(n=ffpcedh8)AN(&God4kCo%?Yt~3<k`sI=7R}A+(Y47WE&Xpf(K>Iet!=D9H9P1p
zC&l1)+ye(_hgdxji3;aZI959bG3mKLYwN+ZoxC3gw9}<UP{?oy!YNy&b0P`Ly;`z3
zji&2BQr42`;=h}5G9O1O%`!c5A?r6uSMO6*G0dE9sBUR)TK?JQ2D5G*zGU)5ci&bp
zqaaF(x;bi$)J9HKisN$(mVmt8im65}#lbr1k+u)5N}h!*h0N@No<0~7l7%O5`I7Xq
z1-R<cE2VcEV<_;|cD+HL^vgCV3Ffx9yWsl91_uO=`g!E@e*UtVTIw*sDCsNgpGKz9
zB6!ANn@zA(Pwv+GO_*x`BRxY-iT06f58RDm+o#$r0IU@Xc@&PM@_L$)3@QHfIGvvQ
zv8u>)w%TSrBbF1}6p$3qU8X@$_EB3{EoZu@0HC}|N}?GUEi^;c(pum~lL5hNxiL^Z
z2gZN^L~2niAPa*MsZo*wUBFzH!QB+R2K14WA^tln^f2TLK^$^c`i6m0R*Yb1dgiGC
zz#fK$pn?McVEXVFrpCQ<y?L!7D>|R;AKNBEq9HJ)5#3zhL6l}~VFJiSeP?7hbrB2(
zD$$h3<{cG#yCE~RF4F|A8*O}Y(j^^bjPoZCl67)U8X-qb^?jR+4W1o|OL-`DOMQ%P
zeuLty9zdIPT>5+|!GBottK{5nt$Gg)w*w1?0lma8sLKLEF)}pjwX6iu&I@9UgzG9R
z<GndtVFSgV#8@Ne`*>$U?^-Dr_)usL@C{sELp1`VxiD507jI<cqC`Ag;Gt*DBDvU~
zDKlY6OhUXQo{@ntv<)~`FOB)b8;05^1-Z;cdQ**DZlp$}7VdDtPnqZk<e&nMDN9*;
z#!I)2rsdJ*1cYcaqO8J(TgNNpvXYHq03qkzhT<R-5@#^Co_2I31y;6E7(`Ip8oPnE
zc{}?HUl3)k+J#{4GZ1spr^c5k%qtUP=!9k|P?3B!$)C~oVqEWH@Md+AVr?;`rXK6U
zNKJ4cPEXhsu=Z=y-c7aAI9h;>n*1br?}a)X^*_iMsq?W%l}8k+!fY8)d_|01%Lvkw
zzFg}4ZL$q-2cVn9Xi$h<!y+Q65W!7TkOo`Ti#v6aMq@)1`@z-h4KmqVQe=wE116(F
zzEq~f^4H{1h&L-!6CIf483&AIeDW}8^51PQs1lBr7|+8e%Vb)x1p;`1Jypq-X}O5e
zn_nJ&Tm&XuWx@?X5<p_FPN!{MlU(D(P|q?*$2{h1{TYQM?{&qNmFwae!r_=`$k^BG
z6VYH0la;XS+^jQYdU|^+PDK<lfaZFGbqilSIQT~U!#qYe1z{#3tP>*r=Gcw6)-)?W
z2#pH)Q;2H>S`6svmUsmSK!e=s16i<rHYUy;z#%dL_cm8W7C;R1#H7PrZ%M?zb6^nR
z)*iUTZcl5rNGtU%fs&xINA9qRu<BtBXsl+8-WmJ}FAmTc%;v6ctwn?zO^epZCy<}<
zSH{<_%78oV5&?Jj%iVS?;G)cGxhJNHJ4ewEh=gh08`B)ZGU4|*RV}gE*z<mW#_NZ8
zFBQ%dc_6;3&cXZlU}Cd$cLxSX99`ix`JuPKVjIltp`P-PwSbkI4WRB}e*^9%-Es_d
zc_dEffulza8a8g&=uz^h&8C_zu*ZVOsvFDXar*;<0Y39Ye28GxMmiA+@}&KayP$wx
zJ{4aMf**fI`lDo|%=A@ir=(&1SUhW1Oc>}%t;~)IZQrWM2Qle6{;kEsG0cJiW<c#_
zZjvp_6whO3Hj#8fqmTyDY&*R&UnBFZ;g%T_a@9;E<MBD)8O)D298z|#GySt#Bp#Lu
zl45;GQMGR)?7f9COdx(Vz!*Qa$OqG2C*?QDdO>8{dRZK6X?8h!peKw4?^y|Eg)DL5
zZUv#KfAd^!gV=c5sS_1rTI&7lNy+ViF)d3QX{JJTOt?2bTizyYhy-S3MGVzrGkIQ9
zMPmLBt0S*8qb{41Rk2?xa^}vL)y65RuCIw-#~{}Le4mxIJ}`6UbMLH`bv`gkaA*W-
zx!7s3KGqM08RMd0Fhdkk4OclJ8?3K9^fTEQyoY-v%ByNJ^^;^%>Wpzc1g2NX=6D97
zL1D{teTzRM)9|Qdskp7t9Q5=0>O6u>Wn0Xq0#{Q*X)VZ)e0;Vfs`;QB6xh;HE|(u0
ze~pBqwLQ!`4L!3+LF23TcGC_U35+GwD{zt2Vv~eVaDwcN%~0P6RQ@MH9)_rbB<mIa
zvhruVKC{(#I^!2n=q^;M0riY8{h3c%HFb=dsnI`=88$RRC1(-cUw9V`qoh@dJA}w|
z_}7Yx|I%Awcn3mDfzfP~U&Vm0=_J4Q@3D3oll9-FyMN<rZ)!MKod!5B>icaR9Lyfa
zy|BLvTElS1380Xh>O**c9gd*i2NR>#tyb>+L&%qGdAU3te`KmF<(V{fh!0EzwbbN~
zF%cxF(IQ+8MeErZyowa=TP>{!Z?x^3=VFNPYO+GlpS%qm_ev%zz0cb_JMMuUdM`-)
zLSl9c=z}2g)?TzX5oS<<>!J$sQml1t7U%!VHUYr)I7DBupAbOFI%coNiwFbD%4=yN
zOq(_p;q~;4SK?Yxus4iekaUQ#U=ZD#wrs#X%jK<@ImD!GIT%!G(Sqdgzr5`Y0z6e9
zmFb^<w(pCIP}Jp}V1Znf{Keh{|3bMc$3X!x@K@u-F&PGL*UG!z9IZ;$!DbBRdp-n}
z<X8~abF*!uUUqpu4UtueXuOmE5>P511iQo23dQJeCbGs4l)neLY8WZ|p?3}81FwvU
zm5*ZYPoUTWQ48{qn9#^t`8Y-kUn#F_f!->g#3#d0at|J1V8@@vYrueM;w*=${8{vk
z(00_dGx0E`AupeIYdscP#{Y7+%YWVWB44Ct4JSaXqx>s&k?v2xx7Z<edj30DbU=M}
zDyrmtX<zedb1lyZKau<=C^aaTsjQW+g0QKM61SUHV}KKPp-6WH3eOuYyLC5UNmFL=
z&TOWzwaaGQy}Ln^L_B7(tU8YKJ^UFYuZ6Y4XxOv6!-@6TN>*tBz4!9hn3BgJKb|^!
zcUMEZWyX1V*(d#u62a)2B75KNjsY^KrXc%uHyCpUWc)1IKYgH1dDVRm=<X48<q&tc
z<KSy)Y@4+R2jW}g>)r~keuUw1pud50m8LGrBrlL}*s$Q9&B#H~3ued5br}5f&F%)>
z^8CT(gV8|fmN4zNoPMmW4le@YP7U~@kB~Xo2gihl(^V{o`0M65wKT?gJEuc2XElJe
za%guCt^>@@D{EPP3tG%!-5t<l7!;&>xg4J6$y5o*vh7Rd+x`qz0ydKRGUSM$PXTu=
z!)^MV?yX0gDxB&HW272~0X>Fu!~T))Ca!>?!;ML|kOGyEC7|e$-4z~gnbD)DmmKB9
zv7^v*X$yWkIt|FejF6}$$M`d#`sV${&ih_a1xw=h7vl6aK#TRgzVAI~MR1Z|5my}k
za%_-ar)>0eMprBIxM0hMd}Dc41+N!4-nao#rM%npghWSzv82QjInkdHz#|;L3JAnW
z@of7nbWXE;o^*&|w?im>?<j)VF@3|#9hFYfDRC0Tq_%@zsnd6kah=WkcDfixBDcRF
zU1Jhs!t{`C-5q+QlEJ9H?%tQJ5^X);lc|5ieuE$An=0<^_;;2O;J_=GlHDVA<+y>;
zGgvKGgXA;DLG%jhLrPM4hCR;Rii5FtOh!Gw=GQ=3Es&;9cXcK7M%kafPNjBXDfpaw
zYc6)sCg~TW#5qM*9zn<c(XbkXq>=;Wa&mfm8Q6!=;3_%A-!_H9Zdfwq^Hl$Z2Kf@U
zttJE=si!4+&7`gX4qqXs$1~N!n7Fn<&hTg4p1gZ?7WVX+-m2Qon8h;0P_tyl94U!k
zG+ElpXA$7tA_G#}oVkui2By8qFv$rr$ln1MLJAx3KiIee2`6~~xI<#!75C{YXBmwf
zfQStZ-a()Rit#0Spx%?8`E(Xdm*Hs?yz&e}2*|<N(dKzLTSoX|aVN2SxkWl_rV^jX
z$QUxx&8RMovN81ORV1Tr4CPr~qbg(U2V8|M=Y$9>ZD5hDvC#xVm>JYr8r;of7G_3P
z%DCXcQkeA!N2`}}t#KSpFmdI)m~JWJXJx$q2KK=e&yUG#G?tcbg2Z5<kPCeBRCrj+
zh0z8VCXy3*AXhJ!i_#*lfSO3QWhw#l9WFLf@!HA?xx`;H(1P27weji3ztjgrlVXWn
zmYQR9q()(48uY$izUwbIpP8z*6k2O0Ekg=Jr2N4CAm=QWvN$FX`<SiQ>_g^BmiuQc
zoX=~jc^|3$%g?q-g{=fHtR}#f^~qB`qsqPv`0#`f1tY4Z(&y3KvKq*Gm5<g8B{Zly
zG&RYb|Kc}YGtm*Ggg;#-S_^~J10j>FZ5&^?uEK}PfoYfyAsq$JfPJ9N8gG-odp#sQ
zi<_l3Mh)qqx;$u{4<4?os*}3zt%^|-A0RqSliytKemfU3XIJ?8UlGiQc$*4b(+ej?
zy>ae{(F3IrQWlvS@K|Q0A^ORnKyqX7bs3;^0E<3i&YY$576Q;x52;M?5t}1hO=OvB
zwCK?bSLQYayN{kgqZB6FG(V-=-A2nZ9{dAo1M#bO)ykD2%Q5%@5JpuMu1XwDhndOg
ztJAY(7!(Ewh3%S{N_|yUu8k4EGmFbdz}?r`E`?8f9seiS`+Awvh_)4#{Mn8|tdi+z
zJ()|R`v9}oGZj8P1RX{XaP8t;!5nXl(T_lQ3T|lEx&b%ENO6Esk)lO=b3DU>M+KK3
z#L7&>oxx=NTb%my02=f13rcm<z;1OhiJ;co{3WwdOzknx-0jZlI4Gfo-eG^S!adTB
z?zDBTQq`+T?y{AGH8@l54kDhFJx}hjAF8q_Lvn8#Gkt($;>~hjJOgY&1|fofe|Pu!
zz_6TyCYe#QQ67k=%<`z$%mxtb!Ng#MjaZ|IM<}wf<cEp)ZB9TMD+Heyiu+J->WFL^
z0Jw+aPXI+Qs~)jQTz2J^S6(cSdMDhA)cKe$O^?kYc`Qg?o+;F^M*ri{%vdC;*UA%V
zdc-cNfh>_H6Tx|Fkp>olcNy~jl)tU1ewYTUpv;KAx37gX;FS>c+OL8>Ff-cL8#K&?
zHY+%ajy=F34Kmv%g1<qnIdROC;bKuR^jvS;1de+U;E`D_&EB|T%~+QWIIibKhp<Ks
zwp}dqqr-}N)@8;ywIEH81t+Z7Gd16>ve18FBomKvggE<(FEZX@7z6~$V&fSEjyA}W
z=u<6dDxb?`sXsH}uRax7=55n7iBu2crsd8Dh_F#-u&I41Vn9};-w7}+fO5r<qc-?*
zrHy82zAh`P(hSsWgv?-W6$=ut_AxQCj<w7ftF<PLldd~UHCUTkXv>!}2gW-8g~AM1
zvoPik?XWoIdLJQY4~kLE26NbduYjen!385n;^8+qHjoVtwz7M&nccI|=24TYXpl|a
zTUlWR`)UpTw*~)O&;ORq#w_D;kxH#>i78<aBq>WtZnbB?JdjuHH#2Taq^Bc`njhd)
z&1wEoY=(0haap<i*hZkYCv0z{4bU9&-vZfTzqia7FhE-3zb6)CXQCOy(TLCS?qEua
zpLi=g)wq?I4A2_SEHWFR@+ivZ&#bze^11vh(Hp^fIz8nZxQKu5Zv$z-F^jRmrus$j
z`M9bK?(9GnlwbO)7-Vkci~KA_T*DmsRZJfC0%Y;8W9vb9v1Y_?Lec}Upw*!sy5FXm
z8h;U_qE$RseiuZ>nCLC!ampoGEWh{B>ctuO1s6QaAJPm^Q3&E<HFn?AMp5L<733Lj
z&D0j)jC7H}Oh<;Wf9V-XJMIESpS7XmViUk=P^_QxcL;Sb{DcY9e@Y`?2^EM+4?$=e
z70>(IyuOl!{!y?}UNG9A!fCD!h~SI<3Ty)mWG<p77syMAyvyhy#<r$qxQXs`{JTtE
z4s8#RKpj}I%<|P<F%GUpM8pDlHE2Gl5oLsG<u&g|EgRiP8{~C+XC!nP;0hUeBMsLQ
z0dVn)yy=tU;mB+v)6O`#w}LHj0j!m6dpi+=8OH5v-T&Eo8U<IA2FAP-43Y!Litt72
zm~dhZgK1ew`>);@^>TWmwdw-n;pAOc$TCypJ^N2o_mcM$A!k!_2RJcbK8S8mC4UO=
z!k8bHjj#d*$lr}<S2QBwQa((CXMTuD5ZgCa^7W%=00tq`T9{V-k2GoCy*wYMEQ5nc
zppRn-fH<O1l6;a5%sli%97&)0r>vU7kgrv`kty~*Glqrohf01y=W?XhK)Wx<=RQpH
z7tz<itY_xa(2=r6zKAcX^SJ4o2h+cN;0B^lWZ<G(f5xj#P(h3T%9qh^RZjk6wC&Ze
zk9_3=Rlhh0$BG|NtYAY)k->p99pQSy5j|gaD@x!PG!8qdlyPMDqE;(jcXk#f%ViJy
zv((cf_DtW=_M*=v@@Dpm?}ngAe?j&xa<r)!MJV<uGQg5n9P|1u13yI*$-b#o-XX$q
zP=*lt*w4pKm4d1b{+0dx8DmFjzOgpb0p0_vx8UryT7!^6a6duRuavLF=msF7eTsbD
zU$>!9Gp378br<*{mdk-Q`ttfFTo&<-7#$>STQL__;5?|v5glZ|R%an8zUi-qRn&Ih
z__ut}foH6TP&(L0(LSRF+(FNz9@>W#S%ecEu!s8h<Upc(Aio?|)XFO;p@z=I|CY<)
zHtOtDh;aF~eOv{fw;bWaGSNVZI8XV1xqK(+tKbe3)Bz+$SAba&u>s?DrX1-*h1nT3
zM=(gx7G#6AB}e%;nl6YnA!vKFJws?tHjn6>x+*}R#}qjzO1~r>Ao9H;MV)%~>LK4R
zQslU0hTfZaY$7|T0d`oV!6qzjb6gCNOREdKNsjktJbIX-00A#2SbGd&gN=A%S{te!
zu+bIMVS6VTsm9>kln$|^BXWoigaGImBOV8p0?}7Z(#ad5mCF6J34+wl`_752QY8fZ
zzLFtnz&WmyF3wiSRh}wceZ(y0hfo!aXT5Z@kxYWFD&4&sZ5s3EeMm46aP}10Th&<v
zi5A;`_v#Isv`3NS#5)<^398|=im$0$VyIZEp58BG%IU6Hi(|D{Y!A$_zFvB#dEgTa
zjnzXr+sGv8<4q88qMLmoh}hSO$SmA`K6K+Z1^k@f`p1{9|1E=-bh1C=d2Nw}TjUgD
z?UotlnUPcd9lHQPC4`*j&n(hPYgWnW!Kd23EpmIS>J0ye*$0e9Vy4@fiO}?=vTf#`
z&=*<~YY!Bx7Dd!5WI&7qzzZvCYy%`TFxo_<%yppFLddN_?RZ4*$Wu*;I^kgdCBoF1
z@uVoM4Dogd?SO?DFC}Ni!XSqLklN5_7PV^Vo?$Ez2fRdv6}c}5b^yaD02K`{3cmyk
z5i_R?W>vME9UKnu1pA^OBaACEh<ihsz#u%svJjs!Du@F%XI_<;Nyqs&I>rK_PNNoK
zX#5w3*qGSILxFS2Inm-khR!s|*kCSq8mhNKMH0jb41v-Z-@r@_zOR*WHoh8MLewDV
zI+2LWXrXY-{k$UejCDUl#wQYL<PI2jrJNuAV}esMO>#l72o8zxo8j2NUZEdgwp<wF
zgCvQYGu6<+>}c^`6w9s2YIMD?l8b}-@U>RJiwtpdiN6MEtI*mCY`sfuHOJKF5$%GD
zaD%;SvJgEevLTndFf8sbnIPgwfVNVWrj4QUYy{ty2X9<}<EKh0{1r3ttjtz6dSJ-|
z#`ektf6K7~NYJHHX}w<Bse@ESulXFx4CJCoz=O!o0ME~siN5TH$Jh?Xcy`_(ll*N(
zX~C8o&~&vmotrdCYD_>hqM(7)#+vA*lr@sK*TxPWCv~>ksOk+!<_d4ppg>4SXwCI8
zL?qtVNFluetz7rQVrlSa43lOJ<uy<N8?A>5tjy%}Gkw=;3&<2}6id8i`y82SLq!5X
ze`)f!nHw_)5`%#&(~2CeY^$v8;+F6kVURX|_?58)W2#UUopE|s#m1o>W`W5V<l^ls
z<?2M+l>C5S5L+$Rq%rdw1`#5pFj_9J&g4*IszI)^Sq#Z#t1Aj}eelOvZ2Vf89yEtG
z57)fh;7wA9(`@Ahl{d!FaDdair@doG+fDv9gHUKiFW$aLZZ2}aR4rWZNa=^(hf05m
zlB==#Zb_fn@kK1+R(qp1GZ~l4$ZfWH2bIa~{w_F&-73y=hrQb}V=$tD<jzE&2=9+j
zs=MO7p&6uF-5pz8L2~dY*nrjv&Mz?h2D!)E*%cP2biUWeKvRl0hcFWEK7Yn^1bP)%
zo%w#F?J(U|)$%~l7U8O_Z~S0<qUbm`$#VIjKV!N8OA;VDK^}@>Ps!nMdDw?#1hB2e
zVK0?OjC=c_uJfb*k{#@tHlRDRSswFQyVVTg$Cx2v5<K`9$m7;XQQzM3g#G7)+yr^j
zw<Qolc`Bv>!AlR$jNl+#Hjk9`nIYu{Ldt2i8<}OhwPi+aezMFCvcjXnN@dh*jC<w;
zSs~{Mts-Jh=K3>clG8=WQj5(&H#|c%Sn7N-FVT(HBpF6IKL&=RV8lNzusI-vt{@AY
z3K1=(tB%$5SQS|mbA%T)i*V4zv5P9Q<xuc|Ixg`RSavFkNQL^v!Y)gFn#^Wpj`<>R
zg&B3rY#jAD5No*)n-T%LT(#&GX>8OSsZlWx&vpj6Z)LQskQR`1$S30ITVz%GB_V_4
zWgM5Q6M2?n%Q7Z+xvcSL44u--nXpyXdJDlfv;j;{Ul$Y@fD79Z=v0vPv9WQSRYEJ!
zy09DKVTDx%*I?R>-V@AYg@4hiY%-pJIt`!EW}h;#k*Hmz=O`vl18+;L<Pg*ptCFq$
zx|+X`bBv*%iyg<vQX~V(Hj@^Ybe11kFThO@;QC|Z=l=cs$abG6qXMZ#A#YAM+j7|v
z^DH#hR;x16;_s$Y3<xXM9Q%8x&kKEPzAxmt{A8EaGqi(WVkMb_E3LNrh^;EfPyKuQ
zU8RZq%r-k_G81RW&tvCQ0n7YFbO)A;Q|y<!kk*c=U$vW@g{((K#9xObDdegr$#05U
zzX!|EQ{euu|DXIe1~D8VS-<lR!{x<riQoI2Jw#$BYpGJBp_hVo;UBD9m`aovPy5^8
zI}PYu4ju#Y2^#_u2)Noa)*T$iuy6ht3<;6U{cj?jmp(rmvjlPJ;H8<b>V`b$77n)_
z#TtPGgE~ywklc&qPi?~haEz|$jf5NxGwu248NU!zrT0rA5P89LPs9SU2n2M67h@<m
z9vN~#OUOjpOF_E&>=p8|zZSr!A#f~q^-5v{bMvsC<g5M+K`)x)vk6=FHGf9N=bcPj
zz#^}Qs2YF@G$8LAq0lpTt&WIn6>pk+Gx)kX+c-(@W_ZgQQ>?BI#d(=s@wUAMloV>U
zLR5dY=6M^VdLiGj86gP_<!=6x#)D*6uq+Ft|26h%Sq}FP%eyhW29zQMfc0MdJT_B}
z$jSGEbb94O`_te9-w3=WVz&G(r~&4J>l^q-v;4itU2G^r0bbGZ739O<@gb<WE+5%n
z$V{ck`bVf_=7<3_WS82J0SSe*=Z}vQQ!iC5CxO+zw4cP>0TC)3h0XtIOm?i6Sdq_s
zcIqug>+5g`kkeMkKZ6XsiaaZyn`k->e!4HbU8@5~y=$BND`rZZ7<JTbDgGTCjB^ml
z6;QkxF9y|<;tA^Dm$Br75FJ&WN7mebg4!B=W8qc|&|M>6Ikf>n04S4UQ(X-_<A}Kf
zw(3UZJ|T3*+)yH%yA=mWs9>Lla8YV+k==_GwCPvWSN142Y|lWfs8P|T|L$4r7+H0`
zDN_y3+N;=69@MApCwmvW-8!+cLW?vMWS@8hCa<PvyI%J7PB1SBbv-#i^Q{WUA=6*>
zD>jIeyO=2L3iyKdw=oWZj(I@(nKA6mI2C~i`ZLNgn6La-zMkGQ1ad3LcVN5+M1jB(
zRATu?v4fjEihIjJ{sJP4z-PFpj<JU^1&BJBI5fPz6?B3;WKcXjGWeIpxte!Q2dHy!
zv9kQ7yxOB>b$9#!<d8HX+*Rc4d8oHUoM5G%CMqO#bi5Dyuo&H#{3JQNxYgj?O#S(&
z#`5i0jSQ?*MkyYVI-tz5D=)wEU(aA4$aiDaQ10eDIkH%hNvH?4a#YN9T(-OlN);@e
zqn&Zw<EN2kCN+<Vv7HUq{`caKlIk+~e$a=pXlYguholwO;aKYu-nUXYZWoeU(AV+a
zr+F3)`O&CCIic7w#{%>_%`H07zef-s5KL-82o@|h@C+VSqU1P~n9*uie!&j#mG2Nk
z9(+YXIu<MbiX!xl8t~M~IBsCBUOEQ_^l)oe_DZ^>zUgn82i<0?bS-Yx8zmTX-0pwU
z&E{Os3*NfNsylLqQ(A2i4YWbhL6QG%#~C9eAo|+H6BNf_2W2K9m0x;9+toSg*)}F}
z5x$)zy;6IO1hyb!Za+)KnJ>NlcUmCKP_Q6cT;vQ&@e<BH-X_-0QN2~tH<+;mWo0U*
zpYfD>rpl#%46r)W#N*`T;5bB3R?8{75LA;Ru{?~sIyJsS=Eh_Y>a<v*;(op5bpMPV
zB2NfcaXMShh@DVZ&9Jx%IWtYTO;c5C1Cz&;q$EuUX%Ehn1EOz){u0|77_5dXH&77)
zl53Fh#7N|F4UU)9#abal;_rdD(Y&Ti&aw%^uwnq|(3r-!e3J}I#O4j>Dq}DL`l2iq
zeYlU!Qh8NllCxv#?aT}=(^eS~w1k&|_#w3r>K$ukWQ@NOweAs4k846=keaB8#N`;)
zuu-Kl#=BhrX+~}&W_pf2gT>Bj-vAt8V~zFUYn(+K$oPgx39tSx^#Pf4o&$=c$8kmz
ztl0{P`E#T5;=X<4JpT#Nq|i;qr#Zopf}#d{5_4iW#`!)-Mi#No05QG`V(~$Gt*U%s
zFk1#yN#!Ej8z_`GNiGi3Vx&}0nDcUpKV!HjBP?LB%cW@!+Z2@baBjSPng2pL&*BL@
zCOF6Wa(VO$r6Mbt*d54L1hQXdgUH|t2Pt>q)DBo;6DsoC!VM~-xgB={GsxQaKxJxW
zLJR@g5rbZMJgZ(Ll|c#Q0B8Z<s+cds)6ogi1te!ooT+M*iHVx?wIytRnG{@uI5EYd
zq&nWBuhRueO}vHLA_$N|VS!p775#5(z&^Zq7s3aQl6<ivORUa_cm_SH>VoRbUc=(j
z7Qe!uQKe#PH`SK<IFL{~6p?0>C51G9ixuIh%vNYH(aZ1>{Ley7?!^h-7_3%<O5&5#
zKoKVdyL&p&1bhz6b4ofj2tQg5npxO9RhGgYp9KhGYBW|rU`A7{QY|7iWLk0Sp|I3i
zX8ez04PCp+m9aSEnUjWjT;&6?E*LRqwcFRq)xkU!fVq)^Y+vRrVjPaQYlA522$;ec
z-0R{K)PDjReZ9ABhHW@Vc(K#;AUxy+qg{D1!ws>T#1f(56y(O3WVdeJ<))bAAXYND
z*?OW)MV;dZ!5x5Qm^utMirkV!1f3rSjhOMQR-W8y4KkV@TjjQ5cWn$tK$e!-*7{zz
zd&6WXo<nq!-jO=9AWUM|cN#yJ3>zhP#oSaavyg<l(@6j?EjV%`eecl38SH}M7jjRr
zT?x*E&h2k>#<(JE7fp4y*o8L7-@Qfvf(~59`-0SdF$8bmR(GA;AL3TSZl=lurWN8|
zlxkLs^90!kW1f&a%*mWXc+t9;fO2!p^m@pq*uQs?JnU~%U=Rgc0~7m^#7J}!@)y+>
zu-K1A+wea@7eHpyMtLk2TLV8R(I5BktuPg`5V0e;fmg^A{ylD{2mkY~+V%2ev<ZRC
z48|3B_LM&(yJ@Ttp3Sg9_w3tKW~LKQ3m>Q)(dc?i7aSq8(gY~)8N36f8o<N?nH>#7
zOweNm6;9?@!~KeiWp4VJ_^3ir(j4<cdF*U0?tPSbzF>&cnUCbY`O%vu=EhpD2ZFmG
zjWd;mGsyxTMv#;&j9oDS?u<pTD<;4Om&Gv>uo0tYsJkzT*LxK8lcoLwL+a?PM9F4Z
z=FbRkvNko!Q^<0c^o(d;VKc2nki*L84Q$gStNcT5Dnd*w(Xl!;#&Re`%;w9QHs)8+
zZPrm_6D*ar_6@KF%B1q#G^kGN+O^FxTO8<m>m3WQ1<7k0Y+eZYL=fsmBQ6R6<QdDl
z$-mY+*C7s=*0I^%M<4+1!QB$e<55iIvejCEJVAB>F8MIw#)J$=BQ0<7qyJhlDD+qn
z)=~IztQasP7As_0y27(Q@a>5xglBnBFOePojAk42`C+z6OT4Fv(N@u6XFRinKn5Dg
zPh#{qUsmk67p=AW2&ly}(3JhqW@X+I6Qq9Xt?>GCB$;U_;m`aVll<I5dR9WIH8J;d
z`=({aS-kfIhaSlC3#S25srBHMrJ7wwuKZ<Chi#(;UdC<9_ABq(JZ|%luQyMA?R~c^
zR-z)qAoc6yH~tNSLl}~^28{gM^bDo_IXi6c-^Fo;QmtBu@Av)^4SFsG)B65kD}VLX
zS6_CeJnb)-Im7+ES)OsmB5JWz{@B(8*xKyZXQQ9e2JYvAJZEhVz<qJ$@+bSFWyVN|
z|L21hvvqkuS}%BO^I8}Rz~t)d<wfgn2yUvxHhRe^XHJ&(sMgD|&SMbU+*k=S_!V!>
z9Qt<G@~dgQR0Gv(F0U209tIuYf6)N&NM3j9tNRsSEW|2OTW7Z73dD@OVe`Qzhm-G3
zAD+h$v}3BboNU}*bLH(cv@K{zu^ahw5T-seP2NeASOQ8wK*5FZm&8v@Cy$^p<nPK~
z{cT$(rH#C6&sgRoUx$j#wG|n8FTH2QAebufzDcaE0LVV@4on3xCbuK^Z~mddR-8V<
z0^_V4igOjF)`zi{RS04!$VdJOvpS$I!Vq66|A;A8KvaI52E&4kSPD$>lW1cgZrjbu
zr=c}vYsy&43GDNkkI^@x<^|9bvq1jo?J0zb(-4Kb<#U@{A4u~rY`BmDOc+=z|4OfO
zhv*S1{|>#kj(3y!Y5!%+6V7WUVf@G19Ec(<Us)S)zhxVypwtJI3+Z847{O}TKpJhh
zTMtLsGL2J-_d-M5y@%s+h(_S;cCv?mw*Yu8gMOfp$eukM=$ew5Dtq}0x(&e+^yTi|
z!y!T(v&5pdQa=X4m&-oUJ`xY~^1*$3C=!MHauJl+P4@FIc^(mX;;rol^6%fn0Ya36
zZjb|_SJe~Dr2CqG+r9)Qe_1A9?_nWxh`E&m)0apJbTRsk9*zZ7;@Ub~OjD49?2DEe
zLx#vV{R;=?F)}0z8wmilqaN15JuFzQcTYK_hvA;>a$d0|h#L}tS}lhr%2>6{<p}+>
z`5orJQ1BC^-l#$xM}ZvPjwbXEGVPFd#A^ArjY2QAk|S)Gyj~4Ajluy6<l#Fu4J5KO
z$aiDnYUdezcBDUxVn@hJlcTIrq+p}=$kB;KDtO5RlfjRPb`++xbLe}i9lInI6&K|D
z@o9D{GFw%{J+_BKW8-m4687eC@i*>9WlYcU-VC65x^R$9fN;Ik9ZXZqn@Oc7CMwbs
zs2ZlP<VigYuLI4=u1wCuze!2wT0nE{;Hw1N8!8m8=TMuL97ldxDjm}*wxz7`pj7(~
z-*@u<?MABh^>;hR))-lGzI2J{O#$AKZIG@$6?VvI37&r4Vk#|kDUwx)7BhjqyT1)(
zSzX1vGoq-6#xxB(TZ-FFl<bnAb!8B$u#}=bg0V;CP|ln7SWnwVmZ2%VVvT@QK|x>z
z^xiQ@7<9-^*v^tveY`X3g4mB<ia=Ep&DAg_=o??c-bPrn+FAW#*P?8AUit??>R3*t
zNlp%FK>9E~_!MsfVv;Zr;u`16sqqpotYv`IX}&tU<d9RkfP8cmo*s=r-J<>qKupfC
zE!w|#Z#gq)0ufiCwLMDw0|SY%wP3;zh$Z&0yKNYc-E}x{muaf7kwGye4J|PbB%Q>Y
zK|ofp9PF)eUKEHZUyH#K>=|P6uVuzSTyX-l&lk8-hQ<%oxtcO$0$Z;Ln{9-z4)!gm
zH9SV*Fe~HG&bEmT$DLCnywiEOu8MdR3#QmgjkK3?wNqu3w}GP*RxQW0Sw=^@h}hWd
zqGn8bwwmBLNV`hT@n;lPhSV&Tv9aKzO2&<r(r8_yPBlzZ#@REZO;zZM^6Kz&;~B6b
zgvlwebDqhrp7=i*AG-t_n%vHR2^+4K^WzunCh9)GE6Jd+!7hkRu6nK=Pp0!)67hvT
zHa=qwDI^`qMMjujkmMJ~M9!+u;!fa1%NuZ(Q5Hmu(WN%b9MSX3Z1)42LU#1!{)uVw
z_!ovVo#+cB6FtK~hNceh{%R=;!7;w1RLcDY?l*JjFOiDGV#}}qs?u`MkcoP}O4h~;
z579cPG!tU<LwP?yWg-&G{HhGZS4b6>s@M#ulVW&UF7_xgFD5}oPfUZQmy8zT3YiqV
zS%6j5Al3egWdPz|*b+4%S*kNrSThK!SFLU2QbnJL_~c_}!#=BIY-(LR7|L>WK*g>w
z;v(UuPU@WiCVqHrC>>l3e4(P%ECm}>zMKkDxika?Fo0TV^dXYX>9t`Au*oqqh+F#G
z5Ii7LVi`u243epd<jkrj*&;cOavEg=Qv-ITG|Qq)^JjbmQW&<$I=RxxPXNlI<Yr=5
zF%A8yv@ljN&TKveN66JZS~d0AX2{g<Hm4rb{PcU>oAc$G9*&D5PP;lc&DQbSSjR$s
zLIbdw>tbfeERpNe-eAfW$I5Cu>p)Td_hOkI#KYC4wQ@t^<i(1RZc!+k((6W_ZDcl3
zUf-0cLKl(Qk2lBK@p!fMLQhBpbAj}OSa`&l^VS-<B?eI3zgJ(mH4WG9*<e~a(WxjU
zahrFUP;^9V-fr{Hmp3(__(ehP2!=#~-&|!QmApHRyIN*ke3{&pM!ix?lCYNC-9|FV
zWzg>)e}U#MI7m0`y)nuudBhjqmll&5CUmN>9udX={r)YJz3q~|`SO5uK8`lZgE4(t
z=#(8lOdp_X6}Bp)A0G1N&BQQEQ|ng9!?B6$VdBHKt&m4zC;;kt=?x{EWj{Qceql-}
zbBAaRFp%jn?~^fiREki)OdkKQK4&SOwhUJlJmGKi#Na;UZFNuj+lp~1CsAD`y7H7!
zvb3SGqDf}>r=|vh;eZ6qj46yS89TOQ)VK>|R%(ns6EaI=I?lG0-EkLSg}VR}&xt7k
z?qk%I)xF(>?KL+IADScx0F?+^hRJcwX|jvd>Cp_!ab6l1{Wn^9LFUH_mm&077DP|x
z-2^^CjS&|g%0mB^Av(-2W2Ns!v8uV6n#S6EK^7-@!ZTFQ0#XV8ZD|Tij71Up1p#Ke
zcxgylUN7H8mZf1^kTP+Y<u*jLu>c6@3Zpa-$`u(|`G1&u&+xpeV{Lfoy>~+I9ov#s
zTmZQn+^`I$n4&J(k{+Q)vMtjgB%y@fLYi#jB3HS08?Znj1PCP5K<E$%ge3F;;k)lC
zOX8fI>w5p3?|q#sLfU)nRc6haHGKlA@IC_;BLaR|>Ojq2fW>NA{vS@4>YV{gX46^`
z?|~*zo#KjxnF!X7$aJ*QFu56WrmV8b41#+RO!=$*OMXLbu9*}qH`sdc(QD#NjsW#<
zY(Q;MSsQ?aL5v!I<bRWOv5nmmAmR!}72p6dlzf_kWP{D60pY8%F(}}f@I->w)+Xbv
z`d%pyB`BDdnNeF0$KHZpB(Bgyl(@N89*KUC3mXyqvf013>V=FYTq7Me?f|;5J!&xE
zeTqBnv1qMEHd3f{(&=L{r-K?<lcxVSY5AxLkK>6%q4Kyx@%s+IU5dfq+SpZ$Jb%J|
z?`Rz~LVjncz+(1Uc`_DZIs|#j?u=5L2ROn*)wuwI{nI`#T{l*bG}G7uy4(KV)&w02
zjpEM)UU>$&tUTKVQ8fS(CV#MRke#0jh(8)86Sv_NF3$yGXQx0fBhSYbR#{SttRjZ|
zFIcCj)MyeU4d5?YryZ^E6Y4E5xnN?PJL#C0?ZfG!3BwTh<>2ai%AaDMpc2Govdpg-
zoB%VZ$6s|2FpQd%lSLBgW3^WPoDhUI9iTMMtx>nv0?kzlbI&sJBmgeO)z<^fy^Iua
z@r_tU14JKrGtEl10#L8ckhjt^llKu3ByT6CpvyUpyiCMunxWq`E$;*-)J&6iQ(u;i
zbvpWg&zEDMi@-NJC;lbik<+V(yq`W@1Y{uBwM0I!1{K&#`1{#7D9BYIA0`l4N-R1z
zs)ABJ@@|yc1E3pWZ~I&7y4{R5pfw){nwDf+YEkp+?|~Q1^>8Z!y7AurhYg8XB<S<J
z6WSliRV{~^;1k<#B(s-KV+?o;F+)H7sXq&%0lG}beEHmPk(>vs>wFPi0`qB1VbZ5B
zV+$)TYp!Wx4Q%<!zoNtf4)b5JcB-R6jH#I*kFr+2PW>@e6U<d$kG~1@A5&5fLiw$K
zY&UiHZTxpMM0x|v?K|&F>%Xb7zFj28v~A|i0t!RKS2NQVIj+~5Qd3(Fr@0E*GDp>I
z9jy~l9Jf}s%2CZ#3vh7joX&03<UDlH7}+Mrw1w;vCyE}mvaJt8xyzhpsGx&_MYhXP
z)f{wquz|?-Iqv$0hz2b-rd%@>dFETNLrzlB=*_40I6W=S&4UA-Cp+f2BC=v?Nfo__
zck(GQ0&YD9#uX|%=cqo35hm)EvP-mqyB$rmydrkZQQcoh>)0`}o4<!`RP{bIdh8zW
zhmMy$>~~zi<+7)@rNVR-C^0Sed)at!?J8lWAlW;O2l|QKCabeg+Kq0vrQ+Sb>1(*-
zqHgwn{xx_TjRk}HT=uv9p?g9T{qSsmdiUrh2gc6PC1a}mz=sKCCsm9lQyk>*2MX#%
zmM#h>cL%3oqrPan9AXOxP0f=-V^LsWX^Wt_{4m>L0Fh2-hkFa-`qf)hj!4+SbYsSd
z+i^S6)-Ymlu^eT;QeAy?fC-EQRlKy++I4$Oj8DeOkL<C5oJDmla$Nk;t5>cZ-!-$A
zdO5+@P5~V(9KsJMvC-_$4V)t<df!N2j*FK{+BI^LJwvW(<`;!rirJkUy*6dwsBlWx
zjPMnz8Njf9YQUdci&>)55)K_5O?t#OL&ePsr#<~^5E2H0Vm6DVm(P<dQ54+mUa=}C
zy;ql0oDRzu2KJV}ORfQOFubjL2Ox1&nU2nCH;~Td#p$O*CZe5DAozPzp#gc56vRQ(
z&}Jy*3UgG)c=Qp{C$_6z`udl~hntqapHI@1$Hw&I+Enog0in|0XGV1mHK!X?qm80I
za$0)Bm;rJ@kmMA2r`wRD29K9BoT@-)28&)%>;XN29Fqt5Gj}u8-5hAkVSY%+lDN<Z
zS)-0tMC&$zRx?ZD;50s~52*S8fQ6&E4DmJ1=@#Wf1ZQ&o71;Kn-C$Z-!&J7z{EO~6
znY-JlhR67H<;HbGuirBRg9nZtA!pe-FVTN8BCw~JDIKe1qyc0+Rx&D<3$BpL03_~`
z(f({MO|p2#0QWK01rk-1O@c%yXUAw4UUZU-^$%#_W*`_-{uD=pUhv!==ddvghDuAm
zH|N+yk^Gb*>bU_!NIW#}ZX_fz{tOu(bHbrul~klZk@M^ss&pgM0zwIDz{;JUdN6}G
zR|OJ))))9299qZ?W_n_#`1x|7<HzXnGQp-(%<?p~a#5N#PZYc1?Zo_47h5|Z01C;;
zC9#3c4FFIL)KZuFba943iSQ*ZmJ*|f$hBWArLiL=wWYM9l%<h$42|Wp15w@$wv?)w
zPQAjPdH5YR4JAVY-h*q)r7Gw;3bHj<NwuSs8ii~&z;-64F=|?qM(>^=jK_Uqx6xq>
zR?Ft37^$edPHN+%BN;7paOF~Gz|85@Q|e<D*#?HGWdbG;yP3@oHbZH!CNwCMOS&uo
zsXB{BXh<6UBUXK2DT;NFU9vgJ#8|G$rcux%r*k)DG=NBy21s!6b+b+8+@i5E+1E(1
zk^Vbf`BA2%H8RmKp}s_>rq(o0*v>0cEG_;9M_0qWDE?C;w_2uoKN>QNZ<tZ@a+^S5
zL4jNmA7XQ<I5FD-6<$}S58(uW3zr!!u5x4<G;Ww&ZNC>m6DQXM;(<3t+-;gBv7OU|
zDv*I)a$UfUZaWx_h3Kx2g(_o21V9*WJi}KsjsQ4vgFiDKg1U(1@*~^y;iJwi8X-Tn
z38_)@C*GL49ldjb={NebP$e$`5xgld9`0uFSC*R{CyhTd(cgt%o$M_!F#w8&2I{zf
z>d)qYQM(QVgb_}+M!y3pU^Oq7+j2VT;n4BZ|Bz_ARc?=7!B#O@0(MF6@MpX?>Ffy^
zOzw<!;663)0n|x%C17ieN;jwCyW>O*8+fkVle%N3BCxrD^1V48WTa|p%A&;1`@ALC
zT{5iz;6}OMmP5y)T2@#7nSr`jkDMOzK$-wcM(`|HI;b}JxgWZ`Uh<1{bs&xk$V5+9
zcyQ@82_=jCG7c<m8VJdC@++T3l*`+IK&f9_e*<xDTGMhM#-S%Nl+%3-OhKcXG^`Zd
zQD|moBvhv+m_{O2&}A;7nQ6(Iz)VT|g9be$2?WWVW%ymkOi<-A+vZULK+N$ksRlE?
zrCsI*KvW~y-MrlvB%)#}@+3j|i)h_gI181~bTB`T(uko0#>s*pJ;>_?s~K*|h0!}p
zpn>*Xt0EA&u|+{`U}At$3fC_PPF5B>Sd`pjKS^LoG{~nQSV-Faz3~Q0AOO&=WGhS3
z5}`uHmrQW6G(DrzRku@?#WP|I081LO+$XCR81(8Yc%bbHf1}<lHWU1$47oDUwr5_B
ztcp%yjG=~=qR?u8MmrPzQeo_qHU3N^t96kZWo_5Olg_`J3?b_b(wHh_eOYgNGcZ%n
zvcw9x4LO~WlNHRhCW^rw5AuO)!nV7sBFIKxk0o-p6cp-OC!1mks%K-S>BaI;ni@6x
zbB*#TtuwEghhy8R3qVHodU+&mJM|6B?`x?kSIXvY{WDaKyi!N(Dfp*wlOuB(sd^s`
z=tCrN-KL8Tj|D!0rPMHt&Uk=QRWMf~vN!EdelK}E?L|8cOQ_?tNq!p>a=wcSvM2l*
zg?T2KLoreLUGxkG{i-GuX?!wn=5h$^wenP)gfkg?|8(kg4b56WmD%!pUj|4Otjv0r
zajwZBJQJ-+wK1<{aIAaQj%{^Qo%|uaUPfrDzs?`MGq6%1QM@>4*=Jdz=Ny{YRlNyk
zfaDNY!zS~5+D+!~p>r%9)EI7SJ{B*;;$g#N2Fn%b^F@CK;s++5ciK96$q{l6*nqY2
zvbTU^6fh30z;FI3EqEpUjKFq9sNyU3Z5=XQKw<N0*ZwDs8M)l3PxR-2^pFfPwaRNT
z6tsU4-urscs)0a2XqDlK^hRLiP#EUR$r-(AqYi??23qL1(mJWo<_*0>-i~J`Ezx~c
z-tlLQnO7%1B#?R6ZWN^LT_*3@M&;$^$zK9TimEUI=J38lK(hm3mo`G>Le=Cf`5?Vv
zz8TG2;mGjUuI1r^wnB&>rg<@h+Oi#e6qojt5}<y${4JnV-OLKEAE#;4Myk|D{+<AB
zWYzetuBz<+2&BTDK3&Nii7y?U{~4W^HJ8GM`-!a?MK9%3`?X|RjeHg}s-!<_lYHJa
z9PVZl26^4l&dc->U!+w*8H*|MTKv+VK`KQ%<(}j#e-`mOtmgTz^z97Ds7#+TTfVku
zaM_50T*dND>W^7dRS_-U8cnEz+f7#fojO{B%^~NC7~JRx0y>3B_0!8$b=8C`k2*=V
z$Th_tFA8$DA=%83Ev;2Y>-o*HRj%slSVV+%mY|z#k*(tyqTj&6G*gwjjeUdcPW7^F
zu4?Q$TF0PLm~3Zn8Z%I?Y#(nb>L+Gohg@ghi{KXrUL9FAc8sCNm&i^&K|-6#a^?dv
zZ?>ItU46#@u_mC+F3~T--|A#n|IjW!noxL^uzucdxu!iNF=v+kWg6wLiBcJhA-nr~
zGSECGS~~zoCcI|%@SgErH^OY0WYhPIDKYT29%aq<G9Ur4s$`+JvFx3zTF{{gt#6Ti
zya81M<fkYr?`yqOK$M5S6hbQY^G0Ag)Nvd>@v?ueYgwBz<*@r55Pw4Pg`n&M0~FOI
z6?O6h`wQn7@iKCdPv1`}I8riYB;E(-s;V{@=I}#o3pW1<6FxN8ZU1F05X6wrn31-J
zS%Xt9y8KExyla&>9jdF5BkVl{u&f+u<5PJHHsUB7rKz%Bj<&rohe5eE)1Z+w$JpC^
z_^lori>S?Dg~#Q(q8Xm-GdJH370>a3Ig{!$Ws|gA#|b_Y=x@=84xbC;M4L$!%9Y7U
z)_Yk7o^tjWmifcwl&%Fp9k0<%a;j|;*Z>IRJz`_Ma-j6gbrv3ci<g^+A+A@h%bhrF
zk-h}5%JJ?kZ=o?$xz;^SIJoNGz9n%Ai7dpfd6F0VItnsCi{yL5S?c&jeU~}(t(Ss?
zlzgVDNZ-}Mv^bj4p-_d6sgJkrCN<dMm%cFqgjk9?>&Pb|f9EXemqt)8MzuZ|-t>=e
zGzEa=;6QjNc-+(c4LnL!M6<LLaAlFxyJn@-k$H~Khz+mLfPfTv3sa=>Mqefa(x%|N
z4~*t8l|Tt+TiZLgPzHC6s;*jHj&(zFJFTFN-^s92L2@!tWi^$`nq_OKk8E}T<<DhU
z%!cJxfa5bEz6`e^OR5{?On=SWiB~W++GjZ(q0rZr8WcfL<f5Rr84+~IOIc9{wL+8B
zEu(x!dLIHFT@7h1(T$f;HvI8dULvF8C?F^eG@c5;F=?r0EJ_*<T)GS<<kqf~vkjnd
zFo#Suwu|uL98+_^I4WizP|c$h2htZemu6w50=dRTFwO@;`T^ChF=<M-iv+(Dk<alx
z=@JCEShys~xoNl5Cjy)?1$@$@@xG-HcF&6~0UEHN44szG_YbXBj9#36HCzz10#;sB
zuYmM+Vc^OXc$~?Ev``B_qaBxuMVf0Fq;^qmC-a*Qf_z>@SYDiVSJ^)%WROeznHhRW
z;1|fHHVURc3*Lbx-C7{ABZv_y0>{jd=!&8!bbUbxkXuB`(?BMo6Xz@ZnNfeHah)TT
zv2>*arady4Ratk4B4{p`Y6IY?(Zy2ZUzkA!8dFMBNG-ffCMGoNDk<ukW}4SzQqaE0
zIaceWovwRzY4hORrtBQI6?d)FcMV8A8W`;r$^tHvj1L%QVa9?}d}19CczH^L^*<OT
z;pH;h3DEmZp!+kV(N>g)yu(>*hR}#f()3^8(G`@}>{~*83jFaLmg+!=ZDJ(B6dS^P
z2BBJ*YJVeq1gs9oOiNsU;IU8|owrn0rWsw%LS7<zT^{=e?1g*?v(_fLBIcwfd(XH5
z7Ic+dX>%G01r-8&uCl*iO`sa$YFpsw;$pcbaHS#J0xU+<#kD>Rg&?{aF#+*)J`D6!
zREcWh1*<P~y>$$&bTzVJ|L}X$>AE`F<%ZNNE3_FV|Hz(ow8D(YGzDDbkG-7<T83^v
z@n=@7Sv7HTV~m7rgGtuuETs<ga#Osa_6KwIOt_FqZ}vAz`G5^1$u^T)VkVW17Ke&0
z`KbYE0PfaXeFAQX1G~RTZu4i9yy+iDMaAtwyRdzY6{vy2cmTP>Tk>|H1PMc?+!+UQ
z1h|B|42FXl4xW{}ZQM+zLGJM{&5+KiE|z;!Z^)d6V?dX;`}~<IxRn;JmHQL<QO7LR
z428m!(!|fYFpw4+dBFajQo%4@EdS>YEPSKlH?n2`z|0u<i!=i&uSw!($}a<r=mB5F
zjr>*G29Qo|6;6`B4wOOQCJOeZxi3%~3H*g*x<4a@S2s&28D`j5*ie{6Wv2H|S9V@A
znxP#E);aQE*SfL&YCGrXsuyMI`dQw08)`)$rb0s!XQyXvR62qBwaXmuhk9(v`m8jH
z-XPCg%+_4{22MTonn)W;C28|pO3@aCCOP%uch<KsFUomA>mu#=d?Pa1mijVTV7uS{
zh)ZN)+_y!nj?@Iprz}c<A@`=P@Uqw^8k^sB+t}%8H!4zEn^FIRCAQB)kdvV_gA_VT
z(>ltKsG~~x?q#ueiZ!M)u`+ziyP(V4+pr2c3|h|U+_}OhXb{kQSXOqMG^KIWe3ezf
za>8XA?i{NF86k}@Y3~}_Y#6%WgDz{M@pw=|MAXL4JOWV*M1L(vyVWz%>s8jr*%~-}
ztZayH;mTLvA{*0Os4t<L%2L^s+u5XsoQ~H2zyBu>`Fv(WX->%#S|52h&C%q!HE>mW
z#GiRyWjYng=GY|=KSIFtuk-=zFeEBOK=`BHh)Am;FJM*;FCGiL!L1nH+!m8-F;$&D
zBv7B63vD9uo7_%Unz8^24<9@dpzotmU_O{{)qdcZP=EKh_fIyKRvH?g)BprBLJ9e|
zwlgDugrpZw_(Zu5bvRcO3^}yl*=mZ0kB}#WbRgYxR-W=1@*E%$B<8VPo(`ZOA4-G>
zC}`yOxt)9KHvjiXRC(rm;HAB_U7ih`Y^nx*tdKth+8I0K#w*6gAX7DXwLiw(N}9oO
zJ!h+9kqx8|l;>>?Cry|jFZc{e&I$J=$mxaq3yi~yK^Wi~Qv()2Udrv<3se5@6Fym9
zw$X|K2xwFPl=jl{OSmc?{bLj`#;^GLa9PoI2<6=>c{M=~OrOu-&)w!mhX#sL%wW1m
zUh~#)CkCR@8-A9&Zu3P(e)t2B{=E^mKo%+-I{(U>w(gw5TzM<tsW(iuyzOUv8dKoO
zJ8^X)nyJz<Y2f00H&=Z)DkjQ%)*7&0hV}d<27(p0wvid)-cRU5Trh4OauR}6KS=ww
z2(H)a$p{F*VG=*78vAR^OH**S$cOQFQwt29Rq|1S2{Y2rZ^g1IPM5#MepR9z5B=Xi
zjt3c3KC7eKU;1`El%@=@3pL0;{EhK6v*C!%7^@jLH~)<Bn-Th0QzxHzOJEA!tJD`r
zMIZUp?v0MtBTkUd(*9WfUCM@^+k1FJ<;xfTo@!+jI0DIDFJA`6o(sl9z6!Eg1h(*B
z@k&4bI{reonhN<QAr6gftV=RSzO|ts7NTrN8fu-U7+~q5@BA6j5n6I35YpTI{nV}B
zyl-gYSub1kcDFrKSn;Q98P6b=(VLOC%~ritgrdrj*;i!i-mV%!?S<y95Rz?tAPvOC
zrz()E5Wa11Q$`i!%XZ#B!OTzvn15w^|1k8?(757I&;-0gntO_31v=3uXvgRWybEjn
zZ33^jlRvW~Z!xbsr%5rJsA@9M1GjCL-ksYjp&B>#oWZhdZ&jp1cfM41^L{AkfZ{U^
zgY4crB6X=KG@00WQlRP#*`s&oj@lc`+##MeD|`CLv<6;F0>M=Q#;atnm^(E2j0j-N
z_}=N67gWHQ`}mmV*g!U)N&{Rg_=$u~vTtwK8&Uvmk^N$+Y7Wsv2!03c-@B8}nx&MW
z<$%~t>KDL)9+*a9S_*ns(TBl`6w%-L2QkVRX8n|dtS{!PtCoX()@=}2m{Fd0=poU6
zbqTohayc~GKmw(TmWh83^Jlz?X<OR__4eU04r-#|vm?BTW;_5Y&jWLMq>Vieo_CYv
zsNSk<#vkyKvOI)G_jc_jHF*e7Rb%flvCk^q9qZk~g%JBlvan2!bHG7i03OtK%A8n_
zH>{{`f#>oG-b#ZR(Q1uqNO+D;Oe<WgDX2C;5<MwKrMX*9wthNV&l)YK*dmI@6^$Jy
zrv@|!j~y#L;v0k~)JRV|G&%V_rB`pI&yyxfjt@ntk?{_!Gn$*0#JkwkUcJ+jy4<~0
z(I|P*8VYE}4$r`3^X(Zj0I0ot3P2|Xffj0kCs3#}XAV)N`cfE68a7(`cw5W&G#k#-
zRKJ0#K>xTSrEiRh`-~bhxb{gu`#v)nCb0hLJ<Baw_RG_J`MEvi^xkeAV9Hs#5}aY*
zLhj2-k*}0YCdec-GrW0W;FSRe&gLv6`>YHM;BLMZit9mvq?NeWae)pt2w!@<42gAB
zHp$TVedsV5mgcyKBLLx-%5Wc>$_GSdf)IcM(>G@Z^l)4`ABK3@0XDJbEL#%H9c2}&
zD<9$ABaIa<KFd=*(lETE6=s<lM5V#TFv?Z|#GMIVXta-__95df@F_iN>9^D-V|r_j
zssHRT^PiY-IXn8NQ<jX4H3J9eY)5y5V%wJirKK`1y<QI1(-L-^<IhZ0ppvkho4{mg
zT)Hw<jgK>5j)W3ba$ed?Mm7@X^d~rx^Zku#TY+0p_7JKdPZ!vXTWF>!mkaH$F}2N*
zj%7mZGGm4C_M(`asyAt&St%E%F<GjS93nJ_dU0NorcF$ClQv+mR4z5VKuU@dDe*or
zTNRPX2bMY%+wg3bl=&`EFNUiMQAw8Gf~rwg;wcYY(NF-rBCs|FEt&*T5kNxyHk3lC
zv<(;ul}D9P50v!<S5zIOWZdX6Qsc8C+d%CM?S~V4o5GwvLwdmmm&Enbg+WzllMH5K
zan-9zxVCqv%OZ4|=q_xVg~Q@kuzH-TI-6P{YE;+zqR7a>%EvfhENUr+q@+JrGS+lb
zwniEZn#kbUB$vgGZEgkUzeXB;9bgS`f-KfPn|5)d@}N)_NRvOKV*rb@!4XuN6I?X`
zywb5*&{MI75yJRna^MuOf$|kHDI5KeDLx}cm`x2DaON0kF%ToQB4UuHr6E~d9GndF
zGnZR~qULg>FSx>%ScFWjSH_edS<WmXSlW8IDlV&Tj*W74+9Op5f!<OwfSb0?4~30L
z-{cza&ZHWH{k3VemROw`9!Le~JP2KdT$fg>nQY)|MK*%k@9P~~=L0WR$`8G7S{5L>
zGdrN%U@bt1SRnLAfsduo6ST;W<K>tPc*CE>G`htZY0Mi2<8i&wz**A(Vd5s+Y$$|R
z<NvGVW?#HrZl(+BcI>eI41&HTI4#5|Qq_XMT$@2Q1KVSx+!`}bEs_O(8#_({JzH)|
z2i}MbFH*ZOZvWnax6CP=^*Xu3W_=Fw7R#Nf74y)62LyL#2lcMBEzA?Ec|zoFZ{6KO
zOH#2`?g>;cM;JEdeQ(TLMOA<wnjb=fIq_&GxX;k3{}GqV{XP&&0V84o2WpA@EOrI8
zeG$X|fR77j3Mvn94KWMMDtM0k-1^C0)-1p9erA)ZVZ!u)`(;c+-SDaYrs9vBh27=9
z@{Vyo@Vo#^$PP2X=db+@K}OmJzfR!K3cP8Jalrnv;uMQgO^*petwNP&R~@k#LCC5a
zn`LH#sVX*bNLR{(>5ypV5NJlh$>G{%ZWm}yabRGilUV@}sN@iEH9OFzEYsWw{=7xz
z_)4KF!$<a$hBDmOI?P-{E39M_^eAmH5u7QyCeej&UTk4eZCwVgCG&$(8IwWs5lHF_
zV$hK=<)d;>i7ZT=ZbZz(q5vlJ%gB$cKMj`_`{Z~XP<>7tI5pekPhl07cJGJhW0oxO
z$(condf0YZ8WV<;0aqK1EMJx&_g_>Ogj|ZmSRN<<tYJV5Yh-r2G-O4P#F0Z~rEM1C
z>~O!X@=4Ktw@_C51k|-iJ6IuW0w9$cxC&G5Uz?^m6ITQU!kMxzJzEI$C*RR7>upf5
zYZ=)PaEC?&URAQu#_ebwR}G`jCjXifXHeaG5Zi|w(Da>Sxs-+E)Raqn*!u0&6SeOi
z@h_Fo;Lt2Zc*JJ^QfFaj06!5>DIE?brN6+LZjou+G<h`U0V+!eD<mnB$L#P818#JB
z$I1|5;jCh8d;gnQ-nk`6#F>@Hy(3lugmzpR%-=ew=@-!^Pgu8f0Q_CxAQD@k-dnjm
zX**O=SBgKLimq_p$j{R??`eNVF_Q{TQi0#w6iglf$Nfxv2NN=*`RHW&Z0ra)_&@k`
zn0;uT{Lx!eT}*upeM6tK;|Nc<25L8+4;&ibEHBt#!<mMa9$GK<c83h+{i%%b-j}>n
zstxH2PfGA|obmE0C5@Hxr^GO4>TwmIJ9#Ch@+fW5mhAgggI0czT={bss?#h24v5(a
zsBAF8|C#|0RdcZqultVaJ)#Y8IjeR1jTi+iE979+7`$mB#qo#BTQSQ4Ma9F5<!x{5
zN#Zq>`kizx8Y?QM@m#!Xt&m|b1D`Al8htO;Hvrl7n1Zcd{t_eOZe+0|77lpdpPBxG
zyw?o*z@L#Npz}CnartWuIvk-!jSP$VFySOwUXyWYS)M5$#Sw%0u2KGG5QG&=K8}vT
zMQN1@@PU7ij%yf(3wxXVBR+@acZvMdv4uv84f09Q`?H{I{xt19Zgv*kR0rD6(&x8=
zPx#!2r2d6Ygk(Lxh`phAB=(sspY6?;u_l}wrpWB7DgISV14k3Zbkjm;@zWmUYkx*R
z7&;xItcrXS2P|99+|cFnZIEiXK(M%*sZLPLhe!(E^xv`lqq9(L$am3P?N*kmMbn<3
zIrSit=UN?T4v>~<i@Z=gQT=1hyK&*NWuB?`sLEijBXj{#3^KTEm1pWbW@_QNhG|K*
z_Gfe(rI#u-v)iP03xPtwd+f=!_N=2d113wIP|9VyJlC$2luv~D){N)d=eaJBTaBCy
z%%%hE-67iIAtH7r*TCQ_JLaj5t+E1fOgrVd_H9H4sd0CXDL_PGc7gRUg6xvlNym+j
zR%nq*Ao}ha3suE}<F@X|Zh4(lZ%wJFKozImqvL_2$DA*F_(#gSfOH}evEL_q`jF=8
zVkWA+Y=ZQPlf518YWl@Dbj8Ac>g@sDyzJxOTPPiM{i|i)JXcg85(YVBv$CIcs6h$)
z+bqkQ%YZWn_}Ao|%-wmF9B2)Y^d5m-Kd@!6<}DNv2c^Lnu%Nu59GquLRg^|P#C{z;
zcBC9?-(eTZKphT?2_g6$>EN5>a2sO;=vkv2;k^dRA|OhR^k*hosM=hPav;v?VA46&
z2`fMr%hBGNE&(cN;G#a}d;7#|lWL|_;9!rnX;!zC%W<(U;Jv6HF=>~PK*!q*Yb)wd
z%JPJmKa*0B!#FVx2f{O<13bybw}2qzvO+dyIXTTq)6d|%FOyU9I#cM-e+wUyQv-)5
zH`l^(qr}Us_lW(dsjHHn23M?PA((<*@v3*vUXtT$Aumk%p6>8Qh;#i-a0qn5GUapc
zSjPY$_aw=){%SV=5L=L+7GOX{K0hfiM5<>zbz#5`r<a-gXfiia*C%EI*9q#`DC60i
zzWzr4!sapEp}NFLKOc?RYU%LV2Gw|fZ>>Qo2(4g^Q+xwe=V@^&aPX)J<ar>)U^oQ}
zew-e2*L*DGu-fH}yv_slApa+g6Sq!^3^9sE7Rvx1fXo{Ob(a65;)H3@k<&*8`T)kL
zb0KSGP<qBv*{YV1!GSr76)R4XAqi?LArR60!2D`7d=B-`S<;<!5@~v5m_Jii8qbhJ
zONIvw!FqT+6yQaX^vs|+m%*uD&Pv~!H40FW5ovVh{xTl|+Wl6@NJlI<)IPF6M)|j<
zHc^YKj85N1C?I!(isv!i?grwVG4)IjKwvG9vja_GUXroijb2KC63RN^6Eb_n-3DMu
z1Cu%F6*(?X70*@E<ec~#8&4zfN-dE@JIc9voo4x{tC`x=%<{kEZ99um)9gH74^;rP
z7p{k;@O)cwO$lOh%H;y@*f7(QF2VYSi||70CKn+L6Ko7P@HW6kO!u9OQpZpyV?GQ-
zyx5<?3ls`2HI>UHc5sjgu|zKQQw9;DUP=PeaCBh~YHC7C?HLl~*H=QRrg;U+5^xAz
zWBpVplk%7X)e2Y}V{|Iw8Av7dE3oNHrCns_BKAx(fmB5s6<l%i0Jo~+8O}6iXGW^l
z1Xh7oH_1eAVRt{HkLSsx0KG-4O{<XF=m+Ny_d5i^byDZgK;~in<T+j__34;V>`?p$
z5EtV#QxCRYGJ!e8Wl+%BL~6l>g8W?OP%JA!GWy0u4@^7AL^I|_WdSAGm?12;$Z5(%
zd*9n{byLCkbWUvcH;UV!qfAgh{UtPYX;mtFGs0J<*fh$t9Q4$*N!00ZXbTisDJ?+<
z22~(_8+L$c@c=ulArzOVVa*T33WQRjafQF(T}Ztpt##NMZ0VKh8;-=#DZBxTuuQH>
zGtn?ecj_P+ygDsH>z(7`-z3+herO|9o2Q<?wQ{YsX5#A#I;E4Ty3Tu`EUsQGxKpps
z^N<Us!fbA&jQ7J}oJLs1zZ>iblpsT|{3xzEG%qQK?S;YlsJ9zHLAgTZ-+mI^j)r8Q
zMj^S;umBZ$YhZG}DPV(DlV%Whq{j2)X6uWb3;c?ZKH-)avq3Xv@KmYRE>0#wG;g)d
z91b6^3b`!-iZwIHc`cUP{Tb;kZ#(G!<&HQX6UhqPnLuf={MZ!5<u<v?zon{VA!sjp
zkh{|un$m%?yWA7~jUPT@gxs52tipY667PL+YY){Ro%@5y&9)#_G+Yyi-#<$;wfr5J
zhdf|ClaIyz{5*OY0x!O-{31beb;NUXI56<P^f#(*0CqrF*CV||e&xr`_%13<=$4yG
z6w0p?FgEaN1`p(7t+b|PZJ>h*=qCIC(=$r1w1Uo(8Ag211f`rA6Vzis4GQQbw_P6e
z4=Iu(=?#iZFOyk`vM{HL!VPagube{8PVX0Exuj4EUHmi^`W){Qx&w_ZhIwD+#tcB4
zsHvNS84%?bNL#?NcuEcU<ayo@5rL|t>q>+$-@8S9DKOkbg0-?BFc4zw!ZZZcIOI5Z
zUCn%CzAW-LL?0R_Ngs@HSe%x>ijkzyI&P45Z%tJKTC3K9Uv7{k-kPijBN!pO%2MY`
z|5M?flVw>Vu9RuHU^+pmf>Jfv%k9*RtbuoutS}rygQd{ouT0CG37(UKEt6Gwojt&Z
zu_JJLwB^G4UIk?}+~YZJ=NLdO2NZah%W7Xze5VQd@Me~*2`UlkV6(C|h!3J7;VdHS
z{1VSEko7j*j@IErhscK5IHc96YmklJ3@$XA?AKE+xkMhaMyNB?$Q&XMr#<4KrQVE8
zE_wb(qHBg6ZjNR^dj@L~H|;j+unmXdpiUkQl2_H3X{H$en1KvNj*<rC?80qVC!G#P
zpee?pN69+*jqeJnwb8uD9Bx?gcwVR0;cLx8;DUn|B{AAbej5XhWxjQoEuM(Zao}J*
zM(#Jn^V;Hz;UNBG+CrRi_Z)dDpoW_Z0^HNao(&l#zqjuyp=x=?zvFhYGCi10dDfrB
zrUN(S57Ac<oa3A1kM<|Xy%wQ+&N{rTIU~=fiNHr)@8y~Df<FUX)!CZG`d&;YyDL8d
zs5k9xO0QpvZ<JqxP3zI?<Yn8DhUyl0*ZID6g_Y3vb{#a7ujHBKlAHm>-{qjl#xHUc
zcr~+h@~eS*bx<3^lkm^Md%(D~7XRNUui0qLRL#Hc=Vc~1wFUBqA<&qygW;9+rcIRf
z$Ta|mK6h_9>XcMv<!xUG)1fl*8%vRQeDP%2X&zt*qA|1Jw_)yh*T13m7LFTm_M0j1
z#WRQ;6dad>t@%sz4ZMcJ0j-zH`{@AE>DXA%IiM^n<pbNL9yz__ul^w~3tEnOLw=YB
zVB$(U10SUU;7|rl(M+&2e@iG$*DFwXAdYH849fDcf5Su)q|=M2;|u5_j+8zwD4h{c
zCc?r3eM46M89S;#QC<#IhT8m3(uic(><y#?8gf)7`_#Y<T}YXH=A&b$AawBOZ28>3
z*Z36DERaUx0(II;fQw()Hyy3n3ShVCq`nM<N2y_)_pb~A=u`=#!M_rWEy9=f7NxlI
zwN2%C{F84CO^O;&?@_)De4A3Ejz<6Xd6HPr(TS<>zl%d(1_c<k@sgiZM5rqVL1v5m
zaL;2-Fw1OaFxWERG=J19@Z#t7PHqVHeCQa~$yWKU--Bs~6{>laZ=HHG_LBi_bh%n1
z+jwh|0D3quWAV26syIXT*9WU*yL^@Pt~^Gz&v)G%Ob89KgHOXW$EK<5?q|Paey8c_
z%AqO5f_Ad6i^q)}K5D4!oUb~+Olfna>|(<qEgYgTc8zXv&r?!@D2bV_aLeHu+0D99
z!q+Cdr**M-w)$t-H?l{*s!QR#dj$*K>}h}2IUenmuS&AXH5IbAcj1Hs7aub@PyxM<
zgT^R2t#C_ZU;D6V)cLYsdOe>}eX7&gKf!(-PL@IQ0NYx)`!})3T7w)Idp)eC7T(FI
zsS$gim$>N;4vJ?qkMneZF*(>k+E9(rW<VM{BtTr6ful@|9O|&MrI6&f=?G_l{#J)Y
zpJ2$T#wRm*xIfbqk9G*qlOt@AIfXstNQaIcLKDmE!=ut0WNE`v`BOzSIXd4J!th{d
z#3fzNY@5gUI3~X-qR6qfDIKjy<^>h)aW?EU=t_<^^nhsABq#WnG-+AD?@~E2t%Ph3
zZD<?hq;Ah3D+6Mi<m7-5JToSjRLCjT240FS4H<CeS+F9f2F88QvYoPWNss7*kzk-_
zJ>&1<!6OC_94EbeT>3Us&%YjMlH<Su@2dhF<zJrH<&v95>oQN9P25g*@3e=!Fc9hl
zL9$z3TFDET71vZj6PoXX^YSALjH=C2;9Y=DDYr^-iSB<?AfV`%6vhHTc_?Z0u`b6@
zqJ$@MUxR%G@^eYQKvhjvic{P_-?chW;G<|2DCBAJ1jtQgY*tRUBZ|Hu*T@+OPb`Ue
z^4OK41UP+zJFQIy*jyT$VZeoGJJ99=dYzR)zO2w^CC6Dm(}QCmoJ0yY^f4!?ULZsK
zjSA^F+(<|5Z-%DzHJ2iAzEOt7Al(>XnnS5KLU^MLH)?p{sdA>zKJ;ct=PPHWDRkwA
z;S}0sge?S#)w41(wNm1SD;ca1R=-+C*$JqrsFKls1`B%07$3=$j<%$;{aNfQ6_#V;
z+(8sK1_cl-HsU&lK9VvnkY`+DGhA2C@rjVYF|H6rPaT%$+UQx>SH=g3G&T)a)p<dI
zJ7z)#J3rqIJ=td7ezg9u0@MX*eqhwd8h{bI(4J9^28XqB7(B8v!P^1FA4S#;s{Q36
z-veSZqrO<)qI+7)i(}eqShM8w<^Xj`U^Wu!!2MS)jo#6dIzOWKl7JHP2TsE+UtQB6
zr3RRiIxS!&Wwt~X7lEm*+<T$COu+%RD5>ygBxFXO7f5CF+zjPRl~!g)V!rgTRa9OH
z2kq+k822?}P?#{RCOuovh)VFnvt*)8gKP>^E0Y{7VbKFN>z?tR!S7NVlOD*J<2vt8
zqkVLYWm2D>tpYKCYX`J|8Gi=R7dI{BcTlf}05E)YqFlOBr^{lut6Pv@yF98yG<qjA
z*f83Vhc4^WWvxytUP?1088YjtDc_iP{R@+!QN!Y^tOPZ!%=4u=s0)pS8mK+l#)7yA
zJ1A%l?bB1@8IZgYrpYWQ)zsiBdgd2Oi{bMC#Ntj1k}`g@T<#0yjH7e~XR=7H2s!|1
z609wlqotT5!*@u=*U6OuHedtSMOTKa{F$u<*n%tfYVXHH6nc+CR8k8sGg!@&Yhr!s
zuU?5WdTpH1V(87Q<+|9_&9{RyyFTC5dJxTOD&gLa(zreX(TK`^MtG=l%;9`Pz+rSL
ziskwg!+_F`Df~DFM3{9oYG%ZYK+O>gjRSvUd<6FuhWX-}lbalJE_GmCVOPJ|TZ5>e
zwS^1c64NdR8B*K5{M4K2al*l;|Lm={OSuKPa+~*W*kQG?h@5S5dqPes-YvB>_7CvC
z!&W-3XuycUa%Ze+ENU-S$X&MU%&jDM`_N>07*9&td5^sxKCq}t?hObadYeY(TDdPe
z8^}~~a=%>y4a4MT_E|$SqQD*q;sbL&y^4PxU;`&Y25A+<!CxeV>r(8Rwf&c=^&D_H
zM978ms{kAF%{4<>L8reCxC3C6F`>Z0Pnf8rH4p_!k1AxkcZXrq8fQcc1bLOo%mhbu
zOn^e2ReA}F2mRaG$Og+`Gd`P@@R)v)<N&EMWtCf+2Uf{!14-FLnd4t$383iaGCeoV
zTJz;&tVq<;7W0I7s=2m-<nz)q(pysg1v20IA6Fv_?6>pImxZwcge+Lj*U&XDN^@AR
z)R{3*i*2iV^yt-F+WkFq$ZYT&GD~8QQRM>alBM~b>ffwvztT@5{ZuV&pesJT!0<ag
zIhI*>9jz$s02W9c_m`(7TXiPp*fYj?MJx}O9Zi|YY$+>a@X4r`*eI(Kv?*8{!(;wn
zWNT3WHc)LO%u!&j*LauAphs%71{Q#|-iN8i?TU{`mbGA)*QGJKg5yj#1WpIS|9Zm~
zI9!tsZ%DnZ#Nwf7g>a_P2uf>K$VOkThAU!GU3KU;rExT;4DL`ky@8U%J@Al^0}U)P
zDm?5fAq`zZfZ?uvBsh?Ul9pOV8*WapT?Joc5(m2KLZ+3DfGvojDn-FkKANV^<Z@(1
zxQG0bBy*KKW)rBbX=;M>-s#gJ=RHq;<NXoUm!dtrJ0A~V<Nl02{jEPE55?3dw8B1N
zJt6%Ms?^EvqQxU?p+&*$u<K9y%+&UPIX#$+J-j03$x}uTi$<L#PkU3Acwvfgx~tEE
zwun|O)?EF)(LP#svhqxV&muHnsUhYHTZe1y*}yXrGPvjYV!=g1@AOCijMS8>d~)S>
z`k%9Q^qqrS<@3H2r=OV7N_oM*ovY{Ai~(jve$f_j<tY>7rI?pW^;Cm-V41w^U7ATv
z8D05P>Jq7maP3h0WcJusVij5vOkNHAC}jW`q6PjO#{<>yP-<0Pv-W42{z+apO3+Au
z!#50Wh`<_rzP#z+m<#AHkhg5&*~wM%w(}}_&ltuh6hgUX>;ZFryc1jOj5IBR?*<Ye
zf^dVp7XWw!*;!1B=Mer&zWX+o)l3Et>eaX2_gyzve?=>RfKme`@ul)X+6mTqVeYAG
z<*)XPyekw~h=hff^25}Q*#k+27Rg5e#EB)7QGoSt28fatid`T3Ji&u0N@`HA#v}gS
zd!o9C!U8C?{3F0UvAMhocho-}jMqQ10da+&#LBAatr(ebKeau8Ywl3_%m6(AC;Rib
zdMRFJN+Di;5gWv0?q9|!fHIF>jg%!#N>D#p`O4qNyv>u}Y>=dR|MJ-zmeK)kj(i<M
z)|FJ%)2&p#3CK|6ot1C>Bi_kS_CwRQRQ?^?S~Pr6v3!@nWqMHNFtk8HDd^lzeOv}s
z)10wI{8>?3id5`d7N|lEFyYm{Re@>zY@At;t~^>!wk|M@9*-o|$D5$(+@`=eK?F!u
zH>t3^ZGmd)rqs!H{sGlFq!3KjvweZ8yn6S@ksWLr7*_o<cJzUfAstFYD$N;_Lm!Tv
z3S5(uZ7wYbVrrQA&IPI+giPLCA-fc)9`EAIkDDO7#_E{tA}hPubjA#j-Q$b1z<=*i
zVCqZ1k(3uOxq#Gut?XG4b`rRGCO6-#?B%nh3Y)4<N*d}QCVTs8Nu-(mtPMu3eF9`E
zvzeFsz6OE*6E2ti{Iy07k<@{O+258@g6mrji1(nKJTP=8I?#3t;m%Cl@q_Q}mL|Rg
zpqUi%AnONt{jzd!^rJzpbdOmkhxjv^>**^6enJi{aAhgXUQGb&VcxsxeF+}QYsl;@
zho`S;aI{<R2!k(@|4nL<Bdr0h9d&kw(yB#{vZ?C<%*xSe2N);eP@_Ev3q>XuIVKIk
zSZ{i^t(RkCg5{a!QXGooY>^Gj%yhiBC)uXTiu#TdVvKHiUHzKVOY}rvAj95FZ$)=U
z@-`<K+LTU|lf4gWUf_TQYbh+WxNW?Ip`7CH?J_q-IHeBSQD#7-J~d&C#ur*OZMEI@
zC~z&7rzOlv&sZNS6;f-}%h%-Y`?%HCOHQ{frwNKwQMoJ0jsB43B%>6xx8d3qC_J8*
zrU6Ep#21$<FUa+h?}JbhXICrW5msP;?NxxdgTk1UW@4+AJ_UYBBo_^8R{I7R2tEZr
zEnhV4+vVM_zy$4XT10!}R~UuGjndx^Z&^h{6Jqz}G#eU)E}CGKESJ-5T74(<ogin#
zYKJkAF^y(L@kE_F5PvHJ5)M)IX$NSaPqh0)XoR4r3@Eh>GAspmcrBdN2B&3pm7cL1
z&{_|1u-Z)oW&)(qEb2pJHk!GI<-Udm0XpjBqhz>k4N98Inf7Z<*)%!JyJP33lX1U)
z84&}4`m!{NvawWT$jF5FOn}7eXqJphdrEFoiIt4@g(0;yyn)Af3n~)GgF*_Bvkly^
zMl$4Wt&H_CSsI#kVMy|jtQb6FvA<DjpezInW=sbQHZHKG3`K>oW#`za#~**{amSr3
z=h{-oAWz(QC;tu56Jo2*b3k`j)R=3Dc?+H&>sNk_>t{7z`Zr$Sm_>%dop}SZa6wZc
zKo;mio7a$v(nh#Jw#bCkA0wtoUg*t3Tf{{^6A}_5MizEY1tJ(<x%hv^WL#Phe3Z0Y
z64y|+78Vz|)UW~Rm1?Pa0JfIc26VKdeg)IR@VuACwt?W&lu4JBvS`=PNUtT7`9k(S
zI|m5yEt84>ILpDneo+|<Q!N((MJfTBDplS)NIh>@{*-EehW9YWP^&USYT{Ba1tW{l
zmGZ1ij6(xvpVsYm2(-!t19D8wJjuHOVN`jP%BkA)tR1=pqXKo&b$JDN9QBT?cW`zq
z!eEUtLyD`ad6kU8PvyWF__<mdynCayCPZE4)1=FwmQ+yGrUNkLy2aAy-^_w!NHybn
z@Qqn}#{40GF!jOZ(iBsN9t(0JL}WMs#_8?Vv!_fBpf@2gc2h>CM8{xHLzy!*)(Xc^
z+~(}JCB~|$udXSD<`-P<v_KSi{DBx;o_c1|psZYxdL{vygXDs!r)y`bqjgwQQ-fUP
zE%e^ULJ>nISI093``Av&HTKn1lyX9<-)rrHJK|WmE+&NpDxR4_t~Ufj@;a#Yyf3=e
z<F1E^V5!`Yo`IL(bqMtOkw1gxNG*|+QhyvoumL#@)tT=nj=HeH%Z-kz^qFFGJJ@H;
zl`P*)-j}Hp6*=YR=$|ePV7)d%0eDM+8mt+&1^+fc^QYE*M{D2X<<>OG)hfkPAAFmA
zK#$u-<&ADP@FV{@gCgV(|HcG8YPMF(ooV1r)E8i}>*X$=I*F6U*)snh)a5he?l^yt
z#^_eYWK2d_?uogbewy5ymX4#M<edBmkIj8CniYM7j$)16Z(El^lmsN-pLq+5TvnsI
zJP_E~Ge1{;9+;OY1t*8X67b{m@vM3QSUITF?MVJItt5FBU?y}7{FRLg4L2S8f1Su8
zuO4W4saL~_$xN{1Olxd4s_vNCY`V{LE_8Q+I5PtNsJfx@#!LfB15%XAgK3B^%4>=v
zz3*gJnyO6!cLe}U&*s?)B%n%o#sq?MY-f*9AN!RuH=3w&gdE>=X-m(R61Bn2FOqoy
z$;lPdWWF^(Zt><)_*`j**9AfUp)+G9Qgi>OLK65(>ImSjm4)6lQHZxD&Gw68IOyx3
zeF1_jk;QSAX|<GgU!mRqj5{V;a5I+p8=eRT%rYj$1pcMzjaeL#fCx*0+C-N5c$g1e
zb68JAmM0ji!yaxdT6S?6;5|?5idcS6IK{1u>Cl6n91-0+R1>=@_HhaXYPe;|>VnSm
zC}SdG;(z0Rvc~64JrX@LDR|*0rln3FvNo-i_HX4JWSvi+OtL4bLTmgk>tiR7<3g?F
zP%dvUOdNgAs6n#Ps4(q1^;J;NYzjbt1t%lPYTwuia9<C3Pc91LN?9Tgr{0-gg~kv#
z0VXbmT$-kTGEicFS7|T_%QkPeLj;Q_4sCS>v`8H`s}h>oAGLnf?YbOtGqWT<7Q62v
zEJF*8HSKhW5zkmIls_rw;bP=Z`AtA4RU)REn`Iv#H=JxPt7Wd8-=@{D(gqXSPM0Tw
zlwCMMeis)Q(!EtQ%acB=(HNX5XIF{Ha3z$d;=U)hjRCC=4TTc38kv;*bX<dUga>;4
z-X{S=A}kK@eZ`S@CN>3}4zETF>8Cr<vw_6Ji$}{JVnQ(PUPn^PD_sp#8jvr43^)+6
z(GC5yY!c5!H&BPk^F9NTB5Hlwv8gWv@PQVLLZ=nhF0dEleBerDDm!^8`s#_4NiTa}
z)SpqBgw!j4iZ8J52rRCYS5l)*(4111S|+askwwPn60m0QBXs<zpH?ET`4^m=8sQ$j
zUS2nj`{dIukvD9kz-VfoqBn!`Kq<IN-tup)wj2G_mdV@JH&V=J<(+sB{sOt<ux%UV
zU0*&0BI>>uP#vuI_j?X=0E^cIC=cCG8OuWWua&>p*wE!<z>B|cOCOwa2!0T-fPQ!q
zoooLZdkDIgDk*eh4?hfascWu<Nm4$F*<v&hE|9MW<!=c(vw#O&;R5;C=7Fu-2yNEi
z4Z39MG~yW5c>}kHG2T`F8T*Y}o$iU#_3j1-AfH%A!?TSgs8aZ;{ngRhl0jt8XMtdl
zG8tagmNeR*$2#la<u(uHxW6!n!=Tb4Uq&YbXzlwd@TXVLJo#7j2!7oF8?oMB`y#2W
z;Fd6i{+n1hY-9`}TL-fBtt}O6U^Oo!>5zZh07vqld}p7~Nupx1)(CD9DOA;6CYzBh
z3Oo0wv$l#>*iv3E-E%N(S*Qv}I5(>9a;rjD;FTeZCUQ|^D8p=B=ze?%%bg_K6uNQ|
zx<V$LMlhmm>(8vP9ZC8U*)BZ;_l_|cp0s`RG%izKE;|&eF0i~An$8{VcPyhyb}H;-
z94;P%oa|ianlAS?+NCg5F*tdM^DmcOt$AHb_Oe>p&6b{r{P4RMdPK8Q3sOTTJ+g<t
z;SEg^#R5`xS?(F@r$9xI?s|w+ds&}_s6Vl{gHvxUzg>(DG0wkT_K7i71*Jt2uyh8;
z?dxCizNaaMY#mi;9GOz@ehI8}Y%v?}{sFF%sn~`CY@N7MGR<<JHwQ-tA_zS`bZ9^D
zXAt`E3{~78St<v`+zaJkgXGyr1)*Z?A*l;yWuf9~t{m#1hm>0{hxuZls6Yb}1@gzt
z337OB0?shfLqJA8!p0d`G)|5T=+f2+w4Db5PN|C=WsSx{t9x`Tm+TL)h&~j@7!H7W
zE0JRhRb371W?6MTGaaJ<Z+wD<@jAl&F2|=S(tt?Xz&)k``h;#fL8as-HB3MwBPZG#
z8Doj+9|mH`sk3qZPPQRBkf4Z#QJUx&b4sCV$xpseP7PEnWBgrKdiaDW$kCx43Av<a
zJcICsOgVVAUV0U(=+l)aHOCt<tb^uw3cR^KUr0>gF{m5jL8TXFZ=3G`ChC?vg9{uI
z%{l^JJl~*t%7hE0AP`1pUT9~#q@JFOjnXGZ)$5!=DeclXjS3|!AV+Nhd8V1t&s)3I
z?a^z%yZ2ATg!eBrQGh837m&kddRk0w7>pNS!Ex!G9y1z*XazYVb!rSSLPw;?pD`tY
zMW-x~0fAqLd_bakXru=k3=zd&jWdbW3^HWt(F2jwgAD>T6KiBhI%ZaR3Th`^7EQQ}
z@E96<Sl+05*I_XuNOZ81Vnf<wcpS8e@T)sBH2_|5otW}j{)~DyMl4ep8Ijt-O$S4=
zuGC2D2;vh~Fv=TfB7R^CQwWcC@bf~Zp+*a&X@;FKws%(ZyI#)rhO<Gj=nw>of-z{e
zAM4!$c#vL@<r17ku_HCD0SjeZkk)es$~mzeJqvRR<=pfw<G67~0HE<{ns)gyg7&<4
z1`-hY`_4~*UeC&r03Yn<5Hcfk;RUt;hT76@dSRifqQj9Eu5{%xA*O@y^?JD|Ez%4f
zt01qNviV}405nV*T8)G}a!Hy1!`D6Yg;G8fk1loK!{sX@C4o?^oX5ytDK(tQ?_D5e
z-aA9?d5%}WkY64%+k7jS{R(S55<F^U+z@0Jk@8V>8CB6e@&+Lq8ML7~){At>%_;h$
zCbpusStk0Z5ImSH1{}v~ndECC?@Z@<=2$TkHx(~4r8eN8N6&(H>td#7Qx9LCTFiq{
zQNNW;I^<nWL}q$w@Ha$c)iVORrdhl!Rs<K{Yz@erG}?wj7s&b+lJ$XHuMAdurZm}9
zP@8E|bAvS7225sHfJ{z%tKrY?I?Xpz{0)UPioTiz0EE0<rW&-SwSe7f@m3aw#X+gr
zfbB-6rHL?SmwJB5<^Ier1iO5$uszH}XxNnwW<(o7IyC1KJ;msfDdw;8!3YXQR3SzU
zh)9j|)d>h%%j=q$yHO9Y+7#&Nr+01I2Q3W>!n#SWORc*yz?%d8^)X+NUJCiMfeSy3
z9+Vl!6Zo>-;5|@+qBYa>Rx}9yD7}HlHVH&h<o-ClQKAi<7?WWl>#Y1F=v+Av=SCln
zDk4Vjtpf9KQ%nI84I|T;D(Pl_HV0>&jwS6lEVuX&lr<1y24o_Kvlep4Pi+fP@j4^7
zdMm24>AJ;=>V#^$+~#k{o-&4h3B*mJ?o7Gez8^REym4|z%*J>pDmvuOAm<<}6^*5G
zm+vg>AS(AS1ZiM@^m&C}mE7%p@X7{*jazN0+!K9FY=$prR_-+{21NpAD))UKeAK*f
zZID6k5AcmdddHsyy{^PsE93#2_lOc?CiuCJNrNQIM9qNP(l5MMA}H&ou2lsj&_aHh
zU}1J6di|62S|q=+Auunf<FoSXv=q%%2UcOew5D#E582L&A@1}z&y<eJsDGUiH)^Ex
zpJ@jqr?5aC^nT1zsx%m2q_uTc+60n)GbzmW*0i-UWjlB|yrG3|j)ADMwiz-0bA4|h
zd;?2N3BW>_<aQQ-5VRSiHEMv&3z9~$MCQjtH8=$E2Mc1u&PT`q+#(jn-(@A)n)-|^
zvQbAenVc*RFoAJpQFpLp(r(X`&6@}#54;*=iH%;|B1>&_CCM9PnSW3JF1pcdg#X_1
z*!km*J5E*<cB<oFHx-wyo|IIbz%<1P(0R42jD?ldAy9EuAS<W`WAv_rf=l@V{kcI_
zr}wPQsj$Wbv`#0uhh?(HknGCiFTPyX1`I0c-hp^o+srdd%4EI0q9>}leQfY<!_kif
zS!I2vaT|Tb)P#~4A(ygj@^7s4mrYX1D)y1w_Cx*+&l>TS$oO#T2f05neb7;P#QVXw
zYS^O6$yC#nh33d+M<wkQ-_YT})aD%l^RsZ;9`&wi=pbju5Sqtq#l!HrGoddnu$&Y9
zX{mkwO&o*r61ea17(AZtbI6il4B#G~mEQ&<TiIM&iTe($$`i2~${koC;&kM9amjDK
z4fGKy#qi$;<w<YJH9`#I=E_q+9w2*xd41Z!p#CTh06P#=QqPs&NAoO_Q8majaRSC*
z4=HUuYiosBU;f|}7By$&k2W$YG@1V|jhD~)Y)HLXaST_zJRisc?FGs#$_ugFESMwc
zd9d6U15@FM)FdxOC-mt8Kp|hs%MRq(;M*BKlM;l>pRD`A7TuARS7Nwf&2TZRmsf4}
znMoP>bL<DAKP$>f)?SN!9yzE;UXKMaSR?}@HTdW^3Olu4pLx0RrlZKRMip-x{7P9U
zZ>HHX6xBE>L)Eu@Eoy$pyqU1&ZG$(&eMFGe%R8}L@YK}D62sq3TSnE41-5n9#=jTu
zy9_PBUpm**c)3LW5;&=LN9;zL>iY(ce3XCrz!s0<rR5p<Yal@ls!o^6hX%!#dQTAX
zk!?JpAW<FH2ajXpM;a^V!Ja1{2T++a>+ilgR-~h(!qxr5J79!A+-0EJp@tcEyRKn_
z{4)WK*;<wTl1~CFS(3K~IP+<GfMB_)hz9#C;0O*%jW+OR`aJbSr*2ReOz#UPq1|q3
zj7z2Se<z_|1~OGP*H?jW`6_Ui9)M7_!Z!ae1L&AxW8~|2OVhno%K9ccU?CJ}IllD?
z*=-3f8>H*s23w?jZ;|g}8_q$NzDAKgs*j>cOhZYw=;Io*Dasck00JuHE&HfutE2VG
zqhzZ-rh%h|Z-Xv~g1wc5B>%c~A63>hXQ5i_W2!Y?r1XYh%=)%{+#?V+22iH$`j}En
z)lI-&$SS%Wi)4F$Lk5HD5vC?pK~Z){-A-3VmNvj0Q|s>0He^6{@;4TENdp~OGT!|=
zr?EF8LI;EmC~Rfrc8N(s+=FF&SA$VU>j*@R0cN}PaRp)-ve(M)sT(!rQE%oq&mKNe
z9!<;nF-P_c_)?ODPO^rsDto0Bt+LAtOyu4+=;k|sy0^omwof0`B96_#A4&F&2@Xf$
z>PFem-gUJ0f_LNoF)Gg~piYVRfB=GGA?HI+9(rjX=x;2$0G){tM-2R)FF%Me#-cWD
zg&Y*i#AZ@HWpJ%tnR0NN{YKvVstGtGJ)@bPrn60Qs6U&fuAR0AhoxtXb!dZmXo(!&
zN7M6TM`-rvPiC^}o<t5zwoy8xgmQ%8C(J-uInsONbw$+zf|KPa$64d(I5K_YERhm(
zkQ^NwirZ|8923x_FZHqZ3JNuJD9812vA-$<XK*c)h%<Y<gT_qq)KXB`B};RH56sC#
zfE8ejn+yAb8}~#zEYn~@IjK+QPU<RKl118jrQ0X>QAHbUz>PJNOHj(>lmsn0$Lgt*
zQ~%4pEQQYrhg%>$Vgp$@1gZL>p$eu}9@F%)y==jw&lx*#u;lm{gjJiLhuoM5LVMt0
z-zdGK5nQf$XmEu}o>P(t)yWT{F?fg+*a&qc6LFjhyMUtmu#$umd;;@vlWFP`t8{N6
zifMffQ2mdXApQI`Y%AoufiiQYf1ge!Tces2Ts7h=<urRTu<Oa`*2T%kpCo5EDC~Yw
zK@e1jxnGOCbE{))jNbsqj+;7e?6^q=+EU<qgF4@X5=>3ZR{>Uu%V6(_TR|Nz;~s|u
zL?+eB&_qTPJ59+Q?1(yY$}q<}x-cNeo~@>%bwC9QcGgz3$eC%PtD)q8cMgn?XT@yn
zh$4)I9zkn>;3N8|(zmfnMn+E{K2S#DV2|>iSWCd9;!QHzd$QTHQWf(;BYm8V@p*SM
zx?uC*Cx8QmJg;(gpfR)5g82l8Gd6}CQVU*;)xL|<<SE13l|3%@<WXes01(=gM8SwB
z=Xh%?qcRIu`Vu+UTf5N^nl6^iWx=iSf%QeG;|=e=^s!wd3=ls*!JethOheG!5%U6r
z*4a&RVGxD}BrcZ;K^V%AZ2O{ERSVcGBnxYji@n<r1<lPNZLeGs+d~Aw3hC9Wr%W!5
z34y64&7#4)TdiR!F~mX5`K*-M??@unC}nAzpv=)z58e>Ujm1))Hg%Ck{<7F|MH-)B
zt=QT}X_z6qRO!ez6TI(gsS4QQH1e*r?CMl(RQo&)WU=jCi9Kr!_8G5THZe`FD=vWI
z`v#e04XM3sgmhOMV9r8%mAd$A2;vW={(EB@m)xB@&%}NsSq`ga1pW;~hFwJ9!<>1U
z^*5GWXq_}VWB_@EGHOOiHa2$D(1FsF1~dyYuOG4}EEIqk9BED+;Of8|)@hy`SO)75
zwd(Vs?wjKMd@p!K<E~5zJT*bbE-EUCsm5F=Ew;ry3-fx)v_2*tI2}>1Ivxb7H#OxI
zP-dlEZeKLhz;lK7YoMYIO3hr*6I|&t!fJpnCgAgGrp_PwZn?^Osi=ns{?);!RKcA&
zay4BOIIqNm(YM#e6Pi|0b*@6LOT(MP3e|5qdAZ)-P}nfXR!g6{Kz<nT@0p(?Hw3p;
z+k(gt#O%i3Q1c*fl>FFX%?li+LacAK{3K9j*jU^)H`-s-RG4JtCi@<>;b<nz%FX^g
z;YVhZWd9b2GLIwU*K|i9`A}-0bBDrOernT%!3J1$s{yN{_2LO~n}2SJfGl6aTqJN|
zAS+#iZIIi2TYD7rkUM-5d{2K+I!)c_rzhJ0CKDmM-efM+=9FRPLxXvD@E=IA4W~&I
zOr}hHkKeqFRm~cEd9U4c2%whxQYX}9abXMP{+Rgy1n^o@sKDicgdIHa?m-Xe>YV^i
z@#krgYry&uZ#*jz+9u1H{33P>_i#B_+h3*$YygR#qXgzxv0IJJ4Ndav_@WqAHYny>
zyPXSqMpNrRb@6JMZr_#FG{Un`X4rSQ4b822rhgY#0<$KWX7fRR!_g_fuHu>?;3F%a
zzhM$V=y1XJlJHX1H#^OOw9D9CJL_|NMH-lk-oWq4Tz_VmV?som%@(z+Sm0uwx1ggw
z3pOu<ws(H8RmiJa1Ix24@J4nh8Pl~=7RGQm=s5K11iQ#~Ew^_sSsZ<gMH<g35J%d*
zy+(v#n`VM;FA0FxBD_JCMvo8;Xmx;=UzWw@!{eB7rw|PYA{E<UM3v>SNr>&PDQ%P$
z-p3p;yD%x}tXA4H`ZI$$HY>|2!)XdTB`AQq+CMXc2T$8dSrc8cJPGg|8DeFvFA(Hl
zU5q?7Q&$hhXT7bZ7U5V>cWtn@CFQV5!GbIsy)!E9!FDNgxXITBi5T1lhxVjj*an8(
zKNRPop-LY1Z8s&JG4$#;Bac|C8pa!MHU{DRV(EzXdZ(*f7f=GWS9vs0%jFCtYz165
zU^gC%8;ef$S?TmC1N&ewV&3#s@|$>uji7%r5>U$H(IKtq=*s9iUFm=8ttl$eLy;Wu
z6S3v=?~&g{f10e-?y)DmKXb&Pa}NCqc+Ec*AkQKc6FJ|XP5{)BClv5#TBaKem^D<z
z2=ZMFwfHj$lW6a>s3aZ>%Z9pMp7p7yHxQQrGt?gpwr5qKBKIHdJ>oZ+gf*kP`dlAR
zvcihY@Zw<tRmfe>+ZuT<;|_koa0b5X@}k{Jj7CB{8A;?zu~-d6^Q#k_O}o78M4e9J
z@+WWB&3F%tgeR@Mk|4MVY@3pGd9_ccr8HFEuLi<O2m7<5N<C2JwE#!AionEs5@T{g
zug9(;KPB};Gz1z`_C}m+9w%^^@}@0!gavB9<sVYOhszyN1V0%s@OFSj<A~l#6Hle;
z<z0Uk89iX$m-k{KVDk|)0pS7pXw=o`F9xA8BZdziB<}}a^(;gQ!Vlu;Olgw81}iWY
zc{DVK^@j<Ms*wq9l6+*Xhm9LIM*fzjPYkEvr<Wxl{jpCDP~bfSztA@Mdx9dSs!TIE
zFP4+yky<SOh*=G1C6RvyKK9HjluvvhNO`J9VgX^px=(E&x|&r%N31a+pV_B%P_j&w
z&*M{I027&#(X|`pi}<v$;xhF=0x9}3%~7j4;=X1antNoy%JP-3i!Mi+i5$rHFM|kl
zc}?=Qzej;A<<4N=k#7RN$glBj%*J_06E8MEK>D|TNhJwGSK+}T-v#z$;67f3i&*-q
zUT)avb7Twqbp)8BE&Hlqhzbc?^;P{6k~lOp%GQ0|DGr2P-ALf4xs81_d|a_?8?S-F
zyh=HV?fe;a*HkW1Al<%iC@^p^=%uiOH_|&6uw;JW9s8OJ&<sM|)P1s3-%fS*BgB}T
zmf@|YX;#l3*}1RkE)i8iSIw;K(l_ZASq^Den{bcZ9NEo#qUy;LFK~1~OSxp(y{{@b
z2Q_9IWRF;!uAO?Z*|MkgoT&v~K%(6%nvjB^3$AtVc&0wV8dD_u#QICh7+9E<efz4O
zg0aJ;77epse5Uxq1Gi50@2mPn;QksYuMY6J6Ee)-pt3mb1)2e913oapPtEnJuh)vs
z@&oVNf|6K1WVsv^+g#pKF9#a{J6fkyqYz|P4)LydJ^*T01|p?{9O`dicEX`F&l0dx
zV3ZE)S^!8FC@ZFPxGkWg6$$op<OpwpLqx+Bg%3H>01g91R*v!oT5)O91X%!rsT7X3
z-ZJ$Xu6Inpv}nwj!9^pA<yeOn<Sh7X9G4*6RMDUnw2$}U*)c1#t$>!|1Yb^2VkWmV
zi`t3NJyK!SqKuZDlxDvUEF$-Mv7DUHNrgj-V0dFXaI8{J@$OamYTOZsI#bi1+U-lG
z6xAwu>Ya;g8$_Z<>_|EAr5bU$fzowIX;vn^d@{yo@)RtVoU}g6HceY)u0K=u4@}^6
z(AC~Pw0T3r%M*ebU}8Vvn~>MHbDAn3|2Lv2{!jXp{J_o9rg|xem5;7uKAOV#(}aXR
zzUlA^1qtcfZ97^1m!=lGUi$?gz~IyLMP+#ZzM*L?Ltb^gu}|v;SBh)opiVaw?a4~R
zXC(AzEhA&|xCaZQD0NoHusQ}L5Ae>wSu4h%B)SYVT2$Dhw+ymh^LqD`!S-uTVJ{hy
z5TnbXpVocSa~1|2YO@(rjU1uFd;s1qWYd|&Wq87SOS%pXp+OUu8^Cv#eSXP=OD4#O
z0Oddw=PQwsfvmVm>AkrYJfw1xGRme`g`AzE0|3hBk|5In_8_k2F)<k&KbBD-19`TM
zz{2$OMHm}Mec=iCcajvR5$Tq$h>xwI_usf^SqY~r_^zB|Ye0zv++gR%LGPeVdAv<-
zbXl347hfUM6>A-xZ?DFV9xoTfULf@V*$m82F6^4Ix*Z}SQzpc-4mbh<gTP|TMZP6i
zt()yB9GVG!24tp-ZP+nTq-GFaE0@Hws8a)y&=O_Nu$Ox8L>aQ~3^*@I5Vb1_<^c50
zI%TDa1eG+xrk|BETMq*BrQCjP0C!U19grmP5&|9o7eL>nG7YDm&>9y}m4>6q+p{{U
zpi*sM?LdO>@)D^DSb}w~sceyn4)oao`U060Z3i`$Op)67tGJ{VYC8uP%kjmoVs6kJ
zn^aLBr+XR_4rY8h<QAyJp?9mXy$uPGRkDY@;Ie>6B?`~Nn_3!ueJ1H+OW}{6?b~?(
zs26hG{^x(v6q5zxS}4u&*XG-yUZ0$H+V;n!U#)}=5(`ycOucC2Aem~PVpR))>Mg#I
z?(Q-wJjyit2o^(V#V(IC*U-Xh@mF{s)Va@*D--fTjzPa97*|EVicE+(1*5@)L)XgH
zebqusf6baIrbVNCbB&#ETv!OO_3pxllmRu@#rA`JV3Y&H`wZUK+f?8ONw>KlMmu+d
z#%%|rV#1XhqF0@Ynvdj1F>&00w4j1TlOG2`tVFhJ>=q90PhtZHz^+~i{^`a<?5GJR
z%|PuOux8Zzvs%ba0q=3cN6O7<O3Y{)f;o-8w*;;tT8;teKMmA@_eBZpB;|5znqa$8
zN~phXvm=yksFmCMcB-!vj*jX(SeCJyLKJwp2Fe<{4({-2bPEgBt(qZs#`2nwV2MhI
zyL@@z$Jwp=Mo|2q(vwkccgOOIf!+7`6pep?R16ArZ@>p|rBfmbA`wbTt!Os9FBUo$
zCVd17X662P0wEM0uHfZ<7W;q#C(P;ztMLPAC~6QGi?>F89?f7hgVKij_h0zv-IQJs
z^tE0!slWWvyRr1-6j@Pu!=V!t4#XUIL)G}N4fEm3z%&8U8cTy?0(=><C|=8S@-*5t
zJ$8&#OBcOXW*Bm384Nr#y*DWfHFf2|SUk=I$|k_$WmaHDQEesNe`cGQbIB1h$0w@D
z4?0LAuu9!B*Uu!EwOHEHf>PZm%m&JAnddXWQuV4>gCMN={*2DnxFm34w8?^a22BYg
zMwy#@VVb=bpH=VNMYg&jsOTt*4Kt7?3YllyqcP5hW)(#TOMF`D*o#gv5CF#vgiHIn
zX&PP~brn&pZJ8~q5?Y{@jj}v_-4y|2cK8(rpv$td(nmGkRk?9lWmAOKzfxAG_w(r;
zOk0(#iD%F_P+5+A!LruJ(y%<>u$331^WD0jh46v`cd$M|T7?cnG;$%bA<hx}TGdpb
zmJI6Vd4(eBo^`}DRgIusn{2IGULq?G*)eadnJ5o?Kel7G^ISIlG^jmz#HY{F)b#z|
zh)Hga>4OA-D}pRYn>~$X9f4agZdO1o^JsjnxMc1RkM%XH4dsjP+Y{*A$M85rQA=mw
z?dDrSoo>a={hM?;$OV#9r<q@Gb`AG<JOB-_(*i{2p5OXR%^YWDMNLO8Poy`jM%`8A
zalQP``?37orX+qc)&Y08pj1x<X@lsc;gC-|D^gP}zxVdW4J)ybXZ#se$GFzj*Yeq*
zFawaRu>v*5O5_hdG203-49c&s#!dXk=m0oRlEBc@=lmIfsUAmkKYKp52Q8FLU^wz=
z3*-fFZQ`=UrM#GWMNk;P26r%dDV{m)XH@>nvE1?sNarX}|0hE!cpzrri2I7YuF(Qn
zNN6Un`YgWhddi5hKL?3FtEN_7i$8J@#`n5Gu&_t2yx~)zW`l7k<bdCd5hqqaG=3{h
za<SWF+U0HQAQ$Nt-m%6&lo_yzzw5-K0ofYlJ^vDx4Lzn1Kjkk8yGUD2y{|e1$L#lG
z>`7q$<O5$E!OExsWZVa(TmeJ!UsK=eJq`fQmk$G%S<DsN+#(-&D`1A7fz|T2m;nd{
zOA?}DhTrJaCr_7;{k^3~(wI1ID>j#Sy;%NkeRi~#*EG~JS{66TKVs7wE2e;MR>(hn
z;NU7*^5c^jxVwFWMo;h^2n>=>6F$-Bk`CPTacYNY=d)NCs9#+JBcDFE4Jbx<<`-!w
z`nBT*gR11q*c8MbFOaX&z+{nNSONEKh6wV;>*QYn^RYGXnETqN@jY%5%<*jb#z(>-
zaF3gf@~uBpYZ4YR8<vxQ2XP%yL3hIMqI;z8XcFmXYCW?Jk?&e4Tlh2Qd(?+w%YLe+
zLJU3}h%4y}xm7<?#nEg=QoN3rH*bENCOSH8-Opuev%|2C&o<F9*hhxoGuQIA{Y<k(
zPt!TFT|d>jAzN^ZY#$vU84koo+IDv6=jyx>aPlva9s6~%pavbUARSJTo%*@!pxeP&
zpb8Wo?9<NvMr(iI85t+b6t{~nqg#Fpexo-EJvVoaK0&`i0jBR|w|=TV9Ws1`?4Dq;
z3G5vsUOZtj20;o8*~8y+U?zkh_eM^V5phev+V9!V;(;qGVa43b8-G6$T+%no-u|A~
zF*qenG;LAs7TC0pbv7b11#VgU_ET+c!(`bn&7sSQ0Utw-HFd)NzR>Quk5smm18nah
z4pJx1NT~z+*=6QjGNwSvO!YQMmMRMtVO2j!xJ90xtnW-Y$e&T9VIeenp&Z-=Vwm48
zaiBUh$RV**U>DUIG()o==6PWT<3rO3tm_85Bc$lV{27(G^gUz{=;8fb>x?WC$cI`k
zN5qhN_!)Y}^f>gNKQc{Fix6O#HaRMuq3{qb&-f`vr(vjfodYkAHF8W0GXzo%<EoGK
zMs#%LrBBP&asKRfT-SF1Pws&7@%Y&CF5wthvYKutRPD<N{oLBst#&j-yc2yDa~}Pk
zgA^{ClY9`;6*JqemXqylRb-)%I>p};lhwxrbm>$(iOlGcsgfT3oOvV%jr32=n&G2o
z9KePedMNgaPar|(NKUL43>zR%Te0NYpwK5zmEM72U5W{9wLtP>pMgVA4rmw=EzWj|
zkQUJ2Q&Y^8Cz@*;co0Q63*x<cec8nXHw|1R#HKbWjCqb_k%vC9h)JL+$QGEUWmZ(4
zeM%v7q+i04E<=%VeErjef$&%aET#~MnBPic<>fR7h56N(Ky<nt#j}d!jF<o<tjts?
z3b=s_Ay-c<8DLGC>tL$^;TxEaV6_+H92B@V5O+{5ls|(52B0W(YWKw`gUlX>#D<y!
zWV5+Q42_AZ`2g;Q&~gp)XL@~Laf@++hR2DAfkCZ`*oVk=P29HAeP#kB%Fn3$PDdj;
z6`vIV<AK8Lm}$jguk`AZ5&oVCX_Bp$FH5TEk>0bBX3!lq)X*(sRG>9-L5z;e1gCzv
zj4>FN!z+e<CualxTm0{31^PeHXaB#^M#iR1x2n|4#Z>GY(9P=++r%I{3Uuu<F22FB
zrY{gKeK{v?Xd1Jib^xND8_npcf*ocI`1rsl6@~4#IWJ)fi?Cr=fwR0F*UI^U>hMqo
zhXTK<3(_#S=q;w{Lhq6vGLH*TP$qbn-K}Comv*@*CI#|AZy#LGa&bI^&V%V0*28D+
zk~mYW<U#Y{rN&^uILGA6Q>3ID7D6rOf+-!5N~1e)h-8Xkw3o89Mw;QMY1W+lu#A>R
z>qjs!fJv=Lg|QPg6HBDhU$f3Oa;!trTq;%GnF{A*L{(N@2P;i=kl6FilbUYFh%)gy
zb(ytXlsGgp(R-#>Ass_eU{fZ=WN^Kbeu1J%Z4isiRuQnB|CKtQg9&MbBMI<K^jmL`
zV;Tna43tbjt~{e|6b%U$NYMzyS_1J#m9=tNEQ|s?F@dbTiq(y2Vl=T>e0bKF8`P;n
z^0_8kc^&>qvlA$jc2iCc>IZRAMCRAYf#qt5QJn$qZqyq<_cS3)u^}^!RVDS2a(t@M
z?XekYNsHk0^;{5XL0qQ=(q>V<4y<a6T<()MSShN=6>$yK(~b4YXn|b<(+C~;I=QN!
z$J$W$24^7Udan+&gdPV`N9A&juhlSJC3~J#Rj`4a8W82$_zd|3wF1#~2CnAH%B);(
zgM+7o10?el;q3k}_6!m^#kCs@)(CO|tbY`Lj4CdYAE&Mx8X0sXKQS29mdcGzf~iMa
zBR2(Huwus9SIEuQPbSNBUbpn?G;J%cPPNi$NHY^OF}Wn~LI%&0t!<N^Mpqg}HyM)5
ztqwp4tk^)HBgoWk*2=8wT15QzgkQ{mLHvrCu{->Go>qhnU=Ew)PDc>lFytsHsGw(+
z0Kt#RUGW}-hvL8jz0Hg(SqCQS?%3j?Fc!-_0cD6Q>Iwvf-o5FLT7?b2FRg)5NIE;Y
z-zdTGfpEt3$(zS8x%{a6_&|W87}*M|s!M9*=dpdGo0}StisTmwTPeDdd8Q?S`C)z;
zy{aVvso$|Y6GVQM&Ib*q?gOc2JUf!V_UYO&HTJ#Ld((@a{$Bty^=6A?y7z{2h7AH;
z!F8@T2Y!dRC^HNf&Vfi-D>J=oGZ2w++8_@m=;%eoU`3f_@WbJNN=9bKq6ajiRz;)C
z3Cx6OUc$hFxp8@bb*Ai!9cVN7(zr3tSOi-3X)~E0jp?1p+IHo#z^7n26nNU(Wnn@B
z^|!_~z=UyH?wHb;EJ_doZ2<bPmGOX!13lmom63Muo6Hl9DoX&9C9xnxu1?BKmZjbo
z_s^1jYgKP(otFg$V6vLiRhGx%OW}}A?~N5{qIS^}r!B0@Fl}W_0Qnf0Pkb?8zREx{
zXz-Atb4HAl)zK&FpjV-~l&tY@tZp;8w+*t^C<|<HC9=-nJHo+PX1c7mXY>Ot%b=zJ
zBiuH`6$4EzL)4++#*8;6&{>cSW0yeram{T?{W4gJs5f06^7Wa_Xd|jv5BfnH?!yUq
zRBxHdOrsR#k=O{>yGv!W_tPEW)i8r?Ma6pQNbkYq>rKQ!V_Y)WTqe_eG`+WLmGW+P
zXSB!Sy-GI90il=Vdn%p&-Yzdos^zviXgHw+_V+wuAOJuz(lEMv@*-a>kJ~{iF3U6^
zM`GL|-6S`8!mkX@nk*gF3Cg8l9$2H_xW9{i!l#vuxTLU8PdeOrE+f6bQ?}pC1R+n`
z?+CoFkl%Y-Y5}HW6h_}X)AdXfA+Ykov+3D-Ed)iUr9b#y*@YRlhQ-tXduzg4U9us5
zKWE1TmUcMmArR;JpqWMTf_DVdxzZa(UsAT8Cod*+U>YKz2u%~}s`gT#!vI9H$;;6W
zr(Ky?RWj44^(O-~O^J>2iv0ms1gIEajUI-UKrsGuAh#y5LUJ$lPJPW6LbZzpIkm~_
z(Xs|&935QMc>9Lkr_;*h&F_JQJO=|6QD;ZqvL)yD$dkA2%uYcqRYvf>6X@o#ovfH?
zyr5C>cWvSwt<#zt<vpL2$5Me<A(Z>S42sNsdZOCoeQzCZQq&iI5c>n%B1X<bVB22<
z9L@EhxF328QXM8fVLZ}D0c5C1p^U?MHjCC;`CEL80<m@SaiFti#lmr=V&(7YV?Ans
z2NNGN#o9j%Du9Fh(|#?0hs`H3MHm*SR9S+n{?k|rI43xvDVP0Ov;%|6r7*AE=LRjX
zO*jo-_`rAtY{q$2jAWYVFJnqd?q^&6AN3mfDs@T&6I1dmk$=S&F#Y`3-hg|=ELC*B
z|Hk)F`BJ+*c(-cQT^l&gZ(}&jlp$I4;ozpU$-fi$SkUOZ)VG43tc*y1)w04BuTZw=
zuR2szoW%h~PNOaRt3nuQ$TS69Lk8F4R{c%Y3)>J08|_Y5h9*?PT<6Nx_6=~1S+?j~
zwnDb)uiDW;gU^#~`-eNH^GV55XuI^G)sm*0>h|8cn<JVc?>gDRr;&^R8gg$gn@s$^
zWB*QORct_dnsV8xzv-2cxDIUA&S@ZXV?;2g>=G+SLJcN?+|@p>0qQi#ZvFp9a~%L4
z$#!<{@6MYQ_29x-y>kzrJwZVPS^$}bG<u=z=~GZ8Kfs5ZP4@C<bg^QTAAJ90Z<|JO
z6MVd7pZ==89oQoK+Mac^BCYLywl=tDkU;MrZw3vM0}}8eY&`0h9+-x2=TNYpE|nj|
z#^6}40gf?akV*Ro_3xYvQ2yW1SIEJRI|w>E#Guj9S`0<mp|N9dfkJ*?Ijn!@9y-gh
zNSCjrdO<x|-QsWq$>#eYHyqKmBWe*O|8E($k93eQIEV823TTs#it)w_I8ToDZte&B
zaZEIZ5`%s>C>tfm_75E?;*Xl^NSnZM9%r*e;8mjvAjc=jFH}Bp9;^x{q)Cw{ruKn}
z0m&1e=&cPUs5IClCwc2&-H5~V$6F*P#}49NprF*1YVs8CC%j`QrJmZ~%_q?CBDSc9
zp%LOL7?ayG{z3}gP14Jjj_QviBqt5Oj@%fid{jZp^+U#lSDU1FOdX_wTpNi|o^53w
zY@7Lhe9cfmCWgvEnhPu#uE0NM=5pTHOnFo2&t}8QINhP%$EQODzk6#t?E0qlAw5A>
z`q}K@XI%|ezkdQ1Qww&tBAk}yZB=m=VJ)ZoWJ%$}@<5vf{C!nWKO^3QiJ&V75?rEB
z8s}+aY?C5e_c%CO4)E@^P8Qa{YFPvASx4(&1ah#v3^riU_ohM77bRc(ndLr2c`jhc
z5MKx{Q%ViI0RczIB|}qd)nVhdm@C85Go^gWHV+RtWi>GV%!KpJa0|7G7xu`P5&kLH
zMROgi{EiHihuvZZltNStt&&mE!^CFPX*T?)6+v0EF+n8V87eDhrxjSl5`U_}T*ms4
zM&fDEX_Ml#f(@Wg#=(rU0Z~>fgR(CDJ9khrP}Z2ya@*&oUYIkL8G%;G_`p60k<1AL
zKWSdy=#l1zb~(>_oCsk$Bj<Yu#)|?KaW7q9fT^u0l?(kf6XD_Rg-U#eOz@*H2lDSM
zxhM_>jW}|#kFA0Xc8rC2OsT-^FLUG)KP1T1cxh_N1l8t}t<hun0QE?v#G0OP+zC<|
z=Rp0U)J;xTv{A}zejTlu(rg8)ywhY?9=k9a8k?GW9Eb|*YCL|e{Qi1*!%3BQZ!j=^
zhYY}CsZNXA2wi4P`gWxrN8ZE}1HoYVhu^@Y7#osmDcB5zcl$?-Y*~tBGW%OyJi`Ga
zEt>(=alJi*bssfa7_N&uF%w4*j25XTO{%kCe`632eSjLGA5H2@ZfcQS=Fh--Lg1jF
zyIdO60+9~^SVHP-*p-b<P_dYZZ{i+v`azc{O$mQDGD-lnmo83l$&%(k#X1znVV<qY
zKCn%b2-hZ4yn7yCoGW~eoF-T<i8}oyr`pEk7NAOVTI)KDq!GUuuZsbhmb#)~2pu4}
zz-|Uyxjb&nvc{GM)P}hta0ZHb_an!eAqUBo&hSt+p=$dY2!&+SwNw=tHiOJnX*uQ=
zyBeE)wOthrHDI}~@m|1nL2~*Ze#*5m3&mUtqm;7ff^(hC7jy&RC*bq1w^NCVEfBGO
zn7~e>rD`73Rb-~zU{gS*sB`5<hMgU)P~um~j|~%>E1EJ;PRmdFyHf(vL^YN)!x}6%
z#$;d!VBUPB$5g()O>VNqYBe5><OAZ~Y$IW<jdDwTiElg;sb54t?E;f(yAjAQw|Z+E
z!PG^Tp#C;*%{z(S12YjXa=XnN^&T_LRn>Aw9H?x4$)pOoGY*uM{K35L@;=m!jE4fG
z|L*k6^gi^Xy(hk`WkfCm9Pf?Mv1Qou|Bt=z46m~~)}7ExLJgt!-rWnuwyJG#!!nov
zfoy3@mavM^mTb#F0tQk@LP?{8Tx83=H@O8Yqd;f@0)YU5Py>V#AOr|--*?K|@;T@J
zK7UT`J$}%)zi+Ksvu4ejKB@-YW;jI5EW>GT4<bUbLIzvk5k#rx!!U2L4l%8E?Q*C8
zZgvcAujZNSrOxAXSKy6dEWq_f_n54ZyN&X5nDm|myxnimUgX~R&?7<fl@;$xzi8DJ
zF2X6cHLj*HuvN^6l^|4$og8NRBxqs6gS-{>31wDHj4PEvPRt0M9auXD{X8|HzAYwH
z%QbrZ;w0N{&hgdrOZv1B*^ghWq#D`WT>sMeD#Hl13=j@dd-~EUgwV%nd_xWWGB2R0
z)IwhnE#~_dQNBs9eH<hvg@KM_f$d`srHB!yxiG$m-<xiloGgk}=uR;FmlqRV-;0Cb
zXa#MQCH^8*85N#pfLZaxcF58+1bt1|YB=4E{FcQ*Y@(@uc?`(XLpfRDkh6?584(S{
zjChFH@|Cu2<4|HdE314<teUMF&SZ7kI$X~R%RE_QhZBTdE^Ff(RjnZ6Lz-L{2mnA<
zalK=k&49>C7J&PFwyY1j8IOqU4K^^ujdIx-gG{Nc#eD_^Y?Jfro_%Dq>1%}<`CY)L
zxfp7ol>41Kn#@K4*$xKX_4y={2W&DVxYWxQA0_e{;2f9EfUKsbMz*GnVf-#7)abtO
zpdlO$fYrIv<s$)mZtX{{NPorxI=sj5uRLVChSLL^f;=4fff-l3%-|+uRwlZx7!>HI
z3;aBi=rDOga9QQ;qmKqDAWo#0Jm%9<M`j~Cwg7!RE>096f*rU?o=9}HhK?2(J{dSg
zd(0pG7pw$}B^D#J=E>L^YJ<?fvhtKc3A-9BNlzzavQUL(*8C}+!Fa?P0Mx*r@n@_!
z1RIa~BA*S;F}6~L^*<-PThAE81^W4aE&$hDN#60%%I0|=*El9)2^Yy<9CKU-VPe!q
zHy6CTkTA=#guMD$6N@($wg;B?7ky0f300>6|4V@&<LD5RXs5=%@>hSG<OYuy!Z17J
z<yhH-S_IPN<duZP3>qIWdh%}x2JUW(ECI*k4$hZXjnd%z)#|IS*`H{zpceht<KQz4
zv_;<VGZ4e%yKBL^#dtPF_nSfEW#^BSw-VH_EGiBb$=?HNdIKTBz8x34WxpVRP~M3#
zn8EOF!ZTS{6mVv*9%?t}{d~4NUO{`1mzkxKllRgl8(C9VSGeW<SQN}Y*%tXAX0Mr0
z-_WR5fDhwQLt_@=HVp&+w1Yje0UcsLvi~kDC^}a@4nRv_j{U@rN<mp!$=Gr7FF%c}
zioZra?FNtbURV1zFQ|VfGGI_4(q>T^=CjyHh77?Ql#|cn7($w=moEaX2o+{P9l`?T
zKXD8*ENh5H;a{dHNtt1Qz&#^h`RDVXm#`-CI=IijPGD>V7__U*78%q<s~qrz9%kPd
zWZ|YNX7E%a)?hn-#xn5;Eg}`j_JcgQ6!rm6x54D-vIjUfK+fL5zM=aXW^)u7f<Jx7
zL2isIVwER$%HGLGw#Y07qs)?>{TVZ-7}mQ0PTO4usjae4zy7l8AgeS;6rwaRjVu6j
zrm{@64w&$b>x4~>Slx2BL7{y%V2Mb)+1+R0(dTu>1cz_>Gh-<Xe^@Es@-^Zxf_2ku
zpt}${z3eeaZGWR`WY1U!EyH*dwpZC}klHudJ_$#3Z$mP2I#$%mKK?qT4)Px6qVGFM
zO}>b!$;r0^QONB<dJ;VU`wdbfa3*^>JbYlA-|4ml=FL*f-#?9NDirHV9N^E)v;p}Q
zrq%<~vY6-><|mVaP##<k%8vu+C@L(IgZ&%oN3fSd`C!5+tttF>h8$w>b+&40g$6m)
z$6(iTpjZPp4>N=yGMFuj4)-^-`aE|Eyz+=3DoUw%X0>j#@*`)IBMpf#vBCs!ln-FO
zYYs3|j<zZA0*FqIF{&9bpsySovmu|I0t5|t$90>H>LFtY+B%upNjYny93OO31|b3g
z7jnWNGn2<5qwh)U|H+9y6e)@sDmVe$6r~g-JIG1#r3&QSa9AjDAlJ*u1|#||$|(+G
z`2KiaPc^E5c9oIS?A3<yjGXTCQBPi@Gf|*q9Z<E`NsquoZKkeVdd5P~4+XaKtn{)`
zv$!dxcVIz-g*Z2jRTkB;0nxgB9CTIPv0S7?aYC@7zOm=<MKWKxpTW-~uU*XOukWAM
zgnUF?9XPczAa-#?@dz21rdNFz6T@=>DKA=Qo)Iti>e*Y)9Hiz^c&9N%R?dn|MWQ$f
zaV9cr&USF^ie^Jini45$53-NdnA*@HgS)NEdcJs(2^o?$3)%n5LqfSONE7n9QSnd;
zV^~TwjZ!qo+!7W*#kkX`xW)bU8eCF5C@!<Q@`kK($DEYdt~4<0p){5~wq&Rb&EqK9
zQ@hEgGR(o^7Ll%ILc|#E-<b4e$phUo6y%&>F^C#KEDDI?TpI<YRghGuPR^W+Fp$R7
zA|iQYpha0gn+C?29x-Z=>g?qxdSlhiax&Tounq~ypn@^}DZ}O5yNHtpqk^1gG+mu-
zl(E|&m#1f@l=+8b<O{Jfs<3vEjq|rHuHH<(=f^XKq2LfRwDAI4LCGkTKpCG#TLJdf
z2KkxRX<iW*`q4s^%Xj1GjLcv=FY-?)y?H_cK7_(67yD$)8TXC@a*4lf8nw0E<l7zc
zJp*F`Y$kH)pe~K6G8sof$H}lb3hmUP;VL-DTc^uqKFeIh%TjxX#wy?US@I-6^BVK|
zGeL{>2wp*d8Jo1Tb=)|q@PQcjL$Poj1k0=uSp!NSRZ8Wcu07O7TLXVE(mp4kT56+I
zrB$wi+DKl%R3@aMW~+a1CDP)mZRTtxir~ViRuf1>R&Tvr9&j9cwM=xnB;_!PoJ`j^
z7vwg`Nwq<j!_g!{gG=iib+G$NeT>@x@kbgQ5T0`8r81FVl0V~^6vv?r%N72NtEN=O
zRL4f&1k129`m$}Kp^tntAJ~@LW*$=JLTQRK2|<J58WcCUCJTI7EX}@2W<($@LW#!4
z&XdV$`{v-1qli!Q0@IeJ7!nJSGEc|w)L3jO1a)cg&y@G-BB)`I5~ro9mSK;qo}qGU
zxiUC3;u|L5h+pME;}(amV#Z1=l#UzRd#mK?G!hGqm;zvW$q!7_nbs^nG^jB5S~!uf
zv7=859Akl*Lv*cU3!h^U?baaE?O6&hng-eZItPZ@>Sj_TE{N_T#q|!1SyLfVVKkCF
zTF@9U<4fg7G5r<IQ8nzx`OVg#iVg_cKZ$dJLS&fvr#>@Bqe{m}I%P+I0-3{dW89Yo
z<&7EU4E`+e2>-<t7(IXPb1(~%zNO{RPkv#Csky!aTX&PgZboBnhF&pX1UJVo6veOw
zD4vYnXO8ME#)F2JmC3CE0J33llymY+|BBMEW;1|t%<%QAAd+$v8JAx>VB9qJ_Ja&9
zzlp)1b5E{8t?b|0fhrg`ZmirE$E^S@NK|jWJy2TFi1IaXLm(~PM<uJ`g}4}daA&+$
zMEPF5RCmQh<tQMMX|9mF^E*hT*<)Rxj>|nUBK!*cWeMQCH$F#<l=eltFD;s`Q{%vp
zPQWBu{gPFOtTPdH)PO+L@WtaOGvW>{Ypju(KBL)b%v?7sSYA^zbm59-PzT~@X9u?$
zT`X<!Ms*GD#6~o@U=6!DhVwEjggZBIuBt!dn%7CYKO@H>M_2-mF)t1dYm&))UmSS`
zD}hjXQ-7%}2x2|Iu^gnd&_1hc(AtTM47jG6NjX{UL(oDQW#~-QSrUkMCqh0bb!m`G
z73#>yG8+xTWmcBQmS_%uvS(QlZ;u2S%1S>r+zD#=th_)Xqzc7s_Eq*a3JKGex!Pec
z0fhU;<YLC*tnryx0GsR>%B`%86IPC(wu)+5XJZwkh&qg$9R?@59aPHtgl>k7%>hfD
zEgO6;k{F|bnXYYzYGWW1HxmZh<R6puF<@#I5(75d5g4D*hTW*wEWdN1q^3#kHyWZ3
zI42JTI@q5?w#3jC4NP9?44jiED_&?hXIsUsjyDZ)19VYyhs`3vcH}|-o!~&Xf-{R<
zT2b^G05T7@%Yi0oQR>_XeEr_wVoQlSc_?1MUBd8DsA%$VzzSo>1o?vl$KTwaipgoK
zdn6=L6#hZ^PgI>i6@y2E>?;~_b+p4O**_LTg6=Ej@i<+*5Z?Ag`lDAb`D3@q7%au=
zeOBf0$w0p)#PVNx%0>mV+5p;kI@n4HY%<^@@T~mFU*LwWmuF&d)ud>3<c7oUEYBJV
z)j^)t^w&QptS~5oSLIT9F5pCPyF4E>*NkA5Dio6R4~(XP9z+?uV3S{f>i)>!dNH8W
z_6lfQ(<Cq1h%BSrD1WuLVPK*|_hko_ml|z(TJsUs^fmH|&qPrId8Rx`@;85bo71Jm
zeqQyr%``#7%xZbf-)2QlIz6o{atG|RuLm0)nt^`}ops)b+m2yvDju+?k```(QU6+b
zGci*uFu}ebk^0I+eM(pB<gGNJS=yLP4DfNI{XKsU*!B4BghlWr>^{Rf-|=VkPP1Iq
zdcgT^JcC8t^smkGkGPF+e?X(ek$cal)$Ic)Xmg1%*&y%z^b1-u^WrvuKnNeiE;Xog
zoifRXwoCL8*2+KqOB2nk<`b7PUp|U4!8d7tU?z<5k&hi(<u~{RmJ|8J5sMrIS`IJ*
zE9I?amj@W*XUf0gXjQZ{HL#>GtHFJmc7ri`ycu9o1^u$u3gd<TZL-Z(SIcLCjIw6L
zzE8$g@VNuQTZN?)7~1)Tk(`$8>m&d1v%;P>OXW-38DxNHobXp^b=K9AT-;XobsUQk
zAaWUO7C`I6gTz=I%WwEIx(}K5$IOxK2Aig9Xzb?m#_y|@*`vd^Y(H2ncf*TD%MOFp
zAXYTCq+nc$>}Wp>g`Tq0V6~%FSIEvbPiHHo3i>j^jdvN`#T;^6&lY*R>tM6OFx|+l
zjk^5{WjA{h$zU~Gt_QE!J^mhrmYBfbH~mYJks8gk%s#3N$+rfZHPd7ib$DPyXH$Mv
z*=!GAKeF{Tz=&*l&%y4guFXK`tEQE1uNVgB)+S(K*o6?;+7ZyYK=uv*n4AYHqJ3)~
zf?t-%K52s0yqMs#oGJSbc5^90E37`yw+&}K`u3Im`~$`YP$DFE`;I^3<-w=`MjY)Q
zgCIx08D{eeIUt@)fNe-~R}UPlW)<KKR_7);Xs}u$ncO4?`-)g5fpm?!0YD6`t%QEH
z9FjiOW~zre1SWsZ?Sw)ihb5@!4WgXgCWrg57Cys61;@7Pf|@rcN5nU%k3fIca+n~G
z^lz{_H6qN1@v>V@06EIY3)=4J;?Z(+pjxk6)4$Y*C&#2;*lVY|#&N9A%9F*ik<gUo
zIDgx^ZJC`~RWLa|sH?G|yaBB01p6B1FX+ywQP>1c<HW&kdLD&(uISYyCmFtYAnUUY
za&n*#EzXd7bxOkaLVASgKLT3pw6gx(sSb<=AOk>?5vV+zmO!V`5lDjwaJnG`<1j9)
z$2Q+GFp54WR=w`&-$II0#G=TP^h(cWsgZ;h>fS~d&`qnPk3-6tSOA><E>oGw9A-&h
zpDpYfbG48LV;=e$R8wl2aAc%^AOL?u>>J?IVy8Npk;4e1fj$*4OV)X0!k0y#k?Aj#
zGZI}O&yhfA+Lt-#+3@hp1gh5DgLun@Z_>9A&LL-Q!^grL!+2XY%!BAxK)5~Iz}A9T
zIT;kR20m3MgM$-f5!h_b+#!SAz{g5nU_KfbRbZ<lf0_WrO3z-QuUHiutPW02DRPJ{
z9R@s!x&h7z<1p8QCW~z&w%h{fOVUQsaLaprv6LE{IPIttWvK51RZtqcs7aX=J}jY?
zh7Q=hao1_FI{Y`i0K;P|(AOPb!v4Na&M`)P)n#&S+F#a947r<D^ASF7)TySLYt07i
z)*NZTk4M=T8Re+&=7~0jixC;=V5lE$Uv##faIB08sGy2+gDAqD7teqq+NNnn8oSLi
zRkNim7#~Z{fL-TgTpUlWYK4-2=MV0pE3C71G?Gj%Ff{h**;mFVH1c>EQ7edYp--ed
z0Sl$E@4Nm?HFN@mXBYXiXsZm1m0X-gjne((<jm#1B$ixKBHuGUgHZBJIhn+zab!^J
z1C1FkO8~6XJZ=0d-#5rn4@olVXR9g1d!jtSw?NqpIH^?FC$(^!WP`X73(!w09Y|jA
zJUDrQyUKr2Ng1R^fyQWSf=$rbT7?j<Y%^-|Hc54`%Tj2bkd<o;)lh43?k|t^PHT{f
zKCG$oW+c_TN~yKCP|d$m>SCwLD{G`a{utJRGL^nMlV6mn$)q4kwpP$n#1aQf)uhWV
z=~wvgk&%RqV@fe#T6X_8#+bO6nF_NODp=0Y1aoNv0=}9Ccl7|}V7KAOQXPz>xCw8o
zKISenC0NYN!fNDX^Q75VYjz2)nE{7LHI>OeC#6FcC1r}jlpt6Hx=f+OF*Puawmk5&
ztF^tw0Gj}>72NsLd>Eb!nkbpUaiu?_pum^}it$$&0bF|0C33a@OtqeHQmyzw+H5A!
zh2aJ9-SWf4NZ^FT+pLCfluxFyNv;X*sKsftNXfPSNw>RAkB&7PStrwj#0yZ=q)x80
z1reXP>aI5`KyFN5xglOeT3usCev|-*3k0C71b9FGAIOO}=L-2r(D{~|aR7d5>qoBw
zxzWK>Sp>Un$;I&K00P)^<Yz&x!$!)_gRC30JE;637(e3|(e$9993t0E@j+p=-0VCL
zLjgt;IC-^Ejb&Vz5!@{fHFZNqQnHeY6@R=npdQhT?sC6OI|)G$D{hC|`d8^0!<TSw
zai9M>FoV!3#>}yx#Bb6w$^tg;Z__rjLJ~+9I-LD-o4*Yw3}isvX1U$pCj6`{L<+OF
z???v<QX<9yrD3foyT}zo>AW++2-h6GBn93T7jZWuh4UNt=@aU^W6LKYm?bCo*p)Q=
z9Jx1^qaukFO1Uql!M(@~{-u!HTjLDXp#3gS$c*^Vuh>lx8fN-4E4oYGwnk<dg$%E+
zlG(;^vJE_gZN64l=2)HqdzF{k=EQ`sB4t@2d9Ks*T$sZaNqd|RbeDy7ng;55MlW0F
zt)CwR1qx?Ddic|FvLH>#6xBJ1$Y5A%$|9770ls>gsVU8oMK);Bi1W*2ae$PmEpKM=
z%_Tv`^w^>Lz*65?tCB})(fVm`lw~$F>(6JJat#bpmF2bxBSy-Kpj$OSQ@5dYf`Qj7
zW8dHsg3h-pL1<P7l!ublzAESl6e?i|(|U7IQ%!VO<2;tu3C5DMOqO3e5N}A<+GW{;
z2Q<(QckgCd>vNfyp?qJ~88(_xep)&L7FcE>Tf$tkO4g@u+Tcc@(Pg=8uy0O2_2g4!
zqb+B=H?-d5pc_!`kb>kyg@;tT*@q)b)tCxA`<>BKKjiq`Z}e2(0N1uW5NOEL5P*VA
z#&ZOM*%H4)m10`tH4uyE;NoB`>n8l(8FQ)imzU6H*=nG7ww`m2JeYoGMkiVuL1gS2
z+@)<%ZtUFV|H<#uT&$aBtk7I$8c{0`+0LJSv^*Sd>M6>*Ygl;*HhK9&;w)^MX5uV)
z;B<M!e^JhXH7=vX$W_Us0rd&rm&g3jpwufg>Br+#pu_-$X}$QK@PVw(frYFZYx!dk
z-?>`s_esZqbztFTHcOteEf~$hBJy;+fWwUHTU)PzjPNG|2FX{O<e7v*`dM(PLJxU1
zju83|HleimpB*7!3smXV0B+hQ0;*x@dCtxtFYfvr(<YveLkWAl)<FA993~JA^R3v!
zhsazfFBmea5J!cgH!nIgx>mfwG?bV8S;!Tu@$9cL9taEF$#D78g-Ct;W&g#>15;Uu
zAk|kwwCXu<p!_W%SX-mwaK*jxs>#;SAl8E+!Ll!}8M=D*K%Ua;fligsRXC70d@NWA
zu`Husb%V&8K}pI(%#=Xy4sRItVtOl(kCZdo$eE}4_p~%7{LrbYDMYx}-?jx*RZ_ov
zC+J|(mK$Iode=uXRG5wW9|kw-fmF(S35jUNfDK@Ei}&qWXY1ve$!Po9ScC3dA2`I=
z02O{{-2o4PPE6L@E*}PIWTqfFpkDqNJ54_?uaJ*oA^=Y97K!}waU%F72(bFZ->`#4
zCqbM1D{c;UCIKYnmGY_alybym|J%Qy&yb<7h~AUW98G3Gpw}9_jd5-AdF=DK@cMrd
zpF&)QT*4q0`A=XN1mg|Y{I35pRs#oP9sFeSmA^_?1&=xzl!*mj8~)*vK)1ZC$dE3o
zTL_BVP%YmW;<l11D%n8FcJTncbx|R@Lbe~G21E!S9l*~HL$<L$YM4KDWkx|;Z#yJ?
z$c{sjH4_dMiY*Iery*`wLkECblmkXf>mh1^Y?&sz7)YJ1y?XbRUHxNnZH8Gfd1<#H
zVSc1^#R0A1(M;Jrh8+n>ydnb+!#9Vh@slwS-x^{T#%_WE_L(W_ma?@+8a5)jn9xVJ
z+MW)CHQ&;^PxeZHF{@k!FWEcLRgaq6#NR$XB+n$Hg6WsSbqQg9U;i8-<Z7qf03X`7
z{cRWBwbh7i+J62Eshey;yB^R>CEppMmO|)>eP#cc9;IxzTnD?~0Wk?y!SFg($~|z1
zS{j?s=8;8-4sxWZVGg+DwN5FRCO>$M<luxfhT2hiVwi{Lq=1DUlF(qqahHRT4>g?k
z>DOBhGkWPn<yi#x9&W&PwjOn<91&ZEUKjPOPAW$ZQ7dz2>qzUCe3Va(MMKJ9%m`DD
zSaFi#%qBV7mJ3$hgkAyAZ;y%fSAZ*^@{Js8fSh{jQF5HaN%etVKdeKJ57==%Q0ixw
zlN0Qd3RvM!jN!5%(0TxPlA(?v6iskqoE(3aOhx5hIc10&qY=&p{g^`JsYALfpl3oQ
z@;uRdR-GTGZ37B#j7@TSe4K&0g;d-LEz-je7qQKIKPB%0dJS>qEiXNtGN{EcArOip
zP(^vCm#w4)_WRy;qF^M=wn!hJ-a>`jSnNUi`mtua?#0qCjhU*CJAwC47(=!v=>WE$
z0k&<(aqYnWZIXeu$)m;=l*t+SEd&=+_lk{#|7INGnfc$b#U_22`f6VEv+{JWA}AeS
zN(gdxkkTc`$e<vy%bU?IZEz5lO6$DUhotqW?Tvz@%_Rk~#W2mIP79jQBywlov`I0v
z9x+0MEd+`TsUwjmRczmmugoG@Gm93IC3#?R1(`)1A{r1{8XWCh#Ip_c*DMl`22Pl>
za7V3>VTRq#*80jRIjxw|D8m!RwRa4(=7n;O4Tq9tU@_;$3uYXGs<}W$_!Jf@N)?hd
zrC2+CWGpz_09OR~|0sVoD2F;Mxy&LN9pqisfPTO-CdO2@M7|h`@_B}nY7pAk00!y-
zy%ZhL<I4OQStM(YGj4TU>`~?9hFXoPINt_oo|usfd=F6n1^RwujzXdWdC!74F+LGZ
zz5xaYXN5xa`jQKWxZxKE5zGY<9AG`)4SG501i8ou=-!Nb?pBtERTr9E?5ANSOyDpC
z8Xkhld@n6gk0s=U&2p*Z#$2RY-4yh~`ZJdWzR`CYmP1rajL$VJm7IQwWYU6|v#Bl{
zoN*>QDL<C-U=CFcjpa321kg)`GY69`HKkB7WwrUK@-WFTYl}v-_cW>WiS@3d@DFXO
z%10rEL&G!IlE=$3@K!6T$pj-e`cqn@+F>Q8SQM8~68WrU01t!I1l7O}2$p%d-ReV!
z%S3}^SO#)Yt)C(omg?+3`}G(o^?7K(qDh2KVYV$GCze2J2n4Fo%22zRGAZznH=^*W
zU7S}Wa82oBxIIU9hlw=CoihnmCwP!@!Ai<tD3B&61a_2KBF(XvbR9J!x?pk~3Eb|C
zj^eN;KTq*r%zzr9i40nt8mu6TIx;P>7OhO9C9!2%yfC(*S*~;d*i3VoT;;D*XklYN
z+WM~!fJ;z}Q+|*JWQ%s*2@B+h`DYBtg|xI;u1RQ7bsg#oE&0sq1*~9m*9OJ0z#7sZ
z!Dpt&L}U5abqN&4PMP|4eZ0|gV2>VhgHJ*84X3s4$c)icr<(jISiqK>KzldQOz2}$
z6<`JjrInvJBF$76;X7RTPZI=`H!TO?Mt`Oek9dY4B>7nymGT470tMpdMsy=eE|6as
zx~r91Z}N9p)SVq9SZV%dd$%GtS#I&&V95V$xz)ae<UFHMe(68KTm)7BS3XV~{_60n
z=GXo#u|aIjZ)|L+4)R<7*{Wb0AG$4lLRTCmm$`C#Ft<@ihd}<3<}BS2G*d8gjNIv8
z!SVy!1g<TJN_WKp(QQzl4wgsH&E4@fmW+UG*h?14JwfPYsJBJ`|Ggo{l-D%mTIw+^
zqx9~J*P2n-XVRn=X-!+c97>H+AGR?c{y!rK1hG6hnQ5${YKqJX-hAHqGTWZkOp&%&
z)TGP=nd4xp0?3`gd7K;5sSI!KDvL;<;<rlL;~UhHSOZqd9t`sw5!_oA(VjuCw}!;c
zkJV)07i@t1wIIF9tU@YMkx#fV2F5K%a{!Yb78xTMSKZvG!ocE$aXJOfWl9FNOqTf9
zG_FC`rcNMB{cYnR)N&$Nby+Z|Vl=4&(wF<!mPnxfXIWvOBOxPKp{?{*#z#|-ScltU
zRqWrEo3Y6^0UI~r1g*9ch}iQ+^oU0(NLk~9TZuoivQ4r!7R#$j31q3PvyH@cg!&EA
zkwBs$*@Vya`2|6+vjihkVmAcqq2n7woeG%yIoW7q!$E~>OE#t17r?bfHn>hU+o>53
zv*GVzFaQ^cVU#i%1w@<L{e}q0rL4iSLLLa38V8%r79ZOpu^3&jP&$1;GrBqnYp95^
z+pR$*=vr6-)3ZG2gOCOC+FgP()#Zz)I5L|j5ENk7(eKkQvmp=DEJdA`KKX}y9NKFs
z$0Gqp9=3C_h3?2dq<QG4AeqmSN8&uSzyvCf`Ww9ZY3ZMZll+)JV`mP=F)RbqdEC%f
zh}-#zSdDqm(1qIH$6_I?nF#Zw!%S%@tcE5GKb2-=O?IS&gQ%Vkwq9L{zIZh4{wc;(
zW0&FK8Gn{%>+9s%G>-;@tJXuFhmwR&637Wf)v@Oc{yqcy%JYVxF%V_tF9D1lN=Uo2
z<%Lk8XnDcrFr`gR2w2)uY?q<_#ULVF;AowIFQMYPBZ9vEmwaS)T)Ymq3c_XOuO{Yf
zfpMmD_7Hj5Ujf0u9%o(NUWwg-N>?p^^GPhs*v>h)JY;_(tXE?zkeCO3qDfwJbl6K~
z_O-~I*Zmij&h!L9Jyag{hLJboFL5m2beNd`#}LBV$VGfB7(=!J!7*9+dms&qW|9{B
z;k@n72qUYGGxYGCcn^{Stx-&;eK!F(Z!2{{wH(m#BmYRhFdfc9xoqFx^KnQ^>#zyD
z(cZWH9076XgE*eUP;d0Zxap!UFugASv;jI>i@nwjZl{lo3Q(shD<6m8l9|+iS}v^2
z@rmt28A>BUN&A=0)~{Dz`PBZiD3g<a`|f}skUon&VoWI3&!T}0C-^L_h&4gox1cEh
z+(yTVWz54D2~TWW1Fa4x_dh;1ZxHJN!$L0V!>3P8zVuUm8~*#Npqa9U=34nWWa-)r
zBJo5D%<QF!A{c`1Em}gB$u|njvV|?vz-97b)rEk3W@~KcU+QI}&kz)>MQ67!FxwpH
zOLLNFUfUr((>B%kWwq?+Gms;a0GMLmt@n!TRG?<W&en4gm%OvTY#=hwo61Grd$LPR
zVDyiBaFy)p&*(qlJ&SZg+0CDU?4gx1B5#H4o|eK64tjR5pe3N~Z>CXMkjL^zzg3{-
z%1Ow&Z>g0%3e*-@fopWn0=0tCE`^G{vR7Ij1vVR%vJg%FpaR+3$EF=mQw6Zs``BoG
zVO-xgeuj)eBN=+#%eU?4-UE8eegU&Z?P23AMy6f9<MUdgHMUF@^?d1~{SAJc6t>nq
zApYWABL^0^xeRfwNcok6VoBv_YEvr*+m4LPTn=mUAq8%3MTe?NR1BRchuT=3t)~w<
z>vTEH{<@$hhYS=sJVu5NRGFD9M--@yk;SCtNMEXn2v(HtsK7kr;%Gj2w6EFto!x<S
zkG9D%{tNjFPZ_&F^Y&aQ$EI->tN0CjfgERZUw!G7C(H4HN01+C-GlYgkB6isIl+)U
zqOwj-OaQ8-5S)L5oD>6t;FbcJ4A(n3D7NQ-{&I@p4vF$GBxdE*AYi=DN?3?K1Cc?!
z#i^q8`ZRmJuCfkED>*s6AQ>F05N=~lLwXc+9cIxeRO*A7q?kB|qNf2~P*o+pViYhc
z7-ztqQSYvIS~9a()l-DAuyE<)#4xLK7JL(Z{XH6$pl!KAhgeDJ=WC~dXBCjP99z)e
zpM{r>?0i5FL?J@{WuTAOEy@oNGnVv>z@;&Ky;jcjXC_sXiFU|Y{)~<wiXJo-p6xKx
z>poKkrIGN3@|3kQIIUOB$xa*`Z;1b5SEmVij5M4f1u+wZA}YP>aQX_nz0Ft|w$3n?
zP~>~dad>ai|I)p?L2(=-#0EB_V@!}qamgqtEpQVrD)pmuUlro#hUQ6y;xVKahTiZ5
zs}wK83f!eKp|Jr@lolCo1l`$s%vo|yU_fzXcsn;RkHf^F62-^}|D}60lUwx+=5ftt
zh#cj^TCBhM)JFTqDky-M^Hmtf#IP`^(DtX*vl&4l=Ox5y&0@5GxJJeXK6(!5CuOmC
z;1|jxX~C8aT48+lI9n@9p4Z9w_TQO0T(=kalvuG!WFC5hduDuqiPck9OU^vG(11Zt
z!Ab;KnYa17X`!n;KytZU6#Iv&yhxn8*k75mb-GGsD<4=cmlSlFS-uV?)QXx(8O^x)
zUgD9cG=NCITDdg#E{80AT0Jg{#rNvjN51bPYfO;hZG&X|8AHX^0|_(%mdBDTTg$F(
z)%q&@Z3ZOLF9aqb*%$%09g`DakCjS0HlR3`r>pWeXoaAenGFCZ_(o820fVE_8Q)S2
zutH7M4ucoRgCsJ~8Bk5Y027zm*0INz#|*R|fdnDn$wVIk!MPNV)e+8nd85=C!W%O+
z+T5njL7A&TIQB0EVkRKrU#Sl^m#u{JmKzDgHUwgajVO>wJ{5OgZA&VK@`AV`c+_}A
z`ATDej*EaT)EQ)v^JhHq1QXqZrO7!ltVSNU49`1+5*}APb3`;7C+^d;mrS-5O;E2=
zPNw9?1j0=qp6btB6vWZ#kQRRiXA-V)Y6c5sTD+&ahXy;ym4QkR#<D)-Owzf^V5ljt
zldJ6xZo!4bHqG*bKr5ndSwEHO*4+c=e`ph-v_dv|3CcA-(Oh~Lu#aqx$~$<oTpNfy
z>a^2Nlj#m5PC4T~u($1UT>!2#@qV5!*ZX*s^<nqOUo~=rzfHr8>6vTfN45ermcmpe
zKTgYLW0;WhSQ`5$hHsSDhB0ln{4{YDm2lzgqH#HuB$68wXcQfYgw67^AfR6GgZ@0^
z7kI4E;zE8AKW0&R^(OxsR(f^%F<j^7G$J~~f=*er;1-915_DuJb1s+`sG7>HzMK$2
z*})hnhG6zfAI%(DYWb931;SBlG$X$@fXZM!LI0EZ>6jCaKU#hpJf|u%899Dgxh>#;
zLa*)<(7MK;`_aGM;UL9RtYo&UNjP@|9y?cHrGP`tA{!wAX)R^B)2J<5(;{~nK%+*F
zD=Cw^?fE#Q?B3%`iNSGY%DsWjo&yHReFi*nhLyM4Pt9njo|PH?En5LnqGy8O%-HFX
zl`XJ8%(AbWs*!BgB(oEF;Vb~l7Us+-?Y7t-u*g-)-WJLne_PjzXI?KB0M50WrX~xK
zzrEW6C}V)4WL_+wH^Nuv+hT_omC1rUz>F4BDRE&u!<D|AvIXLW7ug~*<w%FZg}&H^
zL%~?+qDveq6?I7upk2n`E|;Z_sW#0YSQdMuk*$_E-8~w5d5~}+D`i#7iXaPAX|9?A
zJ$9wdItjnF$SPaJQD<B%tK)7%JUv?8(JzjhY>iFczznWNSzDl?7AKu4>l`SSiJKuE
z{ufzXhpbOaXSlJ2+HWv0pnOA2M!`7QXfyZd-BUK%f1=o(Z1%ezCBo!)K8%|{!6{eE
z{k}s?u71EVL8)dA_7TNjvG?TEq=03z#W9NSHIR{J2HTs?01#IaXr2@xTW#cO{7)YA
zFOm3#UEtZIgO7<7UH&DEXt3zCV{aGB?*rpwkR^nI$q(6bt~}-vdDsWyDX?mlAVMXN
zKcwwOc8>F7%b0j1mdzsQI38GxJnGNL5~!iBf#&j<0X$-O**JOJ@ZX8*J;;4{BHqTu
z#@ierZ1TtSj0OM5mDk9V_6!A=>MI(W>6*&PQ#KMBKsDneKJ8yoV}d0JJbJzS$)8az
zSdP`q!sySW(P_PCN5sJ@dDfrNTfhqyt{M4r5HHSjnAOM{*?$aTn>-g+d^M{!;f|-b
z;CV+CcR2E&%~t#u|Ao8G>!$<SmZA5BkhdxG!A1@L>5Bmm;-VOE%Us@<f~(Z){qfiM
zih&q#Hv$7M$I1{zv{s31y}aVj;3k5-)Y$aj;u!>TDs2qQdNn;`D_P!f>*Tc{7KVNx
z`s{Tl_$Fk>zY)+BY`F=-=9~UGWnGvKaJQ_Gw;X|CXE!m66_WKW`Fl(W%BEuu35oJ{
zOj*OKtT}nd4=X8prM&B3&;SW<m}+YOh-X`;gWL?G@5QsuMXFu!zIs3SA<6!Om>P<Q
zmF=D_AKL1UJNvjx<evdLs29$8NDN4227{xj!pb-w*}LV<sNeT-fLxE*F-Uoz_}FIB
z#A$-9@L%x`IvgR#T|Ujf!%F~TE|Gt`07sJ+<B}F>FdZF=d_oI575dL?=CLL4B7L4e
zmvr+uQux*v!EQ8Eo5_#ULA7XWr>gxj`A=v8Xr5q-{FlLc(E_fDWqQ6!ID|O^gQJdt
zd~KgWM9+HDP$_f^(}+wt%rW06G~*qh2OHxd(0Up;E!!2URf-O@?F-c!hx$+TS!65L
zH=yRz4uxub8DCjBQFgRfd-v}pI~AJA4^}bUu2hF{Wj)Y=)vzhT8riu}Eng^}Up8ix
z>{95qqLCT)bCq3fA^^{Tb$&5Nb_)n?XK}J{7=gK^vb%u*9VjE;EObi}>XHCIh_3lo
zVV5Or_3qJA_HeMeJHC0j@Ziw+vu9wX2@-rx_A)4SpEt|i4uQ*;&|=^>fl}Ea`vkm@
ze#s)hDr8@OMmjQA-fH=_j{p-X7`uvNdZgNAKU+gHu2cC=%;PZ_FzMsnWr$wbKk$U;
zIQHK<Ai!lfaitt+gLJl*j+TRbdJDIu!)CD@?5htYd#O6q^X9xm3cJ)Mq?K-sSz=;Z
zIxG}LWl1@-+rlgob)FoSK*otRTsqkw9w3$$j3_ISBLY)e^MQ2MCPy0Cj6$1=qkNn>
z>S9z+ypMCV|H4bojB9iKmt%sIknht=j`cNm3qr$6P=Ju=1}8c$Z~z5XuS9GF#^iQT
zm5}4}^MJ+bwTAUTp^y`Tv@)plpk+}{^mp}u<1GtNU}VclL3F)PS@7hT5*jbl(V2)~
z@x)4oa*D0L585xETIl8&?KuSp<7tjB-i=fXm%y=nx-EEYC7a<%k3u*4z`qJ*m5+M*
zK6ZEJkPNX~P{-r;>Q(4w283g_)Il!7oORN>P=&9_<<cjpX+)v)bud)<QmnK|zj#K{
z#In{yH|ihYs(;ENDFy`IvJGIf<WK`+$535<mYlH-7Q4YIXBuMq_w6TV`M^Xq>n@-e
zj+`CHVo)(VN)3t!to_H3+QD{SN+5Y?yz3C>6lh=|1p&m^45IQ1ZK^65I*S}0^CH+G
zH5Oj%K)?t^dmU_Zv!ukIf!=^wZs-yzjV&ocfr9dKR_YoW1Thi??T4iq!Yyq5J~znl
zM7_Lo4Hs6^j?~*O=lE<8*XWu9D_kPy7Diqn>!tA<YEu!8oNRQeuaLbn%48M<=n;W;
z&;%X*@Cx8-I(tV3e9)TsjFdwbzEVa71EMJ!)ucuT#wRsqpfTlSjAM8j<oNUK<y1Fn
zY`k1kk923S*D{|6GvNF&uiC>hWSo5p4IzVzu1U^MZ?cF7cZM<C3{AKo_)BdAV3P5H
zKsAcf(9CGsWx7E0`IHO&X#6(=3quEB=Cz!a@5VBRBkv+xDHjE;jLg8vcd>5**$Tc;
zTjV8i#<5MbFwjray@|v3d;#=AW0ANOS;waXir1w<EE;N_nE+SGWi}m5vQyxi{eJAZ
z#jG<{UNT0mjV(DTj}_yhr<}M@Ym}z#mx=^gsw4}T*W*4`wkX-O*)(TFYzN7p(kC|a
zvu1uvRXl^x2Sx&y^EEOdumhDmC)EkuO{&(glvPcdW1WUa&(?rd$~%__6RNApA;=Y%
zE*aRw!md>R(9>XnQ!3mlYywJ>fnC|P!CpopN?YTs>*B19LZ6p<BRV7jY5GG}8Vp*<
zQ|sW@n`GqET&HPsazy}-tzpUu4Wx~ZK)O)1=@71Xv#J<>%LTC%pr{W-#3tLDiO_4C
z1CpUoi6`6hA|%&MaeUatp?+o_JxCH$15<i{X_de_aX>dni-R@??iEY9T`$voEj-q<
z<;pl;P%G5Lag~3<_&~NNL3t>-I##H{1?@3Q<p%-gh#FMyOD`}Oj%w|CxyDa*t~w*v
z2EMBrnsE^#d}n%L7rej!3ysF)x_||NL=&r+=yZLs12!tk$qn&mg;vk_QBdkg6y%a0
z`^FlB<0a3MpY$bIAj3}_6a1Y<E5z-{PrG4<DKpDq_Px=dK>4R8R1Rmz(a!=dtCR^3
z=g$KWR1bq?=@;7&fYxNdaa@V}<ffQ_9w|&f4d`Y=KWd0H$t^}jO*IqcR)?QblvP~j
znWj{#c6RxttrliFD!9KgGDSgfRL=vM{yNAOk=*R9^&1~eg&jQ|3|L-+^J<sEZ;ePs
zHnX$GZ4RZW=dheZYrH)MENg0nk|B4*-&H6v3~qm?e`B^VwLZ#Su^|_+dh*@zhTcdv
z?L8sOKy_zX<`ppDazhyZ$lMq+{M>68>eUO8o%aPpMYmv7rKJwg>geW&yf*U!{|x^{
zqvY@la<Mo(PU1`(66HI}WtIs=SHK9k&Bp1a2oRP+b!dw{&%wQ|?wC3LE_HG`3JhrK
z=W~6bRxD-?JTL9i?r&Sc8ug1{`&kWdUW|;=RMug4etZw*dI>2L*ZG223D81;o#g^8
zOJU(QFwcV&x)wVNT~!wOp7SQ~`v9Y{&MbssaL2_#wX6a(QI^=BJZ!w{m&S}>le~GB
z0T$b|Wd<>v46M<Ce`UE*4`Qxhj#!bvXW%FSYL}Hk30rOlf>s%ogK%Ya{4gm45pRuu
zYCFj`l^wFS+cUNbql~=HpOHXeTmWy~EFEdJD`CDfLl2#TIA#=AWxcP5o~tDG&}l0h
z{7bbVV2u|4sIfSV8<CCvJ3VDI!_J04tocd2=|LNt0vcF_sI?;9Nj4iskh}ni_;<m9
zN1@K*{XUvTu@R$y(7a`O6Uzg!D{R-4lP!TDWc5@_r%k1iBM^uX1KA{7<0O{XYQyaZ
zgP?M3SX)!oB3*{>;V1+A`%nVmPYv6HDN}4@NLw85?T6B&4O4RRu&<ArF+Iz)`~AV*
z*fM?V^v)aQk(d&qdXxIp9!=jY)*adeN#Zg4X1rDYdE6l(WzinVx&++{W8eueBv05k
z5FTix|D%se4oE8~h2|&y8GXmJn$gn#l;hZna2rRUb<m3QZjh&AeRa^P<WEk$<q$FC
znYd<AtwJjxKO5%@#$yWY+%Gfa{yE{23TT|HyqsmYd@eAd>x4XsEBX1L4pc^wzu4|v
zdh#*yLjGMXPt<{kxEF1Ei(x#_B<Gjx+wmYQ6dW`*hKk+w@^V5B6P2{Gmb~K6z?EVD
zLp5ZO```STr3D(0uiBsmV~3WElh=#_A;lwtGApl#5Jwf1wlvI~l{XwM?y~oin{?d2
znHIrXaV)PlN8a+cd8JZkiE7D6+%<fcr98T~{Z;Et1x^M!HNNvsH@VT%@UY)`d^66J
z>H{if$h&DmW<7>y5bw%A{Fg|S(I&Yt{w|aE0x^2R+mIu+S>E?&T2Tv&nTOr`AdQ`R
zM)|q<(CLn6gev5E%x&D4*4%mW&$zs>T(~eA!SX)x34l5HZNYpn>Bni9Zc!p+H5iO0
z?oZ;l<(hMoP;e?I|FTgSp8$h5{y#X!r$*$ZD9DZd`nS(xc}?>$+6wtBErdww&MZGS
zdO$5TD70TVIEHJuv2cShpxu(u7t4Q)fG0LhmM{HPV~-|ok)blwYBn;!zKWq#9W&FY
z#?fGl*v9?qfD#G_-Lz=BD^ldP%)*B9@)r3<QI}c#(3L^9D>B1oG+@jJO>SSL)=U6u
zZO?es?%>ZjE)BQrJGf(dZ=>U2p6pcQb{bH81$!jyY;)GOAc#hG33$fVFzaR4BDZc9
z)Ymh9Z8ryL9&#Tn{+hbrGDI=#UZhsh8fec5Ao!+@$*wDP@-2tL%8f8$eXZ<K<Q5zl
zr|VHW(irQWMP|iAj<X4Gs*18KU-#_Py=*E>%v$Gr$9kZCDNozSrqgSLNousp?rZBf
z2hpBzbyUi?10=*@FpqsRGCB4uQfqO)J_F@DMQ)0%M*R@P#I(r%@dPEGYEb5<MGmkb
z%h2Hx3Ai~qFb!!AE113Hpdz&}cecX)d~lJOmF<cqJ+i_dqa@54a7cWQ%1k_1+`mKp
z8@o<%`1Nh5rN9RZ%^c?2VagbVKVH#n%yf9b3;P)A^dJM_2*+A0uKgu)q>qem)SPQ6
z82nc~rgD@+qiV7pb~!pQU5C^qV)d8++3K`X|Bz!1i)F(K<TxXO@|s4N0*?20Er5=}
zS1?N&dYY$AYmpO!@QP~?=O-tnp&8Ch`(%flY(p1~D3Vi()YNz4S#oMZ0mIHM3|B+z
z<g`R4kOAcMBDXbWvZyu+kJkk0;dAh~Q>>&o)H8v}n)d7rI$wIlAwV-k-b=lmO3e~a
zJ(PwTTD>T2%y)L##{naYq~DcMz~)rz8=xS?r~<TvbJZ`l3%L}_x7-Rq`p1(d%&PgM
z1ByHra!Llpc`NHDG*#zhpwB{~l6Iw~Drms1ENA#iEE(Bu!ZQ<0n_&VWr=Y?}q3Eo@
zH?$O}?Vyo%Ios9^Umq*}W@V6n9X5C*8bf4d-^{_=U`@xBApx$IgG47U^i`9B01AMD
z{9zAhmqJG;KVvSIT?=`+$p9=0;zPk#1o7vjIOYT?F-8f!>!l=+R!3E@2FA<M;4^xc
zAY=tGKP@uUzoiwD<)}LzSRliK^-h{pC&SYpwHY}l0o{&U%)mJ}ogeEUy;;5LdEfd7
zqg9lGkda2&MGY`)K(`#_Se)x37ejS*(K6cS(fbSZI1fJhF$PEtOiPt=UfO_JdR+o+
zC8^Ca)=pA!4UE8LM)zzJgeb~!u??ED4H<_OA<p-Ki~?x?jo94_f;Qa_jcp%qW0z!6
z^!`HoZ^%l>`?dcbJ_`Qzi)^_CwKa0FJ>GH)k|8g#r9-x}IKTuyWBsWRGkAcP*QLI6
z*u`imgCMX{E;I5Qk4j8hhw1yVu3-&WWhVVqGPXp@ZBj%XW~Cy{Vi&&^IAbH^?6$vv
zaF{)`38$?x5yV{8f7Za9Q577i0BLJ2G9lrNb;wOg1=T@hs!sG_4ZOiQq$cqQ>knui
zjO%pU^5^DqLtDSTePp5!&x?+-2eo-d0nU@!fTL_mC4#-`{Kb5qCWb3)1W&Jz7t5je
z=cFNzQ%p;~i*(+=fI2zsTG6G*6^Wqp7ecqd5NQl12zNyTtU@^hy-)vv(&Q6aRF<l(
z(rk1K6%O2Oa-3b(Mufm7Q=F=ynpeuy1m{vcM63(e5?>(7pppvvwDi(UFbP;rklu2o
zjndhA$z>PIRSAELptEWlW@WuPm;#I{$j1W%|3MIWK^-B=q6x+iZF~lZOorzJHw3x~
z8;KRdb_Hqd+B8Ab2v>d!V5bKhS6+RlT=!qLXPp{)7)*VA`X#SX8LPY@aErK2_Otv^
z+6yvh4Q$}`lw{H$J2Jdz5>C*1`AMu2oB|GDh%7%1a$quRqul6&T7hFTk5K_Z{Q~g6
zIr1|HkGzQSn@n6b$>isO8&=Rjj6^dw?H6_d4{MN{?7v23<z@%VGyHIf5$hMA;s$4Q
z9f|N(f5z|-$j-_De(BGs&XW9B!DI6)e@2Ly88aW>{@M;Y_(@KF<G)*es%`dfyX_Lq
z8I0NB=~yba6?Lg$UUf$pH?Grzp6VVd4`zG^Z^PStWQr-A8UM;1X*QG}HvGt)J{#?y
zQ1FrW6VYvV`P=lu(rd|B$h%`yU<PI`jaRuRMul62ED@(i?hV#95w>u-FO6iZ2~wc6
zcH1kYEJE;#$gLShN?UG%Cdy3InL%0%UWH&jD_}s)jp<m~GPpQq$1^~reSF~BX$#^h
zs%D($92<s8GGg~DWp04Dr4<5Fdm4a3EuCR2WL~^6w4ks+=KC9(8innJLL>{)R;>Z}
zrbuzC7s|po{Rj#H>7r3^e9q__U~>VkjK$q>K*28`54gnNZimT=<_yTNvNYC{L&$uE
zEK36}!W}?1-XY74qaA<Qm9oNr)*M85df`2hmHx~`OaKaDcU78cDV!ZP)9Un|N(0#W
zc3ESyH>1Z2i`=d?BJOOxV2rHu*TIWbc*mAVM^c0G9yt@b7<R33W+2k_{v`t(kbXkT
zuxv;GnSjT7s2gMHjSXyiwJ9dwvPd>N4h?%`Mhr@)-Trq$H@QqL>_~FIv6o^Lb7^c+
zSJeYCEVGFlWlQ2&`LyHikpi6wEvsSuv8cVRzRqs8XhkBep6tYDpgb5745+UPbLs3|
zF`O1uGH2fJ)8d)HMpVs}hl16^$YxE{<zb`DF)b+6R4;$9T}P6(JmT0;9sq>GLv36^
z9*x-;v;(kM`~R^-zM8y?+312~EdAqgAM_m9U!F+PtyY;)wU2xo$_3aTL!Mb?_K!9J
zV$5?XF8#?k1>+IRoWXtcltTfdJMTbnky-L|%vFGZrauLk>ilkF%!WK;>tpxRXZ>fp
z))||*N&Xyb29G4#*TXwunJ>>dVvtnwe2j}j#(aGi?)pnyCm4pgoGjI4d1Lnb3qc;^
z;BIe`7ZbLuOzL{Q4G?iEFU1OqP&o6i!8L}Vv#7jmq*_`!e5||@1UUumWSRAA*Uq_P
z<W>6yw)$C6d|wMB!4ZRaM4YNs@_G;fJR)e7kOjHE;jfY5Flw94THZ{!KnOkdl69-`
z`_}8^Eq@!nQ$2pmFwx)rnYGE%-U;%y0leip9O8FkDOm_G@@{&2G43)%DPh!q7>Z~e
zWPtN~h6p$+5$F4UT<1Upa&n+Xd=LoN%{6_K{5#T|KXf!;55e#L8S}xC0*ZyDWSxBE
zZ_xO-Og{G2Mif4sF0jpjt9=ryg^L0Xf6TX1{*~Tl-~b~o%$K!LKJ~W^B9lb_?a%Z8
z1BYZopT#q1!thUHduGVzu`#u*0|uL6i+te|k$N>~6Qn*D((nrTPrS?;VQAR|4gSlR
z8-{us?|FHA6?p1g1lh(K#(_CPSEYQqLcTV-u56knQfvk`{Scq0lqvZ}u^DR^GP_!-
zh$a~>n#SfuvR$zoh|p80R*TkdA5TydpQRzPvO}>O+6qvDjXStwv03fNB^U&}5;U-r
zKV#uf=KapVZLxE)n<`3?rGV{6yyq^(ZYCUw`Es&r{D}fj2*uwm4MLX1(i?5EyRU$2
za}rUO$lC*JkZ-1M@HWphLM^`)tL|pN>>fV1hkpZK4t7Pw4%xF<O<W8tlf425WYLP1
zz2ix9{S*Yl04w_#=moV|EUa4gEp~%v5qjHyJHfOVRuIdt+Am;2JOQJUDgM)_YU@Ci
zzGAnE<SMFJO^aP%W&dI|<rRz@BL~=g+P)9dA6TqrN(_VuLOI9>?{-n)Y_ePx>)foC
zgNxNldG<E{lS6!rS#WnOlS6$BvKlt9qP#~nmeMDy+#F`GoO04ha=3j+%OspJM|7h;
z3w7d$FeAy4vAwmlR?1P177f|OCbARc(f+YYy*evvx-(h(m|`{AQdGk!$;q+)rpYa|
zLoq3sT<y4+rmBXz{qa7H8SCLo$DutTc3h)w(5eKVo#+G7u*C>hR;Vzs`J@EsQaVTJ
zftQnwKDJDU@P0~ALq((+o|@()8@75crv;V@>L=97=?MZV*XlTz9<k?8_u*%?hEzR+
zl17${lwLM#XDg~U_l|SeT#pK+kb(R7j#2Dm83x?hn3_TMzzg_aU!T$(m&E!C>1SB&
zY(@ORMAVn+Uu>4@?haRe4Y?z)TFZd6JaeqkLp3nIgsqSPw^|uP`4f4fcRH3c9LXk&
zpg|32%c4BbOb9_#8*eV?CTIDID4EbUfol5rmkxIF<4tmQ;AS!m2!mq(5Ot;LSKt|1
zmske-FLw1AL+4;KWk~ECnwDCZ<$`Shq)(o;7g)l+FmOLECq-$*)Vv;Vu`f%pgG@Du
z$s;q>ySxleH4{Hd4B=X_kzR^YU$)9ic!yS_p=oqWXUu!GhiwBLHU^ijI!k1@k7I#F
zyr4Sd9A_5=1#+$dUpjj1NEwla*T}tY9ehUmFDhc;E<+~((h>X`qkLJ=<8TCcBH1J^
zUsQ6m?><rGDWv$FS}09EENlT@Bx7Q+un|%VN4Y6EFSY>Uc(cmDV}mB4mp3D1v@Fgu
zN{iJr$hg2T_L+m2MK$k4IzM5;s~TgH3u0mL^RwU(l`3=^fU9A=k7fZs#3)GB-4}E?
zT<E{^8#)o_*)(o#yy3gWYO%lU47tcqjLv(Fa&at_cM?#5Yg#U`-_dfZUcQ$`&07rZ
zNSM{?Qd=?`)UiY8Wog^liWx!!WS@4NiDmNrSS*+l33V07Cu1KsL#08uqLin#Gjh%R
zn`92FrNTe9EwTz4S>H%vg4~<U<hoV&c8*x-U*at&uH49#Qswj5Gs?`sG*kX%LTqSZ
zV+||FRwtm<O$3Exxzxlavqoi;TprA_wh{#}C;A8U{vk;kQr}{!brc4mTD}<)p!K7j
zEp>r2STAVN;>Odbv03WvSd6Hwlm;KfYKj9tSPig26AH9}l`_fq3@sJqiU6qyL3Yw;
zf8knXS;U+HNHZJS=~)&?Q*d*{e4tjK3B%3t;sh43tyg>8<k%o2DKP04E^J5hlz1E0
z0sJM}V0Eg2eCq$_f6@}8L9k%wC6+p-D}9Mf^IvGJQ2T+B10~}tjgKN1WonCD6$diY
zIH4IHk`}o-_HR6cxM62ZQ}7&_daY8X{Lp_1_Ra!eMEU}`#_`2W!R2yoj6nJTc(e_Z
zirQp)011Mo@rRL5cG9jhAdyo7rQ&*j4ZE+*9z7y30B#AnA^nnP&1CsA<wuE%z%q5x
zP4eSd8WI#~wEanJ8mf#UGY$*~>;Gv$3usuhfGx;)GUmvQMoRSvt^S#h&Ny%|d;Tjw
z_YbW^AGJ_00}$CY@(Y`{v66HFq3I_3dknk%GKc48|I#G<?l<TyJ`b4`GbmWJ?AA0l
zlf3DJT`9lxXB2c<%Wfvf`B$+z^-i<mJ&aJl_A@XWtQ*058l`OwwXAOLZ*1q#w3I^I
zZ+%!G9*#NwJ5O#)AR(L;Bfz{Qw|84QRc{uhP-2uj(g-YAPo-_6+!?YAT)GWVndB}z
zc!eW|7s=hR5*XlUAD98T<(}BLDTtha{CjV(k7D%q%*cHK5m<$nnUGe8i1f;eO^ie5
z?XgH^_}jE!z*vcEZ=KAH0g;M<WzUNFD%g!0L5Ges24FEl-jeHCIbHpt(v~K(@J8N&
zc0}gHQjradg6cV$>#v$z0wTseL!ox5v?olF%hI$vQ|9^G<gd^bK#NRqpp9(4gJii-
z+!>xdQ^^p4DhopWf=$Tdffo9e1?55(#R^BkI0+Bq;xsvoh?rstuF)k%?l^(6)PJ@Z
z8^%^Gk!Ai2mFOTIXd!u7?(mXNGLnl)`G5_FV?~UhCTp6t&~GGwKr>jlG-Q=2DiBW_
zE45pY9+uUC!d%T%4R>A>thu-bd$ZODrYOYTsYuh2b^eTJ!xD0rN=JYTWN#po!L71B
zo~bIx1JEHG0<n6Tlb2fP=uAVeY&24-se(ETE9j=+pIHRY<7{VTvk!_B2K|mI@iO_H
zp`PuP<bLOkG=Qy>2mEcGK}MCY!`^L)6{8jkJUcn*jFXIGIUza#Z#6)mF;=o>%7cLw
zAg^A!Vpu9bATXv${@zz+4l|F#z(&woN@e;X|J^)b#2wY2byx&l+#T|8Kv4nL8Pa()
z$m|b}j(X%0qrTcBP5`c_sS3`(!P)YtLuArDL+<CvV}=&^zK)m217aY8MpHy_$P>j~
zdum{NeI-PR@`lFhh6V&={W0jI2By?na5uF)JsAk7Z-7evRQ%O*AZoEZ9lL}!3~YW4
zD`U_egisn`o{7JU!IMFY&)WARM;D(zqD21esG*9@I+C=4&bG@)&*^gxys6-}i_gd6
z#umz7{C61rRf<KC1$n{8R5u_NjB`!a0y~B|T3!q)fMl1+HcXY5Y(N-|%H^+d(CFy_
zV_=A#37#*9megxtZ+RugfXSXobMiM^7R-#Z;of-FNAabA)>(0rby8l7-B8~X!*}P(
z>ozh9)x$siMr^bj@|b$3!^oTd3&TVh0|=h$cluj401DW{dGUASE@LxO<ZYWw<tapr
zYo^#cjvZX1Fo2nsZj!u}OL#ZFsY6;Y#3c};{38Lq0#>)VU}B5qz1UZDcH&k0e*B%w
zWV3MMe&FBmo?vA+$cFM^+8HXRwx?Jdv@ddHm?ixilOVzD5qR-Y(A#7Ln#jiq2Q(Bz
z&IESsKK#TtfPtQ94`N8@jjj6B2c=?3iJ}8mmw(&OOiB98&RHQW1rQJAbNeMzQ6pcZ
zRnuq8YiE)C$LH@>g9U4w30K0Gfma+bR`68v!6vI@2M6O<@jLDkSjj=Jj5uBj?V7UO
zX8GEQumaufrNl$~+<==(i^(@i)O0lfe)a82x(;BSGvf$EAT1P(HnLdP?Muv_2yu|8
zQk8gz60>i$g9o+AjwNn_gdu1W-G4i!A+^pM)F9Q>G<1nA_;&W+sfO}~$DZM2QE1e?
zlwD%Lk<G{yz^T}^L=BTDQKY<JH(wX`93drJ^QN@j9auA}m|pcw2bOWeus-@5swDw`
zaB9jPu|%*##_3ZC-_yS8Y~6B4=PcQ)L=A{=LE(bf+g|T%rP45koUcjt@gdOp3f3O;
zy6&4tgKiJ?d$p9{UtYQ2F6r7InE(H&4*LZb;2NCSx#|S@PJTNfOqj@_MikjU76m>-
z>2ekdKO7KH0>Dage-P<`C2B@J?SxA%mV^A8xzI_-(vYYv2gfR)Ueh$n1|^52XV#_`
z<{UXRK|+Pp5@O8VdRR~fM7ddVxPMDntw_ZJIl@qh9PHtxa-_dt1gDu7a+IUZ^?q$3
z9f1OL8Xnzkayk>3<YHRQF==vKnsRJOm%8F9fsxPgx->4L+XMcyNlASJB%R}oz#8ZX
z#tt7Js4qtZJHx<ENOQs>ggF;eG6j-4<itP^k!6QSPVyCyJ}H)1@j80r6jgF^eB+lt
z^Da-x!wyz{s*gY^6rp~&&n(}987puC%V`de#vpPAB?xBKpPt6%O{I919=7dR%w*~5
zzoQ!;wnYDxUTJl#$>34=8bI4S!J%z-fNQp=%!?d-{7arR&Sru)JQ&iq#C;c8tu+Ir
z_VbzP%7RD(KDb8u+cUKDZARo6STn3O1N^-<=%K7gLTP%S&qE%~!2AW2%JtGaBSuw;
zR`DTc#s~<QCdL@+bCy4&0!B;nTHFa|8y6e~m*Ajn7J?I`jDV>vgJUpFeA6wJA^tYD
zN39v=uO-_kaD>qU&WKViM5{(ImFlE0*b2m+0a8@rKA5qpd9pxmahh(engv#YI!o*s
z<!<zPMOY~<l?Hpi9=&C#L!)8|Hh~T%6H$hxMY9DagL^t;xY0Jkb#v%bgIoEWG$UI+
zS<Sn1eMn+o!yZ(~X3pyff18M4E(iO*jZ7o3YXMFgia3o*&)6WBP0?1!=ol5Kq@meD
zuIY;yW9P54wH(DOYvsHWcem8k<3PfU&p@fMhB~;e%h4LHEKpZD6|xg_x2%kdLC%H!
zy<X0bLwo_`Pb7TG1wK73+{~()fdgyCjPdEfb`K6^&n1(rE({FA48<s09%b``eAj?t
zp={^@7X<`WHB%u3T<kai#WQ3FSHn!XB=Ca2nBTw#Ti=Uk<S5)g1WH|6(zT0<*6=1H
zoLnwTghf84sj2`TJ$QTx-**tXFUV9wD)cn~l(la&cKYEiUMuA>s`b&73ZEI?M%8ms
zGeojJ(j45*Flj*l4*9S$!HN2a7@gL{Dx<^R1A5AY01<bInPlibqX<Bt5W(o5lbz8g
z!I<6}e}-6K++6y$T%Le}O%-SZyO4=#?0mxujDSUIG_=$P&|pp!CJ?u-jB2IS`In~9
zn?OYq7d8af{Y<Hk-?6uV8rMWb%-h}CV9+4YbrQ0!px`yv$)q^Ns9#VgSEO}mX+H{+
z(irTosRaq8v}NUDRiHaq;2DA{O(kAW#4*CHCFh{+m=!ry-H>LxF~^S`KCVP2+aJSN
z-$bSuy`g|YBXy*yL4HLe3d+i)#SQ>cY&F)fNu~uEt5ivZFjyQ4v|Jg$fx~N{Csd59
z;u(&Mb~wfsUTx2)3Ba(9Kx1&C9|TQZkkKyHKa6+iRs_a!s5*0vA?)HyPWs*nN6WSL
zg~#~hWO{s2r50)Q>dMM>vEi=fwy8-mESw#3y;1eJN`zJ3VE>Isr!x6bw{t-0z_L8)
zvixztIn>Gq{KVH6Cc;_JMt<tg%*3a9f!r9Rr8j(d8)m~t`Ljd@Od|ZbzemZE?ika!
z$TNRod)5qDwq9-uFwjY7a%GF$91DT6bBhnrEgzk%6HfD6ZELH%FWoO=kVe+(Ym{Hv
zAQ-<=e(h^eNr4;&zJ=d(gN^2QmPIxX*>8i@;{HVzG_v5pYH#x)cy7tiH{%|>J<bWp
zhOP+K-R<t0xx=SmvMr+|tkwpiTjb6(8ifS1t`%}ukZKk-qMY3AgOe%KK1oY;cggOa
z7+eF0)K2Jp@7~xkxVOkp;eV3*VzoF#XgUFx7u^uhLZmg0C5i*Vn*-OzjF=gWgBEUN
z3WMd$hk;O2uViKbT9TQ79hw!0N3>KW>MYMr$Ywza%e8LvXLiph-;p`SChHOPKi8-R
z7tngR0orZnTG*y#p0fkq0G=TUC`{r^NSFEcJtR|F{}<R~EID;K&ZEXdEcEryff$3p
zQk8;KERaRMT2QE!lqAlSjjE+n7RQ}w6%ra{Ngyl>&Ife|NzL&J0wDlEPL?I8Ni0?x
z$q0v-*CH)AE6Wp1Dl)03ffC#tS<ww$Bo9C;E7RLFd5|9HilH}lm9GGjEm%A$+3c!h
z`mA}g+M#K~-1K6sP~OELt8PzZO^NymvlC>k&#rc6i~&ts*7+Q5z#k3xsK>?I8w9Vi
zv2?__ub}T~eNf2cn#r)RZSVn2rGeKKZq1FcM35ItxUyF4CfoQ)EIqI}CZzKXQ{}~d
zTh#B;n2ep|$jjyaphVPl$sk(ifmjGk0yyH3Q95Kx90VAIc{tL)(wPu!MhR}6DJNU~
z7i=NS++=1{b=d6vL0dM;)z{<5cE$X#)6^j2<o7Y+1FJC?W0fOd<DvYB>e5)EYru^1
zu;HM&enL+E5Z@DHFbc2$j1lD{A?i^KG%v}c#!(xa=?!?yAVI?tT4U?w@xV4D3~W0E
z7<nRLklAZoX}kO}ForNtrn_&1sQsirL+T;kQwOs=W#gA+kmtu16i?gau!)16|H<ji
zBr+QMn9xP54Nox>pV)TrnE;8I?TXYkc{UFcv??QtKL;%pXp`mV60sW}rcME_g8Vri
zt#)}nFoSNdRaLldbMlvNqf+<mz^(Q|Y*k|=O4-%Qi?)D<iWUf#FZl=XU_)@61KL_E
zfAzg^=A#y0s7%1m%W0R;Rs+K`FV$q|SK^ng(^W=B+Yu}ZfmHIh_yt!H98xg%fGOd>
zm&&WgOR_K;F-5IKUQ58xcSW;7o4g(j2eEt1tdck4%AzZ`)@?#>`j-^gb=^SETSi{!
zpWIXaZVY%#ajCpr(lu2@>(18yjsGX_IH;z)gUR5IeAl0m)tOUsw*15KW?{<U3RKUi
z=E{3c7kan&8SPflO5P7NqL>oTBHV8Bfjy%V2|mS8$eer_&qicuHJ-p+wtuE?m+<_<
zv(P3V`F!0?GTI~uV<R3-KK3i^x(C|pD&-UVtDwGxg>L_4B#p)u)s#^`jooDE7)sE{
zzkQW7A@j1%S7`ps$aXkOqQZV9pQktT6`k9lSk9F%f+C<G=PE0ZK!SVeKXH~Iq++dK
z`qw-X^qX{0b>e@^SH5DqDDB$++DNCN1$VxbMnolxV?(uqAp8w~1|f=7ZQ&ePDBG2q
zC6$_?b)Vx=N6#x@5hqu+FLgU9)<aXK?BIj21_mrtsyOTz&yerT`wc0|vQw#=?W&Q)
zDLa>{>9wF%S=}zBT{|l&XDiFWlXtZjaUWF5Zt)Ye1N^kReTNZOfEd168ZoJQi{=$#
z`Br|hYOiGi@E*P}Mq#YMq1UJ!EGa5v?HR~WJGBWxd!-qY&yz^UXnUu(wa5Zk$OhTR
zpo6vb22^s|H`WUq_P5hGWP!9jF+zcXh5I>F*iaZefITQo<~lQm^qu_gNt$9tIhhZ#
z17!bL73z^1|5aHfEOv50scPV;ibeB5tsEF|W1q-|nHGMKW3?5M<}x|h-)P4L&&v;1
zDmldAx5G}EmhMV+#X8i+hs~q0p?+GW9OiGjAzp71{`U$w+{W!}EyhI)36BxXN96IM
z$5cZNR6;s3puk1J6|o$`Q2_<^j)`s7>x^FQqa6#hywgSjO8|anV2@Takz-1`tk**2
zR4ZJe66Uc!2V<roDT6NB<v4%FiXl7_>;QCpdS;z}QEx;}D0QP(b3OV>pzcOiPPErp
zdm)oUl@25V%SrwzT|Tq|%z(k}<We^(<|=EZHM0QeDM2_RARNf4{tZtuV=i$4%W1LA
zDwpuwqfDd{FI8U{OcRJR!7J$ioF_eeTj`*fB|ZHw#yZZFUPhd7(l<$OM=P!8thTuT
zH=oK)(kBrgr6|3~Fe{Ib`D**7-x0MAglk?m?pa1J&>+??zrFOtU~@>Ktiako$OfTV
z#N7aYvkhD8?XI{X;5P#kbQDLrhghHCYmUsLwOgmlnSm!5evw|Kx47T$Op}qb{CC}K
zZ0~%$#b$RFmNs9`E>)LOzn(p0P-&MZ!D*j!&^)XCXq8tOM42d{U7ENFGC07;TOrAl
z!z`2`u`w_W!A}ppV6zk$@;h6HmPlbjA+79Su@Lt7B`v<te55F$ko<wa(At4C8Y%W)
ztQMHoG9$5ohD#C>^1+gMXwe8dphctyQsR`1?8~9)%Oxs*AXOH#C9Am%3(`~`Nzz;>
z!~K~F{Jf0N8#=#g#Tq&0xrW*kPdrIRH~_J842Y$BXtj(?i-NHOw8%Xt#ch^R{)<u`
zeo2l+8J#9uNrr{Hew~a7K!H#kZM}fbGnzxmfHwT!JQ*A3usT!DM)esfi!%Z?yiUd?
zG@%s*X-RG2boZX`qZ0vII1MNCf`A<B&~Oy2$+mEOjDu0hW}q(G%H+b*uD$i*>sBUT
zqYS=VY9=*3h5#Q&7<)_oFRZW(_`1jzR91<`TQqoF9Bc%<r(7;^Aek9WZiwn}^1a}L
zkXzu+W>g^p#V)m-LGe_%EMUb3lf<q20XrUlKYn4V6o9evwy4)2nK)CBhMCg1T5Wza
zsd->8HcPqxZvAjAh#B0FT!Q+MR0QV0ANV~g%wowXnIqXWCvDY404%O5U{r=&a6YQP
zRZ5kuU|d6kOfaM{eF>Hk6snBFHuxy1DOCq3jCnABULM;&hRJ>~1<FJRlwdOjhD;Va
zpeh2Z)z`+P<268}&Zz~?U|b0Gi6(e>)hR3u{>-Ymm>xFCKrJdOl`G<BD0d`M#s)Xq
z&;9%Mmt3j)51>~xOsKD!M#;0uQDR494p72YZwtZFn;rU~85SskI~4MbnSdwTY@Mx*
z@MS^i%K`^e90FAR1V^Lydag|MXXFd?nlOU0#h)=Ig<>Z|AEz0lxm>MW=|CBK)Sz^*
zx`i;?Uu8&U^GVdSn+8sQbzGdxFVF(l@&ntVhGx*>57X=xWw#XE^_uj|3<n)Zhq^XR
znQTf}E&3hFbpM61hxD%{bF^IN9Gz;tS6at6$s(de4Mk8ditE#gH-Hhfq21;(xgkVf
zRA{Is3H>NG1L3`Prs-E%j&1yL0>a3LrSn)eZj=1PXs`ylKu&&Ye{{Bvs+=S@`jikx
zAo8%wEIP{gIM@apVg@!cXMvHoKR4>rjJOK<ML=Rkac~+sGEt^ZZb}QW*j2O!m7CM3
zo551(2<FwsOXU{Z)B$}5$gSyELlVTXjM738Rr||yS|}Bn<-lU`OjY|;+BGKl&{nrn
zejQ8HGGt8ax>kPUz*reFvzn?JD8G%j!C2^Z#yoPHKO;iO4dMNk+haKt7+i*?Wo^3L
z5%?X6MwWNRALAi%-R0vTVjp9ZB4*0nMu%I#=;a<mL}|@bxi=62ZbhDG5=tt%BKHL%
z!1=+sa0P?&cMJD#4OvIqd*@_^zu!I1li1gwO9TMVOv|zM8mcmY-!7L~4m1xcq8+fm
zn`L%-MsE;Znd_x3o~b{I?&f7OCsqO7JntSS$A-EAUOP94yNDG_rZ!1?+$K51hR8g_
zM@2=2%=dfrc1Vl2Lt$7DjFdgATNr4t(C4DK!U&xCxR=$_2r@z!Ls=9U0LSMoF+(jv
zseqd-4h-N+hSM-`$O?Qc@s%+nm;F0X<w=%0A8y4_UoOl1+c`LgY|X&Jkh*8GJU9>B
zs1*#!$;ygAl)BQWJ<OAp>DgvgfLY&Ul@H3>fbkKQIs|c6R(D$}EjC)>j8O-E#jlCY
zFKDVp307I_qZ2jeWkwZCBTO{GI_)Uw@E2%NAvb|QDC+}X@H)Wu#Djo;8RSRVXpU^i
zgQr3yy?>@YZuDR5x-jr*b(<}lO1oqS`ClmK`@dnW`oB5<4+H;)f&U2uvN?p>DNqH_
zC@as;!09Xk(frR86XwbN`QNu*2lk{E8(4WLZZLiF0cXZ&t_oRli`^P$3^__VgRep0
zvZ6!nvNf0^A`;-al?US;Xm?DvT!10Ek}9-e>vW0{^sA^EZgQsax!*fj5Y+INsnOj_
z{XgV<kgiqwl~Ss@JRJ9)M$MtK<sT9d)Vfs8kVgVZ6E&y&(Zp5NG@veZ^%cou{xR$y
z_A!i8kH>%-drD(dhdhx+u+Z4WK(_fN{us9@{Q|H9H_DT75u@`-PkG8$N(&1O1{7}r
zsM=TPh<w^W9SSEpJ2w6)CfBMH%pF}K&-gP1nYS4P)@S`0qZcEf=+A)~%kHR^=VHGq
znh~AdAkX_pCLA&dd#?N?-f3(=m*=d!;QP1j4FuB$?&f?5-irYy^g>EDT**rbBx=ks
zRYF(%tB+<bQ<JhxF}50`Y6g*)9Y<VX2>KdWtoP+CdBw3$Q!uskP42o{Ab$(!QAVvE
zX<Dzw-(xE?6W~{PE#Op_EhVRA@_Ks4_VwJ6mGXvt2Y-R34ZIn9SXP~>mbcOu3so;b
z!%O*lJaZEe8@az7y6bl<<sDy>8HZ_kpb6A!D6;g~yZ${6tD4+l$CH2fGYT7Y0fScL
zJ%46L+dNqBr@5KQf-?lE>Vu#|hHgP}sgw`>Ei*My5ATqF=Fe3(7dCBwADK{t^-ijk
zkGlavA+-ZS%_lJ_j--WHEtY>7zbHpC(5E4-*VDby?On!kK_8IMd}@wHA#tt-B;h#8
z=k_)##K4CF+GRxk7mhwn-BI@)5yG-V6sMNUe_|nEPc%Q!;bE-hOaH|PSWT<=odqzz
z3J`k^7${#mh_^s#z6GcGR=D|PsJc8LCBxS;H2Gg}ZlRul!nYgRwG-?=HI<A(knR1A
zHa&vc!$o!&YA%jB*g?iJ^Iw|Nm&%U*i#2UG#Y+PPWT&BO^2hSgAiAYab~Y#)P={1@
zNl=(+57i81SNpxQb#$TZHq>nn2>%C1Mr0)p!S41#QKoK^eAE6LkrlP_EeC<b$YcWE
z1%|Ud(lpHIGc(wp0oEiGF@?=$uk<y;^g*JyPweZx{p-2#Fi<hw3~_azKm;SbT*m_o
zg#NhiP<KdGf?rMn9=|;_asjxBwA)$b0!HS|gkxquTS85JeT#g@{<|DWsItHPH!GLQ
z0Yl9IXR<c6Dvd^fs%V^tPuk=_2g>baAY>}SOn^9OsN47v;DY*V;DZPIqG(Q46D~;j
zkOV0udJ-}Wbx3DpHW`zt4;|`mi%dlY5;BlGT8r&CY^YkXI$O(7&*X3;A(Y{xEA)uK
z1x_9EgInZCUxyKIx9ob%qk)moA(zQf4hL*^nA`7FnnCQom7^2v3`8^jaEw2*I8ufT
z%#~x)dyHQ+L&|aKJ;rE&2jPGo@6ULc;9rNeVz!(RtLg(k_lZN@yP|!2S@Kv;8ro$}
z!ukx|s3#9qLwy7M7pKImAVW({St_R{;-dXUb3i~y=$<O4*)C`5vCF5YFB#vdoK<?*
z?^j=a$|ce>5giO;7)5(`DtGMVc%^O|Dg6Jf*^u<Mu`a~_q>lkS2GJIM6L2i5p>h=l
zXGP$Cfh-NNv4R!-ZHIwQviSi)SX9%szTChd62$y7N2#V9m3QTg7zr$cso$^x%bETR
z#D#h3be=C~`7=)^hS_zaoE_LN&mc-h>;De2*E(C-B{#>QshkW>)bDk@lh!%J>B+bf
zZy#8hwrWHLY!U5J;J;gCMLou#c)0ku2e3sH4(+mxJY*T7GkR1&DT=k9Q(0fn)JQ2F
zsu57r%B93m-Pt;JxRl1)H9_ApqK5j0{r3o$?xvbyMubIZNYyCA4e^;;1OVjZoLH@g
zveCap57h=a*Qe98V?D)7Wki6$FmgrpIvHu>!R^D;wo$&V>;^nvM*I7~2*x!wI3{ri
zR`_m%+(_zX4P7<O%6S11DH7i>`PjZ3n`UQcFIryOq%1u%k(~w4#`!a?>Wgm~&3nGl
zY5)EM<bnhg61@SyO!yqe2T(A|z+eS`7A~Z2^8balESRckt^Tf$V9m(Q#dDGER2Gqc
za<MPL*b@nbtP#<2a*6Hi$Qt>czr#htJ%AZe5K5WsrM6q-5m~t`Sl^WD2CISVJI+xV
z!oW%uc2kA8x)yhMCTJZw^30nD(B;8tGROsOl8S@{+nRNdma`5Mbsie~^UkPB$B)`O
zfl?|6+Rp<JF!Ha;M?>9ehzPS)OqdXKovVY)Tn)BeZ7W0FCz!-*f^N=b9LMFhKchz#
z%S0cG`O=K{1cQ@WdxiiRq+B!Y9Xddr!(_B?SDG4*fDiWYsE-9tLU1pN;RYtrKS0wt
zC#BW$j;311=vIb4T@lm()=Jaoe>GS(2Iv={C5hx5UOnV^o1ybfu?Ch-!>9<zm1cit
zei&MWm&)YWC$`gN(<j7@2b_~KH5NG@5(ayuLTYTWzfmLIVjrfZb!b-xUVXp-y|h>Q
zjBSXb;M&j~Rj!H)sl292t~RU>?9oqt5bxsHF>itK5kGW11O;1<DT)K8kX#d>R*`#R
zgRk{LsOd0XAM&|O_h%q(4y<1L>k`Ox{#mD9yMM0_FyIqdh+oiOLvBdVSU8eK^tJLM
zTLx<Rq9xCdgVNB|4?PM1$WQzXX{XDdCa7H3hOPESUs`wB0b7Th1y+4rB|l4ZGL4my
zR_o>GcEpR(jZJ<LOJ|%ETTkBP1J2ca5|<irBf)`!f9J`~MvV=K<(h=5?Y9JRvUX;@
z+?qgScqu7qnfx+=0Hm4eU49jmsS2{zGa}7GQ!p0!Yey{j0?<d^LIHr<!*9CnrN+{z
zZG(A2x@`HaElwMJ-R3jfzMJ@c`+s=`z9V<UIVjAOx5%9dr6%^7aJg1<^5C<QyV3~j
zlzqchDt8C{qXi?$8658(f6YQ|+Ch$+<=&X6l-U~h`R0N_)=I0dm=x>bbdaMNRyxB6
z0yg34u?~PU6O0Sg>A?-}?hc;iqeTl>Gtm=&v+bx<BP*aSF6GJ1GAE#ghL1Z5f$j@r
zu03NkpDv2rx7`OOYo+^+Z=3yoo((*FLcPq-LyUZO<6;Z!v?F-VytxbQZ)D{`eFfz#
zipdHlA=e$kG^l@ZS|vQFI7FKu-ZHI*|H3IIOY*b;Vc6{cG$BPY(9#$a5xq6&lfwlp
zbJ%eFC^SN-Unt8B5p~U|8L+~(2o+X&ZLRbfNQb5Z(C4~DJ+I6HSmmQJ+KS2ndr7X2
zZ}RspzR`l}YRg&U-^_tm*Wx?cWo;ZIG$)^zlXZs6EaGT8;w0!v0VRXS8(plAJ=S^w
z3=gC$g${`gwg$wlk`On>eCn2>hP^>HCAfL}s*EL@4GX9`R4>2t-yse{w$N}6xj%>(
zFwT?*g5pciX&e_>MT=}PsC)DoAe}bWu%^0N+3I82L7)S6ojjOeV)P34bhC8Xv(DBD
zD3o4rzUSZDkmX44c_<L!5?<aT_i!3aGp;};8{`k^?N#cNK@uvaVetMV=@)LP23`Pd
z8=>Po>f_MqxSqlY`0HaqrnL}sAyq$~zNY_*0u~JwOo4mC&{J7m!;ptR#&<)xm`Zul
zzauMBaR+S8MBk@k>x-({5B=%1m^FBXV)s1xlVP)wF?r7f0MIuqm)%=E%d`Fqr7tZ#
zqmdCT<NN1Ck*ReDCfCd?7zJm`b3P|e1M5$qW`jI$3$QR0WX%30cE2ezwMJgZFNgF9
zSilC9TfG<;%Baews#^GGU-H3Wat686@Z2tcwFPyyR^{a7n4ESKvpUKvw(VCRbBX-T
zer7%4TxA_v;6aamH9mp3O`)5u3tmgBpACuvFW8ET5p?s)>oGzF>KSP-jhws@+XE{K
zik!-ujv<ZJ00ziu>^`ze6K@$n1Nse+zsG)+UC|7w>Ft0;ZBlGG(=P7>4#w7?EDI{1
zzZ)03C6+?y`-ch8#plU;KEPbCn7O1nL@>Rdo-x({{y=Qv2eAPcmZJ{HhiOY$T$X8)
z3{0~YfJi8jf2Iv$j|T)wZk>+bSrOo)IJk9aziB-ZKep91Pt3_D2KtnmiD+w=#j*QW
zEVmw+4=wU(Ag;jlu7BGd#=b|@@>u|-w>I;fNWqq*1tn8HAKJB-22@>6StIL#rnT}#
z42rF2u0vFH!s!wX8DreYm)k7Jd>$0F*2!1heqqsiHXSl){%aqHsThpNMIMjHFf;T~
zLZ-=S32yIi3=5MK0}f5y;_p=Tum;&~nCh`-A1m7rQ!~d@2t#aUxWh2D5aNbpXQdtO
zFWi-EFtw9U#E8ln7>Jke&M`TKYwQ{$D0lH^6#1FWKs904cm_izv~E~n*2!)TBm-+v
zjR*c`kSMLZSOgP_c2A2~1YrrZfP<+~J_;+sv4^QiW(4Y^eruSUsG1roWe=Z=q)YEL
z1JU*zmWq0)FMBb$oSH!P3MjzGV7+^QF@2;y+C6c#$lhs%GqDX&<zQ{!$45hq4CP+H
zOr?}j@%tKLr@<;X%*})deS@j2N%r%J=HlM-7%2>=Wr2Xa1@fH)qPcS5j3R+=y-xPG
zp-{8W#fSrjshyWW)^cEi&jOE-Y$XTzGhTDZ4gh91$ie;$>aH?4MB2$A{>(0W+R8S`
zp@!A|J$lJuX#>>I$dpRm8cGaI+dn*p#70xI=>TUxBA$UlNe~?JNPk9o8;%ubm&sA_
z3>=N?+<@!q=wWV!L#P%KD=4lU6B}FGJVA~%TJ1z0Z5@6)&Jo5N2n9_rI3qMxfps3A
zP`!!5B(T8HwEXWU_~Zmn!)KueoR~&SrAi4FFn5x_-Hs#xN|LNudGau~d5=Lt6N<N-
zVpu!*%1h+b7#!ve+7Ea|pB6(ruu7A!=>a%BR#TXvfYQTW;o_R>bF5T1L3;XUX0D)O
zx=wnf;h8V4%t3kw4a2q*#%fC`<QZ(2KEvEP3{SyI{NGaP8>6TYM)$);=@-xN7cUD^
zP5)SD27%*nYYa$~kJAsv!cZ%<y2-%Uslw(KIl~wSz{7nnXF7OF&7cW22yIQlO>&lf
zIu5BW6Xoo*7Ah<MeeVVZM5vnsm}GF^2;_%yM9>G@Wk|e(z&T`mNI}5XbKn3e47`JJ
zpu$*@e{WGgs+dV}9L4`f+<QRRSsiVo1B8+Qp(P=e1VS^R*v8oOM3$RuHA}`ObV=t}
zI<}-EbR^p{y_bZRP(z8_<=%S>Sj3b-=p-Q!N+6-vPyz|}d8Ta1|9Ag;*Sc%n_2*k3
z%X-iI?lOD!?AgyWP>+6g1Xa2qrRK{}+j(WViCGzDYkpy$!)17IZgaLi-<D&7h!N3e
zBRoAz5^2eb3QfvLdjVDKnrmcKyg<+k6lJp2QtPW`DrQiYMoH+0r{Bd=KEiAUkUtRY
zCWVYs)Wigd$|TPfA<+jT>RBq&!mI$#k4~+!Mh5*wFOe$iwKm&4RjPxIBiic`k}hNX
zeFciKkW{r!@n>;iU}c%HG57v`kCSmROf_L{V?rgI3Bod)(5W#@*pM>vb6)_8F8Z5V
zQc)s6rv(Gv2gh9w87dgJ>2f)JM8sS`a-g=6Wj#8>1oDIyIn#Q8zQj~hXN8E+#ii_G
zN?=CkP$<lmvx5cn<i~A^#~W_4i{+g7g$82uof1_`jd7M2Ws};NQe8VbtcxiDHlvY=
zF8z$pSR@($MI%YkNe_&Z_0_@&3e0wV+8HSv#yA0kfJ!3u-VU`QuI60qVv9~~@HL`2
zo`FEL!sUWxvr~|!P)$Y?jG_CSeyB{eo=fxf$nPkPHnH0FoEi?Bj70!iXQVl(L3|Ed
zMN1ms_+BS@V*=!HfOKo|5yVpRd_Ol%$Z^?P)S$~stG6|$Cz4#J)+L4X%L2u?Scwdx
z4Sdt)zvv;S$1@G^$dV@QsY~YcLH=@_q?2OrBg0-3+>wL72e+e+RI^MO;p!X`W_I8f
zh>Dr&9^c`eP?(cJm;#Z=`6ihfl*Jyj$bl|7FFGmfs2*J^=i8rSbA=oubS_B4TBv6a
z11K&GN&rQH1tcRE8HJ1U&>JrH&T$h45)EF5;%nOAB_q^$njhcT&T8y+>Xd(JkjgxB
z@0*TsS#a59)@XO<A&1N5!Ap!>MRU~Ds5&yeC1Yf-@ZWjn5Gad?6|@EW=4!dpSfw`C
zBv)-SA!5TT<?7grP=07NGBRFcd$Bkx*ZMa+x*4d$JN_?RPMCY9-q0~{y7l*VR{f=Q
zTvT2z*KIRXZd+E0oCVIhe#BM{rP1?Sw>46-pex!rsk|%fi)NWV{f3}@6)e!qC3K^Y
zLxnIuP{{z#VSF|nH&7gR=;9_D$9Mo;zlwd1>j#ob_^34FpiYe{Jay%pM{NBOE(<^Y
zfBB!>(u+6%`)Oy9Tm2c=gJuOdbINVeZy}46$G0be#&p9vH}p;sr3r834r4=2cL0$8
zH>p7g9K4RhR_+XufGQrqXKD7lEBYvIMM}fFgKMbNfE>(JO=~K1qf?UI<BbAE0DOs&
z1Go6CFW8Ww(&<0b<jgulGhw)%7K_*`#kF2AWP1FfjBNZsIXqux_%KKYN`GbwGjZ2N
z;VLt2TTiMf)U~7Rgv@frzoGjwnVqHpghm5{nRpIP!piuwWlqo_0~ZvAR~MwqKbZ@6
zf7X?@x@<hB?S!3cGy*-407K?kXYGycIhmg(tl6s7_PrqShxzEQ$JtFLTPF)c8Ua#l
zX~@c=I7LUG{>I|;885!TC!qY9Zn?yNp*1AZAegOnX__lTaLsC_xdQQTIy^1&;c=m;
zsjde+v^-SBCWsWK>XQ}T3QtEUmtZN2W@U^IL+e(?UN2<p83JQ_eU(jSFp7O@g#Bt`
zgJ$GySmOhx;f>lklTGO0WNi=sU@A<nv{u&n+muTTTH!InUEC$>ePHxoi3lM|0cvd6
z1G2$dL&m{YxKdd-%SPYa=6PTaX=>c#WU#aBdb}wlWofl+j#VkNDERohV34{z)7~)-
z>Rum+DT*dB?BA9Q@IKpmFvR9sVei`FGjgrImu;gvI0-29TDd>=UU_j*k^DXu1{g=S
zo*L%X7+GCwJI-%;Ah-=sGK~wgu0EKUgDMooDg6p~`aTqkk5Chy01PsHIGzC{;9(9b
z$shcgXElH{nlF!}_n`IRTjt??)SppU6LMoAsz1iW+aOSC<T0C2wz-x`U>@Iw4K&jX
zl~{PcKM{;S3LZpHntVp>Mdz%iyd&;=vd<En9ZyF``4-qJ<e4-Ui^^o9&)O>pOGbs>
z4*63Mw-~^U{5g6oF0YoqB)aBNvnDIg8S4%}@#(+DBGfb@4O}h!>!0_vrzXk*A-wbP
zzXrE%loxy`ykbFQVh>G|zxm8feo^7$B@`APc`-&cIM<qym*OvIS}pQ&YH2ABYnm%o
z$}7gG9W1;4_iflpMG8iO2qt;8hd*Jxz~Ki&hdRsPzi@Ar*P@@+?1a4hBdrS!Y#dj_
zhW>M#Ul<Hw&W?-a^)xnK2yvsw#p4a%@u)KXW>9o8DzxC{{#JB2i7sAm$BsH>qUI`v
z+v7XdUUw%lmqHFhsx-*E*4`joDc<u5+2APbj36c%mdN{QIIDCx^RfOP*b-}QJozvN
zl&`C6Z|RVaqCbF}IasB+z5ZqW^*Qtm`Ploz{=vTBO^04f3uvMOCNNz-3HB?X{BuD*
z_3mwbsW5Gj&uq%M4Dt^l(XGGXQUcAcmd}GIB}0b}l`s4aDDaT?XbA_ze^R^je>0QT
ze}i-TBQxTcF|E2h%4Q&yZHdBSgJH_7m+eX%UC|6=sNa*7?d|tEts}WZiQ$N*qML~Q
zioMX?*$=sVcZ@%(k!pP>`vWFZIEC$8V(6ORK5>*YBAd62UA$2{uwcOjyM47Jz)4<)
zxh3eI&D#p$0^ow)+ALskEJUr&2C6~rYOQs5PJnF}!9O(b?v~zW%>t?kyx}l|%I+nK
z52AFLe7!^gGX(uEkUe}zE`C)|?OEcmXMxE#_KHF4jg{*DJlQ)vBMwgma=Gl2T4Lls
zFm~+VZ&*)AI#(s%EOFSSkZUrY`j&q|rHS^vC2+*u*U3Q=z?}~^2YPs*Knhx9oaMJ|
z8pi>t|Bk<Es;L_1Ksl><Oi1awu|oZhJzBn360Vi-U_gmQcvf}D_hX{SNZlqsu;IXJ
zk#DsTW@NuKQA-lV9E3lNexdOhPM|Cq%9s2ot!OX5LY&@Q50FOo4?;k*X;eq517g*Z
zK1)Sz9Z;wr2Z^Aa>v{q#JkZ;Rq5|TZ#%UObZH$A0wGmZob=&3OAO#B)9%6r>?g0!}
zhk6H$|D*bP2?*Dx#GU;LG*`-D@o~T72Fg!@WCxxBbnkHgnuDb`#(Kny9ud<AY)r_C
zXMpB;1oN`$(vi`62B_vlIVxH&#{Ic(iNeGgl<ufwC2{Fjvei;pmgj5pnGWf1A6K@*
z-V1Bl(cVAdOqRNZom>V)gQwBi&j+LIh!>a&ms*il2f$zk#>*%cC4&+zrgO}=H?Njs
zje|>(4PTD4&eb6fBII~~Jv282YRU;l9*Cpm(78@DW*s=7zntVh(Bij5PEO;7?F0M`
z+phdH*dkvG*B=@ge`b^KvjxCV_r^n|$i|Xs%}$oV_FD#-F7#4fT%tCNjwv$4+r~6i
zy<rK^p)oV~81O8H-;@k<rt|HCpbqXqdPxm8q6}>&5<4P>h3k3&>TA7}#D^LWN3D?>
z6Cv=C!FkYcGBAy`PLxr-SetdzSeeiu(o&mEcjv^Ml==9O3J7Kgcsii5a)0K>8Y3vA
z!oE2d8lF_f(haF7lhJ9?S`8XgM}~oD6k0M}E^y6~YF`HG*_&jHH3!pUjhqrKKmnnh
z2K!CMMi>2#9U$X^uL%V!EAcgRyP+NO)DlPBhE9RiQ8NSL%J_5RXsF$_xPJkmJ<SP{
z1TNIJE@-&Z<aB$rrk>J~Fwq%lNS3<}u!NlH&w%Qv#|pSq&Pu$s4mXgqqemD-pkK7f
zIk5!F_t0-?l2^-3O}vVOUOkv=V~+4lgSff>GRk1cUF1+!TWei31tM7yA}YL8pfhP)
zG!~)AC{V+!v444fl8g^ZHfE<ty}zrTBcSAZoD>adt|UJ$H;^Pb?@do%p8626;8yI)
z3Dz9M)3i33ShCgpRT-)8i4~w$HQM6BQMFZ?j6&U=lk>ozo9!0_CbUW3`{1<)P%?HW
zEZs26EtZyEqvTZ!3XYuX^l?1qemd3mieZ;llMq$y*eOul3f?#r8*Mh7@X{8)5Gg>n
zfP6qSp406%xF%MgpOmB*k5t2K)XWKD&;x#QA{y{`43UuVln}4*y8)oz5!)A8&yeF7
zv!9w8#R-jdW%hqNZ2@6pj_1Wvl@%4s`QF1ER9@k2A2}f8g6LK4f@UebFm^t&?=(+t
z;NUMRaX=orJ}rE7L&i!%CfUE$(OoW%@wGN&fiK{?a!Dc$Pkr8jX6qMA<kGYv7QH(g
zB)H6ZAD3RZ+g%>32`-^(Y88!mJ-)(6W5*>(384kuj?zY}a`4@Sa%Hemq)f1lcvb3{
zs;1HYYVRt}U!Z4T&Ly~pToV)M1*W5U0It_M>GdzHC-2z!w|b*6a{L!@+6}2`os8Sw
zFB7Gi$=B>0*Y#RHqE6H<)za)z<a+z2yAz2uAcxxMP;rB|!?mJOd^v8nH~J`p_oz#l
z#rmdTfMUXV(OeD6vV*IM3I5uL0Ox6SK0q`Mrm>MeB4Fe>a<j4U(BUQJa*Is~J9QqD
zj?1lS`0D2cg3;O1vOvs)a&=p<@0Kg>+mw~tO@vfKYrP`~PIOKe!Kzg)mfytg-f|<f
zmpgli2z;eM?g|crlUN<1aO&mm=wmSQ)iugJF(lZx2*0iaYlKtr8UTU6b?T8tSVw0u
zTnamFityO&kZJzAY4tRLERpHXmdp|Wh%SbJ8ZWq#V+b?TWO!{Sc+YHatDx`CGV0d?
z_%6uoSc<Yrl$G*+SP_;^TW}<lE~8H^D&NiZY528+e%y}b*{MD+4HFZ=G#QRIKWJ5e
zLmXu00-IKAdu@j-v?(A}27&rT@wG~RGj=WZwY?GT!IUaXg04BF8*h`P{t09gr5jXv
zfNaqrQq5BQF1L4>dX~xP)G2m_(}whBj3wg_byBj@TXcI74&Ozx%18v$CQjScL1<QC
z($rRK{1fn)8u}R?xYnO>>uCZ?Xh`c)(@Hl*Ue^0Fx`DBZ09~6l1haq(dCoBtE$_5s
z8P>bnXfyyQMF(2hWCYFD0`S>v{~l9y%24^8t=sV8V!1c<jBIDh3M=Hk*qmTC?ya5x
z!Yw5M=C9RjC6cOQ6pIvqhNnCJ&HV4UXh1f08)=}nVI$vfy~8>(MSky%Xh?wNe)`}2
zSGGDQcLEukClA>4TJpMH@}Q5?e8uU@!M+K#Jrt@%X-OHbe?#P9e>K>1EkrLO>h4=3
ze@MU3vfRslD~|;GkE)p}k2;-ZVKqFgRQ?#ZI<5VHYE|%)kjIR+XPkeAJRaYuE==(^
zU7kp*I#-eM>DZu88e5_;958YS%%>9l8DL}zz8%j`r`00Ofnq^%d&YF{#wqfwb=`uJ
zJ%0*LM=Thd{BsNiN);7H;sJm0XUvlk_w?uN8E&C9wKxuey&#I$-?JStr+X>RlILwb
z`}RG$ue{()NqY~$EE<FPE+IC0w)`zrQ#Hos;0M4{;YI(-oxq6_<Am>t3VtcpMU4?`
z4^|A?zU1XtTvVJ=Jo^=6l5tkZ=I<c~>NHsFRr|3CZYo%v*J4eJhE>Wxf+%R55e!Um
z`A>VDp2P6mdp(heMmIZp--umagG~=eEF*8GT}eZ-874L0%m5R4%YUc-;d$t>c3gjh
zb>H^wm>nzcq^42=Lsvo+o|1Qaeuv}+vJ3qUv{B)L^IYDurb>p6llOywYO^sn0T2+@
z*bfqa(Yr}+-Drdt{tsim@S|leMY!z9NB)e3fp|bMmHNND1zc3<?x6h57P{nPV~pA?
zyiGqbS`GuC0o%o=iQ4MViMQ9vXZH357o2^D{5zJlq^z>E1bg=L)CmobBp%fzMsa>&
ztB3?Wp#FcXjV*UzC;c~84=J5n<x3+*zN1EDr2CwKBcX$+vjcf*p=>wOUG^|NIxFM0
zCfoZnS|H3$xm0%W7U0W6$QIH+Y>8hPselDMV8+8LAv^ku9yEbY=*r84SUY*MRAlK%
zL8WYG`xpV{@S$mwT}CQ6hNMuX@>Op)+op&1*G2|ZM_+8-j;KQ-tP!MP<iM^Yx9$j&
z=9bIvTiPtUc~{7_MwEuZIkNjmzaly^W8nMO{Y_9q8Iq{q9`UtZFrla}l|6%k1j>?P
z%tT9jr3uqv))JiBwQO%25^~#|d8q6&(s8|f20QB;@t2zQzG;6kH#Yd@Ti!iY&dbms
zU1#<U@@v?rf|FPTnsUEwWWsr_o@n10sUQwgogtQ&%FcJ~ckF1Wfv6BQQjy5~#CG|<
z_XB?t?FX4v<p(h?cqC1f{etGb%v!y7_74-Gm@`q^ksqbe&wx%!%mH3^BOOSp(l|{y
zV5H-}gt*%HANjGpf-In@l3SmZ14k;Pa@J9DkT+<ij4muTQk6P6m~M-3C({)fISz>-
z0|KRw<a*W67wLtWxa6TWK~(02U%mA4x3TIx{S;(XF^7NiVtttLV^LdMtNg@9)Y>A4
z`-sc|f)>76a)h^JHxji$Q!pe8;E~1)r9+FVhssgGijbiiw5nQO-(dctk<u?skg1Zb
zAxfAf(m&B0Nz1?;$l)bN2hEW&Har*y#7gE^h3}Z)AM6`O^ALnn?*<tdYXRq*Cg{h2
zJ_gxXu?wO29-G=W_>NoYIDdw~D;V4%84(P1y!SvqQbO)}zlF=kd^y2M0~xHbl+31P
zn4~S)iD^a)iCIFLu8@;PI;=3hy*X2(wp?oaCkKhj%H^l=8T{<%Cp!(-=bw#qn7b0b
zK0248G<^DGbV0B$lEMCrumfGGP?1K8qcvD8=TopiI1NdADcc15b3uj%A4B75g)wQ^
z$gTT8=A86D|0lzPAJjjAvNqfUD9P8!2wNUlZR=rBDzRO8*inbc$esaXElJJgGRhX@
zETm^!B&CVa%-}(w)ZFyfNSXHuFkiX(e>KmSd(*sM@$g-M!?40Aio9TXsSHsum=X6f
z+Fz%9MFKvB@TwBG)WOFG%v&ebF<_h)JQ*lK#{>r*cGTf=ioZafQ1FXxc`*s)6Ke9b
zA|c>I#u;HSvDG;0PK_miMxUYQ_s?UDWpL+dkkg{0%F-e^J<X4W;RvfNk~0!b^p1^=
zXs9*!bH-br=>wrM$&-;+Vua-YX*tU%v*mWs{OmMGsti;D=EylV(7L>8i8Uh~a4}^{
zC)CT&*QHafO^3CgC+9L4f1NdgO#U<$XVPE|hog@_!Ch8?UWqQVCW4Aik@3bb!+>el
z+dmr+_1WN)!2}`vOpuZF7Dz6piM>j<F+yIp`zH7=`j@I1;q*Q+MgyoR64G+BaE~%A
zYNL%E#feZhR+`dER1KBp7zp$YI)^BNMleb8u||53si%Xq1cRvawuz;4W1WdBHcD&!
zSww5M6rvT}_nC!uBQ{B!_sI#<Y^6Td>ha&6MnhK>+lxv9nH1}a)1P^E76D40oZ6;m
zk;-$K65UrfW|6M4!(SoXKrMvUwyEApZ|l2J_&lQw59F+z@2_>K?`7{gbr(3fZ6W3)
zy&u7Opxi9E(7%j@Ldq$f%Lch9MmGjg(yelFtVI9*N6RI?o#5yM>R)R7SJbQ#dl2=P
z8S@m6DV599ax03@hy<nLN^E`F7p{nA3oR{jWg;n)5ki`~C(Z)7%4x%33wR;mc0OIM
zPHz)*(~arecTF@}iui$R)2G~J)TFxP7inLy2;}VU6;M@(=yFu>4#f6Yio4t|<6Ou#
z$aOa3?#@!0&aaPOhStgrL5ps>8r<mfr96iF6_tIoM(0{vwwprKO-A?)%w?0C<yWyp
zAPB=wcxmh9*eAb^PZ0i-8{a|o>t@@sBL>SYu0J%(t=q7Va)R7uV?)|2+9__gb?WX!
zs*3_H#&So}7=S0Vwhcvi6x?@y6USWvfE64X8syHPG~KeXE;vguPP*;g6=NILit}`G
z9_FFDjW)II<HyTAakQyrG9$nBhG_UD)Ys!$-07n?RN5Sorv(SLq7Gt*OpkF5uIXU;
zux6Q&rmRUx5L>iSX8JP&P;D2@3VtXn8Y8pQM`l;%Hk)H~vTD`R<!?|Aq#NI2Tr=h-
z5+Fk+cZhNf6Nt?7w=FX!Qz5L9`TlknV~Ej$6`X>5nJkDo0kr{=xiEdKkt6gylSQ#A
z$@IY`UmUa7Y{RWMw3Y-*B3pH%EKP&9EW=B2*evsBgbXN%umxqg(So(#8f8U#L36C)
z1X?C5;~7pm>MRR!-mOZrS58KnS!9mWi^A%d4}yYm)~rc?BGo}w)|wzF8eA;vg3Iq`
zRSUa#_A>L!`rxG2Y_19Q9^sd?As(St9m|kGvTO`0z@ChZ%aBW(tc!6#CpX*F#shLh
z-aTW($%K#YO`B2y6jjJ+=i@k7sq#Lbo(XyIHx(V_rY%kk*dj1x{jYTU+vd1ory}k;
zAl?18B8_>NN?Yak{vPl1WFy{Bjc8lF)j80^kiZ|(Y_&XK<7F(BJZSAe7;(iO@^%c}
zpd%q~K(o7N$iqf$I8(A}r~JWR1Zc=xy*g4`hmTm}$UKvkM@PEb1oBa_W++nJAV}6m
zgmg4DO|?1Rotovb*fst84U)$bSInY|6>N=b<O$o|9_HC0PkR5rcU^~M1TXj3Q}$-D
zhI4}No=$91(}>=mvEdaFUwAeay*!U9%36EoPl;UX6&ghVz5F@##T3QvW`udKm%sSi
zwmf!ye9k)P?mXfo`D=7wsgQ7Yl;_hYzz=vZg5_U`Ayzifsp4-zjO>&Kc`<h05G13L
zm+Vax^=`zTmY03;=ybYRUP%%#-;}F^6Pu5_H;`dq{;S5?^o@lpAesHO7-<dM_4D<}
z_AUP~aX+eD{%KPhUz2N;*S%5G&8RDQ+GKenx*rSQ3xwgk8Gl1nWX#@MX*el38^;wr
z2HuXH2BT9gZjv4Hj_tHkSh3zs+<;UfSYyCl@}4)P9H$$c87CPW_I~_=&Bi2fJZCK*
z{Rh##=2ZrwLhFF~9en6LP>Dx634rGF<s;w5R9n@2i}PIXJ@{rRw(-BxH#1xxUm_pJ
zGUArAvP?el4(Ytkyz6wzLK9%9Pt&$&C%S97?mu%yrfi7(I~Jo3Ib3p$L=!%@CCFt!
zkuOricIWSj-u+L|3fE(HMn{eR_6Q^>&aOK`2@W^OFJmxSI36^^?7`z(Mk%O<xN5|F
zG2{R^-FBlKm>HXE)fm?8?Ij2ZMt0%9xN+_<%7KxVyt)2m<SU~TUem125Rf~LQv9X_
z&WbyE1Gc41uI)TZQLuuh0oo;A&{KfG4WeJIxf0Y4XwRU?(5L_gk#r5y+d%7-nzDDb
z7rQ%0AdopPyLnqYA!+%fM!36mRx)I0`55_n%r6IF&*Hgzj0(UGObWgr;A<k$B;B6T
zwVGLYkL4MLcwX7dy3R9?4^FDRM=2<H*kOon+s8km<(y2*SaL%bzu~`F<_G8$I2jB(
z`(}KDLz6(Nx=a$Ym2agbScx-(i@QYj9pw-LKwB0j&d9g@ElaaW6XY8CPRw=~N}r8K
z%=mYMTs3q*knee)XcObKdJ0tI5$HNfaX<<Vq|yI@{e_t<2D9zwU)hS<$*Xo=`C%FZ
zF+rWZ{KyGx!CET5xxFBG^?TVrmQ1Zu7zr5cfSz$F3)2u}%kkrwKq(S{HOqlP8n|At
z!_{yJImmc#K)-%+u>WjGq#7~g5c~PKL4)MbG~AUqHYw~DNFVRood3DIkZ?r{N*!hk
z4B-WoEx1O*3b-Wn&FINEc6j<-{m5t$0ax+gb~qoAdgc|_V$P0CJ%{cTZcs<1X{=&O
zFC-0_BYjg3DMhBT1^`aAzw`^TYl;!bQ_Rp@;AmytB>jVs)P!o}=-ASAH7yVhR1gP@
zQaBU|uol8}yG)Mp@sU9c>5rK9fx(QZtDDCKvLJ)f{5FGSHSd}nYx@v*9SC@wHI8z}
zQ{V{JBF7tX$Fbbe3F+en%p|5Z)e~cU5MYf!%@B}tQjABz8SYH5l$M4#+1Rzab1)_&
zKaIcOe%>lSvr*JcLKw0X#lR7}0mTIY{DbZ1G4Rc5m0}+VF<n+6f^e51F^m3#j*+2h
zq)6Zo^27Kc!(yb!n*o@ZrAvpWN9|4Uc;mS{BFNv8n}oXXC8HEat{N^Q?O#pfWmIb1
zcm$akr8G8lE0ldMQdt@zt8p+6aIKWbhgfu?vP@!95g!7nWRQ+j2Dd>(QBpF4oe84R
z&N;n`nDNGeb!5hYl~R?uT!Ahv1!A>-V_0PvyXkOg+&0F0v$LG22JM@?IiF(m1PqFj
zeq(KKw%69kxEL^~s0Cq9P0P*X2aFS8XgOS=<>&EgZLY3^PLZeCis#x;E*<63jkVh8
zsdz?^M!_P*-Q~;}8*Yz8>3gNGi$y|s7mnn!t)c1;lmI)&KT>H79_^`8SL45cZz0>s
zy-+P+%33S6X$IJi=m;$9QmxL11?5sL^N1{xj4?ucUb6nHmlKC`-dY(y%7DDyv46dL
zR5eEG({y3S^W|*_W*j;x50_@@g*ktXJ|<QgcbZJ_73Bo!VL>R)&>+*mC&mUYf@5H&
zMjB%)!EG9Toj4Gi?2VG*qI&+nG#mLQ<QtkLpQIB5$#{`OW<-@y)HAxI#onzfsv0Bb
z`r!jV85|9*X{?ZfO8Dtgh_MVtN^4{=X^UyY|An?L)GLy9=Lu%ZY?euBk)~mGtcawA
zgwXzE|JGDrjq{W#{!D}ZczD5F(2?k~0lH(H06m7jsWDFM8M-eqr^b2lj2I&rU;*&5
z^L-?<5IaPCXTGig4Urc_uV5s`O;97fFr-g0`0*lt!44MwceY#{&j1t~guF^F@uAMe
zjZvdn=xB%`z$UD>%u8cf*iH2I!++&6AJ&|1bHZTxNjrfrPd&iB2+w9fA-}@8k*q;x
zp;s}JiCB`4aLu_gSfqpblqtQgN|UEY3V9YPCMLQ}t~Lg!$JJY|v40~c*hIP3nSrc|
z;yW=G4GCl6Je1wBx6eNBh%@B6G_%DJayI_!V-&e&I`QMabwd!ffB!*pqkm#<yYwPj
zDK{B^VF2dq_bVGjcV`_Wy!<+S!60i81YC4;yiuOTb?TN;5m=CARa9%_R$mpa1Sv$v
zmR_i=+?M9H1OSoxE6D9JQWmGWBl>87i<$f;O?8115sIbUX>XJuI|kBZ-erH0?_dwU
z+yA0l&Puswn>kqKhTod>1CWghs!h@veOIDV&$Jkv#uu2Y*K`xZ#SN?}tZ16Q#LYqd
z7pk?)^lqVx!6G;-)&j^BQk3BApPi;bUt2>Z=J+?LehGN2%hs1^f@H3}4cD*&l8Cpq
zP!FHyZxid7E%QUr03eL~aRw)V;y8X5cz*~AQN<t0>|~)oGZm2BJYN>YGbjO@F)oM=
zSsW_}j&;8i`~u%&Y*xD2F7e+9{NtPdakW^Q+GP1$K$I{rFAFO7JGP%J_ibgTCddfA
z%VaQ(?m*0VW0Dnd{0+<1<62jcmGPh~-->XsdRY}BA8;>xo%v|>sI6Am4$F&80OJMl
z0HzFI6AJ`wcC)OF;S^N^akHNY_rlh(K6*h;)QQw`HuT!ebT?5O74b^h81lKO72d@e
z*_4Ef<wRj{Grd&1WOMY_STjj}=k(E)(gv5xz4i>MU9J^*K<XQ6n&iG{9{$Um{}$tr
z1CBorMJ<BE3JvXTnLN<(`@Pj3$F|M#`{=eEEV$K_kbwh_lL!1=DhhDEg0jCt9vr1X
z>D~WNMV+th?v#gu=-_QAFiigaupQ~+TICNhjZC(oBh$(>u=0qr3ykF&fkGV0whrS(
z7s;bO80hz^Sg_jiADxnRt>W8ca^6lKi|&gi0nt&*{o~#Mt=dfV4KtHG;m_uFuheI_
zm&ubs&&J%j$de#X89lo@x7;F6M;o})(tDcOX`hMTN7Tx*M!Y(N2>vP91r^g<<j=8L
z)Ja3NLae!3Z<qo@{t`<t=$O9pT$&b*TQMyrANcF2t(vGH1=L_0&>~BpLih&bziae@
z4&6$gk1jzzq?LndiO>kMG*bG^kr!N&v}C8s-{M`U`czGM7*m<0p8~IiFWS3XF2mL7
zrCu~Zwo;nF<mH%bp*@qAS3+@wRiwRE{_aEpF!HK@OQ$9J)XxFz`I<eWgDccv+(U@V
z*^vPftyE^%(7hH_DS}2Y^ULco3P!@%Smll2_DUqh(5?Sww2kX>TLbJdNW=4%jTLoq
z>BsZ7)0$LfggSj&?WBAs*7cBs<lX2^0TAj9YvsL=02Aoy2|(}tcz|r7njhtZpvo}D
zFMsGXrU8W3I2OU;C4k}lwElkNeUK`=j37~<{3|GlGJ23PAG<^}FB`o}+hFl}k5}R+
zF<5lO$O(oDeHueyT`~DANg#xyfyjEp!rQCl-~KJOJjkJ5m-4yo+hL>Ri}W2OmzmZq
zSIB>2pg^yY<Kw@v63k%&IaQD^ZSjEVBHl(y6?)?}MiukTfLyjK4J0ZVuYGRH?6Q5S
zLvu*>HJ-+h9ZD5*gJB!TAIf}Nn}G9xHD(NN+%ekh<r+mLL+vm)Zgz@s!081M@Gu`N
zlAZk-d{uNSA?QnXiM}Tx+_$l@M!s6A0AJq$1LbR_dUrkTG}$$^M)!MW8JH)#l^S%%
zC05HN&{yiRYh?G-4>1C4VZd5nFICtHryK%1WRG|Qu+wVL0(FmNPj9cc*8)Q|X3Jh_
zuJC)%?ap|YR6kMn_AgO&h$F#xIAfM4+{gRjWsNZujKu#&sls`a5qv4%O!K1PWZc(u
z`BteynBASD02J)&P17<%^Y20o!1mC${ZsNZD_8Qs{*Fx&XBfPO<h!YJ7P(PZG5MZ7
zt3d(<6kf8FLf<d78((jr-xP!1jtf5s@~A&4@1>mTW?ZTE^L~iEnLqjl`C&ZMBwF+x
zT_ZnoGTIiyvZ-!DX|J+D_V?eBUQj)#p#OtpOsz)!eZCwJWX2*}R1RH=2vYyp80CP|
z<-q@&rJ@`q@sU(j4l?e#;KZZk;2_o*-~e)ne`vNC3=XIt_b65cQ(Y;C+J_UcCVgUV
zmUmZ$z+wJ5m<LBFS(0#BvXaYB>~kg{J3OYL>bW66wxma-hLPL~L_ws59BJGFRH-0G
zl{z#F-z(mQ5we=TV6Z>CT`2v$ee6aI#UdEYP}1Ljp)r$XSJt_AYnL2t?R0kzWh4Zy
z60m6wF!~IGUWG{?W8<i1a;Sl&ir?XuVi%4<X%q}jG#C71{h8G?A!{s`<E)J;`XwA6
z`yN$NprZhr^AE71*tB{fmJ^L?MdihEQoLUc*rqL8FDLtC=zl;Yd=|c<cDzP@ntowg
z7Y%#X$s!q@ZLM5$9?*Z0jRr;OD`l|FwXPBRRGSnBHIObELLIl1f(&umU)OzYH~+s=
zhT7mxMHb>=i2&*zWpS+57pro#$O!9;YLAq}{_b~de;Jv^N{-P5B&R}$80Ek42%+`Y
z%=oy=Noh1tJ)~61VlA>wwXGd_DNkhN)d)%nRIv(YPK|;AM=-onIRZjq35>y&-W6pd
zVNnbFpgZ^I=&CfEmnv_1&i$9-@SB5^dbw1`%wg-Ldp7sj82bnoM>u$$66*#xa2zig
z8S9;GQ*cHN^y5mmUJ7qQm_Gjx|4&XW-MXueq__bvL6(N@pC=m8h{AJ*7f-rzoE9Vi
z8U&>cWSK9gCsJJk>=@Z|unK3yN)1LhH+IyS={v1U2hz=!vy4x$+iT=(<4JgO(tLJK
zysZ!j?!{=a#>WIj(yag#i#+XWOSgVqCjhU(9O_D-oD=lsIv4bFaegugi!8;1g=m-F
zlpv<pNyZNUmiAWGpp&c(2{!H~8E<m7wn6HHrPRHXz`ZA)^3$my4UNSM{D!I*v7EJ6
zNBClbb=BQj253Jk6Mc?m?6io(Mx#QubrN(PY4SIiaD!$Z9t+LhmC6{*$MVU-D^&8<
z739<;X^E~TsT=sYK~4xJre&RnQ>4|NLCvX$w<W+7ki8H?vV>EBG&W0{HAQ403oAgo
zu{$zX<B~hcIHwJ+#4MpjW9VdWg4PpiM@%k1C6R}_)ZB>GFee@Ub{CXXq>ut^J5#2{
z9ztwup$$src`+(gO}XS<a=t&C0o<%>3;qjsx*$lS#v}$UFwh4DgyljXG*ZIV%0>1E
zKr3w6i-S#TAh|H3OKbt~-3;7DE;W)O(6%6#rJ*v!jt-<V<Z^#TB^QiOw<NhDo*_XC
zd=_v>&5<jkm*Pek46|~T{aw+TL-82&dUfo;_U6e@3%T~!81q#%$hAH#Q&M>8WB?rO
zyYh>k6SCPLxL-P>nv)fwOnN~xYk(ne*Tq2Kk6I7?`FbBLuTf--MZgYj@cyhkfcm>r
zH-IR<(ch-X=8jr|{c}@rnmP!}uhJ_TGJ@^9PJSH=S_~zukgbuMjjXu>7WbC;K=lJ0
zYv6!eZIr1V-)->@R3$=qaQTtj?K~<zrAqEdKeK2YVt#O!{!L7DaBBxWjqdb!42~p{
zv|R48ahb;wU%K00qLu<p2<|`A<ep$ET-52+z%XaC)%-SfYau>Jk0_n~w&f4kI_N5+
zx@1}qp*D}8Ff8!&Xjt!WT~M&8Vbbz7qt^n^NsXDBwPTr?<j@RNpf#x$v&3NQECGHz
z%Vs?kHkuX$hs*33B5^bj4%Z2pW9*Lz5hOP5@)zjzO*jrAS~54d3QCxjPMMdQA$+Ze
zv&;_y4@1H}sNW0x9bQD~^a&?oSs33b&zj@bqEg3jndQ&YGZtiV5Ql#D)W(*?7@&Yb
zumC(+8n28-Mb0*0mdmVh$S=eq(uW88XnE`-Iy_>7BS2|I8UvG9Qe~PcD{YhG;($x0
ztcpDY<vSIofSN@Bo{rC}gTt%G6pa}wYg|Ii!YRH~)~0FChXmjaV!5n~F7?>d!&BB9
z%cCqtPBx_9nInA;r0f#e7^`ylfd|Sa`vYNRbqHnkx4A{B`K<<Fe-|7A<z^a|ZB-0h
z?v1y3@xye15XgP;OkMWPvww@Xz?%(0#+0l@x@}=$s%E6|O1Zxmd3fcd(xiA5{>|{T
z-v=Wg-a7;EVr$&5m}mps3#<?Ar_dODwFF!I0YB*AYzyV!K_B7V`(_Z5gKS+a57`DE
zhQRiReXw?0p^vIV(DDZ#9KBN`I-*xRggg?2QaFb0mh?|&^1w&EOHT79++aD?KgKY?
z2?S22V@E$03xgU3@_6u9HN3k4$H7tSiA0{B)C)ZOB8k$ICdY@i$y3GxTdv1h{dD@O
zH%6get&o6o<(af(xU0K)=~<tP#wvlZ3!ur+e7j8klopfmQ5Ge|x8RKzybEBd6=iI_
zSxP}-cua)wTs#ADW}-pX`}}Kc5cp_x$n!x&xIoZ7!?OL;CFliPBUG-FzxB)rx}utp
z&;h+ZW`j-U$cx^M0f)T%FO-*3*95W@&XSjH#Mr!9T-xOoBO06?+U4&tDwgb&SM8Ug
z^5H|}wGbB*vmMy_nS%Voxlk2U<v(22uq<>Rp`(Ygg8b75z3m;`mV||>UXPhW8&G>3
zzo#sfH*743<vb6$ncj?1*44>d_6vGghlPAQrikURXxIhvj!`X(iXHE!HBCh}&;d#N
zp1su6GD+U|9eF*N@&;&bA0$%J+lUfjiF{~eg9hKq2<VUO>#B-!`Bx9sYR1dQ_7^ct
zK;@r!OS7OqF?%2#Vm`G`Yo>O{XTk2s0R`=kiHiRnU#Js~<=gu_=BRs}C)GmvA|{4o
z9hWR@S@4!J!hb@LpW>&>f33qo{rbw6zH0x&#)EIw$`6j6CsO7B9%hSa6lA+HLje_c
z10OAelfd@&th>`rF<`+N*}>m4<c^7(*U4A>+3fCR>OxP)-W|&fK?JW7Q|P(<U>Nk9
z+^Nj4I5X2yPaxC{e@NNcTeE9Db@X|%OY{n-62v(`Qf-j0mMQ9WBvJvam#>v6@Hb+d
z?CKp`Vo?M46&A(XKqHmi%C=f*P@>WevU{0c+=ez!%E2jvNTqz;J5%Khg5wf!@g8Lg
z!8PV;3$kbYj$er1Gx~0?GQU2-bm)mkK&-vfR2au?LUW%)m>y%jrGdls1<>m^$`p1)
zZc;)H-?VjVf+sI-7dRTeWliG9#x)a}MCmiJZ<(V!aNoq`t*Jx4ZNtSqCL`ZTl-B4R
z9)8q6zy#m*;hColt*`6kd;W|78}1pvT{p}3%eD^0-4I+stFqmidYg(}t5)rk>wPKS
zVR8o^sx#yVWeU&M;WoIREg9|Y1^J;(l!5awoc+lAh1UUi$oOji7#}1wjgk<!a)3Xx
z<)s%8B4OiO7+;v+3FDLleb#1R<sFI+0uWzvP<k7Hf-c`2IoK$O(>FIMr%`i<*ju18
z>U1=~ng?_4q24C1MZ9V8tT7y}kIxZqjfbUa)wI^hPhuFV`r0QJ0CITjz_NCh0cw#W
z65|px4*wPcn>=Jz$&uE_DOKfiRB#)7-Wk_o2@?9IiP43HSNt{7&!5qPfg?rD3jK{a
zkWw%&N5?3Nkm-u9n*;m<iXdi>!F^JWvCTDVung?=tLhKfn}gCbnzJ;ggB)9C0C8{I
zh#L9OVlq(6aW<nuORF58b_Ehekv+Myy61wG^u&=9yc1mG)z^wnDJS|fHcWRFW@kOA
zOo7IHJ<V6JOr4y*+oTXXkTLmbY?pSp?;)BYBR{jRP&lWdqkx4jN?S&&Qb7@ij;V;#
zVE=_9WC?&&qC|>&F$X<@ksAWU0M%tkj1xK;v^gO1C`I9GfJl8AmVU<-o?l|!hTBd-
zT8(x%=q1h|7jtjXPjd<O1b691DG4S+tS%kARx5bD6xXwn@r$w_ZlKCpi=ncN@=jF}
zpdB87rM4c&9CNIc1>MjXbzs&TE)ULzAz-{z1kGBrwYZW>rFW$_Bs6Hb-zn-x#~^Tp
zVbxAT)ke)KpBVIM0%5%A;!Ap@0cM0OtPaXR4`4}cO4>0!{VL3LN{G5M50J6(@3!$W
z&fCUC69R=t3n9z}=!eSPY&kVJzio0BS!z+XA|4DzGyzg5$Z6g?4=1b}PhmN|Y^&LP
z{lH>64LGEs2l^O6nbb2@b9BiW&S!vryF0gB3Pt!#rv#l)jmr^Ju1W9DvXPG~88br8
z4my`oxS!+SkmIO%X{aGqWtr5(_+Uhvj&>NlTWjB{{=Y%$jEfLL)0UTvt!@dR>1?ky
zH4h~Op{>YxZ<EJ6MLl`3-k%}KG#!MYG;flISOp}jrD)7qzsL}pukVlvF@xfYN>rY;
z@4GtzUF7PJQ3|(}#?<B_3eo$Kn?joW8O}-URqz6?nay!p*8=lFwOMNziPr%7NQ+Oy
z^2n%Op#+*I=lbIFG8wridIw~+MpuZ+V4=j7V61|5MSWsAjIV7mhXNAtN_$$lncxL@
z4}p)2A14Ldpk%4~cutO~VFHwq%*iz+o`H!Nc@I2YIzkNL!p`Dj1(|Bgg>+Q7zMmKU
zH|A>*1nU#(337!XP`yyiHF80(6{X`>&p~*hca0~&;F;5Oy<8O1r3l`V7aQvV?3gH*
zcr#>RT1|WM{9Wo2UD-sL4QMtMDhg{{gj62MWx*CBP!<W^HJ5wmDDQ_mq7@du!k$6c
zqSzg>^pnZ2^y!&a!Sjy3pC~PZZ|E9zRgfRR6l~<U-$en|tBp)}dlRJEHR(N&0OMy!
zgIzYRO=Q@l5ju;p9e-iEUw0?cT{Qz$0HXBEL;(6}Sime?k>P1^oj2ezA5&j1*ZVUI
z+F^*#BDukz&4qRkpF{e4$c_Gtrvlg!Xwz!BDb^R2K-=V3iD4-nt<Da-#>^oAqWRZp
zVl*bGECR_XU2l$GaIVmkVKrot8kFUh=wZu^82_zltSdlVUV3NAZLwTcc|fQcxjlHf
zzR{TQj-XHtN(IuT3H4}Wevmgj8J6NSyVFNM7nfD~Su$o>aTEL(`O4+47)d2UKjdy7
zF_zYmeqQt$N_UU{LfyskArd8q=ZpN-T7Y)O=+I7MS=?M|WLj*anmVXLxTDK-{}pJo
z%&-|!D~D=dkeT*dpQ8>vRA$Ah0E8tR4Ciy19TR4Pxn|8DJEv@`*O!N-uf4Tab#4@v
z@Uf~+g?Y|%50X_f*Qki=9^8BJw#<v^4=EZ`B=bY00ky$xVnMWnc}9A2!{-<JFZ9Nv
zVb0PS5nEW~zie}ksHI_XtY|f2uyOZa;wz&qVr9%5mJtFb@0SJ*Xn|B(&yZyy9r~km
z>hjb8?{T_JK3O)oL{`LXs*$z4Lsq8Qs9#qvWp7pLiJ@RrQWnW-e@2~wq4O+axF%+8
zwQ}IJM?r+bld?9}W{6hzpnA717`8DBpcST*^{M~f`v~})Z19N~4U{j@ZVyDc(J0rB
zvYe<k-z=N#Z-nR8Xo$z=Sj6hKT&w&pjhwu$p7nBXa3aJBGuam8KKllh%4mbx;@_At
z%`6?=i7J+Nn&<lcF;8UmZp$HF4;k?6HM(WdMu4doWNQooiZ6{cEZy^fPoK&H&mJy<
zE%bx&He@FOQ>>mm6a&`aPRikjgZQ{tG5sl9p$J2r+SD=~NGYcSQ+vb-+gsPSFHLoS
z)YwUVapjLjRX~O7@&8NYvEa2)xr{s>t2q{zE{5(r;hox{Lw`CZErJ)gJZY506j2gg
zp0ZutuW$bW^0ZI13kotrbQ!AhOfV@{259xO(N88%6EoZcV{L0&;a&Xa*h4s%+w0oo
zFEPdeeUFjnd`S`9k#CW|M#HT+*n;Hwm@4FuykLXs?i^PnfAc|6t)h8_CW;qL9~+2V
z%`XMnsn>z*Juzl4TO)av9Dc=G2X2AY|GR&}BdX`5V{q%MHoHmW6N+_t%{wP*Oy^b<
zm6Cs0yTv)&QUB?`n;Vlw&%N%?U?S5y3!%n0QaeZ~ioL^lmp6?ZLs7Ayb-cV~f0f`G
zjUDuM8scizx>u<VPsIP7)Y4i;0YLj(Chw+a=FPbdl|kN1&){@QH4Osi{qzjFDY*e$
z^g*mxOCu7DH_C_p4!|tP71NMDvP<kKP4X}QH8u&bSot_f2Sf3hamj=o<?cH9B=#h1
zxD(p7g3G7zXe=r!Peq!78u`rTuN!BX{M*|!#gx4f`Q%J&OMY%|!)8twnJ?_GvGCGF
z^wED}Ycege#wYwY_5#jEoby=As80W7+17pk_g)Yp<q9YbJ+-m|5iHx4J62T92zlhK
zm+kHM0+d&DyhFJHT@CrhCizNviWRWJ%M`gg+UrIK*{NJXxONz?GqQ8Jd->><svx_R
z8>VQ4Hu&wU{*3a98HnfLIQyDE(+C$V^D?MbyV|qvPFST{QOczi+jO^byL{m*0HPX8
z145vfgZ`J9O;`n6Ni4fZo6G}>1!3}uuiMa(WtTqUd)R2pfUBw5dQbbcyAzv!qU`10
zvq%7QG%=I(-u_JetW*bHD*G6{x;w{I$~Vd#RBgy2+Xm$SH{-`4L*!f0{Ly{;_LY5O
zdWsiXzP@jlD^NCQ&_MZ4{HzsEu#<uOeK(#V@k@9Le6QS4RE=uH@y!G>s{G6Mqq{mX
z*AHSKC8K4(_@x@onexLlCnku6eG%C0kIDnPCClOl1FsKCcNV{7UY-5@8x~=p$$lxE
z2@bGUalu#%_t_uYRNC97%7NvE%$CVPMsQR>GF|Lo8x*o3737d$9T<hi!}0x4`>iTJ
zxgdSg5SQq!3*`mmFrPRVmS6~>O+A3*C&u7#zDGeg-8qLFk8;yy<cRbk-AvS56Q9$X
zb!5y6v0pUjQBWOKu3&L%2TB|uu%)kmZ0C`hhnyt+?8WZRqN!9v`rD-I$IH>yDj)!b
z0SE=i0Po4v4=DS@*n$9XR%2@8f#vQ{gq#U*WT}J59~6(^Ytn|W`Ak-hwW&2Bi$jMT
zXMX?{L1~KP?T_wGBzKn+Vrps|w9v^DW5k2u@zWtE+4|wumW5&X<oIzi0v<E+Q(FzV
zlH}!Q-WE4CooJzFNl^@bI51C~RD)xJ3Om_tq&Ph@@3eI|s)pEijMUd!Z$q8Fww|yu
zX^a?q7#3fv&BXHa4EMJo0&wTSc+|#Ed)x>cK8@aVNGWlCW|SN6jZ8vB9}qabLusBZ
zBV%`rEtOHmx%t{gDGhof4Q)%Ku1Z<Zh4;m*l=}$Gs0zy~C_=F`S1QujXb?yEDsdv5
zb8Drt+@MjycsOIU4pZnivjN=M_8lFYG{fBVRf*=xwCXKb?UTYS3W_;TDT2n-A^{KL
zv>uZv#+WU`Do!z`XxI8dV`GTe(~O<n2z7~ZwhLvP|6;>3=}l%|C#PBiaEMw3z|6>&
z?Er_I793EMV<?QAZe$$LZ=jrE|D04)CujPY7%@ll0N?{T%Vr7!Yg}v4yoeIN6dWh#
zSU0t_6xYP0aABYq3le)st&i5mNq8z^!Rr#)SlR`^F}llyjGhb!Hy{v_wIyXF5{+}?
zV*uFkZ44f(lloYT;f?TNYOqFWfKf3_wK*4uD(u;8s2~$!`}H3<KqguPSXve=XtbY7
z%1dBbY_gwVaBXYokY?wNkS{$+38@nj(}zMXc_U~&Hg=1V9W|%hG_2;_MBtU6k8SkU
zSZ63~L~Gz<gOSjGy<iPBO=^+0m}42!=(HQjDT!oK56d^>CNkOC&Y%pa<cJyQ<*6f6
z(h4y<Afdcv($TX*idn0Qg<ZHrrdme`WkKate=7`^#GW|q`EtIEb5P%-<$}~aJRl%_
zXy&~zo<VWY^i?>3E;1q?cIJ5(94Qya3vlJI_}IB}NiZzJv}$lvUmCrG#7sfN|0=L6
zm)X}qiE2=a&<Yb@Zp}ABA<D=VHi;%iuv{5knVdpx$~AITu++HR#2oY_2*|6;lerK}
z2U)J*#5LAX148BGTAOn_5{dl6+cc*?8c^oRFZ~&Hb~=|5Ww|a`+lp(q%Jn`cg@(Zm
zi0r_M5NNp}G`l*4%dn<04j8%7x`%J&Sh>mDw<KShQAU0hbHkzre(-A}1j;GGPyJ>a
zb2U`8TSDaa#>pXKykwZ%>RSS^9j>jn1%>EP*bdAD`P{rSRlhMoIt<+5Vjaasm=OFo
z);IF4Al>?%i7!zG34CKmhZ<DC_iN>@_#IM!cUsLfp#Qc^19#il^C*aRkCV+zhm=*o
z%am-tjlF=Va?H0g5ro;q0FT2KGtEEdal|wxGk9C>rWt5@Y8D5Z-k?3yR%ZAwyf4#x
zs^=h^nR=ioP*E?l;u+9c(`=W@>}VTsT&i+P?_!zb^XAQyQS&??R8fLFgrt<N)Ecdh
z>H{Qmy*1U&F$bn^GaKZ*5Ofop!R^r1WWFr|>RMve7o^TC;~T^sCQr<hr{Z*3=$$b#
zbsEWnvwBf93B3>=v~W{fY)dg^*f3cVj8SH}?vbryskgyU3o8DU0L#+iqK*KbvCc>D
zSc>I7A@f`$tnFqrSrOC6u3Dp$glk(o(?}GGqvf(HEztr54S^|nlcqgOR!84ZE_nH1
zea2Y|pBU4wNxuWT$2W>NVVS5nv&7Td=nW$<%yXfv^JhG=c%E;76j`5^7$__F5e7Rj
z1oK&Mu)PGi0^^HpO#NA!CVCzaL*;nkog|w=O2FZ`G2c;;%`r9Ci#9VjNE`m04};UC
z*$l-S<lZz4L~i4~jo?dz4DL&B+eH_gCR_X&{Rk+g;QuMz)<SnDvVXAt5u~3375Dq`
z2SG>053np{Fwbt&*(|H(?@hNpaevvmO^*!gSS}C5^i=@s-UhLVgPA|&gZ@oa2!a+U
z4}~;r$hEZO>rmDbN9)6Jw8Hbg05{)1Shp7(B##)&*VH#deR$OSGdYFOBg~WN8ePMn
ztjQlk(nGIrE<nDidi0nzj=WuS`2X^FYyxD1%xOZhCqfwKvht+84fq+y4-TBCY=a!z
zcc45S$Ji<5TDIbuw3t@HNqLUEEzgGRqK{#X{3&)8eMMVUdH8es)-o!pRTB<3i{vjp
zNzY!bw=Fu`YX&InzvqmoFzMjhrKd@U{53{anuVKGMxKxU8?-R-3&ym_z|M@we+yAw
z3{dUG;83QEV};6>;#Vl)^0F_W;kfYcfmQaE7&%0mRoR)ZdwGWZ-C4@?CuIR+ZtMw5
zhOrzn=+(4{8p^K8KU8@F%Pxb?{+g|Q#T5BR45c9t;4FQnH)L!wTHk5pR&J2j6Mt$*
zgJK<TgrZf)6Z1`fL(yCCDf6wr6}$j*DNAt_<n36i3^MyNmMJ6eI21ySkSG`|PZ%pp
zz}tO*dI63X{CBRrYh%`0WAa{*Pp=;kQF!6~SSoM^6cPEr-+>fJ9xfloxDig9k&j}G
z#e}P1o0Wg1ZveQWW0t7?*p~&wchSazF^t0h#3^TrE>CL$yRh_ua&&$Bw1=`hJm{H(
z-;rV5chzUnn*n@Sq<Ql17*nAE2NN$LpQpDFtfBkAOTMsu)wgfIqvb!|iwRS!jX^0F
zxvu`}zpEao(hDX*tTMq*=Ct@S*bYv?3!Lys8G;QX6$<4vA;a``6#=G%Q4^;GaNf|M
zwy#il4h=7cy?ciWh2H?5X5}mPZ+Lk%%Z?R_v%zf}MwXqtNjqjtI+Jc14hH3L=Zf&<
zo2AA;m;_~)3PXyhUGUnD!pj&jkB_g~SO*@{U%uuq8Nfx}XIg0NDcRLW2@wvkVmJSb
zY&&0e_u&w0*8pZ;>=@41ZR{*mnv*@Ox1SHKmOX8LNIN`S_Nq{zg~eefLbco5XFwUs
zFd-J4+{d4Bg=lX^c5?Yfg`=5_T7vgFAUB_aMXOP3VG0a7S%1q1O;C8X?CY&@k}M?3
zIE8PgUe_>oK~=u**x;d3qweE(y>ad<9!*SrYlqtRqB$s=ym%pPntb1%`Q~PwdijCV
zN&~#G8IZ!zRY7bLn56ehL)ZXq!n{d;SmBoiSe2UdlN+-zH5BAWsdZY&)CXPm4@w^m
zN6-T*9F0Y;7^V~Zar%IzCm8iKO%99?iV73upjbzEiE(y}i8t2@yQv@Cvqm6?98%#3
zRdK!vP#Xouq1FY@fjkhaLLSIopR^zw>4Bpz1xB;O5;YMTt^F~vE3FDY2?obfGkTeh
zdgg3EjcBGE?!Oy*ko9RNo}oGpD@h#@Q^ha*9<-GsgB%SISin?{GIpR>bRCer_-kwq
zV!59&X-hsM{i7*JO4ZGAYdt!s*1!L8G9bM$AJ=O_OdI5wG-n39Fc%gD7cCV7t>r$4
zA1Z^4{UNPd<XHQ6(U^)VInMrl(g8pFi5%~Z&jEBqEoeTJ=o1oga5SM=gfnkLoD(Y?
zg>HeiKwq1aV&<dU;i^?9C&%AIkt`CO|1{S1Q0O8{<!8YLK4YYuASSP7Al@B|(u5a4
z*;ZXg1{*~X28?`U#laYLd3e|<95Tdy*RpLgG#bF=9{yg4-n~qQImNJ+pj9t_Fd6P|
zL*y!K$C|k#V%tNhvPhtkSiddPAfHAC*>I$+G!n_Ev?c}utOd^z_?;%D!H2lOk?)x-
zhIf-CQf95zG&UimQp)Wwb@RzcMf5nNq<V~0+Fw|r<xu)Y2ZbA{>lCEQR{fwpCrEXL
z!+&b(GeM%~K8DE~qe^obV>~d7mWWgA<1wYxGS+%&VWHGk85eBW0&#F^nx4g2DaxY-
z;(s2?TZZ7wf}9p#4KEubr#qnxx+EAmUC!t=!f;ZH>?o+==PQ0@EJ0Haa5P;j&We7<
zWLujGa<=uu0)kC)PVf}4M(Rk6A@Qpaw4#MlV-p+CtRq>ewHc$btkn6mDD-iU>G2bk
zOz;JK=@_feBO=?g?tB|?&wLpl+=%pA@Ux{^xjtxxMwz?ba`w`@xS>Lg6|Ivc=S}ex
z;I)o7mkH54#6J;a+)rd;%o-&)ad&OBh1)U@qiISbBtFfhUMbD-I?9Itb<g`NVZ}u`
zS7}L8UV_n4H=)xcExYGB6)6Yk=R8kyV$k$~%>fbD>a9U&)6}Y0<3h|8#GvOHVoC8|
z<eQK-|HbYs8Ze|1X0Eh*I}ll1Cj?<Gmq|8oT9xxqK_&->*Fq;pn!716uLAuOq$7Go
zf4q<rld<!3YDgzopQzo+c}DW4bLIRP5Y-YWeHbkMi)bdfz!;*as!T47j*9b)ZMn#}
zY|HicZ77qAW2v*vQ#;^Rg^R@{X$_zP==A5wrD=LhZmI%QE_0?g?^UV|c8>s5q$if0
zfS1SMv(1gP@W~b46zgR%DvG(Huk>eTpTwTRv454XoH_bCI)xi5u^4!?$<;wl?j__j
zyCw~Lo!*=AU%A$ZP(7wheqlcYgT@&PgG(Fe_{(Tw7*ajpe7nx6N#KVL0tSdNpy_%a
zjCq;ioaYrD!iP7A1+ZP*kS4$j4F&uLxiRD$rLf%Of6>@COMaCWo5e!RG2e`^zYbZ`
z@3?+)vo{MkL+xwi^;@=Ca&D1GyK!r*Gt78E*l&xL#^xr<?a`8!;j)<eJ7W3>iKKVT
zZ+w#EJ+JnI>Gc4RcX}Ht^Kb}lf{WZ;X*37~h4^5=jUGsHw=wC+3c1JMQEt{2Y4X{C
zHT<o&XtI(lv|c)679(J6K<U-^7~aj>MD)q@^zl3{xL%)T*n%I|?--eB1kafbbCx$_
zI<B1|a7l%Pm~CXjsSgP}C#DDiNBxSLQI~&8(=GEG&euQ&UIXTu<f=iaugtT5<BHlU
z^HVFdzLH%x%YxVknJnDlK<$OG^Ke7OiO<W|qO<^v(8Mm%tH^ZyFOwiGwxI&Gs&AGh
zq2ks7<e*oAwz@Q0MT&Ununa$17Cj?3Z~?V1WqCTSn0kr<7W9Q;P%o|FOfGW?_-utw
z+wPkdSitjko~(3MX3myrvdSl`${Jdu;<(x;LSK98Tbq=an^pEQOnQy~ZmzlXVK5s5
zb!@5^YZD=?_Awz!LyK71c%7_EOHaoall1Ft=?4txCmVc#T@V*5WMdkjd4p3M+7ztc
zi24Q@M7u_Vm=ZNoxIoMAY(+=VDfiy=JJY)H<Xk5ARk*81dvi-WvP8i}A|qSuMTSLW
zoAYfQ&^x3%Sh!o`d^8Zy_Q?I#MR(`0YWaQY0$C@p_)HzX)#h|;-vRPKEQjJX*p|dh
zq1i8x2mM1FVbBB3Q}&_6v&_jEk_>Ji^05D68$rGBKw56RKSU2;2;R}u)=IXJN78cf
z4rd;evvI)$9!<4G9yP@#>db+;{}}WwLOy9AS&sz|jx}#Qc|2yT7iLfc8qE_%4PsyN
z<Tk^^R)9pp7KhsVR16dPJs~SzWGIBE$<y8@qsk4Bm@dy41xL0w%d^p?$#tBSYvoVI
zDZG8PXXMZRh8AYixlNb9BvvFsj!Q3grdgoD1rQ(4rB+RnQ>plCYKMxDp#jgkWIAS$
zyx>1`?7WAdkh%OVvFt`19WRFHo|Ky;FU1&#Q%!i;KemiJ%!Id6Ua8o+r+PE7_$%O8
z++j2-@9);<Sw|cpuX_6o>suwS#d6>Sh>vFZM-WxPSc`|I$~0I0>HVqMA3KBT#9sGj
z)GeV5-hT;}<&9oE!Ys9PD>AcW7bNq1dDFkKlxY@E@>XyOvg{Gkd^?5(H{@E@v3e)y
zFr>U%-t{qBf{0MBDTm)P#;F)Lv`XH$Hk-0d^ia>r2SJ-L1d~}x14<4~#}Dmz>d47Q
z(XK1nL?{f5{Fnbik8PNU0IAV>LKD%)J{!hSakJ6~^^+KE)zC8e)X71Rfe0RxR8h{E
zoo%svW(_F(%M6bc|NoAYZw%7>HOuGTgn6zTaso97qEul&`NI0E2Iko){|QoNr{t-`
z{Wk^(d8t(~pn`lEedY_SO`?V)sZ`W$9KzkVt8_dGZvAL+`}h-{GDva0L#3i@)nhQZ
zuT&aXX0Vc7zjmxN%&vFZaf6Zw6m5{5DiwCp45l=l7G!5@vEOk((RQg+M6bKE8u6I&
z)k*{2G%+o9TaSCf*Q^!TTkD3H|NX9&4%xKjQM4)}yZP8ufP&;JU<caW+cQm-hrbn^
zMZ&uDw(eBr<UOJfhKsf$t*Y$lZy72@ZdfjR*@%$?hWTapP9tVJu0L`Q$UZSMcxx2o
z8<h?K;VkA7;9Bs_O2;0`;H|-%@3$($c^;MlY<cC^g6!*qV9E`qTx2?$Z^uxpvK>s#
zT9EJbOi1q!44nIJ3=RXQ1AtkfxN*K$sSqjhZX>hm_hY1`yt)4%=Iq67Ss`-2V2KQ}
z_aYiK`odAcyPf*M`7@3$5&-Y-tf7P>^Fn#{>_d9W9gx~t=5_#h!}*xOVs;V4(WW|z
z92m2)urAL*%Gn&`ztH;1Fp_zy#ga}32c6u-4M<#Se-q#?-UW&6vL5H5{)<UxqDVYt
z=(pb|k;H0Z5+Oaz*#K4I07d>JZ2}gOF&6BV;$05+7Kn)BnxI(n5&p~sEG`Z>4UP<o
zK)ldBgL6147#%l$Mm6_M-!XuMP~HaVXS?~}gU^uu_Rn%8aXZ>J+9V>H17Zmali|`N
z$3#aKh7Q@bKnB{B)K>>(T762rutbQn9>D)t|GjrEgWkU%XX7{!=?acdovA4oH|{lZ
zLa%2G{+tPCofG{T1wAdGL_1DO>$On>J6K4@=y$Ts?vOq}ihdgFjO$Der34D{Gn?=D
z8q`~W)GzX;&HYA=oH95zAZk;f&Pqm#eNCx#tdSw<uQgcjp)p?d0JAK7!)yfbXB{HL
zW0fZ1q=E0xh<G2!pgD3(mlA7{UW({#Wb6)T-Bc|ho-nhXoE)R#cj!3bwM+gXNK39S
zO(F#WJC?Apta9u5st?e0^8d{LNqG<g;fn~?mI~v8b`&eAwE1@<4-Qo8(N3Xo1*HC8
zm8izVVKnBilImEd`u3dqsE+ZMbaP^}0hp3goK1*8(B#C^Sv@0VtiR1DSxrU?-DX^R
zwpe{2aMhJly#<XqMho;kunqY0G^owcsd1~rhC3~0S;%G?6@7XdlSN}Q@$(spb(vdy
zHa0Wk6(~^7^tX92j`-fQ(y4f~oSk}P{0~z&ESGbF425jtc-%Rp#`(iMVdg+VwVDaM
zOEJ@FZSW6eEu}yB&{QbPl2#W}0FJ^bWTI2I$|6KpvOWbfZP<N{cF-;v@2zWJm=+Co
zimdq*>+OVL;aT_r`pbCDW7UWisdtA$nz?Q0fm;i7=?Sr=p){a!3<fwc1`HexNgkvz
z4S_%%wvQeaJ?;xlF(FtBx$?+PA<b!(8Le)<s5YT|Y<B<@tEDBxWm8Q7klnf76nz1C
z%rozAt2bp<Y->s_o>EBDSpX@HlLUxITTBOMZcT=PH0?$q>;@!>o0R6l>{{w%jU}_p
zFiZ+H*_!O`oS18bAAsr$Q_|uvL7`<n;_<pvI?`}g;hM~6a374#SJZ~%;Jg?jG+57?
zB<IH)%`De^K~MnU3<&hO(6%u^47td^r872huti{$i|yM&wvH~{m&9mCpe|lp15~@L
zTxzQa7kksgFZ0m{%)pp)#EBxV@N#b-$Ednt!T(jRu-(xg0ReKQ(~8P9U6fgRnfF;9
zyhunSR~cd9&J4@j)s<U^P$VXg^8fE*QLd@<i$AIcw;`ZX{jjg~(PC$-&5m*RzX;ZV
zHwnR9)IY%bOoR8uFVoyMfoFNdb;)%>_wlHmD%V%K1#`<ySXE0EwMlNUcWRnyrs6ca
zF%F)JBDpDARAfMb7x`7l3KmRdfb_5ZCG%?}$8MIJeURJURG7Trmf%$^p83w-YRpi?
z_;mnDx5W!Eis7`wo#l38k3JXlkvmdf%9h<MH1eB7kmb-AA@;Cjrq$gUlZQW1LGJSR
zOb1g-g3FV;y;Yu&1Sxn0zsDKD(iYf7sdq|*-^LE>f9$c+88j;$BhzfOb(v<F9;7Ou
zPA#?Y89m1;ZYw~CGcwa=KNK%8^n6yYy$V+bbwmULqm~s!8p-UmBIZ(w>cuiAO^ohg
zoE{_3R|8b*ie;c8hhNshM-{RPM4KBsuq0E164>+NPuyjJ=Fd+gweQG%rWh<p`@<^I
zF%vXB03+G*LjTf0Iz@7@7>tiz<RfRqGs>so+*vP+jaIN(%aT~DqILihOJkY<RCuxw
z&tDcz_cDtU5||~+;}?Jn^zwwzz<&X)$chl_^{4`bpsAIyYNbGCRz;T+bD;d{=n_W|
zOFq%Wx+cbqG_cI%OO<(Tdd8xWmd<sZKf^hJ%Z9m0tPegOgL@DZ#TMBR!_;I4yx;R(
zTB9}w`Rm}(zg#vsk$b5cAae2>mCeCd#qcFUY^wauKj7LceD%Mq%DsuJG!P%Mjh+fC
z)xHYHSh>$f#-o;LgmKhxw%D#)Vit7L$A-;Hcdz*qxklbBx!>0Df<ArZ_o*pbQ7vP!
zN&(rL2EkkqEYQFROt^iK2w2|-{L8?Omk<V0ZS=ty1Zr6$$0!0kAF_2sTmV?-;a;;R
z=M#FSe2<j6fAHV=jN4l=H>fNuo3|3m?jykz^@!Jb)Tak}-;Xj95c;U7;m^R%c&5sQ
zJQj=T$u4`sqaRPd%rmAT1`eK`Ay35i&mhMm4np{PKWUUZ^q@ZSlnIe7x8ctBbTGtu
za0OT;&!i@;q_uj-DQTYdq1ly|9J*Zo6zqZA0k|*}<j?-n95}iXk0)C9m((P|1u_mW
zdiZB9kmvjt!<XQ`OM9gzp82a2(h&AwHU_-#-y7t4?}z!v=sW<FN?wSWRKsNsC2lL#
zuv;`)UNmAL+YxD@@YzeY5cIull9%H^ggyel8AK4_=JiUYgOa$EAPc|7hF}lK-+j!c
zC7LghMLoZo=8n)=*REcRF~OGzNyGmU>tDnm+*qS55>Nps!KL{1c+`pj)5+?h`9`81
zt-eH0tnkSUdDDansJsBG-doYTo-F2#_qIQyT0z%9Bu$ccY_ffT{7lC%0UY26de=KK
zEryFjH5f>uycZ(|JSFdk7#p95#qxt_9s7%~6Y>8rp7DBtRYpXIeB{q)9c9W@WE7Ks
zRc_r6`V2c!Ovh67AVY_VL(UUkiXZ!zWH!13(Run4f5!d@=EDU@J`LFm6^(b&&(a!M
zrSdsctTl-I-^Tgy1sous$6KnP8fDEE^F^<I8Onu#L-|h-Vt5wyHIjho(Z97+I`pG8
zxS4%vgepT2H}nh{?RcY_7tOxCoj;=z1NZ|AG)K0#XUs8MM_(%5_;(m>xa9xv&(n=C
zUA{6}0njt{KSOpL?NDNCyZQ|8G};^pTUkSO=h5bK$cirWWtY*4h61}BAYUEr4t0h0
zTI40eCjMHC*YI(i8wk$zXCxEhyR~pz-OZnwy#ZP;&X?W&8SxY#%Y@eC>oG9oan1pF
z+r!2@3@f##zYa;KiZf~o%U=EruDjUV>~OK{?PRda1de&XfDx)k8V=bfy1;E6`j31g
z+T|^eY6GPHH{%)P41xi0^lXxEdAqz^;GnwyV!f#L1oiD3V*x4H7d$8A+x84~5Kzsy
z6@??AMSREKbCUs+J;`^C{&03I9wOiK7c4KKx^c?)V^XShfu_p-t05#choSty8YnI;
zs;-v(MmwmC`NC$XpAed!CO=Gl!&MFpxB#s5BY(!ToY*EjSY>}}4t4~b1_u~Fz?*<s
zD}J0_C$<b_0JpCLy~B_RG=*w$C^^W!1=<aPDF=^M02udq&+KzZG^Xj$yI@0|Cx-?t
zE&!0$$LCE^&iHudQQ@73{K6V~hsDc~i}I87M}D##9#e&0!;Q&;phx&KQ^qjb`xeQO
z{tVZ8jM`FP9Ob?8u$ZaGq-n4;VIdQ+^tE|(ch=_A_$>WKJJ^lGEt^rQ^^ZqY`3w?L
zwWD;?(WAFoa!Hgnt<TAT*iGYs#i6!YL5>-1qK^Kb6q6c+3JFV37jtqZ3^p*9*AIRg
z4AsY523d;+{c2v4V~r$sqdhK7l~?Yr`=$fVKRyP5i!6$VVxOJh8-C!ResW?gEH<_~
zj6;qx<<3biB)TB+rpw9k9*=*>WqJ|Mg(vMM`DqLmXAMz7Dm$h>{mduJr3HcmQlkcI
zPb>x&m@$AwK4#l>MBdlPU>hN-@D-%kf4B0F%Ih-3pV2od(jW}=ae$1_HhTS6hS}S(
zM>W`OxVJ#8hfdDWHe`giV8{<gP2Gf-h?2xEblCOFFAeLM?;03JMj9&=xjPPJ^(dP{
zcPC10ltyRwtumJqSh~xCpuKG1>T)9G{);B_HOUScz`*+oYZJG+4yoLxP4g5U?a!D*
zlZp^gy{a@Rb1$@fHNd>3+JDi*8cW8oXcj(QE@O;k=t~KFj~<n$SkLE@$;X;_-Eun=
zu-mbn$A!3U%TMNQ^VB2}c>DOdlg{k06i8NfhKU6hfwZUf8U=TLPx8XkeKxF~$aLr|
zd3;7<N(P3W>2H83;LS<n#5y@^^wz_uOmII@*wGlGGK}gzSeGUvMx{ZS#B=O^IXm_-
z<Ge8e?2vPO(3aSX8V95-gpxsgsH0(rRwd$5YGeF*xbz->osSNT1R10DKgswrT<?(v
zel|AC3dx47scg-Ur;0E>nlz^w52Bw5Qr8D*k390o1Ej&WK5cIW$%SMAcFe1d=KI9@
z@k)yOhVs!unUKcBuqso@Ct5S3+u>F!jWL*}9CNx9q$y}jyD-*_Au7!=3(O3`)^HBq
zBzfZy<a0w}KK~uG2Pi__a;BVXzvGx}mDbUY2U$=Q6>ZpW3icJe+A?%A1n7qHxi&VW
zAA@6Ydm0Mr7hovN^ExSp0^osnaGEnG`!l^}ac|>XoMMt0^i}h8NAPSr(z(DwI5mAj
zF96MBCg<5AA9vgUIo}2kr-iIsV2=@EA{WL&^`M#*4lsIM<P1a(BY4ex--e8vd&kB8
zrMq&1!^4DjiOm?9i1HKVQtP*j4t<x|4~(13%jNz(g%nGK*m+C{_lj7+3~b7j_E#Fu
z(9(dvuIlw6V`6CAqxWf-TpcnF$0T9PP4t1o@ZI7s*Non(sV~tbgbRfRfl`PX-FR!Y
z6}sn8%fHt8gcD5`R^eYban*SrZ@^zpei@`g4?L7na)R0&^!)3Lz74s0x!$JC;!6d&
z!N1_h7ErihWKbw!PrK3Ex6zO!)8!_g4MPMN_i8>1w4$1G-LGOy>Qlf)>5^Z^m_j&E
zGsKwY$j#A%(nA+jZV48mg9oxq-0GsKQ}5ijd0VD1>6p+K{k?AYws^<K`2j6+q1+Ms
z8mYqa@|$RRWEQm?I^<5H2(To%Ynx_yrDC2UWKWa3)8LuOtyg_zxyJ`aPan7`rm4_Y
zVc>pebYC$UxW=^jYfM8;PNw_&mO*%~o)0r(6u@eLd_b<kGZ40ji)H3$hm)atXOMyy
zaUmI*6)ahnhX+7&jm!>-o1IbzTN5|`9Ph?dVrF(&AYJ~9W)5&1tvxc=o<T@J(y)9n
zzMVIE>p2kl|9{tjlKHVN*huYAn->^+wzn~H#6qWuo0V`PG2blR#uo)W;RQQL7WXt#
zSA+1|C28U)$$|~ZfYhZyMLJEidXeU3se=_DIys6-kJsb4ULIO!7OA9~WrerMS5V3v
zTWg7|j9v@bmI4aHt+G#`l&UJb+II``bdo7FuV-)MZ%xPs{e?%U;!%pU(MfD=OmQ4I
zR@S988V`9(_vDPM_d!`$r7F9!AqF*0^Cii~I2N|t2w8q3_BYbR$tL3-CQ9Osv_m$B
zAS{J?@w>$KOVrk^)!Q+Pg>tX=7{@+6=<oAqvv65pgf2bxwj@cho?i2iYrN7@_oHoa
zo^+?VCr1Gs+{@JO1K6nC@4w^zMMqaiRH{*z0i)1LcJlktTdlk?&#(q0na@o%*lDW|
zT`^yD$=Y*T?nWMnVd8#3=Vkaw$%FokLt}t8fs=>gJ%}&Xh=qzG52xX3_yG1RgLH5k
z_j}|YyjMCFvRutdsOx67cqEpi5_%k*rXEdiT4*6X$W5{TV=Q-Nu6d$7mPSF>S6%=%
z$m8}5+Z2_t9KCoV+8+!bG2|<HG6<uWMaKF9sUvef70tJ1;p+Hwe3os=(VOxa?~X&K
zw<I$|JR9A?iUq(!rNW>5J@PRP=c}M>{@I@)=m_g?X7;}%4q6EXNADN-g<)vVjaJa`
z<OAiezSEIXuTh?l;bpSzQ{)AwcQ4B(7T)~3p<~J4V)%uLsEwYL7i0Kv58;|&9pt6h
z@6E{l2IoOIYjw!W(M@GlMX|hMX8=Nx<nO`%#fTPr)tfVTnde-WycW+eOG46%z&`&7
zW@9`8mK{NLpdI(bKcl%OglE3)uhFHQM<p{jsQN2!M5DMTFoPX#3-V?Rv8E2k{9FEp
zBZk`e+wlxKBC-EX5Ek!vAEsQ8U$|}P{rheRvI)q*S&;V<S+eag*rKF50zBW3KFjly
zYUG2s+~N?1W(PU^Vd@c?Qn`143Vh_xEasbc5P*(x1|eA;mNkn)W*_@&VSZhL^X3!N
zz^b7<$)`SW?rv%ZJ<$Z8r7u|_VNN&GmHs<Ag{H!*KJt#p=l+b61@$g)_Ro+njQ)l8
zmKL}^+8-DZrt;r-Th~!Hv3zMSFmq)=q{`t2lp+HlvR#$(>!An9_EieQXn+We6$RO$
zN)bO`5IiuyQspS$WC~tB*|AD7z)^@Vm7VP0Z8Ui7TxDpbt*KpxckySMOWGyESF5(}
zf+M)Q6NV95%fDtL!>XV#VWaG7O(NKGoa`23gK<hD1$M6r4@m=PnH$RF%-5?FwCwIY
z{mc_&5APPMgDV|z?xnJ4l>wrp%~B{mh=IV8YlJhjutl<$|4v91cQ<@9jqJ>GOLeim
zjVv(z(R*<p9|$)kuQfmgeIVaRQ&!y<>%UCCS*2iWJ%U8PRpnYwiCR7o4u}1_ugw`6
zT|vHWE7LeZzEc%WlPDM80){QfcWn?1>cdKXuV={a13;TP_Z-F5TKRsg`H*Zas*n93
zwb#*xEbmj`owT2~079tRL^idW>IRzs(E97{JS_{2@<;wQRW9CCsP<D>65oR<O7@SH
zt#9n8%NOK;SS{){SOyfTw5=dNb|MjXpbZRFR^>n&ZRDCf$hxk|)^n>K9Bfz&Qq<%S
z2XaX8Mo|X$PcY7*-X;)l^kycoK8X>i7E-HSDTk$5z<z3r=Av9(H03AJa4~E=P)OwP
zs;!pQ*$*Ez>D0*)!Eu9;<oC$bIg1(4rDnYx<sJD28^VjXC}vjDJB;+T85RMvN2SMp
z)?6jFQbGFLPk5bP*GKzbOrf<|26$KIIH{IUImT$0ubrSK`hjUQt5n7?%wbSG1Ic-R
zS_F*XSYIFj@#dVR4kd&Yj|*M_49FuG66E;wGN))4W^Rx<azaoH<&fYt-61DNUy8R=
zX;~{L`7?Y7JkXSqlY>Hv6jEKGwn`sI9wZCpr>PxQm_cQ4_{06opW!`#7rH~GC>RgM
zMBq*`*xn&LfMh@xXIPxxfy8WWmm$$MbOUAx-2`QNs1X|8={QAXnEiE_{!fPc`|5ED
z`GJ(}G9orJt}wtS>R<vaNQu2z4u5owP9pn`Or5Rb?!={WmW+zdz|yr+8Vp|Bn716C
zW%m1!p<_#mhf2ALh%Gmw+X`=u1S45kM!eFW@dzhcvmVFFXcs%p?NSv#W1cKc$TUIK
zRSvCW7<$4yp&aU%VC(+<kCju>9|H!-SZ~9kKuseU7gUAD3a4>7)!)HpS51nUrhgt{
zK<y_AKkE8+$!Q^=>hhqVoF0E7X8=O$&xrnjyK+wfzm+qCFU|H@hwH*wsd;tu<0aWe
z@!2sLNJFH70OPNcbDZTU`ZZ(^;J0CZ3O7rQ(HZ&Nf!F!Fps60>>h2AnCf&x6cXdXc
ztk!bKq?avHSRqoJBFqi}uV=~HU|5c}xviebY&v9oke8n4^|s!a=n4pr2A`FI61)mg
z^PvKp>zRRW7y7sKRxl={SzCEJ0yYzal*nAq9OSG#)|dv&+dch0SK=35T&Q88#)ch4
zL&i#T46}d#fs*%unLj&MMHA`4gS^E%fhP>4i5<)5hU~1%1A44wN*ZPcZVh6<M8e{k
ztvCb={yDjtH;_wUFlh76^fUs~sKbV8i|vU%z19(j?OK`S^Wt16Al9m7ioSF<q{%T%
zoJV$j+aObd7MOX23eXFPY)81K!@to??4Z?RXoXX&%szur>Djp$z%OmNL|GUNZM~h>
zvs(HVv1E<2Q+<A!obSI=NYbW7wWGJ|{008IrL@s>;5a@AQJG$SVaVQ!7N|(rL>Ku7
zhGT_qghAL0TDv&(tI=(8N&GuMRW40rv15@+`(^%2ktYnB85%E71EO_Dt*qEU%yh$r
zTPRl;8w1UQ{wP;Q3wk|OrCP2EnSlr%_;SkCiKDE76@{9e57)%xP~017-dp8be>I|@
z7*b)Tp<h@>-JKQSCY5A}b2j-`)Knwa**`043v#`;L5dTkPo{pk!N}3?I21FuF_8lz
z%xQg-Po3^}Oe=#VXL?_z@s#<-e-$!(5?ntJ7Fm#Ar!Q;)G`T43GrVokDsgk{tQtVw
zw^;YtW<d0}Cc2sT@+wgNwwOGwj9@RlS@4`(BDaU&Y|iHj9dbvUEbvBcl;3z$)E@L?
z0PKBdEc~!axhuwidCT4LG7ffL_UV;=PmBes9@(1tKz|$WLEWT}6pN~M24O)AY+(Jb
zZGvgh0)0I|au!vc9&6hQ4X+8bnh|UPD{4neR%Y4@T31$W<+FOV0w*c8{>`?RhQW0-
z%PO=o$6pGY&T8D5yZjj$6cDu<+2=ZM3?DA@{O36^BC}!)HTL=Tj8znHF##Zsi_ZdI
zOg2LuZZX|j)wz{Mv`b}SuWoompg52$a+;_o0Tz9O@-{WB#s5od^fIH~2!vbRrk1#P
zu`*F9!ki`tMdIzPr9C{MS`0lOsQNCIWwyk~M4FN1w#3K}kC?|5#w>{VfUjIeR@!e!
zEsq<;s^A5rd6U)NHhC<ZHhCf5Bx|f~5G5~beb#Esz}SF-%Q}BX><r#Kur1G)^|n&b
zbwEz0*4W_Z-E25+&6bV+31uHF*jiy_QyliNwt-7Hlg+_3wSdN2aeMq-8l(o5g6J#<
zgk<A#Zx3mdZ5di(IjHZ8Sq_Jmu_dh<r4L2xQt9?*OdicE718ATV=xGp$+Lh+L4I#!
zXselABU^noU<)+xfG;je8!E#yAB`aj>P{#RS_f1#Atp`Vd&oF)DB!e*edxNzI1bF*
zj|s>h(rA&T9xIKcNAiew=4t3yG6-X)R?E9=mptmfo2!IcaKMR*YVj@kqjh{VG88>#
z|D2eaAdef7;NlEZ-xD!;_$D;wrgAqt8Let^KTPy~;=Dc17*ADgRT~gXanS5U^CI1F
zX&7AMQh;X$#vxBzx3%ziea0tfZkd$P^W|A5M<5HV2+vKpK>p-!&&CCuZeyDPU;WuP
z4lnxHQZ91n7r`q7QEoV9%U}FUtI^a2IAFd!7p+0i(f@Y_tQ3Du?;$@Pg&&-To)2+5
zUL!|d2u6eQz(^(dkfSR);{MG)*2^gN0YYHp#YDHBWCvIi=DN|uG%tnl9$A6l)R$8W
zaI80(FR#QiTqBslU@c79e-B+Amm9d|Ti$$m)!ziMR7s>C-)pgh8uCa?i!7l3=vmuF
z)Gz@eA^)_suZODEEUzc((~;1yXWrXq$QwSDZSa8m6SOUFdiUn}NoHLyZ`o4!MY@8w
zV;)eiwFo5w1m5v&1lfs`=iP0(W<fhD>jaJ8i^C${T1(H5_v7fzHtT_rkq>&!9Ts54
z&px!TU@KUH(1IoMQP85A>8Oy4`(HjI+E(bOPh0F9uqe-YTtehy8y-DqCT8UmA0BxT
z22f(^Sj45cSbv%pnaBq{GH8UR+rQ?Ol+S{LCc&F}w*1@2idUc>a!)Ll&;1!o0MXf-
z0q9?Nv%QK*;ul`W!PM|n`%n628W`As(*OE1UJL>5(6>syj4k4M5>pmoP}g!!a92vT
zA!fFP=>39I9kP%XUnDS-?W)7GOed*|6b+0^W&89C4UB|{kU$Ht?t0lF+5{rXLWJ<I
z!|z&8LQ%G_RBzphE*YI5=?vL1-Ubbr+?F|yDNmxb&ra0_Il_fYr4K{lcD5F3nmSNZ
zOLnPNu(B8hOXREZHY6k=GCHe$&7aM|naZVR(a&A06|Jkr;jmi_KoP87Hv%jYw0_y$
zKexkAHA4Bie-nXy$ng(B>^K+hZ+rMR5ZCavutKwYMn5!(gCG{ulZ;5RSGA&>NQ(fV
zYj3CS|HapPz<F66eY}(2d+&)cu_YD+1WOXSC?K*+2hFm(z{;|V``-muV`529Y^kQB
zbfkmylIR-Jd+)vX-gCddGc#w`-1px1{Yc>XKhHUH=FFKheS}x-vas*`=1DP(6Ru7Q
zGTaa~=E=EQC45T_B&6jjbqSLDrGRX*r}}gQt4}!b8hcv4fZ+xd72)LRu}ySNg3Udn
zDQv{c`x>n!E9jZ|q`ib^0Q9q(iv4*iLHfEqo}DwCO{1jeIKmg4aIQVKu0Tmc3e8hE
zyjSqYN@dS$3Y#bg7H6jtf}dZ}ahEhO$W#G1s&?{%rZ8=j;zy2XkDbsI#@FRkl96==
zZfP%c5}8OGx}BK&KxR^wUU8jH!CPzWMZSuG)hZ^}uove9i@FhOSk_5R$KAHwMl&wb
z#!k+E6I&{O&?)|XOQ7-8K%-Q-yX>?ij0i_kD%DH;4JIVd(q0-%b+0RXRj6L^)2HW&
zL(h^QM~Vie5|Px;NK6kKIl|6N0>C~b4TgAaJ1b$t>LSDHusW2amgqNKLJ!Wan^Q9r
z7-O&GH3I*(*zougU8k@yrNYacx5h@qyehO)l6euRMpnAoj%ZLSH`+N4>_zPj1jJGC
z0=FmY)aAazMkjEnVA3QM5#oIA+}xTm7#fz^dAZdQB;O&c{rNc}J{28G2-+^lEr&_V
z6s=%d$xSk>DV^?RIqvfQ#+9A+@_Ze7kflHY=rZR*tX%JuL`=#jGO2^Ami;!yse4A>
zA{(3YBP8O}k>%0G)o0?o#iwZFog|4u9Agvw^O3{Ow285J$uTPbBR96TdXG)2=W#m%
z--Ssui%Ea8$#vIcsTZ<T?+ynD-%^UqF7nA%Eb8dCiyi;C)KIpk<aQS8tS?o^>rR^*
zFcVjWp@XhI+||?KnOL0ynADW_l0;UVV0l|)4fAM(JvP1RxEFBgrd>ihyI~fWIvA98
zBGYR|A}B&_)Dg2OS4f%2<<?x^P;WrU1%YU0Q;MIm5fl>~G^S!*X{?YBo$Cx1ILp7+
zA!}6?MxEr@x$Vszi?C75i79md$tS9eg1Nc6A?87saLLZ&CYt9!s^%%q-1#wrbU^IQ
zsQ(*mfunxdNYXkNCNF6%EdjA>`j1qzqC#R_q4z3O&gm;FHMrSaBA(03EwP?AllB>j
zpSCD5Mg_=TYfS>9LwH+^BZHA*4|M}Mmv46*z|s|AXN&!#vqnRTOJWL@Z>0%S$v9c+
zuhZB>b<nLN7t};?ux_~J70HR_lIOWBR^+Acy#8WmJkwDZciKcq#qvB)fZfwAr+!z!
zOYm>8>f)j0*phE#-OkBo5D7zbp{2H>QvO}=C>?5r@5|b$Q?T2vjQQ9uesK=elYAqK
z#mJ`C$SIM`#9ig;_Kd&YM5HpBeSJ7lG2pC>v5Qd8%Zj^N<u8+%+hJG5?;DT;G`Dip
z`V(V|`W8`lmM*_dVWFo4q&gqDh5nj~Ca;yKREbw5nkcu@-@SNo!>)C{aptHI_Ufi^
zV3A?FM5YuU=QaMuA}rn4Ip}@lEOpuSv086eDh3W;;@oey8!AEvdnN04SuQYK+Kst@
zQcaeV`PxJrHWw*@)#GNb%frK|&aOv(K0UJe^>vZn;rRfZ>J3g{%i8-f^|aX=>npqA
z=IddKhwV+dR1pC7jaDcZhVtg7YO4(^YY~}HL)ElR-mGn0E^o;*ze>GV?5#0==@$&k
zRQr*e9A~2_ntZeWwR=gA-BKaZWmp`jcI*9~iTF0yZ8;*UkSFgPu(t&wiVL+6ih`aG
z`%xq)<n6f|>;V0z%Z200?>iFPYN?_mx_8zoJqUR-2zl4wGx#84BE0e4xz4ejR9v##
zV}^}f<wQ$zG$d6j-s3&_VtXG}dV8<)5Y-2QL%uIYE+q{C>>%iQe?0S6cxn{2**;L;
zQ-Mke+Ux9tl`f<#BU>Skx(NHB`nEd$N}}FqtD8!&!i1T2NBml@9JYK72Io8LSn|(A
ziZ^K5VQVV6y5>5_CyOd+Yg6)>iVKHYhU=1eB7lthM_0l6#1FT&@)=~w;FjYW+EAyK
zo`bbZCj}JmvyGL!CCcmWlykssGDpZ@+~mYVkeY3FtRh7y1LZE~5m@6oY>T5G(<Vo@
z)3(OqwTX(z!4YNK0s@71%3iypXlLrbwLQ6Q6Lai{PuGWbz5ze@&I+RYwdDIRY9u=W
zrfgUKUd+vUrs()uNZNM$uAFzi?TI%xFuH7an>YlD|4`5^z}nqOsNE}CD1gv!djlNV
zaFN3lMQr<WI_OW<T{W*CYBJlOlN4#B5HiH1%Uu?&wRRvm-~{Cp5j$rG^UAxT9mDXU
zoD+N1b$p{;IGp@-Y5$7u-lb#@6HjGFVkRwFam}<I1KUby<q41;^^x!{6NbYVs%U)7
z=klx0KJ0{rjnFTmkHp+LqU!71;~Pzd@rK=7fA`X>I(q{?hUZ(ZIoa-ucXT9l;lUy7
zqrS(}m$p#k;CLQZEQp8<9}CJ*`A!va)1QS7?c@GpSD~wheWDIrvNwy5@ssh43@3<P
ztm%6DR9<x*&GzX)QAs`qMO3<D%XXBryq(2;CO5MP)uktM#8|j-<FiSe!_Ge2K9@V^
z;R|ry&sVCJ&A~Ziv8&k^94pfiFiOMPF9x7)DaFKu+xtuLOxV^{kEOxB9M5#|E89s$
zHNFxDkK6zb-dAG>xg5E9zm_oJpzJI={PiSak`YO<Y_)GBJ|<1JZ-%t#<`d47vuYIT
zEXq~<KOEWp_N~M%UPPo(VaUE+pKax6D^g~?eaE+{ucy%`o7{JEx^c8U$*m?tW$|MB
zz9WT{qW0yB6b?9u0Qjn${k!dl{&@?ku>HvYK5fFBX8W-(8|i`?O7@eAcsJQ701(Lp
zzxhvd`K8I~l1}^C5R7IP*UyJMlX^k!h5aJW7<M_jREKNsmnmkIv<{iT$q4*ad@lc}
z%tmA?*sqh4N|^;1hHPwws*0*+zX^!7+0uzr!dEivI{WS5F$N3iA(sP6d$Q4fm%zfK
z1guEs_c2;XWmpkz<ZUH^{-I)~6f+H7gjfsvV~iukM^$enN&Kn4Js2sqK@L0n^ALFD
z?^U#x^zgm*7vBj2-_%#{uYN+PugbRk&Hr%<&F>D_-*fXPO_(-ehW#U!Dd9>3%7I*5
z!(~Z#+dq?%i{C=`*}rltq3ym-ILE(zW)cXx33vKWh28z5Zq>xdWdDs{2uV~V7b}&o
zTa{9+xtPpm5~B8i=3;RhSWO(>1DhixP}Wy217+7ds5vY+EsNQ69J_7y;N}o>ajTTN
zsX*DVhs4Levyc)HJT#su&r8t?cW_TVtU1k)N*xbg173I7!xNPDS@wwLVpvoMxSr+q
z$mXzEsRL@s@_AHq*ou1lmUnaJ?9nl7*&K!)?1J)`X19y#&Xosn8@E0{ipY@b*4bm5
zkGm0Qc{_a{+w5`8MSP)S=;IT75XhYZn^<E{$a#w1ti`7K^~A)KDkYa@k@ln*fD>7S
zG#L<{++1vfFeQ%AQ*s$nWmTr+sWE)oWRwFWs@rZ)tILqINZ|#HqP`?&wNH2Kkwn~V
z&&W~A<qO_Z(z0hJUb*UsNX+*Ji)JvJWU*%_V9^@Up5wzW?P*_Z&&|)u@ZO2bb8nuP
zxTG@)YNb6tfKJ<&vcHG2gBK)*sh@Z~i#m)@dV(WpVs|S>r5E}?<k($mC&s*4lYyKY
zc16XnUlikbWSmQTFOFvtRD~*1gzcpIOx1owB^&MJgaJzjRvzx_U3QA232Xmh+-ave
zMO?AmPD^mcw~(crI6{3QZmq{&8t^HcOP!_F+{9J<(<{alXC$U0pA#(5&Zyt0e|(`x
zzLdxL%z#VRlk!;k)6NRGD6xjjF)S%=%H%O)Cr_E&JlD?lZMye%gxcZp_buoZBOGvO
zjL_eaF|r3!cqxN(VyBeQPl+vLW8uf`lC-b0QSnXjlS`ffn$bBwdcJBAaF?B%TO_%J
zn{m)?b>0v(OF>mk-T4VfaW6=GJZu-#_Xq>#J*5J^%%35`EvH#*gD}A!dwHNj7Ww`5
zilpnYYJ4}w$?!6Ag*OlExy;7q_shvQQ=2Uy9p{^R^656d&Zu&QYzso;gm@-<fb69#
z*e2%MIQp0s_$BS@lj0XP9C5{sN*;!Mp~E$`1$%OrP0sg<X6>kUCUlYS)xhdkoM0E%
zw+73SJ4s9ls9f%MC(=}(fzUcx$EPJtVoRj9W5X_q!Cc}Z37CK}6|^~O)0>ZbM*GTk
zHT}6X##S9AtU0CRz>NB_EOG8j^l<m4_{E>ep1ivd#bIl%m{Xpr6oWN3vp!RDj$#Im
z*{r%sl~__do3wl7zRb?Oys&S{Qk#>!A9<;-#JJDR-)Sk=Z}aj<4m*3e&96+-d-`~@
zhnoFk3z9!65rY%E&@aN_BS+h14&%V;@$Kq$SE_Pc?mv$kGtOR_s0go?!V`Q`T4GL_
zL!xF2zu<Tdi}9k0Du+WBe-F2Bb!eJaw71Z@Z;7=f2&#9j8X7>@UPqULtpg>4yn-5w
zrN<WMWsH|pnN@e=DZSH{#8*T7gM2ku*Li8Krf_g^B866|^jC*d#d2D3NUpgecFxr*
zFE_KSE~v<Gd>oc_>r5)X<|JDlUqYn-=BVmQ>KsR1VR)u>`&I(mAv~BX;uCRfDJV+t
zxH6uJ7b(~sN3ZqdFlvm6_Gk@fMA(99>|A=#Td#wnv~TOnF^c#m0^VRN>obvL>5*cq
zeC^}3a^$S5a+)ssNVTIv^vA+=MJi;Sx^(=h9&A@95zb$Dja`!zOZ0^@ARG3oSmp+-
zE(h(}_}4}T*kZ2^K-7UzmMyj7d`&!qN5W6|6K;N8GFxsN<v|gad42A3fk%8?HU&Jo
zVq*8*Xg5@N6#ypzP(c+c>2hOTJvAV}zI3kFIzv45T6<ki558EZY!&wU{6f1TT0)e5
zLp;-*^7|po+Zz)TGn>ZSn-b<RRLGE$zr^0``!cSDP`9__yW?g|u(t-V*~!(LKIkI6
zsZJx0h$h=*HzznUbe5xV$Zo0cZ7CPIY>>C+@xaQvtewV;#nydWeG^6)hw*Ls3?6<L
z`=DAOZ_lrF8t{$EveU44G>2OfqmQddtm~b*mN8Q<vUkN;5!Hrwa;v?2@YEE`|6pGC
z?fL2OQD@qF@*iUwSn%owYL$o*5|WB=@IHU7jSkmK?fuCR&l-L9DEmNsM1TWMQAM$+
z>KW<$_Q5*b<`Uo_8Ms6%*@uFXoD&pPj<crR2j^6)V?l+kpd?lDxFbiCXCA5Udc;y>
zL425ZCInEN9BrbzA|=|IM4gvyc7xqo=du&VQLHdG0hLjK&+0j_KAuUclf6zIjyA+I
zb$iyeflJRe#;n=OR0)=jtWEKYqB7D;t<jF!<~-t4TdpdZk$2_UB|yHn*lD)pRKCoC
zyOCSA)}`0*3mrM6mu<c;4f&;Pd#+Yi33Wb^b7Dt5v|{3FEgiAPEKMN}JM&s@>g~r5
zwJRZPEQM!6XuEy0(KA)uU{65D<)c_BPQy04+m9G&2u-%ve;-&)tu&R_l(A%A4mp-`
zBXkzF{V|A+mTnhGng^1mmbSNd+QCX6va88-B++c}7CKasq)_Me5d021oQJ>BPLc->
zkv2OLuaTS1H70p)s~vT=bKy)omROZBn?VjCBzxxThx3<WNhq=j(@V$wNM7h<@u^AV
zJ@ICQ+(3HDc2bAxdlRoyJFuPDK(22Q>Z9G~|B`OPCVn&q5smCAR45PWF*g2qyhnI9
zZ2Dt$ITBS%<d1#4ej9ZOil``1Kal_+#8Ou6lNIu=qE&w?KJ+8#0i;f(pRUN=$sQ}5
zbenx9_wb5#th6ZQpY_p^8Bo(3_PO{*IaIpZ#o#}m;DQyoGl+4qFXTAHTi_S$v@a&j
zQkbE&r-L4ETr^+u6WZKzm3`SUK=_0_^j}FVPM9=thJ7^wW&fqN68~ji3(C=SvdeB(
zJ=7s0!`BlHN<!nzd?OE_O12b-@;4JiSojj1fN#Y#8!(=Di3==pfxWE0;;}E1J-?Gc
zIUB+UK^v3$Orb@^Nn_uufT}Qtf^iD#DtCeMFTNk&D34n<tF;I$KS*Rz>>igN(g=#h
z51WsBR`}BH22zd!F#Y8J$g#MJ@)tj@Okfk|Nq4e`!ie80gz_i;>X^2b_S2lBO*^Z~
zhKTt;OPpNNvaDDFf9@X)KYN7zBKM>u&HgedRt3fP*st<-xumDVejPup$2fh|elvK)
zx%3O$XS)W?=)d(DPCn@j`(2=d4^^!NF1^iu@6QHS<7+~EqF4Yiees^3BN+n^*dG(R
zrLqtFDYmWf>9Te&I<eXQTsKtJU);vv-25+zkPgKTsu>&<`qx+sH$T$B-|A88!WCwJ
zkEOY7$VFKl3@uYpO8*ErBSZtKXSe+`xfl8-K1tP5`&WHU-cg;y-S+RgfW1mF=OR*Z
z0ojfJ(|p{6B|KTHQo(-{0n_D`vzcykL-s?2;_Rb)K47L>$Q&u|I&lZz1LN)DJO%M`
zp+0D4sBu^HEvF{&gJ*`l4pmp?TejgLGmDAuY=td6G)9nOASze0fQQX22C(tG+zkJ)
zhtCY#+iZ3Benfl`S}k|fL3`xPFsTg&t1(vb)FWK)vqx1pc?zafzwOa8k3We}!4_D$
za&U+8fXBpevMR_V$PhxspoOji?Z!>@vV7dkFbR%rU)r)#$tI6aura%?w<pXjCeBgk
zjI<~E4~m86cs^-nNN-Rnh5ZPkf_$FruZZA5=~En~-ROa>6ry;l<8NT~<T>`Vna3Ye
zvZ05aeU?4l=Q!cobL|<iXWBu}VnT+*ptQlB8NY-qNGPBDx)RTy70WIWFz{nlb;yID
z(qzx}c?k~gvgg#ry$1p7xr0!QvgakTVTW#5*=^5{Nu^Gdia)TiLP++4#3kCGhs+(b
z6Y2<ixUB$=;IJ3^+WKgHaH7wK_YD60qWV42YV4KTA-_0D5PghG;iOpC|73I(K~3`f
z$xd$rtH*TpQ$@y3@vV#+8p<^|HTTHNJbELZ(_&_Yf53pzN)Fje5-sB{nL6EG>IkFi
z=SVv}FyM6NF|E>ociI`bT5cq6ZA6hf?abJJ9MxQ|XU+8R1084NAz6uIW2ApZK|dQ7
zGq|aiccE9$EIGU4XdA?!xewUzT(l%Z$Z$k`M>hZ@aNi9!GM;%SI%?j+49<yJ<c;>r
zKU394<vLh_;FIZObRAWV*Q7`9wR2}4e;_3Lg3IkZ-^kKv-p-F#Dcne6m=s`w>*Co4
zJj2wgFBO4a=76_R=de}W<mI{9WBYnA8@(cD#4RIwP+^QQL$+@vZRCsGJ2qcL6;Z;2
zl3vEuXGCGM1lq*g_@o}VhP=~Cv~j^y29GT^Vdn9tv78J3@Bg!jxdcuF;d62p*`&Ht
z-rU<mim<`jg-+!wX+1ePPx<(1c2N+3+b`Xa>z1w(fhOYP<BM}D@nIbsYP3wr6@dud
z2h_x~sj(s%h2$hxXT~Dy>ssS9PYdwGFVrFjcMyL9q+K#Iyc`7jU?FO=={Y6hzLFu2
zAk18v&+g$uU`j<3%t$bh=*3ZqK(MBmMm|+Jg5(D&4mxX2luIOJRCQL*&;x4T*i0vZ
zF%;~^YSwPEW)=y@<-_>1C@Zt`@{#$!K|Bv70~7a1)}@d*imXtu-R8zG;x|wZ+A5T-
zycC;P;c@PzaFC(e#{5{FQp04Qkb1qqck;x`7TCgEyr<R_1I=YIq^=k_)>TzU$3x;h
zb>Zyt{DwV{rcFMpy|N<A%Yi7Z%A=iH^7292Zc%X?TjbMUOb}IT9uVZ!&SlmXBYWhn
ze4PiaJr{s=RE>ZS5X8N>LQvZI77AMuP~){jK$Mt>=d-9?t`{whZ)AUk&FeO{jv;7O
z?kZ}FWPOCcE0P6K0a)SH%PLM(Hlv&_ok8v=AF0z%#I-M>mg{m~-_n+D>l(5<-oH>C
zv?}FTcLGJ6sS+_1E4{)$A3bWcU76^)sIRkw)Z#>CX?Lp&x{+uXFJZZui=F#gZ+<$a
zuZ`%}K4)wvon|ZZoM2`zX@#o-2BA!466MpmD#qOeCF+?v%(z6^@6gR?Uu9QUj1O_o
zLPyFyKa4l`ni#m$yWyFx%2!o@tDysg;6kx$>n{qADQ~kQy4zn}$GHR7VyU2MugS}-
ziNXwaUH+><+grOn(L`3Z4q=<!5Mvj6Sjkyg!@enJ{>Hj^nRUHX)obHfIV^0-5!Uj$
zoS}(m8hgF(_9@q%yujX2;l@-7+($T?MS-(727ZchmS1GjR@9J6ZXvwcoBWDtp@9-%
ze05T)ZLvtSYkNx!Ek*4R3{ZcHXYH`JR-tC}nRZirs=|V(lEoWwvqRX88N_Z$PD4Nf
zVUX%2cxx`_tPu#hw<WPpr;Rmy=xxE$U9l11ipuc=wF8nbs*b(AF7I%$aEMJ{N+{be
zYxFyETGSgFhDsHCXHH8jX1l$s!mT<{uB*R0o|T*=CXzet_KKRF434CPi0~c<8EL1x
z6wZ5Z?n1Hq6(j}0zAx#W;N6uSNM-L2xU~u5?egBaw&Mv*ABgGLN(A!qr+v_={vxvY
zKI8|B`cx%?Zgu>#op^32?T&y$0;KYyh#Z6uO18Z-59D;Ydu>hr6VZS`#kGmdVxAr%
z@B@I(&ALILig)}N&(`NHJg&_)B#ugqvzzTV#+WW!{=YWUHsx+KEhPeSa}oi4kqC>n
zyAp4NnPS19&R<L+HYvxHq_C~dRwm7|Z4S&u?F8*?4<rPGlo2{MV@FN`DJsikXZ~A?
zd85^KRcTW8vSMB1%MuA!QqVllaJTOUxuv!zrvl&FMg^%#6kTU`=g`FEwL$WGbF~vl
z4z_(WkCzlqPdIipSFUEa^@xynu}H?jVX;4Ht60?J08o(^fU*NVCY^MYGkY))#JyMS
zly{RIawwifE~eBuJDlU8S86`Rp=pmc*^&6gd4&Wr1z})lw4=$>xG23&?6IU6-cq`V
zn76z1CAx324+nMJG;ouBB&nlV3xS<ebx%Gk;fNkLac^M7$B7<V3L4NcfbJB%B-v91
zN(XYaj+Js(-xu%`5)7MAV6dC9_S#32nv{&Ki2LISehIjd*->(hOeikbd@L7y@690d
z76iPHCqG%#-QA0ag#3?B1PqFnC3Qj4qJ1*=y$3TV*1b>VJB2os_Po_T?F2OH%+dCl
z`m>tH6qMU%<C(J&SIIx;9Jd44agTjIcMz5=kx^Z}{d)XD-NC`ca&fWE_QhCm#GxQ%
z$VI89kvi?0?Mp5w61}0EiW<t^urDW4Aq25WZ)7hi^(%=Zf_jl6A;GWa6ADT7tir!%
zU#mC8*~9GXu`p4aa(mRp;2T3AR*6`JZ~Oi9%|rv|LM}!`UHeu&MP!}Xx8qv{f|Lj~
z)UKkT+jktpv#=h2H#yyMLPYl3_k09hXQ)5l&pE;xw3So!gLp%6JBRIuzJP8C)b^u_
z#L~>2gC+}l6+i69IUMLMeN&Lgf0BzvNs<9mjlh4JIF;9=uvh@cCw<6%=K7d#CvU@6
zTI}cfdWq3+9mg-?Uj(chi(LB4iVrzdWWd#h&7bzG079CPOPV|E*YUQy2*l^*<B(qW
zo222sE_Suoe(PVW=rOd>($HP{yE>wBMO@4JeLO21K!`yH{6j@qE!18X?6t5D+aHra
zE*&@C{*)89lGeg?_UAmMc(H^b*M^kp0J!K*fAPH?H?yf}@+H&kug-J0Wqa*!d0g=-
zFYYTL0)Nj1%WJ}DCC9HY>wo0DG=Y(P^3RzOHb?#ArJej&?%IDPCDE4W?&bFH7)}xS
z>9W;X+h_l&c!0*CX~ok0H=d=hM0c{ya*LvzXITE(sZI8PSrKzrxUXM6ezqSNZ?A_V
zsxFFz^#{!gi{e}zS$lB2f`<<VSkkNQA@PiV6~$d<orn5oc-&jnSm$B0LU(PHq|$zS
z_^e`voFaSMBW7hxp5t8_=Q`{d8B6y_fA_32&m3-#s=yCr+De6d^sF%Dv9s)mTY?{x
z;G*5>xTrAmV`B*EI4ZNCc(}s-?QuDTf+=y&6%*m(1G*I0OVvS%Z5(hKli3p-I<lS*
z*%N0KJ7df8HhYpkhMCfsB%$EmQZppQ^gd-)*bbZeuC}Mv7rb+cdWSwOz$`=#=(11g
z0^ep&kGGYlCv{F2>oY2p2c)v;RP$$zd$DK6H==4)`d0SjvvO;D+o@-0&rXOYkDp-A
ziT6E@r5s>;ZoV*V_%M6kAb1GRDpe%<BB;ckf4}W4rYP}3ZV>JPY~e4s-|vzj<?bjM
z2gRFxf@ALFQ%*VEUN|cZyA&8MA@6o#d|fKRFm5T5hZ3hS<`Vo@;gUCBH-Maza=E=Y
zeo<zNc-LNxcizVBq&jAeP~#VBM*$a!J9ctz4+M^puvo%QiEliy-NP6qV@P^X`K43i
z8_5W1Mm=qo%Vd(h<SJ2Ny!9v)GAl~PS&x#NobDx#g;TCQ@pOBs|9#w;DO2q9ScTh$
zwfV9zE9c{k+%ndz6ccPmc4j=&r7F!yMaa*}vxq&Fi1J<=Hp?xX63Qa|8pP5=l|f31
zPw{V_J!E)=?-DR=!vh#CAL)i|P=yewv?D6Al{X{fAtGl%+{mOldfnqkIcHY!F9>TK
z;!#OrQ^|rD?K2LnCJU43CwbKDTz^4z*vJzOkcu(89LHUDemrvtK!&Y@c7e}<i(K;3
z%Mt+HJPmt!tbRy%BEZ1ISff%S#2W6lSHyZE3A(J*6}mCA!pGu9Q8Bc!i7F1DOj2r0
zIWDHbK#V1`M4E#tm5k7hC>@IIY8yYRm}g&hflWwEEWqG0F$XNitQ;AelnX5d%Sz1{
zyD*5-g?=Y08Md$$!{!dy<lOq6Zly}l$T@LAJOX0h#W7~q9Z;;#lstmdi!+&^g{cnY
z)yr&Je(v$Pm&E5H9tElH=JV-6^Fzv{K<$zzoy3M3(JzhfF+?IJbH&JbnRw`<Y;KeD
z*Cj3O);z2DsnAJVmo4hP%4Yf*>F(^MbBX^($Js(SHaq`Ka%V9^CfEwCYuMa+q}1AP
z{lFe<_pT|=^ZBmoXtDVfHdUijmHflDpgxmRz#A?tOmr=35n!m5m&K-tN=rnkBx<b<
z%8;q!@+2*S3CV$@fPe~R=n{S9AU+GfQ0vncr{#gwZQX#r$N_AnifL;MlSV#WYcj?h
zw6+9OzD!6DABMH(GmH&VtJb2*Ee;Zu-CkAAcZKY!Gf*ugw#1Q%sC$qEhgMr!f!x3u
zF9s;<NFc}2!`!aOEgajj%$DU(7gE;2I%ge!CieUP?~?wuJV`bxj9ZrvK)VsOmh8@T
zaP`RFg#pD@#4{mWahcQ+wk!Q=XOsAUaB%FN%`Mh5_#%R{&fzX=I2}!#IAMnM`roHh
z)zkXwE|Gl3GN~dMi(Q#J-bU~RF8Ov_<-0P4GP+kKd?jGW4aWWPOc+-BEIw_!+DWpw
zhe7H!x$3c)QdC6!RlaIi7%eL8+8`;dH)20Ue!5~MbK0wO)<R}p{^m8Y0Li#4Hv9uz
zSaw}bM%`aBC<rB-b=-^QW~mJiS-2tJQwA4_&~{^<qlqmYu;bUptn%q97E-k>UYAFY
zR$;WIwAbhMFKS=1#NH5J=mJx5r-NYPjq$9+qA`jZCA=vwTO=NZ5tl+?T&zkNRnzD<
z$2Vx!{9gXFw>WTP+bR5w;QH2-?Z%gc3}4YrzELBFjj)>&uqM@Nz9mU^*jc0P);c;A
z#ilDt;I?>%AqtT|IS3o=ZFv(YzFQH+{r2|y+BPHuOmXx9dPi_X77e43wWT=cG&Dq#
zvOtk2?47x3MLuwbUc5#2uKG(B4Uuwx5c%ufF(-Zkj<(`r_S)@%Km<Q7HX%*M_`yn)
z?}=|DCm_eM9c%5q0jIVgwi9UKO#JD|7T(p2R2X}oBfF!E#6)h0_XluNT$CVx!2d)b
z`9RXpN`>!mp?|P)M`?$ik6dgR%%XeqLv<)H!oj@*<$!I#)wMcz3}K*QchvWEb?fLW
z9Yh7%@65vzICQ(b2M<UtTvLDdVAhiE3op=`4-(h2Hs|T>CZu9reS4?^qWDdMNWlL3
z<OZ~*#Bj?BH&onG8<rJ7wze^zDdJDfC?F!+<nz$<dMW&7v;T|VHgf>(idi=nTU24|
zP?Pjs9kyMz#p$(sc?&+1tuZ({C_<kC<PO+2e>SjsMhEHAJ#A!Q+V*%)r%xvPqny_r
z`7NR@u2$kW_u9_fGb%L_`qHpn_rtYY(Te5FcF)ReFea%HwtI2_c1o>Q?e2tTdQZ^@
z6$-i6m$0JaD$F4J0t~+=H?YP7wm*+oDWQ)F-mn9K7DQ<o2k+T~tm&n^+>Uy%uEt{&
z$;Gilb#C(R`)KEf4e4;a2XPmrU^_<4Be~J#fTlbw0pFFyqq!&yb(D0cykTG^WO;HJ
zN`jdzl?i4PAki80G(S<po%WIBqnH8@PqBOQpOnpqt={XD(2u8ws{I2=2}u6wa6_+0
zaww&p?Y_j)xLSDrqmJ1r-JCc(9#G1tq@zjgpEyq+^V^P`nx*#f#6Q<14YG?T=@ZEl
zN1t_;eKIk2&qnUB!piJZJ^<E?7W;JkF0YDWBnVcv&%`rPqy|;r&(9{(imk*8<q9bW
z>~j_WHO*4@HfB9{r+wae_sDZb+ZS>xR3k&3)xKC4>3T4AZ#XWBrTdb@Nahb6NBr*?
zB(Jis#1}>I0=6ARNY_#QYQQ1Zh(&@MRid6cla*V`zE<agIkE}f?^fd5*K;e{XlO!s
zI7I%9x?Ek{MOA{&i3@$xcNwFv%2#~L7e~z$`?mi;gV+^xN%>CTP&wg#RqV0vRz~eC
zsU-UsLpb+vk>N`IR*BsIUT)%aF5~Yz0+erJKgbP*WvkTOCM;_|tV>5cWVv#X4b}92
z<WP>GE&Y#^W;@U-8}<_ix>GvdGW%&=wa4<-0wngcm{oU-mZSUY=l+Zy6Lirb>1PGy
z+<y^}s9F<vLyyJ3i7+7TN;RH;dB5$GU%fcCAh2J>3_7&rpb_}E&wd@xoSV|$6yp5N
z;0Uk)aaCeol-^^%jk&#eM2QAqZnta8Nd3D+=~&SC`&<x`*?z7w$##E;0R)yD{Gv(G
zj?5;cMcE%?YTr}I?+QndQ|wPM8scWLCvN9D`|}{6Vk3jen)a8Rhq<Gi5D*OZb^mK#
zO^Z9b-KqGuiX6#bGN~M~zvl`{1v;5DcH2J^1tq1eoKFz1I$Wiwwf!>%#uAoF{GE$e
z|8mUWpW+()x0A$#u2OI4KTdiCqPN+9b4KREB*UGv!)$hBBKLLJ17;srPVTg(Cd&B|
zCi%eGVLF>mKKz4bAAcf03a!O>A&-)4PY<5$woi6mcR6&j%N|mnshX(E5fAm@$tP@W
zCve3cHoNHH7dGtSvy1I>Hix&J;=7NSopws?AehNY9t=dC9eZT_uDzr3fi!M736&F_
z+1sN6)RL4Bz*KYO%L9<mqZ0&z;8szM=rQ?k6?{S+c<gL9k|^-=ustpop+FABda0)K
z<6~^6A?X=vlu6MRdqO}}3_{RD>2@+i^m}5ArgTN6Zc#SVo|KTvd{UlC2b9qD<oHD*
zh-3y8;(`*mFQ1a%lq*AY6LE6()P#Vy#XHrzOP8eNW*G#ZmSb|$Djr;(4d2422W)B2
zkb&<Rbqndx&D*lhJTsnQHOBghT^336S+m0~qncKvi6gf>J0^0&XtkZ;yLwJeB)-S8
z_7ku1+<e9nlFCT;VS8RI#aq=CxrFHk!vLS}6f>~;-q+oCeTTik_ibSHqIRY~p<+mU
zK$8^R4Om~8@Ut)4148@l#C%rtHJ8p_6wh?(#4c4%?Zvahv`+hYOr&jgQoN*4Y?YAQ
z3xS<nQ7q|rsJw7W9;V5Z4?H#h(KO)_J1v19i$m64l0Qx9SZps%*p&MNB_G4$clzvN
z`yPJIFgqhRoHGZQ{D~8bOC6^=O8Xi+GqFT&9TvE={L4`z&#__t3&w?w^7fsbAk31(
zWq8Hk+A2Va&UV;{x@4JDOH8qi^ijt&O}BGmf^$aMs2IcB9O`R0Wf5;iCmtuYdrjwa
z{j0HD+vmktGsfBZF#y}gR_RvJJ-!b+&INgBd#FMnJLCOu?$uV@0{ilus(C4KBiP+u
z;qY_EE5?FQD7f4hUpoctWSJWqD^yneeu!~9#5k_*0_g_;afn?Q??99mdJ;G$%ud^`
z6rxlxA#eV^Q~bNQx+l&~-w9^GCEYz8S3^gW0{-F$;&oAz%PJREBuXbMWLPeOT{byD
zlJ8Ug0E(R1MHO_)+`^RT*J~H&#**#YuqjSk&BY!&b#|Ei)w_mHacwp&2EgJ2hvGj@
z$0Z5eppS#;l&ds7-Y)rKJd@A!(s-uWubpg~3Te%#@2Rc0zt>d3>LDMsIo=R`sI#0n
z<IFnJCYdOuQrfKg-X1xbc(%u8*Jnz(DZ!pLCt*ei(H^R?iSxO+@wR$jXB^r*KSTqo
z|8L`%&G#D;pTJyO;Pg3>nDmAI*M+B?Vwd?}FTZBKUG8MjQYxyUBEGUNLq*;ZOwdAF
z>N89#AOr`TElR+g?BIkOtTmnqja$lva>Uw_*>=g5Njzb@wb$V`;??(v<i&vjzc)N4
zNv5=Yw!|;pm7TUU_n>+5)Cty+qm(Q&rGp-|E0PsWq_5Soe4TX1k=B{7BfqJsmnwcQ
zPi}**iC3X3UUB2QEA8C@m#l+|m<pE^V&RH9pcmgzdzUMNy_NRC5cCne9<sly5Dj{R
z^*BfznSBV+jTpPsmE=9ym_31`*Vi$9`Xw`(tuKiKyb}t?DakzPmC0Ae_jOUQ%2wry
zC(th0t_lEAOmr5DtEN8ykhO$fCw{mwPWF=WiV~<gS=ZF*sP(DCtveYGLN&tN%H?=L
zX(fO<b!{S{7%|-^_Ud>h4oD0DV$2?UO&(oEc=U^jU6+V0g}X#&dv*Na0M`d&a{HIx
z5h&TgHKbBX+_+)3clmZdp1_o^aIh{9u^6&fY1RBGU(byJm+}emv*_-&*CvkS(dNC)
zy!X1;M0sM6^gXH$JB)666$pKOMG#d+S)VFBx;X!aq$d($xC`Hy6C&3r<UosiQ*6K*
zq^&#b&AEXJO#~@#iJ?6xOTJa*vdfP7)&NQdFey0Izyr|gPkd9%B`J?;cDjP>=J-b3
z&wHK9Kvxi=+!?#2E<`O){05M`y2Wnwg^+Mq`t#iu0}Y+l#YIUtm^bs*BJxJ^l)XKm
z(aodUCj4pds5gnaYSA}HW}0^nf>4xa#lH%~E%q*FQBx`U`)-HE3)}bG?T$i359Bv{
zPmJKU6?xF5k1EIZy@`HyS`~{4u(J2XGtGhVL3fe8zrs)pmIDK}SNlLBu;`$k?krnM
zdSO18>n^1kxQjoO{HFz9-|D1>u|(x#hRhveXrxojb!+U-c(#EcF72v{F0Bc0+}<Ez
z0&5%Oy~ftw?-yO6V#a7jwl0rXaokds!qx|!fqUgaCtSlvjk1loDhw-f3n&*&9u8^D
znr4$<AL`m_n=5uRm6tMW3P-fNa>%y!t87cGL*+R{X@?kQYkpcNPJz&Qr)_iC5QX1O
zy!CcR)A*j2RXPGYeD({~ZgXc0oBf%Tval<LT`vTJGFP}{C)w>^b|^}DPX$OxFSu&~
z=<a-W-*phB>KPuiy)l|wmrDW!Ct6qczC4<$0J5^f_B+t<J&HWof%wkxR`?#U2)=^}
zbBUKK>WLl7nK^cF3|VUJa6W@)iRK9|wj=S38&*oa+%|SJxS@)%NClE5TJ)_l`%>77
z#BRru0LI}5QWrA&@a(X+lKv>&*vnmfB;UhisHELiyC<nZt7iO?GzGroiAIUEdvmj<
zb#oL40z%R~N(z`*+I=yC3<@~;flaP{oh1?PLHlU@&P9*3FMfIhzsr&6uGHi4yRS-r
zYVD<1fYd_!Sj;IalQeUcSpImPJQ_kTCNR-HQPJsz+>pWSlNH!<;)*`-sd~An*qN>s
z#a@<r^xBp<0-p|$N(KtR%3UI1wgiA8OMfOtE@r91IrrLU<C`Q&*#O+)3_-EEpUZDJ
za9%Wbvwc3E$!?;%Up}`l#5f{D)rOJ1QJ21Il#;n;U(5}eF=5Pj`%?a^34Ou79N!MT
z6m_XdnpY|JuOv%aqzF;_Y6VI8@3Ng8u&*U%C-q3)@3*hJ{B>o!eItlI?C6~9T|lPJ
z6W^=@WmoOP_N{oP4OOx$QjUE)MkwxWXTLxui|BW%HUi18$qFIO5+5=&yF>iD^|(rj
zkglUbhDw$BUV<$%scQazKlxKLw$UHlZ(blmAd=d$&wf}Tr?)5b%3AZIczb9a2(4m=
z{kXneiWUuQgvx$0`?$MATj=;nJ)lK2Jf|4yPZQO>u(gK$EdNOneB25@PXdF&(^l`d
zt8<56B*08tGS92<|H}%(okeRV%E^9J-`=NN6yYHH64|f)Z4B9iUB`cuprBQCw%Bj$
zD_yiR@EpqD)k#!VzL+%a_wfvSgMh)hkNq;9ZGfc~8Y2enkMT@QQMnUh0e`AAE$g@l
zO|^8&zLvA`=ZbN9bTi*JZbPSwzr<(~+9bWGXVYKfZE5Q6+Dxm^-;!xJh!y2L{5`(b
zB`5Jxu>7Mw)46cT_Mh=gwHb<?30nPE63~MA^X=d9g3K!_G^(tf{|q^7@(9W}B{Gre
zMmaA3O~Sx4FQSO_ew&lV$7peg@8JP+(p0$*Pl2dEn_WI|&T%CYs~}CAJt#h12Qrl2
zNlg%Y@SISJ@eTHrj?E9rw~$s<J!2ih*F)zN%O$<qALawnu(ExYJ$z2F9^wyLPF}_%
z{9kzTl&bc~fT<|<EU#Kt>``;v+Ns6*4x`_|JtNM`9_`}{tR``*%^s8BU=<es<PxyQ
z#@M<5l&Xe0#xJmTd)%C2-Ng^tZjaBmipnDhvLfjdVoY6%3eQkl%oB4S_@<CU8y0)g
zoMIowx7n^50QTe@5pF16KUn?oDe<)$V~8pg4*JwNak8l)jJbi^vV?s;&6l@|269V#
z?diGLnd$`cj0%`)oqE{9Gvk>K4s>$gjST%P-vuse{5a1Je7eGu!B29;Z{(zGx98+4
zs8LILtv%Oq)I~I_Jug-iVd&-B<WwM6J%3JE|5{YsfNFOya2jZCu@mAW-`e%}-N3!f
z<oDfZFAPvp0sr4v?i1@;6hNZ`b;Mre;FYRcQXpTP6gigq#fw0hos={uliSG&v21JF
z1LRv*5>JWAi%o=IHgiu7X$N4Z&T)&C>?5T+Dby!1P%N*+Qy0PQv^mAvIG)_Nmm~@r
zE0)6A8urqd!415{$HHMosVhr4Xr~8kQvDQGt>PXV*s#0oj5%R5CQBLquE^yxedpEG
z2$DXlE=zSkBsi2PE-v(8Ng<>XVYnQ`Ak(6Ua0v}h{E*RSBkC4-p;n1y@KSCpc^Eg@
z$T`LJKwArP9J=hB1X)pneyMS>s`ksMoVTeJq8Xi^&A<)XZ|4RWE??lfLSH;DmjIVl
zAk$!s@cDDX3dkL&`lChnzaU2hkY)Y$GM{ctGtux(_VUD62We^)Pau;26?Nqj2z32!
zvoSHH1St^%9Wa=9D=O&N_{G&AmnX}`VdHY>ahNUC)}!6VCn%&lA^lI76S6No(>l~$
zq8$KiV$3hW3U};%H^9RuCFZA-V|rn}P$V=7n#XK%o;~vCRv}hj6ktjC77dn46va(@
zaS|IQ>dn1id`jXU=HG5pb8y5Jw=|H^9nYj$%D9IPW|!oIEUg>2;dCeGb1zt6m)6&{
zGaKo@c$dw{1<XJ?Zwk^8GgcgdlE-wk5Ywb9ShGV!MOnGsX69kwpvilSJ_OzB@50a~
zn^l+UX*b)dQF3;o`@;S<3~h6qA5FoVVsjIO;s^Je-fa#gUfkx@-}eEQ%c-s#>b&o;
z`H4k>WO}<XQ!j``lf2Cl(5BpF3o9ZNT>}?G3%RVm$2;XXZ??<pGwDuTs}T7sbJZ{r
zwJ%dGY!_7!@R5I8k|Nnu1@cSK)J3<&7R6|a$Hlsd;Cr{V#xrq3IaXDbug%f1xSOo*
z_DaRDC?HiRq7ZS=77u<|`~*8#*OGW!>?bZE_5zvCmio>y!(nvOj{05=bSo=9Xjjx<
zs8`KyVwGhCr7JFrUy7CrB%2Y4+}PNe=c%iG2@x7954t=zn5slr;k%N^CSEefx|6JD
zkc(z30x-1@*kxDd?ok7vT>d?=O8G&=QWPN9$ot{mn~82#dRuSoi45s-8Vd8;Er~+1
z<n-l~_yl@vrQ>&C_3~a@6{`~d#c%x@P^X})@|%kZxU>FbVAN`9BYm|nu&alBtOKx}
zE3Nq0?V9){&63g=TpoQ@(l0DlK1MJ9<h(%2@U{6R>c6Y8sxGIR!ml2jjnO<b3QZ7g
z@--Dn7!WwuM=%du7teG;<RzAlbbW5r#a-R2NKry`xWT!_i@cw}TDvhZ33rv8O1KP5
z8q~F4TZdgkw45k;t-US)Q__l(?5M)SglkcEUhn&^sIXppLwuT?PqOqIoleL|#KO?9
zHzl1+>AT9_TrpLP9dX<Seco@0-+hHcZkt%`Ta%VrFpjD`(@g<DxNpj7y}2S)-ZrHf
zm8+c7dW*vXGv3><TN85>C+Tl3PPE%9ro0-yT+edM%M*#$*4~y>O3@&%hxc~Jb)nju
z*|vA&8%P#p1S!Yc-yn7zgivqvU3rXFw$N<J-W>~&mRXFHVDR?DYO&iw`dA9ylQaYy
z;E>*PurNMo(B7MuL!*;SriQ&QQG%yL2}?&&2;T3LmVO4j@`2ow-i}4~LH|t!pX7)9
zP)>w$?zSIM3l=q7?Y9oOYj#HtFau6-XRc@*=G$eq=708B7Re!6v#kw@4VJ8{Xh=-*
z(CCwOu?kKRq^Cdk*m{3PVRw*;+fjn$h8S(A8M?4X*o0Da3%j$8i4(9V5uCOQwke*8
zsYuh%MZ4LTIC|7byDMpddssaGfK0A3w)knVExFT8WFQwq{Z{{YDTVd7CG<!Q>a(+z
zJ=h-4+>->eg&=n%J$IE1G7`Z3CAtOE0ZXk6QI>WkerG`SUA8-M-_$h4_T+DvN*DLq
z-8o0`L6(?k+nY>}SC-hme1mIAw%5JL3;P}L;b)Dq1Nm|j)p6}$eOYyur5bIuLqQ+P
zF_x&U05=_1te5e&wzLHK@%E9#Qftc!&k8^4lbu4bm1DjV3UXlM{cx_NQHlqDB&I0`
zk3*=w$kKr~+C8x`LoNeALs8*RxpeN$CA9W~t${op9B@r@Us5L`kZcE(2#>FwpMKQA
zXlVmpM7!e-A{`@1KKhs|ON;yM;|c5}3ib8cCjwaAOp1ptG1fXi_DR2&X0AYL`BZLP
z^Nca}>3Ym$!$^Tv$Luq8eKH$xa#&gEu?LzJ{`uMZrLJqv%zfmDm%HF|4&S*a+UFAo
zbG40MNK!6TQ3N?g9Bi%p#as<h_Y@)cQr-34MgG2%qxWSWa4}AHDDo@05+wzFHUBZ4
zUT<H^iRUSt;OlWXWyDnQnQm|YyW6R?xf|^pj-z?Vw)V}G(u=U*WltOSt$?WXP6NNH
zZ}e@SX=!_>$~1q+f#@d1q+#FnpGw!zHv8WH;8V^47Y&@Az8{0r;FRmnZK6M@q^xp%
zo|^Q-y2-Lb`c~^=+-^Tggca>os>)&eaq=aeDFj_nmMMJ8ZBm}tPXaa<?$WlYlp){I
zpQeaXa(-8}7wf{$a%lH+aObIR{3eLSU4;JW&vOY-qhwlqq4taVOd&<GrxObq&s1zj
zIq))H{>l-klIh$Qzs@}+{Cb7`rlLVkD4$WzogMbuWM&jSTtVqW`&}M>g8V(1?)S0Y
zk_pUYG+G9XQmhrueANCBz@TyzTQ621`(sY87Tory9Fx{5y_oa5?axCNrM)7;*#6?=
zu!!!K4f|_A?0EueA*G8}tmJP_dCNN58us`4mSUa?%Lc2MgzO*ji(5uu=&<y%{r1mT
zPLY}U2EmQxNbvJL_*eZ>sZ_#}G!8ja|6Q@J{61Z8WNF7}E*HzwqQqv1PZ|m5<iGVB
z0#i7^>XKq}2hDB@A{3hxF{w%zr_CNPw-`Q}CQq^l&Mj6${G$X>K4`8R9Np-Fu7fc8
z;JKk8qnt-|OwRB_<`&}{*>Owlq4^a$#2RK$=*q+9h7qK<<*F4e{c_+wd~UHaVHZG3
zluzLi@tx|hNXtKjZRC*-&XNu^aeI`%pthW)GN%guK6-AkX*3r*+j4u%+%Sh0=Z7wL
zNu&~iA6ubP(P{<7Y_P}0Gl}jBo|A#`@i_vf3|%J9Ue(_E2@VSpaR_WrjMtovOSx36
zgW|@X6oZmFT<BJgs3*q;s2`iM0rw&-VmvGMLe)QeiVqDN#{clt7+P0XF`>)mP}cCY
zxy9(pL{v_&r^leuK8xlJLlC@5#a24rGh!U&=eQfh*R#c*S=X&Ky7TB+xv$-=t$l=b
zz?l)8p6xSorWX^?{+znB;`5bZ*Pc7~xHq>UazuCG^M<@5++$I&=jZYZ>qFRN2SSZ0
z>w1CH-#H^j+X+q;q#|}KDYbZCn3O}HiYgnj=O<R&E9KKSS!SRE_96#&VD<D%nwss!
zbBpPE;v754sicWY%*r)6+0i|)`s_1@*(m|AZ#B0k64$8-@Sr6Np5e9q_7h!tT6`lJ
zQqD>d(@W|&6o@U)cG*jPA_$dPPtW~8!Be26^0M3?aK>CWPwK2lns&oob&@fd_|Ta#
zfs2M^*>;vMm>h-W%rLBi?Y?a}5;tMvki$!UhqL{Zw)8AnX~Po-&Euxqh{PK`&f9Ec
z?!lO@4w8Qwc21?p{l$=;7U~T)$`L!TdP!$DynJ--3~|g}z*beO<>x!sr=t-a!I|eJ
zx0u*x=T}s!Tb9%L1@TOjB28YdwU_14-0LMwg<sYaDtg#n?%RCfBzuJ~V^IeljTJV=
zXHmCW8|xId1Y_Pf|7#1KKgP#Ap0%b^wZSIjJZwFdl#*4f=%6h&F@AAji`ZT9aooTg
zZPMKEq%5LHPfK5Gk6joemm7dpQ9aKlCz7YDl+Q&uH*CN?=7;R!cqXx5C;t$3qAB?v
z+&~Vd{p{n^dh!n!H>RrMh!KUTYSVm)i(3}iB^BsG5~#U~{J2{&jdd~={dZ&0rM^qe
z-F@n_951j2qLt9`Q81h8M>~l<aJp<xa#3NtZW1eI`a&u9Eidq_fFL~8a?Hmz&(06A
z#&(hjZgUbt6UI+&wz<BF;iE^}yquOr3Rw!Knjc8Wq^A;99=WM1U9uqNm8VA+>fu`Z
zV`0UK{EV=_qS@(^xU3$aA>m+*Nl{I;ezVIHj-tF!C<1l=mGKOBnb=w}m$uYrn5}|H
z7R9U{rzmX5Pc5yzHQsg(M2aG&uC|;hn1M=0v^#qEias#5*#CvT)CEsmk_T}TCXa?K
zjmhNVmdzM}hd<?->Zo8SJwq-dQLzlDSNP_m#Vl{wvWg?gu^zqXDV&}0i&rfuzHqQB
z8^iMWMf#$uIS`w~pUQXY@~v#|U2NUScgA)vu@$jO_uJ4{fqlC&R}PL94zi!<$eui4
zC4NvQ=ti!0DZyR6)*Dkxo{@WQO}Sd7==D{g)W1Lu>0`)XYlzQZ>5wcYMWfAD1#4V_
zneeK*!okuVB<dhB_Rl?D3ZOYnRk5oaLYaZwC@kS|y<MFbBe{HH|JUTgr?yV9S5>IQ
z&EWV&^HFK&YxA{PgxI_~UMu)zvs9^kJ)-4NdyTK3>e9s^d0o7yyFu1Fg-rRbULU`R
z667^fEnYvjH`H5)PKOMxGzT}hO9IQzQ#Zz2g){S|@?MG=1gN(cwRkt4yw^I^RE{K_
z@O8f1FFWBpd%ZtjN^Apmz<Rp2B9O>0FX3abXV4pyA1v?iSgAJ+l0u<U$i&|48-+ck
z-`<iRbai)KEyeJy$w0<+)3^F2-vL$4D_Z@{feOD8y<A&~Nw8ZgLoK3Iu{L>>fji`H
ztw>aGQ88qIWW*}F%@^3WOm(T>=9^MFVPM32dw?%dU5*r~vG1rG<wf_@#LO?fcUFE@
za#?gFb*|u0{0X_ft0Ke88{dTi_T703E~PtKm))NKoZM(>X|?wxDRomL;k}6(Vw%sk
z_tnYbsB*|lApzZ0FtzvlWLTJxNCSNz2+&0XQR0zr6uteyINnRT?L&_9dvAs3-Uc6A
zoqJx~B@$Ij&+4+B>JWZMV!o&_g`;vvEBCcK>!j)WOTt`}JYi8!_bMDy4O{D^aZxv6
zGVAi!CeOSeuCZcXab6Pn?*OxkFWr!sB_Oxp9A36D-gaS3RFIJ(n-a-mOXAq(iq|uT
z4Y#}UOcCOw+geenesbIo3F+V;+mb_&n1LyNtM5~zvmMQOTLPm@b=TEwdwr&Gvr<CF
zcI4(Z_xIRNU-`i5`Ip(Qcv(UXch7y)&#~R{OimyS_4lmd*4Y!Fy6z(CR7Abl5#D0O
z=|Bia?e08cOS`*D;>6xWtak?`?tfo0x;A{<<XP``V9q`1q?gzM|2r1RUOVW2zjq57
z=b->7SQ=i(wRX6oT&52XhAW)Lj?^U`#=C{%8IAI&OMO$P*|8Y+cEUD39Ao%G6sykV
z#pdm`k0j0^7ZRZFNwtDzY>K(Y=yz|-;&&#?$0Ik8FaaUbppXgCC`RYLBwWy(rgm2m
z+=}v1r;83cv2djvuPbm({!UoI$8rRuJ=M;X-K&)U0^)6~<>Qs};?)jTfJX-675`T7
zibR%ABpsZ*;C%aJE)q`#bp<|^v`OA}mwnnX*57KMxgVA&Qpass-1+v|3KMCe%pgB7
z1X2zR{N6t2xL(vviKow3S|4h+1FB$<!p0{1LLyq3I9_h+i}`43i3m_^!<T&A$<vVg
zzU+TrvIL{#S9}FZ&_P1^YFxj%<#ct(>LFYxw$l@o2QQPa#TuQ3hP9P@_ON|D_TpA7
zFt_n7E$!?ZbuFGltza@w=Neia^_%gf##Bm}>MnjOo_T1Amqhw@?#YA+lk7XW^+nJ}
zCSvQq>n|*C>0i`t-wQ}3DoN8<V?W>9@B8XWcqXvQeh>pG^t2QVzy|lj3Xv-3>a^^%
zAH{ePPFkYNe7qm0j^EeS=qoXYi>X=r6UX-|LLTg=^;>1~d!s8ch|Tu1`~@ggpHY>}
ziml6j9=~{}DJ)x1Q;A}+U&Jq*sA7bc_~M@9Uk2R778E<1EHA&R>swngHdMv<*D->w
zKxS1MFC$2<`%U6_*s!zgx0P%MYuBiaj@SqqgZ(bZh{7ePPrr}1!#Tk{t~xO&TOMC%
ze{etsR+ml+-QCNO5C7;$qT5cx{*+TsZMoY196v0gtHfVo1;qpppO()3*W`|+IAo9g
zE#8qrqFg_vjr~1W&NGGV@5RpdkJ$T;C75IEpNUCmUbQ`x+3tJ#uVm1r9ScDGJ6?_Y
z6(#S;YXto#SvMKg-07_9zlk>nEb)*f@>r><3hml&HqXszE};6Kp0M(Od13HtyV@Q&
z&&`C&9Z`7-4yHXQzFd!`^$q~mUij8OIDaW_GASiWA`mNkNc@87wFsH(i1_qh9y+gB
zCEGhYt{@I^UW62p61J?pmv-!UT^>HKn8t`4RP^Q}=DERBGN!wARisrVGPJJ4_Q;q?
zhgMaVWzKS;_)$JicPn)QmfNG}73(5p1Jr5J9^+s@7~RY4vH6d#_HKJzzDPqR*xBPN
zMs~@KPU6QM_Jn{5>|m~Qb|JFXQ5F$AF}`u(+b^vr%_}xiT7auTB=^XZ>q}I>1))+j
zo-*&aTj0E*l|9v$z_Q30u&2eex>=-7Df37Qq+(7KgZK2DVg=oObPYZuccHK&X`>jc
zh$h@>&x~ILud>rrcmYO$6#T4t#nLwWVtaPH1B*s-jZ`Z1p{@3u03B>2_iyCS1NPi`
z#qfMJ4Xd6PKWiJPg~k+z?D@G`^~3{BhY4?ZK}?`qT2d%cJ^ZPVf)nEH_4mLb3kTX|
zFZ4~qE`tN^M4xTch!OT8|7(AHue~@aXKXk5_$TGR&KiBz2s^m~;4Kq&U{*b)V!=(r
z;VD9MJv9N~kjmty<nGfdB}$icHkN+6O>m}{B;l+e*n}`cKUSFCy6p7WN4YtU+8Hqr
zjH=wpQVd+1b=CAtpIpXq^&l>h3TMp=yA+Mb2|a{43=0sIIH>+XiXA%pe->0bLD=wk
z8?a@OFLf(N<bpUlrAlN_lE_@Oibhr>$kZ+S@AY<0Jk!BeRYD30*r;4jik$4Q!t!MU
zWLItUJhzFa3?k`BhlaaRj(MRHG5y>+a*1C@@7-qS`KsG6SyLHE9>()4QOX>XSYzYt
zg52)u-2^VbEJpJpjU`yXUY_5Uur%!ItiYMlD}1l0(9vdNoStS7Ryo#5PEDXOHs<!W
ztGT_6kHLzkpdeIY1dbh>5L3(jBhiD?V-o{K63tZ8%^k*4(<LhGvq?Gl;*LdF6fc|?
zJ{Ei>1P3ccXL25HwWd;G>x=4;|Lbt{HC>$e7)P*+P4QJNc+rVAHM!kvu8nE=7QDCD
zE(y@p#;6plLG7FFxFI;9%Pvg}%IwI#Dd56)Z$?5>?4707D!5f74s06Pu_^Zh6x{&=
zRCS;^pP_{+yy*zC$;?VCQeTF8IX!Y(>A>4;R-&u1tbIkV%}#Wo;x_s)(X`o|JTL8C
z2<=OFGB;q?36R&WWQMZNbvDmeNOy!To1fp&P_V0*8WzNNG6gBqQ&i*E-@@e9<eI1x
z)n)PK1~d%SmOTP%6%X8KmphhMwi89vW3P;LD{NVHUhiPtEx|1m(6cC)Q_3Tg+0?Ms
z`~*)_5tLz)Z9^7Rf=SDPqppUv=MGOEH`5j;;f(1c!57_niBCo7+zMM7<GOBK#Bl3~
zXU-~IMZLn8s;jdYsg@0yd(aK7h?-8H8#M+0*YcnWT{G$b@<c1A(XPZ;;T*D9DIXKh
zwskv@v}fC7D{?tYJKF5ZfKI+-><?IUcUn(khpiRaiXAlSGgVP^H(GB@N7`Ds^#)^g
z`uupT;-Ia}*~TsHwpFnJ_YS#I==Z&Fwa%^@^1H%f#LnV(pjdDHF{AF#a=F3k<+MFu
zR}aCmf-sf5c#UJ3aJ99i?!&7b36xnR$vP17CTys$P9p8-QoudL{hAmJ{)60BS}NFe
z_1T~ra=jlMFP7YGH`LSLM^?y<LqNjffjRH6z1F9txb_HpU0qD=g1}<r5_)~oSYw3;
z0lvWpKIN2CPd)ji_Qn`kd=GgEL(`r1rre8Y|FD;9$q&n${cRQ8Y3(aLliTerdEjTa
zcT#`st+_P#2<}&EnC!Bfd<#!lc%I!HxWXM{w>TtlXvJUM>c7yqz1eQ_55{&cvbQB3
zCa;9cy}gcJB=ll-c}F~RtsPMp6Y@LbS-1hX>vq|@>U+{V-6#L<SgIlg6os*`7;aQS
z>-Kybu~M91CtAjPlB*P4_S$=$9`WJgk9l8w;8AoXFT&oRxLGE`u@CrOc(jD#ct7Y5
z#t>4{rM3$8AzyeacFooP*GBsiyQ4xW>$0S=bc5eniF~McdN9X&O$ADp2Cu5PHbyJw
zoM-ZbdJW@s4zBDNls&Wc@v#CrrL|p;jcY@a6G4@&eUSG?2Xnz?^KDb2XA&hwaX@bN
ztsCF5Qh^6|#k6jWlF^7$snQ3wCFnp=KSyk9;!I$)*|wZ;hUyLWwcB>5ldB{}?+6I}
z;wZ)MZD)Sp)lOE4?FtNK`yZJYWwS5oaFE^Z#4UbFm10q<y>q}lKKPh6nz9r>Jyy88
zlLHN-WX0Z?UD^&8=RKS8m+h-KQztFIVD~3pr~yGqSUV8!VCX^<bWQGHQvbNcb|^qD
zA}7Bqh+cstI!+S)4#)Rm-QLjZupPPI<vi5E;NMFYqE*Yh<Bug<O%>_Tx4FZ5R88`j
zZ{bDV?e^hZ5=T;(*-=FCk0gddece}lKg1Q<JqaihC5k7Mw23=HF*k*&+P(2*MkeiN
z%Z04NHc-LE$;LbixYalElBwp9HWce$j4$Rxhd0zr<%MH0j2-&u5Xe+TqO`IE@{M*p
z)<G*E$uCsmTyG!qXE3iN<jyZClH<qo*@`|&A~(qCXxJwbNqt>P8C~JQ=%0)cWo4lk
zH5bKE`&2yB0a1lIwK@NEeWpG$eq(-S@EHfG+_?7H#0{1xZLF%AdA!W$@|Sy5d0%T+
z)Dh*8?DKiK5N)Nd6%G4BeYX@88&Xp4i#{h?O_#V1uZ#4hczgZ*WkgQFm*W=&(J8Y*
zZjqt~*jM5gq%!Q2ZetRQTVr3Xzm%#FW%p4*;hkTLUy{SnwVJi5n(fz}%v*YUTkIQw
z6^*9i_2i}7H{(mmuf-t>r{?e2;Jy{KpvYI*#N2JU7MR2jH`ur9mxG-F{TzSCv9_4d
zmWa3cZXU=BTY46uk$f+uy6MCQQMQdJsP88cwD$LQ_jX^^(W<23A0#uPsuh8Bz4k+A
zHe~Q)5NX(ta!DXV5lHGa{^Pp&evhix<4*!uMOiD?d@BdX1=pV@D5X?tsl@WLx^m?V
zh#qvl#1y<f#?NCgnKX1$dE*(_wf!RLrIk!2T1~avFB8DxeB%y<Df}v*xxEXA@*(?m
z#il}DJ@HWnoJ01T9FyWBJ=6<q*l!bLoMYXsN>H}nxus&&BKv*54!RVn(6B!wJ}`)@
zF9vQd`(x$VEvmx*r{E6d%Hi}XnY-ksl)ZLgZHxUmt_?+$ZM45Q6B|8Zl>ODW_uiX1
z`ut~qi^U-PgKcp_U7{*d^Y?(oJ<6(%ahLrgA(p(PjB)#Cd@n1fgal%ff&u+2w<gTL
z2RJ-pU8=~qodf&tx-}kAQJZ+#f8xva11cIMa_5FE0dfj9+++WZ#i7`6JZ-+){)*Dh
zwyW&r1LlWq@bok6f%DxG=d!F+Y8jq(aIxzs@3jZT_hp5^M!!vaaD8?_hF}?b>>=}u
zrRbCkPPT`}&zJ|Xg_f!i_OLpQxB4t)89<~mRgj$Q;qkkh8)atz-CT0v)3Hb7?|2C0
zVUyP87n}Yh4$+rHg@A@V%0U@1a+p23f}$co(%tvlW9l>6qCGY2v2{fyM4roQpFJ*~
zg}lLaaL^t<-;Ew_x);t=tmA)!cAt>I4q*OchQqLP)qhktC+$A$iSvuqvb6`VJEF*w
z=7$yN%0BW3B+ETH1`t7ndznokzK1^*kn)uI#~-Ttmp9&bBj)a>&UYg)Z$l7+mj500
zv=~R5SF#Xgnt*Y*C1;;K-Nz~AT8`Q?5^Q;43bA{^LO}rMuxC2h1FH!_LuY%|{9;p_
zicK*wT4>Q(%X0##%mUsHXbMn0YR`?gJt;|g-7$M!qJSb&Ev3ou^L-O%kC|&P$a$x9
zcK5g03I5BqXDqN6&Oh#!Rn<vGqC7FzGi&nX$@Zcgw0rSlcr50f7su3Y$<QVqwUgqR
zNMDkGq9snQ@0D_s%38Bia<nn6*dmsrQ=S@2^t)<Eof$hVzIS0t#-KaxCHZ}06>enh
z`%8!Hn`{szcNLxCkewdiD`d&d<6G>EgcE&3y(m=Q=*+|nVVB7LXT^Y?f#cl9ouR{W
za&m4kfdKs3b+}y&=g!69j@fajBqJP?OG}E5jf{bl&B$w8Vp!~)Txp4yE@G*T3OaKC
zoM&zgy>&*<KVE`#N<1I*s0!)2l{wL+vW};cxfki&++jkBWuQAR7t(+RN!jVQ^XG?P
znl~^SM8hsf%pmXJO@3K`>02dHeTTi=Sw<gyitQD-3{JGn_6jr{6VD3QV7bc49h(F;
zUKttV;*E_Y(J0svVa3MJ505}ouUgbLY(hdzC1;6=Z8p(A8(5ur^=Jo^5?+J_^rs8w
zANMBn&4lGluBW(|>|rYDo^UDUBHy;jU9EQUAW6YEWU5npV?jP85!p&0C^Jm0Z^~6I
zvw#=Zy3?j5C^JjY#wGdBzRp%QV0uC}?Cf*wQlI(qYfiHn0VJ%V+#TuwG?YO%1+Ge7
zB*cCY$4GO<uor^fhS_<h<8P8|q)KJ8Splecz80gfydtmI?8M`xeKdQZ;0Ze4oEXxJ
zb#E9@FMzpq8kO@CPsKV|ajI%Ya8+(9jjd=98NBAlH!><w(iz+GPFs*$OZ^;FfF4^I
zgUIB6BgCr&YuyH!n|9g#CieKE%i|dqWDShae7C)_zNb`qcTu<0u{}{iyX5IM+{>sb
zi;@z@jK6Tawa)jFdG1DMi3OhGC*<vtjm=dmyINuW&}rM?K908b0H9Qphmh4nc=7z=
z#Uq%W1_P{?pgydO{%~M*D`h8WZe2>kEls?P?@%L%HtUGt!mNThcAH(1&t%pC;iMke
zvO2Bmnv+qEywU0FT1*Bq7+#)3Pie<gLdHXvlkvT`;Ag%WPTif<+1dh=w-x?%Q~&Zs
z)bhMCNXJjT<Qciv<d*2EMCCn*b+FtW+;EZ}Mebm)a~9fAs<LWkG4(k}SN37BFZHQb
z`kYhSl^JTQVq`HiDIkPKBZjIa<yDpB6da-ad1>%_Lm1y={q@Vi&<+tg#T<RnBy@<I
z5vf>58cb@`S&OuAdsX6^Bnq8EgeAMy*VWs)!d{(dN6eF%mpEyAO`Ti~k3C1G_zUd1
zB(&*0*wn6%Z`Dt29|TGHExTd<@y8d_z?Jk>ke<|~S|$y<G2Znk@r^QB*lV5aXqisg
zlGo+Xgn+NG*T?I~lPIqW(%EWn2(VDpiVRhXS>Z9i+uoRLpXkA^Mq6J|(Cke<MoGN9
zf-6xf!JFr&+XV!6FPGu{Eg{B<7s!Sn%Z~mo8-hf9=c{jZ@bDQ)cDN}IOv_arc5{WR
z_!wg^IcauFMWmO#A8aXdYpkyvuQ+744FbZ64P}Y_l1En-9)*(JX>W5t_y%R)-kxhz
ze^jEfr8T`HW_3p9*R*O(?zDHtw|5{)9JF^0fm?1e`PF@~?@oSG?u;UpK^pSD7lG1l
zuj|65T}&}5_@;Kj@9}$diu9fLI+7?f-?!Y}SIKl}ro^Cy@%?#rr6hOR2lArqg_~m>
z@39Z&oRBC2M~U-x<Lig&&@~c~v$PEz%I#*W1J`mtkq%d)7w(8>T9s_`81n4S<S)zn
zC`*Kjxh7vKl_1ovVu!7*lMR;U^o)cO+HULW@9K+<M-`*@(0UN-V@xq_ovqEt(i=iw
z>1@B+Hab%s$F(+SAlu|GEK?2o&5o!(ykMQk3U?*Yie->ha7zFnK7;#|lOq$1tfcF>
z0FT+$x+pT3!P8y1VzwpnS9Z`n!?ycBJ7{O6f+{=W3vnsQ1xGO1?TlHx#=W}%bzGF3
zzX1VxSM1g%f&o++LB=+hK6fX4<mn3YWMT0;XPUoL)ztQcKq%Qki457jAd7gg77ir`
z6b<a|x)zFGfwBECmA%Oe@Upfniw+XDFW=;AT=8IKqwRMtIHs}84g|O;@I|87A$DAR
z4hI9>GWx(j4%Knw$l3@;-e`woMb0$Jol@4Spf~l;JCZy{mj|Z-1JF@dm#*lvV@cNK
ziWnLT{^4YZv(=B~Bl#ZI6_r|5M8F0dR2%G`<V9x>Khy5bot&gNb{k0QXzA+j?Vz#O
zeKCwXG(5n>1*nha1O*V?$r-zQJk}>UP5B(+15zBt8g=GB7GLJgs*37`b=t>+--v(h
zuur7aNe+N4IG?QJ4yy4(RfA7C)vsvh^ncpdMUX&={`^dmWVteB{McZh&9%0*w-Qs)
zW1mYNLJbA^c^mfmN_z^GEOG~bV)kRRl_~ZGAC{0IYGy7+-Qf=ZqEpGLmOlGZE?^Ok
zV<z~r1E}x~`-=Y^A4Q9OwL-TA8#WzBkTJiO-%`?H6>|UAb1JgvX^&AVO?|`1m@13l
zHyw4&^2L%Z?Lc&Oue5LH;9OCJprUQ;u<yjPVm4($gk$?|Jc9u#iK1A8z8C9v3zTrJ
zV#@CFIgDKJeJ7=9ps!&+aA*iqeyRPi4u-{6gCU6gD7GEV8BWImE_rei+kc#ImjH4x
zvYqymTt5lOAl81G!XrV*<!;uo_*u}4ESa={sjH#@9+RqG$}b!d=ZqL`zpS&AJiLLs
z5#@dr&&1Z`!&j!vuYKkj?N|08iL}{olI#XH-6yNBg1Wp8snD|5ew*Z6vVsSdp}XvN
z0p3P>+1_BkufT3Z?)gJKg|w3UWBy~70uYlf$)P5FrPu!Kh((0zwZ9}V+-C%km-sE`
z+<u+>wVujSKMiiGvqgmi`&%rvNCj}+-SE!8$1|N#C6=m8(Ld^J^qfxaKshIj)BnoX
z(IRDP+yn;yoft&R(bcY6q5ttAdf6%aZ+;`H1{8obTi|9znmnPs<G=zR4_M%KDD-!Z
znzM1gdLI~XD?1&g50LBZLGjG1E2469#XNX{8zaloBsR-l#oJLw(;gDvAhQ;65Ov2M
z8nb%LWqNW{$Npgpij@py1zp7+9^Wc5gH>SA-DQu6Z-stom4m5okF2BC;)tvqk<K2q
zpcpzwpEJ@PofAx^?6o~+f!hXkCwL)WCH=uK`B1I%{4RW~e*+&q%#J=TM_8e<(vMFB
z788AOq4AYMO`Z^lkycTn)VOtp7)5g`F5-y`!s@UJ{Y!GplM>^_aP9v5Cs*vTQb3~=
zr#&T}aS?MeBskt-PmL9h8g5UkKP&ZJcN=NVi8PLyS3pP~|1%aGcXJ%x(=M9oJaa(=
zf|9{caXn(0kzSs);P?a47FI2`XD2Es9kkP)<Fk$&HO!v7px8$*StJp@Vb60QNE%*B
zbBIPax`RF6Um@_5QWg|I%=OdMnmnT(J0Tx*G#aGdEU_2X4b?S@g&Si+N%FT7E7lcu
z;N~ii+;~y02SY7gUG2q*4c#2t8RU_Z>dus~0uG1bXa}9dC)aO|Y0tPrciAcV3`t0J
zGj*k(x*%-w6UN$U@w-d3euQ39p;RHQA=x1IQhyugWYJSgK;7v>W>7ZgF_`xm0jaW`
zbO)#=Rvs~#(9cW^i7BwU!S1hTEeHek3Y<Zx<%B6N2%{Up`N#P0DF5xTvlkqXdbw@I
zwAJv+)WdyeDc)-%60UO8MB7_!WIih;zDgR7ofEL&dPi22Q8pd7qjF|4-YL@=MxxO%
zhV*U8tJ;}!<C$AMidJdo<y;&KmC3+vW9KgjTidkB7up3$H?7@hix@&*mUu6Buiu}(
zBQMXfV5ROG+H9{#xJ&4hIDz~Fvhj@Zw+B|MMo?>eiMJZ-M`6-58|S-USoHWhuEGjr
z7QV|S{Li=wyeh^Tn;5&QODzj|NX00%O~AsW+&iwKmF@jph8O1e(hs%I_pGzY^_c?b
zN=H7sD4ux~zI%Bt9-IwoF1&}`!O6-vlFp~Z@0ijG>)yj|h`Fk4jFL^J#+VY2_(k?8
zo3<eI^AZ@>(uq)TNuF+EUp95xbjJ>zDSGYFJd=GWDu}x?a>JHWt3)AHO@S(1lCrAr
zV52<?yg5HzT-^0G)A>(J8+lU=n-$+_({wxrU0}15Kt_x>)8;rakAVHpP1LrvuOKcC
z)Xqzk6?>w~sPp5Qh*HYUUQ9s?5)BAyzzDn8YYP{o36DOkJd=^gl_(X#6&h>jR#qjm
z&}BLQbfP><5!}lixT(mnuXLPYLTy+}EX{){6p-W-U`4DvL3(s%RC-vSw8Yl?(Vdrs
z070CZwpfDX7rEedrMK7JuMPqbSqeESzrYqJz)gyORm$0txHEOai_0l(qI}1>MHsoB
zv$NF6VdBKeGp0@$ZygC#x!c@&!!f}LdBuX`#j~IUMLdh1P=K(jOqoN)l;KR8Hhs2i
zL9trFxQ+w^+pYh#&bo(db*woghi$n7-oRsrmZ0c~uB6bh7^+E{j*kjw0O}hN6jsDD
zPEJwZ0K=|~XC9<bPLB1&vl7k79?1LKNK8?-0arO;sJ-!)#43fdpy^v*?jAs@#=LHw
zl6(i@kTz;7hmf@L=*2}vSUYT0j=zc?&TT#ItL&<H*E>oVEss%@_0eBP@n(wBrV-H4
za<5L9fQFG-=bd&<z*MND+;{4JSZ&busywfRWEF$ywMq8?CZ=_$SWnzUplD&~67oY1
zMkT+xA~7s>#ErbBlHOp{`vzF2BAJy9c%7e&%P&9Ou8+@=s73SqdBkq;pGTc}w%u5N
zK3wA7h#j)m#^@Ug#T8l$!gfF0>yoozfM~zE-Cmz#6k}RZqU{aNMVd;@g*R3#9Vr3t
zy3<`hS60+fdsB!;TX9Mrv^OWQP3@*ut-U3;intgBf9z7mQK@tB)|gi-m%vFfuiX@H
z4>6+fi!R8UlL#mxjEMrN>6YYl%GkMe$Y7F-)E4tL=PFI)3EJE8<*|KB?Ct)K?h-2U
zjs?e`g^2fm_pmBY?VT|=kjTJLs)xQSm$|sF0W*DfonMU4gZ*!hXHc<JmD1+!Jr$RV
z%*W#{yAT^Cwy*fx_xj#W9y86}m!Bj1>SytR_a_#X5blle+-4t$w<IDdheM$_A55|x
zj{~)n)b<a>jZL>_TU|G#L==SdCpXd^0b8-fGL&*rL~)ASP}1I=@uk#IZKXQ+cs1QM
zK7L07Os@?tSx$Y({Z54L@9}fGF2L3W#tFs#a=_N-Qc8ArAz#}Nv?as;TH6?J$XXx|
zJ&rKj<Zq~42qurs`37<-b~{MtnB5i6)}eGOjd~k0%9fP&TTvQGiS%lUK9syC(APF!
z)WGUVO}5=Bre$R}u>?DOJXEz#u5tNQ$e9i}*AX1I$#w;%oWpv>Klu{Z*=`>cD_e{0
z@oxrJU##2Z?t~KB^{V@uY_D%5!rkJIp5^TJzTAI&O6&M<gY8drr@PFBsUj8BYtjxl
z{1j8;hCP@wfuqum0A7D6Aod8I(z3fy<Uu=}>;j*aBF2s+{gg0CSv}+smmTqFlHIg!
z#S-@0u@shCT4mdiX7k~gTQQ&F|I}f%kL1}LGrP&|Ng~6BqqA0vX4t)X;s`?~`D7qp
zELNh@GN_#U?RH<zQDg_KO$70M)bUP&1?kg1pPHCA>|^ynA`*(%fIk;4AX<!vACKQ<
zLKg9;yMjAK<SANbpNQYr6Yr_2rSi>kK<-07EkUfG3}EDTUjw34HRV%@DAKtpRnxFf
zhbT~ZfG8B7l7s-2Z~EU0<ulG9nrL_QSzpvyqcN<0E*D**XS~7W=i?cN47ou^?U;Qb
z;VqIz;jJ7}br{oI2kncFWpAKGDs>Fu{k%xFqISfW1M@-MuqRj0kJ(ppi+Yy!6Ugw@
z#2sRL$>l*+lY>SkA|?8LEn#MnO6yc;*4JZjL0d{uy4e6D;y3bbMb-kdtd`%5XUbia
zsX;XXzm?z(R?TB<_44|5z*C|%V7u2Nv^ksdEc@>yQy8-Z?<kd?R9fk~z9!C~)I|GU
z%qaLg2yT;oKc2}a&mJS)ZL%LYku54gQ$Nfzge<9gGe1g6l0hW)yWK^l{A8x26d)Ty
zmzO-+ML%QM5~~2>Ah6n>#D0=ogtBU__R~B|!-fyHpVhrCg@b_0OBZR%ZCJ2=o*1P}
z8DX3f%<LBxd)hMvLFuyB4gAYsahPinw0@QE72ssjmGq-K{MWhIW_58{YQIT7I%B;3
zHpZYD73}<XIX`C$fg167hy6ZzhKhSBu7XI)JTBvu;SaO@W4=S9n9>HJ-~Qz5rtUu3
z6Mr7uBPdQRYK#3P;7wQeLCH&8VF&E5@wObAGPCSMuKrtKP8$JbA`+5)C@~f+sjB7Q
z>pEn+`$xjVWzyNFkQMu93=kzL6bQ#K#|Tv`U1jWF2~IiRF8uq=`0pfAkSW7&^&+zn
z#m4+6i4;AMbPDfd{a;_cbi-b=g_)??O{Z4EA?yJQ!`O+0rh8K!5(pU0=z$B}h^b>K
z#>a-U&|L_zVrcv9K?{okaoX5%_TYubJ<NxE2z$svH>v8#Df&{f%R?82MXj~Fgp56G
zVKJU{Ew+a*4CCeuRm*rpKqG{5cQ{Lx=4FqJU!<YQcW?)Hiy!Jo#e9mxlc#MBDR__e
z`BbzR8y*SO3yUQVj#|t<kBv`cAs0_p2i(WS1fu8-LP3cgl%@3XzB)xIF1IJ-QkVDj
z+Y=M`C7mq%Nehn`txc+=3qfixEUisKpNFGnHuRHop{4U>NdgHq<tYK1vZ@tzx*N{-
z)O@>~D0h{VQdagfUl`^XYT-8Q>G`n}XHl=8Q88C5o0jG6uxIASi#j<y_N>GTr(N;*
zY6tr4e1?Fe7<`<u_M9LL34n6X=zI^6%I&!e!+h4(zO)~6J@xu&vtrNlJwN$0d;Y>=
zBb+Bg{tFU@Uj9X!=@^@>E%in@L**<g<4frqFU$oZB&dc30<xW$TQZJrV=wZtsmeXY
zUhLb$eM$C?ofMx+^OQoYY}k`?IHucE=xM*5l5}7E8hLFGR`EGCepdp7!tiyyh{_d!
zL5X)e?S7jjLy5=M?zET0Y;l1V3FoCrLg+4Sc6z+A9$`Zf3(|4V$o<2$pl95-{HaUw
z%y?VRPy=9Qo9wK_=d_NiF!c^|++blsx;wj$R0^qh*o;g!HavcDTfAoqx|?7`g=lBt
zQ?i_ojA!mNArcq~@0^A1HIQ3nBN$gx3fBX4!KOcURN&4lle$df9jzR9qZfw#zgPKJ
z(B!#JG}pe|&Pzbt?5h}mJHH~J7^+LZ#0wmubbRS0bn0dK6%118&Z;f=@_43gP_&$a
z&0mq5D#Nx6A7XQyz-%8k#u*A3p9k($;@(&%v*|M~8E4}bhI?SK6tBW{8#X>yGNyBp
zN?X|kXO~NQ`fQ?qOc{k`Hp%}^@Ft1X6<D`yv8y)O2d6O2oda7>xt9uzaHNa!#o{Z(
zG{q^n*q@ceL=+dLh1-+_;==9~%TzC7YJBJ=xX{%3a9ZrW(;S0Itm2@ru}kuE5@>q+
zSG3#oWaDZzg~E7g-A+n4#VuIy3086H*o;bWvWako6nBF<l{Do=L}hst4^Dn_p5kGn
zN7&4y;+}R&z1gfvo>V%CeUZwx-Dbyv#B-4e6~JV37KR~KVFRiKw8G|A<jBIQJi~^~
z%V#v)Kw2o4?D;VqCyEQb#C+R=ct+F`5}B_3gSIf9x&7CDSC={A(WF>0KbPklCGgUj
z=_`}S<oT8qkDlY@GD?4CYsr})KSU<{2-n*pKLE3*TWkC(5?FIDS*nm>o6p(XepN5U
zTiWw`xS6a9oLE~NbLv_H`ZBV|UwRzplDdT|*x@xUhiDeIG|n}L1-=uO*P{Q+HYdBg
zvPe3T<s%CdZM%{~u#!E!!odTS-LQ>iv5Wdp)G(;Obygs2nG&Yj0}?Q~A!;m7a_|!1
zUH(rfr!-yY&R<&*%G+#3U_t3rMRkG3uN-_P85)UVv-Q-KgUe!RQJWn}W5p|FjY|E!
z9X%Sjw|=0PbTO#1#b6dfpC9EF1e@E+n6Nm-K<g1KG^?BgjGI2wu5yrPs}&@~(w`ew
zgfQ`?9d>mE8RLkHLUv6+u;D&<P_S3tmhQxPTW7Cw2=2Wd9&v4+RU*vWmv`B#V<U?J
zOv^1K=8n5vTx2d{zNVg7{USk6#ou>umaoebi$`gh+S}Xp0gd3rU{$OPie8YdiMlU#
zgRgfwu>v>dVk8ZUW;jEyt&3KBS)HqG_PSi4lIUP4%A((hy|?(nP#|6(Z1^@z9Jk>*
zctbAol48%cHzp}4_Kt2ZNpXmi(p3bfm%(_mzqq`;1%DEzNJpIFe(kM3W~)+wWtX@q
z>0knQ)SYy50$$A3+Tm?>O91Zdvlu<tAHwgUeVMRtbzF?bFne2$M@wd69((L<F&?y3
ziUAx&+<kl92Q_5K<Se<!-;vzy!V=G^?A3S1%MvyfVxR~Lv<vL7@2bP>)OI3q$O0(~
z{<|x{-UAcw83-CI34eP4B|%KV61w-_<Hwman;Rjv_xc+c!92zte}cVl;qfPw5ZdBj
zg|YU=b@={@vqKm#O38BjKvL9_K0>0Uoqcd{`(PvXq1@fEL=R#xUhO|qGrqGI-}xPR
zV=Q6jI|o<FMK8?(LjWvhkF80lXXr{?n`4sKpdt|Kd}CTxEwlAigxOLY3|@JxhJ^Vp
z*ajz+fz=bo+QvFGXUsOm8YT896+!WloBa(Ew|iL7T|U`JoNin4Q|NeuZS|FL87_jz
zY@7eHi8jfSk+;`@6%SR`W;+s9J$*|S_1jJ#xwB=3?Fvk|4Mxe3B{2YCb|>ZeX=&J=
z7}!NH=Tdi9L=KkLcHXczQAa&2z_fiiTo<;g6-18g&rOEHq@SvC=m7_8)W|dKVEl}z
zCmUF?M<0sWbT;MP8tfN+xWchb0&n40?gKgE`;O5Q|KQQsG*SKiJ1&-7_(Hk+?O2Gx
z?F7HshXam{WuHn`ynQ6svXs!&RoKPv$^9XFq|NU2O>b_YVcI~PoFTU)v(OwHvWAh`
zeHH(@A0_E(&kk}LKN=Iu&`=zqoT=k=v|Zfle%geLtX1~0A)}SfL*>qu!1r;N3|@7C
zeWIeuW3{DnN)F#`pLFc}|9CL?$v##0ZHq$BWMi~XJI3z4`JTg__L%_EU3=jQ<xZp8
zKWv-E(gA=w21w2<+mAieK3DhrAY@1mB{q@IJK$Hg*ca-rrOK2_xY_?h+<QRTSyf@f
zTa@05NLM<9mR{(SNrg-jGD+x3?#$dN$u#brNiu;T#R8}l6~R8KGbyC^Nzr8D*c$>O
zQUs(|ktU+w^Xzi&#kIcmuYaw7tzVZhdGGt4vd=#I?EUO=9u6*p8(r^zNkn5h0p5UO
z`bZE32MISjv+mVlB9Hpp9mt3!K)|4*$80Lt?GPG&wsjrF>uPiIxW8^5VYPeYFFsjb
zx52ZD$18CU7CNTY^+XIXhbvuRbN@9>eibC8#vHI?oN!Nma(d7@7RpnFXBePutfy^{
z^hcuSGa(ZZ!kO2QpueTX-3H02^u|DV0mQR5km;>eHF_>RXH43g^<kecbb}*QPbK-g
zPYhGyj#3d>Y`DIVR+)|+>Irm1ZK7tcw^+RB!yMdCUh+W|#Xn&4GCt(xIG!?9k#qk_
zngJ!MlEx?gs_i;JQ_STD`uUm<+w@GcKwr1z>u&Gczn{Dj`!w59iJO^f<jo)$!ljwb
z)F^MI?x@3?7s43f?U*1&%Af?Qp*9=?-~ZviJDv*AnUUSRs#<*r@=nNHmas!~qz~2<
zVXnzyc{dTJ1Ink}@nW67m)_ovjc)qd`_Ul0Jax_Tfzb{zzw|oI%)(FdVN4I^TNMD3
zy!<Py&3s!!a|?5tAO%12kuQM>Y$aS-ypeG+mJRgpkY>0@Yo>hcgR~REw~rdn|4Bz6
zRTmy%v?JE?IKw7mYlPPS>r)|o#+ZrK2z!$$jx5sX(sZg%O;KCE9MQ<AD}}l2F-1YZ
z@ffihV4t3%0Ag7Q0K_SQLh6Qw>{4AEddi;u3z?SLMuodf_L^e&r0q~bwA6i=|HlY|
zO|thC#UW8is;&~}DsElP$UaltR0j+QTcQcZTAS=^FJdCBGp=Tt?C0GYL=jpo6{r3E
z7xnQ_JU~>d`%(^wZ!De|LY^UCL=X=&QgpYE%Q3sJNe=Q?iMO)in&#RO+>?WSFdfX(
z#>QO<8}1Mz)W8A#<WQgV5@_2@OW+kCb~bX{^W`w3F>*4I9gEoe!=vX}8EhDsWJgSK
zsCEqMOnx@r&NpcB(IdTcAlVG6gBFk6>nRG>p>#x@9Oc6{MUh7?he)mT=rkcFub6D%
zVYX3@394X{und5abAec%>e%#)>Pv8SV=fS!9dew{(llrTy6u2hoKcRC-;3N~#rKb#
z5c2?NtsW)BJbyl(Va2$uIf5@taioBm)m9toM4K*v<2E_T{@#SrA#!pI7CV4aqj0P4
zDN_{7zW97O)!*SziP3E{jRdC9X;U2KzLGNQ^jKr)Nxaf=8)7NXuxD0R3eZ&pw2m{Q
zPc<Zr3}m=f(#u<*Cs4sxVC-kb1iRPZ#G(ndNX~Xb=~9BY^d3T7(R_t@-6iJ)drW})
zHwRGUT<?;M3AJCV<}QX^(jn(faa1=u8`sBS1H)T|P2JmCXa&gCh$ODO^zjxT|6sS0
zQ<!(pAc4M96ol_?pM*LXZPG6mk%AM)jpf@yCZSu5QTm5O!5u3Av>^HhL<5&2EH@(q
zje8Xe*>UFCGRPaGNDl<x)UOWk)+QMotPiwahS-WAfdQpm<@{g-28!q5q{zwzv1G`Y
zsF4d}X%)4gu^L{{i)>K1S_DT}RxY+dRaVv^=5KAIToUBKoB*IhgP?hgcrdxtcJ+{f
zL!=~@VL14Zp~6Fr^Jw=-X&N@J#~c=KB575H=y3bBLj7qYf@B#SrcE{QNsNqHlxL7u
zIx10!H?7dgLWE@@lu_as9WvS)%M-_x%Yq{j?O)}2iI+$Bqp~@yc#DjQ_L0E{c+n@z
z?LlQ$eJIeuHpti@0}frp*(*G3%Fj3-llu{1Xfn|>#R$sy=rU6^6O|=%az%7mkwKRe
zoa?cnb*pd3PAE@1f#GuuokWhdQM9{#>Pa#wk-=<RI$HGjHl;@<`}a$BLbT!+1sdbz
zU_qpO9?@*z^{YaD;?M&*<Z7QKHW#%n@F9ax8MJ(jQMN@dyt%e$%ybS>fL@nKM@88R
zgAx;8FW37o=0c`WrEKA`G9^gEbkN3ZR;C(Dje)H^&Hg$arg3IUkJTx_dXQ_Ont~|m
zGRC3G$OUXSJZF4aMH(7CFPdy4mFXGocmt<wM#x^34~he!Z$`cJ44{?ll%AUv0_bl4
z|L?z2oyciXumz`hjWNR9xwx*pHWnI7K-WqqL}Fd+p;Dy0=@y&r>|!it=TI02T%Z|=
zofL^vrKku8l6pHJE-RIW5Hn4=8lKOP@s0kmNo6a_&L9L97G?xxlMfNg15-x<IcfH1
z@E2<W5JUTJh(+@A3D>7cE)Bep!)-npJKFQ{JJ>++X4Quopor06AGJ(LnIJ_h!zfUS
zb*p8*Qd)yQVZ5UT9KO^9L|95QV|xG%(wuVJxzJ0Hohh?@#F|S0;?v>YW*nW<><Flh
zGROAK_>u`FWpZQee62}zllQB57QWXEPSv627SVz_D<l4Hj&H}Zy!993B+0{SsFGV!
zQ@DSNJ49_Jthe0yiRHmbsB%s3WrtgQq1-mb69||fNs|p{Ku&JA&d~J^D575~@FK~m
zNCp>jo_yKfYi-oy?JKdYaG22r!|Qjx>Prx+dU42)2?F}T?yxVr+p~?iruurd2JbY=
zb+?y{DVMu~sk04;XyZLZU-OAUNR!v#+hB5BL81G4OuQ&mrWl_2h7Xo9iiS23^}GEW
zi?r8jfS3j^=A{|AOzv^Aw8OI6DEG!13`e9>quggd;uvMR<2R!zMZ_)WjEK7z9FT9>
z+s&w?gUNh5792V^UAwqmMeesZ08fq|Jx;!pR(K<!7y6CCGH{c~1MwazEwspkX${s;
zg`-X?-;G@hr4Uz-(0BB`;Ot%P5Tu*r`zCG2w^qmxOup<|4iM#sX<SUF(Jd}NiXJc<
zMgwfaD*xDLOBIhH?KHlqV=xg8C!w_aH&`Hyn@7L`nP=n1+k9ibzrlrI2sOg@WPv|p
zumt1RkTNL?Q}4@hHZi^7c3Bh@7?WwJki{`uLbz55t|NWKN?n%O(&^pf3K2joOPsv4
zwwXsgPpweid9&1iF$N|=Ml_kBZp(~8>tL4TWVutI$SZ`7a1|ED2-cK`jKI9Bc!iIK
z3MLP)t$>+U`saq@Kpw(kO;P?3(W`?05&E7*&~KHj4hGYRQVmbSxOj&#O_0U1CVhdj
z#4VM`pjjLJ0;W>w3}bMeQK6|4afgW03_?~n!9B9Uk6nzC%m!9ss4nyqFoKnj#wG3;
zB`!&ojj>tFMwiMa?}(rw6cdyaStgr<j1at50gtvN8j~$)peS2`pzSK@@^=X{7S+K9
zAhk7G?&ie@i=fwUk!?2Ub9-HX<J_}ld#Ds6>(p;8JM8c}={)&~{cXfZ`DqMB=Rq8l
zNrzA<WM^~@Q(7-1!VI~27s@XGWihN1R>fcjuwARz9dw1g#Hm4^$nF>hyfDl&lAi@>
zM=)f&C&<;e{}B1PZ8#KoY?5DiFGUxM_{e_gzhE^0YjLzhs2KzDt1xU($rvU~-5hmU
z$gk5>c;KnC6cl6C$KSY2<}P!X0$lpPca8ow)<tu=Y4-gt<^ao}Lw+B79=koNt};}D
zWI-8r`Gb+AyB#<s?~bN><&V}TjkUa8diZeBw5Y@~@R}HDr~Ju((a<pbvJ{e0bw7E;
zsD~37k;advVHBJm%&q4QgGe%w$4t`EylBH6o$_bfh;!$jC6C9ZLryV~N$Qg3hv$)t
zo3ul~!!HZx#}jF&$;*U~c<J+By&;}I0X$mX3>#V|PX;$&<JQrM^Hdr%(?|_<=E5?e
z1E{4u9Usi7s|0XXBhT2Vhap<yZ_aLJ!V}P1D$m-=VDzY*DbM*pHK!V_TYw%*X+7`1
zkPi^s<)VW*On;BL=8&bCmlqQ8S(MZaQo|5m^zLb3GhzrAoytpT2e76P?&*cQ_2p13
z00!0}CcILu?N`DIforXbN<YKHU$rGdu^7g@G|FqSJ1US5^Lm=DquD?Y-tcE|g)%b_
zVq}%PX%w1_K+(6-Lj8~6lDDJfQi$w-IMI0j#6$9q{R@@M5V#}%v>s56ptXf}{Nm7*
zcWs{{6R1Mov);<eN6Py_Q*22pYwL_)V2cl8V>UO{w;^ecj&<A^`Js<K$#)!}|4Ose
z3#%Z(cgRQn3}!BlM9VV&w|C~*6*!HEU$VZ;$Enq&gom&Z=gWU=PKf2NL%;vE$(Lwc
zl1w$g(#|VL2Z(s0<V)mJQyuE5$e{TD9#eZPmDul!{j`6#2uBow4hSGS4&*aaBW#j)
z-%=CUAbU=AP!&$X1-tIuxkmPys%YFW+(;vP`xmN9VuD1@_L*v+D2^Mb;lN#X-v$B9
zaw&jDan|oUwP$a&&JdOahwFaUfhC0H>yYxZe>95H2rSltdY^K@)E+C0glbuUoUxr5
zzS|}TPVKR<&lXgeILJPv4qC<FE$Y<=$7rA=;I0KtpQ3=*dl_JD9um~h<Dz(*92&pi
zq#_QrMWX>}^d1)dAY%d2T1%T8?)}(ROmVBm8Z}y2Smg-+-Mkr=6Y0k?M9w1l>{N$^
zQF1n$tEOPAmm`hTxN5vYKIg-s(BLV~vY!TD95q!z?1Au29PPbO{h%7b-4jxV!41bm
zgF{Elu})@JIaCWgZmOe^K-lV=P(QCqjt^$KoR`9%;2qc*1``t<5a43;1hc<BKUI;)
zkw`1IzfBk_Ct3?&g83lCNil@+)pD{AL7j4#4J$63;?Ka=Xj?Pw<kVo{30Yo*B&Q`}
znifO&$b>H^(<P@*^%%Y@q1Yk5LeB6;@e-spoouK=8GJalwE?oESL!#ea$h26c|X_+
z?$3ngR7ho!v#0hbtYI-rRFrw!j}^#9MG0;p97{w<JLH_iPK?t;JSeskqf1btGPW23
zrkhZLt3}T9cj@M&bfC#)*NpY{QJVHZP_sk&#G+ZY5P{9UMsg^Hh-#32(E-B!rZewc
z`ln`4!B6+?4jJIjXyceNwN3{5(zSzb>t#@oL*Y`RrH7z0TR{fLK(dX~o7&U_91<%>
z>w89<obR0!E~MKm7sNu;!Ys_mg=uW8ooW|2UKGzXLE56FFOJbbxPt6ztHl0huE-_n
z8@=5cV@6WB&1YPrx`#@lV+bFZO+#b&%xJ>Rz}SGL#>A7$Wtg#TpT7P2%kb24CzS=Q
zBRe8!tRLg3l#$jKu2Mjhn2fT2jYggZ)b!Ct7Rb|8kgKcZGH(&*x#D$n!(8t3vpK23
zg6&Ya^>0kHkCF<&J*|cAcJz-rZd&+tY-+)8QUsmoo<nF3{@ftrZ2sjeV+u#jc$@5;
zyj&3zglfVV#07AoObBu-GNi5^s2|9X0s-L8kaGXsOT)uRU;sB2GEQK%ObqHG!lb$m
z^skXgu_g4}<pz#OugNCX5uOAIbY;Azj7j!Hz`O?EUgi5wU76S+i%s=ZCCuR4g6sI|
zn3D$$u=3qCLBh*%37T9R{ixT{R65fRuZv$aTmw1_b~QDZRdRhIu>u1K@ug8j8xF1}
zK#^18d@8AEm8m`i#4rH$X4vIgnHHjIf}U8&Fi&%01dC;8m~7LlGW9$(V_=<BSnB}V
zLCwnODwA)PD({M$l;-?O95LCL1&(~?M$!jkCrEW{1l;xjM7_p8wnEkHe1X&&|HD$Q
zm%2Esy!sNZY|P4Zqa|YPsUFX;*E0~;Qt!<XyJ5Hn3e`(POdGLp&GksiZcMb<7?DrX
zWKA}+1|otg?O%F7CcwrtMv6!XLLd-1e@3y$yEzbWCiz%m+%b*2vwgWpmVlg!V?|oi
zD0r8Erk%)4YXK#!W&q!w<!_i7O&kQ5+yv!M!ZpxsPGwjw;8XL!8Mn?c5usHU?Xs>L
zLpt(CG~6bAQwWi|#%i1cD9Jn5TLahQ1R_+!Bp@c}-y9PgmYLNkUkq}ncbdc6D7VBi
zX(XHm6K3VsL;<D~0M}sjN|7>gn<*2$Q6lPgn-*%J!4>l*`v;4Gq1t3jxZj`w@|75f
z-c$Wm9|+?Ph;rfRk~{469PH+t+!>>y{g5?BhTIjySDzoP$9eL##I8&^U8GW;xacbR
zx{a}JMvHvII)~K(876ldR~2dt+!F*I+FH-+&hPbp$kTe~EA-j>ydN|D_zT=)&|MeF
zH$!A80U=ZK@-1%wI|>ZKLuZwII|wi)gZ$h3qd5Sz48dbwgkjy^iAh4fFrf`Lfjr>P
zNG@LVM)$>o(OxNv`^$IjUzN2@%-s53a0pycINhNEf8Ty*I4m~F58_)mV`j?_jqtds
z9P;5uv1sL^%jCx~HQxJ)Os!g6b>l-}t!dt7$1`YNbcYy^)*Lm9`H6olJBjxkAU94G
zz!lK$7fdz4Ue9ezpK1=n;OH<~4oiTptc9{LO`8fFq5dveWFiB_<8bMXO#_<AatN*t
zp9ZemXp~OxKp_UBtW#OT`F!9@yaQTMU>+b}Yh-EKOAj%nhr7u4(lUD+Ws#V+Rs~ZL
zl_1mGyo5zp8gEl}JruIb>d)k5g^!QnU9!@cvz4VEV-PqHPzOcv_SLaKr<`%p$+9Ni
zP%DfahzqQ}V|t7kM28T3>{DcK$vUUKsmtmVbhVdu%L-X<U5|j%VnZ}cFC7+uiBucY
zGsdZ?Ss<IdGZO)%373O0gq!0R%{eew$M|@Qx5GqZ2oI%gmk*8Z31q-2Y+4kZC|hGv
zj@Q{qfZZY6Y|#Dt43h0>fYd<E0kp%Ob+?!D@(k3&MIgja(g4U^jK`QSKQ->@Zog{0
z?2Ms9G1=i1B6F;iT}J&9EL3Ywx}zN(G>!D7vfI|O33_v*{LEia-vI_hZM(-CXPgfc
z_sqni!uj*moQYOsvB)pt83YO<{DHJ0Dchd`7E8SOtEuk9VeMn2qp%`-qx{<6glCEH
z7S?5r{3eOC48k4q^4pj!HW>qPSP4&cB>7$RjT^)e*b0qot^D4mH!)jT+bDnVLpp~F
zvwsW~paIa}5_vd%i}HhNd#sT^#Vd#k1RnZ`cg(b)g_aDWDMHKTQR`#^Toq07nEj)>
z{mO9@<j-kBEc8TnS}BjmGX`-{;Q`tH5^|@a9-;oM#rZ@$fqu(WX9PX^cU1crwL>fW
z$uz{a9I{cK3R>Z+I!e!{V*?d=l&CsHPV+NqD7=EqF6#Q5@#hf0yw4VqnpP7pvojyr
zbN14}fqmq8V=<UAEvVb}cbokfM4+|E3#MStY?c>|@(5yIFE80E45e?Bmtz8zP1#(f
zyy64W#Y11~;L&;2M_~}YrBMOJfr|Z_Pmi%+wElH71MA=&2e{f+UJt=lio((e@{l*Y
zc?)(hEc(rO2{kJ;!r(0%>)g3>Z<M!#=0tL^^B9Wv58uFp`p7#mTm_?elrEQl+Ki?n
z%7uvn@A@kYmQzLrxcZ*Yh%Q~?PF75du>kMKU?7)arB}%Zac%*Hg|*DAtq;9Lc1Nub
ziA?gZG!WJ*qJ&r>AH^E=m48RGxPhw?ffVwwtpPIjFvS0)26@>Mg0!%=|C{6+bCeC*
z>e$T`m<~0iWSW6-9gvo=m{3$v;b<InpPCk+8Y=%{co12#$214kFs*C@1KB_AZ*lnQ
z%LL^<GtG`tTj(NbN<9JAl~pg@b6O7(u+inC$BmV}>>rSpSC5yyr#a9z63*<r>@&?#
z!O^g8Wnb?QcXJ`&8h8{awBIyE3Za;x(jQV%_Mc`*9z%v$SHkLe>3uVLiXY%@n5l$_
z8YnLZPIKsmmpIcAeNcL7HCTq=_!c?XskInv$#j2ST}9QdQx1s<l#MHwLv1{~=T{4q
z9p>YKfrYR(HL&nm%F>m?Q&+4)&Lew=9ATt}-CYO43gA>e>#ew-4>&p_QW%D#3WOXv
z&5<}>ub5Y_M_xW>#H&D55?1Rd`xiw3g2|8eZY>toK+a>PDJIu@z(6@Reg=Os!ecd5
z^W$u3P$(9IYLnyb?ULrp<OCmGm=6plC25FVvZCbYV-?4u^qqVm{<Z6FxU)}8WMu*+
zz$LIRb_KMulY*@MdiR%;jjYxFA34Pufk0?#Xp>XZqO%b<4OG;bSMoHUj#=C4kC)S9
zieMuaX4?qDpOLt?O95u$0m`wMNFm(oOdm4EH??J!VKUXTR}6VP?p{30{?UY+DQ3#q
z-kaquLXW|*eU6XX9L==Anb#QAub0TV{td87s`o74BIm^_LQUwB-qBwPvP%2-utKI#
z2ZKC?7qo9syeS9BpkI0qwLQp1^p-;Uu9N=h7j^bw)mbkD`$z^%bBq<P?z);*D%Ati
z=y<mkI)%g;6wF^<TbHe`lEG<RU;?2jT)2ZjiM;Zb%f&Lpe<xzfXmMJKdca~p)H=?$
zJ{XJKh|_2e0hkM{k1`Y!scQybU+9A-VxytGn6~MO7lkY-(~N+NeZahsNqGz!*d_kV
za%}Ay^-H5cs|-%r?8{r?Z+BpDdSd{%bZE#9VEbBnwKRGiT1VA-Sdat)4K)j8cw#ZL
zBDT>P50@d;b*uQnnk^&JSQ&3dymE((jD{<5^F+0bO0=UpmC{v3qB{0?as235G7t$$
z4WmOYvn_iW{9Bj%Bnt*4R2y{yjtQPB%gm6nm@(BzwJ^J6Y>W*64dQL^%VrtpgW*M-
z3>~C!3=ELzxa0kIus_;?xKRj>`IDG6e8X1+m6$pR0X)HnN2%FK=2GV~ic?Ct&7uF`
zJ~A<8SXq~=1Wy8bm}IMg3)o<G)gaDxavH3bltY}w3c1pn7&4@fT;=2X<Q<=@(Hoc*
zC&SggT6S7+uXt20e&?a7rh{A)M<b4AlurSZUhCb`24c_y0xRUY*o6oJ#pN+U=CPy7
zWlEZEal{$jHdA8(Kqd|fMutp_Px}t&D;XCEi+2L#mkNKKP$vO50Ps>7qXTNY68~E$
zRnY)$9FuHNNL5)@TwN~J(_FJ>C`%Kn#%N{Ini%EI`KsB|G&Z?c8`D)N*koy)zYP+B
zg@HU0N7MMr=}vUxS#$MJ`cWA1Je`py!JCB{La<coqX$Nns<>Js4Kbdif6NCDgDIua
zwhHVK*d%Sz6x@}sg<77MX6K>)edUH=OwBE45gkHHjEPe(ujGtW*hA3X^Zs&iNU1s8
zZDK7(X+;0n4aZf|nutqh41_y^o@NHyHUmZkzd%Er6~}jV4goRDVgV1$wn?Dgv6?Pz
z)6|8z>o%ASbD~)k6VgC^c(HD@-%ANQ-egSCoXgZSWMpo12=603k$B<e;0Y#6HMU@*
zd@-a8z%n=jZV9o5Xy`lxE^l?3_30<K*<T0s>n*p(67c*0;h6eEyMDQRDTYW-A=rf1
z27EboTR{m`mX7GOuh<wyqTCD#@YSFSu2Y=}s7Ol<+>zKym0Cy!tx|;^@ATHq0mWPl
zk_Pu(gI!pf@~+^5;jBH0TVR3)6)2&je%G)2n7Ik)!a_ZC`G!BE3}US(#*p2eda|^e
zH4NQ^vAh7^z<D6|_!td5HU_s+L-#rrVRPLV!vWlkoF)0Df8-{Tx`wnGwS22E1T|n7
z*!AroykaF*EoBi-))jKU|HAlCYA04xkmUc4iwX+qE%Jc3%7Ro(Jwa6@dC*zTVg}^9
z{z^E$m~#8QkR6jtCdl_~I%i&Yz5F0Gr)iW}_096bka#FFh3j0LZIB1|<7qudt2_Pw
zl3^+BvGhP+2?E0jn3Z|ng;_EtW#$K;vMM9eQDnh1b>^OVnk=-xU2?H3vcJ_-$zmIQ
z6U%2vM;h`n2umg_JtUp}j1hSOrZ4g3rWzb~I?*`wVnwplzrp!Qd{u@0vh<CG!FZh;
zgpV0SwA@Hj4yPL+-iQ8Y?biXI?10Z|g{-jm6jI2`$`~=sWXdV5t*lDDcd=d!j;O`5
z+Iy!~Mod5h;?xJGw39W#dqXoV4Zv*Hn(R1x%C(ouIwRKH3uS$3g`w}pm%5Q@)C2t7
z;G^cIKtLCuLToiXp*H%L)Mj~LZiQy1DkfB!?HJ#tppI@|tJMWPS`8#*AIj$RJFG1g
zD9oy3bhrBOWJ_$_Oa&klim$FTR(j7kZ;A!t(}wkJwOL(snrw@mqrnq~ky;wZ_7H(N
zz^QrJ5if$wR=S<`43iFi;v}a`ASekiMt)lK%!<yflAY1u<@H%0bGu@t+|2-OSGs-N
zX4shyTb#zRJ7$`PP*exx&ukm@I`?epv1O~`Z81NOzFHeu`sEki7k4-5!E^PO>Dg9=
zEEvM^tLPuPpK=OBnb$>9?)tTVLx&T9UByj*^9gP(F5UQBf173pD=n_ZTK+BuR$2=!
zLVh0;g2SVw4zZW=hj@apEG--_f3$hkqRNLnoSHQFm4p7%Cv@nDor74!L<?c~kAxC~
z%VC)5VWD7;8tX$;Sh&h#E)<DRZ;?Oy)X0*o6vB8<!=WCJY1C(M|GE4neYOS2;}cFr
zb0AnY0+<8Rn$%e*fA!z#5wn`ZL{*kRe=@`CNt^JPCP<~HV!{X;xZQT0JnhdA`H4M7
zwp%XGc$?U9H1sfh$}4(q$kQmH7~iNWNSVJ`o;C4@Ye;~FK4&v(0F`;#J|7dz14L_-
zzlUH2b7|!73pPev-rOiJrna>#CvF<vATJq3SSSw5{IV00_qq_I;GtkCJ+Fl9Da$nC
zLMwUI=V5p?P<R>|MfGG}OY~TxcH<Hpu&>)5$ino0!+)k&Wwh2jAQf+>&Y710Jsskd
zoUvNoiq1#SKF!P9ahi=9ccuIzW(8E55d;Xqkaz3@fH?>T!D#=PnpUbq!gTn&-VH4a
z6>M2V5hQ)jNHQ9(nVOuu@6G$cOvF$=@Mp%urX_u7&$`<Y5eY!OGTT%w|B7d|s3(P+
z$BAWs<lpil0E$<-OaAS{(+#hT0stR6&~o`Wv5S|e1#_9tne2q1J^x*gFa44gLDeD!
zA#km<k$eEaA;}mzh{y$vl)zmY@~Mmm>`lmWpUEDXo}+NWvmtP#A^eDM2lt!B@acFT
zAp8O-XP?Qac3Z}H@I5n*0pfx)G__Yo!NN0p^*ZAm+1uVzAZZ`_JCOB8**D`r;TXb(
z`}qh-2IxSLLknepACYN9L~W3x%}wtkI>5iN6OJ0yQaLbV2q|MVqIx2ebT`XE8O6#*
zPnakNTi5V7lR*y2I8p*+yow^OQV#W2BBZ!b<@vCTW2(3xg)*v54!2P-d|4q)-avMQ
zk1TE&WxasUW*qQ?W`s&bxLyUD?nr-;CKIO(;Ub@lXMlWJwQW1XK94er)~onEItGlJ
zfjv3(82`q|g9+pCI5yKWf-2ix|6gRM$#Jp#4Xu?;`HUQ&ao8B6gf&23rZ5A$xRXzF
zPEN@5@Pd&k;2E&wdG>!(<numU=x;Cv@aN8j@&)6K;TiDLiNPEBR&2_X>|d~{=s-T%
z-s^53-6W^jnl{a<ms7n-?j6Re<8CQAEwOYV3ZeL@KHa~u8X7zTm&h5u^d<lmcvQjC
zDQBi0n8i(-X^ZrVfuh)RUe2;r!lG2><!sv%5Zqeu;hcE0&k)?hdv3G^nL;C*SuE$p
z8>}TMy=`iZEpw!glR6-c0@P<Vy6w-tiSr4TX&j$6j9XsPl+^fv1Y{^VQ~dj<Zx{oM
zx)z!&!NBqXvE0zRo2v6N(BJHUKEd$%!l6CLiCk2?AM9WeTlDY__Nf9`Rz$H23p2zf
zH)OyNIX@@?6Q0K*CEo?M#C-?#?kyJ@RWu;JQ7-bg>9Jo6)fs=v#Tkb_Q2UpaF^t}q
zq^U7MLn(}5^E&xU{TECEgrY2sT3IDA78I1z3MBA@4NXF*P(ch|CbOW_Thj`=AS)}-
zs|b~08MSMAT_?khFD9UdMWc+ce{{DGEg2&t6LA<UMHpj|j7p<moVbF_GTM77Qr-0Y
zKullaaf)p_MQBgcs^1QP{PGaZ%}vd%^)kj^H|^B|h%M-aI>>e@OU?2AR@($-MJcbz
z*fcoa2Sb^Q>WQ{Q<0@DXPsS(0<GLzv0z(a9WJ8``k#VQZDBO?1({_TtYDp|SQCGNV
zFZW+)9MMsgvaDrdG!KEo*s6K-;bZ4+kx6N#3fJ*+QZ&qHOixZ<!l8m^Owh8ebEU1+
z<cwSu(^It!c|-W_YGc;HgNMj9wkzPfo(EKCi(KnXsxZJXwyTqeu8`|uuG6!#vh~0@
zuTQVCpn@uVTrD&*C0>Pws{x}^V<H7FJ19+M+KicLags2qW)?!_=(5bDk;7jKf^P)=
zT;UCM;OJb5;cZhBi#eYfGAxz;4fHMO7QEnkjZ_(XwxX;?i)8%;vK4ENtbit09gSjX
zyc9H68P+9O6(_XR#6WNeshC74Rc$)45ZQ-%%uvVcjFhENA5a1!C(~0`+>gNMc6P`N
zf2R68I$^=%dhgyyZt<$%4<-B-c&-}K@2Rwu4xIkkm7qzZ|8A_EOk}PJ)_kor`8pZR
z$l&BMCMY+@(LWLh0B#zslpA7iLF+@jcrMf9>F0$|;bN&su|Q_t49S~M-7MG%FM!g|
zn6bqs-Q8{%R=07Zw8pcU*&44jCo|JT3Z9zC6k+(<EbjrrfVc<1E*2+evc+sCtJ{1K
zS<9p?JtHh=0&Gs4w3AxoMsH0S2L#!OQQLTLij$%oP%WYh=LS1B*IBiun|+}C$XrHt
z;%OEa80w4OH5?lrezHPtu|0?AZn-sjR<EDVz<a*UUMeXsmD~L#vM9NK8_OXmbp0j&
zg}??wR85LA>g>xtE`o5_9SW)yFVR>0OZC^H&Ez_iuLfbpx5ynqIGAh9G-T1#JA))u
zRv_pud#Ae{eqNONLc#;E%s2)H^zI{HPhHUkt`R`s3EH9Gh~{<u==)hBcSrMOKt-Xz
z-V^UYjiBpd5e|`iO(w(Yz|DSgUn0*2XvJo8a!+^4H*FZ>C(^I~t<XRkQ5QqLZL{p%
z8!=P&JFQ3`s|Nz47DH0TLb%X(VkeYC9M#JM_UQyf)yRWsbgTu(SnYN4T_2C$h!*}@
zfHFJedm$-!vr7*34zu$8cv*wG^i~JR2cil;NKMlbYuSW9^k*GVW_Y_OMbeLA=X%sS
z3NG4An<^2*-X`radLSH#YMy6&Ro6IO=6m0|A<!C&-_gwMk_EATBg-esLVr!w7YvTK
zCs_=}B5N8&h4dnY#py-Ho-x)zHyaL&4r{Lo_dDT&=~&_M^fu&Vi4B)03|x&%)6OS0
z0fknTZmPYuBbUYb0Nt$x{&pxtd08G!A}XgE=fy+TUuCudxsfXhRA;;_W0+U^NV#B;
z$2i)kpP<r{tTOTpLZQRePHccL>O@3Rr>u!ZfzaHF5rIu)ZM>%mT8p6dFOhX=`@!jj
z?S%^?@F!0>o5gznhM8Hk*tuml#6p#~Ah}sK7FLI@)SdI7*czO=$r$j=xu={ho6}5I
zYMKH4lK{tXvTunsEE!fRT`@+e6m*D^!M0j6vzctTEtm*MBlH}QaC`cKVcgV-SILgF
zQK($%s6Vk$eiE}Lw{zA+-4F^(9G1D_KlQ<ym)FWw^5EJmJ7ci*K<0OS;x*fq226GP
zpfrdQG+}Lp-DyyaI;K&&92CHRF+i5T_*w8qwxNOtS&z3)tr=<XU@-bo5Q!^4_h0Pv
zwZqMVGky^)1eDHI@L&2+%zmdPI8T0Mrvow#S<ek*{dEuon}%WgJUSud2?A)q!Ee%7
zSiHtERTVmu-zM@~pec2!-x-6z2G8Xo8Gj!mP&zY?VY~dn8#J-5<~6$FCB%OW?i$vH
z;$`x%zi3&7JRg`YNzCR?#uokh!ajK<mL(6j<)c9aLQn8!L(V@I>jAdn_~%2f`LoZN
z3(E`*mgz56p&qv#b|xxT{v|~6c!tM6krpy3#1w_V>!_jtwT!oSJpfbo$rw!~P)e(l
z{!|>!mt`BZPWsb9A@!8tl4Jl&&-gP_s|;!XTcR4uIFN6_v6gS}tdA*@go)tWH4y3_
zkmsy5l;;5KfqX+Ae$U4+s#>ia`QL*g=5*$*L*xSJ0~QSYf)A-fL2pX``SM~6sVHI$
zaURg&UW%2|`vmFxrM^!A@^Y+J-+_bV6@QIKJ9Q)7i>e{L>hhLl!Pd)biAF>aai2WA
z+potAaDwDnZtjf`51FdzraZrCyJ~V>cGl3g(N*$R?6M3h^5KiOU0tMAic~!$=`rDw
z(n;CT1d2V8f24(H1R3|@L)7Pe5w^=a(KaMLGLz+>X*BgS;8DjZ`mVh(3`s7Cz{<&c
zg(C%$@co#gf&)+>fV_SX>;otjp*yq(K8$YiD0PS_{L30ah!L{_J__y|QHRi#e+PvS
z%EH3;^09IC;K9A+Kk+8+<H*T>({k~W=!gpHf`eEp47ao;)cf}2Qx%5r*<HJ^h(J8k
zW%fwF@Lpic*0N>)w6%kpa}2E`LYtS*R0Q4$Nsmov5nBa2X3q-8H0w~brbYJhp(uuc
zRl!x?vUi0-n3Jrk%|6jRY%h~xa2;Ze*2=!tR1@Nf^6-4^SE2A|cl*TQWwL*T!l-?E
z_vs@C_>hQRIVOQ=Ffzvk)PWTa*_JT}AR`A=xItZ=Mb<dv-N8OcN?huZJbMpG&rHol
zy>U4-HB?!efuHLz@0g;9*7`PuOghz7MzMOh4H6J}n;hZO<iM;X!FKe7;@|RF@5xF?
z*lo8*ddo$=K<uEx1>v7dQ#M_OML+mEgLdh-JIdMwt>BAlt*J$ty&UcR*s(`1=4Ls@
zpIKE}f-#HbSbqks3~H7kUdILP`wkc+$NN;6i^33<MM^Zvi4$V+hgYExj(on*Bj`gz
zbiI5bT8G@A=EAziC&n{)lu4CkhzB~ULSdlGhRMklia^%^PCF%DR6$C2>>@eUpXo8>
z^6E4vPsh%s08_$QRB$$*ZVQ8il4dz0+EfJ882`+)ux48UF*8CaM{ST^g>OJ2gLuyJ
zZ+ONt%#z`aiY4hcOXTdt0A?!k0%yvt9db_m;^+zMg4#IFt#H_UJaY~RkDeD#5dH#9
z0HAsAV9Ih_CYVEXq><~)bI*{zX-J+@uTU428Af7mmVUMoD{yQhT(!SViUGZ}(-D%}
zCIgI6bLXCNjtulklYcNksEr$CPz;?;9u=_EUC9cAjblns<+o9WSVxF_9yfNPoSzyZ
zSY}7?1^&!cVyzT+p>>XWkTr6VcZ8En51qnAhZp-K?F`@zA1r51Bu|f4xx~LL4nDOZ
zUK(FQZ{o(W3w*c3qX3Z!rK_aET@MVRK(xIc5krl<Wy2>*sgD_70tcc_xL$_E4%nrI
zv4`8l^|q~g8IeAK4GG?7UEGnje7Ng9ua{Mf^5HH96<I=r4g&6oB{JH2M!65UEWI?J
zA!j(LwemGi_sgyS#%wF1tAJbj6dldbR6^({yKlpES=y;AaY#}81okBRq>4M%`W!xP
zgp8|je@{yeW*PL6Dj6ROgqwk{2%(xq21q^JyAzB7n3-^$l>5xp6vE6AfiDyN8IwwE
z*DsMt(FyP5v@&p$D;%c=Mc3kgoAJNZa;0@$G8>s(SH(Kv8X8urK&8dJTpcWjl8QBQ
zO^h^$ki|q3yV)Rt($2Zp#fEBZnllG+x7Rzdd4$0?rWY*p<Xk0F(lVs$GO?vKnlY$>
zV#csU8)ceNqZO7yon(CG3b5h}!q*k?1w?gyQ_~EojA4RJ2qHo><4@#D*fx`@#7xMY
zBBMZ&&H8|>%z>dcG_cl7wGn{jt)wP|c{6fgYSZk@!wn~x)1fX#G$xBuAh0>68*k;a
z22jl~3TyeItknB(?R+y$*W7bf!K%TBn?OgPD2?7X2g+y+diHH=HpLM8_U|Lj#)Q3*
z<a0xesVvJIhH4}iFODcJm3-nT3SSC@E@`Rg>1*!1|CbIuX-y-xENupQ%?#d{Sc{<N
zHkoBZ%g@NjY=3<Lf&`W#*J=SK*k&XG!><I|GsmY}3~nmGC^v@G7?aIFq~GLCEy7uj
z<Aut*%#E4Dj;7mPZnmyRPq;$9=<id(XB_`#@Y5}EPyoD{nPm~YThjnxE@FWAQ*N_o
zT9^Ro7mOp7+kKes#n7Xc2E<DFQtYCkljO^(Px1}V@zwGbf5x}zsE1`BU-f6qkYOSi
zO0dWssRt@-){p6;cUljV%7+h=yV942ZcrY5&A&7cGg4P6xUoTycAU106|SS#Os^3E
z{b-$i!+(#;smyzzA<09SNgsEoG4WC_YJ!{aJt79>9vcT}#7KNp+5#=_tuQw(O$w?&
zW_YWEWM`KiYAm{UUohvGOr?C&r$bvD=?^R&E8mJu1J<WMh8GmFNxftNs^Pao9K!m@
zs&y{+`zP2(;Bj-*SUoOv>d1GT<rrV`kUZd>YD9{z$U1p2(RDi!8+uFRyU`JosZqud
z*A{#)coKQeO*QiU_zE$|ln@9)kso+p;3Y8o4=Z|#?sl=;CABEkkJ6$ofB@A{Pv7Q0
z4yMil@x>h|(ry$hA2oW!M44yIkikhN^R27y_5r<t9xO<Ur9skWgwqC=h1S4z*P~jU
zEJ_`BK-nU66W6&dj``wfZvfhd1Ea%wDAO2O=}arb>T5h<nfs<!t}>KDmc*%q0BELX
zAjS&fV5#+vn|EMLE(=+~YkW}eZ+T(>)^z7yT`Ldy2(97_cREx@tjJnf5#+B%l??<$
z`p1wfIPv-CN?8>hAj}uR%d$GQ{CEP~y!2*`&54#Qf<@LEJ#)Y`WL>akN$GG|A8QLf
zMi?0MyLqy~pAi_`2m&xkZ=-!QRLehaiWj;&!SK|2x5#Fn4uZF!WKsY#YLI~WmJl7o
z=SWvf4H>9aS=k!X@SD-LMI$gTh+smk!XBT8owz+F2(%S7O0f@j_#mj0Vu3Vj?<Y2C
zR=9yQ`)N#ccw1I>`s>I^g+`C@L2&HyXFRyn<qF?Qw>8lU&mub6o!$W5bwFM6vouIX
zH7tP$V)XSE2)v%ckg!n{rI4Td+Z-mt*2s?7GFawcSWjRFY-ssqOb}WF3y<gISE(O{
z@mN|{2d4Ru{5tl|NL(de0}s+~>_8X<py0Pg?KV~>|DF9Ek@|W0y^oE84^}qXStox8
z0-b*P+44v4&Jx@xibxX5@rMi3!YQqKoczf*NwHIqXE{K~-SC^qBZ+WpXzOVyo=5!~
z2X26FTBQbpIIL0in9ZUL2J4@LPY|^Z=;d*n2=Pz(OOk@-Z>Q(Lh->WjMDz?OmT5IC
zK=Ri#11pTn6Ya_L%+C1LFesm@(BPYHL?%HnFd++h4wdq>4Qwzj%zh@QL`w(sLA40y
zy}#LVcDGj}OG=(iqbvl{V3aTx3@3js(XSY|D8P5mr^V+TBU}w?HGq}n@3HtZv9(z{
z<AvxL5dwL6(LYoRSwE!1gA;x!IBX0GR49`Ca;$&9e*NSXYlg+@!8)(TPyKN%*=zBS
z;S(;G*W(`uc$YU~m})o>_+&vWCbYk4D~Q-64di>vsZBO!-a6|pzwIC6iLN1UDAP(R
z@(-J1CRc;xV&*Zt6M8;S*@jlTIZXcP-&?*R@>&qx%c~x?%eywlp|D0~$$N<+tj)Y+
z=Nd|N9A(?({TMyNfVsON%@i+}4{XK9O)8fUtu46efNA~9r-7&yRTYpHDIWzvN`TD$
z+h3p~fF}-7m5&oI6r2eb4^G>xOa2qg&Kn0%Ot7NIo#MRagEwyxikVMq55xl+fkR?7
zo-c$MB9(5Lq5dH-N`j=Hs#MIRHxhsLu)kh+(^UC%rNar#9#;?fXDUsPF>J~9I}<zz
zmZnN}vS+1Xel)&4UIBw4E=KXTMfQqzt5FO{kD9$J4S}>4O{d?d(hx}OMv$0p7uE@f
z{v`YQ+w=p{Wx@kvzrx$<Let8-nxDOYrJLo@Y^!kZfPxnXR5~~~61OKF=-)<C+Cpfa
zhQ}XN=m`W>TSX4`XCN|qw=?7rdj{ka@rv$pIn>|t;5H&2a#;G7r&I^-5#At&d+*dC
zgP}O5RdPfuR9yqIIG}0ahLw0!I&S>c6Go35HBpXqYFo{1Xt&S_EbfRp^lJH>w~rF@
zpdhPj5-^1flA|gW6m81S$jZ@`4szzQ)ve6LKgKAeH<`$>l?sWD031D*1#^$HFT2}M
znIgvrDMvL`$qDviHNYk$o#*BAK4BgT%nDq9ljsXR;9`)WxDeloK_oC141H!k%SmZ@
zH_?@+t=Yn#Z2UT+X||k_*3lyH3U~IMS{MdoB7HgY<g}Q>sG)Lt8WOh*_aV|wk$17s
zx6Bzft?u>-rd6M5{|2<Me%C#Sx#|^b2nQ~Lt1N6z-6v<og5km-mOI25eYST-uQ|QF
zJgF6Q$G}(v{u~=D<D9E;njn^;P0sZRFsz081M6d)7u%?e;Rh&sF1^z*wA7oMRh`&!
zed1kQI;oCB>FX`hy+qw>E%b?gK_*B#MvpMcqJJU{FP`&|Luf86cMK<#3@EG&O&gsJ
z0-ymv4NRk9kT5Ya6mNzDZMzKeUo6~;MoHX$GB`${Zf%OfP8nh=JZMmVIlt1x9F`vI
zWG5UoPH21p@(6gKhy}=qiXU=;H)Wawe#c(BFnBfBS|t}nUkY<6`XLur_UuLTzkS4n
z;lt&Uv{HI~6~?LA%~WWh5v`C*eI6aV3-_m*@G41dFM-&DxP&unsC7OmpFu_2(lk2c
zB)b$DW-nZZa17)_2lc3=f~8o35>tg2X+-MPY6=x#kdZOHe703aMFYU)S&D_X1M8j4
zGP+WIE){5x8t7&J(;mSU1AH%!SwQ%bQ#XN=#`xO|oMCYca+*OlWkx$L0-PUX{XG-u
z_!At19S;q9oVUti8_;fX1aFn`{)`6;Q{j<rDp%OE?)Db=BM_5ZjljwYv7IVWXAbHX
zL}IyrY8|n{N2g5mXAl(BfpB~;kV!!t7>hF5Ux+Km*CJT#$`FCD@UYDR6vhBxkReeI
z%Lbe<wM+z?UdnlLjZc+l1|8}&&P{?}8{4j=g6XRW__)sBRb&Gr;|XcN^Ytc=##RlV
zEmPtxL%OIPW3j0JOie7wI3Man$k4-S!5~Z8VbfwPpaTP?Ec0$T&Vh_=qp{=jQsI+`
zV*=3tQt2NO|1;&0Pz#+yRYkv03&PDl{!2D>whl6cM?AY+h47T>pydeE-fTq;s~Y1u
zLi&?{>DSs|Mxl^Soz1%y1;J%{P@EbZ;ufe?&WJAoh60VOPoqJFCoI4`5SvHw7HP1C
z#zP^J##rnT$mwqirpae#BeRB<ce4p_T)0?=8`iUOL(GKG1m>^}I+JNjIse6$mCDXS
z5S+JU<o%a0P3Whw9o1q(M)k8g79MR)B2-gAaq3&ieTrDb?aIuUHG+m%%7+0Pvy4EM
zaEr~hzqX(rU7JlS2W7lo=J+ZqV#{3zy5AV{fUH8?iHFLff{NVaZz~joxuK$;%nib5
z7#K{KtlXUVi*}g>f|#~vHvJd98E${#I!nP>x5Nr%XIG-6$E`jFY)F_c1a(0db@JS1
zD}+O>QEvB!)LcYIC<H2B3h7(QI~=|oT|qHo3-p(K#ah^PKj{C})B+t4+;B_fjx>B;
z-(&;lCZM8r3r?#$gFb$^qwE2w#MIEc(jwWFTHA?|!;!+OP%F^$*Zg-AGvRl#Dhyi?
z+WGb1c&KXdWw*&UZ0`^AKph}FcgOT_vrDzylgLE<mW+eQGX5#|20f9(g8Zoae4bWt
zg(uHC`KEvDdk%BA!t0&#t)L2KO)n3@;BSWt42`0`mGBbw)%`Kx%d*ulKilLx!40Tm
z;vtm}cw@AP>Ho){@}NKKfRF}y%y4q~u0I1%!hzuM?vn3C`@C<BvitkdK1|N$I$n<_
zKM47V%Qt~vz!d$V(WtJa7Kw@Sqcj=29F{j}{5a8#v0ufPdbEd@256d~)ja!J6UHEe
zbYIO+BSmlvM!Eu^$^zR}xK#kK!9s6-5u#cs%yFRLwXJ}w7o}h9GKy58L>7CaECG#(
z5#{xcH1wsk$l!ZkAf5h9FN{~xt(7Il(HRufl%@9f?)K~E$})eK{Kk9X>HJ&nJXoAo
z(1AK)NJL@ujXdPvP>*DO1T-#LQR)6Lt$PV1O#L7$qo*s&WtGn>Vz0JB?X?K1)v-L_
z5Q}G7lV-Js7Uu3-;ip)edR+#*i}8VrWL>OvJx@~9c%LEbjjKmM)^O=J_*c{x=(Jb|
z`fMz;4jshy<)*aUgwD;WreL;g4hFMwnt9n`J(T$+>|HT)oV<-FHpKPM%huEfE4Ofe
zER$_cSJis3ev}HuLT|e<CGrNC8nvS^BCIS`QO2R5-h%ujMnp6f#<l!3k!C4&H_?Xm
zveQ3TL>JE#-G%U8E4%#dkOi9|qPt@WV5A{H1}DRAe~XMk5rwKj^0PSZu7up|iES`E
zqp@*6_btVO<XQQJ{R<qm0%!X#<J}hIPa>c4SKc^tMMwjdh52=?<*+(jWr&!+Hu;S;
zg8D<OlKR_9jlGy&A;0s_)t83PVerZC{TZ+7q%hqC_~Q@po*t(tpp2O0KSrBlGPCmX
zu)Tr&9Y*-EI!Uek$!2*q@Z=hKB&~v7EU?9eXoox+4MEAz%46xhg2U3|1MOVRW$Wbe
zSh$I%YRX>{M_2)Q=)Nio;N*erxmBJ>i%3%wCouM)?fbvR)J$jQpYo)?t?^9UTQI>D
z_mZcg9g+oH1V^{>(n5JU$Ofkpy&$l-<Qad)-9iT_wUEE1_ZC7<7*P0Zn)7yr8JL&C
z^RZK&^Lat8sUfU39X%8}yS4Isup8X8C=~=3%ioPk^&t5R{;{*X)+Go`ipTehF+AWh
zj90*9<R#ma!2N(<<m6?4#q30$>7DXQn!yGD6qLAK@@h1Na+id}7?A&(b1qb8u*>T{
zn1G%&p%v@|q#U$bEN=wY68qPS)a1=H{-Pu<2<Ca{U;~!vt+Y(knu{0X?KqKh$g06r
zCH|s;h#lPvaMH3cSC_mK8?j%X0rJmaL_j8u@@`BM2l!?Pe~2=9&!@!DsLk?zno=R>
z$HFr~E@0gs1he3Pp@=3TW=!{?w+%J{BWgO0{401Gv7Z%nz(YTZNiaSIk%QivWj8_%
z!v=y(fhq8DBJ|qbx66O*sOU4K|3LX~EDLN(${rT7m8u@QlBCMujZal6w1$i+lv~)N
z%78u7@@PRWmrwgMD}W6LFA#yx_%kY*#GY}}rR-UyW@@=tz1b^Xf$GY%lhsi9_qM*e
z+xre0Ec;X`gr(J9HU4$qDmUhl4BB7zv*sXCpaSJ(|EdT~WV#T;99bnHoTMw{fGP(+
zVP>K>CE`X7tWq&|>5Xzw8p;}I2PT3K_Gc{ms8Me4E*=ukATwC7i8DRapW)ml?t&gS
z%3=NtK|DAM;p>IeJv_)-iXvXXc8;i0u&#78@)<vCQyI-Xc;tZ`>7QHBheecpE<Uej
zrS2*@Drmt(c{#euu+_+Sqg_H}cfK4`<?2FsGk`nh<=AMgxfQABc{$EVF)1g<S2?;0
zL$3w(h!m|l!D*{;nR<>>DKrd$&iBs;ZO6k^445!4Uq}myTIDC&REG^O9U~{js#UjU
z(ecSu3RDk380RTIhYkQF%#~HoI=xA9YBXPisk8o#oR)~boY@jM)|SZW)@5TWse6XM
zpqLlN%0$XDs~jiK%xr>{$e6cY@u&{BGs5qUVBNE9u&DmJK+ZNYWf0A%%H}zV4$C0d
zDJtQtz`92g-?_FZnHpAilk<E+rg<zzz~wx!If`9K@5CP)RSjnxV;_G;V8?4-N?&Ua
z<6i~|*e?hK5l3$Tutl`TaHIa|+a+qm@m}r$@eI<Hbq`m8R|gukhxG0%gMy~Gm=gEo
zfTIT6j}=Je8R9)rcC&Z@gN5-Yu&VQ&0-O_L=KjY3mJ8xb<QDP_vK$DWR2TY}Y8m27
zrq>a;rLPEALCQtxOUpc7pn5vZr;D8mP&ELeU*hv(+srIj0W0v*G%DVSNL0$wF3F`x
zkdYGqMsuB@<=zgO4fSuBK%!v*s+D6T_=YfhsgH-Zaq!Mh=0%c<t7Vum+C*Gr4_MDh
zI%tTD@GqzVv-k=BR82xg`V4q1JS3y44D?*oUAW`PsA4rK*gqXG2-=j<zDyRpgg@o7
zShxx}ahhgf;V$=nsgSLQR6@IcjEr$!Mgco1a~jxT&;GW^*w~DDR8X5G<D4X%ETLZ_
z26`T0Fm&&a50ZSMxFMrGuPA(pWykl}5(Z<~c{U*iU6Vl(2jUXtw)0e#$KC1~M7mEj
zA#};PGAX176G|q>AWPs=&B&Fp{u)S4KHM%>#WSoWgLil;U2SvEWUDfAjSpP)6Q7b{
zVb_KLXwKx2TmzI!uJf0%wZQxYxGC0Y4qWfQpi?lId1h>WQ)~d;?e2nueIiqRP&~aE
zk_r!}O!F4(sARe>>#_5fjK9rt{F)yO#;>pg5EvUP3c-A?G-ew-u%A>#OS^BQmq`s(
z1u-NWv_|}--ELQ%$g@U6i&!VW##hV^N~mancA)Y=tXaLWTK}Eu7*$gyfEIbAq%LNR
zGeV6?FyeG)bKKtwg>FWoBZUYR2F3tTsj83DXCf=z<)y*LM`s8WFU@mn{2z{j2tgYK
z-*}U;NT#t3CRB3_2F@-pmoDQCsewY$4@301GYRtM{CCycxy(>PC{ua;D9R#A-Uo%y
zc$KLH##*AEus|{R)|iamtCW|S#&@GDU`WmKHaL0K=tHdrnf-}pi!tR*(&o=>%ka;Y
zGRKGs^#(|)+*s9f0{j>lC$B(Uhz&-_k(K%w<YqCU%jKqM8x}7lAc(z4=K3>!$r5AS
z8#fyt^+s9EFUIZ~Q&-WJky~OmV?am572IlMo^su_sF-$}O^qkF+-`s8IVxWYJ}q~}
z{L8U_dE3EPeCTF|;~3NgI*Xhrtmjv)y-Ferclgqn2bEQAxHL3sP-_F?{7xG)?xRvb
z`L19z=viDnD-ea;R{5IGoq9cC`7RhRUytok!U~kgH<NEfcXTVbbbuR}?240OjofWC
z$-`dP9H4u=)x`*AjKGh3(-?4>JO)eL$l`(Sv$p|S!RgB|F+hpm3?-`)0w-5j-CBd)
z^Q|-t*1V%Nz}S+7^j~5yi{;xf@O+!cQ{0~pMy-12-(wd+^H=c(VtBC*Nbd*I?!+-D
z5Bh*jz6UV67Ru{)gCmI7SIYNdteG6Fj$9^JhY5T?#sYPffh1Y^K`;&qpUeQr{X_e#
zhRT@=D3q9%9|d1Dx8&r<i80qfow0<qc3;`1dO#ahGS3Ie8~I{oH6&i<r<G+5dB(YH
zlm$)zRebdskv=6L1kwgsm<U`ba^xb7dW<YeB+(H=H(`_((@Ph}<Z7~rSj$UCntlUl
zLU_M3o}e~6UV`{v;v=+UbfFqzORIX0!!G{66ob^qvMgwHIS|j~Ap|N~ftNhwO)_tY
zOg0~C<BF=D-V_WLKU0*WjwcKnF=vLYa<TkgY2AzmY+EU-f_tGk(RM0;TdSi>*m0~7
z2jpGW*mh^&07@xAi(XqqI0mWVn8kWBG-aI$8sLSpKJ`I&8ud21BM`VH8xrF(^3^aq
zyB3^LE{kNN4OQ>mkWH}^<1;AWk(15AtyQfJC@Zxk))-QNhc{)Op)__-clqybBrD`m
zT*zf>Oa)sG+9UQq62jpA-yqxK7pg*d)}<};!hr2*W|rs9un#UyhwQMyLXvXudHIR=
zUK{~lTn7B7Hq%DlsU<tpRJ2(Dn-D3erN^rs5VtJ5tZ|h}EI8O5Qg{^LG-fmIHsbTj
z=%1NtHT6vCF*ytCe~|n<SkyB@h-UsG7+FV*vLo_K<3SYRq1^wK4H3o1>YC-(sR0yP
zb!qt<>i}s)mGWC#9Nw7;(J8<4*R61kW-Q9@eU^~#YO`X=K|B~9m@5Ri{K0x3hRZyW
z$@@p^eGax<Rvu2hQ?+1-C@Ll<q7|1g)fyI%L+j;{|5>X7&i7~<NO4>Qu7HFj;d%U+
z_r}{87{ke6#Xoy%7HCACpp>G-c-&||DpMtYu`$CliMuYJNUgA{FEPAT^4CNbl;-CW
zLs&l<b5joFg0GaP(tE6~LdmvTp7v+zbw_(FxBrYk!w~}T2v8I3w!g(QEjhsJtvcmd
zAARwGv5=aDJm+r{Ke0I3CGvcTqAO`7|2^7+#>AA#1z4&Vj3IYDfYCjGytWrpD;Dww
zG(=u<wV}SLMPByb^?1WL5#lAU1VvE_YaZlH;R^Iu(`cDF7k*+IAFuf@=0)KQD8sn_
zUiUS{Gxevb#pwD0V`1KKX1{CKZJ-qXCvVzfb+^x*dZWA*vIaJCL$*QQHWJY95699!
zyz!!2eLivj=sQ8MIZX`}^3SRsyCWBFFdUJ&8eZ#&TZ7*%5R!FCp=_>|_o{l1|9?0p
z&XM;Mb*%^)!a?MN7=7O%gXKf(pI%@r*uP9V04l6%Rd?Y>F*!uMHzEH0-_{=Wl!jIS
zWUZ)4`ms}%@#dYH)3R9q?3Dk++QAB=ghY)b`L7d%_!$)#B5(?Hf>(*)O4hJLnkl>i
z2l3YOsjP#I(8aQ-!@oz?A-)+jp+4<j7$0yK;Nq2RBr<PTBNdoABoBSTd-~f{7-6!h
z_}I(erUL^=2Hvyej^(m<HsOT~DIxZ5<j2UMaLPXZC3%Nl0*#PDD?4Q0tm2Y%LsI`l
z6~?RsOSsV$>Rc7Vf%eb(a%Y$T3D*&7os-H3WEFFqos$E7w6Xihj|XLY?84-J2;x52
zNP+WoDUP2*e7Lq*_z|6iEYroQt&>B&IjTyuR~I5;>aaw<`3UNvu<4Y;eOi|B%V{yj
zp4;e%M3931SX4MoKAY{?4_pZ!Q7d!Wa55ZeBjhbm_}AyM3h#E~l2c~(AC=arAWwB=
z<Y=2E%7Lv#SYleBRnTd;&{Pfx*BxuEBdL{O+Hq-dsnEfR3swI3_!dwk3&JqT-;Asi
z{B1Z^^t{yDj|<Q6wq14i1%F1Xgh@B^(CUft44!a{5#1~&c~2DkynBOny{vG^$w8GY
zfYfYrzD-U^)0@{Ir)Cv?{hv*LT2@idv(7t5PS0A*W@TlQoDpjbDL`nC_YUCV9yM8E
zp30fVJm;Mvy%KThGq8{h6^U||_1=ttiz+!g8&)G6cW96a6}A)1w89?6nR!l{FflpJ
z9h7tZ8Q)_b{VMFQ^I~{3M?eI!Ac=11=RE1{zZb=RMsI3I^+|(P4N&DU5x^bNH(G|&
zW_TB6iss;g#n;0=?w3YmRWngGNBa9SZUR;*!=ExB_RpxHlVxD688$rueC!f5N+n=W
z+F1yTl)>@Q#B4@}1QTD@it}2|kJX*(|B(y4Ti2)+y1dZaLlpq<&qdKMkRv)1<`X>K
zeH($yL;%9RjI>~(LISv#gcyYNKyJU(-()zR6`dg-$2Uu?e`aYTIfZ%JSs5DSQ{WJ2
zi)xCz+kp|=tEAM&O!3FXq^KDd-$2FI>zH((6y(WpW6_d{BV~jShUgh@Z&}H}78MyI
zgFFay;uT;y8D$fJRX`9XFQc8kb1fOUEQWziPQoBmUoQ7&l(K|=0A$ISkgs{va-*Cn
z%eqg(suSC!$NHCm*MNh=+<+H%oGpaLDkD8P*0MC$lxdMG5__hb!7zwIqWFZgZpKg@
zP%}(smU~x}mIQ%$m`${<S{w73>a0w%A>V+DCCwh3Z1aYrs<&Jjq{d#_gk`6e$IXfm
zzumXO8*o+Xm<&UmiygCW(AC~C5n&?!s3|4a_%oKs1}uV0uI9<Lu?`4{uhhh->#WW0
z_F&xjf4Sa!v-mf&eduyoD^t=h7Pt;C0#*cjRi?%k83VXvS`0wJGNVw&KP@(F7&Ww6
zDvS^e5R%G7$1Xi}7ywq4THB&N&}HyDXPsamf}+X9{)&^Ps$)2%lZQ!-H{eEvM)K9h
z%8$i;8n_bKN4Eq>D{nN@qPvVFmFfPm1v7U*eO@6mVw9LVh0!*s+4a#0h|WC=p(qV;
z>X&<2V%(E#Y|vBz4H0?z8ENvKtvU-YVbgE`7}3q%qSZR)B_12)hBO*gm|56Qa;XJo
zt4HiJTuG8IPzk?~RvLv!LrfQ~#aknC#xbL=iL@HmcDIkjZ74D`+jESb=>IoBLS}g%
z<~ZSrgz>BRr$n14v;B9Ti!@16P6wEkp{8v<1~M48H(*Jb6U<5#6KS70xiQwWtf@k7
z@(C|0g!;O)nd`q8qQo7-m`<uLH+wfMSwiRkMs)qfSRRz*#0Bb<s<(ucErSvaC*iH}
zZ-AZDU~V(&&=J^NE4LS|1|uUk0;T^_h^8!Bt${iGW%~pRvl%=2D`^~^s?6iOm9Iu8
zqllHwk~@MoF)U<W-1&)Kcv|d`yL_Ui8j{<vraTnoYyP%@)~4%x-JjWo;D+9R!=DlO
zWPuWfWZ(TiJ>jArxyRQJeP~DtH7FW~5kYZp^oPAnqYCGi+!r*s5=kM9>-=UsU|2QE
z56ZXf>^P%WFZs4@e`avW{V{V~^3l{F-?5*D&u&KL!&r212v0^h*1;`2?;cFEL(H?f
z?v}}SV`)?Jt$feAM5!$KzR!u~cvM>D-#w>5eh_m)Ada4pFs*->?a_2A9Bz6Jg1#!E
zDR)pv928^@x*9X`BkO7=u5kWwuyQE?OF)4cY4?eABEyttoL&J(w~EY**`t0&L$l0}
zQGktgcR~?c5c8VFgAQ)88d(@T53evv{DBGV0P2uM-hfr+woH@7{>-#8Di|xJBRyk6
zmsU-a&KOHY6YnrsVy{rAK<y5Tt6Z8sW6>-o1gwx{=^5`}<2KwX%R@qdOV_x3vIG#C
zYY1UXyzDh-G)xI{LRO?Tr=E*ICj{NtX|giN3Tw4R6PZ^Tuawahvf95h#0cv{(1kFs
ztnu-2$E?Lz7zVsH^$+JU7NiSda9W&4*2UWpG&Jzw=8^SjKNM7D9h03uvcWm~q;q9s
zyb7VmJesv?60A_A0R>iMQ!M8A<XPMtWFOPkh>2`T1TaKYbEehy?22D-RMBuNXark*
z(40K6qBS_Tx7oAq_U!E1I^xmbo9)&W{tA_(R(3>#P_Sqk^G{PTequ8kIdPQyH1;uk
zp;~Bqr~Nh<S0U_5%d8%M7(mSaTICq)HtDtnI&&(Dz3jHX_wL=duly{CjFB?>mY%hq
zIE3<T2zL2-^cvAGzp$U1>QTlvC%?3_0oR(~Apez*0!sp&kN$32HFQl<vEMGg_TIR_
zH1T-F&TnD_V8``XMr>{QZ7f82EB#l$i&t<)g4glc3i*9J!(l>Whi<b!q~^Qmjbu`l
z{4ucsVHnkUlrWYYvrQgO?P%tTp3HypZ#rOg0x3s;0mI}p!x6CfR(T}8LEIf{tnqr_
zN6q#tX_d#~v>bt<|Jh$<^pRBpS|^WZdwS6&K!=+}IZ|s!<xc)-N-dNGN6@fww#Z+i
z6|5*tC3s2Y3F`)F<8T7~)z$=e-e<8UPp0vaamkAF<*A~5M#!BbXFlX&dD`D*rXpDY
zOC!(34#%Co)yOEx%ipZK4A7FiJZp?s-<Fr>{A*nm(2BbE^Uf0}=7vk0;{^x;mhnXp
zyS(7fDE7Ha2qeE~ZI+ack(a`rU~wMY&7vNhm;D=hk5~a06$0dySSXlLK(t^!8NU$%
zAVa)r11V{Sppw^usYYdTRc-Qm?Eg$7uA+yFtV-TUjGt+oA#Wy%Q8HR+f@w`}naZ&1
zUIcsI2Y=_=v3t=eohHnY2E#+h{38weA(|E#;6w6GOdjR}4dZQ4hW?rD;e`@Nz0%O>
z$o+oTUtQGANE8aXZStN!v+M*0QZA79qe-B$JPdJg$_GvxmS2hbM2j#$0vC)e`Ov?K
za6}pd|4RMw)&dJB{wR@wuoLO3fr#M1f7{zWXk(L)y;ll+#%MxWL4Oe5Trq6h|CsD)
z%~c?WI=!1RR`Av;-MXuR7D%-ri9D`&otfo9p>T|RsybnVXg@FxUdzQ9E|op19km=;
zSHT1H(}nkOKJ(y1+&u74MmSMr`b@RrJ^lOjl|AjRjoGa1<zE?;P=s|W6@XAyJD^rx
zRx)Oc>{G2sQ5~~;_pNrIu&i{1>{o5LDwId)9FTp$`}g-}sQV1XoBO^~4zLDhp`eHy
z=zSOYzVIb+S;z&@@u1Wu@EK5zs@TERrX*2W^s6xNHWhz`H!#;!#y}1U+A3$6jJ9$*
zhZ+F}ppf8UF^Hm@oe22Aw+U8UBZvDSxWJ5Y*$(UBh-!nhAf!BMlqNOnJsIo-of#71
zv(`yjrbdo5Dz-MxXl$C*D4(nDVL~Wi?1ivCkMd!{rXes4>y8=QG%1d*cC>St22Nv<
zj|q}e6UW52K|mZ^UC2d;lK}OT0JR_IUEr<+s5!@1J2+ep#m~xFpAf{wb|QGb0v6`y
zgBBoY+`ywwCSQ;*M5EBe=gEn|fKWVJ8ye-L_&4&?03gZ9sWrPOh1$JsM4R0yG43o&
z7bE(LH(#D=Ooif7nC)pX|6H~Pb_%9-dMqiJi7p?6dh4I7<qR8Jc6KwX?VcH{Mx6{a
zrSjX$n#MIjK!(rqhS&rHJD8$v!#dlC4JnWee-s*o$_4|Sl@k4T%WR<>qVGSY4$HYl
z2F;A2;eVd@WKzX6WfPpe(;#_l(>&cEef$}7=@`AS0z18L5E3e7)KOlHWkc>RqRci+
zKV!6u&bdzd8>3YqC=Ug?WI!x3Y*H4#WCi4bu}n2E7x2!Ycn4AOlx9#wWpJVa<zca5
zrX-gk{tY1uE&xl2o^PWV0hH?kqcmc&zy=pae@)GG2zix@g7RQF-tWPCiR~o2IQ44s
zSBsP=08L5TA(zBRK`9>M$YGF6{TWvS#uen<LMid>X);Yu3f)!=ms>>*nL!CM)W$os
zwOUH6d-lM7oLwrzOg!ycj@>uh{*Gc6H8R5Js&@h)vUsG=A<is9C5sjH8RdO*+b~y>
zFeqn7wZ8!KjrN%qOMNOAK)ykh9dcQS*3nSWFAounD{K*=IHuY)t52AgsPc(^7-?>X
zhx(74qswe-FtG-@?pWW%#8B7BxU`a-gF+-S-e+gNOyYhllmcNVSH!%F1Oz!^wM?jX
ze^djn)&g6V+Z2%5gIzr_eMeh@JxBKDq@w;AE>2(c<RoUWFXhS@eWn3`$W`7FA{Ibd
zo|n*u^gO&e_>5k><`%ie7WK4K<XT%uEMARV=lxkky{?j6?`^utucFmt!4%_>Y;}uF
z^$){4gd<$0`Lo5S2||^Xb>}nw49Rt>QBjUpcn7A|@o=Qmf}XI-APf&TEEki+aF9XY
zs<L)8WQzsBiLqy^U4udnCoU31Y@7y=SBG@Z&r)~?1h26v<L0&*QX5NCnnTbQp~&bI
zAPb^Mu$x#Yk2=Nl5XlH|ugumq)e}jX;jdH3*)>J9PZr@YuMhd$+6=ggwXhmu_E=(~
zCy2w5#t>gX`5CmrWDGJ@gcmp20*~cAiCJkjZZAije%5ZjA$3lq&0?@LJ4bS97L3)V
zw07uG@}Ky{Fj><pTl|;eeW#oxtx5Jow#GcP$?D$Ry<25wObkoSJgZf3;mnGSiO3q{
z<;_k_P<#b`lwfRWGlm>>h0L+l>25EdBR6`(ar$qC;&fBcl{u$ca(Aw`U7Sn^xCPpW
z>MC+`&{)g&DsSn)9g~$W1~;J$K~t`>jp+%u_%<V2w^DA6rAI6+P^6kx6ug$(yjiIG
zs%`OJZDuzzD3C_t?f$!KV32@_e8)<lXe=DUfGTYw@CeMbjAi+9>}Ggg7s^+Hs>&3s
zDzr?#>TU88RK~$`b?tmM4e*W_D1ycSPnqF#XY341NNsWe>8>E;2wX_gN-g<owxh2e
zH%Y!;V6EM^JEE)(Mfrvg%Mf!Of1Bj)SnNJ1{?}XXiDgrD%~HMY^=FLLQXdu2*!z4o
z8nth{gxN5RRHe0iGd8CdJjM?CmWeUM!o$Uc{qXHH7Q?mS>XiG_$irNL=+;B+JKmjz
z-B?}!Vl4H6APDaRt!C|r2feMhO_drIWf$_V<6VSK<y*16<$FfF>Dd|beOt%MTCC#|
z`GF6T`wC||PQ6v~!&s+0iaDY_m>Rf0^4<&!4V9cS^2hNDHD8cO>pSb^U+p2-n-Q`C
zc{I=GNQ1)gd{%<C`@2|_V7~uOk-)@`g@(<cRaG`*f%inS2!@g#AxwCgFAGBwRW!kP
zUKAsOfV5m79P&^zHp}8PT^?kn%yszNH12o^^G`cOIzv1&%LOtG+UJtQ6_%kw*o93K
zJ)z~ZG!`940rwIsEiMZN91D6j$#VN=DT8+*(;qV7a@81FVUuXhVIOB@Wq}xK(^99h
z@P5sUkySxloZB;yufIApkMi0kSrfxB)ezEryR7wROJMsjqhL7%;W~e2N(XN~>6G>U
z3_FSkwN7h;52h&GkbH~HqItZsG0{+?^}*4s!^R-0O~H%U^qMOqn-dAUKs<QO6j+<*
zM)U&N604{D#N*9T3+eJ-Xd+X7?!Y{3d|T5z^i~M~QC+gl$f+WKi)@cKfYL34jE84K
zcK9<SKZDbthynlji9h4gAvQu)`=7=$l-&U&rkSie6Gh1Nda$XG2)VuM6SH8@65XN8
zp(1w2LX81HwL1;VYLRl_KMTo%AWDVtd*Z>2I^>c8pZ$3-re5xW*w7aFg`G@gC6^DE
zUnbUJu$w9BCb|ehEtX%!YUx=X9)B9?zxLtKCT8O0R)~n-7{6Akt(ceJCSo9W3dCIi
z%j9?V3>O9<o^&R>*pS!1k7d>%a&}C$!c1{gFOoml@^rVCwL-W3W1<7G9kW#)j-DWO
z>8nGAzWgc34h<->*wnhyw!7pJ*Ww9taqD+s8f0W34sba}<iVrfGI3u*kW4s#EcIr=
zQE=tRpVO#_`%v^PlE?iS*$-$IHdu%J#pa@QEL!9Vdt-DxixN=*1U3E|E$OLfF4ia0
zGlIKl9W41&>}upfWWXO!`%)E&Rn7{i+B4}FhJ)}9Oj=f4#J_pFOaf4L1Zw`QwVTN|
zqrN$!d!GvqLTyDVsW`u$w@Fl^-Xdi6-+lKGQ(P}Eq@mIb=s*m}GI`MlU`7)E^mGz=
zDc;6jGeuylyc|LhLEivUUh$7%EyGSlP8}H4kk(fnC*mnfrL#j<kVWS~A-=1*e<lN8
z3pfXWF~7H4rkKTO!RjW)UytG7jN_d@wA0>*;ZUinl{eFHxSOdc(bK;Y8r@sL#_CLg
zt_M6H?1_O`<8MbpD7psw@*gJdypj-C{+()1p#{KE-vo?D`Qe`?1o~(};&;6d3j|h$
z9=#J<pux3B-m}wt2s|n8$8fNd7+cHiL*SSgE*Ib-`M_T7-?zVfXn(Cj5b?j90;cWh
zkt-jCD%*yVld-F;ekG1);+9(L5<+;pe4KivZ48kMU7^TYmH&j0ABQVanyOHI@4rEF
z9VZiS){E3A>{tn105v9S40fcXp*Mv7vps4II3@m1NH*%0!kY^1c%6K@#vx5aNPx1R
zsWAlBG_>OIh&{b==(-3wu=-(p)hGZ8EF&X(*SO7#a{1MCtMB7K*5e@E*PA6lsW-5s
zgMrQad9%2t62H*rvPAZ;F+>m?13_?r|Em*qJi6pS`xnGw4vN=7HI4}~@Jh=e9-O|#
zHjqPtBJh^-vNR^l9_nLJfrzm%wZeqQVKs(iDi`AIVy34XM(Y*I;i=7eJZrTE;Sq^A
zyzm_-KM=vs*7R6>aY`vNHa*hl(i^I?d@g;b*M$MzxmAuz&uCy6oOHB3Bk<1qxKL^*
zC&&0a?8YBn{zyM69c;(eD1tM2vK$wy%Ot$K9PgimBxi&drNK%$!Pz8aBDbt9mIYOQ
zJ`IZd0{Qod?fQa0GwEUK<cY?Q2xplvTuzFQAhQ_7K)~(fc%~T+y7%Q29~M|EoV9BA
zua;B&Z7mGz(&4lkh3?Ng;|w`Hm;pgw$S_Y$GY_8U)OtD7pIHqbI?cCBuk_wVkW2-S
zoaN7`4H5y}0p#rL#7b-7!8*sfEp3`C=hpOCDOkM5T9;PXfRP%33J@%%cCcK|i){h5
zUz56fNAHTwt4NYQ(Yv1L>LHN6{*01=?sS0b($A=yZ)LHG{x$CMxULufD+8h@#33Sm
zEh__kO<{a1^OK#irt}x70y@ZMGy@prU~3rn4YW4M5POd7GC4m;0DyI^To6oxf6e3k
zz&<<>J>FF=^!?)oa6vgBU7*-#7KvOG`ws@PrExQ|gJp0o_9;WO!NtL9svG5!^bG&P
zdn@6{x-@1AU!fLHC@IhiiY-l{9WpdHaRS0ZN)w0i&I2kPbm9ZhkYP62Gfp~BhI=bq
zMEcBtWy^?YrSITDGSXRK;zSwc4O%cT_7KE_xk5+#FBmGV3q4sm?Co+{EEpUF!1(2I
z69g<6En|Wxs-WXEF{}kBV_9rr09F9<#`+o<r<tU(jcSZM&Ktru)xAWZh}7?t@jkr{
zEf4PyW(-MVr}8o(QG33kHc0uGa_`al!gp#rpnN2m3CH6^?~(svOl%4mmPv`AW@s&k
z?J_wCS`PK-N`K*Am^Sx<FYkppd6f})FuYM$r@2s(w^M^$H(##tF;V&DVa*MXKvf(}
zoa-{V)*89~`itc{qw)w+`Fih<PzJLO;1ZW9sgtEJy{HG0V66Wz)tMitH6X}oK?cf8
zlbUG3^2_M>&D6N#32-Dbu}s!iq{g{dSOuHB4m(IHeXkJd^7ufdC(%HFNiOTEe84<;
z==5XRfNTi8W>j02YJUT}1)H2!`~s<oT??IZhtwuo6&xnTv72>869W}tWYhd6C}Wn*
zbpKuD8Gd1b2vZMqMP!Dps4C@BZ-gHTxU9hl4}C2ojWOojgs~;k6tnX91Kcqt&Gtr1
z2DdrZ$PNAmjS0H`*FyT_5}&0MF|Ib`LEhh{yT*>KP0|vxR_&683X+#p3m}NLRa*Ub
z3Pi$B1*3+Wb)C!%KJAA}>$4J(adS7824$&bcI?aXK%VQR?GtmNZnakCL>r~Jmi5LM
z5WB~eIzUD{ZB07dWJ0YGDlK#wY>v69#TDq(0J587zn0)~Mp&p{O#H<FZp?+)9oTxQ
z+>)rJmzUrfUEe&BZuNE%Sp!z*ep)8C*;-Y$VkWnzHs?cSwOwbuf61rdF|W`iS%E;Y
zZE(#(`LfM-=(usXO86_W<da|>$yfbTQ^$GtF<}Olb1;?V4*Rqms(fDV46P2A0XCz4
zLxbEE62G~=72Z_bIvuZ!N8uoK2YfvZn-|UCq8@BG3=X*LH+;et`=^G8+?}2=wuEks
z&2mp_!I0fT=>5I^3*<b#1IkinS>0#+jxyo$%~%`c6|&Caw_?nrD-ptnt@>^6Q!5gI
zi(pyF{i$WW7f-DY^2P@Fj&EYbtzr{DU<<@cALYRqJD?LB!>DS4i>QOU)UBXn!zQIt
zNfdIse9yQQ%%#PEzVF=@xL>QhV=w(6maVxKuKFLw9O^UZ8aD2aV!Q?<0f)fNrV8l%
z*j5m74**<y3?44g#WK$*gOmgu*jV5B_UiE2u;gXICt6&rcsUh80FXg1SUL>*hTS2H
z{231$t{WovWwFnl&v;DeVM73ze@CoGM~(aYkthyboSvCZZ>AGkGfVdaNRcJp44H<k
zvjW@Hv>P(*Qdw$qf}@GMb6FZ3ayRto?~vs_70dG`o@)x%7I`S>SPt8m#i>_#XJ+<#
z#Uiz=0KUTLSs6`XFbpkX1|^JG;3unMfhJ_(KVI$cnfS+h5CA*^SIL?hw~+&AwHs$b
zMFT!V-y=a|?hD53m@>N*xw7jLDd^-<;|RTPQB{IbR_kN;fttvn135PMIBX3}#efO|
z(X!DNySsgSNvUk|*Jyje8ZB7<vf1g`0o{_ZSCBZe#kmMK<3T&@vaL1*<*T;F+-h5q
z1P%LNt43_|CQ!S5H_N@Q1|qgSdY{--FFS$}N)b_EC1ii%1EO1li=_!`I7YV0Ph)|1
z@{SqvrSJ4-wjNYk*U7GU4_rz+9#MbN9czThpcdJk-dF<0GKk}6{)`-oC}QlOP0|z3
zKmyiLnh&Sq&wV9$bTJuk3*;vKewsA~De;S#QXQ((*TCV~CcpG%xD=+!bjh!LnqJ-s
zFhrLMUZ9gPIEFgl*?I_z-=w#B0;udqr#vTq>%XvVA_%;I+s=mHB|;K>G6AF`klz=L
zhaL)4)s{a51HmpuG%QThKicZ{9?(Y~4(ipwJtu#13Ydcr2e}?(TB{0{(E@qInpe}f
zrUp44P#+%k!59*zN`h)^^W`!Bg^muI%cwew72z$+e-1+6kks@r=zNbC4V~o*8F%p)
ze`Zc4%JPNsMC@h=jfmpl%5oK1*Y>Yz6&EOREJ}7I_MR#2PkNi3D4|coaR*L@r$S_u
z)xoh1vH!GD=-hKpl4nesXS@j7-_p>TjE8DHa3($*j6W7Ryw8QM0tr}^mFKOceANv3
zyPYeg<I2X%3#l2+%q0#cFIqEzmk<&9l6}|f++Ol>jBY}v4bGA(c_r3g0|uBIL6nkV
z60h2FqC_My(AS(4<PB~(3(}xQhbK6HUiX8pp}9@ou>Q_C{akr77-nd$6`@h6Q}>pS
zmes&`-ZPH*?br=C<g&cW8Q@`B@Rd|Hpf75m5RUtI5|gO6LG@k*;N+h+28_2#-VHiV
z9DbF&7wd!#v<;0dL`dBGiE03ExSrUxA9&N@Tk4?7Oumv2gG?I915Fa&P_iNt(>fZe
zN!K5xQ76404XlSz@^2g4NL;uhAKTw?C(3mBj}JrloSt55<iGyR5DMspK&7PCFhVF9
z5Jub@E96tPh7od8=nd?Q<I`9%nzU<Vk6K5vQpG~ztw;v1HLy@s5S==65Ky2&K2xi}
zAbiX<vS;+D#~UH}C9+qoB7EJbJ=;<%dwV;?4RDqkd`*l{_OW(w+wCc`Z><B2xa0%C
zaFy&=o01Sw3jkUsxBmXMjtI84)YaF`0S6pV+rt#@Oun+NP7bVfU7!x7dYSQGDF^xZ
z%r|E$#ldOJv_E*DcFG|(7TgumD2Mv*VP*68!(x2cWO~8$;Z9bXFVy(gz|mEF%ft~z
zgi403<m9t4@DXvd%aK8(Q5l$1pYtyDo<g)UA3SqZ8ag+So?&vd4YRHSYS=OM_e?V?
zog5qALz%&ChpIUpmk7@)eXJQkLxbD!_!vzdSFXYHo|hBSlxfO#VApMt&&LQSWFQ{C
z5Z?d+Mwl4tIG^b48*~6u0-8Dg1ju=k|6*c5!3f<UopN#-r`06b4BYFKAT45<pj_wW
z)LO;9a+#TR@U^taY4$=Mrc+&Y9eCsP7)uQ*k!tG18MO{{XIpA<%j%iY<QNn`<!yt#
zj8?lI1jC=@W6<<CW$Z3FJLWp9tbClDV*{9=$vFr{I=8k*tARn}aV6A*SGIwYlLCF@
zyr3b>(XChm0$jZlB^Wh{O%81XRDq_~$KO^j7TP3y+V_7b%Iaw53FsGXt8K-i?VZx!
zpJ};T2nB3;8BnY6>A->gWT3MLqD^H`>dXpd;U-EM?9W)fk9FubL!lZH?_u!Nj}iGU
z=f|KhErhaxHMz_TIlsVqKy1%QxiGc^v~84dM|uwZzN>)rnH%yV|J^+{t{WC`_lu)v
z1=~oxrE-b)9I+`{!yY|%%B9g8j%0#)o!zWaV9DwvWQi?jZCh0iu9cxCBf8s5CY4HQ
z8VETZf>P0B(}Y<`W>~yjS2bIPhd@DGNF8(h+hjyMf_a1NEnZkS(ukf%lHDjDvS}c?
z(&&R}n4^Q-ju31kX6VnjPmn9a95%T;-ovDLn87NQG5*XndO#fTS(e!p@dj)Cj!gt9
zq=Z_4w(-xn_+1@EsFkUKB89-T@!mrROgh2>w5+a3&z5m9aR#rI3AQu3+e_i}L}*5N
zNNF4%IgML}V4CP3h7yG!DVb!?y4#WL)S@}td713*VOl&W8RjTg8ZECs_e{CUW{B#$
z#I3LP8PIfN{<(TVi{ePG@jdD^SLY)Y5YO$#7f=M`T9XqKXVp<I$aOItlwRYRcYUp!
zRlDwi27C`bz8fa!ln^evZrj;)woDDSA3Is51&y%A9vCl~m_AmOm&>p;eT7Zyq)RTA
zO8e`?;bl@4TTAV2gTqh}6|0*~tj`d621BUd0{l_!t@Ch)wV<un_}jLX7Xx}XFZOb&
zP28>-?XW#)X%^PK&Rav|AURNd^V4m9Be23U!+#G?F++#yV}8(v?B0-7(hwxdqb5sM
z8q;UX;83Cgv`m`(83YwQ!Z3hVNplR3)6l~SDMes!dX#~S$qmsJ<Ov7>$Hr30g{VgQ
zYrW(XV`Hn}4d&7^RGWS}#;CT$3YW7UV4Jks(J>B37y{~OQO%6@0AA*0R&4*0(lIhS
z{sH7l+7gkM=pn!)7qz%#juEU0fq!yioIDw{&I^xj3iS$6WsU!bwD%6Mv#QpIm!cF!
z5m7|3&>;vRjTEZMq(T}=2rZN|Gbb}AQ_jhmnIr=Q>Aj;gDU&+sne?7OG7y^d-aASM
zMT&^O=UHXX;r{OTeSds<%}w6(zPqfw_S$Pbt863!6;TZcH<`)Q7+5*rWZo}rY>}Cq
zlRw6YP|mSR&UG4qv9YW)IcCXu(Mw|%P7)vv=i3m5A;U^8u(in50jPqBbzw}RthrGx
z@<~LzGwSsD&bT;rL#1dA?P9qk#&|*wPylkN{cG|bb1(t<Q-WeZm&FNb!M#u~|CbnT
zxjcpqqp+4YL|sv>zRZ*B<;vJp!;Y1!V#G}i<#Kgmn?+FZ&FHu$2y{#yQTNvx<@%2p
zCf8Ye1CN*M{WE^R8XQZ&IX5`3@aSgvNGh=h5+K4${$yX5og{y@4#rzl@QsPbJpWDu
zM2VD}P`Sy*Tb`eSDcx+N8CljKw}f!4Kr|o1kXVvPZncT_%T?h<7aydW-zs4j%Wc*e
zJYR6w-|ju@4$>DaUXb<gQKv@kNIc04cJ(G>aQ_<2<xX3=eUW&BYHWK+OOULh89Xt~
zSOl?B3dm1S%`i0EHXzXw)OZZ)v+xQxGEA$P>Fso@uM=XHzs(Q@Mwd{FoShbel{$%Z
zER$BF!AR8IlR2Tw_01JA9#C+uw{PAz77k)L07~z9HVK@RaLUU37#nj;F>Pi6FYs>}
zc1OF>GJRP^dZGVfi$ANqqu%u*|3zVaw6hMt3ZcPSve>yA6rp<(7!bqnc$1u|B&5*-
zY4dT|aYy@pmMn>}!J$TvETTYUX?l-U>TxkT=CI5cn8gHGL3RdAB<?Qa1k7Yx?qf60
zHP2v8bcGSN!uSOD5rL^zMtAsBy|!q2l|R#i8>ArYygG(p9urJ*j;u-C-Kv1S)jwJe
zzzp_1bi%d%I~-*+bm5Kj>dKfxEBaYyyyS+BtWR_{ytWOQ!CGToHbg^eU#piIG6ejI
zKXJ8*Y>WZpcWzng8=I{0GP|u#?y`U4zF)-R&XT*mMe0v<PLetnyl4Ef**0`-QMQCg
zsX)k4dk_^MJxfQj1|nlq+T@<<t#+q6Qb?g9R3nGLZriyxhy>l=v_`u%jgCp6Lb5do
zi?bC+5BA9nx$j$)u<#@;VlDUk+Z~AHTzSBsSwbF_p*iwk43u5~>>Xy@(sW-g52cwa
zujuhvf;P)v{Y#|-ISNX-T@&+gqVgJ5@lh919!Y)kS_<y}1@dS-1C|WU2KyEX+38!x
zh0!6mS|0bF=;>h*0N%~QqiU@@kw&jk^4K>FeqDlH`ed8~HCV=#_4V?UH_WOZ1gj}H
zp7zf<cn0cez#6!3xjYlc5}2PN_gNDsTP^@d^qhaiEk`FXF9+pjUMbJV;ITms9G@*O
z_y#vJ8Db6jiSyk6KD$})MgJ0>T`)797<1(%8^f6Fl*X|++>!Tkm<8}(0e>b~@k$Ic
zQwhr<E3c*jEW@A(zHE}$f)dC=Y0PGt<n`1}Lmmot0V>5C={uEXAjBGZ)2FHNi)uNr
zKpK50Z&|O|x&nL=uvgypuenDUq_Glq$2)Nt3}iXkzr{Mi+~Y2?%6u$Z_4gPI$|$SL
zyFuQKO$=$kk_3cmsT?x2jv(E8vC#l3%KI_9dKRm~R6d9kk(p+i&;UAp7<&tv1d$Y;
zK`sMTyZy*Un+FmJQ}*M;`P}n!RQ8*-1??B&>XQ%=#cVk)HUBh-r-&D~AFx4h;{U2o
zq^pkXZ9jtbH2TaN2<grc$A3km49LPqP-f-xZ;wk0n5aIDF;e<};lIP4!ycnkid8j<
zG*UTPEngZF1J*3cSJ6IX3u{+xf|&hxoRd~x2WIEjL0v#*>X-YEeFht=n45&^cvX1m
z8(&;YA|rHxNCRFLA^)`w+gpwwDv~qYkfC1~oH`x9lXI*MHwa8p&)#=)h9nZK0I&{C
zbTtM?q|vJ7Y~#P1D;7$C>d7=VaI!$ae=p}4AJX^XjLypUa|$HZWU-02^&!|0=OA`@
ziENkKYV{e<gaNp{zreg8R(@EDmE6I5WXunJ89?f=94XX8evngGsM8h%XCq$hhdF~Z
zxfiG%D4+luuaq65r%`2N<VP_SFeLN8c<s!OWA0ci<}6!0D=#j>8HnF^$|>-2z+U6z
zC*F&la3+H6JlZ*@z-Sr5`Vk@Y(;zwuPo0NumLubKm*^T~rK4;<IOk_JwtlGMgA1&4
zTNR3;Q<34}ih1BQ9-_3N=HTd=ExTGroHp3?7bVliF`$qY^A~2yZr+_85K2G!WekBB
z4|)K$BEPbQnmV=9IN9APK_zd!F}HF-#Q*-fq&G%YLW@8vOQ96~#>Y-w#0v=EmjzHP
z@uxXL!6m;nO7=iCli&GJO2ZWPFlw;yQJw4=9M#YWBopkuSI$vmB>Vs&wmJ}d$6I6I
zye~%;k9~r*fdOFWKuYdwt)g&UWrOTjLV<8T>>q7Z0f??gPY0xzcwd2)+31v0;8L$K
zXJEsD@eX(lUPc(Pv*q{cnR#<r>-ZphhVO7|YX!rLvU0GI9MS$KNM~P8=o$dG4simP
zE&vJKCS81Kh!c721$Zb`SC_8-r9v%Ebc=xRmKMz-D;ShwJloxWF`SaxF|nX!K(u<;
z=E7aiaL824p*}Gx%e3Mt^m@j}3fK54tA3+iX+Mz{DDKUajM%8&iOU!nNQjfDqI!?D
z7HJ<R35POM4vW@2u-j@Cr0ZVfa32Q|1n3gLBbdleXZI1-4jkCHp;(UeIgsIqiEad3
ze3Wyo32T0Zy@p-J03Sl$nlLX%$Lx{V$6c`s`|c0^40wmaP`GGF`o>7m1I<~M(l$c(
z^EuGSqRD_D2@skpE&p63W&XSGbApBFJJsBh{xNo_phzPGt)|I<oZ8oePm+P&uZ>@o
z5AS!7|DqNG`Xz<%7EXx4iK+=O`j)CO#D6a>rGfK66BFhjLh2aX4q23y)DZq-V}!cf
ziDXQdq24>3Xbd)i*rSh##@;a70A2gQX*@i>#JD1sWkj%FSvBGtWn>T=8V*%R;;YB`
zKEJ&E^7hO4Um2Ci(1FMEz*sG#{d>mMFhq_j;}{b~6}h||A0tp%3-b{{1Y?7DkYZYt
z6H)`bU5;ShVmUF^jOHY0YcQT&m&D|^lWdNt2a%VP{avmQy;4ZU#h;dz3-w)2Nkg~0
zWO>%<YRWkOk~@!%9wy?K;Oug0jI%K_B_pT#+eGJC;!Fdfbl;GxXUgfmvyj;ak#a_E
zt4Bkdn-gopRZ1$10fV}uB;!I9`P-FJ9-YE_mzN4B4oScyjg>f5DnnY1MzjV3hq6YD
zCSdI{-rwg08TG@}tV#s7LKkG8>NNF^8??=WQhP*lwt|^rlT3&K_0NqTFEt@v2lbcQ
z_yxt0GEx_ko*r#hai|YQ8wokrDETxD3zmmfEQR!po^dkIT$$+4IAxrTkU49l=(Gin
zyybjph_CP)?<GQxrZmPg=+`)XDXF9>_K3p7CWll4$)xl)Yr2`WGC4hCgc7UIu9GSL
zjB=JYAI^j6(d^HliegDf=mm0S&K;;Yf^h9{HS(#>Dq$B2YO*C4UE`Lp3C{Lr6hgx<
z%%_?f-!Qfo-<W4`VLL=TLq6r4Sb5x%30+A3=xspNpxE6!19r^0-UfOEA@Jut2!6z&
z5}@08X_+QAWwU1j0!^$yrP}0}`Men)x>_!9+5>WcY`f6Aq+68%KmeKKB5wf$fLteN
z!KliP%=3$5*nLrl5#EAJ{C!LvQfUs%a~4+N-zid-%BA)`^b6FkgkR(`>$AND|FsLo
zTIBNJJrs%u9_lNt-($J}u9SRp<jS=BH8uw_-XgRY$W_4|TGtF7dMJ~sFv-<vX&3`B
z2Z%c=P&yUhB8JerHobxCmVEK9^JjLBn{je|8W%OqfGt>J>V_D9*WSo@`IA$mv_`*j
zmXGMqp?D?N>W%gUC^iDCwZy|s-WQQuE;h1|@iqLBa&uZiTv{);e2e-#y%4+!<9HRu
zN7;L;&x476%*@1P1o9VurZ6PlhAuC+nbd;41V8cZi6B;P2PJ9cju`YvB=&0>(w+Vm
zgDz<Q@nV>k7!I~9y*_YNfaE+3Y3nVOX^Dg^yh{xY;S_q%O%M4pEX$G=GrS>&(J~Li
z%39Ig#gAHLX6(`Ilst+iuy(;Ln?ifbC|n;u+ay~q3w@@wWT2$ZJjjAMu}0vb$d)KI
zROcoV(7I93N#^-CY9Vklf(tB$#InTa#}rTj1-9G*8v^py>SbYc2Str(YmD5~d6TiR
z5f=q@8>?~E5#6baV>T6qCTWY7Yqwkg<Y-CkdR9Wky+5+l$x5jhg|RGVZJBcm!meQH
zEcanh!-ybXJNqoPX+?C}?a=PB(moxc5v{T+SbWP(kV~sW2+-37^khwpzYJP?RUWp)
z+GuJpt{Pnz+zei0z!<6<g9xahBI}*b7Tn4GI9E3K+tklkeONR7uudCgqcu93xqz}M
z@e=cVX@W<Uio4Q+(<<RFGv#i7hT|LyPRn|gY|eRk6-HQ&$FXJ_+Lkn0f-vmg!xE+4
z8{yi+CL#hL_e5u~C0n_i&=>A?I$NC>?nGFrP^m0bd68_5ZvZxt9q^~z7aUEWS+f=a
zxj!BukOxamSL6d}(#$X9ekbO{dKC`_kDSuEbLUg#q2M@+xTQap2PvG6nkXQDHSUB5
zrYH{wTL8(?h<z{-qr6aGNs-D&e7FPw2s$hQ2R!NnnWJjGb2JpSRq|LO-(vMX5Gv3u
zBaf%SGdG3d6L8lcm<lLq>VZBro6wU!Dvje&V}lB%MXxeJ=Bbj_z|0_gB)&c!<Rwt5
z%d||M@z%`iuXjA*aN|uclwYglSy%B&inZF<%yIiI+C7(g=1Ghw6c{!p#h$laGO&pv
zQeN=QM+J7ayqM^s4m9A)s!1Tk<E2=-ehj6MmlH7zTVS=8SJLpg!@+;_zCnScp2cDj
zuX^LHEJcYaF}zX#F0aMbL?9R{?jS|z^%#hfj0QBlB@3agy^(&g01Bq^YGj?f8QUMl
z<qws&62s8J#6_W_OG_1!#pG>MI7b1hdB^yyn5)Ui-$Ewz<3&8)2J0u78U=pV5t)17
z-JC}u;KF!BRRG3hTJ(Dzq+&k$`}WVy3T<ta54=s&86rY+k)EAvp>ll~U!xc_u8039
z5!(0w$wTsS5K>J7FM>eFyL@6Z0Rei;ry>8wTGfJocuQ6hRby`DpZ<)iPyg~Ftn_DQ
ziVn)jzx>yza76#f=RW)ZlzES>Wpyf+Az#=cjL>j3`7#zu;eAGGGoXN+_EpFYmXw(U
ziHvKzWBn?5o&DG8)1@lgz{18e@*i8dE{FD(Z~ku-3FR$=f+Jv{WJD$?mZyjflCvim
zvPi3gfH-d+uzm5}2?ni_*|b_D1lBh8ZF|d>o58@}o8Yi$KZLvGtD5Ee{t2(MMoa^c
zf>znqJ23r{t`}z2(@x&be<5N(eU%3+e_t=#dpjT@lxNK(mL2>VFa@0Mb8r&PmLJ43
zNNd)3p&b8V`j+*w8RxJ`cC=xm?lzMGel)?s*}h|j$&b_UsA{V9j6b2Y$WHzXecVis
zqY2!4&-p;Kf8t$$GB95BXq27(ZOUuVk>(G6p`!Uyd%L}507?Zops>pX#gE%tDhl$m
zpy{zqxQXTGF(n)mgsRD7yQXJ68x=s6U&I*nr~~$l_ja4$+piM0n^qu*0{?4}U;4->
zfLQAo?o^BnYbw98QJ@lJz3lGo;VqDy6#-_;ul-B&Qk1$Ie`8}-P-Uk4c7nm{X1ZGz
zm8mPF*Gl=Fw+&<%QrF_T_L!gmH)>H=W)W1f=LA32D{$#~Qynh)#ptuLR~kEuD3!*A
z?H!BQtyeGE$2*|SL$!h#`TP2yFkw|_X<Ly)Sm$OFmSw*nAO=P5gLWnRr*`I3gM?h3
zDF>u>xbIX$2l*I&V!^7NY#>9k2&$6<jiR|~RATsjnzaI6*jdx%pcwT)<b51GAq5J<
zQ=zi5K{`)x=S3yp-69kSImFtm$kbLymqZV<D(TLE$V9+VSN}!NW<2XCZPd-jp^@JB
zh46WI8wXTTgrW3E12t=K72y0s<0EKIxWJxvrSx=4lvV~RLC~aEYKLY2Y>)T0!4DWT
zV3hPRrYSV$sA?ZJLH!O>r=B8*`-e2#2sRO5IwE$(SX|yJNBSGoKh!$GE<7qYs3PBl
zt0|9m0+<D=xl=xdKTL3NdmxGnBK)`S1jQEjI#~Ke#}LQVZgI#<Sw~mk7y!<&Jo?9A
zpgGsefC*dSc#^P$VH5xV{;v$|=#`2w_;{ua3jSz7EO4#glQTFbQ;CbXfj14AU@+IW
zV)|8QG<PshAs*5Uh&aYb-lJQ0IX0TfHr3FLFx08XC<$UJRM`ww8)mF>K&MWf4wvEf
zj|2H184(nReNNHYQ4@FM1V>{k5av{&%YEF0t-F$FEnBXIO3ke5>tIdV|DP_SY)Qrp
zm(flL4kV6)F-h7ih8BCgzd>y`s{D*i!w0ser<exifoV^O9oKi{NI5Y$_*-FK^x*LL
zTOud<@234zU@CxvCK)oeQBF=>Z(w$`VqB;AGs{2+(V<I?v-#7~gRgiz12#{Mt`!O^
ziFg39v6tmE-{lBWm(zpYkY!pcXBeL&O|c%TrewT7p0SMM<K;-@X<RE{wJ7gPg_A`0
zAsB#pRAhtcQfY6uw=^`BH{>X1nkDNkU@(xNOz2u6<6{+p9>SB#$bc$ulw8F;$F+nm
z@X1W6Hag?>9_;O0aE4wi8JIn9@tI&O2%T1Hyi1Mqf-<-U23761*AZ4A0bwHAb^l{J
z)Z)2li=^J$AqGbS4`sn5@6^VDK#8In3{?TG7m3yiA!X~}?aDNwEYL*bz+q#h=<UHi
z!bxau#0KAG7Hz6VBzn^Hez-Vz`3KW24MS`UxiqAqp(!g((HyviuK<mfNij`E4&xa)
z(B$Rhcw0?CtQVBnDe<=MO;bmxG|rP||HTeyt{sndDjjDgcB4rM!AkErY|c<&&WbNF
za1xDvCmX@pF?>|_g!6i8>YQhdf-0aVYv-S196K5wcKM_IeKIQa%eltAjn%m#fWh;8
zLaGUvMJDIRp6q#O7rDUxF$CdC8M!b--soHv%=;XYP!RfQJX3*K(~HxzOj}FUdch@3
z%(=t{-~afsF+)ZSmrHF?PaGwe*#NtA>DpT^_bG)9M6{U{S}a$1^Q@f8s&;0tcVvNG
znO292jKvOIm4;-V_4$syLT2DRSu9rv0eQNhAKHREY1Ya$p{^z8DXpbzgJ6hIp-I0M
zHr{ncfssflx!xvH#Zt33*z{_U!)-KgDxhw5QU2^rSYZb{)@}@r$6W~ZHRGFbm5vYG
z+;X8BbvV&(_Exyp5nHKo4Y$~{_7()y)&qOV0eQUD-{Y>J@&(Btf0?jVo&BMWK<7{q
z%Kc|Cf`qMa^DntenGAu<L%BVKPNAt>Z$U?LcM|_A0czeE!m*JzLbW(`P0D$2J&wN_
zGA-U4JaTB6Oix`gchoS|8L@KMORy9W%E$Z=*0i*ol9@3vAer<_vQEe>Bmcl0Zu6Y&
z-*ewF!>j|JGq|QT%%cqBrV+3`CmQs#jnSQRW4n#bvy$jM+YNct{VtdJF@?rFgzAF$
zT2FVH-5X_LnilmUE6TLUj`nI`vb@q&7TY^|>kBfr)Ff~F7TE|`duWI(iSbmZ&uFQ?
zXpj+S%8*u;5?>aZxD3$k@|db4&9Wkfj)KnPWu><Q69SYbxEJ8dDxbhyfL$B{Q(jl6
zNg?D5E5ab9HU5kqWe9!<`DwD&=}!^Oplv#MO0OnZ=fB%-wz2b1dck?i;`+4z3I*sX
zjdl^?hIk6)WD}H>=~Pl=qs{h&!2?i0YLgF!LIW;-xGy%!T`_1?hRnXY+lY`aG|&#)
zoJdJ?SH-ll+Zx%D+FOYoWKOvD*as{Ci=Z<!aF4&H_Z5NPmIGG4H(rCu#~K3Ik6XQ;
zlJqDLhoRN?+1l=-$jdsp-{oq4l00A|)f082JQy9Ream#}HS$o}EG&;icZKDr5Qf6J
zDt`_6(vSt`G=ke7j#;U4jU|Ik{fIq-{K%r>Lj_=tN_jNs0=EG*Rit>~oIrM(JmxKc
zU4VpAoX(NQqaVZ_BIX1-$rJGe_cl~$RQQv&1E4XDhX`qur)<-nz%Znzy*EW&Kty0K
zm^JG)JQM6h_dX9x!tDtF-YCxo1zdyEgmOKm=|^dm=VBc24CagF@V`FqUr^>z{xOHn
zYbt;ybL54f#w4Uyl(TwTv%DDN(_VOFTj`kT@=|Q6>Krcic-d)V^3-6HS8PFXv4gzY
z(KSeC8W@PG2sqWX*X-3H$QgV+&D)ZrS)tNfRf@mizmwae00sVzRWi`;O*@4qRLNT%
z>vaz?gtg#{xBVGsLU$N}&v$Gy;o@eTcYm{KRc3+C)XU!!Gw=k)v|#p|M)R)!qR|(=
zJm>;S<9pr(0@3icRr7k^-)2M~bRue9_>(T)4-(-vQ0QTS=gNnvSBC!wNf}=EkxlNP
zQ|066pG$|S&s0f>(_TIaN!YDVANe#4D}*v)+sZ#eo<f^KoveSRp3sZ3A8(tZm-n+o
zJbF-=d$>yeWi2D?3^7@s`=@rM;z}tsvT@s+d|~|E-g0nf`O=#)Y|+mCuY3g!KjAtu
ziZ>m4@^52?-l!$^wSNw5S^ah_c=8{kL3XTsV}CodQvRFf!GZ-=meTAFsWI4)b~t^@
z^xb~PpIMzpRm$bN=@}%MZ8X`&pE2c^G0=?5{a#IgOS~*xYuZzsVOvn?Zj<l(H!Sx~
z{Cp*zZJVA^+d_p#NXqSM4CpjX0Z~Snp9Z4I_BD<NcI(koc8Gs;@7hg%P~%`sd8VSK
zfd9!4?PpZ>fcJ988pS^$XBcjeK#(8RxZQgk(t9l#{m0Q>9hB*vY83Vx#48?tVt<E+
zs9ARQvDq1AqDM_c`DvmO&piOo$Vb~HX2P6y(1PL7y4><J8&Gx%;CIE(ex5#GjK#3e
zclFOfKI~HRftP5KUpT4Zph_|xRuGI6lkvONC~}Odzr3ORm)^D=sobQi<yX<4WgF=B
z+&wLzsnbibbli~43uN(*U;8(+!77GSOqbsT;VV#;9%bRP^4l7Pi^@ifl;1h!m?O4F
z&DK2(i}ARG{~e2wJ-w5;!1;(2^SXe&(sb7`Seb?Oj1Sd)v`qH4F3Yit_pvTnu2S}`
zapZDL5#c|wpSSFB8_@4J$o}>vO5e)?@iubAGtJT|)(&Vlqt(JS)+z@E8Gtx(d(!Lw
z`*?;0VI%-b*2_Wuj8X~*F=e_OoW_7lu&q_;?6ZsFEL5?%JLomQK_rJHHkif(7vqiq
zl`h6byuZC7-vGyc*BS>ZQS!M0vHfs$W~H0W@$|FKlJ4GpM;X_((!-lCDK2N-%c1@Y
zg*QdBRq}=CmYzOjatAXm=y#_WA(q2@rCzC@HXdp)(q>EV8U^>N^Qg$oAcQ`)-&k}3
z7-JB>7Wa}g$l>0zwP_NNc)@Bp!ds>Y<CR5pPaJ974#r>%*HJZQ(eNw1^$oD7N>^nj
z_YcU?B{Q&a4!)~(OVILC`9mxeo>Az5=+f8vVR2d*H2qQwTIvq6njX>>*n?nN;6!CM
z-}V;0cf4Nu+cF@ZL<WSYDTDidppPC02}JlH`wAuDP@)2Rcd%1}LCor{z@8+UvP^~~
z8Y}g|UQ6W|>!TL}42}&#VR6WjYoQtrjY&ajLrufU*eb)Ud(^`nJYcvCw|_Bv8e(&V
zzfW!?>dO4Ck!e7PzQwaeP{C{&jtk~P0mF(sgaKAoM)~aM`JFAJW6ZEXIS(d>jR`tx
z(74V`y9(F)@zEi)J+72xBVo=mw#M%6AqYqZ5<<@KCBq6#f#TuAB;5!N*eEBY{t2Gi
z@pfW7gB}Uk2j8rdlYC5ANvI<n6Du|{=n>W&=5eyk3X+?3L{5q29F-X_<5F{6FdeR(
z8mEm`waAjRr}=BwB9!Y|7`piXa=P~lZ4D}w=>;XldF70t1Qr6+QSDbbPBPvaeP0Zo
zqOw(PoRou_kNP?dQenI?2AW)@H)JQQMFDZ+ZjfwnLthr%8}H3RAy&z=6cV#4CVSYy
zout}FtVbYdZwePHRZ}Lcmz*_+3$$u*Tkr(mncv1SI&@VFF083>-&n2)!<MxSiJ;V)
zTyJlwDo9<B80$`Rp4AEJV<c+ax6+#hdv6*vS-LY9Nj`RCLv=n61)yMU09w^*@LKa@
zVmzzP0&Qe1hO888w(>Ho1CNr1)T;(yB4lQXG{!UV9@8{<SwWM{nRY68V^WA{SW%NQ
z0Es8ZA_8TkGzZ(-pEAW8bBCOYwq}1j;&iO^BTz^=(<jU`f-RsX1&F?mrv}c<nQ~T0
z)DaaGO$9kS=7utSm|GLaHq~fSz@<2fHJsy}sg;DrVB&+iG+zJczm&q4*j+dY&J7A+
z0L(U;0g-Z^4~@KO5iweQkD+F&K%XBZz+U5yWxaG0As2)Qfnkb73X2%JFlJJOf1)TC
z1&hMPQ;>^O{{;GYUaXZ%;#oW5QPBPz?C48N`ebG-yexQ|c=cueHUK|0N*Gvsxj$oQ
zAY<qVc3k1lOpCGtomWOrdMLA)&uY0UwZ~0Edw#xLoqA`urNNcha8L-YvDxU^HB+v2
z%2}OvL)#*cp8R&5w>?{fsZI5`-k;H13#|;6V4K_!<AS6>krmctkU!ZDLJ5t6{5hCM
zQ`QXDrmK`D)q1S=jj;}RIh(1d*4qMaiq18IQZumR=Gay>_3E*^B_@iFOO90x=~n**
z9!ike%GB7R|0PsNY|SFVDH`QA8|r{csFSzH>i5Sb7PWGR{j&+>xd><9>0{DjeHKh4
zEx{mA<dBD+WG}wY;^0R4$TY|m>NcWx02(g1d4@H|YqOy}sqm4RHkd6JKoOngQs}z&
z3uJZ(!UBr*w|bL?8tc~Jp2U*Psd0Zf@_iL2H)~^V$U^K0THAV4gyMxJON})1V(<;9
zp9zmB*nfUdYgkitrmha|>jl1;s-&YSbDk*vw|O~&ZWfy0A2?u)EOKtcDueSE2WiWp
zZd6KJOsHF*Ub4jAgan2OAxnLFW^dRIFp4d+#)l3#NtVaRaA{@^Hxy-MMGU`=H)kLl
zvr1MP1;=Ed^sVw%*&YqXR4!ju`?7knAWwE#^Q|$MwSzAa_p>&=t*$fV!!MC_{tOZa
zY6|zza#^2tm-^&!E)cjNqDp&fxojwDhh_#Ohir_UQWBhi^Ggk3@QrNpZ#WX(#m+>&
zyV5gW*2B|yzT6${l!U}W--LX!(|WUg)82x^x{;AZpw*VxTmy<ITF@x%v0n8wTIHUg
zcNyq?CZomgwZrPD{bj4IWgXKV<i1!~d?)wEKL!E$dmxQ;gBn)TA-x}r6KhBvk|=TX
zJ`}v|d3KRU*Cv1U&!HJ&q72NMBM-;tb@@pdq?*bjF^iFTEr|1|f5fYeX+<u<CV9-e
zpnHpYI4dZc>hySgZKc}fiI7YU)G)JfJ~cMWlQE|P>LWds=0pc6Q>hor(?PkxtTrmo
zq{Zj}(ky&aH=jJ~%-;k@Zc(0#w+XUga0`${L|vFF7zzc+^D!YXFtucuLGnUOr~}Ev
zglHDL=&hOVRT7l}CZuERC7bY;YY^{x4UC(YW9Xw%C!kqgN#unxK_+=MxL_=92jErE
z@>(JQ6X_IFm)C7Z$N~bz<PCp=U?c%<RLzq&{Tb9Hg|?wI%UjVIQ5AsJ^0qZIU`ho9
zu)LE9!xP+4N1keC*e;O2{VzM#l1lz=j9Qo|@5c1X5y_pE_hKdMGIjEP5+6;qP*E6L
z_koF<9=*HChw<tG<BpP#Y+(5uFZh#>t@*R|l}~~?U`5j=KD7>;%I%;3un)QKkh1en
zCxV7bgLjtTO!zD~5E>j*C;694Zh~I0VN}N~sdwa>%;z@H{sYS73vU#$BH$VRxlF#a
zMiD7l)lfjZ=2t;VRV0*Czz-}RW<@&w9g`i6e3`GkLA<6f0B)22qy}j1v_iRL2>Q8z
z3b>m#$~QL8(Te%YfBkK1fo>k!xClA$(Mzq_&~}^|lt&|w4r<iqMVs@1qkpH?;kdGF
zg?zVGA*c41u6?@ok!@<X?gq6PJWs*GPl@=ME0|pVJ^x*8O70`<DGnxlR^P8RY%;<Z
za5TuawGQV@XsW~_Z&$nZ2zJ3HsPt;Iu>RegS7=U8_m1inBq%i`Rv)nQb^BU_8dU~^
zUg;ts*`Zbez@~DzbAC{}bx-tk@c-9;<cDe0C>xG0=zZJK=z`b<p58x-dDHig8K=$S
z1NyNKh|$F?*0NNi0T#o7xRd{a^IaDe_uTk5KMBeY8Zn~3?Cd>IKVuPq^Pq52l(s5;
z*UL|9)u0*HB)ixYkrG@ZKl71?R3h&5^Po5KkpYA5>aWm=M=ByeOMX%7uxp_SRg`I`
z?H0p?$;cq;^=cOprTJwV7|R+_FSfgUnS00Z;a>$ejlq2)yGLgTf-^mGvHUv5NP`b^
zNA8IHCcVcyRMaIUzm0JZMxbB4{4Vv&a`P5$Mp4V$ck1<f#FD5!pw+Ws^>kqFd#2yX
zXI3%D_yv0b?(4m*yMBI2^xm-=CuC|HWuG)=f>Gq~^|Eh_xqa?E2ufA#d_RAN=JAa2
z;?re+?^w-BY&^^CwA&MKZpi`u4b^+Enj@WJK`L>rq23{Tpf3nyJvjzLW)?l;eSaSW
zZ)~cPgJMvdag6f?9OVb6K>-utibIX>938-2jFKI&A`h`<$`JC@AYEei<r(B_cTJ>Y
z&`oL3DT5oQOSf7_;j{G<^3AF(bg$jI6ZRVH@;q{430>e0_6q5d+QlU}a%inW%$b@R
z>6yAn*qyRFksLBeuUbVN+glKzUz4qs-cAmJt*kGm_c4GR%cPHOj8TK-u;{D;cel#n
zK^Z;%nZ~zJj)-TlU|~%nYH5ZX=_BC@2d6pC1-e_f%^7~uDn|vmhi9AQ=vsA1w70+w
zicseU`GbuU#Uu-i*pR-lY=LL79Z0uz(l01PJ07pYW@0o-S<o_r>j=%p=x_7L1C1+W
zWq^+vXsjkcgCS&Knuj{^VAw5?L9u-XB44di1_u}FmQWBD17S@FEr2CM(m)AAB0~`a
zHQ{=U|6+A+5_|^b{McY3hCSfE%?247vZ!&24D&&8w=gzmC6LSEJ{8zy;D!-4#HwuL
zaNf!}vf~5Y2#cZ1ALq~LIO4DvsX8inp>IV6@3bBrLx)bzyis73m>TWpG4^$P3xb_^
zUm%_O$2%c7EaHH?WQEuQea}RP#|GDoga$7s1UG1y3IQ}meef3gR-gp9r6zj7DgtcM
z3bQBqm{GS&$xG>Uav~z#qg2lHvJnJDoKkD(r{+W87rYH(5!a+r85wZQfS{9toHNcw
zI2b-B*mS2F$r17bG%|RzZw-PPy-NdoU?5@8Oqy6DX9NLZ852^>*uPOJm}!+r$hLJE
zFeO_7fJrK99m#8g>cl(8lu;|K-&4n(dbnhrN~TK@_~OaZaVO6B*e01u^<_zwlZdyK
z5t79<M^c^0HV-mZ%k@aE)*T_$IfRU09~a<pm|$XK%<zFyW1IbqaZ(#oAC3FW>rzw9
z_n<|JsKuCc6A;1LO39~p7*j$-Yn2qzDq)W*R{%AF4mMvV`rB}40qM1kQH)uY=kpo{
z(GU~|0qAm@r}1PwG#l-6hG8^Rpgt@Yt|>-bfQ~#Vc56j7tEf+o!9xD#DjH-;w2XKj
zUb{9^*^n(a2fL!xf}H8Y<Ss=>1;D&za#jozWqu%0>g4PsV#@Ql-;~hz)OaUT*Nlr4
zE9D$-2#SXmzyK%9AAJzmCpg?_C+G<S9Q|D51*G|8<UG?k5v=|H;vkgsy?rKNQUoEC
zPc8^?0;@QWi`;AFLTh(OzvIWqMIpc-msz?1_&rXDbsz*0^d)j}@F0vidbxO027L+i
zdh!OiOMLXEtl=B%P0e$YOVe03s`U&RDwla{^hwYsMo(&hLc}UA_irFVHI$9HPglgE
z^+ghQZAPy2cg$d*)Nhlk{23U8urXs^<m&j2g=(NQT@$=e*~I%ZueDFeJG}Xv7mCVt
z(KLepIJ}L3MXvWDhuRFlQdytIIWH`@!N1XR)A)u4^`DHupmwwT*}5HxRQ^%fNphn%
zXD2M9Nm$KA)5KtptdW~+>rJYWo4sM}2Jf(d&n?j~82wiNeYWDn^ct*`zj*7AcdEs2
z1V>RKlgcY`ir*GA*>W>v`pt5CusQr0S-HdhMTrf0c&E2-D{B=XT7rhSC<xZ|v=}0R
zWiZbs3}Jd=2zG1ECshb#h8@ifO_MMhN)TFL3uUIsoi4q)$SnKop$KlB9X*Z7=WznI
zdQS%TYL!`;lL)|EL`Im-mbvK}!E+quz!c`i27_|K*m0K5Q=MPt`!Au_uZ4JGNYFZ2
zkiOA8Tm|3kXj&NC2^Ej(fZ$ZhqV%p^a*}L1S!~PI-ZFT&v<0m+f}P91T$aQn$3yxz
z%F_5}78T_iWtsgGkTkTQT3POEOACYNF%2EhmOzuUWrhE4=8Tnn;@+Ygaz0S|l{OgY
z6t%J{_yG4cLrs;{A*hN9>NLojG|T!-0myTsto7ziPsJgP1+)|?cDPp7B`TykfiBS3
zhhXemi}Y+1b*z&OZb2L2DNt#Kt=BLRrUOs{Es%}zAuiypfNBF}W|K{#OP3y9<Sy?~
zEA4PMLU7V>YL&a~twGT7WwX;Azql2>RJIsp5Af8@_B3?Hh|tcZu|#&dC$ZCVO&(u_
z6YpMsW<`b6c$cj~9QbtT-2*bh0a0&4YqRA(W9&``A0YSJ20u0<52OWSWOj)o!|C>5
zEDGwHLpi9BheBgO!3_<|QJdhewrl$j88rZ<(;oJ@Sh+cRO6Vz`3!(Ig)7-*xXkyrf
z<TQKeyM>&<W_dJ-swrsH=czX_zU8ry6<zyu?IVxJj1`8_4X(nEH(~PteZqeyo}jRx
zrUrE6bw3%~!8CYU5Au}D9|pfJlBc6Tl&D-S&!jb4g0m?yzxma(_8v5#ChU^u;yuzG
zGe<>v=ceb=%orL?smKERv`t=!xp<VqM(iwV#*F1g`6qcX_GJZbN2!*Vydj36QN6`Q
zLX<Qh@yo#)P<s%Cf<M>FEAeq2^H?LVntH-*S1qskglAJ6%ImQysdGTCyb-MdKPFk{
zU`xCiL^+lg<XbUW`cg0%9;bBjEt9wXTg&D#75tshjJlzq)8Bl~G^uglu$K3>LI~=Y
zl)uL$u@9&VYy!GP{dtzW>rHdpX-p?3`CbgfZcayXI7<@F$8LGw-}ZoJ1}Dh}i8mPV
zOvnag#5qoslnLh_dTT(ZA?apA#nwnp*xVoamxi{1w=|w4)tHx${Y#dDAyRIJ82Ry&
zSYwuc!0NLs0z8P|Zk3essV!3uIW)wL{}Gea%pW$lQT}P07)q&p<_pJ|;uZ2Q-&t5j
zw5rwwkk57CJO+S-Ui-Oi{3!$E3vWE^-$gjtzfANdwxTH+pawt<`6`XP<koabe6R#^
zcI4kddyPXfFp&=Nweq!pITw<Z)L?R;o#rNA|6>!%WHRzibnhr6ifm#KGvR#rZxYoS
zbER0M!9W%%rW;(O&e2DN1mZ&K@6;)-KtNru`T1_0A*Qw;4G#|4pGup_Hg$oQ)n;J!
z!G!u=odTUKs7Ze#4i6+}M^`7n0&kXW?VtI2hC<ZKc6EkX(r3k3Yr8O-y65(F2D(}T
zHzgL&8J?m%B{#?pbz8Ruwo;8tPrv{_s8jS7rzMQrA6g$6Ks8Q7+0pxe3;|rE#~-Lm
zS@a{DH}RQTdgFiWecD-|IvP(0_4S_z?Rlqoe?((rW^zV;;w>9()PO_FyuD&)Z;8Fp
zhGGmhSJ7-N0GCE7js|;A=P0|>1%PDt5@IGl^Jajd=@!)xaZH1;7L3&Wd5}+?c&!k`
z%)}xdwrl*N6=$vNEA8I7@(ZI5oT02!D!UmS+gpJ7A-dB3g(!F4a`UUYt#)a5rh?h7
zjhS-Rh-{YKW91S4pMg80L4IxYAJ}*3=mGMZIz?r1%gJwjn$)%#qQuK?e&^5BU;zPm
zjP2p0)x1|F3SkZcl6xi!EFc_0d=bA>UACa^y}UJs7x3s?fMwb{*jIzFXczEaHp?(1
zG^CPO_Ni0cZD`+qvTx8}P`-A&?Dwq!TKM+<{tSo6J&0b#IXhPli1iwoDZ=t)RC%Xh
zkjflXH6&)oycAk9&wwFKzmHBZBSx<+fLr6B#C~a$VaD@(O=ru&X-#pjEq-CDfWf|<
zqu-`___6?Frj|;4g4&ZQI9*ap9jR=jP@8o1UvynSDIUH+Suwh9vFIq|GoF?HrF)&a
z6;Mb2eCgrem1e!@u`h?(OkoJKq@d<3OV1#9L%uR2y-L2-8htQYxmngpZ~r9>B*H@6
zGIOMltzHkfXAZN*`r*z1?BT<0P{2s5<%sA8=3=89=_JSU>j6zf^(gPa7|$!{6X%qp
z(*kKEvdZ>9_*7b|k!t~O7h~W0CQ_?x!o46|PUMvWO}PM5=$9DS9AK7;#8`Mk=gPbv
zN;TdBOX<<y8JJ2iQwyjl*vW5@rRi(~Qp*Yd$Gc0=ipJ{8z}N;VHLX4gxEbmXjV1yk
zNV1nfP67BqD2X;>9QboBbhyEZ0*Vd6M%^Ss{O#G$y>RZ8=-$U9IwU<BXElpnqc?2v
zWBqqJ!I<YlzZEwvH#P6p8ye@tmWwv8&&n`c-%7}&;XV$Jrqr>G@MrMJXf!X*Z5e6L
z+FP&}8;ecMUy|cYF0oj6QASyty%A74I^M<JMU<{y{4pllbVYF$v%dWq^h?AZagCPA
z*hKe*DhvK+Nc9QPc3q|dkT!0P&B}@14qG5D3)53hiWygCvp76Yj{Q}E<Fp|+zFAHQ
z?%%u)Y{0yOY50_3!Q+D18YYju3Z*Zy+`w9LYKSHT$nkbhWbT|6GXZZZst3dC^ytzq
ze!@jyXN1HDk_(8ArYy+VEE)gWPCDviZBp*T!@tycU$6O^Gl9fcME4z~M=IkHs*>k%
zrv^@!tnJi6BZrKT@j;8x8RRrbRcu^1Vj877hzBK~o<AC!Ie$j;nh+OjuuQP&<Zy8e
zqm!`WHFfHlpvNCUM^M3QqbURvAzBX}lVwt8oq_8&VBgm#(z0R+IwJ^CB0BTl0#A0v
z)u8B{6r2)TumU@kg%IuBn3#r%!qP6Ri~h`F+)SJ`Bw8Y3O|2PT9fl&^(&*#iB9n70
zV56gKUy~7bNIk+JM>pcWM8K%&6Ua~uk!frt;fpdk2DN!UjV^5YSu!Qo2Z1OJFkf+a
zHQU~)f-kUM&Wsa=H#}tP<Sg$)jZF-TSc-udcDkNj=a{voQeZuJUDVXnF_UJLwsMX?
zGdw{9>g12X_^RGADuEubwZJ#eHJ(8dUrx@mztf{3=ldG6j^GNpAW@Ss1%#mK>tw9|
zg~3>bCcsn|8P}n<x~7&~Z1SOEQlnfF)2*T+TPv5w$U#hcN?6YJvLFE@ny!q3kanQ*
z;-$;|8;kog?f(jY#yK$|7}kSa>Ca}X^{Zv#<*M`ymnVR*ZLln^_GdV&)QO?V-09g8
z)~bV4UL)70Zy~`zHz<vAovCDpA0*e?s^Sv5OtswLgferEs~CA4f3ls^-U8zAc3}Cl
zx60Z;v`bi*^v38NX-yY^D(w9xqwt7+#}MDR*(eO;W?uXHPI61h!tfB_6&H4x-Ri%n
zJ#2eimmDBzh=xKw3r+Jj+vLMg`>x8uCU1{*(9#rh5d%j{g1D4%(;dbqwRwPaccuwz
zWIdEC1|Gmvl9uSEHk+wwkZFGYlvl`fqut;pnc+Rz8AfLwi=Oe%R`^F&%S@wfF3Y=`
zW(C2}2?<8+YIb^tqf<LxA+0X8z|-6%07d5bTghhyu{75QPG(|g@;op$BtBSrUV3{S
zyp`Ah*v<2Ophz9y)&U+$b;Bmx2Gw<e4V3w!jk2(0QRuPa#bCN;WRbnrh>NAmWpOah
zs9d?U#Xl<ZIay-=z{SOw_R^Rlczcs+Ewap>(JlmOAPiu4ERU5?Je;}^JQ3~laOSSC
zm55OPmBvs5QDGMneO1V*0Zqtrs?6fBSe@2C<?LYm9d{M1iM<R8=TWj}ZQWKc(Tt-M
zivV37q+}?9fqFOxs90nC8)co3GR{339^58K<2J5pZ^3=M8=?nX`vpI?Y>Yv&T8<t*
zvMG^;u5^Rw?g|oUZ81}C_~#P2+gNUBvuw6*w#>(>ZV96Ep0xJ(SaD0g7YNHeOYZTm
zELVh5c7@#Q&kV^|1W~ruZMEnGT&{rdbSgni1->sPQvt6VOzZpoQyzE(=Ws`{JP@7x
z>C0P9SV{MMc`)XQ4Ty@E>N0pJS{zwK4qv_eHOMoZ#VRK={PSUNkeC(2VGwKtA~Cq?
zk=U1wNK~qoM?+;nI1KNOZ;;1~D&}~2-2RbKm+cer>3}8{9Dgzm4ch{oi8wKNDvqfF
zQ@N_4d>|C?X=f1bSt#BRD2#-pz?f6wc~mE;JYyZ^fJ0^ESz9bXP8oU5dn301k<%07
z`8xL@muOe0{TSqd$%uMf@}kqix;LI=q}qIWDHxzUi&X5OVy@18*78cs5T>*Gp5d=~
z)myXcKFynMy2)$N00Ww{@YWUOb${8^RExegknfF>p`&6YZiOH~@TLuD6zJO^Z$<BV
z*#}3zP~MK^LP@boURNgX_**EwVCN-toWI4XJTg;TCx8DRgP`?ec+I=Eprh%6hj-yU
zZ%RW}z%Hn<A@BRg7R(SNB%t&`bkdJ`gdf_!GEImwZIq9KQ&2bw%FV|%SwzXN*>XN`
zcm4nJiLGv?p)f^0^{zF725d1OsOmr5-Xig$pth+;bi&Z(MP&1z@g=r~m#>n~f&);t
zilX*ku`@FT-0+JeeS=*-PlAr7WU0<&yYh=*j|N<@^kpIraYTAwm&jNC46z%qW~foE
zl7Gih0Yo<<!Wa)iA;qHHU#D+yCg2${<^LEZ;N-~5H}*TyHc_{pce5km_`lw!Nln_U
zj4fsCigAV1D^7?hP<dB4N4`^UNT9JEwPMDzvuy5n>lNw4>5)Oc!#4h2M5)p*K#%NN
z`JT5*G>vK8eD?i%4^7fcDRB6<{y9-0?rL}furSNPs@qvNkYjUk-fmy-2<b=^8gG&v
zV#J$KrVgLsqF(tyj6`oQxA9Onn<+nxUbR3kB+8EU2B;cAqC5?q_muZ^_rhNwKS~n-
zqu>`N6#h7Myc#1Vg11g~@@MKK#<z48|0KrCBGOseIeoT)wg{qua3!}tgC?vNI7hlm
zy`ik&Q0Pk<o&;W5s<{Nb9qVWQO)K0caJB+oTPHuauF8Q3Rsw6<wO+wv0EkUUacPua
z#ByaCn+vjAy@j6UkTw0wdIyZlfEi-r;Yx1%6^H1-vU>~}Y)A76m)gm%gOC-~SbH@7
zn{SO5+yvRG@RT1+zpZx+mDMG)4ZH~VcfL$&Xek4-QZ$n3WRD;Sj!1fjX5zr#Gd<(w
zaa2golfArCTRpRKV2F`Hc`C@>(Ww?+*7IziU^94K7(&Jqhb*^N_O&S&8SAy5H*8y(
zhAo+qz`g1tvcIz=9O&?S9Z<j3Zg=X@wTpC0<7Dl9Uhp_e4)kZ?h+&Aq?|mSe%)u=W
zo-;voPz+JwC2mX}yaYB6_FrQAn!A^GYIKfX!L{s_$%u#evsOS~3?EzqDt7S}OcZi@
z2qIMR*EP7si-mU!5}<!N3YlaFZ#&3EnYLKETkGvD3Q^QH!$ToGymcz6Gz=NU!trmE
zLxTe9=4AToYUvp}1G$m#2K0(o4A;kJi=}tGQfNY5Q0Y^z$na>WDu>x$8*y{);r^jf
z3cC)RafClJp=I1~WKd57z^q~izgaIw#kk?jC62*FUjqS;b}B@X4O;B{h6$mp5c-GM
z2jd}OYmn>G*E=RMLBA;y)TCdqDHJgVpwqM9d5?&aoF`?8)XY5&d8|G#wSfA^@W7`$
z31(u^FbKr?4Dc^GB>Klz`S1X^W5H&~Kp&0OX|t=w21Rcen^pB@_M%$PP^dP@VDHU@
z7&f0SIicMlG1-RdCRU)7W8wj-vt}nlY#f`~=13VRt+lz;?+BP*hNfkKlK><njxsF7
z(oni`hx;4u&4l%~R7RxvFV>g_jl*G@kc{+S42cEJv58DOJucX}qDYKMM)}B8BhVcW
z-9>%dM1`iyXm5$DqdxpJ<}v<DD?^e%<Av~N+m<;#{Q}bs>xL@m*eJ|{80){-HnX?@
z_1Vb@A!-2#LFI;+KG8=;jfFK7SwooXNkE-ze3G|izA9R{p5!1W#~h(%!hn)fqHU)2
z!<1`~aUlS!puc7i{g9PYZR6~}PiHyJ{yJ_iTm>E*4JF(#rZapMDESEDF>5hnQZ7@C
z#g=jtmZL_D7$X&F(Q)`<(IF`DC+|1Xze{DkdYuMlB%21z60KPo?^DA>VRZxT!uOg6
zB30IYO%`TwwF_HbO}0{UJ~Y_QPN;=4A+Z&2?825pC|gZpr1=~b^oeCs>(9c8j%#qG
z&Rd^>tvFZe(^oWEEm$ib;{B+jj*@~;mZ=lG9E{}>_59GizDy>@6o=(2vp}+n)_mvA
z<M4mdU=;7st(P=<Pg+zQ(s?GT>^8*$mSaD$WV1|)9X7Fvk>y31Y@gwpuZCs>W5U<b
zEK_2DwfQD#_IAxYnF|GDwVavQ5!H9FDt7G<>lvM8pH7{6)>-4^?C8D{dC^*PU}{=g
zaC}-?U0`Bja*oZmNh5W^H-EI|iU`q|CFcg?Y7jG}Dm{e6gU|C{v^En6htmaHifQfV
z`{)@L#b5@y4=xBDjp72tE&&t3W!&(92;@Q=3W{yvhEDL-ML{_zr*x*v#Xc9~X%2@e
zA9}d*K#)sf<r;W#Q@LDfTcQp~2CePOe9XG_(I*5r7ak7p%WY9vkNFB8s>Rj(KdF;f
z24SF+68<4VNF%9Lu1bW_Vo3_CvlHZYCw5t{O4uBj^w*>znP!5V0knf_R^;0Ht%{`B
z2+Ub%Q5xh`T?S~4oc(&a&bHd&dmVLzTpyg&KU*$0c=LP>!V<9?%q8MM@TU;j0Be!v
z9fT`KY%rMMMkCw+f*0V}o6?eUlhbmTBRBgqLxTWNVCip3Gi3C+di>;88(SX#k-r3Y
z4&yDM_1QD!HtV=QVk2-TYopw59kYxamQ3z2xih&&?u=>G=c}Z}`=N14JdGZoX+Bl@
z->e*vN?(j>x^oS8x?N^B4U(R!9b>ZB%ZwP6I&>zZq|Z!mpY&!hCkyw?G8Q>t|Gi~)
zz2ol)#i&EFwgQW-LBcUO7U0L9WAol}E%w;lv<&8R<h>iV4D<XKt|=Yq)Kaln&}HZQ
zFQxh`m$6M2bWj;H<!a29g=zMi5O-rdV;gl*%x-cv%hb#US?urf?AH7$2w{toX|t9A
zOk)e?kcWozXGxqtx%vq#HHm$>G<c_)qHCFN80wU&)Tq&DrS=$DX9w+aBL^Z%YGs8_
zfF>%VwA!%uSEgq(I03EiwaV5U37J5#R>!Pyda3|~SXpBO!JV@;VA{1wuG7SWx~<j$
z#1Oh5+O;mk3CvDpQe<U)oR|Y^nJly+QJnEt76_rHh-{1rBi^s4Y_h2g8H1WUcLjBk
zb_yW_ljv@rXd|$W%`qu@y>Zg8aN3rjKJ+5O%=Ao~>DeA_;f7)cAy>*h{w@V3W7rwG
zVC1~F-q9|s6r59<Fwr1e)B6^*v?#d%?u%1Eg)+shC0im50<haMxj*{Ovz!n%qdZ{z
zUN&@qJZMwO7vPeAD5iuVv49bR>*cSh+m46_ljsjew^X9>Hj}nXGMVL(V6Fx@#ZXop
zLi$mobrHpr;A4I)w*4rio;@D_sOg`PC(^DiMTAmBn>^_Q)jduMg9Dz5;iw=~B?G^7
zc=exl=>vnf3gmw#vD`X}88|fO%d;_t9=N37xis;Pd{Jt5Yvg(Vg?pJg26bh_*<J`P
z1k9GJhsE_`2o7vcURU#y5dd_W3A^)U9|N~Oy*dnGcqRCDOuh->`>zHs)Mk(yO`kHe
zkY4K;0qhD?2f{H+6L5pP9-kxf2zL>_VKbU~8uf=agKCgR<W6dhZ`E&gEW4PmVx0%*
zr7)A;P94y#$}|p!Me+!I#~U%!omSRD`I|q3IKvr4Q9M`v9s?f_Rno*L+I=_ZftxIu
zrT$)td|{XXJ@qu?X}__GuW1<X#}>t{4ddklJ2wtIOFr~b5~e}HP5^^+OVBCxQ5w-=
z9tBVa=g7y2K^R4<$R9{Yb>0BGPy8F){)!EW1qFFN4RQ|40i%$A#7?SXE_wkI`lk)%
zl+N;53=)+=fkfK2D(6oB-2T4hBIpEPqzQM_l!B$Ckecx~5AZKjx7kXVZD0Aw0Pce_
zG+Nimzx^3%0YN5fQFi&-=@BOA8XyBKZLvZA6O1_o_zuzJZ-PLGIz`1YG`Ug!o0bfj
zBy?(U*+lYNRaY^0I9ZiR%`BeH;TQx*OBxl@<U4r<txz_sDBrcex3`=yM7GI0ZmEz1
zvmzJB_wt5Y;<^vTS%GU~u&i(MeSaGkF1Dt6US!+!_8RC{tg!=U9yJ8n&fjM24J<#L
zdvj!adj{lFy%UH+F31jfkBfwSU<q}QCCvkVX!$`t9PeCPT<OEf|IoYTw22(2n19*P
zn=3uS881rZjX@qi@-OvZP|juuB@h1}$Mirl>TT;G33kdWYI=O363S7f{KRN?LIyH(
z=b+=TM)|3KN^^<65mx2iCGW^xwzdg(H_(RsS?YmA<=7X<&pSTF0Z6!YmF()zEG2O^
zu&J5yi@bxz$3jPt-J)?g4jQsW+N{5{q2j)1EyVvTdmn_Kjk)a}E$S`1R@?R0dBbqc
z;?X4r;dzW(WUW!<H)(<S3&#wQ1OUJ+`E7LA*i?X>@H>A|^=>r8C?Y#I_vly>4e%q6
zQcto=YFlN`d|=v$T&e?^_lgI{H9=>EF1&YMP2nR>mwk-th>fe4eeLhf@F>WBsdonW
zxmQpDy8Y8H9R`pIqXY5^i}gf>(oQy)vAJR+qxcW>Z`?@+{|7EgQ~~z;{8qy(#~n0I
z4sxoJQq0?)iJY2)eX^8o@D~Cg;#ORwfx6_f&LzZfDS%a9q7q>f&bmXq>!g-~-fVxK
zbcwF}R#iz?@7i2NggF;UH-E<Fh{Z51rh9N{)0xsEufX<}Yq7(wMcnwI{_ksWPF{_j
z-!pdF5GbNZP}iN-%NvE~6J`i7Q6Qwfjd1z$y!5f9;$7?Z{$bWAgVx~CY?Q-0XoQ&%
zi&!j2lsqGK82)^uHAHiO<v}ausI&&yO-ceI@mx7N<_YD17#uHC{DYH~mp!sct!)N)
zgt136;4khQze7`?TCtkCfd&)kfhqmGP0~|?ivWd``7?SEcz`kEw|{Ir4QkT_=K*QO
za71vRoF{l;t_(EdXU@cQ2Bi+5f#Y{Nf(9EM%Lb2-A&D)_gG3aw1elj&^6qNDcEF8r
zIN^@<FD#~#W+pBea_0*)9~o*L!mh|@Nl&J94U5sEUKLLgjB9wj)_~wXmcYWZ5wV7o
zkoS*ZzmYc19^HG$acQz<+%oCOcKIlupaQ@k1M8=a4xt4Ire4PQh%MWLnjn4RtJEzq
z6L}fOrxuvAWAq*y&k!O_-G~<N2{An4EihP5^moWeR4Q0y?j(Q47$nAr!U~m>z5j1V
z;HR46a!Sb>w8A>WBB3>^*Jf3Zi`fiEe3hIUvw@^TWKIX=ogSx!c&w;Jx|E#$Ey9_W
zqX`&tM*3wPlcmTx7%-PGV{~VpY`OPFy0QW>eaOaYCLAh^)o}mi{!*D}#<GL9I@uUa
zCDPbXJJMPn-G8W5#V-wTXg5oB5CH;OFO`HfS-MEMApY=-MuboB(HqWD64Wa-J`4_%
zGeX*d)TW_9R$^!yh$IG60u8HR9(q>lZFt${tmJKCxmu`s1wXND;*2rZ;1w`_$rG9A
zedz55IR7?587cZR2IxX#=IxQvQ1T3CT07(R8)NqBGN2iQ9u)3Y)k>P;%tT<CRnDeH
z%K8~}BGBl70Fg;S6K#deM@*MD@=f+WF<wYhFc{RTDaKYsX!u~OW`Bc{7J9Y<`Dey?
z2x|ofOog14c#ouIh!k%b2Cjl8nKga3NsERiUYCdq>>6cinyeKJz%G?@f}fG7I7$9!
z|7x)J&y8UyUP0A%g`Af-1vuq0Ip1F)R*EgA7n@v=y4k4Vs7Mfz3lo1+SL4x40n3cD
zi~Mc98bQ}^fn4lwW2>v9S20&brZ4fgxdrHbgL*2L8h7Ag%3gAre+eML^CNh2@({c{
zt!in)7t5IG+Q}8(GCglNJTTcdxiXQM>#CvUz7Lci<f<6E9yq$6<!XNhiwC03gw8eb
z42KB$gW0{;rf16(lw|nTb^eTgVrri(iE+Ix3BpWhq~8$pgJ!`{0_M(Wu{-&bj}@c9
zfd^IF6s!bQ{@L3`)`Q}{ZE)P&=&eN}CKXqk#Z9ryge8k|bEqA|`i_!Y{3{jjTrqO7
z+Q+wgAC|;Ndd-x-L?4PdQQ${#;%)wm7Qe(=qJD*%2(#t(paXUdONP#eVseLxgWf&+
z$ekS`KsTKsxK<B<n1WuZDlPsNEjS*=*c14ZXBBrzz+=-A*>F~=)xw%E$>x*k##|$+
zb9tHJEy4)ZIt8$1$UO8OVxTmRnF}{FNY|Jrd^RgcH+;kiGCTE_l6~-PM`T{B|IUc$
z^EqbDucYlc>HCgjf?|&sUoD0>o14BTlExjY;UGk#Q0;e}%yYsrd7Z^3766->pPo&F
zp$?gVIV`YQS21*MVJs7)WI$UjvyPdz$)e~DI+WMG!I;-j0ukbu#Ua#zcM!dR&DItx
z3z=p?GV^3f;t<Um0xZTsvxBAnHc7;K^3<)DrDxOB0A&#I@>rM-g9BXS$D{huX}GTl
z)fO%a#9`Wnx+@b!DMcwNm?@%GDD1ddvMOe;mkrb3so8dDZgtEal(zxXUz^}pSz|OD
zlSMVxwP}hBxHII7D>hfw`7gGapvS@}q#m(I)~AWmb))gtvLQyJ2TcehVyqkeZ3Im~
z?AUp~$)90!0y2b10Qv3;?gY$-I~eXxG%4n*a|oQ}B5h8iWw8Q^LUmpMVA|p;Lt{qT
zqj7`uaKg@#dz>VS*-$co_GCgTax5KjuQzNdH15Dq18r+e7AFY5!PVJvUm`FPpwXJA
z=0on2<=4yo{tf>HX^DPj$pbbYO{(A2X$t|EMe?A(s2~|$Tn5nPA%DiWAgjQKVwv8d
z1wPO^WiSjA<E_T>!~V@&s4+Zc+n`TA67(!!vuEVdv_Nz<5O@iP^JDR1844TL%i~7L
zLT-vY5hwd(RBWM;f71I=)c}nwfl2sOA_msh))xl}GL<nl@T@#--|%KCxbRS}1?@az
zYFa~neB<N{0M2JaZ5xGTegJ&Wd7C6Hh>kx2)M3c5bHOWf<@vBTkQNHgK#MQLjz&yh
zrl6_+FZzrP*W%}K#lTC!M%cHsa(RlAsB`3HqXo>%NzA-_#kCIeM7<ib$NnI6&BOaO
z6QX5fP8vBtUJrT!Ek@CuMtLJ1AaWWpv54Swp*?cIaCysq%vUzo;|?zr+<Du6MD46v
zc_;NnK$AyftNhKM=`rnx_21(?jmjVhxm@0jUPooivlSH?dC$qIMhljWu~eGxiNB)=
zmAr4$gk{Sk^#f-B_~JfH+k%lKWM^Jh$7KGG(yDjV+u-jWuOc7&xTw-X0^^ri@`<fl
z0bT>I=k#f;8oj%?vJMFc|FE%~w&ncIi%ye&rfwFfzuy8-pfh9q)ImNA-YrMcRi*r^
zV_`8~z_d{5%#ZbXjK?zRu;~}e7yh={sHO;;lJTXH;LvWp<tqbs@<Zj{HnR2>q(E27
z*WMu6fIuc<+~hyb)~I`}wVA$2i?f(p8QRq>`ERr`7-?rB1;x!!|Bi)5zGMGFYb4iq
z3kH9Jti*4y+1g~ALcoz=2`mUvT*?<Zr`E$I`MrX{qpY9;VT-Y~y2$;$|ALAP2v^d}
zu<#3}F57yW2sQ;>S@f1B+&XSx{e8RiyWOg%S7Ru4&IMm6+uNv7#|0&+WCwqn2Mz;R
z|EDSV10RRO2Y&Yx-N!%l`7%&e{r_-6?O5<=$C0SAg5<;>72L^%3w7#H(aK7H{J5a7
z7rbrQvpW?O6rF_p4qjmMlW3s~@G?AnJ9{q}8N3s4&A_husnh*}%`>s}b_q&Bz@DqA
zCTeNoi27N<FiEqy$Tkqbib-+}SQ>sMa;Se^usq$15Dm9$%$622)`%F4Wm^9t^+JUT
zkPwsAHRfSAn+?j<!ISe#n+<Rhb>aNV+RoL17Q5Tu+gnBrl3)9*Obwtixlw*oPyjJs
zT`Rx!pK(^JhY_A-`JE4-bcx7h^JEVr8HFf<IvZrqG>HwWyHS%MEW1qh@?WSUc;#NO
zInBnsV|rQ=!Qz|9E(>HI8`kJzh3xAu84Xz)f;JEBjs4OpS?o2gNRa)5&-xN0IKbaf
z9z>_S@Vb)?9CZz?mhXY_#g<#p@$c;m!29$bvjSza9ONS~4`d0}EC+l06dzjj3$RV+
zG+nyzRIpGe;utt2_!1ZSF?t+mBMzl59q-eU!l6h>u>^$cYHJK3x305vvxThZ6%XBG
zy$9x|K%n)A8H^e*dW;-u|3Zub6h-Ojt<xq$pc+)OuwQybBLl_`7(PaN+h;I~7b6p>
zPeIY@L(Y=Jyn%o!7Qg@^nz&95k8Y1ac$plL1VBQ=I1O-tnrCukA$$q-m3qfwgB%s>
z2XPyUz|r1}0bGhMf$v#P#(W=t@ZU*MYE@d!8#JTY0|h|(#=gQ4*Dq0);S=0_+&v7b
zDKqveK;`N0gYaVyh}jw$;2$zZm=H%trI3M6rCF$Ax=IG6<>%dj6czwP{ae>=Kq+e1
zbP^7+PJ4IjBFFfL$`|+$fWI6YGEvPD95iZ|+QEt9W->H2yn^_T27Z9F*ha(ROO@m-
z9p`p1`3^iiJQfO(g7q@O`vyG}%iu;FnVunCP$@ZGj`L@<#etgh1d~zzjJ*>9Mqq)A
zPTy{XwX4vnj0sLhQcVWFUrP!+J~$5PtEi(s*4VqfrSIteazgy0ThHEdqEi`~3LK>;
zc?XQFp!Kl;*~}-W6`BdV*Vah|gq#wcls8pYHp@8sm{>l{I0U#e4$>B45^ROjf;7W`
zOUdcpYJvt-F^5?vXT&5uIE}dm8Gpu`Xb`@RWm_iY1@{9Xuo-F+F2#uleJwCkDy{Ph
zRLak$X)=U=e3~Xzb_K(wD!PD~&WdY@PL^sPB-enBx+M^*id;%AXxP0=FPUJCA75X~
zJ0fcmz32jCE(wun-bqj!8xHPvhCQ;Lk0}cjLaU`NW)3A#O#%qsdOzaKtH33PneZ*h
z3ITtT_b-w6#O<jtV4<KsfHK^@G9Laf<YY|@xq!;yKx=5S7mY-F?X{OQBmx*df<L9v
z=fap~f@n4)IwP9?$5@e*E0g@0S(dbiE|AHIgbShM0#9+<d96%|!&@bn`J(MWCZQE5
zMf^~+2^kj5mosg!g$m?f2mL4KCdk<_N=s71revxJGCE|c|Dv1Ip;;^@IQ^VpmVQjy
z0ek+@UuBFaGu@ELZxfbtec2Q}gFd#5AwHOhT`oFD&WolhGD!FbQO}RnK>Y_|D-4vq
zz@+?`X1OqiJ~}f+E=nY$kC&d(nR2ncP?^CkA#zEg^K_&a@Lsq1a%uW#A$AQeS_C?9
zSqxv}`pKkRFZ!S^k9{|^86xQle=mXyI^0iJdaJxym&gXLIh3o?+YGSK3W!dWt8FdL
zgn@cZ8jj8Zq1!9vTK|x`Bg+VE0=~<j$r*B8bOQ=QhUpbG*V`hux4?CAgYQ)0#0%w5
z_6vMiS^2ZK$;Ct=EY;A6{>+jaOMao2OqfrD4CN;OMKw7*n+EWWC-V}yIrLCCT&URH
z;^T7n9c1tdxz)$gf)j9t{Kea%8N(<S*gNb6bGj`^JcRD&+r6dMc0?m#P0s@(-Vve_
zVU@@lyVIuCw-V(FS`z!<P7V6crrG*uVB}7b>CyDCJQSE2-Zaliv;W#;rmaNJ9$jUY
zZTa>V81YQ(!)@8|IW(?0(i(Ht!0-l{V?8yMqqvdG_307Qp!A_v%}6ybk(Y-%$xjdk
zx~t4ji@5=&3XB{OV?i8|Py*=Sq}pIEAw{8Np|?gQ!fN_4NR7t+qO=4wQn-M#WU);k
zgL~?ML$sy&ldZtR;O`~z6&8dwCf6dDVQK1vg*7>@K)RL{JU)b1dKJgp#dgb$4)usI
zsgo5>ex#x|%gX2w0-7!~_$g#nG>&I<GOt#7(TdZ5wG9^2odZy0O;8)!XCp#u*Lqt4
z8Znc2U91nH%!{%<O=S@;;b5-2YzWSwCYzOwvBW6TT*;_P*<??6{c)ih1*jY3u5YaX
z{MJ~t5@S;Cj_p>D1Z$fuHG?s-CHhwY7lIN%aeJEW24Hl2#)4fq2;`oSi(nNmV*_e1
zhKnq+Q110FOF&Zpa;yJh?oxs&nko-J!+mi?A&(CQxri~|Z~T2c%3M6)^yRK!E)Uu?
zVA^Noq0oW=H-Qd}v4a=&uO%ZS6D|N4N7DzosaBkN*vDb<V+NubD?SqAU`h63rWvaL
zqfQ)7j*cn1Tp#lx)Ys?aap%-CG3cB;5xcsOo0R7jnNOxgAt&*`rv_(S#cRT!N_6HC
zZ%6{WS|Cqb!-)7~2<$Wdo(%(ks%}o`>shA`&L+T(&qaGsqli2(C7Q8yGvxU+l2n`<
zBe7nYH}Hag$zmeZZr9;^^m7mHyYVkw5r7C_a0JT%Tkh9OJ_5$EO_!I0bAjHWX8kMP
zEZLvNS4S$|tHFcadUluBTrIGRN;C_UYX%nP%BjtcxxU^p2X(Dja6f~2G_YFUu#up&
zT@7geX6l-WByid&T!^yK@|L$lbpgLXaWc;m>Qf>pq)p!TMaP+f5DTr?A@5k*kgv2U
z|K{W6dO)4iEDTRYDw*QnO%`ln@XR$(a^FqNWC>{8Pw)9NidLdT%&>dkM_%IEA=iK_
zK1eOl9YJTxa{17oxnCUuc!7MBS};H7T1YG6B{Sq>W8SHqjvObSBs$E;YLn6o;ASv0
z1^7n!)HdNrMo*S>OqODmuYU%)k2xM<<g?)YY-JPE1ms`#^&tlzBA<s)9EMv)aO{5J
z!($Q<BIR*p5|~>H-Sf+6bIY~J#Qn<O<EcTdS*=}G%#?p8R`AFl>`tAawliD4HZ~le
z=RHT2@}C%7gI~JzO(Odukex9jZSvp3R<Du4Gfd+-zz~RPM9=i%QAw~kkGUcfRnPWn
zQIuHUnP{LOH?g`ku#Ffg#0ncowUzHqR3H(TwNxO~k0F}1vdu)bX8Q78Nkz-QH&M}|
zA!TFa`}WsiLq^NC6BWxElZ7g`-9*RYs(2PO$oAGOb^sC)v2O5>$PPXf(^;s8X{fy(
z$&3^HK@0^&oLN|$g?oSK?L<Hkb9{H4xU~x}=+S&kN>XZI(BP)2d|*vMo(qher0PKY
zld^5K{K$xd^mv5caG3IA9}#oDnHoVc5ZW2iLS&~HQKr70Y6|z*PbMlVTP$amq3rDA
z<BrpNSil57wKp)PHpFl25|m*<65NsXvxy!a4O-(OLFQ31Cg<lq6bi?vV3Y_2_;^>N
z`<|UT%P;J&Wp$OZ+e8I7aX~dk_DlN*%I%?e^RFfv*lP|9y;q4O!Lhcxtu7TaAYJ(G
z*NMg`F$=2F<@%e63Y50DlxqmkZ>?#dMANF|clH~QvSxn1$3#bX#v&sJ!Q{}w_w>n7
zUR&e<OU9;X1lh|6XR@2%I_w+hrn6*kn_IRvFZ(zdk+ut~d*2u&*oH?4oN=<B4Wz&M
zNMwKeHdlY99N+`M9#-9yJ~3MN468s0vvl&-ERK$c?Rq&d$k!e9)PEnVhDtX%EYCp`
z9o>beh!G5W^)Ed9jCZiGUrNn`(;}Ly9YzW|#JTCZMrZ$>GL4S+MKqAr<6_F~A<^qd
zuqV#fE;dYLT(h79N~l63?`o6otj@xP(#<wud&{`Pq`Uoud=;Rxb<)Ep$w&oD-<TtZ
z24NV*rzj-~r<G_SJ#7T4edeIE_cEFe8!%dWr$w`<F*7_E4AaN@g%?j#Zw^bdL}Dz~
z-O`uTPJ)nJE{DhKW8v^?lq1rR=xXCu#3n=`G!lqJJu>)9;|KLZ-J@cuP|I14j<uv`
z8swY@wfGPIj0SY5h<zuzwv9?agm{r6*ev~`bBKI^IhyTG7dNB%dDfK0SjQtxwpsd@
z>?$5gh)BSTo1kD1FhapIC<DDsEAvS1W==d!58GRV?A4yVddOgJ53zm>FoE((hGNm!
zAt9{bp~X#Z2uD4}n<cC8EmEuGY6UrVqMZ{4RD<m>Hr-U9DQjFpnba@n@*bLKwiGN1
zg$cgpVTlwh>`FRPmEt+Zk_N3Z+&Ha*<wi#Yqr!U(fJ;XDx74Ss002y1ZfKB{F|5ZW
zR@A5ikYuHdvS+wPg4ukyni2=%Xm5v`hE^9HBMi8*vVtpQO!SLya4(4*AF>Qbdo|K;
zWo$g?hx$g1`C2)_DQxs04H)o7zB)1eLg0-(y2>OcSuYjL%;On)a^fbOxEK$s*PY_e
zSfXSTX$zZiT+E1J^DyUF!06Oi$1>EP0`^rYry292U@4B;(-T8kz6mfFAKn?pc@;TE
ztkp}#`{O(q^4A9Jr93_cq|1nOW);~;Dol2QtlY$vF)oymLX3DL^L?`Zt&3oUuQ0-h
z!SUl`a}_Z4D({v`3<1QAuyv~Anc7zxY9P7zXgGG{gfvz{E!<A4q{dga#MKT;l3Ucr
zB(<^88uB?LQP){}h>x#EOl2%8Q+$d*HS)HDM)vJ5g^B91IQ4Xyn1;ppBh}udXmn_A
z$(Mt>8te@y;!Wi&smZIz8?B?!igq-mEx@cJ1<OFt^=eroIZh_|k;GbBYh`i>6abjY
zz>L425~DDv9J<*eY4&H9SwNL!iJTdnjM8bWR&bV0X;7vpXD7xavzF>3Q$yGfu99;S
zLA8V~&?K_vG^mIy!h*|j((i3*5pzZ*F>7*e=pjrAM8Y&H!9gIN=VPVf#-p1BZO-@o
zfJhsyIekImJG*B#L11#BcglUnunA`Ncx`;1#Bx!r@^MXBh|`M`vDT_OyBxOXB^}}(
z8R<|k8syU0Lk%p!M4jWZG|K7VFzyM2N~TdE3`eDClvlYTn&d^z>V%Rjjoihma=9v4
zw+iW)SKB{OfDG2fHNkh~x%~JX%3)pW>yILsP|2=~CCrU!f-h9AH&#64bh#nT3)T^a
z!Xn0h@@Ldksbw(%{m*HQIuZg&I<ha5?QisNxT!Ts8PmKeHm(Ejw4n?^xY^%kg>njT
z8YH)*Ub(@!oEzm<pG7NTJ=^3j-kvF&K#&;mOXii^Cb~bJI0ks65yAVoH`{oS1=8dW
z`+F_^BX_28vDhG&p2CQ@Qj5=$b}Pm}=WyEkgpj9MD^sVAJ4mJn$yu9_Wr=3^$CwV9
zWyTnT8afL8%#06E-w7`NW|`$PRx1>=XY{~qTd6Hq!%(^g$6sq|l-2=9N$;1{A7CWK
zoDNE2tagOWO_ZeZL`ZU)%nQau5$f#ZJOHry-ZGC|`g85_Ksp*Y0fZ=eE4Ios@TSW`
z|Hf9pAqoHTT<nNNA;bXP6=iWS^QbHW9V@vT+Jf{Jc&ruf?TlOEzd*(+?gnoI2GR=n
zWU1{$)Ph5rQl%`5O@LBTvfRJIBHjZ(D1&76beS(Jf-n}6fat;vveKUse56%L$Kt9W
zFc_C;r-I<D--6b#cCy;GUuHs9)|AYikt~dwEA_Rn_20=+JfL|69fm|L9B0xDS?A}Z
zrHaV<G#btCfCplkZ188gkji=+Wn)aZd)HpF$@?%2kY+va1WDP+a+j?D;@q-Oe(nw;
zcn&M;fNeG&E=L&;+2X$=p@;GtoUmBhW7UX1Ra5ztd;B$4nI(P<uZG-fqk{5+cD6=8
z*y)Jn2Pv1xeQBg6?sh0zlo-5V@c#G)Dh{Cm{3#FE*fQg5GV-7+i9Fkg7t_UU_A>q?
zbMS;mTZo5~hk`^yAq)Q+l6xppoEqd|?}ww};EC=$;?KbQU>mE5`lvl?ZvoyhzNuD2
zMjrF`Oh&M5_!4<Mwa3!!t%x?-AW!%+%TF)|(vu-`2NhZKOrCO9)r-?GZdS3?vN`g!
zz1c|<Sr;B4&lt<$2$N@HV*RsK@|;V<7Mz;%<oRe1?lPR0IOmx@MRoCo_yy9OB@=K>
z2&6ay0OZBc0}vyJ%8dmW<Bjr?zphSx@EkQ+++P8h<7Hn$ZuI}9Hle(d$U|rlVOG$+
zUky^ALmu#JaiF}G-lKwIBzryBLvN+wVlu{>RiDiCc*DQpB@o1tHekwc`bf1xFWO=9
zg6@2IE47Ae`>>#Euq`5r9Hy<j?XB@PUZPP;upQs=Z_Jmj9uTq_HZ7MMcXtOL74j@@
z^Y=t1bF;7{J6^=F-c60SY4lVp8h<Z#5j8HvP(pCM@BMuXlvHKa(2ft%vLTNO>%fS2
zR!&?g9~wiHjTs~#d0!+I!|7R|@#8pH1|hc$D=VLbuxEUtUK;*s9FPso^%d1cK<!MI
z`$xP7H<1-d`=_%WN`-vp!!eAT*g@QC`Y->*6msrN1Tj;elh6GZ#+2C(W$YYJ&M#te
z0Q8u#X=TCycF32(;@x_6m#<RO<Okb!rXKt|Ca$>y{AIm-9nT;Hm_C9HF8_&bR|_Mi
zsaC$3xYceWX+Vt+plMMS^S`k`b-3<{(H@9mEGk|GxG4wV{X0d+35zK2i6GeT#;^5F
zu->*QDxwG=p+>%EzhI{=MSR!yiw0vF@`3}5VP+`G;>m8ZZTg-13qgb#pvM~M#1`VY
zowb%jQ6)rnZC_M8^Rxp`mmP|Ve`3*jUa-2wQu%@XjFK|)L;q`Sz3f;FL=7h#c^ZpB
z#SFzFt#JfiG5hn!={<9HFb<yY&6AyqTP*{(d)Myr6Ki?u)PrSb9|PDAXoa=;(-;?y
z4|;eANA6<Z*Q16Ea73T6C5%#^Ab3KpgBDAh{M;Ha?5L_XUyg5f^+p4WTLc~R7cr=A
zz52**{(^;o0{RE?_RIJP!jMK0D<}S{=<zO9fYez6XLp-T&+gsj*VbU2UXuNr)OydJ
z^4p@K(+IKyLPeI5-#M+oc{nl*(6xQLhxbY;Njrr1IqeyZ4MmT86%COq-T0(}>&`B<
z_l_>J^|<DY#{51`w~|toY~Id6oMQWW#}u-bBC|yH^I_T<X~;6A)LPl!ThqKew1Z+E
zU=3stIiU{lQPPhR7j_XaBDh7RNJk-^f_qi#)AIQEezBfq@XE=7HgQ}W3BBR>u^xzn
zflHr%%0afRK-UQfZEcW)W1FF1{v^QYo&6ir<<to(ha^5tt`&631ePxTHst})GsJI5
z*OIplR_FPF3k#%MYMXZ!^Ba_k^k*a$@gr0*l^*`gZrx;Uy%}<-w`w+`aRKqK+0ruz
z%RmifY)hb^IkgnQ%)N}^Mjby~dV3=#A9*HodqNIKpO7wt@+dTTm`jvd@JRxN^bQ<-
zR_&+tp7W`$!t`A!M-&~_Zp1wftfSW;NBUxd4b*HPv++)4?pY?bar@1XqY|Z9>q}L9
z7uQFZ(3NhoS(wKk(le$Vl3uf=uRlZBFg7(&gT>O%pXnPqz1dP`6KcppmR3ps_zD!`
zSeTkVz^GJ?h~R-SEQRzS{sDTaHZ6nvLmrE?Y-lkJj>bSu9s&%r8d6lK<;c!*OyXAy
z0ZXZ1y8H{|SntcuGO`3UcVzg<P#aL+vc59RyI`6DLwgZ9DZ|qs*|3&|)%h=y5&kwB
zWB40EcNyu=Ec}LYbG;mA9puV0GAhKyu}$?dx~L8f#IaPdy7m}*tq)Ge<5MebdT6la
zu|5u(h6}KIdgT{UHaQ^#3WM^2S~kmxi5R$70rSv5@g1kkFHZ8`t?VoGIf#hTqv_-z
z77jJNG7Nv(8E{Ha1Nswho?!K8!a<v4TnROZjnV`;H9ezB4w{0Vr=_vcN2zXHyiI>0
zv^5C-)BPJ_wNRWff~9gs$1})e6dQpFkuiC)1+^FOza>(h+T}fgY>TaPX6Aga@ZWiE
z!Bwrw%xrM8d$y!9#G)GA7QSRg9jj?lgaVG2tgRVShl%=)_eMEgs;C;wQ_^Uat!XbL
z)K<s&lqt`myhFX@d{VTrS>}ZS>Jx$odPt4!q@rGjBenjrMZc>CFLnNmhCc(kHe#RF
zhj<w|5(%dHqC57qE*$K<$|m}o3pNSpH1N?k%0zF6M+K`q&%oX;n%V<bYhD_P4!VHO
z%kjT;Tdu>_YxJ*tD>2Vfnw%Vb#<FOvJr{qU57|1&TAh$DZ;;8sNd$JBbn4odDRELD
zr2ue(GzXC(B<Y@I45yjnD(U3R)Do-R+RCxl^>S8BbX1nk^s~Jo3#}ps#vrkj3LsOX
zV-(bB%vZ=cA#W$;il|}FVxfQZS+|0}c#6ROkaJVBnqIA%zVkI%LC%YnKNfE3ETsPV
z(JqkbYLt@$gmr<pt4cJf0cu|8^Wt2H9;P!<F7mgHb<7yN*xxQqno|CkOM(heZ{fka
z)D{7k(w`=m`A2kUFz%hf<NR~3TyBiiR9B!Nzrx?+R;OU>xQg@2APA-kr4VG<C|4OB
ztFjYutr85+t3yB`SuWdz#C*BN<aYm)<XY<o_tw*Nx-Rv@uwaXEp(}KWTwgL$EI$-K
zI%A+C$qgY>yB*p`{^YMw;i0r<jYP`qwen}*-=HRBh-n>qI#QZLRv0jNV+^%EKRHvG
zX>O34e07blD5>b4WN;%|Lc2Hn@BA0#9(4oSozw~-apjg+{Q?qtNXuJ|{;=S(@|W1I
z{qiX6f1CY+Vp1o_?e@!-797bl<c?^h|A2vgj~_Zl?liKuw+ulN5sSqjQ=-L}f))d-
z;?YMvElD+o;gP%ODrPM|TEG^bgloP{>~uPNoXl_nTD7E72ldQga|PcR*TjM&jB8MH
zUS`D#c_JwI2&5%WF+14Rs@_1Pua;Ju>)^h_`^%hIG-#w0$K)<MZsx|k>5gL%$Glh{
z_>3$#alR9ZaLamGVE@WaLG{f_S(qk+a|DCMKz&Gy>^)SHB7nR&Ha<XepuE!N1LkUT
zl8`vEB;Kh+bT|ydrAD)I*oDiiG2{UwYf_dQZ%@q0ir9E4tcU{lE90NAPk~Oa3aWrT
z*$EHP)jpXP#p2gk7j=bdSsP4V)_07ovxeY0(_o49-VmJk(AVfOTPGX*S-_Svv86Yr
zSy{DyR!!U#?e@=C%3Z#~B|0ZogDSMfZOPqfg|W-=tiz_*98E*r1Tf1#WlM4EfsBda
z5pGx@Yzy)OB+NYBBwXiGwWB=-y?F-g90YTtjLkj%%)lFqUc1+y@erq10dab=)o6A2
zxWnbX7!1yfY1juSKqmLcGr!4z7UKh{8QgS%>CnghV0uQ-Pkq_)kUxtoVMq)4tFbRi
zW%Q7T{ddEmG!9B0aZ*|+5Vdpq$5U*UJQ}=E-c&^3?qfk&sK`{q)4iPXxHrytJLXHF
z_Pab0$9fq&9#%1tX5^FJ2swaI!zOrTZ38~#ji87N;To+e@U*{e^3ZGo*jfNKo{271
zsbMbBB6-%InfINtZl*jJ??FH_R0&sS%k%!sgIY1;CGvtXHGG^<P~}Bir~#07F9i{B
z-7jpJm*ZOq7za$`m3Rhe#T&jC;&^#A<_Qp*f$mnPXeN4pt%ET2R_9R*+rC~*LCuVt
zg*ghZ;v3d7qRddjGT(?CwPtzKH?mdke9P7t?^ny)-=Y@I=M=v3j*rkaPH@q5`J2D3
zMv6KG!teKYn|+Ty-R0eorwG}gjk6j%gB;CQS^na^kbUJ@*1>L&_hWminGUbDCXsd2
z>G&WTRE-Kb`jiH+hs^?BARpR%ShQ0<O8lA9%=NgXSA|s&K90UX_ek6=)oDKQS!=El
zHuM76$Df95&Q2-RLPP?04B9|<0mg~bc)I*EtsW6vJ2i<it(VW@tOCZv068m2#5>=i
zdn@h?ctZXaWEqXR<G3zTK98Yj>3chLzepXghO$WJUMOFtDYsEwMhF1Q(1FtU6RC0X
zRfxjA{rk(mJG2E_6CGzI{@zjU`|Duf|HIu|N5^?x@4`Dt!^{mclNp+%i7d;u5>qTg
zI2lPJh$W4p(THrffu>0cfi?_{j56CYv)fp%%gm6JnVFK@=h<*((%-%7`{P^ppPR0F
z+3&pXfqnMbXYXgj8PkMV9C^cga8%T_q_xnP>@<7RN5yarsvxYh@KzcX6$?U}ELEm=
zGSWFElHYd-q-a%?9@sDMkRP@s!+p&6g!-Sdf8Jn2ZhRPRnAv(cz@Jf5qTYrKVR=Vh
zk;XX}-!2E{9sPo))6`hs&@Kn%BQ{gDB+Of*OEkU%1C)dF`!w>?d<YcYJg_adlD;<8
zqPVKS)ekAQMavU%m}#Gwl=Il3HaH{%tJ@HlMCIi$pB~eQSQ7>$a^f8x6IUY(hYEr}
zlq2Pcyh5~y#Hf`ceZs6zL5vR8j2sn2fFFUkAX1$?I<F8gDjA^Sqa0%-8I35rWBp~<
zaj~6u%W-K?o3Z6cSzgx!r-U4DU1I}mk`wX{rIzP9<V0_Vl%QJ7BL62@p98(H`0M24
zm=APCfaf}2jiZuxI*m<7wS54L1#5u6YiyM54~BA{k~h?wmnWC}PfpD{RHRtAlJ>N`
z0;Dy#)*E*=qI&D;iB@p=>iTSzGmH`~*?Z(nAE3dVNKI35gA3T?tm22*s0z%=*}m8m
zOO~m{kRZa%WReYXPToyX_(|&Em+-o)=Xxi7P!fn)5tiX**(v8mC+hH`60uIsPea<6
z%n%Jmkqi76GsJmjk=;QpxiEMfkxmF%1i*ihEmV6oN?FLo-i@s{_TSzI<&x-{zHqE9
zD_C6`t2-3eCR}F!K*|l2#mkLkBZm)>EBwPJ@!1m?A!<X)m8k<B>UJJqWuN3{SIeLv
zfNDmzl{6kh1{-goLSa^hcq1SU9BYc{%1~=$)X2dy%s2}(-U>N9+?Rq18heZip>(1G
zDUcD)U!XR`kWs$0?yc=DF8F*IX)VJ$Otf)SaBw|*EHXNgiQzmZWZ#unOba(dAonIO
zR~zwhAZXQPSU=aqM{qo<3m>}MwKkJsC|~jJw2qAEq?EC0BD>`}pH*og2HL+2Az_B!
z^-0<Q)Wf3brN}@aSc!$Vl9f4)aMvifVvG-itueTSHD=3ggRo!*)G8d3V}p<?$!$BD
z=T}K26+w=w0^-fbd6T|vS#nCo`!g~&YZ~*?u?aTYEDD`e+FvWj-6A(6x}YcyhL2E3
zxzU~tpqRZTQvj}!iHToW`Hcb+Z$sj5l}XlES97yWw!hB9ohnme4h|nM{E7IoJ04|f
z^wE;V0f0~oL>Av<i>Pkuh3CpNZ=%$pz_?H*G6C?Cs#x<n)P-r1={`d0GU}9K7^&KU
zGrS>l<v=M`-yt;~Zuzw`)6P1$4BB(mg*urP4ObxW2B@a(x}k&cpWG6?KpHcxjVh@6
z#^qLDXt>Sb+P=-{$`ymV8yjG;-0sidO5*XQX8Il0H<E#~DlPBy_PKIY-IMt&cSVbu
zPHV1+yZsq6;1uYF#K`zFrfyKvz<Nn_NRc5UM@UV6-win02R<;M*WUmC|5Ivh5=bY&
z<p7{qHZf>TL#Nc4K)K?IOD~oBc(V#2r4V8b@#aMMBc;(7%A8O<mFR0{9c)Wdy(fqd
z{eX$znkWSlEt1*si|U{}D!I6Ho;M}3F?-t#x#n14q#WSTY>9u~Q)okcN6xr-_~7Bv
znwn+VFJ{pZaL1o=Z%MPPsle!xHh;#X5(0*Ft>@pio_awN_yti=(jIL`91lZ+h;B&1
z22j`DDjoR<wr5J3$sogGsZ?pKF1B`Oq8oEW12MuT1&u&U>58uKFESMa-W8pf?wER6
zQ?@H3J=SwmAtQ5qGR}z5l-I~yTS;Uk&X9SrayPIZ`Fnf>6b$r+(FSz869RX$yf;om
zzfmSH_ZcTtbY+@Rcseil`xxnL)T*dBBbaKrMji;A0>ZpjPXu`|s1J`gqtdnk!+u|y
zBuRyHRiyxweR(L>1rw)Q$9k6Uw~3=`)X%Sxhog6J4l!*8sw-kIj~GeD!*qMprr%7o
z<pVJS1;MQj#0P^sv)koEsX<B`GmUAKuapn_moVGGB}6@!%42zlx@$6R)j3`&N4)&;
zG$TUQ#Jo5!M%z6Ry{l<&{PbjcMsxy)6J)_h{26(X_B}PCkH&jis*YP_gFKa%k3HEz
zOlv|J44TQ)aYUhRIc&lfdBz(7cj!7Tzydt$!zITsl#n`=)=ZJ-QuE{vDieMBJ0oH@
zA`J6+|Ggxl27+q~4hzQe^vMg}GR}JSFlx*EySx}JV`-@pYCb@Tj{Z`b>;g?_Vr1FN
z-kJwZkh7In`{iTyHV^|8v2Co=oSs*_1!(jL2gBZj;?V28l0z8L0PP10xl(;iNM)4y
z-h&=LoT=IbyxDg_35^)bz+@|WgVxEym<e_et<Zkx2f!Uv!(o8SqP&9%L?ks00(kaf
z9~_;=-0Xeu^e#!AuSK`KMU<%rxLi~Ef<Mo_UY5p+PDSn(lWCUO5rL=>s94KWyCkUD
z94l-iG~<HymGN<F8$$Az<z-cfkq($tEwVaMIMrO|s^h3#<G*mIu#+Gti5M{#Yppe3
zU+9+pB9SQOv7F#tRm9f?k)V1pHjbOs*v(G*^(Iuxa|o!R2VjHG!)nu7B=N>HH&{<-
zgGlkF5Ko<mNmS3nW`7Gepei!55W8TD&60tl1k|_2iXaRymFV3T6I8I9tj^fDen6w!
z;|#thGh23~iCAJbaRzXMCLgoZ+)i(%4>6>y?y^!%YbKcPiW9UvgDlRx?Dm0zqOfEb
zLAg{u?yd7{^0tNP^1QG*h@Xfy0B<AHQC<z}8f2wKrgn947nKSG*<(yJHk)mg0q=)6
zBYBnvD*uc-1NO!&vCGv)=N<$|YhK+Q@=2d%WVaOOzc0;dvo0~O`TUf(W-y1Hf+a?x
zeA>Pl=*4xE7K5CZ&je#m=Pg6>*;s?A731Y|X^A;&Gq*nP&-4aK-I=@O3;v8+C+iua
zeyV&icEEJR-jNx;WbE66S^!`6_wC*Rs-rvPEB*`;100XL?1A}yHJC03%eX0rOvSGy
z3*4$1ZX;D8M;Qi;t5os#<mr>ITZ;%MZBhNV5$O`&u(ip+`z7B@Qw8|>tvEvKG9AcE
zq)qkhm_b#c4K)n1@|}1vmY3}T9R6;a2M}mzPEbzdd;ZL1no0NX`!nWAFuiXRq}gkU
zGnl|_zEetPA~8F`p|3}0@NNRK%gYbazR{Ew>>I|6GspFZ-js=Jk`vMZ)_w#E|Hvji
ziDkil9PdJ~P+zlB7OWX(`}`;II~bk%a?}cmNU@*#R4l2OH`cC{pT%Y=wB&iq(9ccr
z$+b3g!<+Jp__eN)si*SGP}rdM*deI!0uD)2jm+Kmt5}?gaI5^fNK@=!fU?NI{Ea_z
znV_b^Z-Yc`7Mgfp3pnU^iQ4KGa~NN1536|ledua~hropSL;MQ{Pp$khT7`V3JBSxX
z67TMlKlv~4Ib--#RCo#QW&l%vj$gb;wFb3dN!H3=5`pbDXJV=D4TY3nCV%zcX_l~P
z6Fq{^36iVlZ@!pS<$3vge&7DEQznj+e|Ud}@=-n!8^x7;@=v1^_Ixhaj+;PHJyiZ>
z^Mebs4{G+mos#5BuXrN=3G&Z`y8nj1K;0V(3v~14@}@r{Ip}d>z3;c8Av8;{XbVx&
zcEdi+)>5@s_G=F)lTjIpsequxjNkp+6%Rw`A5x8);TS)lU4hZ-uDed&(XQ|y-iO^T
z2evzyma9QE91Yn!$lkr<j?3lXc8?HkZI?sZ75=atQ!R(K8(2$~k#HEff<ss?hqdps
zxXWxyogCgC#s%Cj{+fvxZdKa%JR6T_cLWXjzirJ?hE<OAX~MqHvx)gnc3N3n<WWYN
zf!?-k9Y7eZi+i+9Z=koTY+98Z<HV!JZ)q}X<=A$Gou_463)Q?4<~W-e&Q}fQJKm;W
zS#htN5Tr-E4%J=WJbPlyTg`GT5&2dpagx7nO6?M$w0&}NECe&wD8t_w#JMrkDyP^#
zut*3ZJk=W}k65Wx0$eD(CZ{<Wh@cbwCTm89KGQo+PaVT)kKcI<*%|2@vcD=um_RY!
z`%M2`^?H1RY#ICx1r)-8Ve@DCXm}J+{WK>9*Xr!_3)2m$Sg};$ZYaIy#N37s87k+-
zbj)~R9*Ujua$c;^M9^E#_rYt_BDMjF49W$;l(P}40II<*y3og~dLqV%%a_qN)oKkC
zk&A3maj6mPH)NJw9NnXYAj>dQ6|`l##Fl%Ymv~5CF167#u>L__F0&u8$#b26+%LC(
z)-m}3yZ(xpU|X(7u1s`DrFOYvp~GF}zaY#D%u$LI53&X_by<W(5BAsmRt7D(VnZAf
z{gjQLEJK5Z>Zk}WZjoWepygFnGQ3^I*Yv!Mi1|WI;<ZM_dxS;?6AxI8R63eQ*}jI0
zceIQy;T~=;<od|F{x?I@+Z*KScDEw;JO=~od5G)hA;z!q_ia1y&PZAc%<a0?Mqi8A
zgC<05<F=kOP)*i@CP%x5mj87z>k3?3bbX8fu9fwm>vmOBh+LHUc(G5Qt`M4CAY)=w
zZODm7zzFOMZ0vI9nUYNkZYf5ok1bLLJWM`XAQfpQj2h+~;#?nRZx8f(tQp<X<Kx-X
zVj<-T?e2I&>UOS;;ro^Gq!ymorY!dQ4Y8&u)&=D*FE>U*NUKA`nk+xj7O9mtEl%>G
zLbiBL!3vq2p0Ow~CJ7o&@n`N+QG;e`dd3>eYA(o4M(aI~4D4)_X)&6HOb>$6@=_Hm
z(OO$4(~I5Eh=o|&Av2sx3QvP)^fEbxqGzUcV?`bw=uuB+mbb$dv$Tj+a<ji}nHd@l
zg%zcf_Ldkz)z};5R)4K;fMwP6N)R;7Xz1I#QMj0u<#r$v?{@pNkY~-8JNyL(knDt-
zunVD^ccxw`X{qf&O#rDj%U$*<&O+c#ciZM?o&hCy#(rZmSzfA>c;&|wnN%3zr`P6|
zNsYZc9e@`S!b9S;BPORwvi2MGi7u%tp)RA1m{nFEFAPHw(}ob#dd`xdga;Yk$M}~G
z(&!D)&V_}626o6jiCzpzp*_H;T8gIG@iqhm!5Rh{H$@L>H8IvwtpF@G#5#_G=J*|3
zQLDoNYy!7YX;Cj!i}%Kb1DXQ+AwryTu4=G|UZ1pjJH{@iv>=Wv4iULGmI~?-BhZMy
zw}k?7{q>TMZG`k7m=`z{+KsS{eyL_54wmyT#$}lu@jB`fAaod5S7-db=LM*VUEZBT
z|8T{vk?!CO7*E9fkWLPgV$(^Fy^Z`1IOg-&E+(JPNrZvk4Pv3*pA8#)wj5T)+&IRl
z)gWOKpvSy;G{KG-z+CT1eHsF{0*tIygS<D!+k|`U<-XveJhJQC<bMCmN~i@6$_VHO
zVmnvy{&JXX4;nvZ2%O9N?DQBuYJ@!GP3WPcg25p(dH>sfvEVYTLoN^7f(#loe2_fi
z6!f4w)~=FAgLj9F7%d-g0vXt6>rVBCW)*e(55|hC-cN&Tg?#93-<U4tVn(abd^moG
zXb!ZxRB$}zZ9<%D$v0k=_jphwgA=qL(fLmpWf2;~(t=Nhj;aN+Av$$6?F^$0Xd?3?
zP7~^_;0+z}Qt0mM<fCawmVb;aVsHbzG1wsMVck9DzXNnri<1xrUAyzqGAgXe)1kYJ
z%_8uxS)PgE*5_(rw<1o{=VV&8MQcBo=D9<6C8L|3_h-x`;!)1D&KGQ|Wi8!gq8DSR
zYPQl|QP3BMt9qB^C2xUIDJ-yr(?(wQiQ7gY5KjD`m1sU@Z=>vQGtw~g=%7|!iNT{Z
zzgBtZjn855Dcq=r0)y1%2Y=V+I$LBx@HgN=z-kNAXO<Yuf^=rAFA8#%^S+}RS?puR
zXw_jzC3Q(6In(2fLG^?*#aa5ibt=Y)-p69~<4;ttk)=TaQ*06WwM3TLo?(HL<uN@t
z`a3z_74}<WeWR=l-XD|A%Bq-LYZKJb)u~mx2ZrAJH9m&Lkbz9Bn-3M02-|X5n|`s}
z0$y1FR)HzdpXSAqZcz1It43Td;S5<9>x*`xMhOi9L<$e=fc3IIem9f{=N4@@UZ%rS
zVySHKo_QH4M`LIk>uPVbd5+1VGN#FZal9ILbBJKODQ8QNpt8JNw)&)LeDP?47B1To
zA*`k(z**TI9}OKeLUx#hWcVN5CO}zbr}e?u4j#n2VhO5T?(a_IVyR;8DegR`R(;%m
zDLH2FJHzKc5rViKE<#|_a1Fm2kF-80Vr8?k$7VVvhm{|&za#Krf$X)v!`hf2pNtuq
z9)Yd5Qug^X>fz*hh8cb;o`JD6IsnY}=~#|pBr4xl-_klP=VxMO2wr4LuOU{v-{P||
zSj3&9$a534w$G)P)xkn(s-NHr@WeISVFjzu0>3T9ab?NdUx?o|FNI}Gm|4PVcjk2Z
zVo)6Mf;hMG@+F^(fwFozm3-M49^+pInfI00%?$+_w_o)hc(4%%B||I)_pSn4fE@BQ
zV*!M)WZLEHK`F2#z+9Zn>*O10C24K3$PM!wSnuMSCBIN{VbmBq=#y{xQHR+1Z^v*d
zTJFhp%Xj>nK5TC$Yx~8+P^Z58wt+(ALD+#~V8@wh_B|gPiW`8-^vsv<dutW|Mlh9u
z8T7`#7OT{W2yct&e%%K(dXW4eMnU)h3&PtCKv6h<e(1fBS0S2tOPu`3M+VPuQo~HY
zHf`g_i4Mg}%J4np>`!cE%C)M9{4~Y{gA5stOilP%u=Aez^7Hun^h)`Kcd1EgAbP(M
z14A-m{L&i61uKyK@~bey5E6q3btG-Wwvk`kR2a<<qpZC^T=+LW)Wz_3(Kx3@U{1&1
zhHPgElx&myE?5EDEoB^Ct-o(~FV$2e5M^QRvuMp9e5jnU;T1f-timQO-ap#3X7WOx
zKm8Zo4Rpk*E`Rp!EMm&Aw!heyWiZO*uZbWlH8_Bvh6VCa-mI6u+2W2EHCq0jM!r%{
zAYM=Z4_7oaLO_)~dO}b?9gP(4!~Ulk^<yT;zpU#$bm{)v2Td%D20nwSZ595L21VST
zK6iS@=~ScdZ>hYICbd9yc2pCWHv@3fTmu@NT8`)~f7K$gHE>U)u#Zui1HI#NxKzAN
z_A9u3ix7PTyWxs~{R;{%!kZ}v6awy}4Ei4c2u4RG@f`(&G<iQbiXC%Ca-cusqA_v+
z{t`K;pzz}e{3iz&_8qRnzzVXi1=#2i`+3i^0BH~P&uKwHrO^cD!wLpb8czU+0im@>
zgB)H6d=!~(Sww{#;h*!)JC+GYY=InE*mscjctZ}HcvSprOj-Gja<n&XkeumJS|eSK
zDHt^CysbF8#_{=T9b57{ZN{ZqYG{YC6Rg+c3JU4me!UzYgodXWaoU*h3H~jQ6{h6T
z@n>cSb!Q}@`z+w&0w)m4*(N91;@~zfIoa0<Hx<0oTe8C@GAUT))OhNYlGWIVTNfUL
zPIan(<G8w;NJTj1G+Pds#|?6N%o5_5Y2DN?O#j!sL^;Fh2b2`M!?F)dyE`)$5_9B{
zM-hLP(d1nh$k~N`2V+gjF>OY^oRg-(P;uK<3PPi^wE*VN4c;A7XoqSo=lT0@3kmcP
zUxtvJAMGi)3tqwzX;>pM>wdYwC^)IHRxS*FfDO?s7X?E>$Ra=~FBhk20GhyPc<uHj
zi61d3Oq%s5jog<O6f#FSn{1t27K0|L-B>S|r?nuwt2<k+FgoDw8H9c|%a#5L%zus;
zVCzP?Dn@|$Q3rvWTn450V3J`k?8cl1TT_UfM}?0z8DjrDUkhGtl%Y1`YA8U%V#d>Q
zg?1V4lc6Lky15vwI3mUm$3qs_)ySYtuA0Rw5Nc^%4D^m^>xLI`bntmihDu~1FYhYs
zvl_hrdn<(w6xepOdHdyRpHg@rc*)B(F?IDXljC;DwXyUVuhut&lEr=7&wT_{@!hE%
zmh+-_3W#Ld<t$dEQUN=5CtADS7L1n$T`pxlTn>R=o1H185u!~sm|{h$F}6v{Sr50|
z8tJIbK|>mw`sZFzF(wryH0GhkTx{L3GR{8DLmkV=c%Pc5729!&NipYwfol`I8Ra!L
z!=nL~>l&#HelI~;O9M}D2tq0j$^mSX8)HxqZS+t;M?yagJfG-qlc3b9*u$hCVTp!I
zLv4vnPJNrD%rcmzR`C@7y~KP$n--eM`_-lf!L;5g$`l|j<)+}QG8oRaGA%f7Tn6_=
z19GoQBcyq&9w?da&n$A2y5Me^5$#OZvcog&6~^qqi(S)}scV;6wgNX5azK`D_LJx#
zsIs@%4=A*msh3;5amy{D5LqU-#r~*iD6}Au`1W{*VYLtosBj>6*gIeoz5YP%jMHyY
zV^e3QFdGKuUH%y>uCLPFx=Zd3Y8&HYo7kNft0fcEWU9Aq9sJ26EZj_+r8-Cj(5Shw
zwq0ssd9YZlDruJ>Tig(Dcd4~0=L=1Bk~MPi+Q51oQ*}Nv-d{&01+qr!V`LiUA5k?d
z=hEQ6P|v5Qic!40%xZx&`Y+0N7y?5v2~%vBd!h#ni_irT<JtZUo-yW`XxY#ve@4X-
z{vu{0N^^R)0D^$)y+v9Qb(rV@dj^u(v_~C}<YLJ(P~xEnwie&V@n$NE9miCX7s|aR
zHRoM&v9x)2w)vnY!Z68)B(2ZFm})OJ7KKbev=mh%g=hiDEGEmNNjeJdhCyZ4e6EIB
zve3XgLy`j4;;n~y>564-)u6gI=??C!KrD4nS`>yUK@Ud;2F;80GAAZnhU*Zqljiyu
z83YX9z1F+3zf;FtBJ=!rb3*H3g9U@G$5sKKzQ?*{40gS|x3KSMRjd$V28^gB3z-iz
z_r5e5RFP9gpl-;Bw)>M-NpF$LeYDQMEAxU{#C#sGKH3ZLvB-nQo(Q+j%lk_Bh*j8V
zA*xCy57|16ub3w9_m_N~bT4wh_Q}Kki#kTpuVn;LU3er>#R91jEr5nMV%Dlr$)mAy
z(3My-Y$49355(H<c@oV3!B~5|L0f=-+7>Xz`;fg|SD-N^ACASV1?=@$;$t(P7{~Is
z%{teG-=FZ`xgUtW5Da|MnyUgAW1oJ+C&nY32}M*YZTUYM@(-n3TXXW1kCLxgeF@aZ
zD4zTBbjd7;N-=QX-lcK0Q=ajm5E|eK#t6x0gERA)TCG#}ob}n7>278+!1Fd06tc?6
z3ndE(2RBd(>b<-<@kRe-F(%2A+;VIfll_v>Di6iER$jJwtN6~#$9%YSJ8BsYhE}3c
zlv7j?>=pl=1>2YtMJSE#X+ou2AZdF2cT1R8!&N8EKq{ui&-X!LF^T0v8-v1uGjW|P
zh~a8bimBZu@E4}lVr6}*^UGvWY-bwk?TyV@S?q5uLq7;K(J-DRZzDdHer_8pW3?L2
zpgx;BJRtBw$Wr?kX0jSM!!r99;zSy{5s?|2z6rO77FuORj8<Vl+w`{8R>o+-1cYUg
z?JTRJt%(Tll-0Jg?m@WHnqa1C+>wTheEeI|%>#9%-=A3wG4+R?vMx3~L=IhE3`?^G
z_Sy+FbY;DNnPMn$cy5T1;7*^cZ1h<GdP0Au#9`&Jj+>%a%OHe5d7*4}YO*IH2znYR
z6lOEu5;Vf)4ctRpUH6M(m%MD5C&420!~`DBh35Tcn7!NL5XWY&X3>py+3sY5mvD#u
zwS0V;?6kiQ8#Y{a**0isgvzlytp{yfnnqgc00z>>6A!R>sCzK>7PZb#*r*|L76ZTB
zDX%6Dw)8jy40*)cEcc|%i8u&it^+|T$S^L(YM|76{h5W0n5_I{de7Xyeb^5B(gL8W
znzA-4ejqPhJ{1$fnP!kERTIkGPkTFz(B@6n%b{L=Cin<&EL>@y^^dtmcC|f45B&<C
z^IuG0Te>#gqkZyu=YHw~E948&y(u>(MM%Er@4+wyk9{fL!^uI{75;~wU{%SNLm<zY
zce#8eu?`Cdnq`f_K~|9omdaOsPE=?CQ7XdmwY0@y5FlItB{mZUHDX4$);EIbrp_2M
zv9esgX-tvDJ$~}7V0K&>UL)W3uMLV+`$oPK!&kH|7!qPlzMDqOXnDngLH}YV|M#Np
z3E7Dm`F>iHB@mdZ9?5IoHqU9Ea8T{h4v(VOePhsvM8`Ix^h`T2&oBK*hy}!nvQd8E
z<0vgI#~>1(6+AzF7@|PWUW)*;bGkH~<3~O|Yz%BagJyqR*k_d!QT8T}|F?A`J_HBC
zPohVdL5vRM{B3j^3oAcOG_$0fojB1*pEdF`qy5B+ant4J{uXpH93*}msOtI)8;jQ7
z!{sXUP|GjV+Z(WP-I|05-z2}X?KYM1XnCGjzYfArMbxhR#wVt+qJEP6Hr_zcUc3Cx
z-vF8SK7co%vdizIJ7qV5`dk2p`Trrfwz<1T{uq>?NF>xu<Vc;H{BxQ7$$zJpD#GCz
zBS6g#Q3&#9?|LaXooWUz<@<}#X!4Zl74p~EwYa|)fy;jj8CZimzVeJpmcQF@Q0cbN
zCI2WWxLb>%u7Ch$S-!RM&omX4D(Hi87Q4YY|FWJ2dZ*OOzkQ@08fVv88^WyU*ni@8
z@Em!SP>FLp)@-x95ql9qwpkXJl{dX#iWm*FRgUD<Xm5F2_=P|UNW2p*GWwI=*?y4@
zM*-Ck!ye$W5VFF$GeKs*4ud95_0sq!N(l(B{o_k$QH%tqp{mL%4|RNbfPY!yv_Q`|
zaO52w3d4>qyG0K4msstW#mQC!xMv9LZaFA^!I=iz48oQdG_Wkz!5xl?qW&<BrXD$@
zLy^dtSvj;rk<i*4g0|b_unt9K^Eevh@D9b}U>PI$vRjVmaA>{`cQUl2AQ!EYBYj5L
zUpQbaIE>mG<7>A8**+>d2F}@r(tmQabqCjcC%QYP!@y!%R>T`nR8@|RXOP(lNx}A8
zD#!UVl8H7P^C6Cp_drYvAr=fgp+oU(oP{fJ7H*UiI~-W9MRp|pA$d8;W>r9MgiufJ
zaKszmEW$Ut<efG`CaF!IAg6dQTtF1zp;jrU8j%oTH(XAOQQ(-P@w5ge&FR4aP+XBa
zgRMpfG^$q08McfQD<{jDF*IV-yuBqWXL(ypY9d&GnyhlR^@Up_b8?Q&ZlD*?0X+V4
zt`nb(M%B5rUJW7k^HO_s>su<+`PK$Px2sV?s9P>Da-hH#;+fjz!VU#14Pv>-z8}zQ
zMNqk(<wq{|xtY#JMpeHSB+n)O3j_x?U&)_xX@@xu!c)6L>vvLrHtQO83Apz%qfIjk
zNnail?ZhpB*sE8>ziS#>spwwm6Qe;xhaszuT@~!f((bqdMZL>|VlA`XfcXc<T34{p
z2kYPs@!1o*glHb>?b^<@bh}~xjHV@{35fCxkNzP6OrrM72!BS^kd+gl>dVOJ9;`A%
zMmZT=hH2ESopeA21LBs^v14j+cNYtgzRQ**(~Rt5x!Q@W*g53Q0({4#W1U=Mufc7N
zkc4Za5xTosf9l;fQCvEqN91)z&^?Uvd=Urw^(EVwC2<r-Le@8KUxy2Kft1<gaDNW`
z^cD9aKq@a|VmjCp2t<R-q0f}7N*o8)CUo+JQtkteBfVHxd2A3CB0q94D~bdgJVeGN
zUb7orSV5e?!x|ZH)6qmb=t&cjTxCpyYA-5_=#6Fp(#rT9x~(OjTiO>x9(gzQ4IK_z
z%%CTisb6{-z>P*yOE8%j^t~rnFOy=NH)MKxWU`H>Cflm@%%;T7&KFvF-4OU}YR5hW
zy)Ayhv|2r{F-$wEinZcz^0hI2e!EPwzVzM*6yJ=kfb;|uIg#nc)6-Z51p%iJ+cRt*
zRiOsl%tT^JWXrah6-33}F$ZT{u6T3$r5|T#NqK0w#oM7Thh=A2#rD=De3+?W;t_c|
zdhXJla+?u_x0e$3zdgncZ&wam?T+BkX2dsekKY;et;%%DU1{Jv7l`cv|MX{l*mMLc
zR>L;R1RqQR>6@k6+8~gNd&B*G<3>vEZBpyfbT+Kmtp83mo}7yd>7~v|paD5}wwu3D
zyOMgN>I4j~!Cwj~VWHPsa4<B+9EJ`aEcc}Tw=3d^WGtDTNDK@BgxmyjS^?H3|And%
zO(?>3X0J6{qx0rndH(s*lBk}pr@_`ESkGvVe#r${$L3lAm9<H0$h;d00G5bv-dmhE
zup@d-7fM@<3o0wK$gP?afw#QB%@dkV`mKnGYd5N6(Xr2)r4X9|`PKE(VZ-H(Pbiid
ztKL??eGawKWv!vi2EaQ6M>EH=J9<!cNR0{UNzdq<F*?lgXJ9o=q($vvnHw7$CkSKt
z7!yFV0#F@cOqrM3*`;w6yP-<GCsA5+YE+5CN&qo=Z){xn090FPko&v=hI~+0L(O}+
z-#dj2hUTZ58%Y6O(FXc}Q2^oU@ZUU`MzCD<v|VsVy)US(l^6)A(Yi&#t~_L1)`>u@
z_otP%#17uOR|;=FoEj+&S70>3BSG{`bB0y)AGPJ03XP4h#0L@`srUko2itrwjZ*PO
zCVa|=f|(~a_Q;3rA8@q5srQ(7U3!pNa>wJw3LwNP5b6m(AxzRK?nOZX{w_W6o{ZsE
z;$j>mjeNxRDBL>j@=;S$XHAx;;w{|th#Lar=@?xn()ON7!)Ijz!bL0P*?0!RGD41C
z|L2^hREk-Rdm#h~V&`~<;=Eoc&-=v4*pY#*G?y1b(BLLCxJzFQmY;-Dukw<0#Z*e%
z@(I?#E_*pGUx`6NR7=gUb@DN%6hfw87`);gQ{~{EWZ6AEOQkp3#xb7D*W{oR$b4_x
zssg~Mg`a}L$U-+)$^xG{)4zF#2Q*Mw=sl>8ja4U{O~ZbnEb_Ne1646%s3NjBs5=f9
zTrUaI6T5}2+vgvUak)7e{j$`bad|ZXVDE#nEZ77?Gt8d>`LvLhCpv3-vse0I4Eu^`
zTisVw>Nm?uV+bT=x5z605)M6Xbyj8rK>_c|>KJ7et_H;dtZ^dIM-lhtxD^MLW~r?0
zuriNafAS1-cx`_%k$^S27`TPpX9!}8XzsVoIc)SOS?9yFm$+(MWqqQseaLJ;g-&Ec
zP;>00GTE3u)eJ@&^|HyI5o=)4^c@fdo4q-LQSKR5fxAGq7(E6LLPX`(pchysG|_D#
z#-ea?UbfrN=iPR@?0DNSX+E;RK1<{8jA2$o!`YR{SE73<POAi!-7yX5ibRRw;gyeD
z1Bk<3(k7pX^=Qf!YGF9M8vkx@z%8Kdvd1?7jh(eJ5c7cL&jPwzu|j)&q}+{+cVYq?
z9ED)!PsVPn&g20??+fzbEF)Jhz*s(Iugt<tQq(R!orb$(fC?(pfj*PkkGz|RefzAx
z&7QrSuYAs^j@;&M`FzlDZz^)Hf^DgM!GE!^ddr;Gvq8St;a)wBenYgJ8Yo{fs^GAM
z#e!P6U-q5}X|jSH?`rr;8aoaT@a1~M7k@S0gTcvI2Oia5OA5J~gIrULiUG5wZvU@`
z@PNX@i$D4S9q9<Ym2afEv9cKj1#8dJocyNg79c0=RU%p6@(%5DR*7Os29#XEXT?PG
zZ-+_(L=!>X_3|C>fg%g3Sz7+kX7F8OAVmJ5xZU@hL8x?YmhbzpOJT&)2+7xqOXdu{
zP+m_{EuQq24M`1Wnf$;90tt%4n_wofW@hR9(3XW50{*{Sev}vvWl!7X$962um?1xj
zCY;*D(KxT2^3ymXaF<%E{LF`^_6F92H+lX%c0`$49rBClWIV3B%*!wR6TVEYX05DW
z`7@rV41K3D`s>i}Sd2)+)_)TM04fM#*BNI0TPFoMn#*Ic=bNy$x(f0;<L-Q+HYdN2
zvFclZmY}ly!Jk3J+WP>;4>eK#7;`5?gJ-PSgZ3fc|75M<g2^WNa}2Kv^@o9jcgtV=
zJ44Y~nTcmH6m<Ektv}2!pdEip^QX;75w%$Uo`%7aDU2joCI3iMD`vCWF{GPN{+YO-
zG`ZcbG-GPwzihCR5p62}4iVUi9I^j+XQsXoYgi|5B&oFl66#HVgNwr=P75)dx6<r3
zD{j0J>}VMk(ixGF%)(lN8849iIt@$olL-(GLfxQ?Ab1RtxnK70RNQa$upx4Q{dFYb
zPu~$O;gT+3*Eq&^$$_1U+)T{0p%}|Soetg2$}(W>V1J*h&+Cgf;{8MX84I}b870V}
zoo)xSehm;O*EbGJpPTP%6Q+5%KVvWzbH7=s^oULeK9O`_dIwP~XgG3Yr{Z!zl2FPJ
z?0A&F$*h^Ea$wwibZU2<@5ya)Oth;=gGPCRALxiU)_+0&41UL1M5va+>bUrZtctfS
znwPO8$9F2k2|qe+{z<(d@C&mbf_`GB0;0tFz~m?S8{D`I&0PRx=VWgim7bwA*ukV0
zqr9`z@kW$v1mTI0o?^}Mp44oOoEk6S#@Y^pP)|!eEQ93ac3UN<$1{j`P4a_kbVjFR
zxV+_@mAudFG!4d>ow^duh?TU!aeDZvfJq^5UzN^Xa+a~sFkCHkcC1J{3f6%0&#`|4
zh{W-YOLET*%3zNX3}CT@^ZXf9J{mi~HFwMTP79_K(9Oh30JZ`b*i?x?K@m~3_`=|J
zDxP?G6~_wg%_1My%SF)vH98Z>a<S2Gpm*@-!Gq)ye<L!=w?aR>G=>Vel>tLY2bIh0
zD<mn^V{(_<cDPt81Fn=S`~$8j;i9b&a98@!$OD9h%%x?4=}f-4DuxD|oyjw6a3&4%
zZ@8`K9--@x!OblAOn>v>7<w6OaT#L7?}7$BG`44}#!L=NeXGY7xIT;d6swL5?<`Rr
zP)~q*II_aiEY?5>Sm@};MBwe_S^$CLTw~76C=(JWpfF>MjE>!oI3qAY4lBju`Cax-
zrWWUkT-})h!t>c?)||g4b;~m7OP~Su%eC=L{fo4ScgeeLn+^0%oF>=#@2quC$+rfQ
z@A|Zf7AS)eNH!@{=5KRLQCH)p9b<#4L%>pA$~z5DR@@jAr*9jZK=I@qU@Wf34jCK!
zC4&oAbM@U)5oBadh8Ey_4UMh1%*VzFNd86{pO|@p?xK}2h`77f%7h?XYrb15eSlP}
zcugmK(Q-o&suKL()`&&EF;>4N)6p#xW54APtq90+QjiSCfR;AGNi?~0pM}QanIi(a
z%ji8=hqt_l5BW@{1Pe`v<FiGk`p7AWc?whO<KNfFP5z5{an-Ztbe`tSLS@L1P}&(Z
zBMF&pmnt6)lUE2+F)!I_xzN?Xw=7wO*XvG?)voE08EGs$a9Q-#61TR?OrNx6M6ll_
zGAl;0_kIPwm|#;9a3MFxG9tPdS+;q(rPKXTs654m)f+%>4f0ku;;Qhx+?JMz*1d({
z+-}bXP?i+=C(yvLNOwf<ZrZ5Ysx}`b=>u}7x1*eicAzoIU9qr}7@mE%{R?B>i1#!8
zKIzP3k!O5$8q00~S9bSGO^ml0kO~lt+RlB05H^P|NHa1pRZ-p>AC?3mdE3e%Zg^s6
zO{m;7qeALZ*NE3uzS%DIX{e3u?b2XO+=e@4+NCjGhrGl{U)=zKb&t<wA;LQEli9&y
z;}Oo%B26J%z*fd6%*aC68-~a&(i}UWBL|d-v7#-pX>$2y$;D7m0wyP|vD4J6!Wq$(
z3A;t^we_A--NA5isEcihzzd;(<)h7M0HWG`4j#Qn&{PWkI^1WdgVtgHf`;9e?Uc^+
zKJR;@mAP5ET$oUR*sVV%O!#*i1m@H2Haf%dk)G7S8bC&Kd?K6z>y@B56_*d^^<3{v
z{chMejM1AH&#)=zO=Km>_q?qIbL^O&@!l9NbQm*QH_LsYKv+EoLhJW?1ID97CF%QS
zq|F1tWz$iO8QlJ$H%RcDE>qrj_dc81XjJxpsAMG6YwQGKfvERKLs}xgO&)e$;F4(p
zLp?`}0eX+vBBsmX<<T_8r5GLdzj8SvXFd@7yAIdPa5^9KR~3{{rtlO5`A{5KOvC`z
zhT_iDus&>jit-!CTI!a^Y(lk-t@3y<OCt`hCxV?D0MfO|leUZs*wxBMoHBNdEQTh(
zOg`$*tR{s87(Zo=mo>G^(`j|q)7yp2^2PE@=RTI4I`?*YHvMWDpVU^LOV22zmqOFS
zH9Jm89y6@c(+6Q$3h2KO<N^G}Bw2YeMg@jqR3yXgU+Ua<ng-oo@qhYHUQT_pA0Dmi
z<zq&XNriU3P3DzoSr?U3f2;JyGsPqL*-Dx3&!}=BG6rQxx66XGXpGz?t#}zLO9wBE
zjx}Ii)hk&Pq6F=E2@z}^>H+E!k;U=5M(_|g=5AFzfhwCUvD2n83p26L+lNi7XEurl
z$<lZR`3vSDBq7U;kc};@1Vy%4?)#V_M~h`eOiJMhRa9hUYKD<FdWE2@3NFghMv~QO
z%kb_(uq-6DslRLdV`QB{A?(A6xk1*(yk12}DmR#F!Ip<jKV*OEa1j&$b4si;jvGE=
zq^!5UR=}d$V4Pjhx>FmySzC~jJEpMC(28V}H9OFo100ZRDd5Hd*=!`7MGwrDm{VP$
zr>9%C#wc)xG3JG_INQ7n@Eb;M!j69uePxQlcBg};t72warEu>9Sc@Ir8sCP$5iv+x
zWoLBQjJs5uWLGF=6YWwMR2anRvD?=Nq72gX<H2BhV3>#6O!80o&|F@qy$5E&irMe4
zrU7ilS!FkU?TG;(%^ihW212Ojk^R)r3CEP|H3D<XW#yAbc-VBZ&s!sK%9}@d2!G0?
zBlHn17cZYq`YGM|G|YJAU~<y=SD&%@RW|Ec`PnqTT`>On4dYAUhLF$2azM|e8b*K3
z=hK`mS`^z-#d*Ydej!?8aFm-^L`xRQ7t=5Fx?9|@`S`!&qt?74v__XPB5MGDQ#O&g
zVMEK8<9iDwgds;og^K1^(m0X)jdAiu+^_nxl4Mc&TD*sqWTBf?V2H09H`e43A@mI+
z_PLkbA>WKwH8zm73>M0_eCozys(N6divb7U_HSUFFnoc3%6CGDW|8Dd&E&iOlC!Wz
zDts?JGYQF(q2G@Wk($#)q48QwA14IU{CGrKy20yl)(jmoM1J6HMSSWacujxkpQ>#L
znxZhVbHa~e$lxZzKloFAY*d&wWupAVh%<QbprP{9=;f}v?vkI`cLTi@ET_Orvuot%
z##N|@-Hze@A{e_om%$}QoWU=hek?ht^@tTERs$A`xki4K=xawi^U|<LYGUpLv;R7#
zrn$|;sTay`($rY~2{In6!|k|Pej5ZVXDlQ9(Z7o^XS-_PAl1tWe;<=N_mZpS58l2>
z8iRTM=+791$y<H6E&dehCb09`TniM%KZjwU%01@F!ViD3wyWD%AN8-^4w)NU34gAY
zzoq^N0#k2VCV!9j)G8qRua<upVIeOODkT53Ex+e6(B(0}ME{BrLZw4|DsDwtApdr%
zSQQ^yy-{xLKS@BOs6mW|(Wq9z+P2yop>#E6GqZtWyctxcryuXHXPzVs!!7cbi8G{m
zY6`w|IjC67r~{|z&9VS#7s`JAyP8EHXS#4#|Gr7~?^4il(u5v4pi9B9f!^sz|B`of
zdAOds>g}96Fg>HEj3?lFIjGAFbp02)W)0&1DF6@dGPuz6R?nLUKP{F+x*SHst&?DC
z-ZOrvccTo4aqfT(d|2vQD{o?%*2&@V9uTXI)UctP9O2KX(op6B+mR!^a|e<X-Qz;9
zm7}^8ADlR)Y^)q@l&GqlB*&!AXf11yfgGDUTcf;#LW**nKVxHzfo5mNTayF5SKcls
z*zYJchlF5wiaEIxeL*Zhj<E)eaW2M4pVXyr?gVf;dHUooM*{N&M%p6d+kVFlJ(*TW
zw^L%MP=AQSE(F|rYVZV4EL$cl3~Re#d7$i`)}@;3y!XiIX;O5!@eExrXLKpJjF?Ec
z-p@?UL+qoeHGo9Va&pq9K=T?@LCs*kobA7upOwGIA$5*X1=oZ%%ekpd#vW0EY>@N1
zBH$E3R1GlNiB6sGFH)i6nZ%^z3#_3+wmv5p22*k6!6#<ap~9eC6kp;@<mJDFl`c-t
zRzu7oLI_bHmv{^It!1U!Ie%%_KFe)m(cwZH&TY+Ry3F=SV>^)3MRIwpby<a6k!HnN
zt8Fh=1}TS(7$R52!d0MtSBng?Z!@!Tsl;GkFpFHbxOxpgHR{R`AEuQW=3q%T7#1?r
zcOT>=<*=@WEwzPRFt-nPh<p18sH1WVp0hd~FhX$7u8`qfj{MDRLYitbt_d4qn_<|n
zQ8Lp0THBbBQ8s3PhYd11rZW-J7m<rGeP|wbOzxJeyY?NcrV64P;S6X@2+lR}?c_`k
zqO-1z;UTV;Dm-r%Lo6`=PC!ZC9fGN1W(5F7pE6mFkO{Ba<uPHfR%GH|Z=?ql0FEmQ
zM$rvndrnhVWlRiPl_%14m6XSpn~v)3Jjcc+&cQ(i^63@fUGC**MEoFh$Z^(Rrlkcn
z@5d*$Dr(Rc`Zd93UIx`OhuvN2EeG!sOSr*5W`F_%1d8|F-st4Vox?KGzXVQ*w=9`-
zl0Rd<F2okT-zAe{oUJgJ0SipA0hCXiQB^r*vP?~#>3yrv6I7i;9OX^^iy}2>jFs56
zrl3rVZm_NBZ{7*-SXG+HR?REz$L5-z+U7)ji4?1t5i(zoY9zeW%1m#7bz$jSfLbB5
zya8e-#6;-#0i^&W+#C(S&Dfr8M&za3VnXfOtK?Q=*z&STa$A}N<(ri}TY^opKyLT`
z=zyc-Llrl<Beo&9j<~<&wXqHjPm^tOr?*DanAyUtWvU@za#v6pUQote-5vifZ^$*G
zUV0{Gr|GDG51~j^$Jlrbg6m<WgSTj7)O5KIAdhfyG>1q&s64f4mk`j23g?@#1m;J|
zdV3m0j~SVVg1}~7h@l(c2CI*~UXK#w8jabIWQZxlYU#n`VGLZ-S){RRpJm{fTMu#9
zR59~%Pg)8fEsb?dGM;U-CHTrJiwej$IRzk=l-}#1dNtd)P_+gzJS{QEAtMG$&fliu
zkC<~9&#R=>pV9loRG$?%?eC4PN+#gVFm2A>4FHUALga17j2eOhob4fWP+}Us>Voyx
z3I9NkbOi0x7X*7IMPzpRFH5jW1PiwU{j$K6uHY7haa9~ElM_bV4&}Pr#+J#?&T6%B
z=}9CYOvH`dFLP4!ERPGThlvHO<0f-M=nWn{WVp-=g21$7H5Fj~@}4w+V&Mf7B$}Zi
z?~Ms};t~p6qSLkSEJ)-36Kdqo<^DuwR0q(FxK<wUp6Dix+#+V&AgPv~0(mfzj~)f?
zbXo=osFnA{xz19b;NwOHDL&*~^ub?@vjQ-|E_r{N%x3oz14uzME(?(opnlky^na+9
z*fYo^0go^S^pVsu&UXv}(V_BaFleS3$(l&6!-YJ4x}hA=2Yu+Km>MQWY-Oc<$X+Tm
zwaSORSrb8&i0kArA25-7R$wK|Q2nBw&Kf?C8-F#z<Ju-q1ao&I-0DgHvQz`Ka8^Ua
zJ`yVcF9~6H#EHpA?b$$Y3ltw-rVTPa6+)&Kk%CZ{p7wS*KAM{>lxtc!;egHZOmI{u
zA^vC6oX9<#*;aYZiL8inXm;3w%wJ*xt)W!Uzl~(}F2(XHc_CyA@)wYP0B!e0ACu)V
z8sJTIkN%QX5CJb4kB+sn((-b!3#4AQjkPGTr$1(!7;0EvUWpf`q1qq3?z)`4LEU_&
zAoF8gz?uJDFrDeh3(}4wj$qaCtR(irL=hIaRPis1{B0(GYpxlN5j%nw$0~u&glb?x
z%aZ6D$~a@^c^3k;V5l?H%%m?#+T^;#TU63j8sHFs>ra*j!`i*+l>N)RVHVkAUIb%H
zDT}P^-Y!{gBpT?w<aSx%1M>q{@tT!Ep&_G3$g229MfoII9sd}U&B&VIs4-cjfz`^|
znB(3DiMK)xUnl+c4Eg-9RK0-LI)BeLr(ONH-k<TfW%v#vPh>;1k01fWLCHp^he1<>
zkq8oz(%eY|+IcJDCgYKOO<N=5$~OlsAV6u@lNKd7mTU>~Wl+-$5(gZ#)jweXJRPir
z=NRhGLmsJIvdy-79t^fUy-FdC%E4MufpKnjHt&e(1O7vC={&W#oz5hnIGVH=5h6dh
z&fHt$mSpjjerTz5&XOR^Y1r+*LzpUz!pQbgWANj}mI1%oTzNlJzkI^U%i8%&LS=c4
zSB+4(kEK0}lYWm=*LaJEDDN9J&g6xJ%KI`9<f?;v#G(!MCK@oe(44_qKoUHMsC)UO
zkIGn=0tyfXr1&b?=fBXk%nq3OphuE?D(tB!9|+|Er~apH$d%9l8sX&oOzN4gQEwj%
zgi$E-eb#@sV~(R?YAIk5P;C>G*3a2K*z+Jn=I4DZCZLGn(&xh4Ex+)#Uo?aqeW8Z9
ze9?bF%X$z~HzW$tCtpf4E7Aly0TA~u`<M_Er&g%CbFtnOsyqU}VpBxbJ?I*>@>S<%
zMuma^U$YTUt1O=`UyqjHEN2xo@Ce#SF)H#6Uvuc9;CcO1z8U<e&gE7u^71WzkG26T
zM6E&q;<t^(ac!u4Cry$DGdiJ}p^E2s{X<)9wLs;2Hr|2Wu_|l7AM#?xxN>>T{!xo)
zzczWjYoA6m*s7W8$lxR`ct!5D?ak0$`GI#tjhs1X1mb=eE!T9n$dA%UEa?y#r}E=8
zZWDYg?qlQrNyrt1d9X_9PgC=}Sb|x-yX0s7H82d=z_t%G-&cN~m=1a2#^S1SGRB$#
zTEB=d)gUvmstA_+GH6rPo^5TAUxnVj=W)#UF>ui1;GAE_;ILwwfXe^I2B+80@=h0`
zJij&LrM^>sXMgMHlHc2LjiIi`vGoW47eWMJBmObS1K)vNK8q`O{24V>$_kWyl|TE?
zDN`sb-J-#y{u0|B7v3YCjBb#>dIJnSLBtk(>Z|2%iRNw_0C}L8TP}Z(<(ghTP5xp3
zU=~b^{L`qyy2?~b{uSiYgV6bcVWjeJn^RQ-f=6rRKXDj9k}%Jd0Vjkis46q=+?M){
zXibmz(%h0aV?Z77^1Kyu(K@wM=#b0Z9Wj}^p{r2~*&+LN8!kz6iN&T^*@D7)ne5+f
zz^bS0VC@O7F<6Hu+Cn*?dmnGdxd&nkU}?PU5{VIIc0N-^ES3YkYx;VKj__DJD4qf0
z<$S@^a&Whzv4e(>ltbcY7^uW)7=?SNKO-cnm2+Tb9M-L1-mJVF?sL)mZJdCns~q9K
z-~b}RuhI3^;*mb^r4XMmW^q6<)1r7$?NQz%^G=x&yBu2L(f-VCkKT@;=`C_h5CEpF
zWn~gaC<cZfYoi=HS&mC%F&Ec%%!FcjmmKfE7~ID-!`@ZA-~|6gDbHs-K#A+iiHS<|
z8*_a&%SkbolHfXPjYscNIoW^rLkfG`!pP|Vge;VIb}J~1omnfVB#Lj*(-Wo3<kX-E
zL^Bh37C^T?&7Uy`mHrUsvYZ~zpn}n8+zsc~89o!n8E|%HHPG31roYW!Lav?F?YLY!
zQwC5YP|i+ms?mc@vR=-yXQ<vzrJ0qNvvRJ#hq@e45_wQ=m-CEcku_X1R?fG7Oash}
z#PDthv!G*TJM(g3cZ6NB#inclc@=VzH_6HIY}*0=^x_~Tj06TJLyqf2BP#L||DA>n
z$jdVvUv7eAq%4{*mm0s9wU^0d{%T1zY48OiuYKcRZmTjKIi+$%%w#;2yDNhvv=enX
z7RpueOjU5bR7?hW2Nd9O8mVCL4h9FwRADve^$>qXVL|O-0c7t`Z-=u5%XxwIekl27
zZw|A8?s*m{({TSDGshm+WFr~T?TCH{ydAV@Ms_RW50`zRy)!GLynQXOs&w)wBOmAJ
zv{-QQBi;L5&Qf+T>4xz)uQuv-WE<p~=mR$~;efo>`><`%2N44u(n9}m*U|m*?jUe|
zwpy+W!XW0Bb|tEI$o1YS6xY2EKrUGDEN&l^GXKTY04KHv^pCMIWIGA*cT0KD79Lq{
zy>-w5#s<STWU6rqBSu<bvjxW5n!$ClGR_3ajkm~no83fQSvbM|id%Z5(x^XZ5U#Ge
zA!aca8pVx4R>Vj{@7D#<s1+yH#8@gUD|5t<sx6Zesi9Lt8cepv0fjfqloGPjA-^7+
zN@=@ErrIQCLZz*>sHL0yUG)lJs>Lca)1p`P2ogA2E>(%P4DM4o2iBu~Io;<#jo#+q
ziZObo+haz23AsQignCPu3DaF?`j;#hswW4ePl>ldW_6nv&-{064H^Jd2#mHUp>Gbk
zfM_h>5C|@WNV=t4-G->8TZ>qeThml677A?O1isDQ8|bZR%C*bwu?y}+;Z3RucLXIM
zvDi1W)b5OzP#hO&La3D)JOg#ru<_k;xA(XdWl0#5U5rZ4M3VzJQa~cqgp_J0gsHd;
z%b_jC#3u%#;=-Tq6xCmot8W1SA+@OkR$!o|Pt%6}<ZNmz-Le6yT8)?`Fb(Sb%h(GV
zx<Mf+_5KSs9xQ7=*Yred@Lxg)hX#tFu`olU_r`VWQ(hy=#yzPvv)p65V9>~sGCMeb
zTCPT#V((8xHOY)L$Dm1KWgLR|3sojFD4-<`V3E?3WfD#I${G7t=Mak3>TgkTn6DIM
z^b2#Y?~P3}b=s7fm1CtXvDSY99cZ;!04N`W9N&gwLeg#%sATM)6nq`wd9jcHC>$Lz
z!=ZzRNoR3N;CS0*;9-b)&=)(ZD^WXD@dC}LQFZ&5ioD^=<?!2Es)xpzN>8-xiXFY6
z>L$fx;VY3j-R>qt4fQOF`{BN8GV@&jjDui?zgozI1@U(VYtFM7p{x{&C!q*Bzj;qk
zU`)0h$k}^?j4d!N5YKv_ukMK9a=*<T`5TB%c)%6~b@|80gK6oQrB#}3^}ga(Fbcs8
ze#kZ?aC3xF%lrKe3n?Siuu~q6VIm+AF(-Os<|F<duOz1=p$Yt`wS<3dfvovJk=m-k
zQHq%h_Je7BEZs?Mji);+O?@aP-;hPbzI-?agOdqrImF@)c`Q0syq$F1Adkm+0l!NI
zNj~-RM2L(gs7%y^o{SZSZpR=lvYZ~&SZC&me<bM4`;EbQAB_g83vH+kn3tzY#=_g>
zcwO1kX?;tBw~IFcJ>%aHcr}JxBhT6bRuyXHxmdID<K+1mpN>oc3fI8t09-FGIL(Z!
z@u%vSls+#8%_d}9+8X7hkX%#Y*g^3eh~}4z<f)kr_Xx05`=&h8$}%|eiY=V`ZF;>c
z6IX~sRt$teKl5Wgup@EsXfXc*e?}mUrV;DqF7(N8X?ZBJfGF*LQ_Yvkq8L!7u?-=-
zVBf{w3Z*l@0Dvw_iqEt+UX$JzBXDG85#ZFN-kbSeN*vxUua+6b=iPp_ERVrr0@T^*
z6<^_8iz_qH>B^vHSv3XFD*pifL10WYxR%K3G%f-Vn!qh<Vq920VtX49K)KfGsFB^6
z6RmEOp#7;;XkHiwF{^b!GYul42}<pn-liq+aBr6N_6_RaP~U@(4Z>kV+DN8;@la<R
z>nho3`q89uvdNf?4g=X7(z3cKCtHH!5lY1!vNe{yBtefB!y4J<zfiBCddnkStw-6O
z2CznxkEm1et+lcveWRJ&m<Y{ysL8U^e_<9ls}$S8g3vhGWeb%@IsV-?I8=T@<^n<v
zJ|4?bpeZ7sNK3#N2(xZqjrpKf5eg^B9`A?of`kXHA8J7Z{<TR)vgan*>s)}~Mrc@{
z^fxHo7htpVcC&q{Pc2XaN7EMhRC;d}#)0fM9DJYlmgeJ3-XxzflFyrW=bVe>v-Z<K
z?~R1jKj$CQ9Hj+;LOSw!f5uDMsdUk;_Jx>xdpFbezG(A=_KbLVFu|8X8=G3pagr~4
zI}mX&<xRwWB@qOB1a0%GL*vJ<MiWhqtvS>QQ{4D#HrT0563ffi6MN!bDNx)3EvPyu
z94+6lwHseCS-u$y0TB+~l5ZIY;<8L!A@0P1QVCQ4J4GgeW=7<cC?Ly4eAnK?S~H2^
zduim!5v@a5E8q8Fl{(SPC$3e$<+ViPHL$$YuP3jksW7(0Y`;=x)DL2n${VsB@<VR{
z!FqTbHWHZoN3P4zNT;+SYtfwm`G@#BkL8sg`>0Kh5=JG=%4@lP67nxsSC>ZwNLGHD
z#>5EO#qI#+cG55)`I(Igq3j3(ke?^IL4cu6hMnm=<QLuh?q;mb2L^ioZ~jky8FO=C
z!4YsPuasXoZ5-5aXd{9zLjwNVj#re5%JDA8TKSFfPbChg-`ZcN=Vr_AjCE#BnRcW6
z-q;EShY)Df)-8YVA&`;jp3n*ikXM`KkM`A69xQ*d{WrC-i;3yha$`+F{!-i^P{%0e
z*2-TK<2Ge$ky8D)m?WeCv$z()#)az*k)eP0<??bkTAR}FHaa`j%0Hr|ssip0X_tS-
zp^R%^XwYK9|MDg*ADU>2xfJR7VHC>0Z4xspD`v@mf(NNJlHcH?c*CFBiAjZFx4h|X
zTG>xi75FyMf2z4}`8Rz4b(y};E99idafoTzjz*$B`^AH{hHQcL8~5)~AZ$|Ev|HqW
z9z`Z+A~x(DJ%*Wb01ODHN??-sz@CU1Mc}%!=Mp){pRsH+4mKQ8o8;j1-eSPl>L!vy
ztTW_!G|HjY*|dsDa+tr0eX1#mG|Lb70od(M%sLOi?1&zZh6OrB7c)dWhEX)PkMu8b
zG|=qQdTbo}COIm#NgtfmI<!e0-D9cO1PkaEW}GF?d3?{)Ue^w%FExF5WAW!2ImU)R
zV$>)(wnwp@dY8Ayc`qRu=)pAO?f5h@lXNNDTuw+M<9cdriL>pA{tL}@UWKB-1{P(x
zoa9u~`;#zE1}vN$&lDVB)utWt&K?IC8(TWykAnVuiod1oPL4y|)2Tgb;Sb=R7u5MW
zEzN;9h7kIp+#?u!dXK|~v}c(4%2Vo$9yh2p-vXvayPO##G1L$i`%*b8W|`oIX!YzK
zgC-a}KsmZY9r3g!-A8~gmdQDZ7Bunf`X{7tLV-TlmSCV4QXSc|=XrBnNEPAgJX`<*
znxEA0t@DF0%$P@*RF_=ftVIsi3P<>&GzIWNAHXt5rEuPHhwPAxjI&|v)yl;^j<lk3
zV|AucE^$)OPNg${1*URoECCJ<nBIu??32s<*)rhEQMu;wv=C;~u*QqU&|l$i(?}w)
z0r!bq>CcR4mM3+UEiC##T89j>u_06wA!k^*!N$hJMh}r8v54>!H9|DtXdW75&Ewv}
zIvExd^ECY7@o%UqP~~LA+t!ERMnCrq5;jKqAXu$sr;G|-n_M=jLPp1+5eh@1BGpXZ
z<t-^BhBi{Tnp~Zxy#shA0xse7xyFb;rV)3MUmLRkn%Guor7(WCuWE!9(XYZLY2~w+
z?sYxtUFjMp*T=vbvQ3C6HQF@e@{KVuMkEX~EO!&eSRSGj>Rz@B{frHAAg3Hf5v9UD
z8bFbZ+Q@Aj7w@>{H9j~F!i<MKL;ZMOSysdjncy8`yatOvFRgU$HUU%m4dgULmT~tD
z-ZD)yLiH?5N$$5elx1>bj|U*CM*-&;3X3A-)aE=D<x(aYRrjpGc{RB>44V5|t7D3<
zksgX%XWDjP#$_@!$k^F{<MAeIr#=g9ahiX_Kn3E#+n~3ZmRglooN@lt*ch?N2+`?1
z>V!j$14RDJh=owG&-rYWnel8di|Av2;S8S@9hD%#Htd`IZ9O0{Y}zxo_%nu?v0I2|
zxz&a}Y}BAZa$8!ZRZz(oylnm~fMO7|8|3zoB%uMw9Wm%x6Ds6R|C|eJwlYsF!;2WJ
za92<gGMv9q<uR(<ZIncia$9pwGHFR!!>bR>tG)^Rom+i}RNIgl1uQkeXf2FlZ<AV|
zVljMYp=9HosYo+`hq^9FVWJ$mIq>I7Kx|xzh>ocbzHjJ8S)Z<MY4G6_x}#57>k5$5
zq|rOI$x?}3FZYB@nTX5Udt|naxw56MNt%L;s2eyN(fGB}94~hw(z8WcV!(icnE#WD
zb-}hoSy9-obUj&}o_-h_t^Pasgxn8*07V9&3OLKLZuk25l+i#?CQh^&$uR;{Uy^*h
zfp6)LTn_uOJ$l8SWhKI`I3-L?A@41hLW~mV5(^x3B(f?F3@C-fwYmUB{Ib)jXh)#A
zYt$N-t`KPmt6?1<2<Gm1#2e+{gFrrhk4dq6>ts&!U&w17@LZn*0bqlH7t6fFO8sCt
z%hfTN$Uj%fdrS_@$^fJcB5DvD^Z}5XS`lhkVVCyFecrrX5Xtn?6>@*9B#J~?2=N2{
z7Sa;{ztXDNC=bRnKf(xrzb{t229Zjw@{qAdTcKvQygz<oAaGtDwk-x1E(%LN;!XKo
z$e{5Sc{JXig{##O5E@(Z=9_Pp58C?!y+}fVj{}u7^YWpfi8?>QCs2Jz4T3npjOD{d
zli{O=%VXA7BXWM)<?-0QfQKku^`aVly-1#jE{6;sB~N<ucDG3w6k2oz8)vb6B!*kp
z*oK=0KWg9R3Qf2J{i$eZ!q|!Ov^QiJH4j*73shWTC_`93<Fv4A&n+tN+0?lOFlmUh
z7Kb%K^IY1)mO=y_MHhLgJReKRV7R72ZB}0J_L(h7u$#umi+*lYOqZ8z9*9pvq)3~*
z9Mj6MaG-q5KX#FzV&aux*nAeks$Y83JFIGH95p{EFa-6o7WfO?w}f7?qh(<zrifZ^
z$zx%VS|y9pw5UsRF<IAV1-AL(L=e;Osa>l}A8!#BTN1;@<<<aE`_dY3M32TBHuI%v
zECBk@1mFT$mR6`JJUu)I0LX4x9%Id+h*Vn#cWGW$*bEu>o0pZTRhkOwZ<19(79NWj
z10Wt*9nTDaCC|#5ASWuPlRww`XgGa>LYt|~>y{%2ZIgbV8z36kz@a6v`tUmch1oRp
zhpopEPx-V%)~6Y6!ok49pMC-s>ZS7DFB@#%!kv(jjYi7aY#lwUvdQLIl`8-y+Z^M!
z%P<&HC|ms5GUz<>F&;qFvel&9W!K6!|5arp#svR@Y)`9W@h0Y}V{L;SHjaVb@h!45
zSZMr=O4${i@KD0g5rHSWLvKO2%L4hhzd;p&iJp*r@`=PXEP+ZzfdX|CxXKWkSA+D0
z79=G#$(}S=l1YWS4ESKF+iK6MvVlKkuWh2r$z}4%7z$DfYh|B*!o5!Ya2G&((_B8~
zznCs)CM!$0;Z`2+I&2_b!1S4*#?2Guv%%mu<Qm`s`CK%F<zw~PT^O7BK|b%l*laCD
zTJbRXLa30S*AD!D7yiFVz8Lxe3V^`3+8|%DUZG+($d|E>`|*O}|Bn=f|L1>}uO#+T
z<3_Iy|7slTouHMLt(UI_)8=a$<?D$LptXYih}CoNZIN#T(VJ2HgGSFc<GTq+B>Ps{
ztwni6_GA5q2XXX#+sEPwN02W|<vXb_mca4d1}U>xz8kD!3IWbWmab&!#_xGM77Bzd
zgB|#Ne;eKbD5?~$JpLIP_*&0C4JI{k$3lk4)mm>DHcP2Eem!kG0zvjJ`J5M`01}ZO
zq@HOfn~eIQ@f|LKtd$?7*=SZcPD|!Zu!zf#lXkWZcC<!P;7=YGW=+9Mfb^=L1cg`}
z48PD3_S2v_OFj@+`I&d3;u?e0D^xUL_~+@Dl(9uY3Elq}sbM0_G-#H~FHMS|Rz<#%
zCE9*va-lVk%Z8i|q+_g#W#U@)_HX=m?s{r3Ok&eG2K-4=<F{!b#p`cP1NfcsULEe(
zj%>8whcrV?Zl0!pFrH!!jlBFZ^#u3`lVvEU9kBQlSj3;~W#q<S27eC5?r6;8vI%4e
z$zRe;)bI4_S_MBDje?XRsE~gRSv93u{^msDxy!IuUi9_%Xp*U!yp^F={t@GC#=jwb
z{~4db!_0$=hs(cW(I&x{+>EO={+%eo?5w_5m%{=yVgH|Kt`h0NxSZ|{AF=Hm=UN3h
zNbSY)W^B1(!v@J)(YKXBgG#woWR4-1dT3!9WI`~!Vp#y3WWPC%*4E(;1sY`g`v)f1
z)zJe=<8v|!XnepNMINUWS~Mcz9ditt<pQ7tqvkUnXs^I}W}<`~G)LiHl$aPV2hZ8Z
z=+=Q=D%yw4QS5g#fS^O?81`%63y%V5JPhlT!~7SrspAM-a|FyC?p?#VtI~isMjYX9
z(~wMFCA79xj-2CQ;|)2)5+CKWBq<PJ!k$1JtwoNy3)|&r8(Jme;Ess_=mymqTu=~J
z!SQ%(8UXz@4DVPW$3-`YM$60b{sx5y#pYIgb;2Bn>WJYqx5$Y>k}~KmCnf3=zJ~;a
z%q`W9PWIo~AQN_|d)N-pLsH&p9S-zDaYlhFy|?9*c$TTX2L&|LK07s@RiYwPUaLXM
zX+Bs)<6^b-PdVMzuBN8FL(X^`A(%qR5h6`m&WuUJvT2vIyd45>1moxlIol?QD;r=%
zos$*`1~Ho8R7{sS*FUA_oh|?%({i3aW6C*I5d10UdkZZ53H1*P2bh)XBNy1h4H`aF
zE_7PaQBDMt?sE69Xc?M|<`@9X^N5jQq^0dMqo|!z7e^~vW0YqOOPOv3*O;nu$sEVD
zNN+1B2C;Bye1-QJQ$-MlF>s8{Wp#?nf=1Y?babu3>36wN5f?1fpoJ@f+8UExYMWna
zQydOAz*Tb;Ml8Qs2F2@sK=K?QTW^)YF~G(aEf8*@gF}1(^c7o}@){W$ZIsmfut1z)
z{x&xur?4I_so~LXEx}uO{55TN#2iQd!+9(t=QvPbNl!anLNY3Z4&wcA|7u=Fhs2mg
zsW0yeej3Y)dsoK>8CQ9;Tx0)0q8A){*QQCZf;rh>i@e*PX{|LM>UF*vJfpc$xq1k<
zQK)z;WX$!kB{31!W-p8W>oWio$N1Oe4W2rdz*TgllzWe`dg-c#ZMs#)&T*Foh*N`I
zh>|aeIZZ?Bhwh5A7x|~-oaXea!|2B`q^kk(Hr{3m>vTbjOo&ebe{d7Ps**~3VQgcj
zzC~^@rWi4Jgxr{TkI_)fu#Y@T)88isvB#i#NUcnY;X|pS+jBL18<TC@<I1TKGR24i
zJBku*Y9eJ(sWH{dfPkBPK2W!C(&(;6n>=Zzc|ROH{Y_gTII7}ksGKd+{YxFBliB<~
zGg6!C6;`v&G?|&{8>%LDc%Xe8Q?p`{@D>4@yg635tQNXii`-)W9yeoh`E=GRxHWie
zOr}e2E6x$(mj)2jNA(kOw=9y|eFF|3CU^J%G}8blBxCGt(0BTD>27DT8tV<+l^TXU
z3$i2!BEF%Q_iq1&0XoEa5kVxG*eiur=6U6%+Fvr44Jo`@YP?~>FSPF#DlD*)$D)e5
zS|6s~wT|C;oSQ0|js2Pj=j%Pob;0}qsG6nTdouxKL3Ir=Bm~dGtCN>Te~G6lqgGhh
z_#UGu4ut9^neF2vWJHONYI*o`B^Xm9Wu?jbL?&1R@Y&{&EYJ!Y3VBpx#kQ0d>jbeb
z5VT$xI~Sv`;k9Cb9;7v>L;tk)u~P1}^{T1KNt+YGuuwbkZCB*|G$7JWM=i7f%;dk%
z*Y+41kSRoFX6)oFM5|zQR<q!J2s|kkNQW`_tPDcGI{htDiQY4wj&^8u`Jk!MS^<c~
z(j8(mQ%j{hi)`JVm@NYO>k&m$FLS)fkm7cg>MCGOa}yoxj%uVo$UI*KoD1p$A?#L?
z4IuA{1Cizpj!IM<0||&0D##3?;PjREddt{>fNp3It&sZ?W1EXBjKm0(z2Ck;tXToY
zw6UBZ#{=FDQAtV)ddnZQUCSbzoZkD4hsHHFHRt6a;}MJq1?%PgiI|LpwF)~F?_1^J
zG(Igo0%dufJmSwZV*?Ac6MERAsehn-XqBeQ2mBc&nyL(t&>u`xVUVsV5MBjC!AvsA
zTKSNTu7F?+gpz#Nzu~rLP6!2Y{C!*UKRxEX5$a+nEA_|6ZErOoc}^4P<cW}64Y+h0
z;`&MJs0up;_Q6M-42G?#P<18zqagr$vej)FdCFfzeG9^BIwE7fr-KgEY4w+N^30rl
zR+J-^S1_&N*`N>3DV9m6V*OlvkGM|y&S6x^^ZpEbP<?LwpyLZ^sDu}}^_a27Ilt&{
zqyDKXNV_0DUP`>0E^`O3(wrwYyHQ^D-)$pv3x^_X^8aH&HrkQcJFi6V7QF&F&b&qD
zGWI%O(N<e8^V7)bE7K?hSrFsHYT9KN>tvxnV}>+?7ZL9-i-P30WUyGWxY#9?*%IV?
ziT|R>+!zM8m`*|ZqI)z8N1IweOYIqTdaVr!Jxi8_9OAXb;5jRRvD~E1m_{fYD{Svn
zLrB%iN|S(0q*&#BlO1W2Fd>=jzEoDnlyD;weN3odE^E@prp&S%p!?wS4AgmTj2BD8
zT(k{1di!k@7hiJ8C9=+YQ&vlok$3;C_b(|8nNz~jH5-Cr(1&?l2{jixO*W=qY%5WZ
zTMS9POg4Et3~)t`A^owk+25v31U0yXzFUkch{(nbTw8r|422-T4l;lsu+h*pV}rMu
zc3syX+tXl~u1C-r0jcThqG4K^)Uq=TV#D4CwOtN{DS5lBHDs)yh&awrC|0uD-{!K=
z!BLED{<y8(?RQ))pYSeO`@y_Eu1NWwf7ORcYKP9TCk_o@76>{)ip792Jk0ZMpy7Li
zX%V=M8#zDeO?Uzb4z!iBFEK%}yaa)HIs_0rAfJjZVDrvaC+Vlt+&ANl;d$c;Me>={
zR;sFoR#Hg~`E2Un?nr<RBcDstrk!W7k3}U}EaT^cp6!J;lw#3(kY5NLrKvlw-i<GM
z!^Zdge6@Tj)XzyX<jX!}t0-a)Gah)`<SQ;9mVoN(<f}12MAo1@0My(K^0l;h)PERg
zKs<-$=+|RtAU!J)@KVWd#B}r+rfo<8V&QrO2)-FB!aF+h@~xmH*qB~WrrO%&<=?h}
zO)r}^9zL+|_`p2h4pAyw<hwzIW`Ie!cul_NFCnE-9r-Nh`h8btk=gcIlA5_D-~#fx
z57n&`g(FHK(jUY*!RuU6q{*sj=jDfqNb7Nq0=@*C`6GWu`OFvs=2`qW##c_CV{^6q
zBvDf9&Xq9PPyKTXem4aAXHEbV6{*k5&!bcL)OmaEI`{~G5wt98sYi(VFKwzLMh=o+
z#V(l60N7vqx2B+4y_DYsG4j|MIC+1YSeLOfDl@U63H?BhV;lU=$BqLE#{}(9GAnOW
z{C)b;)FU;W)ULN9=nvjI4JD!hJOKX~o2sF^HP_w%9PLlu%2MSU^Ni6WXE6dx<<BLw
z(<@1!h4PNOW%8G_jI>&5#H&A1i&3Gjru;Qp#^H^n(Ub!o&>Ra6W%*mss(Jd9Y4Z2@
zR~0m1`A1TZ7+qi*wPA;J*n1eG{4-V$b`Aj>FtPlrI9`VF;H=|0KtI60)7uLOP{U|g
zE&oZY0pu0x;2ZJpsR->v?WH$^UR7G$3{T#Q-95EhWUj%KEbrx5zw9?RFh^EZrI2D^
z5aapx_qU;3028td>jUOGP>QR#G(Yhja}~X;9GjN|=eo5|feu4iy&(sABal%L#*98#
zDhJP1yyoJ$a!3qF>t|Blua-l7K#F^*fq?J8U|Zy{Xbb0(1pqCT!{>TTs+uZTR7hGg
zL;2@?IbyEkF1Yjpad)WdDo5HJP~5JrvA%%CJ1UK0H$)cQs0-xixen0|897*v@fR!!
z#&R@{o$C-!eOs=e6{3!dX0#@`X%ol$@C?1A(B?j2q5x&xQaQm$an<E=Vsu!J$d8kP
z%&hI(*4i#7dplImc7rH(@wa6wA*5E`>3t(t0$?HwtS^vL{2La=SfJ%G-u9=QI#)65
zSqK$t!8NI;%~gPUS_LZmogRZ3I%t%f5$hV&v1pW>X>ZS}tg4W+(%=XqM=f0HypYvc
z(X;K{Nm;lg&q)J?y@0@=2e7W^`uE)UZVZA*rXQoiiJTW4L1PjuiX3W85mk8Q{L~ur
zF@Vw`wCIBLjKvZNDJ+u<jZ4Q&s+5bou|DiH7WzZN1U^Izpj@1$Q9R&vCr~SeI)8~b
zOxu_U*J?P7E=`?jG8T8QTxN?mZOYgg2CH12K14%2EMenaG1p*MsAYQ}pkAaJ6>l>z
zUEU7PD~<FO9XPfo0>NRSTe&J&+G-@xiX4=Nfr^9>`ow+)#~O}D_2D751OvUd&5@x_
zfh7Y}$-=bdIX%oiDQ{>*7|ifk6(CJ4zGeZvWI4?UlM!<r!Dp%kRPB*Ih9$V!i|8mU
zJQ)?l#%dAe=u0ZW=-57cp2soqJhsBSY%AQ`CRfK-isG7;X4(DM1SPQz2u$pR9(k?5
zO{|R$LSC!K({-D?JAOe58Y)}@PNGYW7VdS4UdzA)roLSNU%eJ<KbH9~rgNF9ZtF9~
z+o8TpjhN9y<%tf2W3kPFVvhA^TAPwe5kz@~bAYL0sOzo~4Y<w_>qi|&qj+VUuYwy%
zU^=>Ib?*R~*ec@_(-uSAs1^j~Fu^##yllKw8V~Gw9DI925I`FwX?DnsL9UXMn}b^`
z6H_PKH44I#2}x7}rCViEnjZ7fERc9|^q^s9tgSm=rle=^s9~CmA-SgdGj0JoJ7Dq2
zO)(O!cS1dc=MRGFWtwqlrqG7c4pku~Z(x#dO-80i`zlBbO`Kt~#+_%QWv0)AXEGBt
zh&_2hF?uv)mbKo^tB2)gZwx^&P+^G7^vNxA_Z<Ys9R{-wp*FYrFl`0-XB1|<Ew~#u
zN@>|Ix!u;GvAR+2NDXV90j_}D>Ca$DgGo_@465Q?F&lv5xXJ0&g*cP%jwzR;ghLAg
z3^M+byL7;Z>!do7PfHA_az-F-v(z{NRC=lkMB$f06QtI^wCowqh$<vw4YSdvGK{I~
z_#Jzx&IpXM4fRrQqZ#N$N;B`6g4mJ<dsfkfNY_T{-%#Qj?HdBma!<&q8e9o9+XrZd
zO=u&<ME0gcaz^5qia=<3u`~zCad!km&}yY6(GU6#rpb8WoC&QxPhx|xvQDd=fdjqO
z8Kx%PJ9nQ3Mqp+t8d>ciuoZ@!GS4isVcMd34d|kJv|IB2%)EJ!6p(A}#xuNjZ|Gnt
zq&Ie}#NQ4@ufqo6^=PFtk=hJGYRhD})zTHKU(>+rdh4azxa!(-uaq8Z7;%-kT38_(
zS2@So6^1ABe&%}D6gAW*De>o}9l@9s3VDWA@%~ZjC%96{f44YQMMLmAEjwN&yGGs{
zy&`b3&2QelFD6`FXqhedryh9Hs#1o2SPzPT!xnkK)(zTNEe6e+w-3fj*1?~qw&MHZ
z$uwlCk%i=;M8=gGW4|6Py*~~lNCU)D$ip#?e4({X9!YXwlsxKRn~j2*!q@$Q7$?}>
zN?sGgHeVOvxeo@XG`7$T`H(lEo(B>R!s)}#JqApu7*r`<Of0ZiN7m+fER-q$5;aJP
z){9ym_m$@X3$HR#uk6!}@<fbSE16hb#V38lRF1ix3S`65A*v4jP(BjhV1Lktp#b=3
zP_nrZ1@GGn)$){|`Q;T=Rq}Mq3T7G3*H`BQv3w>*I;pXxCCe)SQB?fdpq3rP)Q_l`
zDQE*}CC~Xxl*=gTHFgNVraT|peaNU0@<J>bmWY*HUknDH(ZV3`7I`Vn0rj0QAO`Nd
z?0r)_Sris$T=YbK%)jCNm8j5y5<-3QN;J&O6zcrFww#kP$b&{c<NR3N=?KZl$buNx
zm_i-ma%5rr8)Y1uAzBud41sa2^ye64Vy4^T*uo0Bsa~L4c&#i+OUyzh+ztdi8R*m(
zL#XLS4HsEz|ExoL2zUB2|IUsp8m~OMSW0bquz5ok>4Pg$`^zyV3P={(+X)?brH|Pi
z@NiENf3UO{pir_Z)*dGvU`1$CyJU5=rlIpz&6@}Ha#<5Zwu@VEI6WY1{TXp!-nm2R
z*PmKo_%07#D;v91*7+|C+osjUpl|a@tT(APeB^N1V7s=p0QcEO`#b77Wn`04Cx^Nj
zutQ~Yj7AMC3lHAn&v;fb<7Ef@C0i5ojT|Q1`~?KYtMPziP`0OMEICEpc)RR)`(k6b
z%wOVhU@~;4D{@qC={lf_W_(Y1>ak$5D_E!+TfGjKGWqmLPVP>W&(T(K`MZx{8Hmnv
zA8vB_L@XY58so5tlMyQ*F&P#7s!Mt<_%hiOEHNPqFmoUg!V0T%S@wn;XoMHBg%b6X
zY34LYc-zJ<*%#z$%@*WSK48zq20X*^M+oMSPy66XAs-wSAcJw}XOgY66GXH0AN9N9
z@9=+Pimd<s*=S$k2J>QDaLMP2nlhx4D{LNsQ0(RNF{N7EC!tY+Ux;y0sZ>H~0HFy>
zcu&3<Jz^iB&O0o{a7BJ82v}%=v*61?hM_}-$yZ`^pgkgr0)aT&<*UZ96)hd|wY01j
zCYYcm`Fe~GsU)}p348h*-jp6ddzSH7mTyMSRk=EZk4Dd#&WwE9I0#or;4+tPR2oeq
z(ZmqQ`~~ve)V*PBR^|~m1<LnQXFTbN>@Jq?``fxq;69iyzHMLgw;fr-=3FGN$IwtS
zeU$tleQP(-dU+^V3qLH99quvi0p3j45BH{R?;m;FL{nIpKq1gz5g>Vt34pGKk#OUd
zs;57RJyZ^*8t%NG`WTFvaBAWd(&#|>SsDZH0{Wl93qOzfX&8l7qWDD+d+4ZP^2<a>
zdO}PewC(<@wA~?nF}%%?#6TYC%+^!0Piq<c+N{}$H_30_Hpo)n$Zt)c4fIYyWl|$9
zqv5a_fA7Q7%EOS~{SZTcNbRs9FHtJS#`A{mKl(3RMj{^2MesXOhCewObb=twZU{P4
z7fc=avyA{Xr|RV|_V+xC8~-&%Hlb4fmPoaNen$=6{d=MgZ5|%Ph{Fd<ZIpk+;_|LJ
zyuAy;=AZtI?{Q?jckN%+Lt8h(Q2w1p$<m<=AYv%$YP6{+yZk3cfTBo*O?gM!8*!jd
zCGH85y_q`Mg$0A64xWE2UV=z~l&ck)r}o=G@2m-B(`CPTjtv5^&%i?h5M%#&hLBp~
zHIYxmdipbkJ-F0R^xrX0Vb3eiy-g0Br$`|Rm7>P?LGu(-L*Xe6Fg)1)Nq;!Qi4U3Q
z=qXZLvdx(Jq49t|DO6mS!@Msp68RP$J~@1zK}EEo=?SFm48m&I!bY%QAI^dGazsoG
zKJhj=(%YdJX64ZpFbj{GXQ}p@LVy++=SlUIAO(VD{LJF6b6j7Jjy9)GuNo`I%rm6e
zn28Z!bfHt*JJx$+@Dpv7HPE<@o2PK%xF)#uj*oUG(G7cow?iO>DZQ*CqZ>?4G;+a)
z!IcUPa?(7*cFEBkBy|UaU>X0oQ%;_zIP#>M<(=N79lMyco)gHVkW-v^;3r_lR-np!
zoaVc;<MY%+c8UwTocOdfDTYaFZFo7|TMLY{c;6QfUhK!c55R$YhD}4$JK&{0(+39&
zRM*c`GFC=ED|O3&WZrd#7}4~M_zG>L-EvN9f!83KEvD{5IoC;K$g%}CtET{~z-@A#
z{|<#w3!w0ZOd6*wYQ7tq!udwRi8JH^??Zt*6&3WIQdhWeo+I73r5Ryln9)VX{nK#W
z!o~L28StTAVjF=9bw)1rX_y-jD}wEKS*$1awmC^qw*&!*)HVr%T%IQLUy(CjtLchh
zBUtwZxzgVT>nMcFL@9#qz|Ua1?v|^3txC`^YbP-}wcbJV+-U*K0Fixm57l7fUBIqw
zG9-8ssy~#ep+P1$#4N0kd27RB#*78RjOe}^o}SU#hx~;-HDaFQro8(KmdMDUil&g8
zkC8>3h|KUl)lu<23hpD0K}Ltns8-tqB}(IkJZnwo<!W08+{z8#<u!>)DZ~NoVn=~N
zpr^~VJ|DYZ(HO{N{JRq)(rJMm2R!$>^lYU@Md3mfx!#@)^rAXC+$%8I+odd8H3S#2
zN<92oCu58a4hYehu~HrjIvW-6WNgd>d4!1TkqY~D@F4yt<6@!eo0w?R0H`1<<L4P(
zgRMgR70QF91;YgbY?!xhGO0izZkga?r?$oDJ0`DG#tDEfJ8-!WLOX8=X{({A>X}3X
z6bo=kfNZ?}X^Y%wV^hZ6EE8=Lw6-?OBp(q`BZ@-U`uLDeLDm+r@Om$dA(MSXGyzb?
z3qrVGro@P<FdVyHcdF5)ENW%k6eO%I<Ybz$>Ok+<SyGh-#oabNewk7(Gt$nYSA|-)
z7vjd>S%hn5;*#PW3mRiiF#*eNRvIZ|PkD39YPs3JDY-|6Z!2^}u<927hGpK}t0A|}
z+jkW4Q)&X{cG(K+?Y7`(TsTtOi3hjGT7#FEehwrtFL%6+rwqa5F}X_aG=ktg9I1E3
zer0KloZKD68b57{WPC`}YN<kPfv~QQ7iMQ?N{zpOpm#k7c4H6MT7%Qb+~AL_5xE*^
z>QE@^oCu|mPiZh=eGmx{0edu4lyyT)nQZXi4f3Z2sTuM-SuwS((rD~}osBvmfVi<u
z?r{!_P=2T-x*{+i0psHg(x^0Zsn1U9X6C>;$e*VF#oBv->2(x&zeQSd&T+vQ9Kgnw
zEeEoaE!mPSB-thy%!F5(u|_juf@B$6YDwg1lQXiim2(a-8m-7VvxFsw1(vX+W#Rt*
zRn=$4d++mo_qq48&G2^jIeogTtE($_1;U7~)A2hvY(ZkqdQ^LBtv8?0px7yxgb!O^
zLf;Q+CgKtkT8Q<wuwoqt`8Io1J|}gx@<q#D6jSZbsmhWeoy1}I!wAh4WjJaB`BhYq
z3Ps|u+-HON4k_PYEQB?(Yvf;8v!U9~`N0_tS0FD)Qd0mvU2nZ8#x|*~x2?0y7T1}F
z#?}Roo3f6Pq{zYDQca!M8K1G;9J}f2Yo<@OS67so(%)~lG(#XY5P*T~J7$p;K^|mw
zYo%&E80v7rTN)j;b&!s4S@WT-E-C4r*V2GVhal4sH>=4267=A<JZR0Cr{7>J@^wj$
z(r}Qp+*aOom&9GA;(hEjl^%3XwAbcSuB~#)oXTH&UG^E?NaDPX*z2=SOtD-U4$|8j
zVp^@6(3dHO`i<EmbQPSbqMOo}zNu-^4RanMOS>40Hz(%Y38~#{ZwaU_>egL|Z>=Dn
z-8SFe){It^n=|TUweY>Y0bm43>m0r<UWEgPlH8sT1k!|m-8<?O*<m`^l!xS<3H4Co
z)#!#%<-)t7Q|=1pw%~U+@Br6b4TG;}Gm@>M<!+mMZ!Fb?<1VuIRjNe99QNAxH%KE=
z;J3RlkjCt69|*SWWzVw@=G?VMWbss%Dj%MGDEsi;!x}0)GYlrI5H$P4&D0~kB*A3%
zkvekDKygb-5Bbq%XdUE?vsKxxyE*egDJW$zfG$A@NLE_$xjj5_R|l=0#P}QIpsmU8
zgs9-EX<J)o;vUsv{wrlz7pH>!qT4$-+<RL}+`c}bckA`H*oJy2Dp`umN}aYb7Gy%t
z!mc*Evq7=r5>Ww_uCSAv_NIEYbRl8kB4eBDY%d?eCDyj&KsT^gb%mk)FSo`c(z5gs
zX|OGdm*M^-#qsuhBZIQtziLN=pt3ZjTQL*dU_0yghhX-^PVUM_NeKe|<FC6YYB%hz
zBO4&7kR+rPML0R+7=GAUHjmPD%gx>#hnR=BI&X*V%R0YZ1T^73Cfi^8Q5Gted>ab4
zF-Nu?XvVJ06WX{ni0okH`xTCaUt7x#)wRE}P5EW*aK0O27oLGWC8fDO5?fo692T9U
zbvIAOEBsgmAmVKG_dZ@fB^Z$dt7RXH4q8%%x9sD!JEZ*~8?GoSf(W>!!w$@kRNg1@
z3mb7Z4!7Q2Hd3MM<}o|jbTJx=5fZV-JlMZfl$?e_#_fr^+3r*w#)<c#s{2Hp2;8Cc
zSR9x3$?RA3M$|jD>lRmA$m#V{wTBK|1?|&yy(FtSB|2)K$vS$F_7@xZM*D1i3(ecb
z_PKn*Ey8?PDGEOyPmrgEP!#(@eF9f<?T58ke!ke?fo{cc#~i-sW={Eh*_YyY=;$N=
z)|Z?8#@#WxrhA@!C6=(GkL!r+t3jS#^dkFO#fXJN0}5gLdc`IV3FUTFXte~eE%uED
z@fDIH^%d$J_?@lx&E}oP0T$dv=&)~9`1q+R{mXXycHL2(Bvu@N68vk>iSfibey1+-
zRRr90*>~%s6G_Mn*|6^gl<0jwZb#p5pd=m01~mG|?B8-Mq3fw6P3D;&)R}g+Es!ed
zhqc_cs81I~{3zet0P85>R*4S73FgNQ<O#xX_C)&7CKr+Yq-l>^aCW+(dGua@pEeSS
zn0Jk->T1LzN_duK?`JU^^5AtX8Su-1m)Os9hEko#iEcZy{zV;Uie}O>1eX1>&S(<O
zdl2A%&nLBhF1$h++}vruYWkHz5q_6d{B`rRqPk?*lG0=!s=nXk(-Jth(8EzM`adc~
zm@sRG{Wf1+&oLtx$kAZX-^F7y$XGIHzpw9H&IJ*F$W8+s;yQ5P%zkSA*imBn$7Tq+
z>ODNL{*+H^Sh@viG^X><`LxnfiTQ#g*k9@voZUlms=wwFI(q}KTkOAT9qxVgR))XT
znT;?v4$$Rrrr}TV-`OHUp{#2MaY)gYKV*N8enwVxA_2~eh?9TRzByDxC=esembxC{
z#Opv8oNVoGOI;;76}o))XhO(FkA}YH9!o>3$%z2#6xzUhE)C_HuMv|}{<^RL1iOB(
zd?(Hd$7lPer5VG1#%Xr%Y}ffC6Q+DH?vs6nG^6PVV@B_zyl=dJS&RL1zkFW;h+LYE
zGWGi}ZJad+(U~g5;{i)U=_u|tVw~Alp#64xU>%@uIPo%f+JoxToD9<Ai6;_-kJy9r
zX}wE}D?}espLSxTfmW;KL-T2o0XS>EaL^vsJiQB6K{CBqayd~vJX_M`UDulVgpS!G
zvPGZ}Ar?^##oQy8rrrsodf%c98-CQ%)U-|Ng00r{9~}^DCkM4XCOeDMR;D*O)GF`L
zW8*18%KGeaqo0&{0y~~mJ%l{k<Cli>k`n&!vnMnF*)9bfNWRCOxYV`Rc*FGJY;8|k
z>YDAVA)X8Dp98%v<c5SN$E#o{0f?!jcuIE1j&Qk;{5*DAgJQZ)$=$JB)~ICUsY_EG
zj#K(#9Wtlq2YZ%~7vX7RCgOq+@YWiwwWnu`+#tjg%d#AV>5S1)P$yaAn4OtV%bfyS
zqiZOH6I<-8e44wwm5>Gh{N;9b)`=b{Y+oGmGqNr)#Pf&QGn;pHomV(yoYU0lI&Xbr
zwVhk%Nai%OR!i(zF~@1Mr(JcqotJM2Hd4uQBch$Z)O93^iy>4T>B{}6$K@<W7e8Wh
z*VqNItR$A;9(cR38G{n3Mr6wP_~gi{3y6!ps4j|xLJ(LiRH)WP2{zirOT*d_d)S>2
z1I_OuTwIc$5n@a6fb*hV+Vm_jJXDF&HDj$n#yV!tu2aydPn{xyV*@=WdlTW-NyA@z
zZnMZ_{-X{Hq(BOlcwV-o$m1CeXgRsko?k&R8;5LrL5|0>mIN7KRC1k^@Ug*O7*B^@
z%_hVix^l{_t8C)ZFf8=iqzX>AjmmR-lTFU=yW42=B_u?0n9@A0jAT+iuD8pwPBOf-
zI^qH^uX9T-i1QnGz!%xn+5+b&1omhGC->8GHgc{Yz)&8W^tMU1Ag!cbQRhCThg4rz
z2HMRYYLAOuZF)4HSOMF$!e+z-mkileIXq#j92(^Ke|3H;j0z&{nS9K_x6x+CJESie
z8sr9Yk7IpNGw=xE37J#3w9cwPflrr9h4Qh?4not}Jx@`4HYbW)Z|>P{uBlV$(j8UT
zX3v~@F88OD2_jrqdlve!f0h31#aV}S$DKxDZ|o%%|1UXjoV_$3(76VLJt*g8b(79k
zx=#+!>jN32akZD{_v7&Na|SqVf$Y8_yK>8Yo+&p}5_)-yEcEuuIz&Z-gT*`XmTimE
znM|79ZWU|~$~t+*1nY?J&RRG~-q8GR?0Ec7IiRH@*lw*{AFY||!<dLRuM!-{rZG1^
z8bLzAO3$m)md6GI;pTVis&F~u+-F&Lvs<?LsZB~q@ULN8?8c@&;=0K73g(S+sq>qe
za{0XD=RxZkGf+Otp&MpALSEg!WI@xX?y3o;k+iZt9T2g%BEsAruBYmYFQ6vCn#u*(
zy6k*mMY}v&)sG*u_Nsc8LPm9=R<HGsE~Es;<ehbG)j$OfL|>{yF8i>-99t+}M0w=C
zgkcHUN$x5`6=$&Zx?*@yRLoHlMAGjM+TvKuo4fjk?56lV>6&SCY)P#5-0pU}IjXyO
z$6kFGV8l5Exd7+G?d_I2p#f|xME?xftrgeW`xXO>rAx!U>$DJPlVQb2%$-XbvCo)%
zl`u2pHRLedWXobC#P3^45Qq^VU<-lauq|)u6ak>gudx+bS5v;~#9CRO!~83MVO?YS
znxr>G7;~t+Hs;B7Ol|hMyUf#bDKNBAUj6kkD@?#{vVTM639swx@0_VPQF~+bHdP^O
zZ)z4bIAiSyA=e%D=876&se|{H|E|o{Z>i&T)k=`$x7Me_bzE$f@3zDCwtPD1Ph4I1
z_MC&$%Yar2b`kM%TeiDk#kpR&d!<1avA%r_462Jv5_XQtUG|RVn-b8)B5$>K)=?vK
z$W@GcXzg9GgBP@<WqJ1~<IsaG<mM0^-jfTV^LVGdx1!8^5?u_~`?60RxjJNVi?{bz
z&a|DR6Uv1lg7|?PaA1{AAB9o~E&O(I`(R+w4d*=HK2)j7)syDfhntCd-Zuwp@C3rc
z1NM<RDQDtD0k+#mvradl-VS#nKkZdHi#6O9vC{5{uZ<snzO9a3=h=A4k1=3t^6m95
zD)AR^uC06ZioO9d$&iCIFla#^Oy~Ms`Jl7GmV|`3A!gD>EbX9eto{0#EUG}Z(Mq2t
zft$9ePJGssNwzsBKE@aXd?hW)9=6rCRDk+%?hFa*?1~78(0bb%U2|XC5V@kZ)sKU}
zl7a}8x9#-{a#Bg`6QeKJ>}`N-M*!)X%ct7Td|hq}GC>m0XS?b$bDdSco4q-RcP~Bl
zU{6ej{zP}gr1-g~`ILM$HZV*fVEk5g%e{5x)4S)}zWkWTFhL$LdwXnuGk~M;5<;SF
zb|5=VTaq)P5!ki?a<$zK#xbd2LIO|w?NFtE1YAPT=Pj|rF~iHI&9Wo$vlr#m(PUsl
z(vzd69EW9i1IjR-aYi1?uZK5mgw>4=%sXBoj)GTqWUbl9>WJOTG<uVAO+n~Kkug7B
z(P4UD-$EM>!i7R@XhHAd?oI@%6BP_77L*U`05aD|JeK6He~1v?lMMu=BnyYOf2nmz
z#}pTLs@CDaDy6M>w@+l9LIC9qcW8LCeKN-_enEC3(Ke9-<xYjGuutU<2>B)=k;JF%
zw@=rFojvVx`%HGNZY0hr<MC%>%ET=zP2uO7k4GkExuAVMx)|&q0svpAUjoOKCohuM
z%3ATos<P<r<pzY7eW}87h7@;Sj^*uMh=XP9W#rRV|6*Mr`PG~q)C&$tG%UkyfrzB@
zuf;c?_ka7xzFvvpgdQn%zL6b?VUlS=YUT9lrDl(PGlr;pD_u45tvckE{4SQ*w`0h&
zTCl_|`_~xqrQ^=C@6@^U4#5_Dw|<86R36XA*i2)+-M$w<98q2bVto8Mi0>z0m5xqq
z*A7VMzr_b;&Y5LDsLM1J5AYx6vvKys1s6YR76zjTOGnlM`*E%Xd&A@AbUk}`e*7e#
z79b*Z)*kz5eC-wIoNhm>V7sat<hGyZe8zfMxsEI0|BGxN<Lk=|l&W4L(Qf-?el5lU
zjl>B0?{&N=ux0$v?)z2t!@d?9iDFzf%O>=}ex2{g1|YvoXrcWkAkTA$;Mf17F4$GA
z#X8h}TUm>NmI5qW@QA-lpu}O|m&Y&-MPC2D&KgHS9QFQC*A#nYoBe0L29G33WK2ZW
zA1f@~+M)aI*4dw$*&UTfh<3F7IY+7<HrQX9wb50R?xRMp3Pbs89pLb04qwqAISk0o
z`d>|p;??3DlVjgT`&%6=xpaE%ziX+jhr8?ko@0@}f3UW@?H_f_oF)=N$$(_GEc9^t
zB3SFWdc-Skw!1BJrIBYn;5fbevd}eZ0P=FcsoL(5b-_z}dN;f0vd}*YSeXC|f)F`Z
zSU|g1w#Q1y+9sj}*<;J#t!V#L-{&1zJ{JLNI0GJxcJF1WOT*Q=#gzsFcAu!xfdXqJ
zf6V*FxRg_-xl!kS%Yuv6QL6NG8WD1igE@2=jm*P%l*jh|%R*xkUP3ZNp_hZZsaaWJ
z>_F)UEKA*H8-85M$Ua~XT;}?<shB30*n<+lF<)&B9GXrRX7b==u66D2Yj4>@@~P^P
zv?Jx6<Mz-Bcef?Qf%ULuq45(_y44<@PY7Mah|06@5w$;-U&gq~L~pl8Ht-!KQ#!E}
z9|oIZk6LyrL**K?HSa5ZU;0dqkqsZc%oTkE1xAdQJtn4ff~B{|1{6kCo%WQ|>~VEG
zArOQrm65Ku$7hR9pEQB<rDJe1PskP><*5`)((}Zog@b?@OE0w%SZuZ@)fQZ~<w3g~
zHe0C8RY}WE-3Jy3n<wHq85W+rELozVUb575c60BkQ0FPR=;>hwO3F7{;T~g^=%-aA
zoQqkGpu8pa)LaDllj`0(l62VV&0r*cYSG=}9IEiNY+-1qze2w(0PL`*2T1X|Y~e|Q
z^mj(CmkvNl?!r~ceVX?+R4!-6HsuQ40Y!|TRiQqcL7g3VLauhqefEs#dSunip0*`&
zd3t7!!tcOjmO2%JeLp8#M3V#EBOexe&!$CjM&d}73=ST}o>d{hoFuF5V{y;Rr-7Z8
zh;5G79`huMz4>-rpCI`T&Jw&Cw=6a2y+fFY(LyN)-v!yZFjcWJ{Iv_0r9u__^Q8ZN
zdXHb`I=(As*+l^g6ws{e#iJd7W*apxyQCSNJRvn#Q~^4F?9#yFK3qrb*>zqcTy2yH
z_?+A%;j-&zv6iDzylPz~nCm=#ZoZEr0$9Lb4xQ&k|C3lXdwwqPs2&VX8i=~DDGU69
z3UT+Y&D$EC1A5^Y80z{Q$R`M*(_}*SxsGq@JXQ)WA&?Zq81QUjC0d=Wc5>E9M?Wd>
zsGlPfcUhZUQD*@r1DldxavZ(OI$oBu6Bj^^2)?()F0Uxd3BwmrfJ%9zU}tUYLpC+Z
z*D6Bzj@z{80hfXnTaYM3n5--6{=y3c#a+h|yE6OKv4W`LTyPv7`tlx|9*f=Hujs-V
zb@1r@dhMzlVtpKT(ynfRs?-$j`w8eap=g{La2Q!Ny=^{-_o93`&fV~;oRu%jmY@?^
zDYSW+!(euN7#_9{kBB*S7KDbuNy1296YGBKSvS1eu5BQ9$Sn<IEU@e9in3x#aw9{V
z_RczcaXzi}lfvpC9PmqO{{(vtap?;8$X6(FSV>;=z%|~GWkz?wysWOf-_q{;<@!qS
zIN9XP8op86K=pLMfcWjVmsi>`c?p5<IAFaZzZ93VY&fZ;X=AbehJeJ91+3_Ry|RuA
zRiP-UJO#(93)-6H@Z>_^uwq1{dT)<;%L|#?@sJt#c1J+*((}$c-#YWnIO4+Z5CsOu
zM{AkeO{Z`d=G^#6Jc-R~z7%}sohQTRY?F00J?XFm_8c94Xj^xMJvjh5qM&TM?8c@a
zt-c_w`=cb6%&(wY&`lm+khv#@`=V<mOt%G<X7x(AXzv?B88**)^8>+KMPy)Kyfw0F
z!b@#owkuhI?qiZ?+p7ZMi6&!K{qgWXt2N&SD*c2Z;)V!?RS(!;_J9hF35S+;sG0Ci
zO%=Yb=IL9s?9^3}4&kWcx#2u?`imtG>rsd`7}18;!{gBO|K>4U9LUGzKny9ASU1JO
zEXE?br0zT7a@jjK*Xb{~)?Qs<?HWIhdp8*=8|EzyVDNTk71kg<-&%1Tr!s}qvF(=T
zW9aGKD#l-1RyWVwHnOUTZ7Wfur3d}kVzxZrlcb<4JmmhoqSir<C1`VjrLBy?wv=}7
zHFbsM)5cZNq0fG818e!+jm?($x;pRKSjGu%Pui#)oV)D}`7yUG181<+-q=86>{W~~
zc=o1XaYk0n;=)Kk`sREaIYW_G^ti{~k~=ZHz&$O$Gqp82)mtm}PS)iz_(r|0c_GqZ
z==}8?B7g@6y1l)EdkPBqK1joW-PTMt%v{oII&HAqE6}@SU+RNBen&G#zgR&!40~tg
zVP*{t+Pku|N}0kjL;T^p8$b|%0A`XAO`l<Jyr%&h3a8w#@2$X5YEkaWeP7e>A?KR)
zq~1nRT4ywMh`1Phv$>!tJCy7#C#U1cM<1*-4W+5#mq-<#rqs~be5e@&`ViU`(f8qe
zXYAD-x*|_z(__fBAIa56j>f|F(F%t-Tp+rt;z?JlMFi;``Q%1u=|Ni^`>6ZIK3h}g
z#j&dpT(XNeMsKmT`Ed?pM9eURqlALDjv(rF*)C~Ll)$q%JL}*axUCNePKF38eS>YN
zBbtrEeu-_&56i7cX?{I#jRX(7Ge&pmCFk3wX3g#tGv%B+IwjoKH&@WOgyeUF(@B#R
z%eDkK;a>au$TR3^GPY)K60Q~ck)at_Vz*T^Uerx~Sk%1R>tGl3=?<+Oxj3=cCHd^O
zodMF3Rj(R?X6(woLv9mhNH-<}#O|g=o|U7_`w`nyNiP>M;qTZsU&MWHFr+sQ+P>yR
zi7hDGMOK;OLEE1#A`H3(PN%<{=<Ps)#p7D23up8u1)m*kgw!p1M`q|ahX99bNL?3m
zDEo&6LO~hqj_Y#laNQPMt=EmF_(<KMgjOzYCm6zx1{R)sVg*QltbyAR&JlP0u;W=L
zg{G1w=n(u^4%;DnVk4rDSZ7juvF0+Ae7sq0F(>%wq5hUIff5VF>zy{7eaajqR}jVP
zpNMxx{-3Vu<^pdUX(p+lU9r2$E+k3*Wb}srORKAGQ8%Y9sX*&QNqiA^?w+bBGHFRW
z2=Iv<LP~US^at#dby)D)q{oEwTxXxEo$F4?RvYJWzkND~!j$}IJq!hmw2Pl<CN3jO
zpq18Ercqs%_}PktC*BK6elEW!G!3`YP4@X%4<e<dcx}2HJq#oris9A;mS4;cC_x9D
z)YzA@PRx?b%-9a>%M}{e!7)n&u&=~tZsgL3LHlaXCpH6d93Dz470}n>=sWK`99X}e
z9S3^=rw(+!k#EX9P>P;?5GBp#h<!7M4hQM_U74R{D3ZE@3y|zviP;c-P)*Fa{O^6|
zjA$}qmUh~=W6@@{$o_$?>|g7AU70AJ<&b@+83IO4hOpE1zX~P%Zno$a0XYS^5Xy$x
zZr^MA5o@EZwB5d6VeO8|u89e3W>|{g__u%<(R*Bc{DU05OlmqEj>!WulrdvmM1L5g
zCw=Ll<dz@RlLxiJ5a}@BgMM5WhMSM^EZ`EAp9I9^qh&wMVam?8-g*2x?PpmhExhVv
zr}=rF7F=8kMfydZ2&}zErKrs@7ESwQ5YNk%l@11$+vc;&0Kcc^w_jx^RcytB2MCTP
zDEoDNCuzKx=HJvM4@!Wg!39=vL@YUq+`0p~qOH5ze%lmVLx`lm3;2$#x^Uch`+c1`
z5{0hzJ%YC354oHoA-bSfw}c4257~d#o;tew`@8IqsSA{wzQRh$c;pYtFI)0YfJAB^
zZiE`%jZ)%IafF?9?rHYt{BkhFK?46$nZBW3d}qj;=S~cNtstNys4ff&-lX+lx^bw}
z<?g@5Lpa?iyOBgPvVZ18!mucE?P)RbUif>U1FY7KLm2T`n|}m4T-Mh|7<_TPASMQy
z3Z;jMi?0R)Ilcv^-mQd=Rys|oq~txSGhQ+u-UWqr*8xf<aNq_^AXtJGsw`-$sItVY
zyO*jLJ}T8e$bFztcV9O*&cM&D3$6M+awu#u=plFOf^mMCcA@0~F-QSEB3JI2PixzW
z+wwR7sXt3?am`A$cq&@ES2GL_3tG7Eyin&^-9$T%+CLT7XOFCE-Gtq-P|OcYbdU;X
zz1=(O+!fF<VneBQ;V&yr<FMVQr0%ejT^l-PmZDI7Pv$~kD_1@xh{cphmX+E!7%P`l
z3++-zj;!kDp5CFG?0z|CY2l=+!J8`n7FzDXljHI6yMF_ixU}AbjKne*T6{K$p^zK@
zoIRk{u{pFip>Bn``0RqA9yDwAzyJ}%1y-d{@?MnjH>DJNP*X>~B^C(MU8vJUM^;Un
zU=PlA?UrzP*ixvAue7zJmRMj9Y5F(}`S(i|O05$eg#-2d|Ip^SwSIQP<rnH6!?#$W
z(Hn=yh8I38+ciJgu<>=IH#Z+i@l<NNE5`ggG9u|ZW;wv^vxhg|+~f8NJP`_Ycjucg
znq-eCuF5B=Luac{@C~I6c_*y(dOj8G&LfNKEYBqyt1KD%d%RE?BhN!iS6hbtKwQx_
zdsKE8R8{=A#3c4&p~crhBUBy7tv$MqlS4#~sjN_;&Ji3lF-PHJ8kEy5I`pUK#0Bs)
zr&ys~yE0q=NS9@g%>ks(PjrlkytdGCe_!AMNlVD;9rn1pj5aJjVqcDyySDhT&ld^Y
zv&RSOwzYQ_e)^PBTjp6zo)9}yo*Q^<6p9KWR-^;6A-WStM~6MJSzoycV&ec?fRZyz
z2u_9egcY(I^Tpm_Ps*`|HDwGxRA}+1gR;uty3_u-Qcc3tIt!&IF^VKEamq)^H@6q&
zf<HOGpxhlwK7>r=lE8*Si?4bxEP{{+06+v-M?zblk|W5ZGKi8#3Uxs*CtW>b#)PXT
z+iBT}^h)B2nww9&lS74e$HbUmFwU-vx2I-{KI}#lze64kX`v-HC08oQsDFBXLxa~8
z!n3})!m`lr$h8KG>t=ge{UZIs$GP7{P%>Hs(z;+S3qHL8&k;TCiOwC(3oW8%u$6X3
zLc*<K90z0#E!6qMTv@?t>na_s(ndS8d3ps>gdgjLI)`<*SgB4iFG?z+)M(r6ton|!
zBGC^Si3)YPqwmCPN7lC4*%c;Tt@id-Z=vky5uRQ6N+9w)qnY_=fR#4u`a(-QjWi%3
zW*^)1na$Iqa1v~H7g`cAbRvk;Rb~TRDfkOLv2*IwA=lVpn2eJD!S9yZq6vtP5HA_h
z#JK^@frSHwpAI=u3rVV?fR26ktmeZ?L(I~Jy1KH{rM7ENcm^q45kx&iHxz7rUSKB~
z0^tVXPYM;eNQ42U&O?0e9d>>*`?ap6ma@h%vC!hP7l8|ZS9Kq^?_l1uaX~?0T@(}4
zk1Za8OJ%OFZohDPU0(tehwOrSq+ijuq@z_R>R4JxP9VFmZtB@HXU{H_;$p_>ZHsJt
zZJIvqswst1F5(@WoKh|xwTtq*9y2IdFk}Q6cCzp4Lc6~sSWE4wjF}h5Qe1fH`Guc8
zY!P42is#3#JeA_KxFlcKdWu&eZ&Dep3xX|AuuGfIpk(CJ<dG=UHCUu@G$s2PoA=|{
z6{81ww95<SW~~&XwBFG|Q4-IoFqtxYPNCp4k|YE|iZ8V1)(6IqyP!}AFhPLjurD>+
z^KvX=niU1}KO%Vw?b_=vKl6HfenkvtZ=~i>5W||`&4rdcY6hhXmrLFY>H_IvfI_JP
zTfC!`Q})8T@B`clJy0md=v|N&o*cIcStlG8UkeQ>qjidk@)n!etaX#Hh8FD}kH0Il
zI{<@!(28phtwU)uDZe1MHU)WbrWfjhl22^)%v0Km$<5G$$wp8t)OnXUB%_VdiRdOx
z)?`Y3og1-~wWKr^atv1B&}FeG6Lqy<q2Tj&dpXFZn`bU>>eRU(4TZWy0*TLBysZKx
z8t$oecu+`C5pRn_rPc{Pk@H}kO^c<x==|}8pShE@A)+(p;Mxq~)fL$#?i&D6f47uc
zCu`9mHk9JLuKXW^iv0mCD77V6%Fqw6@mSR96(wg~J)=;_mGQJtZftboXfrATDRQb%
z^5=v)QpCy^!ndYfl~d6cluww4+Kc8nZfq;HonshDRVFl<Fq?L}y5g=Bl0h4((y1v`
z6DhG~*3#6r1?`2>y}gY~4N0Z+q6Uy-JRjIZQvE@t3+;hR?cDt)P2#L(X=NQ2JF%xw
z7u%<;or`wF#oO!v5{Xf+nOP`Adw;84=?OuhIdxE^#_27T96D9CEw{(m-q*zE+7=NZ
z?~eI{OA2-O^C?V{_iQJc)N5-EuDbF;=qS`URt7H)!H1SyR~K%^l*xrcVvx=PvBYV8
z0EML`e{nM;iDud#_=<70#i2q=bPeUyt*Xj#^O9uP#%N80&aza}hH7K(EIbuk#dR<W
zT-s+Z&F^Sa@8BJU_KE;*FSO_+VMaanvWg_GQVN!QOt!K@yZ5YAh@ygc#@qE3>NC*+
zOI_Yws7k`UMjCfMCcS%@6M(tBJiF0gHnT<W;wjVx!f;hNbnz8|5RJoQp+W>!SL@zT
zPU%iFI^OPlAL<zSTNdioMfU>28PFEIvJTe$u3T?ksPmD<{t#M$ft&!M2fCxUt-eED
zL~QLSwNCc0O#(OER(s&pgq{UNP83QBaB0bPmC;)3s94Z5)K;3pbxf_ZSqjfMEYoko
zptfu{DHHRC;X9>91SQs5E5G1r-GsByF6t7sVo%-ad%01K+uVw+{gH8cpitxZFWToi
zXbzchQJGhdQul}$D3o0?XhIFernzi4#s$*4as=X_Bxy>uD{)@CimNsY?ZJjGYLk}6
zy6cE$vCaDX+Byq0&3Xh6CnbnNOeMFjvm5Ih149dtn@~KA){wIWU#w1>A77b`ld`jH
zl!Gk6UJY7LLN0Jk8)SQ_b<9p#58$FV+k$2dl^)%<Q>k^*uPXvoM*7}_p&W_iCaF|0
z!Mf6-?_P}q1WZ|9g^jj(^vBZ*#dRfbp*==PF!mx_7}!0rZ?N#wk%<AbJmJgms=7Jm
z_OvY?C=~q>uvpeDX|t@qoH~uKpYy-{r|_i1Tt~9>y+i>pK2drx>c?y#;CbS0XusYD
zZKkZ9cM-Qr4OSrWxTuIk?OG?!NSZF}IX0A^5F;U7zA}jM7YgV;M}>CZ%+s&5MKQHW
z@@VW`QYZ>#PP~n|znwUj#dYQ+gYRt>ig8Z(+H(Nhlq2Fx&gNhODb$GrK7X8Yt}SWS
z3-c4R4V$h+4O&9_Wq5*sZk2X(rthx}f-6Rfn=1w`80zOgFp!#DIa0jYtLp%9K^W>S
z6oV85h1LA+mS(y98a!?2QtQNC$%A??_Uc=M5>LGvCY9gWk3>8%X$Ng-9SR}w_znRM
zYWyh>i^&Pax7f0*)9T2dO{9W4!FwW!M@pqj<1QXoC^$igt?kbOB6;F2&t6a)g9dEp
zksYPhxhg?~3A4hAW}MF7q4(WfsEdPBalhTYLknzWgL1lbTjSKR5Q(M1@<O{-j6~Sv
z%zI5;<0&tmIg5<$g$kb~A-BFGj=&YaHt=McQqAI%SSTk_>q0%*{$%8kXPV^B-S)aV
zrhzt@z*+-^8hvxp)z@4#$6nvO9`0j1T&`@K1Pkq2bM!;=Q7611N=#w@@RmZUp<#zL
zInI<ZS`>R@wu|=QRpRzR4J|23p*?VV4nl$aE8mp!mH(Ig!?7b3>T;!|&XcZ05|NUL
zyt&S-A%N73+2CV71xkbX_Lc^o2qpBw&3uJAe@EHVB0Iw|4xrKewQtSuhnzQNdW9BW
zop4Ifs;;xQWgP^Ct>t8;)H*5f_^KbbZ;w@!{Z$^Rg|hJUvhKNdTQrTVn%YN}x+R6O
zmTO+C4;}mMIogo0lg))2kjaZ+o<h5vdPXpn{-68T-jOg65eaLHJVmz*uPd}T57mud
z+B++Gn9<hRSt#YY5pB!p*y;S0q4r&sa4h8Nt)4<L&dQ#+Y~|f`5l6Wb=TvmIN^eSg
z@JkcA7QpPhr#|jbV9hzi;=Uc&5gw;qBg8%S8hdY!N&8U2tn#JQ0kE#n;%gZ#vMn93
z_hnt=!9oFmey7$+Hzevn!X<luK8J?Kg)!2qlv*e0U#vdS)AoUU4iOHuOE7=b#pJul
zo`@p!gLQIZ$Oj8WM@|iEtByn;HUWhUTh~f{C?_P|WV?5{8dS%Hwb0VM=~_v27&&c-
z$#VN}4n|TX2by+esda2IxElA{u#eP1yM<3$)?<Y_fwY&)9qps}9q~X|i&!(|Z}#W`
z`rK7$cQk;L;Ft=oDmLS)=FL&RbCE1<e4LmH?Li&7mdw);Xm{ibvV+P9f!WvTf%tLm
z(=D`X4W(Ub3}>rz_@It}roC5cU8r^C+Ofgb<S;#{k)P$e0RU`osV!sGMB%!=Hv86i
z0KRMSs0#`-yc!0@V}Bh>WPntB=DKp~JjO_!C^U=c$~aF70Os!Eo1@0KF^(~X_XP6!
zgzMY-3I~)`TlsF}F>V|o3hj~x!_+uM*IL?!czqFm@g;wctm+!FjXCYmdxSRIx!FRC
zs1AuKX3ViW^NT_ZA(;M(MvDq?QFx(UqdWbES6^?N;`jOIoqti`x9_T8xFQyAbMwiA
zVx2K19;QcJLkcZHPaU4n8kmqRK^Mngd}-ll9IgW>4?86TZ>@{wnw*gA9k7L#5Quv*
zGW6pMLKJ2`TsM*2SjhY_vbE6S-wj*3BItJ8_U36BkKLfkUolU#_|teb!$PyA;9quB
zRJ}|&PY0G1Dq>sD>08n_*k(H`X;e0{LMbR16c2`ZR|6%*MnN4;E-%yt_$N;qu-z3f
zFq-Rf7}Uk?z_HO|0`}Bi;HyL)xIw*8=h&!|=O{1Y)yiK5S_B5~%~>YZb8$6hIocwP
zjY3c)8i;!Di@}rmh9wv%6k}D;3@mQz?;NoG`BCW?W%-vTMp^=AOJJsG57<m`2F4CF
z$mLnlK`Re)^zDPYd7(YHPFAvI2XmU?x#og;sdeFL=6XFl)GVMR1y`gq0j?7)w8VjM
z>zQ^qds~mxCgy@T=R%!REc$_T$M{jcf)qyr8M)20@H3Wh5=pfO?dX^llJcH8m!?jd
zW7Uo9wPVd90TGaY4`Kv${ym-$BhS+znQgV>@x8O2b$#JyoTkkknHugLP?V2_8T0B1
zg<p=DTFoH&2khg`1aWWxTTyP6T2~W5$iZ_sfYRMNzq{}|<~El*q5AAZGd|@nm+AqX
zfl#H;62#8Qm?Mr8yJNoI8*OA%mX@^X(yhC~0S&eftHR}Eg{P8a#ub3TUE3<AFuci5
z#uQ{kUjzdyuf&pdoF_E}<5v5ZW~zJ8)I!!dL#T_XOz6ZEZl`i8(R~g`5?!Gsb_%8e
ziH7n+5=U3B`us$;E5;dX2zy^5u;2JdzJgtac5TNgokPg`(!YH&M-xuHVoUZF>V(NU
zS%-Jor|Nj=1~SfJbEr@k+x`MBy!&*s=6-FAcoH6y%7~V+6bp~&E%ur095ppNpHdLO
zX^I7BAzSv@TBU&T1vtMA6sqpdfR;YRJ{RLBI(w)?S1c7OVwWPH%H8+#0ZE16^t1*`
z7Dq@@hWE35VHB{CbFoFb`vY;R(BebY(W@{D?V~ToXJ7El!Y^;lkVIV(VPC3~b5P@N
z0OY6(YS&Nr2$#Xymn+%fLiskB&(1=Pui#i?t<GGjfys|xUulLYIU-ijDO#y5FdsVC
z_$vX0oEG-g*nsVAHy3_3n@hVH6tS;mm%;T2sr)X4kt2l`A53@}N$18<2R&36n7m8B
z9?(7UK0ft+_5kI814?s+bby5Rjfx91W=|><>gNHlqACvCH}msVsXMAvrX{g?)YvCU
zIi&>qR<@7jHr!>CjVs$lZ5MCssna<(LgI+<&2N`eZvk)q&q)~n|9dKoNn5>RC{k|y
zYd{#~TfaohLOCJaF#YzOda}$S|7FP%AlQu~(Rb_Gj<FiV(MXGHPqh7=uGd4`S9lHY
zOVs>c-RrOz2MVS9x_q6(Dfj!$B5Z-3752zMk5FfDWE<>kZ5Qp3VX)o}r`QlBZuz%n
zgv!X}%qlH9d1w)xAu_gmVatAy-xjlz&g+G`*euLJr>$$KWf!6!7$f^(^w8JdKG13v
zN^$sTW<mrP!G*^^s$W7YHLp<0lT7ixemv11<F_9-Jt)Q^$gdP;drEDQe*owym4)Uy
zyu*HyV{{=}?>NKknMh<(E3|9DyD-6s8T@H8rLE$rIT+*)rM=FA0~FeWQCZwi;*hyR
zJ@&JxL0ZV$E=$&6*CE2pozH)s1DD{V;pw8UKm%G_2<nEk{vuuJf{kV=ILO^oBL$+!
zP2qqmJnN_+AxEOlhLbDpmzBI_p5j79oCD*y&FJ5AF+7Z@Q1Yb{SY)aedo?y8>93mh
zn#xnuE^dlL$ysO*!cL~*0q!9EbsZ0U<8GD<N3BAgo_U%Z_ST+hzp1cmm)9`y*oB%D
z4<=o1|B+otqnbe=g*x9QqJ4_k(0MKjsAE%c&t7l84ImEmp#II@((~HyYWK6J%qkQd
zVp5EA@hdBFz<!@~8z{L(0;)xwMty1iV}Hn188dqZQlYYqx$feu_dtolv(mquar!gu
zKVwX|X74Th_D^g4TX+<w4f|u(N$6DgzSJPKj=4Ffb9{J%{i#`qG0qU~sTG({yZTlj
zkK)&zYxr}b8@7a#E1(8<@>T8sjfEDSfI`+Ggob(cm(ihVV@TvK)Wyc1A*}psb|ywv
znk~7Rc~-GPOW+_Jq%=~?N4DC3HNF{L{lx7ID9*i58OQ>&-$Ft6w}8zv&pr3t!f&BK
zIF9{yv%y6og56f;qt%5LXL=;h9kstFP{NoUjtFw>aTGxxN_!<;obvyuXmAa-H43Nm
zJ*KJ{ezuLY_m0@%za{0AY$%|**>WHEiQB=)+lG%6%G#_m`v_sV+j3v(zV6<VKZgje
zyZdtga2R0=B|l6^JT6cFdn^ya7H6Q4LE0k@^+HRG_bQor?zuen=9oBG5Tv>2z0e+*
z%ti=KV~^b{>*PBhhKQquI)Qy?kdJGv|7rQDn%M`BaAZiopsG+Uq&@GV3+>+VbACs6
z;is6on|tg&&9F6KU2=<!jwQdN)RtI=(EUf+u>0n41Yb8lLk_6ZclyXVb-y|T_#-(|
zbH7NTE)ElBB^oIXIW~;L!R{Zxzh?TJ*@fSE@I=(uHhH!{^u=GmG};4V=4k8Chjn!q
z%KG%v&N;^(7;Ex^aK9`R>()HEGU%W^C?^~qbMENCrsY5@v;-EyVC}L8=W{VhT{7yb
zD|AD(MY4gkmI|cwonsGKUXOjbwfLq4m>XQgRT^@3AdDYRX}q(hOt**TCsz(185k&(
zVz-kCVrXDcsk!W76`1lqk0})DlDwnrXYd<b4$`e&VTF76a^H2GaOJ&)LOvY&ei>B7
zmWW+AHoV;)5u<4>94!2dojf??cb`0R6gL2097a06Ru@_lu#vSMN@S0UxeP2OHC|Vt
z<mY^x@Zm=tT+gEuNCF==C+2lDj!VEJS2^hcHWr@JZfsZ3-D6@ti;0ftEfmzy-$H<T
zY>fSe8_p{Htlt9~l~E<Bf;}!C>Viq(r^8|_%N;b&anT-M7oc_1!tV5Oqb}j67;LFj
zaV`amoylR^Psnyzd=zWomNxKP&OxO8EVL(Mr0aVn;;>}S{yi~Uq*D?@bKFgEIx1)p
zhAbj0?~`KgBde~OGN({*u!pu;NlUiaKWClvXBr=8RiREW8d>$?YwgL|E}hOUzKlat
zq3(g+-bQZXh`=4Nr^H8x+WHGW^**_Y2JE!E^dA(BI-e65?#G+<Y0T-V4YZWcBH#?$
zCXFD^9xb#dG!*OTVMvU57f0dg*%A`BRwdcwLW{3}GAT*w+HzN&pgk={J9)yKDTUu6
zY||&&)3ZOx>Dr4r*<8>rv;;lI&#ceRs7QYq4(e@#g>oPg>KyruL>N3|(avnfA@Kk&
z52)K{?^{Z37cR-CQFdKPVWJVT`<#_Mk3At=X;f&@dUxO0va_3E)3qaoxFLQQT4HzN
zz@t<Q&sd)FF85V)1Pm4`B+qmFM-mof?q>$}^uUn~77AhyIJH~4fnmVT32KR!*PX~H
zY7~9#OYGbR(WJn1$-7V&V0+cfNwe%(^|--Xw#CIo#3;#bNVQ{VaGQ=TcHZ*js_{7J
z?8AMnP!nC9xDD@{^BaN?(&<_RUdJYOP?Nc;@RYBMlQ<`O+Y{%;MQ;oJ!lpvWN3hY;
zj%!z~vkMv=4tp_KrORmDX8)etgfDDHkJ62=<F#3+3oQxQGOq{k(taDCL&=M7IplLo
z7h1#<;IIw{yudDMzAK9pyz}tdLY=R5Z@XRG?Ct5Z3I#212cgD<TQ9Lo>UXDIHEH@a
zlSw9Bs0!4?D7BHvANBBBMJiufae-Ufi9Ku;Dq_?Vd-}*S*5dllXIK20gpGH8t5DWJ
zF1+9(drpp1m@BP4TK7VoXaH8~4ts8eJ*TjODjILyLW@H?)T&_)dS0D1r=2t_ob{!t
zLpA7G-m#}VzrKsdaNH=gy-=q?T{8YsdqJYbD1jEi5bNm4_}GBlU|e`g8~%z}_QHUD
z$RjT(SsYn)<q)a&Y(h?9tPM&N-B)PQC(pY6Y@683Q@2fN*M`gf_Ckx(Acd?{$x*mf
zgGVOSIb*|wDj=wCF4V;uv^u)_I?#J}+vK|9IQJI{3A@M_qvYIE@;k1ejuV7Qj|ODb
zcI_-Tr|YnDZM{+t-(i<6PX;(8NBJED>n2()wX<9#NwYP4r0}eebz~J+c#_`c@;U`S
z$krhoA1O_p>$%;r*QVCZ%UvZG7D^77@M8;>C(ww$!<%ec-Tb=q&pnW+@i!P2WKwdD
z{7V{?vMU1dCd`~XweUNJ2!{&;e&zC05f}o8+tIxcFApLk3$+q5CiVeho1U|vS1`ph
zXq2oQc`GLGH^#p<qxL*e;%cGj*xuiFqg_?I0NKQj;G$Hhi?y6Yj(6Nk&~{!OuW*OM
zg2L}S)*L_Kk%X&qVVv14vSOa4a@bd>%csQ>j4sRudr>~EBm&xquxX{%X}?Q$fwiz%
zb-xcT=K6TXPeoG42%}cV?$l?q>$bm+oFF|*3gy5$x#}@UB~Y4^b;>>{E>(O>){U&{
zfE9(4yt&nB*W_2mWbvd248et#fDhTKa60Md?76nV3`O`$Ya?K8u8#19b{}%+$lUdb
z87pbgbrolbY#nl4J~io$B!6s0r74WM367^<;BGI@9@!i+Wrc?zn|NQLCGcowtCKqe
zU$XqvLml&3%vOKj0O_Up+dWWdaNwkN+7rCIH0#6)Ai6k%MO_?Va1ev`vdT`8bFh11
zC>INr@kkGe2kiPf06f_T2MR@@?5l=I(rYiT8v=e#DTwyU|A&*$9pi>^{>=7f&$|q$
z$ZNm<$X-#AfoomBg@HnadY|O?QQeTk7hk38%<ljkwNBiLVv?1k?3E4hwmF|DL=IIU
zExutT$;n==+nZgINwBvyU)bg>3pe)`>inI%2INR_>23tp-n1*3S@23@B99C0+K*?v
z{AFiY$MRE;<#*%l`agdZ9uNh8`72&+ojI0BMJ@^FP@ygspES7WjaqfMC>!J!plGAf
zx_y#=VWoH4+~zr=580($@KLDKqRJ%8c9>U}7{}guh0>U#Fml9nWu0q))CV+`x}caY
zM0mIiEwYBlH+GkG*LOJIM6fY&7wW>cF=^T?PceRDo|6}hFZ>j!{95t(wQsj}`$h(#
zkinRzj$zt^@rQs`f^0qcl7?43UYbJ}_|<k1uZ-De#ny<Q=1kdd3jz#NZt9Q_Ny)c$
zxN(F>RFJ~nIx(iCO^hj|P!~+?jJDn(e60KGGw|z@ANHe-q)tG;jPxC7tQJ<FpCE5S
zp;V1ND0#`|@T%YkF6%A)iVZlkk8C{swPV+2Gd-LMYMs=u&Yq1NvVnjm7oxQle*21g
zf=rN|OGIoiK1i+za>RC^{-Po={g~9`evpCB&|l7Fq=unbsk3f*Md4?qE+<#;D1pzm
zsHqc+=x1i3E{Hj}sno{Jtx~o)KbG2<q)75q4cww#J7EfW?ry4`xMnZ>)Wdh4mVo(n
zZSx51w<XQaCyk|}A{tK@!fEjViyK+%=y`|TTv5zpnl?ZVLDJC_ou#&mb|?%}kxfL!
zanx&wTkO@%*OG%_orV6}3oTm9g(TO->dy&sOMXh`Hrcx+T8G1nCk`vL2N@q2BFB>C
z9t@Mxh27c!RUSR=&tGa?)qRLtTACjSi7fQw8l}|G+_YM1dvJn*(N#4~h=?t#kn92X
zdIw5VZd)kauoB}Sq`z}P)hAGwN12BsaUe?VCT~wP2+-F!%9aNjPGy@Gerr*uPP^7t
zG!v7qIZTt(IReVMBi{J1+uoJO!o9F<W%I#i%kpLu*AyQtJf+=oVigOv*Itv0<OG!E
zU5z_PI49HMU(;RY5(P**#CPPJ<Q|;%+U2Jde&AOJX)Vaq#&zNI3iV>$PH@rEi50M}
z{{Qy6>`lpwB%Vu2CKHI9Q)IuP-4T#)yHhN#IQIG+TZ)M6XIa<|4{tBDNEX0}?+y80
z=ze7@+EZ$s=A>K7bno69^SN*a6pwPCP#3!oCSLAfvUtVZ-W2C*U%N~Z6q0?Bi+BVD
zhJNzRql*NM1glVwQs-dP`756;oJ?}PT_tUAX~3#IsM9sBFIJe5&M?}8aQBe19bvyq
zsrTMmJN1Y}5La5dQPotT-G^8`TfhyvR?^M4<q)M%mevqMcA-vO!^o<>1-<t6z&{e}
zU0V1ZG?x2LJQLV$wW7Dx>f~<NLPZj+W6o3Rj&_&b?e^$Yj8R*m5cHH4mRlay5qn4N
ztG$g=w7;`Z5yKdF>4o;r`bH=$F+!{_wN9Qm5@onO$lev$%e5TyQ4vy#@AyUSv7z3b
zuu3R_<+>_rFWZjO5x4l-u2A1hJj2q7J%aB;&G()FNZY`Ifx_?XUkYI7U+ulIN&ANu
z6@GfND2afJlxXkEI!~7jijx_uP#1KV%fI-~-k;AXszItB^j3vB$D1U3Fd?F*N^_?3
z^RRuOZtmd9T<5e7g|Ku^<Q(No*s=;wg<Z3=dr@D12dv8n17Jf$^`m&E6x=(}K^U1J
zzxJWJ3^Tgt7YcPoP-`V=-fkbxx-iNL^LG^L5)TxqCHYt<iCFWG)D~t%#sp?4-5L7l
zLQ9N|{AwKtQ=RtFK#9q?Y=p#^6(sCX4Nt7f#lR+3`0d}U#t5mz#ZMl$J1XBd4+BYO
zq39Hjybi<bZFL3ybO3+=MOm$ppwrew3C&mwl%$mFtqEyax7ynLtavsl0kqTsY+AIe
z_-t%B@xEcMqR1x&BGPYSjPy{pk>iCH@0YvV7_{){kgcy9W*z}_6oQ4Bt?NHF)Y4>}
zpbI6(jIozrZD8eP=i81-Xk&hMJ<6i=evK2n(Bf~Bb6wm&WKP~p)go}n?yPUJO4=lA
zAh5oQg_gjRiEYI9d4T_>x_`NxY=FB%3l)LYmv;|N9cs7DF(G2Pfii^{G$(rX;0ncg
zIE`${PmeC9k{*a_t}Q{|g8vC;!>#!NQX+K~3Jy{_mDN&*h;7SGVe=r6Iv}fko)(>~
z7hiauZLj#=wzyR&h@~D`HTfFbkv+0}V%q3;U!gAXT&x*#na*u2`YUZ`ojvYTeck<q
za!|7w9aq_|eC;lW0GnV(p+zLN-Uf#|XVrAzWzg2y0d3SFN^Ri7OtHFqYTsNTQkudA
z<1VqiF^bkr9k`5RiKiS;VJ(xlvwiW5qD2V}Yn3d^<yT8(`(s~TE+Mj4xkC#zf)<xZ
z+kx7lOv>;F+z(o)i>XfLI;4X+Re1nQk-<)9HFgpKTw7d`0X+SxINCSbq3jgiixbti
zS)tDLGTrlAb~yS(iKBm4js9C`@%i1!k>VLFhBw=hW{qT23i_$=yETOt^*N;67e^Bz
zBk@<r$U2b^vYtDKcbD3(_hlImM};OKx*dy2PrCXl!jdQmK-@%Y$KxAV;`>Ve9$9tf
zQ|)8<z3OAI4<g|@YWn7Pp*>_(iDwJ#<59%Yj)J|XP>hq9Xg|U}25dMd9^w4jHCs!q
z696Tm9fby+sOtsk-lG$3_(-XB;<rNCxx+^42PRLLG2!wlvkMi0aaZbQi<3<^5v_(F
zVX1ZD9_u``{YyS4UYXUGrKHrlKz^q;b}COVL3{_pI_L+wSW{0a%Bm?oi>I`@x@uZC
zbFgj$_K5)X#0j$sKLhV3O`2?<tmB<X_*tP8O&45nv3;tdB;2ZExCsVMvQ}HPx#biq
zLunjapRVWud2}>(dw0`bXm>&&Cqj`M-EJ6HMSdoD(WgAC@Jk@1$6avNpRLCoHkdw`
zxg~{~IKVg`6#jpNt>Stz`&@!i?&#}jQ7{;FN%lwcB?fZ4NN%6cb}9dKBI(R=J-35*
zPodqnisu9Hx^4D_X1Sb$^yQ*Xo209o%Za|&Y^6%Ik1#@n+d_-?MoKe3L<oPeFU5R0
z&KI=}6iR`zbGk|N%JrCEPEsce)F79j7AgY87Q;>2R~m@!mCV70i^L%t3oSl#Ip1q5
za>%hsbxDRhhuBw}QHlu__Dh+uk4;RAj}lDxQ%4q)uT_S6)PA+3P#L>#_S6Z}r`y+K
zWCN|6h>s|gVy?kw%aCE;XnK}sp0wZlO0BCmODLb%H}g3`O&1g)VixK|EpC1FE%vRt
zs7_AdUQ*HPOu_J$GTI&%GC36?toUT$@9f(t_*_@`72Fg#A(!BA$2B!+5zzTmJ<i;D
z4zy6FTJ2cQ1ipNz@N6#U7W(#Z9Q#hQAQH|L;p9O=JWY!~tyoHl=iIz(-%Vs!cKaQL
zQamxd*|kN`Q2So&$%)rYn?AWvI9q!|%&zvFT+n2qaJ`rBB@(;~uG9!>N9_9vDoPC7
z*O$MN1hk{K7usXX&%Sz+{adpc#`?H6E<lBLA77{zV$1)ajv3wo5u1;$E!5?CFS$&^
zk#0%?<PY;{omr@AAi%o{bx9PYI1EIUrfj|asICs_LHki>%`4Oh<L2Q@D8<E(^VLvA
zxF3$sYiFTF^Xl-6;C_-{1@Q$WrFARR<ySM(Pg+{{Q2uGQi0cl`#vOnfuDs9^*uupZ
z3%cR1?PqldAmbt!P;>xk%6|O=zHL8GqOJ8%yxDPeDq#h{=ps=^%)(QmAJZ<IHpzYw
zn10~}=a>9Bn_!z?)_uPafBix!zvNj5K*v4y?{zIFPnlFG<XS0Ts{0_?uWH?}lF&++
z-B_q|a;xZ;{r2mo+d~|D7}ANG$vV=D;q{KuAC<ZNrUAT=FN_fZBU%DRIEfwLe@E;;
zny*H7Nql_@b((H>XRFPAn^>o8PRb3-WddM9WZ*|wEwuYgLa8A_yu*H1`NoBP+)3J5
zC<ob`$rurO?DzQ+o7Vk+wxj%{E-2^pX;)6MKg1Iw<b3QWrvYr+K*>7ygp(hltSkSS
zJtcioz)q=kvhbmf=LFbde~eYTbo{u&PaiQEE#Rd%t>JY4RAm7IEW7=>%0iW+ps+>q
zeO$*(aFqSIe$(?g2y>1X>ip>-A7T}x0TLtlmwftmjF-2u%Wp$_L!ERxPkFBWH9-9$
z(PVySA2L@+?2}y_gOvSO_7P)ooll`IfGW3Lt@_{UAwb|qzw!YqY=O#vss~>Echk>t
z|AL}_s1xNL!d>I<*&GK|7?f)rC9sma;?V;Cqf+=WiZ`V_g7Ff~FSLi$)PtoM?X9iw
z?GDpL;4g(3cTYDKqs{@l+lo92F%d`&s2wk$ig==JD6|V$b2{wqE3)6Y?g3pW2bFN|
z7U5=W>qB;r>=P~>1|;<V(gKSY<7(SNdl1$O&%4;}xgrokZXPn{!%-FLT(MapF1?&@
zA=ja~(Vu{Xd)3dRyRlqlL_-uWv>VvcF6duIFgalOr<lmBp~5f0ysNdy?wxbcj6H@4
zZHXxF!9q*oX{d^i3~#dgG(%P7tzYz1sB>0AL(PMQSt??51KxUv?Y{X1KNX<{Qcn7@
zRBDSfbdqR11<*+pg?7IfYR}xlFVTXLRb01l|K{sbVGF89ORcNf7Kv-L2dqfW1L>-F
zexVq6J?0XyhzNLKog!!>F6`*Yo>J?4-!drd#P*<gj)*7{zL0?&nXXV5y4ki)IRtUH
zUfc=s;EE)3y0t9|W&e62Fd9YC1mX(qAqlt;gkpzFLz?+g3hl9X@C_Je56!W%Q0#4h
zjf#l6KtL#sl!UsJ%)RyOVfEb^eVyHfk}stctLS$b!y0>dfcT}Y!mmKa>(4&h9?^V8
zisP_x)Vg}17+I0+k@=h`uIHsk!8}?g8D5&TjrOP<qO3TQ_{EgOSwKs$pl;x=dCugz
z;4+*&I;L>V!XBk>q2yB#mPkFmA3E@u+$5pKklkSgm?tJT9FzeSOg>Nfn;V%tewuEE
zer)q?_aoG1VX398)VY3VsqG?lo9NAL$*`O5M8?uRuFeW2s=VRF>TABc3oQZ6YY7~F
zd;?5*c1rTtRjBhdOC=;!RxwJ-hWvzV@y(IPrbmhtTAUgL1qgjjWDAZ-%kgcyJ#j^~
zM-Ji55n*wKO3fY<0yk=LU5?@~o>XT)hT>s0i8(;=Xcyv<$4WOf{d1In&tBZ(DWy!^
z!i&M4Tmj+Urjf#fr5xT=nwAk|h08G`F|V<w)CCEhGlY9bp)U5Q`!om{<XB6ng`E}z
zpuKHqLE(4oh&jYn+f#EI@+Clp3sy@)53`RFp3xpCgYSaO0Q|Mn^Bs{p&rZ#fN|l0&
z0>s)_Xm7?5%3+9=ug=p}B<s?-SQ)=yH>nY6>XBFN)AK#i6%XLjVWsy9Ej|h)EeEiI
ztj`GgI|=`6ehOa{XE|tRR$_PdxrN_861QxEJEN|XXVrO;O$eVKF++tqL5BO-$r5E}
z=MyLfIPsmIEYu~L*36yl5aK$^OCBP7M!qA)Pi)El;WdRiofBu>`aFAP9S&fYFN3^P
zrFU3eXz>Z^GY1$bd*9B<Iyslg-yGhp)H+d&2+mt)=T>6LB`@<^orQAtC6^s7(iVGG
zGq8hDnvEimg}MZe7&w>|_PqA$Av-Ty+$!L}k#t86#X^hsEx|Q3Vn^)!s6!jl-Jz5)
z<b5jCX<08*2<N!?je%`f;di2j*dx)I!#d&6-d~VCJNLn;j}LDv)H%2XDV=TdUzK7*
zJgfqfFU;OVZzA)f41h3fg_eYYU$3sQ+xYtJF_(<6W67gI%PdBFR~_D{Ht+xYFS{sq
z`gEwTusw?%1ZXJS_Tn7nMmRD#p2NBn>IAtqoT%)QW)_Sy@TE}a5Db6TF^Uwsv{KCr
z#$Qq>H~=JrLdr4cuC-_9t07DPXt!P;#i555;poes`7(P>rBrOZ!-bM}vjGAFBYScM
z*zw$473n4wZ_Ku0YzhFE+I3B#{USP}Mc#@vW>vpyL|G){7fY@v3VB{M>ub-8`Qzj{
zSop1d%~eGJ<N0;sF4>JyHk8^TIxc%WGYWLI7vwvB0=OkIGshKL0zObQF|3<FkQZj1
z4p=#z;U8A0%L5j7T;S+jJv@Rea-ysgDkn^Mi>I5SLPX>QR#hxM6Kiw3E_Vb2s0h;E
z-V(n*DF+x?eM3d&=We0J2e@wd4wo)b#o6T11#|+5iJ4)N*1Oaej|<DQHrzv|#CRqv
z1i7bBij#H1jOlh+B^+(!#ls#&IU28<@DjT`CmwQQxa?@1#CS<QpgoR}8Et5nr`FNJ
ztLws}J?JQ?bI=J7!iz)fm25wvNn34Nz9auD&0{A5WUUj5Dx?%1)vn0rL;<z`;Y~`d
z10!8ktea}Cj1dj?-9)T)$<oNG*+fU#^a^&^&hEK|VgeWR<<qEnZAOI#>Ae@VdJ5$T
zUm_(5|LIy}?b)j;N^#W5(I7II$Xrc}cUq|$DpI>TpBBA}97Kskg_eX)Bz$#S4rs_O
zfqUO&Gh<7RN7&`3cTDbfjt$S?ZZ8_GLljWmx<Xw@WhYh+?}MMVS;_Lq8%B{J+ye-z
zD}nwPLvrCMNqTt9^WN@Wn;koFWYu$DQ24D?CL4*oN9I%nxoqO>LLp!%VFMWA&1Kib
z<GM<cya1G9iolm0sl4kvw~a6gBn2$k*G6}Ib<bd-5Ohu3iigF+I)iT4H7(*zwed4N
zr_ka<5EYcTru<qj&N{s-4<E7k)TJ+Ic<G|9e@XN0wK!Q;z|HOM@i4`gHk}-Ca|m~=
z6zbAV2K@wEY&7K!#19TglFt@<*@{!AYx@V29;VQcN(AzhEO=IY63@nZ%_3|OcLck>
zf)uTW7%Nn`g}T^h2rw7f%M%vzu#UYSN%(h^+9C#7>=<@^J_BcNuZZ=;Xx3Kv?GTd?
zBV2%Ovm0XlNr<-)vs-_m>?<K^BspWButs`Bd*zDsc2+<q{$`!f<<3H__o-+YUDbPJ
z_?Wd-Mz5d52XnE@7HR@raI%hjTQ^&K#ZTvi#BGRQX%gtO|Hb_Al#fSj=?cb!G6nD+
zV#jyHu0}(Ji)(+Os85Wn8fuqXVo(WmI_p5@qEGCV^Aa_&uP62O&h4Hz6j4>ys<=IK
z(u6|62OFknsQ&Oan_Gu5Mq%RWic(w1`oLMDWsBgV<2En*!K4V;+W-TzrBIi6v>gQX
zOw&48<%{h8!Levv71Yy+ogrI=-1`O!wPFq?atV!f=Xm1OBVf=~H*oHSmOz==U0lw7
zW3vmnc!ckT3Q!CMT=DSULVL^=BB5(@??A4dUpI^v8dVglOqC8Uv{LSA+7smIl1tAk
z{PgvS!%>rnbA6vJsMGhsNw!;8sEg^3QMYK9DyRonbbtW2YrXlV$VJ`l)Wx(2HyX0O
zx_P+5!Nc&Vh|_Bh(%HiL>;PfPg;L_4Yog&|xTG?{5+YzI>R39&3+Gr0?LJQGk`jr3
ziI_=Dm@Fo&fc0n3;@F{0qCkZ@CwrqB);-Kh>gsSQ7|3>EdYWGNX2Vhx+5_KN9%9?s
zd2s9y_zgDL%wsR(j5vU-(>X_^4lbY&x1su+PHOZIx)Ff7+(o)pHskex_?@;WpO)!D
zq?yDGrPd+hz@I4y_n<9q=H<z%LH4_NX%yPE0H;0sw9}qxH`V1qnXSu&3Kc<v7m^XM
zPidHJNi$5PXZL+mYMl;Nu?yV0W;e%kI4yF2RVP|e{MD&TMh1_LSLa$ul3=tcU5(oL
zeXR;lHT;O=1lhEz?<Ce3yQK~pb+ztxE>wsszVL!e?AAI|`EM{R4!Bb5f=>FmYD+8L
zu@>!xl7BJO6cMAKFG>S~&BvByuhPg$>O(v&wN73x0*c~`OC8w++!saZsBnb74axI%
zb~XQR%Y)J(agkS-(|Vv#<DZv4JG$5YkBC+w?}|zr7q`fhscgN}IFbYn69Owk@EyY&
zZDmZ%IgKNQl7ByJE&kVDQ>T6zrlgW3KPUItYx8+mX8NSIms;m~7e~8|_PQ}BrtKn)
zoYbHa6_0UZ((d1tGrfzS0P*!X=E&43W-~N<8w)LAC}}Ned2S4QL(ose)WwD0K@^BC
zl8X|2iM_G1iBj||C=_%3BH16ZA`k9oIMPY)wl`IlacIyJOH&T8uxC9E8?1VBeWs(Y
zf8ju@Q1)lUOUppNn||I>K{u5<YYV0337zDdU@P7l=-!P_VyjT_Vab(cKai2HtL7~^
zX6Nt$ds}=GtT+mG>@1Z1c@bD~U~+(VBM-k}1MKZ}`?z%Dycz4}cNzh8ne&dJZs)1s
zR%W!_h@pIm-BtmQ1y;0iD@!7*-CJl8rRF2thG(}ofL5aDaP6cnr!bm;v4%*)8reIV
zDePqmn+oMvN?pFOxUFZty|cm<4?@uw%E5SXJj+uxNXqCHH!Lv6ZBXrZRe&JkNsqn*
zDw*sVD|evKuBl!JIc-hnxU+ZHk8%9<zG|pY);T<~YVKTnPcuCc+t4*qm&8`5Z5#yg
z>Gx(!*t*<H*bj%}rMAexyB|O*`z(w*oX7iCxL5h}PCNJ9XN~=@(BN?9qZp#!p96`T
z-{pX_won&jax#bF2l4@V2r61>E$k(ANtPi1%nm{OgB2UE=xfIlv%Zw>8?v;0C?_%Y
z?9@56xzLiEDp-CzXY9i{N(C)!FO)PQZWbHpvX2CXz4Vd`O8&fL{Q36L+6e-SZ16-m
z6zUwJWAGNZiZZRr#dP>OVO1Wr$W1_~ZJ}L2>S>XOt<&zPD|v0JmvkeAvRW~Xh+AA;
zC+3?o)+m(Ow8V3ZaX_&(`JCt^SLHysBR|r*GPWhtZ6s}5J31@*R^nWBv{07>W;DTz
z8_$^=sA=ososm@p3L$<fpGl$4-w|ODBP?C*m;=VvXFI;cBg)F9Jbn^Mlb#mZV;^Ty
z4%<*w-b7);sUruHd5TK^io8QTH>$qnx>Ct;hiqfcMuLbg3)+iNUhA~&<<=;*-klYq
z5%VKPB1r%(G6_Cwscot_I2!gwm9V6nIPR8h&OSn#5~YSs6s;<>=!Cvz_Eg)_fF<N<
z?$r}ANcn}yYDJ-4J@%1@q|3HukH7`!Ddh`sbP_2FEkfYQibm>g_YC#6t!|<T?fr#9
zZ2ha2*!Bd1bhvvNk}i#D>YUN$pn}8NQGvjzFZb<mP=NlI+M<jDqFP7T`fy-q*>+Zh
zn=}=dGD_~k`(E1LJG{wuH6W3sSWVUZF0_Q*mMf8T_wnv#jH^YSppIKgtrJZY{gV-X
zPk{2`@#hzQMi(A?vB~zvN?kFb@XNP%NVM+vWc#wN?iO8XL7jk&WPisZ+h6euVFC){
z0WpQj_{Kn7N_rqp-bupqR%f9Y+l^@Nd3LaV0tzDi3Eupr)+uwI8-Iq^*`a)nt>H2v
z!h5N8AziqI(+<ZPUUJ?g=M{eEX}i^x6ga=yk*Je@ANHPJ+`yL<>S9DhJM`Moifs@8
z`JXFWPHYvRd2)HF?c(3uL7VomNA$6Pc0X6I@H6%^c<p{Gh}&T-b1i`#ZzjLjQ3OGu
zP#0Q6s2F7HV-=WFh6qY-70TKc*IhZ)K3<32w@B`q&9|Csn)6$(R<q$s8_)-~_2aKd
zg+qs|Q*bMu*3eF5C&3~Fs~C}zy4q}4p*;x9z>u>3blOOEBA#BrIEWZXU4)Me5DP!(
zMuP!68S66f>Z_+0e#aEa-m-u@kz-2p`g#gKeN3SV3I~l-^=Wc*&+9J~{SggrI~#5_
zYK~9jmJw{p^N7}JJ!6q^lNMp?v;i{P+9zvIGuwuG`U*uC1C=;ef$gAH&>l8xv-I|<
z`qFfQVR{S2Jght*1okPle)j2@#>DRS!p|I|Uo`_)w#h!z5Q-2q<yIGzZnQ=DOyh_b
zLhZB9R_={?DVip@5Yb4T5Q-BeB}(C`IOnJJ>MR++vGH?B5z##Llq`&_8h6PB_WAm7
z+X9%@E>6ZmRZ0e<ER{2R_J!z%WLMn3MnTkZHolhj#mckICZ>Z-rmaFvV1|tSu;lio
zM5SPcrS8?eP&%tzATP9Q)6Ks6npu;k*q7@HI-e?mOFnkEkd^=0qLVMqpRdG#xcBs;
z!f*A)X|)TX>Z=uv97=A2FQaw3FhgV5WnatZ!XcY<HDC;N2@Mwvltf9}n-lx%`E;Bt
zl3k7#>V#ipDaEY#jb_=UpLDm{LS3=}?k|8g6ODljZLS%yZ)Up&DK(OONrc83Tvuq1
z#nsKd?m}VT$`M9#lB%YbXB90$$Yyj*>alMJ>5yA~Zwr}`vRWa#Bzv-d%@2oLiPIw>
zu9PL(>e~wKv5TAZ2fXy1N{#TgC9!3p?6X_%!o+%Qf1S**4enoU-wlZ1{npP-jS7D0
zEYvwSFJ!0}|H%-maCad!Xs+2yjcVT;T?`~^^+}8fY|EiSOH!{G^0DEy_Wi6A&wPL$
z#70mTBgC&pES2uL3(BpthU>FO5s$vbK8FosOKDp0(h*RUxz+o$ALMtq)s8;(cVnSW
zh%nL9qy4b1kYi7^(<##wEk5j+3Ijn`{-e5ekPu17GFYhahY~Gew)WVMvrd{+Jcp$H
zE!63oBdboo{ucX5MHIlS6ywNERH+|W%&u0)e3UbUTKzQp(XzW=FNQGc5=Ksd(1S>0
z*Caf!pVfDqA=N4!F4X0Q+c&(yeqNt&?Y3Kj3UzvdNx^s8FX}f=u5g(7&Eya{Y!}M7
zUTBXqf#?I$y8W_FZN`*2r76tno@f7_pIYx7ZD7F;7V3P@hUr7dq@^Y+pZzM44Jzcv
zw0!nL>!AC@xy*~dipoQISa>!B(HX?|w&eK#>zKkEuG0dXSsGc@KExI2zljk_$4vIg
z*cYrGaURg$|H#>n#c?sW<Y+=ed7#klD~vwUUHQeC{5Ic#G)dnmKU>Fj?Fg{8)b=_R
z-N6`k+VBqkUEQ0P!zV0iD@}QB8*z98_WL-W#uLS#KQB4|68l31^whRRr72v0%k}o3
z6)mT9^x&A^DpbUv-76HW&G13{W3zo^N>~q*o56sEmYgCh1cZgee+qb_?dxrI6iWVt
zZ$xKRSKFVnu4<o-5t2_`qNgZwnA>Ed(MBeC(f-o3`(4M*oR3jBl-gba8>vS}!)pGT
zr>q>D1ky-5q?^**(JBcEPsITjYVv;juV#XqIg*@DFVyLb1P^-cZ`oe`-?4p_rVc%(
z-oeA({<{JNE719clB05RW$X~03b+8--<$pv7$|vSb*XjIQ%aP>erErOb-sGeoWf7<
zOx6IoRyJuZ%~mF^lcfUL8L~`Gu9wu+Y;rz%x0UsM?VI4Z;`FA)*F-{xU`7%b`ghKg
z2-f9x_v}sYNK1laN2zs=IsWu9yGQ*3oD3q&Nx6%K8$2x`jJ<f8-LrY>h<kuW{vMHc
zT7m>ln{k=lYo!Ywooz!na$)VF9F2nm3+<n>4=3R|(V3BBS)s*e9=;Q9@w0npop?#f
zk8*mib&*0ZM7jGU93&=2l0yjtJH_WipVk!Gy>AI8y0%~q7nt66r4z<$u9`My%H%>R
z&RP5<6_hG3NxNTNp2<)bf>~OH8VOb1y&XLslXU+&w}FLSg@S)2cr8JDjXj{kZ}Me@
z-vNjV#+`2uYzj~U3i)OjU==j+pb9CDnwlae%yqQ|t%&QxVkhaQvIo~rr(mC_5IoCN
zf{r@c#2P&$;iO|!3c1y2sn8Bf;qI0y+I<|t7g^YLK`kD-(vJiz9|R<J6pDW0pg!Tm
zc~~@Zwzc)boKp1ng^v*$vELqE=Y|-m6!~&;qAs_W<`xz+PeE&sXmCs%pXgN>Om-An
zToXaQ{zt}F`~6~|zCuw$8d)_17ZH0@_8`Viw8LE^xmRFcp(UOh>LttKqw_hje2Oa3
zu@J(0(}Mm^_T<B`LtrHiV~>fs;9*NBhg{&<vDAqob&^x62i-@%JvJvSi&9mTQy0QG
zkCE%b9#=6-X^k<Y4isvFygE%1UD5WBZH<?=J$_|~NhyBtl(@6-3za5DOLnC2lu)d{
z8&j-3p+Y&i`N*0e;h7daHgWpZldiNU*0C)b>Ou4_l!L?)pw1nb1NNkb$|$Q7^u^69
zTV1)1-L=pz8ga=u(E6X7^^*}SQjowz4(}<n_yl9;hg(&+>T-slTzzr{8(e@A6miNF
z>Yhl?Bdh)wm-yLJ>hL@~Oni}!CxsS(KEvw^?cps2b=DE+cg}Ih1gZSr@xG_!u!8J5
zvJ~oKBhQ{N+n(Aypnw!nv~^{)j{8qJkCiy&^jttO=2b-%uox5FR(LA6*RGMZ_OvLm
zxyadBDEaOZCo9Joo%~Oap$slj9t8^ifNv#+vI%bdjAnScbKEmAm0G9x({%W?Gh-3Q
z@lWA*GIH{7tZ|u0PIgw)B3+JPBKs-=YoR3u=kd7M@EF|L6&W0l#b97?#6Nj$p~atu
zDRD`Q%vbh|tgDs<a<7(JCuze4ZN~7-TyK&74T1Np_hx4x3CS$9i%`sH-DKy~LAkD9
zaNb?0b3BxThM%S<<rE#xb+Ev8Za{Q4In7#yf<r<I4$|<W4tFyu8v18d>N*!jzZ=yL
zi7*3%CstyVI<KP0<$e8yg3ng>A#Zg99~m&S=RkoVxNZe5v;>)2*w)`;<MQ2_r;Y4O
zrPi&3Jcc*ZLAxNQrQ=^j>kykXe#fCfd+h3gB}(|vWfwNEIfMdDC*$ryorA7>(8kwM
za0$X-p(Kn)wFp~%Q4U?3K}jj(bra$PEmc}o+6Lqw*!PR`aUF}Rfgvdmpj;9dXwjI&
zDTgZclIGC}qk(>Cfzb!hqLEFTKE*E0hr+^6Kj5%KvC!h1=S~PW&X|?1O(w+z(gEdM
z__(p$&9pGEnZ&`KlkH+^31#vKkTEi9q1|yohmyn>Gz*S>TkN?NRvi;&%q)}yD;xz-
z(L67qEJ05m*PPgeI)8xd1^xE??6uMtMc>41Xpt;NG?>!s*$ZMi((W#jutPbHcS36M
zeUzK~3oEV=BPe!hKdvgZPRpd2PmVd8P{(`zrRQHzD8<Pm-wGXc7&XcE4}J8VXcOb<
z1(9`-QsiUoUcmL9-8YBN-lPia8Ex~pm90>Aj92bVnSPH#4JOy`NYbQk8}J(|AkZGH
z=HewbC0A0q$2&`wboC)D<Yn>P`IlaJe&Kf<Dr1=>*D)aBVHdbOL5<2|uu$*?<h*gI
z<)HHi@-dwnkBzLF*wIrc#Qqab+O%fX-6sRNF@3cPEj~vW1nITKG}{$99$6dYW9R4h
z=viH8mq}*PT)VPzPS<(vY>IKDjJxD~n_l<oTtq&F22zpa@{q_7KsjJDn%RuH&xlt!
zTxij@>cwqzK)kG7RcEvq`#_<TieVWBI!OIO>e;I+>R?`$Up>cXp)QmGP&XnhX66S)
zETtv#xyg&L)D{s0*;vH7IS$*4DxX7+NW62&8$?a+B60?TXInWS{RFXD6`==*q$!iE
zPgSCkxCwVzwAoQdW|$r>>~Qoe)M=z`;L}b>_?(K3Yz@>Hg>tOu3=%QgH5JbmEhzl<
z(P(>09ofXe*|quUXkk~N*sFC?%f{+%vFqw^WP54dG^h~8rq1)CW6vb4?Zx?}&??4d
z1#xoH5@7D=>%{%_C3W$uOFI=JeRE(fUs@mS;tG}}g<^td3wu@p`Md08F}riFKj-?w
zZ=aoVWJqJThO^emsa;<o)CG<Ybri}$Ou8|?cRSgBc^$`swoav`FI0GMabm^#%XGZK
zUXd+hv{j5o*0<UcVP)cK$teexY_l73r|68Tkx+7$@T-y__+#6Z_4bhSxN+qTOqRVe
z=MIMlYQS{yHR`m^WU>c&tS#Vn$;B5Iema>DOSF&uvl>oKTxWz(9<lbEK!jD&Jv@Ok
z+J%-pG<3(N%i}(^)w~ZwWgP*fxnyf;_20Chr%;hwRU(u_$vP{1e6CV<Z2=0HjSDUQ
zwA39$4C`i6Yvt4WfV`-5-JJWi6<Q=k;IFan#BCTR=Ei)l;C6S!PfCf)?mp{NmF{x8
z&8x&^^3_)r3c(0;uzv)nt}!E)_YKY~g*xHf#TQ>{-3bBN(>0JiVxnPWBr_@yg!Z6Z
z6K7Ah8?!HILX<gR2fo3zt_tV6X<>dom+_DYD7#Co(~RXg%Pg#??p&N<dkQ5t6^+qA
z2c6+d8=NMgV6g>te^0))@Y`YHJ59LnYa(B}-fRgKmvnmyu97l?7M(-0ySOa9ufi18
zF}<bH9j+W+TWU+3+ltXpY6@ILZDI3Gxh{nHp8dyJOiS!n)ENRI-Wv9*TvG)--;T&b
zIjF&0c44RW*H<{R2xZdWLR}u%a&Qt@g^jWrIT&mp-$5UQ5G00sW1%hq;a5%jk;D-R
zGccG>W7_t!Wdk}Q>Ou#L`W2@L#xd0FAO$wM7y&^MQyywj+7tQY92B@BS!kavs!NEe
zcd$?rwrIC3vBed2F1T=fp%8AXkg*!Nr5@z+MA`-Z9vt3iH#KwPvTXsEZBJe9ZkcU_
zm6H2)0k~YDHrSGS;1T)HG1R9pl0s$rJE7~*aO~!~F>p*E%nJ|5N)+QRcJZ>Wj(sr!
zt1>^e8E1vEx!-QdQ3R5OrOjQjfYxewYudH%2D`8fkWRYOZf#y5GaL)3L<%+LNVK!I
zi$apnB{&aGo_e+}tutuv?prvZO#IYnNPYeDNx(>Ehh-J?GbYTuvSbOM>;g;67=*yT
z!Baa*u-sl~(cUKol`U_ejo8799A1SwUug|XVGar=(BRkEiaMZy!J+o{zW&}qjkeNU
z6dbm)4rX>+e_NrHgip8jI9;~a#M8a*-B_PS$J1l4jZ!$@7E1oEEC`3$MhEb;d|d)l
z%$O{N3etApaiFhI+1frP|DV$`VH!aj_WFE3{F=o0N{^_cMGAwBh4xCQz5ktv)^Diu
zC7z|%y+)~V*u)XJVMJHO--+AST6<#yZuudFi0h|Gp(Tidrw$+=3YPRY<!E?<!_kR{
z&qL`fh{P?m9bA)Q8c{?pDy$DPN55=uuJ5x(q5yH7@Ez<6x$3MewClJRE+R_!De#uy
zOwV{;;g{C`LbN$=&9TLIw95NRtwWq=Cc=Va_O_f|KwKJr($46ND73^{Wmq!-`S$j#
zLkM&N$=^a<U==oJPdA}?vwIKNZ4Kn&I*(w~<cL#SJjDl3yUx*%joWK)j84H)yP$(X
z+-cDkA%!=(LUd&Bh%)SH_khgnl!KPIkquTwoPZO5rz`f(N+guGP3k`?@&yu!LZlRX
zfdv-gqde<t>|NQeje?pHpm_cr_#8=)YYOdx_C=STXYalXv}Ha$Sg4ERU|yei*Z1Tj
zEHgs8dfQ)WowPY}K1E4x@6G4p?qXefy1!5t6c5)E{Q0p(y{{QtBf|5(Hl%p8V}<s3
zgA4`1Q@y`Ve?o7kGHVwq8oDp0W<5NWC;ze!)S-%@ppHPZLY+Sy7VfYBqE|(!^TB*t
zgecgiErmK42z8b6hnnw*xrq}&yJKUaC7Ao`2r9D==R2$B_ZNN(_eL<hDt~HQJ8bwz
z@;kb-O*vloATLoT7!$1mm(jA1#;N4KdWAydPk_Bw;ux1V2-I3tL4#eMoC;jJL`{t2
z#M{B~J4OK|32U%W;dnB}kr1*ee%Eghx78H@Y;d{3a2I@`P5?lmw8_>~L}2as8pWhX
zk&s3XW^3c|TW@{F^@ZQj?g14NhIiV!m8YIccmE4rYHcb!8d&Y9FD1HMpCd$G9s%oP
zH2rX8${s<{za|=_JiAa_+pzLfUTJ=elRD@s)IxzBZtd0T$-`VMz1vFs6ggaY&v8WJ
zjBFHJeZweddSkYPrLu;->SCAL5*K`hzxH~1-eq^@$n=E>j$Bu$3oK+Dk}CarC(%gT
z)O<U<ITd;sbb}Tx1rEaJ+UD$2swZVbf-fnxF0u;CZTFaMiLHIw^PgV$nahEbz3<Tt
zwzaY)pcY)#AlJ<e7AljRDD*d=BS@u!-?n9YoG>n4Fv3uQDLWhOKDS^?N4=Q#*!G}w
z3S+^?eXum;n?nrPQ8Ai}un_lQC#cC)6-sitfQCD(`Ug?Ei-awa5S?1#ey3kWX}g;D
z<jCLwv!&KSo<jD*I=Q<-n^R1u9O<$W+Oz~O2<?zPa%E%e@bnctOGJU~iLXwKD?uoU
zvwYe0XLGB|-bw~K+6%vZ8UT|muOsS!?ThyF&cpb|?*y6nFqu6?q0vs*{%nz0==n2*
zye=cq;(a1L;H7~%4Y}<=E{&*iB&!q46wxf((00-u?|^nnVk~cigAE9Fpg9R$=6Y8W
z)|O!G#W3g=KuOz&V(e`T+Byoqy-%0g94wpdaOJ8m=YC=eKDm1Bmh9=A<AC5uwnz+-
zVo*MNp&_TmJJs2a9|3~uIy+i<;hAC=h*)bYO;wn}$UbGqVrU35+e`i)S@nvu?ReDR
zdh08=vOIg|qyRg_sqM9oHCW%+G8v_ZilK=*Txd_w>H1dLZx7kWqYmnh<E8RafU(qt
zEV^K5o(<<qu@z(--(9FnD$Jq6)DA(e!Xo5%*+XjMLmHgg6;DT<gAk;H*{Ul_RyZ43
zc`5=Or}Zw9pp@Dus&%BF`HZvdWcDYS7WILo%u?&fHHS9D^-K0I6##y&qswrn+Kus^
zXo>a)cObA}!(ltsyff;L9s24>&Q#ljB!yOc%-yE;i696a(C{`nmQYSqK?(*c$0g_^
zLf&DYtfWD9aY=548XcLaE<J4dQw`W6&4AKaQ#X2XPahgRw;g;spWMJ1;ik9YHHEsm
zD9*=f-KBw)Jgcy=&jdWV6szzv$LO{sSZkq>*u>9P^0rW+R-L6OWBZ<uboaS9kURPo
zF5yzIuE9cuXco7?wBs1{`MSt#OScN4ba8@FXz|hcSQwEqoN098wb3To7jj0@CwW{1
zIbCWU+ZM%}9I+1D7xOvp8IipO>|S1Kog^zs^17JkOF5Fjf8UXuo=RLniz7ieE<z?o
zj@y^B4n>N~zGPV_)P;tplU$q9@O&kv3hh!3a!PT;z#8asAu-(IMdi6d1dn~S8K|@3
zunwftY~I=K4zsd5D;!(4isNv(U#r7F1ESCYC{>{@_6?qSa_LnVgR6|}>&??r#K{P}
zr%<Os;|n|3hkU%yzOgd$L;?qw5JgDgV<40D?Bo9{mEfG%R(M1E?&AN8vG)Me^D6GX
zC%qRE(koz^X|iO?76cMmTH6AvAjvj_y5*6yc(uFivuk?|B!fF7F(#FCSzU5bx2$F*
z>s3haof3L)={@;=|Cu@Wu6e)jy*~N6*7uonKlf>K=1f1!udu(V7Lw-6Bj<=WSVRHR
z>QN$R{<d*dx*%=PDW)293%Bi|6d)1mdm3nx?5)aAfCp!NylmL=RI>NB!G<EQ0wkqh
zacW}Kl3&4tb#AY{FBb$CP#o2ql6VWkbw=&|^&Kgk<#mvkGs);jjlL63oVx0j7!v!t
z*12o&JRoNhcRc|T30!}(uKD;twlgx=$3NI9X@1@?(YFtFNS10NX&6K&cRDpGa71D-
z`%u?=`4%N%0alw(QgV{ly{Ha$+lL#P#EA_Ou)M-k8KXGez11fZEW$)q-Y^J8y?vy9
zpQb`2AP_W7uL^!U;MAu;vToBRqDMgOk2XlJPtF%04n+H!b3qBzKh~|)d5-Y#T617}
z16EGl+{e2&6e61UYXFNa!l@`kXEchbS-4;06K$MGx?GPGqm`H1{mh1BWa^XlZN}^O
z#y*)D)z?q9ec8K3-q|A#K*H$!j;$o#B<Sf=T~~r}(vERpMShfzNLUL&sp&;yWS`FU
zn&A^;_zqGOE^Y52qkN`5r$9AH>dL^_V0?nJT>-Oa?6V8A4x2A+LM9#xRs8%}r(Spi
z7TM<-ng^Mbc3HFohc>51lNt_JXYKRdNTi0#8=rv*Gp|NjGHqX|HXBnGUf!YY6Ta+p
z>g&MU$)T$_%t!6-+jQacEE$Nhn%tmEAgApgaw4WH?G?rNKwvdUhPSf)W9xlrdqr;R
zVrZwPds_hk<bJS!>ZWBO&N=C1*oK|@_>EFPOQ=y){Xe%BL|s7oPW|H4#M@*eMHxY+
z6*#$nY1<c#BNi4mumeh#VyZW~9S_&c*JuBlOR#zU2Io;LJilYrsQp`O4uyksL#O$4
zuoX^Caza;*j@iF=ldT*mIU_;qSvxo#Z}mD@^qEM@$sB<+*!Paxe=KrcQG#ZrnT{Mf
zBFLerLHq5zRrEd`c;fy`#<|=6vuZe`AmW-sqHe`aIW<~B%>P3~QvFvBZRwKBmN?Hz
zWH^(MhZ9|675}@NcO?pr+4lw$@2wkeL2vt?w%x9p>LIC&JAfo@Ls*EoMZCTP1I)&k
zTyWZfC4sqVbn65;_WNdc#Jb_~H9i5}!Dfzl!Ptkl$>9?9(mDH*I~q|G_XJA~Y1Ae=
z>eOh#bbW$-X+BR(PY*iJx%)Q`4vRc|S$#q5IVWZ8@$m3+Y6?3wr)lH-@;e$`DB2Cf
zk_VDV&@QJwM>53rkav)MMSW!-oSozkZuD?q`4-lErHmZ2ue_r%59kKBGgtt%5H0rx
zx-<p+*;n21?u!^03o4hBPQvH8wn^|xpxYc?o890to-_Bm7`LykZl!rBOo$GwP8brY
zg;Y}dJNud%_5wShNKYJVKt(o|Qy)`x0R0Yw6x!I=wuxR%78vLT<wVq&Vws>r%L~eU
zUDaqrXqI7S6n3oyH`D|fW;Af8wdHA15ybFj_Vr!85=HN?6X6lq_~6W(`dsXO-n0D;
zcf|f2M58qB6pAAg^C@p~sjvx%QbTm>8}Eo+g8#!Hcm7gJx&v=6_Za0b_@;uT*Gl}>
zf#op4?C{m1<H0)FH`m)b*dlfgCq7`>9Gk))pl45Nh|xqF2O^3xsJn#iTN;+oeHe|t
z1JD_9&A<lx)^0Q;MPqu-a4dFl%$yX|=Qdn7F@$da$<<9!udpPr8B@rnCg*Ik=4{Ho
zts9UmJgNH#pJ!jFQLtvPdqL*T+PBwG;YIk3*7bk`%jpi~R$K%hZ!4gK&1>J0o0R(}
z#+~Oneiw+|V*S#0ws~6Do5tFYvjLFgzQvztUw>D1;Q)w&8H-gr>%cUbGeZ;8GxptW
zc~=qm=#*52s0F?Fo~)u{C+2oKrC3oNnu^F&5Y}!Xq<!xl#kfwANQ-`q^c_&16RT(K
z`>G3aFvHl+u6V$K#mLe}Qv^#oCS9c?UD<y7{yWM|L<zk{dc?%w0{~FCH@^eVit&+*
zW%+^nxY#+}Hi?&k153Js>(wMY(6s$vw|G+E(rcMPP`%yn)ChHY+h^=4t)Id9F-&$|
zgN1R5pXlzVr*=(}{(y;4+z2ooCG$IO>T%>ft*{kFe#9x%PLS~+!wG~w2koMKZROY)
z`6?(RVvbHtjL+KBTM^qD7pio{u^F7h?HR3hU2hVX4~j{5!hH&n6n*70yIv~gq=tOh
zfyJx`VXtjb`b-J>_N?~VEeIq%r<jWr;r7x3iQx`~Djuv4+q1i7k3;T4MD+&KIRXjf
z%-mxaCo(+qHP3#L^PFp_dtkVQN*+SbsqchH%ZY}Z=D=#!P{G7@leU7W7<+E6*zoL_
z^BCinf<ty;x!*jmVEdQ_9ar*60w!#{nl$D1{Dzs&dZF`Nzt2by>9Jpsi-I@Srj<@1
zdLb6DrG8>6=MLJEu12xpi!!e}H5b#EcnsKcd0I%zzL2?Q5$T{7Y1J-iW5uM1s*`OX
z%1T~Or>1pKo^0avY-z!yoQ7DjT@E?0BoxTLx&vOEUxu^_%a7QywnH~U^`@pt(BpuF
z_)C^wZkM)JXj$ze_;?44O)DMzLA$Kz!Cqz9bV?!>eYvXI@`fs*4bJHhp^%;e%yR0J
znyy7#rqq~Sha<!+wBU>M8s6&VC!4B))KG}>3cIgp?IGpLUt1ST;F(7Elc}UW=^$n{
zZcy*piu!?+9a6ieL!@qtk+o^RQ?Kb*arshvVf*b$7~pZX7l2p=d6bCe$zl9OZO+)q
zN~=W@8wVEHl71zq3_?Hj04w}M?QJ+-2^1VQz5r7mb-boAM7_Aapxq<$gIp}~IHD%W
z!^*LtneFzH`k>^~OsIa)8%(KQOI2(?TpyI0aW7pz)0|tJn$$LMjp^9*HZEhZA87->
zaTKSk*}ek^w9@_(gQqi6OhK^mtisB!7s`hVDx-J-mNOqW9JAC{c1xY2ct*cwt{6p4
zwt8_Ma<@?Sgs=d{-}`M<L!GU-v=Xa8xd`R>(8Mjay7~|qHRv54Al_hNPn2*09?%B5
zzG{0#6pjNcB7;Sp<|I1k?;*RYP<<D~=%`cb<{`6XBt=z}3f6DCy6vT<&STQd*n^^I
zx=!X4>lo*aUDJBMl6xS=onoFFHw=#SY)zXx_#RRdj5vVkVEW33b+y=pmsUGT%5$q#
zP^03Do%;CYVyP;v4=?Km6Dfsa&x@Ae!*>CB%Bjy^1_jqKD3zp?_Se=F?uItWB=Zog
zHL#k($T=SB6uyJrc5S_l>VkFEOdNJ#F+1w7SZ8I2&4q&!&h4^w-69>=E~TAa4lG)z
z-+<f^o{YS^zO{{%hF*f5-3(^b#E!)z2(n6R{+MlOGmnGK#H3S9CIMbz>!_@=FHeGo
zkOfOgp~j^^vqe5%wz0mQ{`C5L)Pd=MA$o6xZL0dzcJD_#*yq3sE4oa$%;|y);O4G1
zY3j=hI;3N2ayxAz#Ah7O`Rlq@b%iTS0b?B+E2kzm#=tl%gk9ebvSpWC;S^#*_HlfR
z36ZY%hHf!rdJ=8OTT123DFQ`({02L9u=*f^y&^Gv)vA@w<Nt@5Y%vS`%0zHd?319<
zt2BZhpDRXwv^tP07g!T(UYRYOdUYTIp31yougdQYPAlnvQz~2~8?PpVQ##q`S6Aop
z7lx9GQQzkcMtg8F3hO&%ujwX1hgI86!tdDHPJK??1!sCf#<kbBB>?Z_5G`>nsWde@
z<|c-RXC~!9_qv>}f&Q@b7#~uqU1%xfewV$zV1%kem(xoDSZ$HWL7uD^d?%tAobyo|
zyd#{rFM6@_6klZ6pr}+Tr){YFwZ!;z_>cp9WF7%gsjo##TURi`b33I+%#h=GZfdjP
z`bzqqi1!t1t?(0(lZU;lk8>f>H5J<-<Jd?|P#P$OP0a&_n!?dH;;U@E972Y)Nx#i$
z3oKumCcn%9+tP|R_eg%=6muHmFbegpZEe#Sv@G>~!18z_Gcvd)o?}~$Dza|n^0RnK
z+_-Ok7oLsO2QsDi9<vHELH$w2jn-+PbMTy33G?%|k-oj5^^brVq-lyl36?Cm!ftG%
zTSEjDVH0%1foVUZ78#$oS(lHFwa%y14l{DX8;pI8gh5X(+-}<&iMjMr=UJ>5ZY2zB
zyoU7zni?O&;)gUueGF?c3y^zcqJD}0XdLMZNsx0(BLLK^8I*k+H`rwLQwl5TC6L@4
zm|oa~5He+Ntgof|LgxnR^YHqPoUTrNuIpy3m3^DaiQw9&;Tx}2;#L4jkV#xY&KPcP
z>$iOt`rgbu8k6DFBvo9sZ4~Kz+@`ysJQ#;f=T%V~llt&EV~P3-VqxB9>L;Ol6G`Oo
z=d^%-mQ&Dir#{=j*9b40zTK3gCA|adOd;?vY)UjD(c$JC;lw22t{^hSe1*@+Q{(vD
zPP?TW(gLApNXTZ?#J-OMgEc2XesU=~mf4fDPxxgfP!2tmcQqlmlDraK!vSfB?AF3t
z#nK*gO3}h%YkXTQi8OC&{RPYAw}&IN1B>5anc#2*g(RSG&}wHW7vY<8$g8;0b34W<
zilTpy-wQRq7h&uzO^oME6}B>UV-B#6Rw<ms^*3AVk7a-7caWd4=alNbxb&u2c5Xnn
zAJ4rozvJ4goadNjsa<6jWG9;!Y%hun@oRXqII#K}DY8T@IXyWRh^u{TS0hzbsy;<I
z_j#+)c`3<5Zf_DD4%%(mgPW8)hvz~C2^COCyi0EHdXTOYXYeCvut*BScOIHQZg;dT
zHaT^xQ^*fN&}HV`jUwdEg2`kpZ5ijB1B;==E*7lh@`voMsu4;_6{ReRAw_tr(PT)o
zr)1p{6dty_8>L)D9%3aU?mM8|s@LG>(YJdV2ad0=l(Lj_r*4`V8R?CqJAYekje)Dz
ztn(`1>oe1{x3|7Z?L%MXfD?u-rzUx(Wy|a*>a7kP<NQUO_g2%CLwSEGfo%4Z^&7Yi
z7<w7VWT9dSof^q<Tc9#O)i8=+vrC*(ceW}RRLr^bA9dxF-COl-y+g&%R8|Twr#^8A
zxdKxK`%>4o;=5w+==O<lI-M)xtYpYgZ>PR+MB2lQHoq_XHXT+vr&RcfUYf5S4KCpb
z3XgY|raKi1se4YZQofBMOTl%@xS`lz-i;9^L!TM-7I_`pPq(Eb8PB%SAr!hmQUcjL
zuxYdXO!YIJ1B3E-gVhPotR1wUt@rkWkP>YewF9dWL>lJIngqHXx1XzCNvvxLj+!~<
z5Q|KGS}bu!1EKnHRNBwi_wp1J8tVcg-6Ts*_8GJ5cKd}kx3nl{a#(T)CJcsyxF3Ck
zy({r^vr;rUg>JuUE1aBr(0(zwDmEccQ5)Naxka;YztoLZ*%mbqN4&u#-YD3FzxK;*
z-{8^970DEncrJ1zOY$N|>OJ-=T}LpEvRk=0*9%5~8FPWNrH;y9ZToN4V6RV%Jmp+u
z_#Ih5BM=s`wXC7`YxP}R^P+3#A>_bP0W*SkJZ#6Ww;@H2pM9`+dmLExH`9L{2^agk
zu}#eXj;h}s)Ft#F(QEC#6WDVW>YHmK7Bcp|G*=*q-CxXS(yiQNna^Irhmu*Mk8$3L
z1z`J#(O{FEvz^s)sPouTQb8zjgB<9U1eAKwl5H3SY*)RT>Wb_u9%^woD8hz%p$7aX
zX7DN7-LRI)l&(O2-uwvBmYV2@2(0$kAyk=rTIUeXB%msDF0h(5IWO;Hz_MU32zPrM
z9-|7BDpRR_Qe2=WF%f%_@Jcrq?yIk*o<FqV!mdy}5zp_fUJ7sN#5fP>hXi0TGDmHH
zlJlw1nIkbmuK}r<`5pBicUjwkh7S;ZWdfueo3|->*k&3Y64JrO2aWXxx6a__20&b&
zF3srM1GyE&GtKRCO1XBB1!c;DeYFSM5%<Q8o1KDo?zIE!uHJl&J=D(eu_5PK?3P$9
zRQ|{Y4`=1*Bz`44=Z1$k%-Er7MTAcxiIQpot3}8gYcpkA2IENBJ>13z?#eF2O*SkA
zx;dxjA{y3I23DO%4ook~u&N6g5U~SdN4f>u9dE%3dxObfCnkXenRc{#vll*G`aUgE
zKI_mq<J2c_c{Q;(LfT^uumul-30&I&#Cp)WkVY&_A;|G=7qPb(Jzq?8qMgIigNOli
z-YVoxUClWdofF+R6$YZ6CzBU}guYNC8*e7O$!P6Ix_J%FC9Evg_smcybCpxCm7(nu
zurnv?Z8?sm7b-hg&i65bV@`cC7TadFkB-=*U5BI&QpcORlbRIY;5Ud8MlQ5d-Kuj<
z884F|&Vh*&LBiWM9+z9&>4w++Nla*cryTR3G#E8ccBWg!NSY!3!GVcB4%|YTefU&6
zn>}5<e%0nzZWwTiwLRo#lM?kl)MdNuTrwMLHV!zCi*XPJx#@CZSJ`7#zbMX%RanSp
zM|}mME&Os3qo1p6=X1taj@>+nik4FD?TNT^Dssa9M(YQ-D0W?J4nc8H%iTfPsLxf{
zzH!~kzWrv~pd%9_P9f(jbkEoV(EH7G^Ob3FNd4i!V%;S#9=FHa8)8Cek*wzqOkCJ7
zIb7yz_R@cmPK5alQ=C&1%|S1NGdMMW(k`^k4!b$+l-fps&27i02+MC(jW&nqv;5Q}
zfW)b(h0r#I;zo_kF*<b4*}EH(+>pvX$~kCyA;%d~7uavNS-JF*<xU}9$UPGQi1s_(
zm_uBG;2-e@%h{GP`k4K0bp|5Nex(EOsiG!wOM8^u=hP>vH<qk#zt@fIe(eZcwlfXf
zX{RPTa={^1;=UfV-%moha?R>Z&U4|Z;v0}yVmxxJ{DW>zI(}9Mt_(%mtWJG=S@KOL
z4mxLlSYMOwN4L7mF>E)7!a1iVmS5JEJ@!Xkt7))6;Z55N->K2Inv{!~oNWHMTH$nK
zy}8I3QUy$hH#xdT>`xZz!(O4wWr4+*zBs^pr~PSLqsuN?;uK;4V!P!PPckn1vu;q>
z;n+W*k1xRR{G<y?kMrlX^CVx1Tgj+!onecaYu?Izvvq8Ca-01{Te#JPJ1XDEtOL?e
zR!x&T<u4QeU-P_|IFI>5>?~Wi*<a;bGxEdXIeHDd8T!RtsGI(}eVFJ@atmU>azH6^
z>AIcmat->MT2I|N6AE$U1?_>-6FT)Z9Ve+FP#qh{-?pz@iQu+PN*5sNFac>(_MWVL
z9M>!M=WXBwkiv=Zc>rB^+k4xbU%la4r;r;BAs#0jbgc@B+|5}fjq6T(UmG0=q`QN-
zk<@|ZS$_2x*WA3nl{VdsR)vyi#;f1_TKl_veCD$+avn8(WV~iRw%Z2^A5zV%O{TL$
znZ2k<h1JI1&Gx}oxOB-<rw}-V;x}{`viGJE>_gdsmE--J@rI?O4qSfuGW&4#U8G57
zFD07-7URwHO<P4L=mM+vkv5vK(cX+x%AKt|sgu(L{@O>oe#lWPT>Y3ZT=AV6t;CWg
z%a_>4@-s~6Ll6i`i9H*XUE4n1<}{QNoGj@0faSoini{oF<oo1!B`RRpDb~;*2{ryH
z)<|UVlzlRrB$vrj=Q-cPprsr8<aHQ<ZuY6xJoc)IzEcz?atGLMwiNa;`*i&@Rbh)_
zSe=^e)>hWzGYy#((t&F0G0yJ>6FpQ?DVcUZ+lceEz45Kn+njRSAv(cj@3zmiby(Ef
zMeE?y=agMd!qctx`7+5vmx5|?2`v~SbK_p*NjV=YQ<j<+11DLQU&M5zsIeQ&7b$=s
z*caLoUoBaZQZ_U^G0H*r_c^1ex^8uzv)?f95?s&o<9}!v!Q$X6+QXa)NON@6nt^Lq
z+dp=zuvc2x@UaOI)Ts%~Q&%#3tyWP7pE!u)_D|W`{0^eSb)y}((mF7q6xjlY?4Pq4
z(tyrPIfdMUk8>S{{Y!OJiWCkd1g;YfEE!bpHH1|gw|{M#JX~=|Lkv%bCJ$gOjk0cH
zFky)Q)|L-%3%GTstWM1DSijc(y=$p18pKu$tQ#223X+V{!Ah9bN%N5XN7XN8DN{QU
zp^Hv^qfv->={@wH4F>Bee?TR>0G69^1A6)hB(eWWq5q|W&Qp9OKMYd9*!_2{MIm%(
z)D>K~)2XQ@Gw4@=82^(~I^M%uY04?JL2>4%!=BWKVn!gP_P;qP%a$!&;ymZqs<XT1
zcAMQ<<UX%8BNtc)*{+)=$H<ZOC3luYf72k&=F?6&7j<xw<YQlYXDuKO5RthSKy*oZ
z+lrZ|GYaSzgLv4!tm;!Z7l&J7J?lWCHE;FOoa&m!$XFO#S(Emdefgb1IoEDl<vhlS
z_RdMvr5aWAS7h~)70bOIU-GPH+gG+eqm;kVDaG$9G?!*~+gG)5yb@pb9t=GI;nU;y
z(dJj**><V+XZn5WBzVZF$@8+0oFC`q*VMbRUod8*bO}`CvU6$@GX}?oCrFNf=iArb
z8JiL(Iw&3moe7sMyWGC6CQ0gf`Ppz<$43tE7j1=8e~~(;+X$MUSvl;nt>gyZ3G(*!
z`Pm_;2#?84%^$jOn|(v|G>n#-J~R!fFQ6tzbu-b%3JA7uY^#PAQhLWD65ph=5?42W
z!drdP=G1QJ*lMTWH{DtEi=0P9)lp1>JvKgU-(25*qPVgfMU5O94m$Okmx(PC_M|&g
za*=*UnPe35E)-EnL&!23xAT_19toE0Te`(Q6bFtnE&>axs|};{wz6C1DgCVtQPxjS
zPdKI27xo}_B*WD8om=h6IXlC%WZ58Z)##uDQG){txw3D|!l(jXokDJw%@`)^+pFCc
z4MNTZT35^nHA&xJN;=<J`;K~1?!r<a%WTWtJTSV{)Gn5Ooh`#j_ML44<95)z!Y`G=
zXtfwF_3;}hxTwV3{fHgk)nV8|jGzQf+-nrL?5#c(uUb$AvI(m=o=6Gw-POa~#@f14
ziQ)z8)MzrUoS4A<{CnE?r?w6aIwh^-ic6N*_qJsTJ76fPy}`;5OxXHC`@U|ej?jYC
zvP$Fk2&z|VVpNLx6UGxcV&C5ocM@(^_Y*jfSZc%vi6|mLwbVb*cKjAnaVRPQfUthk
z%;cm({#fZBtS0y|n65$YbYQ9A#CH)>u|4I^u#>;(`OZ@<jdp%WQ})!ht~A01?21zm
z&k=iC>q01udp!q`XaISb*lHJb?MvQ?1HyrY@Gc%h=TI6dP)^pf6c%Ug>8&s9L2(>9
znSd46E8k3<YD|a{5Iv)<63zs-I;EiC$rhw6ozd4J727lGJCGtce$frF9s7{5sLAd#
z^^=qKtR$hNti8&4uEz05jj1f1g+049CtfiJNr$_9{od+h`uU~oy>YsMiQ+CVuAzd7
za>SHa>A<ohXe*fK=j1$dH7UI1NN>aeRBsN&A(QHMh_pSo`x+Slv^nP=@disBiC`z}
zd2Q*g8s%n9BC-G^|8c`Adw#<aDSz>nrku^OUFAUZU^r{Z)#(M@HWc+mv&yE1zloKV
zKVz%k!!O0h>-G?>0*Wb}EViVzL}VC9>y-11xqfg=`b}IWFR8{;B+=oecoKAu-fE=K
z&0wfvjBIKBMsp6UiqQf6Ck;O8)C7`ig%oFwzp5?E9Y4Qg+47}Mp>{!f0-k0QUs|6_
zS4CY@E<X`viKHeMY8AHZo?VucNR;)M^PKw|y{pn$Ghka@Fz6Be66&4w2FsxP(0kbB
zZFM&3LJHg=0OYP+$qlGu6Lv-SafC_c6-A0S*j@+<gO2%KQR|}=BhroNZz^!b4xe%A
zMbWkl-eNE8<}8xmg!(<)xt*Fk1V(z~uY6Ir9*Qpw9#HtqK}St&yi!1vv#K&$%2#)f
z{ZKZxWcf1ZQJ>U3a{V!T@txttz6zEid>oW&)By(4Ne3-8i82|6j?`X~Bx}R8*V8LD
z+YfiWlEA2SkG!9A2c4Q&2Vm$dlFRy$hV0LJuJauAI;*s-gge^G8p)!}m$5Uoqt3(j
zQ=be*@0LNhew-ezEV3q7AoZ3Lp7UK1;(Nl~AQ=<hDt=c+I%(u}Ixe^gtg60fp;WKL
zS-^50=%12s$P6oPVVAA08Zjzzf<RIfeZeF{P4Xw3NX0tP3Rl7wRzafAtX);zkqu9W
z6OR5g3g)nfWk9Dsr$z@O1KG_Hb#>S1egtGqw4B{BdpR{7!lT$=B2L>i-4`ASE6gtN
z6j;hk$mygm+h%Lp+%W>hoay*U4O4pANMb$fycKho%tt0Sqj_-zk~!?94NG+^hEqy%
zh92o<)d?7-;6`u=bYStdeG5DTEr(yXFwL&bN$cGr?GJ@oHW>+s_eD3dsB5-1U!Fjw
z={p6rJUoFeeqFbAdFME*28N-*a%z%rT!oM_Y3o~&=vb^w6l*HahrUSHb8KkKNV-g?
z5J=J3m(;d{n)dSgsa8hvi+pB)#oGzn13QlLnH4^18{7J51A_t5H3LiJS+;DsZOYLM
z-U3DAS$)RX8hqsDHm!UQuRLUg;|?sR74-vdg7H^GR#B>)5O!T#d#F7r#Ks&zj5HbV
z$X;XDH)2PoPA;K$%9>GFqCvZ%h5|1KT}dr*V1ZK;UqVO1QKLO^&|Z<hB%AXtujlz4
z>(<#T7Zz0b10$d_xr?-S>Z?~ul&5KX$bPh7V(R6|j{CC%i{XkdlX+3eQKVdyXBe(9
z_Nw~LeeCmh!h7Dw-UJql0auKT2-z=~cZu+?ZeN2JjIU`UM)vrK%yQ}zE9e1MVCfI-
zH3^dwNn`sQSUE2)sIX-v7SojCsC166t*_x?$x_MgbJT%prs?Z3dtIv!zJ`=%C^{QV
z+_?h4bgSR%>%Elq#A!d`4HoWI%n0<#2HVj-I6RE?Qfgp8F@1{egLMj{*<x}y)Ve^8
z)TYvM11pJSYGT%g8?LThzu76o=Tg5KRDlso(#pIcBW-@!;pL;sM`a5{WzVOn*QarY
z*aP$2OKe}QtQe%Vq?kL7MHZ1_ukrU$4vQ@fT}Sxk?kT67$P<Jtm@eDeMhe>zR5P^D
z3kXO3MKs+~JN@7OZ`-OTvNptN9?C6*4R19Pt0f2>3!R-@5=PsBwQ_tG<vRQxkm}{5
zArT*h6@5c{OS*L3jO0L)7Oa}Q+HS0d*~%AWnwQgMo@><9eo@p&aKAQIFi~>tmxG*T
z4JJoB*%UajY<o=&HaMM`z(U=~g0psb=u;o7vI7_2LpI))pV+oBr&NQ;TLbX$u!($Q
zbSjb)R|y^-(~3=YdvD?JO3w&tif}2CSxyWhX(^V-W#x@+YFOg~(~9r}q9x<qaAkro
zl?z>{<regf)WS5{RGY@B$?Z-dIt`k_r-``iw0`?McT-H<!a3qV+LY?p$8d1OrWaw6
zg4x+OOjvg_hPV0@p)-Gjx+Kpd-nN<grqswvZOVb@4TeDyQkSV{Z%T~bG>z+raD{Si
ztP67cOcP8J;OK+hoMdAI?3?pkdzyf2O-KZHamxxL#k<dLsd_PV#FEexNq)M!S;YK|
zQ!h?eq!Kc;$7UPZxM{R<)jJ?f&HRq5X2!!um6P_?Ze<^ct(B(CoqUj*WH*RT)!lCP
zriMbgs>&$|!-;jkgXqoO(qt|*5}G-%+yd8+4%%Dl1#vvm2=2ddn*+<ET(@Np+mAhw
z`hpjt<)Xy0k;#9FRQ}Fe?MTi|%5aJKhMfF(*9R2M{6=C%gXIA(bH{1(9LB{LKks=L
zU+k0;e}gh>xy3E(`d{s`hwQE0vdDs<@5l>EGEkxc^>Jp16%^-uhFuWqt9Vz;Av~tt
z*5;DTgk0@D>i~3MY*1L#?aAKXh`7RI66qn*{Ma4!?Y7J@w=pByJGy5;9umA22>%XH
z;a;f7y%3~3+qNLeU&3JDfpji#^gwsOVJ^L^W-}#^ysj*Q>zo>OaNYWK19o?vZa5w}
z&(WHYQf6HEal5C@A!l-nVwBsO91(RsP#<fySZf)5x3|?VmBB&*mJhSNNM6W>3eLb=
zz2;p(_{z_-NL^@ePaH#Q!Vs%KjO@gW{Y16d#b)C+$)=jy?bIZJz$l|{a&eFSWCv5!
zxmq~pz%&D!HV<sHpX%0SbYjFQWsdo6Wb7yPBHHp<yElca&9^vD*?`i=YmdUvyrTvh
zR7<E4XQ5P6!LU2^`VBXNw=|4>yRU{IQ?vZK^tB|q)FiFQ{m;h87`+hLLAQ5S4^l;j
z#RdBxuvi3%raHt!ywo@Cr`txobomme5S`D6FfpHp?Pm&>Jl7uFXbyV^Lx#xs*$3=r
zs|kfrDq=+*?=qO2a%$qM+LdCL(L&hIH3GG|T`uH+LKD99Qv3P(Md~2Jhu(qZ?sys6
z9Q%d#1{|7*y1aQ8U~;Wh0+{{wt`0ko!udk3BN#=7Y`MbkcIvZl^fW!7PunlH{kmn0
z5bNHE11RdOcv4LZDNcz|5GCF{chG*R4Q7y=N~gCeE*p^WXX{qB_%Bx*T7IS8<$56p
z7N68XQ_gkNJP759{cgX~&8DQ(<jUgAbZV-{iVfwPXh?pw5z{H8xLKzd)LA#B+OO67
z$rFcZ#T!gLl2a5q3H$X%fmt7&;-YAw$)HErB8uMdR`wf<wd_qflS(PcS2!$pG_(y4
zJnu0{N<~flQphKqIUh%XVx?HAxyS7OHi@JV+KvSiItfTiP67|<Wq0O42<O?3$44Zr
za$wE6f@7N)Fxyq*l>uMgWar3z0xa4-z|!ej7l~HFLuJ?3ts%P;^|s9nI*&2Eps_rq
zQG4&SJ+0+|vA8`=eFIkF!WWC*n&RED%TL?hZo0&^=A<qn=uUl}g4<9O58A#~92(t%
zC6%I(Ncs0{fAPz2pVxmooN}T8J=9=i_&d-|v1~LVq6)duE#-TlIgdK^r4J*A_4KqI
z?0To*#`I)L;{$9ZXEgQ61Z?U}hid$RHr>kVqaeAy14^aNmfqBMB47IUU}9e{!ZE7w
zmS?@`O|Q0xTA$&u6+(U#Mgv$$2+Gp9XAWebC)kke;kE@J(sKC-H3N*nLrzW3+9pDS
zZnQ(WdTTaravo!$#bL?Xp1K)0D3BQYyB%)C8?7+JR*7?G9bgV2Z-_v3e55|ZSznVU
zb$Z8+IrTY_Ou-b{-j8+@g#jEPzyTx{(W+z$)Jt}(#uGv=xX>xSz11i~cGC_8t+(Sj
zv?Z6Va2^xY(PhrUNx(FnXrNWwXqLM;#vD+jrWJNq7!A!YvYUa{@}Y|>qIp&vI8H0}
zNH#P%xYc>ot^@0OxX_$zZJ^!fs8cR>Kso)4eSE?mZCE6Q5{*SN0vLprYPqYYOH%L2
zIZfO!=NS^}UOUw$bCNMZZ7B=uH>_OsGCQ5!y7;*lJCAC06ZgOAa-lP=)u6^=y^{IR
zo;`^^hWa9thhf^%`}u6a#OW(JCQ(Pe08@St5wRn7uD-3WYV~COTpUqA{l#=8Pj4nu
z4qZ|%rbCyEwwblOi=42>a!(ACtf=o4iX%dl;3JV@C^V?&JCv716kHHp5}!v+QO^Z0
zsL1cY#*hN*H*(R~RA<rNQi{QgzU$P3g|Odz!WT-;mv&zhhL(c*MA?Yfi+f;fuKP{r
zoX67T_2TFooq8SF8#ZmU#~XHQS8F!PxbUF28jgMJAkDNWbK-n1=}OM2L831yYT;LH
z8=STa?Q`KbP1<dVP+@eYX3@fGi|@DJ>iXIFtT3-i9W0S!zf)ho(M64+bFp_9Y(Kum
z3e-R^<G?P`!b+f%VJ00<w0mS0%h2X_^Ym}GZ*s!u_b^kw029fQw(Fq%PTMUTx4}#}
z#aR43FwdyGS^Ur{&Oc(m+ZY@PkWQ&vLMhk};{g48_1k3F8xyW*0&28p=XY$KnwhlU
zujvp6D5fS&6S8*)@8Z<gw{>w|P*(ec_HBBkJ0nkqDgooJQxk7vpF*`Jw&{%hVH-tZ
zeL6ELH!3x;i0jv`wLhx20)<g0iOr9gSf?fzBm!+C+Q9y}-p>9?jUip=nG3gjtI_V%
z89~ZnCj1Nw|0i_{rwuJmK`lAIT;vVxycKh<?I2^<ZmI5A+ynNfRg+2!W|^84KB%cV
zQn-muq(k;+-MmU67#E0B(-B&4^}&=;qccBE4`HCRKd;}bAJXm0*(&3k<Lq_n#kciu
z9^Eo+e^DKXOCV?rt)T-;J_aHR9|OS^@??GjT@~?{?Hocdi4b-GI@*Tu2P5g5uKD<@
z+_EFMDe$P~AoK`>e_hSBB>6b5`T`8@f$=EJ=5G?6!*d>862+ABmn~Ugf7`Z?)VMIM
zicAe@pYA`XbFlKa975;KyZYgkQxhCA?@96L`cda8zlVo~y|=9n167n<BxeY6*lNUp
zN?)SWCia89ufr?}AeoX1Snz7;!=o7a(8qMRkmBh3YjWX$gdd>>Se$>2`s9Y<&J^4&
z|E`+aHNT4=@E6^7gY9FR2z_^>1^Ymq$$9YqU#A7MTyU@9cd^?0naK16mB)i_iC2)=
zS^Hp3Pj0VNEv2|ejb?FTavRahAL?c#lL>Lxkps(5b6P8L*G{RqL)jgBz&_l5gD@;H
z`N)O0I<Pu|<X=}@44w$e-6F^0kK9>G2DDJ;(L(Ly|BpJ*_?56-k+Eq^l53mP%ctz4
zPZ&g=8yR^_jc8g7cAtGLIX9-6R5s6!|Lt~wNtzO|5XN@d$6MEtvvjs$Ja=Hx37u-X
z<uW2OBGn9%;)UDn6V(ZI5(y2-LGmL3Q%+5CUF&AZ${L~FpKK_)WoD~WNJ0h*V(hzs
zux+2JuO-WhZWd-d&mn5`HX+EmT0xfjt@i0gx(7$$4Jal4UwX+U_L)U9u4t^2G1m&2
z6W;H4>KF0O(6FXjw$HXMgx(+oyxhM|O|9aY2m_|e3{^Q80?q7mNgpm<eyQ`Q8~tAM
zO#6I3-FS=hl)T?sEb6oNg&KZN9Q$f+zXR*2ZBlC~&7=<g`+9p%TqUC8@stD0^DuU@
z&ZoTmW1sy)PQ}pZ)~#bsAzov5(BZvvGA8OWj$QVTdEk(@?G4Uz5p8+eGa;x#|EX(K
zVf$+IkORx(a$suoM*HXbHn}@PB2*HS<hojoWOfdQIINRc{7X$^raTe%fItHer~7$P
zP&@TGEX*0Yo_rt5g4hD~uZ@fhj%>%Zf5ZXgloD?`HfjIXEzp8_IZ2lA>4;OW*SX4o
zyU|AM-`m7+cHz$kd5|Jr7D;L}$bn6p)~sJ=|Ir4?=_rqWGCS5FsY&r?)%tatH?CZ@
z+5R(UKD;x$3iCU*<D2wfU2D2~dXe(Nsfnq_7#ZqXxmW&o8!Cy&C*&;-Ads(Za~<a?
z3W&+xYyXp?z#EXF2+Ha^W$lK+$V&CU?Q7h@gmr1e0aTMxj_FuGZg$rK@=49V;88@4
zrh`GBqHAArSKDVxu5b#fn%_afhe`X=yV?#}V0!l;KbH@WQy<OB{-#C8>)XEUt|AjN
z<1>A7k04b73Nc0!>X+XYeDdhXtn(PZK7dk%GZnAsN9-%|)rnikI>&RIRfSmP1`$pp
ztZk)&v9D}BzLNacPAR5U!)<ml85>b7L68pESG69q3#HoDt!WFe6a|$s{HwE@%P+so
z>+y<9FSD=7soF-WlPRwP%8D&p>}#_Zqr`#YB21~4gIHED3`yO-u3+MSq@^U8fCKC3
zgIH<_G|Vro^4H%LV)f&<(QD~AT3Lo*T?+UO*}q9{zTmMiNlFgZK3Gf1Q|U<mjoI-0
zj+;lvlpIJI9t##Cs1{RKxP4P~5W*w6BF`poFzsqdVMpeU+Ba9HGr2!-CKm|<Cr)Z&
z*m41p(goLMdr~c;zL0kn%P&PfHflkK=VNn+oVRKWCqk_lIZa?o<&~nG&fn6tc`)q8
z`B%~)HT6qr2h*W_&qAUd`__g;(WWN1k_^yi`E=_0=bg8TrizbDhFO14*^_gkH(j^Z
zd5or7_vJI8k3@DTR`}a$_%ZFAByv;+mc$QN-IU5D87qtWx8D`SYe<(aacwO{Z9HxO
zAKilgYgXb8a^o~Pv?%80CpDXWXN^cSl#js>?qb!&OekTm2Wmawyp^Q%%F&7buzgp4
zA+Be23X)ZOw+!2N=i|V}_0D4qFTcycq!EEQCWH8ZeNS~_gbN3pLX0ctB3)|sLKzO&
z_qM52iH^yoAyuu^XcRidrtJH^Xfgx!8OvV!j~>^l7tP?R`Vss7?Bq1rJ9&(DQxPlY
zF<C(IlC~eH52iRDXcg`1@RQFo^{J>LKIq2Yto>lUuOW)B<MJ#}HsY_8`Ov9PN;q&6
zal=okc6BN%O<VfG!|nM(P2OB4*8tTYwx_lpX?p=v_OEO~9OEf(_2~>DiXc`?`jT`$
zd0O^h0!2DnR!XrvnMtYSvqf<M7v=Z}ZoCQe?}$^@W>iXRs1+(nQt>>!Aq7{vaF-}W
z%`rDk_w5-qz!WV)?O2CmtI+^5xvmH+JJg=p2Do4%X37_IgVZOIgrT{4x5(l4Sw-G5
zf6Y0k{G{kCu{y)#jYDE1F)n7=XXhM%+FYIElony#9`q|>Vb39tUfk7Stc>|roPsVy
zqb75$hp*uF-gb5zUZKxvv?>)~c$hrwz~b`~yJbz?2YuQTk-5+9-aZ>DW{LN}gdkh{
zSoHVU^KveSbgd!JK?g+~(uw~ed;VP|GT}BI_cQ<~!X?9j^nX&5Ea=N$(CWhiu7Fsv
zU2&bc%nMF^bXIB~nUca_4aafDmeem1pitL{&1)D7+XQA)FX?U71ozoQYKBW%Q{l52
zR{_zKqQ~Tcc-p&qj!XFtijsqdLF11>otL(*a0<vPLut;!-r>|}%g03)wrpWq7>l&=
zw>dENrLLHQjlQ&A%GBPGzRiKfP+RIrqMl<@pRvojp>}~$I#KAK*uO(gy{I6WNpZOt
zx8>ae=U(h0bex)4vix!UwaXifwUO%Z81BCfrX8L}p@;2?dJj`C2WJXPz?vdG?;Nxh
zMYjJBr%pN?(u`Vh-b$*oa>JUwy|6nx^Qs`FC=gOft0>cA@Cwdb`I~D9-LV%{ZxWZJ
zX~eO%TOjf<(ViT(f^yprRgIWV*-r9CO*t4+(;Wxe22#T&)lKckix*A3Sp9V7IR)Dv
z8V;wvX7m9x$I{w8U@xiNjNuDY6s7hCy19rGv>z_mJ|yA@wU~R%J6LGl_F`Or%zmWa
zgU=JAgA3YT2Ua~!kw-d`L{n|(L$<Q&WvR@QTTXimI;&Q{m$lXWqyxMXLZH%|-)UDS
zuHt#9Fl0z90MXCct&rAL-4tiIqwrC}<ik#l_5j`oT*kP%kpZG(<dEV35>W=XPW5_{
zf(IG{jjD<(*+~WBfKuPGWt6+e(YIVxt?3G0`TXatJJP#SqqF^rrI*>&^->7j!W+&X
zVb6o!YP8<yAz^{Ac`>_hwQIV06&VjRwKtgZnU(R=*0gnEUD0g8ZJ%;r+H4q;dZSy%
z?WNtwu*YE9?mLiJAo)L}fkCFWF6o%PEc<>Xu_!zy`?7szd}P9|Z79BpnRiNoIFOJg
z;@R5vgEidQi4$!#1=lanz{p#iZvLT*v~^XVUbxAziM_3eJ*QrKim5$@j(7cCrOzQl
zU2iHg4E7zU`pC^BqgyaRZOGnVe)$sTQNJb#XKQcJUf!*P9LS}xQbfK2E2z<84UTNH
zjfqS1J2qb9JO{nwLMb+#NKMPOxv7mDVF&{PMBotz7H_91rx|PEjcs#&e)Yfx=dq?t
z=_#a9-euR-57VPU2e+s)X;qw>+?k>I;{?I5e*K~ut9zRD9Zs0oDc<UnH)=Y49T>vD
z!*)ZpCdo`Eoo?9TU06M8)Ea7wS>jXBEuouzMfCuc5$l(l@&*%!*YNO_^`0zG`l`}H
z0~4pWp4087(caDWqYZg(BCn293aPUC8EX#BmZToMSJK5-b#wE@84jt*xnUi}HzHfw
ztGnLH?N5~Vr~^xN*}%*cVRFPN*lQX{7f)g0LsA5k>vBOi^@tjsv)5J+WoQviJ`55h
zpRIc3z<OkEm-CiXQX^N|>*^ahB3VR^IWUo?!4Y_<*LOV@Q3^TRfhG5V?L}HkSzxh{
z%$>KvdRx0)E*iRt7+CTu!4!ng9zDv7!E{mRP}QTF=R2|*NJl4~0awPL;if)Dss4+F
zd<+UcTrka+WFqvY4lH(^u;+}8R3CHq2CndJED26c5)%3&Zdo)OEJHIM7+}35x&4VP
zaAOo|H+2_v@UjksE%li+^lSQ*_(XTWvzO47QlH(Xi!y$)*tWHUMecV{=3xgW3eVoX
z`byi@7Fjnc+%!6cWdKl4Ho7y3L`uIC2a}C9)Et?GK2nPJY9WIe4ED|5P_3qgR4nfi
z2UcG_7A_0>nFNKT?YA52?YKdZ>ET{W2PRUpe8nX;RzH*tNsQEkDe}(kcWTOq1P}eK
zabxuMyUNF_KZ9%OD3+7ySqB*Fgu;PyalGrCY)S~BtYm|wO-D+<T{h7Md&w0mmN+Ht
z(*6vAxro3v*$qehaFc^*!Ko3QUCr&y2sZY{Zt**fR)>t-Qs@6~_2oSUEBuH}wZjz6
z^BB2{96(Kbmq(vj!Rg2Pxr_Sp^W?d3m@||HPY@S9ons)_UsoIToU-2G_O|rG9VDb{
zrkiakJs{g~V5!xoCQtRvxK0dC&e}~iG7wtKes~l~ld5}*4#O}}pPz^S7)8*NfOK=$
zt&IOA(ToF2%m{fpRo<}>c1yjT(@RVl)7xOuBuFbmW-6P_fk|gHGQF+ul!ZMnz;or+
zT6-lK+3l2KRw^(6Dt6S~R4~b?<xArYrlCnPgHGC;^P|^oUhh11oIxrVWmN1EuGm|$
zQLbR0q&Jju9<Cdk9z}HMAvgS3n>+T#6H-RTQG3#<*9b0KvBG{l@oCH8=oq||QxvHj
z#(C1bx~@GVlk~$V>43#-MwSi`lZ$|<!C&0ode^)2+Guzbm^)pT5@-NC$t6>y3CJ5<
zx&TMJt!qU|kwpN{Ik5Dl)D<jTGCHc<Gq<;CfgYmX37c<3R`DZ6v{J8qh_?2+jce?V
z_SuLVV2-g^FTj+%llv%f@w&4H1dhvepr{DZeuheorh<I0*VtW+7QCUHr27u2j)4U&
zTIvhnR4{wpUG?%&khzz`xJECV?ZZxeA`1i2mBDN`yQdDjJm6y6(Ip8)VwB|hto$Xs
z6)nrS&2a^$Dqvk|-Duz5mL0ziGl26P>y;u&-b@tw410ThMMs7FmiMAd4En0QHur#2
zuOoH)Hv5VC2C5}VxyZG0oWTRisfjV@(2=1KErI=H2b0JVsw4+iOS28D?oTZmImBb`
zL6|>aHJiKBX+{_qU&Y`G6~DJJoipQDI6~q;mku=5So1ri7`At`vu6$3Zwf(|rJ6l$
z_qA8I&Emi`<&?8a1O@l)o!O=N9T6mj8$5pczzT77PzB$nsp-Oze!A*YyUzAtTGZuo
zuX5_uxza&R+0V4jZCHP;Q;3<<kL3_`eD0+EY;Hxou){`AF%QxemtSE&SMSRRUH+)v
zU=i6!({g<H`L5HVluiA&U>c-e!yX%)l>qh&ZBjt%t&nWWdE}8ubV@fCY`1rH4fA=B
zk@L}k=?vO}7ESl3{bIeMlUgkOnYj}VEEP?hE5zQ*K#F&y{Ze&L8XP%XN=}wxOjq!c
zR?(>!(yiTUzub1;$cXctBwU^bycxU{&P%85SF$JbJI2vFk;k-6*_w?j?N{50((jSl
zl?4TmdjaKw=vlhh;I3=G)?s)h;!L#}5+pT&1R@|(dwfxli2ZsuJr5|~9X!oB2bQzA
za&l4$E~b07qZ(k_M5vBbL$UFk8l8Kgk5i@ryT4v-1P=cJcG0mpa+K!ey!-~+*|n-z
z?S1n{5F8pzs~|##n%8zUBp>G9Ri_l=U${eJci77xwA~F4#wSKeI_{L?&EmZxD??Bt
z+fy~=o)Q^OWxIILsmZ-~GdH~9-MqKW0@uGOM2*h^iXFKF%Dcz*<sfvsD<-6jn+KK(
zfVC6b29bY8Y=5pW0SknI&pO3ek33>iEs+d+IA+*^ni=t@QVfZbHJG%QQpiXeJlOSa
za$<TcZbt^B{Whjc>Ft5mIYeBN4vpU08*yM7A-8!>k^mlu)@n<$B-;(U%z-UJ)pWFy
zdg`=2)YYK#1(Q18OQ)u&sF=O*1F?tO%8}Zp@08+=bZ1Jo{!lw319gU{Mu+Gkz+40z
zV28PB?r`p+DI#^8=kiR>j1-|;kJC|l0mZ#*Vvcl!Iwcbl2>pNqOG+3_kz}wmn1?!d
zZm%6pc47e0n#UNw)JkL=#EU{vZTMQ)u{PsZ_6Au^<=E&zH%F`nNeb+E+t9102zGIb
zF>vV-L$3xOdZN98p?iALDb|cDv!3_|*=RVU?2$IF>~!UqP;zx($)%8zLa|}ewaPj9
zWcQVpFR9(?LQV1_+fjEwnP;~{@*b_%bAu!`ixa}BQIElDGCn1XW~b`ycC0H9j?8$V
z%NNxki0KTG8u@f$;}pSh<P&W}zJ^rQgLbA3jMJukSRGhQL)e`m!e>KE%Y~PAwhaxd
zft(j*=cnIjLc1CmJj`a{;2AY`u4@5tNo)tU17PBZmMvSd#2(8IL39t|@$D27KQSLk
zF|SmX=i7R(;>wENh(td?IyVTf1!t%2Hxg}bnZ4C{)VCkM50T_I8{Hn{1e)A7>j2{G
zGHpq{b`mAST<aj*l~P72m=##zUWQ<t<L$7^+T*P+lVn<%(xrtCD0j-5akluh&1Vlr
zdfO+Q=Op0$8+#-;m1J?Dy*{qg6oXDV59guYt@c}4kc``R3L<jcTuHo#y*tOara$gH
z)+8zaRh)PB+c_hV$#MIvQ;ZQLTIm(JT;fJ-ztfFZE^QJV&N{H1&JA(K{BFIXTc$#@
zhK&eVj#8mrP}ASbxh0f>gtH*N7hs?P?DyMV7tsPwuqW|XCLq~X&U;rM+7Hz$1<xMN
zWog%s+8^XRl6mb`=Q&4=j$NMiN_zQ+MTIJ@1z)K=Q#ub>FtY2gO?fLRPP7t+7rvjj
z2w@_(KWanYhEGM`DMmvf-O~_fyZv!Dmw5UD+ZWa+v>HxLd^{PCKzMBk_@!_Ww?C<0
z?wvad^$MRl1~1ncYEnq$y7uX<GvoHBjm<&JJC3fGD=g*m0Jt`zuKP|tK8}0)v#LRL
zgXkM;3r6Fi;?%?#A~cM_N=fkNIZXBsYUJ%sF=r*RzbBGphPtq`*k9BqL}#R0d6>fw
zn8?@4p`oeXO`}Ea{<3@V9GsBkk%JCQGT|G@ZU0yKI6XM&JjO_pbcwwcSBV3s{dIkV
z)Az#dG`45%5pOW9gVc2RP1)bHIYVC&4ln=;D-^7+!ILAjo0jCC<Dv+b?66alQ%)#D
z_{F`aZB9}GA_s`E1C%71V3~>WQ7)Coa`4`UBAaHgn>$5q%P{iYV()9)ch%ZWPC+9l
zzo5OpO$<Vj!qg;J0?Wmei<q!fVk#`pZu`4h55k{gM?VyYU6>!F*>LJ(q^QNhGOuWl
z4}8&thP@Nb^21I|A~7=}iwh1J_Q7sm!j%d0fD)W@PCGU6HaieITWFZ=L+uNsLYAma
zf(&o4Kqz_-p|lU*RhLD^-6FU(hF*HX_c+ZExwnIbaekrM<C{lW(vM^b8DEAUN;%3n
zdh_)9Xu;BnK7~MFxsun+q=)0ja^A04w#0c%qN)4r?c?3p!iykIp9T{ON>3*X+9&c?
zqqoj%9~^Q@(GLZ^%m0csu}>CE(f#NQF5HHX0<i4IFuCQ&af|;{y(9uA^9bIe4lD`h
zjnf=*$L!OsJiToSPShy}GnHnPFeEX#As;UoPV6%alt#`>sZ$Cs(lGEj^%}O)dyV$&
zv(*pn9$j6<-75~PD9K^T`lzO5rjyse=c*>{z#3YHr0jKSbeOK=pMAdVmJxbAwN=Ck
zP@J_7%<Z)=RFhh3`Hm<k0$4Ped|tRYRg-_8Z_DM642RsT15XdEh^YbtQy-I}i}alP
z?H^KTdEx7wC$&ow&k6g-j*=-JL^%g_6PeDE^mwN})-rBPq+beV{HL~VoYxcVa|aL|
z)$GgYF99_+hy8Q?9BC~iJ<(TS(NW3T3gJH&{$U}s&z<J;|I!U^aC~%&qMSw?So}tt
zhi)eO8-jkUxczH4c!CC4oir(s3PxmY{*Y6jSfsS1B*6Q(ZfYgS>g17!z-go>YH|at
zB~sJ=y-g3!fJC}OZ*|Cu1(rl}*t>ds6iE9P@)^dEaJD(c2K|p3zf`OG4Y#s7upCSH
zsEI$9;Y|iz?%_Zj`_B%a&ZP>2BR^#tYSgzGlyK-4|Es<&9z^CWl3+P7ZK;W=t;`*w
zlKpqC`qQ7|JZaStM*8+Y3Egs&^Azi=kH`w6WU~L8Zw?OkW}WBgte8D1$3j>VBgA0j
z)nRu>m%te@ew2~|A;-vT2&JTb$=&sZ1(T|hmIXug@Z3&s^?{{f_0?b$5hD@eVf)hV
z_lkkk{*X&*DpKIjz163WblmAgz3grhz0cT}Ri8xrlo<Rh+!(M7w;vlsI=~ZcViHs7
zm$%-FuG147UcJFG;96}Qvah&1(Ovz+ZX|I==s;hnso8uCR%fs5|Mr!42kPn2?~$3%
zDVJ%TYwyQL$!PsmZG@a_8e7H~Vx45Pr;tsk*RnI8$61}PZoi+IM5<A0CLl>C79nZ)
zklNRDBRU)_2C_S_SkX`%vf_#yk$PR8o~N;ZeQnhzSCdRdgyA#BedwI1*CD-p#WMT4
zyNiJLBJ1$10}wMlf}(iZzP@W5#~_A+HpGEdSL=?MprtZud_$W$4vpp~11xBJa(>C9
zi8LZzvIj+FG4|Ov-kprqh*%1=Cc*dtOtlvt4ahhnk!54nzUl7R1YAQr>^vuF)1qqH
zncQ!_J2?5#$>7hNV!Wcx6HzYJu{|lrh|oIJa|$t;dEzwwQf_!!fMfP8`GFy_-;Yj&
z5IXC?>Nml!h0Dxd<h6>)w{NXEN`^AnaC(Q1%&8F~tR2{V&HB~$<m?F6kp84osIwv*
zP8X8c*taGA5qUN?gLcQe@)q5z(YJ4}?xZ?7S6zS0yT)66c~i2p6-FVWOqg)&J8F@F
z^I_OhH)K-7MK)sRE$I`V_f-4Nn$u4HMP&v|1DoGLWZ^dZu0{BM!YL(9(PD!};+wv^
zMk}J%?k<(Nbf;63^a@;ULGn7rB~`=W+4m&Mu35M0+Ur)YS$DMqh|def#m+ni7JP5R
z>l?0Nss}bYfSBF%G-7r_rKP~e?`w+{JXfm0bg-RrYU0CisFL<|blCT|#-N>f&GC8f
zVC;Ue6kOL~KahKD^}w~xV+<%JoBrj=cAk?nh=cuLF7F1k>5`9~ax|OEr9JbQJ*ANh
zj5wo_786Je2zxKYD~vKUEpFziT^G-@Yaw9>fxwbc8=08FpxC#k)z52Y)4@*BUrtS~
zq;4Rxi*h@$F$o{-JH<j$WgFcmu|jkKX8h@GvLbP6>>US|!athpX+ozt&7M&oz{NzT
z6|94_w~DW%MqPX(Hx#8F>6s1fb-N^JPgyeh#(|e#r<=~6)uxGj(7Qvl+~>fu@7EAI
zW6!R(<h-r$QbnqVUyf6gOM(QV7z=qMh|S0(Y5VQs`epfVV12~<0!i_ZQ=>EZYH6aM
zliOD_GvZaiWVULfJvY0~&Fa`ksHCi}qp1lE+4HLF;(XF^;)nwiQRe1Zdwv`E)ucZm
zy&dI1i15mb-X6CX6ikW@>ACO%cVKEoXOeEfSyF=xfffvrOM|dUmO;#Z-s<s8$b*XM
z(Gi}LaNI7*Cgyj@Pk4|k=ZP)>mPg;v%+yvTkg%oMD0!@Kgq(JYg_eZRiN1h>$w)3s
zB42*PvWoP34p3c2gNU%ww@VwEqFSQCLGJ+laNzCRWo>T2PNnIioE_T2&WP*7mM0F}
zfMyEY6s1_{#Xc{R(zH*(J-fX9e#6G~*RNSUu-buWg=PD+E85`Jjgn==DW;}s*@`8$
zqQ0ytZ4n_lf)tf`J6>zl$0>XHCHBI6ymD&Lc}zM4&!K>TOy?yP>x=TGD=uH^Jm$VZ
zyhGsTN=N&lyWf2gy@Gs%*ur8e{saCd{z^T(kZp^1^XKRYLQ&DP7Z<wZf@lIB@ditC
zy1IVUUQ&Zfm$l@E58$#OQYWW$>T|lsw4=3TKiqv;^1(h%!G|4~w(`nIp&$pPAIY(;
zT6wMWSkzD~TE=t9>Ye^*582B4UfPt@gGU`$Q61%@PY+@paA)nxu5~#G#p&(9;%!Y8
zuyQq%-C$KWF(^Cmwj+-mprr8%%3zNlMBTc&!3blLmY2T{Oo$3rV8Pg816?l<%QlFd
z`hWxLsGV*=>~FEFx~|A$I=mTq8INkGCMngx>dh<d>iS%ss9MVd-e4LinG_H*cG@+q
z+2r*@k=DVWqMA^6Pg~Qz{`h?tZgWaWKj5b2@Fx?(ORHyKZ}_=Nlbr$)HNt=06=E-|
zUlwN$ULSQ}xe1mpS!&m|4NLl%kv`!j0CGM<T#UK1wfRLdz2iUG`jRIOf6v;w*4t}l
zwrp`q`aWj1(J5PB*uo@8Gsc#cQ*MMxGf6w0`WzZvlZ+)kbVGg)<P%+{lmsA5oy{c?
zT;@9@db5`|jNTY7nEV1r{6eM(sbUF_r{#T;<XDP5wlNu@XT89As)^kfu1?42_S&X~
ztHH`k*irb0Y>w3A-XXPpo&%ff_Xh|#kak+WVZekKk7L+Ffqz}MOeaLLVA@)iAZqm9
zM#bq(+4c3Sogmj4rxZINHH{)rZEkMpW<x3ljS7w_c1Rq8-s+{cYEe&#fVvTj>*HIG
z;bP=IiA~CZX=}VN44_BtmATqnYBB6QC$4Oo9J3#7vlUA(KIjn#mM^TmW|h6FI@dOM
zr^8_4mCK*{+SuY6a<8P%_^Z2NN%GQOQ8pWSMCsT&;nWMKb&T0-sv8U|kcmKK2Nu+I
zv2?6qV~PaLB+;d#hGgs4<_QEHz|e{|lyWZX;PfmqC<cz#wKIwoOY*X&tfiXYaVsa5
zy}p}5`G)RAQa$DkhN7B%xDQU{tPQsFk~BqVt*5r5w|0Pmap|WZmh_YjwS#7<^O$TN
z7jRFB1sZNmBR<{elwtxa$O%1{GOy~Mt&#i&$FTEQv)n?b&=hFrTd#ecy@}<f?*OVd
z;hxWNbi%e24Bj?OWH{+Nu)@a-qS3;}^4ZpUJB5xEmVqTsT#eOrTPwU2;!*`JW5S1N
zw9WFCsnHQqn^2C<z^8%Q5(D<z8@jQSrA{&tQp5qLCRdG{Ug2Zlh;M8y_0iQ+$eH4p
zjw_)%kD#%Jq)8E={G$K+;laan<E>uGZTX_>l<S<5IfJ%$-8>lX3&hDgu<Fd{;8y5v
z&&Kb5_mhP(Z=ZkL7yswf1$~lRZ7O`|0ef6}PC3m^bhBM;!-9ottamjawOC=I?~6H`
ztl!8>RmWg#pg3NrNxd5vN#6(q|HigrL5o8{<iKLKM4#nRBU+tI-&FNAWgpE6v43$c
zXQXwj7bh$_sas3MgG!&KoICxt4hZgY@YP?`T-aEsDcY+G<M%4zgiUv3K=~~9fwWSL
z$U9ej#xAEmv=Pa&Ei%=s&!C1#!a<v<HYJ*JD8|VWWX7q<xj@@Jg=fP}ZQMGGTE9WO
z!%j_|I!bbyMm=PrFwo&)VmEhKpa7f{Er8W7IR!Q$dS7@eJwjP#x7639dY9)~Bq*g#
zof)j8Q=k1Jrc>8n%(huW8=3xYEiizZB35HJ^(K_$>DDBT+<8ClJV&n=v$PFMgjsLu
z=3KaoL&MyD2PWz-pCKIV?alQ{WCdXmQkMdY&x-lP?_9KmA9q2z_LgokIAB@cQ}9Q?
zl4h<P#;5GZ8s$kBs7VJ9tFJXuwjgDt`0?DpIK6Ji`LZodXWvW5zT@T{5_Gv;WBD`n
zP%(_oTisMh9eq;#8!Uv>LPS{DTeE)~<hnfKl*%7Oadg3k<OYZZREj(up10LE;Q(P{
z<>xNau@A-qA`BUzUJ?TeZo55sjE&bjPchQ=QP~0SXp?ip==iu(O4jt6-r!BMc4xH`
z{$+e$_CJ0MG6k9PR^OsS(Z(>QF2kPP)n+ba4D}B!3`fNgr>5pa84$Fwbbe`j1zTlz
zcP)2P5D2R14>3-=occVSHX%IrkR9)7V}mY;Fe}ce!MOB1)Y;_1w71oJS^#Asz-z%9
ztS(Rqzo56*dtvYcD@eKF4JMPj%p(X+_7mMeLO;(biuljB%Y5S0$4=CpxT5PPMNLG>
z?I*JZoaA~WE5W5=bj*R}?&DG?ncRM=>n-YaCWs`Y4lK{K<(Dq8d%G8OQzyh0zn})o
z#N?W_{q~M-KE<pF2cPJ?N_{ST4-3+y-Pb;gbSI|>k|sDXwL3I3Jv(Oa?7l~)E2bhY
zN)BUwq=$6s3)SRRlf>|+Ybrz_HA#6NsZ^P*rZmgq&}x!?CL4t$oE>ut+Q?&EM={c~
zpKUS{tkHOnEL^h=D6yhHtLz+O_H)%uIo^nk%S)sl<7AL)_+w6eiNc3CmDMqA%FNKT
zPwECZc3t2DWwf8KZ|y^=+0u0FgEQA=Mos6`=Q$tLw1n%>qWwa*r3tI&`v@wur95(b
zVA2$Wr<>$>D?2|jO5V!tgZ8f6xT~*rp0t}s2Z!tz+d@sw^!u|;vF16YE__(*mpWMN
z#4t@Qz#fR5hV!%iayL1rLCsYANp*c&Alf0PUdZ_PJ6Wy!a3%khdS4<)vH@~ZIIKO3
zVsMXBF9caNF+n{1ueQ;^a*ks>@&aIqO6M;7wOq|BN2j;LB~g+}c=Idm*V_o;h<hVW
zF&P56v)AOsSztTbMi?OUfkF_JMD4WAwcK~CyWO8t_?qXx%6TjbtD{!Vd*x-vLPV~l
z5axE~4^~`yiSrnftg}JRcR1VHt`3=XU1XDc6zfKfw$jr=hS_Div-oW8XHb|CMJ?rU
zZ`$^>k%j-P)SRG_bm>k_G#E-)8l+dYw}UB&LPFIsZ?GT?pwuY{+`fi}u$%G(6f+_v
z-9e`&rYLqJHDW3Qb>sYa(Drvj%54xC@1!SpYSif~mS18AsvGb~95(v9%Yh~4Zkr%;
z2l<r`=1cQCrg~d+PpR%f0Hy_g%?s>-?h^`*kul?#11nt91rq1&!Cc#I6VmKaQ1h2B
zUuq8}($Fw0gHuX!IfTz3<nrOHl>%`R_sm7}izY3Ivhek=9jY2_QkcMq<J~7KrMDU}
z1I?BYLa!a}z6SYb_A+c<m$pz-=)rw5&|r*FcBEiUbirlJ8!U2b6n#6|rj9g<igR}W
zg~yoRKzA)d;jxZ-JJz_a7GL8)VuMKH5m}Y(Id*{^uTG$&L&%i@<{@veux|rXkjshW
z+m>IpWQp^fsK@E1O%7vYk5r%XE<fyVkcdSg;!b^0(FKuNJ!rlaaRjls2-~!i3lk%j
zg;w@?gTbrE7AJMAJ=)Df$K27|qY;&+@K5U1PYiE31@z^|&aLqEo7Bw2QpZf_+{N+I
zsgJSAYIqEjt^)W@w^L-4*+<ic6A7SL0)k>fqn{B3ie(Q%ZD%^{k(M7<0RBeK45vnj
zx^;5K&Nft>ls_M(XgN8r2o|(+?F~%9V@@gFh*^q#wcj3Ved}WhOYJ`(9f9byh41IP
zA1UBRSMKfd21B(4RaY3wF8huA6^xP+!cvN+<P{>?&1q}D*=S<yk`VD2lT>6GK@lvA
z^H!rEq4JTJRHB`x6!*_Yx(9TrjZ>7W<nh~4gg)M?6nfovirI%qJgp<~n9V-{ru!=&
zb70!xGh<jHWFojw@919YbC6WRZ@j@4i#fm5C<A#cup&C;Y7h4v`h1S1e|N#;{a=}y
z99XV>Y`kI!IHWnMIrfj(Z|4dVO6NS*C#6C{`-KC<@8l@^lf+x&0Y@=jfdPfqY8xE4
z-)(dx1W>6iB5(vMOHB$?WRx7=s`dT74$}x-q6hf<vdX$Rn;PK(%_^zd?^lUzpL0i?
zQl6Ny!GWKO%SaW;AG8*RX7JD4#swchqKA73_Yq%skS+R$`L+2SkH3RnQK~m2a)ukE
z{ZX4;E}PZOZIt6g4J_d+(r_VwivTQgyZYnmrQ(s#I;HNCgPVHequ94_WBHSOXBd}3
zh#{qHyRZ_=t^H}inmSZo48X*BpxU~*Z-3S;VV8oMZpcwDD1uWTAEgAn42N&~^KPdp
zf)owegaazF`dGwJ=QR@wsr!pI`VA9Y6W({qdHmsBFoOS;{bk}kinD3wIl1+b(Od1W
z5<S2_td!jMUOh?f0os5I?O%7z#;tz21*H>mYQ$II?n4Y8cfPkDhYdayAr7np^pxEj
zvWR`w{<h|0u_UhT+MIf+kNMESg<kps7Cd`Tn?sgQx@yhjA~zhTJ{>B^+%(-A-Zm~@
zxmkN}4z@Rb>uAp@XzLP2f>#Zig!kpF%<p*iMb5LvGTbBj7(9P}bs=d<NVpEHEKiIh
z3`F*K^_q@yWfp)Rb6~Y$rJ|LdUDSSojr%}btQ&M!a?dH}XGu%1ltWyi{=r<oVQz;A
zUAhAh-FnMg-tuhwP_?NqD^z?ZTIvQPxerPP5zABd;aWH8l@w~0>F9B^acW|#5)a5t
z2p?(4f)7Ysa|s}(VgkE3v-Ht!Dnez<`X6;*`oiG$N&M^VV-2HLPmDQ*SUgcGIRU_r
z*vG3YnVdBFyrkdM)P7P<4eiZG?GrVDQUOXkE#==LC8JZX{*4m+Mi1?i_3>n<(nV0%
zwBM;o<ipU6S~bp6`&6|kdVRL&HL#>}90aT^=H*Y<N8@4$wyA=)f@yGS)cX96>sh`*
z`%KqaCF7MGE`@^(m?8&{IrXt=D_tL^EqNeHef!z=p}{dkMX%yjJ%VE496pz1eFK67
zk6QfEY53UBS97Iw6EB<`I(;O^pHr{-RIHTzd5DafvM*E@<Ov4CC>{<XL!L3GCT1U1
zK!^-n`DlM%u>G7aq@H8sM;(}Y@CNOSe`x3nZ6R5-Q&!z2D=uAT|5(4>$N7W-2E;q!
z3o!L{FEIya?Vq|y%T0fhEt1TUrslL$AN^7~B78+JC@JbcH#EJuHx@o30OF<lIIiwP
ze$}-_|56_gEsBU*NY8b%As9iu*r7gX|F6mY-?Y_vO6e=49vO^7KM>?dx`Mz|`?s3K
z<j|#FWw|-)c6zH>9H|gZ+rQ^rO@xI5dp>2&)k@r&?LX>!fvVBf)H)q<YGPo_1v3^-
z3-+J&cC-6r9uOfpH9>A<ZI}I1S&M01W<9CR?Z4VD;k%R<Bf>r90_)Uh6AxfwgHrx?
zTi0!)aU(nBKwS18a}ZJ>-&L;4|H)Cn{ET@$vm4gi|JFPP<%cjS%}k!ZX+jX+cFK7x
zMz<fCUALg%ePnh|EXzuQae0hp^M^U5BELuMOYX^mvadoRuwe9HxN`@c`uK3@hUhA&
zWnWrflej6ZMyKm?YJ>@w;lS`^^+Vk-DYG9hIk4)+De<h#tyIEffU+;YrzAYGD$I~p
zhjL+$B7Rd8i8u9r<Ur>-HT#P8J;YPxu-96t+xTR*E+P;p+)pZV(ZOBzmD$B9jBGsW
z%pTr0L$0c?>Lx<DJDT%hb8R`})a&J!%h$-hI=d@}LHHS*F{?Q-k8HY7Y$ZMVnl>=*
z#uTeVS;+eMyKvw5+EzzA%P?0HQ?BU&G3WyFPi>R0tIkC%qt5ONrJBupZ}oY0r}UxC
zqSIX2@xDG=x(W3?PE|dptX_;t{r(O2<hi92?GdMxe9W??E9@KF;zvH|pg#^Q&oSNp
z(zkDFJ!9G{6)O|1q${+{LbQ67@TEr1#3@f=KrF{$`{rz&d?g$^PARZWecuI9C5*tH
zRI`%qzxkcal-X+XAd#pp&u`ra`7QZXE@CF)%qbNPOAQc~6Z_Wss^&*L#?By?=xd#t
zcsu;jf>7+TCl@S||FAMj*4<zljhv*9;z#gpi&~P6<~W=SFijY0yl~|D_L{Jy0GZjQ
zEtSHfs4uOR^h9~_R&HA;0BS{W8EWwz`B5|=iX?Z6$sNH*NJcq>2LT`L$klh=Q>^PC
zEbkCbSpf17=eV6=B^63y-<4Rgf!hV0=S37lH)`sNC2X<w-Gwjc6y^>&rI;;oox1T!
zcUpW;ThkjRU|qUWf7{H6eQ%pdjx1T{q*nr#ZGqRfv%xccn{?;jcTaI~-1!6#4#qj4
zn4b*A7piGVWwg@q)HclbCmIh<j>6(oi0?}aD}e<<z+ckI+Ye+<uUfxx?aIwgDS1XN
zXd_;0kdT2Nyr-!=@bqT0i3O2PF7KOCyoCG&r8s`rd3Vv2rN&Jz5PR(@HRxPt5N7_Q
z1B<uQCxP<_H;vGh+f&;V1p!F6NMN}wUk0cDw0a@0I1%3OabUqaPR4b%7u{2=7s3DX
zS;7l55oTR)RL7=0y}dBfn;awld&Gg%;9>u!X2S8ZVNG0pK=}7Fl6GFc;tG4_69y~O
zoNTfC=t-+#u*hOT1^7#<H+xq7M(P#uYw{P-lsx3rh#svT+&*s4&PFa>cDeJIqrMjG
zytrZ2wd*&oc1kgQd^tQ3b<s?C1lV)h#42+I*d)!pcKLp9H5s#m=1ST!BInN8bK7)~
zGhl31Xa+=0EALR>o|oObd2npRdDcX;=A7X6{MMNKkZDX@j#%bXPL19pTBo;d%3e@Q
z7TQPf^)L~=z!u}nf?n8`)OJq17U7eVa;J3jR<BS%wMPys4IW-U&H0zK-^X<aky?k}
zchM0}{o<Cf&_rHA6dOz1?&2)fR*^YM1Wszj&d1z&D@i0Lsuq%5leVnGpDkPPauHPm
z)MU1>K@;c0%Xtr-?l!wLr}Z-FBYBP+k|I?xZ^?oqp8}<WQEucs(^5JUo=NtSSaVsM
z*h(Fvbp|GiyAjmJB4Nw3*uvf^#^9wV6+8VfXHnpXU7n}p<M(osy#vZkS*q-pt+6X|
zF%ij?T8UDTAT3PbHh)tAuhhLk+HFPkEW+=oKa0Y@!S*XgHqM7L_QHk9#D_sl%wF$c
zb+(*=<$O`Sr#X;UuH4OmrDA-QGKTc+hZ-K=fHMyXvic50JL|dpvlr*iTD5MS^H^*M
znJ^osPWCVjQ;J1nKw7Q^SB7FmemH-L@qf&DE-rZ?{gt>hrInFm&yQ4>^N^L|<3VpQ
znO8V4_&yGt^&J^u6{)NU8+rB<KrUB0Z}lxA<)C!5kts4I;VW~6O9kN+#f74rvsLxw
zMGmPs4!#eo>(m!AEZkU<F7d1DmqIojrWt|M`E)t;A`pYy$suh6^)erY8tK#F)4~8^
zwVZmrymCyj?pJjud?(O?%#oHNw4`}KHgCn;O0(PMq;AMkf?&J4qeI7W9;p)X9)z0P
zH4UXAzl4tF#~fJFqbsqY+M0$qTwFc{TS~b*#$_!+GPjpj3!*aToKn1zF@cc)#9mgj
zmsg>L5rF}7ZJ<+gF#~2Ql-Frr2Jk@_#h<q%FKl7PCjA5%E$rI-e!FcmB8!2k&jagL
zUU}_+t<7cKggTSQR2OjL9Wh+%y55uRkQTVq+=1oR08<GNu=O>cY#f+pZ6B^3FH1JJ
zZ#wlkK71_pYP&+9HZ&Z#phSG>L<2rZCBZ08JN2pIfU}>9#G5Z~RFF)&P#;F&fV9=9
zpSF$l!N$;aH%meb;6R=F=tL^WWU4)GoAURAoYT&;1~ocWw$;tG0HTiK^%X`M))lAb
zA}v6=Y{YhCqQ$thw&m#@9#wW-TSiXt-7J>O@)2e3)YJx#pf@H>%XI+4AUd_nuFnP9
zf*KrhKq;qVNRdx+YrG-x9#{Bn(@sI1Cy_^Afg$#a)-3zI+>*3i=5{(YTHeTSgtGXR
z`4!Sb^(UM{>{t<8N#+rUAuiB<G|BbFPxX3QKVYw_w{y@i>2Xq?Lrc?YBD#aA3pE`2
zhm-c|*4%ZQuW|}SAR?z-#Cm4Nwb#^Iakm>sWg;E0dRzAXeelbo1}6v~d~M^U)=y7Q
zY{zj5fDV_HH`wdi&Kw~(mQxB8NbUN5q6r_c*B4Bj1qLnj?hiY#ID|wk<!vO!Upzp|
z&l!xD4b}pRPNtO@H;-`Mje4g(T5IX7HIGAG53mw2Fpggbmefd8vt~=&3BpD=xeT}O
zaAeEzMgCa8>Nf|0={q5muxjZTJkoaPR_^RYh^8ze`Lgkeo5!u!eeh8_A%)>F2Ntu-
zwynXCR9Sw|wj~jUkoYjJQq-65ykEK2wzla->(_HiF<>1$I{jgqY+D;J_?`LW_j?Dc
ziRE5m8?8N6M<X@5ClGH;`X~R96wpE$<h&d0NJ%3W3SSd@LoHb%n_w-CwbTtV%RB^P
zwzm>xpLqOGV*f#C?Z!5$#~++O@03Ji-uya}B93JZ@15Xll)CLA<9H{T+xGm;mL66~
zo=c}YGZY2|CFL3Kh9JF4+z{vuCLN-DNpOm{iS{1UTtYj7odXl)o*?9%oj%!CdaWXY
z2Ay&?$F&bl8*gvy_Kvje5_@zINu@?jQhNyM@HU&O_L5IvNU(ejCJ{hadnngQzdo2d
zN&K@y_|Y~wHQIXvGd7)*aLwS@7U#L9N*)u7&MBM84%{UFZCqR^)-I38D6G{A4U+cx
zrmjf|0^lk1QpfKBrzYsJ)Q7V4<K)QS#diq__oV6GZ#UOhGz=x2&^0rWxzVXf6BJh<
z6W%myx8#{PG%?~lYL@zNf+YO1*<69TagLG@gb3JayS3&oDW8r8E|$-?=XVT@z(2gH
zt=HAT$vfph=5#QC)Zr%e<~G!j6jDafoIU8&<l9hFrOuQf|CZKqiKhsh4j_-`D|^H%
zjoFX2vjaDhnFzB1qArsktY<%--DAesHBKpxhe@uZAmYSZPhzWTQp@wr7it768V|iZ
zVQ=klNO?gsr*nhtnO7zejQ9Aj-<JJe1)G;H;{c*p*gov`YFJ38eJPqRGy6%WCZ-^a
z8%%*BmhWg&e94NX%bZf-knRFiprO1Og+q7NZ>8v$x(5p5z(mlL3uCL@)%{ebfh@J6
zt5UX6lanfAI8y%I^(HIEh>{hE!^^3O4vM@;8KNN2d)i@-fiMMg&Jtb3Mm<>K)CUST
zggge?Yk;9-)C54#GC*X%y)E%_<?2<=V|EHV=@53Ux3^<~9j@$of!$%-*{A$|PJPY{
z#6>dwemcOS!jAn!!x1j?#@z@t7LZbw#bp#sf}#_2$#4V)^K1NS>?hlCFf=jb6yj8p
zdR1Z+6pX*hwj_!Eu>DjuzZdM)MwP>$#!r&Wd*1yF7U6mSzy8_1iKL3)AlHCXjAlVX
zup}E7LaGa-KDKwX1!29_?0Mu*V98Hy><!CoIBWMM4h@e^4bN<MO7YFbG2?*<j2lyt
zy|c|ccyoGe+9`|dJ@@JM(+w9lO$?7ZCE;|xH`}+L$%UI5yv2FUQ*_y-m)p;_wjdWG
zu%~5;3ie&db(>Qk2&FOUuu#^lpG(;M4xB{B`{?x%G=S-&gSSEl?C0B3quG>`<ni!s
zJ~?;4Q==}84-bvnFXS_tlo98-kR!Gn%3iQn`ZRl2bwF~VA{<GN;SC0o>8KTpZNJ#P
zcaWX0t-0TU#YFB8eTkyYWliX~Qv0Qxt}7=<uG@D?xp%@-F)lB+Uv4lkMZVL@(+Mn(
zt$|Rv{YqPnmE$95<(zVThTKhb<OCLhU(KIjS3s<}(J7`*4(l9OV{6KOExR(*!^pkG
zDb+V*x)Sqz(tf>75eK6Z@g(WP5yq*}q`ruMwxfQjY`Ib%p~H4yAvcp*>;9^4r{FoP
zH^5?+q<9ezC^;^<&oCa@&aN|&L{^G52NvrUGAO@kd2SBduC9jL2^oG~8=*KyYGP0%
z#SMjs{F-;S6=Dq;6nj&C@CQ?^gkptyb$#RdO}3}&26GB!MUC#jaw$f7TgOIiZ?z2@
z8Mg57)CZ;aIyKSuJ}^N}b|g)<ecjufWWc1Fij?WzYJ#Q)(nnP9{7&27zQgo}?<4aE
zEaoB^yR;bN=nc&|&}J5VmR(GQ5gs`<n$$ehCz%7gWa7~L=4qBpWS+4Ha}HKck2#Np
z7yCI~prP={4-^db0%kX596sW}k~U9Hk&<@A9&BCgxMPXB@^+=BzL=RhkZkOs>h>bl
zWSEteqvHU-p}r<l*^x7e!^72r5JE~vU1@Lt1q0W|*VmGQiGIr!>a4=(dH0+h%HeI;
zw90u5A|i{L9qzs;QZCA>G&M)&A9ZSSbZi#*Ogqx(X~+p^xE)BKbOk0vJ`Pw!IReQC
z-;U-N&<5^({NA|-opOvwUfoJicId+GcC5`Alv|mcm82kL@K&SU6$;!C(d34Ays<+g
z82;GS4nX+3eCbj<@r3d03#ZkLbD$<#@8*tKdE}mQuZpX+@E#b~tt3Fjej{aThE6sz
zvZ1*zIH2Uf^$Kc=@T+|Eo_A+<&XwGONg_AMb~s7`uT6Z2nTb)y3AM*gm70eGvJ*UI
z5|JqbTaZu{#vn;WXhof0sAuy#Ubb%i4R*RAeh9Y73nXF(TjSK{3J#1*^J-_>3c^d^
zu7J*=AaKLCaKR%N0@&GZjq-Fwzw5wa1bq8~l$y$3pa7HIcCPj31`PC0Dc;)yTOe!Y
zBf-$1wz9|CVYu{)rA{FhBAE<j33|xR7fddw${eBaU10x*qBwW|B4-x+jW$SSPG7m!
zDKDZEdk`XYHj@3(ezSFU^$eN<N`;K_2@=XbU~{c!!4>WW@g8$vNw;CLS~e%q4Y|0J
zQOO=}^|2XgDGYhIIkJrrYn^(vBQNj%D8~*Wkj>|Ga77AdI)&&+Q(`LK=t8%%WEe>d
zSQIkk)F*GVnrH{U(zoC0CM!)TI=9j;D{e@8>Ktl1-m1aKO;KtM`3Gf;k9fUzcPsd4
zky4$lvg=Z#({KIt0~_tPyWUBIpZ!h|Cf}t#7EUa)Tol5rBgY%wMVrGH6%hYBISJPh
zKsV(S^m%TM8M5E4PAEKqt)P>zI?-xmLKegSNN|_RIQx6mrFP<0?2H6{2=FPVK6!zm
zmQ%=nzm2twkleRGUnStS8kmu=l;n;)2fg@%Bs6b%@w1%A!eUNe`looaKWsfm1FT?f
zskZX6W@<!pQT1VG_@johSV|SbPdPDwu8N%|-M{^DwzT5%OPoiEty{Ql4obs0^C#W>
zOP!QFiZtcQhNiWmKB>go-uR6DX={yp5vFGPPB~hG-s&Wi#l-%sV3GX+ykR)rVDDgF
z?&YAr7yrFKulJM#hYOIQQs=#3^E(I?B&`S<@xJ{<n_H$phP{WB3Q7#5Q&WervTWuJ
zDoMFACxz(BLHo=4eQx)nnH=noI5nEci4oMB_E$Mj8aTwKDaF8I#%T^OG5hO!L&@|W
za!UC|u;J2d?6JQo80y_X*3(#*fz`pLZ6uCCUCm6BNQ(BijsCA$KXA*i((K~i3sBVd
z>GjXvlPEoekAU;66%k>>u=C!wdspf9Z8W_OB#BmX4vQ%4`?{Gt6k?_PErjH|oEi;g
zeg|H_+#oV5D)9d7!*&wLj7sB8S)AK;SZ}P<_IKH(;X&dxd5$L2f(~iN!9;%#+6Quq
zaqN8Uo18*TYJ|?o<G?<cFRWR&df*jKA?8Y9t70AGQ4l)TurZys59P}P*AJ}Q>=eog
zcqWk9KAb|;Q(o*m<>3*woUm%}HC<Etk(xZUaLy@3hs1~>>R}i>Vjt~(8@JxU-<G`M
z)Z`>p2n0qw`&fNJflAV8;nv{5M1Sxn@7u?#UPoz~ppH7Q?D$m~sAHd~CdJE~XTVBX
z4Gm;JEYyV04m;&U@Ks0bliidngp*a2Q$#3`s8RFVdSlz|Q>`i8J)rD&0Md<^&gm{6
zt|%j0vwgZ@(J<E*lfk#=K=c&?|HkYyHHdJyqb*r>I~e;`yB9@;eYW09tDC%1?ss6Z
z6pO9Nsa^S8H)_eL!2_^JM^RZW)MOIAwC5t{+UMJ5SwBSb%SZw<>j1^EU2*wk_J#Up
z=Dt}Z68%54y?2;hS9SNl=$$}9NFxPf;sO}kvSgdCYRSlwkYo(lhFt01DQITyFn31M
zn3ybVAaTW{0jcVvW_7FBvYd>@aeD8)_e!tf_xY~1_n9%t?|I%o-dBcs&R+Z6bM{$#
z?X_3mPEYc&wWH%>_Fv_`+y_hI237IT`fShK(@bOy`c&aIlKEsu)~=T^4NA|G{e+#!
zxhpq^6;VA3Yr1_pk-L6itUG~$ePY~!<PsH&<x!->^7kx4&}Zh*3(PLq)RSQ-`7@8%
zXPZ50)|-$PAjv1&o&Kb3WC$S_)yqDYFiJuxC}4~1QVuLuoIa|Q8PUpx(T3n`pRYWR
z)%=jt62m-Cg``XD3zfERASFSPa|39TTZ?7UzSs=%u`mRv#<GGZYw2G4=b%9@RS81u
zL-wV7?WXP)PFZqVDSHx>-cOq5m&?axlzgxW2PV7|vCTQ$P%VeKex*JheNr)pnD&Ly
z=@U**PSPR{IT+{IwE_ET5&N&b(z#5De3-Jhd>9qiDA}3sBuGbw*pZD&>j5Syo!#s(
z|9_PKAB~;uYdK!(;}WFLE=fDtQOMwF>`j^1?#iB!$Qj%tm}$cb<+VrCt#jHV{84@i
zd&*sfeQ^GZ@)8_C5yWC6Wy=rY14%+Z-fvGW{nckrXb@<YYNc&-w>SNS;Y9r_Oy#La
zhFQs;);wQ6cTtA~1>scU#R;cBH2&_mJ-vKStZ>L~wulQov1`eeddi+rhlJQ$<2KTU
zC4<6tr$5M78mM$rsx#U%%d^3Rj1A%x{H2ne*PcCC!6f8K?2p~vo^@A|Z8tI#5wr%R
z$dg)nPjbdZP+$0r?Adn(p(CKKU*R-j=sUP!zda|1hDVFslPm?~z=VP8mfvR2Eh8_M
zXv_$wyuxI3k%gJW7JFWUDVcCGWk^wdk2*c^?b4s6tFm6~`2}C0e}i_}n8e)_K8?ZD
z>CfND<-;3@?X(xvk(A`&#}I3PiL2j!;U$;Z3o9K_d@a1f0D_KmP`gTipggdp><&vH
z4mSISyUMDOO3L96grNb-t*j5j=1KO$7nPUFzbgZ<oQ1&jQi>uSu@~1L;CNLOrUNK5
zon-YOshr<fU>NVPIb_@hu;c;jNZZ7A{Y`ad7q&MK_LS}lq^vSgT`Es$PJc=1K?%TE
z#adH8($O)mJ()wI_bIfHWn<r*16@PGMe3--DFMV+d>A{_OOre<z4X?LU*R<33&eQk
z65cU=+P<Z{v9S*@Q!E?;Xy}Q%r5z|>jNDQC)`EE6i*ul8QnNaa(_emb0E)m=_V}Il
zZ3&y%h9yBqd&?UJEH$}hL=(5cuzh>Q+Vx#xK3h&Za3noWF`vSOc+V00jwIm)S6ty-
z$~e7jxxFkYU&Z8MceFBptks+3{10Y9r;&YU8Jt8AElAntftA-_y5WNm7<d_$Jo~On
z$YG7kr96kDIQ{vKV5T31V}BIE;=6MkGuxn{V@@m3A>^uF)c)y%_B}Zxtjd)OokpPq
zC9q2>k{RG3>|`%5KTYb0b>wqkkITQoK`(YePBzx!A?Lb0P3#r^g$x#|p}nG+x3FbI
zJY7cBW$*<31r4*A#UXLng*n`n-7dlAG?KJ+ND|xTD--{(f30(sua0E8ssqBf?0cKF
zI;$3^ax6UW@!5W5_LN|6<_1|Hg7q${q=H3!km}1V2NGOv|Mc<4@34#OI_2yGRd;j%
z>Za;Dd$L_p(S0TRC5@PmeV7{Ma8Q-rOA~>Eo4T8w>v-~B^x9MQs%$On_2lrNRZyfV
z2<YN$fVnlkx{hFZaHpOFh}Ymw2&&aQz2>eb)-x)ddG>$%Kj+ychNc@Pu710$BAtC2
zyJX~e*MqD$eZ}d|iIgo7Yu{_@6WFOa(W=vSAZh)6D`85Lc6oVVRHXrP+PSM>$^Po}
z$0sYsNv7OA(^Gau;x8(2_MBFski!v?t`beGC-llIb1+OF41!nFPVTPL-gOUaD_5Q<
z;;_w22C$^n_Hx;#RN_^2qBpcQIgRo%<Y5MS4tXtY$@6ni1i|xNPAl=SVTe7Fu+Q2a
z>-1A+)M+;hSk^z>J9c&1movmwL{t)lS&>b0t8-UDh$`5k*ug!oDKPR%`LNiwyuc*t
zc7{iXy0WNWTXE0%kfch}464)?al~GiU*BrK*SU<ZEINb&g!Qdhv|e8~z%`K^PK{Fs
zROnY*6~XxkAW{3@-cV7p81dLkiw99f!rs{YYPQIwM45w5PZHfms@=u$9dD|`?%??D
zI?Z^YWC=Ok;uM_7w0LvHAj@3rRt1t`mkXv`7QS<rM3u|0oo{a`k4nQxEyY9fk<(KQ
zEAnfnNhR+RFE44VwlKet9Ct-y(~7AP7mzs+Kb0+NCRzj|8{9Dm79YSih9QUmHq5&z
ztSwG5AzM8S8MNbxQ{1C6Tx?dhq`*YE#1>#wJLeT9MxRx}58Bd-;mAag)uq(8aeCsZ
zaLX`%pnTf03d2^6d6O8!<6dC|4Mhm)BZySjl{LiMWxNK_GD9+Y{;C!}{UHYqP1y4C
zNZlm6rgY&7)8NxH^02Ka1I_tLnJ=Ukr$-OFX8x6SePWt=+3Q!Ya$1Gv!lK0fR}XW0
z>s_I_Xi3SCAbz~n0gC6K>C3caE1O@3R)okbF<6dcdZgW9SVnTVp^RT6ipUpG0!&MW
z454nXWvi+r&4z;n%j81(q0pwB8)i)BuH<+G7*P>$b<Q!h&m~p0opx}gRDeJXb~9-$
zX)BqsHMx(hZH;Vlu8W>CYdud9hM7ksdRsG#$I@{R%7K;N$X6Ivb|xbam&dwpYx9%F
z!VB)S%Hk4tEF%2lcOdbtE6<V4kw;PK?@4UJg6wnp>o1{<6z8>~Puu!3D(PB^$)aM6
zSD0uA%7$y4vm473;@ktMUBZ9V{c(B<=gx2n)r?LznO4SWdwZRTf_$;T5>W&tE5w0P
zfzDkqLaYReK*?hP`Ta4wsq~Zpxs)}O73sA{1S=akwHM1;eMk8!xx=Ks?3~``z{*^z
zNP0L4z~K%y>&Q8@cUF=jE<i<@u!K4=5&3+)ZSQJ+9^qloQ`DK1KBFjZr$5;NFB3&7
zuHxppvrvy}7$JssSsL<O)Cq*d?Uu?|+d8&pOKZ}B#Jvhrkg9T~(y!fG@7>tN?MB*(
z1BrLb5w4VNB}%@#u7-6TLMCa&9T9X52`fAEJ-MKwxP;S?{;+J}ik0?#<>ozb4UA{-
zs9g>$hJ)!EzLdu72w}2TekA2*NZx17(AGMg*|v;=s2lRJU1dXXove*V$YU$N>BfX9
z@&hX_*1qwM7Z?-zpo~836x0x{-@H_XFghqaKr}VCqbhdl*utKgr;y<)KticKxml(G
z6FCN6*;l4hv?GC8>En`c%Hg3-f9x^;GhQ|~eA#_ZSbvoQ+Z`+~xY_`uai&dd>QI-H
zB?!eBsCTXB^hdUQ2Ot&Zb+_7JnS&0e+c}M7MjRj5nY8!j3^Mmf5>6}8dKLA-Hx*QB
zg(^@R+V={J{Q>0Rc($SX=HZEPLa9d_Ku#lzMKTgKhl3i9*l^C&c>*S!>sTs6i_CMX
zPZlM%`1>fVY-CQq=veWox)+m+;-0<sV=alDWS`zuDj-Nb=$27PPl*D6@q9wdoSqB8
z4A0D@Min7XhL~JCtuV16*l6Nsko_3-k5*X&`(Yr2;A1R*Uqf+O7l0}A3QOiDd(TlD
zEB9pu5J`L|y~1FY;f*K0z<SMhKNdm+$FBq{Q&<Y$<@85xl6F-E3n~RWWgyu>6knmQ
zie<yBVow2UD$p=qS)?rFnvakaWdX{P4YIEtaPE@a8pZ0B%rH^t0}}|+J*~8L^mCr!
zhRKd?PQVDY1j7#G08)6{OeifzlAdiT@0BMWO9qTy3D`}9F_0aI4cwJaZch-N#n}W~
z8+2)ilFxVfEG(<?i~UVjl3lxWz0-*K5uM1&bcU_rwt~-Gp2a4@rjTb!6KvX0tE7~u
zs3SUq0^eVLIoD{&yoKgP>1C(C@H?ukF+@4VzrQ>UXFMW_k~EZwh!e-@NsVB*+nylo
z=k|&z!X)C%2nSN07ism_$PU^E%Ja0vDkxA<B~ka=>503UX=qUl=P~<%`i?ae-6zwY
zc44TBeAp9l1fsuv<~grJxVt$skyhN7VkV+QtQABA`(VL=tTaTH>vUm#X<V~Hoc`n(
zH*{~aJDSGIb>1}Aa=y@=Ww44-kX%C$kOR|CTO0>GZXYTm2+Z>aMYG4*I!;e~XG&KQ
zd2AnUzFl>ISqJ!ufHnLC(|uSxBz{PESNXM%l;5O^Db`JhY6DXSobY<Uez1XUXA4hD
zG_WFm=$xzM4>C3AJ;WB-59M!gw|9po@K6%@;lPs7QgQ`F_^$HUbhl_=XB=3fozGQ}
zxyr7E%wTudiGz%^(Vvst6jcl8Ns;&}yC>(6BUDGIWk^daXcGww1QYdl?B0T3ofP0S
z0*f^Un7On28jMM04$KDL*xivl_Bj2CdG<)flpjR-xW8foGlf)~$S}juxg9QoDDLzZ
zwP`juY);iSK1ICwqZKzjsx)qLn(2#B&YaZcF?*oAr|>t#Rf&S)ctIOZe~gV2v5D-_
z-~INnDo8DGE(>PGJ+S%c=di~!_QM4$l=kX(8gZL=y{wMfUF6}LvL9)_GdX1D>ch#Q
z5*^k{fBZ5x$7+d}kL*Wtk5jUzbDcCmZ7tl%KUM~v(RNWO)q&-MGRP-HzKGznAJ36S
zIR>W@-@AJn=?YvE5NIE7K2qU|d)bXx3$aJgqq}FeQ9ox`9&v*6KT*C&7GB{};fiy$
z6#8Qz!8J9rAwt7j?30Pt%NHzgE^`=6MKMm;Pt+eGA&87n<M=BlX6>meQEC(_i2IWj
z6PZ5Bc5uckOwmeV5@9j^RJm`8_o*<#K2$JZ1@#Kbo5G;2BDE4nf4Z{2<=7iq15P{9
zzY>Ld#D1m@K0=SeSnI$Fy{3A8CvN*tXCNA#@18rs&z3zmWP<1yIb|H35=?<dZiVvF
zatD#3aK`za)+8qsZdJ_Rq;1PNU8z#5PD4w%&9Rf)bL8ET2^~Z^=Cma#ldm#jJDQ2p
zt1?zA2Pe^Lztf*8DJAdjw4LSdxdK)3WX0X|i|fi)F>brc*H@)bRUC7&*zZI4*o7gL
zy8_{oFXT{@)kQ0RukB6*Qx2I(oQ~5B>R_WmoKar+G27Gp_+*4)KNO_OwbPS#BeW@J
z3|3-Ovv-HpJQk$1F|R#(y%Om|bC)67_BPLXSUMML{Xqv-a6;^Y?Q14}yUf?n=0UG8
z#nfwy=DaZ5U+IUeBMyId+M0zmD_1PG19iKmA|Uo(bPymZhR<~c9ISE-)%ZBSr|kfe
z1_>G&YTJW_{c!N*h|?&miIVFu-8-ju+o1{zK?4Eead%)TI^cxaKzhpII>#Z#FfDA#
zbM*@P>vJW!M=_8i`7JB(=Q-E0^Wuu{+0h)}Xlqicgwss#Lr+END8+$1He|=j8*2gg
zW9(EZ3dNFe`h!F&@nmYJO0FNT-@mFgPGk<vx=Rx*l3YfXAxc0g_rgvz{VHpk{TYHR
zV&CcXYt5(<)kFDY+&*1)45z7{`B%@glXpEaz~%o7llaX4Vq|yTm-E9ZUj$Owsl@fW
z`R826T<l|$l%3-!^mw}9UtV-CjhSupu32Db%H0xQB=strtim!rPaF7@JzVZn?jlKD
zxa)Q~uoR)!kq(5mVvi((CVDE2Hs&;QO<aevZfA=UG{?)0gfW#$GqaoYB<1O(b8lIx
zr^30cLu4K;GtRUy0tvt2-ec|2qg8`W(6e*pE{RrJ<K^t{>AhZiGTc!UJ;ikENTkP_
zw;c#urff?UCMAL@ZESys?C0t#V8fQJ1#06B2VvlG?ewQqF`#;Hs+49wUmmUaBKg!0
zAvXeRxLZ~{C^kHs&JUsZO3%yn`Tel8^Yzn}Ar*s;91~>x+M^vr`E!a3ie+Lm&6|Z{
z77a2)g~@QL<Y681{CK%fr;lq}W1sg5Q;8}`#Rq3j+Aq|P4}{3vj(m?y7P=LNO!_l=
zdU%2)M-G!jCKXLQkye3)?{cn^9LUidw_hyptr|4iiv!a{Uc*28rR<NQbM>fgU|MlQ
z9Bu#0g^6Hbo#2!gnyCenCK8o~>{l9`QSpdxRibqKb$TjCz+N>a&vW3kT-&dfM{|0q
zPAsOGT$hrr{n&z#t9tfp6&oP~Nt4a=;*yc;wI59w%_WI7_h}XyT<6#8(!pgVbC)*&
z&Z{J5qGP{N@b)(6O3afmoPlxs&GK1Qbd-@IY&TAiZev+rxxjv_jBjq?2Q8dZqsr-z
z1%@}4Ei5JN-!3pM(Bgn_U@7`U5O3BFimlLor`)Z9NGq9|;S~Ai35qZYBa{7ZfoT)e
z+mAxEz%-BD(XRbo<zx{bHSDy@%2!1}k@YeA{fewns&I~MoOA#=+i^Cdj8T73r7FHr
zwJJ3qbA>>S@33=MPUQxCNkg4JDiXKt4-2Xk?EEC1h=s)ADNcWkC_gk*TKl6Xebm0`
zoy3V(n08eK&nw{Mk1P5xFcJR1*MyUZ6iIq=C`7a{{}39R-k&61+MC<XWqu;=@KSZO
zE+0zGPx_zctrQluKdXa5ozquis&HUhYdV02^8x<68HP$dOOjGjj+~wJXeMqZar!Uv
zl6X1iGG{&bjnoEtP85#vm$^bOzx=BC&ULQmWka2<_E!y>YDJ>c4osp*l!$-u_4e1<
zY;DBH28W|LXA4jC2K%?aDLvv^ArV7iIbM2V?P7yt)k>$1K)cWk5#8)>D=tu7wMp>T
zffasx7~y0e86}EOcohBZ?@EtsC#3C9V`LO|5dP>1n^)Si#J81=-AE03vo<c6zbf>E
zV@jKX{e8uW+=$+^C^9x`|4_!pa)l@CMcUut6{c-Q9GNe+e{4pj<XkO*$g{_APfsq6
zx(O3B@K3qy5wG`;I1Nq78nWk5Y5%!$gx;2}(+DgG&-Lt%SRwW=Ij3aKY;vwI)QqJS
zKSCj8rI5BwOta!(6&n^O_u9YabW(eV(}pLVW=;)Towl_MmzoWs`?q`z#}M~>xu!->
z&;GsnT&cWC1~OwEbb8`*Mbc<^3gol@D0hpI)GWL-chQslZ~m2++ke)dGIs51VA6r5
z(9|1@5~lxE=V}Q7;7%)FhI<Bs(KP9i_NiuQ5)2eDplqz|(+9lvBu(tb8lWHo25O%!
z&ymGZLFcCsrz?z|S>fpP9JbFCbY=3K)+8z@tV1qS<Wi@<e7x#6>)q$<vvtJFRJDLs
z+#$VEqRvi6{JDlzlF=p-1Ry?wAEQg;u}(+-e9psV7hH0ga~+S!Y2BwlHv2+>$xkBx
zG2wj<EV)r{VniPDj(xEVzYu?vtPg`svR$WNTRE{oJ^NB}7Ahz^mvi=BejNO$38%F$
z*B2?aYIFek9f)`chTZW_t8HJ&=eCpx!F7y2iB+{fkJ?u&6)ak@&}kGtrs|%0n^rM0
zlKpqRtu@+p8f8!lHp)RM1W=+4hEe-kmFo!TCKDcGmjlwPvA<0c>OjsJmTS8^7T@`U
z(7_I;Ro*R`O^yp~)SRhhPbv3Nrj$L2R@t`aU29Km8dqQKG|J>q03H!^2OG-M?k*xG
z`LGhSaLxcq{b`sA)x^%G@#X3Hl9_Gi@dbL*8g6lb)idtS?<M&o^n0CNEkmz{siU~)
z*QPMoo3LlzUDu2}Ega-t#sr^&SLTv@Rvq**9DFpyfk4c_*q}YTyh%u}RitteC=%d=
z(;uHGcbF1Pa8cNE%5PGn%`=FjCKRJX=6Cw_21a~n`h-3A?$ANTJ6p%M47MqO51?@S
zjI&ejQu4Xy-5r!fzD1;@(=7Z-7UHZm$J~QzY5Wh_^P5ji0Y_1lb767|;a#>Dloux5
zWUbt=mN-30B-gh#x1gQD(Q$iW#eK=i#=aDrNfZnv5S9Kg4;8&Of|qe5AUY{kSsV}c
zZQsxkS9;2#h=4Y0rzho#fk`65H)hR&7bVJwK3Dh*t-z2BzZ6j5BP0PT3G2mo$D+ks
zw%559U3TI;f^V#-TcqqDr<v1+OrQ)Towxl>bx0BF6!D`hpGSETJ%Q-VnirZ>PA@eT
z-fb@_6DpJ2qfRTi^g1M>!4doBf{>gb8KX*^F}u(XxMgIjI(Ow{X6OmFMi~pGpeZzO
zr@gc;lE}qZV3?RA(7*(0BGV4{g};jv@3n7f#w@)`Qi+a;XbP4&{l$u>!ez;TAZ%%0
z;e(VEalpPc@wo)Yr*oa(MP#3Hz`m_6ly;Y79$Lu-*5kYQ_KFY$x|2>TkWGj+OiVK5
z8T*c}8(vbVau(AJNPqkW2V01GtTlUC^RaB@+?V>Z(n{!wj}dKYdm>Q|VMykXxv8Sx
zdG`}9jg`&rIWf`=9XG7KWG_uRz2%K3m1T*)<&=F_GxA4udKNP99S$sQ20G9TS>C?8
z+_V?_pX3}QJ_nZAxMB4Y`<_aBBuT;aX=?emlp@f!mls5&KT_T#dDIk=lpclJAIm0@
z!(aK`V#l=$>T-dXQevVP0oDNZ`rB`RHDd8A%J|YI6tU|LEa!5P<$$I8!ukyg1LfoS
z3I`Un#WfTwO<6$)q_f*A^Vhu&TZyQUc5be#h}2WYv3+lYI9Yt9n4EB6d6!DFg7>yv
z)DVFRU{;IjY(ZA~g(<?431F~`>(0@nJbMEo2bz)^El$zAB&VmpGrkT7D}w`|c~ya8
zxtb!H>C(EGR6RG-q`4DlPTIOvYwcCd7swwO>7HI;MKWUdqw;8OHRGL4&Sla$k{wxH
zNPxTSHT9$MNo>&o9Y`$nIKVK-Mddhhkt+A?vN|=&MuAn)PUgcBSKuT^1jqe~#*~x#
zwWVK#7w6y-!<A8aQJ>rCk7>;0igZVY@xv~!J;9?zR5jd>o@7Q5L>WE=-lh_*($sg`
z6|-LoxysCV+AAz7o&np<D!;P4p(?VC1Kn$n$V!nwCwPu_Eh0+u`=faM0|Y(L4y@{o
zS5kS!L+q+LzKB#(gtKb+$`C-kWt1QK<2S+<wI1L?kJ$Y38+~V*pmgr!DR3I2bNb6~
zvVJ%=b6ZegQr1+j0@tVmOZlK5f0bQb=a+K;A$>Nd0wL3N@U|#`XV=vA-$vQ(dDCPo
zq7?bH&HPF2)FP0piSq#XA}Rwp{gOk;o;ql+tMBGu;1Q=)7btU{A+R~KufO|=%-6#n
z&4yVlSq@l|C<(|zItAnnbwRbGX1fLtNNcpafgD4uII!|JR&pFbLEAOvv?W???%ZZ?
zYTjSP(&}oYCvX{d9Y&_pSl(Rj)^@E8r-`g{dgAU3aY#|;E#>Z%>-RE)7#|7K*yHqQ
z0Jq$Fvn?z)Jvl0coM5LX9*Gr2@g3}=M{H3pZ{kuA%1dJz=?LStIPad>HV{GM7|7ZV
zEV&itxN*#xJzG+z5u|biB+a_#uxn+QL4+3bx~3GrEp4bS<1H115w=+FQpO$qWe!z2
zwW{MJ{MfPv13S=0r9Mfr*KS9TFtF~%MRr}C!od*=3T=?J50L2PmfPR+ep{Z`Iz`C2
z&V4Hn(CXNVBqm7i!RbRzOSA!-Bs*-<t}o-B6AmLbuFVOJVyD-BafN#Rn#7AzpTOQ)
zr$+KP-<9SEg%Z(|v=PFOA{GzX%6fNDMY?5ElBY)&=_Tv{L*3C~KG1Hc6v@E|L!C*d
z9VjdUXVmKxD>-MY%Bxb9Pvp(QD9p&WA4E`I-7Kr&<^4`8m>-iNlULDbC5a)rP<m`l
z)32Q>O3X0*$Y@-1RHoCP+sM0bev7@Weyv>nybdF-14~q|?{30bxwaX<EW9yF{=&GN
zp0bQm;6WI&b#<&9VLj}$a#F`+TV*3(-#koC`1G1fs5s#Ch-wyGGtX{p@In*-qdqXb
z-+=`dW%E?9U1Yz=!e(!;?^wKi;hObJ*E)bicqvZ6^FfV@Y}`~(Becr6rHD>(v?wFa
z80gm+=3h16-cgZ&gO24!Gr8(!GAr$!Wq`@6;Q^QvfaRwm;l$A#@2WqMQtPzJtjVlW
zBA57Z>C31Uc5@l@PT0?`IN;{7u<9N3+N0T3z&Ixd#cz(-EzL8bEnt$AwPA_bV$JZ`
zep%>cqmdq>YbB2#uv;6}{jhVL!|(R@;H10l-DN<Pa~wowfaK7i=qYl@nkDw0d<Nl(
zZRa{W7WOvU_tn8cD)0wBQr2dbkeGV>4y11SCE~G<9^?N<tX1BZLcWB$8u~EXlP8DC
z>zs=5hC0)Hv?|rf?<@X2<@5w;YR$7Sh{1T!HkP{~UU*pqseu$3Y3WI3ifW%HxwrFE
z))8Df?ljV7r<vj>M2ay0=_(S8R6}nzuAoz2oDo%g>P`n11SxV+RynpW*c6UF>zf^t
zh!onEgwBawPEXKF2vNL;0|5Ao4XOxk{Yjt1ykLKBd1-E7bZ1D49w?(Ik~$=Q%z>3B
z&WR0*>UgZ_=_wmbByi*&QKf9QJui?k${Qi<>DavYCR(;G<Db)r2j;LeM2y&l9m(;#
zsiGp}w|zuLA9G-Vin0JAjvBp-%$}k0wA_;;P8L)}!Qc@0=}!WcVP+6V<ZuO$Osxzv
zp9M^gVAQ9;Xc+(zlhPUxl1@0VG@GqjWZgP1Y~KjPGQ=^j_&1!M*gR$bCYBiTkskbY
z)s$^?cE~e|#gk_wa2a!W*6A11psd4+g)8lS<!v?o=@7q(v@)kBO$uX#_F!__sBL7d
zAr0-U$~;xQ6m&nQM_g)AtKWL%3$;xlVn!4#u+*JYZ3F>syxfyP1)|>F4ooawd}eyb
z#WvB*WIC%b6!5M*>{I?~`qS@<q>(z6ZLV*@Bu0jx)7BC?za3M`N&U4g71QfiIoHV<
zZg~4*+gck}T{X{X<l108slf1d_B2Hap0UX?sNLn;b_KI;Qj5IyYrV*9qaczwMr*sR
z&Q4ZY;^YvC(g)$b;hS*!6YC<Vu9BW1nfE7dFS`8}=Q8og`Gz{u;`<wF3PD&Aat<uX
z8DW8V2gV@b_$bWg_Hs82CM??oKsm5vGwn_v8D;2<AE<K^%O<G-SlKCbxLQWn%05!W
zFZ+S|U2qe5Z3vU%yH0CYWbIEm{jrH%d9`^n2<`6KU17-@q%iNX57w>fdYP|irEtn8
zX5DP+l-6lV*^%K&=0k3Fdg680TxEBbk(I)FEL&|EvuCZ+pAW`LtAa(CQtU$&0|Y9h
z?SHRVSXzCA!6WwJ`rA;cLSpg?ll&|-kHg{YBNfNYD$JW*2}oI8gw=OZ9Ap6a!SX$1
zkrXyA^dXZtJt@>6J!o^5{P#nNAC*F$;W+qm2fJ3)0`0EKj8;j^MJ)ypkCAdGZ-)}c
z?k+D)Nl-r23<x>l^rU#Pg#y!dPhB||yvn(5_6|vSvVe)lJZ$&o8pdvqf4=QBwZwWv
zTlVa}D!mZ0gf>C5d=uw&^K<G?+Wlp+YAO^vE^8j~19ta8SOooLvW`U9vr+{=T8ByS
z!Kl*`M;_`>x;7Ti0}W+zOfjfaNd{RaA`FBqar(8QNZo8xtFvt%E8iA7-rRlB>B%(^
zyKG3y@Jb3b@;sgX`@^&R8N!|;lL|w~LOGS(6;BCQF~6krJEPh!(v>7^KT@W3FBAWe
z(+W&Wfl$2)B{9WjWIvjyz^>ibr#v$12syA6k{3<1+XMDv<;N+}iah1AKaSmj9&I?J
z1&`Q|H=i{t6N$paxImWIen}`oos!h#08{WkW#;VT^@C)vVsP^3DcVgIbrn3KKQ^35
z4z*9zA(F4!-*K9Sk`703O&lI51B!jJ^oW;atb(G?b(}Umg$Se=g1ea=_7eqbaxk2B
zT5$)4%n<NYO!Ft}XGW}cM$FHx1wnuOCbxIVs(bCH%5Pvx;%7>Ii!rDKFpf^AU-%AQ
z4T1l3d4z^sjvR1cS^;ZTuUNI-ex`0&wW!><14{P03ZK+ahp4Wf&AI8c@vCs1AEHB2
z1K-<iTivOyy!HyG5jTihm87XM&)W;;!coCgq!LAzmeZrVXH@g6R86!U`3(@!#uh;q
zG|PmjI&E5#mELsHcIJBt3?!1DMhZSuHqx2LwyUy-zP59n8|$*c4Ys?P$Ea{OTV2rU
z(QSj|^9<OY@*ydMiE?&3u-xI513c1J^`xnCTb7a1@D4dJJ!j$2P<Ny4t&Ei;P-TG#
znGs;hS2wh6UwML3_4Yx|I=G|4Fj)o9rY@}Q&oAUW9Udba0f4}amb{W(&OoqYdWY;l
z?lyfy<8xW&BRplS>jQSMd86E6dsx@U9ay^W$P$Y}V50;05+BS5QJCGiOy<jHkJzDR
zR-~w9rk_;pJDi@f<jLh7x5Lf*tBGN^S6J%X8KQEeE?P+pg+?$K1&34?3H?cUq2Iv>
zh*dwDoKgn!z$)#KNB^gfMtn-)aUjW#HD9l!%t$_UV9C{DU4@B|9)(sPud5V#mflE}
z7Lc5NT$Nd;KgJb4rjUWp*ojIeS?EA%cYrkEOXNF#sImsO;_Xf=m%s>7MO*L)4NTg}
z`qB`1!VwQ!QS8nX%J502Uj(MYYOk?VWdvGr`+~)tbYNxCM1U<UP>Q_1=XB{wnO%Io
zR%WCp&5awoNL|_?iP9XmGqV=5;Io=<$pfl}8Bvo6|3{p=QVbj=nih}i!$~;A{WzCN
zS)9N-NQE&X(kWf}Nclvi4aRx0;pUl*JD@-Cn(-bnBX+j=P39{oZA1}`r52adpCr1L
zG%kCz+^wn}@JCixdfjCnC+^T|zi>d@Ld3F&cWd|L!|Yt8iP%Oe1xfg!Ct0L?7>E@;
zd#q04dTKL_P!^Bn2~3m3sg{_Bey)<lJG*0>2wfmn5K!#9Vl6dKPo8<jP*9A?e!eme
zUVsw>i45-sz?|4QDSVk>J6(H<M@{91a)YNwv?l|ooi8sfDlaOVM35#~$my4ydNp$D
zOmk~<Tnjr6b@c;(`<?!np**8D%@dC|-;^_yRa2CgC+Xve)1M4<<76DjOx2)Y$kAa<
zIp<tAWUGx-YGr(bJyCzy8lJ#oCHw-4_eZ=7H?twxFE;Nl#~W$44NHP75l5Z=qOIV2
zhrj(&zPx{Qu<Klx`ItR1Pfn3I$b$X)m-Ajuwr+d5-omy0N?FdL#3`p0pCC@5qJZ+Z
zh1g=hTE<(rT14aki@S3=GRFl2&CE{wwfe!FKd!SF@NEK0@@`Sk&wjl;PRm}3-NRVs
zfW_kki-0tlHpN5iH|7irUy3rS3U$zvXoU1?q{(lVyNel(N#Da8K#SL&WE?4ShG1#G
zmFSt-_WIX5*NNs0W35drkKZn+tzxI7Xjzl7FG<{V`pZ<RXpe;9N9=bBEW(4?VN_&g
z7Iu=~#@p?8n=ezO=Dw&yBbo28)05&c%UNeAD6(9CYaaXWC9N-Maj1N*RrwBj_WKpj
zeVvK6(<)EF6UeBQ0w^&~k>cbZ*dLT}ioMG0Mfj5gE7K~>vBjjP8%N_GRxyS6v!3FL
z0BFH2?C-Zfx;qk=*KBaElEDn%lpohYJbzr?5)y0@lmpXGuUT-V{Yl+)k^DAvoo4>N
zPo9xM)bc;g*=cjMqjN1fQuWTUmi<{iWx*BmoJ(a+zjwg?Ja0e$02_L_#^Eq-e^EXx
zU7ZN!AR}X1=}F?GPG6t>W%B&NQHGkX1Jy?anrq{69Z52W2<@*b!SseP)b*Tp-WkO?
zk!`U#X+<5jzb?~(q{S4(jt+5TJw_QRYkvy-VbJDH%DNQ+pR>QoZyej?Tm{kP0)k{9
zy1%V-u#(zEH1tU%#N%4Czxz6RJVf{%FtK<N?1?=6`#NT1z*IF9i$@5L(^EK$YJ5iY
z6<9;JqF+`0{-F}ZSf_7HMRE!NRKW9f3jIIUy?NHvEz=|5IiDaj9kG9Uk{LEqK=m5Q
z4q?X37$!a%Rde=Uc~jG{Dnw`t%0R%`65apYAYMKRt@%<5Ca(8-(~rn1mZv=>je*Jj
z{-x5SilKK0+dT)8s|3cWm8l}hdhQ|n*NWGX!G5PPhc82OVQ|PTOY+>m6(mI1V9|pr
zj0n-tPbqk4|6ZOmH;yzSNI?vKA*a7+5D{3?w*RO^xjd`f&=x&}sIb%iv%EG9nv%}$
zbYP+ff_vH>`>*DGrQol^yAahCp^u%)>CgAw)Y%XP;y;y(xV0Gp3Xzd!VLY;@&voI6
zVh`G<bK$OAz1X=7OlnzXeJ5$eh|Tty91$`X2rmuvIwHV6+n`mFf)<BrfGYVSHa(y1
z&)`UT{i_fpw12Ky4~mMF+hodTVTxhbhIrCGfA<rw1SkJT$F(>&7sjPSK9K9}3;6*Y
zS^~4Ap?&52$3W!+gn@rCF@|8-*}A&VY38=0L;rdM_N9E<Iy}8N&p83cf#e<iSbPv}
zzg!TPzGr<N?3$Ii-RX&?MM$9BCCuAd`%1&(MkmIcmY%h$JHAYVwXe>u|JWR;hs`9&
zSxQft$v8~`8I#QeQ-*|_M<DnAzWa&g`Yg~2KMk7#laQx#N45gf8o41s`BJ>m3qf#*
zM!ZAYTP$dPwQKs2^Sm%NE|_m$OGZwiUxG!SwB97@hAM-L)$YlUtqK=8fnH(C&X(7n
z>PYsKa-Ra#mCSOMmjO%KV__<~O6L*9!0f5z?w~{{1S5(AOO82j{?+!h@>3<pD|J>Y
zJA?_RXAbsbQ@5w*PZ1Qi-{v%O@lc?3gK8E%qZ!6=*`-*~U>_VF^V%bWo}%Q_M$Ej}
zGw%sibul(jj^A{edL~5`Ik}6nlJ>0Z=QysR!S1-zj6sQM%eBSf1orHE3PGtj-uSrF
z4s2+N%IvHO)#vT7=QKSLSHPB~1QDGHT6^G`U{wQDs3>XX_S}lCt4KknkvO4CH6Cz0
z*z*cZuTupK2}u>Ef_exnD2&JK`S&D6%$w(2#$$wN#ktE!`ho(>^rRpquP||PC2KG$
zdtv#846BsMNddv&!Rd*o$|Du29atg0p_!75713c1JO|Z_p0XgMZpm?^^FoC&{e940
zR1wcIroIkMF<y#A$Q$tDds53U)IZ1esm(5o5i`4;yHeADnNrjS9o)XLj8Xhe2WSxA
z<qeE6XmpCW{ib`0DV$8uG})kd0gQ>wtyH%3WU(jhCFNss15*Ztyzx4BL$-zOPX9|7
z{mgy;GcD7;xjeOg3oMgaL`Zdd0!6h)hd*a!yqvO^HV8izDjd_~z{+pr=w<VTkQf8V
zP3h)a?g{zsRj+lfVk^U(0GWMjMK?Tg<ABr5wKX`hnT?DX(MleKXt7SAsD~X`;#W~e
zak#mCdqoi<AJPP-zc3-A6w#BbeqDEieMfm(<*Y}^C;^rP!Gf3yZ;t#pX`RAaXfI3p
z9o#hGT*j?)j4)V5u*I|AQ0vORGgstw9f{702(4eU4x=<SM*FTpKgsj4tHR&s?yiAx
z=dMD~C&Jf8;ST%md&=%KF|xTcsQ7yz!ehIOcX`adr{a;?A_%cG^P|EW1ZBq#su;oW
zr`yZ#dE!ET%!kTkFTcDvhK!?HUrN)QjpgIsc-GvK61iO{@c6J@P}drEftpqmCS|8b
zL!8;RYW4c1_KJIoF@~dtuqY_jjcpNVLH`lvEG0lmg9_(V3E~Tra29TIuC%obGmoz<
z7}AoJvdMXo4ook@z|n{A`FqQtf~S`SuHsfloSwk0Se_(vY=d@DVsCJCfH7@#n(2^Q
zGt`<S2s=VPFRmm8alv|^7KJJu;8ZPo^hAl`Is{lrCw57a-OBC+0edv^mFZ_kB{*xB
zR=T`u-sMgsZqWKt=&+pguev8qCj+f5&UN7c$;ckX6So_ZJ8Q3Qe#!Y>{E`*ozyhU;
zdQj+;GPI)n3xCf*ORuTpT^eqPuxJ#R@DVNxY>jqV-C$OB`k5@JEo07d4jQu8R(fku
z1egM&l{|5xV1t6xFR$Os4kUK9WBRODm?As0O37ljE1LOl&JJV|s+_d+^L<W#B6qag
zgJvjkG-+2h&xVOG2}*1PmYk!_Su#4IWL}Q2yRo}b$5YT0AHZlvaN1Sn0ZH<arRl)Z
zc)W#(=1H4>&l4}xq~85MS?A2Q{Li^3MLIB|RObcdu~|z`k>e*FSehUyZlHu{*|vLj
zbzK<@t<Pzt9HZl;AT68~U`%4dHBC=)G>K$UI?CzE@7<(qUnuU{a%*z!%rpoN?DS~f
zsM);AURP14wZKbsx+<{zj3G`y;z_Ts?RBF}?wHdSW+vFnRDkgf+3ZpC9tp^4<{}^O
zw%hi`=J~4IBHG*Qz~b{V6IZ21WURp7WN)fthVJE#Q+?Yh%ovBA9z7LOpvs56xp{)h
z6N-GQ35lLW!{(N1%(d+;6+q01*suf8*KKUIi4nlXw=fA|No$Zo;xuC6X?s&bIVrrh
zDAC2bA91d;AM-sM8Y<mxTU>@lU16LTv<m;q^9XV9L)o*{2ba{9p%}`@M*~u3JAM9B
z5%z!{YD@1aX7JAN1}a`qr76}X;3OfLQBleV6klq~Dy9*pB~pMl2PPuF^_E+yo_Af2
zvo(QXrsbtIvu*wAwd-wp`Ei~zBc8w4D@-O0#ST9<z28<eFX8AVChfuLgAOcQ9W^FX
z3c@}PFBHc<2<5V<;M><H7B^6~*|{$4OExy`&bn(q`4MleoOVfLp6@xp!VM)gD}eX7
zt!y5l5Yfzah<JaeC-^=NLj<iG5(5+xge%b~n9k^UP9@-{5T>mvJw$sWR)D-vjTIY9
zj~=%$&YQ8-<vv1!B7XzZ4+MG+URHZ@xrJZ^^Hll8HFc<w&|0FZWoK}Df>u)Aj0)nh
zHA(VTB>vmVhwG%DqkK3<C$BIGbdp!85o>Gf0==QrMgei!fw0}6N_9-7qOYqXXR+uA
zVioO*Kt`u0w})^pq=r;%HIkle{p|Q*5x9VrZltZm=}AItZG7JZ#v)b-?vxvIU*<ed
z=Q1Zc#ISU=%Sf@$-d>mGWLlOSK#Vzk==!?dc2i|kD>*BEmI%UdMVhm{qrmb=yo_!K
z7O&K9At{<U<ZH2o+B@qj@zNrEO*()=SE(Suo$OutO7gM0&UHLW!i_TfWJO1-Wnylw
z2w$Qqo?}iso|L?dm?(-X-%?<E5$ENmVv~1Zfoe$$?9SS#Qpe}H_>Sz>`azmjmIV1f
z-oVf(WDFCZe0P0L_!A;Bk;&94*^!^@^v8Hqc1D&-A|LHNb&L@@El<=*2bO-e$N}Rp
zZMJoi?!T|{sP<T^FMK^f#7IU5$Hxb)Rq&FeE}pI!Dsgsta?d244D0oV^2(%d3<c=&
zz_j^L^KQ&G)?q?<3dB3d!R5djYE<n3afpoYkA2YE`J0(-iz4Yjxt(5MW!;i+#B6t(
z$(s$H3|ZIdkzr?H<mp6_;l3w*5NjGlry-7>KqQ~{INCE3?MdrTT*$dL<TRQktccyn
zdq+)`fA0{I<Uo0}BnxfgIxQrJK~FrIva1kuD5Ex5VM_aCUxPI|u-uJ%6IhCR_TF4x
zm)v@>b17WjgbU9$J?R@#DTKffaOn{TqeR_AU!M(?S4b4lv|++=U}c%=K&&9~Tsbye
zW{dsDxsJPH`1EVh#QfMueK*^-=IjXD_Fe>vT~1FL(Z*X_Q*7>5$Y|CnX*-bIvRg#;
zjSRMIv<xw=^1(G6SXqlYEi8u+?91MlG^Y?)6`!S<OEFCoeC2K%D?iS|pJJQH?{j)`
z64=S)7<>#C(5n;(_Lx$ZuasyRc~gs_KlZyq{&SRm{4Tl|EbOuI{Hn1Irh*Xy=^a2J
zrPK1U5^8k7ChF@U^H=~#I=~Vgoe9n_>a=Wg#n@t;ji?lp4y24a1CuFoVOt6;6c`YM
zl<mM0`>?xlRqD5`NxSFopFX?NX^66}DF1A-Jc7?=bCAFAi~|#iU2*07%k8$(FT<KD
zp&pyw<G>1aN|@@{`wOZSua!aH#l_|H%!!DAd=wukhahN|X)4fZ-=EK0%z&j8I_+3r
zlJVHK%Kef9q_3T_+w=Zq9B(e~x6{6m@oIWeaRUv-MJ5#%bypTurbyiO17-OvT)WE4
z^&$>ovLDD5wr2H8=Q1_F&Sp&2s!M|^<CWamV&wXh&KVnx(-WiRT{WgnV<>fj9<vYT
zgdn#~M3J7h0}*})+5}_Vk@!V##DItItcR)yEEipTxeTQ;ij$shcb3`6u-~YU>J^s8
zGl^7>*@x<kz`V2-&K3)&Cvh9RsRV^X!bHg<03NpwC*uHTL<l>rvI4d6tFR(d+(!~$
zFTeR>=Q3`Rd?AZVMC63#@`Kqkv+eu?aEFiahzGFMKR|#tJ!QqJ!aX_N$lO1abSjeS
zIgOyfYp=Y<?kYD6--^tam9fIISheDBiQFvTmV3A6NZff3uZ|tm>DNnVwkbXhi}WCR
z&OObuQyw^tI1j9R5JyZwqsn@Yxwd-?J<kzx!LEv$SI-I*=-d^*e^S+<K*)W~2#b$D
zW@k3ai2gteTo5N1I{|XX{dF|#EsFb7P@x3M$GrAbTfMRzl&*QyK3aYw94NpdUF$0T
zN>8$f4#d_b^41=xtNF%$aME(x!u%z-+<KXPEN6K+k$2_thVEwj;WEnNwLru@;K1UW
zWw6n1msPZ|kJyitZ%#d1D^f`yY}bqxm$uViNa0X8xH0?Kk5+!zH-tM1X9ob;eHP4H
zU_Vygl&lz`89qYlCpHgyg0|<BN#NV%{|!H_{dno0LMTzNJ{4ra<ZL!O{mIub`SdvJ
z?c?=3SySxoQdzvhG~e3q;hOe|DgiCT*}cJO7BZV0@9M0g;f{T>JW3Oy9ds8*a{&vK
z#nKLzqS9Q_0e_;Rgk4nxs6r^f(uO&2-Zl1<<x8~X%iJv&nlP&b)`u7+{knO$(<aB~
zr^?M@kFu^Q)O}77E~h^xK=v6K_F?q))0Iqy2{nYr_8dqcR}l<~bX6J!HX2A@zJk5>
zGtE=93FzdC0}d=FB#4%82W2Vir*!_amD3XKMV!VKbkq?Cn4-&7mtTIlZL6q&7osro
zh!YO1AWrAZCI!gueaf~sGtusD=r~P1ij$k*Otz!^EX~;YH}LKBh~&px8+x`gug`yo
zc%X8v3TU>g;(y`bu+vDHx1}RXZFj{Map8kH8`pt|UYD+VyX|R^phG{9<k^eTV5~@+
zy-vU8syBq4#ir_$uIrB6)*9;E=0NgeL!A=#OB9Li&5`2v={najhw_OW0>`pM@2fK+
z^#$@cHcj=-f)SO3I%50l9ayPk2!blSfwAGppC${`f#&Jwkjz6w^9qypEMM!6$M3L%
zb@T*}BbQ-u08&P;ItxhD@?6C;A1pl*>l7(Q0GwzOVThhI?NhC10%Iy%`cR^Cq`TF*
z)SO1DV$Tkjr(lE)N}PQPJt<O+O>EepXB;WKI}M9jMkGaMLTr;u!n?Ukv%#!xqC{uK
z4;a+pZ7?+lmh14Mi(YTX$_?`4D=`1~^g|9TZ@A^=OYL|W-#!?g1d-VTC~Hp$w6XeS
z58)sPJ5e=k%!EVTUYAoPNmT%<$I30;XAjknTRV~GUjisAO^jD{DOJ2alWoY6wv%O`
znbV)v(h6gP44#C&%1+f8By?B_-;)kR?<I%_Dd=>;VP#`Y6h921ITzcxD<)lu@v4rP
zan1*)kJ*`K|CHHFW(P%#W~j$(>?L;vF{`?C)|ftQ4_8(((&_Jxmn?aJ@zqkacfpO$
z$>|y!KJ+lX+a9S5eenbfv*)x6r;|Ne)~zFUwxL69ta7618L(BSC%4LWXEU$l^B-+~
zzc)<u3Mv97dc;wPly2wBh@|YxN2~a&YIoUBr62lrH}TVQ@IIEfo!Pc{^*fwKnL_QL
z$kOa<b4oDS&*jZ4u}qIR4RI{u06O;b4Z0P_0+s6L5*?kMB*m+)m~Yc%@F}mrHBr?n
z4AW19(-3-N=j%s;&Iqx`h5;<b273uDKy|O@_{(i3*XOm@T<Khv70&@#+$7;O*0nue
znWk!<^qgj(VmQt~{e|o#A9;|lFsB(WOhVINyh=F!WcDN2%ATk*J3css#n)-aTgx7Y
zU?K5|lb7rlE6E`Wt2#tvmBz~88cVx~@T7BBP=*v7c1Ql6v|q|arXX7pf5!m`o#?BS
zH2CEl{>-*bEI}_V)|>)0ex*TIBj#vcLX21CX!`Z(y!&B1^S|04JjX+RtOJt{!Z+|=
zQ0o5o*{?N&JQ;+(pDh+xd^n5`NdWt={JzLH581EhoGhE5s%D`U@&fHQDow0gc!M{M
z<?GhiZ|0~jxc$;gsfk8QJJ<!Ud%yiwT?kaw+N9`W2ci$?6E|wxZ|AKm-nQ1c%#CrJ
zIEh{MJ9U~UmL@L(?HFw?kMw$m3E6%(hd4+Pf9E<TpVH&nD_M;46*QFw?e}tGqEL<2
zr56Czw(#tznBMPK{>bTflu7M408JeQV_F=ye$NvxrlDdk<b4p^ydes@^qh`DMN%C~
z3UFAK{;+<?1XYg-{-s?w39A~eL6so;ql%{9#D)p#MUOa;_z(%1rKSNh+SIJDKdw(<
zsp0yNl;prdWv-OWajLO9Xu&nYp8ZLA@|*x)c9+NnRp5@(pNK}LQ?-S?Qc;w6YJXbD
zqTBUyIm`(Xe^$S53pxN3MOzb$#Z58ze_ozOLORNCTFRyDA9s543uhxzL-m$ujNom5
zk<UH<Nv7}<(^FpBsJnw4j@Pxn%xPcBv7XMgC~hl^_OI%2dYyg_?sM9K?#$o`Ze@R6
zhuGWL+U&IE7-*7hMD`Dqqci;SNT0t+GJrOxx=uq6o!Pc|kT_uxQqTUj8b_IB3jGYW
z1MYxhXoH*W@9NOTyZzK&a@v7`RBuAjWWLUXk3h2g-`8QT!DZvLk|SR?F=qczzDkCF
z3C6@5II#GRxg=A=q6nyq-Eyz}WAjof=|O(V!wTDtSOxZD`ZeBb7tFJNsw;BQ80D&*
zW<-N*%kouyLld&vlFN({|DSWH@;X7{v<iuHusow9JIMZ}PIqWcS+!!0SJ`L#rEv=L
z#mmRIkag`}b6oG`xGXNUgHe&@g#BA$1=k{#t!Tu1Q;XljMz+)by^baA8CbxNJFwK&
zI1jiVF7zMuL#Z=1(sG&#CVs<f>_78r#f{EYtcsMm;v9&Wt7G<G%}Y};S9B_{!ljQ)
zAF@x?dA#Aqb?cp03{uu%8SRms?b8)GbK)&z4N(|A;*Qgw2*tXHw1O1;kbNef+7|&|
zPD@;0<_T4gPRrA06H$xsnK;)P65@=+htK7QVaei3WYy{yB0gW<BT~>VsB^Y~#g|X9
zhbrJfyI|^IUno6VGDU6>P9DF*Yme-p^83v);4hXpq~}n4X}4Dx_E4T37`87pWGatn
zjZ~ers?wkHfzwyfohqLYaW(ek@=HafpxUyb$*&`$TWW0d2QK9pRF36Y@?O7^$XVEa
zFIIl1Rs2*sg>)-KBG2$wE05081TfJ8iS?^Bt1)nbEPqcazA(7@e^<eiWzr$7UhXaj
zsCiITz`?G4tuE?bXVhsFp5M5R7~d`wRWYwAvwKTF++Gr*jQ-M~rIVAG?oEHhiIo01
z@kLL$w@?wuX$k;&5UH}l^qgXxw5Q%1L+o-M0t2R%`*MFv#ZRA>%{XIx-BPD1c5(F;
zSK8C>ErcRN1VoD8m6zZiae9(`!z`}&@yF~L<!)J>R8UN@s^P|Sdg5-`CSv!V-fz#G
z15;@|>IXQm^hU9161qR0JFz{hzAW~YJxFqg9avCHyZ~w-mzI|)W&7FX?lgf0?Rka4
zkwXqWI(^Wd(+ohC5J`SVyuu1~N^Z92mN#hy$)Jo@s`ugjq^E=$bcYCP?b`G1ZIpyw
zC(hXgl1nWxD0`SR!|i0zKR>6WGZG$sT7e-IjgZ+zZl)L98%%L#+eH^T*ZG_hm@dOn
ze6+pr-m*u491Sn3KeKyC@sl$`CYIyQU2~R*in>rR^WMTUIjE)0=0#g{yn5NPbxYUV
zi^?w`IxLNP*KP+EqtHSWxhbwyb9&fbTnJfTC_<JUZH~m|^e2B_s?#@z>>JCIWUJNv
zItn-!Gd&`|`Bz+R-*j);ALq@#(rF~J6$Lqg740R>+L2K=bUcPMo5xf88EkBb$|RV=
zJbQ;oc={3h=K8?J<k=w9I{<w@)%6h)?WJXeDXL3JMs$MpoOF6(QO}9+40a~=D#y>a
z6wJvk99bcd3NUSOgB(U;-&(&NfnW7R4l`GprK3(yF3g4F7#;1~>ZD<uRP~#E*ld7Z
z$b)Dp!$@3diYW+JNA5nVvCiYjWU_BhRKDq=*E^Tm6UJH;y|VA9%Z!*xK7BANEIQyK
zY-YZkG8)BZWj9wX@G$Ay%L=y;vtxs%Swb%(G-w#-@po1{wy}VD(^}3UQ}$gAZdAxg
z&c{c+!s>!m6x(;#PZ#wD-cNgl$%6swXGV6|_cU~O07Rt0OGS?*CKGz$^oyESw71&J
zD{|r7L?+ObLXT+IhpkN#&jon{URG&&PBVx-EL03Na;>87%`55$u2B^<S}DaXBen9v
zoacU24d=QDgR^Fs<N)b?dS&@W))9J|N)H@#V7Xz;dXT|B1S?!kKV;uq29;ZPPJgZ&
zdJ64e1nSsDN!?4)izcz{Pdb30K!Szr;wL|a*%enh#QW*Vk6DRUdr28gkUayGi9GDj
zWvSAiuOv5ra?~!ZXz8hr7_D3-b4<h%2UPh)qVTKo$8j{p=8n@$kv(LRL{paC2_ZJ8
z)?S?`o7qM=p&(oIJz&XNhFWoy&TA?YUPnj@^?pYjNbYF!?XtvzBGsJhvMa>~APa}R
zwoZB|W-(7-UU?JrB!@<EifRW4Y?oK!han+SNc#jNCelZ_n7)o(QPH_@;X<boPpDZ#
zs`Ytg^FC#b<!aNaAbhbIoqQTvdq-?ueS%`^!!dovfd!eSWi9wLd<ZX)U6n&onB>H;
z(<(bi#+<N;G806r+Wck=N1-$Eb4>a&FniO3MM%tsa?fS<7Sy?3y=vt<otAj>n2JCu
z9>cD#n^HLGQX7hC=R;0}8jp16+?5P>3t@(KO>J$Up19LWWHBwu&lIw*Z3Zfos5G*O
z>A=iBr>9IM5hN@a9nAK+0@D^FhmFqA0an&@QyWc4C+pz#m0ZzkrCmo6DMDJrqzJEf
z`g8js`G@K$U3)_XtfMZc-1`91Q!Kl`4*QXa`Nrmj=?#|wM;hK)rzaU_n^16-{q5MB
zl3K85cE_AX8LIq_l@B9)*qigVbpx%jQKu1Ku6m~m%0EKkowwxBx`?{t9g593fY{k7
zdy4|UFqgwnXTrHGc+(D&I>4e_B0Y>F&UFkTHz{F4?mT2DTU=KiW*_RAIqimVWlh)2
zyVKz>xS?%H^Er7|x#CvC5$E(IB5?VqU)+|Kn~TpsEI%orH9g2SsvWNh5~ts5%j$qQ
z2d%?&&@PCHokl{VT~`JPHW3NrYGV*=o{H*%u{r(8N|@j+RQ_FFSxJ=W>0)>XB3811
z<78Ic8#?8djm}lUqQo*87Y^C=1txc_+>46C0ajR|ggmOp5X@f_1;K~CwfsinbjF%Y
zIj~$KS6p+=<+d_gWC)G3q-n+T(u=0gk%_i3z1t1>JQ9gpW3kbbIsz>1*qb>ONReH(
zsvvtR%!N_eHdxC!r>D%Rk{`qvbl^kdBDK}cFHbTVvXGS$(mvaxtwn)zilCAAXifP|
zI_knzn=HlYnParc;xsAF&fNgtR&f~XRNO=<8uEjspi95-b=}(4H?CS@YjdfrT)lYV
zN~fW@TeM=8t;@^di6Q4w`{K;DEj?RbCZM{Z6xInWkf3d-M(k-hsBo~3v$-)J)#__G
zmvO5WD+Q#8V3c<^k2l6IW^XV3sj~z@v*z?zjs1|*A9O^WV=x^S3A-t`$<{E@%%p|U
ztfU47E8E~fdq;lc%(m-Wo18|hSS6OK5RWui-S&vRv-vo=)}vyq1B($SZ`S&jQ&JQg
z1{Zr*nUmz;(t4zJMAU2q5$YQbICqJraEy*kj_Z&YyZPSGw-=7E{|>hCOO0d5cc6ub
z#$mf-c1z-JbhPDMmTwo2kq>8w-P+(&S!y}Plt03$@AT_8v(f->es}#!xVolDK?Mai
zvkg0ng0Zl#*n8?a=w&8O$AL)nS*eH<`@STj8(NBgcUom|$_fnsp0(=7EZsVQp?cf_
zh>$3Qa;a@7BbUZb0$}`2W|ieW=x;=AVN_YINhyV1LPXkl?-Q5uL}r9Ap4gKzM^3B!
zEHfcnj8<H8REYD7GM``PY-0?lgVxS54iHx6Tx+FZ5I~N%PW?pa5kXo`P@jeA@bEZo
z-TKOXiV#<@!C9gafK}y=L+{3|ztCU^5uob?+6z1Q^j@bYx!FQQY#S)g3R?)@q}p^9
zR-NJ`kJ(_kPex7gP3-FqEWb3wOHwS$#+vr=v-aNloACaqp7%)y7FZXZBgnun%dy;y
zF0z|gx4Ugq`K4BIuHT&wtZc^cFy`YCHjW7BwV~!cDitJ~%U-WA-J!BmSOsmk;^=)7
z7@YSx?HGTVNk*xVzy?z`Qr@UoR*7R|!SY^JlAtGuQ5Jr9pA<*y7MMgw`RR35C$MJB
zsx*Fz@o6vKYokf&>xQMD(a1&FBO*<H&G$9Wl3z5j4NSD%o<O!6%l;XR#&{grLvY^}
z7;ll7BM!3Ldih{9A%%=|+72KFuA<Mf0&#SQjh8{@!8)XGh7_!@dwK#vng+RN=f+hL
zb!QXFI~H7VwR5S_o&N~d<IUwI!j60~od4j!M4yCWFn%f;Z>e91P(YBB>}lS>7+mtr
zL$<XHq1r0<vkEFFB)Q7z4-98AFH&b%hOnI4WR+=%#vdIRYxNI1pfceyZ%Tht<dXt4
z^o3{bw%X5BXZe1eM<%01#1T9FscbFocE{TG{_+jlKeV^17F`45G+wz7PRtyz@2@Ma
zk8*ZSD<}K<2_m@(z`nhb8MA>MCd?<nVtuwp2dT)!ARQFKvJcdUv?oXKbWb{flqr^T
zjQQXQviuK}(dp9^1o|+lC9tGNt#fAPTtu1Z05z{0*rMy#IE|o^(0Bqj584OogrSEV
zNUS+9vFqhHC?@TWB!>4A9L8m`m_eN$cV}(Al}tRRm7h;M%7~nmA1V*7I>ITm5@0b-
z%vZr<2_<48h?BAp*EKv!A%3WycH!cW;L#)ASu-I)5=3!yjJP8c`;juStYCIA8HIGJ
z&#^FjQ3=E8Px+TKqQ|=Z#G~5})^QGCGtmYFAlH2C&0$)#AF2egq}4{hry0Y}qBDqG
zh-@KN*j<&fQb3eZlVQjvp?ZM(oql0*#SIIuTWWXL(F~4owj&`};|@raw?%~@h#<Zv
zv9Nkn+=2jj+VZVj@E*Ij453O8xo}w*UVD%s!9gzR#Lj)?%jECWQHaQLUSS!^7;|Iy
zSCq&XN94Ez2$W!zX2a3~c*s7QFkHVSSsFbGnCM|pk>%s|Kz{3*o~r54D5NGCOjR@7
zK9;<skAoDP>$1MFD<^N~E=jRJTt~B+uq^oi0K}sZj$pI;zz6L|n#XSsgTEppfeF{t
zVxxe~kLJ)8QJN0(BduH?FaQP0sGel_RP4v<`dYGL-TJjF7O!_8ddz6ldie439xb!1
zG`h=y<(5qG3F)C9Z?--aTF}FFM)s^4k~R+RDqlf72%oC+k3Lc9ij77V1HLek;I${i
z|MG7JcYLz^CiyvA7y>|OX-<!bJz^2dl>9{V-I6|qI=tVkqQgGhb5QcpoOrRHEF`UX
z{aVB(sY)`@!(RI%k(1Ar%|2=*!0i=^@KfcTMegM(;_M_5Iz3_$DC9bX6u;U}C#uPn
zT(Z=g#-baR*w55;g*c(%g-Sd9Fa)km^OVz{PABY82*9_c2k&R=%nT69ik?bao7dWP
ztJm4K#J@rp21nC^;J^x5WOY2u9l|AiczY#_;coAJ6N3c)0f{{ZfhsIDjiVptba;A?
z?MOB`|JwP^rK}bGe727IlA`sn?X29A(`cz;SzGHml%U)-OU+DE>=fR>ShjL(QnTE4
z*YBgQ3yG~xJI5c2a88m~>lxcq7ZJsuH~^~O0mQBFH}pwG&y-`WY$^x7Y;4fE3>r;E
zKmx}p+gnGvh7<mrRy-zD7)BarsN*-;Vf&gMC50uap~7H5qDN1i+14#p77&E?7i{L6
zGV%l>ticmbe|oFfb~vhBMe!6ha3Cl2ifiXx>E(K68#S~K)>n)uuOsrU99RrBbXUIk
zAvCYZud@fsXD02+tas9Z#oZMqn@v!#9cu0lQ!|r&%z*{t8`(s006ScMtIw6gMM8*{
z<PoPQcki(d%z|nocBJ?i5K@qkwGm6#Xp+2w)C<8o_omg0<iE}sqMoFRqh&bq9bu&j
zhaIpOyrQJ#q>W-Ep+O%@I>2fZE2QTDG;$78VDjvE^KB0X>M5P*6(;FNBSHGH6O{;-
zQ6tM~<yOCF3`23dV-J=0&pA_zPdhTb$LZ1QX13kbMS(n7S8bkYBXcIJ6lkujWH{ku
z(aYs{!hU3@>Snx{1DS}E%rVcXlq~kw7V?IjuCs^7Up5hC$;yRv%;^yeP=#ANlY=3$
zh(J#-Ei$(VOtOdTOZtW=!{#)#39QCm+OkLL)0H<FbA~peW#+XfqYtuzEF?X>*Ur|x
zo`m>`QGD5x4oGNgQseB=f_>$<D&<Ee5nzchR_(Z*Yu=o}{z}CTpSjZ`gUd)~lRZ{{
za-L{3r=<_R{q|dKv7f870{@h#NC^N3ma;BRRi^Ie^Y)o-)Un_?mVYJ=#c4}6oexL5
z!M)(L3Zco85GE2R<S=^V`T8wIA14&mffY3)^F*iEs_U~=bYZ4E4LfseeAqh}Br()Z
zPthjPG9dCyiAqN~7C?=L?ep>a&Y4|kW}(SDuw2Jbf(irGwYb-Qq1-Lgl8}KL*ntUI
z>z1y4yFJmcB)z9`w6mgNW{=<L^e1-0>Lk0^ar?#kwpel*aTZ*|>503w1*jqexjObs
z4L;&XLP_r6P*MJ{cx+Vk=iD#qPIBUS{W`uui73Jgux6<sVm^LH*bXHN%6>Fuzfyj)
zm#;^$(_aTxjp1k%gj#L1Urid&=?cTxI%x_Eoss^o{aSLYc9#Q+Fle*d3q(38ei+5U
z{(2qfVx5XgD|sB!BL;kmLHtIAY@jepYm#RGiE-<6P3iqAhk#We)HFQE_M1t~7r*&+
zUM|-Uj@oZk+C)(1dJb($6F6qS-Hc;mXxwSV9a%<OCw%U--zhMJBt%EDF`4sO*im?1
z#I@M(){S)|v0GHep&ieXuNNLDnzrAobFC*sAE^V&UP4b|7=M6_G)OkTpF?ktP4qk0
zIjtL7NXdUt9$iE#j1&7En7GYgTk)=cSZ)Zh5w3vp=)iQt%(i6{L-t4cJP>#eqM?<j
zh3&vG(S7qgauL6>n*KOP(i)j;IoCN=l$#vfILPw<lk#hkq0Y&X`j{h&jnL^Y+oHVG
z+M!imP(H>#Ej?*u%?LmwA;i4elS_zPXwaO$`LpI}5tR<n%lo3riM2=X#zh3Nm*d{-
z&nr5@S(Py_6&7Y9o**B^UnEvq6ROYQw8{iNTB0cq*<V&9VzPD`IVsn**qHvR`ItDh
z1#wEqz~=9@N63_&jy0|xxYu7-#Ip(&f_E6cG`9pT{mCmy@Ka8MB2W2Pc;(+zYD0AG
z5YtRM?kprBUcvo3e)n(dog2Fxh0u1|Pr5S?U*r@a`@5W4Ou=zl4Z0(cq?xO(zTE!4
z3|1nF(nFHfHia{s{^r4A!m2_ggV&V(L;XyClc&N&`UD#xpM-&u{`h4ToFwjxCLXna
zER@&Z8FsFd@<^&c^+jm<r)ET99^|8$@nD8waCG`LL1HRI6902~3v4^ASSbIDPkvN-
z^yKq*z*0Ez0tND4D#cJTM8%b077i?tap8p*+rO3@l-8-U?Wl$0z~VE4Py+#BLSo(j
zTfLiX=L`!y>%a;h#&yiB6K|&dTe0(?>4^P%k@C_?D;Y{iR~#B=4uhbQTqp!P_ok?e
zJQLv)smuIF9r2<YZgLt)tpi<LXaCvQMjGohD4%P@Lwa@Y(r<)r3bFiG{R;L{lvLSG
z9aym~>o~Fj`&4;uy1VO6t=+I%jFm>kU6MR{tu6NHd<CB~#dcp>6g}bOVf#!)XapU^
zVq){>;3=mkSvlyB`b(r4e73-{07g^{b6|nwFf@ub;3F&%`DTIxeXjIt*@U4V^v_*@
z^e578TE29ZeZI;)<YIEJCrx$o0}3|T7aAhY7`B5UkMZSRdqkQMksa}oih}%N8M0ol
z8P&-^z=WI2uesK~R36D%kXTbiJV=e6MEh_TiGIYsoFm)Z9nzWCPBYLWl`qOv5N`05
zh8VhSr%}Xj6<UXTeYN&3Td~S%6wW7`g2aM{?7y4OlRg@jtHhzxlk1ZFPaU-0Z(pkr
z9Gh*qa~w!n;SoH~6Kk13yDxjRD&;XwZ?FU_dh#QPCsM+YicdY|zMRiO3%I(bG@mh|
zM-WtMzsOH}fJ)i!Cfq=k@)S`2)O;oiABRFJJA!F!VC{*q7AXxXSoO5~iiEq2le)G#
zO`$8x*U1R~^!s9x=3jA@bJ={FYRxpkI`)kFLM9@5Z_v3;<}v^B1@_GH6>>Y*0tVWF
zG{;A0PCNZcb&ClhuxI7FHf-4FT;{t#miB|4Y<@||&n}NtZPkd0z|Jo%LbY<Vq0*mv
zV8kn{LK<n3&&ek%bCJvP6{_S`Qis^R9=7K;Jt{Y^?OkjW)05YpWVyx6gqpD1o>$Pb
zV4iasv#3MH6+weaYtO&0P%qWn<h7?Q<Z#dr%IOR0v?~sqHxbw6z+!aitxv0xzH6tw
zu-q-bclbu(hQQ)(gd5CcvSUi0LoBgxXr5DpCXwP}3!vvxt>uui!CamOlLxV<Nlh_+
zu_<wn^w}5XH(s+~zH^!LrP^xt;^rk;`KU=Dae9IqrAi^iU8ru(EV6GbA0wGWLFY2E
zRhamrHi$<F*7&CTo-8bO+72jr5liBry`+-w`qnTxIy95`2Cxr*bM0G)DPY8DCMRH%
zM6TatFRd7rGa(k&X|J#h8m6=Wme_A8_Z5mKw)ucpm?V{uevm)D^}fO$I$Jjm>g){%
z5+s346M_{{;r4BHi0f{AyO+i#Y{;kV+jA9-bT_x0YklR!7<s!L`;PJ)&59^f`w_65
zih&kl?#s#@(sC3+r?URQ5*eMb3H#1EBQQz~l^9f5JP}(K)-tC*`Rq7R=@^f`D<6N$
zrTWilNe||*>aZYucT&WHdGnpie91=a$lud^Z)&_~YhGc}I4LO^o6*b5eGv};m1$=O
zRz!1*cbbhlC;Wz!jO~Kv1NarcVNLJ>61`-T+UfKQA5_NRz#@A^o%_D-P`l?ebCl~)
z=q@ZzN*Yhs7Yr1gCp~cV>t-yEgs8r<(0By0N+D-(u_rh^$+bh&2@jclZ-W4xlcWWC
z)MsHrfC{%9wTtR+39wOq+@mm=28LxD*Cm41LA$s<kF}#f9R516-06i%R#j6w8#bwc
zs7uOkv|A7)CDqG;Y5ux|HFc;4cxiKsSTXpVKJ38CV-Lcr<Cr^f$FzjQ9beTf_9zIX
zI1xTjvJ*Xp%9Y={2cEIRUY)eewlm;d2bY46g+f%^c&H(-xi2jjB1dL(ZiWi}VX6Yi
zM(*5Im`Oz7u?9rCaxsL-*e=VD+sL5t4LZ$odlMi8h7V%Ul$x;9Ui-iFSD{ow4E<^I
zA?uTJ*_W5EE4B%3u09Jp6ZM%W0%KR)_r!&u7@UdOv2C+b#U1{)x~1H!S0MTk;X7tm
z76gR_kDqc9FUUY~c6thGl>n-Mt}`~TfgO^`kpZ1>U~+xEW1(G@Tn@6wG$-W_Kp|6=
zla?=bzs+xEWcG0SSeZ^wQUQWX$>?V&WGV0>B0X$DGt`5?L}Ds9N;Zts6ThsspmSvb
zDCh1V*$`J(2D1cF1b08}LKRBhMIkEs!d|<kzJncBmOg4es~;MrC#RMrr@dLxX(+j*
z$!lkR%&yIq%VAJhN=pMDMt9U+ml!WOWll4h=S`{<^!kdT$P@`Dx&upDk3vUy&L?~J
zhI$w4PAV*+oenJT%DgpB%zR_sMeKfuTqzDFAi^71DlPb5h6Z(lSHw5GDW_^}hhkb1
zO&vgTsa5Y>Y;VpXE{^jkoR%i|rbYIa20L2*b2%VAX<1n_v0-SiH-O%_up*2NLHtM+
zIbs#?Lox)s_Gb`3yD?gWB(qXr$`;i<lVTHtxEQ?wEE^o_=}_jy%^SYH$YcsM^yg|K
zzIwQ8OUmfOyvnO|dXsqXNv}N;jJ3x#xv93a{sxIt)<K=Ooz))bku7d9|7=+^kmMXX
z!zaDGu?n63bb^icCN~a5fH2c_UB#IaeK0BEOb3*5!VR4<Dm_fv@|>3C;a{Z@WSdJn
zWNikkl<pO!N86DYl<Wm88W<!!qBdHpJahe=)^Vp*R`_926++EvdutsO`n+=RI&B9O
zw^5&)H6cChQCpeg!f8v+42`@2m02EBLWS*y`X$8QPPADP019iPTJ1(#Rr@LLG_JZB
z0EEx+i47fFox>c&-p}J{#p83uD?aF)t*P@8F+=hoWW4fTrzh@KZAP?wc=b_xTP_t2
zn2$(%q$41S=s4a4YF*m^Mg~VmDUb@FC|rYM!`O5>wywUDU4~6E_KoCR^yHFWw&J>V
zw!SW<b*q;+je;9_2Z?EC7H(`l=nO;DMz+g=74|Y)a$drg+z6`U^7f`*2Y+e>A<!sj
zC;Im~{X*YDP7t=6%61?@MkzCx;8}Td@-JpXtekt}z1lj8PnCP6TuWv?VDG3Ovvg#@
zX=o1P_}yc#ws$sECxJ=fav2tO*y)L{&f0Z~6+CY5s!zJ~()T!x#%~Zs)Zo}6YuTK&
zm{V4}xf0`gB&(hbBn~Kk7>vK9LFTv3F@xAGb@HflLXHkAt3aq@gu8~Ssy(~4A(Z`y
zb1K~9USZluq?4=o{JV2<IYER{i%u)am>5sWj=U##7K&k{i;Q+G(Oii7B5@5_F21k6
z4l#~>nMrxfD=ft&Eu1~ps(TfO5phnL(+*VT-X=4_Y1>d>5*?KB$BOg{LkDA(D!AdW
zZLE{aeHm4tl*A^3&*_iRQYt(o+K8uQ+8?xbAtCm*&_u(zABw)bTAVI2hlIE{Nt`WA
zf{tpIcFHuxJOUH(<phPq%`6|9KI7b#tB^3HI9au?p>o7>5GQw=f^-!;t{^u0wavWo
zg3B&@ZT`>t>oar=*~D<q8vu2f+x;x<fl5We(sX{Y11dpFs%e@_Iqjlz1VJ0jHN^oy
zL(X-fXk~CmShKyiz%+eq8%*pO2bL&Bu_*B#&=I1FZK~e}RVi-}8$|L-{<6ZI{`jRP
zH783HsI-is=H2I_J|e5yf#rrf){((k2BG29jBZ@9-pl1kYm<%SvfDZ^K0M^5F|%z6
z{)=v%w?MGy;+O+VNs2fVnUASi94dlRULKp?X`|&g;;Ra+BD&if7%6A6#fojbuMTfO
z^@_*~r6gqqaav^uS%eXfl~1lz64JFIT%4ZdsGN|_ap4V8br8MeEhrEZ<vFJ(@56u9
zv+;(jD7G~;1Sui1&V{o&6e;Jfm`Wj6+m(VoCMt6ppB!@<7eQtJM+fA*<~6ptjE>@V
z=pdQcBooYDrze>!Qwo1WKbAcy4itR+4vGhCX|SpEM`3Qtf#tp<v8S^I5q)bjPn@yJ
zU7Stez%CUh4nc>%*h)k!kKf64WKjbNLr?@Pf?u1gd}V!ySj-WpoqPr9SGIL(WY}$W
zEipx1xdR(Om#}v70qy;j{S6Snd(vqN{k?H2WN^}WoeWB%_dBh^3iIH@1EH=UINI$M
z*=t)HorX59nQbdq*$2v)w0CPiN0avoQ|%@lD~`v<exRXJW!49q1}})k%(G(M=`Vi1
zVT@fehnSbqPzZ|8)nAzrrE}4f9DP~0+wR#1a}~8lCY;M8uJ!L+v()Y=gDdr$r8GFO
zyyGT#wcXjEG%FIyR8nCuyigrgs^>#>ilf$wilYH3%dUwCDO@QGe)i!q`aPkVrj{0I
z4PJZ1W2Cw3yyGA~`$&$B>Zt=9l|w6SENdyaPet$_tYaliR>q8n!33~G8#0>`=2)RW
zl(TdG<E%-p3u{yScZPZePq(}BRXm35NT-!teSDzBQJZbMyWzG9*3sx>|Bnirl5JSf
zO7rKgd^}bItVHz4?0SY}a!(FoDarZHb&##Ntri<Qgx%X<PlkAHy%9FR0tljT`s2w{
zEINgyHYcpwefhMRZHpPIa<Yk)1Qv4<yf<;l-wd%kIK9j6ub;<HLUjTVKaioh#P~Dt
z#wu(dP3-h@WF+PwTJk)+6;H<l<y(^d#n}Wt3zMlLEGYJ|`pqTy596@|OMPwLRrBnJ
zn;UjUi~=<X9axTP)>dBj;i!dXKhhvAbNQuBhc*OPCp6vZPejA7WF}Jf%#Y?otXR5g
zy_f5yTXB{DSdkWLKrK57Pf>~`N(!n)3Tv8}lg_=lWH^a-i~V>riWE<G24yL3IC89<
z{=!7W+f)i#W+k0s_VG$C^5Ze7Dz74rf6(bEBiAk<HR+6fqQGRimxqr+!(L%BMXEe6
zGiINhwcr!_OED*LGJsk;D-_%{f~oN*%EPqgWP&&~ecTI-6}@qAl+8_tW&C7W76&k8
zopV~6_(c<w_ER}|M0LQ3X@o42Zlt7MQNP%CbI_apbotygR;22k{HyjP6$u43==x_0
za`#2C4GipIvuVysar%=dU~E!Q^UvmY&1|C((Lk#g;oZPAU2DcVeXR|)t$YzAg7`*D
zzd!!~+ZzA3?G?%JI-)5V{Z0p#+;w3%lXo=py(4Tc8j|)!2}H-IcR2l;JC3ZU$R?ao
zJZNWKzF~AMbj9AV-=EwQ@-xijD2Wlf>aNlrb*?K7ftpmFMXCF3cSA#w(Gms2w7;Hl
zdV+>D9a6DixwfZ%9*U0A?D6J%g~_X^Bj2cqVN(T3oP>d(Ml05#_GqM|khL^PrS7Qh
ztp*x=Mnn!$5-H60fKv?UADSdsr#oc(%Ii}>l706Pudoa@Pj0xs8M-`nd@jrYSS~t_
zRvoYdm3K^x_MAq1scejrs_q1P2MY{$686NX)#AW%bJ$48?)W6U%pR<ByK3Fym5ZIG
zcJg&Ca=s5Gu@lqnT<YB{u(K!@hnv^sS<yRKh`<Wx&jKzIV1k<MNTpC}hw-1&j=?H8
zRvIU}xE;Oki8tu^)f^>8k-G~Edyb?X2b)DHK-oG~YyPy1lG2WyXT%6n2at_D8NyyW
zR!2X592Xg_@_y}G(wQH&;{`@3Fz_9!lmlzj+b4&`#aTpr9jnn!)SZM<df5B>96$_1
zD!+0$l{uB^CJ!||kzWDNE1~ILdqiJyz@k?%V>(#;WEq6$Q1xezMoG`4bj3ua1kGKt
z?!9%1ovOnbr=X(v9)S4T#4I8ij-u0bt|XiVBZb8(5XR|=yM<SUUoa1LCQ0IoYp=Z8
zxz4AduF5<O!?%YE4CO;17c9M<USWBpIW`WS^GLZ5wKHg%R6qw-=1hieNflCik-Y3|
z!(>!&NJ}`XtHcI}a_jWRNRX!33?wq{v`5Qts!DxEf-_58X=lu`(=X!Q(jDrvbLGwY
zQUw5=lz*8GXs6R7Za3N??7Q8u$Lg#xlaiR40};ITKr~_YjDqHoRq=Dp;P*q~acBht
ziO2;8y2I($#@{cuy8V3l61h^-RsEm?i}j}JX*+GYGQ#VZIM@2_nQfD*g5Mb;9oo*<
zN#8(-;6SJC00KG6ZB*S)6`!7|%xL2z#yZT_4kVB-6sQckPMS=rd%W}z$BjH5F|dqZ
zBd13j-~tZ8{DnHI^*1@!IUAYhg;(mRJ<$w82h3=0k2|n<Wdv{p@hT=o`ou5RO<O79
zrRsnKENe-5EjiBJ*z3>OFBJ?czd|;AX}WU^&rW~*GB?y&c6-KtxgxM{VyG|d>-`Qa
zHbKr3IDbE+`YZQ6@glzR17G*gf#*JexS%M}?ea&#pj17WT3Gh0W!i-CDKD*=Z4`hT
zvtO%>cKyUK5-QEyw|fI)<R99v7i?uQLM58-aA3Lb;vGPT{EY^oDts@e-)XNfJhn0r
zLhL8)H!G6CjVg;`6pjUS(CN|a-7vV-eyiLhq^cA`V_9^1%3Q`N>qzqF?6)iNbo=@u
z;mm;qg>KXOv)^fEuZy{vQ@h9?aDZBFi@Sto{BC8YnYvZ}#?p)h8uR4b6|c^j#)OI8
z_xI}8NKIg%vM<WcaolHn#1FInfUE=i?Dy;DxZ;L2P9w3n7WH7j{-D94WZSgiRhVR4
z9W^I{{KIAqM~YF%En*O&HdeLMpBj=>GpY^RA5~TzR@gL@A&|lv=Eg~JSCTZGSQ~>*
zyv!H=aYJ0E!ve3u9dk*#=@0)(J~>dt6cfQ|e^Q?O<Rzg4lxt6zhj^v}t}^}WPYW`n
z8YobWde085+SNkXVW)-2CG5zMD1DA=`?JJqi=gHf(d;xYWUk{zq(L$W$$f%CCA%M%
z5+n-^FGYT`Kd(|qo60gA`$g6nphV_P9S$_2Ce~jxGsqrBqljNkd=igsJW}?TrC;Bt
z3T$#3<aIcpo&Lai*svL#m>Ojn@OA%HrRx<`PNGiRqyve2Q)rZ)85VU!!oN;(zIy)U
zmpj*S7feykMb?P$f~aVJQ{Pnr$~fYev0&D4dV;vwy`<aD9De)```ZTBV!_%0D18!)
zm|2Qocv&d8EB${2hG<t(wj!>xzpG$r`6OAD6lTEj9-Vxl<*#Z$e_z)?53}7Sr<v#m
z|HHDZ{6nthnQeVoh@4ig7|hpWo9rKxRjH&&lz5|Erb9K#6t2`{!Af$me<}lzvXb80
z?G8-r4&ivSs%_an7c}MGlEPf<Q(k){2otx?0#Wv$R^-1FWJN1!3BG$WqnW~LEisw9
za%#|B<zzXe$S#n%%l@^DD>Kz2pTa9F4Ypwqv45-gvD%9l!-0w2;S3%^h5Gk$#}rFZ
zu@3y&fu*otRH;YpKPp-3?W_O=?1d=LsEIrA+Aq_w2olqC-D?rXcG`a?Q(ZwZ85uTo
zT4%=r=F|;z*eMljnwT9N$*e7>oy4BGg1DEoe79q~{#4U*6vEIMRf>=$n%3#jzOZ!d
zV*7M1ytN3k<4!}f#$lRLC42Un`tmihNO1#z5C=HC;j{JoH?+w3kr4(+5Ms)}A>7&L
zD&}A%at`1sc3?qI;_W)#HwvrMgZ)9Z;q3FZA01PN<-pRQh}lDl>5ua)xBdScd+iI&
zw@0iT!br);b9zMVH;z(2gO%{b{A9}Dk}F8!674x#5~U*;(3hHrsS0A2M}gQOK6<p}
zVRM+YFXxj;dD=LDdEaRlo}q1TAMP4r+CuyJN_mHvq>gGlJiXg1OnQvuSCx(WYPnBn
zBkw>0sRN6rir-bmIG4+R7qn?JRj|X?ot5hJm+@pk&Z)>bkS*?OO^>P>$}CfkFLQd5
zGmwlnK5lk@F<`A~t#TTHDG6ipim=PrQ|=FZOt$b_l<V{Rr%&5c?{D_xL3R(C4U%=_
z$6?s*)9$ZyJnO|sU&a5{lPSJWzrS)OrA}zK5jr15v==)+>GVtPZne9FbUdTHAbjjF
z3Kg{vf^&NEE0<9aYto*1e__81XKSwF3<n(e<8*0eNNH0lnY07n@U!kO%z{m&-)W_l
z@4}mJv1ivVTpa^WtNCsev9NMI=l)o8)Pk8D8*H_mcB-M_xsy*&N|xgIa-(B<kyxKw
zzC3c#=~vBJNh`Rt6rJd`U%w>7KdOw(DQjuZyFV84n&t0wE{l+xc?qT5E(i@zusuI-
zT}Wj?E=BHr7!&LT6=~}v!8=WLDN@p2STLo+0+|pl1K>d{p!CGMG^{G%D@h0!q!_n-
zL!yfTpPM=1wBl}=Xt0w9LzQSRwM8LRFRI8Lh5tA$&EoY`9_X>KUtI9`|1tO8QF>O_
z{qIii#U<_)V;iuIB_x__BrzZv31mrtyEB@3C5<(j8P8}WgiW#ycA8^A5Xb2e>b-z^
zBN0PL@4ffld(ZuR_ul837bkbE`}^ardtqtTbM|@O^PIBJKD!@n2jPPpqU$7q0~Ke9
z&C_|STM>OicBN+*Os7K##qiz0YV*k9JkA%bAN=Zbs*|#C2`x?pd#!7}5(odcdg(Id
zqol7rw|=i}B=y7zB)F*I_e!@$eIlpgS#VOV&P76$Tp<URF3gMMs!EKGqIsX!O?5=w
zpd(6iXDQ&F`WTTWS7NBFGNDSb=eO|(v!nSt<iKJnG|J}Z)sIgvXh?tkHFKOo$~wz9
zu?ENNh4rc3h)(@}z2}^oXjbNM_)R7}S^j2#e^D)BD!Mdn(qHW3Em6btquAlRmET*s
zWWmCv1NP#?*rmNq&ZE#0rggQwq+4sba4^T5;=oE}*Bc|x$Z*eIT5n~p95~xchrNTz
zDS{42h4ivEBf-R_cX^+4rBk06!zzcuBK9#^t?cF5-RZoId&(&%uYq~)qP?P<l0$J`
z6E)m7*i%k@U~Jx4BYx*mdu4|so4^vMn2QxcOdsbAz9{L=WmS)^Bveu|(3T}#&frdc
z&L=rNxXj32)ggr3JktLcSPW07ibMTV-Cb06d23x7rq~2xgSe58iBpOL>SF-q(ghK`
zV6QHiwyCZdIPlbc4y@ruw=vR%Zx5`~X?snZ;MqtwzJO@x)Z~`IZwr;@YpZdsI1Z5V
z0rCb*X<q7yGj>J8h2^1=7@2S&p=TShD;s_flBtD4wol;s^hUd?bwnSh3z-c<NnBe^
z|E@`Y``2}d9oc=sDE(4?<r2-X-9Y9pI&YQGy&ueYeOs1Ax-5zOp>5ko7SmE+y&-?H
zcx0Hv$SEcFI6gkKMdah^ZY~sUE%oPK2d3p#0<^d^`;ED0h=&|;o?{VMy1vEK{pOLy
z(&pq=d+TMFJCC(qodXRY=;F_8Zig{=Lh|$dkia0w-fAR$l@^j;D#6Y*ZQ92tS38A5
z9Oax+k4P-q<{lff4w3FcZl|V<ZS&^a{5B7=^$_as0J3)eyld^+>Up~2>aU2mY%&?n
zLA_ok2RYKwb#3u)LY7_c6dQle6{mw+?E1u+%U<X8boV`WLpFegX>8aj<dm}e#OdXt
z%NZ_N<{KXf$3v0`5bJOSep^^6u5_R=c<s~h$-K13-qa9w&B$6ZIly}W$=)b~rcQx3
z=PQ*JgwPfTQv8Jk>f|Ys>BRWM87Xe;l)a_VBtnu0ol<<9D;zjWL)Bq#ZJ$~=I692y
zumg~6aOaYh_O@CTO(hxo;x7?=FwE5C8bT0Sj*Pe0+wq1ZxZvfG2A!HDL~EvpWR(8e
z`n56)%9j<>gEv@W=q_wE=WRj$4%={e&N^jLd3*^;HX=tBTi9A+k~2%$0S8tHEZ<^e
zDE4F@MYl!y8(p{}2S~|iUNHLd1bbg2ecWzpqet+UM0y4f6Hd`f`<<F@i)fF>9dvOe
zDYOao#6V8rwbx(oJc{%zyUT8F2)9gugif(`fx>5$+wL*DCD$Tch2zFn03sY*cQuVM
zZgGcka)OtI1)~u^R05hqh=bHelj6k01d#pPmUJ+MJ?)v%P4>VF*U5`0<KsK(B^}(N
z2*<Im0n5`&!LoL1y&+ym$MprQYrt|<7fHL=v!%1|z=?!r)n{Q!-<7kotPO*m564p8
zK?h(5!5KIu>eJe#l|g&KGT)X>lO09wR&XQm@dcwZx<)wn-0ihwDcfcAS=?9R)Z~)I
z=8!D|(H%ZS!$sF;%iA8rV`^PQ^a05USU|EZduJmwQ}F*3a^w@^WbvA`6<IkwMsi1|
zAXa5&+k(Zmvc8%j(K@&!C}aNQ+u9hdzTf`bPit4TK8)cX<CNm84wI-fVW4+1F?Y0@
z$fDM+ufrq5Ta8-P-K%G%y=2CA*`3w5@O!6MN*))1FuZl-(spn4nqP5y`kVeYn(L!>
zSHq%B<3n&f`yGG?2T9rNU5WJ{nslC$O3t1jAp1@zGj3Ct_S(C<UdVQsjJ;S@#j2jg
z@rC-pX)e_ef=I7t?`a6a4HttXNka7nD9?c@qH5opJ8ewSb|mVe7)@wb%kp+VQAO{|
z4vwyKo)X(uD}5-jrFf^-Y-SW&@JWeZ-fG0P3V*~RXg1H-2kIkO7ejDRB9IQotTVSr
zQ3vdU?XVvUl7=rH@>mBXhG}Gc?W6(^?CTote0FRsvxM;C)N8u$Sg_Ov6HPw!&S~Av
zOi2}^L&LVZ4V}|T>Rk>72Ub|5I}|cDAk;}V)W#5|5!vPrIIu+Zp+OS%4cc&#VC2IX
zz?1{g8+YH$OrQa`Ubdo0h)Hhp;Yzfg7mS`OEvU@*`)y5qjn$5o=dLgZre(bM-g_^<
z+ScYcKKQ}cJCD(4k)L$)kwOVJ&DP~wb2G9inNv)33qC@Kq>Jy<_5h08NYy8zXEb`|
z=#n!2ytjHyh?KDMiWHlwO(TWuhw^h=sl|2HlTJ}nef^DdY<<l-7yI#;ltW^Zjds|i
zyb=8t8)^!-rH@Jj*@lM5E5_G3g&2)CpR@;x%7RAXG=@xx8hWgF0M$#0)Vzr_=938K
z$(Hd}%}J`>rB+Ha1X~RDW4t<;F!_Q=8Lylj)To0qTv3ecZhC}FBwI7H?a2p`ARlCA
z9=LSDfhH@xZhV}pDnFdX@lh^~#G*t|(}Y6QCL2C+5sWw;$}tDI=OZbwO?9y3-CzkE
zm_9)21_Bd?Y-3I_W5!PAluC%n*x+P8mGEM^#-8UE4Jn%pEU6S-*fhM=@5W*(AtwT5
zkSumV>ag?MrdjGRN))6}3{ZSb!bbS}=?sAdIA@!?exHdTH3UWn7L(8Tgt<^wxvHdB
zrcxl?o%?BUh@&pAj|ZajLk2n9(k)#eV#w9Ow$;*7pJZ>z)R^7VI)+{w;sjf&aC2pB
zrKT`IEQ97EvedHel0E-O&gROIwaTYRDOa5hJT;91nLz4~b}giz`dR%JOoWW9Q%=2B
zc>ayo*}bh#%Wu2IDRdYdiXxFA=}UycaCl!|Q>Mc}H~7+D2bSAsCDF$A4OtojuSR-K
zN$k|!3(WH~WK_obLm{vzx$s2mk99BttXZMsoN2SLh=xTOz1{9>CqK?L!45f)SYcf_
zQO4PvZ@WLwJUM7at~Cdsaj%=$WFODRjp#*rOrm)UC(7useWDud-^W4roSIy~J5hCg
zV@(zd#2F$IBwuu`eEiU<uTfqMztO{xfN#nMNJ0wXp`hN}wfJ`X=5{(SA6Fv3Ne7^r
zd)IRNmiEn!>&Bfzv?<dD*2SRU-`clm%E`*&l<GoU%UP#B7ibaAcJ|5EE69_~hOn~^
zEDu%P$3QN@p(*>ehMl4O!O<XuGf2O8HDO&=D2t*}WI6g&`w%Li5vP<iW);TAjrKr&
zSCm63MbxtfYigV$GY9R#dQT#NG`$#_y}_i15D%pj{Grz8AujAfUBXG)fmV)BZ(O%o
zPS{(!jeyfOzdS9%3&D^Z!*K`-^+k$w@o7>uXg(tD!#VAXCbvv%bP8Ij#oSu*NJEz4
zaQk)2wIXK`ICOaJvv2RdDxoS8VFk@;uK3i)Z!~!de-nT9>7*m87B5-pJSRdDlOu-p
zJL>&7LwWxK)=FSnRk(yB(6Y}oBw5IXhZG9&GEbML^*cLQQqHsJ*MJqBAuui}1Be4k
zFgY@Wd{>V81}<jhQHW2V9n|H!`w*^axp7YbLuB(i=B-{n5z=;?ruB3S4&9}FPj-Fb
z9XH>Mn=+-`Qws>8|K93mN(3TVXS~6*I`FiJnkVf0+Meg&Y*z_rCQdpvYHo0?+)uwh
zaS-D-dEthgVoc0d2<}#lh4RGwfo66XoE%z5##)j-%Qy=-mtc5$!vr$$wl;rk@Z>Ze
zcVHs@cPOjfqxn2NJUHw;|4%B)fg*kHeiVV*t3D;smWx~mj+XX>Q=i&VqEJz>%dfLR
z$PZ0o+F_DyR)4tMC>U)Sly>HN-8lC~ug4Wck?-nua%e<@y9yn<Q<H)v>KSNc&vs`6
zZ@uib&SMe{Y~gp?p0=0QjuPJB6k~wlo5unx6-DlA9bji{Z)=#K*`Q_sG@E24g#BN;
z#rC!KZW&p<ddex*E|g_nEy+s^_Q>|PWf)vd%+i!oF0Tp69N-75ANv)&kOeOZ(Se$r
zaDOgp3@iuq9t@{;utuN|E9rG4ljfDzxdg3VI#H)b3c&B*8i(==vcZHGl`^ve3#Iqi
zAWoeRq@#u$2c<Gm%D&va;-;l*?UKnm?+Rm;)t9`R-BcXx$at{(PEDE{xY}WId}?T%
ztj2b<+XoQf*ut<vGiOoxP?OsV1*08ng?aPlIt9(d5SWFm{#do31HO{s(^D8Pr>4kA
zI^GeoWRK?<=iWHyTIady<&r1AjAM4Z&GJe(Nv9O+%8Hg8APWLI5IfOU2=-MTj)(C2
z1{PQqJDfR_hif1IZ6_Np+%k<{-JnyR-6m)W$Lv)0ZiSrgc^1ij_wrSCx;{3`$er^B
z%aBh6b)2v>t@BSlgeXoaI-e1L$!&)*%+9u&G#ay@;eP|Eq9&$98d2ru!{lq{TDKRD
z;}5vjDaSiPRo$y`2v^bb+1ZcW`+Dawuwj&3+IFG6D~2^T#!&~Bn=kZOY5D5Yblb)H
zjhGNwd5>M%>A+$vS|pizGR1{5&^*z-eVnbAffc}#_DxQYO5gN@^~FxK7Ms=Q%0_*R
zh-HS)4Bp*-s9^b<G%Wy&&D-aw1=<{vAtNF$a`RrQUrNH0%}dE58ceDV9hgdh|74A6
zFIhZc>G4{S!r=+dpV^u+cf!TWfVszJ+DPEGrN0tWpxu=wtyA9W`)+sbAE`ant-LIW
z!JP0-rGTJD+|Ii3%@RB8him_f&ymzIn-E;f@(iQ0^KQ{J@sWK2+H`O>%=GER_DFJj
zioZJ@i0I8q{3q;3Qq0p51|KSM;+O-~LQeL2@+g|JAFbhwW0l4xFJ{TzO`>(Dz7{gP
z!9naN?8ove<C{h%o#*tBB0N-nN|nL*$7@#D9jFWP_JdeUr$(D_(YkT_i8gf-+!>iR
zjp)LDr=~*w<>D9W6bRu@=7hr7;T|>W6m?c1c!f@Jt^HJPRxV$b$Vw@h&6_95)NViB
zFqm}fWbbs!h5sUxnjFY6xOCcnreRwshhl$wgDD-k96&DMjQ_K(lT*6B3+Fku@{|LO
zcC{O1SynK3V)4&)n5KI^QkB?eba3ips&xQCJhi{==Nn5qMI`7rw=DJ?kj6!b$A~g5
z|3Wv3v-zUppmb`2Du}sNrdO0fj4+Zt2x7n3zQ;aM4lXRB4lE}P12$<=f2p0?3sx*y
zxddqtKvF=+bGV@g+pL?pe>vCnZCv%vqp&cNu>4AWB&L8)C~I1S<xW64L6`KaU1!wU
z=%xdUk7;wO>+*`o0O?=Ll@_Ta?wnFgoKin=@sBRKv|n$_fT6g#%Ef8x)ab0!ej(Gs
zZ{&bi3@ja3z|FN3W6pHE=fz%fWbfQU?Kj)_>5LMPit|gnrl$Q8ra#rR$<2jr&wi`T
z>=0=KC!AvRId8ibfxKY9o%1%c?bb!kbF&{QANB_Oo%%r}4dk1+ggCHV;qcrFD%(#i
z?04%ojb%B5!n(m^U6R%52*Szl)g;JN$#s+zlTX0=%znRp7dB4r%6JBQgNd+&o&r71
zA9R?i-9YYp(s;HZ<Ms|g`oqMKd+r%=9`l?Y8YhJWh-`n9@7^@BR+nx#MNx_Q^KP_1
zt`0~iphHGiNdwCsE?Bk7{v@Xj#(2zm&S}H58}@a|{<JL^2d^&4;!;!EfRA))^ocDS
z#_Z4P6XJm*4>Goq4#tj=YLENN?9bb2RnfAYjWcne8s;sb|C~A3++crEE$8_bo&w%r
z@~68G>_Vq!f7z%L=0ddbPFd(dFc~>;)}hmeE&HqN@VfDy^H|uUu->!={B=Wy!uUwR
z0})9HbGK8kZY^Iiu+sjf-W97Ok7V5!lfJZZ0D7x$L`oj9lE|g{!41m(wr^+3Lqxv1
zxecgEy}>O5w+*aVuxj8Y`@3A)cPfPrnjVVL_oN6?iEg*Q?^a1?V{RT<ea<*F@pkI!
z;{Hhchk}Lg5hzSTBX2P21K5%}`t2VROP>7r%#JCiplv?@8HIoOr>wgC{U5pa%}yb)
z@lL6Gm1zB+Yko4&KqpfK&O?7h7|uKOB9p6g(DpBd7da$}9lCA^Z8g)S>q?XG@Rk-}
z5x241zt(_d(UAX{4p%izZp~a}?w+%Mt6szibuw~7rw$|ti&LL-eOqty5J|7;_`i44
zb(A9&ks0y`EU1KV7`@D_m{Gm|kNQmu(<LCmfhC;}Eqq&GsXPC(>-nkAa<FeTnl8jH
zP*38Av_SurV_I_afb*zjnAx^kYPJ8au4Tk#T9xFX!fbQI9(3xpN0%*FVfJ9m-Zbfv
zoM&zS$sU*F+UGo&7^3+RE;3u*2ZQ|#EuejFy{EyTn?skh+ks_oFrBZr&wH?)H*&-h
zISofNY$(*^dy9}^?DHRNw3#4g6ip5wkQAXsQDkVRFWMJ07>ANPO4vdUOw<Wm+rBaV
zu`jGI!h}l;r6_<{=L1en5;B-7R9U0;McEz`hbK^PSQ%3sSau86W6Hj`8(>QH%1sZf
z&TDzZNN=kxtQ;d>(mpUaN-{Id*#MHJ&YM5izO)stoiooV6f%e{7JE_-eD-DCCvhCj
z#r6g(0|v8r-oCtAkfbM_o1)WMFI0!rB*PJRk=j?Z24M*aK<C8+4lF*OsIOz7ef}#S
z4D`5u?mXu)r(z8jH2bPVmYHqCL(X%fAX;}FS@zYf{<1ea&rzSTh0+|HvS+p5M7R-@
z4__u=-C<j|oiLK42y9%8#sT>3)(*~yaw*V32`nkg_-X~`z{5VLeR=U>=Q#$Jm@5uk
z%!Wj;=jOn!yJ4>L7@w7Y0q89!Ezt{(vCzQU*HnFE7-3SSY|bc<u#h?R@sW)Dm9%$)
z&K*WDe_nos3yz(~nxh~QFu5n~`Q3PD^8)l!X6kdsZ&gap7u07$+d~KCnaj8?opoxo
zhe^9@FRagKqvMy4hvgmzmPCvQg<)NnGi5Jovo?&&%P`Sm4x~1iGCXLdK4vd&0~e!A
zV<(w)p$M-0tnnVFK1Itka-bKI?M{XZiT0u}B4;nDx72S4F}fKv(P5W|!3rl{e6XFj
zG={Z=i@h|5jcSl>>6Bu2bx<mdq;sXRmp%A&O3Z_+R}*{Tlw$~K`q%qM?ByLy?zuvP
zgWh0L8*9B4SN4jWqvgFxWpN637eF~9sNv;FatgWZmDQOhVMaDHSem2W)zF;uEySYo
z8)As<vK$Uq3n{$<r4A2BsRD=mguN<B!6I@$ZEy;CmVn-s<uUp#rb<-h@(yX_W*9~f
zYC6Wjsn1!!h9GSSf9=)P?cCpN0CC<rH=LRj1LM8{=wMt8^O{DR=3jrUQ>Y&6R-IH*
zzqZyi<FV4(CjFIEGIChU#(ArLc|;mRIEM3fMb{M$P90%7J)$er<W44*di~_|#74Wa
zQ7%-n$deQk69N%+pdqZFiG9qj>gG)O_S=26+K1A_9ChkN0xNdiUe^e-vc`;Z84&;>
z-;#k9_WIU~-iKt&mFr>qZ1`kLewQT$CBMC)^-p@RDesEGPTJMo=!FqF>&`i_Slm>-
z=$w?)rAUMb*S9y;Z)9#q;f+!KF$WgE5%S6jN};UjW){2O=2ZXW(Im=G+cl%q=!Bci
zMK!mYCwDJfF;$kp@)#ty?J#*3?3xENbq}qxHevjv_aPa8+lu(wgJn%TgcMx}>dr4>
zVN%yB`+<-Zb(hU+wD__&InRL&$-IR(md!7iD6>Q#Nglv*j>L5q>Zb}yih`4NZT$w3
zOXND{2$&eZc{j|NW7k!^=vBsDAMysv)UB|Mv~zv+de#m7AXBmrB_uUErw2!+8@}Pe
z(g5rJ9W4Cvw{n1arXtMYOmkz0BD#-6&Fim3AZlXJsbI<lds8=PNhe}S9`#w6%q>A4
z?akSZUJo0yQ&2;QD|hQ=5Gd3ZdrLRc^WagafE<{52I<6kfRvkk$Rksmt~<!ZOm|P}
ztAEnbh*})6x7BE3Te8GTM-xhaZ#B8l=|?Xv^n|^=n@sWKVy6+BAdLH+nmC+#6ZW-j
z)>bUJW$`MfnC#H}>#ny2NhSHw$xC~jQs6*o|8-3Uotvep3#&fQky0N(d87yfN1gh>
z$i~;FszHf27X5EE$gCHgnmElw&06?CO@Wj-OdHr~H`VW(S}7cuVI1~3H9C$^rI9KV
z<J3Sm_1Zgmu4@|E@e&i9`rLT8a4^V!@a9C~r8qb_&#^8kGK$LWwp-eu!>K|>4%q=a
z%b`;rzmdjWdTS{>7I*6_r=U2T9hle(dD}9eCAmHm8%CVRn6<P8RQO_u7w_mky<P0L
z&NVfr^FS^?PJPZDGzxBw;Osz|aG}5nhkYeKwp-hX(PT=a=m69c?i5or!qWOkIAt=j
zW1Lcfk&rV@F!kEx6T`htwyZ6qy2s=y%8(t$K}=1}g2<(Woa1&|!J2OFJ`}mYVz8+i
zPU5P-ok#8VTF9PGLoD2sqs3gjAhWi-gApR7-9gM_gK5pA!F(K+^_?{_BG_{0n`MzS
zN|8x2*g0?I(YJgM&z*6U#w*%{vexOisjy#ob^acf&x$-kowu~NuDfQwt!#K1yFe%)
zS6^j@pr-bc{M+)Z7XMc2cUx7T%7_#-->hZ~i!)kyaKd>@lg||_Q+7u#+sw9&%#Tyb
zZM<^%61%hUWcUoPhfq?^%3DSz?XFgsckMi<5c3~I9R$g2<*b>#t07xzQt6Pg%IR1O
z#B=InJ!GX22ZXz(y}Jz`tkI2ah_<40-~%;!n|sN&plplK0DDhcJq{01TmCw*=uep8
zVnt8VD*^}Yz4f;IsKnqL#h0tW^6-W!Vkf;X$s5~O$pf8Y%sVG2x*f=9Z~>W7Rl|IL
zeNY5cUL_&Y2Pf2O;t1iFj|DMB<g^d8!SwF#4G~m>CdGlpcM9#o03~!T1&Mtykqd_n
zl=krpASDro+2reL7p}p;H|&&R9vbP4T?7k!&IY@h*(*u)p@U9M3M!-6Q*Cw4m+B<U
zvNyfKb_XeifxtUrs5;J$f^h4Im)jyvNIajsw8N?I+Z98&2@WTAJ^27!-UFzvDCc)E
zP7YhI+RIS9%*B<@$ZVGmJN5GJdHw5bP5T|Br00}s<>jiH9)v;oZEeGo*(G#fwXCy=
zZnfP|0xC0rt!urOWEIm5M7>09eg-4#Nd3N!mXJ1^y-sd7)N98~3~pREWgkk4Nq$N$
z#+5z-KrMDA;7Ybk&ysic+xkXg2dCFYKs$h3&LEV5sNk9uqN8nal$42X56jeG;)yET
zfNkieHHmCoPt`|osV_YW^5x+0l#M-OgpxUwkTgSm6|zKqbg6<%rSGQq$_O-Gzf`EH
zTnuE5m-D`MIT?qX`b1oWSOV=Q+A=Pjo?P#ga-%a(BUq@!ynQ&S)|FRYezo%~Vu|yE
zP3Geof@65p;*94Fyi=_M3eFZU0Wzk+GM3A!PtwC=AA-Q(Rb?9s2IA0N6wn+6z;c-&
zM+!oekKF8iABSzaeuJ(?yas#!qywv^m7J|(M!!7A<RSPGdAq58qhlyT`kruLLY9#Y
ztBLvB+%2{83UIE4>oO`frzSN3%WwrG{~OA#yK`n1<1*+x_u)vXF;IXUwk?ffldavm
z!jhFM?VdLG%P?s;rCgX5+@3RPAE^d~*Sda94knUOx&6E~IB(uH_R;La%(go=IL|S9
zu`MDRazfSomXEvL+rBWR`$@bjhL~nw-_1=(-{`e3I<OoS^c4h)U{!I7`|TUr+k7RA
z)aaXm#oIcN6?vfVr@Z0qNF(%&d!(JR@Mytp1NO1%bcP*r+`{VgUD)i@Cw<~n-(>gI
zCo(`lh5-psFjP)W3_eX&c~qnlP@KgXyT3#-Eq$?%%&O9_q^?eTOl~CU_zl1A{32Qm
zu~N$VaJJQc65oH&KAuxBgV!?IZgtZj9+@>QsZo!^$3D?5OIT8nra4xE=L}iGsgG&X
zSt}t|3x;xpN&d$AWoRw5f58EbPi78DMy;OXPuj*%1oWBg?VGA)Z9iR@onGqaS=g>d
zifA^E%=JmHzq#ve^dmTjP*-a~N#sU-k}3%5@F=#cZ+Y<P%UMe%=wnQch>Ql2VK`Fr
zHHAG9yUx3@_qCg)+d7Q1-nVv3dJcM_t2`CdD8&jjLG<OVb;LfII5R%6#(7Lc<73jl
zDu0G%=c0XE)o9Ixf}Mx9QzOyh$*pwhAvlsxb#p64e!5Uet*S&V@-(Kt?^1=l=ZT~$
z4>YEd$!uJ@xDPpeoI?{A=;yqZ9YjYb#!D#%a(W+Z(}zE?=&J)r@2KA0gSxFFXzoKb
zawXw7<CJRFq`#Jn&QaUi4OGdVK;Dq<h_rBOf*Ri~$9H?UL&$>>#dT=rhyyExoAv(<
zZJa$)HA?KI8zz+jPssO#Q<EnH*Lb1Fr<YUq?Ong{+v06`Z_UD_Itv|*eY*7lCNZyz
z-RZzW+R>3K4MV?Qhmt7r4|eH0+C0WWH(I7t4b+shBsod(-g5Pp!+G-H!oaiknbzy2
z<CtVNk<P2<1=D-_4vW3VzB8%lLc#{Uo}WC(x%6G#Oeu#rTs(i{s(Pm;srABe68`Q+
zlh+UToI<Qu$^mg1rF~CF0ehq!-sqI$9Y~!{NfCDY-a^UHuCpn1@1{hY3iEgBMK!Oz
zX0Cl-KHjwCPUo?(Lpc?tTRU#wpW{)8uJc^I5kJBhxGMJtTJJ$R(y3GYpA<-r47RN`
z4kBVhhgzKQ28+91Vvlh}d-oAc3VeixcVKy9l1~pk$~f6lx94kEJbDPP1opr>7A(Y6
zSr*RBoU$E_0bDaWJwAzG>p-H5````acCw$(Ko{^o-q~h$luIbk@=&h9=*9xgtZcQq
z19Ml^2<K#&I1Jwqc;eL5$(h!<yx|^(9l-`&yX~&wX)PsiqJ?*0!hkJfgS!9S_S7q)
z#kv6NK2#{ct`f~b#RC7O*;9j3gyhN7Z!Jmku(vuQVuI#XW)}XQj%*zVKzr&FZGsiW
zQGiooU%}WOPsxi-PSy>U(z@c0cG&)Qk}gtud`g8j=m)fQn92hUi|<^r+$q%Yr&zvZ
z&)X3~q_G`rU4dtlBb+EUu$X(1UnK_YuEDfJ4Y%Xg+B{I0gy77S>5=;E4z1<!yvq(3
z3=WSy!m)?A4lH_=E=zgLY9-?UKv{4k2Z?)z!u2TSbj5V;#)`k+juuP;lAP$cdBB0i
z&bncq9cv#5mR=`_vbJ|0aO%6gB8#aQdY#-S?XgCZV0_eYs0zXiS@B&lMZA@}Vz`GN
zY$U&ZJimV3yz8CE4iV)he44|wGS@LNXYF{K5+*nl)$B)L(W4+V^p%{o;2(#voyhg%
zZUg8Zg_^k?;IOER?9iP6W+&S|TFy0IAPD7pKa$2GLTFu%@HddxPPMt^bQY2#d;2U*
zynSGSov!z=LP9+w6Q=_Uo_h?lv7M<8C1;cR)utLG%-yNa@nQUUlrAX0lVqX8cD5a+
zw~om;Jmml+p1$o3b}lKP@W6RatPIwvpD#UMHH}f|P9dizFel7^?0BN+?Lx5-53=MB
zz<yCK>=Mz#-S(;b?P3ikjcmE_L1U;_<YB9CdLapRiN#3(p6F&vo<QQHk2o;R*8AT7
zzV}~kKUgn>OA1&9d*#4lsY*8owu+<a_(c1muA|(&4Yr=fSO`qYD6RmQv`cLsr)0Fp
zm;*pxBqi}E=|elZp{xp&`X(mh$!EInKLIn6VZm&#F!FFWmv{B0zzHJT$8$c_mLarN
zIbBImT1{#gB}~A5!$v%2Kb-w1>%_aAXLWaG+q@gEwWqtnb@QA;^cHCtxk5bR4*QYX
zr^-$sUjC@4vQ9gmMfjO?7#ZG5q~SMuOWEc^8E8M+W+0B1$X@2a;%!MB&HkMW`eR)W
zxK@=GxWn8VOyaDRygF%ryxwC=!qe)a<a6F&BECw$h`!Q(qUO3(`Oq4q!&%nbsZXSg
z6C#~TezJ|FaH7la=F|k{XnutmghKOI*@$F;IbuK6t!$T4ochiw*r^XJnnd^h)1SrC
zNt%(iF~q~>4dJ|1s*dS3TvgVypGiOj1uBHa0jSL<@5ikBv)!!7<c|sfcNkzv50KcT
z?ULJ9SX1ohy1Blfx1}%f4u+OU79}Nu`uW*WOO`_2EBA;aoVdc;gHC<Sp>~RnKK6_K
zLcwG+kT)UrUk4^uTOr6?<@k$<c#&k+Da6EvHSEc~{7AaQy6zAs@Lx(=LZUp>4#Z$k
zu7!}0m-HwXB}v^c*XPwi@wjXP2d0Udd*j?|>{nV>Zq+?3l%lm5E&~vnX&B0*!Z~ff
zT1(Z3I9bC4;YGq<!seW}v{d87&a9qR2;Q&dL3RDy8=S{JG|L0<*Sk%SZaj)ii$lq&
ziMNv*OR3;Ds*mcI+>w-juJ1x_r#_ZB$%(dR83gS&JMI;W`=nE>u4IzZ<Ck{XZ#4#z
zT!TaFiK=rT)tsyy%_j6X@BVhK+vZ7}iJgMx^c`^N_B&m}`iZ!e6Amo-{wZXIDf`_T
zMEKJ4$^j-%)2)0n(n?UDbfHfIXKxcr^?TJg=N*=g(2_f_ShjTXX?_~gQN&$L_WRv5
z!h18dr!MVqVDTGqiJ4YU$!KJw6xVXt{-DhS>OAt>a<BtQpO1yT!D;f>{b74!O%H{s
zLXUwYkFp6P9AVJ7?EO*KU11U7aSX1(WXD$A6B$D6k8}PA2^w>rYhjdPUI8P!?N16O
z&O`axk@g)}l7!o*H*!f{&;GQ=6b5kingk(j1}XDM8|c&ryEgAS`?LHW?(%m#&)IG6
zOH%$puJiu$b{H%j+2j-w3qzVrQt%h8KTMexjq4;FSWXcBYSWWz?Jt{%J~GC=-~{g~
zW)q+gMSgDfS2=@YgPTThewV`okY`|<)WeZN(EhsnVEDL88|1+9{2J|z*x$4liZ|PV
z+8tP-QRfpxUuDhs+qV4j?qUt3p3=gn57{24zQ&CC69h%2Z-3X}@D?z_t2i<^x<yGm
z|GrTYt}GmNO7Z*fQlNcB$^ApY6h<BXH+UQXiyo$LN1CD}XC1JAtRCuE*Imr1mX1_6
zFmxxJ`c#0dAuT3B5pgHYKQ){jT#MFz3f2xtF7C#`4XfprxWWE8mw>yk*E-K4#UoSl
zJF<VNZ^}leR0iG6(Mjc;dM)0xw7CD;MguDqto)_34lG8KAv#cF3}TX5cG<tx+uG_%
zk4lCS2PT>@JU+5o&L98YEr~RJFn7(Fhi7T+sISJwHcEZXzU2%$X8+L|=LA-$4SyY&
zaGm7C_MbJzp)iS1%9UYzD3TOU5Bsk~Bs6oqQLn<xwrl6ixBt!&&z(DWj`N%}j2m&F
z>h@4uVx`NAMRQ=$M-GBc`5wN)_Bjv5L{H%J11+HxogmA)0*2-KVW0a@trU@d<5&zK
z(SV{kc_uUh3EfAa`Mhd?Ykgy-uyi{xttE~{BWt$U=Vyt=&th_>q?H3%d-etOTd5nQ
zb0ph=1JiFKZm4Hp_)rqZ!~>l)uv?*k6_MWM)Q1)+j(vu%Bxm_7DM4$F0@1#x-iou6
zW$Drpuxy#?LSN>sB;K?vzfsI(`Wt#Q#CiMTnuDy-9xautQ=|33OLqixw=a39{K=Gj
zY#iX&G&2n}EqRgbT~3-`+NLf{Ea7V4z+$4MKVZA_SEdm*fPGnYPpY!8nZw!SfP&w4
z>TAWpQ3fu9sV0m|<{v6@xNLlE61RTJ^`_jHG-(+MzSq9up}^IZD;7DA@si?PbU!80
zE_CE9zI<O<HKP5ZQWBHH0PWNy8C*X;Wnb0CyJ37dUGjk>DdQj^sz=hGeRX?_tn-=D
z42Te58kIawn$N0_i3G@fj{BaA1hg6@tKnAZ_?yhpv)lN9?E&rfa@Gd_@6==mr%2=X
zoW#W5*!pqjIo4h{t=lNNt6!KNL>7B)*L4XnF>(i%r_A)&gc4AEP4~`$;M2ryG#G*h
zU6JL{?Rnk&5ely*f`yHad#eflikT(V7r#;Vn&)?IXLci{RR$))a{H>q_JSPBP2<yR
zooCJ88t!Xli@mTm9~=o5drEZ}><^wBe%xM^$UC!bvA74PsGsxo^N0^Gt{*^0854bl
zYXx?>-bk-F8E3R7#WD5?SvewmN&P+%HTYo%CJHdPmdgNN+SXur)On8132AltrZO?V
ztXuxLnJs2f?(YYEwnh=o+Kr0Pmxs;E>o=+G)moL8y|<b^V;t*y5L3`Aa@0!ohQ^gr
z@-q{ITSThum4zOCf!fK@F=X*G)K}Mu0@cvr+b(NVlpuXrFqsMfY2<4sdt2;P-A__u
zsTBp5=&4Bc%e$!y*9i`ZV-BoSx`zrVqCVu3if<j1LdbO1UY*S>8dQi=6>!Kx;s2WU
zfi=WbQYd7IwPV(I+H13iBWnilcAkR*<P4^XNnTBd?27iwPys|BvjY>}ko%p(%dX6J
zxKrj@=Q+QFMN((^`Af&_s%m?Wyv~@!?%p47HKhg%gHBTM*LAp{l)U0ov^$g?t*@q2
zUnD7beK&@@8&0;06f3DIYA+AVrUpC%wNJkC4Y|8kjN{XqkI1%mb>S;lj5*J>=wjBT
z*1FH$*!n0wRLh|>LQs=EPEBL;Ls*rEc>k3abE-8hq<m}9>p8IGtaKeUmx_=iX~^bw
zq$)fo#gWkE?xrrMKF=8x^czQ}xRmOeu2aF!z&y*~qQqmZCh#UGayqXcw0Yg&WH=EA
zf8K!=Lg>pHTyw9P-@Xu9&14~U0PlBdWQF5<B8pOnogJK*uxs1qz>F|PNGrL`03}l+
zo@1?D*L_UN44wJ%s79sX)a35u@`Ba6dFA^2<;pvjFTZ`oDyJCT4F0;zzEY!{v>VzQ
zV2oFIgD98G^RidM2RQYyqRAJB5!-I8)-qXT(lM#&_B%DAzgNBS_4cOx;mkH}i-QM{
zQrLm%c*rC%=;+@3P^dQ7-t9c)HXj@vvbWSXW!4TU8ed{yHG)V(ML%VU!>DO*ZEvU2
zOWch&m=074jqpnLw(dPS5QfFyfpz?izJz0{mIV@$imA7^^Xw)r{|3J)$82!uvTGtF
z)4sOaNX-U&8AOQ0u>3tRv(Ks5pnGF+RU`MyR~wCwNBVJKF}h5umZ2uOTNhT1q!g`>
zc(agLyw&ud%%VF_*rG)Dfk@&Pe04#z7!rTSC+(&-pP?iaQzexa>0z82>6-`cvVnZM
zb(Qm^b^X9qSJ}<EQtNtyo3=QGq+?}O1(ofVf~CWQdc4PhB_a>&`V(7RA7@vvlXBMx
zosT#*iEf~yY~6})UebO%Fg(I(+!a@jjBXmVcXVxO?@1Js3Hvb2D>Zp+OQ&YHwr<I8
z#2d)p-e6ifxISD#cG%K3Bv^0pNSq`NETbWC;vE|%&qxIDFRO)z9b(C&|NNbe{%iHQ
zLHg7{yhR|F-PUlPgm_M&W|F`Jy2l+Wc6)t56C@A+{ZM6{>N_=MNOY}U2d)e!9J@hT
zwUXE>Y<XhdKzN4thMjU?yWB#sf#cLFCRa|ui8t?T6TO@`|DIEf*F(aqRm$hkfZB>i
zutvCYk~>w{;2j7OCf{W^`ju_&2u>ck2i9NrxHwSlcQ)0QQCpQ`M>K#dSUnxehJvw&
zKoa_j$CcgD_ChF0H7hz5P<1;sY7TD|(#hOe{g#DN#!@Bc&W#f;{Z4&D+A#m@MFV|T
zo0<@=XGe@U!|{H^sn6fQ?n?&QOUTN*syn)AL_s~6Lg+4<EvF{Q@f2n%#Z|q#0p!yS
z6Yl_on={*Bd?xHYt#t?nCzlq+f#sI@@bqfr<oDJq;?3mt_UNTYy}<}J)fzI~$L)Q2
z8n2eMi|1VL($r>J{rwURG%5D}hJX_z!%iW4v|K#$2Wsyq7V3mks?kepB~9E8`(VM6
zF;`xeGu~jR6`H4teO>WqsU*=(+gAdT)})dIId4T*)Aujj5rlD1RiKFt*4K1_r^Mp(
zMArfnVV&%KczT2^QMS6h19PE(^9wWE9heB$wb#$Lp~gAfMJ_UhZ4%t?0E6)69#-s-
zTzI5T9)76o+_R!}#|?z5W7<Od#=tMMVgyTVk(Wv@KfQ3}P0nL{B_oqFt8$3Zwx;#(
z8t1XLTSUr*hhK=X)uFZ3h?odn$b}O#6QRImYEtT2OFR@?Ze7>X0Z>2Gu>6&765Tk`
z3iX;4*(pcz@!l)m;5<erlPHJEB-+1VAF7Gfj#CIDT2BWSy%is-2y|J!<342mo}0O7
z>$Bk-=FW8<qv5ptS9;-?<I$?w6US)_xH&z;HSvjm81<)7eQjv`8*(!Vo>?~_fsxAB
zC2NPs!&t$B`m5`ZNMN}Wxpt65{*yM|A;}q(0pWV<z-laeMPZqFxs-Qdur`rnxqj|^
z=dmV9?ue-h(VZhZ?8EtzQXKObos%nr)DDQ-Hd#O1%?3bBNm{K*aBAZ1ys0Z<A8o3e
zGI@W8iH^S<8cvPYdL_X=N_A};Tf4!t;ey9OmRLcJF7#V4w0>-S^Js4v7yUT4rc0h7
zh{1UlCQNUbwoPsJPzj1bSMGga$(7v%K5uRWVg-JIPAPO&y7lMo`l+T|hc8NCpM}Z!
zFbs;er5zYhdV=Pql?RlAhlD8zR94V?3I?r^&4B~&fCGzpmzXNeqEyDo9eku}<W|OU
z5Ydsr<~lXGEx3VI=?6dBy(#fG%|sE>;?yKIVTh7S0xO5zTQF%cb<Ak;fu$UQDw5n8
z<bwP968v=_OKtpEV8yKQCqwE(&FZ|X$?i=J+Ba0=+EG$m;iBiWu)3N9Udr17Ts!9)
zrx0iwN<kJ`u0Yv<=IEq-tj4FWDW`#a&!uEJ?bM`W8M-2!Ub;V0wv|lx5~&sV*nQQ0
zG4%1Sh=F(;#9^mCTKS*#!I07-p~rTAoAov5{syO<a?CCFzamKT7d^no3-%C^m=D3a
zJOr)*)9hS{JHRLM@y2;`oJXBXcj8sT=KaPjrGVs=3MbO~rTh29!yU43>Q-78J)-4_
znUGjaP4q_M^##!KiA&q<n;VUoAP=Z6aRd+zN2+!fRr+?(+Y|OJ^>f*MWc(m?kpnBb
ztxUw>k5jrA@oJCSx3+_P!Q|lTDW_Pk%b{4rQcm`V!_EZ<Y@h7=xLlAfBEaO`Y1C*p
zE?T-^$uj%4T8z-vL&x?*f3@?(vP$71wct7DElnS;YC_geRnxIUqftyya1Cdin%ag6
zWeF+79%#5XZ~lCzkeH9z5vwVu?SpM9(h*ccbzm`Z7)V1x<|cq7M)pv?&1HLB8%e5M
z?R{W{&*?jXM>B6rg(2nlPTQI!W*JH4o#*JF@Lj$L+$UfUXZ_>|gpB9-vg&m#rW}ql
zls(c7EwboH8utbh+pD1L{g?LJx93!i^hT#J^HHh^QIvt)Q}^1ZyM@H{XsuHe!e0H}
z57>8fZy#r8agO1#3an-fN{7*4(ms<v7+yE%Jc@eGY+JR|zB8NUo)%54Q;d<M5U7zT
zfjmw(_Fe5OW$Keh2QV!#Jv_?^_}%#hvSK-peUqEUQ_*L+vM6W&_jE{)X^$L14j>3R
zrYig1Y!x2>V$PgW%mqJ`IIbDA?`yl9Ok`_O4^qBd{SYQAJJoFB0*ye9iOdObMXG;)
zwYQ60aq?(5g{;<K><4H<#?t-3Gw6*(`O?II;>mO<Q3-EF@1W2j8iT;Lt!8B(*-j5R
zrI^u@!@w=s%3-1H(QdBOZrN!KeYPgO6~Q<rPJH3^`d$(TP91!K8VouJbFOQ=j@yn}
zfJn>DtVwX+!`l+x&ciEHAJjB)weK`ymO9Oq?W|TLhY}Vqg$I9Ub8xr%3=0tpAqT6Y
zP?k^_*seA^1ZRsturdy;_KH*$am7RJ(%tQV9=IDrbKK_+C=l$^=vzK*d%CF%FGr3L
z7F*1hQ=|80wk^bJVSDqlTph_lnThEfSiW%+b`0CsO?_Ni1H$A5A!nVMTu_i+K6nZm
z*q_*nnoReZaJ>@%@!HD61|MMu5_tzV_lBM4JoZ=j*iyrG@S&%l6+J`JnjD9p99caQ
z>5!->uf-8=E|A}@9cnVd2Hld(TgpBQPzsUvU2cb~Yb6DT#-hQ}yDvSAjx<Vfn{K;y
ziZLR{8rE1&dp6L~f{E6qv*jTN7JD`sWpQyjyvo@K|C(cM&JaBmg>)V!D0w|<YR-`D
zK@h0AJ(hFw<R{r)JjasHZcKTyJ_=ELJm12rdeC`JHlbhSk-a<m%wZK9;CMF)@`VXU
zF9#Mq?-%mud0GqEXeSDb<luGODMfE3W60jj`D!P-`8Xpp46NS)2Nt+TzBmZ<sd_`v
z7$QT&VR(b7XEF++8?n>*-i$(6VqI}v0H&qBb6~})B`XJ(k&@KT<iJVJKkhtN_q*E;
znApy7(#{rjLlMo8sR^Tc=TXM=%*O2UZVXPIE>hRv9%bjMeeLxYz9ha<nONYOoccT~
zCI&YJmu}}9DscWw-t7!Fxd=|Z@L#SiN>Oy7UhnG&or<dvarHazt^S!R5k>LFknCdH
z3aOeMxu@rV+5rOPQbLmQz-5}r1!_;UX=XnTPYn%Dfat(<B(9zyO=)ruBU3u*DMVWh
z{VEEFWla0~5B=ZP$HC>)C&Iwgh<+*z;!+)(Fl@caO-?a)fTS(G0@Zl3gQZ-z@6zKA
zth%DRe?-KEfv|`iDoV3GlSAlDt#=-4(}eX{VojjfNqedqmKj6B9Eo5YSTx*6opW@o
zF)M99T)%1l@v<HOt200cbMv<nH!BrgdHdPZ`AdlD^ae6;vN%&-FxuALcwtGfAL;tb
zg+^)t9xQngYLfKM;XnJ)dM{&)(*HEJwo{`?oY@Acj7v&TCj3~}$KbqzG-w@>pfFpt
zuqNKp%HMGP9Q*NlyzV;ZQQswim>?kgiTZ8|zkM8mQ&YHhfyGPen(Fk>zS&Q9^Sy`t
zCu7wSZ!n>}VnV60pK7|sSyduRNubfXzN^I2+u^mvcS|}#;cq{ks5*?UZi`cp)GSRo
z2MZ)bQE*Jd&m_8WLvC0yC{~YC4-gJya5wT(!!gl*wq{K;r9||)Xf}0Up+4a)nT>fX
zT7kELZzOSv>_OnTGG%b{mi=7LA66^r3@PQoMp!FXc-halGWnM`(sjzwTE8h*-xO-g
z&F2@|*O-Tj_iIE#2Z~c4*eGcuLz-%JHD}Cm@F@0+xxBA^?Q5OK*dUUFBn@d7g$m7n
zsUZ@g+SC?~XaaJ8#oK{uya#Cvh1x0m<*I4qLD}Er)_mNliCvxEFj}s|_ABjh9cC>C
zonnnTq`<KKCC{0-^Q+akczxa3{TNz(=`8m-HPyJ>Po)w)Yrpo4*0P;A;sWaqd#jhB
zUw$IGxlr{4#6fO%+OO9frXE6mwV5GV;i}Dz;;i#l5-aQ!a?arf8vBj<au`kHZQXep
z@fOlCI`yLXx}IFJ+;4WXnVcBsYIO0KQxlD=^FsG?2*khDWVs3U1BLDdDIYNQ*>4xQ
z-Xu4`D&Ts4!?pH1-Hdd%PqTQL+Ep=y)JJRkU{vID{@A5`_PYgBkc$+MC(*<Lll1!J
zCuVkz+wZmWhLnt4AIX3mP_&bZN9mu=+wa%6@zZ8ch*1z->+bV$>T^dBuRo5v$se>f
z*gZ<Uj&$$9M1#gBdShe?_``ZbOoVj%`=KPjM7zeuhwYD&DcC%+@gC<{6F#$z&>}MT
z|8Z8$Y`bZ*^BkzGzBCVWwfKbnN%c6jFA~nBY}dWAI`6kT^^&qa@V=`)V1HVD(P1S+
z?gcb84HjPEjQAj4PJh;oilqcIbgL$?9QhhT+L6Zpyv-E7&>fIXk6qN%sSo+AI9K|;
zM>rPhNm;-{s!QI&{-Q<~S~Pl3cqM-yM^#BptdTeXX(hJf3i+3_FoXe7zf%sZsH42<
z(dtR-6miMPLjF~^OGG%bPg-}*(jX#v8u6A^PE0>6&R>7loS%jhNJ|Md@wL#~@ttE>
z4T492?h#~Ve^WiwYDra1*o!w<o_;c&?6bd33`H|ZVo3@$)a`IdSef9z|E|H}APF^`
zcQECF<2_xA`}c`C8wT$l*)YAqDF(J^zUA0)1Ui&%Isedva1L7~ka7S*55fcv*+1qp
zp73*==je@&hzxc)X#dnE5o3#(2l<Pnbx^2?um=Nr)Okx-jM`;DS1tT=Ho9^GnVsj<
zjc;BzGPNH6n}4Z^6yel%OJ!zv6M<8&jj{;t`Cqf0>B-?O&U5YDjL62OYyZ~4qzOo7
zeHJDYmhkw1{d>)g3<#XZ(*LqTu&GW>wYnX=&^qn0|ELcB55=5tnN_7;)Ayl?-dg+5
zYB>*;#4)|K->Hd~n^HvnLI2fS2B$guBC;mW?!-mvbFRc$;6!5oof{AtacnK=b)9l!
z$a~Wp%(j-r817s*LyZlMz_kI*F<~pr)EdX_bGF7#eqH_Nlr_mKH;&uq*6Yo9Nvc9`
zu*T_=?9V>0y@zE84;mfT4lFjjOcZ&<C{C7)zib*(<FO>6eng--VxPaY$Qdfy@V;Yn
zr{b8J>agy7lxNZf`+}_{Hc#PBhbQOIlmiO;__xmV{cm5mHTJJmjOg4bC1XW8LvHKB
zb`f6X_C+~<xLa6Gv`v)L=CJ72o9v6*w-()g+o}~y7T&SSI}oni;)ch1_9fN-Bxtz@
zkgGu=YNG#1$^s4TOWT+@9<*a66#y%OC6dUo4ha#vC>I1t1z(n(8n|u6?YB9FK1L>$
zsX^WC%eR&sw_$`^K~}TkK;qp{lEChWA4vDbuc%K-I}JyIBq5`oPKw}7o%&cn<-HfK
z9LFdnYl5PKp3nk?Mk$DpePwp<mT9cZPNB%5!l+Vv-9%O_3-(proQPtiY9E+LEx5xr
zw6C5m!K6}(In4YYp_{O0b#3T?SBLjIu=uWsvixU`2bS%Gl}sV{*;{K1p=;nK+ez;L
zJ+f?bXNx_jCM#VEH6=<GCj$mGAzsY8!JgYKUn-7HqJIMx-Ip8War>IBK?R7SdzbSZ
zoli?$It)Fp4_z8x7HSr!MxPnnG;YtYA7<DJ%ZHavgQb#M*mmAtQ13}Vu83j>_P-zq
z*D=@&8?vt%<s)NGIo^~_S3b6JAjA3`vlr!%xU^j{zm#$x%5Kj-lp^xQHE6M1$DC5M
z1onfwS`o4Mm~HGO_1k?gy-J~^v`fH>eHfe`C9T*?+ooH-a)DFO#}on9-4*<@HcZZE
z%*Z;199SY6lw1*iX=RyBkn-|wH?ZNkCa32>+HMG`q%IhG>VA7gLyje*qhzF-bO7q%
zA{<7D*qpLg=4iRacEiX<r<4tacB+qQu*=%wFL{UaoTTK6E8lFdYK5B?yxS>M*E?T#
zW=4!<h-Y?r)%P)*)XSe%e;@T$9|uY?8`_FuW{CW@S8pxek)aTJN--BFk}zv=WOX16
z-(J%?GD~b|Lcz&#208ULW5=YZKwY6Np&*&p)-O}GgnL5T7Z0VR9Eb!?eWA>Ou&myP
z5&DWY%j?D`-~iS-fHsfdRobo?6AN->ZiVY^yv}*lG)|$L8n&w%MTTD)9Ve8i=YV1|
z`?XXl62=K@udAM@Uy30=2)Y7G1e8vLW?$d%3j?+s2{g$Krnqm!%dic<p&QlC*a(O?
z4opYy_-YOoaQW&i-UaSzh!jN|@a5+62YX{H-8Rj23zYIyCqq?l1DR##bo0@v;8|T5
z8mA^P22om4m0~``l*n6>skgaRpFR-U&x#dC!^&V_b)5P{f$+siPeZXld+nNTdN@j$
z9_WYzOW6bU@zfTZ*R8d@+hToLEFx!hj3W(DpR@Un!69yXwE0;^dl8$*DF-e{Q>7z~
zJWzIRH{;4Q6NzgbSaCU%OeNk0*EPWEY5WH`odHx!kY$F3!QZZLoxN?Es5+;Zh{v*t
z2FbtwhSouRZ#O~DD2J*c_m}7%d<kNLd>fC-Pyv$cwHw=F!4L0nO4+%MIJmO1_NIo4
zxcd*RxC_Ou0}B56Gne+*n`_1cpJ`EIj?TedPEEG7i2G<1E%cV^a?Y&g^o#>b;<50K
zmG;)QdeTbJmh|tjdfwH*b>UVBFD~9mV#nJW`CEbK30D)3Ignf??$D7=>Fpg(br*ur
zYAQ#ytki2Xq$nguYV{y!y7VSeJV@8+)YOv1<+E`%EXZGO*syM`^PITE1>0!h7Pf}^
zsW!tp<J6>UHTovqSgV`7ZBgs<0C@sek2>YR;}HJnts;UrQ!ZV!n;Oe2v>Q}D9d0>t
zpxS;TW2<{a9S+oplPT9~NYkSWcta8F)W-olvEFX34?QE_Ae8UCQ<KZ5+ts=F$Zlyv
zimP$7e2PNiaB*siG@a)g!Ar1rap(hemB(yxL!q_3x;okcW$RO;J+c5!*^(T=u(;h}
zrxfFlGdTPvWiKbtz~0g3S$7xqoMMvJa8?Y22D#BB+pTS)LdD2F!##S`f#x!;#8Hx@
zr-^>!%ERImqOHcpD%)3NFtlauSF6_b#)q9^V%YQ;`GC3mJ?C?Ti(h!GHp&e@SX|WK
zcKhtQZ@}$bX1^&1nB8Yi6^o;Q-Q+u#w+UJn-1n3Nh)EWc6b}Bu!$QINfgH5oSqLgL
zIi_s`IV{-BHpv2DmM+9v(RSR*Wh<RRPRBy>aZTCERyTj{HBKS-#jOi%Red}hmS|jY
z2Ym`mpREt4mrzB>RY)-nxbfK?Tc7@T?6H|`|6l*_^hM*s33jdB+3o5RU~_Ps-e9`n
zK&A-gth;OL)7Pl~pF;HdKN^Aa%b>xD$tmqiQ7rvJ7wm2CYBYFiaCq1$i6|1%yLO$u
zJFD1M$6<~s1sMtwOXp;=druog=BMKZ(!&lc@sRUeUP@v4U_krqz1be|o66-wDY2>#
z0mWl-=5e^5w)eHIJGEhaTuGtGteV(9HoC#yUtg)-fQRY#bi}FA5J=2Ukn;y}#__Nn
zbDpDPsX%I~2yxmUtghs#T}F5*=ouyv3QoP&ZEEx2gneBNS%zSlDz;zR?!Xe4Zz1VG
z&j$0AnQhCtonVCMAawV@(!(5IsI&aL$iEA=x;`xFPmUXCt%F?3Lq>kN)VH||U#`ga
zA+gDZx|!d}Z(=YGELoU7eVpbp+{uZF9By3?UyVr7;0*=>hUF9eiS_EeJ+vboIJxE=
zSQ4HU>$u!?%GTtRknwuT>v?9|l1;WY;WOK=_=xjd7!%f$ryRS<y6U5jL5$0NJq<K<
z>UEq<tP8<^q>U1^RVqic(he+_5f^$<qh_BuXdlXz-y(yhQ>fQu#M1tf1w%K0tncfB
z6p~6K7%^V_)jfXHYqjUi#gsPMR(28LZ0nt(?6|O!qvON~lk9s#-xwvoE0I00__W5@
zvL;4QwvBb`ExmHsFIe@1Fq!ghIpNgDj44D#HX;HaZM>~2dPV68lqso1b+CTFQ(v~_
zNnQVrCUv44Ae6YUik*ceGwwBU>NUrR2#PlSaHG772X1u=(SckTkcV9UlU=7m)fZaF
zXTl`#oU^GmxKOcm3b~+eQp-`Vjk!cYuC}q}NN0w0<B`EqyWgn^vL!ugdO0fMs!Pfh
z6y>Z<H+&2Y)wom6of0>HkxTfg`|32_v^CsA-ul`9ITghv4vwzHEMl9pgE$YYc(+sN
zHtYp9>rwtaVRz@$&1_q+Y+!{`%1yCU;ZNhXr7Z=W>Wz)z90^4H&M>Xt)25ak8Fni4
znFR~4xeeUHpt#GA)HI4$Z5q=QO{JCJ=hWv&gAGb%C?&>^b~U=)K5s)26MNpN$-yYj
zthaI97V??)?A|s3k-8rG3LwV=)1@rLUte9xgFkr$vMkCNCbtKtJ{@de21&4Z4=nFD
zG+qPe95PG5$;0{p46G6-i2Kr0IC_~EOdV8)7w)t528+0>fIVg(%Na**BDmWr#o$q-
zK|a)yj*A`WY3{4OHo<h(^byv7r#`pI*dU1*j^GP)f14EDnE<D$8*;Wd(8P+n){!3v
zO#OH+;p*|#&SOW=gp{P~v`_R+y0imQc}OO@&#6fPW7+Mu**Dft)#XQFD|e9Z4p<B(
z59lTsh(vu;)1`iRTKL@(c{*^R(;c@hvTttVSvk!%hIMrywFMMEApTuCezc$VEj5RV
z@!R2);+P0CJ2Wp{<Mu*KA^04;-`cI#@vwLh{p!HtH|iv+3rww|sj~@NKG`s51(~2x
z{ZWpYN>jD^a^AkJU~KHrQ>ib&9$-NKXPnLd+i%;a+I(?j?QlvluNr$g_b8|n!>~P2
zHEG6`tR$8gj>4(Qqj+W?gh-@JA!$0kquw6O`C7DM$?{c8P&rf9nX_yJ&$)*ZZ?O6i
zqeUSbLL01$tmRFN-BTtjY#pa<Yfb7tPD3pC%FE*oro<H-dZa}?Vh`7QGRrG<!Kq7o
z9awIO)uZEUaq?Vik7T!25d}9z9G_DzgqKHD2=Mmp9qcsXJdV-?ZvyM4X!gWoneEdJ
zDZ@<B{2Mhz)FiRNdqfdQ-_c0xLULz1r4HfLqON=iq66B=p7WV@FyB17Wo*nTww)PA
z9+ySgch)MR(7g1hw~`#lkK1>3ySzu9Pjp-Wl7+i#aCB;&eRsAu$qf~XR^x`P2?wS%
zS~{?5z`m!wF4G!`QS0-f(28D=hsgJ~?LU9cbxt9MBdd<`OMg~63;Vv-0Q5t~FiiH|
zVA{3HsVXhh_jm8fCYV}lV46s%+IjqzC+!F79mS}k!a?73+zU3djXdDE01&R*vu*9Y
zU_SR@%0B17l7_?$5PUcMrpRJtkJbouU@MK2d>7ktigr^Uy?6-u0`}Inw;5W1Zg|)!
z7IDgJCUgN(ZiOA~yl+O$u;F$du=$qrR+43I$p%xWY-b;WG$p^(qf&n<l!N*_n+Auu
z^^@@q+pfN!RN@3)CTf7cx-f(K#DReoD{fz5yW3o+;`=FG9s(?B`n>Dr+n#zMvY3db
z6SD@aTWBpLsSx=)>#w`FePaqe-jq}9lhK}$r3F4qN_R2)``Si*@&PvV1D7s1knHLj
zE~npU`?o%wd7mdndiRWnOD_DX1CCuGw`?q7(CcG%pgP#^DJaGoUmfY%o%;BVa$3uc
z`UKxU*tK&w+%8Z;JFpmNx=d;sVKqZ$60@??4%KgTWy6Eus(KBqCMR;O!)+^ZpU!{D
zSr_ea9&hvK-sn6AzUW9&VvHT3o<KgbBds%ZI-*=y7GTkErst4Lj11-x@E(!4N4pP)
zOKV6S{YnJtV;LkS9D=@fgc|G<CaROi?N}0k<x4Rz@SNzi8LjzRZlUPeW7TC|mg_=g
z-DO^!nwl$;WI)Cue<lP`4k|FJ_IQqa84M$jiC`;N-ekuci9}5W0z2ib99X{6P9zF&
zo6q){lUF#!ST}@<K)I82_+))ods{xpaRWK4BR6}ZU7Y%kY`6I1;Kb}y(avpf2fLk;
zC?7X*OxfwS5pQ3<<Tf2$01_1!kw<FE&U9^RD%G2IdnGlkhloX?iLkQ;OBbtfV{>59
zE8&qGPCEYtJ6CVlhSR{k!PJ7Ry)f5yzFH7QE3_+h_0TP+Mhw(~$w|A=J`Ib6TN^6A
z{SGWOq-)S`j*dVmNi}z|?Y;^4=JB-zj{wQVU)-BS3p(7p$DXK9rkjSm_2s2=7@3Zm
z7`E64bw-|gKiG!I5MT_zaiwiS*m7#(Z8mE#4)US+p$3EXmfyd&DKMeJf>C^+?NaL@
zk~vCTu59Qa8}Q)hO>!zcnae~vq!IXBr>KdzWx-OLspdINpzK2RQx2>!Lk>QfVqORD
zetW8$+dRo66DH3?O?Dj-%!O_Eh5vAC8xpAVid$G6SnLK(Na=BTJ9CvN(DLcl4or$p
znnMh?gE2`2@N&RDVL#Ggx1368Efk(gteM$Ps2A4WzI^F`{b<{PP$=<m$}O2fRh)57
zz50$IM*P9}W}Wx;W8EB-ETNMd_Wit5qr+&!II$1Yql1(7<2kuV5yR3`QH<H^*Mg^%
z@)J4j#K}94HC@uzhM$uCWc^S_nBv6c<rSWWPR$j3@(uqF|FfTJ1H!(%4$Zc-%mo9}
z<R`0vAF-d#NgDhRw1?;92cO&uR_YM{nd+lhTWv*Yc7YWU&&a_DJLddMH-PQ-v-O)m
z3gJ8Zs=)$xnRsM=`?;<i-qG@fu0;NjQh)An>SGdhG-PsZ#kc5+?4PeMs2%y;$(jR9
zi!FPP{X*-(r32vr4+VE%*&5ui99@+2GR^0|*tK>7d0XiJxC3ir)FI4KET!ozivChA
zBPW6Ln2gB~iay;H{L5_?2c{=+N`!|8lE{@|v6H-Lzf%3lS<0|peP^FjBkc_SGI&kc
zujaVWt=-`~#~6Y{GX}Z&Xcf=cuhkeNT4`nRN%jVlcRKS2jd{#|y%Das^>a|xJV5#!
zgVcWgjW(J{NP|fWK+T&rb;Nh^su;rGtU1bU%-^C-6-MIJ=fsdFf}A`X?YDB$-hS0p
zUXM2|SY^N6&a`<qUhfp*Yl!zT<I-P=F7B}3+4}V5>Sl)6NL!{jy!6i0m-|n1-W6?J
zjT}L;)7JhQ-}p+W5bFx&i0(4q&$z6>es60;0=?}c&XZ2HK_*VQ^nbs#vY2C-f>>Q?
zYwjHBZMHvXmGkD`;1uE$BJ|3!bc|1gy%0pj7s388U!F#&pL7bj<EC(RhIgN`KWg*M
zj#JvE?UADdbE8v}ht4n_5_y-!9~T-fT<JX4%*lbI|IqxC>X7!H7><k5yY_iuQXezh
zZ-<MW4*kh|?MmtmPjzX#{b|=$myJ1gt{8B#Fgx`@)8u!n`9&FGf7Wo5Hx%wgB(noc
zOvaT+z1201X*>J#ruiUx9ZfSipw`CySHKGoyJYUs)=vA2+#$=yIn{X9s%_4ku=ba2
zE6u%e&b3Y{Iww9zS+sP!!e8Z&xvTyP=Q&o6dYH{5NQvBG1+c%a`j!Pjo|=)eJ_y=&
zr(Q?uirbgl-&Ak)^R%MJxqaz?Q<J-UfQvosZ`=CcOaO1htO3bgbJrFv^IH46_6qKB
z6XU(HK?jme;21n?f8Tz<$*-$e9Dpbx{t$QCKh%&#B_c~E^cz_I{3Zkt`^RoYXCpc#
zo|Y8v)W@X7`iBzkg8fs$bb#uL1EPf-nD&H{$>Ghr(f+yGEw3TiNSz<a0UhB)pkBRL
z*xN!P8T*%-ukwe7SrA=^3wLU2(eztdfCc&2ZdBy+;=NOs4m+^iNdtEi)j?E}{ad}I
zTmtyii2gY+^^%;ecs?<?|DJO-XWsR5o#*&c$_?l`W)9nbWH0%~ZYVRQ{KnWg$Ekk*
zpZWVGWL$Ed<M*6i^gi@__FwH6cnPIL4v@rr?G14LC<2w{<*5C4qR-&wjn1P8*e$(r
zOzQS<&FoTKHn1d7j_*XWLRREFVe6k$FtsGs8tUl4;!DkBs%yg|?}~kH)rgKug6gjA
zb!uV@<f)YIFL+&tOR`*9KCix`8hKjd`RTxF+XSu?{HT+4Z*Z?%5^SIUaNr{$E`!c<
z(K{vt&(2}{f`^Oa9>ypbfu!h*0p?~8jhVQ$4Bt3lU-)n%R(*GS!QX2bx_>&3L!KF|
z^KPQ#0>VhRNBxT)j#<Slw+JuPZlswliqn*JP|6Tt^-=rcho4TtJjE#v0SQpb0g;ye
z&>J{!6;)CcUvt?-=C?1YnF8;^qf7Y}I#|%vkZJ8pyB6X~GEOALA1dh!HHlC1iD0fT
zqEq^^g6%B{mMov%U~*wsR1cD}eR+)zT#i#)w?zea5nRJwr#{!=CU#1QZC{Zs&um+X
zBOgJn!svoAE`^9{8BW<(R(~_3PIjtDWq9eRQ&am#vD;GDU$C#Le)K6AG>gIZGuuvm
z;b@ngpMHwF`08p+Ur@?zcrOQ**x9!xnsZ$>q`N?$)kY2fD4S>4l0%UdN(!gGcBqWt
znvbLQ?1$sTSb+==BkvTG{I0<{L_7RBiM@+&UpC<NEC)q<?!$E&hmVI63p*gu|EYD8
z_BE~doNwYbamjXIqIe@~;r}M=c{TFfY3OZHs4H-Ik5d!9%-iCWWbK$ezkRKDH*S`Z
zKMRm(1a9P@_6w@Dc7KM}OAN?KZ#6kjvH~89mE3PHY}-Qt*1d739J8A=I8}SO%3wVE
z_(iQ>_&Dlz8Fs%nnDo*l+tFoO_TrpTFbNK4y;CeSmhK}GN7zf+EN>!lwwzG_6b%WE
zFEVZQIoH`ss}bq4rLB8>X1fC`SsizZb=cdqTrbN>MsSD8_A2p%(^x@X-cXkf&YfpS
zf}pO5pSPM^j6%@3`t%iTVj~xD>K#;n(5WeOm#1KOjM*zYm;@{gx*Uwa^0=KC+&pHN
zwMk#b9iba~PB}J!^lahy0J_NUW2e0;e;4YVolYs<lef3t=t#C*-sYDrAodQUxOXr}
zLLQnI?A0~J{krGIK#O&wKFzM;Qe*R61aVQ+Gq${71YbydP4_jdLLxSpv4GVWx&$qS
zVz135;^>>?l6iSLbAAD=*SW_aVh)j5A~W9=)i@LU)cp)nX@-CWW;{cbhEAu%eA2FL
z9SS|2X!AaAFcu|Fp<|c!+g0_RY#8N==nW<&U4E;(F;cy*z6VCb1nMl1Vw8v?rWU&F
z)N45>#y5*we|<w_l9Erz1_waQ*Yas>2yZBQFgWBq7P7=q1^(`}t80OEo&8y-r0qPj
z?Y7?C_Qvk@$B-Um_So(X#w7~TF~#=J$tH&2k_p15l(cJfoNENIIM1z_PK1sWuPhl+
zd`?aDXSaAK7MR<q?V7f4(PE5kaf<OuBLQ+Loma3_iteT-?hO`VTHwkqn_rXIuVWSu
zt^F_2lzI&oM-BM2YukrHzK+XM9a!K9jdHXXB?!mI>vG65+unH9Rc~=hfgP2qo5mG)
zq*Gc<{sFr_mtbt<Zs)NODU-)$Fq&+if#r5XTYyc26HXys&hIB{p-g%5pg=caH+KDl
z#_EgUtIo$=PEF42%npc**pUnNrfMy>Q??_iTo()IwioJawEIyUr)(R%h`qVd{i$AW
zIP4w{NUJ=h97)6ST6oLDPtO6sx75EpHNaFj^OWDxk2v{voe7z-=cx0$v>6~w4wJSX
z3JQB^Z*6EWGPaH|If^ND-aI{)c|PoIZKmd3JJ%`1cx4jS3809<x3?yUS3w&_F@~3D
z2&0wK)l>J|*LDM!XRD}@g2QASq$U<1r>qVmj(rK0ig{R&3p>gkDLmGU%cv%);j}F*
zm~u1}*XzKF?nrMU?toNxfm4gxNrK50V|(VO1gO*e25wiosiBUt1ZjOh<n&kxUeT$~
zmV@`rxzCmdvL7?shVU81Lu3P5Hektn;gh~$WSVQ$Z*EIU;!!yvIDkAe)=aZOZmGd&
z{uIcm%oz=qnMC3~I9uGh5o$vYX|}c&<h)Z8Q=}GfQo3}EV|q#Z#wya*I;BMTV4X6I
zsj+>alD#A8+BE4Soabmwhzjc!jNYx?7tVpgp-OOIIf@a)cw5>^w=v01sisU8y9gl8
z&0x!_r9Csm(=gdPaCvDkFa&m10_R1$ty)M|pR}nY?od;7klP+4BHv!GHWDPYa`Nxg
z<f|Le$H0l$@_JQFkZkqmVyP8OPEAS^gP43VSHH7s6!jyUPq#4S_o>NoKlu<|;w#$B
zLJ;tlXJejXq+I~MQ=c<|X1Pm$geLx&t?W)@2`8yTN@0$53=)Ual=D{hD?-x7Y*qI?
zozT)%$h06Mr^Jf$PJOBerq<XU_3hl5DJ<x%LrzVozc!3W60keF<qZi4B-!V{^2C}q
z|3<s3UdU_L^R`i7NjlIaj-XM8J$qN{TyW)Or<-LsysO8(Liu_RilqrABzBd(yFSA-
zLPQjuaN2<d>S`~_Z;~i&YPy^1?LE1XNt5k72Ih86LP_4+%|Ut=D-nVy(LSdp-WJc5
zQIkjQeQnT7Z@<MUh(x^a{co`M=d7at9dVwcLlW+V!LsA_n}9!1^_uChTZn8)$r%Lu
zq_=w359L@Ujc6VwOuv2b;d17}ZwWzdC=$vz(CBR*07$^>+OO-z%W)D|u-6+b*?~Pz
z-N$h@*a-IU_-43qib<hHh}{np*y;}1&mfELLY|SS4Vl-e$wekq6So5L@z_w|t*}bQ
zRD$%ud*!*5m?QTHd0Fz7ChuCV10U{M*CpO5F~>3C)Fi118vw@uoLsNlvy-?y?45D|
zdI=&M^P!+1TT?%j*dy<D?O<SerjXxM$}kDV+D{U^c438G+ji1|;jrRV8PO2!H3#jw
zZe5iWLM|yfVwIyQxPNc;$yIAbn=4GJHb>fEU{g>Qz=-bh4i*ZsU}Nk<^<LUe<z;rl
z8%#I@@nZiUw)OQMGaLdL*yg};>DFx-p2Y28(ni|?v5`5AM|;B#NK^kog61~Vn38@5
zJJOCu-DT|5Cv_PndlpgAWBJoXT$aLP?*ATI_VM=KErdg3RiK=EhbUwmmQV-pw26jg
zLxbyaeV~}XUpX*fAMWNR89u_`9GGYj>k6MZVUziZHIsvDrJiugIn>^oHJQR?s$m9Z
zP{Qqp9e@UU$Ep>!v8yAP8>_AZ5aPl`Da0hQPfq71UPJf1o(IPWlG>C69kPHW#VIA<
zG(@Hjt_VhxIK?5exf{XuP>Fzd4lIwCWn{e{$7}uW?xiEa*yEw#z(Vj!9j$Cyk~Q$F
z3KUf)0c>x!r7fd^Zx|cP2uSlIra^wA_jF_>1B|3`g$aP1X(JwV>g&i=@_!+cbUF$J
z{7C&$yi;)7kUND^>o{VkJM}r!x5wp&$UPrz924XR=SgBbSWtk8EBH$5POf{a*-ZJY
z6pOH@&{ADogAvZD*Xk(z>g#J>)Klp%5zZZ0Y}bV8ugHil49M;D4fPu+MFim;#2KK$
z@;!wB?Kdyh1z#@bxo-|c!88-hj(w~l+bwv;PC3Q+xQJ1?*Yl#HnS$)ueO*H`x3F|j
z1R6q+qb5lfJ{PKi%GmvlVm~SW9eKk%2#*UirxnHL7EE!&_VKpJvg#8$0Ga~IO|Xcl
ziE;ZxD=wva%qi9xeB%xFjrHZ!CG3UAIOo7(z?vi}ej?+X2)BJx{&ayzf>S6udtR}E
z%-cTu=K84!Stk9}5HX!NHSsnAV8|=z80}kfF>hadoAa1^pR_P?bbeeur42V0tXxTs
zHm97!T1jrNo_(@@64CB_Uzf52D{|dM8nFX)DZLirq<vfKAGRzol@q_YFj13d(x9%>
z8}4x?eX8`va`eLSp7qjsIBQCR*A+EK&AZ!qx150z8AZqCSSeG73}V4O*aOu)Df)F4
zqI~U?tPAA=*nP;U7oSCTzP<KfH@%8@PoZxXCS_$tCOlL_$`eE_ODCjbmYU?+u70a+
zefa5zXnAJa|G)ls`jva8PyO1&!n2o1=h{Y>Cnw|5R>$Absmam3>N0yIF@sAULo-M*
zH_0lToaHEL-`-YtXk>D7-8huS0p;%(z1u$B=2{m!Fs~PJssWabU_;YkGG*UUZ|RIp
zW6cEz)(8%qx>iXpr!x9W4RFLhQ@@cq0vB$SLXSJJ%eBhMb@iX3`c1P+B*uJa{a%-I
zi7x_Uyup-uOOhD@F7{n*uiT=`C~!^9t#vCmLw<KP5!wxii=s_{<sq|yxO(zK*!Oh(
zRLX4;%UupEUkdUDasJ+VDLLuTiaD^_aY9oVL?$pdMliAVeQg?tl`si5%YoFAhu<r7
zTlRWgikimm@2^cjB$)GDq)1A|WWP`(L3@9oS_vN*ey>)7|4V}9)F+i&Pm<kj)$vql
zM?|{=i>6b(5y^b{00vEaw5vJB{6aX`pPIb=PEFzjxs!XV$H&*(_J&v+*D*dQ$CLv~
z9W%1KJbgk&+|j7{0^&eNHo|)-NFHdOemGB~*0!AwhbzJ5_c~8~0}sxe>5*UOu4F7m
z(`TMiogB(=C~@Yr?QTQAb9&S%WJCCH3$>KED6#|Co~n_^E^A3~0L(fynmaE1pig`A
zdvYf1ZE#9;UL*TLBlqR2gm=L@QOIv$2bK3ochiRL@6hW=D2MUoabV3_CZy-k+#S6T
z`gNHz?LfD|y70rOm;FYn@>ZX}DaE6-z)^p}4t97VrGZWXxsNLTE5vVa_1d6HiE}(G
ztmF@=z=lIPz(vgWm{W)WYUid~o>ahu%G=?#4zNS=62nix8%!HT0s6}IaipycS28(;
z+&iIh$`uepkQeKnwWIYp)yN|_m5`vQw;FEq6RRP_U=K*ib*z10U~I^%uoAn}ggw^q
za%g-|hO4j$0W9?BA{UvS3sA1dt43pyh^!l(!?^3z<T<Qeu+ffpo9h(Jnt0LU4y>A$
zPX_{H%qo_5JCSeWsRZ5Gl$`)3rf$))JMCoKPlzF_$K|o&fP`}7h`aFQ#|fY~l_P}Z
zLVry`dJ|Rw{I}D66DEfh-CG6!0*6IS?Gxp{k)G>_oyp(MY`X*hcBd4Jky!=xw^&^!
zRsqpxJ8mEo?TD#N9V3!pC`c8_Ax+x}=iU4olc5yRyV<wr+I-@hNX!pYBBoS+iqsVE
zvy1+shCl;!{u!vj7vTTG)Ti;lDNtX(iThjRg=rTWjT!E7V4)R^!w_hkHjrZ`kal7l
zBqJYkEGS<FYP*;#J4%<OP^B2F$WF;aLD?~3$&cC-Id!YU6T>ORd!gB6DU^{fkFOuB
znkdzQ7l)iv69Z7dfg}e4D(#25B~UV7`418r6VWRXi0stIV(SFhX_u-6oeFYPS0+7R
zv5328<O|l6k)3Q~Pu6enWfGeyDZqi%D2aIq<X8Nv3|BMVBrA>29x{_3^9Iu$5%Nqr
zkBQ@{B-^k1sPhz~(jrOgdy13lhue-^v|y!EsE>u>8Rdj@8{EijPuEYI^Ei9ZfoZ@q
z+it&eV1@li>pb{a38NfX4LGl_l@pHipGpM<OKv~<@YBzW7LX!N@b3u!F7Qtg0cV`r
zJkKUZHgYVX1NpJsilfs*&SQRjYI<zF{dgn%tH+Vmhn=$a?9Hnd*iU5D!s(IG(1B1E
zXT=2@WA>9dc+#V79&-xW;$lVb|5Pj8JTke#DaAngne9H-$bLE-T0XUf#Q#;e2YRr_
zex}g_t~GZG`H6)%zSz&U8I!vw^8)HQuvme<R9Vaw+7SD>dOLlZgWNi>=vIFw0dd#T
zjEYMBe6^Mco+KPiHew|;feI4QrGZC*Y`>7bV~d<XKSQbB;CgciCz;&r7js%y47}5M
ztm?yQ&!YO5+VGb3xZi+MydlX|fy$hE_R9rB5r~3Ee9<`vrUl^>kK3<Q7Zj4Y(<znI
z6G_x1_55lzkVheQ+)H~LSaGZhsmPSXxw)i{m0w)vw`jlC{r-e>cyP3*9ayf))mLBr
zdi(XpSFV{B7c}JnVoBt$f+7`OFw}l%8Ga*o*1HFsr#y59Nwha;zgb@iTLoV^#9$gM
z^=pa;-EY6uzWn6<GZ&pwd??*=W!_Tg{BfLNf4eawp*ZeiU1<BL)W+De_d0JCHek>T
zZJap6&JD$Wr`4DGdDgi}lEGo8UW4H18(Td+YQI~9NISJq<S_@9eCy?Jc%S`Vy&(M5
zMFG;THJCWkNDR!O|NFJ5n!3YIDU<49EFvHey%z!ASn2+tdM6e`&KNO<U@B5xq(0G!
zvC1xS0*lEXw%roUoGvxXBmSaO6LXO`pZCxOOGV2l)$Na}H4$(r*YUk_V7d3g#~@rp
zm_IbW_Qzd~Zg&?<8udqU(xxWoyylb}sFQF*q?I`3PwJN{sYVepBL91w8e!dn4TIb;
zXMdU{<*NbZD8*P)WC-!W{;Z9^!YI(Ld%<S5amiHhOwh<bFI;JkGh)kf7E=1o!;`ma
z`^F_kY|(x87uDBXJNZb+F-6yZQWI~hMu*uEv@L(xjr>AHsL4etGbHQ8IOfz}rESe<
z`Au>mc^HOF7&PY@UK#dRNnLK3H_v%YJR6)?XMddqCHUUrRUpytb=TS7RF9uY-_D}o
zYc)A?om$c{!(`jvwz-9qsenBPmfLqZ2zihFU3O$>^|13;E1|d><)Pebf8WhU*d9ac
z-C!ZWLL_v${X@MMu3XFjT)6`)d?A+)vm2M^m8Fd@*gtl^;l>4U?9`<r-e9t^$rb%5
zEY3eQD#gt&+z~eA0CET$Fd;l@|J-4$bUnM-EyX-IH955_R<5*v>4qXQp|H%o4oofK
z1*WSf>|bYh5-tJ3Z8DR(t$7cE)Z<Qlw`&lq*j-JZ^l$A@zG=ydfkmsFa*Un*Tw*v(
zNSMkpgWbP3YPgW=U7S+PeA?e|KxI__(P8MU$N_~zWk5}81IX5eBkM-{&-yuhkqn;t
z>%ep?#@hAvUsWr~UbmBQ52AOlAZMY>u>a1pN#W?ubD=_Jw8Um^_DBvs5imW`gMD{u
z#Jmg+NvBRM-IRUKBf*oPT6P|5Y&d3WkKu;+xdlt>Sg6h&Sk3LRFz_o{J3eKf_edO?
z;|gi!Ia<?cQ?3E_`H$pW_wTsobD+Rb>Z4@|mXgdbif5BNU|&$bjBCb#p>3jG7tq9Y
zpx)}i1n@!nMJd&hSmLdn_JxlWA}+>OJ>?XmbD6{-%*6IZ2~+4UD(n%_JO`Gd`w)3j
zV2-}{k<wEv!J5%L7y!22!|uq~(p2Y!p3}bMk+LI_re=yskF&*rW*4An%H4$}#=i8C
zVl5OD7rC4qP<WlRBlQ4eO+pQPN=W~ep(J?ov}vT)_RDfzujD>L9!1mn#u3~Lzx<Kf
zX{+$Rp^*D?(d6KojrJAY`pi11cj^ID#neZiB_GOx1_hFRWoz$F{I4j)0z#76Od@k)
z<8;r_S2dhu%g7)5VCcFbNKQ?y1$UaoZw}g5C!XTv&Ye~0+NK;xQIZiH&+S>&)70P+
zi787Vm{O_aeYXBI*teICh}qLU9r4$mozwIF4}QdXOhyZj&nbIOhumle+D!BF{oZQQ
zH2_vA*gE;PKDYZ#sIZZcL?orJp{6D**$4KteN8QgjI<Y>QWCy-*Uh!()u@=bj;bm3
zxnyK~6`gub!sg!UDfD6X{Dz&2l}2i#Q_hWvB!#gwJa07nd+i0ShE1Isqcnx@cWR=C
zTyg^OO#j{&RyWX*(2f`br5_Zr4v14<-#N-5Dzs8&8JGxrQ8$2D&6BpNR5_B4TYVYZ
zIk@CsDAP{d?qn~nmJ{iihSYPS7@{UuY2}jl4A@KBKI9@eJlw-!$^lllq(KmWElMsQ
z&zE-H`s`~lBy?VI>SGeXCL}%45Vx1L)=)Ib1s??}HF7VfCZ|kdJzqG*+b?ejyIHDa
zrQHIeg<U#u=fF~X#UpVLtdhPnjH&>luZ={9y<Q#+_R8+NDKT}_)v1Z^hV08mA>o%@
z)=&WJB-T-<oTph|UIB5Kx=uqU`KzjpcKs`O=D_L%RN|xN`I255T(@DDKhl(r8AKHP
z1rnMl?8)Hc&Ji}%-V*J`nv&di@SnL=u&)!?^y>C4xixomI;1^u%TB$PS>Cu%{nzAq
zGA$>0o}-uPp`jC6r;1eEnu)Xa+O{fBrIrZGlLO1mb?w}_c12CB1R3FTXEtzZ^cpup
ztnb;C^%^`ADw)*bCGt{J^E7)cB$5zhK=RNIyQ<;9v~t{xOs#PsdH!$YrY`Qov)4Uy
z+50F3HF?tDYbd3{>l?l!_-t@WwM%2#HAH(uLx&V3nUKTYVBs*#7|AVaS2u)`wg)}-
zuHY@S8(U4A#?<+s<*_$b+g#xqYLO$vKL87q&CFQIqiFl!9@};!!c&)`9bgV<qmsg{
zwYk+~%6<~RcX@+hxehr*7<x^8rq5Tch)1i5kFx6Ou4JU$HZK?O$}8W<aYrE)G((ek
z2kX=GYvEEMA#I6VQIz2~Y$#5B5e{AZDF5WhzKAS+Z9BR#B=6$iZd5?09B8Zt7eaAH
zV2UK-^SW*|BBo1%q66#O<kevq<s)`|y$!QagV^<d5qI9<bsfjqpY&dx-Yv_v<RX(|
zcV`zRl9CCk*>br6E&+%DE`$q^f@Ila$#!DLHBoaC+Z20mqS#9ULUDTUz4zXGzTZ2u
zdjKZi|1m8*XJ_}Gvt@R6`Z};cFQKzcVI<6Ekh59Wme#k0c2BM;k*ow|C^gX^O$KKx
z^!So3E3HLC6{B&vI^a4rd6KU0kc!q`U4C3V-ZeR1VUi9aRG0nrn&u-X<CLYwft4;d
z@;p0VNgYDRVXrN}$s8Bb{ygWvVp56?mt8Njf>ejEOO)q^1umm=O4^hp30u9%UOzjU
zA~J|Q)Ew8EFe!iz(&{HxzM-yNTtSz9yj=&Xxr$6+<qjfyV`9uk>`4QuT?Zmf-J%7{
zY<be4-l0y{d5(!rPc;rUcKwQGjQHp?dQixLrH?;bQ?XJJCL%eHdXDA9XIt6KggiF#
z+H*xqTybiW(#$!?ORh~NE4EcledJ;Y)6JZ#U7Y&przAtIpIo$8=a5#fUGF?5zUw&C
z9uNr&;c!iTaX3{a_K5+}$*%;BwLXkBg*b932D5$)^1ilwSkzU%0*I4dVZjmcHE~S0
zH0{C@4a#)bf#q(Ae1jtJvd<7PVCyPd98N^!V;y%uTA=l-@3C7OBovR8DYL3_4Pn5k
zk6Q=dWw+JiU==yWC?rF1J05kmzLsv~zCx#zB;?-7flikU^R}UMIJq$p4+$7pna0#I
zoTkG;N^N85lCI{GBvlf{DnwPT%T9eFAsi<L8(a&!c6<4Pgc12UoSWY7z~VG~2DIK~
zchp|pPA*%1Htaw`Ny%QcH|4hG3U~2_PBA+d#x-T#5I1`QadlH2>TRT`bxPS4E`j1+
z6EP>&Q@WA@x>o+61A8>~&HuGCgl#U9o0l5|Q31uZ_JQb5eIS2qCGY_`$?mK)uV)fB
z7^hfF`2;_=$k2CJPGq-(I-KYDePUontK8k-nB<zWp^;(ase`MJQy=tV>3qAVG!^G5
zy@<T<g^o%PMtqM`pWWl|mwEmi$L+o4H!{d-zh?e>dCExMJ>b-9o!1QE+V<xB{mhOw
zSBnzsiI&oV#pf3<wYSvOognR`Q;Lu2?2*C+rMcbLj8E#4@I~_qgJ96<jBR{tX(!cu
zO+5mv#xh7D%m3R-vhcQMdO~ht0~ghL3aL(gXx%iW8HOw8RuP{xSjBz<qEzU4q0ifM
zM2nX#a~{ijC{v44L?5zuG!rI4HTjpEYF(Fc!E3$FQbm7MXJb@;K|ovidK5s;V{DYv
zCZw)V=ycwC%ZCy4blC`crIQY<P-ao{L<DKgm5~sQqV<(}$@B6l=18VymP@9-5V;K8
z7ZWN0YoGPkH*;FR8B4Q7jV`?l0`1f%KQ+?fp5`fQS87mF=nQ)U;pW(OKNo%5mK^3w
zZ@kaT<I<a#+Cbe515<EFPBDg*H|GjTL(U88%M#gOP6M6|3dSB|)w>Q%glTh&L_Om+
zR0kF8xLAKN&N`i`DQhRU<UE(}FR)aUh|js`z=F`BR-4{$Tg&@&@n$%a6r?%g)RgI;
z8+n@2uGB=s<(Jw*GWRJ5NkkPnHOckdG2D8nWh0ekZQ*K(vL_7t0K;UHyJhZ8!hsO<
zzj14^(cF?NCbw*Ho^`@(p0rNgeUR3Umy%p~An;?g2Wyl{%qf;3AD`ZB50rt;qAD&5
zX?+G0&O7xnk&;KmD;!5I8*e7}1QV{4<$?ptVGfP<O^#c)G?CF<91{$S1B*E<qeL|-
zL~GJ~+C-T{=)P)@HI!HbwYyW3m;zfU<d+;3y0biWZL;ZPB;FEQJFs#{X08F!VZ~ND
zZ`)@3F~_u`@Qx7iyw*$UC|)9OX=v^_v+WI*uuZ6!gLoLQoEou?SKl{%xor<N%?_d4
z&Sqh5cWM%uP(a9jqN`zbi?AGmb-ngfGrt@H%(?P40jo~LXq@nhQZ#|YG3pigpOoXF
zM2`MR97}o5vu1T`e0T`e)x*uW#WbfKOPp3wqB`|jLc;XpFn(w41sg>M^$XJny~5Z!
z;Rv8>Q6DL<O^uiqS0NqJ=hHvV5b6Vi!df7j0tUUS>B?pJ#9YvjWi=a+QR)+$#@hXO
z1irg8&%Bs<G0{E*1|4(qB&B}NTB{40(QUtZsoBr@qI(Pvz$%Bz)2YdBVU7feq2BxK
zJ@q>T11I|=Nliws918Wt0!sjauAA=KqxC1VDLYw_*$iydXV-!_i8YXSs_Z6@m0yOt
z1YP5b#dKqe{ddZ1eLZB9M0zjVB~_J2?7j8PGsifdA{RbyMj7VR%)zfz*9uqU(}-o$
z-j}o6GchThuv08EaVXqeq)&T)67aQLt;}Pt1xGFy5`1UnWw)Zar%<TFF}4q`YS_)b
ztG*6hIDCeZf;+IXK*#9C9%zZwgxttvA4qBx2{+nKsdR<ImKMk~_Q87o^l=R-Nv<d-
zkF47qBJJaMH_OsTZd$^mx(+B0t)3RyKZoo?Ii6dF2AoF`&3iZ4hbz&%W1RR=O7U%p
z5ZvN|9_umtNPW%V<OrFLopK>ziqc5$4++E}%uFA!kJdM?=JIi;lrwkxu&yTmSX~xS
zT&_cV!8;w%LI~Isr(Vn!r^2XxJfDf@=sM43@=_rvZ}>dXz9-k>4vah#PC@<O$dwo4
zWCV*#7rr-pfP6U&<L(s8o8YYZsuJ8tE;?Y3*N$_ncAQdneEA5Me%KR9NjiO`PKzW8
z0Ai}N)07iksdAsJONVtpTGum>iwcwQsf5MY66~qkPR7F=hgX<_+jRjgN#yM5y39TJ
zpM-N1kg}B|yUmdV?GtsGZb?T=%9_b7!z6k8zVf~FED!UUS6JLr$PAEj&c44AShN^H
zPT7t+&7lBc?+7l>X{Ez+Y_OyIdw(FYYT3eh&Z8>0>7_l{XP>Meh1yA5NJec37PynF
zZg?`pene%@PB~_uD!;)B+Nex`#c$?XpTxNe-=N6$>GB&H^kmVLNB}HRf5XtQI=Q2s
z1LaC+2cWL@V9-QI^-P)7B;4t@m@WphHPQv8H>y~|M{Q@~U7rG$okH%}$Z)`G*f+bf
z4<$wTxKoZc;V;Jm6QFCm6Q)bbl+}>fvQY;XORR07Aem$lP?*`CTvB#-*LjY&Ro@;&
zFx#75;-*GSfD|$tIyt6D(WdQzd(*!1GhvSWR>_#;6&5%UC&N+O-*n<QI3Q74S?ZOs
zg_>ekZDX_Po3aB*OSataJXM0zPvQqVSj*k^cFhuHq2$btyYI0>r6ZaCFIYyeur#*?
zn%d#gPAxdL;P!WWg=wM0u<U|{9?5pNX?93_B(ua;2UhrnwDmXI9|qniAnj=V<?s+2
zJ(*Ep8p+j9Kq5YYlJ!`5ldSLBb~;Ft=MbeV9`H`9h9qdx&fscyyxFj_i)h-5r@d1X
zy-S^XN*igK(fME8eeFbfCrltrO0t5$M4E=%_`pG$PS$PL-x=+5N}6Quby#4}Hd{w-
z{#b(OLd{x+`Os^<*e^|PhJflWli2BYD&HKpn~giA*ofI-kQ}O=Y_j&vMSHHPZ**4R
zjAY+1Yj{)7B;cUSfHX)6SuDr&VLM%?Dp<mBj)<D`PE8_v=u;%fp0qQibyc07NR2F`
zq9zZ=&g4WMj%Um7(^HGD!7nv1w6pT8Jj*hltKW&_?O{iCVAl&3z<%Z>M#?fuz>K7M
z%z(_F(tYl;^Ud!SO|*aJ+38&ltnf>uh~}EYcE4R{w$LOx1QMhG$c@7dFaumkaIw5c
z-&De3(#txqa&E~ql{vM9eCO;^uK11ksyWZOM$EBvGLf$kX;8s8IQrV<+C#{SL>c9&
z1D1%$Z}NIosRrZ^s>~TuohX~@mGVo{P79GqxP8=tCBfg!{<a^i!_uX7L>4%JLdoP~
zo_;VoYA@IiH3JMIwG?CyELv-dH9$EL3O&x-bZIS$c>1cYIeylGWzYf#a+@TG;oN_<
zj2<fs{gy^b0(=BLanA4b+4^#|Fe|6g$#DQzHJhnxMGi|cT}s#n6w9AU7Sh*6(^?c>
z^ib{q(oCMT=j+cOY>(<9t}PP|C<av-SSh4}fwdp5D-o)g2#R%JWx|qIsG5@l_L<6Y
zF5;Y}km$0qHB5k<{YaKKq0i)Ax^br*+eI&x{axM-KbkZHUUyt6xhYC%HQFAuAFIoz
zLF?+(2GA#UCJMSAuj5)d$Z<KOgqnboJK7*eT>FWN2zR%8d!15zA8LoqUo6%E`^maA
zh%7ZWZBE#PQ(uOjX}H2122*m-ekxab`Kr5!spERQ4Ok@{VBop(#%lNJS}f=rh@-HD
z1NPIE2J~?1g_AE3jrh)0tL<mXw^9q18l+TTUR#s29YO?g`q?_CNT=F3XQ%f%upqU|
z7R<Mw%NbyCPZ9A%DM=6t&i+Vw`SazQ(sar1tOOsy@=~Ksgj-ODxTard=1yJ{dttlw
zIItM3HYNBUUZC0~GAJIvz2g^iXd9Prb{^Fw?mF0Rzm#i4N+Ml*;1pw3GICNj^>g;i
zb?r&`41X}}0MclCXU|IemGa$?$ieYUTYE`U&4F>*sSi>AvFX!W?N?_DBTk_#k9NNL
zTEjuo<L=j*5h?54tQV35(^JeW_1RY_gY>W(<*(PiZW|h(m;{Fb<ThAC2w7jpexuN_
z+-v`9qWKK!1*etp;HK-|ZogU94A){98%ps_JZpmay?|c-x5}zXS`|Ub$Wn-|Q<EFu
z4%B#E`|a{$odb$FK=^fFT8sN$@$h~2JEdMj(Mb&7@4#}dR}=XCyUhz?e&Lx<J1~6%
zw20W}->ch@(4|hFu8wj*h1{cDf|My_DrY+gxcz>!f})&3$myKYA&EP=;<a8LggU0h
zourXEY#{ariN~EW@+&J#>68P>N#BOQIzGC8STUU#4tQ8ZX&^C=OgF?%4k}mdj}pc$
zE@Q)?Spn66X(8^p=N|jxI+-{s?iqKAs#$dNGW(NUBE?#R!lg7cikWi6{xrK%GAiOC
zoAXI|aKiqqVql|R635o0ym#tzwdLWaIfG;W^V*FS2zf217)>;x$}}fp^cNLnrg0Z>
zN(E&ku7pj*{<1;sS?zb#w(2+3YjE@MIQpyXa8G+;a?B|tUp>ahxun$oy3X|ab{E~1
zQ=W5_zz>I^nw&!RH`yt8Uy{%`rRY=@V>r3&Z!4<xbS9mG2#z*14C>#N*D}h8<I>eP
z8u3M^rc9B@P9VSieSOhd4gm_$K-54#k~_jb)UPZ0oz>?A@Cr-)sa&tl+dtOPp`Vqr
zD&_1ohjv=(PTM~<*sV}EY0czkrud<t2u^+C;@Bj28}-{i*C~V_7(ff+0Fn%a`cBNV
zBGnp)Z2Om7iw&I#=TVzgDgyi0Yi2rwgyhh5#;>h`JCTbl*1}WvZ*^k&@e8F8->pj!
zflM&<iW^l}`oGr&yG05WN~LvPm~tfeVjY7)$?QK$chnJW-1MM8-{90F&Tt`7hywpv
zn(ig^N8Xw85e*%aQ<MCX?wu9Wsk<uwt69MPaVVdL&jgn5Tp979{r2CbQFNUFkbK*L
z1uiBmPZVV=I&1cL)>Lo()9fytqvxHPf|qhRBf}fne?IH+L`5#~!9Sdm)?jAGy)FCf
z(mK5hg<D4*SbjB>5fU0iO0ntM=ad?86i8#Ni>TcpWJPLXa=}rk(LR2e-f7&gi5~K@
zeeUDsq~^{TQI)O(Nvzt`8tJyD?DMjZTsk7p2ugXfvnM0jt$lv>dIcxFmuFlj@3Aka
zE$<lTQs*{r1`>Tq2cbO26=5v_h`c8Dh0Q4B{iIMyDgWgYLQQ^^H5uj#vaWqmc~=N2
zv0;xfr%m!Ur(TB(-zKxs7e5{}=aHMPcOH{vzh%|(mG&jgEGrsUyXvR|i&;(`i#QPF
z?0E(%@};#G2r_+bj8P6G)>N)aiXtTp&b}<auz`e#@S>D*%{X<cZ-1<heR&x{wM)jf
z<oJ}Ih7M}#i=tt&cXGV9YhO`23sa!WR0tq&V0C$A28bOFe_fSF_LT*}b`Su2z$q1>
zftZ}3J}g1<W^H<SNt{_YWCXEuYBX2lS08)+v6g*x^ZB$YD`P>0iD1g&D4*OHB>tez
zNnHZv=%Nmgnkxrt5`vX3t@@fe1bOlEec40iSVFwk&()F7iK?)#eY}cq$M9PXMI?V`
zos{Vtju^DcUZ78wR$1wr=ZZdGS7!lnNP7n{5AGB5lh>M563C3L`!nz+vahc_k3{H^
z0G3}$ACjqPsV_VJNVvMlp!p4tr#w~l9g1k0lV3dg6b^#ZyPUQXojLgtL<WNg`o_}D
z<nfZ55~sJ{sfmwC9xo}9?VAb={R8F!9a9&)!a}hgdd;2o&CPFR=FuCDJ20Kooo!S0
zEloA4o?}*?abSf4YUkvgKr)XXux~B1!p?*SXJ)9hEQV3jsn^b&eCQ$a*?e0L5d|O)
z?aG!v<$#j@teEWYA8gsTH-iw4Y37ePu;f0cp!*Z{9ZeI3SV<>y&Vl7d*NxBi!qTZ!
zx9Q$hO>&ce%xk@NRg+a%T5*<AhMMcL)7ZYp+D@U2{WuPam8He0(O#63wq)te&SS0z
zEWo3D>)DyZc75&kTI8aZQ%u#!LrCzgEv&J<I0sJ}0wV2T5-Dr4?&rYj+Dqz?He!T#
zN`cXV-+Vw9A%!lG9w9i}OG_{Ja&{;l{7NKSQ%)Xg;+NOhdpW&ywH&u<+YNOE#&kJi
zzf&$>BCdo9k>6`Z@Z>(YvCc=(IPEIAWXs8unxxxG#;zm^_OhIz{@wxSu^c7J+zpo7
z%gcwv>L?nQ8xh#5$_<bjH0opGr5sTdJz=;(`j6Nv$}goNDSj5^m!Y|L>Jwr5hC8jT
z_R8$VOW%5vm&ZFu0d-TJWLL?S<~%3$?H!6lVfLy*Np)ZybV_Bn$<aGBR5tHg7y{;b
z2bSNx71cI=sP(O+&+p>Q-~eK3L*K|)q+=5a24ggtc5@+aNK}yrBJRAJbCl#X(<;BR
zv9ql`Yzqn&F1gux%w0erL$58Y3k4}wA{$w?rDxpf)Z|)lg*uveCSy@)o1{HOhY;JH
znlg52;=+f=7B_D`5ZHc#!!AT0r=}pAe1Ek!HD7V<m~@}EBr*KDNA6=WD8#E_9L22&
zhiqw`f<jp{v#8W_qR2JV2d#21^>nzwid2LXwydJ@Mov2{UzBr}w(DwYdv#6?EhR~o
z5_^Y05PMDi=_=mU8F0!uJ*ZS+LD2ZTwzT-*%m?_l!zq`ZXGTtXX0NO0jFnS9I?@Gb
zS&M6-Q(x$a&OGsMB}s<8zPv5>cA&chi`Lr7^6Syta)#bezXQRN8e4h_eW&=SJN4Q7
zJ0{1wx@*+lSUJt{3FkQ$D9`8Ay{9Cs1>9xJbAbpi<o*N-(TO=zA6R?dRy0^4^+6gc
zfaNCQhQ%==1Z-vbIYbt&Dm)?RDH5ntqoxx$mUSn%xz<*drYT%SFjMUH!%j^hM+!O>
zt-zcdEU}ouM{RZe>Hsz$I2;F%CrX6EkK3AxD}9s2iOd1jaY_2%ZE>GA?9AFyqiHAJ
z1p8&t4tuSslXZ$kjAy$gC;ZM)+(jd6odeN6gz3geZ|j=5ItrVi9dy!xl~G+ACL(Wf
zh0WfYZ&<f+{kn~-ol>Bi7$vbYIKA0ziQlB*7NvEHWt(ctrM$7eng8$yRtyz5#E$Y>
zlS+N5w}}CkLeVxf?}>zv2>#DHu(GhY;SCX6u(9?BmPtnOU9eoh)M&W9g>!uzh1~6R
zNh9_&Rrsk2qUL%{4`zt)#TWlSC#w`w&}7-PJL+nZ)r_zPrz}Qg<(jqjrZR_`75QHA
za{h)Gp=YN)5n&L;F$e0VLRzX3P7Fq&ba2x0S(T-PRuaqZAgh`6JRXh%_dnn~1;)e{
z1TS{cKK+<Cm$jCjd!JJ(gh452LeXS**4YY9Ug-<YdW9))P;#;?@^>{;uRMKmhqzZ5
zUR9iQ;pA&~H+?y+U~9OWdL17%nw)j(@3ni%V6(#_>-d`Jyl0qlateE%(<+C>QKZ)s
zZOey6hwm+uD|A$B)Bn8eh5AAx9~|nBYogy=x-ULM-ItFTu*96z_)~XArtB@{gQ;W{
z`>fzTU9C(_;&`~}OF>O)*2HlVKcdI)bBbkh&&QE@oPNHw3OQk*m-Uy10P}e0BDY2U
zWqVs~$Ge8OmnlM{0BXM0w0iCBWf@UvB1cmy1VT=@V!5^L$WmeNC^gbE=-S@dtiROc
zSe3JMl9;(x?XJ8Vz!q7bDlEekBzPRO-lm<DL6CBkNWh{4njztb+*Lz?Z++S4<!d)P
zk0LY7tGv4V>j-Op8C`u1P_8f6zslN4*cZ*?e@<ZzI?HWQ61ajVx8yr#cC77?0gF;O
zEp#_+9_QiY2rj|C*g$!Yt_+s`KD^W52#-58@k=pF;1qgp{f(=Rk)Lq&%t6m6!z-~o
zgGx{ZY;LbYjiq3VQ#MqlDA~hw_OCEuwX`6xk9L1;7h;hx@sI;cridG(mfF_Z0tXf8
zMwAYz!gLVG`)>F285=I`BH<)fxX0iUfh9p9<638=Z6i6Q+^ddqfI>k|xiEy%!$!-1
zM9^d+KN@F-{0gYa?Smq#J-&r(6P-k#xXCwiAOH|@Ziay!OPaJ5nZbEZzGlP5P0M@i
zfo9GWe<3ZAXuP;_YEt38VDS<gufG~ac{t@1bF<Ct;#VOVShv0jbXB6}b4Y*rCTh|Z
zJ8m7sy8?j(o^hi508$cMV2UKirgu3tIl3VPfnFJbC$lqeW~K(5f>xv}*NuMLR#!xh
zpTd#peO_VVCdG<if3odmg79@p{WyR?u~=q?ekAm^I1|!sNpTO>4Y{o~G6oaq0Ll+e
zhZ$bJ)H>)Dhe;k3o63I4D}r89nnMGs=a)1A50#p9A&?MKwSLsZ7m}0vU$oaA&Mrk{
zcDU06&}lQ+*|NpnStddOlW9ge;=ppXm&cj<NF9>AXSj&a0SHMJFI;5rs`D5k6cXah
zB?p$ckB)zfy}M}<94bCY=e&U-B<u=?<$`%>mzf7NnJ2x%I6ct2AfO;~z9&DvZ1H^O
zF$Rt^OUUF((HwRmruG7B@n{{#?IhT1JH_1Tk*O`@N)`~!h&@(b6oN{_=*_C>YrU*L
z$~t?BvnE(7J_I7zd-Fp>BfYKO?QN&1O+aWjsTSW?$DjHh*oI>cEN5u_5cZw-mliT?
zDeI^1*#?$F;|vWZfuV^ZmcuEowDz54Eb#3xlftCpXYbU+E>mA6I*D!ju13C57MBPa
z$z8_yLt=ntU2t0E0fxN*k8#rQ*azyKQhtj)=<*y$ne&r!8V)H>;y9{F@hA4dT%MIg
zlkuoypwnaDU1fpoL(X%-@bF8Ze-h)!BxfHgHAUl{`ZdOX&}$7C4yDmwY1R<n?ZXu}
zMbRf=)c^=L$EQZ^Bh9KdsRQ|KMD}yuo^k36KEy?9h>Ei8X?8za=1@9HITVsx%L@!O
z5{G*p`S!7FH?yN$r6}h|U@<}5jsQtkY=y3eknIoGHjF_ZuRZ40y;lFkpaY0*6g6-#
zN~!etp7JU&QcCo2im4KhBK;F`u~RRKhr=D&OS$i1%D%TQ;m|1dMCIIfK;?)Cl`Ut=
zA$z=;>2zA3y-|glXr1$vyW@%ahHzQbNxa)DOkES*CZV-GSx+kNl2T%B%36v??t7s<
zRYp>i%F8q_(Qj`|_!*}@QLleUbE_+Hp01n)sk6Hs`Pac?0TktWsM8s-Pc&T>n<?k+
zXB}8>tPMllX7zogMY2r_Ks@fi)Pk>b&xn10Gqkg?8j>jYJ236uVEyd}iqs$!muf-B
zjpk-nX(a19_&dGdX`AO8I%;{KNmKpF+MyMkxK1f#5b?Ll{1F+*V9kpl2&QGO*uBWa
zpQ>-et5dozoM;_bpo*06bsxtLwoliAVDr&=#tGxVwErF$vmLb;>)?o;Qe9)tKH>Ng
zd#1cuh?iOq@#J_=IyJd8GTkWMQQLOr3~k0$WxG=dd{3TBLbR;pv)Sf$RrF?OJ;O|8
zN(A<W*P7hV8Q7R|!0tLuxGdd0aUqO~c<ebf$qkJUakE<2_SD^UTdO-bGBoZ0k^vc=
z?6<wmtSI%LJpV2@utHg-hE;zrGw+R6$M%)qusUH1JdOzjSh#QH%I9@B>UGefIGG3)
zvHeZ2rAU)pgPTHRRH!v@O7QggmQXj^f$}vuEprm;uLFx!MoR>i@R!quznoJCo4F=h
zmiH)!!5lk>Q=faCS;i+GhpS6=sKW57k-OO`>^{Jfw-OENPHf>)An8UA*Yk2SD(J3L
z%#~lx!QbK1@~$1Jg#5a9IL|rL8`s`$N6S}qoedmZ_>Te8T8VW+OEP80N}KflkO|<h
z0~4zeDK79`KVCaa*R`ifoZ!InJF7ZvJ8=ym375b;cvxLY0Mr*om2j>pJ6RXIY+PLd
zpR7}4!*}XMPw==JwP*8rsE>`xbJAq#MOZ}nuW6$kw^OAiX>x3G>LH^}O-?@FQU(}q
zwzTK!w1C1g-(Q6B2Uau$<k7%#czVto;n$JM<6?2_Bh(~Et7`#!b@k+#=6x5j8;D$-
zaA3JiE8&#$hKaNF6KE<m9Lg_am~-h19h}V&mbhG1-h}<zxw^0VrY1UF?lAxmquURo
z7fM0ciu2h>S45;84uBfX>=@;8f$^bjXg2LaPV4&HSKj74=g@m#Cw6u0VtHvQP-OSQ
z3A0Rx&KaSAWQ$k;yHwW}*DC}!im^soiS!4OQB4N0%k>o@>8oCh5ND24lj3rBE2d@S
zfGcxeBH!Q0vjgRR)~SgBV|!xH%Fpkz{a_v6&}b2vrW{yw1Esy}v!IG}8oQ=<+Yi-l
zg#U}gZe{P$G1b(By;3}fcsSjRHeD(8t@x)nrQ{`oKB<N1b@=VE3<vCLp0JDHg?SVv
z54VQ;ZKfd;QoShJLB#Kz2(wHuUhDUBJA{<mlB8<7G!!WId|j3_l!Ch#99W#21eo#R
zA1+@I;)JUmTCGel7b)?b`kd#L<Pz_2HPvUz52QK1<~}4od}?~XQy&!e|Dz_Glb-rV
za=bUZ<*m+ROnZt$3SJB2`1Bt8(WduGYa2Qrudwt^)|D$K?Z-+xeG<byBYDPw<>A0o
zqu}PRvY1J>4hK5>@$ywRPt2)yb&_~$0$oJh74|HB>Q7X3@1Mjc>p`bnjvl4=lCY7q
zMMQizX3d|h%tnkx#7%C1dC5$qh7E>62jvx~Z4$cS$pPZ~f2s~9xLEiUl%v36ieP`j
zM-{DuItCH@>C%xop0(6SLx&C*ew+IE<s5rK*4xk2&yqC&_kW7zU=&vxZaT~qiU68(
z#PgS}Yd@QvBcwYNa&Zqgu<ZW|u9w+rKUezCWJ3Ou+W{tOab08}*?oikeCbtg=)jvo
z0;$O(V&VLS_6rsFS|qZUObwurE;;weCuFDnVm$%nkqZNjccXlE0Sn9V#M6aozm%^f
zTV{K^Qz)~UHkONQ9jSkh*e{p&>tMMIJ|oL`10$?M23}c9e<feuH`(gvZZ=BMD;*Td
ztSRNeuV&XF_v8T_BJvhM9FKT9pc%7YtJHznA9G4&m7tqNYEAy>*DIpIL5Q*h?~3%S
zc%HLpX{gWbiFR9h2m6g2=?J;U`kjI%fs0oWb?i5ju*_dJf4=ja^So%mJo~M>f;iP;
zMy6OdKyoGMEPgxN!#PflIE82+QYuO<kN4lnm-jE9SnfRM;lPDMiYNZv(musOov~MZ
z7N+3(&_dhqH4`ZTPU^yo4ouW=X2*tm?Dy+%P)AHSrJQ!uZemn*A;TZkVJ)1$z$run
z#FG+`bKL&0?ibh+S<zC5gY5zgI0h$6ea_39Hm|ooN+Pvv(E{f&*o%pYN&DkE=OeA*
z?JX2T0OCXN78vDt8PRBeQeU)T{R*d`b?8(6jA8rJteV1Se!wZjZ<GHJXL7{;tj^!<
zxSu2VQO=Qu*rr3~wEcPZZHmMgJmxBs*RIe0qTmuc4=NCyiOD(a)FiHKYLj~dB>c-V
z8C5wY=S~!N5_#f`rCu+-N(d@6?XOBNJ_a-VF^=+&BRm3=VDj>t?5~?~9K>ZN4hOF=
zZE&UFV!PSjH0`A636p^XlXwY(Vt-q@lzxP1^Q+2!?3GS^PBq#Rx%cUk=fA7+<;u?J
z7TjuKVcHHbad2asXyz&V`&#Y}V)pKL%7xI}ywLukyj8x<^01cv6IhUVrhC@V6F#v*
z%q#YfWu8SUrNdA%g8U^l%?T(!Xr1ypUO9$9Ra_B}+)vs+m2Z;5DS~nM>%fvtgq_4v
zhQHN6XRDbV1ZcKooM3+fD@Pl7k(pLGG3foedrUg`e@R>=GuaX6If?W3q4sdU{cDbz
z>kh~*LLpHP4n)3O%A)yiwO3&g7KJ_Iz{*S{zoOjuDY1kUWdB~@!?-wiLSm>edHICg
zH+{<fBR|ePF5DtYp)4fn^MwaY>3<fOe1l7lc?YH`nWL)tuckXJJ{?iqiiTgas!$)q
zG$&pvdWA8IZ2xz@AH@$R6_j}dJOCE0Bk&^x5JlSJR&Mr04jZAB^Do~f0hZJ`+7C{B
zP9OA7yo2<VpH+U7$FW#M#nj~GZcb7T>Vtg`e+_CBDIUj6S;jdW3VRT?**^P;f~Nx*
z2`DrtmsYH?$lB*LHDCb4ZQLiEn%Kl5*5~YVv!ocx_Ex7<rs^E$j)o}Z6j}=Vyi$|?
z19d8iMsjM(`%Vy-|M|^;<)@mqlsK2en)8fPA76+|Nf@L!lT?X)!4t9Rm)*R`c@%La
ziT6(X!Y2yCvPRvuQ!Ho_{(jnj_C?toi`}U6TyR)HKzpRGu`ez#eMhlTXW)Z?RfbLM
zVE8X(kg*uRefA}#UcZq%uHBS6BB`NIJM}RRVZ68nMMtIq_e)C+DLmNW668|Ld&p~z
zb|?{Dy2&Rggnd~qD!$xk4Jjn08rFf(wl6O)q!Z6SMkkuYTwZHrVv+~FI+6V8S2W#I
zRtgPI37(YHoSGy6+uC~hS5|ad)w97V#N_3hQsVlm0wb*<e7G{k9q|ea6*Q9rhiG42
zCJbj*sR^WDqA!i&)oXnciuA@7AAAr5lTz#jwO`%#gRpB9qbppv4GmE!-81$zWl#!y
zhr&*idynZ*dJk$6E2a;{6zMYhuWhDH_g>5;A=D&Q1;NhYuXWPCE-`2wp)k&~4j&wr
z?dzXN!d>}Z=AI9-AsCtFH5^TyR{A&r1f0##5c!6GL%wABsvhStFfC-@z=sS9lmMWx
zxo^z<)8EkrgI-EAJ66c$>YExYy%Zxxw_Wf?oXSC`J{1JFU=Gha{=T_x=#{!4fl}g9
z`$3YbbnRQ}a&KC>(kV35L#__b$hQ?)_JVzDna>m>ad^ap3@lpbWE2X(0`9YKE4>uk
zAU;&WPeoR=s!n~bPTW?S7J+ZCzkzvC@{9{wH?&^bODE9aQ=jWkR&lL0kxcd-<u_H!
zrt6u}ss(~`XRcF!y~u{nQ|;FD1(wJ%$F2^Xlx_CHIuUD@zsV^qjvt2qW>w|QO)MgY
zG2J6$(Rj(h=O5*@TxD$6<yvo_8g(9XJ;u<2obM%bw!NryI^U|3A}_nt{m^17s87CQ
z`Ksl1eVOB+Bv4E-vf&-;wPqh<3q7NRGFI%Iy||evZiN9muwXf`pvi>uLH}OTtfriM
zgI#oBd20Z9+{U=zlJpXA=3RSfb|-?uoI)Xz2r-OVZp|3SQGv!)Hg-d98`9-(b)NM}
zr7ysH?#4P0h$Sw{#qAlE$Ek^h%_v|3{+@r#URL*h4>`Gponr7y>Br4fa~AzCzchP!
z?Od!gYYy80DPmcKn}?nH+?K8O_#`oWoo>rsQFs32=rFQryWfE%4~(!qz}>X=$|u6h
zy~S{wTP_@6%%^g@p0}II5UMpxSF8ccjW#ge=2|d&Rnt59t*0Jn&KR8f+!JCg@@%wu
zb<!ZRQs^iPTPCBbHH?gNNJ4=$`JB!+zYv<W%bmwUXcSnZZ6rT``7*_UV>g#+1rwpV
zm4az|XiMm@)W)Ng&fYCZRJNdWQlC<AAt_oMSg!MU2N7~%`6l=QPlVG1v_FpX{IXNu
z9Fs!ZU>L<3+oC#H4CXq!cJlk&5w#}sgUew;K65OoEv~;&A_GujH{(bh4K=xp>xU)=
zaZ<D;&9>@odnw?gGd^HTYf~=imZH!B#8PC=EsgSmEo=IHO(+0IO|8jM>e^aJ9)#<J
zy}EhNS*V51<Wmkzhe>z1W3Q>-g>h%Ibsd0E25*&a+;I8Y24!>~Ql9B~OLat&JM}ub
zd!Y!Z{p@w6i%AWW!IXC8pw}8LbZS0zM<l6A?Db6-BSHaA=#m4gXaf=mQ<c*_1K`Op
zZEt9hKv_d#D|>;#HFCQm{}IJ$Z>;0Hwbl24B1{1!^(eYpc|c}o)467O89*K=bCsyn
z2nSmdcvh5Fm7L2hB@81@O<6Z>R1N)-t!z4*26gVmbzpgL1?kM(Xsj~fcB?8buj0mE
zr<9a<?dCgdbuBDhw9qN!L5UJsN*DbVMjo&=<$Ln%6hDBIivvscqC>Jv)b=B`w$e8Y
z?=W~2OULD^tr<UKw^W$iIa%S$)4Ls5uFM?Wgghm5N~8C&b%|UrzUc<%G4X$7sy*C6
zQG9Ft<mzDrU@yg22LpH5Zfgc4iJ_)k)(t4NhF(P^VG~tj>nnvrYayR!%37qE9WQ;m
zZKzWm3c}pxI&=3sHEMOkjW@o+HdeaMrR_3=5S12_nfa)~&^!H%(<){M&M%};T-)sS
z29-2p8ohEw?Su%Vp@jOJQSeqLKl)I+qmGdomGX%D+L>3~WkQWkY19JcUc9cosm>pF
zsSV+5>;MuOHjs*CQ!Ol5y3{F@)sx03SE8q!54w;2fKw{ylJc15FqIvf%Q(vY5OJ~<
zq9_j-r#{|||65r3PEQ}OJIim9;6$zjAz$!I*e0ENjb>GQ3%(t337p-P{p}eVfd!;c
z=JP@zuY4=+PV88>@vi0T*R67jrNd&GQWlkV`iR|=%hw(m8+4wdRiZ|ATP)uxySGkB
zI5A3baN2=s0!jHq4um(C2D$kn!@5^k@wSl!Wp61MA}kkfNV};INtqrB`(T-9l^YEG
z4w*7wx9@wRr0ZFUuL!g?ZB{_bN`=4;H9T)`&2EwTg#>q{FiGZ`342@98o%`*zX!a+
z<gqCjacB1SI+hf4bnA@+%k78TA$z>v-cdITA%x^_cFGw$hXVxk9>opzvsP_1(jl|S
zs8iNnnK$p|`PSP^c6y8+t)2u<O`+F1ni3!8Sp2N7?iF~2R87d=e^wBy^@>o$iv|8U
zQp^N9yA$8Y{yGm_%%O;3G=^SboC{)v<=1QNX0595sojv?;*B8cw8}N$mr7P9QEy8b
zo1B%Uwco}1hScrU6#iLm=r|?y+d!F=C5sk0&$%gwC&z5CB5er2ahD3;2OXW76mf~<
zcwo|o>aP|oyxA!P5=)TXtz7*aO7U#>m!66NkbX{@E?~ilrh~RHQMWkG&U+kPwzYoX
z&D<YOA(n^6ESVG_*SJXAaOpq>y-5zPH%!zTeMU<E_vkOryKSWD8dq%r!@0zP<(>_T
z5)+Ov6bz<~mYPHgsZ8ZivXgisYOV*tlHJr~K}1l&C>km~f*G`K(l|PO**Zl)_~6Wk
zW_HZ}bKr$m9N~9j2XF+}wzRR*%UKa+zXQvSfiZ?ix(R!r8NO&^y7!4sAx~(7DaygC
zI~UH_c*O^<=_YPQCLciMjcv@(7!$hncj*;}zU7bui%!bpPFmPQ(z0r}HqmrnnN+IO
zAbEkBSU9b}?3lqAhkY$oDw7qfR`+9;bBYDA#f64@k8R5hjW+zk)sMp~QZ94VWI0l2
z@X_90w|*BK-PUr-Wi_>RQk^7@=eRxCZ1+qEaY`{>!U_hYX{+@qSrSYhocgIyA(OGx
zrjki`<P{G)kA-r`eONx!NZagKeUOE>hjOcJ-n4w9^PFw@lzh2QLI0A;7Ynz~9xlU|
zmkm~NJoZjFFfqI9Zr^0@%*WBzp!2BRI7Bijd7?a$A5mTv=ef*}Y(8=o5YZ=5fxRn#
zKS?TD9?SVfx;5|u|5|%@o&4YgWSrRNz~W^=omdcMnn}d;hVPi;TQe*2WGkmto+=aD
zBdTk_-ZP78y-E}50Lm(HB{@G+Le2cdquC)C;tm-Oi3`HS4@|@?^KCHrzOFr%gU92h
z?L5cJbgs+tRQ!h+-FutSr8@W^7!54ZeqwxT4BMx@uMR2@HJ65E0`um$Bs#71Gmvax
zG;*@OKPPtU(1`OG6I%~h5&!S3BaFOCibs+Qr>-}yHHA9GaU9#nzN@?^bn2|)(dp+L
zSd!qCgboha2lA`!Z5`)PG<=H8e{K6<GY7ilAvO`+M~Zi90{_zdpX6=t^N)tv=-Bj8
z`|f(6N_uMPmThPQK)suD1E%dml_!{!C%NlDwaSVo)n)M<Fdu#*Ojb9%!+9#`7*W{I
zzX$Rdr=)$Pu1MH>VzM1r>7h1b;+rN%qKL@hN6T;09uqva11rCo^-vSLByn7B1s|(D
zfd7#QufIBp4m&jxyGcztj)UaK>)8tlz+k`$iaP1}M^Vx!ie5;ms4Q9~&a7w+4EDGB
z?R%2yv=9mK>}fmYGGOgLDdR;76gm36rADMvmbxY<fK!v(5g{~a5OUPxiPk-vH{I#w
zxl0}vd!lxrGrAI8*j5J+2M{AeiKN-mc`D3Dl@_LvM(1Kr=1XUGtm$)}^Wfij=LUPK
zE`}6&IoS>@8>}Vv=IN$^^j#n+h8AESWd3Fkvj6KR`$X*}XN>Y`T*eir!nEbYZvwOL
z%e`}F&no9Jl}tVD@!p}9eSbY3`^YfV##L+B0cm-{!r#~4-co+CAIQzPfz008B@~mi
z%l^vrr_+Nh1YswzPc|E8baZHNs9YTmRK$Jh{6+SuI@!0ZxZNp;X+;ctqkXyzNz%1q
z;oxl?m{#WMqtnM<VLM6#u|<qeVQLOcFBl}fpFL9%W>lfbaOeP%AslTJJ1Zrg?W~gk
z*W3?Lrktrd-=yGFvH&QpvhAvGYW3sf!NdSi|5vYEWxMln8<u(=_3>8UP{;O^&dF%1
zEu@sI6_z&%1m|UY>#l)OP*Tcp2G{m=YO*UsL~M<A(Cm<P-uBh^BGl=cQ!f7XSy)D?
zv-50!d2e3rneSzb;O;|>Hh<6dNEde?OZ|ktAlXyW2=EV_u!H#}8S}V^W^6)Q4PZ%f
z*alT8u8k)aSL&TZ_1MFegXm=32Y|emH$qBgk4t{R$s$XS0@oVugErD<C)ZK*ek|%~
zJDgLzaM=>)u{mbs1i1@E-jOoY?1%`DJQJXrUTgNk$x9ESIKrf-M;n?K`Dr9CQ*%jK
zUV;VA-_t4&9+KtbfqAU?5!Lx3&`tuLJ{^`ZSO;3g92M=9C}509t;FmczouUErmIVz
zK{+3(U7Y$tN9C8MMBMV2A_(11l$yLI3V{}OG%R3FO-zgQd~p{C4A{wvDVPR{N`|-I
z;}xbAkZ~VhKYO;c6TdIL?Qz)GS=bSUZbLfoXf5qH4p_ZBf`G+XQyD!g7)nLB6Zg>P
za^=>o+2}lK>yC0%jN9qlP~)vJcA^qw1JF#9<Dm~9Vmnhvdxq}G?OI_G|NpDfXqyMq
zmc0MY&eo1XG8F7~35pFYSF#WNV%yGDqI2)$R;Q3JXb<BE#&$nn8zXObwm8L@T)6}4
zmS9X4cHtU^NvDpq3JU+u40@5`(`gk;iHloUQxs|$@`+un9oTgHI;Sw_<eYPXv|Xw-
zK^;<85+a3DQ&wDw)Z_!R&o0-O-i@Y)LcH`^qd&ZlT`7wuuf6n~jlRRVr~Zrj(w*i)
zBzi55;s?t|S$t01KsE=andR~z+==?_hw5ZaKZg<^wBQa*6lZ3~)lZ-{o~~55HPV|Z
zCI_UW=;}uy2p>g6y_y5Lb&`v@oI-RbsYWKVf<b!zQJcwT2!n)tP)O1Q7caY1*z;y@
z|Fh@oe6XPu$j<iRFLC5fjk-s6rUzk#f4FHUnJu}<oR?aYHl$Rk>sUH%pDDl5j8=Vj
z2<uLbM#c?WceREm+x8=M!-x%LdZc?~VDN_F58$<4Uo4rpsgYDAx^~=tv=DFH@mXU^
zfdr6Dwtbk%?Z@f@ptqE8cLLXMVCqGO%q9^W`QxP*BDjh`i8Qvfl=48l9`#z&<#C^6
z*M8!fDao(~(T)>NO*$h9J(_`#vj<uxSCMFCKbfy1Ds^ITYS<|ze$hk<Lj1@>*>K8!
zs`OnK%I!)`idR@3r1DEVZ$DkyWyUVvKqMTncx%aa7NgPpp0=MUHQ+?he>^C&t4>YK
zpRh$*0!5eoY-K@SeS`BX%vk;47{{6YTxrd%2JCZMPw#SIiTx9Uy}cv$^Q8&y7-B=Q
z0AKJ5Q_?7<lY$4fU$|yQGfF-jMx2`DQC5#)W+WJ>YrmL8bm{!X&SUN*u0R^LU#b(y
zF_=>*=M$1`I5n{u=16Z0S*{C7%YV7_leBa=q)0t-!hw}@E8FdZIHXD=kiveYQkQkB
zx=x{xpQAy3FlpPb*2AA|zKj13I}ou++gcO$YfWEMD=g3EM0skG4{ddo-OGMG7hq@`
zp@BRnC%$0eV*8Ccx#1HaygBC;mO51(rpN6!o55wadn959mJ<^0R#HlFB*xLKwZCk?
z)%@mE=t24G6-FKkd>Lfoykfsy$JjI3dp~DY*MVq^OSn+zck*#)V%&Mm^X6_+hT@~z
z%<Zh2G<jtSWV@XDLUON!5OXwm3MFy;y=J1%#WuMxeaL~84RodPQA_`t-!C;%V9LAX
z1WFEBK2JF{4XVrLB^%=()Y)A-IpGw_ib`V^%!~bDeI0Wxmjx;Rl0s0EtxHI)#Qi^N
z&d~=6DjXkj067WEhhZ-5k86E*7|k}^CV*I!)Gb1c;*exxmHz!nsn57PC96COuFfj2
zsgF*G@W~@-H`w#1^*3R{*I_HVq+m|*mAjmJ^{MC1l`B`1MB~rOE95_=G(>nkIIz^+
zpy!C}jP~dCplM-grBK!?F_{B^{+{RCf6>%rn5|OOCL>vE<~p^6sA_*%exs5UXbRm3
z>!jBj`8FtaLh=EJ3Wva7HOnGL9wDagcv7S%H8CT)HY;o|uquDu^h%cCR9}V@o!1(j
ztm0mAAj<yQ-_$R_6QWLttQW!Kd9C5A^psz+CPC}%Zxj1(xZy_UG09Y4i|D$)t1pAz
zE1`{Uy3A>*HL}`dsw|y}QdItZ#V-U6hKTz5qytNI+CntiR{Mv#Ot2oQu>m4RZpDx?
zeiMgg|5%&cIgIgCj=Mne9Rrah;h*Xb&@~Mdk`QeshdH@N|2YZKym^ZkInRNr#hFb?
zb9yRY649Lf_Ah0ODMRd|*Cl1BQzMZ|DJ$(?Ykv@pPz$#E9Z1<niuy_Y(7)AvqMOFH
zz}f(aMp`H7R4}*Ozc(v=hON1G<|L9Gu)xf~tXLai6w<T(M`_LG<umcufhF6!jwH7O
zSdjj+jvwtfOu19exmqvJa{I6PMYg=~Q634v@*QI>(eAGOcQ%^YkvZ0+Q+Hr<=1wRh
zBZe~g$bxgi?8!pp*KO``3c*pR8NDw)ROh5zX3{ktp~q)ES-Q4n-KL(+PBCflBKzzo
zW4cGVVaj>V)1!|&P(J6$xsxN#r@GqKnjoQUF$t=eFc6_Wx4vK_OGQCsA@aS1Nzgv;
z$$F^#PZPBE<TFmaw!v21_y@Z7`K5J<Z4PJf3{J_-K>hW!W>REB$$cuFsirAziDxB~
zu{~yA@MLfalX0UOJ?d*ap!_-bs4S*>Ns5(~Q_ZXA@cFSX%;{T;*0MD^<rIaYSQ*aQ
z7d=@vL9Ze{yG~h1&>?t3X*It1$wG3OKamVE4izSqt%9o{#P%hn-5wZ2i6>m_;J^Ys
zW4QECp;38keQ8;AY4RQ|L$6Gc?yi!O;Gox5$@QW0L}ZQ=pLfX#Df!<oOI%oi`y-E<
z|Eq^#oln`9m-*L;tE);5aWbTBB<G*{y31hV<t1~$zM>8nnOCVu&dlue3QL9*BQ%u%
zE9-Dsueh;;X#p0))%Vz;sGFsplo<6@&6gr=0D6!U4lFi_2BeGE&glG}^|ZCGZuW()
zdaB-oa`zIGE5{u2=sB(8n{zMdiP6~$axOM+$HYD5lyYmc{sS%hnu^Fp1EQm5t{#N?
zIV+d4m5!og_O<nwb5u7ez+#eQvTD6}Eooq+;9-@kJ>I^q{619&i5X%#OB{pO`eXrS
zS*CQEIc#5_Q#rFkrU5w%A{7D4!L>(vJHz%3xo_oBi`QzKYeIn}(T%Nwh=d{j8=IL6
z!3*r5yP-s)rjTg%Bs~2|*u`(k*__#-yS1HCQbAk@`X|XO`OURgp<(ARBI7Wyn8V~c
zwP<316^8mP<pXm2Q9{C<AP=w<1NuAL@h)t^Lw;*%CO!tqiob^)SS}0`iytvKri9(3
zEVFN`y@qOXf}<xq<-oEFaQ(u4#X=@_e0%*I>5+B!Gvp3X-3~~nl4hT^@2JCG(V3WV
zO3^|Igw*dVY?v3;4uv5}jF;j@#CTC78ohYQ0=up>A~+?CRi_L8vfrtRT_BOS`i|%X
z6GJLC{Y6g}&xfIrvDVOdxJNs{oGSsLQXFO1moMlP5f3JRLLIIroEpUxz4~E$aT0f~
zZpA}~LQG8_J2FJ)5qH2|Qfg`lRio^5YVy0cNnx%_0PLmBtnY@2PM!={vK3>U5Wa7C
z@-y>9m4pbgQ(i>u&J&yMK|hOdQutSekcFny`CY!JZ%Q{^*^Q-RXagbm`z4PvQeC)5
zO)QMOZshxi!P;Jy(?2r!kn^Zxkti*g@|QQutjJ8=X@A*)l~quZRi=bROx;sr%PZ=p
zPreZGTXHSEOl)%Mb)InV(K35w=_Q8L@aa_iR-lDb6JLP|phm7EI`z^}t~3cZ<y!_v
zTK&#*jOb(VKOckX`Z(TluPSd2n#w2@qM<`&pVyjH1!$`&=^!Rto7b$55^M!_IItua
zTiV1w+5G0U$2n8PnMS&!>B~-yFnDBS+-@#?pR1;+WDzdfYrPI8oh4EXX2`OV=3lY}
zbuCwM7alAU<y^7Q>MQl&QCnF1#l_fU#itk;nt{xbY>z>av)LA9za}3VavtNvQ(w;8
z;#_F5Uz63JLO}}A;Lr-%lG+VYI?MV&IXzTDJ5}}qh$uJ3*-RZv%bTR+kiekivcOVi
z>ggMV-I;jEmeqGJXG?O=D`gQ$<<-iZL9eb@R7iz3R$@`Cs8g@8D3b_2;IGMj`_9)o
zPq|(M2@`z#+Ga@=o{q|8dcRkgI0eNk@z-8g$IleYoP>$WD=cn+10BfcY_Bg<2<HON
zK`CZkG)~5AqD2x(?G2^-ax{=w3ilCY7LBU5h0oUSl&%!SOzULP6WbeeaaLhj8FUJ<
z7o;PS&+}RKqF&4q0e5i5me=W(;N02<Zv!OHoEH8gc!k-DLe6mi?K*`*>E#$x7E(N^
zt*nd3vdRfVF22B`bws8G{gI1286#~~?!h)Gv7P7WW2L6pGH9zSAjE85u+{a=p(Ymx
zyvKnFEp$aUQ30%%t*L0SZHRmDMxAm!JI6<5=3QF>plxq2NZ9~ZbBQJAj@^=-(jAh{
zb6|DyxZ#*&>vATrTH^jsAsfI^DH`q+8|K!gr*eOdq%K}zk}{Hm+i$nk-$Gk-KiUyo
zCn`*NyCtk12DR4LE)!;hPws>RD0`Lr@M4l;QDGD})P*EGN13zw9Y_*e<PLc)gr|vk
ztAb}6ODFcBg@V~bvOMj;a<{?I5X#B4-k#f|{{iQzEPbdaU<;Ks2qlx<(bUL5FRQS^
z2l3!^YP8iDB0d84raFHsI)hFjUYGPAE)(8ETE*kGDVr&3bkr$`<a1xdYU`=U%oilP
zKN@Hpxdjwd=G5oxP+wBY28tfD&E=OMEVMVd2PhSULB3u4o%-nK0g_8xwmY+gO&=QJ
zl#(p0<jT-dyQ{7p=aR$>{yMNAZ=5omymEGw!dnSmCGKI<f}HNOyBq8{7af<~3|PT^
z5s$i`GCTywW4kARGqVG`=ax44<FN(<i(wXdFV6EAhTYrDsoXK)KOqp}`;R*{1)b%&
z6}DS@bAf5W#7yBr<rS9akAg1`+gqA;7@zc3xE)~8?=*M7UMkuQ)wq1^4o~l~`^s;m
zsV-N}JFr9`nKqb{CgjlGT5CW;S$}s+$HbiNcWTlj6-9a6w<hh&+bVz5%RMI~^X95d
zK)KuTRNJNinYTarnanVRd7H=~w1Uz#IN<1arh(FYi)@OPAt<0~en+DF>)+u##f;^e
z1W#+NioiV&3_FENA*5R6D86jH*`R+C9{?|fnH>ukEVjPdI><%bXyOeFUO<wyB0Bp^
zyYv)|W8Q(q#Az3yFNQ4==fM3L)-J=z?L&PCpi-?xjJZ=E`}xB&ADoH*Y|ETs$^3ho
zMX#`Zp<NPlx!(p#yLvvI#a#+mqIvO**TE{nCuvy;C}UsQV6$Z79SJRQr!7Eju$dk6
zmoAxSL;3QV9g4wFstL$dfiVp-F9pA8_gB1NtLkd*xNMtsy5iKt<im%`Fb1w1vaQLk
zEFT+p9`!j)zG9w-OJnKPkpDQLFFLS9#!!BMq_Azt%8k_h4s(OMGsVO}RBPnZOa;y8
z?5|<KbiB#_Dl$$@%#l_}40dw$qB?0ZQD)cO*2%-FuQi5G2VyhjMBhH$D2TOltP<k9
zu#~)lKs2OH+(d>?%>#8lM_LcHol+T0<U~yOukZ;NuLu%U0!|uMM*0PvFmgh!_3CJ`
z??=7t^=D^xS-08ev;MczArzNxr#{vpb;Z&sAG3)%f$TXXAc}R{2fv~unQ{TVY?F16
z8*rQ*Y;W&6kZ6nIp^!uAf)ok4t?7ua;Zk=MRU6*1JRfxGCB$%ZcS}}V+g{&|7&bXR
z)g5sln!u}vICL-CgL$rOKv(ZPt40nf#1>99{Y}Rioi%B5r6vZOgcp{QoP!UQ)<w8N
zygWtkxD~{y(ObhOc&k0!AS{Y@Fl`@7l?tmQ<|Jyech2Ifgdb%JbznLJUw7U058ES2
zVZZ`#vlPf{1E2P;#G#N>wdovyrsV2JvH5%iZsOf_t;kxS^+m36V2R!9+XF)+bqTK|
z&z`SCmCbQeN(&a;Ja6e@drxk%B_w0vQ4;TMgZ5~%II`ACF2ddAd56~wxyz|9_?aHG
zjdjT$tJ4QpD4F#Nr;}LT{Z386EHTKEU>~#hHn2lzB^1@R+kt76V|eaKy<qRlHe_CD
zj}UB(-ekmq#rMnQ@sLj0LJ0b`_ct}o#RR-p`HiT_@5eP;G6kQs?<}pWG%3YrD0H2g
zXw5kYk_#Em+ILkHBw+*TOyRA8gwGN)1p1bMCNwNE0xBIwI5@|zr6`WM3XKYSc}e1I
znmzV`ob@|zah`Isgq))=1*Jj2&&xho&#7&#A*Yby{npX&ME~y6RoGKb9r3rPyuy$s
zz~3OW_Mv7-$2cA%#@d0!C+SLfe(F9Ic~u{_4_6eJbvWWT4HNIwm*qrZ#8;I`_ajLt
zu<4PgmP_ZGCWqnub>g0}k2W9FHcahf*p{7|RFX{g5<Y>y$EbZQadSmy%6SxSZn(uh
zo^vt9t*2ZuUrE#5Q1c1<o;pvfah0b~_+?$28|*-!a*crh-dZ1vmne{_0mRxLaq2Zf
zRCi=T2k&HIf4upPqRlzYgDsQeR;Znv`kYVsU@%|LaF*`pL^@<ol*6n$<~)i4<t8h(
zXP-S;(F%cVpp7%314(8K?v|u{l2rND2~bA4<4)OAb*ff#bL>=ar+>-;C8IywnjB^J
zNq5?GRiVJ*>txT7)+D87>hp-=RF7z;Pn6}-JySYWmG-E@lz&dDn@dpc@2iu%hI{ZR
z1Q8bFLH5&T#dU8$qGE+fASp?`?FXv*0Qo?YtBhoFy$e+uC>Z^8TFsu+n9$=gsZW*}
z7KakLU2r+{`A*ICOe-&v`A6$f6?*BQo+c70!7kgUvfFQc_>ueG_7*QCJS9LX`*axu
z*iPqy?e_|kzp(CXyEJ{!cGL+Z#AIAob^s_;NlPVdB!Nuwm9uBcc<FR_sLJ<l)(1{X
z2AM}_RkjqD*HW)5Y<E_+mqamSC>2fAQ2;oXyPr@`(%BB$uEej68#XzQaZn8mgSEE1
zS=LM$LTHKu%e5bAf!d_`-IFI$cML~No~6B7gd@e?rbm%F1U})QSD3a6m)SyFFWbJ_
z9Waj$;_VBh5S9?_WX)hn2=27~%{wE-2Fr2IffX)3FO5{1{=7O(XU_#Y(ELX4kwIs_
z1B)-{uodi3m<f-di#V9$m+2H<ic*<XX%B<!*`Z28mydS0vvQP6Zx!j3R4McG9iFp`
z<Ss9#@vv{RX7Zn*UPt-L%|xFZDPO{W224&3E3kO8Oci^X0VP;PFEYK~j@CDGI@Xl5
zaxPGC9A9zj<Cmcn2b*-eRO|~omf!3l<D~PPd!*A#c%Qg)B|R-2Nd-*E5)MpA(HT(g
zCOc6l1zt>QR=Jh{i$!2ULlD*nQ(Dj)&e_Sj$b<N2p=)hBkepiH!UiTs>`8mJyoI>$
zum+?jt&C!F&%WZ+r>S+_9oy_w*|MQ0&@O;Q$~o4$XNJ_0aKdSu#~ub8u3NR|n)#Fe
zm^9zYVtpdg>Hu--%epJEedw?3bbZGbQruvVp`5A!uE0jceWsEn1n5+*Nk=EGri5-O
zh4wK|w8|xgIhMR6N9?rEoUJu%eQjzvT!#4UO-<3==%%R1MP%plV0+0cou}NCJxI2L
z?T74q89|z~ggfUPSbSVqFP<m;xH*9ALN0JD!%?Rcqfbp;#K_pi=6%wD!-}G>04B8H
z96`{wOLN{{uRRp+bHJ%dJj(!<*yMJ(zK{eAk&6p1xx(~G3H@CBVOPp)6&HsJg@Yhc
zWnq@8HOU->Umd?utnd%kxh%p@!{RwLxoG70g+KhEickv{FLMexY^f__qg}A+@}~5+
zjWuLFnLekcv{qi^P<`0ddbW~cAkz#xpu$LGPP#L2gjgYh?Lif1Gx?Ksx37Ls?(d?{
zfC};*5^D~Jam1dlZ$|Hg_Cro4fW+0j<zf5bYi2$TpJ7~bY7*}^<)nY6mix(B-tKqG
z`6(=SLPvOQ`;pRPX@VriN76SYQeDB*%c+kOVsxrCIB7pxmnWz+o<%{au;n;4IdTp>
zbnan0)K=h}W|0oqkCpm7j5SOhjyoljA~>CN>UEUd-<q-?uifr#4-C+4%7MtKqY#oQ
zf4P#D{zR#fi<8h)mvd(-idrMDVx{dk-H^neobwwgU`dtC(X^W`P*d0(+0+R3_^Ar$
zZYO9Cp$k9~+VF%(LX9xRQT5YJO*pd99|=Qnem+~HgOP-)!70F8+0T^a0HYC_!lK~7
z5@R<_4)jjh&*rQwS+K}?tlJqrCW$<MG5q}8?0XQT^qwm|3sW8qIT>8GpD*o-A)n>l
z<-oMOGdtFCJ-q!weG~H`7VyOLk2x^4AbQI=Fn_UG#tgTRM|3W>;w<+%_1S*{fKbK#
zQiByz9IK(ECeiL7lTLkNtc<^ejs0@zfRr+FgOV@}EHF#dz8E^F$*au8p#4ho8=NUw
zP;`(EtegNHn4SUq)$)#XrAogyY1<Wdd}!vpH}#yVLas|<Ya^Qw`wQRNUn}HZL@M<o
zGGC>S1UFwYG9>_~RY4D>RZYLUJcob1{vJ-C2mtIq9UV!?ayDp{i_uSTk^M$p3{+Q=
z65$lR!o-)048djDZ<co2yGpO5yyX=pE<tvE+;{A^k{4LKWa%>JS-Wk?()sq=`SS=2
z#R%L{C=xK8yc^r{2>zWiDtQk{QFew|4tBw*$;DPmp?>?_{1xeIQT<TJI>kC}wco20
zZM?mW%WSaPIG{3HI^tMe{)*V{wBN7$ZWJvd4j2wV%ZdN*fvZnX@3TLs%iGu4-WlKO
zlyl<VjGo;7up$KKZSWf8#091SZ@hh@{ZV$4q*B}tNTGHDcCjumX@=~N^B0}|Ugt3x
zkp)Y+vcCC6`WiJ#ey1kLE7v<{!H|HD*q;^{dMCw<%K+xUk|S@8kUvj~i$ANau++#F
zNP;>CR*sTz(!``7->b~XcNQM&&+8nl>vomQhH}mUIbOHeUzA=-U{b(OE*ezFsVVPm
za+xvhl90l@W`EiI^0?>>E~0Y(&@ZXUeiJn|j@aF`zp9X{AEDJpU~+-PeD2}UQz+Rf
z=}6Jl9ksu%s{$<+6Mkgok^@VM+nqvN3%&T8=I10VEjW=Mr=}c}QgkbS^%48qG8C`|
z22wJlVnkD$avq-p>Z3EV!C-cWJu5Ci_ILHew_q^rZ%sLXm=$Ef6u2Z{ibArrf!N>Y
z=dlY7O+19-F%hZ`OgOcE)q4AfeC$D=%VQ$TotxI#KQ_IFD#c`8^a_*hAx>PhTmMui
zl6)v!H-$eDkVMSs(=4DSQl7dTANJ3=mAaF#@mPhx9|u$>=L83?R`ay|OEZ<I$bl(K
zq>QQuoSK~Vk{ikXHM@NEy)(N;ycA}3tXjRs{w-^WjBoWig-Ak|XKG1Ch00$0CbQ7W
zbqrx9X01tF&ajmh^xu>G%<Q=CM(4ToHw8`QfcuXElZje;fY5aZmbe<)`#i-(R;h=1
zIcEP^ezR{zw|b?c2C(=|I1r1VFv-$Jl%Cmt)pv%N1=*qh`<`zilb!mQ_#8*r-T4}x
zo%Y{#-bQf_CREh{Xc#j)?&xeKo!$W!&+Mt-({7t`o+R(VEVu2mo+{L^g{HOb6cb0*
zAW(nyQ#lzWl`n?=D_&txeKwF5?}UBMQ)ymE#X>4A(!ERiOeK&++H_jQ@Kf^Cl<Cqw
zu7zWINd04<TfQcd15G~Dc*udpkkbuBG+5e1Ce=Q#d{a75vFHaOLa`rtmeduw2Ax*<
z`yhNGIW?o(=a;WBfov$*Al~B@CVhZ1yB(a~ZC_B{Ab%=JwMzLC?4DB-s3M{$1SavE
zePL-G1dE<F$>qJ)2nGYA3_(HxcGwp+<K7h>QrtM{z>=WJW2T?rC;Q^3;-JB!4^LG#
z7UkS3Yla9lMtAZh<<nu{=hJi*{FWV=nwPQM3`qC%h;2|p!ekdXPw2XYOWB#EF=Juu
zOG}sXkWMmJVT5Ti;StnpUzY8lJ@_(ows*-2=fFgJ7A;$9UtW=EEly1oVkxxb8NXb;
z$-iRGqzR><5%dWqiBl6(tc9<!tY2AR(C**_`Rf&?{wm{!vhIG>Q*r7L(x7W7DaD&m
zjLU^zjJ<tz_OwGlCXZ2FEX=IvbM}JfL(u2sR2*0##<FEcC$)uTSzsqTXJ6CQ&zkfl
z7L==uocdffC2k(;PTAL%ALzr0+7YcKc5-?*C`o-a;)l|6Hl<!#2Q#$g>zdCtw3N;@
zZGE<W);SkjZ7BDdKl}Pz0g_QFM?!ma+Yn9#aV;k`!QXrwbR}82eZx~FO#2S`9wLe;
zA-DrgwRd-Nd~AHEi{|zlOP3I1kfd~AX-Mp;pc!YJdNBn(t2f=XdXs%qV#;`%+u@u-
zU`nbCG!EI;l+68`n`IK$5V{Em7OlmE%X1>K^WlYY*1qMbBJNL)b|=TgRIzUWmR*~g
z+C8BAw>BS?$y@H*icZ#rbJP^Hff{K^@T~r}>?>pzxr>5^!46orX1RTPMffe~Whj)6
zN#B<ehB)g3_8s{nQNQQiDaALc_J=T3dtsdqH1<v*8dL^?^UJQQ9YYTx?`7Pkyuz3(
zsb1L&2kk{k-chKDjd6<VF3}6`;S#><>#}X?NZT&;FQBCVH;{Dh#q}d#u=0MNpV{ZY
zVpLEtsBBO$Pz_z5y`;3>r^|~AjRcl+0HssT*Ig3I#Io5->xY)F-|Q5U+>H$*ZAmP;
zp>{3JRO&k*=^1WaVK?R|n197pQ4)OHq`j>4AT^XCapxUaF3`&5>(|@M>sQyTUE>sD
zUbR7Gb7BkIE1vqy`(r(4cKqM}b9$3pMrhnZre?C`e=OX;Vy{ecGSuxK3YP{4kSs=Q
zng}>yjV?leME0Mxo9aYwCM}(O4IM}j7$J>Pv5?!H9Ux~ZNui1YAOgo;m89~1cr6~w
zmPk|JL3CsD>Ld@g$y7oqTEsC)e{qeo`4tf$mO5^AkjFAPHAODRFJjDYuJ{!VQ+3Oc
zfTF9`h%u69&`#v=#H9uK!I>Rs%iAOPbb~FxbPTm8Mr`3V{n7by#w!eF1;h*$lPxOk
zg27<IbQ@xYiCs}HWlSHoxS_UHi4;l$rZ0WJsZYFufskwAA-4RIGOV2D99C*Es3}-j
z)m-OoX=$CO!kjx3otiTHCI2Od&a!0ZhT4-4I?u7;#4zcURZ<1|4oP`+`DLgsn0tln
z0854F@KCE)*OB3!Yp==0Y>jSJcmbs(2pd`*du=U@P4+v5_>`!ke99Vz!d};OQZYQ?
zOz*QWX}8gZFp;k>?W)0&#5N8r-U_V{i4ScYjLY+n+8gpHLLERFAAB7t$5bfId1QB>
z%^S<FwOq>ShR=QCh*Og;FnLU9vu6$8I4oVL?i;b?Po?<{)C_th!}DC5-W)E_X`5#`
zBBv6r*ovH;p6%`Sgi}b$7y8J#;_=G5SvE_|r<AjZ@SwX}<Yy|*^N_7-W<q_{395M$
zTSQILw2k;m+v=PLg3!1TavX}-?)L&?geb}w8y?xVrhX|jJ;{N_?sjT&Bp3m^B*U?_
zrKwzJ#ELh8mrhNh!<)%qd`l(Ly(FsKGNj`bP(hdF_u9Jhn&itP(HSEkH@H)ih$lCz
zaZ>t<=HFT=0V!Bu*(g@Z#45~<0TSWJEF>50wt@v(vDDIBbYNmy;6ysaHcr|4N=A_k
z6gw3D>5x&$z4*U6_34&}k-T>Zhi!g_&T&JI<?eN>Hs9hD%EpW9w=s$zvW*Rl8|R>u
zx_1s(vO+iywB<H6Znrm|KLAoe`&h_1S=F8TKmoaECjVPd)9xrA*iS+o8B+;;bzmZF
zo$YOVQ=RPl9=YC|!s_vsZAuJb)`*92O1V=|gz7p)@<m%uT>--VneL1&0Vu(6?a=tJ
zZLUZ$L4Gg=i~`6RV86@qs~c$UY)GMm*%Uf750b@BPw#i?%WqIcgWZSNqj%*1mMxm^
zJeEz8A&;4tM(n}e4eIE;5Jk?UeaF4lXdab+7d|*QvfWdDlbI3s;IFU2o<jFSjj(bg
zjPLi>S?g?>a0=1EI03ZraP59`gLt!ka{oJkj#g!83RMqY)!tHiASG}~{#H&oY7*fG
zxm9bT!}UB++WTrxR&2P#DI_7_WS1vp8iXYvzO_v9+@wN9#S5!?#;Mon@HIm5^|sP2
zX<KDJK*iy}isUVSCcRp2+VZcB6|=WDO#W_|{IgCuk7QlRKtk5a^CPg3P^R1IV|O^v
z-09FeYc&JU%Qr77d6-TFr>00uvy-VF$hm=yV7;}!OueKiREiEPU%i0vslH}~PNUA2
za^aX)n4~!*SByr-t-o$OWDwL~iZcOJrc}#OQq@EdTD#5&`#ALFh!`1pC1&ub*Ltzs
z@_JI@68RIuaZrE4wlqsZ7#MFxR-j&KlGUZ1PJPbcdZo&@fzn*Zp1gSer)G!xBtTn7
zJKKlb<berH9n7~(j*^E9KOu@$%Q-@n42wOKTWRUd3!F#InXWqO+WiS9pSAOt_68hv
zM#j3fwf1a-^g>QCr-(j<`x!@}__otFTo+-{;(1OXPY4{BCx~ybkviL$?D0>Oxe!pi
z3XLF?m}{MRRhoxKpDG?^k)S5+<g7j?lji)I>?tg;=y>5Nkin$FBi*L+j_d(CQK6-E
za*Rus%y%9|wtCpmHdb)?Bd9b!g1(A!ZoEmRjHJ$!T`sVCpg~vdG9jv{3mlc#n(#NK
zGm7XF$9UMr^CRTV8F8Lts6t^h(J02GkLE4bt+UqC-tH7?Gq~!+FP&W%ZK5DG>=|0H
z@N05l$-9RQG}U{}PC@vctjn+TkG;BQ5Qx^iKd!0WR$d8t2W4?3$1*(&doCoL6YP`i
zrBh)Qi?tSuM9`d5Q-p_|f$otH2{IvrD?ZBhU`3@q^2u`6Q!eepp@gGU2WrDu*;LM0
zAFdA0b8hKd*7ev!wXkU35~olo@H6ZU&Ok)Ths&GOJ2`F$XOa}97>txtUnX`oU|fjf
z_Rhk*ZWl-4l!O>FJKnUy9x30Em@VA_25<+K+jL}VVz6uPs(e*1c`uz(Vq7okHIy~>
z?(Fsu8rBh~5M4+_3562aE)mk}1cwA8dQyCteYPfOo%jVhf}e-Ir~Iadf#WxshniEP
zO)vL764Qw-c{Dp+E}nCWu?jl7l+#LvYaRO$u7eHsvGQthamv;t9iWtuXPlZO4P=1e
zCQf^A?x;;`S2&N7se8BD`?66BQ;zJFPB}i82WHT4K8O1_{lydB-rsbmF+VZJz2XQs
z#GiHQ%e*NAWrjG1n$o_r{8HO3%>rlGdO9gW)#cO&qQE*qIYOUe-&N;LI)!Mh?X1*@
z)FiR&CAlL`23`9=F5iPw1L$oiBnyo&vE4pc*d_erruR9e7-GmxV4M=e<`AX8I$_^k
zXB0+KY~K}#KO_q3LCrOjLz2M2Tjj7<=G70CUy30|SA`5#^JY+^!OiUGZ4*?}9+cDD
zhigB=IGwI!nh>0?n4$wteR3cmk~Cl1qaP{10U<*}mKT{QHMOYvvrc_8{Yt=iidM2C
z9kh>@nw)ZBQ@j(!JvG|9T^RhXeXKM~o}QIsQ|d2iR?8%cr#@gGPrT9n^sQl3gp{MR
zDSpaSa?HMGwlhJVq{-J>AM@FoV{t4=woF<7d&_&|6C}l#wB@rfDZRr8WRKT(LvO{4
zWa=>-pq!fMiHM$}-sOPs#Ozdt@>ve<z!HPdOZ0Mj4BM0Sm<DNjLtw_vMTXVfH|yKm
z?WsAF7#YICpG)^(Q<DU6^ZFipx_*w$FA~5iIk0T7YTaG-iP>dj<Z%Ecb>W%w!)rYU
z6mtf9jK(q-PLNCXePu27vb5)&Qk;*g-*l&ae?BYz$axNQO@mab7egZCTU<b4KagFz
z4a4}TmqPE5GCX}UYgpO#fKv#ZN_DtSLl%Y!zhIwg7D3(^NhG7oKBp#jLU^HRLG07z
zHzJ&HBrJ21fg?U!6Va#Wc#D&*?a28gbf<}iabQW^NGTxc@|h%?eTA-4)O5?5GGROG
z+%LR&nN!F^p{0BywyQD_L!<p%K8sX`b2Grep^RYFmGTE|_chays}UMcrzTn}noT~n
zawpqUPZz}lXoJiGhDf16y7#@6&TmJEZF?zhR5&rXxG!rnZQP5TV%c6|tHk`{q*`E_
z!{-8Va8hkLsUrw(TKiw3njNU~&MGMIOgw7ZEvYGkmBYVkF4@6?IvS>&?37+4!|p_z
z4m$NI2KONBY#r<jkJzDtGourOEvFFgO#6SvsU1!jQDGmP`9S{j0s{|X8$(*o*^#GG
zNfQEwdLM%Stol+IR+5T1ZAItlAy>%T(YmFVhgGQ6bs&YU<X-lE=-aW{akME!ttbu(
zP*EL8GSk)w3%tA|i(r5?-j3&y`TDmwPr2No`bp*0i8_aAmy-&@D@-he1Pc<;*~w;o
z(wd<Qx1sKYl+>EKn$lY)Nqjcx&k&bT@)%n&?i#$R#DW*&qK3<MDkp+iw`KDdJ0)e?
z#hHgZmIH%?%<C9@u68`4OF|9m6?Oo|Ox|<c<87x)J5f-X+vKSVEDni@e$oM-sk3_9
z<T&#<=m4a&co5m-Y~5aBV@I80lD)Mfy>_m=SYckpH^eI}eN_1*68ibt(NJ@TaoB++
zxyBb9MJ5u$h5Dr%ZhY$v_c_Hp)*&pjiAx)QLc7>}{5Z1)nqz+kmeK=CA!HnK!E+_i
zrShW60%cgMB7RdnM}1lUG=0IB+2yibb!*1yo19{-lc;&oWP#!sYwb#%k40SCK_Rj8
zuJ+(iA1vAr7Su0+6!H+`qm7|24s>OIkS2<@NyqvXX=*=IdL<*hGCk<v1Xi$rd~$Rv
z7NO}Rnk72xIj0<R#u0@`kjc_@+KYuG#L})d=z5lODk5_ON3J+ETJiNmTWls#uQeVL
zCnZ%7A#2aqu`Zmy#3|&bLXn)iS|Ydd=p3BcWj|c{pXPS;|AGUH--!2TH$*}a#eB%Z
zAam?z%FfNRBz&Ae*V)^u#WNjuTIKmg4j-8(iJRfCG%r6==Z6EIdj}>QfI9!`8((QZ
z+ALuTvSLvp*43#AdXe-=r#lQt_;lo`a@YP?>3c4pbblNT(MnE1JDvK#B4`Y~iu?TR
z#~aL2sF`T*VF#uSG||y5)IU)=)u3h!h?sAuCZ;O95FuRY1{#EiPoJ}&tmuKYC=ylS
z=mjh$CSA2+N3iavS@)l6-X_mG#EHf@9RqRd%P)6<-wF|yd?3;8rwc)5Bf!GelD=@%
zex~VtI4?*b7owuUtTh_4nCS-N?PnVlh-CmRB!o71v3RY|-iu(9Xm&OP%zmztjWI-=
z@csj$BZKTlL>k)9H_ITg8s-K|6tIGJvxgsZNN+g;r;ph$B$ljMv&MN0Tu64gxC%rk
z@;)a`-G2MU@<Nf<)DOwRmwp;#PIBtSz+L~ghwrmr%JuGcCds5hA&xi1V&?CcoB5M&
zkL?i5C9Hr`qZhpHy6awVzfu{a2MAIl=NaX)@d)w)sYK55SWm{_SIZ29uRSBgC(?nH
z^%dC*o|L~Z6ZUJRN4tpCde$ig60`P72!cuaoc(&zrOg<eB~VgKx;;4A?M3YA+Hd6a
zJ@U@CIgc^E!^-qyzgc@8UY%1;IeQVdlj=Ook!inG#)Rh*X1VkpQYe&yTqQbDpLAU(
z;BZH&f(Pukn~9iB1e6<TFTCa%uk}d=HAN_Og4>0@KrMc!qDt{mmS-Zc=whzFNW8Mq
zklNFJx3pH$Zv~FY9#>(=oJfvRf*rEoD{n~dS#p_#B$<JziG?eDM5dJ9=Ti8nNr(RX
zIe9&Oaq&GRiS(~}vHd}LqkbokADQo>e@;!4pP<`^RzQbd*Z#0s#mk}UCMmH4)5<72
z6I|0DRZ2eI+U69J_ZuVnfP)im^^fbyt?RaW(MnP-D;JrsAj1*BgL>*u^1Y<593-^}
zr5Iv*WXXYC3~0vQ%FXUiD}nAE?%<I>?hRl|s~00u*Z!=23A&cf={3_CH9|aGL$E;M
zv(j&WUa?|1cZ!Yc$meEvfN_wm+hBjuyep?nv!Mv7OHPeuV`j$`8RqOSORqr@nDKPG
zN*}toV8^L1q+R2d%gQnPs{-RUoU|I_J_nY^v*b--=wFu(gpVQHDQ{>Jp4A%GGy*&1
z=AD-JzbP+Hq8aC+H?S1nbqd<wHl4Zz?uF*ffyFYxQOD$Epbm-D?C(mAPF7)^SgJT*
zotoUwQc<FM8MD7{hITQ8d6wLP6{3^ISDm~mhVIuHX#Y^B8U?X@)8v*Qu5h<gBT}~h
z_Kj=pA4{v)4-g>rGR|11Cf=XM+laj2_tNvGG4!8Gz4pMYt9#zki8q|r`q`L&v>|ai
zbC6&@`{y#I=q=-iFLhvv)S>3i>rL~l|CdsubyI-XewiV8+b*X@QtH)@Ayde|>R&4&
z-PWG!I;FCcT4XJNtL?LYs|_S1Ii-RjT>i`TM)vQss0T66+#KqiLYH+*BK3)SVZ@7@
zQ|v#Q^@$XD%;G)=mh-1xYUab&T9Q`i!|XqES1ek3v-4Q6SqcBs81r8Rwi9x74he+6
z4y@R7BbQK+oqEV=VT3+=?7wq)u~tku&nX)$ShUdW=_1d!xw?c5Ta@+LNJ7&d7{_0|
z-#+W<_`zg<zg$_Lo=bW~!i7R|`2FnsDGQ<4523)b4y+)IguKYC!cZGl(LU$tK)X(t
zl<qv|ZoeCm(LT3%Tk=#SNpWC8((%bHTkP|$d2@=7CBj3_?@oPL5>S^fa|xS$ehy@2
z$319XC`1FX_3|eniMf42fysc7DS8e&u)IP=ce#+=pIc6_5q;s)u^-{RPs>gKjRYc!
zwq<hMzG${@!M$kg)no4<TB};GPb)<e+yV7z6H`K#FK)Uaoh=M6>@=`4lgeI@B64zE
zddsu+B{?>PYfN+$qH!iBN)19-VnT>HW?%Yr5n{+XIMjBEb(J_s(~~%TD?*`fUzSL6
z`xb(XWu&B>=R6d2qBYRLQ+C3>ym^V(NU=$DJ@$6e4(iK3&Sa8uFA;4yYF|-?Tirki
z`|%3PX^A)3S3X_1KHjZ+)+NPfXXz#kQ=Yc8DhnSDEc9THeN_&CJhd{r4>;v$!)3;D
zKxDPg*jJbSiMf#>mpuN(-Hw_>%vS&X9ecskg?e=RuxD`NPuBr8Lz8+}2>^mGvaiVx
z&5$SnX;IgJwjG!d<)x2YZ(n=OW|EZ$s}LIHlTJ-;(#`D&`?|VZ$=gxPv!FZ4di28x
ze0`aF`L7ApGRQpBwn;fled%CaZH;+WURM%~zoFEGqYf}hX%TVh)WjTVjg%Nb&WPVw
z7aj#=q!8(lCk|jvczB~K0{f(WQ{C5$w>(SHXWu;AHX{DXx$X^&<1X3s^Y$&JU2?}E
z9XYVTQmMtNDou=A-tT?czBRwObkV$d&U5L!bTw#t;eo!b{9Y_(v6E|8#)NVBIQ8)x
z<+hUN6PZ)&+ne{CEu!I02UZUxQD$ic67}pm>SVH)^O#62#Brx4-YqRUIYVX+VpMx!
z_IGB-n>)_4Pz=4}9)$AiYEKwx9BCi~u-v25Cpj+4(akX)7gIvvUzAw2VXDu0yk0L8
z36{xc)~Q--P^RJ31e}AVzrGoaCIYKV@HygxsS&Y#<Q4bbXD@DADGP{pUdBCNa%!4I
zkxl`74Fq1hDuUeYCHWTktgiDc#McEWFMT@L=7&a{r(#KmmhS4<4Rxkq-L;oUdFa3j
zeUkc4YKdygPDbFy{K))e^A|eLu{xkFssn8Z8Cs#B_Off%KNt?`&%n}!*OpVS0Y7lZ
z%1!q2Tn}Yrg`AcW*?*<Fensu^hSi&IxqX#WE_jnx8-)+Ovcl-UD7X}p6(-UmJXZ>^
zn@VrfJ}_4dp(e%*KF~u>&Ilu&Wfflabm39hig@gH9YAzL^aC2eO1RBS*vt;%@;eVr
z4Pd8tVCBUM?toQ@6+bMBDl4<iPwKJ?38ROcYa~(wQU)mbNn+_?ySdUt66Fs&B~2ZZ
zdDa$G>a<0{Ze;v&AW6%%l1X`C8F3yE5O#P4nbmL}EuYC)4(-w-wy2JB3)d>iIu0QE
z5*F=Zlv-TT19h6XPRVMyuzJZwul0c}d`zLH^-P3J+miZ`L3nFrz620BBvxPU8)7fD
z;||->#M>9&=R66Sdq_91tmzZ;M{^?K99WKd1^UHTmj)6F<@tCXJtDB8{yr2qodJ2x
zrN|5eZH>LA9yYqUJxzXqG}o<(?xejoztY#y_M#9|gC`x*xn}k`dtIIFnCE~c#d2!I
z1j8+mGOs(-OE+JiZ=Tt)@^0rji5!E^RF*n%Nz-R<xMmv#{|in?WfKYur#?s-q#gS5
z#xhK>G9+jraIV7~J)YA_T_-A;LjrULPi=XFr;7hNz?mf`=_oi0(~13*`aA_OvcttA
z$E~PuAyFuYEdn}8OlzzD&SdYz7J|H2HX}PuUt|PS+zwvXPEFA&DfFmeqS<(o?~unp
zfqk|rXQ_))#d$8XRPGLDvy1TTx7Bqq*o_KtkOLO5#EDIJZnQO}fyl6OCL`)Ou!46<
zjAath)E>6ANpsS&K=1<PY{^+A21bd-6k&1MZYl2-zRJZ<a&(2U{t?{6H*H<*ATj;Y
z%ga6tOuFqsqNCXPc5C^F+}(8Rr;MU;YB)71LAA!l#tCtQ1-Y%fkKG$fEbS<;GCh(A
z#Ab;TS30Qm*VJ&<YM%*3yw?0b-rh4n&+54Qp7fsBiPNhcY-8L22_)<|iI4=6DUu}t
zc8trtySH{1ZSxj^L~zF7G$(OMvuTM?mk^Q=pk9(?3F*D}-t(mQ_Wu4ebFN*I=ga%y
zeZj1+Gw0mbb;_JM(@%w3p$E``LsW#_(xz|Ob<3QBcEMHGaWna?iE)t{W5y|IV@~u&
zZA0}5i6qrKXc>W3pQKub7Sl5g?YM1hWQ5-6l+n#A3R}g8O=@qaUQ*%pAY1H2drN+4
z6;ct8@zz3ZBHiQ}cM9#{rbZ}9;WX})md;4zI@IbNw9Q>#6#$@Bf7}}^nPAxlPT7`*
z&~Kg`nTS;NKrW+SjOA?Vd(ssJ4KehFZHb2zvm1zRw>4o#d5;xM$~M{IK(+3&UQZ+2
z++Lle6QQ5zvNk(4iNQCmS!;K+k*$XOIHk0U;C4)4|69A=`y~5b!;zUwNgz_oMt$is
z@aNT4q4u^8fr5U+!#?4_vei2BSwzaBZ>YWrz2(=m8(~ODe%7f8@|T)79r=6g?d_=E
zsFb=poN}~5Clc01F3Pbx8!V7eeC}ZfmV<#ja%R1w6=l1eaf;ci-ryjY{Xu!&*}jd^
zNu0G@?t#@(D_l|b57~stgk$fjxk<&_0S<UMLna%^vr{h&UUBU;_U`uOSq$nsoMQGr
z*w#2atyeIaB`@Ut@!nu54@9CD8|c=KMBo&nbIp6aqc=9#1~)i41`*)M1)@b?f2-~~
z8fw+35=cHgg*#}s(#p80$Xi{38^L?h$Xmba$$BlL1)E&V&k11`__g1uk1>hDOG9=V
zBDOVQ1lsCW<e7oF>0t*JT^26stWn^okouerw{z=`>z(JCAz9)>=(drD1#!pnuv3mU
z^eE&Y<zgJoHfW2EKl>3dfO59s%+p1cm;~*vHg;^G;4IFWK#~rwAps%zJO_8!wr<%K
zOfE-i3Byk<oOWtNvaVNV;j!F1tCgsB%qbOrtMfmMQb8?SDCc-vy5(11=M)k*2PfeH
zY@$s+2qp&+1;>b9==n4%OzO3Ile2y9-JR@CI5~Eu`XLEzoT;pc?3c8P?j*Ya@k6%G
z^E+&4Hr3{s!v`}Jt9;gh<(}ZKvTbtgv-e~xg%1sCkIR`An0Nzb55du<vunhia8OeS
zDi*RDXu@Sp9l=1LWvO-3s}}!=&E!npci(-lah?M$UJ#%w7R+X=CPdfxrKq}6Ewq|a
z(wBgX&|`D0)3`h9Do}Ji4our|_3ACQy~DwD5K7l%IU-S`#T_O4qut$5pp<dx*+7B=
zr$*z)qh@-?nB7zJxNm-sjz5HiOAf5md7`}{`4DhcK)!NH>mIblcC`0wTy^7Grxds<
zdX2)5)<<n;!KB8K0Cb#pH<-R8m~zS7+olf@8PtP9%ifAVwNn#s)#0mXW<{iZW8xm_
zXVNIlIK^mk2y=}@39eAI``X~c{X&eBG+a`qcQr9B5N!kxNC57=?`_*?78jivt|A(8
zK(*WDOT%HDa_jsN`=+)TmaVwfDQMaWxgRhuv|XxITgqfopS{K2kugwD*!%L!SS2$>
z8i_jr1@?n*5;H@YnKTC`347H2O$U;fIE|4=ZpZ;9>-b!kSb!-LXP<x5WBY-&b%*9e
z%_-&#Bdg(EB)P=i-&U4Av(+i(d3!5I>NnR;#*b;MQ_#RM1x?zw{9iw=3(|AfY9#kN
zt&-&HwB&2@E*8cdwr@=Y->_w^^B9d9iyo;gL7_|bZOQ%MsXl{xmTdtb<w%$e-Cdyk
zpOOGt2+_CaTi2}J;ylJ%8+GcW9Q#1MMBNdmf}9Kza#+Tz&O7xnMGPT$7|Dd+(fUZn
z@=JX9F~t?@_ugtG$cm{Cep9I`>^s}EuD|t$4NfU_4jX$Dz44xXSM{Q7yBWF0!ZS_8
z_Uw1+Q{TC|Pk44@TI|~gb2&x_r)J1!nuA@}zsEk*x);u~&`@C!QovcKCU^g8oVV=5
zU5OxJr&PG0>@Pu>>?7T@MNTIrXs3p&z@3_;w&21**s#94+6q%qn~2HIHFj!ho%H5p
zYlRnVSbBBBzNZadya|)0%L7xrB);m9Q{O0(^aaR+(*0D1%{c3275(1USJ-AT`SOj^
z`SGN;8of=KsR)zeExLg1`~D|08kTFI!ffu2AGYtWzF`~AFuh>OGrua%2u^*@8oh=_
z!FKPlk5*6Q10dcof+azZC!Cs+T81L29fqBcb@&zJ7EXDO1B(HR<VE&CI-CR`oeRPu
zZVzTR7j|tLb)GfF;bOPze+M|*Lp2PY%VfD>KS@kX-M%zK)YmX1r$|~kY!BCOv=mrp
zSyly8%b`>KI^opUZ<Iql=&U`Gu!!Z<MO6wN1eUmt!FY_Yxku~0iknz80p`_-&mbRB
zFDXY(nMy=3Yd_FdiLF1~C;hHGmK|tJlH8VrX^{X+xJmcL2MW*5+Ycs=N0*N}rNVJ3
z-9p1|yK>`94v#pGK_xW4x<^1SLzsT7{g6J(_944h2ZQ`1Dxb35ja*Jl4mgD(I=ZMz
z6PUw&()P5P#7GE8+AE%PYJ__n2it9L_H5;fWzJ)=T&reg$XB;7YhL%RH#(1rQX2*!
zB!d`VY=7IEW0L~|J7jGDlJ@_%khe*E5V9!^59H*mTf1t_+D%SLT*j8w8|`2XSJXMU
zWcX2FH5!SB>6)Bk@u5}&iIm5D*h1Ogdz_k@Z@KSC;keHZH>$&Jm(xxuTF5XiH5^&(
z?eT)82Y%Qx9azwiP+I139w-;rKq9;zY45>XLdeM4JFxss2Hq$;7GH?8NTjAbVo!8*
zMYJM`buzA*GbNb@S{3zzs55Sb(1)HWdtpb5ZPWok%GofeB}dyt@kI|p+NN2Tg&OLQ
z@33PHz-5)&ly?k3ZWf#z(EZx+9M{6GZKFdI=sL8Tz-lE<h=GCfC@n0~n?0G>gC6^m
z^Bk?zV>w7j0TQ&=PSjlO6O#&8huzT|EHk5OF3#Fh4e16X_&UY<tWLF5Nby1%KHVmt
z8%ds)N|ap;ELyio5CSM}`$>DICQKbyS|vqB%MV1ZQMyE1BBKUGegda$PRGrYZ2P{Q
zY+qtAC8uy;c)=ESp$55!juPW*r(W=NkVd+QU~j1jgiCs=<vwYrJD3iYu!cG?ZA21)
zOxl^8h%wSU&tZI|s8f`%9WwUV*}@NL0+b94CZ11qo<>f6(R>`c2teM618YM}r6fyb
zq7YLD2;AEtr(UCm%JD(=>G`%0Q@t_n;G>vl@9GF^MssnYsN4hdhoOv=LVCYyg<WiG
zCiNdAsIG~aKkU@RJXG`~+jYNP$}uBZ_MGQfzb4h>_QZX}eyGv7<;#~l1z|MQB=6p^
zA8ucSMv5cn#K?`})WjEwL`A^e!+(2izFNnL!kR1nOKfp(H4+G={uOHwlWWiBU=e{)
zl{=-_o{eXdyZBO@%R)O#IrEg~DGZz;qb6piGKf~SF?WuRvrhJ0c6MPGJT}qykpT&q
zJgwgQPWzD>O71H8u<v$Y(Gek*WGN~4(nmBtWY2fMQT|;~x?>KkXiInfa<{Rt8`zK5
zZ{(1zI19Y-9hka$!z8z!&MFN3$J+4-a={jdoIB${6G;Y9*($2kemoJeUpKOgk^s_@
zEbN*kZAkvoexkPL(ilA*`%0tv4Ck2m8S&Z5wn5wIza+jF<P)H;#I)N_c4Mv#RY;%>
zmUKyp4DPp|YVGnn%-|&lmhX<sn;QZ&oy$r4=|Y{P$lBauT;LtO)vyvsHR3dt-Az2=
z&$LMi-H{9}Kyr7MRDu5FXB!9-JuxYRn*$2em2q@mm`D%X&m|h-GR*&+lH{b>Ny1p|
z=W`!%N$HUDoSZu!n46iBU0}$5p^cYQPsTa2TlpeP<~EVnXY3cN9p{bEItW;U?O#Zl
zjq}WYspd)>Trmua+DHN&8anUlG5kn5qg0+5$jdvK|6i`T%q_9RqrqE^ycy*OFPTk$
zpRiv^qQqHqKzAZhj46^PU9m7^8nj<cLPSzs=P@OZb-J$c*IH>>&aRZ|r*Ye_(ni~_
zcQd5)7-_u#CX!5uN8f%UzdpyQsk4b<P^)Ak8Zt<|@SEL>g0qF1>~mnb+s7j}O3!|)
z8p(%VKH`#Afz<&f{e{S+LL$zQaO1bT{_4aJlHv`PdPi91`Q7$AU8fdBOEj(+`l*Q#
z%U~imYp%Ap-z^w-ra?GGMRq%|JpR&|(SEPm=<;mwW&Cwun%QS}L2CEf@3(Kl2`ddN
z$xm1e-Ty*O4ti#C2e%?m*dOHTth{!)^O!5ZR~0Z%|LqSOPC>}Epo$uf<DT#CnSb0{
zeeTTIc{;vy%tTD&A9e2x)?UVD2UcWuiY!H3s{Y5--DH@Uc)1Je2z6>SbaFvqOXJ?|
zKWQj4(HrV{m6*DV3#aT)>n#xCP||cN0H-F1h$J3Sf1$6YLy9JDxU@gZPD)-Wt-_20
zi{C`lZ}uimPPya|3T=O01KSf`Qz8k69as@UvJ7aNg4L8r5|(}Yi;m}cQcgXo_64Ot
zZ4L@$Qj|~{IS8D#u^8!8oVTL=Wx-0VO+OAeu)=xWXCZ`<+FNAqui9J@6Lg(ZDp8$m
zE`*qoM77Nd3Lh!-QW%mdG5ru)eGrGPUMZtl8_3dMcdK?Ph%T3mLuTIxVpEf@YNK2s
zC=oZzx_^_Hww@etJSH6=&F&8S+Zq<9Nzf(ywj5Y8UN>#Fzgv{?<l@3Pds+hxnIw`E
zVYEEyw2FzyDTtjzTi8nb`!@9Jr+Y)h-wrttq1Bx)&42A5@@voDFAvl<@gRtpQ6kCh
z9~-l?ft)@RYTLpiG8f!|WB=5CM7Q>XAy0$mz)E3p4g!%_UD&jUhxX6i+jO9cVxIK|
z(;(&OAny8KYPC`|dH@cQ`;M6~`Q4DC)v1r+YYWOxL)TJcFhx=LuWh@np1gZtQhuko
z5ZmYa_HT7Ykr>E%E{u~ECJqk!_xkD*g_^Kb0s}PkkW*h0DKh=6{YMd#l~*~BiEl&)
z$}@4s{<Gl=V-!;)(iE(dQzLYk9wjx#jQv-&BTIbjhPa$viW+Y<K^tPX+|T?g+2enA
zeNQ8S;y8fCZfrcEJg}DTdV8?y;r<c)q=e#aarj7==hPQD4Cg!u@?QJ62Lo{s%;Ty`
zr(A8$>VT4Z+dlrmT#2Q}sCtPPYPD7`WXw6uG4croOW7w6YG9gGW$KrG;uEuo+t2vi
z@UT-9lR@|@0W6>NV56mYh3fK306FR<qf%xb-3Nd=-9EYMr4LsAHzgEEzB{&*Q=fR0
zM@JYV69Ye`ekpvDa`ha{ND9g_rtNX+wd=9oLPZ-3YM)x;)8^K;B`u{l7)G|R6)_w3
zY2Dl?>QG+wyB%1KCEWpKQ_5Kl46sjcJ%ZI~o_NXdL}Fm4KFQ3cTUU+RXXM%~?7~Tf
zTlJiB;_g;fbH+Y1XL58q0fx%Y;sA=ENQ_HHHNX2=)mNg<L-am~=nPCGMfaN^tj*YG
zcjK0uPZ~ZN45ui@EfeKg`<xnsJcA|m3Nw~qsZ$>qmWQs$+G%@HV!*<#VSJt_<hHHL
zyU<G7i>vFYjmZ0Y<x_dasYy<^KLI5iWMkUr)(awf0h&SxrZ*V#7>pM=q3rY8Pcuc3
z(qsVA6w2hN>;W&SHz_R>m!(Q`?Z9FdWwKjzXXR31`}~G!841Q3rpo4^Q&Z<?c&I_^
z<P_HF9PJCb`b=MPi1m@QryFf#;aOTw(hPQU<**J7!zKH|2SbE=>1&)P9Wyte2(vG0
z{Rt_#%=dA$P!m%u98eODvuF#xxL|1j7B$=Bi?CEaYu3J`-l~;TD77SE?ZhMAYC>?-
z%5#1P17%;@{U)Qel5(NCacW{_LNH*o++E9W@-9&&V*Il1m#KbNE-#JbkW-V`zIN04
z&8zIo6NONtPVaCE8sEaMJ4o~kN{|`Qz9MG^$Cl;JbD@}=1EdQ$WMA3APK1~Nuj9ay
zQ5_v0pCfnOSGE1O3Js)FinoB_(A)THFRk%Jo(9H~dml#uMn#4Zr~V3h#*DR)NX=3B
z;)KIsQUir)DIyMRv6r>65aHBEY9GPN09f|xmeFC7`Po-L7^;!W-~LLk#~bHH?XsK_
z<;|XP3i*?}M|bMRt(RvdWDOq^3ON$aj5J=IwO7>4O93R0m$3ZkuTzu8vum!r#$LH(
zfhyoIhI-bi$uCYRlJ+UPyuA>l3-4L6i4iVIjgD4aXXwnFeAHf*w5vCTd)}myRshJ=
zMx)5>wO6zv67TSYQ_LQ18XdA%FZqr%gqe&3#PU{?N3{?*-I2(nS`#3Z8YawO`hXQ`
zW|G6m%pMF%D3|QDUHxJ*IUORMTWXF<On9UXoUzxnY33-CUg7EagWh0-ABOFCKL_sX
z+md628k$4%fldQh9FjdU9@~{UJsT&d$Y?BM1b{?%0te(CIs_MbReepG0<pl4&+l_!
zddpCs#1M9MYryF(9+<?%4lJk}YW>)0a#4?S)RsM1693vDwEf7p3^12vW^8n7%9giv
zxPE%nyTY<-uC;5rMUi-~eIb_?(Y972@k+~)i*kDYuwC2w2ZJm%0WqTvERkxa7grRo
zXan!b^u%5EOa`z#5=KUKft&DYWo}$z<aGHnrR??f5r%VJ`wsJ<;FNO<2OU^8S%<IA
zjJ=@^Y}qx}IfdHyl2A0Um+XzLAJ`$t_Nmjuf$1Fs#BxaI{WUcOnofy>VMh-CNNVDR
zlF((t<9N0=wJ%;jH|P|q!3FfmC@tpq+1IuP%dfuLDI{}>9~F~;82ojMUjxf=jIX$x
zV@^%H3|$t|2_zTW#MkG`xSXfQefRBUq0js<y!xsRD|ClKr;b6mpeA0nPbLn!PKqJB
zzF<&987o=6v)*7~3Ssavh^(#+6xwQjtDH9x1jrY-XT_=S45ZEU+PoLLY)rPMt<<Le
z6nRM~N-;GwIXEW_4Yi@IZ6hLa#0a4(;|?UbLW~7u!)J>mDL801w6Sv9XA(>GO0qjO
ziF<3v>Bfn2WAz`|1SvdoVWs<X##K<Svv@W69c*1raF5*heW#GfwQBn$1Z(}`f-nOq
zhX7L(;(wr>d+p6jR#U7)>Q&|YN=+<)7;Vf8^M~gT+D+B5I!rjqm9?j0U^qId*Fc1p
zW21B9c5{}|yHAZG_Bdr}gl`<NTVC)ceM5K3A;_>?9Bod0;gU?cqMoPh);58mr3ufb
z+UnG3P`7ZED~_JFp|Sc<Fzr$Y5N}I6lmyYewy~RJ*_)c8K$b13BaS=u^&4CmILeh9
z>Md<+a7&F-h?Q2{kQ|Q1wa5^3+%`R!o>7o{HbQvJFt`f{$r+n14z+SjJ8iX=_r2@h
z%WZQvKyke~$@O8lHK(SI3?-{?d5StUnQbYIPBvDAK*xaG_j0Kc&uA4x3PUb$(e*dC
zHO^x#DJ`8+wUDAs1b)BWo`as&{j^S@(7Ov{2@JSkcXVCUB}WPz6p|ldpmD1C|E)g7
z8XXXa?X5M0hFiK+Tw7O@c0L5oz4LpVR;dv~G=VJ7*xR}}BaVp=K9S@Iewv!1=y*KC
zy?;Za-kT$$Tl5-GHn|1s02}Y^OV(T$my?Idfk`x?gbKm#tQO4}50mP^Vot(qlkIwr
zjUWE$_Kt>Q%tQEZg`Ok=ut9<SPJR4RUh*uJu2C)&dS{A=x`Z8nL)1`lRs#(t?*QD5
zyjI1uYMHfbGKB>PAJ0;e#SIVku4Lc{?ROr9X-I85?A_f!<))R{nHo%~SuTcK>PhQ$
z?eK2ai!ovVD-_U`Zo+#do|WhHK(!9v^uXV<4lFPiZ4BJ6ymPoj7X}5o7MBhUKKT5L
z=+(lmA@WmA_lI@F$?*{Q5++GsQB+tsse~qSsFAuMQmy0DnyZ9Q80u~7cU|l}p<`D+
z;jKnWrW7BQz39pL)3!CI=kiy+?)A=dYy<A<jT80p`NwUzduQa3)Yiz&<%>1NZbw%Y
zQ!vu4*W%!poX?uP;_76}m&TW0#IP~nG2s{mLMyFjHZ#K1XgBOp?qzn0ft*5!C+)7*
z?dR^VmD%mUVzp9Zs;H#kyRb`ctJb>hSAMS$cn7BK1+zCuHd8DwV~yf%ylJ&l5Hql5
z{pQuTZQi_|o4UqZcUBFcpmd5I&D62SC6_i)Fp0|wn;=TlfyF47D9AuUu2rD^7%Gm*
zs+V%EQ2{x{D@G0PET>);Te5%JRE@8NP^PO9*;t$!v83Fa&Q)#iX<x)}F14c$R*Bu>
zmk<Y?`r3)IM=z4mvo_r(6~7s*7o2$~9au7goHuX+Qa#ONmqsQBokx9`6AvrmZ2epW
zL#G6p8+;L_ls5b1R$z0j9ojR@;&Rk=V2KRD(j{S3*X1&*bSc}bUb{n1r}8IGvwHYF
zIrYiBDOGLgTJG)^0~hKbAO{>+ItB%?%`6S#iA6{{qGI>t_IS;`_ulC|$1(?z3nbld
zI|`;HWyz!&@CFkfrkg+T;<TOZx9GRHQy%BEC%nNJbr?-fFI=>HTRR-|2@%t|>cC<?
z_zhG(5h~6okcU)3_Kj6f?~yyCEyx*HG6YEk)K|ZWEvNoT_RM|t8znP|qzRDdbdB8q
z9Fg))eX>-fqC_?O-old>v!<Lzn*JkBO`au@-ia)!?O@%24Z$gi_!6x)Yu@1+Av}=8
z_whGEfxOkot|l2aI4fC??0xkczNIrFsE@vT-l?fcIT>ER;{Jud690+Fu={h+8|G%G
z=4Lm~;@Cazz|@;fYuDan53~~%v*WM?LI7gYCT_9!FMbsKIuJ5^&LlN4lZkB{M5xlf
zxy^1IOF^q0m_D?y>)y-kTiQ&Am@e(SHkyVQW;drkR|orw;-H>k*xy>P1CVQ-8;{`x
z04%mDp{oov{|2Vow{=sYV^Gs00p$>HOX5m>Zd=lhlfIio;r8v_5ar5}teReS)~V4h
zUf8vn#MK{Yi@6S0TBlTaq`L~#{GfeD!P2ujbY0$H(q&<TfS)^L-`S>nws%h`PXW}h
zb&*(R<@m1r_Us%<5xok;gJQ@1V6{$hCbt=|?CB<wjD4u>c`gi}BU^6Ify4*2|Hw@S
zKY7YN+{OcMG47NCg(UyUN^(pnrf{5mwvV(K*DhilpkKV_pi`4G%84=68%8>{@6J~b
zNCKX5N<~-mx=<PVLP)po>1yEq_#DRP+NRV9X|7ti@=E*OoG$8*J$L``(3As+zM>~?
zA}uSeh`d;D`@Z}<(L>wWc9d%DWGm+-xaHdScW+o^VVAh)IrTZFn0^H}+DF?2Md~Dx
zV7c8Wgiq6X)TxhA%Agv4lKhnvanL@NYeT>UhI<OxEELe`f~iC3h&@;}?3&QfN+byl
zs8gfGB`FvEc&IJgO%Nle6oWemosdv~C!al>Emn_APIAR3CC%je%^Nq_BT2|4i{ba>
z6muRqHFOAurI($lBN$f09&L1QMw(BjSktZxtd{btv9;_6O5|ZHLqPMwom^8bi!KtD
zf`RAf9&p-<r1pk!$YBfqVA7k{yw`csvHs3i-)XxVPA@0@ABEf^a_C?cmEiZWB{F=b
zfU6{^MI4h+pjA$ha<0V=ykNU)DCv5b+Up2nqFynO)YmG7KRLUWKsaH(?dfWy-BpG+
z`RH-IqEk~n#Jmp~*-yOY-gY3MY?J6jxwb;&4S`CwYk&8(meZ5rs89Ja=7DL<ySWs?
zPvVPkLeb@yfyaDB_ulT$IlHSj=scE(Z~B!UNl}RI>%5Sv73R+lR9A($y1$Oac3?GR
zF(j$HPjBCY-30DYjJkYi9GI9+tZY*#AP!}VxO9lXWbR#=abVFCE((x|m1RRhDUP{|
zC$_`ggq&d_B`eDOkKxOy$+L8JWLg*8KHkbVk8-v9tW%EuN;R%XJ!!XsyxEa92aqrM
z^kk5Bdf_CUk$IF>v0Ny^!~Fm+DSM*O;su$G>ekH1;aRDVF-cP(qAy0XOkGQi)JMCw
zoGVraAJ;`^O{ZSCuPj}%C><+~8nx*}q^WTrvG&?F4X?u<$pk4Ct@spu9xrq|-ll$Z
z0%A88#|bdei!~D0?a4$Al#|Y53^}(Wx-+zIC)y^vVQ!mKh&jz8T_ck&og>tqYS_Wr
zBv%rqc9faaB&kGOE~yQH;puk%5SVw7|DJXr*+@5Vp5zLZ3412#5^HwKdCs+7wR($0
zij#?5q#+O{QK}6rZ&9ReJbzAg{gL@S&Oj#Q!2C(?YL={i;DDW8GCI*_Ie!F&c4}(t
z%IiGnxt*z16uXYO6f72#f-e(`fS}>6KJN-o#=972Hswh>Tff&bwX-ermP-t~->Hv9
zjwr=qta8EAX(9KD7|_S;T<RVBq+G^j6yHPyGY1%*-jDCF!YHsy+xa$;GjpMgrW^~J
zgHUWC`bZ^Qw+n3oA<A0HR2wUdmYVcO$>1twSVrN0vAsF&$eDG@`E?Y%a9Hz_{V#RX
zNM;CacJ!<bhSU${!ms0F@k8xh1CtzSGfp`bI=ZfG8wm8nx&B*6BY71ip)-z!=riZL
zZ?M?(`wZUzrUoy&?;V$aHP=t)DJ7>m`R$Qwp1;U<5OBqlP#;qww>O#2<(N%9PD%^)
zYIZ+#PHXXm14}wLfOglO>!wp1GY+Brt+$#K(DFcu6|^5|Zw=NtY8+S!6l+HBLFM#(
zLzQCbD^+eFwp2va$6PB%g#3Dy|M5q&W2nHyjZ%sh5*m{0Jz_uBeuWJo^mxovgNd3b
zJ&+tBf4n1V!P%j1z-bY^pe9e`QkM54POIFaGgI))efx>lXLtuR2XVAVAS%DY%;eN(
zj}UDXUZSAlpG;D;4e5hNaV2AHar>!iDrKj91IZTQz=G1U3c-k`FKn0gg%3*D`|0*e
zsBb0p@di7<&e0tiWYe>s>5#p<I4n`#smYbtNKoF4{cNLLc%7pRgnI;%$4v6QA+ZEz
z+s}1~l<w=g15(P0^G?kYMyVY@atYKvJ8M7R@LfhpNrnK55nGJ*S~7C`g%@-pDF)0h
z_}OYi73J+LY8dV__KWQ!IQMh)){p~8O~4Jx=QuGmX}?sjEjv3CR*r~HjnICOqn9LM
zzuY##%^PlC>y&cmqko8P!iavQ-n5UC={$>bG7b)yU#CV3O-#~G`_&qztYGqP;VK!<
zGPF+28>zs&)h{8VfjLsoU4U-<TI<KN51^c<l!w*)Gx(KgyNbS?TOjXhwtL+bSG?oi
z`<zm>%MeQJTTBt`fc-|-ACdUny1_H@y{Ay>QXeBtSM8AJ>^JK;-8tzzYLeoL?`iw3
zZVi>2NBoFfg~ZSBGB!|Gk8=~8I$OQu1BmB=`91dA^)VUdVZd3*a}KP|);@MVGPX_F
z?^LrKuk;)H99WDnq)<MmOjsN{82Wy9v9;tiWb_l>U`?@piFUtN?R4B|BVinIV1bkp
zb-J{c;6JJ}w&*_l{q`IBAYT;{COEM8P350OzoU~0HnM*62lX5A+G5RwAq5M6(!1u+
zrf~rLVb1dG_5tTH=U@o8yMal$7yMD9MOzSEX9k^e91!wQw?A&njS@<fUwofFbit{K
z<qW?7YS6mapX3^n<)b$-;gn(`Q-LKzLV6V+u|KWfN<5ZQ412l*)A~=+iNRU>v-WnW
z0+Q491`D$rU&M}Oe_r26Ym@Z-8K}71srg@2AbGG*l-OU?7t*sml`z19PBzEe9Iccd
zL5dQo?Ju*ZED~abQ_^(Z%5~5~_E+6Y({_c1(t#xh0DU68Fd{vCo&BHO$%7+!d-0qT
z6<OC6qokabu!i>6-TTf-+*sh|N^h`?rIA+QxcyBHHg&d2v>`7RVXm0ZeNKIzH(SSa
zDTMuPD-*;q0M9_Vdhi$|FI-n^uz%Mkm~EBr5)yF~20~3Cz<zB@B(#0@_xWCV;VCJI
zQ?9*~Layu}Df#_F`!zNTNuXr7bYRgsDX5H>O~m@g#Nf?S)6S#tL|P;^@IU1=F6_E~
z$m@9$&58YUy<WqYYZnnQ4lIAU9w$8emu|{r4{e-=&O|BNE;{u|n4gpCU(p9=>|fht
z#PL;p3A%y{-fF}$WaNXKqhz}Lx9p)}wVdZ7fx4eZ`{=m+dz&K3l3HS%pv6Hw<<tnX
z5Ot>~>_58s(uR=aa@c{zeAVG5ML#u<&tJ6vZ0jqha`IwPj?EA$R{4m;#}vDuQRvYA
ztLhj1D_@Xo!Ksfg$&z&`Zw86KN9@1rmq|g>LW)g2v>x)(rCx(r*T>tz?4iboD2hh*
z7im6l4UGY5-n34%N-p9Cj@ljeaSz2CMsV(!bqcwLcZ|-?+Q&bX^QoD{sU05lm;=)X
z7k1s$8?aBP4@$nz*UN;OUXkn@XfytFPODsM_z8vab3KfGV!;sKQL{u269<;N2wwov
z)laH62qJtw{yMOjX?00~FbV|L*&r?8GcZ}7+y*zr#o;@gVo|9KG9y}?S;%x)r*Q86
zl&TMNGu_R*+?lz=Sf@Vc;}!++f9gZU`V-rr)C>+FcNAliVKYTg-Q)9VZOlv>Iv4TE
z4lJi``Bf|J)9a^8rS~*?x0;OdQFy>bCg>1S+Go^D6_tC6-kzC1=)m&Cnd;9@Pmc9*
zV))E@by_v?sgvOkiR($HCb_Cr)5GBuVxRR;a6F?km?U`)Aums6osn{5!*QN9?eg{u
ziQsNNyX#5>3u7-i@4#|=%dWf1KBpCyUwe&HsEwIdnGqmoFKRG$i*!?R^aqwI3pgY7
z`CLR85_|{!OT&d!lrOGdhQ~3pj&Gnhm@W@3wXA(^>m#^Qsc0OSq_@ZdBHtwYybeS5
z25zL;RxFk{p|eiCRD3s+cHt#$NE5wjrw~I*;-DzE2ki6P2SYb3-RfBfmb;Ar%HEWH
zLBnxeqvUT$S^XK2nh2k)zI|aWp2QR7(bPdFWGBOF)d+uOV~j)=q2~Lde90X7=+YJq
zAa`77z9r)vurIEG#-$t}l(K|k(Pp-Uw-=mx4RngDyuYLyN8XKbWZ|>}6Kgyw>i?xn
z5Da0B?U`ovBTkK4ZJy)^{<55|%_EQ*r%?Mg49MV?tmASg)DX|xm)9$GHfrUS6;X`M
z5vN8Ih-CnCP~X0y`bJ*=XX`%)rmx~%IA~wl<}Wy#+>&AxPdPP0PomP{?!T%HG16E>
zsGI}SW(V6(AhW(STac!@w;j*7&{YA`W}E2C*o;FRs)m<6^!$qf;SPZC!I?}H^GLL%
zB4|S%eRPERmKaYS)$ZGIsXz$&z=}|%Lr>@X7qfKMF00?@z}2Y@_u|0v3-As~FWt!|
z_VR{hYlraG9dgPV`V4pcOxP<Pdj8rRX52K#JVg>H=@(#**=jI0@?a8ort?;@VR8o)
zD|FiVby7sG{aI(PY+Z;GyYz<PGfGVkh_{6+VVwMnAxxLIbqaNUhC(HmNKGKG`XTI6
zTJ2X=>pTE6eQub1otkQ`;25FIlXgXY`+suA<>Ttq=Wl}L*CF+ky}Ev*^HMoIlxt>5
zgsfAaTS^KkCR--(*Az@5m&Ccp=8rhAK%!l6uV?JF*#gZo>3C2dh-m{BGZX41=3JNZ
zq<F)g=ID4`J7s$Bk;m(x1JKUAX8Ce^eVa89Nz4f1kp4m)s0q@ZH*oRSuIzezj7^^R
zDFf3HJH(w?x--SD>b|A46a66v5ICj40^}ghHATCfncr_$H$p~4z}@6yZ78O+aU9mm
z>W9VYfiYsEU<LezHHs$G%<@uO-f(G+d;uxdVfsJw{9IEb5!%b?T{;(F)zMQh5c0>9
z8JxU+c5T({k7(wwjRui9;?(5vH_#)fls#-K5`v8xcSN@B4hItJjNz?JbWhmIwu)Tc
zLe7}JJoJF+U2CSX%TCF?)~?G5f)0$2qVc4dyR*~9Xj3Up=M5dIDIzJ{p6H-1LW(Pp
zoO<onNfN3}+Z)>|20e>lK?f#!cl+8+TkLDPxq*nW%5Z)TEO~;ByaG>hB$78(GfB%#
z7kaLz9GLLx#+%pL*S5vI8J*^gQ!Hecr=a}x&)U}&Op`2Pu$#jTSZ;FhH>e3v|FR9j
z2NW85(7wJ+I6)YZ_|bvXvg!JW%uNs|W2+kVg_>P*N_lL)xi^6a!u9oW9awTuB+0D<
zi&t@gr05hv8O|5Bda)*~x17y%Sdrt@=(wFEOnZcdoUk=H2P^O#=25e{2|J{%t<LED
z*z1(4(+X26moS;>?S?i{6-*`m8dzh4uCp6k1LlT|GhA_X)E8m$wUx&5u&sNje99*W
z?n<>exCk(oK?;g9pvVPoP+Q-o9J(Q1R{RPQ5-$=*aLK7J{Dn(QwYI%^$$ZG|1~VMX
zbbjF(rzSClO-$0j-rTI+)Hdba_}w|BcwGr&OyV)SxnKvLQ;bCDT^(3rd%$$EmyXJA
zX>Wieq@wz`H(00&Ag0G4@wc`&#D3Ahl+F`A(5Y!h30Ou?(8p<ED#e&>XcK<p*ql?y
z^J(okfj=9&mD359c9A?~s7Wdup1k2D#j87=t-Ymv6ylektLtYNuw0<2Pqjnbtq$K6
z%EwKuE6c9C)+r<o&y2H|ZF8d9t6udcugBh9qqe13-VlqJ3dz%wLnKb5)gCRW)2c8q
z{Y!}q2XSQLy{%F3{yk(0Mb`iziQEuZ>CKLi1n~BUo|k{jO}98tIXSCIp=ft>sFJtK
zV|sL8F)bma3%yuJxy7}VYj15Gp$797`yL9$nxAlLI>eK3jtp#X%OP)Ez212&+M5^t
z%-A>7tMn;(W$Ws%ObSCybt2THtdIl*u{W{T?d?@B9lR9(`U_S=P~xw*di-1Y%^|uc
zUsZf%LeM;IcQ%yF_0~Lk2MbGFYQNsmrt20ZX`~bb6l%&D>%_tyduJQ8B3KwSeyrYL
z=(i#1;x5kFyV_|vI7VW&8K;~mjaa5TDs)$>xRnaNwRhKhab1BglPYSD0}J6u;c%#Q
zpn<AU7=qLa=}33LdWnIn=5*aT1!3UkSu~ZLb^~n<bPZSd=K@K*Lc)+*RLWD$+EE*<
zn$SElGI~J$NsbL_5{Yok?b%TEDsfgbAm`f1%^L5_-A1b-Bag`^k`*CqYnwaHX|eox
zoI0>1BXwvhjUxOFR>iFywg@N64jC)gYEo=Xt7r7ohTB)eedQpRV?OS{k~-fE`!kY5
zVe0W;cS<?{h>N@0Mq9I+CUIJ=)k<kXI2blkRy?+e-Bo={#TqLUT2dn3otj!_vYs-F
zQe>36ndif{t?Kno6#X0$a-8T6P^UhPjayl<ov;IAZGu8V7MjXVBk_=$+V_Z&agFA9
z_wh>AN4Lbpb9#%>cj|LdFwV$>6BDCN)NiB#OYcI3IS4I8cMm!Bnxk=D(>a;DyGNc$
z9%J27YLIGCgj!idrrJbq9-DLu-4>GnlfoL_lWU+${9ukL<$%#}<5@jz)A@UDAwiQ{
z1)^j7TWzL$^SZT8K}dPq#OQm7eV=V3!w@b(UUzRan9e3`Rr%!1<p}P)?>?``Rhw_M
z?cKbk<GcQf;g@BQMRV%)t<}>6Ys}i+`Hl^n*E)}yob7~K+C5ztBKZQG@L>lg?mM)%
zx}I%EwP_({Bp7gDg)w_W0)dTU2b^YqYtf#xo!tT{PZ>2BuY{L%7nW0>O0<n~i5{|h
zvrk+~dX4j}j`t@f#1ejEef<!63PRSQGzT46emyWX<Ytyh-YRkG*?sLuyLA%BOsA;z
zlq<~;awa_W4tsBYgf;4o4>+Y*@kDB^_(}Vwf{BnS@IZ-Ufdv9-LutTar9?|`#=N?u
z)-B3bq*7U+AX-j+d?8iCq20IlwOz7AZ<U0In#<@hUr?Szy+L_E7uBUD>S}-3{cT5a
z0|Pu+pB$`RU*r}TqoMy9R*9RCJy7-OohNZvJc@L>5rOTkzBy<pc}tRg@Db+j{fSFA
zp<m)L>C|{{#J;%|Hgl3v5SA_MS~Ftb(khlOyUHmP>L^2xc&B6#73HIw>b|vRO8fal
zAk86fFgfl^EE7xdZSA1R=xwR7c!TtOnG+z;&1qGnI+dy|@z62*_Qq5qR-tf^`$q@b
z*qg?IQ-0<jXun6GmOOztOrIZr?tZ5xvFc`ik5A1ADyi?tL9a!{<2*}l*$NJSXNNXn
zd6$DUi>lk^s1dFQ&-aXd*AkRVUx<_5U@7ItU;AKwW^8QxxbqwX)B*?R0s7@G_)zP5
zoKnf`CK0A4mSCwCVacVY_Tha0)b!kt^W5z<y<hh~R6x5tw(KL-XEDiQg)efR6fAt#
z5Jj2XA^Yx}j*V+qJCD_6xgaE#F|(*DzNgXExd{?%V#Nm#^P{_{F4^~XsFCR}Ps|^2
zVCB`YxIJ<crU#QwY%;9YFT=tgC)gPWrgyKJAVO=}zCS4vac*)Mn|8|8Njzj@;Y+t<
zAMIu&x!k;-M?vVpPCE5T+cuFiAA85gs(tn<NlcLrNy|=sX2rxtquivS9G$TT+gyxJ
z$W^rO0J@ddDV^yLWj%VRtwHD)QisxD=@JlgVGlQqfU^m=%`gWjs+pSH$8!^-_DH=h
zsZVIt9auKFeRRwo?bhJ{{FXHQ=e)tRiAC#=;WY3A-Ar}ipJ2#Y=2N!%Jx+a*-E-og
z;15sO54PDuZ+hU)w>xDqhwI0Cc=q*eSE44dW_{<m&{)R19Ox3}#~xe!d|1Ziw5V+$
zwY1_usE^N!w#%wXrU~0!FcwR(gwVz_4lEb+CT{qEyV=v;8*H%zZBggW|H7%S_v)(X
z$c|upJ8D%a=#VlS10udB={~LMmrBkTN+H|V4m0Qx8lA3q10h*EZGY>_lGuzT{;l5X
zGlE+_NFgIVhpK_&c%L0;U!2E~Lyr_j1S}=Ob#voWBX+Qwgr1C_5j*4!rq!1Fl$7I#
z+L1~Yn7&hrq2Ra#T>#%nUURtNU)+nO{h*s5B-c=*DSq}r$oqqc*^k$-lIlsaNTFQP
zF*-DF_31G!Gr4^Hvz7R{Bi*#5&OtYeW**l_)8W)78;U*wd6Yn-C#r8DQ_)sA)k2T!
ztw#OT<uYB~K|5Ms3Gc>07MRiJy2R&BeawL*A01nwm&XbbTKKV?Po?V++m>lDNw(p%
z3e=WdmRHg$`s6qqy;k9Z9dCyh5%FVvM4TZ9q$!jmIu?j&dop**s@t}#bDnh;ZJ3&|
z6DeljHs(Bu{w?gfnf#shRDE#?msZkj%!pH;t^4Dn!}fHR*4~3TVa6#XJB;l}zLfS%
zO=j@43~vv6iRr`1nVJ}ww6&7Ik0FR@H|o@eeYBG`Jz~+NEK(9~O}w6^9FYKb)M=F@
z6MGB33tTq9?Ts^bsv+*=PN`}s7Qqx(pkK<h7_+IJZZ##?hC6Dwpi&b{rfcAplMEG}
zohcaol<y75e#jSLCvb_2!_Urk%dK0<^}Qz@SQ=jN`@r`y^_-&^w9h@1F{cbJ^@6bb
zFk!N91QR1xOZxCkkE9qzev!ionR<!V-_EzSSts9KN<~C)R}$8dp6Wt3EU`hcYrVl(
zU#M@`Y3*XZiOf8@1C>&)tNf5j*=&~@CDhf+9M6<fy_6%{8hMNChw>$IOP=mIrRa`O
znXJ5Qf`j(MZJ|a`xz0GnFdZygY4iEGdc!8?QAf_!={e+=XS+dZEmKVjEO06GZ>)jr
zuqE7viDDu9F^J8wKjf5>ienobfFXG<H^Q4|oF}oWS6{j88vBv_{?`70^H{t0K)Br;
zL__j?gF$f=y?+tf1x)?qt|-zt{Af2LBn+IMo{KB49cayg=<hzL>`D8vwwxG7lsG&>
zm4g^kQww;QeWQc^ar^PsIu@T%N^L5mm{XJEB8#Nbi2Ov3Yx{`vlxOzl)vN6%YclYl
zAi$MU!zF2Er3M32&fo;{Q&pqsM5tca8#r{1P1kCWn59%hXl>U${&f8&H)`gt2pQX{
z$s;;^y@g{Df(DhB=zhe0riS_edWZVY8%&wXbqF5gZ28&xg5Iix7E14w+!i&WOQO5;
zr|jpto}h(}!(M*oN-a-K^dvQOvhT>Nlhk>T1nfNa^KIND7m#WGf|Lo`iB65C@kTCp
z8n$1^+1|jV-U?84%C%PVZIYSnN&CgbC#6|-o<j=g-)+CtmSg!f*E)rwgPpN4wxDw6
zA&=QFx9)O$R=@z=m4Ged$!@1UkZvdOr}iry(NtV-x->5_v;xr?ORHFO)eEhTiu6ji
zXusM95oenerDq*jZLTh9O3akdmlSvQYwb7UVR;+Ybzg+Z0-;myfc<(y&+X8Iy<<+f
zFi#g!;u533>Yx2ay)jQlIjmlEV0lh+;nZMchWX8EB}pb(sHYuRy)!JstS3>Lu@t}6
z2+gV?($zV|+!5=C*V=Dq&vs4`Y^<yWl(keVu3llklPie1e7o};6QSJ}p;n-<B=Wiz
z*?u>PB*M3{2@qlrATUyTiHr7oSz6e2OK)bIQ%Z_D#GTTJjrRN1bsf9NuQ8}_JoZ+z
z)b%Eq7yE+_ca*MD%6M5nFNORr6)g490kXm&H2q-?^X?H8Za9%S<-}MRaVbnt&DbCP
z&oOe|%2O%ge9_ub-+8>S@qz{=(;}us8E%eo$oz3b88B36BJT-BQo#OU$=m1D$1>^2
zlg1(MhJ6|(>z}kfg<3oPUV-IW&2Xj7pH_q9wA5g?1IryvNglzS^}q1x&$^c?XRP+a
z9tW0-xrBAsu`cfw!~^^D!~r6(oX6^IsCVhJ1UyvT_80BNB)-*+D_j-iz@qg*#;ZK?
zpdYPu<H}#QPsVtYMCRX^L8o5pv9PN@Iy^E<#xT-;|5d%MRMhZC817>eQ<JUn?h(Eb
z0cZcG`ok*3Kyfi}U_$2M!9n}$wg+aOeE<%OVyu4>2o&cFhwX0)Che(ioVZ9deFrNZ
zO>T^T+uGr>D-uFhAn#ysdBQ`nr+=59+}7_yCuf{uO{EYvPYxMn_S@gLMP9aYnJ*Sr
zTy6i*O{kpKS{wu39!^aRK-`k<tGF<K!v3*QnTh`0@Ptk|+9Yd<85qUVKh;9y<U{iz
zrui&w{L@aow#~w>XFmqh_s?w!I0uE`$)bu6qb8T2f6vs|B%CM5!oRfdUq8p?BC}4p
z$@DOOL>G|~od}PAt(PW?7t&UA;k;9mgHrYwW|oY4_HVg{t5)CWJmz>nEV<f)jP~zM
z1|SXRcu&_aIUo%IQ@?_m|6|F3D~^}7#n+Ktd3636r@kiX2@X4H_#U_atlu1jFw1;!
zObfOoT#x#;hf~jq?2Co~-1x83S+<bW%qg-M`LcvdbK1t2HT~)Qi4SQvH!|V`rQ^*?
zB0)`2;1kI1!B`Xx3Le|+;iY{_CkwH-$DEoxKG5+I<;6mMT!Y-zC$f5aXomyQPPy{B
z754ED=ZA#EB2bHJh1`79$KX@5l+F+g3zh08JX}mzW@vKC)iO8Kq9ybbA1<=CY{g2a
zP_|cYOqsDyYHuO_;qKlJryOqyw@8M38sDK@`%mtAiczJoSO!q>atalqK7T2qQ9CXc
zz&_>SI)PRVZI^e61IRvb2Oq|!PklI02y(-{$xcZ`n4^35tbN+UWriOjt`gdVAp}q|
zpKIo(M|$J-=?@1hy>5=WdRFo%`;68O$zr%a<dk!-$tBo6v&|MXTme6_MJOHRNpCe$
z215nlPeNIYqUN(|2H^YP0}jaZG(-R*cBGJT>T5&E6Dr~p?6d1PdCV?)<0SPz>C_8d
z2FGOP**c1l@Hr2cHB%;)&>L!o<qRB1&lfoyoK}ILV%=I66e)z@UJ-Mjwing-;$UO7
zv}Wi5Q7uuE^S61buX_n!ob&hWN9XrC&v^ijBDQ~SYseLl{ob5Y*6<c~L0$&A5?WIC
z=hcw)QQeeulGI^~Fd<J!&-Rk`E^b>u&U4C12k4Rn!Kd+1{``g&Guz-BoMPg0@WqjG
zfqB5S+rFT^Y#g_vkr}6~IozzsRa^}RXY31egP}}v9yMGpvxIDdY+saBTYEFkW1`{R
zx+ucFxIUO$Q3_|-_Kwc)c4~53R^vqcC9QD%+AU5YmmE{OGJMH(`b#@l3c1oAbTGEL
zv=6!?;LB<W(~79Sk5Qw9PoYh7>h)D6{U5R~ua}DLkq7^298-ZMXE)W`frs;$eMP-7
zRGsu4vsg4$h7}<V^?Kvx2>tuYeB;8daJVHL5JnAHe3B@oY5S`BB!?DbLx;iIHkc4u
z_D{K$y|jHb93!Q>mEpY!uunPl+IaNmX@s7aJ^cI>rPu!d{Ezcp(cDd|?y#?Jlwfvr
zYSJm?EKX1E=#8N!xvT+h7@g5gL=Hedy>sCKdwE+1My{wglz`q~A_yqss5xq{XvoWY
zOV8Aib#FCc)(^sw9wf)@m2E-TAXAvtxG>v+6;YMrK-%L%DzM92F9-VE8O@dgptYME
z$A)RIs(#9$K<I*|*n#DSVy`OLIxn9W?{r0b7Xm`iSO7s?SZ3HeDX^7HLbmG&1G87R
zAN0qGs2!ek0PzFe6Cicckzk-Y-*|gX_2)xSgb%{we+WYqFl|2~fnY-??6p}W>nQ$K
zqfRk)mQY>F+{dwxzOKFk?70VG7s(Q@e872FwtB|KsKuVk^-<>F;Ct-#Z3ZB%qD&W<
z2l;!y&#8&89fGq?S!=&tSzl8eq;!ENlh>5y9OmtiQ=fZ35`6J3%pzAMF}?3q_uc6{
zHy%Q|Mh=n+N9LJiV0Lx>o?GlteD!do#+3$G{6<Hv{Jb%92%{7T1bsMW%NpsM9KfY%
z#wka8S~4S=E8T=}QO**O$_utUQTN)bS9(3(%F%dDHxH@VlB1}8d)}$ZL-LLdn{Kje
z6Q$_khSg4?BXtthE-sw56&);sosoYp_#%u86KIETiLI=5lJk`WoDMp$L>cVsq4o?+
zi_5ZDvI}<I!_U7O^!R^4C!C+uDoW=S*QePV>NC=pmY*qO?F@HLy?(Q>YjvG!oDpx#
zRUDo~-^g>kQ(C^q_~jWC5nt2Ix+rXh>k8X=zmldn^*M`cZdrSay{T=MjUzp{W>U32
z9ByYf-)LW31Q{|j<W+zmuNttgYu%0wtWzG1hy2G2)v1qeMHo4o8O@k|eQ^$Hj7enB
zU-2Re$s<sXd8d^Q#Ah?9!!x$3-h_sWwMk1Juw0u+AS8i3oY}A_FagiqZ`XGdkWLA5
zwn@+MOv_5W)SHA}^=x&$KR2Zg6VkRevKA)p)<+zDZB6S9Y`EfU?}uY)Fio+Tab13}
zwtA4(An85uuHr7&`(xXmbLwM$_ATs@&XD8ioZZmgjh0B7Hbt=j6K!90%PPCE`-!M^
zMkWEP{UwabZ1Ld`5_|KYt!uwQAs}-P;p`494-`BfNzgNE>l>eO6RwaWPO+9wjB*Ht
z_U3L!MqYC^lJwTnx=~Z?F8*=YSZ?a}<n^nag68?E>#ka1H#b%pt7Yc!a6t8?!(sl1
zOS4<*G{%Sq4Wks})8dHHmv>1n2Y+FCZ>=|oflL>?6Amn~o4i!xwxR2}f)b?q*SUzZ
zrc+bw$&TsXI7uHi=3K9v8}WMF3Y+<s+zt0^cb<|TyE$|Wo9f%R48WZ0Bs=L1mS?Tl
z_T9F*jRDq3+9Tb1EajjA4=}_#^(m3D{f5G4VoUuZ#g~-sbVe!+f_r$J`lX^wsHzeD
z(X8CojaHO%sZyAlnClF|2>NZexA~9oU;>e0+JI@0Burw{&)6Ngf~#&>yV-e8pW`)g
zshn~VzBN~WaC(P~A7f5A=Y9a*#NO5>Y?X{7PBFT05NSkHaCrW-eM9RO2dLb7baN{(
z;r!*Vdbhp3{UigTA-ww>Sh~@?b=2<cNXnTIkb-*+VJY1*z!LO3w2Ju>=13zgf*xr)
z>>V}J%rX2@sKvoD-Tf!Ov9nJ7WppK0O_VeOZz9*OC@uW3I?TSq&(7XiU&-}KWlD;k
z)Z_}oM<kd3uKI3j6vYqgykFEqP_JEe1MbHYgaX*R>)l0YWrG8IE)r!$jo7M9Vh*jB
zQ?rGzC+AtS*qa@*fexSQ@<ZXC^pAXV56<M&CvL<gWJ-*5)&|?mhbmE{-S5C+H$tpL
z8Tg9;O2!qmCoJiH8)~!wS3dB;yYkKbaV5R&ci#~~W(DAEfF;J<g3q&UZKZV^ZeHV*
zG=$OqT{henYB2*+zJO25Z-V+9LJUi|b=*eUg0NsOjD?~mTEk9-xWi_#(Sm8CmG0Al
zC2^aX=xt*++*RnXgkc7cJ6&la6%dG|(>9PUy@^xoNUi?1`VvGf8C~ww1D3R9<4m91
zx5io%Iw<Mr06Odj)4WIO==r@iUhNcyty>@tE<Ejvuyay=h>y03Y9|awKdZkEEU|J^
ze;B!xT)mTRxLfY-k8STe<!qtbMfz-To2o&<)gp6MOdGW9ai=Ejo+CSkaA53B56DsS
zJ-MFjA40X9l5lll*OpD|ZMt67bTlbgf*9UvlF3X<To^HJ=Hcg+89hR*pU80Q%N<~0
zukNGJfqKYh+x&sIlKbT)++dRUb(sKHR@q#)NJ`@q3Run9{7I)K--=@f@*?YS(Y812
zylG<c9;cM@ipJ0Tq=w9dgLk)+W!>b~t)zT#07<gAJNTaV$&TU9BkhP&Qv(sVDnaeA
z?Woq$lSl*9#W{GZ(UBnasIr8mg=c4On=4mb<vi-cgmTW>z4`pox8p*@bM`&v2ZR+9
zroZ-$jmS^;$0xT(KqR2zw!%?JM3}~Mo{ZpLyD$5_uxtGUK1jWA2QC;SGjbx4oa2(c
zH#>0kl~*rwo)fj^CI&{itaZq~sVx~(U8#W_Se`gDc_RT*zqje519_XF^byjDQd4gB
zyCa|=ydR}fXH^zc!M(5hJr)><$NO?GVbP%`KX!AEfZUO(-jLnjrin}((r-bq*koLV
z>eM9ju%?G1o15DnXun2@h`~bEGe9D3*K_tk{NA6)zk}98PC-@ljHqY(<~DRzLoxt@
zi8q)?jf|3FpueTHgNu<!avT#Tu%yW`#ys)E@t!>=pWSb*-{g+X0~T1UezTLEM)HeO
z%f78!>ZLZ;O!p&dO?{#0UM3`6=)Qe%lEhnq){^Mu*-nkHkBeM}>;u&b9D}*dspRkc
zVW%cW5NCCqyBVKSacTA)`OSr0@FR_L1lH|6Y2!n)AxYc5v)(3VT%IWkhRsYr@-sf*
z)YmWNEQs1u_yO<sxP4dm%L|<8;yC4lkV{T5ty5p8IDQPnT!>;HtX13vU$NgQ<zX?=
zpPQZ>w-42q5DM6!qJR={9s^UU(Ybaz*WP@%k;Jiq8K=<gN+lbVT0`-A=j|iiybBw0
z-hj3Kpijoj*}l6QfwC*4GPfN($MMjiK2{IM@eqBnX4>~Onug=jtW(N&4uZ5K8kn)~
z%?{qpm2rKi5FJdda~dQfCyaexHd}w|nhke2g;>!%`b23<hYC?ci)i294!gAjlVd%n
zsAFL>3WlD2wAJa(vPq|?y8HO&GW%HT>s2esO+YC|Bd#uF1B?z2RtL+EkiDaM6c0#E
zVlGx3VjW>O57|SFVvxQ~MmXVuNT?(U)Z>B0=bcu0YLPk79<C1*D;X@kOhTu;)ig`U
zqNd=Hd;!U@u(wmFen^HQz90S8qa7C;8Y#vjJ5%UB7U`GMs`iQONYd>^1PM5z&YZU&
zc=-9uZ!a1gL9v;^4|ptE>O6J&RYF4h!S0Q6xJpmRh-BrBe8Z_%UkQTd${w`;3cTHw
z;?SD6I#0>6U|l4~#$#=9Hjoj)DMe>Ve<R6W_6H`FX=j2-9&WpnH0;31+<8vI4f`U}
zkQaz8xyK+}m-cj>T{Q7SOgJe+;`UZACr%|8!{!qVihMEUOR=|`_*9B;&7A|wUWUU_
zZuwGT>Xfs6ImAuj4nOP^g$Wy{6-2Xt@qP3`V>{@;N~FY3T)y%L+J<2lE8UhZDMm8D
zA)K0=y&EvY&Dg=h5nXKM6mpv}aD|7_C4L3{utW7xt)<SjBlE``STU)0bC2eX9nSI)
z)N)u}UX%mlgqX@Gq$>y>&&3+u8Q-H6(|`gM2{j#ogj3RMqUW<CNv~F{T;V*%%gR(P
z=UIEAtvn~)j8p15uNTOhGX!@#x}-+I$*@d(jJ0uUG;bJg$j~6&{;_=J=DC4!s1l_d
zq3$w=3q9V-xTcR2ZcABn$vsDL(mt8Eg`&3SJZs5NS&!oASsf7JLbJ*aQ%*SLkEbSv
zmCh8Jqf_=&!7_G2ikx!}EM`hnRhfXa2W7`#IK1b)J>BLSijw}<`y5#PCh}E=KY@Cn
zwIF4iJ=5wX$nc)962lkB{S4Pcr(UPUz}yZynUAxhTb)Opt+<FCwNo{&m`rG_T*vcD
z5o!`4M<>{&Y~Rz>H1jct7|4wR9@42vG-1V1?MlS6Gi}0Pwlq~;%%fA2=!?ae>?mj3
zxgqBUXwQrTNp5PeH#dXU9uDYSec>P+lt>>Y8V8nQL)@cGA{7pS89SfjeCMlP<2)wy
zDmARE0-;m43sobhhuGxkzjJR<6WkS_K<9n2O8Z!|{Z1)|iJ%CXc#ceTcBu^$oR>93
zS;8a+o_1=oHLEKZev;<d4<!YB#jBkst<3W4mfH{Ka$GMZ2+s=J-7seJ`82W9dD3!C
zN@?<J_fq)_ij6zyz|>`8gJk_(=t{kbo>PiWqzNu!M(w#m8ttNr!T_BXl}glW;dN^p
zxjV5M{YaZfDZFWV7|OyPM^L4v7G4g1aZ1?pxosA9y{$Lm6k-XqEp+6ZW8VhLZ9iHy
zibav<8LslYnuE)!(bR0%vcZ0=brsB4?wAb1?0o*6`g;8#2&oR!v^X8IAJ3OuPs|OE
z(J0-2ng&Ucm_O0Fx{VMO3bFFaLHmULWc{e|Ueb{w$Ib6`YHD5Kp@AVk+fOa_Q+!wS
zmVM4<NFqhOgimSW$I020ikzQrqzIv5x<BERwf3ur>}RrH6VJXMtr~^+UPks{(*ZAk
zwi|ccfPtKusRE8UHMJrs6iZV5xq_ik551Ev4h5FSSGb=xtATvTguP?{&CNwf@rb;v
z2V)yL(4?lsA?fJ(g|<_5`&@rOnL+{OZ1Wnay&|<MwBn%sV%N0`A!O}Afe9>W0Vy>g
zF~8J}Rq0X_9e~9Pu~l(y>1>g<?G#GyUvA?<ClQt_qzqs}o^`9o>{r_AZl0a=DkyvJ
zwEb%CC?c|EoacDGLSbbdlNc2rv0uyAFYHp@lu|o)upWNDqqCeTGxqCkDnT6aAiE<>
z<r$7Z$=4^G`uIX}chWC~kb|P^H?oJl0b+5SLN$}1o;pa;lVkRqRU??JS&EHGyrL%g
z%WJN_#(t~%q29No+`39rchMh(jcxVfa^zI^7-UH*kMP0ZsQ&G49E%r|pg{@$U7fV8
zz8w`^#4-m4N^$$0S`L*`S$)xgX*upd!uZ`rn697Oxzi~X{)yS;v<lw<ND>nW`(?k^
zT4QM#b4rEP=a3!g04;vMp$_UnqU$L}6OD^&_kwTxgZ7d&)3{4e%1zjt(#bQc%n9~~
zZIQQ*a#P^|zACd0EE^(4Op#2XM_h*eQN3T@L*dT3@Pq>kWE&s1Kdv>($q_jXTt!Bu
zCdo5G%CP-OD{W#=Ii;8dbvi5zoI`1B(((F~{b{kctbFisnee8&(_M(l!qQ4ufYJ<$
z%%4?9@|qeM?J6kHsmU4AtMyZ2%Kkie0^#r1I?qwB8iio-*Z!hl;0Z_>OsE6P9TV;l
z$)CqowZE*^kVLVVAnHB`mLvnUt|Ei}s)qc(<UqctSRLw<BJ@Z+jW=e)lVB}c?ii91
zvQ?FZllo)=HuWpiPm!a4owNMxW5{LuoRXI5n(MB$zv<qtB+*VWh7AA3HXsutHkZH6
z*7x1_dgn2ghq*wk55)wZjU*Uy1F*lVnOIciw==BOh-hKvhD(anXHdbvZ(DZVz$&K@
z9}urBt&DcQp5ZQZeW3kAH~Pc8MoXe`FQ#&#KF=(cdB6Q*jYjvJDNO_x4R0_>XsDrJ
zoKDz3<*(NCw~FpiD!PM_GQJT;RcyikTwfPgDAK4iSmaA7XR(<%^<qqxUw!43_AmAA
zq+8PD>~Uc6X2>AQF9~mA1W93G|5`PnJ>i{V8zbY8x0>X$BX=4#*x&5m>I+GtGvF6k
z(!XWbuC#ye8Yro+x=6myww1TwA0EdpGQfYdQwvq|P$cYiK&682gQb%@cMKyC*nj35
zb;WpGBMyJ*z>+P#Ljk6+rvKIWS58WtyKvA2l1<>B=Hw>%-wC({J;<OE95?{2+sqU`
zIDNB6Y6X~EG?I{KLVrWJX*DsQT6qr6KtubuM`~NWxi^TzgjabL7xB#yCu<*H?-Vf>
z<HemJI-2GWc&pKSbmx$CnfsI>%0A(dU_f4buk)lGGdw$OpZG{_kz9U$G=Io}74epe
zTniHFRJxst;7_XGC@?=1Ve?14!Q?Hd#Ehp<wSV#>DZOidNr|FM#gc%Ec3_l<)Je2T
z#nRTX-f$Q>Kc$;6lxHEXG>2|#gb;JHvtxbx)Ev(pw>wY5ROLIePs?Xks^>hH-goiL
z4OdA9q9Dg0N%rYAT$Pf))SwPi&rXBBwECrz-O^qB8Ev}2WBGY2kStFKS!a$p^@%S-
zVI1Y|`Zj-{f?Q?J6T1N{8_AiYKQ(Kg)ivT~1K|dWqy}T!*|G8hI&7cazFu(#i3!qx
zM?w{+UTZywqV;pyHzp@?TL1|97-Ak{M)tKA6$}Lmi>s6CtT&j(MnWFO`INo5zMD&#
z`T8U0BK_J8r$&9fTRuI^HyHT29qdBrDshx>VCnpgcsDu#+u5I1CX3H&STPG1hok?D
z1F3UKQC5la$p^foYLrP$rY>!XysV8H4StG{&Cl;vS;SuUoXT><tImPI)JK0~bgVpn
zR`vynZUp6z;j7siNa4=Ja5UQ&HW=uF;TQxu?ZCA8r?{&P?(vJ-kVDlg4c0*imS-la
zGchS&Tx~QasTP%21X!{TA%};u1tjPqcI-=DFy^#`B~hV9o8boJP5aW;Ke%5xRN`RZ
zz>;Ahl7Gs+to1KAd9Ffx?tTZR7Ld0|On+ZqEo5gBM+vX*z@m3@QIngOu2B=gk+UUC
zo_)n5CE^q46BiYaIiMseWS*0Lgd2-^S$CbHePw+{a<`^ZI|o>PCTy6h(awy!nEBW#
z`>L*M5-+8pdD<H+9fx8M*h{Njdgv?a<fJ!P`hAFbzhp0~@8wR{8N`XUgvD<4__u_J
zBcFmiwd2MC`|9rZOEZyl*-@SPz%ea^!d3Cxu*=$}oSEG*CMO#Jxjy(4Znu{|Qv8x4
zl4hJ@l8IGg!xG5s6>SQT_9ap&$PrkOJz<;dahXtAN@HP|M0{P`^U7{=Bgtr}lf#y(
zdIVGIqqFQOm{a{dZ<i;@eaooxRENKmfd?@4E8JNA+pnscR8wnuB+y|{aB6Cc$tPSV
z#TmPzT2~l8qtIQuav!25TBAD%gA!ZPUY$Ef=P>BK!ztGrr4#0^6%_risd1tGpija|
zG2Kdh90bm(ubwMAN60ohn^5+=Hs6Lsi0xr^bPVN)cQ9P>vFDMmiJR{0tNRhIb)zm}
zPTs*F$}$ohM@{(pZYHFCf#e{@OH=Z=Q&XEktWmPPptpS0*imQf%G^Jt^+ZE9<A5~O
zg<T|_AT9vq=MKB7A;-Eo-Ni;RkptBbyvw*<-3ZNQtoW$jDaX)LR1JLx6Qm1LmnESb
z-P&^=V@@+4S8N4a-u*BgWOU1;c!jf0O|fVzSJ*Y}dxQ8Nf~}Nem`nG|k~Cj-ZGB2E
z^2Hi>gE6ho>Ba~oCtHzQVioBJCWf3+j6zB;u@#V1IwNwtGDopt&D)&EQZmLBC%RkZ
zy4n`Sl?K0?q$^MJ@H%%|i8$Xv{)qLn{c(FkzH`O$mCmCm>vl|P_QvW&I17mx7XwCG
zi1SWO&dBnsm)Y0U42Y;pONcKzJ?^*>U#wp$0V<X-)o5?(W+)S0qMLJIOIA?`u7r75
zGwo}uW<P7H6bWhjlyKkGKoVMtshqQ~>()qOW2(=zX?12R6u8xk4GjNcM8EWFOC9ys
zw~1aIceP050F)g>VW(3^!dITus~W<t#-nY_DMrJnWZQGOcIaJC+4VK<;<=D54tRr=
z;YbP{TiwkMq>VQ`jaD33VBHH`aimIRL`0?9n#BcL;S>Vf`MCxrmnB=<x|P@6g^nbZ
zRn$m&g)nm?T;FbJ8*YkRk=mpH;?tTy*)~8*{+<lV3@>Xp*4K3KDeEXM1`aGS75QSQ
z7p`^d9_cg`Xy-?#g4O{PpN!H%dMSmltZ$zmNS8bZAk-r~1`K_3n``CJ-sTi*QFIU}
znGtFZY0mAYRufEuj*Cbh%`s0+O^%Xz#GvfvgozGeI+?^7Xvl%(j5A>phh?Eya+KID
z^@U^+nl?muNllWh;W>mg)JV5B60&7-n)f^9+)Z1M`!}?O0oxU{th<H-%bl9&skl)|
zp|T}!%vT}DO~H5%Ipq{Ih!x;kP-MHewD(~%6X`t*x&|(Kt69`JhrtS^@}?Rq^8~Y|
z=#;Rsve4ow_dE4Mkx7gyV1;eY6LaU?&QoeWhDcRAWLp;JR=<S@J?p@-iNe1R*=<So
zpeuXk_c)~(V%#Vdr^yMsy>*7J>vC5qkd}B{JN3y+Yy@lUj%uwl4@Cs*lmdFXSS_bM
zz94r)Wsif1*jrnd)+nOiDdvuaF6&s;J<xi^aeG@^$tk(WFx>#kc1sLBP02HvlgnQz
zs^#Q&zkS0a&tJxL1;f{z+{0?Cy{Pj!!c`*v-f5dVh1HzH+(%O1w^x6l-3SGdFN0T|
zn?LH*B;u@Dd$ZkH&B7xP<PTMa7SO3l#DUFIg#BSSh<DUHg^^9?To72fR)@QYw|b7K
zI0wV%4#O%J0bE1uo%OXPsy@2*3{oC7u~GC(U4yH@jC5VjR4eal(+d~Tf@WYA;mN6q
z*<5mWgvKFJ1~RQ+5d}28yGFAQM<LnM>5Vs-)>G}|+R<x^wRu$TmopAvDRE(M477z`
z-P0wuU9mql`JOo&%nxDzCwLY^l5hi<gsm%AF1MjJt4I~Z7%I%f8|)zcmG<xmhLwI>
zmk5j(6D=KsyxXZMT+B?eWG1q$x&E`G+qyIafC3-mEMr2#onNwr4d-FKu<OQKo#zs`
z=15gHWFy_Lq`AriDpQG46Q9)r;W5ahp0UxcwfuBMb)@>2W|o@VUYjS!Che{^sWaOU
z;88yUNZi8uB{Myugl%g?X#K=i1Vqv?1If;<>!IPDu(6y*jQ`GK(QIW?$ao<Z6|c?l
zu13^c7_al1t?bm)2r}^UiTOP?(LOObEwR)ow_{4Q2V_G#xfr)>lZ9+TeWA<&&gK7|
zB7;`iVXHQ6TJ<)YYM<CRGKt>HDOWe1fgnrWebnC5!IaBN*m=|&OrMhFAGzJ8+ddZ!
z9V1c9tOJT3r!Wu#LYzktGXx^d)T^~6WH>#9xB)DFqirnjPnkNju4NRp*(IA%yn*Ul
zo(1yDIQ6-@gT0xa&9yT*ypNziLV19Be65=sAe?N-wl@ID=;63096-KgxHmU!cQ>N6
zdTi7wh@D*6weBXnr!`wcqEV+5lb1q~#=OUN6fDRx6H1CO2Nolg8BSp#kw^rtPbVPv
z+1bu+)*@d9qDQ$e(-w^KRv%hhek0wHPWeh4X7?ry*gV@ulR4oOb3#LVr=yP#Et$}7
z?7Dp%Vyu^*bzmY28&_?y`*OTQ-;X)ZdD3yG8XA!IW-I(koyQz|i9^RRq<vExJ0F&5
zLmq^{3L`{#<3>0-qcknaYI@G-=8@je<lRmwTGL(e2RgArw#Y0ZPulw)DFrcCoRXK8
z`~#%Oa-fOCFlw0-fUE2qqOV4`F);0vW!8Poz4kyZu5Oq;<P`F!c?kfr-wAtvZCVgA
z$WE~ixcpEPt#wI?7${|B`{oX|CpP~c2u_2E`H&J>m$`pS8(Qgj#Cd8FVOTifPdW8T
zU$B1-K>WV7-m6p`+HuIH4lI}!B6fAEep}An>dDD%qkX5Ov!OTCY1qHLVe1q(lRKrP
zQXwHU{LwM<fo_$#Lv6+>=3dx1H#1`2(RTUV1f1^}Y0Vu_tWx^gh>?*8Z@7eriMH>|
z8D@e{jyZ)yAub1p(fY2o^KPA--a0w9%>l%m>J*mUhO^Q>STNngB+ba<gh>D^)HoI-
zpTHbyG8hQMxqT>ca}X0Fj~YM**V>1(1F|bETzaEZjA;;erBhu;5tG4JKhn)O7d9o#
zfyGx3uuEi(lZ=%7Ld2ojcemkSqMRM!00tmx@v1kx=}!BewxsM#P7$Y^Q$Y$};^M!z
z6-RbVP0l*S{1kH-vY+fvk{9_aSGMnaq`U)BH^@O5bwdaoz*94eH^9UUy2>eZ3Vr+j
z98E=EkCU+gKh2z@Ye)<6(GJPs@AaKvQShjBYHFf&nzd6$DG_CSIBg&6@JN@CMtlq>
z8w=&s#8S%tS+XucefD6zO)AJrlw4>p`f$OiPh=yMuWt{%pjWv)NKU?R$f-#ZKFAq9
zf#2=J^)X3=BC40>_d2j82XV6gyabiQBaNbO>g{w2xs6LcIf|#nqiwBbCJ7W7a>~h<
zuDs?-`+<6|kX%QZg56R8p(dAfGxwSd*$=igLj2o`38<|RaA5_W^^>;i1)Y>1LAod$
z_Er;JX0~OBVdKv~ZjWVeh@huy6p~UPFv+2Fcbj&oiEc0;=-3+!HVEEV0lsH!Pqo7#
z8M-FL#bXXEcLRdAjGo%WLAY&i>n8J?VZ}0f!fH7+Ix7g-wSCnnH3Oyo=BRnvsi{s*
z^hmpJ`x_o`e(UZ=DX3cBXyqfNUTd<jYZWdX#GXm%eW2mS4Xf7Q>{S}z<RwYM!G;g0
zU(o^f9YAd#nWg1Nf~JSVNjDtg^=60KN=-nQ&N{`ShDzoi3R(Jd5sW>YbTk|eab_Fr
zJCG$LV+n_4k2l`nj!EZP7^Cz#Lw2No5_)!C56xh934U~H0{L_Z>MjsmmhFjxiIghW
z1{mhRVm#2IW+Yn5+}Y9YJr`IrQth#t4TfHGLBXV4wQ9%eJ>uEYpCE04GI8!k3r&4W
zYeU2bp-jb4dc1xTj^VLeWdjgb66@yFhmezbp+pss%Gnh1$do&;_U(Cl@{#8=OJr{_
z8N`_p*%)S=3Qg4LtwU??u@l|5&LCv9vxM^>WRqt>tJgc5JIxX&A$zKR4+cuoMK=X@
zL2Hy==@hvtsgLPK$0p4lLd}!*bp3{RrmhT5s)#x?>zq@c8i*mzO*FDY_Dm-vizcQy
zCGrjNmGnfFO@me;Sz!e~S)Wz6<)?C-6$K_DaOIUNuC!C#_P`B|+=Y~*n{qG14-b>W
zc{|;$u$<tP$?%i|%WcJdF_QzMcBcI<E)vjze9(afx~gk%7mM+BcD7neArPGX{B8%<
z2=MTBJJ<b|zzznd*q`)oq$cpOoq_T@($dR`Sr-nS&-V~?)SKSslr)9UK8Bs?tp3`C
zV!&SEJOwc%#gkoZUuV0GIHl-$p0-l@O8P#BOnIs6xl)lOYuR8TBT5MJ421HB8akoE
zmDWUT<kGyi)W<x_lppSsoDDy`h(e5}6yFi)El<sZzF1Gl0rQzoh#G&6eEyN=?_{hC
zA6VG+|JA?FH+0Uh=wWt!wkAq4hZrPf!w9!hr>2%knra!6Wb9jHau(VI1pAXq3oKFK
z&7?T8=jsiiC(|3)(a8!&Y?4zSi<BO4VU^_zM;4CQk96NwcDan2(U~AJM@mGeJ|`cx
zO?tsS^H16HDX+nuf#k2x|8sybh@~2^vb)HxX+K&o5^ap|5cor2mkHx?b1Dc#evcJy
z4s)~QptT>%0ppl-j_poK^yvED#I|Ys@osc-K@`1HW-(MUPK{=1)#@$lH{5DJ(NI+$
zA{;b$aq~ISwmCJT0;2<y_LHrj5E`AHrx90y=`|>f(Z1PFwf#Icsf7DZSyQyI3%L~6
zG)|vs`{{Oak?xR#j$*FX&B~$lGxZ6P2YjgM$K&2$TK90@m_KMg+decf-R}?0IOW<1
za&8c-w9kI7eQ1=lz7tL{PloNIxXet<l85!@8*XsMiw-UgCzMH(`ecXhyyDI)uCQOI
z<{8umvO)`+@OO4$w^N^lmXuBM93Qh^Y~!DT@FL+61L43lp__@<`lU7s_$yc+r(C2i
zQcj1wm~!PWxAoBy$Kp3{O=@D~sP$v-pntJnX|ss3Be-r3YG8Q)ORfh89+O16XcMwu
zZC%ZH*%4$X2UY}35x(d*I+Mz;EkS8raeGcoD+I2bB%S)?Zxu}<AM;<YF6m%RrT!BR
zELaQ@zxC}mvVcYdq@b{5IYaSczgeA0u0q;ki4}4_q9%qWqq?-ILCaI|68y+-wV{W%
zfoP>f1n&M5qR;Q~#d;AI9ke<t&N8IP7Gl4h%d~1@$LxsoFO)Sc3rg(8vd+#m;l+NZ
z^*crV;7NhSB!=>%dY6>;cdJHXk-nIreL8xnDY7R8Gm17iT!q)m@CXNNzt^eCbRO$e
zR`7GDf}!6sbI>YLBcj66mj3%~VYu{SrjIcjfY9+xm)~iBkV`-w)VXa=p?<7r!H71o
zKkSCm-BdyiUS3X3wEhT;+K1sxK0JTg{-_(o;*~@}S@{i$Xs14YBMY2-jh<ZCZGYUh
z&;%*Qa7>u>f{@4&7oEt^_$S%pjVM@n6!Bp0uz}j2cHQIdK4zu6I?kzyPw7x0yRmpj
z`?K~bJh7om2)jUXi=e^CZ6l@KapoSe5v#$U*Q_TWo(F=$T1t|rS*Mlyq};I>>;9rq
z9h~&>$QJql(u_k=>@U03z+5&lyVU^%vWV=<6D%Fwm0kL;8ZDZGw9Goi<ecVE-=e?$
zKQ%%*5-9-$_lSCfk$VVS4i^^t>(&m^Af5?rkOM2a)x+HoI%R)TGm+x5I1-5QqRzmn
z7hxonoVyQ)>~HI(T&ft#Iizm~wgm0#7z;1#zsuHgfXv;~A97$Z^}-D0L)pXmlM5J%
zdHef%Tk@wECv+mCSvcs_B#mCJ*l_!Y>WX%G-fVr;8w?Z+HSJUL$L$}homdKa8&c!I
zqRZM(m4=>~R6eGETGG9i3QVko<W5OJ)F&##r%B0xEy4b|extW3%OH2jh;T!sgAaA;
z<2Ulqh#Ocr*8Zj8!!|71q&#r|(Ix5H6`z#}p}F44{<Ug^faT6t2-s?*WRr1AnMe-e
z3HOZsTm2?UQOY>-3cBdjBsy^{4Rc8Ud+Qu0Y4|zJ?{#3p+bNFe|7g|FT7_!k80f%a
zFN(?7YyVj-G-8e$mg=N8n9wBB(86T>S9OWQot>#+O8SGpaB31OZ)7K-YwO#8C#Ss~
zQ?&D}vtwb`B;Bz`g9TsMH8JfxtCyq0bM|o!{IYwU=Nuqj`VxUY{?SE*)0y!UL`y0P
zt=?f2G}NaZG2CNw2P`=*MRVb`KjG0_p7j2cKZf!;>2^43G#d-M-~mtDCssRVCPI1S
z%NLi9J2g4V*%88IWsCl#)|Q<t;qMtKnqY4aI5ol;HqP{P#g5n~w|+8z&CNGXML2&F
zwFL<|<g`j$lLDU6DuCxxI+z?bWMJCk!1CB0A?1Lk=~Jsw^37^=#(@=1N}y?C4M_;>
z(;lrwx$TAxP9eUDgcU3|cy6D*WY*h05?d$ry4)eD@6cTo=Uf<y=(BxB)iCc`sjxog
zDc5S`-JtiRpdfnvncZ*H1DPu#gc{;RP3^t=B^R>pC+)MUZy9PW6MT8LY1{D|>R-a_
z>u7uSBk&F%MhB9|-C6tWN1qRFUGi?43OJ7=`b_=RrEgLYg|WwJt2xHiYiEWbeomqa
zg56ofYDzH$sTDvi8ygHZ$6i#O&KcK>FFLS1vKGZ+KGR+0V=rz4MbMCwD&dt5ENQJ)
zlh$Fs32U1aJ5Sl?<__AhWtH=o#B~a%%u)Nih91Mylbp|Z$7a{xaqoS1zU%Gwl5D-X
zKP=6NQ|uBf6WPQO8&Q{^UjtMEZr#Qy55jO5_f`|@N(~19_)BoIeL;4Kv&nhPjeq_6
z8|@2gc##8xUz{Uu6j;z9m{r_NzNo$i{gvOr{Hb)I#EeBvzG(UKYwU~LJcc(=Y&Slg
zUd^0IPQ6}Af_iC`zNBlNIX}?of_R-Dw^JYJr6X4kIdD++rEMw)i13<miXu}hu3ctd
zR;`6367rv%-{Zh?T~YSQvZxfrY-^Z39T%tU%WGJ9k$`+##D7bj*=mx(4MPj4h!d@S
zMfaQ3Eh%_b(|^{fQD?3tt^QY5XFxA~Om~wyutdpH-^;H)#geaTT@UopQSyv8n0O|q
z(<kRI+DmJH=y)$rOFH#*a84IOk#$1*L$uau8($NLFAha&V<ibYV=wClfNh_exVMzs
zFW%?W)Hc|Sm0!Q<;zIf5Ui<1t!(8_A*E>&1m*rnzm%U)L+EOPRSmOMe$sI5hFRvyM
zguv=3>j<#uVzS)gLs%n7>v4NUk>lid8jBEX$O#O!#P+e<X;s}#a*+uFx;wn>m2IDm
z;88gWlMWye3V}@;`#ta^mnZ)2=#5NHI)&`?G+gtby{fgB(SmJ%3ELnr>8U1%akIFh
z2CxrCOZro>_%fOt^H!6Jj&M!{wc%jw)s1e&P2gOj0HElIV#gI*q%3A~29iSSxV@&U
z&qFUY)1pJt(>&$W$1EnQrT$TD+a-H#JJf_wyiF`b9i8DY9Qqrlm6mcNt{@n>UzfvM
zLvB~+S=Q0<+wJv<*>{g_a~=zerA0|yU@V|r+0{hSwU{%wd;DKojZR^8Tb|8qoB92A
zRU2HKI-z(fQkvWy^;)roUEJc%9pTDBWLH004tAA6c!#6`fEva0+;*`z%NDzWr7ZdX
zjGU*YXo`eXWnZ{x%L|rnPI>Vrux?g0Obu34;EY}KXkKR|3L*!dU<E?oBgAihqD!ox
z&M!nF*WbGFwk>vTEy_|4=jNy;Z)c7BF{f1*=7<P1j%|0wRxH^Si&isvd5GB}g6ugb
zq?J}NLg=7ii1x>9WjAUx-L%Fb#~X~?VhRHQc}ad+%aKP%>td3yy&ymJwSo44>iW%5
zdqcN`67i&14xx`{k){Jq{S}O!Wsm$#^hfc)WUY-2Ax(#HU8k+MAkxR%8*?p{%MLz?
zQhZZ#3@JzA6!@CN$KLi~=P|briW|<k)Fy%IEq&jcmMo;K?y`%7{NU8YFH<j&+edya
zGOZq8c-p?UaU(^qWelaQR5&0)>XysZ*J+ztu_;|Tgc*L?zAjniXFr52{~`2mlr@g^
zw{F?AVa;u+FZ}wp%ouyISVc&y1FRh;J$q{9<0!RNIqYkeFLNHFQLYp$Gpt!;{j=+9
z_QVh#cS?26ui9j*Ys5KLp{tZv3LOmHiBq3c9si~=Tho5JYHX`h2u41?x0HoXMUV1O
z9kR9E2bLyt$;5jX=neIac!ba=RgEapA-kbl3>o)>e$gq#N{R|T;nbI<x3^V38g^s$
z6$dy1{hU%(Z|tEo#9w?}PSsugzVjHQp*H}}VC$FQgES7ukt&m+qb4^F3?1|-oN(UU
z4Jb847^kyCbWRKn^)Vp1pGr<%y#3&&CXpzf8NZYX2NG=$A>AUWq8%YbD}1YNZeNBb
zDw!0{(GDyn8jxrIvmeAjbxR`D5Cl%wNjv4lxz#u?jbiqnv0EFj7?%Jk`(Iq$1UN<p
zV<QJD{k0AG>otAu07#b70qOKB#jb74DZKL9>zv0F`_QlAPGE1Tp~xtyy{l`q*;Ft&
z)F@bUtiP4uaBg|w?v}~BZBsjm7SA2FqIBeW7=$9vX%!}*os(m>xx)v!ScK?6+iqJs
z_2RWwUU!vk>DEYo7{X-5@aK9d8jJc|qu`{Z4435^vT3)~FQraPXJiq=(~<30t4qB;
zvaz?#Zm%ii#RyZPMQ0sYyg0~7ppsJlD#6Mf)jG{aA+~sfB?}ER0f~8Q*L9uEvfIR|
z9}GKYC8s`-VC_9}NQ2dSTlb5g$`Z0QSbh;h`hS>v4=BB_;_iP^Nw1KEbifeXfU(?7
z3$i2^kZd8zrkRjOyN_0Dt#-v0%WE(h3@MNV3@Jd0z3ROeTRmB?<MiHp@4ff@Ki`?T
zzkLkv`=0mw&-tJ8mk-NxXMW}W?wvb#`gQOr!Q47FF*|H;t}cqnDwZ?L3G*5Uqg8<|
zeKUpI-;(Pmlk6up_Z)!c;_9ofzuxYxuhDEv)@2Xz2Gd&=y-0+<y|wx#@l+`$k~(%1
znW&H^Sx3MJ8Ae_rFw&Xq8`?bsT2wl>wJGj%@JYJpR9av>+1}Rf36?L5=T63*`X4-#
zd>C(diDIq#>!O45uzh3QpPRvMj{i^$w>&%28jd>n9Cd__kr}f5te1o^#K7<~I+S+g
zq@(qXZ^FcfoK}gYI!-!Bg}>bd+t4D}ge6)<UqgL#Fmb1zDt~Wo*2qwNI6hn(u6O4%
z(?SBQh){&7DQc0-{lZ{J(*`RYBp027iTA)};~sVw-UeTw9|Tg`hQ>{GXvP#pX946c
za@U$$ZKN>+tZVX4(H5zU;?K8@-8&@~RTdJQZkvOtiB!<5gp*x1+8%rD&DS`MxS$+D
z$TElQzDCWk`Qg6PtY_#fHAQfouuW|;8R2H@VE^XrQXVci_!^%K7p^pYEZ1X2^CLt+
z4=G&?K(3vwy-~K1&21D`=(KT))&Yn|rB;G;#3V)5ijB7qxn=1(r;&S+PJEHTI?)IM
z8K*LPv(9&HB9A-xc&;>4Gyo<-B{(q&+LrEvcSJZn`9I!Z#M8);k36-<-d?iChgb<8
zQebX-BiWMMnTGu#SgM}2jH(B1vb~WsJEcBx9C48bm>gmTPmvcGzig_J_mdxe>OD>?
zw;dg#GaIRF)9rO(gBlbY^F%`MH3t$4%F4hCF4|0^rL~;chOvWot{wQ5l%YsTS@gQw
z{cQ@+xbj#!Q7*q7j5^Lq+#CBd*nAalvo)9Pom45|r`R25sA@lYpkCAVpss8L3l%4j
zgHL*2bL%pjts!Ytbzy$DH<*MM*&q@8?ZI3J*c2z6-$_cF`^sy7N4*cu7=ojs-n3JM
zAC~z7pCq#ZkFN}d@2u|>kyX$vH}KoRWY^Y>-enI}*Y+YQXa3ht2c}J{ho|{n^%S|C
zl*Q2n+c_8w2P1qQ_28S@l(ODAVh8i71BxjV7fxOtfvE&UdN&*wwNYZG-R{7ey3(gG
zm<&`?_U^iH7fV^&2BqE|SbpdV3P}%7OxeS2pqAcp*WK%!W{H`LQMQ?j+VTs?y0G`u
zRXWvj(JCmu68sAa(}A>Hl@u{XA!H<`-q5VQHwR~;zuEa2lwK?o``u~#<|IdaW>iLo
zBhao<=;EBT&m)X|OU-DrP)INw2<!5O&uFPG<bc%}S&R0q-IJA3tz)H?0mnIo4kr4g
z8X*!;ad_CbwPnIl*yt8cJFa0{M+en*an#;dPvm=$B{T%=eTorvFlj-I+$m}!jSV9t
z+xy$BQ$YjuQeI3zA`2=xD|!6eyC*91F&%Z#13pJE*~c<@=x~PbsH-K*$y1_ejyPM`
z!H6Q)u3K*3*)2THuXZNM)7rUWDLVLEH%ke3p0y9OF2=!c%A0dQ#UW%rYmO}KuJHu>
zU>mBnGh4RsNwo8NL<60h#bAlKVp9vYXpa;vWiO-;$U>oWUk6Q*TqRS;`mOe%HbiK7
zQ%)-hBz9~LSRwu=%n0`38ln{Ua<6W%%4CA=UbOG(o{;YhpBX;r6Ang`LAVD;FxYqJ
zy64#Gi7ieeUMrKkR;MyP_biZQYTr|zD_4s0T_l4w&70a54t_w7lZJ&Gprmq65RzJg
z9{0Unr*@(u%7j?X9S0MP<300xIivm~^^vk6D3^>1yWU`8jZZKbr^tBuXkDkUEve5J
z*hhfXL>`TON%nIDacO_*kA1A8v2<q1X96tlX?#0&O0KSI5cYjZY1c2g+4)&ZL}~_I
zfpvX<qm5z3&<s26q&CGP^z8@Q)UKPrqC_hmAYm)D(4+Q)HF~L!q!^aYil_<1ND-Fi
zTord}JmiKuIKR^#EtqyxRpH=k9axP(sw^L;T-sxGPnK#CX+Gh=QZ}7Ja_8hcDz~<$
z<Yvk0ld!j#r3^lLBOA3+=J{*e+WfL%v&}Mstz;*dS}u<(UV-O?JCSE1vN4m}+Zc`a
z3EHF;&rUO7{IwnRY*8f!Zk9xifrF{<?Oat_z*sJ_qSDUxw1~lxEAx^A({{IF%^gcu
z+OC}5t8cv4`I#E&5cOoRw(PFAJ|j963f%8t>aB-jEgWOTU{n#yVF%u2d)oLAS&H8P
z_3*d@6O~KNCcbHJzIM&hRnE`ceeYPd!uHjjqI7Mq3btK@DGpvn&b_ujZ@O#E(mR(s
zjTka<7a6jZM|q&z&obLMVh=tc4uz{?(GRdz2yPGC*1@)`>NKZOavcD~OC%#6VCHm)
z=Ap%T;oFnZX~r%&81d)JFTdOlrwnsD8Pd-0Sn5j1I%`Mjky*kcGa?&Ik>qL1IAKS-
ziC_zgLCJ#4dmT&)jxtNI-Wb_qX)RC;!>}q|;nRU-1C>H?#X}F>YsYi!Hco9C-QqN2
z1akT7tzt_DIM(+>!!Ui)mSl&K(s{(elqk3QHanROOx@UwXnX=WrmG?IIZtIb7UtgZ
zLa*NwTajx{*L|Vt2NCJmfyk#qi9O`tOO3#;ve6!IT?=_n@`8qiV#wZMs*b$^Nzo^e
zxSyz3$m5j7+VnX}yk#ezKjPpoV}#%%i&<g1Rx3@GYOD+Cpqqcf&b0T3y(QIFZM-KO
zOs-zSHe?7?K*!leIf#rnhZHH{4MxZqYn8KYc<Q;XhmtnnP+?b=3p}`hgU|K3UJ2}!
zeyv_$DL~#yo|<+MFx3kQ`=PVjg&e{w^PkfUO5734>0+C=U=wopYVn<d%j0b!aua;k
z3&V1DxU6RsgY7yX%`|2C@P9j)q>$xE>OYkHb?G?gvxKA7f#jD=_9-ngZ9iQ7<&{_*
z@}w1DDb~U-*jhH)d@iX?RLte)m71&k=J2_2*J50I?PUW2QcrdwiK>7}y^-J|VjmE9
zoU3Bo;L5CVS>MjuLc$j2h=klc5n{LlD>BzElX>!TNXab;FSMs}aCH(W4pmwu&>Uw+
z7Y{G*hy6(VZcGGW1;Y6YEUreBVo-y3@@+~NeR}b&vcZHisCEfNrQ;4h?o4wXB2y_S
z`|L-%JysJH@-KAQjPHc+GSx?&tKzG&vPD|p)LXgr*zoPgy64M?&e~z0lT~?#H;fE{
zI^RQQ-pE#d+<v^iFvT+ke-Yo?zA(%qxw^t*VL#EXW-hg&x3Qxy!gS)e0>+NoPu6u(
z!pgJfz(mkQpiwizeyUN}U6@QU6nE`N=o+)1PGY+1%2zl)bJ&n)6nFV<kw$#*&(wf~
zz*Q5ck5jmlT;br8j5o+8(YK#%-IBYT>(9@hcVOz)!rU8rW5f1yjgPJD&7yPAEXq(d
zj#MUP>Dy&LU#}jd>YQt6o)~HHL8)0zID*9N7ut#$8tdb09CiSC5q2s$;bDgMi}gzF
zl8V8RBo_qZU`oI4QP#(PsWq|X5Y<Gp$WpNpsXr=4Z@2w&4(07j*E&D*V{f54ko`()
zExGyX8=Y1T*4k-V!0cD^wg(@a@cMb{%De2>+GIq0p{OeaO{f!0-ao2CC5P<SYlK1!
z<r8$uhCW0-Mb&8mU(=drMJd;hYmw)FqwZALjj9C7EK*Mi?=SfLiq*X>J^Rh}-Yui7
z=3%E@bEYs|l~q&ztitMkt8H#i9&vsrCkd%hbb8T#yI`?0m=2wt(_p*cNZ5Q?Z9DCE
z7QYPXN-I#BNT<9z`0m-5{WM1QyA1~Qs45LEHwPB$SyEK+9Vzy#TI~qzX!d*EI4s7B
z420@B_~gI}Lsfmq->)0ubE;#>1|+2K^M+3wWl#r0(;cR0f6xtGmd3mUlMM>=hLH_Z
zLGn70AlKj@)`zjlDrO^&B?gus#-wMal>#%77hpY{8)$#jI!;iosvS9i7_m$!4DG`H
zxJmteeBR0*bRaRa8YAtw+Vw>j`|VHie!{9}D8iDt(7>V-S%n~t(HZ|F*T;7I)9z_m
zmFWU9a(K4AVM@l8$+bTA&sxW+ut!P4tqvfGO_jo#R#<ag`sYccOsHz2(@J8;cpw9&
zCX}(k?)(?kO-XKI{fjw<@qrFrJ2hvRV7~avZk#f1<S3GCgUKo?%0Z^FziMk@3Qrts
zf_8N?%;M6-s+)h^jgp=y-X;A-=MBkMam>Nz%%IcAIJtA-i2Y3`x-aT9$*81NmtN{z
z72{7DCLb0nJN8NK8+ae=ZwsHRO_BnqIwIK0xTSnI@LJ^~W3x*A_`B*yhQ`I->cHZw
z%H$58kn|?*{P*4WDrl%BDp3a5!PJ_=Pao@9J}3!N+QU!SKRn}^dYTf>V8Y%o;sH8E
zOQ$%-s<(e^eG8vpc=a7va>G?K(=+(6{;5$8Wm+laGwA@L^AZS;w$(wj^kMtwoblc?
z{&A;~yo8ZuBT(*z{Y#S;#EpVx73`*sBHqncA>%k#70Kyv*vv=FJ^x<|6FZe=CgdW8
zbAt&PTTvBqsmRnQ^HEFq)@qX^D=*f+wVsi^q*{i!ZyZ=|QR`<Y_h<i}7Nk38oL^FJ
zSnYlLkM`xk>U6rTw$=++n874t$HmgO|E%jY2~`3^oyFV&Mz6sdz%e^x_FrxMaRkY&
z#8UKvQSXJ?Qq;JWf%(6??>>iL4XY2%XpP8O2a^^C$_^>?WXk@ho+&G(RzKhD!17cY
zq8(Iko~(ig<?FCVV~O^3n7Y$w6BZ1Gv&`(19&M~h>W)H0Bsg}l8xFpv>Nsku4o1$3
z2nX$xA5GI$D~)yl(#r`MfH+s_or-8iea}vPO7(+@$7Le!a@2u|m$1K6f8bNwOT$4*
zY9RY>Jk-I&Cm?xa7)&<*X^#dy4{gr2)6TV5Hn%M2bOt-xr#~8#$hqbZP*R3gTFFQH
z<l^HW=bRIJ)}ys?4O5bWVRQgFXb^)G%RLJ_>@yy%IoYBjdW^FJ5qYp!umRd<K3W#u
zIG|6dM@;gR1CB?rp~JBFi_7`59u0a4W5nI7oTi9zVQ%4l_SyAJogw*z(~7UkM4<4{
z=RRYf^JuM%v>pZz2-`J7dtNe4tQGC#l*Vl%E4!k`6avp3{d2p{={&=*19^kVGfS$4
zR-S!cP84Zx3?YqhZp@6?=V#9;yEg0mj=pMn$hoiLkSV2qLA?c<jnqtWuLhGGQ_;Nk
zg+-PypmgZar{u#SFFE*J6-!AhM`oR}FG?y{nA^~s;rw=|t*+g4<Bj&ki6UW+T90fw
zw4-}sFH)xv<`^U-3VU|fJ%w=zf5?I5ln|NwoUU6c;3s^9uA*{>#NyzKSRPN2QBl4*
zU=KZX#bsAH4QZeM_s3b=mnMObb079|n$h>&M6Jl6d?^gLuoe5VZY7J|4p_^%O5!4T
ziDdWm#zu)*A+7hh?a4zk>!D7X-KdDC9YJ{uMbEyxO=rX=A$o+ksbI&3i2zVWQk!lr
z%CBf66@g^I_Jh$YzzE6T!Pi>Aww{}-eP!JfLJ&_=y5K<ix`U5<!i~X(a~w<4SJfh+
z0?Yg<rxjbM!dTk54VVNpMu@fh?W@~c=O2%rQhU>Z#XVGpll%kA$*57pxFGniY4|Gt
ztpkp8=950a=Zr)ozDzVy3`nT8uT3Jo0~T@rD9IjU4n%DF$w$~%AAzuS`Zi8tdtU3e
zgndR`N3jo__Tu2<%O!p)65+u7aeIC@8me@mz8rC2HHTd)5l<7Je?ehdaIa#GQ3{zu
zF!}xTJXnG@*3#`H>R;A9QyzP<5mi%D9W^l7E5<4H+SB&JdYTIBXBmEtrj-78rh^ZM
z2mR$eDaHnU($7rFAzjs8)Q#yuwBfxN#FWP^W&uWwcGb+(*yv{ax?Gz&(8~F(9Yoby
z$L;dgwXArfeZ+y)t0m8c8qQF1o-^ckF(eeMYj#B&DC{_4J(CCk5g&FisnZaLqN7^v
z%6#M4J;Vw+jTnsXlZg0zal2z;d@Dg=Qx2dm5?g~kLX$6Puq<}28TJNCW0L~9j@V1P
zk-ZQB^`RRvgH?jm8(y@hsB^g-<SJ5X^kLGZuIk=%G#ES*zIQMLDI%^A&UV_%l5wtF
z`X=Y6OvKAny@OItY+`nGJw?yNw}#>JpaV;b-OAB1ThbaUHf(SjNg+%_CNQ6|YYLWO
z5Nu~Sl7OX1IyJL?W^%n<+sPK)0Y)0FkhM_I`K`KkTvbY@j+o&<r|a4>V#ibf83q?f
zRkN)))gul*mdE0br~qI21g>u#5SL|Rqa9buWK+HtK31Ce4Q;s~qp2hK&LGe%$cU%a
z!Ph(Fg%BH7PefTaHh#N&W_@qeX_j5|5%!AAxjAJwB}GGDo1EXFUm+8X+s(!3@K`OJ
zbXuh=bU_3mn-HlYY2033y%<pE_)@CvIhb1Ri)0cye~cH=z8#t97%DCIqYf;mXL1HY
zp0!t&#L+Y!SwRNnk~H{7kd$+63eLCaD_&J^mqZ~pGH6jHS<S)cpsb!*&pd?5cgkMf
z#xx`ZEr$%kIO<@u%ipl%I{W%Z!zcW}a_3iiHpt&s^+5Q+Yr3_sR7FW2(yAzp4bc;Q
z&0mfybgl|1P8!2GTUsq8mK}9kF>YB|mnR_Zl9B<P=_`)fEd%4G__t%w8?YQ}olwHn
zs<fI$kMu$=t9wWadJ0Q^oLC4f=?Ci7J*wN1Tk7)qhNLzyoe@s@p|(mLj>uK@&QpA&
zwgn~0#-R~P__7rZADi`n&88SE_*k-1dt@ob(rdSN-=^b^Vx}FK#tI=><9l0MitxrE
z&PQ|r$?5nETwco6wcG2}gO&vGemUlJ7;lG1FOWei6S)>`KzHPNxZ<I=z2jb|(Y_N7
z9<BpkBW+nLeviF2`?qQWL4L|<sXGgETiG9Q)XGQsy2U<I02MnWfFyIBkxDSk%53*3
zoH#Mt?>p@jI(wYGiO+1>Rz3Q3me0YmgKUXtmDz6&BB`Ew!^8o1t5J5LG6Rct985Au
zwnH{wY7rc;yB>Y|3i|qgEh9QN#KzV_P0lH5nTeD9rpWjR2)5N-@9~IC$n*k4{N;89
zyW8ut$7_csrl()$v=m@LHN`d6_a+ih%~4-K!~mGosFhybv$c)!FQ3`s)d(kst!p2N
zaVx6G&r_bakvWU2Tpvsvi_`8dj4C3?>coLVd;!US-Z0po<qQu~QP{QYo+5@Y&xHD{
zP=_=Cqfk5e7@^#plRc%pAqiJIla|ICnnK7??b6E?Hst_y>%!dsX+G?Y)oB^dWQ)=!
z6(Wv<$r<Jol*({?;gr3p`^25vzFSHqp9n_dzG}0*x$Y}-Im_s@0~^>|QVf=Ca?#$>
z=3*69Cx)G-$d8nbhgk~uwnv5WEbXlgmN9)9X7bi{UC5te{f8Y`F0+v5@+eLzbR>!>
zyV$;=eI6~DvlJ^2thyP;Z%W>0gr%drt<6)|S2RGX5V2UC34HWh<rj|GH+Ij-R4Zj=
z!*~hG0wcCRF*{+su7k1<N|zPC>#Wiaz8KLF@l(cem#weY>_vsx<Fw)mbw_)(grK2%
zwY*Y?YNy!4)dw6*?oj==&y137hn)w{`Ed0jr48)}x=U{F+y%kkq6r~t8f7H+&Hs({
ziXJFDP&Od#kg^BG>D1iC4q^QwWt1%jkMQCc6e+T?%!`Y2RSeMmqqd<wPd;&#SE0_O
z1B-Fi0Gzgwx`MEwxPtIS2NrWE&K$$qHXGa6*ycTCYm++a0OHcz)1^3)0E-w)KyiuZ
zoBIzsziS?}d&f80eI1P55-FT~PX{J-dvp`h!ls7DcrDYkL{?x)kYRO<Ivu=C1*c-K
zw6Vs-Zy~aj^-DY6Tl`D;P`0@dR%gzVQA@rtF!6MqYzZB)j<XO+JIi^zo~FJ^fkMsl
z221-zuo#<Yy`WYb$0$1O=!HTg<vrF3r&|(+XJB-aikJ~_3oLmmk+IVw_V&6>gQ_58
zV$vK~h7aOY9vh{y9*l|t_LDi|>nU<lKgT&ScB(PuHEZv28aWhkeoi<PY&t1{!nu8>
zl2-135H`me+e~(+H>m{SDW_R8tvYgwcGI|#;%@in+wWMu-1%9Q!(wBI^3W`;{;lm@
zEAV+bEq!WyJc&RLbR!>I7Sf08dwEhfMh*v`BeRV1xg$2)25&=eit@p1-@^_pxjab@
zGJ_wNEt4sh1{%Se2z<Nk!R+ew974g*{9Jr9vX;K12B@SWVk-}LgGu63IU3cbduQv*
zJ>>NCoaQqYfFzdC-0Y$HsOA=GxiB)y$$kP}%XECCXYX2E*CE|9s=1b-ZdGs!rt4f)
zb}z)59#&@irUvPaPC{%La)4-YFU(Ozxo6(iOBgn^DXP?@(&l8|xai=MpG9TTEm3QE
z*xsFAI5IZF5%qnisYl+)Nsw5-AI@Hnj7~T|a|GdoSjO!=^){%2=&etbF2M&0N-mXH
zD312tHtO(3%y9?=r@g_XoyoX{C}`hY*GaYA&3MQ-2P}qO8q7s_E&nlUWH8#dG$mkV
zUj;a2$^{fJXK~BOE<M@4wM``=5!`}iJWoa=CB>>hWuEx<ZMl+$l<np;;<8{Gd`6Zs
z#L8js>lWP7<xZnsxDWY0;`;3U$=@Ejj8m8S9eBywktjaD-M+ov!x_fIPDAu@^-W9c
zJL;8+Y>-ppBKurwEntFYN!M%%R~j+tCz$H&JG)+}D6g!^Cmh&wf<~xWKQlhD5jvir
z*4gH;fq!6tYEx@h0I@3ZhJ@+h#iG|xiPAn;4;LFqd(U<UmiB<_uD{V9NeT)3f$DV9
z7LudL#)s@f^`@}CGJSa-Z0A#P@FMrsea`5KLap}UZdwWKkt{#yfHX)qVCns?q*G!T
zMrQg>Oa1Ik&2F~u?gl~Ss#C26COl`N_3V49C)#8rre(FPNQl&S?2Gu$#tC5@2}|f<
zrtNzh9qFLRNvD~n3EWB>P>Jj#9ievC)7W0NIhb@q$Q=|pryMWw(XP9v!yT*qHegzW
zvTu<h_OZG`ap}^3l)(=y72o~?xHCx49<uLi-^Fr~P?3k9pVK$apTsEE;So2`!&q9X
zUeSdQ+xOSt-gLe5GuQeC9oRByKhQoY9L(vz4(o$=n9`3!I?-?Y!EWKmn8-Z{-;%q|
zI+*0|{W#3@(QW`F458lfdBBpf5frk-9&5D2CY*~$(xEK!l63ZX!wc4gqYO_9bDzz%
zH^34_*_d#=!DJbg4kc&&wr=5_3QeIIAND#JEnk$%js5-@7O2rl+g?{^D0g$}Q?J0m
z=mYVakPg$g9qskVn=+bH{J?>w^bsa@aUZSWa=Xu8w4H5?dSfHKzSB(hg6N{k5!AY|
zU2W}Md*jVcBiK~>6SM3=+7foxt$cQv`S;ErbYO9HR^LczF7^LC4a0oFq1iV?aSVhv
z49A6qgw`2)ZEt-G&x!B56nuO!4jO!H){-L9d>xXOWZJU15!Yt>8u?Olgp7{zS|x)Z
zbIQTzI$OJulEb7Z?5`UJ{U5mv#Ww2TYs?52VAiQaWC!Z=G#K(4$k89e<P9V9g7oM-
z_CR4|2XmaaazaDjX(S(KB}n_Yhza{p+m{j3D1+9Yb|6XNL^yFu^5Nt@3v<tVzVkad
z<`gUYNOdO!G`g6(Ke2QUCf*~GCQTZt__Q5uw=NwWb{e^vbFi5_&~ROx%g5TA-tq$H
zcRk{4<Wwl{_IQn%wqpf2;$7e)4k-N~He>KX7}B-Vz}QSuC-UV_zIS2AwA0AB;4D3I
z5l8G~@{UcU3i2~4G1%*X1<Ow5q9#9Th6r|A(Y>-*sWzjXZqpN6kTwZ1=tFG7U~0WA
zMjJ>2)6&JDWRKU=;7V9wB^R7?VA+$PJ4x$udlScNPvrbeqS!HXPP4vR56`rMTs&v8
z%Ok3}HSV;MEv}`!{Al0Kw)-FhnT{2nsl{~I!Q?EjT(`o`Ro^%yn~hUzbEgBVQ96i}
z)>*T#E}rk!fh1zFr;11`LIp3@I5as+LW*5zFIwN<;51TXyt6liMFSuA#d>v$62Uhd
zSPW$r=a6kh9ECG)c-h7IgLbKP4FyVJ&)Ow(jesffqYH>u2PHnimyX*HwGVyj-C-&x
z$JK$UyBOO!t-yY`>(9vu3+1l^%U-XY;eaqzwV7`aDyK2fLxl3yecE)w56j!0tbR6O
zXuGsm_O{2pVUn@6Fkaa(7V0?>JWSsiVh5(Vdg!5tE_;bRRihil4xLueD25bfVHZ?&
z*nXsYts;tKh0vBGn@z*OvykW|Fv{n^Zf#HJE08<;!(I)7LN@j6N4o)0iSRUe0LuYc
zMGpTc1q$uQ>ME5L*V~R@r3O|j;$m0_;`DDOvq{B2-Y8@3?Mt181gGn-yU~6kf0EPg
z{FGG0VWU&_ll582iexd9EsnnqW`GrmAsw}!YFD#mRaFnJmg5CXTrI6fYMqLk{B-w`
zaiEH}JO`FL{CzzNto%%CjZ938IIVmJyjX$xD$Vw@9qgDEJ=?<x2d1xDwq~{cTsJEz
zfyiYpcmFX5Q;SwoZ{L2tjqchNYwjVLWUT{HS5`&Xw*5jrbZnyU{1hUVQr~`Y;1$}y
z<XB)dpm8v{+OK1Lf2leuKU6xS4|{`&JIUGn@WL_s<$7EO;z*y#z454nsR6&B95K||
zUG^&-S5Tsg{2Ah9q2WbT&>}wEZ!8l9>~>kr_N#3VE_<2s1yZF+{+kP&tAZ;PyHGVX
zo}$w(el1C*uL8jQuKr0+R^8|W_Umm_brf|}PXnUKx$60^u-~Z1mP{I<C}F~JFdAX1
zbo7aP{>|?7WGHKF*|&Aic5fJo5O~Frf8_BzkD&Bhbx+I$ol%7&NrjgjOe~P(U`@{N
z62a}avx8UPc#ZQ@3t`#PwJYp*+LtW7e>C$DfF!Bj`_8w$!+!VCr=QEU>lv5`OhOw{
zaGnBhBd*lBATLMV8n)kS-yV!&z@_A1l4{m2U2VVLs1&teW@Mw&PULIvUT%Mo-CJ_=
zHP<@7>#MXqq?J};lnikv5xf8J8514Row3i$4xxDSoM^7f9gjG6^gc#fA$+pj{89Fv
zX`Ez}rd8H-xy~YRugFHTQ(~O`aYqm$SV_W199WVPi%_=x^I_z{uWx_SJSoe%05;W@
zo^lXzi(J_Xno{<K!q5)cpO!!!0^@F{sS&2f$JosOtll2xIAAi}0@D>U>!~j=Zhzhe
z^L4$gP9vx{wk$3ude#1-1~os5uod2{te<e)&+m2c(X$x5#oc;-kNsu$=}M*5iaPHN
zrtV=Lq6WYz`>UM7g}FONoZr!Xl@UE?f89Mn)%s*$xa7c+goY>`YJZcu-jt4jLZO%(
zeZT<~6&#Ep<O7J@X+YKvc;5cD+_n;tw9xY)%KWWGD^qm~DdpPYwin>6Qq+#v-=%Zn
zwJV%oTE?qZtg*k(UfsQ((wZ|)D`mh<IPwA5KeSPPZEvH~hz`mM9u`*~t??lIN=Xs+
zkNGU6KhI1#t(2GX1QYcCPwnf722o51QJLJfXe(%V>GqPiq`N1x3$Kv<bA6HO{pbKV
zMUwzaB_!mToaijsHsF=8e`#tpPBWIk29^MzTq==kSix8l`0QWnk%JsODI~oTp<Xyw
z<!(;Eql(N=*}pY2KxU!jDI5TEI8o*&HW1<P?;Tx;va~;(a9}a216igF&&itnbM_xe
zWgCdBbbiOQ$~Y?f7GY5KpY<(i!jY=8EuMVD!PKfyMGDC~d+fhjKNTJ?6+=!kfb}^P
zp%2gKf9F)LM%6`@rd8|y2;GwLN}Nf?zyE1>ayHNXlpvUNAY!{~W^n-6W7UC}pAgm^
zSPi|L?-~MJs@m@ClOC%DavO8xw6gOEXsWmfH?~hMSkg!wH15E%Z!9teib!-LH3*^Z
zQ`++=ce2T874h=6d?Vk`rAOGOwti6rplUBC4|Om(L2#jv4iL=-oJ00$)hi9XirA~@
z4Y2%5SRLj8_BL7KwCR8PV`VRjlJ-c#idO+xEcpzJBYaLefB2aZO!lnmhzMOe)ah7-
zV%bvnPCNKiAD7)lS+qNDxqIDRYu4IlJQlLjFbBV?>b=uWKD26LV%R>j4KsGdtQ5Ax
z8;pQDI<GUmF4||+yEEoqx!8K^nF^*m<lwt$Md^^@nTymudticew8s;vfM5^;Qv(~v
zcH(bK#-O+b>=E`k*>jH9n&4owzSEBDLwaFyj*%83c}b<56~pnl^)1N=$o*W{?*lMa
zoYHMC*ypux=8f`YAvbx0#hOGn*AefZ-+l9T1m`ppOAD0;kcyDr@R51Rb7Wr7yJeWb
z9A#fn?}=IhNx}{+ZF~x5g;${8e_;)f^3OsoWI}VTfXM-Zwp%>pK=Pe^Q8(xkjpbsH
zT^ujHcNl0^{=d-Y6f$C8{8;fM>dRy5oOJ+~Y2)SlBubE;EZ@*##LFNrdUm5zNLgk*
zh5RLV%vj_U0J$nT|8fqj?AvqN_aQvXm!l{G3DSpr5z{4a<pA23w5cB0Ewt|`B#Wdd
z2QMNfK~3kdeQ6CeX&5465t0K7f=pu2exOsIR5$(09xHA~q4=>`j<I(jMTko3Xm+AZ
z%i>w~+}0b!7jdhPuk;FD`|0!FdX;^7{v4lJ@BAztAWK5l=}Fb`D+-fV`EZZ$R-HZw
zCKa2ZjpUs6?JFM(Zn|X2jW;^KYfqQEHBwLPs|qHEjV5FtnoEOedZL^ep^RVMeIf2~
zec@9N^MsVUv7vdxKeGmm3cJVj+Sk;3w25fn-j9x5Fru_LOLamq^27GE?K+}A6bZYB
z{s7C}MW+@Kls|0GYmho9z3)JBV`Fb=wMU+xbKIN4OWk)`xtCG|Z*;?kDSJVE5)+IC
zCIcZ92qJ3?c{C2bcE9uJRZ1V*W|y_sL^%La_BIC=(<IB3R{1tG{QT$_Had<&;GAYY
zFzgiwK}gmwYI98Fgyisy?WBO8gGn~K<fa?0wy$f;US0w@Q>GkHth-Fth~P~Pm{fGs
z4+YnD)>gZ`zO~j`2>sIWWb$XFI{4HqW;U<4D_WzsspmAb9gmK0oUkifmo^e4<g}8U
z=(?N+oNiz*E|?aLiiHzH=D=cnm`WLqvM4<SaK>KJ21H>a^qqs;9at?pGI@#{Mv$1j
zw2@%wvCXs|9%4Zpe7*+{nG)VqR3_9syDHzaFvoct)JqJJ1y~Z*(&>!~VjHrT<=4@V
z!}B|wRuNP<_b_{%u&djAu8gujG~<$tei!?D+?G@)VeoR$J@v3wOIkfxVW+&|Q*sqY
zl5`!ajf7&mrgdwYu(N5UmH>&*%FIdG=kVjmdu-RXE-oD5Q`v+V1Ynx3sF!?Q;YHvI
zct%!sV6`K}q#?gR+DwK5D&@$(zPl$JG;CJK99Z1bxQ!yRWK%C)!EUH~^aknS7_1#w
zuC^6JGZO3U#x`~oxWph#Grnqo2y3!#>b`d|1VfwVUI!yCxfbi`&GpJX3kvR!uy}7U
zsk9;6Vy1t2c0*3v>G0eF$nA5j4&ucb)3;Y-&p1D6Vgyl&rq&(bag-p?S2mJgF^iFd
zR`pstQ5tCZ0<L>iZEWfH2{%~DreTp_Wh0?FSH(!`d!z&^Bq_DAWEih*vs1C)SzJ22
z4l%pnW6A$-Lx*DFODer+U*Ci|qS><aD0J^YLr_dcki9K?0sDt^a)K@FHT9iZ#!}Q|
z)XZHGjJ%avtjGmHSzByr8y5JXDlR`B`G-=lz-TbnP=;)5%x>v+f8`y8TBbn=`77*V
z!2!4`21zlpk`a}v!S<^SM{ZeL)(sLtdvut=q$9aKD_Xh3hi)cIltkwz!jpdruCu(6
zz&fHhsSbw-1E^SsqT2WIy(vc>vK7e-Hz52@IgR9a5go+DVTHK0E!h48iiScM0Fu(n
z1~j1|yRGO6-wFm=TIucOsMnEkQfJ}hc6(ba>!LR8kOPPpF}?_$QEbBQXfF@;r?vLD
zcQEEoEDwfnuWb~&MukIZ6_3)+A~Ao5y{=%;FZ3rZXpaL6s#j3q0bAL2fa{lB?=+HV
z8EI4zSeje|i+jte_DK;z5;<0r4lw1AVJ$mCU>7NhNh1hP#eLZAE;yLndY0l-zN;=v
z$uR#$4HgF@#s(Qp*y=XI_wn#;PBXqy3e7pbhcmS7_3axO^(d#!>I0@X-80(XI%R9R
z*N)6Mjd()E46*W0+uG`i?20OqMW1W41e0n~Z)8OE2nfF-1aRFVS;n_1j7i+Od}R(k
z_wx-ss#)IM5qZaZ8Pq(@fxiwuxjuGrjg+#ABFxS1Nn&Z@|D*$mOT&vr7bT#DmBHT7
zVU)SA#f)=YMo9+~A53In<fY1``Nl@cm>SU9`@;?-nfOx75jw@?O>I`G5iyQJMLQo;
zn=PYcZ*Dz@CKcV8h26yrz|<%!=|-ad6ZV!iYH6bl9#nNES!8D%e0`isCy8Q|9?mD&
zy*ZxLNSr3!k5&<Bo+qZLa?Kg^AW72RnpabfU=+_et=#<A6YqpbIc48a_YY)U#em7*
zfkDI@zFa$@Pvw9o=YCuF)e%n|fe(_)_c)lO@4N3<XWv*1T=9eQNMdG)MXB_ugKXkr
ztJ)G&Dv4|3)06$?i{W#8danjji#3E~@=gX}`e@=4ZGBy*tzA>Wusg6As7x~3MX1~}
zjf6vWk1DCk)5wXB4lEaQ1kP`^;Wm*l5QG33CmdKyx`3&T4>6|w2AUu}hY+mEo&(A!
zjVUo>+&0vc@S<U?<`~k2vkpw#{7u*1U?bfSNP?-I5{>ShH;kkr`2+H7vm;n-Hs(xI
z<_I&Q(~3cCx{7=RvOAx((V7Zrc7kc`1!b{>^y$TW!%oPxLAeZv1l8aV_vPXkrYKp6
zlnx+A78L~VEgOCwyH^Hz+msz!js`O{?X+rkpHP<|)sT&u-LdXv=|wy>zu$r7XRwb+
z9bfd}Y|f`H%-uu8nA55SsY0k})flgS)!8YKrkuK1Bl$#RYzLq0Y$HXAh&#okG_~0#
z8jsx)>x*WN03)qRH~h6NU8mxBO1#CACC|?9axkqMp^m7e{M!r0hR>dcvIH9f7N3>I
zbw(dOH^$v2yG}2Lb<0Emkx2G+@VUuiEXEf)l@I4gPU>m)@7JMe4y?v(JJc`kCwGS;
z38%Z)#@NF#F__5F;b78^hFmG4ILl0?sMO*x2z|Gi|ApoycT3g=Qy#}+Wpu8MUgT2-
zmH7MHT39u4zXW~S$tbS7W{GXhX@wTgz&UB<vO&Taf)78?K9?!$6Xq;R2Nqoqu_Qh>
z^ClEGtRj2Kr)Fyw=tu-ik^<$xig3xa45Cm{g4Td|-(GvLaZA-h8df1zAT@8Kl*PML
zT$HQ~dq+JhHybH#N|M_jhJGA}4qgckSG~yI*$8)RVq)59CB?816xxE3&mPJFQ4-0p
z(})Ywei)}K+Peyt$_R#gOhUkt4`H7c%Yn&!Q}tE`I9Yfzc0othfGIj9b_zzu73R;l
zy772bl53a<)?ieVW5|2;?lwPnQg$2Cqg~@8Eh{wuG2vbIaQ7LDu3cafZ5HOPe3iYY
z@VZ**Uysipc3|-|&1`0+%f5$W$ljY{NI<j*h*o_~e&^AJ1NO~D6<wHBECu;|WPw(Z
zT$qe=RqdWxS~QOX(N#$`-;z&WL(Fj){Bh?xu-rp?oLB%|j@Y-hp}#d_XohATP<5Sz
zVF*E$?K0Zt+ls*D%RAt-a==2k)h?&2sRd*2E6kvcR6e7F4kkW#FQ=E(@!j_RCb2LL
z6qp%z0P$d%jkPe9Z4_B5_U&!7BA7;M$Uz5|TxBcKx6}3=-5aPT21n;OCvPwert~pP
zdG?(hvyvk@#cFYMomhT^uLCbRio-JT{<41T1J$cC7R%7g<Bt&~2u4yOn+ghoeX#X0
zRT+eXgAOd_s%h;h)Ty*-M)W^Yc$tOH&;ga!99YfX@aely2W7ET*`1&-`_SS$Wr2fj
z?sQ-=iYz{9_;3YL^+`hEKitMQHWc}E6dEX23MQsUkvMP?j6cWiyV~^JhJ1<Y;{bxO
zLC|r*cQ+n*+uGZ`8WWo^4chnA<KVL`a?~448o!bVF$~-Hc5F~pCtyPm@h8Tib5#tY
zEO@+IBAy0;p-^<=ar;Qy2a<I%{Qs+`aFsq$^)jdX6lc*<ogb|M$t_K+S8@E3xxpk8
zpPU&Pfr)>tu1te`p1~7|XbwhXI4YCH&T0F;ZbsyvjiV|YSkiS^<})-S%1PPxw;{ac
zj&)8WS%NZ<83$xF`+>%ISN6xpomP@OVT!mRf3OBdyihxbBEYg>aco@<J{~9$M9R6`
zwvV<CqaY(Om`*#c(d#b7?r_i^YcNcL&Bxqe9Iz~3Zb(&gHSprql*f&~ZR_q5UY)Cw
z$#0pMqLw&^_Sv?_>RcnbB>#!5uyhiwf6r6WSj#WLOiS8Q>?n661Jju`*nVbhdrF=+
z5>dg=P&X%gBetXaW`*JgCv{*FQaAPQv7LpM1TPzfj`;+Zqk7-yMx6D#n({i;bAHF@
zsFaoJEpWb{?e5mJ3S~*)*HImp=xXe$4!*b;vQHgWkgI)9VN!C<!~J3%JDBXkddg%@
z+1_pfNcm!Xl=%isoAIV0+m}Q$8Kn$p)cld7uf>7OYk#Av)%_u-k*e&yijF&wA4i?~
zap$*QI<rN`;tbotL@l}IrmLObML9YRLVke#cBt_x=uFz3yeGLFWDGvw;6tXVGFnKT
zk#djHlLnvg<=f#J18M4V0LEWDIlH}Kl)9$_?$g>zY>pji6f`k3G_xhnEbBmP`|EW=
z9scK|`Jq$k_@q_ias(2SL*Td_>&72BfOq58)8F}%4klQP_@68udsr_i1TsJ9AIICO
zfpI9LfOaa|%a*RbV;$v{Pn4+i0_Rtb_vV=m8|-AeaArJm5FJ4DOT`GZZL;y$se(zS
zR>0Z$`8^ISGsmcQv~glM?47Kf(>Z5Tvt(0jo^qN&YwD%K8|4E^Dd6$8U~mNYN4L;t
zAX+8UqnK`<XcV$#W8Z0Hk5|qh7@Vm`N*{^NWOOSAEHE`lPk_-v*J)%F;$sjgMP+9j
z5wBghbj>=aSzmexvqda5J6DfjjzbyFGj(Kr7fU_l;8VSs>=TMTJTo*ZHhsQP9=oob
zATl+I+f?y62VeX(4UbZnB0k10<h-qE1!-xg*gnNc78moo#W>;kw6ePjZshCE&+oQN
z?ef)k-03vp0zElv^j)wY8W?~12F``)M#+g_;%fSdsUW-#_QM4W2R`4y@p9f^yI2GA
zKFBIK-{y|o7y?ew7l0HGl^0n-sP<&{>L=nxZ8wpr>0t8J)F@CzNzU4|g`AsZdLm__
zX?9cD)enkD(4?NHl4IUN6be6Ma0mD02-vkB$?+Rw@sTS;OG{$u*a#8x(;GM2(`{Q8
zjcZM@4pW8_x!}8I9xpmq6}Q}p>;s>agu}kG-F`ItFf%md{H(c&QY;d%?8jQaN{s1h
z5jvQ-TK)#<={Ric$2$^A%Z<7ug-5|5GIxQ`{bgE-sEEw=6K%5afhmESQrQkHI$OI3
zlO+{QVfjB<&lfw9m4`rD{>pNmUUKjQ3ZHD_a^Mc0VUO$Hekwot`WvqE`guFN<)<6-
zne4Oek9*f%OYtli-_LZTqxp_t6$e(crc^K#6eJ3%DF^Lm^Buh{6V6X@ge5mGx!!)R
zk@|JlUh6dKJC8F?x;w&U*#v&ReP<{nd|L!=b!8~M;nQaEi2~~{bp1Li6JdzTIRjv7
zgyhW1T$kKd^LJWFe=+HJc#<kjXi+qib@eFUZoibsCl$)F^OauB1#)(k?)b~~>eMMy
zFO*z;*cUMpK?Y8(wqMD?+Csqze#V%jg|Ex>SvX?9+9-uBFY5t5?hPjDP;3lap#56c
z0p&@`qY?oQIL*O)n2A`J`~Ta2_Ukp@P%QkVgq`#N!!AxZ_+(6xl^BBAZ`5$?V;`4y
zpQAM#SkBYB+gH5FezT@JCZ31N*h89iI)WkL5Cxs9a!C>n)ElN4u>DqpU`LwnleywR
zavz4}P_t{we!ClE+2fdMgrcA#2U9#mj=EHcey4*;q=8jqy=$<n3nA{Y$9}g)A$47K
zPlDf~7YlqmTbZa5DR9)<@6|onmw0w0@Wo1!MUWsj2d~Aq;<j7Y*zYIdj#0yZlFDN=
zCF;#?us`UCPWru6@=W61<zTdvjVi9l{;)0!1u!_Ca?ubc9E=8@f{tt>_D9uE`W_*q
zWal`rpzAcPN(7OL$;NDdTvw;-R80Gf11nuj&S+Wa^rqY~B`Ez#*WrN#lc?wgCN(PM
zJ0fQPX+2$Huu=kb+77VXp0B*}E%s+^|3EENDEuWPY$fr_rdt~ujGL>f`!HoXegc-8
zEcwst1xi1Xbxq-yz%(3B&hgCc_7?*zM?wr0)_ArKvy&<sp(W5RSH&B&yUP(OU%KS^
zqxP5iN%u$iECyZL)t4l`jAcBFN8_)mFJk#peh(ur1526zo*wby_SbcD>Z005RQm`s
zl!GbGScdiLhDKgHP;S|O)6JG5yJXSc<-oK&MyXs%GB|lTf7{l}s)<R2=?MoAUDn4-
zedw|Z?C+YVknr9(K3K7%0P}LxCrW9-6m%@gehxSJ`yz~7)0v&qLWh+O2R?K(ZCL-%
zNP%SDEm+`a=d`b8-~Go%7ZHIXvGJ%kn8amSdXHkp`=`1t1M=lKI^zwNm0h%N9kqYX
zhc8`s_nH+>BZhMD9FP!R`Sjy#NRm&E*}t?dKsl19592ZcCdyfH$DQ}ECI4%iyElvu
zPj7UZvHtPrv3E%k-(mmOU^<zK2NSa8z;eo8-y^We{=JO@hGGOgryZj!>Qi=vqGf{@
z+JAJ+B<ru_5fzalgAs10gO{h8@f_MT0e9pqw5|4^ja@$ZLA(|ppe_TDBG8MHe<JK!
z?#d_Zzp`i4Y-d{~M!5BCh=gkbNB_IMpTjD}D5bDhwj2y{w}aO<(8C6@hyR^IQ~94-
zY8u#WPD|5I^!#mRbIC&n{L=8ph$@Ob1$^`}xy!&{*eA^u562Fv&_5mZ)WKj1N;QI~
z*(cX^`-pFpT~B6tV9B|-Yq1uKWDV#E<JhOn#oXv@Ak|x<SuFij*xosO$v$;1SlBX-
zIFc*bY1e$od?ntbSYj&Q*r&|}2fyke=a-hCiZO&a);@hMS&;k{r<_*qN;1QRJ?o4;
ztHIb9<;Lb<EC<#J>`nK|z?Cp2*k^QI(HDiGSWFDhBp4Cy!rYUOVE*~cZtUZ*uCU-r
z8f|fO;Kik(mSa%OPW!AJm4&(0lMIa0%!j>pX1HgconQIJdz@c2T4|@I1g(9}T<za0
zs5v#s!GaD%L@?7|KW3j>cW9AH6yHgrSA(T^Ec!oepI6r@+A|Bz9(G_w##!<~`H#u-
zKfm<@X(D})dmLB{-C)g7ZPBuqLNfLR`N^EzFgi9g@qp8gZy{|(8VStin0;Z+;=){<
zeuYA|2vFr3)UhvWBQr^QxzkFH#+h^W#q9-CD&;lhwBrSmYtaJa<FjWMY&X_XWO#Xm
z9au1{!F5vX6>cZZLU%r=?o>xr9$$Xr2Ez^?0m00#eMx=-{8s4-P&j~E5J%;`U}utw
z&~4lHr2`}<sY5}6-4QfsxZw2}3v<&GGeaAPr8n7^&4s1y#+z^U`q|six93(D(~A*V
zOx|Ewico^(G5hjb8Wi3_5TljzpduK&`5Y#OuPB&mACTEfIOIVG7Sy0(i<v*F-Cmhc
zUs+#}@%u#Pus#O@FC4sxeRP^iEF9osUscyP>qKUdEBKix2jXVIM~@_jVFAFVe$c);
zX_}mmzVlnFgXO2_!mr8JsH|&KPAi#O$Z(k$9m1A<Z9O#=Y&L#mjs{CBTR83Qc?0uN
zd0SaPk{wDWYw$^P*Ij#~J-_Q}RD(tfIqtx6@e)WeZ7--RQmm1)V6OwyBZ!U~nzqa8
zeFLU;&A%9?Y*!AxR$)YXv!WDvqj7&>yEEz@%SbO9VA{Kgj&|^Ir#uj{C96VQ?sG5d
zx~)nQ#L%#}0Sh{XQcy!%(8W3KQP+>y*R{7nCkipknCHdN4DkBwwb*vi0xz$357-xZ
zCR?C)7#Rd=_mu#4&aUWsAYWT5_Ys|?ZJi#1FA5})fF2wTd)8T0zAL*iN^5k~U5lhv
zF!5DcmXe*kQwdX?dvRkLTPPegwb82$6F>XphhWSvY20-@jtM7}4T!AVcv0=8?Pl%g
zoBB>W2sd8?+hf1Gs!_w#=9v+v^(<~yN;9NvFKhDv;b>L~-tz_%Q&C>frTGJPb$vy0
zhm;GTk>C`p9&dQKcSHlIAYmF7k-DCGk1fd^<d!!$zhVrN!G$4<_bN&2n!-pV?&^56
zigr1e6e?c%YP)u^Cvfe2ih3oAiTbnM7g3i>IzFP|bp=xug?3y7up9_#RE$1g*XJu<
z&OhgOE`JQixOU<ZHzaIf?$(JJbW%9$MhBJ~PUwlre8k`)i;mac*fx_b)Mg=f+5r?_
zxfJ5acET2J%GrFdPifh10P;*1CCjCG^X4`Hm%YaMT@TUHfrT=>FRwe<9K%Ma^EgDZ
z`y5Pti-P+Uw0PcLQTvT_fy`~eX@rCziTm+zf9Q-#u8KQF0Mgqac6()`H&y;o=xsLw
zsss}3iFauEVn=)GVS7~@fmQPRL%RTK1*x8gxL(t-UOo5p)et5Rk&u#f7JEHBS;~*R
zq33G8<%dwcASzHh_qA%DB;-c9o(%o_`qqEM3(1gdc@AuV_%v5iYp~DQYjTw^+uM<)
zXw?|+d+OoXrO(*Xxu;)B2menMCqDt_+!PO@EQn+bBI$P7EjfsYUg!AVC8wRXa4LRo
zp2-u{rfgZe6Dzb#j3`82Fhb_<?-9WrMP8RDX>8z}Y3KLA=Seq{4vO!^R&-22$zrtP
zx-<na*3#}I>)WllDzM3i*^Op(U-GrK!y|TEUPAP6pMiFo;<_YrspR7AMG#^=3hYt1
z|KQ0L;M0yXN<1UEQg+8Px*P^!Z4nDQ98A1BY-&78m5N?lS94u>CS)uT>ptjUVoR1a
z<e<H-?#%!TBrUeF22<B4sKV!-wv}yIh@>2ITJ^@mXbyVr<F;y%-9XiXu@>Zfi;d%$
zgU=6I-oIZZ&Fs#(GL)^QRv!iXFkS-6QN*Ph8JqJv?XGqindbycO|X~%rQzhxUL2{p
zy0xSJ)U4Bv$H+MK?&!)cdwt_#9K|=Z(P_pdvf{~ULh!h)DOgC6^qpK9?_gXZ;?icm
zwv8>DzH&5_MHf*M-Y^hK5OHMVNTaZINqn#7pYuE4En$uB!3XW`Had5$S-xV;iscSK
zd)HWR$nI&M$HCJZXVyFIAW8|Sx?g2T#m^`Ko6?F>>&Bf%OtiMRllI0$4H0iT<usD4
z;oFO)2UL>cP5Igtcdl@Litn$G=mWvNxxE$oQe@=;sS66ca4@xTD_lw#EhdRTh$Haa
zw`Bis>2K`aKgyz_Ei>kDA0rB0*P*`M+rD0viAR}X0J-N7Wx;umZ*4OZqM~edXB}8x
z(tlu75wPtNh;|<@CX3#IC7ryEqk-_6y{&yVCX_5V!EyGHEe9s1MAE!`{u<10tnt%1
zk>{LN4LZl#q*B3BBhkuwb5G~tHFs3aqOAMvz_k}EO5WDDxj^#U=d|)|Xj@OvS7qxC
z6)aLG8CW(P2bKeia(}*Y^pNRf#yx1mIj0nf#zab^C|a{1syb%<Hm)0c7@~2M04RQz
z3a)}FrSt8!4c+qnxWpw$oAbxK;hQuftDSU46*bVdFw&-Dtam?iOf!0;ol{m(_>gUE
zJY={(Gwrm}nB6BBTP7A8tuDyIrF~Q}`U;u8$PiKd)f>JN5LuDh+PVtzc3-_yO2l#E
zCdk+F1{0^0N{0c?Hr0Dn8C`{zFGhxjjH#h;ICxFm`cX91v0M+sllMD6^+^P2DsFxx
zA_O+KPg*&_ildbzxrX}pvo_v17m1er$%*v@3<0{5XYg?s6u2HKyxO}X7lp-~=4$O>
zxrsbs6FJAvfB9S9`Vyz1B~3XEL<HN?jhBu-8&KntRJ`FK-+F;Km{Kib67B6J-DECW
z*-oBXj5d*6wB5NX>>oPx<3mjJWW80pgLHLOc#!30cfA#Sk(f^6*8Ehpdz-5FNn}wN
zo75w1tg48y=&R+b9K;R%X-?jouGi@%CDf(Rs0ZlxDF<IZk*Ksps@6=Afx@xF)j<Rp
zFmcljz5DI{>R%DEgm&Jqa9%J$vwD@(ucH{mw`Ok;WhuHb;xtR6yzXXupc{!UUY76e
zPCRq)F%p_!<(zcoDQvd8GfKN?_hDuheWTzzW}|vc;-z{D!K{VBTf@gm-BZn+Z5YoR
zCT&;x1{AZ{d*0CvNQTK#XVZbj=CFtPzGUyLD^d{&oyZqqX;NaZvWME-tfqSFxYLXQ
zld7!Fh-<#9dz8))QrAy8u#i}_A3xKpARYG^jw6Oa^=D&E*f+HWlx04LDF+5f6F|j6
zc)qvU*xDO$8d|kWuD<prdv}em#s|8LQA(LRLJ1wb$YF-UwGU^17v}P~R1$P%9ay^X
z#zu+IB1U=2-c#eA3Z1N|r;FYjM*ChIBLid1_4VF16jU-$Q3m*x11n9407%w=>IBF$
z%<!Q^?zeAF_3!re&aa{YrB7##Xc?mp+qXOelfgyI_fzk2Fu54EA}OG*4coW2Yh|?K
zDLRoc7`B6n$?HU0=J(sWuZ%<kR)DOa(5-{1wUJ5oIw0tM-M5GbC`6T1ALvdA)||WG
z;0MS<!5w?;{q=^noF%(JrFbg9#MOA(q^`~%Bt`VwTYn;CvuRDouyKPC9bl#+9`!pK
z9jslk+`GoI*>U^MHXzGq$H{La?HEW5YpYrm`mjFHKG4`9JFI*SI}n`H0;R3!SWP!q
z#RTdD*t^leV5d;NTu%&*eXwo0WcGyS2OwTZjXuR{ve`=9m#N^9T7xJa_$6uObp!%Q
zzDk-niaGW}bCq;AHB0D=vLV??0G7~|%!f&(H~8*j<9otBoJ*Bc(5E)`hn;4Pu5{r&
z()8@Ry7|Eoi0*a<Sq@l=Oe};^jJV%j*T^7nghw+p4lJ4A-PD5mp2iY}N$j3<n$@jE
z&Zd+8l=vXq?DuvwqtJV;Ohr^*>|{ysdN$E`y&3yRy<E}`=^&~P;lQ+XCmx)zj}}@R
zfy|g>VmdJO9|duAVwkMxk2L~VJwc|M@+yF&9>M{hQ>wZ7efh+@m%YyG=lb4y`~Ef^
zY)$gA>R4vY`9TL$OE|L!J}y4aexSM)%2eEQz=6f&Le}i;dXB6xI-KMO+kn-ckL1d*
zNQ4We#2bEJu(FI63WbaDP*tQy+xWw!NRXu|zTO|_Afgd_tX`XXm|h#sa`+IKpa*dT
z&fk_yZ*#Q(283W%wL&&S7h*|B9^k6Dlc%8_a)K+IU|YfzNsNM##eWK>y_MO|p!Mzb
zvDp(@N-pr3i$w##*W#3;PpOb-R_KsBZAVVh$mp2!Q;dFk6p4Okf*+XO?EFls+|YZ#
zcC}F%=}YXT8P83HP~Ti><F>mSA?Z@WloQ0k#CNLyDlQ_=>7If~Jl7N0&b(mQ31^@>
zj%FP0+unL=i0NEM0o{CpgGmm(2}@T<z5D7~Nv*1^lC`D8&B0`{Eyqd`ATShGjQx##
z*Aw?9k)!~M&hJGL^8+==;ko63D!?iStb73Yg&e%bk5rf;JJ?<!2gu|!n>xG#V0^g}
z8Q}}zbUKuDh0p8*^E;eYV}ijdHCvn(v7i{9T?;$xa6Lg^L-w;gI1WskD#bD;D-Xkt
zw80ocF(e9e%7K)T0B`*C##uXBPf|d4RM15a045f&jQFNwxm`{@;QWdy4htz?x>&vB
zcs&B5V`GV9tG&TOfn|a?G}2DAC%~dLR>u_g8li9vlLsCvbK!Vm>|}QjLKF8WSRk(Q
zhN;tD7=HzwNrXGqCL94;UVtonPMnp43HMgHA0@o;OfhWvP@7nBy6%U#kcd-^4NI_t
zNs{FY6#Gh~vOV5?{e@Ue*aE!4G{Fka*|u=To~WntB#auh)yCI@$%VC*;`wKqXvD$q
z#LIIwwgXCXQ7#U5syOJ`w*HjjywhpMCFvkn6`~_{u3)=RuO;756V`#{B!;w-`dP|S
z=i5s#lG8opuagce_ZnGcn8Zv)4&4IQ__$q2@>)*dED8K_@BtC&BRx*q#qM1y#3hdO
z1f6F)cQ84OD$>a?A~oBkdRGQ*%7G)Zz490(P3>~<(Q|PS1$mQqs%K*T_@R0llOHZJ
zIaYv)ORrq9YQ>%F?1#IaoR7$2X|zdrVA9_iW{Ip7fU|{RhLb@YdfQQ(ufApRn&kBv
zOr6CW!6(mAR4Pejvqbx}aSj7O5L%po>|k=TXU4}+b=yuSxm_QiFfpCUCy8f|H+<9;
z0#BJg@EFtM+OzgliR9NTxyJcjRHFF#9Mm5vm}Gtl2-qRKgN0(2g3{C7y(;;o6>-sl
z6-$?L6<N|?_AuZ-+F`WbBaABc%54rN`asd*+>$VC{#d~@l-hELWOiUF{}Yv^BL%R2
z{dhf8<;_%u^&E4F`<#m#rbT`s25J#39I~INNAG|q?ji&eL$3o9@4Eh`8|^1^jjTuh
z^@T%DGdkB;^+Bw0Kh<XA&b!t)jUe|ld1%b!hUmm?`{_2RFl?#UiX;z?<Y3~_Drj?9
zL_WXYex|$nAbbI$VOEmpz+6Tb`N;0DEbxm7rQ(c=3L`7ies*!Jr7FZzWMZI>ZE|wT
zr670dXyTu%OBoxCP742SRNgS!RZ$tA(%HUIXUBfN?n&E5N(~Xo;SG~PaEhpr?!o%8
zU&t}*A;ir(jU3d_DrK@bX}{RP<hc!_xi?tot9&^;(0-||!`CbkiC0{3V4A+W?p$lX
zoNIVZzbCniRtklzB@MfDHdzp4!NOE#ztRo2wg_NcmBS4tr&V%JEG*V_CO5KO9JgO>
z#E7e*YU&h-sGEq*K?<;6>!uhzBe=s+UxX>DLRQnG_Um<>gvTc!_tOj89avgiMhILV
zv)`yoBvzlJLr1V!0!tRMrcbFW8J~W$x}7y5lG?N*X??2V0Qg)(Q4mte?DkuAi#9V3
z-;BRvDjE+tB*EuPrtk*b--A#6b~2D#FkU*p#WPps=?%ZLNO<u|MP*9^xa43uV%NFu
zirf6%CcX5>@1J#AwbCLTi7Eg`?f3GE&178;^_^CcO@_cxgu{NnU@Bgtm~sVA0?UzG
zH$lpc{XtzK^T1Jf(tegTuvDn<R7?!d+8?&AU`*TZv@-fmCEOthna;%<6&~Ap#_f-i
z>V`JXIzNl;$-a}@bTDT7<Mwh+EKtCwtUSPSheUP48WhYzS5skwUF*F4N#3(CH$*rR
zWg;bl0n3TIheBh@37oP&tw--bs@f?=h86^@I&vX=ENQYuR<S?JNkjeZJHKnFWh~Sd
zea!y+889vSGY%}LerRGdOqX~5MPq%-SKQ__;$=-&I1y%R7{C2xzHDlux5fFL!sFfJ
zqv109s~l?tNo9=Ciif0^t%>HyEcVw8qb!T)rRwKIplRzX3zo%+q2iMLO~Rz?+{Y;}
z%HMNf@oYIyB?_msEoaf+7B_`oL#Pol%na+CddY0tHH9Jei*&g*Mhh6SXB-@Fe^;=`
zjfty|I<Qg<w!i`=CL=rJ?`y6}Y(z5nguf0xhnQ#vjzMBQ{zE>Ra`dxf6GO@*n{r^f
zYM2spkc50Q|JVj6q$~|fGWu-}CKR!am}&p1>t@zMQ>a6O;ZIjki|k+a&)s#>EyU3;
zIxwy5g*gIJHrv0{`(#g(=Rky$oDxh)Lpo+!_i7(G%2HFW#B~+7`LB(Br}}94QAz=j
zmg$`n(|68L$LZhN6-%cH23e0S-2ug;d1yFF&j%CL3y>7pl%nRe{d-%>vMo@Pk9Luz
z;$t#WEC$H_qdrpvD;KJcHf=EN+~HP*?fhq5Cn+UXK8v=&l>bHaR%`<Q)!w|hkFSSl
z)OH{KeR=%#;s0GQ)*xIY6mADrltNwV{>F(h1^oO^QA!?Rtk`?`02O$&mj~@~@Ii)B
zmE>We%#qo)qMps8B>d^)f#@pSAv(ie2Tu0wleUHQxKSs0IgNNsZrf7!kJ%@0t2I~~
z<fW%1p*@a&2~2LF17f^pqq#OerJg2}(J6K%9kT^2+1sKNM0YcFmQl7(-B#>*f*Q>C
zJI&~jV)?a5sjOw6)*cJj%%ai<9hf%xJ65f>Pv6$K_b5T%eW#u4JZeuVFkOBc5%#m%
zy9f^MO*zf#noj$|Qh)^?<{*U)JLoa{jCzoyom>a94Qe%jQL4k0FSgIjek{Fnsq<4y
zZKVeJvvQD!I2v}?X+(!K$XW-A(-2G9YoFZ>e}<VUsS{XStuCuqa)ZX&vd?Mb!0vE~
zJ}ULF!6d<`h?9=-{@l8bG;Z`T$-HMASZ;BOts2^7pVtkNHnptiB-VMr!NjN_!iE-y
zcbGoQu@hEi`}{@(5Q{u>N}>)2Eq#SiJNP0g`I{us9JeoM1HXle-4jkTw=Y?tkRvb;
zf8oGaYZr_&iIF~zIhc+_Wssvxm3>inwPZMbwXy(I(h`g&A>2}5++8KzGtXj{{PDPh
zDMFT4uEtK|R3rWD0m@K5Lk3D~CM3TebMVn8)sK@&0b$#o(>=3eMo<$*3-CJ-&%o!-
zw@w-h;^3FGhtMU3VVs!X?!b~H;Sb^&as;L=BXZ;IOADXtH4VP$&I$p;!RJd>jSg=`
zkF+mKp0HLS_PiLm2}oXYzfLeiZF+9fDzd=nu+zw?iRkAIh_m+Ptxp3hIJj93mUsBb
z4HiF8JR?ksfm7<<SJeI5SF;e5Hm7uRAy_*2T=G<G(*&?t*jE-z2~XO8mFx(t2(vqN
zm$8Y8KvGBStJ-u!m>Tr%R7USG!O=1jGRVQcI>}{W?($c8{SNl~HEodA;Q5|}sQ}1{
z2|Zjk%7=M*cjkqCZQ;fGWJx{6N+0-O2cI8Vx(BHo_Php@$D4aNKFxs@Wu{5{AlgY-
z5TGTw@SmUSc<I#Cj0&PSZSlbyIRNAZ_5CqgxmdaN0&x)^kD>memKb{0Wo^3q4{+8j
zg%0`-DCuL(=rAWhj@b)aHx+oqT<m920t+UU_AODPSfg?XUbL<G5p_VVFq(@lu9d60
zvVLF}6$|dm*WQZmLfW}~U3)E-s6%i&BJdoTEFUC{3{TtT`RTWgPCGv(oiK(+2tvG~
z?i6p5=T#*Dfu+r4_1fijWnG|!EggUo;@)6+G?@l_yS=!vt)=7R{Rf<8aiZ`QvtUoy
zOWL3lYOQ_aEaa+XdECL&Oe#yS9vdn|a9ew6;aMc26<(<H?6fXXG|?Nr4e{Av5-R-`
zdxBk6y^(c4Ww3o-FmwfE&OBtwUbZd8rX-6Ta;eblu6T;7gq?F$(l<6MNg9N?!ZD-p
zJ^kBlS2x*$IwYJfN8KGDDYjs?L$RW^$Hu-b$%znKr{*Os4f-@iP&e5%)nCLcMS{-n
zz?%;&xy>zTjO!<Qlf!mx*O5$`&a=C<mgzbY+B5OV2b+BsTdQ4{?_7&mxzTCp`Mu5S
zM@K5c@cQf_BGG-l2b@;Wac*Gww5Zu{H#CCTbo+x&quY~Zh*=E%xUs%Ux|&jq$fI^(
zNo&{Nc(vX1zgj71D!Uv^>6MX`7>p&A#Zdfs802d=Z+rU9e8R%}7v}!I`OkSnecTZ-
zHI|U_QsCnD@)UDkH|hLJ@t>fXDSJh7&EELznDaXpg?f7yGM8BM%8s;C&6P_Fn`6l6
z#L+nTQYN4!dsX-TW3i2^L{7Zl!6Z>K;Hu|wNt$2yc~hExb-Ob-e&kiKgs80PU~Xm9
zJ|Tql{}sT#zI`L=m7MJq-*aHmw_H%-B#Jms$1S^+y{4P@({fkE>}o$h>R_b45qdpN
zk_8oWZE0PL%ETBd%TIyK7h!l{QXw23okFHZ)w?CfUdMAezoj$jcn=8_%NpH@CDMrg
zX<k&ZBqeTbC+fu!TV8mTG?G4jF$~t?$H5#Nyoi|C3d~I_x{o;&*7h#n(!tc`O>!vw
zkAgFHYX?)Ho+Rd74op*y>42bMyR93%ZsX9DOU9Jd%@^@1?8Nd$8Plpe!5eg(+U<3J
z)M%r_N?9#W#8VG1>~QcY?LLY55WCPFjaL!rJ&f<OH%|B)Ky973P`>T8ZS$EP9oeYJ
zZU9Bq5|6{wXRqr9A;Zuj|0slOFnZEMm%YMPwr{U=0_0N%mJ5uXNJ5=78-48|Th)EX
zb}c>BDn-<$r*7x`9tWS(PgFr~i;{irT)ZRfuo36ZeTu~+$iTrz$Hnbs-4si_Yg+}h
z$tcg8&nhM?s&Yy~(Zb52hh2dj=Z@sd<gm+<h?Kax4biemPI+PY0BA;`Uld7ueX)e0
zp;52jq^7O5HElc)2t^CL!-3`UUvks+SKHb~{4gH*ctnCJ@PQEtPHmOKvaWm3c1S`d
z<x37MyE!_(fg?&(4(RTW+wDYOd(fMek}CN<YteI+dVt?+FXx)ylaF4QL)^ho6v3?p
z<E|(Nz|0-CH)MZarGH+Hg}Li)yw={>28VjgA(jD%)i=0#D?Wh`Gd7BI_NF!jcduDF
z<+N^4cO<hajZPV1m4F|lJy7gXTk|ud=+pM*ZUC`RFqvnh5~0c+cQ8^@$uoIN{@S>~
z`6UJE$&X+qySMdX>DC_gZ5w?P|DfPhY->TL2kouh*F<bs7`}r;(_<Ntz^ClS@`hpb
z?Hjr-X=@V;iX@2n{oXK&85P@6XqndU5qn#`M6z|u1;xC^{Do)1!6&DI51fQTV3PKY
zjR3-mM9R^S0}@YMm|NdhO{!k^)JoW-KZAiH;FD8K_I0eXt#7w48|%#sJI#_06g@}f
zUbunI_C04qtvj^j5K-JqgY9GVGnnw44c8O+;Na?u!Bk+xHQw>i71l4R9TbjJUQ2F}
zBfwQ{1W>NDmM+<bg0-As9h3nqI<4=Flsl+DS=dNjO?GZXjL57DtgcqwSCzj|QZOH6
z8{1ec?@z9ub(-;dImu*UCZyg*Yk0G>xp*ra1qQdn!Rs5ZzUc<LuWm1SHxv_Kxxvfu
z#9N~P0S<3PsEc!LYW$B=6?FLSumibF<eM96ic9(O0b_iUrH$37>Wx)LCOM156ww$<
z>6~*_a%XAryjR<GJkd56rX^M?_A@XJCK-=T&BJcR<~v@W7sdcOaAba)55Q3Hw4!#}
zL|b#?oP*)Cg2inl57oBRXvnIdZCaTJQj|qXN>>LT6QHzCX%r;A+S~JS2;UQ!^Jo>h
zo<tmp6N0#xBv6~oiMaQnmpDJ`gJnEX3?1r&^l_VNF9`cy8jgTzfskB2id0XmZ+Ew%
zH?)n(Oov0&!9@3>mb~on^A~NVV7g8bT5NC*EMxH}`y-enCr2MlDd_%sGNi}nXG{hZ
zh0G78EWfI9A={}e4BMX>17Tac_wQ$`*#niHabRVR?);&AYlb3p2U^+#_2#f32Zz-L
z`8aQG7|BPPj1>G8W;@%xO*O$}Ryywu7Ey-uP=$aWOn!U2jz;5mjnDzUaGzYt_KwDY
zw#<yF2rccBD|G;K`0$y?#l-1%c0`irL+F%;5SsX6LOdOOGPHFhwaxVHp`@^dxpf;S
zokk2tu9$SPP{Jm7tfH#4M_#aZwKayItAH3KSLdfo(UJ{J)1a3gb*|0Vt)`42uluHF
zykQ_nj@5<`%Du9maaA$tnK6R!y)7p;U}|=vyE>qh5mwx@;!b;a+h8zWZpOUs0FrT}
zQ#p*E8Igmrg!qDoyLrDP`va`9(JI0~I7d*VovXy1ZYJ~iJ^4QIou}U8v|^%?yk+vZ
zVDBxM{KoR?!GU}c7OPZ*W#3%HB#mD2CrT3*Q#j;c;;Ayd<Y_ji?OW<YlE)A(5H{D7
zlF%a03w$irv=691Xy@91?OW^46z##VC<cUm2OpeLOu2lCkW&}5$igQYEX4);wr)68
zUo!nUQVf)hp;SHa(R10T%57oqYrGjJ>5eBKK^y}jW~Ku8)As&0Ww)b!6X)mvG}t%X
ze7${pJ&w7Et?3Zk5U`xl0X;Y)d-ZAO>^lmdF~OgZnBd@R%H;f#9(v5avkf>Vk~9+^
z^$x})D6AFpwSA!V2eXI_$)vM5u;>re64t~pF+;p*E<Rx&tb1_BgjBZ=ZLPsFNJ5i!
z&>m^eo+mJuXDf*TSaIMk`HUXxtfmir-0~4`DWYIgP#C(B3Asv(j#8+zTXjI*huiv@
z*@BWN<FW&im^Up0-go8isp-wr&hHu~ndjnh_T2^3x|aSyP_+Zg^|yXzV%EN=F_D?c
zQKu2TuiZY3ph152z4ai4%oBhUbq;`~3Nu0xXB@FZ_L27W@IU;u;e8XIKk5yG`Bb$X
zfMFJmIFFCEdxCfG$^*?q#s!n43Efn<hJ7qMzhxtJEC?p9D_AJ8uUueOzOUVdXIrv1
zQ8eCQs`#mb`jl0+?@ubeDoz-@bj4K;V330j7&EocN`FzN;14vexSqO&l&hs(6Mh8F
z$u$V-DdGHJH<;Q;D}p=B-C!C))146L?a^FA%O;jOKLefPC(qqkd#qq#17X+L#-aok
zbc`z^f}ZkRIOTnWsoLh+IwjR_g5B5wB;gn53jbnMpNSBA+g5mqIFdpnNo82st_1>+
z9DH&t3^8o^v6*Y=_uBTx^@+zBo^_h>BBTO_Lwb$aO&%n!koq0Pz*NLAt$&$;l^%NO
zCUFe3*s*hM8Y6ERotPRzQc?2i&QyBtn{|F^H7r>|$-rH0>Wj!o^6~~t&P*o=0o~mO
zGd$yBD0>}PDLTE$A%doCPhF(G=v<gRguXV|?r<B0ifwzlhiwmOfs=Y1m^u|T0QYrc
zp^6G2IY1!!o6W+(=W<-f`N+1vJ&gq`Ne7B{V6~-Aj!tc|1NA`p4kC6O*>xaM!W+J6
z9O0)&PA=bt9jsR<+=T>8!eq`nu>5+2nK1?a9!hdtzkbU38RVFI2RjsO^>BkpxQyj|
z#(@=eM07S}D^W|X@gq4L3v;*J>ipJDvM{&u_GNao9wSvqk#Q6Tbzt$BmLU@EQz;R4
ztRv6PNXx6GY-WZ%__z~}^_2YDkpclHIbg>J?rhdZW)JcXo!b}jS-D3}!5ls<+p~6}
z?oZ)LqLe;bZ&tL`B?q5#xEkBEovg1?plziN`65ilG?~42+NqCYID;u0i#q~WX-9bS
zW9(laWfV@=SE_#s<HG1pFBHH9CRSa_y6O1UCf`hJk9T<GOvFS+|HDblN&}xOjl5v8
zt`Nwu4?eLd(oal|aO7Ev^bWiYb_5A5niS47+Q=;xQF*%qOZvKl@(R82X*-)E7QFbp
z(`paX%&~9CYf6m|J6D(xjMQ5+?Q~{0OqdPD@nR2F1l0L_S)Tyu8Ht<@pxBYZL4rGr
zyCFDA_S$0?>LoJHORGK~8ng_$U{V_-`fnVoE>_x$O@x*#!D9}fV^u9cb~W+d3*0Ua
ze9^g9OGi!{Sx@l6P1&WK&%4%7^(V1NIc-^T?i#iqs%|8eDJ=(|DV;g%U~<un&y3p-
z*JV<<6y1M>m94e8&%xv<FXs#voq9dr7T%hb&hHpweSWjG$D8cQ!mxQEB4QwjEsV$;
z2a`Kde3L?-_t-+CZlovC-^uwS4lFm80f|cQL4Ra@Y_q58p5USk`Z2}^So9|Y5mk&s
z4lC7E`jPzFsTuh!X=Kp}<j`Ztqr3S3llF9e*t_n1*S*f~;@l)HP|3i4w5>+A_KQyI
z*)azQO#LuYj}s6Zs)+qq*Ke7j^)0CasZKsV4F@mIJ2kEH<kR-!`PM5Qy6iINcgVcg
zUSmH|Uy=r$iXvn}_BfarKXztT?`}5#6ZVsBmEp)EgCxQv9AJu4(-ek|sEMCyG#j_3
zQ=Y*T%aV@Z;8XpD!C}}}>0*Yj%VFgMh>*0O&Y`$fS!k0^ORQv+H2-mhviwZFW59|R
zq1tTLA%KC8?t@2n(eUiDpDh@Sg{7(u>Z}8c=?~8zy~xn#)HhT<`seb!3v*j0Mx_YT
zj;YTBAvnDWT~vzjCHwgt@Kxl7ku&c!i!TsrQoJtKm;FMU^etmO4riWn0MUos2}?G~
z%6ju>sqpfPT^}U^68BS4qzb|~gykB%w(Q#_RsT}{CW3Ox`JGE|BhHOqZXaK2GQxxd
zi$01UE9)bIb;3UIEA47&2J~LX3j$LWb*_uykqWkTdcuCSJp|2rHylW|?O*^7Cd<YU
zJ39(}XIu?d`(JBMCs+uZNlaLSarz`$QM}Q9eQ^xMA?X>`PUNn$4koEHeE3AWk+ks}
z**^^C?PMw<P^TQQMrR6Fh5csZi7Vw%QSAT+6oaphM}<fwW`uVt#>0Lq7t2i~BJfjB
zTD^4H3j6K4Eg9Z`DI`YVkb}={b-1^A%6_LCqcrlQuZ6AJ!ITKq8-LJ#xBFn3SyD%q
z<7>Nvi4Sgrk3ed_*I=PPYsii`FmV)&k`VX~4A{Tly;?jUiI7P`vzXuMVCs<vFhpkL
z$se?FWfoMGQiV@}B^&F_5PrkLA5v)4AJzb9$5Z%Qcp4#B#l;hT>EPq(Nf5Z(<`3H+
zEq)#gPze!5bq*#;b?u!??T=f7`Z!J_2WUBprTs}mt+`{h(<nB)7%djoTKm)X8CoCO
z;vjZl#prdKV>%ddIwI1D-yE|)YqX&-bpjPBb^@qgunn3VJevK_8(Y}iBWrKmX~&g$
zf~F*wGRoRtH1>!NJU#<OI*?rM(DO4+ZD71PyTSglQQUQ2KgsyBzp6K6jWgLc^%2|Q
z)L>$yMYSz|=C8YPTCvh;)CH-)!rtv~x)G9>QgXu%Z!n!qBwyfpbjH8U>lWrR^^Rn_
zSqGL=LcGtk{ax2FIg&KpG6-NNcQEl@giba{F~2kR_jR?z4lQ%#7eq~R2a|hYZ}`3m
z;yLUelA$ilEmfqn)6BP0E_2rYu^Yqz$5V*SVMz6cmkJ~bkc&tQ8kav>-$nbUHbBFZ
zeW#I+7Duput}a$t3|WWuS@UNdOw3%0VY2MWE|a>UA`JeeK3D^Yu*MoX=?$ierrPF^
zQ|w<mQd~T1Cgu`648j{;C-F9A6)}mx%PcqaZ*{+H0t&Ch6v&!Z_>V$8b~*T@+o<yi
zEVeuO|Np(w?MRO@Uc*j1uGB(Mpf#tH$Hbqv|0s;y6sJQbl&mhT7K|hd$%WWO?LW6Y
zeG5%)3pYJomMuY1)uCU9e77eb%J0bZmte(j<X9ew_@7wkNurm9Ih8o&xHaW*{8#ea
zCqF#D!}+ag;!a!y_TPzFm|OmO=XX$A?%(@hKqu{gYLJovX#yq2EZRN6r`)oP*m1;)
zDYNahP7ouXV5k+|2Tbq0<BrXQHrOX^Pst^H`wAeF=RC9O7K>bSRdO2WURki`nfOQ#
z3rXRuee(7kboETMsT|+%iZ~bv)GvRfeM-J}VUA$O5vLX3m#65Z<o~Jl7EQmjRRyTX
z#KIwR@UtThKIQ+-vmAO7)#N{IdwufS*>Q@%Z|*ygnE%|y2-<?i#JzUer?<xs7?Th^
z!YhNQ=HN9jlsD19JgYj$5{rqKC85Ds_+isj?X%C=9%39NaC<nZ#+<hJ{?aX55DPxD
z^>~`hs4WvRxC0Wc-8DA5c?(9m&#LjyrTL66UtEUHRWVwcb5-wFV9959uru=bGjlr}
zST5(%LsVs@#SF{O_c^Vzy&;7mI_)Y(A|;gST~axaEHkS?aSaCp&QZg$0`mNV^Y*#<
zkqdKMC$>6`=v=uK<-xJf%drfNX`j<7V%`zz5xTW~eh1qtMkYq*z=|6Uoa4#DlbT2X
zuzkVy5LXuFp7%WGw`Pm#!8&&N3%92NaLvumPsudvC~aq7RK$QGQ-;#;dy1i)bTGN{
z*hzQV7q@X?lEw-6p&cE`m9`s`XS(Ra%R9f{o?UoxUS({>Vd;yNqrKq=PMYK@smtL-
zu;;Y*5sNp)woE%$?(&R`n6)p-E2z{EhhjNx4dKEZJ7<^!P+q^Z8!q{CQ<w}FVXbuV
z8VL^loUkuzBawz(R+z#?WvFPFFjE2xbsur+v^}@(5s#5^MaGMqC@_s4S?_5S{qln4
z&ZLO!6AmnyF&jtoJBOFpzM^56*xKTuFl|X-iq{;CuOg#Z7)+C~r0pvQ<|e;d#kVd*
z&YL1o!B^j8Indl~x38*u^mU77f@2UPA_f}(_}r<{#E49pvaeqJbUL175b+2OCc38u
zr;<%9EB<Onjos6}CMWL6M-UwznLp*ULp79ah`XKm1u`}yPjdZH?WuBv+t(&JuED85
z;t?(Jh=sW|y)C1|_Pn;*IOoY}<XVzWt0|m6Z_jV!w2Xa*cu?ALjS8p89HDdIU(lX;
z2QjvjPP1l4VZxE)kft2FtT0qCh;5QMFkXXY(ZR%HCHGXYvAwVjM|DRuUJ^ZKwX+Uh
zH2AX1FMp}MsOx5rlF+kGThCjV<B$rXfg-o~>)K4bp*KdY*%1ejCtWF3Yr_Uab&Srl
zc6mN4f~0gA?PMJ=ckN&>d^p*u*hh$>vnyIZHDF9A=QlC;CnHu$fsEj#jjuw#g8NU|
zm2E|?#Xv?7;DiH-5fhc_C??eqQsl>JdvQlBDu*knL0MWVy4)~nP!SnrMA|`nNjHrM
zKs@a{S2q|(?J?$A8Q`VcTLQ9*)Jc6wLJLntC_Y`7w{vZd=N+6($#9N7*te_NcTli^
zT95r<2a+=%oIo5*$pDJGdRd#Fjo6*JlD2pxA-v?m+SRRf8^VRts;4MCA?qD!862}E
z?Ts?2kQy=NK!R?x%VTtBZcf=X?Onai<Pr@#?YIVON(h?UsQ+|ro3l<ERCWounp(AF
zdVr5U<l7||OZieEnd?&Oc-=DRS6q|gK>AM<M6koIZ#`Jqo27=F)2@Y`$sie@tzg@*
z@z@R3Lmkf{+B%6q0IaC)gnYQDKl1fug%i|LyRqGaK~1ssS`ol<aTAt`BmbssPGKHZ
zMIM@xmo9&$-CR9Ss!5-rY)ad`VOoHN;szw2p0<})f4Bz$xp+AoP-3>|vEen}A!dil
zrT$X!V19?aqTX3E1-(iPrzV7h$<aZFVlPm_d$Gkh3CCX9MsryR(M~(YUHvMlA+A4Y
zuiD<CB&AYjA5+p!pGHa4AxleRJMY{P@0Jv(5E>bEVjtzV4mEytzFXDvhtTGo_JB1`
z6j7>c^gwPcf;sH#+uV@)60!bJ6tH*`@0BE^=q%oW8I<vhVB=f^#ATva`D@xYuct5?
zWwDnw-bMcKIBdfK$Sir&H8<ET$p@DcJ?Z>T)4+8%+-%G0mAR87nTA#(LSo>-$HI^~
z&x6Fa+wwL|^HeJV7Kb&!_#SsKA<*rC9k+TaVpvgEPFr<EP9ya){W&#w%Gp6yWKSPt
zgW+d79-?j<S(4HqZ>?THfDDXCa-T24<ZRW#ykxi4b=pt}-ca0u151J@Jz;prZm$p4
zYLSsyKC3vzBIm`y#|JAx2?M_V+8ym4teH|!<8SOZIG>O+8a}lNxtZCN*>ELqU0@x)
zHlImdBDz4BD@WKnFzrnE!QNo6Ycui&PV%FXY;VcUS6^o<yRnY)+Kilh<)n2um>ei}
zCY>8E$-!2&%R?Fr`s8w=K^#ncmkiIi<Du+Rc4xjfj{k5PDFpVmlAxhczbgrh!!>kh
z-mKHs8>v==?{y2|aMCAN=hRJN802R%QAl2cdYs_G*B2~pDJ)#PoDHUV6ivvtv!>k(
zwLC$W4-%ZvU~Hz{5?EUwn5Fi_C=^2`#S%;kQRKgAIBZ>AT5(MA;FKT1L`PLfLaFeI
zu)8~9vdd6AYI<T6GX^m2%BapLS7=N3)DzWlnKmQjpOM_<U~<xs2)5cA+I(Jj{Sv1U
zi&`U)>tH{_-<=j<Z>-LUf~9ckoNi!o4<fJZ5s0ty|EjcKuqt~~Q|w50VXpXI2PArW
z!|hAg*_)H(*W%(BAD;4RMiBwbKyS(3p=XlbLnBuQG6ovUy=@S6<h2S=0f?c}Cu_|T
zPGWB@SQyK53Hc(7nF9usv~O6vmrY(zxahz%qziLVoomY8)?T<C`D=pR8bFfu|HIyU
zN9TPUcfXU~tJ8aI#j<2uRv{(Ii5ycTCD9@&nxrHbxjX<Kk`O@v#sf%%WI1EWP8=t3
zNpX5Ykp#hBR9HljN=PtH@4ffld(ZuRXJ&tZ{J#I(bI*IuJ?CEW2<O?^U)kU6?CkXA
z-0iX6&;$q5^zE$;y_ep7`BA5pq#TwM1-YKGUWesl(2q>F!dq02(qE_l1@T_W536e5
z_%;;K?QzcXQ1NR9$gXt&TLnkC;DO*VOo@)tK!S6wikF5hfVn?<@q!I^1Cbpt%UJ<S
zG8|BTO{VIR4hs~Fgile($QvwjkQnEaL}m8t3&a<j%CW@A;q=D<<`=L+bB{`aZ|zpY
z<K+5f=1i^zdSVmeX>dZ}USemLhgTcQw%QEoN0D4=bca*JRp;>}Xf<}JuFf54yB29_
zpNY(t?F|v|cNr^H*$S9Mi7Q`Wqe+@@sB!0aQ1%i<FDQsIGm3Ap`#UUXGIBpkvsYw{
zI{k@WPO1+{c}KlkE(_^`CrMy$FoH+5$Fo75va$B!DDRFuLXWmMoc9XQboz6CL71W=
zbR_T2ZgGt5zyn1y#$H+%Vw?ihrAx^jcEHBlY(&X)2`L$K^o-M!4#JylVv&c0nZL_i
z@Ya)ZRr#)vX`i&o_KkWCqdswQ-V3&%96-`pW+?w<Z>#IHOl3Ky2$TcM1=m-`>%gXR
zD66)-&iR=X#a5^o)N>0@*`T2b5lzV^8TBS7iz0RRI{nG$R*#O1+jM<QdfZjROiBW7
zbf+hcWt`F~FDB_vGi}G)I_dn*$E@$M(L7LmcIRl;(W6~+p#vA9)}cL^cgze;Oi#}^
ztr~jN@L{8W!gdvm6)eqL)yW)~5Wi{VZ8qBwv2vIr7Y0tdX0uN1V=K@+KF*g?yVD-Z
z$3-DUxg_`s2Nw5eB}ow?T!z1$3Dh25oLn#}L0|D5tU9Mh1G(yvH{09W*tKnih2We6
zD|gMB&GtxLko+r36*#bBEB)S91p~gLVesD3t$n8z_z>fZ;&((geO>YSVd_mdjko~W
zE<_j|ir^jWB-Vo99u6#(8aBlioh?ym-rm_!#?ukx@ezrtPJg_CZ4nV&!L&#0(KhNk
zW+=eQ@&}OS2sBqQIkCD(mrE|oZRiR1SPeqtr|^VG$B<9S>52QPV;YteZHY~2Rp@u#
z|E})C(Dh(zoCD#&;=_{V$_l3){1DsJyW5Aws*&43Sr?siz*~QEiwr)AQXovQq`R+g
zg3}1nHM`Y_7X+URc6W{t8L@9@XhncS+ljwn6?A$MZE&Fs9yx9AsjKtLLKA1Kn3A0y
z(Frx2c<47KMUS%dbw-rat~r*wDFj@5Z(DQ^J?Q+-H7Nm32k{N;n;Hn2M|FBI%>c@E
zf#Qa`9AZ=OUOpEal6`aA0ai)DCX{7LIlw>zz3DYc4k@|iynRa}ZDMAeMh+343@76i
z`|N!U7O6Ao=?0eTVX$+;-d`6;LXsgEkB<Wr1DWQ$PDH&AbY!K3=gtxg9WFIUa+O4+
zM{areKKo!&W^VvzlI6z%Byy<mnFpBLx7K_4`dX;_9a!3fD6}v>@gR1JZ4#Bgt$jmi
z1sdpK2No#Fy(S{bp=DfH_U+x(3Ru^gea3+;L10>p<pe%$-_iA?KTzQ*fncX6rb;%A
zDpC)6G8JE;HXmwOTz>5er&0V)yLrmb8B0&R0A&1~)vpH#hZ#4g?2D9H=#K$r{E9fG
z&TWvR_q&p}4UJ7XKZ^;dG>jf~!M?kVlGVa$X3fd{N5ThxS^wVprG<&*XlIu7Pth1M
z)qhXpbSsetomSyVwIXHE%%`2V@2z)Zi3z>=<i$e{EX7;N@7wMBYAESK%H_3G#-cwT
zj@PA<shFNU7Z2O_w~rvuRjvx<iRklG20#Tc>8}rubKH;$p0OV&n8NWTQYfYySnVLX
zSE`~2Al&WfJV)#Y7e4}RK`~Mk2=49?(zLE+7U@ozwcUyT9jklJPm%b{40hYc+lI$6
zH$773Xcv3P3@8}9?Md8%_2=M`v}$!`@PSH4>@AqI9c3s{Guwd$n%FWR53OP`3Nk)y
zUwbE|eqnu1yZF5jCWTx(bC!vSSk&P6b2;jb)B<&W#~6rDM_lH5R>1yx9|iCN#(_la
z7piMckJ!KfF?6ui=JPpQ;knN57$cK}AgOfn@G9XfKG0|a$%!G-j0+{}N`TCAp&-Bx
zcIXhsrN2Gt4W_x*-cGc&9cs&L*$vk_jT*mledWHJv%~Fz!Nic$i1&rem@A04VtBM8
z)gu8f7DvQG%LGMFJXHQ`(F^`(J6f=$LTwV~9au61t?1kggFZzs7^dx5`yv*h4CRD&
zdxL37#*$;e$J>&-Q;}D+IyNKggM1FTb?e<n?L<SU>W+^%P0`>^3<P#EN%QV4tDK)n
zTo`_3{5!*Vo~SR@NK{E`Z>Uk|4^B_YJE~c^4eQpEb-OZkWnrfXw*%Ac&x~-Q%!obJ
zU4a{4^oXeen8dD|Fo8Xt`pA}honJZQA%MjcRp9-pHbxm8!Lz)s9gMXI39*(>+37tW
zeh-gb*!};Gf6jPnhQeu$?4=?VT`6}vQ{P1a8s<3bX9A0RQqd6GRkD(fpAD0xoz11Z
z4sH3K(f)%@JFq5vDHLH3)G7Xd%$`XkUwhLE=co2<qSH`E&NZ^1-ZkwsG*^A30M3(s
zwjP$m%U`5c{XOKZCnUKd;b7EuzPm?xo*IVY4^)eao+2$B?iXiv+l59zk`O2j=m3Hh
zrRi2LmJ?|)Uoc<tT)S=i>=1{_4je#AB101s_Cxt18Hpa4$iPtt7I?%e2ZL0lA~_kt
z5ryQx-+nlk!*$nQf1UF?#-L9{QwqMXG3axoZ)*)wM7X2Rz06|@4lz1!J+NVU-(r=A
zx@eag1+YKKtSMS$jq+rz)FV!RplF^0Fpe-vMY1h4X1$Smn1fNLt;KTfiWPP_7t5Uk
z=U3dwkb{qMQjGmb!^vvO<#dfr6o&b5L*LfXDV?q600KLrFO^iMwj!hJYA)K3*0&yH
z-1`1>q8z?B@ATB99%9FpPIkn8tUZ2`1i^`^5eJ})@DZV3Oigdu&e@?qp6qC6Z%2QK
z6LV>b@Fpjw(c$bTQp_WSoA9xnk;D&3{A%@G_inJC%!{{RP({|Dr5=S2+fUU%n{X*9
zMxMrZIz76IxS8ASr`rHgx?o%b2ax2m{QB$dXF3^EdcuQfl@2IYmFQ8ffk>B-?qxsQ
zA)aEfF{7uv!C2d}#<NeHv!CnM@$p!fhv3e_vm|AlKu&*h(#e^J9ulAY`JCqEH(clS
z^QL9X?H6*!H{I*}k|B6zTtyNvq|VweHsV^cT#_D=Ea`8oLH;TUOv25@1NKYp&Lu*N
z$QC~e+wBuhf9Z27hnyw5P&u$r$NtVID7o&J>nVt<;VY38AgitXU^+Xt^=IspwmRg4
zSP010Y?Z%KsLUAS`0Kz*p`O`Exz&OFYBv<6I;CAFr7Wpy+V}QAm0Xp3N1Psk6)W~p
z1;qSX4U!?V_9XK?j+EVH^g8`P&k}Q$HAfjR`}L+ePHrC^!)3GGffOp`HqtuQU+lm3
z8+CuhE3yrqbzq_{f@v_4{${&?n}}J0xg78YlW|9)BgQKGt@a7A{RzX0D?#9KdfJi;
z$Brbr-!7Q0OS9x22UgpSDve|zfGo`{kJ(xKoesAI9+5)i0P4ZY?GcxI(th`O!-!La
z!7&28)1$At^zMt#_U-qQE?=|W`6c#5{FeQGdx1()p?^65Epn!9AGW?fSRAvqYzaFK
zED_7gW$ftuxb}x_i4K!GO`~RMZ=eM;gZ)u1@S&-ZDd%^5x;T~gKZQ7H)wa-@Kh7Cl
z^FVJbD^WQRJ&}O-agG_sbJy$JpVV9OBoSzh0WQJ_r(~+0V`u%-dYjZC?N*|RoUlCP
zkp9Fgi%B#`Su*>x_5}*B0i#DS8vsiq;DAamgYEX`g^UACYQ~pDa5Mx%PatwP^n1eo
zB2jb83t#lYS2>OR5aj0J84d?T@BPc1h?Se~Si9l&Eq85j09R-tv_nanhT4OQC9=t}
zEvZ-JMv#L2RlX|=!8xt?mUN<pnrMIBrsnpUF{cq1q<2``_*wg#e8j?TPL1B~v}!9>
zHcap)5U^;T``g?Zut;LJn0A`AmKUQ&Q}+40hW8k)RG&DT0k)#?KjriX+e$4uv-gwj
zWGIVR*dhD-eBJ8)V8`^tq|+3~S;HB7(<`SqWbYpuNoFVnvJ_PhFasTTdg4=)i{uXf
zj|EfEhhkU>Eb;~;xgVAoWmN2+8cN^_<6bQ!yJIMG|J<GdbCjyhVQ1c8Vc%sjKXLJ_
z{Y!nH+#I|sB>s{OrAH)covTDE1khmJvwy8;%iX51KSD;m!4~!9Bld6Y+i+|v>sd!g
z0ZaN^-RG#kar^hhRmR(r^@ta$^wg7Z2>O#0sfRHg@o2O5A4Pd$b<(y~nsKRo7p4~|
zlA;FAwK4Mtz`W=)4-z;FyZg_^StdsnuQG4|K~o1%Sfm8%Ot~<K@`;qxu>WfQDi{&9
z9l{^t0F$y(!$1<(=0@!Z`|qR)%=S}#PEV#86CFM#o_HEb<A3VrvSWm$EXdMZPi_lK
z1X;7ajbLPV(YEu9H&_aoQQyHns=F>ztmN4O7K2rq6JHocwj|(>-dodxjSW%2X{R-1
z^UV18sC~@d;L}8&a*8VrNkPaxWA?H2mR!m*172`o@s<!;YGT5yXdjo4S=hb&x@AtI
z(2YhcF2Y{iOh)*q?BnavB<_eK!lj;aU^$zhN-IOOoonV3_Leg-4g}iSo7&Nzc7V0G
z_QFo*FF#5I<|lSAlKLSynHYiPWAPH_&L<Xpf^GYg>T0n!*^0Gy0@Grc8QY3jXP;c3
zf$N&B0b?3VJD46%z?$v!#~2P^Zb-41h}ya**?h7S?aNtT`-sz@7D~)uQxgx0{eEh`
z6vBCm++bRT;ze_yCjW?iTK75%><eqe!~@H#OV?F=AS(?Ve|q&0kc)Yp&kK%?rtkD<
zonHC?vib+u89$>NGKM}zaW=G$mY=oP=@(IK<j@!U%r?ziCZ@e>JkT4l&uYlWZy0`J
zh;k98I{c}deRg-9VA>O(abWQkj2!hOy^1pUpR*U#)si;k3?qQv8%!dBI67sN>~rcm
zaXuX$Oy-j}*gWdxeoY+il+WE;0wZD8V_FRWv>K@AqEaUIdHJmqW79q7cXGyCNnW$h
zZ+C8@EHbH74j^V<OkCUqTey8ep;A;5vBJ<r5`HjGr#~D>ue`;+aBm3c8>c2lX5u6&
z2a*p&LWtyyi*xoxd&^RtCpHqVuCqTJu%_hJwOcmZ7w^q&ECV(3^kgwxoCgsXlhQp`
z>0^|hNXqM%bZC{_7;=;Yt4RpUD<~Ax#&Bg{+9v(>Th}{{*fk@5h56G8R~Cm(U)ArN
z<uA+E%?$5QZV#>6(lPfjtOx>zlPdg6B^bWEh9jX~bp>P=0+x|1JNvLBGCb`o>M0to
zf+uC(tXvROil>}@`Ks64VPBb(x~sN0n!-8@&R*C)<+jzEoJI^QH|W?%d0<z!-Ck6~
z((;oreJ@%{gK_c__|dE$uqzrzWv`)f5Y1!@5riN|=05wXf+1P3x{^MD1;SehO2*<U
zU-reV2M4w^hrJTeQI(!^dL(|ZV-EPA$ek;fAVrsIq!Ay(w@WafUt;(WK9YgGBxhxy
zI>~*fRj0qBN6W)SGGgDZs&|W^CGW`=#TiBH*YxCiSf`x$tGhXkvfS_=Y?`v77t-l3
zYF3hw%sHp+rS)t%OXP3YNouLI(32WA=Ooz6x?4HyG(LRJffb|6V%6#1z;0wOudDfX
z)*&TG2)${rZi87kY}d3WRAt4qTr22^9x+7f&Q45@Ve)!K`_KmnAaYtkCEAZM$HI2;
zSRfe5(X8Z+Yuon3ad9)mi;PYZ00S`VX5n8$rkIYjbvD)O>R}oDD49&F=Rz34P=%d-
zZ4WCsXJK??Vp5ZKeM8M^&g!EPSd|>Ek7XEkL*q<%1knt}96%y-IhCfD)n_M5NNbfG
zr%yRO@vKNe<5>i;+w#1%PuS4FX~YGT($lhq7woYc3#R3-fIZ5*I<VBjw``x-Nfe{q
zlnd^ro0dC2W8PAZPpd%anzo|79>JU)9AY@*4HiV9C5|d|b9*B~HAQ0-bmhQePS`3S
zuGsLH4TWY!xS+kV`V%JBj~KGVVnKg?I`1Y*ngFfS_SMDM61DLhJz481l%C|HTovDw
zPmcOKDXH1lBo}z?3?_h?olaAeu<EXLcWts;>aGmems*uIQlEBuw5SFg*oP?asv4K1
zEf!w{Jpt2$uvCsr*w;2z(;Gnxb(%TT%Wha^ukJoths`z(Ozy_eaGm~ILW_wMv0d28
zu1Bu;C}3Bsm>v!9zBgU<riUM~TkAU6BwO}d)PD2UBZ`*UNVaRC{e-Q`<uS?3sEmSC
zuhR}N$Gz&-RknJ`Qx&Ha=@pXyou0gGFf(Cm+NW>op?o>bB7sHgFFQ+U*yzl5TlH%T
zlNBv2#gGUmuAyh0bNaO@Jo3of?e-c#V_!moN*Ww<dh#XAm=F_zG$Cr`$KBCrGn9I^
zuZ7dzV0jLozM9z6CWEs+b*KXEc&@0Z^89#u$3pEjUBBwRNnTZ5G7gSn@JjTiUxR<e
ztFEy-^U;cnRc%A3T?G1gNYF}+k%j+BTh~NS!mCwGVzlpoBu8wW>c7p_*Nv^1UvOB^
zMrGTEg|JD9&xYlm*$g&j8`@UQxqwtj#^wqnS3Lq-C0+$N9I(6ED&xt`KA5~A>xUAz
zhvSH=GJcRUN0%PE{21BQJ=?gqqW>_#AT{YA(msKO7&x6S5y}XwXvAKdG|S+abEg$U
zfdVe$LS$NH@+Dx{Hr3=M`w)d3abR(^gf@L%JBa+4ZEm!<{+^9aBfsym+>K>slulq<
z5~B;d*SyyGoy;16T$z*9Qn$MsEX1EM@j9@0nvMcgO|Rf%GPT=1^(hA>S~;zH9!0JB
z6b5JawvmW0CU*=<Iqme+)wt<+I|XR$bqOOJiqg*GoZ^BMUohSqIeX$w_WE`&JX5m-
zA{<!J>~VGj)-l4jgarAMd4s*7?vXA=peuj(Ixx}c`n6l^jnyrZKhMi3%;F6uJDkF_
z6&`<Iy+#%QZMQm*eTf?7^rsYZ-L)&OwKvu0YxbJ2Us}SP(-Xtx30+ONC4!#4xh+W&
zr#al)X~$6e_t;z7eulGQz^S+nAeFW?tlum&^427WdsIKqX%vk~bQU>eQ2Jr(B{?Vq
z&-on_qOmsleyF>1kxK<V`eM(bP84nn8*a;OVu*~7J|$IvEFlb0_dy$JFvv*>4@BrY
zu-vHEYx%R^^xKE^h@Y9DCgi{YrFnqE%g_|I)&NtVO(UAZgDyBd;vEaS*Y<2%_c9%O
z$bNb8hyyEekt7w{-mQCWNY9_#V~Rh(GIQR!D%VwqLQxmt6*k)5kOx~Zsq+pjrg-V(
zFOdlB{<<g6J&rR*9av434%dv7(Be!x>S;o<^821+JvDZrBm3wVfvnzikB!x(@_5T6
zZ~@C=2Kz`E+~@SyXK;2a4@nK2&$OM@lMW~qfypx3N1UFR!}2Gx)jVnAITzw}lnQj3
zwE`6%2%qNRVFut8c(QDwdw6$FVn{^?eX%Fq{?hvL*viTFS_nkeR`MksSm7v-1tBX)
z8z?@Qw^fe>NcMom@D*=8^6p3lK=+^@u4Pk62UJC$<`4&3sdTK|MljFJ*fbIBV>U?D
z?KO8gzv|sOe`yK5*>rn1tV8aH^mriIIX$(o6#PK0GGfO}(c9APUltrK1uB>^LP`Ey
zl?#G(O~S#nJ<y>z<NTunv=}=#97Z5${rYspX6Q4Q#2(p$jWQUX+*^|GwP7k{Ph5%q
z7^)<9rTMcv+pdD4c*>BV;W@C3P(z}oNcS$pdin5XmtaSa%=jz*dh3yci;yF96<NS!
z2zn^LQOG975N|GkT;W6<b9&js_4+Wj@kz+Z4lF3G$)P%Ixf$?eMtOVPBWX^o?#RU_
z99Z77B<MIKYv=&$y*<*shrC{%#@cdVHQQ%p(qePY<?xKXqeDuR`9$_V>cHX+955pH
zH#>|LpXfka{RR8Fx<?UbGLRA1>A-4KidEE9DCAc>h{<6;_9kg4##0BeNrnPgpD!zT
z%EQuGDbK({d&=IKtl+NoE1jS51#;2IvIu7j8TYI`nt122Bq)%k-~d$k2d}}5YW}R*
zHQb+!!28GA>xhG+sy9Ya9Sba?i_^=Ueo^V#Eo(Nd+_Ltr4fd{Ny##g4I=_qfl_Q`J
z+-L7zM4qMx33#6aOHv>7W|W?4Ute#P;#awU8h^Z%#0#gt9;{<axr+7w4GB}|1cfp%
zTY)TK(w-?09b2Qlr+cm_QJg+w2<0%6`H}uGAhj@++%a@GI6{H8PY`|5zOil3TgN6I
z?2kBY?Iujo1N||3Z|miR^)cj4+Bpl-<QS{!+vGO*O>Kv}Yf7E~?sq`(Vd>+LP)|H;
zZs!w``$r?5R!TN94)JanFep;n;Y8bY5Q78D<wUthw$%4^PuH<%TJ@*A!Nm1t)jke`
zdVibraC8ebA|I8`2&0>~{zl-U#NZ);Mrzfz4_nQN3uKR>4Fb|g*2&HfQ~h94>Cn`S
z^HY2E5M*fI+V<s5l<sm`sba0U{Wklywx#cwaeimVYx>*l+Y2Mo(u1#{k0b}BCy~k8
zmBkY$n|()pS`J+qHBlR#^#=sAF#Xy|hx+&1hng(&hS8lvy`cyD4kYQ6Jc5ZRL7RKS
z%dYb8%==c}z1qGjN4J?A<AKwNG33@6?AX4$Mw|x9EI<yQ!XrRWN7phSQpCu<r(hZ2
z!Wv)~1Qt}Fy-s<#()~iIw(qT;rTa;2=uVHm?An`dw(qNpB<V`p%ettF;-N>=G+Mzd
z-(MFBTe9782$)z7nj}3j(p0K5>i7rRw?>c`Ws2EU8cf_rJWl78{a_<6V2e@3$|_Ia
z1^w}C>8#2|QR3z9f?<HYjL9xk@CFmlRAl>JIQirGm9yjH&QCpNVRtHHkuJWcJ_vM&
zq#p;C$UEw-M`k%0K$P|oo0#qWh~Y*lob-KbNqg&W2AvdO%F}>vWM5<B96mq{hLlh$
z$<{+tzl=aYVU|rcbX=RO{tTdwWmn-?5x3^_YYA+|Tsm(1bJ=j~Ddvz?5C?`TP?eU<
z2zs~q_VxHEQ+Tbd)9Fb@F%^d!A1G`ql!T%Vhzf@sn7AA-c^b>+!EDK{h*ghP?tYpV
zcKAaf1%;nMxfOP(%^vcyk`tdH4i8wqlFICxUT=r<gEq3wO*;+ogb|9B*pYnG%ucqd
zS*Ml!FHFE$dtAKkfE``(4iPm}<G|wTvhP44lJ9l$&#}5X(G$D71Jeo_?@!qA?lo{?
za41tqE~D0yB1hH~7BfC!Cu;M^DP39=Kb8VdIgT*IRq2V|)E_~)vy+87>6sW(LUv1q
zCd@caf1sU}x^T#zXsQZgjY{X2N&`}t8Und0rXwA=kwn6iyJLCm$tJ<!y`V}DX5j6k
zlLH5=SK)3aeaW6`^RtdKn#Y}Hu_lF5LZoC>Ki&B7^6Qs5jhs&<K=$oaT`%E7MroBj
z1Xco1q^t+SP6Cg@4|ci@;1acGomR2|&aqMMAS=_(v@7oDJ-EYZ#q&}XlE5dyUKs^?
z>LELuS8wX+5K6CB>`}y&E`E=MG)21kQW#DZ=IfN5OUa1nnkj0WZ5<<~FIO-f+6<7)
z<@TZ&(zD%6N<H{rqZ#P$NH?QwWWJ?7xSh{&LYi{pw3bdrnX?N?lsDgaz4J5ChaxBQ
zgJRv6kd{4{Gl1J}TB%SplmB6RpOg!o(7YdNj|hXE1P2MV=M<BRBzDs24{C$zaUc9}
z8*#o6p;pW1;Kfr;Pt1*|PC*OGXS>+0W@(CTKXI94SGJ7v-g*>wo(UKgscDxQ1ua{)
z!fC`qQxs7i`59ZNhsNTIsgtZmP&@|r!bS)wQ97BI9$_-!S8%Ay)z9LEAlNY&j`n!D
z=MY$&{?zv1VHH>bxA~FoeaeD~W4W0mMKREm3wI?JKAcP+Zts^#L(3}`q}+$WVVr)U
zlE^tuTDKpq=a!Qrc1Usk!%k0qwKPC>M2POk8Y+3V7`66=WIyx-4S<6nGU=O9Op~j0
zOvsOS<DY~1W1QmK!?WS7CyLGTO?q;$D*K7<OSDI)e*@+d?od@wqF;N>0HJzdKUo*2
z`jnSt&<aY=#b=%V_#9F1vpU^}b@)^56PYTN1|ZGCfhGDU@a|2J<b@q-U_af4f?y+M
z<#E&vHX%5pxZzQ6{UyGEKnWF!`w~L-+Rrq}lhwM5^CdAr0ZOj+0C8aUvqj+SQRor0
zLf%hZGHEQgaed6`!q3ljvvD45w)hofE~iI4k6k@dUw^(yj9Ys<Q=bJA*bt%>3W7*v
zzmS{TDk6-X-^CF-J)D<Hb`a5>vtO+H8xQ9UYA;wT@EBEyf2n(p2#~oEC(KnFg;83+
za-c<^p)ZznPXLJhauWCY2@ZoF?T<L^v?@Lb(zdHAtNlt+!TSCvyp%@C3lwV@wqH%G
zZjJxETCX7h8JYgq5-B|j#=$660|tobY10HO$$q^pD~buoJWo3?R-rwT&FnV{rdgNo
z1_{f774OeNky+AD!7P%34%lzDsYUoNX8`o1G<qqO;zBaAa8<l58AF@~V!u_ec6x-0
zT{oBnF=EAKTDRXWQaZ5kG#+SL$=zdNX%WdUF4NcVR1Y~}SdtG&Sz~tD0iB-w28c~g
zKt<LsWnsVDE#^2cjWv-DzY|W6n9;g^Zz?Ka{$A5edc#w*I=-v}O!bFmM#rY@_q*}Q
zVH3(b?^22rJra9vym^KF!6G`8uq{M7up$L2(QTJ|-u|%ZVz=M!{0<C?$!HhB?qq+o
z_!!oE(5?eZooyMZGJo6#10p3@MV*U42|YFa=isX8G5wQv>m)_BXw>@@^G47dcKJUo
zJT5K!BPx;BinQ+Z$F6n|_Wz{)S&sgytFC^z^E>I73iQ35_UC&uxO8a>Vx*fVxoAJo
z{t|v5uGK?_ds8F<*k9zZ7RZ(-ct?~>+1N(ftB|tr-T!5~=GNYTI58EZa9}|K;yM-L
z3&M`a{!T=-{Z%95@VX|45?RT*gu+&2%vI@=QZapmow3oGo%Yv-<TQlEWP%UrI`jk%
zMd30^Dhc#&>eEy9OeUyRaVXZCjFQUi5mt{l*9NU?W|f7mhzY4==j?BDu}^R3JhN@v
zCn-AxB(9BkblycoWDeT?E^*J1W1RRjaGJ@dsiWVwzt1_J8l~R0)2Ov2_g+wKJ>wq=
zeHpVJ5;HUrIYw}|lTLq8i_U<F1C{L`yU`PY42)_5jO#3+8T7=}lEr0&+k5e#{ZkI4
zw{6?#gwx2tpRGO0u&8?XKj-4P@+J2<KSkjyIYV&T{w2Q#i4kIOTJhlA{guLV*#5N}
z1~1}eDr+LPank9jL8mi(57YB+-4j%(UySU80}Hfd1<cYBp`Ek#@7;oc$+0NW+=VtN
z7_h^m-uh8D(7KmsQCy|CmH((GtIU$>gDI{>mP-O#+mjIk<!MtuWXL-r0Bn~3*=BJi
zF#%31@pbzY!85qB?7tF8_dRmW6<0V7sd-aKtM=dZfl1u56&`nBNfN6kCZ_Cv64h5<
z{SN16VhE*3Ni78HnC(lL2tC7d=e@x)k2C}Y`>1{KP|R&gkEzze?n|6fWgoq-5vyu5
ziVUz4BFG$fdSabs0pFZQtw06$Diiu+_7#^5m8!RmL-YV;oXuw9{x3at@sw&F!972A
zU+~+Jsh!Txn88#T_oMtB=Y7hd__%$=E?3^R&S?lwXvRJJ`0OAPk}xS6c|SqUvaE9!
zHsg`T{RxGQEUOh;o&!rR@c@VAg+UZc+$SbF;vB}kPAeWLa!lvH*fXM}eNxvW^SxG0
z+e7KeNiNnmuupC;>W%a!$wZQd=>T&dA>fE|Prb2$eagP#zC;L&PdH6|CN8<YeQFMC
zbgK%OJFOU2rY1M51ZtmsTFuK+o|w{HnkwlpO39GE3m50@)AQL2yD5NmgLkcsc%1Ds
z8WoW$s;X#*7tT4b5;f#~3pqPhQ-+b8x6j-cn&7AkPu{U*{W=Gb7#UE<UxsM=tPXqR
zU<VhK8Y2f0J;fSG(&w>qWqo!c;$B6<@jLo+Mza3}-GHKSX$F8SaUIbwME2h{K4G7e
zgH!$9DHTn007=QLll?Ii{1q@u#BC_~q<wCO>x?Vujw*9{ayhP}j*@*|qtF{}Ugk6k
zMLO)Q^Z`j1xI&yA`t$3dd^1iC{cVTy(i7j8kzO(rp0F>-M^g!txv$2;?wRSW_J!Tx
zbXw|C^@g50x_~1-5v^5EB|I2k)V*32`c-#F^D5!5^>749m;e+tnPJ^uT%Qv>P2yhg
zTt#1|U7dc3OjB5jz9f;-(+YpwX$7I~W1o}-Ax|w)IleDl(i8QPVrNz57J5=Ly7%5S
z_GRq@LyWBS4=DsrPZ2^1gIGnsJSTc#_dQ#@`~Aw7+gBuf^Om(+oZo4*h7_Z!$5+<J
z$s)wYgRUU^2X=p0_yvxkp}&1Ycmz%$a|gogg>|Q-1Ig5~`!AKI>5n@l7wBk1u%Ly4
zqwhs+@o(W=6l#M^_8m~_)2tgUkt^!qCF6i|<@{%Kie`)c_<kN53q{uauWHYzd`jAJ
zO^PovDm`v5Zr`u0B2_T~kfVf$rQTRBSG%&!#IhS!IE{Fg#-Mp;zq6Ngv%>}j8LOT|
z<P<tR`8AaN8t?b)s_x2pRF0-o%gsuUXlHn$Z&!DpoK<!+T9veN9x6{O{W*0o%qb<8
zzO-#A#Jlq~qtq=GjXQ9yObp$m#M;a1-MOoXVs#9>0**eiC++V#hV<q2wd|hZ=X(6&
z5pOVwga}HAc~96i9abb$u%mL_B8+o@nQ$pRuW0xnkxTVYX$OOd-6{BLnx4f5bZX1B
z^<0sY>bBrNbzmu+Z{l#2X}hj_wa%<i(WZS4tR9P88{&x0MC!7w?D|CZefQnx_4CS?
z-Dfu>x*nn$6uK-et-l^t%PCvd#(`ubC%C2oSiNMiaw;p5L1g9B<*gqDPAoi*c!f43
zA8{J0(_bqWSs_JWyRjoed42amV>w69IX#-oH6yB$XgBRkjYQ@RZ5d}#veHLP*bLH1
zM85JDzq4~~?y!UcjZTUK+KPQ2zM6+28mQ1!=&jtoSZX)@9NOk&mM2=oxuZ1u-a`mo
zH@9c@cwU&Y9FPcjHNCG)KEvsRedl*$8zGZ*`iy<`zA*KS>hyt*FFVl0O87;?{Agd(
z2!5>Ji|WV#k~iM7af98GZ@X^Ajn_KAQ<lMNIzHGsqZA2yRZT?NgY#rE{Z%lHLa*t+
zLfBXmvz45UZ^tGTdzD0S@^I~I^UVnGxMi!ca{b12YwXqa7A8&N8WW>nd)^sZ#8Xax
zAR?DvmgJT0f5BGf)GqAaKxs>-6$ogS4u$LATF;hgP(^B5IEjSZd4ksHFC=%F_k4lu
z9joeTu}mRc(C@(FY8fXLXR4yFs;GUyR<|DZTXCDTpveJBPs%Ri)GzEQA!AKa#P-ph
zgZ|i7rx_0o^Ee+)5myFB(6-%{gt2Dlq}R_KlRdk=mKQv%?=))sByTVeUBMo^qs{cj
zm0O%f^4<ldrG`Pw*0v8sG?fh(ZN?i+vR;I>;az-9K1k=sPdSb5L5k{EeW^Pek)f!O
zakAY31X(PpsLFSVB{y4X=clxF-2)G^fgiwNzt@4~x*Ketz%4vZ=H>bZB1m*}xQ~h1
z0VQIoD?m0j=?oj{Su#t=@5kY84lH=hUiSYJ2<0E~*Y4U^!8ehZNa)ohu8n~M)|!09
z3twd$+th{`KBZMzdrvq$x#A+KLY($-du{t37Qf6l1bsO$iI0!m_sA8tsUb_21qC|*
z$cG@;>4+G7$F{k92p%pTB0m?fT)ErECTvT6l?XBH+1TQ-?;}3I5g}Ci#dNSR&k&|-
zcXyX}4rkH71OipDaQYLKL)gCp-}kgp@c|kEb-^5%K7-t_op$fO$iJC+(D{}8?T#5-
zBKEq%R%4Ay#MSRJPEV4K5;_?zes_C)x2yHWCnjf|cFb1p1}VAhTTB$QH?+B6(Zy~V
z`kkbi)}yF<=@F77BsQJ1H||TDG|$BM3Ui3Ku*EY{web<2nja_YligP%m7Xrw93=xB
zSS_?O@njYJNg?h{`3;xehj8}3i%&ak?KhjZ5Wlm^-rS8i62ccv#7>XUxv=}nhwUx-
zn-iC2oZp)0mBfyZ+gs~)S?(j`mrt0BMM94*8$w{&X}!h>uXxFo*ElUr$*PFB>)TM$
z@{Z9F=cjJMP==9txO+CiQ=){0D$VEg#8jn1F8+PMMhcA;;RxNx=hSUVpj@G8T%~pM
zmbbjh`t8f>n`J%>j5$3T1se?nv$ZXV+sLPaS{*=OI@x`z!lruj+Zw4|di3JcV@@;K
z1p3Kje;U8JZLjTFAs8Co(}?0(A67;?d9jrPB4(~KkcaTaa7TD7=Ai;nln#DwGFz7k
zYNPpC8%K9JKZV@U0c@Elz55gMJnb2b=+)v-_o2b0?PzqAWn-e6JFt{%BUEhG#unQ%
zQzNrZyLd{*FKQFpnfPLx+qlVT2*X&VZ9Gwjf06=PDt`hXP$v^g26`x#k>z0%jRPP|
zg*XTu!IDGnlk<-L61S;=JLubFn*?bvAW}6-!fodC==~c=`hHu@qoVemMlBF6y~+|)
z5ROe%kEkW{6IwnxJ@K(=XU;&4!Si6@;dEdkn9;BVPPdPQt`tGv)rD|+f_tPq7l|;C
zWklL&L$;ZAC#0?%zdd|fNPtdHVO=LO1b4Rw3Z^Th8lPO4cVO`bgcR(2(q@&y8ixUD
zcJ0Buhq~dj<5Xvs)5;6Rb!X%(p`g=I+tseyN(~&ej)&M<f#nq3!vT*h{n>gK*(uML
zXrbt*2=b#}{A2yy>$ccKMU*4Rm^g%J#^<DYCki0j!wrUnqB?cYFzdi_sin*n8Tdlj
z3441ji&Xr8v6(H2eXWJeRrwyILDAD0So$M1?6$~KCtf_}lz!c#Bd&TQY`P?$**n@j
zOSUudBb^o|j|%;XnxPqv!W`JwwfC|!Mr;voUtqB=hPRq`EJg8z?B)Q3s?i59NWHV}
z)wIfdOZJij6U!#N9j0rKc4R^f1GKM`%pw-T7kiR`(#@i?q8_WKR`@y0;DiIK&1`A-
zc~@KC-c|ie4^CN9l*Q=}?4lF}CZsUGyV%VE$-hqP3dRi)%ZnDu6~v+B_c^vXZBX5j
zIQI1&&UEr{oaE=ga_=4<-PW73Z)jq`1`3NtjZGjz>W$Z5XYa{3_s3(k(@IW*SwTjx
zPMNZA%v*?eb$-SN<TfIv^rXGFeN>!#7oxfY3q(oOZVrTq)nkRRt=Km;@+jA%>e?(>
z<KFu7G~Inctoi9!;7G{*=I))(u>!HK<8oPqWmPf^W%ey?L!q#$ltrX7KruAsR0zX3
zxnTueI+MMxt)TER=d0OvF1yoHj9!ty!}k8VQd#T4d&N=3jlK1h<`b)s9{l_t=r$*K
zI9D%vUSsoOeo^(Es{+l4&RoUci~H?^ZCuQZ^d1u2YR$GDja44iBT!4&5A9pqh^JXK
zV*{t1){>zKl3iJL-&POJLUp=Hd2-r5=n<W)Shm8xJ!z^p#zw+^R!x>W_Pja!j>V~v
zvJGoH><v}{=#2G{eW<Q$20Mvhz_bgHZnbHpeP>O~($mL+S@20ZEtGyOX&uo=g|Y9d
z$8k;^4!U`!V9tT1_^<(2<@juG`;2{ehcyD~_%zNz;^IA+PFfE!0I?b99B>8RAF}VM
zdvw5y6t~Bbs(Aq%iJq7jsasNgFkIR9CQe6HOrJ=nfdi;5Tglg9V-K=;#=ftuM$Q-`
zBLp9G2jP=+2=JJFf8Cw_6KNYQ0FfTCylbz${$~4uVj3#3@~qQpY$*s4IXp;9=R!hO
z_Ja-e5YlrDo3r~ISlx4m{ZSi%QX!rS2etfTyXzhWdvc-;HQnFoz`A?FL=Q^$v2-vF
zLOF-*@tjU&W9;y1EbLx(<8`(t36ZTat#PyqM+!=aiXOb3vi!EUp_ny|X_W6f;J`$S
zkKFQd+gC$KQL;tn9>u$=fgExA)1#zV@o6G8vO=KE)nnNs6rciII^e)!5Meif$l~Nn
zDCDs1ug#Z~Ow=c@=2aA-Q~tpG5;WvtPBJ@%woVSnb56g8d&@0X+JWx1+A!E+2wnr0
zi!cI7(d7=-MVf47A|bFiuq0(!gF}b4Lmkd!ybBGi7yv!Bt(=1_^_IhSI1zD)1IR9Y
z5S{0P7_vWz=M;D<hIrU<f}>*}X-s1MC^b!{oVKo@D14v8A9_QxcC=ZNu&CZYlZgTj
zF!`u7Nv$kccxaVOCkm?_E5Tj<3WYE)#sp*2zylIC8Sth$5c|=8_BeNlzwVkPC~0bR
zmmN=nAJ)+t{4RotAnCBHryP7Dm%_sCb-kg1(+X517nE{v%1$=)AX~t5mG0=k62I$*
z$?+eTpZbZ02M%SVV1?7JM;&F4L0rI_@MOU<+eRdJ)En#oK4h8EBr-jfB(P@l8s}#`
zrV>L@V2|0;1;eI>ye^f0j{^&}CA6-fW>J--G(A$j0!jU++SiU?*Kk^e73lyPP9|$S
zYp2^l1HU1sJFvEVCFJIeg?bbUvz=*#iONiX<iG*M<Mm)^UCIPV)8E;8R0@;jk6>{?
zml~$iU$|3@opN0XCoRz?gAaIS=xwsR=Lc4-e^@)XHttukRB8SEwR7DwqwE_bbHafo
z0w4&bppt1@h<dg|0L3f#SpGUNanQAE)~vDf$*Zv0B2Cgz$92~(zt%2vSW;q57&X9b
z9tQOkS`q#6%Kbzn$t$9onzHenv*)TumQp1!>2M>V?ugT)=P&Gj%S-Hs+K_Q#!}=nc
zIfW*A^wA4CPGW?rG6cxj54X3$x+5tnlo_2^OHT&o$?d{8PSBsUvqY-5i%lLR0c~`f
z(+ssM>WN(HP?Ra0M;DGeupH@{SzG83j_eC2Fd#ax+_7(1ah+Xm4^U_@OPrd}4otiB
z#1NUc1N)H<B}bqgAwwn)<zv!Q#JDI*FEKjXhr2ronu!m?{2sJLPkgvMvQchEOfwW;
ziCZv}aHsxg8$4IBgQ~EQ14{u;p2&mS?8owBsgyhI^_#kM6ZYfX6v<^Gc}Q$|-^F>S
zCz+mtFBp{+<IMZskJ!A&exlxb5CK8vWHeO=CX6r!(ueFP+p|O0oyUeD!!^N%PES00
zse&d&B14LvwV!HJxpi!Y#4VaN6y-4J^-AFB!k+A>yP?cOr%DgM;0>l^6m8l_mi9C0
zdI+kkS|_CBcThHv=lK7EHW2&S4yk+5pv1MGa$ur}tsE}C-F~hvkeyD&olYZY0m}s=
zDGn~u6`8T0Pm|*-uX26~M+>{BH*r?_FVs`CI8<tzkVFR-Pf3G+@f`|)K5oC*28Ea@
z9)PFXfoTh-mW%yTn{?EliP?eEF4mG^LNd$$a(g)924TePz^XDJ#}+-gS+5)$p5Vk`
z`;~S(Dx3ntp1b_0H<%Pq)s=Y;N5-$#^hl9P4M_7StOR3l`t!k(D)uZqZNJtCgU?95
zpg<QfQ+jfy5#|&|ty9wHBx#86KvKV+$XnGLf1o#TTKRopHYM;GVc<6!4%SYoP>It{
zE%CO|F^)(2&DO*UckUXenM_rF5I#o{Z19%_%9ifSW3iKA$Duup<N)*+1q%2p1W0L<
z*IP&D>2Ls@b}DU0&Sr6{eK2mGqzT6XC&%`0cT*^r0$&AEfh9?d_3*I#PJRKQm3VN6
z`%XLg&}}o*_Pcd^`og31m^WBzpkj<i?e~iNalq%e(+E6OVW5aR{{6zV#95K*lk5OG
zYI#2stvclN$31dHP%8@#f5iTvur-fGO}070C%9TF6g~Qa!5C3gM1*z(Ath@ctl?R%
zdpa~Wb{eNY=LBCcmbgEv%VnkzCy6+!60&rLB>k`E;jj3A_@Di8n{3n^6)jfTk=RC@
zp7_}0a`N`<u|H{(a;w4wXvGzXeP}ugzd2%m+E4_?l`<-`NrQ=pDj}EbbNjP)FNnp8
zdznMH9as>$g!FeZ3W?N+_+oD`wkPb*8>$gnQ05RUW&cGacY4a^D?-b}qVzH^&Hng{
zdZz+9iYOge@)fE0JW~Z%!qj4aSv@J;WQ1TYNP0A}!WP^9UsX%Rc%E@u9Zuvn#3F5f
z-Ea&y*AZaa1?TkJ>8Y!kQ;3J-o$POtN-3>PXb+8Ai;HOvdcYC;+a|$|WQ}NI8~_Ct
zGW1NzR0`!*9&g|ef+b?u{;p{N6rku&an!K`Nl^lV4i6XpO3j?~!bR-=eVh9YYwq3b
zUF-JIi3jW-+B5H`90w{_&w<2Tk_-cFM3H)C{)CEoz`*A0AM-0I9=v1VG~z4uWl{$f
zp&#<8{ZsoScwcO&kf$Us8T*y3CG~&axvKWUMSDsXcZw)(`{$Y>rDdOYTFE<>FTdXY
zB^!jGk`F|qrbDr?VHvQ0ZFr`fdY|k>m=7QcNkhHS`|aOq=)$=i<CJ-0P-5GyCq={<
zNcPyjw|iL6%E~+);qE8Uuv-tsWMQN-Fi6By4Wj?3PfV|CNDM(;GL}jlq(4w7f2`&(
z&nA|{@t<|4?op`uV)Y5=jd#*7#=|}PF3u0^zZwqoSa`W|o;nyeE9#fnCj0NkuWqMW
z?6lLYDO5(P)VipdiLLK{62m>tRUbHwDh)(@4(g{8!_4N2$f1lhxM1j%ozUrt3DNVU
zgPmZb{HP@`R+t262Uc5TS0YMdh`EsK|D)$(#wNCo4mrPTq*;F<WKtOVLHn4wz{KbP
z1r!C4W|D9O{IHOi;KiCk7wlteAmr0U5UeI0m;{fN8`s#!wXb5OYul!fpvJ`3Mvv$x
zj&J+;xn}T9IatQ^G?8Uala@qDP0qC{k+}1I`-FTA@|yEAMky4D3F{zu2^}IFMr5DZ
zp-VN-a{vWX#$JD<zZEOMh<(!Bhu_S@7v8(D`~TcO=PiMJS<qBdRDVTyPuV9|4`yJ=
z69i9Ja*@*`bPaDO=>1dXV(sEpr07pI7IyELnX*sq?vKI+l(9hUk8OpX6tyP%lQ<R!
z_GxY6QfpxAf6{@yjF-Ip|Hgmz>2sw|5Ui&Y)oE*Az+gitfPF^mT}gRl$cc86ewD;z
z@~Fs~7s;rJ!ZhNvbF6)eAM;rItOnRJi(P`?Q~-fN1(vb>!BUkiAnEM0+h;-eN_m$h
z8dzQ(8Z73(uoFCkQSb%TBLWL6R;X}z8J(V3bS)JB@Wl)EIf?g$-RoDabsBL&(kZG3
zp198~m=2w*U>|R=a{M5J?YGZs11gWL=siq9PEYMZCm>X90%A2HXGsiESp+xM=eLy}
z>`5hDB(ErLHMuVRnuF1t=`UzkhYpt(8l3GS4J8?Q>rbxwzZOFM!ftUXqEfrLqywES
zAEfK_2YSn>9Wqa>Tl=CmZ)_QoaUFn=wy>LA)2hE=UtCB_C$%1Y&KoQ@;P4RGmn36a
z*xjF;*fr}k1OE!*V)CUu+LzWyatD&Du1fYgJ^5k^Ftgq{q+?)T)@A`MbbsM5++t2o
zJUeAN3C9QQ%je2w*c%~~2ASP~lq+PC4ROl8qV7_(k3tU4<A(>Pnc#%i5#IQfg-Pu{
ziV#<MjZCtOGCWRyU=oIa&<!e(lw`FzWiPB}OL~#@dN1ZtV8QgzzZ9Aq^pU@l>Ak4C
z=RD5|Dsf;M&!wa6w-@Y+HgkyTAxnlZp~?~TG`6qIE9NxKIBD=dZeP_f1f|O)qeCE;
zz+l2D;l8-8TS6P&V9DgN<iM55-bh`Ta()-~<q^woc}e#zlxgSL+V2`nD~p^cISB2l
zj#d{Hj5$_@%(5pJ_Bj1HXPg;J?GU@V%_=l8t^7p?5*T{y;+d(wy|kVuvgDhhoP)?w
zn+OhNr@z>$3ea+J@yl{kddaI_==@A8^CYnc@`~8Y8w0sx)fVp>>t=S?HLYjaie*lt
z$Uxhk6p>bw_!WsJcq>I+J2V{&+cScTUEA14I`3nV%0I9#gwGbAz*YLh)zn>^+G*F#
zl@)lfogk?Qrv#Kqx0(-91<8CyLb<*{G0h2_`5HijQdCGxvOZ<LnKMNPD}?a8-Oy$Z
zTY>Cj$C1;3N%&lkmweWi6~Yx9c;0Cxua*yy%kvyXl{s2oJ!#*OWQ=f#s&>v>k4#^B
zh0dD@Gn(Dlp1pA5@}o{md+5q>LW^!{*rTqb%n?}b9aw#eh&tVYX;TaroUO<S&14f?
ziG2qWq$*;FjMK}H+0ETk7R_U_!7K2PEJUY2NpOZRfw2*LWjB&T8a?9O0bhivoKX4_
zzPg^f)R;nCuEb{UdD7`m%!O?PstIQOB+lutX`B=LCy_`rlV-Kyl2&cE)Cl$?aElF4
zEXIKaF7%u#xI#}BbkZ5augYiQm&K3xMyDBwh(lt-<3?y~qI>g}O)H(>`OeMTdy}J_
zobu|Nnc1j{;Iv}fli7&eSH__zZ7BvUJ6vzz7tF3~<3uf!-&fvKU_~$5mK47$wUYU>
zTN|G5m>E(!F>THMW`ZYdRl^9Bjm}6jMlXR24OD+^b^9Xth73S5eKeR}pfo{}6>UwM
zZg@780MH&IMw98z-I0DRwuRldPLa5NTU%_x{TM||1`ep$-3jE?)VFT0uga}GDLNRT
z(^K;zKZgdQuU;%?a7T&K%@$6zdHNS4xtCn)Tq{bXQv5?io{rnvS}k)(NEe(|N830*
zF*h8x*R-)Jpr5x>2F)8Rv?R%5Y*%+S<iUYd(d*d4yaP)T7>rU(41?FYh6|#cXSS(`
zo&$>6Y^Rqg{w65d*0*^MG>aWqNg!YB$%4Ziyd6CAgl%XqAPX+*lsTYOdspAJ%I?a>
zsLb?EBVJ7=oDBUeXc_VJF&mele6c`hr`vh!*JQLR3{pho5>XAA*A_BV+$3G4M;%y>
z<8`Aw+tj|Tc-|4dAO(W12N9)3zlhg7YnvO{Ve=NjCByAFj?-V<K;dL*`hbtXN4M0S
z&!Y-*ArLr1&!s=<k*if9MMkH)+kA!{G_0Y%2&=*wD)!p$?F?xL99U3ny0+4;T#3{0
zj<a@en<dV&Lotz<2uOta(wFBydtD=)Eu%w<YH$GQxnwIM^H;&ZukT)+87Ihe4y=$M
z+nhE+Pyxc`><x9b0`56(5LQE_*Tyj7OyX&KV|UGSVdy_~@t^~XkIATZwg@B(38``o
zk@aWyb@!<BYe>`%Ebd8r1Ph!AlfMMX9y#!7dsE(nP>hPCgY3Zou-qhWRp8>Zy}5k}
zer-j7p6X<MdeUUNfm*nIdrRIJ`TZ(uK|4?o92Az7B{H!-+glfVAj>5b;`HQ7#pnKz
zR%);Du+=L!u60^LAo@D_qaZySDwrf&=@rCBJFsF^@(_t^l?0weDG_OGxCS6oib(WK
zY~IKZPEUT5su>P!q>YE!CWWK#07|l&h@)=$*01-=t*&}3m;oGEUNtm9uBg6aYa2OM
zK^f8%+6yd38FoijY|4$@h;3WLJTj-u2}*BIR!mQ(5<W09rZUd3ku2h}y-_CtSN&n8
zZ{UEghN4+w%J*TTjc^(jRMU2gfj#1Qb!><ziQUCUQ;@uh1F!fglgHccxBI)n=qwQk
zDx(gFHlim9lSjxLwnqvEieVi0!ffy8CJ0%F*YVeZC4zQRo|SS0BR19`(}=^IY&h;f
zVzTAcP$2Qs?87@78qv6AUw9`md%(KMma9^xDoy!?e3y;a-hwJc_y?_^%6T}B3UMzM
znVHC4dexeZTb$qV%oM)U6s{mAoz{fk#U`6Tx_uUtMpQb+YX`K5gL<gwnSFa(TTGZ2
z#t3Dho%D@;0k6bPo9aI3AU1IvOvIrAyH+#SwBm5&Aj3)+3Zo#@pBXI*raT^Yeqx(#
zkPp9&dgL>BX=WXO<c@{iS6yY(i3Q?vq-s=)S0JI=Okqf=r-!_?IAEUT^+4fEEG%8B
zawLm<>B)JAB`DEE#Z1#4BA4xx_FzXbO3>B9gf4InA9Z>XPirVtG%<<0Wmj%WDqS_=
zG;)=*A7(Y&Ih$>SP4?n6DHR-V4Cuu{*wCG7s$fFoEvlT-g*ZS1@|B&%((_O~l$d<j
zM0l7Sn5g@n346E<dN}a-nVS%1G1(0ze7wCrr3GQ5Yw%!fBTi55rZ?QQ%pR#*QKCbg
zl|{0_WcJYJPA;pxqkaD-W(u?Rhy$tZ0$qw}R_(tO0lqHh2DKlzl+#MgO$-muOsZ^?
zHwxQGji2EiPWuIXQ?K}mVn1ZkCw1H2*@(6`QhowSpg`OA%HIR8dBPs;V6tO}?c5tI
zYK!nh#AACb(YA8aM(1ZTPR=hQ@9te~SZh^lZQ!(Jn7xT$w0GyvHS1pQ{1kz%?sLTb
z*XKJivW(4ehB7<IhyznUC!7uJ8`=~5Gt*?_JM9>Ea`%v(>^&V!F)E1f?8LwdwXy&!
zIzg6kyu$X49phMX6fKOc%`N?*8l;Gm>WFq3u?>kk^6>1n_by=-I@9A6<dt=_i7r6(
zYyFH96JImxm6;fM;F}ua*6Da6ry0Wy+i9|U(e}lI_RS3e+cIi~c2+q~r5HO`6fXUi
zB$$QW!!ujAI;~`Pj7z#iS6;~8S3POC){>ERL+V`Xk%}u_LZ)0Pxclw>?P=^)awseL
zFH-ZIp0pk!w8{UV!)Zz>mR}>*(JA{tlJCmNNNm&LxIkij!mF$~&vCb$*=Oy8)uS=B
zIL=&iT@NWUN(tA|^5NmKZ*91wf&;dBnp!QKeJoK%+wiy5WC;<c0v%z!)vIBx&v@$>
zx0BhQ!X{i5wG|<pv-a%`xs(B(c3L7&(nbgN9UbW@9z+I+Jq}FFfh0ay+=trtppwYH
zr>w->&FQHPbg{y=LewSAuzKw~yFsgbpuQ9eg*S5c6i)ABPQP9=J%Ku9-&GesFJF|M
zOG!9qFqP4N4;TNR@iW+WH~g&Y%}hD1n%q1NmoRYxCskqWdkS0f)XQQ*OobN=1~9Xu
zH#HmOYrnTmC@6%QY32j*^(yG`*_a+^+P<%QAW^J*s-jSUN&gKj9klPSE8yr6v4n5W
z8w@L>f+YwXJ8wTw?@;WkYDmi<z&44Slb(*_Kpg1hhwTR&|KChfD2-&LNIxRx2;^<k
zj=-*IPZ@~>7vJ4>=d0ghk2mK8sYlziL+e})fSnK`kw$u5D6~9=8s4_2AsfraxYH7c
zU9n=B?d=A7HY`jmKyNT90ZLAN0x@`B+f9&FI1GVTc92_V#z<g5L!8UUKK%CkoS%XH
zPA-YPfmBM{-+g$hft1kj1{0+##*kPfo9~bjJRucG{1=Cvo<N;~?GHuB{Xsj>7~aN-
z{xlV7Y3mEGy?L1(tc{_^5mWpY_<kx|A8Htx<d`|9^}P8awayMNfhn4j>@{D6DefWS
zv+YP5X@#G3I+N2=Pkn|>UZ%Q(cC?9g$gP~lIqm=gRhe6@Xjojo%D=H=)x-KkLQb2A
z!o&|aJta?_1Yy#2RYFro!06!0V|KiuL$c)rF}d}m)TWiDSs{9MqI>%@Z0QQ<my+`w
z;tD-6c)bxLgA^Y-*=UanQWOYJEfAo1O4JOoClcEcp3HAy`>JcMuqW$bDG|hO=)mH`
zGXhU)g5(E+2kfc(Jnh%oI~7nMexEf6PI&7Ve^cm{T$74@ky%x7W=|)v;>?Is!ibax
zlB)P7{4!*H!QW5S>$ObIGL>^M%>78C7rc5X9XmQu{T|L{rzWEAIo(#lW;Rfnha5->
zb(?3W>`bGnRcM4xE4$XM-C$=M0>Wpi!Ol6bK!6k~Nxj%Gv>OnH@=PJ1DWbBlFDY#0
z^yj1BFfn20+PAT}Ao^)kfa=Lo0>lm^*byRW&o=UtZF~m?dH^xa5_1)kr=aKaZR(NS
zw11q$?f^_17+d3ZyHJnU)+iJ7G5BZ-`t(S&!%dCJbYRbQ6PXN2vZ|&?QE~EL(k}`j
zvfF;BeFdqVME~^>1OdssK<xWAjB$PY;d*Us?W~bx?3{V(^u$NZA(F}&1>3fZIpBre
zw{38KSAV2GBXgfY%T1VwOs+ywk%%g4*O6NgBz4Lby0u0HT|QRElzdNa$@Isw(;cST
z3485w!CHt1>a+t(sdo^y)_$bD3cezjhjt-gv8*$eG~>xoJCSM9QSjjoo#_aUFakDg
z1wRvx=k&*dd+AH;N83EF-LPfN?Q1qUfOvNLxbm+AnG8lhVn0^Ml{_Xtqw+5G_k_2e
z_*_P+G_3e*KVGvvtsvcj(~MW75Zsc37On9=(P8Bj^gIV9_c*ZRWbhZw0+F1&HcsMU
z!4rgA_h>iFBb&}Ru%hlX$wVj=Tf-CfQyqoKXaQ0}p}{-j^u#Pl3_|qAcz(owx~>kH
zS#O1QLN|4KQukgnNkLZ%y8TQ7DDO8r2;V*cp?dj^H`>o8p%I(0&G{X#nb%I0@|XQw
zW5&ZAgt~3kX(usaB}1+Q4JYj98*yMu8W}iEEjD!XzWqXBNNSW^SE_D-yue?%a0$2V
zFxVBIq+dB6`l2KDi`}f~43#n&BdtczFMOVUt+G|P%SOj%`u0o7TR823fzZ%rPChhj
zzue(ABbTFOj99bNQ;e>+)qbU$Fde)t3YvFdHDU6*X;BdQC+DC2YQtBof}lzVrpHlk
zmGVEoR-;$=F3S_eNL8355bx7zKu&+Hs-5Kj*stfka(j(CjX-o(Z<aZ0&VHkYauC^4
z^hQNzh*{b<Jwb^X0s#%f+R8=&$se@eOjK;28FGHs_bBRoNgYl5t?rdFE@XZlFnv#2
z34go$9#uOl6^|%6$$P5mLjM(9oao^J5o{=hDpVvLONVU=I7iso@3awv6`67M$`%*_
zA9VU-n7MG1Cg4w`ZvAfEsla^E)S=iqae6W$s){)MqKPp)9)$8t+3&TD?Li8sIISdL
z)GgF0rud}&e!c6!Mal13N)cleyBs|^Lw9o6<*fZdcXLFLiHZ9n>}*6G>Nxm6Z2LhS
z#ft(6D4(P(t4aH#wn!--#YM9YAj5W)-mk=NZP|a^<|=ecwn9ZFK(+^*9?i?dj4Jp3
zNexSisET1J&bNb!IkVr$k@lyZq7Ti0xy2|(rN${34pV}2RZ^a4uGZn76->r_)pdf+
zI<S}_$z}6gm&ZWKi}>etwQNw*IMD?hSSZfLAB%SLbmzmdc`J(SFPe-p(x+sp3@Sjh
zh_}y-+F#ZSlmDlVpB_MZYO$ysTI%qB)dm5%Nw&gx=D>=mlui?lKn8yH;(q(<#^)3!
zkJ3UruO#+gZm_gwBg)v{R8Q_d`$Nr$oMLZ1<dKHbt-2B76mEap?qQ2p@f+DPW0X!$
z8oCrIKv}0*`@8O2_J+#He(JztZ+z3s?C<MQZJ&vvfCxFKeX&QT^|TFY>pf}z&|V)o
zJbBPU@{QJ$EMI#G23P(*VgFdGOnewrAtp(F$jmmSk<g#(D2x+PxX=En5p~2aAh`la
z65+-;9Sl+ozmENLyL1exPq|S5Y6_Kwrnl~~e`)VPCQ=OU1ttcVl--5hSQH<!f6ZmE
zu=~d4%bi9ME?K0>{_Nk{Bf<y32VzHPFfDXtmSW_!f6q<phGo}Y>-^SIeM|ml|51-<
zT$j^y9av5qCvTCyNEqIK=K6()$rBt7mq`aG$vw`|4By49{a2x69$urhp*V;XKU+^z
zJ5{qWLUZlE8?Qt?S6Q>jfsEghL?3th6DODCmH(eb3JC>6BAVotvrf;|45w%#a|jgQ
zrv%t6vKAac(k*XMwLse+d%1#ZoZq=q-gfr}`>6fFWEG;qQA@K<JL6e!6AaIwa`dTS
zVIRFe=U-$fwvh$8h~Ia5Qgp&ltd~X6F3cnBNEfM{Y9CW~Vwa#lqI3&Q^u#Pna)+8&
ze|zm?>*}-}q+>n_oSq`9V-exUj&2{<#(L=kaBO9+egHlWG{vU%$n2zU*vHop=b*(p
zv<N)POPqX?ev!$b-&aJBeM0*pu{)s;k`*y1+(VJmtv_w5p+wQ6I47Ck=Ij$|r1}Ph
z<zSU`V6lGUPz+%yLRC2b%s#2Eo&#sveq~q!mRCd3u~>tB>umN<uCe9mE1Czz#*xtg
zPEl^0eogJA$1XoMVV}}a#%?Xu^(57sfd%*O3MH_RS)-URByuaIK-~LN+W<lmp3_E^
z%h2hmX_(V=gT=?~(+Z}H7Q_uuamIm#1krZ=C0+t#8u?Ha$SLk{)INQGFvXj$yV3cX
zzTBOh^o5h{GrGCRxX#Qs(8H1{=nqWk6I5LH9Fm=c(-ZcYIjIY~ABY-TG{tylMh5m-
zt!H{x!ZOX0{Ueu1-w&?DB{RYNv%9Z5BAI<b;e^0a2I!9?1=tJrr`>Khqj<_`DW$y8
zqiB-<35m^byYv|H_M;aMIBgAS`AsYAbMxnpi5<?*_@c5m#MZ8Wk~#alZc@@>E$wSR
z2%L9%l8N2gLyACVw$IN2PRvXXRFauy%$i76CPJP4{RJINM=)rs+v^RcdBGbg6T5w3
zejKHJSKPe9Y2}xrf)HN)MICYo*b>@cPQZf1mgMI$Cz4T1kPT<s7Z=VB6F_ssX{FGF
zb5$GYsF==|)CVUG$-JrkOA-q`@jV)+Xz2oG&@ZhAN-DkJw35ed9NlJLR+md`6_p^`
zIIzT7e;h7>@5a8ou1e)vMvn^)EX8qJ6+#!JTZtt075mFqjeu0{W1$xVgNBw|O-lkT
zgP|)Ai#u|Du@`CWN_dUXCHu<u0hpNgI<2B?o%xi$bSeD%?S*w6$^juowBkCcBT^OQ
zGeXEFR^mnVJt=aFr1v>6sdNjwAAY%A(NMH(#d5F41Ysq8`>J*))F%axj4}sS%S4J-
zsw{Fd=OTXAUfgCdJtfrWLFNOPbjN2Za*)4vWwK{Z!5E3?Vh0cxDcL4HU3={%-PE6f
zv!_EnxTe#S^LFX|;P-t9MOP(0Hr&1bRxAv(H1W5sUAJblU7ephq!46&Cq*cE;idI_
z4M|4LZs|KcdD*g?m)pzQ_z5o-``3v^!199S%dWGR*9933$7)2C1tuJ?<orgvrhQu|
zdc;~_-f1u`!<1T{wO7=2I(J5j3p<GeD^~rHdCRGgW!(6+-3ZQMIG|P<{&--yn%0iv
zK9A!k2X@{556gFa+pzOXU$Bn7eb%ne(O&uxZ{+uqWyKt$8{g1oB5Vm_+c>Q+EBlU~
zxLV6kajeQkTb6UNu=_PL*uQDz^z$TG3=;SSTi(&pg)j#q_VB&(S!l&P>-5(>N^4d0
z&b-~&Rw2_MXLH!1u)0bGMGklR<DPl=q&$DnrQOt|0Je2Pg|<0>T*?uvB|8RNMeuq3
zK5Z+SFfpvkz*Jm8oE>qXX|g4_W}1v()X<yjZK7AzZy<xqfyFA+=BG^}He88>_R3`K
zo9|lf{7lN(uyU1sbv;EggTf9m7CNxvHl1*S6s42jWcNR1UsG>kuZN$styST;^u(Qf
z6rUwb%I>2rI<RH8w711pkG6{_0IU!aSiPvE{-w&stGa#c%%u;q)|Ax!0sJJuwQqgx
zRoB?pCh=g48aTfb1;NrHFejrY?A7)15b=ur0t;6c><Y1FWer6G7e8Oz?9&)Se=)<w
zoTW6<-&)V^YK_5SfyIX<3vLHe5?9HS<%U|--J=qay6S=hOQDes^g&61tGkOhViiK+
zoMQ(TgXRb(R>qn}8pESg!+1E5tO0doATOajX!?7|ZfiI|zCg^wn|0WMwQ-`@mK7$x
zU3Q<7c6$?saL`aA&1sAE5ewT2vEPxq@enkRamHyjG(jKQhZJ6faA#|)r%9fR%1Y~z
z6s~X-ardAsdrb{5o*i#c(VPmECSQa8T6LUsuO76oI~z{$m@7<*Q3I;WNJ}hoTvxDU
zciP#HI<UY7?1aZ@(+OJy$cG+C<6B?%$m*yVeLl_`Y#xSG;i9&odyewLq7+mtNTneV
zMB((Of}`U7)cYY^9}C}IZEfC*gRX1bJ~3n)ThFo^Zgv`hVL8Lt-SPbqcFj6JX|HXZ
zj}(9*r<G#T@b-z(Dce+IS2RsU%R8|6UM<rU7bw4kQBomu3KQJiX8(2$q3`c>+E+9C
zU}hc}4*MVi$CAn&#SYRc$$J(~+m?E3R%6fte%67dhPjF(GwtqNzXV+0?)=VA3r3Go
z201t)Q0<<MtaJ(=JTv4)9pFw+uA0pl3aI~M_cj2=FarDr4j|vhNu5*nx`qd4FP?<B
z2rM^rI4mLN7*1Z_Mzw3y`5mK@VMYnA2PJ$+ld(5+&w4ss1<zbO<_$K-Iu~m`3<G{+
z8`7QQ6I$60AX)lOiLg_ONxH9%l9;!hPOFxY>@V6QG}Q_;d{fO?YL8O6B574o0@53t
z{<u@yfYQIDf9GO)bEC>dM@cNH45&Tm^yha*m_*nt>(002*k*9CGk99DnzE!R&%Suf
z-dZpP%2q)kZ!qm`a>A>^Qm<i}im}t-Hw2QiGciq`!B7($5Q%%#O;BNR2N-0Ew1F;~
zl9IfndHT&^8*UHfR1rcrcMcp#O_^K)DGi6{WFu|xa3;x+$Jx1n#q)#cgTE;DrmOUB
z{TyQN{!8yd)1j5?YJmLMx2;Jjv;61$PFjS9bPkPe>r~Jzb))e9B@!#|?_$uMYqh7{
zeBBMUJ%9G5C!L>xY$WwAylu3`yMz-mk;F{}%R1on=LCgU68QvDy}x@-9@U{^fq^9z
zA_PB>%I=OvBeK{O^`uu_D#-z-zvwbje_`zS4zL*Rv$5n+>j{A9JB^SVr3cHwE{99H
zv=;Gt%yuSTF*Kn!IIYyXv4BYf7;l83>YPq1zDeq%wr3ctO|+R9=}%AWa+*md(3=c|
z$|p`<6lr!gS+Ao8`^B?PD;}6)h0He3+S}R#*Y*4Od}zicIvZCu5b^|Vs=Xw{wpduQ
z{R#~*ttXuRxQFEvv<mUtAW4KM%n|2zAW*&%Wr>Q)M1?18y3NK=9O<zYLl_{TBo01D
zhPIjKy*wqylyc&)(=QUb@{yO>1NEL9p|nl~81I4o($n-@MF0@GK$+JE+n00D(bx{B
zDgEKP8*jSac6FHN=tAi1AhIT~K(%&kX^?4&)^8#Fyv^1<ngqp9%R&IG6JybKCEz?%
zkCAf>-7E~_;VRPXq43e4L&Xp{I>EWYh`;u5M<?5GNN#i5wbtcGk_}6aEM?HYy_>%$
zL$1`;3oJ&yKQn3-pvKHDZZwZHys^9L7yzBX6gYHx^3Dic;5}jQRm`Nsx1;usHV)Mt
zjE_+YnXD1?#A?VW4Xwwr9Wh#@V6J+4;kbQW<6_$=gR0|a9Z0+}g-BsptUfA<u@yaK
zUfe|}6E8a>Ttkasfzn9<iE2}a(Z{$}s@H2@WADs`I)Uu5-Dwo^B=0U}&mFKw8zF~c
znl>k3B8!=^-oPH~`embzXblG@dL=lQ0Nvqfdsn*#X06&wk6%3Hz~ZA)eV4R>Ow5jy
zfh(u&-E~i@rXr`~4lLP;<k)VCWhG0aXauO**VpI>`*nV6B%)@V0^95x8h!W1w#^U+
z+g^g^BW+FP?%&gdtaVhWqmj($e&wso+Ben>SZf2ZoVMt|VpJ%Iz%uts{xW&Z6f_XZ
zkG;41UhT)}l~kE;N!x^PI{n2_RdgwhF%o2t+Bdb`Wb0Uem!ggVh__%*9Nosrsxt$G
z#&7O!7ei(Vp-xDGe#Yq$Ggvv&vv0|tcTG+@KjU?>2jNDBRqFI4bYFX4N8hsINUW3O
zzF0(#{@nLoOEh`k-d~p&gTa&(W|(Hhb^3F#w~__YCk$$Q)IN|{ean6M&uJGeDEvYj
zdmUxRz@)K!FlTgPgaX$54#t|Y=RDH(Ui;R9Ni~&5gL3b{;+ab~a{`RT8o271eOvns
zHVqPPXr}lvG&4J}Z*Molx2h@|w{?4%)1Ta3A?Yz$r|df#sZ305>kW1~?Gz?=j$)XV
zRqsQM^W6m>%TO^usR2;kV@hcb!G*uGJ!*J@bM?lZwx)31=!|_=Qs;<5X1p4dv)#GV
zzPqD<2o_=evn+6=KkoFzXp`e8Z0?AC&*Ce{3hq1YvNc_Ivwd$pDh&eiRpcqqPdGjG
z;Yx*()Op;#uV7>qMywG8<G_-<Sn;T=u@lQ0de*+bP#b$Ow4a53bYgv0o}xeQQ~<g{
zlh3odgSPAk>Q0fhB=6t>FgrFIW_*v+uTOZ>!;jqeX8XbCJw~stq9RUDF6l5Z?y}uA
ze3hBouTQ#o(1FF0PNGgVqgYhslCW3U<4Kl1>JdA?lYS<9BO|t_-Y$+EM&VEiwePf^
zuxN#sMp599z3puTIo9V;C8wE_h5(`DA{}73ufYPV+8gJ*!SIVCXed@=-sakMSk1-F
z@xVH;+;*v@NC<;0kNdlQ<5<}1U{^dK&m17!Sql-VL`-$Rb8W8rTjRXh`Nor0-L-z>
z$_=k~0P3IMWadc@12~Y|%tO1JUy1GqN2%8`Wd}?8l18-OX~kE{G$4-;{L~KBY9}?w
z`JGsZjPek4?Qo-m#mH>vNEtM--8=m`RL&-u*l9;PO3<N(S`o@Yzz5*;h+rrgIx=cU
z+c*_VfvteHI<HY(IXjjM@WvaKIY0C1QS`))w|S?)(zMe`;#C+>ZBQYJ*oo?iAwp%c
z$V+o4thBAXsqF(hS(l0bOH^i`$^FCWNjy?|XzSL2JyDpMNAXX~(GdrhEGHd1M9e;!
zxLga{b$*MBZ0>Ehr}F2}@QCwMd-*m3H0|m9y|DYiN$0l+I`-s~cB+jFZ4z-+BmsHI
zc|lKd1kz+tmn8h!=^X2>U3|d_#6;S04J-WeV|>pkc=nkbtCDyg=&?6XJCNK3V7l78
zbXd!xnsC9+Hq7rFB~;UC)<d7xyv7=OCNCMn+e%id)6_TIg+55>oO2C3o2l#&1rva1
zE>}(sZ^z?o&$bubF*;0TFPg=}<f~m8XLP<z*@lg)oJOvE2xr(XbkmY4RYFmNN&ieG
z{enH$K9(s~rL(m9V-id$#OY78f(&^Du<Y6o<@EHnj`p12sbcOVSU@L-{&4#`v;^cX
z1Y#hCPnmF~Zeo<@aTPgb7aK=dvsK5~Drz2(zJ)?9=;X|Ok=%|MjDF*o&4YlJFT36r
z+DEVAWOs<6P;dv97rWfBGsOYHG(=vb_|oi0x-}q?E!RK_`1B;<;;|VeRd~!k+y*x}
zK{bmg@c}3&9L_Ic`~sTVkG5}&*q_Ftvyvl`aZZ2SGq<oeP%p#ak3DY~U1T5=MUSpV
zu<8vDE2kGG`QuIZ=<ga%#G;UOz}|qe2(f7o>?dkbQK*xjK|1<I33JVcaLj%(X^bje
z+le9cYEJdHV&~s(Kh-cEJZZvd3wu}H^6-83({09Aaq6nXKmg(h)BqUWs)`gp({K@v
zw@lWOF~3!sqd##G!dwi>ezw8bN0s`)7U012Qp{!p`?=~D>58EdlI{iLSjrydAyZ=c
z`FgjeGlS0a2x~>Ha9)qoFT&cal-_~;LcU=LYVrF0fR5+;#l#@#D_%bzdD9j4OO2kk
z;>V#8L@ms+vm)iF!2RR$wM4kUFL#(%IA#e6@~k;M_4yI%qVOO4m3ms5X47Mm0%JVQ
z>5qFd#vyhM`_=Xw#1I`_qM+SH*g>o>=+}rvzt)tMki?ceHQ2fi@i=vHuX9y=Pl`K;
zLomptzUg1D??Lej<C%8nIgD#3ot_$-0<1LMXYDuYYNb$R9J9joaGg6nx&05evlb(E
z|2OO66mphs_VmP~lYC|GoU`BR5HI~CO*ztB!ii1r4gEPB5F(o}{z?1oM$WJ#Wopun
zk4<D~HxCzD$bJOK+3$4Ip%9m(;5LP*$WDJPD~&<YX+-`aRsL=vd&w~X!C(~SulV4c
zPOgeEO3{}7sPvoPOUhK)@COG@OKVDHK9oshzuy*}e8gg^0CK)$1!7ZRNyL0sSquAv
zdc^@=5iz}Nf538w7*fo~AJ!9ci<gS0V6BK2L4SYB>DM3^D5cH0@?)H&G;4p<eEVEc
zY2XaMs|8lRvi%_tNrI!sYFjj%&24|&&5hC#W9c|BNdu$(QTvl6XotF3SrYfV!?c|K
zkcL&WOEK_afsBYG*&^iWg`eA>*5oX$u^{90F{dYKZ;P^TLc{s9L<z}C*vFk#yg(z9
zIwf(m1)bzUWz4?)c|$D^RX`@?UL06VM=sqA>`JOcI$1bqe^LF~{nAGvzmLxH(Yu~?
z`ZeFAn~x6>i*A3JSRo_9`6&VLA!%jySM_f03Par!2yVc1fsX$hx4&+4yg^k-Xr++2
zVPXe)V1Lt%T#1^Q+R$K%ypuM=ny|lJ{2Ju;me0{8>N)*EzZl>|Tp?@yU14%RT>q5Q
z(wJ6q9LKEveJ9^XKxg=aQ_|2zDMW<h8B6LJ=h~Df24jR0v+w;wzH4FkN{I$hA;E#w
z`r{yb%~KE~lDGY1*Ar*|fNcee$PYnJ%B{EcC__DN|CAVBzw&nHXQYsCv48G9Tm}Hr
z9K{90`4$R|)2}TFR1fT5x>sp$Rep}-d179io}{x)Y|Bh13+G=8JF2z1-)RN@@bV#a
z>0rh=`?of6J22Ket$c(mZ__gz=4k(3H%Wv@0qTHvFl=~=%R5BC_kXm9NA+0=S|=S?
zGQ7?gt`N6#VV0I@=RdoZl_>;R;}dOM)fI4Zu8OA+kq6J=$zfbMA1v?*7RG;dgYE=5
zK=opkR81z6CUd=h&i=bYc#?oZAD}>xA69;+zcBSAFN<_e`=2(GoBO*s-p*+!$!x+>
zkA2N-zTKpmiJTn=b~Q7{45>y_+hJ4FR2-J-zo`lwCHffvaQmow3YD8;eou0I5irgF
z-Q@1rN9XT_-MyakyUAIx5v5<{9PY7?sh-%gd2iN!AWg#QNrLQ+;c^?;$Ib_LRyiRE
zfL42=qN>XXq5R8_YhSmZ1B#<YkOPaWHQRD+$wDM#AFz+F9%Z^J+xr+gO7u8AH8tn8
z@L83z9CtHspU{oqS)6%dIKs~fYzOoNo6K1Cd6s`hgJMDibIMf{AO6Jro|S_E(Kb#i
zX$RXLrnSos?a9C{XCDzN_DR*BwP}){x8K@w=?_BC5nsxqMNG6$ZW!E3I6d5sb}kx{
z4v^ocx9_x1$=eoolO4N_bZ>2)z_bmHjiU06*r(3N{&PDxub-ROZnjUWk5BukK3<0u
zkksh(#6ZGH!r)I}#QOAt3H=%Te!_v(Vo^p<1hMj0L9VfU>@(WLAfZI1(J=f|v&`wK
zH)MtZ1zzkk^HW>UiPMe=m%TwNMnOI~;h)uz@<4B!&i|#|?P;>#EG}>@j8jQK_SuP(
zp&3#Sr=3=eASSZL`+|C7!nB4ZU7fsm+UXHKqXz$J`<%KLoWVR5kafm^Y1vFo>>RSs
z?Vhd0mqH=1_?nclGWR8k=<`|+d{erX>QqIK)1zl{Zt^z!{2GBc>2ar(3yYL$q&@qB
zhEE8U?~UC6SS>FFZ^(YgvI)E37tY7dHHm{~%4x(T$}cMYNhPgw=L!n7FRK3eFn;K-
zHgqM6MB<3kuc53N9|`r*zIeXOtVHU~@&Eoy??WyFnsf`1qIjb2*_Y%OZJZfekGx7N
zFPIoYrB?BN`_i^vSkY35QU+6^Y953_ztFR8<%ZSvWeoxH{?Jc7Dt|sqNiRq2%j>$P
ztl=f~4i+wv4B7vR`C_<tQ+t})KMtU1JxbBW0gx%AX_Xwbuk0SIQ?kWm2<GfiM_mkN
zC0w$hqJ$sOUf5o;u|I{8WZ(dDT8Ael$d9!bEj|ElC-Ftr7oCq0GvxHANqZQbf2b+P
zSLAZvc*jcTXYH{g_oLjy8tkib@*l(oF>o5S4mJ5=Q_tCpJ6PsJ67Ar?VuhtCb!i+#
zCZJ_h)|L6_h20NLOgN1~m}V>$j3|r7lH$SklIpK&yD@LMccfHJzr?GB-H$wSpItRy
zLLO<1Tc@37_N_whxVjs!0t6)zA9rAR!R?#xw3jBKE$p6xwwzYcyVCDsBEs;drN;!>
z%j$75or&ACXF9N2N_pad^kG@}582BLG08H-L>EK4A`KLjKJ8o;_vcmVFOA*Eu4ywK
zrleV?ouk7xF8v@^W0uCmnPsnN{UJ}Pz@x$$<K!8Vw~u)1*X|+vmL!{u2$aeHfL&Yn
zi+&{jcRB?Wg*(z)e=B6E2+etp4=MfMuFExtUAyo6PL>u?-g;wI|FY}bbWvphhey|(
zLGJ6@4Ta%|CxsfMTr`+4s9ME{QnsvJ7hyy47Ch;|QoqCCJY~!43QC5sC&O%!1A&QD
z7ruqkA^Ly!|L&jN*k)O>BkwzKnc}A%ZcZcnv60cEH5UbDNv64JkqX$5;DPZ@MRm|4
z>H5|?ZAIcIj^5#SY9<5KR<xVz_S~NoI7{gtM3%iyPm%Kk#ZK$83=q?HAx{Cz)9H~K
zIf?l9)eQ}IuUfa#Y32RPZo1CCrkj<Rwcr-XFX@TJu2m>$@DLK<Ee*%SvM9y%8BBA)
zk_(OWhX`-9S2Zp(m>EL6pk|Fk44^UlgmTI*6i2$qsc_J~w(Hr0NFleAD3LM}-g-1s
z+CP$$k^5q=o-eB;Jc>*JETBMyy$AZ+M_Kf?GQV*%w)1JH5ns~c8pTg;Etu#~%oYLL
zffa>P2sPG;%IHXPTUA$UEegL1@dg%aFSlULuc$zo#GxPA>gpGbMs;Jv^(9pbo&I=R
z%Aw2|`8f*I*<)+E@tzJ>9#X3VOFJx~r^7pjCU)6v3CS2rRYP_lN$tZ^GsAX!M^7jL
zvrf~XhyNXU|NpT69|Qki!GPV-7ScF|#U97uIgr$rr~3EfZ5i0w<eXSWou9EtYb%xy
zH%*K7nzp<~a4XL`&5~ahGZwVxD^5{s`p)h_64<lgM5rg;dc=t&CuBG{PDWeT81<I^
zJx-&0fs{)z!?}xlB1A;x3D(bl_zD=HY)|1VkMl0XykMyxQTN%oHsw4HWGolShK?I)
z4_DC%#c|^N_pXOPRhgEkOs~He587S1d(GgHaen76FiKb`39=jOd8y?i;7am9VmjgU
zlyDeU1|*)xwPQt1MByehW_xYN1EV?th(G7Ra^qUJ=62gu7eoRnjFs|X4lEh?WqIZ^
zlT$s%Y;$AQv4OXAiKyaCr1E1<e+(+;In{VYHl=?*1ZYZ@ZK<b}GeCAsZCDDhBid!3
z)4!Bjz`G)YsgP)QH}tH+$iH>qw8g5a+0)xOYWFl+(&5LCJI&k~D(G7~OYAasZ(^W_
zJT>Js^o$-pRqUXzD|{gMTz*s<1*isAG_z<Pj#?BUA{?~W*FCCUR<2BMu&iRD&HpKT
zLk&n3pOmPkif8Glr>8UmL?U2VU+s<gHB%F#Xm2~6W`0fB<)i3R?W*_X2i$e{mW_9B
zaau72OYJft`-8&frMqBnYKR<YYacl6+}jzbtT{TR_RVcj>8jVB?F~kB2*w3TGl%Rg
zZQ<ilAOLFM0P<_Fb4M)%#JIQCx+-fhu^c)jO4+F?C?lNc#gOp{=i0!xtN@EvVik)D
zAFuV=JY#Mj?mNxGOxBBv$b|5g50zXvI=;*KUD6Brec*!Au57q5v9)7kGpcdyKx$Pf
z$2fBdgTLgfV}QqOqzzydzUf(~nF7ODkI<@qjXp(5l;RNaRK+A#h0`AcQPc{-v=ZZO
zYe!+?H!!c}#Q?=}ZqXl~D2&J$tiPEtyRCYJnMw}?uZPGArYHGvZ!n;sf8VyZX(`VJ
zyvKpH2mfERy?1z3*Om9bruWJ8-jX<z*lETUGAV-v7#zSDaGc_li*%8$k#rSZ2`neh
ztk}t9YT``KR7{fuq8l)nG6P~fz4zXG?|t6Scdfn8y`Fb|?|(m@2R&!6eeOADuf6u#
zt8Z*|)eM64ZJqK-@`lK9s-yD`!LOd=@%Zwc{s^|%vcVq82QTlXq72@gna=El1CyCx
zqT3m_UCpEN)Yc;ruRo6|I7Z_1$4BVE%{ZblLi;GHXXB+urUkho^*6CdPncaL7N$Q|
zadrGAF)GnF)F%fk0k?_DtyPANnPTY!$`=;~&%x6%!xwC_^yu*v6aWX(3$cYeJ;^;c
z46U(l#qoG|5~J9t14#K2=Nj37goaKvQ`;Q103MYI8Q6CEgB%nH#^CW=oUnH`zd0dB
z!P^x&DiV_BywB;6?}oWBHwRfv3XqXicDnpB{iE7tL=;M`(7w3j{r*gvARLL`KO>(+
zNa(#fplw^$4sBSy(E;Qhr>v8iHbqEnrrt9pg{wW{0CLb>qDo-B2W>VN$huAzC!<*`
zEMpvOJz~4dnClYYJRGmEjMvQ3>}d$6IW3IItc9K~c)!0nT$zbiegHVZJ^OIQh753l
zq4Y$m;FL7T+CNAv(vo!41GajbE8hv$u>{HM{fq+(+{-{Ji+DrE_DG#I@+eiqM3s-y
zzLz9Rf1S1Psc2iwalR|xv%Gh-4MQQL0wWQyJ3<g)-jtIlpJW&kgAOGR<#BI{tEpnl
z)36QjhfMMAit!9c@TR?<y3%NukJeFyi#e6!2vl}@MBh`Ki`BFDlp)ECA#p}2-4P?f
zD`mK-{jyEfQWY{Is95zC@~Xb3L5T{IW~NP*8c4Eq`U52(ok^1|TmEcuM6WD%m~%O1
zoMtTdw4kJ50q$6vxt0#GD@u?%j^u;y@9I8#Z+W#eVHuQF!>sC+Y`;!_VZ%DbHzE@3
zu}bU&`SsdV<Y+q}q2tj<Uv|R{_IT}YPg27pQg#4k<bG#jLgy-X9)Z5!SMS4}H#|$E
zQQHB<>+~T?{nJ6@?{5e~)vY3rESBj$r>EX6wVJ^lwXZD$$~dQRZsdN+u&HY9^h<R|
zu~nv(eW2`Qx?NQiRUJ0yP&5~8k}_@1UHSQVE_<`%lpsNH|GGMB$iRvVQ=#7h>!Z`3
z>wOGY2nEi&_Vwj^Q}|Zd%gbJ2ndU00c*edVcZC(?{y5j#KsK)3Zr|9<mGUA}^se>*
ztyB7AS59VMlQ%{AecpD|zNwC7nsf<D+!Nq5>HvenNE42v4dpw=SK2oh*>Sj~+UvB6
z%t0l2Fs}L&Ihs%<?1OnXh&&DxprzeydGGrB>_c_;UqhmS(~9?n4;+LO!^K|}7k#4O
zs6>4bKC#~`jKLxg#lCMJu5-^O0pa3XxZuEqD^8)XZz&&|_=#>_a6$`*<$2KQ&ly#g
zAinO%Kl#==V<SYbj&%1-I*??^543yZq^5sceX9hg82M#yV0?)pm2nH&w>NnD-|Jw}
zALGxeU|DxmOzyCKNApWXM9VXxl=FTNM(x+y5UW#6-$0RR`_4KBLdzg<_8lNOXCc_m
zgDJ%s=yze#w(m-;aWu4ZSw#2~3{N~#gKR`5`$#VHPJ{u{h?}*>@bV+ot1=X;Oh05M
zzq_KC`u<&3K#xW50EN)Ib=aUBE#FfRx}T)$BZPq*ARZT(?on01C}8}(bx5qf-R(J)
zXdqg|6#9*0z`n1}O8PJKcCRpDP^nIo=(6vxLdmTwZ}HNYA|+(TejstMZTo8HQbfDr
z)|*$@-eyIq;)QbP6_ApL99<5V7239QS1xp7gwN`~Jy{_m9cp$m6AmPW-w9A*`|1aW
zKRP9*3d^{fK&tI;?o&m?6mo$DkJp}By<xBo7OZ}Qb{j}Y3$b?TBX*z=o0zKd)|5ON
zJF3%@R^XZ0=^b4=m{{0F#vYf#!lY6f59Rf>$63-`2jPlz#3?gH0}HUtmv2p8GmjK_
z!7EJPD%T%ndu*ZHmwS{5@t9ZGyn=c7!sE+VY_YsarLn|hByYrS?(~#BNlq2{5%$^P
zqBvFF-S>51KLt1G9LnvOdxKAg#epX)S|ERyatNC;!r75>mu4qzsizBh!&8uHsQ!vw
zp1YDkjLnWxtIC$@jIqqv>iK)bD-1b~Z7MhJqjjY*R?SHiPlZU8X>*Znf&PML$w&F6
zDWD%KztmUoMaXREX_37XJ|d@IGxyS$-uNy%-i#p=rH*L>jkpxaNYFBB(^Jk}xs56o
zGMLE)J5gbQJsoOs$}6lEF(jwNPB!;tITDg7y~2WC!@IWMPL=z_)-y`+vI9$@nKBmB
zk^O6@D?zTt%0Q@k#cpwKKUJQhIafjh^WeaO4Mo%lm{3+?EO+{KBJP>8B^|^IaK&jA
zk>Ql)2Z8N`olTf{lw$sIAq+dP8^9vi>O1VEhQ-0OVz~`3ShRC>Z4Y;QEvJ>ELZl&L
zszqqY)8(ZLEQ7FUpsNDYUUm1@wRXP#K71Fn;tMmT6^VrH<U*katz9T6Hkvqm#p#K=
zlMa${pQ%`D4U?KR=d?u;-Ra4$U94IO;cOHdm~jAw<)==41)&FWL(Y_vb=oe~;Y@dS
zbbIi!aR;P#F7L%XPN5>O*0aln(lj%QcPtSwjpQlw*na0OaY{UNoNqDK#iMkkjI7Ak
zm^ZSP-*KlW`CMovnFb)uy7)qak0ECF4UqPO6_4xM(<4qR1~wp)fw?&4O;K<5Lv=H_
zfBTTr$h|E0udCEf1`D!dg%E953kn}5<euwTnv#2vM%W~ugdDI4+O@je;-gaZ=rbfa
z$<e$y94vpYAw@VqW8YXVlc}^l8DB`?bYMkNP_ep@8H30#*!9u_$unDGkYXMR)}co>
zjW=}Kdp7B8Ap7C^V0f84O7qJ{9az%oEK*=b0zccJOp2hGcCkGqdvyBC;t`t;Gl%_1
zffdq}-0l@NuS815!69AtqxH8iJjKdfVlx32V?+K4QCr)SdQRGp<y5xvKucQXy(dEc
zquPW0c;0c10%yc;(kgd6txb_PF4#}hO?ZszIV!8+K+@JDiDVRR{gc&us2auXX@X8<
z<pWxGMQWVbPt}{2_x4@34zw_6ZG^bu8Px)`aLInU64th%^-d$F4}PII5yjLp2dwFz
zX$GH#C$eNL78IwaSl96UF%63K4sDHrH438rZ2EQI)OD`X*0P(bl|NT85lljJblQRC
zxG99LA==NEoARkzvsWBg84wF!EI{HM{N{xHLg|U((E1*H>0)J<ogPUwd>8A`ezA~G
z29sbLU|$^&2%Y}K>$Wit;j~|>sNqYMkR|z1`p+S!M{I(#qifj6FIV=W95z1TaM-@I
z38b<wapr_`R~bS^TbyQ2V*S~#G(8kJWCAWqVwEY!>B(iVt3%k@bPIL<R}(w%H>?1Z
zYgCd2FyVNuvZU?Tl6SxT6)$iu^OYe`^BF9Ru$tJf*I8LJ-EtbqU;2_fqi7OI?KheU
z(P4ENB2;0rip#8q#BRS?XBP}W6ju*lU2tH9(a8gl$WL)6j6C*Rl`XEgg94<q%9kj`
zAj*0%*->L$T+4pDO6>^S=tdc`2e3#?ar6dUL_UIa_cQ@;_B+iWq#VXW9azr!9=6t=
z{cbboiguRyD1^wX7oDD@(H=2ytVa92a_jtZS{|a7Crb9B(<AcW7=l?2y;W}C?^g&#
zvf3)2<3JkOVw9zgx#16zhi|@j>n7(qrmYMjUJ<&M{b9w{Ms{wel}l^m#x3?o4Ho(`
zfwX~$XRVTfbC)DSw8}m4+CLtkkZSx1mD%Y@*^xy#0i*eoW^#2rr;JI@IIx&5Eev8c
ziTKeJFTp5PZs&~sX`N2^OU5%uY0S_+$ie9k#;C1Fq6f#U+MiXJN<Q%(it;+Jc#rI-
zk}z@?CwQ!EfPY@+u}A5$VW*kub7XR>9xL4RFPd+?027{PLlVb3p!^(E`vXqDxCNQ&
zO8)uFY_(>ty^rstRn}2@F;BCpB4=JbXMa`SjJ0XUxYI0ae7@P8V)ff!SM;$`WSUeg
zC$OBq+izZFf79S9l+@Vd)BYIpwbL(=Y{Rxq_O}&9;^?q9>J>P!^5&D_^UTyu0$J?u
z>O_JZygN=ikS_MBU7wRn?C<L^c69fQ_nf8<o?@=DdKDi#Ix=PdP<}~11md}}B91sP
zp=~{(gd5k|KbEiPFTYUckSpjZQ=NQKRm}dWz*H|)RjZKe99S;N9h~PhZvWiOJjv66
zHi;1zu%%0;e8%Yy9<Pn|y!}i4@!hi+(41!AL?Mgv=}~BapPfPb|JTx!4mJ^<wlZv%
zPER3e!kPIE7EZ`S*+TPDVC~<UXL7I%ze#~!Iz2r}q|0#*T>Pzp|Ky|EzgIRep%`wA
zbPgnr0hB@@Yl)1(^OQ1FvX107Eer8h)Oa9G!A$#)f;S1-DriE6kONa7)jN0Fe^#oR
zqB<&#T*d2=qOd6bSK@1V@0#vt&uJyET1OF4jE>Xx-(`4W|FT#Yi5G2|o-(|?12qGn
zrQbbfwooRx=uq#azDxtdaw#(h<|g~7a$nSTOm5&6rr=&AXhm7sM=z9#K>U&eTSc@6
zOxRAp^oAR5c<@2{m~yX38bV^Lh~j19DXj3Po&H>CZ+Xj2_OaQZ1|M;wG|H&@Tr<%g
z+PHn(LXuF?fe{Ae^U>B!PYQMLB&8Yid;9q2X;EfGTNVT-UuK-3<MhXGr2EOrPk^+2
zLWNy_EXt0Yxq8@v6~a{{orqK(AXRJs#PWtT$fjPBT1^=1o&MZe+7A;yIHBrbpR`c-
zuCb`q;{cK-My6XkF%^Gu{emEKF@%c_Ov`X<)^)T`X$E$pG1%6fUyPjoeB{j79)uA4
z)bhyuOl>Gfy}~4MYbk3Te%eATU&s+j#%UFlW{DWmA?(xZn;{{sx2O<?;X)o%0wtXO
zK$WCh1szJ>)mk}W&s!*hDkPwkm@JW_)0T?vBPWw{R}es^8>C%6|3;M<`;2BP`aC2x
zC3<r45O=@-8pX8jGm|9m-gxg_&UIpBNaxaZ?X&9F^z*v-PUU!M`%%^x#V4J+^4M1@
z(qPP|Dp$&w_SyMpi9BAeuYCkB<L6YAfeZ~?(T>0h^-u;7nAD*tmSdk=?$#@FKi=;Z
zM)iO0j=%PKm5M0s*~QjDN-Ll;p6&!>WuM>7@KKgoWH30ez-8i3;yK%7NH1WY8To?x
zu`DCGbI1~KV2MR#-9=@=FRV1NaSqXzmL#v<+(bClhV6^$hs<?3*L%ikmzOFHBQ{%B
zeOUhBCMWHS^Kom#&Ed4fHMYzW3i>5^J)%>e%JuTz-iUo^{ebntHZ2XDIz2OC6Zo<^
z@Q9uepImZadg=ybQv32|kW?~*>{$H{EM_D(I*D>fAZ#rQ_7xS=1hG@8NHJJ|64T(F
zq67B){2U}y;yJyvBJ{CmFQ}`dXzS#shKZ~k>ge<*!@&fLYA;pFUf74>PGb*&;Q$LH
z%2FWvJM{vgNEH%`YrCNipsX($d?g$#dF|1lqM%mxVAy?8d1jayneAMesAh$M(qE7w
z<q`%jH@N=VSC(H6m`~sr&C!TXfBaIFi720Sop>Y0ahg!p7Y~d>Cb%>xDF651_4jh(
z6#apZ<P%Zj$zD?5hW4g!V-fK-nbCvWIQ{Qna${+n{r~T;-B|hb7HUAvLX!Y;X>6&7
z4d0YpS-M4Mrt38Gh7CIEY{%@ly|iw0+qyfQMv79S?Mc*-mz8nq$Vud&uq9~@i|fGY
zPsJX2lKB`gCk`6YUtaE#+yA+`uDE=f?k_oaHH%Ukx;Q}<CUsKUEArFAwAvnVnuTM@
zA1Fbn?6UUCTv?Cabkj}FbrMWk_amQP1ar}D&bw~D`PMs}>q1s3M~l;CQq_vQYsIRY
zS31`>Xi`&iQR<*%j4VNsMJiQ@LM-f<-BK3}l8EA?`Rl-nhps(5MxqPR-aWfDUog81
z*FV=mX%P&<kB;Cuy=`Cy<Tryp;}t6U2!@>gf`;^@pNrrEm7~4A(rD~u<l#ou5$$UX
zlFy-E%E29X+-7&w7fZ#Un>uV>V9R?|O^A$bX|Q+pU&fM@#W-x6*B^I!0ykO&vb9KH
zXL8ea5L0|UA}Msx@ydoA^6ZC*b%|1|F*6KYC!}##3{0D_1o0)F5(h=ws!GBf|2Yiz
z>p74>oTB4c!sG%{6S{KbjqTlYPD^(z@5KTzLbmPnuAY5WB4FjJRnBF>V*T()cO-u8
z)ww#NSiaMc=CHi?E=B6Urg>}D$ituGz=DWtinGkFd=+d;KA`>f+GaY_i_N+7!R92s
zE4!V3P0F^l_u1>pH%VrZ(=rav4Kb3Q{EV%{LfTg+1&xhA)N`&SwB9#MOuwzJo5Bpo
z6_7DWJD2`lvm+y8#1`6J%?KhrU5bfUSlZ-O<n@9L)lUg)Hc=;#5@nv3=bfJXl-IYW
zCv8phJdB~puP7uAtgwi_NF@SP)COi$W*f=a);7e7B{x=>${0|IUjxSwM!$ATVy%eW
zulwIfcbJ$NRc!{O?k#J#ZnwM3M=XR89WtH+OJRQX&^B9N8=Ho<IE^43J}Cqrj54;N
zeoP4CDfxv7EI-!i*V_<3C+(gFHT_5!p>DzH5o%=5g&bdBKYB<-9{K2L2UHA1DxS`+
zy|Fw?@irmrK=>N|g3}XUseCK-T*0EesTpV1az24%1gyOC6ub83h13-#buU4H(il^^
z)-EV5P}*at6ud_)OqD0*o!=F9DAgq$T(U}DuzMS9rBOrU1eUnnbnjN%QfY48Y*-m-
z$9kB@!lq0{amadHAY!@QR~{HX!lLm2D=R-vzzII9jbh2RmZ#}s&+xIz4GyIs&~()4
zk6)%$DU0GIeQnF{LjFC>|3gk&Iu8YtRh*do#}V6}xa-dDZntKgmKNg9k(S-R5Nh%x
ztDUPruS)Bjv^O;0uG59J@n3OZiHzHBS!oYcI%30-VjQP__BRZtKR#Eqg^eF&?ji$`
zDgBL=G}4Se;{bB)^2Oub4jR~-@=dEBnZ-UsqcFe=QWBxtMSF9h(ov$2xK8;{A;qmZ
zdrL)e<iFMl!3^&7#FMmt%h*Mdx;<E6pjv7*;_C?qmJ5H!IQ6^VTAl|jNy|c%mRlCO
z_u7*wOR=ezCI7bi1Zon^I;|j-v>|I!<#Z!fhPzRG?d{Ffr<E#5BP{GtF^TlGUs4*T
z#)w!GH5WU)y`%iTY$P!YDJu%^Pdfb?_eNG`YkFgAj#FK%RYn!oGQLJxo0t>P0O`rm
zuYLpr$c}=awA@D?w^vx2Q)TqHY{N-YlkG>G%UsY{OB5EyZeSw?rhQe~Kb{e<umupS
z<%z;-?JA>fCk%SdX(k()L4PM}-F7yU8&$VL=t}jIo<MS#%|d`s!Oq<B6n{Eeewn+t
z_U}AzY2N9{m&nodFu|Z>g{?>Raz6F~vac4YOrtW^ron@|M0eCuYV9B<)H-!}bog)@
zi9||7&DcX_^p!y=qYPNtO>qoBR@xiXkEDn7_^|D&kGt)TTb)KhDP<so+a0p;ii&|U
z6^Gz@XT0`Ez0#xOuY^RKs7MIv37>JWmj$mq)eo9^#@fCn%ez%HS0SMDSC<@Eh(y9l
zN!1JtDfpqJgI1@5b;~=I;08;>_9)Y~Y>V_HL)pCT0h?-GGmqY?n)v4&Sea9mCaEWr
zE8tpjH%INA<u^}2yfQlQ-+`sYXERaTV{KGco30bQQQitFfdP;vppx?v{d&nwum*LV
z>-?szYuDIJrN)g!053Ytm<2heHBA>eRbaNjG5~;W|LP&HF!4S_1vu7z+g*l|vVZma
z99CKI;3f3uvl%#8#-elH_7u!16t<@S=$Qwdo`M)5Bj^$IsAxqh01wxv@f?gDAx#4-
ze3Dvi!(%pA*)9~?LDU#^AlestV{`UMy%`}y1;1#=3)8<zmMlyV^%?bAO7uB<SA8K`
zh=_!3+xsxiI0uG1*xsF}=G5)3b6vIwWf+T;I3AkcpRq@)HqkFM*z*R+tm>pp9bjwk
zX;73hXXHuAR_nE=D%#BnO7=D7Hwz&|iS<dXM{ja^QbQ!tl5NN<Ev(tF#%TySv!k>2
z-tvHyUL`|_Kj1xZdbI4v=BNuhVvkiOaCdj4=d@zD+DGI~i)5;ZM0mVGuIe=ESWo#O
z8dIm!AFEAC+EN6sC<ID#%C#y1WVT7BWxub`@@{rcu9FGf-9Z<Ae_a^u2@?DxkQ9)J
zPBMSDr^5{Jr+#gjNnXIm$XIdIfraGW8nF*N2YDGY5H*yYo?JOYL)-1^^3^cUv&(0k
zR?04Xw9Wv@Y&HZi`}*>>G%#p52OU`QV{{-QRwnEl8vJm&BqKi(7PiI<>h$XeTjIzF
zQN=f=RP*u|Iam5}&JwY2Dz6HmJ3iuRuP|{n%!Sv<4YO}9AFCOanOQEM<oNU?Ye6+r
zp)tIGA8bCY?`C3raP;`0)1OFr^X#;JsC<E>QmOUG<qj-SgPo;>$1y@IvGzpiNscP9
zFarHyi}WO!4zu@6+lTWV6enztIE`H8#6J)LVc*h>T8dh1zg}VD7?OjKXzW|deOiDT
zugX~fi$z~1EG!^N^XvpyPuaIEJgcG+!&DUSDP$w^1s!N8Mah-&6wDN7^ExCG`}R79
zm)uD0U#A(bk|tSff!a#YFb^>ePuh1>)Kg}+I~Cj*kf!#<uXurdXQ9ujap$sdMlpm1
z#uiSGW;gn-f!S4H*fS7$;MVC$tgch7UHeG$Af;2{<c~Pas3{PND*Z9c+|y+%Df%AB
zQ?8=2RS2uC-!11uCXN6qd-(<E7|KID`ks0lzA!v>Ix_)KE?!I&Blf-JYtj#u!kq|I
zAwO*FPJcW!55y6RS+ehIM!?xaU^8-c21IoF^FyvbhU{@jikW?XU3+(v0!xi;2N3jG
zW|O^45zL3L9<U!MKTpOjE_lU(73|A>6rx|OG}*yqx7l0woYm~aN(OZx@r#r%#V%xR
zR9)cE*X+sWEs;bYf+Mh8R1aVY*jJZ{+`f#0gP+-Q-wW5FCvgHzX}3bGw*7U?)Bu+;
z9?8>z724|uZG~#e(opg11LZd=Qz(~<R0&}D+2KtfQV6lIsv6S$4h|qf#zUD>bJik?
z$LTN2o;UGJ9lEMTa>WiMEw5g;#ktftb+*cs&DVv!ZE_Abh|`YI)x435DB=qRmWoDb
zk`64VB;FDEK1@u?P}DTD#Y*;51fNYi%|r$(L-BI^!)SIaI3eZa?Qs1JO6&BvpLTr4
zLF}ih3Z#VQBlVt{j;hi*?Ih)q)+3MDQrQ|5ws6*K3mpn=>}d0}sJ9QXKIy>p>-J<%
zA-l)w9r%Tnw}Y1Nz@$KuxX7Vv_$7LFyb2JwAlfSAuBae@Vi+qqqvb@y1;RKKwvb>q
zPLH?%o+cD$N;}ur73Oye@vaY(@d{$aM1QK>2hNd+BvTx5U^1mF@4a`Uoi6>6a!kLf
zL;%Z~iGzZZMdpSs^XCP7s-p4rItgpqoA#cusA_bkJX1;}ksl$L11qzgv6op9T=qqV
zNASbhiWr>RWR&3V<iP-A4z)&e30<~xl|UgIZDzt#9Fy$y#NB#e?pWvT>0}$*W~X#2
z2~!Rr=sV-jq=K-$@{q%Jz7oUG>TPSCR-AR3()$72hNt*{*)HT=Tf~FroR%<&-JXKb
z_4mUnk4;N+I!)1wi{-o0CmJzT4lF(+Ppb{x1w^3w*`<nz-YkqR4y_q+fLd@j-gx6p
zH`(Qif%K|GvZ(gT!%k1qJVrcuYVhgVm2$VlVHJpz5+gD=;q+AON*3qs=j{iR%7#?1
z18mVuiAM(t=;YELs#CwG&1Ted+9`STI1daz7z^QQq2^m|aV`t8)2Ae@A}a-7%K=fl
zjY4ImL4wcfZMNKuF6y!Jy)t)m>W<Ts+;wAXmtC*e-rU*Mc3Lr%bc+e8m+Xh@r{1F#
zI_b326M#8pyqhwCJzL2#>W!z-IA=q$(9=$T839$vLQ0Xp?Hskxf24lE&Mv2c(aMKY
z>X|}SENp7`{Ae!!wc}K1&`A};4kW&F0c*L!J)gOH#D1*uIW`T7GSG}SmfdSvTckAH
z^Y-Hd^QFu^isw17#K{nX^iPzZhR0G8vci9nlTUf=3HL9{<v0(GjnjUzVhJ1lFiVbh
zN)PE%z)Ta7U@S5rBz@6W?58SStZz%Yn|1(&o-eXtXq%Uj1KXSZbbj^nUQYL$8;z(f
z2Nrh^hNE#e>e|m#dcui}u7haefC@KKJWG1XWyO}>Ew9hd<{DhRar=7bIu`&epoqx!
za|I^Xw#-`?WF1&C{yi<^)Sn+fc?w-cVsd)o-7H<Wrw&<^>$8blv|p&`-!Qt#Y2;ey
zOsXb_6!>3E?6xO%v`0o*PqgECDVu4EF`8iV$TGS|?U%~8DpLY$6qCQ-fh7sXW)Q+h
znY&-EZy(ya-D$+ovl^*b+5!8O0xKTkAO&EFdk}$r6cH;GpA+`0b+Fi#lxCy)ooTrX
z7VY%MFBd|eKg5|eqzS==Puj2LlJ8OzjmxBPVhr{3qX?0|UT7-Ar+8R|rZo9rIU46K
z@zTw0qRD=vDwcenLakKU18gD7GzHc%U`Wn<#<{o5`9*LBWgravW-0*7dv|ca6J_AY
z<%3%T)f?8W9$IU^m7lz;JMLUYMAnG?c13mUAm}|3V?q0VZ+ZeZ(lsh4`<>?Xvcd*M
zdmpyIwxJn0Y`@z)I}?c%$(Gt1JvjhuG>YWs9B%u)x-3U<A`#p`J8&<<A&U^loC+pj
zeX-v!JvwlNLm-%AxpEX5L{AV~&ZNX~h2%+H&{8;Ve^B2si}|%T=d{arQ0)=qk|=I}
zSe_weN0sMPh$8Glygdp-_i3I}Ktd4uqbf=d>1YWGT{{r%B12@a;)G#~_~S%=XRhU3
zCUwj3#zS#BGlvRN{*y}Gx8HV~(@5=cYFgz+>`xn7mAfj1lR;XgKQ}HZH#^~te^v%Z
zJqmPSX>(T`SiCz~ZBS03pFgh@Tim$H_ee!U)*{94oFtQpyYeNNEp(=?{Y8CA9OIRt
zWWaJ~xQjGISY^bf{AK=<D9_<exNHDK0{WoKhR+HjC;5(-_mBFkqD#NP$X_Kp6ZKu)
z=d==~QWBVP<N!Vxk<k9SZZ*`XK$D|+1M|X!HYqBxH{hgV-6_zsiH@_ssXrYWo90LY
znn5eFoyzh<c|QBwiVh|nd;Ww2NG{9SPLuX``IF_nIxrI>g-(cQU|7{Bd&AQmipc$a
zooXERUEGYc<5eR4GeNp1>>ui@*0f0hYK=JniCZ_Xywm<MFSmB*oXgalXYeev?Vk!_
zipm&z2T?y)kNb0b<Tjz+3V!YL_RkG!aRocA@+~SVlRN0YRCGfVP!_rp<}XYr3r<hm
zt^HOwbu7dGnmlNE@9W>>T<2~Od#yA(`K1+jBlY3D{ag9HCQqwEl^cLnpH$?`_w3*E
z3yA<j#C2LJ5Iod=$o`{znv7T(Vi3MAyp8mTbUBHGZECDFVgK2TB86wgqRDXgl+zQV
zsQgzO%YT)-6BZjicjJ+BdeZZZ;3iffUeC-$i0%cjkU0DAIx^I<`Q+>lEcaTrE;;V9
z218RD&}=cLcxdar&Sl|-I<#BLaE$BfA^WJse25fD)hNw6H$s1Sr=0#o562AIM>mgA
zDU&pOYSCSAdh)ido3`1<l-p8;N}G3t8aO?bG$SnK3gX<yHn7Nq6OVRaF*F{IAe9rX
zYsHf=t@d&Cb*l)AppmPFx~vb|$JZ9wE)z~GUibt~<tJd$PoRx`LV5T+b`n*x)Zg?q
z<ye}2S%;u685ENsxPl@-u?#O<ANn|EKKQWL9z7>jEs32RePW-qSlIC_JRf5W?ZU++
z_(d99ndwhn40<I<ayayv83z($5LG|L`$(d(Pw68FmbVc0Z&JZo4+_k_?6p6QF+!kB
zj*SFipV}b(8H5_0uW{ZR7&I0l4t9u7YwlB4L5$Ubm2cMunX++i!k=F6#^tGzW36#u
zF_nifL_$;;zJ<*==@Eha8GBxxul2K|U8fay%e<L6-zvTpg>?|3j@xJC8<zKy%b<WC
zBo!~PFt@9U2EGjY%*DdUrera2ngxXmA)c!k&1W@a@Ju*2b>>C-Vzt>Fa{9G{rpW<Q
z^k<i6WIC3Rx#Sh5sUc=e0+)TxV%g-zI0gwDwgV{JOFs{or2!d#Zutbs<w}eSD=DIi
z%r7!2(;vT-kR(dX#3i}%KCkp>QOE$T{WioKrzeH#QJuM{VlDRhiy^PAzQ?%=Jm_cm
zK>2&czMwoRSr6nc282=VlGCF_t<&yz*%#(KfnQu3URs<?FpZM?Ma^SnXpe{n2bKZ|
ziErp~_QlN?UtnelMhw}M2y>?=Cnt22ka_J(3TC7j)k;^n5>p?cKQJTetGU6kdhJU~
z&tU#~h<{E`Y<YLyVPCdbwNLzpT?u07O#tINWOL*=JZN~{zC1^<dIYhE(^8#wp?&$9
zWFn9-rHWKvQMm2eZO&zxaLE%@FOatO{O9y&7UToIf^C7Ge7SsxEHngldqIIIqCw@%
zu+)2n;dbY@3TnP=FRTb;aa{uaEFA}yOp&mKDZ8OQ31(L6emk&4A?9tVPGS-B%Ouy^
zi%O5OOXT*i(Rvr09)0EV-fgoxdbsp3(b-ov>*!Qe;n|PLoH0j88J1qBziN47_}6u)
z(~HaJAWMi0ahP0mVA3_FW+!*qOUehNjcu^57d`m_Gfi}&-B>>tz6I%q)+crDiqjL^
zMC(V|4~0IiEo8sS+u2QZsg87KcgWZUAhAaH%>jF9PX4tI!eT#&w}!TQ5db(b&0bbn
z%ob$y8K)T&s7X~3U!)&<dHFIa8Bqg^T~qxRot`q?nb{|$@D=55t#gGcE9@y{1bX6b
zEzXd|uRdk3Y+%&G3xYZ5zzX+NuroH3=;?F)n+r?hJZ}{2X{TKrB~$^Lu@#F&6bXfW
zKn`Y#`X@wkSG+{aQ8n+fTn}POTwkzTnttj8f(UAF^LIWN!y-JBeo0c>m3ueSnwYX%
zbH}@<wY#+~Bw7cOdctmWmqV1oxGg!@YhUSHiA**TA8EIjH^stdF&<`%nvY}?sT=#9
z{uD~Kx5n&_2D@1rMn&wjLtS)w%3P=l<smrpF}t(5TXJ3OdmsVc=k(-UfF$jbIC5=e
zg~bBY%g=d*Nu$qzwx?`WMNQ=VR28L}AE#eDGmwC^Z?7u%EAtQ!E?=e~5M|iNZ-3h9
zPuX^Y`uLP}AzE}XH5BccL_ZQbE+1U~*Ml+}_XY#N+o3=xEGVfWAFx+f{I8uLSq)7W
zKnSTK-si+GfB;2C2o2}#HRW@qSBmeQM^r5#ra&%v?PsHlA{feRA@+ieKyv3vdu<MX
zh%+o=<D{**X6@~_*VT^=0a=Ve-+^Jt>B%*_zSWtwudWk^2&Q-z*<`fvPk8OoW?SNt
znNnHYR+r!CJ_QhHGnG7i(CLZEmOYjF46u#^c2`4n8BVS6VLeB-VwuY57uB^U37xi~
zx};dS3Tc)KIv9-|(!Fz6vPd#;3HxRAwx+y!5kHxpf8^?t7Z?g5ep+r)4%FI8QjoIT
zQ)v`vQrDe+ZF;L#th9A`IZVzAmocssYr}41clSRvoRO;HxDb1T@N%@UBJRpvBRtOe
z`&3cZmuHGyi=0Sk^$MGhl-Kw=+fcqZWi7?EDdVY8;=J|?+hLx{915~LWJkGV_f%wY
z5@*JzJD^}mxz=*?8Njl=KIs{2OF3u}Kv~33v4{`G!OI&L!-2uUHl*ZGU@mBo5F_A8
zUsQVQ*{0@~3KY(`U0}KcTZK%Rn=2lonuq+-#Hv<UsL&bTac|{Lx8JhLY2*hL4x$pP
zTk4Bj6LX_ZEBCT6Y{)pJl+SsjBZ;Et?Y@SHQ)SXFgt3v6>gXxI)TW%K9|fSu(1xeg
zwpK1Ojcq_V!VV;5x1qTy$#>iG>z4P9%uY$Bh0{1Nv4q$BKmXbG9K!=04pN(RS|Y}+
zh=cbxA0;IV#Y(;wh3+uRs@6e&GN)Ww473DX+Z&QAtlmB}<XjiDos4V`Sppv@PtMpb
zMIj@@r0F7PlK%1=)m#hnsJ$^aLdvXja_NZEEaY)Hu!KHjZ_1IZ9*J}4N{g`Rg9r+5
zuK3yA9wj78fmRMI*EHUf*61{6SK3>eDWW6}Z$2It!czoT(IcL*vqkM1d$1XL+I?lt
zk}K><rzgHkY*_My3Vgh^eyg-C7AYq>c!e#Z7qW=ig3sIA@)fhpUe{@qJzV0o9;cYm
zCJ)ixULVSgln87GRvaCa#^Iq)u{v{O6nD(oId1PLPfi66o2Z=laR1`Aboz5^*DAwj
zVXaDi6tsyGp&W?J%lA&$j{N4)u`Z%Fja-L&I*5qF1E>g7H%bYx92Mr!n)08f7IClz
z8_8M5QA#v^&uJFRi$cz0p~i+G?duX-Q)ms=t}>FU)wP{wp-W`%a2BZy6tTv(v%FnO
z0p3Ayk|DE_e)huy=+Di1?F7NmbM3Z`R>(cwF%Gd~Q2~hwCOS$71yGOXq>Ys~Y8tgV
zv_^8<peM$w`BS1o6zxV~q9TfB$2z%;N9I(2dbsDbg~4$JjkWEeB+!Sa6(keqFFK%b
zna_0Cgxa<%SK9L42i7>(TDmvg^q`G5A5>$&O5mv19;EN^94QnzQ9n^v11GBYFrLE<
z0F|%Llugyz8?ffuWJNsZ8BMjFW?i;n1%ba=w;@Sc2!h*zAVmu2hAiP^$+=6Tnw*u)
z+q0=kC81i%6mf}<)LtHu+)jVYFuQrwM!pm?!BFg-rALpIPoX)%%4<*GH6@XpGy<Eh
zZws{}w<b6o>tD9#r=0#~Nr%M|hTJQ^39kcEhpGhY*cZtrXp~NWK}DE3(cF&MO!J#4
zSTSI~40*=sk5MH{NJT>H_Jqw=D&*k7@z!jw?LgxBSOOGng{OSUc2}%~bTl}2dg5+Y
zxjZGI4DV@RC&8ZL%1=44Vmwz2Y@H^>7^gH>j98Iq^KkQf+5F()Fh5|$kjq9)jhdd#
zH49kzCdiP%Rm0hc45AasFd;j4CCB0D5{%j-_DDfhs<&FMgebAN!sw69ntm;lYfH%c
zm+f6i_}huX<T7v=tB<b|b}|9)-Q{~j0*ux7v;#}+V?9T8JzD9XOpe{;O=HRjl;hbx
z21AqN@SgH~B{4^RPYLAgd}p1WG9#D5TzK~CA^VzU^z!(#cLuep;G6<J<$fjr&$+A2
zPDT{$Cs4@C02jg7VW%bhs0a^vxS;jDxn7iKPk5x$40)Er4X5p~I#8^TDS$uazy{cm
zRK=|4teSQQ5U(F^KH)GQCkN<R2UZrDY+45q(<Qu;Z)NXW%=%+88)qDS4O-D6Rn~V}
zO!rBaL6kq~{6eg0H7BjUKS#C|5t>kETGAib@d$>^>ucYA^|*7Lj6Z035rs-wqbheJ
zP8u!~l)4X;&rW@To~swU!bEHAc&PPH+t<~3zL&H3XatcD2K92bBAaY#Uh<10$VxA>
zuP=Cm_M~^h6vQnv1Cg^Xz4q5l6u%4oQn&S)7wsD=O{Mmyz$id5+iB!T+Wf}mH9Gw?
z-Qhx_yioYaoPMG2<~wh*Zz{LSoGE6MB$ZJN1UY-tABYnh)Uu0+IoU4lo0D>Oc5)-9
znVvj~Z&Hy@p<Be+mUl=sJHxLfAZYbb>k$S1ywhLA4B3ctyliL8HW6>ySmQjQ57nn!
z-G|_)h$(_iu$jt|k5$>}mkhCEw>?o&LO85e_!-nXVDS#&kzgnSDKctZTlHBftzw_{
z;Ut;XD3$lQ&c#ZO=>+FFjN7+Vs*ZZvAxS&1T=JXSUHjJZiK?=uFpxZF_C&~>)KT%;
zFZz-pfruUbt%0ESZ54aU&nHg}m9OUjQwSREjE>s3SJu{^o}H5Y3P2ik_RLO?wCy{p
z96#OOIX?Ta(+={0%0d%BIkxYt<A`l0g>PUvah&cH2QNz+_^y1m_E6<%(k#qSVG+p^
zQxW_~v$jQKidN0j0T-N}!i?ap%mwj0_T2?d5zq|HMbb&X_L|eL->8rVGl3w8a{N8z
zIZB6Aj;GG}0+t-5AIink4q$Zs-kirCbz+=r%}{4_ywkPstJAf1+g7JhXd*>ct+qI2
zl)tL-nq&L^@@9#5$;%eJ!r1U(*tzI_ATdusbqCAwsMC(G!@4BH1q_}*38?RY?M<{k
zfYAdxw9^#N+SS?BwI}QJjkO}UjCPs6u-;3*w0-5lxeOOTpI9V2pVL#2E!%|BaGtjP
z^)q0u5y{KPq-TzvxVx5ZJ;A3P$Zr|#c1NA-R0+4Vi3FXtgSEX?(IPFUtp&Gd)DBgw
zuHL-bX%s#mHTB^AnPYs;=F1E9$=WYqs17W461gITC6x*qFb&v3vZ~j;;APHb-g;f}
zI0QL`TdAqnY>Ul{`)<_`MZNY6N>^YH$Kc_*oJM8|gPHAl1Gt+7u!oGqBNa<&A%z(q
za$qq;F%$7uh*F=ve94xYyD!S@BhQKhD`YBpuCjh|!R%;dF`HTsb)kki2NGjUr&&sk
zlIC>a1&93Ev9eXB#p?fKALAWQ!hf_>&N;t`S+2`fO#68Kq7uWSP}s<Xae9)St%jGJ
zC?Au&gRqzoH~I&M&|irYgP>HGc{^DjuzACf(?|*$mOMXgr^*jB*<${e@wpWkv2SoJ
zyheDQ>~zITXL2{kd9#>+6k3&TAE!;)Q_aZY&<J?y0SBfxa@0jeo1Cflgf&Wn-#)J}
z9W^X2jZ4DL*5{+#D>e^vyaUSxOo)f7@!%xlukd%SejjDA<7g=d64kFVb|}sH1DcH}
zL0lgFKw5E{hq3b6)AdQiV+yK=pEr;PI@1`hdObUz8vtfPwgsnEXipkRhNVdETeJ&x
zR^>=!d1<4=ScLXKPhyX~Cgq;9_Dq3giD6a~vQC9b5YveX)JL+5<*C}uWR;YGBA-f6
zS-J`hsPv^u8qsf;8r(k_V<$e-f#q5VomJMojJ~;KmkTpaMKWx$<(Z5oM8z}c``i`p
zOu;aVbS586tIn0W8!GO;=QNX?H(-DI!TN4klDrsl5XtF!+UdzXbR%0I0lekIb&JBm
zbQp4lWf3)P+Gp)*`IubN1Et~f!@c%rr5r6V#dgSx7{`;^wPr9)JOLkc1!+XjdWI5V
zl>jfx3(yzfY`KyRtEz{SvgbeoG4kuf`uTDdu!-98Vjp(Bj7ht_c&`dwRagW;i(sxE
zv>&cyq3A8%sUow$gr3#gH{82<^+tQP+@fKrSeP(aVKPRM03X)XAE_cmq=rYh@0<hF
zX14>Qqy1=JFYkTPjm~w>)Lrdi`?2x{35=R4e5MX8C`ZVa{UtUEA<KTe=@HkfN7)nG
z;`HRaPUEkev!BSvQWuE|Gn}8U_%2|nD(;wWV`Un(pUe*-K5Mk?Gztq?!mARSET1F8
z(odD1`DGQz3rB64b$TSPjXXr^W!HYXe!|WU!3$2axax7!@5W)!wx3yiR`m%nLKly7
z*pwY`S?4nRg^>^9Ao0b1Hn+q(SKjViY75>zWj~j1)%JqniB@5kvd5@i#nbll6$U+&
zlasuG!SJZ=CU>*_LV2+?f?6>p^OWdGNp!_ZYP0-ef5yg~MheTy_1?j;#r8{$a3xPd
zESS2U3n62gcJ2!0NuQ0yLlO%VC1jUOdtCAvEOXj^xxt9UHPME`O8A}AlW#9tN85g-
zE(fU&nmCDWEI_4=lf&%v3vq8)XTO@*9PSP~mw~lpWqR~^`?Y-Z>?GlOBds0idqCo@
zF!u0hNwmLSaVBF8_*U(mh#Ycy5@(yp1F_%87G6+QPNkLGV{fcQnH~GhdhgWqBShjB
z1m@QyJ5Mx<0(EXG!?NG1^hK2=gk73iUiWU^_$K@924DT^*_pdtd*Tsd7QtTdBiZjX
zvnF*xZp|8MAz@u%YDZxR+?C5BGQQY9xhr-)``v;ZC8SBI!qVoz$}q`Co@(3gm3tP^
zrF9q~dw>JWWvD7jDk1y(4VEtivsSSYU`06UC!^O?TWmS@2l<<*r4uKO5bFmlS-_Uo
z++^GSuzYzUH8MhCd0cdQ^2Tkrg2wER>Ode?)$-sVFb5{&F7Ms7$Nsqa;IS^%XL}AH
zK29DDsidl@5h5;Fl&$zr@&`KgJj^Z*Am~WXmf=_RV*j+DUplGM@#US)>)_61FXFCz
z6~D=iUddifI=cN?=}*&Sn%&YkA6l-{U$|O&8&V@CCe!}B^yof~TB?Q0vmA1Ia)R!j
zWduYH{KaDAk18$%BVx!VkTlTvG(A8bSN#>~hT#aw=`WKOw$03vqs6vGyI^hLggM;O
zQt;*3`Kw%=t?8Z4W#B7?UY&3;Z+~52B2)xk7Wh$bV64K3LPpH`n>x9Zt;y~lr<uww
zg>I)>qhUI-zb#`|B8klS`&bk~CQeV1hIsg#{arG?EfQ&+hD<3BbtdfZ>+;>XY3l~3
zr46k&GrM!A{ljy<UF?WrvlXM;mCn^`*+13~kWU$^%o;e@@SGmKWW(ll_D{{+9SHWP
zC^FH;1*a!asUQwTAL#F3v9o`!)cd;EIoG))DfEkWZ~sz1{ejN9j?)sR*^!2aXM179
zv41U37qgJCS1dW9i&J#h=?^@#C+y$KC-PbN`KyE~`H~*tq%#2@wSTYs#{-=arxBBz
zcSj<u{YL|nfk2YyJ_nZUloZ!V`_Bd$5mLkMp{!vNcbuLSCvLlQmHk(FP#A~7%waYj
zU`eR#c_L0>OD;nxay-obJNZ5$HxvRA#vMqkn#5@sy|t~ev)bW8nQ&yq!V!t%!0H+k
z9kGo`SSPH?KI(Abe)ULf|07O2SJw2{+{_qB{2zVz+4W$F4^bFu$hGA<nO#gbua?lL
zPtIK3sytWDc1!^t;Tq(5+8v3jIiMn(tlf`E$nxIR(+p6qOCSaR5)OiR>|+Z|>sR4T
zidh3DyedzrXCGI_7#xC6QBfR?{-o2B%R1E_R;L{PiXD|5WS@Qf;h>hC<6Y-6Npagu
z8-2$<;c)I?8}HrfT*p_59;AU?z(Vnf38QRaXN2`QuDn-Zfo!cA`Awb<^Mb^*Pim$?
zk;bXQ219gup7V)Ewcia$8=rhQB!pML#<@yb+DJU~r&I)V;a)W283V4Th!~M1!@9Lk
zZ4jio`x460s$f!woSqn&%vGUV;@17N3PX{`gM#<ypaT=Vv)~rlLq5HEA4?9H>PJY(
zSH%h9B%J<S9aK2jWzTD{a)gP|+cO|;!E2BFH`4RSgyCsajj_+j+1QEidIuILr>$ju
z$BI?<nTHD(6d_GI&A@qx!;E(k{E9LU+h^6UxVuZ(Gp!sc8>#43X7JCh7`^R|TbxGR
zzz)SDREd1QeNHnus>c=weKD;~GKbTj6Am>=yi4bYO6-1aT}^N3au_7uTL5uu7Agu4
z3AlX>e1BdJXlJR_M^i(+^OjZi`T2c!QF49GX~Y9^VPynY#A4VNG}Dolikhe{kV%Mx
z)1O}%rXv}B(WrR!0sF$z6A0t(s5yW*bl&MHWGTtL|27rxR}}CU)mb3>lw-tb=IW6#
z5k~&S<-rj|2+A^8RUcD(<ctwl$&+xgJ=>R*2}rB8Ja<<dSbWBFY#99surD2$RRxpl
zSPNh=+@uI`jZBC<VPDoDN_pv`H;FZkZ^G%<27B%C>yN80$Cn>YyKEZ4c#yUcR38yL
z8wB&M&@%24u@jhr8Qi|2;$^bM30qDxzF%)aS75Z_h6<obG(NvUbDZcKkxlu<aweR1
z`a>@fY$?V!AJkpze-2h{dBNdQ`J5H2(BKTYQH+c7;tugU@t#53qlAlj=XW`G>$Bp_
z3v&p=T_SGUBTh5RClk*;)}0!gBMFMiIdgVH9V$h3$6HP_Hr1UV!d_J7Aa+X-6@FJd
zqgR}sSZ4HuWPez-Y3q|(#1igsnq{eyPrxRC8`xf4VDiD>{*p<^8yNDTlK)hA<s}V5
zkA>h*_(M!_3A=OpHCWy=+LbTm#<~UV>^#hb5eNY!rYz@M#R%jIJ<pm6WZO**d8vR?
z`hT>jk2yWLgSI%v4x7KdG?(&7XWY3=*=@Kr3TeG8hqJtQPj`A(D>H3@6}nXbWxDY~
zAh4GwiN!(O(@txE5lH=0I2M(B?G=@u?PyOpjliO8EEH`)5xQcpteBN1uB{t6wZbC3
zg{eA$mFDI``SMds*^+!NF(41{fS@4ci_Tr~Aqu-wf~4vMWTNAW(i35lh5TZEy!K#*
zL}1A9^3X5rmO68xxTMgb<j9hJ;Pl59B-imdyS3h3c6J2>%O(cm<QjGQ3k{`zcL|~Y
zw(^?+!zUwY>eD*?g*2C74^hO%Zm*lj*0mh^ODje!ZCk+s=fd=lf70$~zUQcR0w_{O
z&lCrhRz$zJ`YO2kohe!1i=T0>bElk~ogB86<sn+_ic=87$(0|LRqyvV6^3Iy>xHPN
zwyF-AK_RJvsHYuRell{Crczcb(j*8pepT&3ERvI0OC=%;oSr1sjk6CUNPksrVyj}I
z(@fMV#4orJRbcZLDlLWko8|zC=JJ<jc-~&q;P?Q>SSjWwy~5`4sOmgtMFze$*Xgy7
zu(Cgb&P_X2r3j=%U4jqU>k63)M><zRl?~JID4)8LBu-<#<gUaq1)N&;)#U^98%2&G
zk2tXS+<cDy$~`Pc_v$)y)@g6dX$EhQX{dyifiiK3l&SCPKRk447`<1R#--AatSB3*
z)DkY0TqQgq48hd5aQd4M(mB-fQDZQ|*1P6#G2AJ)B}<3`k~yH&=@FA#SrA1pR07BW
zi0;ORciFnc0|8h#GM!c-c43Vjf^dnuE2eg|kt%70TT+WqPU@(f$)|+lJ=g;FFI!)q
z151&O24>^HDw-PkFgO&Hr&H55)XB^fhekTXybhq$gY`I|9!dAhJ>@YH85L3^E4Zdb
z-qB}x4gDG$Q)91x4#6Q`N9+&z*G^CD>B?8R@+#X{5rijwcNdKhK%wSjb-0XeQ+Zn2
zn$p6Tq7*%%C5VAy@wU1CR!lROyH^<5Ua>u#zxuS@Tb?JWQz=vD(t*VRIkl%K%aKeM
z?AEq4zj+E}FmrdqZR_+TcEjqWQb3ZypAIwm($&*;U!6AzY7nL(zQQ`r-}D{9T%Ege
zRZP$Bn6s_rnVe${7x=&aI{m@ER@`FSQX$##h;tRF)10WViz+;9FAvhTl!T(#bGe(e
zlb&<>6WuT!`4XsfWB2F7-~O(5JD1U|G6G2H*QxkwZy0!u7LX1f0Tv775EW2Gbxmzv
zd!Vk^;pz6&jMIz<N`_NP0H<Ht8>_(6dRQ`m(@u7`ef?SvXM9uLMPKsZdz^;Q)SDqs
zXU5)KS0AgZE$hLE14`-XA*^ql7yOp;WkD~@i)vtom&j{R=ww02LW&mkgNcC7$gp#n
zN*>M>du#nL0(rYToMue?fDKJjPsETNyZ*Snt&CM}4|X5L4mdDP+_sq!dwUKABwV6{
zR<7=#l630C6lo8L-RBwV4!olojSMXc`c_VZ{2=rMV$spjDV4%;%375&Wh2q3Spj);
zm8fvP(_bckNz{ru@|5kUC?%zwV=?4p1*DR2in~zc7nTC7)Hd8umWuXa6O3>cxuio2
za{7fiW)#c9NPQD4O3Zl?$+N;j?BPdLrLuOpFP&my;`_XTMHIbsil=R7on>*X)~M4C
z`jdR75WdFzWup}@#qv1MxPfUBUiRQyY^-!EEkVU%<i-b9h*<|CrM8g?3!Qu|+pdb>
zw4HWg$^B~4Y+*=JDdIzJR7ZN~@UyCiq!SG|k2J&P74`=_X=_hs)OOX8hj14Q!+|x{
z?+J=-*?4o4%*e>$R}VO_96C%-QePz0AYC4{iApSR=R=G`RiBe>(i4<B$Q3gFDwwNH
z*7t<%M#iKw4lI5%U>xUZIxQd<O}F$5Cd-*h8AF<gO!-dw%Xg}pmdqFua&QvZRC(JX
zQkq<^9L(Xs;%!tJMT+Gwa;L8K_W)yiXT{y-ZR?zd=5yoL`)oQdZ(q5}xlHOVYBt_y
z>m^2KCl#t9>m7iy1Z0JphkaePncTq1X&iU1%d(1+N?_uo&E~u1V<keiu3~vMY<K12
zqp)$O6+@PDSCW|q9>}mg&1+9%V^LYbiw-PzfT7m(4tu!#Hko<Gbx817a(a?I-ZiIe
z+PQM0swarGg<=|Z26}SFB<X3EatDYcEuBjJ$iOQk-YPmvvj1hLCq`XNNsRiey{q};
zI5i8Nw%`pcb`%Puu!P@To|<ecV;z$MnjXwSO7uv>lXjd(6XRq|%sAIt6-2>z+V-CM
zu*Sxq$V@5|I6Y->QZ?jDk$0sj#$jJmImD*cG-ainb_wcIyr%r7wJZs>N7?_yg^rjv
zC}~nH)N}QM*PgU!X(nPj#jf?<M9lKu;m$NC4#O8l9GK>4rrXtux5vuI4iw_#V{;83
zx&D~bpHr_0F?+mXx(f89tN{~NrYJr#XYXrnNNWu~IR_T_&>k(D#&JUNk{H55YwvIH
zl?MZBht?G2L@avkmn)G&LnA~omv2Ne+t(J%gxCu)Dd!1g8?}?3W+tBEmny#`Ymzeh
zc)oq0vcHbvB{}uA%PB4-SpzQ&L31a2wjxAi%=)@&1lZ7<Zjn0Qb0B3U$x0y?tX7Q-
zdtYDoD^|@ETN7qqK*A=4*s#;tH&lAs#k|l6!d486+&HS#{*C4PpOb|)poY?4Sj0Z|
zVeQQ4?VBoRo+AI3Xgvpz=r6iCmMG94nm$9iY~NhTabtT_MbK&M8&?nObPxMr^H~bU
z3$K6!ORwVEQTtGXs_NO3>s-Fi+QU@APql^X^cErUiDIe>4TsN`)8IMc8T(YETVZqH
zUcGC@t+(H7A1>b#8!f+3zyz>5I#iE1Cj+siIq%ZGrSyy7Y7GJrQbvC-IsMI43x%o&
z8^mSb+93BVQbbsEynzu#sd_Go1^>3Zk41(6vYWy`t+oRzA``+6qt=<6vQ8<$4*LSv
z{MNobm*P(4vq1r+xpnQf`!{aiX5UesA$yGWNO_Pmb#lOp;@p)a9(DwUBFS1KDO=0!
zoPB5F@7iOS2)P!yEbkp6GhuQJ^U|b!R~<jU*P&94<J0L$;tIh?yYDmhkph#ORHh`w
zTLViBUfl<?Qjh3Mu8i+)o^V0o5hxyq9azp+Wc@~UQt;Rmm$vUIznQ<fAF5zuyn505
zF!JFOoA$lst$cA%SJsx$aSF93_I{{@xZ6T^-`A%FsSG?@;TXIrEJ0_T{<55)kfP{*
ze|Zm5n50PPLrG{l@t&T<Jd2&@AOVQ2HE~h)19e?(AsLZ+S+rAR8k+9np={aS2ARhZ
zE)>;v%7H1$<GwBSWJPOc%Yq_tb;_W5(-WL5r(Aqz!S?kB2muH+#4PrMAiegdMU{R;
zpuxqmzj>!ruKazJK5}|eAiRyj6bI_aV;4wo{eTGJ^cRW<kvin52rRIJwI_7C^!Vxn
zA9H&2f*!GxJv-FAAkRk0A_H=GhKT3%mx<yObS9E$4x3N$YSo>$d$}Civc(oE(r3sT
z*^SM%0jz8F#)$Q|#rmwmpe~^^0*e92St>)qWfnW1$znfT?glpzZ>VDf`Y>jvklK+l
zg8A$HQg`z6)`~^+m&L{TQL!6cu%*HlQ%68tj{W4R5LtD~v+}EDqSm70-uS(a?4%$Q
zD#9^4dbosT>0F%Y4tYe%WpKb{Fk0JbM-5rrj+LI2gy5$PHnUaeX`Bh-Br3zKAy=YW
zz^kEc?l{}r@uYLMH$(*Dk!i=LEnuS)$K|j=JCXOy4tH?Hot|{s+G7aC=-@s+S-vAs
zoQURaSCF-9Pr54y<+4*rFNEn5$wEV#wbZ{b)t_$YN1=&qCZR0#qk-s`RKD_-JMF0|
zvhIwaM4Ia9OzgBX$vKIknslxU`QNs}&Ng$Wyi<|?K?kNA9(~)pUh<%wE8~H)Ac|>!
zlg0TMNj>x=57^V*K^RBdp04&S;&OGsc*NlY%)8L#l|*Le%Xq{t73+&6!vlFRe-&d&
zf1;DXo}OJuY`o>cn;yLBtuJ|@(~b$r2^10!+B5l!-i%zMPAld1^(0M9+r>ISgeW;`
zAw=)Plp7>#%;R>cd{$c1Wmp!6RFZU9c%1&Ays&sOuNd@#sa&wjr3Z!<+`Bn(&gs#|
zs-D~!Q4FNywCzd`i0W#yds>|tryZQP(yeUyQMi!5RG+dROeT8ItaFu!SwpO{{ZM@l
z_8w@(Y3Hy>-XzlMYTk@Z>oA)Rt-zZ6iIGE(j&`lU<Q7n%4pHb1EaM}0cQAQ#nt0Ea
z%lD<6!8A!Y8wd%m{gMUb;#Zl7EJTqJa@Whw&ki2GVx<lIU*!RvyW#`0dY0_W7wm_t
zWH!CSxo)OiOpK#K?AbcvUYigw?>9NQWT*W|^F&369%KhT;lR>p6(<xE0b2=nc5)|X
z(dh^<1Sm9<Xin7%`>`rj-#3earC+HZc%mCG)<nyGyrBW<g%uZ<sgO=jqF&M>m^q6a
z^AiK3QE(}tRt~Jtfii1zyHWs&76&wIKUsbQ<7J~uO|Qb>HCR;^*#h5dKUIZ#V(_JW
zw;f2?+XMHg6p{cbAP%eL$Q%3V@?;8VDwqXMcibyXf$7p&QReJt3Lj+*Fmp)<N>z>I
z*Ep@c%A4fdIQJ%|Rp*Ctnm=1HO~@CL8tJ5PI)LI&cTf@W=W?z&b@5i`Ixv>Bq`#-_
z=j-$CX>suVw9_v4L{<b`>1q3g{{@5LePi~?`*HfqOvE8)?AmAT7wa?<_(+yM&A3ID
z%V)wBcHVxedDVs3w~n#F1Jmq};Zk@6d(khKTV(84wUJX8zktcEzq~gpZ%W_%mC6b6
zd9jxbJAhy*F`2L#o#ZFJwq(EByi>JM7q3#`$tz5;CL(Bz%l2!9R;7oYby`v{upD;n
z*UJ-RiN#XKrgzBuFqw`bYQTP@A&$YQx8r<COtaHp*!SdaLL7*aEgvmWP4#@1WSIj0
zXPus)vm8#Urqw<N#n^Av`3Y*tQ`iPdAUOS%X4*ZBnZI4z+ao~FX$QK}|EM>UtAurZ
zpZ!kaT4__`?bZ~EERfJy#7UV*wi@)kSb;*Y>J7*M{ksK8R87D31pC^D(b0hAI^=GJ
zE6biAG^lk*jga4a&ZLFc4w-`70;ea4NB4m~Cg_6wet`vhWo_Z(b6~muuJ88j4=R4~
zTVc}ZIRKH?$gGaAoU=b{9?F3EY!puimWU#D29K5fQTe0{1XI|kSgax!YmbB*4O;$D
z&iF&V`r|T8c6=7IXhBlpd>Us7grQ%ww6QfZXMa+jFQyc=%&KJQ^pqhhlO#g>>`$9%
zQeKlL2un1+v?rY&aaI*pWjXv=`E8y#n6=g_43-#)rI<bK&nt!y%2Y<d0p##kkMC$<
zrumCHV`yv27rDr6X6$V&Ca1s1j3uaARywxD6ZV(o6ErY6lQ;{&ftC4B5nU&Brkt6-
z`Ky5;E9`m^0k*<)0CE<4J8pkn-kRR<4DN}<B&SEH?L4%rZGTgr7Q9zQSuQ!S^1T;h
zC%A|SO}p=s{cY)i&u9-|7gq2O;dxGvxJ`>vL*ta57-`$zHSgEHlN21$5MF!IA(Mr-
zL;A44uar47w9#qg4!4JBw4VJ#`O?I!R{TM)u&P?9D$YMPI8~Yen7#hE1Iro1nFRN?
ze`-dpvpeL3J?j-F*MbVusJ`+)*YCOgmX%H;XieIoQW`nu#Qvqg(4^2D((etdu!&RQ
zc)l2^Bbx4iZF)!=;M2l3C#K-^46rT<d?j#(57EI3|5gVu!4VCuX{W7kS>C%rmatg4
z_U{#6Yu4U%|J_bAJ@?b?QKa##RP-M?^=lDKooi{w>o?nfCNv5GaUIf!G}V*#U*(HM
z$W>iYu=NqAC$LciVP&HD?>fFnF_F8Bf<sbwoE|;3)4KM4BtJWn7}9!|uqGc@#1A+<
zfd`R`)GM^n;3N{c*w*c%j+EVC^|%r;<{UtwRb|;GMWzquqnk%*J!U9eW}b?P%{%>C
z1$Vt^`&#>$BdMV&%u;c>vXHA@rEE)4vP{A$PiHH0?v>E6b;CCM*t~4PHM!IPNn@1o
z`*BC=PNEc_X{TM-ltR#g?(O3XOq48dh%&N(#qdRawVaJ4pvZxTfIgu<2Y#U#^%IN(
zSS}N^R7m?1^VM71!)>Jf5vQH6R_+k5CPCXisUS_>1_}Ez1c?Pb<MgD^P@`+@lPktC
zc~22!=m6q@!HxI^6}kMB#3f6s?OZ1&)}R&Gr<M_=B`bA|bnj;>3H{}1D!rnRhl}=U
z1tua;At#E9Ik4DI|F;vHnj-!7>2<KGsn5Vv7q`MVW*R(4ZF76xk)mcyb;suvKsfDy
zGTe>|MHG9ULbQU9F_)il<k|G<$*!pAj;NF~;=r|wPW9TeBlejEqY+XCQt}uE+$2s<
znfiT_-ryZ)?XwbHx2#xkvvXbgRY6bg)kXX4N<O{u*%8vvXa|v3<4_nR*<Y9JbDCeD
zXa9`gY6qqXSi>R2L}+|&ol!E{W~gG{b0A4%DLP3q3vMRM#}XRH=jFf<bcti3k<`_h
z?84MPzfSIWYe&y%C8vvLV3cBC4wb>apiGyDo9Cvl6S+WH$=cRZec-M#E>+@HaAErF
zPuLgM*RC7KyyrCIll7qL;2xkN`=S9VQ`(IfYdw5{idkO!(`OlzAc7AE5r>G`7w7Ay
zW>Mf5drq@p=qMI3RKnPT8=h=mQpX+EMETg+5BPh~=}FOQ`|PlNX}wMS1{5o&M$~-T
z=}D>&_e%QHzN~?%=83*;$tz4FiexwTq%Uu95#=Rdp$e%zi$I-Lb^4`3<73+}$;O}*
z!>?#QL?W{`%Tjm3=_!Z}wLgpwQhrPRojt$O7j!G5GiNF|u=ouQ^R#O|SK+i5REopc
zGZkTJKvIUkcXno6$$>8{A2b-IECzJJ=_wzi<)D)URMRe!=@AhR*$wrBAS{KhbEJa<
z8(1D9K=E)bj~C^<@*I-ehMi_G5rtQ({;jG|As>dJU4A~+7on_QSss^B7ik&Fk<$~?
zmQDl((Ot3^S5^}2VGl+99Z>o!B0?Zz){0Id_N!MCN%@lUFnrOm0OzmncVL>k5pfv;
znPpMEvCbX4u2N7>Dr$^*xa9OCqaAwV&_=r{_o9*Blye#Q$af}f90CKU?WOf6Fo#`E
zD~a{q`?uQ5n$J-UPnn;V99XUc1<&!;a4lhil+{et2wq-Z1bIftfif~<6+1n8G#Tuz
zhwT;3V-yO>x@9?OBa$75k_b+}ZXUXK^X9eo%5pOiFWeUPH7)4+e5b!m_9;ZrObSQ`
z$<3ul^M-#CB~S5N5|3+7Qg*uQ!nt58a@5OvC)=~wE5=oW0a!sF$JFwMV_~+#VQ04t
z5GRIQNK7$MiUXE1c*Sdfru&2!fx8s4m@cYYEB(ML<bRN34_Le{ly<)NKzLPgz}Ri&
zZe+z+;zZ3mu((^+Z-oP&sJzr}FLxheH?Ofmz+(QSgKHyIf=;@lb9vlR`e9A1fm9Jv
z)rft}YkwMCa<QsT9Xec`a(QR&V_odV&UNl%;o4C_?K8GAVKP5b|4xT^^c+}>AlYw*
z69u8#s(jj-`-iqW*JY&<4FyeMbGBDCqme=p^9SBl{ZyDd>hz0@?q0ij?N<A$Y_Ntn
ztd>S@V%z7~)9lr`3V19vH@&ntS*tVMwb#^P#1<~C=bQu6Lc_217&6#v^D{Z?Ytp$c
ztT3&PY=KNlll^P2>px9t2r!&|4lI}LwGUo@%)Yt~l)fqj0!xxWX)b0E5Uy~j)#ZJ%
z0ZS*7ODbh5q4T2Cp95VBCAaOa!V@NUI+tl!-PIYHr0g3;qoL+umldfP>v{oZB>r-g
zy^Q{t18q-XpT%Xl=1Axa+oqkX_{hp+j<7;Sl^1Mn{RmYQ>^aRc8dM-2jaLy;BeT`k
zB_><bEIO{sU7UVda2Wfz-CaJT6slmyJmkP6yRTTe>SkMCh5-hGAp}zofkZyrC!K!j
zemXZ`cc;C_HkA7_ek+$J%OQBuVVGX+7j{!gOT$IU=(c+jsW-j+Cg(B(=8&x>TBM?l
zak`_u{zxcZ13`4e4FF91w<xcaWwAEqz?S!J?N9-jFc@4Tzyb}a<Z1vV#=|xhG(ci-
zrL1!YET?;Td=}5o=Kd|Qqu_RQV9Dd(KwX7<%h09p7Bxgf#CjH&yXf?n9i=<D+qN`_
zClrf`BY(?*<xY;fuRV@deqYYSx~*&1db!@w8tvKE=gj9o8j;fzWEHjuC;`8*ZEL<f
z{m*#M9aw((ZFj7+?NuVyd64Z9r=8pe9xSiyB_48reeiW1%oB-L4lMAg^dV&<al9wX
z5dL9rD7erLQ46Q%(hZr_=}%ltb*FmvK*}^#mF5e8l)X{nT6)EV%Tnji-dNwv7Yf%(
z;)pjpJ!y=fLg#dMlf5ZXNEwWGIoD;)MP*N%2FL8pb*wnR*pHaIQ(j^6rYcbbH<G=j
z&dn$X3&|q`Kp4FC*wr%=_F#SMj@I-Xtc$kZ6vZ$`akO->h`lve_e&pjuJqH+<XC5i
zy{(ejln%Icnu(c-F28wu?OZoKi*`se(bAW67`6^(intAXM}wBE5hjEQZAYCRJ+{M_
zW94m?=TxJ#kSwktiw<r(r@!!vEApL+d_jW(g?1ElNdS-_qbNgQdUltJY11tmuHS$O
zM<KNOt7Ub{Yfmz@-1b7Aw2|_g?2%?EE_MW;PVI?Nzyf7Nk4dt2MburCcW_#P8F)=B
zxPGznB*>i&=Fox}i*jZWf8a~8M4kS+9&WSIW{^iP6v%^i#49XX{PZ%i>{z)^tTeJe
zvEEjgOu|B%EK{8_yj0U=(o_0_4$Gw{sdZbMGe<}5p@Pa`7Glp!YkBW0Z+iZ#ZC4pW
zx`V=(%eRTOCgsx~)GZl9K}<MwZM?t~si)*?9Uuct)5UK|)wYRBVdGsX!!&bA^ZD|P
zX6y`xBAYBd!k2vSM9qRbY=TZtJ`i{Re(P3D-hS&XPNS^jzKlB<+&xtustHkv0ajlb
z4LxZq**?~qAuVCr-kB>f&eY!AqGBF^WYAkE{xWCN6_@uA7>9G8c5Vuj#4f7(c+Ywj
zq=@XS1Oq4w#Ef9mAr=8CRX(4Yg1CeD76^<)$9&3x72Ko;K)$g>YAhbJ+2(HTzw$mD
z@dg&rN>DRCw%z5vj1op!hp>R%py|l<Ai+!Ps*kP^znH9Z1jg>E?*UP4UGndFg(<gJ
zq9ZvZ_HaSFgmp16opU6Qh4RFykmU4>+LbcXvbl1zmOimP%L)!k+hCp_M}N>2TS}Pv
z6vp~U?!qH&mOOTxP^KJU-k2IWV)!w8S2JRY_p+5-0FM=xJ|yUg__B8w8j=`%!fC~m
zivI=9ble_oupg$LNczzDjyOFjD5x%3FOJ#8v;Ce*TJqNg2Py%pPEQgBjtT|B;lTWw
zW^KuSBC;0S>sQ92KiR|T&6~|Dp$&B@d_C<=Th%+*wcnfl5A}9-Mn<S<Lp$iEpK^q6
zR)$-9tZpj9Z1_8=jSoQYndv@ERmR5?3u%TfTQe}xJaO|>SQ@qWHFFnMt2hxe7#c4i
zq|WIt%CeMvoB^SuC?QQ2cGYtrslwi0&>g4W5E{-&I1VgFylMM(``VnH<-PZ?DLJjO
z3n>&^^H(gWP?r+(KhPkvU(-%^X=NNoVWdBo_8672sRG!uuPYzGY|2!{8aVC13f7*K
zC$j88D)#dA<@vIXs_gqo*lDhtFgiH>`AhZ!Rq|1Ua2WUO8;)cxT)apTX_v~@K-PhT
z7e>+85u9HnZE;W{i5xoa03GKWQ%Zf;qhIA*merBx1fwh2H&xO@91lHGFJoaIa(d!!
z?HCH)WLvavE_dsMOWBZR3=gC_JxQpPyKl{?#=3p5PRkpqZAn9nH!4l^>_e59I)pn+
z5*-U5_ry>t%cu~QN_eq7QGc+e-P=X*jspn9X`e_9TG8NC6SNPP?>@xORgjCgZU>g^
zccQC6xNoVeVhc4z$qArc%s~ngj+tSu;Bs86Krn;9ddR-DJa~awjQAsTMqnD2r~|hD
zZHeofSFLm|rI@S{$$Wc*Nhu~eTx!vQCCv_PT4Uc)kw=ADL@OA+11f}pdctbp6y+jJ
zd}p;!kMxss!nj6<9;wrmJN8{=NPYLCDo$#P)cz$=i(p;nE`9#$Nle+}_K|FD+fJ(O
zCZ`o&AhAahv(mX`s$a72uESj6ToyjcW9g>Yw|!4tbMSm+rOhLR7w*pfa@6V1se`Y`
z*U2MuBem}>Puou%??o0j=iPXPMZs$JFy$V9Uw*&~Uhx7imn(0*#lAmL&e14s=UM|B
z?qI&3wjZd|Kh>Qb9`mNTMccLQE#H&=T8SUWcn>LfPESx}vMKFu$#6w*N=|(;zjt{r
z8!|CFT5k@l5atpJNf_wj?xVJ^?4D^TlGZ7-{?CJy@-#v2Ey5PnFk2(GKide__;O&r
zrM<kD<60@XYzLY#Wc5_V(}pY3=}ARxl$u-8_79d@3$uZ@${B*i-RTj{4kM)6p}fY;
zz00}I)k8Ir-jvNZBUFU0sw!)on1h_2LJeuBRgK$)`Z~-(Iu0T-*)b2CcY5M()wPn#
zg41kmv0ywb*pNLZTEsRlLnCPnPQPe;h;*!7cDOu%a3eh$hJDF_#Zs08pj<`yTt${j
z^-=~yir7bTMBAxRcUSjerx`zz*+LfM!D`$Ymxt15OG$N)v`3xGSZcl2u8tk8_$bm-
zzKku|>4_0$6k3!yvttD&3R1iZXK#Cj>HT80EKNJ!JhdOJ8={W)dort2$D>5Fnw`i=
zBV2ymxz?mbVRyvIlXVU`%7Q#<n(<}QTylrsZ>I`OCQ8tW;qH`Im{QqQwB;mX_~{g=
zNbf`gUz-)b9AJ{!<Sdb1go`{?c^Wj+8JTU3_Z(1NPDp2LWqi1XtWyGHZ$FctziIW>
z*E`p8u5#k4oh_J1jVs(M`Wd;b^d!BpiWZ<2d?R+Q5N>;<?OYcYHc(SoZYO&>(Xpqk
z12B3{Gf6R=yz-u1vh#K8uQg;Hh}Op~t8Tv8E|gyo&=Xq*o*xI6OFxRIhqw*Bog}35
znYx@g15ezYcHEhar79={x!c7$Pn>8q({Y-)ZBDn_JMB^(Z*jAs-FP3S0*TtYPTA#h
zUwXmAP~^axy55=fJv?JqD&gZ=W)WksEJ6Js^DKIzX6LwbR|-Mo+fM2T9Q(n1ff(P;
zB4{|UnB75}_n`jtLy5iB+t+S$uH#<00OTpm8hMl*<-HhC?P}?l=t&v@xk)NBt|~`t
zBrF*E^NZHBck)TF_-oBCBDj;CPWG~<9<9?adU~icW6Sv=szK;n=Qcdj+HKb>vLWoA
z)2eeMp^-%?I;Gx{{c!!!P-~~t5K&CGaQ;49JMOxB$Z6z?SdWAwcJ(6-$z(_ab?v>v
zBxjsq@Cd#3qvgI*4+Iu=(t)Lf0vD~7J`}Db2dZ-O3Cv>w{aE>>axpb|xIY}2zOMzI
z?8hsC26L1@NXO^M`9M!hNQww?6l93ycoY%NFWOHe`tBqsl1n{sTNgU`$-I65F5_G$
zD(?c7Kh=yuB8NEIIj^t?9b>HUcKhi%ZPKN<FJkEl!Bzmcv@7~^g(Fid!id6I_A>>h
zJvz<NCmdLAQ92iUhy84MY_6~LWGI$K^rMs5=`RA`S?sC`kXKTx*l7*<kY(`aDhpdD
zt8VP>fJAsxD)l&pEIDI8UtT4vmfZG)#FTMF%(mBlN&QNMRsn#lKc~^0v|mURUOU1p
zbDgVwh=5u9#X4KDaShZxa(YC}q{`VZH3(8t2s+Z$1+TE&_d-LoUoH=;u)c~zPhq0c
z0Ak~{Un#JN%;a+~LLwC=PaE5@$i{xPuEQyvgGv0611UUE=J)Wr@m0i0*{?NIEfk~)
zAd`jCCPaUHhYkytI6x^c`}I1l6!F^CbDBjPo|_yY+8OiUZ#2IWCkdCL1C#KynZ&yY
zbpOpn@pxzFjB}me7yhC&NQyl7+i#U8m6R}$zUaV`tll8?+kU&;B9(~vJ_PyG4y>?=
zu!e-%D*a{2ey1TqSu3$ih1eZbAP1b&9|K+_xJjZ4WwGsd8zeQZ_pGPGI>(#jwI9NU
zDaG2=pk?M}LeARn<<RjO&v4#N&uN$8D&jgBDR*8yWWQhDPE-*t-AMBE3PVZ@pE`f-
z51LV2kYxr@D(V)AmSYsPKWwf{D`k)vYh$0aKdkS;cQ0#^^6i0%&aWNEZ1JT1QBKbu
zjMdKd!1N4mAAel8&{95ycDXYJh_G|XuV#NzCyn@QPQId9CMZqJI?XZAXMdU;tFudl
zOCwl9N*QUUk1{=wVx)10qw~+oXymIWX;xK;9GHk?1Um=kd)uEU-(S-mo|W;v=KzY>
zqwG+fWFmu&vijoM{vt<4$_;jO3d7Kj67t%7B@H&EHNfltD_r3((`bMHjB}OSBY9s_
zQ`)fqs-df>Bpcg3NI)Hx1^bfIAGFAMvI@Lq^FL*OT~`~PD!hnNZUEh&i4%cKUXTT<
zUdyGg>l)VZH~BfQ`by_2rdxdAr2TC}5Fw945c3W!_m10F+-iSUCX6V%u$H7-1%~YO
z1STVjo|lC~C9EyN(sBFyI)~z{L7!+rxzMrmN#w*>=G+xyMNnb<sSx}_gP*Jzr3?}g
zNwovfpKF!XqGuvm**{i_+|-)vP?m{yN>~&N8OF=4VrKSF<&$_A+J_z~=dMf(q41PX
z;;tlS?oxEkDHf4@_2=!Mb8&3jvUbS1j=K~gDfe>ZPcsJx?O$?3Sc&1GT&^ce?_U$z
z9vO8mlaY-Qm(sI;tMAj~^9@R`OfF21W`a0TY)AjzY?HFEi%n87>VJxg&Ru0CMq$TL
zwlvTS_8)aWdy_IXX~lg?7?%s|RBR9gecOLF)TdC52tIRQMIwpw;^A&8SP~kKh|K=0
z(gJg#?MUYpY4hUuSZ<Gd?XSKY1zN~5t?)91to(PKy*O*4=d?A$%X@WTM0cXoLshn=
z{Dy%d*1SmpzDj*K?({2gcGG?KQA@ES-#fJ3xh%tk<?F)`Tn<v2?T~%+QWBA(HB`Ys
z2R&7o=tJH^X$v1yo-lZjfQWs;=?OBFO-6AXr+MnfE`>ZcGSZ!Kt_zu7mN!U2{`PT8
zv1xLgkAlFW-Z-GZt&-xk+I3iS-{O1BK0bef4Rv;eqh|3FI-vMTO6*Z!8D`&)g!_c@
zW=f&NqB({>1Wb}5C$+Zh6PJoKOaXO-n{fvapDN9@l!CNRDlkzGmLZnOX$O{+JBHpm
zW1n1bBW+(eq&@~XCQ8QXNt4MG0Vc4#PiY3(-%3Le<yWv^@AvDZ^Zbk?BUO$%W1m`n
z4-5h;0cr!bv)Z3JN)R0wQSgR7{J4Eu{qSwuH*DPKGy{S2*ZZlfNt=-qVxL}b-+~RW
z?KE?(7c;Lctq~<{&s&NWir!6_01aV>Ffq<B=-Fp16#;U^s+(_dTJcJ4qf!l{I$Lbu
zn(|}zna!NVf%cfFPdc#t4I_;tMT7&<CX%r-C+xGzZ$vs-!{ML<D-<smP7SxjY+;{W
zDOswyo)a4menzi75*SqJ|3bKRIn&TSXQ`O6an^8v(jzH&1C8$z&Fj6Y_a%#tT%(YQ
zeQrY+=fW%=^&<H015Qs0jT{mOnr0|1{CSla?`ci6@GCih+<(?|XYBKvkB;pPGlc>R
zh#hx&f@af#r=;n5`+@?~<|cbQxt(5N@GJHa#TygIT9FjmO`>!j^_EJn^=sObBuz}s
z*cX+tiZAD$syZ1;b&Dh=ul+HLz=$N~#YzK}o9eWEarq=7$%0Gl#{v#4<|u>6!<WLK
zR15A)nzu<jNdc}FvY_ATm$Ct#>`Rlp@Gek()M*WnC?BY>M7c%n%PK9yN;w~jW_h<_
zw9+OQ>&?Erp{BaQ9P|nk=M2Vx!S^fb{H*KFI*nZDY_jB@+w+snEbo2GJDlr6ipr9P
zG(xLHg0dGhNKq9FWjZc8u(&&n4h&rkJ7j8mVS^NH*x55^8Ls^)k)g>2VJoyF^fbGn
z4h%v~@BQNrEDZu%+Yw)AFRC|@8Ycf1;Rg<^41oQeSyjE-)XBcG^vJ6pJ~1hh5_p{+
z@!2+(H+yl#6anK-LmeD?GGZ@jejeop<y&PcB#qM}9zwZ{Ve!oy>*SG&%DD+N%j8Kt
z6y;az_f1I+2)#Nli&oypqEy9?WGPakU%FJK(Xn=(pbR9|k$h;XUP2uYds&?yhQzc(
z)bkFkd3T8w<H_>dy}S@*yX9O4XGq=~!h<GJ%0`$|UeOGk#mSIWNIGR5dg7OZr)Nn-
zkZ1Y8)l>G$$~hvUm(o7F9FTVW=RayU=jiVa^OMsO{LQOvMNU9#a)JP08iip<_mIDA
z(QYYkMOLnU80LOn(Vj~3cKTz`5w`&e6TOk&YqwSuvEQ+~Uw1?hVC8NqZE|-F6FFhG
zmG>Y<BN~aUReh9<1l~p*pVJ>w`8?v%VY_`PC3`U@6?{1psz<-dsf;#YPliY5-gu{s
z`RT@&13=SqN5v?dMM1s#jjVf;E~UQGANY{!h#U;A7wpcuEO6W?rpp1O_A*D3)))>B
zTbVyw-iu0A&maVrHhcJ~d<2Rc5IrkkJ<riNg7;xn`5@H^P-PV{)O3jjBXs)3_*HUf
z#$Ht^e<wEbT~0GTR!WUTxWgp)e^r5r+ewgDMMYqRN-6%xeF`m$FmZczgQyG2BLW>&
z%%(@rCQArmMXD9<x7Sp{$Yh*Ao(NfXci6hqU*3bP7R+1Blqg7hZNWu)R}gZT7Fm6x
z;1O0SJaFtY+!f?rWAH*74bedS@3g%x2|`D5bh=JUY=6})E9|Q)Pua41o6`ugAs&P}
z#jBeUN>|EFDi;?$x^H>!I)oFutGQ45RT6cmk4{fwIm+Wk;;0Q3m@FG2FwUiNVDa@L
zP?1*F)oH@PZB2PYq$o2B3LgmF&*@1Ca=JsUep}moN?M&o;~^J1Jz|$sC7rf)OV29V
z=h_G1!XJWfd=Me^gM_dE*TXioW@g5SA-y}Xbo;8?yj<S7>ULXS5w?1Er_;!3V-h6;
zWsI$=-*2cxhrpDx$~kaeVf{SV<91Ko@F)PT3?<s-i5DX(w6SvA>+{ZQqz_+vf-ElD
z14fMjc|8rr#kjk%u0S%{JH-DvfI?}fr18Nn#PiXeZBveHxI-$v(}<~y31><atG>A+
z`vLZnp3{t3Q>TQKlVRz~xsw6$-by7yV;i<Ot$1Ks#KJ*jTk5w&Y)tqWp@<_elYUR=
zeu{454Xks!uRitO>CULr5+&cf@>bj0e2#>=TrLCY<4%7eMivYsqHLXQTgpHTnQOJ=
z`|da`G)3T&I0SkmBkS)`=NDy(T*N59y&>EzVx)r#()2Df!a=9M=q(+wKf6Dv@Y)lw
z7Ovxa^66SIr|k^|A;1ti<ZThtQG^~nDa8(L-fj;xul*mn#(bnK{`8kQR<TjS{IoGZ
zf@kfG6?=CphlN&|WF=>69#7kw>Wl8#{CcNRsIjR{$M{ge?adXN;ck$ua>;=;%UIbE
zhm*j<x1vz+x763c_*C3oN2nAS5vuLUJcXub4_47;vg=%@;U%5n<1)M1TPs?J+EnBl
zcG_Cp_b4m#Z3V?o!WFr0rd%R}=;!Utkm3|CJfX4!NtviU;!Kn;LEV%5g7e(o(eU7=
zBq2}BwAQCXj901-cjbDJ0tTA2Dh8fqt<s;Qnf8joBtrW2waPB#EC9st|3ln+fazTw
z_rH_gyVHwfLTuzhF`~I7Rs&=tkR=&{!FIO1k{0dmyWV#fSP01&(&M-#j-9SvKwUx-
zl0eC771Dd}z4zX8Ki`=-zg_arll$D~x%as^+Pr7xSI+O8Idi5Tr1IgqC%Lb@*D^b(
zHon{eI~{y}A%m$DiDyXX9qC4A2+!iQ(@rI6Y*Ob~syxhS8`ZV={2nY9F2E$#Y5c3y
zvFo+vKBY`arx~9o+m-4sNp>lnn^p3ZZOMVZTLxEt=LRvQaL27R<ZVrAm4$xdVCt2j
z3EH;$9IXjLft9J~i!jwPRPy^N8>{QmjHng&j5nB8o~#FG)%HL)<eE4!9oaNBaS|k+
zaq#NrAOkVB=NGJ)oE&m~$G~Q}66sV&Y`mT(=}wjj8LV>khW_l}^B%H+m>`K3Hqpq9
z5QxZr45dYjw8Kd5;yFrf4t8vlFBmV9-}s&o795OrC_-O`re`OtUl(c`bl8UM^Et3!
zk}i-TWRWJ5#HQM$Y@p<=(~2wNdWID)VehEt(T#|eNV~hNvkeo&s=<Zo<E=K`K2Y*f
z4N6K0V8pV<$40hK+n^geSvxWYUzP!d@Zte>@Uff8A_Mo1Qg><It>4VuVVfyQHZ3xu
zCuKEAmCt=j;-40X%Di)JeBk@I=6$Gz?}fR|=5}$@tCu@JlSDU<5&HjNBgAzC$Uwlf
zgYjuv)ZOm9PJ5{F$K}WcG*a!mgGxqHupVxoDUQN8$%>ujwxwa7a`4e9$wJzF;W0MX
zdNQ`5KkBsN3I(|+k4|3WN7_3%LCk4r%X`HuUTKdmd7FqUE9P`1Q5}3f3lftT>FE4U
z`>KRt3#H075y|Ln4osghGskj;Hs<WHMhFPJiW*4!gi^G@<cgi>kIs(w?W?<iO3_WM
z3DXBwiG~U70ck|ue(kzv?Q2@MV!u_`*Kr3X7EZSwHgBK1eQ4Ba#7s!l;4Jt>-fZt|
zgRdh#RQ?x$o`HRAxjkO*$S@;0tuA_lRsC}&<cvMh`oe0J*8|zX7h%Mym;3!|>$=Q1
zjHm{0Fd0S=#Nk$l?Ca_}tydMFL@sn-!8w;uG8ipZEwo+s^>xod_^1?_qeyqaiU6hU
zX4oQcMJ)e@x*FLeHf<$J0MnO`^ruGc8&h3<;~Sk{IsK(>D6H+9>bWKBMA?C`7<t3M
z(IQ?%j6*BzN&DuyCzUJBq@3MsZw@BjP%C*sRRzB#$(b)aGr!+yi3vRR*ke?Adskhq
zK_tu`HoDszOk$2c^!WU<_O1WJew)6D)bf(Qa8(RZ6y#v|r5>kn^6q@k!VVoRO^mYK
zTfnqz`dh~BJxMrWHCIZE1BfSy`Lw9U2r#ktR*zNmS@{|=ZvxXN-9Cq#+TPbTSM*99
z^v-Gml>G|xuDE1on4PirH^LqnJm@rH?9xIZb#uRcTTx27Vzra$+&3NYE9cz_@9@|Z
z$U&kYkQ^UM3lC$JeS4#{2*u;ou|uV@NZzi4kFJqv3AJlJWr32V@B<wy>1+X^LF@!i
zsc{D{b~!vfIz6(*zN6@EcDU#K*1!@gWFKtbw`tSuP9we#p@fN1ZJ?9(p~mV~_9+=g
zE3T+)RMoDq?<|-cv&vS5nK`iNbqKfJj;pY}@2W{9HI<+FX|OEOR+hU^Z0~p1BW0MV
zBdi@*E=dGerH3OO^D=R34eCDoo_bpDcWJ`OMGvMry_@}RKX%1^_PyP^m98w_-ijZ9
zk?m-0-@dPVVn$YG)m%1x2NO@sl3U>uw(l=k7{KY@ZlYxyOp8sm0~F%DBLxBlo>9-j
zX~v@@Gh|%vS=-sz<eN4T76r8fXe^vGlb-FWvB-rVOFsAZ80td~KIO-ZsM9vLyFKD|
zPK-fXqOH%o^={kKghpKcqXVberjD_duq1ti4cqp1Z|$s(keW91jD!TA;~ZOhh=R7S
z8)rpSu>>k~Y_Ef<`CL3ZhPN>rQFF!i*VCk>N;%YV*1*zsuyq_k*$%Xa!^fp8P#Pws
z3I|jCJI`TMd0g#kau!RaI9Pbu>02NMlzaK{;N?es5r353R4yU;q@p60WNCY{Iww{k
zyD%pPIk0%AJc$HM!E+^RE}BH1s(WP3Q#y^bO<*}u<GtDK{gF8o&O<p->j^xkqR^bv
zE&@4&d=eX>9d3(~nod2Z6<0Lh`l0zlcBH*5{J(Nd9Q6iEqieV;>}b0#Ji5uHfyIY}
zw5X2@r=9|uMCoVjSoc01l%$}YgWh0lsklF)wv`=ky?phx*I(<jQt%qZyf^udSv!$v
zvS|7U4YYDU9Bi|b&EzPXH|e{h4kU}caP%n-IqZYmLajX2u3S601^*NY%RqvB<*!hZ
zwvJsFr`}~xH{v17i&z|*Y1bSY%>s{hx^9)jQ8q1=P)`#pn4~)@>GU7AGi@zR%*naw
zv}3^1=B*{8B@xPK2<d0L0gc5JjtvcwGS(=P<KUAKDujnA&MZH?Gxkh-vC7&%>@+oo
zBu$Vf!D%aYuF($x@6)rSQI0yGc$l=!$e)C$E9^CS{VLSQaX+7r!nIxnEiTl3;whvh
zpqDBMQE}v8g1*`jY|p_zpUvrd{S}Ws>ipJr^4MdqwdXq0&|ZKnMNlO$?L&0p(83|R
zm|rwTw3v)%G;`5p1Wr~IVd+FgOkxp!pzx9+<w8FjaR(W#LsW=2ycV4TI#AR2+ijxu
zgKesa2HMW(&@S@OVWgU4`CVjKny;>hWephBx*b@VaFDPs)dkvvm2`e^{<H(rx8a)a
zjgHwu_x0K;h-0LK4W_bHPe@Cef6^|u>lkEx6xE&_Sagin!=*~alQ?2O)IMQic0jQw
zr>#yYuzS>=uiH}AmPt|ZV2Y~&lk4rFF%FmI(jRU^S%I1=d#g*8gQ6-K;I+wIl4Sfi
zF1jD-zV{e9U*@s`i|JCi9J%dOUNo1BqBeKgk2Y`d5GM3FB?+RNI8YJL@K}GMPkgeL
zpZ!?wKQDQM^GhqA1ml7IcsB}KDNRO|I@>U*o5b71t6@h!(MSj`T$wbHFjD2@9ej{)
z@^`VbBleR86IsbINrIyT%SHZ<*)cl!Q}tMMbc7*k>L^1nZON460DHq{Cnd$Inpse$
zsMvnGULl1-i4Jn-1B)R}<@UJ!OzT?k8gbXB9hjz<%>YCF&n|u&d`#&9dOh|WUWL@>
z;ES9U!xDN(EP`Y9b8Y(Z$&QVXO9(*k-ZVQ3IY7=o-!34kgtO8n96&xxz9XKZK(F+?
z{X(O_O)MhHKGDt%C$7ou`WL&N>U@jXT<EWIj1`lA!aMwuX#6F|A=odq?og&+D;0(v
zKz&*UgdD-#8}fnua$dS(dTWLB0nm`GrmD@H{YrLjB{uOfRh#ByJO`!=dIQdsv0ts;
zr?3&4vIC2kHLG%-o&i<ZujQ>+JEw-6MqHNiTn4E~%Q<Dg-Y&cGx@(=r3-$vgDy8%x
zf$<spjp{+NXslB6=NwpalIfu>GxnQpgo#vA4G_r}4HmV?nCt!XyY08CLt=a~w|Cih
z4yNWvM>8s*{h0lBH>4D3f`U{~F)Jv6X`Az~O7Pn6)a_D{*#}UUPdKpn`rILSr7R^F
zy!YDgHUd!=%EX-0&bh+^fLvXXgYq!^USXs_kPaA|q=x34gGm{8Xp(Xr)AswV9pyWw
zownw1VaK}JN&AEDF&WS<+hf=b9E`Zh!j3!E+aKnN#+()UB+Z<?%S$UxD>nF#3d5Jh
z)-88yFd7F_%wtbDY;{ujA9sc?4zuIy<it&5Rx`Vh#N}L--?L_v6u43QlX^#L&3U{e
z)#Fi?7BPx*6zdlo(`fx^caN$L9AwME84SH$bTHa;Uv=gE_GkHX#r7GmpLdg7@aK)2
ztx$%r(~R$+hiSwhB<e4UOjF&{TyQ?U%%=HvZ+Pu3E#5FkDCN7qtc_jYa?WWL!O8L_
zCiSHKRWn*+t&LOwDKh}8E2u~@%;)T{+b|<~Nq^;VI&Uz2w!G6M(b?ZLp$#wKzDw`M
zx&<Vpi=AxxC+%-*z;Y047llywJDBuJ42@E8W`=-5sQmA01Tvf~7AMPMF(d(eisvh*
zXD0^_VHUN&ubyfk;Ke93&p5E0>XisnBwPJMTgw~AMte@H1||=P(!~B_&e^i7mpMQ4
z9m>p7eX*_fPaQAq;&++$sAeb8)WOFSbJd6K&i=Vz%E8`ql{WEZ4y^c47a1rP3W@U<
zIpklu_lWR<52e5hIb%O`@b#YZsj;p7Yj=;-b?v~+;iACi;I&LgIL&on|JLXixkTB!
zsfHb&Kj9q)I+2E--Ymit?cdu(QN6yW5CqW-z~~7{Ch|ngKRJKE{v*fmZOg9q`Z>;y
z`JeU5T#)gd8apVUVNz6*(ySN|MbZ6N-9zXj^0#u?FGeo=9tWeZ85`+gakj(BKiZ?k
z|3ZLpFe1VMQaF;shaJw&vP)=KL(u~!wzT|ulGQ%E!>c4%Mm_*bIXL2@<CqjiMbZ5k
zQ}7Xo)7V}y6GXk%VNHCT689w59Coe>4ktNKKC+8aZbOLt$a=n}UsvlyGG!zw$y_Cd
zz$b4E6(kpqp0d+Es$qD8L=2+mia&H1WFtjooe7Cr*$f|jxPk_UX17XDOM4#BTqc{z
z0-osG#~dyVjCLO-tmi;-8(P^L-vR~M#~uzgw}}L3e%786VQUdlpopa-_HkVgMI1}C
zRtKYb!Ib&-NAI_fuW`?M7P$2BAOfqW#vxci$Pu?}pU^PueJa738ULpc>A^@))Lbi0
z<rBN<jDmp#cN}zJIh=GbeA3ANIT4?9I4ElC5Eb~GMve&CA)H=opIoyX{FRwKhLD})
z8or2xI9JJZGyM+xlzL6>#i9Y7Zp3L54n|*2tQm~XKDD}-4lwcNy$&q*(_3dBuutnA
z6osB354ok35(h?I><{xC7^i)Dn>5@ISqBA3tWCMArGly0^JlcHNTUet!+`{!K0uVw
zUVBlt#wMp`Ny(s<lYliPmK4h-N9!|l+1#_OKji$*B?1rB7M0$M#Vq2_N?K7uBR{iC
z%vu~NgZls31<NF(3_1o@E-clClgKMkDZRk>eyS6nQ+I|6r1c0h0Tx}(jb^_hdcd$5
z*yq;O>fpub=0yjV^Q^2IWP#7CYqUZY3Jl9}V5xp=>dnkf+2^-0!E_^!n?f36OdL$|
zLS8Dv7!h*z1vNG@n93{~X>G7u%=2*QQx0CcUsR=zB9>dx0KTxTw$&6_<N#j>Ai}z%
zZ(p=X4~jpN)LL%`lb<5$f23zGZiB<pk%5R=@eUR?Yvk1V1NOx^lN<>%aDL~f!Z0~S
z_9flRx-dar#>#du@u||d6iG$~rCm{1?~C$_>0^)t^SC#Ru5RHZsv~#EzNDU`00+Eo
z$fM`I!Qh_~Lx>E0&aP}7TiNf=Y;&6NdQ!S&dXolkU)nti*Bqm(KxANPO1gCn=RAQn
z_R<DH+6Ym@f#_M`;E?=5@~*wCUDcml73td!BtMw$ghzb-DZ8qBV;q6Wj&;C+X*+ql
z4&qtXt*mEAf1=gt9w!(wiE|Os8DzHGUf$+!3jT_Dk#@3NA_hpwo3mHsGP-lkz0S|*
zjba-msp(*hSGGQ{nBvHeA*cpOF>MZ*;IvtLRRhe@gxnoKjt{dvPBH9(UEO_W97U--
z3kQ~qaY-;n^F(x-y<JmxD$FD5?V-5?tIboXVd?FP#Rxqrll|Hk;G;$+qY<SK<gf9D
zPhv`Omu(RXRxk^@t|QP)e#)|GiuDCkJSp_=NO<AcsoY{w6<%F;N(hj&7)B?pXa`e$
zmW&sAcCH?NKZQ^3`okp_t|Iyo4IXDHpj<Z44oqZu&ih>t6Bj({w6u0NPuLB~*tb(N
zd5>3PVaL$+8N0DPU<BMwrzw_<WUdIEo7$9=`hS*zZ7?lsb)V!JySeQfx3AdfG=f4F
z{fy`BmV%`Rbnn7$?A#5OMLi;b(q5AfVy9=R(TE4ZgHcLjKeN|%L)1=Pqz)8TO|c@G
zO~Gpio0^*@p!#*$co4t9BTgfpCaX)N#Le%v*Ehmn#}R3iKcKC<C;C*Mv@c7dzxleG
zoS#}+mp(9m*1o*m17*w7PCK~1bR|?0SRQJ$bYpu%K44_q!1<|RjdY!!t;p9b>{!11
zM(-Lgd)+m*vZE@w|79_W(sl4pFsTxtHsPSe0B)<=)V*nz)Ly3<9~ejbFfp9pVXL#V
zs(IO)#_C2p)>9e<^RI+S1(z)DnrqsslNAyS;v0RDh^154;6Z{oS7}g2ky`B5)|D08
zCNSL7OgZApM<2D@TF<f5j}Kww9d$rO=sIzQgV{s!?6S!0_QGdnv(!%`e;xq@w~K=p
z^VqBkT@jmWYir<Sg$b>Ky*szYhAH6yIY&X(3VwTIV?pc2Cb0vropc~2bY$?MQn<7W
z9Jf2#kVKSn%2q+LJgY0m?Oc`jhuE`33b%DNIjK;Qz>F9LEJi`%QAO;>lq=JVtO#4*
zjZ$Y}45pBE6`ytRG2!Y5F=}!l*qv?Um*0G&(?~8%&}g^@?rOcz0_0f~Aod1}7+%I#
zCfg159STm6#b4&TeVB$WIGA{+v=P}p<-UDW^;1k%!6B-Km6n6la~fW2qip!xDTCrZ
zDi`dujcpz%!!SCm7=9o{ACkZc1PkALCkfc5_F*fBhUdl!3jh!|z{jNvqTXnyjI=|o
zA={kvIW;~+uH2;4%rzU#S;k0NSI8K*H@6W&qSMJ&q#`@8npMS1sn2+fLxZ}z^?cK|
z%;I$*SBg`}Q7%`LLT+-yO*)B05uU5c3iotlc_s|G1Xnt+{Bo9_f<_}06>&|Ajvm=p
zv^U<?Be`CiHIN#sXR*~l<w!G40wTxT?rmRG0SHgcpL1ZzOooR>&@#pdLVHUa(rF4|
za8Qr~$ZNI@Q4pG-qMqH?u30sW*`r6*(%fL~nz8$9jCemnsn!~`(&)=s2NM*do+&Iw
zu~KhM)<G@0N1We*kEjgpq~N{1t--<uF2BoZ2bRRVngUTOUiS9x!8)rl-0lu6w?mA)
z!vp)uM#oT_?CsiwGa&~|yp>%e;+<7BaHvL7GG)YBY3_nTFbNPD3O?vSCiigg!nNXQ
zHe64`9~#^T;nso0yrhevm1&UcQ3mDbY@~H{<&gY^H1&By<6Ef>HnVNQM%$J_J<7gU
z3pa$U3BBrxMl<aV73t~?lNs?DpcgufZE2k01~xPrHRwupkJv=pS{;rl;KT5+#G=sA
zhG-BDUQBmYpR*c~K1e>^)_wM2hAQg2dxPzV#iY4muZ=ZQ+BC9lvWNX^+5zQkQj|pp
zyx9W@xHnvYqeF8JAoq~9TUXfj!_U8ncfbCMtFON1zx|(sOGy$+#&e0vJl-ure2pTh
z2yD;HwoHBE_|Ako*TzWbXrqga1V$&DsNse2MJmMp1?5)4>Ku$ZHO!%k1DnkKy*D=G
z{LJyJy;M~#`*|(q0AjRhB*%#vOM67D2%BoRqN!^xIXKQ2VRFbtB!#`BqfYod-O*85
zc}7lz9y$2pQfJXvn_ZY<vo@Wpt-l3VIKQi-Bs_~V&F{5A-6@Aa#dWZp(2=}h5ZH<X
zN&kI@(n~Yd8~KDKA}HKQ%D(blltu);hV>cvLGruV_HL}%iaW*V>A=ziyA1|24S(ho
zX?w8y6dk97TOcSA)0%^c$7*9z*bdy@9%=&F#{M`Ph||u!faP9E5eQ^mfVe%J&rl&u
zw4HDlIj|(h^+U7pww}$E#J7{m3;eE;)t;5wJt7b$+ebQ#h^wV!a<;g47_L)Rh0Guk
z;%bjJx?Z#LZl{sVJGRD9Ss8kEE^gtw?W@|5!91n2Nxx{YG(D(0N9?h#BRYgM%y%)s
z$Gk8L-@LG6VaKJ%(S_=teRcQjo!ZvW%lA02k}^3PFUuo*O#>-2o$(sXI3RIZvYRO_
zK4sp${CLRe;{BnQI~aK+Zo2tqduK`0%WraiE*a0N@o@Q_DAJEtm$ZMx=@VEX8;qdN
z*mpUB+MejvLfUeW=J?!9T~@6O@JZ#dfEjo8oUg5WQpppoM<@<!CDx&XPfDcD;uJ;o
za8&o~>#D0-ZpsNc6eb}NZo|mfs|Q_xv1=|)+SfP6zG_ZBR+=)7a1!=#e{yczzM=Y|
zTk<F;JS-@#%FUu$ls9}zNRk(^B1M2_${gR=9zTBR2{dGy`Ke8~La_Lz?iCTU#}ZW$
zsq&uBA9e7_T34@FW#7DHTnJGi;!f|2dIQ=59DK}{K2&@sPqUI6^jqrA%;!R`Mq<iY
zNXfh5b1TUV-Q-)LOxU~HmvFZ2gFUBNWFe2Gq&Jyh<cj;&!YGWcN_~RS;R0Y{{m9F$
zeM$^0Pkf2BWbaO@d+Y12aemh0wHBq&S8-^2PYq#Ywt-Pd>>@0q@u@gz@2!{ZMbaj8
zmIy<=QruQ?ZMz(NOoilwaIelkXYXtGupKJ4MA?r^DHVM5N-}}MLn+(9GY=3%`u?t0
zN_U{J0J0P?P28Hv2l{jNZQY~UYgup#^Jt73Yp8?oc%_66VhQcr3#Rg8qzWMx9dlsG
zgP;&a1WTQXV`h0Ko+bN0<F6Cs`ctikwmLD&`OKsC9j*T%E@)M$MlOe~c*97Xh!DH5
zL!X;JWFPE0p>(^@uN;_^$%P#+zrsFL9ZDS{rDu@?HAvZK&N=ub-#bWe`ObP+B^5FX
zyBt{V0E*=f%8w{YlnAMh#GAG+UQ%^IgRX<*PBS9lW8_$(3b<!uK4jmWFO7U2rx9v&
z2K}-!B>r$&hLtt#;Cqt3RjP!@7pED$3gxrbW3$C6rssQ;R+rs(rSmg7ph(IvvQc#5
z`)Z^q4~7ky@c@(S3`}lbB<+knf?W3f-J2A1EG0(K7urixN(QgNCctCJb~G__c#J?#
zTJd5gB{eAIn4R5dh$*AQQKY@W+Qud$_pZ85L;!7Z-DwAw<XzHamtSYQ4?jPw36FhA
zd(;0{|8w4?Z(7(fGBkzy7oiM$y1wpVN7RBn<iK(rglZZ}2V<x5rm}MGw!KZxL`4ms
ztOLn%&Sa&W;omX-U9^4eeH0l|{GE6kuowqyVi=#qt(rT3e|zdBWL0Krj{}jdh=rj;
z<`2}Jxr`*1A&bTWYnaGd%9(|M>_O!_SYIbkrfSd}!4248$sJMlajQPru2W$%C{j5H
zz|vC1x$Fabs?A@wlEjnc?C3Dz7fRyK4eC&P0*;L-r<KzAs7flx4u1IX^I0Q5>d|2i
z8Y7Nz;DA%~P;G4L9AFDSa`^eRjKnoB|9|N}=Yf((whkujX!q@5{bGiXIj~xxS%5(-
zN87YEM4SzFtVSd9OBib<v`d^g>R?jue4wYqo8$HHw1^yJt2*Jp;^9lVTFAhXl_0MZ
z-91VNi1qCa*6Pp6k9x9=Q`D!Q7(-PDRFr!N<v)~F{w|7~r;1&*jd;=XpA|cCIgvNb
zRk=YbHiGX~CYz*&L-ustpVmKpfH<Y}Lj^|2((*K8z*Q+gP0N@X?%jc%&Vk`5RO-ZT
zb(#`}Zs)khah0z+lbi_Af86<<OponV^&L)Pc01b*T33U%-W*)i!55=Erz{$H&Ix;_
zU~=ba8)hgSSgt=zDbfi@D9Gp7xklk^(8^Vm+dB6}FzPGh*qf4udcJ$Eg#D29yunhq
zNfXkAM)B*1wi2yCVk?k%yJ8OVWRJ?6v}fB^cEfcyI*nNP2e4!xv*%twjJe2Da)@A4
z2cHX78nj(p@_{-cQneJKe3^rJF`u!p<5mCLfA#}y_9&yM@<!O0XB=RX^(ve?J^R6K
z1R^zp(LU?IlJ;aK4SHd(Rrr8}8H_6P?F(<4o^={UV!87#x>GJCA>e-+bAHEFl8H1K
zo%PceI&xUlfthq!X*Kl@KJTpUriHaUXP5K&v*TF#`JE%Tb^w$4p(5_4y+v66;yY`P
z<*H&05;1cN$DYK#A8Xv6UmX7!hKPexFC0wWsrsq=FFy_eW)z3m_&NLG98_iU4bP5^
z%U9^Ya$}jD940cp_u!CRgFli(DQoq>Y2?(bB<5h;ezc~h>Z_>=v;#{yY0ZN}_G9gl
z_z$*lOhuL%0~(!^U{%HUlA|%~ih}v^HdL#&A;Hk9>65yzfzQLDkJwK%3|>*~I1xfE
zkAZ{9!G2@EH)lWDO;DaA%&9W0q;qJsVDfuZ<fdmo)m<nylR?F@HJoxV`M|01-oy6O
zjU<I*Tu&>awte@;u?OvE+U0i=(eJcUzF~rOoDkj&``LO`IvXN#wI;n`Bpb>of1a=A
zFN*5Vb%ZDdRBZeOW#_%&A>G(RA`&yki%BPh{pVY+P^t$`E60@`MWX&LWD+g%{q_rO
zRdUQEWpl-c1`aTH%8k&*)-kds?H9XwJb|5{lZ8WgBzM5UmmhVp07~r>A@z&nowHx+
zNL3{$HJH*7x(r7LFMs00EA5xtYz=MGu+oeV$aRqpy|8iFuhf92MLIGFyuqZxDV7OK
zmi=n?Y8CfM(n_T+#S-v2ad5Uh!F-R{uN5q+yvDG-(;H0n0JY8}SpRyHkI638kyQ>L
z2av=<+KO%cji%&m;m}hAVcIn%M~PsP*vwn8mYufWEaPI8o1;HW8PNmcw~-W$e^J&g
ztliGFk~bb1>)UU24-CggWk$<bv51j`t{qneRaBqux9gttcnc%egPFs@#BgM-V6ppy
z_B)NH27M0qqE%N!k`*%f5&PZlc?xRL{7SZ!X}e*Pr8PMLCTqW!izy;~oJNpQUZ8Jm
zmA90ds|7_3Nc;UZ-V?C_4je$-CRs`9%#&d#P}aDZu2SAr&HWD=2@nRXxaH7S*i}wx
z`Ooii@M32Ok%S!mVH-}UC!;{90Eggkz`<y8ISOvJ2_b*fvG}MQ5GFa<doh#BVsq)h
zWk-JDs-%ixQb3^;#FIa6RB;a_Ifou{+Tq>NZA$akE|hyMp8$8)fu70~mv!V%a>pUp
zd(!zWo~0r}WOMy#O^xU~1<0p-5vGiCMNl8KKg&h7Wo)bJlse5+ksp58{=7b4brh7C
zql~sRsDcq6y}dt7bXL#)qAtv#*Iq@62BDIBykXMLB#a6J9>-q%m+eJ&zQy?+WYk0^
zeC+mDjlX``YhL3ta(m+e&G}cx6@_@)U)QHk4d#dop_P$B_bAhTV1HAO-%n|na1PBs
z<-ls`X{DrwEZV<df7@NH;<0jHKIOpT>U6Ays{`6my8Yj^ZZ5z1dZ$tRL%}8K<^21C
z!L`B|LZqBG*rGY$fc-<G=N{6w5|jYMH_JUIQ<6&lMqZO>o5-twtOsZGn@l%H9hhcs
z+5NA)-~Oq6A##$8=kZrsr-KoVucB_qr2TVI8^am?CZ#Ge$=Z4snRabeTwBZiaM)BR
zYGMD<y2(0~xT|PIDYWhE6!4miO71yf|C(JyZQJ7fR!27Iz)$<PCOb?IQFkixCy;7^
zin-m2zVq*GsJHY-oJMpdS-yz;y#2=#GL*ISe8h#xq7SAd5CTi-q)re}!DDq*g=hb>
zQOw=g_8EKHT1s4beEvE6ucEHOKpP{id{zd@GVT>N!hjK1ZgwQ@RRZnEjMJ(i<LLQ#
z=d<=<1;g+{7FO~dU};{tvp0*$d(uAqNPPIJ)vKJJMF|<_h-H{}DR@Ae+XefGBgIp0
zB~)Simf7hU??B^Y6iS;>TG2-yN!obZeQ$7n>-(mMF+un2qmCqXH*G6R>{GMF!HAwP
z!T0Q=>nd%pIzuP}eM!r7F!_xU11LUw&_1TSnhZ$3n`cX^K~!xR38sX^DUBn^%RaX5
z$=xaConR>rMp|FSMn|0Vb645N<wxLfc!E$WTIwo_#h8738xTsL3{4>G%{ZX0kE*ZB
z8nRC~QqrI@o(4|yO1(t7zyZidAy@iXO%Yk&u!7F>WS^LCTRlCr)%l&W;<`1f2lh$z
zB8(kuf8mXBU@?JOSIVRn5ia%7e{y~JKK%J?%aO3BQ~VA%7?J<ocdfHeITCWw)I-j%
zeBsQt=~?^KZloogGG3?)T7ah<jE00Xj8XfvBPHw;JC{}ARp<a{t|E3r6=cB$Zl8Xn
z1p4JSEq5AmL7rTrc+uVV8SM*ok|x~z85Fcdm;%jo{OTEdQC$aDm#YZ|^prPPmadk+
znv$*Eb#iZL^v^pmb$x{#b)VJV$LJ_K1`DVIixEw?ouVZap|vZF1&SU2?7Al<G!fB7
zZ!pcQVzYRXeNGL_H!pls{4WfI4m5R|m2;2S=O*nbeln6@6;|#5Yv3eoz}qiBZlBkd
zX=n*zKQaQS+D^zV-taP?U1XM%7CMW(eSUk9EK|L)P{$l#ya>{Xc_r!Hz90wZ-XWAY
ze#g#~a$D4w7W*d1<O|ynU=PUZi)S5J>OY&eP4^$NFRGixJ{5&5(Jy{G7}4EBy{-1*
zdJaw^GzOgneb9lW6hl`j#-%KyFK(b!+xirSMFep`ddM81TQ4~hy8s!OnN~qNx}719
z$#}x9Z&w^yD*dI&Ptv;_u@csjw7E(%bS?i~XJ67y1}RZ|fQ~x>76ekUWDu36f$C`d
zJ%*wC%0@s@UPPzdXvav)G2)ejk8xM(i1u9y;Mtcp9t~$FIAq3`03I9~C;#T9`I)4O
zjS)mK>a<11>xs^_mmPWj75tuI#hhmS%MT<!c4ibSq_n<OVYYNqnQ7b-qm<)Qv<~Kw
zUDYOXFt&AK$Z5uf89xVsGgEPsEo;1D6?=ozs;`PVAmZ2d^2YWsZb-e5JQRzS-(rM{
zb0Sw5qTTIWThyjKDETznC-(bSwDE>*A!9{AOX_x7>L<P7$uNg$KxCJn;F+O@L3U5t
zE30QL7{*%*=9o8_ELxJ$&{pkLZ4TZsJ2Z;n+W|;j9KGAFZeub<VewI?8Q)4Tv~uQm
zv(I1C^<M|V>MNxKP?OykF%jppcu1LD+x1Kbbwx2k2bR>uy_ki9uC!IiWg;E-x_S?L
zyIgIEMp8jJ8NtECow9s%R#YtGa%d5tXs>P*L|8V*x;pKmW1X~^CG*=_SMBcWYq{hO
zTk(_VT$1?CH4Cb9mBwOW$GDD-8{3BVY0Hx`R!$E+LRLPl7_FTD=(pX_Xy>}?uD#Z2
zCGS{I*{gxw*d_uNfT%O<0Fnk-tjJX`W6=?rO?FdZ<N=WkRE{j@2~0}r9Kbd`G_ad<
zy=>mrpKyLhk6@R~63$AjD7Vzt{-46Cd=^sCh~jXriZN68AhHF<FoHG9=E~x-*R;`K
zFXWiY!2rDnST`@4WpM(=IsWRz%hzVlF1;T;mf!KE&@aMr=Duq7Puc6bHy@Etgl({$
zHy2NI@Ofth+hxK>RIs<#=SV0dKxR0nUGq*}9p*ntNlQYI?aNw!*|T-RRVWvzgl&WD
z@Zm6o^K}d!nUR(7<<*nipyY~|*st_`Flk{&8<*K9&hUIgM`mIV3=GRx0yO*RqL$!X
zRRbe-D&IAQ#3A(KwxUh+rlFbH=}D(85oQ=JiUVk6BP|^!HE^0ybM?JY3)M?6y^s0F
z_hqZP*;l<3Nh=Z;#M31QOZh;qcdn}0-;LN2j5Mxo)r(eFC)=izM-bz;b};G`I}}-(
zwx;MWjh$ira9~L#3O`~PB5enHuc!^XwVsVl8sSmaHEFA=Fb5{rSLo6apkTK(3TMaK
zmxJnyFmXtUPEXnGZSV#-hT-D0>kE}*QK7-1TBqlEZP&}Nphno#(&CW~pY*v{Q%Xnq
z$R7K~>X%C2$YUoB3RqDt6>AxF6;v0`Ic#^-RHC$?Inj#mY$QT}s;%q3Q+7;g0p}f9
za+8s5(1We7ODctdDBx$j!PqPq#$^9@HYua`@DwK<$gT<K8fNW(^E><xIJ@gexeRec
z^oDi3B@k`axFm;dLpOw>1;Es`T~Kz5AyN%y@Hw5hQ<V&I4wvkk>Yk<Mt+Z7_p7T#S
z_!z?6TqR3uyP(XEZ7fV?GUi`R#R&%^5$Y`jKyAvO6aB5u&w5h06=Uk7`jc(0eq}sp
z2ru4XVrPmtMYp#%x9frti3vRGz~ZSeQNA|RIEYKdSnqCM8>uh*<~iEIfhB_?;#YxW
zs`I|YacuV#K1C}|5<=QZ%_c_68(yqO8)4Q#jT5r?^5^U;>Yd8RQYJr9Uk)r4A5wq@
zc5mwyj6m*gYHT^M`i4c;A08@uOZ7+JrExzlGiMq|RW8!O*O)4hQ$ZO-E@B=$Y4>&S
zNhd9&?+vE0lCxT&nD^&(6l|CsW;Y||dDMZ$QyF_WV#b&UP1sv=0T3<#L+m-t{It!e
z@b<R4P25X$H$;BHuN;hs6bbWQd;5}4Y6<k>Q&>2VY@TuOBGQE&#0U%stD$t2eP!L#
z)@S(Gm>h}Ml*saiUut|@;(xWFx<BN>;BcCJ3FeWp<lu|X?LqobM$riyF4zHN!|>_y
z9$-N}Y4Xd7NPe1jwUNTh)}=aGSPq!tBOwRkc69JHT?%2({ZCVS(nebk7`51{P$z*&
zU6@tzN9)xEX+V@xaL|F}u)h30+j68r2g>&yVaU{}Z43jc;lVzs9JaOT)Qm65;!*2i
z;6Sry?Ah|^(|7tuAee2d{VNR#B8jw#sEC6sQpst#Dn>A?$zmeJO*>XF9XA$5u28vw
zrNwB9Q@KXwChUPmH1nq-m;e_KuUE<!7>QW~M)xSh*|Y845JBJN)dW_Xt^!P+<=<I0
z-s3i&Y-i<))m}e)BjYxa(Cuh>{LKC;qz!FB8tP>03NTrBbU>w;;6VqIks5Vi?7;f1
zGhts=;?x-jR<o;0A>npS`nRe46uS4c{yFV73VFiWJL*;qFYdGuoYFlHrW^#_$-K3+
zQu7KS$)>YA3p*afJ53EdRYL1vYaSjU^n&wQdN#-fy8=Vz=)kMRncHM$+f3_zo;e1G
z)MhD-4oohO(DJm%j@xWU>Rqy7kWHN-;^6Zh_8L8hrEL#3ScrVm*^hgJh4V&B=JFHv
zQ1v8*;3yrHi^;)=;S9*{Lz;g$7svq3V8&^rlmpc^PhtA;Qm342blk;E>l_ko(%J@)
z#w3$DSJkJ6*+#B&d!)f|a;1;vv;$khBAE*GqBI3O`T|VAVvBj&!DzXzx@*I|_Ep_g
z(n%$gpJFAEpyFVb40h_NkL71)DzeiOUl`(Syq<k^cexy~S~+_iSdy2Hf<WiDuc?c)
z)hW_VaXCw1u@UWL1$Z07)Cxle3rP?<R)q`j+Ss>Fj@dh#co%Wc1E-y$(eUi}@W39g
zCuJg7m_K3^nzjR8Pr7w{)}E+KTlkey>crn7ypUxDKA$|^NAk9>J(3mem*{H2gLL^J
z^wyM+;Yx@&p?f%Y=v|vRrQ++_K(Njf$is=O4ov8^R1*36LQ8_p@@f0M!L+MYnzVgG
zT_?MMIJ}O*ZZH{2B-N|#;Wy@tPstI??<EexbmQ_K=>%|F)df2}zuUg4=t2dw(wR}N
zS_hN#&EnCftdbN8JeCPj$LD@?^{G@F!W`UzrAcu44Y%00)C1DAkg6pjJsU5Rgy9u@
z5?)zIPt8AR@2Y#2<oYl-Dm5+z@}M`owhBoeIOese!lbJ_!f&nnMcSCM!s01Y7lXzO
z?{)AoC`+<&Y65dBC)<ed$#*9ONZuvA(`kwu6Rok0B511i{+_1&t(}<2pg9sc0T$C9
z<!DNURK2&+tfB*YTby=M);dbjPulz1vKj8P<!;^P0Ak45q9t2~C7mV2mqhhddw+XF
z#c&NAh-hggY2DwJbH8crZFf4q1)iq-#kc3rM<09C`I+1IihJw>O_HKa#HiCM9uRJG
zM7?M1I~q*h6DEOR8wVD{r)`666ZuK0TnY<GEj`wqeXxB75|4^#L!n7CqCCvo;NxwY
zB$=2`;Bo#?J-KWaC=s&tC^10XL&jsg)6P}8lY;&#Z$RY=T)5xa%|+ypWGyLx^+C#v
z7TdjO-_=HMe3mF+WE=<bavnz)6ffM=e=*3Vf?Wz+GSeNf@2=+(EQ5T!TSgZLmfLVh
z?~$9SZJcO_?>X}PoqE0~^#8%xq+-nMVJX&zVK8U`k%q*l^UhN>xf_S3?0f6g``I@#
zNkiolDFEpWlUB|M?4}!vBKW?#M=Vc5c^n54$(3M|fTwi!o0$Fgx93;vOw78x%ML~>
z9jDU4Dcg~R_0~rpyU+QZG=A4K@l&?5o{;tdeHh|~hSI@^8rSt$D7(5o9S(Pr4B6tS
z2fSgLG^j|(++TNh_sCzZt6LB@nB0!YLQ<To?a2{az4kWeXWG(6AHXrtx4re0HbR^T
z%`o#c2UGlSi4_AOMLbz4iu)R#y;#wia$z7&z(qr-aPTFaEf%g1cb4s6LYzqd$m>TP
zSiEiNYQhGHB_I`{1Km9wHY*q#NrNeI5*b=~1a>gzWdrW3fz!}jP{jlP!?-=!%mOOl
z)_<7w<A8FgHobZM`V|}RwWo5ZZo2U%ub+=#dOB3!lJh6KCs9!_@#Eew(kfc85QK~E
z>~Px)R>NRu#Cq%^bs)qe1uHE)_5@(Lf!r}WY)9Leux)k+)Gk_Z9DF>Koh-J2Ggx|$
zbuUuhv25b|9axQR6l`Mm(_)7gvG7q%?RepdbzQW|%O0Q0&cTajFyGEbfYOOJj>FSq
zTelG~J>!5x%o}Hi=j>#D!|EYQi#Uy34x89IdUk5bK&G^-VxkJE0TUmSu>qmX+SAps
z7O|BH|7TQ~Rzu?8gT%t@MrYFmyk~xoozDJTcMGws{4QQdKt%AOy>_O-!d<13$7dW^
zD%Zo~{SiCc4T^$;M1%@T)Jfi2YR4RW38b<V$fTwmZ6#ejQ+LYP+mfA>_#n%MggWqX
zCw30`c~RDsr9sUXJ6F9@(L+2vEW=X{EcZmFlC&-?4vIrn`a{}s&bP@Re2n9IQ92xu
z800ppO<rireKWa)I09)W$%N}ABnNx88;kU1pChOXST5cOHxxIojn1BHy+<sP5RbZ^
z49OcNWeW*eVq&V7ezDyXTV-2B6oPXwqIb@xnI5wrsE%mgQ?UagFC19X5806fKL23O
z(WOH?>9o^|vEhlBL36ugJ4I<o7CFE87044=BMUPGdlr~{?C|UaMxsk?>Sy|66HY4z
zDIc3&j)iWJ=ohbg67mC<%wuSjieaewm-DGqqQ~7a>NKN!^4T>1Pxv0_ih|RAsPI9|
ze6RdJkXX2VwL^eU!e#ii{!bw%JYSzgU2r1)6}&65is0V%G`OLZ!l%RJ*q!rdKiqaV
zG-4_4&pD8o<n*wm@|R|?A1QC21okxYMRC*M{VFFZc|8IPNzCP1^-@8@<FoiBr1t)3
zJz3W&#OFCJJiH61y<svgG;_Pq{NY?bmQ?$aWv_I8iYi7_3EF<Vjq<HkClf#oOV-y=
z1*M;;kC$S&xEWoLd0iO|4nFRQt&71qK_JjiHW;m_*+SGduyku^&mitRzUHUuLRqFn
zS_CRMu(pVU^pcxmfwP}(7^Ds5BPhZ!Eba~_>4lsF!e=M#XL5C3dN*rY$Hmc3?!OTU
z^=GrOY}qweUF9^yUvKXd;q-GkQ0s;U&d<DI?aI~m^X+?NjpfB>=Jz?U=w@YyMQmuU
zjbA8?bO;=0Y+q#@bTB!-kXUTySRa0|Jv%}`#4n$8VEH+Cq7)BhztkonOyH7^CAW44
zY;XARAo0p0&~C1=h;+7Jt}ZDYCW>FH;D?N+4ko%J)+@ilS^PY|(mc{jye#0^!GeZ)
z{-P|ywKe^Rl>*H(&)ToH&t#tD@888I0sBA9TiB26*K(W*f7~|XG;*Kd%+7)RdV5S5
z5`&{Uuo(3mTe)5o<8xtQul+{j))L@Hu%nDSpy*rFRbrJKyZpHQW;c>YB0NMf71B^u
z_WfBzVDRw*))!vu9M7mIWxv&ZDcU(#OKT{YR_JIC%2PP$Z`XCiu7w#xRTY2*zuJp%
z--?xM6KWYEzmr6`dGm%%&hIjlr9DqkMd?>=c4+!$zngbVZy6hPe#g_36f;aCRSRth
z7wq?vtZ%yhM(3x<`VmH`XTRUk_=9s!Lu1c^I)lsW5861(q{}eG0Xd<FgQ-suKS*s_
z_^9QHaQ267n2@^U7$g_gfoa=(r1yyZQK418h~qclsD~X`O{(Z6t&QsWA^YQcS=)e)
zGFy4ha>fCG*Mo0)^^Nu?ZQmIo`e5KRYll$UxNH=aK2jUlpSEs>4k+R}>cFBOIWu7g
z=W4}N|5^KJK1xMex+A8b9W)>B@JpEF(vskRUT@>85KJOK*BdOY`=Q*~Uv%#Y4T_Ck
zaTiA*@?ftHKI49HaZu}ND?8&~R;Q&5%Yf6JbmCx=i%iavq7G&LRl6H`MFuzm&;V)H
z*A0=6vCaOv@n{*;ODJ`KA{q3tXD94$s`2>z(_5WJYCH%)DoOvh)q923Xm8Wsi1f?9
z4qmKwXyk!0`@3$=qmC>~T;d;d?qC{|1g227XYB9W%%F`{F>q#q@HYoj<QNVt)Nyek
z1@%P2uKhz&$gS&EJ3rOk&3yVQ4okLwY!3~wU232_XcFteG<CH<wXHWYXa7{!hDVQ2
z!<qx}Vk2@ex>o5_<5;ZOMgBQIGwSQ`JGT)Ex!J#TFRU1!u<>D>b}+?7wdT`Q^sj9k
z%1WlgSC_zo8*}=){acNIx-OkAZ5n5A>ZwxRK?k2)koh3!oE3M@{=I{phL=T*mM_96
z7m{;f|IvN+#n27UN+gv<z`@kGM-W18XkqM^)#a%DXZMDqq3vj>3bwFg3t^iI<AY}Y
ztKI-3QI5cZGIdi90HdK>bL(9j*O(ox86kd6H5&#FL~`b=VzqhBhaJuSXoa<Oh1Be$
zfWL!}QNTGIMS_qt?86Htv06)q@$n8uW}=E>sff=<<aG-idpwS_FitiA7QAAq^+Epn
zllGC>v4tILMjxW&56vWJn1&1t#An+_wM&R5Al`zeE?L-tQ+*r<h7MEz=%c~eR&HG5
z{ESzn0?%QNun=ub&)CNlCMP{B7HAP2b}(9MoFY50k8N*TNleqI(~J(4^A*|gynWo!
zBGY@uwtLrDe#<TP@ketvTADkSM<U2BA$lV=#<{Avt?Gq_+R1o!f{%SdhtIQq<Dik?
zZV|-d;Kharhn9@{i7)tINhXK21!1!RlSWqdR&6diz&1jB=|_}$pic+wlj?0{5(u}A
z%)=7Ho<(#6A0r^OR$;}QQedClMqt(YyPZZ0uN-i2MSts@eM<KQvYm@7g;$g<PT3hx
zI(Y3wU-FV`>{E{hpBePWXM@Mm7V04+?oX>Xi%K)2<G2G$g=sy(H+}o`9Eyb<8z=$j
zw32!lEG=FPp7t5ttfwDMVQ)&WJM9g_(8F4?im=4wWA>u%le3K0Q^ZgJ%l%>SP>*oY
zY5U9<jC+orG?*tHjQS|gE4Cf`tgg#c4eL40e9&D;K=`xlv)ebkrH3t)mTq3Lg|JPm
z$)A(M@#gg#oS&l7g&iyJw9n0}IP50;@ifJ(<;UMHL)+)o_{*26^?iiBLDia+M}Yq0
z;A5cnQzyKllI-*I84EimhDM0GcG_Jg|2gSt{JUUZP*0XoLDgL_nLtwtqsBlwf5gFG
zNlzF~9jzVpuXraXrt=rs4$6F1+6XfFe*40srQ{$CZllW4BeDzt%;Dj~Ci+mp7v%`N
z^@>+IKQ#hWj~LjC^EV%N-ua!9@YL*<E%wE2yjQMUv1zk+&D&ONSbe*_B=5WWiYu;g
ze#bC~3+RM(#7(=RU@0vtuH`ugRtt7<L23y});m8>-Q2oIVUoDVkt%jOuw?n^Qj1gD
zWQ$OM1g8|(m5oQUWlqW`?m(h9O2d)@PGpCDX+1?f%uJUsu|{MJu4@OMUw-?V6|3!~
z`Gg3eNX3z$-W*R#?!qm`Ah(wtjS$oobIvbO&5B!ZUAuApn$>nyw^TAGO|=50d7t%$
zNmG&*iR1>n4z{ejN4{_Q9a<_4n4~d0Vu~z)bJ)xCjdDpd<Ks>{u91md-cV(?yrSKQ
zFf5Uom<$J&w5UP|cp)E&<lk4eA;u+@kzdj@x;$J5pLZdk4NXmPCaBiNtB!_l#s0w=
zJ2TV$$vS-pcnv?NXTm&?g6-<|b(CU~EUlvwv+N9*BvYQkzBI6Fjz0ex3?|Em5*IzG
zRczT=&>fUm+CBYoyS5%E<wU}2l<2wqI2I@{@%mDtproC4UAsC2=`cXYVm|0#;%ZHW
zwiONmwO7~GV*WC$c7^*LjHqM`cO>Sm>)ZN>2n(kv<Y4>QG)ECEZ=_zqMv=81dlS+%
zTP65p_T#e;W6D#6$_;f(<VVE9S6u#c(lv229d__3AKt;~&*MD~-MF!q^kesX{hXW{
zv6~t#Z-G*sR_x>X)r!c<(hIxn=JxGj4P;A13<efgYn|u>K?SRX8fdpPB4Jb{EM|fy
z(GB$UJZcR1?48P6VJV)l*W~6#3G$&4rxD9I;)O)>Y!y77K@Z!Ly|!^pinaAerk!@N
zimYqUA}eLuKm@keHJ-OVi-|>k2Ege30azrt7N}pGI)Ddsr@g)~NiV!O)Ibi_axh5&
zN$_m3&)Ju?9pb8&I=>4qV|vu`wl8lm1OjpsSlTCj5f)537WEt2b&71{LnTZ!m^Kqt
z$vaBm_=>J~-I<S!I+X2Ta_Oy^9HHcw>LQUFvoa}f(3`+iL*#6(p-pIKN+P6sZ==kW
zQIvpY(%LM%YmCgOfvs-KZd;!_F!%$|+PZ1^a$A!>Z{`T(`l(u8x8`i#L-FT<S3?YP
z*lue=%;tOUavHV#<fxQo=78PaF4(YcozsX5B&f->Ln6Pet?%D=S@lmvb#f!2@H&{h
z8VkB?>nfr6#*UN};?JmJ1y~ZO;^1TJ)GndzQmR6Q%fjHaY4RN4Bz)K%_2#V5z-+L!
z>b}zsCh2)>QU|v(s_WXAAwJJ?BA^2Z8WpS2GS~}m%S?Z?bDa0<tMiKEhLOOYPCBsE
zUdd9GjqT2oEtfd;>42h}Ghr;Fo4G26VLx)Sj-2PO-IdEc_KIUpOM1(Q4nVLC^;Kz_
z(YI&#*dYg#V&Cfi=+>USsZqw-%2uCqK#~opXEb9Q+r%xu`4*=UebDg8ML_id+tk5S
zuO@Yw29r@1Gc0n-3EN!PDUDiQHL7WNgHda+rtzG;xyk*VW1%sj4C&GY9dNFSo@lH!
zo#@?mcfk~RuHXpeD*?+XjB59?lHHS|{{VJ}5Rz%9qP$_~LHmmOa>mg4T_2lUBHE4J
z+tJE1iuqtvE8W7uBy&U7D0eB7xL73TE$#E+u?k92P8_h5IM&b2Y_t11cBk$CKtz=!
zO~(px@L$ZV{NGesd6nH?eQf%g*k*7$2NQkF#NrS-idb1^tg5$mcdD#}CRxsO%9A-5
zjfEmyM@ERQ(B}WPMHmI3M;P0I1Jf6&&Oc^{x7RlZQSdy1MZLik;&USVWnbA=l`={w
zvFZS7=r6`%Rp7x;n@MtALx*=D!RkdzGLCDT)QUN7!-d(W>@8qB^*O3S<zSK?Mz)Pp
z372?)kw*V3>iocg1IiLoeU2YNHI7V?Q5$XDTuB8Yl{*KJ^T)YCTdkM0LrI=V=XdS6
zGOMJTFIn}ZZK+R-c%e`vc6x)!X{79NC6sJ!*F}0uC>-8k`^8$=b6~But**m#!}5#8
z<-n2_6+i_gslciHnEHJRQEaSzNPo+gfr>E$NEgZ0sBs&jmOYSsVl`XCbf1E`^`h9q
zSo^XCZ*Oa7<?I&k8ml*Ku<^QDEM9?5IuaY0_8=Lour-9qdLkb{N+Ksc)2KDAKqm?9
zdu_5{nctWy1h5iN*i1!23(DrkN@e|y?dJ(}&4etCS0Vh8JY;Xzrq8vKI)`RwP#Vx4
zZK{#R7D6E>`vkZGNly8uSrXOl9gS~Otp*FI(-xcDd>1*o)74L@GRn@D+5#-;lO0fz
zz7Y$H7<bwRxxG%$&Nx4d)D^E2a)ixv)FZD??nuaPc&NeT3JxhY8DcuzVp?pr?riSI
zF6rID<a?M9Ibu*b?ZF0PdgRu5>hcr52vd^0cFtY)P`x#WT5r;BDy1~tl;Gppnc~M*
zIKR&x?w+j^du1Yg(t$<qb2k!yQBTrj%JJbg*M0Y*e_qS_n1d0MUiR44S6*q4w8yX$
z$Zx6n4;7EUSZo}81;k~h{E|GW#dvjlw3{7$C>><m&(B%JP<K$nBCo4<Bm1g)vqUyo
zf#gD8s;7a^o&8RkCkFP|3woOIWRdk1E8@I^7lCcS3rZ@0eRYkQK1p&vxuFg$x2Mfy
z?tM+t4H>4+PZ3{#D?u>cC|4@ow8|I&5q5Sf8p1o<OklK9HVv8P4lHIuZ<2J6hNQs@
z)2?Kq#|y746}d+xWn~ONxOW_Uve6kVdw6uF?TLDeD!6JONi1++<xrB_G&XMDrSfJ=
zVj=Boa}Y>r$4EGEn)26ie8@Dra8E|CFztL@BaZnKGBB`VNletD0COdG@aV-=FI%wA
znpK__vQ$Vb%Ch|Wb}g99m}4GgH|DPd@65N$LE_r{$Vprf=kR*iH!MyNwm1#gMF*A>
zw0^}p`^H98P^~0eITw<s8YWLKZ}Fc5Alf&z2Q0tgW~ZTdj`o<j9w*WD?3?S68QLow
z0&-@UQaTK@tL1wbX0Hi7X5UiNr&t1gUskH<j)8FSv0&4SuZ?++y{nt7(@?#vQHpQ)
zZ$(F)s~Wq4FoI<4TbB?E7BI>C^hT_U-prqJ@VOqt9?HHE#FnczldO<+?cMDutSa?6
z>hrMnREj|*zrn|wIlK?^jW)jv_MXN05c6c?Jnz6#(~(|8>3d2-zPEvHogV9rDu>$v
zN!MA)oV_pC%0^kt>i66TVSev#FD&PRG=c_WNFkXZKl`?}OHi|b@H3jleNKcWMbUjr
z?bc!Ov2X9LS}gVpwW2*p3y|1_h1t2P*u5AzoQRpxj;gdMn5}&v`OT%T1N#Kbo3;d!
zOFEfyuYE@&Mm|7UIu}`(5f|cM$}p%HT?qSwb!o&O@$xt(lUM9^FybVuaBJI#y6(ss
znTBsoafT`$aPTojBEqQCi$C`}+mq$Iq4TIkkS{g4988iDvarNAX@TF>V2rPVKMBfn
zV71-LF`hPzuo&8R=jY0DI~Z|Vaov)mr??vV%f6>Dl0!9XUBP|_Q<FuCHW-Dhr`bzp
zG@@)(B*4D6eK-50_Oi4bC}=3;a|a(^qE($$;?CRm)sr<o%`KFDUpiu3RYaXfvrDM@
z{w7tf7#SONTJcqs0mq)9<07BUGAOpAv8QcA!(%hEm;iwUBh>hFI4xqYwlAecM05<J
zue4Ed`vbN!=k?laufN&(oh)p{^pNdpdl;2>v?OS!`2mGV3E#w@q6qJ<j)z|(*!=--
zFlF+@5iRpi+MY&fL=ufnzGKz_)Xt;bCU_&nd?G}51bN%r&9+!~csl8^>>yxbuJWFA
zUds?J)*WRJkZm8geO<?tFr_TLV-75r5sZExvm8q@HvCA;u>JK;Z3Hk<)CoCv2yt~V
zNu$uC;ynmeu>);@qNMIQr=3zF#r9ZA2ist7x@(=&sQYzPM3`pn$#y*%I207=nrz;f
zl%J};#16VN0m>UD9P;!<A-q|{=ujJ(3Ta+Em{}G*$!cU*I#*?1V8F78pCbn7aJ?-l
zBlX!3rM+RKwbU6A+ym3LBlR4`mTLy_1vxNL(D<0L<Blc)Z&6w5fmdr`2O`cEJC?Zh
zs-rRPw37Toi{B4tJ8j3?_>FOMHD<@j)VC82a}NdRXymBR5~nn2C+kTvsqKUyR4N8o
zN@EjqI!?w;b)-kOJEW2MlKa^?^btoKd@+zbMkR%3$$d|^CAn%FDG#*bI?k0;&Iz$&
zlJ4bRJ>A|&Iu+#>oVNI}%yuXkXX?Hbfi%Gq4vt|6XqdKZO0S6^H#^$~8zrt%p&eNE
zlLanY=#g;C!iMdcw$b4)qs*L2Lj#J7)4G*0ttrf&Yma7=Rot%Z%fRAl`AwuW>69V{
zhwb@%VL}gqqEwVdob-@`X<R3Ef<=BOfh=qt@S%L?g`<^t6zPmhimyS*U8%fSCkQ#L
z!+p*jInP_iXR+Qq+h{=L#AyWi<d&El>vILuDyQ5N+vqU|me1Iz*nYcMbdX9?u=rv$
z(Mbm%kJU$NKgN1sKT!8bh;7+!l76%uC~h5m-J?Yz@%F6!U=10D76u^7I%x^9PlHLx
z5{{~ZKRugo&qwa53@P5+6!()3J|;$Tk+zdlAzeWycB#=Er}oVvf6xwQ7xAcEscOoK
z991P$DOn52_-Fd#qm(6#WlPu^i5HjKE5Zscx4Se0oQn=7?Hu=Ab^m?#Lv>wc>`S$e
z*Bo>(5~^0MT4m2SO2l@evV{F<2b4yJq0v5~87BbZasT0FYFVXvLemch(GDDNe1tw!
z!q8FeTW~y)`x)|Hek42oK!29HY&2?Yq*_R2RrC(0WZI8*#E^%EJcs`cSjs*tRSp-1
z{bTucJo(uCL8ny{d6c&y4`E~Z@kVr;x6L_?xPak-IzsujpJ>mN?G+n}1Bi>1T*km7
z(w?xNEH0gv7%l%aUTFV{bI2m(m#gAViP17zsYK3C)o94boMNt|lsq<XH%3nI;=br&
zaLJ!;#J+iUnnJ!bV<a<hKyS*(cR87fNVlKqW@UJiWBe#R!!bSpYeeN*Taw6cKijAa
zrXlly0>b4UtMF_GU)@jLn-Br}x$GXtBr;hvBu_r}*sJchpYQsn6g-itEL2(YR%x8z
z<6&{q7Z~<(`-QFp%Ajl3?XH}VgAWlH(;u73FShOwuC0Ww-Sej%ST2wV+~}mS;9c@S
z|5EqX6EfK&6P|To#p5Kk$RvRMt$5%CNg!}S6__Mo&VIR`F9~HAu*eP!=?#X%UJ1Sy
zKS$swq{n`xz47wnD6K(K4op1jnh^F5+pjhfoFcA)MvO}vE5gg|*V-DLBTw0B<pi?V
zOA6vxW&NGsZNJ`)h}2z~#r8O`7!fD}nKyQMrZ$2|krGbZZxmh_xsfIL?z@}?IdukG
z!%Nl_b!p|~^3VxVPup*Hy;0ev)EijSRQdpuUqTR;g5GuZ({JUcqWz8!ISrA^s&R5>
ze!Ct6Ik2tB<*29#ofQct9wV(tF%)n~`<=R46RhGF5zQ)#jf2VUfU}7acz?GjW(PC9
z&uPU>=iU{v5U*OSbP>BuzSe%PiN;hLROuM}-w4b=Yw<+Z0?yyF%&GFdf4_}0U6(YO
zl&+&V=_K0%c=68_Z(e18ke!Odv`3v*E`%I(iM#fPjUsMXeuLA<Udyi#%Thr?DZpgO
z%M3G6h5tXw&zIp-RV``8<f9hgkJXA6+mTG6P9L#9&KX$t*emaMekbol<RbfO&i<qg
zrRup8`&#|fm@5j9Gj{DyTi+Lyrz>INkOPY$(he-|FL4q4Wog-;)zvB9XTgvL)AAC_
z<ZLec^VSp2;Ug$--~c4-UHTB-mk;51`-}Ej44IA>O{0YPguKBg9DK_P!0$D!1Lp?z
zmpSVF0mpO=oK}pw5+lz<V&v2ISH;A&%SioG&BffObHCMs+T~o83e^;2^w*6Zf&`^w
zs^XhcPr<|sWJn}jU7NK1O?P$b*b3YT2J2vAqS~r&R#<dA1sK@hwk?-T-8gn^;DF)<
z>2upRkHOgfu1(o+A2*HDOd**-3u6B5@9P$=-}J1Wb6}x)A?l?5%p%iwAgF}5HUg*Z
zAF6-i64E|*VW`><M^ldHLHKUN!#DB#q(-{LrfC1T_~v-a1tJQ7sdsCr`4;|ye`;d@
zzfkU$Lb8@T!@*Zi70#(X%IL3T-lPC&Gbiky+c)3QA5hW5X~!_-l}hkGWdG9jM#8w3
z#c2nY)4GaevVZL+N}NSQgAqgP(;FtG`C$6ssp4z@R`+O#Byv;GflhL0k%Q6Fe>qi1
z{=MET`Bx>RbutsM+_M;O>Ga~TEK(M%ve5S1f3&$LHb<dyupvP4QIJbGzoXi1y`L>c
zj>|A&rOf-Ed3BtbqN3&Ul>S#OsrO)>Q7Q}VqNWS<UMu-ovtwl);EW9vyiz)VYT8~&
zHDdkGKI~X<ab%00^E(IeimR`%4?kASC;S=VCSo~>F9!2*@WCjw^reZ&4jozd_7TTY
zwd%8aANFeDpTq!eA9*Z0%1#rXi5B7wmcB?$zy<rL>?Wz#I<a8VX(ub>Z3?fGfG6J1
z{Ihm1+DF%O<>!#aN>&MAb&sMUwGvgTzf1VCk2zN6$eE#s`;(m2G~$3_9A#$HG|Jug
zvByFvSUon#LyV^#NY3A)mT4bX!$I71Sb;fJCb-YD4knlJhBbF>ShvPLzFk>S*($G;
zaRgwJhg0#NFwsf-gkurJ_7&@$Un%Er<`i4|#C+rwkzL{2r5zp8*vZ_i!Zx2&uhB=y
z&lzE{(g+-kHf{pC#_f}9jAZ{Py)I_CWha5J?V!mYdOpTZ`;_dg3{OKiYeyYG4mU4J
z)ifQ|I)RM5IQ!INv5wYmT<!c6p>3Ll>g>~w#hy<HO<(>Inl)=GoEKY<=8l;TQ-FPX
zP6vT))6<jvHPh1$L{A&<%|6nz&u9%oj+{pA_Y#&<!j520d(q;E@!dha4y>+DW#t66
znDB?$XXc14>{$JP^E;OsTaOIjD$n~_iNEgJS6}P=P7&sfNY$hE*=<eTc>T3bqixb8
zy`x)xPEM;TE%qjy){+sdN*jv>R_$}U$$Bb9M-;I1qibRA_J$WrR18Q4;ACkn4xzKp
z%h${L6G4(*00M<@+-sNp{Or$dxT;75p(R3z!wbLQ1%o4>97{^E?G7e|=GBD32Y33y
z?mHIK1Jg$$3;J#cA490bHhFr^%%8O{YWJ`}i5f}cAw(PXwCDFbcyZy0!KA&ox{};T
zG%qKi>}X)}!EgaF(PNO^7avQz2G?DFg8ACR!p2UAKBGV#DWzE`>nUCp0qxEmu~X^!
zt%MhJSbL3>F2aHBCC5ri+1MLU_71TNDGOlCU><}K_6HscQ{u%E@UAH8#MrUNX+&?~
zt$McH;Pxfm7chzWCI^;l5WQ4R)y`RDR~Fsfe8aWQ@3`t9GL$%^WPlLg>`UvJi~zE&
zxc5;9R?mzbIP40<J)lv%H0gJg<Wd=_XhtWcqUcNz7LvWJ1}Ful2#R1Z^&l-*>@E_Q
zoU24VYj0a~r(M-Pi`f-lJwx>tV4BfQtM9O7g;xD#g$H47^aj(epp?#C5XsB)mfp50
z=VvaZ;b}PcXwP0z&)CJ>K8b3k0*}DNddZiZ>a%~^E3-?mnMvn&eVBq%WVt<Ruc{9V
zI*5^O1xLbc8FO^TxhnYtl&7*h;+!e^UR}@B7_0mwMU5R;vi~96l$aW?IrjY3^yz;n
zfB3(EoV$uEs-l02FY0soFzwog4~MFR9VM^qSN|yM`%VX6w02NZD;}RCc3t*tdk=4}
z>c!BG?`kVW!c*{Hdv!g%<TV_8Djmkk;jdC2!N=3J7BM%VH0-nM8y}qLJ%EXcW_AD~
zlT{ZJi!HCmb_+Q)6V<(o*E;x|B9^G)*RYq_4edkY6D8!z`?hFg0H60nx`Vp%wB6Wf
zB|Pv%2LZ_4^tzjGxxsE~x4_Y{b-+80I<Odv5DxW4@Q&aCc5~|hR4R4s0^CQoBzfKB
z@&=#ma4Vs;VhVOkcSm>{BJeKgC1msq-tZZIkUo%{!C>L`n(7N@C+CdFPQ>T$aWFMN
z7a8}+R<qaE)j3zHZ6HG^CExso7=}(fkmo#SuWNIK5go;YrdIC4j*So5>+2QjmRO%E
zPB)k~M_Esm+x2D1+2$fTheoXsWD&+F{)QLv%e%3VzbE}*OZ0Z{@ZpeF-YIq`<|UX0
zTi6@A`-5vsZGzn_rtyHUc{_%`uSIg5K)G5`cZx=jGMBMygpL&R>fm$UwqR?cBzn(Q
zHVE+##Nm?E(m__fdDGfE*KFEktJ*c>1oa758g(GGiT}6LxEVa+3vG3yJ<h4=Ijy$l
z!d9gu(kz^_<^>p0SVq_$a4<3JRpTnkpWK?mPYm+RHm4C6h-qmPkpJPfP7b;8TBlJL
zkXfE%VYfGFcxp%vYt>%{6rZ9eD7`E!2QrjFP>1IC*xE)mn<y{8-D&2kT0b<`w>NgZ
zJXr?uV_q<H0g0|MJrC@T_Fn2sLww3c19T<y*GVvuNX$ycb1A1U)0$`X$yvItUUPt*
zN5(#mhHzjiKmMi_wm!Q=-#I1FX{Bg3I!m&v-Pt%Z6`DAZ!)eE;#16^Jl@54U%|fR1
zg=DRV5#|aet`;kkH-plUwxR1xmKcYFIxq!2MBU0Qd^IxRo3dB^>8Wk#9JFGTVKQuV
z;$?E`9gW@e0%p~X)rHJ2%4M6j3@}A%&#^Y?fHVnlJf3n9lHb($-ktdL`uOYsNK)p!
zx#6B|PSRytnjtNT!o3be>mDigT6=R|D(!2=Y2<p=_Rk=n!~U%mbja>5Oc`3(byY7#
zaj9UUa~WDCekbQl7qnvcba)+wP?k&7ZiYad$-$>{d!j!)M%DEcEx#h?Z0NzE$syG+
ztM<Z<5iIOeb9Qh0<N+eq$??7e&^K)w8lSPZ<Ts6}9;DL{0@wG9-B<IbUZOpQ{c{nf
zP<e?u=j{Hrh&Hf^&<Fz1K*+GS3oiK9f@QTU1sy!)z-pFcmkOf*S7)x6(qG=z4Q^Bl
zDpiLUdc(-<LEI0Ki}Ic*oAK>+kNo+nqkdGz-^e-#BO32l*4<aug*glIT&TF3(lfz`
z7PgFylV&#5h7AJk6EW-n+Q+f;A@MGpvf-|y(vSAh9pr9B;5eALI^#1GnYPnL^7&e6
z(@sOZioz8vfYGj3|IJn5VAQLH9XBk$*?Jv`NEpWK%R;Lu3K9dKuWgx87!J17JyI51
zeAN<Xrh_k1R^C?{YC`_7t=*lHWLw<?`JgrayBvI)CD$Mu2Q{KUZ)+^&64~$ToMvr5
zSz}eKAxuIxmV|pJDIZZwo_0`$s7J1d&~y;X+R@|>q^P@U#cHpg>sD;G?Tuw|>IwPv
zPP<sH$auWo4(+flE>-!#=sIBg=XcvgH*Oq!1;?P$g#(M8E51&rawumvjhB<%h-o(w
z3#?Lz4kpGwW7WhK&se|Sbb#49=d|JqF-aU_n9UB`RC_yLn}&&f-obdftOl?Ndq>h_
zobbZQQxgs#KlaigjO^1jfFbp8OLsyCzgZLx-u20kUw-`m&wn=P23F~*`ZQ&-p?^A<
z7#t0$!p3y&&P*dR>?#V~pH<;NfOTODSwCU3i?hPMr(hgq$}80gOb*iqw0wK85$NsO
zUY%x=Ka&;JvSg)=sH2BEOy@7FoeoSR)sPnCXUhXf6qB~w9_|)Sra&rUStba1=&;*(
z!zWove~oe{%Ege0!shD!edu^{IYe}T!aiyz051*d(tGgNzWeeM_DDAz=cQOnQgmRU
zvlj;n>vE{vjCJRYv_~81vA3(}wQL5!a=32qP2FZ+mE*r+oOmI8<7mKedVnF1G{d+Y
zIWKVG;3S|XDjmOIkLBZ5QF))Ly|kp`Zl|c-S9jy9k<VVKV9B!%CZ@N14WbSEn&g8h
zwgivS$W;`<SYehhug7V)<ZC1IhQIQ-fvKy{DSBEq7JFwK9;RBE&{`~ckAzZwYh^lD
z#hr4)$cV~jVUHI~)+nwE10paD+2-3fDjoNUnm$Nd2D8YhToSb84Icp?47y_N2``V}
zr?af{ukEJ)DM&*6P)eB2Ne>shgD=hmv1u#ft3m{`uWO5CFh^lj<+U(a0MjtAYP=`?
z%f7zu4=W^ZJ|Dahr;~#T!fpZ|1S-N-z#Aj*jD15ll21v^kc-8E#Xb3)#=T|73Z>k>
zu^X?GNMq4k2Uf0+5O1U=<`-~}_^dRjllDz1kgTG#h4Wi<vv!mq;{;O;dAsvFcgM|}
zR@k@Hn{ruck2>THrn!(g3pb;^tFBWNb!Mg?bzl;H7Ixg*w{LBG3h6Xs(><qM%Uxl9
zd?Cv{%$$6UqEz19<^!ICNXL9>SJ2pisYSS0!!p%SM2-kE7rS^*^+!BQ`6mi)0#^Gb
z0)>PM1qel4|9ji=<P5$Ur<D@?s=MyIbB(<(k(YMNvw=FT=(vK))n&Mgx+gI}zwP~n
zm(MP=p3BndGamMt`9luAI<Y_e@5hL;$*hlW%W)@pA8tS+E=xwH6``na#h872o1T>%
zl|w^AjL0))A83f>%U|s@vYXAi3B5P<f>E^Z$nja&5pm{9$C>kjAt#(d`^t#64|Y_Q
zm7B8+StnrU_&Hap8>_}St>Qy%y=<ib@Z{)2D)8U{<8$^dU?SFZu*`Glzq2N4Kjjld
zuqwm@EQO#4sboP=Wz<>uuJ%Q2WYS<{Xw{CdQ`M=w=io)QGn`R0Y2V!@bX-R)&?;h+
z$Z~);iM0~hlZ0pAQ+-IbBF44Tfklt^BS2`U$t-6iN&DXJY8@UM;pn~uCVIe?{!aV8
zx-OX?3wL3!1JjawsJG3&zk5tXGr)CEdV}HgL)K=?#xt~|T?hA+a-(#Z28)mwU?=AH
z+s?X9I)cu$C1BWrX;Z@npj3iU+tpU#7S!+g{Z2c1G43}-9I{$$cQZDmzkux=NYRoU
zjw$5J`zM+87;oNF_z17j$4d#r*vl-**z4f+7Tjz)fe_;lrD68AHMXfo@PyM8v!75g
zwmI9^K3UP*IY-`LNyK@E-~RTH!M2Teu5g;c8IoU$rzn_a@hIISxC7k+l#H2@ev%!S
z>?doYKQjk2+-e8At|*sG9O@Yd7US_kca~~pMcod2vhdQi7gIsIvs@i~@fV7cZnvlE
z6=GG2lZ2!lSSB$jkur}rIFxgD>3xVJR1b05X-XR4;6mD;v%?){>>R_npy+oOOyFD<
za)!QH0e%Y6l;uLqOkw|ucU15QFgubxg%57tNHjA|nLK75u%o%)`?FY=_#H$Pg9u5e
zQ0}qDi>7CLP9rW57f!3g89Uzbx3jTex;QcilUKvJ<Z?mBu@g0-l1dbQ6;_xC`qnI;
zbMSGGVuIvsIgBECvS3;XnUM6f1Jm*w+BRmVmQ0d{LFqzi@s7Ag2Op0_NQm{XngCC?
z*Wg8wQ%q+|0t;eb-eUXCmnqFKB5v(;-6O$7si^$*f-UTr>~Szj99?W@%AB;x`IY<c
z@YsZ%?HFe|7mrH34+#uRij<qj2x5Gu-L!hu3a3$fjN-8*48<0U#CNV6_7KSUYQ<~8
z)E$h52HT#UZ(WXjlL)Wy1}o`Qc5%DVC~&m@5H;jSdk!QABke<QEJ>_Kn8(PjvS%AU
zLo&i4B!N`s0=nwp#jl1&M+s&f4D7k~Ay^%7pM)VbR<VOg>c5*)EaYdji*;>QAyo1l
zA&z>FgNdmWCB&%~#4Y?l!EoWAy2W1ymV6F6Z7U-~;^V9B2iy8eXI_|vGykEaQf(5>
zRa)Ic!xW;S0>*rf>#CKjoS#}g>u$G837i-sT(>%s!fb?E+Cst7W~~Le(;G}C2bE<+
zX0XeRopaFq5W`Qq`%ndoQ&!A=sFPGlyQM&>14z1|7TTmeU!%4p>qtCt;jDv+(a;KO
zcG)Glj`qXd%XF4;*i0Q*@D{fD7(s;whx0~!i2YhiMhl0zwjXJoCwXOJXv%5ljAjLD
z_}P!v)6};Jm4!!D+33(MG&c&PqX?&SZIQghF4-sLDiJICu{Ho0^T;I}IDldn$HEjU
zBlC~v@)_2d22MlL%=R&u^H0>pIXz<0`yE)kORcDOT6}7s{bY_U@4qmAz-h&cBF0}*
z$~>XG`Rrwnz5agZci@q&c}}<??5Eo(tnZJGZ6RCDfwXy$@xF|s@^9rh{!H~oLX||T
z<GK4n_lp?SEYnUW$y_nxKU?=8g0e^I7zrhab4Zngsac@>Q^;LEm$Z7zb<3TfxpDP|
zwsM%z&o@e1k8t6%;%g-QD9(?SXTQ*3taQ0(I8EMxrK@t3jSZRP7wZ~~v!O1lav-o$
zuM{6o>h&*Wx5B<1l{g$oK4wA!=A`{{H%t|^BQlWPkaeJf5#XbHvNC3L+FARRHgsZV
z44ou;30)T*OtOMtPRXKV*Ua?BllH4QC~M!c#`!6B-CH-drDwmE(|hS%xTg8NWa7de
z$~zX1Df#t!ds`G;L1_n*pFb@Z{>+U1Mt5mc0>jwAZq2L_dYdo2?v`t>x8H25FH(at
zA|R}aS_!cs#V)U+C(gAYE47f{#grjYh)f5VE$p}23=gO%=d^OjSM`~>Y5VOQ(uEx>
z$g{*&$W%G76q^yRDLN>tA3arS-|uvHZ(O@+!$$Ag>-*Cxocy~DKc%$I38$?MY1w^m
zu-|K&+lrN&oJOtm&_!fMhq9i-jq>}A+8AUR({-jLFmXFp-+6YdKWMZTep_Up%a415
zL9sFhK`hVOAJ#i!wTO4c4#vuIFnZqD778oP*&nq&V-ryASK3Kr<S2_X?ImpIkMoY9
z;W^4pIIR@ysH?3&7Q%D>Bp>pU`p;>%59GC3pjz>wd+bjayD$45cCTFyEUr%FS*%2r
z^ihQE&)VdT_6Yl-Rm?_~uP_<dpBJpL4&v1vSay&?a{Y&b1N=o1GJGufdDU$S{Y(36
z!;8wO=_%8a#tbuCIvM|R$wH77`V2IYnoEiv;G=7j?!=7{lI*YAXF)r%u%E%sAzRUY
z2NTOm5-MZNb}0MPqOb0+^Ru_k5Ys#1w35z4;>ej)-%=owTq6J8-{j7`YRvgn!xXOD
z@NoTYK9NB?2Sw3}0oSv%yU8^V*TuH-cX{0y2Xt^&C#{^AU@o$sJkHoi_x}4_vtt7S
z7x^9U({pg4OJcHrs5?1qQC|%vsvKN?!oi68&`zYA{A1m#)DE5EhkeZ(Y#+Q9wUseC
zX#Z5#=}2^?V<|2MSYuA??lQ4T-(0lE|8rwO!~MAt{L&6UY;I^ULg}b~Y2PM}&Ri;a
zR<6v04kjIH<o(P}lhPQmTL0R5s`EVP0kNnKOq&Ugph*Rr{#$J`*I)1aEPB9_qa6!P
z<lk!^m3ooY9<oxbqW!b38rr9GReTj%gGM{7gy7LygLdsdI()_yBp*GAtx&Xg!oe4{
zNSG%)BKT|u>UA~``_KA%c`+0*!a+PGX@b%GEbLf6JJ@dj)m<lzs-4cEgsO&-=1!F-
zszza!kZfUg{P`F01T;Jy!ncS2uhZ`0BwPolr8(G!tTz}Rnu7&@*ztPd5W2O~lHj6a
zA?(AC#~_knI;LY$`2Pb3mdp^t=Xv{x<F%WUK8)~4Grm0~XUP`IZB#azGyr8dYDfRb
z<1xRg0yr@|wl%}D9ca+Hj5uKxP*{YdGyAB*h{edla4slCTmwwZvOHH>+sKOc(XC6H
z?^?a$UZ+{SO~-(1|JBB%Q_w%=ct^f48&zbBW3<@{^1M*+wX7)B3?axo{ju$75wss<
zm6PINVq8e&#Ps-a_3Vr}&g)=FbaEex!5IgiBu)lJp93uLuOsJweBIUbK-r6=h)N{m
zuY=D$J<i4G&)6p%j|DZNgPXh>3p-X3y!DA)=fj4Az3YSni=kj%VEvK8z^7t^u}{k9
z$eDSg)2Kn^Fh%++LHm>I))4DiPUJZ80HUiLOf8qiMWuxhg{~1+Kcz?_raqR4#ySHH
zwTr+4VM<(;L_acw1oWwG3OODT-yzNFcb<Zz16yNPwNFcGCQO4$*BZy31JaXd>TKvw
zZ{v)9H#9XMS_TuA0~L+0WIwgfI35bsJNnMAn8Cf6)OXv9+5k%MVEh&T-C$zdIz4MA
z#Xvr@^?Wr2aGh3+9nKDzhwLy1?6Z!SIE)!`Y(j1tAUVP;dxel<0=Lg@-~5i*{>+fm
zR2Q#Xe$#dKIUS+qkun!TF6b=R-th4{CQQ>t2M8?94t#Dmd9v1M=TgkA)(@C2fn`AT
zt=R!8)8}RHNVcA)7(JQ#q_Gr?hwqNX$zS{Y_Fecqw`enQAW5J1Y#W=I?9JI19Dn{s
zzJj6JA3+rd<kMM9Ibb*J>XT-VlBgJxL%5r`Te|Mr?e>K^8XG7X>ikYV5xa5b_=wA%
zurF#;mNo&n#aRcYEvNqwpJ6ZVF3{kGtnR?lQo=b;_Ql=A$SkYq&{Ga9-Y4FQO;S$0
zywzS(7(GbU=%fH>7d+_=BkP?MZIvbn)?-(+-Y-d$%eep(L?PuuJHnC6PuQ0<q7a)H
zahkQYC}bc@1WIg;(Bqg|ZF5()k3hIrROKF2vVtw_K**#HxB{B&OLMj&q-cC-a?}CT
z!dH6H!G%-vN9?7ozhRmRzY5=j9m&Dul3j!M*Isr!<oZpMtTL9KS9@W{BYk*$&#r33
zvYxCj^7bbkh-CAz{)8>-1~TF`7F|u=Fjey-Y`J3dl&-DLM(P7}sa&WAOLCLH{hYm`
z%`c1HX~bNHJ*wDH#{9ps9=9LKQ*|exR|i(CO{H6PIDtamWTVyJo%X86ed-`Ff=2-b
z581)=z0<C4+~=-oYSOaZNs#dlO16}oCX|pP<@TC-nH(nyk&2j=@H1x!75LPf*N@!6
z)_iR?XDPX^f=4t{t=c-Is?By?eW(_D7+&H7bW8-8RL)mYU43|n;&rd?MpA|HC?$ir
z9hlh9twR)izP^nhNB267ppleYq@6r#%L}I6Ix7^P^9B>El&1xojonb6*z`q7$KcTc
zV(mEi>>4I*hKkR)F}tyB*|Mvh-&&~4mc9JtSJ_R8*tGes4bJb{9%LBGy*3z&-Q1?T
zQ|79MfP;z8Xw$8L-CJ7s5QVk6F5p#@E)*wj!7RbY>eHA?$Cq7HpC|XP%#SJDye28&
z>bKp0-+gazS|airmS1PDO>TDEy<WfGyxm^c*3c@6$FxeG&Rj-qvf3PE238%r*XKvy
zbi)nHo!{}%(g~Der*2>#K=k;s!h|%MJ3w)JFe1)X81b;$eR&Q6;zGa4*bXeoXdV1u
zXx84)h!WnX{0Ehy3bjvs4q^hI*WbBf)2bC4ZAIGvMmd&?C~^lNI^q?Rwz4A=ncihy
zJMX~iGt#uTSi^f&UEQ3Q7y}(xE#EZvhM6uz%GLR4<c4g;SwO26uI!vae3~y=bK>eX
z-R!HHnKsF%9aua&l}ZhsN>?c_<JN9kmeevXg^VgN4nAp#m8%GAvTf|PHZVvXc>A4p
zvDS+rg(?^4_PplSvGLyKIeZ|rt7zE><qaqLqrol)wzhGR-V`$S#8{N{ccAf&D6cX|
zlYFB}=ZnXBy|LbyoKu_6aR*kTt>{6~vAoOLGwx{Ig^J5aA2j1a1-d9pS~)bam)W|;
z3~pPs%4x(`g;7pVqHNgux>ezWVwyWy<n#+O6PUajZ-`7QO2NIeo2iI>LCRvgRGwVy
zTl*b+`LKG#O~NYY?5=u@mbmOfJlq=$>pnbMe~;US_DOe8%y`bb=1NM8KWuO6x~UWa
z0<&2~)ZN8@0H!`l4*AG;vyBawmQfk`yun&&Z3INyRM*L9m_-&2IItMYAiav#6HDJ?
zo7<PL6eQ?lrg31&<~IznP1~EhCTEU~JI&Z$4zfk=vb&R`+;y+>D}|U%w{En1+6xhi
zsDFjJ93eFG7aUC7E;aw1_7&AqE$>54OD|SLHA0`=Tg;9%uaSwBpJf*D4Z|2E!XI<4
zDk40tK+D>=Q>Z0w)ZS7O+?4aH7J!(v^uyhDUv)HV*|5+hGAruE!Q?1hMYQ66yFYun
zu;W%09HAADl3O-%88C;uwR;pRHYbgjXcGsMya4C8v@gA0LZjS*%07Hs!-s4kqQW1q
z?LD93;A<&RqJ=y2C`?Y<vA1{c)W?ej%ZQ|eA~4az12AN1j;sRv%I-Z19HDQqSH<oj
zwbj9km7=ikma=X`$vH<bfH=RS4>D6|@Fa+Y3*CnEs>habj4{7O1S`flyU<3CmluT>
zDVP9zEnQSV8)m@RQKSWE5uSIhjqWmBl5k-Q`)o8}3p>bK!c-7N34moc77j8~VHe_Q
zxc{uzuBI9_hN}sJv>lj;ZT-+Rc{{cxJ9_DZ^ZyrX?*U!cb=-STdM}RS6g$nf99xPl
zOR^f>5;jSc#bT0bt^t8d03r&F3y`90Cu7M@aS~Zhbvk<&8(735GJ(J-z4zXG?>+DL
zpP79CCF_0PdT%{x5$l}U=bp36?CCR`Qnl@SA;CkqsKPf~=ocl*zTCM@#0t|aLtxI*
zK$jGJq`K1FCl?irDMOXh5`&RBO-4VE)kf=W43{+$PnVw-`zC)OZBBb|_C+Q@AtLLI
zagi_QARnr?Wc9_QtBceur#;^iuI1-`E5af9)<xUWHdbW2HpVTEm>3jP*?Fq0_q}b{
zcZX~&XLE84OTmP57t@nJImoMIXOSlfLO*xX#%p?#=4kgS5R!N_rzNH*e;NDQCc3v}
zaXd{77y|{1_#LM`xje@Vj!ka2$wqDNA02gWlDsW4|Ao2HwzicB1798i-(WI$VOyZH
zr|n(sx!BJN(cSOBx&ek!5h0PlJ_eXZxT)?pY53u<1Cy+^eEml2H~PVm3NvHQtp<(u
zgXqICSJc3_={6{xxJH&;-LHI@<{>70<~cnOPZBdgX$B`)EpP2<luFJk#-j{C457BI
ztuv-688(H%p*GZYM0z<-sn48=Yug)2D>^<<PT~rMH@vlI&6KOHb_fd`<l#053F^He
z=RWzsb=SVmc2o}}%e3rJIZ`WcCr9HvRr^JO{tD7OX^&J_Kp6$=AnPPqc*<$XeYp@g
z!F-h54Z(kRy-h}Fh0Lj1_X6yFf`vKxJxLA*9>(Fpb)jUGCB?#U+}_(whKzk8Q-oq_
z_MMiZ4C!LpjU<Di8X~71yU)I=?VM#C3#3yG0HmpHRJ8}GOE9Wf%dc*_#xiQ8ZXU(3
z2}GjmweNk$yUp7$GD;)}O9x{*I6f_-RjV^u%qLSrM5M(FWpR67F7FMu-(-)rd1eK|
z+>Mq6Osl{0&O7a~HY|iA?LHNALeb6GQ`iIT$q_lxX?oJ$-{hhY=CRcdJ0Q_pBBUk<
zC+!36`4u}Rh)Cx)5rdQf`<ndh*0E{lGT}Ebxz)aQVbnqTU{nq)raNthY2(o5-5-l`
z&c3dG6X&xq>OI6rHkg73#A7ba?X|D(p3AZ|r(l!7l5`+|$UC8E(@1Ht4|YAd64n;-
zQ5=}o2TSLGeM7ZJVGHI7TIImXIHFC8fD=kudXrSSKuC1fzOlZkugNl#!T35hkCq}i
z7Yms>HC!%W`=&*|RJf*SZkh#YN&2_}F9%U1k5^ArUsmitvYmo-v{`~^I8W79k~>#>
zThWzSaFP<6_Msf_at>+nx*i)C+F?(0v!-!s!;&fu=RV=BMekB+sXb8{<Y}1Bhubpu
zd!ydntnE+RH@B9XZ@<yIn_FIUyM0T;AT$wtf7}Yaa5(Ao!EdcjD1#;?_K2}V8=@t4
zbBe%n<XyCH>y}F;<if(?4Tg+_N1QHW?)vtV&t}d1%^d!MStL`b9B|B3s8-DUW$c@a
zi)DRB`=nwIArn||99Wsk`#VS)9<-0tL;AW5IOnjZ1Jjr`hMm*CvylXxs@uu78g?M@
zVU%Ae*Z=ry-*xiY*YkTi^K|GfC!v;!###bNB$;;|`M0PJN=M-onOJOvZi}Ar4xzOo
zknFn~B5fq1i!yQCr$MCmFhyp_im>mg9xbNn66=~PzqL0<5o9XJ2&gQIY~S0yuz7%u
z)N}3=36Xf^Cd;^oqtMjUW4|wZH@|aci*p?-i;fl};jevvd*22O22%sheG;K{>(}04
zJ3CZO`Ax@>L<vKurB<5AjD@7RT^(AbNUc(J5xwWM)NVN#vZo5rN8)2nqwP*?-9W1A
zs<msJ+afL6B$BqH7$!!jJvpMuDFO|+PU=T=u<h-BqWl3VuA;!u2d5<;yPpD}``QS{
zDNEwq#he5?=0|W^+h1Kx=W+_&>;_1=;;t?`?YTpgiOKNJ&F!%R1rzotxasKJZf`I}
zc54jqVmsKrhd)TqA<BD$$*8FioU@p14>kV4aaf$}CC440n!!aH9q1_Ka4%tfhjWxm
z$3}-IoEu>!>9Skx$fEHmcuJL)Gap*xIp(yd-9jf_Y^NkEm^F+a;b={0IZ{IiRmkrt
zZ!MAq4#aLyon~bfJ=wM`LfOO&+`9w3g8g>{GJLFNKywY&hg}0$thR`uWEf&9nQP=l
zJKoJ}B%jDt>cHx0#W_g2DGRBc$gflXNJSWWLk=Jw-A62jwntZX%TCsdaT~Jf<19S~
zRx`Yiu_r!Jq|r`wZ_`<QIzUC1QN=9M62F9ONPSK9lVz936baB~+nYfTxi2O3ELKu!
zkABJw6wwoD86k)5bR!|*0mdWjK(s{%M+n{P*_rw!GCw3NOkqqa(E?6O^hrW%M&K#j
z{wVHI>G>z@Y&W%1e6_F#9hlx7_S~oKTs_gathio-g>uf92&=NE>N$-|NnY}91Is=4
z-fOS7^C_^r;@X!umlaz!HPc5G8nz2<RwD`-5gg)_B04Qeu%ga?CO%*6z7;2ahky!9
zy&mI4ma)EF>fZOXRyJll0{xv9@st&45L2R__H<G|{9nqCrj~3zuoZ7g*zW9dNdk(X
zlDe}CA}euON@yY^1y!7<3L&X<&r_$uZ3FADD@oN<jf+Z#&TYEuaGm3GpcD>@$`9mL
zy>}Jq&s=NPaH|tq_k-O~g@Q>&ffYi__bN|HS<G#&o|b-vSxs&(2PP^oxs5=CXYv|b
z+JJMNR=KS@Mu(ve+I;u1BM}#)=+6d|iX@-e@%c-3wLORaA_9hI#evoMMf`X=)bl;C
zs~^gzbwat0?^BkP1FKzzk1&Ww*vCV5ge%yyO>S9+uea~qCWTJ=BVK7gT(ER!VaO)v
z1Xv>6-OTBav>&nKb-3DPBKCoWd?}q-{GoGk2kByh*(Yc0N9&hzBFo5$an*t4bVYS;
zNf!2F3mabk12#Dp2~bxFWeN|aUA^W2i=l1J8S>-pHKE2tU3EY)bD3Hyt6IgZ*-unU
zDrbvgFub*h*2pIG6eA=g0g3Y`>o=KEtQO%ZB?6t6m_&9m+&7eW`>8fz>qjP?n_Lul
zkSxm*O8GQ8Z&=jqr`v==K;(IqKo2apGGiRoAq)01O`7O)_ApVl+-qYF$d%BypKS{e
zTqAi$TA6b*sjCN_r;3#7h=>S?r9TUmh<>gHDbArJz4KR<0whU{miVQVIVE)|El5KR
z(>GG=&$sV`&SIK6s!U(jQzDVMay%7FlJ)2#c?D9fs`S6mEs5M@hr&e69DyLTXrB#J
z!oYs9&5C5;sF?<&FgUG596DmZRDI3bqG{60>%eKJy%sF(p+qU$FQ0t2PQ>IO=uwf7
z(XAlkz|;<>DK!|Zb<(eXCHIhmDmIM{IJeaUNf^pW*k`}m<`Ehq10xbOe~B>k*20De
zH?q<)VutP48h1yypP+)X1CRoI@2chY>p2H~${#q_i5b*0;S5MH<2R}YvZ{$0(GPF1
z^nip9!hW;;k_Z`yMQ_M~q~c7;j+r6*t=4UHGY~PXqamU>Eip)W@NxM;t@hb(w`rU^
zUA+3JH(11CCxY4Uw1L42CRCshz%>kvxeE6taz_%o45w4QEZMGeR(>}pe|%tgf^)wH
zo%>W>WQo$?f3NzSmY+B}lK#mD-&!;+a#>36i_zHccRkP8!o%~s9ax~IkTNP8u|2a3
zf6%pbF_7`L6HZH##ibKl><`-xfl_r9B*V8VH|0}Kd$4~M-+9XZsGCrZ)q)yG&;eEi
zYj2SH#2wWv{o{ICh6bHncLx3gEN5a^q9s+|{-o<?9&nlQEZX+81a&^dRu^N_!N2KO
zkXKqJMAgpx(;9#TSIt-00I*W%Sq_14+M|2Q@=%w>IEjOxeD<F;*1G|J-o!9b^>`ux
zs(*KpafZ}9GHrj}kZt44mfjBMHt|^gydL36_7~kKwU;x33Rrx!F_shXf(&K*%eIKf
zsF*ZlSaM)FgJ}jNoK?mq>O<RK)zc|j?T4n#dV}F@0D*-dNA0h3NvQ>gR=|x&&%h|h
zJ^oE2aF8bD`eIO{i~lWp&v{DwM8?#K3B_g-e*CxfX?_9G%mk%^aR6@*c}QhlcG%yY
z%o2A}TyzRe+L6Ml8Q+oBOd4O;>ii-nM}odqDcA>p?eE*<(9bX+%pY}N@w6haQ|eam
zGp7XGKQvU*QjAjT7eG<SQ|L1}GyhmHQDrSZc1;JC^9@eR)2~3$;57D6?UnZrjEz&g
z*#TUqO|O|t<-f9rc$I<$vgE2{Q)T(_68q<7n?zq4oFx2mx`%ZiXwL9d4_cC0m;X|j
z8-@#J4)X;P$5Ca?FxGQUyC^9_z5Q#O1c~YDar(OE&N(f?;X+U4hf;A<G}V6FZ~s>9
z&2AyLK}H#2*hAB#J?2cmQPGXlbBFEUyLT#C3X@?hkbE!d7~!<t{LO}ify34469ta#
zKiZb+b2uVFEF(RRj0c!ZIOIcK0ejY|RM^JEFzr7ZryAsN3RngAx#GpNm3<gY)&8qm
zGD0Hdco}M_L7QX>qsGM0q}iz)RI(PuL7(*oOYMzb;Hv(pdM*x>jeRle4JNmrhDGU8
z`{+|Sf{3hSoi8E^X)4GgMPxYdv=`zb4~M3--#+G48%2soI!jE3s`E}uK8>pctYHqW
z*vFm<GCRL>jcP7%7c-#B2r9f4sa?pO`|aaSB@>DaTY4<hQy?`o?P8ecJHwRoY`i4}
znuzXTJnT99`1Y2FBv17}e1x>`4a=p3J&8*cvOQJIc5H?k^Y#hd)HW6IqytNa!ZCC9
ziS+<VM~Di9PB^ec(yYHAH9kfqYmAOM`=surST;dVthx@YM7M==$66%%E~xh>cWV+c
z>exF%#R0v;1Lm~X=%tw}dqAaD_u|?7lvBlmBAF0<0)XU+84Ltlr>E>wPle>P0+V4>
z%i&)1bi;PEh)>I*q966gotrknQSnm_*$6|kPd}BcODcFfji51fg#e{=+C7ErBSNS|
zt!PPV1iBo+j`SHdIr4}I8StBUgUJajs;0BaKeJ5;@6mI6*=CERWT!nfO%yL8LJ-18
zseWF)JY$kVPxA&7o-2!<Jaqf4dU?tt=@m)56ejL-+G8q|s3?6u%CHcuaNeGuvq{NZ
zm3i(v_pz?w^ozA(T_9-XoVU-eb{Wj1R%qWqS7Tk3p+tM^o&(x>v{UQYUeHDXg*jQ4
zb70XUkvU}p1b)m@1kpa{ROxP;sW`EHQm0`6s+9~IFM^^tt5V`Vw_eRgk68tE9auaq
zMk+QXPqqFYx6kW3H_J3+%sDVg3|Q?3IVEcX#rpHRAs!}LC#+ihKH~;(5;^T6{I_AT
ze_^6#{20lX+hFZO6dHd~H$agGZ5r9Uq@2={+fBAhuqb&#ocG0+K5t);#3qOrA-Qex
z!n+<v2-+8RusA0Q!TORHEQEwUxmlxxBz#fZ`AaGA#!c;V$rm}S@Lkg_q7>|lt36e8
zs1~xSrcEJtI-isF5|ZH^gd5nG)DMD_Gmb0pKIxt$MXHLtw|4B>vWdgT`7AS_*xAb%
z%3j>9YLuh`RftgwqDD)8YKlyI$>T4nM|1nNYG;V_PD=`Wo2dOC5sWX*HHq3#7%#Yu
zC=a4L=l8zqRd!7y<<#^Zaqf!0N;(uepSPE_*+t$fJEa3l8hZ1R+wDuco7--8Zfa{~
zUB9d;LV}IGJl_)q6m-~uB2j_qJp@)!-TD>Xhg2?AO7L+9rlITHwP8hv+qE^z5@?$C
zn^}osP5Yeo+8g+hd5LPLr;2f1;^}o?^-AY5J$1OttFXT~5<9Q`RrQuH>-uvc=)5xV
zfQk5`m*O)RuveaXwh~<~hk5hE$TLr{`JcebPFo-t_z?acm6Q=nl^K_y7kgFp6qjb~
zK9wMfBq*n)SgvAyGQ{=z+)L1aN1W^SW=Nj!EmF@#%D6skug;M!UBA@1O!@<V3(FSt
z=Z22l%MO@kjFu3gqQXh;d8)Rhf(oV2h<D3ADog9E-B|6<(PSXzRm0qAk1uNrs8XA%
zzT8y5**8b+Aiga5UgDv=nHR1;=Cp^o1;icQWUpzsyiq$$B~XB*Dt!mXXieG84R==&
zN$lLk%8PhLl0g`c;6HXtL(p4pzrndlatOJR4ytTGEyy0bwY?zh;*G?y4V@OD=@@p`
zzTK8P=sx+4R7HvV&^dQPb5GdqHAE%bH(@6kLMr*R=U!evFiLS{O5Wx?&F>tet|ZnF
z9RmU^Ht9#0&5t1BzP53taU3ZWSLHs|Mw=kzZz*ili-4f8TK2kbZDdMKgKDmmG@Ct6
zyGZtWtRAngWonT}1n@er>?3~5I4AK9^-#nd1pPbV4JP-6>=-BK_SqZT`YsC2K%T>5
zmpDax+99`-)v*apW7xjDUY+S<A<%h)?ZaG+FwNrZvacwdUa{1<jM2*%pfKL2MM9A}
zWh=adnQUn-2%ZR}Iqo#p+h23DEjtzI96L5SSBXgQxUD0$yjDKx=>ayRtf|OtPK#I=
z2Ta~<D;gSMJ92K)U^O{4Fo>;tWgACiamh^iJ_nZbIMmxLerk7A&n4rwdQl<|m1F9(
z*T(0RcjoK}Yk6n(f7Qn2&SjwPLioB=4JftU)xB5&b{XXZ5k;m-9f)>K!d)xxu~oUN
zD9J+EayC`pfe3-u^tLO5`R<&WwKLOE41?SFCQJrrL~}}BoI7Z5${AQr^*s)1nR4#4
zcM)ii+vgyq&}QwPZdh`frdA8AwqYjHV&G0BvDIywZ@KYy=cXQ$%%yk~Wu}mcXlv@F
z5+|k2o}MFEx8`cYI$PUNjT*(h%}6rS4oGXca^uSNwyxeAz8>DFfJV_ZiqLjbxe(VL
zTQ@syZ_fF-f9d))tJd7<+{TjaL0I3*@j4uaVe9MVIR`=~S>5E)YAvanrM+EUc*-`^
zXOkSGDzRx8n!LtHI>(&$WbCK`SZO(QZexS2q#Dr_mOBTczL3hkiP(i<?5JaQZ<~8~
zxlXJ;3|=-^cx19?c3(XwbW{~=Ojh1tkpvmufpd2M!s;w>Zn9J8&lEs{I6lw~LuUBo
zD*Holkb(A$)1K=|*oG*R>Mx}-{$Tx57M!l$5VS?+Uh-g__L%6%ZRgDw=TF;P>NkNA
ztfI=(FR~YTYv-g9sVve-!J+O<ZF_5C{-!Bv>5Myfxux$4d%wM{_@uUzRFv>qFZ3wr
z;6djpZ7r29dV8CjVG?n;Da=Kthq0?Db!h+gj)r?^)KaAIHZtK)&K-4HI>s*jTfba1
z9(!laK|2Wv(}*{iD#l#V-oTIYm2D(_#Knmn&V9UB(oKlnLaXv0;a#zT>Jhy}%FTHq
zO>}Oz(<0h1zjI>`*TANFPa>&yg>)LlhzZp%dDot0<x;N-Q<h}r6E@gJ&TrHM6h{3L
zC4iP1`Ng2-5Exo?ox6RgeyMfWN{SkUsmEK3)>WEwS830BjRa8UVQ_>Z6o(v8Ad#q|
z?8m5(wz(UWmLmqjaxKX1w2S!@mcXVP&Y2|?VE5eq>#uk2W0hr(kiqEyE0$v&X~Z8Z
z<sd;z4j_h@2pIM*Ca*Jq*l6o6#RUhbM8t^_fD$7&j1DQA@u9@EB{$vdT#A@c`tBjy
z((rE8UCW)D>XB^6Eh2C)n&enFc+I<lL{EEz<uQO+<6=B1To#Jl;oL^Ql|3xv(V{a;
z@RrO@nW*+mi`JqE9Te~;W%jtY_AD2vcJTnzaDGowU7M_JmQqr>1|-KS1XY@CL{K|V
z)ppWmECYZxv~BJBCj$UQYeJ+~#EOvRkkhW28XQ2_dRIfE;87K90DV5?w4`XVPJtJY
zvT2*Dhr0}XSZm5LttDj}hSY+;tzUC2tk&6cvL`i|IxQMN^#Gf0I~jacBEWI>T7w;k
z7`&Kbn`u}?q8w4|eFxBO6#2W93L&)9oTxy_wl+^uh(@Ljo!%nvA}ty5R0e4DFUg<V
z>xJ6K(z=KmrH^UrzJj%o_M`yANCft9y*^z9i{eq7mQ)ERwvOA5_86M2^xKGChMb*r
zTB83lk+xc8l2jDVPe7wRQtjHI=R)UiX&tc)-r5sSG|h@&It%HO#C_4;-MYJc?TVGo
zU2X>EA<uVhS>f-g1<-C((E1*HL%?coC9)=tCBDXR?Y;H3?7OhHoLEjvN_yKk3bVI^
z61MI2AeSnoat5ITi_S>tR|S#F?85yhE2r(NyKd?r1Q|6X%PN9_mgIVHiK5)*-P6kY
z1&KwowfIg;3_xn8Tm_1Lg;pZC$rkyvy|3OTX-R^$d~(3zH+kN93)@To!ErG4(fo}p
zVid+ql?RykKJGzGE1MOO|5&|6`o$HI2ki))j<SYPHq>cPfqVPN=+v0Kzpc_;gdI9}
zMY0rJw2w`rBoS#-A86AE)yPnROWt7F)nM)RHTgbvJOMq<T}+zjhw2A#D!qMeL(f}o
zy3x5QVzN+)FA8ef*CoXr>@iY8Se*NAMoV$$g-^n<lb3&e`xHAyT~~!xVD<I|nKgOB
z2fMdLb;8g+BSgKnDD7dFSVTT~r-Bw2+Z?~4y#;%^j2k*fh^JWVuwFRrB`L|Vq?JEz
z-&lRpR#AD0OLC9M<K?t;Bb0O`Jzpxc>;Q`V`ljSLZ+yusoy%NA+(P(5AFuDEBq3pP
zw*%{DHjivCBe8v`$%?&)2ghcp{oz28V%)QKqdn10WKs-*TfD)vg36-Q39%op4?K5w
zQs2;NsjUL4GY|~=_RTdoN!ra{ESgbShfaHkDWX$*$-@1X`b{$i%DNJUe|m5c?4mvQ
zx~f_#v*=sf41%~4vB3<me!(X}d#tL&d3=jtiez5)EwAmeqU_t+TSzjI{Hv2ZbifW4
zVx;8#PJ3>WKx(ziKu12TeS7N=n`w0Oz{3t8ZRe}Tb&g`zcK(hW>io{pi4o4EjYM!@
z`A%q2oFpWIhEL_PeWaw?O+Du_=BN!!*_z+k&6}uaY<CBy<;1ngoP1Z4TzX@hCb##U
z`@*vHr$w2tkfjyvFl*nP8<=8)5EbVx<sJ^A9><UIJ+1G=HYnN)mD7QVFA?vcWSQ^n
zMxubu)HqVnY%SV>uo7si^hnGN_I>Rut9xXKjXC!r=us~D9{c|4uxPz1ji`R6CS9g1
z#hlWfn8RzuHdTLbXTD%S`5j(2n-L{OY*+i#kkT6mocokED1wyZvE2(JU~i(<L)Akh
za$0hqC^2MDTMY^ziR{Zr3alpg7(%G9I6TBq%&0SKd#gR7fEY!Db@ErbrOGYQUZ};}
zVl`zd)KNwI>X)L)lEjo9t4`^N3FR5mo~bP;fI~f5b#+e7{x;fvZ#xl!&b{yv3o{3h
z4PaEl$YlqTm_#(g^k(Na=2N&Nqt215kJ-V3NyRUj(HkuHe{zXKZGs-6Vih+juRJuc
zWzr6JLyuaD3hH2PWyOpD7^l7VrVd%nBf4XOlOi)r?XV;D7SYl)>MJgR6inWA8oMR!
z)p?FSL1-Zt!j9H&ppNigCS3w^EKLYHPWyFCil&<u=`FIN5n9DJGjUKlQu^(McvtMn
z`bt`<6p66kfeAMT#)-wWV~swq;~dZ(&TX_!?ZD1<yc>did{IrGmy40XMCPwJ?bRKf
zMl15Aj%1JvIp$oeGR7j>X(#H9FrB~}g_BDCl(87v>$Df$*c;K%IL7T{VOo^gVv3K3
zYgN)IEs5lBnOJM5(onZ?!nq363^P{-g)&QuOr5RQXRKfdx(=+KR#oIe^x$HrtFt;_
zD=U=A{w6U`0^U)lJvytIRuZkMSe;qm#ta~~IO`;Dk=&*gJmu}F%uGsYELW)P269W-
z*>3v7J04Mt;<d6|s?rSYN%07}BRI~^)kw4u3I$9nwKQc~qNlO}NIO1r^-+7OTMA+a
zcq0M$jH?SdblL;UBkWOMga)RkBO_gqLE!Y~bLXrhyol5U?vl*~zktMEs4fXTGc`;a
z+!11YXpsrzZs`8SRJJxzbH(eLtPx_G+w%5iwmLU4gj6%+^Ehlz*V}_)G1*c>gl!kR
zwd8V1*hr`OLA%^U3gYD>XVQUmlOsxa0e*g^J+*0Ia0>-_M;u6?3F=E2|I?Q#dGiMv
zn$V|q0Fgu{ZFE`0Xiqd@+);#=+}<Co-#o`mgy^gKOvmO<`9gc12&`B%Evw4@Dei2p
z_3th`_@gSi2PCH+rB8X!_Dma=^a#3w-^PIj(GMQMtJ0mXvn?bq9JKjd2?T&)@@|4A
zQbKJrR-CKtZ8$6xQLpU>ELlL<q(p=z^hlxnp>BAxkxDXE52!Q8Us9`__WEUKlc`@m
zTffZRkW>{xEZ8N|i_<P9NC}$$sLrjz!T-bE%XEz4K9v)x4$_hwQ&E+$Ok%!%B*!to
z6Z0pP8D!{nV5->ks+Zf3Cezsp7G$ymu6GBR+boX6nzkRSsi;tyt5|9sm?+WcR^Den
z-pnlTyxzHvk*b)bQfU+y5?Y%5M71Q9((I9sfC78SX-R~^W}R;L^Y)WTrk9h;+=DxE
z8?Qy5!QE5FhM4-FstG&<+fZBvsvI9&*w(b`<??ZlGx~7%{&XARhA||+3Fp4s5k!5;
zH*P<Z&)iQDBr*=BocnN*+_c1gHZSj7y2k7B?tyXpxtxWe(cX}A9o^N)3GNCRPU7`0
z`}y`M^b)1(;@5LvN!D+^<u&#T-2-IWfIb{H?!W?#Ls(<Q<v<jhmC~%UV}G%07vCz2
zka7d~0Bpc%PZKo9;f<?`uKiMzNk(=IO-&9d`;?egpz$3kN}?YT^0BDw>Mu9mxsou?
zY3DY#;2nf$X$$^J#|#!1nau$82w;bC9Cn`4OkVejSJ<!SoUNe#vU8oYuwh`>ey!as
zUAEM@spUBWS(A)9Yro#Qau+tsv(9Y_G(`3e*>BVjP$SqO2LfX|SQLy5!Nz{GJ*TA+
zB|=PTu&l@to+10KdXAapeWF;099Xam(b_ca!V+0%MJdXI6L9<OwlyA@ajt8252j<Q
zB+$&Qx@5mo?J|QyI9R#J%??v8>X6ee8{KLG@a%W%**uR^3|SfP;s+s%JMBdssi~%=
z`Mvsq_MHS*k?u^$mVN4o)2^4VA0>(8_jB^6C#lBf+~kI4RI(z)ab3*7Q}ze>lR*VT
zave{J?;QmzMZduxNbDcBm7(7JuydDNJ%UHW<YRx-Rw2`GVB_Z;SW+^A8kAW9YyIP{
zBbUR3NnxVEg@aCu`o4qkRr{0fWoa{1a40aT<Z{(*8kn*_?fQTU#5!F-Mh8~(MRh;Y
zYNDw+YJb)~8$yjpyHq0?l3Ghv-H*}6o;hoOUbqn_Lh%Jv5lI4{!OB=9VG2Bz8t8y3
z31QAwfw8~nCQ=8Qwhhm<(~>B@jz}8&%N)e~&V?i+2d0g+Y-ZU0DyM{W)xL9G#7H5$
zvE1|%@y7nTjT$nXfw)Rb!?xhG#5(18rCaQ>zp1Ae!o0M`3UH?-FgxXr5<|nP83VAt
zZHq!WzRLa*_U*tl94ORjjBbC|p5W`ULGagsEm{-OW+Ze%HV@k0*V8DdG9XJiaA1is
zlY@gZOeS{Nf2bFvh>q0>?vk9G7GcB+q8o`Xuz$?=&F@^Nh!q{e(Q{zQ1xW|mg3I@x
z8o^mJI*cC7Z7$wA$fwS>)^X?moQ(C32c0V|#o*|${Yy<wXf#Y%9uF>iWcG#jAS6Mi
zfGwOenA=EhmV@@MHJ0QkizIBPC6-E~gy8f0%l2<=g5W=r%84s+VELUe>uc*Z@0p_R
zQO_lH*uS@Nt$bLy$7uoqnCn}6AnnOoj|`>wk2W{!CULwvx3zGxzp#x$8vw2BKU)jT
zI(0;4Gd@k=7cKD#F?f|8<?J*1eS%+#(LHVdmFPkpzCOYqciQo9LWMJ%&9CJK)7K<4
zDkHeH1d@hUBI13SB%hC(ZRo8iL=p7M4lH^d{)k#9@tTA7(X;V2Duhis*D>3XN5<Gt
z`|M+8v+t=l38qvU?1I|*b9<cj_|1orcnEsp-zodp+5g>Yh<(UMsjzIG>b7o$G0=j`
z$Ud&?t=2Kv87<hdeW7JHa~Qr({t~5ZA7A}Y23kZrvu%JSnvd-uxvOWNFk9Pb#U%A=
z5hME!D5VF^ZqP|Nefz}O*t;^D;celrmRpK{3a_aweNufe4~CGB{;~s$r==c-_7ZwC
zwj=xGHnr#tir6|2l~{nOuG4<#=4tzs+49kZ6RMwurwRG3RcIs&w;DfKh~od$Zk1#v
z*KVWsm#h%#^4uk-J$Lj7b$W?g)0VbRt7S^fKFogfEDJP=Ku1cR^Hd;jaf0H;Dh!|d
z<I`)**&g(dPb<f|wMfB+yCOfx>8Kp?&zKDgvgF2Foy*+enNr7Yy=b3VFkvY6RM}6w
z!DPA9*ip&sd5OEQ@@>v_;qGaOtV%juvd@|=K|v0$8R#AWQLd%y?y~34K6?$9ynVqQ
z$#q1Ms5J1RrFNdpWkqOIXhga`p>_7zi9STDQFViZcY6+`P#`0C6>Ffl_kH$)n#lc3
zj&`vUJR~EF@KZ0>X^$RNOUTtA3Kj~no>sCL8h~N!b7srzhNeD19f8QuaG)s|-A!h$
z+{K?;osl^s(|+ZexkLmAEt&}(>Hc}MWz?s<1N8*Bjfs?mpsqa$_I`ewT#-@M9b=R?
zSP<nvvvc;s_8i1b;WXs%I<O*s^3jDxt}TnnOS|BNy(l}~*+CsZ%%D0TAC|1?83OqQ
zZG87rl$p)y0K{j<24*B4ePMlFnZrt-6ZZzDCnl#hvEA*9W<w0e3b}dOxyh{$cJzi&
zyX}h?eOcuIGHDFwk<+5-T5{8E_9cmQ5st`pbT!FydSSJL&fAN-S2ZFWk+X9bytPQ$
zl%~^_c(<3-Z=?gsh|`s7cUp3j07HTRHTK#|YhHyM(i%zq)!X=`nCfY#y*304k{Lfc
zx5ut2Sd`9VeWdsUdl5r>!~gz2ds%xGF6Hp*;nM~d%PQkdvp7M*WLE{*uGp96#IHvR
zp_l`A!JjLZI%*xJbxwIJp~}tex0lbR2{LO4Xs4?jx+*_P0M+>zWYB>&V$mV$IlqWO
zVm;NsRU^S((WX(!YW#-E5;as>UcgX;aLIsuh>26q%9H(>ZFX%Vk>dkno582>j{vNl
ze@f;EPU2K?txNFD>k<u!3tZw{=Ufo_r?RZ7SSg~Xe0zjX`?Bg~D6}lavAJX3V3M4e
zKPcT<d*$r2ucd$UALRd?|2O|TZ|D$3g-$>udsT;u;d2RNNQU~;QYeyV_sz~9vFp1x
z9A@4@6t+6BqG>AKrNSFhmM8>huWlT5d~#v{&kc#Zfa0ZEUWsQ<5?*ve2YWh}5oePF
ztGz?)I}0Q%jXhyEHiATsZhz9bO)kC-9fykZH+4PSr!abXM;%y(HmvXIutlvMNyYY>
zTqH~*TfDAE>89P>eVZM?Aa(r10f_W@rzIya$Zr@o?3RKh8P1RnV9|RKDG8?O1tTrM
zZmn;UIOkmF+t93}n99O<TboQ!S`Jdl74a6QB?;1IP2~h?;q8s!Z5^8#9_|e}fKn{5
z8QYRJ`-t&G`!i?<B=gt1-)gVzdad}%)arl*qVhfHZyAGON0LYs0R6i9O>V%0SSOD-
zuryr|bY!pZM27C9J9VBRsE&q8<B6p3RJw`#C>(uzL%mfN3nlua9yl-!WPay6-}%m0
z+8bw|y^Xt9{onoP{9G@C6GF!A%X0zN5$@w$XHPaG{|(t!w6Cu1bKE_5u{z?2as$Y(
zxwKnXS%$@!AwGyxr<|7dbOiAnM_OCf;gN7S);Dq3TZ@7o<?L0?rmWouZF#*%TV7cp
zI$9Fg3mL%w2MJ!WuyVEeBFrrCR#Q%U;iLGC1Scf};>5R=&k@FxoR*~#A*!NgoTqY&
z&qwTq(g^uItk>?Cef9?W#^_GS7RxLpajT5goHq!aoZ1r0ipM*@6U!VY<NeNDk!RjN
zcgeXHIt@^1pl5gG4Tq<o_i$4~kX=z3ys7a$ZL8X{Jv4wPx$gMnIl^Y|&UbTi)NRgn
zqUtTLxy{~GuUl-AO&dNfv650Bq>oF&xu+YyoNii4?bqX1A9Y%Sb!C<NX0%XlPV<K=
zS!i|p8gWFJ$FDkXmQsGnyB3M8o#Z3cYfVGW2?{NbIk(yS6)W$vwXJPrfXo=@Ha4z!
zmL@oiaJf|S9@)AYQg)AMGPW47eBbJ&ciNj5@CQ1pEh{J1LgEGOL0D9~>6EQ+l!!Vq
zBoR?p21s;3G_d7pMeZQ)LY%b?4Xa_=N;9RTssoFe!%QBID%3RQ498kG3zGjf=9ll)
z@i{%`F4pD%l8%JGV~7&>wz&lZrIjCI9)rm`ElDx&BKa9AKV<hcMu62eg7_3J&~w0=
zi#yk^T=^!uKT(DPA3f(fkgg;+dM7zq`H2ryPa;o>aL>bpw*rf&6Q>s0yX?VyCui0S
zI@k47))HSR+4PpSB+Dm9oSWopEAUzmPug4SgPCI?A#b+>OES4}WM<0V)+Ptm8UZX+
z9cVDwK}3aB*8A<Xm<J*A?%c)FEzVHH$PO8iPx^iKj@GGVV>8Z8Dk03gb^roEQta60
z3QLibCj*1t*}4M{lQmavOkg$k^fD_nDGd!@+182>74ic~asiU+%2H6AQyp!fVR@M4
z7ukV87>v_Sdj=cyh+x4*87jPOs&4358@WfMS{IDW|ADCpgC{t9u-+~!l=5UW10wp0
zc09>E(H^K*9%9Y_+HXTO93i<TLJ{Dh$T}@Ct1@WI5|Nd(@&F%_)~(mxM#u!(-c`p}
z-(U#iY=Vdww$0t@sD!B|JgauMq(R!n!H7^NQ+&dPyTOL~2%S3Xz!HZM);eAj{-cq4
zT9ZOcL~H;{TOy9Zhp;b@Mm5^?fm3Qh|EyTKK+0P?0)y&NNsLKMLl(v#X%96_UN$*N
zJd|^<nS|#K-8b6XzHG;;A=}ax`yQbEfgKJ&dslW2LU?2ZAFFPV$RrLMfA`OKr@c7s
z+Au_2^(Ka=M{K-XoQx>ekwWqb9ro6OtO5RJ5LGDpHqn+PP)u?c@yYcbG%l#R#Czop
zB9_}^8`w5V-jiR!eay6eSImho?u89)ZJX%6mFu0Gj###Hev>kXkZJEq*0_0U+_|j%
zazZ@~>YDpys{5AchrXyn5~|EZOVLQB^r>3$MeDckZ=)bGH!1ZDLcRMo-P+bIU*_Dz
zkLC9l-ZH^9Q=bw;QxPGUJh+Y((_4#5<1x9INYhPn!M5c$a4u8McF?(vaWvzF%6Ej%
z$+ox7j6vZ0&TTxHxefclAC_9;hCSR~!Gy~9oWZq3<v1-lo^ANU_Sue_Ugo5eViz)^
zc029$31t$C7m4p;a<oV4mys;QXXPwUqck%Q37}nUVnXRm_U?wUB;4%Kk+T4j+uSE?
zPzKI>+9ShMBqcf30mya>ici{m8}jtAHn1zWZ*(|R9i3&lD9ymWs=a;yx)z!IKnjl*
z@=xkmmWV0&>KbUq?#u8H7HKS>w8SdQ8Y*`iM=Y9mXmb`VRl@xh?_d!Rd=P*BA$wmV
z5Xve-8RoPoKuM`La~d!KUG`|>OB+^f*x=knKcxmKB;+Y+jxcK_rBmS09?O|{*9=^m
zBGJx$_3pUnAf)QFy+5a(1jsEt=O)?S7-#v|2kP-qXIV)yXdPJLTX=KPUdq&MU(;Gb
zvDRJ^tAs*0Ey-{nnv5jsudVl%4F<LSpf?!d6v6psk!`-No>QEY_#6@U-e7?@Pz$`1
z_Vo+zW9ZTv&N;A}n#J0lWOOzla_0xTu_^IJr1ydYOA5VVq_;WB{Ma{iuaP}5)y>Gb
zWI8}{Z|xZ<3oE|*C@)hr<FFFjH?|?5yUReNI_$u5QAyw<t##VIDVN}F?|J9D-{#z8
z>l$g6I9VT0qBRWZ=CTNo)Mx5(Kq@~}(_VxsEU7_6y|tv7Ifxs*#&=*(G{UrD!+P&-
zmfySHKHM#0(gH=J2d{Bjw8Q3izU>|M&Gl4MqGe{5cVw5-QdCYh4=w*$`<8AuRoy&x
zVP?jLquFU!5m8R<UP;`<^gR)Y3LW^?`pV+msMK?5sjnQNe_}lc?b{Z8rH*eY;+;`{
zw5L=)+?%v-?<Vp|$Y2=k9ayeDp+A$@q41l2M`xN+*)(Kw&P;TGiCG{(L`;bLNgm*@
zqKZD!upLhqLa^8UP;WEGdcYOFv;7{|nu0YHHXFZWQjUO9v@fEfc|iC9xbboOuC_Z@
zuivoLxr=^)R7%a{_)L|wkRc3>5%Kc7bHE5}Lw)B)t4H~xLHnM(M(Uq%t`~8Il!8RH
z4%+wD6oq)nD-R-M0!#8RG|ACj-<R(}16<-<=j`6{np^Dq>!+$-n=>@)z><~p-ZeA2
z&34u!R5QYsguG0G3b0&&rBjM7P)h8sZk?o+$j*Dvfz|x$=-w>%#Qeiivd;1D{0(W{
zy_qeOgF94A1Xxm_A&wQ?+_ybBkY{#sI?W#EE?cR2H^Eo;w!y5L-00kB%p96INkQCw
zIc5%MKJHxSZ1g$Z&-T|$=00pr8ySRY(cFxVV@9(B`GOm6xb;@&I?))FBfJE2N9|z2
zcnz{x$Wh*4DF!lGbBF9ueQt4BWh7EKE%8=uTa2a=5_>qifE)lj<t84G*riWkVYMSU
zmU~nVV#@1I5$jayIa(9mypJOMlEoF~KxCTJu6e-w<#YS&$?8~OQQ%F4EY6*ET5^4{
znkmbJK(b@`)1etWDShWII;Igy=8`Mvcv8RHZok#JOdQ+v$dsLE))lBd=b{ipqZ$u~
zxsHOZeE1}`PM*r2tXaO%PPQq7LrR+JT+L2Pd^m&Aq7u8EYSW0MCX22Z1hDevao{BK
z7pBG?He18r2@;=k?rQcgMjY=k1gz5?#vY73N7WwYg~1<@q@4Dgm_DK0#Mog#J=0{8
zXFdeAe4PAUAUQcy&Ov^Z6k%s;L@ldCDt_39ofd7kX_gmOVdvU|x|8~e1x)5}f?B)R
zD!dk(43y)kHfyU_t#EGQyT_4oo{sFx-FCiU(&FToqr$B>7_UM2izG|C&^CRaa{_lb
zx3zN@LV*`4n|85YoB57nvnRd5vSf0Y!0l4ouNlba+y@cce>HPWQjNkHU{BW@rGhKN
zl*%LyEGZEOgiMav<@z!FJU1bPCb<kP8QY+|41O#JZJ8@VUC<ubXII)vZPX#8Q<K{r
zNOVWC`<3LL{95s4KhPFhp*fReYxEsZ%?TA>zzXF7=m3Kz#s6U9|DE@)axP_JUN>WN
zIfU`aN2n4p;M^wt+C-)pDPhmF`*FnfLqp!(Pbuhd+U9Fk$~wvpI^_+PN@%c5yV~$z
z(~h2VQ$%uXKsiW1)V)OhR-_#)UkwJ|5Bc&4gzDM0I1ySH$@tow^GBVQeB*YCh`<<s
zIET7n<x1zW)<_Xd_9OXZe`b@8iJx}vW733wstR>7_Om38kiG0j>zmS>q){E|tTwcy
zC8Ll24Qn8?4iW3en#?sa$p?rG1rSKT4~&wHUMHjccy{ltZ+oe8S(I#YW_Uz}L+mHo
zV8UW`*triRAe9?oP+gEvqIv^AS??BwSB)8+Tm&qqV!4EmY5S?9DPaZN;oN0!plR~t
zpF>&r>6{^IGBbd_b6Yq^sdQ~)1YudG_Sw(WTuI_m&f;k_1sRi}{77e~o%VRST$6HF
zAi>(t*4vVorARIb;DFPTJa75xwHxf`5<yf1X2iKs7hoFl`4a-Qd32C`9M0SM`FdSy
z*|KN1AUj%WWE>^MT*w*s3+-uiZm|w|Uc>a^T?-*8qGA#D%*+z2+UAeIHEa~led5pX
z)aX#(eyP0@>G)g_KDmBG3}}fCK=TpC4v3l9FSjw?bORZS+~xd0ccnb&oW)<M_lYJ5
z2Ac3b{Ekjbe)vrjliMfkSCiY13~VP9wd*52oax!G)r<2qs)MSk<iHYZ=1-zwYx*m=
z%znLyk+iT@6ljr6FbR|Ml$3{|$%wMTwDTK@sUXr|4{C>L)&bqbGymr<AN$QVxQT&n
zqr=Xv)N;`>Y%~}d?6=wwmt*2Z^yOZz2<rXXZ?{Rkj{lsyw00_10AWI>11TL%`A2y^
zv|>+V{PsI-o~|keBKbP7v<M824U8*%$bNVBS=ECI7w&&^;yP`ygpsGr8zAELdj)Gn
z4OJtggGJ&{82jz_v)?dWvTU3?4QSKY7W;#|T)J+9b6M{{7dBe#?)HbZ?ZWy6iikML
zn5yU<M(DIlGP&-$S6+9W{ZTD@CiKX&uUrDk1UoH>eQ;F~3v!k0j}x|mF!|~J*4`wu
z@4%9PaOmV#*8NW!McWjr%#Z_!7iH#0)Ofc)ZDed_f|xw=n8qAXAYiMaDg&RaM7!<J
zszt8ZRI(#17)K>7iSyW`!0SJ6LK5W{AWkzf7(2jtujoJNO-zJj80Gr>MYSjkT8SGR
zwJyqq*?^YRgr|BM#gP4F0#;EjxhMOk19+Ym9XVB1b!3*#2&=zp_%YJcX%bjcI>-P#
zniF}a?XPnIufO(n&SmvZDn&Ry@~{0(H*L!04wP_UA@3@f`jY)^_XJKx7FS1`Hdsc7
zQbib{>hJ0~Ne%J<DE$vuQd17wn6|%fZ-Q=<bRaVC4JNCh9FNkc|Dk#tPJ2-wOw2;9
z(@skwVvrA*IOW3su{nb<$!2IjtH&zH>3<e<Ob0bjwKbeXKl-P3t04TobDR7Q`<lX1
zk=*~eftK~K^G8q+pmf+_!l@h<)Uz{S3+ylERoP?z(yg~d(GuAp{HL52ZHX~LzkBwt
z-5?e2nXZGVUE#E74^B-!V*l3WkF!ep&Rwix8K05S#Aa}~*}u1$O$(>os}3wz5!K;>
z{YUqNLK5VFeaeAJXU4*~#s0J2Eu2MZ7b7o{OG|D3$U{d5I%5CT-UjoM5e%cXFTk?G
zitKZCy8ZSP^Ap_17ho!<5m}D*QQdP9Qbdn;Ik4;KU2(H?@1@Eo!CnC%QH2DlR#=Io
zWFLJx=SNS0F}O-4YNE$EExFS+PLUtZ;lCeqx=E#ai3aeTf+kLTBE%3>L!`fNA6tFY
z!z>tr(`g5m`zpx1gvzW?aM?btjfh^#^TA&SmNblWYoz>teCySwnQ4v<pK<^p_AQQ>
zBu$gS$oO=~KA~Q!1U)4o9+=<l4JNA3!R8UD^oiZa({+ap(}C59sCqgKIqcadb@L_0
zB->xE<&#c})|?|6bYf0=!#=rQs+vl&6oT^(tmIV9i84xM`RWdEw@+#FLjjFhrWqZ9
zE@H@WTCzJh65>FpzI|$Yxa#dIYb0_}oR*xp)ug--efVjo198`5q;amLG;CUCpMJXW
zOdW+cMO_>q`Md4prDN{;jBW_UeVG?vNuuslpx$S;Zf)0LpE~Nt0VVEG>jzHsusyHd
zBHx$F8XQK{Z!o3>xhAc)pH<H(F`VEg`0D`&79@aZ_z8Rd>7YsFq;zg$GE=xs(Q8rK
zu6=esKELzM^=t24=iH{fa%jX}&<2KwMVY-Sj?-YW)a%GAszll6bk8aMEv=})f~slP
zwAl940<~U>lVec!xwSZH^+=H}wMbeLX)AVH){{J@C~AV+=e1?cs5t}$O`1|cXo+8D
z0EFCKEx9~&QTzP*rD%UD^_e<x!f6SbBwbva<G|b*d*SI6xfm(@hlq7rGAJeV0Jw3)
zdxkTaXB*z~Y+3+AqYtTM)?U<>4pCn6?-kO)fN8!bhqesZ7c`_HUbN@j6@}CZ1A(&k
zg>9z8L8_S~WVB#uh1ti(K4@Rm&1=*VLot?llfX--CHg3%ATt($KH`!Ja!@6%FK%B|
z?|`XF9kAqbI%K40U($Aj!ieu5<*+><5p%Pc!ia@b*dw3L;5^0RNWF(ZLu}93iwj8-
zF(T5677bS)Zo(W-)wC+#r?Dv&!jI@4dr5Wq33SdU2)?DJsRPSc4(vsF8fM1-DC+4D
zYDY-T_R_YKmnr<B@7%|)g7xT2_&5~8F8``B`lFwA&B7E(%)=0Pz#FVXH-cPVmb-F(
z=gqg@<lN-GdeaK~(hdVjx8(&SdIqyN>a^7ANaRI7VhyUuHMIViy}V8R(A4N=sjv<t
zDIN*Mh;upt`xR|eN@VU0Irs6>6kIeCUJaWU1JNeD_H;-F%U3UVE(7Ple8PevddIG7
zxCD-i$DfFurejKJDGW`OT(!@?tUeZ&Z(%L_rFf<|fzuwm4JyGhNTSMsixc+BZkd%8
zmv$p9go1&Y4W~T<v<9}=t8!*l?RmhtS;Q@c9vJC#M50~a`WHN1Jv#2d(zLl{V28cB
z9?(V-zs4-$4F<(!o??sH4b`<&G^7G+!4~pbX^)AJSzbZ%sta{v{YIFqxVdu(!qS@L
zwNPenYZrTy0526AD-swd!gauI%6F1f*4qX*Q^bJ-i=pC%g&%~V%3pLadreNx7V7wM
zS)GAm^WxoeN9^W;WlT+c%z-5b!q18Idctn$CUO@2Ij27Msne3?A8LbaB?e>j3|>~d
zwHx6=6^}&@Vy8X%Bn~@@p4e?&ClshoT<X=w99VW@=>#X&;3>Gh4V%O4@M_1ffu;SJ
z5D)@Kvpmd_dYgoM^_Jq7-e3`J6&}q~_S){Nitvd8#k|2%cb1X*guSjkN7Sbhp(u>u
zz_e*86tcx$U+ox#B36-jBL@~cE9qB?j`&*<phL;9H&i>~dg$O3Eug1SAVJ|gPJ2oT
z&peJw^u|VJmriXW7?<0ezNH)oqBzkaPLfk0gHY~B&V#bc*(Fc9>iY9kb_pJaWhQdI
zn>~-34E7am!q-o3!l+49X}wk!BQd3{V*q0*3=kbZY)k94Vu~tfDF1a{Yu6lc+H(`a
zZlf#G+nI?2wk+q3x_@KjElfe6rW}}twrXNCb$xJIE>G-Tx^jbasVOI~a@tnZOX&jW
zt`8|mL*clz1deBm#(&UO7A*a?>6`>sv_Zu{q<^Rs;Dx!Pc1NSf>xP``c$=gd#hB=X
zy*t~|F(HyMg=Ar9<F8J{pgn#gdMNpUT8?&C!I;F9*jZv=(F2i-{I)EsaxSduz91GQ
zfm|VuarO|h@HwYF_i<$J%P)T1?oR4O-Gv=}=O(Q=%O~q-!Z$UP>5q<$OwKs>$sBN~
zqHfCIb5H9dJ}4YOSVs0cu-t?MHnG2lY;{|0D3R1Q_6Hq&l7Y$=xyNb0l|>SU1g|Fe
zW<d}NQBf%c!YBz#$KaAYDP>?zWgXYFUM(M(A>v1|w18@zWZ055;3O|RHMh^!wiXDa
z+?JHgWB#6+Kku|?^NenuA^^qK)wuK_C6*}H7+CCu`Z#MO9ME9E1T?!hHyn8LM(0{@
zyQeoGW6b()mM%(Gr9V$Pu;^n7-V#$}$+^O+Y^d*KMY;$>PUO#(cG}~Y+Kq}okb5@@
z%GkzkbUF@K;*4-jMMP-1rluI23@b7%vfHclpH@yWHu0?8+X%;-$Smr6-7X*4s&dJ8
zUkPWy?CEhPRfj&#`_fz}cBcFfh*vAm==@G9(1Sg@f6+W*L=Bm-<X$bPxAvHIz939i
zR*jBUC0@}UsNXB@LAg<?v@3@R<Oc3^!drX9RnZ5<>uJp@@Yf#9@lB1Q(Zb-kO}r;{
zVh}&apuMF*IFyM>!o<k}$?b`qTLC1y;gN66C0>qEY<h=tCvjpU>e<^853qrcIoARY
z^e61?ZCdXjD09-eEs;`zd&-eh%4AnIR5*p3;vF^CRu-nRZeuJu7m)Von6^{Rg}t-M
zB#&T9<R*pfO;a;y(_dLTeq1hSZh{{T4H0}=-AUdf-<PPje0Yg0EED-5Je3?~h~u?v
zQ}^-N$ZSw5GB7=Hhjvll2HW#@acH`8SBRS@`zbR+D1XI<YM9~p0GVZX&xH+z$7xS7
zYx#!zte5m<67sSYgGAo}sCDz?BwmKi^%1nY&>l|(j?1u2i`L>L*S_0^yNARI2y7J(
z7NR&UHK8)-!Ci60j1)}rSSDXyabQ6V)Js{-qb%>jtl4P&Ce4^^Z|EE_^>gV)3TN9x
z)dd1akX#$fXMLTPL^hb84AsiC+)|%QIW~E=a9VCLRUU5bQW&%c75lUp=f_%i!xtx!
zQ7$aiQIH}0gwvkV0l`0GJ0@(rdY6`92r|@C?>p+PMd^)?@eVTBv-dX9p&CWqCY;;E
zmDQu1TsK*RN<@xYxZYqHQGvX}uG-r6USt4MAi{wNOQ$E7_3T~kozFaqw8venLGx(o
zM4lvSAr;wF4FY3NkjN|HT1Hdx)*?-^%Xbh<uztg|wNpc0H}Adn9q)dXO}B5aM3i*y
z5+gTkTx~Or1r3sYH9mnK)d8v3@4fb#YiwIiF3ulPv&bgkJ{jb?fgLhi+4kzF1kD%%
zs}~1aJ1tt5ab%r|X^wDtIHz%b=LU)|s+`lrkOND3R+W~~(ZUe8qh86PX)hsXrD^Mg
zx0Vb+jl=cWYa06_tzXnfRNX-bpe3X)=McT#okN=6dG~tfI=2bZ4VA#q{%7qyZG!Ia
z?HF_JYO||$m-wV?KJZa{Z~a#M3HvvnKJLJ3J9U*ec^X|6&UAoquCMCwU%H|2Tg#>+
z93?T$I_>eb#rp#*5SI6^POK!yfcVQH=Qcj1Qg<@Ji)9Ia6r)4bmU;J~?o4++q&m!D
z9dOz^<za-fO;L2{eGT)m+*0m_+vp6NF}|T(g4ihbXilMmz@6*FpoxKp?6I1Yhz#W6
zr|~)fi>@xF{pu>}0CwGT_Wm~CZ;^1uUCqEk-Tsj3?E@XB>B<Tfy@?Pth6j;zo=P@I
z-+B*Aw)>j;o%X$4f=8LEw7*vLGIHf!=cz<yf>oz+0NB^oP^7siwG&gm0~6_eEeGy=
zU0$vx1*9%fM(yj{3@WN&)Va;M89{pbV7pm1FzDRm+)6~IQ(aAN`-bkj%3cao^9GXx
zglfw4=!kt|dxP9~3<!rhy(Eg=X^9Ch!6-_KE%NT#H&wgr+-dCwr56rmbK2GGfz2=_
zdpx@{zY`phKFr2+V8Qc~;Vkm9(cP{to_(l|cj=0C&P|Q?bi~ed=aJbHZQ{oVROWuz
z0ccO^tY+e~IVXjHhz}=@ka9vuiq{>(rL>rRb4^$tlq(jt&w*(OIPB5Lzoj0Ka3Ks<
z*eo!Sgq06Vakp>nW<&;6qz2g=LQa<f{+!dEtc;D1j?Rn|B3dKLGS5P2MP{JFI4vow
zki1AF%D3kLRjZ^&8q1&qNRBc-Nu8wc=)Mga2ALp~UXB#VgAPofy?q<f6YduKNH?1z
zM@1XirN^9>ME_x2lcPG%>O1TAJjK|=rrjG%J6k4B&L+0+YQ0*9XWzL?!Wh10QI1RY
z-St&*CE0W2#FXvlfVUQ&lOG|B!l{~lPyHs7Zc;o5;^wq;q>@l5$fINuZSAz(d~fz(
zekU;)z5bvg^&OaIZq2~9WmHM}zBbR(Gt;Q-+-h^)vg9WF{-jUOd>f7+uJs%0WpK)m
z?QF!KtPQRMC({jA&Nz+Ac6D>EJh;FF2bL690cy+{=Ug)SGgxiYM7+C=M+QCCB+Ek_
zm7_P@Vy%6V8V4q+xCe7McgXhC?^A`TB&C>1{o;(%u5sKONd>mIt#f~7aIn|ItqLRw
z*xF4>2<qFuHi3jg_7u#8i2+y*LR+Ef)fa;Q$iKV4>x2jmf?^OSMV^Uz9(CHUqZ7#a
zPr~74(vxqwJ7l5gaiji+C{A^tXoY+Psqo5TfGIF&a8yA=SK_H64?<{7L7s<n#SS)!
zZiDP3edj(9Gy22#sGeena%xmQqBrE+bsHw7#}jrqpP1iy2PHMU?k0)%z@BoXeW6d8
zVvgnGUK1^8fg*cH+uP=Lvy>9Sj(UU1NU3@~#I4$s^_+N~Bm~t|1eWCg<u8A^9c%a@
zuOZW^U<eIDS@WVAw5!_$Y>g7@l#vI=+bWYoARS@If#kj<`AikNj!~laL<if&JM(P~
zwr_q<2^S~rWP4jbPP${NItcPt<N(>JdRInvXm=lWU^$HSOYgARdf))OTU-77+yMtB
z=?)je9{wG()7?<>5cjA;S>6`x3{HEjb`C$y#|qt?wKMstRjZe-f6%!}Hp5R<tuFI?
z<mhj~!%9CmTOU*Yzd~!H-Ugg4#58y|?KNZPl+!B?YUkQs8RR6stymQSG`m8WZqi0f
zqy?Wnjkqm&0V@ea{=7X^FO<DlNwg>C_Bt@pp5+fbVCUOxO%Dt?H;wIul|>#y<P;LJ
zFD%Rfp|awni4;%!LW?9I&9D{~yt9jq;N3qkP1PfAb7WyO5Qjf$m)g`LDYxTJlj;UN
zcG|_8DDzB|^pHK>Uco5kO(RX#I~YELF#p5sF1HC&VTNt^ga9Ou8=g`y&6OGrK^0hR
z<loSFd{|_imJ&uN`_4F&%9kc#(T&sg1J$l@3>n9i`mor0NPB*WN@$pEey|?TNZ&lG
z1g$`V9lYa+(;jnzQ-h8m7h_qQ%MFdRx&@rCyZN2VA6UNLo@qE1vCNtAfPD4e_&-m@
zGM1GamoKIoWry2*z8mX_eBj(kTcVuEO?I`LlBx}X-$a8v?5#y)P~0QRap;@&Lv1;*
z&nWX>Ivz0XzR7KrCbDM}&sK6`6_?3B;klv&^2)`OaxFvZ57+la#Fg=nxh`9rmi(m_
zNbz#9yOQ8jAKiYWEjGF<MTGkfK-hijlH2S@^Ku#ymdnKL5F%>L<I_J@!^pzWp~O0{
zpsG4r>%9GVo6z`p)-&Td#2s4Fp2T+7<cMoTjxE<{P&K%psHf7nB08d4Y_gs0gPBSs
zlGBQ(G<9$V<e;CdQD?|RXo}tIrFzj`bm9!=2sSDflb@<?>H{+ICEHT1_ST}j#LMvU
z$WBBqKVU!o9E?hxOvZ`G{=7<wa3yOxPsMcXWDO44&oocchM6gjg;%B)!4N=eg%67;
z!Ez;WXrn@WiEaJarmk~lm(DBUG*m!Zsh3~-TKl>D0`k-7#H4elhADKkpU=yY=>g|5
z#+-%`v9lBQ3yprQT)V-!Df*=_+Z6c}`+LHEv1?JJZ~8LR$VE$ZI|ZGr43wCFl7f5t
zmm0stX*D`9?%c;W7Ue{WF<yPle!04b-!ha2onr$mw<3C+jCwjMR?Kvd{YrgTetW6J
znF=Mf=n&L(x6`gEpWnG`a<p&1+P(^_kWz42=C5GHv}FIoXL<mdD1YH)rCC7N_S>&D
ziUjkR9B^*ao6u(&wnPN@uQ%nI4MeeKm^pN(15RW-ZJ0NHBiCXjai3g@c8m>Rk^IdX
zxbRNFLL{*}u;^B3gqk3kBiJNWuEl<<TaYkOL=>!^I_#}Q(zQ$&T}3MU?S}K3Al}B|
zR}L&DSeUErbTTqI<YxJudRo>djI?~=gg2N>8DfywA?$bC1PxD;mf+mxyx#|s+V8dP
z@jsaQVx@<tVz8mEWXIC5L5E2Dwcqc$B84M0e_*Nl3?j#|GGq1!waKW}Hh0>&D|Y#$
zxDzDyqWxh``smgkg<rFMe&<AQ2A}J|^vDkTqtg`=s<`Btj0cTeS{g!NNIn#PP1ZzZ
z5#>x)ctRR|<TU%^1$v5aadf}|L{9?C`57KO`;%_kC3}Er;S-?Xkkb-#5zgFLkmnXH
z`O`MP%P61i+~x8?jO5pm7!cpFKWpO;?ks-^1!CJ~zj?`P?9baz5o~mf$U<3tTB5^j
z^{^bHrP*J!Ri<dv<W}dlX7_|pQmn=PvR)`XT|1J4S89*ZlE^HHLqZwkF_pixcKubZ
z%R{~1R_8iK1Un3Qi?qGs0)L&mXW6}XcwOGWF#e{EY4c>yxzVnoZt8>fxAmeFsbnjG
za!UyDu7&-i%pOsDS@d_=-?f+BeEV(AO^z}AZ5ey?_XX2AY!X5w$^(nh<?JWb)Jvt{
zY9swabz~vkU5RGooj5J=1yL_CbusdP?8d(kJ{j!GX$fA*yTh7zk#2(}p?)gADD9O0
zRPU7Vf<Xz^G3&szC$aM)xY$3powKpWu^!U0fE1M_51f8TDpM5qar>7-{Qq(C4Y(G{
zE*K{b4?mpqteDQq2iDoYwyDDgBRL2gr31@_je-=4!%>JTu^#qswQ&!L@H=->F9=17
z!%pqrTX%7X_c=;^$UDeZm3)(X??1YC%dr_YPmvJ}eY7Mw8ks@)`_IG=vaN=k>sYpw
zrS?bd{{w`)YB(I^YX4PZ5M`8;>H_Sj!9*uTTa^uNXW|PwMHH6?cikvu!Ip(77xtsh
zv`rA<XiS{!cwO$P)1HeEL7q)wk&E`D&oneqjzAoD;=oc+UI&pGv5%=A$Ap6BN_tQ1
zq$P&gv}_%49xa8PAA6=@l}LJ<I)ye!FjjPuigWN(8y34VXqX0vukYh>`13ngkB)6|
z?s6|kn3aE!cdC%3%sjg=wS0W*_L|<qcxmhQ+R6?5gfoFbcdT6RTqb!ZN;b9QPdrn4
zDT!##P4p*qi^Z`WpVYNP{Gyng#sI6MC3ahi{b~LG<TK?#T7m5m$0PSSPN=5jN||+H
zzuBjBEnV4y+;L7zitPO@JM2^IBW%pTM525gSgzH*I;7A(ty>X=@1@T<j6R`cTz1-p
zvdrFD7XQ<0eKZEu-&P1-7u6ZDAfiD$6=Re5-%7oIMz<DOEKU)pAt|$R2n|kqPVvYN
z;%^Z_?K97WQHk0a{r(zUp$?#yT?dsVKah%i-kz5qoZq=pXC&j*Ve)|0;%m~C+L?|%
z5yC_BhwZa+U_-1YhF|WY%Sq1mV>R1j&#&(#^Gn$$(U_K8Jfv17Mv}K~pWVU2rGlh$
z!GUS=Mjt`o>e&mr_qCD&Wt4JS;(dHWN;-DmC+%|zb^wHJ)?HvR1I=D2oj?{(!gjl5
z)Pa3k7*RTC9d8pOr(LT~F{3R#>~o*jbtQ}S#>pQMjo(@nC(t5hg?W^4&GneydDq%|
zR;*mV!2#%<_w`0bb$s;;+uqx%lajl;5qx&`q8d!b6U4fE2NU69uY+~=1#Qr1QW@&^
zqyvjV^Q+he!62~!owF}&gGSy+BaPxEWCE3U=aSQ2`rOIj-b%JpPLje86!0c&KWksq
z^^H48+c@cg6`hi)Rm-A)KxMPUF0?PMcHR)2SLb*vI<DPmuc-Ga$}SP;Gi+Z{KMBnm
z?3IkBkP`(^Y17cI*}NOhpi^RB+&v!gM>69wwgU<hfInYy+S7?fq9DRQPBdZ#UecCe
z*@`uHcz4GEV&nGGGa-$DiNjvk1APS5YjRm<rbuRwilqP+AvYGrxxMzX1sKv4n*>W_
zgGsKyBq6G1U)r8SXOT_k>8p=9u-di~=UdgDqu7*QUQY{MWi5riIk0#-4rK&+<y?&T
zF?&VLh>UHre5x2+3$j3YblO8K7XOcWs3^RP^+4IaWbnQ;ciFBz6OlBpz23QM8$SfE
zi#QvuYty)T5+@^f@kE*>_tT@p_GL+BN4Jgk$DJGPa)MZfA*ioxKfXr_CU-d;HkS_e
z(_wsRgnLzct%6{Ii3l#hf}=+q>Gk!74BwIDST;zfB|FBuW%xRU%=qd?-N$E03U_W3
z4^>5X90&N2-H`noC(#i<I(LESOcXYp1E5(<DilXNW0&37Ua(FDjzwA>Pz*P-Wp3(L
zSMhvF(Idv#X{kN5Q1CA;)Zh;q_G?=ArOX9RUG@$prw>}!EEe6H+jDS1g_SC}5m-%Y
zDvzme6Dei4)YFH+5?KvS;FN4IMF^n_s4$P+TF=RhT_^E4u;{UNwZiMf-Gu83pHzv)
z+q$m`15<C$g`-8A7dOYS>fqd-jC`C>sLk9v_X^zFvelN<E1J-wWJu<I?vT^62oYsv
z-e<3EQpN;99NcK$IY%3v-d@+{oe_$V>>;FC&e==eTGHpe$j@xAuTLPSAWO-JxW^kz
zv3(&+p)|gsExNSXpezoka1Em~(|U?%`ZpE~J%+6&|JykSCh=qWJ@>9yX<vTk*&+4#
zzooIx@BDxNzw@o=AY_%bQ9R;7`-)tKXC5bl?s2LRbFXLSch=GN6jZULZLFIxpKwAp
zE@nVUu+ruhe<MuWmURfA2^4C`a<LaZftDmbAZ{TuEeNaZ`D0;53=|a^x8=o_U??bB
ziD0sfNY*FLJZQlQ=c()k143p2?WLiJ*RSY?8nI=l>n9ypJe{bUw&`GHwz7LqC`Y^v
z7NYcUrzNMXJo}vHXm@ll#dg5fu>Jr`#2XmgF=ltx`_cg~qj`{S%%`+8S|U*}k1K>_
z5pf!Yc<yTN4rV6L;~6j5{LVok8i$;Z_ia_;&_;^74mvkY&jiU1lFulCOWa#KPil{p
zlDo5u;PhkU-*6|zduD91Z*OW_=pOR!xGAy`BF928M!Tom`isH`7tz|*jyA?NX}ecS
zaYUEf>NYEiB9F1B7yqc!60;nu!2pgU=B(-ZaRsU?jZ`KGdE9A<Ib7`7mq;VgTte8|
zHiuAq9fPqO1`8~H(^LrsWh>-U<PQwAb!RG9U>aGZ&ooooddz|5W+3E=EIRVkrtQs5
z0a+%%lyfh#hB`I4d~%1a?*_@qWDGpiMJy|qotES?TRFXx0@)ZjH{>Eba|FGYYhjQS
zD}tOh));dC$O>}U7r{hJq+Zy)^_)y>WNN5Zf;X6ELxl$AxVx{qrJ5?LnakO#j9+1Z
zttHh>IYwJyjr+S5I2xL-@+%1CA)bJi+7+{bURUVJ1Fa|YL8)sBg#=bZ#)1d!fR;R1
zFGWt1wF54qsA36sw1_NHqZIqVTN+BPTeZ%)iO;227}}}5wQaaHc+Q=>B1h5!i*`p`
z*4rA^6aTx<xr^sC*d#$hDtMT^z3Enjdk&7O${>&$WT^itP-E}tV5;RVjdZs+SVbyg
zF*{-J?4DDuFXeujv<6d24EgINI(%g}vXep7<hlSRL{KeK=4+rKZe%1jM*xQ;(r@bY
zdu!JQX?90Bl@)yr*|AM$p3R!=S*SzR1_!q8aKJgG^JfB=IG5c9+vJo8AgYu_+h~cW
zpMY&-zS>0_YEvH6s~yvd<<V)8S!wOEyRBEf4+(%CEgbhK#6Ie@7qLlKo6bH=*<*9Q
z<Rx#v*0~JHV3`f))w?!1SGnLyo)j%n_&m&dzm0SgmMM;L$SXV}(ik9owAaYv)N&nE
zY@=;Ya~8Q$;JJ@y4n#VNP<6LGwCHzI=;7|(!9XV+hBgbw-_qX2DU%ovoO{uS0cyvv
z18l76B-@djdsC5<<N$^G%kEjV;Vv6*q=>0L$-m3|yXZhOe1zjG3B&5!ME+*O^uXrL
z&P{GK)R-R8nkYw;-3&=zmYqzZr_P{VXvy@F?s!z&+Qyj985qTR2exQyQItXvGf_C^
zUEM5+Hb+po1Jg(-(l9b^Q{7T1n>KMblW?^EE<5eXG;h3p$r9_gF;K=~YI;*|VA=tN
z#1Q#U%<nE_g)m+7G98Z8+>5G>Jzf12_fV!Fk#@1k_Iqo|-$WV)l=3ugika*Hj_V<U
z47iI`PnsL_(6+TMqWO(EcSSxT?+x6M>*);f#M`T-e#_QVEf<{WPD{L1N_8`gKinjt
z;PSGaw9ss+_-t*7oosL<8Ryx=SHfL5VKAhl9R(w#8Wf3w3l1!|lE_|dwIHnaNG(J2
zf~3#tDcN}$!EUEr_S<_YQN<ySJ$rXUK+@l+!8%Dy9J~-{^kWZ##Voc{$cw$Fn~%##
zLW%&P*ZRKG68u(@pSHA0mgd|jq2s8%H#wg?InK3q$mSXBq9})7l_)eaxt-n9ZdBOA
zzB+$IvA+rDx~4@R5Q98#-e@>(+Uzutd<r6^62;IVdtbrAwLqs&Ky!hWNP&U~Oo6&W
zAu%${5B6xar)fC?${1+y03o8&9+RK!bun5#R!dz*>xkpfX(%!!UlyW<gpl3^p3VKh
zxe3Ge{+yPL@FSIb=C)>3_0%N8N*3<20_+1dEa9UtpU`{<R`?o(l>yZuE%r6}sby45
zq5>9o`EIOqFb-IylDbr*EG4S1ZNtZPE5@^sCzbJ+w8u0*PV3`f<cCPF|GFgL%h_05
zO4cM(g$UfXB9Su&N+8%Y*nx~T;P0$`usw(DDq)dx3LKd1pQLk2$;N4G-_WQc$(n=m
zGyuq55Y7}~Tx0Ox*dD!;ctz>x0BUm1@{odU^WcRm_Dx;OE|eDu_{}y?OJN-Wxtb_?
z`gp-)T)>E`zYa`WYJTU|fvprzun)E0AzsM`geS;>h14F3Ms6G4s1%XdLSP}8SN25p
zrCC-wE1lDlI1A@jw(&w{d~W*>w|;?b!n1>L5n%Nj#Q|%2M3<v{H)Qj)eRD$uq!X!9
z3MpUoW~aUOlf+Hxm0<V!mNv2IhLfBn;{f6zEvGW+F5*)A)_QTCu1j2H$n;&HxYM5O
zjjDB2Hy^TZtCu{1oTl`6L=Oj+3M7n4-u`2#h8b9W5Cf5je&%IM3HbJudzKDvp*Gi8
zZ)jMV4SffkjB=_+AP3g{A^VQ%8mKM|h}lSAXXN&HYY*!k!#SKTNW`X`=Z|zNq~k}B
z^kpOpiI$d_2&N!{oruP??`#9zI;u3`zApeI4M2Xst37ewsN${r4&W}3iQ&nX&7$#X
zS+|vrro7j%jleT13Qx%nkV6@8Gae?iW(c524hcE--AP~v;Il(sH=7A-n6mFFedGf;
zkhreSE74QG+-KigzZ{$)ZlK+ejyqxBS5M0Vmr)7l99Zo|AzNr0fpWX;`#UsTNJ?uF
z^2+~e?UL)LVgN-$i49(vQ@(S7P|gfHciGe6kQ_;S^-<f^eLH4o{veAI#&M@5wty@P
zXsK1V{4|`LZHeBwyWS=vnzj{&IR}>8S*59{0=lQ^eCsI**mG{{mAiONBDV;qvPx8)
zVsDbd6`WqnW$q0j2(wDqqU|eKZgB!)m;qp+4a$Ia-uAcgj`!dMbq5YWZ0taNd@+Lo
z3E9{pX}i3&YqOt;`B$<q;um8_3$%mXOv#5LcBhDH__endX_Yd2MFzO~ZHKyV<`I?h
znd+UD+Uc~XP7GP@_w4YQXI1l=P^TIER@!xlzW6_Y$7Eo4f@K^CwsEA%#{-<o&RGoF
zwhl1gcONHtPutP9LpD-NVR&TP0Z8iHPN}d7d$K*edEgOdi`&4$Ot6M%K)B>?Jz&RL
zOUfwpUpq&K1k#e!sNcidLaOHR8t@{{&UOr|A}G4wyLJ_0<|W;^?L{L_wBNHyWy};`
zIdGLAu|z<pJvtep9Mc~M<EUG9vgSyh*))V*abQUYp?4ALuB4MJ1`X26f0NZ~h{Ui{
z?ai^3WHeB`fCN)oV)kVbMQ3J{?XlU;?50C|+Bv3dI!3El;fN}5o~jME2f7y9%1#%;
zDq!xIa~Drww&w};=Y*Xpm`10dC`Be0j1xpgIWBP1C^6)AHlIVv!(gx}g&$yA;5%Ub
zBX%yS6eq6}7+jwPW3N7DPqh`T7%Fu+gWhP#4rxELK!ktW`Sx^(A}!cnF`cxeAVviG
zcDvAZT#GBVv&(@chESJpP~~MV*33zJRn)OCK)xusGV<6m<=^v^B%%A4u3ur7+Ao5v
z2VtXQ3xr#<cFfyT+R}1Vs*rFKs->Xw52E#S&JV_`!D;6vkck}+vMq{wvdarCTGE9`
zGp8j-yjrf$D{TvHLwM;qx7m2t47ol(&`pO_>r^R$X<X=Wx7!akTo2o0a#E>V+FYD=
z;<V@6HjoBwbM<t}JTiES=q`{F+H?EHETyk2nZPsEqCF$!Lte;e@z$bm%lb(~TJ6L6
zuJ4J15I_eebH)94t+uP(#*$LeC}zt9K;8}>BpcX4=h-CsY)lq>e(zO6FG5(2LD&z~
zr{pJBFvaEh(+*6ZqI8Kp+x0|@S-#j^-e8p}MeO7$`{6d5yat^lB%fsaamu69UgRwz
z?Fv5#n6)2i9Y9Hx3H}0V8L;SdQix_vlG{Iww)Ug-8x>Vk_`qfCmcWwuZk^mZv(<jA
zdj(EH=18*>oJ6R=AZV|hrivrMmF&m6u0%vL2Dd8?EIJ+znOJf-7Ch#h{Y0%oY4VUs
z#p{KwHs(Kb;UWvkc{aYLSe`RzrG&58Pj*uhagY)Al)Rr7JxwXMq0y~bE#7`A*W{L4
zmpGTvrM4Hl8XES~iQki)SL|HZjLWc+*7m@1`<dK@#5W-YbCZw3CNkhzUTmVk?PqfX
zj1G?Iv>WbX1eiimJ+-w@+s}2sO|xK^mf*AqgE*+le!l$>{MMe)2^9?{;YHDfN5F$$
zsOR{&L^CMUJ_nYOBz4Bql>K6Za46z9sc@9lZIDf5GmxS*WWSUjUvkq;&ZTT}53I6Z
z&e7nOz-I1sH;U_&!sfrybup`S<9PB0YZ+8n6YN*pUSi|xSZQI&qKaEiyGVvgfgq*+
zS|cFSJGPEEcZEk?u9tQrR=r=ZuPcr~io6Vtz!I5>qZ^;J-#C-`QZl-q4b%Jo&tK;k
zIdjCi5rsg|!f$r-%Za5Tsu<OQMW@2cn{<x>;gPZ5s-Y^_hBLl(@+U2%269b;h|``O
z+&1}eZ_IwXNuE*Nd)m2=7i%;M2$4rx*Sqa^+9$&$yU0%Gw1|1i5kGCeTfLRtAsOKz
zZ!m2Vj`Bg5I%L1s`YD$=BS<A6qg5~BFQ&ajBrN;rSN8k$Vo?E=gr|mq0}FB^bYu@p
z#3&jXHZS{w`i*vX=8Ye8U^R7$_0)clJDS68><_!AkD%SeS*8vwzL3IHaWpJ#yM;gO
zwLfaZh+=V#$;dS(?dYJ>Uam9MFU)pvc=@a{)>KaIKQ1(hkW%`jbH0*D<T|kZ(aFGW
zn0^Ri&My*8*f&vQ(*7iO%!{vguEZ?aWPSV7Zq{TNNJl|~<)nvC$NsFFZdG;f4LN`q
zxFm+OB?<QyDh2&{_hJn!2#*7csZIWp-jw8}e^F113&`k$Z{LCC5R``3ClJE^vTZy{
zOHDd=899WSp&pzHTY^r9(FVrxq-v!0SIw9h*=MQEUvi+)lMK(6M5qbZ8H<124RpH4
z@f|}BpkrAQrg(_`t>EcYuKuQ8CFGUe^I!-_I!%BU&GR!yG5Y;&ZY$iKGksDwxDBGK
zFDvC9L#q8<Qn!JvQ_f{HCvlHRdi(o=C9XyQqc6aYmM(e0{-K+eNVQ(%3tcR4@su>m
zO;nzUQ<DDi%(KteJ%iGzb&GN>oTiLo;^@U*zt_{DRrXIgw)vfJe}!|M%bGf6i1?9=
zWB=S1jOw6UoVy%!XmVL%?OzI(RaBL|vfF{h`!ZWN2%1o0ZlC>YPSN__yPV6E;#Om*
zw|{H6g2IOo$^pedV%bt8!ekj)dJ7WkzvqWl&3TC8@ziAOIj}U&5Hw6yv54z`<Ty7E
z4E4OO$CS|apWUj94u*}1Nz*B>w8SVya|nq=zdvgKRTOZTV=A4S8eUQ}xq|aBA3K{r
zm4Q<>M+y4CvQJBOzK{IE_EFtqXOafPP#a8%C}bx|f3}Z4TU!*nc%x+6uJW&v1m(2X
zdRz)y4vg78=4=@SnQz%ll&esi1d1wKzg(p}m75=V2#SjSqW2$*>!*0j$DVyIMkus5
z%85HI@lE+sWSe`EsyH8aHhB5=8Ru%zwiFqVjTtMFef-(jOYjgX^PY{RM}d9c+-3WO
zdSg3V8b;v_CO)ikQ@AGW6YII=DI}fHfyMWrqL7fqVK|?3wuq0=1L<SjffPb1yHAS1
zIj~ZTCmMXoKKVIwlKw8u3^W1L<|U_H2&)qt2x$D2v%wSB-n;Ccm0mZb9Il}9ZW|8<
zRi&<Ak8xnytzohg27FpQLjV;vllbev5>cZ3zG9Q@({m7$xU)tSy3uza65oa?jQ1I5
zi_vp9#muyGtL|-8efz$BW__S(Fr)-|gQY?j^1MASe}!db5@ndXm`>?gGXF?j4YQ4X
z*4g6Lo5mExIOG6w6Qo2f^@DjrQnlx|eqm)NBc0nq7*4knLQ=fsvvWF#5gHuilq=^x
z5g8#|MMtU2YA<LXKsr@is7e}Xp^iE&`9=jp(SjGmKBp~SIG03~pA+-$wC5KPF4(P-
zVeNAZhU^BC#a{;|sbqPcQh-$Qd1q6=6NX8$JPYR*N4*HIO{b*naGE6SVPO{G7dbO?
zC!(;qj!ja>_t@uGi<DWtOM%;9AkihXq|i3PsZV`-VQv;et!JETt*eT2Abq{4UJ>EY
zbU{)T+UeCty|w65*oF{cTDamD)Z83oUb9%16eWo$Neo*1B4Loug*N~h@3ehk%@gmG
z`2r#3Ne6Zfy-K~4x5-(mIMF(YhBb^jXLt_w+ZW~FMkjF2;mqMyT(yq@-M+Z?l%hRO
zICr(1r4!23Bo#!py1%5GQ<*o^g`Cq9PD>%fVY;9dIB72~m{v&jPk7!NOx#E&Rst98
zCEdhG8qmaLwysDw$v#fIrhb0sjkn)=lfASdExukjsB<4*mK1@0j>+z*UDJ9)VfQw@
z!1R_)qr;rGU@z+sH8dJU4`}*8N*N%YcG@*N%a?B0XkS{-Y7P{XqT+Rnyw~2^!9J9L
zX5UtxaO9HM%k#^Vs5$Id?vmz*;vipZm;=GYjJtb9n+>qD{LIYLo`?zpQ=NA8jPQ>k
zySCPY-e~ObtX6i5mY5ao7wPq|K)bG+sVnRf$*8*>SoEgtBIJ2ipsOuwU)Fwu5kr;O
zpMsSFi{C^<5?{qg5nDjYD@k?s%C?I*c@|d<x498P7h$|gbjDwMRbvjzIDu(Gl^B8K
z<OGuxf@xT)Kp%@^*LU4PFa(Bz;J~zLIS6t6$~7D9)eS#34UBJ^BrG_aH!%syM7*I*
zSqNEW8HIM8^VX8KJJDsK;c>*J8yf;r4P}dWcWX%EvYWcmB?>9YJf^{ENn%313o5w2
zrV*37mpa$6I#H|*oF%x=Zthls$}tQrGC9JQ5IK%cyYQDnUOi0bx6~Wtw^an(1&orw
z0>jBsK(e~E>zG_>+Ep_Co@C8iOImEwSu3&sgx%J<tg~T8oZD_w%ldn9Zm->5uhoz1
zs5@;5gn-tfRSV~`&R$nok1eU8Y14~Hr|%Sxu{w@=NxP{1QVP-IE5I`F+T`kY4-8Ij
za&C$orK(Kru-s(!y63cnD#{z#bWTe{e-yf?b6~IUCPd*f88#YR)fZYK{8W9)UfLUS
zO68Qp?8jZs2P{QV`Phc-jRnh!^kLC+U|NXzoey$oiG6wA;p<j38h7!II0uvswI5df
z71cAbqNu<L=TJz;E~h1QGZB?$HB5yy4QU~m+0uGS99xO8Pl6IH{a#p{{Dr5~D~?Z8
zN!?}H4`e0IKclX)qW^)3wCvc5uV;DNMiltlhIxzo>UUM5hrq8$Dd_EQf1PtF+r<3N
z5h5N|F02cjPEvqKoGJy-l8@cbiD!4To11UD$+<~81gE@=zKb$mcNWr!mx$L#q)ur6
zX(FS&3>O?uDV}^++h8wynRA^h1<96&<}uRvSG5;HSal|Vj8mZ@t8ehq*4}(W$`%s&
zjL=e?5_fl4tRx1y1%Xd%H37?<ghG4ZDaQq3KRQ6kBGpX9b6{`Ed3rTE9+#ScfvJZ%
zPxGFJUEtGUue<r3D`s#_u5LIaUx{-U<4H0taevm<RDXnX!kVZN%nnl+Hk8q6Px>S0
zN8i@ADGADwa|o<>kwOnbH-tuX=lt2aZp$d{=fJ#5zou=2mOv&_3g-6No4a@GNQ?tm
zfzCRxdbf=CveLxCE^NmX`_?Z)$0N>Nq8Hp(9&f6b*@m2}dpE9HyT-Yx*Kv+OsEAMj
zSP!a{Ja@LSUZCt-1uf4~7YJCbNS-RAIxkxHLw4_?eI<&dz^!yF=DNl)0M1jA@^7Hh
z)qQ!n<mOwP%jjx((j;No{cW{Y$au<KVc%J?bu4e^?SbkGbdz34mC9I^gWg)g>`n{k
zTrdG7*4cyIq=+RXtpb*-x+PBBc}w=|o_ij2E@OriyOg_53Tk&yy1liVp*_&Qh$VMm
z@f(yruvHof>RV(-AeoXi`nL9)xy$qbMg*M)*3sk46jPF}B>A_uSr{JOH0|8Qa~UwO
zm|nf3S~8ln`PdMyytN=#Yv1a;rC8sS_RiK5$bl*>Ly^6~6lNEZF&K!xvYsQ_2LqBk
zxVwD;rs6AMWUztu9QY#jkZ(D#A}LfLP<HdCnjSUF(x^@`rA(hvFLpWYxzI>)Oag@N
zf(>?v+8rp((~4fu6dd-}4%)&*(#jgdk>8V=XhTVOS5S+MlEXdcKAFizBC4&IU!33B
zij7fQ6j&@Gwyf9=7{TbPlJz2S%r-YMYaO&T($0b8W?Be!qIWu$IsV#kV%6<SZu7ca
z&!L+mjU{Z@fn+u2+>2e@aKr64*l0t`<^PMh_m1<kI_~`^&FPMvIL&E}jcveSgXykO
z17swSB>^tDe6{=4YSC(Uz29BL;^2&poy1OJo5XQkQiOU#5(xE{td@}8d+)vXp8NjH
z%z1Y4y}#dm-TwkFmhU-pp65Ab&YbB7T_29OjkMo7I%)T|5!_1Rey3F1N8D=YMs0JO
z$M9N*OF`?`)x}{;eEQru8*3!~wzbahc#jlq@@L0Y+O~A>#&;ySW~VODaL<#SnI87R
zi8>w6PL4|N?G$RyrecRUKbvSBzx2&qe@98wVBN~4HrWtma^`^tNSx?YNU<;x3FXlk
zyyJN3`ew#;WdWJS`om6T^hDUC9xFN}^b$7J^*f_vbg>F`+4^acedL^Nkf!<%<0$%6
zA7nR3sUi1_5u0u+4^xH82V)R5se}v{4lD_1=xnBao)=<a<nJ!0E<WG5H93uhG->x0
z^61nQ9g(xBS}9tSctHmqGpVyvS?h??vaR(tIX8)J>pBWpT*4(9MBemqKuCfOX7v7g
zPkN1=y!hd{vre7nxi@^De4~eLw!Jr^yhMexHKAP&M98^v)r#fzK(5?0*B|mz)$`k?
zr|q2$pT*YWVL~}Elh5PRgP^k8zOl6-I1E*$Q_kJUh1gT};G&g~x;$bNb&<RS(K(OT
z3AXWf<wyJa2)}D~6b`Gi;f#G#)rEK!r}s&3b=)L^wJ&*ZubJ1jmMfQHJQs<Z4m{qi
zb&&EB8ffos>MY_0`Z)2RG^Z+-R<ftOlNGHQETX-qAmX_&B}3^RWmKsDy+Pt?GoI1G
z7Ac_Yp*BOylpAT-DXSB>Oin2QkMc>{dmA3EnHnW#hWzK$bQGxEHDf?JN>GW|`)VjM
zG&~GFdl*z{b;$tl6Drv^w;_i(C(&(>Qy1fFt~8SBb9~=Y5RF-syt{3}fe5ikl)MS?
zYGB{mz4w6Z6w)C&bwTg0zWOSAf43B{6Y(bTs0xJ%DcpfaAFu|;5vJ7op$D-KblWGr
zI;7oB87$d!;MtS)a-p2EZ)*$4MJz+QD+TA+#OGVqueWbcJe$I#Fzgi6K_Yav*moqN
z2JicrQ%cn4&_VYkHKBd5tCIj9gxIM|en31`oV*zUUGj*1sD`CQlFK{+UhI95D<GHi
z?Zx<*lP=m2twcdW)TKtAogZ#fl*vfhbSe8BaUjuKd1{Im5(Rm<Aq*USvr@&W%{x&g
zd4q?Vo1RHEsNbSMve$lR`<lY@VkxABfSfvzobdh>IU$GayXs54J$9(f`P?i9WsExv
zyuo9}c!#ojiiZ)J$V>1ewK}(~cYdbnZ5Zl{s~qmIIW#;MP#2T=TZuHO)4_7%wb$E6
za|M?wdnUi1LG|4$Z(n}L-S#~R85-rl=XXb&8AaWfd;}%J$GSC5#}`GJQ<ta+x@(5T
z1&9BUeY_j_{g`cropSP)w{Vfez&_DEn7QrKLc?zdB1G$Lp6b~ryO${FV9Y583W?Ef
zJ^-k{6^#ZZ*<s(?78lk`PCZxwoH`8^goA+F_dRWRnp(L9!NNO`6o=~sO}DIbe1~Dl
zVJ^h!@2}^ZBI0x~_lh|Ok^@`<EgZHV$YXq*7{O_1)sRz>R0*Q16W|<5CLyb6o!siS
zwda>Y);2lCIE@G~oE7FgQjcp#NZGOzSu8O=xozM{8aMXFhDOO`K!*11t!@&@0k2RU
z($r~s*G{kqI}$l`buT|v{Qz1u+u0bbiLp_q5c@@^u<!w&ZrfFLngk`nIfF#l>cV8s
zz|i8`?&|a6fWj2HLmlZ2-rg-AxJ51<wx_<MV?z6qX>#fc1tOyavmw9xBkbk83wLiV
zP^chz=6EQW(aJcGL<AJK7i?d>CvA^;(y1Tva^jHi1$Z)43T%|LA(vRz!1lN0+S1c?
zyA-2aZAEp>xQ<fY>V~tU9jI<eT_AOS*`!cg013RylSgDn)Lh+`ASady?6QLm?__@o
zLRcyr2a@DBh@8A2cH5(^b9j-Vd!d+Ahz!DH(Xc~pURgh(pR!9(7du%JucBZMmiqmm
z9d0v)?g4KU=2}}z1B|Fh2d=j-<yusEW7?7K?K?xQyPwTLo#uYk1W`hLdo1zpw%)As
zGZqE54AJ@VHp+<DYa;mB`I8PLsU7|_y8ClrN9!}8!DRo-u1j5#BlueRgDG|({59-}
zg0RH0mMS?3Ly?^nNF4|g;8=Y2;<2g|1A36v*v`ADi**u`pp8ZzBYKN<lzXM5B9H>K
zINoNSjB04MC~A1vsuizrqB@beQ-o*IKwrQN0Yqea33rv9tZzeg5SUYmNnPSdc%q|h
z4lzGfiCw}lJC%F*!3W=Ysq?!W0(yf@?EY{FINd~xJK{>c5vM}Cehr!K?M(hA(d3Zx
zyW8|Snhv65INRnn941rb7Ms1_fu!Q?cG9DqD{M)!%<c-FyGvK-4X$w$lf};FELEr=
zGeLTOJKrtG@i<?V=)Kh?Z>ouT%r0~rPTnNic4_JrKPwGYgEx&^sHnkrNeJMv{a~()
zXe}}prQAd$36pgaVcLGE>h?*)C%i|GM_nLT_Biz|!K2t*PQqPFR;^r)VW?39;Y!-x
z6Z518J7*U=xKyY*3*=yO7R(ezop9h<m(@y}KVRL*gu24M!`|wW*y=EN!k(<IiE)x;
zP5wt>UyeAC8eAAVU_Ex(4_6OLl^aw?ItKLy_izWEy8@CYwas>>?5X;~K1hVHp3oO2
z9#y6TN#(}v{jIu$+<v5?-&(v;$mAHB0V-?qQi-p#GkND9Er{MGc2R1oa9Z*P38Kt+
z_OM6unIG#$aE4<eE}h~Txsa&0dmOmtaQTXL%k0N<d#xIs^!hnG+_Rs^7W3Qgd7<;W
z!}z*iz<x4OL{8=yxhdr=f(}xF5gBYhRdp?|>nYZbx;p2>w<l1$#R%*AoS*KV*&k}J
zb90Y7b?WO9sI&b{^5l_Y0B@cx{b8r77BHg;6<+bPtqygjvblgfPMsVsFMGTFTuo>@
zRQ76r;ihoTf#hc0h>yb0=eps#)!%}>bJ(e<WzP`*^K%F67pg81SXQmWPF*ae>Li@9
zWGn%3Y4(dr013fZ-@kv_Db{Q(WZYN$s*)PX0r{8OFhsvO*9d)Z>f#wG3lvk5Tes6|
z*e@@@!Ph)-lamuAe$+cSR6gRFIfWPP2m6(JUeCz)UCQ|6$HZbBcHo+#&Ed%Nt99_a
zf4=^8YKl=*$V&`^&9D%Jl<4_u?I0j#fh%t*rsv0Se{yVMRtl6~Z<}Tv(PuF5PDP$x
zamR&vCzcuk$bKX7_r_~(aDEmpRG=58wcqUOlub?#o^$F#NkF{rg#A|6Kq)C<b}%VW
zSF;=>IyFHuw($b2!}i++m)1_kN~s0XBoVTv1J4r%Bfp~3<l1Y$lNfm0nDa~DUw!-9
zCF|^W6Tk6W892XnB+hT!s8i?ns%NY++Iu-29&+mR=5V#Wa2XEUx{3Ms+YlgWA`Uy|
zj(b<fh(+?GC;o%3U0kLDZ}T2+b!m~&{G72r?Apcc+##(|Cl+3L@03>Ik7@>VvX}c8
zn<!iu!Qdm_;EQP}M+fXtUgZ489iAb&QAc0{&npfJJV(Xc;LY$7fe0`K3NBVy&Yu)M
zB(pEZSX6f(L5@I@-eO9fCJDU#X(Gt$dLzzH5rKD%+n;sMhc*<AOr&K)0S6?INN0s;
zxHj&eS8MiJ5R^2einRxlt=al%=iOs}(a6XfR<3jkMGdr-()tw~wf$wz=#uG`*k12<
ziiNy~U=2}=lWOv>>R~vY;A#{anY15BN>r=+Q$w@%*Y)sj&==-67(VKXytK#{8LT@O
zTl<^F`9e-N^zL=ac?5;0K<XI6Dr0|}D2n!P;QWqm>dcn{21d*Nu6-Ts5mU~l@m8m$
zQ4*O8bNlV@+ZW<66^9_o3*mPlB_Iyz`n!K<`=@_@+?1uu&Ya3>vufonx7$C~@JbmG
zdsL^J{NOl$9C!>z_#%U#YzRTm?4OQTN{`<7IPr)>akK2esg4I2Gn{|KZq#x5&o#c-
zZ5MHd<5;IId$WaC4BNlt&Lt1w_|U9Vib*QV9>d*d|JwHUsyoPTPpKw5?;!3%XYJqW
zSxqDEM9Mzn)YVu%#q-Mky}c*At<xAD1~mr~%b&guAp<~=bx4Z$KVtvU!57T3DHw(k
z%sY5#__jyxXK5tTx!^zBKI3x62r{HXvd&b0WE>^&to>JRgS8XN@<TzCZ3Im{<o&-B
zv;u);i&Ln1mn)ezw1VuVL$wn*jdVV|xsK26b?WpY$f`1wecFj&4oELF<ovEl(uo`Q
zvhg-4?mzuRt<_S#zS${8!?e68QM80MJWA~|>VraB+{dt1>~QLubk&=ZF5f=$#8VlL
zIW#&oG_wUaKTPLjc%VL8qI)LgCsgvYPSiOd8<KjAam%SpiF0VGw-J@+Gja@Xedpy5
zI=>T!a5^>pfPHq3b_;xCe~VK}(st)G*Zp9u{+xXD;wQm(e&?g>A~4%Nx5g5rpFQyi
zRGGRMxlVjpd-<2CW?Jp@s<le;0@n^|b=qONsY_e^^G_sJg&@)K2W*MJ)F&Kx4Gu4R
z3^HHPEn(`sNzv@oWgGBh0tbRGY{ObP%p>e{DiU?s)QY0ijWdgmS9@m83dRIExKk2?
zI5ESK`9(EAF-Gzw$-8RkxG#cmn5LIe4*gl(L@60a&}OHuP~<$iq|k(lhRyG@XIJa=
zKvXWJJ-$$<6eZ{fFrt6)iR6Tu-cRq2DdcbGH{i5V-&ee0#WMSn6X{bZ`$C0Zun-HE
zEw%f<Frqnai@@k8La&U}_nZ^Oy{_S;qYw?$X?bxYO4^s^$8f9M<or%dkP?^A9M$=$
zoo-)NldmZhf&bJjbPXP?0#ht^qA)4(7MbL~yuK`ZtuD6}d2e-Mvz0kk84bRoYbTok
zt{p5Abw#t{Dqu~xzQ7RUJNA_e%aY!EKth4$4Wbk!;wSh*jpw=DJJa-kI4IZ-NIZun
z0yOZLDT2Kas&(1XPW!5cV3S-`>XaHXgq`86G$-N)?W-GMSvSdkbc(eq%Izphw#U9E
zS7nG)<fBd@8ibmR24_K(uWj#`nwZ##duZDX{rhp3hK77y^<R>*<mEkK8|u{F4m?=y
zTo0KTN(!Wp((#0ZZO`k*AnSN}cgz3Hf#gAkjcBmNzWzk#@T<>(t6G8_PZ#=4y>wjC
zaUOM^h(Reeg<^KGF52@KAyF+J-HF&+F#3oCk50!?#4}20m_l<NdtxgO+6x*%o!BtY
z4e-NGP4YAdLs9@8g)&IcM>nd(33f>hv&nXHABbQA2cmN$$lYmsVYLedBJUCVDfQz?
zZ;(2=`3_SE0X$<bYTsrYF(&k(LQXDs3uPJbT76wCuuBY~y|}BBVLZudtIJp+#jfqJ
zOWWSP{>H1ELg1Smmcya1v!h*BT}XWg2nIj6C`Y0L&uPsZM&ez;=ub^-a(Q)1dO9h}
zF#vh1OLk5Bd6&H;J2k@1a+{n&p{|4-9UVvQiq;(%^Ja|GyvfKewLkD8G9gCENWi(g
zWB!0$*^q7(QG~dk*2|?%QdS2AGQG6T0xu782<IZ5%?WRibR0@+X+HC^Zn{yi(^{t(
zb-D2JG`FkThu4e}9_SP$i9N_am)g~BUL;{KuP0E9Q<n%wpQP<O&x!=vv#+EodwJ5-
zC2Mb4=lsr{gusff_nI0&*pR^#WtL8B9}qRXb9|Iow1HjQ&O=-kdy@mFT<$Okh4OX1
zu92g4cP_itDJ>#Lsp5@AwCigCX}Op2_A#eU-?}11#e;T34tV2~93v>`^tq^bgB#m7
z2FQ2bm6nsu(QfL#m`e9Uu@sNb9dICe`$p0_%Wvlu)ll14@*LVU{>s1rB<3-xc?M3m
zzdChZ+2$S&UJ0<}v`AgfC%!FXJ$u!OQWL<$4J*m`z^T&O&TqT;G04TMPlTfQo+~bK
ze%2fv46mhwFevg#+GVdP2pAgJC@W*`Um!W{BQPQ4-LTg-$_lYlQt2n~8wuG0VzZx9
z22w>YX_ZLMNvOp8xHR)uMDQEBF)OkzSlUDY;219*cs!XG$ZKZO0qvkIIg!e$7dpSf
zMtRfqwltS+lsrx2PNBY<Icj7c0=b@iw=HXPGTg_DiBe63Bxrp@Q37DLEziY&@WEHT
z%=z6`N1BG<_;sGhd?fC0MXioBjiGm8+|;#a=r@F%C*=YwBawE-%_pAvM6AvHw*Q&`
zIz0@=apOk3UfZmZg5escB(g9;DAKsyS_2gq$oU5AjNjTq8b1e~Y(JTdrz4_yWvf{R
z4;-ogsi|d9zGp}>OqXcrL3>?(SeaK494SRREtG_kyp>7;WVdzG-319$UxF4p@Sx9<
z+okO%#@JTnyIDj@<dkB;l=Mz0nr>hjWJ#trTV3#U&yW@$WdoLVUG~XA;=psWF-6zg
z?QOS=D&OeHxKk0Y*I@}8&8fl;!D1x5#NWg2XrGUL1S<@taDd0baSW|G@Wg0vNE6Hn
z1e{LVnu3&U!Wn^2A2@cvfoRb4+qf~c-J0?G)|x30GcJ8<Zg3!Kt(-b>(TbFEYa3>x
z-IQ(d7#bkzvL_kVlLi><jXP~!*IKUZk@CT*i}OWJ3a4#-Vh-7CdwuT;x88B5-C56S
zxfFS>)ETWVHDfIL+<v>OeIu;kqBpy})n(=eDO7jc-POxHSQ4c)_s1PbpaOFo`~ewx
z><w+X6nZ=i?Wdv|$&4R3VQ;J@Y8gQ?5Jvvi>6hE!IfUSE8OVO5?LDxaPuQEfsn?`~
zLBV}N-W^Dad|X(r8+qQGlZG=KIqjU1pc|T%^qvz>Ju{XS%)(wjidQH-rZ$)x7~tSA
z(%f+X{!ENgWtg+KRI8^?i<WD#11YVv_3^G8v$xhvB-xYnb<U|10hr%*$xAM=x7Dzc
zSrvz*ix>{g9djVr?B*p)?Covg;?B0vV>oqrrYLFC)bv*Cb&U?k*{B;ls1uFfz!|rp
zTh#Nq)KFRw=25&95TV(OY!L$+s&>r>CDsT&__za!ZHOU(b(7RA@ovFDFkEo$Z1|B3
z)B>R@{i_2H<WI|9F+6O>{nqcMLHfiLH!@HRNN#=fH0c+vt*x}vu?n`a^_F8V)U05f
z3~EQbLDHO`a)zFH)JD3m9iVSW8^S2`ntL5cd@XRYuuOzTDhfnS+f?vHQhOYH;WN!Q
zffwfQhe=Vq$9WrV%M$VX5!&rt9c+wLyV4}x+hNy&ZAUT^ZU;fIa;<0;Uyv*$R~#H}
zY;#w)KiqGLFm>v3t|1AE*No6L?HU_vyv|@NZgG(@o~qnJ%lq7++qW%kQYFP{x*3d)
zje`zE$bI`Aq{kmm9%^;({?RRXFj368W542E#iOvvA>*K#hCQBW_<D1nOGTzeoeJUR
zRpghQOxVr3f(p`!^KD}PP=9oC+TPK1CtbEN8u5N<UV!8#x$fF)Y^s|EU2GVZOJAtd
zjTCz@UmUhUwNuWEaM+iZZGA$cY^j>#w@v4WSFR@*ncsDe?3JVsVz7AHLN-%_e|jLO
zlr146J8<!vC=*GKxxwzs3CB{WbZf&-x#&;mDP@kU8B9Stohi0<pO-y7oH+CI4kWNb
zDhfmi>FDkL_Rf%JMeJLuo8VUIm;%RWm3un;OT>uEu@dux4lMMfvvmY*W0ySclw{->
z+-ncy%uBkt$0@|;Ii2G~RPMTWwpj)pn$HJy!k+nUFM5G}V?#qMtx95xZjP#A%H_Fb
z57uyXHXmcS+_WBy%^?ROf<-8?bf)jh$sC`!_`dmbL!(YPNk)>U*gQJ_q*Ne*ihWb{
zOxhiVR*|aGTb-_0JrKEy_M5j6K#!%jL|x(3X#~qh2b=BPH8tWl&N!v){CZq%2KJun
zymqL3wH7i@0?EyGixPMZ?4g`VdD@OUg`7!voLv|9g-`o?vjGm^y<w*i4KihM+~HyG
zt8Zm|gbWpJ!hxjuv%8nkzPag*m1Jjl76(6^<fzQf<4rU=X5Z3b^G*@+|IHB_xH>Vv
z4f7KIHUq*u?OStZ1~Y?6=XVZq!^MZ@j!)V9a|AfKp`)e{BTzqea_T0X4`hS+ZP#A+
za;Fdt<o}`!3p>d#^KA`H7a{X5)Egp>WX^Tq@r;h6c6r~om(nxsGa<&ZiXtIW-Xcq&
zqig9`Xd63QxTg5?wCEsr*mtz4xfxr;z$ps#xao2W;S(QhzyaZ)1E*NMfF|%QtSt7S
zRu>0Ya#7SJ*;+k2un)J%8f+rjz`!X-SJKK*lpDtMv^`u1#~Fg_a+M_J<Ky(53#SJ|
z6X`;C^t8#6s7$NuAf!t-XJb>d@2V#k^X83T>73%6R<$Sx5XN+lK4u?jixNUzXq)28
zB6?U_9C%7E_o90zP;%D3JMr-jR0L#U5UF>nbPBK;?4vm?!??4Jcolk68%C$?dumSN
zWak@Q@(Tx2BT`1_)Kz?JVJ_%va9z+u#SV$SiKmlhomTOr%wTyR(*gVV!rKtbF)*XP
zqArnHNf$Qs>=T_XUAhXz?a2>B)_^=b6s8ku26U0MjkV;v!7(Y7X`GDq$@-GirFb}o
z$U$9_$IG5?-&=4UQu6!fBz5Y7uw|MKjaAW5Xjpij#NPIOjpE08EH;lXibHc+#arah
zr3B0iVS=Ihv+r-?MZg>aA1NotERzZju6V`bOZWqc`Eqoe=J2OjFFjvQ)2YT(yb`gV
zwyj&h!y&<i(k>iJfoRaT@?v`=fA@w*$DQA~(G<39+w1w9Xq~*u?|?ecf#fnq8jC;>
z+fl9A$N2h$5Ub<vNp~R8TJKTD7Hxo?$$TuqXLuF~V&GKdF%RJn$E?62_K)p4@sv^l
z5fV$juDIv6%fSvPw=EPz=Akp;<03DT-HixC{dxm*Sp%o)65d0zaDoYeRstzbA`mDE
zi(MKl+tX;wT5TAos1eL>8|uwaR>~yXn|tV9NH0ID8}LVk`@+aQwy#?v$^U#Atue9`
zc$p$Ba7{ShK0deI_E&4EMPoQ}ZmWJu762X-nKULWR(4?F3+x_f1V?VElcjd`25$nT
zm?-#a+KS6LX9wFm!DeOa#xm<&9eA3_BaYgmUEfmnNTopDV9uU!;JJQc%|Jj!ii(MS
z#10jt4uJ3#(5`*dfoSwn%V0J<Twma}weU-fj2j$B3J4^#rUPHZucb0QM0Q3kqgF($
z+mWuH3fB)a6r&0k4Af+29C-9ovlld*_s3e;V{H%KOn4n8ZpwL%4^7+$>&%|B$J_o`
zjef@|<&ZfIMFzznrXA^MJ8P!<o1H=-kMf3L72>+0Ct8C$*RAocu&O_6#~R0u)i=@(
zQW1-;8z_|}MA&`2zN{>dr{J7+Ab{!j83z)r(UqgUWQlSAIngcg0hSmM5}iABK|5q$
zYJvk=3oZAP#P`|B9I0ZDCa0ZJvMR*4b7q{%-cR5~H{ldwSu!hAinz&UEQq^D>~!uH
zqS~FGc@UH2%FeVmk6|pEb&73Dp|FX~L2~SDTaDXgbr05+su)A)PqeMVzS-r)v~%?`
zB^}V=Aj@XF(t#A^tk@wA5>Qqm4u`@{JKts}tOU88LZFJT?D7WJ+jhpvOIYPQ2;lFY
zKV}!&A-t-OJMt8>L`@7#syDIu<P70ml<WuFz$USslRlt7>{RJ^W~BB*wRZF#f+!LA
zfK!*NF#@Nz%jVjAt|u2&S6H%alU+>GJ~2Ex=KM}H3f~JkNgTKNR>xjQU*SDYT^<2%
zQe?-I9qOD0oswoz7n7$HF|rS8ziU4$Z3(DlKb-xU8oXm>!YR~Di}YpGn?@lLe;%#Z
zQ{7A-he-eut>n}sYtKFhUYlNFehndV)ifjN2=^n^O&m7E3zBojsf!s`XrPF@Ze7($
zb=-cmO$#m<{I1<4ya`jj{aDp$P<c2Ubm|gmuf6F8`|(z|?&UW)h1zjU4{M6Qul+<L
zC6hB#EUQzlDT~V+*p;$4D@lcvYWwUbtFM_VJ%T?prGzQ~Np^0-%;*@=rvv+`ZpLJ`
z6xIjR#JF>Lfaka&-qPL6%A5r5r`yuLeg>h2l31*ZAIGxxG3?eq(^xEsWsX`>$byw@
z(V^kA75{MW(3UOsvt57E2c@}c%Erf>J8+4w<M-On)vvA2uj*qYe9Gib97?<F=exPr
zK^6*gr%qp+8K1b%exXhMt?QRLh1yg)T_XmRp)$Gli(RKW-wa-+oDOl~JMef?4h%Z6
zv0($zy;r|fPo@kYghMumCy`Z2H=6(tgiOvMn3+IjIZZ4CNd9teQS_bnjrO-X#aMqS
z^DFf-vm^iQu<zTiG<?DLY7zoKxps~0p6U6n52~xAGqGRo<}cO#igwmUJmNsKsvDK~
zTKe!`%QZ*u3iI!j1QnA>ZX4WkJKUAGV_cu=)M>Wox81qPeq#~Kl)=e)lUe8ZDA}Gb
z@8{&9ZSDgwPS!DTpugFs89p)YW#C}XU_l_cpKiSNYWuB5(2&Nq4mrj2ww117Fba`7
zI&QyR^Iqz9CQq&q;lYr`JJx8Y(<-S~f40BXey5REeDa4jkHJ1VRT>@fD%1A64NOGn
zm?A=P*li23s)t%^ll@-0hF#2Cr;hS}UHEmxsZTxd(*7`x>%;c@ZJ)xd!?KSKJ2koI
zVTD?ZLVE3`zug}skspCP@Kcj9I>mu%f7s2V9OvYMe$1&$ap)E_2dvv4b?d)?-eP}_
zI*|D4Vi7}W)ml!eKdxy^WDR{?5%1KjWy&R56;+JbLPl`P{-n(o1|r<oBrZ}F&GHfx
ztPJ>5{b}ofoHpatVvQE=AO@}#nj)zb63w5rrG~HOPWug`N&-M-ycRDtX|d)GqCWof
zZZ?(YQ2GaKlnVOJ4et%khLwUN@{n<si?@jr`7dh5v;#?9p;KNP9_tX0oZF2vea-)0
zwsGR%2dA-t?MPJ~xErS?9_ZU&byzK5XG90~s6^eJ4rI}qD~weXI*Es_P5SFx=p_VM
z44guvULtj|ZRGW`*Z!s;>69mFH%(^;yg`^zkP4$yY+Gr(|F)h<hA!R{TkWhjNF<JH
zoH=aY3H!SfPdz)fdS<lUh)PgIW8eUyYw6ycS<%uV=kL4oIFzRBYsrL>Es&Gwv9zQ+
ztu&`8xRSZzAKJV?LUmf{mJCr1Ir}Ix9q>dX=!pdTQ@E7;V}}K2Img6T?{eyD%1&`w
zi{q9u3$G~qr~Kl@hY_zIM$u5^4dj|cupO~~ZsY^eUy3@krA(iCoV^L*&uNvouykTT
z9yR-yh6|)eADh_1U8YV|ogA`3E%svk*XnS}mtr1{ICa%p65S$M&i<`yEscPzZl|2O
z#4|#Oh6ncV)n{?>%K4m5Hwrx0x+p)ROmVcTIgvw3GGbCQ@3j9Yh&T;7D~dnOh$kS0
zr-7SrV!9VoDt^uve0BaaJ2$^=cm_u}u!2oMT|BdwxD!zm;XYrZAK`ar|CMt$HTpp6
zm?&$b-!;nBW8?PUiTSvA!lZVE=Rf~4vy+Y1Xp~a39P(DD36uZO;V0i?pLR0GnE^Pt
zgM|SH9Z2+S5z_OFLL#&oPTHr}7sOFYpguhJh*KBOXaspni)4fDW&`_-f;ZKGVpJD7
zra5q7EZG^vwWCKN1o<=DGqII*&gryV$fOB8R!W?dEZK)kWwqBns~~bl5SJ>^S1WnY
zf#gat6T)@zSQ2+-ih~QbXVlmjEjvfOJ>b-(;L44xn8wL(WS@Pq*vKI{$d8T<oSNuN
zx|8uTm}%)1_Sxs04C&*>Yp->FY7My>U^BV$KDQnTykj<W-$76X2cmYPiY6Je&pTOX
z`)V#DqEP4=JQ`Mx(ofjupRCO`Gd_tX*C`hf(A5RWjeSAB7_@<*aZ1r5(L&7W6ZVBC
zYtN8K1YJJGnq$pW+_Rhe#GY9*`M;}NL#prvxIAlej49Kn|H~JhtRsI+SM0DMsEDPJ
zv0n*@GP8^PK{7h*SvikOriPrKi7n_2;6lRH{n_<+Qla!l(!hj=N;d#pv?#p4mF547
z^K~~|bCdHk2AQ4A6$~cKz9gr9f}41!okC4l8Z=32YR}0Ay(#4-q#$OfuSNUPdgnqi
zSUG*j&tw6jbeD^{+%KzVB+#LM4j)n-z<V7?&CNn!U~+Ei?Oyxx9Ng%5WV)Vp%E48O
z32DN|Vh&u)?|%D=YNZ3>|DZ)x-_Y37DsOmQK8h!ha%XZ0ec!&aZGqdlR4h^mQWMk2
zeuHw#q9hj`0>tgP)%~<d<X@?CH6=200~fx}Z{t$h`|PV~p0#7;QwSU6)Fmk&pT+mg
zzPg$u@si#9s8bhnw8#%!n~FK!0~c`EzNWgD2n(9P4yN-RkUXZQw}MRr``Yg9ClGa&
zTllC`cQswS^8fW8`?{LsxMGkg)v1t(#v+6O1&TYZ#Ij=j+c|5`tC1||0(6KkWHMTS
zD`)^bdGbBHSb}1_*uK6-Dg|!jL4ig<BsmBg1RoI?Hb@+LdxgL5vFGP@Sh;S6^HUP$
z8)xhVH4&MyM!rHvyz0>1Zc?rzTmh%-l4>U@MOIE_=Ao{zp>wXF6Ue9b!dz`=9cQRh
zie05+Ufn#1tJRCTI%#);BXR0dL>QWgj6wF|Zci*6=NSp3xj5>;<HaptUx8|uR-JY?
zA47L@(z`l#8Uj9q|53ZF?H@|AxX7JcsH&h4sZ_KYXPs7cm}Cao^xlwDep1`ZPqy$D
zUHVN9F4>BxObyJFk<9_lNr|PTFxN}V>}4;>j?Ql*!b|==6QsPSE@xnDMhT1tc15n)
zP1juK{46@18US&GA;H*{ZO9=`M}&d5It@(WaCo8FORHT?VUQ~7+5kHRdgWM>6lO2$
zrb&FoVq?uB<-gOaR#>vcp83b12v@Zoe=CxtQ>xx~Sr?cMTHDpt1I<nIX$aTxSdJqO
zJm~@nt+=}D<<&eHQrMLk9?qsThY4`t^&!&3*NX0u>}%Jw4qbiYjW;=^#2pOlc;sN8
zI&IfBHh`cklHg$>cdDWpb0UI`{It)mYjtzmgDWA}#i`Toz9sU3Uf&uZE|H;!J%|@_
zqp=WJRK&dz1ZunoZpZ^=#Tc43e&<MsQ118b#(GQAh`hhF)oHR|URc)Cc2nDi>6giY
zSckC#QRkKNCqlTz*1w{C<d(HdZeQ;dYY%7iW~UyoSJp)5)=2JE*Lez?lnozvJ-M*^
zl}_PtdsWWG{I=_^yWT0p1d`E6>j4XP(q3J4<ST_d>I3_ox?Hzr4VF`ijt+ZG+Z02+
z4I5-dph6nHyU=tX{wfOWwe6(BM?<c1B)jeK_Ar{BJQ%?$nTJqNNgk%ozM(y*37?V<
zf|}S5Qp_qZKX*USK@cdH(U!EX!*vU79*45GI`KXVl_D6!mbP~CUepkwtJKwu$t6v~
ze|S*CV=Sw$rVViKlP+03EOdEAJ8)qa*DH^1wB_wJh%7h;L1dI^N;ka$$<E_cN|<nn
z0ulScZv;PE(bk^h9BM-``XgkLe4?x;;jm;kpRD-4$xRc}6J(zoaQDD{eWzZhxL6Rj
z;w|lSn5(7hB1tnHky)8j0JZ{n9lER&4AZUkz5G)8HN<ak5b5et=UiB3E9;p&Pjel^
zInx`YF$mkSp`sFC8^131)WlZOdh<KRn6jGKE(is5=eXV0u=H-4QK+Tq1PK-hVbNAK
z6q$r3Z;%d$ntUZ9rNl2KWtZ>Uep_7-vG|JPl;c&r#Tz7i?L_#3HYu^w?cJA^FMwyD
zM&e&iI*?e+D8+ZA&Og~qGImEpvG9Y6bU033JkzEt)Iq8Fi+tE?5(7pjk(Y_rq!?Ze
zYuDK8bJCy+tDWCs55_jz+OBt8z(Ma)UyFO11Brc|W`gjwv~^Xdji7`61ZoB9Qhhix
zZR_i88j&)wq;sxB@&-{rfgD?OhhHlF?(BN7;Kz^((FT!dFJqKK=YyBFyV}esl#REY
zMcG<)Bo>gyY*u%a+1=Gnq2@AVAxb)RF@p#~VVufTL5USu!R-yzAEA<bK#xePB)>%m
zB6KHGt3P!gYOy!wEZ#li{93ds%nlAcx#?m(?M=B1S6=x_=VyL?!RsfGf&g+R?XbPM
zbs5sA<ZmQ=_Ey*E8?Ib-PrXOPRaP3A+2P7#kDP=h0$vEOB#)T}C-=o$>PdA_ei7my
zT3ROzR}MTDKJP@60DRQmT5r?z%PgR;VX0#}8zk)_`cBJ;__lT&uUo%lty7BrFqH~d
z;XTrR6GLNfFNmB|6kj2iRc)Zt4kTrYk=ZTd)~j}s_!Y31Ta5I14n*y6T!Ry_4egKz
z5w>j6rNvHF`#5W!7#$?ch7x<Py6W2No!{9i*t+!BNuPz`niB1g^xiq<)Me+egb%D=
zy!Wz!^E2j8S`{u`l@FI~Y;zis0bvDp>dI{LDQnklLy@PE>Qvs-AbF$dL4YJL3B#f2
zbla5p89pJpy_~W(|4Uzdg^gxsBe4IdQ_^XJ1)mtMd%H7(gd78>n0OPUpds;TG+>(x
zlKPt@WJ&yhlzn80jddf9*c!0>lv9^eHJGKFTMGX&QA1jPl^$ANm56!{Jmych7I#^*
zE5_@Ya0X@V)fxs7YhkDs9eDKm>7-QmQB8DT)9oSQa-}B`w4`&|<-jjtE`qs~tScgu
zr%+;;w5-yrZl61AlR3`ouf6Ij=Xa6Ev+xPQaoan(UMSar&P$1QN?Qj+oZo8BReY3f
zs$*W1nKF#tIS`r$vxOkF-D#Bzyk@Guh2)GjXs7lRzN+X<MmNayX56XQdd_cKHalt4
z-Sp{Vdl?K4J9RZZx(P0odMV$`GzfB(WX~>>o&zZ~hI%q_aH)1*tCN>Hq~qM&ZeOU=
z4KWfNci7gp1?5S&d2|Z*Evoc7f&``}hG)_ecYj+OcEv(4Md+b*G&}I>f@~9#d(PQx
zR~NxMP?PiC>hfYVDP1112kPA#Y)cTfSX&7^-VOdRaNL}rsqd^h;h}gx+-1Dgi5e<n
z<0<UI-<TU~_4+l=&zRY`77YXz;|EHkoPDq@)~5a>!k@(KnK2ScQ6C=`X%~}-nZ({z
zZ(+MZ0;Td2krj&tq^2OFkqLgWZ(3-b?ZhQ=B0uUtqBTg%1`+GE*Ssb|KJcJ+8esTB
zov02xMxc{Q7=V8gZ_{;L@9w5UH}i(<?9_>-Y}qno?<sKdh5Uu3p{`!09N~4sNQ-U{
zRcqZyp6Q&oJ9TjurpaLOeiMUr!QNZXC=rWze2~+ri!Zd>yClzK%F*|=?n1uFV#HEh
zaO%=&Ff|S#g%9kT>m6d?G)+(NSED%S4Wh~5RE^M*WD~!oqt>62B}6D6e7=&*(@K<X
zYj2EW$G)|Z*46zHc+jr6j+Ceadw=y;hYF6Lx|i0eiy48&LP2y^$|{s*LnXHTK=+J#
zjusZdle(f{T^c=}q1T|0@;OOMCpiRlnSEOuIdddm0y(OL6~uug>SzNe^4YgH)|hyK
zjmiPz)KvfEiy*z8n8r>xx9@0A^oEA|lc+qLns`Dw-Pnhvda00k`(P(SC`HehQ`QXL
zoiq5MoI#uh5O9Z`q7EAJJWtq%+fZT<i_CHkB<W_K+_MsiA5Qv6A}&5aAqJ`?mj3L_
z+<E)Xs?%HGCuMPT>S||;sR~vlC)B>H4pZsgPB^8~({`6zX!<}{p}$lL7_IFi?F-Ts
zB*dvJ^p|gbGF8E%fl|5~=)3F9rOgg$Bh4KmWV%@Zb)M5IDFh}f`DL6zHS^J~ZfEGH
z70A`<B$mph?IFhfJ=IP@twIrbEmD_k34(3(w8<k-Hz__t(6W!UWt#yr#)iEsZy}ms
zbi_VhV+&&@PfCVjK1AI(ymvY9JhGOre4~A$y=-Yu!Qf6=L&Db9CuS5uhEzMBY>P@t
z8SIcG1FrSaETgcLpw7EbW0wEklTTgG<9tsJR{FLM`z{KUdeJyC=mZlRw#GDm#-YR0
zvF<t3GN^|$AWF~gD@fax3Q9@_2uQJVGL(oGr|+1YKEkle7mwNZ7xoU}YQ*o{ZeXvJ
zVt9hv4<v1U@RC<JKNI;D*=9A+gSM?EHaYspco&DfL{u!#4m@>C$nb<iJd*RgN%w*}
zh1?jVtd+x{ZSP(td`de<Vy(#G4hODPx?}yVE7sbM>J!3QOsH}-B9Wlk2XeV4JUKEQ
zAslRMrE(CWXG|GwC5Z&TW;@&PVLRlTpzN*GW$&OJQ!~g{yBaYeJ_(}&DPO1&ad;i>
zC5&x%%3tre-}$BbSdF8$CrdqDJ~Zu=YCU6Zv>WLPYiE1gfFk@gRR_|uJmw7&C0gt9
zi?C4GzV?hJo55rPXQkO}kaYUgVqp4@qi=s3Y)Hj%x_hgWkf5|Q3PC?mt07TJw;n3P
z@h;XOjoXJDxTqhON(}5^TRF&jyFMf|1U#V>qQDEkP`L=#2~U2s>a-k{X49$DJ4s14
zutVM0xSoJ#q{z*rH9*q+NHL!1igf3M9ZpOe=(b_@BgKxIDankq9edbgN_VWdyCZFr
zC*SInV_m>EC`@rDp~ro!wH(7w9kit!Ek&Ir_$l{D9M|@EH==Z<JOT?sT`Hx9v31Rm
zD7c<h3Ih3S>TfumIS{qH=kom5o_N~$W#++h6igtHn$43TKN4+j$GX+xXl1riX|E^@
zctJQf;PFi08s8;<tQ~K4frYWtoVq$vllKcPQAh#W7Dw$wTj_;kmM?^|8mu|wz-xA-
z2*}w~Z}!{CHlDyS;kTUB5ceHOd{g35x&iC_k>gZIy>_aeRC+L7+NT8Qq04n3F)Ag2
zRZ{fxI8&c)tzk!$sa^~`J(xS~Kxz(-v0IXnI#b_H9hyRom2sfkO%6QRnWWenk2FaN
zC{IOTyoL88G~3xmA=dSkXMJkm)QA!g7e!hlJC`WBM$samD8*VvnWRGrW#_XFTNnw}
z;4&mnQI{_n^fyh}g?tItVZ7MJoMOC0;wgE78S@$Y!LE)72gb>6cIxtkyZ`?C?T4~(
z$xGk-mUlXZcmv{0NN9S)T-EL4eP?rCeWC6V=?u|zowbYYS*Sy^JWwZ|eCcf8<{R?k
zB{+nzK{*H`sw_m3C+lGyYL(H-sjH)F!Ea84M0_WnBuq431NOrmD(PYn_`W#$h+1(V
z+C;0l<Q0dYS$nEOtz?1DpaWsMar{ewIO)IxT|@b&w`=pT$S3Vb+R3$sHRwCV!s|U8
zCt9rL>i?r{kP&e{=9Kf`Kz~6u2KHn1Xo|B<w}6c2K$4K@#yo|YPi*Pik9W;Ueaw!R
zfWjVz7iy3+6X@uYt(6p@_7gehtCp;DerlnK;m14jCu@qqTX-Gq@FPxLjwJb~qGB-6
z_EX)EF?$T0Vxd-Cq=3l%nEiBDCviRuCQe<vP__dS6~QA&!>3TVpJ^-1_EX*?d6Km{
z`BR7;<*2uxExb_B`AMe~-^GD0e9<7<_H%6ytl{PWuTrGl+iXAI_33oe#5r}j2t-g3
zRyS?G(C~lB8r+hr$C{8dgp2=&#jsy&5*l$8I1a(E%iE6H!ji*aTH=p3++O>oCRT&q
z!=n>hW+#SbWe=i0$^LR~q{F@7m)q2CnO=VT5~mnLOplD9*7hsiF!J`c^YEqA#fSAV
zQ877tC&&A%h1e3L<W{6BmKJg54!pXo1aOO~)UTa<>LWpu=C}P%``77P%?mDIj7l1V
zUoVo71{|mvXI>jmC;U3CV!Wx6OCdPq68nvssg$NN<`}LgXN?mYczjdynhBhb*>869
zo+*Nk$5|%786YKJ#qF+=QQ#K1WD5l9gi!mfhJF|fd$V9H6}6ChAu=wKa2^`JoowS3
zFK~Wo=c3NL>MHx4+|f6^{2J${m?H97j@a)uG+uVcI;W6ptWCu0(z8iw+TZISc^8ee
zH6lOwMHeI$iLKA13AW$w=AK(McpDLvPF<YLA%n76<uHewCms=z{iOXty)BjHI+^9B
zs5nv}b*9UgBCeePKjW5+2zmO$hQIJ0vYX@7$T#J7C!^e62Og^;g_u+~r{Ea>sEvwQ
z*KSA^m{tWyZns<D%u#0iaocW#-UhB2rx=TtWU#5yFGSXxoI0-Bu|LU`cwnRRODpl<
zWiP$t68qDJKS3{><d&|Oa_9$Mt3bXA?M}2{yI9~0_Gbm5_uy<EV<B+(0t7V`t(CPx
zwz}>1=k<){SN0x7=H?0BAevvf#>yo$`J=z6xk~!58<)yQyw#~S3KX<^2kbAq72Xa$
zOJ)9uQzzuQ=PmcV+5W0VDzYdfO|y!$2AOpPc#M<-JNQm<aB-&CU$;461@&f4s*ZpY
z-XM@y-nkHPh7Waz0peW3#+Fz`D~XQ4b6%lsI%}oAPM+;=b8o!!$_JgF>i+Gd^|!yP
zQ$+0BQKuxUy_^x^dck4)`}$xm;o{*jh`Id^yb+&>kHC+{{$XKi=|p7ock0r66GA`D
zO+XkC{;^&yQ?ara>K@oMC`g-r+8bQ^O%}r7eNnN6;RsIrpW0Z%M=@t9_)7;8`)H9`
zk2j}N3z?$qpX<pu6<FuIfkx_TAn{}zyM;-xIqWdW8{6$)+Hk=n@x1(X>T1qK1B8RJ
z4vV+_*Mi6brO+>VE`>tRf#kH!MX1T#ehhT~mK$hQzeoH(g*=N_aF^Xkf6V^9wqt4|
zIJH>W#(*tcIZ3Nj9o)UC&qM!_t2;8)U-$Y|PALgfFzmZgl_-KtC?@W0pZ#Z|$e?$h
z^HZBS^19f6CB9tz7>mO17;GITsa;jhe*5o6W!K$#yHm*XfmxH5OBcS|sd`lWcCB~3
z#(_k~gi>W6j+6>W`ON0y_Gzc`ZF|_++&DrKiHDrJnswZ$d0Wfpve!PnTI;IXG%}v{
zR+k3RP*&Jy)K^0Xmrm8Jj<Rd8QaC9CFIS&U6Px??nXNgBAf=Yfi&`aoWkeY85_%S{
zSr+k@)?^2oV38#$(6D9LXVtqz_(YVI2R3f3OQP)^9IiW~j-=zwuAwlrXXFs)(XOEt
zhv(`KJ9UXM@`{>-o`3eKpznC2k|T!(lqF+q7~f!@(=CN!1w@V{nkj-7h-kz7wx##l
z=T@soG(;;Udy4b`4kV|BXcGLeKQA8|?&GS=?^sJIi*e6q0u&*h7eFY;>+`c$;qOGi
zjZ+Rx(MgQDTol~CVBsCChMvX&^~vxeLyFw%z(XY3#|d}TzOcF}?7=Sq&9BtrV3q^X
zh#$P{N_%E@as91J);qszlgo4$h7$gE3P1azQ$?$Y_3ZZ_kUvRRrGT@D)3s;iK#1cy
zi(-4+DJShD?pv*T_Usy(R7^T{@K<)~w6U=BA(nn|*WeVK8|Gmo4eDwyNJmoJ3|&Ck
zhwMvgwCyA(sDxK5BL|R}Ntyd)Sl~Rc=d_`6x~NWk1Lo{@h&yo2<P3S2dK>IZtMxvl
zB>7;bp1aHB=fEp=e2UA*n3sWlS<dU7Ygepuers&76T^M`^6tw@jGFFQ$#*NTVuu6Q
zmr07ZV#RX%iiQOkV%RX+^<H&yzQp=QsJh&pyV=R)5YqKG^d;$|1J{bbm^`V^tyfB!
zA^$KeJibtuQPEP_eO22xOQwcN&OPB&XbViujLg_qH$+!*&w*1cWZfRp_gUgmzNV{-
z4DC$&9;Yt%$NF2>u2^AT+ch~98Z|}iQy1*jV#{$8bLA5qT#S8Pd(qge_Bh2r{B0oq
z4trj96url^^IOc^#DJ@M@LaU7@6bdl(?uK5fy5ZJ7*e-0h4%cejynL@v`UviUG0(u
zVM#W0Z7#F`_JX#67e5TkLMf4tZ7)mQ*}2_zN%cXDkd&1Y#1wU~-GQjL7aztZOyHrt
zuziay+(fcP+GlU@bSiI*FLq=4B5~(MtuM<_U^pe=;mx>7t*{q&6Cq^==Q3L{437>(
zi-f>BjE>SS?G`B=nMwBS)a5$fGBGnYY?pP@6fOp$Ro?1kNK<^riMbtid20tUL^u)p
zx~>k>wq#3^GTBR7yLspyb@062)$y!O&LbGLuc+w>M?-p#eTa!v2y_H-IPjozS6pIO
z<|`quWP2DPy~9x7tB!Sj0<2X`-vxVVo3Zg8IwneScI5#p>j0UBSxFHA32SjSFgS!I
zds+Jes+vCeu6184b$M#6zs;`dI=i5a$}Jj9kpmA*QJqK^E@O6eI}~s>;5vGzoK*6<
zmtSu$Z?hCx@4{)tTV2|w;*LPOrnS@DLgjRkP^(Kehfm6bxH_l~NHd^o;1&gwI`CX{
z`G`%}bqzN;A(Bna`?HIo^9A^VV|pq>(_#7g1q9?nu%R#|h&6IIZf}Mm%2T)nqJ-FP
zH`J5bZ_=Qg2j5|aIYoe^O`P3xlK;-=|8`@MlpL7MopR=t@j=`QOcDd$@3a!K+;sH~
zc2k??;i(y?5bM&69~>3-iq=DVP<Jr!mUI$9EcFF==BrmY;G)E>_R6*-Y?Vlyf_jF!
zVB9)cA}Fc+3tpAo+0s)a7lmk68e<&G>NK=h7ZK#j5maMUB7N8l*;ZbYui#KVH0>0U
zCPC>U7b}$RwOKd6O@^f@+@<hqp-$L~v-iNhq4iitD%dG1uuzu*_x|}OY)LnYNKVM{
zDo#R1Bp`XPU47&ASKHEV4G)LVnl2y+d=8}Mn{{DzF`wA7QxW77I=Xe!gpaMqaN2jO
zlh=itR(!Y?pwc53M0|qK>LkQBdFthDAk?8B;O{P5(T=)h=;*j8)TxL*$mvBQmeQr)
z-1wLbo5cx@8qLpbI)HCU>{$=d;AicV-~?{1cS+C^w<A?ZtJ6G65QV_m$~Nvem&KAi
z;nWo~e=u~On&<GJu-8><!mD5fC4`-E>XOBja)vL-pH5$uYp$4{+v*u`N{kMxvcst>
zWe1Tm#K5j<y@eXWfrCm1oVw@*`v(Fmf)0U=D9zQ`dVbqrGcFOMLtH`NRTpO~2C`sW
zZ*OmnJa<yf?DwuN(uOIQ#}2!rc4`Ws%mHHyg(iQwxUiOxR*6aL`q=8$v=I!(X5<-8
zIoA%|u~cxfEWf_$xG5K!!0pm7!Bq#Bz_qftFUfCZZOx$0WW~BDgB9No;Y-&B0ngbY
zIG8{PTbJ8v$*NU%EOQF_FvgbQQ1!|^V}18R2|wb#Iq*5^AsSHay$(DM2z-j|&P6zj
z{Ygm2!YQRzgQQ+V+K-cTPFnW68l_q_vDqoqyb#H5cXzmxp8Ya21r9opm?3fA(WyBy
z47XVe6^~45=bXA^YuAxcVGHp=L-xiRkERt3OvF1x(t<$2T@JiCC1J?MOdn1vZ(1P0
zVVsH-%JN8HM#3Kqo#%O{RZi|S*(xx%yt#WyJgqc9if=jwv0uyr(W<tH?z%{;Us#Ln
zo>NafkM)@U0GjFlpZ}b8WhOcmy_3EvH6&t@6*#1nrf=z1GHr=+FG>ext<)h0uE}S0
z5pC_QZC&mdY;_8`FWJ*|ucVeQ)Vv6Q_O><z++cJ-cljF}NKR*`TxTHm_HM1XU%mF|
zLfwg&22%W3uj{gIZi};%)(nl`D4892aGQBf9<vSA54n#pm8kyYXT@dW4kSt1`hJgm
zRzq2w8IQ|Xy^8E4%p7AjT(1E~n8`4*@fUY32U3)AXcKmfep~R#p0=M;ZgeIoOqisk
zg5H?ZdGUR7J4T&Cps!4QN@H@ABXXpMD!xagO;{z%mO_^I%^mj!PrMCDizSeMOh}iu
z>C{tK#p;EnSjq^Qax!zARvK16YC__c*^p%dPn_q(kge(ynrWle|5U1Kn<~sn&46T=
zE4CkR5xciTySP}TPRfjUlgT&~+)gVIh^w!<`sKE{2AL~K_r<%g1yK~s4Z(rezE-$F
zprws<byCyrgMz=?sjDrc?75<HN`I(%-cn!C(sh?#G7gzz6)n`@g-cQnYXYSjLJhIg
z#ygB#FutdTJRQ1#=N8}G*VT$P(Y4nVE?Q7=LsCd;36FY%D;6{25IF<Vm>)}IlU;wc
zR5Hro(3D<&$hG@@0WNb&oOnEcMn-&*E_+8qND>ori62F&ik6T(d3F_E$Tn5{zr>BO
z8jD}5HO%~qx-*eAfE%K1hrKc~=rVKtJz|4yzJ*F+Q*gNnVIPQAmozT}o351-iYnZh
zTyEa##3v{}uCh(dG%Q1Uj3`RhT4t1E4kYM|=0S<8H5+gSoY|4Z!S1X5{?OdMxbls%
zzD^*)z_xbJD&|YGOWu@NiPUTX&y`zoe_Vrmf3;85g@TafEh)SWzbyxzTYVDGf+?GA
z9M-x%*+y{~r6OogT*~+SH`@dCzSQPP$k(Vk3EP3^`_@d1jt|*8Yx9&I2t-u)SFoJ(
zM}&@XJ0`92p|vv`aI5^r#=$9x{kYUuUCmu&e~D}c4|cN)6^vEVY_QWDND_FGuAi`X
zwKbK*5-THDCmF?n<V>w4#t#<YoAMOJ8n2LKeZ;8>RN2cJj;JTE!P<b{wrhw!Oel`N
z3w82>Ou_!$^#*DGrSF!dM98++8zdMs>J+vtjnxHvPrX~>sw_m0K?<lVWroDr=Fz9i
z8$uyr4;5TfpjlnWR4JD82?rh{*#}z--Cc+pv-fr*375KXgQG4Pim}MIG}#-r_cg=_
zAC=+0Q*jAM8OSGhj8lOKx{Dy0!c<%78q)3d&22$9PQeM1Y0;^P@i4|<9d&6K?~sfH
zBYM)lCE4`rZo1m}DdbSDEW!irTf5mg0Jo7!--S9+d<mLI?EURzU$uU@Qz*Kw6IPTI
zKjx|nw<t|LoD_A{t<K1-1&EVc*1~TqbVeM3;Rt;~?53a=OhAe*w2HSR(x)CmZcHp@
zm{q>L79%N$XomEApqK+moLo9FHqFWV9gS0j=_lcGw^Z_Wj=gUm>_(Pyxx&k+%eSnX
z#^b+dAFA)=#hN6AfSc6c;|?Txs-8&68TZda@oF#Ghua*{JEe^}jg&#1<nNWM*V)5u
zT{obr;WDNvr>2g!eGr-;!S<c?kzMl&{|)TtuRf^X=Nxz)bNe8H851Xxa@@YF4If3@
zNvG6xmZ%a1fD32Z8E+rS6<@b?lndS{#7m_MgyJa<T}&yOqwns{^V6cf%!4c#3fh;F
zx6F-BtL#*qy3rlpvUl!r`)GS}xZ6qolOGegMjeRg^!&C<Uuoad(C_w@Yn_5r>uepR
zsK&d*KGwa2SI3K`dt)pPBxVp&9Rj>4!pA$bTk!7&iN#iM{FD{hX_W>PLVUR1bzq-p
zxWOsG3B;*=)TxU}e0s1*ry{pcwsA`@8N4N2XgGN{NSDB!(^<r4r+sgZd+h{H#rzg_
zY(&)lzWRKwHL?J?l&H&PiOCaHluq+ejCbGP*1Es6H|vzzT8kkFlw9NmVn5J`JYM6c
zoKnnL@Q4f)??2ntVO?ZZ(2*R=RDT_K&erJoeG{WxVEjlsgogWUJ*Sv7ken<y?N1}Y
zZm-T~M_`pj(%Bii9Y{HdD9c{COf2ej9tJzAOGy_h8wMw+oM+|o+u#}SNl1U}@-9=6
z?QBDm&O29y!D9KWUI5R1Pvk9bg0`!}bg7PYxTNV><hr|{*lAT$Eh$<~ZWybA!^@<N
zyBk7cx)#AD+|{XzWvrRdp^+Qg_SD;U!{E#N>tt{#oebr`HSJ^EH#1B`S<m)18q(V`
zJ2H%b;nd{T3l&x}H_{eAX8YQvS~Jmi3ehZ&JTVp8hZoSs?ynxpg+dx&rORn`+UYt0
zlmhuc_lDG%i;Iyb+X)8}Jj6qDkBraS!J0=MrHGmE*~;j3h9gVAlZV{vv}y=N?m?ul
zN841OoDowf^{TYLGNR%u?7(ZBd549FZ=i-9Iu$O0JabHw$k<P9bShJ){KD6~&<^J|
z>ydtmpZVRPN*{To?WB!Nu~Uk#q`ESEROt2r4ztG=4L?liP$DtiN4!Bwy(8qL7xDyt
zyxY^6S4o+DsLKv4m5~8y;b?Y%)Xd7HK&f=$*IZ*yG!%xdm(oj|U}l5PfP3I-d|!Wj
zbjpskb%cIN|Dg<F)D^Z$xl$)=Of84W+~an<eIO`)5)LLE$A-8~&Vh?Qty#9(PSoqg
zIO_mbBq4P<mJK~=H%9Daw@7NFMO09Y7&st}L~)cQbC{$qOq+J9gP#o}-)XpfJt;;<
zLF>SET7WA9JDt7yMlKNHXP}<8|3N#`w)9=2<aM5Q%84@^I6BsZIf*l8+caN){S8h*
z!@!?$s6VVL1J9kx@Q6hem#rR&3J~BhkYLpjrYT||-se1#Jo!d=kDi@B_0&gsW2fKP
z_CM*D)4MnTkZI(mroVQfrixu1)G0zyB|STkJYqR`q&AX7{e#_Vi7`oMrsVH(I)K+U
zN;8~vjgo%$+7C4X#>FDI^x`>6&7zH+R46FC&7FEmN%NEYTKxB=4yc$mg_-p2VxB+C
zcwy7?QxzD;wh<)`&PBTd<qc=eE}O4$2@A!1LGkgO&%R`x#w9(!qEnwtgy{{7wWN>}
zC^ay{7q$K2Mn_gD4_uYlM(9KBsp?8vH^d_AK<dK$y}?B$+cw3Y1?h(^iAT(jv?-!P
zazNpoJTEVR27wiWRhPvu4L6V5k0w=|-^P_Q8NksWb}H3z8QGbp&}G_>)ysw8I?>L{
zy`M3kb|5iKH0!}?9!9J8<DKnL&UMTadU(zhX((_skZQMuznpj4iac#1PwlAvM7>PQ
zq2u)!%5v(|pf^IkS^LS>fYr&6zSB<KrRwhoLe=?y-9`3Ojk8#gv03?}afsxpDt$Mt
z0^`cjb&;U@(;Yz@izKy9MIon-=JdkIE$}mowm{iu0>zcE9`@LQ=aZqz5=*SrgpS+K
zwoY=o>5StFSe{YpZbC|#{M<scipdwUN~4UxWk0tFrIWC>{3Pj=1U#ORzgYAZBh$|}
z%0wDxf*~lzu8@*f+LVQKc%rsP>=&~6vQ2nEjP{+PNaNT<WmUCbY#Wc8nH8)>Io`)1
z39ZIOP8yac-vcJ-ggI@$)P{j%kp}LaPF+mR(*~`rWWSu0eck$X&d+#p7*BawXe~K)
zq?R~_K>91~#o&B;g&cJ1@)-^u&PxWTRCWrf%{BYgdWIbY^2YXd>dNsoe3Nx<XKTKb
zX3%q-EWehRdf(`M6ZboX_)OZ2Bp*vw&FfHPa{l!k#qFb;ou4|N=C_UX;c<SW1}<S#
zv1kh^6oBMbWBX|zb9~xw<~P~FGs8|HUxdfUz<#Tmr-uY0Jd6c((JSra3L9dG-C1tG
z-A$NoM*{cf&id2q^yCivovxD$5zWGreGVjjp0-X*ZC0WH``vCOGxPn0;F+<fIS?I)
z+zN-ZFl@h9A5V|GpkPj2zHa%-b<0*QS-IMNzm53Pmt5u)YD(nnC)w#y`-5%_;z7B}
z3eJK$aR#G!z1bf&?8KbPhn;e6!mvRkQ>PQ=G5e#|Ax<SZ?_xV}>I&tgK~AHK80rcd
zwLdOYLG%>!2kq0{x(Y#&u#Q}zTog{LhTUL%DW62F?N4%e>(?$>w!$go==-F8B4~En
z{xrKpp3qIB<SGqS9(6kBNQ^=@EavggvOL{Wa!d$1Wlb|?V<p+LKd&AxXi?O+l;(9a
z6+AiKJG4ssOK*sTqZPqne^LET^A9#1>8_=LaUjW)4tn?7U)IEBJeeGYA9w0v$oXh^
zLTmf8V1LyJY9-QeDr!X(tfwm+<igFVY=2#lI(xxS>4A~Gyg{TrU@OOr>*P6Yf79N^
z*&}I9`rUvy5D~7M?^wHf$@-Od+-`rHn{xBysPi+&bLHh%+TYbHNCp_%MwtloqBGtg
z&<eC}B4??mJYs)eh@xpu5l1$<xU>Ui0ypXQfg7%Rx&1@KNiyvAxU7zH;I$@5W>Z}^
z@sCx9?m%h^UUJA=9nvKSrm*&;{ZqBWwvIsv=7OC@U9v)oAVXR5&)u9HjZlQg!BUA1
zr@cW+3J!rW3FE)CMdbkEVoO&2s8g3K1=}1MR+I%uRL=gjeQ~%qLRMF&ob$PAW`q4(
zdo0xENgQL#Igs4Y!B(Mnc=93p_nJu-Em(}qxD{*{TlRnhPgMcur&RaBJK&m1Qa279
zbxZ8%f7F}Bf{1#ap4;Wr6_I{4*uNdLj>OEYq!gO}*)$fDxP@%wwwzg~D))cjULrG}
zWbrdh=D*tP$MH($L3j!3>NC2xUV?#6zRpHw|DAMuHCKZpkWlKva*<F7j7ODXr(<5`
zx7|8Jn%D6W=~Jka;Pn7uS)Z0*-kOFxGB|jvOOs&iRr~a=U1Y*QdA`?M9s4L0-2{c$
zXPhp4A>kN~!%kTz3!c5o(EFLEbHe2-(1kQ(c2KwrI$b9G2qyupY8R(l8tFTO2A;Lg
zI$d)p9F7HpMS+EZI*OHb;CYU5Ve<|4jOw%UU`aj4o_FeE$zy4gWzu^PeRpFogK>_~
z{Lii@#nCG}73uYyx}x9GH&;MTuumW$tkCCl&pgU(q$QF%?H;^|@NFh<;pd(%k#=%s
z%j9OR6QiP>uRaJl{~#8b4|05bUVSR=S7X@^vqeR?q(H&-$ALeW*UoQy+yBQu_W7r4
z(<~=Ui9&3RPyvH&1P0p|baN~#op4I-5z2W2L?<2x-UIf9)he~I>VgDa<@W^=vjI7l
zav6=fOezwA5fkQi*)tm+T-*j-J?NAd&3i~|^hC-GdA5l3p^~vLs_(^&QYRCJ*6oM`
ziSISfaooRoUa1_g&GWry<;ruBJ=c0TrD%;H%^UJ-d4YMZ)wA0=45nw}T0bg;IMX=i
zj}PpNPlxN&{5H&H$b5)6aaB3hi2-m11hX?+GT)p7`y$hjeaY#klt%BRuXcXvb&QP>
z*mJsBlzl8`k+p!jHc05Kg9>Xc#3_Gitr@g9%yi^J^H!I(f#8Yl%No{&BbP+9olafq
zzF+)i`|`H-Xc1-e#^&hMY5UI%_3SIU&qSUd7W0Brmz(Im(a}-+$~ILS`kM&t8g?oQ
zoA6j>{3JcJ=hn=W4H7(&Hnz(H;|(6PpA|*H#N0^xw~Ms?XYH#_mx5!}!~_8=6zjx4
zEa4gEOzarQVqcxT$uuCMs#FwS$ppo&MiZ_iE`(e$b352PN~Q5N?fuK~#2z?B4T043
zc(&pTE?d^u)(|8Sk^D2wFD$OCBuL;6T$8#1V~~AaL++v8_$>64a!iSCn~+G4a?GCB
zP<hcQuBAWXKvJ7Ciqq4;zP=i2_t9UlIrbxa21xA02Z+9TetR7!wGvn>mz;>f315Ij
zNDw@j_9(Q@UQpfArK}h#rTTR03WsI>Ox=pUtm`x6k$XvgbBsIKoZm@@OGuPTA&<lt
z7Nq$Vq>kgn+;#_&{K6!6YLTPtMTtyn)~>jD<r|z*o%+dtDVJJsllJ23YxX60F{iH1
z0l60?$DCMyX+dH_7q7Gf$?+{vW=1I+doT-j+37TuwirYO=nJ)&Q)rlGPB%d&M?>4{
z?PD?L%9onszr1>@R3M7R!K=WjtJWuiqAE-!#6x>YL7;TyXcM`x*wGE5aY*I26Mi+n
za7Eiu_>geMg>z0aiJ=H~;Nprlj7<<Yab;rEa;}cwf~lP1bAaK;dxZavU6g11!|ce1
z*+4JN_33Sx<wz#MJ!R2?t)SS;+P>e&g%(aJ7ZMscM&698YRHXCVtYAtHDvj)htmU-
ziP6HYZW9EeDq>YJP_0f&EX~9&a`U`AKR-3|fb%mp5!u$T1u`RW;McTHY$8AQ*r-#k
z4@$Qfw1>aCru^E1KrMq?h)h&*;sZ%F@;Y`a?saxuHO(ABntJsR<aQw0bhGY2Mi+d2
zet(=yuo94*a*9<%MUh(oyWR~A-#EXdVv!b89La7*AP)%Ox!G3^lfA@l%oi-jlZP9&
z$GI(q8g24-Y=qX{bh_+Z+(j-&Kr(DmrS*B;#H_s{TW;)4ZE=2U`IXjm%3j$91P3Du
za)@;owm;@TVop=dAv5MtdsSl%ZYBjah1f2ldLot_XGv)b4PTw#1XVUVzoTtru>rLe
zow`U@ZLg`1X+ul6BSgcgQ}?H)$LzHQhVu=KM#OXK;+5)iXbr?iLII`1BW&Kjp}le*
z<rWYE7@WF%2FecA!i}XQ5TO}7#FB<HGVxB#I%Q3jTvZ0PG=V6b#%3o6eW#q`M*?Wt
z9A+VWOLlS$FDrNp7Ey$yGD0X7>k~K$0MC=84~7iu9-V<LuOmqZxtNbV%&f3XHug-~
z@c7&wr&XY`uBO}_S5wAc7|9h43rPqcmq>LL%ljLY)x>UYL+o);1DBRj)`>lXCKNNt
zE!|d#oZ&?XrEG7I^hx265Ydr;irrchofbYF)5n}Tp)Q|5!#iXv>$Rdq($t@CXS@Rm
zEX&J`MWr6G*VV*n$f9<lv*MLX!REjdGogtjI6=d4%x+7z>yGhd5TMO+CR4UAr#kr&
zTxxN)ZX0S-ZxqjEirPLe;vZXmy5t{4GIE1Rq>iAnWV{F{Q0P{YWxQLs#*^a*Uq`#W
ztz!>w%!v^k+J~LGq<~`+!y{-%?x+a}0?j)fgLxG)AMpmsJt1l*%_*vCThk3w2c8(#
zT~1x#T&_Y+Q2VVErLWH+O>96J$nTh@P$9+6+-Ga6jwfRyDXf6Hc&8@6#W&0ER~|Tw
z*w)oE5GZ=BCr>$b@yw$9w~Q)t80cn&_3hhucOX86fJI%*W>SjK8z-fR`H*U#D=F>H
zoVh#ju)t!q)hQ?098m}Y;;`LS-<Ek@2cNvT^0$;o7aX`m*+~?|xQ%1X8AUI7ckbjR
zLrNrAg(;ch><u|(AZpL~oo9s31(sJVnMl%kdt;)GSo%rsSfZFHh!rKcHtwo^Q|>*b
zha8=r?)ROFq~CXrkD`)%bM>2pJq#45==~xCy#ucU3bq!AToARwXYHPPGL@!c#HH{+
z*9=^j*#NH%!Id1$2&!&-OM9jcln~MDw+@#E*O|;^lA*JRfIFT?%VBS=ckX7(>M|_(
zXSBMoWP?%X_|Dt9UT1*Tp>QX`a^MS)V%YeOFyVx~J*S)${j<*R=zAr|1P3ozud6#5
z>dX^!JDoah*7<EOzT7r+`|=2?K;f-82<gWc@exj|niVB}36EkM$}{EC=fBGNSxa~}
zc(cdnp0MF=i8Lun+a*tlE~TOa&uvXg$==wg^&7dpm0*qs;O41_HO!r<o=L+m5r1QS
zPWwkDN3t?Hb<ui(6%IN-x8Fuu=TK_}!-wQfUG4q!82Z%o_nSI+=f<h|2r+msYCGWZ
zH6gUvjdXiaF8a~#Ynfb$`*EDQ^qV1JW;rC>+eSY%0e+7=<(NT{CN$MB0DEk6o56}Q
zIW3zVa-N(b_i!)(Pd(Gj_#mztmP5{1lbJDe4<`F$^u4c-d`7kQPo5zI+tMEG-_N$5
z!m<ycrBZ#E-=<`d{qb?h4ddC*%fc{w>{6$k(+pW94+L9EIs<&EY$B!ZD;{)yiP^Xf
zmT0oglQekjGkK<k*y%uGz`42@H#m6S(FP3V#@Ysc<Evh=-*R*So|8mt#aluTZK|zQ
z?8kJ~X@7O8YlDjhB*hh->uhp1XwS>`2IO?GjG<tOJDPyUfYVGYu~6v~VJdC9n~Hif
z3n!lXoC7KBI>dI!c*b^G!L8X$qr*dE2nsS`P@~NtlBheM?7n(KC{>s@UBDZmNBrfq
zap2;RxpkC$`jcq1wzgM>ss#SSsfo!FHzswdqAKohSSLk2KE#w`aLsO>>2aiNoo(Fd
zcz=UasI`jQtYqY~2dXYTRYJgW>T*{g97l=+n8g&M-r02~6xF<8k5eaxkSj69@o%mM
z83du@@32>$)IX*B>Dz;C(jl1A`shCMIPM)to)GIc^|p@NyV`_oA@S9$21`v!6$FA1
z6f|Yu)Gd=ZsZbclR+j!#7Yj7Frd%NzLA+s8l0Y6ePcpi6Y+{4+yAFP-wRNA9Wb~6r
zP<aW|ySwSq?RUy5e88!z**_vJ0)iMZgY5r3_MV)Y-t;tn2o$23vcby5VP`naoSNHh
z59N@1d;^4Zim^;7p^I}rWbdtheM(4rAhFWn0Z%Htag3Dw1AAYGczH3C;>oo|i<mm_
z_=ebSse~Yj_RVb`S=J`D?1GaFffKe2cs#QQ1rDK%Y_S9OEp_bVwxBiN(dLsjrOVpj
zw2Gl8W0HDEogw?|Tif!Yx*3}4qjjgIHl2>hMI@Qo5k|kidV$7)X=F?stJEb+d)JD!
zZ?q3|Efj1VCziK5^;#B0xETAku3g^h$)%@Gov?493qF{}Z?7hy<c-t)u(vwBnAfu{
z_t|%JuR4!(BD!_lsmnn_1?$?;8o~$LpyA&pH?j4dnp_ShP~j!0f$c-Nq_{bc%Hrsh
zwa>=;BNNm1;T-X5GS$y)aZ0%*Rwu=`hZBsEX}Y)BDJ8$SzBm4W()-$Xwh1G5;})k>
zQ{F9E&?m*S%3J2U3f>S1&t_2OFjQUiewPD}-s@0P*gv7__K~Vn*m`<%9rRWg*KQ;S
zY~P*j=C|RZz&(+q&K)>)g)UN5>3uROqyDsywvpV7#|JJC1E(gUrWU!f(5sY66wpJT
zLn)>}l=Q`sd1$YFPy0sP4HtQ;V}tB-Ahk8}Zabmr$8s$o^O5q}Dd&u$5Yqk3=k4PQ
zb0mGIwm<KS=q|qZgagkfR@@JtFw6=6iLM7yb);$?83S|%B6QvgdSGe(WShx2d&SQm
zcIwoES%F)&@2wWe>4k07sjE5cly$Ks58C&&BjJ`c1E-)D^(Ik#4D9>sMbg@(W$cKz
zx;nxFA%37HJ$4_jMPy>I3iFUTSu{ooV_~W|IH0;|ZHHasIiWBi;QOa-TbmJGE26wC
zPEGU~!8cZ1hI_5wX?vs{3afjhLve~ZnS+@P$S6b8wmsjxlo=a#3JFG9A_)RlE$wKt
zAr%(5EB6+uDGxc2JZ{#Ej<K(H=0a~lxrA@wuu~3wQE%3EwRzjXj-ybUTe0YwCM(&A
z-E9!z%96Icpj-~5wux*Up}n*{ZRajwo0C}4smQ%pnm(n$35kY}C?(3?M(9^fggKXT
zw2y@R;Dhon3!6n6F<INbHW!3PlR=XmLrrqA5Zs=shHZbF1Z4ysc8bwasobO{BB_!c
z=x{9k;kakWTb(GOj3KDY>|pDEP(2y4@R545(+oK9Tw|F$*(bV^R;abt9?fI;f&S>U
z^E-OLl*%L}L#-X^ru=afT50SG>~J8L2(8M>48Mt+0tZ*-_Yh;z`D@wX?)&n-3*#Ci
zgc)}r!UQk~OX7$f=?21iN0SW3h_GRVh20KZm~jhQu*VWJl(k^iDbxTn@aNI6&G6eE
z?;r{WqU*V9aLsLy%;toI6;9QU5iE4HdJ&wAmK5PBqw1s<dBPif(JrHvb|2P%#KtEY
z&P5uNxTV~wQ_nBE?5%dJj&IGQq~yba%;NDgq%u91X{8~}Z-dfZK<&Mq|4S8)OLCJs
z!z2KmMp>M!K=8M`5$fZ#%H!#dDK2V2_jMv=j7wkZ_3K@)u#@db!ovB4Qz{gc(od*-
z&Q7(luzk^cF-ymsy4r_w#mZSyE)~)yx7N&)QZXK%JMPp)Yl(XV)Pk01NwMRcv@_LO
z8fQhCK8B+Vb+xr3`8~&wQlrUtMF{(doy}#RK&vwD6eNq^v1Y|uJC_r@e(f8bpE>lq
zN5?kx$F|t{d}!sm^()pog(OrXQ=^lU!nzC9LycJDPVGCP#J&KL+CnyX-OPbkll@>f
z2U5vN_0=MM8bky{o+4e)yZummCh%Ig$dghRc!TH}kxHptuo5Az&vl2+4%$)Db|Jsn
z7}BX-Y!ir5k}SS+#N|76Mav>5W=LzP?bcy&*5<os4u#TLmrYSu+b2>-KtC`5gh}Gb
z?rrJlOF|D8LhNmv_YOP{mnR?EXg}P&<^bIR4Rf|IvNVVoH$`@2ig}~~d#aw1^;sSg
zQYIwj2BLM+B{RM8{siWmAIT+Nwr-{KGf$A^_-u~bk9LDPf(R--z(J=@UF%INnbD6W
zrt}7c6!AM|C}=yKiDf{$q$5Sl?vEEdMVGJvfXcelAUg&;W=kYNoR{oI$tnFrbt8>&
z1YOTP;?!y55BK}_leLzp%V9Vu1p!UVev;Owe=M|qNOL>vry9;e?l{OF8Fy+*_Q3IN
z)2#h;4v^rYf%CgiK%#o8Mp1Y%GqN;1Vn5Tm$q}O5J@~yjb$Z{$53%$gB;4s|s|`rQ
zb8*%>b(+y78z!(d|6Df<+MFp&rkev0ZNKQ}#fNZV`uWpOy-F3Lq!E-F=o4vN*>O_1
z#(B;$x{xKIbNmE)WPXqH4sG>&hc^t_FVv^QLL57+ET0ZUJ;YIH-Cg#JH3a3y4=p!t
zIn<@8ebL8K<*#K>_&K*7*)KJ+xDE$=r&MH7S?4&{hz9-T|APrO_7^kpSn5lkB=%4`
zVW(}9BY9DA)Z4Fg>vM`Nh$C`%kmS+G%N=+jSR^%I*5spzt=oPzJGKIsj^VyjO2->w
zS>yI$*>3FD5?8p|2HjMo%b_NjkF^AE<TX{lp3}Kh*WB_u&_yFlXAoJD@*oB4?KkqH
zDBF~*fl|I~%f!TZT#|0TnO|I^=#WT*=hSF>VuzTOM&`F_Ky|*cZ=?r011AV1=jI(V
zy4zyRe!C`Hxmx6Z0S$5LVm`$4O8Aqvown^h`<<eP$TdoAQALt{5vhdlZ9SJ8J!Ie7
z!a!n4Qz{nSKVrXEbs6**{Iye;gGEeETa5jF&6ey-%3#kuG|X1#LG2B$msV(w7Ee5#
zV$<ypI&3->yb}DPQx_|sF6t&pMK6Rw9*$r;{Nx|DNkPws<1z}1$DO)7#-=A`?uFwX
z*dH~tVuBPE-|=5Op=`>)Lw<SR{BBYl{Ba{7w{nRr94ckeshh69-u@&(FykW;woW-`
z3ziL^7w~bv{b{}oA7?Jyb4t-7gS6#OyT|^lTeW>5mk5{h8VMj}&dL<b;_@_f(S|P2
z-2OaKbfANt-?eQs9fuN`YlY;mg{qUmD*o;-x-Ukm$VicuK_Cqx8(t=tKExNfdc*!Q
zG3erZ=MHS>4V+@2ngaULkwP(SPp}4m)pkim--U%^UT%G{fgN~~o;B;%+h2FD5i>6r
zU?s01pN0cTgdZB?W@wP~Z#sMs0|{DlG_*RBU~wD@;Sbs0wnYfpne73?MqTw*IkUAT
zO4Ve4*DZ*`7homk4m)*uEG&wZTg<%ueJ<(zHWGDR<CH{jSFT@ntNlZLCvAiI9kdi!
zZ-D677`=~sr{$^nk6n+X0pL^=6=e9^y+M*w)p?=B9!Q4vPxXxS9$0{+DWJ!wj2uXw
z^0&=$rLX;ScIkB!gH2w)FMgP%`Ipv=ybE|<Q52rz88ySzng80>Wj(Qt6l%icE!5q&
z23z!RZE}LoZ9~*$xIxw(crNei^>^97w=b+^4V{ueT*$!%KL1hgS!`UCVms;9K?h!k
ziP9R$*Nr^GZD}JmM7q))PF)Ut-OvO!yZ`E58*z2zR4BLg<Vk?%283Z2Pa@XpziY5U
zbkwe)63Es-3JE6$9McEQ&LnAg1VVPuDJ9iI6Edocl>7E+XR14E`XeK=PEo{SQ-7;{
z`k9>4T36OuECuc_-XINY7F>Qm<4pE8cue8<+zF>HSFL|Px7>bay)|tLi%B<-pkkzk
zncHQb)!qcP6U{shoue)mSsFn(WE_u(FkQ6xjMi7SB2+v1Tz1z%2NHWj1fu4Yer1xE
zhw$<F?Cv{zII>d=YIQh|!T||o4%p|MNv2LD@Dz2YgAtY!VYspL_XIc|JfqV#M||;p
z^zC!c6p0IsQ=T4rrxd|yfyF1=X_aU(uEbpSd0FJ*D_!*B6t!osy6T!6?DN}LrNLlk
zx^yt!AUW2>vcVVBEF@>EzmM?s&Wh&^UTJwkz!6+_DWx*_!fuvyfw}hmIj1fdOF3}i
z$YsxL$Th_6d(%BGEODxmFo|7a!F0V?$dkN6zNcSQQ<DxpxoZ>M#G-T5g;yZ0>JVK}
zM2gIacS0F#&nh^|J7}_G1Su&p2di?_0>7Ne(@+$x6lPp4piU*RxFSi4f<&gJ2Hn0r
zyM{~Vq3AJxz18X9R=f*Gv3>EGq8*oh{iRN!Fio#+ACP`Wt9``2B<E{>+gh9q2zRC&
z*qleBxB}s{<do-ROYwAJf~5Q`2C#D3l99##*_Y;Kz5R~0&d)-hix183vM*~L)9n%z
z@<WVV9xhU#%S`^beR&7bNu9C&>iAA?5N)zB5OxR9SJX3NBystUBZrln-|j$)vSGa2
zY+u=4&Q$G>InWt~8>s_{nUZ^UcoW2ql0Q&T6nkz>Kl?m<fG*G>>S7S7QV0ypIkvBA
z^E&8_k<G#>$J^UNXH-l6C5N<qb=#ejv&aCHlH0nKn*rgn2lh4XefaonK!-><=#qBL
zF8kUxDo${@K8giOMG%nO^=lQ;^mR3c(8fVmxX;_Et8p)63=gS;aUVuBdtSk%=h21J
z(Te4gAjJi;4qU`}Vk3UJ3mY;Bi7r&7E_taS4A{HE+!*SoaKVV+TYG*R0Mb{;<tg&d
z?R6k=)W|nkR|{=NP56RV$JwQmPSJMM#V}GmD&!Wc)TL9iOS%PENN14fE#>Di2i|#r
z7V&_9oIw?K$X?jK$I&6P9$$3o^3+<h31(riX=2h|)M3K~B%8f3iSm_#_B)V7!bk~a
zFV2lHp+pN#K`S)XC+o$)E^Sj70<<{kL*D8#T3HIC19n-PFeFQ<3zf(!a{vH|2}=i9
zs5ujMdDmL_p|?udQDhX5Xe~ZSOsXs)FKMkq&BRo~Rpm(zBwA~F#ZM_>=!))p+@(gH
z!eOY3#h_hM9c}KoBM`0&Ug65NSP);uYjMfaViN4YbJ3XvDdfbrz4U2yLCKstoj~|y
z4D4mq>(o%@Ngmo_2a<b__sAf;o&CM&%(1K5o3Yv|1+Z*6)Wy!$NEN1co)N_S+SP3>
zSLix3r<lx5L=7Sry}aHAkqW(*5;IYE3H@Y8Ey%oe8w&eZdH{6o91`*az9!$lI$Q_`
zUd7?j-WD=`+O@gcFkLu5lSf>-ewkg@jUf!+eEV^yE~$tj&Lgtv`WjW9!p&{?%-k*q
z(&i|X*-C>dN|(M?N9=|+Dz==?+5@QBsLS_H;F^o)CBC|iA7Kt#Ts)F12cXzmsUH`D
z(P<kqoP9|j>BPUA3L=|>{5<41kn7zbZDXIpNG%P>L3>4qQD7?l!>kX^b}B6gF1}!X
z+ol<8D#P~5EVFBNApah9%1I|T^w@>3YMjlrH#)y_luK^DdF9&GE0)`<i=^rP5wQa&
zBzZGF4j$Nf?XP`c5Um1-m4;f*K;$B~*R+v^@lDFu4A$A}4I-5p&bgto<VkkpYwJB4
zlOiF(N2jjsFJ?8FfBBpw+keWwp`KyZSS@J=cQ|!Pu*N3FM@FY-hWoapdkv>+<55E3
z6lL7)z;g(Y7ey1C!0~Nq4Stc~Gl@eEkximxa<q!YO&71w0~Q;xtSuB1CP%;Ic;vf~
zbQO4fQ*i>~rAYZ^%e!fNG{lWZBPIZ$!+~UL`YRnU2b!(OEwY@%Y4sBqDZ9CC+f|qk
z2TrjbJPFleXt++_k~pwzVhrbXrxY!850pX-2+y`#yOq~E7mpMk;EAP4C219d55ApN
zN`IR?)XHwcMHkC=k7H1>KjB%v0N0fj_*Y}1trJ%_dD!dPAkv|eVp4!xXL@#9^;c^y
zN=_Cjr!JQv{ozokR0cC!RXthkY$DD_x>c^hbKtpOa5f<h(N?!HrBY1T9Jsj8fh>q`
zlY?2V@VmV=p4*O2zBIy|{6`!}OsagGBz$ufhTYLkG@f)Uh5U)A3k1h(#~Wm%l(lOi
z@^ei+Bgb-OK~tt>>gpNs^~xW@k;vpL=9`Q)_WFW1=0~EVxP?8uA5(<`&qI2d5<J`5
zYAzJj{#8=`++%DG40^z0Zo{{YwT$F*?9QFGu2GEMRvdN4`@>F+&Z$*g+hOY)6g4Wo
zy71i8ghamu&vd)9TgGrjmwv!oT^vm;tq$qCy6#oVL><UMkc0qx9e8vvZ9IvpVsv-e
z-Q7HNh82)Z!PcM}oD}fd8eHj7i2H_mModLZ;3%1<(w{l-nhZsgNoA;nDvGgx;{yK|
zx+$?9i@uiccUt9Ui+GF(;<Pt)Bh>Y_Iw0f@cp|jZL>D>mTrWHVMn>$-i<-BjvZ)IQ
zx!-|D7sCaT7s)v@cjY~~@7Ilv;eG9t#LrGm4%u7sCpJCjXHBO%lM5l_dBWb>@PFCF
z2oBHqa5_~nnhf>Gtf+6xk1#9S;fg8crjfTvsCw<~-PtL%r$j#KrBZjDH(+;O9!}fZ
zMr8D66uNG**N6dLhE5^55?o9F7jy3&XLohne@}Xk<2b$B#syb_sjiWPC<Y-}5@5&J
z$<^+acF`7j@2;f9u`|YT5<78A;@Igb2o)q$5JGiDR!d0lz4zYRFFn8aXJ*d#F3FSU
zujh{kUM%-}=37pgGkxal(D6x|&p-9C;D1ZI|Nr=})4$lfEg%lfBIpZSlyt6#Y^0k~
z1*?Ta*N1JQlbH}P+98rE=0)h6)0UGxh0%X{r@R*Zwh0O29O+}dcUqzYP2givc_qgX
zCnd2hxrJBv1_S4J!*+O-o)AiqiDp|nBvTMdBd*e28zh-^1YGcbPO-AFT)1&gj+t^U
zB!`TQDN@A7yVphia#W~{z1r_U3IiljDq45H-B)e#TH27|oVFU)IS5_&oA{gVawJ9D
zR`3H)cJUkG1&}iSE{H*cGf410*|mfNFjU0GCbIJca!yP*7x6KPJQDUod>iWGHaQ$l
zID=)|NR|midwDIV*GoE^dfEhww}!TXeWyVfF)V>k1UZs@V!EyJWS=^GNW255Nn82)
zS6pq|&p)L^<B9v7UnMrLT4itRrbm$)%UyHn!~kB?tjt@~02Gm#dM16Gi22}1TU8a5
z10K&Df(puGaEWOew2$6RdR1Kz?TSHuw%(S3U7FORd<W@^cT%U_fy>Oord4tI?q8d0
zJACC}AP(&`R9BT`ETlYUJKAS&=}$Wsu{tS`a06<cWIDE(lug@NaM_X+04{Ee2rcml
z<T&IFo|PSCpOU9m=DrBFvRy^#!&T21RMcK-mbk2<UBrD3J9kO2kPmT2A#(HeRn4t}
zbyN!{XCW0-;Mx!y=ds@Hvim#fGvCM&o6M7HpvHjp7paiDVx*~5<f;<5B$v93JNEWM
zuOUxqb<lw%2E$`3=s>JJ)&#SteSJ5{QF}qg5T`B43-PU+`$Q~mv3Jx5l<%NYL{W)E
zlk7ke`|f7#-<e}0Jd46>URO93lqLBMT~B0RjB$ZWaa<wW_q+p7?sv~v5AD~hWBUni
zVFKe@w8dbMR-{EE<W#{Ke(Tt&ceTN=Fba}ETk>jK+D)zF!PrhLeKYnzyN~p9CHhJ}
zq^+E4!?e=$OJ=x5Acnm=cipRA<opVfnM)$VsFFx3fANjoT(Lw#fY75r83$70?V3C6
zJ$0MbS<a{PQg+xadmTt1zL>H??++6b^-XPeOmChT=NLJsq2{6!bol<Ij7*~!-&`*d
z!b*@Ll)-7!_Xot!P1(1!uOI-c+#;FHUTrvl=Vp}yr>K$lwl6CZ9-p|=P|QN5e6z+>
z${p|P=2bhEnPGd-mW+EaK1rbLF8kIF4<6G_LjICU(8VrzgQvqK&OD;~Szb{=y5E+A
z7?a1|xrkW^s}Ns=YcLc^t{H##{*rz0vDVMo@xFba`l+cG21$USO%g4k5BJyy^Aikq
ze1~&U)K6lwlBEvXhnBrS3=9W`(<b~Cwu>QuxUJ7JKBPEUVFaYtf$Oz8IH=V=vv03w
z#FY+ciy#@O0IE{XrTqX8tBYhk>3DHiB)x~XMdYQPlbCwsNw@D<)<xMAC|KgHO=nq&
z$3oogJ9CCsuD!|m*<`rMxhds8+DE$i<y3ombWWOMT1=(CIq=-HTgVrfw(qKsa;iZH
zm6z}f(3a+g*|C{OskV=H8zkMc$1pmC)82t(57um*=Ug0FE<V=qXX7*k7E>|}LgLA7
z{h2BI?s`=f6_otSdg3Z~AYzqQa-{C}<P)pLXGnJELhC#|F;0|n>u#9rgYo&WO_rUs
z)O}yRt50PoekL~Lw1eWu?faAJE$yC~rtSz=u|32uq`}J@kOJ)ww8>m%cwWwD8$2gC
zQ`3|81hupu?7n?M#0KNqg40$TKo!OGq|V<7w3gAvc6Xaq7Ha7sIF!Zd97v2rd|EC6
zs*u~B^Oe<w<fi1FI9`Ot7?3JWf^{fx!W}G+wzm!Z@?jz>xDqa4f1-}*%WYphCIYDU
zXDq5rkqxp8we@}Ca~?eZ)H8U+(rzh|gh;ai??;rSIZ;T5C@99+_BU;Q1SxZF;9S=<
zq_<7jDp^^PI7HOjLv5uIL6i`P&DLp4WLwF}jLNd^+ktvbiiPF!rD}>o093yXxQLl%
zhnn8(kAwAqD6bAk%4imqLJD==f#)vg5gEr-0O-*0BQpE2GsO<ig$am~&<<P}dU?;{
z89SWg<dhA*NLdH2W1MM*%JPyydwCD)q&-}dUiut9p`-^7IFMNGp~`*Oj<hz2u);c2
z`K7gK3RDUWPmmq0`!X0a1xeZhE0X?$RuO(8aq^MeZcDqjZ69$i63b+iWS3)<w`1*w
z>#knmTx7pMX9bTU#M$v0rMQ+>P%f4zro?X@?*UwxHba$rlGf}*8#LzqhYI^a0D`@N
z)O4loBrABIfl0(r>}2+2d}Iud78gNXb3T_lOC>-KX)KNydo+7ZnbWBq)VE`OMxC~p
zkqpN}YKYhqow9Q(_ld{=k_K0GpWLG1bLX#2Adi&`by=Wnb`-c|iR{@jdBk}lW{-7Z
zdIdRQr`z7xz{$Nx3S8$Tgv~71FytxTr;OIe>m6E23AlK=owj<0uw-UE<Fg({pw+HA
zXlJ_d?_#TBekI3^HfbaqcG=lR7iN(ooGalop_OBHt{b}&12J!gWKU#+11bDeTh@#}
z-;IP6V<POij(aM^7%q6pE;QOOq6iu!Pp2iNtZT2k#xCZ^cVaSde#f9hO`Aa_2z+e6
z<F=6ZPWHxn&hNT6W12-#gGa#<<PE=+lwi<D7K_*c8Zsev{Y+nV_*8koo@kw$9EY4K
zO^6mr|2P;%i7Ee~nvOKN%N8Wtp0w*BBx*LXk!2O+09{*blXAyxcdl8v!MPTryY3ZN
z+2!09_(mX4Tb;|8uBNw%WG_`)(;+xs)LWa}y&{e(TJU7`tVC`3u@n6qNX!iLAWyz{
z34bH9O|@<8ha2id=3)^iAaV>SLLGPtgd@{Y+QOcyF3B#Pne=C!HVx{nFS^oxq}@Ll
zha=7Ood!+e((cP2TikCynolq7-YoTtlCp?;)y9II3S;Vn-1lP*D^}lpmva$YP_cMf
zDPOv>QcHh4KYaZwu5o^f%D{v+&tX3OiH5r2zL6p%=LM;d4kYI}lJA3e*iUx&6R~tH
zC<T!s#SGx_Osf3~rCG3_sy4}!5@0C^?6lQvD@+6}0mDJ!VWx!rbbAKei4^t4gWlRC
z4Tx}xwERpPaip+_3A8wIAk`Dqc9QIH)PA<j`Ic$QtAa{2BnKwoe>*1$&Dzi9DojdE
zn0K!7CE;w*OOD&mH$?4?z@q!kbxpcZ8iN1>3RTyBp`OmdZuq}25z`iT>yGm6f@T<x
z?H9W#kx?eYUBo{*lMLnmk^@&?mUfShqm;{jIj~>qo{8WTOc-KDUCAd0p5w!Qt@Px+
z{c_?9aUN%1=v)XDsQ(})Zokrv_OT^-si~FVw3VL2F$&n_>{qj4BX)tQ(H==QG{lI~
zPa0@~bdM-!zt;6m;hU*Zrhz-q!y8-?5Gqx5k*_Wn+kQO>*G2`e^IKH-ZDaNu*^29D
z+WDP(W<9wN_M6rJ^pPu6Xur3%kd2sWTn_eI)vrW5si;ps*%o+=NPI`eq&<?Y6!2ib
zU0=v;61F&}E$+rCLg-#>otV4h_B(aA2%u72-i0xownThndX<fl&$M|~mB;?O4ZEmu
zNl>_V*PA$wb7Ek>*Wf!j+5!yby84=FBhsLPIN9&lm#PLz9JL`1qpzw$DDd<gZJ_oY
z;>I7eaYWd?v<)(4kr0SO?!Iwe%KS2`<R2zBP0x_A=v>rhQF;HQ7k`u(#)+WRP1xpK
z7W)~NNDljFMBOdeA19ewNg3<<xo*Rq_9yjy)w9tdy~;!tmOtV^;(a>xMQbDiBi5%?
zQ)%2k?RqZOsL?3mQI&UrX!P6q^VDP>9oV1MozxRSl~`6^z#Bw{WBJ-C@ya)_tlFQ~
z*CoR#_-h};hgck>WZiqhfv4n!w5f6dGNa0cr9PlaMhoN$CUnOBA~BhSE4YPoRWh55
zmYPc0gjG$UF`h0v*#5HZ@$KXNnGxXxEx99?j`9^^k@*|5Vt>`R#cXffxzH3$;|xKA
z?b~11`lZcA({##dyOVq9f|efi;Z>r%s!*1xKZ@1Xwo+K5oTcf~e<<AJt+V`J(9qEy
zL{k2n+`>z{cg#D#i%!eP5q1;%+lHDGnFPtTwmDx$1+%|vlfzUgVL(n71!nJILeqBG
znU0)$3nxXKL(I0nuVqU_&QJ%DcF717Zgb$dF);jYo7-;xQ14K-i%xGM+skPS#8bdY
zER9k<hrT)c$65l6^DKS3hZF!oz;oLX8^%XQ#_gZlid^yPE1Zj(i-(au6j67|{!g_<
zzz*|6+KRWf^qnHoV;KAA_M-dH8@Wg-Bl}`4TdC|Y&xh<^8eF*tZQ;1L6`KWKR7D9l
z>?L?rZs>olC*yFtz$#UX>5Q-rTr5Rrf`f|oZ*{wTkhp^&fih%C=#O|ZZ}7wv)`zW|
zYaxu=XaC;K$m0-cwH<fbBzV8z%9p;@{-eyBNSaaSKwi(Nwani<E&*xIxhqxT5lYo^
zEb;w)`_Hy(HcU?{4S&jMiOCmLm;H~D*!ExTu63I=BCLe82o=}C`-vg43xz7TOsn>|
zb1fNttqPOd$6d%pNYx}qox4K{;8_P=lN{N75Eg|O5bkIne<8afTZ;_BibJKXHubPd
zCQE|M{t2xUTR8QOE^-|Mk?JgyhVorMv9*DZ@>`Ii=Cq}7Fox9#nRUcIskK}tjh_E5
zIxQ*ws~TT_dUl?|$M(q=LMA{VnC6&72o4Qek=tf^Z?jLi5F###h|bUC&YO@N>{Dw1
zsi`MT)6|@BAkpD;!-gJcpLQYeXleH?M53Q|E_1&v6PvU|q&G&;>!)8RGk3DpK>KAS
zs)DW@_wKInj*Hg7EIUab{ET;fMt$oc&Y(g~53fQ-1j{^h%o|+HTboNBr^ML^kh9Oc
z(5yZ3Sx%7@ACp9L5;2zGlzmo1j;d^x$7?BLT+t?{9e6Ca_*iB?q&4NBeRf+U(i&Jc
z7Mzy2myDD!JuChwhw_Xz=Lq5o*Ce7oAr2%tXNd6HEFym)lVD3aW1n-OZ5e1V7IIdZ
zCav<DIK#l6+3+bI5mP+uv<dmA2NTmf>~q_KgtU-EHrH3^+)n@Dt|FshXNoOQq&O^b
z&#Je`Q!0Kb%W#qk{0_V&PPFiLSJI(;9-=5xlO;wKoMNBXVVOcrq?#Ub+T#1N#bkt<
zf>9w;Y!PMreSVwk-drBy?X=`JEME<=d4U24?F({OP(5P|d+_*nr=h6%DezQ*LkH~(
z+g`mD|1?C0Ymv*P-RpKr2K%C%!cB9y6r2mqHz+i)FK);a;sICDiEhCp+(b~~m*fk2
zqpHWjMck$tVWTi381-QmXYEV#)mX!KIlluhCG(uLFKe@e{3l-rHb1AW-jhLKasCd3
z>WF=LhXy*1Kyu|dr%m1N&268xuV`47B%dBgN>ja}2pw_YIeg{<!}V9zhnkS3wBqQz
z11S=*mxV%k!ua`B?d9<0IP+Mqg}o_fT&EKQ&pgO2C`?<%w_*%~WIwyUp)z&Hm^Gc<
z6Mh;85)+{vea@cKeNm}3VfA*}Vj_}HWrBzDu^+JK=Cez??;&{Ixk{yJa$eMHU{@4T
zoS1fg=GsMV_{iTZ1bSY5Q4)=OW=fB0ZSqd*tYPZn*z@Z?C_4hB0#MF7ZGp&$c@jTE
z1eRY`W>VQlUXX}9il1=1&d8W?T4J;hEgr&H_QEz=R8i%a%gRWb7KejWDfB<ED-*SD
zR)U&y5u?ZoPV#282&5OaCB0QOpq)!uJl=Bez4qdK0Y`F744eyno??X(6JC-yLk!lW
z^SdTlC#&a~xZ%dLmv#deL?YiIAGYbJ!9W&!S=-g<JQ51?>;q{=nyiBkJk9Ya{3_)?
zLkLHYBEj|g{fpYvqV8wBwJ8=&^=J1Hq`jg&jumi>!g`!3BKd*lWT!V#(^0-iyDA5~
z0~|Qyb-{r)BO`WoLY8)~zGIDZk=Sx`;;6l%5chpc?;QpbIIXb-q+ybOqs6srI%Eo)
zK_Gt!#NHs`CxF`NC~b)yr|sIB9a$hmA<u`+KvCtPA9;hPUKYm|Ft5^tHM>Xcx(?O$
zhPbbC^=MvDW3k6`4qR;D@_QEd^zHgw{;}S6=ciU2j?O{xc0;?`CJx4xXvR2hdAqT#
zvM6wFXs4|<O-Z8Cx0r`h_R6|DD}sj9=Csvj7zV+io1=qAWi@+MoAC9-Ov=PfL%lvS
zPtjcz1wp)?y}HRiBfUw^yC853Mo43dlfA8!$x^P8y{6t!vTb+;wDS%+kXSa<>7Y!f
z7SGsgFSKLwl+qg(FHJ5>l$wGVU87~WoIH^_iJY*w`d`-&Xw%J`oQv@B>#-;m6k?GA
zRAkK5x=Bp@>YU@NZd`G-^E)~vB$ut}A_Q+`4MD3cK~z%_PdbpAPn{(qTtE#1dlbFO
z|7}&lvFe}<)$u-|_ym?Z@Fd=lH0?oobXM0h^)*CQRk)H<F$5|H9$56i;_;W+P2C(`
zlmI}u?Fpx?I;B{P%>GJ)e#SGdX)CvNVs1n}R@DTgu{{Lt${HfQJQMBlz}U?>M>mr=
z=loU=MrV3kX6=^5<5|iL@l$v_K6RgJZQR-xW^@{|I_q5PsioZ;*Ke@f@>VA6(BcK>
zDkmM*DQbBZG4S;b71vJR4_)jz4K-KcL{@B<-QIc~euXfqGV_v{gfxd8c)j?%>`V~j
zLv}}w<c1rrz0vs{$diY`z0cNm$lCesH5ZtCfh1+@?}99js{*sg^PSZJDT4BOMnuw+
z#2T@f%)SHfNVbd{XVMyg%WGXH1}e@@RsoT&5HmwYPpfgeOJZ1X9}qCLxY$EN8T`@f
z+q*zOiLu(yBIqewi3q9CId@&b0&t~Njd(wQ<t2`+TE4Eb?Lxbw_J&5Xd!x9!bU-03
zTE6k#{r&B>q0Re}EIR~*f-~jVmMzAC$EOtepz}*)cS@!#Z<0c8PtrN9)IQso2z325
z*Ev69wz5~+Xwq2`ZEaKc!JtFoTMjxze--%*JPD7^5yRX=X7ycd>cIr*4l;nJBOFLz
zMIJmJanE*l-5vWnCIQ_7qsm!t5KX9T$jT9l?PT{P_DDsHn$=tvwvn&E=yVLz8ynMK
zO|0C!a~Zer0_98+Uw6#j)cW3|G}H9d^yIwLB-voqT^sCcx+fx=QjFVai+M_#O4umY
z7j)Xi!}jJx?xo$g-{Jg@Z>h+Xq#-5r-rKqu)t7{Chn%+D`k}Q2Rpo9%N|-b<&Fn4p
zHOZc0E^;x)3OSIR^~pYg1DMOE?5#Ood}2Gs`#YV>n(EUsp(udMzP9UqZB%)>2sOCm
z4U(qtREi=V!oIFso;<Tggg8Q@y+IUjAPY)oaIs$P3SqQTBqQuuR)kvKBoW+|qmNxD
zoRUcpjfD40KHOYiuG;6!Vj8E<c!Pi#5k)R-H|E7gilJP8?F#32yk87hWJx<zn?Mc`
z_Nk3FYK+=5J(A~=IE|HZm}!P8iIj^iaa&c>;5626r~#g0Ho}B-!f8vrVQr5bRohZu
zPw$^rP>BGdGW<q?0WJbanAZe{9c;C&LmCm3YFRoq%;}Pz3L~#hrQ=>bD}ip5V-aku
zz7?l+&|O9S<!^~vXT8B?$<et(GN37zJjJg)Hl9d-Bc3>Z%BL~1)9&l0KMB?O;FHQg
z2U1%}QmppvBepFko(f2_&hMD`#t&2vQmjRl=Z+xlO|<7(c%g@4v}miXpkf2EFeq@K
zOA@fj>P?^qotHy2G;jy9Y{k;sp2#G7phAg}DQ&90iMjU4cY|QmU3c1wMJW5}0P0M{
zI3mGK*WIE+kuQ&lkG9(S@)HOf6y(F)*!I?gN*RrkAKKu2=PZA<4URZ>Y1Wo@uiS!X
zYcxc(x8*kOZP)n}tI^R%ou)*i*P{RoY$h+sD?-W|IG4HaS2l`egT^pc_i#Uv>*usc
za((3$SKMo}`TQWe>RiUssY?>FPE$`}t_GID6w0oVR;S<(ASpPkUb)J4)T4@zlVeoj
zMYQD&_rCe9wzF=~Ie=n1oYd>BtqvGuIuF>cZn}nNSry*_GnIYArgH9z=}v6T*xn@m
z^EtfRHmzCf{0_89yBJz=C{)3{;OiIc{x&w4kQBEp>&&7#kV4YVyA0J;2u4~(-d=Fc
zR0yr?%5>tTN&-A)Qjt}45|Vv=jZAJAMW@T{nj=fJO^(T3xrW624v^RSqxOzAx5%8D
zItm7PYZEh;1TA^_oo$zth;f3IXi4sxZs0L#3ZIwxLF7>;*R%Ew?F$NJi_L$`X^Sb4
zi4f;Wkl_p;mmA`gdA%*=YE4A?iY_>i7`#Z0zC%E(y{mlz<E0AfJ-oQjY0F(5Oig2u
zp;wF|qn#dTqUJ5r)1%nDorY2ybf$esAPhjxH6BLP$qo1J_F7)0z$w{R*z)mmowx<O
zHnvJbiP}FFF0E7cjrC+^8Q=ng=j2{E1Ytkv!1KvSRn?O+3+c4R_q2VvMfC`r%b0=G
zZ$#J++c&jadJ~%w?73`{1Wya+pnY>&d$v4BVPT7Pn({V85^B)u2sQhbYLgBtw-!e$
zIBkjQY!B&AaTbCyrc?g2_jb=<Mgo;7=is!75)c*8v-h><5tuby8Mvr0Em7MAZ*YkK
z$|6;&d+Fmm@wk0!z8|+K;q;UP!eimIiC7_*&(6(kwr{I0h^}O9yNA8C?T3G9Z#{;h
z_5N<6sO>uHT&9(xKZAiz!8{*GhID&x*7+%h^v>tl2isGsTR(6vQ+MCg8=tTb)xu#R
z1oIVMqmta@z&a3_GG<hKB3}OCnuwvhNa@)_)dhGgh>BR|IZhASx0ejAuqWptZb$<d
z!HSCy+jkV=E)TS9L~WS2%w4H6ubd&4t8d?#W0+R;v4Pjs9X(7A1N%r5Pv)km=#}0O
zr!kN~%MX=i%T~GugFUl&(!Q&4^6Q+R(aT(E_3JVFXyOA!SjL^-MJwd*hvqPEl5URK
z$Lh)MgbQr{tu3roV8@X~Y8rQMP{@%GT*?<BhBt_IhbWYaxT_+>Ui+RlcbGcEHikGu
zTh7X0JC>`yeQ($E3s9G2B(w!#c$!l+RH$Xzk9}W5k@2Y=l*r;ThKTBbtfMuS2qKpK
z{RPRrp8}prQ6F$1F>S4m1LaKZ2O5HI(J4_QSZ8QSE=`y$^!o>^cVV<;4zkYTi4a)X
z49u%kCGLt>OC(Gq=t0}vA)|!UOnFJN7Xq9E&u6e*h;E}!+Ma5IN{7fSe-&*>9cp)r
zSms#>X_MNU2z>d0#fM+wT*S>OAZ4lUJg+eVGi+bMMGAypDu={OAvi(kSYF`TllSek
z2NR(P2_tm4bs>zKnf)Vlkz?f&3}Synmkj2RJ)8l_QVb*(SKmuFr}PPGo)uGK3Nbl6
z)D2rz^&<Pohi!*LxkZPt1MNQa4vA<OfV{Pd50h^v6O<jS`^1*SU+~s@Ym>AsYPA;`
z;!vkjXQY7^S92oKHHom|twWx6?#<<*tds(dDVS@Ahw~DPq7o9)^3ZnJ1p6I$JVPB}
zo)Kd_ZVxA6Am^|LVc@E;jlj#E9jS?wsHXE<x>L>_NbV$Z0Q(eGIoeI^uuxEZG(XKG
zQ`*wGOH$nZiY~QB8ri`=fqRvj>sS|QR2P?a<DJ1i9rekNbx4#}JsDFX(>+<81JAy!
zo}Sq}KV`@BE8BYtP~$>sHW6=jqDCV&+X!(uExm&<*-};57x~vt)|6#IWPFT-KhTyY
z>upoASr6>dx=A>uGV$ouPFu+r-BCb|E-3suvD{9Tjcgyx<eYOA&tyo{avYU{Qm3=q
zV;7#vTstOcrbp1JT&(2t(jKcT%~QJwqqUvR^<1+Z1J5W?!ySS|&1<-^Vs<j(j+AJ7
zJh$8Gd%S*a+G=MS_nX!sOkBlBQo$4re$>vk?S}j(XA$xt{G0>xyg`=fSIjHU9XnS&
ziOAzvM=a>+a#omxH+YJ(vih;Xpl!;qL?g5F^-T$D^3<gbN+vKM(Zl7{p)a&vQ;|zH
zseNqJ)+WhARe}`ZcCqeb{n@-q$UpD21!<S4Ax@uvMcNm-KIqVM^-Ej{|FHuRMqd5^
zTU+P;UaI$E2t}d-L6D$nD+XJ!Ymz3y1}w?=QF|hvQ7#c?1FFZfJ7^Q;rS(<f%MaCy
zll3ap@~F4Arl2bLezCeL=BtPt$+4lsIFOjG2ufsXPA~4W%hjgkCFG2|A9C7~lE<@(
z3y>vbsP--ue3@%HP(i*tY!ME;uv6jxWL_*Dv?uEsQCQK%tPL$^ca|;%UV;OK-$4@c
zka;I^xBYNV1aTX)Dz>@RX-LU#`}9u2K;V>5b*rQ5A{vTtM@|VK(G@~GD%9gK`;j(0
z@K(pE$nDuC6%OaXMHo4NZUQ5p{b;_NRDqE(&K%@W$LTmiW8Q&C5a#&!TkXf%)={};
z=c*7%WNev+XFpyY(gzm;a?o3w=2&Qjo@YPN1{syEwdZMRt0%1y^m@#GvTNH1w+W8s
zv?X1?Ye4F<{Zu#F#~G(mc(uFoK$25WKH$J};*o2W?sNgxD&}?2e!6wx^2cxqemCZC
zS~5t6TmU8_W8{CP>q7=DNIa%3$u$0j!Stm4Y`dPLaxc{YxDE^pqQXa2GB{(ruxlbx
z{#<oRiz78P-1?HS9LWE`&9p2Frxxtzt2cO|z#z#8dz`lVS~&E@ANi)VJPEMp?HB56
zDU+mTm}R4Y#GI?dsj}}+BW(U+vYge7W5T(pSF5H99Yvf$^9gH^3iSR`8#rgWj-!`x
zom+zlPO%S}`+vFJvsx7a2hMecBGo|?@=W|nw@fly%bKA81L|2gkW^XVtVAu>eYyLk
zrpvVmiFP1qR=MfsP4;WuqCL!uWCGdmwB=YrXAfkm-NA1t+pl-e$e^N={*F3r$scb5
zPv;0X`;B^u$h3?O=TMoJm2L7NyekgGDs^R%>^B?sPEtc_+PMtud<er!NbSMyprZC$
z^`S(L&^GzpFi8!#W|5-IQY7bZnVBZS@wambOS?ndm{Ex+r!9ufE97lKLX_u*+&ku%
z;_&u6ZCbGSmNPbbtjz*bb>Ok2DhjD13`FgIHz@<T*IS(50mp+RTLWQG=k52}uu7HD
zX(d!qT-@hC@@*(S_|9OEm+beeEqfS7LT_zp!Hs$|_6Kdlk|_-X=NvCu>TL^r0UOie
zgZ78HWQ-0bJ3<;h4Y^t8sb)Mz9p$#BoEoDfKBpnxqN$dBPsKg{I7c`pL4n^zyE9iZ
zD&`S>bJG5#;pU*<+v!}zdss~AOp*Eirzx`ywr_EMCKyVT!tgo!vwF2yflT<CsN}mq
zgwDM!TgZj}^JT~&(}l#GtYe))X2A2+hy!w0QQ*k_q9m|XsAa{IsCis+v?5aYt0Tj?
zE3hL!o!QJXgJ``jES<K$tT!KAdKA=@s&>I?D|v7jY!-WvcSg2NZ2P}z(|%unW*gcc
z*Ba19#TwdQw+=HG>D@l-w8aesF>klO8O}MZEtyyZoH-D2rEB=d{<c|$wnMA&adMa+
zO-1i@A|X=|VKlfw?C%Over+}a%sYrK5WVuMt5#fPf8V;jd3q8bgL574W8gAt@mEXB
zVO{nQ&E7KM{8ZOBh-274=G##3Mx5U@y|VgCp2R{Qk%FmG|5LqvXe}&{y#q<IcMm0c
z?f=w#wTv{`##>w3;InYoKX>07Mi8W#CQNA#yfdR>Oc&+-m-cE_SQCc$>9lD)OS`YT
z`da(fZU(eHLMsPHVQt<ZS>;5QqlgNCK>S-Hb88iJ=v*erp9-D|uOQF#-&<c|JCBdR
z8fZ$<XJnq^|EMTp|4|<=KVR^+jB08S9qgr&#0q?gj@W<JGqPZ)%t91n(0+9w8A-Nk
zX1X^zf(KJG@Lz4o`!n+x>745%`gh<QfVi4n%<TnFVD%+zKIpU+o^_{wg8t-V*vAzl
zH=c-!?8y=wpAHcpbudtB`p0)~lSZg*q+}@x1V9Rdaz(SgW2Z>lCtR$hTD^XQb5WE^
z=gNi=8>EwC_Y-RfNu?DZf5>TzyW8$c&0p%kC)M3y2S{<rX{(JOB9#WDMB-1byJh27
zs>wr$QnU$u`T9YOFP~C_C{>HCC*5Cmi+GO%5Bdeoh8%!{Ov$I#GuR8*!ijdob2Y2A
z58*<~q=}N0;L4G$_GuTBN=e>PfkJp4e<`5J`=dMv3v<}HE4gEc$>{}@8YyM+C`9Aa
zF9!3t{6WMB)F0<s(kD4nI(E@MqYVu7lI;i!A0LtS1gT|;f#)vXu;vYS*=N=;OLzfs
zA9LDbTGK<6I$}=iXVu-J4azXPfHX^6bXUwx;S#bsr*9y-ZlB$KL1lT9Inb7yYOo#E
zOC>X&k%Ndb8YC=onVlSL8=th#sd-EE&C0&&mLjwjpBXZjb62bw;h$1kA&3X<nbkS*
z9}zqBuS8<4C>{pjts|&_6kV~;ZTO0uAjYS{jWGodB>OZoHh#Z7>tgN5(YeXUmZBl%
zfm(3EvULJTN_1HJ(8)h9AK%1zUIgHBCHj6JVuO8teKNe6d~A3^HjpHGYd3DRFUWOQ
z9O(V10?u`DSd|xua-8;s`4L12MVxRIcQiTc+~OJgqUz@Vw0xy25P0Fe*i>kZ_Qmy#
zmQ@zJ2zp$$SAaJY5rvoL?MqtEq!4od8D^TcB=c1XfS|}Pt=Fb9B1Wzl80m(|R)N=o
zo?(lH>T6$?^G3{7&-tB;$B`e~$EL?e?aR9-4x;8tA$h=CTVz5qAo!GhMZ-4OBf-pE
z#)rbH1GQLy-h5?sPSN>_H_+KJ((F}_62GJak7u-KEzkI?sx6Hp3-UTRZHYB&$#Ax3
zcW+Hoy6goqNMJ^e3dF#xCt)gseLZT=X|ubQfY#oma~<~}^(r(-f9<*5mW5U_<8tzc
zRm=Z7L>gn7pRcHALiMCg=Sk)E1fo9Ec8ZyJUQRh>UhZ;!CpBNLupq_nv*))d9fO7S
zoy&NI(uEJ$3%XAYbMqpE!-3R*6)&K!oVFKMo9LBHc^3$?pe>1F1Vv~SRn8-E1%=<P
zY@`+OU4A2$iYGJFrAy^-?uxHbGE$yWk!E{Qw<MV)L!~(|CktPR+X2^3?ad6v?Zs^=
zMhVWr{K<7H{t5!0vX?X*S9zL=F_N{NrlgMHxYkxs?~b#LBr-f|FKvs7fQ~TC-$%UK
zl-4-DMaQ8`%-hTI%jBtNS`61Q5~(W)8M18I%iA2txX)J<=OT`dxj&Xm8r>e}uCShc
z4AlD=YqR2FI00^Yz4NQ+zL02n%JfwYiOF4_aIWIpNeIG-U{~iX%_)kJ--#sQAjp11
za+1B`=@4X>^k80aAn{Jnkd=}8dcdy9ua3{mZAP);s=k^i%t_sLZF^t5DU<fYnD0R1
z?qt=>!s33rE~zPIZ-@sMfuNuepzCXyB$C$2P9hbWI%wX02Oh7JCaCE?i7WnweBI{x
zh%#<-*Ts(Lq#ojz?Z$=_w<#iAo>8YsOTeM5${5>juWTrYLrJII?7`5RJ(2u!$$=*g
zA>d?SuWDE8r(q;q1%k=hGz`#X*dPG5MSFGiykq7VM`?=%5*wHAhA6G%iuRg<L<Sw>
zROQtO)^;E@{W5NcfN!sD(~nJFyh2h>$WRU>?v}i(U>sQ$Uzekv?e7AK`#tBnwsah7
z2GXij#;_dMS2uX{GK0zWjfh0U5CS~Df%Puf1MU-BSuElGRLawT?{!)VsX=J^t@^Sl
z%&fXt0jb1qNa&v91W}j)&B;~nAY^<-dC{vIt0%E$Y%SsDG=#R9@(4tW`Ac_<C~pOF
z+D&aPBOpo2S`?OH{u?B{M-^DflLu|h#qg(Wo8IaC%tfEVzG*kt%QEAdldRa=@J+<O
z29XdXmsl5LX}8n_iVn&!1S;`YMP)RNM;v&5;Z5r{+-|pa-4z|oiH^dB4kT$<g}r77
zZg<v{-PZaBEz-N?(vC@SAVF(BOwV)@$Ls4(@nS?4{#KHm18G{|4r1vj&uq6hat29I
z;yJ~awQ}q~q~Vasl`+R%@?pN}&2~q7Il`_i3My<BB9#M4`GiWSqvKm`ZMzeqET+>j
zB?l5XC9)b3Oc2OBvrP`OJfj(-Vztp9Ekv*~TUXPxERZLrF{?{maNzlFOnXWtm!)fc
z_YB9xcQ_>GlhPRQ)cYv>uIK`LLv=lwcCJz+3weX&rq!0q4<EM;^<fDF+Khx1I&H<5
z(qN*$R0In+hF9BI-)TO7;Y7G{9Z2pE&gu&LhHdKJpTW~77xy}CiCY^ddOPf{cC~)P
znvKqtBEhx(+PKOe9L(*oyDx?&OgAV=0T9J?WZ<->p3bqF<0I2~fbl)`MYuTInHo}%
z1KP4*o3ZK2+i7o10u70!zzG*(;}n12Zf|PqAM&Xp(<<)Ofn>K=Q?=Q?rag1T^SpkK
zOiT~#&E4l!!v%JZfI?fY6v@0q0nXaJ-3-g|l0jt2je>7F@FI~H!y86omc6B+13R1S
zB4P#fq-qESsd3;%xcA{XP*}1|qC_a#Tbm@ZaZ+`A>T*L=*c{l`)^llUmCZ-d9Aa!h
zvfEoH<~2|DbxDD4XPx+&(#@?Zmu|f_Yq4vZ=UZjLfW%NBG{}(g*EZMPa`sV(plrO#
zQXfKeW-fK2l@TRdh>g^HglMw0E5DjH?OkPw&fy%g(Ogu5rJ>ba1Qs_6M>HttgQmdf
zz*y_o3nlO>pkMwe+LjUT%w8hsCL&aH;z6qFTM`jA&QUUP#<@x)r1BuqM;uaxNVYX!
zPC5q5T@H5I;^m4M&<CYME-M($Xcaj*R^3ujeuN$V9&_5{Nx$+fud?w(J3@&D&hL_?
zGTY_?dbf7{1-q}UT#r-idxY!K5UU`uR(2=C4Q*T76gbmo!<9!%zEI4AzMW@dN1%&M
zG=z=k)37gg0<98ncHp_eX8K#PbmIh^%(YnBef@PeI2W-NjVWf)gX7U(9hEoL4SInn
z05QD<AGYm*ZN@6vbl1UQC9{m21Cl#taUWty*feZ=`zo%e&TR#&RZT-e#J#~Kf9iaZ
z%qYtw8CkjBR_~08mMn!V8nhMdA~e-zI_ANkF$i*90)<pKaovZsA_Te*sxK&XY=a0d
zWrBytMGfIPN<pHIQgu{?BabqkZC~0rM?mQ==Q<~RBbvbM%-p2Sb)%Ixk$1-%js|0-
zCZ0EV+H<AlC@N4jW+HCecC<GRNFv5bGw-y>!mzc+!B4icCL@<n<oTe}rW<+_6k@Sm
zZNWF*wZXYa@_<G^J!13qa0q~`4uLV=+L{y&=dt_SJFdOvM&}}yRB|=hKL`W|?d`eQ
zvMAyU<9|n-#;}K9cf$(%`uwoO1ew>ljCaU7rj4(|Qr}UFBoZmvRv{@#LK{RjIPrc3
z!$#bgy|e3!93NVqLrz<ckqXQGS^I`Y2XIAB67{Ppc1~m7vqr0C-XxhGG2ELmE{`}(
zu{9L7QKwFWILBEwzOs1O-gWV**E7dUyI=JG(Lc@`3rVHQX{>b=i#<@I6x&cCCJ6(r
zEzu@yruOc(OdB??bS`3wHC?hCtBTw=wtX6gAQ6435?Xa2MU|DHpwuYRTj64g;fFTc
zdm5fkOz%{2AFfmOyN^tG5;&+F_D#9mgSipnZ;Pg;x+x;jBDUgnyp&oA_RZBD2}YuR
zN<S0zdf0*FHX5KxN4WmCG}5_RXRN|UXiyJyr0tw4!ExB|d%JmxqNcL)J8dzm$!?^^
zi^fx-&ED7DovJr}3#TpaM)1ZMA^j{ZiAU^Plh!}y9dC7h#$2VLT69i!4VXltt9@IW
zo9Q@H{eGAcO_@k1i|^b-9vv{S_qPQjjMlkG61G_;;SVfZKINsKC04Js#aYa;56;*J
z+f#R}+qluWYKtC0Kom5z4^^8g0_e1u(^$e<TbP4+6UWQihwHv(!Z}CSWNVYTPZ>b+
zX?}aRGnDcfOgo8byyu_;k3Pr)s)!is`Pp}LeK?J&22B;|i?*0XnGfWLWVU%jSGUK$
zGf_?k5B@voT<84Ej$^O<NWM|QKq|#DJq}XR7OU6J=2h^1cv0=U+L92NLOD%UZlOs^
z$Qw7>N2_zS-0YW2PFuW6RtO3HRD)#WDT?-EPlHJ8QGyU|BL@-#kH{&8qHweC?)KFg
z<id~wn8smLGVq)%GCNggYA?35?`g<S*2(lF#b}(C>J5@5&nr8Y_dof7eQzUVZ|G4z
z-?_{=g;L2(L+Gopy)iBOzW)U-3o*Fvz;ljondxn|?{A-_ThhbD_Cz#}IFJ;s`ny0@
zXgbBn><2DBbp<oZ2MOJ*UZ!qKM0+FVi7?K+)zQe+fo&;W<G~1_La?+Se450!Ox@%8
zA$R7km=~G+wfm4S_zU^8-33vyp_tTTVFHq?8c1>IS-j$81KFN7XtEId&Q;q&tVm3u
z?d?`p#~p?q=fgIgugaRE8t&`J-q5g<ytUC<NweG)^Q3o|WXNo#T7m<-n`l{KAM8eW
zLN@lG+p^v*IFJxqg;O{^U>6E_Nn547I^n6c{nc|ZEoFWnWjSpz_X=T`N9fFd^Zq<k
z-&9bxgnhYFk`V)`EvtSC!-dmAGVOy$>_C2vGj%Dw&SX+X!D*8)qB6_Vb}-RZE>m$4
zE@Q4$lUK*j5OHOPE<W{4?qAwHP7qYM$0Vmi@Bv7ixcpH(frs;VZ)Bcg!R@1~DI|c(
zG1i;3hpVGZ4s^U?%nN^+RtFx46GfaaqhiDKAmfNSN0L$yu#pvB7y(VWuUR9l0jHMQ
z(PRgcbGw|MF)2xvL+s7zN>sN;+NAORtW&8}#t0I}CUVSyr#iQzH$sH_yd7%~L`e{(
zU$OaM{dMfU15a%Tw#wj!-WOimC4JiQ?k&e5xuQ_=2upwn%d!Jc5omgJ6wT;F+b|4L
zDJTkTj=(Diq6OSZS-^pv%yGZ--nTeE^995QDde~DG8WCbh$&0EfL7t0ovIFr(lo1Q
z3YmdI4m_R_oDHR}+G7nBoAC@T1jiw^07xy$IYq@Ww+HQX+uE`Y&QR>yX-O**rhbAT
z?eTVf4+($3xeh##6ssaYkII6C--yz`1qn+5y@?~d!<7ys)|<Y^mXofmlau5Z)|nw^
zYk6XtX-g%LP(j`j=00Sf3(j5A9dDsi=L(sprgu2MV_3PX0;TMH!g1!j$@!fG`c|2h
zFErGF2_jlk6gbx;1n{G8wu{}A3Wp?g9dg=g50g*IWHymnzq!j6+Qdc;k2q`4X-jEO
ziROdtJrd0?b>t_`u?l{Zr4!^jm6AkCr*l_)OddIr3JNy|VN!e)^Jh=waNz!rE*)^L
zawp(DKnNPMA8OxKP5kju`A2Eeyxnm1RkoNI_v+Wa@XA*?7ljYve34gSm%HJI!3^?p
z&}oYvNmq~+Q2Dx%sVn9SLtSbZ(i<JcY~WnSdv#ox7_?#vp6tG*5|G#(pM1BsHc34Y
zO*wDuhwED{A43UavN?xLx%99D5A4%v3reZgahhTjr|qe3ky0iW+Y#PZte^u=yv01C
zg%b;idV}^OZC6sEq(9+Y)&g{A5Iv1b{AewsrbCej|IZb2xJ#41YTX;{$GSN^Dz79A
z<(ShJy-_lvY$rMkCMcZ!c-}30%k1n#Uj-zbHqnz1SnMbA&6DF>ou7sJ>BEs*@h68f
z4U>T~EI#hE#n(jaG#_0sqy1FZw+OqJ)Z$&6qB+1i?6mgN!!Ly`kR9r@saH$8uYJXh
z_A_k~2O!>*bD8UejH(>&^D$LQ!}{5J6YL8OCyML%u<blzP@diXbKQCl5Bz3xE4URZ
za>{|%n3RfyN*1QX2XGMne0>4oGt6BV7cV$%N!8Y=7WBY=p`OkNQPD32!zt<&NbQQV
zV*lXgWA=-6_dc!;LMis~Uj2T^fds3DRNimDRJS$K7#Sigz*z^P&#!JLp#5^ANvmgP
zCY-AnN@Ik|7T0z>X1`LMNuM8*EYyg$oc6nB#_d-dvLbloEJ#<2vvY!U0gqQNGr%yv
zG8+C`y;{6Oz8tM+Yf}V3SrY8myUr=PxuOlVR~<;s6suBkN%kAnA7PquQ6Gm^=y!8~
z0I%7lPAEMkpx%D7zM#F3K|Iod<dg=o%!rx3Cs9OI#E|@v9=G4>Cg~g;Ne6x`IBmI!
zp&fDNcFfMh;?rRDN&D?aH#Z`4jp2u)AwHE^Xkx$Z46wAc26wUFDLGW87J?FF&Semf
zi%|svrc5cB^SE<w3jCW@%#QSUYN*-oR=;vR^iYIK%0jHD9S0ukC8N0Pma-kowY*@z
zmpzVf<2xZ#PD2eb&trU)4WNE2$b$NC_WRw^9}U~xc_e2!ijF%F>2d3K*&igI9L!Ne
zzkXxM*=B#(_~)Bn?fkB7e(hEEM|nZO=86^0Md9sG64zoMw?FQd85;@%Ai%?Ei~eRH
zTn5DIZkai#xnO_Nh7RT{4n|r|+OmPflj+>m-4FqxAWtfV?oV@f-PD`yZKAxUa~=1K
zF%(&4Kb;qWP^BXKv+A9yJ?3Gsd!4p;rs^GnL_)%_S}H=^{=5;8wGz^2ooh)IR2zmx
z*<a)vsVv?nc0eB1QKxN%?qH0tL5)WS8%#B@Zk^myx$BTYo$}WmC1QV>B3XY6K@o)3
zj?6nPxdNM3-DZE4OS59d_18MTV*$j9lyW9Mo>soUZhIm^UXh~U0<=jRSlWH_?e;hI
zWD9@L!Ns}ffYh$0PHwQbN9}Lh<UnOIb+${i66N#;7cHqYS%%t08QR~~wn<t8Rgh?f
z{ass3yIGq{fh<zBxHrBp4zCcYN3nDMeYc>)snj#23OewlaypBJ$y8>N>`@2oA99&c
zaOB|Nst`%h*iqhteuShDRN)`%+e#wSj@*k~jke?*BNTZigzum79_meQ=g1j~;c}i?
zwXv}KAq$c%&fEX#-lcCT+VY&YHXYz7iH6bHKR2`wuDndfav(7ixw<HoXmS72h{|Yx
zY@Ps*_&kl0HMaN1?O*GeG?>s6wmWSx>U45Sq?Lt34s>m$<MwZj8QqGZ(z!|s1Z~nP
z=?tKZsQ>r&ZBiuh5aqXN)VHtaM2-Jwt03CV=p<!e&1Bp_r1$hlyO+oEKf4aK5IZ4_
zEUW3r23{x_u{}z?wf}0Jym@Y9%(-fJu(w1-_L(ho(*@&}U1-14wrs&gO(e0&a-zQ%
z?c)}T>L)wVGr0|Tept_syHbN7mH$5b`0kx@V5c8CqJx+$0_YrgO<?+&qtrKFrp$0X
zxxC0u+kohr4!qD<?3ZG9VdB-%vY)t678nlGQs%_KX-WNJz<D8}NCW$%oB&E5jLc1J
zXRfgT(3Y<Rd+-`SW=M9CTxXx$L9|*ajme(;xC0TnfsRd$+ox2YWvx|6P}oFb0}FW$
zI`E`T8;D^2)P*u<A)zVGIN~B9)UYHX4HOI}Cykcth;y%&aB256rbRj{Pur)}5~{SR
z(w#0Y9&p-{Fb9nbRuS1N@@3<Fv`;TMT=Gc@+{V|Ix%eHps2-{C<MtW3kW`c0Gd1dU
zwX}Oxzc*)}S?|#onvFIL{R~>(U*Or_5rV^d+vTaX&sr$$Yjb~8+N9GEcq!yZsDd0^
zvd>;90&?FRiCP2ay6_`{dc)KhGK4*&x<F)trjG&V29!1RB&J5-WkMttccwz-K4&4>
zOiD*7j0|<O=9A7S{^iKVb2*;1Dql|#PR8@hg%AwxUgiAKw%$hA*5`I(>M|*uwpbL%
zH<P{}f`>?k*Rk?C>{<Ev*8XPaXPXjkRf6C-`@F7AmH?ariw`<&^~#7dQt5!eny5>Q
zdGqsY+LkB#1{HH4>g&qY>+K8b#Tfu4a(uyQ3(2mPHxt_t(N`dfOcX3QUHZcA^AX;)
z$O(AdC7WGX6dZW28k<M1rnE;Z!2gTd4tUi|uXHYA6d?jKCAfcijf9ExK7Vn&M#fLo
z){A6xY#aw1h=|O}nXPl|gfFT4$m~MQp(AuOZ1fH!zL30DJ65L(NW#|NL-wT&Nj8sf
znVT6&DovxNXk=#2zO3FH+@3dgsrC+JnY5h-=(JC0Pw4xIE%xQz2xO(v6v@9x2&w~#
zl}0Gz?u-HWirjtFNttqfCnnHOc}6&F-oCPE&GaZieq1C?(=kcJ1&j~utLhoervfG)
z!)`!ZEQLfW<y;Y0ko3i#-3axL{?@I|Rc*8*Focw~(?R^4wq$GPN1TgVGNvSOLTrn(
z;rM+HLiyZo!jB8TN#GOr=+rF-E&?+#zInv1sCzSABA6c^P762&J#TPD^hoVjKAKJ=
zRR$uWzCEw|*05?$Aqh|Hb>OL`NQDD^E@Fqq*NCA=lhM3s&u`pf-HdXxu*cF8Qz4ND
zwTpX8yRjE^6AYgT6x!>w<pf7$Yt~zmIcl%HutCyz1pyYF!eih-VsTk0DOoZO+Lbk$
zG$v=TrO=`rBMv-y$Mvtc-d@x?jTMQMBm`($>XnKwq0R*O^5TV%B&O%6nSkKMW!r9Y
zToJhkTNqk*?IraJC6;JaA4Pv|Z3muwAmRoHb+wn)eNyTVpxQskh-iy<Yg-Ayba16?
z_EJ72xV@}nb`ob(kcHmB;QmM2v2(A4$W{W^US9LayJHinOm`3n2a<c>9@zJ6<cY0l
zO_L<>Z6)=3;52DJzu+~mwX2e5zu`9LS8Q@3(XMVAVH=d)yQ{l7=+a(MW6hwxLy)LT
zPFv0+<J%LXjH=J@upqH^O(QCtAD*=$n0y+ACQG}O0dV<4&@+M``R|lnyAaiRm;T#H
z&RqX5|2j{^S~seMf77mO$QGJ}x*Ijx5jub*;@m-)<Ms8Zuv(#zhfyM0TLu|3%P^1|
za-R~zK5%{)YH>iTb|^~5jrF;<cRM4Q1L-h9qH~y<v>(7VdHP>Dd=l&kGc9+?nA9M`
z8_5G`-(q(N@~^6QYIn)>q?!^Tt_L4oaNy}Gn3&U<QLpY6NTmj4D3Y+8H7E}ei~w9~
zPcXi{rj2tj?fg#V<)*%B_StJ|+;M_@kmyI8wh)j;wvF5Ch9gA)6LXQdP^Qu_r8@AS
z4D#!W@+k5qL@xX4#JG*KeU3tMuGD+dez#M}+g5h3m9IJ-z{{*T4qUwNwy7ON_1UWI
zG<+|@37u;}DOjP?)>b!}iz=9u3m!NPi42_2WjD2H8|%xx&1LjA<rG;9<c`JrW@{P-
z1of29^PtlfcN5AMIrNcM9lp_<YXDGm6vGG=hZ(6zrbfTl8$3>yhN7QG9+E9cwht62
zyQOG_7-hvW@|K}fCsp^<ez;dAeRv?I7@d!OYc3GSS4}#<6D_F5KR08ywH`yANG^{g
zyXb>=kWk~oxDH&w*ROB1VR{Dz%eWHa+<5g3c6<K3jlf@iYRXY1dRu#TM>mVI)&%wA
zT{0K~(S6LtMJVFhZkC24lOCoSJGywtf#+LaPmGz}S$8i}dBJ@h7@|DSM;v&qT`2v0
zK@uD3mA0;}9r!73hPBIS>o|eJ)aCDQ3a|F)`i3RFU47@G`QqhdCH(}E>kS=Z%~Lmx
z^Mq-Lsk!c2+mK6e(`_3!IlqN(bCBnrZLCfv-HM_xbf4w)1`&ympg-+f#Q5X3scp(z
z$43e7a;_7BH>_J_chw7JGn4D1D<<JUf@X;yX-3f=?C$pHbt|rQE*hDFQZ}ZmTnjPf
zBX&=>Ji}bC5PsQbvbF+`55&=Pe0mRL>W$SM=~{|3I?T3@Vtd{oWeNq&(g=FV-n7s~
z?r3tQoa&T2n$uOaxraD7n6$5HLk-a-h%{e-<T#K-l#LwkX>VROF0tU|e^SsD%C!S;
zhyuB)5fS%!-tKMFv!S=u>w+E;EA*BQg;EZuIEm91^S4a6V-_wdMi-KC!rt0_r;DTq
z@m0<YNIWUB2m?@6j1bPyBM~wV+1KXK*X$Z0wqDjl8sdRw1}1Zoq3;XQeh!`D>zbSw
zwosuEIt!FWBvCfWTrlp+5UK7Mg^20uwRuH0SE{NqZzHt9foO%7c2lx%I?iSt*ybj8
zshS1*a^SSYe2PgZXog@>W?I&?k;L4Y>5Rs5t~0B9iUS)E*~jmf7CG8x0~A&DWgQVK
z4FaRBu*4f&dP1a7uxf&+^sH2|emyTlm%vT#A8;kCiA_H0z-7LE+3Rde?&H-wDC?0P
zIO;S;_k#nobjk$DG=!~<c<MZfi_T@;C^b)HNWBqr7;AWQ>&-Vg7eYpY!<8j7-avCx
zlippdp-|6#PwRHB*)s2-15cR)%afcdiiF=*T@Ww8U4lO$D-<9129fknlr3@w&oDg)
zY@*%+oiB^>es67YngCOd)Mk@)p9Usw#Cx1JjYY{xv$HnU^;k;`*DCJV%&HBpO)10j
zaU}Onr@`eq-4=Q|aVWDN#zP05(gzt7)P%I{ZCm$tAwF^y^C9P_VrEXlo8FdpT;9V7
zE;(0;t|YV&zZfM0X4*T~O^oj#&Wr0<#3aC~Fd<om8ecZ(mNyL1Xl%+d5TR?3oNsYm
zc~0a+o~;o{-E5`(Wb>-RQ4U<P($enNkK0@~duhTJnb$TeN4+<A^7-sv+H^9`Xh$2#
zc1m+$oaVabL-{Lm0G+m-P4wS9PU$tOLybC3aT~iQJX`xvQFm2W$hF~7?Ik6gbmYQ;
zCp}%=+tRoB#{Cops-;2Eb{doTp+l*z>pnK6d^9proB~tt&sU7EA5$<eSK5M;EA8#=
z)sVKHbCs-Y^{s2{>)QZ{(ZRasT*p#L?TxDJ@d+}E>>cf~>#n=bxrhNuT288X$==zm
zGBH#1NLlAGm);;voK(ha7NhuvZvE4WqaCDubkTuiJR7M-RPZE}$LroYwK8^A;tCMq
z@Z}HCmk&Uh-_<BH5~JdMyBJn_6=!#7?>TqXT0RW%g{Uk(Xb+?~@q!mz;rz^vv7K^P
z5S4eg>#bzLvNgCaRyj>}Ewafs=CfnHZO+e{9?XPn%10JY+Iwo-D1ZBibCp{n$ceVU
zd@FLXeN#cS0!oCs1aA|eI^{re;jX#yYWwD#ADw6$=A(hrBD~)=pd`+>ba*fFtSCHL
zO@?QQHMr1FmNuQa+3aO#&-UJKu61aioSSJ9KJGvyaJ}QLFShs9M`~NkkHNN<N~Vpv
z&w;DIclPIIrtMq1HNm^blvGj*k1VqVIClkxO9%spVsdhmL?7+j8dj_%25)|VmQBkQ
zT6<VnM0ksnq~-)Qjf9adD%IYf%R=H+kEp#k<<Du;wi9J#@;{JsNcE|{*Y6c8IGg)}
zwV=%vM);l67E2iHh`v66W~)%C6ZWCH8xfPS#)-k++7cUOwK{DdZVS4G)3mtA{Xa5&
z|Ga&Bdj%#XRrmoLytPSOmxzyD*1n?=AKAJm`m(psR9pCXIPYYt4c=?t*}e2cP|Jf$
zr<}IjXG$lNJHkFvD^=7mNR+fP1Pu9O0vk>_ca?Nh<V>y$L|B6R<<7M4s$OISsDeTi
z`hbzuf#et8HrK;DMt<5yyP1q*1PWU@`N@H3lc^%*)PNFa_OWgV3n33&TztrBtA&t-
zReMQszlxvu?rtiSjwK&~JQ90gdq8scuEJYG_KAH@Zf~;X<*pkz*Rh40jE`^HzPH*W
z6shVsYw4}6=}=mf?`zK@sHmveIjlFWO}eOT(+WBN{zei=C4JDjD%_V(Np=PWU#Qq^
z><0URf=jEA&iaT9v~c}AtLT^m&mJ$HYPJCT!6r@JM2^~&a~Y;d#j=S6>xjDCEUNV^
zvk2tTc4xQXBD2#2=Smy?HCMdQ_T=uE;uLD<cM>phYsRR;`--eRW_$Ai&V6-$#!V^a
zbuv2}ZC`b^g&1%QsnZsBYm3MYFWa?f;z4_`y$95gg9}ndUnM30DI&fAAxp2>{<f1=
zD`<qPnwFE$Volf)d#Kv<c4C=uiX3p-a{9p^89#O2t`Z(~wg(Gt2O2y|K_tl(t|>?v
zv#t-}3N?|KtMhjl7vEIP--G{cTP^ps37z!B5$CRD)0)Pnl(-Mo_oXwcXBQDhRnSoy
ztIjY6p8RTu4jG)c!%18?3zj$(E|$GEU3U>zZV$Jq4c%RzQYwzN6_A<`S*@|61t*YC
zCc0%u61PUfYBg{!;}d#C1aw3cm~^tE-6yj4a}p$SGaN_|Z`L~I>>Ok%*Ih;&ncPmq
zU*g{Ekq%?d%ET{xb=qpnh#F`$bLN#=EVs|Gu1_*5rO_}WIEN5L+iC7)C_KD)#Ev(7
zQ34UTfw1bdCFi{b+uez}A>+N`A*U@(KE16vWBO$4R~g?Fqn9zmjEfu*;7I~kt3a~!
zu}9k+1$jUt4I3QNg98ctZD~u?WwKN4{Rp7wW%ORB_;fW|9eC}E$ZL(cc&r<jRw0Ef
z+HzdG<|piQH!ht6B?bAo(^gyZVYrC&icD!aXpdJH#2I9-%YqB4?WAQm?!eQ0qqwCW
z;Z?%PGwpk;Z(ZkHh_Gy(-DYR&Woa!9g2~z+fWU?2!h!47>#w`c&UH&FS)OgkB8Bwt
zKw{9bkw5_0fSvDN+(Kn__CaoNkq_W8J_ri_ndZfjrTIelosfX=l_*>Vrs6<KFqHLg
z#x8afdoiRaj#Tnt+cEVIqO)KN9kstmv>{i~+on3JKF6hwJja=U2!JAXoz$k$QnG=&
z0_i2OD1|SJo=4P#L`Hj}do!g$7}F6*D{Z+N)12t6a)7Wn`=Pp#OeWbNqUsrKg_+X`
zxk`UoY^aTGb93LhOy9@Iw4$aJr+GP7=ap~2*ZHY^P|<X@)F^Qep1~$)6haQ^<kP9k
zKUqDI(nZY^=|)al%0sWeb<KMF;d)k1KY42v%^#`K4kXpRyI0v$-Fq%@S5{A>P4KI3
zxatP`k>vW6vBSea7OB&qkI1o(<nyE5%u%+VEuvsQ+H^;6Y8#>U_G2~ol4``Fq_vcX
zPZFyG*T*>0;ePw^t}~~_so52eJ8ik2z`>JD!GgSk@<P~8bOV(2y8IK8A00^1;o}ib
znH*mPOPGGtelk~NlSHHn_@SvD(MdF|f(mK5LZQV&_ET+cSCZ!FT-7EZ*@t28i>*M?
zlKphOr&ZIGPq;#Jpi}!CxKzxqd!_wM4t#Q)qC5P~m0LAUnTVh51|}Pm)=yC<puGc0
z>T}=pto>ZOjTfuVcZPm*=&S>Y!6a|WYKqd=TMFaP*E|!P8Qnv`bJ`M(p|7KqYZ$d(
zXm+^497eP*DvahDyHq5J_ZW0zL4L8lV*3omkEV#3rzP(V<wRso7DUL1{ZhMk^)#j0
zr#VWbYvBNj-n=OMFSj*GWh?1?*xeFOJ2<GNWH@CeO3;j0J?ve-(#Q_yk&lcGoa>kt
z)(fe&{zi>KQO?8mtJzyTe;-6f_#i?Ws<BjL2O?^E&$<ovYi$Xh^Xiv77aakWF$LF&
z{dzaKs*oDurqfnPtn(C@0q{zhA`jYcv>6I>Oo$PQPY$HQ22e4ONjCTGH|yrqGm-_%
z?{?OKL`RgpE^Dg1zrR&&65{1gILmJ8+Auka4MNxLx9hFx|A8qYprxn?B<c3#_{hxk
zR{Nc{Xq);y4mY5-I4v<S0xIBXS@R-|s=r&16it)c!TCvZ0+O8nratGC+waw?Xg;(-
z#a2*uF)*aM06aUf3g$sw_up@XnWQk;&LWnSMq%$gD>vSH+nu*;T6d@YL2jwod}p1j
zLfSmWs0-t^Kdf2G{So@7x3)AT$ZvMo{-}+83>6GIKjO5+oD=yQ92H^N{<yVaz(^y+
zNvBPlwl@Pkv_GlQV#9`m<PvHb<krE?<qe*-m-e&wRSQlQjZhfuPxECf?_78HO6MX*
zm9v}Hy>yEFSsN9CjKWQi6Uahak+?KCr2k)aCZ%N_v_H=guEEneJEyv3PD{K$HMNZI
z1)<@)e^Kw(mz8<;Aa>2xwhwF61)>dzEwH~#A_uO_I=^!bLQM%BpD*M8vcGB&`q{)$
z-n@9&fh1MhiHkwyr|qxX$RGz2YB?;`TboF=!avc8?Qe4ZSC0`bH0E68NDwcTqp2X^
zv?%;-!^VxAr`~riBfMGkR%ISkb25Y&{)Rv_)T9`87Sfb_`FHu6m*qdsWjlys6oc+4
z^T|<7qxSdJ`wW1RsCL|GYs9}dYyZ&3$6is`KXyu|ElC0Oj}Q%2Qtltyi=e!sJp6Uq
zVu6tph!++H)iKHRRKceIl(@5g!@5oDye=-k`^k4v4!EXAQD`!gE5KYcaN2=L>mO`K
z{DJ+C+CP_IN&N?Y3bAjRn;jXmf60~J#P)N33$Up-*fy|#?G{&CPcn=QBI0Hpgeyr?
zs>Pbz;os_+17Xez$vfu%VFwbMF8wc|YuLY6o0NW~{gR;Kw8etUcct?aQrKqN?LX>n
zMj4s_0r}qARP9A4o>8I7{<ERpRt|$!Ng}5u3D?H4KJ_I2D@VMv`;BjMepk01k4)O*
zusFN)KlfW&(!+`Az@v-Fxuq${ttH9k<1RI$!-<+0StL`>ObZq69Oth1CUyi_k0G3o
z+Q(NL^-|bL@;je%+KP(LQ-x{DKB3;eEIOa>N}ZIWS85TbL^^kA4Jk9b@(%mNOChz~
zai8-`%}cviuU=`NbSVYJ6*t`A{I1c2-Afq>yicL5pIneUv0aPQ5@CP@!fI<th=}?w
z1afPKVym%F>EMb=NnJ{sG{h2j>0t*R-;5m(Jk!mTsqv`=(YU0A?UUw%ezwnn#5W}y
zWtN2U6yzNU1BQLtr80AEo*`<3U<(>@)dxF!Gn4k|IRVJ4)RN<(2G9kQ@G>NHhK_J+
zpK+<k*Cfsmf*uA=OD-DL2<-#3Zu`uxv%_@xWWYe;ZalAW2*@<-`m^fp<RFqSu+bp<
z9C09pgv2Flx(L1d?6&wgtH-x&ajt9l3O~py3tFb~g4;9N@JD8PJ16?1PQ!~>a6YcM
z5ftZmtjqMKMcgC|_OafV>~k)q>UraJUOy+Oe)G)goVFrssS<5YTdZGhb5VSWW%!!x
zbK8`ZpEHUgNdR;p(ie1)i9M^`MYyM8<A>=UKw>JyxpHU^+vl~ZK;)bvevE6;-7#`&
zh#&j>dO~I*rGx4m9NMCPj6L+m2kBpg3D_6ZG)f<lnN+^CWrue;@Z?*F!g%;7&+~=(
z{-xcc?F>9h{3b53F4$?ms6Li;T$aBKNU6_6P4TKP&MBCggGKW@)>A$b)zdg9AwQUY
z*0uYR8WtfYZ16Y*#jA}l7&*MNMCq5-*HiwIMX65BbRf&(XSIF9c*%xcmP}w@*6;>S
zfjwu`X~<6ZCdWq!P5AOQBb)n}rss*?q$M{GrkjXAW8TAf0j?>c=PPoUxb<h9-y%C4
zXoJ%EmDN}AU+G7e!bT09_Xd%DQ4Z5C*v-DGTi!@<gH&Ad))tOS_JZuD&u*&_+&-gC
zm=lyLAZ_P#aVUyHkc*jvVeC1HP6|y$bLT1$FkLE{h5X#E4?6xc&0w_U7Lf6X9(TH)
zUD1ZWYvBBj;ggR!WzTC<Jvu`Q9#c(2tdKS#WDIRc(qz;HCC>;~r(EFn{Pq}f^9U*Y
zPJ{5DZ1VZhzP%tnv9mX^ZQxua&ASItcdNayK0xJx@HQU_FS1OB5n;xv$uh)UC8hN7
z52(DLU0I`p-$OREwIWf4l>(BBABuyp3XKSooKZbz?M2xG6c3JQoNzAN<b!zW|5B!z
zZpTAnFYZ2+@xB*As?~W@Kyo39eo>0XUXo*_m!KgRaYJ*eU4X{Ev^~U(O7})cb=tHB
zvtyjcW-se*P~~Gy{DRY_`9w&fV1&KATUf;(NO%vU>Eb!>AP5@@5kdCm4YIwhsJ9}{
z@k|`y>$Jtl*%_$Y>Ld}t#D+yAD+K<kdM3+ON1&?HR&?t`bbkSvL`8p2+tmeWBJCsS
zdg1Bv1}WQ2s9|A%wO2I6*-C+k-sF_ikT@HT4Az!q5FvcnuDMi3Gip7~&*%g=(38eQ
zrY(I{Ubwb<T3kv>OW9)`NDNhP6jR2O0bIW$snD)VlCmMrRwuN8YiZ?U(>v_?*2kGX
zr{m$$qajumrUK?g?Jm2azAQ$gh^xoMfnh5997y6oNQ<Eg+KtsF--%8g#;oXF8}vgH
z$L_aRcBr{L2=aW`M&zFWp26+<5~JE9jV`SYd`%V<nWSD-UxS(<f@S%;)}|ha$&lb>
zudXrdC-hKt@=wKHM_zRxF?(r^D!L=LJ)4ws_Utu{0Iwr6DNd)Sr7-^yWMtJ{I&ZH{
zs`H|^yzsfsZ_V#b*qUC~O-4q7({XQYS#DKU*9-R5xk;9G-@banxd=Rwq#fDv2(v*^
zZDo5M^1Z~fBZ#TAE!*OWpy^E#H+tAs<@SP#XtQxu+hVwhq(MkfCiK`=*B5w0m>n;%
z60~UtbFM3BcvHP@Kfc$~&Q+06MQ5b>FQ}S`77K=Zfq=I)IfXORs_q>J3(^u(s5lKF
z1U5>$xf^zOQU}XU;GP4iA$ElW7<_mwrTo|}jc2Z2ch{Yp-sD`Tp?1wZc5C)G&N-NI
zuDanXo>JKdVuUT?=eB~g{9%MrlxgZ72jrS_;JGJP$C+)f&zT_x20tYi@os2RHy3HW
z4V~R?@4l@(4Y5rrSaQfoAUNf~i=~~XSMqi|z+bkZ-O<RyRDUPpu-wQ4r!hX)LS5wR
zk|hoKLn@E0&8>~lFsHbU5vL(#t1UA)O18{R*qya{u#t%0mJd1DIe&F}E#d(B5_d(v
zvdEAAo{oY)3%0KHm{nFN#wF=fD)Zq$>RTv**l)Jl`dpJy)l#qD_i>1Xy`jA$h~hGN
z*nuRU-9kX4ZD>2>*11{dA{Y3jFSL#If{+#Ha@LeVg7FR_y)<@<@=#TxWmAWK3!ws%
zP2{wt<3d%3r{;*YCeZ$_n&f5YNQ(60ItX{N0~bEr)89<)Z{O~2?}a*-Y1wJ3-7H;N
zds=Q!?3s2?8+vEc5<g(uIFLjoCcDyO*j#&{gQ39L8|ym??UVA0kKJjL7@>;I$_dq+
zzp2AGu!-9!jV=_44q&eLA$*9DCBlM2582mL2cx@mT|!&%T?dkr6}|v{4h!R(TVsFM
z*!bqig&sJKwd0U=nEunqKv^9z3HZIaloX{p<@`?BS6&@#CU2={BhH(r7Q)DI38J2o
z6Y!!U=`-W36?^Nl>5k}0Tmkyh1qTv$Yd_=qk)~o_+r8mI$wvtLN1V3Yhk=oLRz^gE
zSMBS%Y0p9oXm(CpEibVbd`a6dDy~|u>%EFwG)R<%;+^{(csv=VZsrv)d5+6Ax1OTy
zDS`tA=CtM9tnbaQ$5}Q~_b4esHD1ZEa@w8+3Q+lrl8*Sn=C(_X8mO}4Xk+wLxWJ~?
za~g7ouB5!;=HB>y)=#$e<}2=XekN@RAuEZY^z#LVp@`Tm4Ia))S^MPdRFY(n4+oxG
zcfHQ6+}gU1jzfQ(>s&U7tOEChYGWNK%2dp(KbfwtGt!(wROYTEznt3FpRn=jM#Z8^
zH|BL+{*^a)sPxgb6kG_TfPtR0`<}MaRR|QOzt4e$IaB#K3{4s|E2Mp&*<2E=6nY@f
za1^Ma9o}x+S_khQhg3V4ImyUc{m7}EXg6%^&5>Ebr9_1-R9u^ED<luLBIIchksHj;
zZMLZfL<!kS0Y>i{?i?3ip0eo<oA!kWm3XMJ0BJlrG;{^bH)&N;cucKrPx?wI^q~>w
zLZRgBLB^Gtg!{n%x7EV6{E-vbB4}Gi-XxEsK!(UQ(}s!xh_g7k)Nn|0n72VgUY0?*
zp=6zUYg70~vd4oqoAWoP(qkyyQ=?8xk`ta(*HsKy&Kp^qp@TNpCTi6!tJmGJZiCaX
ztbfX?lTb>bDBID^m^6vppDAwuN%nHvX4~0Nd96&DTouKMj1uiS+m+*-rm7c_2hOD)
z-YWJnU$={h=oqMz5M9~?kMjiYFL*f$)Gcpq5p@UZQ7GQt-u7GQ&53prEgB&N9^KM+
zl-P*qU|-*GPshoRs`Iqu+@T?OYzHNdNZ$>jZ|^90#zus}26%UVrUOr+uadFTQ`;w0
z#_XLr;j1TfP$?D+ry(X>K?ACOaVfMFl|}i6W)ce}DHKrPIs;5TES%KA?L>qhXNk@)
z?sc9B)J>5x1N<5_8k!#cQso`QX_Qi)6kkkh>JL;#Q#Efz`Q(^FD`Lzt8NVwR@22sA
zisU+%t@BXxgO~?p4ThQ918q82g?@&rL`&-JY{A+42Us@k-Sv19oW#zs_Bas9(9?q+
zd0OAtEtjOfRFi2-`VIr1A!qYFP2Rco71uZyN%ttHHND-wsV(HNAR}*436CgY<_*qp
z!R(f_E%77Oh;MESiJ&t@UGoshX%x+7Pn`pMzNI4vd$u_j`T6xoa|3&Cjgpl@|3Tvj
z3k;TAZ;-TxMdDWA*+F|>bvaET($>QKhbR&$)DAo_m|aYl)0p|KZOTb>9V19&tJ4w~
zEH0xOkjttfD<Rsxt$WKkv{zY2E;?<RSfVZAds97of4xJ-M<q(@ul7x*SDka<(G!hM
z!4wyk9<&d%2_yTGO1YzfD>MeKq{^jNkh2*9P#?^x#F9gedgrR{Q@jORjjqR<*oSg}
zSU#!}nA4JscgV$}b5iTiBQk!tI@byY5c%${O_L(`IH_9p?d?73B0?P1nVBlFrb#>M
zz!P^RAaEEs(%W}*bC;%nNzkqBKv3JzknKD3zNOvx)=?`s5o*+F%h#>#vr;M^X&<Q`
zppruX$|KAiRV1kc*ED0LY)3#2?7NorTG1{l7$~|dg<YHsc+Mo2PbHp_7ZT&yXCG~3
z|JJ#|X7LFciXtjTSJ~o^+sBgG-rV2m{1mdUo!%ii`Mc9ivumsKOPEf{Yzm?F?Ry&W
z=ZNl6=Sq8+3U@cIu<z}rOGg8VG+l7ol6sWnC>|UE)%JZgGL1$vps1_z46ru)9C$pb
zsP1Od)h0M$-`~xN%0{Hto9cbkm382fAYT097uyeX6mMANEGef1cv(lVWjlA(9$Hp+
za+zW52iusp4?te7qGO;<IK<^^jGzVE-R@KBjdPW}Va<ff_RG;fusz)(<o0?T)rPhp
zGqT7Xu)S?X*%k6Y%RiDD29W5cMx1GUZQfE|+1EbL2X{DEi5cR^%=xLn4;g+QtUic7
zisYd1dDj+<DDceoH~Oojyxxp+t$E$Lb=)3mO8^g%8l`AqUKgm0`_F*~u4!w_)r0NR
z4zvy_kdJ{$i3m~3fe6J&Jf!OTgdNPw%@k$FZRT7SB52Ddzmg5@pdD%*T#XkMa|73s
zBQ~wsaHk#ap4uDSTn@pQCI^xP1wNnArwS)0=$7741=b!ec#6<+C4$T$?qWE7@?8!*
z`XdfaN-^V*qf?PcM=n)FZB>xcX;=|oR7fi7D>WJ0wR3OqV(|qPOhB0AZ|7y8N5uTt
z(KZV3RQ*$7@Q_R#NQ?ptRoKGhvk|h$=dNQx9?2D;Y%(#(RHSBNoi@?^-UMZ5W_klV
z)^_jI_-5y#&{}&mmq($nkK6Ia>yS4Q2)Gow_U2`Eu@epH)=JXhDp~fbK2H8AJDFe5
zLEr=DqQ0Q?P?;r!*@UXtqYYgcfXW$jOqkOa6QoINZlaJyQvj?fKh-9PWoo%JqWT9N
zNTMKb6H7kIcOL8RmKRQ1B-)FR*MY>{t!(NkPF+9UNcgt?{N`zn+8j7dDM7JP+Bzz8
z9asTgs^CfD!Q*vTTxlvhFOO(O)#7yS(ssIS?K(SCUkBwPYevzEv}r0Ye~=|rUe(zS
zEi>$0Wp5OSt@?_@s(6DFM+@5Jr;`s$&^v7Bn&^P<nXn_GNomoxTf1iC#+A3MvGW~9
z$t-~sz-~^jRYXWQ@Y;t`4q&$kg0i`vv<nT7R-?*tA@snl&rykXu|33US7B1kI59pz
zV#d>q9X+*$hCa8;pVIU0`fh4R5&eCsdt361VaXdHu`oIAWg0IlKNRPm$c_@J^6nAm
zA|6R&W{$KTFUmkXqWw^PviVXq?OM@u-XJMV$}p&W4v8d-Z6?8q3T0(Jhh9_QN$`=h
z_S)r|2;K^F7XngIxV{=+VlkBo5zNh&8d7eYN0y4rM4EEO32Pg*C+l_1giYqE)0TH2
z)NivNZdW(sq4VzQ4kS5ys-A|&MC=}!GTz#P@FW4YA4yVs)r~8zbAH!OQ<iK9zxJcm
z7U9IK_@hJ?(pH35F1jRk;jPjd;qv*hdPe(KQX#{3+Ugk{K$J>E^ORF`%znI)+$f)>
zwU+TOf`Ird#@xk&IQN$6VU9X=3W@)TQoq84$g<+s!ag4r?0HlA(j-)8oWZ-%?Rg@_
zx*iq&GU4_oyR}jRl`6W54yaN!km$U&wX!1gH%*56)DiosCLxeiPU*xvte3{@B_-cD
z?-D2VPp2fOa&tICF<aA6J5i)j3cU1H9TxC21xcd=h8d<H6H0)H+HB->+@H<0TiSis
zX6JWxF_R!8kDj>xVqErf9X>n`IS8fPTbsN!It;@pQ2z7vp5>81i9-=C?!W^_DEJlI
zM2P<j4M$|)WdNeXw8hpS<_}i<i?v0y0%SOaHLE$LR9QvL^Ks{{nvN5>IYXvjK$iNY
z1}W;J%}byR@fbi-a9w%lYWw95@gi<)@d(C6+VZ7g>`9YP=;!t;b+>3mgcw2#XiIev
zjk&kYezp3&e<>|bL#0f(EyFuF>fDtABH97IX_gkj``5aO(y>y3O-`FSdmnV`0sHmp
zSR+*={70_l8Gkw?_m_Cw1^bPBfllh1aW1m28#&ux-hQ(_Sb`2B1~I@-`$plvh>-nO
zn-+Fg>DGnpK$7(<4jxPOZ`XT=Wsfd6ZI!kgT^ZQ#WXIO`W(lh1A_%hN;3$O<3T>*=
zIx`&XWBk8cn3gypjI1~_`f+6Ov;)t6^mh@ALr`#PGS7LD^Gmn|Z{a9S`~B))8y)^@
zr!5w}_-3r0T&5T84+@fBmGU50Gs3R}NldzJ<9hqUHa&$k$F(!jn$I{8$)=ogK5Ku}
z9$7_RITuMigKA}{%^!CZdX9(%uHudsym)DGzx_$=6Dqi1DC8>o9rYBU&{^+THIDzZ
zW=k}&%FwZEB;f`AIPiM+`G{kcp#EoV9f;{y%_Y7E5z+6)Dmj41xvRcOTq%Rp-qO6B
zv_G%si!ow`kkG44RmfiaJxjzLcU{eda5(Gk|K;ER1!8~EXz-xFRkc(4loD=DUCmef
z%VjI3DD<c`E}`0i1S#aOvcIaEYBPn2L45X{KOG{QpHx~37TI67sb75bNq4UFvTgk>
z_BS<`nTe}ee#Bc_S|{{^llHfbXTbzi^GX%dQesiwm#9N`2&i*!Vh7r^Pzhbl{;t{-
znF*i7)A*p%7IeG~L>6iciD2nNL|H##f1eX{m(D-gGCniOoYB;genmJvj!2wVDk0%P
z`-i8^jTou)H6_IW(MG~G-<!SP{;}Q1;woMCNLo^vLm=9Ud=mQ|29STsj^m6O9mh62
zI|eVSHl&-2yk9<cDg)OW(-N9}l9OSt14-I7LlvzV`{%aTXZo04=bh`o$uMNbnpX{S
z`<HennMa$)N1bc!+7T?HGxo1Fp)K!J8J)CgzjAWhg#BAvgvcHd8#?5}wh%V32A#Km
zug?q3(NvqBOYr#+vOG(2fhe*6s8t9H0?5D&BrDW5Wt$D*d)bB=tDJ5Z&C7!Q=cT8<
zl76x>$09o6BymqNjEQ$dAsj`NjX1Xj9vwt<)$z(_u%rA}o6LKNK%I3iwW^nof;Okj
zo+vS7<NV}yN?Oc2ElI(s$MA7a1lGUf`S&_M)l3NBjD39m?oD!p4!?6PAb;|kM-4`n
z_n3Xc6NS^zB-vsyd^&B3QX4o-%RcdmaJDS%<^<R2sqWAmnv>MtGQQP5sa;?Ib}nLg
zvZYExMtHMNt~SX(GA}AIlD5EkVG|o96u@Qrn|(^xHSDBxnrtAaEt%FzG7fbr&8OB&
zWX{h6-DIbcjL2UW8+eRd!9NL6b~Y-$eOlMevte8n|DY}9J<7}u#wpiepI+~mkSOo8
zDyF2=Be8D5f#)<J#M)<c(;SJyFe|D)5zBMbf#j5;nUC6Mw)=t+gb<|h!R&z)i9w)d
zo;du}KC9Y<Z>p4aig4N1CWa{1FS7F4?LLHNF$oIbIBhw{uyL`OjH=1AxiNRyGwK;l
zpDI-0nekzpl8zKK^RRu+6UE54al#^sCD*z4C+FladuG07X*V?lsLB*I=V=ozZk(Fl
zX`h?JXXdCa;anB|E0sz|Ae^*kbyFfKEp3m%?Yu#j8)ObB+QmMv4UY9zVngYc@`)e^
zpY#S_hE#?7;0AFkpWhaxH$sXCSL*Nh#Qc<fL4&NEn4R|S>Z%)VxYoXK*)&TvXD>Nz
zHO=8Nj<83i8#FoTiyGSA)0^4Jkp)geprstRa*xZsFOS;e_QlzaJNmsT=XX}D!Y4jO
z&8mo5urFyJ0|7;a5GMlV<sqOu4qUiK8KUjIt@fq$CB>zZML}4U@*q)29f$_JdVXZW
zzAVxBmbGi{+~iycuyqb+%9kf`=}pc$KZWX4X2e|e6}6(2a9%v^Tou~J;WzA+J@%FL
zVeLF6=x6O2C2<TP$HUqjCy0RD2kfhAXju}w(lvR$ttjyP4gvPEwJ7o9+4YQey&@_?
zYXWIymB-NNz-v!arx)}8ls%{GnKn-v8`7!^B-t=xM@?sL!oa8ndv4q9+o$h$F2ZKC
z{%*VCiIkR;<7cQ3<PZ5FEh-{j7o5A|?O5_cpI6k2!u_QOonYyo*RW*e+BF+)TfNe0
z&`R#;E1cV&-#)$>R?S5qkVdT>3R$7!>-K`SRcFWMCN~rG+l{sf4Uzz5B7NZ#r9XpN
zlPJ*WBQ$C0dlXx=D{D|f<PaBw=M9S`=9F<)EQ!3w8V-I>dr^Io(2*w<k|T?Ba!?>S
z1u`;3TZTM5lpNAtoNrq{GrdKTA)G@p;j|^q-#)GQj9Gh0PQ*%8r*kf9A`~1P(h`5E
zY-cZRi;18qdZ+__bLv4HxH<4=GuEZuFa3Z2$6nUvX_Wj|;{1uvrz!1Kyqi%+Fkut-
z9zo^`vC&@M<^irJX^dq~>$L+B?(7&Jw-q@&_BI)M$|a#eu(8=OyDG^n_S^Bze3i=>
zE0SNBQ_6PFCb+t7j6k|jADy<iTU!rz4v3;M%8HM9ML}3uHejj)(R&Ay3LEZn&dIfF
z>b>F|5>+lP9&y@Y4Hep>C~77CWQ0Hz&a`V=FPRn9ZBqJ15L^c$Tv^(^Qel>nd~es)
z-D0p&6c0WXwp<4in01IE)=KZcuyn+(Z*5U?UqbZ}r|k;f$~ev-ZzVlcq@mKHMPQJY
z=|KcB+6_gv-o1E=hy$(?v&aM+(|Kz*Ci+L9K>fb_5wx&Zwyn1b!Iq0$g$THb`WlJd
z_Nu&Xj_89~=OVhFBwqXUg1x$%co`JZj2TSc8${x&Hp|d?@S1w21+vR|mYWGkZaq8$
z;+coB{k*o@V8g%#1>#nol>?8?mkJS!CjsP`y{;`Q94#d!h4r4o3@Vf7AqTF>l6^$c
zLi6_3Z646nRGtA(G`B)n97wJlY#0@a+8hhEvd#5+6^lTGprLv~oog0Je`VS^Wvl9O
z35H7BlQn=gkwWI|(WNuCI?;6VoKkH1qt11pWT2U7wQ@*@$E(;)Pn30MOK(8VUtpeB
zV?-oo6Y9m9Co(T8-xRSAyr~N<<M%M)DuY0HLfdr}V$RL^a|3l(>*vNhSK2L!R4Z28
zaINz@ks?&!P|EGrZn!Fvbr2EhxYHKHRn+q(yRGi1NFrF=DQ|79BEd;?p4TV-z2hBM
zUhxv=BDX)p5cfzLCZBxQlkc+Ii+G9aDC+{D0eE+aY%L;p#hm6GNIWGYBJPcJB)cPr
zb=CD(UGMykZ%NM<L6Nc}96fGpYan}J0@T!PV&ffsvIy+Xwk!&~#wkWF?Z9a)yx7Mo
z%9JnBh%IdEs^4i@3AbZlOol;lug2iqRnIGTUAQfe>oMZP)_3C$r%4zXLUH5$owLum
zOC)d&>AmEj_w5b!K8hE~wi32z{XOnL#4N8~aU(T%Yn&oN7oDrxnX+F8X|s)O1EBA7
zQr^I6&>G{K8}Ciormov!(~`%P%b+uThLDI8W+5KSC9=C}Kp-pr^C6)@bO{kOcO!R2
zXJq`<n@L%*yK8YX2BzgFDbM!*Hh11}dR52WpE$kuUQKX;nBHw{XAuYp63DCo+koR|
zcOPjNR=e^(yAs0K8H3ZDnAqt(iv*~nh7dwRvQ|q-@4ffldr!W<nK^eK-sH{a{qz0k
z!*b7=``ml(Idi7Z9603CNe5oY1I1^0I2f5{Dt+s&j!ftfD0!xYTq|d|0heZQJ@JHF
zuspuG-7dnwOvwEPdy+!O$~^DD)9#Adorn!;?INJPrNj3umIfJc`cjOeLIEd~VV!+z
zTNn&ED(o-5n_<sDa{5#BWQ@D*?iz^HwstP0&{c8JI&Tzs{s!aPp5SMv?VkD#!)E3}
zh<gdD0Iyl<sul*?@3(skqF>6tz7jY9L`ZY*JKyy>ds{aqd5oo8$_1*RsN}T3<M}ce
zs+gau2IN-w${g+TVqftLvk10*uRfk|IJPzNzP-KPkUC~=8Q~U@Z_wr^9e8aPl@Y;|
z$d<`B^y-(=I^m(gV#ElJa~FcEufT_ku;p<j%8?yw>ovNWP{U1VGxX%U(y1d|MRU)_
z9PdV%Mu-z3d@I*)C&jgoJMfC$P#7C$$xKf5Y@(Z-IC&_<n+QjAAe!PWy}23d*Ig-`
zkysw4hNp?v0TT2r`RcIvk%QXMc>ayO%}yh*1JcIsQ{dWm+gN=%s^uw03h8tD;u}l;
zpikk}3d5Xi!vz=QJd+_e**K7Rb|olay*gr3?Te>2Q8oo>iF8h<GY7!}5;6!|J!to}
z$3n|ve+g09{0AIJA@nX3zYLDLzxoJal?P8EyVF;^^AsB+KX6$7m~PUq6iSZhAcmB#
z`pOevx4h4ak`;;FR1YJ?jDc_h3e+PijK>#0=D_op%#3UqN?{|gd}*)Ebi)uW6Lv;|
zZ}_#9@CQ5%ErgESZ1>g53(N)3Ry*lHM7<R@HYfVCx&7>l%P(~r@u|XQ<x(oao}VG&
z1e3FOH0c))pmYn`V(q;L7i_M*5lwDSEVe35py4?XVcYHOL>qMTOMPHIOu`U-DN=9Z
zNH!#l`6T-zs5cvDoR*+xKZuU}0nlV2$>$u05#V8Tn)&1Fsi)iT+m<#AD4YC@a?Xn8
z9d;lZ#)DJXP`BnQIIU_EC)v<SH|Zv#r%z(OnXv~NZiFV9+G*=+rvi{KVB7Mc%Zu-P
zwR0W#RNNQ7eApgreRz8$y2w~hUrcf)vJ1sE$&w3Vh3uVea0rq2JFUXLSO%H26SJ9J
ztUC5keXy$Ns{AL#b#g;F5HW<?2@suwp;ANST`SKLT85PqQU!8Kig?l+Je;r~pVCrg
zq?9jJKG1hJ4y{<>1t=;VY2{kqmzC?Q>iIIOhe{L3FBwWZ=`jZ`bXku>%e!HV?dkD4
zxPuC&-XIcI#92Gh+}_hr^%jEpXe7I%jvndUVDiH?*GlM<K^w{3=?gVfZmr1SKz9&>
z!^8uMA4?3hNAj11A5ouoAv}UkU$IkcyfJ(4t>^0;0&16nTBp4AsbC=N0b%qx`|4(t
z$}=CP@Estv(Oa~M@LbmE$T9_dt51%K43#?Q^aTMRG6tPRs%K`fJ!N0pCVPZfGm7{o
z>MiIL!>?<gRZ>2jrUWN4oR#7K_1(C|sG@ME)0c`}e}-T=a_isM4MpN4Yyul9eOg`2
zaZ!Qh8`{X&6VLFiadMKFp93iw1&MH8=P#fteq%jR<#V_lkIB<cU!r~pz;I4zSW@*6
zW9|LjP>)BX9!4LhFKH5@krbeGiyw;J4c)UxYgp2-QS!J;*$0Yw0TRDd)~QmLq~(S(
z8PxNXeN(};?L?(lg#9E;z+)*=v_}Ni)`L4ywl~MQyoeo!2vmvYPG6#0N|tab>Dpat
zVr=gH_Q9kzoudHVjIQ{h>|yX1c$W%1Yu8EaRJH?~!@jxhdXimxN|+{VJNJsnsl;7M
z+}!mtmTdAq`<D8+l*iM$U)owwUEuL?X&IBR@9EkN-`XvL;!_mReZ*T|Qj47g0PeMK
z>+X}-Emn5Q=@XH~*>;P4dn2n93x-_JObnc^WVuMnna=*SeMi?PJBSFy5vMQLe8nlq
zwv{nFJR|m<^&8o&6jrKW4EhSC-?4zNVa&d(TM!9{8B0GD8kt|~+!aGY&yfhJC=vVa
zWGW+(wnC-=?HG>4%&a;t_o|YcG_Qy4dveEz(S}H!^`y!}q3F#4@f__~``+S~m=U=Y
zv!37tr)~Cq9iD~r3H*7I1y<th3M6X?osveOueI;5VQU!j!5tuwyz7HNQD;FmF#CZ<
zGq1e(5~mTEHf$+Lxi7asB!byuu3=R5=Q?Ibeu+9S%O0!Yf(t8DUddw)B$rp4B$o2H
zZEthOCQ}e1j!}l21&b)0>%f!wjx5ZR+J+!!JL)lHJVUP2xI$lYn$3NRLD}QY=eD*t
zv#<gAFODL0y2W4~I`=|*qMKA5kShKv3X3S^Kx)NFKxdRYmiOC}ZLMyib`t@_PDd<x
znwy{<Y=yL)YO40x&U%QF|3r1<9HTGk4IAtb<5m#Ckm78+>NkY+gs%zR%IQlQ!>UTt
zN?A*pExQ{eLPE9GvuMH&2a<?&7Y+}R+C6P@I7J9bL{lFKJBy0dM8a>hy}77U<5XSb
zy51=@UxI@|&zKzB*RTbnk<*CVf|xRQUG|*q&)bks@Wj%{Mb}o)6ff<vr`mIFU%$#}
z)Kn^oN_vF+kg~po$<+?Dn{Sz&a2oN_49$upPkusTvK{OmvyxS>p=BE19tR%dfE9!N
zp*h&0*2nA}i9L11Tc0?N^mEDs+TnUD<{OZdcgs?a`hCiQBzkThY{LX9^gPl{pW<>A
z=MyR+45vXb`-b;LbUAbBp3dJalXS6d^Ze{M6cO8rtb9O(va1n}k0#maVY7=9^Jt4m
z^ryEFSNBXAadS0NNf#=DV&Tq|%=Ivzqr*+Pw-z{98q=rTekqd28hxJI;#|jsg}TbZ
z$=m&Gn=p2YBt&99oW8;X?f20A*a{e4?07*yTiGUq4}}Wt9i*6cloNqeownziz>cg>
zR-n_)jeX<7ePrC7Xkr%vzfPe$>GUM=)nItNuq=6fh^cn6!;4H(P>jDOMn*A;%scQP
zdRfys8JH~oBHq$5g8<s|A9^th*Y%KgW>o<yqII19+-s}R(Y4o3)i-C{R~Xnq81zAJ
zkaRv|Qf*qcPS>o$eHgq}@VK|W^q9*w{k(ms-KSM#U{qxC)+bp&=*s8WnYvGI92E^`
z<(<A1KOh%}8BtXHjXI~o71d}UK*N@5vZTePm0U#SP)mz&;K?Dv2sDgeDT|;N;}<(y
zbC6s?Q>vuJ9oR2{1WrlbPI|l7mfHdb_sjs9)JP+a$brXi6zw4%b9(8xy->dq`IPNj
zqFYj+G++VOa)BVH`0sH1pGq+HgDc3;lvaH?vl%0D;3=u1tE2iMtx{9)#fE|$p+?>=
z%_Ko#%0M~iNhFIb9kw4z>1xC6&Xu+%{`{FO_QUl9v9+8|iA~DC0+P2~KeFC_q<v=~
zpUBU%PG7t$T~H#$$ATaqvmfndGR~b7n|J#1aX0lJupjHDs5{(L6QxgOutywtZCeqE
z^uk1lr~P;%0&iO7T&JpV_ED@VKhcH=n}#z$2ah>@sbR0`O{<R0PqraPwZ)cSisjRR
z<nY#w++;u1em3Y$O;dG@c9Cb5mW+a2_S1!JiLJ^JlXfO&Bm)J1!r<rJ6(5q4taeY4
zeEXUF5S@PITqous{K#xZB(?o)^<}7(F!;37mm7K={w(Eu{#=`Z)pR(m{DvEF?)`j%
zU<G@Yw>zz3?@}~!R}I<z?H9Vc6>&(bc7o!}W6#?!))Qj4GsZjyrSn61gG<nml%zBu
zjas{lx(oJ8Ij0*r-g$l!13B#+k<QUzxd@(OT!h0FQTyeZ-;5H^wAqyR<n2(jj*@h_
zE1sgItRj#aZ>mIyI=_<J;{o9nNJz7`hcx$H3=GK#5^~CZ^(7Dx9F-zAtpiCc8^gC{
zzm{{{8*G}La2oOUe53yI<zTyJSW!s7UbBj|1S&@>aUp`C^6ubY_8V<3A_6I67Bm-+
zF&CUN<iKM+#5DBf%&`4thaXByiZCmuFUF(wm7KDY{uP|feyiXi=<s>MwnkwHYVrn;
z;zK~Bt>w(e5e)X*Z5%j<MS+n)oxa+tE43LzQnug8xyOwsi@wv08-?CFSW(N4dWJsm
zyB$RN-tvPfXhORZNFm}W$!C1lar?bo0hznzokp%dlCU1iX^nxKsy0dLwf%mBN4$;{
zV2QO^k_B?(4XzO@QCo1uiONV%u)qC5H_J~%%*n~6z25r75#bXI>IwV9T<QB3IF@4G
zY2`jeP8jRmH*0^?aOkf7^t{umuZqkbnZ@jn+lwIhLZXAq&pCaCF!F(Dy0w`k3jIlY
z-W{XYk2o#ud#bpuxxxOlexwzbO)dR@2+Izn_DV=RNM70k-H8YGXEg-Q;gxK-w1c2q
z`i3xE-zv?1KY#PP|GbmzB;q&Is}v3mS8B&jC?*aHh&uO*7!!~2kp8p3sAuP_Yd>cJ
zTWv{<?hZWp9WI9XNsiIEA5roz+tl94$;`8GJUZ&DvWDq))(!^t!o&Q1eEAvst5O~|
zQs9s4nD`u9IBx8(yGc$3Rt_5NjHew)E>ncpWzCkqY3%2^Tb%2Z{%)Gxh>!1Yy9a4i
zMgDiQeH9V~q#lIno`+Mq!v1b$D10`_JA@XuK9P{qki@6Zf&V@yP3C#dqnU8p@vJao
zaF2>f{6mxfCZZ~(%vg~mPJiATA{pnYES<7{Y<;k66~QB6&FM>XA!jj8Ww>u*6h=S{
z<^Pn8<;B(0v-ZzJ2tfc`M965Xke>;odD?*&&Qi25O%g@W|D{{*2$_aJvfN8utcwGW
zvM0p3v3N;FL-6Ctk(gXQKv8boBN8%6!rpHG+LmDpB;hfqSp!JSm*%I(9JGI{PY?GS
zQ%<yj(^uy6*$Ej>XXotS+c;N5OGW_e;ja=}fXA#x1vDPb**EqdE3*QD7V}fSipVFZ
z2AKh#sJx02R4PNJgDL*AO*@v(@deDZPS5$03}swYz@{SDg7ZVoDBr<#$8T#=?MxfY
zQ|vSHy)mU!-}2%dRXb;OQeWRbtnQP9BJy&Ky-i=tl;pQK{IVp?6ZWz*q1V0b{MR^_
z@wM7A&Fb~xXM)~fL0q@WX~=3u!JmPBM0+9<G~rnM1Zu&78z&t|O`5VycE0d1-0>s3
zJ_&(xvF-8JC%L?xvGq~iec`BvZtZaTO10H_j6I@O?W61EJGnuYrp{aGK$0Af#Tj;U
z_AzH(?DUNYOJZ@Eb7olFP)0e5`Z{;z{N&dttXYft!o&8lXOiX&%R0I~UK<{GZm0<g
z$g3cTeO&zlv<@l<B607^Q8vKiT?#fk3?%B;$v$l#-~Cc)kJ|9UiN?VWIPens?i|=B
z)MI2d&Zw~FhXj(6@MKgqwol9@KIei9oJ&1njqEY@NoRtnO=9v?><vw=F63;VeR8)B
zQA;972YoSaG4;J|+^p_y`;>xYhQP4S1SDq%M;vw5G4On9Jy|3~)EiC0>B~jCZf<s~
zivN9Dd*D!LM~F%eB#=vn;4JL<oPBx@oYcFu&b1^s+|-;gX`fMdi)v{rpj$hA`7%7Q
z9NI8vpV>_$k}mjwcEhc32nV9s8W||A<FoQv%ba3MPMM5V1E;UlP@NlrkN}&751<4l
zH4Bz#pWT)fDx)yfv=3(iZfLs$uSp_$2~mH-KIhDfm-FKPJ)H4>hC06uCb=T=iH$cT
zdd@z#=7eJ@VEysetB-U8%1Sc?E!5|A)55{=JVD31gi!84YPn{ymlOW;`S}4vWTjhW
z-04Xa+&~>BOh#W&4{LI&bPcpAY_UMn5|1z73+t}5*rtml{yLE4{PNE+n83=I&d{^%
zi|Ust%qP#!!2;<JMw&imQ2XL`H$+}^U?qPFNZh>>=}?KsnELHY>fN%SNz#yE2bZ%n
zebDNNH@J%3%b&N0X-(@v$f)-6dM1)OzlmZ!PG3$Tg~!>7+w7dYX?byta5CfC=+jNJ
zvh?c2PW#f_u*-|19Pcr3n!@#)$eQJxj&mDuouZwbX6)8-M6r=FI{ApbB4?Z|<4c_D
zq$sz{-e<3@fl_A__7I=zt&ae0iI5mQ_SmcP0q=VK1@}1DNt_3q@rqTKT<KTWV_=Qp
zE#xG6r%#_Ufkja{1p_;;n^dWgIAx$=?2K?R%l!}hTz(MZVaBGa5J9TUVvLPABHr5h
z`Q;;|ZY!mRroL(Qn%gjDd|A@ITY8(kF4uAR$ZNVK5%UVQ+v$^tN3hsN5<2aI4i~$y
z0@fRYxi<*I!4C$#kljVR#4gN<+0>hwaIUpD%Zry@aq-1=QF|eDUZFu!Q0bFJb>V@&
zUEIz5ksykiGWv20jYV3gUD7Qmr;ad`aIo{*V-BRmVny{x8a=c^g&5YQHJ3xiY=$74
zdi4a|V$<M==H7S~VJ3VQL|WnKFRUB{dRhCdRSRPSr&Y8j6CuUOcM<V{^q{|XdCuwR
z0#iDecbfX*>Hd^mu`<278)ZXALpk>wBvVJC>Mi3S-O;YB-$?b444P-8LvI~OY+G56
z6vMoOcVE@+hW53ncS)L0IFP_CwIU^+ndqoCB7xX@2K$|`tIvc1Z+Y=%4wdi6q4JXG
zoc>e@M#dHfeY>Uxkgg!=dBHvCt7S<IC5NF*Wif~S_S$aIRYWRG#ZI5FW|Wh65Kmv%
zmOO3dv9O`*IuKFKwY`lqc5U0k6ALp=BVHP-$(In@&Fh51kzr@AuSZt0L}8JV#a190
z11BT8_~x_rh8n1}nb2u42-26|f87)%JnW4*9*#bla;{TOsAZD;6IMj~@|PfD7Gq;@
z$YPOhz#L>zuH#F)oO|O#(!Qa)6!fmcC$;=1ZDi#`Ls*pt0M(Ur8HyNi%_Lg+IES2E
z*U+{{G^o=Oh1fQOm|&|C1mE3CY#Xh@=;1RjRf9lE>u$TgIDWFA;_|2$5*ZZGolJ+j
zlK2h<bA^xHZfJymFe`6--|5geL1#|NvDLXlZoY1<*X3)^d5zuJW;85u5iiF5*etw3
zv{16&M1gi~BfBYAgp7lpbDdQFiYqR&n`<>;vXWLLL3`^<7mdQM_t;m|eKJ8Qp5QU8
z6!gV+39qEli3v)j*lD-a-RYT9)YvJfuVz7q9BHn~1Z8XLZbg(o!{VPJQkuS!f)*%$
zEmzXoS_{S~fm0+l-X^;dwfnrmqi7d>;vV4`=Uq@DyS0Ab#03SbhJlj5PZMC=;Hkr@
zoQAlbP*8{5AGX`t*CK96Z$AZ%iWoo#q6zIi*uxdHu3kg6$1GeUOB$7N#4ZVobKqCz
z3a?b#kQ^dgiDY0AY>_~$h>1Q7Pr{)-Vz0faO+tl*z&n`uOif|>9C)tnSpR;zJs17b
zi!Z(0x$ZVli<6}eN($Y~&S-}RGUg@Twmw-yWuHGl+{*g4vYK;-iIc|Z3#65|_ki6|
zcVzsaj?CTZ^mPuJItQE&Jx-jS^cD1Z1$Wq;ZCl<kGu3xmC48wcHHK77I(K!SzfxXb
zV^A?AVsRaK{bsiuIpKDCb9*WgL*q`XC}JKuz{wKAMQ{nCf-v}&?puyyh@zEr4Tn9a
z1iZC=oF>d9q{<3Y{c4FR!0Yv}BR&e7zun!?20_1h2ts3YAY$p^4Oe*dJ@spCF{HE5
z)aeT`WV|;w#x3{ePsZ=t+M6f9+-Zx2UVYg$_O^OYOeNEG0#YB+sW*r|Q-;?S+3zdc
zmXv<O0`uEidLW5M=UjA=y}j+lg_+I0DJh55bC%+>e5eXo^=j~O#uJY^y7Yv%KGp$3
z%`Sc65gTi-3_A{c5rbhEEFDNpo<ee<0`TdlY`k0VuB<yB6@~<+^C<@&G$sA+Ej0v1
z*(TZtEI{FD#BEi`U)d525l&?NZrJ;z{X@uh`EZ|9Y59LQbb~{!qmM*zr%zmW&?87@
z;|iio&-a{We2XLyRlwV^^sG(R9*}ueua9DULo!FQ@^)7p%MD3TRU$=R<QX#|nb=hO
zy3wiIoJLL@TS+95oEy8ZE!nC>ub$JR(UR{yZuhsJPf!4gM&da2RXHbXx*NL6wPmhl
z>#GV25)f@u-6t^s&zK^Ry!8q1WTYb?&t_@~Gm@#*t5``LY;gJEvtIQ+7GDX?;_o|b
zwjm6TnUoyj5C*s%kofLo8%X0AO*c0MWNd0y@zorhLRU)8b3NqF342Gw8POcp=CRn8
zxJn&}rU(UU*5;B9^yU<#?bTXdoP>kgpuITEbQwj{`Xri&7mDf5cT-8&0a(5rF`quE
zK8mD^Vnnvk=23Cq+6oB!tuF&fGGJg!-4_IvO|XyT2Koy9^LRqV<jSA%xNS|saNa%F
zI+rm`I45Bv8reyEp!#@>s*x^Jzt8DQC3F>zI+Q2dRx2mEa3H9e*zO5f25cQGBJjjJ
z(DN8@0&?uZZvM&1otvG--A7k)o?vehO+@dNj7F>x3;fP}3^qlb$Bxl*^Q2cFFUn$d
z_;u}}HXJse0@<?Gm*UN(jh=AeqJ{U7N&2q598AqQmnlb3kC?DhRJnII3XfEB*lCqi
zqNAN#iZ7?8+gCM6=tHSxh|C}K25F(ta{h%q+&nR;KeMstT%XH$nB8H-j`NI=Cln4X
z`%pMD?LBP|QrL&kZkjb^I{P>emz3XCH69em`Ebp=$jf%7@h}2AeI<g=axjEFk|P<P
zotty6<8dPW3V3;5E5Jg^LWnYA?=846RCY0rZ{{z-lH9v<>8Lljvdk5F#$j&k8|n}s
ze7|Su3H$0Z5xzP4Cg)0PFgrJHUsKbpXpenPD>va}4}157eQiTP6hqNiUh1t+Q48AX
zR0+4QYhN`sJv%>enu(+R41=(*ujlCa8Kvj$Cqsh1_*_vTAz8Q$5E3e>-riRW`_e-@
z5%hcoB0qlOeW=Fv4Q&d8M>gRwQ(#Z>PT<-L%Zpn%MR_nub@U1Q#+(7nXua_{r<Fs;
zrWU5p_h-i}WEWkMuyi^!c{fpN`O)@Ms(a5nt^Da`CFyfe(>Jx*E6W$P=bgS>|2do?
zPuK@?!pVRQHPvb7W+ebxaUL9WZXfI>LMKqR806={a3G8EIm?UxKmW6DPDZ99<{xkx
z@rhC`YTfNya#p5wxDlDwG`m2W%0wU7x8`Ky@WSxzv=V#5i6g5ORvk8uNDdsxzAZP_
z&7<o_ooh)+*AdJ<Y2V%`4Q6(b9Bid0mt@TpNh0<gZ5b)7GsVFjPDfhHHcSo3f5Mz#
z-&t=L??dQSN>(gPCL3&b;KGp$E;!G=D<4NyGbPM#>^mLt{gPPZvcjZe-<=~|Uc7#E
zjnj~vzPz|`mV+g~r`-<W)(Ifktew6bF|qU0^Y*>nXfiNuC8|K<d&YqW@oegEMnPcB
zweMSjfNrUYhPBs$BqAaJo#G26=6!$LT|p9MEtLi&Mgi4B!2qs-&-EUh#d`b$jh3t?
z<(H$VX(tyfhE8Qu$QwGBEuMKX3vEzcXRvt^JL&{gbE>yD0S%$pOqy5iv2HU6d*q(f
z=r|<Yf#~f$0>YtO+gH8}HB?`&TIz?0hX9iEABIVVLLadmZLHV#SfV+nUH&K4Hc5#*
zOXLj2w%FtK4n23J)*_{O9Z0=n4^HMMnWLljL^q>y{RGwX)+eS<f`24&J=tbil$9^c
zYC+=N5(;)Z@MM*oRJVDvjQ%_8$&%d^BDNnlFMYK?hT<+F21APXe%saIOw%?*chgFa
zu<yX@Eq%&K=`Fjvo;jEwJx)(CENQmFWJOqs;q1i2^o;GvaV;+*Ms4YD6OnWJ>Nn|L
zj$b0&iHTv?x4msnSy?ImQ1^VY1IcYMJ-ua<?aLjun#JN$+k1mfNVNUgfFSoLoW>B!
zi9F~$ykjT|PZdOxf~;d*$y^5#OBV(yAfSvLXy`$i2ntU+?HtOD9HVCk+it{}%ke{9
zbFf+GDjsV0-l>c-TD7fCV0DcxWQV&?mN_gLp=>3m9Y~=ON6)bQJ8;S#=_W9b=ZIo%
z$SzV^6%+wHX~PJr=$t*>J%SQjArw62^p(hSNWL?v-@@^b7*{H<#6aw5K8{?1nck+p
z(~KFFRX`y)8dq3f?3ujrh8dmgJK;2o=IljENFTX4^+`L{K7gs!-coXI>AW)LIB-qw
zhTa&druOWa7j@VOh0>^;MwW7(tm$+s6=nP#c038aD!Dn=!i4GZ{?xQR*J$cZoSH@>
zS8VmveN%R#h7fWUFaJMlUo^N@UqmKCkH8VxDpVrjWW$I4419jdX{Q>F>SE95fKWl^
zo$Htdsn*KFJC5!kVeJ_^)gHkC+2D=Sj(ss))?+C6!cKQ+sq)xiH*oqSaNO8WaeSM7
zsIjg=zaPi!(jh){&Ku9MGdW|gc)4>GSe53-$d$FF_62v0+~C#VFc~(~+4=-#31rb+
zDRf9x5`dJoNJae9{+`(TBoeB`MZ8b8+>z3p0_7~FA4C=*@LepxxhqkUJt}!OBQt_D
zg@sRV+#jEPz+R}S3Qe5pRGdr%lQRaLVh3D8?0pLpSw#H@>w7=I?)o54QQS9uxl3;E
zjg3v&i;cDs#xqTd*|^gq1lzLJekiZr{d(t0il+j7?1vMfR80dP39Z_w+EI}KV?R=T
zSuip9q|;YBPXRM|($%nm4RtAMiBKHfezbm{XHXtw*vFi{r0Q5IMX$4rkNA1UaQ;~P
ztaXI3Y;&4{exh<2@TEC?ZfU3ectcEb&WYuL^cLmu1`&!KjTsetS4`VabRV=Y2ESwJ
zai_0)gF^b#K(%k_u>E9r_a5%v6#_!_X){x8CxdK$s!jboHvfo2q(_s9!!%vfDm3@g
z?XK&4IA6yX22PJq`_|sA_A?C;;i8dJ%nKtEkq;wviDAj(%UwBlm^zue{9BT_992JC
zYcuRr8bYhsw^G7!U^(~3XD7Ogc1D#}B)Ok!gl5A6#e`_({+E;&YowBNL<-r@x0NFJ
zXY;(%Ox>Xe4<U@qej$+<C1E!;-R7NkJSpr(JcAH`$jvxmzgXWTx*`i_$_uLFmZmD;
zwM5!Zxq-u!XTMax<SWD&BfFOzQ#oqo*Kpvm!G$uiEVM&@#YYaYnbP9%%Wb(P`nX2s
z(5f~&eSzm79(Fi68=7<hT1kPILdx%n7qVZeQ9vCb4GLLM{7jd@;|(q(tQ-Zcs|but
z751x5JYU-zA0ZBLVb1A_?I;_hvX3~f(0=XAi|fG`kU`mQ{|^J>cZybt%0;}YiFo3(
zGNoxxDZUQ7xbtjLlU2#Q?bnMokVKC0mzJo;n1q4fcnRJNS0@7!Lr&}yV?AKM*?oov
zl5#nH$!gZF8nNGMOeBs4nsM5xsZb_pVVWq8zWsLhMkQjUiC43C(1GYl<h74a+V6CW
zkiG^jxC8@v$aKtOgJ)cy^irAsNU=)_VZU3iQO$Op&-5Jjs;uAMAo`>ynYEK4qACaN
z_v$zD0bw3tlRlOc=s=3h?G?ie$BCr3;tclt?Yr0Xw#vRv`+TjfXqHj~73!Z`lOWi%
z>4hz~<L34U^@a*pfw$}3`W|iO{SI6+d2>}Gm)-vl8wpa`R02B&PEUnV5HZL}@c1SF
zC>Q%Yl>p*e6zZyLF10^S9<t^x=SrO$Cm*ywNi^xtJh%<Xik8rXgC)l7Pix^LF+M2v
zA*U}F{?^s2?9b{3Ee<hfP_Z3OU+i<3uZ(T<!4<;F_~#uCD4tEGe`WDg8QFovM9at(
zp->2=DRpG*?O$}u+WB}{aC}}4B!9WQgJlt`mu->f#BeDMTqD$iKh8GU0mji6zhUbH
zCCUP@a?=JUw7)92#D^ALscj}y!#d=^H3c_3z$q;D*KGh1r>cq-isjTHlo3+xz>@|*
zYUJUiAh!KY_kt|peo{`qkYQ0^UwFiUr&K_l(EGR9-`3YRO+5q+<)Jx{G*G~cFjt}v
zL9+j@ZrA(eA61^8`~$=dHwZR6tU9PRav(|!{QLIC`5p&@IL+i2VMW)s%Weo|w0|gw
zg2EIS0TTzJLs)<$m1gTnKuZDZ@vzM)GcIZUKh~#9Ov88>2;W2}#54z<Z>~%~ZMM({
z?Vk#Q!<Usz{vsUP=|F_C*Y&Z8{Bz><*ru({Wz75&%C?r(cloIOOTE4^OtA&AQl$=V
z=k>q~d5^LaE1xK2?SF0S3_>gLJL~f(u1z1Y%YkbUtFg`4zvVUM6la|4+%M~id;RzJ
zkRbjUq|)(uFcWxveIA6<MI_jNyaXaQgdEMxzXM5%9ggli)$>2wTxB1IbJ-_Jif|w~
zoGY%vue{V=mjz4ZP>xOEK#C2hs*ALZo%Uf%@rCzZc<#B*bsA_#rl)kO%**P28JzKF
z$YtjA35zjFQmm(EAHGy?KL7l4oks2oa8RlrCMf%erKak0To;kL6HZTROA8#UXdk(h
z+csU^GH^?Yc|KHr<;FMoOFaJUD)C6773`zxYgd9oL=dRl=s=Q)Fq%y1ZefC=!n^FF
za~!MI-+bGxqfSd>UtUDh-?T{wB2j=i=jQBVd^gvzh(-BJy~Gn>ADc5GLxhg5q1l`b
zgFe}^efzlf!4p#tKIpXKLE3*Z4n&n3>>7vd<GbmO{RJ)9>8-C(Dp|@tp?iX|E+lh_
zbwSA-NaDeLvkUf#_5D%~(yNh4JKEqXpA0-!O9CwZMgBf(pHzKuOeOR9Jl+xdv<EoG
zdYgT6!4)~A!UQ|;X3^IOZ!m>O{Ze_>B&_x+Z5_}9A|f8)Q6xt6O9zkv*Yn5GjqFn!
z2I}CD{(`TxsltGy{%LJ;QCO5ddD^?4n{k_L((!MfzEq5Jgz6yZ0T@Q;5?{RHn#=4n
z+G<TK5C!Bk6B#G^56s(Vw#Taq;shBNPET<n5r0jboUk&8q}uXXZ9?EwSOP(;!%koP
zMv`f2402_6S*!NhOC^qjYFN4kPDlJ8wIXpFx&AYjO|G8L$r)N+yyzb1I?!h*bte}8
zpnY!pGNPEtTBKQa&&UD*r#UmjKCeyAK<P8QzVCG9aED|l`2;145_M>wU!NGo%pv%v
z5#H%bN->%!{srCckQg5aOKBIQF9W%ZssEoNlmFLV`@;6AOodq5j>y3mo;QdDXGPsV
z9||ZaWnWYu!%D#f!g$0$M0Ybfs=@!8&4dY8&Hpb>OnLSB=Q)?cOO8C7urFz7vYAsT
zoz~D6Ru+voJ$4Fvc|n4O@yhOi7YCAvv?6Kpn+)&~3qGfQDWOYZ)IPKTA=+sN68ki(
z11Rz8fPHD3|5cM=_@o^pN{?s;_y(o2bL+iQdQ<^Wl8jV)`S|2;1D?VL<}WHWrrP|9
z`cW33j4(GoCZh9sz+nfTyJngiX7<W#QCSHOhto>D8!U{E=Mm5(D!;1XJU+Z}h^y04
zI1epAstOxp%uSl@U)_ycidm~jktSrm#8=>P>{ukoWV}~!Zac4jKjgOI6S~?v4t!`0
zLl0p+i2b$m+djH;3ZIVC%;^vDKg~{GRu2~$OVfu^P81#plZ4{~7ag9Mo!hX%UehLI
zqR*iSsuM!`kJBscx@+Vuqjo`F--uJtxz3%nK`y*GyRc35CbF@dRtzSHPmrU>?V?8X
zNAI6<8VwPGsGpJzR&jxLaeL5PCwsF_s~&VL)Xn4g?Jud{CW~toY;pHGkQ~(79)Wl>
zc4@wHR;d9_L&(4yDnIkG_Tb>r5ftzA<<H3?m}fU#o^2`+L4@&<(N2_^zy-?5Ej*RG
zuBhkkgBBfiTB+|`Wmo3q$o!;psaf1el89YZZw^+-M<c2_eTk@Bh@rQuTk9qwFPv6j
zpduU<%csL|*d@|Ek&I5-H64buGG`T|=0Ivk63r1&w5RR0b$7-aOS@GSI@!#D#N9G<
zC>P`ibmDc@mvy6Me|*;I)3=P`jIwLnNJplLgrJpsROVddPbO(#uWtyBsE_WAda2yk
zI6j35+vUJ>(qUy}i`yG&fKly;VGu9CM^j(XfuyE6-do_{O&!?v#`ZN4$C}J5P~RIw
zs4url5@1E{efiRh=L!$P=@>kz<?bg!rNu*~R!=zhrY<zViNC={8sFvIw5UMr^yK28
z?8{h?m2f!EhLP)K*VP1yQAuzo?9l1UFDBStKBZMTq05V}f2DJsFU0UqsMhuMeS8yc
zFztJ(nM&{hqE#47F3j5viD>6uc#m@#NX~g#EZ^$-dHS>nZ4F5rNzQ@9xOX#RWw1fQ
zd3=||U^f;VO(&xZ!uBGv7P5r{7qaL$NtO7Ijc`+2Im|)I*nlZB3w!a_0m+rSjw(&}
zQviA1ZqBW#gq<m;5vZ5xa9JqlE4r~u)RxB8(iDNj-6@1qtqXp5OZAD#X^N=m;PjQY
zaoLr&rZw*BPdN=u-lk1AZM3y{ec4qPU+i25Rnk_|7m(at>v(HEe~tvI=7sg@6C$zR
zQVgE5+wy5Qk6w4Hb6x(^6{G)<4>o^i>l!vpQkx0SIPDmNgscjBVGQWF_NIa;+(ofd
zlD-4W97wEY_Nm~`&OmG(WO|HI-QEb7=o*x<Pvpl&r&qYbP9o*ZM%xUc+NiLL$P#^t
zjiVSfH`)63nBF+aFiJ$#W9W?mM|>^A5=DvNJG-N8m+_uXu7iTolN403KC`&y0&(uF
ziE2rtc^Ww%%||HGTQdJa=dSq3EdHn+{ZybATa7(_#O_)OXT<}P&Q&rO1S$lw$l(w3
ztG&5HoveMTtXKMy_MUU$g%{adT5BEAqfRT{DfE+-Pjb*uaG<@li3@|xvRDtCj>JtE
zJrW;wpJ(mv>I1uF_&_Uu%3B|{DU5Pqhq8Ox<O~)tgCMa)DmA@{lW0zrr^@|gtJ=K<
z848VNLUt@Y<3O}-bNx{D>}@pwHlr%X;u_!U^wn%CXQG+A@*q|f!1>CgDXZSH&biDH
zfD>taJj#2BdbGFKgTvRu4j@R{=?e;k9K5v4dTlya&vLdTt=y5f;+h<*5hy%Rt{b9!
z_)8HE{wl^DxHjqX;;UY5<L!r2lQT{u#v|oU5oalH!<uZOEeGK-p|Q}B%Yir;HH(m7
z`yEnGlCtSEgB=Xpgqj&vrqCh#kZov_7VIXX7nnL#sf(&T4m@6`AXuUR_<PzmcAvRo
zxkou^K}-%j_3*(~PF)f|n5^$p)B`DA(1AF99aIHU4w9|7Ly$eD+K+qFV>pG8*XWV_
zan0pd+kLqqu6>1brOvhF+TEXDzkmctVSJ~lNk02EOOM%fQlhDGzUBbCpZ3r;!lu^Q
zC^vPGv<j@aFgr-3=j75#wB2blZEm2~iuRVgMqlE?l^0)bv)v686Jh6*Fyr(Emg+%@
z%2ti)w7Q6)K4F{lYkHeGE4n}7G-Li1EvL|ZNsI4jV+C8ne1N_cc3lUOvx&1Fulh!t
zYovG`-g~GO?J{k1GQ&K?nQ+>C1F>@);JtBz*rbh4Ukp-$N5-c_L=f)@8gBD70`LW1
z+FU7MhXYA+c=l0*x<}cZk1joL3vE$Jw<l_d>JjuL9lm91Zfu)vY0Z&|9&R(5THVpx
z*4Wne1zWLV;rpbWpE%Chj{3w0+H3@Al#_sbebFEXl4}7wux)i)Occ}AnF4(U64mms
zxJg6f)M<OL`%EQ`rFR?Yi)99q+alzbo!ouE-dSXVA})mI(uz9}ka(W_NvG|hHm#x1
zk!S#+95x~~cp_#hf0FhTDnq=wy(>|h(vaJn>slgtLO=`znIWOb-reSDEde@cFiuBd
z(s8gRn3{c64f!#4<PN7LDQ?}=X39|z#OnFFJ8!sgwR0^RNZHcqg$a94O-z{NL1I-W
z;WL?YAi9ryD!7L|+z8$c_dC}y;7l+`$*j4WR9EMbZc25CY(%Pi>(l2cQG~zt-nvhi
zDBlDH6P><93W^e_aISrI!x|2^Bf8UR$IBzTiH9d;?ut6t*R<6l+X~s5H9qF_#czZ$
zsuRz_hWt{3NF*2A*LH}fB6{g0$?7>kVyd5nTuPv!D6f58>*LwVTv8!m`f@ko*VFmo
zWEFpXZiGolEoZMgO-<+W;wnw&`#Ky}K44Qka3Xqxq~U89tW4WhT<;s|HxlOLDBK?Q
zU6n;@5Zxz#AehScZC~1B-`KtklGjqA!d>A&L|iYq7<=CPYko4`qbvXj-)LHp`I8Pj
zW<f=Nb%qdL0eiIigxjiI$?=U&Up({bYp%3!>V6}QE%X_uFSqg4S6^)(s2k)Jud`mf
z^<^$lN|qn2`xqR~2$^nnIeit_BCVjsE-B724D7rP3TEG&D=^E^(*(mi%~WmHF3hQt
z*|&5fJQ>+?a<b5uBV6Bmhka}N!P$NAqdiVrz_|!s`nH<c)XStV6Vf2pxdX|SRvd(V
zds~4ZYVraUO>!VPp4k~4b8g?!w%9puaIOnu!(YZnr=?eCC19$y@9f|!ewv|V4{vbP
z5O6_}^D2Cy0%4l&6ZTzA{HLYZoDA}l6I)**DF>C&xvP{xROWXq*mu{AX?b*vs)9CJ
zU!IMi(^-z#_tZCsYz+P@od!H%3V{?R?L|CNjoKady>)k^mGb$e>HtI&F30V_zOO!+
z>Y6gbA;D=f!UT@avDbkIPDlq->HH9fL-U0^$F@K<weL@Ubno5wI+r!a;Y2`Gk_GY^
z#v0m!{XoHU_I5=XCd9gk!;w`Mc)T!GJq<<i;IjD+eYY((f_7^Umo({kbSQ3V&8_xW
zUX719SK2rm6qCL^M>uucsMn=(NOrsgjdCzxIST+JJ|iW*GE1`TB2D|C#*ZgD2ID&E
zw2GifUKZhb!k(xn3~&E*B_*i<UW=xqI%PCJXiv8BpfLzFrOb$-?s6bQsN1&33hUe1
zjbJ+?$luVHRG&h=GlXc`uAGF0`xl(c8nGN`LeFPxcUv0BT07}C3_yty4m^Lub`#lC
z_4+;CxFXFY5>nG18(8GPQ*w{!=iLx(8IL(+!}d1fu1IIfk&#|ZSKO@Ibl@dX3f!!M
z{QDAT*HRdq8oGsT?>g7^w@JjFrE(UkO-f%(Rc5fj6vcf)agfiOI=c0nc2u%NA^LKd
z;H$KTuE>TRsJk`jNa`lBX{AqUzSM=?cCg;6pag|nD}_u;U-_gCyzn~d0m)<B)kS9H
z-n2t)99gbmNa1e~#sMYCuVUF=c*G9ZZ!}ZNL{%<g?sOp9>1W@&w0F#oEVW<}4(<$d
zyP~nThwvnsUaO|6_*plEe&-j7KO7T2(c3n#r*q=B&CffRr77sQhyYA-oDN_|n=CsT
zMp!gtdJ5N+@e-a7d!~`|Rn!c1TJ<zdh%6DxE|4|tSnV^VY(DR_l960La;H7ptxSeg
zbJD%jmk%nQoyRzx27`{*{40y76`B;<D+1>Tq)nfH+_`IrdZjHpkygUh9(%40U}};o
zyWPzh(V>TlPKu5qGT4cvKX<IZVbr<S94s%cpF?NMBQ;Ovx53W{hI3jPI`%a}XvuY)
znVp{9sIXsqzWb<S5hMUp5K|;F1V}zLnK6bcYUw6_US6c6-}t1{%tc&BoT03<r#p(C
z;=D?8h+YV36+IV02+mzO(;KJe>_e?Fa^r~8h$&7ZONw<NJlUCc+x4TPP9ypCOrKM>
zmm27{!MM|?)h6DP4X)rxk^AFzwjjbw#q#KAqEHPSNZ}9_m61oN`D@FC8&57hiQ-5r
z?vRU}&`sv&h`rEQt=s{)c4+54a4^+g(0TShSo4uIRU|@^wANC3epSqL?vl=~%$PBJ
zqZp!JtkrKw(&_iL`GCi}$ZiO>4tm-T)%WGQ5E?MjEBj%G11a(l<`5{bLTR#m-O(U2
z&vFXJ4|g+n2)0OR6@)}>`#laMr+eL+k+;~7<hz#_$2cfMp}G`{==#P9<eet;3S9Wn
z#L!zf+se5XqEKdP4$%M|;l~;xP4o%;a+<aGZd-3ZULW`53pxaFH!(juyz65?lIl+&
z&;CSBQksio7O1KdiE<7+mo)+_qh5jiWc8sNNaVoqa>VH?lvz1&C>hXCb(>IHNE2Xc
zE24=4*ThVaUQ67_PuFvXwcKP>KqM^OFinA!pQ&e(wil#Q$EksmSOU}206dj*cJAB;
z``I=#>q(BJ5nt4_;b%d^pX>TmbZHln-`kzO<Wiv#G*;h{5c~P|8+e4gki6aL%O$wt
z$}8*_x<Rg-6F3z8=D^F8LGc^qg$(Q$>zR3&izwPt-ulwID)Yb&`=z>1Gb3$}$Wf;+
z#~Yd9GV&{Qwo3rDUvA^&>oiXXvFFjJ@oq+8p0i)+mQdjXQLPFLmS+M;J^B!|Gj(Q>
zY2`2NhqeA{+r=X|fw8{O4rFM%NE9KYEc>;B$lxh)`v}UGXxlCaBD>4v?Ckw>%C=J$
z^;Y}!9Q5+y{pj;?PQt+HlXxC=_=$?Q-$;4#oR>RSxyROH8@Jy~W;p0?aV|UTluPKb
zz(iQ6{Z=<Cnbf41Qu+#eIMQOP{dW7h!6YWBzSFKP@igc!2iGq9oi;<^r3ej<FYc<r
zI`Enid@sx`l37wA``vCvy8PUrZU-Gm%=&Om-JQ@?#Z?joXTMkD*s1gzR*FnRufFgB
zNh6oi{rmMCZFtoZ(V0vUp2Pf8UEP6e)>cDPW+v<payIYkIaflCiG@v@w%H$c9{|&5
z29qVoU<f4lQwYP-#&+8uwZ4##q`N)i^lATc%Ill$kK1}Ej(}n4(=#X>NIfl_lAy+6
z`;*pJeheKM+4@4F6dF?8<WIY&N`e!DV();~9d;mz7`JVKUH@4<FGCbEDwDpL>?RSm
zeXi_qBB|}q>)C==gdcBJ$rK>*OWmh5%RFHg2us2tllB*lzt0mUq=PExNc=&5QgH^#
z{n%gT0BMma>a=vn@*+V=aoRB@-To?9V+L<m-)U$IoP87~BlF~6H<9joD(Fp<K<e})
z@=(lJj;VuUry)l{vvFMgP5Z_8EHy+G4o;8e=sHf%VDtQKTdnh6>s;4><!lrkI$(d-
z;J4r5T-TZ*Y=HIz@Y>&3U-*GQ*rQk(T3;H`Bg)eLq1^|oSMiAh_(579A_BIaNVB$o
z%=>O1cJ|R1EIGoA`9^uCLhOpa+GYd?+dsVob2GF(GSMAK{a!eYav5Sk+HGX<3F+Vd
zxnUI=jKpuPZ#>z7XvP*Ov4^<vFF8W+dqO!RMDGopzE}lOa#`I}idCYENFR!;{cFx5
zr{@iv>%`Hsi^SjlTWfL7-&XHh6mi*R|K4D8{mlcX6{JO&DVv8vZBjJ<kD6PFyea{E
z5S4+xygS?_$?r9ygZ7_oCD>&uB*AQW>(dkD!o^+!*5ztz4nr!(0!Ckar|?X|YMg7y
z-|!EB4Ig$k-<qT2FI-E6^N<5cj`{YtzujI|k0$CmtQ=_b-uiUhv+NsqD(u7aM^n9d
z4DmGT`;UpLLx2@vM~QCxh_gY*H_C#_jWo5~%kN))-}2)B?Vo*Q8)|%tP(W!$Aie`h
zuh7hFpGfb`^Y&3~DkMsQyhJrSefdFGUUG?jbUh5$ic&<Fb0xZn-2sUi6O$m?F|=CY
zZl1t-%04Cs%R2BcFfq>el4u>_8nMF8U9~e$l~BCTKDPRVsp&;L>Gb9Lun$}9tS}a8
zO%bs_u6_d}fo|;}ZFz^&7r#;Tm#{Y0l-aFteEaybLC<gH*iGkJ?vAPX1^a}vA*J?o
zE};tCIUTvnU|q@+50{lntaDw#4Xo8Co(=5hdrwgmkyfl}r`a*85aaEWYNX;RlKJ*7
z?Qr^Hq>@LQ;tY+6(3Ydns!y&FXy#<wRXGyDr9kAFMeKj)u9!>g)*;xOTH0%$(vSfo
zeM^grC>;k9cSBQyAxjnh)b4JXc6AhrN^l)|;SmQCb1=NxfulhU!j*{g_|w`rBbKd&
zS7fxSs)?mFnY(H&RWMRJQ-Kc8DHw#cJZhhwYw@nv-Q!%=t-I0wD>q5PA`^!OZ=aEC
zg%+Y?-RCBpj*J%S&rn$tl}!01pIJ}Iqhz#zNE_1cm_D77*5ER+DpoR%hQge~fZ0PQ
z?Xz+~8>c3m%fi=4++q!P+GlsO7Df(MgmsbK#et;6FuKJ)r<)rMU$2rISQI%|RdaLD
zxhtNpJ_V>iY3*|pvrxB3Ze8!Ra(muB`+$93Yg}>F<xWHN7mmAt7x(jX&~JO^-Ogn$
zb;OElX#4F8YEFiopxPptes$7;7Xqt_8v}Xa5tau%)4s5Nsf7!>UM!!oSb{h@@cIPe
zM1tP!v@hz`QJ<ahS7CPJH#&=87_M~eG`6ckK)~&b+Z3~QQv9$b^JG*p2ND>fRh5@P
z3B{?Oi%9RbFFE^S9uPDd@ehQma5@~Dn*R7cO|kS`rCUdp27CG0u=B3F%eks~6MK!k
zM?0tbGVV!z&pCa$B)3gW*q7#-_wGj_<~kk_fiDaW9%|=SUppg<5G1EB6)A;Yt8CC#
zdqqC;lB+K!gosAk^G8*pzHhH=Xs<HT6R>GIG=izYCdB1ebxTVG9&c6nSbDj&dk;8p
zjY>t*XE-`1w#cj72=1bE4Yf(JWY8s&c<+Vh*m*VPz(?4`33Lqll4~KQN<I^>OnKw{
zf)t&Kjh56NNTG&K15m-FqxNNO@@|<Oa~iS`%#QWOF&OYIuW9?4P~VBIl=6~xMrV=B
zI4vnuM6F=vXdbW&iWKW)qT^0W&s>;^_+F&v3zMrZFJk?g9XQQeB&s{`abTxi)bK4;
z0+iIW*z9v4ad(Ej2*;kWi`xey5a<wPSysaN;6Tph-F#|S?n$gN(gPF+l2#~T?RLAQ
zap&G<Rj}DLaC)=?Z|QHgOLLaU#qUoz4UrY<at-XVdbf6?;seQ7V=>q{4kRZ#c2nr9
zc6s#$3*hZ?T+*kf^|n$M^onj~Gb2gzy+o{x&^hG|F7ox#xXajoc4dxu)pXCfEc}qA
zeaO&dSLKI-y9B^4aCETKQ=2%lH4s*GaDnpSlgwY8U%$W!e|@JBHwzi1zekK@<yLY2
zmR(aX=Ih`YLP9YyF-RbZHwXdxE^2-Zf1SNHUqb~hs;1G%J-2RQY%TQkb@gL8Z<L40
zv6(z}Z)BHX(Qx1?Tq!D@8sBqvZBE8{=e^Up)EpCfyU|{s*VL$<bgs2;=lZxO><vQ;
zrbDbbUE5n<lU3m`_Qrg`y53yRxt6BCu0N;4=j_Xy5CT&_?X;3Et-5`!jWqrPf~z=A
z@T%lLAf%EsG)3Gc9&-B)qj#)XZ`ZYNg2qrvVcWpz(IdAakq*ept>>?sdcbKEy2%a^
zQlDMlA!izbpTZC<E#+|sQd@Y%w3MVc#$h+KJvvwzn^<s~n!ih~zUB&B-QIfDC6_sk
zv~-~KVQsvz2BAWniU-0p=Jcib#*489mwiJNeMoJ$n{v51NsPcns;SV{hmM31=jLwh
zsd&a0sKCxp0=qZ(e>c?1OuyH@qB!&B&C-)-#kVUpQwS&%uw*3_pNO#2wn<qj?Enn@
zeoMnfil1)<8`sp#%32{xs(@zs{Es-0z=$69SG<yI>&6hh;EXzY2*Uzue}m+-r)ZxM
zc()c7q3#h+L@PJKO;gkJNK?19SC=?^KpP=SO#l(Ez4jqeZRh4^H&CTzy#LS}IJ|pV
zma%Ieg7?0pR~zQqht^z=U~^;MC*-?(?a1gY_v%;rkh?d&v^nUSTA#aDtr?A5g&x;l
IG>Dh|7q`q4GXMYp

literal 0
HcmV?d00001

diff --git a/models/ggml-vocab-llama.bin b/models/ggml-vocab-llama.bin
new file mode 100644
index 0000000000000000000000000000000000000000..c1a3194c4012e36899b0bc0cebc47aaed2a261ac
GIT binary patch
literal 466955
zcma&P37A|()i-|FcW5@*;SvH#AcW4oK;X`r1hO=lfv^O)J>5Msoy<%()6*Fe#P@;=
zD5$L>qM}~hQPE~c)(ev;F0Xdncdxkb<G#G8-|w8$r^5Yz@ALe>c`#PpsybU$ojP^u
zRQ1G`p`pD9e>d?z`X_nI;Q#l-|1PMHZM~4c?ZcUxB=_ayd79pz)7cxcSt+sH{8Vm1
zDz`9|Ta?NzPUV)Qa!XUWWvSfqRIVeH>rCakQn~I_t|yi2eR*zviof|O{^qCno1fxu
zeu}^ODgNfC_?w^NZ+?ou`6>S9r}&$n;%|P6zxgTt=BN0ZpW<&pioXRZ{uZS8Tae;!
zL5jZxDgG9u_*;<TZ$XN`1u6a(r1)Eq;%`BUzXd7&7Nq!Fkm7G)iob;^{uZYATbSZ+
zVT!+nDgG9w_*<CbZ()kRg(?0PrubW!;%{M!zlACO7N+=HnBs3yioZoE{uZV9Ta@B&
zQHsAsDgG9v_*<0XZ&8ZBMJfIkrTANv;%`xkzeOqj7Nz)Gl;Ur3ioeAv{uZbBTb$x=
zaf-jiDgG9x_*<OfZ*hvh#VP(4r}$f(;%{+^zr`v37N_`IoZ@duioYc({+6WpTaw~$
zNs7NEDgKtE_*;_VZ%K;3B`N-vr1)Es;%`Zcza=UDmZbPwlHzY^ioc~P{+6crTbkl;
zX^Ow4DgKtG_*<IdZ)u9Zr78ZFrubW$;%{k+zojYumZtbyn&NL+ioazk{+6ZqTbAN)
zS&F}9DgKtF_*<6ZZ&`}JWhwrarTANx;%`}szhx=@mZkVxmf~-DiofM4{+6fsTb|-?
zd5XW~DgKtH_*<UhZ+VKp<thG_r}$f*;%|A1zvU_ZmZ$hzp5m_~#a~B?zm61t9Vz}g
zQv7wK`0GgV*OB6{BgJ1wiocE&e;q0QI#T>~r1<Md@z;^!uQSD8XNteh6n~v5{yJ0q
zb*A|1O!3#5;;%EsUuTNH&J=&0DgHWB{B@@I>rC<2nc}Z2#a~y7zpfO2T`B&$Qv7wL
z`0GmX*OlV0E5%<|iodQDe_bj5x>EdgrTFVg@z<5&uRFzGcZ$F66o1_*{<>5Ab*K32
zPVv{B;;%c!Uw4YX?i7FBDgL@s{B@`J>rV04o#L-2#a~Z~zn&C-Jt_WrQvCI#`0GjW
z*OTI}C&gb+ioc!|e?2MwdQ$xLr1<Me@z;~$uQ$bCZ;HR(6o0)b{(4jV^``jiP4U;8
z;;%QwUvG-P-V}emDgJs>{Pm{z>rL_3yNACQ@46ZP&E7ipCEgh#IYAUYyEoFMar_f=
zsys&I$L;bw%H?YbUn5CXkcwm8adP`%qC_jy@a4Hf^|Rthqe8j&TpMhWDEk~E%Yy)4
zQWX%8({Df!Sj_@aRlb(dK9tmWiRd={hSrVecq=QU;QQSQ0k9%Tjo{ly4b~)4)6!L<
z$E8N44tBKF2+<2xK1kHE^eEA{q(%vJC-${M^b1Q55KSvI)cyN&yaO+@oalK;jtvp*
z`@9Gj{2VB$Joq_OQYGNcNX60^7&_+pwrsg0j(@%_G06F3sXYc7JZ*={Gp!sjW?LHV
z=1PqkYGe)E1PERvp9=uC%qp*eS3tgwquhV~conAWz?Ua@+z1muw%Wi}AgQ$ipa@y#
z6anz|8-m$HSIE~qWV6}6cA<|A1UD>NK_AhurO`DbmhQknu=Hl42}`5x4bQiSK%Ry3
zN-OEdfDyzM(0hx0-A42p`C0}!uhVjndz+Go_HU3>9pCP-Ht4=L$=4FvG_)K;?5&b2
zqs^UG^K7EK^efujEva$Lp?4XpJa+G~3=iq|N}cP$X(Xu%2I>0^+B%{S>cHTh{IGrA
zjxjBjYk>AKNo_;w6Ot+*^+`#Uk!nh+juG*owayVeZ0Sy-N2Nv{<UX!-A*?5?d?nFS
z@^u)no@tL!@Ec3A!X5X#BqvyZ?38AsGKxPV%_`vV1q0Si^m+MO#*q1<2q8Zq)8tFm
zfrG3CTGjyRuUL95(bufjcAm4}Fewgzh;Iqp5eVixLe4fO=X>%okGc5+Nll<)Vy(&$
z=!=%&srnPCGhQb8nWX9<{TGIP2GOsq@+4-@Z;X@zH~pPeTMac})wsw0D3Hpa<<C~7
zhv=`C?uD+98ilPy|FnjAKzvC)vvU2nq)O->n&P>a2akJAiB{p_y{Goo3A+(~KmE*z
z_m!kfhyA7w&>9oNtZet6+E252^z#AMdM31zrOzNbNKjSK(FaSaj$wa@m2ZOHlCQ%6
zcbI%F$VfQcYOEqULJ44zXCzswL7qoijrBxF>1Uo0M+>+rRPr&7I@CVaGCa&*G1Z>q
zRn}9-8B>#GmcsE00M$+~43OxFZ43Z)l2!o0CrfJ1`9!BkY6Q?uojO8g^8ZzKZIp!y
z`1h2fQ6M?pkOV|CjIJ)ObcVnzW5&<a>a5VuG(JY~{VZ#HBhf4?;Kg;eWC}VV=D3m@
zh|X3*(HrMTs*Vvo*T8HcI#<g9e4dtr#Ph8DVypl{N}09(e1lsA+Y7AD2BL*lz?x{0
zfGI;$E|#X1ad5Fj8RFzptIT?3nPtwvz*%nnGBM1`ZG`~Z3VP1BTEigo0z-T*(MrX@
z>h?ktWj|}1i!3)*L78vu$57Ov42@s}ciMMe=(_}60aLhJYK(K2^k`*n*DK}cokz4v
zQWcE2)j~^Yh!@*6QwQrfKwqjByLc-5e~*lpkCOj2&FkcTXg8o>*9wZ=H}AfMXq`}#
z=gF{MXejR>+90Vp^RRp>U6}2c8e^?zi8fk>)nlItW|B>aek<b1eVIV7)KTkllQE>R
zNs%&(SJ>xEh^~}$>*-dL=qmeuInmWZ#1I6w*?`wDDz1^wdC2Tqg~v6nlXjK*2u5ch
z6<E-Dt%?COAgLQbTR~~#`CgP{nY*kc<>z7`3>vQ1gG57?8GyPQmO3L~q^yW}%5M?G
zyF(^^t9%|{oj;;Ac&v|Vl1Isyf$PWjihQq+fTpT(jiZ(8Eqxi$xYg%Hpk@W=sR?b%
zDz<JzeK^mH?=~qJDHn()r2*vnI7}1ktOBd`?ZU|f=-eUIMX}r{ALq=4<zi^ppzcjl
zqs$h~t0Y-sPX5cV4$9yT71cA4Sr{sCOzfMD;;Sjtbbf$^eT%@Ufco$jL&cJMwZbpp
z`>UmCd5CrUYuXgx^J@&^ETY#MOL_EV_*w%b=9Fb>!xX;G+FU?%s}*#NQRp(eZ-(6^
zNb(bT3N=~g5%PMg4_Sq;mnyQ-zCn_e@gjwqEKgv7+^#VA>+MrV{jU?&rX=PoluLMr
z@i4%O=Zzw{N`XR6t;BNwCZnJV3f?3L>P*?2C0WH-3pJVN#k<j_Yb%BAB%AziQM$%z
z6uw1-(R!lw49rFafgTCpYIGD!7-?@)IsoKtZFtaMcS>@S5#A}~dEk4yq?oj~D-K>F
z?=q6Hp$PBN8tC13Xe)5{4#QUGQFXUf&jXChux<8EX;FtbHB|*|?-D#^Ea;&oWkkJO
z*?Jmmh3}RcgFFlGQRE=w9&N@m<UL9$1QTkq1i{dI1p?+VAiUQos$menPhcPwzE2zS
zq>JR^>^T%h(yqpGp?l@)c%8y~6&(xn{Zaw_y@JB`ONBg?*asw4*+$_94Dll{vL)Sm
zh>7^138~1SKV&40Zl};?Cc(~!T?s_;VFAmF{YUI`9^(6mkW(MWbpEJKTo|vA_eZ5<
z1+PW-DSJGb?latZp6eeIpo4=HYN`yWd|YajYGNjYA6ItSCi#Tn<$>}E;b%manft8(
z>L|Q_>Oh?ylfz;Ud{O{Z03d89nTJmq8C4z{pE3ZWSh|}g)hj7%3Vk!6xE`=7EWihp
zY`}R?$%CaCJ}4gx(2frYgbEMA@F8o*%Kc%(I*#rOAC|TyR`riai!CMAPmfroJe!M;
zTKhf}J*wDv|MQrA$CfjEOupyQnjW{-MV=y$3r$1l<xdNS`~>#*;inZ0cm5MoGyhj*
z@_#S-FH<jhSZ7Zf{u1bVQh|@~?&K*!Id2|?Ps!J+m{w2A*D~h)(?%dG7<$IW40Im=
zJtIvpRj{r+D{1JHCj2)-agAHV`uR!j>sX-jKzgpt(j*)xnjFH2d|t}0yAhjY`B-ID
zI3>vn7V&UOu-C`hCh1P&saU3Pr&P<2gQZ>ay{rRrm%!wK@)@P%k=D}`enx5!K^XK|
zg95ud{H$@ijW;7NNDIjJQVL&?7UR5Te$HC79;WbfQl~V=o&0%|&SYgAqH|d_XuhB_
zty1^}X#>p#W%@;dyS)nbzG&#qBKmJ5gss-l<?5_eza&jq8H8V&I!MPu$CxUfFFO*-
zkaPHDsaoEFRxN?P`xfq>mV6%Jo&9cOrV9(?Zo?{)`HD8imO9jAg^iD|>en$a_*H3J
z)eXYe1b*vDhOg-fQ8Qnc`gK;aUl$3KFy+1>sXD%WLz-ci4Ox6ss>)>3<nRPK=v(r&
z!j-?(#<BYjqHi1FW9YYU3%mmQ>^t@uo3ijb#t#o4`mR(0$nd)YvcOyG@5#p^H1+qa
zWgYYQ`z9@@kMB#B*;s;qU{xvu6#hUyv*P`sQi!n^{?IT&A!2VY1tl)fWSKo{Ka#IC
zHY9!|FaV8Sw1(H$A@3KZN|{I7kFB7B*%1END5wI^PpoLHgo*SM70wep8GkBijNEf5
z{HfsJiSaX~6B;c1nN*k<<|+Df1G9~l>(8A+RvY>kR?5>k{DrnF@dEP8b{e*H_)BZW
zAR&@pDFbYEX>y!53cr>X7=w>c`0G7zA*tU;2?+Kn{Edy_*0a#w^jl+@SNh)yGbQXn
ze`iB~yd(oH{GFi8mm!$n8|?fTME`pOum*^xrw-8>GiZ7woEDfj!iDz-NzIx|;U5Ip
z)O{H6e^j_^9Q{$KY&D@m=ubxF<S;hN;h&^w89n)DLpOvG_-AR67xnNLqiUE};=fq6
zOJM~3)quhNhJyI3;&~EsqQ6<a)?*a@O$tWyya@e$4>+iZzq@Kr5&gr8s=S8(LmHLF
zxUc_dh_H~N+x{uMl}2D<{g1UN!b=zakJTF~^Je)asW%Q+Pxz8RE06Q|{@0#1Ai?EE
zdEEY6o4~6T{@Z|IaG>NrHdYE^4}|}bn&L&Fo%?C;R4e4u&bDb#ua#MS(9XS4SQ?<1
zb{?SB)XlTk&V7)rOpcMimzC7SgwT=#mhX5k13eQRz4y+2(W3Rp?vQ*<kH})aw^f2=
z7HdJd#v77-cJ6N&N62@%7sIzvxR1~Utr72|gs>I4uQkXQi{$Ss1?3U+dpn8!cC1NR
zYW5R?V1rDMzn@edVVgrgZ)5*uyx&f2b@*7}|K&ek{+UJ8y^8$(g+lD%!u<vII5dB(
z$r9!~9U#Pt>$67w0a81Use6FbW{ngdAd;yM$bRxb`JTrb;2&rbt6|)_%%)+A4-}Tp
z#UI>v0<wnsnqqx*kOCQkAP$nM&<jvq2e)CPGY>Wyz8Ky~p%#=dY~ywvUULtT>fA^E
zA=dX}B{rK5G0LVQinUsqhu@)XQ4gVsrm;RlAsuSiilsq{4;6~baAqAQ7+B-^hZ%-R
z_FRUC8HSE=iVu?pldy;nm((c6iKZ?%kHW*P>PE=YWq4O`guyJ2VM_T&NaF#h4o&5E
zP<(_@b_RR2G6t&k0Qs3Vaj+`g_l^;Y_4^1P4;?A|wjRd9>>nxBCs<)=vIY|<Zr6B{
zwe3+x4m8my`9}#ktj3R0Fb@JutZ9rbY!;5T21VX7`$rpf*hqv&OY3W9Q+%|sz{?&T
zBV<%y%=pJx{mVd;e#da9_!wylE1Hg#%JX2N`o~)3*1Z(!_w5i`?7lz$arkAWei@r!
z{}lqHd;^90k#`UAD-6i)SHo><)KBJl>K!KySlJw>Bx22pkFy4Bm>+K$m<;~$A`{4A
z9)-tC1Kvc$#~aYf7@OE9Sd$v$bAnVW;d^+3w8p^zX21z1A33TxQOKAxhkQ-Vo<-q_
zR%5gR6ekLp{5IINCrNSwZh!xzoo#clHY|&Gc#@zliR&;v$;eV`>txI3N3f^$Pj*xm
zC_Gt6gVqIUCmZ4``H<rj!8Y{}u$^Kpwxa9w`)s(e<5Ns3uY@DpkYHPk-ab_rgYFAY
z-PvA;FRi!tj`67iy38i-Y1RxY#&+^gGfdrBnog5y6&CbqMp}&z)I7@#BIkPsVzk1G
zpetmPC-#iQ>)<;#;8XXJf4UUpC!kM4O%L)xQhd6>E&v$KkXj=Z9%D14`NX&^Q8Nr_
zKilo`3}Kh8?=ytxJe=IVChKgTYjTJYoFP^D<Y=ZXdlfOy{Fx4B9^Dbn6nu3ty5pHD
zGTup^X%r4ax%+1tgBbf~8iRcp%&uC2UD{_!5e(rf2LD-xXiFWFDm=>oR&iz)pQVBs
z8$zpDk}eOzwdl_h3dTzi@hoBCrkg09WsoMy><O5y%wvD-&z7dGr+42@;cTll$d1>z
zT?;!WnxhB`*o66WOdNw^bA)q@tTHc(bBwpvTX8I5nbjC-XA3p##naDqodjnKEfvf;
zI!7Rl=<M*%QAY9@{pT1QoI%Ft7?s5etLwP}b~}duTp@D2#KzoQX<P@r@my=X7M!1}
zwXvw4YoD>{3(qwk#&C`mpDT^G@z!9TisxZ0gubRLd~g`fvj#<Y!{T`Y3!VsQ0y@tC
zuNfr&JgZmZ4dr=?9VpH-CNQM1(aKsDGQ^0??mYGX*Ro)PqHO}PWfEr%Q+WbXnJ+Aq
zSfS6?zALe<GvD-4RSk!Dz7gAcA5MD=L=g(#Um!(L=PaNFLOBld;{`^l2!a+$!MPWb
zztCD^^ia4^u#V+1++t0a$9N0B$Pl8EubKR4m2LS&)_xLJZM;YrC>Nkl7i%9ph}F>7
zH0(!Ayv14t`^Q+5bv6u_*zm+QfHm9_ZT%GcJC;a`$~Z>R5}^b8S@v8mHKMjb_)Db$
zEE7z*rP=_#$#|*3g?R}3i<U`I2}_f|%$l#o;9e#Ix%NhiwZ=G%7h0~QV)gfzOL@;m
z3YVKCYZEYDVy!b+hF5rn)Pd;y6(*EzSWdzfMgxWk>!=k%>Uf3Y3(l7`98}}vpKtXr
zx^T1{o^MbF*t&4J5+C<mAh>SYOunXOTu9*sR=<XW#P|YZj}OCET2YbLx0TWu4))M}
zpTNd8cDWoM#a?J_CJ^Z0U+CE8!4Y04AWAUS_f%t}?INotW5~Zqdx8hCCP$c+iwts&
z&lY?E1-sYxjpRJsxV|(Uz+i}d5fc2;sNJE6v6^+5xC**RcNo+H_Q<iz<=HpcDZujE
zp+Pjs>rbf3`Z%h0DiQExbnO8vmSmSS#p+67msGlGCdFMs(dA_r;M8q+*%Rq^8xL%I
zcAGThd?0o;v8%%Qr6I}>uusD85tt)<wjcHg(or5QJqBZ>jOB!Sr9@PEuc?LdsJI)$
zb{UrKxYx)S!&F=)bR*?!Y6U#!tBmT&;R?m8jF!^4co<d-EwIS0AYaq@2}m_uE!azN
zRm7`}u7MJ72i8c*)O{GoYYg-_j*&wxki~C}5yc*+i-oAO&nEw3``mgLg%=Cv`Usnt
zn$C;2?h*$SArt;3hJ5Q7mg&$H*06xaE;r1>YpnnrW4pp%Yt34Zv42gAF!`W}H4T#v
zN@ATfK>!Pe%R0k5%%0_S!eyP!SywR5!*acY%x4Y$dTX@~e1z)--tG{Va;)iXW!`jc
z5XNyX#sg=AfM)MPxWRe@W-t!{EgEO9NKT|S3@PUn=&iix%9#W^z)8+<moSp(QYk8p
z@*%3P8FVXif2q~Wvo5|=&_mnfP+F2T))X6Uh`|<QKWDhn8udYI#~X!)D$`GWRx%Dj
z`F%!FAErT{31_5I*-CMrw8{_gO5MM+eKHTfg-me2-{5S6`y<pGb~_dHU)(R;7ocY@
z6B?^b<7N60N?bqYi+PIm<BSzJFBfcZv=_<0Tw3Kv;dZ=STHQ34;>(3p?9O?5Z8Ge6
zo|2o4eK=G3Tsd^bqrBQ}5*QoNlUE25UYE$fLcY$y9J|7RbpUyMg&-g2z2ucvu$qm`
zE2RNkHKBgreY2{IE0tPy-d-ii3Y;UpCJVeZzslq~!H(~EPj;9+M^{@Zc*j7#TCmgy
zDZE;U-TgYOt?|{ie-Qp9W^1z)kLRn{*l!lRI5nnlv-N1}9*Q>`1L9h~Mr$z^O|m5t
zUSkcoKd!L`62Nt>&{6?{Ypv2sEMV6PoN>gb#G0&g9NKl(v=<u4zs?$9Q3|gUQU}Lj
zZN{2zJ%yo7flZ*iY&-lw@ZE$Hk-&O%RPDZ4i)PG0Y=dBjSwnt1$@}%ZVaF_`IB(5i
zK+1vWfI!}`o_tN^u~8Th60uCOWvv+;4f2F92#PKqwVGtBFDwWb>&58O&*J$j8s-6D
z_KVV#cN<~RM9_l`eo+X>!@*KAria8w>6fHr&3Xz;)&LeqToP>gA-2K?4F&dK<PS=N
zL5xF94M6p{8a3YM3<(iz9r#1mv<{UK4jG2l6BKK$67~!<EDc5~(7nTgx5DXg*c#Vx
zWEyLwJnTCv8ytAM{Ibxq71K^rvo}**7J#E{0&TJJJyOQ8gzvI&zvCcli=abn&YLOT
zB8|!gaJ*GMW8?H@@->YC2}5YBRADD$yw&=+G9X8jG@^nml0RZ}ck&^^h!If8!(xwJ
zb~Mkv-chS7YxStrt@7DzIBL}0L9teZZ7j2IOb8i+ndFZN-gz@995b=B9->%lOnneu
zl!|imFxCLSBDMJxry{lChGdmlQN}n%xGJ?(TAFS>$o6&Bp!eKBan;C{Gv@1!Y;mvn
z*DJm!A-(W=YXwadUoTkdqv)1#`QF}ij|()8km40bi$EA#=33OJJASGe+7axM{hIIs
zM|4=T&dBS@T)d|uL_`xt1oWRjA=F{kLBOtH5U8}EET;!`6;qk~x>PUA0bpI|uS{b3
ztQ*TP3xHvpWp!V&O&X13HKfoLp@(5|wTtS!7)}~HRraAwY9Fu<f6~AWaCk{PDb&CM
zqZ=fJ9`J8a6tY>;uYLU#x4)Ko`PeQs;3x#snuZJh77Di;)*?@IEs;?`J4{YJAZLfG
z#(Tsaf_TnMiuHR7F-NRJZ<Javj#-WQHwwy;JhWJNqW~M1Sfcnw;SH4#<s^0Fw8?*^
z@mb2V59pPGq<}T^mC|@XoIZ4uF#^+6)38m>r0^yqmleCqtVGcAs|4NDQ&<UJWo*Fs
zrBDm<@Quf>5~%rc-dFyYrbnUe{Qt5Btbju;*?lW`ja^A)5Zk`cXdmPtEiV)hSVSR)
z&_J-+8EA=~Qfnje{bR!%g~BkalTdgml-bvpXod~22BNu?Ha09gO|NAW-@n<Au$>*;
zEF9P2X$dvS+9hg#WrxqrruVBA_T?lu3+j%3goWw2gRb&!u}av?@j88rNf^ddaEmdl
zUKp2!;a#NY79mHB5*Q3^m{76FErMum57DcIlM?S@y;loT9Kn#^{>)2$py?W)k%gLu
z)&}%;x`?B=_|*ciegh?%+5(0B8l$boegf|`M)w*p@fsZr_mwEn<X|1s(zPhEG$Yqy
z0B2EJ#vucIn4i2x=*Doslzpur-%=v)wel4V!+Z2vLl66O8wFZv0QRF+n#?0MEqbjG
z1-A|0#O=&D+s}#C!U+#Jd7U6Ykh%9dp_{{B{nr^ko4|zD*oyU3Ym8&-9=%RS2@Yo|
z*3WP>Q}Q|+Bw|M2s*-pV^1oHk3}S-ZYPNOh26fg1w;DIlg^<*(3VIYyoyZkQ<Yj!T
zfqyZ)hmu=O3&RlSRfTSI!)dsT<16nrLD~<4?lx=44yfQZp^M`v!`lpZFGqhww;4$I
z?s#t!-)54iVQAeZl)>u>xxHRsFgWk^f(8*=zJ46xk@b41J5tFP5m6`=peUl(8#>s+
zMT)gxN1l?`n^a&{>DKZMMlv6Nd2bL9>{5Dz0G5E~K)<f(MtGu%`U4e;-e6LID~Qk2
z93Tt}O5WfoD2>1cbGyLB+T`7C1%+YM^KUoCVQ>ex8!c?&gtsf}+&{MqG2<Mk7vC;;
z`ACreDA8&q3=6trPkYSzJET#mh`4oM)2%142@UQrlzq_6cW58sh(EeRXuJl853WY*
zEtF`}G9O^l8>I?NdQVgEllgBH2t|Yw1aCBvV7_oP$Qun)>pc{`QOFq@NAOnc3N}Dg
zZ#1fhc>Db(X>)zO0$#m03G5Nx^lP$$6U5+6)_^s1_$Cu5ybPG}uCU1dsW&N4b1Bi(
zHu!knYz?5{FlF8>bU`b_AF62_&u*o_RZ|=B&4LvB2A%@Zn~l@_I3MW8Z#GVc>Xc|z
zKrC{|c0))A=8@MBOuKJE<c;4D0Kg=R1_ccPFSZhlGHp}ZN>M{3P==z8wF-lU>oHMa
zt!HrhPHz#w!xK0z^)wA@66V2Mb{?ueY2C1}h6>(dlw&smHp90V<xm=t%k*GQzQyD;
z!Anu1fMGn}g;-ZXh*(1mv$qNWOaea0d8;tnvz~&t3gP4M{)cZhlqC^mq?I~=;;p6!
z21??5Nfa=297e+1tedeCljm|pK4N;CfrPgdb%Q;nFbu=D3GT~zDSn$UQ7CDERs1%a
z|IiEKgH7~{m?S#r=uVReF9&xDr*gRB-zi}8n8m@JCL0_HBf9BM9UQe9c2*SKX~gA+
z@*q9FQ-qSAq(oC)FiqYra3|S&=Dpnzw(cbV?Scf-pg`*s%HSi^bae!SEz&9+mJ(~S
zj2W4{-NcoHUZJ}LC#JP`mtev!84J!`Ms}@^gQP%<uzx8cvRH~Zf`fM=k*ir`iy^*C
zS%+gf(Ikvu#2_2W3HYSEcL<_xIl<O6hBvg@JFJsi&uT#MJB*4E_UlBhyjFspRiyYG
zA`ct{;ea<$NH7<m-|jZS4oaY@cehl8KOgpkf46YF9c<k##LAh{-9{nrJR_}AFH@{l
zIOHo)AZ(RIh>;c`2s-lKsT?C(&(~z7xSfJ`8a!U$!*>eD@H#v~kya{UcN}YS8>E_O
zl3nujE+bxTXYXCocpH0>{dd_!RZknMnBZMP$}9v$ze`A&fn(w5T_(C(o~?X^(RvgF
zB$|dvi};Us3v6hbSCjW{X<dOM$!~wY8Px;*oM+4O-GXW6Jc=|mgq`ENZA`)skhDQ?
zm)|4Mw!`%FG|B#1{~kvcADjmFNaK+b?~FsOhEYt>JvI`G>h_GatO)-@a*xWg$@=U)
zLeMBy%lDYPhu|0TU9NNkx>;*okIC^KZTm3TjoRr4!DIX$W1<u0-g|^{+yO##y41&N
zic$Ms!L_l6d`-av6ueie!i9t2bpx>xOTN}v*Fo`njjk>TCDBSSI02B}XR?^&s~z6^
z3;-fne9hz$EDlZF&cL>Vlr)3AF*+oApU|`e>xu&=;Wf$o3@sd3M3InLho<*5xt-@h
zB+T`*Q!t23-IdxWv|(h_!4rVxMcYX<gsY1KC(%oZLSo&CV0>ZY`a0j^@a{FRnD>Y&
z1=4$Ma3ca8V<XV&m{c1e!+V8JXljZyIdwmrg7?~(=O{B>Rue7W#slsBM(8Ln>)!iC
zBsDmy-mhH4krKS$#8H6C3*T>&u8mUUnv2SbwNV}plH~m&pOMlaxcY!VMF5ER0Yf)2
z4nxP+g4r`E_<*1*VK9cS&M5CjA}yJ^7wYi?M#y^BC?Bw44l_t1zCWli;I#BUXhW|^
z{s*lUPk^9Z!f|$?Ht1)G>1RmtB8s(v7zxi%@<Abf5VHD^5swq=0(l=2LEr!Y!_ohc
z;O6t$U{5vKh`B75WoY~l31z$Q!lk3w6=5F))$$=Bl&?O)_d_2xIp;8aTn776|HExr
z4N&l58xRw6Nh$oW@mj)F#7J9>$boJAVc}z<NXdtV1c|Doj~IX|HXRrMA29$t-u;gV
zrt%=T`iPN(1MN`1!v#u_YcL8`9lKl+5=tC!+_=CbrH`5b;s1DmypI~lehAQIcE5&#
zkGAn*#<r6@*gh&8;8I(xAIEvEZ<k{sr~8cdQLG8xeZpmBd;~+_J{u|=v&s|aJ^?j}
z!xq<wcNfupf*1!O6yImmwcbXF%dA6?)W-}DZxX$a2_8LP+EaovfZ$`sZjs|X!;c9-
z#KU5|L?07Gtkz-$&@a}Xrp0eX9~T;D&L!{T2538nK>8mSj93q!p+GBD;f(#b(Yp%Z
zwF+Df*advt$Q_5No@j#sxL`gZBuqjG-Y2ZeBur`F<v1K5_=LdKWA)G#btCdN`h*Rh
z0i2g(HT#56JYyau+Gs@2pYAvE3v3#D_Y3)3I3UcwUuYNrSAl*;Xd60RKj-1~j_wzj
z*Ih^P{dSCk2r7*!O71tFTkplaeam33dY^PX4k54)?~{V5jQg#DCb5j*-j1|^LPw8B
z`kmui<4+2WdGRlNQpm?igx;L^ln~U{L*A!MYGY{ae@YPKF((37fH+B90a943CL*nL
z9Rj#xO>Tr9N<L)*8f8P4n%ck5!o={JJI~g=1ys}j%YPjEa-g<=Uc)L9G!1xF?oKs@
z{EZtZ(o}%cIYo%G?_o-sB9XRN{sGApB&fuDzyMWXT9N;NNm$9Upc{@s<`w+`VWPrU
z%%TTOP{UYH;|GL-I`o5sh;21?A`glLpMMg2xCadvc3!uTuO+P0f(M19TA6Rqg%1i6
zaUw_Uf(da&XoIQy;2vtX#dd-o63na6bq@&?_@&X=zGlD%8*C35TVh6Ilc<g6&86re
zW21<>!|_8xEyvX*T4nAGT*I;|ytsM~3l;#zd5{0Fu>{4Nr@)o9-jDlQ)~pLNA$nLK
zk7GX)yIOsiy84}OKF}i$7;ja)M^xBiB)TFw4+tI+(AWU8nIAr4CI!1%d9X!~2#q-8
zMGwS}81wBTgX9q*W$Ipx%0~_50L~rAdsIrcqO1K!MXVhgDR@+najZH-89r+4ufYyc
ziy(_KOp4e6!^(|5dQ|8~A&v+iGd!C(aM^oIfDd8@KPJF&a>?^Kcua)^pGEkXF~#p0
zu+0}eW|()tAE`CBA;>X#Oz;fKZKB5wkDMrbk4uyM02|Z3D}&#;NWtSK7C4hQsxW-q
z6d(JgVM%IZ(2v6SabxCmN*=e4mH3TMOG$YGL)ZJXfnc)$gHX#9pQbU8anST>;lHnw
zqEELuhDvvh2XRZMNQpMsCT_zgq*@>5o%e)c=b7)j9L!4!o-p>sb_!jg_;VG81ed#d
zqj?+@Ct8nf6!b7XDa2K8An!>-#pb-P-*HbWcv7gzL-^s7!o92kk)|cuCf3URuo<7U
zZbOEi64+Z{%snLyVY5R_PdPliW;`VgH0&&NMK6Z$rAV94fSb*IhbNK}ZGx`q#58(Z
zsH<>)JZ<tA<+B3+X@SgEH?|N@??DQ;hpQ=@%1B{wWMAw`V9&vFd|K$n#aNv5JY(Ru
zaY&lyvO|?hk^E=0<^#N<KO=m1_2bH-wZTy}{1MLxu>+8A>`HKq1VnqXe5H||wc%RR
z5Jm4<X<3F@?mugD1xC79BEhqU1@RZ)?pYy4e0I^ZB82h?-xH6Y6}s|p96xJAcT}&s
z$5IV}xY+c0u@z$GU<~@P$?kIGf>?m`^--v)o90mDYLo^ML>a3n*x!@HLU)OSGoCXl
zYh_%&^q$jNPhl(SKPNEP^yAqp!PWMkhgt`_Ls%*=YtDwsa{{=4lZ@m!6BeTLWd1yF
z1mgh;ycK15tL$^~pBKElU&|fkDt2)Yjmz=Oi=G#n&N&Bf{TP5zj#N&bH*_*+u_jF!
zx~+&F_of8CEN=dkz_?}}o`%r=<16}8)_>Uah?6%`0DM{s#PO7g9||6(NaDcPuujlU
zE9Er^owZZY;q(x!?UY_bXfF;#f}KVt4wE?YD0D3c^Kz82)7ZkjD6HwOem^uMG-9&T
z*s5}9<St?5CM@ilDqT<hE~ABmmEmsIlB;J?xXWN;qYtxxmw_6GOETVNRv>nB8Yi9X
z5-xBiin0G0O+Sf&;b|JH1+10N2nF-zP@rG2FUQL986Dsij=zmG!}};KP@gfR6L>Ko
z(W3eU-&~;2S`kN6c%K!75**}z)(A()1(qn6g$WAxiEB5BdtuRMts@KV<KOtRLJTxh
z4#V=ZIzsu>`UUyUkuKP6ykN*L=E$BGq*jGpr7xH`OE<6z4_{D0JPkoaFBn~ee5f0{
zER^_Tl)NCK;WuqQCorJgpOM@A-scP%I+^^>8I&=;5D<LMhDdn~Zq!j+FE;vH@22Q;
z#vZ#`*wq;;aIhzfC?%g00x_$ghd*zi;Wxst{JeGIUF3gWSggY|4L)xOR$&QtO@=vg
zD*C*!UmoMD*s+4B3}Nyo3Syj})B1uy8>nE#(j?+|$p3=z$$Ns}3kC}3=iF~!5c#aZ
zNc@5^g~=y2ajY#dPI&|+Ul6)+>WTP)FDefYu@V18!8&yx)|D?Bsx@%I1};~Drz8BL
zAj0YsebHoFz*YwVAMqEBCY%kj!}E(q3159;i<kb}L^!}|-1~2l73L<y>;JdX!tt1a
ztA`CgI_<v=Fq~6}6^yhUdI{$bu`7gG1+ys8LflqoFY%WI=+sBqQ}rc7E{314MZ522
zd%B&z4RQ!I%~5a^eaSkr1NO|9Oe(xlPQE1K!VD$)vQZ8xepy=JxPd)Kno)mNptV5s
zTwEhJxcyN3UpBgG6L^p#c4fHAkbK!>jN>T`Qe3eVTyRu?p_UQEgUQztj;W!bC7m<`
zTUzT#D97UzwS?&LJeIJR6<`r-nS`;oh6P2tjiyb!Px5vfrXssucMA>J1Y<_K67Iur
zw~@<!8{B4zb{nJEAjK}z4KvDR)?-e8MIeGl42rJ^jb*;s^A(|jZ|CwD5440Wyznaq
zcg-4#v^8AJP#a$nN_o}6A!?!}c!dNkebsp2%bMO-1q)vn@xN-y8i$6uWedJ);=n~&
z4s`jdq3f(rq=4jX1UlFiVeTbg6}fNNh?neyqusB=2=%^Zn6Z^3|7(JI6t64=U)wW5
z5M9G7m=<p1v8T!EFitMNW@5#@f%`i7n$eTP@zK`>X~$afzOEg^k)^(V-MpECuNyPv
zQ8o`ktySVP?nu)(G(bSd*R9X7Di2cfb)l0(YUmq+4pRaFa^5!t3Dyp1+iz%JL5hKX
zE{zpoL~DHv^$x5dTBombJ;hoFZpDWw(K>MU@~Hl%NJ!4*uzY+|p-$Av*W?HSo`Y`+
zo9wI5)YdXZ-!uX87Dv20-`vB*C`Osf!q0%2`YnO64x;*&;9zsc|CVs4cc+7I33nB_
z(V_+A3Em1t-!ds;I6`u<R)jAVcEq=Yd>*F=2%&EqQRDChh>!f+f-KJ_`?qZ@@K!JQ
zw)NuxRQR{8mw6kEE{wh{w5;KPhgd<>u_;e9xedNc`i=m`g)8#DWAs5o;uSjoI|3Tv
z-n__by*f;R@H--cK5+D%m($7mIMz1e9HHbpCP{t4<hxdzuho_iV&Q#PAg#pBlkXZE
ztq)M}U7=|h<0kyBFaVnX*z_}uA&NCwf~A;f4eUf<6@AY{GdYZ5_C0Ak&SrxDJ(CPP
z=<MwZTvfis8h+16!)hV>?#Pu-Ahag_o={Z9yL5>r%k1y}zI-hsIMvgnguZ^?h^S!8
z8hl@ffcC@_C$40Sm!R(p^Rup@SW^?&a<n1xam){-VS$4*ydQ`lF`^_6*#Cib6mEMU
z947dIK!Q0M{y@aYp;y>QX)Crw_~L!6S)7W>k!<n<BQwY6BR{mRnnYL#c|SA^dOgbj
zp>@y9nH0Fp^H1V+4XZg&=ezXL58E=q!Q~H4D?%I8BsMb9QbbbYAr7h7frDc&5&Gw1
z7$iaz0x>vzGDuVeY%7JD1~WLe)LKJ0EP_Q4Cq@B|v!Q7dh1Q8}z>oIyYDL0ty&nnu
z4eIvue`HL<V~kDPkAz9~enOI(5lSffkr2281K~%8jL+~fZ?&X=SKnTg*5z%yiS%AH
z$mKDJ(|1`IRM10O6}%wAFnm#%#1lWzrEM6}%duuExTXiDv=$D-(N{lK5s%|+=Ev4A
zQy(S&$JV1VVSg-eG58Szrq$GBj9i=6-4tsZF@--(i9)VH8PiXsd6l2F@P1<4;C2+u
zte+U3i3$S#eqtna6e)C>RXCRUiIbH?V8pH+-ua-!0p$ZIJdYwwv6sNp<kUy8I{eg7
zp;su-f&q-z@Tbxv<CsQ275vzvu+LR9x+PD3YGjX=_zuI*gd=R99wATDWwDBXW~xCv
zp)losCJ-mf97h=bOc0%a7)9+ie0M4SnX$$f_s~5qkDw!X^XTWoO=%2<k0u9T<@rAs
zSUnhO`f(k&4fSJX2Sq<OW+!<c@pGeokYg<pmz4tr`h^Yc%f|GS*ZYOFF2FAEe<3`e
z_d#v&3&DHSY`hgCsjcXob_H2H<6nqq@-m{e0E3gSmj6-;u%*D7`b#4STSXWlzZ4e7
zag#sL<Oqg#=qll)1S(tG!`BEi=a(jD+@j-v<U|Y06WH$3uLKe3g*)$8f~dShW10M4
znXK`e7vdoTtvbMQ(&4X+X~fA>^eg4_T8b3}nCRmBD#@=z8VGR#TKcu1Vr!VZUkekn
z@v5f(YiSI-Hu$vx<%^r4exI|MqF);W&}RsRrdW&c>}8@bxhL>$j>!qtFq(g3tq^6v
z2AcmH>pBkE<Gu}kBNQQW5$4-(bmZ{vNI%0WhVXtPG}ZK4c=8+J2kiA=NAz21k;k#R
zr^#6e?eKrwW|>brgWnoSH4aq`UzTMH?ze&-!widrHiHcYTPJb#I{5Y={Z7Xee6e8b
zcY<~5A%vUgcTm$;r$E2+HMmfdd`KDn&P2KqPKx+<+5_y5OnztL;V@EcUg-A%7<!Yu
z-%C9dAj1Cl1_4^0g5MiMa_}DhUiz*$NRhVAu^sPPbYP?;E?40@oYN-24oJ;q3Q%s-
z0tycR1=BX}5G(>|gwsZ8pSl~PX(f*z?TlSzM3?avd0NPo2hT9%{$S*`9>6Vo?+-Qt
zV2p@Q!T*C`?LeU0AB?hf=%_!4Eb18bkpltG6BOMa1b*uwY($%sxWd-^;IX4WI&^#|
z!~3Jq35O3X)jvAam<;TY545z%R}I2H3chjPcKy-h-NTMVE#Qm2i7Vi1mGmbe9JZ(T
zC+)Xlk^DbdTfSHlXh8*bXQ;K;uBE8`k<Uo}WYn%<3qSdji5<}(xPeK377SYuL-l9D
zFgA5R`T7wnVDM+dH3)J-ms=03E&8+d7(BadfyW8~vD}ns5-Nu1FM`oMKIQ$z2$EO>
z|1Tmw=tfv1nijWr=&H#g8STmPz9w$B5`V&9goWH1JSt_(;dBVwd+)D8Jz^bsTjl>%
zFv$ZA!CzIJ_#A37kC>6@uQocm;D*=gcwGzZwCjxErG~!=wle0=-vr+7+pyyKe-n6o
z3KeLDOV{G<Dia=Jt0>a%8`n~-l?$*^lfN0p8Z3{$TYbJ8;kn!x9OAyqO|WD2?}8de
zAN0WA1q4hf1ocLL7docy!-nMVCg)*(r!o1v&<&Rr{X<d(XqkUl4<d?4Ts;4<HWH&9
z{6o;-4G*CChd|x|4x@h<(X7&3W)Rl9YY)F7Y=VDEn>C2R^8RVObz&g-|CEm62M**E
zE%>M5MT91f{QfC;5k&~GDU1R%WbA6K<|FAuOYj&Pp7Jt^YCFi&46cqM(9-`OA)!)5
z?1?7puxi5p5fR{Hq$w;7@&D{ROi#S!DlaS#EH8<+$?t**^^!p1-LUtPG(Y1E@?WwM
zS3oFLk%E_uk<ty={)I2?JX~R-3taec^pc4k>oF^a_$4DB5j)9C!hEkh&HOI|2mYZ-
z|E1l|J1qZSX8!XfI+;qrzl5yr9z0AYDUMG2m*HaXF;rBnFmn9pWunyvaQpAyu6vlj
ze;Znk*C7Aj27}EpXs^J1z#7c&DTJ<hA4frHb8J}QvBg|<(5fSRbmr;~^D~|QY4;Ci
z>wk;}oN<u<A17IPBP{rju(S@${l}Cn$Hg)q(SMA<YDpXeu>&<wf|JQL$dBmW4Mz!k
zqhAm(l%ZV*Xb<7J1<*jd_CdCY4Skk&?Te%==|1h+59#~}Z&7o5nqVVNfx^0G4pnJa
z`@N8Le3avIeA3aNUF~=Z9axde^FFIeO_#5YvgsA?X^o)lmgd-$M>u*yyV^IgakQG}
zbH-`fb&ztw@5VtD(_TWngbil)65@IJ2ZMVVjqHreYPB-2HNK|L#vkQetwA<G_Y$Fu
zz!g(<1-KkZq2}Zm_tVtS98|BwB=5EBP$daZ9<+DX%{Dr5&+N6U{Tvb>63AO!t-Xwk
z{D``39JY(`HrurF!4YP-*0Q%SHT97x-0sL|2Br(|=Jz%z(2o4zWp;1t;$ilR_<M_J
zWapUEhU|=lVY;_<k^0f9?TQ>49PVvmW8{s!MIcpdMWelCBw<)IU6oN>KZ+GHthU{6
zq?STr9ZQL}En_N9yB0(Iy#78)7dw_R?Q{XAiMNk%3P_XaFfG9Q1aHQhd`9mWhw4*K
z%kV4$R?>ZpUPK~LbswR(%xYXKVJp9t8k$D@2Bwf=*;eKkPn-Lg#Mm~&BY#R+5z=ez
zBmDNjd{q?uuu<#NG_CD5evtOH_O188e#-1?#9%2WPjd(>;oA_|eFZk+A>d2gSI1dp
zJLO!H8Z6piU!8t(d!wou$fFmEd|#7+u73?zSev9sGgy9B^-$B5Y{lBBSX&Rkerst#
zm%PlFC<!Hmj7=*E952p~U+yP#a^N3jG@YM-sJ#7zW;nAcyPt8bcLVnmqI=|(j@*7G
z0vz#C;A+Yp`09QlQ`~tDH8pDmH8iyaF4X;Ow&Mi?>FB0w4k03NdOwo^?w7%H-g3Cu
z`;#cKyI+r*(+cz^JOQ@9kUaobL`IX_a9hCJUznV_4<jhMzfB|fy*Yfv*Fp*R$!QyY
z<t5P1@Y0P?b${a$a~>}*3XT%u^Beoy7_Jm4+TY{^$R%oOwNW^7Vudxx&uq2!-^I%@
zv}IwdvlEgwGKhhu9n97{@SdY#9>e6z93Xt)*#!6^4lq8}@RL4cls!PCjVp%!0Y*rz
zJ~Tu*m&Mx`=p+R)&PM{(1C0Ke*ySH!^2YjudDl2VhX4XHA|-?8#O~Kp(-D<Jwb<o~
zu;*Gz#mZTfDD3(KyZfgV)(EUDI?&Lt&#*+91C7P<7`E023X3>y7^bW%fn9|6DXkrc
zdDbX*peb4oxRdQuaA3P<@>F$g`+y^KZEGbCqHbu#sr!Z~I?#kQ#v!In#b4s!jQBv4
z&iWFyv;bk5Sn3ZH+4Hk$>TH<S4?Q>76(2gtnCb?g%s~bS{p%eh9E!n_JxB=Vt(mWN
z2FCf}>zp=kzlR%W3%uVjOjWH`mbEB6NV<HqObx9fw{IhdyNIBMrsm4(zKs>()V(9r
za@F{4+~gn|P8=gPgdWpkI2w4>ql1Oesrx4=bFeUjXQ0uU-ofqeLWg7zHV|!pgMaX@
zww0)7;3})!!Pbu)N`z1=g$KQdBcrOqlXKZn(|CyF5o#PPoOX0kq+bhs$l5&ER0#HF
zaw{or)5EV<v^3q%57#A1e;Jx>T9b45XL=41t|nk~XAZFr8^O7+cZl$aWf#j?R%=b&
z2VHT9unT`viE@Y7z>!0QK#S!1UsYSwu@MS2Ssumesz7<^yp<xwho&5p)YMufSlV$L
z4$r_=8xC9-$sxun#9rdy$LT{<B-mdaDpd7ym_g=HqpHp?(|CtUr5zQ@9x4pVlOQEL
zmul7E8yltEp*BFLo}%DTk;(|RT~$|2A7*x0xoy-?kZ=cxVWc@c8_w%W(?R4TUnQ_y
zfCbmmYI1$xK}s~oiyn;0Y1gRrLA=%~!-*%Hy00>a30-W1dYa@&(d=QuJg&QftHZS0
zMmPv6*UpID7-$A~`U+IlCiO9><M1#O1fFmiqsCz(FT7wHX=+EAnu?4syTqF8fVZM`
zm<<@bh)0R$_|Y8PT$pxcJV)qoR}-s5#^u;w?j3G(2EX*eCRp}xqX##ZanIy%p$D7j
zoNLD$lfdORVjffv7a8EvTM6&eO1jQY*@mX&cdVkrW!lMrdmM&MQz64Vi8a;JPb~*^
z00vrexX2q36*R5MJU@ANgm42_ACsFoLa-q)8?Kxqq&~l|ls&>mE3V9o*VfmHa8_bL
zQV{5jB0_xvEh&xQ#G$GnX3q)mXl@(N1m8St93ef6KEh;<wALlC+nbs!!($#FVWPy7
zZhVT`QouNvgFoU36Duz1fFDP}CQv{bA$$hhxfxAfh-*?_#?}|ybk)GuY{q(WJ?s%*
z>#o6eI+ro<dhI|<_+n~RlLNeQ(j-TgH8Ro}IHgsc7->-rC(BJumf=H=HOUI8rO8zo
zgNY-8_11JoIsiS7fp(<uF~ElQkv5?(#)@;K^#)96=v=7kBc)aA$=z?lOTs!ZU~(aU
z2jz~GAy-7t9BJ$hz=)}8n*qLo7%KP<F<cwkV6;w=f)a;0tAyr})+I9C@ydl#SV6pI
zOOxe69Qg0SgZ<j{ktWLyc|PMPAr%9sMwz38v^h7Dca)K~4tTOh*^*Mk3W#vFqYMPJ
z+aTqRGXD7lnRlVVQP#K6*ZT7k)uTjsrBV1B9q^m5NHmmy93N>#S}Bj86lor15<crJ
zJfkd)M@rkMb(HZ82QR;`kT_`TupCb-G<hMKjuv*#-AtLIjbI#=LA@Pq++qJf*`p02
zoGo(0<!HgiZw%&+7U}TQ9u#OAc3qLG+J<9^)L9W8E#%^zHLNQQ&0qrNaD_yXwIdmu
z4jJ!9;X#kJZjXdvw;Y{(FiVM}a}*Zyw98$L(N4$cz#PXf%Vdr*4p<f82*f+a7(~1h
zWwk0Itl%CwM(8XTi<rT=V?->7J&+Y9I7Y`ew0>1fScQnvJ4PgV^)=Ku#)blxCCraV
z+aO4nn%agvM69l2#fmt|8nv{QD92}@922eD)85}tA7cXuuR+M|YC6_BY>Xe5$sB78
z;<7!DK<`*%5L<A{Y7RFV`PW4JV{PQ%usP=d!PdaQJXS{`6nXVn;ptVI@jkWz8-NSC
z;hI-5<6IGk-ZYOj<~8ywcED@wPHHLe8=y~;W2JWxBLag+)0g3B3r}tf{xUCAuQ04V
z95MF_X#l@;k+SVHzd`7`T5AC+_X-n>9ujCFUJc}`Rc(eOHiTLt2aVgQ@d}eKAFuJd
zgwZR6`B~>v^A#d3jpUCtgKo#0Vl9P%7%SE_iBirl|4qNb#4sf4fQ}QUIIc8voDhiP
zJ$X*RJI-pdiv@0o;{<ITophXVkLx3RV<M*z5C}_wCa3O&Mm%m;23t9O0%Om=cyZjI
z;?)zlK^w;j3w&W=lp+NT7HU{nn~D&zWU*3)?+6%zdek<2q+g=sIPEPw*-z819*&IQ
z5IbHdzO0`z$D0s~*hX@gvUj}eLCm@A@dBh)K@^35yhwzf8p>(14=YP>ybVk|RbSN%
zKd6K5(hNMotUVMEmRPu9k2ghH!zrb_Bi(dRz-`=yBF=(VJ2N%`t2#N}h6+L=pgfd}
zI@~vOf)EP#4K)4<!a`o(F7i$gviaS(tX4w&SQ9U?+w7BJk70O%2n8>T1SgnKbX$W?
zJHa{!FRfuPhYFM5>1b%>wie(=G@}ztB5R>Pv?iaM#G2fW=z`V>#s_<TC{em#G+|=7
z)&-o2pD0`vI2tT-qVTZ>Z}xd7S|4>YSDM2bE$@qKwKAq|?nI-SZ{WZmf1(kFShMPh
zMglY_j^ngh6>6xVNO2$rgZ4xd1>QuJw@8~>nn#E%o}t$ffuq!x!-a!v^mn4w#Dg~N
z$h{c11L!0Zd=2Xo>%o%*N*?<dPm|b4Q}!eyep24h^iPuR>41&yN^q?ZIypE=$0+*+
zs+!Jk2MOUx#$kn{A{%Wqcr`aV$+(hdF2H_M+pWYO#d`?U=MP)f5(=Z8#O2WGlgy{f
z5j}dqNhe#!<S|P!CmR7Ze*4%v**JnV0aGx0vUCSFU|7z4&G2D3EW?wHK0f{D_bLJf
zEHQh7R6SXTh5W|J$s#>Oyi-Fz14f0SlSN#UaK)Z%@~&59-HY7^Ocq{|TkZo61mt?A
z0s{Ckns!XI9>jSvog#3%vH3m4xRan{PYd8_%${O=*2F{SYsm(vjZ<vmi%rG75}aZZ
z<75BoDMo8)a8M#lPZ1d*WE*x%Lo@XuzV#8EB7B$O@Na68ErGcGbz%p#ls>%3D|SGl
zC1boXaRhFXtITw&i5r^$asOvd6|~z&IsV5xRirJi=VVVcrbqL<Dr>zy3CzhkFx+c$
z-hHY`f}?`4<x)u4jzEQ+YK-?_FVHwu#JawlA_d7N9XNC7*6^Fov6f+%4`+KzffYFF
zCeZ?{1Goh{?I5q^)4<bg(x1MPGN&1*h+vWz_Po<XjN&=dn)<Ajf0}q40TIg{+yAUv
za;FK+@Xc{lU2vMn1utV`-koN0*(5(a=8*Pa-=g*8ydhEu0Lw2UH8m#@VNc=b97F~i
z;KER%@^DQg+8Az$B2AxWqoZqBA}w+CrG3(q*Z91QX99=U0NVo3GqP~LKS){6WZs9<
ze%}+;XJH4I^Gq%Uj;Uf@=|IB_;VUXtPvnS`5~#@VWqA0-rG|nkW0w|rA{88PqWknS
zd@T6M38M@zE2607C}Za$B`!Ns9>fS$LPanS;wLJMO+JBy*EVyy;D8$&5f`Tm7qG-B
ztDkEXj$t|7IG3=joYrDrN^rU`qPn-L)nH}FlR}|Zy9w{fHyqLu%x}!i=yVg_AYO`S
zo^BH3vm@SZ$65|=Ex&2gQb;(ArKC+Eo;#V=<RpGskY*S-zQd83p~Db(ycs4b?7`ti
z&T3Hy27o_9n9}GSSo-Zo@O=gjS`FU1>I{(&3{}iC2d+cj)NQz$MFd4eE{E&()YP14
zedrQbwjOTuR+|i%W{Hxaa|7p%4g+E#`L|HdPyyF?9A(Zhl?S6v-v#o{5bET9cJ>V6
zo1ee(H8~Cu<g^CPU~zbLhIKyl8@s5g4kGrcstQ7FS{;T@SCK<<d>*IY`65hHLBe(h
zV_FCFKmQoDw4h$ZmMb|!W?)ArO>4_MoQO12*nznWYBDo*<h_PRo~Gro$?QyRiM`lN
zqqm^97jrYMqp`K6Kr8n2QB_kL`YF^JBO0X8m}&iwwTTboTpmv+V2{|GX(PLfi{m|6
zewD7J)Pgsh?<KAbMbICLho4n}zm3imzEJbL3n_D^(7zQYjAt5pj-i8f@A9=0u8H8M
zRix&2IM>b;Szxq)tuu{D*wIuy(@4zob1R_(q0fglwBZ0dy&~6;uUIs-1S%B9)tOF^
zIA_CtqUFHUsuFyxP1QX^(@NJI{BZPHhP_n4UAN3xLK@a2_5pfl3HcZ)*|UsM9C#0v
z@cS>8*J#w-Stg%V2!uRK#wqL%sygueG#h>=#<1Z=0~RuE#lLt5L;oz1(02T$0yQ0Y
zUCd*Jh;t}vX_9}$CviyG!aB<=takJwx`Sp3Rcx9-Ys|9A1tI8ldvBJYEDymv&1zkK
z>k{KqGkEk4rdLjD!NK%u3N$Sju&T3622)SLd4%7z5o~MZCQ4(L_B>9u9A1v_mHS!E
zStf=txFqlsIF^M~3QycD=>zQgdHa>j5@F(g3r#Bm3_`rzL9>k=_-w{2;v<+X4N=aI
zK6tYYD&ER^gtD`RpbdzBnXOV0yE3P3_!<EAIl*iZNP(46Rnxc(1HCv~6gXCQHZb{~
zYeP%%v<)~^fWV7jxTYhb$a;RZs%pH!)N-|^K1qpY;48z{V_GX!SbxlU`8dW(nwcYH
zujwLhj(}oEVRnwOrmqFe5vma-{c6g&A}9pxrCpYv;-cytlh8O$<ij~42n{1?xSE@=
zgi)k5*;@c^n~FgU;Mmew+FPy>eh3jN#AV0WpqO^qZ56EQbhhiU@gcoAn>kx3faLH-
zzjwBAB|ndmJ==yC|H=r)<k=>giazp})5f3{#{<FHA_KVZ&~s;tr1JbLs167Rw7|9F
zaNreL#HhP0e3W%+o^9(i6aZ!zhUVEOu+ju}q6Qd`*Sdl(ur{q|%80s<-9MdUtQ6Rt
zo;gR@)a#tyImQ?NtOdLV=a^XJw2uAizBcF31`a>WxfZzcf<YB1IDTavaW&@%vr`|$
zwp71Enyn8};~b-Wkezfc$8PNAIXdr)C0qt^CAhaqEiIA9u@VJ$say@4Ry2I$gXS9K
z4h-JRTw`@>2|Ijmu9?E^_;+4fvvY-iu@=eKjK*K(=31v#^$nb0uCT+uMpM;_*x5kB
z;anjZ;VXAhL+ey<3ovrn(PIT_x-4D}gdM365TJmW)N*8RV>@fE%&UuL(6m<43v_g@
zpy!vrGv^AM5~YG*fOB6qG{m!&)pAI?2BqbnEAZxBM>+SMe}E`ZfY<<0RjYJ#<2S8L
zg802PYPeh%IyXAk<cTQH5o(^RUB%npSYdVa_fX5V!ND&jS}^q}cA)2qsO#7fV#duA
zqH5PuW}bk-m8yB<%`@yfAfv37$fWnRghQ8cz?hpSvM3-NGni*Y6?rCCH4Bl;&MkDM
zqx`l?!)5uHh$00ubuZM2Cd=sD*b%`;^DQN!pPehoJSS;v-S~-^X=P)5AAZ2gNSDRh
z<?=<uHJm4$>rYE%&odzWYb0`Fah^>(4&#LfOB-_p+V0m<aGvm5#|!#Zg@b;bdOdy?
zOl1I5s&Sr(5EqeQX1O}E&!(mpAnp|gmcxRhPaICQv|_KmlA;v{_^#fx7O<YdwlgcB
zar+fUW7b$)$3a3_<AYzzU_P>0;{%pGX12elK17vsrB!+PD9D;}>43`s!4xowvoa_~
za9oxZiRsRvp)?@q0H%m*wO;%I7<)~JrVlXUtc`~&+27vM>_(inNe?P0wnJ!IaX^|^
z!Ie5+cx!JGGnzwDegwAGe35h?Ja*anLKhd{!i;{$fTrAh!I~e?2SI}QA~hVUJw?^|
z#wT7XQ9*^;u!L2$p~>0!k;tgs=mCxiPy+C5HoLcDM@|)IN-dWg<V(42%<y*%KtE4w
z76kIH0zb|s5}k1kWfquR;JQWcEfAhY5QmvvV2Z9oT{6BlS_99cD}wIl*MEWq*2T36
zeu1;<pmy-DPlnoZ0B-LE#x>{|0BP98EU@9T0mf6)wdDKH_<=Yf^s+V7Qp)NuCK6X?
z6L;RU*5MfJg;M(*9G)!{p!huu-rRc&MebYJ`ITKL9O4J1@FR2bS<a7h`kB4)fqpLI
z4Opt`=lsYZeuU2CQPAfB8w+iuvsx*S;p9bY-+=92Q<MDaVXW}rVZr`IDdaGF*BTF#
ziX&}WX+|G);4G9D32{>&gw>r{B;2rj5Iz%6bEwDR-PuLffwJN77uiT0!WKqhAg&P`
z%|#-cp)yrndzgm!r95kJSsyhtqc8SH+FG8QYPvS;iHuziQ76<|WDBU?vzH&qN*vhL
z{WQHuWM1LO&&9$P77A=#77J=z_QG`a7Mpn6;YHcSHZFJ*;V(8}JNLk18%pgbM8Tsr
zVZ#_Pqg6#E+k;S1L1vu@9B`0_5pEG_wLy+~X=(}d0KZukFSZSeUFd+hu~=APkA1RO
zB;VIQo|<+H;Aisq4J2A(J+qd>>oZHFWf`$z-V*H#IQN%`IB--8D}RYFj9?kwzvi?=
zwi|(_=e-!dhpL*E=i@_%0hfVZO$`MACi$nNq9xX`YhfoYu?fl#qvGeGUDZ)m9WK|w
z{ts6dN0ac{Pb;<8;fJ-C3bkUwXO<fHF6^q88mr>+&$=vr(ij&9mI~)XFm!VY0q4jt
z3KasbHe=zhY6jOBIY4Tub+)*eF*+Iw0uzrSt%*Yuc;=c*MUvOx2ND%BFHh83YD1?V
z<1^6`sXx8c<iLsxeY8wClGi3Pnl25D*U4KZBzLn{ahWiRv5jr6zf2&@rF1xQGzWp>
zr*<e<CbV>|rs^`IALmKfyW+5HnT>8SRUV^;BC8K#AVrD+ceAQcfLeex;?wxiN23-$
z#fKr(YGc?yiNoB16?R$yfR1(8kS!N_5rTqXt;}-a4`Bfa`OtKQeHPi}(qR_F;x89O
zawy9yOHT9dy`W&Zu(lSOeYptcrt9#dRSFNgEtne1ZK_~{h%K>rb`%HCIlc(q)Rq-I
zsTD65DOPKuV_S+0^$;t8-cOdB^sq+AWwL2SzXs;i3X>z>AI8d)Sz*I)9X8M_gv<*M
zqwEUPN&GNq`+B#(!iELkqseJY*cfc|t`IrNi@wzrA`EyB?vNim3ms0rMuuHzV}%G$
zVni@DBW(*$0HP0-0z5|)J7j#LsHNYBFdA2g067eES`mTqRr1qVbiVO2F-e*8g_jPT
znw>8#W?XmyWnB$C0PLS{vKUo=Nlr^A<h8uud|_4{-c&tbNJd|2kf>%6Hiz9B?0KyV
zLI$DTBL!2fvS6Buyu1bG#`z*6gtFgEEr%O%80<by6k-*9&z_ZOrG5i?>;h8_*s!9%
zT$X?3g1ifaw>++BW?jiJAAI>2D7L4teac;6^8mkmjU8NYfykzc9hYl_`3(QS1tLMb
z$_ZW5PzVKPBf3DOf!J<g%E7}315vS-;*b}Ciqz6}YkTmZT;kvt%SG7r+Rz0E(Ms)_
zxicuE$=P$@7Fa1%=ug^YSK6Q(fu{9Wnlzx~AEw+&?c_cRT(ufI;Z|DL%2Evdp~XY$
zI%p^YwY8#^whrp$5-bEw2do>xA+hG*K;v22S}Agch@t6~<{jK&n!F4bD(ONSh7fS(
zLZvWA-i6W-u1PFg7aD!oGqdZ-zfk)TKChf6OV~vR3JFj4@)pdsDUC@qaj2l#>~KZx
z<Aq3Z!Td$Ahr+v-AAr^5ARmuuvJ-dw62*dB%go1xrabui0{^TlT_hdA+a~-9T;?Jn
zbQC{`=V?Lv=hU(n38S-bB>y5EM59BP47qkjejF^gNJLN{g9fc?Ys{HGICDZR;qzT;
zXc`x4@s8L<(%(>tLlvB3G_?pfn{LO?%?hwGKmFZuz;K95i8cc04n7K<R+4Xm0PqAH
zU)kbokr_?$Gh?16tLW>jue8f`LSHk0bK_x@bD7rLDe#3^JbPSqpF8>}^xf#>SLz#D
z#`7CHEZ+o+i&$`TBC^OA!RB}H*#X7A%~3><@SO&i$4^$mv`Uo7si!cD92$g_;+JYW
z3@iKYDAQqM3%_8>Z<2T&!ZchWl<hDiop|sM$6y`8M^889TnT<S7CVrj!v^v0+o<X=
zT+B}>haEO0cfkH@D2zIy-ZWK(9jGYT*&l0V{6Gy~k8XA7>=}W}IdNb%@wQQ$z+T7C
z`bs7FNxe*`P*}%RO-)@rgR-55A5W9vmw$B}a0o|E(|9Zqp50E90|M!I@KifZPI93!
z)VlH(X+vRO2H!=eNl2a*q^4`Mj(_qZ?i5*WN3zu^5*n!R*-4_n#EhKQv>ZuOmx066
zj@Yedx`YPYtdLW8ugf^$_b0Ml!s~T&$#>toVc+C5*LH#iU1lnXD?u)WRuwEp>cu?b
z!!Dt69X8VqP4Z9nMqM@n`8a^X2bx+q48tr|9DK(Jj-Zz2;5Omk7)l%yd@~j6uI43z
zoUI+|mL8Coq%++n<kq{;30}8Q!Vi#UwN@AV|NU;^^rrJEr(f6LK|cz*t;hH#KC5wU
zy|x0ku7fM55{+&X9O4%!QXrTo@HlmwcqGzB9?*^*U_8dePjI%{WZ}duQBu}cXu7*S
z_gOzuk9AyLe#<1&WAkC>8_DZ2CMWpzb=H+(y!)ELq(|&^u1DA(XX{pzTUe93=D1Xb
z-<avKe$ntybXt#y2M@nTjtqXp1Kc@OJy@6&WCfd*R*!T6VxKV<_4Bnio=el3tivfs
zy^dxMLCEw<gVw|Rq*<@=%C0jg!CvDPCvd)2#Qmry<$8rK?9_u^VT>;rR2^Wr2Qc<}
zt!LJ;4WX^`LwuO*Di+wI)^rv75Ih<8nh^MTL0M=W75IVBMB$V;_J6w9dRX=p^7C}G
z%7zaA3>gAvGOJYb?4Vv{?DVgu>?$K~JJvRTl?f%!zoU?I*+Q8gG;no0@L)hyA(r^I
z+A0;^gY25tN^3Q!F<RwBiT;8;>mcU%+$YwCfB{3ok=2RUrd&PzKp3`+t3=RSF%DM?
zD(DVe3|MVUBBYA#IB&Hv!?!20nuDHY)5u?Klt3VSq@aNCYb_Xyt3@~?<tkMjh)bY^
z!`0RcUFZeg3N}`Ygka_BR!dvv2M4LCNjzeO9}PD#$}d&54Ezu!j^s)3g~e)Gt4dS^
ztBy57NdZf5W{q_d+;5(KFTrP-U1Ln<@e_i+%Z(vICbveUvzq+_fol(=r@F={f{Bg|
zv1>Jm6{Fz_VX@*JMFoQ69>f8x5y6k*&X9hUH+WlXOe}c2PA-!r4%v#8G_AyL+^`Nm
ze=hu7i3fW!7Yjdn80}qby@6n6;K^QW^oe)bzgYOd^%Q7(S0vZ^gLV<W5m~)hM1i2l
z$Kl}!H3y|%Wks(T<Wc~?&>U$yiNk6-kkHZ48%h~ox`ku6rS<ychZ8l8z4H?^?b@{N
z!PKWqtQX37IX82Okze4~61+=<d_;7hyDky7@pc>3mGA0J@|4b9V!KLyIjMaGDR7W$
zScj^Yn4t7tN~qw#B!1wnafzAgh$z(S$C09zU!`kaV%@|)J5zy)5np06W(?Oy;YVm)
zBK^C*gA#36*a|f;?MT2HOSD${aBZbP8BOv}A9-s{08^g?mur>dVN4-^ts%kBKJ)Ld
z<lKi`+r6u`@uMi{LxqD^v~e<y->SDkf58RRSZh*6m>u@TE{mNI%tZwvu??{!W;JGQ
zs|}vTpJ;lrjJ^G|0@o{{v`)tkejXfSKC@2f$Ke{@z|r(tK6}ru6W+1mAb*{)#g9Zm
zE;)tJ_Ra<CtfTa~=Bh%G=L$ou$s=<L=aQQ1(|XZ5V-lOi$Emr_1hDx6iq}c+%$`jx
z_Zh#<n6!z)O{r;x#Xn`s7p>O|webE?X1z4+$39CxKL6xg%B~k&=zFoY{PiLNtQs;H
zH3uTs!CN1!7draI+h1KT(%Ha6I9zXo3P0W~cMTfa5W3@8o@eWwnDCeZ0+5;tsXWA<
zfw)Z}+|Dhf5ZAC0t$`ozf?YqYxbUb7Tu&QJHvH@(Wn5M+2zf3mFCb)H7TN|=ZG(-^
zL9Dg84MsJ_i@azPY_R^qO+LOvwZR0BHypwZBE5AvYB(@(5K**21iYPBGOe-#yGYlv
z^$}`0M7(ITHJE5It`I=aO*?Q!sB6j@RWMy4T%0B6jOy+@4)k(@E?-2zgBEO*%T?rS
zu6>O*=Nb*b1rp?h_8VZ}=S+ME8pL6jLhFDe8`=W53h}TCMLA&|8Vbg2(*fJaE1Ao!
zhN5gK1RQEo;$RO>J&N5*&U$@V?pqLDx(7ENkI!5xoWQw5-lc+#2Nu@oOO0P1pT6es
zAV0suk<&JOdoIxA)O|3oF13DBzf|ZD;YLuM8kg!sz8g9^az%Iz5WZb4;xI9aFBKu-
z#haF+KrX&rYCWRcc+9X%jWagHc!}O<Rn)DIm3pJ`x>lZ5+h{Ft*;{VSZ?vJc1sCTx
z3g$I9f8A(y8<c<hTrJora^ihl)j=Jdx)&ZBEveLSn5Q5*OBA^}xR;68?Tx}b?lQq&
zxX~mF_ZPJs(NJCzE1W10Jd-%>fFOVk!3}-FjfP0!n4wST-Hwja0w@Svyg}FXnH=;A
zIpS0L1U=rF#LlzNx(RV{IDicLgt4jnaR%0Bl#Rjo4;B8JL25Yoy?k~O^_dtEBE&B!
zHWj=CddCU_EA!L4Z=sfhD93qyChm*DT%QTqp8CP{(td*s2LVUzWcsZG`Cx{;e&cMo
zJSr<#)>V)n()Rm>QM?gdrd+>q!vR=9zu-oIEB0trS5(7qeue!a$UGc64%#3eVMYBW
zD#UQ}o<?Edxkf&)>=(K5n_(?Y^4gau0fSTbVd^<(_(dk%m$*!5gM%IywJ$T;YEWL@
zWde$q^z3COB9Q63T6K1}X?wg$f%nl|wGCUTss&R|;g}@UbnSYmtcGh2c~Eqj$rcf4
zJky$rb?QS<{+i?qkFCp$XY2!@S(3}Fn@a9|gJ}n<$a@-$%*&0P%MtsRak)O&NuJBG
z)Uub05OBPMQK@ODVszH!!gocywt;5KMIKyL*B+Zr3NN=2fM5#oM>j4vzIc_xej;*J
z^p~BnA}Auvk5^HwuyOvtF4D`5S-6|=cx<A$@CFLHn5JDR9z}qaLYv%R%=5<0WozY~
z?<8-N&>;amvMtPRl3v{g)79T(Lew)&-XrBUSx;YZE(My=@Q<oC<X?CXH<@6oSUDS7
z3m2_07$aB28@;B>;X!GN6#^^8mMiGSek#!de(85w>4i{mAB3OArYnq~-LIp}6*`dy
zpa{Gxgpmz>{gl1JM1VWZa0a_3Q%_Lt3hNrV_=XiLP#~Zd-Pi`WjUU!>P!Pa~mBB&j
z>!Iih;h2AKi#?B5$kgn_J#}rf4udF<O=3&INgPt5?Kx@=Y&l@HZA4dE|Kh1nRwq{q
zCy3l9?@F`K+s|b~Z(b>=5uL=b+y0ftDwHqQgey%tcyJ@QQpoDXUfb2m;ms-vT`dGM
zG!z~r+=;`kNO5BTP}AY&cmlcV7HjDS4i8osdHJ1;L^0vA=d{+SAPk7EGSY_f9Kv^%
zP_qN!4_Dc6*@U;OvsVepcpwEqlm1o0`T)eA(^@$4;->%ttp$q^I;MJ+2@Fq&<Gdtv
z0JpG@L}B288;V>(AKnCPUL~UiP9i+q<A__wj)ayHw;j&8t8Aje-m0JrruTf{JNs}U
zTy5lU!?4R-ZS{-#p0TIJa<2LyWnE?b9vAsn3wveg*4)*`Lw<yx0t#FnlNxBNu3-^Z
z1wxnGgv$(#tF4#u^NEqd#622nYO=z&|6G-`&!d*h>?l!kwRH<lxnM6}E!~2kblNP$
zu{zCc7D<5;JdwOvD&U6*vYQ1h|G+ua;${J3KYfzhY+}GT<g2;CX5+6zPS>k0&o@NS
zXNqms+*#DvYy#yQo{{2Pc_TFyAJ#?mYV0blgLTq!xg2syyCRl9=tCte50%gNMCckL
zj5iwql(|OO9KjFRcv=Q>+OsL^D)YOISbh9!j5Pkik(@Ri!`;{58e^|8!Vh6quQ93O
zGBs>v*N#I38rN8V;i0GK8k^GkYA)*!hZZ}LClM+VYXb!KV6ki|))(Fhx0b_-O&13K
zv=&vcO`>a!0DWV{W#O$KAkSrCOkun2vUsc>Q3TgY7hiKV<*qepv9$^(cW|wYgvvNO
z{51_1FkV;CCJ5?jT&tsEj2{Jxv`V1}d!wmUuEh@LS`$`pc^t0FmNtQIh6YQnH4%4X
z#!f3`tnE0KimsEEZ7&R-GrP`~D1E+>f27B|&ZPUwS5o#m69YVM@FiR)z!Ad3Cl)!)
zLDj9}cn?>#j}LgOT8L@NA_)~Ss^;;WDNbdslTlO0n21~vMiiVPO-EB7w)J}696Q)>
z+do4sm&G|d);FaXDv!tO%UEpkMUg-{1HYD<2?QtJTp=$Iwz$`MP-g?{ruOFB*G82s
zlnaDoIbM960{wpa45|jwk##;b2?JXLH;ES!Zb_}zQ=~{!D{i8uwtxt(r&ysP%|Bk&
z(h?jMLSF{fxB63|(}DDcgk9j(Mnx)Dlk(P0v*(bPHwLinXZKn*Z~cZN5;&5zvOZLh
zb1f0T8n}vhmk_@Hys?6#b8H--Sn?vj)tGj98(CLE&XLlR=N~9&DnvXrhmMb3wQjaV
z@*?yy&LI*_^6LoG+PE}Q-a<5Bq>i)sof!~(P_y2E;p@e)%(~n<EEE2KN<vQ621F9-
zM-Cjc367Gg4wzu@Yd&}&hr$6H@%mHXJdPU*fMamL*MM{YBBk_!{-ze9tMJ-MtZ4ad
zD{47Rh-WWRqD8pn1U)h!Oyda|qJpq92e&s1LNfFPVoM9gvxHk@UAC9Kw0^<H0L)cP
z>0H5>7$`~jexRVrMeJUxn#P$Teon|$lsl~$hXoN0$Ay$Aa#(l|NX>!}xg8d9tVCY6
zo?4n>ehMOeKJrG5X$KiL6&tctwBa{nCS{5O3a=}XSCnQ@H+;;I)f|L3f?rtli#nA0
zDOWW9VPeBaqV0D;w^ubehI748D|NMlx*J84N|)@IqoOH(OpdW4A9FUfanJgGii;*Q
zd7uerhD8xSj_wc`q3JyTYU#A5TaV#5pGrb}`_=AD$yn@zf3;*3OE71)B&@96Kz_-b
z#)!hg3NLqEbFOI)=Z2cWt{J+bS`spF-h2y%nuZ;Y7e`A*FLsI?V;+@6rUhK=Yq~Nh
zUfe>YSm|Un4ys#@G`yIK&FCK77>OmD&*~k)tLcNr2sW`WBL{^ZJVOM{KWH?_Pujtj
z)Iz+wACW((^dQ<Pr^yMtofl|wD@JKmD;G*wfx|(Yy121L4VS~{rKk;*!%~{Aq#Mg<
ztiXn$QCfo{`!Y^L(OrW&ZgMpJf9QI*;Kr-t@w02IcFQlnTJKhEm8<qeKYr{Ye!l>E
zbg}>Br@0F<!)V6?@qbBvkv(INJTtZxUuHBzCJ_)o0Rm1S#6SolfgngA;x29zC1w)k
zt_cZot|;W@M?Ly+Ki#d<lZRcGAHGNDoIZWJ``r5U=^obA2TyV=wJ`xnMvq-$17ZGL
zqnbHpP*%+-DtFAFoM0E?Ux%)fV!$|silcPQBurgin!#4Zt0TP|uSzY79%Y-i=0(Z!
zPyvm|#ZjxYB6XPQrQ<1$$ieZxcPtTkWf&OHvDtl42FDV!R41<Law!|pnW|*eKbLsD
zn<*z2&gvVgTqJK$okzZ$#CGEeDS9o7iBV1syYob;l+ys>g{-f$wW!P8_nIe$qb(Zg
zeFa@!tDFWS(>c#xI_1O+UR+2oAn!4i8~o{)FZ}QBLl>I;fBtK@<e@KMHzhzbG?$>>
zh_CU{NNJp^%v|b@KC3>)yAjW3y3m)B`ME?R3nN~fb9q8E(4#~iHUb<{=WGakNX>6+
z(Z&kATAMSGx%_fz%o!sV7P(lAA{MK{D{s8T;@O)^M48~gHzR$ov|D#2{<%7hLg9Gg
zatg3BdOUFq^X<%W!-jc^v+42F{~c&i@?MO|fj$(E8$v596)7E0&Eg>%HFd?y=@l$`
zN(XmTBg3!x_hEFx((poh8_l(%x5MnAQ+p@iv1Mcl>m5&&UYxe)FMWr@>kTyM4u_pJ
zl#~;OJ8HleW%NX10Jr{}yicS)p5<xhiPSHBk&%xSfhDAP!a6pk`Y!)GuKFOauI>cs
zQ&hl!O4cH!VW1i(5;4rYNLrD5ni;O+sa<a2-;1)?yP#?5dwY6mHt<ii{gRWZ`w&YX
z5R9HQso1?&GAC0Xr`Y96?xgdC`<lF$DZ;2IbDcevi{#QG!Br|J4PE4l7hbCR@p}J$
zPS&DXi%JW}yeMM_?{I2xdomFbJ6q{oN+ANhD2wYK=8=8?<8Db$34>uDc8bg=#vI^6
zaXxkEteS*o=B)`l?x=sqoTorwMSD1Jyv7~?-Gb*J1Zp^6@*KU{kxD?{fd*(k@dk}G
z8<MqoBV+~pO$Ttm?nR~L0MPhwPMtTJZal%v>7N&FW0yRirY`fTV9!M66Pq|;s1F5M
zu+|7Wjv|(w=%}YbU+o*X6<)9oD9(~wNVLS3Sn^R)gP5}3MeYPLR9Y~C@oE7&Ru<B-
zgYrBbduvf&eNH2x3)Y7+!=SN{2)sOn?OQ97vud2u@ub>JDaJi-cfx&P_M<EgXxKtW
zk~p?Zm5Yguv)t`3T7y#+*7U_hDcyM5>&@JvcU&(3^NR_ej~XMrqU!%ji^iN)=99%>
z8(l!%wU{Wq$8FlzqJUZfvBF3Cw5koH6%~%D`(qaRA$|+o_c-2D94I-+MMadDsILZ&
z!x<*cO4_!gG-6^?-S%k3dx^K0OvQT7a>yY!S4sHtJOCS!d?hjW0J{2Oq%f6`lEaug
zffg*PO)$k`DO52su~>4&<Hb&}+BUqt*}!vH4QM-&{EinpsUt9;O&w*7NIwb|C^+zf
z`b<JTl^8d{)bc6o5-x@%^QqK}Ln>F2i*h{zctvK#;sO;vmAZ(}LMcUR9|%{XEPl-x
zthGqH2(NjYizosc-UiF5a_v(l8gs+PZJkdgN#~5rIS<CN(8rIuD;~ac>pMWa70KXJ
zMne@7ktORW)|+hgMwgOU`fU{>J+qWL?EcGhOU_=jBl#r*W3Nq$OR1}s3lB?aDaj%e
zs68}^>=TTjTEN7@izRez$%OhCYB7g?9A>l}G5lMw@$7h0bLeQLx0D74+M8+Vdlt(c
z9gAkldW2n0SYQmRtI+7O;m$p;WR?wgjBJ@^{2N^h(LAD#UQCl&TuvnMlc}oGy*wWM
zSpgFqR>Rt|k&<Z-KCLY$EU}KnTqu(BOoY<%YFuS;dR<Nw)?nsv&#}a<f%KP+FvUGG
zSWY92P7jUBWyRWK*<?x|U9rCH;2Jrzk^n9*PHX>GtgkGSjJlP?rvmEN;z~l2*P2ME
z!(ePlCCV*hqg(R~y`Wq3MwL2fo>vUNUAv^Ul4y++IqiSPtEfi07qIrBGVObv9dIjJ
z?`eE=Sxru=QLdUTr<vrfS|>4W;OaKBntDR7hTrPqD(~OeCEg>&)zl|DrIc1vhi^av
zDyv2%UY1HNGES<pdDUp9u_^RV8mqP)9IE8ZkFu9wp;D#cjC33^(nr~2Is%x!vS0BG
zd-WiL)k}P1A7=}QoU&f1oXF@YqhR)C$@tgay^=d+xM(B}+9wRdDMNY=2XjULTw?!M
zI%OT>n5w>FE2mPgSqx^SHf%yq9&27lA0;%R6ei$Ep@X0cWyV$KRFZTy*?ynbivp&X
zb)ahAD+32Qr8?Qu#v#@qri;<j36E21jg&cUs9$%(KFN78{E0HS^HI#Uz(vnP>A?$u
z(&@AlS0AZr^jL8ydfKR+HrnwPgpJDSB<H$CMjAcsZw~Y{6o3eUrI%RJxzy~&Gg!+9
zfCY}nEOe)n5VLL2m-GW~copS@R8z-*e_lqb*0FQvJ}#N6p=9ypbJZkmds$ub)r8<)
zd*4>{N?0*SshT=zh6bERE7jDvdDvt>9kBG>#M-bGbrIk2??v4^w*s%DGgvxyqDc+s
zi4r=p52*U0A4$&q?5Bqao#Gzrj0+i-`lG{7rufVm<E=k7$epo{-e9)ao`L9Gd!-c5
zq<&7ZL%=L5G7qBbi%g#D=`q2faP`2%@834<HN28){8{BZ<HDor_x4PpJhxyBwScF3
zP399XYSEAeXA%u@gQ)M0<*fBeL%@!nwUPGWk4WaMb;^&(xwDC53oQHjv(^?~F7YtR
z{OCz~d4s;uDV<GBGas82sCaFgCWtEKtsT)N&D)^n$FtJ#rcW{Ayu_3)X#txv8c%CH
zXA%1MqZw6%t9!A**+k$8i^m|h7}BZ@rIVw#SSL50W`gvLWwi<CG5IalQCya&_53Xf
zd0j!&r+o32M6HW=VI6v2N?zUr!B#xMttMNLn!}|#dsrvdZZSdNE*V+8#mIWa)zWe_
zaO*7{FFwU#gKd0oh=#=}=zHS21@*o(a6s37Q2V^dr&Aw~ns1}gPg^5QE4;PwZ#4{%
z+^4O<PZMnR(@8K+&v7g-e%i>O(Z5QmF1yvrr;P|Gnjw146TSmo^9qLm+VIaQ6alSI
zCxOtA(;bJjpk82l-rmyGF|#S?2g0V-mBFQooa*JAHH%?6YsIK1ARjLVGUp7>V_0nE
z{PW!9S4;j}VusE|$aV2tqK95g(<g_4Aqa!05~Z<+HM6yIMwK&cJp2m>ERBGmgkwi_
zF7<4h9e3wkq8{Uti6WBjT_eiwSe=zYKtgMP0LW*o3v;}z9Q{lJ${AKNkvf3VAorQH
z__4KGSk^UN@}Ef?R`qnn_DdJ}VpPLiO-i4!zTm#3@|lDqYb6%+kt9SaVi=(CesyxF
zTx@*C=;%>4^s=rMun(gp=zPYc3L-L-1UP%~)b}iGlNoUSZ8u-?*)(u?r)g#`qwxz3
zug-Ir$$U0Z0fSh{eKxU<;8w6Q|Jk&%S`0TORPnRc#{*pT_?LQ*{j5=SmHVbzw4h6y
znu@J?<_c3q<Fh_8bpO!upO<-f(D|$jy2U2%In>kaA^m4$@8TUY_-rC5cd%`dT0&d(
zS|c@|E-d`PXk_i=>^)XX(m8d~p6h062~bhZbFwujE1riX=U@<2rl9cvbImo^NWNx-
z!HD@b+u0N%7n27mdPTQ1os6tQP_R@>JkUTos#qx3j980ob}9~)9^&<oShQ8gZH_mc
zIz84LQtveu?v~WO-VG>h*Swxh-f9@UUNfHT-ysdJqH!tEIyIy7s4(>|w&exqSETLz
z=C*Q1I)MfUMfP1?M<Yd0>jlN|FFqri0hBjOx~}ydlnutMD0|4Xc~Mp!5KwvB%i@s*
zcgD9G)|X!;Bexn8FuZy}MsH30G5;H5w;H2&xof3NWMUmOFImsTI9hj-IfrubKCVK>
z{Tm$$LhRpj`mnm-HOzW|nd;Ug?jD9<a)`$p7jKo4V|VUmDf_Q#;xCmzfS7U9QuQnz
z7Qn5@vi))0v-sw?f2+wSeBL)F>y846>v_k}h*~%3eA^s$*G8wb90oz~mZk0AbIXjk
zPE^{ZD{%MLBx9IuE$ddL7X`RlJSm&rmojQ|hi{K&-jc0bV{jZbRqY_^&8mO<ZLe1)
zFy?K>p^aAnK60D!Z36Y4|7wPbW4EQLLcM^Q{AN5G->x%~z0IV4iilUa+tNhJoiHaR
z^R+kbW!3eK<h`thNh<i~xywH!#oK(4BXdV((tqE0i4k^NnhJKKD=OcXSjtm13RWUl
zRnJvVu@_ll@~+*Mx{tAz>Wk`LS!3SeJ(GR<ZAlhUS+-?8VCfdS;VC_QbAWF~9==nh
z<>)Qp|Fa!1aB##}yDjl*dFXSu8#QOazZa=x*5XZvGsn)L@1KP|hh)oX#bk7=3<7*b
zWqmuUs}Y#oo`}H)l$cc`w;QQ=iJ-H_=<TU@YRf5Op2^lo=bp^%#;=notFoSfz;Yib
zx!au$IDeXv@c=P9!zLwvdm^X>G;c{E@)uQ^QoP+H=f*uU=^2Yh+)Z97%IR)L%H9So
z-hdu$h&0hpLAvUAse`MUw`LD#?4?3Ipy(T7SsPZ=D12|px>sDCW*NFYiK?E`EHYv<
zssfqa?y!><!>fIJVi}PEP(ZQ+x!p#?;tAHnZsfYkox$hr4*i*Ldb?8?{Pm-e(^Kpg
zwmfH!yQV>qSe5qM&NfUhFs@_mlg}9&Ev=qvMn0DqhjAPVl+UFBpa<76PcfM+K!eQZ
zjN&{t$$l=)K?JMLIfN9rWIO=&zJ#3TKWDt3Tb`A|=ThHy&v+QNqGwJqnN0d89yO)p
z&=E7nLOBpx)|1BPj94rk3sUv(ooBA&PB%&&!-GoQ5yjkc?Q=<p^hGq(UXOxP8!t%1
z8=YXCYC2t1Fc(j%_S=y-J)^II+TIk3V65!|!aUD1y3v5zUg>tqv(Y5$6ucQ-X8Wo)
zztSJlgIRbhnsgV_18)TsJ|Wv)Oy5Du9f>7+)jD!V>XAMm&~win#ue;m$NX#W4$0h+
zgi=HQvDO@w>>Wlqw#g`=1EM|;9uJ7nL9HP3UKEx5q7?2(qyBIiNAyy>Bg9271tueD
zm(ry>e4w+jE(cV9vo95gYVUlifq<UtYMvTyE9wFAra9f7c&0iLoSv0+@8aczv3GEo
zM%*J!FXP75(hA@#L(;y(M3px|FsI{LsP<OM^V0RK*{PFstmmjR?qt*ZJc}y<o}Iz6
zz^HKez<$~CYIdO{15d!S1B<eKM-p&!^!BVy?lgj6(;_2xCdM(#tA1eg&O}OlaxTl*
zoyNy&FqpbCF<iGu*{H|qw8iVfT-3w7yd~q2L>jQBx-&6Mot(@_!GYc2zSoNZ%se<Q
z%Vhk(3qsDrQi{fx*~OOcG^s&1$aW=i_VYrZ>OEsyE;aw;wpQxVIzFwyX3ewk{vqpk
zCgP~yL){fO95`F#DQWsQ>kbp<ok@Ca=6?lC@U|0YH*rZEi=lqC8#o<c66|@Y8xP55
z{JyBMIU}`<r|_)+Z#$PaP`VDhbv=j4_V8OD(aT-NzK!Qt?(Rz5y58;xhB@fZ#_md!
zAq30Z<r5TxU9xu>AE#FA$d|h-F=hqJ?*$o;Y6oW}e^+!_9f;kE)pEgWBU$P%u6UP8
z$HAgZdJFdMrQ}HAD$2qWm+wj>Hm5#1UwUD@qR&Y+5MWD!Elw@cdNZ~lb??$4X0<hk
zt8asK*sMFMOkw)g#uNH5dP$mhC85TB!#%w>r4?B3XRd7Dm3C*009b3gfsW%uJolD#
zz0EVA+KZZ(XRnaW0K80$Cg$gU{6HK!bylzy3^}Z?mIjgJy~B!ZJ2ZPkid|7X(f){D
z^2ta&F;k7AWVCK<p2A1>SlyVZ2Xf4z{de5RO15r%-KB2}b6!R%a8$-U0h|8g!;<$3
zME#Y5e=c)*SM<6T&vUYFF?uL&7hLYkg5H<vY0MMlK+2KQK~!;2JfD`TgThe-uc4w-
z(-vm9UQaT_J0xfXa%)~3^%G2t;y7?8-S9lKYixRZoP1Q%-U{mMpPOazt0x{4T0<XB
zbm9lb$P$68yAFz#k_}7U1og^K$Yyk+!gj9j&Fi~R3-!7+Z2k-?j6nd@kpgbpfdTKh
zdCc?q#6E5>P-cyM-iS5yg*Y0y3-&s5%+nYhYPf5h$$vhL3Uz6t@7=OqA0`~q)$sGi
z{UsclkAFUO)xB>^-b>EmD49vB5RiuAEX8OWH=68wCnG&SrS2+9pHG6um}61&rf|9_
zm4HPw;W=(7rRr^QoUmv2feME{jeuPDy87sBE%4_Zf~-dy*Q-C!#;_b09nH_D;a6F#
zq-Z5A&!6FbyB#=aXblZ9(fPb_kIh;)+CMmbTsJPgXcVtjvibQm>Fs7O>RG%G&=cfV
z;B!#T>IacD4_DjXyhcKoyM2E2<Kl>?V9E)#W4YUyat{9$W0x`*8kxIoAhOk8lB^fH
zj0X+5yAzXViL#;UiSekmgH+y&9>GON;qEj}d<Y@}#k*5)=E^b|prAZ@cus|PC;7us
zH&>tKNaLA?3**Y&Mm1byRK3xmd*+%qI>i)Ij~4c#oV+_pg+3^db<e}HPp@wR_B^J6
zN2D3uIEr$p6*VmobUO<31Q_*LCr~H2nE5>G-kn6i^r!BQy%#VKK*ddmsredlxbMm2
z|8C3sut*;UcVGD5{{PJ6)Ijck^33BrC6IfJOWdz5$jCj3xI50u=sm{Q%dWjl#_mZ|
z-V`*<+>>a+^HkPO2cS;AC3jEat@><Y3X*ZJ%KELFgS<Czx~y8F!ac@|Lp!ACKclOK
z|C0g!&?;xLQY4ob*+Z8D74&#gaU55s*yTjI(=#)?WsD>bi_oG)wM7QVT4b(hx7QuY
z5>eJ0QI5zQ()6y`bLy7kh;cVtul7A@_Td4wE^(cxH@i3|-N-gAK2xm&LZx?JHvOAO
zmh;kgphfc<xfL)lV3L7n;R;4)S;z4BJDdvdP1INSMly1*5pmZZ8ND~*ghMKggg17t
zag%kB!)4~)G%$#qiUawqXY$k(g-6a4NU>@_?=@ztuAujl<UNbA{_j^x;a(F7!h}dM
z&^VyJu_uR>G>R>8fW4AldNPxG77)(rWO}cO8@Iq5>I36LiyX+f;*O#R;ZogO(|B}i
z{&}3KXgwOn@`<%E`c+1WghD3pz4lsZd1H9nIwoz;;Oaxy#b^k3($bBTs-=7PCK;}n
zqLUdkVD_Gqek7*nWXl^s!-H*RWPI|IvK?5ejhe_867{D}sh6mcFBlV5o2fhb(Jwe%
zv4t4>f-!ABH#^Lr{yT$RvR^Q|;k}np=HKWTSqB3$N&yB{-XTM6E)>s5;S0u0ZWnZ-
zEjsXh9AYan`Gq7ev+kU+^o7)Mj}Zpzyga~eLfh}4NL+W_byD@LMNZae|D#d8RF}G=
zt)2$fzF^{c=9sKI(hPxDq!FNYV+7sw96NKjBIhupsvQ_{S8UPXJJDU{5GJ4Q7t#t5
z`?H?6USV)-2DB;8<NYYB8U}qOBwJpVm&h`Rvii(%J5o1tGL!p^__)<NDI@nKRx$-+
zfHQiZ_3&eiH!sEeJ?=H{OQRG&a~$bCxr>Qg1G(kyOY`imgEAgPu_}2;^7ol6V0j^h
zDE3i!=fBV34Pi2BFmara(tXBby%A;Xl_OJ)h@^5~;_?v=$PAdO<69&m%|F$9QR?2<
zoGFCYqSlm#eqiAXsN7O>9*ktAjMZQ>Xu`psHd;;zLQ}P*9YuHO9d75o#4wgEoJF9X
z>_+W99_-QaGK-oO+4K}*x8ukW=|^gr-RxEn=;*WzqM==SKe+AmTBHHF-<Uyzwmlm0
zBoLWExM#oLnXBGj$L>$m(m-YOBojZdAJrFV+54S6SYPDsPphJ?0OJ9b5J4R7?+-Cy
zvoJ4($UZuAgp2w}9#aF($!JDn(bI0pkvg-?-mn~**bCJ-16LdeF1n@a<)Abk*6vSI
zmAy>rkx*L<N2cg{_cOBY09KcoKkrY2ewnen$XU1<m}4WUN^Nffa-gE|Z+C|pUBGox
zTe|n#IAB-SbLju^A6Lp|z|u!x#FmP#%v4!8QLH%)(RiYqmcji=IQ+_=0ls|+zu}=&
z)-Dy&pv)Y*M%IjlSljVPb}fxOvm74tEL=}uL$_wUQ!D?h1KEM6KY|C?wMA|%u~Qwv
zj0YIiM@fDy*vYe+lTvU{i%a-H@)p!sL?+h~HO(WdGIA|R)Dkl#Q%>1yvMfl&TO6Jm
zs?ok_xuZdKlEa+pxh&A?s(vlX?&P*)ea%RJ<$p=Tv1S8cVO*O5gu%p)ptTlTOf2TP
zzfgmSb`<iV-ib7Uc9=oBkydR8dl~Pq?i^%uNcdyOP2X|SphT?gTaMF>MH%?l)z`|l
zmpJVHk>!hqHO^B;zUX5LfA*taOax?3hR`V6zL?me4)Zf#G)|MXB3aMDz(h?ma$hvQ
zs~%t9_>VjEE3S}yfK&(5YE@PEq74)^_+nlS_`6Y%O*$0j2IhUZ=TMGt!Q{p0omx{>
zJmt6PQgwJUpTcxoYTlWRmzJdN$eJ!@EdUqQE#UgF3K%R%!&})gElvNKtu)udtw3Du
z3-w^vcGy|Mn0Fk5Q*)-M>3Rx|>TPAu+vTjvk-@7hVP4wzW|mp@yg0o)!IPvxq%<z(
z_7{_+bM3)OC|^pm3OkRHFD3TOV=1Gqv_`*_h8yp#xJ~+!vGPg`L^5A8KpIU||0Me*
zpJ(-vSI)mJnCQ5FV=RsVT0mooUXj9=d}XU~A9BAG1C!lLG8vdKt}K>7=}T!|VAj&<
zhOy?@9Om+@66LuhXWn0yYP5y%AlFI(ofWnoiEI8{)*M0IqRP5M#I>z7ynhf8gI_G<
znqNvIo#&N0-LwKosU+=a-(4y3p36H&{Ry@N-AElm`_K#63{h6F&4A6wUI5#^<8<;2
zMsZt1niB<N;Pp&fit{kp4meZO>b^i8NZhh;P9qN(Z#{(6XyldEqGarW#Omc$S1V*5
zFy3B<@9OLWiPq*hT#pSohedbJ{fmr8Vglcq0eDtVB85oy8lYHwAdyY4eOQkty_vHt
z$E637EULfD6H<P_rVN}VRUWwTKl>T>DC<agq_27bf+p7<FljSSBXx&3clNBTJ&<OT
z<BQX>?uG0nRKs%s#<K}&dX<eQa4_XC4<D4a|2Tz#Q75RwE|9k{>|z4)K6ZsY?<C0S
zf-ajuFTGTmQ_shJZ<`etJ<PCezn>mBMdl;w*0xg<%Nit59!xaB%LaDTBM&C}o#vK_
z0;66)9rK!x(6I-N$-0VQ_w%5UWuu9FW*<x(*438yGLn1HIJ~;?xQst&q`vx`<R47D
zw%CVy<S%&UjR&OY@VK7layQ`VyF-<U(t{>$*z?&6TMpFs@w}0}saIy(d%sj8?|N?I
z7><rL;<S5@)E`V-n?N$x94e#SoE@$^(i7-e8&PhV!*w%2OtFiJ#OX8A4w}(x-Sc!J
z7w5=bPjDrJhk>5sNN8Ekfe)tT(=N)~IJzHIOP9-5ATZ10je#TTLF|`hJ2DS3yUUl2
zHfnS*%Yk9Ulc<BuR&Vsn#tF>EWXwM~j7sLqi4HrFvFw+9+0RPu%gJs{{eZdm`tdIt
zr}6gwisYkSg&Dr!0GZvT=pfErCX-)IVr7enlpG)E&ndGa<u98MfrRqSL1Pp3kW{1G
z^=jf=bJ%;hw5Uf;W$7$j3mkNLUUztN?3_`$L{_m-l&06Clu@GPB|)TgwY0ySgq=GI
zH7)DJ4_u#O<}TfUwX6Zj0|#|oA)8-LqO3MYw&m=5dA5N1#A++b7f?{kz<aat#97()
z&qa-0{*Y0Y`v?}qhZ08)>(G11sDzG)#F!`401us+mm8kJvYthscrKWG$XKl&p2r_b
zeB@S=HD5hN<sV8@HXe80mcm1cPk1ng;6+D<O)N?n|5h6SDS3k^<pCt+htg2j&88Z9
zRJ?3anNf{SXt3p4G&rLh;rc^~<g6+!ePSM^86dLm*b~lJ8vcn3Txkyb#6yLae;(uP
z*Y-B`3AuC}in{HilieZaO9biaIdR-Sz-F|*&)o<11G@>Vgk>uz$(_K!X`=^WbNRXb
zkPVOtY|g%t`fqOLA@#^td?I|7iE#8Q#x~f21^L)loE;qLGG9rwhH6Z;*{>v0Ozg3^
z54o>c13UK0c!2oGM<gHRbYf*4D|{vKu%e2HVqn80c~TAJC%vUZ_U<39mcEk2YA;)`
zvg2y^5|yu{Td5Rw#rau{CWwWGCM}x4=8d5-Cw1@F##5XgqF-FPA)M=x)+MGQjcC2X
z6-?71Cl8w~M+QwTM~CR>&Y2~45z_IRHnuntI#HHpRNLG0dKbTl5N}33HF@L;qaQ!8
zn^W=JiXU)g%=#*W_+j6SImg(J4r9D>USvJ-@UqKfWIZvC$a5GLjjkL0_cFEVd^5J5
z_-K#Pc{Q+ZJYVFMVs<@I{)5`O<kl|@NMwK|lxxJOeHu#hLq2{edaTF7dYWcz7(;(4
z{?kvG5bEZ1GBCyyY$;$&>3~Jgy>9ZU+GVM%o0#!*r5eT9F6$&ybIdCk_tXRPqiFZm
zqCt&XL<~O+Y8>Qo^_|~XPXoe@VbI|?I%e=It$<Zw_tK8k5t{4-4(dmpj&y@a=DOm@
z3imEOZ_89rn}JGLuJ!|>Ug~Z+dl-MP3ui-KcSe2MYzJ)fJcMYV&Ux6lrN($84=4Ui
zO>@a@&bvn+PJ?Zy-kpy<oCM?MPe|rrqbi4Hwm%OiPOERNT-4&I%uF1$Fj>k!oQS2~
zCUNH{g@=;_?VmBP@5QLDH}NtV$-_MTm6BsTM<BLx0N#r-+|ilacm~#bhB|D3tx<ng
z&p~ysj|9qEwCpYwWqk<y$XRJb9-eOWR=oLedZ9M6aLKn|>)|vyxJZH!ZO7u`?RaYL
zI8j(0Ut&;ogS;qb^<d55>3NBv_s-2A4cakN9rhiXMd#V_KAp7*U=Yo+6vMmi=woGW
z<8_fo65ALr^D^>C;;fq8y91BWD8d;IW!)o*EjWtjUL6I`BdKSo@8DuKDlm45VDN}f
z*9T<$kt7s{_DKGb#LOb+LMESrSJNl%>Qt-vND`DeHW-ta+Q6|U3#|hs>`yt$eUwSQ
z5~&?%C#v3{N5-uM2weWKEvb9a<ue$_;}gVFX07?6tOuqOFgRfD(}{Ul5QPMLyw(uN
z3?^~R%ubsny=M1l*PWn6?+K1w-A8N?efYAx^d3pml|7`<jr*np^GDLsk7jY`B3nT&
zG*i6amBAx6s94$19{4x@*zw5sXyPdTSGn_eG&RG^U+zyaR3CjbQFi&vf{Z<CD5<q+
z=F!ARPK=C)Y=FYu3#P=E>T$;~9w}@>CI6_gwz$aLD}_fBYt^usxvJ>7YAm)WliuJF
zrpr<^u&ig)a?~4!FsVH1+l43b&ryxC=D?{IWsB~3xgN-?s+S=f<$GU;^6Qc0M1E8n
zf&C$v-;B&+88u_eLFv4yyDw=+_WTT(bRzjN4OiNYB*qtPej`ceQKUyh$1v9Gd#1j-
zS|R+vmeXYsJ8T#lB|pk&*pB{jKu5hV@>QcSjsQ6Jj(jyyn5T3+50=rdCe{_%BA~zf
zs#ET5=A%sH@j?)#c=oGFBz6&<#iEtwzG^~~A`py!)nq~~K=WTs!tybMzu;hZ>7Kv%
zRU6lz_=HS4YOEw@PfICKQ%4TeDSNs7c-{;|ZuCnLZ$>>w)*Qrcd%s)vgt}6~rJg62
z>{<W1*HHJfjX;dGTbhwtP#vD9h%2HRthRT~4b3^cd^OE+deUKS)5SD;@ZUZxy+HE_
zrycIIqQcO@BP(s+3Bk%VBU@g1?yQdCK_G(BnQVJ8ROB3B9&?twrK9q(w4Y=#FuP9Q
zPmVrjOy-Qk26XH(<AUD%W&HD`UWIs(xs4`G=N?Pk*~2C6xM#r!-iznG${rn$k0s%_
z(e9dyk0t3b_gZhsq?g5-Q%X^mlm6nYlp{U2fHw}QJeH<^zk@#`)yGT@h=5lMY<AqR
zTj~yU>J&cO9!n%Qqi)+ztb4V~Z;?hc!N#3gv>B!SExqncTaTq#?fO+T2kmI-W9*bV
z-XElL_~gCoP0#2CIDi)|n9HU&yG;0y0EzZ#O18X?s-XstUHD&I&81FG?Lh48*>(bR
zLCtuTucg(DO~A<4jJXw!BRBfBL|(gAVl(8K1PQ!GGG9v_nN_86Hj;=4SCn5%)MaQc
zsp@F_YbFGTb&ma7V)pVXhR5DKcAk_fdIEVSG5IxPH}e4GV5d=X5CxrKqqzz7vQi0%
z8eE_nDV?)x(TjZ^%%kpLR*z!rA6?(gSv0EYja$Q0Iw`S&Gy@1U)5BbA2!vaVcHn|R
zCwm|#5|bAPQ~%~g8)lEu*7c|oqAd$X>8YWS-B#os;G{Y5hB(&9b~N<=%rGO%<3`b&
zP*0EeXPHy{=;KCt?uwKDv9ZS!x#yVlGmj?`rYEjjlC0O!9W+5l)odpBcw#-ihV&)%
z_~RxQ(;V?69{?v<KMUUW5{LET<0eV&Jay84S9f!z$CIp_TtrsN4vXUthZFzCPK8~L
z*O;2dB)}mZH`Da`<2Eocm{@x}O#qA2dcm^pn6T^BDXS69*ga>{tDXC>w1y38V5GMH
zx^tIw{8zoWl&-f21@QOW3pnP&XY=u-w;MiA)a_?K%5cqvsc}${`=n*{SUE^!fp6Z>
zo+{LmhLMQHPwhAwX#pcf18QUlKiWuy*H<PiC}WMpy~AjGGF}}#v;QhtPoOP@6uGE4
zjFd3m@MY~<$v2F5dNz=P|5`<P*GNq1J<i@>vSHGMy@8axJYMwgm9l@^<Hky$vWy<9
z+Au-W3oSdC;rP*EfAVUnJ1}l6Wz7K`))-*xjl{lP7AUt7IIE!($JS=FplYC&_m2Z7
z{c8{LQ{|6W^BaDhRJ-1W*~avuTNgjaz4B%>fkGBH-O_iop`dz6(9_nILskR5fwwrL
z&d_Dsf1iQu@<ieaM+c6QG7>+q@~Ex;=o3ccGb)E;PZ%i)79kn`T!q=$C(@MYQJxrN
zPbBv1Ere;?#{(`F*y_wA?{F_Z#QI<GZ=OFfQ99r$0+LTYVIrlj&Y2^hNaBJoN+~~K
zGBClG#p_y_wyJ+~G-ROpH&Zxv;ZG!qSkbNJT7cTeQ=4^fkl+N;h@PlPi}twbMR9?x
zIJcq`1c_I}p|)dKS=wmwtY(PG46358m&By;m@iB{&n&9L_su8LOr_^~rYa`$exSIp
z$fCIA#qB&d@J3Y2!mcw=z`v57W##KeiCI+TUr)Vc$-!mg*Au~%Ut^Jpy#^=3uP45r
zUbVml*{>%F-tVLC>&9v<8f5(I#s+R|n49ySV97;}3*H1WuD)N3j>)baGWqqy|3$87
z)%lc^0{&^Uk|+m0*PA0VheYI|V{CD%f%*<+&zh%DUuw*xx~JF?sM*O{l>EOD-gO5$
zGYpN~@IEXskeb6jtSH*8z=hZ@xN2>CIpV`e=Mq}0Y$CfsCUarO6>{`ULy%Y~?@h10
z@e)JOv0>(DV{x?QZ89Klx=se3*mxGaW&7(sZ|TNBzLD5;8q2nkZy4FH;4X1AGUhob
zkA1`VbLlbZH<ApgB7KG_gRROp3@kTVJl_6>F^$H1cgna!)s1)l8%anQ-7iYPi?N+R
z-(CDh5(@UbYD75c^){Zvd1t_zyh+NQ(92{F^xsIsUqjO90k;~ztm1S?0|?Z<kpzP6
zB?fp1d_4djVEJ>*3mS!Z-80k(ft#x!0xPC813?bhI(N3bEE{1$9<;w<BY_iMCu;3N
zM!Np{hpv#G7gwTuQ#OMVl~s%%`d-IP{AA0?P09JGGdPnN3=vd^klPL~f0@XW#ycEC
zsbr2kX{2TOHZ#-FCyfIf>e(-ic^>d^ipzE5lg8$aXBQ;<q+xv}flZ$@_7SZVGY$Wh
zTlT!SVVIy`dor;|A2dPP;**K8Y_i;k{G>Nbjhk<l(xnO*s6n!_H$BbTUWvRNm`Gyf
zQT4nz9C+2D*_mT!r0%F`@DN!`g_t`1XFVZFBPwV#hNgoihZWU~HX@r{m$bd2oz6Rf
z1u=j1FyDPL2^7|@W?j&WBF9cHD(0I3jfW!Aj|%vJHwDU8kefJhxc`ad?nQ>&b`a>}
zH;X)F1Iu2Ms-57-Q;EWQNuh>QPdSrcVDELmj6G#c!b0PI$vl-19q!~M`&8OdY0q_d
zAn7@80e<kp9Vs4v%9yXN5AzNU%NLFdQH=O+%(TU)e2L?|l$T;Na*ve!TerAUc0fni
zgIByVswP+ESD!NBWBp+ft9cJVfYZCw18WqD7O`c`3*+ya?cKTqC04dHhQI56M4Fyp
z7pUXTv4GPU-*|>cpYM26h@#FP-KWxIgC>&6Gw?u1E1N^QzHEn!zC)Q<Im1e_6+F;~
zQSAK(p3BuNN6T$b(4F3^BHuLbaG24Z&B!-XgS^h*ed0HbS%;Tc8^*qA<h=UdCG$<A
zjr&{4dJ$S>+w#rCdXAe5%O_<#z@ZyFFL_U5xNPU+!Z&>oKZ)6*L&4%3TeqmhX&hS(
zM^F_l4A-*PS>Q}j86xxvseUu9y%Q@M{jBEDxc|RL>i%sj`kGhfs3Gg#2WW8FHPUcM
z6L`4r#`t*$s@9g*-9^jNj+_OyeI2jvo7Ju-(EXt*?s>Tx4^q7u6hYv5svvzw;lgc;
zvgNHFJu8Fgv0gCB_K+BA&?HYAjTTR{J$c%CiBk4Ols<RKsAny)D2+XByg9gEGEXNK
zD684%W#b1PEa?qI&a0_6{qd)bk&JkzguMTLh|O2w=`=2F)h4s}bXsKb3L}$V4=SCz
zQc9k+@w7Q}C_5%t#8jR(+4)$RqiWSrr$WJk7S&Q)+%64qu(@?Dki{fWN8Y*@TzKbE
zX*``obI%+PmzsgmQI@DyR78X{UP;?a@5J|)uCX0Q#)1uWqbBwP(sR_pa(mP9vDcH*
z_vV*4^>0NNRLjhW@~o?_x=OYk;2Z~9NF~igstNQDBhAE*dGnJiqfKKdn=ct_rY#7Q
zoq8??(M;2)$CsHNvrS_URJ*Pyxn}A<+cDO+W?Ht~O#stmJ{nfj|3WkIaGw2w8R-Xv
z6(%8>bVT;3m-J}pFng1zh`^lG6Mm(chF1lbt>~$}AyXbdB(<i=9$xh74$(r^1Vq%R
z>(MakWArCMk_lFN;5EI^=3=(xu!$qUsZ81q`|u&@1PtzgxdHZ!{TM_=uk}5|W)#DW
z1Izr#oM4XM3Yslzyb&F(14o$`4zleaue(X)8Dr5|{F9G7lc>030Y@Ah)<&O6#F$uB
zVvIeL*uAufqt|DQeV7Tdn|#Ix<-FvcNxb_QdrKMjLK^)^^1}*CI4Lg#sOte-jMO0(
z(aC2_j?9OzVP1MB1i)XrVb2{k&}T^{Qu}$%TMZP<KSk}Cw9ZcP^iEa8NNzo9U{5D&
z-XhY6-K}2Qtv{2*b?W4?;rsc9_i^VAX*$tPFy*xTM})TRKYoG`Zk?c{I?G{CCtWXg
z=Ct$zIaO`QW{?;mSmte^<K<ClWh<~$3y8rpHbzvx$;Q&NFg?(@L!M2XIfA&4Jevm7
z1aB*zO-!U7I=iuF(<oI39z55|JexSCmJ5=7)(0)NSGi{s-R%XHTCI(H?zHYlo;Cj4
z3{mj*S(5baX3-0+vH_eN7Qz#wlmZ<3FMk$W4oqj!Uj`y*A(%+L>Y|zl)FKNt_Po@i
zPkQ#o@!qpZbZsj|<#k6$*CA{r8j-K|%4)9Dbl|FTkXAs$1qZ5xw&(F=lt6AlEkK>`
zy(ZlNwJSvti0WpuDw|&2oNe}lFE_B8+VYg{6`8dLQBN;x=VUv|>Ooq*m1s|tPu4eH
za*o7L8XS?WE$8-cC1xs6<Z<j<iJ2!=A)Wb_$;HqaO7>fcOw0*$IFgdww~U9F$iZ4K
z%OGGuki6$05ld2toSR@{G1|n(y=p=yy(k9;c<kS&FlvzUx02+k_fYIjzh#reo=K^C
zDWWd0feDPcHeQtarAlYieZiVnnqeCg1(#HV&=_{@W!)r5^IImhnC?j{AbiYBD%-<?
z2m+gujyJuJSJ~2yejP_A)(iNYs<`U=R+<8m@S}M7mJKW>CAzca33~w`0|%>SDY6}a
zma$b6`F8LKxki{rzMa@`0HYpt3px7jpp@PJjeXmCbLHh9lFYXgLz$B+lDz~B)ffv_
z?%Rga%A#IMjYr{QdUnZwJIxB}cd+nnBmaTjQjF$MnV|5PjB=Or>ZRn3ALb^t?7$I;
zBd{-;=3zV*zEbsq>O{2`4HHk6+0}FAdDgTRZSi1^31r=~F2m5e;mx|&3u#6{W@O}|
z_3bnOhKFP}I?)PG2-?#5w(lR_z=X2vn4)@P<J|KM)*Nj2qi+-Rv>y<-RX}q#ET~)Y
zfkR-=283<zYkJp$=H<CGrPA?{=Mqos&Cuv`iF>;^$f16DE-@Ti2lbnqdCn-m_n>5-
zGsZ6J{rhuiSu%};d-WXme5}1BA5f07fTJlXJm>T60&1>eH2E>ExS3`HJwmRRo=cR|
zwalCXl%F$6*vp;-(pCb0l%MQzJQFTYAg;9lg@1^;w|T>k1AAmG+C+tkz1_OwiUdo;
zi=Ec--1P57^|an{$QXfe%^__k1A!4c(KoxnI3?Zqfw@p!toIzs#oN`=?YU%uqqDo&
zmZK!>JG5yGP6K})RLCGwJ5_rYfYp{;?~9&K%;FpkRY#sr%);Dx8rdCv-pGvViF^0w
zja?g$Gm}1V0x)G)uG#040Gw6>)|^)%Xc#wq0gtn&9)I(m$vX6!6fV`%Kmo<)O)jP}
zh>47f+M|~O2&RN;gj04PJ5BxRNX=l&Uv;ET&Q%cLTEKwEWvM@(W*N2nKf=Vl_I#2H
z!qK_=rgg_)c3Bz@bqRCr=JSc(!#7ZFL+nNOAC|Va%2fqZS0`GvlXEwcga+e;_`Jy-
z2pmI7H=j=;4<mgm#w!f3aj>F0GnnZ$5JtQ=%i#Gm3A=Jw?>V-IOnK&n$aj25;80CQ
zzGDnhV;%G`-$@L@`wk;=SPnUq%y$yQ^;J0fs%(_i;2M(qP7;vQ+=-!q`HrtFC(vhm
z6X3S1AO%k_vvQeR3>fq3acc5A#@M-4DS6ANYcZX0NE@%Bg85FOgoj};3Rl0ArfbZ=
zrRJsQ*wEGkWlVh3TlSjg;b>IWqY^jb((u%9#%Kms_~#`g*LM;z)nWzPAvOT*sD6}*
ztrHC)x!kBpH-50#zAs8INH8%Wn_iinQJd+w?_EV_4Ht(5*EekguX56Eg|<CGIA*q*
z(lYk(CJ46)En}bFd1C|KGMZybCS$F{4!xqpeBbNrS7Xeqe=|#CJJCuUC!jOKFsk9b
z3E9s_T5k=wuZ{FU^X^(~CE6cB88YeZZ#;8KN{-i&vOZoYM=3SGuC$UQVXCD!yj9Qk
z$b(Yznrhj>dC^f_IKwT>ntz)ndEE;!@vw?E0wC{Q^xWF?Tm<gslC-><ni5Dm5ZT39
zpwqIUZa{nz>qhqup%Wu~LX<y%_-wXJJg?14Kd81#od<4td%TC^Ofd*-SS8dFVY`*)
z4|{lO=Z+VQ>dUJY)3=PgkV?>J4nA~t^aUfVt)wW{UNDZU38-W|ht5s&urBM6Na>kL
zaxbLW%fpqz(H9Kqk6$MF$YUD8nwGmI2WyjY@rA?{g0U!$lP{PsRoL*AB2QmBU?fov
z916%*<pq<$T`Z|brRoU~I>KNaHvS>0zmVh+Uv5x#&C^3q#H@-Ak&UQcWH%f!6k@ka
z({pUQb6#2w9;f5Hz-R{=2WNQ~6<DIpm}586^|BjHy>#sbp^mEur_Dff?yU49QDA=>
zKzGx(LEvpS#<K00@q~zR_@ecD87o0f884;|F3##oJo=(BkOPD2r^mdgKJ4UXB8u|7
zjcGRfqOs^BJf?XuF@YdHYPdN5qA_6=XRmqx<Qhf_{;57ViVk=c<59+60EWgKV@h6b
z2K&6SqXSzsgi|HjnO<C3;bd5K6s|icwHK2Fxi=u5k~)+*{QRxGm}ZN^stR6rpz2x|
zId4Q0`xlQ$Gm7p~x3#T6O>b0D+q`HJJvVjXHtF~`*VJE;uJ`Al`F7|9E_Npe9XB19
z8?k2YM{|4Bj@x12o9Z4J4B_303BvY}KBm%;mlFENG0~KfmwXviqt(%ujL@fG`Ivv6
z#Aqh-k}=WZaB;+cDY1Bka~fxeNNcpWc^Qv1rkAAip0*2D4%||3Xf`z1#k`aTJ*ImE
zYUj<*WZ=h2!yNjO$rt5!PD$CZ+IS1@y<{w6&QI^hs{vWB;%lB7s?O>Glx;Rn+Z@gf
zdk)LGf9_IuP%oJXq8M*_ar9Q^B)1i{XDX}*?UzhI)l-~1yXyE4;5f6uT#hkJpkn(`
zN0nEudxo#eHY1-HIns9)TzGinIoXOFUd+qDYpQMG@Tz$`$P0ClK6%;5!$w8-qc0n$
zD(=y2)Uy_GKhNa(@+G#dGPPzr6O?bhLb6_u1bc1Cy=-LI1211TQL*3#h)4cq6OUb#
zDR>Swcm;(4=*x+X7P@XuhvCa<2!R??DaZRj!QxFxIa+0%k;=;^hpeDSq#8LD93<5O
zxV}xXeR18fpXN4ZRo1+wnuW@`<9={j8vb<{joZs<>teeAsD|Ifz~6RH?Lbx^jOguU
zCs4p_K)Rm7qq6F;dY(d0txLpabc@XcvsK@5Q+s&!GSN=&i`+O49I2&g+4e6Ec=C!7
zN<Dt*eLni^kyp|NkL|@g7mqeJUr7wyX)f%>UP&Y$Iy7cVeI<=-jYcBbNG`3iInKRe
z!U4-}ICxmbUoj?sbdTh{eSSyWEyzaUl{Dmf-;rWeo9BQt87V!llwR?<Lro3J(b{f{
zH{dW1+Z<6HmRmsETZ`t<L1A0Y1-%zNq)q~5EgHZ{j;z1p6A~JehQr6fu*ENLdL?w(
z$Xd(u)K*p%Q*Ezx<^T0B(s8O0rj-j{ompQ=YYS#IYVO-}ILqe5e>3_p%_Wueokkz#
zS?McDgzUAkJ^3GaO*J~<s2DYS4SfvtY6w4rac-8O`Kt9~2J?Z@SB<>bb+A((do>}X
zM@kUNyqfsz7y8)%1;uTV&An<=M;Xb+CQ9pr3+^^vHMZjAz3^&c_u;}GDZZLS<Cwm6
znT$*|=qh=onX^)kV%Ux|CcP!3ko~C_I!aK5SB)54Yusx_!>aSHd!G=Z35+bfuNzOp
zK!>nLgI71aQF!r?D&3-~-lH4<T9KU4LDi0a;T(ihOvgd);ju<HN*+VCAw6&TV{=D2
zw{ALWps}mWqauQW!$n6y_o1>iEUX59GKe-ftWj{=3!YwNb0M!8yHsU4@|qD+1&CAl
zsHbS*ay;vfy=JU1C)=6V5{dB<h2zB6l3lmGK@0Clxz~KS4~_m<oyYwbK&tP4<Kk;+
zY^PfXDZG}%;(+<OD@JQLs$#Zvd{;JJl2YX0N}aoV&(W2&61Z1*dtZGmNgnn^s<5cN
zX0o#z%}r#LIT8b8&4I86$+~|oap}{DVne?d%|P<7f|Qn%=i(zAjoYtXQVggqnjP7t
z6CjAOronl-UIjO@xKWHulz}tEZP<*iF>9|%-%;TejBLG@rmPF+W#GS~g^_Ll)E7X~
zPW@H$5BAg}ZR4>TJ&pRO`nMbN&ne!)W&G2;wY3xZHS!0h7Hy-|uKh9|IRwhXf*|iX
z69=WxHo-xQ_9i2w80GB!q)bMdD^3>VlGnr8UpaDSEM7y!;UFIcdRmR#Ln!HaF40b6
z*IQGowYEtlu`{lbwa8+L;B*l+51QMGhG*|MN@x@SH64MguaH*Th^ZGXTxbN4{raY*
z<23o#tFDr+e;%(8($i67X;q8l&8Sjb;@v>MohCzdS%STWY<W}Xu9ZRHgi?W-d^^(a
zOF>?@?y#-GWX->K?A*UcMqf|8#_JjnN?%Xx!6`bERmO|jZB6!dV>4<i$$2pbGM@GD
zz&rlBamZBiaL!Y<5Wg;k*AvzRrZG<|#b~p)hGucn;jW?<D!m@k_o$R3F*zv}M`Uq<
zlUUUuriGp{QF}d&a}9EDuMq1Fbz$bXtVOOrm4I;T(NFfu9HJVpCz;mkH{`PE?XeHR
z<AQ%Pono@)-w=musi-uJ?$PzkX*G%OMYnyw&uPQSGSo=+y=<8qN7)L3@cJ0<rGXr0
zGH}`UipcFwUS`M}X+7G3m%WiU68~<{Tl&#Aj1ZjOlh$0u-Y~Xpyv33jRcFC4`$pn5
ztY>483sCHOI4nl?(2g{+?cSl_ErAp#R5$1-zL7YsA;I*XVe$>1)>rlT_=b)2TW*o^
z8;Q$1DpCostY^1N)k|HsUus^L2L4Ox-UkabehICa72dn8doi^Jlg6byrho1i10nR_
z;KO#+0jVCe?VnuPNXHRkM~s578wlx}YU#a^Btx%?8Qz=T;0a}N-y0aN;B9;gY&rh;
z=@<kYoaW<)&oi?Br^uT=ovF3<+cNTIVjyocIXS**+}g=pI`*a!4B^kbX=L2J<M1BI
zdI5C-oAb}Bue@5u-%Pw@;$R%-9je|D=?!J!&D4ouES{p{0C6WcGE92uT|7ZkEA)~#
zhr%7*%$sRs=`Dr5ZKy<DHG-=;me=i*THpY+Ik48Fs@ioRBDgD!BGYG-9qW-+1v~yT
z-ZaT@8^UH};)e#bTG8%JOmOX>+etNx?|9+eDIiDJX)~j4Fufks)yvpT$9Gpz`okg%
zr)2BRv@J!4Xc0ZIw}}R{dD(W<7%9v^@>UvyTnfv`TgIpr;v)E0_9_{B%lLR~=2Mb+
zEAeJ8K{VLVyp=f4QRcklB8im@CQ<=<fy@8=TZ#PKPa%D@U5Ek)n7E3UQmko{(bO_3
z0><21Mo;cQ(6*KxIaHjLs5F6|wfa^Pj$LKK%HU4%EfcqM|8>j9rSAPe5h`n5|C+01
z-M=OlrSVo8^0>)mYs85^x`Tg6b}pBiqI2!Y#P)_YFfs`&C0%do!dn~+9RpsM;j1Rl
zBgT|I|Ll9Y2@NFkR&t)C8))p!-Cf2YsKyoh@R^xI#4R$hfZk5r+cAq@_DJd8dh~6h
z7(*R#9DCb11tslpn|a&la#}4vvQcX%N}k-?i46MWOdXt!N50x+<|B2AlUu>tVDMdi
zwG=&zUGaIQkIBgTZ~rMJZ=Q0TcHrOJCT1s4*$k0HqVfI}RSq7KS|kcM%82^RWb8d1
zeRY~I>;8@2mg7VqhchE-4q+2Bob#X;JD?hYc0fZdcAs>H5Yz&-`?d-DQ7k5U{_~V>
zlQ;e6GQs@%Z>LTC&}|92Cbh2I3ZTmsL~-Crt{?F#i08|<lk{5@PkV}eJ`J8d5L!mg
z8!LH+qI$>C^GT8!Eog~k?0k%C4LYsL9m$+GGVa(R*??s(+j8d<6L%>SIYW=1_u+#=
z4Rb}WSHgL)|IDF)HKynoz$Bd|CeJ600~pMreAc_N(s^G6F=3Fhf1mp#W2<sLb<ei_
zs)MdZ57^r4>&4pnG<g7!>8tL|;g?^jvF4ayMXr;ngFd7t`3?U*qXwDH=%z+Im)3b>
zma3!Mk#k~!m(-o07Hw%D96d%v_i!^Sy})Kh7k$|b09;1%{4=oUq^tM3TbCf}i}K|%
z5C@JL8z?*{ME%1mq4M3tfupNXlMTekcMaK%Cs`T4o2Z42Hx^_NV=T&}PG`f)4(GdR
zXdhxa%zoEceG8WQ-!&Gn%AAn#?;2O(*o|maqrwbpd%=q#d)y{|H%Xd?Es;qt$cFbk
z>W|V8-W;uzzni4)I8R#Wb;UD@bjHIN|Hh(ePHNsTPG69edVpui;ab=`y2v0{_bTNX
z*3gE3vpwT2g;&=*aW*oZfz2Xn)FEmF9$-V!@f7Rr#!I|M3*^|?s=t(8w0l}TNNzfu
zjVExL)*s@4naNgA7LhThgadC~j~(ntzni3T@dyvNMLI_Ip);Y8j?25c4<7CK5VDx>
zW1U3OoqA)R=_JBk{DRKilI<8+W<m~@X!&wc9-AF>G~Nn~(-#H#sI8hJB8e5^2QFHb
zo5iTEkriaJW72Tlq?96qI6N>g;NrboDvmL&5cs6(UEz`%^Eby1=S9R;<2tH-sXeu`
zKpF0EZme*C<%kiD9G#<fV+cu|&o-k$oEb{X;nLvTf?DLZqawC~r=;WAYWzw!@0V`J
zhQ!%-?vS2;ZajktV*qiV*!^hY2qL-_xyvi7vqTl`q<N8%lvc>-H}A*gd&a^k4V5(V
zy)-H-t4rz$T}Ho`__$YHWsH3<4XA_Y!@rl9%8OiG<g@X^DwdASO}X!xuwY%N*EQpw
zBIx1g(O%`fUDOO%r$)OPL{$k~{9c+}-FlDx`WcxVqN=Z#OW!jVPvbB%T4gRf0{tt2
z<OaUpyaV88UaV{1OX7?%92OmQ$BDzADiYT`1%GsnS^r*|hK4@RIXuQs9QE9!qZvP$
zy6XjmyLAa8gqbz59Y5e9R`(2@AZulj$+jC9AFN0(h@qa>R+(isJ$u!jRrR9>db5rf
z3?~ZeYAzie|AX#bcKdsY^=2N<B*5Lr_f4c&9%SVE#vHtKs1u#h?<Zz*zETUnvG_r+
z6p1I6@v5kyB>VkDMdo>iRqp$VQflD2gy(h{|9%oBz4~2ZE1Lg)5~XF_FG#_gXF6Zu
z>{a}}Ng*AE&7K1n4kL1<fQ_P0uZr+3=FNksUI4r-mGAq^W4qF7G{fm0&8q{`3saWC
zy7yrUGn6&2q_$E_#QytnF6$fLH^Dl1K$`xw2bU&ZmeqmL-gZ=3gE&q`W9GY{8+1`0
zx6HuN6IF<921U7_))-9v_<;?-&O2LC-@F6KAku}I1q5N+@qjXfostb>CGQ+%WFwK5
z&Jbm9Bk_{sJ_k7&+erLjyT$U0{YGZPXvS+M2%6nU{5);W5_1mEkf7}N#wA|b3qi?y
z1@3Eo&s~TH$afRz;)Ze5w2qU3kX>MK3*wm<??&><8zzVQQCKk*R05J(x7kajYG7C4
zS`MXa6w$c>b3Cq^H_~*$)XB?evn*P3z)-7*xUYMksA4DIMv!btXR2mkxVqr0?uLm$
z=|80%&8eRP)$mCtDl=5xmadmQ_X+8FzovDIvl(c!C&$2{A8GF7@GQ9H98hL*1a%O*
zcFVSdV1a*4<ekLV({r;-&+izeEMDTAj7BD|axmO^Cr!<}cS^=f0(%#Vlk7Xg!Nr1<
z^Y1AY_3?L%?aU5t>X3gY$;~oPap`BlVPYajqWF%<+r|@U5xgQV#dJtAMja@pKwb*G
z<7@9;sRV+(2bmSCUVjmb?^@(Ax15%GBoAw-x;2M}M(l{JdrNc#X<+a*9Qwa~NSclt
z0>x_33JAPtW{AF%1Xia|jmgvrg4~2Gbcg+7@<Hw6cpYLE+Kkp%JPCQ&k6M!l_RChJ
zc$+1IC_6pFNk_IF%sgf<av?Q8b@UWw3vlj&(e5V5IC{aD!L-YR#S3ZNK}V*@%mrgG
zy;TRi*$auI7oLz@Bo0o<cqDGY*R8|m_DKpClDrgjw^Q^S-cieBKwhSwrDzvL7}8gc
z+~e3&SG*b)C{hi87Q~|#?fxGhk$U9e%&;YE-meAo6dv#vxDsnbN&_}Ez1%_lZe2(M
zcf|D(Z7*tfWbAJ*(UDQ7SMuEpX(H2c3~zD0c_E!Q&6A&cZ84v2n+}Snk7^6s4;=JC
z|5gCz0h2xG7<gx3ke%bU9m)N?0}%Ov&$Ecx$PW@@Esh4u+~^OCn-?B^=PnuhLE;w=
zeKMC4Je2#vKcS7|cf=s~gVgmY%zkvij30VW=p-ijAJ{<JF)0Oagkw6E$bo{64?PS{
z{va_On(~-Gk~n55OWEsUI)gGO>gqGM8C+;pqwt9t3}kCRNb-n<-B~WRevoz$b|J@M
z9mSplB$Ox5!2{qCwlNwHv*E<rcop@67c{=L{O6e!qJsJ7oIdvWfr&d7)A~faJFNN^
z>!x@8zy4h|gDN;O&`aTd`~Vf%i_2DYfk&g9IUUMbHC!Enf13@?eIh?hV{0D8%*YQD
zudcX4MkDKhUN8SJb(&41y0Zal=7+|M8?Y$<VG@Di;v=~Lb=k*dJV3EBO5Rhmb4N7*
zQ{jguNUM6{^bDed=m~E!DovnfEk$Z4t4R5Wi6yA&b$#^K_7O;9N~)2m>*^7y{m`WG
zI`uSI|6vjflj1dp%4^Iy92>6>3#x~OhJUlin~|nNM*)e!U@IW(Q_mdjNa6#ItxqI(
zv*(kpQ|LoiNzc2fI?c_0!cX;2eQ#<;m04Rswu3BxUPo`>wxj#AJfdL4595Aab=5~?
zq-*`t=QuLjO@!HtP6SK2ZrTD2g<Ee3Gf|!A)ROh;xOPO3*G=S^&|B{DC^*5@c|KCp
z^t9mBPOGJOv1@{`XTMB30G{%gZ{m_y8HQjfcasdMf{VB2OhJ`y8u1!-j9qlKd*Od|
zM_HFJK5YSkYtc7bh^6ir{O1R?$7Id(_7bVj8eWgGbc91#qnqZMDAMdEVgJvMNGm$N
zQ$3Niqq@!~oJ^uRqsgAEOV?XpS|vnGFKVAw@06R~Yxc8swMg!oj&@niosV6Lb-m@y
zySvMQ1KW58-^bftdk^+#Z2RR$X=JGZn%WqT{K(j?bB~PvD0PeV9Q_`<*0CR@Ze>f7
z`H=|<?sxQ=bM{Ax)^5fkxgXisn_@(b|0oI0?2LK_&wEW4-4ALFDMar4hWc49UTWkD
znf#Fn4SQaWpl75Mt@tgNlpW!XCjndeQ4)dS!!>M4s}4+Elb~eq+Fb1o9b(l3oKsNN
zym+Oo_kHWb4!TpthGPQKC_LKIbgWr7m`PbXT7h-x<I;}AK@{;$h=V}vI=1LvvFY)j
zr<pc4qxkn*(*Kdk`+<Y971-faNy$G5x-H@gbvv@j+jt5^(T@{5x&E=+q>&$|AqSUQ
zG8)O9_~ezb9~+0%U1{dWiPxOyk1>FjCF{9dJxK1y#&+}(OgrNN1S&D%dLCrKHnH$y
zV>Az|5Isbw_+t}_;Uf!{%1JLhwLrLgDMi(fd|1j3)<OwGl1j94Ld`|0KQ^LNm~ek=
z5<0DOP2IoG>R4ZMbXQ^XdenuBXb`;A#dYOcWa(QiX+^U42)S)9X`xp-KX&Q9(`{Wn
zMK5n+<KOeP76>4?IqZ;~lKYpyo#)1M%k$<~8Br8CDS$9_%p6>7#}68$TNMRAv7S-l
zUu5JbiIwi|6%DQDnyzl_C&mwTPs*|KCuzRsqFk~+NyE_eFI+tT#Fs$AgN*+qjSr$+
zq7GA$%6s<AJa=CXNDY!uz!d|KK2C$Gla9b{4TV_xiOH4P(&~$ra+KZY<`@-6kgn|f
z?^5-`B#x_MvgR$S7nJ%>lGG`(+#+LH=dG}<LgyY})k+-cX+&2wgj3UtBCA*t{KVt~
zuj*|Fe*^obP88p%8lS*v7aWY1ncEhY)&6xeO7EJOkp53%%7Rj?ozV#mLn{N%Mnmjj
z;kScd^u(_4<)>-r;mjS?&rcJV^wr_$PZK9>OF3(+X;hfjcPW{l8UlLt&cXB*$^O&_
zE6RqP!$MW6yU+2TCQ&(}FDv;a|I;)}-+Y-Aeri&M@fU7wy&}6my&RwP78vk=VKpdu
zNwe)LJ2FYNRQaiq_sVOe8dw(f#9eb3@b+WscS-8eO^xNV7P-sl7@1IinuN3WIFsgi
zX#_M}f5Fj6!WDG4QJn4MUCdwY$UkU00Y{jrgqby{u!GQYT!2Z>aXy0jHsGLM)(7By
z&pLpO&#+@x$-sdf&_^QM0Zbi9=rEC=CCrcE<dGjner6Qe#cMkMI(LnX{me*xK&`$q
z{_DTuz3*ox2uanS3rHuY*ie*Z{AVT^SWlz*mV9JizsLy{8|I%S_HuMK8?55bOcIuu
z7AL($b$}ox|9NhfAn)Gz4s0X+M_@}eAWx&EMoelyOJc(HN!_u4k@GycT8pyS@o9XM
z^`E8TpvO0^HHUC9wQEMr6P)u~UeKO^s7I}~W4H^ok#w9A2o*Omu*|dO)VQ<f@Q&*^
zOr_kdOy2<>N?Nq7Kn=$!`dobw-IzEm+dqqewRqCP$?&lC=ZSX&%T#NTk@#T;a|doc
zMt^S1JTfa|KTo6fz{ZnYXGZdvzNXIp+!<^!Y=55SE_KdLf0$n#%n4!{{XCIt#aLEw
zI5X-d4*4nu0?Wf-wm(lhX*-9%W#h9HK=znlqH?75bvS3spQkBhnvqxyRA~bzN0G_>
zveYBBfP0m-D7t)d<4IZfUX?dqWrmG{x?OJmJW<oU;At3UX$2N;jokLM&ig1fXQUIr
zc|ye@#c6Woe@gG?CbrW|%$re$O($m*&p9|dD_cQTCby+U_5M5vl=g8t*>+@4XtWvm
zg|UOXJWREIk%YPTyo~<B$geg<whbBkg;7eU9Ok`@rx5j3UgKvSh;D!-_lwluOIzu2
zhoy6*dG5<c8Gkgd7wv^#q%m3^UTYM8k!0?u+COsO{Dt)%S0YmSMUpXlWL)+fy{xJP
zX2(#VRlQ3aPfu}nt3{<*eSTSw7IE0mB@<o*9L$lk#BCgQm!$)vuf{J-xUarOn*LQ@
zm6mrJMHw%%qC<LZFpr9dEo-39Kz;~)f7hv?(*t`d-k<*>Nx$~aHUXPagbNBZT>a=w
zfz@}*J98a8;q2+nYhVc3_Kpn?{(^Gpf8>OwI!=`AJ+sEWX>R89F27Au&GV=i!TZ&5
z_OQJ)()6;lBAH$yJZC9v)g{|AAyLZ+$@Pp08?O?+xR+SAY&U6Kmc%cM)3`}PD_ZCo
z<xcT(WC%n98%{=x$51M>7wV;9$dgYg_l&t3K~$|1Dh_mLP5Cwo$Z9VQbr&h4UCnDj
z-aQt`!SP(wJCkUwNbnrIemEv5w9x!SUp1%_E@zsK@MXNDjLzuHt3kWky~Hs!4q^Lz
zLOOxDhMDYo>WE)$_X15;K=q5RI+@M*QTJ=ZGhsgfS2R|no+7s#lyL~P?XWla%h-18
z2?{wSa?zNfk3L2&I*-&cVAOv;g^-P1H2R-qHNBW<%-yzRFQx&s_aq$7c?~WHW!yi{
z;NUucF^y$?q=_Xx6HOs1va;eC_hOO(>kb7bz5ahM?vYY7Nt{?IUrf}qSHf>Ht}0%3
z@f2oy7n2aOb<@fBp8+4yIIs1KX?i-2w*6u<GCwrKLOe9=UH2mD-(f}?QJx8_8C@B^
zLWLDAZ+$0j*W1ynf<AZYc#h6z-RQ%CspHZMED80%o3+h<s)xNtKd@rrohC{);)an)
zfG9HK4x&m?{Wfkpmj1*}erZG~dQ?SBzkZpzd<d@o(io`bGmPM|UmCyBDWNa=rHR2w
zHT(Ky8oRpd;;4}Ga#O^nk#Psl5YfXe!_^^jbAKx*x56)zNZ3KS=$UGg12HE(Mb!zm
zG9_>D;B_}h`Im{SOaiOyx-l>DT&$U->YrCLbJn796bYE-)V*)&kCq|&OB2)5wXz=Y
zbQxEZ0%<sq6ZD|@%cM5nIfuPY+yS*57=W2D+FpJNC0NIQ=Nd|teF1ubrPkc<_M%%A
zB5Z92xqA;u-|@vNlk1{cUX(Z?@!i7Y>)#jGk>PCzQJ7Jm0wVAF^!NtHm3PyC(=~te
z-6Y@W)ff$9k*lxAGVdBUSu%-z^==wcXyNAd%$9rCI5UUJPy06>RhQNMs#h%>=B}p@
zjT~A$!-7_HP($-ywF{bbQ1DR=T&47`(Y-h!<;Ykj%BNHuD!UQX?E!TkyJt?HQMC66
z1gO88BtV}`ah78teb?mn@SOV0Uw4umM)n(C6Ztj`a?^{=t!zAr9>Eh}R=P&o?<O%u
zs#Jg9aU9IDN)5NV@207#!qJ6;FRk{1E<3po+H`nq)}U|S6H_=3;vD*}EgzV7a6_0b
z0|!njoz&HGfCYTp!KrBux;gokag&jvW(Fg_vJN5_x|sed4IDN&i!%1BL=TPpsm~N7
z<CW=~s?f5(N}Xd8Q4b%vUzyOXu)iOVl30RD-kWh-*+OLPhY7`B4d;70J{dVg@{`i9
zoToQR`BzEicAP|ACl#-ASRJrb1L@vl4E~z8G)EvjW?IkNaTN0srl_?5!!111&HAsB
zfO4EdE6kI<=#g6{N;9Cd^?6%b4*i6QZ#%GE=D}shv7IPL*Qqddd{KJ-iK0_BJ^vJ8
zJpISgC$h2?Aa*mf2GJPdv0stxKzCw(4s_+$#w(UDwoN0yHsb91sEq#Fh|Y0F#(td`
zQaQP*)+w1^C&Hne;Lx2Nf|%iEMRHLZrNW!gZrmYZwIKPZcpWx*T%LH=F*J2Whr!{Q
z(Led?L_&-+?3Sw(HL+s^>+-MD?6I3IdnIy@ptM6J`|BifvqVpkn!{!rsT#dsn`Cfd
z0_*(SuGrQC#`U<moRLQ0pj*~ede-zdS3GKT%Yh=ay7lpn5N1a@{<*k%TDk!Oe~K1E
zu@`6;*;Z@@sU~<i$}sMGwjFKTvK84x(o-k@gZSYPZzs6r-VSgd#k5J@GnO4w=SuG-
z0<c81W%NB`>&-mWdoNKIr#S4xnOomW%p`~q?<KQQVg)CnYU7i8&qp(}9nVSvGKvn4
zq3<Qxa9`9?crR@}A)Ch@LDTo1(VPu88=&`WYjWuk_|7xe@n#`fg@4RDSb%vQtROhG
z__w{&uLc4v6e!%^OM<`-OzK{6;~99e7MSSinHOZL!(9)2O+WOCG~P?oksSy1imd5C
zZ2!@U^7hOe>pmx!KE%Rsq%55vAN3LOZd73G=}xWpUYd$E=9c^Y-i+Qd?;O)-*8S)e
z7v5}do#Y7DL7XW^6RZrpo86AGd-b`I{3i7P<JOjGu||Gl{65SL%Wo2S4lu~Zeq(%4
zZL4HF2l+tBA=xOWwhp=9Bq_lwEqnPhGX9&P45$_&|C_{P;;C?7Z6yi;fgrn53<w<g
zX4Kes(%YV5(v;F~lFTjWzCg;JcVdMxQ1Nf`ZBvbMOT=_rl-h5Sv>j!83EApiK<61<
z#bwP4AlSO)1=V%O8$w>UM;ZZc8HX6y2YIGn)wcp;;^)}mrTrTlPaI676PXo#i3(f0
z-dzvCAU%hygAE3}BAY>sd1m1Lka+4TTDBsKy|fG*uxbs^$h@gODk5KFUdU!*E)mb>
zP9A5!F|uiBZM=lWely8ddRB`y=VluE>S0VWURrgt*-hiVx$ewGO-xLNcbVgx#$#yp
z@efIUGm+Y^_;54gwUIcEHIb=Nq;-k)Ru-nFrL<|XSH|(3|9C(TsR8;FyIwCbu}f;v
zI}QZQ2rKB3qU6R`Wo<Lbi7usZGs?4TSvI?|hBw08B~7oo%A$m3*MHxKXNz`Fgcq`$
zUN%j*Z7KF6bT<kR+m&mxX!U@(qTh_PPHC_+dapjOk-{w}jb3WWz`xD!?RFHj$)+vx
zTca@BCrXa|HVsX@{K)8U6TK(wrd7s%YiwR+0nS9;JiaxNn(S{AkC*j4miw)deBVwP
z_XcjlAb>aUzfH2^E0}t0iZ(e(OYyfR5WH#O+AKi!o<M+0QBbcI%D*)}xK$}y@`~ej
zf<>q51-U((!6i)13z~{ir<J<nGo??0W$jWE4Vqa0ZJK2Cow&LIYB)ajNVge%o|%RD
zY=FI!SKln{sCag0gxvAG;tuIX8#k%mujd&PyJgccVM)VixgWjboFiL~G;*(V{FDp=
zhCVKpZEwzuO<i8(cZvLa6|<4w8FkcQoSjfce`jP>_s%l*J7*1!w*BkM|B$SI%@Q*r
z0I7318UJ092=!W@_hKLC!Aktj{!5C#OOnD8uD)j{J$o2lsTA4EC+*Btc66?v;P$lg
zyF_!`bK31(^>=B|b2U(lY?XA~n%95lbKr6A%>vnJJp9OdRNQey8qp}rhnmGSJy-8S
z(AY=rvcBuY>LfrBE&{E6fZDiKy51W-V^Ff^L|`+u>AxeZt4zOruYz?1TBa>$01QMi
ztz$6Y?Vgft?*Xbjk>C6B$9YIbexLfyv#n9jxKY(>W518Zt;&JLowR3upC+QI)B1)j
z`}^VGWbpkyjWX5Jo8n+RN^9Io*2@5i7Kp1mh7_8j-y5@cF`n587l)|Lqhp)&9Q;=6
z$*T1G#IhY+L#Y#j^01~qnyUOhsd{uS=5EHqE?0kVa(9F~n5fVCflh#@E8Z46&J(g0
zs2*h%pOW?8C(+)AC6+Y2=%TubYC6SG(=jMoPJn41O}2vo>IcS@Bc1rkOw805Np}cI
zj|e^Q0Q|zgu53D}?3LF_KU&B7@D<tePAsC}3vTT>CEE@PM|P<A{*V}h6#?79kv|w?
z4(Sc(AB@9n%h<n+{UHq+t}xYca3+2*f5mf>{e$ttVlu$hKO{OFTLk-YhXuwM^X9#c
zGK*aypqTG-Ox}l$+DnkhKP1-e9Qr3O{lO#<10)RI|6qnSy!Y2I%<(Lx5<Maa6`Ibf
z<3-a4P+-*@FTJfXR(f)R)lidbQFlLb%X%4g^)+C_bIp;6DhZm7JO+hqXw-#mE2`_&
z6jpuSQCxpX$NPl@u|?|!an54eB)w>bR%O$>c*81oNPREJmdQ@aTR|;7M}yHI8f2<M
zFt!75;qc+2$RE>uxV$<mBY#W-jbVwV0=A6)(YUP<T*sp9ZnjOCNKGBZ0g$I`ReD`=
zk*BWg$0K!&r9b~i8#Yr@Quw1u&mo&ai=N;CD?6&mKPIjoJ%jp$RkakDX@EP`T1(k6
z(@s|YXd?P=!#}A;i+fpvYkv&EEE~V;@dG<r)$gtS(PV~YxgqNg3t`whqiW8E=W*S+
zAk9A}k)6fggR~rj6U<g^Z;^{<ZuO+|PYtN5S-Me?V<@-U4im#T+~Y@kWp>Q?kcpxP
zR6DyBUD?AlK5#05ToFgykyWIg{K@)oNLAz`e@Zj&M^O%WhU)HQ%)e*!1}5_-BihCL
zdG4xFV6v$k;+OM0+F|-<{T=^P2!@`I^tDj_Pl>9oOqar+jIr5IN-=7)8wA?qpOTQN
ztx@SuY4+boq%huzl~Dthz4B*1Bb9*5CDlrqy=>H8L40a|N-SVI-<JBHOguN<K-1v(
z>wTfDN1ffc8fthB(YcqU897I=({4GTO#2{h2d&p1s-){gFE~I!;O?K&D&X#tnWdTa
zdX9K;zieK@>;YKbkA`+KjJ6zwiw`sB4*YxdBnWIrZ}HoqH?Q)3nr(@L%fdGD{-yH=
zJMxN#z8-x)(Et)*!5PV&8tXjseqs#X)7TbD_I+QISI)xN_l-&Cu9ER+=;#bbP|06P
zFRo6rc)p)lP6$@kT<$1}@0(b$nI*EsWVDK%?2MG6I$BLu+p=ftfkPd#RUC&^jW}0z
zFz!CP7GR2MyI+4liOPPIRBP`WcPqR}UVlGL3Ff^XPBy$KsObi$`F?PRU5)BpR2R{T
zPC$y2XViYS9S!c+o6C+PW<fQ&K{+)B<h^9ie@GWR%)FO+xwypq-VgY8siV5pEysD|
zO{~QNo-RH_sN43uRm{1utp6XlB0X?GM*68|99b|yAf83vx_#j`7S(>DhX%^a^g}qX
zfj}$hxw<Q%S}w{mXl2|}MVrVaAC2f@t(HoKeiEXcy1^?(5)*c~+c%jh>fkR0+}#{`
z%Kao4u55=Bm42FT7O+t9nwV&CRtccWC8>M4otRv+Eefa=JSVJsic2Cr;Whe6geQ1U
zf>xsG)Vlg|X+=lOj)so3omx27luooqyxuwKIx5WEJa_PKbg)F)il+CmkoBWO`sj=2
zoJ_j$liFz8ySzcP%X>&(YQs#2$T5&n`{%@%y{5Sz`Lj<;Pr%H`QXLtq=${k)Y~!YD
zY{pAp@re&fHcC(7Lm4+he>O3>k#naPItKpvKPP%)6;k-KNeFw9RmN5^pu=aLltG`#
zD1G@SrS#_{Z1WnGo13)qpOYN0{I#X>XJb4o!o5=cb7F&<dh-6I=GdVNVsucT9-xn4
zkm88o?MUs?)*a528b3Dz4u%|hU(pOK)wDwu57P45tT;zmAlrc^cH*pDJVXi@SkbFy
zsq6Gu!kui-+rm}=ZGFHwplYE=>E+Lsr#NHZ%6d0&<ay%;Nw*z@MQo4{hVv!hMm|X6
z?-1i}^n)}R6Ho^aR2nT}?1RLN!v`hf#oZb~vLBcf%n@}f7o~73KmI`?IbI4ezWpGP
zSs(xE(DZsob;^yR#Ox!A{RbuvH(x6yFT+a+DF=?z95*VS($$gI8&z-DpS;$*_DWUi
zQDrYwTKgbL?*wLHoKPJ06cM8uA0&C<T*bn_BF!j#9UAi~X?Y6=iMWaC)=7so3u@b8
zYbfdc>)d}xFS?2)p;p)oME0O|?R&u;H_2A~UMiJj@Ie|gyg~Hni`zjMy^a?7i<M9{
z#wv<BU71Jz60AdXNB@%gxWn8ajQu6?V6VQ4$o$0!`|npu)~j(epF$m$`-{nf8qmo2
zUlN~G6^Ony@0m+;T+SB$lEh%+Nh$ut7{B9iK_(qB<QfMtL`^(7E2TgMWeB_1av-V>
zfGS>h<H@pAqab$|?A~gDv&N5-x+mQVo=4b`i#J>rOh?W<sy7_wA>5xd9Zx-Fpuvj<
zSbLRyUS~?39d-gMOs1skz|{ph)RkUTWL{S{Y?0o@Ne4Hxfy?FDEB;NkBDpXHY6C~-
zEVJQuv~oG_KIE@y5oh*7tL+9WBY#bF=emoqlrs8PUmxFMef(=;D}lq<>Sdy=Nhf-|
zza}X-%<eH4*%NSst<?Bm(`K~yvU=)c_Tq`X1C9B);4s(~s>h(>UrloCqJZ%*84X}E
zDy6^&SL0Bz9O*qM)3Ho&)X#nB<5CS=R2#W6E49C-*+nf`)Q}p7rBQzYn^ty2-WoG7
z2F0@Os32T~h@6pzCp~m6Tiw6<JjU)oTK|M)Roc-K^M^e^!|i4KrfMM%mDddv3E`sN
zRC<oS25;K*#&)gZ;)R`7-*aYo{kb*l<GEX8;3$~S$L#=SHlMGH{LS!F4~;VNH)H9{
zrx>Y!Gy1yLhTW|v>EnrwS7Gqqk}#;#I7Bh~H)Aj`-(*tHMQId8C>8#e!~=CLWXMNt
z4lNr0z3?|1_Y*fs(J|0rF!{GM-Rz<oa|(Qoil(c_mgsM3Vl<7WvNoEAeZ(D&<n^mW
zh^Pesp17;2Up;=9;E;qMt^Lhr5VcRl`p@zDa8?>lC7ckjZbluIP2Jx*(k%Vb_B<6G
zPRBZchZXg()(tvLFRBxmp66kv#uV)B;-CYp`a#WCF5V_v!_;d~7mvc8f)hu$|J)7&
z9K?=O9fQi>ouBFnedO<nJZyZKF*t_)Jq<)m@ntOX2_(WrE60|<rwO_Db&gNjzb76n
zPFsf<yni>=Us`&{9puIv7~&j}d;q|n47<Nb>N!%1j@+UiKqmj5q~Zv6;P}M&d-9P5
z^&kn$oT_@t0f7*4Ow3X1Mg|DBy572u7B<PXC~RSe>rr?=J4iIUe>eF)vP^uazo#h&
zg%;XReU2*)@0+?^KUvnNz|DZSN1bl99DRt1so1b9r|^r~@o&zR#F+Iq=9jq<jpp}p
z?%-{b<GLFbqZaG`-6ZWgJ0fj)0<K;^Ap-~TU$<msI~bt$#ddEf|1buBl0O;wN16rI
zN>WDuk%lE9*G@p=v413Xt7v2*`?S94XIlP8Vu5bsIhX#!=%Zdo$NynN^6Y8J|07ZA
zW0y2&1xEtc%uL1pjW5&_GWm}rFrj@bMXS2Ahp7KAU+)&w_jTp@uEf~JHjZ(k*u)_*
zma&a&#o<8@RHAf1AQ{<6q6N0;37HXBfrQ{cTqLB(r|52V4>PTEMtzPd#dPb`kRDA9
z=~AlbL`gGg;#LiciwD&yT%`vd^gwl~YDhv7k|+sDNOC@FfB$<aXC991hyRbg_u6Y;
z*4}Hc>wxMzH*1AJ(4T>m|5L>0ma{e$pbn*oSPWn~Kw*I3*XyNd{su;lGs1cIt2V27
zVLD0jgE6Pz@_FWr&8vGz8Ii7jBcB?!8!^t66sPBXH-=kQ2Z*TMF>gg*u&yLAvDN4V
zJEoJiCRjW#c)DmJPa!(k)Evuvi2bqkKQm488cRw>9J(KtC*^VXR4vt~@`9lHTu%$}
z!V6Zho?F5eB}=EO^}ODQC*AC<W<9SrPIy+ko|f|+kGr^@Z#6VG!?J^}S8+dlS@U|{
z2p*FkI|jiaoS4&C#~&c*LWcFUO3dcXi#8BtTtIl=3)`j!n_SO3L1C;p|E8js16@PN
zNT$`xzKP^RaVbiBQ&hGYq4(;iK4r7fNgEM%Y%V_GPN2#d^c!sX1!2(3(JMSh#zxRJ
zEC=+n#KPk2RiP{uS5-JZf?Uvb-fQcL#JXIbzb-T|8Me*nH@%&=^yzU`{GhV$Y4gG~
z@M6(Ni=QM`{=GaEB1=@m6)MN%?W<U(d@pq;zX<)tOXtt5R+X6OfAb})D;E4IJaD~!
zA2`^u-<rRdcMr_SqP2c6uOt==SP^!pcx(cXYi_h-YKa!QU<2Pv`{@M3RD5PDd{0CV
zC8}BSw_^lT0hM<*`+NF(d7mAHpSvVHxq?~L8C6XV0<IaIjjAKPSRw?r$wp>gC@?}W
za*3ubM1?cXq~3@rXut+n$UCcTD{ARiB9AP$249U?hHX*^x3xeU2^>%0^}wag8DM>l
zBSGP8--@0ZktBRSZ}9`97yf=)9_+$li@u+y7s5cNQ}O*g`zRMMnUBwr0@A#i_=N8y
zdWoPia`#pGex~N4(8}t+pC=V_7i(5eoI=>Y!fv;IKWH6Z5ojIXPt-le!FNo62a@b^
zAW(dB*!N{2z<&DWmu)h>PQ$2AeLvs1GT18yXPFM%+OSvebHm)|-A(*Gyp$&8yI)!Q
zn$3PcZJQA+ilUm|49*(G47B%g$1kW=;#F@5PnhU$;#`3~H1I~YfQV})n(}&r)BS!j
z`F;6+`Dbe}pdR#8>w)<p8BW-yx+LZ!M%Gbxi70^fgFN5kXW*57kfyWiI3@~FFx+*{
zDt?fUP7*MVa+>hYtolK|6BFYBeymu{57Iv2#lcelAg`921u+qI_hGuQAz&hCOK%`p
zB_YrTpVIn+G@-9{TE`FaMkK!YF~MNQJe~Z&5AxO{a$27iq625cCV!CkX?NdnAFmq0
z8A{VI9mu&^BjkNapbo;$x|zp}29ktuvjGULDo>xe=%#^A=YNnlF=OmY8t*7?KZD)1
z#^er*%Rk76U{Mi>G%!Z6-*4$vH6<ihI)jA-4+!i|0%#LkZ&R~V=(jEXdb`bRBSVLC
z>`HfG=|-L(1hG)=MqX`#yjaCXn%AYT_24mAwV~slv(0KY^1?NyAzAH4T8jhiRu=#d
z4zo+@H`0a~#BQZIV4s1(Xx&J2e`2f{Iio`rc^ulYKr>}Taa)jq=+KQ<7;GWBIX+0j
zpGjdlo>bFQQO20yhc@!o9Gk$p4t|2eSiO0=wwdVd9O2WjH^lNpxWSEJM+P@Cr_mie
zWb+$&DN)q-x@u}6{=zF8we2apA=s0M(v~%6OJ8yAi-7EV(^gesH+S}0%*knpd3__x
z6kG-!AvT2#$A4SFBUEmwji+wouCgDdWx*m91<Vig>Um_c@*n2iJvi)!0u?{ZOI>j8
zt160@{@yc{)I>26Qs8n0t62^db^0O#wLc*0qiXV+XbwpIU8fNewF)I<0EirZn0E>r
z#hn+9{V;6^UaL0n!@Mx<?#s*L6$1TX+DJ`JHmP5q{j5#>FmGElWQxc){liT0M@jxx
zM9yC)Xt@I_rtp}>I+@I>Qxu~%7sz53h2hG4d=Cve`@awfAMG1+zsnmjSOg0^Pc=BS
z@i=wIe3SCZRs+(Zl)!sUou1(tw632gIn*{|Fv&>KRk~Y0%$us;?*nXX?MHcd{aE<@
zD6i_$S7DQ)=*7=jg+6#Dvumq<l-9Su7{9&dM|pOO_FhbWZGb{lZFN6NTc^Pzmevar
z_B23h79@5YiI1>iP^%G)gxntGN4f&w9ghhBZgucU`aR((LkiLUwDZZ6YT@b3rUVM>
zDXh(H`bTN8JI~uC{oO4lT_FFXv}qbcJ}_Gx=a|pnoLYT;m(2@s8+O(UKgydu*@D^)
z6*ZuiUXDSND8^O-H?DCc3h_yxp*~xSy3J7XdH{v{1K-X2rLYWJ!jf6)!nSgbRr_%!
zutPp=r9Vz6x%L6>gkTsy&fA!?ou?#hV8xH~I<q)cKTZ=nc9O7QYU-d<MAlkC_6Lo}
z@T~iB+7ZR)OufFSJVwa5<{#(HghleW$8&BKT<2>#0xq=4ll+*9-g&3m27avLeH2|-
zK@cTcv&kQ4_J_le(5x(yJ6cZ(CqmXHW0Prt>SFg^ih;W|F`~afu%+va&Hgx_2#DnU
zi9h$_yit3x#qqRdP-QgnaHtFMb(#oBH`EifV#_fe*I5$hEnp8|c*Y-A;|~YNaLyt^
zU#t<mLzo&h371aTrWS*wD&)xCQo*?MW44(VW_0PQm2Re48g$R|^3A;E@P@<^q(WuL
zT?r<&nYID}=Scp>*&0=oxtG;$=B?0#R42_}-DX<Tqzm~=t5<v0w9SqGn$_M5kt)wy
zt1#O2Id~tnON;UyI~MI&`?EHnLckDD$qIo5y9yRzGAeWY;_<tg7lwZ3-B&5f4=pWX
zPPCaeD>_72gBd}l7k|jj24wa%V%53LMA%=%FLX0+M<1u_A{PQ`GY+YCLx6CMNdc;(
zu+h)n6~XmA=T9YD72L)VTMM{jr!ZB6lGlZv`;23>xtaGAVk?im#s$1pTt<TU+fVZR
z`x}op3j&3o<kdt0ju5o`Cuu#}P)x#gRp^U}#0JEwev&rCA>u36{3K7>Y5f0L>e`>A
zmG3L&%&1dUwx`uc4+OB|7H$5CZth-mLajeZ6V}>p9Y4tvhItHc7y@YuB7GBqjckq4
zBQiPP(gnf6G=@;P(FchGA&@DR!EJNe>NhHLWPCB6GnCB4*FHif&;BGIN(UNk?kAZg
z;V4aRx&B(3*Qh4216_!=IM>*XK#sG*omQ8l*gZ#AG|W$T+NyrN#*wS;&ZMN<>-sY?
zz0IHG#{<@0ex-4_+m?n$yc|M?yENga*~2ltFxZU>3|RWpw0KOJhj=NEuTg)g{4@{m
zERx)xrj;b0oYm;(Vf4(kKh0|ZPnTqdbphg_yMq2S?-k<jp@eMyX<AA2+9THb(|ie>
z@t6}Gf{KTZ9s6lo<B1bCAQaep<N+%DH19?7*(j3X<WKXaN-`|0HlurD`Pj6&NXDHm
zuMKRl`Q<%56P3F$2cC^mv;j=vTy%#|1o8IZj^?8~oS8<jmsyB09Cex94GoW|LOf9P
zo0#XUraC$^*z3{xSAUv!N>WB~!K4R@#^2=W9RibH)FC)KM7JZpz*s4$Zr3S1Lw}YQ
z30FrbyYy#yy2;GWfXjcDr^jOfICaJ!dhzn0MOCyoge?FPzn|qf_Mo4MO|1P{-Xg7}
zAg%jZUNz#Fe$(oImbS&<3D#Dau$3CmTI<j9rp($Fw8|aPSr^{ei0FTocEi$DOdbaU
zGj{~xJzDr#+63)y*<^Hi%Een#fuf|g97)sBHQMn@Lg~AI&5K1Hpm)TXFrbir<eN69
zBK!gfQKaU>oJz2@fe6Q)pR_mP58TlyHsjCoHIHI3rMTxB_@=Ey2WMPTwi;+2?JIZ|
z09*~@Pk18`te2%-|5+j+8G5)Sylo0GS1j9BfIO0PG50dH`D4-}xt8A3^p7EBxR;hG
z$*-<k#l19pXJMA^<vl?(7Y_HDdzq@S?yrqP?7b!kf*ap$J>;~tdVM=_C5=Y2`a22c
zt@rXyaqX1H7qE^1edK^06FzJQ*RKx5R|5=R;a=XCet{CiVp4F2y2!$EFK@8JNYZTj
zUfLtPc(_-ih~UE~Z6><o;V>%*TxvEaq?263HNW#QsN>0NcR?Uq`jEWb2pG8VEeBdI
z@`HX>qQj(_T@64z1VI|GH32I4n{i$JbDP<wU^cWLJaX8!V$CqL9`~B|^R#CBITuRx
z(~UUGf1cMii%lwio)(T7LOjC-QKj#6yJ0A1HEKLafT3Fb+~tz_;$MkoQvdTj%ZK`q
z!()Q-^R)CXUSj=J5uu)S{5)-))7<_%N{{_KFF?v<IPmkd0O-;G%nDIN&c$!qWZ>*U
z%$?t!iV7H4vUopFyYQXI?b6TxaJQs$WhT%|P7??dRQolao<fMBp1_Q3UKmU`#03qZ
zA=y#gP_?3_eL3dGBSqo29k`tII1j5DZ$sODTMJ~kSP&;55Lo|tW`i2Kky%ajGeesJ
z;58U0+X|!_u=lWEq@lGnSm`hF`FO~!=YNrB5A7PFgecsHdbLW04FuR9L5fi$7&V`<
z+F#`3a^=4$NFz@5bt*pGXZ62GTL#6IHS1>=I^NcRO@L$T&{s0f?XzR5HbPkTfxu{Z
z>A%3+{z8ZD9Cy^@FY>6-D<Hb^2gQ`iU&m71ru938!@VT<e$?UlG4qRjsvrh(D}`U=
zQ^gf3fmjRi{^x~CAL_Y<K&6FH33j81B%!L7)e|=)R2Q&`V+3cgsxm*mY^{ie2bj8U
z>w&+g67UGPP~R`|u1|KgNNEKw9S5%QwqNEIA~*wP)xS)bd*Bmf2EWwgd;@2vieF}m
z<o+PjHDX#-Wm~qNv6^3|4M4u4+F$0mkiAshFY|^N;JII~FAs0VOQKmI<Zj&>J-7sD
z)LoJSHx5)*@t0}c+j)Nt1W*rmRrqBlr(sVi>~>3&s@sShZ%PPpsc^+Gs%fFxi+yH5
zLG)}h(OY}-{=+sK!>fMP<}?$)IBD~0e|Wzw=vVU*yAhC1LifvpG(t8bx9eLGE>M82
z{xU5`KlAGwg_sEf408e13EaV>wyC~H60^D$cpdd@IkuG-qL>cJO1IK7B@syZR-Wx6
z=;121bZEMg!}H2kZKc&6>hsq~jXqdEl&`gdvGg@wjdfdjOVEg5(<<um&Dgw^mS6~F
zKx@=0_*U!)XfJ)*j;TpXc{i|?cE;{%D{N&>&rP2^j<8eM<W^oh&q>9;oVO{#>R(zM
zKn$xffaalOjCg4)?`T9uZUizDh^+iy9K*AM$Am^&)H$KE&n3e1F+nZtjRlS8r7wQg
zZUi364++69N7rr16naIq@jkOv{p=yx!di^{jS*Yl%4djPa?aIks{26~jc-Nw-N?r6
zS7|*E|1!6w`tX60)%aI=9m!%~6~9U=d?v+P{8d_iaq)mB%#a#F?!3@yf0fCvm6s91
zx?iQ~@13-IH3U9xl8J$Dj=#JPf915%s*X60Fsl7kp6JBV5Pe#p(-rJT$H1@BX804f
zZXqDKXQ^A8PO5O%uH80O#1H<FLsWn7I*BKz;I(JP@QjK`jcKzne5~tjE-*pX>r^wZ
zmVQL18CdvL-m&o4ye5B@H=6vZQY~8+8tBga4WhmpNNB#Qzq^kH*0jZ$c5(i?rtYmq
z+l)!|s15_RrCJ1BLYw*PynkI@%DIL-{nr^C?5YQ>{MTuox)Ets{5q}10bY>5PD^^K
zdjw&?uk)Ir&tk*WssYXsRu@G$JtFp~SJB8Bg1P-VZ64=+d13rI?HA4)mz8t`PRZ`d
zH{r2gr)~58WkezulX37sV2K?fDL4h$Ep8^GQyikXWv5g$gc0d<6ya`*YtJS1%cgph
z%YQ~+5V4D6XQQmZSw9zF!z}Q8bdD<pFUR``O8~+{L&NKGGfrF9Yz?{7=E|=#cknvm
z)wV?Cwi@ur3dO}3XksYFg}NS{w<6ux)GB<b+P2h*qkq`!H<{*+BE&8IO<qLL1#acP
z$<**>H&4R}tN2Y?z(1<l^;xTmx+F(JZ1<bIp00Jks!M>omj0Ww9Cg1*D<PsvBB*-(
z1uQTlK$H(}8mJt=G@|u4X{D3zzIEL1i``<!0x!NiiS58|(rzQdfE85c_WdG($=~F+
z8k$j$>CIKiBWUV3X;*%s+NRZ$2cg6MCwk(5!l+9FM%2&qu{ff0Y^_0?3vBmssLlT-
zAAt?{mcmEy)?HBB<jZp-YPWOiF2~n?)_FxUhYQfj<d?r%Y&V8I)mqeS!2f<dki#nW
ztZi!44Mz#B88gCJglE0kb{;kNP7)NSL!)#%Eg&pE?+yLNnE<=J?Yz=s6G<3cwVn2b
zE6z9}YV^fod8WhKV#A&j9JSlU+ZLg!8#L9cY|w=(&D%v{^~0xC6%z<d>#ReCd^dB-
zs@1qV(+q5<g~7YLu$^}!z5~Z?ayu<@7lO2@?aUA0#KdIt{*M93R>Lj{Bo08B76c&T
zGxpkSjOH+Yp>wK;d<ky_m0*B_?$Z_mrfW7~d2a-qu?Y<Hmo<dJWNfq&W$xf*tMS?6
zR-tRfX~N7I;Vc1aG{`2{p6G=5^KKipB^=y)&so*qrUgR-4WIhk;x0#?{@b)XFyi0A
zU!&r;`8eiPj3|kRc-3#y+%vsz@J`gn(zk|@E7t16^RN(;=C^qw-ABRKss6W_j@yxp
zu!!mjK-jZdt57(_vEK39ygDF<Bqtyh#YQ&p+q|dnzIK^aAwZ#Du*qnT7zzF-`aM43
zrmWLZxe4L?C6y-Q9FL4NqtYWvHZWTp(xAUs=7i!PT*`dFz;DPFqC=#wT$nLwMECS~
zw4AuA595fK$5sTjw=X5D3Q%o?09(_l;6CtIwytU(-U-{h7=^s)wuFqkz}fHed|}p$
zpze2hp3$`OGB1x}vIl(GDpZVP0Vfs<RrR~H4@eLAWvlsJz6@}?k(&S>4ZqWQ=}~8_
zPPNgGVpZ_Fe8TTzcA8ZJ4~b1)l)TNp>-b&q#^(JmScBv4RW<Otyo}_n_Z?CQOmXca
zh?}4!a{xqhg73GKO~(*DFdz43^tX@wGowz}SPc_48$}0Mnz<Mtb1~(zK&-Lmf0sA;
z`+u4C{h!%_TFJb?Zm7GY0QvKkf41d-d4RiYC9ocG$@Z#7JA#}4T9lAo0+TM`g<A}h
zvMCq@<#q#^t>2}i=`)`;`+ZstM2D>8@AJa8d-(P8-)E9aIzb>-{5~y>EMlvEpVu7&
z1&E?1TEEeSaYpU$Qx}Z~t?u`kp=-E*f?K^{DVBKi@AJ~YBjUDXtpSX!`RCRlVEZ^=
zqh=EcetcEq4_I#a`=g){?HfTQHTnCr72erZZBxI`vtE3u;F}vGfJZ=FdFl5$i@uTI
z%m~4Dw^Fd#XkAoG&D&hG#zdRjZ$1!lt0vkk2<csq+l`n>4_fPysFwwtJ@X1qNAc14
znym_nLBE{)Y%S)2#4yB!iarsF!<=raGGGess}dLxS!|pzA*48RAI_#WE3x+;lOe61
zR^7Z<;!i2umJ>voz4xeU!3uM8z&HuQz9Y87-g`u4`&{h;H*D{f@Q3aR@TiQ2$gz29
zLgx0~hg3Z=-9;nxYJ0Dk^`(EX-)aQ1FEl;71u_LfskQgq^w*Q*dJv~Q_Fk#rkrWLM
zqAt)Og_%ED_S$<7s={kHAja4Zcl9uloFq+wBCppTzdf7P1C!V9mQ*bPk%$!OqPO>+
z5JbNu(RfSf+#dD%(F6H;6MsltVV`xzAKdQK*>`*I0U#ZLzv<NweIM9Bj7copkjlRM
zXDG}I#i=-P^Dk^XW~~87#K}K^?&p}Dk8$9<bjqf*WS@TCE~r~qbWTTCGO@fEKwPy9
zyK_l(v5mLO>e_`+HWOX9V`gwgb3B+HVY2~}W1khdDhRDj2W&1{xjy%r7GrP>D^V0T
z9w~G_x+fHeo7-PkpXuUW5TGWWHH#X{mp)}T{$Tl#B)}NlR448b2%ESZlh86q2;f`s
ziH2msR)n|^y{!)yQ+3&DY@3sVb|>(D9cF7yIHBtswYz~ZUOk?Qaa~9*edD-o2n2zz
zU<Nj0=Mhj6gK;98-4h(Vp3rvbce5K?ZO4!t6ORC6A9&A6>abIO$p_MY1}4h5(hp?v
zJ>V)fH@+@YIlK2@o0Y3!*D%Sqc0_^au-Att`Wv^(4<s5&R<nLmR(&8-<LgjZ^#>Bc
zIXMim)-?fUY07qgARi25yde4_*&(8Me~0Z6V47l4$+MtN=sf3XjrK+-N8E@kYGMV#
z&D|6OJ3<t^bmcb(WD=FN6jA;T!lgwPvNfQMk>RRc{WOh2Fdh1R)&mZ77HeIe`j^W@
z&k2A>#;rF-2`i3sar93Fb=hPLec-(^k^TLHsSvG@Qq9|Ve5SZrlOIUrJ&dO8{0GvB
z+tOxJAIRsWr$y)Tyr7|wqzZ17)tC+-Z0A9{C@j%Du-z_cFg+uMvvyg~&+_M)z!qlW
z-S<fvV^>sjf)M{U8xV1Zh9Kh;slsWS)1;6=leBDhO;9ldv-udXn@#Bcy&i23bwN%G
zv24g>dDAYcSHwFQvPM;$b9?P(eBzqU(?b31$5s568jbo6S<&y5$?kRl?;jxxysZX4
z?u?Bs4JYWGfZF9&6>HHXEt}nqi78(0@7j8-I#xa#Q6J*<do#WZpwMUcgbo5!gnZEz
znKFiKJ9b<vZ!G&@zJ?Thu;hbzm3z7fI)_Ec2NO*dqokI7Fw<0fAG@vmgNeS0z%pt(
zqNuuN=kr$a!L)}4LAFx04z}$(Y*qTR=ToSTuKVTCq9QjT-K|cczZQ7<T2++S9w_Zm
z#n065w7RI+3a7sJgPE$4`YnBxGob#1>CoZIMcUc)!AuCfC>PvF7SnrGpr^wP))GTV
zuH3C3eD84)j-Std>mu4}zhCEXp`4CDpdC?8r!ZTZ<BE*V3Lb$TgRmj%Rc9y(I2#9w
zsQH$#q5IQ5IO@u}LNxcIRHioLnn%uStPww$=n~tO^B>GdX)y~f&NqQn1Jidw&Ctri
z_-jawBS_MYV`7x{;>B2qF_I(V7kPh(|2LF%IX+Xulo^e&_2?1168&`3&umQPm%e0I
z^)WQqxAdQFF6t3s$F2z<n0N$lf;}-tmO!TAx?uOAGG7R2EoW^}0MeV|UAv)LxO>^n
z7<@8vPLY42Gq@bPrFojbayd2!+^vg(>~{PGwg&S8E>EE|C`qo}i3+>~Q+mj?Sf`@9
z`d#62l84KB;NqM{I{G&>vo0@4aWFSyrrq6Y%<kzM_S`KJFWOe%P8=EA)<kjXylVD9
z-r6I0Z@N!q$peYNTyS6dK)&}pD{Ywy-fU^N@(1z`#t8ydf3g$U@j#~GL(GI#JdkJ@
zCau>hAIN*jIX{e*kPJVNHrme5TXppEM<k4@Q7?atWzudz_#d<XW7cXP$ma=$I`DFP
z9uNWg9dC8|z#0e2+pCYmN$p=BuwQ2<H3bkq9MSUxTCe3?FfHm6$3RR&Ro%@~SN;*2
zY=0o1Vb1Dc>C_SMSZH2$o$9UK<#AXvX(WD9_r@$A#4Xjs91LjquYT5s9>@;>jM9=<
zzH9ah(HD{l=RwBRfXJ+q0c>OxOC!Q&oL5g+bJ)}a`7Ck;J29I$*DeH_c)@VgOlyqz
zC!ey5QNK6E+HrpVQcT(r>`^YqvLfyvAd{w-gYCsmyHbQSh<J5Yy<pZQr`xM4OPTKG
z0!A+>7VMhvZEt9_`B(y;N{W7SJt`+N@&$Fs)`V?XjA==>Uww8X{@~9iNd0E~!KG3}
zoLLTV*cFW0t(c6%IGy7_A~gHm9PPHca`#1)m}*yi(%78<^ZEobkTop_p0;<R!$t)e
zT#p`b)#DD`hze{QUyn`w!4nEQLtC(W!eQxKXo0qb!^pXFs3^8$B^og+v64iYL>IG?
zl1vU`L!RWlR7HJAf69ur5ZjiQBobWu3hc&?l0-19TN}X^qPYjz8na4O?Bde5s*+5`
z9xL9eOY#}usZeT4;<&&?$v4YxmG=w{7F;h^k~cGtNB8&OygF`sq7@sGJ<h8%{zAQe
zyn<}6&>qGbsJ<lAS$AJzQ=6hTK~9oAO|wShal*%}B}U_-a6H6?5;RL@HES=9&7%f%
z6idECDEa?xoiVhduxiKDNfHIRKj6F8t8x%d!99fF85VFT@ZqXQV27Dv%r3(n7e2gH
zCX!TnQZ3)X9{IdL7GvQ}1p+4!LR<)_`zNs!V@{@HDQf;@YSu1B1<t^;iBe<a+;rUi
zm!pQ{5+q-Y8i!N9s4MzLb}zB?Y#_dW+^%Y>+@Z_nO7fANTxppEyQa3x4H@tdj?L>E
z&*$Um0=OQ-;TE!m5^)kR!6oTTa_zj|Seyt8b|dNzB67Z|8A20o%L2+-gSjea>Meb<
zZ#mGA>8`{#?)}K3ZU>%8WWVTk(6I>#DXPnfMY|Kbiu^lg5YMg!3XL$@cQtm9BbKI6
zf349rRA)lUuuV<Yc#1!FuZZ*C!wKu>tN&Z4Z3|4ldtuvyc^`W)c2FpJFipwlKX0WE
zW~#!x(aIjoBzFdlfR(F|d*`1pZQP+RH$7zyDxzvLni{KoFj2Kf6h2{9QTDp0Q>l*6
z6BFE+H4p0az^HcjgPDpw)-WN2t+v=DZyuzGdsOwMUAwIA!F<q=K7?dBdsWoXdeG`)
zc*q(2Tx)vpJ%5!aAxCo*j`gJsXf0|ucmkQ_gZYk15~f~PCbkE%`#jfR$NlEI6gnTw
z+fl*FV7re+9dsb*ysS6=(3EZ=8&ID;XM1cYh71RiJeLZ>mNaX<Hm>2nEnD45O)By*
zuI}@yMj$Bsa{@^e7z=hGK0UIzO~<erJjKyzLZI12VbFC72T1{fANy0gtc~zC2{2UQ
zWOuJ!32;cSxKCp=ZsN_-U)54@!YQ@JTnxz_K!;!&uEk=tB^&zrfan5ryDkif+KQ%2
zBZjQTvD05HE>7>D-3X{TeB#aG-uP!HUR&1W6eip(_f|j|_KZd<l4JhuK;La8Amr6p
z!Q_pBv5wtQw@X*iuLPuC&W^kB)kgr9^_bX{?g!2HS=$IS8?j;T;r+QOVCT^4+=~Y6
z3_mN`oLed$_Lr)rDKkiB4`pI(!I-b)p+r)H3F*>@GNbF9v}{7G?4eB8;+d@cp|tS{
zFn!8)sJR>WBKmzO(HxILtBg*XxtXifqI;0%LUr_!6q|ZG)u?4xcem}1?vkI!H?8)e
zbc~!lWqb7N9BMrQT>4gjzwOmu-`;Qa0XkcqJDCbYmi|lRBQ5sB3D^?-V1b8qkhMOP
zX}{MciPj$9?mPvNK9uFY%Amx3-_Om_`A|MxkkIf@L%Z})I^T*cf%QIgpZ>{8v@{3(
z4+yFLQ4jGlB&6E0zbdFtS0?aE`*R!DII*Q9K%u}gkbaj1osWL^4<ihOF_?-9jqbXC
zAqEI9NfvmUjx`y8)43RE<;w>JRzP4m3wBuzc;5K!JEM7i;j?x{zrOU2&Fa_r^L91B
zB~~@Pi05Jk_f7aoe=U{*N9XxKZ{<HHoAc`r<ugxt$k9A41jcO~>5KYeY8FtJ-O#L!
z5;f+g;Eqgq-iPIYJM0T`)^4fd2+ZP&5W@V_eG9*Cw*x)_ENxX44<6cYcVco+bPXqA
z-da>8%!u7pj~dpxJ_ZSXJ!Tt4AXpM^s!6ZHF5c5p5*`$;H*j>R25Att1;XFQh<hK-
z6EV`wCFo{bB@gFKmh44v4ljK;5g{(p@W2lzx*H{~LAn0C@{c@&cRZZ$*+20OQ}J-#
zhdCoQFV*;)ztjdv&%smm;Y<fE2{~)k52y2@F+H%X=3$)|#5fza-O<Bqt}3aGPmdUG
zd+vW?m|}G?ERS55EPD0|7!iVTd8`-eylRK7NxxAK;dG%&?E5#-#>9sQ85CxeG>jwl
z)fYMwS6X&N**;F#PR-Cc8XeP5!ogXuet!2WMA*@wx%U*`WJm~L2-9PQ0Of&r$5nz+
z7OLTg^U=6JCGfKIg2<+^DgAt_p~Wu5M6l}?ZCVhzooTvwzsRlcF5UmU!i{k`;QPsr
zCGSkE4SI1D09XD1HAF}1a@W5`a8$%{*u9+f=QeZ~*J2G^@Pc|0%N~{XW3qN#(1wxq
zazHE;SDHj-<F=@dJthy&3>aF$clnFmd^jD@N8YgIKp&~#DZ6za+cPdOSc%>}5r^j&
zv)k%@@aF$+s}H~Tq~wI`JMMV8)#Y#}Hore<U1_%#yWlgQvb)h~H-ceX7dD+Q*hXyb
z9vsyXqp+Db<8S*XJvX8Yk?#qIK3;}f8rGP{d)`(jykW_NX{CwWQZSv8(o9!Ky%8Ig
ziu%}FWu=*N83HmuD7U;cQ4djGUH-76G)-0e0jns@)aap<dXR~aSfwBh^G2>J&6iDM
z(T^rYOQ|l+w26X@w_i<ZJ{NLKjNL*MW)6CcQs)9aC*vNKCPi6Y3}a=Mt7flIJGjs4
zOB0DhR&VuLlkn%`UoL#vnxhCGt`-5?xeH^t)<CQ^*^#%aV%IshSMP{vN%yjAIXVTF
zy@Yp+9lMWc(N?ZE;2lW8hX&Lav0uk*C{SwSg;a<ykKZ<4gqxx@*rZw!D*QWkUXx3B
zTAK>sCkSyjVi!s?8N=atIE3la%m)>tMEEwjs0sSQ7wl4?E7q5nsDR)6Pe_acpH@Vn
zSBjgXZ@^|_7Lv=Gi%zc=@w<S=W==JGPH|7h%ADyM!@4=pEeyB{@Va`yKibn8EELCe
zw#yb(k;@HBHKz537QtiOjM>cq3ipm9P@s95y?!eSJyf@?L?K}g>~_os<nMeT@n>vR
zsKaieg$V>t5WX0HT=k1lF0^`A75iNRxvq(J_4sHqSKo+!(@ShK76P0-)7QQDL!awW
zQIu*CS{rRU76BhVTtK1GM>1o(ABrt`B#|1yXG<SR`x#>)F4wY0GV#J>{Rb<5BvY(g
z^>e3<*bY_qTji)#MD<n|<W%ajA5;GnhALV!E4*Ks^Xf;^Y48>oZcVg3;xB^TDmluq
zYK3Vr75bh>GEwjMkT6yk2z+<e_6i90hMb2ZR<E+Q#3(d9lJ_mvtf-Z(SzVmwZfnuc
zgKh2Bs+l-4VC@0<B%0BfE)UDp`A9yG#&~I;N!BgL)VE8v`e8YG^@;G84&Y(cUd)sY
z#UOCcLEKP?!8ebO7i?VUoWP}c@{x3uj&^bRoEJh!<+!q^0tMH6D$VExRVFG4@*(|3
zzxn5OF^0~&#=}QkQU?SZvdcm&h4jD->ydm!GD%5Waz$;EM#%M=Hv32>_w+(1h5_vE
zRRJ7b`mZ(@)7CIx*J5CYU2{J#q(Ais|Ll6q1x6|CxrLZ%GIZObzFPgEM!2pR$Y`IN
z)Y;AAw0<Lr^OghqA+~^DD7RwIwel+9VnZhr{UvtMZVO+?4<XrVbb!{#x7VE*0D%h-
z?W(fp3hlGI`fztYsI@rXe~lf;Ms&ejzpvjm<9iIVCf=Z$EmFMLR-nvbkmh$g{?gxv
zV4KI6J(`Gf=Q~#NXxa_DF(A}O6YU&q-*07)CYp7Y9+#y@(-Oio6m5uYhgv}rIQ*&<
zdL@ux&q*%%%16^d(9(6P8~(9MP$xM9s~^pjMgS;mQ)*O%?ACS*B1t<i)qgb86=^Zu
z3T98V^Jvyks5(e*ZB#IV^2#fpwfaZ%aYAlTVvTc9Ha(gToUtyH0nGu5krBY6<$jBI
zizczHs?vn83JCzhBEcW#4(Sk<Z+YOIP9ed#G0d=Ifip5aH&pJ8X<-eBbx7PK!EGS+
zO7dkL3d}GE<{}MDFiF7^g}pX@|4*Kf*%h~wfxp{*_u2XAw?T@Ksesz;Ss*V6%Ql70
zntn7>TGEUSPq@tKVhrRsi+w5jEDVtNz=iUcqZRx^&f_to9Ws%EuwK!;B%XNo(R`8>
zxkv_bHIRa6uz85jeVzsSnkJY?0yZzG1ek+Tu4|XR_`EILpG<B~KOPqIqF`vN&fS*H
zZpQTVb8-yZvdWNH!VX8T&8{(A5%hj8*l3h8U)X9)f%8j%MVzZaB|S;LebxdgB%bi{
z<90WCc-xa9tjE0Fw^Xx@=*-mz$8A#>!ev3lDs!WFPyAzB8sx~y3ES4h^pjbZ#ITPg
z>SXJ^0|h^pC^)Tl>0@c`m0X_q!_MNDJ(h`%h<OlC`D2+NV4u;{JeH0EcfJUyJ9k^<
zW9clwThgi?%lG<mbU)Q9=s473HIKbl&Wqu5|I0ty{aB{ow+2T&)mp8(BsMuZGb|tU
zsqiuglLQeds_oTS&iNTtAKkve9&CzHv?7wkQ74e)f@sl))S|5d*9*(o{#d?yh~1v_
zM;!rlAG}*<Ohp@Ex9wPbwhSKcvfda*D`F)M<N-B4nqq<2kiOhsidij0+oLWiACI0<
zAVL4w2R@dMCQeOmCC<t78dKt3{9!sJ%m>&>7Xn-_F;Q$<^OD@vFKV!-lHblHH95oM
z=W-0K0h*eLFCv!P6@4+|ZY?q!z;VVHv8w`(1slQgRPp`4KxOrHyY^T<E;#y8q^Hc|
z^RW^nSYn;72Q<zG;sXhzet3dK0d91m%8h7sH02Au8Q(`-t!&Hj-AVu;6sOvjjR$Nc
z;3wJ5?Wj0{&n#w1u}UNT{jOZN6BP)0jDT=0mV&z_rLMReP2AoS>R#8By8rC3ZD<Mx
z`;*RVQ)3vNU_AE(S>}@Lm#x5z&<TCEEtKCnWLB05ASDeLvy!q*LYM+rX<4GcvqYtI
z4!x`_QM3E5@?I^=TNWu1kM=R!5p6wHNrl?_tuqMgR2e<FA0zWt73Ikqw7M)&_p6_>
z8ufxG6_~rr5_J=x*=hs)DNg@AW$CbBi|XxP7vS1@2F}>t_yqSpt5<E1DU7(4WmCX9
zGjhV3)gz@CceIqHQz|jht@{1`U%meq)*b^u<N$Kk5fk7hoybP5GvG9K*)g@n2L=L-
z?vcuKBM+$i3Hj>{#ZZ`JsI3sVbN!Cncy#=h0;^A!Wda2g7w?SQ`LewKxynwvf(AHK
z>V?o){wlo?VA9i!yqba3p|fp&++r7na6g(KyQJUTsa<wC#)(&ax<qGGikWktT?u3e
z4+ZV9GiJ4Bo=fLy3?;=ro-4~YuG~2suWNxuujiq+d0~yp-4EpJf;BPPZ40qx2Qda*
z49JJjTiOlP8zisO&A>NZW+1ezy8DVZ@-5ZvS^Ao-1k9tbcDL`h1YI{PUe${B42@tW
zb0?O!_nfV%W%9%My4?*pg~3T%SH%=#$~Izxcr$odAbev25TdYqu`V8Rz-7L%mP9x7
z$a+p&+ZqB)iCYhtJ)W;d9-HG<5}!ygzHOzCXCgjH-LfboQD(Q5Kd${fgk|-P#}joe
zeG6xC{dw5s29=NJi}*yc3$0RFO3PRsuv?kg8hw$+%^A(zk0%;#Y_nRmfYL{7Pq9Oi
z?OL56yK$CV5!>E?_7+>A{_%Vo?e9%FOqw3g+d3t|R4A}!^+G15mdDe1ggr5~U9IW{
z_GH}JW5})kvguF}R+G?Z6ybG(3g+=FOGFL-5B@MuDwPH66>RtX;Xc&|2Sm*mY$#?2
zO*@lc5I%=rv+?4n`w*K<>aTnE+WF#N{REwQJd-Y)!L!Yb*@c*b{iI=-*4OA@(*9i3
zmwWGX$z6($k0zPO<rvCQm+#EP>><T*I=rHC%xL}nVYAUUhX%RFxqGh)VtTh@yJ~Zf
zXZqy9=lVmSTzfp9^KBkZ3;D!+07`~-|DAQbuD;=TVC4LCUkKdVJ$C(~20~CF54L$D
z27)da1=P)`e8fes%c}hT|KZNcc1vH(J+?Hq5?>fF*ln$g{Ki&ekq#WRI|8}%_ee}p
z>#pfD=aRpx#ra~jt*gLg1E6Lb0Rt)7rust0%UNLeL@Vwbu%({_>9cJ$@W8sq%|4W;
z3Ym0b{!2cT>9p$v`sC6N<@p`<V4sOgDf>{ON2Z2=^dHIuK$J$G#vRcfZzc4mA4>EJ
zKVg*t$~hi^s`^l(dwdKXShXsj`|e*`&4&`9Fo*EPyVW5oFzk||U}V7dd?=CcXFg|j
z(QO-P&TOw*U>LdE>eUCJP9qFziV=G-VfV9XzF&@^R!d;nyY$r_Vc8mkB$1Je>e{18
zAKvvHfzq*0S!WcrK!(SnrJs6ee|j~<rR!X&A4+GI6r#MQhcuY>HY<E6(;V^nWsx%;
z<4ZwSCgamh3haDz+KVP-Dhhc__t^!_#rt2zTwz+M%3buL3I;ue^(7VXCgtODV2!c7
zZ?YNHgxMUmE2`N?)CA%lM8_?C-F8()2sjBwHCODa$6s650wQsgTnjd@!6H7pUYwDy
z5n!g+q?<RIE$Rcl1!}mVO2=J@RQ&tW99+_}{@z2CThRp!g&zYefm;*j*zK58C<hG@
z6;=hwbHJUpJH^Rxc4ke#IY&{4#(rr;(Lly-TaS*4tjvZwLWtlJu}wi`yX|kXds+(S
zkh2sWOf=;Uj$xN=$5M^ExWJys#CIGsRx5cT6Bnj3{?I9XB5w?i*Kd1rBrAI&k-*90
zR{li31Ca)vK5IMTyX&;aYpV$0I1mk5<rA6ou@_D`SgM{#v^qiFqv|ITiNnJaFz|^)
zqsfpB_V$TPm}C{LRW<x2u;UHzy{?k2yAPh=_s%x$jSBq)<hS}K^2ySG^9}C<_EwVy
zGkC_D1vx1dThwI_BF<CR8gSt?hTN8?J>Yu2DC>w%_ph~1VeiK!I2?=C$?Jr*?o~O<
zg$&CE)a~bX*1uvyG5d(h&<$8Y6Uv)%0X@VwZCu|ZNJ6v2bv3ER!yfeE{1aKpm7}vH
zCD&B+*2Y*aL=Puj$bsRJR=}^qi^au2YJVw41Q+sUyBsiFf-s{mZa=59E3xK;%<LPo
z+4zP&S){80YILIetj*=dDKzV~C`@q$=Jg53&pRB6QE{zZ7_%Vc-l?v(#c0&xZZkKc
z!!e$92#w;K7b0)@<tN^AkBR30kF@7jRGK_B#HFz!@J&P(x-AS-uw%>^gae|ZF}oAf
zz+2AN)TH5XligJ%x7fFJtv6?^N9ytXTpO`}k4)O8_FH?S-4itA|IYMm701fE+_nQb
zWmU2AOgG5+`g!(SNqHg(_h&<8TAoRG9|=aStUQg^@3Bdb<uK?GTFc9I@;kfWX%Bal
zXYRKNucZooxi|-ssEqDhbqoh)RnbEani{JvPY1wj6IP>NhxXa-s6v7+WJKlp*wDL=
z?X^cZgV#x1&eh}lf03BBy@5^lS^P@N)1g3c9&0Mk#}V<mJZ?a<%E%dFEgBN=PH~W7
zajzn>-&(tV4&MAv)=``(XTGgd=#ozOSTu%6@RjTh^q0O3Z#+<*4kM4jWkdIoBMEy!
zpxr5aJPHW!WRo%N$x>TP+<A55M#@QgHx=VyVzJY(3qp_Z<IL-H5$%)b>|(%f<B+}-
z*m`(@al0(^+B^>6j3BsbV9c(>Fq$!(v{{vbP%j3$`rsBtCp#zju7X_oC%dLE7*iyq
zHm@)De8bTBdh`wVf@8%(bitiBXp4a;z1R&ElCzX>r^9wr!*y3}7#|Lc<s#xPZn!0U
zUc-oez*b^n{CUT2xgFiUF@Z6St*RyHF4~>gpqSw&(dt@p=K3b=t}6b^)Rk9lUC6mF
zHMi(SG{b5fju*2(p@2YvAawVlM}mlP$Zth;g3{q2ja`vIcMmZQ_GBhHc*HYS5}$t2
zTj`Vera9ojY7jpRSlN^Lvh3pBMy%B*(>C8fS+E^XW&+0>GQEB)p3L;)?;4&E?^>n0
z7(yKE88@q*%oO|<r)>3;#SP0iYoenA$>C@BlbMLvbS`A8RRy<mOAP%UH75ERk<Fqp
z+&72I-Y3&Z<K`EmRv&o2mY9jN))cUNG1GyOYko4HK1m~}V8Sgi8X`1d`_KBe#te1$
zVPrJu{`&0!ei+W)Wt$yQixdJyC-u&lY9f)t?`j&mhK^frAk&pX_t-%E0rGISnjuYE
zy4R5*Y1~*1BI4CJx#5vbOvWTgsBvC>INnpL*OL;NTu?oiOtNVS#8#|#7T*~Ce0A4n
z>{6`M>oAj-1NL|dZ9k)LH;nf{wss|IH*tB*#$Z{y9=jTaE@hn)E}VexU$$#e)W+dB
zukKGTea)_`gEqV;AVI8m10-s72srR@BN+JJC)446bi{5(MXsF_wj8U21tbpl@V_y=
z{KapMl_%fxcm!_J0I%p#aK1y(#oIy?MxS|J6=LrY^z2Ub?5W#$f5r%o!rI)8FQ~`X
zh1DBSab7eo%1!-swBPRO*RC(vR-j2LgybB!9ph-n;K}U6nR4Gc*l@r~KAh;tL+Ot7
zTj__>=$p}*emK)4BDdtyRsP{jldQ$~NDr29J3gFQQ*0<IKCC<TFJU3y#YU!5-*2IZ
z-govOnwJmfLjlLDV$!bafazB|*BIg=A%xJ34Pm<j4cTm9?(ku~Z?XE^^WjAGOJCuD
ztka+I1G2p_LKyyDtB=p4euXv(J}OD~_G{J`d*}@erCOpl=n3s|m$gRYm9N12w`)wU
zG)rQRj%bimfOZNXtM6_w#{zr{yiu>dk_3b>T+y}xV+=QS8H(u~KWzmK&INupuHS=h
z%08(|`%_T2^MM{Z(bF~+Ac;AYw4E1(FynJIEoKiN&WGWc0-s;hym2Ggr6}{`0`dH1
zb>$IX34R%nFw+B%E5fPWO>t&pm|RwNRW+TY69YIGa8uTuF1r@h{}p*N=3@*9*RZsY
zLgMS%f_ho`7L&jD;rE_{yco;>|Nk{7hyR}ksd%W%)r&U;0ZSzGGBE}O{OFT!#e9tw
z6cKzSmZ1Ht-44u>joWH*iy^GP6AgXey6$4_KH5v)vAg<m*GDK}T@8<NpKJ&lI3f}$
zMU$gU_Pyx2nT!I(_P9@?R@jaS<UQtF+CGw3(X~J>+>JjR>f*Y^Mf@ZAww*|cR?9w;
zD3ve)AW{C2L^tRqQOJBGQ#JOwGP0`pNPbDN=kN3Jl^;ojH+Gf`KOael1D+8JMK9Xw
zK!D7L?xb0xDy$SU1dqTw_9OY=!MmT>J*tphGB_ai1Pnsud;IRY_`}#)WN%P*fH>qq
zKk79|_<WcKj*g}%PNsd$F*HsaEKOj*TA~JNE4YRAd7NbS1rTW0yc}|en2zWhQvNzm
zhR$e8F1cPirU_N1`(8D4vrHS%G&RDT4(ZRtA4CHDk#x@bMTvN4Jb+<FHfWQvTK)v-
zwewNj?wa%|;o@QxMDH}e5F;a^n=-*q$9(U1ewUTK7$YFpj>p-#6uqFVhdsU=y|luO
z+l&w&;OCWSib?s>*T~GOGGay8RSkrNb<@&06_EYgt_gh6YYgj$0MCSt+jUI_&t@FS
zHF?f%_u8U9aHrzqMy`cC>Shdy>;dd467HxA$jXC`&D*Wmsq9lv9^<jUziulr(7vSl
zxGf~Hkh4|6ALxTd?x<r%a+Iv=fo-clWADc2sX-o<>oFk&1N6z=h(C0VCQp>j=##nR
zsJW+U@+1MU&Rgm#>Bek3+P<M6bG9QBtBcg3<{fFTkd)2$Y-yA>cTMnSD%<g%yNXS?
zaZ$NyG6Xk~-mxPQxyNBa4!k3gYj@X#RjL+5;!>e16?ioF>K&Q*Bc5VSKzswE-`(nK
ze83HWcVzl4ba4jk(I;k(iHWX@(v`pFtlk@+Fq?!`6VwzInxGHD(|yiKRuVNiCAzIe
zH4_i@O>5ne54vJvr}jW^j7jbgdP%a<8E|0F@F3SHPhz0c8$giG4tZ?os7eqA$W&EP
zz*|88Q$D<L{XWrma@;0&WVU|apld>i+JUuFfEn&1x5ZSfQ=x0=Ybb3r$_X583B(e3
z`{U)Js=9hB?c+;3GP9NP=MZ(~a@1e?_dIgdlbcPs{QOE34h~>&s4qVW5C>h2!KSAM
zDL*y3SK&CXMLQ4AJg+a#2H$1Zg{=#J3I3;#<Qr#;fgI_u(hj<zmhdr7*f-U|sgZNG
zte;%Gc1v}}#?N}%!j<A&aJpfkq>1H;y4zNT?jRunIF#6ucht@q2cEiX(RX^wA<4O`
z#uyb6ephwb{jJ2E*w~Tz%LW$?ddg|rjP|_NdC-m8y{Lr-uBi6j5*!bngvfZixE(l+
z?4y}5{o#fg@JAErz5nmn;iVtVTk|+-4Oqppk0$Cnm7b^NA5Dxjc^!5{J@;$l;aedf
zhdjblr8;m|(qx@iB^V>ek8|u-e>Crkln(<5ZcP#D+qV0o`T4sK0<YEIKmV%j`Dos#
zyr#0AvM&DUo&=aTdvf)?AI;><a&eams{z8kSd)G;UxYx6j{FK~36v&=PY!zkzgE?4
z@Khk}F<tv`xrIDGnvS4`mKN*$XkHvP6~~}&xnlvko2OoH;8PgkSQrqj#6)^frXkJD
zeyk{lte{R0xWn6cv>Wghp9~zvaniypDT?>?VW2h@13-`j*A1Bpi*$YblT8OsV}z}}
z7!&7jQ_iHMV%4Y`S<1`NnVYJUASXII=OT|Q!Wr2+_vUQugMUj91h-VuqP5~vqk4@g
z0mij}(dYrZ=7li#+n>7)FXHy=sx4ox1<i!VW3a{I6agPQLJb6*-M{T-@#`N*Y&>Pl
zAI--;!8~QYc}sB6k``<wFzxDfs~x)?C=J5Q*=m&2)`cW@G?IO7wk9xaoH4r_lY=V_
zv#~DphDgA{GgcpC1AVOZ=6(2F^mb1`QYxP<)q@*^wc5UqH)hqxT^?aii6SrJ?OpOz
zBEZ(<*HHRYCS4ad^;+3e$-IxVf`r@Y7{fvz)4Os3Kn6xsv!^mqHDMD`@l+=8p1}!E
zrB(S<rpF!+R>Ov`dMZt1(;q~Vua15-wwk9hK}M`&T-~Ch|M5?+TCMN~wViuy&r_LD
z&ZMB4PAGLn9Nqn8?^8*}K;>|3VjAiN;v}aOhD%SSvuC*5ngcS9DlO3|Y=^Zzm1y&G
zgoSFqk8+=re1`@uM{qwXI%6i>sEKhO6DUju-2STfseF<QP2fg05F?Q6+ea`IRo-%I
z)q)0uY1~O0FAfg5tWD~-i)!q=e*0?KRG^Mws9g~1g9v=5Rp}IQ|LmeFVUy+()k_*w
zQ4+IYmjlGnzJ8lgI|4h8+m$GJjazXx#tM<TGWTly;Y45W2>W<WW6?QA=p@bdbDcIH
zHQmRTRk<Fc##9w$Ys|ABonyABjtQ3LqJkS56$XWNGXRko&3U9{t>Y;_IPF$IMIHHN
zTTz*Z6&|zOF%}OfxEgR=r_7CVN5jUjaNO2}4SE4t=G_&XZa7wK2nA>G_(>1)x*^ms
zPq5ACz<rDEsZAHjZwbY$4|ZmEmu&}<`}z<~+S8e`-T;k~r!#H3Sq`C^pUxYo-;cbq
zr!xukVx*9ga#;D(x=PqfJDyHt?)>j@tI*$%bQ97{eK48CdA6$PgNqhsty-mh_<X?}
z*XS!bg}-aNpU&qY5(pFmSd`Qjt2=MoXM6O;&|c^Kt9v?~Ic@uow%gvP^O@i`C$f91
z7e-x#dTfeup-V>A^z{7$WweKEa4k{Zl&(?@ESZ^DyE=hkKn#4pp$CA&KuI+rR*uqR
zf&vL7xBK3hJwo#%s&{Lgfxy4tpQV}_r0R0NEC}OwIy-GVFk;-un&Pv~ZT`;(9<XQb
zpEMPt_YAha9v6feflN6_rti<|)no?r9lIFI(lX-K_?MyzR%t|Q;;y<Z5FBOD%*@kS
z7NK|n3@35G(J_uz$>wY}{&H?&bZL=u=&Jt0jWt~4Tjw+l?puIW;x!Gsn=rNWnr!kt
zkJ<H@tAoS+w(xX1OV993jk(}P{ZqT4CZldubu;EA$vTiT3G~gsg%1`u#`evXf3%gr
z(&yJla$8loL2Na8;6;hLRXt!ghZCI8c>Ac`)$F)#8=L%Sf$0p7(hRF$n`)c_E8dGa
zmcKVA!<K3`V34;RFuFW+hE*g&>tXdvDiWa~>qIzMst+l_%Jcy}bbiUoD-t<O;HhOh
zDiS5*_`>)`tfC^3{rl5Q*<V<tzG$^`=T55<3}(#ZIH-tC<?etr6=~8BwAt>8OfuMd
z`bk);&hUmHG*S@)T5lY}h4(>jm%XZkHt(d>2LLy>f|i<up`TKHyz`rj7<x3gZ(B>$
zP9`gaLyA9)c)DCvs}=d^N(lk2qax9N5<g(0z`LMRh`@E&u^8~+kv8j9ReEmZUGJG#
z24Wi9kJ?aFego1e#Mhf#`s2~m)BF0JI~j<<T%O^sIUfy{zKS4NU${uP(fEaE?vi^y
zq^6_i)_%JfV<NyO+xe14@X9N@?XrF@eTC!>(W4VL7vmL8GtmM5qs^+`Npu2NEAnAF
zhP~T>Yv1Qov#XcP&ew_%{|<|!fH>-JF?>;D44pNdwh&)UtnXrsfx8TCD%pBf-U0=(
zt~UjD`ELhpIpE60)NTbztRlhHIm1_C8?<;@+1rA}rgffhH710>>^w2<1SYL+C#bpO
z(nDl_S6dY_SXU*w^aOnyMM5BafEz4q3L?$N{K~U?F&ZuggnGnSi9U$TYuf_b@S54j
z5<L$09k-H?<!cfy{!gv+V~Os%hj4=|Q!SVUEB{!Y|3;oSi0-ivjD|SL@BmfiL2jtZ
zkLBx&J((=gsy>#8p5vTdSpBhlZ8!2RsQFkT%P+oSy9HtCTVym9Y&IRo*{JPNi*!G{
zYjr}AD2Es$1*}uVyQx<RIti4h`t3SaYc6)rN)Vh(1<AXv^<#<fkfmTS!Wy*)=A($%
zJQZ?>Y9D^npPMlWym#FsnDp>532sDYy~Rm5H)sR*F~e(TC;%fZN_(XsoZP$xQv6sx
zk;cz}BsNIV`VjHg&POpm^q%QwO2r(3&Xrsc68?VhT{tZyx_Yp9_*i}?L~1rDF<>8s
z1TF{c7UxazMvWnz;r(J)ibFa+XtSCv&KmX?$L3Yd4OqNubE<Y0PWqaD9-_v)7PoKE
zt_LPAL2{!ANOK{&>L0;=Vlh6kdE&7tIQTr+&0<e~P5Q%S{U(b$E-;!zFnSYjwdex$
zJe!11#_bqvKSH_HsN$((32{Aw;m4f-!V0eGqlGsMj^XO6KmaXUkHu&Ai=sCIU)1yy
zwi(ld%{aUEUVJZo%Y{T+nv8dj*>;q`O;KXc#QC4j;gV-EwY7PQV1H?rK9eX29UAji
zrVnT1d22qCD3%u#H)S*=8i$KRa9N8#$=8tPm1@Beg2lr#nN|tkoPx(!3pkNhxD9Kb
zNe9K!SBZM`OlF`t>oK2o<G<Qx^5Jo+3kG11YGamaby0HI#h825Wfw_#tzN&0$adVC
zV&u(<r)*Z|Ea;!~S<5r|py=lf)f!cmpshXQc?yj|M|9`(Wt~FA9Vw5gx5R4p3ZUGE
zY(M}gIn`xD&%Ebab~+}*bhs}S0-aXEst8hRLwmDLMt_LC;9*sC+Hb)r;h8RPm+V|n
zRhPkXrr7lT3HPX^G`JXlVb{1>k6ntPzKu~2I+d7+!Ba#zQfD3#-mU~5o*KYr17Czx
zS5*(LfK+Ayv1?@MpNMS#Og<RU#bht5`PeG%)X9r?$gbaS2akZA!!zl$e&;1y44eke
z7VL&P9vV5@hp}LB23&XI&z|J~0;e!ZxTC=GZ9KyR6U;MZ;jm!0pULMTEUf1rTh&-b
z+=B5=%oi{8ako}n({k^6=Sy}s&}xO=*JIDpGu|xz%-K-ooqzPWZK^V{rrb*Bo(j$p
ziGE8zyG|h<QXMzbg^9F}=MDe1tNKblF7jBwW9{P^WmJ2fpt$Vgi7q_}xs|IP0qAXq
zAb?MbvRU!*yvG~Z3Y8yERPN^LBUTlC!9!zJFL+oV@zPYiK^`H<eFDTOB2Cq*tAmHz
zZBL-&VP`mJ>I9KjSJJNR6%_V7X2$gb-|W$<nm(S-qu00x$u}XGZY<NHpJQ+itx*eO
z4i4k?kLNw*0-tf@g<OywA5R<Rk9J#Upjkj$a!h@@vJPwdK$Kf_;yAh#_43ZY?zACQ
zbVVKvY)pxp<J!1DCWE`L+vLadsp%?^WZrc?`rL<lZ7L8(Qvt#kqT4oi0E0F)2NN)Z
zb}^cs;C#9C@qA1Yco}g~uU!sc4MW{LyJrG5@(d7FB_@M3Fj$(Qc=$(~)mYqejoVY5
z!K@(t(Rf)VyB0`sUSs!W^U)C@HF$wukB(A;g$3c6+)=OFViXRH5Mks-0BhvFgVDGd
zz#4j8)wdjI%A7tW{kRo>B#)Joy}!X%gf4=RZ`f_2OSlJH)%rGd+nt!O<9+?MR-8&7
z>0SNiF^Sc^rhpmdxY`J8lLPGL$MX$CzaH@p71zC3RJdG8XSM{w4XZHg)dyT6H;Yx~
zRU-FIaw#sU%=8M2-eskgiC#}|43t&IZXv5cBJT3aM19c1v7NS~GHu+JCab7SguHZ>
z0Oyqg(A3v$RRJ0k>C;vnpZolYSffT**kPj(MJ;_`SzDQ@i-!?!<vo>I8sUcmhZKAZ
zM{iwJQ+)2dmHB{?I?w7=Azd_n;4}sLZ@yv8LjUk#YpKj93`%!v6<*k({zq%qU{1p6
zbVQ|pnfg2H41{4LkOip?Z|V{2jf$u6vKbJB!G1y>sZzH`(JRC{_^$GZZR62w>8rgq
zSzHE8iO)yDp{1|dRAt@)Ske@O^<Id1Xiwfn(`w8`X&3LW;XLLP1QoJNDqZ@{DZ4C;
zc-Wn@ndrv7*ssL+PB}WWF|#C!OG*^AglezboLV0Ag9_>_!t8@3n-7@nepBkJ=wZDO
zNDp+m!TzG!A-l%5KFZ+kAn2O`2O}F>4rIo#Htr)PLktoFF};UYD&H$IE6guKGWprw
zj#0t=4%uq7BWajyQtt?hHY{6gO}!@flfSaN+B+PAu2WIb5jl%*sL@d(;@PHtuUulg
z-P3RMAei+A(&UYFyWj2jLz`cKq@4XkBF&_EOA2Z$`9$8A|MHjrr<HypQDzhLQTB;U
zvqXO&^@K{?D}2W%5+(9%MGEwZL`BZ(PdHCp8Q@yT22k~hOe|>@d31gvUrxOHK{;K3
zHBsx$f?qJUTR=GhNE*aFTN|k0+>FFF`e@}ksf$9SC72NjIZrQd^`FRx3um~?Rxl&_
zgFiWw$KBZciFDp9eZ}<;EyZDCv)QU{(xd-7)*hXr%eD?x8yO&CVD!({#C9kKgG!Ig
z+}10EIDs(>6p*F=*5w2}q>gc9;9xK4qbuEF<DW>UUf&6u)Svw@{DPeqC^&6y(U>FD
zlYqVuXzYFNQ=hWw_{7Nin|ARNd8;D$^r(I5Rdz}622Zlwmt%ZxBrp@BKrQ0g1Fi%-
zs0CV$-n&vT`Ky`_AlY1W)x+Bm!Et&^UDJprF&)wtWJD|qVNg+}#2B$-Sco0+n(Oq%
zA}vOnhN6A*4R!b?{MF6CHHrKeY*~Hn8?sw5C?2}DqT#+jjU?}Obd!|BP7B|%)z}<7
z2r<56E8Y>Tqm8x}SQ2sLZuHsJi;pxcZcGTua$Jr#)b6v#Y%{7Nu5kA6Ui9*Mm-`j5
z6Sg#(og|{(4nTdA19)87C-Y^>WFne_+x=uBOx*raAdqhOlZk+xyFqYN_Q{|)m+c}l
z`eY(j{EV;%{A9jaN%mIp$wawrDX$`ISmh`4p`Y|9i8H8D<D;IQueu1x1CG?FZ7-UQ
z6k2Y#zFAYea(Twpeli^+gZSU<QJ<tZ=iIK--^bV{do>E0G>r1BtXDO>{+RS8b=7^A
z-_82X;Qq{7ieNF?X;qILWb?j&)*eG^=(mnfCi1DNw@ytCy_{m#9#i{~-fruSdFu0c
zS;RbwAz)t*szU;}p9_bWDS_;v!EplF3#Yf;PGvII2Qj@LX?9+{A3A7L(Mdm&6wc!d
zpUel6r}jdpgOgey-Z)_w1^A@LX}P3=Mi^6!5u%rIeE65X3ZomKJn%OnBmBLKbAmd;
zxe=Qc1ZI@C_|=$+7G#!l`WhecnC*5gaCoiD=3^p|KJvJ_9$j<p;Cy)87NUm71>sf=
zR4|Uiju~>e5fjE_j@ixV+|4DIqkcFK-MT+h^4(j}Qk`D9YPX{dJ?I$YT#Y}R8iCP)
zNLBeHtjk*9b~Gh9x*K(m_YHVti}g?Dlh9K<AgxIeEjD719@Gx0yKTl_+8|PwvfK+$
zCsE@OpKUAtK=w=5iET&c1eS%Y?b);sKig>~&*pm%Gu)J>jEkrA*+iPnT>vk8Hs7{5
zwUMFX*-XaJp+}|J5#?`>kXNDN*?emcFMZo8pUtf7pCH^gZdC%Dd?2c$%;TcqLGx@r
zPk7O%vF{GhuX7q%t-5wjYt-{d*q&$8sX@|ln3B3@^Yk?-rxP3^iJ$_VH;{(ZKbv=X
zQaHPNQBzcS4WGj1XVdX;m}zbiswevfMy*xTRr9j7>(}Q#WgXG*ZTREP_;M56WBN*|
zjBs-k(1zFc+d#}7lkNBJkZO`G+*z{1v-zZx%Z~5q@qoOK7wu$BQnI9H63z>Hm)inH
z^X@)!j-&vp7ovfuR&d?lbj<r7Pq^pvMGbTWrGs4(Fib7&vi|)3-*AS_=r>|~3>SnC
zl#4RZW@Cwv4ij-O#+M>Pz#Yeo^IG$Jg0Kzh;~cE(d{l2wwzhUXz+p#)?SiU~z|JiO
zAdlAPX}s)43=MT&l8fCGZvI%~HN7lg9%#~IZQP1+6P%lv2rJP!UJOYo8Z&(y)^1gO
zx9xAYI~xAd9Mip~-%ux^2KBqSy}i}eqcP?%wxPyEg|<!oe3O8=_q1Qh?y~eX+Y<Ka
zb;jcaW3z4H>!}IcP{XPc?QmI&Cn~8*M2WNuX>V1gm1G;k#a338$rN+H(^ei8-saSU
zTd^HgnJ|xY8dXH~wv^>mg>M%FAXQ}=?;&(Jb5UKD>BbX3yG+Jv)M88<cXtuD)8JOC
z-v`l=+a8U9TW){y;jXJn^!3G>=WTCb(Cq0r>jU9N&M+t4rf7?nh(}{{Rpt>2z1E^m
zCj2?lDxhuu7H77qbgsSps&&M0+mR-9M%#82q|tLS>+6k?I5i9i7~`@b^>NOhCxvKp
zU}<sG#;fvSb3eK+>QPnVT1@QJc_FjwyiKVK4t39&dO=;B<Jg#vQJ*1uxRd+En0v&8
zu9m(eC<j{Xaugv^@~Ad5F~;V>6Luwv;XC}jKda)t6xPbF#u{=Va4)l$=VFEx%Hf(C
zKs=sDab8~vRAScyQ(st*oW7uPY!vYqu|>`B0d76q*`hLr<=FCJtf0B%Es8&CU~&`|
zGi5Ej6_})yTDGF<*y$0`LDSW__g~v;z-^Vd*6zfhU4+SvyH*?nX=RYD-;Lsiexzz$
zwjS%)fNaz@;uD{4@;qs--+96AMFGzY0y=1P?J1AUcHqx>$jQOO&z{Q!!lPJjt|iZ9
zg2Z02+e)8H)Qeo@udM93d@CLt_K@i1Dr3{&l^SJ3U7ljKBEYQtSEji#J_|13ta>id
z#Gz#UTdj|zx3U_+;TU%RgazCET&C$ZMUAO_E*~LiY?JI_&vWU7@Lh)yblr1_R7YR|
z_QvQGSeF~FUUebtM!3)>!P<wk!o6A`-sU`{(Ftpbu8Cjn)`YDw#>0KCDQ*u`Foq_D
zB#A0wnEU*A>eR@-dx0R7c1#~0ABDSBAK-Bg+SLcI89)q0Cnzd#r!NG?T_hA8e=Z#q
ztw(J#I%g>o37(JPCc&}ivY!fI1%wF$b|Ho|b_%)2fK3ZJH_rm@+ZXjkC?g;4rD*JN
z$&d_da1BL;&x}5slCM2x_ECF0si0@0t0NQRzFV$pYB@Eb;O8>;mRy2#QCt%QBoyQB
z25Iv#9^~-K1baQ&^`<0p90zA@L5N@>mEPcs>Ic36RjYpUSb&rRTv2lUbz6?*g5SqQ
zHKy8=ZLY+C`iNH#J!@lip?h4t-)^YyxZR1R;I>Etm^G~^j0z@acLOYu;B5W5bOgJJ
z6Bda%R>wvl=l)p`=df+YpOln?6}T7u_ITj2t@tEM(6?+mK9dU4c4lf#Y=)KW%opha
zr|X^hR(fmD%64Y{4I5^zNoPCCqx!zSLEEu29Rmn*32YW69S1wCa%Vm-4mfXH6@{lU
z52)UmiS_6N@?TEhny8+940j7|pK}(qQG_-XE_r9Bd`?JLCe+0r-1QMgY46Uwi@12f
zIZb{2&O{*%J=PTPI8*=Gn$?xFT-KuBt{}74o$0`sz;LWxKanF^M~uT$8F#8aJPrBo
z^&2uKAF3J*QDs0A+Lmmzhhp4rZ0q{^LLkr5xk`IHmIn)LUZ<1y*8txdS5#uA6eBo_
zo0LrnCqIkOpIi_gN1f-I7O)6zcQID?;1RnN$Rwu@r(~f$ifPG=x=rjH!Zo(#yj_X0
zx8a;V8(-tNfn8NoPb54S1xJbHcP)B>C^4kONFx5?3F6(E<9cx#7coFyh}NDcc`;_A
zfU_OX-Dr#TR7#I^^Zo>)AYsCm1&mt?%8i|XV4pmgY06guCO(CX`nDQ4b=hjbC!`6{
zpbK^<{xVwRJl0}G;5>(&pT%-QFr4eL^?=`;vR!UyYL>n>YMVRry-M8N32X{_>|U%T
zN9{LkOMr-)@ej6L{QVyhG1&8&oK|Mo;w8^#$|MX_I^U2_J)a31lX3{9EShxnFqP#|
z%*zJOVn=*qSIY%c@qDJMmA_%ZaXo)NQOsLSR`tB7+?llM=owC;7-p_UHG43ccs`R}
zWWGbHtR}^$fxHX%L~n14MXD1z7{dKk+H8BH0*4_vPOLtLgIu^^O+}nM1A24R^|0Ao
zqb<*86;qyPd>HXmje2fpj9{QW=3pqL3hIb&JWAP&ozLfUice3uYR8^Wr{z)ELH4SC
zBdXtSS8WVDpUw=liZ=9oUTTinl*q>lF@pVwk!)PE_OD2&8r?K_Fdi1;eEbE5BHgl6
z0{yw2U$P5o_{C0}jsZ22<BZe!VswF<BQam%6YQbyfO?c|<Km+kK^;Nye<ffQO?ziG
z!w@I|m||iG_{HmME+*I|af8GwO^KbZsXrc|r#;@%yr6e=_t<qoPh1YKA8etXPZtd@
zBM!Yq&1$p9-X#FPIz9^PWH)2Gk*Ir237wW>RA)TF$Sw6U-1l#7B}!KQjxBRrB_z7A
zRlyu{(bXOG29wn_gkmV#VCRX$jVAZCHd~Kshmf^w#LfoAe0pz2TeLDmb}we>47Vk0
zc6=Y7;9jup_$DSH6l&FZd%Zn4l!BI(ROekoa9Jd|+)1U?X;1BZXSbDAr#bTgEcl;S
z=YX1?Ux|a2_dsHFb_9%MZtIfl3N>WJ2M<`K01ji_P*t5s{9uDsS7$B&b*|M^$64Uc
z^Ha7viam~CZQzYvM7mw~#9tJFA8nC(fj_go)%jo|BQ8`@AK0`auu{DSLd!PmC;Ys%
z1UxBGxQ+t|3MQtd{XU9_bU?AAIvrA2GeNiNp_e#>J$6iktogk4R_8;s%~cKq@!8_0
z>_dUpn`2gpGJ;{Avhf&y1F<nD^_Ao@|J}|9COAF&icJNogg{0zbD=mDecd+w2XL^2
zicVmlpjIIO&dl-U0Mdw~NuSLK$Vo<bMI$}8^erM83Pj4vK&z{2_37e2n^UVUf~H+l
zEAA^y_4OOk2bXW4U+^&7+`xfVEAdQhF+SgRquU$RSyaFcJ3Qv5q_8nJW4(Gfchaj^
zV4dfePwFk9vh;W3wi2_@*!rg3j*-H=)8VQrwqeeFC;H}5*Kcd_Nf;j(=vc6ZleQi(
zyTNB82IUE%nA^?jd?R>NA95++H^BvPuQ(Bd{wP+zZ|wgQ+YXfc)JJypLZS}l&L7<+
zFXU~s^0)58`a-^oIq6gOdG6j9@|K7Hgl{R2Iv!6ARtBQn@j{--o)JPed4d&cR=kk5
zF);xzTICCgJYRaps-jDdg)XZWKG1X9S*Z1eeCfKD2ETr1)^@*;N%)9zYSl)YQEZ*}
zM6p{LGm1L(^_ef&-WQ4|v+Hc@Ur0yPA3Up?0uIqytXbt?0<E$6u{VyPbAZ7bwF+?C
z8EbzbA8ww97gx-GwvK4-2@*Y?LT5C0ktspe)ckc!F=7so6%Zogg>)`{{);s>6m8%r
z*+B(;C${xn-ml~L%elt~Y%&H9-C>bz=LOmgF5E(PDu#HvkH@iEv$!r+fT@mQG!;1@
zyQpCh&Bw)pP7s#@#7TJa%UaY?KmBI(8wCQ&UX5V_CVn=cqBABA0YZtZ!fJmC-ii;p
zI^mW2=P7{lf?bPo?AOzLK7brrS|olNH~01UlfVAmR?ij!w|+2PR4*>K@>B3e>{PBz
zm;`@G-xQkfywYjQ!X7H-;te<_;k#D?nU%lcWSqC#s!E>7<F=|jMJ!8}=T5Y1=7F>p
zUmg_Yt{QR0VmlSf_I3&kyb(RXTypT(rcj^2bjs1Zr%vFiF~Zsku%3ByTVOGp;?{jJ
zQ^qNzdCWk`i;22dzQTXAR{CP5?t>_%t?b48AjT%KJU$M!*$#au_FqK+<u+!MS!IC1
zCw;)G)CbYVxYe!t#YE(&s79<NS|Uv#T$1g6F`pg8jDsDjeKBqHKCF@U1khe|_Er})
z3m$pO_No(zagbPB^(sYygZ+rBhN85Gujc4BIRzntR&nx>e%D&{1?oOS?ZS8ID(~Ws
z0K+T8Is=9AzAk3ySj^B#Vn(PL!7Z@J&~H5ShixbtwKwc**>44b^mB`gd|V)ox|h98
z#)xr&Bs<lMnPcMll-^HMFWx_WJxqveJ}yKfLiOMv5T67{L?j)bZ}^k%QXqp{B8@Mr
z`!fU1BFqTu=5g{TUr|A8dKt_HkbNG>@2Y0=2)v<~u(>F2@GXBWCe4%7&qrZ$zP}zA
zIN`$@$I_q}?!iE2@x^oxRPVAI(en5Z4E@chDM^Pd3rE*X4co2a2ZC69bu<!=C|uE1
z;MYg1tp*G(YR(<IqnZ=`fL?nsAKBy|8|FzoYIjw<GPiQc)`hu?2@%r2Ya1`-o5<hW
zZdSrQw;3H^;pvg_?rAcP<8#CPuqAjNU)i>`)z7gE_EI7dNEuhxmomlf>w_PJCw?iB
z$jTz}9+w7}y_C7vCity#l`=e>n2}BFcqvmf$1~ops_YMEJQXXWu-W~Rt5irY?hI00
z?8i4I`Dv}@r9_GoW9Mx5OL=A_hGL+#FQv^%Fh<`ZPB(i36Ji2z_Sb1pON+qXD_8~I
zN%g`(jfm`096F9?*8EaFelTxsQNg~Ek(1UMpkMFm;X-d$=L{NMf(N{}j+nJU7t6yH
zYi!Kg0%Ue98sG^r>I_M55xZ{?bz&eIr+m0@$MFXjs3Vam#Kav$!EfWS1bt`wY*J`<
zC6?xVOjR2y<dmdlDr&cIbzBJ8SStD9IUTT*k;6qH1`migPmG(m`J925qpi#{Y$nEa
zpvaM2324MW>9yJT?Cm=T=?b`KKef34h8g41*9fwv!UE2)HXl86Z-Di6Enqii$byE;
z`;{eIjCu@(?RO)9V(Nq!m)(3Rk@E|?Y+1j)^pf2Qq!1br1Uz6M_ksbv9rLxHt+Ohu
z20c&Y9sN!soV93mAo<PQjj4cgQ=sMbSPER9ltO(&IC$PE4e0FwuF^mUwg&?YbzeLn
zw)K_b?p`;m$uytH@mncvWJyiF!}^E(`YWwT)QrBPU}ZI#nz3ndW7+bWOuep7*ij?)
z^Q>p4s;J4NJJyFhqcXrA>N-JjReUGz)2aiKKm2NH(maz`akqZ;oVHrkILs-{=Ghad
z>`Q@6tWE&^Q1Q5IwpZWo>kk_<X7$kkb%#rcnriZe<j0Bo)HMr#=PY4{^>F~>EuOG3
zn6@5Pu3cqNY(tZE#7H=IAg4}Y{9>maQ<vDj6|a_Ff#J$eitvG&bS9m0ue+LjCMAP%
z&)`^4Crmtt+qi(f^hY+SKVxDDL7cD2SLk?O&%kgBEix4|xt~#9h+!X@<Y>g0Du%*o
zFmD&NyyJsJ9loTN{M}VYm(@`VF0>ds27au(M|Xe&4nTMY6Zl($Aq}Ekjgn-BIj1fl
zV5m~RxxER%bRSU+UaqS<&y~kRRuI_1DFFVWT6Isr{oGJJ<{tLrW}xK`o7hqH+H#Z=
zkoZ=7eD`l`MNnWBQ2ppPhW_VlHSjoyT<wmU;FHHLTT`|7|C%VtcVi}x67ZEPVm*M@
ze9ksvH@P8sVsTZx^xxcJ@Ln{=p~IsrZRroNoUoGPOJB9^m^2=)I0)>@wCVIww1zF&
zm3Il|*<{CnJulsrDCKOAmFd?z=dFBKCcXWWeLc2gSK4MZ)&IsS^b=XDRqjfp+pTN4
zYF8!(4}s%`9n~uLE61<TD2EZ?a2&ARyVAKp93;Hb^_zDvjIO%qfbFUi2$rfH{nXw7
z(eH{`x~Pxxez=Jy;opx?IELBGR`Z~Y*0L*~O8b+zvQ@n_5v;F$S3Z<nsfetuBc{%~
zAq$3^=ygV4CwRRb+m-L;!$e*R*qD29_zVa(>4I%2ieRCL4p0cpI3&>K!X1ytWS{aZ
zE|cnq*V?yvubmIvQjA3eOoENZ$Sy<~qE(MhJT0V-?r&({Zx;jXSl{U`yA+l8@!Vqr
zUygDZSAz1#=gJjC0#^ho-J&+D)oK5e*8O%h2F}ej7mY9x9I|Usw)5ZEyjJ6-V|HE5
zySoOkwAhu;8-G(v7_z8ljmT<vgKB)ZM%hjM^n~BbF^z;R;)QoBMgzygHntV@vGN_l
z7~NL8u2CqnIGO+6nMJ!3OZfT-q#4@@wH>0vUb`E^X@IJ1Jvu#=((P<$M2)Z6rh0q-
zKS0&@qERww-HN@_4Tr(W>pF>TVbq52l)ap2sTa9;$;)|5wh{;EoRz+u2ns0zhK*MC
za^96K7%Rd$l&j?*zq#LbsHR6)$G4?IaNJZT#s8>$Iql8iG*eajfP|c<4Vb4elSaf-
zBbaX<Y_i>gNktI1+9H}RE>b}X$h=^6g6G!{v6KQH25|%xuzEG?;x24@IUfkT*jXg3
zj9<>D3hF#Up#^5g>aF$VbQpBGGP(WbEUzQq8x}fDj-Qqt@#mwSvMXmD>QqZCrpc9Y
zA0N&{G;F;^u&$ULcsZYaE(zg|de??tPG{fl$E_e_JbB}wjmK0FecdLb-_fP7_PY!<
z$-2%*Uv9;T^lmDK$MH*uQFTr9r9qpHx<js7B-D}$c_ShEjdnZ$kr$~brKg<{e&Z+Y
zN>oCkddg;Fa*umF3s&-KR5;@ihvs5$^^CZp{90fzK9a(D&8wSc_msOH(0E?(B7jVb
z8!T-pTMXEz{3Uba<$SYvh#8JdHz2$j9X2A*9<$|W)9!Aow=|NbR$GaR9$NwS_T_vc
z9L8rtzK^TLEr;zEe8ipNjJOrvT7bpH7<W5??dC3x0=%M$xD=n=gPd-gFTYpxBI-R$
zB84mgj&*KAsj?MK_D#SMZ%1J-S?lbTv_FY)Omxv#GS51U`%UR9X%Fphw6a$c-R;Ao
z9&wRAAg8XAwnHu4krM`^;+0I<sD?buROKsq3P$^m^91l~x$2cnEC(iV#;cAt7_@oH
zg&HA*#2u%*K<Gr@39Ef2oiMx%ZO<#21}Bo!52_heO(NKD+g??zdFj(uAN?UFz+&;0
zd^WuX8$%RB)g<nUwFJ<9Y}y>xR^c*u9Q{bUD)tk~p+hiS_v`8s>r}OFcji8(?%e;w
zZ4!H5dGBetr2nC&*je#pkv9;?Vd{nhdwjYHxE12F*Da67W2&6{<H(&<zY{J{J1?A^
zLrT*;rDbwe#%a5tis+%gY|{dYSOnJ`a@)L;H_&|*Afy|!bjT&dc3E8~+tU`emzoho
z*I(L|Sf%}_<Yxov6x!!%ab4U3alz(<)x<DDSi2TGspW{xYq#}v*>$xTOG%DwK^?*R
zp0h>$b}@$CxZlkTH{ead;_&A{T@LKq20hczEw$_S=&vj4=hR8NU7Vp5kZSdnd^9JS
zF>i)s-F`=P@IqkcuLZtKe_?l{54b`O<n`D*NQsE7fZbwDOCtt$n^E4#jABV{_g=}z
zb_(6&H^Ei_#UGzV-7XG({Qo8EeS)jLvOLeTV4(#aLBuc-ELc#KvMqPXSWe5fY|FNU
zo<M*E&(Z^ATiE*Z1QPmNNCH9P4waLPQcV$+R5jB@PBC3nwJKUUO?If9WR@~Z=~1*~
zhZb7M5iD4+ASzgB!GZ+~7A)p-?)Tk;W13*0-@E_L{d3Me_x#1?#A?%#Cnj7H4{<&2
zsI6Qr(9NpajCxoTx@G9j+Jv*nDwy;xeGp3rxOUfOEGH|q?WxV^xb<J~0NPud(W68p
zWFDiHG29TdT4Qv0I^3xVBC^(~TS!5%!)jIYav8IsI(5qB;c{!Mk1<;P>@=umOIvfN
zHP$8~^WH1gq+c=Ahv8MnFfS}hTWww^(q;4fL+I09n<&96XI{0UW`n5EV_TMDq!gaA
zTlL&)xJTWSwgP#sy1$tGLCUcllSrV3RK&BC`(RXy0k)!JaOse~4^;lybj<&i9-Gxz
z|MTkub~QG`WtJfFoCfE?O~|F%e93^t2osY`Z9bNrkzv%Yt2578cq692bt}=3Z|r7(
z+{Y8+R*aVNI$-_(vKVIbP854^5RUPJfa!2O=%QvoaA`=vT@5<IIxf8&^luVNv26FE
z*Kw|m<@ieZ;bPedAC9eq3gLO&%yd2GMsgz37;dN(!MLljHUqpCjtgSYErGZ8TP}zD
zf`}KVJ&0u+$7|x8IH!lP0YxH=#s{wZk83kNZ{10|hw&lwXS>oN2jYpwxq?ZjJ&oBE
z3%CFTz?ei}MDkpi=fV$eV1I<a(0s=R_1?GF64H2_40zQg+woe$DTE4%BdK~VBispa
zap!9ZYY9!qU8;&85G(^>B~(#C%x#Y<VxWkONF(^A>rAxmi!RT?B2~YZkf^28YF<lN
z--lpg*lMG95Q(SgYYAzPnOS|MYa}Dq5M8z6<=Ch%NUZDYo74k*#i(OqK|NgIE^gDG
zXSqS!RSC1H<Zw}ZEmNXojf0_~1|tFv;@be}n2Wr7RD?hqq}1PtDtLDa3OsI-xxY$M
zB%{&jeRwMI9Fqcskg7HvWf&-U+{syWgIBp{_PVNW&SFz!a{&izO4+VOOZ3=%t`3bK
zP%t_}eLa3i#^Y#MV%gmL<mPLc9~l@XfFpeBEx|=(P`j;x$v#7HmzPJy$%-vhX6k;j
zFt@SIM|hvl+Ff<h+;Ys8w6k!Q(35+Wwek><%K;ghKG$EY#N5!PAQg)8B<X~qt@H$;
z9kh)Yad4KvHMXe{U;VPjwp3Jr33eT_-LIUjX{@sB!E2eb<DKg94mgw#qpvY+I=L_7
zyEPdVKaM49!*^*rN;@#F7_leno+JMK-`Z1wvHlC}faCZ52kd!FW{ZHuI1jvdEmKo}
zQe8;4du9g`!m)L(dfRaz-yd#Dn;5pL0~vZ(1WYI4;LiBXNiaWh;CCHJ=mbn;JKudE
z-;~XM2knWn7NVscNVs*9cmBQu`TP?H4z|~-59BNL(%>M9ubQafncYE(TJ^+`oXom_
zmb<~tpK#F8bb#VG$<-PTB>KS}^r|&#*x@S^))b=*Tz1qg#J77YqH|Fv6^^w%Mqx9{
z;&nAWr)X~PQ;PPJuuX`LV)ldM=O?U3P;{O+S+w4$%FD&EHzw#8HuPyomC?hYCXJDa
zfWc|zK{**K*>lFGqp?5xzyF!ds&Ubo>#M4XhTCyi5KW9<_B3~PEdV?`OvtGD=oH@R
zdVCys-EQckwX^d?r`^<W&QNoDP%z!n57Q`K5I%85-d1%IlVT}-M<2u?@ryJ@Oo9Ct
zW1BEjw!8XjQ%amAZ4?%Cc2Dy`J4Ln&{oFA)W-GBSE!cOH3Sit;59G7$?p77s>UA}l
z!h>`p#^<?;U2u$l5QWWFY|1chdl&5fflQ%Ns83GFgQ!2Y{&)5;mWXKfj|3wO{GZw5
z0E`X!#$5(&JNj-u=jJU>V$X<5<aGUMK#qR|Mt_ONc_zrQzq04*Kcy*q5ny!U@NKVW
zBqRG7fkt0Xcr}Ukrt0;KM&pQ$u||45->R1|w%n!C#1^<ZYWM3I>D_)MMGD%Z1_TX)
zhT7iO^LhYb$`P|K2EaRS5_|sm?z=bvS5p&R62*q-P8{so_=%PI=kF25S>5ZIv?%?D
z`>)nVPyLAZ&ssx({u;p+8e_^GeF$KiR2kACb!*}EgzxUKWNpzAUZUqP34T4(nnT0q
z+~~X*eeuZP>eVEZ)q!IBcSnUXG9T-S4q;X}?7gunu%79CSbjZ`Kq7EJ1cstIK4W0O
zs3zjEOSs!6qwm(jk&;ada7>?nW3#U(4yB`HS5+yc27$KCMH}7}<VTJHCTDOvvUz<M
z*dtgz_+p)h8RAIa(4XUj<Ve<^1aP)nu_6#@&rNb$WhXmZ?T)4bHv{!p5a?LMdmQ`4
z00Yb$v%4ysnqFIgfnAD1SU?b1J-Y5<S@qCsc$X_$ydvJ0Nv?{Iot9wcA~a|{mbMR>
zvMbd#qTjK#Un2T_{kvTf6`b@a8-(s-#vHOOt=(CS`qf2&;0@ePezG1^02TZ2FjfzG
zh9&GHee>Z0Y3z`+030m1lI>V2th38l7(7w!ikfiAo(90#MdO*KgN^mDJ+CZ3YK9lD
zClP`Z2X5D;@2^2Oer{Hmsh9NMQ2Hs`QJ0aJy8yLmRb4t=L_S3{v$HPaEN5T2i+0te
z!|#FYY<FEI4Q0PI_tYi)_sa>}?Y+?!1p?}2Vp8vmu{=KKgjH)SuF?13x0<?){$z7N
ztHUj<{`{7O8>v&}(nRu*RZmc0x!p9>C92fi(rS%$83VblhtFbLD2RtiS7`-xLSh*g
zG*t?)$+z~pOeFLMbQygy!0qtkVWohlnHY=R@zu!@?1`ly_zKctjl(Sf<tqn72qN6t
zXh<EiPNT_8VKge>8tLSFGJbHgXQ;?@bU%)pzRd<?$CKp!s%A1Uv_`Ne{f*jo*seu0
zQdsxdd~6A90E+{Mcs+h<Av3&FIJ=<^&?Bzc%@}BagexF(Ou88pp4(B1q6=DeCpvRJ
zirgibJp=?P+~;DnFK}Jn)s|-g3RGz+%J*0Q#O`TV$(Ccwu_7>;Zi}-LKQv?8kH6lk
zI+JQ533JzD?EcH&eq|f_Bq@}?@oh8C8{QBphx1lt<;Zww_cgU70d+gq2Lk2b$;0+g
zdkbR)DR>mU;9!6c`Z;?Xqo$j9JK#i<=!^6uTEM3xl_IxXEZfwK+wIZqGmXK~U%O_{
z1sCUvSpCuaF=&X{8yRU1y1N5tx8sd`{QV%Z166M%)cE!rF6cKBZkKpUqT`8*Q-dhQ
zxuD~Rk?FNxL(2B3IzZs_us2Gfgk#(j``$=|!`;b-U6Ee>MkWT$9%rp4AnU9+G}T5A
zg$bOJu<uukBx?iR>fcB=((#klAdou<f7kd%A~(42V-=yl*+2{;-pG^*0~6N^kZ)P7
zzI?UE+EtSr5Try8KyX}e6Gbb%k?2(CAOb~nJpp%{do}i`kV$u<AnuK?4)~37R6Xnl
zbVHhuE)N$%j0%Ejw1Ja?fD8a+z0o>i6$%c{st;6kkk}Es8jFITlgkn2DxiA+^J`I<
zl2l+~s6tNVZ|!=3)`nl_jaU!pmq$0i!cfrkO%Ys1e-f?GZVQ;c^i;p|M#yAfEZuT9
zMHXGq*0KhpGbjmc@r`_Ol(DSyI7%GRyMo1|pxKfpwf1|rl|)_~>om%BZOfJUPm$=-
zRsvAZM-S>R*=n3K@=L78b`=LN;DLpDCsyGgwt{37+SJHLCs6WkMRhm<IC37vn(Mb;
z;V7$)U1ULs{+QI{lt(^!6zg|-h}huzLRs;1qSI~11PJ(f!kz>K#lZ=u=ucx3c+_}S
zHH;>r2{|mz+4E?^W_e!87aBnEJj{Nak;T(&+KwOR(|l_9f=h~ssTegn-Ggdpe71OG
z++9CTNc|<=I=iFfuqTt;^W%*F80W%9+uk3i(+<&eF!n`lM8=n_IzEd-m_^t8IKQq7
zC#*I;_P=IzKhF1KC3vMHtSX@MgL`lIaY9rN%Vv#&4z+|eX>^WKVW!^-YJqzL*wUsV
zT<OpeSo@C?_Ls5b|8eF>AWIysl6qTP;m*|{-rjmtjdPqL8d~p<(;;nfF4^;mVFUrj
z2G|7nfnlG7P7DP|Cs4VL#tJwEwaF-K!3<<NzKFeJvjLd%+88v&0<Ha(%|$;h^R;Vg
ziN4aM7W10IX`Z*&)rgw{9J&$x<DhIeH9J=j*e(5i<oKZzc3ZVj*Z56zr?ToW_~fP1
z2iy#-`C`E5CW%+<Zh(ZRAQEJC2@{XFN`H@g82IIw#_<!j5)cu_$W~Rv!w-<w<75#)
zsZ8mHW>vHQN48l3@oTIEzOpR=Lgf3e5b%6s_oF3G20tth0#biPkd#WzIa;Ie<&Odz
zlIU)X)A3kS!IO5<iR8B6x<Igav^p`(uC*2PukqOjNw8-EvUw5)IlxHB@H>0)<9vH6
z&IOiCZzcr5shN-oJKoI2PHH>`N!6PP?S=<OZ0DO9gE5cfHhnW;8Rz^Mcj4|g^AYX`
z#9`YLm4`f__g;-c<%<Z8)#n*v<yEVhlm^_nHR_b-3vS3(`)0y0<PgJFSNVGihOfTz
zH*PT2pubNz578JPt^ET*oB|{)^;z;lfYgHM#M%VK)Zmb{znM`Tu@m|N7a$ei%v1%F
zD(-(Psk>xx<?L8@bl2tlYLDuB=mNO~^#(8{7ND#zj)#i_hcxyyhIylJe)qrW#kF+<
zDen(cb5B-A)URLJv=%YhSKI8HiF%xvw5u`DD0xoj-pqt`>vt7V$+g&3vazOE1M}($
z9|R!39_w`Y!UPJB8|n|#`vXs?o2u$YuHa~;=nu(;^Y+Vuu~@qkqkt>eUSMb#<rlHr
zT8v@F`k=&iR}=W)Gh5OYqU_^g8{qqk5fvp5+`LhLYAagvlsE*OUX99ANq_>LtiPF^
zH@E(43bnoT{of|;dE?FWvQ6QQD}ML<{*7&Fa3lo3vn_S+hNABCdp~ygh%;snRC8tw
zOS6Z9=`s*Mih1>OZQ(o@vpzU6YuizZX$cI~lPK+i+CEj6n23>AE=D6TH{vNZ9KM*O
z6j<=bc}!xS|HG_4b29Eu#76qViPvp?M&U7%*ExG$Ri9DQ&7=v7TA%T_xs04)S2Tdh
zaXFdouFppwQIQ*j?WxbGewat$-uiU%S`XX4AN-E}Q+>kL3u}v9t1$+Ttg_YC=UWzy
zZC!nwP?w*(W}{wRzV;{9P@m{cD<-Fn>Mz+sH3dkBAw9TBA%^zEVr?-Tv~|?lqi|sA
zq7~H-?9n+ZML*asxIs|23eU01))NcGQ;wzVt^aN}sOdp!|CeJ^jv=lPQfVlrl+$C_
zXe_IH=9D3&lL9gmnCbfODpoT;*gu<%eoHL%Reh0v)D_Ef0sJY1*1W2(1;88t*zWmQ
zhLVTMx*i~S1`@aExuJb<cUtnNM|V6oT?T$jYk?Of_7u?;;{_Mn*&US<!lU07Vj#Fd
z9O$Aa)VnU&yc_EwYcO0cm!b}u)Y0K9b}vT2(>)#C<){L|9e~|fi66MvF7m@_eGqD7
z8*WyDDk)$=#yVyj(FRo%?7eO3L`{&rcT0bg8prO}X9hoo0~g7D5M6mbUwinnD{gt{
z!y_#L;G^6P@I9ih$4cDRA%11QJ&A^YI+J9SPh+E8L;&loeL!9cem;*aZ+6QleN`Te
z_EyGb0*xi6e#cuGIR^;*#IRNILqD2sxPzT<<wYV7Fs23*<Qv@eR>okGm)q`WLc9t$
zj@a{7!nzsm(7mdu5ID9^H8GtXoW`dJB!8>IF7)A7UZZ9`UshXfbm@vKSi8FTp$j3p
z8(?TEhybz%yWy=&h@9|PV|=pPm>+8-SO&PQf}lc(zxHdySSliTIw;+)A|mfQr76CZ
zh*QlUS?R4zX1M;9=crpC@CZwBS9l!sXa=JgW%jCNvCqnHB|<jo{Lj!^nRv3tQ1{V*
z>JXu#VpUyOHvMwOU=VF*V?rrF=GFK_%{gXsZ^gy$f&Gc4R<X_^(}zHL^*`8r3^zz@
z)fn#IZhmDq;_DnY$xU^hVncx_w=@OvB3-uI0T&Om#80cxT~};De;;%Ioy9n8cs`Bz
z4DPCwchtYyQgq|*?Li_>sPH7225mWjDPh2}QrUz9wi@tPU$ymEfRf7{HezYpQ^?&-
z_2=<6Z0oIjxg|DWm~^=Qs<^KnJO-u<v>pV&#6R$iXZA4QKwfN*)FXk4kUIw?XbZQa
z8nH<2iNO7Ezda3*&Ze|L&)&*ZqGC%oj9=69Xxxqv?uAa!6&PfDJDnQ@gLu}x9Z2Y6
zL4YTY%-b26UFGWY+4*+DCgpzK^>#YCmx%4Ais_U}+qLKId<-Yyh9_)qd?V%OYS<TD
z%VdpEjBjT;0Uh*raE+Q=BqU}s4Xb@S&rRnJtt^Qp-_A(G4KaxQ0-4kWBcX!f6&v18
zbR?<o;J5V!ijE`OY>FwT2XhKvQP7eM&vGW)0_LGXPw(FTb|N}wS*>DpbP;JsDJo9D
z0TSd*lk4#OqGKt7T~E|#8yx3J+gn+AcUvf{FO-CCl|B?T@g4w4N8io|%Y`wpAHPl(
zf|G&;VYN-`&m*NzZ1(MZMf;IQCC{P|LMB%B7<Lxee9UQulHY_~)9~c7axb0vn3&t=
zU5_t<z1R%_4P7UIOr@1mrhm3uZ)d{T;o|Arg6Jsb;&)z__$|E<pj0xMFRD)*HgFga
zSnSTp5nK8}O)ipYyQfkjnQ&h(s~J%iaa)g?Xf53WbhQGfn=Y?c;QXtkXWUR9EDjcR
zo3YZ#oyxXUh_$vOZp{49AZ~VX$_O1bjgd5V8!<7MlUaMDF)ytB1DLD7F@{EA6fG0O
z@2apTTAe8(XFZKl=$sp&FWa;D$(?4w>Pz-K)}{J0d+~DI6I~^<ck(^if+%N)zT9R7
zIaAd;>AcqLf8BPzld)TIDV@#U^-e||LjS=)LXCF6lM$YHRq!gvs@};vJLx%`_Qr7T
zbi7Yr#mih*NSEKqDDO-aEOj(ZQMYSV?Di!tZ?n1>9!(&kABL}w7MM8}tsy=!9Uy%9
zJBi#NxovtUbLZttR?sZ_5Nfu)lSxPmS<s-^?HZGK2BRn-#;vHjYyS_LB6TiDMo{*h
zjAk;4Ny<VTWn-k*@V|@hk4#<Y<5-k6|M7`${|>YMp?5NUN)H}Sf%z{s8b88V`X!zm
zW@GBZM1)OdXVWp)Lw;;#W2nROjlLR-J56T65u1w&XJ=9dm}~KaCwRkecRpHRzDbMg
z0+tZNE|IyRm3ZZqSL|jC-Q|{tx8kVyP3C4WNTk&k5fgD?>`s912BZY6CO}236B7AF
zef%kYkl06Rpc5xcwxnLb?{*k~d(q_6OSY^JiJDfT-YFJ&HRh3Gk=uHlQQSCOPh=Z0
zfm3dBuo)E+r_5?@35Is4$bI#VP|aQ8A4E%3HH2x5f$`k7NAYzOuI=$VnY7{-MvN;-
z4Y#9%LmsdEi3UP5ZBJuAP=2_{;IpWj{6n6{v^w1Y$X=-F7hlwvHDoNju=Yz_NqHkT
z<bz$nkr*{?RYO8cR2Lo$({@G+{5aD^y{jRih%>U6ZFfV)bhn>Jr1OpKQ3vcrvL@QA
z>U<vKZf;0)W7r)HqS}RFywPeJGKwEenR;tu+>1PJoc7j517ai&SbYGFS*t%J8r1jY
zF)~<2)x#GC&skGL#=we_<EI4RUABT6oWr-eO@E%~?67tLlBlK|ZWkNA+ri5hYez6U
zfaKEVcyy~-5htUb7?SrAA>5)SZXxLX%JJ=*LU?KQI_7uUXhYh6_iTpwO^yiM){_m1
zCY@@v>4r>*i?02RuRh`+UX6awrm(PcDup6A#lIFn`7_t%qcRo)j<?zx^u8S58yfHO
zh}~2liH{_&b2RGWu}XvrHEMA#!0uGe!T@g(TZpqzbaM$8?JmYH@@7MNOA>Z<4mohG
z-BNV!_=aRU0G_(S`Yy*FBTnkWB6mfti81O=>(z#Ap&>et_9>mvbwQ=mzY#-BT^fef
z+N`XQdmR12wqn2$pWJ=5<ggOu_JKamdq7Nk*pRQeiGJ6t_;dS_HjdhiPoQ{LG2+}8
z1Fqe6PmpaL68A)XqS4@gk~hOs4dA>KGB3g9J+aHrW8ruJCac63>c+#z*}LgT65#<K
zp?5PzLla35hwH(+>9{0`A5kmx1&lm?ZP&XQ%i!tJuI+v|&#Kd1-IAr@o@g;Xh$ZE_
z31g37Te0umd|^Wx$E-T)ku}9?qGz78YhPP!REBSzu{u>wmS^0?QMozUNHpjhIrqhM
zg*3jKi4rO1oM&p%jIm`y+$D%6Fx_l>H{aGHJbpD|Qc|)FMfHgR@w8;t>H=F#@K$B+
z7mHkAz<N}MD<7B}(<-}PUrYnXo3ahXY9FciSdB*YlPKsX-^~}bpGXXN{g6)wl*JjF
zeK(<Xfp}2SHF*EmHWzbFu5L)QG?r^C(kuO*%|{jX+6C>`HF$E{CqsLHQbq=SQ*%PD
zf_G31g=6}--Hut1Sqp6%_ij``^8U3g#IZ;*ER*1KF&cHb+XvE?yOoU~GwG7r`vzj$
zZueq;v0=yIdpW*4dC?BqiXi2>@Q1;wzKZa6^tv8Pm}dQZq)lJjhL$iLaoY@FiTi*D
z(YyKj?0o4qY4`POB7`2BUD0J9R5scJ)I5yUObNn~UTZ#IepzFWqYT6Elo2R?AVr5W
z=}%%)_rGsX)zY1>o@s=&dyMcr>RiDpOLfZJNA_OELQsQv<vZRJj$!Fi^<Jj+ZY;O+
zy^L|FfpJezy#b;y?T#+oi~;NC_Yz4!>&b)oy@Wky@d(vpd!IUW(FK=6^?MoTi9do{
zpugj4)ZA}yt5t2RtBBrdbpgT<DIBf-y@YWmxeB;(1K2Lt`@R?I28>d9NdMHCaWp(l
z0V^(|Tp9)SF@=7goA<qh5A7}1{$8eJ#K?qm#obR8QE$R1mI9EJk)u00Et8~%`}e)f
zh;Y%!PP<oKcbBZJ$`yUpk_~CPFr7FRsb3bDH#I(fYBVxn)0K6QL~~Ya-%4*+)w;R8
ztJUTLs*A2?yQbDfJaJXv<lK|mT-Q{HzlkmIjrTH@fDa*VJY31L-PEo@sFf_hZdHKx
zY*wxqxE<@myFcA?chu>*luvXawiVabK0+}d#b1oyltRuI^{$}5Fi1M8rGWk@!2#`F
zfZ%Cs*6!G{mf#3M5tzeDY<D$D-&X@1UKQko$D&d_dlnzti0wc)N?0d|)uz_wl^WZM
zjwZ25xvyC;@nghgjBQ1W%?us}NEMH=N6~fwdz#15_7t&HHNcMuaiIl10*m7D0-bZk
zY$LkDG}vRuJ&V~FNOoe+<FgH09eWX<m{I=J?EQ?2V{5;MZGJ!DTgk29tKQEC7cT3f
z54H3Cd>E%L6Cux^=DXB>xWe7<em~!E{-SfP`TdNQByn;#1D8KXHz^Mdn_~OkPxJxj
zao18*zn=+-OM@_p4oF7Uev2T2Jhp-pc_7Lr{SCW;^~U>|Xc8+D2GA|SL1^{PR`g*7
z*BDbq-h=Z|d}82DY~=VP&cn2|#V2WYaY72vNvGXE;Z8lp_Y;*P0H~D$Zq6>w5ZzIm
zjUl=^>)jLU0RvQVXz7hH+7JU+S<t%bl3)x&?<dOS;eKGLV|nq81nF#2gC9SE9kl)&
zgQm>BpNP+yFYT(j_RzsY(6Tu#74kvu;rFF2b@vB1r~L}c20uRYssq14J|GQ_;R!N?
z8!<a4u9!o|sur31t(Yrv=EO4E?SS^QiyFC(1r7Uxu%5W*>Hry?--U}Bi%sgSvAZ!A
z4|7^B#S*!-sojg7E9~HMWzGDp@s+KpwZA-2aDQd1D#bw9Eqd286iKCO8!?TG6PIl>
zpg^3m73Z#%$ogyl+-LU#j;^);I$;mAR_E6i5N}nsf>ZKHwFp5-tkui*SR=4w{(|0q
zKT}Co6;}|f6iGB4K#a{?wx<GO?H@eG;j=i>a5U2?`YNdST%bW{IVQ<l;zeu#CyS6C
zSQA-eIy>Z0W{4e)>716~+^QNgmb&PK<=xqs(Ej`DJPda=CW_#JYy6eHyD?J%NKV4e
zJ&hSV$&7~BeQ)&EJW6C5gk0(^#Q>>RZ;pCM0j;Jn^8xOETN{NQ?cM6s4x@%YSiL}}
z-TptehQ>^H#`}nH<Cd6>FMk?@kMk7GO@h=#-Q3X#&8!fkU4q86X#tA77usVAJfsH+
z2p3b}jJUB^Np<@{$L@fntHpX^*zi~D)gUBn^eA`bsC~N4h8pu$r}P*~?KB#b-ug!x
zOvcbtQYlSGIg7%(SmjdIc__-O0*zOZ&1u2#g&~+t>{Aih9k%8+AHPh^xH%~?lcQ@q
zak;M>6PX~A%FW6g{CQ1^Fa2I!W48l1q7=FQ=#B>WJU9z#1FPypcu@uK@Bh&5#;p5X
zn&7^vB-B)S5uw(-SfcZ|CdC%DalNi+<jYrVH9$S+!Cu$n6V){#rUEX{(!rtKRQHp}
zJhs&Rr9RIGbw9>x{Zo6OPFzUm61az2y%Z<(kq!hjbzr0)n~ujZ0G7C5zHO^AtQEFd
zQHTlvHpZT6bmujm>Cf5OS$iH+JB;h*i})JFZOlGMm^nP{D7NE+gnd_B=2fM?SvRiR
z5AxlNxEVHCMI&xWx4Y6FLOf`Dq8{i>cH0j^EaaKZf9<Lv4%$978G%DWb)*l_>nHD7
z%?Al7VO$-1Yd^^7eb}FoR`)@^Qj-#(!h6@N4ILCtbXY?`K*(rVW<iR#CQsZCGVWrD
zT=DPK2;hqZ_-!9#5(6omcITvaf!FE&8CKL#1sAWwFUP=xM78eLNXZeXCxGDQzi7QN
zMf@zv(Q0b_r#AFKB8BHMF&+)@(4i@a?c@iUI=XfQXr|SZ$oFh4Pm<l(RTU1qx0ucS
zpef<3iOv|~UR?X7&Bs>@1o?XObllAg?1tu&bR{=cle6uD|6BU<N1un5(sCspn~lYQ
z>yGAAhNLb;r<V~Q;(4!5-Ezq8#yU%vipx;|bPVjfr@BsIZ8`c$e(<)U?}{A@W*S;-
z7`Evl?tpbQ#TEhK@&}pat7Knf=*<`!F$%|mR6tQn@wA}oc0Ybg=J@F`@IXBivu)TO
zzC78kPbb?d5ACn*k!Dee<NR2QapHKZZEJ6khme2#iGC+mwmsFK7ucX@acGj`*Yi9r
zN!(tjgE3sHF{Au2pJZM?WG@3(c6^vI6wj0kRu!K^9%gjshw*ZRN&qbAn2gYNeV7lt
zOBP%MyZgh0f?@}3kIJ30fx2#@xA((Ds;4j7KGg*A*bA%vP$!uiu_ii2d6D?++7A<5
z`SQ!3^jMubIC#P81IRWvlx<Ld@Jc{ujUQ&@ClDK$+@${ZpQ*J%4E&=He`Ia37*|LU
zi%baKIDSh0w(g}={4k?Z^C%>ZV94=9daQh7-2qLqH|!CZ2oHK=xDtJr^?PTR4QZNq
ze>rhm0Xs$mX5tF+o|}x8Z3u^_HGqeR@Mq#|tdhHn5q;3EYKl(k<~~eRnYI0_$F2pa
zUCuV*tgn8ZNyb*<dH|002<=dOPP$0?W_-4~A=It-BsUkzp#WlvjOT;+=-tuU;lvVY
z0TyB!7YEPT;>#xFadn%?yYY*wJzX2T6r<p81v+asYb!3TU5?(mxPDe*ZqPi~e_M^F
z(ELeTkG|1}yDZH%w8$yeg42i1s4+NBRw{nk(g?5vaEW$bXOHYA*l@&XI1{>o{=*n;
z@&YdUj{-LCdV3tTv3mE{;C5w$`Z42(vS!cdja_s!9iJStXHlHAnFP9tqj8v5Hvu&S
z73h4y>?1un@#}W(_@iLe{+^TNjL>ha>Z62R<Lf`Oogd|+aeT&heU#B2A4#H9eUwi*
z{Jlpg*b|kT_S;@{;`D#tM+qU1aFJT|N12)=<&Bg18bQSIJ4}UIwR`>516HU0{2A6~
z^&jQyU29NT@CG$@7gkt;Mm7HO%P*}dfIUA$4$?w=Zv7YR5ZXS<#M;j-z62mpd(3C;
zFAy=1WL|I+Wx8yoDBSv+#4UDzlt{)Af8lw4xSp8BRNsvC>dTF;F@P;c*(5U@(zH-7
z!1#;?T!cOuv&r~8>~hNKj}rasXt7y!;rXzE?CM9El8y`xro7K{0*6K9nn9^>?d3W<
z|Lro5`Pl1(%Wgyc2fH52>Uk<X$3?n+Zo~k{lwr!S<IvLLv0}FZ$f+Te8@DTS<%~np
z@5CW!J#GuJYVgY*Pi`@Oz^=KU8Qu*zg9%-V?nzP)J_jhSK=!A=a#X+-6<bPmQATEh
zw^Vv2t!n3Ri8ycT`Ws&ZeEMH*1k`F%CB&IS^BaAl3BZUl;4Jl}sKO88z;wZ>pjvk=
zu-Xrgp{98hJ)ffI$59)j$gBZY?L+>ac%t2ME}G3*wx=<A{{*f(&th~UZ28fA9=qFu
z41>UsG1_T-H>@e)*V;eQ^NyyBu&3}lYRVXz0#hv6&Zh5jJUiGpY^Kv@?^1it76w{P
z`K5Y{_uwA&hIbo0)!vsy9D;rN8eaQvE)l41N(c3WUaN^QNvv@Ju~)77aKjzs&ry0|
zFnKi9H|3iby9<n58w3lXYI*EN8Ae55cT)gR-SZPGM5*hP5Ir|#oInsT?eD(!7%Zh-
zgr$xjItFKl33OL#$`lPB7$VO_muLGBA6t)#C&{egraQfw2@`>imZL9JGGvpmAwdUg
zh_u^ARXpU+EcZ0GNi|;km7Ca3H)Sf=Nl0I;FA-j8;@Hki*wvWmL6=^|S3j4?HBFP}
z{&kMe^7lWpd9@sL5%2YwQ2Pvw`Hg6b376e$%0#aPD@7d4&;Z<u8WU*CZZ{>$k4LfH
zQLk(YZ{7e`zUrX)Vl36cBo4eA14CSvZ7Dv>ykG2|M)3z3G-o*qTQD=TmH5OuXvS6r
zLxE=NS~H%acvo(y4=kbmcHGn#^zKx1yKM>BE|0@}KOj2-t{{Wf(L~7gP~Fj{G-HqA
zxUjbuU9tCA#fK0pOxbn-)i<^FD|_;?Q%_89Ph%<F%0hYSpXD9Iu<v<*cN{L}MVu9~
zNqZ1oYtH-MHaLu#ACz%Nb6WaAXi-&j+NUEpl>60}aCB#L#uF49q`lqMoKHJiG)z=*
z;8zEIMDp9yoR2lxAc(l!oaloSciX30m;5T$&+s<Hn$yo`UVYVSn=?y*Ct4<6b<LTQ
zkgUlaC+Y(pcRoed9PnUpd(M4z8&w0lDVK-vBGz#5OnCe@Co1EC(cH%&+PW!-Z$-P>
zA|?Qx_$~^(X<{0b)Rax!zrWVH1B~V??)TgiU8VH)EM{+WqD)<5R<3k{&(e@ONme0$
zu+f+tvB5m??__Kbm!j5xI_h|+W=tbz1CDcq4Z7N#$SyAn1Wa@KaNQ`WuC*5~h;k!?
z`RH0^er`J%yF~y5BGL)Wj^=K=sg7J$*>AU0P4)_yPBsTA_h<-7+0X4;fdi6)pjj&p
zpMKUB^|!lC+TB>Raag0vwiG-5{k8D(yQldPh2ABl%K;-oZW4T{x9@)eTd^7oQGnN5
z*K)UAv5go7pEoG`rbfvrAhoPp>K8Hq!;R6ETDNxCgP5M1*}1^T+3$xM%hhlY#z(Oy
z{xreH_pyddY%E-lZ3k$_Ipa^%r9Zfw^{KRZDrDn4q3P%=V?a1<&(sf^H~frHTRpFA
zsH=bMh4$wnr_Bx~#2Rz^(H#df&b5#d#i|Zw4EKwc3k12=L7fq30Qei#3Dm%~doW@D
z^x%l?(VuGzW42fQ`;}t*4rUBIg}n(EaJ72iV#M_c3|3tZU7yve3r}QI7tLV)3A<$V
z2Q%St)5#JR?hR^*J_@^M{XKJz>~Bqi7x*~Fg@eHta2(pA6V$aH+t}I<=Hoa%=vfZX
zwQ%<r4<_U*v{*^)J-R9Q*IfbDBPR45%#^L2RU$-stc)9gTUm2CPt73#gmw>rM)i9K
z52?vm>yF`Rg2{69r(@bYOdJ<x)t0cgw7nWb62gNh!*kI=CAugv>keirNFc3b4#ZV5
zAH6CEv0YcM?gz_TSP*$))SEE}oQg*6R<vVEP<y+rA6$Ck4gf^H5=1U6Nm`8sbxdR>
z80^JZ`*!f?uKI?oI`y=r*a{v3Q~uhx7lT*VymG*nW3q0YFl{SQ-0E&0tNJ{V%%G5H
z#ZvI7A#WTtHez@Jz1U_{fOue^z(aQ{*6JuBC}kRE_ca6X>TtU~2pC)&&P}ZOC&IS&
zTeOLfR2$|U_UExa2>6Rp(6&Cr<2}*2!DsEK_B6mc&oVsI;tyjo|2zts$yM}1NAPg7
zSxY`>NLx8sJ6iIQ?uUX`wImEIaGrMRPb7S{t0f)a@{sM0)`zeFz(k`Z9r$JfmP`zn
zJK@`$QtWNXSel+QQ?^eXjDWAv0a7*i)f$y_wOegVA|nLe?6W%kJ<@lDz*f<s1P#YU
zp(PQJ<}zAQYi!AQjJOtaSlDCDr(y?OP)CVuw!N%7gKmv2Ygbn&B$0L%TQcb)@7R=;
zG%O-mms2lUcg)FuLMqpzfx4I?fn$0#4`Q^IHKdaS8`1!5e@nBCMh`q6kt*3_OX6@S
zb(blRo=!dkwU`*Yu=Y<j8&e;IOXb0<EqV3*UYpY>JfnE!2IR@YWL}{B`+slORsTzj
zv2Un8A|`aVQGd!`Ew|#cuS{>ZRg+AC=xkdu{hr}bNrqsq)2QugEnED-q&$bC-Hr3q
ziYez(%mBd^+LL=x9c_(=OjwRh!KjtXjc^(Q8nm?4sLcz&qxi2!=Vu8$wGo}SCu^=v
zfi}rB{9(GKi6A^iA{A8$jf4*HftDHbNWV{GmTasm5U<;#n5Bnuv&S!MxG2~CinpUi
zo7+M^3CJ#z&-Q5r5`v9*Osr>d3|?up=TUP6Y2XWeY4T=;jGUdG=Gt}?Dwids)v5{!
zH^2O%$96_N{AJ(_cNOwJbzn~DB0<|-NH}}&i0#o|%@=KNA)y$gmX~WGANMJ)VyiA>
zMCwpj39E@dU1vs|62eyr91iRef=Cwf-QcvE_s%zj8L3yVZb#_%O@kUhLs_E6LZ%8X
zxkWSIq*4+t(sv=IL`)YRZEb@363M#)kY*I-+$7wng+z7$q!RlG8jW;H5Vg9qYsAaw
zE@UK=aeg0ZaC!oCFr8za-cr4>ARWWphUFOb(%J%qhd0$w)cE%AE)i!VYH(<hgP%=m
z5^gr`RKliXAcXRs>p5M*F;kaV6IH#CDR48l>zwuoTNo^_Vg$5vWpt@<&asGYRl(W1
zu2bQ6+zmD5P!HQp{oQwt^L;C}?J(YgG5Qd3x43g-^eGT*AppkX(P@%DCl)mXe7W7#
z5@5Fk`+;RfK!du_19set-p7(p*m6`rVCb^umCBy~YGCbGwi*jMHGz(PJ<i~z!BN<a
zLEF$TS=;5h-a;0D=AEA=t$=Ms)Bc$hHp=d6UQ-esJ<za7I{@;b=JV|@uy$50$f!T3
zAFBY{WZRCV?<-&Q=PKeV)%9b<)SqgBoiN+8sQi*yj;Y}q4!bP){3fX=b*wexAJ~e-
zsx=(~PK^_ns@8<4_+^vIBq|QMal$Trdb~@wU)$Z9h{71BWlw7+0v?<cz6e9-){NEe
z7U;T$eXSV@VOpFHRBPZ<yv!i@H8I%fzR?k@ZOzCt(SOmeOmMGGu;DUAnm&z=G|w6|
zl+5Z*4dXf4sEVIBtJ@UAwn9yK7Z(E9i`+Dc>uzh!#IFTW1kc~rMEpSV?{8UAAB5fH
zsxDQ=@=f!MRj|^U#o8iLZ16ieck75mF?yr()^nH1Yg$$VIv<#xwV?p-6g(`m8I7^q
za06CdWMPgC)9Q(R^3;ulozxnPuKm)kzFa+5gUrPkMJ%YUX*oQ>h0RCFIH?}4S9W^}
ziOY@Hu_6)0Tyk?WI>m$!ZBnH&)WmM9n+nH!M<w6?ISCjRVnWFdc+3_BB_?!sR~??J
zoS>zE#UEJrRIeAs!?M<ZcPkh7O2CEhuq%RWwKZKZ$B)>0tSz}UVVyPv2ysXM#5Usu
zRXl7tL0hr7FogFtU`xw!d!XO3Zb$n2Fs2QKMo=XvG5Pd|_+u5q4*N66wxh$ItM;U_
zKo`NCr-F>9@b7uj|CK$9xpqtp+4F#{#be)+6%N0msBsK7-wvfY;+wtWP{L2&aYQ3k
zhXOB>pwpwl?mUz*(UT7DQe_TT=U~6>K2$+w9x;2=;rOWSJ(Tgk=m&V;p?uiSyl>Uf
z{t%aYO?>UgEW1|2j9nbCx<eV4&%%NNLVf&Dk!v=@Ak%BV!Zbr|N6=O`36Kw8wL;Xv
z*AYXh_{4Vw7C8n#*>%v0hcbz2_PEjh0w^8IS24a;$sfi=6x}f$__Uq?3(2PYP4>nQ
zTYt~3@RgMXm@{KTTH513)WeL%u*s~`v2Zee@~usBk7<o5Hgz_3$Tb;omazL*H44Gm
zZ7$b!IrOzdSyhjx8Vo_`f9ACo6Ql=o2JCvQMOWdN-8huVRU2uj(fS_B{18`Hj--4m
zR+G#Zb9Ou0xa|&tD(ym{WD7bZYl~2aMFA%b^<7nYeZMWm@HPMQzuUd|8Um-6V@0_u
zmx!pm62<39p|e_<E}~3)e6<WS=g!%NIzQcPo3V|c-@u@oXfoq1LFFOg?7r&O*4l&U
z`xr0rhw*iZ&L71Lc^E;W9;=6qzr#3dJ8Je36YEKQVUur9V?UA8xh8?2?5cox(wW8Q
zv63!Gvlr32A33lcPPlinwaa!K&R3qlZ(U7PbvR?1+XmXs!x?K%L)oq%gi>SoDcOBE
zvtmc!6ZWVfVyhuew(X665U$MFzNp)eF1k9tE^yLnqAF&>@XacmMTC!GX5Ha@aK<K_
zd9PQgE1kD~Zw-eNAyC4X#>0tpaJ&#i>F=qLwLe(laOQ$0E|9uw_=@C+ZHF^@!jil7
z$ZxFu<!te80KbZ{J{(glZMaR<@vxh?cOOm!@4eToCmPZ})Wv!QCJz&DRMsRCr%R4P
z8`2`4XN1vMiV=>`Wb{K_H$|JStm?!SP8HE3HD9dES(>Z*06DmW=hRP%do*F!G{6{=
ziTPOF6l?Z+d?M$+WH;0YQWfa!%~)$J&^c4LVui{)KgZlS;&wFmlq+^eqe2(lh;|_g
zFFHfM7@u1UOy#bIa-!5{OVPEP5Mb5(&vq|X;y8%19F4I%#Si~*BAnb!%RgWXS2fL3
z<dj>lY`EL0Z|Lt!Zoap9I1`$Riq8q|mWIOi!x`E80ZYXh3Ofu{sJvbu1}q2t)_)ZB
z{Z+yH0S8`<`2Ek(t!)daBc(QbayZ|7Q-dg^(u3w{bk^?I_%pQ}U;EXVJ=Y4!rTRr|
zhT9h*IBwIO`tzjy-O-kNcB>g)?I>HT+A^w~W+Y6N)#BVBxr1Sa+cK)dQ~IONcDE&b
zIZIet+tZfuydUp=7JRP?l~ukiuc5y!((CPCthz1V!92ZdqI37Oa5_>O9aiiQ>e>?3
zIfZ(@UVl0-*$~y7l}G;GsBTGD22E*FMUtplLGa+lfGd7XbSMS#Zf{F?*jcio8W4Nk
zN^O~tc{F+JR(s}7u;xl%a52`aMS=vlqf$8<^^^J6hE(V`bwAN=1RBx@Wa5)illXZ^
zqGBE#GpxvMRxrI%ZC8I-gdc;=saGTiu!eR`U$8%2e>!UO8iO*9>2<-0_zhN9Gn+&g
zb5nmh!+1;8M^Lzo*=>P@Y;R_ax9**Qkmn3p`(lg0V`9gnkByPbgon8sU^;)t=u5H6
zojCLmj80<#=-S`fa<rWq9JZAhYU(n3zZ!)X*A{F&1}+RrN4^o|)5NLRtjr9GvDKDI
zGRpj6Og&M}#!tys-|yK6(IG<3L~S1mRM1rpZjS`2d({##L;W9Pv$rc7e+6fOC#vHi
z#K-NaCXg5=27UH&K12w#=kYlM#@h>ZIKm6ijs#}7V@L|2Ydel)OdB4D6Jp$|BY9q2
zBkm1GX!Ro*S<epklZ|ZGk$jm>yGg~7gj^mX7_qU6z&#XDwpR_^i9Gpa?Td!T`z~5_
zjO2=OWcOUJ>MOlqtu|l-6WmL}>Y~L&iu>hv?2(K*cx9lcJ(6jSwjhb8DlmE6i4|))
zB5uAsYXyzQvmZCNw&<Jd%zfCFtv&ia=n9ZxK;Md1w<PEig-ISM*8Q>vmn|bIvYsQ~
zxjTMJ%Gj&Xcmr`R%F*>X;#$~{KDo6A-Hd27X3>hXnpl9zBZ+VnPuTR4O#cvp;1puB
zFBdA=*1-@9)RY_-B#yy|vrP(4tpGM<HZS;aRK|@Urswt%H{xpy75hy!!9&Py#bNOC
z!BbUT5H|pO#{eW{qQ_8dK@BI++AjtaY(6oW)y@&zmh^W@^m|W#Pq<~>vRcDW*oyuv
zqR?B7gFH!4a$H9|ve<eoK7stOljPWLXbR4L!Denog%ktHV@_>pD8G^HzNW>#VX*Z5
zpW6d9_7o0m*TYziwST%`k7Amh8QZm!kE7gqt?ig8c2p>4D>L$|>S^pOo@`i!I_Z2C
z)iIyLnO^6Fv~zsCh_V#u%Z}>so+sk#(R}dn$l*dgns5?7iZRDn6~b|k`h~LVXgWJ*
zKCs<K6SDTFWSo2SF_gs0dynQ*g$CILdi&HPUgsX0uKH*?$;}=Ru14)V&^ec<3W)cE
zc1M5XEr8G7(Ts@|@ho(`;b`V=%WJ=cQ=r5z_eMdR-e_z~)6qn3u#?6N93(0n%@oNM
zV+hi|v9_4}m^(+eYkr-dSW$nq7p)}lxdJ@?Mz>~1csh)+ebIX~2*rmnFu{)EXr_+5
zLf{b-VC5Luwe-%s+E7f(Z71A9n$T1M4Z;7;*<^H0cx~i7(}L;1f3#UG9E0}kx2w@=
zdWf?=7p+EdFypS(T9zkRk@?sg)(wTKU5~a1ONoXN<w?u&Q@a^I;HI8@>Z5n;X;`t_
zN588CujL|0`3bp>zUH<4bGxIeT^?!Qb><7PGT0Eq-I12@vMt5{1WfSQ40km(Y;e#+
zEyX-L4%$6UWN@4aY|Db%tuCB0uN+OBGV6h2x}qz_2|+t|iP1s5b&Uikj*Yb|mNtX~
z{y5&$I3AZ0Y>v(1YIH!kkTo>M*xbbW8xb%uoR8*P&x5{?AjW)D>6HE-Yj`d&KYZJ<
zaH#U&RGz3C#P8?qsX%ki`U1wAh>K%e4tWHj=dnnh1DlKUh2VvO28--iK5<-cDA*8w
zAT2(YAv-<TpR!U^9m}U5p&^O^rn2)`!bR+DP!JtU6k&Y*7q(kJ!)Wx`o?{uCJwzyI
zx;K8<S|RZDpV_`+`4TVtU3@HG@mQWXimMb|o+~nH1FF_p=1Tu{#}Y-EK5O;IG97}F
zqZkTsn?2!)=Q(PO8f+oX2l0t~e8dVBsNG+pO`}fZQSVqCU8OX=eqf5x_Lw7HNs}MM
z@T>b+B5a46Pqvq=CuW0b4Y<_%!)CZ0M8{jV-NcY>9#RWOA6I$VXtbDg#lmDPR@We*
z*7fCdl2^d2mZJ3&yBcdu=zmw$&BZ{(u;Zcvo38y}*WLCt5u^Elw>an-P_JuVXW8Ey
zS{|hSsCV`El|hKit>`)(KhKYHJAT5$imL?S?j5bl+AoP;V+&D!+B4>~Air3Qc@5w|
zbyxkv>0@^x5W3t#xw1B#nnWO$1KNV8F0z%EonR@vs)hv9w)I#<lJxP^jpczZuKlXd
zHe)Sv^amKatvG8a^d_AG+*h@XC1QsjMBiA=P1wVL;UM?SBW(#>#$)|;vfZ|$UsRrK
z4ydfMKdP~(0Z3mzzG<jWIJVE!B~$jSoX_=zpy?}U<O{7Ugo{*F)}ApGit&w$XKY7%
z!qAjUuS$P=jA`51o(Ueak)tp;vnx9A>_<uDxm%6MaW`Rm^fxBGpt1fY(*@r3?fHgH
zylzsERJSK`aFMX6HMt9zwAvWq5Om5dPpqyzQ;5sR1rV>)#}8eKu-p)hl38qGr2~Fq
zs-UOxTtf)3qUETY`;J;6esBjQ5`eS`7Fa19irRI(JD2S$v0_YP3|C2HQX~N>wP!L2
z=Z$*_{<?V)$gS2>>47*CtoX9^#&Be!^4DTHATL9|k_*^Sd!i2>L}$!K+cW9K)59^)
zADxq0y5XtIHr<|SBY0%9?TOaC&huEmd*pw%Z7yncoNBRa?U@cvul;h&=3@r#f#KTW
z>rvsb+ali(KpdeHV|G&??j}U6c1@3x0yiEx1mhdoB~<T@;Q9EZEePbHY1BXgOc%o0
zyDE6)%mG`9Vaqe@)xB6NXB(j!%h7P0M1CukWx7Q68Cz8^lgK^SE6Wc-w++D~S8{sT
z6g&s9Q;xAZ{I0yOitgK=E|mu{R0n<y57i}11<5`IRmmNt+)2<LN7VvkYCHOMrLR2^
zToW_1_Eeo9)y7XK+7L>^p2tL6dB$=Hyijq99s2*8b>!pUEQ8k_9SLQ-j+{7dRUP^G
zk-l=u6;V4oQdosQ{{Q{6T^$(*E0HPK=-nOZ^!N)H!-+i|>EN-OkkY*!`I;R{=1}`O
z@<H^l8c9(68>@~=o-C7K&HDCrNutZzXcRsncpk)jB=@rVjzkH%1~IIP4#{=NTcJ?^
z0@692XiWkT@#4T`D|BR91`|(qQ?B|pElj~BW$mi#k)kK8sE;Hwvr>nU<Rk0u$cS*z
zjRSkMmiRkze<Tx}UJXy)N*;e!Rxg*57;q?NF*cJDEsh2}&VMJSaWbGkb>xUm>&w$W
z)7h*h4WDXP)h}EXbDRs9+6R$JU5f>Ab7R7<$0!)0prQs(uB+b!BN8pm?&1Zz8BL+!
z<r%xBVOpS}SXgPPQc#aODqxXcskMcem&dx{@Gk~HiT$|ya!$w(msIE;_)c{1sn^mB
za=qnfhb1D$j)IE>>Z_Xh`0$XeS9ZSZxNQWqErbsss*QlUJPJQyTQTTC$p8IVgNj~_
zsF12n7;={OVT_Cy-l#na6iGCfi~F%^Bd@|fVOzC%L?j01Nk=?4Dj`$+mP;4wQ_b$s
z1nK0TsTC{B1<gx1km~|6efEXLc%dpS6>on#o(~%^lAP#xI`p`%TGjE4nVu6s8RT}V
zcmf-rUHY>bdb;~~MoJ=d;aj%nc*g!u2bFtj@9|75lG#0Gk5MH>3xa4@A5Y|Cc>FAS
zUHx4|)MK>)AYxB)+sEg*VFEAKAJ6E95gk#ES;mHdCguFFMs>z|{k1j4uxIh-@n>uy
zAafreh)<ga;MyA>h8t^F>m!aFMg2X3y@-{LXGX*?CoaXbP2I6(CtNRj$#YcoXrcNz
z>%IE33-^Gs`T_;;4AS38-Z*+Z(Y{J3`AL0&1AgmH2do62>Ib;l_`&seaNSprXIe?>
z>A7fpg6r~HtQ3xwF4teSd4c0m0j>)Kq|Z3q>*uv!;nb&}J&cjvdfCC($maePqk-GU
zGxg<KLHET2AC@_Gs4#=I*g|Xvo*{@h7NeUE&kuwO(eB1H$~anL0<#otj$@Yu105CE
zBoDi`9P3>6NchRPb0y|}c*s`OgPg566ziJP=;d+Sh-Kk?*```eku;nb8JaFk6!NTq
z0V;x0_JJzG+Cv5&#vCzt!qMP=SqI*-kK;#fXj})dhQ(Sr1}`)ClUR!s2<4mauh~<9
zP~^F8&yJ@np|QuFs~PX|v-p|45GY;gs<jgd`Pfjb4B6BjClW%ZTpKWDRVUJk7@qb3
zn>$ZrG;|Ry=VjN4j6PG)hLjj__lZOS&LI}vb0VEZKP#~0d({(X@ldnvi-v5VD=0x=
zl}`}T7USg;37f~s7HG99AQBWH)Tw*dZ`KDiWT|onqv1rR8<H^gS)(9ukrmK3=?e(l
zP@$ma*!sIQo?~tlx1d95kFRf65It*^Xj(vU3Hk~Q653h!55Pbq1;VOq|Ep)LSDj4E
zKu&q~YEpe}(=ZerBH3bIqiQsKe#9nK4IULoKK%`Zqc*E1Oa{AYENQQ*tmf6%KD4>$
z&Y#N!$WYm^+cwV!fTViz*rC^>GFneq`x}+9Ni}pc_QfBe$r|aF=CHQtk|(<z%fcJY
z?nF=cR`3#CP^n)Y0I{eqfc-hEe^+1Wpy9w5{Y;h#B^NtK{4W$4Wm{Hx8$1X2vZ5o<
zoWkg?YHofd+4_lm$+#j?f&4bqnc}(+dBpb36WI(U$rs#EP!E@_IFp>S`x+AEwNLax
z1zeBg^_4voR<QpGxgQ1e*!0?C0o3>HSADjvwQN3KvM2dKul*WPRdm4F<D!i{ag_s%
z#Pe8qe^EJl*$Z`xt3NKd#eB_94H5)zM=>Fp>kg}m2~#lY8L*vE3j&X5s;GxvvIz#Y
zcNa5t-~Wp3DJJ6ZCwS5q6IM67dMw>S`-%yVV2lz;VEWY>ukXr?)u>ZeLe^xp#f0R;
zVms<$NM!gJNz?~K{RH>01`S1HTs!=Y->9MZ!R=z4zi5gU7`=d{F&tubrr1^tibF)H
z5pU5h2(T5ODEC`28sLOF38re&ZX<Wux}!Cz8zo=t(fZZA+Izry^*1*JPuQs12IEaQ
zZ|>P*K5X(T@WDp~ZcD{mW-?mg)-yxyRNk@zk16&;R%T<uB#cV0$*U@Nw=u57IemGC
z5#G_)^yTU7d6UiSivxotmR*nC9LA|nkh?Xt-BbZ01#s|Ithfu-P`}-d^=VGdKX;;=
zzVo)AeTG@H#mXVbVZX4+h?a@tw59m&!cQKZZ+!D)1diqS_Lmb#{1Oxo>P0Tk*y@9>
zbX!;b@d+G%{9&?D%q$Y9?(uQqb;`CGz4EZYsbWjtLwS?#N2lm-&e?;Q-bWwV!`Ph)
z9r#y!r1Gf&R@EMB`tWwPt+i&?1~6H85}o3$cFCT`Oa|B{*xqNctNnxIw*|qUt9`PV
zCqRZeK8W(nK2E4RL^nG=*8TWr$n}=2>f?N}$1o>IlK-6_Cxl)5A1*`P^>IRUUWvoD
z`{T?pe~fcD*>^wA7jsI-i=U828D9T$+xKy1vXZ{grC67(`r~|*J^ka_fAc6yH6JIm
zmtc$yMYW`1QUpU+C)Ww!fBQdg@ZaiHpCyt(I~TA(M5u~RNwCA;z!EmaaL3&zE9H|d
zs6q;lWo<E`l-JSPH4@4JH-;}(#`Wah16I-~&==y|=<h@LEL%^&<H1(kl&bgRL^TPM
zKRn5$U5*vTD&1Y5FWZo6VE+wR1cKBjQ`))%!JHI`M9}4~n+_1sU;xIffWrTHm}7bM
z<3vt>RBLnk>qNn>1=vTi6r9(V)DSi3x`x4dBRT%uh#ex%A9rq2@Z1bgre?6Xxb<;H
z25d*Ef!8wR^LC8T2}E~ZE<iFR!?8Yo!6++PKP(Cg0w~(uXw2KsmNZwUevG8N@lj~8
z<>*8v{}*gUt6uDbnON0yKl;+v^=p@}{KgL!3oVLmYS@7?uU=If!Q|WS2MB$aM(lyU
z+({aK;h{P%9<oPjJVS%Wu_~R&KW#fc;i*vcs>$&_d#XWrrNRd5?^pR}&ts{lk##1c
z8n9Cpl4#31GY^R#fXwn}ggP^>x#tT(FFG^o93;cALPG8AOo+<9z^Cns1_OhmGazec
zM(pHmKX?v&jVd?;z@qK#OhmzR4DRd9H~l%6?N@hZ%qH5_Z%Kqu6F&@cVqrIH)wYCD
zbzNs75xk;sAkg117n9ZyZJ2=jHZ)eogMDq%pQIPJLT92!&K-S!%i03$G0(x`Lb!JI
zgVAmmOgq%Is1_ATW+_0&XN4egFPD<m{s8UKSm)4%^r|j?({8t7<rvI~6#i`Z1O?D;
z^2-mS8jKU#F^H14Gn2Il0{M+i$KDK0V3w3Z5Y5I~A50FPSL55G=OCKW-Z{ZQ=^Jnm
z<3rrgd<^T6=dNo6?ytmz`&!wJs6kdbyBS~2_$l7fSK>H-Z?~iSLo;iCpVGqJ32?fw
zxkuX@iwI9KX^Vo6N#i$gH_Ey&%Ce<O-!9_+gWZeuLjFCvhVe%HFff50kw7-FWKQEb
z%$!$Sjq!=3F+O7JF$Uw0*+y(4%ppuYMv^fHm<^~V9dE@*LzgEf5OCht;rsSK&0Mqx
z(cu^;)*eQOsEGa5XpdsSik&CyvA%c{xrNAflwI*ybLz^`XzlN@Y>uAUS$OLBMEyKr
z&vn>1UiKp3<J@`_80*TMaPx87(Unn*ljI?DtD>~4!*+Hhq+kP@dBW|AqHq6TyQAnc
zTnqN78&rQ4J^9|~7oR<kq`EKax3WgnU74;Vy|UGG<pZ63{9L(=^h)FL=wPerx)PQ6
zVE=xr@5)qy%(c_j(3Nf<Ohg;iYZpPI(yO|u>pM4q>S*ObQs~Nrh!@7Md5A4pn?OoY
z<1lHCUXS`4tQhb@EUl!k6q^&L-6$izzx8xws_F7KZZPZBxX7nqrPVES;4MdRjF`Bm
z_w2LL0L$Nv3F^t{{R%iR9bmzw;cBlnA;SutUDd}Gq-}jRr)F#aIPEg$YXQ{AI1)!r
z+I&|g-Q<h(OBCt)bphJUxv(2*pFGKaV>h*^X9w(7UijDTc0h(Nquo*CzG;-V3;G)a
z_E}quf!tOa7J(JH+m&x{j;}(pb6G5@Kd$HGTCagFVl%oNr)P@Pc@t-CB}T@!!B(T#
z(TIgukJ%o=zZkDg+Xx_%=N=0H@<y``GA6;*Yzb<jDUh`{3O%Ggc>N%D29)D{#esYn
zn}^sB(b}WRl3@|gbMLWwOR<l(W7iI)^t~z?aK-7<C_)hq;zplG39FChNL6`<b2qCh
zOL)N#x|PxqJ~B?$lL{{HC?%}8gqvzrDIY_w8)y?Qe({SEF-+L5QpOylniICWq~|NA
zf$DoonG_&Fa>tgvD)zL=@H+cS3C)gTjiuiq>&xHITTL`O?6H~<S(Osj^}9bsovK1W
z1^}^MC4-nJSVJjMA?zlRWHjo7YDeLh)uwV=7c|4xmXfup7I7``jBnRe{9(+4%Zf1&
zf_f`yrRa}~kVp#9kM0;^;G*>e%%Fy2Qg0~>NjRU<0_p-R$BMyykJykt(}_UVIvS-Y
zR=Ac+nW}lOaLp=aib=FETgpg~t{|=t$IGh$`5}(sT=a-O?i;%nfHvc|Lx$w}m_~Dl
zU5}>t3fYYqojh)6?r&=4j`Qxi6-5Vei@apFOW73erIR%T@s9d)(>U(B1+{<ewf(mE
z5-L2a?5;}I{s-4^fG|x4tb3IX{NaM2`DGdDuU6EkrM=l!H3|1LMY&tbbSd%wT+A3x
z#>s6S9h~Jf+h+WtKy|!*V~-IKdsYQhz8cVzjze=E#3aMY>>I2n9>&jZ#yggsTpsCc
z5>EiXd>!=<du=-w?)2~oy1OTWd>p0Dus!|3ej;tb+Q*e5&tk2R-w`e{KCgH<+864&
zJYaS*A)kwH5oqi<nJ?u}XRPXEK1VRH+&(+?oie!l19q31`nBATBjCxrfMp~nqb{DY
zJ<;9%&uwpX*Ofw2?2B)tSiBLcqdSz3R--TXmEb(qsv~mY5I?o<WWsE?;W4X^PVlh-
zzZy;^O0@PXGV-3xSFLLXQa4TNlxz*|yi+)t$WW1N1x|X~D&3BwcWGA<Q}Z0iRy>)B
z8rBF&LsE)g{PFH?x!uuZ+UaXg0Co!LHVJwInlJ?@pUlU9qMz%7JQQr)lnuqAeY;|#
z>h;KY3e{qhCo^#*_2mqK<)<|xcwqv?*{o`#Rfg6OV=h*->oTSq(bmII!kt|U5Fv8#
z$@2l?lvI}Z-$xHq7;)Z+uj7YX>}F-D|2x5NV(uwS-tAbNG73wV7~P3qP7_txE#MXc
znsQoFTZ}%=_xZ~>#`?GUzp<s4pt+3F(fXb$U<~^Q9LtuqXrp~DWLSy5Pw_+_ylktn
zq&4r^y7u`30)5+vni!RGYHc&tWRj3a!?txYh-D?lfg{)bs5OE=^`JcvEU->4gnJmk
z>WKoWdQ<_{bGF{I$1ykebKZ{5a3q2$eR4A0aPNO<Pore=!WCBWnWjNrE2=$LCq<p*
z7cs-GQ)Z_U7ULApV|K@>d}KNicOvqtI+br<xWRL_Gs=0w{2oE`u2UK955aJt8dh%;
z(<m-%&#Al~>4qZagRjTl=n|*D3EOunv(X$Z_bwVHsDAaS%)}qZi>BsOKKon$w`A>H
zG)q={DxvT3lO0wUV4&s2Jf=S2tT^d4oJ#BTlh3SCUALyJ>eh5Bkp~XRuoX^a#DzuU
z*kfL)5g<9%t`06-vSN${GlU06Ddv;n|8_?&9?gks)Ow-<>@Y$ubvn-bs2p8&O+yE5
zC~6X^1)n5=%E9pW;-tPXg+*^?)9R&k33IDcnRanAdRT?8?P_#<3X~!OT~tY0xwD+4
zYnplA43gY=?HV!}wA0rs<18RX!|F=opf0kTDuB*i;a0d6^OPwr#vCyyf-*$1cPbNw
zIbMiO#yJB^sq|J{RDUj!;k@3B{)llpj8f5-qQ7$l4h4zT-y{TbIf|6NZ$;lpS49XG
zTdiy>Ndj!WQbtZluz$%mDhECh;>}p`>EZJ{jN=n=6%2NO-8bq$rx@=KUaqCzBF+yz
z%rot?N2l`5M(T4ihj{VhIE@6W?1$6ejxH4KhIhr27@17w$ivj53;LdEE|Wuq0(h=d
z==sK?e!FK1u+w>1?8(cD)3xJtzMqdG1v{OP<qAGoJ5MK^Zk<Kn!d15GbVgweDxhk+
zPiGpCG?%}#J*N{gjt>rTQ`la8LCaifweNI3doV1@GE=Pz&Q3xQYECD_J5Xb_r!%5g
z3=+Qn{x7ZWbizN_A+BdN<a`nJ%NkB+Y(x%<!XJ)XfckmqCu_SksYVh&;mwWurJ$LS
z<s@AjZ84KR*NkF~rSg>F4oobb&h*KpZZ2YR_MxOofN|b!`nolN(eu_LSTDH?c(4A(
zSpYS$))}FJN5)ZWe}K5f?x<Ri4cMe=!9dKSV~Q~*26&Foo(=-X3C(&uyBZ5}0iDL2
zzT7Sh48ImV9AZ6eUhwb;;vSBDZ^bRC0vu|OLA$A11mfiyRsl4O6hyaUrCNFEBKo)!
zhXOM_`do-_&wqk@VeC#zB^bzE_1E6o+|_DJQ5}st)-(6C42>Vya%Fugf$vr-;H9ug
ztAZlMkxMMty5L1V1AiTh(Z$PuGp3J8EqEKp5K*X`q%YZh9e5m}?Lkx?b*z4<o*+b?
zPUMjwU<`=R<EZIjwrpGD*Sz}vzq2PX{^Z0Jd#YA$DTTmV)!etqo~sA<zVqw-h1SX6
zovtHBpJ(0a#2-bUVLQ4LW?dTWAAyUl(idh=_y*Zd)%%b9vt8X8yIFh>$<5ugyF1^^
z+*U-1>`usjA-SFI?M}#kk$1tq?u^|o9-Opl4RUrm#ZanI#S_Jn)mBD7&pnUXm3l|A
zMZ90aX{dK_Vr||Ux)Vxwby{P0Mrl{EaC18k(WH*3OCUTe=qrU!8n?FYOfGUx3~N`{
zxMIQ^6r*hiqVAG_z(Eyj3@|ZfTt=K|J!*)3ray+YTE0A1j@3#JAb)M!P_)G_0KHLn
zCNAQpaad3XSo?sTQHtr|=p~HOW>rdHI)X(;SA(2Xo6`~wA$+(NYmILxiK^#SRGb>N
z>*^q>rEf$BTmQsebu(I_mPN2|E7oZ12DZ09gH^vBKlyti8ExH(9r^apY(XD{d*n&5
zsO>p|QFF|rZ_F7~3?&7IbG~*j*81Q<lt${}V**E%Y()^dHU<{PR@Dl*1d*xq_u2x&
zosEEeZ2|qaZB}*!WiV<ZRc!pD*Iu*x-T6vEYUVfM1C7nrxd{EC+PQfckso7qnlE6#
z`8YliWG>sbK2cD&x(fP<I!1SJflRO1!9P{GTpGd5%IR~BAn#gz`OS0=A+vVp6x7Z>
zNw{=!+y#$t!JmjF9vJK!w5m@MzG9+^QL_q=7^5EF^-1PY520t+ts(@!&iH<k@zo6h
zcwlj;_I{EOaSU>qY~1$g8#L)kiHobF4MF?3k32d|jfNOfw2azl(McQ@OshW0tIC7i
z=U4wpKGiO);puG+YJvxwKk~RjqXq_NWA5eE6rFS;(uB?yv=+ccrfx9ylSE2Jm}0yB
zoPwV#3NT+uwz?F(I@@{{Qr4|Sb1|0ne3Eg#-^~Ji`+H;QJibCXK$x7Ebl!GI15FS$
zaWrZU5dV8Jx<ak*RFI4aFB@N&aDmF~Cy6?p`44vWlT4j(sz~7s!L~W=6VrR8+O7#q
zG6-^mYS5bfsC#1x5X%t9HDJIO#BNrmHcIHMTk7mQmj~=`w1R`fwKqn0Q5sSTe^4w0
zv;~Y?iKMb9Xj}0IRP(Nx0cJs>u;{IocahzzoWf~1_vOmpF3(vB*bag_tD3`knD6x%
zDZMLKY$G~sn_&w!)rsc=v8^aVuYS?)>yuy#Yrh2M2eIe~K1p}UPY=}?9SPc?SkQ}@
zG{>k@$m}@v+nV1YY##;b;fV=%ReLH(J-+p`Pm-qqJ2F7n=b164i}O0XZgl+0kL*Rv
z^CXw7eVUJyVjn}`cYK<$v2=OTs`N$nNf*uS{50c5MN?zDqBd~a?oaa_Nx)3lq&@Ln
zhW>-PpXRHU_uG)|Q*m<H_XNk)0Syw>@hfY;1#zPVX<+asiW*(nNwn_ML}S1>kgPri
zoVw)pZw;R&n$vaU(^hK?P%)XeCVdec2@+w2N+;Ob@K}v8;eh_Lwa26f&q6L?Q;MG^
zR9^qt6)Q#Ei6KsS_e=QP-@$RZ=hM6^$(nW6ZOxpF^$LLG%J&B=tIL{K{@8{-&D3ZL
zPYq7ks3tMqk3xI$(@YSvB!K8~F|oD38n;<}jmgL1YQQ(zcL{&|=o=qrcNe?%X+~|H
z9m|R1`;E<inlOKw2<rOlLxQ~uh(UBub~8F1#17eR#R5Ss0ShiThH;-4Tt4rpHHP;;
zw}mLg7sqWJ7gb31Od_a9*Wkw$The3>@tlHJz85VKb9z{kWwj)>CT{w%3YQ?Ut8oy}
z*Yhe|k99-UR<T^!i0*KHbAQUs7(9hO+j?34b8@ZzrQMJ6aXi=TfeuaIh$|5u#v)+b
zmYfqG2_W1N{T=W)z$5I8ZAVYM0*JB^(01cla>Smh_ySP{lT=yL<Z>FY=Q=1jb=nKH
zO1?Y(9=Fdj=8mAB+VNS!3#h@?HLLn8ADc4yC=pJ;fb5LwEwCoLKFg@<5%9)r_h<PO
zZvCfZm9povjJrfv!OrTlOgUUTPLFWNQ3E2~XVvO|#_b~!J$;riao`FhwN`(lB1K6Q
zjrx%>TD?AT^hnWJ8Z-&`;k92|V^kfug41!+XYtls#&YIs-Aq=9iL{Vx598)&1*48z
zLVR-A|I&&v&C1y?#k?w-1p+}vo%WgXCGOdtsB_%um}7abru6Ms>p!!yAUNKIp3d=i
zC>FX6+!&3<TekojlU$pOt!h1I(^2~fuZ&ssOES>^Y*z#HW|uY1sbUgp{obxcvo;W8
zJ{AG2f_1q5Sw?gAADNspF?Qp#Ow9)`V9G^WmYV?x);qVN$1Xe{Z&yY`IDv`Moyrb+
zP^5+E2H6`9JL9&fQgqQoca2`IfZj_ng@Y~yy%#G>ypT)W2+L9XJZ)E^_AyeK+NwVL
zA)uZ#W<5%qr?|H^0*ni5zx1%rw)t5iV;{V3TTxcT7n)dnc0UlJFF7N+X5~T5{*OPi
zhgv~@0mJkx+oRYv<ck+B`;mSe&<h1@J394{<y^mP;FH)1faY>}8iOGFe^y!4dC1T2
z?D-EiDd}?j@Vxje-8B>Jj`gHJXLxe#=*e3*1_$k0hpL{8o?Xs3@9atFR)S92uAYqA
zm}Nn4clV@qIoxS`dJ>K3@DPf&w<oJX-GJFOkd9MG`S<nY!{h!}6~C%#HO1n9d%h+{
zIPIYmthOgpq9Wp`I*s?)YgVrrV4rLaJ((hTc0uW28e<MeJ*-@lhH#kSrZoiihyVaR
zZPO4Kym*Akc2)Hfu4m01Hv(i(YL-mP|AlSru7K^qRjj8cXqekFUSycwo=o<}SVpet
zvO4&p^3R4enLm2fMti=?fh;`Q7`FbP##fM$pW9^gg6F;O`m~^Rs~<LCR$s{y<u`Wq
z<zithIrkDK_l@==a7`5|Aq?lE0w>e6R{qAW3r2~%82(07A@jy1lw9OZ2#8WdO*rnT
zgSsE<J<iPS_-S0mX-J;6lvtY+rCm!mihU03b{s&|C&`%UZfxF^D@&KO)oZ^;F&zE6
z(kb0&%bL>KFBuqL!dL;1iMXm2K6ud9E1UIeu5jCk85~Q3GTRIgV70J{Kr5_)^2?AL
z#O_C>ezc=*2=YL1p5@qh(9egO;mp7Y&igTi^E1S|e%zBU7`GTAi*(9uV%vf?T{+*_
zlS+5IP5cG^RNY~yGGNcNE!@B}Za`+wqx;f{!U=m3v!1{o&(5TS2X9yr>hCy{QJlA)
zX9u;aGa1FPogB5DXEIu%@yE3IvY(Y*XA(x^K?_^4`%FfBgry+Eo-?|G*M5D;_Udnc
zBiOz(8C?&fsj7~Fla3tbsU}A9AUwktkueg-SpYM;t`wrx>ds`0hEc(4<}&D1uTeOe
zzp;ksmuLRqS&||*#)JrcKpZ>4-k!o#6$BIXmsqUU7K1r=0z$Q)$*Aj!-4xxos20RG
z_Jt@#3kfo<JBF`B)UlozesJ{Lzqa0(B_byn7@p)<aK#d~4v(WE8j7VtHDjYMOFKJH
zblPNm_qk!UO~?1Si9QrBvoRu)Nldt+vnkkyxtE2m%D#3cy&L@UNyawrj^KuC^JfzE
zK7>VP^nNOZy}S_}5aojRwbmYE-ph8Yax9WD>TMNZWQotAK=5}xH_(Erv8SZ|T8t%z
ztsh4T!_zuOY)xYDCH3U@zTK<z^sh;(z8p0RAR2j%0yYBiV>=x)KZ_7?J-$ZqsN7Je
z=Ln8vo6!xF44=p?eTy{%<L^heZVO}&V$=OpaD_ZnIk*0ZoAEwU`4RLWgDIZ);~4ok
zCOq33dFDL3@kD<k)h8)vi~?g#tQ{3$hzU-gNq4gQ(7cGga9=n3JhR_xzp@>lCj?6Z
zgx^`!=Ls#rNrW#*D0Y6H(6ZlEAiF-#_YWa#@LIIp`k@PzuH3WssIQa7LUXCr_I{r6
z0HGrCS-{)(c_ti}+#LWQt2H3ois+i7RTDKP#>t6_2B=oQq>En(g6lp{6Kg+l*y;s>
z%j%S#u_1swa=fdxvt*5x=6}H{a>KBnTa%i52DE~vhfmI=wP{s|vd4wet`E{)!7~;$
zEnwscWF>t;W9iPi1AfA%TqG8Ie7of^L3U!TXTjLY5J)wrZAfF2jsi!+XqB9pY%*4%
zj1<$83r$B2zqYT~tiIP4rYa6gpJxiEpw1}JY%bQ~DEBN@57(lT^w6~VXauL!C(ir&
z=lNz~ejX6<H{bsP3VEY4Ywf_z=!mE(2<UF9)Xno?-fw@NslOXIu~44y_0H$%hHw!z
zDj8dd86t5=ilIeqOB;`{yU~4$qHarS%LWiYd)e-3<ydayWKp#vMblV`Zv>sJ|Eii6
zCz2S`)?-OLh6U@op>o$Rdeq^~mm~S>XG<H#eo^&)bd6<{Yn)<9JVyG%IPJJIyB!56
zEjZaQ(i^EVQ#gpyU|Vx75H0+PDqR`;_V4UzEL>W9<rsMu9YgERf)e;K#ZTizsY`xY
zFVv%(*xDBfJ=T7E(RO^1Z_hTAi&bCbOL}sGJeWJbNVrW*5}w#!B)muZK&(XcF1x=-
zB*No^b2sawO#UQ{FTcoCAiYYLZJ&B2;Hir*tXiWFVl+_`qqhuASZ#a_tlhIZ^~rkF
z9I$$Q3|{hpT@3<-5~4E<p0&n+nq4?+O<%;}`e}ADu!074&ZsShdF7Q?tX+-A(S;1T
z*;e%JLPE@^wFErMkdJ+lF0V`2a_CoF9bvi!n)t|aG#^5VgVFRCnZl7n&-pOY3Tk*>
zZq5o!2J9G=fy9*+cP-@X7vKHw*nsOydlKd!hDE*_1Njqv%;xm%?^}dpOLk2iO(naI
zs5_iI4X=NZX{@_aa2MK*fC6(O_-!>tc}+5;*oM=?c3Ug(dX3%DEO^@gU*P_N8bH*W
zZ85gLsF<C1V>k~Yas?5@(ifQ&RV1-2#Jx&~9_Qj~TMm%N2QMKxT+zTE)!J&5bdmiJ
zO;mthM!#Vj(X|lP-!Gf7;a7OOZK(rzZgKPX<G8@4dX&HPnvP!CB|^Gue#xllk^XiW
zxIK==`SxFJTODJRPfUibU)z&dh~&t@>sc*RPT6OIiKmylxIfo$$?}3z^CCdP{>tNO
zxj?{rg^V~z+m7DM8=p^^VXJyG${m55=IrcL<%toNWLIxO@d20Ta3%L<6hGwL!JcT@
z(%G8s<mmbmd^#aJqCO`U3+mp416?PrrZ*9hOBl%3_NM*BjKrVwTGl#2;2~+PUJyWk
zUG&kQFV6K8TxO%b95pz$O;LuCnibT?fq$F3ZEd}oC24N4_5kXTYk!Kp`9Q<Gj}R%|
zN-z6CZ`-XoISb&T4eN>05|7qieI@k#XDi42yIfXcL%kV;Nu45b*{DECN?4nW1?+Ga
z{&AbGtP3kYtHt`szq6|vV?v(%bE>lT+Y!4Kpp&5jwtd{@)u{RlyB=VkhXc5w-%IUw
zQ_C{TKf5JZ@XLlk-456ZuMS&sCptJrn7xI7ln0Cx;>9TD(P4L^G4Z>h%}ZKiteO#F
zT(EokBpIFFz?B0=5NlgeKb~@oimUo+@%s|=jBRUawT;*|Ou~O{n>sce1n7|aPa?_P
zQWX{&-e}D32T)}Hf_6XX{jQR!y1?#WF`Q(94+RV-=$<`_NnL<Be;i*^_}_1<9%(&h
z>`8CFB`TpL{N435e)Md0!<g8CY|mngT4!Nu;d!5HC7az5Gy(8JP*=Q0%=&WQ$_fvA
zLxhgLe2}IHUBa=QvZ}s}wukyKb|b}LU&5*Wegf_2&olUs?e0sM&ujOR?TM}(l*HK~
z{EB|qt*I%jDb;v*eAcS_65dP=PGezI)0Yv&IR%%L*Y@T6|11vCb!ztkVOHy-puF}^
z*3g%z2BK_ahf~(rmkIbw`qZR>c-=rI3IS0H<Az#F9kAC$w;yPa4jhn~6wkM)4*b6R
zg_Zg;g)4LZu=<E^%oB&tT90N@K+jBw0H=?=S|y&QWCzvXC_=D}j*+MQfiW7};GUvo
zSM*P6V4fGC!L$Zy2BSPEtIY;fV<UDo7UdxNow@ir<U!t};_+6S7kF~l^5}$iU6sjm
z>vrTfqVnfIv70LLz@NNtZp98`@;;2|y*sDg*2dH{?YBFzv>dCi5icyLFBTq?vPD%q
zi6QG<{m$_woWW8Ixc@V|7at#^hKiMSqCvXBSRiMgXLz~UYF{GyfAqSo$KV(-dz6xm
zXiXMep30kX!akXpNujs4;ul0#?%-heV@F!u8s$OsNia8$AO0}PM@C^vA8C0A5rqA<
zJ&xMEkzJp;9lPLy^e5VRZq<2v8YL)j?O8NT84pR{tKwv?gvXMKfsd!+1&+M+XUuSA
z@Q(hB(w8vzMMC9oz^eXym%6)v?NkelJ3T?uuKtYgFclbxLP>V_XY`vI#OKswRH{kW
z<uMdUwzoeYdMsJk^?m&bclGYAj>0m;f_HHA?*ja>NtU8E;6a^ClHdMBP0m9H>tiw=
zH@qRb!rvCne`CK046fAHq?)e1gO4agkBA7t;{HUihLNaTXe)KJTYD_aG}aH}e(P`@
z$6!;)B}<wFj?U~}cK{8&8wG8-jQi8!`utC<S2YoD&`DV^z4Gd-@7qvx=ugSf_#(TO
zO-7H{lVVwiamRE3eU?;;Hmg-nj}Z?vx?)!~|MZ%eCWY8s>`TYmFOwJCwHU)s;Cy_o
z{RIIj1sR%{5xdc!2%(D}V9@cWKn`^?R(F`iC6<S##Vd*{`L-H!(PJB;ZjsbGW3~{@
z2*O0OMeQDz*>*RI&=|0POEKpZKf?vg_q2-5=WIEEokZ=jq8)R(3;Rb5iq*=X@Inl-
zu0g*2YcRR;=RY9062vq**=k#A(CVx)Ph5dQRKrSaN>wFVD3g1r!eo8!`ie*TP9_rH
zuriNh9*L{4Z7s&P*Vxr30X70zdm1&Jp`Em6s=2m^;^Vmh{j%p(dl5T*$&*spK*q0#
z3q6odc5|tv&8h|xHW3LNP<QHsCmw>9T?6?#<GtkJ;5hla2NLp}LoB{$Afq{pPb}Gi
zOc=Uwm){p(`#GuA1L-gi5(#=Bqkqac<0jR$0uPxmFTwBsmbYM?z(Y}jv^juB;FCn|
z4bk->mpALnuV9P}o756_P=ZhD?-`etwM8#2m1uzNnkg#|yI&m06s8N4lyNJ?)JVtm
z)&IxT`@lwhUHiUE6^c;Btx$#XP#%hS9GZ|EHwRUyLMhF~#s)(S@v$93Q#Zx-*dE(s
z&tzu&H$WUh2qC^H&E@5AT3#>7<>ZozRz4}s@yVsRD5eNisB$V)p$b(fLKUh|g(_5`
z3RUm3_L|?<C5q(V{Pu6{wb$Nz?X}lm+Y^xXzwpvJ>x@?<`NtWB<27F8J)=%;acm=7
zk_(pQgIwR(^iNiZ4!iN)B|B^or2Sh~jTXaPvM1DQql4QQNMU=UoGH)KF;WzczGx2h
zWSqCB$c=^IUrwYW0c_oxf3O(=tIVG#D^_&)OoJWQSmCC8b3&vUcdahOI;l_b{bE0=
z0YVD}7ul21psgwVna>0bKme~wq2+0A`YAcZqM7Z<1iFq?n#0Zo4)7kozW=Aq#nA9?
zaeK@8o=Y`p)-S6>J0wBoLiB<bALX)^z&PM9*QRIxcVDtAdb91>?Y0n4u)trSzr3pX
z@yy}L#D7iA^IV9o#}wg|O{SR}0kzOI+GjW81v|fq)XySMO4M>&<Z0%JBdpHuXt+DQ
zSRm!LgZPcl+P&y6yoT(4{NCJdmdj={sb)TL4<cKV%k~Po?XIb0F3M$_Y$pnXWokf9
zn4XkAYI!d0Y`-<wid?p;lUBu6>f`GC2ewL2d4g?L$CIB1cu=g(C7!@|ch=jw013`q
zY2!tuIyV0kTQB%na1k{(2x5QlkZlx<_JVDSSGa&-MY7Gg#6(&K$kLI^n8?C$w+ZF4
zl|odg3q>s&Vmhe2Qc<hAZ5;LF@*Oq&2o|(GdN<VDi%q>Vm$AOql$nofZL}uUbkTA`
zcnoUhYa}>`_F>iv`f{r$`9bU`<uZ1Nw}qzwDN&3emo3$#CUs{<2r(fnh%H3*w@-uI
zjt`toYfQuNS*8`|=%C2Db&uJRpc~Z^_<crAVUCf~@L29r%|wlIz+E>zA)yx%{)HV6
zD5Is4ozT#{UfM}LwX@7m37v|^I-C}^4dmO#(<3OvJ+EkI17fkCx@BsHHUUzf)BNVf
zZ0_&z^qR(Vc0Rxq-4Jjg{zk|eu#55C{JSG487>QzoQEo22|QD+R!qNo$z~<qNRQo<
zNI2I5AD=8gS?zkjCCr2y+YRC4qL2GM*iB7nm1HRH$$l#aH-J0CZ4syo26!iW8mUBP
z+q)V$s&N+8y+ufg%jfLAmMt7>6sg5}(>C0R9mJOOrp+`~^qE-No0fj5zt5KSW_!7d
z3+eLSY~{C-TNh*diujB@$YxsEo3VnO2u#VivZ^=RIV?{sMEc(S=s1tp)==;%UZ<{h
zv-9f$GCTt2xxP1J65ZIh^R_`?sW*zVWqirW%Tl#XY8(k>Hpg4HQ*cZ+_GX)nVsXj)
zpgGz%p-)Yo?JXM9-c8mT58>C}Aq?49w&(Bq!RgsLqiJ7(qt5omOAn+@Za=wrNtIFd
zR$gCngT}|Z(3>QWzHTe^X4{vo8Uf~3715gFDl(xTddq#q0>S)2MPeH;3k8n&UU|Lc
zLjf;Ix8#sbYyODF_bgRa1S=&ZnX_Z@eQ$2mHAlxarC^#euh|LFo*CvG7ClExup4B&
zQ~HL$YZnS|npUHvEFqleOe`^qB-q)&zQr^2&k8u)jQn<bO!nhBVg4MNtAH?4qW*%K
zwRaW9?4p|WaPI2e!rL4gSE94=t~`l2q8+L$7|PXn!I*~an&!Ecb$MOLBbJRI=h%(t
z`r?1X6-(gro}BuYW_G7X3E1syW^hZqqj&A;26{JoiQ)@(Pj5X0oz(3DOneLCabMa<
zJvAlHlD@QciFPAjY@ap<G(&&fm)ZE3Gq>7=XtKO7+q+!%Qut9{wzb>1C#~#D41jgR
zjbv2~JUhfz>$`<xNlUjjzDp}4XY15^hjT2e>r1?&k+W)jKw;w(8K};1(y@*4rqLC^
zoBGn?|JJu`bF`@~)5ZYQ#R!}GvOVijgKoHK2>_#>S-7<?V>>n1nw)j$1NaOcUDWUu
znlL6#zs?vga}U4UE9AB}Hd!t@o_}}H@`4j}LkT;Id4S<n2sE)$v?bZBRRbc~NH7Ra
zs2K`~aoeXy3W(!C8BGDwK4ymkL}y*69TAf4Xt0RmRreL-RM$Bi(`2;tauvedqaVnA
z=|+JQ8p{?q<H_iW&Ey+X;5Zz~yp4>k4oQ8FrtVB%#(t)B!9J@to-~OgHb_6>C-1qw
zOtRR*p@ZHd^vi+5Sj4=X?~~N%H~9ZV%6-!=2yOQR!o&KadO=5nRr2y;g8_G|vMWNZ
zlro#R349?)x8<1@yIR{H$u>(GoxaSaaN$If$*|NU_kioQHTLiRY&T*oT;Z_c3WJtj
zvVhwyEh}!zT)-gq?VzSJbhfkFSIE{zN*y7XSY>&HbL(Efr(p1a-B<hOq!qRPw1u&v
z;WFKyZN(NsbC&jJI~5%v7b#oTpOM*DK~kAA`sMv;$GW#6SKj`N31AB4nzm99pX2SS
ze$9brrkw`rtgY@(JJrh<ukFu{X*Z16x_I<6#0w1837-<4I~X6<FaE&ojLJM41fNT-
zC+gr!+t{D&OI&!$J??$8sXyDuNO5Er>a)#(ClZP4#Txsw^%sfGT)1u)ZpnqaCCJu@
z0@zw(1lUikL*r$&*d9G~aEf%sC&Wq>ZLd&w?@z9YIT4QADMQIe<J>6vv_gRKJh9yW
zN<zMpRDM`|Yk*Pn+JxRx_>npe<fjzkWLT=&v>w}?gB^-*Q6+q6N8;}`XNNQK_mqcb
zAB+B7v5MyFc+5A~TPFWRG;4jPt;0@6v-yt-uAV*>{bbP|r=#IOKl6Gf#_Bqq&)%dI
z?pckU^^NL%Ha?yIsGpl!-p+}N%}Aefddjm#&j(m^m8h8GZ=8<V$7<CB`?(ww)Y{f$
zSE3Q2HC0<sztnn1hj}$7r9HXW`rYK3piE&mxgJm|ID_BN^r2vP509IoY6Uj{yA_We
z1no7Tce$-eX{se8y%T7VycBn;_`5hxMEyOX#8QNc_X7itdp8<cKJ8VM7(=!spSE4E
zXB)Dm`Lu_=N<Q#q`D_Q4e)+HSKe6Tcw5M@A!DwtN^4Sh+^)qW_d^gG7UzN{xyH5h5
z<7zcUn>K7~0}hA96nBwz(L3oGF<aEd16}kMY<+;NjIkZ{Nx7E{(1F+%7GWCNrg&Q+
z#-2&u9B<*w&Jw9Sn9rhyP+g<gK+T&>=gk2gr(0_YYB*Q4)_k_3(Pp6-!SOkt?O`v^
z0G(k4d-53rBf+P$jFcL>Qv+@2%A=(UyjY!{yFOPVb4tIE&u6T}^`I$;q!9QpBr63z
zt<IXN@prfD2`6GGiT}GpIo?fv0NK81!+C)3V|1FTuad)EbBOB#g2xe!j_PLEVmOjT
z+HJd@wPR}E#tnwkSa0h|y?sLC`)7A=IH`$l@5xhG!A|LYJ26~#dhz2HoFD}nhgQBm
z8}wiqeEL6aHbyW$fJe<afrZl8hHXwiw4tg(GC8jwaKg(mtP26_+2lTbF}n5ZCyo!7
zg~WCi#}y&r=5f;F<g;^RONhe0S+52%efYoHHFdQ;Y?RkE#wIW0cO%*oR!#xYn85~;
zN!^Oi+%wWG@#NA73?uc)o%o0uz@RG(xHJ27dajUrLV8;-c{Ma?u9++MQg(14Z7~n_
zMiOGxE*VJsnMhR(xdYkR%PO>GdioapYxzKuO&i*5Mf^@RPTwRe10a);iZ&2mIFM~`
zPhD=S^|<igQ5dZqNGyZq3uLa-JH&<otJC8a&zQ14`e_)b4#Uni=))~jwoxsI=l^ZQ
zHZ4B-)q3+lwh@Oh$&xI?8e@=X{;BjEfN=7HXSQgJjk_DIRZsm~A3NfsJ~Fx59*vS3
z+Eo~4=Rmd%F^fiY%3if0W35Nl<kY4CJWv~9#Z?bacTHJAPaYjy()TXn!d$BQo*Y$N
zwm3*9)TWu@G#D!P4J5AX%xyZL;e(4qrX3o{crdnMzC051S1s6#dUd7Xv1pf!8_ore
z$6Kz+UR&*iNJ(ll7lQ0$(0V6+N1k8gl)Axb9!7IIp4+?<98BEK1Q9<wPImdT8V|0Z
zsNQ%zTkGwA`@wGK)F2fQ_Je6I##bETmK$w3j;pN43!3wk3G$)_)r{NWWuc9I4%G&l
zDm9NNg;)eg9~j}UY7C78dlH=>G}?Q}{p1(P>+yLRd#5kW8-mZ~_x$*FvxZz`ow=p=
z+t6Hjj`Z6xjx}}v#qJ1I_Yl7;#J2C+`i$MvOH?p+Kb{uOC4X)k%+@t!(&3v_Vi?R;
zAs5*Q++I4EHW=3rIOVdzY!Pr7AQ)l!V0Is_sjZ2*Br<e`z@R<%plBSb2IaL^^NwYP
z(z$_Re$`;cCp?A3>ROM)FCjh;X1gDh|HIZTwxnQdot|3pj9o9}iiJr;%ngHSDc85{
zvW@ZQ{LhAM(_khOAd<T9xLG~)VXC!8y(cirt2{L82mH0WFg~;lUh*WgNJ(ARs+X>3
zL^<6NDB&VOMJ?MSlu)v8Tyzd*hqd!hd?Uj}BLH^w4C10OsN0nnrQwRgmEui|0ePXx
z*80Q>@wknY0!cR=-MSW-pu(!?zoY0A1!J11H9%1emiyEIVE8JdyqnffIOn+g@|+zC
zs(HS;BZHabfvMGtWW$bTVxC~Vo1U^`0sphZX=NRc5i$GJ9}FnHaJ`f9^ZaQvM5i<!
zcy+&>uF(yifGBWAbn>{4a;?bn*+mqcHYY%FHa@H6<vFL%r~ryP0Q`kTI~Tor4oy2h
zn4PjNL8VAsVh}Y;42>h~VgO}e<Vd&6dh2>tA`k-{r(ekyw1Y?tL*eHEyBb7cqZ5d6
zEgmLETz_?aFdca1F}o2Ekg})jrpVCY=fW*L`X060iz9aBzq<n75poq6)7>C0^-}C!
z{5i-Ge_yi&3-x0e{%u3qLHXztwqz*V$6m0YWJ`y#ee5gF>+!&^xf~B=?7%f3xKvhb
z`B1i1fB6YslY)cS6i%cohcb@RU#k+dYA9`%j_o^}Z1qsK_1Zn}5h`i{+cy7!trIY^
zDzWwJ)O*-B;d<)@A3o`c9&LkQ`i{aZuyH8udd#@CNv)}uoAl9})w;cF*b@df4rT16
z(enf~>pkj$cMz3Z1Om~?J^8FP=!kWbhmPpXbLi}e-#L`oZJqJ^XrW|#V~i=$^ql45
zMK^9z`Jrq}k&lJb*yX)~`uF<~YN}E|MRmaFtML%(&xwFk?4~xM$msPQrv;<Z)z^!A
z&7q;R@b<l6N8%fbD%ng8Fv%K8G%Lz)$$MIp;{wqb36D<b2Tu;*0(Zb1%G;@!;Q2Eg
zuBU|ve)4uEo*K}iaFaM2PjMjFY+y2iOnFY7Z)<zz*;bni%&2OO@B4W{+=j96LVQ)p
zk=Q*DXlSy_nnGuWc17S|*Jwc&hO!-B86`JpDo9~hqmSgbXV)~xBpGL`U<VcquRlIw
zH)4!rP{q+I`<pqxrLGBI?zP*ZQu6S|vpL#UM$v)Yjgg=tMf-2}q6bty`2QdU-4BH3
z&r@y0tdRCtecP7i`Zilq$W8?ICf2^BZFAMt(n7X7he<<~i~-ByOE}jQwE&hE(w3*5
zI&5r3A#Hi*Askh6wlZE;yNe`gU8V259N6kY+7!t(f=#_Pz65BAP=KPXE2K?6$-(Xs
ziLm&(Xn`{&32b;?uP?kbJ8o9O6kq6ss%<P}ykM9dAW&peA?@4!oprXkka&iB)>xz3
z^>vNmOJ&VM6&sGV#8-`oey!1|Un0EvsdWUJj<$QJ!cg`^<1#n9&VWLEu%Bdm<5Qo+
zlvpAc2#gJ&d&~=r7u|LdwGiKOM7mkA6m%S|xRg`XK$F+izjJO+L~DeZD(rJ#^fXE~
zy;Lk^I(i!N=*>d`353b+btFDR2KMXrjNoF0kD$<FqH#Y}B`55-2=xrIFle4AWc(#X
z1f&|zC!;$qkQ~7A0cO1#Zl`0~=|9&CYdjNe3+{??R{f(>PAUeQ75LVa-OTl4=Y#`s
zZ(bgFu8>&IV^7(6K}cyJy4gJ!3Yjn@jX3-FVj(+<LZqyv_P;FR<R<J&JoI=%Dq9Fj
zIojA-u4V+Dj|`pH0!`1$=U3o{$?XW=8Yn7Na%#muQ<1To0eOTR@4a?Q!$GN!vczsj
zSDuEIRdXl4>!RSI-K}AiLh<ft$qXW3y4Cl7d`Rw2&naocnI34=Z}QNSg@^87$#7ct
zRJNVArNi0&NXZ$yY}s(O+qbzN9A5Xfyw;vbbz3o<oxH>IAA2DeTPX-IJuhm$YFKBT
zYeQELC+?6GgSIw4hh_E*btJ>hy5Yox2tsq$z`E!PXWAZHU;EIrd)kJ0=*qb?yiq+Y
z{E++Ers0hBa?eTEH@3*;;j|>0TdZ+7aU>i)qJ(&}hT!qQPT`i}Yzc26Lo<95BDKa>
zOgH=?SGJCL26H4nj5K=!Q$i}J&=P<5Qd7Y8#&~cxM2xdsJOO1lVtRg&RJmkQE@&VP
z^XJP}5@uUm3i6|PIHZg`+UXe3H`ds`_|d0kS`Z0UOV{V<upGJTk3>rjESp*E$}3GA
z3zU;l>9~*_qDI1rKw^X|B_+sRS~wX{THU2XtHVVd&H_UFg+h{`hwMyD6}jlhrE@l5
zbNc)7Gb<!I2%5BW(GuzqN-?K*)HSDCQp`z`5i=#x1x<7c!;Y}Ti^JLWCKj$p^23Dl
zWg(g>MY;~+N^L!1)+yUUOecEyNYi>%G$7u`uIb6+2(PQ#RQ|}biW2Xy^+B8!TkmE}
z&)WZLw=}A$F}od*oQWf6--$*|t#((R6V{29|6b5|l#s9cF*O8Yazu}0d+E7e<i;f<
znQ;A;{kC)@ZDY1HIq~9ed>jY^9?3-KHj2}&7|HgtD_7C=lNe7xw;=G@s*#KZAlh~l
zG_ramqv+pdIk&aZ%zcu=d3Bw>!$}FRO{*J8i?!r;WW7W1$IIU~2&w*G9<+^t?$f<w
z)7=z*<H*RSjlVg6$yq5ZOK#+8j-lgzHvfUO=-qR@y+hVIl3g?Kvc(wDF_QNFQ%~5Q
zXt>>j6gcHOMGUwN&O=DvdxhAekJedEcoRj27TfZAO+t4o1b%3_@O%!0dQ!UD>PW^G
zI|$w<Xw4?{VK>6(KD~ESex(obIdGgH2lPnBzLHGg*@&nTtRa;KCO9Q$qUQqcOm<A8
zci&Ju9<L}z$L5WGkiN2D>XTYB^&NJq*7OX~N~Z;7Cuf75(c|!B<wZLiEh!y3VzVMj
z${GBMovVTQV7SZXL{bdiQ9b8;;7$q7uvT0UDe9kVw~O)dXGc*6TwZ+Rs_G!gSDvs1
zk&9qyjEs|Zb>!0j715X}ef%6mav9b0jv=kuwHW0@a$vn4>u{8`b~n^PW&Ypnrk+|M
z>8*I;dgn{yc6{_~8yU1jG86`P7iWi9>2MCTdqFHTPVhXEX~nnSdE(n<#k2<qO|>P(
zY)!Oa@^v4YrNwMFV`(9FKN>U=wu?JS878v4m<c6mDMZ$gpW6z3hhrQ2x0r2dHe#1Y
z{o1NxVhogkWUH($CI-+&=1z3<Ym13ZkYTQD>!Mjt(umd-vt8q%b-lK}nDGHNhpPZK
z=!aD3F=rbCF#NjAS+J?7b(>P4Im)(KxOpT}N+;W>uiA;|ZB_%g4S5GRYFmow_-#Se
z+*KxnhsP|os&AJ>2~+9N4_qr|QReN5aV`7{y><!}6bK06X;1Er!6J?MK^(2UR5}{9
zY~XG4=Sg5th&h`-J!&PrLY;x)u^Ol`V&qIb43V;IBH;2{+obJN&%WnakFL*{*31+8
zjn;|64WdDgQ~iiO!mkgWoW%VZ4G{Z@-`$Q0iVrHeatWK{1&VHC+)fB%I4HdEq@KG9
zCB%kPfvvA{_IywQ9!029p!C^WiO-!Cg3VM-n$-}uwAs0MpiThWWKJl=9oKV2o{w)m
znGZbtLQwOu{cG)_hP3d%e7o9ZA@@vMBjshT#1PwD<H4ctQg*;{;~HD&YQWmR|GRcg
zut<&7%^7fA3kk1fk`6I5H#CTyJzMOi=!b)UihQ~iFi_Bu^2XhqZ)-FSTn@X7F4)}(
z8X=l<nbyWDV2`nT(LBo<Q=Md%?ne`kZ1r_(rECY|=ivc@<utiVO4-g&_pGI*v;y&-
za8)0nOQpm$(Cyjsl6K*{>_a5(Qaa4Oz0OvavV;A}uRdn00%U<b?x{pq$4ieYWD?ev
zGTsy39Bh64f)}8Baa&yg?)a9ik8k0mq&>F{@%xtDwlTUWBa8a6x~Y_yOsVLmsm&VE
z!Vg*Ujitn8I*5d_<^bn+a!_dz*8k5pziF)+o?9;2m>n_nK2#oi^lppm*id7aGC^ei
z9oG`T{WvD}20*_K5?PniM;;jlz<eO~TrGb|A>fm~+DfIwXV|#RpcjY+yQ;A39wo!H
zW-qIB%o+1N+owj2u-|Dt^#kBgOl9o+&raBpc-u6JcMG;=QOR9_?N}g_N8mdi-*LB}
zO$N{t@eFbexV5m4OBr`f#&UO6&e^GG<RvxzmT9Ne$SwH<q4{m*jAptziTfs-_H0Z}
zsY0gg*_cEkxuDfKeZTOL>ebH0{I_Ctj<#4_Ih!u1Esn=1PLo~lVr@XA{JR_j_cFFX
zzM|ofvU#j%3!0{y+T^N!*j?X3zH7S{ok4)@Q+7Qbw!*GS4lg2hv_0EoHv=?UGSBQd
zVz)G2PXy=`xE){n;Y^ZSeNB3Ftj|J>{4cP8dwQL+XWrKuNzO&g6npx`wC>%UK(_c7
zv%}Z(dgOeCFMaWn8)^n~-Bet9F`G8_KUbi18Cm{fcHj_Ogy^{9#k7%GGt_Ufl`m#H
z9>**07Pcz-gA!~0(N+g=3?6UU+7}a>z}er{y_hi+?1a=DelanL)?Kz<2ta6x!yx2-
zXd7Nk?0`j59_UV<qZ{@8Ru4?tBn+um4$(G8ub2N7g@0rGJ^$~l+UEGXly1!d6f{ZJ
zm>h{)1B?BSSVv&$$v*eQU}?6CYN?$XEQoQC+unfK;4xS^Vd2FldMy89wqel~jHG~5
zD@6Mw6qhs?u9Rn7?B{sRiNK9H{^r_KwS9qFbG=Q+ug`h%xI-F(AMc+3?}%vsjmK;z
znj(XuyB1wrsjy!2*gq~R=E^vpozUZ0%84iR*qViu)9loX8AoHhUTG;s^_|vtDJ{*J
z=+-l2*jeGT_Nz7<?NZ_#e3;J(nD58X@KQW2h)Au*&;gtmgzXJ>K@={{pJARa>KRvh
z?v9sZvdA6py!lGZ;!ec2g}{l5GF7`8e{-M2&pQ5g?*hB7nSN>iH|&OJi(WAK``lD#
zuHD1aI)+14mtK-u+ifk4lm`U0wthgC97JJ#S93rDgAZznmF@)<>Lb7RWCuA`$ZK(5
zBt>GLGApMonmnVIl+!LMV;fmoP8-uDMx-$8&gD!8&gS_Ll_2?t{ZzE&(a<$wREfkO
z74SSeimfbX`xm<qf{3jur_G4V^6GMSVn2s!d~E;$V7k26x_D-1>~=MT)v5Ki-acC&
zZ$mFhW`zxUOCfvPSWY~N`#Uy|O=X?-z2UyFIbOTCOM>cv<QWxGlCS1~gsI5o=$63t
zt98~|PCO+kP!PU40)|&~WETevCN9)podKgAM!-S6SGai`buNfA;O2n5#@JW(Anii5
z#LDL`!ligh3G;CLDrc-U<#u$zXd-Cq&VVV8v+WbH@n7+ROw;-ZU1|>%l-boW4(PhQ
zy?vJ*5p~8n5j;o8OgR(MQa~?~W;!^J#lmPJ;mvV@Xexl`35~+bijrl%oH3@j&Zo0E
zMvdiyV2PmVvyFBph`1HcRL<e(VE$)=HXE(y&tR=TC+f9fSDTABh=wC}UTBa!0pAg!
z(L#c|f?bSm+K242$dbZae{NUG*?k%DEZvd)dPdNlC&!t!1p(#|I&W8lMD>g3!?i$_
zbc%L8&|@cHt+gAXH>S%s{Rp$0wH5T3-Kx>RZIicaGo6aC*&QK>&S?bi=DUGGax8$=
z-;4h6cC-5$W&LQfC|A-ZLV@;u=6gvcZP}XKG;2#MX=jglz{RpkTENJKD7vvBS7M`M
zDOwS&DDgaCE92*4sn1pgXjZ(7w5uy=PsJI%Hej)cp2>3)SXarmSc*AfZ`Z}QH0bu-
zTFKV#!v70l2%@tMm5eXoneRoMHwOHH3K{)vQ(!UwGdx2Axo$6u)2NRYzMrm4&6P}`
zaF-pj1Nud<CB8tQ!F@9b0v_(N!?qRT<iHxTJp#xc@-5Y=_Q)bcr0)$xb^<o1K_V&Q
zE-`8O=xC>V#}|Z2E$rPz%ThFOS7oaPq;9;?Cq$gBq_*4_Pn=S&X_}5^o^$a~;JTAL
zzDpEGf<O$Ve$r+%4q{s-?N}u{Ide)nf)uSUURrA>G%{2OoaHA2Hx9Onor;DXMbF4}
zTG(yFo9&D~^Yw-Wxsq`QHjcCZ6nJ2>YKDm(I-e6^pXMxwdUG+Z{rkUS=S3>cWxta7
z=q^;!v9PtxE~@`_SoviQw88V=UkR)^5s6Gx!yVj@u0{u?AxgIdtZ{P3>udD$$jb3z
z7uR1+raSEVUFTMO-@&}zUR<v3R{I~Q(C(<O24v8?8r`^OWkx|QaLH=IFZV@|WY9FL
zW?O($!jov@q*=|*KTn&rw3;nt7uVQyuB=*CO`8s_DH6nTwco~VY(+I~XTl7~BY|s|
zKG%(?Y^$o7Xbl8Rz^mgqnMFMZ!rFkrlB%Q(JkcAE)jq4MW*ZbAAJ15_UX7basam>0
zu<`o?_Qq<6-(%>$^s1xih;I(K>_BF%v6}IU()`DyO+x9VU$9Ximf_r0y_6+@_JrBz
zL3pmFc54k-lNeXI9t+1EcwA%g*&`gc`enGYnw><PZ7E=Ylz7#|5~q5plb8#NdUj7{
zH6Kr2#0OSdh>ox*b2TlgBNSQgP+8Rwi8G?7iTD<^IgUHFFP>0&jFM_~T5a2rNZ8la
zw1u9D455Bqb^jncqCR$Z^-r>*X95b-F$VuV7I+WL|4(WVYohl4gPn*mqjoQN?A}Rz
z)rR-Msp#A9PLnQioQ`MuEKm9E?TlzoKpT<<$FQ))JD(P_(H9CbB;s?6FY%o(+FUj5
z+GkO7Extk0z^*LPySl<HD^<IwJ+Ldw;D0#)l-SuCzGr3PO1+>T21dp_MeSAn0LOI&
z)wMva-qU&8^?2}!M%NX?;-m>CHD@=2eBHkBZUwG!-V>~KTSV}*nr_s)6GWVUH<d@Y
zE0PwGkM6~ji>U~n_k|Q#y6>5dW_Qoz1-N81ZStC@`qFs9z0TvxmyM>Sh%;-qEg#MJ
zgQuNH*2)#5iMVUl*viptA19A1TQ!>YP=6{^(^X<#SC6LMH~+ICTPsKl|1;8g*t*ed
zFE>Zwle)m8vFN$jZG8=oUee8M7)@J-Wu5}I(Z0;t#sKbVFgFDxR}x~)hiz^i4N;pp
zASa&2PMJmuL<kd74uo%2OMr6EgI2vw{K(xnIs%c=ew535V$2*4A+&dnX8bX&kAY<O
z+8cymi;y-BNt|19pioQCZuy|$<8@Y0cg;l)4)o(W#>&^V^q>4z6{(wuh(pE>gvS~C
zK8<dht3sy(^rPRkLqP*{8eD#k!~;OT{gTbZ*A?#e$Mp3Uk2*a*s>{?l7Lun9AU=f~
z#|a@&O6e0%YTS|4s<%2-wNnCCo&Sh*wCZ!$fSnNlbbg#>XEpf+cjhJ$h$!LdAdwaE
z(H6WR><%^;^H3fb9J2GZB^$+PFT{YJVWD4)KikkwT~-&Zp1G%DSA;`@Crr1600L3A
zA>Ce8b0SsQuy!r_n?DVUxnBGGUHmd{2+T;)kHedy1+?xV&dB%bE&b@S1XpJlCvIy%
z?)YYR1jxk`4)ExNljYy-Ui?ihnUdXCJ9kkaE@~|8-pC$hOUAN2+UIP>!_=0Jr7c|d
z`Se@1EE+S5UYKY3SjGoPtCqJF@re7|pmXV!W7#Rl>B-V1OlmA`q;QQ}t&dyK6WiLc
zv?;4-O5BkY9n^@ei{5<cb2`+;)6U4<vVJUagUdhcvJHWt-{5WISjGli<3O<BCjHRe
z+t1CzHVbGy8vVwxY|B-WQ7n07GzY+;u3sLqmT2xSSX_Qu;|&FcFfxy29E)T#enZYi
z=LOw5<1<(SW_;T#Qq+(7m7I{pvBdZud(`rxxQE^+!(1Uq<2D}G!jv=win_P}U5$_1
zpSB6T!Y`%DJ!hYgLfFQ2VLF=DH`$>;uRR_1N7SsfYm{a)LCq&V{oi&h{@hO{q2n>y
zz5>C5CxoV3xLjVblL66L!ib&92JGI>c3K0byxm}hqwY+A^(@C{Yfb&OLNeOG@VQ-f
zZY-0#llzi8jQBdA(?a0}_^O?cg~Fm>gLvvfm#o#@b{Lu~+Al^QzH|@~E^C%+zFuCL
z<Vw(>J{b%VtQG<Y$+Tx{X?3s0_(>^2E@^d<?2#O8*W-KaXKx@){ioefi=CVuZc@G(
z1e$-t6?wM;BixZm(`L7W9C;M-cSKn((>N4jCc7J;m2T-?e4FalV{y{t;cy7Ev+;B@
z^6FZDw|SO~XFCB4bvKS6OUKhbuAks6U8a}JB<G+lA5UAEb0=N<R*YwByB+fq38u%>
zUS?|!6Sn9qX_Z=lCG)>}Jnqg(u2~xoP#-~Mo%-T>Get6ny6B6y*hVosk7sv;T}Xr*
z^t1yvl8w<XP5Zb(=UmmLUodke<HqLr0*g)_%Vnd0L(TtV%`xbK-d~=#mhntfCMSB;
z4fli}tGmyq-?t78%CiCygB)L{p#08whfL(Da(`}n)fV9l0_37C!UyGtYbcHM6Mt!i
z@x-dUcvVT*<CRo$E?EsydHyD%9RmhZ>wxVWPaI;DR0HvS&-@2=NS*$J|Fa`C1P9pq
zGtsO#%w1W{m_DSz<I$|!eGpFQ9VaE9oeTt?!MaW0vF2(U1+~EFbO7~$>@(xpjb-6o
z7ia!%qny>SfSC-Zv!Zp4`D1xG7qGh68FT6<iS6Xc3k*j*^ZW%tbNh{7JuYgNp275S
zxke?wYF!!Ed4){b?NxFY)U|8Hk;z~XS7VU8XF*<z&Rv#?xR~n!5G6A)6T<8{1Z7Ba
zQ$6^Bj9!W3SZfToUOY#}GrrYRq{<&kRyY^h>wx;4pZ<4a=mW?q_iCMyYsBtrTsv4N
zHjy@HUy%*EWFp(W^opX}mQG}+smrsD#j=UCtVZxZTt1ODpL-vZjVpfdMYvcQ5D~KQ
z5UKs`=&YW&lqEhZlXMC`>$wHjPNWt0-EY{s0MQ8Oy689YP0u~^b6Y==?N@vjZ9{b3
zJvxAXa-*=W@4@!VZk@;^1X2_X*k(PF*X!S`F}_7L3KwhE0Fwt518JGam=LQ9M$sB0
zz|W~-9f5?`z<7~;D!8PUD_N&7Y$J4Kubw`i{#VNhXctZvxLxbh6roTE8tmO`C5?p>
zh4R)`T_l<pF7Of|h))wjy9qsjE0_1hs748gB9%{c{#=-*4+)i?BC$^KglK_iY%>#?
zv+hn7-lDhkv6u^@xb3)bcQJsi9ItQ;wv&3**xQXI>Xfi&(|Zk-)6sV4_FZ-++V18m
zmwXe>)}}C(XP6D7(0fr>!p@0095o!p>IFR{yySd*$(52b;=)8GN|imowZ~WG>|(UQ
zftDlia!lZ7uYdM^yHe|v3Wnr_pU8G`==OZa>ZgX$94RYYgw~i?hwdYHT|ZQbu_H|8
zhS1HsX7r}sBXXjdv0F7FRFIwnm#)eXX+Yv{^u8l@SMv!waJ8A;n@~jI_mht1XGxPz
zQemz#_XFZfFLjztrcLNkl~{=<vrST8=(eSk*=}y1%#GlCvrIpDHXJfTFV_z}>_{$8
zD<(6hl2tCi0{p72j80gR%>Sy%!~r_jJZY;ZGYV3G!&9~W+}2KJo7{t{Jac0|On99T
zYk`N<1%xRi=k<cnh7)zNhHcPKj5~^iY}Dt2^B>!$cnWo}ZB}#VD5-Q=<78rK2&V*8
zL}P?UKR#RH&7<G>_M_H1nQ@6-MJlv(1X9StwkL+m`M>v3>zvH^0X45ces6%lFig5y
zy=(LMDa#8a7H_Jc3N<Iq!bdEgl77OD={{KO@Wqc@hccnDcEN!6#nbMAiie9#$5Z=g
zx92=Lq~1MPA0x766JtRRaeqdj8nEyjQ`gLlS2g2=j<Im^ciD;BxcpXkQcc+zY-G07
zDRt(af8BOEnv+p)!p^9l778_=ji<uoc##m6*~!G-JD#(1@kMz6>2oe9JV31<q88)D
zR!`-2A&Ak|6vfCds#gL^kd`k`X6F&XrRj{lq87sx9K_>8s3WA7yu6ED4TMI+Z{wO4
zXic%lsza>SBrLqgg1-@;Alkyn?527hng1ZA^|+;8l1m&HF1sy6ipY2z#DWO1%XU|j
z!`mJ*?cWoKJ}R{Tq=4e){$wU9rhEfQ&a|ntmbqKAy{8f_zw=MFbShhvuCb$L%T%_O
zF<qh1j)vX+wqh!6)=H&dE2k23NSp#|u^^Ca2_Ns(@iw_9qraZYHa7gobGnk+ClL7x
z|Ew<lrtCU9IC}U_-M4LnV0IV1_SVL!i~)7k4(8~Djn&TeZ}U`QT{u8;uWg*lW|r(C
zDfquxzc_1hd8#GgyJCgiJZr6jPv$)5XB{F#1J&_tkH+NrQ9FfQgnsXhLHy%CKK7F3
z)R#L*xOZV*u&J%%T)r@sF}CpfNcL(gMHi%I@!XPCf#LWoV6(M}7$esgEcE-Pva<@q
z4O$K~9VoG*BX<lOAINcL9tyn3S4bKH4aogOY$l*VzbC$F$3)D^{7>w-Mz<xo*_{yn
z?Jl6AzK%8V)T4GPKFo7z0OskKKOFk(O#C@o=(Dq00R6pQK?4zTwl?Wy3Z9*dzum#a
z=7dMf{0DYEP=q~lUArKp>Ol&>k3fnwMUu)%ymT%H9Ey-M+LeIg^%vmV3mQ4Pw0G>P
z@TrwsKuJB7?UpvrkL~7kyB@v69GJoz0<qI`C*IT}et2Z*657ur{ofY+n#}nhb|<LO
zN|54R(f!eXuzUKj(Vej43Ax4H0_>%1Um}~ZXqLQ`mN>WDp{gx?Dchu2oiIVgBS4b(
z*z%Xs=KkhaY{g5p;|smn%9pZjfsmb|`BuG@ZK`V5&@azpKYl4Y^c#1z+gg2qhZZ_5
zvSYrKSU@8OTb;f|lEJIo*6Rnz=c{2uAk00K>D#EcqpZg2sBL;FBe^RDhR7&@4N8sL
zS+O{1H%5Q+?~+_V$oQ6X5u_!?#sr|Jj=u2vVRdzA6cr-R_vp#-=!}nWsIJ=HpqfXQ
z&%aC3kbvRD>9xF2*-2VuRNaM_GB%cs3NG-L0&)_jur@6=nEUK4G(F$52>}_#t%e!i
zr#3GCq|;(r<08r=rOZFH_{+#-3OYR!i0*D`Yv2CNE}Mzz-P&ZwVl;%5_1W?G`-vx>
zsIwC<UCJ`w)>0~^=rqsPdoqx2DN;g&>+(y96M2D&lwgbmv7gP@>7c}OZQDDV?2PE=
zs(`Z^(pC&-9B8xAEzS!Vl;b;iO8qvQ(=a%tk{RN>7VR!K{9Sk{V_`MTzE?Q7C<;B}
zsVN3}Dt1|)<ve506}9l{DhMB>;0qkKpiZ-^g6fYU#oM(Sk^}?Vbs@R4)oy5T-Dufv
z#&RQPgWZZRn&H=WJH9~oikx#t@M})gewy6XSV<1TBvKi1kxu0<4_CV{v}^nvwb#CE
z3!o5(|G8|mM8Cm@Kp$D!mhOvP#|32BKJ8f4f~?c!YM<`Z$ysxS8to$80lac${NOIo
ztULUhSM5s-;_-F1dSAv!7S6Cb*6vIE0-Xl6)uLN`2dyst9N<R2KK`tB7i>fPS&vAv
zabLz;u&^PG;vKRt+b@s<pG}Va&H4%d&(UOkY77uN>i1Z4;I!$h)}k?ag5P4LNc!m3
zeHl0OB#&I518={zyXV3l4H+dlnknnl4<w8w^_}gFkx>eYE3{s^#)Qeq?@N5o4=2<s
z1@(_EklAuypi+E3?hCgXt$mp{Y_W;J3`lM?K!GiaqE#opabLE|(S!84-Jc0Lq#j1)
z({?0C(OVd{8R17PvMaHU=?6q8PsMUv<mf3@=Fi%RKo5i7l%3SO?G^X?Jf*p*&v_6J
z24;av9-Qd5_ocIi5jBn4*?2%x#MYXPv66VfMe1__nyS;VXU@pEn6tiqJ1?4a<JEmZ
zPbB2Bi-O+pj9u0%FUxjCOQM9QZHj1E2$=btT@~W(;ooT20#ad~0@ijt9uRz}#&jbH
z!1_kvwFnfip8*f?z9moy1mu@N=N06J!`L06MDj(utNFutifpZyl$o{rdO~q;wm(~x
zMA%jaY{~w#J5zzpv=f)A0V;i#>9WOVH}>1|{n~v$;>cgIKRbSRvgB6o&sGld0C|)B
zoUv8=vyDc&np^?yusS+r?x=H7?(MO)`xCqH%h$U7npdJF)VXVpu_y}^KDz}kF~4Dd
z+Wmvh*Eb4m`#^y$ZJX3Qg~@>Z=0LZ1gnat=jKs&Bl{i_)-v};5PwdamB3H;@Z$C(6
zLYPro13estn9h!XMuQpD`Fr+f>?1kKx<J+$<DwF-?Trtk7HCe0)l)#obHn8KU#e*b
zw<1=c7EU<<3xU+Y2r)$^fyY)dO@Zb)tBTxCXwFNwxlZht1sLnu-=V%wGl$<79$|so
z$owztQ2goTbdCr!H)W$#wHfuBuJLZTKUN#JCwcl0J06pi?oNI$IT6pP+=MH_Psa0h
zjv}h%keeW$|9O;^@AA!Qjl$C?L_Iw_qq!vgv}?uA3Ncc5!#^q3R>R^PcwuS|sH`dM
zr!cWO0YDn^g?wH=)Vs<S$4)_Xw<w>ki)yk9w!}A=7Z+ozU5WmjhxQ|(i<-$ie^rl!
z?Un3WV3UPw*>wSGZLu4RZ9RDGX22cnBB>PHUO4QAH#3mi>UM-u4tGQ>SBq15eEuh0
zb~m82$RYQ={pq0G)n@kt6mXFT{WP-!+2(Jcs65H*4`e&Dx0dE}X|&+XpbQmK!m<Nt
zd$Ub(rMKk=(q=<e#u@uSwolhQW-H?bb$7X1AIKO4)v<bPb-dd0Ok<O+)hiE8vvmiu
z4MXmCrhxwH4kYCNX|JspZkQu5Gf*t%Kw=WPUQ%=FOK$J5ODrXu4rG!^O|iijn-65{
z$ZMX}LZcc3A@VR2-+UmGO&U1WnD7?$iiy(G0a&Y=5FUY!&N|e0+7DjEcaP}m<<>aK
z4`g~omzh0jrx!x9y&~Du1C|plwjyj;KCo)>D5QeA+)f%O7=I~VdRVSiV@6noB`hZs
z@rK}MB=tpz$Zs~Sp;S<cIB^dNZYhtB2A>k)3HSIvX*hldm^mO0g6-k6@8OO4^CV;+
z6M%YRYv(_(<I%^WbEusNB;0?|yF3{~#vV`WQ-SPfZ=hm2tzkTa&&wG-kwd}GY9Jk>
z=w@d1xP6Q>;9NYyBOx`B7kZm~iq7k?(uHr=1<gzY14cC-DCBw}^X2%vSms!~qIO<j
zm4uxF)6G@NuIh2&e}Me8K({d@jO&Xp|38j_8yZVHmB4NaFo(h!?%nYg3$SNYy&Wj9
zV(|@&1x#RQEuxk8cVo%30qmXtMa<@Xbwj>cvzOC8?aBLH+m^hX_USMGJ6HRqFQ;A8
zhthJ{%W3;_DBzvLooRVIZ$n+00^3%+oNbo|bacdZxejLK%V|eeus*+>af=F>4ZVEx
z>gbiV$ZoQ=(G%Oo*2N1BpLYm^s#CM>sc~EXa<aK}<#J6PswF?g4S{rG8XM#9uBj=U
zNdn6*viZBWmu(JcZYbouY833o@~Ab(cdj?LmUx2_-ory$U(R+qh3ZlW<jdKCn~EtW
zeH95t;-lm@^&i&xa^i7COkwQamlIz@YcyIxlE}p)V}}-4p94NzwU_uJR(?V6pZBAs
zr2g?)b6?bI^o0LoO1Z1o=mthvsQWZ7PvkKDa>m?9R)ha!HvkUBSV^B-wIeTQoRpUw
zvnVEK^a<Gesmd}w7UN+nFkQ!Ex<-h|IT3$SUerzo)ScWh?39r5I~cjhP79`=)S(?d
z6NvWUoyiGvR-^87>1<Yy<ToI8N&r12CCc)-c%LJQ-Y_Vg4?G6vPuYdq=)Ac0MU4(I
zEb7f&{yXvzG9|e6!(CL5w1rqRzA!xzwOv&M6vl*g+chDUY*g0+930Rcj&8i1vG2vG
zGbnm9s22*lTY+Dzd;8yx7YIyP9qtHngJ+z)8}GoRjeMO#!_{R7E%#${9aw8NowjL5
z?;iWYblS@-aJ2rmWI8)eNxM|F2jZn)%3$D2r_%x@6$(r2i}6s8u*0o;TAP(^gQQ?z
zn$Fh9FllhH6qB?hI&FDoi+x$`(<KVG<rUM33hotIZC^|s^x$;jE$bemNbAbzk=q?B
z_Mz!)jiMbJ?)AW6TNNENZF$-r);PLw&$89iX+JJDT_Xh0EF=@bT7A*rBYH&m!&Y3}
z{e`WYPQ2;Kuh`>(6}f=O3_P9K9PutmUu91SGTzw9s<1v<kN3hDzNWWib7k)!X9UvS
z@RP5rPg1Z!2}r&%`h4L9doo7vK}P5fVqPfO;W05!1&BxYlkY5Ev{NaRL|zTiGf{u&
z{NsC12t8t3)sx$fQ?;+=_!0@nJq&gP$ga|mwFK>0Q`8CbM2EWq5<h(UzpPbn-8U9a
z)E4yMt%v-LT1U{ezPcSJvI*NA1i+>0;Ab<Kph@x|2$b=2N&9)tgx8R;&P9M+gR1sI
zyms-v$mZLtp>*|-oXSPFZXvh}&Z*BaC6LUdFK9wez#+@4r}ip}(7_-<s=Y}K5kYCM
zw~=_osK(hArGS9~wqj*1M6Q;E&Uvg{HK<1tDH~H4exBHb<{xsn+o(+m|GwVdoZE_R
zpD2=Mr(y@<ha`QbOU!iO#`5A`c~CGzn+3BLOx)~Tfs?btfimeLsDB-Mjlqzj3X-#(
z-IwiXpz0jo?`AW?e9!jn+wCpUk-#)A3db~Vzf-fAnTxmeNrkc4@gSNTS0KteHR{3&
zfPNxKL?oyEK=?NnN4aO?uoN#x3px9dx~Zh%Dju<PO5Jq3<@Y@`NeZNt9O1M`h!{IC
zT(S3K{B7<ua7Hh8cny)f{a9n8C;*6`)iYTNS@A#B2X2mH$IYrs#!eiP+m=3z9=@~g
zA0D%F`oJ@!@WDs=!809_Xkji|b7SqYk7KxB`Nn^-^V6A}&@@8Mz+x1?{v@W`Q_4=+
z1%awXW`3$4Dpc%dhg=jv@oq?)@{)eZxvh+CdO3b!Nkyusiv21EofTBJD}f3DQ;`%U
z*xRoI6>>Urqgx2>R6k|E(UjogN-Ce2YmX41uipk_;zo0LHj{pSI+OgkYtJtHr%U|y
zJ8ie;yX<;Ur)y-e*M6_bAQp}tbwk6gxqW{g<Qqc!bW?5ftopM3L1^LN-C?)%NH|BY
z{V~uV?6%vY&o-)}_jCOINxd{7jgWNnPE7eTqgA5N+wodMaroyz^E1LP?XF;_q9lI_
z3b4oOIi~OFsWM))zsA!r2`gBQ_haBm<7<D50T8|O^OQ2eUde<5);4p=eZe={7hbuv
zgmeUiPM{EfC7tD6L!<V<E1B5o2Nrb#mPWe)zkq!4mFzm+TA5_>mg%|dzLNH(SF)oY
z{|3z3uVj<th@fuqydhuKkC*`?@QG9h`mp5j<qrnj0Tx2Sb*1`oMOBYI^h$PeyP)gR
z0&Xs^q@%UvTlR4LwQjGiekHr|fAfH?c_j%Li7p<y^gUbqN~Q{IqaeT|LJT?~g{4)$
zE=K>$Pv(DOkE?m#FTZbf!U<PjHuw_(p4svW5#y}j_0gWOyO!%~!WkDbl%s^(*#<%E
zB8KSe(HG)2raZOw^2GQF!vLO)9^Dj}CU#SN-R3@IWqT^-aQ<|^ZH_r?cSA4*9s8}}
zl}wWApZ@@#2=3mE(I(w!NCHr?tuX~?;}gl71Drbp?GToYIM}yn+#Cx`@-AT-wiNHL
zRbLDum$iwoln3gmW!CQyn8vgZcGs35wx~T?3|pMlJ|ASkq(VV?0d7E8e?h}yI^RO6
z`90ek4bTI+Hmp0Ik~-jH%V{-?bz7ex4dYst*Q+Pj*`Q_;7-){lRtV%C{pM3PqQQLj
z-#q9u%|uBU<C@xQWj&%bnzgFnjk}vP$JLky?<(tln@|gAfgO0#rqp6^)Ln!231owZ
zcpuQCPq0mEFxxnK4#r@}Cd-L@C^p0JgdKh*oeL=^x(oARhCH%JLjtwYsk~}OMU3rx
zY(|I>qR8#~Ej{nVDS$Ivn6R;C?d@06U720LuAtVv#~M1fIUW=w3%b1{WcyegC*sM&
z1#oHqfuKF-u|X&Gm=2sDE#gWXq%(dyr54<iD)wHi6+&GG>~#Fygl~?$uVx+GMeU3p
zJ!y^oI9g8mE%<EUgWkp+6Mq_L_h5XPtsx5G;9?&Jjr{hEf+t$7`OXveQT*B6J7ja}
zgxf~hK8}ePEZKPxpba<FPyTNI)Ew~af{0wgGrMY^3KV&BxL!u*7-(1-m!ij~)_lV*
zYi%x^CvEF1Set)tzY?q-^r{8BqTaW(@3voyq%Kl<+6G(D34p>1Uw->dbcZ>L+s9R*
zvW;Wkek<tsE0DqLnm%TO=FuzvF3{`1SbbeR_u`-Zdp$Kl!W&w1^}Fr!n2#;K5pJr3
zZD=ie?GN#eRZpaHe1tiXas6>|@IPW{c-Y>Db~}2fK#BcHbY`Aw>3HvGFsNjQ?9bYU
zFdigpN+k!99f-WRfc-_yk$9;jrRJPd*a~v?SM^avDrNjgpZ8+`Ma(1kU-j7EH11~7
zsM^7F3ZtN|*cT3FTso;XNiAYa4rZsZD-!UldEj6=<yyDd(u3LAMskra9?TBJlzTK0
zWLY%eASS@+OYty-RgX-)(J4`y75j3$D3TKcchZA8u(5^_-$Gs>doWrfI}~l@!6YaQ
zpjvt8V0PBGdo77oLY{38J-8-4tQMp?IBKgk7R=XN=xt2^dx|Q<YY%3E#c=N+T;&lV
z!c=-tm3vjJJD3<4yoG!mo=r6+i+x<Z_(jF)L<w%{>|u->_Jp7}Cs};GFnPAMu^#t}
zuL*fJ&Z`UmK`88oKyH+i@#`WNxeX6yB10cJ0iF!vu|;8Vn}i4&{)#;n@4)tHta$ot
zJ1+DMYRNsBqgd$dE*1#;wypZ1e;|c`HtT~fD$>~wL2D&<Zi}9F)wkO&J$0}>TVoU#
ze^{`#7)3i-bF3r&9`fAzy9K++EsT3ad1uKU!3E`?4=_byxI1fqGD$D!OM>btEf8;d
zsKL`MAUFqcJxKYabAcHroFC172eX61l}4%1VoIo-7yd2eU>po0q8UNLE2zH;nsgh{
zOy^RJk8i+|nh@JswQ>L><i~9%)fjd1-Wt=A=-@~$*o0O9X*%F7aJs4JoBM*FznT(+
zLwCbDa4<=eJK(A9>9%Xr0*T6#5gybh+feDUKn?{qtY{3)c390Z<`G4DM9tga!pI~?
zqX!mHTQ4p+%*{*;fMfc+y%nD)1)wL$I;Pg1N6CY#-wswmx%1<qN*_5A?41D9(X^+-
zPGl?C1<W6a_8sKsIjKqHN?4v8$Hd!fKZ<Wz;pB|e2#V!yPV4vdT#pU~4+w$Vj@jdV
zAyM@(t24qE_lV@o@?-r{^QN`4v7*qB+E4X0p=Dg2)s{@eoUW7hp~zQ1l!8jmMH3E!
zl;akPf20LdNtH%jr8OtGjh+wcV-3NBY6&>fYn+6O_Q~RF<OX8<gBt4|+goR!2D0Vi
zlwFKJFTPr}OX31MamxsTkg3*J(60hxxE9tf*GgScXM?>&S^xTAc5m|qsdVuR;(~si
zOBEzo+rQEJtyN#Zv!gb(ByOnKZ#5Scyj-u<UU8%s?RV<-Ij@y`J>GF@Q1$!wF_QYW
zM!O-B7fZMsax{M)qkoPhopw_|uztT|e-Jv@2SWn9RijVWfc-I$@oj)k^tMogF!rYa
zL}Jg1-vsXj#kX?H-G!FH{;V1B<cy@=3>>^>pZ!HRjFC6O?gd;NSsmHG#!~}Zk-d3e
zU^?8p+x}L=h`>trmc5z|Wup74_Jvo|iB}{5WXY@9iHB9SY!B#h4><vCsUCBF#(pti
zw8LF&S^VAg%Yz)m@poki)A;gNv(r6gZ{{$Lmt&6bidWM~Gl<s#%lkn!B_(Bn*x8k@
zW+x`0a|m7!y_!h?IW`}6I6spE1RP`>M*r~Q7u;{VZ1t;|!kuJjXwCAr=G9C9soibP
z+u8tuIvokg9*Hp)hbvf!tF}%<CiiE~9v8r*FoPY}#n)K3{BHI{kdjp2wqBT(;qqS#
zIJkkO6Mcga?IWX0pM71=Q4w`xz(LJ#PrjOjAY8Ce!EF+axD=Hk;Zxb$UNp0t13M<z
z<q3BRX?QisQtdmfG3ZVrdk){|jl7Y5aly?ddK()UvK>MM-vn4%i|Do8lgxWW-md5p
zk|S2M#sm5mE*5Q~4t2njXK+VA-bRMo-J-{y?YnJ{nr`<HVUTz}C`71Y@-6EW;hQN}
zVlN1<;n6`>_g<}qkhDC0wmXLS=yx9bhUL^-7xvb^z>;WJ7+yYTvh5igR3nn&S|M<)
z;uka$fA*s{Dg{tZ=dzXK=f~ErwW{!U;~z}M#-br!KyTTEsMdn2d(=%(QyM@z3~Zm8
z;Q2y=Zpe2)KM~^_xdEmFL+sC9W$d5^-7w$+Xem#vp%@LJ4u=D~QN;EmS|;p@Ay(;8
zq1)$L1GdVHI@^gaA_Wdc>@A__Y@uSu1jXHiaIbkgdT1dF#Bohez1!m7(Ib}TlAVZ#
zn278L!a3>Xu}7k~u9518oLE1KPQ5@1OA}Ax7|%bhv-e`s8j{8Pw5F<^<Nf`3RgV-v
zo$q*n1tQV>I6zOr9_(zqV4;qb?WX~Ef`e7KHn6AdL$&&Dot=x%DST75k3=))!vh{x
z5X0)iwesUvvy;oOVe&*fzldggi+v&jK8@j(!ClY~@GlQD{<MZ}KeE%sSF@W55r+6{
zru}zG)5KBOhh+)>C;jAFu@w9BE6v;%WVs=`5{R}ox7x4uYHNK<dy_2)!xF;4Z-Odo
z_uExbg&Xv!{r2w%pJD^92@r9&kmq+AOCF06vwB^hG}Xe1e;=fUb#e6HP<wWZFZa*Y
zK3vRlb~C6k>dqg3h#}-zyte|i)+7dv*dG^3A1c>~ytys1xVJCzQNjKcKUD{Yrtr^9
zZsvD1%$_2<;Lm!*5rrbkTC7<%LvF-sg1EqqSoHG(5r5U35is?E-_l_u_cc;Na0;x@
zzX>(uflt`Xvu20l{O#-Ovo9RV&d3fI#0k+jln%a3@oNtp%5G%O<K(q8KBW$OmwoY&
z&TCIov+PiI-bD&%)YbZQ2YG3h2Vi!@Xt|qh`{hGns?NJ?vEop6G&Z<5v`^84`WmI5
zt<=}44wXHm&YQZ*BsDpd#0Xqe?BV$HS=QTXfxtdyYcy2ilfC-s+87M>+FpA^FBiV=
z#$;QkAIO^I=H|z1t<lO_-J!5EueB$lw?Z!|6k<T=1zC_^6E$`q==$|&gODk5rhHxC
z>#I<JsBOK*YtNnxSnX(DFePiOg#p~gpAsp$us|1V^Pz0-`VGD2W!#`fsF>)z@leJy
zNE%bC0?8S@HPH6}`sNrP`v1J`&~&6Eur4pP2w^U*Kc{NME_ItyNrn{K8hD3aXPfXQ
zcWvG}L=rCam`S)BM*}n(9QebwCvc8j9w-5y4`9WT7y9lzlnG$?$BtIG!=a$opgf5p
zaZ({t?2VC7c)nuY8VNfnOuxB!jh(e%eL}I(<4^KJF-Mj&*WV0^ju=H*sD=2r;I3OE
z@i!UJIjZ$#TW^J!4y?}l1?ev5QJ&Ss&Inw$F@bJ^L6dhPzzqzebePh+O3Ev=PpjkN
ze;cp^@#KN0ERX3L8ty4~@X)0!r+_EEW=AJieL)-SC2hDJt`RnQ6&w+TJR;YvA9l0`
zA4e!Q;!JIpJb092<E`ipau@BGMl@K#%j<1DA|Q0z@o3pKgxL3vmJeD>hk8OI+DSF`
zALyxqz<M&^P!@_f2rk8%=ETBWW91pqPU)N7==k3gxiN-PKArS8df$PQ!~5|>4MMlS
zo)JY5H`uN}j(yO~*>X0p-obF~r?nB3Vd%43@5r~VyYWO~9|oq)#c{vH+c|-a$b*l9
z(8T3X!(>h*@Y*h<K>avSO2K7>@oNBCot#fZP7gk^3jvIsgS`2vX06>bUHBL-s`IDI
zBp|j+F{UTJ`(3*%NLYJcv|s75esunS*p<Lzpqj^^Wxvk4@kHDUH59{(>NoKO3-3oE
zbyWzbU}O%<-$svJxiPyIO_SZ>U+j0HUB-LxHoPu+pqJ;8ZoiLjT&<3+;)aHU{)emS
z=P?(fUV`xE;)+Oy7K|;PX7UdK(M!1`KO(yoFp^1+LPPe)KyW*5RJVnF3pP3TxBQbj
zWQoA!?}$n*qvSLG^Wvi`$M5QK3&FPb7qO?5hBKw|xEGXk`z5;T7+jFb?u#It$p<}q
z9y-y#Mf*Nn012#mEo0cV;;ON5zwlaiFYQZ?G)rE~m_JS-z4(CP4q*?xmd<Q6VXgpJ
z`dX&Rrh3H}U%RBJI?lPDB`dV&6_~YU0eYLu_8znLC4o-*a>sMdmcN!A`Xx8bd^uo8
z#^)8njc5oK?1O5HEg$8!;9(UlPloCAq1Tcug02V?^K0o4-@a{^J*>9m{_8DwlYEXS
zk=HU=Wh)z;+tix)c*^xEYt?x@^=$g>k=N1@{m)#Zg+`TtL3{kQOiuAMiY^`2sR6<p
zmgay2uW=jLoUN~o(QOv?wbwF1#p9w=IQ)j7wp%J$wqF;p=X{xL3}9Tez3_Y4p44P~
zv(q-kLj?tY&YlV|eq-nzw9PRz9L{^Kq4u{&!!(KpNbyX!ZB>8xa8lD&WN<C%i0ufP
zK8tHbi(Vn1LA_miuE)?-u-53Nm@7_Ln}*qnb5cqt)gk<LqA=Y3T6P?#vSW#<?hz_D
zlya#Sq1`>0U@Pm4@sh!Z*?vLYvjK<5fGo;vX|Zl~Hi#l87l65BrGEPK<@~#-yDcx8
zq@>Yh8`KXAr`&nk3PFXg@v4o~Xg&ZeD+%L4obs)#RrlFz*eN{6>PJ=$vJ4EtcgBR4
zpZOE<Cb=g})p{5sXOza>O_>Qle-FfGq<8R>Z#p{l(kE`ZJs6aPF@~GBXMH^s(AKSa
z!VX6#^##mBM}+P;**@%OylQ@CXOqo{CZ3U!`?Q+oJRszl9!sP1AKKgTNDN2SbAlch
zui4R+FXx#n1m23h6I-b9X*(f;G9Hfz{6Tb0pze3<q{ds1sOzrAKN2!{pQFb;C9?J7
zT>M^*8sRp5T9_5OrY7zE=yRgW&Im02>|8br_G5u{H-Hg48&t~Sj`mZn*9z6ExZaX!
zL#?s9a$WjRV_0|(uJ^H>Q-d93nCY{RgbC^Sh!?8OuB)9s)+>a-VLQLL_@mGjv-*iJ
zNGfpz1iPRfi43Q9u6?Q>N|jzO6MQj7`(K~3OZptfhBFh&fy?@d8tro)O=Z6l%6OIH
zE`3D+;Rc>@@7J-q@fl}jEd+h$f7Wfki9a9v>R0WmCZmZotiN4c^6u!B(ooyAXc7rx
z?RW76FC-4zuIqD-8vA_^W^Dc=R6RI;#dzvV_IXeb`IFN+$OT>IlFida%s+^P^}}wz
zch7}ef#hgq$o?3;z=K`Nx-BG=Wg)H_e-ghSfc1{BdagKVe~wpF6T!}NH!#5qdcyu9
zlDem2(Ot~%)q3*l{CWFpG#(g2jecLm=C+8)^*43+=vU1SXV>EA5I(<fIGxHkcWzA@
zqa}ya8H3+kDn&-%#si15bDEAh{*umeX#n9S(v9Ht#dvaS6+B~EJh|wI2>hkPaZDB%
z(DK6>>vdO0&M;Qrmk(!W{PRUy5ub9Bpx3eo<H=3DD2Z1Fro2N8#yI*=V9GO5-}n_1
zt`)!J5NL6ta&O;<52rJK{wIEsMX`N2ovizxvNiF7k9QE6VQUX(=QHxKs{|g2uG6G*
zqgK}um-cv!DOZG_)CuKQ1cxUMCqdxx#~-)#f?07R1n$|<vbV=GbK8c1fsU-#zAhNU
z<i6jifi(_PFlSHNlX}lJ+BO|lX$C*$(T#Wr41!P?!=E~wg*-W3P)QO);lsDhqI9L(
z8e&K;ijh==TeiArAcA;nyxqbb-ph4Zv)+!v%y$IpTqJS&(wjCo)h-c=Ya+X<^>8M0
z;Pw!;mD)sk>S9k>hrS@6Hy7jGdgda=QM^Y~^l%_RKCcd+M$+w!F?HmLU$PfOINU6|
z+;L)WH0C?!!rj8g7Y8YuDwd0(47nYkPph+cBC}KE1AY-@+?WljL6hGniOJ?x9X&Yb
zM1&d<_5&m1@UfBx>!w>PM`PT`N163%tQD?tHWrxIwD^9uPQ+539JDFzzeR^?+ZT9s
zp(yLJ1H#Mm7`Q^XV$&ki@Gt+}4(h2?9<xKjYS@p6!y<8i@3^P9Jt8{QvLDad(de(6
zOacQoQ(OLCY~wApt@pFnnfEclLrnMU?%Vo-yAm3DJFbbM2wRcjd-je{f`_09R|6aa
zQz^F}9L~G~T&qs-t>-mC4RTVjxv<T`aeov;-Hrtm)z_(j8#=*)y%&$|P3<X_$LZRr
z&^C10`|*&UKaCsRnHVA!ueo8zq@p~UOZv65>T+^`Ouii9(dBSgiP8o(8=sQjkCQuK
z!e;)l)6PXR&IM|=*+-hRaB=s-+H)c%GL@?eeqkTSN3|-r=VQs#!_z+zhUD53MxGUe
z{raisSDpU}`YO)rXr6LOUkaFD)j~klWsx3jR@Hu`CA<Z*PnTVZArRGsWb$iu)X(DQ
zhlO~H(w%kq8*!elTqGsO!NV8x)#z}0q#yomJbBzXy0vRU6$cg^9`g3P7~~ko;dRlk
zX^Z_{6dOk5#k7Eq_=YIPe)s!p&OSe!DU!n8i7tlya8skiA-T`~5agJ;_!^4x_?w%0
z)&3YAIwgvBI}jmk(*6|N@yYMlooG$YC2Y>|H{Ehr3svNXzlcr*7PvFyy?E=+U%mF%
zc;KXA<=+n&q{OM%-)bw(8MzD7_If%SQ!YgN!t2@nl?;JpTN3XE`rTvTf!8xWi>G(0
zZedGbzvN!rmw$NiRr})W*^%#YSIb_{F5>ygoO^)1o(@0iFu*09cPxKB9h2YQY+qKN
zJ3Sm`h2XR!K|T0-cAEE6<0!m9Rz{2Y^C?dKp?G>0X&tM{s_55OKkV|HJuKi`JwyBI
z07uLOw?Et`y%N)!z~pJNA-$g6eKFs0U#!?8@s6N{b+2a<3ddneUI_HZV^l2i`lw@D
zCmJO+Fojv{3ANeQ+f8O=TdyDRcyeRd*P@epih0!NJ>~xTUst=taS5f;4=_+q@b_ep
z-z8$(B-{~~2(7EyQvuA8Yp~7Hl&fvk8Z<X}vT?pOisHL4C2SSN*7^mjIUaWx%PA-L
z4$TMOQk~3Yj+TJuk~!>o#C8R|;GxzUqkCenwFyOn9g|CKhsd?lEsP$+v-|aIXZ5C3
z_jnS(wD!cO=vauX=Ku5dyoiBj!FgS$@E+;P58DfZqZ17{d*iEAWhh-QyY<y>cdE?A
zV{fW<o3p-Xp4s=v8TEP+y-0p#gE2}!B3RA^(X_$q4BALMr(1z1c_?Y@q&t{&nY8?R
zrorMajfM;P^Izb_RsEdQO0IkI@5h3QQ<OLf6dQXhL!K&sN&~341%01BC&hjE=TsT@
z0kvxB&6BZ4OJn|Lb}*Jk^36OHf1~6m+u``TM7|9>65W%Z;AjjCWdnSdRx_H1<h3zp
zZ)v3DNiLS=Kd@u^rK8A252p8a&~E-`xdEI<j*E64?NhXOg2+t2oe<ur@5b#1>a4vd
zj|^=m1Fv>2)d8|XnidYOQ))=)uRHF(CqN4)G1s4t*6a;G4Bl63bhEyq&jbcNg9G;C
z_|OCNoe7>594w)n%0Z(PiZC0vG;g;L)yTDAcCH4E%fUyBZ{|NJ*<8FSF+(2*B#*YX
z^D#YB1NiLNC-GwbJT5O6v|o_J&#^E*u}|amj;1Xgb}=?5uGbVn^9vj9^p~Ow*VtZ;
zro)~y;a5R`mab7hAFsrliu*YIS||<{yHRnAHsl-ht@s<wW-iqpAzE0(Y(-fAtvJL8
zH30K|L$hnO-4JQ$36&DilxT(mQLktBiL?=F7Y>H^d#w#z;y~v{45tr7K96p3{HI8#
z-4tmmqbVdXXMYf0BP^9$f`pjO!u@05(BuWj5-YeJKcsZqC<`X-Px^vu0`quBN5B@s
zL3$AlS*(B7aJS_B4m6A~cQ-m79>{sv;9p`kix^&p>|VgcgVc3(75l5GiByYhi-}sj
zl8}~_`!{vKg~F4}u|Dlcb`-;!APmjEa3pc2cCOn?j$}ubD{}bXGh#mw@At2%v!zEw
zY}G{PcK1E|;*oTmrwGDjN3tvP_7?k6d`vKiEkBYS^Ut=o+n0}IciLU49)_)m9|j^S
z6QVtMB%P`4iMx~IvTQ5$O?AANyhHX-K(C@RTourh82mkZ_((bnr2C6-m(`1d&#sPZ
z;xi`(2jAKw8D~u<IX`1vKzu|aYOS|*@eqY%9}gVi21GQEdf~geBboSu5fJJ72|eQ)
zh|AD=J(p8@x~~N|u88cQ4M#E|ij+WHyrV4+T7EiW%8s`EJonp^N0Jy+p0G_p1&Wf{
zQvqNcQ-fb{H><NPsg74ev>fhkw#LP&j`UG}71|opSmAuOW(|6Or|r-~%XVvtR-{!P
zwq5aJ03CeJT7#JFtZ}e!i&vxE0V(tq*goEAyVWtK29|wN<nD>K@NhWH^MR$iICZMu
z<|cbVU&8IlyO#FeUXiTLsnV?_LJ}FT!IcYSspM6$z9X4P*xt*jz%C&jRCES`ZBU?|
z>mmT65Ga3josDSaZf}3aN`it6$i8#Yvs@zsCpy_#MRmr5seDWyu*`hGpfV93;6{cz
zXDV>$=DM>_UAe@4Knr$jrU{uA@GS*2vUX5NH+%i(LjejCi5(Uxa0+k_ts{a~i|hZ;
zj%xja8h6`nn+e479;9P$Em9owS9qt{F>MkQK>kg{WxgGR#z&aCh3cA9!mrft#7p#Q
z-F6~StZ!_zAH+6g`;<n#l3d)!PA=k!Ng5gFBm0p6!j{}8^OVTYHR-0N_cRbUU}43w
z)B2>F+;OO(-xsCE=0Bj0xW*=#j?1*b=1^pkASa%Q1M#!sKkcWnCZ6lCS+z_qaQ2~~
z<7C@o=QIXg%s)~K&pcvtK@v}GXCE&DL-4_5t<j8QIP-P=i9Swu<O@N=F4SG0#-Dzl
zzo^e5{~M-iedhIK?XvnAE^=hpue35fH%cF87YTZVIi6|xq)z>HpgKC-jqY1hhL)2m
zAIMCQ|2G2sbgDUbRnPd5vGjhc$rvk6y65DzSm&OLn{EC(q1iCNez>l$n_1#j`+dCL
zf(n=oXE%gldyhxV;X?a)fSZ3eDN=8$x30Kc{UHYM!b@N6v|HL??S7VvWK`<`v7Fsj
z=Ok}|EBs0AQ9;`sVLFa>#r~|%u@m^_w!5)^u)(ke{~|=Z(mKHJ=?7GJ@G$$U0Qd%U
z+w1+n6${WB`&;zQ5=!pfZ)B$f2}%frej}aR{oL-Cypf&Tjs#vZE?+$GMt0tIbLm+6
zMt1PmDp(PG^+r0VyNeANCYQaD-Cz-r-P7N`^hS0<<2L4zy33=(i?5dL%Wq_-TzURy
zSjAQZC^qs}zG4r)amnxY-9FDN1s2`>{6{#(JtRa3R^fn!Dy!lH+$~1z;rJW5n2f_3
zK?gi`P4tP$o6`mB?ONdw`jAHg2OMAFK<ffbdq;hTJsx!1k|z|L-s<$7AMz;0^;{Wd
zHQM^c=cMiXT7bedR<#ZB_oI)Iui)$Pcb|)?8}&`K=w`Dg<9UlS)gqf}lLqbiwUg8O
zQ$fu#p0%6b$i$8XPqFtAf|nP7XowC5s^~}>10Y6Yj!1%Bw#K`KA8^BK7H~AQev!`E
zj-Z9BV!Es)9;o-gUfQ+T*SlPbUDx3Ohpj<sjucO5+NP-|g2^SI4vmjAC@j+5npRIE
z>T=GWjQURE=e1HBT@lr(<w>Zu*SK_$FU0r@SljkS=kxCtNKInh8V(u2aVP{{P79$z
z=xv{#I!N@A7b-|R1KdAtPyn8zYSa)4HY)`F$$1F3L>me6y@<$WB@K2fS31^tS@7XT
zR*k2qwqs*|Hxw_=l4OvH+C&kdZd0OE4&xI>>v-o9xE+XhuHb>}KeuUhRNqbbb5M{y
zv}LOuiq9}ApaBnR(i;x=<dTba<GmhVee{h?B3E*)I7DYOs@?M+xC!MgVO~=Y_%V7c
zy5aIswYT+!`|Ti&9)BZAm^;5^@90%AH{{mj6B=dvKyRK1EHxk}wJS7FYG@Ts!ym;{
zEkNv)K1P)`)n)J1#@dY{__XFO{H2-v_oI1nWD7nC?E5onLYY-cZlR_8v3hIkBB#8N
zX~3Gz_4TLG0VnwXw%PdmvB!uf_%QzNAuF?;i(W8(q%aiwNJy4Q3P(0uE$O~4rt)LG
zF3+EXX`I)#;ttKW@)SR8;ZLI3r>EjSyAY_sf|J`f?q)HW^JkKyS>Q9}C(9*`lKtsd
z@?Y5Hn2)9=`&I0nVYIbZG+36Hv$J1oo?Xv2{{b|v+Je?rOTGOj;QZG|?W#U@iI6-0
zZ`Et)#!;(X3#7=j+-JWNorj?Mb?rP)4OxNf{yy4n^>VH^1bg9FN`m)!;6Q*1NBm8_
zO+j--`$N3-OiDd=OK9M1Q!S68>inZVATgl(5#Npp;k+C0vRZ$NLA_YDJAoFvFH<Z2
zInW*{{_=+xU$?vQA-vg>W!PT?xX4|dsl4}wZV7%JAsBkt{u&S}TtgwleGQz$cm6b<
zet(MzsYjJ!N3&zoSq2>U3rEvAoTOw*UmeX(PU@Fqm3TnSF*SL*yroC8Q@*bVqyOU3
zP_j=_-qMyGO-KIx`(^u*zC)vqW4tXtnsEYG9mXnjw*&j~(Mx_DQ-GK1Sw}M|fcUYI
z3gO`Ppa5~lbO*VW(IXAGi#;T4ynLgr3IwKld+cF>YG&>+V~~OLXhu0ga>t8#7aG>Y
z=zR$jL$g+$;Rxij_=uovVQ-UVl^Q;3!~Mc@`9CfkD&x3E)&+hIp3B>w(D#&Lbi;ea
z){DZVp+=AJH9c47e+>WFpl8er?Ch@xecWPd8}+UmL6mb`xFA^Ju4S777sTGA!+feX
zIzP5&ZF67-2XAV(hNGEG*2Xb`kl;JKG5TuSMxm;$(R%(oQi(Ndj9bd1Q??`E;)d+P
ze~aL**=)On5*8uO1Z$19^a@F=O|Q2_UGa{^*Hq=&t$vcMvqwmF5W6zs=FUlb{%GR+
zYyQzX16{AMWG@H_w3mAYbSo>STR@vVXet+9jbbISzW95fI)B>o@i$q8hip)UCfy`*
zcp;#3m4oGsM8l#Va3#ThuDNNKl{Me4ek;4YP}NUf+G*I?-I#{uxkN_eM4S-L9XO#(
ziB4lgdLp9jJDLe$Y2PM5;Bp`k!y}P_P79fxh{SeKFxY4u>UJm)cay0du6^h(nRY}U
z=8G_pqe6KrV?chQM(NSag52O%?6_+qxx6K;M{wXib~Nd=(6*LwrNM_F1suICV!9Us
z*Kd9EY{wRRN4pY@t*>&H@rh{RzE4O2Kd8~$t@S7MEm`>OM}pkvxt;Bl-sbSleJ`FU
z+l;o!PRA2m7q26GUtdxelSOex&m9Cg=IqC@$XZ+NY!GUM8!1P%I%vf-HLH&bDU;iW
z(X^66;m>JYS<1kieH0_%gc-28Xw1IY)@C0^ci&lK=Z|I=ofLwB&g7F|Y6HkI7vg)b
z>S&)v!<xIQr>ndeZwp>k`BMCilhBA=7Sf&H?zCTNu2IZ#pVlLL059+u61omwsnl~6
zItAFM@AlusszL}TBvr3nRg;lPS9^}wZ#6V{)HN+8JVrf3bJc#QFPf9)cf_vimoRH^
zRtF;7ost}DHw1>VlA*NwJi6&C;!<fhgMvLH_J{cEt8496EZTI~xQE*>(9OJwn&gi`
z9GD)Z=tP|HNqoUIX)sj$O#J$H=jf&Xn*)l8TKKyw3BVS;+Ws6g;Tu1*`R1;u+Kmdp
z{vzZ#{5*_i+U}{}RzlnV8W^zyxM9bz9Dy$T+tEukziqy(`RVzc=(7`MZ!R9yW0+sR
z@a82yNGa1qG}oKySms6rG7r3&ovJ>i$!~1wo9WP}vS<~DkuScP$pjtYFS1OaTT>j}
zm!i4XDuO*Me=`#Sb`$evU)Ce4w4ANbw{U+}(1QZb$Vn+@E7cy$<fJ_m*mVtKA*!*$
zRYDj;Hp`4v_wbw9aRt;=Sd^>AYJo<kbtmjKL4qw+Lc`X+nQbIkcnXJk<jo|WbP*M@
zE(Qc&`~<%ExZw0*OXvPr7vy8gD=*p;0@UOcemM=+$Je~}h=Q*PTgL*Bv$o;QbVNP5
z*S;Q*FneRt+^BwbcH#I)>M{M$&qbSDWt(CE6>?43Q-bRnK=^RQHovKAhzPB$=#SH?
zz#o9YH;AqnA4jZFpV#7her{U>Zye=4p?UM0nK(oo29Yj)dD$TtTU~A0vN)G$!U)5T
zarIHU_svTWve=kXkA%dNE_Aj~fKVc9)3~sbdDXBE(GW(>U~pAa6Qm$STic^1uCa94
z?0NN$?Gql|8R+BbLcT(KAy7@xUo4QlnzAWQ{Mt;T`pF3h0S*kQFFxzWZ<d3>^6?pA
zOB^4AF}%+GR*0`^u8)L#3ELD#_j4<Wx)_4}UQc>5t%?c~Q&nHok4$>nx-pH<%_UwJ
zZbECimRrF2@syyn)$g=@@l?x9c|c<!uN}*DT1a3r_Lwa)j;a@ug(Qv2p%~Qe?RL1f
z7E^A+BbsT7BoM3}pkcR^P=@w1BIk|)dn?+~C0E>GJEm^AUl4o3x_LWBn*!5P3i0Fd
zJyH-;`%Y~Xq~m1woCs(HcPF3OA85ckVT333<l6opX}~GXlbs5v?hr##ve+xQ&kWgV
zeYTTwJnsv^cBXuw3sd|VHE1TF_QwIIC!OhM7b(#-LP``(T*~`T?&_E(KT4g%#o_Y&
zzuAY;IVtSFZ|4M{;$8tC2>@k+%4Ct6)BAR$E&EvSxxtKZ6+IvC8@Nxo;`Ea@Gf5An
z_-xu?p6Kv`K5NH|1DbytLqhfM8nTNqB$S$Lu1liZu2DoxyBzz$3+etU5Fa8A!mfxY
zJGsaFIvR}+4BCP)-sZ-*-^8LqR^=`iK51HJ*qBKHk99m?c}NGl@tWQvNZ@KVV!so&
zA?*rV-d$p@2QCAisr&cQB^q^>ALj7Iy^$)zFkqb1G_QRaL=z%61L?M=ZT1KCIhOK0
zkz-%u+ma_e%9}|zg==#fgsvW#w*x?dV8lPw2<jdy-?KYHn3EXE2gRcOIfx911K+m0
z@jxCChW{76>HLP>Q@@Rfp??)#<SB$1Qh8d9%X#~o=5-gl8ZKlr8DD0a-3W^e_=TBt
zbXCYCx+MOj?5{mAlO62+>?UH3sJ9hwxq8_bXOav6vQLlNvUngOoiTrDCOhw`2H-d2
z<ujQszP_Cl;4_)5-{kUwOXTU=vSKC^I6ONjk=pUh-PxVe&e_VDOX((09F{#4{h_er
zm|HcI9o|%{p=J%OKemSjjq=3k^H*yy9Qq#jQL#0`(}SAORji%K&UDxcW-|QPW-?hM
zRm@nYFHpPqz2)&389Rx?(&_@M;rSnf)e~w<;+-M)UlRs9eR=pr<ZA-Il?WNz5Z`;4
z8*Xg1;n(cbwo%C8Fv|$4_N2xgbyE@*H^mIKvi_e6<XoO3BkbmY7e^_|u7)6It7iac
zjK4|TQnsyHA1Dym?j9J~EMj^+KvJ1SdoFrp3D>()vBafmSG*%?Y|>h5f4d(^Tm0Qq
zBoA&!^ntv5m;>D>_!HY5*y0j9|1Ju3jg~O&RJQDS(KAOvng5@$_m9o8%KCm+j3LDM
z8bb`n7-Jk>h5;UiA;cKN!w~v|^21Vk+KLRrFx>6k(jPZ%-}Y{yRSAb-czAgb5wRj7
z0wROFjE6yn;o)I;2t`5+A!v-@7-9?|#t>r&F@zXG&S&j)U*CG(zYa06*M051_S$Rj
zYp=D}Pp$#WO4D;e<(MrG+OZwSO|%t31QNVkqkbRWWqCa`H(E>dr^l~;^{>{dX>e#V
zt@bV1G3n~0L_-(73c`Yv5$;gkS;7KmI=<>|L77J&2+H)BG>c6I;`kefs6p%3!msQj
zmG~MB<~CvuFh6`?)&V;?xRA_D_?BUxUn@kYoaXK_>q>}_uNQKx4+`|bdpBsCz<#|D
zo1d2(wHZk`*VsmWS=$u&!9yBtv*y7U%588<3HQF0eYRB_gqtrembj+cHW7B_FKl}>
zru-V!*o1T^2Be`aNpOAXm$p*`=xgMJ*cHvlKL{VR-SL#9y_2>_Gxd~f;J#PLA>?v;
z?H4R=1Srwf-tUyi#Rr2H7<lcFnqo`pw!^}15#Q4hy+nR)wvS>O4W2*xs2bM5*p6v)
z;=Pkr>bSaMEdP_82o$g+;A7+r>!iq?ZWkU?e@Y;$;mW54z@u30OpINdBb~w7AX;&i
zor^AI=o>hT&j%T(i4785S1!aeUgj=s=IvsD#M2|Cq5D)wQw1q`G+zq35Q1ozwSF%z
zvnwLt0%SLk`&<OCC--2%z7Xn7tlgJt-rKy!-D|G~Fyf@xIbTIvzqeq2xTc;dPg%^8
z6NX$W`UJ9ECxi><?79%^qk@(2SV$}`TJw0hW>tZY<vUT32(oUwDT?Qt8~c)O`FjmM
zm}|Em1gsq8&Vb!gfAqj)_%7kb9<m>W8_dO(%f#<;)4WrfCovK@kqg`^jDDT=lc+M6
z=@FoEFQ`p0qur08_}1Nl!0gqGJ(5Sk9(XmeIegRbBY!oqA8h>TY#;Ti&LmV<kkq|g
zPul2LGqzRL*6H?0dnh1szWxbM>W2g3pfj{bUd?!AH3^c2|1$h{<c>#Q%@{UU>;iVE
zF#!=SlNGmk?A44pyPN!K8>?qNBrffb2c{gRWga{GL<~uOFSO>bW`aW1B6}(x+Ns0R
zX-@|b_p<}l^)^mG`Z^)g_?U$Ut}~|z0dnd%Oji>%V;D7gAUbVw%qx$w#z~d@3v5aW
zi%!zOP1PK!YviBHG|hl4kJxjjznWwqcO!a6c+`=|*PeYflY=~~0~xnE?75f*fex5c
zXT(EqD!x2ZebHEw%uK_!HaB_&p4p<t{0b}A%9o?wtW;aCa>JKj75ESDLv1rJ{-z>*
z(H3Y7lvm_X5fpL-Ta8vYF)o5D*&?;CW<np=Ti2ZP-PWmFJx>i9d=*AO&ZotC{X#>P
zEzyvbSki523A4YV<X#qkH~Lmy9)IUC;@Jv8B+;mU5{;s`OYO~;kLOUCa)oWt_I~l_
zaR#lWT~dmrv-Vi1JZ_JL7!E6Wk9BI5NSlSFiX)+0KP@B+5%`f?R7hczaMeLErs?5$
zh|c{1yP}%nr2(6SE#&o%p9d)N`f~id+I{5KibO3;Z=Ig%+t%26J-xtvW5drErIa7R
zHj1VVn3rr*4AO+d*yiYre!CB0VT*QqDH_SPY8IXifIHr{=#V((-)(z|VxBVZ7+%{u
zO1stVR5tYL1<&Q;Y~p7+v7N#tReao~A>nYzQ*(ET>ac@7A}d!H_asC!9wWLq_t{Rw
zr2PU1zbn`Q;ahTfIv5yz@(x)>?2t&+*w<o*W1v2~zK;YHzOocf_K}eC<TX@Cv7;L1
zvklDi*vKyLYxFM#E_giF3*TjGaj1DMA?kKgSm9Grz<PA5w4L&t|EEV@w&vW4?~Hb5
zXZUCLV9dzdS*<K~6t11g`|ex-s&=j1c?~+(3&dYg`()Stn8WX=UDS^(XGFVml@6M6
zq_(7bvzG*@!Y?bAgKEzFk%_KoO|I+#q=!Cyt{HIJdX=m%1Q3%JU-*~tJm0-*Jr~ne
zy<gazhmU_1&kJ!C<z7y0j*&6`lf?bUk$1liB)H|dwd$Lg_FSjSbxrBBe@lwXzmbaM
zh9Efa!^x?~zSW>!@`ivfEwa0~drq30`Xwdx<#YW0)g(q%mD>-mW@cZu`rrJbnC>@Q
z?3T7d3fG26O?EprT`p=_YB}BSoh=3t0DVUorn;n5d$zm!i4{BQPSy64cI<-TkL_M8
z47n3n`~kbK6MkN$*;XCA=%0NMT+X*94$mF5U=ME1E|_yY`fSwJj8`_a+32n5OJxJV
z<&wGDYaBfkJ%&G|<`IYJ!&@`v<Z=_aAc*SLPe~KuxjG-+nu!Rg&x-E4JZ5Xg^KsJe
zZtb<l)T_iINM9O=lDZMQc>GN*E6n2ROIC<ZGSfb}HSx|Kt};*Qokt}?(Wkd2=_Xxb
zxN`^g*miOrS~`AfTFE89vkCE{BEr~hVnBT54-;*20C4jr8+}SVp`&+4o2i;t5iPdw
zmTB=GqM_Ba>06W7!uJqWRh}7{BtAZ=g8Z{F65qHz7jq;+2ivz=UU}D#QEsR+1v*c@
ziCNJYGi0C5j%DlWz%Fj(qCk_EqQUGfBt2=iTU)E5?`PZDOY;PE9+%Pu+7D%XlvW+c
z&0jpnYV-yZ1$OTMQ9Y+&j@3mgs@qvZKuZ}8j@shbeQ4;Dh?UHpOM)bcK(Jw3s)>@W
z&12OF+z!}F@#nVXnoV0R5lyi5k=b~eb?!}#0g*3{@EMNbydbt9^CV?POCXOf8|K_)
zt=i^G=39G=My}qX6$E@$%7TMxJeqnU70Pdk29-Uz)z-5$6J{foP+Fy8bjYPdn*9MA
z`H^JdB6JPs%%DC@--(1TaN~P9=va?K&DwzC0v@SV)U&U1wXN6m5U{*N+)3M@2Ia_O
z8}+o9^s}2Z5b8Iy8=mW6bF^>;I!E9ZO}ZS9U3@ESYoLuwV9~Y(Qbc##c8%;l30xki
z&=DWd;#wOL6#weo4RWVOPBmq)>G#^Mfb9Mqwp)0<z&LyKL?v0<8<Zw>V*?+S?HBMQ
zJ;NLh1TLkTVZ{M75C;R7msZ#zom^P7WKTOBV^`0wnPW#{Taj^zz+flBM;g8n&-0`4
zg!JN`CC6ehD(95HFxQU9t8&-w_u2{V<_353x06Aw0>|>H__NY)*V;E8Y;I>XRymvM
ztjOdpV_hly|C}hDPDc-EJ+BXe>z_wbUWn=7Irh3;RC^Fc7090WGzKB3Acvh@iYLE6
z;?RFt2vp)Lb48u&yIP9aT1X6D>dd*156>4d$;hPUH}0SZQ{f=E8fbXNTl*?-EFGdy
z^_q}gN$<W!`#LCrkAfQ;?3;i@rUHyr*8>vQjaIv%j*WdOaoD%|!Rr@t5Puh~7b2?L
zP0^R~Z@u=tDC*1aw<`*HXrVm5wP3fRJHZJ#yDiw~Uo8ypqnf4Y<FmRW=p9}o?rz`=
zU&k8qlfZJxCQa<USo9{C(tYjOq%h<lw$~E3>BoCK{`$ofd+@c4e|n}ZgqnZ1QSovf
z99S>$ilLkwcZ!EzOB}_0{!-%KhhNJs?}L35zkDsRcJ9O^R0%M%ersc1%a}YtuGH&L
zmwG&S#|A(+g&z$CdpzE9l49<)Ctl0gG~;{d%#*KWI)J+24_i1;-S!OI)exj?XtSsF
z=tb&m+-sRwQUAP+FAWf-K__SccQz#5#l)Bb*X;jhlLdw|k?5}1GAY7ssBP%Krv~yc
zld7%)65upF6O&Gl=^A+Tpgj{W`FyAhO!};6W2BS?=ehVBDS<4d8r4Nan;Ea#R^iJ!
ztAx8}eE$!dT>|JzWG{G8{#p`4+R+_U#=yi8Or2^~nj^uZ_&eD={otlw7}5gGtlkaf
zh}YHe3+#|Bh*N#D)x_BKbLtk8`@2>>QsFq3pf2!i_R2*KngTA>+@<x{iO^w70uods
z_R*FGGJXS131MuRUVEyV<-(Q2(w)jyl<0v9#TrH5bN^-ectHvqJkPWcEkrZ4T5G&Z
z?y&7b9bFOSAy!%8wTzK+0Q&U(1Md_FuASUmx;2TpZ8$*m=rslw>c$jf1o9@3I9yE^
zy5@V0CRE+G#s)PZG&eoN|FoC2rRQRYAqodlV6j;%H0R+@>8U5z#cXl>Vq2`&_!xxP
zH5=k{oAW<xquy}*4FA<O2@wAM4B2U$h2INvs>*CjppNbmk27J%<+h*XPTO<_WZ8Ud
zyXG?wcE2OALgT<)b|@gbJGu+MJJoq9i~!9`r|pVOT2C_V-7&6v;MpD_O=(Y__eL`U
zhHSr3y0R;Y8wUinWBA12@Cj5x2Q?$Sw|ec6TDSz-HvDm?9gd%VUS0A?jDeSweWd=q
z&BK4QqaqnGPN_!dv1mKMraK;Jdr+R85Ze9Cp6B_bNb@4pI3=irOSao-O$7H8kB>Q{
z_enxK8)K}bpu{=B<qIXP&H2}K(Q{wg<d5^&zkl{;yAaf8+rZ~93LccEu%#Z|h-8-p
zGv}2_?Q#rPHGi&M(Rb9_VxOxWTc8DX_iJ*C;a;7%f6D)o#OW`8_UWMSC#O0`!BvrO
zu3JE1cVER=<b1Yk8Vftf*RN&w!)~$_;}bpnv3(Q2p%NWR_Ax)&t_vdKm{(k|8=71S
zrTSJ-`}^#>SfPb($hxVXy*Lcu9b?~XjmlX0A2eGo`5dvgN(AL-z#^@tbD6}CBTv}2
z?!*&E7#Cl=8$*)DkKpWBXSg7&$$c{JX*<F~!=wGa_5yj#JCIgwTjEikgn}#bw!}wU
z+$#Fuwv0q5M?6ZB#-pM!C+IJy+Gst%A0OJ5@hI0_yW7pf>Yuh!OB?+k*_Lr+ch9#+
z1420q&|zZ)!Oi$hoL`S^%VY$<(4o!D+t_W1)v#dp__ibgpe*+Y=qJ?H&wv+^y6nlA
z3|6*}?Wt`U`zHJ<37hGxeOfq%T<_=axNVur9pQuL@xqy}zQ-m623?K#jF$%RLg16*
z=arp(HYGYxO}6M}nW-AXBfn61p;ex?Ej#%AhRESU^p#EDmc*oA{fj*lQ%jnazuU8#
zng_m8F>1h`6HJc)MDjrHni1HLPP@luY7Q?Z9x^KkO{g0u;p{+lJQcpn<4;c=U%4%l
zcq+JDk^-kn^R9K}_dLBv%j}6B7U+EwrZB4(mR^#R4G8BBESO&Mcd^>&;TIG46RC??
zvY9HeUpEAb#V$YI7DrE4{G=$3CGoOi(F-*Vwp5cV#|F}2%e1y0O_J1mt~OX6%~Fi7
zD-u^|e-U91>%vAYK)(7#%j=mVfrF|g5G-$Qaha-B^G#LZ+k@OB6CAKYpf`NNjYpkY
z8(;4p+-uN?>o>9)H+PaSzr%W>(?VRc@GBE_lg??C^@}(~q7T;SwKK!d{>=tU3&;L^
zIsR<Jy?m`YwQy}+r>E5e1bPX5GJ>wO4Vo**JUZHqB?_nT?oAr5yN`lM_>rk~vCpII
zw?u0)$FQHa#>*tfJK?vfw};Zf6}OA>xxSUjqP9cyK;}rUps?^DDB8!O?bQ2xbKjb_
z;X|BFyJDC)$2i<}Ys0M~M-1{*kL`&WmsIC_1IhYsYJ}SU=<RCCbc;L?O>wjGWIP9h
zhLe^pvqPd<B~gEe^`ZVOgW?f2!F}95Qj-K6hgE2BR84AfHFiuCA;rJTbjP(A&1)#|
zb0UU+p{CMK1{Jz<1Oc8ZQQ502+G%xJ<+1-~G{ArV#c%AaUXq%(*UkkHf<SY2zBKWF
zgKZKbW*79e%<1&3izJ1lYK~njZD&HN?9&*R5A#>!>{7fKbm{i8NZ#9rjEDNxu4v@X
zUd0vX^H^#~9Ov3m`y%?cG_#GqjCod<{m!n&VOPcFIU4kN&B$v3ja#vQq4xC%BD3uq
z5wR?vjQiJv1Ws@Cc{hXzxeiv^w*eJKe#X9wztg_I87MR|&F`Z>#ic$ZbJq_7?@4vI
ziQE#Qu=aYyDrsAw-|Ya<M~2KF1HkY;9A$ELM=R0*&%PVY&^+2tn#G(!yQiJ)Ss1tk
zp?$tDvKBoVldHb$^{|R^xIFNB;&U7&Zo7T(^~{S4OOo;+;4~H;s))23{d&eW=eCui
zBp!M_@f(i8o`HIjxupDS55Jz>PJgq^9(g_EdiAht@}|6=n9<5X8xze)rh~2UvDY)*
zI5vXh>FL;xRc8{ua0`4qP-X+oBYORbK<(VIDfZ;+8I$pd0fKO!60WcuvN0h0JgqjU
zFx}N>TtI^x`=LBuuu=2cgm~)YNaG}$7$Z%Z^pZ_}J(Dz6;>&7N1T;sEe2$8#dVl2|
zbf?pz=O=Hs+4P_l$a{8`WajC$XEZ}tcyqTsD=ZeUub+E86Mi_o`~B)Q<Mj*SwU6Rt
zX0&bW<DetW+3T5D<#ieH3W9#KgTy&FL64QI0YV(C*jZ&|tj5IOSXBTX;GfNlZlCNc
zo`1&{#0$uZWm?#w8Mhiu6~PP*ww5qLj$DzEx4Iwzk*(Gs1R`ay8&?(!Mw7b&EfG0j
z<}AumZKb+0Tc$Pj8KS2ryP3K-w4yy)5#8rRs0n2~V(_N6Hp{D}C+)NrQP%T7JELmV
z@*scn2_#KjyLxrHZ(KpEU52I{-rp&FNy*38*{$}i?sC)<^s1ox5R)sacO|rk3G}Nq
zf&lfW_1^7dNMnQgp%uxWtLe+J79B+-&$S{=vBTEAet~4e9(3~m|G$YHotH;PC#p=L
zmfWDp5NC{*V`Gdq<@&m}ZIc!^?gw4u_3yFGLErY~e%qqCq|bb-9?KTsV!cf>VPb8~
zEw)`_dF~E^yn<4_Q5tth`)n~@0Xx+>vM1Rt{Xiu&zZmY;59B!Yk9kk*>t%niy+Xu2
z)op)J4f$>rr7>fIU7oAzpg@#E{9K<M3J^_e>~QQ`_ZbXpB%z;2gumN>i1dp61l#wx
zp_A~9qcI1>4n8(JCh-4fpVwco<N6^iBX&Y_UWGO4q|kOhZ#$)zHGc0_--V=w<tsX)
z*3aV%YiHHEFud2!>1iHeBImX7Ip?Y3iT8Jq#C?rh?V@&73;a63qAp5yp-cMyU|7zG
zkJ@En0^vQ!+ZFxbYC(A9=P^27-0t@Mg>Z3g*Ox+jVINY!)p+TtIuRVcioLa{?ss-A
z)`^l{1bS(JTCUBWl~faQlVKytKd#5h<yYAa4Uhu?`&NA`-2nmPwBF-3+iy3usN~CG
z<9#pC3n|g()B7PXAyS{Bm^v`Z@d4$!=$4?{njorw`%zyFVh6GZQPZ3>cVcGa$mf1H
zMnuqX=db^;p8_Ed=y6NjJt5TX)%NVZCg5oa_-t%@#`kdB@C(`l+Y^5n{?N0Q>SfY#
z{KH0V&ve@z+&e~Z&p2~`D(w5v_QctUEJ|0{hqos#PJlZ%nn&UthYM<;f<3B7qGSea
zj2^3-@sM~d9*Y!3^c%t0?MdXA`1@bk<HCr_1fKo$i5LWzFlhH=yp80teq(+rCWFem
zU{6O&k03_>9Uy-@ZoG|;Hia&(A`@bMz7Kj)e8&?Xt>*%mygd_MN(!H&HYGrMI$c7q
zw`Z)@Eg(MnH1$a4X0LrReS3Djb}7SC;BrboQyRy0*vIVIzz5I#PJ1q3#+Q(?5YC9V
zp8C~hYBegkN6Zpw+|QPnv)S<juHZ02-wfr#r;gki9)WF@T0+mrgB@B6!4XC#@pxT{
z!%te$1%XdGpfK!$ZFS&tWjB+n5qvjIq?%5(nlEPnx|#%~PIy-n0@<La4sI5UHH&J@
z`@IEQQX&o#E)Ic9^}3$ixiFAr`T>^L4~JX6J<|)i^}-7raBB1E*os&z_#}jFjE8op
zK!CPjq(W!S)*_VJuqd_$5yG0yx!kU4dhlY`0CI9+d#1#x=i1np^lBj28N^}xY_@Ju
zA_YOZPhHM>1d1i*h6-OOY6XbuZMObEl<#KF3$`ZUwKb6eXHY9w*XllrFKZHb`r|ci
zYfJMj5_x9pv^iOa<YT&Cb8Nu#X@f?sK^U}+dP#JRZ3;+SB~Tb_);KlItG!<O7PT!l
z7Hq2k)pv0|Zqp+-7npWD`Zd_4%N>GMAm<1l{*Z`tWnVhdya3$J7=Y8y{q=TfCW{KB
zpyr~tTb;-dQKS}%JOAww=sF5u6k2Sr=*3A&X!HJ{<=}uvb8XKg-%@-$V&}m)A<q5h
zBs&yMlNa&f7#<>aA%r{YJfbNMaz(O_G%$?YCwVle*PM*n$As&PDB((WjpH#vzXWqp
zJ+b{lskB;F`qK0Aozy5eT@>w90L2LiPu9~K(!I~&4`%}Rii$aQR?~K~&w!oNY7r`C
z=eMW36j4DJ;&(Tsa8g_h0I$3>(LNObzUkq=*d;wLai;>iTsjg`RU+2xxLt|PxUAae
zA}mgS!++&!m9sAdU=d|>dhAO*LtwbdRecj&<<XF^7GFh^<~6vd_(kkmEXYE)c7Gjz
zFP_h2ztP_CngH&=ab44Oeb5b2|H{Yc<NlTcqrcU7)ba`?(04)!cK{CBn}S=OlHhVf
z{~26N*=pa0zGs!)(lhcVR*l>0?^aS*PW%{H!+&0~JJH%>UXs7c-B{bcqUUk>DISOy
zfi(AGR5xv6Exj+OSh14w#@@(SR14Bksy6Vz8;N7rxLXCzWp8908+SgmWuxB6cvffW
zIyw4{*bxM2zmeEC5mj8eAATb{RY7vZt!~I3c_U+Jl}M7*=W@TgN8iYfLwAlVO^nPd
zfeW&qyF5JhMkaZnpYsUyvBJLBg9=<5^|+wddO}2;$N8Y12*fxN8?b(B1_e}jt`U6d
zjU+Aj^q!6ZX<0@>r*V3kKc5ukZzMUVYEj)nn-E>dpJWqDIC@Q{f7)c>m{PjrY|0zi
zNfAXW1{rW_;B)>}&%QBD$RS5g4`42Gu=}17@e6&VcClxpDJkHGKep%I$g~3K;x<DA
z7s-xH=b6zM^$!oT^cDv!7usg)2WsQ`zAx8U&+|>?tWvbadDAVr_>u=i%-bB7^MV@g
zSZ@oWFP;Q`PE#}_8se3M?!88Us5UW_RN%E*0_=vw{uHfFzu=R>F7CAkAxMFe98SH9
z^@FQj*_um62;f?>rF!gjTjer6)^l;R<*{Is$icfJ#;U|DPXq}kvNcBAjM3$_WYt{k
zkXuo(wZtx;S7EKCMIgZ*ML1*TY}lw&R}dAeyF9y2XHY3+HSLa3s_@@%BK3^S2hA3W
zg<_!C-`rvSdRIe~AuMx^X4BK$-erR$AD3sg@~j0!*yAzuvN}>dJMnaV56BB{ay*V;
zx-PI9K7jAudhKz1i)}+`Gb35xp0qKhy11<Rg*mn<I{D7D&C!f~KwP<PDQ#LeBBDZ%
zPQ&}K7T7iok8z<3>#kqnw+AGuptjnMXzltxvX94$^MC&2wC&W4pZDS|yF^(0@JXh#
zTNCBpf$GurXdaaicW*Rb*^G;35oWkwZ=YwD2jVTxY}E9GI~<I0Q@O-L8n@mr?}tZr
zZ&3QA9f{uQUhbZI9|;wBh%=a@+CNc{yal=KSPZd*e|B8R)L<?-5l>uqJ@fm?*aT~O
z^L8q_EX1YKPRE|bwZcd6GwqBJ^UDLDbeo;kN5sM~@1mX4WO9gbc3wn8^R+r}7h(jG
zSQYJ}kg7z_`Dvih!})Y6R){09qQWkZ43Qg1=8r3yEa`I#1FPL}`*XEgTrwPfAsSIP
z1^tPADa=qauCl9w;)*MG)Euy{w8?*7ZUkZFnx^jD&zXgN9W%kbm+Yp2Fg|8HT-OG?
z@;*L8ZoWHdH=+g7K709Fp@wq|_bdBOU6A`&7WcTen=t^6yd+Ljk0g_mlSMz2ybW&&
zM9Q^>s~_EN2M7SJv>&w+i&!fZ?(we2<E!p!C0gJ?KM52?8Iq5A(}3NJ<}f`j=+S&(
z@=Gz>k+`>Ot=QNP#1m`R%U$)s9f@PWS$NxIqjn_Lo>}&6bUf3zxy22;5ADcggH-Iu
ztzQrCNZdaicc&35PTC^@6ImnehDUcKRywDL45#tu@QGG?Oppty{=!&6u4k9Bn^Cjx
zNK(Wverrzz&UkmbVDRLQ#B_<hLBTFCRB`0p(>pTzfvY<BT1$7%ae)&P@$zLhK0yEa
zmo{NX#*0&ak>rdwQH{HsiLjr%Ba<;anJ*i4N+3Y3Px$tZB;cS&W%Ox53q}FcRM&61
z`jUpmztLxc1e0EwV$TZq7PP!rz117NGO8Uyv16dCfBAnl*vuW7j5Jac+hz$wJsh#s
zW()Jho-e9gk42XiDg)P^J{NzhG+LTCQ<<mM@Z{6BAc*c#75t$(`Xa$#Os|Po?nePR
zY6A&3*d%Rrogi1EtnyoPgCJMWscNvr0Shv%?zSbd6u)`NmWo!m@pyKHW$|j(1Y53o
z!uvjX-B!diI$n2mM#HZ6i2rEF^FcUtVI9^Ybjn=EjwIfyf!r7bYiL(@whOc=M0ZXO
zbnE(FxEEQsfK}jF=ixp*F*hWFWGyagKc#}LNmbDw;JC>VlO2B-+FF`zutZZ5RNy5Z
z8-r>wwl@Ckqt58MXzn$bZM{HaW+e{EFKHXJ-#H-N8jT9jh5U`tuX;|+T-y}Ga4^6_
zH^)Q78*Yh#dQqTk6$$3zaYaQhP5b%b)3#j{<FG_zupN3`jr;FVP=!wpC0WW{y))jo
z4GwzN<Xu{SREmwZTk8+6+apN7n)(~t8yImZwEdb|I*wHS>wtjGb-&$%dPKG%>Pz5i
znj7t~Mw(l0N8(inuF#w=A8AUt6x(S>W9S#>SK2YXOV-!pc0Ay8W5q4n35|%_F>fbh
zm)6(Vsen+nu+~niO?emYGiL;FE;dTgJR1~qrvmpnLkD(F3*a22%gzTdVy5guET-RN
z?P5#<mri%S`7}C_bFrIvy-vHN?b$^+8V{krEDC(`wq4O78=;`^|IdYJuWM<(5TdB1
zS|RV3I(Cu+M7mzuRS~%xb?{d^vWr?<Cl(|>q&>~RH4XM6suRw$mG-sJL%ih>!c#n^
z=Ei%wE>sqDp$ferRJbd8o~&>6o`pwiuE*y+AI;5}DAsNc&hO&^>mB6#L6m*I+0#wl
ziZS~-QH|dgkQ`;|>_<IT<a%Mj(Q=}D``#7q1U9wh(YXF31d+kqCG6gg3t749?Fj$4
z>%$M-LUGdW2c%EV@M+nb+5MR<%JTzy=32#)Js8gw1YYI+d>!>>VxSaow$X21aJwsI
zdT>dLpL6ZxNqIBlP<0-JORUzL*+p=khl4%(W~OmpiJ&p&&5V=q!6IY(v0;y?#~?;2
z8>>e@!R>JkKr|vJjXj|sBKM6w8JIDHoIMp2sKZ#mSwt!9H+AnP?^bdnVZC80<1`96
zV!Ca-fWj%AM@<NzR2C@M#F%a)iGC)lTao*vO;NYx<+jqMzM1iKekcWOPSbBCff>A}
zznOS974e|@GjC>E3}ik+F$vHX)u`NL&&3n|Ld~eSXXsrO%Q7=?Yb6!btkMd(yT@$d
z*@56-<?)ym@~+gfQ`OiV(nh9%4U6vc-pqupm0gr7qbP<g2#nh}q6=0X0FowzbHr+5
zXisltwE|vDv#S$ugnm!%-=GOJRMgtyfD1=PIlDxCJGrl}v86FP_w-q&Uf9lUxez2t
zD(UTEX)6K;#%Q+2((3({D`mcfI5D?eoJDGC93~xZt9~IeYWPF^zS~R7-P~k___J<K
zrFClj=aG=ROM`i0wI0#BrGPhhQOoUK07!!H%6_5C7W~s^{}VR0M)O|rM;nZBx|#^b
zE|Hcyyl+=qtFE~wL<Sg=t<!*%ufuO#Gw3FM4KF|2Ad=%-<<i_n0a?Wrw6n<lep87o
zxNu{}7Tqhl(5Kjz7$GHIThwA(N8oie_X2O5en=iX6v`G3ZUl9~Ylr4Hyx)dmeiilg
zwo}L_d*ZGjWP|5P+bt63*l>HIkGq)d)o>u|MIH7FNQECV2LcG<!GGGp=+=mv`JvJx
zxnbzAdT{&~?1&(eRSmZl`$!W=-lgpEqk&C@3zh{trs?#y+wph^W$TGI(_LZeOgkAp
z&>4E###4bv+wlLg)A47nqt(uo$WHyEUOO8IB2#fW*5n)TFg!o9i3+(!()}SjUQi37
zWbC4rl&>&O`cI>QOI{80?2<sZ%*JMiZ(P>2d&wclF?U5!P-XX|>(J+-6gj>ySjVHE
zZyvkyr8+dg^-*Yc+trwbI||uXB4e(A3_zjNYcb(mr@Og*9T;NCb@k>qnot>mPS-Uq
z_>!Mgk7Oq+*tgL(c@>_t?_!w6BMH`Ss(BNBjrP6J^c+t3yI{GFbsG2I?3Sp|(#-b9
zDLmGv42kteO#qEtw~Op|Vr1^p|ITN7HzwQWDGPs!ZXPCeuk_WrJ3Y&9_ccVTM+3rn
zhY}y0UqiG_gFP^m_%Qh~!~XW*P{sk``^4)^zYL#BmR1`*lyO>8n7cddLqmy&xJPY=
zJv@|7><W+jPgY7Yqr{M@2m&clVo1(e0ss;E$7q0D3so4M4-X|~KXv?g8>_}Rr@>Yq
z*VqgC5YhAYL<|QssFWN28^_?2F&sZFdZEFd8p<RygtFGSJ|i_gEx`B#aW5P<lt~L$
z4nW=UF(uN~BC<^w%7g{nFUU}rE;JL>p^36tHd%O(gTIs`C8ja%x3g?&fEzx*{m7;X
zzwUvRHa(s=iamGTGx6j(kx|b*tJ$*aDQRHO1sp8VIEc&`!C?5qg3VN8z9;t<pQZN{
z&T6*Vnz1_-c38R4C~{I)j=bdqY_cjr!?DGM=y~yj*KkBXjQ(K3P?E!jKjubN9k7N^
z;q_6lns{=xA1>FmT3-(qbb8hWVaQ}?4FQCkE)HlVBH;-P1C9sEeBk}Dv_uv3lAQU=
zG%p0Gf4Akr6e_raW`%x0a`c%u3WGW<fsR35E1K?Ao_q$DJe2e%zxm((S*zetzRQ!{
zw-05q6*VYG@P!0s1tFE|MQ-Vgd3p8*>(;CG=G7>;1+Eu!GvgNZ19~?PMC^~bvw;`>
z&emvSRp(YSA$$!6131oFWEYRN?t{{4Yc(zG2`P=lXMeVJG5Zw99ra(<j|}Ku{1~nE
z1Lg00i*Ae&FfhQMwNJV{eObK0;8?INnpm~_IBeD9YJc1|vZGRpknLK9SeCr)2w3M|
z1@)nr2mW82bUR~6u1hWf?-B*dRw3)Uf68uAh)DiD!k+9RFjd>Dk+@o5R%o^T0nqcm
z@mU=ZK=-6*bF;`nb?Im;xN_VM1#+awglB7nCcH(FQ;&>5@Y(>rC?Ca7_%yreH);Kj
zigJ`JvSXr08CvL0J04#_8+MBmqFl8*R-M%TiV67z=~T2P|Dm1M;0zT>V$P@sl081e
zu9`io(HGztbS^07)yI15{7{k<Dy!{+8s&%oY8T_FKH`Wz)o!XC38%RfWZ@)*i3dzL
zIu5~CVh&j1(Na^mOMnsN><c{=i2M68nhsDqkLt^IRVQ@PL;FS7z1Y7h0i@=X`v6@F
z0+1NbzSeTAYQsYr)wbw|ZmSpH%j_&KcURFHL3r1jas?T%Z-q>CZqU9HFx*5u-Pp|-
zsB+E{`(7>Q6O{5pO#KzAec3H7=IW#>x*eTXaVP&Vs8Z&e*X~3gipi|ByBeuHH{h|7
zKb5A$r*tnm4Il7I9ruNQV_)xp*<0CNxwq)yNe{f0O)phEcu<dsD{E}jTiM~nWQl()
zBBhRgD=}ye6wc#^;sIvMwt6@^W8?SmO?%|6jMKT8RZ0{8=v#5n?_+C@c`K6-%A!<&
zJ@!_TCs2?MA#09(E0YZ{a~0Slj|=&5^n4<kgUYHd5}3UeoF4PzQ!&cF^3R?Y6wjx}
z0W~h#b>gWze&iiiNIujFnlO7Gkpe-=CI&E6og@-{E4xCzIDEjBwN&Z~B#4LibR$#M
z05yf%O-|S}VaY|trpHs)Dmvg~_LgE6|2GW>I*w=65o@bGCtT6=bl?X#Bk*z?tIZT;
z2wf+G<E-e4#gas;HhToAWaURDDUSxVT!Yb`RmL*ne}-68rPjoHxQvb_Tj0@0ai%Va
z-h5(Q=d4-_-9=C~8dLwoY63fC*__pCBJ~5ram6;lP+8*)wIN_*W^$i|leRdPrv<5E
zi8hGuAP?4GD%?EZManU~EP&UuVfikW2Q&`dNQSf`&>(qypEbUfPO9HbvApKv7lbxz
ziTR^&N8)JJ7&wEiOew9}<Gm-U%vm8odhRmo6d>OV6aoq4S2i_SPc$RNc!w3^?}DFg
z{iW`t^tCn7-5viZnb2c{q9l~(#+i&~FKY}p_UCME(9Z=gx9Y45+*Y@p|MQ@&kMX;3
zklPU5!hDAd+r}98{M(3Fo3zb4n^z<1GMa4;I=Uj9r!4`fth~au3aN!Kr&c~8-fs(9
z{{AJ~9^;p}xM4fA(l2t94(T*`-Wlgk5!R#9Qp&Sk`ib6XyW!Bgwa;AvkH_|&7{G5F
zdD~lB350@9+aK^bjJbPZa#qjeMs(5+>IW2O@KQSz9bnxlBIvM2Z{r@w#?IRjp*+v^
z%pb*klGKXRSiFqmY>gcgc*^eiavl$OKhxUUiMKA4{Kk4B@-lrKCrf*vlOheL)S|63
zYTLWmb~@<ak2?5_7OT9ufA~a_8qaEI3<Ukm{hSD2Xy&3FPdFm@IeJ0wdeQ4%6gYP(
zi&8Y7#%k7bzYhA6;sD9*ay)RpQeXCpK1OyxirD>JkSGuXLES_D3n4cAA)<ZJz7&o9
z!rt75z;jjj;L>Yfy_MamlK7m=y&N&uq7h$(8<4*)5eyz?-)IODllk1QmtOXxUcVvO
zjX2wWtEU3O=XdeMMslC;oAG^cv5j&d-)rS@Zoqp?qj-RU2V6({eCBm~TkSD1;S;W*
znkjL9Cx}%lkmK_0-9R!|z#-$OpsUw?oMZPypyWa3iFED<Xiv1}XPWKI_+)=GNiL8C
z9?&m-9{1sq`ryvQIq_f4v3o}C%s6K~zBin}qjx5@OUZ6{b~OED{rNwmnR|F=#?eyH
zfkzKL5)D{(r|6@C6>)7gMuT&CMUh36*Un6cprm?}jTOiqg1|g0pgq1b6FYkPaB8zB
zqDLL9%**~g8R!*VHS?6lTjY_ZJ@&NF!CS<}Mbkd~IBmSfo$J@k2|F`^BUzdLktDx~
zBfZ`9yfd+Y+5wvqzmrS570-;R8k7?jXH}aP4JW^3(*<CVZSjmI(u=Z->euS%YG`nJ
zPB?Y>VLd~S?!;;{)!c){xa!RkF;{!0Bu=2&`eiYR3j9zl-<f?O9OiBY@tDz*Rw-Cq
z<lO1AYG;zn7E;`^VDmIETZDvVT9Z8a$5wf!q-sqC6#$sm1Z7iI_F5qUn@I}#I>CD$
z8-O+JOv2W0$4|7yB6}R`$!&0nn&-RFdo9&tzYFxsc4ntymi7-(-<E4a>>G+Z*a~5>
z7{6G&o2)T#M2FC2c`b}*Olr}SC!lH7+=_i9p|tj$nIK!$gPQ^OKBB$?)+g&Z6gvYy
zGy<G9#9x<4;lUX_LDgWAMWM)byN!dXKj^Sx(iB@0TPELbgTkejkM-pMz~}VKDYjMs
zaAI9WsP;N7OeE0dPO*Ljd0Y!9m#sc52A(~(G0?+E*=n0aS!A4oZPsIxI~Huwqvxx%
zt%9_`T@89|TcF?F-DcZ`^lI`0?uaLp8ScC{6ku2+Wcyy*xl{J12-_vIJEtf$x-?uZ
zcW>LRxg}xP9d!1@fE9}t+Fl`p(UWKZ+ppfF7Ae{RA%uC-4r&FFdpu7xpUR=21FD(<
zI~=WtPZ!+3=ZHZ0Az~kiq8uocqdqFouCYC)r@5psIj*PbBDK0=QM+<>QrjH)2hPU1
z+GVF=ZOOmQS#Vl|ax?C*Gci@t?%3JVe(D`WB`!E_-skG(NBntx;vTH*GD5mtUJ!{=
z4j%O5nh2$WV7H%YNOJNN?UD#i2$&xy*sEh`{4Oc5S)$>Nb=R~1>~o=m_k3UJe*8tC
zPT~gpQXoAUl-pnJsy<7u9>|Wkf(wmWPh)v49+4_Jdsf-kYEKXbhho9L(KHrwzrcem
zXzog8?h`lk7BPfP`mLas4<H$Q7c;^ci>xUkP(xZ2`@RIpD-mtAAJlU($~C(sM5;Uw
z8w20osrwgRXUxy<AW<UhHTzN9`~}a<!0vBu%X{e4ofxF(Rc~7DuF&m9-GeuozW4Sv
zyC?GT36OIh9ol^zerzEigi6BRPKP&=DLd(bw=;%X34gZ--%i|n_%E>IQEw+k&qDZR
zY4qC}_eNNVIG%^z&bVOm*oty(55Jx9UmOuRJ0E#Fi4<_iWRC;VqcO&S*F7Ecb|yYR
z5zkEg*xT6$pJX)OOo8>-w-X!S?qQF|pYApFgwV$ki^|!7a&i`@B&Sd=dD!3)F6;n|
z{&?Km*=g-lK!8B%8z1;3uYw6`zdEV-CdL4qu#_dV$$>q&3YvS;4RMOV`2`N&K$dW7
zbV04cPOxb~eB2P*du{sL8FNCt<q}L1*`E>G2=dKIFnM-_On+i^kJ@uVRq!AsO_Ez@
z1lEK>64xDnx8TBSvor&8M`P%iEl_1txCg&-feKR|mAs;hyBK&G8mkgQFOo`XFt?g~
zX|V(zGQHsKB<Qpe+#QPn#z~J@6H^7p%YU<4jpDg$dij#;v~+{H)`B(Yu{YO59GK81
zXEK`UC3<d#Z;~S|fU~JN&X)y=E27@h@<8Y}f3y{)J+p?4D$yTt#=pCxmJfUoG&7S^
z%i9;S%6C62gGBTHuvXCw|09=^+Vw-3AAPu;1qPqI(bvyM)u~-FA6Jy_(ok+@B&j`v
z_o#cy=zuOgCXcX?c0hm3U~a9g(I<u?e8`=?;eCTz4?d4{GqINi5&6_P+}hXzb#A#^
z7w8b&HhhSRUFz*pf4W|6h{l`<eYR1A#;XX!;U+CyYjYv_L2iy;N;eJ{T(-nwG$pgs
z)|iXePURc0ZNjW7RbsR4`hkdAU&<X?dF+t|8`67^2*%%GXHcWC(sl{;Dg;YHgQyIm
z)~o$|@3lR`zZ*Y%uJN%4a$Z<F0R3^29f%*x2n{|c3ga%|yTg5c4hf}&F3cR(dY0qe
zb|k2Ujh$P&8sgUkx7XlBkA}8mCBi1fc@Qb*wL0uXKqMiNTa8Y}QzA#Nof43WMT>E1
zJFPLU{DTwnjAradCQS0I-ha07vn_T`kJZh{Hs@oifB3_{+J(S|^xIqPVhlo^A{X<I
z+NVE<>OR<)1jcusUDo^NHGZ_&6|F^C5euq&_kJ!?l3g5bgF>Hcy5OMn#{MlhrSw==
zb!4LK@N>hNd?4qu^ZS)u3(_HrqZIpEi&2f>{*A!D;3=`Li+0^e0OUlADYIEHfd`Z>
zO4Rh&cd-{<c;0RXJYRbxg74qH@EC4de5fi!<dR6*6+Z}7k}seTbKlQf8W2Y^Sn_S*
zll<_16tb&usJNpiw;<n*r-i;F2=t0|n)K^gcCWN+$irgy10L2o*F2f+3bpn6v}h0P
zN}Sv^pH$M`m9f&&t!>n<jI|Tk)M=x4Wej{}b1MpY*6^WS!BbZ?+QYjN<Dc}(OJrQx
zmGSi#+dA61hd=aC)kkAwPIHgv86((98TnWMa)r>?z=?KqE9~+3omc{UBKo7(bEf!Y
zw0fb_uk23=a}M+2zvk@eU5TYMG}t)3a#av0j}M5g!~1N4CX{Z2|H&?$7+_sDfo)rT
z=SA|1DKW>s0vluMu1wQGf;yBm)X9#%E_;zCT(kZCAEwwdyDpSeYuyC8r^F}R1D+N9
z;oWRaF66s1(Z@p^@Bt4ZRm8rwnWeRJf$=q)B?=-rkoruIe8fXmF4*jiRjAo3W8O_G
z+pJ2%G!fB3{<CbDPIS9y%wv{tk|wKGQ!<e=$C`kWTis!`@hS_MwYr#o7wS#QXU0QC
z!|!V6*y3oJ4s_pxOGZeC%tWC8Te>TgqLR=C1X6};buhz4N!{g|<3e(u*a{(sHU>eW
zQO{*5pgu1Mt`5S};H0%^XsmFFd9;d9_;PvpU%LjWcY(GbAfEp_*`YgwKoq-xHw$3t
zqp%(Ue4gl9zN=!;V!(Zkt$$ahAYoQ59=wyZ_G<#<@P0RQYZ@)+|GMnupm`C_u~x`<
z`Gj8gMqDR!TigeFy*jU^J_V}v_!8Fs#x`mXu0oMT$XSxHHff)(<{R4_PtM0##4Y-a
zxe8J6D7dqxhg**qh1w>ZFi^PWn}{wEWlkyti!<7;Oe*IfL?TL#plFC}sI)DSiFU@|
z_^+Y3w_O5TOJp4-yqj&e=H`wIq*xJttpj{l^g?~JSB;WI+xCk-ox=xu>_9wHawnCp
zJQ(Du{hb}k7PUKPht;P8qqZH<lwE}9O7fA$=xl~6b~G@D6}vwCm`GY*?6cz~xSlZQ
zgciRdHh#F&oYYj~%ECgN(vUB7CJF1bfOa#<GeQuT&L%q>L>%{rNp>!V!-k2uTR>cO
zg#*9^txK*zHZ_z_7o+*WYO;KN8c<%DG|4V$f*$tLYnKHnibU8IAwUKctloN$>5GE7
z_C>(OaE2Rs{Eeh=#;$6r#IX^t08(G+r%o!>^|<rcwZNncr<Si{pt@!<Lhx;TqZ#zF
ziQU!wy1F;D4(}(+SfAYxHl_N{-^Le6&?>loC)9CZ$lFaldOV1IFK`^lXuX*H4?zrs
zrBpfCZV89(+(1(P-HsVvd5;<X7z^cP3yA-{6K_1)-|lJ&YbmPqQ{YtZZh&?#9>PX&
zU+8ljdv;ZOM~wM2P*2zc?<BT8{2>-Qd+?o%OCvsa+bBKOxanZ@I~hwMPc&@eA;IEq
z=hwT3qkU2m@!371_FQ?<mp>|Se)+I5?_@H-{B&}|SRQ*P`?S;UjW}?RL>sHN>54@9
zN_$+RBda;}pU@8-TyHTBJgH8(d^;=7x$x9G8FRoF&Tq4P`cDfj&v#_w-U&^^KS1)d
zjgO~vx5Z8#PsIOu)WpP?AClow#I^M{SqN0%HOu9F$~#GPSoTMos&V<~lNN2jrUfvx
zKPPc@ogTnScA3B1Gl3Oe$mtk-HXbY6^3|>)elCWpEEN>Pkz9D@yBclgJ4u!*XFJS_
zj**524>q-HY#X$4O})`e=v2mDV6ycuSXIElMr-p#NQ_x%Z&=F(?_BT*f_~43Umbt9
zd0~>8cSb}Rzn|u<R=Z$f8?m8Q7Z?!F;{iquB7#TVFAhkUBJ#E*#_z+PwNyy=f3nY(
zMI$d8ZOb)W!)nit&|@ovtGm*2!#!h-0cjG^7x|I52YnN=YApi623hSI@>Wr0A>7N_
z)rT@pe`5cgvO*wKh4T$abO!taY_2;fh;kisb5H!<!l_=24#;aBIodCX<QZ{MX28~H
zOb<n~!RYGlr7x>-c>MDrt_^7OJxeuQTdOv4wylqiQPx8yqiVnMZiqd&sKPc1(3Ks&
zx1r3Y0F>)Pano&^MF}qLDT&gS(#m?ix2^H_XS?9E+q9;&J%a<?wmsmmah9S7-7zu{
z_gBBm*pT*rJ(hMVB_Xozj825^`?YG9TKHu;qR(~<kgsD(IkZRL*n+OsPFzX$>c}AN
zPt=6x5=l_V{t~|LaG^X9f0JH-dYgjaRa;n%LqWf;l<EJlh6J3KRy}1$^iz_g{Z9Ci
zenL^!MrbgR*`?Y1<FnUL(i|&2MO~lcA{RUb+5AL_tnM^>GFF<x9(F4B0LM})opD-Q
zf*sOjXS6!uMBHU(^$VA-B!QiKC%YTGFu%cheNqITk1a6i7j{8grL)~G#sk?F*>Io!
zY({>U<gC0T@O)S95NDS~I=+JwEbA5haAhA$^mqF_=8X-;8O9fyH{u+Fe5q&7UOt?w
z?_>gOM(h;A_La5*cG8mn^R++*;rnEg`u=HO2YDL8EMnhi0_Dw}QNZzf442F2NkjpA
zx)IO5sEqTmefv&!wOrKbnyVxneHZIN54#zY@jw{7lfM@^Ts!Lo`k}PQUIyl*-4ZQ`
z2=jD9w}sJ4kB#zzyFZqCpyuPqzoP|49;cF5ysRU23AFktRtM`Ur|Lby_6+5)p8N5(
zrELwn(ROEi83PP{iVy70=H;aF?~k~^2X`lyux4P;M(xhHe5G6BNAFHYxrY_lL-9Aa
z%7Q(-J3Go55_5aEJ+eEy;&UE6s!=<Tf7@+LbWM>tkL}K+4CEp|qsHn7A~y+pv&VN|
z7!%nV@j(eExukF=iCPT8S%MAa$#@w4u(P?tp4y!_GjH6{=;;xVO~Z%So#O)E8r+>%
zhVpv6K%Lub6V%?-704eGH34Fqkp3pe^9yW+<nla4Wby*qHdW+6+E04cX+hF11i|Ss
z`PF_cct&vZB+B4wzB`jS*n}L(?tlGUbbV?3GMf>z@(d<$z?qt34?3Dz!nr!<uAVkq
zI4y9COnFSOnvavBB0|v9CAdUXRRZnFu-Dl<0lV@>+OZ3SD4z`bsCuLU(kIE2YD#+s
z!L?R+&E-l_r=Ddv<&jZELpQglJY=?5Eqa?h275{LM<47Om!<L6@B7YZY}xKi6y!2P
zF?Te>%VTOiBn*J52mJKql6j32y)hun=V<bxhL#tRF+t$K?Dy3c;f&;At-`Ts_~SgG
zkJdi&ow!-)jK>dHL3?J|c<WT36sC={eRsUBNohWNv>M%>=(s3&^ZfuNw-TuL3q&T$
ztr_V>*?JokmaZarSuZR6acvNWyCSX&@D7YN>qb^-A6BZ@Y`waad%Cm@u@y<YL(Wst
z6}=u;dUoEXSkoE5{<Uo$f%HLgnA{>%sxi@T73S{w<=AYC4k>*?GSqGt1pKj_3+@mJ
zQaB+z0tstKBrQjQUfKxQ-q{Gd)V>kBW^+oIX1k;PAW^IxwnwwW@=NrX_7#^t^lJO#
zwP%%Z?cITxJgO*6=bFv)9N&k8YHunqZHJ>NdeT-qqTQ2SMYxy?wRwEBNWJdPbW{y{
zhCg17;S4eFSfCK88c2m21RlWfdO{Q=ux9uW9Py<3*EhAUZpN9C{N=mTEdlT7)A3h*
zm7NLX@FK#B9^}BLjN`?*m{cwCGF*qx$I#UDNjkp^8qcrVPTz~WGl2jvJN!Q)v%;sM
z=^QSmc1cJzp{l`i9}(rUns|t~T~QNKdGW<Ltv?TJm;K7V2yAP8LB5P%7xviI_!atc
zF&Fjwx;<pcuEjbMlI;TGQTtl!y2xb*`$prxnB#oCuGhG#LbDq|6OI!Q`BpC}a*Qtf
zyFg*mAMK|0e^L^8>G|(Pqs~4|rSQBTv`;9tZMOn49D!kO3+;LCFiWlN062dnK;7=B
zd-w3ic6YZ#nq%-8LP*@YZ$Ame&c3exF1x2r72MzNYZ&)x#xd94%{Vli>Xbe3Zrss7
z&f`e&;JX=HPPT6>x1-+8_<tFdYkF<;yIJy*GOqkEGf(k=!2MA4bW(UO4gf#=ZW0P6
zt(anuyqj3I$0j8YTYFUF5Kxi=U&rVd(y&pE5f8X`vvZrH`R@p4XKd`d8IvwU5A?Vu
zf*Tm+M`FBMHo%iH9?X0WRY;)qQ!z>B`(!tdhvax`<1`9Y6}bJvs>Z7^$-XD(Jy-;<
zoA_?VbGhO7#n~`fWNPT@80fMo?<Tnibujr-^)98VvuOh9Zo!GeOpoU9?|ysc-Auyr
zWC|RO{;fUxZuYGWpK#q?PqMK;C&)GR)%f1eh-Z$J@ZOynGx1Cu<Wh?<7A~rsW3%J$
zS6+E#nU#xJ7_8j|urjC&&vL1|N>rYBecc|L7gVn2yjq}n^d>oruf1B0SGJK)xkjL!
zuOb%JY5|+vYdv65k<l97&9?T>l)k;TIN(*V8~tjuB+y2)(qT&jQ~{$o(&Mt&KC4G^
zO)eL9SGFT?xTSAJki#8xNuCz45Vm0OB6;akcEfp#kncsk1Iup}wJ}xn_E~#?!#c~%
z3mVStDtXe^SZ7SnMe6S83Ks4NbH;kqnko};b5^*h6^59e@E~2f`lDlSRvmGTK9iE%
zZ-XM~Jm~mxAU$VI%Am~owN@*#r~~nRUCbJtJ}J~P3i^h!4e?>4wiwDvFLA;Q*+z}n
z-5f<vH^saNw6x8FPE_Vm+fv$Zh=^UbRdY;NB42=Q`f^f?!0;jNY1>6+60Lbi&JImz
zA`<M7y5v^!H6eTxel{}d>tMJmXymz!5km+*dpGgCMez5SEy|lN+pA7|9jk1=;Np9N
zRF@dMwn3FUGzKR+7_*)9$CvC-U<L{+kqQrMex6l`kjp+h67#PpbYYMGNZW;|BPMz@
zUUCNEJ$Fn)Rkt;hIuJ*1N|Fd-Iw}396T%vH(qbpo;`x*ogNx*;nC;>+JFT|x*+x4P
zlcy%gKkaOc%~``fJ|~K=>_TF0w)4X5zt6G@(Jd9)zNocF@W!l$?me3N{^_+#dY#fB
z^MPI#ey+3bTE#JbC3>`bY>#~&z`1&_vM;m(DVJDc`(MUnCa<uoK|Q{NViz^a?W+JY
zanhto|7zD_O$dgxuQlmzuZ#PQ=EiYNwGg|m-Lx9r<qfrNbuIn3ngC&l%nzk|G+;&3
zh5e?$xowl;u-|J@J=1%^euxR-0gHuPlc(}8MF^0gZpTma7tKkrw!X=J6bhbN(C!Ez
zS7fXVcZDyCGyAEu`KBzhd+LX=pv&%;D4B=d*`CB~sP4Vb9@vvuBm$-1Tprw$cs&N?
zWH}iXk8DB8&9~U-Xp5(%OJxu3$&TP0$(AS~?HM=GyQ_~Egt14W4<K5a`|MG@obNR$
z#_ZA6oJy%2YL7+hI-KqSHFi&u2$~8V_V}Jm&=^EgW&hg~d(v0C;#c<Mo=oIGhepDi
zr}iWXg(0gJ+S3}boGoVKM$q9n8owvw%;-=&Xo`#)6M{IN^T{Tvv)}D(vL5SQ2{UC+
z#(48&zVZtVAIsD|nH0jo9U8Z30TFM<oK4q=Wt|uXh#Me%T7dO*nR$-fXEjw%`|IOA
zr-AU??CkZV19_VvWa`}8cBUSQ?L^ZaW8(NqsS?o#nHaIXJSbP*W|cwU;nR&)6&(tm
zT6do2ke1pPW<k&t_YT&jTFZ82SGvvD1XcY?hzwgBs1VMKtMq`?#qh4vO5bcl4EhJ$
z2Nnl4%1Bu5#qZ{$eX%8bGJ!FSd)Z!G8er-@TffKWE)$vf<_A1Ze7Q)#_ta@Cv<)Wy
zi#2N9pU*W~UQdG@EG?Q&g=YdH`b>aulqy)e=vRZgd5;zJ>`S`JI%A+BU%+bX7BIho
z;PBfMz#2(><cnQYXEYJ)m;RVWTG)u%TNA&Oc)yn}7!*vr)!o<k<ronQJ8GD<v2wMf
zTCjET0;M$sSg&piQ%+vnpm}k7?Xr!r5TtsyO(IAu`moJvmr@^X5z3xiqt~`ZziRqz
zi?LHx)9ulWT%<jDpZE^#BF|<`s*?d55?};O4C8@q6^0}1;u<Ag6l}Lp^2}j-)YCIv
z+uoQORmMu@{{4I8MC{(Qr5v!X!9~nCP(qkwI|rj9W@0xTwXU*5`nKpqb``E8Lv~nW
zj~h41j>On;^B%B|;`8b5u%kliMdFn0m<~keB}p=LtKD&J)aTpyKsdTi1U@}TyeH$&
zvf5fZ71ZF^n)ypR9T>R#7!3D}2C5`Kh@B0pbG-K3x%eB_hDV#w#u#K5^i)IC=EZ2o
z+H~2cF%qju;;Bnorjk=F_w7KciS2wP{v6=i{kayXi|?z~zEH<*DkhLu4V}7n9PH7x
zcvsbiGVFZg{Pa~o%&lp&Ynm2OjU+4&6x&EFWZ&pjk<3GOU1K-6Kf2wB*IZP(Fq_!7
z8iqrsxs#k8_Fc5_`~vQpb5ku~+jv2N$oD#mU8RJ_0$=A3YKCsRJ0<$MrJ-wkJPXI|
zSpLc5?8hK5oSxoxN8pevotwZ*?#5h)4^dG}kZU>ADOXQ+Q&Fo4of?wteeHnBQ_bGX
zcw;wt?YU+>@LqNg@9*OCdhor(92b#VW|ZEPc`U`~_c9j5k>nR4tmE$`mPow$>7qR>
z$fVNg<BtBwdl>^oRmd_v`rZY%r?z-Cr!fJWy#|5$?6G*P?`v%LBX_KZVj$EX%<AzH
zUPPup@m^xuT!Z-<T&TAvM=(W6WKXF(zA$n6JuO@sutN;kIQ8JjXtVL}WjA_CJlceK
z3YT)5_}&Fq<EIbLBboo{S0~T#XOjb``WMP=O3V${o19G*>{`OCZJLH&(B``G>7vt>
z9VuVvGw}n*I6E03#GVxZBD5$<5>&>Ik_*F(cp!fPdeWI{NNHn2!{5s!heR3F1Xy5Z
zzn27>3BR@S5wf^NvyGeqmG5Q3R?7Q<LOk`Uiar(a>>vfkG2BOO!ASE|9wb>_PFl6l
znAeBSxkitb&Ft*jAWcKv3v;YabI+snW*;=fT5>w!PV6^UTddx5U6;EgXk12$o~0!^
zP#z9<w-Hp%B;4I)^zXJj=J@mQixux>oZl}<e2#tLxYnq3!6kSVL1)?nsMxF}-lkkt
zn6j)j-li0J@a{kowvG1{x9*rC1$~Gc4}^a`>m}>fR93Uktw#fp48n@<WkO;~1C**1
z^amWO@R0p9W(#E&R@z`Z;KPO`;$?wFP*`Pa;~h}~FpG71m$FE%*CRpXZ`+1=bdiGG
z{kBoWuk)n$?tHE-vJ$_<Bs*``lDVD^Rri*dhF2wkr)>=wo<PERAAEdUyytuQ6Ro#v
z(p7Bb9V5U_yM5J$v<QT&^Xcx?bD2y3woA`QtoQ@&j-5H{H@2s=Kwg`{wX1v8vZ2Sl
z@%C%rMVuK2^pq!x<)CQK!f@+oK^!|2{kbFOinzkr;Sp5blyuaN1o2!}@-zJ-!C+fa
zfY1{$9o3qZ&#_}M<eYFwJ1%fl9vVdWS-j>1^YFTp`gHwBq+F-F3Z4p-++PNH_$@oF
zotzqR)pyAk_qG_e{gZvef3>rkX*EUO&S^nYEnXMt&#M8kqIMy+>mWH9E^4!&$6RHf
zMmvh3dp@~KLHJ+%$}X#!Ysu}3kSb+1ELDX2Ji1rl<wuy@7qLxzm!}fjDA8lDV`G4y
zi#TQ2kg+R&rLVh@BZP0v>o4`#HMMK;4UD^+7M(y>xaBv(t<qIX*JE<U+#0(P_~)qy
zs`<~O$#?Pl+#0(nSV^F=?={$!*OTe!hv+yTL4ZijTcRulskuY5yKigselIQfqjtz@
z<Re#H^x7Q_GQXx~PMzHiC`nE)MS1<Car46;wj<2l3rbb--tLPq?1`fsdAPV>_q4s)
zVZv36^l?3?%pcgB@wDN6@b$eJo6k}E$42eVcr)S$lFw*8lP4W9;Gw;ln9%EW1F7M?
zH*x>uW#H~vti~higSRV1upix<U8YHoyArjFjS&h9S)9iL1Sc9y+kFGkq-&U-W^xT<
zGv6C`bB?kn)I3Gqa1FL6<1LZz_EhwAn<aDso2U0C=>k6zs*S0ud&Cizh#$M=PY!RY
zh3c`(^^p@bl^hjtnrw0a<AT#{Q(_X>=LgXDPF2h2N!UQmv1rNu#6rF|bL}h1=_#5U
zzlmq|W{ep@($k{(0rITKh2J8s(p<=eRE;O&ov}AbSJf%9Wu{0uZrMzmrKeU@^|J#9
z4#+Mm*VM^X0eeFGQW;c3b7xfnk|iLM!8~D=G@~AY@poGg^W(ILY?0M_llaivXf=Ty
zd72>UfYl0xerk@^33cr3E3HB6dgZU^+|Srzp~clHi5a#e0I-8Sk<HS0s&A;9kLbBf
zz=uz@+w#CE)$ziS!d3)y%oN2?Pd5sBn@j0#7|#nK>Tz{iOFZIXocw-TgJPKBKeTqW
zghHKG5G=B?c~;kz)~S^u#dNZfc8{#|f599O2c1Q5E9(*Bi!p>0N62u3M#rpQKh$##
zV3V(jweeunL4CvSKS9|RGQGU4kx_NxD2!j6R*;lL>-JvwKf=SrZ6pYjp0x*Z`~TT?
z>oqe}>#J--jMWEo+bAfDR@$ccy}a0In@i|mE7&4bT`9@uy;b<38NuM>l-U+^WJ~wk
z_PrNMK^xf8s7?~RX@|a-daUyuHY5U76TP!Do={lvVc4#KmTRHL*6#Sb#qT?NO1}fq
z_G*-$NprS8T2SAE;$kst&xGvRW)F&lB%}6|#16$IXYm~!j`3ZRK_PNPOV6eZ=jD$y
zX}6CQ`v^g|qcPYkuS{HK$3)^f_gy(IqLIto&&U%R-6aH1^LH|O&s$XAV5eeuHiEN{
z(}7nn=g}E~z48|p@@zc1tm(?0b9*n89E(te<@|WSaGs9O^8uo(vC%Gw@C^)1S#K@*
z?`PPj(T;D`E@_6z1%mq$PH^#h(!VUTD+1$1Pq2u`4Bay|Opagdm5Kz-WL(Ig=azIa
z`%=B581hvOmR;`Q^+AwUENFHukR()?b^2OFAg~r+cI})xGQM0FWCBECH8<24u35Bi
z18CRk^MAJQ;_p@6NZmKp?8-(A65q!on~xmu-S&gfYwPrjs4L2EMem&3gl-FkIe9X;
z1(p6t(Y8B!F^>=JZge4tpx1tizxia^?Vb)?OsYtoUXR4?M}u0AFSmUe*Tw7EQ%F9r
zFEO>@1I0mma9`r*UQmgv->7{Vv+e5)8~^Bi8A~V13-{%R)Ph|?x?*(b0f7@32Sa;g
zU&fS@-PB!qDE1u#5;|J2G5a!+A?4G2Ovrhf5R4FG102kR)4(3rYornnlYBz2NgZQP
z2H;Q5aOgcH%od=_^t?e&$6Fl4UbArmltiX38^148JS^m{&SVB`g67(Y(X40_)r6UH
zE8-)Wyf1N2{A{7)lzo|e!~sCuqE}*_D*Wb<5O<oMdI=nwuBqdql(%Q}SPPT2X9G0n
z8Mc+@0+nQ}^fVYV0{!I3oT`}5)Y$Ic(v?2-StT;#xnQ$nWXge_f4ket_a%PYU$9CI
zdSxG+rD|W2j5@gS&5QAK?WFm!1u=dS2l?!*I>4cz@Zu9S@#LYzkfAnU<grxO#h=Z6
ztF0magyq}fXznUJj)5hCC8>~ewp0{n{}J7|EL!6D=$eP+B_g<ogon$o5bkvY@YKei
zlK>|o*OTSdgrjn$wZvGYg26dmui&ZwY3)H#Vk)7Lr?)BSeFZrxtWz7N3I#pYt-1p@
zHf9Se+Y=O>G0BQC1}RVfVf{hFs%EqfwniH=U5rsT4GIMHq3q=tu#XGAyHc&ybQWWg
zA0#<61wXX|QM$R!YL%@Ix{jN)!ZrlB6tL548}~_0^o&2ju+PANHbsx&k8xBFT*+%l
zg|L{xygJ*eRY><L+ZJ$>JpFgu9%!;JDc;0oTXQH!Av+XQLJ+`8>L#U~fiZebzQE38
z=HC_IsUST3m*Mv*6D9~KJZ+EYj1bnz3A0ypXc*{jo|AHH?iYsacG?^W(uRbI40PHK
z3TQoZJEZnSGEb93RZGF`l8i#s`_w3|I)w)#oP4CSi@l7%o+fluKVr%08Hl3D#{@ss
z#J1xxYa*@O4bJbjb|MfOK7hLbCB%a8NEaP;N*m2pCU#nw*N|}Pj9zwkeX`!piaOkv
zh=sOu@g9lPvwqw8Sliz60lN@GP$!(iVd~b{=f+z5R0!ANN8fChqF*cd)Go)Kt>_{F
zsa?^A3r#Vf_vg`caG=A!5E(r<(WC#q6kKY@ldk@%$cpp0-%V-rRSbZx51ZmOVMekl
z*GPRG(?|Yyc;Cbbex2zqrbN=$MUXijbaz7o_LE<faBv~K01FeV-D%&IKFB|@Ep7&M
z)&vRNzE|UlthUDw(I7>K|A*a*r}?<Gfa~oTnxm=F1*9K^i60quM>9b{`JbP?Zg+*+
zi_JZr7+uujE}8D>_e33d4hA*Po_Rm8B`G#BX8W_twL5S=us`EarA&A^dvJfoJULg3
zHfn#Sgk<$v+E!+%=0*!nBYWhb=u}FBWDoC89GxDBhkP54?9Vth`XArCMNX<m_b0ya
zi(mZ0#zcE($gVfH$AlJ{Ncf2h#n@=#icohDvd7f~W11)WBXE3wVw??S<<)cSNrA-d
zia7IByhOt9!aaTe1y|_vp`uXVpUES5ed4}FS{G3ujWi(|gE>V%^{S(bOgfVT=<p#w
zmZpdzb6`kQV|FAfwrSCd+11zCbPc!K#d`MRGs2YMTQrBysyPb3=VEYFl!TF$KvFVl
zX0$;gz^OhfkV1iwE`76wBhqJ+mG93)GqR-nCA3oGF(sH~m7Z}Mgdxm}XN2kqb6ueK
z9*0Ir63xIf+_^JvjapDW7KpWKvAUToGia3R)Pgh_&0W@@7R{bphApyKzfiB8nm#?Y
zB&Oc!3HX+3679`-ez9fovM_uYZ}cF%=NG^YV}+K1bI}?_%*ExF*BDL6>@KOd1Q@)T
z+pIPI#_eF0wTp&%e9^3+&4!PO`wZl*GZv#jhUIP{*Mupa@2w}^4Xi3yQGl=>^;`e`
zBqS}Yvo$dqYJj2*26a>?_2uZ_L`{{ofj`Ux)zLa(TeS58?0nERXx1H^%^PEw@iS~w
z^v4krif-2Y%e-a;5Vu6bjz-%Wu<;9}dT0FYsUK{+2>AP7+YSMU9EHs`6sWebRd))7
z24w7A`sBmd<(ax{ceL&!OZ=XgvoAH53BtMK6$zPae+*ymYV-r5JOvk!LO3!H21aZP
zR{N0N<1B&J2<#8X`>vFV|H%H#PQ*3Xr9&=VaL$kRXJWP2FX~P*#L>XOJz0*$RPwmY
zLAK)}^_)EJA}97=$SOeXP4~wC!XV(4j!)Xj{YjIMsp37)I{i<nf0?@(oE8z?JH!KU
z&gkn{)a06+vzk=tf-_*})MORZ8np8PlLI5Aj=CV)SLd4SVl+<oOQvR@?$5+F;*;SL
zzDbjB&!t%Y-%hp5!m-xFZm(##D9!PC&=ik%4v;U@wv*HY_NBIe7<%ohCP)riS7O>%
zLcqnTlHBO!Ag%??r_TJ1eXWgFQE%UrHcH2UT^D!`1`p%BA>7=HzR$i5bQs(fs`g#9
z59Pb_)SE)Lsu>Oa_aZLhO<(d-`yr?YYxC>_w?xYY(C)UHBOQBggl79uo5#;cSAmcR
zU4y(x=%d{g(!Kv9-AI1Y`v&A#yQeu+EPA1)!S2VH&K=o)_I_fY#8eVmj;ZQ__Y-65
z^QyJ>V7!EzV8@3Xqux&poxHlQp%)k(P2C$0uB@i8HYBfyqp7o5u8NQ7EgBI_5|8R_
zN3P4pynn%2B=;2b-|>d}I8<h|vFgY71Y3Ii{Y-|4bP~A8KcS!et0knM%bwIv&UJb`
zA;41_u#zM9>GzX3P!6{k_kJdVa7`)N`1jLS1=_gx-~xtWoR4NgwEx|ldYf39G`Hit
zO^(h)U3BJ97*2UVu|F?TIEYq+d@j+i18<#aF-MOcvFRGK*A4m41W5Qgyy;oZ_<0Xr
zx98NB8lLdw7y|rAUCtHCX1<?J`IbJL6@z({#cXvVnuY|bg2|nd=~R~Hz;2>$r(W{G
z!XV}aqNF<`YfCgFD;;M^_50a5hTjjK(5|_w5n1P=H>wqKSKde#mpc7W=gUrfm|&o%
z<-Bqo#Nrqbdo>Bwv~rlLyowI>N(6&UVW%kxwoE^eci2;iF4t^(If!kAaGLqc-&&&}
zvb^Z}S(SVYm=cM##NQ-z#%DbKp0lQ#vLr#3bR)wEYlZhSsgkS_UiBS|pLJ^BdiPH0
z)+4dpR1njn+cK>vL;(7qevDgx42aPrZ)=1#1)XhBePJ{Od-?rLZY*;V6$hZT(L8CX
z{%Py<-u1o6IJRB@su0a>gBE#Vo8RR(#=^3Ve)*DZ(o1I&t8BASS%}GZOFV%G_ea~R
zS&)~|*&C*8b;ltHfndAVC*O>`-eo%i0v<eEX>2IAX<KuX?Nm?9Z4A6iz|jPH@N3?7
zYi5Ke54r+)PY?~J(rJ73mJdB$zxKylZkMEG5e<evWX=amyB-p?+M&QA`GyON-r@im
zp^pftJNLUi@}tsckm{El4XBu++!w`;X|`w-{7g8mQ3*by9Es*sn<McHi(@AP910?5
z*m|FQJN$z<t&hYL^6&*i<TFB^24~f$-X$6jlRBrL;KVEytmAx4v4CCof~Mqo3w?}>
z!m<(LzI`ehkkSIF@=^@P{$QCdt1U@a;nx}}k8gN#-TJ(A^dJv>p~mewLLtA5<EfqU
z+ZuxKC{zJo#dEdWVx8Aqi)V+K0tdd<Fhml76o>dXqIEqR6<S}9dAZk7GBEvz-3Y{y
z{2|J>YLBtYuLj=*HdANYO`-YX@M$u%Xv?<%_6POzyda#3xAc1M@Cmyev+aXh{U{{4
z?NH&9FYk_4FJ*~Of%kXSjQk1_Lh_Twc%I{djdbt*OmHeC4*L@>4W1C>zSaV-T{{pQ
z*tOF~?SXi5ZuQ9-doZ3Fx>CM}Q3tZSSXrJ#0!16GA3VU458|N%iFI^Sx#U2`1bf}Y
z`pAKwSu2iy_UB}a^a`8*7h=?-2NLgW!xAv&K#~<0o-6iag5h^t8+$<4@+2xf9>}FT
zF-%*HYY`vpNzG$5*X^f-Xr*s3@}sJC3Pu<g!=mr!-S`*>94LK+Hfn-!T3BHd^%8e(
zEKie5NOYjz#%6sW6DIP_Ejb($QFljgk|A=&PAiRb%rnSOKag=hPn_%C)>JPkk%Eoz
ztWcbTLCv0vfeL6rAkB<uSI6NvQ!{lB&smx&``a}pvkzz!{LO8ZXo|@N3f0Snt9uEq
za@ny`ldDZRC9493#e`M4_HUjr`s^*-Vip{@;4)U-qD2i>9Y|n?Itl0+A;Ax4U`w_7
z0ks@gH>=a!dr7VmmV<^Ey|oz5rMCD$#vqfDJ(+rKi6C)VM`3154`hstLnvLSm&GRh
z#qVu-kQ+IGFLgy=U%4o<y)|n39%q*h<@^CjYE5pGLmkJb+amgw4}WN_0flqi1HY_2
z)}fP~P!MKZ{oM7XGw4HfEUIw<DdJ+@Bb-~@S(-tLfhpEHRD}InhVrD8TN8t{qFNe^
zcI?fGFWJj6(S`wAtFisi;;OptKz4EDlTYDqDeLh1fYim|xFP;*@9nmYfoORF&)Q8|
zWAYI7+U62o@akUM5)btr$6#9n$-1Ps;0xa-q`9EMqPJ_t6&(0G^t9rSf3zX3{z76r
zn%cVATsxyhp0It6$6adQh&R=4&APJ8_QXI7Nv${E&Fg#B6a|s(*Ar3Wc0jX90rLl=
zE1NP!;@P2~!K5h@?Qjf1ya20uBpzI!1gj3}R4=M{-j0gSh^lTPKNc^klSRcaVL7+K
zPH5zX1OwSgtvE`#E<_VMRT}%1SL}2E;k!I-XX58sGiKP?n966HK6{-r`<yl>x=xZy
zLNhQuFN83Jxj)ARjmtG7Zx{85s}%8DqD}?&z)OKJ_fT41j)zsbc9ii~)XnRO+vi$a
zybmz+d{LrLQ^)W>?MuDHcZIZ#SL3DUH0!jl4kZ1nyGgily5!7Drc-^}h*>Gx2DQf5
z`Xp&L<n8SnA=1ia+^z?8+DO%OBc{TU<~bX_jVEHV-1YanAZTl2*BZMSe^Y-O?Y5>r
z=)u|XrUPAq-HN~4nup)B+doIcrGUTNkJ_+vobi=;b@n>~HXnuQT|F%>FSDOSlX@69
zG1;01-ej$IKS;z^TvcWEL3V)E<pz*<^Y*|8!I3@5)JA*ogLEqSiIco;BJ5_=2N|cK
zFtZ0a6Px`(#-5vok9*O>hoUFLq)V$iH`46kfX=TK9QluYkXf&rlKSA$5g=({Y|IB4
zN2adYV;^K(y=f4EYpgnBxF)IF;{n=Z&$w|sp|_;I;)ed@2+E|tOKG8=5-!#3$EVc*
z`P4%)@hB5<>ibg2fsNNM*bI}XIN^gNK6JMA*+k*D*uCo}>#@U~ex}4@UpGQ9JASGl
z;aZq%MbkdWI1UjEq)Bhhkyv=TaLl1Dct#-cooTaYg)+BbZj<<VJ{Rbarq5;s1u9S(
zaJwa{D>}R32;;8J5(0j_`b}nbG$37dO5R%@LvvpJ6ZKrB5b4LMp-R+^xc+%DY>xUz
z3xt%uw`%?5*ExuRyZ8qeoPFm&#cG8~&NZ5dEp-}fki)ql+D=+d*4)Jc)jL200_8w$
zi8c*}WB1x#8tvOW9ATN-!(~zbu?3fpwD*|$S8atRvnoXeH)`~`C}Z=n1*vJzHKRof
zPzOz|RXs88fNA>&nNY=rFXGUw5TyAP2I0;C#cdl<-2z2z2e@+S9dVQ-ToFZ@&=R?(
zs$aj<VdU}tYc#VYU!zF1!I(JT8wpVbz5_|n*6KsTWVB8@&2<B@+iksYPsYRK2D70w
zcS1ipZKFUmb=xM5-rq*z3fBv74zh8JE)M_IwnS@p#pISPidGT`yDiWpb76;VkH6V}
zq*sepC6nlomdVT6!1H$ovC%dVTg#5z^+6K%3zXy@S%%I|OjLVf?iE-YT}8ZCkXV&g
zDhli`O^Y;Tl;w`jc#T0fjAaMa+UW{YJ0#@2j`U$Yb${}I+mU#3nWLx0J`xG}dQ%*J
z-i~S=stWie9E<tpng%G)Ewnkn*sCYv*((v_k6y5ofo5$gP=Bg~v)k+JbnK}PltpLa
zchdW@L}vpXmeln(=b}|JR^0PJE0hQqq6z%XRh}0$n|WARKGhR?N4pfLR@G74<#M2g
z<br&MjD1C@qx520pX>QWN>a7j7uv2aPGc&zFH7Tc4Pe}>q6+Tlt@f3=R}7+9vuj#|
za_+0J-mmpT$&cxqKyLUr2f%d!LmSX-H-aMRD6wyY2E(U2?7R5gO`4r{Q+>#qQ?&23
zod>ZU=RR4VvmdmW?4uqUv5#+kAlFN-7U_KRs<}=}&j)l{v##|@tJv{>6h%F`=^eet
zp)`G9cQvkiv2<`S*iY(GhiGswo>E$qkL`ZUx~Yw<YIZQ=!F<FhijjgJIG9*>80hRl
zHNYtYe|#=(qYh?*Ij&qG`nt2l=z|&Wt}k+PB*g1rV!Iv-1<O8|j%rH$)Y>Bl6Su2o
zuRVG&<K6Ye9LDZ3F*uyP1{vhB=;SeAHdf%fP_m&(eEeYga(^+^o``|yRJ+KYj6TE1
z(P=&ve`9YUXUf6<&-L9?#-EYr)3|8Z--X~fUNcU5PC~_O!pKh^8AX;on;3KGBW8Q@
z!6f~_B4JCO48x`z%$P8I&z+XNGR4$`*|AQnXEOnd=whbDbh+U8b#{7SQOUB}Gb63B
z9O9P4(e~`YOmrbrK)*ewCV1^}CGF*FpAk*UakwN=o6Qu_>H)xqJ}c<m?2ZY&&0MX6
zh=2H_m5cKAT`g9rrwXoYRYBn7zG3sUWKHbyURw~9_yc^kI(ki=`WybUnxG3C9FNZA
zC0koso71R5>f&$o^#j%re|s>NEk2lJMHbWTg*|W<7j%}yAh~vK%Yh6j6>yVY7Wn_J
z-j)Xp?$cPHV`N}p|HHV@YmNG)q&KsCP|6j1&UacymiRrtZ6dh0ioRYy0=bqWuw8o}
zVaUCD{R70UOe-_!31&Nk4z8Ma9z6n@>kYF0N$U{}IY`*{#Xz28uFd+@JlDkC*&}V%
zXc<v6coK_2jW^fhs$Y&rSP!#pw6y_^B<Oitr-htLH1>MIb_FUW?MR3geMyqVb+E}c
z#!z#UzSI*iYzhj`U1*y%C^GASZP7xxJjSIeokd$?JbVZ}W9>F|Nr6?{^;k8(s?2ui
zGvTB|eoR)Sp%`!ZuWhGhO$LxtXasiY2M!C|Ic>M#m9?X-*b`$%{VCfk8bcRU1pC#v
z!b6@9=rt)3`Em}%4x<2@9a76i@?6{DY#&Q#I1*dXtIcD{79jk%x%wQ9mnlZpj_D;K
zRcLUIYdKb88afeUHsJAhQgd6_MPz=noe~la9*=Z7-dD<b`b;b`X+GfC@i(&VukD<6
zdOZdzKQ_*5(IImqS0voXN~`4e+1Ko%#zqn8Nk2Z-*zRy?m-N`!hmYmu7>$bpAzXSz
zZEHAqyy6~1H1^M5!MZi`0_yUs@zmFw9B#i7mGV%DO58CAzQd$3`)q@K9qn=1b$@_w
z)E=cD*Wl|hLj)jB3>Ym8@P>wSZ<23?3(2Q??K@2vXG<!++fB8`Ptd;CQvvVfA7YX6
zja_ytmNw@Rj&@r}kfVV`_%U8~;akyZcScr$t?Ff}@0NCvGag)1IRt(Z0IE50wZEtH
z#>HvRP@cE@YD1(CGqt~ni@l4oWTZ6NNv`<+FI#T|n`OD>{jL~82yu)d9%78ga2Q6B
zm(4cD7{hir^n+5MlulcL&1SQC+UMy9efqKc1+7RJhV5Y-hRx=1co~M*L0(1y0V@Jl
zEEO#_#2ACd7>+T-7-Nhv#u#IaA?EznwVwOmJ#UUt)BCxvdtK{V*T-7d`n>JMj4`vB
zoL&3LMzxU_6GP`TtU$Xu?ZxbTCE5nHNthd_znI<NNsW+h7Bk{2F2a5S-SJ{#J&363
zw7gRwJ@);sc$zqlpiByt7Zay1>%!gm?iZ5?(34{EXT`&?PomxPVv+$?;}3A}RNtPG
z1O9oReu?a|IFG%U2^W!!)@H{KFx*H#Vsl>1B$*UZ>8o+BKH`-1RNN1!4Kh61;?A7S
zdoj~3yMK%;Vh{>E-e!J`7^<Oj7_<d1W`asHSqZn;!WiYmNsf+(V=SG7)jSe^)-PRZ
zkH()%>+9{Y_!wKkKcR~mwa4S*z5>n@i-aUNwQ%}8A&g2<OYWoU;RK6=%pJ)s%$5j0
zwDefBU(7@f)?yyf&aK#$nx-ewBvxq%H90&jtXgmTkPKF6@wjlz%8DqK8cjucp??dt
z?W{JCZ-F;iod~KGTmNDv=0rf1Peh||U4@0p)~E?GQ>s0LubJ>qLRqTyW$T4;<<jye
ziR}`cwgK*P8$~+GPf-HPHfcmfY8y1Ym;}TM+|B~!Ev~p~nHnd>f_yrd?(^0*g_B>M
zkhco9S=|v4J#58Tzm#p`OzhIgox3>mv>uU)c$ef9+ZzB<JP(_>KhPlDddLO?7!hVT
zt8CSmo;J%<91aV7O1aa_3~zgJGV^+hz_C#T(G(G9+qGnHS4tIi3J5El^E*T#?9@IJ
zJ4I2lexd=~^<w5+n$@)9L?$uTGooYz-ov{^h(@Y;*t4;+bHst~iFPTDk~}Q&<bIIQ
zV_yuh&DX2#k5`;Iu01&ruh?Am?RGGrBjl#D<&X|2_s)zT3GfQqp|>d75e=fp8J0a4
zqofEjQiK;kIvSlJR(pzvV;QMEfzol2dNoVWb$l=UuL~mL5KaUPPr}PsP6du`;cCa5
z;&fnxk$Kq83L@;)-BoNPK4o+@%WLgie2Sk~(oUY&LUXMK8h-ZaMGdGtH$-ovCHyHb
zM7_#O@zzfz-;_3{Ne#aT8;>8b5S0Rtu4ukE%scIEjj)q+X&y(N&XcP`JgqSo3!-0q
zdA)kTHTBbi;g|#HT`kI%0*T#y{k#`@tZt>fFKG4M;o5jTpdktp-DMxfQ^`jCky>{a
z&<oMq$HKX`y?|jn));C=`=oS`l3V6eeO89FNbr)~46x9;jv%nl^nOiQeMNOyy?w6t
zd3=$+(2#O1Fvz@psUfkATL@;reJ#G|-r}}1`&tb=+XolCf_)R~iSrMaiEo8AXV8#+
zH+761P5hbrqJ6Jkm-V_IsQsY%u4c}V7PKA6-2Ib!bcGeyM2d$71)2(5?3Vg#qTHta
zqT#JYaIq5^lPW{A+ika<$T<HzA{K8yk$AWV_WFH&nqCs~Wz$bcu<+A`Js`!zsLfDQ
z4EJ_Nz=7mj+)Ln2J(5_PFfg@k;O02<L?%}>cKf;6Yj+FBC@C;Y&t=>o2JId_ukf-n
zcCVhdv==%C>^?#CT#6p)dcRtb<btDlc06Z6KQ<?xk+~3<5O?Pa;f6wylr2J-s*5&H
zFDna$9-Dt62{UAA=&=PSk}wl!E)*JWCBjs051&W^h_f^{cJgp|Bzks3EFwL*&jlQ-
zcbh$SA~C3~&GvZU;F<;U&Mwk`5c$Z8^2F3<nBZ-3G^XyXd;Tmrk?~s(&T*-)&&mU9
z@?BH4<U}SpwHDL)TBRSzT<gijszrcK&iWPVd__vBT%&=u;2ZD9Ky9?g2#8kM>hzu{
znfAgK^zMRylQV*Gqu?(ispcBBPi{J#e(N+mN=4#nuE%<J7~3Fp>jsj?^v0M5mfwbl
zY*Rc8b-HUpQ+!zpOR~JaZ09ltKW`DK)*`O-Co#7zNQJU)$TkbL<%m!@NL>u@p8c=e
zx?;N4G+57xOwe|RKa7NEsE~tuW1vZheBJuheIBJS!qQ*>^UPcY+bSZF?Tk>q;rMpq
z3^Rh>T{I{sfZDbTv8Bkzt=uDbr~y2H;^1nqwoqibVH;QPl}roBJ#!)n+^wwC-6t|x
z(Q}n~a%_Z?XZ2%>KMcLb9${IQ?`E&;4J^F?3H01|B2#20nXQyj0@2Z3j1Sp<eax=J
zW#B+;5GIbZ)@xk%*+I?arfqgeD0|*(&dtLo{{JkmBX&fi#X{BZ{+`bTUdjDAZ%6f>
z<Fg5CupKMC?_#X2c3kVg^WwS?fWki3!Nf5*zzHGlvNj3;Ax>rr77q8J)1p@SRA%<G
z8bAdqWcK?=>Gbf}@^gWdryAmXy@-YBe9Xw-qd2>$9$_-*fJXIPhs!lZ%1&gr6j!V`
zKLv&dIF1F>l3c=N&Bmk851vI*xDq|UknL?P7!=p&1Wrtr;=r*GMj8xP;-D#;RKr{+
zp%vHkxXRb?yCN+@!T_;I_FnY21vYN)$KS|tXa-741)Cr_kv<e@RwYxfeH5E%6}80U
zBQJMmH#F{*xHgj*J>H@3#T8CnH!%kVPvN3@^hqH&l*8n+=oWLr08;kn@x+oVW3_z|
z<X8n?`cn9$rF+i4`l~ta>Y6wSsIT=CORNW}sc7E_hXxcCZeIB|Rxru#nD6h@T^$ZH
z-$!>HNHaeOR7EL=y#1&%ByAX&6D;?j{iGhSF5@~$7<kZx<c;Zcw_DK(p8u};_$A)r
zXBRc&?PSJq8;JjOr_S3>W?JK6cYD75WF`}k9mJ+Zb2J5b8>S$P>3UzpMSaG}#Pi6t
z$_(5QkQ&)$UfAi*lZiLNq9{2U?MN4cwq>S3R94p8-SOmp^mts&I+?Mn28yo}Z;$rn
z9^uUX;$)%9iCVFZl7l}f(bdYgxw|aiuU~M)FM3kF*(Vc69qhL`f!q)yv$>)GMX~et
zfF6m=Z?}0TGogu#GuPtz@!Zgr>p?C?nXl1wPW-1W)W<^wLT(<8#=fh4A=x9*E#cS$
z_Go<Q=L1{#v6GXf92aCqP0rGe@Z-@Yat7F<pd!ol!Qa>u0Sc`OSAX?7>>|q&!CD$c
zxr!cUQJ&3#d*)TfgGUn&S(QHTBf`w812Q)coY_xi;(^~|S;I9Nt}AeAMI3hw7`D2T
z8SA)yDDgR0WYudP@b={pX}3moSmP4InwXY8uK6XJ6uATU+4_@{*_C#`i)WBAS!K~2
zZ-^DZ6)qA(xPEJ6fcf15+axq93zVjgM14&$tdxj4e=>2qZuVYF{J9)Su}y>JD&?1w
z%^KeC7A{=)dn=yI7<)PvToU%G2wj?&Wvi`6W1IL-uCl!XF_3gj{pthX4)g#vDC)TB
zVd7P;`+~5_&*EW0!2KWAy-g6%=6k&0b~Q)|2y91u4`X*<A={}A=rNeTJ}6J+Jf9J{
z5`BHCj@s^+ii#YbBY4ThCzxPJy~aY`6T~F3c(3iv7FHiutbIYwhu~WK1A@mtaS|Pf
z+1d0PJE-of@IE6hE#55pTRR+3J3Y(VkpR@pPI*ovNLouY&`0BCHE@re%oKYCVuQKB
zcH40gu)Z5>$P0SRdu^N((Mc+oc`9CU;X$R7v(wR083uk2z+!2hjm99_Y9leuR?eWJ
zozpBNFHKbUF<N9&51qNFK3hCubyO5~*~U{f!vZg9Sk0s-^-`;28cVv%d)mQq5sPqP
z>@8QK8_s|Ru4Qj)<u%mWRrTrlYi&|rwlLN%yLK|u<+#;)>Yo2@46&~pE#G@W$psWR
zroDeM6D{PF=X`)DarH(>I)z6A6B1CdB0nlUcp(D&xU_WKA<^ZU8!>`*qL+F5iH4k{
zKDXk3D#-ZBcGyil;${yA7vUKse)#rz;Li*W*%x|A5h_<`eW{OEdrFP3N_~5(CNA%y
z5K&Z6^BXm$Pz2s{|7EwkGW~;p!vW?yVa)vyjS9l?_v#jVg>(8J^h0TwKWd_eo9w4}
z>z9MAb*K~V=NJMu%c9-Vr(1f53-A29{UWT;Pm;;yRK_8Rj?Uu{gdB9+sf>qH5&oa-
z_EU+6b6s@9^0ZTl4<}Er>1ybjahq`}+dIXa^GIjnrxHgWO2Xcqr_wRGymG1Cbt+>7
zJwvTNaWhY4l0+S|<iQAc3$v7WvesM83JB$hLiYq7tSchHPbH?xQQ~Uk5xXy%twz6b
zzi=YW82h511Z)C~@rJvyn^fnh`$%%xY;$AO?2LNy<UgQq64Oa!nx__xOxk>n2xkpk
z=50Z|b{jpf7X~5LRxhnzY7dKU*aD08NT5$wD+WACz@vfwFs9naPGuaB%Fg)jF~<}f
zjrR?x{TB&uPe|fR^ofi_d0QNRR;{kLC4mf^!gWgJ!nMwwVksgXuMw!)@m8%8%`&5n
zRqH1oKhZ5%;x&jBeh;sS0lMaa%A*<)H!rwi-KmVzr3H-wH&MMl5X?0XSY!N2_>Zko
zU!?qVn-Z7y0N%&#p~u#1%<g93PQ4q13V!x~vW>Ac@MjvfO@RrqZw2>SuqFZT;YeoR
zl7TI#;a+~&s9J`sMQvP{k~{+10=*_oYnx9csc-IYtQc4^Z{-At2MO^xcZZ%Ht(QtP
zxP6Hc_Xb*I<3l;Gd3D_Yrf}c)gZhYm5hXpkZ>t7dP5vy`<qwOtSNC|d#WoGTbBN8q
zJ@7!3@d7eC0>$1OE8TXgWuc4f&#q_*`aSqIJfk&1=L2XXcB^MUs{2VH_^jxHf!vE0
zy=i;&m3zElB;PCSD3Jzn_o)F1lv>-GY=3}FX^4_bj2#FZb8a8BgX$IiI;IiQzX&4B
z0hV<*nwR^=vLiv$D*mzOVvZhM^q?J$Z#;7^cZWVZCUPV66x<DyMR{CZdHz^mM=xjs
zRuzg(t|=+liBp**jRuyG;BY)RB~sQ5!8A?_N0POmFgY8jVC%OL^+MEQs&#!%boV4}
z=cDObu5cFxWhs6dWMG@>+DnR-O3GhS*Ao{hN--t^bAY=u;CO&Xx^?z=C8$NujOFdX
zg)=faf^h-7s;^g5yAy&=#_C?X+^z*2zY(ww-xbmoFvj-+>hi|rt849jq1`fZk#u9%
zHDa_&&_8D%3gvn{#_S_K4#FkuV}Tw<EV?0RtH_bwX`hsa!&-qCeH!!aWlij6P|lC%
z&!SOBN-%Gq3jz0NnRuSNxVm7^+n4I1p??S_^;JBFDhUDn>u8ON!c`03M3>#g9`x_R
ze7!59ztf`^A+zrTp-P+&e~3Q^NXYSHeBYHL+x|~F2dV#)5<$DnZ$AsYDz1RHv{*y3
z8uDZP7flp?KsY0ry_7NO@@_6Sx4ks!&dG#yw%P43C8o<A9!=adwX3EQtWAF@W4z?I
z^)%KqUP`<Z2@wqsen@x3$DVo`ZF_v|X?`7vyW$<f94vEYJaBxVi@E!yBqRJP{bRFU
z%9uG-k*UbrYxlgAF@R)4&lteHFJ%$~W=JF~`1*Z;4`A`df>+(ICS;Q8Y4XeM?3a=p
z@z6tyY)<q`YiF`^O9MxI=(Pu?23|mEH!nV#M>NfRb&*oi*#e<UTrWf+1}&%wtLwIh
z)tcmF5dM)ssIJ$mVY((3ANQ9sE{%}nH8H*J!DE4p%Pp|%zwUx_ohb6?;)6f23!jL;
z$#B8tTNC65EPBZ$F~sB!blJ)^V9zg7si!WqZdJk*$6!cTEz)^3VzLpe2t3kkb1lO@
zQKOGmQPz}j@*rDNcgiqkbwZqzvzxkFR$l_ajh2JNJqa5F3o3HEyTKaK9S+MK{H4S$
zDMhnhFUYK%w++IM;%I}g&0gE6;pCc{Y?COlEcjPbye6`{y?My;fq0jPs=8-XOFX}`
z3)?*TbF59DR~KAeXPec3J~vE~7d>`*fMAy%Nq$KnxxUWhb;lg{mexK(#E|vBlraUY
zusvDs{y~wlF6UcxYYhF@58AK>wkkJZ+u|veiEg*;0k{EnMq2cE)snmdY^R9tX-kSY
z+wF?RZebX*XN0l~t+rbP<#^wjoPM6wPvt1plIDF+>;o?ew^!f0(#!S<$!avM`_(ts
zVwCiGJD@>xsv<PAlezUA42%Ynb^DMgfCArnwfgU+7;O@AjtK6mi7^kbvFFrp7s1O%
z1)@CNAdU$H^URsVW;z~p?dr7`qVpC4H*<C(b`kOwNn!LA`x%!a4wlm~@GkU_XEktM
zG~0+!s$5!C_GC??ozsTyK_kwx?;+ImyqeXZ#CHF>i!WuDC$!G3k(zEakXyROE=7mb
zgRwDnm>-4#jqA}B{B|Wq!6CN1+un|^D!3+G4Z<YHi&2{tVODu^U%NI%wu$5RuAYV|
z@bjKJXef3M+WYZx03#Ia;kuT#Cy3&r`k_XfJh~%q=SQMoa;J6H@NuBFytdMA1dPJg
ztZv>XYQEB|b23l%X&{luZ`Tt>-W2{Q_n42*V!F6uP~Su-^|_{-0w)@}r>gu?2$r0h
zT%+`rzNz4xW*vPk6fi-c*F`S-CO#yI-qK=HID1yRZ#8T9K=pUKMb*9&a+`Cc-L&uH
zIi)>HiP|*xNpXoM)sOmuD%u1#{UnkX-JRiQAwygRk*|cw-BJtRjP{H0XHB~+r=8Ap
z<<19gJDoA=P8V0HO?EnCs-TTX-fh$319qX;1(>c6>TrAFESaGnidYmm!tOXd=`sQa
zm=3%1^wgD^65MuIz$$blf4P~bCo@$kFGDTLk>c4>?>?P`3l{6rDw}mW6GD)cI_;hS
z>fCssFqpG@PiO2M=D`@)yZ6QC?q}U)_iMz<Q6J3K09RpWYs39u&gsPHJKEc9uDX5a
zFZRIcOpD~kKO$@7Cod4F_IxGT{P?b$IU!|Rz+1FogDq4`tO6;i^TShP=_c>!BWjS+
z3V`~f@%2Mb&$GwWmK5(j_IQlLU0~^QkzV3&(PK~O5erJ6E!LRbS8s`iHShnla!ptd
z6>eZNmHPBaF9Kh6I<eBSl~%16_}6sU3PD-E%4#$TtP`AHYvVbDe)dhBdg*i%cfsmU
zXQvq20IKS_R<lu9*1}oV#1l2SN4Z|FQ|BmodTqVnk>k`2;~RobzyIBXwo$lHf02`J
zQvgcyb=jKqWdn{*Y>B+afg?`KX$e)HZEFNu<hZrP5_V;uZ5G}n7U(MwC7idSI)pnB
z*w7Wvh-qHB5$||)>8cdG`g%3cGGc=IOSDRLv$#GF3S1eKLYJ;Z;qHbVEQ(<P%nvbm
z9Ov{2Ml94o+XKtL+C$^ELm#a|tHYgsr?6<n_ik4Z7^f$k9`g2#UN$sT+wMRiv3T|)
zAw18DP$<JZD$;d$drBj5qk@Zhdxbd`3_LU1ZJ#!0BTjQYeq8R4nIlh@9f<D{g*=@&
zrOPzw2xColC}xU`oZO!d$7d7&jGA4zHX@Kdr*X0sT^=~9ACkCJuw!bUXCECGN`BOU
z`2~9+VE+NJ{e<S5g-vQ2SE!%Tiuu*w+v%y#T>x@@_gQtABbgk>_DFoNy1tU47U%Rq
z1qv4rFS7HRK93(p^BxNhhYs%JK?9Q9q_uQO__(GOB@g(F={J^WDontIJaC8vq~7R?
zkbdwfdpjuL0a@&=t0FUjfSAXl_moWUTKrAPUJT;_60I~rw0_HbPiXgf4G()i;BZIb
z+=Jx{Y{wRM&WCZ34C6Lm8WO{IHR#7`N3>1OZisqrOmg*l;6P#79{W^nF_JXfO)anr
zl#IUjJ`+v|Ih+fht3^Y<7ttWKn_A$@!$SHpo|(I}0qrZT4)g~0b!p9cG?9H1nBaO0
zL)SF+WoFdxqHQ-i(=GPBFdfca{a5=z+ylCzCQGhoZt?q3aD0Ig3zxT_0;4h<KS}s8
zWIvzI?y-0>gznfQEN_WW$wEl7bae^WBg5O7j1l@V%;Fxm+s<VCy&o+<)sk*MlbBr@
zs_$uMGD)BnXYlD~5|c#+ckc<C5ie0I_EU4^j(BoKCDy+?1(Z4_&Wr9kGwCe9Vs&k!
z&5WjGdBWkq?mm<FIDTs!QnR897`MlA**)>q!q0acuzSyB$8aN4<ho2&@O|p%`T<|L
zc7H$vz+k7%K9kwFS&HbC<_N^~BhEoM%U`#-XEJdHN4Rd+^gIx~KKRhD7TY{gV7R;6
z=Es|M3L4o0p^Db7+ZGDUKMYKqY$_1t7O1fOSSF9emlMa~l8?sUWOO6AE&lc(Nqap0
z#t^`*T+p!%blDSn<Sx=~i#2S|>cD<k5+kK2Pd15PeDv+vH}QuLxGOxxV^siRPm@xW
z1aG2HW%80+p?;`1++{U-jWZWY)!M1$bCg^6Ybe6%L{qdn7{Sw_Umx9j7XE@YPJKkN
zG!9dWn+Sh$fw&X-IzhpB>}qNcPG27j5xLkl#1QC$81z62Z>b1a+Z4^Z^Q2AJY^`U3
z<^>cV6IT;E<8BE^Tw`<Ara7y{+~8N&%_Z2WQXZo#3Sg%PxRt2NLDy$J@iB^rJs{$s
zdrKRq2>*`nNCMtH@v;rZQ_A;EoiYtR#ra&Y;VC>Xx-P@T8`~CmGQX(O^~Fl0NoH|}
z=B13?Mus;%*CgY@E<t3gxLwn2x6cIjY>mq0w!1XYQqY4vtBLRHM#9^paaY#V*j~LP
zPkN8-(}d#I9O;wx>!)RyP5V4L?|?p9=}`*@1L7K&^tRfe7+s+Wzxu;xlGv9@qY~GZ
z*y)jM0r{czoRIOPi${aX6;IaNu`}5r$Z6)cgU}ovpZXbj+)vfC=w6s2(f}%EJE7Tk
z4Ww_zQ)edKS%o+x&lSQ6>WQH2baa^Oq%^oNS;KX8B#=xszwMkzo^{A1o>v#HGxP)N
zVgRE`vyBEmAM7eg7?-BdejdTpgHgtDiyb?YgzefY8y7{%e%xtSgiPDSNqakx=`A!h
zk?t*e=|(#>8NZ{i<UGA52u(vAIPdE5`u_ym_w=}79)1q*$Cs>^O^fV$Jh%?EU>^o}
zGNjl?`evCcVLsL)S&RPBZiwWl%G{@23z?f$Z~IpJR0L|nmCNVgX29-kZ@15)&weDB
zX8T+#j1z_sChn<UhyZKwDf%*&6cYMn`zn?cs5?JGVFlo?)p{uh<~L_DuK~|%g&&<O
z@0g!&HR!Uf_MOmohDYq1uw_GWm+!M5G~QG?{>OOa65yN8e$sGj3$8M<p93IBap)l}
z-3pwSFRLaMYy3^obmG&`CdNfI>*gVM_rL9I#)oSP-FExg#7nt)58AY7$w*rf(Mcr}
z&*UL-3<@^mY~tBv_%gWP-W_K%X3s+LZ2ETwtYxnLx$Ep?$vYCU3USTPjj)f-Jex#<
z<^qLB&t?)tzMItVZYrA<FtD(alr{PTpOw9&%f46W<?<x8xBJd!%(f3FR`)x*|7>>Y
z!|lO+yFq$(0Gt>DsW}43JQi(kPy)Y3FMjxdkm3G)4o|zCHt%ev7Fdf@TRaS~TqqMR
zTq?OCb=tysiT5=J-oqjc*DDXQedKI93mC}KzX}+9RQNYz;&D^mV?u!-uOX~k=-Abf
zyKIt7X6iZRZ+t?14R_@>ljeJIKt;b5l`TeWiJG9q^0_IGakIL)j##DE1A)+1bvBcX
zJQ-_mM+&&97G9_&JfifUY=w|-7+`+kur(sqroXe=m}J*~*QsT0z#X;oR<9nnK%B<-
zZte2Q#%f!m;dwHaoUPMyOK|}G`g)B86Sg~nZxF0T&ZLdeygp0pxG5$d^zzoE#j(~)
zMy4did4a1(eb2^jIh%0-zZ@iOu<zuyv+3OY_21d%z!veAffNOjiw3fX8{WGDA-@_U
z5cUYY%}AuZ8qJfv?$+NQz$hGyUTILcu!h>%9a}{>__F6hvSF=XY(bs2ExMQcysN{u
z>m$O?J<Yc5P^T1&fdlSTKa^i^E9fpkZ6lMaJ)?Q-&)aT|aEO&tNVV*rRr~Av2>aWk
zM?zG)Y;S;XMs$o*a9_Od9wdBp;*9N&DItSl&JM)iYYXV@4~kODnTJC{dTDYcJFH=K
zQmDYyUPsjJ>WZ2Qdrl2<%tb4!{HVs+hq8;UbS&0n3wD6x(HWLd_t$<wU9VfW&Q53v
zxJZ@kRd0UqFLtVgH|*U`3p5_aNbGi2WP$_0d)&l25+8i9pWSj!8>qa1)Wf~d&I<<Y
zz?~v4YCy>zII72Pu1Iz%R$zaVjRk3{IB{{oL+>8*#d>9-UWuV_d=RT>Z;Ogaw7085
z16?EvQ+CB<5akbqy<bz0ja~};-GJ+^G1$iAfgOje^1i_KCHv0x=#-2}?%nvIP7pR(
zgr;&O5A;`c+s9g74PAC4AaQ($Lgo|Agk$ij`oY5zhI2DMCBp+F{VWhqg~at$d!AaO
z@m0z^$AY&n1*JX}%KS>t4TS=cK3|I{aJ8;JtVZ^YhEUf#al9|-i|pIz1;d5Ay?!Ss
z<(|afzK_|)+|_13=w1KBMQnaQ#+PmbU{Ze4O7&!+UWDxDAP)HlZ)pl}>Tb4Q^tg68
zhoilmDW)kQgLVAnjC;EAFs<a<Urwhtc2WeWl6AnQy`0@#VIZ7P(_hZGWTTRp%+PP;
zDbbIc2kv+|<L?as<m9;X<&5DrK&pxVAZ-2RBs~z?i?f@W`;tMeaQDke(12gML&U5=
zo|{pUjqV9po(Y}o&0dn@-aylfyBFLy{=S#vF7L9g-5&#kgN8)Rw!CDsql2~Xg9CF9
zTxy(a&W*pJ%aA<~t@yMOA!44eME${$ZS!AF;>etZb8Ug>OYQb{TPV7bT!7Q}VbRE?
zp<Z9w_J~N5XTp$_?9qVtAn+cG5x6;>@}e36VHGSi67ExEG#dYY?m~MadUFdWL;SNX
zR{O}@lo;`nfQXk6e&^*e$aeCASY`aZ#kXgbhRFHso`cmIh)2TU<Q7m!9*J6}CQv9}
z-e|RI*EQghVx0hCE%xlQ_0hJXtjrn%#>5HNTdffl(0PNyu&s;N5S_N`V+P#G8A5DO
zFR(zQ%#G>=X~_@5P0>pqDLtBEMu<5`6(aKLr`!wZV7G74<d*d|Tbn?yB?_X;Hou&4
z^U~H%`2?*PAHkyC<knT92FxA{`OAsXm;cs!HG+oa)-N2d{|A@D9X6<bpMGeeZJpYQ
zB~jOgg-U&)i!0SOEuOYi!rIyScCEKM&#$>7<`CsUpY7D=m4!Ym_q$3c@A%*^C{&+O
ze@PzmIo>VOppx}s?)I!u>?cp$p3;1s@(7>3@roEU_t4v?PxD+7{F1ysW}vmuW(T6F
zJNoeAV4w&QF4&=H)J)QV!y->32M>NJM}%NKvX_U)*>h@xVcYlXQQ`Mn1n6Tyi!DjY
zAhk@O>sgxf_JX#8Uj)+){DPf$d9oA?f|jb#gW28FE}jaY9MT@<eOgrf2j?($R?lSn
zWwu9RI1|rr&e^%rmqZY;`_79dcu*5uowJM4mI9FdHX46(&gJZqhSBikDjU<2J4=s8
zyLT?-hwVy`YvLr<@Br?n09K$on_SgOsNh7jNj(mBbFsaqwNo-p;~4U;XyR8`*GPIO
zc~A|X7w+xvYYrl`23z;_sW~Vn2dobTg|XnJ>msau6jZ{~j$(pP@nd0=EL1n9;HI*=
z+><_0i-A^7+)t-kyuv|!9s%fPPy&AfvaQG88@B!7Df?U)QFFm{yI+(xKNs3B1(kCg
zhqfRsN*@n2_*yi<*o~ut2SI(K(NLhAG%Illj~%_4Wb0G2kIMk{L<1mR{vJm3g8;fW
zFK&m3RzC(kIT<M7_mfb~^$k&A*M8Oy%%j(GxTQsdsuUl=Ci_K0sB)EwjU<+~xyjkd
zZ6k^C)GuFVw~u5Tx{PaJpG}KbE{#qf$#_28s?%nSB%WF-$a06i!yyFX-8qujF{<Y#
zyDM690xzq3(q_h+;r4fS+T8(t-u#Db)=0+UBiE$eQ$nijw{~wJ+XTZVolHQ3P<{!$
zKU#J74%lo#z)947jOM7{Rn5z6uAXYKt~?M}xNY0!X<)82vH61IS#J;r7U;`9Mzb)U
zCeAS1{q}HtOC682U>^}E>d#@)j>m>lc*<iUQu)$lB$If2B-7TooPsDGT1{INgKDQX
z_Y=_@V|5MHVgYFGMvk^6LbW03UaUMQ1*uW%RqAcgtH_f1Ek4Itt1X#2s)GnvEONG@
zG-1B&Y61$k2jc%~H6&D7eO4zF8vJr&^&^=GN4Ao7qG{N5jrxt<;DMF=wnoUf?w6B)
zof?t7p_J-k{Yc_@SoJsPEuy<y1VMk}h$56(<D6Cplb7?GNk4kfHjQLLC~8J(aC4Y7
ziF)M7vAjmIq8p=Ki$+pKMA5Ld1=TCdD{OOo$9eG=PMu<jG;Xr%(lAIAit*NZMDd<x
z>lH|JVeHy|^|;<IT7w!?%FOGsn{AEnUBSVvI{Njp8Q{B3&?-}Ivh5?8(A!nOpBfY3
z4#Ds39wv57k5%rmw=19}3nocm^wRyCY`1_FwxV5tyw9q;rXkOExhEq6$M2}^)xa@%
zI^z4HIl3K07ewIwrJaa}@&S#LTMP;<S7#gyB2saz%MPh0D(*tV!y}n+M7=VvI}_@;
zBY}X6%cz5%3;4x}=k2Iy{@@?%m<B@KB`>FJ$4iLgxx!KSf)FpmZp5ZM5txuXjZ84Y
z)XSk#jE9AJS^!<k>aMtFgLKFbtmYvb8Oan=q_XokDHL3GF3?+r!}j@72k2+*q824(
z_&g1DvyDbep6ycndG$|u8;hmi)oJ6JsAZm4;)?Khv5+JAZGrStgKJxlT@^vfw%DY|
z*2f61=~df+y(`Q)0~jw3YdEv1zg2ib+53V0yoY|X$*u<|1YEx^emIikoEr9hlA91m
z;3OaEb5s<6wvY9=ybqb=MhxqB8|{-AYZry0K2@8(AtvNzJYrZ+5jLM`z_r0QKUecQ
zKb^kN>gm9?)n{L7P>8d>dc1PPSNevv(`jF8PD$0-=Rt4Zh)Njv?OXL(9ap&T0^2&=
zMZQ;muwx|39|UWlYk;BzMf<TdG!z9tX>yhonkJ5OZrRU5rO6W~+|naKGyV39pb>t7
z`W!9!D;b|{^ITxJy^?WcZVpItx4)A3uBW3U8Thn#i_Z!1*F>XEe<gF>;@q67FWP4_
z1ho}kF}vfHYz?!dJ^i-3J>2<9k^?-h=PotJp303nx{0(>94&XhlEjClTT!mrtXDGD
zy*A|OdtOP$__8&2Z+w>=l03NhzW6R=;QKYQR<aMEAAcn~z)QK(s3sa1l{GZj+*i`-
zbE(_CS6p1hIQc+O!=*#^`aD70jP9Vz<_lslA~x@W_`nYp=5nD>N0IM0*@pu`_yM0k
zB5=gsBk4UFMA=ek>T`wfWBS;G{aO2ut2-*4d&)nY)n3WY?9C`2CY~pYYYb`d>c89K
z__G6dxFtfXfg`9~9dGI0jH6zqdP3%LyL*+UnR|!}@1*AUbJ(f_|95r{yfbDiO2kR=
zC{`0dE6Gk+>?ZuBHhOO-3vr!h8)*|4bdAF;jBb&(MqyOj-hp*<jqvr-OzWmbiCp5x
z{d$4OyXm$MCD8_TQnP&JQroD<D(AkN)Ir`O6|6}hBSCzV<pW>Ju3AfpaNTfib(AB5
zgpztVqQ5ffP8sC0>9u0CT#Z?^EB+in+uKtjj5}Zc*?QGAe313rud!|V)vs(&Z&3;W
zacjJFhRxHkW~RD;&obFt1MjZF7B`2v259>$nS_LC3pX&gC+*N@D{+|S3c6FnD4D3p
z=B74D)@`?E^n;%S@TJ``*(vzfp4HnLtXG`;d*UrCebrLit1*?61#@4JqNdPl`!x+D
z!bA_Q(Jt$Cm%)S4vK!<7q4+Zok2o9~eBs~Nktw`=mR$w&T(n!o!FV*Fv4X<e>{y`G
z0yvVlP4&!~Z7;-U&HV#*A|{Z+-OYAN2Nls?nK|=xG^qD_dS^?0yqd(oJvO3|R^hdK
zPE=uwjC^pw&PVg4aq-E&7@)TJ)Q-ldt}OQJv_8dF<nq{<5MJSYcwD4HAYzkVQO8AM
zv)|VGMS~ZaTVsH&&}C9Dxh%qp2khD_nFz!-9>E3WUpVAO?A?Hw!e0Jl@5NKX{erz8
z@E-cr2D>gI@h%*$#_hv+|Mc8n*+;@0{z%0#JrZcpZ#RMjDE|ou6CT~ou6iuFJANu!
zpcLpuFK{!yp&%HJXc`R*9MO;zuJP8Di}r=yt|B1ZW5U0TVXzAPCi<1gNOmL+=&$3P
zXW34fuwmCF;_<<^+D~QhOZYuI;k#H^1g!dPT5Hmcr>G9b?FWsxw?F}h9|J$nJ#Ie<
zKU|Vhm5ZM>kxh8N-6|~@KM6Ty?U&NXv3#4IOKfoBB>vxaTRgDsq3P}CvQ3ZWYcoc^
zE%=k`x8&09E+aPmT;k^4100gLl%LBuDKaJ5MeUArnJiGbw4uW8JeP5k{%$f?BDCLi
zE|U&cz{iJd=DBqCaXE%b$GfD*{bw+rbuMvGulk7`RT8u99zk$7%Yxk-!*It1F5Bla
zmK+|6h+p>y#sri#(f@2Uxc(Oi<Y5eR&Sj#7rvT?iw8%C$@S+}X!5+{k$yw$}app}C
zhGd%&3pc+s5QJu1a4yLc4Qp*-{MtQ8ywSq~)7m?6jCp@#s^fjgqJQFAa?T!|`tf;q
zWb%8p$6}16ibkM%TtmQCMF=$J_(h^~HK`s}v7MiYj&qz1wm2v-F_sD}EIF5irr-Vk
zAuHEsc;~0gUX{_5x{hwfs0vb}U2;8St5pXYZp}m$9)oPe!~C4i0$0Hk?gkl0tTs^i
z;{eO6F8;1_m;ZXL5w2Pt)~NY(ZrE#UqAB5Q*s$a8K~H|ZUZ2&tf@wo(^(M!RjWHo~
zJ(_0FkD^Uegp6r*Wm;1#DK=|Qch5i=REYSFQS<9>IRIMX30zyyAhwkjVKXNm?6BW8
zUO1@5;Q!l$Rt!AmJ+$ff^Q=oaUjKJiR!<BK*CL7r^4vRx5?9n@4jZ-pn2(nJ9vjrm
zy0#WF{>iqAI;|;GXjspKsEl&9ExPaaO=8<MjvfsDI|A@KFB9pluX^AF-L^{|!8lMu
zGuO`qsoRH<Wp-<&uHy9M-uG+(c8@qL^Lw;x>UzD*_OR_0d}^68Ir}tF*As<5qU{gj
zG<kU#3OgTA8$a=qm^^9+g_yfrphC*qAx)VJDDIT)(|<V7B}g16^T3YlNp~x5Y9d0B
zwSQCqohK!Y@iF~W*4vs~5A1kQ5l0F1EHA`hoc}s+vlD@LR$S#&O#PD;c3QJt@6<o5
zj{B)%KBAZ9uJ=Bt$2#l;cK%#;?O~h*81;E(6ZfmX7?`}Xt9f|HMzyg@Sp{8Z$1OZ2
zwxdw6F%cd^8~bcrQ`6)2Wyp6$XVt_%=ToX8ZsY-=xo60(1|1Rf`)x9S!E)O=+!}VR
zL}OeR?A^etJlRLyQ!n^Ip<%|qJjRak)Z+%O$9odWcucI9_4qJA;md?UTR2l-2qn_T
zT5Z+s{T={#BY?pNh@G`h^x97nM)c{_!trRIH(aG*H#MowhxqxNeHNo^;v)2U{JC(>
zL-s`s#HGv{`%<$)u04{k>fP`DU|$QO+nU)YBw!0L1*-FXD-s~sk}{T%F%h~i#a8;J
z`aVX|S7fh}wng8pbn7vh6Mu{j-LhM-pG22B_sX-MOJs0^%&6UpWw*N8ei1?ZsAYlL
ztN(?8c{vyAkiME2b&?|gWVgSX_;PPyizg$pX|HB{c`4}UxXw&}HSy(0ZpKh&yqY8h
z527JZ?A46Trx<el%HwS=#me2Kw}S&k7fWZpn#mI6KcuXX$4cBC=#;Vw%u@fI9@I~N
z_q;mkcc3!<v3p<5m^kSTqfW$qfgZC~;nk$<egTMeP8OG`PLbML)1_|CtBDWSmfPGw
z7>zvg-ves8k`fuD;sob;uO_be5Qq5uK&8dqRThY}6>L(E|L1sI7!8Q`!LqHv6LH<m
zls+QlaY*&km1-(N-m^7v4_6bc+v#|;$76mNAK9r2wkWXq^+OBn33b3>OW2$()(<5=
z`6U_-A{9mAQOaAn@J7q+N+v&BDq{}lWDCjEs^Y2k$yMcqzElSSL<w+fLAxF(KDoTn
zwNoinye8UvB|7L;8&9570*!N>AlDFuUSDd7DU}Omqwu}{qF?-NP0UK;fUVQ0TML-1
z*NaZWg9C7g4f?WS0OiDxZ7eO4+KQ#NNu$O|pmzvC7(cam4_$~I6YRWaeQy!$8crqL
z(xR#BW^A+O1FL`1k6KaCVLCknN#N;vHHo8;kNi!X2%+rj(XYt?*;fe5xmTzY5aE@+
z`eOm|@usJ2aOy1{@od(u@phoeRe{4o2jy_2)ZHfRQ)Ob?t|soClu}Xccs1iF6|QBR
zO2D-<2(h-%l%{c)K8=KCtl?)$yV=7OY<El^sz9`H_N=Ig%kiM?5#p5vSLWK@K+IJx
zX(0Q8tc5%ll>Gs=8~%ohT`Q!ABmf5mCqI#rHXhP?h3#`7B>Usx0N^nJ1v?^4Jg0;`
zH+2AThx0&yqtSKs(y~fBrs=~jU&<eTTmYz?Qm_|<P#Ns)L}?tTq_ELkuv5{syPR&d
z(*fB<Z96OSwNORLqugyokQ$Nl&qX)IQc>^o@ok4&*DeZ!N{=2GjYs#n>&`dXr4kC~
zKG+Qsf3h)wE#odft`&eel*|4VjlaRejZs&<9WC6%^uZy9a5c`50!PMV^n{radvDRM
ziTYTjJ+R?jbydBp!QK<`DD*^n8z=7j0jyZqY}aE%dCnR8P*{2V1DV-0LPA~P>>?J5
z*!HC7z7Z|luJ2{7KM?>|x)CEDsNtfCv!9<cHwBSU5w7lm>BLFUB1}9`Ne(t&#NY7d
zUi(u0Rz$`b%8bW{h?>N9*w^aT%R%Su8&NZxjc)~l2&i1%6D@rgYl1DjyxzVqb=Nmc
z#ql6c8yn<D0q&eQ1FQKdx=kVUXx0Uxqug#q(<JzEQ@39<D-Hip<4oGlXIvLSxT`=y
z_4A3HZCGfxpHJLsHIB5?&S(3+EYCpz)v&+2m)~@Kj^v9wg3Ztm)uHvc<9rek7|<Gy
z{5#Jl5up+TjWguC;w2U!zkJO+pPhYp<gg&{UgS`{I}qBk)n*A)KTG7E00m%fj)Z-m
zPqM}8O!B`ko;fhmWzp^zVohYVK;D=g4fx&#BLc<*j{u)nZSMIby{sWXsU$T&a6U;7
zMNTc7r;m}?+xl$2pmPT!z|R&$V|Z`Q7M@RHPDS?sLhQo<*U|3HZFXbFBhmEfr#C)j
zj|%RJloH^vK#J;IDW>4@pvCf)w&;8&{&=*9A0e2<&u1bUVuefk?q0lD5G%>~IAlvS
zOv1cyQVZO|h*K#5Tny7Vn9-^jnx~YBu+M5u1Fn^}BAzg&`(CXH1Sw=v8>{E<=RRe1
z8nW}b+{7zZA4A@J^-s;#s7^WO`f+1&1?QST3&w=8eVvBj<pOfHK0tTlleIx~$7Jf_
z??&|%{F6#)fgs$&Qw}r<`7V&p3o-%5P@_e}VTF^;B_4@-^dyD0S>ICQ&x&eW<vPMH
zVS?_>D<}2@Y?yzTTzg|kn^8Gzwf+F*_iygHgF?CquF=9QZ4ClZhZt5JZAly90{pfB
zi^?TQind+fqC-W<q#{amgb>OQcE-o0SG(dBi?4EpJrl3o=>f&_?pRPLh^Q6gdj?g)
zvw=m08yNQl(KpOlV0%S1#BP$d?TcoFiso#8{7v$Qf*puY{IY*A`l91nPQpXNxs4Ou
zm+s;CtlbaRBPFW4?1NS9xq#A+Tg6e)*p+DLsgLOg7EK2?;N#~rsou4T$%pC%{X{M`
zcOvMu6M+pHX%DwOrHL$hea6#5HIHWItX8KVXUXk|`ey=3en+nvnYVMoq^Zz`&BM;?
zhhomNy4l6(2GJu$ZH|`akzIo;6qUwBlMb9fohD=YrGHB}UyYYYfx7UD@G6<m-AC%}
z=psufb2VnWJ(cO4)ZxlWf^-VOSkl*oWlBweS9zet?vGNP@0I43TZz3dgvbl&Vc>RM
zr%*cVy@IKI7{76d`qkQtB9Gfg;vcN8H>s5#Ni#?7V=bkA&U%WbX)dV6M7`7Vlm9yx
zHkJ)H08hT3DmImT3_a(SD>FZhGpDBAZfZ2wk0nj<XTrFFUFlQ*xjJls^P`AAFSX)1
z;FF0Ns2A)@4UF(84jrG5uY_A4Whd<GfRwy#d=q~Y)JD6A_T<}GMBVuOL*?&c-ro63
zhkdX9sX@fq_Jca=D^S%w`Ahw%VKks2fi?XUKX{P68^TDKuCF57prdyy8oBn+T@$+O
zm*{mfdLp}!Y0MF7sDFAP<H=Mlu-h*rP6v^rqBZRPLM9Y=sWh0#bbUY?2b*yrv08%S
z){wP<e5>&cfN~zWxL-*8E|+4a?$S%ntNwPIc_G`Ou5RaA%(Y!j>fKyvvjST-8{y_d
zc8~Dm3y)T_d!s=||HP}V%)KxABkGXtd;f(bwYWR<><g2A-LJ%+V{@X9tOS(Jof?rl
zzQq%Fx-Hm%=f!vMPZ+z}^<a=*5KmnN*i^4AypZu*w}khQijs5kVGV|2V)#BhQX&TV
zn7iyz4W8zxH5vS4qBZ1l<M`t-xEcgPpVmd1I$jRj69LXw0iJLA76pr!^;;tN#1e5M
z=lgOEpuEswl`*uZAHt)(O6aH9WH;pCi!P9*;D;3&52R-))u@5n%fYu+9dMDfx+x4u
zkb*N<uhmDN?i6p0dhNZXt+qxTtir_DX6s^v4b-h5EL`oI++EKHv_b8u9L{Ro811Py
z%bMSGArp&Ml0qxFCRvj(#x~<clJmlN6(@X4ESMHlb8Tu#LhLPtR@<zW%PT2?8a;C-
zb#_{`E-g&Ikh{xKk8tx!G30NH?^E`H%hs>Ka4bd|mBFA*MOnjA+p6AE2B>};)(@o;
zh1&!Gj?v-uZEgF7$x<9=)xgAmc)8OQz69BKXyIHxh;CqK0ONv#n}&<-yY#-HouNKc
z!h}#q+pU2S{*tPcKO02ua&JuA6BKKxF0;KNkh|0F(^CiW8T*46#lj!^2$k%&1HwKf
z?IX9Q9n=UZFXXcDAwfWaZ-@1T%Q7iC;fNaa;41svg~^hWc}Z=bMBby(vAcdG-TJW@
zTFMi2JVwp1x)_<gAO!kyEuN9@gs6f9^k+Mz^~CCGZgc+zJFQ739V~X!v+)Uwt<9ry
zN8&X?RjB>X1+@mbJ)aLm(BW;hi_t99$jr~EdctFm_+kFDOBd2ztdkTk!V?M6Jxs?-
zou+7Z#DUW6EyXA3?U*YUt@`Y0Jh<TXTbqoBvVndq#@8-PmeOsp8c}6~Ny%fz-W3s+
zc`||b)VZtHk<i}P$w6Fj8YKtWbuAir6lEs3xC=~=dq01aO#_?m;~)xkOzlSe>=D8C
zNr@<EN3f#ov`>Qof0$=CHSbT>FSpO)iB_(K{#+B{fu-be)3V`2iAnfU)TMSk#Vx72
zY+nTq<?gK>f5MOK8zBuNC4ssgksb!^yMPn-X+%(WaQa@T^>I1>L1ai;aBP@A3O#7Y
zl>8LWxhCqJqV%($FJvN!>yg?$f308KZV6=$oMO2C|DyH2d}&Iik<=Y_F=N;8A%NO#
z7c={JMfK8(m3I5ZV8|?;oK3r!WP$dSbgj*%UrcMPa+S@97c3S08uNCCpf(NU;q`Xs
z#f)>iMmhP~+;uS%9mvAl3xl_r7ZcCTbrG_0_r;7qdddvsH#eBf5?DX|agn`8-Bx0~
zw|g%p{@pszZue;zgsl(>c>ff<F&8U=I%>0HY*?w8`8gMpV6!}l>NYp9Yoz*KE??(m
z9UlnnQl;j3(b2@24x1k|_`5l_Abu=#wb;V=li*Z)Sco)H&lDH9ctsijd-P&<0@a7v
zqaG>F9t))U2fVh$<AD^pkX^~KNaN=)@re2-N|S?994mStNbqgZmR!tu8IlkeP**;e
zM`LaX6DO^5iXLHnhNoDS;9oxqiC(knK=#1}e`6~&j=C1yN~}iDUR}Ltwf`01$Oq3*
zo3Sps#ALv|yk0G+qTGmUCiOV1F?v{9-(YJrnM+BJY3t&R$J(r)!s|c%(6bF{+T&3V
z8>b%eU$IR=w6th0SyPZ@z!hP6H7s|Bu9kqC^wF-Pur`scPWxtaKywk7!>6cKi!%)Z
zHhI(qv#yLl&YtMbYw%dFUe)_$yk8`1<CyNYK|K@SPV=qmX`sKwhD&qX?D3Y{w12Lr
zup1_8yNH5kgbUg`O2a1Th?8Y!0Czta_%6<GLcBQ%yOccBFP=`^#iQNoh4M-68dk7p
zV-^-uac55np?o*l^Y?0K#Bp(r*CS;)hHby5+LiE0W}vF6+Hmp{6t2;}Gd5(0)HPxV
z`XIVKte@bYxH@qB9nqL6KI!t|bAi*=9xo$tG)UiB0Q#|bF>&05j^lzvOm~R3FO-Po
zZVqqQiFm5&?p=u$=TtnoY6CWZI-ru>eZbCY$09V7TRqI(Bf^UIZS2!?8foXm$yPhB
z_a0%$Id)O+Sr=_K8n{L(>Pvc8R=1pzD0;*=f|fm|qk>?REAbVc#<0t`)vRKmDQ8!;
zj>CN6`qD}L=FT$tbV0ZlF#7tvY|Oh_56jw%eigO%M7LJ1PRTv`eKkr~G8WBsHNqzB
zm2La%!#ETF+uz$qdfC^5Ue+y79}7Yrox9(`Zj|OCUD556z^Sgcn~WEqYKt_a8{o}o
z%4Qsf&Y#7oJ6tA3fd4!oR}zKqUI0M;BIb^yYeoAq;PPoroqZK>y4^i3XJ4xqG%TD8
z-)Q+J7Z}eNG-}_*CkQ;`$*}JN8mBUQ>-%`+nPJKB^@DI+TUTX229Bwu6}}rkX`=c)
z&+gBGAcSbKTl&sjnSK!teQwDLmASo^7+f8v{cW#hC;V!c_gIbG#$QXky3V62Y}#vy
zpP@hUNZjeKWxOWK#pFiE8Lx>gBW3e17xLG#>pn@7+wIQRGA>=0{E^7e_FCf7E{EI9
z=mUP>$H(3Al(ZU`ZI%F03+o^4o<Mz&+rz!DWkO4`NVfMAg3QnNy_Sh0#fdXG+}i!o
z(}t(HaL$gu5!JbmtD%c<*Du=K821)Dxv65Vx4jIJvvZz?fcdc9#q9Y3thGq=*#e<S
z&=O@&^w@}Fh&`-t>xo3g*l&--AX7DmVS7{rO!wfC4|dsOG0@884fc3^oix||snpaW
zf$!-iv)>cZ*6ScIR!=Ma&|+;Y(Jy5wkUnSS(Yyz@n9BIGq3}nm3XE9TWoS#QU(4jT
zMz059D`Koo9$8YO-d7@WS*<?wGE{Z(VK+2P!bZKir`Tr-j%w7r`PGPxaN&c0Lxf%v
zV95IWM^etMizgO=OKj`aV~Y=IgXRX8vp3jA8)F#D%gSw&-mOKI&q$hHOVZ`Wd6pNH
zCNJ61qDPi&hqdWL;@&V_*k-+NYp1$pQI9K~t#s*ejhEu-5xrIs*V1miuVsR#`;J5o
zOzRIKesCBMy}`f!o(!u~D<E>DjoWaL0H+&IC$vosno-(#mOtB`%?V=X4!!1t3uWQX
zSk4IPO!O{I!-_%|N(`*wLZAbUobA?QE9H~zSy7`QUBGRR4vM5_r43ebeS>O*yLaG2
z$Vk!l>lGEFh!GE5@RPw#c2MuiaqP82@s98Z#I?g32NKG^Bon|9q2gtNIIo|ZYV#lI
z%J`Zc6}fd5kYqJJB#_(nr^r~xW0^FQ)cl2*uR`n4FvnbnozTHn7tRPNKjEpE0t6qg
z^L{$|O&&5{5QUX^R#b71BI?<4evJs<Rg{M(!Q;8;Yc<ox^4((RHGbxTP&BniAD15(
zw$bR7C^@g?cu6yd-Rg|<iLrRkPAv><wQ)fsdp_6j5;Z9g4?iz0I8TLfRV#E0rNi58
zQg9oP9=KN9HPMw#Nv=w6zwbs@UX>I(_j~H9%KeMp55&njm2R}?;-ha6tE#sT)i^ol
zex&XyS%5A{d_2{_jq^9ud(Xrrk2dVKPc+a*&-J4ADAI{e6X|<$nVJ1Fk&=v?uHE@u
zKS2CG*3K8P1bVqMd>IGBvPSz#px78lEIx~0i(K@EM)sQ$DRFbLZ}n~!`Kr6^JM~hH
zf|Q~{-v{E$TirF#e$dFUce?8!VdMJ3YXO4jPa;oM(w)Q8HQZImvUW>-c5&$D?U#7Q
zBLamqp*J>~vAvpst%zoJ+i2pzJvmbEf&cBJ83$gEM&G@jr;TQ;KNYM`ma6IT5i?lf
zg3k;+)(qg1f5&J#N^uprb2MWQIDjyoyVN^*!98KpO#R@=j@Wy5k0zl4$UUt#Ycw%`
zjL)QS8%@H9hpyinZQOsS&+b#-T(So2{%E$Utj=b~oAw^$9d564Ml<QdFWv4>YjXn-
zJ#+RvAXNGatrM5JY@SH53~Ps<h4V#<GAbRnH@Qw^L41~J^@eQWXm+H-m6CD@(dA)%
ziOsB}?R`Y>kyX7ms686wA__YMu*Wnu;t8I$$29^^CDHC@`y#a<TMajKwSZ?{U}?W%
zi-RB?*!7l(wq;dT9uLcF8*z85j7ASVG}o#$W$b=TqtyY@t2?2SNutvVHEBT)PHnSj
zf}wz;t%A%j1gDK=cAHgY^;V~+Rvbs_qchKM*BBGb!P#bO;%Aa_**cACEw@cuul|0^
zJ(@@~%~Zosr|0V27+^fhsBMbAFfw+LXR=A3KFN;7wOurS63=NbqTC|f8dyGU>KVg0
ze(id^{y!*0Q=B3%6}}Kk{=0Pr)@AFKTaOyodDi*f*ac<ee9?<Gzc>yGCp1$yTW!_r
zo?K7PhV>e2HqPnW^qRVUgrkjS;_UVRA+hL!?HJ7@Zaf_2$-h&uHha+n+ofgL-o?gE
z8ii*>P1IcDm<M<3qZY#H@OBRb+>qmf^BxhQJ2PwT4I(rl(e8_>n0VgR>cSkA3^($4
zB+3q!NXm3|lboCEkXln*hx@qDD!B_EiT3DOE+EW}+H*nQ4ZpFY8Wr5teY7}7j>WW=
z;&6@!R!h+i+6(Fj0lmdeXhoEBp>$_}QyO4M*<_|W9h(JRJTlZ-VTSF+!&FD~13bVp
zG@T37=KR{u2W2YUsy+qw`4kdz!AA9AkvIydcPW1G%RYg0P+?3Tp~9h%4!dDo_#=W8
z?Me)-!T0Lhr7bjuHa&*LDs`tKyb*0u-*vzY?3%_+05uK?efI9?<WyA)&k&!oBD+G=
z-qT!Fc=bDbUmue$KV8DF2j<<x@;r}bIA<RQ4Hhr5kF+##;7IvqIY;bcfrN<{n(T(q
zL#H!fpXjk~0Bzo<rIQ-bsn2f4Fsq=?XKIGHMLJfIE13Y14(to{gYOFq<4b+fhN{ZG
zntGhL2s@8{JUM;NzR^22V!^(hf&_aeXic+seZR-4+4uUPzTnOl_CqYpRqXRJJO`7F
z_{Uh7L@TiT?57w?0h83v!X2kO<R;$U3RdE>D*s}2^8NRVdcl|0y$bC0jIpI`ZN6-7
zdp+YpoG>X;jD>Ui>)Bm<pqFhuO^-x3;2-yTV*JiC{AfcRJLC1lec{s`cE{@(BX_qt
z^25BIi4mSggXMJB>)C<diD)|W^(-?AjEyY`4hi?Wg(@;^Qx11En-yrdCyATh?h%|G
zT&M3<w;YXJ7w&sKgn%Yg>;mO3jLCX7JKmP9s<Aoxus3B`m@BXqt846m*Tc$?_p~JQ
zL<jPekUl2pfB^DjhllCH;WgIH+{(dmVE{)@<LrHpJ^Xrh_F~p=;~s{I6?1xYS=O9<
zO&%2xw_zbNKc*@2t8Fr}?y$!-C?bvB*^k6-`p~2BQMEo1)3U`Y>@AKy>d|epT$aT6
zE6N(JJpSx$ZL&%Yy#}LCRgAWCpg$R8s`YX$D|&^-zT7h*akSM$XO(OSs|~7g2I37;
z7ax+*$Iqnt=+4uFHEIZcK`OQ)BCQdH$yo_^5At!axLJMN)~nZsRCy8xuV2__9*A!n
z0}sSDIJ|m@6l(B-rm3GitjXC!UY)}D`mIHeeFHqUX?-kf_sy_bR9=d2DeJkY_DL1o
zrS{$25^<S(U0m38x}MO$6%4&X++!2H-+p1{j>?|E&jzPv9!(#!?*Tho^<@;5C7DcM
zkg@`Ej%^blioiAxcj2PCT{x1K1&zE~kk^6q{P4P+F-!1F4^-QwbvosV#RXqWpnBrC
z?T(qmK5WnG#TKj{dt&x+*dd&9uU=6^(e~-7svE8^QsuY!341`jKIw9I@<_qWF8X=0
zzSa(j9wph>57)y}1bz<mJ*b-KCOZ;c@plpd)Nuobn4|HujIusG)HID1gAXtl`oO)r
z&#QgC5HE<t>ar7o1*S4P6<-r`>5e|9<Eg4WRd6}0$#Dn60fd*4*V7f}*T1rJuV>ez
z719!%53tp2po>8bXCPF_DA=e_UdG+>QsDB3r$}r*rcUrePyQq0(aC?CYggi{w6sU;
z?HCS6)Zad3SEB=xE`S=A`^k9e*7i=j77t#s5B{!+Sy?0<zP+b4@2VQe^u87l0a^V6
z&BUBs4+QZT!AP$5mGs3P`AF??)CjBJ$LgbJ;?;L9*$q9rZmZuuDZwk^g7Rr;P~22F
z4{yd(I+oD2YiZZ|t?+YEZB_G1`$9l_Fb;7;7h=Q}xN80@HCRd@FNs-8J&}iw{JY=8
zTi50$#nHDxBuu&}<G<5%vI-%a-*3K;>1@Xh_Xk0)N1Z|H7(EZ-qm0E|YuQT%^tGhh
z^3U=8a+2lW5~}QbPb~P0Xh`TZ0yc-KU7B=fFpL%G-7jUlJ%vHIHS+dLi3j!&SUoNN
zTuw2B=~E3o+|R>nXI#p7b`G1xpxtq4@_#G2D3+K!0j$nBy#Ugk@%`HN4(u78OSaGM
z()ZO97xD3!8NGq0-F+z*{r|J?^Xy-nbt#x?(QO0wT*_pPs^v9H?cPh7bl_X_zDt>O
zQnJ3eSaW}Tv5J7I9Grf3sR!(!%^h9cTYOW_i5|$0fj(>QrAz|Bodwd_1A(5$+s+FL
z;wM868I59Vy6-Je=S;o}I-HHRFgnjh{;*)TcWw6V{>T({UF9{#Dth!%rYM0=!|*Ys
zM^CF@+g#3z_IN-+2~{c^x9HMj$p@%yU~qs$9Nac2q7tL>@ZKR?EU4&1(&AYnTskH$
z4&vNUeo1N6Fq`Aw;S1??{_oMBuUlnl^?3M5m1ZID-s#oxNF+Bdd3jqA1fyD`)o6ao
zI0|d^*%l9%sng(H$&pxCy^v}MT}fkn;tBg*{#+Ad!uZ3vu}<LDbaMq?uOapgG&g%S
zwGEo<X5y<6b3txnAV;-C+oTz6?j2xhGzp(lw&HHfU&^9B8mM8>-qN33ml26tgeY;_
zUN)mG;1;-IY>q*D-5_>HQE0VJoWz5s+qy)Qnu#&%(R(yM=*ruyH$bh%nV|ntVh1$?
zHmI>6-lOou56!k}0M#jL?r>?RFbfn!i-)|2t=aY<Reh9Ww;jS@HD}LGJ*~v<w@V{y
zbT4TNsbmw+6?J#~jgKg0$<>g_@hk#+M88M^%2wYS?UOEdUm!H`Y7cp&_6O>v1W@C4
zAaKAE;CH-(mnO3tqciB&l7Y;9*~sZCI&cF(sp?se4vP*1iLgqJs7)#R0Q3Hwe(S`3
z5f<;G+BXb<Q`P$}+A&SYdX}j>A0LlVaFgz`7xWDQpQ!cS+Sq3&gh5BqFFmIOgLRDM
zGH0iQIEXo^Dy6f<voR@E1f}@e9*Jr1Vr8C-Kl2F3=LNZUz>C{n6l8A0>FzT+MJyNk
zye1>3bgEW#NmB%qB*%)4>FZ?_O-#4`SYZqP&aUVHgMaP7JnZ4nZ%;wsiVbtWsvlYi
zgl8vA3c>uu8KlK)F_Q$Fc*N?v>M$9}Jx_Zq1B8s$HhW*MIh|7y-s_qtZshEUX87cX
zYE{@uge|lAk(QckM%q2Y*~c+Spww+Q^ugLv^0-f?J{Z3GZ}0p&+T>58Q6ko5^6I>9
zH#L-qy7^2SV<~LvbAj+M8g3XP_C<W)S&CsbWUz^`HR5R1XJ5sG3rvIdwa%(^#s4Od
z<u*=;yLye%ncs;5UAbYvdQj2#YLoBwv+f5y*D`(r+%@2K_cX;g{8KEU<&>fMS&b|3
zHstm%Tn1g4@Jry*=j+zq$ixEJ7v*>`6yEkmV!s?Z94fcJkuk@9s$CJdI_-^&9}_X#
zZ`1XNUE$9*<BjY#!)EmhSE7^M@kYkt@K5R|;>GTKBZ(9qSeH_M-xdAfT#wU(8@gr+
zG0t%}K$5&&ebtsVl-Vqy*5o!0Zv1adCUK|8qPsT+;W_$;?7laW(2%tE-1h~!frLw`
ze8lY1kdv?LoHvqK)LhtXa|7afUt$layGHEO?1Om$6>W-d=K1lo)N^Q309B~gyV7G}
z3I3p0e0(^7lKG%$kGzpFayPa+J9P=>QGHDns2Y1LdZw;aKM}jo=JCMrp$!XcQ2_98
zxW7up_=M<P#pR^0&}NHEv|-oTl0cOWS>%>gu5n#|jx~P4D%E?l2lZ6NOwl^u?YZ!(
z-^gSxqRoovHT+Onp+@aE{SR7=XiViXavg9?)aobf^5Lakr>@<V5e`!?FfD|};5x5w
zeAy7;*Kcd|4NJ)rudNfh<sJ*XULaUANN^j}9z_H~xNTGeOibXlDZWoz5lN@%jSxr=
z{$ol`<|3EnMX-h(>0$$b$1$cepiMn^o+`w*&7wZV9q}Tg1iAoV*R`&gK7t>uM@<LY
z3C0G!UJZxv?mp{}d2_vt4aRqy^BuNTgh|;~JRQ}r(4d-#L)w;Y0rw)>9^Y1D8``0U
z4QOCkPCNBOPdMG|5?1&mk^6<Ky|ALc3${C;J^1wR>{($_OY~;3!1CP_@KQ}eL|BA^
zXg+bY%k~Knb|j<%Ju+KF&uMPR&c$RL6lT}|Z@*yJA+>i6j~&+bfp@so2-`&7jtI)y
zlvU-q7)}$OEbSgs+F?hvhF#W$_hA#YV?i$#dPScd*Td7B>;?7O>;5MvG-S?9SJ#Z#
zsW+0?wx--p>m{NEe$8;uvw;uYxI^%Wji||Hyo_Bkv2&%lqNV|1bQ-m9Ycz=$wS1}p
zHyYGjfxhHY{Q30L57}60p)9X!wDEw3cM6$nu80~rZqkIaPd&qoy6kE^lg1zz*Q5r^
z*#mdiBZa&A?Ojohl(&9Q@KCq+)CON8T;+q_$w9%-vp&171r<3In(f0NSmSd0NFO)4
z9r9z1)f4>Q&`Z~P*e7pf(mACqa1cd>7U5%`#^{$fG&a~xk&k;dwa@Icpwa@4%g?9K
zcbzLqzcI#r5fq5^3ZMB>D3qf){z~9Fx!AKMzmC2d3dI)tMlZ4dd$rAPg)piO^1JIa
z>GE4W+KnOmKEOvxI{QH%;C0h&KkD)N^O#zG(qyHa|Ms)`qu`#q0NzrA+8oXt_KTkD
z+Hfv38_TZlY|f!hyKO8n_x4V=jNLw#czJu@QgX)+*|f2YOIPC`=xbwoG+;~kMZ#u`
zW!&9a73;Rd-N(|AneP82NkAYSu{)z5zN{qgdo<tysh}n4LTsiQJpGj2J(fusO@tXj
zw^?J^HUFun7uY>xiLsZ}E<*vJ0o5ka|Gs#K%*D-LU|h`%{<GsHktI2s6Mu7`ZnwDt
z;(6ytT%+esk|A{4JORpMT|t7FKbG-n2A+ge=hF+uGT|xRXT9RxLQ$-|n+5xDjH8*f
zw&eDFWGtN$fBd68s%B-b5Pxhe<EaoaMWlT2j6E*sWiG|IG`~ncUHw<KM_0#yJt0U~
zLAdq{b6mAuT3@0$OQ%nS5R6-Sj93<rN@1|24S_qN-mAuvBsB5jfK><eYS>}6A_hgx
z3=hDqQIDz06pUVTheMpz3FR788;I^Xs}ER+O8C)>@+5mAKxK`f`OHAUby_<_B@hzq
z(#Co<>dB!thRJK?rjR*!z($QHIg+|d2&`T|GQ{wV=(8pPt0#}Yb9(_>LmIsTl`LCC
z<1HT4(-up&k(GeHKSqnDz%eUoIK5m0VfBLLmv}@&>k&N~JQT*4QEx1Zo+j(pOHaY*
zq1ra6pYYV6N5-~QBw-sSqZ|SALUFxs`)$GltQgbwz;c+fZrsIp=q+iaNEEg+281*R
zE83-(wJ6<l_KfILy5a3sR}HAs>{+#63DX|7JppXu3?;E_uRi9qaJPwlQ$K7)zA4!L
zSYucJwcQQ`lCGab#yA+OgbdL|JERe(yH2w!gAR{PW-0@}0mE361Fr{k9toIKX|v}7
z3fKKsI~q`E?C!U&U3lG&MN`7Z?Rfn8yWc-$FQ`A9^}DFn%VBvUAQ7w7Z>O}x*L4={
zw5AB-D(T|zW<0B5)^Rgrg^p+;WQBvy>8rs)TYHKdJ|7q$q}WCEkjL^d8hsNFh^Akl
zYq|++#ImoZc>CD6ri1WrrtgY|!o|7Ull&F!Z6U*yry7G-qvxgd1k4HC3TSps;09As
zr+1@ap>x3A(>i$iq2K(Cy{|{u>X=;@nvJ1;{7{|NP-viob^lSU6jBA?UiPu}rYEt8
zgnD*EG{ULd)t`C$BoJ6X)M}pwRg(VYriO_ixdmklr}}4sJ91aVM1CHhZb94rg@)DB
zJ_xzL{9jMWFcN*G{;@KX(-xp#Yg{=>t8?%ECa@q6YrlOfu%0QZ)4tPl19#d^`#wfo
zgN=pkHu0g~O6|uW`j!Go)^WBEL<iWaKZ`8A9zLD8uH6z?KM`{FiyF0(u7T8RK6<;H
z@f5Fc2?z1h=eEm<lcE}e`(MtuS1EN062Rq<4&cFgn|?V=<i!q~5x>L0>sHzwmlJO;
z!-tZ~!<~BdhllL0%OM6=+06L31_$Z8qdgyYP5h_L3JkjQNK*G)&IAtZ_sHhB*atM$
z8e#MIUCvYuE8Oq&{($qrUcd6#?0D+$?j-2O=3LGM2rjrDa5z^#_^lA3`T=24gbwor
z8qH;|%@-mqtTPWgTA*LLxiz#xj)el?eD>jZ@+7sqdn86n{wsS_kWnC`$`>3Y(LQ&A
zdOU`>9Cx<g+M>(ZZPQg4Y)^Q|6Ph@rQ|=7uQocC8avMm4EeRU3(mSlYL{CqwUODxp
zi{@-Yt~FJGY06Ch->q8HMR|@Dmosr1HR5>cDMGYsE@$G-^3|T&p9I`YPHjNy>zKHV
z`)r-)QBkwp>NV^S4s&=n#zLrAWozQE#!6dvIdkRk+M$tQ$hq=*O%*Ode&^i~Ln13!
zi)|GBwiKF@rfAdUOrq`E;^b-)9>d(`^TNX`3tLNo;)uvwTQp=FxW;+2dR^{{D(v+7
zp<zp^8PGM=HGI*#>Jji8FaE!~tXGeP!oU{m*IX)Llnn|f$G-c`Z593TJQ)`1Jvh{E
z3tXuVK!|%RD8yLX5kI@K#dfO4m5Eggy4OGE7<oqToBPo7@74&o^^*FMusF?u>zfeY
zOL?C51nw*u*K>}#LyGMULL(nHAn22OHW0z9XRqxK{GNVln;p=IDsjcrnX;Zbx@-Vy
zyhPJ7OeTk;X+=$0{Zcy;4@fi^@}3joHGT;^sz*<IYsWOsN-qn8e*SWH5n+Rjq$GJe
zv825a7`n`EC$vo51)eHwf>9ST<LPKj-di%>#NWMz7Gl{;zoQGZbMZG0)ctlo#(^%<
zE(*OKUlyYR-H;>3wB0VnXJox4qfN{)N?<JAHXgg-slT-=`iLQTveCDL9y!thp>$Ud
zl#WH^9(@lWNL-6%9^}+)?`oe$0q)QQzo$vS56RuXNHZsD5atWO>wy(xbibqz)mzg5
z%ASu(8)bMn2XpyYFTIMN-3U}_*tnkrDp*NSaek^E(5Z5fy%{6@!+iTJz$H&b0@q{Q
z%ZS!N0_W`fA{s9KoqZ_~e#fG8OOVlBBkgNp!Kv43-vmrQ9`P^xHlE-xJ$T-J7hp(U
z=+&Yyx@VI})utaZB0EujoH{PiE&<DB$Dg!koVDU&YCnsdC^xX>z&>titHXdX1~O^(
zOAyAL8f`o~qP(;YW!1(LCnZ?JZXZv~)1%@U`?T?lN5k6jBS|LF>En~G#_q+)VKV{}
z24+vbd%^Ay<eGNZ4BaUJNe}3!_YS)&y6fTQGIKoRi#=ZXtKaS(&v+`qdHBo)I#902
zd2x@R;|_s@aId~bEMcT}UqGJw^l$9`@gyoF4Y17?)CyO?&502ZQJb^5@i*5WQqIIT
z%X}}`yzz|r=L?8nd^SIRNkJ)L=~@uzc|jn!^uqWYCd%gzt4kIH;pyX<Fw;5ld~dTy
z?L4ZVD$qKlu-nJRGj{79-n~5@AOE<TIlF|6qqa!nMjB{`XFs9u+(zZDxg?p>cX;`S
zd2xww;&g7aa*=b3%V}1r*LdKP1YW%$g42;zmnPoJDy$H7Nb=(DURI;2!pYxi;|aaF
zQ>IR!oP{#Y^;7G@Gsk;^CrY<O|L!UQhqE;SVvuVnA>`}g<x+0`<O{R)<4KZR&)r(1
zAZt70-l$P<9m(6K@yU`a0u0Vw#gaa=DL()8cFV_-SxElh7EzIaWxRX6GIN`#m`txe
z(9NZ3P9=qk8W}S~e3EsEFa(h>^VCNzVNNu8JKq1j<C&Pz2tIx6Do(i${O9x`)t6G-
z=b(P!_LVfpTh$u^5#AX#tie*frq8y;V6OgO?Y3QrfAFtPUG9=Q^i$GSyQ}oh7z;N*
z5ZxtU@Hlkp0c_dQTH9Tk`HACL+0}|b$fE7hw3i_<?2U$-N$zU<)E(Z)?()iITKH4N
z%s1=-wL!&~%9`0hO_*y02^K*k)@O&*kl13bClpUUJf2C1s4D-(*AUyob$dqw30NJO
z>H@vwxA?jpjcMQ7V#lHl3Zh{<t`*coK0hq#!qPL~u!c^k0k?fbu~YG}=SH&A0dla(
z&c?4uZyh!gzpkycbMY&NL=@+m&=g<<gS;3^4ljYJyuG6$Ft(j8yQIflv9;UAMDGUb
zA&qOvcCzsawPG7zDXr~A_O^(Y%dTcu^-~D%ICD=1#5J+ct_8k*9asNk?*<__@cP47
z<vk4m8zcTx*vn&)<Vae7U5H|)8MF`86KCth87@@zQ3jjS<zs<D^TOqI$ZiCvVYdD!
zL2zHUu$58!RD&*3RrF>&HFc-^?Poel(lO$mZvSMTM^EK`fPSIrWI?%t`^)HP3276*
z(mPKaLr~z?@gd1hDdnPG*CQx>t3^OS4Y$AVV(U;Ahx6?F7&`)MVv#?@I|@W`Ui_#&
z8v4m5M(Vhjpz56c96gj(G%lwc-z^akZx&SL>X3p$2z&Ns#*7*0K}2A$f_vMWiOq!}
z>Gn4>Zc1)oU#QdG%+%Dw$SBj_%%sw=j3l?PahvgGV!4>LlN@=+n@JcLnt0Xjj5l!K
za<7kY*P9s+<Mg?Joc(8;sUG2h{e+R;tscv{)Xvgt*H6GU?s+p~s%yAWyC{8c41hkc
zY}}^~DdJ1sIE^?V{ATjso2?HhBsoVPVC06G>#@2Y#n%Js9%BcWD4VDEmAN6J+ya#Q
zE^(t)W4}$|9JWxna=$<ngr~om$xSWTmC={KIq8xRMliU!Ku=s`6FwGA>pA({&&wVU
zx<EE}SX>m1-I2ik=blj496%msxH!IEikhhrGnFk7@MZ0sx#gOOl2kQrl>*T-P(aCS
zRc}s~_+F`+QF~8MyH&^JZbq0|p~3hymDsj;<-sMjL6?a$o`9w<F!zP+y7BmmJxj<5
zCWFSQ`EFcmYf1!lb&jo@!t!w1$2h=GO{i-`UBx+O8#Fjl*bmx9jf99Bw{C9=#K;tH
zO;aGf*uf}=eLfJtGtomhTcYhC4lZqiQ{#1ndTsO6$L<wMNfRrI2KA|=P?x5>jPi>(
z09lVdf<<yw;c)8JM|ifdE&5{>VBp0B9t3b@Ab06ElZ>*Oi*&$s?!-Ccjyl@}q>k`}
zZrlE5CKV*jIwDoru66`K&%=TqJU;R&PD$ahOG6waLKM`Vc{7tA-Ms~lz;C<tTgg#{
z-=5VtSE1`f@oIZCk5$|^_X;~4MbYo>iz#Ywe_3XBzh><E^Efmgh;LngMD*9eH^r9!
zC;mEq|3$sY?e|bL^LQa<<FGn>=ij&35p_6>FAMi=k-Y&Y+ll{R864H=)!o~gBj0rp
z=vcs+7^8PC-l7Nz^pN-Lg{dEs0?SV52LwxJ;HRQ1Vrsic?VEg4PfsmVSKX7B<b!{<
zvqH&vY4Tql(H<*s$;5dk9scJ;9W;O3%VP$k62nCeBG>F14wnd-kI`tuAvn}+mqZ!Q
z?lh*U3im$l=t3mHK$Ng6B3~WHFxApENK(dnish@qiP+~Zn~e6XL<-`v53UI$Dv{<6
zR~5f25N_xqsPR1!B{G(z75IK29u6JXqZ2av;7ZJ`T@V^LQJK__H26r-JMp|b%Gt-U
zXX{os*bSlC>cZP6dhAG7c#$H-S+NjrPK~?ZcDB#7*8PG`+6j(?&trQw<1J`k#NWeL
z|Ab1tv?o0c+BrX;>??g;?>Xyw{XqFzzrn$p3*6F(e2;zI(rMpD{~p(pmiTwV5P3aC
z*L**PA=Zz){SeQ+EEuj_-0?NU`W%kSKLvvCJWmpnpEW5iw<JgPTUr{8xVrtKZ?W5>
zP`{FK|E1_IT8Hd5{m{dFU?sl&N_M%f_JXa`u4L?tjm@~GUrC(53A@+~eab=P@ili`
z$ryK&CnJKx-FYQrFr+_3Z0<_d*1OaW^LGlont3Iif`bEgcf1Q~vGPW3)|Dimpdj-I
zfqR5rt)Hv+M)S;dAML_@Y5+Bgc7HtP@U<W&#AaWSc=IZ!>_0h!K}4COkGwjf&DEge
zQvbk}$<StyUSXa_?XjS&i1}A^mW9IF&2nB&Y(W6!7*3&%3)P2XYY$(^1SKvIIG}O!
zdE`ox(*_WOa`tHS`G3rL=r{IQ;8R&>$)Tu0{eERKGaI_8oo+)4<^Ww3U^)J&NEWBx
z6B-S*f|=pP8Ufy9Uf^I!%qf;nD;JVAZu+lOQ#1zsP^PLh85D`;#HbD&KiCTyR>Yfu
zP9}87YW|AOP9^paZbnwCnXf1A49VB(gdEpD3Ut-$nN-(!EH>(k29zBbF>H-~foh4>
ztP8p!eW0sZFO(@!f<4AI1Y(?<o@8R9o_fhPwn@VUA4<eT=Zq~}Yx9Cd###>&Xo&%H
zWhcqJAQe!`Z5G~O&CRBGCF8P@F2RMAF8#)_jz5R9$)14N*FhXvZ$Kp9DXEgIUl2Jd
z@tYWoJ<?ciTjLGO%!B(zY*?5Ka`V|1?bfWe?HV0v=elf1bhX*tzjnqaNkeeicEuA8
z@?I+anHaTabO<Nq-NJjIx6gBtappZcMH<gz(rJ5yE|M_Qv{wx}+qe1|yH7u)Yt(DD
zU+A@XSki$&uZ;q5b}$g?>rON}6zg*srgT`l2Glu`(O4XbwiRAeg}c0ZL0a=0($T<6
zPRYmApn>=|7YBrR71AmZ@d1zZ(lBw-PDI~r6UXh8$k6TSK9kDfw8k)T7CGT;pqEl=
zxW$Rv%t(MsrIN@_r_S;@PsepW7PIev;JL=)Vs!ZXzn^2H0$%RLqb}*&CsQ$=F>Qnj
z7fiW8jO&;6#KSTgyAr=tbB7p4kYT~S9q=26dhDtO1oN0UgHoc~CUs($@?xTQ%-c0}
z-9~-hyuGWAa$Q?f4UqR@4|;79oaag_Er<5@x@Ih#VYu;qC}^oRlYJB<D(2d{(I9>t
z2$nTe+YQat^}nP${wJE73dEjIgFac7{hNV88>ypu?K9D?(S_5`17WW39%bO-%ol+~
zT0~wG`b#0z#9BdduM-S0nMx(xOg*RW*P<D&yCoOkZ%TWZyy5n(P;M%8xL)YHc-ib}
z7r&1_kPY#aMZ5b$V2n|PGR{BhOWYg0>ex@El|U`dpEd6K+_2pes`&4*h<*`?(+!#%
zlG$6?rM_w6vfcJp#?z5_$!haf#;40j-s`y?r@fVNc^m{>3!rC$MDbIZ@z$g>z;>95
z-4X9Rw<#wsJpL{5zXR@y#YBZa>0NIncIF$Y+GeU%lJ;Lq|986I9pGG*ojeO?y_F75
zws+<cv%~IrD+wJNhzq^<t-n%mz=JpdDa8Jkq#6oNI@i~LiTTdiY(18B;n6ThqjN%I
zXAfezYY>d=ff#FYyq*`xFofQTlkR=M_~*yhSX|vLy5AP4bJx-0H?vT^p^>2ooTj3c
zyW%5nNdVbTcOI?gm;Fae1jbxtkLiO7@`F7tWLJ^wlWMMlj=s0Up3q|rDt%iV-!@Qq
zg5AF)X5jxL>uq4Htg^h{6=S@Hc!@E_c#Sc<ybKKt%@|{hVQ7XbilV@(qNb`qhR`&p
z>eTmV*QpPx8Vy6!jF)B{UdNZA8OK2`<Dl0zG$KtiG-Gp$5)rXLL}ZL1#1LbQ*BC>L
zG3Nf(Ug!Cj_q{}udd{<-z1Lp*V|}kx(b7$#neupmj3a<m{3+u;+)Rp+=hs#_1(|a&
zUIxOQi>fre?v}P3Q5!3T6AV-6%ei~3`cE03M<ydgrncmwQ>%SO*>_^+3Rb6floKIe
zhX;MMSiROXH`qf(Fa+@yYYcEnPtLW!Mzc&o%WJ1#%8iV2RMrKOLiB|FST7hc`arGS
z^t(aOU^ViSYa9QR$<y$vur0`0q4*cdNBTTvVg65vJFTm*Rt=np?=fo^U`X@io!7V{
zxGv%<!MNSg8~Qjd9@xU(=xxDItX~kd;(F)yq=DGz4XbNxP`jWKHfUH+)ypbuQ{ct}
z+WKrPdP#~+S9iQ_o3*$VsIAaaVvCQC2T^ryiLFNB*IwHie`DBfwQc`PKgVo)yv0L?
zs;}ykJ&T(Kxs-Nj+59oMJlm;uZpFb37EU0#xxVGOxwcCPB&IvnffEi0^GUH(&qi~O
zTAS?-Ts>Z8d(@6r>E78JC`sxALgCdhG#n}0o}b#z&L_EE5Oav_3q)Ynw*6Y>RnT)(
z{4p>ID_{a*UD-IO47=@MAUtIQwU^@|Stsn3fXlOQReACcvfBmBts{0wFgYGLkB5aE
z6c%JlWXBwdK~qi7qgv$Uh`h(NwaQsVJFahsdc7_gyZl7VrM|4jPDVE=5dDpS4>pXO
z{nVc%Dx?@Y(2)5?jK?=~fBfS)Hle{_Ux}2DSO0Cnf(3R~d%6bo2cET1@N+-dD}GrV
zvhzY_Jt0|rHmT<_m{+RHM31wcc;e6OlHTFm4(oV121h4AeVP`#60a%w6Y|T|`1JHv
zlo%iBQ*2pqS0DdL;lIR>Kz<Os93IFE5{F%jPk%CV(O=nhO|hpBKaNkdI0Y}<Ll~T}
zzm#yZPlfIZ9EpdDFX1z<2Hn`lKNlo~=|Y6RP=9QE&v^3X4{#I~im&v^3K$B=zn&ZM
z{O+=w@yr=UHU+2O=sgaZxPN>b0Hgp;a@%P>h(Yc4z4p7?X*~oHcc+5csf-b*#;xF%
zQ;G89o$8V9F8SViDkGf@ePb?=+;%GC6*yVD?e<fN@^ic8Y}%=e_ToW6jumRk4B8#h
zpo|o7Ha!Nz6wCN_r@F?c%wrhtil#aD0>3+&I?aX1+;b}3u(j28Z~TsHB=#h`FM#Q9
z$8#K%-Y)<=M}sSnIRy`#N(=*x+@VCt{+3zIh+e8lSi!A2^HfIc$PpAVkh28a5G8I~
zajlClp!D^9%{J#$M$bre=h3nqeksfg_=obn9X4NRs_)~#ESMT1vTGi(2i2k{M^gQV
z;*t22L}MN<_2qf49yyh`ueyOgkH<p1Kb6=WTnucX;ADQkXJ(6{EpV2TL@;lQPbDS`
zafD;JL{QY0a*tbij8oAE{wvfOiMxhIkmlp1J5*a${0$W&^G*DXCuf&c>!TcaV1v|Z
zUX%Zmd`oq)r2wdcjpyjDSA%LdH8uvsu3}6|gEjhqP%{s1v9+;+bdhgqCf%Rki|h1T
zog3s5Ke0aMiedwU?1n&6S6{zvjI~P!Qj*+hc=C*qcK%exaAi@<Zj`Xrm|`25VXQqM
zj0A<8yH4S_3l9>!>Pq9D@EonZ+9(ZhXZ^xXN})>Hd3}r60Wt;y8qQ)8>W8B%?rE-!
zO)-VX$v`|-I+%&;b@jh(4ghgQ7_cYfZ|vOUteqN>oPlujTZK=YHG4Ypc)tsVLJxI_
zZM&Wud4!hhc{t||I6H)E<Oj;%xS!<C7;-d+iR|fk&yA5vGA4#~S4;$NPqxG}G0Eaf
z{e2!y@oc<fJ!_ZRZsDu8vfB1Uw@^c})@q^Lo5P;d=0(NuQZpJXdp=sYRmG|BK7GK6
zBB`CGL6!r~-~o-a4lU`60fdVceRl9v#_9PpaW?K82_|0Fg_78(L^kXdVVKFVi1w<+
z1+acQ6w6k>>Tx?P5IWkuzayFhh6db2`S@sDZJTG=F?9n3BX(R^DzCQ_TEtb!PZbTm
zT657B><#sdIEo>GgLO)Sr{H`W58V0++Jq+N4ks}=J0k=k5ZKvxB0q$u>o})(-FUQ|
z*XMsd1Yeq>vu!e(l#}=pLHMGuRG;%J%`ORGoZ(4StM}u7kDJV(UC}PXWLYX=ay3A9
z8=ifnJ2!$3A^zM0AFG>HM5C}VuT9Mw1^l307v^fYTU`_NNgx($7|m}4RCQ1wzrsI_
zN4G1x!_a4%XCE2mi1hXq`#io!yaMlE2=4<J>~K^<qSr+6?!~Mh0HxEh9XaZ20SM*5
zo^Vr`Ez1wvH_>?td$4cgRYy-ULVOobo=F;!`+H5?FF@ojD`tdwcmOruEyZ;4w{kK_
zW>?H4`lOoTuD`9AU3*A7T%Sa!yP@y)VkYR5UYk{#R?OIgnT!6y?kHvio~wp{^ps$K
zx;}yBf?D0F=Um>m+Fda=w*un0GnaO^I{nE{7TZ0=M13jl&M^*-?%rY|aLx2~Uor6)
zOWhOUe$B9Fh(Ln}rUt^gG-fk2NjT?}5^83&NQGg2v&>QpH)DE<%h~FqfWKOo&50qL
zooNFd^8$%3hvsa)+E%xCZiWTL#8_~RHM{hmY*`P+)KF%->hYoY94#dtBM)oVBP`D&
z>I@Qc`k(F5Xu8SsjN3v@v<4NVEs8JUDm~0;aSR*s9Nes@gIZD=2zwENe01$TPN_;^
zMZE0Ec`l;LV&dmqYZvz4D&eH?bNoHn8Y^Q#n3wwl<(wi^$0`twH)yq*M?I@sCnRvA
zaTB@MV12YkO3zuNTK6#PHL<X((R-~G0@7Zxb;ZQujg{H@;zY^oU<K#iv*fxOX@lTQ
z>ams)+Ze-89gqMw^%kKG`M@9!Q;8>O4HPs&&2h(5TN0_T$2))U%=K^GdPav;=(k=y
zV?AY4^y?J3J<xHCwRAx9e7vQ@1_R5W5FL;c;bB2v#udbUXPboOIuxH?0@B9RN1eOM
z$MjaOsIIUlHDuDd{AXg#w?qS~(=^%EVj^J3WB;FjwoN;!9``BRuBo6-z<cwlKzb9Y
z=>tPR-0-;V3_wVXi{ptst;vv0V~ok{Ql~g0wc0bWsKw)CuhR^M5tw(E7O1b68Z!E(
z(=!V0Ri9lj>I6YQr+qXsFvfxKjD61sbc+_-K7EY@*k=2urZMChtPW_{DtN9J1J_AE
zco>cD5j&{QNBY{k?PWct%kdTUwiGdnx^H?$)+Gbbq3ACiILy4LXbuZ{<a)nGjsyVs
z2YEEWQ7tW6B==b&xJ}bTmEyX1(0Dw?;Y#EP!&aV9uYi>wYPOStv)ogEyrFJOJW8*f
zid91m#r7GGb(k}ErcK1}u;Z|&XEYx#MownCI9~`1ue;GL5vG5zb1_))M0>JXpV!#r
zM(DLk^+EO$Xy8TdXTKZ$ZoQ-y7`|Jm1g8#Gpog+6fguXFV?(m5LI(~OE%4hc;zt3#
zTd+S?@2#G!firn6KKJYkTsqfdEY~2x;!;XR>+kIT4UG&zZnjU0nf^IpdMd^5XEEU1
zxf|_sbwCu2eG&T>XQ5vEQlKF1kjoNRdv%`_UtjAn6^gm3_Dge=$AT;UCWaaRJ1W9y
zg>hg3kM^B@NDj7#?EB~~yW(s-Bfk~3Wp<06$hBv;j!z`VF%ojwZF*DBWqA8|#tERv
zY_e(5u-pyqcZ_EgobBxSG=1YuAJ1g`>d0et=Xl~Ba=pcWr}Eo)#uBs+^>#Je-D>TQ
zifs3L#?#IHm$U8O0E7J1Ev({w0%{r6TJG0p&=?#{?1Aw_<ZFpFnW2_mA8lszSM#{d
z3dpG|ywhe!8<f}{kvC^*9<IqDr~%vmyqGtPS=y2FH6rBPUGElXCJi0k9P9_9`@E;I
zen=gnwN9!fPUyqq8QYY~eiA71NC3Ucz42l+j7ctk*g}ow8YVO&i{j&y2>>3NZMrzd
zgI*PENlbFd5-X314Rt5Esv^E3OLWTf%~zG<8Joeb@YKKnRy8%4n>ml!%9t7oE)RbH
zgH>y)%Sa~HWVPB7%|y#s-47<_nZuyjvE&|E*k+9}Wb*FCyN>&5i>=Wo?tJAZer-%A
zSHLK@E+#>-D);+YA6TkH4Yncv9EF+Q7(;O1GlQnmtTF1q;pF30N1Hn`wT@@%7&Nl(
z6?4Se1AM46+Wx$C#>Wt=>Q&3FJK#WPU$EZzyZ91Ew?8_Cf53-pK;V}%zDxLn8m45u
z>$YK`m0Qgv-%T251uE7tec_oUJYSP-*5Y95NsgH4(&M9=$cDOQJj)}Q6yquOkiRf7
zvP@gYGggNf4Bw+3c46D1gPdzSaiJGb!zf~4HJ*w&mV0g49U5QBVQgoteGGz1=xKos
z&kHY#6ZJ0jnbP9_(w@-|)n4|CqOSU(ikur{(%-FKsE0+BaTausHdUjiSKg~VH<lYg
z_4Hi8>Do?vUU1e!5<75V#_}G&rHq2^8}7Q@{qfEC-(&DR5Z!kqXTO5I7yuAR<2g~1
zlI&o#fcW*gpViB;*H%#E@s(Kn?o>SZRRM<18;jeaz;8O{-F8?z19q<ml3_>G9@Q<{
zfuq93;~qCMV8=A^wif(XkL&$F-zX&4PQ)yTo$j@hG1!pjv&G3yO<i$_czQ}*O&;VN
zkH@vp-fk0W3CBS7A3LK3XvH<as}b1Q(y?K|`s|$M!7_8F_1Ss#-RQonlj9R5uOPSS
zakAWJ&Ao#T*+sRibfw`XA;3AlKD(^TamAD)!WA`gQHmV&8sAR_i5=`C^}5W<SA47)
z)ga^)e{0vqGZ~b7%Q-9etN91J9@ukn+dk1Cc<Q72yb)dIivM1GowM<&x`as}!oWV$
z56A)3Dzwi96;QhR)V>H5cXrvAf)6_->FA<qx^moD@JIW4s=*MK%gsOoWp*J!9Rv=#
zBJ7)3U+$vExbD7<E%?(}^X<C;y$uar(qDXEIu3cjw9|>e=H0W)ZaJOZcf>p56iTPJ
z>Zi%QuIj(=w|3j<j7SqB??U43YC{?u{IWR!)AZ9MiFMM=d51pp`4fUL{d7jt*nwQQ
zB%?o_5twpk5Xl!dXm_2?Sctls6%}^(>5Oeargt^3$M)QFdcqkHcmSf@)dyFWlRU1m
z`{I3fC#3iO(}|H`dhXu)fV!jf9(GlHT>ET>AYI-uSVcwhnNu*}M`^P(xi*4ZH{pzL
zvrlK&C$59I3WiBz&gsNhw806@(_~#U=bkF_PbX$&Wx*Dl&UAjhJS8y`d%=UzqPET+
z5&#{j)a+q(=J_u?9pNMKi}Rsg<j_^~r!!_LSdNA2!u9jEDA2%W@gz%&^%75D9Q1{N
z_Kw8BlnYHGX#PivujMI|S`j1E;7*%Xq4Km2ksI(h&F;@{Re^+`{NyLLQivnE3ekcy
zP%SXK3CAK?y{1+PaYyd6x>&q?4uPm%Z-X}?XI=EZqNmN)2uODu9)fCOv0tmHqzKhf
zTNeXH<x=>!^@52#=brV%+-VE3Q_<}?Q?*gQ;Su9%0M@4|AY&fz4e_3Ow(hj!g@aF8
ztFVfa2WemUfgrl`NeIuKYK0kP)Vc+5%CLqSUq6I7)%rD0;zoft)x#K2FM!i*g8>@h
zc(~&UX-&im`{~>ihrSi|j~ejNe=;8-P~~m2x?1iW(UU?DYPis%Z_(ztzV#K`s;Bzm
zQQIbPJ)<xEZx2vBf4n^vO`X>#RAPtL9_mK;*Ule|MA=+VO7V1jJPNC3yJD%S9c0gF
z5RYyJ{m%;K9v0DdtLK${Za3K@6nKpems9p?tO_LUR6^@HeMIDS>!3X!;|-lYY5UZD
z^$=!?{TjY7gh76j9f*^<{x9ss0JzF?IUdv)ei*Qo?6jA4Pc--Sz)ply@hj?nu!HdY
zSGB;C|LQ7n6qf&Bho*k?LLVnD?BQ%3KAlOFFmZpRTB4h8J+Asl095nWb~I*Hst)o;
zI~Hrt!5`dY$EPOJ4a6j{708|va>OOWlWN0>bfwxGF-7jURy!rcqJB?iq;VaT!S>=y
z#r+;yJ`t_l90QcUzVItMb2_tz1F$<6g~|UU<k!jFS#?{1E0O!-oQuWqIKK0O(G&LJ
z*+gad0BJ>SL!({PGE$xxA?{Kjw-PI?r{=sI^Y`$MZo3lSEq7n`tC}aJ1l&dDBdtoV
zt*yg84w%B;>)eN3i>^*@ExtBl*L4aSplzQ7xVSOf4T0+^`tX;<WFNy`_&*WOK8u#C
z>MQK?5@?<w!@dwWq@v-*u`jjOu03Ft|1(`8hwQ8PjQb5|z^`NB8#?S}X_<!6;C&NK
zh%n@Qe=7u)r9vv-#q+Y>e1D(c{@+Jq<S#6AHW7Ask4N?37hLovGV+fYPf;|xbs|#@
zRN?Y>n_heIC|JYWClZ|-!VP~Sk;9G-O3h7VRQ>dC;J>C%l*D{bd3t9I8CxL<CwIk%
zgfQ9NF<4oB?TXsvcF#mc<OpR(e7{$Ny9Iz4)%zwgj)JS4Uz(Z!{c6;X4>)?12h<4a
z3q#}V|BL{FV^V{t>(|fBn8(kF%Ts$#t>QZ=Z?h*d;^)R$P<1>>!*c@iHX^v@O=Qd!
z+ucJn{>kPGl2v5U0aptqCY+HWi*36-7=QO;L9>UVA@F0{adPyqI;(Pq!v(-cv<ySd
zWA<o#jawC_c@0C%EnWn+D4?obvDOw(b;RCG{?JR}Cq|;*ltI$Uqi-%47r-ktQm(1l
zDg!SS16DPW@n%loTv|u4Tp6!_vCyiuSg>BmCbconn9IN}HL>({0d+2q6LGzuuFDZ|
z3#pGilVYO~1J`I3psuByIcsAdkxLf0^mWl01ROWJiHtw<#vHJO#SNNBExh1H?Md9T
z2dqipwxMLmSw8y44nbsGYYCPYddaLrMfiXvc__L>eTWOYQ#**LGi(>A^92OyWfGg_
zCh1-cOIYWS^=m;&N{`>$fG|-{#uKisLH*Fy<~c8HSRXaw?BxZBY*WmhWK^W9(^OC%
z0LbRjJ5N3EWK4?^YBby;Sjs$`?^fYwDTiyD2FVSzV*7)K-LCeyJ^$LCQVXi7v3)@N
z4h`LqWQCpaMM~!jiF`VC$7qM`QbSzK$0$dmXQqQ}@~jS6Epj)}xzu|>d$n5-q^-WE
zv{AXNSc1KRzYM+KbMaJ&$hqg$ADh+7(oj(Xh5A0NT7O@@*>fE2*FN(?l@L8bRuAa6
z$MKx(LXFU7FA5Ua-jjAvxbV{nwWmOxrc~$QnD)vP2!6Hx0k-jvfbgn*c25m5%SS$_
zL)k+4`9G}Q6XkTt`AEF?#ouH{H5BO=xJ-^s9ZlC#`VRe-9WPx@0DnU5aja;9=*JF2
z1y`zG|HecnqeQ%7$eeiX3B_bew{dMuPiP9SFrgo+Ft)Ju&IofUWg}TpM(u12j;~kn
z&F3^Z3>2G^Z^HS~j_pUXnT)@SFP#33U5vkRA|A6#fdH=TAI-JP@uIG*fjB&^43)t;
z?W!JIIhOWOeAb@a%RerSgKMu{3$WO>g_afI{CYI3EURu<W}i&$lYzcg--$N@6a6iH
zIFs6^nmD>rY})o&j7k7?FD||I`P6$dqhK5PLLIK|CwLu_@@0%SkQ?RUtEu*wf>0-X
ztzA;0%!iZ8;->J6D2p2cThqSLw~2E&Y2OB@Y<%Bu-^Df{zpH((PA8x7tNy~TSQ0AL
zy_res?BjllNZ2iJCPKB;-KK7RGb6$y@F@d!+nX89UDcGL2Uz0UqX#x?a&Co<nD%By
zFi{l40Roy=NxMTs6<S)7>3n(&HD~VJ`F5w?7GJ`(1m}ah^b4So=MR59fkPY?#O|JW
zCUn&K=GX1sH-pz;H<AHQjoWd2#Wo<=Nwxq<hWD}u;)CK#I4IkUH#0)QRNYe}`3uez
z6!5D99_&Bs&5RZ%PbH5wCxZXY#3^~rSe9kZn~5=6R&Mj&%($WEA%K`ag&Gf+Lyyi%
zQS9c;jPc?pC#??#X6Byyj|KLSUM}qzLg!`=znPe=6=h_!k9K38!~4<rdG5@)wlJXK
zyp(k}+oG5y(c8GyEf!kX{ggXjqTb88n_K}^t_J1M+F`5Eb91+QpjO5+j)IsES(K{i
zo!OxcT&do0Q*%zcx&)wC2(OKn?Wh#s2co5?(j4vaIFx$9To}eWAha*V9_BS~*62OX
zytX#_MQ-h|b$ZiGc8~Qj3Ugn&#x@B3NMz*L5qM=o#A?F{Of_C%dCk<7^>6}OtW`e@
z<Zh3lxujmOPIc&6^0*bd-^^G+v{&Q=MfKII{eqPAbh3v2!TL*`{Lw4UYpxCbgo^^!
zU@Rp`U2Iq*pwRR*2b=Um$p?2#Lr%Vs_|w;Hvu5HEG6e0b2_iE|98n!?(Q)7cq%NRq
zI=05JWi`uen_eTac$^R?eY@bq{VFMso)UWNeX)1wk-FPdwTbSCMz*K*DPjnzHFLHr
z7Of9~9KWJxrf@_w4YH(1<vcqD1Hv?}>NvRUjy_j&`uAv()L;T>dt>w;r+<HL&k0Y=
zAji3VUNCS*vFKwQ(1{;Hd_ZCcxKZ05d&2c01m<ZFbdD$xUz`He<yp5J{oW2v{TSII
z+&cEMc29ZBaPcTXuF2;8iaLBOsUlttJoqgM|9wb6*Yx2VHDZTjpKzAQ$`gMld+Skk
zLv;g)(lL$3h2k-ZuO~g#@wkZoYN4HoSv544)sxqL3Sd_*xO2c3dt<80QQ!Wjgs*DP
zn}v#gT<hBcmuC}dz(oir3=0seQ$e}Qv$1$!$c=jt2T*cOvnzNRDzfQChsg3nEvs5>
zQVWjy%EK;UR4?iqm)5vAF2xwjYS-9hy#swt0ImoNaC}=3Gd=mn>vmP!AKudGCMD^k
zmv+pSA8X!^=SD&EwKp?<r~xZdZ(I2g<+GXG^}t|<E9UGIp`wQD(sm=3rVbx*`!u><
z|HOLxOkLL@e8L(C;3|rZp@aV-Rv#>R&HFFo%W96&SLz>9jXz&>lxV5bhnvw72SWQM
zz;dG=rRO!^Qq=WiwVzsO_fnxI8Jeo^bzf5L%i!ClA((Ps){zCmDd#Qv1<9d_+uCkD
zlhOU<eZ8gJ*SDR?Zn|aE(IIZ4-)=vX=r8q=Sf^=c67R6gH6xy7<&HCnYSoh<dit3}
z9vA<m-5FnyED3h)E{)=s1gsfK<h##glyP-$k_zt85BaFqd+(Wy!cHDV>2lwh#J(hV
zuKQ!i?v_@2Ky$+SoE)ELOm)=J+6n=h832O_;t#o&XQ{#C?y5Urv(IGOkQ{*!a{_da
zf6CcBbsyovl&FodJGv-Z6E8zY#zI5hHB(n+L}`bIVtC&Q9O0Apu)sl3M-lc&j0bps
zle0%PX&6S-00Om~EC>h%i}VAvQvIr19N>^H&Xz<+#Uq1Ilk%7o?2tQYSHzgOzp#SL
zxH6E5_Z#JEO0#hNMq=q~rGTKST)S1rq#!538n}(AR(R=c8Np+cSbKexN_5o62tWGi
z+_~1M#xQ5YIWiTm3Fwxts;ykH+|~+fI1yrR$D`ENMGK6BZu40$oGnKgN1(mFNQ!;;
zGPaF@8XWjB2{}Yf!qw^(9ztb#!IA9L?A=xY{a7E^ZP(Vv&hN7#XV01J()eMDH248Q
z)G$;4i@kboW(wro(^!uYdE{hlK)<wX^28u5HmKpa)xB=)A8lCOtgiKpT$}Xlv4(gN
z=sBfzu+4h*l;CalWNe4#(JtE}9KnV~eK*@0uUD*Uux)3=%sZ2voOu9XyD&mofI)le
zOvV%Ym75s;Ew&?ihNV4*@nox$CEFS61;Ifmu%~17VMt&IN+6*TekK~R)vGa^JR2`^
zJy8GHy3RmC|4R=kG0k30u&--i{5SSoX&et?I%&_xldG}GKeA5?Udb;I+5JK=8y!~w
zR8R+E&{;?(FP4DxE8w6Y8}1vkm#2`>($;3L2!<xy8SwE_V@h?{A&rXu-Ay%zONXl)
zR_RC#mv8B^qj5Qwt+rz^K~#5al;i5G3}l_q$f@|lN$r`)fQcWUM&mB^L`A1GN4I=o
zpCdPJ!0Xy(Un84PlO77MrwElZ+8hmljBgr{oDE30GKxog?3^H}a@uiTI4E>%!dx?{
z?SmCI8DY5=3U)E(OExIGq={9beY>myv30rX4@6yw#$@5KtAVq^;7Ff+q)8($wz$>q
z<3Jb%Tj12L2?I^Jf@f^KerBTN^4mfRkKV*|K*-gshiDJDnj4zIs&0I7?NiM_>R%U}
zKMP2R$0iIfrT}6KxTJld9?81pxt_ihUO61jZ5^|(H0g$xpWD|l!I>ogzZu;U=8tf%
zMzc1rw{L}LLhZWjyLg#zB~}SX{pev0@Y`F79yY<EyT8gUZzY<{S!%*XKi)YBbmRJM
zZzWopQq0}{RwnA#!0i)l|5iqLM_`_zAa}f#(fdX+Bn)~i*YvkCf{4A*?u?OKg9j~t
zE4vW9q$o-hrv`UNdna%3oA(4TKt=>6)_o0F@7~x0cAvU%XFK?Tc62TGzm;7_?pc#^
zQNsAx1NyivM{yyP{4q5z)nGFPW*z%&Rt(Dxf<x<|&DNJVT6;#JIRXnGyWiV9y?3Z6
zV>Ca0SXzoKCZGLV6ImjhT}iLv#m8Oo@elUkTZz#jPUxY4q4*k9r_UZ1E}Pqjuvvo3
zN8ZZVFs`5!{`+Wr+_2mhmX?hniFmX{Z)MyVIk%B(d)WGmW4<eDS2Ws^_?ycPX+B1(
z8XWAek5o=p5w9o7b!V0FcLf}fRS8W6sQJoxgL8JDRmaNq<VKs&?Z1`yxYdoztWIBG
zX|?)jl`JpTD757WUPk&_6aD20U$M1%Ie=A$K-YEZ$Z6?%ZN&;|e|OjhJyV7Y8oDu_
zSCNSa?LF0JOQ4R8|I+g6sTLumH9&HgdaBK+o4<i%wOeO=hDZRJ*0zE_BNrS?{~C(d
zUiC()c<UG5y-s2md!8zD(aDfktRK|-;tBlQdI?Fg;pl->${k$7n>6uNk*#V>t6tVh
zszwBh&4FHSEPGO|OHqYeqBCkvIa{z*^C?3-%h@*dl?sM}nC;rieNEg9Pes?1|Fazd
zHjWh3nTfwSpp=<;E8US!5+Ix&N#iGVm$obUkep}A+cN<QFn9CxY;?h$L#ep{i%-KI
zyhpQGy1d?5`n~Zf$IZ$9a{;6CXC7?(d~Ev1%59&%qijSw<Z8d5sKY@uxh)>hPYsDB
z<c@h!b0$x0mmSn29uj!(MfdZ6YOlm!%RB7V*d)c5J#6q$K;MJ<I3=_^tTlDh#AuV}
zz&RqsSEV#BM>R%;m)AI^*KO38BxO!?fwI`6SKwNK@<fb@j)4jaCpFX7fm9*qjX)=C
z4pO*!BRt!Qwi=U1(n}dnL|0s+%R22$X)Z6g2m9GTG;h|R$2<2{;!qRs?k;&AdhPc5
z^MamgK&aqj>^fHyT#OBaByN{9EW(ky3S17j$Y#psLxXcg4P5EPal0A~2#R!n#zFfi
zpn>xG<^HioEw~l*S~Ld5Aq-vD!ZvbyB+r&l;(K=}!(@IVKqCskKGn#BTo33%K2zTn
zIXo>t*SWy5Mc6d?afFx^fbyl5xUU(%w8Wo$r50Tb_O&PxyfuF1UI@W%stK36eG_BW
z5b)yJI=<DI1V|^hx$iVeBwBTG<@=aK7yK%gk`i;$QRu_-x4*rkK#{q#8Bt!@SNy-Q
z$Db9E;86HEye(R|`hfor+wEr)xo594vuS5D(HJ{&&h9vy$Thtaf@0I-DRPt8oo5rJ
zs;aQNG$wAUc%<E3YKJ-;B7aZx-`0kW$L>9w@dyZ=uA8~<Y@!YBL2<u8jk>ORd*JMZ
zTfote57><8l&!gn1)F&`V=Y{`;pTYVW=%msZF#)#)Lo<)$2(swI0f9!^Uh`x*bkF7
zcG-LZiI$2(XbWPhzgS=o21Ix|wAn-I?yplBhKKbGBMJbI#IxIA`$k(@>`~!1)r(#j
z@RgjzT-mj#^vgE<e73vp#TK8<cp~h-zeL69X@i!ipC-5dmH(g(3+nZ|$(gGuO#W-S
zkSk*q$Wm7PtyKjsiJNoQVx{059QA4&)j}Me-l?`yZOocu+h=w0_fYRB8FMuXz`0tp
z@odIt)DWx6xnPCY=r>mqkRuS`Wvx(ElLYE@>d3=LA$#iuLj(Tnu29;bPjH1}2W-@S
zsK(K(DW>OqXx{SiDk=0%T5IgE0#{PIKxub(@|0hvGhXLe(C#TrPN^(<^^L2TxJuCX
z_QwF-+$Gq(2L##UOd8=^uouYq%2zfVeZWAo?}(yyhq6tYGW8H#W*G~_x__{3R+kN4
zdC#6K0g1oqmT1=7!bG=DodVBc<qE-VQ(saA1~+{@lL9X{Vo&{G^R)Nb4)xHA1ca8o
z)plyaeoZ6!aR8ncvdN5YyQU`S95t{%6X?%%k>}B#RiEy`nv&@4F3k;T(Dn$+be{QY
z?2Q5Nt@2yqIW@sPkrcA*t>=ZXTw;7|Ukp0_&%JEv{n0@go}Tbn2h;(Q1HR;NCa>Fz
zniv;KVXP~6dYc^#bgW*n0-A$SUSl+oRSEs%E1C-?0e3ORAx{`Z=Y*B%P=JAsgoQe+
zmz{821v?^i*OE5Xjz)LIqphtR-6Ss`i@Rm<FYLGmFLaU6FmETKv$mF24}T+1o;o9E
z&zQX-*f?w*)CeE2Qzbl;D!|4Au1LB!5ig)Cc<jgDLxZ^Oo)x^kLw@0(i{}nfL?^D}
zyoN0I+j{bBCQFt0J*pFMO<fcu)F9zQcEqI^WR;6rmo+h5&T)XZD_RMH**q8#CEL~b
zpfEU$oF1Ta*LOW`AM2Jz)^?X>j-WF{*VHS&w!mav*P^4VNY+?deWD2>z`|PHhza>Y
zB`c(TDoB#cYx3m%O#KvwQa)Ip&*wq}y46wpA`WCT?lxcQ&2p^5_Eo%@Gi#209eZ-9
zcf>8nHw7<|DlQ&>qeJx=JLg*g>gqxC@PQ8w^>O<?{`T@}W^X5g`Qybuvs>O4v8HsY
z-THP$`d1VRFs8S?orrg`ryR4})i^0(roEl0ej|#Gg541h;JblrIBJLy6&d~QM1$dS
zh`EcupLk-{0=xU|ME)24!tRN2(^Zkga&^!^JdWKbu;Aq}Zr=ZPVgcM(6L~rxP=i|X
z^mp5gw=>43j;u|d+ufZ<X2#g~iuKy8fR}XMpmFxw87n|hGSbKK)11K3I;8)3(bSDi
z-8TPiZC{tR|Bkx_bDS*@z~xxY?7`AV16(8z1+ZWSZy(&J?BSU9Qey5juKOh6cnXDA
zN0kI~*+P9>Gell6#BjZT%qws&emmpMxb!-lTU!!C7e?4G<!apSR;`M+GiIe!U?Uw+
zjO!Z2#d92~tJG^5sj2<IuMBhywdJfj2B=@Qw9#q<HF$8qUe(3J^7>V)>#SZ=D-fNZ
zbl{D#nLxzj=t|aTjD}SAp<ruEtLjPgpm0>N3Gf?+Ja!2id%!@x3fmAcq|755<L5<x
z_7iK05!_-2;`8wo*(<E|?TNd>;sn3{gWGe^+M_cr`Twv^VXVf}R(4Ne=AW?-AGTg?
zG!&697py<lvaQJm-p=@}W~}Lc`LT-z10_s(S(6RxTQ4l1vrYP9d5hCcxbJ2g(~sE2
z5NeWN;pV^<dg9&^>3>qqvH%Y*$t^MA|K~sd$bYs~D?m@NxwgejQ#Qrzfwp1Bd@6o!
z@~qc8re^G!9uC`1b>qwn&H%mkv=(xC@r3P)C-O$PSow_J)p2z`I|XJ3Cga}Zd$n6l
zDqEUtkM;qX0+P0EuZ|drx0{o;&9hxS7pSfpwdVu$j?N+5Cj>V)*?uja7eKZHf$we)
ztMk~(7vIh}OTwp6p}JSgLH*`#Rahywr(YHds@xa1*ItQ{H&GZ?BajQB*A8h#>-&0&
z`|Ys4Y{L=+r+Y+GasOR-@`wXF8r_hJrTFh~s>g)lGVC*iO6Bah2I?xF=(ZDjEdC0T
zb5i3ZmERtFBTza-wqx#)QvucrUoRJe#{)?y9$AtJbw!Xwubqi6@n$-KTyR$CoH@_V
zX?9pF2ue5~+ayOm6`K^ua7S>ue0wogr3PK#rD)mX;m4P?zF-|8lXN)IZ}J7B?yp@9
z^j6l{M;gALQ2FR@!Mwk;Yl69oI7Yi3A8dTWK8bt2IyZ`PZAC9?X}^eW#FsxIk?^P5
zYs<KKKT|*O>ePM>&|rI$DS|nCp$0jU2HTgKB^7h&8`kqHA;op7z4mp0+|<&7{pO}V
zN1*(reIqL5#b5h2g@*OpfWErMzEgAZ98%Fnu(tHJo1IHUbNn~?F}vklMwU@pJDEP0
z$Xh-8+Wp01`gYs7M2~Sx0nXdcCBlxnD-{=*b}plE?nJ{@CtyCBcQl>;AKf-RnsZSR
z0S3iBmmQT%-$}Nc0D?8-9zxr2_qoIyEW`B}p5h*LP&SO#@7{Bnke(79BBY<w^@t{b
z@tyy=|D3{2Qmxdqldv_~1L_TrlrEbg=;0aN+2y3o)Bps5V=1>;=Q2vzi;8soH~ltS
zO)5MKAHsV;%p_Nl5NDpA+<-A(DE4GDeYPO5gY6b{JScDq<ZZKu^ko+&wujG!`hiS$
zkHkj|P&HtWo)f1KK$DRIwdO)0!fo&+2ctzApvJR*E!F_d{X@1SW=0q^DuHCuD35W<
zxSa@_)5Hn7gz%BvP7^Qd9d(EOD*dn$(T0m_Wo)jF0{nZmzV~F>E^*YVIVmw*KTs#|
zsJ=<oT0OQBq+^YG>~-JuHSvgxeNyzURc8&&?QR@crypvFcEojMy?*e$1r6E|7|mk@
z*cgBg`Munvaq3;?kk{ifcBHk&M<}tZU2h)iK=I$H{;5xff4z0<eQgU&Zm$r)-Ae=P
z<Jt`Psk3ng)C1~NTq<o)8?ue)#RKh#ti$?gsUr=OgKbjBHTc7MHkGk+iQ|2|&Nj!y
z`k;9Bq=u_WB`de+xyg;kwpH6|WRq>vOGFC>*{(s#Q~Y_8Jr%bDyU^>dpq3x#z(BYo
z@UyI}-gc@BT)PoRpN?m4@)S93yVPYf+J_OR>Y7|d7B2CukhskKmv^gA*RYcnM_U-n
zlI>MvYyyySJ=QWodtQ$nD3@)Yu$?@SPul*#T4jYDh)GZ!#$MC_>&T3AFrL6XjwklA
zdLSWDCwjD30)t#T+!U|sZ6jsc0OyeA3uEdbn|3%x{CS5RQDbs-xc9(O9qV=sgq~Om
zk^>Eo^<?2GARB+@1{nE7X`TD|`=qew9{rANdqdq5fI5h7=9H#Y%^fwa!MvVN0e{JK
zZ%zan=FI)y?Tk8@+@J23voYXGKZ@rxF-R4L2J+LZ`7-neP=U#~>4SveFXpKxE=Knq
zm}xGlhZJna<#1W|SDWjulF`Yo2<l}Gb#}EhXDX}OM=>De&+V}grjNDQkGZDjng%Ev
zgiGU#>uNw27Bb+(##)6jEfL@wnnChPIb@$sVKsS!<7@VrHY3!m_^<Z4zG2P0zW5iK
z1mXJ_+><Nkm+=+4>Gr;%ws!j}HVxV|ZmqA?WlfG8-*z*eU4N6*SKmZq$TK_mTP>R_
zPTi;nKd{@r3%o2|zi5MfAM;L1+T?s{_D-VM9?Ih&-ttbO`HxdPwaN3h-TIDbC59B&
zMsZ#}slaW~kmHVP6_DQkPDThl6I>q-EYqTAZ{b>mJKo7?v*%x6n@oQvV-DDYZloWu
zJN1(*HHZ@<a6ZN<QQ$*%_dAI}@mdphk48wyxHld<f+4f}-btJSE+vC#CGL;!a?NNB
z9*8$Uwz6;3X2hGy>ayiFGv1&bbJl&9=I9A_I&8MUgJno&Lz^QI2q@^adGT0F^sdv6
z`BNR>NjPQ;)Vlb$ynaw{l&x;GhvH>fDUtBQ?<Brz8I)ku9(gCD>M1yD3*-ADJgT2v
z+2|Jq{Pam%sI?%j9yB7q>z8!i!RumyvK-vo5<%LIi!HYF0L=}?s8TBQmbIlqdZhqo
z)>aje;%mx{v@%9wry$1hKdEy{m?zba)kae!EN;ZQcQSUb#NF{*y?(1z1(C)8wWgFq
zaE<0tu`H^BuN8PG(b*U4;)90d(h8;5&U2*zgKUWReuLsbVH>rqcwVRE!`2iqdl4ne
z3sbdA%gU@(kCQJ<{;Rd?5j8BCbTrmdlAcnhrdyq-lt<2g_D1IzXt1+uzE1K-QWlbd
zz-|i}A>jfC)nnR(c*$$FeaZQB(>sY1+=Sh8Ou*!koous~sf@IHPwKlSE*HGw<GU_2
z=Ub;<;f+YboOd#|nO)_2XwL(%J=(XBz3i!&L1#ZI?;ZMfm3va!&VY9*4nmGg=)5q1
zsKhU9=UoDfC^&mY4N_sWV;E|meP<%`Gx7S$$*T@~nHQ4X9b-3l!1?aceEa)adTg&i
z>BaA!47V{1A|GMv_PhoygSp?QCz8L|etlWf;`I#<2<eY=V3^g5YTnmHdfR}3E18tG
zYT8gdfo0!b383I$llS$jnq7TM@(ZI7?K?@3a>h4AlH`hcIKB&?{vX>BZF7%dZ|x%*
zP9x>In};~n$JBL-nRh<n_|%O+&<&^1J<>^q7xR0v^v_PJtI2&_9ly6XG}_YS?6b*E
zX~z9%2FA5dOTKN`!X{!vjG!<)6OB=2c%bIl=nNC{5K_Bu`E!C4g*VAj0sg#D4#&-&
zn~dq#dEg#vc~M_9p<+z&o|m*MJztCyOY}3BV=1WEL{7Z8UGS$xdbnD8fzir7(pavR
zfg8~DAq}_;+O?QU*_txD9)FU&(>{qVpiYB!BWCE@c{IJBszbzD>}sD0eOQ2zs&RkU
z5a`&bM`mBdqC<2z_p$|i8AG(=MQL9J_(W{LyM8V7U*GaCc2nam#ZQ(TczTAvV$XdW
z&!J#tg5Sj?R&p7Bua3G$saJbG(RUWPsjr+pb<6ogWmqF_=UdNbYI%sjsNHrxyPq3c
z@GB%Id#~MoK3z7OfA$NT7N0VqUo5dZ&g;g20-@W$L~qm2XB0lHE3ERJ`YGx4JmT;!
z^@D2n?f{{ql#l410GBb`k8jZKJ)dzIBZdCrYh*sW@4Oh4=aRw9E4+9h>HX)$Z;+&v
ztNDTRnI5`7cltPq6H!xjkQQr(X5BP^7&ucyP^&tz<+Idaji&~kEtneEOw^<glynns
zI#0861qeu`f~{ItrsO~i&QG{!04cxF*Pl-u3bSi+tNawe++NV&hXo1YG0Bbgk<x16
z>W9u=U@e20w1paoz^Z`}TNJN*h!gfRxi}tAZ=?D`OQ$V4pG3m{{tGJ?%pR_V%gjGm
zg?^%-4%zjPZ7MYwg_bv2ReZi;ovkc^G}r+rP#vvE{+v|jwd$t<PnyKx)y4EvNwa!Q
zKi4~KjcUvZb$tUey*3~w%^?p=L0vz!(UOn0v$?iTJ$2%>xIV@}4(PTG>Xl50_^2|K
zjryr3SzQnXoAlENcP;!6HE7gj@o#a8kBPA=sFMrU9+Pf%psZ7f=hNIK>yF{8UAfj9
zf1ln)E`$E_>DDhfB{_El2W|sebz+-X8yk#Ika7OqhGT1uxm2%F@`HF{*;s%;))%(+
z<^Yd$lD(dA^+|0U^cpF}_7?q8=!iNhTQy~`n2zzY+qMZ5*HnY4?fQW$&>f+k(t^OB
z;}nqVL3J-g)p9ZIoLZjHt+}f8Y4!8CSKuTE&eS(V^*j?(gIvOiqIh^#2i3Vvr=vkc
zc58GbK(YXH#%xcZp?JUzbbA90tsP#2s9?{<qh}Q9XL`>^(=}z)B;eY2K4YkT*Sn7j
z-18yZAIPVA<!g3881Hgk#$JqfZH3~$*}+mf&x!T2Fxf-wrdwz2l^^zw6b7^YYYbek
zMvoVsWlYR?b|?UJX9dq_aySmvq9552Aq<EA7I>(mfv{#`K@et-ouBwW!l=Y1^4reE
zJtC5m0O0v_j|=n)c#c6kF$E%ek>q!GGPc(EuiEU5m`pP|Q#%zu&z<#S8;>zdfg%%A
zT`<WM6@NyfENjQ(!p_FG<N~#GF*9y9$Yryg*CRX+d2V8&GuPWiZI`60#_|LnE@`v6
z*#U;ZF2@It*Vq+dhMBnr_-YJFqBr{}07i0ag0g)aPmyQdt_1{WE|Zb}dh7-4lPvTn
znuiObka|Ij=LtS{nmNh<0@=8jJ`)s>f3K1Gd7!X=h|A=Q7?|jH7XiOiXI{wIzS4xz
z`E}6V&)L`U1=y-7TWUA;l;@)UCZ14Q;0kNs1|llg+IKNxHH5^z*91Jmw9D+>jIQ()
ze`~kAn^EB;M0=3nt?y=p9_ttO_Gp2Jb-H>7xyf(WNA7eDm<5~mZpIEQ!ygK7W&Dco
zcsH>Rp1lJX`FAJ6>5Vgd=ewEhfK$qry(>Ushf;sd?vC-%f~7p&_o(M`)F8|gnC^|v
z*jRYSznfA0rRmgTihnn;A;VbdABcuz7BAS0_`A^)BFua@qyNpt-}apToy~eTV^NSz
zAl!*DogH7~23f>8@i!6sB&rt*D?s6V^+nXOyL2uneMXLZdr<u*X8U!B=vI3u5EJe%
z55KF+#&z;pX{<*y#!~E4k7|rw=YR{gFvh4~*3e*!^pX<-e`t$^lD3q%sn?dgn<)D7
z<yNk>TTa$nwpq?9gyAxBQE)=QaAovQf^7oZ16=$NaPV3g$i<n{s$-!VhDd=xv{<dW
z@mk3d+EW*wy2XfWH`-X3FepXyu9!17@&7ShQJ#Hijpmalk03YHS|Jf`1!f_Y@?NLj
z@+~kO>)%ZTlw*>TaA1?}V5K(1_z)!IjEw?tDLyA0IM%9Z3RTE*E{)`4=ENfutTjGI
z0|Cn0HDC&{>(t|@J3DtrQ=~TNaj!Nwxdkxk_6zCwW4Rr+#Rg)S#`>~a8&q>D9P@5i
zK%s$k8fTj{*b2y=jp>7xLwPpmW+CVYV%3uYQA_f=-6EV~aK#8>TZN2ey_wN>TdeKC
zP^)bh2nZyDZdiUwb8sUOcE^J4i1xWnqqeg&SdxdIj)xTHWV>SPQDfy9^;zg=dvGZ{
zt6$K(3^^0ATW}(pdYZI7nt3&0DYjRSH7+MUHw6?viR68zOpRKU=bA=spHRidp`z>l
zI9I4Hb|Cg2m;xZ($Rxhi=iF`Nw~J+>x^>Q8j#uI6-D0l@RHPN6DD74Ku&SF|h1xWM
zAMRZ{{B9;+u$kD?iD`lFI1;Gp@YokS8V?H=Uf;6Aj%h<fQZbQ1e3;#FO@~lZ;5(s}
zhY2S48x$0!zCLBMU>!uw$EJBBrh+acZ>QAFSiTP|jh7%L40=L6Vo2Z~MB`nsGx2^Y
zK%NZ*HYBFo&S}oDT^$^m^QAqJ@3hHiiG~Gk)-DFBaoqKgM>jiOil)nI%IvbbsgD?k
zE4u!mCw4VjuE9|BQK{3XlC9!ntv}|A-{H)##jZsI?kbP^y8c5{z(h6?H<Y(eqQ9}0
zR=c6~URp)>pK9EOjwbs|FVQa!+UG(IdauLiqkix9d4Z02k!#&#U+QGx2I^KfzWhoZ
zu~&xdYk>_a+$nd9-Hf&8{wn^XeG^N?IVF-k{!ZyDzSFYRx46~!dyU=d71eu<#6hJP
z_EiH>GUt}bjNJG4b=s|y8EuDo8ME6a6Qv};!)~9PaQfUgG?KSzlZjgV)f~G+UstcF
zvFY(O{?1W3K0fifqjuM1Mo`<~aoI6<$4esgI_w_x0`q0}#v8;LD#k~TY*X}DJ$Ao(
z8SdyVzR+e5=!fPGKQuODGVv6ksiNLyPG)>WRgU01n-x7LK5@3*r4pt#N0=!?Ti=C^
zc-~~j`qVfB<=)rx10RfK3xp|ZHvQTjRCl=UVSRikTK;s-U)aO(3WYK!v<-Fq)JWaA
zrW|(QN275=#bdTG{#@O-)E32`Kl|ANTO89YQ!1(@lNoJ<DTO0J%Q%?yfaNj8;;TSg
z5x6KH#UyT(!tu&AHC7ccPz#>w_DMmzQczV>X@R_lR;^z!N0Rev)M}%TRC?`%)v3o`
z>KfF?U`v-e$I&Qkcm?W$t%(gYcebr9Ep7{XvvpcL4vH5RTpyotc>aSpfFrg+IIj2l
zq8s(t;8uqwJyHsx&GON6lSi_&s^{iDY=$JKi=I8ql+@p&)~N=Q&oI;<tvkNH@b`ex
ztA+Qw7~up09dkuE^eC=@QsXVaI2eET<zbYD1re7WL2Q8%s+YoDjs*l2c%vtwdb3aj
z&w_62$!OfrGDuP4Edr0Qy_VA_Z0ip|B5^eSF}6)@sO*<i?COzAo)v#e>jz`r<hIov
z>SdYhj@|!wr+%uUIxrzMxXJ}2Eh!t}t{BDHl;rMb&rD{Fd~d00A!K?oV~)A?u<|)+
zw%r1!hctLNz-uJJBl~c*$l2b~exlSXn0!tRnvnY()Aqck;JP&Tq24!{CGem8H-?I6
z9(MtFjjMLQP=a59b8iRY2sMpTBX%-lqn)~V#2%j?vV+lLpbrbY04}t+CgGJ>S=bv&
z#S1nMQcAkHL-FO@C+6DWz!>~GpB>RAz|)ZOGx3fKY4RmI7F{+KU+cEx8m=15Qy*5(
z?l48Y8TB(p>Uw)a5Kw(9Otz;qU8=ZJz}UtEU7ksVhz-{@O~imb9c|>Ci9@(>iJjFN
zw)A@WPZuN43Dqg+8;Sh<)GwQm(O@Pfqc=#&kX_X2Z{ym%B<NBIsVDlntj|{$hESgo
z`KC`iU+q<Wf{n&Q6%N@)fzoyavyb()yZpIA=9<24SXOV>Co^_Fse(DBzTTgx%?f-=
zZ>WzP85`W1^r>))Q&7%66UvdcVc<TG`9V2b><e|}d*RD?LxgsReWkBZhr#fC9q(4U
z@O)DPjCiH(Z`2f5JEzrC2PO0_hp5c_-Bcqo2ejGu0-GYR+&(zJU&x3jIu6P;*ew?l
zm2*OimHk4Z%%pT4wA(Hu`py>jC^x?qZoiPx_cr2(r(MV>qNgc91dq3bQG3|O^b3jH
zuWl%_JEJ4EEjE?TmM(1R7c$l$PaOxlJ9^>{X-|yQJ$l`UY-RVx;}A@T@2>mwwlG$F
z&F&A7yGKbe_`rp9p*XLUv}uI3*^CR>>f-3&Hu$ueXU32vqV)%xrA~24qk3F)S~pZ^
zw>cLw+U@s$Qh(2jsp4nte)aQXs+D9&TW}$}y)di{y7K(z-S(gmMG!y!2oDJ&u0ryu
z;6lKGcv!;_;)Nd%(*7g*(C<?!!VCY)2sBw3PhhAI_H2>X6^(-H<`(ORillJ1C0Yi=
zEKl`p<<Y~^`ekK}%dBDwfFZ0)@BvmC4ai^BxVpirVxHZ_eYWyK;@Y}~ty;ad7Y_`z
zlIEmV{gkz_r0(TXC+xO+m_@z5p<r~kOQDULX*H34B%{5M@gx_v`Ef>K-_Ftgy{#33
zArHS!dZhobbsCY-=)rc+D7s#=C-b5kT{j2>C|I{`)PS8sq;<5W3!(-4xYB>mU{hrc
z^KnGxQ-N*1wF&~RCoWgYz>Qja>=)ea3f5V|s@q1pty=>j*RxrAqov2&SbwzSVDwXO
zcOXWAIDw);p~$QK;jV4NLdoRz(6ACQzZ)aTrs#3lLrunn3F3;`>|8gS<0rD%HYbsd
ztK-QSm)p)uNNy2??t)aXt+C8=fBKiUt#miQGZP0V43_&v@{}<Acn5j`^zZt?#Xwr@
zjE>x_{dB+x=Y`Ha{w9{T#hy{0wVnX)S#^<11r2Ps-nV+n;yuC&#RW!)C$qh3(Bx<J
zx#*knoT%B=8F{kE{AK%u?uOOnwqLLD4#2t84qV8nIz~y7?NSyDS15ZizTz76<Prx3
zLCdE0oCnIid?88OKl@+)*(=&7)TY6Cja4M_P3wd6GU0BpNQZ>}nxS@biyw|DK<0>j
zi@&LeO|6=kX@v*Lu^o;DSe0&tCEQ(8E2mb;iC8u+5yZ5U0dY93xHZt;&_vlJk>k(h
z?o-hRK^BmF8yBR>@8WlGVH1KBhYB25&jfT{=g-b+9Z29jXy=67$^D$)*X?{j5gwM9
z2>#9{)mwhl_0t#Q*%OeZ8~RdAf(ToDPc91@)bm`L0C+{;kM%jOu4-k_DZ-mmGF~4A
z?+xetadhg&$DCc$dx-BqzpE#&>nGNmv`d&UK8b$4J{NF#AkGcVl_G3S_NnGN?D?%I
zRTFsO;yG!b2SwtDyCRVlVa2}?<e1}G$}gk8`j%Dz$L}s)C!G{o8SU$UnJ(6~+f5Bt
z)|yIKe-m%1i;9@}ZH%1bpM9rQ4l0A`$=R0gOHH6-M!MhsEzCDHZVkERVn*mtU%|?O
z_ggP!+(O*KcH6~_z)kLQ2>F@7IQ_^320ix(yKLIU#8(tw^8A^1!~=6^sJ7`BGd7|g
z@6$UkCYA%UIu`hgnLvQ;6QRPdox3k)Y(^bf)d+^y^GYJwNMRGb0gQ`%_r*X_q5l4h
ziAw$W345T_D5a&g85gsYKKUF95zklcTBDg46U)*@Md?|Zk=Hq359ZNe%nnGA5Q&cq
z)Fp<IK$m$_n55db&DSHo#}uTwm}yF2fJn(dXb;9D{2upP+#V9@frQuz_?CyG52wSP
zboh~}7H-eP$Yzgfap2E>hvHzNejshB=U88)X}GR7C5^xUAJZ7<fDv1wUUA$xmUK+z
z>IBY$w2M}uA141Xo#o2FaXs{G(5kc?J-DzD3Lm|@X5XrX29o*VSFfgzksP|+>eMkS
z={_CxY7b8YQxJ>hNe=K@U!%Q|bQDCf4%pfl499N@(Uev`^|ZdU-bixn&kdTTv$08;
zw^2i(_s8xoP>7KA1<IE?*f;sY<UVT^9*IZqA4Kj4FYPg%bP8BU>nwqbZ6Dyfr}kJ1
ze83Y0&q(WyJ@y3og8PMzvWBHAR@i_*TJ8~=g8>qmxV`2UTW&b|Z|rMxDPq&bjID);
zyC)NOQ@4$&k2(x`wmC){Mr-k8beC(wy>-ZaKDNZ+n6ufoMysXOD{NaJl<?H~wq1Rd
z_l=+e&?9jR$p3mQ=g`|uArKR`XIFlDYPp;my9U5^#X!!rKcgc7Gn!<AXVn8fbA33u
z?v4eerZ_5%Jz4>HzI5wwx$adb&dzf6*mI>>l4^)mMDl%JNOm+0**<;JGt@~^eKOpr
zMOm^&9ni$ub9wyj1XcrP8YX@)P=ukJdb9C&DggJ2`e|e&YRBujrw^IJ4r%06HG<pj
za4aa7Vm0jF5iKbDyVt2;Gv=tHf(CZ7&yGz&)6xx#WXA)g9XKGJP<OfRCOa7|U9W=7
ziz~nzfljAHM7gW&a1%at@rrDsP|XPqQRjrV!_LGchg|kLdokn7y;fR6aT2KPT)^R(
zqt3^lPdu@}CN*a~7ZO`>aq1{~d?i*$yChiJ;ruVh(<<r!A$DI0Bzp7+(Z3M+sa>VG
z#*YF%C~wX_j=zh)g?3$=8VXT?Z1}I+b-`0lAcnh{f1-WGHIT6$Hw4j-el*uURg>g#
zNW==<%4b@arW|Q|KMxEpT0Cd&U)mS(E|q+v>6ZaBS7*4|xI@}k0>n)z-r?7p8E)fj
zOpp4%sVz|tK_eGMDkAxfh8`#$Y_o5*&l(0>kb%Do^rJhq@8i#9^~=oOOEecs=SJaM
z-b;jW)oQ!-y^Q92>9b_MyY0Qi6|ir*TH1-EdoNMNPBiG#-b<u9N18&rBU-VYU_iTU
zdc2~pG-c=BOOzkBmotBt+BJ>BsNVfv#!yh&*qv1GiN<g*@B;TnFWl!jyDuPSfV!48
zyI<eMUa$vBu()<|M!YO|uC<hZVCH)nfyWx#L{(boH&W)Tn94HhZqHWlHSpYX-phCl
zL>IRCJYj{~iMV&PA@ie2(uKI^!h-iQnjT5C^7f#9aSv)jyX>L(r4Dr?i|5S#!vTSN
zN<I=3XMW3S>`^@ojoHFjof?wJ$0SqG<6^<+c{Io&FN{>R;0i!iS^WUDb?2rE0pto$
zYU^3$dl}I{$yMTLvAJuB+WK(+TNyC618%jtT#5x0>q4y%>zsg_lMs`iwmS7t?^k4f
z^x&mJ8{f+~I;T=`SFDNGecV}VOIS@V`WODm)(K>g<_;gW{s&9pDrM%iL8B3CMJYJ}
ziS_~B)0#A<3v!T7zRABgk@<0-bXEtu!!Zc5^VhBF4u*5s+J$Gg+T&Fx0B{lMN?EA7
z-^;3PO`<G-+$pmV=j)jD>Wj7^qOAQS_v@#`)>Ci524Wl%HU#M&)Q)_N#T*WxMp~K(
z<?gagv2YB;EfQQK4C|P%g;kwsW3t|qPK75|dNTGD>?1hY5-&*i&}&<xZ?=R7-P^VR
zq<A1@9q6^~f)$@+>|5wdpNg+Z(9ml;gdbvvi~nXj;~9My;hZe~(?VwhUWv9#i;+e7
zlXzYuBx8+x<~>{5bX*_1H2{_wmTr$)lghwdty*kvw6617w1GCFQMlUdc?}K&oU?uU
zv<#!RM+4dZXo0u^ABR?7AF$iH>_zRRr4cK6Fp!AcjLqZa_cGyUX{yBs&q{1voYuBx
zdsT<{zt6El@sNrb*kN^x&qB_#*~J9<>GWt!7yi)Y&7F2EKJ-ju1v?(IgY^0$o+ts+
z=P6uI#)In_bM{6&;ClpiP6_3(=mk=4umQ(~)ViE&QYN(9oMN5PM#6}U)`J={vEo3q
z)z0ZH_i_Q^Wampr<hC7x^`2A%&lBcH^kNK0vL?6G_t>R)X7PwXgKfDS^WOxAb|oNj
zGaVPeRdtobW$^QnR;bY}bsy`o&2zq8i-{vS!?a(Iu3TUM4WGml=y5`SLrXv%HP6)W
zX@E4^L%g(u_*wMiQ4fD^p9>7fymOXc2&tF{5XgqP8on$YDp%ir6{C?JFi~@jh9u^h
z%x<bfC<<8sNHiN0@lwM6#kcW22={bA-$l<)%!LpCKKg>I?8WKV^VeL;s1=ZLOW$%S
zlc$$@`8Yj6cI>uG+5L;YsiZ)^JzBK4xR7tt)LS8yEWG1VMxn`e<sm53FC~5;RW4&o
z-WiX?y)!x@`=yM5DC<he8&Kz8N^HWiZdUJ}_?t-6Ub{EGgyNB)9X-AP^-M4KYgmuZ
z$4vHsTGzWqX@(x_P+HkcecV}SwOKJsB)iCDfYs1ut3@N%&>TIjgV&xXpnLH`?(k}&
z^DiZqsBy#=TuSVV+j$1;!T64|=(OOWc-2pVA-v*aYA}}nX^&h=OcIst?9upwo6TK{
z7Dl77QHbZF_?tXb2;l(%tiF5Ophlo@yq3y+sUsIidGuSxN>*ITXt!%)hbR`|3`M2p
zMd@y<3i#08ql8!)50nk%+N+Mvk=V-yORgzadnsdKTtDVU-8%i!1l93vS+D-<C`Q<*
zMesUWwnl4KLtY{T__g|>wr^kr4rrY|Ux_Jty*}@8AEphLGC!qUo~?<B6zd^U{YEuw
z#<|m);*m*lnIM^Kj6r6GUskJ{=P6NZ?b^>I#ih4Sy)MfGbGKe2SaO(pV^NB~Lm=zd
zwjdGEu?zo%ygm>U=q&yg=S>^b*kdE8Fz}vJ=lMPg8*hq7Upx=&9*d7tA+A5zX5j**
ztvwklTl_bY!K>H75nMj;ms87ntUlW`Mr-jH0q^m$KQcc(Ra$W*#=Px_J_m+-Y^SDK
z<_c4LI@)peDmrbKUgFrrB0m$)6|6wwR-V;}sY>u}HLoF+PP^^VM&)jFn;koEZ+z14
z!7*&N=Y&a*V#vAS<N4TIWU%TZ;~h#3+ox$hhQV;Zp6WQS2VyEi#h2{GSRrhG1v?ni
zi1>z=1txXP^7e`j2r<O2GlTI$;~w26-;~$wP)sG5Fx2^nFJ<y<s1@Ml+zo*0Z<@dg
zm>xXSM(t?8!R17_d%)2()Qg+y@pyq1kHT*;KsBgbfu6`t#<em`f%pJ~dNfWTPi5b*
zqsL2Y=z3b4h>16Gc+SM19o&v*bue9%GKhWtoHh*+60UGMA3b(Dc#~Q&-1OktE=EtN
z+3Zp@Yvm)mtcApH1Pbbm@)dpBIKU-iS7YQ9C=gcQj|594Hzb;K`?xd#Y`G+riy8lH
zrd?M<QtY1ayaJyHXi_)0lK4i97KwYfKt2uViU-<IqhHE|IN09Z$Qm9W@kjeS2CGB3
zB#>PlrW7Xy`%*u+oXC)0#ncPK<f@4;IRi)@H)Eu};qiaAZ#3s>jA$g1i-yiZcH4Kc
zZhgbWKVpmieriq#RenR-`xyZY#)74|<^4oKdZ=aIW4FGa(bGJho$S-w^uy%8xJtlo
ze?QT0cjp1%Y3i>gi%`5nKjr(nQUO7m{{BRAp&LO~xHJA9?il|&yDR?Y`a#xyKU42>
z21!v9j}0j<5WTtHdU+YU@BKu<Py*Qf?`L#@Iu=~I#HqiZh$A$7Ml|ClvYBd@l2kIY
zSuwL_Pr+Y{Y-Y1HhO1j_jvn(vT(|Sy&v+BJarSU0P{loP!5VeJCNGE&5U26`qkwz7
zbm11?w};-(_D6Ce!`6>*bjThSvXmzIkwD7WDBe$x#^2Sr<=H~PGx?9mZHwX+ELEQ^
zem~=}h}gj;ZHXpJCX1v2Dvw3>0t;PMp`NPT{l7ALdSd<j^;T7Chk<6mR;n8h!tg}Z
z)j|Tydg9ell*cDtwy?wM1W6-y{({wO`@oX;;chH}4L4_N1P3lnzqGY_Z1)l>>)y}g
zm-@c^Kq*1Z`amjWV+S$bZ<rd5%YD=~YBaX$@945AU|a1q>MbArdp3RUcx#Q#(M>kl
zb}h~dtjwuuowF;Q?`I_6FFUl>rLf%Yn8W%1e0Qt$#@~IeW$KR}bN!<>AVj$zAU1w>
z_+Y?JD4bti!<y~iI91!Ejo>VIhtr@jA%N?|waA{PYO_F<Xzi!+$ryUWFaF%N2nju~
zq6lJJwTw$w*H^EqYqV`~LaM54yBZe;VG1ym3vi<NxikFU+aW;EPYgv-QrjuqV}!uL
z=4m}s6cB-amp;mD4s7>l)O!a7nL`ABR)|=N?%j3^S`rs@kK&lVr-V1)!SKB|o}dWs
zp7>nMyJi^Wd91{svv&L9=VpS=_iKiKJ;aqyraAq9mx1Rnc~L+OqE^KO$S4Q(8{s`3
zk`IIQasXO9=$Wh$rPYs@2u8+zRTHTr^4t#TeYdOny6texYz2<zP(j~dM`C+&e-!`T
zj=rC<%q&`O!SzX8F~>?1BpBj&%$yWjp8M~F@at^Ag<qxX<mA-uD3#dZApF7J5M<SE
z$~qN~BOuWIN5_TnI@jHCK}_hE1|)3fa?c3T>S{Y1$UzB^h#b19*X>*Ym-6mlK-Yjv
zi!VXpwKtkv!DSbPU2>Syy_?f6>9>9^B@b}IkRF}l{zz!0T~ROivQmEjYP<z|188<W
ziYJ!KdEbu%`DMeNlf)H0*W%+<Nf2S1UDwBrBS_zZw`-vIn%&U+y$G-KD4(jiSFZa^
z@12o$n)7)HC8S(HwR4CqkN>+mMZ7v}U&f;+;&IO7E3Kbr?1rIpBk9-bxU+b?+pV!T
zwTnFO4J+|YtOO-kIGNwZ0|96}e5a<U>HHLYuOHkRM3Gzy<2oJ2VudH&<&56ui%0Wz
zs~!n`#+~kRcKIfC;UDo++-kR9&d3ZKp5>Vq-yo|)^G5BC%b8@~SlwXL<B4tQ^6j0M
z6O%z28J6m<%Ndiw!SedHck4^aM%q1kszWPw@8yhR0kskNa-ZNR?`tAP<NnJT3&7o(
zcb}~XVmLH#lxvxx9#Slgvq3ZCJzM8*hHTd5jIu8$sHnYo04{m<RF~9mZZE!MbA%xL
zGGPY<11@nW*5=1+<d+J3cx{2^P{m#NpeF40bN0eR>I{c&gmb+w5M~OgdL-UMGu;?z
zj|RBq<fv-5g_jfWv#PSv775By<dH2_$G8%=+Y*7{zToB24R;S(hzhlY^BA^DA%Wsn
zIB`3Ps?sL#DY~Kx-P}rz*}z6|UcOrR<-$VO#ObPy-f&b3@!c=GI^n#c#_BaXK?MI`
zje@g&b*-(5ZJO(1o34$&n~4#&bs8EDF@=e)*LQi^ZV(!s{Br^~#>Xjizez7$WrH_C
zUhoX|SZj2ICEpjfU5kr)5r6$o4LsP<iG)Qpm{^3L{^Cd0D;Ua=CHwU$;uhTbfIf98
z+m*P3YEOm>8xBO2EF88;xJ1iAMNKZFF#&%1pGPoC*k=7u5zZ@5PK|*E+#~X~2)#VD
z+g2^ECzOQvaF1+LlVz@0+pbPA1GrB0sQ?rC^7RYbZO7$|ef6!7Ojj<i?ku6L8)Z8w
zYc%)jJ{JylX)<m)^}}n=XjCi(#H_~`(2+yQXLq-HAfyee=$^}omB&K7SABUr9K0s9
zKNb)iunC`!1<8?&b6)_3zRzPP_6u^zN~##jX2gFkw#J|b{=X=o*sZO0FdhlhC-Q|r
z_2}8bc}3uPvRj<v_2J};Ozuz&%${+wdN`iAW2(t{cSK9Taih>rlF^0p$6Y@|gn9<K
z*H5?O8jVzkIL)3=bE>R4i*-`#UNK5tan#dq1WHzf)E*BcU!d#pm_-X26(*t|Sdvm$
z@R`dQ=Rw9+7C-9Ro{i7L&GMXJ@v891?=kaAoa52`+hn}v_;gbt1_Ce0V5zVtLf&<k
zAl@ZsDZsrPLx7^%db=WIa=}s@Cf=czpybR)0rCp2rA>8yv3(rN=aFW1O?ZJ#E4~12
zFMZmVM=$Y7v>fpSO*gb8u%7m*zH_4(*TrZ0flWS)h4*tIhE(sRtj{>etJ&ngT-b?0
z@k_03mC8(frJ>xd@$0}oXU}fN3(s;+R@-mlsepnGw0$eI;q#p2_3!jcA>j>|xaex`
z`nhHwWb~LkP7s}2KFIV#$>o;(454ARTR+Gsb6FE5+Vxp`@hF1U?RtjZAT>|Zb8Dv;
z8Mxzvj6Rk)VYbus4-zK<>yJdQ_Aa*&@%BODERu;H>U?($?wk}w=I{9+Bj^2ygMP=^
zz0sas@VL|F`_!EW9Tn{U4>E3{cs$iFc_7-?m#h~v0u-#LM6btK9k@DjC(QaFF&zbD
z&7a$BjpBwmn-gETCEw;nyKb~Y5OM+7HA2lm_p@B^L3Tcp2OSkl{Q7xNNL`7(;-L=`
zRsPBPU)aOZb@8{DI3E%4jkyl$z}ll9h_P~e^zRefzVL(WcIED1382u!+&qgkLL&m!
z;tw(|2ie<md)pFqKDie!ye(EP)YYQG=(dWf4dHI-RLpkAU@{3*3K_Xg3{<5RUaG*%
zl`)~N;tN)-aguk=pw;U03T`H=i(#QyJ<fhxeW1`O1z9p(b6}0?g1jkyEZ0Q$p8RBO
z^oA~?ol42JE*|@Hqsjkwy}oSiNVP>a2rFe36}B;+JWCvr_Dx#6u<+7*UN~Vxq6BXh
z8q)@4A=@>ZY7+2Rr#_&#2-~w;0FY$f+1K7!UtHKw!s{imA?|ZDpw3p{L2H8_Bm!1J
zh+KTtuxfR^ZHf&C6NANIOvo>XpCZ*cywYaDKvg+N{*&qzq>?*6MkW1BK4m(wt<fQw
zbx_J~%eJ#azg+;9r-Wekl)5HxmGs{`gpZcqPU;rrZKqnSgavD|r!_e2=%DS2N4L4S
z1pbUR^m14aZU|SUJsXqd3d2GkD})T^r+be&YbqYaBR*iow(r8cR*#y;oJDr#^8$Zt
zq+|S7W42E}JmzMgUfVD9rF3Vg5Sr|OV1g|n8imW09JHEfjT@dh>IVZ#sBxdY9Q*O7
z8yDCs0;3hMU~_WzYP_WapB*ZF)Z5Z!hXoAd;232`qTR*?8||pDNGAkZ9E)cTp{sIU
zv*SVp-rW#)b=5!&3Kzx67zfAYf3Y`oKO{<(sy`DybxPA@SD-X-9~m2;8W-JZk4<Pc
ztGc>#?bKScGx}*wHy`zIxt!JTE@pz8bMZq`MI<}!dEwd5H83I*O->!f#yYzglW)Px
z{E~1T(G_Ht*Dq+9*a>*cMf2(vOYN#~f$tyXU}9|4K?9W?o~r%h7{|p6Y~R-eetAoG
zTMD?o9>}UAajkt)x)B(M4Rk||6OIUTw@>3t{<kOWv)Fy7Umz)geJ&uX-BkRA9v^dy
z?w6tg_-5saRY;lGzlsjHeK=}g$H44d4;j9x?;E((?Hj>CQZ5`uISStjCCEw1qWN7+
zV+aGDeXkYq%(|%Y>`F!l`}<P3>Md6i$thqWzV%9?4Q@#COOyR{+m(#u*RoKz>*+C)
zpijG!=whyI6kFRJS2E&;?Z3Ta7zT=Pt?5@14Mz{|0hV@W3|0K?nB65X>b-9GfZcsX
zX_fwefvFevp!B=v%7kkj##&grdjk+>_%Hm*?o(fF*j%^)?!S^zM<jVJg&r489*D`o
zANbC5L2!nCK_nd*#ieoPmBdy!ede>IT$>eBSjtxQOm4F^)>3jxK=tQDC#z8KQ0d9}
zg?SpCo3pQ{K#@9J>8@nlPebCs{CZxX-#V~FwUOrs+k4CmPg6V%?4cMOzp&(>{cvEs
zHRthVj|h@svLM1tKN<)c|5ZQv=;A{r%k>8fAdn{aL%ENmb9>zuM+c3pH3}4%nI%^;
z{sH}+bKJQ!<#P4a(}yvkLgQjf_WFL6`T@j}gwm?6B!&|Aq@>5PmFl3{!w9MaPO6DF
zSZ(}0>Wa@gHEY1wVD)<BVr{a<*gH*_uGVNk7RvQBYxRTMF>y^;7Z5OgHUREk>(v(l
zVkDc_K3-8?V;l9d5l+#X^avC=%ST(^q@3T@7>R`~rKWCISGB<0si((CD%35oTo=g}
z?A0NOqZ$>d`eRjHL)>Nq8okvi;Giago*Eu<IG&xJB(`RghG?L=CA|OGmBcA*{IPA;
z?&h|pj%>l6jNJ`ZaJNO*jCj69OAZ%FNG-+z${X2MK|v{m9^0l-Dv16>uf4sr7`R6~
z70?qG#`)QyhD1!a)BjGv@2UwL<pM!esP~hyOU*|LW22z=ndl4%A-~VYLs@xQ1G(m7
z#7<UZPxRp`1$1D01!}UmKc}6Z?0AkOr25!plYc;kxG&nf;O%E{zp&ImP&K+KJD^kP
z{v2>SP|+8ogJzGKJQ!fDsI6GG)?SVWchDpE?9|w9iHE78lGLlw0(uBZIusA&r^fr{
zur|v&nCv4l!Rl3y+fl*fQTr_jKgZ%bbhR9><61^6*rbPNWlsc}Q)DmdO*<K#{n^|F
z_J(l$cvy^1X@0B-=h00w<3i$ckARxc2>4>6V_<|c0(KZ(=vlSL1_6~1^bEU_{=6D4
z$Kzu%wqt(8E(RRP-X+b^B{h7E<9%5@cDnE9mFTg`IlrrVRffXCKAPIHM@sdR-E#1;
zx~pAXW7qUeA3pVVJzlcKM$ue+B9x$Ba>n3BARN_F@-sz-|1`$0URH^kUVFHa+>a>f
z<6XfWn!m^vhO%V#r5X}4hnwbCLgR(MM`A*1vah2_1J28-3h~Y8YsksbH)@^)EZ1{=
zD@YShUa;?CVM*p<-^ZU`5zIbJblbC&b1FE-cFTu}kdgT{d2)EH;H@8K6dLiw>shD$
zf7^$Nh!aPEdE)jD6Nzmaf<c`YFDS2r=I@RVGZjHoE{_dz`iGg;9}7#b-T7fgIQ$lg
zMZw5;SF}RSmUBJR-O;<J-nV<y!E(QI@6{ugqCvZ_1Qe<za>r}zr8td|vgQYExkF~e
zOt~^#t>gKCZ03g<t>yNoNG2+0n-yK+u*<5={xIVYk}ocatx2CT=fjLK;&${yF;6{n
zce!Z<ss4ZMneDJ3rj;W@kUjWe;-yHMY!Ahn0GDfny=dUWAEv9hywV<tX2t&-L4FF9
z=U6>#VIXc{>HVv{wkT#F>;)ub@rQ{a!>osm!Is3Vuxc}_@<2@HC08K;Rz!X)tBk&o
z8&bMTtD0IiSFEN&Ang5>(J7E`oNd)H9uJ<tqE;v=3{g(i?U!{^D8VJPmm3RneGHAO
z4A)EAW{v7*7-x$$TEa#*7O#z$6&}r%w{>b-zpT;L2X;|xVW-y)W?fO=hBw&8KvoYq
z>YB7wh6dS~me)9~#iQhr)d~Y=k-B;9`k?_61K911(He8;)Ve>+)L@gUCh9cDdbMPP
zjFLwQOB@oshU(3Yjo84{B+$+e3NFgfLk5QRj0_18+9b4k1e=>|ZR`hY!-BaX!fOt0
z7Q}V%XT)bdsjiwZLpR%&crI^gC+{^V*c#87`@e0A8MYvF+jf21lk%KC6_}i}V4m#|
z{H>@ec9uTkblKCfJG^c=s(JN51_Q)}X95c(tbyexuT5YIc|J+XI80y@+Z~;?5x{JF
zO3-<-G52%Yt63wVdZ>j9xhL(p7}IGg#5_QQ_xJi5`{GsIKwn?0?H6|NhVvXS2XqEX
z>SB8_Mn@h*aF3oK0e$2Ue0l0)53_=tYgLlp1~K0DYJ6RM6~Xh6=3ehA5j!lD6<^w9
zN3<Wxhq~MBsP+uS**vDgdA(ygI)xU7NcNQD(d%m1#pF$NLO>w3Ldqb%CpAzF7uy>e
zs3XsHcq-P=6)@u(GqIl@R{&{`OX$z%+L@RO91`?A{%#?6qMZ{SP{O%O)%gI)k<V3{
zv&k4AYU9<XFUC{xB?N|oU5byL`E!NIh+P&C?g!u3mLfK<2<EC>Hx_tdvbDtxJ|F3s
z!-6pLu^{poZTS6brTsvPh!(r9w;rd$=bvN~Bevm&hM#;M68yT8Nc&VjBI#0L1KthK
zI}N%9ibPV`Kx7RY;folq4aZ#S!>QehLXPaZU&UNp*Xj)Q*IKWXA-xw(!cDE$05x5{
ziO!nI5JvU(*q@mL1SZVyG@A=gCtKF{A7<V|RO1=Q;m`o-#!PNkGs(P*%<{KfO@wF8
z%=va}{5f~-Pwlp=8JR}7AO&4~lKA}puxWaU%BjolxEj(tc9*<Oznam)F0`@C`p&DF
z)|lIg3Ic6*SF|r{@*Di_s~La5f`Ytz)VwCwiQGh#n|f?W<yP%J{g9*52Iueos~HnP
z9dwRsN&bEy#u*xD0<IZXMYw}g^K_;&qeXoRGMP2iA{;3Fl9(MRDM$S=NAFVw_3-YT
zCsg#4%hBd*HjVD4ZVRqv#)jex>_B%PdQi~!<9|)2iigz2AUuXWtYKWV=RR?U;t?%b
z9>L(z_@dC#=0b`sj7JKclU-%h772rHO2+ZW7GKS{8dqwfV}Wm4Qd&`G`wm&T`mbQy
zRm3d*tn|++)gy8{c6s#m(H=(<LR@U6W`OWguxbsrs(ZAjpWBB>IL&H#2*55u)kQB!
zz=8B@6skmYctOQReMAZO{>VjSYc$xpTsyjZJ@)3v2(?ZVDIaRH_3^R}wWMocHe5|i
z>V@C9Wj+?RetEey#V%OoVFZ?sHfSY#tyL3Q>WpZrK;IsXdy{)>ry7vqyPM3qbkKcu
zVtQnWSb_C{N~(#A>5b3OPIDysHB<Ig0XDz}^h+!10zV{!@j(SUpv{I$UpGSlY?JmF
zK_v`07SDJrqE&`+Z59ghxbHlv9q1--XyPH;BB;Y8=EH4`PB7bf9fo7JEzm<XU)!!#
zz%$<keX>H>Qvs3lsGDqu9{y~u?G$+BL-~BWJ*}xZhwWqSQj29Mjfrp55|(?ep=b4o
zl5D_sYdQ^V%{}T3*IUR+-u7zno?&|~8W(?yll`DQucKB$4oOte`>tk;nJd=EibuJ^
zaacf>*C13?vID{yu5z4F7|0ieGaML7Cg6jb2gY$C@L!J4L>X`|ctvwryJ9U4dg^#t
z$p&qQ0?|#B#yuQcuZ<<QBkJ4J2T;ddUqaYM(LfN~uD&tbwA=BRg4@(NpeKZt<#l#4
z9<Zu|{Wk)GDaj8a8pq<4I_f|Dn+_WnE*!lR!r`iXSB_NCXY{(J&}L@^XE~;qHan;0
z<atN0WaqVDb*1VOlllZ-C0N>vnwqn(5P?y<6xdFYo4<6+I2Oap`iwG?T*p@gbv^2A
zyQ*>UO-!MSA8E-7EtHRfs~EM9)k$BW_*#=)6S#hWxx23GBj7LX*H1KK4y)aWRpKTk
zC{Eqs7=;S$Gj+$r+<^MI=7u0xaM9hqxSC|S`qlQOT0f2m&3*Tk=8y%lejQC%RMJ1W
z^V7|Mg>#6DpM9ebA^`wt?ld4V&fyjJT>zN`#Qy=&`(91J>j^i}oBdbD5saWsxaGex
z+RqfIE)y?@sQ$@)a@&7p^qK%wg#Fw9E7*s*PuMhlSLqVp9cqc6Cr4<yUYB)n%<t6e
zWfZ5!Xs>57WMEOa`~RiteSmYm+O^T^bUK|*$7edmn2uwN&#}E;pKUvi<2a7j_OGhm
zP#aa-XWK%eK}2>Eiq~uMPeepS8W9l$5kW*m#3Le1f~~fbrao0w)hAE#JV`XK`h1S*
zbUK}Lt>3kNOTU@S<a+M)?_Pg?Yu)Q!_qvz+y&$EfJO%^A#_D8qIUMh)*p{=282Ftp
zV9<3Io0#&zcm8_kDNnr#c<^^R2c3%{jt}#Q5Qcu|Eeb7XsUIBnoi8M;^KNK)p2UW$
zsMX0EFhZG3`3pw67NajtjMA?U&EaV*Mt|om`Z7=Fn5H@17#E#S!eXrBc<t&W9(QcU
zI9Kp|zRJqYuu;v8SHa1Yza}V4PPbFNGZKbQbd@v;SsB?+QeC1~uHmBpohOx1x@IM(
zC}8!fReU1OH7vzfN>4Lfd#vKsBg|Af`buRV&e9L#P=$tfww4KNjOkI7%VsKvCXeJD
zKU99Mv&Q%?Yigvam;cU})4W=wFdMThFH}9FX=%MkeBM4l)$3xHm^$AiEOF)0I*0XN
zDsIe0SPtrC-}z$abgqJ&RhGLJ-`G-t6)Ivfh1?ib$ryg{8CaJocj8n!#<`y1MO69}
zq+VXpdB(!oE#3w5xGX;Q#%?6IS{q|6y_=rsXmokUy^EO_lUxT<22Alx1yaP*!ddTQ
zqYBx=3p8|9qaQpDVJ)|Ocgso>-cFJ(@{EkJ^}FA>->c!0ZIm#^6xuA!+gwrd?a7^1
zs%sr{uFcL$cdfIQx~ELFI^~xXWT}dt($Odn8j-CwnjXh{pXTYQ9G4lNQjMlkG*=a&
zISukGq%Bas>%QbQyx~yb`rgw?Tw(z=UuC1vwJ7z|oNd;l=yf`Cj9Bb?h{6HA-<Rk&
zIin9M?s`gJM>ltq4!h{a_b*NH+=oR6h;(f1UCKD-YT$WU;PpAj9Cymp#u;rC8^6Mq
zW&A0UT~xZvd923O<h0sty74z!B^=MwU{t#vaMs>6ulX=4A(=DRI^Lr`ocm9xbJ@={
z$Ns2SmV7afPsX`1^#QFT<D+@42@MK2ZwH~#rTv(vGxRgBj#y7$Bi9`_3#i3zsbtD#
z7n?`JtmqBL9k0l62{wxjZmCEv5dp8ujL)yz8k?KqQr5-ejs_9U(|I8YcU_~dGs@Ql
zw7N0IC%e$*iZsU}DY&PcrmZ*ruJ60R>2%ai2exihc#l+b-FqmCePoaS@JQJwa^~Y4
z@>p%L&fE!n;+l2FYKn|O1+7k|$!*958A%+Su4(w11&_2j$wic{r=n;;x6<;Jw`4pO
zUcd(y(CfNG2fk+C^-ST>Mg{)dm9x<pu)R=ar*1LR-Ah+lo}KbtvR7{KdD69iG3)2=
zZuu@ajC!}h2<GI<JdyT`W7}6k<5$O!XP3{tCk^#92LAfjKLL;ZHPSqFV<x0G9{Xz~
z-wHPJ1!vqhANy-N6-s>B_pQhN?v>)qD&DuEdz;7p3Q2t=m(N(E(|^as<E0wPNEBuI
zh{SZ>A-XBC?_*8Tg{7gwRQ}=J$G+4N#Z7HL70%Bzn{vrT^ndJ2A#2TX$pBXaUZv(X
zha!N@kG7khQ&<diDW)4^o<S~!u>v>8L4zH|JjQHz8X^{Wb^lYEr3`)S$q0YhieZmE
zp}-h`VYo|0F~!@}k5EB;(t?5wE%9Ab`o43h?sqhv<C_x`iY23!>9mhNiozII_dnAX
zZLC75BBRe6g>Y>)>X7lSaC7cUgAz=5?5*n*1-{yeiQ>dq{4byl;3WOz`H*+#ZnC1%
zB7l3JDaw}?mz>=&&9#+x$#RBk%S3wVB(F0cThopN%yOe5VM78vLCjXre;zRuQI5Mr
zx~DNm{Bu>x0#D+`JeTbZn*DGuIsdUQ0dV@J@u4|4YpcgXHT>d5SR@oryL%P0dMz*F
zIc68EPuPs5inH$DuuK)?%PhPRz1$TRMW1h6C095Cuf7{mFGkUg@l^~~+3|Mit6cNl
zA<o67cMPMvT;q7<GH%3q4hGJw@v5QG{c*@7xbK0pvV95!iH=`BqD2dQ*MOeK9i*tY
zHyEwwMnyHsNgiQflY-XsHZHa($V@QCq2JcmCGnwTdTuiI65AAi=1ltKPIbk65N8aA
z(p>^h9cH<tMjV-H)db3YS!xY_t;{If)&6~*SLL{G`ZKz{@jPASs<+)7{9ll#+*Z+M
zJzvejVL4*tNE9gGPk;Ff3T-<Zhm*#CP?0Ms-W;jX#NG{}56o7VYek7t^9w+!>jWmw
zGg67C+C1z^`!iobFLO2L{V@*iV+u7|e#$T9YNG_c<&6p#ldr~bA*^&|uj8(oM(a4O
zR$)D+&qI}daPP*?yxM)zPAQHWaT*`XGkM!xMorCH)hQ`o>80z``rjqhJx;FZu6yj6
zhdJ_Z(9r&`|B6PXH<CB?_F~(c6U?-AUBxblU592tYR7TIRc1QPYmAJ3%T<@FDrb}1
zE{3t(G@fwpxNe@|-<7_riqN5z=LoOW^_MYY;SKFJg_|ePC`{s>%7|G%(|f)T_g#)L
z-pt2&_krpnW2%QP+<bOl8T+h9jzrsBI+vv*k;kgm7An7<xQ6mXxXTot0thNgcGJxD
z2)WTPmHxszT~PB;L|oRZsH@!2ax@9uk8fF#I)-;I%mI#R(x<8wGc+y((5peS)OdD$
z=7KneXo%+;b-MBQgcl0_aQf7RG)%%v7dwV$L3rhwkISby*7#XDC-cMszqmkSFvQ*U
zufOXcnnY20`OQU%GEQq_fbacvk=`gi!yDgwizo%3wOjz+)GyXz1~q|ixnB$T)XUr7
z`zyXt4e&68Q{+3}`}6S}PR~<m*y0^U^!eTw7v^~0f8x>Cl}abHRO&0jf|bUy)qDM<
zpUa7Sn<<3z3MaJw-+QZ-c@;An1HSj=iKJLspBS_3_rCX*+&}z120CVEv6h4M%RJ=A
z!|K7``@1`8y3FPLAucC#xx~8-JJiKl!0U4L2^i)ou{NH%DYKF*Q>igA!d0p50O$Xa
z-}}PPEY5eEF-pJrhF;2Q({H0)cF`L?HLtQ@jLU#ZLwdXU-XB*?PsM49SuZf|dtc=^
z$0))nAsGMJWM;ftW+9g*F6u7xKbu*I>b!~HdkgT~_dmfT*R*_}-kjv=V0^MGo_p0z
znBsm?_FIW*N<W8fJVVv?JiuSZOqENylxxH+`(w4y(a(0}nCm{qblwwNq7-KpEqr;`
z-nG(NZ%aSVmAb%;r}^LeM^1DTl)PnaG+$6{x$qely3YOc`Zz2SZ`+0~#p3U88Pm+>
zE3`LSB24il7E4tqtHN=-%>Gzu?gA{=k4>A+9%Y4oaJ;YM^KgzgqtxXJuVUyf7I7M8
zAAX88ZsbfiRWw_Xct={#j-x>&pL%!wF=w;q$q9+dEID=+lEkOy{xiizT+PQ}XfuP2
zin!>*1=!>ohx^6#*y8y6n4d>Hwz}VR$iltqHbq~bl*miT>yYXO^gJ@fZz<iyH}}0a
zBh&R47kBRby*oDnWVyDU+ft5fF=(#Odi#fq$Z=f7@s<j5UG(X6$&;rBHJUV@(Yu&Y
zW&>2<+No`?r)y|Jp{qX4Vz;bkj}(3HZxB9sZNggY%7|jUOI)_|%u@`d?kBfvG`Lq+
zEnJK;wPZ9`(__lOsCK!7Di@Pe<LN&K6$)Zknk{>!(nPUhr^Oqk8@|Cqsh2Kql#ZTE
zxms4cV(Dif1~sm6Js&-&bu7P6OQJg07nG%plBM1e==a{-0tmR4pv^lkV$jcxhDkJi
zx1iBQr?sw8hBmpj<oOU)Le#=scX5n8Fq^;Gu|y7O*>=Mf!FOL`jlZZ{E*G|oyxHA$
z?N6Ci@46@Bj+=y9P9!D&u0`&;@AY+Warci_)&8p;U+tyoS({^I{W^2g{hljBwJ`!U
z?mMD6Dfd+Oxp?3jnfLl)@X+yVY^04>3OsV{JcD1%V|Cw*ssD^8t~t!H6-}FhF0rwl
z@gBnssYAVY&Cuz>=hI@G6Bw=YTpqlfMFkqV9l<!0p*;N57WmP0^s2w8rjN!mVOj~S
z#B&wGIp)KU@xoPTBNe-N>AEB?dIjHAO28|{ZaKN0%Cw(d!_heor`um%7vI2P6u_^p
z_6vAU`<shS<0dY9c;YX^pO`%_#gaFkc#A(50nZBd%_sg|@H4LVPyE%D$K-J_bdQ6#
zpLoK7(Gsw(?>zAZ0CO0@#fkD?AD6?-Kcnvxe+lG<nA(zeUHRU%WKM4Vo_MR$hU69K
z?;>+@YiYm$_nTfVsQ-KKiN9ulM(s2|&4G?!lo%Au2R*rE=Kg5&O!i>$(Q<Snh5jL`
zrIE#qarscyVg<9|HgA}#1<mug<ucla;ZHm%V$Hu{gw2x{BN*wjnHkTAJ4U@OpFV}z
z9;0oU6=v}pqsSl6o$g6-V_ns!MaLz?(L!k4liw}xG&|riIy+>%tMvyh_5A23xH4C6
z;MJ;$u1xRNY78bh!a|ci8I#qb)aqf1@}iQ`?A8L9=JKWx#A$Da<0*+-$5_6=J=3*C
zGB3e;q5)>v3_1K}D+BM8-n_FN<(T6)&NF$JACF1AwQHVjJvJ-mtJSIcqoYkMeB$4+
z;5%)6U4|+eGlGrVh>K)6fJKVVJ~O7niydS0lISYK`&gGe@nubOY10@jvFoVEcP+xZ
z&nwq%#xggKjn;{$<H=a=GM~xbUg3Vt=6OYo>f3URZ_lYiS5V4|N1V%-FAbRse`{QI
z>Dt135HsGj2gf(xnsF)KU;N>Mg$t1AIPp$)=ne)+PkaG~YmsNCAYm-U^%ZX>ai_J>
z@%TQUM&0BxT4cOr83Xn$;+1zbSih|<_Dp`0+Z4o0+FW!~9aAy%SxYX{9f9XQya$9#
zSBnL7wvQ}VP2Pp%$fhfLM`2&AH~Q5a*FDA)yiP50)%Pon@mrqDa+UGTp6`m}v~5hq
z3S9V>mZNJ?=)#R(Dt;A3sxLd#S#uTJp_RgKtwf*e-;YMAsGRI*o#&Y1`bWOaldxs3
zVP>tPGukJ&`pL9nK8+#0Da_l0@-n>!!HQ@;K%*=vMaG~~d~$eu@2{VBd{Sv_thcMw
z(i_Zk0H}Wbfg68|6tm4Gu^Q#P<;DTjDmLvljYQz`(j5uZtK62OJe>?ECO?J+GtuDc
zC6%dZphlP4Xfr7|H#x!r@5U;wtK5J5$A3h#8(VB9-cP;Zew){|XqabfxgxHmoO$1N
z!8|DA!hFXKKHqO7?uysdn^^5u;ZMyarZ&e#A}<V37`x{>dR^?6)wusUoaQF9Q-5GH
zN~WKqhr$V~*;J2QM9#EnT)ZE<jx<~FCu(Ynq&%{s{*Ci7PpN`#f|z5jJn)2#w&e*3
zsU_A>eb}iYwwQ}a5mzmv7NAL15xQMVa}T<T1DP9^r<TJtTz-2USKKj~?ea5y8J@6W
z4!h_h$MHFI)KI)|3AEG2OYy?PK)g~jn}Z?Gs^(|+mGL`GF*=N2Tu0EQ4|SuziWPG)
z7{9r}#%&Bu3Cw8c-u?%FZ_g(o*WryHyake0UD0^+2Y>l7R&f3!7jOOG2>{Z6zwJ`5
zGbefP=qDW)k)0Uy`N7w=b4P~0Kln>M?;G$W<XxA&E>UxKqe$BC2XDFm>mQe){|}xl
z;`z*?z{>NE0pgCg%(>W$4-WX5(=hM{f90Q<6w5gXgOrVVt#Ggs4W0Oxp%@|_jDnNr
z4{I<~VO+*&bT1s|Dd2zb1)G*rbQx>D)e(+ll}Hyc@&|w8VGe}NAp-CBjS^o*5%noX
z|KQ6%(Mjg4(`d-XxKv!IJOd4WF=Jf?+K!k*LyU8+MHj0ahvU_>3*+byi8{Cmu4zBu
zIy2GsiT5JIB-bZAK25=7wM6`l?U<rM__i*2GcQ?9b9G9lX4!j=V}|26YCVl<csvk|
znU0nm&B?<nWtOAK)3&*|HrvtO-xSNYF^sM*>IY9w`q%%0xoY1vX8NAzYUfS7nC}?*
zh*mD13-^Vt8c`co^6tEAAfEMc6~tmk;KtF|E-X<+S8j>MQkR`4_ZdGI%YL}^ziJ}q
z!*S-OpQoJmPPmMvtvPF5E(SlOJcbpHK_iE*M2zDmF^OXHDp&ZYu{_pKL7c`aN28o0
z(|?X~V;8SSyz8y0lyMW>uMPCvlIVCPGsc2EfFxJ=RBEUB)SPPe2}O*J$~Bs|zpzQY
z-crs;*z%eqW8=vM-Vs<?9@Jx-iuG=etw*XW)-;is7Lo2qG)Sk4o_dN*m7K(NF-s*U
zuHzh>t&)u<j_=I4s?(JKhd3|OtLUi=<>ei&XnC(&&wR%Dt^wj(%2D7LwVqcCmAkoV
zAB!T5pZEU>#Uitb=gdKge(-(Cqo1Nw?In{0PRfU0=g6C}?42^lqW3@ur<r3vyp{}&
znS;?R@sPFLMWk7;cPdq(^5YUw>A0S^ijMJIl@^$DzA9H_EFV(g*;ln<CrRg6<AS}<
zyB4*si=#g?d(t}9XZ>f<bP-ywR{xj3pNxPjW~LDh8q^7GJkUp@V$rb`nnX5+H@NP4
zv8{~jixHHY-Ap^9rI809@qB%96>cbf)JM~BOX=4qQ@Fh?PH3A>|C@Ilo94a--;Mpj
zS%Gjb>1ltAJkA?QWT@3mY}_Vs^GmHM58+*l`X;gWRPKmBEX92_6344CAbj9Br!Z}v
z3*zqGRi8e?*5Z+PGg^H<9_PA{LXL4RL=#?415aF%=or3m71U2-FkxoHkP4qg^Sw^j
zpDCN;5&0eSMn|`jre^4N)nBomP9X8rwi7+gqStZj-NZ|!VaPMr&SrrzU2kT<=dPSN
ze3I{l8wu~n7-<<VRfgGhcy}mXsSJwE^n2$bx0y%!0{F!-M3?JSef{cwd!~-~%_Z2V
z`}PR>+x5ls@J7&I>z5ei;G02zd5&UpzNKH$>ynMx(c3{^4?vqGZXW3PF6b|1R5{ZA
z!ALrNg5F|1b-_aPRYCJ8!EzU`AI4h-t?cxJ&CmT{|DeCH(sq~&5rwS*j;pEWBFlU3
z=K^kcF;GQ}7&-}q#3AP^bF4Vn<wWbe6khO2#t@g(_?E)Zpf40~SN9o@ABP2f2_$B9
z4A~s6_${Y+M}b#WMkxOc39K%U*j;^%*AJSe@dm^wF|%Sl{o|-s^t{I#DPu%mzI-{x
z25%V+gYoc)acV@P2;#S3goyDf!ZX9?OYn|Co(EB*g^5Z`H;6n~R{=(#|7T2Ay4Cbk
z$IZ+X@qP324I8<<V4BN@L(WJYGhF9+vI8f;nXf4uWb8f5wM7cOu&=;u*PQgMF7Yx-
z`NnZa#mzfAZ?4|~Ttcs5o?|dM2J?e%VxV!q5sf+bnCtE4{B@zrdnJ`C-ioluwJ!B}
z=3txO|Kgx8Z_3>n8;l#rC5};EXQkR^scoZ1+Bhx~Lu{k<#{NzWt=N!co{Mz_O!dxh
zjQK{4i#})JBCJxI@L|+B>~G^NH_nC6+MI|ru30ylmnGLBUcLOsKQ2XrV=9I|xOt&Y
zzs(^yos!YGUQDJ$(~otEa^)Q~(>EJku8HeC&vu)X>&?9}*rI;*Ob0mHwz{Z%Yjr*C
z6Slc_G$+&E)Sl`(mTkya4`Pt6R*2nXw11hdkLLavS+07<PB>eIF&0&qT*ZT)oc4Er
zK(3;hZ7#QCc|j-RQbsbOFSkwku2i0$e!70fG!(eJKV~Hh9p`OjbkdZDA{EAtGUtFO
zcClyj9KS@3$K9{)NS3<P3w442u<Og&vF3qiW4K(V7N#`G!*?~|Dqh<y7nv^csHPWr
z3%|t5p#L6_k^VS1|0n4Wrxi0gZUd^6!D?RrK()vVV%DNY6`;k7rxdMq_4%{8*;40z
zeqwY0_3q~ib4N7bYWrs%cQm-6!;K9^jz(A8HQc$dW17^<lsnCe#g8Bv*KMtgm)B-B
z26v!l`)G7wmvO`8X_UUPxTO-!CrEMVwo=cT`r%yMv8k!>^R}SoL%#c(UApZ?t7~nU
zJ(+`{HdnwxbIy-@ZX_BD%B0oY$lq6)p5l#bxf>x|+$pSbukKobdpSIE8PNG2H=E4!
zu`7eeU$k_8;(ou+mY~~uv8i+97=%<SY6Kh6>Duu9<izNf^N6@nYn?|J>zZz5FqQsQ
zIM(n~y-T$>di9xnYTTmZnPW7G57$3;zqW8M`NI7&ud}`k`kJ|Ed{}9<u{+}<$kBN9
zx^l*msyXocS?xi)8lK5}`<q`}tMh%!B>Wom1=6XW*f5`tjCg+Zo0$CY{RKd~znsyU
z&Md5NwEJtRZDU@WZ}*n_l`#oxXiL}bp99l*x~CH{Vj7pux7$5AfS(S({4IE=-CMGF
zgiH&1VQRTopl`b`1Mmqq<1YyBw)--`)c8*+Zl%PnL%()U28f}VcmH-@pqR$Hh65aX
z-VP&^JB;^SgyDZ$ih*K-lY-fRV9@LMpPK_OszwI8_%t!Z5XGP4>A~suPIVo@H^2Xg
zVUAZLc^Z?<;f`1D@Z7seFv8Z97u+z?HK(_Yp<{bTu=?fEioIYZV~=sn@w2<wc7dPO
zSXZQcltyD*yDwjGgUv5_yqe~(qyB;kuC5D>HT6VQj7EbroWmsjqE}dBS3cQQ+ML8s
zdCe7HE9Rw?X)dQZ#_{J2$HfO+NsYhynT}On|KlB4_UbG#^5KGyFk6+PMD!`5Tvjvb
zG8%K$pzIGmPCU=mWtF*?fccKAPo~erLRYHs&cI`PEOL1*F=E?d7r}VPGghukTz`1h
z8dMW<EqAeusiBbpmWjQ!#xj1nW8q^SyYQKrcJC9JM?D>P#ha(|!!mn>7#A&`S23uP
zcbw2hn0{?t1DZ`fCyO<z33v6zwk2LovxV0`5*#=5UqJz$&%Zm0Va|A=OLCMrKFC3H
zOr$8gmU7QA)J9htZ<?Bmikloa)VgiJmUeHMoi+noRq)1Gz5^%7$8nJADl^aQACT_K
zGg};^#m{W_7i6}hJPmVmt{gs|GZ)z|#k7`2UIa-(j$?sREdAHI4ER9$>J`#%<*D7h
zdz$$!zPZ<c0ymyGKr?u-cM*jy#&pl!jhLn|znUhyDJWKxY^IKlRxBl|^2`L3x&bzS
zIu5%$lGpPTtISnAna{qa;FxPj-oZ19({dMKI*+kpQQ@k=i*@T_%{6tN#j6^#%wyT7
zg<DQ@5LCH3e3i+M%m}5`Vls-xSnS0bMWz-j7PVqCg;$t)E6(vjrHGNX>s|Y@{rSoh
z&FWm{<`mfTgWI6Yy%k|K8eIgV4l(*OG^xW{%J|s3s~8Vgs0wHnD@mKFDY@ZF=fx-9
zr^hW<x;bGmp0#e<=w>f@#}#DWm*ToZZ#QlcFzO{!bN)AK=T_H#v74ywm4~J_Ro(Nl
zg?p;R+V#eF9qv0;IHrsw`@r%2KEK0wJaoTJayqjWW6iZZoO|qgelD%+^n=zw94kTn
zp!p0B7egv>!TU5@<JEXq*Y{(z|BAT!Ei^B7n(aZiN->&uJXKyFQ*V@nUdIka0y;c?
zro6pxf*0f6Y&3O4%aK@KF<yrkuDrFYQ@Ap6fmepBqWK~iUbzOKGQpga{rvl;l`@2G
z->C@s#WgJ_Im4!zOMZ3yL`5%*U5wxKd-bPlxmRUb2>Hu9J)9Y%{Wn7XZhh)HcI2BO
ze{nG$-s#ac<gfnxNLPD%i?>7m^2{zZ=hW0byc6=)WTW7ta=MRxaf&zYIQqI0{*7-b
z;@yzH-hV(r=u>{W{T#)JqK2vdu84*F(mkyi1~|^l!|Sw?cQxUjgrgJ#9l^|3{Dua(
ze7RJwqGvx0b`%~@QflUjeCU?hNb_pR77P{k-kBa2^5;L##(=glJnI?mO5j$>C@x31
zszvd31V)BDxo83PlW`aoa&plj?@-yhUraH2bjWQ$>5?&#?-?8O+%X|<%8pOMSXX(o
zd>T)ET=QM!X>G#sJl@5Lq1?fjw>VHHggg;x0j(G(swVVQOK&-1s^wV1jSNgyfO(A+
zQylZgr52_s9xb%l0W*|Watz;!bw&RZwGgzdbHC~4id(nYuCteLHH`}S0;WBCGFOat
zVXkXA8a{F$VxCeNnT^uMe8&nE6f3dtcbQP6<a<6`5$dqWmCQAdo)j0mKKg1qC!0%j
zS$!ExTo-?wL~}+STP*!u*<6{ppLb*0)8!f;GRs}6b-YSso?KYrSfT**DPr_fY6PzS
zF8=fn<VQBnF~hs(<|e06#H<m=+y|PQQ8kEH8}q{$IuQx359v~2C5Iy|p%PtV&ZBB6
z$^Ckt4=|^=da!BDb@z>~9v_$oT#fEwli2(?(OgsHR^KH_rqU~UEw(zoIbC^LoNZ!>
zpA47BR2QFS8?*^Vx{GgaZW&Y8Oz||;^Z1S|*Ch-2FxF9IYkcvWU4b0OUKB0Vy(`+e
zF8^uX2|DsL&TbxDiG0`Kt7C5NMS=MG>tFx%FH2GA(#0gv{D}H`#~1Idc<*KxD>?Nn
z-qyK9jBGZ0hf+JT6L{+dhi!?C$tZIr(oawvj_D^I9eEnka#d{+g+T7(;!)vZM^j8g
zr8uCxltSfy^1o46@0grLLs-gCRVr!=y(d?@u;jI#8AXk2I3s3K8c?(HwP^Dx*SYGM
zGgb2rRlUoAt~1!T2q^O#+gGB&4U;*1oV!sqZ#hO)Q<LNQX31x(aa|2zoFXQp*_C0e
z@{H~A4P_a{kNTGDyQwkU#+Vx!x0RR*8}4fFxFSu^T{n_>!{0mN;PrS_&{H3nq2H#=
zS5thtC&)P&_id&WBWO*pOy{g$VLV4XR6U~k4daok2Q_4gc<flG(sQ*r&wb)J;Kqyx
znMQI5Dud6~)0sxdHMcpKqA9&{Sh-%lW7wV%H(gWd>^XR`Zo0Zdx2$ajXB|Br<XOg0
zPcP;v#B*EID;DYcg>pd*p1J<w+<%E{$8%MVy9v``FWhXwxq<o>(@c2jn4D+qpzzB5
z_V~olYPU_4*nUyboJeTK|Lg0@^SG7ovY0D!iaWzB`5oT6Fo#UOAvod>bcxg9FZ4Vg
z=F<39hc5@vtr$JmclfKcXMoQ!`HrJ}N{>Y5SfEdbx9EF&k_7a185$0F@ZRAqX!Kpt
z&t?5pF<YX)B5pD6Z_O>U0Ufu@5=>LHRd`Q$ZtQ0?tLyNW-~~J>S)F_{orZsdI{d|x
z9!A*G2}Tc#!LIC?vGk#UAu9V%!!cAmw$ZsOH~+&NCv%N=!Qt-bXRGNLX+(#Ub@-Y^
zY)VTRM!JkSU-B~@)!~azD?K-77_I!ds+s-h7#E#8O|CR!-EU*qN*6e;z-V)NH@?HE
zv&%hgfH|0@TAM>AfC<W(T}+S2nCM!6RZBz43R)LWa+FUfml>VxWHlZ?B|5P0@I;ZB
zd>J(b(_FDDjq&6Rm-Bq?M`yY!QPQbp|ID&gG8SAkwR6GdHaw5Wqf{qvO{1M0=DG-N
zWn?}L^IWCZt>qC8=8F~H-KYJ`LKmC&l*~qFkwO>nBx5rcyE@VOfR-#gZg+jMHg+?2
zUljb7Y7p~*A}n(?Nm(C@<*u5G%wo2}1t#&G%owGb|Is|GQae$z#C^L)%0EoO8b#;6
z#k79BE7!;xoRedb;D%n3b!A{?nMARin8b~Saf88CTiLVa>Fh?O{^5@kvC%dIeNJ<o
zR<Tn*!4?(!5nXq)wYR#q<6~*;<ZZS~&AC%*hpz{W<qL1j-psUIn9^;Wm2?7~>6#$Q
zXbrO5Pws3IknI|OnkkNZcGpCI_s3DlbroE~%>wsIG{Q(ho@0qxcy845f7k!@oY)Io
zrgNx<q;*iCD!PU*0db+$4;nA9xMC&sUYRlCQHlG)X;w~|owq>gs$NYqau^Of?)bba
z%G}s6OET}Z;+QJVOW5>m<GA1`q-Q`>s9^5c*mRYON%xGr+iv^8b68a68i>;oU!8Gf
ztTryb*y%McUwXkZu5N1`!8ozwNl~2}DZCj%r@~x4>tBD@hwS(K`j{%<+LM+G-2F7T
zGAU=qw^Y)_jp|J`sxg*ra;>m%{bpR(FUrIjZ2M-%KK+fy;D(xR#NXkTDrqJmzA)!v
zeY&2m<?gr|e7x4X^?TQa^JFOzt@`<)F+!$Bt<8l-t>!N4p8L5nnS<}X`#FbCgW!SK
zGfODnu6gJJdD?dok6caTtS_j?9o|;bJMS^3a#T&?i5nEO73YH;W|SkS4vhL}>I{Tj
zgMYArwKi9aJ6)T`&{tLxA}$K$g;;vribc1pHSY}5a12jXHs{x5y3OPclDdFi$M7OL
zT){J!h7v(+%Sk+U`C9EFPukVeE#)gY$-ESsJT%{gS89q)=C(6;@d}|oW}43_geAA^
z<jpwuiw}+CH}}h&kE7FHZa7(S`o$Za{wh7)T-|!J(^rMR&+dQA#pYE*v-Nws(_8uD
zDM7u{>8%GWbJ{fY>GW3I`R~u=iF&8M&?m0GnZA|-`dya|r(ok{t)C<C_6hq0{Z;Hn
zqX^~t?Sf-BQE5QWwJK|Y5rr_&ar}un_86p}bcE<R+8^BMZ<0OLft>Z4OY%b;TOS#X
z6wSkiy1LHb8aB*j`)9NB9<CbkjGdcu6;BC@+lY~_-hUd3Q7((M8~~%0g?ZY+Oh#i|
zj*ECEhOv$Tb24r2q>NKmvpFKEbk`4h-lv#7!L<O*mS|Lhi3;TvICj#cPG8cZ_aM&E
zlhrFDhEBm07ZjVo6Gu$zyk%V@a2evEn<onBbbE%_o6A#l)95o@k$l(BIDnef>1$<Z
z0%{z*Q1dd|Wlsr;hk#Kokx`(o-iWy_0vm+O%{<ozd>rFE=6CuMn0FlJiM?3p8gaqG
z`B>yQrZdS*8tWM0MI0|ypHMX6GVT(6n4Ey6uGpjv=Ed%1F4P=7E?483w;XxZ-LV|c
zt6MyNkLmOUKqHWOzIB=ZDwSol0Ep}K#kiJJw1zaruW^ZZvdiJbbX*DIUEcGkYox(h
zf@{P1v7Fky>$8c<nU9zAew=F$K77R2FS%#t+p}(<(9@2mz1pY>aR0zv@Fth1)6!v!
zqxlArJcf5|FqH#noBOqrmzGi;qaV?A*9TEZcU9yKA$oD+x1K3JH$~HbNtUZFZ~XHM
z)~l-78h3yE>uBV-ipYTgEp=&6r!x5=@(Kj<T$$#s5KmU~T|MV>o#T-b3RGlrJRK4i
zy6ChZ<2`W{b$a4|)O-}X-k!nxYMW8wqEffBlizx&YYtv4ru1sQ#=~O9OslM1nH!hW
zjnH;XIeOOB+&-b)Rma+j(RC6kT(O?BNmRNvOQxL+FAz|7?>M3P0B?k$N~M}xYMgzm
z?LhRdHPpB=KIJUKvs6>bS~Y1RW%)Xn*Yvnp^M%&C@{E@SZ?+4#IOc)_r>X`=(2Kli
zbOhsGo9f^u@nAgnB;mRX<^4c!4==*Z)k(=Ip3>rm8#jDx0JmJ1aXZ4HxdOKxg}$M^
z!FoppaAEl4UvO8+spd0U*jCqq3%xU_HdlP|dM>F+xaYbh(fAI&@4{PXBDjLyZZtqV
zH~$ZXc@7kdN3Ln85vF>DSv__n;|9TOit)r{N6nHEtAc9n4XfzjR$RwXtmNu1u017S
zBO;Ee<n_E$(5;x}ekFJ8E{rm09D0>7c{AncO?c+2>*<g2+?CH$10K}jh014R;HArG
zj@fO!(ofdLGa&z2b=|O@*C~EcqCX8Cjb8;<#;wI~E^ltR(l}ZX_E-8un!)l2<_$;S
zodNSY?VDkLf={OS!pVf6>|0@PA)jZiHN7oN$(Ya3g#E1oHF12ZnM2952Iv#^mm=D$
zQe5ln^7IYRlkslY7cr8JW0QViPjI011#f-z*AHIF;wL&lKQ?&gckhM$N!OfBm<v=G
z=n7*eaUVD+>`NNP@HdHv4TD`Bqc?0a_8LQ!B?sfb{0oMLec@*Mv;`O@ws_8KoGD_s
zE05|k&fPrh8lm{nRGIVB8tF<mG6{F?VNVwNgxmhnu7;c*>9^K#!6OdZtzoRoo2#(7
zf{~7KF1HnIIC`%M`zxc_Y<!i&-FpKjD3X~#*{6Fj(PeCI_;HX=QWx=&X-WaY#=jk=
zxR&{VzT()om=^Zb=td`O-cn15@eYG9BkWD8Yp9ifT?bzH!K~N(7+sPMh-NDjwk8h}
zQljZegR?+1qFnXnY@ohq9mmUDN8-X}{9jN(?+Tc;o-Uf_i(T*Fb)hTFv=1A=Tu4}?
z)?Z-EscAOvGBjS{yyazya%ijPXkRL9+#q9_@LFSsvRs)=k2ALRSm7F-?>epG5xm-y
z1E2Sr%pvtE^=VvUBI4Z0`D6puh<@|vYQ(!dC>EJbysK8Sxd@)90;$h36G@U-p;2ry
zQWV0*n?ci^jjk&x{qY<On^eTszy1YVT+9!6uDaFr<pNJ96pL-HEqE-z8#J7FQdJOb
zKDd1s=kr!`r+`d#@cT27<#OfaH)Oj!j6I+63xOPQuz`xJT*uv9^M2`S<hk!Z!D#c0
z(9R}|LB5#ND#_!Lt0-_8^WG(`i1<Aein-17{D~rQ;mM-uC|1a<*!V=02-AW;1*NY1
zkGN2faM(4%)OfmA;N65WH)eQ$*)yW%0((qVSRGB1c(nqjA!>6gTz=Cj#dAwiDK>cb
z<s#*U(~7$w&ZsA<#7DBZ7*wqQV+g{tcJ;gw68J?nq1KIrd1lQ)o%_vWNKUx*j^}6&
z2lM1*Kv~lzMl>2+JGCA4)~AbTbQ$m#E_XUjs?B-|1J~UsGZGD&^^@*bxJ>f8yV&HH
z!g=79tLr?n!SQh0rKMyagFCL_coUX>n7I(~xLw0!_B6D*;CV*HlI*EN+rqc3BN$^M
z?|ICjxO?w)#^(K!`;L7s-W>1`+;84OOQa>8xSYG-<M~l|<O1nQH+C%^yWbSZ*i=u{
zR0&228gw<HaLL9)$hE9q42wpm`2S=kB95sAJnlxf%hz+bhNo@{V_Q(;8p9QqRq3?@
zXT>@^b1CQuH<1s{xYD_3n{n{M_1tIX<sdGFc<DY5N;<Jg{A46_zH({TZJ^_VSp4ih
z!s&D8;ulrYb1CwxTIXhFa>{4;&GBFs@6UjUw{r1d)`*dBMEs@QyQRT1_K3e$>T_YI
zzZLO&%u_7z=`V8+;%%35-D(=ia2KzP>6_fR!RzDl;hJk46YN3Xh`+{~2U~a)hId^g
zs+;EU=@;(j9f_-YiucN2e@F6_As7(xWCfmAtlwZ7?LC#v32_w<`Ua{Xu6$Q0uo-i|
zL5^8oi>HSq40gQCi=zV#I(r=w@x+xl`m(~%h<{K_PK|qHS>+CMj2K_Ko_*W!h_`Z&
zZa#w%%7B*^xWL;+rbgvMn$<b(M*hQp!f5eMUZ)$&VT_6R+K-{MomVF2!@#(RFH~`f
z<dU+Ewp8O?Rx7A@pYS>c?>h2&b?h2UbWFPY6qA(c7Gq>IS*5W`v}T{;n&sno8idAS
zTEtt!r*Fm#amjhW^QwcHj%)MuXd-4gMpmy#z-*N;GnQjI$}x~kKlF4Rv>9_<6+c{o
zd2Tpyc(aeGJ>npnul&sYbPmgfD$N*ca`^C`yDQBY7%Wz4+!y6xi8_EUTw<w<l*~=v
zGDYG&_-HJ5Ax4S2!u^`c$0}kXzDP;arD&{jtj?K_IK||Lw3a5@Yh27lJeotiYY{V#
z#v{Q6x0Q2BPjo-&5GDyp?x$z0nc{x(>K+}^xxAuR@aZ>fa>H{DZx@^G%obH=(`r7v
zyY+X&({orz!xT2*HkS!C@0^HJRfbguAYD1oy%<&Tu6TaspCZeR19N1|tg>Br92FbL
ziTIi@?-apvdEo7Yb6vc)!!&}(Q{(A{g?!h{qeuS(3f!+16m1G!fo%u5VqHa%t1wUR
z%rPN~UDofL5mBP_biQwvvr-p*-uw^eOr8JFIPCJD&5c2sYl@9GkMbPpm_p67PHdQR
z7dnjs?PgTCKL2D6DqSG`v~xi`t!ASGEWR+~c$!L88mjfvm}(mb8a1yg#mj~i%xlFZ
zCqK{GAxCSS%X|*yW}ciY^VKW2(3g$4VB=ECXn37hDP@FIG`g>fFHRXDx)x2!eg-9C
zD(@(~;rjnnpm|uX*%e5Q56Auu{q~gF#udt~h`SYFGUZ{gXCG&-SDX8gxAk=`rTy9+
z*W|1^EdgS1R}9nRHhZM?_YCur1ln9%4dsZ(Jr~XsaN`#2z8l0;%`t&W#sgOjSH8{0
z!QMlc5udwB#v?Vv)RYgog6dWH$C^>hLPLWdW8!M88=trmDQK7tX;A#GjN8C_bRmV(
z<$DYr+wcSHjQG1IyK&&N*@!r9*71UU3?1=zs{v`79?N7gc<LIMCr~_*=yj#e{wJRH
z(m}1dn1@<)V&?|wycr+ig-gmcCUzxWIu582q0zEgVem@HDg184&q{8rbvU|yammeX
zH0tUdudJXqi~i<{i>60nboom?my`9hRez()Tf6uL;mt085~PEbM7-7IElHGbxgx)<
zA6&0yeS~+q{CzJCn9xT#Y@vkKx69wX^Q?kfw0GTa9$tN#f_`28;y*8LGu?~PR9^Y=
zr42?O(B-eVWX^Nn?cIEO&!tMC(<(Xy#y~M3zn~bTJZM4AkACp)D!4I$%Gx1ax6I+a
zXFxNw>-A#FdU+zuFjs3HcCEy4mp-1?>p3JwIKosO!tjE}$k&8;YTbDo&b7_ZzhJau
zR_|ETSDhGG&HW?BigByW9j9O1aGJ*f#&`Kb7v&(H8Bx5N;M&jF)bM`0ntB?w%#)NW
z{d*^1va7;0UWmq&*FVkC(=;Xe!yo@EW~e~V!*VTVx_I-9UxZm*{@9<hI^|P3J2$nT
ztrYC{1dhEZ{hC2z2TpZ!9e3QC^4NQxYh4~jnd@yFH}hR%@ks=u%UCFG%)N?5g3Now
z+Jx5sg`Gs}_^w+9q+qE)G?odh#&Q8Vp%I{_U4X`J0z51ih{qZM+O-SBAwggz5(PFO
zNnjmP1QM}P;8Sc8ScNSDqp?+B1-1$B6;^>!NEi4FnF1S;C9oFR0{=U51o%dazzF0C
z{5|po#-Ko8A_@gQM3KM(6nA+d87HF>k(Q!VU@Q&`Oh=i(P#hDOigKH*1{ESLM5VxA
za9ZH6s1o=GR15q8H3FMZYtsi%C(^&7-jXgNAkr2zSkhHAiu5Nm3A~T%0&~zTFb_8b
z*5H<saap@9(hS@Y_`h*iU?f@vqR?hzr{kVT({Nv4J|5VxG&~gP|G^`H{|%1?hT)08
zzac0v7a@WF1D&=4wTOr`6Wx|{1y4o#KhY~N4$lN8;JG4krG6pOM|deP3$JVj4ft84
z<@iNlJbo4U0KW<RUqH7v>_5R9-RkH%yeZOTyk$v^cw2}TFz+bvzoL)8zo4(cBD`y(
zUqe5UK1P3mNf;n78}AAH83S$X3m9ZG--E#x6=8@Cy@a7k@Q)ZK@Lw?8hMmU<A?~?G
zTH0ld65<LxS_%FcV{Ehw7;6J}V4Ov}Fy4lq#RMhzCrlJrhDkQ89Fs*Fjwv?bc1#m#
zD`p5R#!P|#j9CIxFxzI(h^TIFxeKL!j=2i`JIu3rWnsRJN4MBQZS`2BXiKoz23BH;
zMP*nj#Mcy-`56~txl;TmtgxYZh!Nr~*i}N!R}d%Eb`)zY?F{1mTn->X#IG`uXd@ON
z$)Y1jv8V(a{UoK>WRnzQiw!QtR*UknO{nE0QY|`xbRoJk&lJ<!kR=i=p|bs)4<W}A
zw<Fi0-N+MaD?`4}SNl<5X@^iKMDKJ(mbMqg78Rnzh8{<$MMrR0h&JG579GSf8+sb$
zLT!gpA$0R7DlP4EoVMr~s%&T@s)cxHR%2;Yu~~EwbwVxYQEzDn5U^-38vLpsMx%&r
z2hk+-)oxrDqW6Pl8+sf!EGot=8!rd9Eh@wvp_XH~Yf&{?EviMEP}_dov#1>RZRk!s
z5TZkjheFMl@JQ(P0X!DEnTaQsb`U|Kwi5^mwbY^0()J-@X_wJ0bn_gZ3N_cD*T$>F
zGoh~z;JKxp#0#O80A5;Dfmb$O34ZpQZ6|&avAGPt3bmcZZ#KyppvRl$S}x)Zi_YQA
z9{Xt)<1HVS;cbOC*Ww+ao9EF-s4WY9ZA9~=fhAU;pU_ta(O;;g2m>s-g7<8^a~No8
zRT$)#SAoGI-rS8LHewNmT2zf;Hc1+W3*Ak_2umx)NTKEij1s!}IYtZJNW&P5axvD1
zHej4k>v@d#Ym$x$B0fBci9#)@m}G<Z<zuo<U4|(_Ulm}Q4bCjW48JZpnCZh~m}Oyh
z0cKm6Re&g)|9;FBYNa~^8=Rhw`8GHc3oXjSBB7flSS)m7FP7NQ?O1AQ=dsK$I}gik
zk~3H#)Or;$LTv}IN~rlX;%w+;tP%P$6Y)adW+K7Quoj7eKV7?uB!#zRBgLk^f{ix#
z$~9~f@#a-*5&AM6TP-ac+bk^&sW$WUG^C5zmWfQE8)uLu)Or!wLd_?UBXlzjxk7E(
z$g`oxkuUV{JPK^QD=4&~*H9$%?RFIVwYq>35nD=7YJ;<I*rFVi34NJ?V?xb!DEAZZ
zLWPK}ji|K27jRnW<`Gm0eVK}Cp>MNMBlPebYK2-GQ0F&82I@s@*^PkEe`KS<CaFZD
z4gDNV7G>bN5Iw&)``J|ChKM&$;FeJ9CEOPJ;sEX_Vn;?J?uyuU3avIGr$w7jVJYs3
z*iwW0LihIKfl%{VJQVtJJ04luWjywo;KV56!_x=~eRBXIq1FrNw4oOevFxN(qg%uq
zNAT1}tV6HRce!}x^L+u&Mf`RLUI=}&4=;tj+=*90t>^HwWv3_`zgQxD(2Ceql8N6$
zY(51%)!dklH=b%n%-V}L1%GPH#ajyhk9@o>)LMmiY|PYs=p*98%jjz}+JSd%aC$2G
ziP(G|{cUhD1_<5QgZG4P7Gt1A1sEjMd=-OjHid;4BI1|180r^Yf?*=w_#DIi0=Hv?
zh%H$dDRjFSqlCUVh|xmd6k?1;l^84ZWfsQy*<@k7h&T3Qg3!aWm?-pZ8YWrVDNMHU
zYB0rTEFIG<%*w|M3-|5COhx^o5VI_CXE|nz_)RIIY)<)@EA;hI%oDnqi}^NQJ{Br*
z@ZxSP66{F7jKvCnScfGdhRaW4sSVFKgJlZ;>IjzmiOR7;Fm&V;Vk}HOhgFK&d=YV$
zn4OI^B7U2Wct24k5(GO^(~+p~d-+HbF<9P&6cHa*VWZF;PLT?3&&bCX5nIn<tI(G_
zuuZ|8fwM>z@teI!_tO+3Q!tdCjV!^AjBI4voT`!I6Hg&m#I`-i6Z$p-`F`>pD6rWd
zMxjNgQKaC`Kq`twY%N2Hh`~TDO8wlg;IIupjxvk(;+RE+C>Q#MN4G)`Yf)+QIk6k3
z{Vb28%4V6CiE2f?a|Ja*Z8@k_a43Bz>TGZ$>TQ~g(+Jq`)O<8pYWg`eT59%TG+D^}
zuPcKuccWS8=3(6M6CJ}X5nK1+w$K-+aYw<O4Gp+!gG<or=bnl-!C-j}?%A9&GH~BU
zC~Ck1!SLC9JX8kFRd^)yuo{nT%>A4-Mf^4oL7{JsA*A5Q{uAgF>`2>>h+uo*1iEe1
zGCUPAeEvLoZPe5>JX3hv4m`KS^j&x%7%AF;mkMv$i&r*0V<&#LnSPF6gubrFuQsD}
z{3g`;Iq;)q&L+Ge)Lei!Evm*_LXS@2ZA<$c@BFA4w&X1O2u4b)(AQEk_u*Z^VBi4y
zDb-hp&|m0oJqGx}XYrm5K8%4PcBGxbAi>DK8VnX}&*0=K7(U2nu$8Ly6o%PUsi_z)
z;=ku%gwR9!*c56m#3-Aj6r+_Sl$wDtHgy`OPfJZJ!#JOM1mkT^moP!-K`JKN;Ph*l
zBx1`oO!m{9#uS_8IHvhAFJXpYq^Jfn1%JA{53>~YP6lQR-DpCTO_XsKa|J_bb(kj@
zJhvP3Ep<l$7An=v<5(o(PmPo?MZ8^tB{osrJ}k9~cAmg8pLz|;1v@ehV}*t3pCd-F
z{qk|F@~HvD2}X*~VU1w#p*qA1hIVa7f?!u+4H5-A(z*E->`D6^DT2Yk4r~;U&6U_B
z^ld7(*x<{l*eVz~v=`eH-jah<p>GZ$UBRK$24o6$6;>ik;a}z<Tf}hrDdY(L)VLG5
zg2BdJ$P?_@w*&ct9cjBzAlMn$jY8$sT8<(S!)Gf{EaI&il!zE9-HB2WzYgHAh(9%E
zqfD?PqYB3q{^%UaMGU3xL4_Z_50!$!@=BbxFp!2SrTStos%_M?1E>-4oBgO2u``f^
zI>AtCF6u4JK8b)}`{lD}u&GX>QK+>EO+wA(xUPu7@@r@o?Bo!-VX1YMxFy(;eg?Od
z>gEO9vBb0{+_h0JrlD2g|L;+>Sz`G<+_SKv3HJrVHGA+tQNKBYhaz?#&BY_Zj<j=l
ztnfQ&cw!69M^LEcFhVwBdJZ~$coq?bw;n{di0v7iPX#*zJJD;Srco00Q&AGNsp|1U
z#Bd-TF9o~ucHxy^DDx10wya#nFE)Kz7Jl_%1%49@=QaSn-V7P6<D}Z_O|Q-AcvHmi
z<vP42;y2}ZTj=XEct_~wHS|%$P%5QSn^Ac=-nHR-(9aLwi~b_s&A<Sm8(DZy=-Ztb
zsE8dL9D{tg8-oSgGmc`2O_YkEA_mL3T^H;sJc8kZp`B$Ip{Q*;G18A3z$hD@Rfy3F
zZ#{xBHau_=V+GqYN-$0^lAVw7g5l~COz`8Lz(i%xT#rdY-yFnb1$U&K#1s)9G-8^c
z<|t+eb_PykreL@r7qe{oqnK@prIm;hv9$nm{q&qxE%5;6ix^6~h=qRmc`Q=+7e}$!
zh8Lz`iC_nXnWa8Fjb)13+=%5uU!TAV8%!xNM#KjJtP;Ap3vm{m#Tq{!`VtTfrPU)r
z;cZn&^ocb{5{wjGMv7oCKz*j7Ht)wK5qr*1pDEaWWe2t@y!8;a34L`OsS55$KaO<4
zaA^%P1tX;tO%=7J5ZNMzGLIq0PgQ|j3)8696bx^#LcU;o!%-Aim_f~^^7^6)MIr_-
zXQSAr+R=a#!N`SuD7Cy6<FJj&?k%%X^LOKzPd$cm3+u{JVN+G0Qp8XeHJpMS>A9#<
z_%{Wp7HUgFjnJ*5s8z(yz%JB@_}>noUg*ng1Z<XtjcBl0W*kSOrDm6)Nl}|m;JS!i
z+&DA~wqM?X8#YziKHO5&Z}V~6C+@)=!QhD<xU2BCGibHx8}rd-qh@5`o?vI70rxGu
zd=w7^!<P=>p&vH~kNgO`@YqH;yB|;d2xSNg22Z3SBpA7P37vj~B18m3JG0R(_)}vV
zp4vJzq1RUY9G)q-y)hNfEinTxY))MMUyAsk2Cqbn96FDmEiBH%FE;&Eim8^mzaGB{
z1_S$nzj-t17nOM9Z{GCUk+B1B3jP$>gSQlZCkJoa@bpT&BN*PjAAJP7Ph3M^!C-kU
z-nDT#8}{?ze)Ja%U%r3=7B+IDFW7#q2m@8vy@MDeVzB%o28;Nx0YijZ3NX|rudc^1
z%U}V9i`bEV5hDacJ9!XfsTrvlB^V6s!)O~JoARoK+$oQ>5z0?voM2Z~6~<d?T@5B!
zUg|N?rYg$BB*E~JJWLks$T)&2iuy%9rU^AyV1|Obn({DHu=7eiW(l@ulw!7`w&Wv9
z=)on-_4C|?d4fF|m6)&an^{;W^zZptB=k)&7As<~{2Z3p<Z0<xs_^C}EEBQ4;W(E2
z;YYE;!i*D$QPh@8SS4ck#0A6&cAYqkHG-i76^OUF9z%kNomWmEQ80LUCz33zs6mQg
znC-jKQcv#2Cc%!3G;C2`Uth#l8+GSyY!mF+u@|Wde~^xJpO}YC8#Q-7vIK*HGsqSU
zrPd%vup{j_a&6q4GUQp9cLe#$_D(GdMC`g+jY1n<y%R-(9qCml7K~)?MTua0pd6)w
z!MbxeEZEslg)+fVQ7VqvxETjg?!$eku;n(P(nigu#A@qUhbj@nd(NTShefEdFz*^_
z1v?vZQ70HGDnPwpSAG@(g55h#qrt|#b`*^kUOA2?!C>GbuB+T{D$p!qJGTNiY<PJE
zZV86c0=TWHchBLDh#hG;xGUIQL7`Q!{aOXu1cTL-S}pa;cHFnr6KQy$ylxfXq0k*Z
z1*+h1ZV?^}hDtxj6Pu`x%2UDKL&XRQ{<y0goq`bxCK2WJ?H+VnUir$G(0}Bj*XC4y
z8P5beGWO!RO<q%n7lQ2#*YMKD4HV*)jhoJsbiq&><yOJ)1x~S+nz|dm`Ek>MXWqmb
zIeQ#$2!@Kv@a8jb;(d4wZ;2QT9LC#%;jHa=N8v5G=p*#+h3IRO7gph2!Op9t=w}n<
zXQRJh`_%vj2zI1h!+V0EqDBm~5w6u^kfr7vz+l0jt{%n^8{x_>3{|<`7GRi&;e&@U
zT(Gkt3nK(0J5w>zr|!Zi!S17#7%dnKT)-H?aCR2P3Wm}yVVp|1cMju)no}{s#@tTf
z)rXasr0|C)Fj>T&9l4kyV(Upv6ERqxjv0a-8GA6(QUjToB^XM(g4uqAV~7&$$;iW8
z!N~4v%(HR#@4<Y*@TJ38XsOw!ut=~wbvqUd_M}oVU}45pEVXen%dyPD1H7wi;l5m~
z5bQ`RM~q-QH^{35Bj=AGPF4G+2y28|4k2FXznwyYA_gz#B2h4$O(E7&tFIwNQSYQ<
zqYXcI8Jh%qbIP$ru=DCQY_+K>u3?*C2NkHPg6-UQr&|X0AXCIpb}_O9dpRR!3kEAH
zkt5g{C_=7MeX$RDLSLUlzJl8+<rUaObw^O>Q}?3CQcvtcu^-_gN(3W^OHnEq-kpZS
zHbNkPGQsZs<v3>J9y)<?!Jb3MP$AfnwilJ+=Egppw#1CjQ6<=Y?EtC;drzjKMle`M
z%UK)Y=zi1*b{)BZdP`-G2Q2ke0~&07uc1-sVFj8L9HGMFx`=l!qgm+7eYm0EaDE<c
z2?hf-xNWK1GjPWyD#Tq8f2^oNtKd^cZ4->tUcf!Uu0l$(g55di@j$TsDs`<k!uczB
zBp52%i^rC_lm4MpSnGZSg_;8hDL8m84V{AR8Pv8a{NWjN+whB7cq$m$SB72-bF%PE
zup@OBo?Doai5G%h6og+2hI1<M%EF3t{A?qX9>Fgv?2B6bYQy(j!f%#?8sH~ymhCOu
zhc^Txhxg-63l9eHmW4HY@wQ-i&vCr-lQ#{2yBB>#>?&_SU%_BqHQp8M-Is}eHr2r<
z^cM_epT+>eNOlF@v+z(J1_}o2YB5N#BQ*zu1;g8SVu)bZ0V-ZC^;!mo2?ndnG2FuP
zQj8GnuIBs1g6&r?VU(p_I)~AI-2E6M7|K3@v4VfwSBi0hk<wI*SB2YlVS**@+=YoE
zeti{_MEt4o2qp`L5AVVh!S)7DwSpb#S1`jyxYB@`KJ_GK33diJ)e83R;t`hMkGZLs
zE7*Q53-bg+sYfwiu;(BpS{s31<|4tagVk6p9$SxMiHIHPN3m2e+)#mKf<5~#VYy(W
z_y|_mxSZl+1lzChq+jr9O)BC9I~xkIMkPGTLcEAS9xX+JU`NJDBntNI*o`DVRUT3V
z!+Z8(qYrDbNqIdCV2g-B>J_&NcBJvZ-$pIng;YhoaTe(|YV~Pk3icl2Q@(<oSIdws
z*mH6_as)%E1<1AVzzO6DhRaKkZ?ip#0-<{)C{%F!mCsRRV;(<)Vhax*M2TRiunwhy
z!Rl%pwh{JluC>&YJ8(>}{b~)$1w+~CsIYKf6)FWIRY!5!GIJGGHvC9Ess+2w>_Ux&
z+=SH%_Hcr&vvJE(Q7;%iP>F!x-}V)wL9nx79~y0hv&CpqVGl0jx+Pxy9L*xO?7|HZ
zL&e*1%Z8t8z-_@u>Pg%Y?7VygcLh6WjnQf$4>Q_qgsZ!8PcT?_0QXhGca&*G45v2Y
zp@n<O@kp>eqX3U>D$e3h1bYq~L(oz$?m$Q|R8)^nORcU!M6inzM7Ln~>BD$xBh&@Z
zE7;2w`kA=-_X0c@F_L)@F9buSg?MSH2h#CMu>Hzz{45x(JC9$K!Pj;8Rm6_;^Y~4$
zGf)aV_a@wMelFe+>^<IyHw8Pd9l~4By{WZjAKn(R{mNdvV_{<r`Uv(MDnMVsj`TXb
zE7-{{>n9iqY)5~=F3#fv1cQNWyr&W#@`{~}nz0>&o_kjI&ku3l8*JlN?86Yjc5V%a
zT6pCUh6#pivM^jQLa}dz;E#I_V5Ev@J%Ldoh6-s&D%imt`xwD6zml<*%Cm@Zg29Sv
zjQ6Q|n4r9FG-9F+FQ;DC!T=Ss3cr6IQ~dBMOcU%qOoIWzNZ~civ=O)@%o6O`Ux(R(
zKUJk6%EmoJr*nesmoqR=Fnn?s=GzE77+7ejTp1Tx>b?V5EZCiQ2}^8*?EP43Ar(@~
z1lu!?VYy&O#yPC;sRs}v*mZI*RtZLmPaw|1+ACNi7|!2`c){S+3M5z<s6?V#<O?2o
z*{FM{MG)-B*o}>X?WqB5vW<HRTWr)zC$QCqUp<d)f?Y>WAk{|Leg^3lmenECMku_3
zEDH}bAzLt1oP`|0?#g213Wl5ZBTuj+{VMVWBRq>K5NuDaLZM*q{(KY(1}kb&Y$IIe
zoGTdKegUPn5=U{^hF`saGQkj44#xyLG7g|zFjBSy6@owR%0{JN`_(#}R$dP-qDsVY
zc@e4wLkG%GBiNC?2epDdMW3TiFc>I9y<nv5C<21v?0PiV2$$2)XsJ0gNEPg%s^hw)
zo}fvpSh)8&ZV0s;z%2!bGC#*{%V9n4i1=d;1za0mn2%NqIqcd5Be^?pPq2GuE$&-*
ztq>0cdpM&$wA6F7O%)97JcGwJ!m&y`5p2K8*G2^+d9+Uz?9D7ir(h_HhFF3fX*>ZF
z3}4%ir{er>EqX=lIa`QlwzkLdT*SyhY7zvyDtF<f;8UK(zOt#PAN^S{c(oC~D1)1)
z@T(=(mEbqQF!!DK@7|2tRmvGxFub=CZwhv#amKYUJri#WcGq0OJAyqG2hc|_bl@EN
z+PG=Oc=x|c6_Ja6B1R}~_qS2Aw_|{Xg~fPJFr0k|0|mP(_F#~OJ9l8PU`KjAhS<2f
z_F<@nd+1hFus8EOhO69LoNz@9?W@2@8(vh0QG)Fm)fg?<)pQDD1cMZD#tL?yKZ<dJ
z;T+l*2!^<_O%VKXPbMY`KD|(eNrK(ic3`qt___&GMC>ZrfoT@*yo?!w?G0x!Q?O^>
zHOvz1-gO?c{Rk%zB^azbj=9RV?KI|DVnqP+1;g8E)9+Kuu}Cn&v);vmJqOCMM6mZz
zBbM5@C)2P@u>D8@mJ3E|0$3r~#nBcc*h^LFD#7sX48#exH_*yJu!kq(@#67jITA#S
zlpaE&V6gfKk`(^MaioYC-n|bSMZA-WO*T<kBevM^rc2l=*jtu~ZGxfV2BZph(BLFp
zFmg5*nS$Nd4kAmi=RiKP1;gcMkR$lFoB(nKLpjHgXCssrAYZU&2Pa&gN`-=8q>#sA
zf?Z{4D7NtIPLx=9;R;FxL)ncuEZ9wLYni2HUBofL-mF}d3x=t2sIV}f2V;WW6}xa+
zFw&5ZDz(!$X{feQs}G~b!rdjP75tI?TPN6iFdy}ndZ7*h!BEaVGzfN|<l&fLM`|gW
z1iK0^;ksb3f~s2^fro=PEKJ{xTY^8P7vr{#do>+*1iLbi;;z_jIfz!FyC={lbniUw
zDPn|6=Y1P|91lbc?c0xsK0Jj-HtM0Hcq|yKIEN>K?N`$f6bv`zA|%*(r3jsZU03%a
zA{eZ$MYmvRX9k`M{_Q{^dd0$b`|wP}p5kkGF4&%W9xntV*_ZHAFqC};uLPf7tisQN
z!K=sci(q@|0sLy?76tH|O8B}8c;U^sq5Zq?hG2N_RlI3o&3U{f*weHdZ(C~qLA+yO
z`5yFn;mxhhY3M6rcO@?f*zoPy=qDJWa;v{!J5P58`0xzg6AV_;NtA`v7coe%BkdFh
zTgYw25F6ogCWiWPY0l-xt-){$)2lH;jD39^BSma)%)}@go>7j`g5m7}jIpqx6l0a@
zZYIWw*t@F?;|05}WMYC~r0_B(3btRZ#U#P-Zk~_{b_Vh=MX<Xj1JeXU`wn4-U`IwB
zW(v0RJZhF;S0QIy!Om;dh!PB+4PdTd=)iHz6O7bnW4>TlRRtD`^E<g%B;t?tsaP!7
ze&ryR2!>Ldu++k$S}YR`U)_i0g2B2ASRvTEzXmZD9?HZj!H%?A#93-_8rBGg3d<2M
z7`{}C1i>Dve-Z`TQ!|ky`1Imgq}aICm$6YWQhFMjeCmE|5p1uc9fDv-+68PA>^)2=
zSFoq)0MZ46mvWFP7}=ALEVamuL&z2}y!|Y4Eab<LE7)6Biaa}5QjsrWXJZZuMEsJE
z4~ZD0l~0j~H)~KVVn=!gN(8%ThFB`te)$*<3wBZ`Stc0DIe}x!;4bf3*{D}&DQm-v
zc+*O-iz<xMf}NKSp-QmlAWda0ucuHWV*BBG)CzXfXs1pvc(n-ig1sd*2ndE7%g|sW
zlpjH(4=<s~M#$fb>w@jqPM}$^Cv_)o2tK`V5w`?8(ra;B+}z2>9T7XPq~We$xGWv5
z3jd}AZ6dbQ%=(^%jfJ=`7_8fl2bRhV9$G5r%SVE}nPqrvVR1R02zFlHj-X0tE=9;E
z?nI|xuzWirf}wqd=oXCZr;V4TrdFa?u=h|Zp4r-5!gCROXy@`mu<KF@UJ8bDj^dSt
zM|fCf;rYY(MX+c4Zu~0v<2g>ag56X=;D2~CZtybi90*40PT)<!a5e872zHef;%yt@
z0xc43gmPZVvXBbBzJl#EWPMjKw66sHY=k|#&|k3YOc@6J4{xGwt;BmGM$V^WpkT24
zI0gwmt>xtd!4NlhL;MJ5F;uYgN)?6)Mk<eDxL^-8uOn<+T4;^5)Pq+r%2F#%W3*WK
zx*TIf?BH4JSi#8Ia*PuUQzbm!QnQON!NOBi<62mD8IzRPjnkMcVprZyOc4wQE@PTt
zI|s-N!On(q%oGfNo`zY1-8;`=wqWpTCZYs8(obQojhlS{^8|Z}s8+DeScHWl{#ZqK
zIfC8C&S0@%@6j485sZ|b#!?%BI<94wTAG37f?ZekV1;1tL^@&wJJNSxm0<VPD#Qtf
z>+-Nh@W+$ch!+gy79&A0a;^}Gf<38aND{k`^N}KAFmMSQZPePM*d*AMSA#8roq-H&
zwUx}mHW9=7vXCkmJjWw5!AL1>X$5<VFCfcy$xdX8*nP49If7kh4j@<IUsJ>t@o)R;
zkS`eC!}(S)$lDu*g55`{$F-102F1#tnWhOgRrW!Y3U*zp#9>=NEy_e}5AfM_!Oq4*
zD7UFt?+U?SIVW7f4w^2U7K{YWp-Qmx$`w=#b{*j51i>KJ{8|fXr&T8yX3N#vxToq7
z@TrH;AlSRV3XK+~*PuzTJ+%<m1w;EO;R=S&W#NW}dw9{+!iGJ#ZQ~wD#T^T?kK?Xj
zclA-U3Puj{T9@q-Uga0D{puCm7xB(6JP@&yi^4;}j#SRLg5lilcq|yn+=C~Uns)_3
zOReSIF2PWs8l8e2yp9pEakCDfTd=!+JDv&#tIwfVu=7d|o(YBuv+-Ooa=aQZY}`u)
zcxfZ#?8hs?A1VL+Z2Ou|pNZH`b=t2Y{`(pHCSngoHoTOK%kTIN!5=>_#hWjsMLm+c
z8*hmmq1NDS3ojSpotNH(`5+U0L~Ktzi@t&(S{=Nr@Vn{gCt?Q;R{INfQl1(h7_Q!d
z_bhC@h=GFL=gwk~g}kdVSg^Bc7lsJ7U#`MX!JeI&7$z9rz6-+zL&ZBWLM${>jVts8
z-2vG2jps4i#-w4_7{RX0dW;qP@kArWDb>H9#CQ?IW#=(LFjBf36BT|V50fl0;{qlN
zhEj_#MX)2I0@DP;+*r>L?7n&xGX;A~c=1axlv#q=f{_cfP!Q}nRfV~NT{-obXCci~
z<_q?gmSLe_q;3}$33l>Uz+xNWWG0pf1}jRj)W*GZ1<M4x@;TvJ>M5R^Sy*@uF%}j#
zVU>lub|Ox&og&B@!5(&0yoDt>NDvGLijgQ7E~kiVA=U6Hf}tE*DOf6vHZ}<cE2vix
z?732at!n89Y1k&zd<dxuZohI8={Efy9-Ikw)8aBqFv9&(wvEe=FUQ8+or+w+9xk(a
zg1xn+$QO*9IEw<ouIl|L6bxVHoms*5!zCyd>^_-=5*wGg^->%6U>Obz{<u98W#at1
z0vxlcnm$Llh<Eu2htSuVs8n!>#xkdEP7Rk(CD@x8K($~b^$=<-JbMMTg57CFs1yA0
z94~|k{?y3pvx4FBIy49dFE*l4FqFL$O@iH=cCK4$VFsEl<ejJ+f|1;vxMiuk>6J?`
zTz(99Y=rYWaaXYON*-DTgEXpW6YM!wjC-o^e`Mjli0zFuxfKkb--m~SU8QMwBp9Sx
z`>|m6-ko?N__yK!f;K{B4MKulb!F%j>?t~ih~U#p0dx!g)OZk21%vgx8D`@iy@+Rm
z;qB+~+(O#aybz3Vl)n@V@{H=0rRJpKXTi?P`S`_B^G@Pd!LE~6@S9+mrxE!7yh*p6
zrjl<6cBE(FO$*Zs@s?ol0u>8_U0k}}vDAxA=pz`~cN~2MBiRAGYpFX5(N8eE<0|?K
zwl{o^0hU_Rg!e3@lrm7TlU84Y{{I^8KeVZHjRW|XaT#NbF)m|_F~%4oB11%m$QWY`
zks%^7L_{t!WLz#XhKR@z8EMzl^p|aIo7OamZA`PYw)xSdX?`^ksp5~)Ij_21k==9B
zoRgC#-DG%G#^rLK_51kWhvz)+Nz$~Z=e+OpyyS1?BHke#d5n3NG%^LeN7|_lY|!Z3
z0yb)HJdI77&=bOD(ki?0K3Pk<u!S@hMzHlc^Vmk3QkD3DM&}CHPFk@5c93>r7$1`L
zVW|uws|DGTMRVsP_(XGyA$&?YT@%=;3CS+(dQKNU)7*{$>?UjH^Vmbyhhx~Q3DVBA
zkF?x#_*^NK7W+wKs)PgoS^uwiuLlRIe^29(9*)TJnGVWja9E?VD)R+tF3jNw>BvR#
zOP=7Ste#On$>N!2O7Z?tveG7fx1?1Z!q;T2Z~@0iTRzNjo?vMbCm1v)JUFSTgLC+X
zTpMY_DYD{s<232ah4C$Ed!_f5TxpSd1+w<NbU2VkB!hO+$x9I~S&w+pNmdK8dPY{<
z12{_>(JcBj*Co|qr1_l}gQOK3ftR$qWinTz@+o}E(m4F2Bac!+a&<0&F-?$P{t#J{
z>rhyu@=zO5#$@@7w9^Glk;a1b;gT!T#u6hPStL%7X7B=L$=anH(wflIfs97y`j91!
z`CjBnd!!phWi*G9Cd5*>Og5&|xT4WW4vXYUYZ2G<%k<znb#ERwdD!X-V2N}xDJ+ww
zYZSN1roRn$IBGGDyL!Bs#eE$m6-_^r_46q_Agj_0^N_USvXh}Ej7lLcX@uKR(S-IM
zRLSO$2e#(Q*Y1#w#1K|B!8eH}S@D<gD{0Hh;Und22)~oHk{gfpoqO?wx_K7Qc-ShB
z;SaJtD6_d_qbrL4DJRpwll4E&#d*B&Wc~VUkIUQwS(nFw7fB~RftN@#7stz_H9Lt{
z$cFSlys8O>99|=<St(i2T+bQ2uFMbP4bt`}@h0EGPZ#kPgDWjQysfFZEZ)&n=}CQ;
zsW+rxK~tsRbOTwD+3k&FRhrK?k=B_4Hj`#i_BJ4$=nS@y6{-E)s%#y`HnNr(zz5{2
zd^y`m(=~)0q%jf3hkQ3b_TeKvUdrHOvJ#YgUDEoqM2u`o|L9I~MH;Vm@dQ_82N(vc
z`F`xyQT?6RqYQUoFX_aGu#c?!WJ@e^HQ~p8p6gy72XvGtjDyO~4jkg~TXLhT8pB~8
zt`+5aOB(Wp9U;wv+~<-`N$ztwYSD*(P!9t*s+m&M^OaKO3ce<--V~0Jje#(ZbJW!U
zPEh~s#YrAElG5>@l*x)y%2*PoNh>TZ4`l7n>M+uoOrlj20<CCMCI->23FD*i@Wt*$
z(8*vk5k)sym1V25q&+r|KC(Vt#5vLk7coe#N%@DDw7h<dkhV7tA6b{%9Y5Llz8gWl
z?s5=g44Sh2K!~iB!U&U2xgQbI8XrWIqkibb6oW=AifM91_6LcPwec7d%5(^`WJUhM
z(quglM258eqsVGPAdWm~W#w8+nuQcfWNmO3mq{nni7RA1HGxHCzJP1WQ8%uWc9*p3
zl4j6_C5`5%u&kWx!fn!t2XIHD$uZp3Xs;jl$@)cU)+N{4r0;<=qGdcJ?d&*y(OlOB
z82phvD50XMvk6r7_<6b7l2(gsf1nBSII^nD_|PQHg*<-M-1bpCQs$*1m#pPOc+3}D
zlx?*bbS`=EjKQl}{6Q^y@LwJ_3sN7ZXPOBBPuDN8^|2Yepu9MWzdT*P1mE)EMFuNU
zbn+5||LVfa8kDJoSIDZgG`_0i<-Yherv7*de`U~`PvdpcP6zP@>CC3_rj8oV;Vsfg
zNANaJa6=l)s3p15Qg_Dj9usTQmb5|7RPMt@vMK*vH|d#5No*!fX>WL+w4{1z3t5p%
z=2o(ri(ngR2W8=mtR+*}PCDfvc4)$Y%qnQ~q8lHPtEnPBCRbYe@Cj-4_TW?f4Bgnt
zV6_m#E<N5qg3m~MWCpuQBb~t>O>n2NR~ZXopVBji&owub#eU_y3kOJ3b~ZT3*In$w
zAw8a+$KUk0D~Q9&xek0mT1hvKkZaP5_vKU0F9iky_&dW!SnhAhT5kzQ`4YFz;4223
zA=wXsv^pnoOeuvz$90rU$e$q1-~di)LMeuC$lr?nIHm7=45t}fk)DumNk=wCaghxv
zMQkOT;eNE~sB>s%(2_@M4{6J7PN$Aa$lfxfDb15-HDNx9KBatyb4u@d43f>!L3lM9
z@5P8xipPEAny(vv(&^41sFanFG18D<{SavuzC)N?k)`AaS<R0kN?N_$m{Q6l*)-Xh
ziXx_z|0)U6kQw4xvhJEinrsGSV}$1kGOs{ZWsikCX%D$kB%P=SC9*E1vzN(LxhcLv
zHiNP^7iq~l{58_{lyRMGOu2EBtY*VlB28JNUe;V`61+{W4Nc;XMrT5}OV);_aG&4c
zjW~X0(2};`2b$`d!b8%LYs@cXb9e@ZCZuFufo#aCf0Z;BT(HTNmUD2PGlx~uDdo^q
kmZc<@tjiblh-{u8!tbO#n89Pxl6$u&WKD|IpFO?ye@WP0Pyhe`

literal 0
HcmV?d00001

diff --git a/models/ggml-vocab.bin b/models/ggml-vocab.bin
deleted file mode 100644
index 38f63493a97a7e85ef04a21697f7d2989156e5e4..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 432610
zcmZU62bg3<)%FoVf}$-LP*96yVRvC!S|!Ijo7ja7&CbG-xjoZ8GtJISHxqW*Mf@(9
zFt>t;m>M%G+GGTTD+A`xYtFXlXAWb|`oHIW>+*g7|IhO*ytnGqsZ;sXsZ-T6myV7;
z&hU37{*~|#9|8aKe;NLH+#dKB9=+-n0ssI1{+xh+4*22!+0BC-JiY+C7r?nO>Intd
zvjBS)VDAF#Q-CKH;7JA8w*XHrz*7p~<zN<1Ex>*S*uMZzE5OqW@QeaHvj7Ja;8_KD
zb^#77z(ECgP63`<faet;6yV?jEGWPs1z1>sLkn<N0TvZtaRHVT;P3(*QGlfdII;lC
z3UE{b$_03S0bWpm7Z%{?0vuC-V+(Lx0gf-g2?aQ@04EjT<N};hfKv-_S^<&*3>08_
z0ag@XWdT+dV08gbFTk1ttS!J91vs++>k6>G02>Ohu>h3<oK=8L1sE*A<^r5ufO874
zr2yv^;JgByUw{h=aA5&1D!|1BxTFAS0jdQUD!_07Y6YklV59(}1sE$pqX3r{U~2)!
z3oucD$pSPB&?>-X1(+(pbOB}xFk66a1=wDI%L}lh06PnCMFCz^0H2^Zs9#)w|0+OM
zfU62{bpc*dfR`5FWd(S70bWsnYYOnn0=%jKuP(rA3h>$jysiMRFF?BhZz#YU3vg`#
zt}DQs3UGY^-duni3h<T!+*p9O7T|3KczXffQGj<Api_XG3h=H1yt@GJDZqOR@V)}P
zzW^U7z|95tU;#c<fDaeomI8dF03R(tw*a>m;I;zXUVu9aaAyJTD!|<ZxTgRgE5N-4
zxUT^B7vSRsc%T3e79cOcCkpUT0p<$uZ~-1Ez$XjvsRDeu0G}zqXAAJT0(`yzUnsyA
z3-F}^^a`-60ADV^R|@de0(`9iUoXHn3h>PWe5(N8F2HvR@ZAD@uK?dKzz+)0FTf8A
z@Mr;kRDd5B;J*v-lLGv-06#0h&kOL20{pT7zbe453-Fr){I&q|1^8V7eqVq;6yT2q
z_)`J?T!6n6;I9SvTLJ!FfPWO=p9T0=0sdWp|2WtU|3X;2b@GcuGf<lzbX1?he{x5S
zNwb6~R1r&5M_7lb36$rLcP=wyW_?9Z<NFi4dm#QEV~)`931(9>D-%_bVM5eIh82Z@
zWh%O1vldYUX)~g!y4f`fL)x}-nHfXq4GgV<Plsq6VILq0+s(Q{Ewg)wnzIm`s5)-;
zX~mA1^@JwOzDZPvtbK(w&3+~{WHwLK1l?ah;l*!wP1r+(lcQ$4Kg7yI3HBnYq6AMO
zsw2M=3G0(6!qW(YSMvdk52(k3h$fMDP#O>}R4h_27D9@pMAL{Zi_Hv#Jeq;SU~{|(
z*P5M7G=<pYp}k_m{O`tTfbANwXrdO@k-UZ)tcV$M*eqs?&CVsNqU;wcYz10Oi9Jb8
zfrDmaLa6F-p#e0g(1m6*LP&A>L+mV7FYt>Lx)}{l)P!VLD{PzDONkoD`sIp+VAl{;
zk@i(Yvj}?)Q9whyj;M|lZN;L4ypgDZ6xWIFaI@<bh7>mtO`*5ns6u(4dz+wl#CI_M
zGL))A)I|4s7g1Pm_8t<}y-nVyz#Zs^MAP8&L85IyA0`?G`Up`2s7o}9mc5NARLt%W
zT5fh1Q57QIBRW*gy^1}>?0zC22@gmFug#pW>CN&G;f#;$hgshf?0=tRvL?#)X|Y;q
z_F19^I>P54+5^3%I>YY$MK#likJX-tfxaxX&g`p--{E8H>&z_-A>(h7U>w!(Z3b?0
z5x&d7Dn`=xiDp39SDFT@?omM>R6l0?RKx5iM6(d}XGCFv*)K%69liZm%6`}r{YC_9
zFi=I{?eTjCHjw)t6}`&r&qAv)j){i1n*Ci#s^Ifa2Kv<YZ=yOX-{yR1S5cnF&2hYY
z^ll6co3Lk8pl7%{A-nY+b6$~{Bz?+z0>gvPJbNnLLd+9FhnVe6)I{^`Lo|!d`$WZV
z!7M^F1~&W7?GB#9Y~3d-{&cgaC{wS{5@CHBHTG1+Z!p_Wf!?Y2XJ8AH;?tr;pnSTZ
zch+YxgV6GM<C!YJ_J)`C03`&;v&0v*_Uy<PEDlsOL_COS$w_9<AsPp#=ZaG?{dgI|
zu+OA{9I6vwd$8E1W(y?43MV*(p$&}8g(CF{^iat=j?lxD>I$<(3iO3)F|!+17hV#>
zFE=|}0-=_VAeu#ATq*{anjNWFFkPlt2z8WV&%kU;HVvO+pD&Iz6zm0x-)Qzi1^OIu
zGy@x$yN+RCa|$vaOIUSyoJf7lA1`<ay1)rVckv15M6%cl2~HBl7{oeRnH_0%3Im%y
z?VYO1+U&EyX~L6DgeOWjiQoak#yI-ka)tUrxq{q=G00XjaLTKDl}J6+YQ`URl-cP-
zP4t;H%%DE%OV`@Dy}TLGN}5&9m@E1B|MTanz#a%6QlfQau<NQ_SDUS8URCd^8<<05
zr`blLCCji3DOZfXvm|s6>hC5&pN<Aq*zLZ|ZdR}lw6h7Dv!FPK90E$Xg@IKU`&<Q{
zX?7kN^giC}nw_uEbIdLv8bt+MD5leB(Tf;ZMGagmP9A>=(>G_w(VtQhc(qkUipC!z
zx*ReMOC}$GHNu8BK%MbNqB%#DQ|~sjQNbZhHDin)hxiQ!R(%*=stU|p=&cMM^0{?f
z6y7^01iiH<3BzWDHVJ3PAx(>X!<A;237u^=B}$)Yrxl2{m{A&^)@Ic?$Ev=JZDaU&
zW7uqaQHR*sTuvN@eG=V4CNq$1Cy6=OuVCPkrC9Nm-C9syN!aj}>&1k1m;Aq!-v~Qp
zrc53{{bZRKW9VHaIWMs6D#llbyi%_wk0xZ!1cTSfOT>8?p)VnEW7KEUm&Uvi{8DjS
zWcD%@ql(tbUPgjI<wJ0K%(9n@_GGhHC~#oXvIuuw1?_~P)tRbgf(>u^SBe&Ok_mFv
zdKE*PQ#H#38#8G7SBry(zIyHnj}0pyOuqaC4EF3bl5@x>f!9)^=CEZ#r(l_Uo$_lT
zKfzg-;`Iz|q0eN3RbT$vQHrgWMd%9c8zjZ#v}JE#X1#lR*o(eVEKu9o8zn=nj(&5k
zWB`k6BR9+^*AZ^_?5<;c6}i2M$OU_oa`1)cdcpuzIlEr*Xty^j4a)Uq<uvR4<OY#K
z@6RH1z2z;Wokf)iwII`tq-<be%mmp--m1cU9EHi=N{SI5LvK^I5a4Y}=6(I`gjG~W
zCRj&x*gME!SLPhvL5eVqF8NM!0LtE}WIl8{3|zd#vJO*Cdu`ZF44s;_>?URARroFj
zh7&Cls$%MSH^Z9SEPHp6>77tHRfOAJ0O3$`w(pfN6FV%6a61a|z8H#XcpusMBKm#>
zR#9Psvr`z3AJD*qLWkP@07*{tHQ;7K4vL$ZH&hvn(+`r}$cSY^4b;Mi*khp%y9Swk
zhzT3M!hcvfdT01Bd5m-NzeT~ldn~&}oqhWl*V&ISqKTwggf90-m3_;b|D#GefyJ?_
zx;xJ@p#_*`ZjGE#Nw-Q+q`!@U4QOK7Z44a7Ja#*oHobdgw=0!Tymu(CDKt)Y2ZQTA
z3E#=!OY1(L2#2bxRCg)mAc6%OuIt~eP;8zu!79?(Jz`z+4u22j7)ASijG@&T>_M}S
zF<rx(_1?KBJvIqwfqiT$s6vOiPdU_)?|ou0?z;E=WV>vcW%o0*MMdP}3~gYXeq18>
zWMB`d-7zPD$pef)_r|RLATdUM7j^>XGwt-bg75QI%gNEZ!zZH1+p%>Q97Ugci1C+P
zf$bdwTRvsY5jL@8W^;3UqI9#9QDJ+SF`?G5Ot3nI!aPC&8|qY#kfS&ICnd+7y^mY=
zNhTacHQ1-b1G;DSDN=-O-1IYDILoq6lVHl%s?UhFcZX%4A#Huqo9(lzk?qYXR8{<t
zvhX=_X|yc+9O*H?V5<8(+3aYcK%ZAOhnamrnU6uI%Oalj+3t%Z_92^nk);m<lbRU5
z6ti`8E&CG58at7qM^3x0_IBtoc-(iPy9mPyEK9qT7t8-;CB!B$6Kwj@{uPBzqCj6^
zHZ7^hUuDYPea>5WhLge9NI&b7-`7~AI)=*EiDnV@btc2!5T*YHNjXpi$7awB-(+ah
ziN6{1*!3E-Z!tJbqIJGSZo}w1-&P<t9@)2567L-L9TI?9_8k%o`?m1A46I@H{H{{X
zV)T8FsZc{0?%yN9Vl0E-7eRB#vhOp{C&V8xu!(+^{Xltf_ugl4-Gc=iZn^m(L#KV=
z{vkPlmp!Uvm(8MvA0<J<`_hl1<mlnqj~E<UVDV!GPu4LogfXBnxc-~`!V#7Uc}M>V
zlQl8VWItgtn7T0FekvB*eA@eIoF=7v{!B4GEHh#3_#w;BmAVcMH2b+yI4czY7ZTWY
z8Nn%6sDDX%)Zd+!35RJPOTUtkiDkc1-|RhzDz;y%c)qs&niO^HuzsV?JXL20%6>zd
zs+*dAD=Dj!sNCNw=~^(KpWEAKwvl4=XY=HB1&l?%BU-f7vfq(@?q+o1-;1LwAHOHZ
zUKc~({t!pa7`A`eADFO#R{End8bzP_BdMz#+JBNPW4`wNNhD`ssr|E<Kv%`k{<FB>
zhuW~eh^BY9Wq)Dt1a<Mh7HlxM{~Af|H~X7{TdqI~*C)LN|1S1exX>VfCw+Y!ipxJl
zUxPI$`-f=88@~1XCuyc&cF6uon#PnX6#pvdk!{2i-uM13a@bO`e~S&88{z*^+YfWK
z&Hh6w?(gkkDwD0ICH642%-QLN4|aQaH#ACp$Z~sl4{shB93Dsd=JrWT9;fhWF3t)c
z#v+*uhJ|Rp-5%Z@Nqcwh$}ABc=YqMLa)oY_E4VT3D*EF|j-19ViTKg%TFV|!Sunoy
z$18i+zjhaKwN|racLq1c(UuWnN0|$9(%*x`SleeT*@ML6zMd<1n_I;4Js#fIi}}C*
z9<yPeZ^;uV19mIf6PSJqGjA?f$9S|oS$i~@O<S@jsjC=Vdy?Ach<s00*z6E@O?xrC
zig`E@4Abaw5pEf?d@qW8B>rF;BD-nO3i%wew-}6~pnH=PqZZT2K9MgPXdecLN3-iJ
zQ*a&KE{_zxzCDq|UiZlp)tZxaS349Q<~o)OHoOx(iTp5=gIpMEAtnl8D4!#iKZ%ky
zU`pASynPl*_7(5#ZUxH}9tNf?-<QPOp)Wp}Xaemnbn;P_J(&!`Ce&Pnt{*=|yc&}j
z2FX*HYzRYJsJhegr>NA2xILvL7QI_75v*gK%3|oixaET5Ztj07)AjDa0+>9N>1KQ)
z5%ftgj}-U$%(tJ!z&tTw$$k{Wr_%kz<TfzL6&j#8?=Rw-ZwHh8B@-0tY=5S^XtCw{
zOK4wU>}e!uLPbaf!`YBS5M9;sr!g5cOnW*>mqEQt1Vis8%LI3zg7O&p(1&2vBW(j)
zz2q5W(74<(1-c%c3x-`UfzdCvSJj8hGfC`I!ZRfpX8T+)xVm}(LjxA#<N%U}WtJ(>
zRq6ZyF>6k`I{GXTPoo|L>j=%B#nQt-jJ_-!Xrz8N`7K#uiO}LjmOWcFHPHm?XOmlX
z8+5<}31_g|NDh41Rmm_t#-%YkkonhXpvv_Tq^}$#$*SYnttAIV$%ZXEh*@A3ggC<C
zJU4DXha_{iBM-r_6%8Y}7{<>0IV%5)VC;-F=u2(#Tn1xC%AWf$SGu!i<IX65E?G8Q
z9ekdWV7}gA$@3)5N-PJ0O|Q!5#a`lvBSCn?=gOd@6HQ-kVuULUd5}Ptz_Kcab2nRZ
zFjH1%FzPem5kDTtg~KqI*aFgwH@&wkAorOmE(`jAISJZqzJTmqGd+att1yQqg0rr=
z3XVGSLrCi90t-1&1}HZOPZmn_%~e!(rjJ=Fr1?Tsjqj5VRawU{<tB%!Na(wVsz`(A
zj*(#44WNgSBUHo|I<G#$rL!0e*<oVS!l_3tj6FS4EMj<L1jejn5t&ZaQIU(7)0J0R
zzDPpOa0k0sMa5n(S<HmJkMDZ5Ws5~K;%2=((qPwMOT>H_TaRRkiax?MJW~vg05MLN
zs4TrV!ayy!1|8>cO5ir6!^wSComdo2j21hB9LLqWlOrU26@6GR;M6ETLXy>*K7B4F
z%N^*#ODWh?-4&pvOgIY}bHT6<^>U<Wv78+#nXtvlj%2bhiPMQ(!P{IxUdCX|p7&TH
z-1GyVOc)zEtPRU32zJjHyLOaVt{t)DD6yROZPQW86U>j2*ytwMT9oyHx<U^tKfKTX
zwNm>s?RrM=H9aqjTVn>LeLgd(`&9aT21A*zS@wL*5G^gmxu`<#ov`IAaShXK@&e|G
zY1E7T0?LI$(p<rT#j_VO_{dW&d7)CG{Vfxo^dpU2xH0Kl&!bglkR(Twd3B=YTI|tE
zyB%6yE}F(L=CorN+`A3yO(G2a3d7_WvT0yHlnc(f@^q{^*LJG+$+1d%zwgbDC4F-W
zec@PUi+!0}#*ULX+fb#)ksPZ$hKYhvKFg02PpDAPU+j3&*Rd2Nf?*xH=JBk8i?6U;
zfm2We>;#F3bu~GG@vAmjc7hsldIoAhE{c%`Y>X$80+p4VsEXL;MxzszH=5k%55-J1
z{q*)EQo%ri`Qs!}p_kx*FH;PTACW=CbvO8*%p9(~&=R2qr&@Nh@|wm0P=2y%!jCsj
zQE<&yl2e!r#@H-|&R}zvM_lpau~S7ogJa|5)L1-k_Uu$9twUukNL<Z1O(g6Y$!W~A
z=^avV+{HdEiskm^giN3pCW!>B!UQNdgjqUINDF%=Xa|%b){X&H$*`8c0r4Bg?k<nG
z>ejC1<XGK-(JkonekM3O1>)t(96Rk51w-mwE0_=~m}M(Sa^*tHS1^}z8c<AarE+u&
zSF%!ayN0$>WvBf(k2Kgj;Jiuft3!U4m8@dwaX&rHRx$O2xBn`o9&cbGx77^il)73&
zxiLY*Ru*HhDCUA;5(DXU$_bPRoe1md>5_8$Skv;;B|&|Pd*C&s4(D33hBVa~)LpiQ
zdHO~=Un5zD>aG;8W%%69=!<K`ateofnS#002nM%roI%!yA8yGROcr`KT6PAL&5pYg
zCS0X`=*-9thtJ8G%5v)@mcT4VOk<JCBOdeavW^KReVtC$#oE5x?Kz6TfWh=A3<U@S
zdOZ`sM2k)!j8g^7U+XE{tSfpkc*?ujhRD+`&dCNzw;obw!n-o)j=AtQT0J*10Zu-=
zD{LfF-vwtIRo_q#y=fIO<$LuC>t_sgSt0vfukigvMRhd*=@ks^5Yxa}<X)ff!>{D5
zSf42ES(2>k^P_?>`{LM;6w^MgH>uM=cXIn#wn+&FF^lHHEtl8^6+VS3Ne0D#5FL3?
z)iB;{Zna!A)gfO4H$S|WuR_>Yb4Vtem3|vG=GkTz8Tx$_t&?wNIfpSLoK3zhm+EW<
zVoFtDwN|xUfeTK=NhZ0$I9s#i9Hy#Hz@T*wQ(d{#@^dH`cA7p!wkYqa52-CGE(~*i
z_Ls$|314ltF!d(%g>#v}R|ZSYW#|%&5{1GrS+o3HW;*42rSrsnjVom5kr)Pq>^zon
z*Hs$g=SfC40iI9T#0EYQ9QIA-`KremH<#wc$Cz6&E>H~0ijH{!`Ol76b^&wU^>VCf
z`32IXSz3CjT*%0&Y73jw3z;9z_bd|(y|-B|ia?{}MWS&2g1+Ksg2C(kB9)L%D;HCO
zCKzALWZ@JnAA(aj;m-wU-P`68C0mUdBe_JxSQ;_~k4!-`%Y}RQV_;gUfl#Fao}|p@
zO4!I#wb+DorCbpUmf&QWJfSvN5~1%ptI8GQ%<`&|LgC_p))0Aa++c}N6^io^s~5|Z
zYo7YRQIFdLhRJ?~_eMe2qOxJK-#~Rn!L&8il<N?3O=?W&`gbN6R$-%BW9C&D1L`X1
zD6L>go#AUYSXO7cfK@%OlU;SxwbT)@#{R>S5fYD}e+dm?%82pPuJ4bMg|EiRsFKZM
zxMvCuy?ZSetV55sF%pkAF++}#xal|(4%0aB%0*Izwqy-S2MbKnpcGp%e1sNXXt|;%
zT*0_hJ$AeSb9E9wV5Y_K#ih&!-VU#~{8I9345Nfw8H}yN>n#yRSA(Lll_YM|%D1W|
zn?pQgvT;>f&605ybGaLs$Kz0`LdVVH$3)dFhZCaWDmfvlmY=m{ihx9~v3!E<0<Dr`
zXp)&vLJdhKncK1jmMJjwZns=;?mgJ`Hzofan4gm-Y2A|FRL`36B`9yIDDIosBDHEq
z7+Nhf(-uj>s>>}`aNud+WfGSLf#fpfbRTLeyG-<$J#xX>2~_?R$>WA{iY0RI5?}cg
z1*@EEvx*wjPS&PXKjYX9CethdjGEcBD!r<cn7klDmDr3j$843%FfGOpsxt<UAOi(A
zXkVUHr8F#=C2@lXJ+tJ|+>WJeR)xf(4Bp%HA>Dl&6HH-svMffR6`|6^h*@8pwoCk$
z@Bg=xYt^m3+r?<ekM#5HlmQx!T~360PA(U7ZX^^sxY_a;+VIJJ2Pt5zLP3RLz`fcs
zVQl44`@(Ez+o^h41wnR7sG9GXb~4K)3oYNtCh5U<!{@v!_yrZyr=R2srWmhc_Q|fG
z>{Im1$gdz*5W!P~G(2%gUZnEXt8Tq`5wjS^I;P+u8Z7L}q5x3%t|Un~)Uqojrcc2U
zo&xvD7c+G3e$3Y|mhey{EmLq67S~+C)hXX&{TE3v5u(os!}kfjMi|-(%ziPvIf89a
zCTU09wIyUsj71H_7RKq<c1wkI#+b4D(Ta=>V<)5ZJ{gG9aKo)n{Y*vawc!CHiF{!X
z!s}c`PK1N&tLarFpM||36ZH9^6GPoxaTS{+w3=><=wHQ52R6fzh6!DV2v>^)8y#OW
z6#+xNrB|znvLnO~sIN8aC<)a8EZC73ro{f$%wXLrvzJi%y6@z|OPCo<^p?b6Uwl*H
zX+I^&gfYj0d&D)Ik>xKT!`aKNFLWs;*q2JKY2O=%m#V_nLUh5oo9mVej?7}1Me3T@
zT_;ix;dJ7qY!5%&^#j=cODQj!90TxW%=FT_g_kiD1%>V7Wy%&h;5JJ|G6XG6B-^X-
z=IMw8MhS4q<HwZirF~K1OaSt~ob2J?8eUG}-JvfL3|k<);<sX@Q{WUfubr2xH{jsS
zas|V<X?+D_TMQ-j734N|7i#<!%whyx{}s|)>zB(Qk-kEuz)Xg^d4-rxz--Zp5pWAY
zPgg{EG<%!%U!i#u3a-!nc1`RqV+|Z|glm`)T+oth$Sb%>CB24{xHnmLjq+RVXYrkD
z#0nNB*X(nVg=usgrG^y__47)mc1GcqWDkFrM1kYpZ(hlq#+%g|e62`;;of<rGJ=+g
zF{R+0RqMY}<%cSxt=Fq0rXP%iS23|0NfbKcrq5KNYgfW$OBMR=rgdJWQp3RE=PZ#4
z8r}M@iuu;ZVPbeSIb!VyuU7Ez7-*7LlTm;Io4#84yK0dsc*bjA!Bg(bl)suH`|*(f
zSzlyzG`GE`NR4qTT(7}<HWBXKi)~u^8f7+!Sya#uWjn8-WEa7d9pig%u)dNt{0Pfl
zO9Cj!K?t^<<h5i~!_ja0T2%mgzn{x1BJ|#Fo!3&N(J45<<S}?7szt##MgZ&Au`Vu~
zZK8DHb>unjTVKH@&XLmBsbD^vW+KO&<s&vm)ZE_kI+bCm^@X;<g7JFQGv+09``42o
zz-IS)^22f5R!bv^w6NE+9MHbKGk5eAs#AUpo4;N)H#%#5rN_|+6rwh{rpqj}$#vJ&
z(65p<TM0SHmp*NiF|`w@KT1>IYMnNjHZY;)N{&qkrefjV12{jmH<0bv3=YYHFmyI7
z(QlA7R0_Y(Dd`*33ed(;u<Q+z7sI?0;VO)(H>gHte1YhT9hA}Q;n~M*;CqG6^F|V(
zJNq%f8(F|r8!UYzIZna0oV`(5)md#Fkqm(S8#RW9>NGU;#T0W9+WA^$5U?g%h`8p*
z64#0qEP0?xix?<g*|p5?9A6NxCEww?oL2H{HEuDQX&LNa%etT%ppInMsq{WgU&oAT
zpidNB#TZPlQ%%5h4WE(gC<K%v>_e=hFBs9?YdF7-6;R!7eW4Xl2;M}V+uafr-lVL0
z*I6QnvRf*OVfva0x5m+pIwElgfn2bGQPmeks9-*@>zN;hWw@TXVE2e6;ChKWJ&OZ~
zR1whqYw(z4)%f|k>++pQSM&8czn-F1VIu7dLb-%zCE4O~0tSWfN}f;(qtjw0db1j`
z_n;gd->lrn-HOqPanp6^G&Rc=heHG$>Gj0}OEu=B8`QERbgc_FkOi#BSh<rMSjHVF
z$_*5erx7<uF5j7TL@?X1Tm<eu(-#cA4m>`|0L~rZEh-~?oD#uiZHJ|A5ieg}Ghyt!
z@3f9c>ew~rg4<AseL*(~+l>-h+Gx0uY__?5Gr3U%PSzuzAkrHt#3HzN-bf)9;4rXr
zV{E9Z>&r^ty9-(Og`q6swEnH+i@D+@7T(IVO&Cm)7<?7TQw3LDdwVPMS-8wPLZjGa
zDLP<R+>Z>rY2QX(JD}VJLANL-Z;K-NL0YPh@w)FEGZCSeTIX$Q@tVwbxngSAiTB?|
zVS8PljowaCCa{LRT{S!kOGpxN{c<#@C@zBnq0rtPC}bz%+fcIn?G!RBhhim&dj#;S
zB|ofj=#}qaE}K?aA_Tif`VOXo@d9qQVz3EIuj1DaSg!QY%g3$%4i*?n9hliWRr>Aj
zTpr%3r0_FH6kdh9E9TDl0SyYZq7UpHQTd&BQkI=q0U{$hKlI<J;=@#9)*;hb%*sJ<
zhxhRgDK@*AEA42)s>c(%tRtDQGsYsIG(6dkbU4W1+|m~hpZDN4%ruwHx<`4qNenTD
z<(F}jx-GnC(e~+0D&<Dhl^|wU>j=)>0yEc5>dWrKqs6YT$Zg(<-X-BCe2ES3V)3V8
zka?F%2~$G)E){+l(^aP6Fg;<Nn2=LOuJ9_%1AUQ?*GEu_catBE?!&v4&CC=Oj6}hU
z7h3vmW>ZHO%Ob^u>+&6i&)tM+?cK_LgU<=?R=0)fL0{JQC^d|4;XR~6hpJg39K1WH
zG2DHYGLdig+IbHJ3#VEB9@Zrn>j$i_@DbF@dnGW=D~B!U0|seyvE;qX)z9(Lf`}WJ
z_<>~-GxK{X&#vonl_8H2*gIficrWuqx4H+m4tpPYg$jm2gxIwuf-rhp`aU)F3@<%o
z?^9jWarvn8J_<O_<GlQREYA#PH<8fy%-*lWEvV~g-1jT7_uS<D%(pRuG6)77Z)Sop
z-&rRTPhd*TBd(z)`jI7kzA?b;1F9tK`fs)H0hMJlswl!;FSS&#F@lj6p?97SP+nYq
z$rU)|D{hR(f@(KQ)(JFc5N=M5qiJte2XN0l@A@~B&32ggV*=O7J2z7h7~w4!483cu
zAK`j9sC-b|ecKd-b$U^R!w#K(P(`b`7hI+f_yVH+J0B!_pL%j7*i8LOm^K;v5an99
z)WV0vW{10vB_CqGSeqWOR3t4J4?m=OI~}Y=fNeH3XyGsgb+<3#As8DzOycb*a`>=H
zu^ozO60slurXOaOI#SJI#7g)ac0R1`Foe@mtSKL+j0=`oUkS!_zHy5L9(E-&+`=p_
zb?2Mp7Un#Ja-<4|;}9BE!BtpeJGYSIC6`!!iwyVE&2jm8^hFhV@4$^%6&GVHd?dCe
z3hZ_$#WZkJEfvI+jvE0K2$L+@v!ihL@XJ3!*{a-AixF($^63631sPmr;iIafNu*6a
z%8aY%-DwPle;KY8h>s8Ljz}(nQ)(`_3G+t(qpFe#UMuU89CoUEEOZ}!vY!S0Uw`hW
z@B|OVJcGGA?J9AL_bj>;Y|}>T2&FjD5ks7Lj#*#C(H49wQx4M|Am}4Bp{iJNtEy1<
z?>#g3tn{_=Rtnejizgjnob{ouZe_l+7{8GZwvf=$ZzG?F?!$iOHi?N{$kmo8+-I3o
z;nNMjPm|q7_B3X8V#EwL*-9{XGt3(?EtGD%ow=TlX1JZ)VI4)oCAUj@S9@+(2cY7K
zt&$QfTWX!#nG^O$SO7%tUP^rtEM0&L8zS%}EZiZ6I6X-2h_zkCnvllu-n($8L*y&a
zdpioAazB_zF^C~1=w8=$XJq0ke7KV-;Z=_g8Y5_RPVeMs!lutv;p|Q=(Qf$jZqgAc
zj{J}=zf;AGM(O^Yly~kXSUvAjRzvOu9`0iJRy1UC7prbylcjet8~3h7g=HcMYq3kZ
zOGRp6iOwS%tcqv>r42Ypx?9}0xJzcZn=CO8CwG%2PPKg0rFS#eDeS_tyHyCk2kd%M
zM^RxXY-EDlU=``#&5TEQE8!mHz!SxwkLr*sn@Nm;#j|GVJ#nDJBr}SBbdM$?w}wLJ
zx`zS;MCA8S#&EFp?@{S^`u;KE#tgb(_?VLVB8#r1808_1?ua9bkCD&da_f97mJSnY
zOgcjMpuWhr(U5g7i3Txh!oAAZM{5!>lrT&0RY9qDWHEwvM5U!$Y^0f};*hX^ui6dk
zJlf6fqd=|8E!?LJT(wOU3UAf)K1xtU^=2aA{M!+xKSVC#&Csw!902Wp^1ZZb;ePUk
z7K>^T4ii3C-%r8i+>^zKN3(ZWM@bjJ3=~6Q6|?^RYK-L=1|KJEdaZw4rJryMbfOQ}
zM4?K5JQl~Dej^n(P#vXqPpLeHL;rz>r3Ag({gmJVvEAklKOugMHk&m|9w6J^t-ew|
zz%s7bjO$8K44SbXpm;;5$2^A9iK<_G_(djrP#tet4qo9wrfEP;O&(NVhsw%TJ$+ET
zaej`HJjm>6N$Mz~G4A&@^9LzQ6=uo@)e$Fj!8B)TI3Qwk6LLkN=c7j@xoYSf;529Q
z!9mM}u3TcB7+)WO+nF$}6#bmC*8OPx6OwJZflF556RMi~vC&FCL4IpD<53mn87;G!
zXs`#us{dGGU-Umgro*s?_dlVkfwvRKz(W!f*SGO5iSlc?eI!xDu9taJM4}b$>=3b!
zn$AO%<A@{hCWVqtxG!e^A!WhNhxK4iS!{*xYY^tLlFTvnMayuxTJ7W4VdvC7Q1xk0
z?98!%eqM$QbA91#ViL=FKQf-id~FY_@}U4ji3$%hOPnL2AP-YCI5NVBo<1Bq1~zFh
zxMVTaNR<X~Q3l+jz}gupH)HO?Y}FS>m?UB2eT0HviRDwMewihYD1UdvgfUa$7c8>u
z5mgK}sZf6(QB6$2B$huST?xAtd4Bd42@M1N_md>--G{~w!kDw60eq4<FI!@%Lb12R
z{QXJh0n3nE(e&Z_1}qSw3NyHs+gHTwjQdgBrzDA=ga=`|Y9ybMxHv4w5)nV3reUKW
zQ*6hrk<O>$suvIX@=vk-LIvGRaM;f!KTTfli2!~3)5`D+_q6yl$(wGV`?RX9ez{Mo
znTTMkgfT&>!-yZ><?#cP>fP4=G%LYhXZQ@|z^3W}Oe_yr_>3|^16uMKm1NSLmDA6t
z*&CBE+fCpapychn)jDDk+zjBRwfr*@2v&R4Y+np99xyL{Rt#asKnMG5(LmQ*@>z15
zg(8;<hSRawMe;HCROx(HTpN>qH8NM?<|u|~Un!>iv6jz~&rlPqg&;gCEcu+u<T`Qs
zIkCY>sJF=HSPyH_uRizKprO{7iw=FxdqiIxaK?r6-_J|#+g*+QJoB5o8FRaW!&;d7
z(ukYbk7u7}7MT4zpH~gy`i>vV=bx7(I1zHw+UH|`K-IWjXkSoO4f&)R^no#jN=&|>
z{M~acja1mCqDj6WhA@!eAgl8QiWXqA&SL~r6Q~z`rNJ#xw>^H5%;rAewvsO@OR7tW
zB6i*2x@N@Jpv0N5`xRN|i^^gE8iIm@Z~6KP#<(*3QXwnq>Put~Gqu|>^dXx{`XyFT
zSc=P>;<y=;uPDOw46du^F&s|S{Vx^M&lEZwE>AHBY^hNI;ou!45q8f8OMB!$3SG0u
zD&kT5J=W<lx2Y<YkN*{cl~!K?`$Om>woCQ0#rFwe7xTj@Kg@Z;*vw%R#&B<;Y?nlI
zO9Fi6JG){_Ty1%TE1`-+xB+AE%jAi}7c}*kN!;+80$(N@zeneNHC2pjNSR<*yVg2N
z3o|8#_?KBapY?G_)>jPf;-d&(k<@-ED}04HxPyA~6-}TxB+@n|{ffE}uDH3wfhfcB
zrgaoS+d1aM7=dxq{|f7N<0iaw#5{Jr96c@w2W(p``6~09z{?=%SBp*#?*?Bx6t&I!
zl3;5LCsbcm^+AvJmh3AgRB)W|HD)%j&cfH28SdKGEK%r%7h3u?iQkyObc6f2q^bKU
zWk(nWW*oJDP3?&laK!pwqjc`nU|(lAx;7k*!q=G?b2#R~ud7W^W2u7clQpQD3JwF9
z%>@USZ?IexFkjtaeQ|{8(0l4PC>&2_FqwZtJZEMt5gdnuXZj5m(D&9tTN~E-hAPcf
zep(;CQHVc*UKBsDW5#IvCR49R^?Z}uU0q1NNpjt6OuxxeHR)Ba2)N0+s`E|N96B3n
zD;F8;bXc_&<Gm0E_4X|ZGld;4?W^BnMpajCzeV1`H#(_;!w{y}Z>cqXvxmm&h-$4L
zOy{DS#fGylxD9qP`!-qP%7%q+ODfD~c*iLDHcJBAo-eGTn#JOqeVfG|L@B=gKhpWU
znJWdi_}144)DH|Z`;I96np_=6;Nd&uaSA-(zr*Z9@7<OPj-hX4-%($3yG=*IP&q6Y
ztV8SSivpTDbdT?<Dz=ZIBMDFW;+%X}bpZ>k+g{R0=J(UG?@A!d<lI|!V%!Ye8uIT_
zq!!*c>I*j9Li|03HsH1w1nCa=J!S1~^QkaqJ6!ya;gh~Je2;})bdlvkGf>JSH#gIN
zpDc#`7(INS5~IJ+8#4L68U()Ra7#$PPd-rXvhTBL+{pz;clv;a=`|z<I0NIcTVK3G
z#qH-mV4|=c2lf_zAkMmak^Df7uyCQJ5k7Pu?xBljXx4ACb`;To|6Kk9&0`qR)AS?j
zD-M36cs`A+25=l0`jmGm8o5tSI4E~lv$U@QyH=8kCg6ZdG^02;z<QkP3pdr>6{|0*
z<xrY`ShQP{ZiwNB<h)UasN{z#A}o>E!sx^8Y^X7PPy*}xkb><*Q!6=6e_i+ND|{I5
z3_VIljcvY_3i`m!C{$b$KcMYn{&-aN2EVNAQ5FiF9kX9d%{I*S(ZuB*6hu@w)<k>!
zh>06hu-XfTxer+KBXuYaI$?BWxYhhf5>XxNMEc$hmj8$qK!x&S)>p(drcV1Y8MQEx
zSopE(1$Qu@&?q9zG~q7wV-;YaW?6)%!(jho@(d$9A<rW{?!#L@GV((RJQcviz6%b5
zb05I${og78>fKTW525#E|IIp|!f^O+3Ijdf?S1;t2DGn_Fj04}>z|MhHWznV5N=Sr
z{D}rNEpbqmenM{B8}50P{e+pEw6Aqy3co9n|3t;{i#}+A_ys2qSOV;)EJJ+~s)pbY
z^r+;g<g*H$NP+87#!P|Do!0rOin!gi@1Lq_NBlIrA3tb@wx6l1o;|4(yYMrm8pg_;
z{EVs4vXC$R8FRaGG2Tib+KMKK1YF8;rL3}#{!A_A7eIebhET+@zWrQcVA}*0;OC^7
z!j0Nga2%aAizKkAV^URG*uS7&{9HAPdpPcf*;iC!23tP+1+##hFgX2!WR0Ej9Z7zn
zTE#0fIJHhiIOKkznJ|nxI7Iw~%675kzmSA0+>5@i54am3L;EE&fj+j)!Y?WMV!XDN
z{E|tbEv6#C0mBHMfney83$61@iH%tZj-Qq*0#Bp$MeenZmm-uah|o)ar4%?1c6BKE
zl^V$1oxJtZUr`Qtc|leBmHLJ6BNPlx6IJyq7J6FOIr}0CtFV9gHR-E35DkKhaOj@=
znz`##E7b=)$Kbk`J@(<+(yy5@x)3ICML^?$hS85y1AgboenV!s7m9T~{D%3>-3|u;
z!7w;$si0r1$prmas`DGN$IW$^RfKWefsybVRfoIPLFu>Ok|kz23%_LwM8IEI*k@u(
zf2-Q$k!ki@wnS~jI!ftYB6*}9KtJe5-1Pfh^JE?dP(Kk4V=9^_3p{j@&Z{@!*dFE1
z<|WynjA5O5iQ~_><&hBnI=)fV7oXrkNBW(_>)nbQu;F)crKnGF2c7(mc@4l}=y#H3
zJ(@vq7CpEV6T^CfY3X+qr*}IxXI<+TeBX&3yZt_9;vVDS_mm8#3uu17FP0|xQl^Tl
zxqEc>duBQ1+Y<%jLIPSt!G4XeAA|kcmHmODLYoVJP@B|hmi$4j=<e>Rf}7CfGSRME
zXPp@6R`x$g#<i{?_y3@(#hE;AXxSgh`%-wQ{E@sT=Wek?fmn&sKPs0Ih>=CS0Xk9V
zk7^Ov7F~bMl@|V~))&O2VD=~0z+(@mgg;3Po(Ly@B4^BD&;W#KX3ZiI7phM2;roC*
zrr<96PZYbd7Ehw6I5<tf#y02+9$~%(N&d_{criNtv#JiknP3%X_Z`8o0>)ku<0Tjr
zB2tXwhV);UQv;*mFXXoC8mvIcU&zf*Em9GjwGJ<@sIuU*VjYEUT4%XPhoSlP1;aE}
z`@f3TeYiu!lQ6+15zn}(PZ%l{#_wNQcTia1$k_QSbDz5zO1#3un7{7T|0`vM`NjT5
zG>n<yZ)z#{YS4V~H?;vhPt(6q2;7TD27e>ZohVU9F!=Nn;Rtl!m=g9ctj2#Q^;&qe
zgukmK%h5%Wzq7^t@-a_Y(!VP=*oAQfBtrOHp?Z{l81q;jY1a6mV_)HToClB2FpDs~
z)51T<1{ccUgqA40*@6dv;4HL=>>n&J0y{#OwDW&F{6s&`<V8NL_L%bfN>P0T>dHUK
z-}i>$pG<tnA(s48y=fQ@Dm6?0De>!<V`G&4^Wi5sH&pav4wIdKs&cU&`moN$GQdBg
z|4-J+YCfa+FR?*sG2Q)(5e?sgB>$4)>{oF(DAIpXkd>?OYz~oo_WVnExD6B2O0M)3
z-$wOCG=!T&|Bj8|rTMqA^K*1d{#{6j5sC2}Ke7J#-R3MN8+6YSMPcIxD<Vc)<kA}V
z!<9%k=1;l(hxLZ@Bn$shuCSR~@}F39e8ndH55-@Pg8WC*pnC+m%$@&8pjMsR;5@Pz
zs>5UwNvq@9$-qp#8)_VG86|sU53hS%dq%GI$m8+hw>Zl7$nJn#LKD$c$9<z#DH34w
zX{q>Jw8UDX59|FHAY!+hx9t(0e5;dm;+O9|TGlP%CR~}xi&Stg?J35UUpUkD?GfIu
z#DQAXPn+iLkv&~zf3FBrq&<$J(v_<8I2Om3SrkVR!A(eI;fBxpiSXh0vr42HaTV@y
zEdDr*3#}N8i&K_8jzw_yoZiQ*tuIW9^tOE5BYcto*O22*vKup^x#4k-?CpZO0U86f
zYrV%w^6I!u5HXiC@LtfoBzFx4ZuGmc7<2ERgLipJA5aT?-@2PPVCM42(8`L#i9YuF
z$!@F&?w%@&aFZF-y4}<WvH-OrnL9pa!q{<*T6;GZzlDuhXE*i_G<P==OyF8WE`qS@
zwboOx&wSQba05eVK2neRBgBuFG;SOzMLdk<KRlk<V0Ua~1>^qk>+$ZBWSn%DVyh?y
zmzA-G2?v}ZTT6)>KCLPk+t#ht7RD(%hQZ@mciS5N>RneK!L=PcA|;7xsH@)NS;AFN
z{gkmk=7Hv&R|5G~+3p$~y|<(7O1m?4!18HfcM6L`fA`QXD+Y%GuzBsSKGfV{m6%`}
z+F&Y9@+5DGCalKfDn_V(Uwzs!VtTuElmN@X>8q?8!?$8p5>uRl-eWz{tl-;TebLn6
zbTcoe?(gc)qwPVt-95@m!qpj6P1u8y!lY{DJycHJAkxPw-ifa$LBIi!rICs^SX+Cr
z{&1r)6IyhlwS_K)X>SjWSlkn4b9Q4=R2bYp_fSjYRtl`Py_kdBG5W&0UJ0ssF}o5^
z*gS!P4q>ZZ65IxV=I{g-3-@`Zt^5QH3fNxVwIERxoh~X$;V&hm3Wi;2+*-nT+!t?e
ziKGr+;<jLD)~xdc)d_gkt*h_}m?d(h9PvkEdQW&{U!Nr1PyVr$xgWDZ!EHYB>%AH8
zgsJ;XVjz`-aiJMI?mZ>@T7T|h(#m_XUU3B}*;4{c&yJ2-C4S)UIT}LAr`!nL+Eb%x
zA@;O;s$Q`kU<9=l0V3-t!h86xmsvLoQUOsOaSi%WPdT5m$ok4`cE%0N^WrlO?Zx&|
z25#4?TWK#9s4<Dnq%aKIV^)sg&^qwKnNql`+q6~o(sbw!OWahXdr|oMXw_OV)gU-$
zG1YY4oj}_nn!9<_I(w-~Cf&KNs}yxVP|rmZHq@=BV7RAYaaD%?z>l`l^9opXd2x}z
z_EsTQf=Ov_RVCUjD3}X-S;4+pN%m&`LsR}JX=QJgExvY`Dg|y+k6BA(4XzN`-fYl`
zhP6e)yLO$JZw<%OT`_9NkeDmGxtqqV7m57s!~WjtChmnZidIn?Y^Hq`vVAD|+$}R!
z+J_?I*-kWA*oUQqEec~~A2EnK>SUissB`JmQ?*msN3G>99dK_E(}3fLmiX|LE)&L8
z|2wU{56d>N!a52a_M^A%K5@BjO!LM=9;NWR<2~Wc{-9D{88k2#%?mE^9}ajT%Q6F%
zwe&<a&N$8h!xLF9EP+^N%A%RO8Dm{Eu$$DaBAP%`b*czn6KaWm78`|3urY!4PB83x
zg>|0Dk|Uveh;$WEhn|{8E<U1qkqfrx{U@r3sN}jkjm<w%6@lHylPL3QKT0k=N!@kU
zUwH^mBFWCCm7m1Scn!af2lYe*+s=elo}?}~cfX~A<5(72kwnjG#t+_RXp0q2(W#8+
z3s1TFyx)zy{18w=HHM)r^+ZJH=G&~Vm;i0s=3|21d+>%IN2(7Z8DL8LQZU!<f}nfT
zmG@;~anTLs*jEiS?hYxH_(A<FRe~@$Y%L|2oy6psi6-E2gh^}f%i6(vcO9Xf4eKhM
zJJ{!f1K6GS!~$obL)PDyHR;bzz^8gX#(2-SC#&nUFq4-ec3W+DvPJ>kvUioI{A5)=
z?jpj!{mGO8TgXaG;fj13@h0@_mM|{M)bUCobC`8gVOyBLmC<=Jy9T?G>x*4+g=&xs
zt=eq8$a4tlQ2)uSS)7&IykOOzXnG1u2IG{AS$Ycdg0m?M6;C0(zd=%dih3cgy3(GQ
zhzw>mEEvLgn-Ruy;rckN+btzuyd=fbi7}5EzoXQC3abLGfuY$E)tS(QyMhhaZS$wZ
zDTLat^~4s(I@o_JHQxV6SD24^Z$aKxqPPoSW-JMwimMl)q?G`7-Q*frE~&jXKqE**
zw-%eqN=fY__h^Ot1+|vokZ(2w-Lt1%VtwEQPxDzvQPVhU><Ttu*USZdQt1hvjvm&J
z0{FZ$FN{`3M|vvdA95vDFr0z);i;;5sHd1KF}XgK;=?EBb$CNk9RR8hzF%XNr?Oks
z&<>&sLr^1H!b5&nDSIkQIlya%ZIMsRT1Sj%y7l4ReX6RTLl?JjB~BCf#Cn2_5ghv#
zd0?M2FTQwj6N{4VM-kCErmeIeMOt!&1%-z7;9uU4{Q+A+?{mq1s(Q>zBUaf@<@HlV
z->aqjsr@j!>QV02eypbY1ngvy<&~Js+rkw;8tI6nil^1Pg5j{k@K`1jj@P$YZ$Fg}
z#z(*N+%H7K5;-r-w?A!v3V-B<R@z@8;;;qN(*Ei}?wV8HUz}i2ppX6j%)?*8s))+3
z(_1PGEvja%{aGOQ{?H~p6K}jFgqgoB2DqJPo&8x!xR&mQ?QYD-cWtoX=1R1Rj;OsT
zogds;KT0<NJ#;?eGthhOX_OjP3heWqCaHZIftfuhD!{MA%EIt}fjQu5EOnz+!)#tr
z8k{t91xTN!-io=oB{IKGTE`=ztg{O)vi8%|$XGJa(>tPvBcpZ2$*mtg&E(1v9%$3n
zdzyMQ-}k|k))(ojxZ|ByG!Bz=d(xh+W|{P-8cI)Bp>S2!dsTS4S_#`yD?gp0;HHzm
zvYtGhq6Zv3Rw4te_2_B}$5R2;dOAz-;tTQ8sn`s`EZU9<TNp7h*qtZ3;*Lv>o^Kvm
zPP?I`rxcfC2IwmVPW+*S2%nAP5j>hoX&SzKJVSY{a(^XZ*hy<vj@VzzN+QiVu&F#l
z#pgBu^cgH8-neqImdF6HZBh&mrgm8S85&@IG~+L5b)LbZE;`A&VkZyfTw(>hpwm-&
zc)|GG&{u@NST(PRFv?+U&tyT|Gq3baHU*A_c=<OxQ$4}W3fPK2lNrvUAruZrclY<M
zC_UT<Eft)*3A5KTA1PtuqbunD^IyCa8++H~e$3N;CTqgouP3bYOjYAF7P_u7gO^V(
zk^nakOk?`+!(5CQ(Zd_GVe^p+M-edh96(9W-fX1<RB1JAwcLR-91vRyqo*u9-NZ?G
zascbVALghC4r0bn1p}TAZXG~D{W%IWj6Ptc^pP(%Fz&!ycz`CoX`E^C&CqV-fYRlW
z12@gR_%S&HUA3<dKlsMbSBhDfEbLhn7A6bK$IoK5RrSh8coxO=_mIjWf_G0B@8)VG
zu|uPqJ&V<Vd;ICMR1Ml1pmD@4;4L9^(@fm_hJ5>3G*(eHkyB(xkzpOizo6UbbHN=r
zSMEJao!KoN)|WI;RWMj$%3++2KAR;Oc2Awsvnl&pygL;X99DWcp3MSct>jCwB5GhL
zRutj39mtw0BEbKr^=xH~84rg#3U6T`x5X6)EKn7nt%Bp-J-&U{Ra_O0hIk@aZI0u%
zUd#iBnrP9!h-e*}k0PD{-)B2e)jN$9!{@C7nZAlWKoG=6)5-^`e{bh&Zpndcf&u8e
zF&vkWFn6X0vO=&lV#o_ucOdKRK$WTKK4|Tj0bX3}9H?^ex;cv5Rhm=qXL%tOHgQPR
zD^g*|M~LfU^9RZb>E0MR60?Ie6sqX+rGu31w7-ZM4pIrQibK^ZD+-$xEHa5c{6G|%
z*FloW&wTyPc&c>tmKw2^*wYavJBT$k42Ordg2AC_or738+hIaHNHsg#;#!o)Pjq}=
z8GG>)hG|~g5({|POxk=DymuQ;8SOdbw-VdJ=fpngE}TKZu$YyfqmoV2;*=<SBZmBQ
zG;FE6c+IEJQQi1aZ|gadtv)j1>v<;DaAbw9(ALN7sQayVp2Jc$V1Dij`dXaF(3zdq
z6DPdp$d$OSm`PXbqhMQjUD2Maik)u2JPswT^jwx^$AtTCgy&K;E?H$o_>B!r%K8|j
zV?`yh@V3ERY3XxSa`)!ICPh4;#$)gc;|R27?dP%rH>|Xdn7P`7(!^}m`Wv6QKCs(`
z5xA$s!|plJe=dt3a0_le@?7U9NzYU8!KYa1c`6rtP53%gP!ugf&tnnwtV{B|M|gCE
zWyp1BpARa}qx`ToxnEQIJk|l;M8)V)%CLn$h!Qhf1x;Txw0d`h@dogyb;XH3I``w*
zY?gQ<w(d$=#iqP3a%{<KHvc@e{)#brvB71Z=L%JMMwNo96}x7%QwZvHINKeua!?l<
z#5r#gSO<%+!>NQgXx!t)=d;KVi@!V8x5NVWc}!&4W8D1}lD3#Mu#4%4H;%&4I0}X}
z0Z)2M8oc|3h+dS&O=Q-OALES?^fuA43U0&m-zpMseqwq)m~z9wiWBvNDK|7(D=T=q
z>0Ss2tDJP0sffny2C2d|JGR6YD?6Xx$V7G}-U4pN4C_#0F}^zD97b?cyL+%|$ImBx
z$C)3@62Y?QFGBRh1IKmNkCNdDfqB90_{AbypgMP-^wI*#4u@6?3sf)Ig~HTXR>S}r
zH(9`<%jW``a!dltT^cz}!}8P;h60Gu5?KxKRmOIts^Rdv6EUvIT31w@k1+$pm<=%N
z_G0FkNcsy@i>_?heB=vHGyes&Ll}j14Ox~B(c}bmN$+QbLs&}Q#VjAfQu@O$iQp6p
zToDD%>u@x3h?*C3k(*swksJ1G8p4^9wZ>q5RJN!kKi^TfyYY9$4tgjGqu}t+-PTj^
zYz-Te{vjMk1Iuk*1XY;8Y$1yQ<r8w07OG#p)cch%pFS%uR7&hx7E01#-G{1(C}68+
zsfbn$T1#l-X3G|;Z|GpUtuObwMs;@L7cMqq2i27vVTDKK#fQJE)LY2v3pgw4$8ZE=
z)GNI|QUI%w9V)7$PPNjZly)o5s18-9a{ndh#_?;q4##&qL{F+6FcTfhQllrJ5W=vx
zY_ry(YW%7{Tad-XdZMl^vLQF$bYc>BAnyuenuF?cXsjxn*gymC6}h+Qps6nz9&BE+
zEWyt?9;Q6&e!Hf07{$l>;1;ZK7zIWjC@UP7n5IVScq3nhlt*9XFxA89a4{3aT5qk$
z)*l(dJNn84-r87B^x?m$1J(F2*2E6HZ*ASkR*PsZUO3&bo*+D;tRFLT-Q_UpF3~p(
z)oqKIjjI3{>5DW#P+(o04U3pnV-!kWS(N^|8hV&MaKr}1rXnhsN?u~AFt3HQ7OBj0
z_rnl^m;Ra4T6)#D7pYxwUJ~<kk4WAb>52}o_CaCE;|H`mSdhdIJ6+$J^cS%X;iYf$
zg81bz{32}}<L<x5J@H}|A^4T-uvm4<2an5(DZ)nhFDO{=6jv5gXm@wUJ|I=<VV^!)
z!f@ut?6sH`=`&8e5!Y5Mo`XP{7OQ#Tj?|6W*L)6E@GM>-=|zgUk62&1!`1^8eO}>B
zAM;E8r)Oe4D=ne8YgbrULhf$ZDlbuibmMmkO9~Itmsll6U=U+389)578Ec6uOyOKM
zQ-mA@+L3CD{|Iqs32VW3$f&oj(oluW3yfv8_ZVYz%YOWr^aXrAer#)EHMPTIn@o-B
zhF$4!)eqiN4u`7*e2Tn$xH^o#+>SnWxGJEjXQ(Qogmf@Qr-!qwFgc-J#I5STeHIzG
zdjgCVkvHBRgnw@Qz(!%#x`#*I0eX@94CZ1N>ND7d73}cPVGLe@Ld+{k1E=cT=h_hx
zeArF6C4J~RSU5tZ@MpxaH$Fmz=Q)qxm`ILbvHSqrox3WL8ZP?iBB>bpd%5t7ID*wQ
z_a5lVf}!_bYagKwJ>n*%h~0SF6-JB0l|6>z-k0@+`S43$?9bxmlzH)Ve|1|bo&)HD
zrKKv^);e~vVW||z=<VyvSy}Pax-8L$d}1m~)nF~X6_GBb`2K2tOJvacQInZ6gCoNA
z))q|@H;p<mUreOK){P%{%?jF>K5!6?anOsRZgbsbDaXWV3v6B_y0T+OGGl+)wR9xg
zg8l~Bza05kv(qY5K2pV=#^jP5NnXn?u}Tc}pI}JE$}e)aL@=-tzs{j#_$3%?N4x@!
z*EuqF2{)y66#<Db@Wo|d^D65_dK?8>U%_*CVfUq^v)CD7yey+A)0bIk8FRwbj%5~v
z@fZLWNnsAvWEqR-E}b}tsw`vKhvD3oE|WMlAA_xBY#UTAcQILvoAB2Q+VKOsq?UEW
zVD2Uid%*@8Fpru0!E8@>vl|Qg%VLdU6XVY=%uDnQgZSwYRT7uTh^sZ6ULQrN^aC2@
zqm<lVd8ZxxC=DHVb%fPLU+xRE>t&W6#gfh9rbbIJ%)JsnWuo#zIcX~wTu+4B5-AoR
zZe0b#QxGak<c}jU9BlL?%4)qwAhIF93pB3?R2EoSnFYa(IjCO3VZFO_lqI|0Uw7Hd
zWyua57$Y+&C{Zbu7}MgrAZb|>!T`3!I4XxCS!Q#O<9Maax{-F>76%+IW66(cHgLa-
z9@UNP2a!HkD*Sp9_O87$+j$etW!OHV^Yz~5%dDz!KFoU0r>t?SRC>OesyYt+<@qEU
zgiWa|2v0%$u8Zn^1UvlaYx)|}h5S@$ar|<>wVog6`AJLw+;%Bh9jjSea4~*nsuPpk
z>ZiXF1P`LRVKt9}v~VKPi+IE@@5MZ^habWWI{$oiY}dT;V=}D01s7TA1qz1w5Uuk9
zN<0q#w(<)!Ee*)HkSM`g*u!E3W@UddBYlAyaC*jHbZbR!1OB^JnbHiw%=!YA5qS?m
z7-;t|P~~leO45yK{3axR!iIvKz1Dh?Wflu|KT>S*#+es|`=P#&v`65W<%JXiua5gB
zHoTC`FLh&0`GqVIem(;~gTr8+9#$0WR^L>?4ZIs@Ed^J{NAN>AD*v#as%pRRe<s1k
zB<%R2yc`>!uAslTm5UcF2GH##ojVlAw0PQ%hM##+;g<_=hv{evHTNFqtEHn^3O8WE
z3J??%+`+Ycv>J&@d2%$16h@)miw2%!*k~Qi+8AwEE2e~!h2QTH@!5mc78AXs+z}0*
zUg<`9w~yr!!)L&HM{60=U0VJKNI$Y$v)Seao9;(<3=4q?9E$xh6cv|DFyO*5s>J97
zQWhQVX~W`uj4Jf8&F?YlBJue-l;;@MhAWity=y5c)#&UPm1#MSgd@i>9GrKA`J3!r
zCB_K$ms|2<q`()>=vdyx!tKF*$v7B~HRhwp_|YEz2K%w9`E~9<TsoE+H{iDuj#X{L
zaC|HakK;P%#K%fz{0txVZ572(B~OKyJ(|7UTEdVG%4Fn^YYH#1wjzT6ct%Hgg>}%o
z#MK|e#J(YqL=!&TBOdVme5Arr6)eK@lJ64y(B*NIkIHW8IMv(=>>rO)!Dvb@#}E84
zA>5!vGzulLqO>?Wgpws1TyDf-+tLTFLb=1racWAMHPPyAF~GpFj_6=`hXtv79P8{N
z{2Y_``ch#%;jkG!t}jgL`QuoZFq-*$7Ir*K!FSC{!u28iJb5^tC0gkguj5%F^j2)~
zlH-}0*H2-PPzs#=qhB3QDORkp*71@Xrx4h?;aKH(^;RkccUxPDW=GITI^vDHA@F?e
zDgv|j$MC~ak`g}<gx2gu-W#nSa~!}LH!m6Cc#I#PqKt4HfYVIr1eOVhzBsNIZu)+_
zd;;Yt91oHcC?*f+e7>*f`>~fd(i2#qb(nomU}aqCKV>a`*oR<=KSA9en<Z=rX`hf7
zVYy%b>?&0g&&%W|ur#e{&W1gGp#}VB)cVTSe?o@WSmu@OS|}|i#;Fo&Csu^giRxtQ
zv2{F=qMf?0l~2@M;t$QmYn{o7EIw{>VNp4e^@6qC)kj6}ZO7J$EIf9}ui;O(WHB$l
zz=FM4Tc14PN6+ghCHB4W@lxJ+)FY1>`AvtOLgCKWmzeI}Hh-cT@pSzV?n#n=X1kS6
zqWl9mVNu|MQ%|;XjK@Qe$w?~tge(D-lURxwzWJ1%M5a({{CK-1O0<Vu?(_*~CG2LP
z51*u}7{LtO5tCNStD>vC8kb_37Y*Ebud-gu7ia2j!tCp_h1PYO!Mvp2h?Y5-0>KR(
zB9!8XzawtpWEQfD3tr_IKIV6Zlanb)?|y7EDkrO2@N1RW;iM<CE?U@^!~__bn85Ub
zci=Ezw3T|;Meiszyj@w4$P2v(qU8F*F`xf_aZjn%t-@nT{m8r4s6j72nT`LL3x}OT
zp%*M&U?suDORyO}MYYc7|I4SSCdV;5CW;7{jPJ0@DQe0=OC!m&8;?#=19DNs3~&ld
zJgV(-Tj{0QbWYLwBu9Bn*4@ZvB@Xy=rGxp$M^{fVs7}nvl5?jwi{RO~o@J-1i=i@0
zr%JAhg;OayOlVkCPL*8PLGo=-(N|%$sR-7wr%X>}!SSS{Z&hN7`Xs$CGqLq$KSo3&
zc1O8DrNqI#cwOud2MUh(p^D&g+$8EN2i#<I`A^k!>2BKob8vPVE6X<;_$8atX)MwN
ze#9pzBK|l``7{=A@fDVwMtR}bh;CkqAN=`~^fVS7Ke7$iRAF3U9Kw&!G2GAltSyXd
zcz9jsG*%A=)~NqTdRI}niScUuNF9^l*Up>P6An1mv%c~R1AbICFJZ6rpYcv8w_mv`
z3HtLpL9m6kEhmziSC<lnhb#8AN`$>PS(>njcnY%>0|y2zQ@ViXz2Jtb4<D`Amn5o4
zTo!>z4t^jBE3&%NE!CFmE5H}P{kP*6e&_<aYG3T<?#I}O?BSq?-?bZ{jQBltT$?Qo
zsPEu+(fp-=Fu(%B3}EE}W)zmwWg!`$)T>rnC5Gb%=b*!<N)UEkW39;d41Y>98(__1
zD~;1FrJaShkx&a(Ri$(DaxR8=&fl+0_LLBs19aiYWQ%XDlmNOfe)5mJ`6EE3<&<j{
zSHFZVSYTzv2Rzt=-vm?(`*D0l7|(pc>bP9xh69s#iq>+~2`@`!qVqEcYm4jI*tIWL
zeejVi>&67@{c-VpIcsePU{B#gO}EVV#fa+KyfBX?ZG~9jAu{Y;ODiY}Jji*{8dj+0
z{N{Lh1xt6yQcGg!O6V6AC67kCbVW2c`&OEl%UUZ`m@~ZhW-BP$dTcS<g8s99ofYa;
zZic4gm!ig?4CUhPw<2Kd=;;f_0{`WOeiYnqwqYd{6MEFTu3;-FIo}X1t&|MC8&TO|
zC1vm@@XDfD;g;uQB};eZNmfzlT0B{0=}NVX-@WxoS83NZVS0=_X$5a932gr`MF^wU
z!=kcM)j{tFK9!nBhIrV-ABF2hVQ^B@mjLUUHZRsN8DUPdRcf9pUxzQPqVVD2*IHPm
zLe2PH%yJAz4^9-00Sk}Rio&N{dlbCXXWW>S&;4bqRQqzXL&L0MMc~=qPR!b`+M_Hn
zi&a=O1e;Lyd#hN@@Cro_Q}D%C9BK1{voJ8(>L{llGnZDA9IuvNZ(+4c=EfjQR;yJq
zoZTcM!VQkDRaR3P>|0Y2xHEMtvcxtSeM;0}y|3cJ)lolIj6}n3AL>S;LAa&ms~=kp
zs8aS~&iL`MzR2qC<vhPyZAYz~Kc-`+t26iys=(EubUG#W_kx39*u2Kdr)$LQz&ewh
zuA*1{?dnSW81|<!(nvLctH3QKuKR5n1;Zx;rqa_{J?rEu*3nmhrQ0>c$gARK2)Pm=
zJr?jN$a1{R5^3;rC)iY-&T8C>{<MbjVXVXTd5sE$BDkIu6cyYPyu60FV-|E3B+(ZR
z5qwix5gp#ON3UALN*HgntQC`=i3u`WqdHyzYqM{a+R6zUqc$Z<RUH|zt{|Rz!B2py
zy7+4xy@D|wgNtHrM@g_`%<GdU-9FIQQi5SDcqPFA6H*YY!^&1(s}`x^r;L(_C*jXf
z5nkh)*)*nvYSCI78_jR#XOVscvv)g2K=;EdV2Z#20#13?vT`SIqd}p3$Fa9oRe%?o
zc%`KuvpDe-n-|=)aXo&#l+ybzp_k5J#?w5C3}<L?z_k-Olob_d>rBpIdEi!$**8Y;
z!fqNP{8h7-(!;gv9_;5c#o&4!#`hVNi;m9zBgh@a(G#W{*<seiDAxzxPlVC2C#pgI
zV3RO*k@wnsOwoHA2CJQ^nr`4F#?qOp#$mtl929|5c{GP2@L~}rl_XMackQ!srgmEX
z?nt~Ekw%WwSPNQbs!HTVpPk8y!^2V5KC`gV)paqkqg4F8obH)w27ijbiDf)LQ+<9C
zSI@9>?w!fD+c02#B^us}fjb|w#j0kujv1|M4qHjkf9N5sQ&HzWf)WW&k6{ob>%<;E
z@#t^VSK?<SnjA%mC*;vW>sTCDvT)5Y6YFsD$<|({nu7}y_J#2Srz%jll$;$Tj{>a0
zXzN9`UcG(c?G5Z_=apDj%4|LJ^B?*||1GVj{5X`s8xz9o{FJr4o~4Cijvud8QT#y_
z)I&w-qjf3OSCFfAOC0#%Rwf);b2z&a)u5<4>s2spo$j%&G71--jGx108!TRIy%>mJ
zfa*s<a9?9yT>OXb{L1A9N(if!l{PTpX6!%|_|Sbzt-OI^qkXBtBpX<4%;M}~8`Ny;
zVJA*EP@K)QF}D=4(K}_fL0twv*+}o{wpd`?Kj`B{{QWl--7TamE{##&4d+od7(siI
z4Oc<>qQFm(Ko6c*K6usuCY6n<4}Yl4O7Vl27DD{sz2<WKz#M@Avrz>Y!3tN|C^^v|
z_-;hHQ3C|Gg!~%EMhy_$!_GFcp4L~a9h1X7%{n63;p>SAPQ-31rs}=ldXc>^lCDwp
zB?p|MF(b`KMm0>NR*@VqilO>&qF9lPEAhkHp+YIDH5^`7DBdRhU~ZzXc;&7VlMG>l
zpH?XC<xqqxsvtOG;jl!U22fXRrH#Kd-w_q&6R3gR$Y_(VClRl~WY$w^9C=wkavkxX
zYn!jAljG;dTw$}biX8C_aOo_{4pW5%g?m@Q+IW`A=e;dCi-p1?%6?nEq7;5NCl#E#
z8S08?f~+f9%mOzQX00t&yKcbj*@+Q&(FztvMYvNAes7r-f)^!vF=t+?IZL&!%_>I8
zSt<!Ox&9*(wkb9W4ary=H>qUn_=M0V(c-En@5OFXSGg1yV>dCcwK$F2B>e}IaXg1f
zMdmx1R^&D@cN1(63U5y1*hk5i*R2yNaGwias+(A2xJdz<mN3>b43a1$rWpE5_LUq@
zdd<h=*!JMKdXVDEO$Em6LCUxT%`#Xt7uN9dph~PuqwpsfWTtri47;sC4Gs7l;czM)
zWO?Uq!MTZu;C+H|FTQI>tR4BT_IraJ<&L*};n)LJQYq-tnu{$~!jJE|+Io=}4}O$I
zI0Ho)R7DogW#9tZW>r5nz3%Q^+N@Hzksm)}pi<zs$GL)(V<dkDGug}{;qB0dRfK~(
zy`&<7;~#b=tr#(lUnj{nv#P2vLKK<!u}5dKsslc%uGuRJPl@?4)@Ig?zn0Sz^p&SC
zo+JK??DLTmerW=p-DguG7=m#nQ#edxatUXXjW3{O!2r>cNHOc?AYt4X$4fwwWaCzA
zDR}OFnB6ns>B}&wv}00~-8yHh{@~-|<D;w8bMM6jEa=yxdS|PQ(2+6z`)8{W>W|${
zo{vmwuG^!Jox_5JbKpT$ig*xOM~K+#rYwx(0rW0mOpIuXbJ&1Q+6z;CG-}>iS}`>?
zSNKtJwJeUpY5Q!Sqf+>)0v)>(qxB1on3K<8(}fq~H^|8zwrn@iougvGw24Ou`})G0
z186*(k8!xhjMd+^#BNpf%}V^3Zan-J3tOl#(ZPaSpz;>Bmj9-DvPBi8(=Fd6R1_Ib
zKGITg(Q$oC;r?r@*%nn^3oBGxH2%OBOu{k3H#gmg@u(<%BAprfMAeJIE3upCE7;$I
zoZk{_2YVsZj-AT{Vb{y8bgl-$5Qc0}@W#Q-Rz6q7#l1!Nr$+L*d#!S=svSSWIqY{v
zQYD9UXr9Pwn?EWO)5F~h^Kwi-xXL=`vcmoMLfqnYE?Z?eZk(P=0oS8rRIw@Qi51>U
z!hwTWxgQb=5?M`e!!Nw6?fj>tG3?J{k^HxY!+FwR;{&Of%R~w95cf(=&QsAa8Dd^m
zvVcqP>3I}pHTIB^qJlSE@RRYXD;!0)&r`!J$1zAp!Dv|P#{AqpgqLJ;#cgz#NU^H&
zCGx&<!1b?rQ8eL>VdqPn(W<*bolhBd;_&)><+TOxmzG6>#|YqDmYh#{hEU0s^H~Ep
z5b^7~si>ewVLoUnD&UD&*x9ql>{8#=D?J{Iuucpf#Jg2pVHh#+>_rr5y>Ezn5@`oa
zMCYp$L1Ss6n&*p8zhjH7@C6ch8#+?y0#VlV!ezLCa`HUwHY-QMLA)QBT)@IKFyB=~
z6sqI?d`23-Fc6VdD<-MIsX2>y3mmQ6N*?g@OC3RY?ps%|>Gy*o!Qn?)FT$O5>tCQM
z$EgQ&$O~BIaH_QnDU45Wr3+aY<c^0jFC+mTi7sErbpA7=n6fTp8F5%HU09z1y~?lR
zr5CDm4)Dyh6~A07LTf0WMN1c1TloaP<<(I>r(9uO<%6{d?UhH8^;qqD5m$iwkqC<p
z<|^f0#pLV$>UNRp-xY4KC|$&YjpKp;a1qA_#E7Gz97+A{FwFkRMQTNVzP+Mkleq7e
zUZmm-<EO~6*}5opS-%gT#T0(b+g6HzM^8G+2*2*^R!AS~F(YW<{yXWpc)+a!OH)tj
zKK&LLT4GMvVxZ&BE24?5fnBUj^<F{zfQ@&^Li~U-g6(7cz>~r_{k@n?chLn_xmY#g
z`W%d;>BTH&bIJ|G!Z0Y}ezOw5rJ^l9Fx`TrBa-17*72@LE{1l0u_|SCV+uyGo)Tcb
z#2nLCg0K>!Y+hOWY~+3`b_r+DXfeQJJ(p;SkcHQOupnH*<_a%*k(Dn|p<(&KZuk=B
zhl$ti+LcSxQtSOpK9UXk(OpYq7+zilnZA63<@S29o<&P{Vg&jKHuBvl#UM7PIz7xI
zPndZhuwMMYDJj+!ePBBBp7+0s({(5HlvR#jNh_r+23{|)kh1o@O}%@T^$7zChfbm-
z%}cG4vV=TJy2nz5AH2X?Da$(RHuWsk3bcjxD;UE>wW?+vp%bsPu28tla;3!YN@Hyl
z#*q-_GUWstv;X4se9CI(!MIsfneoy?RZX#Y3Eq5D55mUP4L#+mS_Z~*m?rh1i^G+e
z25x+5B*Lp%uoG7$IF3G{MAt2=vZmHx*eJ(v9_q6rT=hqyyGo9SlF+PqBwFdKys~J(
zR^1o$TgvmIsE;=;H5-z6Q?8(vhL{s3!7!vwvl`v39PxTg*oiO{9ccGTD(hq#`OLU4
zN^3|}fL}MjV-l7PspslvIK5A{MeRNiD9I4Z3!fA{ncN*xUE;lfT-o`HF4l`Vz$3SA
zeMR8D1jfGT@$7=xFbl8*cejRF0E~2a+YPIXJPI$zkJWAmOolZwq2ggMR)!_E|8zDg
zG8GH_mMzALFwTwevkH-nH`maw#L!Rb>(+_feaB<nVHRWu^v7I+oxQ<&LS6=C?uV72
zn2%hcrMUWMHC4ibg;o;83nO^bj`CuHa8r1>rbZgaFE=GMb&}0isj0MRBkbyxb|>cC
zmf$4LaIzW;x+1!%w#5}cGwQeJIyGrU=;V_%Kf>xN;i?UrEmy+8=SgrfBmr<tgM(|~
zs{fYRym0Ss9KBkdBFEQQOLY}%5VpL!<fMyRxlZxdZM38=b0oZ6u&VM-TO}r{;ABXN
zuqVU_Z`E0ToElzjnJ~0gyk}QeJ7Ra_-m)DfAI7zxZj8ayh`ULaOE#ZkA!#oPgg5xG
z^(yj2|Eg;|$~g>g-Hu3bY}TNNjZg+W$btEIMEUaPb)bzY3U48IEE!?BaVlaJ!5O^w
zlnQP|-)o6@xQ>C93Hy(9TRUR(Nb5v~?i|pK;VZF}<zh32nV=`x8aU-blZ>cORBS$`
z^dA+nQT3WZR6uF8sBHYekSup$l%<=c+evwp9JXS;7uvMeDlz>wY|_(FR*gThMpu~D
zsG4`~F3TcW4XRUnRKrY9Hg{qSp8K<I#G5fG<f8KT&9Q0ijjA$dU`g>GBvSq>H`=`N
z<Wux+dbKg;iOI!Ef>XGe7sk{`BYqpB96vVru00V^n7hd;f_!a!OxfunFKdly5a14+
zR=F&u!bxRY+~R17tb1>UEvY-Ea$=H&t9?GEiiY>M^&-DjH2t9F`!OKIKmR>Nz6hx!
zU05p`N_{q7g)K$wzqk<^64b9hj>1R1!7MS9{a#ekP^tM-ofUoHnH#iv8e?!$iw&6g
zEp_L@Eb^W~owplo?u84j(_p1e{vWE|EIRJ1+W*|E*Xp(Oa=Z1Y$KTD(?NqoV(W4gq
zhbzyOV8_*#6ZCK3Qj#sZ<UuMV%Pu<@lrW?iBc>pRTq+Qt0tAYLBt<3)M4*_<jAAar
zBzV%J9{aQRcgy^{p3bk%Ir|Lz4EyYH5m8YV2yfp?Z#>5}HBigN6i=Hu8>+&7zv-1+
zndLEjMq6h$q`>GItq+t=i!(7k&JNyPN*8;0?N6SGA+T}jbw%YbT&P?1jOG^KpD;M<
zD&+z+IAN-O(xa=cDSaHyqP0}W<%WohRG+==$7?U4L1=Mp49+C-XG6VyC>ZFC!*LZ1
zs88n2&PJbMJUwosXEoKxV0lTN)wr%;z@;+y*yXm&G|suLq9QZScTw_;b1DSI-B>Y>
zERM>Ft?h#0(rum9&_hAPI+=5c6Q`$~ebPA|WI415;0O%Wii=0Km2?dmJEx&CDXz>s
zdM+@&|DYA;G#*GT25~ONa{+$+q)HIFw~{4p4-=aFSX#v4eEnRYnbjieu2t1}*;QN1
z<|6tW7&)hr;Hk(J3D0SXqdjQ7-89G@WGNh7u)l}9J@$(dK3Mj6ui1G``VmA#=hgWu
zuJ~4*k4Eq?;XcF0&#RFbi_*wuc@wLYM~0`7s&bsCZ}o(`2kwLlp(o>!dDhOyWYy7C
z)Bi_8x@8?zh9AKzOqDkt<^)sn#rvQl&+9k|o8?eSKQA;uZM%iCEyM&8=2L-WM;A0t
zE?q{Bw-7_{p4zz51v-&7nWa3G8TU%&OtCJi3o#$)O*3Ql1s(D*TX!FGt69^1h_coe
zVj^oS<d!NH%QI*Kb@Jns3;cTv(PP&xak0ce5T@iyZI~r-!2o%A9R_azTNIK`A!$`|
zrp`^dsQGa6EbOz!7d1o|=S(gNnXe8ssgiS@s*5rGEA0M7P0uo2hVinfxjZ0k+qKMe
z{R!^O{JP{%Uh8Dh)9%`mi!*z)0FPgd&hTX7vdKy?%3!2ytKG{SXZ7ULo%e&sd7`8B
z7~Txt5{i0sn+u1>+jw0)L`fMr{A68g@CZEJN~RG{TUE8^&%%ev3R4_>XjiBZml4c;
z6*|j3)>5I+$>k}QkoK~Edn$C!cTYl^ImtS=VHPS=Xj_F`IosYSv`ldXD2Y2sE8eJH
zAM<r$<5^}vV3EwIN34U&8#Nq!U|BWOhv;5COW|dQ4YrzJT7+u?wg}6q^~RVet}n;P
z8v}XsY3R~>qt+@fK1gs-#?_gBIh~jmFGB&t_l9k&eMi07QUFFL4WejtNlQ#}33g_&
zq`}Fz?D&!p$6J1KDTWie>=mVp`imd6>XOzKhgztxXVzI(|7yZQ<Ap10Z7CM!X=rCb
zJ;P*o)ETV7A<KHII}hj22Fmb|JF_-a23HM3$p&ivEL&FFsPTD7qtWH)EfP6gL`QE?
z4CF9S8jb*-&a;M#4aur3YxsI|IRjQ@@tw@AvkQY8XmvRzjYA1+tm&gb^9yB*{5a1c
zWgS)FfxxS5Su@EC!||iOVJ>XL<(P=d0oz{IPGOL#<Z2h99j+>l*wG6yo;`^BiVFg^
zyg2J4x)7t|a$?sgU6f($RaCpc5{osb+UyRiXPIR*3s?2G3t-k%+~)%B<%Q_gzJ1oY
zkQl}138U?)f^#(o3Fi<jxuMGM0VCNe+RZ#<R^(n3#$D*8g)-7g3<^;);<jRCx9^N2
z)9#REQfaiW-fb0CD1tt(Dx<g)4A)m;Mz9sL)%^RMGcH%+VsO9S{<GFf%xf}*!YAg6
z=Fb^cbJi1>;w1E$z@WAVEudyYx7xM^!=sAAlL)Q7i_vpm12%dwW~=mCE0j8T(8e!D
zFWhrEE#uhwVsv&Mr$!Yem!P7m7o#UU!Q4k+{bEcFbdVcIWdlxsuc?NsLASCDDygh6
zl|e$uaduIw7ot>Bn?Ys<=TB)@N3?ALn{mG1rD%{Ffz3Xu5A;;HAc{-T-P`XxWaIiD
zPjhV1$t)CvV?`f4*3n3=UWx%1?kmp~yy}ABwD{_!m@9P5AqXyM?wzP)ovD)aoBcJA
zKnT|1Vhro;Hn;I8db(M4fd?NK&Lu5Ou4uPp-BxoeJgcptX$0yEHrj}RT)p}_D;m*_
zh}}A_G#*bGQ$t-k7}hQoRX~z#)rN)?B41ux^~UaAz$q-WXTorO?6rjLPI_BPFFLi^
z$-bPNL#L~k4S9esn3|b&#2?(l{NT+B?U`M;hy>%ZmJPbWqnYIREiT6hrOZ5jIfiv-
z=!U9H@rs+Ra#^FA;YLP<Br}e}Y03-?b)0}!ZCKPC%1Xx()lw7ADdxse#)}_!?}o}K
zkbSUW5|(q(I0aNJqB(3|)?DF8%WVYhN{qw9y^UUp-SWy;S#c$Xp{L*Yl^8~`z*ReW
zB`V_yEdth7RIa$w_4=z<gp}~V2IA{iw1r?uAYNHjJ16{#QL=bl$y#diB0ENZ&iDnB
zP<f@Gkgxw2ndf0P;0#rFd9$qsimC8xZTCcT&Us^4qxZzHHa^Euri^9RPQ|qIo){V?
z{jRZl&+a;Mt$vvQ_h@de-A|v1ovgA7`&QQn5-YC#)q6B22ROmkltvoJ3cDw!@TOa>
zqYmJ3)q3i{3<Nnu?;tx69-xQ%xIKcIZ41yVURr&p*_*UTPKM8p(KkhN90zFUEA7Ht
z8-G(Y_a*|%y-8c^(mYSl%9~>F9+#xLTNB!^t}-w(ndDV{h^AXpg=2thDG7(zI;xAp
z=&bdWj)=U2H^nS1&72O&?=ZU>EJWMht;GA)u12R&cI7l5y&9b=&`vItJdKXaxRO_I
zywxVJYV!RQVa-;qY6dvjk(pHm3PC(uqn;(v@C&t`t1%0gx$>35JwYpL`!qqnTot-z
zp;>!^<mPZ}_56@!dY0O=a>}-|J}+WWY;TUf&to|?`euQGiEl-wkMJZOf3t`L$d`p>
zzf*1U%_9HYHwy%gU$K>}fEtfg-yFThz`y=x;f@;%{oVW^I2ubULEx*!{h9j0ZoN5f
z6L}+dF-<379_IzzQ<_QQ1V}((2QgFznv`rCr9%Q@Z_x(8_9_$#M)R8n2^Gv@dy7Vk
zVxNt_MF=4j2k4o6i*B`u7vqyrc}tAv2m%15+)LnFV%}DeJ**~Vzg}FOW=&Za*y1fU
z;!%dJqraC?aPGY&rUj*WF4}=$UV<)DiQ?MZZCfP@q-$LxT(?I=DVzs0vKm!sh#fzg
z8GCtUj;(5q&72DZyF%rPimN`9R1O(2W?{-;xdC7z0o-=mZ8o_Y6NbX&eL+2xGY<i@
zRa8m3bJLk|jJc|+QsJQj{J~sXjcHkgn5eI6Qt_Hal(~kwH;;9e<9bz@gQvMLJJwXK
zJ7EZ}31Co8uF$x?8gqYmkF`|AJ$S<_Y3FcZr@4%+qhjYTSXaG6E;VPp>;jJ_Cc3XG
zyRlpbIYan*owLmZ2TJ^o845^jq^&G>Eci@W&W(Jjwxe?RD8L!zz0uPfZ?Tbk179ev
ze8ooZ)eZ=&^|5<nO7_Wiu*fWQFV0&@StxNjRW>dN*WL%qd_rls0xV6X31z%&Wz`5d
zFK3Fr@VLB$I{kF^;F)XpT2=kJ`dX{$Uso=+dNxkHnptZo165`)v)e0QvrT1ySsu9e
z>XbV)Z|nNIxC4#pR+hihPr`N%P~o^cDn`JaW$P+Ezs%n0WtDw?3-<5Li^^k(uvQ0|
zi7@G$ZL04z7_ovJD)-OX*1fsdPX?QCJL{Zs|G_(|(;#5heSwFKCjmQhUtnhnCY1gS
z1+B6BwAI~LO{~j8nRxV@v(kMr;2ENGjNg~<UAZ^&lX^lWXLu3Yq)K|=hq97aZ+y8`
z?$cIfMV+*1{lD=9dqy=8CZD(3ecB|LsZgk%InL~CC<h8)iQ{$kJ`EU^7w3F5?~6%O
zSZYYVH8oYhOzYZuLU9tcB`E&%<muYZ9DL_lM;%(iFI+ck<Gh1CBhbZidQ<FYom2F7
zkm+S^z)jVk=O!L1Sw3*gwp4;c<X+oOuytR#J6YA0XzZ;TG~^snHb(S?-m~xb(YI>9
zxb~WjDGTz$&y3=&fsb=ATS{*Y%z7Lq8&?f>C%vD@>Zj%)1Sa3ANsBO|9V=(vqO%Dq
zZ;kbE*8!VWsl^j=$5hQyPD@x#O%`{;`N?h)^^9&dl;Mt9R^J+HC=~BQBf6Oo+_ulw
zvYN;EdCt~VZDp3LLDdR>xkP-W_HM(@tLW$h|5^04`&Mlcd@REr<DS>(7teAx^)pW_
zUAB+fOJA~^YHtR0s9}P?G{X(OrT^!VOl;>=x>MF20YfB?3KF-Fy)7D!Z}-+P@-`s{
zMIx93Zxasu={EMZz$iz+0%I=T7ITa8dg*OCGfra0|F%HJuG{vB1U2{oD|S-F=9f=d
zSxvGz6a=QCtQn4gX(jQwW>wW@0kNuU*^y<Z%$4RyU$BPKdoSOPtS-x(#<xII-9zPU
zO?^UffurZ7t!JT`jjvcsjZJZHv^6+1H_th@RZU--a~p!LYQT;}Hz{kMN9nttfL))(
z2^K+sm+6P(W)2bZ&!u@b%+Yg7Ze>$)1G=q_7O%G*6>;k(d%LFH1!*I1*8+F7Wq%pG
zJ?0(#z%iu`?y=(SI^jH+5a<3$E2%!@CG5|H<tBR*35yZCHrS-HVEisx`R&@T$7{Iw
zwaVLb5jeYfIzK|JSbe)TD5PjDp~)83>Vg(F0gY_b-#M#-oR@erf!#gtgn+W#HP!4X
zTi1YYBqXe=BeZ_EwN*w+q1H*jTnW|{DD(}P{+^1#h`LZaZ+&Gc@bN%hX7||U+hei2
zc`ha>pe1`#K7Qn|ZRsP#5VoyKgf6fhRU(QOURh9$w0Fc5pjTic?+|i0d;K*$`VLJS
z9@~~}>>Yur+fYM!htTOnNGWUZLhE2XG@dn3f$i8tet>86Q{EwjxXJ&VmGyO}yxdk2
zNO=ERw&~qo@W`i$YBskFC0J7${uSgv{(l(l<c7uoF>9+zBAK&hwuv7Wda$Mp?BZ?x
z9Xg%ddzRCbErEi-oUyjj+yESD@6bwAP-nDOyPAak_+l4q9F;rq)z7{h;h-9*%w5N9
zGyh%mXkD3J#!u^3f(^>x85n@u%Jg@M?PS``Q|<mh&BkL~Irj?}cLWDI|A*T%rewk7
z6!!;S2pSDu?$_ydAs#>D`a<K~&)<oxa`co<-k$?>*EaXU7$^%y;a-X=_Xoz0R%}}T
zhW32*{#YZpIf?Lswfi-d;jqF_GZn=L?YuRzSvVKyJ*?)JOHUT8sZMzC-!(yWyFFj}
z>l!IXjqB>`!_5?ztgV8D2htrpYn|-tVNSB{{i1eQ#UL*3-k!3FliIPqTDwdK2U+XJ
z(oMFR0GEj(!KpdSPvYCS?Z{S6`Ej>w*v=1htS{J(h7vYlXl*oeX}jphMw&vDtC-km
zGY~j~m((%+?@vz@I`uy;1Fh5ytnG6><8hVo(Vnykl_~z?l{Tp)(fh2d<T8StifTrD
zuj%HC|A)V~LL?$vZ^l|x3HJzXwM_d~s0#>&jMmWK7;|GUkpr2HFoDgc)&^D!%m*D>
zQ&HF^ASsLDAeXO}azfLqtsLG2&d=@SY#g3H#nsc)@q%-4SJv+3CoA1!g7*c4I|`DZ
z8B6cx*W0H4E$R?Q!)B~Cx9N&-Zd=*xWgh3-3C!0!J=)23v7^Kk;GIGc+;A8#Bl=2j
zTBDh_5H?n0?~E2u#_|x9xaZY(xN2})m(n|R^d0k?=$(PvC0xi(yi)^{4^*2J%z2z!
za-ftGMlMaOWQ(|1gI=7@uk$l*$X$J>77sgy3rP*&s>|vL15Lqsq*c~XEeP!}YR%?(
z@_4Mmrs}!z$y)XoEofU$*mt;ZxK@^dRo;GQJ`@(~yZ+ca%AG@c+0B;Sskl4b?!7ZG
z1KHEhR*%k}agtz=P2q9HHdVja4-KcDqND6Dr>*S6QCIZc&X4fCvZE>yXZl?_yYx^R
zd6%aAtm_oKE2jJ^mfd5!f3vg;r6GowtfV40piDjfE*(v$aC_qq?uo3jhxDY1ok0Jm
z{4T9?Z5<F>QC=}$v*`rE%);U29i*;#kW>U;wfw@<0%F1XyJBWA8E@$SbfdW{a5EeS
z&4hdq=GeQ0X;){qb$y`g<jnqrN;F%(+S%!ourxYZCxS9|vlvEsFmQU=Q9_r5hn@br
zgw){IIll^uE+HKuz4@+Ku%7GD0f+kB^{2F@ZZ6W}ZRHgDxhhexb<Uu`Z0`=lAYoas
zk#`4Tial3sRLL7&af6MiL@~qc6z|q}F(2c0Qii)!wefcgOYS{{W5Fig9oXk?e9{fo
zRqayEIT_`5#|$6aV-@`kKM-6_C+K4<yh5t^vAhU%S<4Q<QL?)Fbzuf-A<JBvo5Ln4
zKQKFhtCJNLS<q{lalr?;uG&jPDs5#MypCI2X~M3caHjGoo<g*A->nk?Q$@c@dRgbx
z;=J`UD>%O#b;^RR^nz_FO_ZwhHWau@_z>I5`lyB2wlZ*Oz-;ZPUuV9>%jZ3sVRyP=
zBkzf++IPT4-=m)348>#RjHwj24UebdJvtbPdxrDflCm%f<9R-=57wab+*HnAyMrfe
z@;yTK%Wk%^k~luMN;ZGQJ(5mm1&=jHbUtOdV2e}jJ=)J#{8&>OxgO8c?B}sXo*#%R
zv)EBE(NrrQwPsDpGd!r)vqAJpxPfvcM23b0YiAE%@e1n*4xU|3TURw;jr;AB4dB$%
z`k8h{=iohBk3_EJYU3<RScAuGn913B+fog<4``+3-*0$>?PQ-^tH<oUn%$X;?we%f
zy@3bkoH<oH`repC+(g>gdjoNYLl3;r|7=Suy*FkGPmAmlrQvFE+az3`v}}q=)#gZA
zzHDV>!H@O>ucE9&5kGob?H8xysH^&34O4MOK#SVNT+ZrhtGNC4+pQrGi@Xrw8fQ~3
z2(_jvxLxkrdt<R(K5gr&%3^-fS_x)9YMO1O<L%hV|BtgTy4eflei1*lmqTN}b4>K#
ztCRUUJu^_7br$$$LYd)BIm~kYCU#o_+fq3!0c|_Wxv|ntrtjhfV(*K|#cj(u8+o4)
z#Zid5!|3~BG_PT=s0=nb$gSR|-GyHw9;ELJT<+ue^l&fZ@6(3Zchn}bSa>a))c;7r
zt(-+(4OWz#!&-PcOS8$(T2;MpSr>bzmRYW*VRf}voZykiVsEI275wG@FLfTbrs~WG
zgK;hE%(%mAu9$>}l#%yNe$;E|QMGgQJV}Aw(U=K((Xnn8+vC!>-ur|ME(wgLVS@Fu
z-T{Si(pTs&T(C_+DfXN=VZ%%>L*s4bG@P8Z?F7EhB|kg*e-W_uK+G-81X(fifaaY|
zj!g6cA<g~tjj3GWVNK|Fk)P19ZgWz4AW$-o{CQlJ3O|P@5*on?c#=MlXy@8JZ{^H7
zId=l3ko-99%4*XI+T*j*QdRISFGD-kG7AliRo-!RwTe*E8Y;y+zF@2RyC~gY%}ly1
z{DIhk74c^;+q$w=mO0Io1^edw+#;{JcJ>rAH)n$iG{c5zptM}sWv6l7(Y5{qv5KK#
z27>z^|8cWzCLA~ZAW}tkey+hK1tUk9@X=sv+V%rNx2xGibeWyqKIQYu)^-bcm_r^F
zBWr;iG?6e~TNBno>v2pOm|mk}v=(@A{o>MEkSKfb_Ou&ucw8k5cVRG*Ae<j!lWU18
z)b7t&SwI$-@W-JVE~K~VwLG~!Xs{z{O*h{XXBUS;O|>|htuAm$FE_H);EZNN?!tjH
zPG4Ms?#y>B%k70{zP=XId-H!;ORemNRcj{zHUhV8M`J9m$7|rFngz5id)M3JShjwC
z`8yrTK<)W`Y@15spkmI3D&;{CxHq@dle;Rmt<<frupN~+F3*(q{^%>t07l-gU4ggX
z(f4bzI3t--*gNkJ>=o`txKNsm3s!o6%sfJU53e!){+Lo`Dr{sY1o@_$Y%;;Q8&B6r
zDZf7!fGf#x;w9|;uxF-~QE&!>A=N*DP(jTwCf7!(>q~M^YpBZH8&NW~RSjX|i6v_a
zZ14%z5^TktL#<~8+Fr1h+SoH|ZT$_=iEyuz9d?~7f4+1Dw?vrFo?5@CkkoyBA&YPs
z)Ie?W3WJDNl_iwwhH7M)t0%{L7DI*YOnY3b9f2ueh;tL!2Xtd_vyXg0h&Ycvf;(&(
z{Xkwc9#MHr<!(kGU#OhNIrE>DKA^MEO+3bxj?iBdN~6|_%1*-KV!UAG4+JJ5th+qS
zE7^nnOE#T7V4s9%<?08npY(oEuya)QIKq;8mPMq=iM(u$Yyo8|#0m+Mo2{A2+g@gC
z>X4IlwywW%HfSw%r$C7j9isLJv{x~Z@?)Zt)vHzOW{Yy@fh~u?p=WZ=`k6k)Yk83H
z*!kQdn+cD{@&Y{@s^GcHs10oi_RV-Q-xloP!{Lm^b`r+StegDW2Zf0U7crs_1*RCy
zoSIAU$37?oF0bezTzoKadjtMJOCQWL(xs<PNRKP8xcV<Pk;y6iqb0agPLr4OW6+{Q
zl@DrKoY~L)H?2BXxc;gi)QRZ6%g$QugR#YN*irxBi~r(`pp%^G*ZuT1luh`-)epw1
z37h(+pv_;oVr%-lII}ow>q>`}L`$sz3gv8TtHQ>^xHA#N<43KlKW9+W=;cg6N1;&<
zg^*AWK~)Vj61wHrsBPxVID0+9Lk?Az+YT;4wkkQ|wu+v=9=T&jGlcE}E6P5kdBDmX
z?az@9X}&HYT_AJxLxCE1ix?i>#y%7n^|KvqiVww9Y_wT#B^7kd+4>Bp<M@XHhbtSO
zw22SJ)ZTj4CO;%>Mr0#@#+H?J*Ac4-7UD#Na|z46G&#<z>JvInA+D`uhYw+-1ZAg+
zkc$skBXjOR;=*Q97mED{tf@l96sv7b(4Ygt-PO8EO~JiuWvOKzkL|2H1GSLJOP8&i
zv%?6Ts_117?kD@27+CXI=ne#kz;3*LJ|sRu2t}mQIm~L+8*M9_JcV-jwty>m>$2@+
z)-ldn`*2{zMZc#YrAPD$TQM}2Mn4>Ljf#zpDaq5ziVy2P+RNH0eOQ-m$;Lk{y&m^3
zBX7PFDu*xJCv7t8)H!d<0>x=-6;;1_gH3-}izN<otA055w|iWz>4T;nD*}X8$%EF&
zQg^so=Bl6`fSujU|LaRvz)H64jw9BUF%P{7vmi4n<%+d6Zpy^CJAzeIZ?&#+aA=JL
zn<Fki5rCn3YyE_=?BO}HRX63f%@1oqy4FSzZ9|QZv&M}twz6y))~0Q%wT*|b*p8Bm
z9>?>+m}Pj?TtrG2$9>~G7*hi;gO4$NxJnXyoCjkH^CePxQ1=){LE}og$IOWbg{2_F
zuuz(_Nr5Qv67js11){*mD^prgvygzW4U~47tW{MC2Cma`wFmQ%g~jQD^}0${e8L;q
zaSvFynhni43E6y5rvc}C$Bj3$9E;r6)my>?T1!cs5n4OPfO>RC$<w@ix~k_!n$}aV
z+>H~%>MIAEQK&VwEF0U+Z26GfNDZ@BQ|LL`mLQ{zY}*=@Kl#FO+s=cr(5Dc}eMI98
zXS+=4$VYToeV)T=^dp*dG+f7&bg#?BN3^y%>Np5XAJNdK4umi3@s9-HJqK+fE5GVh
zHkoDo(Bnoae?;4%?oaAUw$A}}&Xski^?fYtX2Duj{XsGKwAGXjo^|~rl0)pegcr6N
z37TjraCUZwMuacd?rN$>8=rx~&M_cZVX3X>mml<XYh{ad*zj%jeAiD}M_KR!A+K{#
zd*_z8tF0#(u|IeesUJ&o&P^Ps*0uMuBsa5`t9rrO9_A+$kuSEb{KRP__nK|z=b^bU
z65q+ep%(Ch*}8CZ!wojFE`$(w2DOaQ^+3o$jvhZ3#`Gz}Xf%~oZ4nEk(z*`5msZzp
ze0_H#fmJRoghE;G5;Mu?WPVl@v8lYS10gmZvul;>f1otwq}+7&58r&%tUKcy3|^_)
zb$+GQ*JHV$`rgPwkduBCtg3n)<%nkX^W+L#xopN`A`qtswFG6JaUa30buDEn7Qk;<
zk98KRiq=Ueb!dWarccmNFWYrrvUIzj6A`A{;q_;r!oeQd%;v9a;0+U!i{Q4@Tmc$z
zgrST%_r<W2FvBMXXomIh(LkOnA&q=AFPd2dk1HN`Vf3Tg9(!GuIi`{YG2)Al2EKR{
zLPjWExc`>%EYCxb6EVx9)-m}}O^16A!>yB*KdPO2crLu$R<fE)Lv1=gj$?*rRl%4i
z2um%&9fXCdCg(RE1z9RSANqi+yX`ye-&6%wk*#HWa-3!B3H8JkYh@0e8C*u${-~@{
z<`(|xSE}<-t!V@Y47aPcue}w2mp$ch6+OW==;!po@bjl43x1&d^trZ~lK@>D%&MUv
zh2Q@zb^J<f((P=J>kh2!<e2L78(%Z~Se{qzs%+$A!k4S}%0zuMvw5k&Ec%!x9mlxH
zHeo}2Oyhug1a?YUls$&$<&Wv8JYo|c(_TJyz$QN?Bvp9xa3GZR>1LzuvZeAdZP|G)
z>*-y{xJ2U;6EFe@)UwR0Ib7?R-ouN!p|%xkZZ#|8h=S|AsjubBs5N5guCLHJd)(Hu
zZ&RQxp&jyrF$?jCh92r9fH{<aI2koXOD^bC+{a#yd+;z%jlLkh@`gzpd`!nksM<RT
zyD31hpL{V7vq78@*j7#q{1OjAysa#55MI1W<A)Z{Odl7n@Q4Q|{Nn<TB_8rF6k|s}
zuK8QOykKJ=k4bkex#HtO60agogi?aQMFGl<yEWiGPh=WxHk(wTigSRh{Ba@8RV+C}
zD#~+3nMIpcGbcE_stI`6UthJXlkU>%g1zwwR_To_7cN6qvs`5*WIxU9sxuGSVp($V
zDdu&37rZl$TiMrRAhewsl^QIq&c|bc`U&RLigh#d{2VR!^5ZogAgrGs2(uvv^MfJ6
zbWPx=QAHZ6e{N-Q0hh^JDz}K{CrHY2uVy9h<T!X>!u~b;MBoYIS)OkrpV0ha8spwO
zMn55RAWVawJobse*~asn6h-D>;R7SJ^a-JNAO7AWD%iN-?D|cf_(WjXHP$8tq<9TW
zx2*I&r+6!$h&}lF*W0uT7gsJ>HM`?x&6KaH%wc?`W%GB5Ij$P+18h~J*&jA^O?|lI
z4qPkgV<}8x*HyzE|FyD1+=<rC^n!DX)P(!m<)PM9Po&BUx&8^AJx<kgkN9<!z5XAs
zv_W?61e85ei!7QhyO1Fd)e5)4oNcLB^H+S&ZD;c++SrcLuvQ`2AJW7>=R4^kQCl3x
zp^qkNYxJQ&A~Ze3+1Nud+b+u~lsxC`aTS^0Xw&StAQlG@s!ga&fi32wk_UVzJtWlM
z71F@ULs}Q%s^vMGRyj1#teWL`oi3iTT7JzhV0~`&hjau>B6`^x4+V~iVzZj<@40io
zH3e+uBA$F6(zJ#OY2452s&eB!)=JQjuAFS`EGdgrr`S3gUvb9@@7!*}e+@K7@1a<n
ztg<v7?W;i|0ox#fR>C>_rW#wuU3mh73o>I{D(~!&?T23cKb_U$&gbGur|I8Oam0u0
zHT&<F{(|du<i7)Pb&m%#`rn$u5Gg=TD2tFjw^{Muc~m;Pv6LSm9x>B@YX;d^OU}TU
z5bWcA?g;=k4X8e<4w`9{spx~1$;Y$+uzwjm_jgqQ%6{%;LsRZEpq|j&9daYneuA!M
zuMf$EcvIC^2ui<}!`#oyDl525rS;#jZv6P*1aD_`X2zc=ovaRbyPK_}ga#!c030$r
zu$0Dh1jVgv;|?Ca*+v9|QX8p6w3RtWcr$IQ0wlccWCQ;%k=D&V892EccIJqZHQt`1
zpN!>(OmEj;&DbY(EzNTT7s_?BO>sD^^hsd@juv6Mq4z#1l;ZosEfptJd6vhEO(viz
zZmY6dT;hpY`D9>P4(+D(zdIkQ>i^>0A}gS#x{&*j<w_&@hi*{$nOVfLDt{&vUz?wd
zZH0(>?UUNHi?hyZU)LA3p&u@-Y$BvqZB@DYO6z1}9xkD)fA{XQp8j=Vl=anYfxpZ_
z!U#7Lo1YY)C=H67SQ}<O5sXoJ%-KSY5K)M?cc%f|2>gKtvtzB860RSYn74Mx5RM(c
zb4FX5V%I<CcpqyCKgZ!d6fNzu(!W|sFyOv1?D4FU{+uUTx>#Oeld9~rvX%9316Gmh
z7K2Z62Tuzo5?`yTG+u$<V>PA27^$9JScaF<XvG3@0Y*GTaOWksZ@AT(0)-sP)>P%V
zNBLN92}S!N%v>w`>WU#e$l3|p8514#i{}($>Sn*39ib_b)YT8ZzAA;r!5{~7%>&JE
zCU97MxG}Y1ORLu{i2bFpC77;^w5@7$?iAK`^#2@FY!7SpA*gvW*+_nJ3%KU*=)-}G
z%f7nC9u8#S$l3}eS3qa!VKu3!4{Vr+g<j`W2X}TNA)@!`2K$qOc<n)M+_KUbJaD85
z<P2dqryq{x;;vsg%~YnqvzXN$4m?jm9O&N&w$f0V2N9b_X%1O*Q3VDGF(lerf;oiR
z{kobuh5zbS_TCi+e2ndf19i9!^}cqp>jXP>#fz>0)|WQgDA{KZJREahV@74`v@VN*
zGAr)zcJpCfc>Yr5km0}`X2%v5xt_LEG~P4Y%5a_lG+Z{0-w%Jc*{5Rcr(m>xYWFJP
z_89$?c8g<WEHjXm^A`G)@Oo(_oG+I?r8Rk2`{z@EO61Enp-S-YIYTG)!L`TgEUOCZ
z)_w9-)c(FbHvK7Kco89+yFamN!oC#p>RR^Vj&O~mEDxLwA<@Xr?%||cRSrxtkE_ts
z7k9CX`PZ`C|HFb@*Er@fWME4@Utm+Tv*#DQvz_dN2TML@U6sP~y!CdwvjSPx&q)b~
zM~MA$B<}n?!dGsp-o_JbB>`mLchTF}Qbuw2?Y1p|8=t4IwxgcLrEH%TDlVbZI`Zk5
z;G2*ZjppCyc@~a+IyTEcjl4dswc$+IIkr2bfl4FGz(7(!iorW<LP;lDCqFIpvp*lV
zvOc)R;h0uFEtEqyy7JVt>TEoQ+qHx=eYe$A!$p8RS=ArR=N8bpgL=#F72IREQ{>f8
z3)K*7C?c_Rn+fO$7m<2e_85KZ0_tL6Bo;YfZZ2zQmv~0{8P`!c2uUmyx}S~}&3mVp
zHTHq~zW#mrO*Y832A{P})hp^N2%(3cj-@OWC)?6jaen49=G5ESDR%(2BS@1sm^~6T
za8lxHZRC-dCa%v=of~~5=96bDl*yPrIBmF?ibn#88=qgW(j(E=n+X*5NMM-=V5pwz
zU!<Xv`j=k7`Vo9gy3cy3gUTbCLr59<(47|8;<c}}>TXfgcbJ}<ay+nYJ#+S;42FEY
zp`3YK#;j&jbA+OAD(z0pZEd#*2b}))_k*>vxcifCt1JA^%@*SF%!IbFuF~-u?Pcr4
z!t-Zq{}G+>Xzhi(caYhq6@+&)p`cQ3!z_v~++YH1<%AOP1veG>v40V=btk9cb+0x1
zOu!H8F4vM8`ONM@h9ZDK_>9o^73i7=Z0s|E0Hop%SfPrD--GtG(r2#g$2vXY=6>U<
z>=vS|$BEAbO5Od{q~M@g0CAm#h~>g5Sow^u87w4JhKv8*R#n<bMXL$S3Fz0lx&UiL
zEa1jxVmWaaaGk8G4mM=xteJg<fg2v;*HjR1q41H{1(3Lq*2@3=_P4e^ps#%Q2pX6~
zu|XM%oZ3_OSzx~J`^sWcU@s1`2k-@LGw1QEL68qsY%0QPZ{>`;Q3OkPTX~4tcxdgY
zZ<utNeO6nK7co@C$Y*sDq1}yD(PslK$Co&>$37bqc<aloP!ahMD5(fxyde`lD-`qi
zT39}36A22Q)eAPMJd`cteOUgiw&5cva0vpssOSY{6<(pp&DAkyA+Y6Ys<8kyUeB()
z-Wn=WOkMCOC#!;lg!bLmRGOz1ECf}BNVjz*iyLmUmY_^w?X9xK9vFTdrT5XMb@TrP
zNUENCCi1VZ5BO6sC<p38hGpE$IbbQH7Hvbd_u_qvwx#w?Ua{@ni9wWPJ8Fb;1-k8X
zF)NFgAP7De+Xv?0<1jMrw^3y+aaoLgE--QQuoa)vggZ(hrAqmU`GPb5##P0=2~T`3
zu*&Y`7@E}o#~^abYNd#~ldQ_;blu_az@}BKxO%fyKNm}6<8xu`TvPk#BGgs<H8q}a
z4VA%B6LhPs?zVySA={4Aa4oxt+LZ65broCq*2k@-V#R@ZOc~nQkCR*;ovgwFaZl=9
zeeT8MjNf)Wl?=zq{j7zY*9IEdbv@Hffe+g!8>-nQ-o0DdZRc>ZN0f2PEw|W?GUj>a
zF)`NG#7yz#9%*aViKXrKHKXl78bpqbwRQD#7`qQiG7Hs1>F@z7sbFz?#m3v(Wsp4F
zIBi`fGP31(o6N>s`MBKHF@2u9IuwPo<^_%qn-;_a?nO3%ABP&qDlCI@{=TiZ1A88U
zj2<=g73|9gZM7Y13C~MS0SHI=2@Ce#dNv0O6CObhe2QBZNNsg990PU)6`#qD{jQ)M
zKW4qGk5mttp8h_Jib;;pb+0$G2m*EVgEM1_Q+O-qW!Zh9`%c?dztE+&9f7+2ZnH;Y
z`mf;0b>vaa)t&`R(0Rs;KB|FFt@yyl9u*3g7ID+3G_JLw&K}ivI&b5T3M;RHQnLva
z@put7nZ=iI=vq!NcL1`I>0?}I(~kxg!@FYiwyHmhv}r*DHA};jv3jNtV^r73X2VW<
z^-<k+gclDT4%S{X%c65+YmaK}p-<?-oApPv;AYMd(l+c*TN=&YJ=WI5oaJEX=uZ}2
zSAV{qAT7O|EqCz%on(C#x_rq7*&%<t+h$H8h+W1;NMDNRW?R`amq~A{l5^f5(^L%6
zu<vsFe4Z*S*^$o&y71v<qn{Ugr~!{@?DHC#8?0lxQG8ywaEY>&RA~v_mGRGOvxH5B
zYxYbiZ`N;rwG%tJEZ|%WZf8|dp%qA~=`4gdE~_RmxHmFWt7Y#_!CA`=@XqFNQ>8^0
z<gR80jMMYh%rW>|69-M7*8+56jHj}#E7xxVB-gFXbxliGhG+|@KdP;>I~Q;_Vdy9a
zqb|a=<h%gng~8L-PayktJtSwv(8jY(RSZ{ZL#2y5p;NY0%GC!rQ?|1XR12p2J6X;j
zj`ju3E>X0&Q?R%i$v=2kUx0+=ZTbbF$w9HQ#=amlopTmw@r6KWn)jE!pvhqO&f~(w
z#=j7&0OcW&rgXt^Y?DfZ=OruW|969&O16ZrU+2tCD+`huyeWMVfwR@rXraKg^)JL`
zJ}_+!mBDi~ghBQaf#9MwcPm_W1K%}OAb9I~7FzO`aVv-NRlk6({e@Vxs4`h6YriHG
z$huh!^m5tv)Z`&79<8r*aRv^~AR+U<K=ktkZQ0!g4C_GI92#g`ZRrEu1EynJKwKff
zb`r=k`aEV|Of0ZES&}1P)Kq&AVs`{K`bF(5NrT3|7`?w4ry<1`g)Yv=x|Mc;z{KI2
z7+2nfMHgL6WWm!;)lYs=``7(WmA@Dhd}O~>vJoEsJlv<V%#CQySJmKgBxf}xaT;ay
zYz~uBbQG<j9Cv)TnvD^Cjnhv5pU2!`EnC1ShK;_ij2ln}Z7D;x-qy}yoR8qT^F=M;
zd<sG^By<#(yVg@T@fQD}^_2lz2=a2E3{HErg>x)ZjBA*+;Znob$wE%5ZYu+7huS-8
zbXPkb{e;JK%rVfB$28|*UoiTZFtd*n0@lG}LL<^>_hVQ*7IS;>sFhTG(dDO)=|YJ2
zwOm<EC>Pxrn@k92xFUJ$mmdpE;c15#Pd4=$gq9q8*&)J%Rv*(0DJVUYQd7x;-0@6c
zJ^Ksm4QfFEizyb}YJ$K*x~aeK+;ilBtz|o~(9k|wS0UD+wN&Vm@4&Xwi|&oKqY8-R
zVU=4~gC=lMFUJyG@Qc>ZPfjU!L^}|KYj1TOpvR=|;OA`UyTY&<3eGIbbSlZa9W9bf
z?{&UT0(2cymo0orsBo*l%Qo^Qp#qi6SytibmtrDeksv?)Qs8pqlbmp03S7>FaIK_|
z;*zW6jjICTLXh?(EM8r{izk)E9q_c3ck6g)l**R^-?OM9<llAIjICyMRH<C$p{DA4
zgY~Aa%%MlzP!*#8I=P+o!C%dqU(#vd`o<obckN5UH{oUEFl1fro?5n+;4Ps_+WwN}
zG_50~_o#x0k6Tv_!0hID>1Av2F6`$AfkAi+X-;YQG;412OV?lSB4*I;*4U2LUHl5F
zG`0fYn6~{T9gJc@I){8GC;rj1W?$CA#tn~+d^wQd$`SAp^g&22-%nYJm1V`3g;2M>
zgsW1@a_$h^#=k7oUqVg-|KQ6y-Os`+RkGMuwz4X4<{`AJ?6%^5O{TvbXq#WLs@j7&
zh>%-VH=crh@MVo#4yI!!8!AamLXJC?oQL#lW+zdnaF3&F%E3{bt!D)!oYqozbC0*P
zFL+2KEXS8MX|6qheg@ZUH!Ggx=;<YRRw2}(_46~L-aT%E9PHGTZK?%m3=dE{6bN{s
zU`T0vt8(P2<#Qo9*wF`p2qDm{6X?Ni00ZMrpvUEyJdQdsw<zP-SSR-zlaA7`Y8-YI
zw35=Co=CV=+XET0k+KSYLs-X?`Q?0Pq;vUoG#+6Ufxx#0JW{o}@#saX3fhU9D@fO}
zj4OxMJ6av6yto{%Az+FUWUH#>5_jH3s<m(#e*K!#!LeLdAr1>}uvUUYU*6wkZS|md
zZPhxe;(}o7W(W2`A@@3gxTvlpr=J5l2Db(W(OLEg%VM(=3+@#q8|LinbN5kO`oQ)D
z8hbl?z^U!><DHK7d>D)c4gQrF@bXGMcmpF}Q9dJt_-AONUkU8Q4F<F4D}hf}e6vF7
zVix0>(pPks`J2JUzoK(S!N9=QSE8G*yTK+ihw~W?L8J}>5eH>O9l^PLETieK!~&^9
zGF3Cjb*T|&*RtI*tEH}X_i>e-w1%n_C)i5a<jbw8{ulWDVfNP4f%N^zt)yHCI2TT~
z1Q{01Tdl2}ASGU~j@rRZE4tm??7-1e*h^&Jc!ke%-}hB+qwV6gK~BmU_Z_mCotwX6
z!%UW;>Jrp`#<HC~4N0KwC<g-rcG2T8yUXZML3ux}?Xq~vZ@*D}K}tLSb4&&Ag2sWP
zodq#!;Rq}}9!Pi<3iyv}i4!EiRk<b}kBMHv3E!lWFzl_Yq}!oa1at+pIrdNjWi}z~
zt12~zPE+miSQ+@I;hf7BW))w#p{8#?YO9ZHWz@q@1?FF>KaaPkHMLS4cV^?dT5$)5
ztl3tA_Hf_rEVj=b2zIhFE;EC<pknhgFWzT8r4c-S#QLgoH0+-T*@KYEFk_qQ!Ch!%
z4zrm9u5qd~9#GzG+X?F(2#$AhF5*dUPiWDfK~>F0p3p_)YI>tj#FSnHsbfl>LxG`q
zA`lmGI(WM33xTEk^BK=CkMDBcCh`kKG`=REh)D~v8qY~9t0k5>l)@9*t|&VZU=$mH
z>Fg8dKBt@-FZRw@P05Ytz|IqK!f+1n+H5uwq6<=2)poLSnh6fF6emA|ap!^Kwyxwp
zcSG_-EG*cp?I&Vk!Be^5t_nL@cdpJE-Bm~3r;uE#^i;T*Ti~2UDIvQM6rDPVJ=%e^
zFWJmoyz|>o=@&n`@tAGp{~i)}TeVyZE5(O9IgQK(Ys$VF*n%8zs_v_Siwr6_s;osE
z@N&c{3&gB&{1smfR2Hwj$x2_<3Wqwi@vp|94}hz$#-PK2GlQN~`F)fr3#!{lpbqy{
zAu0mY%=<?As<r~%;TY!Wn$1R>3ae#XocUH)nj7qdHS+&D&O25Uq-(ebJ=s))v*Ct+
zO(k7D&eql2qqEjh>Ns4WuZlo$l@Mp}gRksw&?VjMo~xm|w5pd4Bj~lhx{9|-=d%p-
z1<%v3lg%6k*EDC#P#|4j7m6X_58jWIX<HpxnzbFJc;?uX(IxjK<C0i7qa#m>VB#wG
z$jh;}p49QTHyj*|J*jC;w~-urPijxPuXHQr$I=QU<@l3A-<?N~+r*QBzE>Zx$tUwf
znp+CT;N>TU@4=U?k`>NF_D^TpsrahWcszp1kS**FFUtBj9y_{`r54~Mu4W_fNzj8u
zu&RQ`+)0$JC1_kc*t%N3$a}x_WUMQkIYunbwyMAvW$kp7=h}hJK<TR1&HwkmSWhkS
zj75~^XP-)<3kZRZuBo6bhuh*!eQ<xX?T4C?S7Hz*ED&-N3&8NgwXK3K{^zljwFWIW
zm3S%_F&k<A6x-^lm_3iyJNi^00%<#x?buVga{Ng}H;Si(Ly2Ka352<g3vT?WKora{
zNQW%xO5X6lp9-|$^|q`s#qsh1tEi>owCm4LXND`tR8@TLiq*0Rdavv!6=6+5->Jw+
z+(bVW&m7p_?+K+y=b|<>9D-Qcnvxrz14*iXz(Y2+)IVn%I6Ekt8GIaKxsxALzAd^r
z4*b1B-S-6S01D#$-Olq&w}C!hGk=oT+U8T*p!;f=A`tKr9I0Wa<ZLU8x{8!-Cv=`I
z6x&g;ON(6N_H-b`*=8e8$E5qh;JrEew06Ud_%<9<IvkW1Piz0+u>@CHPfIs7Y*DUz
z@ECtu+ccG=xD_X!)-rHk#et`%^Hxe;wUyO&aU}d9Rk977rEm_ahZi=UuxkEabIa_S
z{`Whro_(!j+1hwoD+wJAe}b<*9V=o#oQKS)@%*FunyPaX*t(KS2m)GJC@sO;*{$Oa
z%sOi3+Q)bZcAwsT^tv8TJW8#Xm5H<BAprU+j$=O@XJr=5sX2U5Z{|QaNmp#B=FzUV
zEhUQ=U$AW@;Un0Nl5R<0U5(B~@z9_n-N2qJlZ-0q9+AeBoWT~RP%=FKbhW44J2`^z
zZp_QR!#0us6AlMi@1*{pI%;M84;%AYR(d5%g*}u_=NF<YuE475;0&sk|IdZz>3Ua7
zsLaS3nR5(wFG|~8Euz7iYx}4ezP@j>wannShW5@nN5iDMrJ?URiJNvnwAJXXH(5t@
zoD~lPJOLbb&H0{2|F5^)Vtpmg)CsI5uv}@bZ@HNjDobcH54&2YZl;g)+_u!n)mPYd
z_JzkX=j%>>$pYEF7UP6aLDfg;J$nxyu+gt+l=$$$80>4Be%!=ztQ0C1-2Bql0*M~O
zcl>M7J{u4pOn5(>P^Dn-f*XCs_4hR^t89TQ{O~DI$<_vI@VTZ1a|L!rRmtMTk6SI1
z)6-U02NoB2lQh2e|GQRM-K$^I{`7FHVRh70&4sx$ww8HP*=CZ~b7at$JStjW(+YL*
z3+t?{cDU=YQdXMd4s{=;nG6IKStb3C_mws4YP^FSq-bKOakZ8Fv7t)VkTBU+4j-F9
zyv}8pc%9IP9TkJh#Yp8Dt;s$3WEy!!__@<%siV&XCBgf2mltbe&jeyNp5sEwYNzN^
z=^0@Xv~mrPClKffo=TaO`g55TT<h{Pno?e65~f$4(OUBmA<m_qRuh;9&RO-D*q!&>
zV>R`m*i*N9g5xfEt2I>O_QSTSRPkTdRM`lbx!aMs3s}Fb3$$z0SZg<jgBVFtb^`7y
zeSpLh2<O3dm4vBpJ+%xC3A?JF-S7aGHh3maTZC_H%1t$N)=@WnCRSD6mmWEOOFhM2
zeLG>`AQ+EL`n&XhnLQgb8``GN+sLy*5B68Qn4S$>?B%2zdp0JAMO-`^$lSl@_yH@a
zfII0MSMt`IZ?y?Q<PqUS{p7RJTbCoaM5?UMBE9`r1P!MXuvL|X2>}--3<hPv$c4M`
zY*Li&Va%`Yde(E*p@#Z)`ypG+mYG?eXw9tTl6BCJ<GIY9z3kJzo?p(k#hzRPmv&V<
zv+%D#M>^To-5f&QoS$>9xZ8U+R@44ncj0}_!JIog69lK#SX!Iv;l7Fuv&h0l+j>@{
z1U#+qHjX|+0{3QZM<1+rP78ZZI|wnKjXW2qxPaxPQl(pL?76_yX^#T`oOa4V9DPG9
zJeP<5`?*iX^MgkeN+t<%0Wt05bDCbH7OW`RD`(~r4wuSqnj2+0fiLsqWyc5}d~-N_
zHFXB|T0JXF!QE>-r?tAThU<FVYANgLf8FysYpTz%;A~BGZ@bmjm6}?#)^pmbxG#j<
z;2oQt$1fOE!EO@_q?;LNHgNl71_3CnuLfRxj^~j6#{LA)3)v-NQMhr}P^G5aVf}M*
z@!_QAb<^?foErqAX@(+LxSb@b!1H;sox(qDrkw&EeLkk04a(9Qdp_oX8H-0;p>mg8
z@?Oe1dtpk9Kd%A1Ns2qpnaDiXNSn;`Mcyst=L6Tpt+!g``M@>Q;|px{>CE^a|7lgV
zUqJJNndf7{oQ27d-Ew)iHM028W44;fGDhQBlamH|P0hMnU|Uxj4j}|~b_Z7*)=rp2
z#pDg7j3fTO?ItAHHSe=t)^<(1zFIqp1bd*rXPlti)ZaD2@C=_9iJM0Ha70|EY%2jS
z*I8KGnpDJ1_?^Pf+4EY#5ko5sBVW+MKfq+#$O|z~D7jzYoEd#VYm(ItH^atW$bIO+
z7o8Dd#S1ZEd-hl<p@h?)@fU=keU2dBh!e_0hyztVsXA3$H|lT5WzPD_3tAw)qo-fc
zyP0-R!Ps<}NmbeOgELO)t8ZX$)L+nWf+TkCW+VFy;p-Mls~Qk~E!>o)sV}?^`^eYS
zKANt6$O-5%R}pS0ec6@I+BvWYw%&OmR+O_lyZQgw1#BIAIWx54&h?$V7Kj38fHufJ
z%=zVNn^_UDB8FYrTV8W6t=rm#<2IekHQ;Ou4ulA{$652ZuQ>a<=J@0aGsXoy^7ZK2
z#v|NDU(Z7-!Uizxu`CN~76K1?;p>{~V;oeauLlb6L0ee=a>}2z2|+9Ff`6BV=eR-3
zO0$lT9erKv#$Be{^w)Ja`lIS2R!vCrz^{E>Yv>FL@{F*qEFzkqVy`rsA@jB>Xt*-~
z|IGv)UM<$tl|`2Fx+>J>xU*VHL*$^jt@19_gh)^Z`dYB<vRgP(hV1O=iyLm^3G7s|
zV~~RGBVdqiUUJ`In}V_NFphYJ**>EFt(->~AK|8ITXmcmhmI5UV(|o~m1ez|kaT`A
z(u)blz7zGJUd)>!n2zbA*z2;^qNl04_Fg|zt<(z)b7JwFg6A}zWzmg*vml6t>t9c1
zJ?Ci<L8P3Y2+|y9m8|BTp>4Vs*uH(*s`)n_{XtSf!}F=tdpb37y57*=h#FB3Qg3jv
zQO6T5@}u2WT=orfRpY&@&O;5NJAt1h6o<7EiaQHzXG6HRvW}_{1;8~5yIBRTsf*TA
zW3J4{kRGtU`gG-6Z?}Px8;_#Gl>p>jbeNz|u;{ijXZga)DPni^bZ+66c=u+?F=EGQ
z-w4#qcwmK*Z)i`|SC%3K;^;TD#1FbL`PesdUf`;JLx{qf&acu^ey*V5#pyBrjX*v6
zRnE$v&=*08-U(M_Qmye=M&mA9^N?gdg_UpU)RQ(7)GQH+vU6@_Sp7zzZWd=~*&e6O
z3Ffz+9oxWDkaFl%cm-Wm+eL(#x2C>$>^Ku^O`zcuqYe2D9T@2umS<A_!NW$P^>+Rd
z?3b^wI6J$2U@>wJb@LNnetskOa^mWX9Dn`YiL2WnhwQ$4RyYkdRSp-ltbk$m*5#ns
zi>rUIJ`pV`=cqgY?tDY&3^i4@E4=Y<25LB>ZRDFV`Pi_#ONh~L=8n2I^qt1?(?v(b
z4JlL&28oqat>DaMJB)u*^WiFN5xsWen_3ty!drrHoBXC0#xmy1R#sgO*9*KkD&LIt
z!~j8`(#3I>Pc>n}VseoIK5GP%XOA<CziM^C^cx<<yljA1D!e^4;DRtNo~C*|gYbV%
zrCj5L!%_dAL9p8TW-OJXN35;X0nGbVmeZR3*;U853wS<dQ(+a{&v|gqgPe)ke`nBa
z=G-A&^e7F({Dchl(_t&?g?9_v&M)FJgXA4`tRMu1Z3LpQU$T)6O(9(;D$a)R#G{o*
zi;Zn){vlbo>X4onDvZSg6Ia>@6c<FjhEN_?)o7TK6C1mj3M(L+)Hkv>(WK=BO|FzU
zD;t5Y;B`%BH-ga(r(9Xr-o;9)snNr*QP{fmgyK3tVaL$OKGhMyz}Cxb-^nN`At>C?
z*5)|FQZaNj)&#t$uxQuSF2&<L*UG6`^3$Z9eOg)2?Y0r9t^TKVvk|xcb?%b&vKmYM
zdF!j()z{lV4b3{yu$kS2W<}sU%r7J`_?X%fZVEQWwkq%2Z#$~Y_4%~fw}i1v^QSl}
zzZFP`*hTX;nprr3Kq2B=Iw<$=wc=Ybxy3%7`6Z=UF1&%2p7F&x@vXoZyIErA$!}>b
zEMqFisLBG30udh-f!=tSH?XQ<E8-i9{UwmRdC*AymTtL&R?m(NKE#>TQ0+xDQCIVS
zPJ&C;%#Y(9WM)lp;LV({bq$g*ziy-364ZZtxwTb+Wx~1K$=cW#va!^waFguO8hV0O
z+zpodDmc$U0c%2iImU^xnXPbH5rk-%HKvaowyjKKonhNqYIY7<-*!~}Jn9AZqE`IO
z$%_b*!OV*>{dY56qc6tva~NWZ^rChRv&8{cD5r2oh$U4hUVPZbGkJ8%CNg;sUI&##
z`e5Z3wFb&gHB|IJR*W{C@Rk`*HCuuq!&<6k&KYz)>#BgX-x>)hT#K(}TmSwlYv%vB
zvg+8H>Mn#w)r1pq?=$To9NJ1B^>pV&Eu9m3LwD5?f22C~w>wYl6&IBDU(_MvyN=Tu
z_h~Qgy0YLs;x=007id#JFkW)4zG1fPrm<TI5EF(_2j5ofOnkgpc2x8*`o?D8*15-W
z8Tq!*6`sDiDo4K^IC}Bp-@4z%RIEV#zt~M;{_^eX9mN4IXdVByhB<R8{Kih?=ip&K
zUTpH)vH15)TUnraZlX1r&HB3b=g#!Eg-GTNquBhw2|>+js)b4d42rDf#w&9;acN}1
zvva8Vt$tf83SFZs2r#~_TRDU+JT_q>2@rv?FuF`YCm;ws(o{>s+jxp+v5EywI{JI=
z0#QGdoOe}mHAFZRH;L?L)#tdMG}QlknQi7AU;ylbykUMaTTrLAmEFd)ix-vZUvWje
z1pGY2!u!m=qr>Vvtb>v7=nT8*CL7JakGLT4I~pVemb+*GTJfEj?K{zJ{f-cuhrW#`
zh#OvK6A6Ml$R@uNvv!Io-B6rm6<YE4lrr#F?ys2XtTF}LuA1q+oZvN;Dq!gPJ@lQx
z;-QE^+{i4yy-wJwD&6kBr<y8O#9CZaF)XI$aT~dw#oQ9Pr8Kk=tXEZm&4sE|C+i<_
z>-27Z;GY)aAwTv*DOq20@$y@2pkX=3Z1Z}roF6h&6LZd1*~$qz%7vq(%UE_)0`rnt
zV~<9xk6Uhem5nIv3i~$N*UCG{%N?zuzAo2Xq%QL*vLfb-R#F;gUT}{3TJ=*dgPh1h
zQ;0MtGd;_g%1U2yoz+S|(0kypP4{Edc^KZ{N2m+w`n}fIx^O0oJ7z?)OT`NhFamwq
z=)XAXj#~cN$4e0kZ#72}mz_1$@_)Xx`jD;lV>ulphE6oNo+arzPmWe!XTtUDc3%tj
zKVM~?9OPcN5$x_Z;JPB#%WCX{Faou{T3lMe8T}yZU2-2}o2m!3*L2Ove)ySD$n1Z{
zY}`@}IW*fAsEtSQD!HTD2hh=h47TrTXSh<1>q3rvH_+-Qg^hk!1L4^RPYLR7?7JFB
zscOY{1I;*(aD&3qcXjGXIl#tMZiYQEp>ix4CkQ76y?AN0W#!B{zu{ghE4!d?vS|T>
zCgf>#*{TU!(hjQ$#Kt3ltAAIhO~VxQ;u_zLMdU{KJd4$T1UBiErJ0~Iy!e_5*K01t
zS<eB<!B$II1rPJCwKYea;GA)s;+^bk^>x<G<WZQmN*;BMBlQuU9(ob_dzK?@GYfyS
zW5e&p+C6gAwz5w+{O~E>&KX+7h22hO@&CqWU|W1osDkYu(u0xjX~Tenj*aHWUOXb&
z*!KcY?r^U7o^Z%}^fViQ$jQosA#LNT!wZ@NVj=-B8ytixRG@=ZR;w6Ru#{L76#-0@
zZ7!&3RYd=viq)+6s#jX=d$A@Wh#rHlC;YQ6j&FP~hF|9>{9dfLSw9(?N?-9kx+Xv?
zpm9BGf+b$RDsye!@sb&Ct6`n|7<`PJu1bZw_ulukQr{tsJ>?X4Lkthp#sXn}HdT8j
zUVp<~P%j|$+S127cM^|Xjj14luFGt8^3wxM`cwD&F{<KUY~=ewm7Mdz_4|H|GPpxy
z-w(98qd6W=$|w+ETIu`RM8P3IJox=U;tKA{C%&(BP9!N9GQLbEl{t4FDaZF?e6Gj=
z>!*@H+!Temn-={29;B$MEY|~cJ@{IdJS6dVT@Z`C|I-@kVR6P8U#t4xt^Awc-z9c9
zI~Le&Yl04$0*^Mqxwa~+qLl;la1U*zS$#-yHAncD?F#UnP$Ru8yw^Dr*^hnTC)?xv
zAxt|!H?!QnsVN(3*agnw8O&C8*+aA0wq_qLqr#K!BsQ4X5UAP@v~h4{3sd3;y2spd
zX!Hk~=ODyR#r>5<XI+=Q_(5RRg`qsUp0v^rw9R-VkN+T$00Yixwh0x)U$|UbO{!q=
z+8eB_=1_>i<)A7-J9(VoX;osQ0)msds*<7gQu{$Hja`;U{RgqoZobVL*}savAXf#i
zc#V4jv1Sg;qXDdC&NAFAjygfZjS<J$3)V^)IF<od`N5pS1<ZnwVE1#zbTjX0yl4eJ
zA}sfFyn6@{gP6kx>fZ_2fC&NCq#IQZm2m|9h#bN#wk<$MT)liJ0l5jaFLV2$w!~>g
zhadT2jC&upGD=;&&BlHh_&ehIFopj8SG+u_;x03IJmH+1;mWSr#18{o=ySp^waLu9
zW0Ch4x|KiF-sLG6Ix3YP29}pNwx-pV8wXoef8(9rYHD^5`d|7JxYkIhvv4_C606E7
zV5c_)1@jTCFR~QcOYU2JT^+-)&>u9dgodI~JL{a~t=drs6$2X8u&(m<!O*at;IUBP
z$#W4YYu;7X27-3R_t9=CIYt-?;IUnf&Q^8`Hvn$_xt(2~I&M2Z%nd>~be$shBVmZp
zgD#OD$<ICPb<EaBe-sEiamvPiq&;_J;}Jwt`ElB<7E3=80wdbnk95MggH!s?$tM73
ziGuK>z!}k{{i-OdojG?f#JZ_uv&$Ls?MI?I<8}F5?5=87KM+1MYWd}Mm%OCZgJPCF
z)X1&@7q=Ukfh5qHnI_<|t!1I*a~qG?x-x1TPjTvHAtx)_Khj)<rz8(CXq{|TI+EQV
z#X9rT7uMOF^>$lsaMbiQ?wkM920x1BH_I`(nI$09pw{)j8wzaYta11(ExM1*?d--O
z==2?>&$<_T`*BP)G8$A^ejIR%!Iy0G$I%lvy^bq>?8h<HemHP06{=LMV8>cg+KELQ
z|FOpRPm*&&aQyrVe@>HGLT0Bf{4A?PQA<X5<;Pk*MCSE7WLkZ~MBb`D)_Ms8zncDc
z!9+cKa~cLtLuEETH^Yl%H7lHQbE#&w<KbA{+1#1{oVY?9r5qfGILa`sAIAc{^)_oO
zRa>!+nt)-0Jw^_}nak&4ML_0dm)U*~!Wr%hTyY-i1dGleYyJ9Y;*#CWY*^RuTZY;B
zGWXn;`g=RG&KpY^?#|kF)IdIC4c5p1&g;baXfS36!3MdTaMJrEM{o}2sEXia<qWjn
zK>Nf+U3Du4nqc=;L8GNXEC82C+xQ?Bz{XPq@EhpJTMn5CEcE<doW<P^e7Eu-X6Yij
z<_W+95KU)Gr(sb-T_~r(ILm5k)gyHId#pZ)m6FQW&%^L)475SDRAQ!9H3-~{LUc6M
zw?*fpXDbz)@nhYtR$#Tjr^vB+=mnf4wAINQuz$+#`-$k`I=X|TTX=m#Xr8rR_SJ(u
z^#%Wg1doI48269+k##P@X8z@5S-yjY39ybA0jMr+GXV(<zz)pbV=2C)mI%@`V|Fbj
z+!e-0t_cq=s~y$f7g;`I*J7TpaFeNk%O0(CO<Uoh`|%%FX+%FZq2y(phfZG8UUai6
zw01cj%2|mU4Znrggx?qfnbWHIKPv~Unhg<W&1%;)PhqR~EW4wwa*G#HPr9aM1tH`I
z<39nO@K|Y06)eucqrE2e)?;&Ad#Qf7uFkmC^PIJ^EJsT_yPXzdOm;_^dvQ(M%@LGc
zF}|n&{oLtikB`ipu|f6<ht=4mY+kQ{Tz#1R;7FV$>LZH>5q?&!a29T7;fj0c+R=v$
zX6z?1TNSht!nxVVPjpfq16MzZS#U)J_TAV|0*i1x;A*JQIai1IiT0e+Fs$Wqm72l(
zy-lb-mzalCo75MQ)O&felz$>L$1|{^ELUt`%BBUPn1V#8s-dH|-)Xg<2w!Yl#m}uP
z5B(2oD0wR<+iEt-qZk#3rkZg-fo!gy#Dc57!qyX(Uzsl0wUz*!rPuABi21sA9?ggA
zvVe{N0hBG?Rp}X6G(G)4k9G+6Sb|;PnhRHmgB(hoC>xuqR16-pq2?S-Geicb&_P!*
zyby3WDt&F86}h9al;_+M*X*Y{!k&SQ|7qaDtr2YWr&_CEk5xUEIc_IXD9x2c42+*@
zQ@|HG?=P0|p9b29o*ReOL{?jNS9wa8kHhybC-7s79xlA{)7W^as_6O^(?1m&T|LjL
zKaKgVOj#}eUM7l`)di1RAGU46I|Q}KYb%QlKE`01KNWu6Tp}VluPOKVy!#Gb*Wiz{
zmRssND>nF(Z55lpu<;1*VihZ@w^>)k3f7DBfP0z%r0EJq)7MyvI(jlZ@EF}7=U^}L
zflUF0{9zu4`k2AF1#hpPhQb*lb>u+1BA;ym^W)Yfx?G$)8mOx}z%jC)1&&zx?(AaZ
zXEAAD-mf!#5YrnT#n{hsjW3?M;0n-I{7i^;22bf{xzm=G+_ruEXR#bEKnG7`DG#eW
zDR|QMD`&=GCcC2aj5}MJ&i{$NW7VGtYj;^qy)5>egN0;ul{xOt1sd7O!9#4_Rej78
z%!ad2o!WB}Q9eh@T7p1Y$}zG2Gc7tE7jPdj#mNy!8_U`W71H8)>*(WbEXQv4Zy6)0
zp87Xcw!TJ8aBv$a37gF}m4A`on)<W)`jTxWfc<Qc?cL6Qf>-QLc5v!E&V0;%9+=<~
zfh-yMc}&E<SKH{%V`?GtZS3co^!m9K*99nkuFb|{o2O+dtIxrMxA82=WBFNTX+kiN
zBim$Fx*h!_812e94gaX3`aBldxzj(_WFZz0iAObSp|?d(Yd_aQ*bk9g&zuvy4PXKa
z2s}qTx~=L1g3LKgssnQ!#3&s?mi4TE2ae0;td;FL$+yBd+RDA4c!?eLws68iN!<Ed
zPszm<!Ubn_d}l>;qd|7G!u7qGQ!<5x7h7;B$nlix+E!)~2ge;8Z|CPRG_jFN?j)F3
zqZDGl2y~owS0BI7dBdUIwb5S$!d{Ev%`Y@}ICOw&<>dQCo{0;muyHD7F2SqOGhpWk
z2;>dBY9b->WaruS3t?NHJ+1r;5pGP4Jfy>M`bFRY@)W}27jaYk)4X!;<=fGL&ep&U
zCj>e&s)0NjdS2<Ub8Te99NRE&l~F*?SX1eZN15}r?18@!u@!M%;d=Hl_!&=H>lZpQ
z;yL6ZlD4YHP2b6~VYrEAle*>x7>Md?*2`(4F3Yr^h1oMs^$dO?D$ApB$W!X(?mQAU
z!%Z27IRZ3MAPR+~;wTO)6D%h*_G>#?>Yy7O*e}&v>^3^VWf}Qp;P5#7_g`v-A7NvR
zDeYW6D}Jf7^FDu|TPaJqym|bWdAzs-Io?|nztp)JZ|TWj3U|m)kuyeqIqMR9&MH}#
zr_P+KdrvEGhGWdCYLRG++`Cp&&e;pB;JVV`k)e^LmWbN2Xsf@}$~g&X#?&+g-cJI*
zEp1KXW^p=^$bMY^WvuDq_5;>RK+70{qh+B%NTAxuUgE7kNcpa6L8@CXv+C~040QDc
zMGmEHpj_W3O!JernL|I%c{t2za?hK#l^M{Tw*5=JTzCh=MR?Y&v@_Rzb=#&;N2HVa
zb7#2AM>b;!8&AOT-PBsyHElyjZBrZ7eb!i^lFkJzZ3dPX{T3R}T8Q&gRyCo5#p_;f
zlbgE-FE+23(W@S72m(K|JYJ$-W~#I>Gi%k&SWPwDBI(Z~{vb=B7oiiCKef+Rb2Jdl
z>|cUXH?!2nCv0s~r-xr#U@A*O33!}WnDQz~=o4#elq*~Z@DKI>A*{N)IeKi>pyoBq
zxb*Zy>}NI+(GZzrYe&Kr?`D4akprr7r0y$}Rkfu_aM6ZsD;*vOcd}R<30<>a#oR-N
zP;TT`+E#d>u+d*>PNz5zSTbY33Iwij%@vt*9`9(ZgpwfOLIR3s{8urlhxXcp{=OT5
zEjFaT(t6On;r>FhCEkfvQ5xNML?#Jr@GuLpnuVMhUHet+Uunp}ajmOAXF2j4zX}u}
zC7Q#f!Kw-dgUb(rrXbF^DW0wEf^PwST_p>*x^y=cEj1Tra_t=B+$j(Xsi<My`daH|
zr7Ni@*;7tskM$Gw-Ohs==<lihwyD3lPEe~H=7^y9ZA;xo#~bZ7+s@i<@MSx{iY*eV
zb6N%VYi*H(-kXtM$6UC3p`h1Ce;sIZ=a@G3Yasx4q)OfVA68Q86#lLgjJqPT301=r
zB~B<*<aMaG=fBV}R{3=vR78Km3*hvxW1=%qLp3v(iOb82qo#UyOd%PotAnt9q^;KY
zwGJ>?%Bz{>Yu1U@rn2VGAT!C%&EkK;*0ai<6V}RxIL2N1psixgHHYK7+X=V4Ldu^Y
zaB~gEH$iONYkl?HsclLPG*}3dP5sZxTVah4RU26~oRBSL6|g&f2PLfiGq$57EG4tw
z#KilXz(#(fF=D7Os=x1YhR@h<@@i>7w97?p@f#f(GnXPxwCb&}`Nw~wUE)06cm_{o
zNsk}Jqc}mqt3X)8rhu2B0{hs9P>Yr9&Utt~)5^iqr@twxztQct2cd);>ejNJ3?}No
z5o2OErmS<=8^4KlbOQc-*5xGUjf<MQT7gPCYirrTliZmzw*DI}(nIL4SWCqg-8D&D
z88D4l+#L;g76Yel4%odI#N)n~e}p=YTiN#$iWAKPrSbX=LqkAHH^0J$*($nMPuiBc
zy$Gu*@p0gy?WiIy7hpC0R+EpE7gB2Ew}G@{F5&uZbRO~s>R{}*TKI^m+(l!NpYV@d
z>%Sxz;a3{0sbX<@k^W8yV1W|iq>?qRnsU|;AFwFfW>cX-KK)yvW^a0Ht^PLF6e=Ak
z+x|8b7_g7^AcWVI*y}k4f>pp_Yp4P?tLWUSA?M@=s`NR<ZRDS8S?@6Gkh?1Dxm{07
zod`pDx9Dj9HdZ2vRh<3qP_UC#T}XkJlZJ*v8S7~*tPMz>eon;|)I@BM@M+C9mA-QY
z-HxF;1t}7*z^$Ace<9M7?QDoc1(fe3!1D3q6|>*v?L@@nk>6>nu&H=yGx4Lp3ru=c
zw6QF=A3~(aG<FMC`dy$tZicVfc;>jf-ib_~=8B#CT`c^W87u!TR>ZM5-YWXQs0<oq
z`gg+F$;&+Yxy`EC8xLUPoKUN&H$I@c%KlsatdVUU<mOxbT`sd)V6vH?P=U^8T2o1`
z!j`S8@5RA6XRo!CgD4A4LUpV-h267t)b3ec23^%ac!(s&de=ca3)A{piN_W)NkKz#
z2=~9)P7V4J-v?Rjh;stBvf~FhTDR3sQNdwp{#~KG{XQ^s%-O6Xzt?&AYS<R~+xbT}
zru3Y1=ZoLRG+uiE12p$LU)n9lQX2og_9=%kH{QhW6W#twa1w9w_nJ8|WUc)BKw9Yy
zR>^u$Q$U#(L_)@*fbn~s#D_e7G(HnxLe*65&2P4P!b4bbp$6@j^)9d=R)4Qk&Ap>o
zQ={5=hF7x6xNOMQv;2PCUbK`Z@+30j{Qm^HqaDGCm$P-%o-@gvQPj&`^N3&?`@h$D
zEJu`_nw&@jb+vNXHh25;K@d622KKTAwv@j1QO?0_rB}|;@J@~myL6YP*&j5s#F67t
z8TrF5H4e>6Tv*;#f6#cD3bfSnW3NZ)EdHQL$IlMrl$HLV+wZ~^(DjEvz|~u9BEd)C
zYn$BtT3ngsy846ANMJMWH#pf9frs?N=k;{9g}&dMRkI?zGj6t;T6W^w-S*Vg@`^{0
zX{eUmjjkq$ifafrRjN1)!({Cbu^Q`G|E>Q)n>T#)f}@sNqC-wJwEvKE52fJz0e?kz
zvfIuMyX-oE-QAwLRH~<D!nt=pXULVIuxT2o;x5CVb8vT5i%Xoc!-N}5e&^b4sh1ni
zqEC~M{MaYf%#LzaP%%ZD_kTm|b>xVR4D%S`N#xYe=r9KP;(c6TL!Ff#=B5}X!jVJ6
zhtikd^;8+pa%?l3P);R|RGUoTej&QPM|r4avDb;CN`9b33*H2y;v2P^@b*LW*M`Ej
zn6%7vUHvN-&;d|61mjTb3CuBIO?~V|85Hs&q1CY=URUoB7Wj*%HPlk{z|g#?+nSSG
zZ?sN!J2VvNc2{lU6wP|s4zca#tuN4=pr}wQ4QKD1ZDx~)xIBhA05|Wz(2654|8Tw0
zu&~+A*05W`_7OCxHcuY**&j9S2ZP5u^2fl<#>1c~Gn`MuZS}|9J&~;JLM3l{{mZSC
zC1>zRi6e_Y2IBAHZKWco>F=bH1#<YZN^`fburDeJ7yMx~!c0tO$s6Bb)jw+CocFks
zNKR^h%p>k$j=1_CW9e`rzsDMX)HJ&a8#?l<st2cn)lGIYL7qVHug3AP;Tk*Zf-&O?
znyrL^0Nr`m?d+Ee)cw(H9hEIkaus#6W9X}LTVRNgb<x^&en(&JE#b~@pjt?8;p!%Y
zBhIkMv@`#=)F<AI_j0>zYmiv3Fl{?37T%@pr98&~HKKHJjD0iul8#24@Z;CyBu0>9
zF9i~d<43J{DX=TKu$7cCPn?bMEQ1T(iI+63c%wpb@Fh*98{7GIQn8bM2xc+j>kwb{
zr9kp)Utv|1K(pFv*{fMd?z$TG+X(INhU&`rY*lqHG_08w4l*Nas(_Dom#MA`>J0HB
zS}$o8@SfmO#R5MI-VTp-#yWz0G{RSR1rMD87=hUe>|CX;{)Y{6aOhAueX^O&AAqGd
zRG~e0+gAQttyXROC9QfahZF~CCnv!Lh-QC^tJ<I5E3k?D{u}v|4m3AC8U2%X)}Am6
z9{ZDK`Jn4q7k`TBd)duaQXw9hGcabxRozt+Y~oMCgfpPv=1nSVX&&KP`A<Ud#v@iy
za?kOyP5()q#gzdOjLx01YIcC9I}~0myXGz^>Z(QDTWe$?_}S1stJzDB9AQoMK)SaW
zHD(^32f3aEE&R#01nn3m{B4zSIveg$g65WEw!5k^<Ib#l*%#DItglww4J>nLgRI0U
zi;lx)e(i(8$Gu<n;KtHT|7KhHu`ok-w$+6zoK!mr`$i1d?a#U<IbGn2N*!S2&ze_+
z8iawe(Ld`pdXC%Z&q5fXKp>EcEEnqx{?VU>+2hb!<C!(ZOhbxH{8=O$0Xy)_#My!=
z>YroL#sksm{<1(rMY+d{is}b5nH4>qpizOdYW4w_tISg^zYf5F#k^Zxz54RkStEPk
zT(1kKZ1vAN16)7A6+3Ywmo*pA%7d=Z7bhZ$kGB5j-3?FYTY{wEZjgfN3ltu&sq6h}
zST|c>pNEkNj>_^MXI7xYNt^xb6prED1HeGt#lwwlDresc#y?OhL*>k2k+qd$x_Xao
ztGV#qwUeMigY9c(e~HO=e_J;4mq6Iu8`)uh2~>)&02TNbjqB{jQ=m_I1V19=3iQNZ
z0=K|>mP2+tOTxh5F|4ct79fM3%zBXQ9<8+em)NOOcUwj6`cj_$i+0*Rs<8XPT2|t=
za4B`|FFLY<XXc2@=0Fb-yz=9Y75rDPW<^XwUCpVPpHn;wSQu+6>bh9yJ*g+JELlsl
zfpISPRMzm7<}{!>$<=CIb<&pvZ$001S<N3N{T%b`qPr{@C<nzI4n?(z;W_;oUZ1rm
zuia-``Smojh?hoHDo!B7+Q|VQMN7>cx7lBXFZTmG@>fj;1S0#JXO{BZGTp{94;SW$
z$ROSSDhv)j#iOD0SIzR`Yz%;Hr$7FA(M|j{kh<|q1nHj4%4pf3@suC_QnHGgTy#Qw
zT6J*x>_KO%f0eHlW{JUby*LA+mURi~!cm#!?_!ywU8pZ#^N^9NSuld`HM8Jhs4lpK
ze~ndoVwo5bfBj#1Yi<r_Esew-&7P~d0bV;H9dNfT9d(xpLy0E4ei09KJ*9a$5s6Lk
z&lA6RkPRP#0>?%{U;9BHteK(U-X0HxE!8RRn6hmp|LdNT?c_AOE@ViA>~Dd*H}JEO
zzr_yYq;g#_8`Wn);FPn><(NwH$tyXtF1ht9$K~I2mO6pUyX$W;8}1Em;%~8`2>3nu
zH_gjy{>i^73m|SSIV6?FGu>I6{+pJ<b<R&UTXUM1>8q)%(|=aaPWhv!@wZq{Sa2~j
zs|n;t_*bFJ%>;t@*RQm-toUjaxz^Rom)~kFf$+cpsHkZi#idKuQE4_I3__*hl@PKu
zM>Pn45vWHH!bo!glhaLY2Ks{B!*RN8CU`VI&e~9LAh;po)zo<&zuSL{9VEpIaKEGP
zf|p1Lk1&>W4MSLa=in7?1;*r+95L2bO!vS4yNzw-q2zPM-Y!&-NCd$-DsAb0!?ec6
zw*p;P-HX<iZWcKB3)0h^ROTrtobr~C?EZz3BkKz{G`>@&uXoCy$*NO)>pfP>|G5aT
zSyk`$@3F>~Ff2!(oKUL?`{2wu-t<j<?Oi$Ly4Y)(cQQQotPA+9x7=c_1U`k=2Www%
zi51w%;k}w=+f}P4ITBT5ad{D5bU&e9#<L1%qOzoywW+zgum~s0#Y{s%n{nRnmYPSs
z(YCY2coptwe%&X4A1d~DEk!P9=y!b__Wf_8f7j5psSxRnDHrvhN0IygJ@6t;Yo*cV
zzt6`19yluAaF0#o-}gc_PAYxy_QMCPEI0+1{&QCOyKsf<7~QGqzsD+BUco4yyQQjH
zNDk{(`+H#g41<$@ORN7~TbBK#YGF%5&sP&B?@Cr|Q$UJiP)KWPnhOx*t^Yk%UZ^9x
ztgDp;7f&J^&OV0rK}QAMpOEv<tecsaoW$$pfcKz5C;(2F+Cbg)D~ZR3>!xnzd_t7S
z&Cf8q2{(v`-d46g<z%CBF7Wo}+}+6*_VFtIhi3O6LG}I-2tY>_g5w{WL>4&Q{jq;&
z-@$%HQ8zyp>o8VJ`2pRAJ-$8|At%VhKeP*znPQWI4#yO0eC2UwxKZ_!R{2LD$=w{e
z5NTR{TIP7F3S=!fWwq?oT?@X>>Iv|+m%YpynZ5`)yZR5^bW&7G*8<J#jfaJwb9KUO
z(bb6Eb&2Y~`0&PK*7`>*f<?43vXGzBPIBZ}kI!+Rc2y&&yx#xE)%ynJomKh1OB-9d
zn^x4=qN4sfO?Q)Sy5&V)P#Q-FkS5(EVSw(&BhKO9Lm(mW3_K(xrf)Gb@f^;X+&g$q
z%FE2;-h*>;YVcI7VjOfBrF&9COz|Rhi>}HGUgQO3<pp&`MU9GzdOvG_*OQ)eNnZT+
zv-e(m?LTYpwbx#M1RSRG+o)>ux)peGOGfOn9@V%(z#sm<$%6B_|KFc&MhE(No!AOo
zkLl6cc9e(zF4RAx+$vt5?1$RO3g_TGewe1Q_aqO+@$+c!1*`a>HZ?*YjYIp~sInjG
zEJ_DE7%>~wssJC(6#h^%;i?sMxvmJyFU$c8aZkDpz<Q|B4!D4}AErTjvD>=zV}^It
zo*$+~#JG5f)1e<~5#d=UXZZk(qd=QdF%H+!+@MXX*huQ`OrYZ`Tae+bl1E_y{YiaJ
zK$0kEK7jBl24A`m;2UUa@rPQ7^e30zBqOi$92%fP)=iDpeyGhU9(RyJ1-{LG9&QLc
zZe7x@C=f8!yz2+JCYI5TMj=0FyZU;)!)#k{;ktU$`BS>BaX`ceoo#ClaZX?r%2@qo
zKQ=I#GVrgBEyK1Zr7?BEYPZu^AL+EZ0CjYfd9UA2OJEp{fR=!Ij^n#+J9YomMA(XV
zsTi?pttU`R-q!uaeJC1owE()75AY|42@Ei$L?JyOPDc)XfV<e%BAb{*vxt+LO;K%l
zz?HdZ;{q;r(5m89U-)pwqg}%pzo6b6J!Xr_!B58MfImkpM;e}tr?xY8Lr|v>>a3}Q
zt6z6XyZ{`2)izY1hVyYVMgYBN+uGLM*yWiI9HMP&f9uG*cd=ci+zC8qg#9SZ3Ho%1
z<Mo}nADi+Ysm}1IKiO9NNYj{i9wv({g~+asENuPZ2$khQqp4M=6y7<Vt*(!PN&C7b
zU<~%2MTpU+hPl&nZT^wwpXqm1=$;>?iRMmbLqF2cce))N7c>a#M`>|1HQAKDKKogl
z{*jg-uOafoG$ROEJ3(6lA?Mb4gA8~m-J;0nOMPBN-G+1_5JZIpg@DCKjS%3Sxe659
za~@~eZbZkpAaey%S9l%8P%_4o02j6q5RRollbdSj99MxYeV%61*$LE>DwaRO?fyvA
zJm`Be`hfOhonM3KH2qk#wE7KLjrjSQ&soKf(?oIoWA$Y}PD42u{%Na!tnmvy7zVdC
zfFSU+I(;{c^TbgvC}<`Cr9}y1W^=V``*B(pV<=bnJHHx67tMN99a9;24Si3#Yw~>5
zKI^>Xl*)T>l4$`#%LM&aoB45?$nJ|atKVH8%tg6BPLtXA;I=y-{hRRdT2QUeAFxFM
z?m**nSpbmex_s@1G8%Y7T#L@)a^BVhD-KNV#*x77Mb0*(R0}k?6`<gmz;UOT!*RT;
zk<EJ18e3pqej@TX=0jBalQd~AqvpiN1pFrt_FAq@FisUe(JV3nnLZ~@<5K1)A~*um
zaLup%Nt!xm>fOs)oic-|um6eWegYRnEou$D1PI<u?gY{Kgsy-HRdLPiQN|l@RN0W~
zqpHQdN`)lsv8kVkonW)VEgqBLLFlK23(f)4L;j2a^|BskA9U4q9uYwWVy(UBY+ki8
z#JPTDUHFL>-vBxd9)tv%io))eZaKcs;)Uae;Irm!EqdUuQ274|h(oC2klcvm(Fx2k
z@XQo*Kzk6C;1PP^l<g=3NA1LT+*L-wxf<I^^Dw@8%}RIDhz`3QXZemM0{bOi+bfiV
zd=JMZJ84po0(<xjtB!(69mZ;RV&6dmk;bV`*-2LZAFO^S4nkLWYS~Fk#IulAE?S!~
zIq*3+3e`$O|EBdsE#{W6*Fq)3pNHk6ZB`3M@l+IJW5nW4SY|+9bZdpkJGQ!l5{44I
z5OD<02?ntJ@-iO~S)&N>7IqRrehI5B;hK4`{FP}r;I&{RW;av`mHFhrGqM_eS6o*;
zUqyaJunj?M9J9@UIB^EW9q4aM)!Y^nRkR(g6Xc}ssfQzPH>L^q!}e2+vEOJE?eeFZ
zD;_ElyOk@|!IKkQOvO({CA_v<_R}=g$MBt1{nIpw8F9rYnZ#NpbK$eQP6b(TRv)l&
zYsBf*B53$u$3Xn2X>i&)t?Q@Spiw%2o5%gxPc_T<>>)l<w1nb~FVMUopl*QE!GKK=
zAC~#Fa^OkX3H8m>CekK-bC*Y%I=&C%ZgO5oW=9%r;io$8+2E34oZl4|l|6}8+j7)G
z<=$=tI?%Qok*yU0@(8r94t=`YHuUup8=<;#E`^NT($C11c77_BdSWsiD;%V@t3kl4
zLkd%D_A}WfC9a#*C8toOKht!$gM#w-9(S(&XW9+VBPY}+kzcG@pT~J0to@nB7|tk3
ztLmciQ8(54ndToaOgt~OD2F%5F>CvoX5yUNTXzW>8vEAsvov>;r))?iSWCnS$p1_$
z3m^FMS~c}EEfbz_3LNFr(I_4!Y(`m_n5TVgHgG^QkvrmC6zoG~aXxa8A2Un~(FA(c
zcwB<pSd1nxU>HLKVL32vcbU-*fx){4%(}ku^2$sJ<hkJi_b$Bgv!cF$bNg~XvZ=l{
zxy$JyW(dnsnvUkieByX5R=M6Y*lYbfjTL4|P-E%OHKfD}Ws@%dc~A>Gi_=sj2ha?s
zY!+pY@ji<i{hw<L-6>mQuxfv<DQ_ciSDo6(@GAFhtN(eL^Wjs>nLe||^VarrEv=&b
zop->lXrLGSM#Q>;w|WgluA#uhmBzVm<$s>0m5}FC(a;$evrGp{k_EDb&qQ;)Ue4<C
z`(K4`7hv9#F#Rba`c)8JPzIWJxX4h(D7ywMU^LG*<U8MuNOB4$e_KD-G0LOSTey*c
zlL6Lx06yoEnvFoLJ(qXC#h3)e4`?;x0Fmk1`gz(2^`P$BwzeZ={2|bG1K5cqm%lFx
z@hg*i&?~*K;ho@qd_T=*lA<hG#r-r&=V3hbdxi%Bw%BS_$j}(BjSTc)CUG<vsg8c)
zx>>#2ags>#?iN)$4YqCfHTV85;nug-6(CO>v7Y-{2L47i6yFUnO8NU*TYgYsvOT35
zBfWTXzc1oFjs(YMR0V#Y^D8oN;CR~Rq6wJjd3|z(*n*Hu_i%r+TZ}H8Ocrv>0@j*C
zI&dTEV|cR`sJL*6Rlgq1q<6KA0ML)?jUn3<fV|&|wgiB)!**1oq4Vg8<F*^~ftqK*
z%h@l|d^AJ9rTTPbwDMnQ)`}PW3Z)qh1PE-<*)K$h{WuLn_N0o#xVWg*=K+`F#;;iB
z*Z)GJc`S$AmM8vSq#<;n2eYGMh?1;JP|tF<a`Ea>DP&Pp8Tv(<Ctf@M-13n@T%&K>
zRAB0^Tb&!7j!e|Wn4VvxHTA~hHv5Zr)<B~Fxj-qYTXPv!$(OiXsio``sx^!Ho^)8t
z0^E=^PHqU(cne*N(QtnxSU?5_r`^hVL%nP0XtvEjflb|wgps%ZLhM9CA2O85LB?Vy
zs=fsCVY`7y1Mlkg%hZvM1}pug7VI(C%>AX7B5x>MgCpZGk7L;{(|9-F8hDI*ceQF)
zf5vKmscmfizsv0w#;0}4Jf5@qU#6wNV}-To6QPH-1zcQ2TGubrv?mq9ob{;C7*1!0
z0+Z3z{|^7B8mkMO4^zKXACWpEOmWtFTDdQyooX}sPVNV11>3hHG{xtX$`Dlei?LeD
zT*MyewBn_BQ9tK+f>{m(S}_{78-W;3P}WqOGYQHGnE2A*+}%))Z!)$SZM*Z5CEE&|
z-TATm8-#*1o>|fZME3W~L>-BY{aw<2rFp<W8TC#j+<c$g@2}Eqt^Ok>4i&!=O|g>`
zp%DS#uhMX5o#(9jS7~~QUr_ri5tg(|>Qo&qO-xdi<4$qls$0~Gj2GJivdbvO-5eqC
zVY6eF)J!KgkD&nPE>!Zr5<!f5*gU_DPN`fYa-?a&V|!<_rx!CqZvefsfPfIw=AxC(
zmgeI&ADyXs(H2yO+NW(%HIFyjvc6hQ*o}a38fsnpRa&THL^g1(rgcGKJ==!x8)O`P
z31C00Y8>BCj({C)w;drdmL5EI1EY5Lu5Y^<?g)%vrMqb=($&3O-<a9)gs*Z*0Pa!N
zvfVV9BRS{It9La=JfhXA&gwV0Bh{%)=1N^wuZ(_QVl65^g7CO4%H(~4bp@PfKW#lK
zo`TT~?WTFJsj~d;J8C2DqtPj4YFAU_Q7Bkzi#9EggR3h;2>W)mQjR4Btl3@du!uGM
z;eD=%;D2R{oZn4E&%iwh9=0GP4!f*zQQv5T!a4}gXTS7WyAiD`@)E?l77cbJ70q>(
z#Ja>b^x2O?=*{ToD`U2`E6UgH#kqzZH8kuZ*4=2d4+*RNI*mA@MMkjn*BbF-#O(XE
zwmN(+TZO*QC7)WqPLRdp0vLA}f&i&<GpYTx2BwYM0ivT{ry(Agvie_ZL15C9SlpIK
zzYNdftGrE3uz{cw_-hSkVj1w*siNulbAlfdl3!pc<^zgbe7XMIl(G&SsIlq53NDPT
zPQUk_#@0$zI*U(xbIKyjl+8z%(dV^=zyWEZum45W@=Y}Ld0EgiI8yFUzYfVN5r0s|
zP1LZa0=$_xZ>}~P2(`JP-(Alet-u&p7YJ`@s9tNdofty5Z62~+mBHNyuT;M&o}<aP
zHFDYCXzxSoljzL9Nt4%yIIZG0Y21%+bNWr1M&i|T!TpWqhF1+1M6HUWZf$juF+teN
zoYgC1>>TcY1mC$wxNwzXukhP`SKyK~SA2o>{3b1ox36%mznoMXhXOlvu?T4;sIJ#A
z70qFO$Z0pNj1ko2W+DUAH<bQog?jcX2k)G6AVe4b%|}k2y?7zMN14FIXb|OWZcRl5
zb%BQ!S8lH>tJ#{Wjktl;`fqeDzl=DqBT=bs1SBFta6koGsMT<&ZAF7^+|PG3&R?#w
zUG*Y+^|;w@MLF#V3QK>h3G$egR{mS9>sJ+(wL<0osQSQXEsJslOF(?|Tg{V8snLQ_
zs(1D8{dlkYZ5m~fMiP?L|5lX6!!E*L_`7KS2oKb40-Nkzt*h7u^KU(Y3*R2fZRodY
z?O=&(`QN5R=DNgM%2U79orQa(`#t5j;$krU+qA;ISY<N;>n@6TsgE|;PEdpqCJ2kU
zJVh@quvB<b3>*)$0WGRWSp!yIoZ8&&mX(_{{cl9cPR_iw_-<zY*41lF15PK4){Uq$
zfiB8slxx6vZR<g&aV52*UN*Gj7%lqG&VgO3*`E3cdys^XiO5U$(nP`>!;|Y9m1VS;
z_B4+ZlS$l|?WO7RXA@|;S_LzYGgKRu`cJVH?xksh^Fw~miB_+?VHXXw?4?zbOHU1L
z0tWz;j;w1>J8}#AplTFc@qTD84L|mpdDV42<&!q0(!B_Hrqv#5S12vOa?AwMdL_46
z0ZT!p=K=(txCU%q8QF2{wifh_90Yx6p@n(_uUobp(ERNRhVe!~n3zN@cuk4nq`I*l
zId0Zr8<FUD-M7sc5R@_z1L90UU21}@2;lhK^~a)Jwd~e7&hGs#O$x7waC^UtRf}x)
zcWDw}U*ALnyyAD-D7g)yAIYUNQ_1WV?n}Q*gT4BlQRG{-O1Q6g{3!e`&M0)DecP@V
z%&gPx>#Ri$XXxr#o4}o6tL+km41|zF1cVTD*wF8^Qm}7y$x1#zAeXkOsE61C-XdB$
zIqB+oGf}Jw(eA7uCUpn*$uXyrC*(MJKF}X_ettnn4Z{g61_W#hY&jZ2(#KI11A}Ka
z_Xx;77~_Uh7nek+>_9FBsVH_BXEK}mMp5}MZA*pRnFzW{fw^pDc7=kQx!CWu^P!rE
z=;-&_(Re@OI$0iB_(T4hRs22;Itt9}Elfi8`!v6V>G_&f|6Ut2D*94NMq^raS;Z`&
zI+f)GjCWssXNMVy7Jb8up`{eRzs^eP`hA>`+;0VExZu_mL%-KF5iix}Js<eBahltd
zP*2ach>ZlzPYRofPPv;VPOD=3tcf|b!<J_A@$*pze<8X<zo%U`h3?AY@3m0h{ts#8
z{<$p+s2+dqh6X2z%Ku{hUu-R)9pN-t4=jmPe!(`>;uwapoAEnduu*<d!nYE%Uj!Su
zlPe<Z{yuFXpZT=eAJUK`%3`MeplR!L2ejpX5J99jAfQz!Q#yXxA2g>t9kaBnquwjM
zJd)S`A%Si@YIT1IDfB!0s@1EgJgv6;K|>Fp<$-kD0t^f7FRV*o4zu+|sV0I0_<Ti@
z*K_C0^TPk|7|#<^e@IjO#(^rE{)5IgIC!k&0%Oi%gl4l!af@&mb85KLUyyA+YQ_@<
zuk%q86;jTg#c1AjLnys0G!GDQJI2eM2f7E(H6fEdk10ZYHojyV0ygYNVa_(A$GDxx
z14*=lh!uuzNBzq3uH_zW;*+L;2|Pg%+}f(C!%FPdBd!dn3q~J2eEg7{xA<}{vs<NJ
zg4+puH*ARN9zJ3F?ACid=dcUp;Tr9hdx-DzSof7t3Hc~jE6CDrm3h&`l!iKNmCEDM
zchIWUYVLvTzebtKq`y|}c%V3U9CFZZJ?yXuaN=RT0|`QsI17KsZaw6gFJVL!JvrXn
z&rm_{O@S0boPIy6TI^O*gdB7Yl-7WQ9k%zn&2ByJaQsN(eNy7g&L}g;tAOtZkxblF
z&blM%T5QhH3X>CX3j^wm&#Mgu_F{NOl=J;R=V3D+T|Ra5D^}2V)_^I|)C0J_zu3jV
z8bgLNHm%9{^iy_8&AGwdOf(@CvddB39|U0>W>pT2T)U!1Tu@?j(O@S}zgN}5aJqcV
z2Qao?M*Es-w>2HHg{a|ga@XzFV~p1D1X^#1^WCG{;)B6S3cgD!oesuj0cnD*UlIOa
ze!*@$m>llJ9Jx2uuABG3V68=`TZeJ7csmm7U*$wwS3>CBQNl$jy|$rdveUzMH*kF!
zg^5jl^R6>)_X0_@{yp^3mTIj2<4M~VBwVU+*zUwi!Yvr8t9Zn@FBn)s@ZO>C7FSc*
zi%xj^?z_O=b<0U2g$*uwS6Xer;~A#(T_U?9{<z|5p=IyVx;@%q<*L*>O1O=E@v}!u
zRecY>Q?SZ+C8A7%58r<>)yZClKB`m@+iB*qIsmLrTaD^uc;H4w5Fgf&)!bzV-zAbl
zcML|kPQae@&<BSC@CjG6iBi0~b22tX7f#>{DIKRR0gnJAt%2T~2tQVsq_%)Uv}5bk
zchfjazDwWd-6*$P(0_1cn*Y`lz)lowAo>G~icw#*iU(P}hL61K)}v0S?`$3UD8K$+
z8JI#Ok{^~S0UhO?<l?*14%pgZ)9=zQ<^ei6q%H|{LR7%DG14=|)(KT;mz6+P&ElCA
z^8I7^^L9m$&-3$KV8|i;*4sFtx2q~QiL+vx4`>)3LQIk1K77^|)P3Sn5n9Er3m!@{
zwiq4shijd3OHuDwFT}7MlY;Dz8{3N7LIi<YUKGPbZoqCv0wYgYY(a0v{PMOcjQhN=
z>pMlBxf5Uq2@{L~ec<56j_4?`yWI_Vz5b%T8HLlN*u5BlI5^+3Ej5>+GqD{dAvZsv
zB9*7foZVOR2<Z?p1$g{*kJz3v2+x%_d$(4qoJ^IxTXWY>TyHkMQe}pxy0Uj`d+*G#
zZpz=CC=ss#<F+p{s;c)tWfkvEgf|Rwl`3<z<M46I^s{?^u8JlHrTDH4SBa}p`K%QV
zxAxtMAZ3XIDhE~UGxhteE{e6myB<=p40+S)H=xyep~At2T&?NdTFL{6qFv1o6<<Ng
zjyG6qAe{^(+uoh_TXbNO35x4vcNX*imQ~*sO?Dz==oSnl`qmQ-<lVQf$EURc)k3Ju
zZZ#Ch_$1p%acGCf{b?p2RefWV;zB`fuydk+_3lJX=nY(aw>Gwj0f%CKpwhq~UD7vi
zY_PxYP6uq`7*d>*C}^ZrbvcH5g3v_RGZnk^e?bvfB9WZ2&#6ys?I-MN!0!Y8d<+-S
z(66cZ$Z&4;zu7{R!Q+En*Eb9;0w48pazYj#ORApZd2=}+wVt;X;gq=Ux9o;8Fz>LN
z(bc4anfy;gGw=*{TO%@wW@St{yq}9G>`tU%6L|3Dm=Y4hk`3D3$mXt-!Z2;dJcP;T
zJ+*%trhO~g_wze#z}u?o_uS;WY$rz3O)MttzH(t#T?6Z~?FP1ZJ+M7}v*EvJ_K^6!
zF>I{d0;=SpL?ABgEqy2v5$?PvtV}<@+S+O54{0@`PR&zndcocIkZA81y|xNfgUK1N
z%7?T*oM%G!5DDW$Y3c0$yj2DCKO$H}wV)HH)@t<gzs~>bywyG=>PJutCv;GauynXi
z)hU6F19WppiR0;Mwm#s#%nE4=0KRP@@egEPD*BagRZ+})P<YcgxT+mt-xsa(p*Sbp
zcLx{uA}R<^;r*2AbZHQ}Ai~~v&Vd-sqZo|25%7?@`{HM9<RP)dsOltJR)5FIM;qLx
z{ZucgFdp)zij(3oc`+2fs1|rjx9NwpYxrX*-a*)VE(L1n%Cj-ds83i0p0UeOb|86y
za6Wf72ImBC|5swtxR~SoLoH;_blcT{;4$6jg$T+oGJd<JycCmWA?gq026J{@^*b9H
zZ83(KOM?6}EJeYj%(|=uOO0cBB?cgA$K>orB>mdJo^<oU6eRx$YXOGW-f_Dfoj#8F
zCTizGYtWT+?x>g>9q}|J5W&{l-2n0OBzN^q4L9~~_oAUjxsuun*qGdKYHdHjA}Ocs
zsCOPb&)L4b-B;VI-{Bo#SMbL!TtINWr;MzTr+h0(RE0+xD=E>Uo*3}}qovBoA(JSJ
zGKd1pOA_U+ex0LypMK6dFhP#0?yhCRDpjbL!_~495ukg)wJK#~?rOcdB*+^>1z!L)
z0?|J*ocH&a5-m!uNUkl;zH-tIMhzAhE0r5pq$x9qoq!z@KBMSY)R%}x`f`b-Yl@N>
z>?eJN79r(cQYWl6aCZ?ZHkV34X+%e{&gh9-*mgzEn>fk+XX}oxw8QN5l%$o9I6~Gu
zZ`pv#f@U&d7h>liiIKpBLlsdGYFROcbg!T+`1Q#oo19YC8|X}46u9t*G#yQ!;sSjs
z;5AR7Rl^v}#Bf*tV0zvzM<%<}`9xy^lX1>)%3g_F-KPn66uHM!%!sSXManC&=6oR9
zT(E0u-;FqIp(L*DDMT$h-gVXEm5WF<h-Vg+#>Vhu+K`t5d5`+OT%yw&c2U#K2>zp=
z8c^}F-H4LI2rX}_b>1v(O#s;&P`qQuyB%n=?I0ywj}%6+$VTo2c1gUu;yRkKtZ@0J
z1A-~J8*2rR{pS$)ZAP<=FrxSL4LbNf6S6Ng+P2!5)bq!VM!b+b65S6x|0kTIK41Lb
zx@}Kj2Hnol-lLW2j%R7S<UMH!KL2?ueNUpS47DyRdyk0b9Fk)ze@`My_W$pp)U{73
zSEysOE23ZvQhTdZ!aX&fvMh35c8Bd%kvKKUX;u9mQ9Me4HSZA(y7xv8BU~Fz!TTe(
zI;eur9ynli${=)14p;k!l+n<3)as)<+?9RDZ&D}H6}=_0CUPmJNvkRipF%qJ9-S0P
z{z0tzoOK3*hdugFSFxft>F)PviOJCo>}gNrBNgYJ!v-SRl+FShQak<U`)wq;#~GjO
z81ll7fGGo3P>)`hp5c_*M?S+*c~Ql1K7tm1AclB**)B!GJ-gaWbfUpSADqQ;j$KyE
zy=O3O5(xO4KeH>E;@5G)tpcYB*L^iWAy?rnD|Q%_QzrGA&}KK1WABCNhMQ`Qa5!I&
z$>>OWyNdzIMX$D`cJZCZ+lIcg0y@DfG57<+b|at_onzgMIsX?YJX;gA{G=-v-VO+(
z9`QgKGHbmP2>W|AB)AbXlythW0kpenZuJ@x_kb|~z2A%P9PR|RV)#;U0A450+jbyE
z9Af-uVwEZ&7m)7XkHXBj?`15sU1g5?txWZbY=_yyBBWN7n@S!|gfyJcEPYt)q4S-3
zg}1VYMUCP^tXx%b$9cx~sj90dB2s)f(G1ris|@fhoQ9dc`-Wkss{(p@rqV5<TD5xn
z`m81zO1k3Tw%Ugi*`GdR2laJ<XBz=q{m$T^9n!C_H(Pyxbw>%V0qaC2u-TW1>J=@p
zSG7h9=p~_TVQmlVFduL!mUTwTjVK_nhb2=~Tm{b8L3Z-)hqV_VQQ)$|%Yxv=0c-;g
z7ipHL1*;3t?2ra%aNJ!Lj0llVbXM{L<U(ieDt}=G^$RUB98;-IHt2GTi_v29sN7FX
zM`oj&yIzX!z>N`;)MjGThu|VE2P(xQfJcj{$yDX-iuww}?`PheM)K*;+EslKyLet-
z7cbhiVy~Ro(#>xnM(XgS@4eS!s4*^F420JIThf4CdN>ZgwCnKVupAh6u*I$j2m`Rp
z<H!vG8OP_rO(8lq>7J?A0^+C-`FXpoVq-8|>l&H#$hwSy%AJ6XBQo1iv7^VD?QRU<
zsov2fR@#hW7}483wWMxrDKU(L&<WcP0O(Wg2+skz*SfFjICO(w4BY&NPB@}H0r0yZ
z!o>Hg>tlqTPr?35-m9gP-sdsCEPZdHK#UUM9p9TMXB_L_a{aviPh9%;y;mpP7knX9
zyjN?h7~7HCS^VO+m0_HOb3J>n*0bN|&s&vJjp?dk)dGfK4Cc8eAiw0o^x8<c2UR;*
zOrQW@b<qj;Ym~Gj4k;J^i<m9d3svrLqt>Kv9t%+BR&TObk<-VgyO>gmtf7EZQJ5iB
z_h+cC$T`gJ*R2W{sL-QNbZl%upWpvF-XJuLoIA4KA5mo#4f-t~fZXk2LEk8_@T~V<
zZD7qQREJ#@G!}$S>+`jSR=X4(hb3FF8P)D{X5?})!}XD7i;1i3;8y~!@3`nA&c*2S
zeu>cd>H}mDb*PyGe}bqZ@EUdNLjBx<ocnr=v<nV+c0yA^d4s6(ED0`wi!qh4Wnlng
z#E!C}(Qxk(u3>IB1l?r_cJsYy`#kZAtp(CZsm|E#B9?^dMYST@b(i2CJ;d&)-Qk=6
zyKTJp*85#p%W7g%r8#ZbcVi*@Ma~~;HdQ10%nNoe8tOyzXj^K(`)S*bh1!n+D53zC
z=1%<5JmnFJTts<aP(1MYt~xQ{9y9j@k-HnwrU(J<0{Y|tqNG%`fg}}CN2%zCRZ~_f
zx&vvXQXB;3rHLl+zUO!9eWhtoI*(XIsp!cad-Nk`9kWV78Rb@(m1^^E3_bkh1)xfh
zJa8m;pz2c5bI~%qM##Y&K~a(CJe5a9I;gz#R8kj+ug|mX9}+r84_mz&&tA(}lS*el
zxb!t^iJ!6KXw}dA51>NS7O1o({nk!_?7iR`v|TX(=`3`ac(=;3KG3kYo+4_YK{ODM
zjwHtzLuvz`d=oYjXmoI6$j7(){3!&yWX)_-%E5u^n|4vi;wsRl1MDdrA&uFk(gz%b
zn-0&EipL1+X`lYf8i+4`(Pjfdv5efn1oY-#BjpGz+7QHD)z9O(A)AjL54eGg3+=82
zPM479FQ{Dq8P2H~`E$J!=wb$fiZ8sR6jq7ul6g7$alY49gbjxjdOQr@4Z(2V3=?)U
z0K(~FJP^p8W!2t}40mU0>yd$Bu-%ECL$J<Y;XT1Nlnu*<oc_V6;qJjU1qHQ2sO+AK
z4Z0+3OC9!S&GDo>yB+PN8_7-#HaL1bkNc6#`Rf=DU>fMQM%#;l#!3g1EvV~}L|s`k
zlveUcA`_eemOhf`47CdmwK63#*p~ln<&TKcT%VXTWz6<HlBh6iXt#<e=#D(C@{u(A
zu3(&;o<u$R&uxt{tx~<$IOM9M-U;WJYxH|Nov3|8L@eI_pi)`0vwKz-&3=F04yh&d
ztk|u_tX|MM(t+HhJoMa=;Sr-!wa#+7wCeL{N2j%^-zSEwGoYU4-75yh-GX#KqFrNx
zyX3i~KhPt%E`cE^tqmxNycp2%e?nDEij4%~sP%J+&PQik3WdBC)ZQr!`=-?1crVAm
zMU_Ef#6dY7jkAvoD`3|p6~zW0sfNCJ#rg}o9G!*Vcf;yg0mmJNT@eb&pdICvN3@A4
z8vBG@)!d|a1Ao7=`A0;m=@zQ6uh#@@eD#0WLJU&FkX?_?9QRk&#Yd80>V<#vXG_s{
z6aiRC%Q2?JnzR+=wE3k4S1^G(5$JKrYd4jN=7x)!)&k!Vrr0<4+p(6~ig9_@qm{&T
z*&QJT0U^rS2*A86__Dbh2xIJvFi{DQyp^+iO1N<kv=*cKRbJz_12}it+@S3gn?f~&
z+n0s{7xK0nNaEu)jp1IT2Xlx`xrErGiMaN^VI_|yl7kC}&>l_HbF{PB${tNL=L|50
z4UeW3#{nJskG4;>SmK!CJ*o-BtyKSM4%W&?6QQ*Bp6P=V%LH?Zy{bwX8Lmjs45(HH
zazm>TG@OW|Cj6-O>Gi8g``}>Iaj&4znNsO)iO^mUo`3$cR{y9r0-|l<la(#4N%#|v
z1JQCz0HH6Kb+;DFz7;BdZR$W1!WC}02ss3IlzqKRn7!t1Bf9nN&*at<SR!$9<;sB=
z0>(OdHXKsA0sjKY25ls;L79<*EZRpI0IhabP@(vP2O;#Qxv65OlS<~rXsba2^yz@r
z=h2rgsbq&7wah#!N=a`eqmwQpxg1Ellib;8CqIPyW*75biSnG&MUUchf<Bqt<y}?f
z#7oX=4d6Q|@nN8SEf8Vh^BNtA!yfDKx&{zCTwB!dxFce{FKN|1^OP+Y2aI#eZ+|Pr
zAxC9Ujk?Z{-Hd@4gwjWCO*zQTU?HQu7Jn~V7hIfC+=&XwH@9s>hn)`r93sR=)25K5
z>&3HRGf+Y12G?7#dja`%cj~+qBlM0|k8MX2E`~m7JHiRh1^N>tSIqT*0k*47jh&vf
zJpmmg`WhkB9@8Q92GsXhqP{ejrH=_YM#?=hCRVuYu|!K5UX@t><&Pzz#0mhx*uKXS
z31etj5%BicSfxI(fU-=Wvy;3LRO#ogW35*G*sXFHSb~24FMn3^nCOnU_Z~o{R!Ka#
z@{Ypmw32d|fCxnJkmT7RHFLptmHKG%71mZ$VBW@!9mW1Ykqx_5o#1rhd~MV3=pI|=
zW7<aWW}Ds_x&qi?xS{Uocn5AvttS$#!zX)fAkc0@6v9?Hq*mLL6}*io<yI7oUOwuz
zy9~Avt?|qN&7w2)m^Ki0G|mQgv5P8@I0V0rO$*Z@R>-9QG=S~D%?Qn8AbMFDXOfox
ztSX%2LU1Lz(g1DD#W%iP?W*u&RQ>IKKER^pJ!aQb3_%iZl~nBQukfJpExZ0$aISu9
z(Gk3;0!V$`RBI{VK-2iM5Eg@+z$*gT=%RreQKLOY;<_2B6aH9eYmsWhX$8+@T4s$$
zY(3y6xx<|(G=_Z}>K-u*V}pY(p1vEIIG99WwHZUr?w;Ii-;2Vo-3Co=39MVXjoP+q
z4iBbxkR4$>J_+{s1x4y&tb*OZ2D9Uw?Fq@(j+vEd87HU66IN1|=quy-)=HIi9?xUW
z&6brVVsd*kZfIp%d`LD}mJ_xw>bb-83Kc}{5N4z@;Ja-SqmxBm(tN8bOBDIy3s$Y4
z5#j->rYw>B$m%t#4X|g}jSs3x7M$+1bpfiQf9RYYiUhdktiDW}Kz_`1%bEhlxv^8$
zBG4I6?XCJbF`aGte)})q{)%-5!fe~1V_nfvS71XfXWapzvDbQ33(a}RDw;&b%gHyS
z)TAtNMjjx}FhVK$z?Gw}V1;PzH94G}Dl1-vISE}$?_ybznPa3mpr}7B=(sTOTiB%l
zkS-_0tqMero$vVLR=X?|2R)SftiCzNdhJT|3ais}aLp+X)y$k-4HR%t0>v=b^Xjtu
z=D!xGC;yTQWx-cJFfqCwjSqM{R9jSv$5P)Qmjq>Uyw8@^T6PrmsFi?t%;QSmP?=$3
zKHLlp(?J3fYwG6VIR9>|WdG_nZ9Sm1!x-Hu)&fEIdb%O#{Uc+jCfrq>Y~X@z3bpjC
z`7OH_5c0!Qwxt@$!-;LjP{2vM0}Hh20|Ww^`!Virw(D}mm?!*OL156xgzc#W$GyMw
znmw+=i0f6sN+Q7>{JU2A_$`-_WTz=r78!)1>a%iX!2T6%-{XlYR=<N$rhXoE_2kOO
zwboB1?YB%hDJW7^Ky736s#S`3EY6tKJf7&TvBPRr2HG66gZg%FepdH*B5zl6aScd2
z6tG@n!PP&m9iw?5`B86rTnjV>o06lM7QrW8L#uk@4inMz7Z8sAf^|kW+nmeqQU;o1
z&{t%jwT`I%aY?gyKK%#3b|%sLvJI$=+XirgX<V}EFWN|q1#hGbab8#)f5{4hk5|1(
zo6@g`4%x-{)%UjP$3=EbwEMK5uuIC!nsFvQBe3Jc>2=_;Qe194&hBh9*`Dt7R|5HV
zmq^UTI3aa#D!Qs9>VeLU+kCVLnG-9MbMBgkJl#pr(6fcdMO|EV{M`r$*Ht3x;3nyJ
z?Zp6*)RF#~Sz6K$a1clZeYaf>d^+7{?}|{x35L5+yAg=<zQ?1)%_w-n1$}GEdi(!y
zLqWS8BZ2;ct;aV{%yvgrWHGW4lW^py-BmwV|0fbU9_u!hEXFpyr{Vunm2D}*Z@VC5
z+W`Uj&5qhfmIw{oePy^>@~%E{;bePi9m;@Ru=nXOLf)B}<C6D@(t1zvLRk7fjUxfG
z<Mc0kUm`aK0(ZRc(_+Rymk-#!sEHi|Z)5LE6vv5el>vkc9hxAnw3WRt(H(X}4yH;)
zFMR*2R;|wrF0w`q@g#!YNBlfCWC!1u$nP_sv$|-mgMcP>NTpE`tFd~0<GSgrHAU~-
zjiYPlw-obGcxes123Eh(FTC2KLj*u@(NSj<&S9zA6)5z)VBPU^D+JUNHGE&<RbW7+
zR+l)&)UqDDxFe1T@y-s*zfbgn7h-8Y6rv}|mB3Ua+|!U<jHU*VwogX}*TbA$QpVd~
z<0*MY88Yj-tc+m~%Q&kHZeJv>1g0qM`jVMbIoPvtyQ*>?B^f*6XtFieXV;X0U=61z
z3(-u!-!`rXG<<#dYsjKHhgfGRM&O$`PKXM9+)!*q3Ah5*^$is|>Ee#~eRToeV@<#J
zQ{;9u;Gqs6-}S(%iT&nIj329?*8?&)1S7@ZwYxC@&Ma)|o1KoQuUH9<h)jt+X<N}i
zunOB+l3a;hTC*cWvgRhcuX2nTdmXa+B3`I9!oKdQ>8#+_Uwgkuhp62sCca<9fr5gw
z{H5>Lq_gpT*8?k9+56K%K7G>4->*{$nbg_ywl7k>XWjQtMS#T6VAv|(FH%MKF?oN_
z-k)e}l9&lq?@we2mx~iAK}f1UFoY_U5vQwGh0x1JJ23zg-55zn5ist4oh>>PnS;0-
zw)*#L55Ujr6nr!5rAep{pR*Q0OixIys??9E{TXWuh@9vnt})RW5Iw4dbw$GM-mF{L
z`c_8EJy9>|urq@LYQQBcHl***?XQ2{Mq+dk+#pxBysB}3T1ImHZ7V2ESOuj192iro
zI_d_|7vFy?+0GQb52UA#>1d||+%E;}(|$!qr9%Vg$I#^%aAd!;#SVQBnfsNf?Kk1M
z82!<7ue=(wi7Pa?@cBqVTTH>V05Lw<hv&Iswd6$adSoSw+(qF7Kg`_{xrQ1;mle-3
zLg<aEDqD#vt*&-|BO03EGJ{|vQrwBSnXak-E&q#L;&v37Iy1uNy)Lj#c)_@%j0l^&
zp|%lmOxWG%G<O)=RE37)O?FQ|yZ*3kY3Z;Bx|a!$*0HUE*@-FJ(JJa}wEJoU39_>j
zyU{=Ho3<Bd$*a4SYyBf}8sr*lCFP0U+?ItWF(tCY1n;u4a`9Cj{1Mr)`_n5|<#aH6
zXrg`PiDI%QtQ#uI69u~Pf#qEp&G<7Q9F1iGxgTl1Rh1{QeTk^$`Z{*lYN7z4I*?_Q
zi_mrFVbL5cPgI(n#-vm2y#1BL{u~Mn`p#nwrb0NDu%>ct9QcTFtJ@ak;KAHlRh<y8
zt}EK645GwYr#^>o{<U?*m^ia)-9nMzdOcB<#g6sUKp?#OUAW01q2b<NY@~=1fkN{F
z>IRsF_!);IHWfq5M$T!T*rAIWFju!sa-`|#0mH|;xm{9gxNK$oW&+jI7wmFC?O^Mi
z4gB0Vw_sP4)8YQ2=2Y7S%ZTd(@fMWFY+eZv89@C|3C=Ly+ZF`PADz~}XV;a2u|qOv
zi%Ple3ksV{(JuS}%G}EVo-=3IRsucX+YR-Ih+sIa8?~F7V>gk6tzb`C3xxdweOo|Y
zLfLl6)?+CAn!{SS6V1Ic$-A9xs1``B*xgukC<rHU-e!zbZqn|l*bk<!zGzzl;5Ln%
zVcSs&O;-=0ZaZq4i!$zv?nhg=tYHJ*jiNX=!cY<24FOGYf9(T0x8SwTSxF@PXtmN0
zXzxGb&KI~+4_Vm<bXXBcm;CY%q_y2VmA8E#5OHGZlx}+!ABY1FULPj=Evx)MTGt~u
z(DYbW*#|^_udx$W=^IS|Y;B`~Bgtr>MwQ%&zKecp)re~bCC2xlD&o}#&pA;MUX|_O
zkh<Yd;^S5yxOtFVzbZEc)B#i?V2xTnpxq+90!iVgHL%BL2%5;uTw9DpUk)XXVYh$n
z44_1zPwK;6Q37YlVG7zELy6~BI7JOr@5o6Th~|5f%LyBbBum`Y^GDQ=bdDnLP!G|j
zLX0SQoT=Qkr=q*Mo?jGn$8=hy`jd0@OG46RV=O>i`$W0(Xxr=aMZ)>c#$3D%Lv|(L
z7LtqKIn{T?-4v((YLsl^AeoQuF*p5oEizoDw;;&b+gI#*{M^CzwWy!Zu71;&1hxa6
zZO8|$!~n&UD1=w+8R8&##Sf$nw|&fRMj^KUQ??ePk0u!AtniC5h@^cFU;n_Z_aPPV
z2WR*#_v7#;?zip;0t_*uxgm7kz@5(BVxPI{aq9}4+hH8;#W!o;wuH$m(4Bro2M1PB
zzm5;ueSIDHqU{E9DVR8Rb$fw%CkiBH9~8B{cC_J$m3%OflRG+`Aj0kkQ(s$n5&NL1
z31Lt&_bC6MsEM&IjNuy3_I*&N3h&w#DvPS;GF!=;R;l!Bye1w#e-zpL2aA{CV)%}#
zfaV7&M+qVhp@B$;X||eZTrcQ6Mystc^xY1AFj4L5*V(@6^fQ(Ob}0G;brB~v677C`
zH3>3LPj2PYqUzZzsGPJ$D^LXQgoLDzYU^KzJML6w(MOkcMd4!5t!{y3X7L+nJpr~A
zo?}2c=^l^km#Ejk^SG-RjKr`N&RSlTTm)qWeGj{$?Ue9nPHurN22#9vowexzhG&fQ
zqIF44W&Z5ObMyywZcn&`p<tKQ5<9caMvfCWBHk;4>7M#<brvvCV504-!lTob9_FJ{
z92Is=Ejt+|HfkXtCa?azc0G#zJA8aE7Kf22<hB$U_&u~`eXoCq;a&NlcD(GtU-jwH
z|HsYEkM;Va(M^>_n*#}bV2C)FH^AF51`~PtO<j)x?L2RHqUEIZvk{8}@$TKI;;YT2
z+M7kBR^POH0`4YEtlTYCYG?o6Rv~x_LM}vwb_VZ$wAX@%<fw}Y6r$(77!qzDz6k9@
znhk#kaFJmokKwc(CcX4SI>jbapt7<LC5ptvmWInelvX$|ObF&bB&tK#P)aEkAJW}{
zHTn)uTlt|xiWBFF5%8fzYKatO<yI996N$(Ttg1isfQ99><#wi~Gpu9rETP&(6>muQ
zVARGTuzNhMD*`)@+y|<Q%E#Quy<U*vcsYI{>88j`s!c7?5%wA0@?lt7BNwOIoNWSo
zl7Omt8tv2w5ZgR&UC|<h_AQv^?x;p2m;viiLa|l{RLPajY)Avr2=_LkpYc0|Oi_dG
z2NCp92vDfm4BJ%9hcn;<b}_O#{hf1KsQYCfku4Q3MPKm!qX?xlF{aJV2{LP!17|$<
zyGMxGfXx#sOyiY+-3D)Jb3$~8&#O@h)z;N-64OjMSWnwEbs=*#xdr`9+)BHyR>NY1
zx90%N&=hP*br3&dtf>rVJ_l??2}~<U+$fepK5{cUfmdYK32yt9hA@)-kll`z$GUU(
z^zJS7Tee;twj>t6BjoxB)xIIfLpj#^UHzQ36BqPB=h-OlMdHjb7pAS~KJK7=2)85I
zJD#-IccLA}j1A$wQttd1MsinR(p!k_MZH(#fXMa*vxpcL8dQo<E4~6tBd;X|tS#Gj
z%cb~}u9Q)(ZxokMU)+~S)4d8HQ{0ybt*>{|Dpi}M?(&LE8SdS-YF`|bMc-M~0qqsk
z6l?Td7;=?GLC*5M(Cxu};!TH<{W5)Zk+=R;_R2$%fFZ*XkJ?RcHZh4-mRoh=Dm}_U
z*L~Kia*6-=wzUa7?mAM~ht5E0g2C+4cai~g2LxCetTff9)2KTPM0F&89&mmriecys
zfvFe|qm~z_6w9Wd?^C(c1)JKZ&HnJP%g3=<&)LNQG3qhEr(-7ay{q3uc(1HU%;0b@
z5STi{etBQoy8KDQpMGps=`8u;$8*A!D7*TfxU>mlS21y^>($5_9zwxNtvU4~sJIqg
zOKdg)wA7my;mNK?9d~%Ss5;QIubd@e=%Pd%Kq<k4BTUgsphQ57v@UL_9=w3h>P-PZ
zGj_q&^vPjow^e4MaNdIotw-<KhtTg33S7KuY(wjcXuKF%x(AWFs^`Zs*SF1RIo(>2
zHr!KHRH|@(7(HpjDtmih2-93F<-rwgCu(x%<l-`J_oD>wE%8XXtCrlI2Li~wm~i$F
z`>@E%FDa-nemIfP+y9X@Tl!(0ekT!o!&;SnI8o7=bnz<xaAGp?P`xioxit%yxe5U|
z=DtQMKb$7pO~8_tQYHvvCr`2^S1B_~KD?0-RtLn_twtcS!;omLe*OH5cJRYmfZQO8
zuTXXIlWQ%bjOigk52=ZaN#QgTpdp+ZSd+dP9h_f91HQ4h1{#y2r$^nPQ=9S{+);UF
z49#Io;vljQr;Vebwbi;mtO;Pjux0sj>Ism2TxkXZhx`}>JtX*v>2fzABg$;%EqlcB
zs&&N8TnbTZ$meq^S}tI+fnrBw9nPV?GaU#c;z%*jFR5tS(7$IhfyV?c1TVicV&7&D
zNsndfLn9M)C7M9rj?Hf_nz-QC^Q%fAx8>xTkNN#m%!ggeLle@59==L7Cb!+!qkf~i
zh*{L|aLW9J>mMSlT~fAeI+lf_TQS>;GMa&o9+gl6SGR4s8L59DxZkv`eOOyEZrP-u
zc3V(*{mR>V;MhChx)63JkQs(qw2jE4mS5@b3jf0$wkZHChzYwFgMkV@{ohjk5gdkd
zO;ciGD93E>6d`lK+I;~?0RncF0e=PSvRA|v)z6cDo3<xJRhO}{EO{c4QQMHe9hW{4
zXIpy6Eqfx}H*qYFLq28&4Pg|K%2gS848C&ooi*`JUGap-vwt`R)~I|Um_m2Kz~arG
zNc7V5Z@lVOMQiFywF<oBRT2Z20ROMwd(mp2NXtFjf5;9#q2+!qIZ5;7UKcpI>inT6
zZY9504-6jYaJ@Qtnmq;ek0;VzG1_M>0TWgC)@X`D-rDr@=WtchSwu1Cgxd8)nsk|8
zI(c_TpItG8KK7_I&$+I(GN4AYkx2|mh5~0vP#x!y$bQZB@A7IMb)nN%h|VDOvnhSM
zNWm`Z+h@b3^_`*iW|xHSFrvg6{d@+mP<C0!pts{PyID;~NDh|mN&slj4ceT(ahh4M
ztMU6woKW-8I|$96=dMNaRBm95Rk@&~_8gpiXjGr<w#6vt7I=*PQgIl@(5zaH@$}7Y
z!d8^R0gqpJZ>S$ARN2h{K_C?8Le>Q0jBjpsJK*u)^fg;o!d*E|*qy-L-Aru+1b-J~
zcfYGFRI7+>t)W1cEKQ|*g3$->hc1;+8a{d2o$+l8C7z9KCxE*R!hO}~C2SndE>(u9
zu)RR-a1Mc(eIyZC_6p6Hd_;8RY7aQU{751>g8#5Nm3>5ud4Ol*<UriYKcbDDHL_17
zor64S75e=|A8vC5k72`mrey)0Lk>2mN~s*yAF!9zN=4wFnvZDL@<6xAUrB8gb`CLT
z2Lt7o*{ACC-O=3MX@@={TK1C*nXlCglV03>HAP=|4@463k>VCG-cPib*2rs02bmJ-
zao;)xoC6xev0^26D1^F)%0X^yc0I~tIVZE$ffy^?5RVXKfUR{XFdp=4nXp7ID>JIR
zuzaJt+X{gRy+jrgi4J#la4}kk<#8*9>FBA)fAcZFq_U{MK>jnu(YuxuEZ(%sG4P5K
zI~&<(@F<4)6;)=_Vdc+#M7M0&--3MNA)`5rFOr7Ne55Z-j<2q;Yg|)@IX6aw`&&?x
zuE~sM)^&BU4<~|)8VMpEPS{e6%F)q5TmDGe?apyci+*#O{h8fRk#W~cxmlb6f}hjP
zH37Z(xA2i_jsARl{hw?-u=7!L?C+=`JGyNIY;IOOq6s#-MVP;+`PbX+o{IR(BwE2y
z9yJ9nW<@x8+fmizX6t^8r)({u>0Om;Kw)Vw>i4=)xm6^B>1S?BDiUd7q=Ja5R0%zU
zl&MhG{beh!NR&xrc}}p3v}947r&nWEQIQDs?HLB?D^{t3+5Y|eEfX-N!Y!axj3wj7
zKh+g!h>vtwjiBMUaoqQ<TGdDM8kZhXpSMd)GKI`hcK$;u!#mDttB>lgwgP1|sS5E$
zx!JY^X6|X|yVe>dlez$|JR&*f0q1x+Q(1gfxr6tvinPO|OExdbaMImsp63qhi7pPG
z=&*qbk$EF`S+6)0L)O`DBT?)X)<-_RU*)hZL>09A(wR3^>;d7O+50a>(baDtXcSVY
zVY^D}rKsq#U*A$@qOG<;yByf#yp6><tBmKLudyroT>UzZs-qnzOpd0jY6pAUzp{Ci
zI?dJTT7|Z|33P&n{JnHR<$4E*a(F$U{T-Sf0d3sxCplXRth=kbZ8^S^=fuF7Glds2
zqT314R%nRXx+!>Te>-ez0Z#^Ec015uc5pn)-nt$W+v?%j?g$FY&$+9O=sfOfVZ!f5
z>us(#rAiz+h<xv9$+4KXRDq^E@XoecM)K~~%(f$psXpOHkKK<RuvsH`!nZ)c$6KQ9
z31q`dW*<%TG@3hUB_GvM1<(6uR{GIIH+>_RH<hUj$Fr4xRHNL;lLg{mbf%&bO7>j9
zRB)KHqVl5;)>qQr$yAQ5mQ__H;OOb+t@@*hpuY6H)u{IBcZePK(L|{%B{rXNJNVH=
zscGMS%j#5{w>A_#0@4|LIn?VLSvC)3`u5i|Yl)^Ab==}8BiYon2}k4vJlr!Moq_Xs
z1Pbg@+2gPJwJSQ#ZO_#l32+)6cl8?^h>l+vwjotQyL{Y60w8jav=;KJlU$QoL4C{$
z=Rgb%kEnMHKR*{E3rkfGA3Ck>vr#z*a7mf{@Y=q<X4G_VKkv~W)iw1&a26;kptiFS
zT?weI&f{T879BVTcW+mt11E=VUaet7?8@oa1c0{RvIP}74<CA6pT{V%s0qvs+fuae
zvJii)hP;-ekwFhhz7h#ENw{zc0(#bVGurqn0gTr4O*~SR=~X0q6`Pd+4N~qI+dF<I
zx;BUqX(O^dBpVJy$9RIj8<m-zO|{z!UyoU)s^xL!Vp}n(tYlH-c3{Y}>ZI+&K%m>o
z8oVE=)$h2FV^^JiqsR7&rAGbj#XctH&*h`&(|=6V(&54M+zO@iV~NCgM`3KrlyJ6}
zo7Tq?y>YYPoQi5#srv;Vla=^E)<jgTRB6bbcj%9a(r}uYT<cZ|EdB*JiK;)A$b0o0
z`0DtWm``?KlmlHcs#YaA!8mXasw7HyRu{h?cj4M0HPuT<JFC|>9)C_+Q}nbY@n$Wm
zoqZu^tty27U;@2FA%$h>EMj&BNpx2<;<RDis_v#yJ*u7<wgFX_nSl+d203SXZRBIR
z`4MCBfssQ#kZ8l1fS|NBbhg-3wB%Moc2OB;{T!Os$fd*5B@~xFCjNlfoE{f%rZ}+f
zb&lefBh6}X#R;2@&b^M(1akT4_V5|JbLiWh=GxV0-JK%ad|-%R<(kTH`V!70AoY%|
zeh+WI9}^iP$uC-8EXD$G(=6_IBeqnm!CB1f^T*O2`o^=i5?BnK&)W^PJTi7ZhYD5n
z)n)JKb=z71;7p#vNk`!II{p-+brjR|^1N(!G%gv;pGP3sP#a^e%XK%#gj;CA^|m$z
z=)fCawtIm}8}zvqYme64-7}=It%Cdi=yBT-D)>NgEr|QdxPW)wU48bRL9C%Nu6hU)
zW*^r{^}0X)lzd!Ed>NaSkEaR5+BrEhDf@V$DG#t|<tl}1UfZWq;FA2SRj5?fz!X;s
zuq)G!Sr)CpccSB~Qs{JW&r-%Pj5>0QsD1`ND79+j=<!ZF7^t{o1@?tHCAozp`TQXj
z%utuCSN#_Eqtf(o?H(_20^vVfg<LtIRi6`_(rr<Ohjq5d&W~%YxX7Y_9E*dg>*HxL
z{85c{2Woj_?mfcUA92v~4dghx2#0P%g8s&zblZrE`Su0#7X#pmhE`C!NhRr9HuZ7s
zY%WYs$}tzCt;0ONOb3cczCrO)G}hr}AyAbHPQqB(<*0Ou{b2Uvaqpa*K+rK@R{}`G
zNFUeBxj+Tq%6JlpPUD6iy&S~X|77#ZcMTuTEP;g)KysnJF%5P-5MdWZH^>&F0gTPL
zeJw=;$$j*)vXTkelC4D6&=_|1Hv&i_Ck?t%H;Y&f_`}ayAST7$<fi;~{6w4!Csn^^
zt_wXb^#5pggdWDxwxM}z>a)8sEGKh=wyAHNzGG{tZ>~{j{|YHak!@u=Figg3J0I6R
zPmAs|iARw8F<Eeik}mA3ysM|6qNoIX7CXFEY8LRcpA7v<Dn)59v%OYYnJDcP<Xcu5
z^N%;>M3Uvo%X;?PzRI)!Tbrz+G7;eFHJrp&sz6h&&$0lDa_3pAio{{R6jiGN8c4h>
zM@DN7ylN{&8C(;%5gx46{n|;G?IejStuBhl@9CjRZ3TMVv3m6@ZS@}MvnkMh^%ZLo
zy2p=Ot6I-^K(#h~qh<G3)~Uj$;U2o8$PcFf&bk9Zj%lP9%HYO0W&=^^3_AKlsyjT0
zdk+2F=N_2yG0whH+z(J8s;quvz@}8?6v~hn<L6_m->~UQ%_~|A;Y#sR^q4sHOv#L@
zaWvWG;_MIx+#O-qtnyahJY!b`l55Qcn+x!6Cwet{bH-7akI^F(OcK1S9JID%3u^SJ
zZzRxEK+D10EC#erKc7@8EF6~u(V<>f8(vX4q@(EhMh+YjM0hhGpyFU_fdYXtMtMiQ
z9bH19Mt6esN-+S8AOaj&!`+EKz$1>>M${l|h;;4l3VH|n1-7YnlPddP+C9}{dvXc9
zvQEf2b6b_#@uy@v`d*)90o~U(vU5}u165)=x!&SlBs=`zA%xo}(!8ZtnDq2$C7(!S
zk^SI5{1+?zM53uC)_B<`L~B?|<4{|_yCvp673YeE%;FP?KAjDobRM!YK(*o%Is1f`
zd>Sz>S)b6+z-<;J)3H+>C0@<@25&W=5K$47fK#nDn#YX;(!qe+#&%v88OgoRA!T@g
zRIC4lHUaiimk6M;@!$t%kT{E4RCV?1F1v1xE}@~;rk2x1_D$=I#(0;sE)^OZ!n0Jg
z%R)m}BT(m2he(t*pepQ&s7MLe>fiS|A&#gPYSV1Bc_n(&8B+K}+Lv;tY)Xw}Jzuno
z0t83OxfSEVvlL)2Me~QAeBlL~i3EmU-?q!2(9%QX<{qBXZDdw(hCQh6mFSN@@6H9@
zh|%41^3{ODdWXuQz24+*_nMFbf-OV?gd-nEG@0%a*M<EQiW*u2?&S%Wb_m_0H@rD7
z$DF_9FR@}ZR-$x6=wsec=~v-VZU$~iV7zQ=YT@vR-Hr}$g|l^izdeI2=}v%84}QLu
zzhfIQ<tbX6>+0VXl=ep347{*AzZWg_4q&MWI~X13p%VkZZPoZ}kL^S;F4@jB-4EC=
z_qz1~bGs{u`w79g7a(#|L)c*0Cv|KwT!@I^2|t-g0TWup&IIE7WLoIXEg;$`Q)V{d
zl0!rqpG?$<H3IrlpG?%45z?hXAmwo$Hm35E+J2L)J)OFlDz<xoovHxhM)uXJIe;WQ
zIVY-7>lt=)TwH3^#4uX>2S1qzK1JQOI{iMu^dHjiL?_`2Rj&kgJq&J>n&~^wcZ<I1
z(w|#v07Y4%O|7%tEc!HAXLPAy(7HaE2%x&&x>c8(AC#p6V*`CQ5F?XwZ!M!X6o{kc
z%T_-kM4H+7@eU9u9vd#;;#bgKcU^(0m<fcvzU|mWwR!BQO$YEnZvX6(mp-Y@!$W=`
z@yARh5N({a%K~`X{SeHmrAC-J9*zTg0sGk1Z@?A?5O<=APy;*5=)K4DM$yRTRhzNn
zmU%6D-HNnqL47NXxrbZ39&NwWYm3ooBwt)EmZCxS2TY4j+H&N&RXitIG>=jdnwKo>
z8!;>l*o57T=3I4bElNHxF}PhE4cX|d3xTt%*X&N@@Xj?sk2WHCW(>vw@>0RmFa(={
zOM42Oa4*W8%niA{_tq!1N4e{2_mJWG5!=xjcZGz^)^;M@0eQJR;C_HO&BF~|es&{C
z<SBm<*^B0I00=4Cr_x&bY`2wsO6Lp;hAB`9+kWY%5=pgq;O(+c1qbObj^&@y5e5CY
zUxj^<_xcz<qbpP?8(sabRenk*(-*jipR`OlNw>Nxa@_78UHeao?zxwy9@YfFm)S3@
z_ETwvIj1u2KC$hfatWQuF<<v74N#M!LBLfJJ|a+fg?oFwYG>)G&<zfnBJ(9|gjzn8
z_KD*RVyk{Wog?mSn+B!&IqTHd=U%X`sP=k}6D_`7v9L!SNud!iXc54Mmzr%T#)#qc
z(``iMh<xcxP5x8bg{0Q+E3yzU4|7MGis48ayma`Yp!T|sR8;KC;S)W%yt@>I-D$MH
z@XW->{_&(+8(tRJF+|fgD=;XU*%kf#_TR9x%;_7!FHgy;43+WZWAiZ$NE%PtwdhZ>
znTCIiaph*>yETT*f_?!eZ!wB?CLKas3NUD*@S3b*V=zH00b&9!!$W)6jp8yG<2J-r
zbyL{*#e}=znxMK<F}F9r9sL|29u1eVXcGPC^vo5bc@l<bLoIeRciLV3Oc-@U9QuZ)
za6F~&md?&L+ls0v9@w_3;&sz@^!X|d4DYKyc$w#dv8xG5Hxk%%)N}TPio0R<2Xo8P
z+UAfFhf$IxqCx_L98q~$(na7<Da%A;s6m~z^2mIh9SJVL_Niial0Bm$3U;I@Vp*cH
z*TXqSCfe-B?JB*m%CtXw&_kCRShY$MDA^ZcO<;2tY-;s=l$T9As8U&TvYM>R5}kah
z`YAgUEw^|$tolH)k)6f~uPJKr8pY+PMYXd0fVB$rq%-Di`nlszF<r^h?(^J>))k%e
z7~b7cual>)Xf3Ha4LrcySN4!f(GMF@<qOWH=cC4v)s=B8WZE3w@pniN7L5<bP-0Fl
z3WWm~ZCc>jW<4(1C4s*Hxz9u|&k+sNN%eA!FybkHN}LshBdvBNGLUt+hlaW6Q_Juv
zyIQQ1Jl5O1+R7z|Dt0aAi0zs4kF~iF<0ymY>q<cc9`k5XDVgTmQsCy3=Kf`um5YWO
z+UT~T!8pQ+g8@qvMFkcOC!Uan9=9&+#sY(O*s~OK=yqU`!fx2QilV=S)w2es`_P}*
zMnG(nIMD7!CtX;^8MhgD<Jumn?7iaP3?kj>wXK+^2BdGc9SQ6w2W&^<`Nq?BKYoVk
z#^70r&g98{FYsm0g5J64*ppi3T=!&xR`R4s37xGzD}6H298!S4w6Z64CLJAh*Rkcw
zVJToa7C9pxEx;-Q!217SAS)x0cM)2q#IdBsSfxaIudwPT6RopFx@PpU)jTO$#Br*7
zL8w(W?`1yE$UoGeuOd`8>z+(RHpXFlD0(AjhnzC?LNjZGXKQ*gogIg{X}hL8WY<#c
z#Q3DOMw57Zb<L`_=+p6>zw&hk0;q5j;t5yV(XGS2jdUyV{Y$tSvmPbf69A{0Fv~7L
zb^0C7p6Vk398osrsrkUM7ek}MlZmj~+HESDWUv$QT#T+I0i?%Toem%bWd?~$f&Ro9
zZt6oeqfAbKWsD>*D}}2K@7iosbbmL<8C9_%g#L3%HYM9t6n~>EaY)g%n2$zIOcs0%
zTodB#Xi(mh;)0Sffi~>x>H)Xwf}1kgV)OvXtzxS!MV)~ZE(QWVZ_8?7EHr0V)B>k4
z4`%v?eTD=d42IunAd<D15V&AW$znh~=+%03Acx;xXiyW<%bU_hvBpU5q}`3dX4g)F
zf=y+@3ZR&CFTmkP$hMxmmCVwdzn*3vlePMGpyRe1kQL$Vi(O4_#2MxL(OSPdw%LsY
z(FeX`dyz<wO157#hUNvkS;_uHec2JG#QnONy*6xR`^5vHamm5tM|yb_Je(W0efmZ;
zh^w6Vz3XU~RqhvEA8}4Fi;S~8(pK#k57|D6B$1u4ItnJ+p&Av;`H5W{8N7tSOYRqi
zvSYcJxh|3@TRF>dXunniha32%A+6t^R(3<bH3b|d?q94$4W$0I>f0YetWEu$M6IS%
zpUADOEAaN<e%<<pKOh3Dz9Bkdj;Z(vMTXQ;N75f1iGI4mpueBy17)Vf9~}!Z^yr3i
z>zXQ#F?#v_=!AMlVBL-hg-xrl?-WSVFRAr*=UQge$^$n@mt)?Jp0L?KAgPHv<tb~N
zr~A46w_J}iF++%VV22j%YV@@OQ_uPME?^FFP1QWm-a`D`j(?}?0Ua_zU5sbxNP8Gl
zzx*tz?CJ{YlFL!gLk6xyzw;Q^aDj_jXqu#;88?fAgs7ZhSyM&M2B<9N81XNWsM3(G
z7nd(YfUUF&J4M(=z{OY}ubO$g8|m>l$v0!X;n9d(=P!KsRNHw6+X}cXDdN<&24MA@
z#0=gaOj<6fviIBl;@Vo~abZ^=uoM3W+l%kN;!<o+i4@l7SgIvYg<J#Cx=XT<mOLdQ
zM75QLUKWLW`x&hA$bzruoZc5H^a0t_E1nWXtp5!Yh!N*gX}!PJWZ6@RjM*BnZi&`7
znZp%PwaWGLK>d`6=0Ox2WRg$4vo4A@LkFXs*TrJgJ*A`bpK#IO%jQsIvpL~o-Rc8<
zWQ%!g3QRn5XG@fGC!-uytxw%bPBrlVILYgHf6gV^ZBfQ`A-UFeMt?_Az=5tvVgJRF
z>J}QVHQ@owtVg-+((W2iZX*wLeJ<!8dMfQ^yb#&QQ<_4yyc86`@_}(P0u3u@bp9P7
z%A$b=cNW2LUyL-zb2`VS1@d$IzigLO=}X-<6CG$Iq6z!t<p7Pz5*`a80lv^;Kx@Yb
zjSDj71Z|9`*Q)^|)DY)2hL98vK#B_hVKi3LF@!Gn8OFC<3jc6jwYgrMb$=g=g4^5I
zZ%gWU;uPTe(G?Lw+y~f>Rs_1my&mFXSdF&BeAvxcLWH55Q0RxX=*c;Ec79tejOPB-
z*5mi}zhfcX(eJq9whcj>a3RWFeP_KRJPJg4wAN6|*C4*sVOvq;7*c`lSfe0S+(&n!
z783apyB{NRj*|)2Gg2of*_mxGQh1_*uB=MO&g;V?$>~i=mDT}{{?JC@j4G{4Yh?c$
zHC9%Y2-aN)VBc9Kw`d-VfsM{>!F@a57xj}`pvyojR6jd<#47bU%1d%qmB{jFgH=@}
zvc%MdGp{OW*G*9gWgA)UL#j6L<n1|~V+SKGufn_z2sRvz9jemCF;Q?W`1)wS4bh1{
z8}MAvqEGl*YYj*eqI2X$bsiJwerFLSL=B+SC6v(ifYvk|1Nfinw;pw<`tvrRpR*1Z
zLk~rw)m3Ll0+m-MEFZt)(CmyAqMr?TH=9!HNoMfRb}?FS!#f7ZOb2SXltgB5N#KTa
zeKzv|0{S5V8rB2qQUVMa!z)p}5ta0u&8hlndU93OFRXqC=2A^2AJWvhrqTIy{MmwP
z_2O#hx@vJwpbD&SuIL<bfpp$ok#fofK5b|S+Dar|ch$ffRcTvwyYJBnSAS#u+>F`q
zSg+}3r}gA~uk6Kc3xU<YE7*GUy0Pt5yAyqbiKUqh73x5Z^lr4srDf1IBY|TH7)lk&
z8VCs;kov%2JJ9z45sctYm3CzJF@fi2JXf*>-;eGN`^8RRUTOY<?FACP_aZrYI;}&-
z$1jW}PseHZw{BPVv`#yA+Y}d-^G${EdFot4dE~p_9T)=0Xy4NstNyWfLLgT>ok#~?
zP?xRp=|oV^zF}E31OfM2m5LBW5bAnb2b;^}v4{0js^)1C+zG|fsEyL&Xj>kPEZ24f
z<2u3r%opvD;Ip$Z5cNXz-&`G<0tOxitmSEKE>H-|7p=$WBwGutMVml&oU_iSwdr_#
zIt-@%#k!)R2P<-G<L;>FA~>AzsN%~!aYtXd*J3XHblRyt|D|dhiSlq=tcJXxC5G-T
z?qP*uE~n&>O+`nc7c?U6qLTiY?fQGu(Veq7E_Z^%r1|w1CNt1(3NFGfYc%jH;bJFW
z&a+YdH2momeUJODG^cL_%RI&j?@1Wi`GCl49#*?toUaL+=Hy5ZyG}KL#>rnKXOzo!
zJ+N=qrE)O<99vz%#|~%XQgLV%IO5y392gCP-HIx^#KiZf8?ov*oH%~{9(Yrzz43gv
zt*IV#$Ke(*PQ#0?2LkJV!=ANhca%k(rjxdz*})$flW;fcv~VHVjBhvfzo%~wGPFrz
zVqQ<;hV6h1BgW=nJ4#NXOya2B7jSqaR8n>WpvSh_(>IDFoNmu(eV;*QM1M=3Nff#M
zb^gs;=`*6pqr9_P*)zJ>p<`7ZpT|0EpNh!aQ$+yd#AD%DWmLzmbI3A5#zRd~EPf^t
zCQlJ#Rvopt_u`(zYE&@8e-I2st(wZAt#L5G4e(}abx|ts{!HwUz-7buYpH)GkrbkP
z-r@Z@AoBY8ZYlQGO#u*E3P`d*e9PLD0$Cr4PE}jI#vQLKK)`2PcQi~CTSlTMM&dNK
z(<+2vH!t1#?)BI7k*LtwaJaSE@=E#M!eLoZm)hMn)uy6<7%CB+LcMUySLbqV`kCT>
z=*I9a0lyT*aVCLsaU^hjh4?ZOulR*#Hk!wLh)P$Kb#BOy-Z^2}QXpdBRb{j#9(g_h
z9LAu>u04~so)d6VVqg{`x4{?f^%xotnz|Sn$y9ZzI8Kl$EMp8jk9Ga16|6jy_Ts7o
zb|dPX9D%vL`CxkdHg0Rm^Vf$_yB!GxdFbk^RkjffIb#Ys+H520a}csO*j<$(s5XlF
z>P1Fmkx{O8<91J(>kI3%wxz*#Arw!sZ`t-UItTp5=nwQPm7M^MPMZ5fy06}yMC+b?
zdRMUAL#6EroeK=MJ)0Ie<cEmlS<%_y99$GU<FjdzuCE}4a5-X`ie^o4E#=ROJHtef
zTlBtXMT2aZSeq$}&Br}NRz_BfTmEIr0=lb-R|R}uW<-Ir>Sq)2PEK5~nrC&CNFYMj
zYM)I@3)fmc`M&lK1_s#8vJ=+{x78Kk9uk~9H-viiNLBDq5S_)8#9CA;Lk+K0KOY_&
zJ8f+N_T}Dwj?Ye^m?2wsXTH`IBQ)${TX;tmW`vd@l%6Pz?)11H_5*>_KVow|6ctmf
zIC#iNy3mz7bUp^@C=aYwP_-;~K4(+P=uJL!F2<mA@XVLo+)qczRyOcU0TDe8-_~XV
zVtPovta9*9=<rAX@C3{bbtP))$=l|l2S?(hzZy`m;~%j3NDSmIKyU)-QNwH@fS{1I
z`c2&LC@YU4nk`0a?A+kEq#^Hv;+F-C8yQov5~V<##dIUU4Z6!SZfMUYGJN`gt?BE_
z&)V%k1rY|$P@^_77udBsF(%C{gbjTUdrYRg`cA@y&8TrCY0lq^PD63YN##}yGiD76
z@wu&bJ>rBq@cIx3SD*kJ!Q-|721)NJmHZkmnN<hB=47Oxu_e_ap1~16bV{oe1tRax
zTUm9SboiKsAj_4NxoCM`b)vD}^B#?&LWyi5$KAU!z#Z#7MOIb>?sI5W0l_af)zxVv
z329WLul}=EtJ1_XcGlrwG=Dgm<Xc^}HUM{bhC3}gq_o?@!Wc|geH2IB?NW;-WoJ(K
zZUNIGB%LFJX;k6}N-SK?0{xDDW~Q@RXLduAbp`J14iG}Ouq3E?Pj%W3(DsS#eLyX9
zu%<`JA%RF%GAha`%Zso_P0Fh+hL;VfAdt`gkxl7mRHs<l7prw}6>|MUqsdr)I>xP;
zeqD-go|t0eLbW82hjth3vdR^P@!&Ws?D^fF39hJED@JQ5ZUt88dAMye++qO1>JCS!
zLZ47Xb}fESDr5@+4)H-%`sTF60Z0)!ln0hn!K2x5NfV~5$>h{`MdkV?;VN##D095u
zd@w{}_-!rnaP4?IKEMArwyxS7$%y*(jk53s+lbbWBH6jC*0G^s^=vBR?XTix@?MNt
zJFfQFthWMe_2+CmR*5UCCWcTTR{!3OdhSOxl)c@%$F4AiRf7pDtbW7xVtBZkp<aAI
zl;!u5(AO(DpmPItP<(X5yp|qFYo0*IW%_#KqLm-eY1KTH>$iOe((0?O`uA3$Ph@yj
zc_2|}pAN$8K%#@pT{QTEd6jZ~+xKHDayfL_z=o{mK%!^-aA3!&Zyp0-4F$dnMy5_b
zXPvx79E$3L{)h(f`p6sP%x)6Sg9sR-V=XKncfMt<!Xax;g0VKiZo+j_=K*aY{xF1`
zs4E7^f#a3i)sVWQg;U&GdQ>qxj<=71i24hg#gH-xmS-c80aJp_Y(6l7ta)*Rk1WJ+
zo^j9cQ!0zR^t*`3E(R{imllE#!Q_p=E=3LkJNNQ9BN*+?4V}$)IRH-N&i2}D6g<rJ
zhJ|<~@?i9EE*y#Vt6XibDv?eun^zY*ztGlf*P@G@8w*hZ<);z59>4GZ_qM2UezwP!
zRI#s@fa=Q!itH>8L+SdyqWX<U6u4)pK1L6AQ=cBBaxDf6hZ)>xZU^@84=hJp7xeWv
zai4KVb$Z94s2H^W<cyZxjp=!L43dh)!BY&vssXzf9cX|eY%7{NlN|JKs}GGY*^U6b
z{hy)3`%xh&-R;I&=;H)uH*`t$p0Mb^D#D&ilrVrawd6S+P;L05xL~D9aTi8e%gUb9
z`f5c5i{rUmwf^|kX4|K7?n@Ru_loBd(ZJc^vxRi!bBXFk)3{`+l{ps}@Xny7i7294
zIj<gVvYO`-Az{UQ!D<6J7ip-O<j8W_>IBx00X%pFEVh0OZLMCFdO3@lgiXdRl&L`9
zQPnsmFv@NXY<lXgO>Or2gJ0)!k{P8K$~Y}am(Q;FvE74n6eAOKs}}m2`0Os?z>dY8
zXdnRkW7^Pj+MS?|<j2NFo=dw^&Eu9AJ{~M-*b32c{LI=^w1>ClqWee0_0fiFqakOS
zj?S<t;t)=iJwCY2M7a@vD8jLavbZZCQj9uWe2{4<BL!@m6HWy}*hUc~c4urp2JocY
zYco&RBJ-SkyI6?z(m&?Um)E1!!dP+!vnb##ZsD>NP++y;k|D8JGHH?6O29nh9Q}>w
zw7a_#1hz6)kh&SoG$Iq7u(c@N>84`0g>_S#tw$lZD25$;PBeTR-D=q(Z7AQhSU9Ec
zM$cW_XtUT~ml)j>z&_5d_!-+m{CY&9Z<zgO=Q&C92F~N?xhPW<z^n=3`fe0HJPGf(
z7a0S@ma*s4%Eo6Te&F<-jiPj4`g~ds&5c&}e4>`a=;k8IcwQV*?`hlje4<P@qJnR!
zcwV%~lYqM}seE3Y9nYPFG4|sx6PWDCBt|n;QJSv1Bj{>1jD!~Rm_XuG?v&N4HSRxl
z@Od44lS#!N@w{43;=S+MAr-8C_S04$?Qq#e|4OB@mtYF;QKeFN2C&uuH;4w0W7nou
zhfg9g?^Llt43@i8%-?TafNtF?*5?M0J?aB&lm{TNf#+|1n5W(>6yK5`K)<Udy`!N(
z3gsS5JR{)>r<RYzfGeLBVpyDOVjG-NYm+Y6xhPDW+epJXt$HrxK5Lg$5Z3`;vl+qQ
zN{?bbp7Ed8!g%Ll2ZD=HIp%UlyYhV63X-;5tLu@>37Wq{*;RdKIFqvZKou7JD|Ri$
z-}S1NZ9!R+qlo$Jy2dYSJz<MlMY&#EQu&D#{J@q47Dn-ct?1h?;&!9h#2ly3O(ocj
zVcXUMXF^nyx~)2c?vHO>ZJar6cT@p$FeZEBd2N|V76~_<^l0y@1X|3j<;}oz^(%HS
zpu<(MQEtWJK%Rj`G1^!$2#u%{?L;0oSR$+OFs9MVlJf}XFLwhJma0tKUUam;OTN`4
zn&9S@gcuw|B{hloP(&mSMU4muT`AXUE2~LV$8DP~mMej02e8?vN*OU0twI^u`u~mo
ze`Sr1N^IOv|HuNitCLmLq$xh;f#D{`@R3(tBQkC~W;H5zegJigTH(M(?ov=Y7|@%2
zRn)0mV+%ob57ng2;WN)#y?#Xr7#2~*VGkIF7Bz`yMKmq&&1bD@^4TvPur?uz5{Y{#
z>x@>Co3SpHa;v;<fyRwCazm9ln>5D<l#%6;kqjw=%NYCgNKNsQh&hDIGL(KX{;Uwa
zezx1D^!cyfdeJV%z`NWGf}9p~E|xg>sL{p`JtEdhvYwe3WLku!y(};uMQAoUygrw9
z0AjKSn!PaJbJ0ffJB?OxF$;!mKC;}*oke*~S;zcsYeAjESt*2kT}_>4#1@N9{g-rj
zFWXYU8)ZXZj_(xPC59FUtGqP}bNC5Yk-HiFAbbJggw~Wo%<9h>w*#d{76QVzbwOPH
z4jcC!VZklSHezr_(Yrn=j^wU#vo<`z#V2O1_i99m>+^1=0>|N3m8}58L~jfJ1ys{)
zC$Mj8LajJZ<UyL}`dv|vU*S3Z{EE3JG|%L8kF^&x`&fa)=`C8x3yGk(D5tZ(^o2xg
z*$L>WOy9UG;zWBvS1=A~OsHQ-BqZmqR-t03Vj<a34o)@Ui!2LZgRmb}FC=1UY`1EG
z>gAd}Y&8KFa^UjyLZUaMI(9ICa=*8_0Mm?k<d7O*8}ygedLhqQKH}8Wr1}>)6I=8-
z$QIYCLMWUhrR_E)5qhBMfO?gH<DUUxVIcF03sbw5fw&aH(l?5F+`5DT8u>|`SUJgy
z#z;UOo=9hXUMS%5&I*yk^RfF8n~L%K7W4d9c2OV((b2H!zyRi8z%E5C-XeSb*(;JL
zw%}6HE=Mw{t|E~al+5+juV{|5{$ZSR!R^fn6TG?ERTaaZJ?;qJSqdD%Z`rlzncJ$t
zA}W;~<sLd^*9D@X@fBMXj&N)2151Jjhf2h;B7$E}^6!=6)Vk}N8<CeJZZL)cgAEah
ze6*CC!nK0kR!axo=(cqs*aR!-?}BzmwIF&p(Tx|xJ8@HUD`1w}-GFljoi)y_NHr(b
z$9pk6Em$#ZMb0svx=-7-+RNVlKW#^!H~$)q#rSpqi?&<r7VaR?wcL9_yPaP$#*-G4
z*^8nF9P=PmE_qRx23H15>`>{8!qi2{g(R#hi(h<u<R_MH`HMPgh|0O|MUh&Ap9dB3
zdm}z|^!X|`+3ZD~f_Mmmb+jtApPkF)cm}DCOpi9oqOK7*nqhs_25#)rt~!Fk{)?jb
z<Ah16(>Le90Xw9|hA$APHlX*Pcam$0lv@@dlUFMEzSe-w^bCk!6e_;+xE@7k)Xl??
zAn1sG=0;DA{eNV=Pf*=QmgjpGT4<q((2J*q7FzsTO^xw1Wm9ErQ#Q6ifP^HFuHj0;
zHn#b5kuKujEdc_l@C5Cqqv_YYE_IJyH$BaJt?Hp;s5jIERYg@()s%-Au!98)9H9jZ
zEr<>+SZJYz7FuZWJ}1BDsygNoEPlV6nI})?FHh#l^Vho6cl+flIqQyUaP>}aF&sZ4
z*++`1h$aRlSoDXNH>;7WR4Hbw>r)$#W`3Q85t~x8oU^3U>KNUmQ=ICJj@<FIG&UPZ
zof{w+$y|VgE18eS*Iu`Ks?yxve!1Q53k7FISq&&07WBh70(wLx98Zg?PQq_2I3MT*
ze|kT}qOTOrX;}znm{Hn_CZI(@^;gw3Hbu6kN@y#H%dMZ=kh^ZgJfuX<Zp6Czn%1a0
zScJXjwyKZe*|ih>a#6#MG5S4&Sbr}jegFrYTlVlZ-6c{`Hop9R)EwUadwUeqLEPR0
z{S3SN$M!f7VbZ^HhaWo(hz&Q~oZv}(6XIa_4*N8)#JU=@r^Gir6I^Uo?785k<RdSl
z>vr5G?RAj{nM?=*_j)25l62eA*TvV3AeO|g<8`g;NsP&kzn%t?4qsPLoOoRX>h=OD
zM#V`rC0qr2nw?Vh46ITl`YZtBT~oqdINo}0kHRff1}JByhdnM-Rs6vCfBr4vA*+5}
z#II1~?nzb?5PA^}U$@#o^t-qj*TvA<dJy#03k?Wd>a!+QbT<WS2@rU0-N1P9b?pXc
z2X45jYc8O`dUL#~`=kq2oZ`+X#zphGbp;@ff0k-@j2uQS-HHmTibpx*T8ygrfv|Rm
z^yZ%ToFt`y+1zwKZ{tD~gUnyq)a&WgZ_C@XkW2x}*>`551)Mh700VSs0(S$O6BbR(
zKMWJxqrUM3Q8(}DafHMZ_w`8VOIwKXvtfE<^hK3iX>Ya%>IMD)N?%eh*x-7k-sM1m
z(BQ+iqN1_!ttFVg)hL3Ag5(6vW!>wlg&f?$ZD?k5el2l%9-cB9_{44Yi|v@o9we8p
zTG<Jh#8*Jb`1&_)MxR}A4{`!EF-n%sp62R0ro#f=gab2YgRhzWKu|~XBfVwpI@t6b
zs4HGv=!s>Ax?<UI-XF%?xQh&0;3ul<H*OR}tM)X|ESq^h(-7c{2JCrEHX`a5LfDC#
z6K(a|S(^XYtXjK_G??taguJR25wbg@96MT_NR@bW2pNu5i|Sa<%6o9UI+3V{G_n(_
znHBuRIw>H2NMKibD%v50Mqx-C&n)`zIBKI-QLTgX-FL0BS{oJ-sFC_}@~MOD5~qB%
zI>?VEFEv$UXOwY%tza}XHd|e_Xo&-oyxLYT5CbHIvL=BbguM$%q5xZ9YpoWE>;B_X
z-&~;B<{#hkfky*zvpVCsou$<kgC=MIQb@h9C$mEp0t>FzZo^R&zRbr^QX>yVF>Xx0
zt(M}|4K$(u@c=u5YpYEKCYO>FcUryo6}Jc|qGuFx12!8?NXyt`b1~r@xJ&|k)qMQW
zNTyQXjqIM9qBq*G`_Vxk;oTv{7+wR05{prS!h^l#K>$fP(0xByiat{)w&iHo#PPJE
z=`0HX*hN;O^mN6KY)#*a`1`gV<AyDBYln?^ZNP#Uf16E#kvKbv4!5Jn-aB7^X*;S-
z!U(^!+HR~J?5k6j!aa4#GA7%t%90e$t@rkY!kH`Q?2$mis6fC60RsbOyf|;z<LED)
zeTRV^S|gv1Cs7yf71;?nVqQ+VTQULcS*+`=S$i&ASS7?A4wx6AcC+tk!JlyhYDneC
z_q2F@K{3Nezn4hk>#sNtzn3U6&$S5+KvWy+N14fSSZ+1G^-H#zlWLV=L47Jp*rkTq
zvom$)4kH7u`m7MptO2#ZRR%8Y;nGkQZJS0hsK<_2uSxy`fz*62oeP()SgpEiBk)n(
z_tJ)diz}LHJ@a``Q-4o82kL#7E0a@DZN7N5%UV^BIMpQW3=}Ylb|Wpze=luD?fnR~
zi0B>Iwzv;cmzrhmsQJ3%xsCI@5S7lkc~MdQY76(P^pN@;M*~z+BjOknKVOZoe1m$X
zG*DDEY<q(?9n*lvgWrf}0$mT?IvYhPcs9mXDq@xW+U5hL7JRPm#pu~x+}{^#D`C+y
zMgRmo5{=3hg#-!SIl&%$FJuWYW%8KARCP(yL6CBU&9?kKZQcbekC4s!`dv}49x=yO
zh1}L}-A)X-WX#JD2c)gXc*jU2WgCH>XGVo&S8Y=`WkkQ;jt`O}1F=U~*@+RK!FGxq
z1iR|*!YHb{y{O8W3&+Y;Pf))85?3{K=htll4#yx$WA4R&5c71cNSsuaAnN`xaf%KD
zVS=$;wkOeVu7A{b*QbFpw)Z%{q7Z!+m5Hi-!=6WFK7{9VdLgj#S2z2<m`_h!Wk<fR
zb$NBblib6W9gWg<_n|lzFO42|@c8!=DSm-9|B3khoF{TR`F&9=CWbJZcIx|S#j>4w
z*ICp=964zf@scZIyjdx5S<_{!ipSI6wd(Kd9ExCLPG)L^jt9QB_WOyvJXnv_sTG23
ztJlw2(?qW|sXm6&to9ZmgM$pRWo!L@qSylV!}>jA^I)Om1!-%8vrUk^wRL?zZB4Ak
z6s^j-zn@lQqjQ#?Hw4u*)D{R6ZUVNp1Z5Wknae2BhGN9NyICm;8d1QH$1@QjY$}j(
zt`xJ7n2fEzw3%q_vP7FzEp$?T*Pc_YYh2&vRg2x4^?xtm;tFW@_3VOWThR0QOJ^_J
zq97s8_mkj(nmZFrehR(79l*pc2VQQNb=Ot`89aWFO9~vUF5(kClZX%TNg#jevTf++
zj1U^OskU5qKxI4D3n4oSl<lZV<>~L(ZXosxbh%&Jo~EBzzF#7S_{tteI~ZzT3j2Yb
zbJ!%N&>PEU2+s8&R&^2{4fEw6t1UUOO1_013LkPGLrPC#U>#c<XxHMU2O_p-Lam_$
zBN?5iW&Im_A>6Wt<k5{)$Qy~Ka392h_{bYN(tJ~JfBK_uB*Gc!AGBktmth>n8TLk6
z%w*!j_vH;SoxPwmU?-zm(ZiaaQuFLK!bm-SN}Qw$bs{k{=Uk-#z{=zJXH^2tJ$k^Z
z<C$PVRuj({(Oa#aFZ)Sdr{`?zCxi_NRIn3fVw(b$My~tTqW;JFi`M#v2o<3SQf3!&
z<lfLefJq8xt>uMavR84|rZWI_IDgmmM%s4VrC06)-GM-!0T+Y-pWQ`KMfEw3$<ojp
z-~4ZW=h(aFCXD|ksbneo(yw3IxcW-gpf>eJqWa4vn~qL~$Q?E#46?o7#5I35z8V?L
zl1J{GU|{nMzVk6B=WdN68o#GE$me%lA?~Z9oAE(sQS`fHyS7;F5bK@?0iRWmq!FX5
zD7%f_%W{Ap?qQ#_6(RiIC$_3OsCl@=1;TzuLLJ3buKY26Xd9aP6s#PI+>EMY>H5!l
z-+n_5S9|{@g&$r0_HPrnx%0+1$K7Z3Prv<}qz_;_;P<a=SKV^^{f+Gjv>V5{|J}p*
ziszl#*;l=ZVJvhW37b1$dJv=O<p{x5CdPSYbjl8+1m!-A%9AMRU_X4SXAEq}3=@42
zs1y;3rfuMMx^-hk9IGq|C;kt!8gU8kcEE@7>jO8q8WG?yNoSmiJz68;a^p^d)Ya(x
zYbbDQJ|0zJ9$bE7Cu+2a5p1|A&B+>(=sB)Hr)ttVYCdP#%U`i1(etgXWsbCf%q62>
zRW-WQp@XZg(IJaC<xi5_yGEeC`$tx*TFn^M)d^CvDX9-M5PrFvep7VlX+v6~6L!5J
zYmI`wvD=o@Gwj9<%SU^3%Wi7XsiGUqp>@S%c!V;v0yW>9U<r6Y!2jPeC<O4k1o0@w
z5XynMZ73$rE%^&<awYXx-VDZTq>TPI-^Hc^UY^OE78d04a>d|GpnDY&D0ju#K#@-h
zYdRML&U@IE`SSM)ADeqBbSF2mdIlISRxU|hP=9zqq1`So2~v>y3GhH=II8#BQgi?>
zg_}&kKuPC<w3QgQtY_drSEC%g*wDaTTMM{&jHacx9>r{deH`W+@ycm(o7c^n*fdju
zW|ul`N8d>ChHWQm@BPFXUfb2$7$uk8o*oINZx3t4^rld#+C29Ii08JmM`a+KqR{Vw
z1`4vM=K@)ekLVGw4z&zlIc-m((GMq*WbA2t0Mdf7ojenGBy-Tx^Z1AjZe5~kaXqp(
zMRkOwNo?+sH$_Th&*h3?N8_~@og}C3u{Sjp<ibPkJ4()<<8O-ANVI7uqB1_(uAV;m
zW}>ADPO4LHCR!%+ku#gwn~Ab80_-2h-vDxaGY(?UF<q%<U0+mKRRDBl1*}nZymlaf
zb@jK9;UZhsIMxb|3s7%cUA*uWF_je141T3axFGi3`X%BN)gnAB1lXz~A_n_jkb5(2
z9hHAzc_Ex}xGLvCr@H3ikzxkGGIyziAq>#E)h^d#1x-%L`H7-nXM98U*P+1YEMZ7u
z23=q?ULGN<pEgr5gcQ(hx;(fkZtXK~#-Zw=h>5e4blQY-vRQTh>VL4g=z>J(^U=k>
z-v83>#q$mJ@B8X5#T;RwFQ|y@2zP8TFyS(i_)QhD1UUose9<lamt*<h12O0ruc(Y8
z>0fL$K=}Q!AHQoAT_M|#t?N0<V<NB-)BKuk2F4ZBwjG1b`<;0wrmi&wGTc>L_bj))
zH+7Jt<8^@aaL#2vRKFfB$^|U@<rR*Nhu@_iMc+ss?Lg0j%0PM<y`u*@j4H&3vnN9D
z{nPd|(72wG1U!3F6dm?O19&<-k7lh1fnEsOUD!u^E0GAoDEwyL3Ig)veIQHJgSSM0
zt}=7%kG+*>K{*YNzm=BYB(Z+fFs(OfI8MH$^RWcKIBKWj4T&&Ya~42l&O=bGw?x(K
zQGUOxRPk+sm?Z<6s<$*2ts-=4^H%+qPCs_&el*8q`V?LSTZFT!Rc}c_10SjZVwc9H
zE(Vycg2{2aNu>i*tgx2Ave@q_M_b=Y8^m?yB^MxWb34yRrBOI0LjDM#r|RYO@O+wG
zQQp!&!ey;HX4eg(3o1bY<kqppsE3s@q&4)GmWqoDIM9EYtfWf90AZ(%>+yX4Bb$0l
z$Dl7<6s1f7c9b!L7tw5^SD(>rBui|YRY&BiaBKd#7>Zje&ByaNZs+&(%sxxlgJ|LV
z$Iop+z_K<MaV`pv3z)AyDF5zf;Zh(F`8JmY4byMTlhq`<GH9zWYq6b^PFE1{&F55H
zR}&&;;RqV_&_ubF+h*YAhL+ob+g~Ov(vBc771$W<#*8M{3fof=mZVN}f{>l<$C)P9
z4&h10G37)*9s}XPpRxnhyhSQ>d#q>7Hc=QvtHl1g0_2H?HAWnkr%}dU<mRyJ_AGvI
zcQ)v8(w@g0RD5DD%Dr9g$eX?W%@MzSG$L>~@^+#aw*)`}a`f%A`YKPqZpYphRVtR4
zGoQ!b*77IFlR`(Gcw1yjBosIR<R@>7A1!&thEvgrJN;$qBzK3S0!gLbh{5W)i7TT~
ziaS~*h%gN#)!|BC9bKa3K`g;WRTH%_>B?Dcyr94(Q0v=iTR<9EuR&ymNo&$$4??q+
zw?*!a<Sb$TY*kPA{0<?Y9<iKiZvDS#2vk$<ND$lGA`+PxBqb28YthGd@vaHL=f`gK
z@SzJD-N@+Ie~$@q@onuQiOKgAQvYm2@e_=gAC$>~ErvNj1ioa5Gae&5>x*D2Iyfg!
z#Oav6@gmY2n~B2fBxhxAXXDkAui!T~7j-d6qwc)k2-N7(f_oamE3dp__oKTGx3XG@
z5AWxPn-w6<R69f-#4@l4fvlUtrM$jNF&9X0mj&pD_%LDprQa`K$=hlyhE@*PwW$8_
zq^+yE1Ronw=PHx98F;6-tF|5M3?p#Y57<tid)1Y`yX8gb0ri=!J%Mk9(jN*6;v9D<
z-H%oX>j|<Io#7E_2l0FXZs+ma+U9WoAZCgrREN><tjA@2BJgMq?P<(4f^avCdKT4^
zP5JW}K)ahw*b6oK{PRk)TG0#!>9~1u!>iS@Y{GdDb=pzABFOeoWp*rT<7br)!sE4x
z!kpE+V<&1wk8Z7th~X<csnRT2U!1B<q)Zw_%k((lM$A#=)&Slnm9;u8$!p8!s*1jD
z!;bp?R~=P}b=7AzfhLA{&Pdk^;+<i#K1H>2xB72beQg{laQz-Jfe?o4)}+ce@LO!r
z<K^}?YgKn?_j2RWTy5Hvu-$?J12GOnK6j@Y<#09Viq5zn5Y{JZVN8w=rx0(h?}4)v
zG-C0iD5(7=#Yw}15XLYWH**q}YSVskwb{m1H_N$(E}nacXf+)WucxqkGkSw+eXpGj
zMEtsDb5RjJty7tb(9N*~@2QtNgLYq)h{sZZP!^&|2bV8Av(=>0&sKY&m!3X&aa)RS
zoO9K*iwBqED{(hL0!aE#0fKN{7qc2*olcNv29jfUnZ@<^f`~FmY;PN?Nu)Qw9Byig
z&|&*Ni;f|!wr#c78r})eW0M20DZ6@O**RCS7aa}yU>@poWyB5K*Y7tyY?(c()qyhF
z>(X_<G9J{5*yxq8x2`C{MJKx1T67EYL&0*tuqVPF4kB4aKNV=_M3A$nk;CB|K94El
zl9Oz%UI>T>60#qpMTVa<o&!G+C9+qMt_5S}A0%o>awWXARWsv4XUA2N!Q-MgIq?IH
zqupItk|ov2s5{b+UDyv2*_}swo&7+^AzRk4RYV=4u3Kfm<BGQR3#*E1aCQ?`t)gUq
z!)X^48<HNnRtO^%y`47jx*urkAkB+&8};e|TOPzL!eSHyv=%*PgIwj*ASs!c>YU(V
z&^RtRrvPAA2F<A~a=iChm&$O;f$m}`1^3Sh<T1-HSTSbweE3%kMa3(qN=rY`LFg+C
z!$M!I<AGUj!lr(Zh_Q*tBLNUv{VSV^5hfQFq$!$tmt3b?=9<k#1-_vRW9Kzr$<cMl
z?v?RGW_e%F-2Cwsh%RvK9I?f~nd}&7N;uuh{omu_xD=l{#T-jwwB@MY;U?n90#;(8
z$<(zfT>SH3KWl68?XZ)@eR4hC+PR^T?OJb8k2A-cddtBN0dYGfDvj{BF3+?b;glAt
z?FNE)KjTO615r@>@7<g1;SbVwmo_SoHmeAJ`<ehQfW$42Sxaduk#cGbUwlz%kD~-L
z)07(^UI_#4EbEi_V5i@;r+RdEi)VtowFczpQR6OFDtat%D%d+ZXIS9GWIghZC<L2<
zqwgeg%-m4u*gK*Y)O8p*t4<$rKTbqrR~uu2sTSy~xZb^!&cAY&>@mw!)i2s?(G~BA
zPVuY9y~pn>m1^v#rB$geRwq~<TXl3^Bo&g?=;aDqGbdf3*x`D*cS87#F&)>!a#Hj9
zz!3K>w(lmj9z(~<dG}6Q*R74#`i^!R;+etW;k=`2sHb4C^3i>Y{oNVh3WTxe#M2=0
z)!=UIHlp1E*_pS3K!#po)rvwDHVwCechcU#WO7HuV@Hh!`)oYMExFm0=D8VY(`wz&
z+R<z?fzNH%qRgso4u3cS^=;O}-_NTS5!0|%y!TFAm;a0tVFL!9RM)DWDXg(A1iGH9
z!cC?ZV@9~Yrqkwuz}-mswwB`K;dIu6gEG?I<@iO3U3>~xgz>F@l1{7!#uo?-WoyxK
zAG_7o16$XS&*Lr()*JCbD@ZcB87SOJ`A4x6Q<+Eix1E@H#BT%?VRPBlJiJnAdjX<^
zrvIU4FuO>+t@v<g8Nv5aAQ2uT2hpYvtAxkV<|^^*1oS&NcE6KYr}MA?9w*8<K9u7D
z1bMs16?zt<Y$ABNJ&%_bEFJ7cykHXcCuZ-8l!mu{3xoM?qOrW&+8%vZix95mwu3+R
zt~kiCJH$!y>%;MP)3O-|&i8~~GrpMXEl<8HiXu&eFGZJO27nZ|p6Ryi-9+lRVY=M9
zLT{N%DKO8B-UqjS&Bd8)ZNd!s6N&{rbDX0me^;9talc@q+!7TsRC(5e1yEfK4><%*
z2l0YwDKRDE1#jnZYl#=q$KiGmh?3s1cht@KbMK~Ii12Kd4=h=E7*TgdO(vY^vCL*y
zj2d<#OxwEyatjw&E2y=rn{Z1mstB7zcE+KYK75!U5-X|8OP8_A)Z;L;KlN_fB0l}X
zrd8d8Fc#TuW;A5vQJj45O26gi({A!dGU2o!=2V9hoh$+B8g)K*;CtnncS(fZkJ-yH
z+J*8bgJX1-%N7I6YcAH}0Msipk64!ouL6X8$WNkW;hl9?+Di1mHIr7WF(qyzXKMj5
zSgG}xA?Lrpv<<BS_6Rsf-Zxc>*`iw;ZkKycN-*0A#BYz17cu(g;=dOwshPOITR-oy
zhtYe-)_)nbeYLr{wS<r*rkoXUAjpKABJSNCd#rBxGJZEad{?^{bBEbDc331ujIM_#
z?$}eo-};HivU(QF6K-KV#lm9_pKFb9-EkjEx3zu|)6A;DbsZZFt4l;l{y%^psY|P`
zz<z$TPF$Q{^_j?Hb%|WRo##q)Ttyye!Ecc#>O{S44~e=?*6CO!j|rmMQvs)8h&T#}
zTm&b1x>u;{4EN|^t5gB{01^j8fyc77Y89YH{+-nbzwGoMT5X;7h!HL%n7<MqsQg1e
z9FZp#tglPkf(uGH6A|55Q}i*(j?f}dIr!+-K$xpyKcRvGVZOW@!Q@r17n16X-a8tt
zD>{4Wb<!;uq(JhRM}??)t;LFU`h+PFp^`WZ#qjq2k*cNWZcJ)~@!t=fTc3K7D#inf
zO$#nJ1DnxUu#qS1LVST7QX{NdZ7$x&Cfpc~sFUT19zptiuPzZb@c`}z&VF?yRfm4B
zsI<jEi8vXqH+Y~<Jx>0TTEOD@=D4hacTc}>D>2F*zZtua7%8Y!nnTR8rpN3i4hiuI
zS~$u!)Z?AIwiyVX@sNAl@q${hU^{_}#{p;Q?Fv$fJNurhP4;-+hlkNa^ABxbEnH0G
zfk+;y!xYi&K=Z<`*EiUUw)}Ab$7T?EwL?{fQNd;<iVzgT7}!(6_f!0t9;c?J?0F34
z9PU&v;&}*{5PL7t6zO##k0b9T%DL;Zn4@}TjyU90i7;yjHIA!b&@G8hL~FK4(q%_2
z$dF8K?}Zq{^J)Il)dMV`Of?4KH&7nDmk2of(K}Z8ULr#nKBv1Xy=CV(kI3r#UipU0
zi5H^78lix>Sew;GuY}ozy;K3-Q(WfW6V+iK8Qyg7CE`yPZs%@W^sX4kZuTrlFnpZz
zIU&*HVp+J$=&+yIpPhP6t^r-qIj1bCeq%sbXBVRW*!G`TQ9s|r^l>OqLSLfLfu;Af
zbGf`648{dc8|qX{Ly|Vww2B7Y8^mT_HY4mOk+-6^+gm@ixp;15i<}P#mt4hb_f(Q3
z=l4~@I&~58g5LbP<6%tR(_Z9UsGHUQ*dAyE1-84T0C@)?7M`jC>6R9@60@14GknF>
zK;$CSv!-X?{cJrTB>z#{P^+*EKq;<~!TgLD*^#%^5UnsGOW~D8h9kP%jV=)a@WDwA
z5H%F3g`L(O#xKd>H?he3f=G0q0ee(lqbc!fx=Q}S4uoIC!F;TtU%u3Ahx!8C=gEWl
zM86YP)Sl|`79ZkSEQjRg@w_|~NrGOeY#2vHOvK*T+T<QU-VtW&$ot|qv0f+dZ@d&e
z9Ne+@<HiJSgCgXVOmU9CuanI0%C1K{@qQv99ldt){qzleAL1LjVbZDh(^fiu+p_l)
zk+bZ7Xcg}#8obGwR2cvfb0@yEO4Yvj;zwOpt;T0as~xDdxPf1-`h*VwE34C6M(`de
zv0ji*e_CZt(cyRA|Bki9!0(bI4Jig(W&Du*-P}VV_r6G?f%M4;7xM9%*!QojGx|>!
zRb4`h3vqXJkq2TyzqWT+QL}(|hi^g~tn{J4(NnhL<rA%15I~NrA9s@Wi_uifhP#3g
zbIPVQ9N&;;^q76pWwU`?hcjL{v<kYjmFTJG16A}0XbR$`<igJT@zU%@BMb3DrVZ2t
z<*!Pv&3?RW9=xAUWX#-O1~y&_q;L1%u;p@fa+<jn*h;*+`pPwUt7Q~O8enUB+1hZa
z;Cevm;0V|#Pw`KjHMY%Y#O_<N?SP2B(xp4Lqxa-%?9{v)<^3b%p5T2?Q^27Me`OEV
z3(?cjmq#zS|G4?@qv)k{3kTMNz=ku`9!EVaYyJLn7!%)%$~;OMT>Oi5W;7Zp4coJF
zBiBq3j3GYrIc||SjH@DqM?(7#bh*Mi(s{@ag5miGR*16}zp|qrB#Ic>{!=@q$L!LC
z9sfY&i3bxAP(IKKh3j=M-IGzR{<NLaqwmAn2Z_qgbGBQ>2cmX2<VhQDr3$li6jrIt
z>#x3M)#}BsA3av{feuiOHYT)I=({T!2X393e(}W@Rv!r7oFGS6Q@rf`KWKqlKG2QR
zR~|OWEc@2Lf9ub=Ad|dQ$PksYVEHK8`|HGKb$*aGne%?D@tiGPf%jO?gmnuQH~GX&
zuMj0AMpRVy+>Br&h5{E{_lK<%FXvn`HvU1{k=hz<N}xPTBlt|K(`>N6KjkQ#Q7K|%
zx2c>h&!zLHep8=|Z<wgg?KgjC^D%Xv;mIRyr;Yz!fJI6POM|XllZQ3CEd*L)MMU3=
zF)mgq+v)?=%9=0PQp^n;rN<arj#um|dl}72;PsMWTMeKjJ%#q7b655`bG#mfIDlan
zrV$jl;o}jJE(n{NPRyup+P0ptRmRh`Jk9JPyLu+}5E{FHhtwFOu)r*@NR9ik06IAN
zQFyxMPAvz95!oLE#H&Dj95pfYD(mf3de(XCCmMtEm`we;J&itlM={`k7JU+@#h2Lg
z_-2jVyb19YeO$xB+v*ctZ2cn;kJO6;yo#qmy=W>qUa#7*dNF;Nlw-6RiaWJrm0-4_
zXRM_$%dAhdiFXPd$*FPypEavbv_7!)U;R#7QJ<FEd)-zUeG%0A76O<mwRe*Tq?l25
zt3UZK)Cfd|ejAg#S^+08AXm63!2|<JtdDLhx_)F$QQ{f}#FzCtr@6$A`~9XhI!FnO
zU^C;jjXbP`JIV`}3|n}j5Cx#?JqRGJOEpVG^LBHGZq)>7cC|u4K{QWPYb&ZDOcZib
z8>&yc4|!!dPhDD3Qq!%Uxf$TNO0#ytBw|5^FhxVb{$SLmV_0WgmJ-jsY~Hi=ViEtC
zJ2r8EzWr01Q)#~o3+H1HtrM`S_o5jlEp}hcuy-5Aq?WLIwh*PG=tLH^8t~Gy2m0AF
z%-d2RD34A^a#^Dz5RMC>R-!Y=f8AE&g@oI;ZB6x@hh<M#kD^8lZEPc6JXx!4s^2D>
zZ0mQfSa>?^s6Doreh%#hrdJzUZBNe~NS^IsU~`_8j{H)K3{lIYSljstJBTH~cf0LM
zo5$+qEZ5;NI}B8M#<qTDPs+eN$*Db!>2j^MGM+wD#}hcP*z-W>5`4gm`2OTN^bm#C
zp!24MT;n*epf6}hgx1eaakL>()_EK+{g_Ex<FSTBC<N`n^3;wuXvLx{LFtbxs~YzZ
zv*~1mjzQv66QQypQL=BamZ{F9A9nf~4uN<>`uWpWU$v?R9kqCqlsA{^2GKkz8r;0P
zCa`daFQk=$1?GS^+!wA+?^u-hy`9^gx`F4wpH3}3V2;~FM@v__`*gSJoeM4m!Fx_f
zjpNUo7m)1q=_>1t?i=p94`EjTN$FCVw(f?sdvpw2A)4cxP*jtofB&5g#c+rs<7w<l
z@ew$LG|S^r?!hfF^_q&lZxE<rTJs7|jnHF8FRtan7)6a-G~(*sxd19-CbtBOFM{JQ
zv516PTSKSa*Y7S5>$L>|k=7VfFfD-$?MSF!Ip+ccwg3c{LJhw4x-IM3-I;79rf38v
z<BqMy7yWkjrB}|HDtMY`7S4L$hLDtm9fI@i-*IGb#?YJKleRUnEq84vpyRc|Uc1|n
z*q?G@B~9U;1_BAgj$)jmw#{v}A475jBNt^jYx+p=T-Cz%br7@ftT}dwkM(^M)93cB
zhk?>1R_&99Z`|0yuNclwM2n{l;-ip-_Y^Rf8n<VHfqn(g0@M}HHQpXq1=tIHi`%Rv
zJCg`y*sb7>oDl^z5_;W^o)Mk;A;-3koJFNj6nEF-Q56?u4!kpog2(#@?W7*JmWJ(=
zAp1sd*%?vORkZsYeieG<aKg<2N>oi2RgYB(fCt2{jz+M?1o*I;Ga`LA4a;LsTdP)R
zC9uZS^TZ7@w$uwXFk^`}sUW+^CvOP=zO=|!Vy$O%jw8824S^h<^EP)Tky2Bm<<F#j
z$D_|dO`U;a3cRe6td&_IL~P828=+c3*xzLSiu&Et;Bw6A_cpE*rI=UG=FKglVB;}N
z*mYmuQ))-pLE246S2(5<#c?K@MhrM&7M;<)MM#!pV!<Xq7toYp&*lZq{m{7Q2@6jk
zc0aJjtzgI&q76e}N47<E<F|b`FD8zPCeAPTN%p;@#zgagC0&jgZiOyZ1c+_WcRO2+
z35Qi4b3S}6I<2UD<uzN6p}Ed@+%}@9*$u@v1?F-xBtgy+!-mNx+Y9A8(Ip`XY&Qxa
zdvwNJ<@RDuE)YaaMl|+N<Ic{t+J5xvI!n$$HBK85iM(;g6B@9&Fc^>ZLSRMAEDrS|
zzUhe&zz^k5>}eo#lL0?dM<hG8=kap`ISpQDQJrfrYmCE;ZI^60jfwWMUUu1|jfsYu
zSRu#sh>XdOH>TxSDB6i=sWh=C8^7^gpaJLm(Sbrc{E*Z5sm3_7TrP{cT#W~zw`h!X
z2d=MDzjw4+l@K6UT8~xhd9dd$K~ti>=Z41cxiM`E4F$AbR@W#h^l&a1Nn5?DhV5dL
zfTS(1rCf6YtqLEeRbWs%A%V&1g`r~&kXKD^pKj>7YMn6-{{iVxm%8ZSU?R{#x3DK(
zUO}BXIa^V4mNhomP_*Xqgyh3YjoKRBVZ!6b5|LSbj}G73`X`%;q4vY^alI3mSKsZn
z8A0cA!M!f9O`82VA@J}2z0Irk7Z{1&Q*A^cXyc;xm~-|E@zPTO&Z26O#|J%JqxRPc
zu2p2k<6w)LuHdlcmxJ)!Nw%VfSu<v9tATM!6>n=%5lOzg5wFLjW30z9gTuH`gsf~c
zYQisjWa8}rd7YppI{~sa*}&|oVhNP}Lbj)h2qlr3lxGW%?2_#ZV~oQ5M2qq8$y`x1
zZwE0R5A<V?%cU+RbkDuRC~a|T%_o7&ZL+#Ojoxvb_xLK$G=S`tI(r_q29ZU*5F#3;
zS(C`B-4hDhk*2g>T`7yWvMJHk7oT_8v8aU=DV)~vCe42v230PSvJ*{-*3O)_lX_~n
zZKs+Ng|Z29V-whHj04arnnVU|a;CP*facl~{Ot*cqJFdUgaK;`L5`c1xb^#*)ifna
zbvr8G^J@i~{f)_}Q(5LRLG-8fDkIeqP@7^1#311M(jq)3NzoDAH=x{PN9P=C(%D8x
z2gXJmLDX*fro^>0yUW>2ukg+$k%vrydkCf973e~lEX;Jj=#HVc4R8t-qOZxVCAJ4{
z8O135`tK+4Bac!RL2}Zul4i<{Se>42JisH!^}L1Y$cs_A+>@xXn)9rI(`-i1SQ}uc
z5|Giu6wr*pZ^lHqUYnINuNn8V<(`_cFbC|uo_lVvMi=6voWs{A`YaN+g%d9N9D^{H
z0zrI}e8+Hxds*$^V{Jv9V)ennj$S-)Kpo_c=xgOAmOMY!%U_b1n9%e_e952nZT-wP
zW75V((V=d~YMSgH;vnz09eotk%}LAQ`Z|3(f*9$0juRZ&_M%zuL<&h_57qma#2owT
zftx#sJ<<ri{yVHPRf~<yua1vZfF-3J#ssHazwQ(GL^XYBFx01-Bi|_PSycT!dl|#Q
z!4mdT==)wC0)A^wBuS<Y_JSkLiCS3=z7ZU4P9%n}Bf-<7TG188$D0$4c&swF>p9V!
z2zi*5aI#sX>tW;IG%(6+7B#wak4vUAHO*iMoN_DF;Z^P_Y`m4x$+e!LL91$x)1mja
zA5u_GweZ1Rf+RZX6A6>ms-ldHQi|z1Rs6`=!}{p38T#iw*c1p7RvcM52(<_UJTlzb
zr&VCHZx<{lWC)bTp_q?8{KI(03fNAy+ge6C3S|aZ*HtGf?+&odH|~%VtRRH4zVRt5
z20~Zi`@nuEdThb<PHm9d_{tvF%<(0ZU3y9thqiud)8%<`VfjqJ&S85pt7i|sZFBMa
z2!W~R^-N?`BrEYna-QH7LcSlsFlIul5EY?+TNFCMAwE#aw|_>^=%pA!vSJ*zWnqSK
zk*%oYYOIXaz`-vXYpT_az<OO>a$`dcy%G4}v*+po+iXso{iXA^9dk(bDp-*n!6H7~
zAK7lKjqrA3ZS2J~!csldV`Jkb+gC@}*&=Iv6a!`NM7Sc@n6&wo^Rdu|h4f2-9R{$j
zX?qeAavK_WDonZj{g$iiU)r-6UE65Uo(D#a9%+!=E_fb9=|xy!JDbLY@8glPiNZK)
z{jz`bY^?Dla`Tvc$Id2N@`OOg&nEIi#MWyk^z3(3J1JlzLw4${XgB8zGE>X!(|4^R
zTAt-Ft&Hc>n2uJd%JA(zt3E4+^g4_i-PXiw$k%EE@c7oxP>!qBAR5Mcb^ZRU))Y1H
zQNqw9Ua)+JnT!svbeyr=S#1Xm9%a+h_2$p&@WbaKS<(B2yE71n+vy54kgs`Q$nJRE
z`v-REFRh@8&e9YG=hDBG)DJ~x$#l&rs1!f==SsKtap5VZZYt*9W#4dQu)))UPXJn*
zDc5va=j>UjQt?aS;C-nXm=kuRqyceuW<JK<(R9)7=__X~BvnLjcUD{m2S_;z<wDFD
zk+^4UF>1Rt1R@p9aZ}!w1aWH_YF<_YX%$yg<@M9H8l6`D&;Mp?@mz!s*JI{5r&fta
zxe-M-3B<n{Lqnj3cb$ejapQ*V2<Wv2+l`L|+4S|hDH@~O69f+xWDixds>=2Q?nQ2x
zkK%a}um>?BE(UBQk5#hs_n5RCMy(zqtUQTl?11fQd@E!}E{i7ctUw?nb5`ql%$Ey!
z>_xQeMe1nh61`k$?yw{0bktzQIqvF%qvu2&ZX0FC&S|}0W6!yZkVMVgCjG=Yk=c39
z;*+X{P^HNFv{L~AVZelCQLh&rVnsaPV%1beC5#*4Ezc$Lo<bCi;aRobvf)u@Yiso8
z>cYL>TCJKXgnXUgu~-n2=$W+4zq6)ui7(7XZ;_y5;I5=uE$2j3T=<;_`IWVn$A;B3
zw2_N>V7Xzxz}Y2$18!Q|c`j{4?|j#~qLD{-1G#P?2b=Cq3IdThLvjUKQPXu3=tD8|
zK^8?RU{KG^bjI~MJbIT^LIg045Q|r)Vp=aOan8IMb(SJ+jM}V#4|Bhsi@8kE{pRBZ
zsqv)U(+lzdcEI~FuUNyfQWj!93S6^>T}izd4Ly*%Jy2cf*i8YKqKM?vIP}EJ-V(^K
z=*2gW9$PI>u^XFv-uSf`^(7YidNjk94$swd-z1@*hbjL(Q?;ofT_qRHc6?5^nBCFy
zq?<AAp3|lf3Lf7G_k=3344wUZ7&wG`1e$Oa$1UwqU~$IJ)`O_){5G65&UEVO+aIH2
zITSMI^DXw|oX(rEepD5S!G9WHTK%|urdA_cKO45^Dv<l)i<oe?(LgZSqBHErNuPJ5
zMKIkeftx7Gw4*A|uF(T#m3q6;Pxd@m$Ck2?Lh&_Dv?S`gPB1__*&-V4#UGZbJ*6V$
zH*L{uIp3XbFaK;6Es63ou9cMm)GfE2@~Z++Xojm*^C}9}8a+B6RvVR^)kTt9*OGP%
z<tVRLHBzHkld!;48b{U`STTh*ZEZ=k)SkDTsuOqC@&bd0ElRsiwT1H}$YsERJFsrS
zar-v5|3WnBB@3PvRph5*FL4=E#9bOGYbojxMF~ksV9)ZwuGOZ5%PSQ&{rduJrfo*>
zxC_Hv*{qPrI(|50a{@;ROK)BUh=X7dHJTEV`}=xymTN)P2NBT^+oC!_(uCOrM^<14
zlTUt~_!v+k@p93-MV|!%9IOO#&Lz?7YRqRlX8!~yQfJxL-`aY#8S5XgjQ~!vUL++^
zaC>XXwxgq_erY;)qHG)w?cEpyRKC}u?Gp9t00wQS-r|R3AM9t_egHy<miEF&dWX>D
z&T=3O+}{l!aY4SwXFODg19ve^exf=aN_E7ZYM6-;p~Gk85fBr~p2y1sv}!K|%<~xA
z`B>p@WS)W*+L803m4OlX3;H{HUL#>+cP|h^$(~Pym|Z8cFUBV4ze&F3<F4pFpH{Pn
zUc_@p)sl(ys+|&4ch*c^8Cf*D)N|V^q6b%SBKzfV6P$Egv#P*{_3RedRvooRQv?}5
z;Z!@r&KphKdF>0D<|IA|Jly+w!|KnA(=AL{liG2a#hs-kpmM0Vx5~P;2GlbyD9;79
z&1mfM>LKk1$)?>p%WYj=iCDwBRGo-IP@Uj#7vP&00;uPhwc`1-@o-x9yNSk7VBSpT
zR2{jc^NG}Rmu*~CBORAbm8T?G9l=6sh$(Fx#21|qhm`bI8ZS0DHYcob7sPNm2ITtd
zd+|Jsn)JS^<B?(u@tOU*b1hRPJa4h{itbZRRE%qu)Myl)>vHtXXC<Pg+Biwts-9EQ
zyEQ$Jy5-Tj+H!W=h8}aMayDaGmI%+9{9;FJJ0_VhPS}UB40nW|Gd8elyHT8?ad?E2
zJ$2$|gFO_&d>u?dzWp=XS5pr+%}05pn%UMr-Liui8u{M}u8DdaWzNMM#@Mh*LG7jj
zL@5wnd^!AAuzd54?O9aCzzFwLy^t=3mlsiz0>{_|EykNf-n$^S5&M0PvI~jAu+tuP
zdb^O;2|VhMZ|}!1q*d|hdv-!?dQ;M?lX@v8(d8*M$<PYApeYkP@(+6aqKXTNVjDbs
zP^DUV&{qyH6%dmG>A#+_Q^r5%f@ms4g6y`n7sMeJwtmXl%<s-R)Tz^SU*QAPUr751
zR-Nn4PFvFj?IW%vLDch=wZxEz-9@ogqiX-ia(ZgbSzb6~?vLK7*2*jFQ{1I7W#>@*
zqw&6=eGE<=o-3hKi2hv5>I|S2V;HXEc56<;p#%nmfx2O(07w{8<P+n<;<f*1Q~F3*
z*J+!M`r|2rbJQQg!GtqTxE4kk`CQCCvxB<K=A#v&I{d@SUnV(E*!_6L6*PIO2289O
zn3Kf|if_cm>>QyV48Pz;`D1&asqOFx!Y)l(iqT^O38zG2q&v18fN?T*ALbR|;JVsT
zTa6L5ov}5c-9JL)o^>^GeYo$N8y9qGM!6VzKMbCQ-z2?4INR|}FhQRnW*EW}0rsol
zt~&HsW6)xJ3XUd6ki$^jVm8V-sAmL@#KP3TG7oe+h#2i400R55x`uD^#dH`Gg%Sr&
z;E6yW(k17x`f{H176xpHCS&@~daR4*F)g0ok;ChS%3wXA1-qy<#6g3Kg{wT0$%{G*
zul4tOKruUdQEM0x79wVlKXx(E5;9m62p1DIk8J;)ozTx5?LBt#qWDV>?FNyZir2jj
zLWTaZWvYn{73bhZ9jsUYIBDt)J&p^(s{)heDaHlV>WgU?7{6{c7qxS+%c3|41RFfK
zx@SbFi&8!jX3p_~)OFCB0!O#UZc$(3_^LWB29T7-&X+<i+Ff*_$!nPX7}9j=_p=RG
zTJzQw*kjTFJ#_!R2}aHVaBtDgulRgLjV3K>x5gfd+9elFmtsmf`Uy5GpuUg6El+9C
z%^%rx%p!q|8H~+D2Pqe;Q^f4cZ|N3(N&Pq%_~rUNi`cwIah>mWPyHb|MpdWhyZvkk
z3js7OC6C0u7(d{1!N!gd@qy6Z`l&}`T8h$Z9&f*qTlsQ~qYt-=6+OcZqXk#j9d2nE
z^T0|-guNa(l9(3ObfXLa%g0SMBIu}X$7GQ}hs#||1iRK2$;5YK++`nW*rUBzIjC|<
zPPGp&roFqo%EfBz2TbgYM(t7bdxo9<KvT|X{#Z{}TJ10(qS#{cAwQM;L8U#7?t6Oi
zN<q}l0(_<+F+BD!dmi8OOT;5Twig;yM%oH%6=gx)U%7z6j<hDqN?C7?>e-`m+Obw`
zOW-f(Jy!kkXzZC0lF03Z8jzP})K2OdlSPP6&qUzHbG7x~aEeKsUeTHe_%`9|D$4+1
zeX0WbS#}fG{ababDEbb!UW5cS@k&xy_j;|3>d6E!F`$7l@dMCYcuoSuEzt&*Ir*xr
zDPG+<fdJ$!>K*0;mxERvqt2JQdL$QU599WP6ooM4dDY0^R54e_o31=y@}+b|Yy7X7
z*>&rVj>tygx1>VgT40Y!ZXm_hw3(z-#WvKc4G}wXry##(l{7^IV|Q#^-DJ>$O|_=o
z=yfiFt!WSQSjK#qnW)otwb5n;IwLRju+0U|Zs+Y9zxgOU=eCvi)Jb-E*zW6hcPJo2
ztzZ)?+gX>zcte85uJk|{e0ar{)Ves1iXgi0KpeQDpI`a(HCv6&3KM*ZwU`lSyrA**
zXfi^Yoek9+og|*UZK{0<+1s}2_y%h`8i-srX}>Ex&R};Ly|noW^-$H^TQ+U(`{jwj
zqWY1ZIlM`mr4N#K7YcU+vd2-ei4Em2Ai6rwo(P}Oi79)kI>=1%Er|NKmD}?eS~C|n
zw&WL@uoU_2f0(st5jV)d??_uBw~q6dFWJ#Hoq8mS8gs?Ov9{!l{D=R`pB--#1%=->
zpYcRn`uZ$Xbh~!4EzwiPMt4dDSuvTGWNlhl?qr?>lfSZxsNm_S2!AOId<K#LttuM4
zv^vN3Iq;Wk#%k18M?dB%0g8-L@X~b}SaySpN_|^e-3Yq+NO9bz9S0Vd%wQa;Evnh%
zvYb}c^q8BYmeWg;>sY=`OZ@}u)X&)&H<{|vT;RRLUY^Vyx`hTAqPPHAL6s+yaH<%%
z4^O0&M?-;$^QnpTDh1A0&!4w(f$=nlfSJ-T;qz=-5aCl8*-YTl+K>EXHYUJLC<#Iv
z&@lx-*~p5YS8>7%xr@ego2<ssjGeVGVGDxT$R3BCiV#Xp?hoRtzEWjNF%FN|z(QUQ
zB+?PTQXT`+tX0)>-&Wrc*Hn}zXRNJ98*H;M`VuZA{M=M0BLhX-jxXDB$#w$6M%;0U
z&!DDvhB&!xFS<L!hWs!_96A;vfvF~;TA9K}(Iau;hU_4gdfE>;EFY^7*$P_ZL!k|u
zofwEGZNZL5pa4H<(xLWLeV-jA+37R2VGcPOx%&l)+(Jss>@)Zms?Zo-w023{8<>&O
z;8G$OoXza$r8u34^`ks6$5c3qO~r9NHn8uVxFnvD=rH)3oV+B;{kUJ5C{A4xktf4o
z48Wo|#kXKXt+<qkeqiJ}8aqAb5L8%IAcfG4e5>(tV}KA`HJ3y*#nCb1Nit!z(SM3P
zZgm2}%>0$rM`z@}E&7G6DR6OL0=9}4y>jr348Y&Cs^xiK<T*VLqPw^JB^~#EOmKfq
zAEGnn=CW(RCOuz8mnNi#b=$4S4vdxyg2Qsg%RtXb{xu|=BcN_cC7Ijtjt7o}LFoml
zsd#l=4qWQAKxZUWoQX!4Ij&}7PH>oVDe=0^sh)e;&Z{00PFz~`^VZLB=h4p|cEc9R
z&3}RH=wDFqE?&~^$Tp8wgsUrzW_*LD3G8i_V#={Z=c2wG&D%V;04gL~38V|SPhonn
z8nrKB_W{Ehg?t1L<+C0$S@6ie$(U^;#(1u1oBDuq%-~aRt7z!Xi0#DiSuM6J+{Q?C
ziSM`yU`>#^M9)w?YPWq=g8_zgAH{euBf;4De;F@#w#V@kXCem7*ll5N8C@3W@=1Bh
z<O(}$PgR)nI1bM)r472S%bp7XZk^ZhL3yD*JJJznmlL7zx!lKX-Hu#Ngpy)_!{!{l
ztb=0Q1F0QTS>~caR>AShB7rfeIVCGPaXC@#4V)8CUQX2Ls{j`ElnS!54VFbCK9DP>
zAU*XkO3*<JFfS)M93iidRjGhDC3Iaa2(H(v2`q3j^25AV1IQ%r>ajZY?1B|=tQSB^
zp*wXQt_)6jQP_yoy)~Y{-ayDH5LpxAGYC&zla9tZUv`b?1Y%Rk=~q9sZZ#R5V6)?z
zCty9UeJ%ztq%VwPNKFQA4q8c-;PY_j(KA+6LpG%WF#v2nu@;<GS>>zWecxsRm|qWZ
zAXmwN+h@)NYNUejxN!4P5nUII@V)4jGzIr#{`~@$>^&A#b8Fe<CAKJ#88;St5HPUj
z<_@-`H{Z<ZdRcEE_+w^yMQ>>l;m#8<N(BNE#n&PF48rh&t*d4W=R2!^L!Y=I1v}hS
zGe00~`?5BDxff#)XFK{u@otMA&35;)%mP9nl+;Kp;605sV+B1_XQ(B8X#0VrN0Rx{
z9;xdr5$Hi+jLo7w)^pF-U-sCcU^QII+mrG#-TEb>m1xYW;-HI}ckwkx_w$%wzhyX~
z*b4!~nHD#?oDMm%Z;&)Hm&nESM@MsM-D6JCXUC!rTQ8!NoGv}d;*rsvP@C-ZD|RxM
z$or4*U(F>#Y;g5SIv28Bnm*W%MAQtoLhX9)PFQ752c@iXs`Tbc6wfpAvOA>1sJ3eY
zmtGv+tv07Kl)IhZXX{iqa|;DO&+DTu1_;nvbb{EHA+-oWZb<z-q;bAstt!UCb+piH
zxu}eTSqXX*9JejHW1Z2GWJ;2Eb!nz5U+sR)y7kQd%>^zhw!pL!9F=oUbCAip**hH)
z{*7USUy2sE2uzTVjC-8GWxZ?%Q!zLaN~BxWv`}>C5)P&ry?Oj|?vAs1^Hi(cQ|9!>
zV!+PD=Hq(};N~OnZW(I#HAXJY(9%N8D=d^NW)=k?Ye-H64+22XO<U3<Opq<d2Tron
z6mo1OS|$F?R^zP$PgOK~@y6AOzSrZ;?-Y<=xpNOIL)NVLh-X|8v#m)O8O4RbS@)fs
zSP)3mV;RCt6l^!3!F*%Jy{8t%@<(_W;E?#=u>Bay2Orp@@@JJW_h0Nl703FRH+!s#
z@Jn_W-=q&U^OFFE|H`C24ZQpK1Tc)x;(PV>lMxQ0d@i(-MK7-80(b_6ihY=ft_X-D
zAL@+qB<HJk^h2Fy!>ECiEcw`niB`A%yG!wof0ziA`_h1&P`m6y+&0Nm_o1j_(9_o8
zX(0Fm+ka-+hh;IUi{6mBkbV3^?Q5QnY3sjwOpr?bA|VbR3+2OyI*~oj71u|rR(JpU
zf8FEXYE+uJ+?mg5wL%Qh9(o613@-tbRUchlavzSAjjc(=DO`)S#2`|(C2Q3i!e%$!
z%?V}(r(}6Ovl|e_LC<ILH?*$6!@~`@$x64Tf#AUdB@WF(%qF(9?yyL{WR+vT3OWP`
z6E4}_12(miUWj?fsWTpkd6w!vo6;Npg99wK=?~ME_nj)6(bMH7n+;?yU|Th($*&~#
z#k~5%a6GvX+>38PgeXqEq!_p#D2z>@8DIENOB)M1O7S;jgIf&9?cnjCJltfUgL`Yd
zWAu`A*30Ud09&>a&A0{Gs(!}d+1BE*snOP>vCNTg*@jx=dN{Z@HLwr9ux<U?;j_E*
za#E1K+pa+M6}auG(jX?Q_Aol`nH;ozHE^d!&gVx0nmcO;YC1vn$1(nPq;Yl_FK{fV
z^i=&)k3AJg?r5-+dVZBZdmhs{j_fU&Hb7%4P9hm=*X;}WG8s_O9<=Moaoc*F#@j^%
zXUM*%pe4uJ6R9!tob<<|a$o<@1jN=JOs0Fi_TPZsq8KA(*q&4EiJUzrP$nSR4VM^K
zw2LZ<`toanKUc<UKWh)7sZy&vrlHmCiLkj_VAQN<j`5P!Mhg&j8|%902PUvyk0i{r
zCgI}Z$Zr>{B~TsqsPHZ>YSj-ID0aZ?vIFK+KO}7VKo|>o!V{IJjFx@@>ryW_kf(PG
z6?`1sp1=yxjjN*x#EO?epz6krybfs;SzAA<y!MbU`Tf9^W?$KO%zkkc6Oa^6V=89!
zOmgR%jyI1GfjC7tI*Fs&i=Od`7oRZ~U3oO4`SwIXoFa*B^*OP7QBKwwyC2UD_#`d-
ze%>&}TMV#gC$|0}C4hMlD0E;Cj7~46iF3PT%bJ!9E9RB>y#q5BTMdXVHvXNh#f%_f
z9oj;*9<P0)=#U895R>A2E(?-XXPeO@9_S;3w*C7ahixZ546GGfE{0>}N$~T}HBEZ~
zy?CcIikR)8#`*Psnz(KI0ce<YXO999ia)<G*+EQ5uKlt-R{z`(Tw{M2C3ihyk)X4O
z_H6wFcA5c^Zw2QWFDMg7?YU-#C1NiEBUYWqKd_EOBU!^GJJKPtz>~^@8Xk?3jy5~i
z@lCQ`Y2f;DJbwQAcXlFv{sgn;lO2f^QTl|I^He~?SI48_Wl^`8S*Yj`#V7rsRd#3@
zCa)e>5h7XXNF)#Y&Gc8R!}m^~wi^AOk>73HYCF<?k7-<8N7_+32xyXSjrHoQtr^z5
zNqA-O+rRW>ZY}CO#U8>uGaz2@`$sNtWaF~Ds-^gAxYk7p@r|uZy=5*B;<T`Cl_5WY
z(bS;9n;Q(y<S|fBq19tUfr{Uf5}u`idY8339*Dr>;LNHQV!i`pS}%8z0QK06Aa4ER
zxXUwV1C_xM<Xx<qxejff<ZkkV3Tf@Ux@h2A+r1dMdzbvm?rZw4_t`>uf?v1AzzB~e
zd!Xl@aTIe)ddAdt%9f)OxAx@dWd>F{bUv~S6||O(dR3iq1iG?vO<ml^Qf@ug#TaRD
zMz7mO^oZWsHlxUifPrtv=+5HzhbNZpM90Z*hv7nQ(V&f-3vdT}f`vyidF-Od0}Dfk
z`!V;B3ik?@*rWI~h@TMn9K_TxF-*;4!Aa3O4&%F=O$qH((C3P=r%`|+6JiT!7YHnj
zMnToMS#t6T4<?FS-DE?~r^SALgox}xfk*O*0w-}bJDS(g$R^MJfn!g+E70_)9nb5A
zh&*l7PUI66xjiaXPv%9uNIKkH{*($mVIX`%mQSQ|0jn1M&I(xncFroJ(K(MXf>0%&
z$h_C>xU2P?^?_E6p8HWETdh#Qij13holws%z)=ctoQO>-Xl~3~iyGhoju&dHI`oU5
z^QXD!1d*@h^TA^Kg~GSp&H&zb+qwcz7P(VIcV3r!)|>P{I&caxJ8-6hR#cPpy^-$?
zMHwP^4nE-u1-m3;Ts0X`x*kvEwV2ZZz$Wf=H63`KWns+(Bs4N#*=!)zfbBo|0p|kc
zhBljzCU_axz37h|0%%(AtI{Rz8w*iz28X9fTg=N`$eqR6suWi0fvULq5a-L1+J5)D
zr){~62mTkfqTjdvJBM9#JWigBwP@^&J7Uao2}v?G)S$7o!8QfRy~R-4<+Xz(PMG5l
z1F~3y4ct;#m9`u23VXz@F1{NgNKEno@lc(5x=i+l{TS3s>zuF9Xg`V{-8gZ$Y&w4+
z^zhHc?^O%-{cbyqDY`Z=h|cDTaP?%r1NQXgmqIdtJ%x+UpT(>oLn35YyxjG$fiDEI
z&}Vif5sr%pxr86Nq66^b2|KES3>J?Q;MkQ!j48*cU(b$Ti9@g#m!B(|rUJM8Ar~Up
z$pCZuGdmSvI+DwF7H>#Y;FBu?2r4G4REb+=v#zSHB-$qWKC$nrRSiCI*lGd*w)#*-
zt@_#e8CmPD1ow+fIf1BGO`=e{13=T2v?Jt*&+HpwOMo0fF5jxYGS3ERIgQix#Yu{u
zk9WU@y8B#bR37(TuPczbid_%s>jHz93fsaJE#c8#HgIBT^GU|67*q50h7Ad)^CKxJ
zftB={k#uT;P`l#-4wpy(BAZennn890;ugisb=*N=9_>5;8QjfmAjReb|2S8k7Ad6g
zaE+E@7_8lk=S$}r?S4$<e<PGbj4_2;T8s%Opo(%S#)EjjM*K>*4_b=e3u(e^IojUz
zT)ry-|6k|++E&YmxD3n*a4p97Cv3mJwRN={BI!7imVkMc%Vqx^+l(oze8;v0@fJ6F
z+bQ?F^;0*c+Kq9S2+J{GdskxP3J+l?hlf#P5O2nQ+t(ZHhYNll1$tu2*+dQkJ<oZz
zW{+be?rnG&U~o8qeS2~xohI*oVNc`t(yhD9(=!c-oEKDjuA-dQ;)@th$5pedX`SP8
z%C+psRh?Sxh|&<c990?mZ#LayQN~92TZELyuZn3r%K?r;Pc=u!5op`Vs~UMa5Acol
zk)H}sxc7`&c2z7U3&ed81_&iwaaGiO3CsJ+tFiI^-;&C*wXa&$)wC~My3%IV(KQ-6
zj1OvF;+$j~wO7+@{OA*_Q?=%l`_t;x0Uv(Any%_xhp}PFVn`;~Yzfw?=C>v-7rnqx
zU>BYb{8I$R&S>wkML4FcD+*y25v~Z>2={?Pfan-!U$SD<A_4*43M%4b`yHsHHw^mF
z&Dgjg<R>vvxvD*dGsrz=eqqxAGG%VHnJ7+rm+P#IS&g!10*T<9`a?<p$2T86EFrkT
zRz*FaEU^1A@psvi7h+s8Bt(H8ogtV(y!Rjm2Gh6{9}tsotW459Xjye#M!?!z2}s1M
z8$elPs{!c-;ou-VK`OCjtw({9?`)`9I&D2X`DT1H0wLRWl##0mdR?_04NsfmZp`iY
zz)dcB@q(}f#xqdu8FEBZ#M}MytoaV^{KTX3$a?JHs!o#rDUW9Eg^y!>5u~gauJ<q+
zD`o?n^^@q0+?&W?1j)ro&xCuah&%0btt!tq617`<km}eqb%uRwK`~*DT+``y0m+LV
z@8Y3x>{=qr<|$<E93986iS$rrv+JBtVU~1Izp;}l9O*BzPunTIvER7NB)g{7#Nm&z
znrj+V?2H#gTzM^#&TExcb*(IlNXp?~eESzxeJv3vYz0T4o-^XhSnV~@5|TAkvGAaJ
z_jS-qW>Krx7?S`5FG@u4O+uKc>uLLL3H*Cp!-WltUQ^~8s3~_%dx^_=Tm;}pbN-sx
z?wfAm)G4HgZd#YHnsm3zZarh5jDk<qx%hgN2Z47++$|4>)OxtjN~*>2JcV{4dKm5F
zf;n|9wj)+uSxaNnF$K4fb<b!P-HHYZosG6<nE{(qH7)_1d+}|Xab5_N5Qq2MebpgU
zA_t!e*mFo~uoyGb%q@;9`vY~8VYmderFiq0>^Kg^S7;1}_!YrvZEomjw$-SLRutQW
zH4VA$JzI~th=55Ofm90Lv8kn!;t?fgW?NVyO@IfD!IRC`cGW7wL=ei01qQ>Mlm+%s
zW5zAf_M;-%hf$>qCAJR_cYdHEV6$~Rj(Q&8#SYa|<*V=hJ9`qnmPYT|Q&nOjSWwSY
z;NB1RTw`*d8$VQEX!ZI1%Jq9_Y^*b_;|oZ-?MP=Lq{;r?LHNI;dc%AKUjjR(djF9>
zJKiZeWvUT^`t|Nar<Ncm2+_jyax1xwo$5>!d>gJg6Tl2%7;Z)tdb>WJqJC7W+T~o{
zs-oYUoU$0MbS7Fy`a)zzoXu4)+nTf5&P1La?N-+*@^rB~r>CFX^+K0e5gcy<n1X(g
zCb?5YUykW*twI+EE6(Lyv};3Hnpf92n4uSkvSN${gjv?57i`n~!ly<)9af0BNf!dY
zM_Mu3VVR8Xpi|p}I8e-_)f9UQs4U7b!y1~zfM-f&gf}8&SODo~1=@@zp~yvgHs%yR
zFOoFQsUSBtVDoC6)QR_^@!mgihTM-9sMok6FT`vp0XuB%V*KE|b~1Q*5HtSuAKQ}P
zctU4LOw)fJqs+j&XV@7%48_#7bB?yA=j_ZGR0nGQAr9Ah+tB2?bRMR`HdPO??znC1
zd20!Q#7<zlwS>OXb_D^GGbDX`s#f<0-~FyV6f`p7C3Wt;TJdpQ2>VE_+_Z*RV0y_K
zZeh3fI9@#8tsM#ms>NnkAU@GVAZfluj&Lk=pQ>E0OHi5mpsKs=J=dF`95)bRYR;~*
zOW8+h{azVyAs8IxM`ADg`g?Fh_$bjH26>ogssQ;Uip}FH$j&00KA{4Hjm}DbR6dT`
zDG8cFE}xImf*NL{OSVDTM>-5UQo_HAs87%*P7IHWP^t2wyv(bjb~}*>FjNt)WTJFU
zlGO+_lZyGr4=1cv?|8NNm4~A($}!j(b{~-XXx@%cja{)xh-c)fgz7#@+rS_LYSrTy
zd{a){`>gRn^8wEp!s~26otgy~w^-LlqS{_p-#Q!89aH2U%Z2E;G+J_gv8WD4iK;ae
zwfcy(S_%-TV13Ub8g#kYtx*>YOnsF0lTZJHO@9=}z30C1jr%K`5zOqB3Y!&7PoQOU
z`nmEnO1l^?f(2soMZZ{^+kHJx3=zy_L6vT@wZl$Ei<G7OLG<Zj9&X>xhA#zvO&IVI
zQD9m4HRA`Qf{qE2A*26V4LHr*A8bv%5$$8h*5lcw5gUQg8E9ZrSl@(+-HtxeEo;Me
z0#wTcQ@pDt9?{(Pq5vJ}ZF{IjxF&D?6g>CK-^Op`Zsqkz4Uwy(DTs-?jag>&HHI{a
zm3yf1oW&AM%i!GTs5`DaRSS<U`s|}`+(SO=BmCyXFvG)n6z(xVzVaP=5u>`oQD+}(
zi6|N+8}E^i6M1F%J0&~%aUu(HMY+)E*vHz1L#A)Xqb8Wy3DwL9lm;_$GTsX8$q?Je
zI_9|76fF~w<QnS<0V|?^WOyTYw|)(&MQsAp<9`sP9c6c~YL!AMkVj2)L#ke^gKC9N
z$N7(&tu9c*gxBg-i=ZA5vo!?(tQoknMPDq}zp>UBUjKDAGZ=#0$BFp1|Mafqquywd
zRoYp`$!)Bi0=qued?$;&DYwY+Ev8%DCik7+SwX-mU-?5Teyr>981~?-mm#4#(u)eR
z^szRo@|~BML@|)9zZ|hCK^&G<^>pAg)H8|SYrw?g&>cu-KNcy%qOT`X{FTjpoal6%
zNVIx-pFn*A-j7DY?gzMjk9WEdQ^6hvD)5j2-tFgc+<KsPm;nFSmZGR_py2jh%PJzH
z6;WsdB&++btqS?GTsGh|*P<0dHh0foS1Tf_;R>tgY?4iHQ;jlosoasaV|Gxrgmu79
z8LwG}{R`WT4pV@gy>i))$!zu)_Atsu@LaKdz4Q#a;Q3Ju7OS7+dUT*paW`^)`Ej5_
zXazeA7;ww@p9gN8SgN15r|Rt%v8R&6Qmr@=`s}$rJ&_#jg}O?1|9<<lPef~jXe*9<
zl9pSxH)}^f(K0BIEsAS5hx@Up*vO%O{FAu2f!?s4_(ZE|?>{BI|H)5+r5wPn>eMHq
zZP#N02VOEt*(!UiLU1PB4h(ULdg!~$_Ex25lwYU}qDe3EE~`;zn2)C@1hp!Icij4=
z)kVd=yEp;Xe-gKyb?gy5?>FaxH3j00q|m_>GwQ?0Vgem6)cXs|#gJm<=3^8gizX0Q
zl($Y4COOBtqWqHYFHVo$LjUV8xBt`%pQJtOQU`h&r?z5DWDBc$C>ri9fFk-cE5#>h
zzG350?>s#Bl%7e$^*1&hh&H%vWJUu@;+NmrY#B5QdM<_v9dHcKe<E_^dm<}xHo)!)
zudM$TW-bJczaL#=|F95nI`Cmyj2^gZW7P3LB^-muD+2&hAx!&6Y*}S!5Q)nd5bi>R
ztAX$tm%gpV%;7CR$zHr3^={H;BkElwWteTMUN2jjhp^u+4|<IAVJAAiwe?dEt7*HR
zq>bgh*KIFKa=5J%v&kODGVV$4MJ{dMk5T^NC-z8{nY&dyzt-&_J_a(bTX%dxJ`P;9
zc<nHNd2m;bG(O9dm}5|6D}NfGkrh9SNt<I+_>Db(Ir&Md?~CAt>W(t+)|IGk0;c;&
zmyWk#cuAK{AMFw;bvO%otSeD!-qRZ&?-BuGuELIYqAQL6T)UlAgEkMPWv9BrR@LJc
z$_34X5M^Cj3hsdsz7-W}hW3|}voc^`b2nqFQm<JKaY?oM`Q*D+Bb>3FwAwE16rKx6
za{0Qz`ho{0s@IF-31_MCVR1#r>})OS3R41)DAuZ~Uc%9*elb)=PNHVXxb>g;P@RFU
z2N1BXuGnGRTI)8Qba!bx8fMZs1Pf~XdHh*XgZP72ZKx|eN<5|wHpMLPFCq9JTPYwA
zE6~5|xUh8l2_}0=Z%OO_E1ND)4~AefWlY>J(LeI6szz|DxhQ6}xnaz&Y+jg26hpW7
zq5@IFCsA_{6*}-pans;IqZ~ynHcPCC#rR=FMmk6~g$>rO#IMyPd1`W4P<53dXisJ!
zD=N&!T=})CkGb_*6ng>DRS@YES=SJ@eoBXU2*qeUPS&R8@yr?9j?eK+j!4@HoG&H;
zj_pRrFe+%K(dudOm2W+sfahUU=|$t^%J+RixXzOBfNqb}=S1Hi?y!OO%?TojKJL;6
z@75P&QB2X-?J&NMZ~0%@lK=&8?)S;30)<IHpFLB>Y{Cs}?0EprUv9c=FJhFVc-+{h
zX?4L*g?RYLry@~qGoFphj(#c<#WHNjj(sZ9LLZCy;~ifE$5oD(8wd7@Perx}=UCP!
zKTQPPxAn_OJEdpmlr7V9b`H%&MRb=mnlK=h(SwJN7`Tmui<p_8YQ1-);Cfd5Y5C-K
z9rYauqDJsoXTP%AfX4;3|2nAx>jHfoh6x_6m(~<4rAfH5I}%CHTA~w=mIz6-YFbDO
zFeZ!coN5zW(g&T7+7h2xXLK5|Agn7o?H~I3udF*pRL&a>^g>Ki^S}tIgJMhv!h0Jk
zm$bKEZnsjr^|*njjmO)K(H=zeQ_%-9Fic$n#2Cl%Ou4|-j<cU8-%8*9lj#CHf-fhV
z`!wySXR$B~s8>^XvwP7Pe{b%Gf{MA}9a~U=%;*+XkWx}VP<wjtlVIsnaabe<S&k`$
z!5u+J&*d<BYEHt`Rl)GH)z$*Wza+Wrdem!zK**I47!iONJI)yAbp%%1@jQed)Q+BS
z5RS!mqdB__-b{P?G-7O^&xhsNK>}v`@frOqvz0&6D`)GV8yX&H0OygH_ooP~kE6#+
zm_i)tdEzEB{zT8nf=R>`@L@NJ4WR-|%%DKcj9b{h2sk)~nti4T+xnRu`79Any72zS
zj((Qt5K2NUz+L*-XNj14T~L1fGf@OV81O!_6Q3m-?m$T=r;L+=bS2l+kZ-nADkfkb
zVn7f4kqLC>IgVWgU!m$Q{$$BiM(OAXS;x=|RDG7{%{>=bLY_XaS~#~}K4&$8?Xn@I
zQ>qQr&R^<iZqHj?H2(9{*z+gWtFdQ9Y0{9e)-PGhXK9-uCKFqGt6n^!HiA#p%D4nr
zUYNr0dT74Rz?(1uw~1C8Z*Jd6fQXpaDX1+#BM=+6qM#8}8@Hiok(?Q<6yq)+C-amA
z<5B8|<XxLm>DCglt)01?*1$M$Jbb46(#*ujFL1J<iJpz-=^|xwQ62uHN8IWBXTiR^
zvch@Kzhn2hr$#Z)_X7k`0T2o;=*`WP=zsCEm)Fm)m=4ce`9KJ`s1sqmEyZ}aDJMn2
zvSy@(%g0IpPLXD9Rr8Y(kaXSFG(SuzQktmNmVC-L;tdfR^Si0ea-&H^X4^4%_vvQt
zcJ%7HB#-{MTkgSc0DF4jYf$lF0L7xjH4l1Y_#Q|0QLI~>30<!ZK?*lM2+0y+Rh&V%
ze5jE%5o`5{e!koP_21jmm?FRtk&g3o&7K8F_N419Zaf;|hY{i$C2x!ug5^en_IV=B
ztzX}^BcCVYBlKyn9sNA6f1~86I;H}A@VPX8o+y@kGjX$!#Gg<_Um%=?dMS{H0VA`|
zMZf9xMJ8Ci612{*-BzJ~`%z|BM!${4QLBpQzO6N@R+X&sHLDTe{z(rEQ~SBjUp$!m
zuUlPU%1mFk`p;v*kWx8u6u+=0L2+KAC4j&3$}83?bViVp7v0PyV7f>O^G;RdVuF<B
z^K^(zqF2|iI4QxNs#o!0g=kzvq>owW=VE2alH)uLX@XSD?Zpq>Qs9f(4df?~$ryO*
z^Kbrlm$tD&JvHIK4N5s39r(3$*k<$z{3eH(X4Pf|WGrdfqvk;J6`EJg%nh<RkL+Ib
zjiCm7shXjBB%wk~^|b+8)OcU7v<HI4b@e|(wM(kbX4_!P@o964$hH#Qxa0g?F4C)?
zi+Kt89)n&BQ112p3tJB)NBSp`Rc)xl52|c4e(xYwGJ1$Wwt&Rjb^@e`&H4FV;gj9v
z_Od4g;7!HvK8$4mYv)m*(v2~o@eK%gtG3Bt<3P_YJGI9#4PXC@9mXX8+`X3ferZo)
z$jK0#TcTQ}tZmPPw)3jot)Hu6(uT7dUIZ%Guy`~Vzvx<bqBCxvckD>F*7D7ieeGzs
zNaj48CoJDFRU92;0*<R<pGzM&47)|1XZ<uj8Lb-Io6|`cASdBvaaWI;tUl~JyVF|k
zxMG#viR>pa_N(enI&D<>eqGfBR;y<Z9%wa!$NuP7_*#`@WAJKqD)B|mGOUjhjC!m|
z&#(RK%%ZiZtE{2XTFYp={6E(%9vs=$AhC2UUv7`?ty5)w;QBS)x}q%41+81XCgS?J
z6=Jj<e&4rZw`h-G?h@e)smG*Fvr<e{o4df005;};*`3nF{OI4=wBU`(7j{NfwthWm
zvw<#gwqfr_Y)(xoKDYTm?k1=0J^h+*wfp*&Ox#npAT02HWJ6gDj0s4~!T%r{UnC&Y
zQs4(u?_1<@6v2$!N;D!8Dm&(?Dq?}jh5VMS>6L^ty6veQPMM>&p?02vgo>MbYxENf
z0*epR*lauTVK7wvrR{c$eAzYJ-w8?Ko;qMk;W~!xVIV_Z5O&pltsk;9C%H(HN<9(^
ztcx`}h+*7<oqZh7WB9Nh>X|U)6ZWKA`*t{3`|aat{N$NvNGCf4u{?`U&^!f02)FoL
zU%SB_wNke)glE{EoAn5)a+|_F;G^E71u}-KK8tY7j`oP;&U(&p(ggJ+y6EjCK!hGY
z#b4w^Poh}vb(3~7fI13^OF~2wF!&ziA-7Bo2S%o>q9@VmSpPVd8<jofOXr9?P+FCs
zX4i2-tX7-%2#{A3KNq(C$!Y}|!6=ffF{|r|FY|IQsaFSZ{*ZT5;E@6Xp_EV=LLXSE
z)@bTX!9<F$m{XV8-rrfiN4pRqj9^eY;|)`-f$P?#hE3=MajW(Ht6K<iB_lVNo)MH|
z$r=5PIr}{n6YYLn1y{Y6R0{jeavs-IG(Zs^PS2(Sm*GL1j^Up{);|-^XFULJR3R_-
zoKTX>hR3V2d7(`1IJa%R7gay|k=<881KbH*|3Z8_Om7D;`*oMGMb)h&m+ym^J{HO^
zxGXQdoLWq9mR0Qv=6Wmoou!K7?`m{)`V(7=#}83M#0)x#lP)Nx!r9FUZZEdkleXbM
zc-^+6OAIjG6MrY#kx!0G;cl##k47g_n3uhHN1)>7)AlgtyV<q=`(@;Rj*xKO9!1&U
z5X{_xz!J&=J5hTaHMtGCcIq(Z*hR8W^vpRnXHVmIRB!gIjHYKOBJG{vB~uT47&JUE
z;(~OGMQgp{BVB=aq*o+3iP<936TkNy?bX@k!7l7rZ(4a6vw3QV<GrFz4sA@~p!gHL
zBAKy%{J=a8nW}f(8Agp@r}W01&x0Rif~)&oMHCevYi@f1*G0g?B?es;IH3F^C0=i$
z`J3!dHGw(|nbigeyxq{i)~Uq(F0EcAuA||UZVFh4r=U6YG=MbVR!hFE)mmc$$Qnyz
zv9wF&qLVRXVR<#ioscin8ECQ-4MAued%bC)eD+7yt$JM2q2+=Ayz=U+?^-c{`Q>RS
zp2@mlrGS+(P~+NW<ALUNk{;PqudchUUH4#d8#b+(P8{|)=|pB?=G(S@nmk=*1GcZH
zxp?0CbApcP89Du+-Rn)9gkPdzV6oChR&qaPaUjKYTM!c5BslsO)s(}OaJZ_5tk>C_
zrD#Mr5T<2W-v=vFTZsZ>&3vKN7-RAocd_M~=Bwd`tp|!FMDrWH+Tz_r9v|$@=nCEf
z;M)TJ^<P1K@%R(OD|*bXG~1r4H#;NARZi1Fe5;5-rU4Q8jR8GUVX}>N3HO2ClZyH)
z=HhW+pU!PN6pCNZ^5vccx(NU5Y1DEiuw>6vvbBtm{JF6DqU%+A5nprCLu*@~sN`kD
z&Avo{4f)0vJKC41jVP)hcuX%`aS#$5@6%zzy}^SCv64^pB|^A?koBZ$W=t~?{{%1V
zz}Ym5=hLjcioUcw`-!vJC&En;@7)ZvO2CmiatHeMZ@C3kt05wFWSN050*WNKtqq`O
zU6$FGb_752FtDmu2`*d&8`JZIOLJNRLSw{;Y8AfBEDUi@AhHfr|09-<p^&iSOY0Pj
zFFyNGrFHeCbwO}r)czc6Dk^#eh(dWhBaO*i#fqAmck5oaA;Ig#4v|=Xnu3mxY+N;l
zxlB$)k4+A0T9~4RA&nUCGeQo+^CdCcr~Q$f;kU4z^3!2XfGSV_q0I-Nv8@dvv8b!d
zXY78AkA08&3o#XwE=go+7NZdsGSF2N!V-`q3Chx#+nZ&JxvVkMEp}vq$}MbsZjiEq
z1+l8ySw007w>1IdLhskp-`KkFz>dJ>@*91L`mtB?80TU|HwC}f<0NkDIhzFVj-J2x
z;`1NbZv1}k%6Z!pL^drK!99$x;Bi-zLUuo@7P%rk(l9Y1C0@dT3Wj<nuG?eHU)D^J
z&6tW5cHl`&MGJ=wZZ!5Z2G{0Ee@HW_>211V&xM4nFzkhXZ|N{A1Uu=Xg|FNm!;TcR
z?vbAM+)avvLL$Aajcf3+C?u{yUkY7vT!qe4+lhi$t|S$(lhF*XV>=bkMNIZH)p8LE
z`s|8AqR*_Y!75b=9J#<15{ZwE-$uS#9dOak!K&2+tQ-A<RvWE5NM)`{n987CuzD4=
z4&aq8P}$yn*A^7CZ{o}4>_%%0C&_a>`#~<cYMSs+yZOKfM<DBrlEMCd^r2k^ZBMxr
zw}}g!Y7)eA)Cz&%ZRV?}R_RDuwV`Nr)`L)Ryq5~vuF!nDe7;0@>4Fwq63)5d76YD&
zp79w%1T`JQIgd?{&Aj|pC|#VJp4GT3D_^xaRU{HLzt5KkS`KV=uOJm87hN|=^T1S|
zc<*<1Ul1lrC0mGrc*c~)c<x}CJ_r=X+@5+VuscJ<QZ((V#_eX*Rs?C3WWB3ViF+|q
zxu$}waNE{@kC$itTel4X7<B>irk-J325eh>ZvAA0>+((@P;dvNUBOOy&`{$=5yXXO
zLcH|efqU53<ii_y9)(qVq(V&AxvLxm9!!Qup8U!l$EO;{x!)~m4~6pp%H}6A6+LWB
zPqm&##uRPvS&S1=D~sfL8AjS~3-%%wAG^r}9D(&Gx@t?g&HK}Ozdh=6eY8K(MCrzL
zJEl5W4;QcF{c+|K<q+lQ3DwFVgOmLt**2JvBziw3aF|`PJ7>|(?TI+~DxzEec$ESE
zDl}T9Hc&E~S#`fixfhLO(P{!5nY}P2QxT^eN3PXHMTJDD*PBbV-KM2qL_$)Sq!n(_
zdoz-|WK-K3Bj~QPoO+N2aXzquBCPWzd<-$IOVzFqakN;sz`5&UGN~x2kj!3CMlsrP
zH^k0#s6U<R*L$thFOq}6<i2U+F+I&<&!^(G|JL=<Ask_a3l`HT3Ix;rkKwCl`xB`q
zo{$KCY4y&<h^`RbJ>Vp_`*pjgw*7?(mu}pT4+eHAVvjAv4D<Ck?PA_0N^A=7NDrbm
zZjOH3V#XO2(gh!fgJpF~;AdMAzO5dQcr}2s*AVq6UQ#%X^}zEQ(iA<8l!zrARXTcx
zZrgTL>EmRJ-}nBS4SQD|O!gFUKUV?b6iEM}3K+q#?dv(qvBn++vamHH@C$Yzz}x>P
zW-I}?=s`XY<BPSq-LX9p1Yc$LG)g+sUF(_N6iC{hN0~cz63R9JVhaxoc|FlfUkDoh
zKc?OXw(2s=|35Lr5W`N0;b-_6V(ypiD7qcK#1KOq#xE@`6fD?nD=<#SrRBE0ZEt(8
z_uiJa0#-ystPbOsugg08Ty$9%4KraC-LA{}B^C)W#8g8JF~kr<3^BwILkuy*{9fmI
zpNoSrxt~7IbDs13KmX6Ho^;&sZy}CfPi${y2AO2f>Ay)fp^5M+>zSXc))|<b<c^@$
znO1m(`90~dvJaro7X(=LxMiZv%)TTgs!=a#&9k;B`fT?b#cF%f8E@c?SRx1<Rs4RU
z8_XYUnSM98U~zd*+Q7g4t*sE4*l-O3<2kCEdUV>{C&<P5&1ydE@f=%vbU}pa0x4dj
zGIs*DFl4<8<hVX~aXK_W)|K06bD*@Qq0zGHZu<3p%LN1kDoIk3SDQ>3qjW2(9^wD2
zqQB#BkMd_UD#1o?wQcc1Voc1%)FRV1V!QOSqqf8L1X^v#RPeBiF3V7J&N=Uk>2B_Z
z5kUAKpPu9#iupj`y^=fYpt|B9_;Cs_JVwMSg?@}q2>6Sn=tz$)i1C|j<IyPh;2k_=
z0z2d{-0>d0)+<v6>5x>a_K2+EPxJ(R<s9jMLIJ*DCj(8lsKDCuR6s|XfVSyOwXn|(
zQS59WQ%t`4uB@EV<g9$8+0Lo1bZi8C=#f#?%z;GBxI?|GZ7&FZTtgRyJq!z)xahgL
zn^>uKIkpHqAPfi7c|~*PS1U&_>XwP-L-2<{D_viuA>NDYf=g1pKD!Zho08<sdegB(
zZ-N=8erK97)Sum}*9^IoTf@)k)l~Y*CXpOto!gtvnOj=H$o1;TqS}HTnIDigQnH{o
z=GdbGP5|wgWqQ-;^0a<60>TT*QZG`ceuH~bpxVHpNS+gf(@UzK@F2wbUm9>>)R0jm
zF~_oAog>U5N?Y8nVR>&nap;~A>e>on$qVl?nucEOC1Tw15<9`!6xbwF(q;{!0Rfb?
z2xFw_)*fSH2iT_g-Odf(q2KsL6l`;#>b7mLK-riaxGfORsbY2*xlUds9;pxRrWoj!
zlduV0U4TSvXroa^A||3w6J0D&CB8?+CgP*bZ<bwZ1q=L+?a}X6zoaJhd)yr%_o-hO
z+#!kCujVjpEb9SPYk8!#-43eS^!s@iavqB1!L-sDy~C>5*UQQs5q2(j`0NGJ$Q_Ll
zvsV$>9#g~V_j}<>a&}w|GfhxyQ}H{;o;#u6NXQUa>Nm!coWZEGg{$I>7NVuK(a!46
z7~YdYIr<|*DU!2u+6QgPT+nIs^8sNTY5E00C}W~}F_sD8tZUaVRi_jctzA}GJ8=^0
zkr7;p>2HjXaaRSA;BuIw=yzc&o~qXa5w;jJc0*r0tgbw;Tsmto-UBu(mkv+2N5Zh#
zxpW?WMa<MWIh~{8$NvOtHCJCK`%(MZJhjQR_?t0bZN@ma3vxQEKG_gnnW%<jXV7ZY
zCpyakTciR4Zz0^S)d!%v^R`65$|D@BXi(}VhGs4upOtK}G@Ru^r;O(+%eX?N+?@aJ
zA{T4Oold?Rh(lq=5hf;OXOlo<KxoYZ?KhaUs4?e$3|1af&797nhnZ$h5jk$k>B=D}
zn;-ZTV6!8JR^ORJdLnGta~`!ft0J^_oa0<h7lBLPl37V!STGDL3X2xMQY!l0<&vt=
zTr%lOw=-tnkO&^6^*?Q!>Tp|Q1EwxhhBG4C>MMg%nd7i0MnrBC<n%m;YL{8P{!!Z(
zHCv(Tc?qMWjxePMVnl!MD(8c-kZn8~Eb*ZzYr}WL4p$pBV*(c7Fw>*eQH?G0(ZAR+
zRodEz&BE~j#$2M?v#I!OMRdZue<D6T3p>c4R9BBCBhOO;?~E2|=rhsr8kYQQbnMDZ
z0zBk&vj^$u2ko3H_js=8^U=i`>7FjcXc|2M&qdY4S(9X4!oQv{CYQBtZWZPlQ+&(S
zNY4IO)rk2;F&BNq5&AN@ul)5ubTx$9jaUqq43)Y`64aMAggaA0Ww8Hd^`*0j_ZSK}
z{U#$o&gR7L-*VT??MqZ$eXGsuOQ(_?SiaHc_vyT)u+#Vm^`)bn@nEi&iI188jX0yG
zFP-bm43UlE?>x|bRvRBHJu<)&!LIMC407!(jV3F{ZCSO+^uG?-^6H1Pj4S%|tPP@8
z!)e?agf!A(GUBNs1~J@d%`uvW4Gq>3zk6Yo+V#7Kn0mG;Mnk5^3YS@jW-5b{OV;h1
zqntQ_Zik<Z@_PD4q=61D;6B_rZu$7_o_NJ*=j0%VzY^_;hyt6z12!7vO$5D1^|Va`
z_iJZDGlC@{CTF%wUl}xdwA!OJspM@k`f?G}K21mxd;2-vuQo7$o;~G2Oc%j%oV2im
z>UkZu8}3biC_2DYmuuj#Dz|!u7cNvgqHaFhPbA%=F$Rp>5dKhjEc*E`e@N|k6sH7x
z{xD2wOoajXIMw9&tFW_9#%iWmU#DUqO_*GrsjjT&n{nw$RFn<0K6v9%lAa5!8*s(K
z_Bpz4+d>2qXPPc(j3u;KzR4~IT#l+o{<TYijv^>rj?&f0QanoR72&<G_HXQJAm?_g
z*8-U}>sLKu*P{Sof!&B{IG*fGtzWy71fSevM2_y)Welq(H^|xj>A*mcb9>F{*Mo;C
z8ivku`%focFWFLk$R0$kWAg$4vQ~FD#QdcHwCt9`#*f*r+E6a&Pv;uG5PRt82cHHA
zO?_lQ@jq-)wH7JQYU6hcwtY(inL>UH;%sSuI=*$S>us5SPQN>7%lmcuA?2L(S`lzb
zN=qQ=D8og<V^BB6Cl;igh@+d;6VyhhwM3Civ=D5zt78<?WYn-tfd;|=uIi5dD(&qd
zG(#aSP@P-)QHb{kXW@bFA=V+XUqDcts}9(M;pWxdY8-8oWK$iuL=nfJO3d%tf-eHs
z#%L5Has;<+QOxPWq7uQ;MEn6G2RE^tvR#@Tk7u-}ze-bM-<^cGlPb(Cf5`R;?q|Ex
zKH0A^u&U$;Ql)P0ql5bM^kF0zhhhxen!R>dt;s&;&+Czx6}Js94NDYzw2GL|!~qaH
zrX~?T;&?P6%N9ynZUVO7lzQ_>Xm+As@0rse$rlc9A_2%S7#@vNs+=J*teuGxm*L?_
zq!QkmqRqtqC!7IE4*Tq!2Eoxpk$qktV?!=exzL{;n9_({RD0;qak~_QZui^ya(v8i
zw(LrEd>)F;4WO<DLS?RwYnno$vfFk2?B@Z!5$nO#<_{nDsSW5+x@WP?8mKaDotO&G
z9?;qIJ>;?cxjkY026T9xYsX-$Y;y;6z&`#Eo3Ln)Z3!lE^9OY9dn2#F0@Z9^v#!xH
z13ENq9)SZPX8>6}{gy2XAVl@xnAS#r(oJy}EeWXHTpq&9mIfq$*%2Er(`Zn;+487K
zHn}9vTv6?_bI^k^HVo*QZt#c|O;HAW`wGNkb2NqIaJwaIQC;*>e6*_zkL<8Xf9Kib
zlyvCtxOv-VjUff+oU*J6x-cin4d^%!Lx5M$DWv>O{0jpJO^SjD`^#7>`jCiiR8R_C
z<P3`G@*GPO(Z1Z-(~Y^pu7R|LwmoipRG(xEHW@&(p0jIT%<syayXEhX<_!wAaUee3
zb=C<`K^w@~A$2)@glFt<bYH{}_=up@!}UW;AJvyGW;-SfhM>QWN2jY>A9=LJri2gq
z8nL%MQC)kKKPT0qoF$B{x~^}uGpe2O%gW9MAg)mO_n85m(DE?x+)`+4JEz*oUdhhK
zIudr3t&YYnT6_M?A-kxaiEoEVO7%CZdO13DzSge9j3hf#tS(irJd8x<nxG+sLrQGd
z)fb`(EGeKWZV1Zs38I^u<<rTkYhBq?*J`u!x}iB)=yH<e$Av1h^Ez9D1UgBA<2iYq
zt#M)*&&{XpN)}Kqn|b+kwEWETU`*M36;`?m1kYO#tvvllCZEn(Qc80qYlKFoh{18e
z7Uk1s9ODsnuOhD9TEWFwj1UWaFNyk|Occ|Uzz;jkgrF@G^vods?aaXPe7Z@tchuSn
z;ouev))2LNI!CZTv8F%|dA>EPMFXT$i@J7#+VhE7yS`oh&*>r^o~oAMWIBWuUV6^k
zZB|8JtmImq75ou$99u4cctS}h?ec2LBjrk?qGoNl>@-9r2A8Z={+;)7RCOU1DqM5h
z1ZkLvODQD#gnAippQ~L$EkneZJ!%By=q%)9z(hwXC$4=eCW~3VbJX_7#5@A^2yzFs
z9GPT$oN^l;R3|V@Jbi+N5+^V^EbLJlahV-a&AhAok4CRZEEAh-#{zImijL!wo8!?l
zUM`+IZz`WIje8%m6VW!gMR#!)PUb~y+?20!22bVxnkb8~#b;s?$jE(Gm~HWZCpII@
zGFc??=c<!9t*C3~1>7SxIg3|6wgy5qhJ_f9SS@Otl8IbW!4Pq7yX|re0HH7pzFpD0
zdgwfM#8tKGBo&ff3tW=R&Gpz0{Sa!dxx1le#ANd*Vm5d>$t#APKV|0FP;kPt2GbcN
zAKZk^9@MLq0&{iRoI#z})ovq%y|2xUy7+<Gyuskfray4M6`LPzxu!fh=Yqj_eOwZk
z(MJ+bS<PTNm0WOJ2HKNIf6-vNJa9j8GpSm2g1zMyTM`XDYMU+9M`zaQ$z=h5=1=e`
z%LjE$Lj5NuhV8y$FfDmgvo#1&RP0KN*%XMohn~-7v+&8RBmyS)8dK6De2C6TgbZs}
zeei{fLXf{n_+hF|9C-ciCSI~_Rx8Y|AdxJqlFSCk+gx>OvPm(R57g_YPn4_}*sOGF
z!=J>#bjMV3NI*4&Uz5;TfAl4o2v|6&BxX^amK3pfkLvM|*krX6kC44j$Ru^c{y=Pi
zc+Uqk^dVR((rWr^eNYfuT&*6f@8kuyHIA~;b)r89?1*|GY7x;NjtURnUw;~oMd$4}
zK-qDD%wVKF6@Mp3CwUbH)5%MeS5>5OG8Qs9K+4_dl&WR$ekl<6Dq(phP)k{xoVPwJ
z!0b|{enwM@Y*!)n&c%rF4Y2d^+r5M?L}w{Yg2#u$Z#GsQuM3XPB`rYB|FX+5hVc=*
zqAGsHLZMz&mBtpk7EN)>LK}Ns4Tf?2xS=j_r{b9$(y7?!nT=-+i7@@;cAGtv4j~7U
zI7#}A?imL?{m!(KU}oNsPNNI*kQfsxDS)knn6d>!I=YaFT^OEchIFm{RUBBW5dgRD
zkYV1UXoIl_)@fENe2UH&MhV$Suq?KvfkyAgJ8hYMKio}}!{z#o0T%~Ozj*<PM;4eB
zT^Z1%4-9ms-?HYL6VN?iEkh#5u^L2(-9D5~>_dxflX_p{#?ZX49Wiij2TU@cUN;9i
z_uNx!*}xPZ66B(miy{KtT3)}A#$c<dez&A>f|Vg%UG2ET;;dq$(WnbTZd;VO+O^yR
z`6;YMiL^PSOD4VG?lw&Zy%<2TZ?qc$jlkqF3bB1LPPcor{VE`(3`cG#-F{Fu&cVAz
z>|ktzx^_Dh6_4P<a5x~W<F2+N@i917e!`Bb7HPSL>{tw+SKTEC#{<AygPk@N(}uEA
zQTb13Xr#K5bo-<#XX-Y#*(v@0Y%lEYnd;wN0IK2ta<R?C#NeoiYH7^Q4W0gfTuV)?
zT938ofkP&oId+=Q&I{ksWR7=10~{vY(Zv8=p8i+66u(=Ty33kK7#?5nD{Am)E3sN)
za$twARp<X^nt8jf(IX?_P9kKBD*t}*@5~D6Y~e^|vkF@6W>k4@tuVWwGl(P}Z`eXw
zjt1Q6Q0J2!-sTp9D|A9l-tzC+yl8{z59h8>C1Wcjt439|K&7N%V!LMwX;XI+eGOS}
zO(7j@ViA>Wk*YaAPaYvv%eWg&w=F5?^s>dB&0QLwDd7^?(glRnkL2|##4b!>bdq>>
zMd0He3Ms^2L)34>&#oy7xfk%;<#B5+q?@4`VP<Cu4|exIYKi`wj>FKsy^4)%F#lki
z1a9UZ!K@>?hdqXDO{ZzIh6R1&&z|afnorBc5L>5D;QpJ}icB9Kwqg{(pd$jSXbCb<
zYFJr8x7etvW?IK=TlDAWh27;G`b4ZJewIi{2tFO7&GLHgiRxH5aYrVScrpfqeBHU2
zebL$nkW62q_hjJ)ufEj|1a{nN+))SPt21BZNj{{p`ab1^s$*^uoEKE>vzO8>bTm+E
zBEQ117|hC6JFeD5L}rsuMaSanJnGj8!SK)%+@~ki?%wS+cB;A_|K^9x&IC%2v^EgJ
z{;c4%Iv>xo>J+KUU8TbIKPNcYf!g_iLr9h`?&1s8ecTDf@AG~!&|bG?rCo}7!k9RD
zLR?m5#6|>ca5Ky+fqy-$QCGnUqpNBT0SoJ>p^_*BbEoUta@80Jl`+=ghH!D;Gv9$$
z)Tu+??k+#2v~;tII;n}*m|aZA7_GJopm1v_rW20L$>tX0^t{gbfif+ohvavQY<@AF
zzRZKay4Mz{Ql3-kq2DsabWGh#h{>)I@Qifp9M$;79?C6CtyUo0e``zBj$4k<u(nix
zU%A1S3494E&L6YoLcLgUb;XJp$IPGDz74{u9WM!M5-{gJL83XZ{U5*ijkUy(TtCIJ
zYS&<UVB|MN$x4?8AuKJX<D8j(#d%=vCob*H0_*G${xjKV;Qk?i%IR}s`0VEenGm2>
zET+5AV^w%ML@=rZMqR_i>Wmet=KABd|ATFd3JqL~6Y;y3hPyO>I<cP}usy=4QO+%s
zF&}QqX#0c!nR;!%TEV*>OA4+7YL%(StbmhP)XkR^oNiv3wL@y)iL0EyvcpluHLf_J
zICXy{R<bgNxf;jrsIV-Si86Xjc;MH;yd00VGw&yNRKJF&w7M;54^<Og8ZX4jsEN@e
zBD|#AITeE-B-EK$Q@CoNo{b?80CS{ZGqDsQ0yr0+8|s>g7;5JQB(rARE~vj%DO>Wz
zn6CELM;q;u#)C}9BY6$k<rtrbiSqGYQDuJ+2{afj2^)(BF;F3{0mt`x6s3r0H)7?J
z`35V&d3+)rJr_z7HTMZUc<$Mr^_@EViPN5rs6X39mWC&^ESy|Vxads8+$TiW;d=n-
zHBVpc1>(rt{3mo)F=Bz@wgpd!2447gmJv{d<u6%{{)~ySEfOZ&BseVBK9P=g%X(W9
z-7=K~r*{IxmOhd8FN<I5>q=H~%cAWncUf2-SdnFusac`k&ioVN;|Ben{&)6ZlYSQy
zXZVCD2Lg2^L&z5OzWr`%7k(bNZIc>g){}9pBL)XLJe{^#MfL8Fk&WIxX!aJ%3Dpez
z?@)3Qv%KmiLb4e1?IK^s!8ueR_cUCge!~P1JVFRH)!Bspy3vCV?FzL0LG=hxdt#1%
zeXmWb8Z;bo(wM;(GJ=M^@cx*uYzecS1MxAEA}HP50=T$&teitn=-y#$o~0pq?i`L5
zDagVRb>;rzb~G@k`IQ~h6s6z~SYI9wFn*FGS0qh}@F$0oxY`rZdQH8Zj9DytLR>o)
zUl{blE1l67k#qgVJgY|QAc1Ct2@&l{_N?FVP*{cPx9frJf_`V7-2Q9381sjmD%nz8
zif&xqgk7ixKy>+Ta`4!dSn}irkN7k`p~XC@{k525oPOVmfFIYNIPFrHI;7*C=DwXM
z0{D5~h^d3V95*Ya<CW|=XO+?cDIr^*T}s>0=`Ls+G>N6qgL&$>cRkvpze+Y&)m&yn
z)+Uq|g65F`Y<@|n4^0A3zAY%F<AlLwMwK!f5x&+40KgJaZ;MJg&KsPew%Vw<y1U1g
zsHP-j_#rQiGLoy>vQoN1;7yU6FIUCPx^5|%SEx|g;2TuvKFBGoOq2RUx!}}ivvB-X
zt+fdGBm#i2X%`I7w!|qE43<bjnhwEe;~4WuZVtTMqcN-BeJ*Cs#mIU}?kJhp7kVEz
zeJtwBJ<$=5R!X`8Q+zKcKSuQjH<wCrGHqK-EjnCJ@Gub#kmhY6gCM6;=w{ZmwXL^3
zF&|sGrx9A%WJ%;ta=L?ROHbE6Eqx;a9QH@!#yprE2&_GU7cq8AdbP!{rdLrTL;{X!
zwr20q20J1UtFX!A6%^>{cl+&_YEK_Q^LjkysTIA;lzu~Q4cUq49=Bt3`O$qdL6Y)z
zN*%Nf*qNAs<dF8Boh^w!gy=~;p|dJVP|qHn34lBbC+wU?Q1|n_eO~AhUdt}1H}*FA
zIlCCMiW=($e^BgF?2dcwa*WzFRaexA=f{($Tn)sKlaLtPuIV$WhPjHbt1qmC>_&{N
zZaC=;%W2yq=J_M*IjfuwU(hDUY<4-F(GfTEo>NYz3Tg|HDjLJGjx73xd8$RC!#<m@
zKNpHUwm_iS$xa?-%IT!&wXD%c=8+s!^dfaifN2Fps1>+Om!G||cI3?eVKQ)|v8BQy
zgWa$vG+ZV?ePtpT+j8|i{Vw`@!61z?b~Mps=8uW7Xe#R|ccTenN;!MftlAI|n6L@!
zj2lgG`~p!TvJBX!0K{qV!_g5neF6GxvpQb~xNHmv>I)S&d1nQFom=tdqxWz%bz-fk
zVh_`66+!93hIUkc&a5J|**0~>`{8nu303l#T)TwQI(S-t%lD`xQ#WRlfjPGGW45oX
zcPx9;`5rnn>ONUx2Vx|!$-JZo)iKWv#%hAnUhtSOhXbS4*aRJkCccZ%7?yRiIRt(^
zCo^a}7FAHcGLy$+@*d_bWsauQ`}Xa>v=cEUyp7JN_&83M)0Mrd)lNm9ZCtizVo3EK
z0r;#6c%AS+ib`wY9L@=Fae!pcg0hu}y3rDeFKHR{=k>6Eu~2S|=$yc1wQXn3u2eUk
z8x4}d)~-e$_0V(IVk}$Pu^ff#)%A8+JKr~A;7OTfR?!yb*7oqE7^zj#p~`sJmf4jm
zUFnfDlB8iyB^?x`cD&egqwZ>WlX;bN05N*T^BQxRXnK=VDA|IF-Z0vt=*v`enEx+V
zwR;iQ2r&J642g<J(n^oj3U21J11#JgTN2fphRG1NG$3Ld1;k})R2V@*69w%BM{|YR
zaAsJI2KAdMPJcjH4}>%EjcglI2ZNbP1lsf%X(y2BP5E2Gec2)?W2n(!1_P_L*odML
z*c4c;a@Mb-qSt_T#ZL_@w^{Qv-c5$1tWfjlZLC#JU!EF(3ob7(=n7%ViqVCO_F!`>
zF(mvG=wno02xBpuux<K+i6?1+O+?K$UQNzmB~d7kC<2_bX>OHadjk5p&fYQh<z%en
z2zTK={qCFo7xMJQLTvt9JD?Vba`W!9b1+)8qBlOIIdpzv%-{6G`lheaxN|Z`Vvccb
z;S%NP3k<Oqru3Lb&E|xjJYFrrp0{9Am2}b`MKDwSJ8wA}!<b@cz+Dbsu~V^+*H<GK
zpHb5yXEBEA@j_r7XX3MOXvBjGp1U~_6k{tRWa>ODs$I~BFC>x~^~pW2Yh=nUX$rdh
z*j&~S^}6EHaV3WALDZZdyDI#rUr)iGuf^mQpgXRs#0fpfw;MlW=XN|coQ_*2*-XtE
zP6sHgX=m#T!z4e|bB5Eg#e}cR<__yjyJL3JP|X`o(@65z`D&C&mMykmIPKKl6cWF)
z$jW7g)4`j5cfe`_%FMsWz1<cKi*RjHAof~;YAAT@T3ZtRcN0`$>2SKw*|^DZ5;?+@
zEfZi5ezshIU5tfRmgBKPgXDaoZfuAVr$od}fdekQvSeu%7}pxNRPpm0#8vI>LSeWU
zf##;+bh~g}i`eZL)=if7Kwr{{Y*v*FVThn!;SM#t2xx0L&DnjmmXD5_3ho@_PokzR
z6G7hiOC_eY5ic3&QX#pOv2Kgetajntgh1Z&54KBxhF^y{*rV#1-#uxQ`dNmL-4`ua
zx)<60;b1e4vq{`37wdQE@&nOrF@@DS7-LkFmF~)T#SR6K%JlmL5sZG;_t}wXh&+nd
z<Y+8x-i=Xlhp$d7LBW_<Y)k{n0P<^_(kz$y`UmWUz>opJZ-kSA{|MXmlzz4%Q8^R9
zTRaX0@yw#<dJi&dGXj=b4NZ71D&lR%fwS`}nm){>a3Oxbj-}eg05DW=N&lr7Gc&mb
z4+TGIm*Y?VTEi^*gMB3i;08o?H30a-&toaj-1Pr$*Y%q?_C>o9wOo~q7sW_Ad5Udf
zvqp5jdi-K=_m$Zr>G0KlJ@JywQB~H*L;uem(fK80PtN9zqyvN@QNPdE{E@Ur$u8?#
z5SI$om5EU%8a5-Bw3?B0pj@El26f0zhHR1g%#^sXYt<)NBHaLK$w)fGXa3Y_OI68v
zRa>UsvE^{e(w4_(S9dQoxvdCrb;ybvMs!%pNllXU<xMIc==^w>HLEv2bFc#~Dj`h)
z>O+-yAO+_KtK;#sEgdT27^8%>%`x=4VP`*aRvJkc)V=puF6P$VT$7R}FFaj0;4+0`
zpiQ88CpIf;*!HlEMgcYg6{!AgfeJ(^YUT-5t827fLaHr2v3sIYOXo0^CN*!1Kl(4*
zr=Q!2gRx)Z>ml>29SG!H2jaAW9TYS_#zS_fIz+eRw8Jq(lGgN>d3KHnq(_}R8Wo*}
zCybzQ+qmA2kBCl9rXsGy_q{t68v!2Z89Ol|(t_>o;jNq&jc(R(@o<6KsaQwcMeIzh
zqYRt<=CHG>UYAry5H2$UNWju#5diU=0N{>I49@_bbW%Jr7xayW<$36p|FnxyzK*-x
zm4=su>-6(3aJsBfVtPqHAiJXB=Mec`jah_w#0U#z$~A$x85-oeTBh8qXespNMp|4>
zw)M1&ZafFaUmb>7TlF-df$Ty}KYMFBk9DKGXmg^7_2NCUx%%WCNX*Z?t)fZWP)QKj
zb8DJj4%r|MOMaOwh;ndag)>{lB_=weW~*paNP|Vu2c-Wv?#;ETmO(d$T2Z)U)<b12
zjo)ig?=Rb`)87N9n04_?e4`3ULh=>r89hJ_4OT;dWX}GtH3>{#H_2q1g>r_tvK3dr
z;x-tad_D1|wFmkh+XBx)l_Xc#4s{3B;b9Hmv&~Tx!iRxoRg;IF<b8q2P%l1Kme;5~
z?L{###sZ}1_gO_adYlnFn*;&sug|t^O{aSpmtM8rGX0kAiVpwQ|Jj~Et&gKSsY-=G
zIK!yK@<3zlSEVl3ARmYlUM!j%RM(H7y~JfOR%10OIKbqvz`5(>k*yKo;&r#c`>RcV
zG=>7Sq&hnmvl>@3lKbO=(#bBL3iyfKCGL{28uB>3Cj*X4{+&!erLxQ;2+_~zH<l2*
zf?MMSfkNx@jB_*5q03;QbU4`OG$cydVCPle=>^&1Ul1@voY-*(J{uu58UCe!?N1`y
z2M=QG9ZVU~>u;?p`FOPv_HoM~F%`6p{_bAWK>K(|u16>M;@b`V&a7n**l0RlJq3=}
ztWljjdPBHsvq!b7ohJ0Y=ZvO9f<w5?9ZiSFEiH))qrZ3apwAaHNEo#E@!Ju}j7lo7
zr9@~H?m~}CRWq7)(m(v#770KDfNMw7sa2u^IHLD#iNIvA?y{xowrjW#x!JP7wvLSM
z-JCg<q4q{tu_8W*FY~X~pbrFf+&N7#sAQ`G^39{VG}tX1@D_E3Wl7oERpL>;pWwU&
zL_!u7ts`)1#cY0a{Qi34pDY`|I#G>dOd3s+6?I;t-Mra~fj4(NDIcv;of;3{?BUKK
zQAPvFMmSy<XKoXuVVo-nQ=v{aO8vMiy5Ca3cS0pR=8%;(IeI#oC$3{F{cHugPjiib
zrR@)F{qf}xssJNMI~WBG-CgJ{4h6m(RiZW7Vbxo=X1yH|nj2tll79K9(8OMf<8n+u
zkkSxD+m6RvaPjjD27F{<oc<Fkg1O?>J~=9KrR4EM-A^NHr&NyVAP><Q4fnI>KKrAc
zjqb=uL_FP5okGc-{LIB4>W8zqe>)Kx<6u|0t<i<}tl)=(k^aR%Fz1rFOHl^-hMd7J
z$B@fVpX$G}Jb=@@ej{fjO?#{uhu8%?c761;JdFOBWDf5pX)R$NA@pwu+LKRqn2n`l
z<Q`RM1;^4!%+%$(Z1$K=BDukbu!fiupC0=GiOSsg^wR`GF>g%QKs9<Z8oy_3zCg3d
zSlb0->Ew1Se9$sunkPawdsv|NtY%C{(;YNCo>wmy<f1^OnNwe@ww~R1NwjUnEHr6M
zmR5mzsv{3|zAPH{Pk&&`$I{7?_|aBGRX_bHxk-bnLgHcKt*Q`W{(Lp7<euOE?jCE=
zFwh1PNxogEKug*t{mg8>$2!zFL)Hw?-K@r_0SP)7C9Uq2VYvW>Hj~U_fsRkX%=>JI
zV*CM3mRq`TF00jDA}|``>f~D7Ca4?w%5JPap|7^#E{_PXD|&PnSrmn_PKcYc5XO@M
zq#kX+zUYow_lzgJhbrWaC5v>7!8x3RQIUPk@#8ogiZ0wzs>=?mDzTYH?MT3HCUxOa
z)yj`;Enqf(Y%E>1?Hld5>XiDR-KI3cK5`A<yQG3u9yaKt@NR8X;>}ahBZllynP<lI
zD&R<#-nO$hhnxr9nLrovuIM4??VO5-rCmQCn=J%S>>>!!nCO`|*tQo{9fFYS#x6x)
zL(^|12l&eYJsB6le%O`hya2Vt(-)AK(vn?^$){us@rSQR(;jk;{+Vptc4JKBck<{>
zRGN*a0|+O{VH{60_WS>7v&Z#7AXDegoHL$IEbf6YEOS+<tJmg@rvp|l=WV{~r!$R4
zFj^BX1Pj%SzBnpVY`L0o9YyX8kIYB-3L)@I{#mVllSq_vC+NSg{hcihcwGh0+_7w2
z*F<MLfohX!g8^D0L@^J8munc;n&Ee!TyC4<+piC&Pnz}Z0s>AT;?aOh945bS+GEi5
z<ifE{H^<=d?m7Y=Ic9Da0)O{+_daRa=))D-ZjF@-n8f$@`!+uw6hYF-Cmo~}1&r_n
z9&4-;0C;xtXy$BGSU}99Y27xicLD)FTsxdQqEbz#8&hrV%kom~5;nxR!tJ-ZZfNjr
zQuCQv{2SXBQ&yh-v+a+utV||K2Lj(VC+ZN9YONo-#}27l4uTq3hXo5WC_AE`!}%UN
z8cW;T?eV}NzmBONp1%=0uHW!H(4WzN^YmMGLO61D!K|DNWa>Z$&Ai~TQwUf#hLyw_
zwIh6bgPm1Xk2uX;Iuj#Btn`YV3p^scG$PS)?fO=aM(T=DyP*0UXBO#VG+yU1wl2j7
z_DjT%2}B=*GQJW}gG&1kyQ-15;Hq>j=JuYywd>KO!A(hpBeZ}kbL~l;DPDZG!mKCL
znSzHKAcwR*GUy~w1U{MeHvp0?&fF){q5RD+ZQhgVsA76kwE0iQh2(a*Y~}@Ol&N$M
ze0&0JpsHpX*0)(rRK@lM*$eT}o=nHKfd{2FT0t9-%D|SyCsX9Byi{1hd9gIhqHLI*
zRT;MBPijh?#T~$l7ac`ZXk+JLV%nfyreDXQHxTgS;tzYXkYj?8fvOMeE!Z#Zfpr;w
z&`t5%`|i+hOhYTSS>3qL=k)71)~h`fIfo$^D6Au_B_h=Plkw`iVA_epqE;Z0CFmon
z#Z#ZXglysuHX6+bF>7G;wne#96Y0*Lh;d-eo&pN*s(ufRC6}^2YQLeewQbEK>upjC
zx2n<h34dHpdThUbFJ8R3)($)w+*||$Nxm!avvyF3HWvtf4m<c{qG>&Vb_#hzsK3u9
z?XX6^v2{&*qaBIaat8Nk3}Y3_6rPh~DniAKl2D71satJRfjloo5@DYZxXgN20G)hN
z7fr~AJ;L*;n2kr=$-8e0uDsD8>(MCAMybbt=V>RTlX_i=n8(fqH2)BNYv%(gPZ)k7
zkXqMb7h@P*$bK&A&v--IW&OE{JJhbITO@wa4_5<n7=`*lxEABWiJnCw(@}LcN}cY+
zaw8Cp`xSA_Hf>}Ct#0iHQ7|k1!d(HnP@6V;n@$X*_?&Hed=Nl-Ah5YnH}Q(eY;9gt
zSdU9Fci8-GX%D(NC;JF%%>~=ialUVnWwz;6ojJma*KA8?8;+bjE9wsWDy!AcKDf*!
z`dR79+fx0k<H29HP4@u0AZR#r^y)9e(6L|Rd0i2IzzS|SX(Srdcx~M#YZ5NYe`U=v
zjuyf&7Rm*JcD8KOZRKIWU<)-W=RsY&yeS5W=oQ(Cb;KuO{|KyPn>7}aO28aOk@NXX
ze@;;Sc|a(Tzd3sl^;~^OWYl!4Z^DXxXX?m_KB{2?#Rbm@M`P}g)8fsvO}iBizQ?t-
zO#7~YJbcq0Xpe^7ogcKxzzOd`7q9G#PY4nZO|U-(zNJu}K57Sq1Qs#lb}&lTl-=^}
zP|SRta)#UC0Fx})kd~^yOyjT}6|%@OI2gwS1lMLKI*$uTO8e>#I8$1Yo?bf<qwm5Z
z^<?}eK#ZLVxb=_NnfS9OXE__oT*R6%Sys;oQZ8%f0?{_veA#)Slb<G`qg~Jkqy*U=
zE(#z!6(LL&AXXK9?s$1Q!0_A>UjaFc5aBMctAPlCy6jr442m)6&nO_Bo85@th<(kr
zYjfcnSMIY}+tV3IiLcTbnjOW6gV}6zs$Z_0wYl5l)V#)%I&XVAHCTjUm9|~GfVbA&
zSUlv+g6%pS1ZW^tG1+jp>+lk<yVSkKYPP3i?F`2v{me9Cr5)Y5EE26?fWR6t?}<{E
zs>go6S(gR4w!XZFKVKeIlj%K}t`MN^Az})nwO2#lCCo$h8$tkk?d_qhM__=CbSIY4
zQjN7}pcr7VOznb2g-Jxfn>1kvlrac&a;8J*lVi>{tC{k0Wdo%;(iC`Hgxq$~pR0JV
z5j3V4Qh6cMH-u+GF#sb=pCEzZlvQF<eL8uaT$eVwJv4dfPW}qrZLu!c$zVYve1@jU
zOs$_D%zsZ{<0?qRb~YKkC3e(Rtove=9+>1mY`>}}zQO791NufB1NjSkdQjhMcnFBt
zkMFO(|9+VL>r}(xz};Pwl{3eV#EKF0)8#Wq)gnVsRBkoWwM75N!3d?HzB1^?xMZfH
zDYS)e$`kQf=fXfVhXKirOngC3Me%wr4r-sNZm$+QtLe_v^+L7AoFygdx%j|cr)cK|
zuj<xe7Xm;_vt6v#bH|cP!m+=T04N;M=yd~kBq(2ru7(IueKjV-1*+uMoc?pCT@xU-
zE0cMBdwQ1Ex7rP@3J{^9`~|TcI;{01M$Or*9XeCp5z=k8zPYhUVZr-5N3}U(nE2Y<
z9qE8VPhvv4BOSNQ!h3DL{!IQln8h7B(B#tCVi^^ze59e#YNB8UhZS42L$n-Ga+zZK
zsojwf{-2v|NnnJ)8eud+>(x)Tn^0`g%Cj&#r(nx>=(P`4oZ2gP=-TkyC*eZTAS8%}
ziQnN4T^aSfBrIfe^o9D#U2Ux;iZRGWI%Mt9U)s?w8rl?d=c!hBCwE9@<rJldX7lt4
zwmGKd;Xcd8h*$A=TTWOsyVpWKx>-ZW60Y8&irk6LD*Bx%B1a$9Z`=`~UaPR6du$@m
zF2lh4<n0Q0#T*g@^!I@mXFTP~P1Nw@j-NI5pZx)=GqM^(cJ7O&b@;(ezh(P1p+6hG
z*#T8>+b!?vpzuIvO6@~J@U!O;dK?bKAHlBRNc_eF*^UO{?Ze0pj>X5C5$@vSg3S46
zGO_TEvSgicB0kEUSj3#H?pOwgC{@6--Av>e{VtSvgw94SPhv(`wgAut>tN@qp#K>}
z&#PX8UhE6iqW_PJ_+sGQMuw0}0gg-g2%Mw7(Ajx3yeq25uEUOIN4kS?@d{rw%CBh;
zuz~G*fRPv0jp%^rfM!pnbGRkvJfqEeDxIv4|2M4Z?5EPn>OnL%=c#l&xtg(jV7cb%
zdo{wi<lHjvDQ)+9<U@FSxRhZ2Q|Tm?(G{w4rcC@RPivS_9~n-9)u<1)y)6<VT>h`%
zFi{(oy2iKKlBZHMz|L&8(cL(1+G;ITopi*P>38S&I4XQ~*hyq#w~w(Eg6wJ_-iwBS
z+)x^}CjEAat2L_@f?GE~Z+S{5l*Ba2$>}LQaVZH>lA+*iuLjBP<Ug$AsdV8KP=0Xw
z)EcSAaG5tdtFjE^M{oF4x^uYoy75h5FXyB5F~8!A(Jz)gZgp2tAJ~AUu!vCzk2}m3
z+!iBoH|U9{bOjNV0qe0Yn(k65K@Q0yuRb#r@vI>Rn~dsU>#rD_bDyxNPkOceT9zR!
z`w!?R>AmcrAg+VvwnKr6^B+XTIUMl(Rg}E(h%n!RT^hIj(HLWoQ#;4v1CL?6ECP@R
zC_~sfrOzxunFiS(C)9iY^dUPL<MHH@r(!&iHOez{=H^u3UJ(%M20O{}WHUm9$dE2N
zI~RqBAaSy^^MOjzO<Yh-u30zWxcHQ=)t`M(n72!+FY(Z2b?Uiv?TQK@LXb{h4XE`V
zhwPe$kZC1OTXgP%8%Tj0IwCu2%qG%NYwzA<w@jq-$Yw=GY_lfxj1erPVz;WunGvqn
z*(xF!0Nds^ebhljz)eo*G4<`wyKKLh&?`4c@Iv%^I6<kq=0{fA?NK+e5g1<1n@IPw
z+k;iweh!0i$3(gp7TrrawE5}_u7*8$=fq8k&0x2?@!0|aZ(RAX-4)n(dSZ)=ut2mE
z6&FU=@W@H5TobkHeGGR8&RhphB>uN8nn?G_gTJ)<gbQ&kh~6j|*v=<uiY*S9*pMbU
z^b*zG+Rc%^Kgx*F$0^`N7ouI<0}n)Rg#KaVOztxE_W0xWps;e+1o&XB0O7~ow*Daj
zxMw>tcT~_u<}QNE)T>uKe~=fey+?E(vQ^QK>;F<tkS4W*l5p+sS^?}V4p{Sq?jJIV
zc)+anD#?8GZ`KkeZu!UU(JBZTcljB%b`{oD)?j)wY8$jV7^dv}Y!dY$4%5+^R6S0c
zJ{EKCQM0Y13JT_+Vvnn)lfea!(dIzCa|=OKoFM87l-6^<MMHPw?zA3_o(NL|mJ9f8
zo&vjH)lzn>vgM<FTb&K5MP@6Tyr|y@`^r{|%>`S9tBZTfRMd!ICpHqG{f@WMSX+ke
zW^y(j*!FaHXI;~0+hW{l=F7H2<%xnzWc`Gov1#C{b_O&_TewOC8U{!%aLC$jA&J{H
znII>0PY^(bDdjM_#m`=K=XbI5D3gKfrZsEU*h?{6IH^ER?~5@y@5shvreBWAWpKCs
zs_cSyhW|>;Acq*J2Q(RYY1ylRaYJDkjuOKU6%FUJ_F8mMPNCr4v*A#5(B;~@H=;O+
zIg-Qi;g~82!M?$=y{WObx-$1j6s`3*=s9~kMnZsOP(2!7#J6I{{-p{9r37c?*w0Ie
zw}R^!-Vxm2FZ$bi?RZr7C;?P{zuK1ASU4<CsV3Z6r+uJ-{qom;V<#r`CN~ZdZ>pda
zkRNJj9wKnuP6n9pwE8GM%Ver#f1ipeL&uyB(&_llx{ic{bw(JoMYymy1!en0HL8gR
z?d-(q<SthY4xlUOQ=veFP57D_-6M75_IXS?27Cml&{VmNGCh4ExOjkNvCJe`?}Vr+
zxJkl1IE`#yYS@{_I_-jHqjRXg+rFwU0M;`E7L5UhdFt0{Gk}!jQq;?_drS6Bpn@4{
zyIqbCOvbzITcO+EWmjT0R+ICymxu4WfZfP@Px!T~n)^qFD|msnVbz1^>U$yf8BSGp
zEuf~r>OZIe<`j-FewViv>_=fWh=VtK>xP;niK6|a?y>)WFS%RW(>i4AQdSGv+7HSt
zPoJKJ%^POgh*zIZFKFk$u-*E!u1<e8h?zWFwfdY;zwK!;oh{`t7H3X;mE1<me(|)P
zTC~lmBcIk1c;~K-|Lj@2J^qBcM=lb$4@Z-t`*_~*bb59BSm1=+{DAJ_lr47W(|RYJ
zJoHx?LxQK%gVy|8yGy?oZMKX?k@?LITc`$U1{92?Mgt+>PR{NQWS9@=nzVlxsrcg$
zr~hpCMb(~<|7f*=2gZsV%f-SHAbt6*?8GH1Wt=XBxj(SP%m!f`4mP$lAa&w(@PK|p
znxeRgnx5DX8%GZ9gM#czpR{Dl1-aENZA$i#MmBx8*H%28$dNWz=aGu3*Xs3$Oz-qt
zXs6*a8}x@n?c-`&wpChE><iP>HVK%USF8>E8Zh^5zB%LWT(Ledl)T;hX^Fc1&^fI!
zfutwc60wKdqj^Kx%^PAvp_AAY%e>OBz{dmxp*zV@17JNw?c*^Bv?(tkQ2T>z7C2;5
zF01L%S7LZSuxzaJNSE~lj6n=<a`EQ}7g>L-6K#=llq|2Jdw%nf4F&3-z35I`X|9Wb
zA;xyyR*DZK9LKB@aJIT3ClAF)4AO<5y*3)9j5wvq9*@%gVK=ha7J%yAwQ@&%_>$Q~
zAibJvf2RtF2*|s)OIvwx)OJ6eF2@vK*2!I|m-bX=2`5RU0?*jq81kA;HW{7cegyyc
zQhcpLP0K48y>lFn*~<c*VMmt=qwY)Ox50ivBATqd66p1?&<6wuKEX|DUkx}L-Fx6*
ze5C9CwJK(KHu=SChoTftp=@smB8KvPc38g~v39UGqf$FOmmP@@55-__tJb)4ct?c;
z(grue`il_Vf;j$IAjkA@!@Q&R&KV+>P_2c(Uu^H|XIJ-tO+^>b-X;4$lig1e^%!s~
z26Z3a-1lLGWG7=Pi&%?R?4tldXiOL>eIT@Fy-%y3hZg?Y&cq&@IYIEfr_m3-XP*R=
zEy(clb~buk*|xzxjj3~@#GRaNCT?IvD%cF$=jsGI7O3!Z(fw+kRr?}fqBSO3&iQD@
zampdB{8H$&qgK2Sy>(+R`c?dHWRfnbN~UgueXT#Qbgkm00J|D~yW74|3HugbLG|uw
zhx7Jrb&#*I;O><3j$IKz0;=10F^{ZZgiO2|s3Y_nu<zsOa4A4>l0Z;>wnGwPBJ)F3
zg(4&mMO5WwK&qUz9|Lj)%8ijiquo%A0-|%QRJPbpF~%kW;n~jg$|3A5+buiA$t8gr
zuJtx+r`|eegt1Y$b!U20T2|Zaof3=USby72Jv|j7=pu7r;B$7Sr;djXhlgM2V*t$!
zQ54k~;lj&yy9x^Uh+`PHGaf|rAXvl`)xz#jU0!4|^QyZ(L;-iI8R<eH@)iWb98^XR
zb8uIb;uyTwGJ!p6Mvtq%P(U6cBVWx<J=BBU{ha>0qjQ$Yohw|Iy=bS1KfKaC9$pIV
zcVEDEGTmxp)NV=u<4+2r#R0P^(WgtI??+o2>QHsxU!B<#P*4AXquNp-GtBGrz|Bb^
z-j#l5dhn)wQ2lYJxPF#L=g5T1_E60J!{}?G!5YkT>!S|59*<lgKu&iYENxYM_Vy)5
zt)?nC^3T}XfYCxU!{+$CzOK#I$M1HIQj5Ag^{2eG>d$S&eXw@@9`K058v<sdYmPQm
z7c_h}nA67uqJXbtNBm^+ACFeJrxILQf8RpP|E>UlSp_6ABtLwhW|-H+pS+%(=>^ER
zAS0zlN&zf$fparaxcUVe?j3q4AN`b(!`e`+UN%|%`=KdDG2UhsE2$$~09=}05y)iM
zG7^j2&U2Ty(O6ZS(mDOyKI4J{we0U(OgTi9n=$NApKH0_+7jPp69EjN8spg+)mI}T
zWWjf-kUfcF&~`^vR5kc{?TM<b+=jgCdv{78(@G+E&lfc*gn0r_*h?x)LMIQ*vM=g-
zgb#OMd|5}Dw4M87272%$wpRqAy>U~!9Z&%x3dpcmV}9F-Cvh;A2!^LLwiN|ex4ouf
zb}13V0_Q@B46XJ?eAU6-<Ek29)*uppGdixgW5toc4~29x4R|}g!}iaPYKI_uv%f?;
zoJL>(qaGH`yj^4VPE1kVKr%T$uF2;bPO%4>(f4BukaHXJ4;Q6O1-u52O7cOV>+Wqh
z8$=Ptiv|0zTEshyPpn3==-$n>_K}b)6~^t9ex7=!VyE>pvkv9Ani56Ocf=>!KHMj0
zI9w)jHoEEW#)t9Ko#IJ+YuyZ?I}`VkGb$-6Hv9AQ*o6^W8r`|T@XRsOU&Kt8vCBIj
z1w54n`!YJ)==lRL=z|-Ze3oAc;kwoayBO12C}L2>)AzNYZ6pAjT?zo`C10^`0tqxf
zqL(hmT#(@XTaDT`GLp6{fe@2m-)WTy*G7VPv<&KP75thfx3}+Oradedfz{N|bMDv=
z(f0^`)OKCfFu!vYf7Dk!$B0vPBS5sfZK?gFFL!|AY}hmD>BCE>Vz)e#9<&0+;<KI+
zr-p8<WVgo0CZa{z?D)v~TXvgTw{a8L9R2S6cqb2@ewPPOV9tF;?`MjV$&;rd4}N3w
zo=HzoKbl{*>m5;%@MC$r*ygLyeH0;F?tCVM?;IUC6o2;w2#HqS?yA;>SL?LQGr9{1
zu7j*5XA7UvlOELk39C_cgs;#UcDMR143^RCk;*;>NG!Lk-4~#epoS~5RxQzHIRCL&
z6GoU+TN2oiI@j(OKulxOo4GWQ^Stmq_CS0o``9u;Koo2bs$Wc05c@2TxquZYF;Ndy
z%SiWYD>M}>pTE?uey@Kf(Ijo_tU+iH&X?y-xDzYPiFb?_)Yk}k<ktogXo<Oenqz*}
zc*tG%tXZ#`86NEx^^7zCBD*zagKU?{cClUcRuhMCLyUjZnhmxos;zO?JUV_%gTcXw
zYY$~=OmGun)$Q@XWN^5jow_+zS=5UAuyqOUJ-;U#dN%s#M03_7Ao!kfHRUw^)sI+z
zR3T`g<%L}Z+lnFm>_v)ER1Mz4l9lx5duwW}5;(hZ$DRE~R1Is2muxiVrx^k1uqz+N
z1L<@@+qNjiqJSW$Ow*3|1O9dLsGU%KG>4vlWM>SrzRwA^Vvei^!XJa_-NI-Xl6FsQ
z0rvL*J8y3w*yFq|TYfS+@r1pEZyvIj0x3VrW!o36-N*uig_i}l8L#O5G5vL}FMTCG
z&<>JeS(Q)@*sFnKlHQ_SLfjb>@+aPn*My4)ZfB!m=P3ODZjrqqc=bs?d00&{Z9JxL
zs-TV+m#mih;HrXNcw4o`IM#Ml7-WV@_7}k%<yi|H>p7A3PSp5^T05?WgxxRM`_<*+
zmh5vwWmW6MXzYV$^oBV5annZUL>0BRX8SOv>R}XT4CZ8fa?iL+-$#L7FZ9T%XGEdn
z6@&#^Iy<Lh8F-9(&?8_m5`S=+O!5NwB-Ujm6w!d46=E$-E%s>?tg35nYqXibsK~?n
zxyD|z-Ok0>;RlE9i=PLX?2XR{06gB9f-eK_92x@F>q1m+47a_n1c~c~NA_aWWvBaw
z{W|K(;3jLAgv+oi+rJUQIkw<swQ5P^#gKhlomdI*WQ%_#hVS->P#t;uPIFf28yJs_
z!FDwUwWYv5{ysiX_>)>Nwp}%v6<l5v^C-jR#T<Ev>mOq+aFBS*c}))djTjXM`JVaw
zr$8k+T$0SkY?t1;o}M1NWtU#1b|+IY>fV)}jH>9+ZdKLHW0=3p)^D<jcG_*b;$8Hh
z3UkybqtKd&H>1&7VkperCGKjt)WtEoUCp2}c(>>6(sNYrHkm&CcSJixOEy2+rJNOZ
zXLQ#{GFV%%D?QIB|LrdQe3X5YiN@$XY+(!vuRqT*R-+<_)^58yie~=k%062ZpTz8N
zRp5P5vxJRX?Jj8nYizN4$#>&QEJTn%v9<1xaj%6ebmm}bpitmGe;@$YRTwK?k5PoJ
zl07IGZLr6v`C??`4u_YEL3g4V%-f1xI%UpdhUH^@R6x{0e+|2IqEpcCl%t6^aFvjC
zN7E*CicC0XYh$ugfIg==nyVAxiRO3O`sgeL2NTuSBGhD?)*5&cNib*aF>u&X1mtjg
zf(7TzLpo@iR4gBLgiVjBZK3E1EIW3Eh#-f-0z3=i%oqmq;mU43(wl`l$x_SK6}Yqa
zrJBpC5S>`wdIGTqw@uFlVp$^OxLlz><_jej8!oSAc{gtv(r;qJ@)Sm^*6uRoKf1v9
zhAzsw$5o};1&*FJ5+ED7B=NBpn7%>S<?$#frx+L8RFXOMVxR3$iMxKW<P*`g+Y0U6
zCD{VkCOi(uj%D2yaQeGFAN}r_yrje36EoqyZ!Z0|y#X0x5f0E~tb@C7ab>@xj+j)k
zGRGMD%h<oX93MQpUAAAfItL(?UWrYCOw8dOh%u}qi|DKIy9{A>P}K-#h4nOyUMyg`
zU|eqEQFIPP{SC-|--rpt@QI|K1ZIe`cGSpkY6N5ta!uKh7(0(Q2kdR_<tA?RqiV92
zq1j*b8?p#%<k+`RfG%C|AT;j?yQaccXGZOKfRy#}`x-j-`NUG6is5=Zc|y~D5R>Wd
zA~?7P($)C;AI7AmwCX4I!S07P`zThW&3(BY?^Ds`!vrEO+i8tr@jv{7oe3D|;-0Wi
z;-hYO`k(Ern)Fq2s4DE!DtHfEIumHg#^m#uyIw>j=K{s#;>W}Eg?j4Dj@WrMO0wvG
zv@c_ppuEyOfnC*w7(0?%SWf#&^<Dgm?E0d<+=vI(>uYs6?5Tq;Ro6O+*%7sS;IMBr
z0Uka+*<9FVA(j+0c+J_jLTn9YCRYOIW;85rP4-=M#qx9MUX3|v9wt`g_tg&<r(TPX
zm7YfThqy#MwilNI@~-PDcFhw~Pr=l**o~W81X<z-P_i%BPpaL6u@epy&*~bD1ZLWZ
zw>+zNt0x&C%z9Q=F-rDsEMibQ+pVgAOvV}6+0RM_O0@0Uo~`=C@WQ;CG=uJEd(7rM
zo8H#dPV2h=$u9zElCXLTzbTvhte(}Pi{)?EhvH_=3w*Gq<^<dk_0Xgu5Dhr!uu5Z6
zbKUuDdf|~|p!^eRYgVtfyP_r$pSnw31SG<u{8_#Kt2mf&5DV3C+~oi@(PbSG^Ly>?
zXVY`|_b`Uhc?IVlyHA6{#@H`sLK8<JWDgoZ6>=}Qo@2_E#E4vuZTCN`=imJdliTQ0
z_2*h9w%`K+XrnLwG68~d^kk(adoZT^HyyTI^~#8Vv-XhC#{jF}R_HfIuH9B2zui*4
zAw~>c%Tn5^=!X~&<cEtPyA*E7)~c6BQC&Aj0VEEyw7xnss50`_BH)EA`J!86P%Rj5
zq!1<TfzdjIR2!bv)0YyBr0Z@|bdD(;>|so$8!zI4W*sUdh6Jnlcy!6J86cKqjDKab
zbwvmLh#0aO8tg=Z-V<%6Uq?)8xfpi}uvxPH_?$W9W}KE6#Llf18&bdJKA>1JFzm<N
z*GjQ}NQnCks_GAC-nWXn?;GH57ztGTO&!(mqzE1l*dxRti7|E&65HRp9YPUjW`A!c
zgp-F|cQxD2XVc?&)6mpo@a+<;MGF_(Zgo-T>2LQ0np^Q(x4kN8dSqRrO~xE}Ofh)F
zsFro}@qO`89G-s1URKxGd{;bv+y1x{)^_GfIaUMWuxzjB2sAuw2V(LVgZuBist$39
z`@J2E5!OKzy7}X4fdDqINC*$9CMHzl_C}0dhMR{2n|$Z^n7yf<Mmz0DfWv+kMk;S_
z2RK*14%t!7hGz-?OYFQd`7B`A@E(tv=*eIc-l;D48{9}A*zqV_OVooNdq4UnBolrD
zLJ>dkvV9N*kWYhlqPl&R^n&&JF#1fQRGwcu8NJ|7Nv1jbC_anjZcn6lN~8K;583Hx
z$d!YM2kh*bDx{YoZ2ZOeBoM^<4O7jt0f;l}5tcp`TG*ztOJ+0&)9-fK=lXf?uYP6c
zVxk)fhWABv+q#iU3fF1p1zzF#>`Q%dN8!h47osWGynUsPN2cFL_<-4qMo?F@uT=?(
zjW<#Ilv$Wc`YI>PzKO}I8+84tYl<%mx#9AFeXIVs7oA?Y66huAvq<mnR5yd;*ww&h
zW1-)^*SLvjj->yZ3b4Z*wI8bU2_0E*W7g}@gR`V3>_=7Y8$bYiLnVZ-_0ae~MMwAi
z%4~OfGcp??*lyXKo+^yeRwW6>tljD6!a^t|AHm@J*4=s$MYtxZ*zDcu$%HEJf`qwE
zU#_FzoX^phlU)#uzff1Wr4W-wpfecsA!PAe;P%~mOCKxPyub<hirpa)u98FKIA32G
zh(;7#cM9K(Cmx1*w*>-9=4Ur8af7P6cBgl4`p?d)AsXGCUbF2F*}~m=a@S(dZ8fTu
zfp&Dp_HK2O78PDGx|F}keL^i;;QZGHnk_udi-rAt_uXep0$NH9k5Q=FcW?3drM6UU
zkkfSA1JQnvsJ_c$C=GpO)Qw~IV3ffi+Vb5IOxmB*6mt=8Do+>ZYRZRp%iY1}0MQq8
zkq^~Y#B7$ktX`m<q`^5J{xLw;<Ds`oWh>!HJ#@G=McFWy>stMWZNl&+{@%)+vc9?o
z1fL?a<8Hm@sG607p*3cf43Oj29(9Ry3d^@4zF<Ij_%_8nxmy5$9*Zs>h92w?jP@M9
z3-)+S4Q67UuA<&7D75jLu`Bxb<t7wJ*|LH&;IiVL*k$DSbyGvG$DAM+5Cn}_e-t-5
z-;}b*<^|htHi(g-z_)K`D>qIt2IeX^E2#)G)nQh$qS=Ohu@S)xX|KO$quO+1{Wcz_
z=VwDs+a{o$h-^A-N8sZgy)IfT+eA#t;K%=NJL7keytumpmqEX-cgIBbc5n57f_q{{
z@W=BglE-YXDtF;m+h>!4LJ+2cy%aU;{H60tbzeY&gmq@_<@khCK$2_wV<AYbQXpxL
zy%I=p){&k@A?8jgHOs5JPp5!JE`i0(&6z9M_JaWt2IUxc;A@)2HJ<F3n3*b+6e@3T
z=(DY{ExDy1j#0obciNl!$W0%{aOjAJL{3L|N1=iL%2bl{9E}dg`iPUrbEysoJBy?N
zvtw#V94B7NI52<LVaHX8x%1h3KNdyiv#x}mifMyFNvX)~12u?zaVG?+j(hgQo0t)&
zB`|THZ%*1rF<+JGKO;-wl~mOff%CL}XPR-lI1`hMl%-;ys3U^N@`Bn~72(Ig#{M+G
z_sT7PMr8<T*;k*(9nqOpSXPwjzA?`U+8PCP{6b&uw~4Ihe4vMM01D@veW^i>@El!;
znQB~VU&Y*zP#evghpD<4bHhn>E-h<c?@rVld39WKr9jx1VoaDt_SiQX?D(l?5wYqw
zC$(bVs*;bpU{};9PEPinj?;s`x2vj6Y!CE=`c0QSk%1s@^bavBm?pcC*LA^kvzTuC
zQ6Jp-?ByG3PpF2n{S<r1FIXo&?YZ<;qzG4bi#{?$fg;(xKKgoHmwW4Tx(Tr_O|iFa
z_H(D*di2bnPCa9{={J!w@Sb>1OlEE@>(()9M*L{*Ge7flpG%L_?^f9D(bqaRkDnJ1
z+MrMFcup)uHyIJ6b)PSo<Rf-Q=}vt;3SEmleZh0-S@+${#dh583b3m@Hf`p)v_4o{
z!~0+u=Ght+3WJA<!J($46eheh%<J7M!Ss32bD{)zl2T|Y;NJI~n94e3f2)o8;4H#w
zAZ%P5<*U29JT3T=_{3Jkl^^#DbREIPV^+uA4)}qnmF`5`CgPKe&I39<s7Y~p(v}CV
z5F$9qRP3SWL{QY#+X_{KjjLFF%szHKyfh6ltLqUyuZnr8aW<n#9c?I-Qk>MaG2^r%
z8;Q%Tn$?}tvRt1-wq9^`-&u=Bvv{+$23j~KB{RVGn3i=R#IOxoflPM_=8JBZD`u0r
zLpp%>BLAPT$70~{$$l$!1g=A!xj}n8Al!UcfUsHJq?`nat?G&v8{A+lD`?#*OI+4^
zRJp3fAtH(9j&Lg0uW|VW&gRU=?A3c#ARE%y5(Vw9&czrjuE1kX)0Li+jD?2?2@Dk6
zKO!`##2=FY#APJjazyhrPJnfx*3exZ@PIMnfp*z7o!g@6&l$lP>5iz;+?^w)LM;09
zyS7tHo2<@u={L-J$#&~ES+aR6)h!;pdxbhoG`AzQCS$&n)_lrdijfg(vQV6U%l5^0
zdjXb*3+iRf%k;b1KGgO5W0u_VzhJKjCzjU^1g;2_w%V)FMcbAfbf+B@E^RQ6YF;MH
z0FTI_sD{g(8ymh6AZ8AtOg$Xi*RSX|qaJd=BvVzdTl)L#ZI!rdt6x+{0|NRjUPM7a
z^+7y%VbHY3-iazMrLp4y2?F~4>ffi|D%zC(PWEg2Kv3M5(N3t&cptV9_Mr-<PheVb
zvU(x7!Mlfq{YZst8&|g5DeYk{3R0ao`-182X?5Td!ZWHm=w8L2XnLAEhyA5GtCF%?
zi+mdB^^^J4W@0{wFXcz-^H{NLihYisp4w$vAaB2j(>+A)o}6<lc3!n3;4bbF0W1Yc
zAaTTVqPo+j4x*d^eHH7D$q)!!6#6}M{<R7)E+#paU5be)4=0D7tbG%>kQ&%72Mh>Q
zw&Ay`+UTi8(lNfGKcrx!FvVl`UDSuDX05KqwY(DZ@oq>iw%_+LxRp8Q4F(~SuBpjj
zU)EiIe$b>95YrFXb&VSfIhQ4s?Z=oks6%K(l#J0EfdJWyKLu!rZx24it!R7n2y&G$
z2~@ddPr8NMU@vFw(Uak#2>$nYp|`4Z`@&kAy(gyhW^T{q&VAdS^oS<^pE;_XS<`I4
z&`-?BZLVr$9&Ky0+xO@RTAyNd+dN^@C%-AC=Z-z;;cQFdW8##RY<{#LS1r*n>`uY0
zAj?{?Cq2`Nl>CF;wI|WDH>FZCi_cVtTV<dYs*w+$C#YtR?!Tm%@mJ2tuDfFtEp@hN
zPogZ8l<mGf>6ylwu`=vwYib25g9-@Bb#Z)QD1`~elK3j6phfo!4D1qTV5vTcsKweo
zAefl7_#1^nQiZ*I_u7Md5}{EVwdF#IL?rf*=3^`3YG?0OL`N%AK973U8|-SbhU#J}
zXo<56tF#bhUPx<-L2mD`wehogjWw$V;VB1gy-?{xvYWLQb<@VK1?yH74Z{<W@IzSM
z*I^r?OGMUe<RnPkq?+8a+zXGXBR3c6h(4Pd?eS>ET}-5&bcQyoRi@Qvp({23p%6SD
zM^-43(W7iV+AW#3ZtiP#Jb_3Am&vsL0J5<Y^Xj~i{M8~Giru)T?GY<RduS@ojuUYu
zHS)KVn0Ya`V2`?UBr3CYd~kF>s>&GMAe0za&t0$t+oB7nRd>V&Tvg@nC#nmUM<QrD
z1IZ@Ofw)T`5Utzp>aO?cC&EcU318TEY;WvEkb%4FvPr?oxf#E`RGm>2BC_GJeX-FI
zUwD0aPrWQWv2-M>P;|(h?X2W0Dnk0xWe0>-T|<Mtx<|}Hao97&h)X$G#m#e$KZES~
zwE)E>;#O#fV!)kau84Re5O=i(x(z!VmAi;8gvj&Fn2(X^x5$ne<G4v9!9E7CEUt`%
zbPzBs$^G8`Py35jd1JdBi+agG$leL4m<w*P<ALu@7WDlnb)R~h(!f33k$q4F0!e}5
zF)-hX3dQ&3htVu?geNse<n^_W^wW9PQ_)0mf>1d|6VJkCX97A2cR9zO#O`u0{~q2A
z!bk+3*zM$!An#M5Mj~gVtFaJB*@!U?6~feh9snOsxns`77xp%6pf6&YM+#$ZA$DGc
z9-oxM_+_9~-^V__5bc`So)!B_?N%ag<cQeCz^QGE`@vuc__b<Jzn;W*m!g+QFa3=W
ze*DQ_b=YP7&b0ZP8Ph_He~`)AmFSYd1DxCMq9!6vyBe5nMLJ;LM>F_TKZ15mg~;hS
zFh4}so*xvD*W(kh61R%|7)|}SxIXQM`am<V(0)?yY=>lU`n=?hxZC4i_<VXBd*NAU
zJ%3ZM>edMUZr}R+X=fiZT@c~3pVzY*87y$Q5S*?;J?f`9&+FxYkaW9Z`^EEm1u?I2
z&$zkj=F~GKyZw2wo~7w`(TmJeLFSjgv^&&Um#^OZ0Ec36`hARr?u^bcZ|4bP^b1rO
zwfB(SrQgtxL~jg?+qvl$3J9u7Zgg~UHPN&9`MZTa#(Ufii`1;Goh0!03H6mZoGIz4
zHd^@uijXzF%7gCCxup7)<WKht2nv#lE!FRP?!|}q0sZcAqHbBVsE~uYY!6m3O!?l*
zwmgREc|()o>_eKK61IUWp4US@;~|#bhj{Uz)b#@0SAoB85HJ*gJVhABuTt5}tME2W
z0T`K}vx-?;tI@j%q0^f6K_qI<!}@AJufzH|x9%=OEkc_M-9yc_#^U1B;M7EWj0q=x
zHr|F<IuG^cRKup~Jo{umrjJa63rsp<!{NM`55nYejW>_xWwW48zn;g*-nwD{#6!ij
z3P@RvE>p(+ttWoB6P6$s-9tI_!E0H6G~7tmhyk?jmRE&jl!0NE4XN=H5Ik0lfvtj>
zVb_)dF1LnNgpOkL+sIASJ=}YG(MMIe5ARMJj~U6LkU;gQ5~omYhkAE`1XKH-O+*)U
zT{s!WSy;KscInTErID=)Df+?amxx1}!mZuzy}0*zQLHyZw&=l{jB#w3e#^z}F9o*2
z?)sCpPaQyvSM24e?RF7RZ~LE5G}*e}+AC2|$PTz}@IZ{Jt*<+W4@MLcztdef4#q&r
zyg#q0VmO8zie?CR#yjne7+Dtr$HO-lHN_Lm+MAl4!q7_e#+=wAQGrBdB-cPj_IC8r
z+DXjZK%gFdB<$H=RKzR!Uv^Bt@4XkB<#+UZ3o$M2xZoo)PohA3KTs<YK8Of{u{p`|
zjnQfpm8Oq#&`-pna3K;h1OiQP_&yZ0j}GZSJ1K0rAd?{~<~ag5eI%LY2#;~U=TFDD
zI9twW{<odcoHsVwC)xyqNC(fx0NFTx^?Z7B&)p1OHxoOixz0Wpg#UGqor`8pNx{K>
zQJv(pH4s>x7b3)}?6EKPJ2SxCUeItd9`LKoo%of?R(UeCive?HUkZ)%wFZV^JJ0E*
zC`-;U1^Y&29=T(SU5?H%Yp9fl5wU$+1@zBJHhx8)Ja&DaLgPCP;)#k~eLitXoS&+2
z>w7K8P~qc0oqEo$so_ewhT9JTwg3mpa$Q&JOq&zsI3Et$k815HTN(C^7%I=+^kHnE
ze$qVFAq23!dbs?ca~s~WH@#Mgnk3<czB1(8M5cXfR7Dlx;ge?X)w|kL;PSmqAEhx8
zO4*#f>5-j&vt+-BHppi&&b7IFgX2U2ti8px{+`{w_q4yeq^KiX#$LU<_>K&fakaEN
z0tDQY8xhS{PZYy{+!;7{0wr6ZZpXW~*j)jniIqhFPPDka65rw`x3!S-CwigA<{Q!z
zfJY}7T)4IB-O+v}r;-?WMEs9(sBMo+d|&ii-imQptv>2K!l^BemZU;*g=yKA#2jW5
z_jG@JRi@wOyf3Yml{_-;1DXZbi`lX$>4Jdc-K@?c#*6f^T%Qn$NpAL#>ZXVCn61z(
zWE$J7ey=E;R<3LwV?UJ*>Z5Tr=_ytT@97h~`_>dAT3H$%x3z+Z5v7yG%>i-Y3R@qD
zpe^8jwiZFlFm!8;qSeZY*<LLo@5Y9JooI+nfmFMDuDV)l%pTjDZeVh<cc>Ln{|okb
zbc@7na{yb#`gaAeCU=_1s$du?y7lOHUuF8R<@B3qjRV#nQ%UGBDAl~0!?<uY45?bd
zpO#|4-PqK)-b%5GE(9w1dt8Y>cxs1a@-q^{a1Z%my+WgcmT5=%KOVC&f@c<#|F*p%
zX3{B30O);(dc-D+VNFE$>mVR)XSC*U@Nn8L0d{qm?T$unf@XW7QLext-y0~d0%Ir<
za(wMg@{>%xYazuoLJzzYSPo%cx=%Hb<CQR3K+~HX5MGW6b1gA!WYmAOeWkq;rwA#a
z?^3qnf$H8vZSx0vRduoT?X-i@j#$F>T0rdah?sULO0w7&zM&G55+eJs!}{V5#FE3_
zj8<ftVyTbBS33?$S$kVsrlrM>Y7U0rxp))<bPLLfW6>a=qWip~YUSimdOXHa4GopG
z_tlX5z0am%<E?JB57f=?7ut!vBKVV8H=O;4s?-N%a8hkOf0BJ9$YF@*A;(UsH19bC
zPwO|v7ejVNzcU@b>#$E^9TEA#CB_H)82eOp3uo@Tj@(IsMxKk!&*tY^bRJoW5A9r3
z7#efo+K_z_gW(oA9~%U#IFE2yu`i=;Q<5bP*@gI)*qawo-|!EK>a&XhfV52#6n!mJ
zdkPqm*(J@@mLdB_zkXF?m-n8Y%f+6qdbbDqZ{!Crz}A1O!E-5*GA1Ts>likwu`qCT
zX!)J-xw`lC|LZK-%wLh`fce?}=X)VV3lEw?i)%4|T`;uvL!gYl#C2j5c0GD)!Cmx6
z{bo1AKdO~?+-X1Uy%}jY?3kP{`0Rk$3u5M60m@_f!fAiDNZ5cs#|!B-grfk3TVK!{
z=1Z0=ZDy;TJH4W~zwHIRx$V;1&k3L{$y4SR`tmFW9O}6Oo7sSGi`^a{h;y<wFIsUA
zvoG!lkc^V>LN>qJ%x{T1Ur5hD=ODU*xT+Qef+!!^IP9alREGqiDZiP!J7xkT6rh`l
zF4SOGR&Wfdc|nJPD>WI?-~B>*@H??GUnI1+xjy9Px-Ve#p!bAVs#U#A`3YMbAR0Zh
z9ryAQfzP<*^8JCG_Y9z8w)BPcKs>nF9?-y0Q=+I^7SPvqV)jL_n)vL6eI&BYazP}=
zGUlq)1xfcgukwl)PAmTq1RneH1NKMK)H2Zem^;);L)H)t!w=v+Tcz<~s_Q{rn_duE
zfo(T7@=n69jn=DNG}c^QBcw7oIcgL=q^*15^c`#*7R!B@Qi|QA*1$3u=Cd`%fmXyb
zfwad=ae1<vFzJb6=n2R_+Z4qv7jwGivFH;`7Pn@HzOX4E5|ur!rpfn%Mc*9DG0v+R
zD~O098vrn$W(YlM(uF-Bm$c<HJlq+$$on;*j_sBg5@Ga((?{T#9A@6LV$33HJm*pg
zXA`SkhVrOH-MXPM4_-79<8kr1=WrQ~ohcRu{+mE>(Ym#6ooy4g?!Cko*b%rB*@*2n
z5#6HdbFU3zzy)mF;S=w_O9R-j#&*ZPN>Qu!#2OOJ88=9^;SwZ4AoDvJleV_cUJ~YX
z1^YH=`=UFjY;4!r3oq*r$r&YuRNXJYP!%lPD}gzo)7bw91Pj;6WC!?a3~?RT)WP`e
zQsLKPfGMbm9TGe@IUt;ks$rloV27*C3Elr@pp{Z&ccLmi5`|4TxV<e1Thi-!v^wt2
zAsqc<SyJ?3;(NZ3C>Vdbd@2irrRjgQcLa$by??ah(YWlo+V`U|X#`3{E18P2HmDZ+
zAj+WMhhVi%=(8R!!Uc36s-Fx3)nn;=cnGVL(WDLSIrI9FCIJz#%W6&uIASyo$LW}#
z^}`Tmc1COEY2-c;ssq?s+u7>uK*@iqDqH*dZ6>f>?aG$Vb+V!E;9O+86PpLEAR%wj
zV5%!O6Xz_Qk5V3LjB<wT%eWZ`B)~!S@PQZ9YoEt2{YnEuBFmOSZCTypDcU?k#E2(W
ztp~MM>i&`tZEaj_-$Y*{Dar^DgkwA_bGZG#AbCFbvLw8~x}v6eoQuATsdLMKKiJiq
z*BAa}Lx{8NdsWTs_?=zTbl@q^V1J0ecl_F}M{f;~Z$AbmM55xVB6DI?E9LB`ShMw<
zMeYil)J@3(x&jRf>Xym$Xq1WWHcLNAac8$q>gnv|B;bWWMh%t0aM*47nR$5F=1ivN
zwlBT-znIi}N>pR#_2y2VPEy>uHbS9JiUw}<m)I%RM7+<N)HCjJ+wg{*Os^T7pquL)
zxA|(1$H$P}seTZTal0&-6hB40d!pV`_XE2tAd!3zDR3r`=UH_d^0F-q%-qcZ?z%=j
zWjOw*L$K`;bga(cWfw(VL<3H}-zUhN1J7k!tD1w;Z-d9;s7ZK;0k`UjUf218J5hdr
zfL(=6y)D&;yR%^06GIIRqAd%AFaQMOiai)3R800XSIaf&E$rZjgp$)j#Nb<@wocn6
z!ArfSt;PLj8}yq11SMM)D-Q$A5p~z5rkFI(%R?X>)rGksLTZ#03q{z5^(w(%WXxLP
zx9jU$^?OT!c!TYeX}?2P?&k6ICH&Af2nuF?)2}0tjWOaBnNpBER$UVUB0wN_#GsrE
zc*h=(Ew*rxZPrf$*;-fpb^2Ay#?S69>k%56d+zzQ<pOc8G=|YXDR$3=9BH<kh{;F4
zI8VFb=TP*EaS<o6810)dD@PD3$Sfv#_VgP=l5rb}h3y?Ckm_i3;u<{<6Qie5aFXc&
z!Ege@I`pMGqI>e5xCz2UbWa>ahP_jx8s@#*6<dG@&l9NlUA8+m7$J$e`iSRedjyx;
zyvz0m=A1A1pUJ>>xOdoIiiL-_!;}nTJKGnNxBP$D%c@A~6!NG9nnNfOUYSg!-jlzy
z12Nd)F3$9;>V;@iyg+tPU(ko1C+@YI$ALgGLv|<-f+XaYc_W}eB_n{1n$A5zKfWn+
zP|n$rsD?J7Xm1BTk8mQ6MgcTtxBW#0?H*(ESX5figYr&nF>D~xDjbhSs4<_h_ceP+
zo3YODr(!CmGf~>^oBBb3YUcer5l!94%sy0sr|h<qnwZL<eWU@_!7+0`o>JfBaHMm{
z_S5>t=GAGmtew%A(Wa*`MV|z!WJgGrmuEExxE0x_u{zs2Y$hgZEff$)eI8TVM#QPS
zoeOlgur6Oj#d7zk8wZ{j_V`V(7rs<0r{IxMUx-#I){TC17Z4dBwsF;eAD6vO`#Mk=
zCT~Uby$;_vq^;^;bTAJccsb?@|B<478?&1Aq^vHVyb>LblO+4Q08)<tV$7~aJLHK(
zI1OCb6Lu|d;c|EJB%XBeQa{9)u)oHfH2@9501nxYG5XaE1mpG_!WxGp`$;97O;_cb
zy{IP%Ph-19zp>Q8n)Su>a-nxkckQh&#=YSQ5YcyGsW)Y_Ulg;os&B|OOm^Fg>G9&m
zdj<>7If6Fs&&e-d)ce#-c#i3Rn|_b?e6Bjkz%u61K-~VK#LPpevP9L)6SRyQ=x((;
z;)`Pd^Ev-TNqFnrtL;v8M64JjtuLmhJG1H$yGwQYiR?aP8KLTSMxeLT76yP}&s$w1
z*zO9-DDD=x#S2kcEQ%5GEaDqIWcO(rvWOLYT9U@sYNP)lf(0#(QIYw;xBQa8nh0LV
z_3syQZho;e`fnvjqCN1UC=2&^CW^~|EmJ9G)TqTC)JHj`6I>qckdYAzCpyb@cX_=2
z6*2ZTE3IBAW$F>7HAF3sjn8RarK0KA5iK<7H~CL1wl*LmP$4ovv?uzat&g=!PX^9W
zEPRU=D+BZ9@>Wh;Ycy!%cYAa=h;oNJWJ44^+P=y*MJEktIPEbNO#c}XX@`ENT<wqR
zcNPVWZN53gw5_{hKAo2Bwyai1LOB?;p4g-t|Jr^z7b_pWGS;tI%3{=&S4U_uOEwg1
zKp+-06CB}U{9zqYF*qnhfeTa+SUdAu5i0I#19>?js4JUE^ChT+9~iT7{dtVkXWR7W
zt!^Q-L%%aUtlC8UZtEMho$>ooDDPcAPkZ<f+wPdXc9aUXN1(T0K(kk<AR`^-DV~fy
z8$Js5QsC`gv~V^1R5`s+@7c>SJ03pi9c)PVs|ylm5LfP%7f-v}hF{o!C80SGSft!a
zuSRFQFUk7sV9YL8^J~=^g->Aqkglq(lS7{!fb4i90P))7?QmcjW+HE@?BhS-Pj5$J
zTDpdN>}_>0jQ#0RHQ|&bcapyZDjs^lQ_vo(;<<})y;I$>Dci5JF~<cI%Ss}EN7WWw
zKW!>7L35u+d+>qk@yO!-SS`wRy5Z7?vEPPdDOy6<owbv(Zb;W*u9Dr+M*`98lA%-4
z)oM>odb+yCu6ul!%juY%i9fMe@*agjj7{H@a<82Y04tqr|5Qc(z(QZl#Ej3pmTYQ|
z+2?^Tcc5p&I2V9WVsI|M(0r|NTR@Knaz0Aeb!T}mzf|cuOx-+Rv0bRb{x1R!ABQgd
zO4Z?OTt{(HLtO1M@O2D<RNpQp`*)n}FWuY=_?N(N*f-Ic2gt|OBbLq$YCW35xAC0<
zCvuFgyqHL@RrU7Wi&E1$TxW_BjX!BuV|kNXB$KhlzSlH$Doa!>l-t;{T^KMxe+ZmX
zCNFXp+V$!XF-UdOPWw@V%KZOqy$^`jWxDr&5g`$si0Bs)5#ROkxax7;h=}NMC1=L*
zPvbbhnaSLa>pFj(amJZB^G|1LLprTbPsd|su9=ycnOQ%#pT}H}$K!g!F^Pz<i0Bs)
zArTP~5fKp)zt?rY-(TD3;gO*4XYTvH?(6=0UHA2O+0?7?qoz_hu_xjuZNtivx`=Ih
zKN5IyEJ0)Dx~<Su-dX)|`8H9i#pz`-ZJVf5Yj0e!O|M}nOHn3fj)N<=6`DKs5;+gJ
z-mcmvvXJ~orh1O6{fF{`&gg2_Y!d-Sgfp6El-CMC1@qsbo{_lD^?gkfR!)c!j0<h0
zev_*LWu@EUAw57`z?+2TFs%H|E=j*4=B%=<Q24^c^M+I*^8L1S0h_U{tkSCA#!MB|
zmWU=q3lO|N{BBmv5dh^OkGc9nYM+J+^E{eNz?{3_7Ga1-=WyJrGM+4o(0*HWXvo$|
z-qPFE2j?>GP&;h*(NDVGg=g}D;{`U~Ks)iiUZCHom&W2Qt<zv1n#-v1s5>%%J*r6~
zmZ3#cXI8Iv@9bcYEHS(#JiC^8trv2jqCp5PUAZV4{dYvzXeQO5VNvU%XjUDv&qS-H
zqWA+Jjg{IR$aV|EQdC&0R3jZ+Nx^xa2S$*e^hv=@QgH9Xnl#oJ|K2A28X%%o{4?Dr
zFNb!0b9OMV=W^^ctZw*-@)++{dt{=J!@=kgnw3>cW4(GIE+UDb4R%9`-GFFS!^|)N
zB-ZelGXro8Zxj2cFc)PVMpfQcf}o8HD%=-ARQ0>k$h|P>qm}y$LulSP4+xbOOcWjz
zgk+Kk8{k%B0~_O^>Y5W!1Ah5rJnX_6IGZj0y^_#>j~F~+AYSC@6a1(*WGYJ#sCs}k
z*chVGj%{j-A=GGW7b5sk#t!$Fx?z}VVuR}-tZa&1hL9&VjlFie#}bVmZR<$v5zN%4
z$6n)h5AgT-H#yc~zm|jGT&*09IAEZK$K<L$sM&S)kI@O<(1(P<{cGc}?-_I%m-Htu
z3wFcO#zk>N$P@bm7rh+xEAF@*o7bb>l*dx06vs5-b&G25SscezRFg8{kvQSKNgP~y
zKdHIt=h{4_pLO|u^o~fE&N$;<YS=2Tc~a7yYOuJ_p7o5Ddn{ee&Ur)g@Jaut^J<N7
z=ka*MbwM!`Boj73xDZ*2iz(jHSH7Dt7k=B_)?tHi!3|pmyGN136FAikxsmY>cu%ve
zXgv9z6#5{~A^t#cGTvZ(C^)|Cn8A;X>^j<xwa3T$L~oqp>G(u&@bN`yaZzw!#(UyZ
zH)|g&A!$DIXXX!rwtudg%VBNf3(q7eTSnr`>LzA^^OZp;aW6)CpY_MrYKX~mTYO`f
z=l->S>N#>nsfZh;H@-8>zRc(Hy*{9(NX&mQI3AJ&?fX#++5wsP$@{P~pE_KV@mRV4
z$FXJAuV2iK%O4XB%ZM(p4gV6;RI~&N=prl+@W+PZzVfj`&C=?rgju}mu|h^GlVc>6
z(r;dWgbNUe1;2@F9}{uK0gag9uHJ&};q?LqT6wJd<A%rdG9}q3G~mX^B$cJ#G_Aks
zF}-b@r(Wvj%1gZ=gwLRHZtjhWKRVEe#Pr8R#9Qu;83wBaTQdzdO{{SKo+W&EM*dyQ
zF#zr@d^v<%CG}#x$W!MDN~M0VA5Q8Pf8pmz-CNz+#>u!%%{T(&#UotH?SfxoINYJ%
z&}CxFxvyljV!jG`H(-~yz}RNA+y98W3@(L}|HbL9QBUi657!9-7R&SIMZHIblNS~v
zvBaOL#Dg>}^H^F^u?a)k@K|AnAV_Q~40^kwQN6Y-Y9^<6G`W6B^ujbZ3%{<+ZqRDP
zG5F-w=#K%&d#qBe3SLWbpSDfDyGn?ISna|Lx3I?8ah;{%&03Eh5kRziCeQv?bn4Fp
zVZb)W)7_1&bWilC1l{pytT(DiHV9?CA^IN^5#y4|q2{3+aATfYaPvb(d2ug_=wTNj
zekTt7sNmvVOgL3nTb?r1O)97~jBScZBaftEn5450cv~(;LXV9;14fSjppjaFwO9rg
z+bm!h$gtQCxlhys?1P7OOcwo9Y^kCN+&{-727v`J9Sy{z#^;_z^$TOG@IhjRJ&y6g
z@?e|*BipM-x%!gcVceHMR(A>rFK6CsyVNjEVUuax?S}VpB_l5wh&@`076Oj$6^yh+
zihV|tq*#3IS0$VX;(!sku&0pogND~Pbz*Yr1fq*W9szbPBXQWJc_sBuy*dytYu)}f
zZ_^RM#=0b4aodruseg;3o*KLyN?xsFsyD&*IW9!gwI)s&(ebW~mwD3Da4#c2WoS4x
zt&h`28xw%Eq(9@b!sWas$m<Ateb&%8QG{eT_n03zY5iC18{YYscm5dX)tHR{C4Iw?
zm`l<``6HDG@s?q=Wd^T!Tb<M|T^twmvog{h?^dUIkeIdc-ec!wZjebqlDd{C28H2#
z|I&s>LQEuCj_U`Wpv9@rg1CGrutd&`kK6<D@sD*0kj@n!{SFo(K6$LXb`ZVoD%y~|
za4uH&Av)*OVSVa>6eo($3~hWYKG*y%PxZwYmllZLxm>>V+%IG$zw)?vu5#IZ?Fr-X
zqC0f8#i0Cc^*7p#@6?2A1+ja4uM)yJ;p(Ve1igV6q!#=~Zz~9u>}P-S>JrJd56&*O
z7gCfi*gOEY7s}P0TCU5t>v_n7KE0TzI&HfM$luS4EBq5Gbmexv<x>8cCJa}(UD+1h
zy|lYpT~yNaH?Gk~4HMiK*Sc*gRy(*=uG=oEo6i>#DRjp5>W9UM8@3lxh*B;63U74H
zJgfJ@uf$D49c{|1ar1T&MjS2*V^;3Git!S!o44R}K_m*o5MMKlC8u*ZW~$%8cI;FQ
zIFnS)(Vy21jK$pTg%*=jZ8XL7^DfO6S4Tfup15VZ-wwQlIR5xs#I4&4qw&jsh}+b4
zkp=uEZuel)?Tnle6LE*3r56u$9G>M)VL{+ga-AFRd&}*;KzQTElA=^0ID`C?HTs!>
za_fX^r55r)u!ee%v?;}>OI(K3GF_dQxs0rzcy$=BOl8G(f+lLc7*4Syh$b~o20iuh
z%|fzLho-SrfJO)N;#sM7u({|h2;jI%g|#d0X>5)AJe(RN<bAQ)YnKX8>g(~lM(vA7
zT?(+)df(uU3dum5Xb`!!qtWTF%qC1&0DaYTbthlaQyqJ5zxBeGa8eV|uY#!)sYDp~
zo7Eh~U;(MB|3aL|t^vXU_r<Upkmi;9bJVj~q#_$)+!Ha1E%PQ9WzeH{{zpuDS=ez8
z=w~m+tq;0EH_x)o{=IPmhYq7l95)h2dt6+%$o{sdsJ6IBKH^S>#z>Zcg^g>Dj-@B+
zR@Efd0Xt-ymm>Mvlq+m^%XEu@3hnUBwvDYT`jed=H&ncswLv6>9ulw7*zMBJuCYzM
zQ-`@h_Ncw?8m^zc-a;FQYXCPg9Q%Zt#jg8>TFVf5wc>!<rf96ZmJX^aMkyIA;*e%*
z_UhGf*fUgz=g!MMW`7ee>xhdnV2xK?T!3P9$B2)*xUsD<j#cLks=#=Td%C+ekq+R5
zI=rW`E>5b+P#bPMr#zb_bNgvG*pPm)XAJni{o>d0nhOYrI1*=7FiiTvIOjqB4@eb`
zAaUONebGS1)*|2+@%G~lHGr5yCxu4km($}d{dqF^zqhM*#uzVTgoAOxJ-+j=7&pAD
zg<%?X`y7q;)V8~e?eV@qHZ1&Ye4s0-p68${jAdS?k2Ed3^zwYg$5j;WicdUY3p<PA
z^`e(3U#WkIPmKbezw6>N!Gmm`iO==B?BXv3tev5Ksn(>Jc!_v0Um0|Uet;wY+7pAu
zF6{^jIlydvqsC*zJNs6Rr=G(Ep)I~s50zFZ<@c(D)GK~4=8GodNAK12MCJ*ARP!fK
z;b?{dNnuPpUI^SE*TH3v7it5enOeBZA1?|8;{W6oEze*KYgD}+-<c~O7vWmmYrE?!
zA1@?`%W8O{g+K%)a?e$2Q2F)hxLWWlEs#mP{~s@eVcle0`*<P81baX)blu~H`=}sW
zE;ek5<MnPxr~+O*H#~m680IX&5A;SsKYMI$-1N8}ifNOKBizl#j>{jPWcZ4T%MqHT
z;ly;ogbwC~Im0j!BF0RW3~>E$cg#``vuEEQbA%_sAo=@TgKHxiFh_iz%Pu^PjP4e-
zdHt#0xb<;S#EcD41pgy(n@0oDB+}XK?s*w|^p3~HR&b+^=Y!);l_3ql$x*G^$GDOR
zk@L9a*kzh?j_wiyZQRVmQR8CRA;@2nsXEQ}tY1gHfKKo)mZ;I|&yAjYI+m$`=|iVz
ztdOuqqlX5ogGAKJBNv2xes@N*hKsLRw7O#jYD|}xm2TaUMir|(bciRDexKS^nis}u
zuU2MwLc&XA95-w2L8`G<^Rh_B=>!;8!;aqG=ybgdyB6J^KP3!@@Okp#QrW~sQncRn
z2hvWc-)p!KNh=g~z}OL(i4!&CnmuW<F|49mauhI(QMapOX1R{XT@zwU-b9zBZUnAo
z(w{wW#5}zo(7g7I!m&N*;g9yk=EwDrsQfSgX7vrcJuR))$smD{>tUmZ-yU!7ErK-n
z;^wE~5ig2(<3XYk567c|zMkalTa8~icH8{1a2ZCO+l?Kl_I7p04)>6r=eTU>Yeeba
zc6mpvL>If;%@mCz_V|;GCHctN>rbN1<<Z)wowxdTvEN;%1#BG92mB9-gbxb%V8%E(
z<mO~Z?`j*1!)o5KDPGnn(-KOWBpj}e7~_=o<F}#R2k}fFfs@?MM^z5t<{-o|gZ{tA
zGAeW&_ormyal#|qfVSbJQB8Z&IOT%Id?&iia>DSjyU)1Z-I(gdYp&N%O8T?@y#%xI
zbKaKRq~yE8x>SwQ*0Cqv@Gj_v>IpUiiX!=c%Vz>2kR<JByzRc3*o^LT!JjE)g5JEV
zp<*eTW~J}B|CFY@ZzPvz+TjmWErX{)bU#$@E1G{BAE^Z?IdIvi_QYc^W5ZSNB_!JI
zxUFB*nsjc6PYt?_ZsRkzs!J6wib9uR$~PWgc%g{Tzy|u#0GE>zCx`Nt`o>?CL{dJq
zHQ1{0@JUJkHy#Ya-}u&h9U={-#={V{@H>rSI+1F>7v#d|rP<97!YR%7Q%>}w(HP1M
zH}R7ZPFGwueC!YfK>?mmj>~owIy3cB#*1@Nd3FDLOjA*%5AOPk9ffLE#!ITYD^(HS
zf${*mYDXb$xO~E~U#*V{!iWC2#vj<nPsFwUXy`)ge4Rjyl4dA5;Olo30zK;=e;GFz
z3!)}xyty0I3R4*7=%yWo8Y_HXzVvPuj)+0~qoSJWhlfm50Dm)mM$Ay1{z2I8nHonb
z^hOYf>Wib6-mr57ykeol(U>cM8Ke+r#T_DBDeumnZ*lFS0L)10x9$)TOv;?W1-Tb*
zGyIfq9*f(B2Nu4h58hEon|5`HCUvJuGss@d*Y8Rlyz>H2M_T{kGjf-@#po2`l8aqq
zJeu(VPrt>e^HOCj2sDG9bgmiVr)ms!7)F$ZTP7^ZHG<j=92?a0g}n@@(U4OpT{1Z|
zd4!x@BnJgTvoUSJA*t2B`{1Ql8tVpp07v>`l_vsD4^yZ6T-ZV^LN4*u{)Cu}Lp|2)
z5Sd7sb%tT)9_0|kS`7s1!}PW5V?86rvC-{If|9Q2HX6NXfqFDx=@q#t*Lz6mOf)zF
zr`W$k0)-ZyL1QDB;1|mqL10rn8bcm)aYst~p|%({5EhSj7Web0=NNxE5)5nORTkY7
zn*^t~g9r?hUX|vaw6A)=C}Sy%m1aEXv5n$65}UoznB`&*_mIcaiY3Uy?x_(5DYm!(
zoBDV}1u)D!X&*K8#;%Q-nRBb_jkfp4HUntMtgGAo0ap%LRHJ;pC8xF1F#7S0=eX|j
zG+cPDTsoPYWw%De{gk#ydyMspenN8cXxZxqcs!31S3Re7pFuVgcDuhj_IpOS<8TZ-
zAS5Rz(j!CLvVvPe`isJR9nz$p{m<EPSQU#M<IAd8VkM@5&_o>ZkS1aL<CW?d28oeE
zvSF2V6{X@a<99D&9!s*13!!|x@z6T4Lo9kVRaC}TI%()Au}9;SJ}{QR%I>tgN~>1x
z>NCQ(wsujx=3%Bde<IF$EpYXSb34l00Q>**s*@sfPV^gU{D=Ez#hY%-37mQ{-tyO~
zRNcngKD;F}0jqL6E~p`9b@8sJ5JM)c*>GZt_uO~|q4tO4eSa~H-0mNEM~|R0$$Z`M
zp;ohN6N>$OR(vG1m#3!jWBp#al)-)CRnG{<sSLm9;iWv{)9RLa6?xCU6^Fvl+z<IU
zB!u|fXrOfE+WW%LYDy?UUXhoep+a=cUZf2h;%k4wn<2;k#_dvjl@I1`)sDOW{`gL1
zY3p93BHwG(#zr%yzz_PscAQE0Q7}{Wgf8o+>c|MolLCZzLhrzCc#RSb`LZVpMQltp
zHAXf1s$i&uTxQx6mYn)Abj1@|Y1|o~D5Me}2_A=6J@J1q9oiLFKOsU=2lth^hZ)f|
zPw1LTBOnijNC4^+&M6hi*F7P+j$Si;H?KFoBWZ@51GvFZn=*J6hN&Eg8=sKaw-t?}
zG44k^lyat<Rg{CNJW)7=%0j|{O!x1lO9{TJ-}Nh+m&Z(1Cl*J{GFF+1;jb~rSe4KM
zZ87%=z33Dydy<akX+qBZYkH2~VuawnZWWj$=Io{0JiYy6nJ4>pwIPu0)T?oaKJdnb
z;NI!(Mu{Csv-v7R=)lJUmtlaCjL%*EB=cMD+#2_AH|K%e#GXBj%~H39jjAA_ROBLC
zqLHU@ZY<Npp{326lThTUh~)zE*hYiF;%rwDk2ZOl(eq&AKN`*cMz76&9EnyVMX>WW
zObb`~GX-{RvQ^cgrWEEre~hF?X|+FEcoxMP&EjmL7OvHN*P;BxmYf?h+6A&oTsplV
z4LoAt^AO#1Yp*V;jUEBzP0OiYFHG5d${W+~`Dz%<NDc#<krFp<NIle}0nZ5;u8slu
z2~XZpAvfOD7vnBz#QO&_wW&HfQeMYo^=PD^IQ?{q42&hk18zbt$3i!~r=i5DBsvZg
zw^^7>9mFE(A^omk*%%KC-@F4b*Vv-IRxWI~XHh(&YPr+mQB`A=@x6?#`YKVh5U{Gz
z)N_!f*zT?oF%P11%g5*r!67?XPwZ6PlqwNnRR!n%Nc!O29_8H`0Dg}r5sO}eIPCS{
z;KdPR#Xe(S$0Y98?62#>xNH<cbiieIGqQs!!*qu%SlrczR8~?F9d>Um`51k<x*Ij8
zC*z1(mUm`aCB0$<xXtsTKkB`qRG8gxKF0)l760P6QD{U+enR8mL6;G1Pil!bjgX{R
zm6l@?7N-RcWpzT&`{zvcltK77lD+Yo5UMc!jB_;38vQmHp*ZKsVn1%|jPrs_(38=4
z!>H6DD|=I&r&{?f{jAj1#@kiEZ4*WKeZfN`@L>PM#xxOq*ELpJ?ekuBrw!$MKi)Sc
z2(3262X38$k~A%wh!4G8|2Bj<Y`{m=8P8pp-c}zA(~`LN6M@8<hB>?F?R4RJg(7A(
zK2<qfF#GZ|qk|a%Z;SZc)#06(^tADeF9aI1!lUt}YZrc%NCE!Do;t0jzOMF_adt3X
z;!oeGF`(APx1Q$}tODX!l?<oP#P|AL>4p;gU?7BrF7~S(6Y-;pAviob1!Q^A=tt}<
zv@#_rY|zVgilB_}9=m*Jp<3MWly$_korQ3gfTVfKu24gQPp9JT%AKOYCE-EpG_KlN
z$Vs`BPD8j)#?^ucJumBg&CWvM77Y_yPCutk^u%?B+FznTUvH=_>^OEMqSu{;TL0oV
zaic0?O`1gbCgF|sJ%U$*AV(4<m7V8HHi3MubJF~}n!dBpdLja$jGiH&zxw-_DWogK
zE54W+&2q(|F1$|XXatoEgUH64t4_&{1+7#)q;d3li>9gqZY8tMt!_s!ggjffX}Gih
zFgI>DuHDE6(LlN#FC7FCxG}1O4_^HL9*+4t#d-XkyD=8HaSIfwC+;$)D>LRujXy?{
zCfDg>cr4k)`YLYeWeMU-ToWpII+h7yaw*)`4Jv{^K%LpBqSR6|^d><`(R`6=HXCXK
z!jyawwi;@~qWacYsqt1e4#g@>^{?-b`#dL@JESM^Y85P)8EZT$D9`JU$65hKa+pSz
zNM2nEbBz47Qxk-YrZ2jU!a}kcr2wkO9i*_IYkIvKvH_Tdel<kagw$2Pu_47O$3SW^
z7)ta0VF7{q7Cqjm_eV)yRvy{oYRoGJPe%Rj@9iejvF8rAYwS=p0YsON2laD|oLrk#
zJ#z`hLk3J9PyB20newprEtgp82oXgkBfLdz8WuG!jz`o5M;hAos6ONov{jvq@bKT}
zVK2t|ge)CuUb&mZ!aF=asU7C|z0*Yv7$NQwHXIp*Hzk7SZh`5YBq#P@?D6U%)M}5t
zUR`L-KEwFsykEzD{h8-b9Pm)e)p04I$3X*GoHm4q`~l~H2ZzBlcD2XLhQ7EajtGZ7
z%rChKUa9V%;#(C*Rqx)Vb#crksdwb%bzGo_(6NoiiE5|m6mwF0uB9<fsrAC;O>x>a
zNCA&>(V!NiQ3A2oG%snMnASGPWX^he^EMxha{_};M4Z<c^3@h^s1gQfX|(XBwipQ<
zhVZZ%h_}4u1_@f9&ZTdA#IHXU7kvH684Ei5u8|u_xx#x!4uMGzQ~16wh+;fho>cLH
zCvpf;<A*!-Oz9m!OOTglMxgk}9o);KpBragd~9T32)Qh<KNW;jo5n?VzHAVY+o$fF
z*JDOi`K&sx8Q<k|f83pM5x&p|+D0z4FZF@;g(>>Vb9Z-F1_S$AJ&g?@V*SPd+DIM~
z-}<ArZ3L31TC>vD@O$Hm`$BIg4~!p-5?oB0#QeC^X%#a=7^c3^pp;^tjGqMK^)noZ
z_>*L(Y&E_v^A}GG7UJ?h6@pibLq%uadDH$>2o}i)<BC6>PYoae`!U7QpWD{s&HpF8
z^P1Dm4G)YziO4KT*N$t{q_Pe&e(j%hmvg}2LGyUWb*?*!IwG$3N8Y7zgMntUTdYfO
z^rvDj#7zPV#%*!45N<}a%w0jM#y|NDOLUf^if0wM)~6df5etT5hQUFXk~+*3IAR?4
z$1Dx3osbf9++>uuRLpggV$-uO=KV?UMn2OukbZCRU(8POms{1ItTfE)Z9)?ZR<2cI
z)7<Wv8R`0`xI-&|Jw+D+(mP#J&y>#>Ha!FqS)e}?v4Q|N#;gyL95uqMnp(2X%SnE|
zG{C8%jqZYmOa3H=W!)etMF_DS%LJ*Hi?Tl&gkyQXbK0X(U1ms>XfoK*GMZ+Cg$>TI
z!>vZSwYDi%3grr7zg2ECz>92C+^2%Hg;?$S=qqE5Q6dZ-!FVAeYgGv3u4vcqsXz00
z>NKkOvhgVNbdp1;NC$iTji<xZOIVYx_cC^MM8AGEFRF_H_j@;d<dD0`&^p7O&+dL~
zOh>(tscVIMf+~$0MYhvFzw<vK8=E}FwZD%^{T%GTebLj$i!?W0Jm_X^9S~#4Kx}q5
zB#*%o)ff-qaf0PiJnSjOx+J-vEpEJl2iCwS@6<;;e9S{pGkRug2YLU;Ry8d_cUE-A
zwko8-t`Q(@_h+$gAcdp*O=VUxEjvAysROY~KWkfBVz*H(dZIm=s@9B?v)5DD#^%|l
z`e{(J-<^P5hLAqs=CyeV98?ns?tjK1HEG9W^{^Kpl>#rjewy*d5yK*y4od!b#RwPs
zH+Jb!jlDKmu>Lq^^hegmaRHGa@5G-9&pdB_oK$@zRT*6Cl<N0R{XcP9Kim3x;*6SR
zhJ%rKO^Csj@Zj@sTd|fpSKS!>ZS4eag4;UpQf%Vm4KEm{M>p=9>NmNT)+O=QrRxD*
z40+25UgKeo5DAJ)=z`lL3>_}dfOkD7M1DcBsLFZ!q|N&KuG|VOMRYb49|)5)o`?@U
zVQu|TR+4{xBtUHgY0UGnutE=)qO?yufQ7g%Ty%eon|N^kAWBVteC8QU>!_3Qxkj?`
z=ioYDxN--!S@ES&%P1gN2cTutN=*N|_}U}x?qU;Ta;o`X2v_&5QARE_lx*yG>XG;1
zzi`~X7nZ#laOwv&N{5FZ)d)#U`XU~UpA5byy>y^*y9$+C+K63!Yh1Rg&^02KNzZb*
zzIpWI`NG1+-&3cGwk)pLC8|b1)-=<*a#x{hY3J7$SM4g4tTw%0i`EJ6eRWnsG2^(e
zuXHkm5x&--+q%hVlvHn5A*u6b%!un<5hEN(*$wV<*%)kVXWY0;mo;<JHuY~@Q#b88
zUrcaT+|wF2?-FMNQ`4j8AY4|(;INk91vFhaS1uf64rl0t(0?#IGc^E2lo^Gnu=KNB
zxg%4D&vDO02e$vdP@evq1u;(*rw;Nqh+B-&(8Rj9Rh@G|W^|(4^p!Cq5f9z&k+FXX
ztQ>a;20BAj@po2xo_eW2=DQ*%B5hF@xQs*>U2&I3mxfV&QRA*hxjXBsB^;iPs5cCZ
z1(LN~qQPViBxH`fe3z-^)T_K88q{*?6dLyaXw+AVn=tNd^17w-PR?4hCV?na(JBzU
z;V`xFgkX~h$<qfR!_l|jCvXUp@<w0nvA~Y}ORO=Dj5!4fYcCBuy-1_oJvXE4$(}pC
zUga&Jr0?CkN^r(s|8M`I$B;;~kpX$v?-HLvq-%n|z{y0v(P$fi9~jU$(nC3hT(G`t
zBSKPx89_P7utxof<R=fyxMrfV{I{{mTcx3GV{r?LNrA+ag~+J_nR@<#c+i7D?p)lX
z-}z%~7IwwwQ8D5ls&<rCz^E7fNwS&TZd(K!Q8e^Wzbk_o#zwz0XLD@zFd9<7yv-jQ
z^J9B;mlTKg9iC&bX39Hbr(m4>S5V!hX4u7W7wuLv9tz1T?(ujRZh$LH3xK_z3;bF4
z8C(3|A!M=NJ-}omdhUq>0+jK7I1&d9C^gr;DKbCg>V3T=d_{qISO^iW4Ekuu9cUln
z>W);~W|U%PDqhjtVnLZIyS&ljsON_i1#!$HUx*ZQAdU;;UZmP5s*{!GCnvpEgnm8^
zPN^nwtKzh~qyvR2IOE#zyVxyekFUAMWtdi+72wPZG92f26<S=^5a(Tm%+m3OkZO^y
zz?<G3O+TB_zop9DZxBntgK}buctNv)mK0ehi8c%rW+>kC?@agmzN$@-n}%$G@qus5
z6gid$T-wBbXiyHQ@{tBi(4+X+Q@)`Wi%3#!c&3oDjKoEM)TBAvr&_m^$ia?J#AimK
zfn4UF8wiT{3_$aRz}GEW6<@m6Qq0o7((gZznIpdTQf(}vi*M9m1Mkak)#t+8hVh+R
z5N3W|eD6ML+9uLh=?4#iqwu4eOdZUKIzJib_OX$Ph$r;|92rd)gv*}PldVKtzT6+M
z0h?mlle#__82(53%(>#pLcA*6k5{TDWR0zI)sux@qIFG|skr*d>SHhUwPizI^JJl3
zQ_r#0u6<Ja$b|}C9M=gS1aQ0H(XJP)3If-D*8T<~E2Fg=1qQ-RC>tR&55>~W`tyJ1
zUsT*aqa<;OOjoS|YzAjkOVBQH#AbSU>|sbXOi9cV7=&pAaeGqq_ui?4DgGn7h7iFo
zG+pE8xtQm5%F<87EyfInK5kXj!3|W#?B>by@*w)(UgdC8x4RKqv$(_9A%*C}`sq#;
zrLHXIdk%2pB+k|XeHG4UbKF%$r089t^))JP;)#d+rcTR*H4ZFBz3XPW=dzwl3>f(c
zcil35aYT5+M1vP`5XVJiacN}KD4gK^+M>y*w@l!VA=FVkH6{UT)%v2`O6~B;Sg9rG
zhHkI&(kCmMuB!JLqx9P@aW+>AhZZ&<yL*kQb0I3o##(hs*xRvaf3n;wzn&YN9z`<G
zz0s`^Bc_F>_ZU8gGV4kZl=UjjFk)@dFK`*pCI$>7*P4*xLU~<BM+~bHVO0BKRKNT4
z9vXLJLL$Z{;h9EHgl-;=NzWbg%sZ>7Odc@&RK=3KSTmN)Ov?B-3#*O2=l(bu52^1#
z42B-QG?Q#HOnbJd&vQ?~ygcIVJJ7We7LlIfQFW2xRKB(f%)*-5*k*K=LoRwa6qIh)
z?ET~Yu|s_?OyV9p)j{Q6F1=kUsN9|8Yj<^<rE_Bj!r9|L&<dF;v~jQZZwha*A>F6e
z$&Gi$eto11$v_-XMXpTVAt(&pu9#(<j6?oJpaVUN!}^457C(TO-D_`2!wcjiMsi8Y
zE??0E6wVI=8#T?XmIUFL0h9D5m+ZLa<ZonyapKAISEcFiu)LwlmOtrL9w1QkDVO&S
zC~ImmL#H*%<A@v2ctL8q#-~mc{_8b&iAZ^zk)HMB_IL4|6sjnR^Ugc(k`bhbZx~D(
z-zZ@Hn_8<T-jNzDj=k_hZ>#GHdDJfWSh2RnH||}-AfP<cmKMS98I`G*V1q~Fea|Du
zY(rNsOvMMr08_;H@JYQc3d32Hc3ci031N13DgZzBv_Mbd6N4sfO^))Si^dVqerl-g
zm<oR8-~F&}p9>h9BP~(C5cWLd6k_~KuPLf#ENu)u4VcqyQFS}3HiO@|Zl<w~Z{0-~
zi8a5|6gYT(@`m55w5=b7_z!B9S(FyVk3aA25*##sQlA-mCvTzHEqXQHh08Yd_cDLy
ztvQB`-Q~LrJ;l<pjU6#<x89B|86v7JuGn1&AL%2ZnX35uL+Ad8#OSKsqA<m*IsFB$
zR#hk;k4Qt2S4v##E_tpYaKU3^w}>{WJ3Hcf1H$G62S)Dc8`Mu_m@MC;aic&rLs&A!
z&Q0oNGzAMcyO$N|=|y^DBY=KBrt3G^wu%4d-WKNrYvyhd_o8d~_u^YLtJ*FVhP%tP
z+%yq$g$0pBd(eu{b64Dc7~sV%sy65TxYdm(S?t@4_z0pH@-eze3OB*xj_R260=?58
zX_Oc9T~Wy(V}U1YW5(X#j=0O;D{<4!ORi?O_BRA7#pLO&bu#J<Ly<_LUhtQb{Hi~e
zxHA@<AU2-uPNLFm$k-bVMw$mLST<@nCF5t4JLZ}%!rf-Wxf{80wC*mv)Nf|Yij|(y
zblpp;iFRR?t9EA2`uqGbo(%8m-MXc!p`2iZV~vq#2V@TGwbcP**M~(&v<oNR!-LW3
z>CVU--Ts}y4|+U`(J{h<#roZ1{pyEN_43leZ$ouoe~2q;K;7}K<u1TyQAH)>*RW6(
zK^m>jDfaPP8si%A+Szksla{}8eM}mMCJxpEs!;j+x$&TzVv0xp`Pl49R|rD*?1!r8
zqT(f4qiZwK*hj<`!9?BE6OVZA;2HYkQGax#dUvZoGLBemGlXTS>>Y{i!enTuH+C4o
zjYNCesa|=6r}5G*wP69^dq-mTZtEW83P(&@?lUrLxSJ;4ld;$1C~RYzI_y(NwJVm#
zexrcu3LE1%;C8sckH$eGf#xs{d2QewGFm1Z{IE(9Crrf4LXq6&X(MsOK=OdcE1t8|
zz6{1ugG@=_u}k1_<oEA}QF3c&nFib9q_-%%1N4KNZ7@y=HgQHdtEb%&Z@#`bqd^hc
zAzl--wvkE1sfLkCRvs&PwVv})OgF!zx8)6W-V;&cJRn;&+9U^nkiF@Si25)ZZ+SX#
z@yMf&CQ;pCPc|ACbWkqt+<fW3c-JUkqdVqT-+Mw9x8!)=;4*wj>dxW=A0(J?I4lfN
z-FZ_67yC$P!g+BoY>1Ct7sG8XsQ&oGXs?9R#-c`FDXN6G{?ug<ayHv%hFUWLo&8)R
zz*GfqeosVene_P5FfwfOw)o1eRwCSquibflM%<yhZ+4&8n*;kB#!`)1ko)mlZ_|6Q
zg~%WvUA>*@^6;I2k7oS)p7`Df4k3=fx<*q}Av^YuKW|_n<RZoS$=4BEf|f<e8u3)2
zPVj8(jLY;<smt&ump@gg;na(exM@!nvdKi}BjJjtL=7RjvgN<>sp=yVJ&aPOT%{I-
zh2&KpS3gzmC@4_TxB%lCK_@dU88@C18D{j!u3q<)4rk$Ma(f8V>z^u@Fno91pr2`1
za-&hkM1}l~>a-ZDlAg+F+C~9;RO19lWRcT_QzcDrAjNP;GlWyo>&|rDjYYIHOC_1)
zC*}w&p1EDa<sCHFfbyY&9TW>RPmK`(u*PwVP{i13{Yc#Ult=>NmgE5n<9ger`RwQ_
z#QpZ_g#0s?&(XN!DKRyKL?|5Hox%opCHx-syBi<ASYVuqor&=KE(2LeUQ}?ZF%T(h
z$ho2KRPts}`lvS&_Y!zyvTY;LaWu*4V7Tn5!YK6O_$Z)YSwsbms#AIU-=fLzGjhTR
z=daldGuhVDAFcivY3snl!)Ovi63ON&e|15hh=`+#Y)`J$)heo0?uIo&;@7{AwcZJv
z@bXu$umyQ;jz*`}0#>Z5Z0LSUiuS)vbwK6+h#t=m_8Q4)tiLo}TuzwasPEUG7#rtc
z40!#QqU0ZP7pXoX{4SjiS0SaBGBQ|wLW;}kjtlR_O|i+x1TpHaG(6>mPkPqb=7lGZ
z2Mi1TEh)=`>bN=eGn<7H&MH%f$+B^2)9~I-i-Lz$m6JRcTRf>)VxSg!#7ooD)n6>U
z9@PlUZ>Oi%R!w_HQ95iB3Yj&G9Ua?^QGLmb5<C2fM_Vq~POmIlrvBLFvh=`1+q_#N
z?_U>tjBGt@7;XZ@7I)2!dE<`6KI1!tr7urVt#4Z<b`*%;&yIuo<3e2V4tX9?-sZZe
z)x=?=u{;@sm%UmwSoa*!Z?vGiIb8`W5NN8q98EvNIOfSKVpmnmJmCXzLe&U&ky?Y3
z`YJ;oiBkqqyJ97VJ*N%*!oPEeo$>hcN`pAP=CXIT5b5HqKN`B=7S3r{zx&;PkMqKT
zP@`Mp4e#rR3)_1%-uxL}T7$k-?TeihZ@a8(Q(i!E!E?QE6dhPvAHM79B5o?TJ|bYA
zm~M~A6dk48JB-Q)zRq}z=NV5@DrCg&()&kh4UdXo=3`HG1I+9v276BiPr2yX8G_0Z
z5l7Wne$<qzHb}tw;rLuDe9ztSg&O3VLU+G>>il)Q@Ht|aQVv|Qk@?Cf;$DnMCT-Zh
zc4rvZKm@-r9>vD@TO+suRrYuOo!Y+d)nvKy!T+FjnEk5-@uMnb8mIV4bAv`EB_I)d
z3LWIs6SgES+f!&~%Gk)_xJN|HZ<J{&tKir$5LfIG!CcqXgBX`Jx^j<b8Ck-w+Ed8v
z>_6O31b{uFqxbgq^$tzFlrCe}sJN0g{?{625qn)H5GkOzUOy{~YvTs}#5y@{6t2jT
zk|f=vw)YI=qw{9t$|XDX=eDTqDc5^zYfN`R3MW8whM@IKJr^@Qg!1nBPj=lb0nC>)
zdPuihA~){2LVB#9g`c;_#H(iz0RovC)h@XiLZNze|M<JPaoZkIMiPLt?RaFvA>VGe
zQ+wHLJS#P%N*YUJf2zV-?bDd=wMmxiu~^`_f<hAl#~&QOT~TA0FoPQq(bj2<9qW3d
z-rzg10VfJh75o5_LL{z2l{>v78eEYGtBj~o5ZX5OMUw_wbzO^Qbv#&d7`N)9w)MA7
zi(;i}mP<Bw)~f0>!Hs0v*!%YARaclGmV=<EUI7#@N))%oqgsx4eylZ0$O<6X?fzO=
z98f!}YlpKbTHlk=?at7+luNwFQ-Zxqy6~(w;Fe?q``tLBn`S(N7|=uzMjKjaK;GCf
zY(RJ8+s2s~6`%>!fIKPp=qV5W-D$>OWW1XMZ0Z1jCyjJBvZld!K(k6p>v+&8Wbo#Z
zw7lAE1j(t3eU6)LB>FXCSncHpe;r%ApEe*$!JVY=E01`;ZRC7Bs&+{)p&46!^mq`E
z*g$l!x=}Kw1U`?(0?~C!O7Yt8;1`v2XBg=0`T0?K;XjH=8&Zqd<qZapv0GhY(Sv9=
z_84FjUNR()?25gYCLqH`>@!9^6PyU@BV5?;3Ptl62RsV{QwN9RppYfxKB6OZl$wU7
z-^StUat*b`%Wlz!@-dEhc*)?xqQ7F;16_<hj_M1dmy*+Cp0<|Ju{dsc8KU8Ym$X)!
zRt5?uJuI>BO!O&tbk9KHQ%)P$5Cb}6gfLI*h}Sex(|<QR&Z-W=V3d+oJjZy;PQ-c7
zdRsqnbP$HTp{f%biTLuSApBwW?0D<aToP|E-Zl({qmB#i7Vmwuk}gBZLGrhFPw-HX
zV6>{=u#RWq0}l(26s(;=<U{{qfXpz%X;AZ#dLP70;A6pT?!p6_o&AY%9$_n{Ve3UV
z?dX|0K%|+m_|!N!AK_=h5BD+f``pN)K<<n${E=4U@uk6VH6nRom|v+Mn2wV1F1~iZ
zLv0g9*z}FYbKzOW^Q|T*gC^r`_nk`8A2z<%?@AL16Mj&CEou4`Kk7rW<DZN&mt#h+
zi>FQgP6O{mT=sM!jZ-h7i;2tiQGvZ2ifR66OmXrRPfNW?WK6E<D-DCUOg?9>(q|E2
za9pl-U7mADRj)CaWWwXxr$t|vmb;Tn{ko_B<_U>lwOlsjZ*;x#I3Sc*+~6MJb#iuZ
z^k*LrH-yDExq4gYdRCG1d-Kz}l(7QJcN&MZVl*<QMobrmJU%k8?~JF<7bBn!7?PPP
z!DA<FtY<x4NFw~yQCOXaVvf75#m<js=-j6ZIbQwynCJdDphc215w{2svWd({eyadk
z8vmELO@OeDEEm-6`mF2iY)Db#9U4xZ6MkU8Y79FE+hhLIg(In9nHQ*;LM33$<67;#
zlTqX0wr3dqIyI@Z+!OWsiB2t+c+yZ;AZK8umZ|+X3f5@Q@7@e&-YDcV26!<=Z1Vap
z@5N6mn)QV}e>xVlUK(}^TvrMT!s*UfrJl#oTHI$u$6tRgR_k}BGmJIvsC8o+01wC7
z>MqD!r5QuFT`*>2wcS}QJEYfLbh|OjzcFR2J)QwF2tz1c`<;i;i2arQ)Gs{iF=U|Q
z0m1EuybP%sB3aad!}@zS_w1+|!11In*SO)~H?N${o7`Kvm&Bx+rIFkNt|(h#4$^~y
zvoxb~LLXJLGV88*NHrG@6Gx*VpK}jucde+6Ee3L7XYN<#;Sm988$-}A6pwo3JVQ(9
zj;&t1jPSEfzu(ydecP^9m8RjziJ{oxMYs>C+s<kSywH-Xk6qr+E$BVThX8}OTa9sZ
z%`j|xT%M0ad6eumh+L-<>|&qK$};kcqp;uagHG59<rhUufD!ot<M%XN`a%89sJ?N?
ztuwAW^L<#;J6Pflzg)$3FDyJUW8#SaP`J%}2fgAyAh_wp5f48{WB>2(JcdBwn7<U&
z|G1}x`yI~pgphy}AB>Y)9m0dhDdEFqQbKW`_Hy&`9*i@de;GOs#%uoOxlwqTvrmg(
zzGrD`($yl0I;X{*{fjv7EikY?-VjvcVzG(dyflV<H*u%EWw4x#G!KfmJw@#36D-*U
ze_nWwCHPCct5G47&+Gd=j}ev#(C_<;yOGWR!PC~<csf&E5+8bj(Qnxv^CSI*JM$zD
z`Mk_OR*$U`na3waxTb4B>5nc7;kM3B9Izo1pXxi8j6PnC&z{yJY6WR?3DlU3)#sWF
z>c<yCGEL>NpZwB;PZ?lN##bsW?v5wpYmXhTj|_V8jWJo5J`tHf?pxPJ)Wl`{ofi*|
zmHZnjYQ_<tnfXCQ=*GDBel*mKz6vq;NoC!=o7iEoSM(6kJhoYv?bS$fV)FbjftT+s
z*Yu`|$(XiRdd0fbi(av}5Suhvjw|(>H&lOIwO6k%x<Uau6j!Sedx%qfjeF~ZXIvlG
zs#&qxzizLnIXp++C)fKEk1Cv5;)cD&q_Ki#UnWC^z!Gpr$W@TA-=uzUtLf^Cn+3U&
zCK{D0sE(<_?6&E93q@_j40D}?p)(BR?EaYPy7|~<W6UzTcs9U8&C$s2VUdd|=3Gxl
zrgDsV9yYvik>=eZpo6eew+fk!$u!?qoo<}Md29-o3Y&_HGA%*xP_z4Itd2WXHU06S
zK=VDKVPpymjAcVx8nDG(#$#DZ>TC4Tz`-IExsgiO;p8^z4Kjm(ZjL3@!9KxKpNwTj
zjcw1qXs8y#x#2X{D7-eo@--Qg#W32<ZnF@xBq9R1yRP*mW=*Vg=}1?4G_KNDreX)r
z`&2jIt$EqU>b+v9VC+c{inMW!1~QC0Ed*2(RZbjHdAMWS)qW|LNTvyjPEREQYYZ&&
zUDj<pq3+RR%sQrCCH*rR%<4W*^}?C7OOAf;g4Hu(z_p9hBPP8=LZz_;VH@^r4`ukr
zQNvl97aQ?BKy}=Wt1Qu`YJJk4#-wq|XYK<osLhWD1&EgsN@MPCg!r4QEASkb!(;J~
ztJh>WoreX7U=YNe^UNVt;&I9zdPF0+^H;x$N2@Si;2pj>wz~TUl&0INQ<8oIV6a_a
ziu-VpgzRuz>{F8R?5s|iP08RQcX`Tc5WISE(TTFSdtA32Jyll;sS|sJ{v@7ceX-AD
zL1%}5h<6Fk9G;N})NA@s=Y?@l7$Y7*wQ20{<{CY0<VH%m!+2RmNaA|ph<8GjHRqve
zbe3?);1JW?<ESeQPrbSkH3=-zF+r4zGNI@+wh7c{Cp=Ra+fN;a2Ay=<mX4l{UFjDR
zr}h^98GFss`qfe&XM_?q7w9=PUNnIyy3cC7O<2?O;5(;oiQr%4P3Nn-hb73Fd}FUj
zE>=Zo+T~mPre|+are1%`$aEl5z&04d|F(<My)rJim_RU`lbq3ah2iR7#(ToBDX0B?
z{dM_pe4xKFlRP**%VR_7t`HvyIV@z8xjGsjd*xOnZH`ae@4ao@c^6%TnH}@<sfsxI
zAn=)sNXdsx^mBDQ`}gsM_i+(pWb)E4Jr@IGD88UFUkNi*V&vrzMDAt4e`7S4q=_Hd
z9<|&{e|GGQ@7#1~>ecvu?<E5d(#pQGEw!%kgOM5-8yp{uA6>nc_vlXsp0<`45XLj2
zfzXFjahX22R8U}F{)`A;(NUszoA!(->p~*fkHi(voG)(o3p3e!z7NMU2F#Vu6k?P^
zP7iZHUiC~Nx3kyIjjNw2S7Lf66idLk#ywzsUYyUa^)~@}NMwZFj~XZ_+`q%!UjIy?
z(UpZrDsC|1n8XmcT)i~0s5d<$!n=t0<;>;HDyguMDr!;80%N*i6U&*G9Ry;AD-ytT
z=Fk0#vdq%3u$3brP1$vGJQ=NneG`K*S9l>*BvzeEN~of^#h}tivE1vosw&iRJZ^hN
z+(|kSb7kfExcwQOkEs(WQyMPX&pQmUv4wyyai=DUqc4lX`5IOFl@n)2(3h{MTNHQc
z_aFZ7ht*NzNkO%kCQfyN2&G7lwB9p0>*D5pu|(6*!kx0rxQ`Srg|pe<itBobKiO!I
zNxfh#n$%mRBQ0bFgDghT`iwT`--r-LVx^kava|A0uu4sl2K2>!!mJ;~CcM^a&HKim
z<Akj-P8W7S##4{8Ryd{cAOUs+35Ep%>dE!-OVwt#Q651Q0YUBYyduUP8H@D-gl>(I
z_E(3Fn6Z<v3NfG|CA&GK1)%qg0eU8eU6(l4&`j2F)N9DJrSorjYK*JVNHx3ACU13f
z!ZGRTTf&?@Ab^WDmFSb)Fb{g(SM<TsZdOfXzy#T;&Q@-gc=(yZRHE<MRZXkFeXuLG
zxUZouCH#6sW5FFN9yMhArjEu|eGpKa46NI#6P-_joapU71I6xm>M-w<9iC2t@T8&s
zP7h%gyxuNXZd=EhfGx$fq}tQtV6;aVWh6*QJdP#L6ne7)N-7|T83tp&o3PV2#DPmk
z5c*w$Kpyl0*Qe2zXu%<kZ1(T(kHeZ8x?Tr2^|D8k@hmdbbQ}@v+WtYbrLTA+;CsNZ
zjtWQx8=_<GqM^5oAW@i>63pC?$}N_EIAJWg(!1lNOYbg@-#nI1X=E!G#%b4ss<p=%
z4U?$ie~s5v!0o`EJnK1KHwahR73Yl0e_Ige)ubePe8VdWTZ;MwsjWcdo*0R@s>=->
zOwRXh<CN0R!3|ui7gT$F+WN=40?Tu9L%ioXF41SoHTS;IpR+bT5L9;b=peZ;<3j<N
zHG4LRaX%88xGu%V`g>(4GrfJ{(G+p&>A0xADjSiOed>BWDc$|dDC0cDco5jC9jjFy
zpI;bE>ZwS8_NA9AquBPxR|1Od9y&}9mArMCaUQ=*E#T4>i;ACH@(bU(;|v}Z-(6aH
zJW=C&<A?Yre(+YByE=Y!3q;F<@ssDP6H*iV3OORH<a2S^zCs;g#*%x#d|x3#r~r#p
zecC<|SyGmE#}%rHC0TMeSMJjV*+wuAQX6J`G~~9fpMm1zYW0F0&{$mKf~6Vj`dW2T
zfk((|=Q_b{#*7yb*Y7KIx1+x=ZrCS6J_&cnE{+@bmD9HNmvNJtRFLHmh~Z{6snjoz
ziu<Wy)5P>D0$dm~_DO!rTR+_-@ED(|(u^Y#v#RLjbu`BxEh#3LyHBL9lh`TA)U(y+
z?b8d1OH}fnTLcN4uC}<<{VW_nkc{0iL3K3R3~biPHF&!Rl|dA<w>#Vc)@TDG>5SVM
zcN&4_v<;l^4}7kW&#EU(_6Te4IpDOw%+_dbYI>tivoUqLJ?d4xKf@L-G1g`Ia@v;(
zFp6|$y}?UzVNZFHHVSq=f8h6;REWDEhMN=7tP!MyrJS`^b@)3JxhwY-E?^-cS~K0u
zQltd2YM&Ucw1gd=sP3%$1f(US4X0{xwWo!1I+5O%YdmaD#6Ya|0?+z~XxET%cW94J
z|DNOu==K0=Ga3aRC2o#UNq?;OOf_QOIvoA}%H`S+1FA@NZY*Yp3?w;mOcRIo*-n^^
zT*y&3<PFU(7}r9VWs~2{CjG_qHA5^<8XmT>X_xqb`a=7Ou;4)<--NqyY*qoHPX_yt
z8!a#K4&z~ugcsUiY|%sz!Zsf9;CqnDJnA|n*v3|)na<-Qu}v)+Xt7-_N+z!znr>pl
z4R_{HcY1eZoKIZ#Ct{aD!knSJ)JHw)19X>$Ul{1vYcw*-%s#it_@A*~5Xl2#1?+)+
z=Vk1ec7Tq(2Z=@#5(kYW;U5mE87jV1A@ppB!#*E0AvOT)`L1}`eNLY~JC10!^rD=I
zS9A=9`r@dOxfgH1IOfYL@AJY}rIzNnx7^)i6wlbBCxpc?JmN|HT-ekUr}VQj|L^0p
zMxLfMkhnAMpbnq+cugpA?TyD-{f4Mu<6~4&3&eT<Zp5SN4gF^J2jfivV!a4x@|G6@
z@0K(O<Q1v@JJ_BV^mBse<hx!3QbZ$He9w)Bh~-Sg9ptgzLy)DUq#wAP)SjG6{_vq7
zwQcH+k30;#jtJ8#w0a3|5udnVl-L(>(fzfiEo^)$@UZTK2-D*;gXe<i8o*mUK392q
z-==}X7njB~g3Sosf9XpS-5UnJc^Q4>Do93$N{E=R-C@&k`j~&CWt}rUzSZBMsI-jl
z3<!QK?{#kV?=`Zi!({s~%qA{clARK3&a={ibO*lgCsnP?nHTY_-d;n*=;fJl*|Q>h
z<AWTH%bzU-X$2t=rupv+(;EAVXGI3MI`TPy4pXK0Yn~pBs|=9<4`aMKuYR^F-OYF&
z*F1YZ^@H7+nc-ShWalsuV{x56TgKYg=QVh}JE04n?2O?CV}oz!jn5X6hFK1KIwjz7
zQx!i%7;&?kVq>8DsbU;jky1~@bl2s8_r?q%R3R-=%v3d*w8pGw&tD6w<<O+3;9v9U
zkdd5XjxcGtrzYmAImQfaG0$+Da9@mD+~u;~l(OFHIk~X2L~*-K2y(TuCn4~0yPM*N
zL9lauU?~O@eWyO~>qQ<n-_=ONhnwZIqR(Xxc=RHTzw6n;Cd~Xz)KsS+^;5n0)Yd&K
zro|b2i&tpWt9~ssQs|-%fdh?Y)ygFRd6C1Pj0ShUWDMzLqd)4p*rQFFgx2PJ$cXHb
zb|N%oja#1;BZJ9fzRO~z+bvEh;VRAJLIOtIr@s%9m<6MsYWHUfwrTA6msq3W{;e7J
z+Gj;y^NGMw8j~SUyJsHL#SOTVczqG|M7NMBF;GxnM2~8g5J}KVVGg0hQib=V8WQ#&
z-%BC-s|EMRkO#1lT^qv&jyrUWYEPo?Ptiz9qsCQA#D3z?2|{IgT}%oBRz1k8A22vb
z@`iX&AJ`#tac_1>i2?SIKXAZ&DjpW1Nw|qV99ula%^Acu4PU+V>#)BnT;`+I&4;`Q
zk?B^A8WCE)G`9)ube~ST!?n9z<?Q=^ruGgGv7U{&vkK^R>d9i4CmIJ&4&82lEljRB
z_V^1v4=bp>+Gh)X9s9fl88tfPcl%whbvW&j4!E-wobiMH-HE;XA<r$74hP;1V{=$=
zc>T4FVi*)JS5Zqv!qGUQ&*VPxj(x@L*(fC=_(UA_dekh6V?tq(40VnhP<^@!;9#Oc
zZkcpVIO!voS605c+0dtiNLnMnKRy|!y(deROBs^RX#Li|{@m1`<28@0fix^<y;vo)
zTr$1qT!`mZoEMUl_$9yL4S?iqL%b=ZNVAzShThWT{^FPMc6G}zDRIGwIC7OUw!EvR
zwb(e}{PLdWC3kB{^N0{u-6NBRc6urnqW41+_@Pf_J4c=ql9?AsmFzm*xfL-168H+P
zX3(flj89z(J}zo7qa^qJR2X-X8_T0zLWR%u*Ts$Tg`pJoB))VP7alJPgs)V%6mp9v
z)z_Xs(j@XaWRHI1KF2d%&$pNM&2VQvyGP<XgI?O$NHW{+wVs8OwO7Xv?yYm`rS-5w
zKMKKmO5>;MwUnyQ<7qsAJ|p()b;VPJfI`CvT`${TNL*#=IVk9Uk<&Kv0>rfaqJ$83
zIEgF#MMPLg?UnmQhDS0_54j`vm%F{_dee-BnZDZfvD70O<2CzruM%`{9U_9b)}1bA
z+OOO1S6W8cOL5or20q+HoJKU6`}NM`lDW}UizPl!ytqka_=d*KDoYI%qX&bE`an-l
zz7RJ|3n83|u$uKq3DZipEPs$?p1hGpyp34-9Wh5hVDZVLtP=F46UY(gx!&D4CzCHk
z_1I--FZbKVP#kRWheTU=n<k0PRzm0GLBHM8f_)Z-n>;a%NMjdA^-g^>6x;Xtf?t`w
zdVVbM?;Zracj-6R)?n0lkoaIj6p#_rX?BnsMZJDj@Z?=$Jc=qlEw}#?%Y+A~TJgRb
z_7}oD(jJXk3!;HAaTC$xuE)u>)of(Zey)pFui1q^Bdt9XD~$?|26UnPVU=c!U6@+6
z`}Eb)+PoBIZMA_;omwAjgh|QAf#N#W3J}8VQHeR*FAWm@HQDY?qd|6!f#?<**gzJ$
zi5@{lJ^E6tcRglfAo>l3SUnj}Z(R&{B?!D(v|U5hHT*9wXxON{rg@1B8`YJF$w#Y0
zK0$?JG49V6*kZQoCQW!ce@*(#rM{Zb@5CW^z`Y?(!<-SzAg>OAFgEW$|9`>8RZ?ya
z^Odn9n9%$`Tj(K=0b%Thcvy|bASYYge#N?YL_cdrNwln=ncR4*kxTIx$8no6;th|w
zA;-U614Cak9y|6I*+A;piFTfn19X|*?Ja0b`(l@;p#ihG-6}zZ!p4q0DrxH`YsOyv
z?oMuhpZ^ZD*l+lM!w1Cyl@mRj1R+8;BS>a!zeAoZf^_D&kHZ?)0#3rq-c@i0JmQa3
zH{I!Skr46R!%V69=ok*1aa7&>;SaM`$1zW4b6Nx(_XH8MHE*vI?lY}OGMv*%wOg{H
zr8Q2eQx-Wd@YBL(gq!1x0bY2KrFu;OlQ&C|#M%8a$3QUCmIoTeX1UzXsrg`gdz|+a
zx6&o4BvnxP$2;Rq{p6g*TOLuq$MUs+DUb?h|HJCIU;r6H9j&EiBdzPCP<n6wc|Gfq
zhUpmsN42#)N8Wc!rR@8lx;y+xg_dggJ(=kBBaIUGJl5i44;j}sENE4$(o95xi-w5X
z7T5i!t_GDFjn7nVaO1f@#^?IIVF-@&i|X`kL1g`<rjIQ}OxvOO>e5Q)V=INMU#n$X
z>IJ@W$BR0N?4$YkJ?T}x^PiV+SiaY9BwOTK)^9fd`uNe8Aqp}$wTyKhKdHDWJvHMw
z(K+lfGXU6S&lOrZb#QbtE`P3&$4slhQ)k+9B5>qGE}GXXR9eB81M|u&UCQbc_7Ul;
zO1T{|sEVtf)61x6!O|px#FFX=cdq`p_Bj#d68G#nBa|-h2&%7Fy$Yu|Zg3IzZMrkv
z=psTI#7+AB`WY^!n~lpdBxf0^qM|a4W*&>_22q%g!I<%!DD!e&o}fMvGd-$y6eOcD
z%f*a|_W+xGjxfb8khz-soHz}RES^T0n{1vjT12?PTl{+j*MVC-K+HNi<2HXZLDJ%O
zeQ@U>X}?2=7sXS$df%z;i+LpmWC&p0V;T(%mczL~z0{*Wj=O|AX|v8fH54_^6$&@r
zA9e2V!gJj3_0JVvqmS3k68*hxJ;5bnnfiu3Q6l69FW{>wLySg!rRyF8ZxW17w2RI9
z*)_H?TJ;kW8!J^k#Yi0BRl<>gl5KIHO3(slRNK}1!niBn&ucvV^gd39)3wICc>;RY
z{`1f=$4JRdr;EAJ)<w7aAb0`BvDNXWJ=}WDA^uOyO9o%)cWEvD*)iblP>*Pr3>HIP
zvu@UJSo3%1>=@Nw2}AeqF|OA2UA&%SlQ&v<_8}0NbaPUy#RGyo#&b0dP#!e;E76)x
z5=@gcS>9U+fKp?zAs*8FO`p9s9@e}Rlm3y|@|?LX<SHhWXCMZTsQ%Qe7|E(*;#ZOL
zLwGM~j%}Xn^0tWW`r<asl#4rr9{aB3S>0*OYY_YG5@HZw+y+v~uv@1Wc^ao_pr~T^
zXyjzfn|g8Tc~XD4Jz}=l=Q)MB8Q^Bv?~%7o40SCkvC$3~FLp8I2URX!4L;*^95S#L
zMswJ8M~Tu&Tsu!VmM28EAX}DUr=TuVxAw;?zG>K-Ft-K$qy7`J{ox5E{5xi-C5K}i
zR|R}6(mW{NaB)HvrVgUUB_W6qf|~cmDQ};A?8j+i+DI6aGcFh!Bw6`uo&(+v_#MVs
zefADuXaTJ~r<EJ2nTYd(Kn6?FfVffnSSsG)O(WTa<!)EJr8Yf8+j(0()efSXj0@gI
zl0tGg-&MuQiN1Kx)0NIe>C5xJA(E+-aJ(OQ@-Sh|ml)+A3NR8Uv_~Hq8$zI@yyRoG
zg-cFIpQvrJ`9q^6^yWnme^ENreCm$Ii2{eCrBPglwt;mVh|e`>WEQ963sq&_;b-H^
z>S(I*#J^H;@wxnu_*!2}F(ClMZ-gONNqaIN-x`BF=<%JghY|Rn?>rvg8=-r<hBGLv
z=Y>~Pd4vAwnSctQPRY1uKM6wuhvHLm;C%6*NF%k&4v1bj3Q}8Ken6y)TW>U`9XMa?
zd03fl{5GvnzryX?S@|mu6l&{09C6iwLQ|;^qs9Kd`he&pvWnb@queRi94PnoFMjch
zxb}brw<&>*>x>Fs_N*T)_w@pnqPKKwj2m15HA;qb!WC1vwl1t`SQIxI7>YC)hMQFc
zKRk$WtPtjqA(|D?3U+Ooo;7B8yrqYkstC5E&2mBCATb~2=ws?|ULSKk{zVXmc^VBN
zAmbMOth6-8t?p!FGQjMv+l(7da7eChcSXcbcL)@MG2BKBiX;*9)g3QgZrTN!Vjjii
z*>jiif`99X8kJRu51UV~I`_i>AXD}J!fcA`d5Lg>4VClRGMA>$38Bx8amF`FGsH%h
zl1&XrP1W&q5gP%iN3%<D3-20?R+n~VL<AOnWwmxsFERgzW0eLukl~i@Gs5e;I{A*(
zD(atl8CzHZO)p$j;A=g}+*r}>30qkcorc{3Uz^-)x2j=r*&99jjWT#ctoJN+VyP4T
z-Z)s!r9QVa1_ZF52$e%dr~_3g=VVwV6C3(tv^vp*OO5dZh51<C9Glb&PutO$6gHQ#
zi9Mj!9Ykw?P_4Pd5p-<6gym>FWUxtn;$an3`nWB(sEZkQ#v`hN0T{FQs7JgoqgexM
ztE%<2$2K*_?v3nM`kijjvBT5#kH3nYZmbAYU9n3=y==YR?z$D8bdQgzlpPuSCHAW3
z7_o5o2}mv`%r@?lDk2<691xI}6v!U*EE5(QI?r`>NW);+S<k~RMQx3w1B_o*>0k*h
zeB{9Szsn~y&%Wfbc!EH)UO6CkG&36v6)JsHJ*Vx!G3}O4jER_}<DS$-ov7<i9B>YZ
z{pHp4UtG&bqIuw*)OA)B>1LE!+MaUTg{cHS?FmcWe46N=@lIdSkur$aJP5P>6LHqX
z8<>~LIH$eDAur)d&U=O%+B)J5!7Xnqh7xZc5DSE7AUAeyy<#`<mKOIn^Wtq|)s*gV
z7u=14gTJdHY;HO7@44OpZo2WlH?vfB@qq_JyuTEo#fR=RSrhYJ8LIjt&FZ}QzmAXf
zvr^j<pXldU-$YzASP}rI)77WOCT(}e;xiS}WeV})b4`&1xJkvnFoN}6NO-^W1j06q
z6)T>v1PjuV4kcfE#+ETd-&E&Xp2J=7t+zwoVEK?I3amlji_=nkZzM<lsqpAOxU3bf
zFMc$}wJYvvY>l7vl{ATs6weoug%=bKo2V-;d%h6BvCQ2Tmp?BGg{GB(Ptz9>3m@aX
zdxa{dbrN*Y6(K9|FjqZ)zEIdso=R7{gf~6n>uX%n*ESf}>Nk8l5>?eAvOXD*;(E8?
zEI=4;cwRJGF4dUl-snH%NyB7$FmCc6^4=d#C)%6cO&wQZ<@rKzYq+<k3*`z-`)JH~
z-ZCSIOxgqF(VnTgzg^T4vphWBft@ku`9kKg-yUc~cr{o3Wn!R7UiL&IP`f@XA#PFE
zX@MTMy1$WBx89}?C@3`Ic8~DxbP0_+T#pP`&_VT8A$%(jiJ1R<xp}+CVu3nIo9(;Y
z0NZ<l*%=mZD27m{qV&ol@0E)<W?a`xgc`weh&!Na#4O^}H#}c`_+~JVlqNNL0v99Y
zXfi?<o+$G8W`8#4g5nS2wnzoJP4Z2>Qf<*Q5{AbsgSsN)?h_N;t%){snq(iV{YmV`
z4Dz<d;~L>wjJ3vL{@?#5+6{+=LsHC)cdD=iqUqLe0$*aks^5z?4H5j@Tbtuvoc#Kq
z7dM9gbmkmIyAuQMuq7=4hWvq-BsoF+ky@%4b@$vH|C!fmjH??2>z%PlzcE=SO=J~B
z7x}**Q2$tF!_Z+$@t{_t$WZ<on_ZeRMfiqz$W7~ETH|3a-16Rhmu}I@SN{I~c*F&H
zL2rmh-Su)5c3V{fdEu25+l(<WsFIkVbaX?^Dj>EyJnx-daIu52Q-#<=z|zMqb=TX~
z5xd<A^$OkXF;L`C>44T5dp!o604GzlxKH?SnGeK%mvJb|gX4h8ctH_Xz+;_yiP0U>
z2V!i3`y5u^#Xi-2kBm+bEAxoKr2%p3nO;$=;SwMHsD^_=A~pVTth&karPUe7-5U-j
zggo~cn%W3gz{JN%l_I`KV(8MMP0!7I+S@yWL~z=e$TOGFrT8`1Z%N)DRnlku2Php2
z!JU3i1L{XBciuzD2vrID4dd00(ly@nXmRF-4|~fi&Jw)sS_Hp?5*j?N;;?+ccvlri
zp|S6|;`%n6mfqL3v7Y3R28>$-57`feLqlrQ@`nFN-wCgnB!BD<uuujOuF_9Dmo02K
z=JKNEBW*T{n&Cg<Q*{e3%<w;RZPXmeD1R<2=G_%v7`c0=P7`6k+qN5kU%5s4y>mx>
z?V`m~C*m88Z4CPGt<i$ZCnFZ8<vXuSiFR9@RKIr>yo_aX{(~Xk%@Jl({P?`3F2#}q
zCQr`pKY6CHFpPtR+LiN37@mXY3lVF3{Y+f0lGefE>pbnCM63%5%mpe}_%q$0*v?lT
zEEo0=G0_iN44R6vs}G*9%Ce8X^G9x^^j?05d3-FcIanxEFH+-c^|Qi5@jAE8ms(tZ
z@Y2Jwuo*W9q4K_iAi8Q3#5is;-Wz!`-fYzBa=Q?z&h?66^K^Y68Kz{0;8jR}fJ|hj
zu~=Rkvs{2lDLRKa?zRsJDcZ$@BFvp#-EA0OAubh20$g$1-D3F1G6wFg2Ss5rG+|nP
z5)<8nW^r#fT8mJg#2xCQpO-q5aHncDb7jr<xYAl}L3PS`c=M{d>!6pmWc5bcNSrq2
zwZ`~izC}Rjd>?u+O(h;!y?|5`tS&JY?>vc-!7_E$ykbRbG^iB{h?4+qbk}^Ife<wL
z6H!pfpX_mt5EMWX=2op-4<|GY<YMJP5uqZOElRprWf&!O_dwjI7CZ*lm51SKZKz-T
zW2_O1P*%>?TD7WMp~ye&9&Gx}lnb)+p!u7QRJ|Yq;;?ml4r``fiXOG+K27(ISg)BL
zV3qrg56_x3CKym7yweb<8o(&#t;5E%J54tMJ}T_c$RLaw_f*#uG0-Lf??KElsh^cu
z_s0VoSnEVQ=pOS*;qkTkpq@h<(c))O;x|5|h9ozRhxK#)$WUw%LNz4fjz_$LM8g<~
zM~w~ia3r?sqb2R4W1H}9E-!6P=60jXGXzq&!y~Na>fPzzYk&Xy*yWX7j$L?1?-0+4
z-D=l~b1N6u9@lNh=47wOShq0tsk7w-Q(BrT(EYB4z#<O#cV-$sXxxi?=^-^@ixq#G
zI4pRx=gx|kRS!F4*6fHrQoaE}6tspFwRgu+&nK*0sxptMh%6Xn|1fTgTjPZLT#jQv
zob+lU2pWXg$0_&uhd;z=HRn{Fjx+l6f;;bw*L*G>dFOHD@@Ku>kfaeBjQIh-bJb<t
zQgrC&-5;!WXT0GLj7sp;@J!UA?RiTDysoJAwmvqrtw(Ho!A&x&WxVTELl=yo|2-8E
zN@MD;@xF_&4Iup12P(=KI0NzF!J^bmRduQoZF)%TIiE_Hm12v}^ZR2TDC#Iz`X}zb
zhv#Hm)I9XQ{#<;j?s#mc&+=#b!V4pf6hGI%dfLe#|Al^&9}>B!2RxZBkNVxm3pl>k
z@7}JdXXBfn5lCA5m-yCOa8Z(-g_#fZJA+w@DDHdzURkp+e((&mKq~QcG~(FU^u$kI
z|N8of7o-7fCJA-2AeX%$`j-Jrw#4NxoKN9r-Veo&K?eELUeKLSa=~;Z#OLt^5yOtD
z<C$LQ$`^F&l&4r4a8c!|7tR+-TFVuDwfd-Zln3uM)y8Go<JuPrA*_%G<+>L{J3A&}
zTCR5^l#7J=xj|p)Qv%0*qss`R!u#^3Dtd%JD&a<McHa%`tcnZa;?h|P8yinR%b4!w
z8U9B9Q<$rn!e^j&EM^&>m1$Wt#~*!ZHZfOY7#o5`WXH}k49usBZs3*|bl+nGO+eV5
zHi9$WYMk15+}>tT*dp}C?M9Gy9PcA+)$R}igk_351-F(bE^jnchw3)f1SsNSfxFG?
zE_wL7R7?=h5)QFOv&4<^R|G+IMqnIua=qtIp07*PVH>#)mKm8#XQZKuT{4D*Pl`r?
zAsPBWG`at_l#;;;G`ovQF3MK@p1p=xge$9SFo7*IDYIggw*YF&v>3flb$ipWd9~|8
z&k!)Nk=9h}rU&Z3##*Dkp?Kc4yT`={^g1=G<Vb}@?)DTnGjGx3HW`Kyi_{mymhhmF
zS4s410)C0oZb0pLDFS53K!`I9<#P!>PJ{>W$UBf#q?Bab-_6b)u*st+Ar2A7#iS7B
zgb<F-VEf>o;z9LY`Nc0{vv*CtW3^l3A>&$<KgFQnVKr8A>Md^EF%*xu$MIeqvs04z
zs0PhTaCGX=u~p5|u#9)2XRQwRv+ep#RI0w%q2KH!0`91})mXc{2pQ}iYPws)Eow-t
zSYpo$h5hX(Rb+Ll2L@0=>{GK^^hGIH+;1q%d=Jr&4|pg9vLRulKw?S59H9s~^ta2t
zISw1;%zOB<e-FL>|HcvZmsEVXJ6`d0apsD*YG)kvT-W!b4L+uZZ5<ONjx?fNFYLM#
z>OJ%7Vf)%2C)KQ}BtAW5JW~S`r#0n$2rka(@6aEJ%xi+j!lhK<tZH?kb33QG;Q@SJ
zRiHsBN_)ejSc0VeP5(x+6>sTxeKWD)-}XFbD2seFTrkoIT^P>0{<@bW6Fu>s>Q<8F
zM4cD!yEoWldi%ihfB|ezeCUR?lZbQTBhPdVZ!w76$JL3K)yF47XX-e|_M(CCmK};u
z^-+#~e5UbEo$iaz_4ias4a662K{N+)B72KYqT6eG{h_w_%FD(6$Q<PC%&%XtSs9FW
zxq~u8T+Ua<Hv7gSY|89oXj8xStYx6N?_7piTsbD+yRWnf=;JWQ53bw{!TnL=FOK=I
z#!o8g=q1)p91^AD2qI>LPrU3<A&*kM#N{qV>5Zi?PnBth#I$1MA_g%P+E*MBb!-{s
zjf5M=p+eQt{{gaisE|bh|2M_ehYICuVE0_3ZYnLKZKxoxRi#QJRQ)>rMz4y9)PM&O
z@F1<a;ZQksznB*{3R0zMMckzNQ^%2)-K^ip+VS~6bpGe3XokT(bID3%x~h#2!fws*
z2#Y2kN069V{X@Q>iL4Z})Ylk3kaG@|n-uZ{aY(1{n0rWs51NxEl^Gj*o+?%F!0N)e
z2zkLR8Z3-{KG<$m?{zF(+*YlN76_vTuA$pCdFe$v9(TAHYb73DBb?JaH6t~ck`cTo
z=2vH)rX0`(hYGLLl?vgJE}o2<u;2YYYSd|^Ww1Ny{JWNiO1-)(3i4RuMd@I_j>IyJ
z`*+Z@1{KVk_iO%(Mq$Nv#Y(MM_%#V48BkCg>Nj%TiD=dDbias|hr~0b_Gg&;KG}#>
z>Y{Bu??EAggE1zYtA+P(TVjnMpxj4WsL@r>=pc9=iFSWmiW8lhj1(;;J?O43<g@v{
zA}(}Kb+U<P4KK)5wcfiKRwFGD^B`er;}w|9(0&6Nc?xoqK2||uEDIleGKM{49DcU(
zsF3AC>Wy(%ZR_A|m{$0kyi`1x({X#!y(~_T#RpUdAz)<f@u1)k__Z@O>vv@_UXBkL
zW{T)YE<^AfwIk#TPoR$2qVmN>Dwod4k7ynjFONsv9UR|8Z1qye&{OxvHjM<wq>K`{
z-PMab#SVYeFRfn~JH6q#li<RNK4F)7S@Wyd?J?rreu@|F9)EJkF;t7ahO=-z@9ll+
zPZ=j-zvqz|;JJIi6>HO}@SuwbbH)BYq#Z-LlsN2K?Zg3%m#cg75{W*d;>>G`meD}4
zDdh=pR7J&CEsnVeHwxqp$35NaPzjt+U#(a<o%BeSlTsZ1{*=3JO^<}rnsw5A+XV8A
zVBo?ImDF#z!e7N%@7EU85qZ?lc{!QNb{-0t)d-dHW_v8&aG!|3(xLrL_nBs7@s>Z@
z$FLlGTYbPv<F4TXaxp?LRRCv?hvUx$kV+L)`VqE%pialguuPK=JsTwGX6`@Ita5&e
z4CkGP<71U$XqXz=Ph1<p6HmO0+8QtzoN!1SMCem@l-l~wi~#{oM&ff14ihqRjK&wP
zhecF;c?r3W_)1IA**+Lwt9e`cBgHrF7++;3`CAnYqQ@SH@2a!Rrpv^N-+N~y3Ba_D
zL;Hgplg)ug>W{t^k_FA^tex?b%d!4vRPmze$TK|ee@-!y=4h9_C}PM#NCKI+4zbK%
zECjDoiv(}li@Ni$#OI+e#nBZn>OGoyTyw^+e6bK*9(s98T=ilhrLcG94!hdG(r5G<
zRh&8qYe@(Il`I^<B=Wi!3&9;Ko~zgEV=|X0Zg{a!+!bhLZ>;v4Zs$<0oBXY4ei+lJ
z=ZZrAMdii97(nLaR3Fpz2U31OiW2b|Zou7_;Ynw@99|bGB-!a_y(oEQS{kI#0P>sm
z_;Z9`g;(=<%+;7G9mJm681poysh2qDw+QnMnBL#2pUsOF#clezsJS_A7yd|>|2eJj
z;|?_*=|Ww7r%?l!Zf=VC#>CzWX<?V9$_vzaUon=6yNnl7UbKxbimbD{I$@Ji%UI{7
zxCc2_z0q!HL#G`}TrvU$Ugj;4;&pb5LJ2o`s%yGNH<EWI8r7g1$`DPSwoALO`9;0f
zkhS7SkrUl&%$B2@h?Ope2Pydou)Hy@g}o`cJK{d$SGTC<9{f1mzGsY=%o<M(X(mW;
z5o?WmKQWlwUo7lZEe1IH>&6tNbXP}2qM)2;7V&uWRI$hh1iPUt*1Hq@3@4*sA0%1f
zt{S+safr|hp^70-2lnvAfNxkR6$^ik->5preQ_ej1qdG!l>eLD&PlFg(tt}Khz?fm
zffsEPBIl5kuy4iBB_8w!z@($Z-0y<;bKh?ET%^ex^Zbx!ji_u$ogRMid@+lwW}3+i
z^dqsw^VNby<0Bq&D{;f)QSHn&JSn#dPVsgket^ExvWu~ASDT*PbFo83+z$Ef-zf-{
z=HA#P#9#@_EwJ0yNZOg{PdIc45#HbZvDb(XC8>ygnif*{L7_7Z!hRLQex{4_f$EfI
z*p7pq(iLo`lrtUrUl&^z&4<-K0~|7fmxWvEH^>+sM}!Az!O3{V&>(1O01!t#Sq)1Y
z<CvOZHB)3_$Gw5_eU4<9^L4_5C>eg^q;V$c06Cye32@#Yd8FdB7fw3oIHM7xFE9mu
z%`3wdyCKecOdANSlOBrY?4PUdr4*JY=~aJN9A?zxHw?cUbsrbSo8D;MI0Kfw@s>_X
z)7|m5;WQ7*aO;BLz((QC5%22DTE+KVcS9Er@AUG%v25%`Wd4CxA3;}q=u3`B2p_r0
zLO3lP7~*5kXcG<}pSW!0!sgl~3!CGj%Q~^$_|yX+ZE+g!f93(eUUg#(jRB7k8{C*0
zlla2L>B-K+rYpV_X53FGtG_au+!zz_b@j5@U#KB(r1-{tEgnu=w)j?{D@#}nsIfOS
z(M!K~FDX#rSwy-mm9=+mj33=cI|-}eClBOan1MK4sAFXzlEcBc?67F-EPRYEKU^qf
zI@IMGbefCs>xe53n`Gxbz#hKzus#`Ax*E#VxazQ^uNzZad$m6ZFp8UpYqjt~y7sW>
zb?ac>X2WrvF;d{;>#O*|MUTe~{#ury>f=U#ZS3vp8;hF^C!=Deui4Em#ethkvZB(e
z!GBEGH~b1r2am=XMz5u-p8x|!k8DOU%LR2^T|+TP1$ZY9#atCEEQrl>=cyk+kr}rf
zF7z#l2nVoemTy&+RHMP-65rK;(MtBj?S~6#*whhs2<?0s6XWkr4+2IBZ~pn!Ca6YZ
zfr`kxmNpZ29kvV-<NV~6MxsWz5(AWI=vvFly8d$C)%%m^fEkKriHEy^J9n8oTvh^5
zHh5UwSls48ZB!}B9i)`e<WlVAdr2M~&2Ed4B}`+hp)Mxm*-F<e1`gast2_`=E@91D
z9aS34tTt**W5rH$&0)#%SI>^Msy4A9)xzznX6J+GbgN#T9MH;cPsY6v{n5e%_6Qk-
zL=&;z$Sj5RXk#M#1vsNS4aR^LVr5GV9TxMMrc5Y(ke9$C4r_>o-TIFhb+?^}7-2QW
z1t{aWbj2o529a=Z!kN_9)SsDe<h34g4Ni&ipwU?b*SL-s(q<QO_mZB{h4*&#vvJT_
zy2m=E9CWs5de*Vy9?_1&g=Ry^TMVIp)PE|TLV9~|tu9gLAp3lqzA~ISv$p-PL}i7~
zvEMW-R+>5Pjva2uSwd;M^RT2^+;^FQaU^zmoz^}OyNzPT{^a)BbNK(HD?1u{Ju<Ws
zWU?kSk2)wumz}ZSrCem~Xzk*FO9!9=>*Ao-H=_+ESw*<3^_@Dv-Fw)mCf~`!)u`fq
z&1T>QaKy7x^K-=6SB!EkVn|NrQ31o~y)TZbE9@+iIXi)F;kbtS8^ksz+!NP25{Z-k
zYQn&i#3F}nO~8XcX%(&0@#wVRnY!~Ri8Jc$okyX&uL(B9F3x%zAoS!Vt4S>B2nyQ+
z8A%Dvc~@%NfD!f^{vg+ESG?&7gy4?ha}aN-myM9dcw4{WeR$6bX%089rFhp9vAn3|
z;yoRP<z%${|G9b_*sSYx&-;oo#u(Qa9%GD;G52yf4!R6uj4_5`SXxRcSZYsOkzp8y
zzTBmyrS0?$=%8U)$Mtv^$K$%L>v28kay-HyRs^hwh!lx2#-K5V#~5RbF~%5Uj4{R-
z^L(z~_j@z*JVwp^@9*vUy<YFvr6!r8B+hxBw_#XG;8&F(Kg0~kgLbjf)eJM>eaeN$
zJ&TL~vDzYRo`CxNL~Su`!#PA#Ukyh1tagIZ>+Dc`etrjshGzHC$rt_u$7Te2V=%t-
zxa~02DQy2rD?xZ8cJpfw(u(21HyUIBdf;1kvdqzN<2&E?a^yh^nNmBxch6`6F#^Md
zQs0E%Dz9w(sQF+woDBL;DosboV!IssgxcEiz)G{U3-$>iPNl)+h5LkS@pF%h_LbKg
z4i{^xVx)bsO9wH2y+qB3aK{>7y04(+Omv@V94=Em&;#kXRF8lLNCW1$gNBB-xME*H
z@@biiRmznb3{howKYpo52d*-l+!0;^Vx_wVm!TL56Z+LP@_MG><hB0Bq<1g_C{^|q
z{53fqb2aXE<`VNfU(z=A#C86_bG1L_3!)0_9}M0N>aBgKa|B)kp|4@zGW&0_(9O12
zF}J@_74tUcD-}1XB94s{akIM2Tg~;mWuGJiM5rrfnhE5ss?k4G0v9dXC)Dz@U&dl}
z1LqIZ>hY$M5LoJ7mMg)=V47mrtHSJ)X_3^RkMz{r9*u4~hObC8sZS_x%w)A^C7UNF
zvf0y1(L#$CA{Dh$(YjA;@!3ZUhDl*cn^zj2Jg$6yba;^F#K&v2>|-#gp8X(P9f_Xz
znNfQv37j+%BLD%Bm}9*r1gj!f=#;W-P$$1x5F0&wLzM)I3dz(ZSWAG;9<A6gx2IP>
zS3+m?d2+B0Ns@2&gnqMd;lkg>;65SIf>V-O%nLo_m0xjtjCeM){|19L>ZTJ#-ZAce
z(W1l0#-!IFh0(Jwz^uCE>U<xk+!FI)kmL@xMAn=BJa@aWxZ}o76@oz1wjy@9TSgIr
zb!C#N()8Zie&qV)k(-J8RAE^sntAMld?3-t=8nZ~Ef;}QN8<rOIQ!%n%*2D5^Yvhj
zhtxy95WE==YxOt&G9K~Jjo9JfS)#u$`nz~ceQwFfJCCcIZqEA?hP+6|kYDamQ4@VW
zwNL1|iq{=X*>LRjVzmt;8GqUzJsCUy8FgNgP3_mxK<*=Q=!^q;1?o_|reg)i?Vu_H
z;HH>XFPnHV2IG)dyx5*0<kd`|=hBS+sK@I~Pq7)VXi^Rt*b)?U%oFJ*P+X=SJ8tL;
zFPb5xPIzH(#6#wOQb1JNVMI>(B(=B1X^)ce0ppBY^fIrZcw?U=4XIiDD~I}~`WYKW
zs`OUv6!7AKr{e8>Lf+@ikn&B&tI}g+YUX%XAN@sO{hmKESaQ7YLy=}zNim!gH2ru3
zQu=|XS7@iytA6NSvF#g(kNklFFWAz{%9OC~jZXwIyR$t$by4Py=K_4@X1Axq_~#xk
zLxI42dM)sfV8y<4MZ#5MGxRsNGexiO{4Wd(zBYVtLU8+#k>7YI<kyK$zxAi5^PRU~
zsw=)%%Ose$wIaBR9|SNur@{Eq&9;tCRbT&?_{ne}0VCSLelaf@Iv#@xsEP~r3saJB
z^uNZ1`-Q$Tb=6C8(f)!riw(@hu9i~NxMaUh{a_}L&-izjs&soHl`h*~p8J&z%j5F>
z1%*~dwr|P}nX_M5x|Q`$W9BOiO^FcJk~OYW+j>Zgt28?21kXt|<l?6oayqV2wcFvv
zuQe=$q~TPL6O_lDBDRPBD`rC;R)NU6W1f1hv=QC+I@d0bep!(DF704LZZP~P#4*T?
z1zMiemr}S;lfJWQdE4^1(d&a|XE1J3y~?VIbd$YVz)qo7yJf$y4w(qCSz`A{$EDJC
zM8b;<XFAd6QCnPFjGkDcpN+S-#Zq<7p-%Zmy&-H(13uDNsu04YA1k{?uT4#8A5H#4
z&K7%jG@}N4RdSWh`beIcyj-3GTnAXFb-%DnIhfGJWnJ3@RA=`@bm%9h|FOnn5Y#EP
z2(X~)b_`2_kyz(Bwll#rIb3f5u;Tq=Y|zqU0WuXERS~ZhCByDI4?3E<bldFFAJVo7
zokZ{U=yi?c`-@JtPe65{I@!FxP)+lG8H4&`r5?9RO%W^CG>RDUNoYV8I93qfsK6om
z5~5u1T=KanDD$>X)=n?n57ZSGcAF>HMEsSM`A&HaPw!4E?H!)#mhD{FyHx~vkb&oS
zs)$VHuzb5*uQ~(^aF0*p2)g8Zz1K`|0&e;~kCsp8I02&W_xw><X4-;RW4G(4<qt~B
z2Mi6VOjy1LT}X(IO?_SQkS0??Ulj9#cvxU`roJa0@mNS=5t9>=z<sgFAd1ISwLULK
zJT8bE`6r&xV*hN>&ti{y$QX!F(_`_J7ZgIGI)-r>hitD0$ryckHlFqZudWVuC2wDi
zXZDNMM4E@2T{(ua-^<m!orn8?;UbU#4E{k4kG?gg-8a5OB^To%uWN?vIjm}T!nPcB
z`{iiCiOEwNE>mqeRp~JgH+z&Y*vGwyJf~^+bV9A60uvdXR87VM@>7}%uU9dSf#Esr
zwzrevie)>a6ME;$ctgD;jm(*P(||4m6}_b%S9Ywty(Qi@sJ*jC2}*Oewn1nGQj_zp
z!ED2ZCf@VMB(HV6?+&LREzY@*)kJ^Y5+4|R_8(mEq3UID^!P|Cln>^|sx(jn<;Eu-
zHZ21nzv5GO&K-@<+<E)>c2LCUswpuCr0EwbEhY{iOSKyw96?HJXxF7M^J{-(;*j`8
zU^HM}^{swR;8gyd+75LRA^3a0Q{)0DK{AsLjvowEBMi@vwHLThDBJUy{mC1F-2^5Z
z5q}XvtRIHFzu+%t(i9fgs=l~T1#qOugD!HdCUT&~#eWf+BW7TR^t|LR1<gTKDU{2l
zs+++$k<O}a220GrFIO4T5V*lPe-WdE<lu_G6ioc{@=si;9xLSaBjem?T=f?{$3>@E
zdycREi?A{=A)r*ipx3Am1XYPCu5~fdNJhGnajR&w>#tuFb2ZIMqP2ORCJ1^ouB%Nm
zbHmP8gZWt3-{7uGFrrvs@bH>r0q`qPcOaSJxY0EUUBlwu^p}DFTSqg4RLc3VsMVh;
z$YC>!R7PXIMWZG~0D}B$=e)7_TdMPS{>U9#tlxNS!%lh=^65hwwp8O5BZ^m|UKK0p
z8{gpHL&$y_4IQS7EK!q}D7}V?Bze_e3b|Bv5|#w%b+a3+Ce0Uamm0GxqqL$fYc;66
zX~Sg6h&DHeTjS!V#Z-rylPNM4y=%O>Bc0GKYyVPE6oKp4`Oie4?1}Y;iDV{|Aga;W
zVDLM;x?-c}yh50xOTFU{+26fAdIZf-I*O&`NUy$MzXxq85gek=jhm_|GK$R}wWo86
z5Cs0%l1@EC{vb>lPM2=ggN+fpH>$GKq;HS$zvy{JYgS^1B_}iK1#9ih%e76D{Kd~=
z%F|d)-rF6jQxY2X$KC3(9b`h#R~3;AIxo*IFL1ip4`N<)kE*6DdVAb!Fq6!S`@C#C
zx4Vj|#Qpw5J;EK19&vhfnloX`18$ijn0U~DH6hD;$m7B4A=>SUhdn`FI~mSF^*^G<
zWa5SWd(^W<=$m-aV~&!1t&j|lyT8@5$J5k4o~Ru~+yVEvX?>xnpK??7j)?>AUaj2F
zXgsX}dvT@ejAzsXGzuXK+=e@d0R|7i0ZnlTx!1wk;$UYI(*~nqd1Kuj%^h*bS8xP*
zDhGE2cdx^)(u|fit%;86C}RqU<7JVX%?P07YvP!OA%1)u_b@$EAb=D8NO5qSRHe$w
z;W(wA%OIOiYtlu%JrifV1gjC^b;ldE#hKk3Z~FHXq5j`;x9x+&lkv8iOrU(=i9YMY
znc+oX!QS<}iY2L{sJ-VkD&~G^O835iS=rJM=TuhRR?AuQfh*sZIjq=)_)z`#U{#nQ
zzCJR%h?H3Ej|C&dHZjr-$XLor36g$(ni*E?GlOM=kCb(M?)uAkGdUMucwofSNC~+`
zU+TdpMACA3cVz6Mue>MNhKvuVJ}kaAK$|-Wh!)>y4@)qO5}?O>3hPj8OTKe$@;Tzn
z@V%FHFxAFC7yzuX@ixP`<wuw1^@*PhE$1?g(Bi<E;xCe~@C65i3b!OlM-Gt#;`jg&
zGIcmEI#AFWoPWkFzSxypF$dwUU*gX}^g}!?m#SO*CkNs(^-~x?(m=-L>ZdwXY=h<;
zun>+S3EZfd5$diuAcVTSF|Iryv<_jnC9YEQ%Fsl0vNNteAjHYD4b0rNYg9KaJBQ+0
z&u%4*Z^hGDjV7fJW$#?|KhWP7^E}k+{~XsHutK6ZvKK#sVtwnzh>`MC#0?rHGaD16
zN6i<lUmpuy4~cOJczI(D@=V8jlM9PuEuh}4cGurEKW=e7Le-AMt(svP=rPnH7oo`>
ziN*dvjy@JkJiD|*TIwmx|IbnHtr#KdJr8?>o7|DfL>mtjL{)cNH2KfiA`HYTw_dqA
zn!UX2I%at-0$HRlXT7y{L{jlsjkW_4Nx;F9wlpPW9fqlecYKYyA{*ok4B}c3h}dB?
z*0~?z$fcrky~h9TZ+;dVJU$tWczHJJtL(!o(dAa{7=3V*dki(k3cUv`63*x=(<QEb
zpCCzbQfxMyIM$9;@ww}XK@S5Jh5_`DrU_A#X^*h5^+3tyW34pmN!4NfKkgMN882Y}
zCf!jT>w&!0TbyQ718nh>JL=iaG%*cwWmFW1B<?;?Q10KH`V+C!4OSy%fD7H_kvCyi
zbdQQK{y?$Z6Zg8Py}c>!)BFm;<nCa!f4>(55ls3hrL<_bAx{Bwl0y#|R;1YIyZhro
zU!@Mrf=2S;dPwU_{FHcDbxXeCv~zgGn+fsK6_4tJSZMGHkEzk@KO%-TgdGrTPk3Hj
zp2V(u{8<QvYCPqdJ)Hi%1|uH<AUzUKtH<9#x<2E5<Zu!cAxV_|J_A4hRUD`_OtK**
zq6a-lCn16`$D0<Ix3@JAEc=ivFGB#3P6Tn-Ta!L`NMi*MYH~P9bwgl<ayE_`hLqPK
zYygJi{tHK~1W1ONRPPG$mQ_v~%%A-sPHCWYui#dk_Cn&SgQ8NSrm14Qq2HvQLrtmf
zGD+^(NOCZ58Q8I{X;<^M0E8~*H3E^k%Ym(&{C9QXr!c+sG?;J-*4}q5tn54EoL4H<
z1}S{i*rZMyi4R>50eN?P<Rh~T0ao6dj|~Sz0=M`Rm$r{*Xl{b{xD>kw4#sCHWpB#W
ziO)3%l05uXG)U$P8;dVhfWaeZ(%UpvQ+j=^n#0K8w#GMxZL0J1pW<7;+^imgnPO~{
zdYkXuLmnBtS9;=m!BK~i8jkAQ<A(#{$GU4W$Mkjj(F;;^JOq99Kwuc6_&h6=laC^Y
z^h8|ntnlVIvR|?wT=;B3Ty+RZFM9S&!LFMyGQC)}phX#p6wl8VY>L2YoXG5#>MO-x
zXmKuchr~Zm3~;%6S$2C%N6b-0B7~sijVqolh!J}oo|7x}QAw|ixJqpnezNo7xcXT=
zjvZ`6%Evg5*LVaH$8!$jT7g`d+!7THP~l)9*_i95BxmOh=Vg5<<~=L20JERrlssIg
zk$!gl&zHn}&v$BgIBwAIev%f%0)vKhXgC(SgMXNq-PcvcFGYo19yb3bwVK@vR=Qcg
ziQ0ohihifVQrxQF2#Db*-7^y7p;+t>sPO(+;wh%jGUt7%MrDi)`-N?->So+CtZLQc
zstml<sCL{(g4Geo$LrBhY=TyK1ajpKMYD@AY(YfVQacue0z3O6LYP*sUCFdg04>g1
zo0?~Y;A(W#8V~`UhmNr5h8X|#G*DgZ_9*sJsF@C&>$FeMi?LoH(>cDHnj5wxaxvMc
zYQsH*epR)~I*wdVZ8Hzc1&bkjRSZkk8GYWD7R1lVhHZAQB`*#08FW`$(jmpG$73`W
zBhQM{eg#h)<&+uq6&M6xYLN(TIz0PaOzNw|G9`~l?G(OH9{Q;od}OVM<A!XA9U7B8
zMC|I?w?TukPdmN(b=(*t`uM9kU6kS;_r!ruekb*7_xgx8tc?3SF&jt6{od5EVJM{C
zo@GOGb3EVz;=GQ;gWfL8C>4_9A^%|oimtKrpLy8TZ%Yg3M+DHCq?5MAqw2NVg^liG
z&la>_(t_cgQP9}qYDz|P&eanJAwy&B5!lP_Xp5(w)kDR*la#XLQ+sQFL~fFAUO7ll
z*A`;}@nk&XeN72hUZwrdo=KBpJPp#fofj?xio^j8)VYOplpcQ#Y}m8_E7>OFke9VK
z5F2#Zou_0fAC{x)hD?GnqoT~_UbgpG4b&$g=h7_w1g3k(pDhGw3;CVBdc=g>5+@Bp
z-|W6PrGn9FS64MotN&qSBxm$@<YBybZ@6{Wb{>c~{c-mH0O4EySU(@F^4n@&*|;dq
z>hCF&tH!(fJ3jpJp6jnj73cf4o(_SM&v{u9(WHH>k0boVX#ckO&<jEC;8beFN9u93
zzdt@!C*#lu-SLS}4etT&32<Pa)>a$ai_i2M8upd=T(=5yh%|-FF28Wi<veF!Y7+@u
zLinHY{uC3j`d@i^b=%@=1DnVOPZdMQRe%pArL*T%(vN<pnLrvB;lTHvSWC6YZGKQw
zz~E7W_V`gytX$DVcsu<jAtyeZ2Mb;xzhd|HG!nkxpm1MvbvQ26Z`jDmxaeR(e3eZQ
zv4k{GD}3j3>}p(cu;9`<bjN9<bg4cOgLY+<-^<j`iWEv*ez2I#QB-_pxaHDp%{f@m
zZ+Dd>TnB}zdxi;Co2rm2)gHYVq3#Zfsl08J;GkC<sLme3?#49&sxpSzU0U8;dr)i+
z3g)yU1<Z;%nmrC6bKMcUG#K;L5w=>HM(8>NfF$z}Rz<sFzJROTi8+s&O>hWEPwDX+
z;1jmQLKR{6$%AyGmjL!6-Pdk9C^_Bozj5Zd+3<EF*iYrlEglwEjqPYKki)wBG&v)@
zLG7!=*jnuVr-nK=6Tozd=8B9(i6y3EsVg8z$#t(+YtAUwIvTVPI0kk`<3Z6GC>==y
zQ4H*xY73A_?q2|jtTL$W6YLrot=S8*@$aKWKT}TH>RO!>X&BlQZ60MS3(%p(xpQSh
zM^mivR5DaRXRP&C@8AT|*L8*)g<_hTu6G|DycQeOe_Ji^)<%^9P)~Guv+vCGN+oP#
zk2`8ZsLM6%J*am$iRC<{<>vJrEbrJa{ysJfK8Qg8A9QDs+~D+7eGUo6WcnbcMhw(u
z=(kZrf5&JV29K*765b)@m^1{O2QmT2HZKpDFXL;(l<HQH`frIHIw0rGaZ?;~w;SM+
znc+iYr-veRJosjpdndqiYOU^ZZ+%3li+jC4<2`Yo=dztsP%Sx7?|03!58{loyVk68
zY$zVkzLU`j8^C;-9@H4alQ?P3J{u2d-b9}2jED96&MMN}N4y>@S?5O$<nm%X^O&cw
zg}}%uba~vv-M*}OSv=vwPS$x7XLOHeG=`{!<NuT<(Th}guR6u3Gc7}(_Hb+0u8n8B
zt?6V~^boJU^G~tgaDqDHfIG#?F@t#?^m6b5^WvwK=(H-o^AP9!kPm)+71{y*Cl0$c
z&vB}oj~e`<&6)AXFswlw(@7fdisPD9BQGDuMaX1TjdRC>p43FR;&=_kDX&!_cH^`M
zN797=jx&8m3-CMKKi}|J9hq?QO?{+E0UAnua96<lZ@a^>BFR6iFJf?{mDany=Ufjt
zf5BV|=+(pVz7I_MP@EI|JffLl1tVCmc8>g^`^Q3zXZ|A%jmI)G`&d1cbd_$=-$R})
z^;TjFactu=_s~{SNPX_F_G%TY{x7@$+|HpfGz0OaL2ezL-8)vK81a=_Aihq6yszE9
zK0^q{HyRfuMqhmElJVK2Xl=gJ7er<T6aC)jE<?0sa*!Xi48)H5(aVhCXm|YNk2_cL
zWW{sBJA{!(NBNvERH}W-iM;T+^0=VUg8Z<&I4*ilD3iyV7v$pSgpE4RJJ?_1ztopd
z5-Bde^f|Gl?R3Nqzw9~jZ0#(_?7!l~`CP%xILTqy<EnWGU{S8nSYV}8IA5tk_<?eZ
zuhKvn!<Fcj8O-}?^_(d}$xcklEzPb_gK_P1LI|Q4xm3rZqVDfZ%LP!U;Al=9HBY};
ziS2Qnepebm&GQW(3_DL~-0)mM?{gQ<iv?au@__coLa!0-w7iHndPS0^7|lZ(H+czq
z*bu0VxLN)DjDBuWKdI6r$(CWp^qRv%=wDK8sp@a%EsVwLCACsa>d&#nwH5uNpuDB(
z4SNb~TI)5DemvNtLBF@=^=kB_csbK1s>#5oO+1D@>Vd#sh!C1Jc>T(bXmPEfiKMSu
z-37Y5jOEv+dduqSqC@RxpGj%Tn%dTP_c8KRtW~`V_13AsG_hpj>)kuZ2fA#7dxz@C
z=W3%`jp2mer4p=_i}iMo`>)Ty`RGl1y>)e?-O=YR@5EwyC^qY8S9~c;T&NgS70_X-
zQieQp&??%>=L$-#|8<NS=Ju5_UR$_-g$ddjlOFA^>lenh+Fb%Yu<ryn@bGQbA=ro=
zJ`#N;jbftByER&+HDjFZR8O$JW3kH(8>(YyP46*?JKy<d1a|j&l+2h{_U=9}0K&dZ
z3mW$uq;aAH?e<RWPpke1)DM0nX@T^hoAoFmc|4>!ZmRah!>X7@5!~HJ)Zf{E<&}C=
zmAY}6eas8d!9#(5|8axW1`V?X_m6nO<)|9-rtL9QzlOYgN<~`>9w)ScfXt|sJ@K>;
zH1w*t$s_U1bD}trn^Z?oerB+&{Q|N0U%e6s^i}zXeCgw$zle^=E>3IU*(W!5#v!-D
zXMnqN*eihDCXR@maa1))nKT|V`puiv8OIEI`yH#}xPPaS<_Z0N{aDXboYcb1?n6VV
zeNL8>{Yl0BY47tg-mN&}kI7+}wl`|egw%#8+TW~Q=+yD1ymu_#QVXmG;6LB4ZGSPY
zM+bSfw$6jaAmv@R2XgEOljMu{G*9e`NX!Fc_r4)0nu&9@{z{rSID`*eI<bj&@WWc^
z^E{r1AuB%8oYCVFEmptR@A%yx;uAwbZ0D31eyZib^Cv!Yf4r+0@@NAPN~Ce}7oI%|
zNL0h=;_#)RCjlm||K-X5O1E$`;bLq1in^ApTKd9LIGqN*bysO)h(Q&^%6A%<n@n1q
z?+sq(*mhFJ#Si{uhe*B}Kl&gbio^b?D}FM_jmg-@bU~M!yYgATV7fg1Z7WyAh10@`
zbuc_*aghp=j=NY5DqxE(afxa<Q2V8BgOZ&^x@@{2LByC{ak+k1R<zy`bM$+v|MgvQ
zh5DU;!}W3Hv{0&Ig2hz^rS8{pwMOg$NfOXNW3VX6YF(>Z!=n>XaeK_$(iUT`M_k>#
zBIfyb3-b2s^qc49)tK+8B$FRExP8W20P!wxn=znYs4-`sWhckuMpYyRcsYJIxfAld
zpmJ1MyG_+^@q*MZUx9P<t<yy?R7pwzl2x*^SfnZRlN)lesxz~cnJjVR?qR6TSZeUv
zi<}_p)vAIxslg?KnI(-_KhuR4!EB|k2*R5*5hU83u}Z&D3XVs!;D)xwPp8F`K)5m%
ztsb~NiT!As7CD4heLt4;+?o#mg)^OwPW)w!K}b0wFY#Jez%SXZ+UuqZu0g}P-enNC
zsqq2e4b%4c%S6HH9F#FUNcoH8{E^r=Eds9m;W&>?*ZDOfDns-b?p4DmW_k@b*?7qi
z*QXU~s9zSFRlp7ZXWoZFFG4D2h735N+A_t@h*x!_J4W3Zav5&hxW`$Stj46LP~sXU
z1r*y<oNPJVqFQDUZXn#@ei}-In!B|*7?WYeiGpXR`yCyg!VAP7jcJi|kJ=S|6afib
zl-3_{pF62;gFj{g@AnX0+cWyuZhaL)%b9q<6GU~DfIg_E$Q~dEAQvCfgom+`e7Lr0
zMYNKP_#;{<@_!7)qaK6oajes0(_+!d8<Mf-^C5#{RP{;&qKYTf@8Il{vBzurv){#2
z2AmK&neZ?68dQa}f7+ndp)=qrJ)`ajnnN@K!<K2k2))k{JK(NTdY9&b2h|wt%$X!(
z^Ry}BKcyTcg)MPNaID5W^su`|>K8|Sz({bDK{&f(M)mlPWBPgek3Df*2Y6^pobXnx
zNRR20hB*l;p2<`0EN?gr%SfE|<m<M^886Qm^LfL6?wg1=4Jr4PxnZh<5l?Niu@i~6
z)olLtzuXvSwO$Z{$sE0_=cJiiT4)f^Jn#8?ZVZ$C=Xl>AE62d|=QM@iY={pu{17P~
zK6DN8aQDPVnhi?NVq^cYC)XlO^@*#t=IimP&%*$kjIsF4lYsQg^N>zUpSw1vVM~1B
z-6BxhSh{F_>9s)F8ee&oCb2AE3$|7i^51w2kRBA+x9($NXo8@~gYlgw0kiwPmwiRG
zYj!VhO8nr#yV6bgM}J^0Pg-X8f^jzDR-~%Wi(Bl=crX_{FPsMj4gGVWzA`8nfle=a
zUgFmw=f1x9c^R_sj{P;C)>vHP{x)GR6PG@(Q^=9bAfu^jzU=veVbhBLa#zh%)G<fh
zD+~$O{|Xf@-<C<cuY6vp@lMg2S3O@I(-mvtYPBln8|kKZjarFLyw<bXgzqg9)#vq)
zIyw&7vjj<9dwa}#z96QgNYg4H$vSjb*QtHdR1gvK-P>k(;Gwv|y$J<y;TAl9rf6cx
z-W>}KF2dgA(QXt_yva1Y$=%?I3WapDYHqD|m29fFxM4cIa(Zv|Amq@@kWHBCXu6fn
zFwb~C#$rt&(;+X>&jHMUmKv0Ho{4%7w{>_kCP@wMAIc$hp^e@`Xz?UwYrb_&g7@`Z
z6R#hSRbGT*n;y*qpn-@|gCyFH7LD3Zq{ddYsz9A#+-k^ETa;SE=r9~D{R3#I))>M}
zvaq%`C$M}zy6fDuClzgDh%nZ><K~sOEsqWUXiN;Z@%&n*e_5A1%5ZNPUam*5kxndn
zy*R^QMR(tcH8}}tFl`kqL{b=yLDjn*W!R8@P9Q%V@e-r~*`H(7gMxWDnd6?x#-IN@
zCS3wcOJCdcl_-&~DX)A}6^ltC;XPkyilOS(RC}a!;%@zgfQXs%G4%&K-Q+e*G-0rI
z8E`8jh}`4yLJ9WAz5dL*3qH6{3zeSqopHa*n$c14`tH`;Dyx^r1BSnz46YCA&&}0M
z@sO99Fl5M2J+Bp`X%zK{Y7L`$eN;btc%>iHm0b8A;_(`sxqB%fctW*S^JMMO7I7k^
zv3p9x^wJqYdTL|xe#X;k)IB~C&j>`4Lw3i0KPY%@)KpRjRNR&sY7V;p7p%OT_Usz5
zCp_de;FwRp^Kcwi)gl+jZ9S^0TXXAXG~pCKCVS|HXbO^^Ic_jlCB8f1rGdEMdY$xe
znZEv%Pb0D;MZwdAHn9q4Tm(Y|(wm4ko)<B}qce?S|DSk*rsGZZTnyj;7;ovTKzk(K
zR$tfuoTSTM5K3f{aw6VU>0R@F9`6}ea1vQW{DFD$c$^cc2#ay;@c6^sDd@5>*N2`p
zEbS;<?MJFl!VZjb3_nK|l8S(MDob%zeCo2*I8LVT<Yxk$9Zp{Bb1xSWH+V9?cwT(}
zUsa4CMlTbUzw(W!15bfdIn!Tj2DpbMf9P|TiZ5ur6Y;I*H(DiV^ml?OBdW&th6xjn
z5<mL~l~ua1|M^j6<(=aFi=RACG?NjB3Tm9)hs9}JpbxGqt9Ricu>|NkHlxJZitVcZ
zQp{yjB_9_bD!6obf@ci#)kDHyP-4V!i%Zp8rD1vdvbgM!gs0=fL=uCXy<9CTtH7Gu
zW6q(1SYY=+imF+pgZ~J$R~{;O3)I~QO=3bcu5$Y{AgjjJ8Y<PsFpr0X1Z9T>8FQ`b
z;ug}y$Q2i#{b!!-G=QFaNC-SboAM`Qk}*&4kVCE-*LhY%u*t9t^J|Mh$TNAc++g4#
zEn~r<g0<W4jD`B^@FboEH@b^Wqq9%3t2fmKd=UEWkC5nx;^x|)o&-k~o4vS2bIN?-
z<8iBjMpJ<sAu7^EUe6{RnOAca7i+jq?sF{Bd}oiB;Lb}A6=LK!zr8-{-3XiC691(^
z70HE^;_gP3BKt_$*QRI^0I8=!oT`a+VB31gH#x<(xT~;e)6t^ge?P&&Y1Qx6G$d~G
z89<La5FOsL#7U#EM%8dgLFcI7lNrc$otreL0A#&BqzvD1sN5w6CB6VVl~2gV+QMn`
zQoPfpYL$`UkqL?1@s-p7m0xpxdj$sY^&-9M^YQBDr2;LDrz8fm9ZmN?{VWCr!Tjqt
z{&s#08Md?k!nqmIdSRL|Q_j(74V0t&Ar&GsF|M`m9UqCw+Ob0>$fEr*ws|!+6<2{N
ze@!CR=!_j2XE+ai+-=xLP{;06hp#_}m!w;a@J0{EJ?@^z9Kj~Ld9PaaPeC=@=kvFk
zcMgu{ezi>7Ahf5ueJEN-GqvMXJm7xG<dwXnCza9+C>boU;vx0cmBHjmHu!LD+G&B6
z#N|vpVqj9$jgX@!9`%+aA&`cmxwVf8K-?=aa@Cj+@53bJe!>$T<Sg&0EgQ~JMGpCt
z|4=u&sVIYEuht56HPV5nHJQYJiD}{)!CDK<zF*sV$Fex!oove$KIj3*$&fkig8G!8
z9`Z+P`Y;?mBxZnFWw!P7#!rRZQPp|<p6;pbG2_eeH>s-OFtup|)oL8~;-X;TrkwE7
zkEDK(Wjg7%W%eI?N=OWNO*QQv*&e3_BqZ^8oKZtiUf)1!Lf-INqEL=E1y6mEqrc@5
zupdEoiIuCyM=_SR$!9elp$^g;EklOCtBw%;qmDGJq#8ir@xIScQ~!86?wnI0uQWbu
z@qr8T{xgpcYeypO|6fk#JU;TaC1%L08y{<CT`&lr=;y+@*T<*&Sy|m2pLy+xtwfMJ
z|NS?=i!a<$3ErOOEnljvN-mtQJY9%G0`cJvQhTh6&r@EDZv^)kS+JSKx2}I%Y5|FX
z^_|8{O<gs<_xy=3K%CGYJmCzMHWNSUAl=>+KY8l;6msC=g}*^2833NVTrU(PS!l99
z#)U5wBsp5$n!zvPq8Ef0mjiMq<o?C_R`44pcgYI{Kc~4p4k2n=DIS<uxT>tPc)ncj
zvdM|TB=6?DAnZ*{E7A+31@;vJ!``-6dfYAP_{um}y>KR3yE^`ft6vZTC2W;qQ(vQ|
zm1UVRBCd5)om%1z)P{l@+Emz^xi1tf*is+!1R_#hXsYX6@iqc*6D$oV=f6;J)Ac;9
zHwb{<G!j_gMO?-;BlK_T<`=3Me*pB7o+|!}!))431{d?XeEo>TZg4XSNHuP8F?z}J
zgvG6bsX~;KYAg~2zqo!uEUvBYD3N9dV~Ia)p}W+;L$Hu)6tQm98%iaY?asHLK~Pq<
z5@4rMP_*2!x*k*cCiR9bGE_Y>TD84nWk<?UimPR_>StCMR;EQ?=|_z;v(^1J<GtDD
zdZ@`@vN{aq*?&*wHr8n6J0@bSXSu10x^bP_RHh~~J?wh7X`g`M-X0s&d0or0<+0JT
zM2(xN>Nz%DDrFV$6>*<E8NIgGeKzxYkfub{xCMt|vsWCwY>~^wpnK){ADJlR#Lx>O
zi-`?_z(%5{5&x~2N#?F|EJth8dzkA3KGkkjHvS<dYbDs>a>=#{w27`1OHCO7_(BCJ
z++kQtX1%!E#cAhJLYnM+K`;BVlym(xY|AbUac8xwY|uS!rcgJW!+SL*@-Q~+J`I6r
z3Q1AiUt9ftT=aGu)P`z`4C4XapA;Ds&>z&&ReO-JKcsfUU>Jn8DyWR$9Q%m-=$S1c
zw;uJ`1Hp2Ai;n3rHBGzYYCP`o(|bCes9jiinKU)tqkfx~*EPgb-WoKhj*z<7O^LQ%
zji(Jl9f;)_!wq4B^!`ZfSH(1;+ZG2@F^SVS=!NPfXF_^~#<Y9sfEGTai3~cW<Y5&J
zq+aEyM{Y>V<QdgVTcqKht~lnw5503A!~Zdk*8qcCJ>lI&)deei(z9((r#?h0r}Ui)
zhj$(Y?9)0!Rj&9Mje#-=-Q8fk;Wb95n5NQidWfdg?eUhEiBt{*UC2dzTipy+H^*63
z?BP|2cMVJ0Q<Lo3Gs5i*4C-Y;{MTQkr{^?Y>RvP9;s<V;A|X62RU2b8jGyvF_{e=>
zW)6Gm)3y&_aJCcx#f0Hg{RSJ3#%CUJnPMW4-crqq+Tcih;Q=#sA*r6dZnYKv%8l`a
z=<La09$)+Nt!#^LJjmD-nJ)F`J`VbK?r_WOvpo0T>#Mk(BCU1nf%^E-JCejAT7me<
z>)!qk&507@u-HgQ$05vKFF0I~$@+zH;o*WUZpS3>qQknebv<0*l!fUIT<oTh->8Y>
z5`Q*{hrjf2!Eu!}JiwQ!r~v_fqQA=zi;=}e6BcC7VZBaRJh0VRZ}Hq+AuzUXi!0sD
zI9q?!VM&(ZnQ)^#TyXjAHOlxJeRIhRqZHQ~oGyH~!R8f>iAz4hw3zFjaJzdU<{d8B
zdkroNrZTV7po6?GG2acLYWv3GhT5D-^!v_#iv_9(w(E?A>T}uf1f<`Mf+j6THm8vM
zCRM!at_^>Po89B8;x>DWdL%1iv0c7ZGhcaIED~6iv=+~6k23SH$Q{sjQp0H$xx{@n
z5Uq>E&z=_U9T@WHL!$l$H-#N4kh5QnhYR76VGrb@)#Rndpe$CYUlgEu&zc1kA<<iO
zDt<NZ57Fu=CZ6e>eLmXMeDmo)c1MSY<2@S3BqRlGYXlA`0_EshgPkFbJ7b+X9mb|*
zz1J6&SdxqzJcckAnM~A<*F4>FyA1FU0Qa~@-20e&uNQ)|O5_3ciQiHNp^D9JMP{8C
zbiKxuS`B$@Btw~zXv7DqTpKvuQPtfH3%xDIHC%F+aCMUgvx`@}mwni#A;_o)dQ?rE
zrjyLwp_&K|3au4)yCqUQNHij=YY+%H0N<siF1kmnRB2ch_o^V3(TF+k(*^pQw6VM6
zenGG-6-&F-aQ)l`@qijuAW{kG^Pno>-qjfo={L^M)p%G9l9_!(f3CFfW<6?H`*>~h
zI6kH_GDMHY<6h&WF3`0+p&t5CcJJ}3roacO=u`gWNRSXW_UaH;Qp#1VQJ(f6h`E#M
z-qCnQAR_@ySGfJ&u))lJbily%AXzx*6PJ%qG3Ov#iMt|JS7u@Ai9;T{tJ;UmJP!Ml
zwDOp2$5DNP<cqh}jCYFjk6{TJFFpIrDW0=G?(cDOxhBRq;VC1;I%!y}08ML?Q|i1H
z0^zi#(O;5Vo$&+6{fD|^VK}637`Echi`A<J<L)ZK)mxf5S*M5MZ39A-uXMMJvwk<q
zGn^T|;$02EW0wqMCiy!N@A*v!C%pR3eJMyNGL-jyK*o86Npa%UCjSDtzz(1JzhPP`
z(8w7?xYW1BRYp=t!P<v<Cz|`?Bg238sUn^H*f6$pw{lZI@i6VI93tMs7IGb;U*cEN
z+mZOp)4*ARM<Q4Jb3-#m=<E1GJr>J`yn<h<5Lp-YreCT1VXTf>%dfRFumAHveB*J+
zNWd%ft%o4ruPsNuGuZ9vG=Z_KPEfk6rjmp(v2LNbQVG<Ls*<v-R3Z<>Pa17A(v3JG
zd^Nidn(~4pXNo$HILmS25%Hl|br!P!q9a1Wnc9R${bE<e$vG}jRa{`!;MGj*yCYJl
zW=dO_QjO4Ad?+q=5iiSlf6O@|zB0wQuz#?mE^bRjRa~hdxhlAGZjY-B5w+7hC9YQS
z!1(M7DL%YL<KRWawY&C6p$yW5vm!_v&;Z0-bx}+E5c6uYNE0EA4PUbr9p8MFfWbhh
zt~G+1ae-<LRY4h}vG9nnUFtPQ2r0%(cB7|2IuLZ!H+k~Ju^ft<Jzc6IT4TDmcuiTF
zRCL~|&t}N0+_pvD8j7Z3u|9H#u@Q1RL<E^sXQ}(e?*K(SE$a<%eRUuj^!u*sv9oP7
zxFzEcgckUz$z9phdX=GPfjJz_E=;W_@U^&Gn5by2jfwL=Ml_?*rouGPjSd&)2<4)z
zakr~cF7?G)&83|P+BjdjZdV#ICGhKAmwc9-tqrP6wl&V}#v_Fas@z6UmSXY|U4{@j
zey02DF@&pmZF_YDdJz`&xfUTpw^ldBX4hKTKteE`2>7r>VS_Q`Lze7jnsSVsUl6p~
zF{-cPv-?Vn8+4v4g$I~a$II&4m&Z1bSz?Wh$CN*v3TTI$@@fxcKDW60i2SY^CuaXO
z(|;BB29|ZF55d_-;C6SZ6$C$u^EB$e$EEH4OzU2cj=NIa=V@R*UovLiuZlxye-^v7
zGVM)u@qm{i4Fn(b?*Z(-A5#0l>hH&JbsCR{Yg5fIO<3s0Bc2Zl2a<Mt)NK&~$78OW
z^g*$Ke%ys4nEpL+M3l4zzlA{DqrviVFEZMvRKLWq*sF1v#t<`!r#<XgXK%(~ddBm|
zL-WtE-=~9<(%qLD=s4ibBMcwfv4d*M!RbpMxoM3Gc_Oj#A+5?J9O7XCfHY}a996Y4
zZCqPp#-m`(L>MCei(^L$KeC081%?h9D{bgbc);Qz3d&Wl%a&t%aY}$EnKBlqHFBlp
z*KtOFubUW07x=~z1##hgAs__;TPwtyp7x4#!F{W?B#3CCa^9{zN#rz7OGTcUvzh{0
zcE|)cmv`NKB<(fdt1UKH=6wM|cCR?6Kc{bAeBdR3$4BD6D?ZeOe>XoqQmx7zZ7btr
zEjD#h#`%etKApjc_TwYL{nuiD=2av+D#@Y985o~y?E18z)lYCjd}&ZYU4)GF2iTcB
z(>h266C*Y$rX$}NoH1C>Z$0TERYhO*odIHvSn}^x+Sx@OFa%KX!x5qPlzQ}&C!w6n
z9}Ogr!=Q|gfAZ;Fxx6@&7p6Uq3NeBM=qxTcD&)7UX*p@kFFab%D?A3Y>_w{CUqbZs
z#l=U<xoKD(m#9_+`>vk2)cti$^ng+0vZKP0DXlCv4woMlVkXMrC}=e194$z@a|l!L
zD~?*&`~oKi^$YaQmF_OzVNCz7@=y(EUE}Jb1xs$4=#Og*Hiqwb`dnM<=Xerg5M?t_
zQA6~N?B`t7URem)n0K^1%N=CC>FjCEgh|&4rjlWBz6P0nb|7xhjQ@UaEYKgTL%p$3
zKXEFH8x2GkdA=|jQUT7iar05V8f_Aiw?PZVEdpqKBI9h|s+Po+Om=;d=gcFJ{;7*q
zGyS&Eg$guW<OXA@08i^3*si=P^{NW}Ji9L%YO5esB5!;&y63ZdSe6%~Ng)1g!GDNV
zhP|~HHGeewD|34eM$6xJH2u!KWIbxtK!}{-JKJ1NF2D{9)QATj9qM^`TYIcg)yi_h
zo5foFJuUUu)nNIze4JvvtBs_E^@iF9Rs^w8z?1`VEV{G+6Dg|cakct1km=P)D>BdI
zxTDX@(|Z2=Z&uq%Cy&LT&mo2k5TeA6bziM@xMPi|Vn*zbQ5Uo&t2yq4>Ernvj!A#v
zJxaxG?tNmsH>T7G;Tq5U4&Q;Z#bBRpxZ6`lYn{aCPEQcG1n|Hv0hTrhU^`5P3}|;D
z4GFC7?+mJ%1jl{uokR+0UPH2Jt??3~^Y1noJ)~rf2RspcI$%uwK`8L4c*q-=;@pCn
z$@t}Ryrgl!^vqsAzC9lGa3Fak7Yz59{{Y@X0nOv|xMxTDrlhQ%P*2-NGOYC;^|Yl*
zuTQC9c5fo}z3wG*#up;+X)j7DpI?q=^wl^#dK(h{{rXDQ9He?ceG<ubA`bedL0S?|
zHW3Jz((v%n$L%`gnQfWfw<!+S4k4aHyf#N&#$NZuj6M{w^O#HOT2~Tg-5-e2;YzjK
z%P?v`slQ`237R_PO3Nm?I^(p@qQu=Pp*!Qhq;Wz|A(-D#gR$`pJO8G4c}4$VGF<VN
z*Jl%~Low}qyLJqc72z11bp>>Lnayo1-qmaMpMDkZxoB(z`Cn=w-gny`B)7?Xoby^2
zmNPyO9Id0ncz%DVqDp(=cs}w$m|`72*3u0mB?oc)iRROQXKQL|>Ge}pCS=cGe5OUJ
zG`Gg*>Tx*DdOG6^_m50~JN#wsr75`rv5%UGuT%jd0^h3mTCnj-ax=bBQO4&Zg3Pyu
z?atQ5_)f5u^c85)zxP&*XGFsvRFA3m#*c23<{>|wf5noTF7#=<SP)Dr@9qUJigv#}
zX(rAV9OjD!H@2qjc3kvgL7PZ9(ogQ<7xmuPqJpG)>5><X2%$&$7dBJvE17~fE_+c}
zv9;KE5Z&j+f)z8YXv|UHyvw=ISLmy7<*&q*?u4ut|0S*xxRX%USHCENpy-~QJ~*y<
zQK)`!_E}5{<64ce{w`Rnihe`Z!ezQz5<+K>#ykzN73)(nX}WBbDe<z+H`pkQ`jaTV
zLEUT`#Q$W0LBp|u(CYr^KrJ6Px>Xy#W2j`~CQYeCo1TiBy-?@#y56E`HLPrpTeS;C
z9y(sqOe}K8Bf~^{TCAFx({G8#xGkR-c73V8>xz?VXVk0e2!;p^`q^IneKZRGN*!{!
zCP7u{$Vk<(N+n%st={Z0Z-a!17B|Vn6RjF#n3XDsyv>7>EUP$9ba+>iP2fNudFOvY
zYOc{Z#8>=10gBe1-`=Et)_J7foWgpqJq9}DppOj#du3gHY;?(;h=PH<>%~HuZI~ZD
zZr_#R!h8L}@f(OfH^JKlL3eC+SzkXHqXzwPTR!hY{#cV?`$xPctFe#okI@%JuBE<S
zQO#mptMSfbNOLFu+L9P}wOv(wof9#ot}tuJP!~IDw<F`uribg@we^Nv+v!5yAc?7V
zsb`2-w(lM<@~Y|(3^-a$4?BQNDDLyeCK3R~{ho7sVJl;|uXK^(QI00lwPEJrCj|j!
z=$Ht(fE*EcV(J4+1au^ixTGF2Q#`5<46Pw`9;?mx`FH*a25)yf?%^|u0Pp7$weo*0
ztkH|H$IE&C&JqGob#R(Y2_00*UTu0e0bQTgysMkGPw@l|#4~<$TK}5yH}-3h(;JT4
zdO!mdD})TT&Y3#sHAwpma=P)pOdHD8q#-4c*CCB^JKMq`+!}{HUsfA86xT%_)H5*^
zGa3!|z08$#%=?1o;@NzykNcujN#}aPy$}T$#?q@s8d9*P4^X;=pZ00nN&?INIOG1>
z!Oht7#2a2eE)o&Q;OpL0^~~XgcJeI`(U_JRZ>uhmP723{1Xt@=X>5ykU9p%NzUM(2
zI82G9->(&>0`;6n9GN|q9z?_Ofv41wu?t)i@hd(lUe=M$nlkibFBLJ<QvUFXKUvl>
z&cdexXB5i$GhN;l9r5}3^NE>z!SP>sK^OxW-<Mu0ObOyEm#ikbC;6z|dlRBJ(hPi~
z!j+rSFe<+FEYM)3*%VHhHPFb!18CoS2^x#E6@2N5t%l-@A3e?xPaoN+RTZr?B27&8
zV@3$3d14!6O<bVQkxs%d0q=z~!g1@6{HA^8MKeOFCHZO5Ok6x8Y3{NlA1?7n^91I=
zm(G+Y4aNLrGh$FtxC8s;9+3#C8M<SRKQs6Sx8;hNg1;*xo8rouf>2Qv668mJPFK*Y
zHAvb+jKwt`5q4rCu2rSgb*)iRul*y?4De%fXN1X<eN2nMn5TA($?I^PK^UuUnmslY
z^F8Smm~Z9lbb}{dN9Ml%uH@YpsJog(X?rZ3(X*OH2`R9E1h~<y(2Q7ec$52uqRR|g
zakIYhbrk7xi|2yv^8InEAx@|2{(M~*xs(_S@Ru$HD;?o*J|9ao^#L?COTB`1jZv>Q
zD_c4+|7p<QufKj?G<rk0qbRnT1bpUmKxR;=s8z1ei}aM7B&v``CPO{Th;4vFiuR_m
znoO~+{=EqWeVYc!ct;)jGY?r`tkItd`4VeAk2~R{Vx7nNHQXQ`DBg$m?OQUw+6Hxy
zk>FyZDxx$RB4SvVo87@J!?4!tdIwgWnGCSk@U(Lz`aBw{ODt;rarS@X@HtppLNaxL
z+W%(^3C6m$D`UhJTQjEasLpX6F$?_XzGRCg4FeJq3{$qbXryz5cm*y(&xl#<4i}M6
z2lta1D|P<&d2-@N?3@wpS@KDj4Ip%Z``Daend2TGrv4$WS>b)}^;#ihIf0_*K3C72
z!dTa;Fh%z%v)%2f4%ZYH4``}gP<an(v1Xr45v(CaL;yFbJ~Fi2FfzYKJU}r8iAP<N
zp!;x>2E_)Ak83;N{f|KO%)}E~zxBV2J)Q-aD{WY~?N6y!TkG!?EX$EX#nZI~P0{}|
z-r;%}rBuc4ukGCnxp%}YI-uIfY@t048Xjct=|VQ`PhQ}R3vx){%>6|i)+#JZEoZI6
zCyO!pk(hC#L2R*ErDG~j1{GHs=6u`@5Gf2NS05(~FQ|MlPHH0UdAm;e=pP@8)0#gp
zr!fX1#5m(t13bs^hG&VhB1+(~cys1_MD2`(TVhKh7&{Sfc`ue_4w`t|%?STnn8>pN
zZ5T`cCy^|6#=DyFl11^J4;m`+66KHE9`Cy|&||eL&KV$N3={EzKgK4I7=5S@P$hKf
zN19Y4EAg=_L81^8&&yJjIpEv))V*U#z;=J;x|xR}KCgYu9z_7Fj|{xq8DF}D>r;)d
zYCVAZa82>PoP8+0QQ{kawpCM^9p7qxtGSbP*hm!}-gjz`<ue>ld@sPOC?9?>tQZ?X
zZ&3E5?q<?*{ELnu@6J!P0Us{9+<3_>A;_~FgbQBM%g-xN{6aVi7rrEvI5EmqyvQHe
zNDsxuFBSBeFLpV&FUBP=6&wd99EeL_5}HhdBBC0+B*dH!;~b95RS}EPshIQ9nUYH$
zG{Qw<+ydzp1`W!ts}qx*xKcf&W<C`vSGk7~3^=d$P&`+V%GbP9a4wUMYuyP(pXr`C
z9Tk<;t!|FFZZujvj_0|%Ww$rSb($oBquJ#7o-tnc1Uq<11jHQ!=`IK6^sF|+M*`DA
z6(coBw7ENOd`T}M3e`Ti3>0E0G&O8pchTD4><&^x0Ref7mp@svq7vH~w|WZr_oZ1A
zVSn5ty;l*|Ef(M<P4^P@-G=;=1F%$cT~^l-_4+xwsVf>h(Pl*ajp}O4#CQ?JH@WCe
zPUb4lbY;fi;Mr+bSp!EpTC^m*PUzQLH72s*d^g%OPzIUlFyP4y3~qwVSmTu>_853m
z{qq2%g7ZYI^LXtgZxhJV-`tXPVvP+NA94rmS=-eTY=<|wtM-R<X~;*q$3uZ^#-rCC
zV-tMzdDmC;=W*WbwOo!}9H(;7HH#2&$Tf%Id@vMx$(Sysf$8j<4DhIdOpl@*Y20w6
z*<^-oipko3B1d9R(`<j6+bDq=VGjgCiLl5Ev%_7G_N;qq2(!Ao4HMx6kkh&p|7HRU
zgI9J5NU&2np4{W}_E(z}m{D(P)9l{3PrHNmF7EfAw<7D_tsyXN!0qV)7m$)D9`x_V
zVX&!}qi1Wrb`N``JCXt|x_r!t)cYN6E%B)5ZE7)}y2ok@_yoXm6*Dz;JZ|uh3AK8-
zc0}%bRI-9_|087fQVEucFrb2Vcr0^uJgowJEr#M5jgRUh_Nz5sQE3!$KwpjhB@+BW
zZ$O$&O+X4vzf=hOU;I1{={>`PuEx<GR+mj&fTL=bk2vwUt1)9>SHLkH)9>H?fl#i;
z-8~ksMH_!Y-T$X~anh5`BRU<Y)OF?W{t&0tgg_yHg!c4|W|&&fE%AmvGEojVR9#eV
z8zi`IyyerI;sX}uZ66L?FUBXju}VKHu+YMzvUJ@B#^9vCS371P#`3DX@193yUwHjk
zoby-8RL0{2_cMq(|3gm#N}c!WBYl<U19_}Kt+XWP^NH7Ib@y%YX{~dV0PvRv1q<aA
z+kWoCOrcJQJoisjBH|o=sg+4NU=d(^r4dlQA*%h_+r{Z(agwtCMr+!Sso=NnwGD=w
zpa=e%#AXuRrB7!DIUegReBTdhxsrgtKN>vlE86t<$t#6(BIGAeTpT--Ha>{0k;xtt
z{wyZcsYAN(SiwCbI4)kKpDPL2d~vN>s{YeK@{(i1jh)D@C*#s%Xa28ZbY(<}8(Ctj
zOonjTv4W59>>og?ow5AJ;&Qibnmvl{e2zu|y13$4!8V`#j|b2H8&@7H$eUOR$cwMi
zN2PIP^YXY_)2QV2x<;driKl2(l2EwzSi#q;ah~V|r&f-MQ-+os8s3_hYOXqK?jJ&$
zGEcQDm<D9J-|GZadb`cnFjyN9{lJiKD}2oY51Q;{QoTIe3)M|+rZ>7*az5upy{Wc}
zDQ-E=>A3lrurYWM)QUn=->W&$FqwmKs~SM#IjPGctx%<JVsZlab7*gZvuhsdR<}Kt
zxJS76ax9h_f`Qqilc<vGk2&NH%1iX-c|cy`|LK$UQZ)Enq^WtMSGp(dV4Ktl-jSFd
z^~5SI4q1?**?UmOlhxuTTholb)l*HDydaV`1JW*@zC)iGa5%~LHEN5npPRnc9jzIL
znOx@yjZJiSXC9RG-p6kI79j}%XM?~YA5m=d{&bIKjKwa);24m@(Q`}|;_XDy?(ZEh
zCX|pFy#@_G$PC!sr~axC+?&;3W)|R-4;qwBv-_|=7>*%NzIpa|jJOPm1QON07*!{?
zqr4qIR-WhPiI_C(5WWb}u*8XNwX0C#6HaO3=~PTg6Mc{r9YXbPf!L<RpRv;;-_AR+
z%fD|!e{+v#(vdcrgl1P!f9mV+({C)>NQ&udEPi_0?)Ew<#}gOy0o5%M!3PD#><c4A
zE_q0^bl8y-@vy*0N}PPnBgf1iC31YRs6Nu3hv<6e_l$U3C>+71o;T*??Ed4mb&+>2
zd-{a`vISYTG?06I@&TDQFUK2<r@ZLvIC5zSxmR$*Ar8gUu1`d9#J;I5BHOe_9T-gF
zf4^IEBGEXJa9Y<HBBEq>N`!P!3)P6(dY;m0ZR!vw?T~)<KujKXucH&0%<ZUqh0`ls
zz)WpXl6->T&DXPJF*xQ;06`HkAdb7;3X&5Q{#B=D!QaP8KQmyEohb0q)$mlUEo>ub
z^t3;FaUkV9pD{o^v(G|ty`e9R(bH?}O%GF)mRMB2rFm3H)c3ZFcrQz!!?WH4_)AW8
zH`wW2*Qjp8vyEMPuXf2Z%}r)Se_tB_5W{iKmG5*=p$}@6r%wN`*Z&21<cI3NEXXTF
ze|RZAGT^el|5%+>mV+cdapN?ohuWTwPt|+IeB`kw-hz&42c{=vp!Pi~xAqHtR0_NL
zrTT=w#4)tCZ^iQRYtMG5bBYero#30=G7jfE^sT>Im@7`9o?L(0#-X(PUdy<W@by2q
z;xcTg;9))bWU|9QY4ou?eetpoFKB~s{iu2_c)1`to(3-eg)fWQ94Dg%u45Oetb(uF
zcwFocETaD$m%J>h9@aD2jItVW>C4uHrmhDU*vs^tZZReF>E#+5Q(Y9xslu3JV0cke
zc|@>UkI_=sUKdvyn68v{!?(VCrjQSV9G|OIFJldFk855oG(nN@!nwN7_T<Ym<xE9&
zVI#52oBMJhGrFsrW1jk3mxty$57m*GZSe(FkMRC^{ch0L@{E2)axjC~maO_h_f4kh
ziMUZ65}KZ@(=Qj2q6GU-^p%vs&2CBVhvv9NV3JE>94|T6;#NU+{rUy5=;brX*QaC4
z-vruP?A2|AWFM>c#S*Uv_a>I=cV!TMquz5l`vfQUNHloV?lkIe)Ku_r$k<3tFH2^D
z|HvR*d0zijuF#L!@MJW5Ey*B)*Dsg4#eYC6Dl6Dl50=JRyzp&q-iv`e#<;GX#{sXL
z@mS;9nDu0kp0x(8K0WT%xhdxW>Sw)c!bC%$#s(LnPymXJDlD1Xpc%Se7Kv5D+9b&m
zJzkFXPJ;Al+0xIb+}J+%oKa99eKvblNt=ZI49Pa#6s=nfX-jZRiV;_v?8l`MfJO~E
z&XHp=t~Mz>h)K2C+%pi{JpGI*!vlIGraUxp%d?Li>Vu06PPtp{n$QRAbhUQm1RRWA
z{v45k!aarrYYie4z@{1!MmZAqsh6MK^_#fAws-hZ4OTgRyY+{X2M9tBbgRm<dx+-l
zPS6{GZT-Pc;O*j-C&~*;{fJ>X``?q$jz?WHWe@R~k38>oY6y^5_r&9FeP@XP@Psz1
z3+;b@`g0A$9-qLZ1i><BvEnH~RarI`d;RmSjq$Wc8cn_ZGqr`4Xn8vJzg$S2HT7}8
zMWAVHDZx<(wOn+bW{J}=?FyT*aZC&!htxAK7;o=k&o8eYQmdmr3{7-2BVbm+-5=A>
zyY9L^j(h7@wsyn`^@LWRcnT*y^G^6e96xJcAb_DbEtqiTO-^+(&UkXX3*b$Ekb+}8
z-t_vs^QS^9r)$bvF30`?lUuE5Q32;0GZtrkJmf#z9q(#sI#$MeZrG7pr1w3u448P%
z1<9Ys2QQ0OC9F0^Y>03Q9TXp`{>t`__V~z)1b;}blK5Dwv4Dr`lNz`w$0Oh=>XlCg
zk4_yg{AV7x9_i!f?q~~aDc9%=6}49fd*jR6L3^CL^HO|e5MjcJTkHEilsqZs=-cBP
zEw4fo#ka1Eb;59b=Z~{bB4+>I+fX7D#}6KZR5xkP^rH(~I#G$oPyT8poh^Q#$HjJX
zm9}q*3yv3**}o;V9v2=jD6@ZTIgxO;$3@451DmjO%Dui=bt_3i$0f&wHPc;&vuCV%
zyr7E0w>lDOTy|XA&jupd9@pyzrtK}<ddzX*U4MuxG<c;8ryo}4%Hst^R{n6;g1Ab*
z>snU8vKsuBLiS#x5{@3+syj&WB+yn=gmrmm%++uB!Ty-%4#Xl{=dUfe|B{K?UGyMX
z2OHm@>5#@n<Xd9C1;;HdFR^4<h=pF2`eBaTjq0wOuz|!~yXknj%fJ79-0VtqDLTF7
zxNsHARBX}Txi@ZguXRZUBrUthe>nXoj`h&ML@ag>$iZPz4JYQGNv$sRUX*u1rp70t
zUb}Z*>fPW`xAsAJ$`7<rt3Uhi6Vap<YUWnNDouju-09HP?6FEF3DBl0Rxt93Rs-1#
z;|=-R8Eq<ryukL>D@(`_4Aa-Rmt1w$X{}EJ_R2WbCAMDY3L~8*5Bhqaz2eZq>$}0w
z7t^*>tAjfIC#EB9^H_AblQv=vC+>77YY3B6CF@hK*Kli^?Dy%TgS|xtEMVbi6P7`P
zJ_=g{>U1y3azZx6h-clNM(=qfMzzf&T`}&W44;*5j4|mypz&rL=CREa=h_tI3T~qY
zYF%E=9Ud7dfd6iNO_68?&T*&92t7jJm|bcHwZp@IkBeFm%XY@SUIp_j_j!PJgf?-%
zYu*MDoQmBlnthh!7V&_aahp>G{b22L8+1W69@1Gk{iptTSbe0t3q1XZ_MEtxgYl?m
zS19`K6hu6B+zf5<Z6K{;Fqke=kE=iOr^FNL0k(8g?C}eN*;z7MDO`!CR27%X*sGtv
z`RyO#X^(|*?+_X2c)p%dKX@LD$A0~;tldJoYHtE6FaiW&fpySRYlYL{22Xo~YiR=x
zx&CCeufO=)9oDEI`Z(%-dys$6=r{g3NPpdB^Du5j=u%w`JBT>pCC4R_%{ghXQQZz_
z$fm(KRXblC{5*Z9HQMqv99i7WDpuwVcQaWMHoU1y)h!e8miPCr>wozl;%$F`vW~}D
z1Jxlx*1I04nc&9*oZI(Q7pGW^Xx{g2N(XF(Ns4n`2TVg#eAW{mXwY@rH^qm7tVjVq
z@{}O5w!-rBcz>*spj70G^obg7MGE_=r!mq$$#Q=7|9!cn5cu5VqV*$AAl-f8d32Jx
z6Hf6<jZMtE@%YN!WrUF}@pWyf;mEecH=18FGIio9s8MoO@tvw|ts)Y|ELXryaK(P`
zB8;YUVZm$hqr1%e+ZjK(!X{jg3AdATjaT%VXErEa?|dOHc%|Sjgc9J?SA;L@$`4k3
zMHm+>*Ao}NA~ogUKwP4qmDbzh(pL(KuEW|3qWLlx{O<a={1p+0jWI`mY{u;J3XN77
zn*F!9Qoo1EBpz42BJ_>k6dDy{Tn)_m!D;jweW2z*a(k`Dd*^{9@1vrx@!>5P*2mme
z#74uKrOVGee<q&;k$IiR9%R<@-6b+Gl2Ci_Ud~Rs4=nKaF!b^!Cce-O6Pw+r0W<I)
zB{yki#7T*p-6O(A1iS75I&CZsLT**zO3ZbB9gAKOWs&SP*NR)T*xLf-2EAO^$0cf!
zlJE9dDgZdM15xi)%aE}RwRS1c<q|>0HwwNIS@(ZNledGg->Y5`fr(h^L>WU4jLolz
zVP1KAh9D;RD2v~sKE?)SU%`a0)w5sLyfWH6$#)*(G40R>FIydJ^sA#G*1mG4sKzs+
zmf}_<_FLzTz+fWzw)N_fI5EAk!7H`3+Eru~8#Q)iY-=t|m&cxh`|R=983-_XRlt?*
zi9R*qFsD><v&UMQ0unT&{%qe`;u#Osh65Aj03+`7+`s=n_r$0_s?~|DG45SDA6_IT
z-4#!7+5~O$+M>0eGMFP6^zG1ZA{XQ0sVWe8u~UCemOplRoZAXIbg!AIkL!MKZI)xm
zdhau|@N5Ji!%fAji<BLYfm&RU-P%G9K+4KyQkfqQs3o$ccF0{~_-T0T9v+B?1l3)C
z*bxtV4h<L>`R=bHHylTpNboAFLqz_VfLhjE*S0(!*9Q~=ikK%1ZgVm>d;F1M1LG;r
zt06<jBddKyBsd3rx`aL2g+AnIL6sD2JmZZ|lUeejxHSpU(}(DQDiSx4m<swmS|wrY
zwEvEbDGup3=9%MhSg<4AilYW<B)5CU-L`k)v(+ER)P%T!L{8B7CC?L%oI25e_`5ji
zCIFtn-A<{yPQptdh;`qjOG6B+F(&}`4gEdc6uRS0-!P?tl>G25FE%zBY5GDaA<rK7
zIJ!EkFJ_fC8}E9Qt_j2r@2RMNYKkxm@xF^PXHJ|G0L@&u5A=5=;|R+>bl=G3V9Y+!
z%zk%6e5}70i!0nW1=0%a@S#w7+dfsrCBKQ!+(B{{1bo+`rEyn$;qiHyHpQ2!mQNUt
za$l($NiZhjYn?_yPh}<u)Sk7iA~VCc>LT6EaoMarXvjI}Cuy(meFqYyWk~5CymYYP
z$nAN`e?0%3raFg>iJ!C(X`B@&%s>)ypR^|@3I@YyJ1$f~`i3&si%tkHf=JQK7X{YE
zC(fkUa}+^$TymmdClr4fy6{L`>Ry{khfb(>H!kv~d{uYG<r-)NI)2Ux;iQqw+A$ti
zoG2JIbAw<Zq%jEpGZ0tRARXQcCcfIOQV+_4U8DZy-Sx}3Rvi?{VN~4PvK0EvRd0Bw
zb;dmXh7BOhfhw)Q5(VR0O$8SkNE#a5aH1fiVva5i#sUG7$pOHP3)L7TM&lbj9NUev
z<_QrCTV|ge?N00boBanY(Rzy;+b!z5QgR88q?hQeu9B#V<(i2_o+p^MA3V9(Em8-O
z=7NM0atmyOr5IXj7<fDTqTY+PH7TBGa1l0vgtB$dkg0SYP2SE7^1RAxK!|}fdx>Um
z1BSuTq8^c}=ImSDNg^&budTMz8TYn?u8I!TNJTUlCf2C=B<vRMlxtPE91fPKF|qDM
zp@`NYW;NVIpJk343^&AiPi#CPgbb2NgP9^L>r%sC+#NkS2n&gp*z2{z!4FHr%qQ6A
zwJPedoY-b{jQ*xN7=xY*>yEcU40#bSC1*Fu5@uNOuk6QSc=W`Xg2+37VvIY%TQq>Z
zkQ<S2p?FqJ`Ww!p$Q`$N42TRwuQBC`5#xR=b_n#-|E)jnHlXkP351oZ=$*B5k@C$n
z%-p4xa6Eu^kGlcKAa_+aE0(v!ecsdAz34by2iGHVPkTe_z|9Zn_hy3T#e*IOTgWuY
zgHbfR<QvIr_OQzlDV6ji@rXAg<xiN5Ba<16M_mbTJcxA?|2=j>loevIf67zA(I%@+
zJg$L3G(;ZKeiu8emtv2$b6ambr8<Z%red$pL>Doc&}SO5jJCk3c*ccrt`If*)huIc
z!~u0S*%b%%SE!(YnATtKY=}epE6O0mtbg6}3~?AoH9fSle~oxJQ`@ujRykH%`N2)Y
zaa=uB+R1`&!iSH1vEw*QaQRO9j4p~(ULX?Ni!C|rKZv}=Z1jv~vqrb%4fQ)VaQct&
zrmFHBjZ54AmO(~CghdHjOKpEAfp*UtI0W>Q@vg_<#hZN=mLlG(^~ziDzB@r$0GqZw
z&Z(0rF6;*?%5%umJ`*2$l7nR1`bc1Q4Hr)CV_*Jq@zc8Jzs4u3T%S+dr`|A*N{U85
z(-2DuZ}oZYQV_BL7o9KEh`@4$f%j<ZAV9wI7Gp*a&-1k=N9cUsop00wBws;I->R$!
z$8q>W_uf8^UX$QJ+J>gi_(A>EE$diGDx4p^Sl9y~Wc5&_7l3ezSB3nT*dq{+nd$9<
zR|}Ss`ryJ>g~5oXn1}D8SI?A?K~v!Wi(j>oVKN}a?*d8rC903AvCv|dzFLs(_SqNW
zGId+2&p6GOzbgE}TX6)s^v^NJ<A8a_an`!R<J7~tUFo9avU$L-dR5%w8o0$IJg?Re
z73AF4cnH$c5fjKWD?<FtmHy|bxGHIw=DI4%Mi8byn#K{rT<75sd&5M;e3vzJZpRx%
zA8F44r|62wO<)qU&~QO>LqdTsUlsMyi<S=wt0shs16=7YXOD7qZc(!~UZu2giCfi-
zWl7VkMH(X&w`n7^*hBJ6q|46|wOx)Fqyu$CEOo~#`gsTIy~L(WPDFznk4#h%Rz~Bi
z!Zprf&_6QLA5B`I&5$Ik41Tg=c%i5u&CFXg$n3Kj>Y!D%@@S@1Rq$}rFh57E+)>-b
zjyq$Gmo7yQv9<==W98;yP+Qxql)a@e?Rc#BWC*r985;~ccJ`@j+$eB}Z5v&+t}@v2
z3{P*5K*Gu@ou+zKa}teDpZjReOah~^xz;f4*a(FlgPw6)NhvVo?W`l65k?X*;_4tJ
zh;|;2QCG(*fCD|QsiKBM9Vy6D7zL;JYQg)r!zjL5sI9~g1Taa%h8^yr6@%vCxck*J
z1?v`>EX0X46+6{gMgu^WDxfl3Lt)nM@suX<#{#5#H3AB7;$Igj?tT7?mHpJY-}73H
zYz@&~?Dift!d*R3n-YfUgYlp`Z%<1;_Us|=$Jr+_rF>X5Q^J7<$0M&A3H~3f8<NLE
zm<EUUs4As*0PFpj2YUTqw#MTgXbL+3_((5pJ7#|*)*XsHep!Y`H+2%-Qw#Bw+K??i
zsnNZjIiZlCmE!5zXA!H!GqrU=%g^Q8uhWH}+7JN{i|yQj+JQ@PFP`%6{9_z67>T)x
z?c}s?NEKEJV}tVi9`Y(6ae~v-2a%(A^p6^7ve-f}CLP6U&*=D7whzZK&p#u89QP*3
zU<r=XLzlJ?WpUCA)XH;6Hcn56P<t6<^0Z;VOK>R8sH<fDDV&KnYUh-Y8v`kLeAA6n
zp@D1BTb{T)=gTR2Tm4Bvepc0q4uKgX{HI&D^UASR?|G66U^x3^ngYf9nq=$k?Qzba
zZA!xD1Ahz@M$}U-J`)_Lk7_flCXfHI&uKE5xJq+zKhbh^WAzlD>i5*?KO%_rEy|EL
zhYGp)x!bm7?4i-T^k4WdV2o~Z53>(n>K68HiLV4lTDO-&`nBPLhAme0-_+njBHk0<
zs&tei0Fwca>M6B#Vk-H)z<T{jT<L$P9UdrvV)*)_x4Q$Ao}b(by(wbYlftOW;k7qy
zj|==c!fK<*yYQq~#HLJDcF{?p5UwC|xcFqjrCn%1E^#9siL|S~^khwdN+=6(%4H|@
zv=f*La&o9>$1Zmr)RDz~V$R9(E=^9v6({vBl@;L>&BT=&7g0^x6klbKTJoK{T9s=?
zzwpD?xR_ZF#<c>h6HPYu;ZZqh65$12Jd&>#`eLpdWUir@=P8OIz3${0@!dEJ&-WbD
z>k6jehLe6{B<@U=OD6kRpjteK#o=h7`%d`c#*-p4Afqv~f#SbOr4tYm<gV6O|8eg1
zzl>Y7I1Sa_PQ(a^Y)_uIZX}8l;};5%CvDNmLN4(DlPyH=(qhkmoOkTs5<`s*O$Jq7
zsy#!874-(KIn}QXt_G(&&MGvj4_?ni3r&LTod=lqDwRxZVbiup^ZE76HQ@$ej1eu~
z*)}{zpbDea0Kw*vOry<Tgz?1Ys>5~L;o=c*#~OdLKIM^LD<Hw=$Pv~VNCKK8myh)V
ztCB9#8~l5e=rS8UF~%ocygFnaGV|+kA4K;{bK+jLgXts)uKTDW2-<8Yff0IsP=~A{
zC*VZ*L;ed-D|+%oM<Z%IHh>q%s9F<2j<6ar?!I`x&`(e5oOjg6Hi5>$kx#>v$Df4S
z-=SJ-ZjZY?8^Q+-#ZFb(oCYwv)NSEzUx|D4g^6vZ<GfcVEj^TF9C@GNofsX<_%pmo
z_tzF8J?gi_ZUYIm$TmFS!Y%#V^6h)jpBd*2Y1%^usW)8?9u`P_BwvX~)XCUzS&>I|
z)}}xTkNM;SEN>({_T#F#ETgW&k-C2<o#Y7iXkpG9Lq6r=c6?S6)f&`BXgYjDy=v{V
z`{Egm*f+a3_Itv^8BndT$_EVf>_IT_LEW3;0-l-%==f=MTM{4;Wypi{bY?h~!|E+>
zEqx`QeN;pJ_E&Rb#vRvZ@`qz?cSi}rf857;S(0!N(<l7BZuU4<tz`04zfDlnQxFCm
zqf_d)WBW*)_7q^^NCKSk3eNZ)bsREeTRriH2kay3Tu;2|0Xm1amB`_53AT)5j-ilG
zPbZSVv)&Pz6~NWJ>wZd}g?LY2gPnb3d5`x6R9$;hobz6s{pZ4=e&E$#1`+Y0ifYk*
zKhn^B1Q!{NkG)16NyvSoiE)<F9yy7GPX%^49GSWBGXvAb89>PER|YbYxaRO&8F%mt
zuMTE}HbeYUu#6G;G`=#JUDbh<SbVLb?hJ<YjmCni!j2C~;kN<}#~#uSe&=?mmop#8
z_ugf4RsG<ZwslU$j|LAbB@V|=UdqA`!W~4scBU|;U9+#m1+NKzLKhN2;WgpKI>H)e
zq^*lyvp_iglj353;hKg1{}Of5&%+*<s$Is3<0S;2zGmF>a2lzw5$RAj9hbjW5ORfW
zn&TRU`hBtdKiBmNb%YZ=78%fpSH4ydGuN_aWw0}@QZMWA#=82o^UOGSpm>>G^O~sz
zQZprf_4M$K?2L*(GMRPE^=xt>(SGVH4J$w**J*CW$aDT{RzS|~OC2z?y+J^tGfQL4
z@mSy?Q|gBe!a|RNw1CtrUg}M7L^r)=<o*zYq<2X&tT%fZQBlM#uDp!sFt-}K)dW)`
zdzt%0aySr+{n3ojH<qYgJ0YyOuS-?bP#wgkxLzM1OjwSF*M!?s;e|NnpQ5n_O;QFK
zRyqwbnmm=k>ZVRCxmS7as3DNB@<v3nHl#v43|NIt#mu3_r-o#rc(;%#+jR&yo+;lk
z0k1}zcdB!I3%NRAC#%sR@CdES1H8s_DsF#kYfyA<-OeFjr~buOfz;O<D#ZQZ#Z=gB
zFa#M`D@8yXUprGcLufn505}T?hP|X)hclE^&ufLAYi)>L4_6S~O!OHPxKU!^daSJ&
zWTBj9tU-;1{D-Fjf@(<pEc{iB7>ZtujZ%9ss)O2o$LbjO&Nf0)O!{~8@`l)^vC{Em
zIHojK5t$`xek68y(;5(<AOS;1sxHS7q3u*Rgv=X?UHXkWt~c)a|MfBx_o@_&12UV{
zjzLju6N}@1Zwy#Z`BhHDZr4dUJ_dVWTXpcs&d0Bh2fbK$KbHI}o-+ph#V_JvRppwH
zDO)RZXHz`tTGT1;!(+Z}$^7Cb>Mp^SaE<Xq?O-O|k#Z=+<FQ8*NBI~}c^hvBb0P@V
zfVf|SYSj@(0wL94Jfp=O%PcanUmrzw9S5{V8I`3eUDk<HBv^a5#<ZcynRBBKc`;zM
ziDSiyIIICnj;o{kU0DHBH{)Yd&tAnbx0)ErTp^s<<AT4fZgrecqvEJ}QjH*Lcx_L;
z_BR$)sWhqSr%$WJ^*{gl{5a!bApGziRl$E;xNu>->65w|rX5Q_Fy>o-a}x5BPsiI{
zx;C6FhU2Wi>LB}wFt5Jm-F4UB$9pcpcox#@eXR%_F7fHBaZZ~_Xn)c3AE@E$yI{jV
zbVGC)P%Iz4R?A@z0!Wun!vEy^nsj)4tY*KOyZArHC!W-nVf5#pdIQEXOh3*ms#jTt
z!^r1e)&|TRCua9yjqAcew3lCcY@FXfdSAI04srVBfBiRWA=&jeuD%kK4U+X&2ll1|
z@tvl~9fS{K-M)7zMrNqnf6!NPog!k9VTU2;i=TXUQ^S^SK^TaPMVu14ZAQ6!!Ks2x
zu&_%X(3Bouc*+9f;i)9kFFGZB%-ixvT&$M7M4fTTDd9TwxcL7QSDsATDy}2cPh57Y
zplTFpj2oA`Q*3S0`gx8jlE`%|u299qbUgT1o+__kOLJVMzheT1HX*K72P1vhtg?L9
zc<}T^PJtvR`r1<kHA50j7mV^>QeLb+nsCMD{hNDADCg|MY<lu(^9<DlsoQ!nG4i^!
z4R^<U^;RK-QTn6x=XJh8<r6(a{jtEy(l*SoUs&r{k<AXrjV|5NNs!;0GzwPh<<-C0
zy<}vRTTT`FrgfCW*Xgr$>#0JJpzMuBhK>2W$V?V%CY5@^K=s5D^$(%WwOZ=YTO2_(
z>NP^cFe}~QZV2r%u@yd5MQP?4P5KR1A^w7XW2ZI{&2G}kyg2|ZwFB@!h|kw5fGZ6g
zs9e%E7dEF7sY8V+;wsC##+7jzNu&2ztFwmox|^^~`F5@IUs_Y`QgHWrbqI5gQf-5l
zV_<kBHfsJw*+mE=4@>kJg5sVM8B_BGOwo1pYC@ZcTM&JML*b>^_RU_)0d)1)7<t6w
z8TW2T%_{9sJR@F*5}uB*xNZeO&cw%6$bBj->!dp24dtqD6WDhWm2t|~tI&%nG>#n_
z0oCwm+^yf}CW-l18;$5sP`<mob{M2>>F>er)}xPYA%kAr>#q(zz@h${BoQ`dMqX2>
zYeL|sMc8f)v%M4L$OFD5>F=Eprjz7*P{3o&#&$m>IA-^a45v-@!<tK_Wm!DpwP|T=
zjz{$yj6~c&Z$@g-<8dEGnC5-mNLk_u^_Noj#DRNUnVUfXBLj|yFSlr~`(KA-?rBv^
zvU@n5Id!H`Px*!YeVT$X>-~NL3VBi<@i^c$1aLMf4tgv=8Hs64r0w=Q<B*5#@5_D;
z8)Agg7`@T#sGgu5^J2ywFzVJg=9Q>#i{sv$+l$2=(qGr(861m~?hob*u`}1=l!q&h
zs&QJg%)N;-o<gehHH~<~ivYnKZ>k~Q&*^UdmP-b)6?oeX|6u~Uu6Q-ZSyiql2pP=Y
zyIzU5&U|#@Jp+gVE+PJ0cKS~k9*|_)=LIE{1D$`Mky29=ANn1XYY+&QTl<kaxE-r?
zj@QSv`60d?k59ZdExdO<AnQ*xG};fUex{bKV1ay4Ki6mK0n^~(3(snp_(6C_=3acM
z*033@>{njdQ4}*6c)+E4TG%3@X4Sx8F|ymiPkd{zc)yWAf9KV!>zR!2)vOp$#}BI3
zzoiJRf7Dk-&xED?$rH`z3GO3ig$gG}5jb6-k6unSL5pU^rxxJ}ukJ;&dXgaYc-HXa
zOk<skXT`h{I+R_wWL8Lc?&4p>rL#h%JOsFHmQcT!xiW+cc;a$@b@t@3nKNs7Z@#kp
zHxpO5vtRvcNnEMUNE6DH6Ge2@Y(XpC^mMgqE=yhTHJ;Cw2~3!;tqp;eVJa$~5ENz!
zD>7H#CGlAvfq5=ZwN$2ky3Vb~u-+Mn`Ll(DNHSp)6Td;@B<Ix`3tY3gnm))2H5L4Q
z;bx02<3^Pu7)`O)O=^mK4Cec0&vGlve~Y`}34i@x;#O7Lmhn<!k*Bp9;a4nHJ?OD?
zb67HK_MKHb1of9eMV5L9v<I-LnpawflsrETD%{eU5ic5N3;mE%Bx&!Oyb5E#N4L(l
zuTq6d+Ex6&WW5cHmQ|MTyJEavL%7Bm!!?GO_iz{*7@A`UA%>wj)CWa@RYgrzfee?n
zd5ik~c&oa;ph_ehnr1i*<8T-*Lo<wnT!x{&wxJPenx<((6(u5KfryA`2r<MM<0Zrp
zV~n}Kwb%cl=iE!|^!wDa_u6ZJJbQhwmG_&uA}}Vdrq8N`eGNNUt!Z=TaA&xuR871^
z4$WJgT6Qwe<*{|k(8;XOl+!7(mBWcq8Y;3?F{#{QzE{sN=n9YLg;`SCG_JMP0g_rf
zgsuoqxJx(0Ug2PpsC`pRxDL9BOOjerh?^7JcK?Zs*gEutj-|KT3VOmG#)fz6JiG13
z*AWX?kLL7TW2^PWkU6v!PU;4PYZ2E4H<GQ>DpsRv^D;>`q`uYepc}uJEv_iF4H_ot
z3jUOs*NyR-?DX}vX*iW=K^FO+{@G@oj~ZM}Y>N;_K*l3&Yb<I#sdHo9a}<K#9!)4k
zj)C7^67s}_4}ts+0mFf#$#zDk;ge)663zpNguCN6D?v7SwQuw6P<sWd9VV8L!hJdw
zYkP(`)Siv$<-lj&eA};nNIcDUAVZpJv}BUkQ04F}hjcZ#cIqhl!ee$=9oKfXbl9tU
zNSD@Yf>MGILA8|(CTv6Ajs%=UlpNA8?5J=>M)eQ=STx7O!o7`;X9wSH6n<7tXeuZG
zT#w5_os3;}ro*o-JCz**baee|?6feWWHH`I0w;Q3uru)%MG41YBvyIqv`ID^e}_$l
zkvpduc)3e*7Y&a((*XPYrW;ujS@;LL5Tg#CYDsp+iyAdmEgw^VBJH6(mvt)r*76(g
zir!+RY^3svIxa?2V%OpoML@A7*mX_a>v6-0vOzZjVYfJbCa6swmdsalGrD@TG1tPa
zc<Xu&m_QQBYiFGSzO4}<eQWHC;m~KM)%<fp#4lr@Y13BQSL%w7y?q_OakePf9U%ax
z?={fvuD~W`%r|;Ssp9XcWl5gm6>x6f2Jn%;rTU8Y4#xo2%D&fQatJ(PKg45p_1MU0
zQtp;k71;y&O$H@<Fn)KDuEieG-)p#@9v%rQi|VZ2#>I<bH;RrQxvyOH{APZL9vKN)
zwIVW}JQ`i|h2ejv&dNxT+ot|P`x<*(E!_!=13O_P-JX9n#h!>}UQ3$&c~XchBy0Ob
zb>U{mh;L7gBqCacS9(sZye7${fK>UMO^&}(`EI*S38*MUJ+5YIhPi7RaErz!J3WBI
z*rd}nLt{ZQ-SuduaIbCc;A4C`05^E@&RGJ51~e%=__Ah?1oPqgdFYPK2}GB=FH(TT
zpy2$s&DAKb_duGTH<I{>6dwzZ$$^_6BS9s4Z9z<I!2&A|Z2CKrG*KG8$d;KEwk?al
zc)cP|rwB^r*+6c-I$=u!LsTvvGWvU~(DW9PUZ&ovwBc*;rnl<*lksda(8YM^F+t5%
zHxeSg2io9a#7(l%maBt1srYxgBBqe<#SF7D#-YH3`%$clbtywN^L+dqgy~!z18@U_
zcfDqcvLD`|AzrmMyCX=`NGK1`{<)p~n6<<lp@m+d-P&T@T*cBw))Bwa{q|ZReh>ee
zkFPrbLZk6`=@H_^jO-F?p9YY{ro#rbh}=Lf*R9h4i&5tdsjbISdkzp=uPs3RpB&B5
z%*IqIBMaTek<pa%3+&zN!Zq%mNXk|0n_^LO_;CB#bH{Dl9G&y7U&Covhzy`MfRt|y
zM2fv?+BS`QKYf91k7)&X6c;ZERro}>I9xk-2-?`cz`Xv#cIvUh3v*DoOpoQnlaujx
zw|Y~%f-075*Pd9yTpb2M+Z$-+hVpAsitG!#T)$>73$GdoQY)^N*hpiVDLa&Vr@H3w
zK=c^-dkk3zW8l{0?9^+A;yc69-sAa`;^uI)fCLrXN9ENxLyIYi^jaX*k&52EE~IK<
zMu~BWr}VLQ*ir2{Oj9QpxgAqmRD{U6$2Iro+~2&%PH3vljd<dn)VrRpL5Q853e1U8
zEZFG)((ic!aXeB(S6m^Eoe_|+!+eJ$@kzI|*k}f>v59>Dc23LOgz?rhB-{Du#+|^1
zUl7)8A$L%hT@=hZ_re<+2_>=HariRQ>2eK_T~@0yS6E%qM4fBuva7np7XR5ieC_^9
z6bj_d)rfw!32(L=g0;|7V}B;xD`EDBe`7aCLe}DTaZbhk7XHC*#rn7yX`c%;p1`Qs
zZb$F@@V^hg$=Uuwz#ztWE!&rRgf}KPoqhG^nWvJ=U&jKrwc8zGgdK|HLvh;D#lWq^
z_5Mxvx}PiHUSLR(IEX<jp+9hf_HAI8dzS0yyEsZeojk+7j}bj|Aj#i9L^FJ%aXdPk
zC}V?r!`K67<31uv31>F|KBzy8?R6FLrT=CRoeknhgsF=~59>V%7Vs410FTQ$kmMoV
zXyes^eQv|0`H{0hS2#&rfh5sA8-~YJPn?W0efC%&l&h{>TxyS>4f2eP>WV}6_MdQe
z)JZ=Sml#p?=0b6T!zJ@%yzOX%;7vT6h&5Ajcf_XzigGqs%kb3evYc>R)Y?~0rOQd#
z$oPxdWKD#E$N6<Q{o0hXiB9Bjo$?aNtWK{@J)6jPGaSZrP0aN|?h!EKY+?qM^xDj`
ziD^XZdta17v;1_tud23L8nzYHj?ESrkM#DmQgdR4&b@hoAXn$ctiZ#G`doE(jjqj$
z$>%(U$9%oOXAu`*O|+#oF$Tq2inVBn*ABnYKoPLg7`GB<z_iL`XY~O%)6bXZ{=!xs
z)BVX$equ|sJj7GL_rn!b2(1o6FG#PeX_P@gk#Vd#Fl)#o6x8Tla2w<%3aE=an{Bxe
za|g$M=o2>86+%5liVfPz=p0qL;DS~O56+MK?-DtqBf)t?ujCBMYW)RH#>H6HzCN%4
zZ|HGsE7~1*g77kTr>#kAhav)bRVxAUaU*|-gl%epQhm@mgks9IgSuCbVHUIQ?A&>I
zGu%JnZF@5Kcy?M}AVWY6Zb@32dgAi@Yg!i#n_!)&9vV|k#sh?j23wyY;e5pgEd*+y
z(7A5ZVYs#FHQS`$HN(ekv(WPlh(O&EuRQ;)ZH+y2E*MwqZQ4=j2;m*u@586mn}=Y(
z6dea)a%@LzAyoowr@(t`8Oys%ICio<wp$RFbh!q7k0$HczMWs+s}V|(98;-_eHj8h
z`KCU5ImYNed)oF3XhlD2>jN6Fw;xmKIy)HO)vEtuhXTQJ&$e|~u>B~bG2d>l>T6!p
z)d_J9=f~FswvVdI_PVxh>|b4tic0$*>`3+}PZfUp(jLBuqi5v<;=awBllz@VC>{$W
zD*xJ!$0U_q{Ev1bW|8Y#XD2hnJHU|ie?rj{cCB>@=(OJR4Y@Mst(XosPm`U|(xIwN
zhJz6u@xGSfUk@Mfkj&BeOY-f(=kC%k?cCY0Q^rM}Tuq#^rmW(2UI0sR#BlG13xTG`
z-&_=4o`?xg5GsiVB8#hP>+G_&h_ZQzEms2LGOTBw7VT=Rr@Mo9*tO`j$o*%pYi6W2
zcK3rDTK9Z&GX=Q=@vzW2Ew`KT`s}9RH`dy%0F$eQmV6#a;$msHg`}rn!rKocYP7rb
z-w{c^jCM<FO6@D5<dNX^wa_440yl%*(O$aziyZK$^tm6gyU~UF2Pdd+Vq<Gt?OqJP
zym9%BCd35sReq->DN3o=zmJ7oSZIjK&_6^|BoFK_HX3$Pj}UVQc^s8SgS;Y|Ql!Kl
z98I(?SB|^WLwe<<AYi5*9!>P6AGhDpM1)#fDey8HV(!^r!NWWf4|!aK3ifCW72BOy
z^v9wbVGQ>8Xd;wFHC2nN7TJV&B5jZ-o_j*Wy3Uw*u_ptL2NUqThWSoZgBH9|A&F0^
z0kjSV!5QM*XpsC$1UbJfCI#j{C&ET;JvD%zk_MYH8f3@KYAnggB<W2JOq+?;m_B-6
zU~)I_f3g`GeJR-;Se2QvU}OYrv8UsAH}(>n75%^%J--u$*#cDV<lKdwIa=!eHAD7H
z{Kg>#lc@&CcaS5LaMZx1Z1D=4pMl`SC4bEYqlsBz1R$oYUsfDz$<^RuWvRx<*RQd%
zScTFaE01nDConF5B0epNmoLn<3T;Lz{ccs!bjW2pmlxRB>VQe2e@eXw*XleGZxGga
zOtBG&J<GM)P&H+BSP>_LY<)P$t_(mBO`MNLgV*t<d@5<ho)_{}@KURF&Tt#;v3jA{
zjHsNqhIoa}fS8viVem?Uj2_ey4Y0{O%O%AlT&it49VARfyNwE1Lz5OEF+6TQEND1_
zullW9%br39Pr&u`Xx24E@N?1h>9M()h)u$7)VvO-7%x6x>tY_HJ0W?IP)Gb_==FNz
zfw>!ENQ#@#Y@^06@_dM!v<@YF9Ge9^-`|9F2d;06TH=`bceYh;sd2)QU|F|mxZ1=!
zZx2LMIz&j+OR?vJt+qqWaC99aX}z969u9q%K9DL;SqjI4TVJp3&eq2d?4D@vR|?C!
zSJ)Pzd)gOodLtU<<?K`QoxpK1InM9bZgh7wtnmyR2Xtn<m=lDD0J4Mn%X9dNwIc>M
z+o5dEzk_u;tU3EP1#P7;jL@(4fGK+|8`Q76-$Q=>h_wE?{^&Mwa^XhCks|>WcXYni
zqk@{Kmdk<1;;nBe)wwkUA;^68C$i7U^<sX;erYFlfq?8OwZ)KK4~Y%9S@~~>>}c@n
z5n~im6+TN^L^8kHh>n-1l!Gdd>aiTvJ4fi;{cnuaTg3AZ+WA;fd|`sOyr8L|!d{mw
z!xwexy5a4|;`i_?XMbgv<2NpkLv|%5ko(azyBdG4E~+K2Mk`Mxk2brmhbBHFyAcm9
z$-(tAjetv{-3)X%fW3|QUE8f_SXNX~yU;$*cE6{q$<Nj8Sgh{GE-$bAg{F!|4(p?R
z8IJ@g7H~(jud=t~7eH_RwLmTFCg_uk{Z5S2lON>aZq^pF00Qwh*`Ww3cr*#y_q38;
z9k&??x6!^;uSB0u+joH+x2B(=?_-6?s%Jk4+}KXPbT9pqg+FTcb`T-XB=uX#arE}6
zOOYk+V)EeIq10Uq7tmu5y&V$i()#3?%oaWzf8p>YH$BMoxVI%8xu;i(`sY&{uOWIH
z8<TnSkr-m?v}rT!QN0`fHRgMaz>obI9kM&&uOt+Rz<|h1h_?iGIj8xiJ@IxT>#PKt
znJ3k>C6!E_C~OHhh5*9J*i+GX_}4g7TJG(*$eEa1$CIDbB#jNP*W)3vliv;*JNb5a
z#3s?MF^1PfW8tU1omh>9#Wp=)=hpPIP&2Yc@IV-Ju<Qis@81pvhCetVdpg!&+L=Gh
zv{`yp(%O&4$7a8s*p|gbWZ;cfL!P_znfUXxNz-g@{5e<DvBu`b$V5cqSTtWtz$vDH
z<^sVj>Zo_cNU>fPL)8YXR8MQD<=ty#F?C#{Fb}cp<pGu1pn+N<IJoRMS6U&Yyb5Ji
zv}-{$hT^`Rco0v3HrVOD>ovl&cK~g&7PkcTfM>y9p3Q;WGArUY5?ZUR)Zf<-wZ1B*
zoGV#w&uf8^Nyz(<O{|~zN9b(UY_T<HR<72C%4#E){>Q+`mKcDm<P~cZAkUM;oz)Q&
z$e~Rkl@3anf=&R^!b{1n^?U19_dj}tQq8Hhn_w;w=?g?iLtq1fjM~ftf7j_zUV}p#
zXzZ25&Anmkg}3|A6WXfR2!kZ!Lj2vR!_37*#T}Q<H^opzl?!dNUc$M1!~oyr7U6|^
zN>XcV)#BCo_HEMxH6F=tA7F_4vzOG#_0phWI|8LH(&gZdu`^o<{OnN(xVQeUY+Qsy
zoMUkr*d5T8@ulw3gs7~*!t9Odr&`xP{!81Z)d1r>-`UF=nQsaUE6QVioBgoSNT47z
zXa@pom*5Gp5pc9yD60--sJX=E)`H*J;p|TlAMI9<S9KPO8wZAu6Ks|&lCKHSvq_MC
zJ)ZpDf+sza?O+!kBx~(x>^oloxgp|rvY8zh2(m0g3QlMgK6#HMdNXNFPR2FyU+3DX
zm_=<}Q4P6xGbCL^c0=t(dn*Gr=*Rbr)}_KTub{{s(atr(8QG{_ayh{vLSSR<ODPC-
zKGu)5aI+1UGt<A2ExDJGAcs;sBAfO%E^V?gZ6)dy4~~MFyPVA)*@RE~N<d#&wcM`i
z4a?+L@U`r+JI3R2m>29Fc3lS;KFlc;<x8V~3kImqG{@)igDmCEw}Y>!#d1(+F7Bsb
zBa^umGi`OnkbSOcSCV<tZpYGB<C$t-1khD4uCgx$v>MS0k}DL;DFlM9_Um}bdU$Qr
zJJG9xPvNd0Lk#g?iUmzn%Wvp=(GF)R`!?FSv4(=B8ma`vE*W{VO?3M|^+M2~{17PS
zsLEmRJr6@obJfH@JB@lk&v4K6+?n>^xuB_wx(dn$`p~%$<Q9@kAD`=Pd-z<UanwFw
zAI3%hg)Yr|W`ObM60N8qh2|sY5;2<pSN3T1AT0+B&tn?FKWkVU6q1jh3re)Cka*+?
zdTfZQb5BI)+}Lpx1W%qz%s_H-nHWQLG&b2&ngw>k<bs*YK(sbBaqUk!7i<Gb9=RkZ
z>*aIqO4wsl0%R_aEA`YE%i|yNHeEm?Jdg5gF<xssDePkop|fEVA)lI+3*V7cdRBDz
zP3Ge}ZL@_0;wGw@IRPy9H+g$T6NJG+h!t`a2lo=Z^YloKFu#cAM=z2F*@Eabe5{Y#
zyf|RPR=9g;X^f0Z0XxJD%VLIj|B$|3n7d4#*dbe@;i>1+Vio5S7s9n3gx4(#Ra%Ea
z<68W|h)-3Al>ECUVE^c+)23OS8p7NR<jMB9Jg_QRT2;1qk*&}Q;Vgyy6TdiH8SgzC
z2a@V4t<NHqLIkL)by5VUd^fA(9O4>`Nx<=_*V-&w?12;3pwW{hk(1aYG@tEa9b0sW
zv9|gg$R2SnKJ)*WJ_mX$9BTV}S*wDc)_^cMmIVJ<;trgI^ys;9od>)(TAzmCj`jLz
zf3yL0UsmNgT-NEyqlB>F>L~>_+j>2DLeys45SzSau-!Il6<}VXwu)_vmy4Iy+U9d&
zZJqT@4iOBxMYAS-G1b7&1;6N*AQ!raVQ-5KPW~`k_#5ZIw#V>X__VQ?Vn-qKkX)gH
zo^hu>lM2-6yW-FJPAF`Q=j;|VRu3t4!d^|PtG#FBSGG@McmTs`dpVw5B}zp7er;qK
z&k(i)T1bv7=2+Ar2V=O&aOa1FoPW57h0#FQkiD8syRo_1Uen0+7?JU8QZo{P+mZX9
zQ}RDQ8n3wG@o0)c8XDW}cwAgX%j`seMQOwVI4M9yEX*m5ml9B))=7wLW%w^?6mB?A
zk8wuxaXSb07cvh<d)Hg~KG<kBVTySs@AY%q_F5<qemabOKBmM6K772>E@<>}zrinR
zw!N+EP~4B{P+&Vtrn6-Hx*T&O%ZOdkB+AkGT-7kxHeAgF{kawm$qQlEqao3uU3Nng
zL?UW*tITIHK55?Jc5Z66_4!`Uk#{Q&Sgz6a0g0SLFx8|Jk8n7J+roG$1=qXn3t@a;
zk?2C;moYjKjs(&~bC$0c7oD#Ki>xi4N90cS)uEH+_@1!4nqY0?&+VI-&Lq+m--|yJ
z3X6EECX?5%vhTE%gv+$s_c2mK6AA&`f1~$uFt&FRg{p^rb=&^~?<69ZtH<>>-Z<fL
zljlS4B>I$+4L$r$P@PJ6VB%AwVQwu9CO7%`cY><c5ii{5ks6Og&m0ycd-R<|VO?(o
zy?iG=VlS$NsP8xO;{n;p3ZL|ZXq_WE3OlQYsc}!x9(z)q-O&hY+k!S=;yZC|xKBdL
z90n6%PpMmTp0WZcf&&>(9c`0@ST(0-ax~*x#}%*7rl=Q=o}SfWsu01y=yx_<?;H=x
z7R-qGx^>he*~rWLj-tzUB+1{)T)K$<5BBssiESW0U{*|X_ziCSE}N|tThr2y4T=?;
z^G>ibToWnW?wROVyU6Co*sekFisSR%37&=wq)2O>9N_ser^Qu^>uf>%<|;xWjpZuu
z>+7zGlp$6cugA!AWM%QY6t2R`<9DtXs<|Y7;~?5)6|o4N`N8@|tBQlNtZt!It35U|
ztBE#A4{vo^s60V}$R5k1T>~!ewnDF{2<-KVRtkjEuT?r+rBt14wdeIj5e?|p>R6bi
zWQah|OKmlwH#G7KYY>bo*z%@8!<}}?QZEpH1L0_~wrB>s3{lj9fzKg8lFfYswO&DR
zC_ZZ4T3@d@)Xw>((o4KeMST18cK8$?J_Q02Y(V?^3gRLDzMACH$b~YbT`g)N@fOVX
z`dC8l7TchfiUQggV5lhJ<ad+sFG4iQ+h##YrL$OqEjpTA^?c1+10>~jY+E3T0h=84
z@tY5f6!q^U_Q0=b2$p-<`<J#uN03|`&Ve-8&KML-J9yd^0JtS6`Vt7xF>HE!gnP-N
z8fSU;Mn^szC$szFrSm!-?DldT<Y$X*zj~3%x&^{<K;u(d-R(6Ld(j_i6H~>_a7gnc
zJ72pU)&n+b{FMUejGx+T@oQ15y&esQU-eL*BY|-zO39Rr?5Ot5)z^ddo~z-Qmbg5n
zC^#<YrJmUPgkCjMm5HPd?<6h~^`m>Z!|Q?Bsc4S8OrGb{npaa#%BTNUEF4S(GNvFA
z_~@ik4ao)OCHO{l@h)t$bHe=<_nbW+0CLOGBVBkW`F$tu*1dl{tkJEi7lkj?0#M&Z
zhiVslF9%TMOuM4t5ER^ydo`AX9CI8hbPm__+Ep=pGS}lZ!AtJ3)n_*Xb?Ba7y`O13
z3V>7ZC7QC7AspP&#?^UcXQ#ZMM_+fg!N7StP{P;PzQ|U9PaApuOF=Eo<3ILQK+R#%
zASjV64qArA41Y)4*R=*;ox}y))!TLk_(pUAUKH<EPupzw^qT9|zKuZ%_NRI;5vCf4
z;KSrP^u1t5$~$M?e+Y!z;q=g>k9P@^64Ll5c2gag;xFfe6fWr+{=cx5p4Y|AvG%X%
zp?G#-nExHMhtDVS$Ejr=<KiKQ?JsZR0}gO;TDM2yDdO(!(esIJl$Y9L8Vwga{B#~y
zD^#kG)(HWwxtZW4d*XZuo`_nmQ+V=xBI|C+o+uQfnpuNAb$-;9(CAouEEm6XOWBS|
z=Yxzp!_T+=rcKVqrcx=MNCFcnR`CSZTh21~=JfOFlEK}`$r4Og>+l=xHbY3EN8tmq
znK7*wX4=y+F4pE|n<YShosvY)))R~zgU*RT-HO&V*w|>#Xa!Rx!@1`ZEtC_DD`Mv9
zFPrh6*y1_^n}1$n1Er-<tf95B1%g2tS_+fif6sP#?MQMH3hA+bO&4WZG$+T+@NcX<
zup<h_$@r4{*!2y1w(p8;rHSQqnWZXb$p_bE)$zN(Fi37fjQ~onP^$~9bCpDoak|;P
z<@$?@z2t|2rC*_C@sjo=@D%_L?t<*B(zvyF%(~*^`D_il-GZ`Odt8AVPJK+qxwr;v
zh-R4JT#jjq6VuDp(W2p7+>te9foO}D4Qy9Owqk!4y9;XXY7jnpG>Y8;x`VGC`(%&s
zd5#GpTxP38#yxu2KzwAdoSY=$6W#4*ohC@eX1`H}VhZlbXzK;6)~n#y1|f#;+{S2B
zXk==eva{@2Qd}{$IqOAMdE9{YL_(zeTHAX6K(ut(HuY>m9zavuWZN}4zc!J2IL0q&
zLCFJaJF>|*2MVS;16ID>E4uCyEcd}n$&_|$9>_tqN9zdC%rd?=(843dzvzA0LcbD<
zKTgWaT9$lbC~SWWHuBE}j^=^rT!aT3T+2aqhOZ`T7#sYi9nwVj@Oy{a^Jh2P;Q+U6
z@nUX$jHZHKPgW0fW3LH&6v^%eiF1-x4V?%!h9faJQa;;oRIl3LUwZAB7PN{)c6K~4
z89v@b0^yX5(@w-aF#iQRsUdsYNnO=or(!jm8=E|=g`7zOL?($LdrSE6nY2=$vd7M7
zH3_7(k!TsImqz0S_dEW;@w>kdcf<3-ve54r@P!zxm86=9s$bM##eRQ|otHu)xF#|x
zbD3Ng24p#>3%h+KfG>5i<Eo~DvnwvQc1^2Is0wLQpuu)M8u#`MAce-U&9y&gpXugA
znsj$dJ|AZ>ZVDQIYlL07rR_xKP8Ky<e6GnLtih(-j`se6k>|_4xc^1s5|O+QzZ8Vt
z{*;%`vHwaFMvFISUk65Oa9+5hzc0d?XLsZ8QzuWgZ(_v$!dkcZ-4llRv%3uRZOktB
zET`c+q2}r-w7Btx&+w%E5Wl@FirKq~ME!XF&+LJB6P3&rC@5wRz8eH~ac?iI+(Yq{
z^zA3?VKq#ugK_athp^sj;{#f*mV6CrgeeU9^W8-4;2VgeiQg~2IC-W${%#_%b6>Ct
z?<NwSKJFxTs(US7`u3!dfz!ogHSyg<B3+Fixz(Q1>nd^#Q|tTPU>B;%5#l*)-Bdp*
z#_ab<`^f>B^rkG>lz`<0Mo2HjAEw3(RwCC<kA`l1=&%{@>L|KQ`M2B!m|<+DkSxY#
zVoz&~9zOh8?}`r~d;(@>c1*K`SQ?Gz9vwK;K}!XqjGzsht8SJ3<d;Fr)VpWB9_0L3
zAY8MWTb<`w5X1GZ<rEaFVT;=}O5Y78fUKcO8V`5nGJ$rn69><7K`SDOr+=VJV%q)9
zd8-J(H497XtSSIwHHS^9j)z4xOP5t!O}21EvL<b6U3AMmhqS+3&}&l`>Rwx+9rJ`r
z&=yE%U{l-gQ43f>V|3!Q?fC$cBI#GhpXdF>Ppm$+*ex|IZ$msq+_p8noA~n_yzqax
z-zeoBptySfU~O8VN>4!6k*&o)W9K_+1sxw$VXyXDckESjz4g2sTnp*)8~u9Xbo2#A
zxrOyMpk7|GG;izFc2T3#D|pm3Hl#mc7lFM`epKt<O}r9%$bv}h1~tmu!@0sX2H^ka
zPk-b;+oW{{N*tKYF{_j{YfGRrz-U|J&-I=qbXzt*k9j_7+XccI15C9AdnwwI)#sG$
zh|!3hb}?$F-c)n7?aEMV#qd{1{wllmx~#F@_GpiZs7+d+z4|~%O}ZwXJI|rAFBY?W
z&|VHyTig0=zZP*#y&cfDcu_w)7!W%=M8+ea4#jFwPM2f>)G8+Lz5e2EC}_61Jzv#S
z%iV{tV6Vk^>nM~a$ixR1?1=WNrmJ)KfE`tzX6yp+gvYWk8;?Uc)`<5w9&Jb@F#PZE
zb|<uiMQH8`kjUFffo~r^)nTXfF#Jm{{nOdv`$e(G-io37$sWnqekP{4*muGOypiaM
z>Wc**6&ReZ3wAF0;Riy|Z#%Cgn>5`nXfEisae2NNJCY|OfQ<=BS!Pez<>*z3PU=du
z>-3PntJ*o%6@iL`4e^}3WxP1+^;pERYP+Ffy9qRnpEIZbmE9Dc<-{D>t!TXZMf*JN
zor?S*T=(JvDo?*OZb!eLkf!kqot}l<h+hf@yew7GVq0Mnllq$Zey!Jek}}#I&5Me0
zz=UbLt0izPUctVJw)KsTX!h@^Bf`@!>|0R+FNxN*E_8a|1-=!P_Pv^t&x#!Q!mY8;
zVs;@BtC3$d4A}!0f()Vxb)tA7U9B~oPxq&ZsoFyq65T`z32qNxNH;%fn3NuQ+=US5
z+)aT)OVDXFZmmE2KRWD@ctG6dj6HfGl!}>_jl2vOL=(9sJs{-qKxQHCkx=dl>Rdzx
zWqTrqb(L*nCxYh%U0ukvjOx71#0zp)PgTWS#&WBYJtZiWUA^s=6Sion{OxI*B(Q{4
zV%xOIfl#i1N^ay=-8Mz9OFat{!e}g9ZYe1nrt7yWyJu(-J!wLh&5RMrLBgb;*6_V#
z^s!m$(~e<f_Jt6g$#*m-e&)(4&wb%OL+A(+hm8_#t|setQn_i&%Vy;nW9AF&n(lsE
z5Q|KBA%bmEeHTa5BJK}DVKhNPq98P+H_`-)3WM%nS+2(=F!x+5O9F6fFI;VfdU~Q!
z=f11dl;CfdaI1w76|Be=sD~z8_^nP41^1F&9=&kkO$x*n0#duC#g)k`^;k(<63zju
z^vGA-$*uo<EK>t2<JEy!zu%Dcf?wmZa)TZga-ys$I-pXq7XAI%R+PkTf=hJ~e0Qxw
z@2VPM1Pk{U$d4p8Ol>V5u^fON^+ZjFo0#?KI5iV#bg%_cY(Re~@kK!)ZR-TC5}zQ?
z0x)zT@q^D*+xnP77qrDTXpqWO@@b=<>fP99n{+tVuCvX00do!LEdpJfA~x%7>xEzo
zJ$E}w+@4lUKii`7!lD}6uFhTxo-#4fnVU34AleQATZ6Q`*6E}sQkpqf@5&a|eJOVf
zmg{24(4gaoRl)YE8CqkA<Nbvt;^SpKQ1YG9X<Ek=8*thV#Hy5)+QC4Wgz|PsU{{i*
z;cyJds$$f)SJfK@P#fBt*8(!v0AKLydRIpQ6;L>$dBL=K2$da;k$&E4$1)VQx~kfa
z>%(io?BmHvxZ0tLfdX4e@wE7z?_tDK@w=;=->0)H1rK3gOnXa!2?6UvUvNg#sNl{R
z(IB4RzZV~{bZ?Eul&4Pn-|U>ajvYwf{rTv+#6O%1*#>!%BqnlEO&22Xa_f%8=EsGN
z2aN|(Uk<>nDC)1McM2il<GrdIrP(zn$;@Eagy+KAYP+uaQD4k%#QY#IZcBpze5Q?j
z)^!gz1-PgWj)vB^^pdOrWMhbzOEH!tvUWS0o?H8m*caK#CEruLZobs9f?5p!t9_-`
z?1)#l{8}gz(25Z>xuo3*_-Hd*x+v;wcVlq$H{2!P2v}vF9H(|Knz~LQDNw$RrVu?&
z<9FGm>v;t+)j(T!+xM{s^H<G#-hPNnH7Wa&^PJgxi3++Sza#g+dx?-eM-j1l&*$^t
zd!j`iIgVC@@63a2A9^o{Ee<GP_;7T}c}Acv90SI^7vd>McH8(E0BHz!9Ez^J7i5+r
z<z~|!dsKh$h&duZ2rUh0qHIU(@%Ivu_k7njL6Fnbp9sjU!NAy)ffEM%KBVZ0(JQ|O
zS@)^<om<j1Xu0@ZR#CLbCPfdlGtR_L)|@=$L#s^@T9DtQ(y*x-8pq^<O^>0gh?n!5
zdq&n3zo{Xcsg}e4jaN@=?4o6LHY<7+DH-GJXt$92c+lp&7xXfPU2JrVD9ike{+Ky&
zz+s-Yxms;vPgyWz&usPmD`oxWYxqU1uPqQ}EpQvyIOEApdx-j_dgmIo4l5I)%+JaL
z1m1Go9ZTK|wXu8ZJ1;^KhiFRv3QEdEQ)CJ5tm<e=+)?6#_)CrcQguH=d0i|*r7}b<
z7xtwKqe|t9Y$+QX*zc9mxHdWDK}X{e!z&R&)a#B%zu_=l+iGnY9-t|4qtyqVUP{3l
zv?^64MMc)6hp|`2{?%IafJ%>C4jQ3^WOL-U?-0n8V8eN{LV!d)kL^=)bAq}y#m@Hx
z@)RF;HnA_m)(;e3xSCm%4PW<O;+ocBwHguv4af<$UfWPaYNZY8QO`wy-);10N5{G;
zdx5tVDI4CC&i*K0XLxyTiMEXt5!f04+qzMaZd1Rd?j2^^V{#=pSNM8COEp_W0-iaP
zJA@2zEVffGQz4TR7}a*YCo8_!1Wc*dxD363+V1GJrWKBGkLK3h)!1o!HGBb|Uh)OT
z6T%F^p7!$ndAHks{U#-W9Z;XjM$f-|P)qt8B#9XuQp2uxQU%6uE=dvus#)#uDeS8D
zT0FwtCI8RYHItgg<S9eZwfB;+;0$4~;K`}%X!Hs{;UC*E9aN9}Z0aJOMPuYU*7Wlg
zp2(KR<4c_mpUl2p9=SQ)`<2v30c3;z*=Ye8+s`NdJ9|rGlq6S)b#_Mc>qd?p(P@)+
z4>pd`*yOdSqt3-DA!>WT)cI(N`L7>o(~s>1VTW>*q;#NrQOgPs$_W|^G^#z^hJCuM
z*7Yc`Qq0vAoiNX0-~<YN!PP*GIt^q&icQAz5B~YOUSOE88ydwG1<+efmg^bkD7zWx
z7cDQcTk(@*YW8__hdT7x?da|rT(pH>WGEBZOip_(6;?W=LfmWx9Gwo;MC|MLf`{Zg
zy{{v_6QEo04YRv}GG5;Bdf&v0+^voOV)ryk37&mqKGGAM3g_s%7&Fv$Oz8VSdkGiS
z4+7CKNL{vziH5R?^<BlBlm{**a>zb#Up^QQME|ph4_ypFvbGTqDl*L$?BR>)T3G+H
z7i?Vo%)PL{#$VJI&i#+<jj7ZgxfpabbnEQlqxwVA9eG^PW9pA`@bQa@$hRu<#e|DN
zz1?%E&z=Z$b8CCMhu<KJ)sq)R^7kcEh-ZiQV6KS)BL~^`1dLBzOy3OkP|uzu{S``*
zR#K_tG`sp9#GFY2OeMv{vQE~^<({H#ig2jqXi(iWaP1)4Yr5v)VmV7jtv0nROQ|Sk
zUL196EJ<O|<Hqk?y<0^y^tfA~bh9-!?()gqZ;ti=w<a{XLS!Kfn9UV<!X|pw+PnbM
zN%Wn6N%Q0J>}E1RH@4Y=i%IDD&o5Z9uuOuj=Nm57A4s)N4kdW|G7Ummz;#w09T%^(
zB|@aH6*^lH;7N*^l&e*OQHu{k;*F|ff~g2ejV7Bf3|L(@7?h7SLK@XU!VuN(PU|Rg
zvm=u{M3djnN<nDDCu>#whA-~0=LL<NR(RHd`fB}wdeEqV5qawMhqc^&@DXIJ7pUR?
zhO=!<f>lTP*lR5@!8MMEwP|tbmS1Nb(W}DMIEDCqb~BN@-Eo%Ww&7d`ys7mB+#InC
ztiJU{2e|xyw}Ch=LoSzUgoZvmQ*0;(CpQ5{bbSCI30T3?+-=a2Kmtz*MmJ_hq%|_J
zZ^~w!td^|HW+C8OO;%-#9=V#_L1e3z9KID(Y`zm^t`?<=i(`AX{Gr2fW!+2aPZ@Ie
z;@Xk5#;1R0paOY<XG2uprH{)wGpAZ{iSO2kNIGOs<omEaG2P*VZfw~b@S9q_7D2D=
zi${-$?`ASDN7Lm+6(nrge^JUAPmhPCAA)tn4g^Bfe|p0XYE|2vH?TwTrnz_c-|Vnj
zc{Y_-wPKybP`Mq=Ui&j#WDi)_zr%p>dQ9%pF1{5st#%|3bMvrA;vbDqW8RPKm=*{Z
z;zqcU<AK8(e8v&hPFx)Q-`%x@!{PUl3pgs}n{pJM6vCy@^gcV4A%-NMyo^rAff@N_
zv%M7)ScCq}&cvUmP5!Zs1OPddk7fXvSc)J$C-@6nuxs1-=tqt=yAbP|gG8;di+X^Q
zAOk~8Y|<*btizDBn^+i_^%WgHS42aU?P@fBuF|e)6_|<Z4zI^hr24TNfdjHiJyhkh
zc#4d<b~AdR8B2!NTY(5xCN}eP;qD?AL{&Iq%g4h=^F**EIE9PkOO1{W)N4$A6^qr~
z&n576fFV-UMV>nX<3;)Gt|o}4sg<_=wR{sjbCpFUc2B<>xI(`TD2UuR@7Z@T?Xng2
zeYC0IVz(cJut$5fo4r5kH-4^j_&4^z`$6@RaOlB3556CwCbkW1pYJC!iLWQ@28Lm7
z*oW1@y=*|N*T%gcBzGacBKS|@?>Ig{cpN&exG@;CABy^ZkV(EY4&Y<)#HT{F6np&r
zL{ZVqq&%e)1hp6uo_T@G69I_hf+z0#i6O|9q^}mU==+Jt4`45SDw>f=wbyd-yUs&{
zCk6c6n&IDcp8c&&j+y1))ghOOrJEA1^L=d7)c8$ID=By5cdnG>njt8}1iD+=Om!hs
zuRSe@i7|Z>!G4p?iiHSQg4yruif{#UR{CO&V3%Nvct)@b&NKGf+;~;9u(sCbWyrT+
zfg$y-rr(^BgA{DR`-zG!T4crAokd=!+P&CHwT?x^U31P^$FhJ+s!;+#13@gnxV$Wh
zMZ_)5Dq_6aeu9CC2B{Jdul&a`=CD>rN4FD@e<2=nsI^I4|Gt<VH@5yUQ6SIov0O7~
zAWszc$qFqNJOQLR6_{KpC=HE}_f_u)x0%Aru@|<o56?#f$OHWMYN1quC7%z5y{OL?
z#gmB8q9NvqhjXtrMOUO~7PdtLr9hcBJq)_zWk)<9wLk|8I;=$7Vt(t^!s4gmwzEd-
z3Gj6_MOD_Prqu1@&43U=%jdMl)@g{v5FZ;-(<S{49JBRWhW{W~Y>3v4$v1GLRt*yw
zN^#qyNiQs9#<|V0XFdH*wnf9jp!d1qZ>unN(*TyEUfULJ^Xmp}y9P-z-%Ih3LL6*I
z93^tJ?-ab=Znm5cd6%Bi6ZAW|-Ywh^!#qpd9?h}>_jTK=hf0@>_GO6RZAE5G(%9>5
zalU@g_GimXHaD^w#S;{08zSt%Y{+y&l1NY;xqHZM6dOPt#=N~6xP)tIqrE27;M|FD
zve)&vw1aP-?Deq@+=X`Z{ZUn!b?5fP6hPUJ#W1bz8E?nqVdmUh8@JgB9ZQH9W)Loa
zW^+;#Amj_IPH9JB8_5L()j|!dPHaDx9#l*mgSP^CWcUU<BLG7UU9885up_*1RIpGw
zb3>u==(Th4wgkk^N4MI<pxOn^5mu&^1AkF_-Oy%Z(GINuT%cW!BZO<A2PnA-=}I(R
zSXpFO1*9fCq^|{%Ic|2l9<7#RF1R6pTa!-vGwm+Ei@(K$ywPsPYi<;e?zr{m$-|7*
z6G6~mp9id=#wNS19W5yb@E01dwzb~A)C;s)efE_W0&UDuv?afDE2&V0_0Bh~vpf1S
zaAb4q3VrSh0B4}zz7dj4*_{SA+Py#_H~f$GZH&tIMf7L<PGQ>L$2@X1jc%X&LE|-f
zEzg3HvLZ!(F71hQ`VWkaCX(9S)n*Tl1qFp27_x`P5@o^(+#ZfLUHxktY}{D7vH$B-
z8?T-fi!1Gsv5+dcz}$)a(Rk(cx$Locm1}`3<QzO6uZXj0wF!a%6Jt-rZ`gaPlSUVg
z9U7KSn<xkatsTR!G}}{pT+`|YA2u-InJcZaNn=5!%kzZS*yMnic&sUUlZt%VRINb~
z$~fvE+4QksB`Td2alg+Q0h1AIrq+b&41Z@&3k)tw*pp^OtDjE&FE%?MqMYQ*q8#j;
zY=n+{eIBdbGtsoR^jVu5KbO^&*u41pv!Bhh`GG=FPct<TTmiRWET|D|9F!gnS6|XZ
z700B8UkA(5K&^NXbETDOEtf2>wDPgU*id<t`l-l!r)`PwsUZ6^nXIfr&zPFX5;JI3
zF|kw}<&;$mR)O5hH8Ds@iBt1B&EKmg_S*6|3e%?8is*p5BWa1WJ)VFN1@WpFi4Wip
zcoiSB=e3?SUW;qB9w<wU%~KDgYHqfMfLQPT?M>N+cVQPIDN#W5z%P=W4q96_;7$hk
zqjf~jOMlPc1#K+klQWH|*&P!=FzVoQ?orc?;MEtualgvjfH2`oA}}Q$sPP3~HWXu*
z;`^Ee$MsqexB>Jo8=`4#V;=?gHVQ35ZyL{@vQ76<A+07}2DUjv`~OTkv_+7(c-hUZ
z+8>zLdbgBqQ}2bYm2{8D?fOGG74UGe$2};<ZcI6*cEnp}0Fq;q?TizF`#ce($&|Ef
z?7pTLTaJ@L+btA2Nq&Trtwtb(sSDF~-uA|8c9PUXwok9?k#>AF?Pa0t`W^S;+&?y&
zk`0gj8}eEB^t)O14X(fgnmSfq=YI~yXHY*#t+cUVHT{P7cpJJNvBUAcrwc1-w%&~{
z8^0EN2HQYsQsLvF2}xIUBzjGIahe^CwSk+Z)3JDPKPpnzN2A=>ukAzts~>)&!%k|L
z3RagYY1yDy*2qz>{zIhQwYM}Xl{CWac1Ba9mKmk_Y$Vp*<Kpq|cAdy*4Aa@#Om2ku
zfaWf+^V*-rf|ts2q3ME_GX+p0XI{*n*CCxiuEzonmwLZlj^DXvuEQ(BD1~!*YM86)
zxU9Dyr4G?6YV4T-ud6Xydv|m`VmD$vS`bG*Q$KfXawWn|^{ZW2W4FeF$xRAIzAWFx
z&-H#W-f*|oH&4W}UuIuu&2Xg1+m~8SWJegFuc8~2s?oj{2;bW~@tbJJR=b;RKr3|y
z?3;MA#0A280=(9%Kz^%cxD+`>%no>H2#!!y@cXO*@wAll7LpV|<8HvI_EMr@Iizxu
zMcV_H5-}u<fP69!UP@Gww3mJM(51LtQx2W<?LHiDbInBOjk^>y$b)tfDC3F1M0X2+
z<WeG4%W8}4(Mv&w(X+QTwqxmx(Wr_=yx`;UlJC4Fu_6=nvJQE|o(MGiA?AKgp47YE
zq2V`dVj$8nNaCocE~Pzkp2IbhNHm))7k|rPY<AmdI*^lMFcFL2+hhU6DTj(K0kFEC
zn&DF~1(oz$IEhlzV{&+Ry0`9(m|Piob7o$Oy9IT7pQ~Vh-eFH`SqOB+Dm+Vga6x&6
z-V|p$Ti^-Xz#{^Qb&k6D4M*ist?@$K&*#P)R-+3BXP)*AY1p+x^YvJoRG79v3ypB$
z2{ElW`j*rzEUH^*rP-kUSQenNRu-?xgi^Px*2-f}9mD%=$)&{iv=3N?Ahrx2?5C!-
zRS8B>GaKgq1Jzo=7I(?6QEN&YcDRgHr&(1Hl|zcZOF`u?ZT8QPJ=(&j_&Zyn1%zq*
zyQJUvgRRsk1R3_VcwV<vnkQMa+-UZ^28CjD*lG>Z)=w%ftG}e1ct6MBcMR|cuGc6T
z)DXR9P#a~2HEC3?4lX*%vJ6^FtU3-xz1F6cb1PSeb!Z6WQZ~O3?cC$uy0uf?RFUfm
z_>csaqEAcVb#-tKwE<0jY)j;zAgXjX>hpEcf52VPhcs)V^*FFx3+n?%G6SwjLJA-H
zhQNfo$cr#-6khJU(QBJx1E>A;uWWOm!g1pk-lEki_78unR`R)4BwKX1*?B_b;qCSS
z;ReW;0y8)lbhh!E*tJI6DQH!m_Gp*7=To9(+pTv^p5SqhR)FNkYhjjbuU^*s_q8t|
zQpO4ut$>heg<LVVKVIc(mlfLqy~0k5L!2GF6ciH^6DbQR=z-&i9g0Or+x!o9Sfe(s
zYsq_n%d3}?>-Eq6KmY7C?SHLb4P1LvSbSaUkMktKDzF(xw7ix5lvcH)ffs~@$d&j_
zoknul2acs4;KD|qh)!j0M<Xm$lPM<W`Ked{K6pfv({ZteJAiY+_Le5c_D3!!mn6>w
zG~BV>?mHq3lfC=5a51C85JNWZcIRR+udQR}wc4b%?6V76#<2r@d2iaq7&|=2P$&Pb
zjR{V}psS58#~ON~nsjSk30Oq4Vcowf93sa$NL^Fkp)OyU>so(wSn#ZrBh3zV;!5Ev
zKMP=PTFTo^z3s&a)$NL~TiFb~@aiT!WIqpFy~+l-xGVl`&5U9x_4b8kHsCo@NM;`^
zz%_B&zKWhXS6i{>?C00Q6hkQ+btk&kG&X@MzDI$0(vx_Jz;EJH$gOO#dm5yuDHT`z
zHV~jr7~<4-F=C#7_Pus6ZhFiQ&JO&b22c?rZQ}ppD&0!glRt1dNDP`a*dSK+!OKCD
z<94)%E{kd1;W+U}!dmnv5bpEXc}~r^06F|?&u}*Wa)^hu752#GM8uJc2kp`LjoB0H
z=;hD@W6y%Q_UqyC%R$7e$+ttWsGgP(i9%|Hcme?zZ=Q^?qVnm)%ZWz(_(gk4jZ&IU
z%UzZNbsr+UXYX`<#-u>3nVN@_H3!eb%!z71`koT_Aqx@R5{r<SEnMHHXDdd{Nt>Yu
zJR?b5fB8OBK?>nMdpe%|cJ;f}W@&jDA5Uxef!P7sDW4~7oRhszrBx^%@QjupzWuj|
zzjO6S@;J{NJ5Ol44lspAV<Zi<ds<;>76=--bjkY4Ru!u|bejZPR+<6-$Mgl3#Y}3T
zCw*3~CGW(UhhWiw>RMo{(1c0bgukwuJxkim7ONIO_Q~z%HEIce0ojd~9^i|2>~igR
zlB|gw>#-Fv1a7CKtdEXdc|cjEokVV73eRUNgixK7NUJpn+F@*?LV$QQU;YMl-9Prq
z*nVr$YV{70Wd+HV6=;e5NoScovNj=!<rS1WvQv{Brm^*tzQ+o2YF;GUVt2I871fq3
zUTi%=WRXWp^#z(_B8^h#HV|#=x|*H0UUxazcL=8YfN+O&*pT{HW3sXJG0Fha_J#nJ
zug5{J-!1JM<B(5XZ=0f3NyTE@9Lq;|*bLhuNX1=i5oGm%_cKy#_J)tuwrAV_Cq9W2
z_ENSbPXAnrwH*Q8`P!X2*f4R4Ti&I<cn@{qptn2VQr8k8e2>-{zA4=hTzh-f-5Efx
z3EQVRk>Y}VLZ*9JOX&;JZ~N7}v%if*z~o!V7CKoW4r&rD`3Ag>gh(yQdM0={mH}fX
zRo3EnD#!AgAk+b$illn#?1IO(BN{E`tL8R38vSw=uusRdQJm9)Q=T;#AdYK%*sv};
zk&WNj0lQ%*W64@^-8dyM`Hp%!9rJbBA6X9L>|4<;*GnX*+6~m>R}`Sh?go__73gYs
z^u|^@7w!68<~V;jxK^)ak`^$D5_Tb=dM1*K@$<zOXWE$Nhy6Ga?8|{n4xz-YNvG|K
zFld1vz8Xujl*#~zk=LTDdzq&?^VhR8azEPWM)ZZw<?XZhJ^UN!#?5RH7-+9*ZMTF)
z4Nlzd-1@mr37>67(r*ioAN^>WeWBOM!wPS9IDjv;?)7<6Abu4O^X5;T_E+|GtW+w&
zLc=?;X<TaIzThr8cQw2#*B#h5nh6f895Ii9y{E0Nfs_zuoeD60t08-a4>#L)IyJR@
zjYxOj#{#33vmfH;!kUF<A0)a3g>fU^10N)Ev~-z0_(9NBFH(}sGY@@`=ql%-y|IP(
zmk$z=Y(w`t?t?@m^Q54%@gD?jhHY!NN8$zb4=Dl@FJKh;P9IaN`azhx#{+PVB7N>|
zHzD4^6+m;J2ngJ<d3!Q`=c*f<ZKC?bKHF2m!1Z9cK)lzp@T6GWNgo6$#g<l2MK)+9
zGS%cje<9T%rwDE(-09R<jU2iK4()WU0e1+op~&$wK1kHkU8o?4#D&H)BDkktX|SjD
z>^_}@ZrH335}QzsdW#KmW_Wgt>;7PK;%9DQr9Gpc{X;f4cA%1UxPey+NSZIaJUX0g
zxSD-=BZgUW_~{Y)=B^>7Lc&#SRJO6Q58~=Z=_39QTc|dmxeMo~B{90!*{u+$5^P!M
zpsTcWPHo+ofEf6aRjY4}UtcxR*UM<t#bj~uLLr(AKg;7~7kA4Ftz2>fzVuJFQb@8i
zZd|uj_m|pLlg#3IjY4b%<z0jpvNQae)oUadV4(1{Zumgnq^F&AKqrl3z2{i8CV_!T
z8?Y9wqT5jMF4B;g;M-Fkla3Ffdd3(+F9ha(4hz1G6IM{`X3DRpFRfdDNUScIkgX@C
zL~wGD$3C5_XW62Gn84b`dIBKZZQTc>sm5e-?+IQD_HIaPg1wXYJ#sJTRD0Tv4KWXx
zF6O^6UJ%$^uuTD(t@dzK+Z-^452pCrf^89&cuiu}Kqt2~`Vx{{ux(loV)KUo&9+BZ
zG%!)l`Xw!0ExuT`Lrb2;y^=;Zo4+gV-9u@Yju9XG?rfoxAKD(ZB#hUcQW|Y<w5;~p
zYta(FbuJ)#Swq4s=54>a6=8hz$N)PKQ-uMC8$t)H#+;vaJERj*60uB&;~f$qn($XY
z2oazpRl|WpBUUE9vgS4Rdf<}#&r|J4bV@0f?Wh1^HO_kmqiEI*xZ|--a91u{ZMPHA
z%`?*U+R2y?B+obMl+d7PccYz-2iI`r?X7sgbA$Cbqs4`h?Inzyt9?WZSe<trz^G1=
zQ;u^wB$)n?+sQp2@4#e}UC=vj%U+0tUDSec5B9^Pj_I}Ml<`mOay*j=#O<h^b|s$K
zB4SNo4X(zV*TL;v3w+!l!^eAFfRaeairk3F<m%ie^O+u+J^#<mctAFUxxE!YoS|mY
zpT`r6;l%Z}mYd2a9-sV0j4;?q1frw%Wx#Q7>c6zFG`z2|^JZUX7f*@+8{pHv6N}R7
zM(VpU4r!ng71cP%BcAc*o<MQK6Cq22zj*EC@x1kS(U-;bgcRQg)EB40vHlQ;6Yi{l
z<D%zRxDxUo7;%q2a3zsMe0E4$q~8z$d+16gH=<5W%GigaUrVEluQpCldQ-We@mE5W
zC2xwmuRjuxsp<$v?a_E7N|Z4XNv{OKE^1Gq%cz87+=U(N(1iF+G+4o&xDq6qbhPM#
zg`8(4n5f}Af)(?~Q)*Y^x{sV5s!^-hBz0=*ZL-N%5_LhciI`sOLN-P3>$o_k>i0@`
z)9FI7fX`&B*WH?NC9xZIYi(x0ckO?VJsqpS_vH8VtawFTb<WA`00)EiC!2F6u^?2O
zv1g(OGNZd^%#B7vgOIR!@tgc6h?-Z@MF4B;{tO7m*i5w{?(-S(?ZpALh}|lUxp^TR
z5=chPiZacIvXWLFn4t$otX>iil-%N~s)(M*rA2*m1+c0sA!@s2(G7Ifdaj2e_+hIN
z<Z225)djGe*UGTv+LubQVjy0w&|_6s&sunlmFl(x1L7)m>vT_$=dUCplgl@7)KJc0
zBSLCjttM-5U$goE#RRwlkUj;uk2&DW(xj#h6lt>-olH^>0;Nr_iW<PLL$45~_z((#
z&+u;%OuDt<B%?TS>7S5Ad!lLE@NaNUvp$VCv=(6;U!rWhE(-Oni+8>$?n68j9aD*o
z-`jf48TFuTi2WM=H<I3F_^-v44G$u<s=*epS+JXiPY^O1z`7%=$kuq}u?h{gEui%b
zblP@JtH{eB+Dq{wSK4Mf^a3Xswryu@V<{_~_<~*8EW8fhZZ)kWtwW3L$-aKf6P^eT
zuiL=5+g<57Hv6<*?kAmhCHc$IgM27mWIRGNxBZ&bvlxdC=yx^W*}=fCfB4sSDE`d#
z(rbqUoA981HCqI?WWT1vP3)d)au6uc-3GeJ2jERR5;F@12QB;2EAn<rVlo_(n`o)i
zBBU3?QsB8WXvbrGE)haUW61XY0#0`);{~=GinIjq3YKj#x(z!WSHS><NCOG7+B=at
zll9?zjs!4Nf}FF_m~cIxz`6L@${luIAC>C^`mm;6(2>9g+Z7}i16rG-Hm1FaO5~RV
z28tlN5{;VZVpp|gSnHu-&T3v$x4Is#1iK#ZlXG$uf4LzH%DBnU$J%E?I@Nf+8MFM^
zB)g?%Br-hZ5xJiWQ9_zsA$dDqM>+{U`Y!^f;e*YnXRk;%_$Ry<j)tYqzKUV05w8dm
z6|fYp-)ndD=+Y8H-HoaC4v<qJe&*oN@5KmR10(-z-)fE(7*t3w6ri0cqh@!!>KYjS
zBbLP<vRNQ(_#I{+1`!G}&VoPiVInr2)Yt5^2R{sY)_@-yC-5OXQqUK%z#jfEQBZf3
z0LgKJRGGybjn^L<y4sV!i#_t;sJkIia<Ri6jo$;UBY$g;#c!?+WXl*n$M-}cg&40k
z#r&X^)jLmEZ%=-hXbHluP5dzKX)<hcEfW9vVIopc&RjI&4zWpUl#-1xm&t+M8c!}<
zh0J181l^SfHdPM|{alLEKMWqgZCagt)6@|SjC1WUnll4Q1V}ugC?Oh);({GMLbE=M
zolj0TSZ9&Tj@WE1%>87Oa{}DZAU-6|#P156mTa!D82d-$nR)R7W~IyK#~?)G;Co~X
zG&Q2SlNO^mx_MgYb}JQ(a(6i{3m7l1nz72t)e3cfk1Y`hcklMJo)wxh3~u6YQNu=a
zFC5fr)f&GJ+ib7Z=vYAfQ59L85cML?%QZR<BEPT|dT8-9?JGYF9>1omAxo;TDxgwG
ztq+sy^Vt|Mjf1v2J9@6A{)N>CsAXOk)f(cR=L+Tywx-zh4)Xc6X!#doH%!%KT%vFL
zFvzT5N@zwEGS?AUU;NYio2(GOyIeog9nkXKgVv)(anC5Mvz*qxz>z>0zi0+DzrS%R
zvre1rjB2Y>h9OPZ^$;8bPlK{v=()upoelr0HpGC>zwno~QIqb3p+h*?q)jSWR#UOG
zy3RJoIWI4_E%$dI)j!`VM2UpuEBD)Jn+8Yv?vHX_wq5IsA{&>6m-Iw|9)#B&8Ln=D
z@!A1LY^PwhQerT~&0U&q3EEuSE!+t6?HI&Wc8^vD3@~8sjVJDBchK4wGprmyDH|)@
z=WNpc`12ZqI}d0+f8Ea|O#UG~ViERO{X;^f55*n^AmAU?UkC;8;8xg=SEI}DVbAx8
z_$UZo4iCxhb)jBOY^NR3yAD@CQMo%{Eyf)d8srD&SRj@g{(C$AVPc+hY)7Hj^&ngZ
zCo~;gtxv{0Ny_5co=#~Mo$y}zW%~3_XFIMW{oJF!x3`2%g&R1|1U73~DEE;Y(YjW<
zR*BDfRL`|Ylg=HU6K)k1c0O=J%uTK!XjI;`3xUuj&lQ0IG~2P^UqemfRnBYLH`-;b
zAURwC>qe0)`b#$#hX*}i0*xLxetE_0YwC}O9>q_u$42IQ&`aEiH*ALUjGx7aP&D9i
z_h`!PW=vseV$V4`x74+6Ei$GsZ0{L<!)|M?UV_f){}*cLb!NWQJI;HGNBByUCombs
z1eJ2RF~1ftgp8whC)U*yN08an?q(;PET$3fVBZL2+wjQ_x3t~Ux$v9~tn{}5jTG;E
zPv6A@fhRnCuZAd+{2c$FN4FPHCXV88PW7;2;BRy_B)Nv+;|=zp9tb|f;pl2yhe?U`
zM?AGQ*~3>uK;_7?_~Wi7QjEb0N;7EV18T0WqShXXh8!@L<sJ>NBtl>-9*bT%J_@h#
z_PBacoXIBWcQu-uC$0t;z~Y6!{F55HxT_wo?}=A~z;c&0xWCI&F$CHnQo-lcH+kMW
zOEKwcqRhF!p<egZpqz^cHfR|>2uC?314Sjmmf>I9R4oD?Xb@VB8vYwI0(ZFiQv45X
zrtmH2etKHda*HQ>J4--tF-73arc5Xp#9>Z6LYv&=WY0v?VzQyM*xbN$X<3=g6JCl(
zu=xUs>tl;8(9rIgS{yI98IYw*)e1^JU}c&tg)DF?b%Ic?O?7e9l_=<mmS~(>Hr@H#
z3auU&2@Yv|4OLgu4TIZ9h@pM~Rcl2{E3HN&;;Q@)Rwuk_mQ~sE_?>U(n5~H4Ylzdc
zl^PI!Dg|<_QlAFeJ+GzW6ZkFP%j$TQLSO6k%2f;4c^ia5U#B(2CSh^)Eo;%1q1wVP
zxJ`rgwYDK!kWW6k{q%(&SwZ6#BQ<ra6G8^uYL7a(+~{h;KD8vW+Xey+X~J!tRtt%o
zs#IJNLqhZHKd(iFZR_<|8cqTmGRSCBJ+@|}7LVU8wn^LN3Bb7M`NB8r^+H$dY!N8b
z$F8;88Ym+_y?JSiZMzz5oo{S1TevK?T?^BJs+Qymnq!4m7`GkS2j<~b?uWZmW1(Fp
z{xk4&rEKyw+^xO@t6=Bab2YKE*!cDeipRLXhe1gL3bW-Xy&Q|3CsEq|0DzXsJ+Kc5
zTQ2fcVT&V$Pf@ILpNGmG5;B~)COaGxC&U#md4fIzsw486(DForIOD3**deBKBwld}
zoM;`@4pxu>>6n(8PlOTxN!rxPKIb|FB0V!)y>_Ra)EFf5!|m^sno=3X*^<-R&(cAv
zTcYHAD^Q3+v1j6euw;Oa1lEm28;{1KLE06LcP_p#>NjKaVo)b`=c98tlwA-GUV#(|
zG;k=xAs3CYjRg)I{|>UtLpX&K%m_9k$fmw|1eoM#0-9F?nOs$kUDKj*-H}v29uTHU
zQ-34ST+GF>uG%l1&tefhqQq`$9bj#SUx8w(V^;&R^ykrTt)~XLt%Zj{voF-g4I_Nq
zU+R$~JAj?;D=h*^Y?T)i*IYFj`<F}GF&o{{&XwO!bA4Cf-I42?m=xcR-HY||jIQJi
z{5GC?Q4z5`-)VX9%uMp;_j-06win3xA;3&qHO=g!pm*dB;ktj|BY90a`$^PtXe;f(
zj}kS_71dJ|R2^D|k0Ve#tS4@IB)@TbYHIUT$m2f>IwbCp1M*0~f%!!iRa@uNiDmgH
z5!z%%h4wujFP#UWjOm1rf>3rN==dFEPka;<{W+(6PiBL{cK6!E0F)a(nX;ij6^}L2
zYv%$1EMi1|M!!~EyZDYL2gqLJr=Qyt!FN-PO^rToX|?I`q63Wuez`)ceJxdc+@Eab
zN5Q8ie<HG=CtIhMZV8h0tdA0n{>iErY<B#4_%|4^<_P7wd@FU)?3s_ml(;4FcZq?W
z`%&C{eCaGRRB(VhVxAz^AxO;sD0l@VN6%?$3j}g(FWyKStyoJ`g_zx8rP(ICV=Q$Y
zUD6=hwoH@GuLF3w_O9f9@YIq(zJ2%=s}S_$^FT&<bu8ufu<F1BTG8pOsMW-5{LW69
zleqEg)V&e`j!$WM0QP|X74ZhCxrN$)wlW^O^MlD}ah3WtwWhq~&ujigrKPqyCg>T2
zh)u1}&P%e<0k=W(=Xjw&ZPH?;!^VByBK#{z5Mga<O3HE$T8D-tO|CO`g%}dI9+b9v
zg<rXQZuAJqV*LNC@1sOKN(p$09<@uC)!4dd0!x5Sc}NRf3`asbSNM(f8kveO5Y-KW
z$HI{VHzq`e?}ik&Z=2#R`5#d6ZH@<bn8lJQw2D(&8QUtLgp!f=bem@0SZJeiL4$4A
z+a)kC_4bm6glX%u9RbI+?k+v<)S+AiN#}cZ)y=M$8W#+f%m9Rp;$P$*0jeK9jwfzB
zV0U#PR69czKfWN&*~>zAXl?7rFNbWu9-npPf58sizd=y!An?}PLE!<h$9tXYjm&?V
zRHYjU`793yUeLWRdo|A1Pgl>h*EDn!h`>1H?e%y^9UMENE``Q+J1P_y0f!Ph7Ohs#
zTy4j-S^&Ul`$P=O2jq%{H|(S)ji)IjR6uI+TIZ8K9q`9?jQxwfrEfV=h*Y_d_=z)`
z7Q2m*>;4Bel8uHQuG2<^{nGZ1d<*qZ?416vyo2VQTmk1bqzg{0!i9LBMCN39yQsDF
zFA=Pe0>%QXTv1)MU5;rs;^lfp>*;~1Zh^e2XELF&yZ8=8(~89<c3pGE^NIBFF(QhI
zo-(TAHrQt|181_aP2Lp3#f=@!DU|V63|~#EK>IwJdbTwV#cefBt3iOMeG%W&e|pir
zj1kYiLfEc-rQs{wu=lkdo^?Ci9o@wEur%NSowCc`jRxEid?w$-fSf`P(z~a=wOp|F
ztwtx!3odXR=kK%tNC(Lx^?e}Tk9o#^(5ia|LzG3-A^%_9=eD~mg^oRNEnWD%81){!
zmWZJn5&Saaj68HLh-wvE@vweBORCp#*Aj8dHxFV@8GkKE3U=6*)&bZMf;1k9LC~0b
zD3m=KgAD&>$R5-1HD1lK#~#0?bRPe|K(R}EP%ce~L5CXIohPm(Ho%$TOTV-y1*aLU
zJ74X@YeA8aMEQ(6ol$@4T4tq-;?6leBX0E!;>I;8x;Q1IOH$}e4xl9*KF|6yMPrl@
zQw?RE8tbzZr2~~6oFY%xsN8B@oxK!|!LcYB<!Tcj<Co@4{iPMVL#i`xPsdE~pu^s4
zvtktdXOiRS>{#2TyhoGF5e5U~x`O3=CYEF5m)+zmigr1!C<v_(89M--Jc*{~O`Cr$
zaU^x@B|<(-!h&l-q0tvPPnu6tEEbf`E>z5=8V&28*VZZ1BWIk1VpbmOhl5AbrPva6
zt#CKxiolF2m$g<EzXx5RR;@<0D4?xI4_tfoRu^ZY9wWta4a8Qs7GZ@R-6n-`cx9|s
zj;+NlW|d$N5<&vq7%8{7xYAbZMIAhr)$4&#^VSdm{J`<uHO0EHWlG}NA|O>@*rwml
zlGvg{Xt<t><5bWwk54SBO?1aFu3c=l9*x=Ll&4Qfqt}G%8i=LyTRpJ}>jb)%x=`@K
zL)Q{5Uj1WRuQSU1NhP&j+Yrclc_8$Yh?^U=iSS*7wxKMhuz_vT$fRfPw9SHDN@Nxq
z&@EbKOwn7T8F3(-{%vYTEMg0Ax8Gl6x3|6&PgFqlQ@%q@*Y>8G?>hrx4pO%(AQlxD
z)sjOu{@%ul?}`2{6r+XNEA)~bbe~RYvdQ>jAfLvfkNqAc)&3aE1xf#Q4ro1U30Oo=
zVh8mtxy2mr0lIZ4I<N6qr^7MO;;PbxE9}*Ha5p6~|7D}PeG~?TS`n|uduS1a;Yd7?
z#S>qKqdEjDVMLF`q$-v^XUBz$$GA2k0Gx<EPJl?KppC-1=%vufs2i<QWF5*gI~{=j
zV%ki5OY8YuSO?B%PS63qBv;Ch1mxTzkAfH#WIWE0l7W0q=nbG_JFk{#q@kzrV!)NS
z7u9GHcJ;A1UJYyQa==2ClxE|Knmx-0c~!95+^_9gyeN0h>AGGNA#B@?>;O{zh-fnR
zjDOQU6PT)Hm3A{j2`^r|6$fFkw{H-M`}6D|v49ZY+c9y}9#Mz*gnbckDi)TZ$kmCg
zBcB%u=i^PUyIX!8BT$K*PwS4F5ekD7+FdQyrN2jFKx(jWVjF64uS|9I?ggBF*MNPi
z#!0+z-OG2vFY&Cs_I+#_X$b6x_~}(W?BhfwJwGewn~%^Q_&CuXGVmq4xkvpx__4$+
z60f*-ke}s;K2AgvZ)L>Whd)k~wZ0$LZd|+|{XW{K@gIk<Tc2;hj`T>p!lM?^_0f;x
z0{3q}79Ug5V;?7Sj@l>hnvll>t|t_>3F^GaZ^S3`fPJ9Po{R@rAi_K+YQz#8X-Kee
zza3wG?&FYZUHRY{Dr{0r-tS@xOrg58$<ZD$hMk!bE0II7&dtsLj@i_}&u=h4+S3IQ
z#ht6jkx>6<h-~%Dm_nYM754PUi8~<yrOo;{Bj<$0!9C9YI9-~>Wi}@o4gcQ=rWpeJ
z*d6R0EYMuNeZ64w0z?pSuJrjabIdhpp>09TghcMlq&U#de1%Fi_~OW*VrB6RSDliN
zS$VcKuEt9V$kD7UiIH=wr&vYIDd)l4Ra%1He#(rw4XQd@0-W3m+%6bS<1aWKa3Q3_
zQzrlen6;N{i|SxSZAHwX)Z;iBY^9pjEUdFt@f&puHdCEP_6-$bc+0E~^g78~RIek@
zPo-a6a0S0<_&8bpw0~eWObweBJ=UUzW(C_~jJiDfosN&?RC2%Wmf!zY(3TL^MGg<F
zEXbuAppYM0YdsmtXm$I92gOdgln3;L)WqkrPRr&oDsEu2q5EsjHn^hN>-(%1e${X|
zL{e=Kh<Xf!YiwieT5)3wxj~ueChejejcs$F(ug=|Thy~N<)GUdYc+M|blWC;n-I3Q
zs{`MLy%dx3+L@?z)tfk9*yx>s6sZ+ph)GElh$E#YX%OQy_K)2Gs+pic+atU@RgHVW
z?bS??$J}|_g}KwVFUE6v1__Pv;Eg@g_x^ZQ-P6_8WCyYhBzFKe^1*=ZeiW{Tv_mm2
zR{>&aK;d%kB6rlQSw|1JfP`jWu3N{6qO{ke=kV(YTt_tD8W-v9XuQb{|9YJr(_R<%
zcQo5^odpW1xUaTzO(%5td#P_5er+cM!(}jC$^YV%21Z7Mq(Jme3w$Nl&szfD+Q4;n
zCU(iy1|!+2bLP!`mm#EaVf^JZI~M@pKA@oSo0{RRc0n^nDda9E7XuMr;6!T1Vv^ka
zUNiA>JPrREX1dp|1d`57xzNAXt_lJ7!R=~JQC8Q4Yk9r{yQo&IsnN{<H}u4@hqeDq
zcz7fv{P4}_=V8;0c1!Qv&z+8+XH&s*`nHA~d)YOkPN3|IY#CSpKW*H=Uuvl3Pz4hE
z#gbLB!C%KP&A1a%!9}eW_vXoGcsB-ht)4TX-(*K&m^`rsWaE3<ksdO|e;Z)dkZFdh
zp|QJ};ThA^_rm|uOUa_~!^fj3l@)pXd-52QA(j|1?Ruh)xpuPXK5#vefT@#a*n{zN
z+O(hAL)Q~A&B1z;qUd@^hRzZH!N%zos*!dZ9}ljPZ?H$Mhv?gm=8ri(dOau#w+XqB
zo9(e^TU77&>f-?uV}LKy1U0S9w;@9i%_2y(sR)@psmDC|zWH({UY9PEYTX?!WjOhN
zDxmlG)PqUxx)i9v`FN_iN%6iWIjK+1-iKqDf7DZA^2LboQ}s4g@C;wI>CrpaO{OlJ
zA>8ZSHP&Wc4<bJN3g^k4E}j;)-FSl%uRTj$`=H@ATcBN7<-2#1Jx80-0Mq<TbnR_y
zcGlhI#v`RR$rvzb^R!BC0Q5{y^RI_?-4$u*2H-*#Xs4WYI%36wT*?tDjlO@Oe^w?~
z$f8(G(e_49jP$s~+Y;dnJKt*+8fIz7AZ4sQhfkGeu_%N*U;gUopV)l}te}&}y44F7
z)u{s|L%Sn4hAr0+EAuUAef2=?YD%FA?c)AsTNNvyYRvY0yt?!&x0=R=)hsHu`slvY
zU6ZXL+Mp9FSd&mMarQ0MIBkihh2)girq|>??I4FCoja}vW4RbzHPyw^$tpw_be<gQ
zZq1By((B19d-U9d`0IzfFB+G!yUjMBo@=;{ZJkaHZs-g&6#ekXLT3rxSg%Pp-~h2f
zCkhtN^RFJUjlwgGAwJBe=#Gz&=Z!vLn`1GkTxVOd)4Mf^utZ7Q)_9v+v(C2Z=U+^-
z?Lw%yzoDVUUeZLI5A_jtWay%fB!(qsk}LKM7`ya<(x%6D3-MYG*&e|_iN~ehV0$%K
z=YZ`CaKpdBnYYhg)+bU*Mle*H`>zLU<BHj#;p1GT7><!n1)2&Ac2H}Adk|k0EY~5e
z31(y&?hXrM44A~Nz8cHGTF1?dx%V~Uy<+hST#*EDq3nfrB#^DA1l-XWxS36~V}j~I
zwp0yOF9^vX1fC$#Y+>AVcrx0$#gk9!lxDxE+D=C+Y@V#_TY-5>KY>sJ%X&sYy3hWq
z)kd<DzygkHot8GV=Sit~POmC^o9(>tDn=RKY!|WxB`+^m&_!)TwK86hsWCns@FbTt
z31{=Teh2MJociPp_Y1dLa?M{=7fLd5C0-MTH7G>wx}f80mBNi~WXH3SauLwyLHkUd
zyLyM;sJEL!&ObBmlUr&|=3Sk&&owVTDZ3rNxzh-45g43CP~&_lFigRXUtitd3aV_t
zLfF^Wlf+W9%<ib=bBKc6K6iz07F704^kkddUc$k+7vu1~;J{?xs(A#;lI%zW5kVL}
zK;K7$Bmn&nNXrjuz`C7sgQ?jkqbcv-TBP6yJ_$++%4DVgBvI0S>SOVpKNL?nLd6hM
zAO0i=@w69hocffxJTqRc@J!>_KcZJft$c`&>eWI@onk=MQ!z%Oc6(e;wSB_}8*D;6
zQ79aZm^~3g4gaRwp8O<;G$~kEL}=?LK~s`aE(q%<VackpoW}OT0yZh;IQ$#*QIkIj
za@(2WE}W6FDKSKAZ@*0q01IBQ=~@>urP++YXtA5VD8U^N!s(TBi38EoYD^(Cn<XsU
zTYFG|587-ESVL7ZggRFU&4~ef;a8sdBq(xG(wOsZt|sM5zHrr@7r^st+90DbEjUf4
zwIE)!HZ@VkMN0@{2iH-m**6hzU}f<@raE@0JRZEd4UR^dSD}sj!72ozh=YciRCu6X
z>y$&UfTGhxfBQ-BH3{!Ny1SCRF6NcTw{v;mcJ5aw^;W1OI>a7ZsZsE5V79B`$wSaM
zoX>v}Oj!wZhHrnhR&`+bkktq1`LKQs0nSb9ltB|XWJ%OpEn0g5!N?qi4mX;jYv<nS
z&`ZoB9gQP@Z-wmmc$%YB&b3>AA@LEsy5Chj+8%hNzGSTJ3q0z|7urAoAHsp2R0{Ec
za-3NWeG-&8yobp`!HHl0$!Mg(+mFq{-7UNX9>q-|c^foH<X%Ib*r;`IyCYk%N$b##
z7G!h$h6CucErD)QZ@Jie%C^Q7OInD-@-h?K1c)+Yd6aJ319==1>?Qrgi(*Hhp6hMH
z8+xbqHObMx1N4zgTWyzC4(%p4LznH&dbnX|k9GiZOAg<?(G+aRZ32(KFKg-%2rsLt
z=TO7}NgoEutw;!&T(%Dg-xO_3_=}K1+Cu~68W9%vkVfz<S?+XvSU@qFaGSpx>r>)+
z;9iU09oX;eb*=GzkJ2OA=&qs&XUe0(wPJDAU)wQ(Epr2`9S=~{uc54qPRSy?zB=-R
zE!fF;>oo=Jl=>}0M^EP5SmrL`=HC(qwP-+x57X<6T4KgyS|h>-_jm|hGLDbNG2)^=
zk+#FL#heQq*R}L@cG&rVfa5D+2X;Y=*XgRff?W)xaunjQF%9Kv@;<v9Fp9f81LzgK
zDj^LmNdmRgse~fb=32~E1#kF2yB_VkxtsXnZ^R&-9=r3ImanuG_UdN5Xzj%@@RnXU
zS8AVY8OWlI(UD8&b^uJG^hx_7`caRi*S?J3wIqMGud+`wndn@?_&VO^U2S?tBMi7B
z&)s+^8uBaY8%+<p7ZpbC=|IU~5}_8~exfjQR}`LfCKX(~yvcXjcPSO7eJ>=Ee=qVB
zu1T@KFTVH_vl~I3JTbms9S__H>O?&}r{3Iy58eoUE{=p$uzTo6-2AWzO+-+TeBnk=
zCNy3Utwz`Rj#C>P>ftwF55`9u`Xtk)kMl^hxwUKGtz8@W&mO%I6ukl=?Xf_LPi!j`
z|4n=RMo{>6SDsH0#)Kon)<bG<#PwU1Qt{c7YFpy25EIqbeNNpF@RXjtq@ZJwi@6ch
z+Kjn>(v3uiUs!LG0~vRTu_@{U1CXTAsWBc2wP6?V?A3y0`5aIP>n~VTU2_h}pBdna
zmoBZar!~)K@yeMMzp(<4jUiBh61uoP=g)|fARD{sjl{a-F8z0#8{@m@a4r5$^WqWP
zK#Gu=AJfM&3)jCOVB$r_Eo8-7nH2C;s#WVIry3ZR1$+{BSh@DQfO#TsOX9_UnmWlU
zgi|+36x`Ta6%A_>PioZySc>BWGE&VAsrr+Hm0L*0j$2(U@QUTOJbrcH(HjG_wi7rb
z1Z%qr)aQ;Coj<qd<Im)HC(dxShN&PI5cH{DkA#Uj%hV7}-ASdvngU<kySDq6V=Zb+
z1h^~5+Jqrw)7f3Nnwzg9=H!|t4n#p`n;TPU$E{m9QX~WaUM*G11z~-$oBgg$8;G%7
z{e@cF*2SwXqTXyMUb)rYlR2%=#-R8gSwh@Sy&+ncm65A-qlQ@2;w~4P)R@oRNfmlj
z+Z^jxfPdN&xbejSWUFSoFsu~Y^i8fMHABjwvOP}e{X6n6?WKU!gs1F|z#unu`gGeF
zfBwk}FW4@vs3#Bej262!Mu!_uaOc$lS>%Eo_xIlD$%(do*>NCihg1HfI(u28JWEZ^
z{hBAyARN^L@#MN-_=AJdu0YjEJEV4GpNAM7)`*?Rio_Gzt1)7(q}pDK#lTF%NBX)x
z$z%yni6D+d%Z^5@rP>fzaWacz(UIzOq_aDIzhgvHK}WR((uKPr_GBE@|Kn$ND*8H~
zik3H!OJB5KHgCm{wT+#2CVrw-wGmCBB+4e+s3r(q;~qGtx9*kZ*$2;S8%wju!wZ2t
z2~$#Pq>I`ScN8GXaV#FakO@qgT@GjsU1h1Xj$IM7hICVLKVA*|!~Nu%-ulOL$$eR_
z>ks%8lHQNt_!}{eArvv6X;~6m^m|y76LvG+=E}P4mUg;`?5UsY&u3n-+XC2xPON|U
zl^*-zMzEj__-?xj-<R#nz$xcpwDwg1TIHF#zSaxQ1<puXwi3@RVs~TSi%W{?aOkja
z)Up9PfZfv)LA+ee{cS+(MH_A3>6JTOAU%G6fAMi3fjjs?uPZ!JV1wDGL6wNTUTY70
zny4f!MLU=BgP)Ej@w61Z(?j}QOUn3%KMm^Yf-Z`2F0gT*#ywfdrCEf_KvGYS|1^kl
zWy(N7&2EiE>OYje0i$+}d*-FA>W>KqSo+7#_s^B{c=YiqQ+b>4X;AH+MlW#-)qdjB
zMCyL@V|!BQl{2G>F$piIgE{f3PlK?cckz(boM5;cdWt-or1!OiyBsgrWFg(&fxg7)
zW&neP`{g@TQyBXd57Xlns_AZm4`)P2s@n9m`i{&LpnQL`=XbZK)fFWb85ll|6GBf^
zaeaPk+w4z6xmD?^Q|Peg76pMh(XS5D009-yXZ@Z)ku>+yL{O_1K4<f^eAGa7`%Eai
z=W9@8vURpV5AaTFtT=jBvWlfKkCf;3_f{6I;O4lg%k|Le+T0}?!Cn8|MK>`875^_)
zZy!|kwdQ+%CzVQBDX*1ESym=xS&U<QJh`^*IF92uwynl^j7N>F#@Jrl_5}?ZqO=lY
zJRWQFA_5{J0wN+JZV?p`5pnY(T7VwoaR{-EF}Bz2y<u-aImYYt+No43HQ)97K0Dp_
zul?E2dSAcaZ@oV2S@N--+K4=ZYBlnSH%(n==`D6|$fEyx!`on>nToADK3UCs#@a7_
zjwSZke!n>ZOO>NtqgG&<#BI&g(pWBL*S`)c6eA13H*%%@S?lzSSLx68FWjc_8pRaH
z>)hilM{c2zeW@I1tPcr#b>9CN8|{sF*A>bQ(GyLunGc8RUc6+~YJL6}o}~~ZDVnY8
z)?<ru=NW5DHg6kfT%I}z*d}jF-(QUFRu!BlZN?7E(}#S?6S33&t!Jy9fL(eUd9yeV
z`h5zYy_^e~5q=G+=GUE)`~vCrJlKVDChHvo4Ism6bV0ixnfk`uVKr}DeF@o?j|6J6
zBgeeHOXCo^a&W6V+%+{{w7k9*`BocU2fP_`D+;WbTi0>Pp0!={Uo~{*B7aQzuGnIp
z$MTiLitfxrm)ajrQ>f6ZN3diW%2iXV*>6>-h;GS{GYG%@&q$;_2UN<556MkMs}yD)
z3vx!~<amLW`Z-FZDF%h3l{xlJHI^t(+s32Ts?f6+fjZ0Rd(^+Ex4P!2#4QXo*k4-n
zb^65tD>XH2sl-jo3ktc~U6|ftUa42;mfl*e#<(lMr3<bbTr+p4ugL0cvn((_)bQ!B
zh-o)|owGi2!$PzCvSoDB>X+j?-6y&gx2&mXKe=V=e<<~B`^NtLNpLQUP9^vCp05v3
z5vR)%-~6dN33|ukR?gjk6L(GI_AR}I-W=Ss3gmg5c-*)AIbBV6${G)>4rlQ_f2fBt
zYsNp|krmo)O;G(KXu+KJfp-O5Sdn@+lAy<27g0-<jSF>>EHsZ_a3KeMvB+rHfTP97
zmdpq5VnC^LX?HcAin*+~7SH4*?)?w(+_Jxo%Y}Gh)t#_<4ewLfgqLz`Ked^QIzL+F
z&@clVzn@}>Ht~fB;Abo6V(zH@Vjijb!7c`m)X{z9=5p-bUwPzpIc$bKy}ef-Y1Z>O
zd(lVg!MI(S5KprV81%^NaGjzK%k}yrb*${UxQFCye-E~>vwx2v@=Beb)tuMAVQG5D
z{MgvMDUY9RU4x<Kk&R7z6NbftarDH4q{>>{=Um_9NA$M+byIE*QHMXe;X3@)?jpw#
zkGyd~`*CN$H%@=zXh$sxBNgKsM!~7fD9ba|kT|_?^W@Qwyk2?ZKVyu=$GvyFV^Pdb
zq=E3Uv2<zokL4a`X;r#zX1v@!T=1?p>X~2#ow+(;Qv&rUCO-PL>*OXcw=-GcldO!t
zYj5DEIoVQL`xy^UOtF-_vtsd>YJZowa9c4=wZ_#XOjkNw40Cf58p^deIf!6GG|O_a
zh0~&SydiS7m3k|WoO#_X=2%pGjpoX;w@2!3f}9tty$v68Pt;=K!nxN=EVMde<zbQ1
z<bo3op<&4*>16X3722fd5`udV-C^iU%wzyd6)?`JGA^_1EaWjHeO}8SX+-7@P&tJT
zJ{Yd}TIt>GC7*nOmG(BeRS@n`ZpA8#ayFmh8soWK#g13m+ADYujLNcxBa}oWSbV&1
zz#Y%oXhq`zZSNYX#MpyyMkqF!-}jcizj(=FBw5<tx;~8(V9O&75ZLm2yTKa3B->Nq
zIRwrlwppIu<0WC+ErO-aii%T@-J#T|%gvJPjFp7n&@MUfcq6+rpJgMS76F;7RQt;n
ze`^0D-EzE`=7Eu6#o_5BzLzwkYyVlto85X{wpF^bXsb2s9HqL}Dfr}CRO_9!**r^z
zjiytO$u~D%PyY!D%+1*@;k#V)$m>q|h<?>!DOR+s5nceJkL$0;p;S^fK-4>my^fr*
zasiWj0OeKzb3Ub+zek>}`8Zy%t%ij$xt%e;9_yObG*&O~9A(9$N*Uq{>fQ8t#`4ES
zN2e`Xt-OEcOctXi_Axs9SR8b^Lm#z@bw|fx)X6XP&0PO#@kpl{4N9^7IQN+X3KF+?
zHkzy?T8MKM(qiG<N*M=|t@d}ZcSa1?6!h=@?(fiM?+Pn~C&D^n->>(f-%zO4YZBW1
zjPqv9E5}tdrE9NJ%&r%1D{mXNvxuGIkKIv)E=xO!%ElbP+_A^_>AD>oa5v`0t$J#}
z-m@sS(ptuS@um%|k_YCE?OqDI#fNf=bJNC0`T{su<JJUMdz?+wMg;9cp6?W-xqCvr
z)lCSi)El`()uR}8xZ?*=D~OxEQYobfeO8;CQ?2Jq$szh<%hyJBa08YX&JNsk@2Na{
zoxW!`7tbtrBgc-!bMvej8(t^}+}y!SeHU&+z|%tb(Y_OYVaaZO{Zs5oP)(@?KPygW
z6cfML>%&1iw{hKj%lYW{nve2QrBCt7_g;rYt(Mhz^?Nn^PS+=9Qu{S|)|%-c3;C(r
zpLzXzuMgH;!SNWZh<JbJMGX1gmV&9)&1=#uS{*}V>%#%to8Nmq_8)({5<|cDx@6C$
z3WsxC`yQsKcpj78sXWHbo{6`WVBG92>)2jkxLn+AKqC}j<di>*MBMk@P{A#mab0>N
zM#_y{2vvB+3Epz;_Zld)pQbTB_oa<7fh}WP!aLt<*zGp&+yXPt;*3=VxMh%!F-{qe
zTfN2YSGhgL@fL!egtrmSH-Ca<y6c$R2Es(E8Je==n>9&}OA=^ZgR4E0mA$y%v8A74
z_3l}So@({Zjjv=(Q@Im6c4NA-%-iFbA9qv5Oe+H|l5n2IHgcBbao%PsBXIe6HD;S+
z>uP@PUhH#BB3q;G=*e7@ytjQFZE?F5p?TkXBaVOi&sd;LZgdUnLMsi2-E24)S>hj1
zW56>`yTr0R@3XZ$S!xx*{Tuc^SZ04Y9ddid%T=hgJ62=Gzdr}3T1{*8-{1IO-8n8Q
z;k)xV-g2h5kKuIx+#cB~W$As6XRsz#!8|U%7V(yYq%9m$uD2vVUdPSVWF+WCiIeF_
zT5eD*d~}<UXis7W$AO!SKciKFB+DC<ce<$oY_a5Ka3zyhzo--+vE$yRP^)>G1KX90
zc0WB~M=TdkCym{x{bj+pVTxUf&pQRO8NUBe6|T`jstITSkBhgQNvA8tE$ojnlw#7S
zY#TF`qFec)J$frBkJR!t<v~+LGJ|eZ+Gl@~8>=Pv@#R@@6WjgBx6GMcy8`)hXHeIn
zNN>b@e~)72A#VPBl*k8fX*&KfN|mlQgV|J;#bR?8)+eLfvgH{cXOmFzeQZ4CHkRG`
z1~<n1=4mxAUvT>ml}a@s300QQh3jeR$O^F7?cG#cGV5rHoqHQKs!?|dEnmCZSarM4
zu?}_i_*Q@J=2`WMX7lH(X_Bo$rTWLeo`!%$KihS_O-i?m8>?uM9~weItHe%q0oP)W
zt(^Uf8%VZU3plI2g$v+`yuWEZIuv%^2QzU)VK;B(K=7v0qQ)z&pWU+TIh}6amG=D&
z?Xh#N<gG=xk*e$%cUr^4DFUZ+T!rJNsMTCdVegL88}pkLxT|RS{yF8SdzM=cH{Hbn
zoPWk@lSgauK!4myFRyg5N5r9nGh{)PLVmiB%=_wi-dhn={kjcm*H(p<=p3pl_1N=D
z{vr|4U$g7hgVs{%27Ol4HJfRk3y-ZvXt56ivD$Q}XMEgre`>Yqnt1nEUHkORVxP~;
zaGu*UelOmQEAc`V;l|&dmi0?{<q(q=jbg8ZyB;TipDYtJAIoKzpY5Ngx`kiNEAJ3?
zjX#2#axPnlSAyzKm%F8>SA#da72SEPuGfNIH@N!KtxiE>P*8JN>UD51O>;&;O+~qo
z#l2=Xau}>QXDnWVA&PS$$9tTK>W@49P2D2>VfAv}Gc>49jJk{L1UPINW_g+66xrXh
zKZ`ku!`q5{%*d%2u6(hha~ne=EDq`;B~zuJw{lxBXPpEigIe6>4CZrg9gR{@Hk$Fw
z*l0PnpQbJ|)z!zy@v}`V9ydI#eD2gbm2;>gJyuy*vzgWzlm=}nQ8RtK<f~S#!UR>H
zv%rOkioh*U@Qrp|^du$Vsq^x-QOg=v|GBP$DGEx{Rov`UjB$(q0n-#_18owqkD9J9
z*AITSjSW3!S`>VZ+(VsZkI3tt*+k8b{r^lVrE{$G$+RfD26L_AXen01vUw6Da9+V>
zAwHf3^3UZ#>hfWsWom0Y76otUQ>JpT8%DC7a7RGhUfGh^BjrMX*E%o#^*#C6Z8!3Z
zT^7_hO-687F`Ozbx2*998J8ATSY5PGduye=k(-^)k*a;oXItEbX%_JeZ{6N)k;j|Y
z`Ae2!y<B-!*nB=sXJjzJT<3g|gpF2l+uZF;pCVC3`|WR6V3TDep0-AL07f<FHYI7W
zhDy4MBY8EgcP7gx&yl#g-)1o;ZT9RKZC8xf55!}KT)eGYzC}CDBX1tt%zfcqv8TWt
zL$9&**+XMp(f)EgQdQ=4+ue#&x;=vhzek1@$Ia6+6(@agd5k?~&>QXi)o+laJi6%|
zrz5#R8xeBU;f9HvJmgtw-0S*y^Q@W3w^%=9kqRuQUF9_Vk%A)ouH2puip}3_?xL0`
z4$kg$Ua8c=Ez!Z&GJ8&Q*SRYRof>7i0&@(;El`zcJ&(lsC8qf!F6BzRgWpkAP<I5l
z@ssQI+*cnPKjDo0txot1)e63Ws^O@ScyatEs8w#LL*O-hbykYsJ1zHm`}2{T)i&6l
zHST;?z{>Y~E`2xIE5RW?yYv<--A0b$S@o@|Q)f+xjUZp<t+;08aJKW>^saNh<t9vS
zWu_4w7N1*0T!$O_=RP40GHxpD{2A{rz%2{Q1t)LH<vzmOu}so9DLSoYr6S_C1G+5g
zC2p4$ckJDAdSP2OaC&-I&fWqLdro`y>_|Dx<1Ez5%vlT`SWp^R;?#<9J+zp)!9ji6
zNA~YMmU<I{suJz>#UreExE9rd9;>bQwkEA^zksN{fp$#V=_mBb#|&Bm;2VU;dU9Mg
z#emBCu`}?Br<S=byb$%7@ea<Np2y^zkGu$Kg=ppm8cuaOIK0kpHD1Pw;tbZg?VBG}
zZPZKP-l#Xy|H*2V_d{*L&p{33W_VrmBAU$Le!(xw*8A@*214pms7K^Fv{ypv!>ks0
zq%`F9r)%RkZKNJi$j;gw)O?<wxLd~^q&WBi@O#~X*F#>f%gsJ&>592lx&}i+>T`L?
zinCIIH$v)jXC!{i!Ao+&r+72u_2cnWG99XbaWi=iXqaV+UwJLB=eUHo%yIPJuE5*M
zU>u!NF+Aq;pxe~om~w>qq#70Cl+F3xDk`lUDc~61L-|{bvb?zwn^Q0wZF%#yGd->3
zF^Y+*Yn$+nRhl=Xr}<s`Yb%xG<a6;_`W$Z=;s?W9vH3|%u!LkGZZ#%`G&|>{k>B4W
zRlpy|{sEJ%TuYor=M?#(dK6WLFjfC(bI)nAPO~DqUD4^Wobcv7Dyz@5IOaPeY_lxo
z-*NZD+0>nFS>y2>o-^aKnxn+uU;F{)Do&0wKE^zYWHt?EVuAd`@tVDbmh<&ayBUit
z7azSl8%r!DcM6#sr&ww+EO$enW#;H?AUU19<yO<)J~)@B*k76tXY#=HuqzeyCr;~j
zl_mNim)~iAUhm^Vt`u>`_r<&XAl#Ta-aI8zb(>4LmOAwoX&=xk$W2xGL2r~#&Q6_f
zK%xq_gKFZNEDt=B&4Dd17`6W{L(cBd7W<z-v))uHL$V@j_j~3=wpl7X(dCW}ZMPh7
z-RCpx2zlMv%vso}<hHHj?JPnX%7@P~ztD}Qks3>vLwvXDn;uf9$J&xTMou*q_QUxL
zkZHbWwzp7Kd<(KIVUGA{572`0O3n>yw91mJn)S{|<(Xfn4~_iTchtguigzNQz+B(;
zOfD!H4skb7oiG{2%JLUn<Dn)%i3&J-6H1k{xJB=x%))PV_hy${5x4TPvt(3QW$~np
zTbA>i<Gb9l;N{$wX&!<3be%iA$(>MTdXBri@r=0r3?FTE>{CnUXW<5KHHv8-RZ{qb
zYUPzHD(g_E+$8h(A5XGa`kcnQQD}qJA#0d-Hc)lQqIG+9o;ATHMeDW58_;6TZu0I{
zaL}qkx0mxOQ7Z};+c}|aQwFwt!Fhd0>;-X`j%~mVOVh1EJNqj)&8wR+-LeGS9T9Br
zX>Y+AVYlqz%E@16r@8aj$IInI+56~HF+E!>xTDN`vf0@x!d=S%-x4?8yJtClkKa!s
z?%O{XkY)junYKD^ygalgx`4Vvikx~Ad;@~|L$w2L7KIgj@q1L&;Zad5+k5evcSNl`
zOWd7SZm`^^_}zLf9xI*?xu&rN1C|YTwloU)RR6p;n+HW>eaC0*$JX(v-luqO>3p&w
znY|FZPerwUHE-^~ODis?!fsdO$A7OvP2SyQ^i;yNDr6(%vgW?~pDhpbRxeq%48JIi
z4Ih8P*%;G6Se++r-8co{SHhZm&iIs%>D90rc4w8AwgJQHSNU;m@J85!!s`6iZTI#O
zIct7h0o>9O7dHooZ+KHzcbt5P#rw~^n+R`&)r<a)1Bs9MxxH!s-GI$yVyH#DgkPq&
zZi8W#6L+yEHE%7?9Fg;F!`t@PwI+P~hg)py1lQA26Gqs7ZZmNt$(>PUiwfsEFjDN@
zfH*3wcFEh>rA7hQ`bS&LoEo^ru`yN%UVXkE@958Bu7M?BY}iH>N4-ryXUV`_b2=_;
z6Dyj~O5$C7PE~Mx*lRo!w_t)5({)kK@)CPeD<*gC`9@AM_jvZ;yeWeZeR9|vTr8$G
z))eKOHvedyL0M_9*wm4RX>xEkbYZ$>*co2IO#PuQD64puV%i$dyU{GszvtYU`a|}Q
zW>GjnnXAa+mb2fQ7uGP$b{pJ*g1uM}dlc+~*$^#MXgA7X-@V9^=X824mi!u=123wQ
zvsJFgQcI0}3~gU6vnTNNZZ-#(X{NIg%kANPxP?ktT*hAUYkKStI2W~d(pwXBRt#2|
z|4(^rz!`O3V_D$v;bX)rFb5A_pT6GwzWX~_<x8;Sc^1>1rE&`d8|8&_9CzxW7Ky61
zxRFz^DfW`GljDm;y^tiUEM9+$EynNhn!sesGb_#=`P^oC{++w5%*~Rw%l*S7cW{W)
z(^wI?<g)b>?6h363GvnxcBw>s+1U;G%rCdYsf~zK^XpDwIQ6@9RqqVXN-Z+1vX}7U
zisQ&s4*7+xLAGUS9yMLPgN!*A?@Vt`6S*q(^&@MMXBE03{`vvr>(3wm_)mXWfdUH?
zzlDk!T+g+<P<P$CyRlf#od0{%z7i$>g&Q-MDjI5Ev$HC*v|4z7x1~iJ6bYzMcpA*`
zR^9xHeklh{oTVh9()_OG@CH?SHe6Ip=F;%if98s=Wo`~t^f(5oRw6rSPqM~bwtnJm
zSl3!5x#1XlS^eWZhVJvLw<5WnFn4D}g9YV*Z$3E$6lurqwP>=(Kc81awusv+xa84l
zxx8NT`379mA7{vl*K=FkPUFaFJ$5LvdHe`(*wdL2&w;zs)4r*ITtwjf?3N|v9B$h?
z$<wXgayO5QDq3%{>t5h4MZ1B6kUK)QlDKOTvU{d3uF5>$U9-v~EsB0Mzdk&$+;fE{
z2@fs9Tu$5IcHkaa+MIrHE6VlAK?VJMGmn9Xty<ki1XXe6ckO1`EnwY6tpVl2rDtNm
zjJNcKwGYnrL6-xt%x*^iSUKUqmIL|$Jq8*dV4oY0r}k9X5?^K;{!A%zT<vP;d90Cl
z#{;~p2l2wPw$N#c;HCZZ@`fK(N835x`AHZ@1>m1!MdcP5?X|muEpnXYI@E5jjh@ep
zy&@d*8yaBfR#(cMJa$R1b*q=9ff-ua?N)E-sqpeGd)@wjO#2pY*?6#mdgJm<7-A8*
zbh#DU?R6$Rh5e>Q`E@bNJXBHaa3%!ZS&?DgH(XClWt{bROT1v+=Ts)@R+qP!tL7WF
zUQeZZ)9`L}Sv&>L+T7$;_AtWIoV|`Gh7qSU|8_J+O5R1Id7QY8vK%aM7NSPmpU*eY
zyvCSr8#&Nc_PXTua=c>^vQ6VhHnv+skhPwf8;p}9m4V${YrJ{oOp5LO1pDVy?`U$t
z5?Sr`WhQms&=RQMo5Q;8G%khmB?w@$iX2B<k(gq&yS}|Cc@6a&r`rFIIHqx{d(-5X
z9|8^SD*oBL2`m{iEwQyu32c_dxQO$t*_M5d8|qm1b1eH#tA&b1=IBm|a?^62QuF!~
zY6f9JEUb6T5euz&pMJt6d@NFmJjqHegeB&a=MUTvf2rISb6xujEVD9j&zE`u-2Alq
z{$$-3oNsWzv_dZ;uO7h4SY2#hhgDXjrLHSkV-8z*V^qBSFZy61)=SLQ4bD9E<o{+m
zHY%fT)sOX?Xz96LobBW~Y_gYViyh*1twB<^H(E^E!r`_%)XhFpQL}bwTtO-R=C@O@
z&5BK{o9r8v<cyE7LrH!>L#3?OomNx4tb`A7mtx`tKU5d$z9G9Pyj6u!yLOP>L#p{$
zOLLg%R_u9h)|+8}II`b_OsnLXE)mW>6^Qy;W07M;TFzlJ=Lp>DPDZX}f@`&$aOVAo
zXUkj1k#CXB=Q0;{Dhia4yNH{eqyBK)n8_3?pl5;J4LVBftEAE@N3=YJMbo;fhIdAx
z%<{y`#87U}-F1RINrno!@dzWWD*XGKMGH@;RDztLu!^eWk4Ch30#pucUKyxbf8yA1
z&>FqPzQLJ%;X|#p7-=WPnGCG^wXPDKMs2;l1Jo9$0Y>&(4YBX>sC|e3z9kW`dgG2U
zNB>Qh4o7H-?Nv0cAXhHTIi<c<EANucU*MWtIC@ND{kB<VX-PF69jc%)e}x+oyXJy-
zxtO1iH`8#?Ez9wTpLpl#ZkrqT5|Yp<xA&b&Bv*90%x&HV&M)rRpS4^0BHgt=^J!Wf
z_mm;mY0-}TeRJae%_TgrQYD!6kcWDIyj6auCd7p=JhBH(om3kBc5f<zD(<`wX3Rp^
zs`PiCv4rlZUXPNCi>ED>Er`axI!<wEiDe!7tb9Cy$HN48tPd`3E6tX1p2T_VfMswg
z4Fce)`DTB+uKg6AS!}k3#a&N%ZuhTcgYZJx;x^`Xyi_i>yYqsahsuQ(DycSP&#FC(
zXIPARnI(QPcDt77QDe^rg-s$}=}}Mkt~-G7YLC}V#l6P|`kMLUfgd-;8PwzTs)-zh
zye^EJITM2wV9|RExWn0_ZZwJ4xPQVI@(qiCO`@}?^QQf!_Bm@CLwmgbc$-^ZWB+K5
z>$Y>MON)QX6fgbd;>FvR!;jpe_i%Znp(@V=@Q5Bw`@9LS%o@65qj8p%58TQKRb)n5
zxn{BF7-iA@-pwIL%YplyoLtI*V-OCk-?1|O?MRHZz@PAuj#Ki^gtTiY##>xVxz~XS
zmZ;k;bXw^X<v5q`6&Gdoht_~O2%T)DrD6xS=rKiZJg&k=Fje)!vs2t|!!%`M%*g4O
zZVv0*)$5qqqeH;#0Jw$Xb+g)SoTa!Ia6i*k>}*Slx8peDA#-}PVnFpZXK;Y)0&^`|
zjzGATGta!cCAkgTu)rKybT;w}t+X_>eF2MlG;;B_IJ`k2mRLnCUa|;FEq^rV^*`T*
zy-c3i)w7$l0Pk<zgcX+57GCHbkCo=)HjGv&S$D?~k9}GO6M5`{+n@108Z5bCjAwt4
zv94EGH-kk&j|M*Nr>W89lH6zkxbwzWgJIYLCt92fxsF9un@v_ni+FLlv+Bj3Q!(;7
zOrBn`a`P$!-u=WmAa6{w2Tz+0-s;IVWtbyq&g`~ZY`?AtVTVa#DqY+FwbIVu%djg}
zavnqQS;jt~F_+)XL#h>q`v0`j!*4QOez&jY+4>AC7|*!!YjgH3xe2Pb=eK_xhiuEX
zj>}SakXl2kaeg{flSZzk<xT={=PS?3w1~Y4Hw=)kq_!r~a8!YLrS=TZO`@pB8v@T;
zgkpP!vv{WD3zV2Qt~O-xn=G|r^C+I(%w8t%u7P7=%I)oX*9~+kl)R_U$%z8|mVMKb
zq4^C|#$FiCwYH(k>Sil73#fX`Rb5M+8?`*efol17r%CugYwVTsl%Q)Z?vL3DaJS19
zsZJG_#8GX%#q(~$I`=&_SUT?Pu-5_w%-tzNvk_^szq&IHE%w)$4CNAHtI~D0A-3R}
zx$&HiH~&=3vo~znn(R%mJM2xMnPc3r$Kj5%(+nGK+J9QW^j^7J)?EGe54f$8aw*BJ
z<aJstE%A03x-7}9o7r9Q?rp2`BxgbKuDP{S;b{#mD=2l(#PEGF*ZbGufmHz45jpW^
zBo9sC4)$`u(27SE5!X1}FeIp~eYTzk0F_<>2VLw-l{0UtsRdEX%GS-)L+;Zbcb<te
zO>^PsECB-wwDk*)DYxUP70z3C#4}5m>&M(~!E>d{g5!ln!3%JC3RK>he@`{_N0sff
z%~ZGkNkRU0<T(5+T$}I-ezABtxk%xAhlu*mB&z6f+5Hv!OI>4kquQ$xHLzPb0N^Lj
z&+N5`*QYLYhZP2iIaZ`0`p6A$n9TJbUJuCE!`pWq9MLqD`g9!L46)c^Dy_V4JEHD-
zt2=7{rv7jzo2NF0>JJZBuwxpgKc9IDMsG#b2)gxYcN_w5TOxc6oaGFUsNZ%fi0&5j
z5mutrpKW*h-f<DH<B$8tf5b>-KknU`i!n;RxC7@*U17AP!9^Rkq1?h4qcm1?VU?f7
zJC>sB-8tV>lpk@LI?l?&W{cMSEM+cmQ$Gh2EKc@uPB}Xj6D^iCtQ=ZOh}d$myFN3U
zbIvAAR#9A&#%Hx3Q!GMv?uM`GROO!+8M0>;JDWwAZZ+^bS_0xz!_0^`s&H$H?iQ0x
z@!pa@W<|V)ZX;K0W4We!Kjy^p;Z`%W2{1RJm0B(=C$HvNZ+^9_G0%!K|1+*ye9AXq
zfeF}^xO3WEytQcOY^E{FMarPJIkv=-a5cfN%qcD}Rb4E0s+&~4wWyrkAFpv=uFzc#
zd}CLLo%u(s6o2A&@K-5{cN3h(F4kD(@?M7Z-2POAe1Umt!EG9@*E35<N<xCY5g&bq
zjl%24c>%YD=g`7UK&>cS-Qmb2IdL_`efV3H1*+?9MY1?<7L^;e#ojxo_wWg}E9#wp
z`~!BFuitS8Y^Ob|#op?~I_$ENbHSLp^K2q~^30ucPCAu~g&R0lM!Fu~d$W;Y@$twC
zGA+E*QgWx8k*#z;<FZMP<!FJst7`*t?Tg*yR*l>smQ@vxJbk0;nYd(r1^E^wPd!n;
zg70;K())t8)=;Fn@J4N^D3;5db%{wR5mOI18Ksu;2W+<ax6BGXBaw%Osm)w&?*q^M
zc}kn?(khhg4XdeQt5UPk;HqGy#q=)6wH%aJDS2wYUt*tpMm`rOxXE<2(%<Ti0@TRC
zZLV|IQ>nYom|s>a>g-7`bUgs-?H{+2*c>)kPI-OuR(B0lK#}q|_i8j*y>uP-dWuVE
zv7kKP#&LeDO0b!o?lmPJ=f(<XQwf|Y+(a(NcgT(JJ-^}`R=R~uoX`8Fh2v;99=EKL
zc;1ZmC)p)%+f%Q=+?nV!$AxYIV5_&t))moM(y821S7@yp8s3RT=<J){wd}IH<g0tn
z{_zw^5;cY7alvBVV~YndkJk^c`vec|9|tz9kVmSJO>Su|Xu0R0j1`2i)t&C}SdAX#
z_M_Q|S~eDQ>k)kxpJ!wTkFBX;iMe{myPciU!2x?2)_jVm=9|U?lW20@Qe+qDp6qjb
zBA>f^0oVoMg}ve&J*?ZpmtF(=mllc_YI6?_KdK0E?=Dz?pTyo__n(!^>*-sQKgTbz
zI$$^XIS}=_HEyH1;nyosb)lXHH+MOs>Ywzo^D({_)syh1!n}&Z=}Hf>FrRLqaslU{
zijWpnomr#776W@YcK~-khD6n;In(Xj?!X)7feRk<Y5twFQ4_H5O!n@}8ES&qWF5m4
z1y`*%f99&@t*F<J$F1kq&D&A0muIij%wdmHUm9-N;~`Pn)W8T!e_;ZTd(&)QoB}7%
z#tBA7)j2S4&Pb#*1V_a(NZW&+7SHIY*X@pLJBu+2O!a1Vq*jrf?Y~c@A<NS{fAgO(
zPA+lpI=w}VSMa!z)WDS|_flh`GVYAvvr+$)dL@%AiZxuCog8!JsUsdoT(=QZEL*?o
zT2Gbt9Zm&fTGShhun5$Qoo*HIVIq~*5->CB^_A~_fmx~wey^V02F$j6x~sO5FvpVL
zux1nHif6Cmn>jD`IA|A_W+=YE0xQt_Yp~E>55A&&dR)igE3-(kx>K8c!ImgZr&`8W
zg6BRh6}PFoOw4&fE|x1&-uZ+T=4C4<I4i|G7rGj&Ozsu|*BH;Bq3d`RK8`9KtFhiP
zH-8Zllo5UepHL}uqxo9OjTR(Y$z7|Kh)w3y<!6(cWPfOaU<<a`A5W<)+5YfY4UG#~
z46E1hiWO|Pmux=I=egnh4rPy*+|#(wuP>QrR*KvDtfXBQoa?b{41IcHHjzcDLerQG
z7icX}evuy|!=9|$%wr^(=9a(-II^Q!fAQ9ny+gg;x=4=s=_=zEbFTi#VsoBV;ka>s
zgM8yP93m80B3(z=H(WuH6^;8Z-i5ry7U6sD>6WM%H1_MdjZ*WvaMAnoXDs>$lv%8E
zIY2JA0&ly1ocjb7`sc1-VDbItHj_iXFHmVu@T2*tG6&j^Wp{H%(a{hJS7$93Gq`Mn
z8s*q2f4bw|wXp*5&<}^Ib=E?p$OIMNwR($oKF5>Xxm2_p*6yI41Vzj0V}IQgi>H*m
ztq(2s9q>K~H@K`rt0JDo@gkRqIOM^#|0yYVVN07O#T5v?wH@{`IQ3?CDE~&(w1-`U
z+@SOHf80StcOL1c>R|=-bZ%M2vasGo%eR$H+7V@~cm8`esW^=;tC^8}XK}~exC`mb
z!Q8c1kqbfes`PNrB4;1-g)@e9-y)zf`>l8o^+xtHlHX@9rJF+^M!ngRYv8H&;1uti
zcGe?HiCwmvY6TSu@2KWEm#}J%hH2wzK!Kk}k7A9pK{T&sMAUNeDUX@P(+F^%{!_zs
z9fQQ<aqOA7>uxY$sm=X+?!wWapB^E%DrkhrUb=;|KEU(X`$IL4wRmBva^_A2N!PaH
zr95+JxD7wbv(s(hTl13zbtg!;f@`^BalBUT7fWO{tuvxGHVfr&eJft+^?D0_19-Jp
zGejQqPr_^Z!?7Rxyg~95H|GPq-g`ruL2T^?OT2?)njyWKeR8du(~&pqA2%aDPR5(P
z>QWabd_hw(RIXLjw8c%Y!+L+6@OY+3y%P*?S%_qsF`<D(ysh|TaSg*2p1P>~P)Gb)
zoNb%9<QJ#xd1?nEdo`oxz7_2VxS_x(D<L;8)?&1UOyuEGKIAd>cLtZ-sowog>~A7h
zWO-i2YGC9aFwQckyAo+@NJ(zs%m@>dQCkn3sDGSFIfKNLdNo+#NP;^8%*|x0C8vYH
zQ=Up;Cf6pXDn?o<-GphD<(WJ_i0QFEZcAaN0{rH;{{^!YlV@474ztbILU-zSPOs*=
z^EV`aO!KU+q;u^NZsJ=sPa$V<6PZoX0?QF6TikkFX!XZ!26y~~@4=#2m2L2j%q&q3
zoObon-W&gmJD54<lTDots97sXGFAvyW2ImNRtaeKK)}^l0o7#%+^`fRVxxfiv4R9_
z60Aj%;4^Fye2Qd261E9G#&*Ga>=2B@PQe=N67YU8!C0gUK1aG>8!`l+AXD&vAX`97
z?Se7L75p{w1mlq}n1TYq`zR7DMscq<ykQekBGC$z3MQaT@Gi;)BT*rk0lx*SMWsYb
zP$l>S&Ita9YQf*2M(`Wd3bv!p!Us?<(LbTVM3)edXa|~1bOkLE{ViGr@8OzYKH3Be
z(IMD~8w$ow=cYuna7*z2;I`l$bPDF7%Y3KejzlwYSFi~8%q<1?CHjByK=8lgp<omq
z3H}*D!2*N@|0jB^0Ck8;G#h;;YQ|%U{x1v&CgQ1JGM>p3d(r0-eSjB&Ie2LiG~q{y
zR^cbXB>XJ+9exq~-$0-DA$){a`t+dd@v21A@S2HQFi1q5gV*KwUocqkj~F6YiZ{&r
zRlF(DhZriDieZAecuVkmyluWOVz@=VA0v#45NEEJF;W5k4x<GB8KceZ0>+3qPkF~=
zjTkFppE*te{sH67+eJ(;$32*6v=@`i^&BQEz~5tvU?rxSn;+988ja}|a5rX3v=g%g
z%P?E;zhRDGI_6pgEtuEmbzEWJK`fB#Utyuem4QX(j|N0VT@6?&Z_BaF9ILS0s2nRq
zv;n;`7I6VqDa3!m8gtD>yohJk){EMjks#_ij*TWei^Nzghp|b*uhWra9`muq=opfX
zO0X>!q!imNNHKPpb18Nj<zbho{S<sgCy^?mvDb8EdKWSzq6SK4Eas!gHsNmM80|x@
zsH+@#qOT7j-(*KoAfk1nB9k3Ju~7j^%=H9HjgFyAM2*&Rqa&y=*E8^oy2?-~x_%s0
zCOe2TMir<w*A~=>xEWJxGA<<<9YMW_C!!imb{GMp189mBy$mf9b{#>h=<9vBCgR!R
zHgi3J4x?h+Fn`&&X;gq)qV@{hHmX6VQ60KOU59YT$d9|`nuU8J9x1;sYP*aFqML{D
zP;@;Vk4$z1K~dL9ghlQ3=rLI#q9$uZpXmB|JQlUpV!-@W;i>5B!+2)0Q+O_F58#DS
zC0?4p68spewk-T4VOu$V7IpD@T?=v+==Ykr_Dguh=saHSw@<Seuf?z&gXG>;hu1~d
zFJQ2!D+5E!qciJl!b-d;`uYflirR}X%%~Y}nZNUR+ho-k9!swhBP6`O4{_$P2qTSZ
zFv^0YV6^CV3dWeM6z_=InlM&${UFAPI#Mv+C<hbFwFwhNofj}ER+3aqmhk>5OcAyF
zFx8w3^Dxaqmt(r<>wL^K=ky}Xij^fBvtw9+Ifj|}m}{7kk9ijVAuJGe@{qAPr>0_&
zIj3WZQ7)E>u9sk$sN(>Zo9k|@FxdsHjHR86RTktd)`&W<AYRmU80$rCXOLj7jo2vq
zDjkWU@6xd;7GWKdgg;!pf-Q1y&qT6?ZpJoqZoZ1`5?;T89ip#NvD0Lk*k!U5_$>0&
z6r@Vnm5y{#$5~{EIxitp)OHHlqU$Nh5p`uE*IZ8^PjvqR^37i}3e5E?ibUV-MsciE
z7f~W%dkIR-IRj-z*(eu%m4*sYTRr@-z<W_CVP^}f%=seDh^`+)wdgA!YDC{<qE>YO
zJnBT9O{kAmLmC<+Y~P1~=+BvGvLIDxG1r4=HA=%Z5v}dE#iFT3hlJNp;)ba6GH!~#
zJd9iN*qzpb+Y)x2MyGjXvuN=Yl;Vzr?X|cox^n>cL~ZAAU-Z>(JTO@!9>y}khEc-%
zXAl&9dl+F+=SB3G>qSH@J1I5jld$6$9-GH{42b?U2Tx=9zKCZMezynDMc)?Uh3Km+
zycBhw$B&ksqD=f`LfR{ou(u=~zew128hEU>F%_>oR*RT%0Iv#vXvx8Aa{qH428lYW
z@w)l+6=JZ2_Zu<9BHDvD%sJJEHzjPlfT8AGjA5dV{dh}sy%=vB<zu+0?FvR%GzA5S
zlklq?jEp5+f>9E79K`5Yg1a$B!uAZjBf43Pv7#@JV4UdN0*p7R!UWM*8JHN0CIgcs
z>^OwUqWkACMf6<?rkd<DrkTH5Opj$O6*CPp@-WM=@Bn7Z>z4(XW5O&y=1TZ&Ddt&B
zc~~I&<~SCLuIFHp`OCu+1rA=?ho!>q)J80m`~7+>moVZ#g%#$Wb`~q;{`E1eiUsmx
zjWB%dG~x|?=doU1+b$u&gqfMxDB*XhNQ?!l!X{z2FBM60zmtb85(fRPNS5$^HMWUv
zv5Ay(C@l{=B<wtgouaSyV3(YG0_Wh9@Y@4OjfE*jx-guYi40+PS|&0rrW$0&gr|`s
zVb^}-ioQ!jUM%<?<XiM*C@?yMB02X2d?=Q%vm7N71_N~{jm6%KGIKwHa-##NFe-pw
z^es0iMfdAaW$~QchcmG#PoUbOOi4$LyxwX?t*9#-b#e};W})7kThL%((#{}Y?!G)U
znKbo0T1=W*hE_wye@zj5wGVBg>t*PO1**Ue2|EjMQ}pE-+>&!oQxk5Ra|t?QvHQ>^
z4Ek$v$6`uL!(H=G)P#G&$hkb+R|IX<cp$o8gNNqx5L-<Nzsp5X^lb&ga*iH4i5_8h
z${|FBp}<M>nb&eWmN0VR0tU>hF9lEK-n9qMOqjYC&xO&VJ$NDa_5*ln?rB;0(IP#F
zpG4m@;Ae{{6~Bl&4+2lra`K*OQCmJ<HLAgDq6eoj$Yclc`V+OVCFd|$7%i>F5R;}C
z;tgRia2RhY)YnHbRCK!m!(z_o@Rm83;cW@KQ_f<zFj`oP5yDU!8&6^62(PkKsLs<E
zWubgNjF#|Exfmn5Pg`=LwgQZ`Af*_mAYoq`##`tVHlHR<DaXW^^cW^tOqVfPbkBz=
z=A3#JQzdM_ifOShXE5EuoWRVO&&!x4j26{mw(y6>Ld=oZTWOdp>S)D03zT*a3xwg6
zdMp$M&+o$`lkUmK5{0^c0!t<Qp@kz%32&BQxdo~(#0m?PbrLIM(yLe{>`p7g8pG6s
zh!=($Phfpa8bE?DT6`WGg#$<HkthuB-HlDc-hx^r3A<A{`4;x497M7(7}$et%41s<
zwu`>=VTU<4`mj?NJ$eAU<ldeQpXl48NR@Ng*MxLoZ$T9@<o;DIG9`@oPa|9SLrWHN
zgu#}*$QAY%?m?cgJ7q8Og*}0NC{SFTeiTU<Iai5d32)S*M8ark7D^@jCV(;te`v`>
zxv)E}8WnPXa2|dM!@m8fjJX%0N*MH4;f!G*1=R}m<pI=~*ObGkmGIj`sFScKkd1m_
z*q4I_!^~3%2t$qM&}5-bp+(f$idIpZAJ^nD=)a0KVGm!S4wKeb;fAm~^(<~G)b)$F
zWx|wJ+%~V5QqU>)|L-`uOz1Df9mC32+!aP@_v4<tetQh}CG0z%g9pOyl=FBf_gg7=
zWQpb>C~7Z5*gU3YqbG*v5S4r95%ftIN@IH}><MIHz`Uk#BpM6Fk*I}gz;g*BfmFN@
z_U7)zOJO+uD1Nl8G~y=<pOS%}V_1n_gpr&kV8Cl3gY|4w2fXIBEfudy7-_7>YZ89z
z#~{%+XYsn|`c({;$FPs1QH#j$#~bFpA8*Fo4`8T-x6?38)RBR=MBioMZF%hG%P~BL
z`!GTnN;{4?3*^H{34?x4*M+?W$1qwL&ML<kdF{%=J29^Tj5YU+0*sS;=P`^o_rNJk
z5Qfr9Fi{xI%)=yMq^1OuV}4I!iXv!hz*Nz<M=(v!-6^LqUBY`Um>COm9J7Qyfisvb
zjO6EFj)gytxh5>F!aNB(^RXZnp3SNW4`Y#p;gm~Q5_7+RrE>rBIF^}vK?;@&yE&Ly
z5yLZBDX(oUSS9-AB-WTSM~U$g-V0#8==xqH7@fn$SUfyCEexkLV3XXts*w~E)?$k=
zTGWVSVKBh;OnGfPgzXacpXGX{Fx0#UJLTSa6uU%UpMX!!-Ki&#DvXrYB3&3Q<<L}K
z+Y68>VK}`4*|AWS$T3XevZgSyyBc}IP}6bb8>Vq(Q*nLSiXsVvjhQI6P<xtCB8*-v
zM5*Pq7-i;_kGI^s=Iuj8Oj-fIVZ9%f7OE0e5{5Il!YS-d%|W%?zs*ODs4E4vq8rCi
zCyzaWy{MP)Uk;-|^i?JT7G*&Tnk>q+6KFAMW(iv5we2LXN!ZJYLz^(vxCb2;Dy0xN
z<n_Be+>8nL<CZXZau06Hz3VJGEqqHJy3A`@2JQ%Z0!_GU*mxZGgptcfaX;oa8xLY0
z_Tr&=IClt-Vjjv76b4WF5Ee!+T}DsLLlL6Fa8@Szgg>;T;IWmV6$4iA^LQ%fP>T=G
zOqhn}78ASw7ZTp9#Y+jJM=#(<!{S{0WZ|!Hm}=5P4fsVE3>*Uf;<cn-R^gSuc+G2f
z+8(?r{2{O(ugU#ZHU^n{Y874=M)n=TU}4|Ms~92-`s?t9`DJVPW(*Hus4&ua5yK2y
zIMEk|t`^~KC3fcshD#XqU&06p?>8Y%)Si!#7QChbqb!5@7%gFU>LrX3hO@X4WYRPr
z#tMUhLX0yHnH;Yga;7}NJowLGqOiBR8k0;~UyI3>mj+C+P(|sODvTV<#WZ1e+A&O*
z*Dv!hQ`A<8S#s`e&BbhCPjdt22t#S5m@BXCd6*}<cNq&}aqh)JVSic`7Rmj329}8a
zl!v9FZ;P=^9)te#SZ={nQn5nrZLL@-VW{Z@R>j<pV~t_jNyN)*`(><`Fmm!D5`?`c
z%dk-xK3s`Ji?srqB<yKEi6mjLF$-G^D{GM~jIe&Une@~?Y!`N?rC^8R`sNaLn%As-
z*d^@Wa{xZM-%CYmOqh#w^O|!A8Ny)TEHZ^*UoEnQ-6<!KV}7&Ck!zTH40($7Rvq#s
z?7dQh0&}m)LXogLwHn33XyyTw2txrsN`=At^C%PcG*zQq7%uXm!u+Nkfj@?YsI+uj
zQDt5;IbyXk)}va&$o})FiD40H4Rf!ePT13wgL+}OC?5^N-n<M1gnfI?pvnARJ&qQ`
z<`ZZY1_PIHP3eAHi8cvCoC<W9yT1}QgyEC`Zp!QJ^SC8pcS<&H3;QZLXcdO8R-#K7
ztl_BDq|Li=*Q6&?a8GgF$j5!rEnd1N=SWTw9ty*y2l2=P)pPMwIB>KWVd0a#e)I^V
z9GFBE*LVBTXL;p4Frq)_V8CMXH{z+VJM93TS@7C=JQs$VuHuFH4HV#|`Ay|cx-gu=
zajP(Lkxi^gef#iB%x^02)N5Fy=T6`iVYsLquRiq}-uo4JO~PQH41<J`jNN!$?(I1k
zEc#OchFI`|YP=!rxl)QZEl^%2h6+Ph0vIOjPPvM=gyEtVyloz?Hek3(vkzm0@P{j9
zh%*n(dofb!ewU9?5=M@cVYINPDFb7K(JUX{iAncjtg!ER6~+mJfr}U~jAUkDf-szN
z855Plo%5I^YV%>T`P|LHYYeL}RqppsVw!~gdvY*c!p>8eDPho`idn+$wEdWE(m*=q
z2*WAOm>cs@fqBCIv|KC@M)%cVq4_<uAB%*M%Vk(%(#+FXD(v&^#xh~Qj|&5aX;-kq
z{HFV{((o`(a~c-rV2!Xl#gBMlh!f=X!svx#NKnzfEy6}o`%xr{{^c|_$z!lF2T8(6
zCI_)5t+|S1dA*g2ZRUQy5!;0W*?#O0_FTD&offL{Ds~CGxq#{uhB)s|wG8Y>x`g4(
zVq^#h*dk^MgOydt7WM>+kfTsv79v;l&3WX>ImA(3z6Gj3hJu*%0E$d{axaQw9xkCo
z7%eMBsW7rH1!d+T5J0)G?~orA=J)7H_=WvPD^MxyPC0-o<))($XH1xO5Y@uItA|k|
z9604etuR<momTU3{1EDey~i%1!K8fR0h69?LX(yEDq2MME72<FC>K1gNqDOfZKAIV
z(IMwZUM_A3g8^P3ZPMLoxMhI~a9hGBmDT7JKBm_$VYKcd?g)DeIFc3iWnaKOVdx6i
zTFt|SW;_sviw@wSNwa88M~QVFLQvEeKv>Sf^C{>NhSIp&D);+m(P!?LGVoX!E-c4@
zVRi<d3cG!K@yswS9nXcm90<P<MzX8$(y%fWKbnWqWB5sleOZT}&3*r6{9-w%1%B{a
z*@5yxydsR29m1=IM*?`wu=W532_ySY;PoH8X85}U7%XA0zX?Nx!TK7!Asi@7$D0=F
zNGpa4!<lC=Oc>3q#9M|(bMdw?SYL<X!fszSMhGLjvk)ijJ<NqylU_~3C}FV1kI{zy
zQj8Jy)$kTcVd%<bj5X=y^B5QNdkEu&;mnhmApA>VDJBY|r9MnjfxGr%vI(>HVv2;{
zT)|Wce`q;|X~Ia^UQ8E;n%L9|yHlGn%RDqUVRlS<3Uh=#0XDV5fxX<q5<bcCVSzAo
zH3JKUVc&5q680bAh}Jyt%Umk#JyL^Z%46qoESIo5^*B}tBTbc9DeN!2j8(#D@iDA1
zzii^;g`sBd^a~%?`j8;(X)3@*rSKpFi4s0JUW!e^?zB@#687)ehb^&Cxkwg9_8-8u
z7}jCC;<_Ke4he%?SKKM=PT_{Xc`e-wpS*UQL#lbLIfHcJz)@b(DD1gXj!a?ysolsH
zhJE?SF+6+{xx$FQ1bG(iDddapl%PP)q2_}qGM^{TqS)}r5tInS1@$Nu25V|iW*+vl
ztu^VXJ*W_duGGRW3}>dI(y*`^Rl;cXah$QtTtT(DA8SC3u=ng<)EaUURwwLd16yx?
z%YA4NMh;gYApA>VF`9%uO@(MN59f-}s>JR!;+hGs97LOh?R(K7VYqlVZkYS|CfpQ8
zeW!3s*wc6nw}su*#^^NUW=5BJxUvs-gu(j5xT_TYm19~7Bfb{gH{9>X17RpF9}g`Q
zTk%K2{=*drn)K2hgoWXv2K1P;rWR3QFGmo4!oD+Qcx)c(0~invu!nxC-25pY&m@ed
zAHj2BxU>K-OnNvKFNLAzefUurtiOPt6u~$3_*ug4)C>4U*b^uPo_P&!BrgZA2nSBI
z;8kJI)uVXrnb)+o7h;fvq2>d4-LRz=gN6M^^D#u&om!7Kggty@ZwjM<-54tDWjj7h
z7z||MEv0au$I{Ge+HMSg<|+6;JIZ!%g!!#3M4T|hso_Y&=A#%TjMQdev@ptH-x%SO
z{fF_6lIT2%u@Z(0s7Na8=8S#3Fv727f=Rg-F;N(-tihz1G#8T<S4Rt`n7g0rvW5XJ
z%*y@l1x$~*S7WAdpo|Iw!f3%&%r+0~66Of|57lF?@Q3OY%rn17Y3NHBYD~jIVdT_a
zEHV$=7+7Lb_KZtST6h@CgnhY}vD`dl9>NMkE=a8uhSDmqO4yxt9&2LK!-yC5o;rZ_
z!f5eHBpB8;W1}#VmxV-O@Jb~%83wA5q#F5>TVCdMKUWci-D&%<O&Ib8u-z*6G<KNR
z%O|nZ+^<}~E@AJnlkk~`-Di<%SYD5G^H9)?48y~%$P|W)GmtIpt13p0Fw%Mmxx(($
zE65W@xfhc!4Ed^2ARIW9hazFHvJS=Op^<H_FtYn1O05vbQD*K}E}~o*=8{8&usiKA
z{K9DY9#jgS?9D`#Fm$CJXB5}HOQ@DG;x9stFnqWiwZiVy{iqZ67ac^sFc>IDgD_ft
z906e@vjI)!p)mz5Ce5Zos<59+9oI~Hk}9dn!kvTY5Var14LOI?58|ffumQIue3H!p
zuDKWFq0^ABU6(MLvj=yCeOYz5Yk0K)_k{gyQSY1dJator;jFWGXdWu6@JJZC!u#@s
z(Ol}M3J20l(IX6JP!UVmox&Y3VdUx|JXX$c*I_`y{&NL*YNb7aXA(w_a3w+5TeTN2
zgpauw`_e*jee_3R@Jb7QQUupe<7X4rm*5v+g!4}PH?PI*EoF-<j2x)KtHSOSwz!6=
z=@=yJtG$fZh5eO>F<2Nrd>%v0Z%Q%V_-|T@$ibTuMmgLbYF;yUW0+w<G2RkJGLPbI
zVQ=Mr3^&Z$gAu~+)CR<v-@SzxX}F)J8ifPt7cg4s-e7|(VYsjo@0fd0J;n+{X*C!p
z>}@@b@xma7I1_|@7mj11Fp^FE0%4dv+hpOB{ppw@e0;GSQ-yt3_h6c`@J%bGOW0eo
z2Qv+`8Zk>4YC4D6!v4amm?P}ldjWG}9!_GOFj#*A3lweF87wqmWdMtWk=@kkk4gPl
zDvWZkcbTyNupi5X14mo1!u*~}!AfE1SUy$>qqPC75%%(Jix&=XDRsRtvM&t@!cY^n
z421pM8BbImulun{!f5GHBng8x$FN22U!Fj+gpqxP*e2mEAGTYd@)qnc_twkUDI6$I
z$1Y*GxCuUCHx*7&h0$|4NEh~9J%S8j|KU7j3M2lr$QJ%3JAfQvIJ*M5=AkqndBXlZ
zY;a>zE+`121>6=B_Lirh*zjByN(?VHqf{8qY(bf@kE^ZaCe65n3gJLT4*bFhR~#x0
z^SCi4?5o_1Gs0+79;#JO-=?6(yw;SV)^J}5>V!}DeCve+NAl2M(u?&72*cThXcG3F
z;^vsJ+gFNKVQ;}@ToVQ>xpZqDxH;Hin7R))gilh7ant->NyRN;Z~AfER(9Kupi^}F
zB)UX*F5r$lM%i`VHRlt!Ct<ko5bnqDG#;4OqsQ@37_2;xN5asR6a<Bl)*OU|J<Ubv
z5%yj=fT%E7Q-?lbI4cd0g?~9*fB|LUUkmY6!v5l`cqR<_F5tN^nt2&7gyGDycqx2*
zsTw~DgI6l>lQ85vjGxVKQ2@Uvg>R~X=U$5&KC~CF2qOos;8nxg3wTY~-?|TjOqzED
zuN(UJWAJmYZEZ`z5DELLctF71cW2^FVVH|sLxmyk?hK3JS-d3-R?&EoVa+8B7j~zd
z#t1`BE8@&UV>(90{8F7O=C>B34O43{Mj89&1m2M_)RK;|=AP!qIALUW0OJjtN-;s9
zZl`0SgadoaF-h3loQ}!DXh9>U2t!xuFjW}Y#~m_ZPaqf5g?+VYm?;bw9>pwScUnDW
z3q#zGnj`EjV2dm4xmtsH!pOM*76`+KPhg=i+K`Dw!rtmiEK$yH<zT6VPa1q!CJZ$n
z!E#~P*NPQ}MRiyyj9e+iDq*nxBGw284%H&w@Mt>L3%gV5kYLi{6l@fR3;akFMlRQ3
zldzx5KS{!nFCANik1w4=viYrP#5Q5H^bEGgq=&FW7^<fpg0MT~B6bM}$~eju_O~8J
zsxWvt8|lL6{yb!;MmmlnQ^LsZbI3O2$B`o(C@)2>y(~WDN!ZhpjeH5e;`JsH2C3y!
zB;oa16ie8hnuZc#FVzrBg`vg@lnHydCRr{FXP-ocBDl?SR_3*tnzH6z#FJLSUM^vr
z5%x45MYXX12vub*ucuKfVW_MDb;3R>?bHi{SBlUe94M(pKp1H$N0WK*A45wFFQe5w
z<Q>2@Vd&~fv<drtS?CZxzIX{Ygx#rixT)OS%EK)Qdzw>lTNo)%MW@`qEkT!rA*xy5
zF>EQoU16|(H}08~5!^Q^+sg;Sf%I}bG%WVxk+7$6H-bu`trX#yFbh4xpno@_!f;^$
z`h?L#)bTQ@uL=XgfulY=wbES1GYR{t=ki?Gd$|NJgpusycxiZyn`MR<%J7r0fA>E8
zEPQgF4X&_{3lRA4UW*%S<e39uwEiSs6-H`!=0Mn6UVuU7;UYB>%!8jtvJAPPH$)hs
zBI_H%aA67FG!OgtVyLk9Y&nMgcdw!DtioFoMlYn|ZDG)V0>g!m>v;G;80O?IF6QAJ
zMhbhHt1(I#tvZg;!hWv2jxoQ~(0a$DN18F#q?KndPFeWIkMR<AbFX!RFnZ39iNXk%
zgeRFavj~$7Pjea9u)Gmd6<5a@Op~xTHw)8+!9XKs3PXHBW(j+m{Fp6_98AF+VPDpH
z%oPT&q+_12JM}acnBUC9SSai-;<AEOV-c1}_@tUfXoP(gXR%B;aJ&}Fh0*deSYaNx
z#<kL<rD<3t>}}qUHNxP@RKyFrQ}<xKu<uGW5`>ZZTx=9RIhBb-VK}E4n}pHx1xOP1
z`^vFJ*?pLYWC??T%h+aK>yBf)us63BJA^%fH0-oOW?+|uk-`l4gu(ONG80BisY@&D
zFTRKjdn8%Nl(6qqKC*?qXAdJs?%!~TE8$-X>yamn>}Pu`4D$3wfw1p5*KrNGWl*dL
z+Nhdfp)!x4RM>mD3T2i+9m*vP1$b4lu&3oH{1%Gkt`r9SY;c9$R9!eDj0Vo5TG-Rv
zj2dC@VIEEp2HEG=8B$NHUKnA`HJIPi4G6@fN6{o4I8==m!_-=|3PZjEToZ;5afB<3
zoX<dq;eH-;HEi0Co96eh54Q|6PvEw&ujV*9h0!BC)@6@`NBJcTU1`Q$32*JiJqdf*
zDcl!!``F?NBRRYAP#8_$k4GlWZAQ?fbv)Z83<qk^BkbmJjHvm|IE+4FU&C%Z76xn1
zV?fx`oQ<c#a6u-X38N=!@Z9`f&c_S$kbMX*g-<yC`_Y~?uVs=j#O1V~CH&J_{32mL
zhirJEF)qL3SA<Uvmg3bHT17pUvk$LH9ObIPAj8HYy#B&#Fz=;fu!JGsISdhosdey%
z+;6AiO$ocHusT%O!|~KGVWegc-ZE^tgtvu#=g(odA<t@z5cah0MVv6ySdEdw{;YJ2
z5=M6K#b{x;I16Kxg*GnZioT>d01Mx80prXk6}!d@d(#^*LHOik3nnVmpH5+tgpu+K
zm@JHz?!y$hcjRKK3DYiOnlS7u!gOJGS|w%*Bb->z682rWg4x1>5+3{#hSN(hR~WrW
z4FzHU>1r$x_GUL=p&`{%76}JR%dtclt>25K!XBOqSY{qhrDM4;SXqh{=J#?lRtkIb
z*x;J<H22L63(h0nu(%cL4fkdtK^Wo?WTUX34=T~HBpaKA!9X#Rgb_c7xQ1MYPZox=
zsik02Ds5~R1}nL)Anb3>$4=Gsy%g*cwH<{|&Y|X0NVV|$xp5}!qsC>1Fv|H-rupT^
zmu-Ic`H&;*XE&QG9H=Wro-lgy9P)*|HHT0jj5PAhtT0qof?{FcsSK2uU#_i}n%^Vk
zC=))}osM$l{9p4?VWC<N!Y|=%UXvjDCLL9B4pUj?jK$P+8P&pp^Z;svQQuM28lG!L
zov<&Z2=&4z=XoGZ_(Kbi&k7^{dNc`xms-#w3}<GcRoKU7=bA|i($HqeGf^GFXigSx
zm~<a4X9*+z3fwXe7qW0$*wdVgPGOKr6<xyqielVRf&ZL=yAp<4sB$ZeTqwkSVQ*;)
z9teY6)_y4LJCKD(!oL&;5Ht@}wFnD)>&ww2>@TW7RQULE0DZzAT8`kcFxbG8VdnSv
zB|H^Ic3;3VL+aE#7e@J(zYqqwNA=RA**^Rz>}kxyPbSShg`b7Jr<(DLFv8sk{130u
z4N+C{6=8R323|ExDZp#O;6*Mh2z%Ley>8M=tr#o}7oNZnVKg&<H%z*x0B;H-d#+%p
zFw}Gq!%SM+ink0oN_kt@L#?mj!XK`-V1%%b)0jA6u%Q_vh0%*y7-iB+S1{WAp0CFk
z^RTZ1?+7DTcVn!ux88?w!r+x+jE`XxCI~~+N}OoYOU;-hjMN;(WMQ=S5T*$G4^cBr
z*wajtEavxeC8nF-mI}-i_U%21S?1xwUd)c64|B|K)<Mh__8xA+JYoN#8Z0mmJkGRG
z7}<RZiwvo0u|ya=(~714!|N5l*@tDK@2j!goGZC~W`P<Tu+pU5m02YWUAcrc!ag>N
z@rvLUch5xMa`Vi5Qh9%)Fr3bFx57wEAvOtnn~xz$7-eC$D1zIkkt|{GWG1$m*Mdvf
zF8oVzI(7)drMt0H*muc~UBYM)&)y2VQ>d>X>^;R34#Hq%12TktO;o}a_E#1mTiDag
z-7{f$_d(<egH-|KnO`5Z!-S!<A`}QCHN_|rMsv8BYf_e?#IW@^N`-x#N|gy8U#dm9
zdEm=mA?#&4<ToiNwUvf7+&&XV>zi>#7`(!BxWaB8W2q7Lag(@K7%Dr4I$`hGM%0^!
zeOYKQ>7_gbgu%=E(Iku(=b*)~su8W`q2?m437=Hgqs^q1yeC}Po!WvM_AoPXQ}oSc
z+>&!-?@`<q_BC8Yr!eF@f-d1eX*%vGRL6PTHTRYP?pY{mG~E~WAHIkO!X6%mc_@sW
zr|m)WaD<Au!k|9`Ve^o&4?V(x!c0WXFCTlK@JVeU9-D`f(-;tjORwUoFv^|7XNGkZ
zcrNU1-HjKP^CG+y-O9y}a*kXr$4|ok0<Pu?pX{x|FNUY<ftTLL*>M@Ky!6`N=y9$s
z2>Ur5cum-M{v-woLyhP0x-e3A8iR#Tcm`sKd1!9L8^WFdRTj)|<{k_+Y&wKt!f5FQ
zyrnFBdjfAu*qu^>;pVk5fDz`EXHw(j^-C%&m{%&CjuM8sx;<Lh!^7uegpobXct;p&
zp=|?UU)4E`6Nah#Io>e!5GDwF8xCTk@G&3GBw@%`h{?j>Nk672H(%#rs=2o|V45&o
z#<8w2;#9<h13W)EOW4hWSF;ttby|Rtu&3z&=31x&*_dbO&%y#>-}yo;6!w?UB$n`T
zZ7G&0u5X&K)Iw$YvCJ?#3(MvHXAX3Y%CS<;y)A5Sg+V^BHNsFc$GO73R*rKOs-px6
zqK8MZ(R@;=C()2=1)GGC{TGoWd~(o_Eedu0D3V40vLD;z9K6624~AT<*kM?60Xu~e
zKaV^Jd!2Teu<!H*q?(7LsYo}hEkK5OIDQ0~O6;3TWJ@?uTZJ5954Wvyh0*fM$P@Ng
zw;*2_^tYfu_!p{w6bU0mr6?9gi_W7&*w5iksqo3E9F!^D&N7rs7^3-s3Sn=nAAVuq
z)dQ#$MvfPtN};~W#u*8NHC3n<cGEscjj;E44QdVRD^Mp4^9!pN_8%=ngD_fp1OfAK
z^gNn`kpSCTVW{~cT7|s@b+{(%Ysf~Mu>ayobQm@@<A&jp-MA@??&VQkVW`Z9+a_%~
zgHFRsd(kEAJAV{+OnRXlcTKv#6!(PvCwN#__?L8^dk_YzuHvCE8aR$e=GS)&K~>p<
zR)o!KT`hXd{V-cwVI+m-56lCnBaaOmN-!V{U1`Eo^P6!5&kUQWkt^(Ns=x~+)<Lsc
z684=f!jBSO58x+JXC{7@bEuj6VHVT5qrks;O>BSpIlN+c;s{>-H?Il)a|vFPFifSB
zK@xtq7q6R;O9_L8Jv`Dl#KLo2{D!=KeHL#@7`a@Jp~7f=8HNe_>gw^9g*x7dw}rv_
zN(@&7U-Dp?=r&tg(VgQMDUZE8mNd#@x|)a4!U6WKV=SiD3wTEu;_-&D!U(mO#tFl0
zGRF&h8Y?kD7%k(*nXvc5MNAU*T`j|8^Kg)>3MM_V8&idkFScNsusbCW(}j`!`!LhW
zkb_wg_B7XEwz(fD#vEa^_#Ea6gY^xVXC8LfV}W5!1r{1+mSd6mZ3tkA;bk9|3PZGP
zuuSQ8?8S0(ufL2H=I$%QO2bQ8SS5^H*o`&9zwj(x{J&{^>FB|uSTAwV&+)CWcYiB3
zDv3YuL862M6|{#SjAWlik|7m^wpb`G$R`U!We2g%JhaweyYPpW1K43XFUL*^yZ`?h
z?(ZdGD2fC4_b-q2!;p|Hah6&VB4H!bEG(@wdXS2Ukcfx~hufTPLPS)Lec^k3`*1n;
zVQg%7_ndopCS*Yj*%ZYRBbFjAiE(8UD`r*(s~YtAYn>DWE_2q*tOk3_h%OILuA6|f
zV{91m8Zw5tJhEb3%pzw->uGEovUt28&I&2)h+(RQk|8rAd*YCfepz%cmQfK$OeOD&
zUZ;qv=%-U~47o{mC`L{dM~3{WG{hlyh@0ZXYT{Ikci9+W9>9GC(PNK=ju_YpT#8{e
ziLN+d>Fl*Q<)ZjTj7#jzCHl<6_r)OD!>t%O3EYd`c?A!m%N+HiiRB@|Cvj#UU|>dT
zWju>x+rf+O;9VWB8uWQA_|1e`4&KF(@0brU-l}0}0#3NEAV$pkd!l<~ArME48H97T
oF%`pZ8?j-JCAs2+H|SH0^EST3Alt&X=yUD%BaT_D{`cGe2dxopkpKVy

diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index 689fb6f2afe5a..9831b13d4b039 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -1,4 +1,19 @@
-function(llama_add_test source)
+function(llama_build_executable source)
+    get_filename_component(TEST_TARGET ${source} NAME_WE)
+    add_executable(${TEST_TARGET} ${source})
+    install(TARGETS ${TEST_TARGET} RUNTIME)
+    target_link_libraries(${TEST_TARGET} PRIVATE llama)
+endfunction()
+
+function(llama_test_executable name source)
+    get_filename_component(TEST_TARGET ${source} NAME_WE)
+    # add_executable(${TEST_TARGET} ${source})
+    # install(TARGETS ${TEST_TARGET} RUNTIME)
+    # target_link_libraries(${TEST_TARGET} PRIVATE llama)
+    add_test(NAME ${name} COMMAND $<TARGET_FILE:${TEST_TARGET}> ${ARGN})
+endfunction()
+
+function(llama_build_and_test_executable source)
     get_filename_component(TEST_TARGET ${source} NAME_WE)
     add_executable(${TEST_TARGET} ${source})
     install(TARGETS ${TEST_TARGET} RUNTIME)
@@ -6,11 +21,15 @@ function(llama_add_test source)
     add_test(NAME ${TEST_TARGET} COMMAND $<TARGET_FILE:${TEST_TARGET}> ${ARGN})
 endfunction()
 
-# llama_add_test(test-double-float.cpp) # SLOW
-llama_add_test(test-quantize-fns.cpp)
-llama_add_test(test-quantize-perf.cpp)
-llama_add_test(test-sampling.cpp)
-llama_add_test(test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab.bin)
-llama_add_test(test-grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp)
-llama_add_test(test-grad0.cpp) # SLOW
-# llama_add_test(test-opt.cpp) # SLOW
+# llama_build_and_test_executable(test-double-float.cpp) # SLOW
+llama_build_and_test_executable(test-quantize-fns.cpp)
+llama_build_and_test_executable(test-quantize-perf.cpp)
+llama_build_and_test_executable(test-sampling.cpp)
+llama_build_executable(test-tokenizer-0.cpp)
+llama_test_executable(test-tokenizer-0.llama test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.bin)
+llama_build_executable(test-tokenizer-1.cpp)
+llama_test_executable(test-tokenizer-1.llama test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.bin)
+llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.bin)
+llama_build_and_test_executable(test-grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp)
+llama_build_and_test_executable(test-grad0.cpp) # SLOW
+# llama_build_and_test_executable(test-opt.cpp) # SLOW
diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp
index 87fde16453d25..a523c320c2aeb 100644
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -1,3 +1,4 @@
+#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include <cstdio>
@@ -5,16 +6,40 @@
 #include <map>
 #include <vector>
 
+static std::string unescape_whitespace(llama_context* ctx, const std::vector<llama_token>& tokens) {
+    std::string result;
+    for (int i = 0; i < tokens.size(); ++i) {
+        result += llama_token_to_str(ctx, tokens[i]);
+    }
+    return result;
+}
+
 static const std::map<std::string, std::vector<llama_token>> & k_tests()
 {
     static std::map<std::string, std::vector<llama_token>> _k_tests = {
-        { "Hello World",        { 1,  10994,   2787, }, },
-        { " Hello World",       { 1,  15043,   2787, }, },
-        { " Hello World!",      { 1,  15043,   2787,  29991, }, },
-        { " this is 🦙.cpp",    { 1,    445,    338,  29871,    243,    162,    169,    156,  29889,   8223, }, },
-        { "w048 7tuijk dsdfhu", { 1,  29893,  29900,  29946,  29947,  29871,  29955,   9161,  13535,  18031,   2176,   6905, }, },
-        { "нещо на Български",  { 1,    821,   4851,    665,   1386,  29713,   1305, }, },
-    };
+        { " ",                      {1,    259, }, },
+        { "\t",                     { 1,    29871,   12, }, },
+        { "\n",                     { 1,    29871,   13, }, },
+        { "\t\n",                   { 1,    29871,   12,     13, }, },
+        { "Hello world",            { 1,  15043,   3186, }, },
+        { " Hello world",           { 1,  29871,  15043,   3186, }, },
+        { "Hello World",            { 1,  15043,   2787, }, },
+        { " Hello World",           { 1,  29871,  15043,   2787, }, },
+        { " Hello World!",          { 1,  29871,  15043,   2787,  29991, }, },
+        { " this is 🦙.cpp",        { 1,  29871,    445,    338,  29871,    243,    162,    169,    156,  29889,   8223, }, },
+        { "w048 7tuijk dsdfhu",     { 1,    281,  29900,  29946,  29947,  29871,  29955,   9161,  13535,  18031,   2176,   6905, }, },
+        { "нещо на Български",      { 1,   1538,   4851,    665,   1386,  29713,   1305, }, },
+        { "កាន់តែពិសេសអាចខលចេញ",   { 1,  29871,  31849,  31324,  31934,    228,    162,    142,    228,    161,    
+                                        146,    228,    162,    133,    228,    161,    153,    228,    161,    186,  
+                                        31708,    228,    162,    132,  31708,    228,    161,    165,  31324,    228,    
+                                        161,    136,    228,    161,    132,    228,    161,    158,    228,    161,    
+                                        136,    228,    162,    132,    228,    161,    140, }, },
+        { "🚀 (normal) 😶‍🌫️ (multiple emojis concatenated) ✅ (only emoji that has its own token)",
+            { 1,  29871,    243,    162,    157,    131,    313,   8945,  29897,  29871,    
+              243,    162,    155,    185,  30722,    243,    162,    143,    174,  30598,    
+              313,  20787,    953,   3848,    275,  16125,    630,  29897,  29871,  31681,    
+              313,   6194,    953,  29877,   2397,    393,    756,    967,   1914,   5993,  29897, }, },
+     };
     return _k_tests;
 };
 
@@ -65,9 +90,9 @@ int main(int argc, char **argv) {
     }
 
     for (const auto & test_kv : k_tests()) {
-        std::vector<llama_token> res(test_kv.first.size());
-        const int n = llama_tokenize(ctx, test_kv.first.c_str(), res.data(), int(res.size()), true);
-        res.resize(n);
+        std::vector<llama_token> res = llama_tokenize(ctx, test_kv.first.c_str(), true);
+        fprintf(stderr, "%s : '%s' tokenized to '%s'\n", 
+            __func__, test_kv.first.c_str(), unescape_whitespace(ctx, res).c_str());
 
         bool correct = res.size() == test_kv.second.size();
 
diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
new file mode 100644
index 0000000000000..122e5168424ae
--- /dev/null
+++ b/tests/test-tokenizer-1.cpp
@@ -0,0 +1,122 @@
+#define LLAMA_API_CPP // TODO: eliminate me
+#include "llama.h"
+
+#include <cassert>
+#include <cstdio>
+#include <cstring>
+#include <string>
+#include <codecvt>
+#include <map>
+#include <vector>
+
+static std::string vocab_type(llama_context* ctx) {
+    return llama_n_vocab(ctx) == 32000 ? "spm": "bpe";
+}
+
+static std::string escape_whitespace(const std::string& text) {
+    std::string result;
+    bool escaping = false;
+    result += "\xe2\x96\x81";
+    for (size_t offs = 0; offs < text.length(); ++offs) {
+        if (text[offs] == ' ') {
+            if (!escaping) {
+                result += "\xe2\x96\x81";
+                escaping = true;
+            }
+        }
+        else {
+            escaping = false;
+            result += text[offs];
+        }
+    }
+    return result;
+}
+
+static std::string unescape_whitespace(llama_context* ctx, const std::vector<llama_token>& tokens) {
+    std::string result;
+    for (int i = 0; i < tokens.size(); ++i) {
+        result += llama_token_to_str(ctx, tokens[i]);
+    }
+    return result;
+}
+
+int main(int argc, char **argv) {
+    if (argc < 2) {
+        fprintf(stderr, "Usage: %s <vocab-file>\n", argv[0]);
+        return 1;
+    }
+
+    const std::string fname = argv[1];
+
+    fprintf(stderr, "%s : reading vocab from: '%s'\n", __func__, fname.c_str());
+
+    llama_model * model;
+    llama_context * ctx;
+
+    llama_backend_init(false);
+
+    // load the vocab
+    {
+        auto lparams = llama_context_default_params();
+
+        lparams.vocab_only = true;
+
+        model = llama_load_model_from_file(fname.c_str(), lparams);
+
+        if (model == NULL) {
+            fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
+            return 1;
+        }
+
+        ctx = llama_new_context_with_model(model, lparams);
+
+        if (ctx == NULL) {
+            fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
+            llama_free_model(model);
+            return 1;
+        }
+    }
+
+    const int n_vocab = llama_n_vocab(ctx);
+
+    for (int i = 0; i < n_vocab; ++i) {
+        std::string forward = llama_token_to_str_bpe(ctx, i);
+        std::vector<llama_token> tokens = llama_tokenize_bpe(ctx, forward, false);
+        if (tokens.size() == 1) {
+            if (i != tokens[0]) {
+                std::string backward = llama_token_to_str(ctx, tokens[0]);
+                fprintf(stderr, "%s : error: token %d is string %s but bpe returns token %d %s\n", 
+                    __func__, i, llama_token_to_str(ctx, i).c_str(), tokens[0], backward.c_str());
+                return 2;
+            }
+        } else {
+            if ((vocab_type(ctx) == "spm" && i <= 258) || 
+                (vocab_type(ctx) == "bpe" && (i == 0 || i >= 100000))) {
+                fprintf(stderr, "%s : info: token %d is string %s and bpe returns tokens %s\n", 
+                    __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
+            } else {
+                fprintf(stderr, "%s : error: token %d is string %s but bpe returns tokens %s\n", 
+                    __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
+                return 2;
+            }
+        }
+    }
+
+    std::wstring_convert<typename std::codecvt_utf8<wchar_t>, wchar_t> converter;
+    for (wchar_t ch = 0x0000; ch < 0xffff; ++ch) {
+        std::wstring wstr(1, ch);
+        std::string str = converter.to_bytes(wstr);
+        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
+        if (tokens.size() == 1) {
+            fprintf(stderr, "%s : info: %s tokenized to %d \n", 
+                __func__, str.c_str(), tokens[0]);
+        }
+    }
+
+    llama_free_model(model);
+    llama_free(ctx);
+
+    llama_backend_free();
+
+    return 0;
+}

From afc4ca288989afea3903c962e1006695d50db8b2 Mon Sep 17 00:00:00 2001
From: goerch <jhr.walter@t-online.de>
Date: Mon, 14 Aug 2023 19:20:04 +0200
Subject: [PATCH 147/242] convert : update convert-new.py with tokenizer fixes
 (#2614)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)
---
 convert-new.py         | 104 ++++++++++++++++++++++++-----------------
 examples/main/main.cpp |   2 +-
 2 files changed, 61 insertions(+), 45 deletions(-)

diff --git a/convert-new.py b/convert-new.py
index bf85cb55177fa..75b07ebddd609 100755
--- a/convert-new.py
+++ b/convert-new.py
@@ -205,25 +205,58 @@ def load(model_plus: 'ModelPlus') -> 'Params':
         return params
 
 
-class SentencePieceVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vocabtype: Optional[str]) -> None:
-        self.vocabtype = vocabtype
-        if self.vocabtype == "bpe":
-            self.sentencepiece_tokenizer = json.loads(open(str(fname_tokenizer)).read())
-        else:
-            self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
-
+class BpeVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
         added_tokens: Dict[str, int]
         if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens))
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
         else:
             added_tokens = {}
+        vocab_size: int = len(self.bpe_tokenizer)
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
 
-        if self.vocabtype == "bpe":
-            vocab_size: int = len(self.sentencepiece_tokenizer)
-        else:
-            vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
+    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.bpe_tokenizer
+        from transformers.models.gpt2 import tokenization_gpt2
+        byte_encoder = tokenization_gpt2.bytes_to_unicode()
+        byte_decoder = {v: k for k, v in byte_encoder.items()}
+        for i, item in enumerate(tokenizer):
+            text: bytes = item.encode("utf-8")
+            score: float = -i
+            yield text, score
 
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.bpe_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+
+class SentencePieceVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
+        else:
+            added_tokens = {}
+        vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
         expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
         actual_ids   = sorted(added_tokens.values())
         if expected_ids != actual_ids:
@@ -238,32 +271,11 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vo
 
     def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
         tokenizer = self.sentencepiece_tokenizer
-        if self.vocabtype == "bpe":
-            from transformers.models.gpt2 import tokenization_gpt2
-            byte_encoder = tokenization_gpt2.bytes_to_unicode()
-            byte_decoder = {v: k for k, v in byte_encoder.items()}
-            for i, item in enumerate(tokenizer):
-                text: bytes
-                text = b''.join([x.to_bytes(1, byteorder='big') for x in [byte_decoder[y] for y in item]])
-                score: float = -i
-                yield text, score
-        else:
-            for i in range(tokenizer.vocab_size()):
-                text: bytes
-                if tokenizer.is_unknown(i):
-                    text = " \u2047 ".encode("utf-8")
-                elif tokenizer.is_control(i):
-                    text = b""
-                elif tokenizer.is_byte(i):
-                    piece = tokenizer.id_to_piece(i)
-                    if len(piece) != 6:
-                        raise Exception(f"Invalid token: {piece}")
-                    byte_value = int(piece[3:-1], 16)
-                    text = struct.pack("B", byte_value)
-                else:
-                    text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
-                score: float = tokenizer.get_score(i)
-                yield text, score
+        for i in range(tokenizer.vocab_size()):
+            piece = tokenizer.id_to_piece(i)
+            text: bytes = piece.encode("utf-8")
+            score: float = tokenizer.get_score(i)
+            yield text, score
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
@@ -278,7 +290,7 @@ def __repr__(self) -> str:
         return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
 
 
-Vocab = Union[SentencePieceVocab]
+Vocab = Union[BpeVocab, SentencePieceVocab]
 
 
 def permute(weights: NDArray, n_head: int) -> NDArray:
@@ -679,7 +691,7 @@ def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], conc
 
 def check_vocab_size(params: Params, vocab: Vocab) -> None:
     if params.n_vocab != vocab.vocab_size:
-        assert isinstance(vocab, SentencePieceVocab)
+        assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
         if params.n_vocab == vocab.vocab_size_base:
             print("Ignoring added_tokens.json since model matches vocab size without it.")
             vocab.added_tokens_list = []
@@ -853,7 +865,7 @@ def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
     return {name: model[name] for name in TENSORS_LIST if name in model}
 
 
-def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
+def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
     print(f"vocabtype: {vocabtype}")
     # Be extra-friendly and accept either a file or a directory.  Also, if it's
     # a directory, it might be the model directory, and tokenizer.model might
@@ -875,8 +887,12 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
                 "if it's in another directory, pass the directory as --vocab-dir")
     added_tokens_path = path.parent / "added_tokens.json"
     print(f"Loading vocab file {path}")
-    return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None,
-                              vocabtype)
+    if vocabtype == "bpe":
+        return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    elif vocabtype == "spm":
+        return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    else:
+        raise ValueError(f"Unsupported vocabulary type {vocabtype}")
 
 
 def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index 2424435607b0f..9c15febb51f4b 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -307,7 +307,7 @@ int main(int argc, char ** argv) {
         auto console_ctrl_handler = +[](DWORD ctrl_type) -> BOOL {
             return (ctrl_type == CTRL_C_EVENT) ? (sigint_handler(SIGINT), true) : false;
         };
-        SetConsoleCtrlHandler(static_cast<PHANDLER_ROUTINE>(console_ctrl_handler), true);
+        SetConsoleCtrlHandler(reinterpret_cast<PHANDLER_ROUTINE>(console_ctrl_handler), true);
 #endif
 
         fprintf(stderr, "%s: interactive mode on.\n", __func__);

From 7494c7842812f58c65b68f82cb3aafacc645dc64 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 21:33:33 +0300
Subject: [PATCH 148/242] llama : sync gguf-llama with llama (#2613)

* llama : sync gguf-llama with llama

* tests : fix build + warnings (test-tokenizer-1 still fails)

* tests : fix wstring_convert

* convert : fix layer names

* llama : sync gguf-llama.cpp

* convert : update HF converter to new tokenizer voodoo magics
---
 convert-llama-h5-to-gguf.py         |  20 +-
 examples/gguf/gguf-llama-simple.cpp | 252 ++++++++++----------
 gguf-llama.cpp                      | 348 ++++++++++++++++++++++++----
 gguf-llama.h                        |  60 ++++-
 gguf_namemap.py                     |  86 +++----
 llama.cpp                           |  59 +++--
 tests/test-tokenizer-0.cpp          |  30 +--
 tests/test-tokenizer-1.cpp          |  26 ++-
 8 files changed, 589 insertions(+), 292 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 0bce659e697df..9d91b433bf29e 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -95,7 +95,7 @@ def count_model_parts(dir_model: str) -> int:
 
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_name(last_dir)
-gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
+gguf_writer.add_file_type("All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
@@ -122,19 +122,11 @@ def count_model_parts(dir_model: str) -> int:
 
     for i in range(tokenizer.vocab_size()):
         text: bytes
-        if tokenizer.is_unknown(i):
-            text = " \u2047 ".encode("utf-8")
-        elif tokenizer.is_control(i):
-            text = b""
-        if tokenizer.is_byte(i):
-            piece = tokenizer.id_to_piece(i)
-            if len(piece) != 6:
-                raise Exception(f"Invalid token: {piece}")
-            byte_value = int(piece[3:-1], 16)
-            text = struct.pack("B", byte_value)
-        else:
-            text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
-        score: float = tokenizer.get_score(i)
+        score: float
+
+        piece = tokenizer.id_to_piece(i)
+        text  = piece.encode("utf-8")
+        score = tokenizer.get_score(i)
 
         tokens.append(text)
         scores.append(score)
diff --git a/examples/gguf/gguf-llama-simple.cpp b/examples/gguf/gguf-llama-simple.cpp
index fa8fc6fef9f41..0679240d31317 100644
--- a/examples/gguf/gguf-llama-simple.cpp
+++ b/examples/gguf/gguf-llama-simple.cpp
@@ -1,126 +1,126 @@
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#endif
-
-#include "common.h"
-#include "gguf-llama.h"
-#include "build-info.h"
-
-#include <cmath>
-#include <cstdio>
-#include <string>
-#include <vector>
-
-int main(int argc, char ** argv) {
-    gpt_params params;
-
-    if (argc == 1 || argv[1][0] == '-') {
-        printf("usage: %s MODEL_PATH [PROMPT]\n" , argv[0]);
-        return 1 ;
-    }
-
-    if (argc >= 2) {
-        params.model = argv[1];
-    }
-
-    if (argc >= 3) {
-        params.prompt = argv[2];
-    }
-
-    if (params.prompt.empty()) {
-        params.prompt = "Hello my name is";
-    }
-
-    // init LLM
-
-    llama_backend_init(params.numa);
-
-    llama_context_params ctx_params = llama_context_default_params();
-
-    llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
-
-    if (model == NULL) {
-        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
-        return 1;
-    }
-
-    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
-
-    // tokenize the prompt
-
-    std::vector<llama_token> tokens_list;
-    tokens_list = ::llama_tokenize(ctx, params.prompt, true);
-
-    const int max_context_size     = llama_n_ctx(ctx);
-    const int max_tokens_list_size = max_context_size - 4;
-
-    if ((int)tokens_list.size() > max_tokens_list_size) {
-        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
-        return 1;
-    }
-
-    fprintf(stderr, "\n\n");
-
-    for (auto id : tokens_list) {
-        fprintf(stderr, "%s", llama_token_to_str(ctx, id));
-    }
-
-    fflush(stderr);
-
-    // main loop
-
-    // The LLM keeps a contextual cache memory of previous token evaluation.
-    // Usually, once this cache is full, it is required to recompute a compressed context based on previous
-    // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
-    // example, we will just stop the loop once this cache is full or once an end of stream is detected.
-
-    while (llama_get_kv_cache_token_count(ctx) < max_context_size) {
-        // evaluate the transformer
-
-        if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
-            fprintf(stderr, "%s : failed to eval\n", __func__);
-            return 1;
-        }
-
-        tokens_list.clear();
-
-        // sample the next token
-
-        llama_token new_token_id = 0;
-
-        auto logits  = llama_get_logits(ctx);
-        auto n_vocab = llama_n_vocab(ctx);
-
-        std::vector<llama_token_data> candidates;
-        candidates.reserve(n_vocab);
-
-        for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
-            candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
-        }
-
-        llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
-
-        new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
-
-        // is it an end of stream ?
-        if (new_token_id == llama_token_eos()) {
-            fprintf(stderr, " [end of text]\n");
-            break;
-        }
-
-        // print the new token :
-        printf("%s", llama_token_to_str(ctx, new_token_id));
-        fflush(stdout);
-
-        // push this new token for next evaluation
-        tokens_list.push_back(new_token_id);
-
-    }
-
-    llama_free(ctx);
-    llama_free_model(model);
-
-    llama_backend_free();
-
-    return 0;
-}
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+
+#include "common.h"
+#include "gguf-llama.h"
+#include "build-info.h"
+
+#include <cmath>
+#include <cstdio>
+#include <string>
+#include <vector>
+
+int main(int argc, char ** argv) {
+    gpt_params params;
+
+    if (argc == 1 || argv[1][0] == '-') {
+        printf("usage: %s MODEL_PATH [PROMPT]\n" , argv[0]);
+        return 1 ;
+    }
+
+    if (argc >= 2) {
+        params.model = argv[1];
+    }
+
+    if (argc >= 3) {
+        params.prompt = argv[2];
+    }
+
+    if (params.prompt.empty()) {
+        params.prompt = "Hello my name is";
+    }
+
+    // init LLM
+
+    llama_backend_init(params.numa);
+
+    llama_context_params ctx_params = llama_context_default_params();
+
+    llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
+
+    if (model == NULL) {
+        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
+        return 1;
+    }
+
+    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
+
+    // tokenize the prompt
+
+    std::vector<llama_token> tokens_list;
+    tokens_list = ::llama_tokenize(ctx, params.prompt, true);
+
+    const int max_context_size     = llama_n_ctx(ctx);
+    const int max_tokens_list_size = max_context_size - 4;
+
+    if ((int) tokens_list.size() > max_tokens_list_size) {
+        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
+        return 1;
+    }
+
+    fprintf(stderr, "\n\n");
+
+    for (auto id : tokens_list) {
+        fprintf(stderr, "%s", llama_token_to_str(ctx, id).c_str());
+    }
+
+    fflush(stderr);
+
+    // main loop
+
+    // The LLM keeps a contextual cache memory of previous token evaluation.
+    // Usually, once this cache is full, it is required to recompute a compressed context based on previous
+    // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
+    // example, we will just stop the loop once this cache is full or once an end of stream is detected.
+
+    while (llama_get_kv_cache_token_count(ctx) < max_context_size) {
+        // evaluate the transformer
+
+        if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
+            fprintf(stderr, "%s : failed to eval\n", __func__);
+            return 1;
+        }
+
+        tokens_list.clear();
+
+        // sample the next token
+
+        llama_token new_token_id = 0;
+
+        auto logits  = llama_get_logits(ctx);
+        auto n_vocab = llama_n_vocab(ctx);
+
+        std::vector<llama_token_data> candidates;
+        candidates.reserve(n_vocab);
+
+        for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
+            candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
+        }
+
+        llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
+
+        new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
+
+        // is it an end of stream ?
+        if (new_token_id == llama_token_eos()) {
+            fprintf(stderr, " [end of text]\n");
+            break;
+        }
+
+        // print the new token :
+        printf("%s", llama_token_to_str(ctx, new_token_id).c_str());
+        fflush(stdout);
+
+        // push this new token for next evaluation
+        tokens_list.push_back(new_token_id);
+
+    }
+
+    llama_free(ctx);
+    llama_free_model(model);
+
+    llama_backend_free();
+
+    return 0;
+}
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 0f8eb3c902b1f..8275c7357ec4c 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -7,6 +7,7 @@
 #endif
 
 #include "gguf-util.h"
+#define LLAMA_API_CPP // TODO: eliminate me
 #include "gguf-llama.h"
 
 #include "ggml.h"
@@ -76,7 +77,7 @@ static std::string to_string(const T & val) {
 #define LLAMA_MAX_SCRATCH_BUFFERS 16
 #endif
 
-typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+#define UNUSED GGML_UNUSED
 
 #ifdef GGML_USE_CUBLAS
 #define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
@@ -125,6 +126,8 @@ struct llama_buffer {
     }
 };
 
+typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+
 void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
     (void) tensor;
 }
@@ -623,7 +626,7 @@ struct gguf_file_loader {
         hparams.n_embd         = read_u32("llama.embedding_length");
         hparams.n_ff           = read_u32("llama.feed_forward_length");
         hparams.n_head         = read_u32("llama.attention.head_count");
-        hparams.n_layer        = read_u32("llama.layer_count");
+        hparams.n_layer        = read_u32("llama.block_count");
         hparams.n_rot          = read_u32("llama.rope.dimension_count");
         hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
 
@@ -1081,7 +1084,7 @@ static bool kv_cache_init(
     cache.ctx = ggml_init(params);
 
     if (!cache.ctx) {
-        fprintf(stderr, "%s: failed to allocate memory for kv cache\n", __func__);
+        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
         return false;
     }
 
@@ -1370,7 +1373,7 @@ static void llama_model_load_internal(
 
         ml->ggml_ctx = ctx;
 
-        model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
+        model.tok_embeddings = ml->get_tensor("token_embd.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
 
         // "output" tensor
         {
@@ -1391,8 +1394,8 @@ static void llama_model_load_internal(
                 backend_output = GGML_BACKEND_CPU;
             }
 
-            model.norm   = ml->get_tensor("norm.weight",   {n_embd},          backend_norm);
-            model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}, backend_output);
+            model.norm   = ml->get_tensor("output_norm.weight", {n_embd},          backend_norm);
+            model.output = ml->get_tensor("output.weight",      {n_embd, n_vocab}, backend_output);
             if (backend_norm == GGML_BACKEND_GPU) {
                 vram_weights += ggml_nbytes(model.norm);
             }
@@ -1410,20 +1413,20 @@ static void llama_model_load_internal(
 
             auto & layer = model.layers[i];
 
-            std::string layers_i = "layers." + std::to_string(i);
+            std::string layers_i = "blk." + std::to_string(i);
 
-            layer.attention_norm = ml->get_tensor(layers_i + ".attention_norm.weight", {n_embd}, backend);
+            layer.attention_norm = ml->get_tensor(layers_i + ".attn_norm.weight", {n_embd}, backend);
 
-            layer.wq = ml->get_tensor(layers_i + ".attention.wq.weight", {n_embd, n_embd},     backend_split);
-            layer.wk = ml->get_tensor(layers_i + ".attention.wk.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wv = ml->get_tensor(layers_i + ".attention.wv.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wo = ml->get_tensor(layers_i + ".attention.wo.weight", {n_embd, n_embd},     backend_split);
+            layer.wq = ml->get_tensor(layers_i + ".attn_q.weight",      {n_embd, n_embd},     backend_split);
+            layer.wk = ml->get_tensor(layers_i + ".attn_k.weight",      {n_embd, n_embd_gqa}, backend_split);
+            layer.wv = ml->get_tensor(layers_i + ".attn_v.weight",      {n_embd, n_embd_gqa}, backend_split);
+            layer.wo = ml->get_tensor(layers_i + ".attn_output.weight", {n_embd, n_embd},     backend_split);
 
             layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
 
-            layer.w1 = ml->get_tensor(layers_i + ".feed_forward.w1.weight", {n_embd,   n_ff}, backend_split);
-            layer.w2 = ml->get_tensor(layers_i + ".feed_forward.w2.weight", {  n_ff, n_embd}, backend_split);
-            layer.w3 = ml->get_tensor(layers_i + ".feed_forward.w3.weight", {n_embd,   n_ff}, backend_split);
+            layer.w1 = ml->get_tensor(layers_i + ".ffn_gate.weight", {n_embd,   n_ff}, backend_split);
+            layer.w2 = ml->get_tensor(layers_i + ".ffn_down.weight", {  n_ff, n_embd}, backend_split);
+            layer.w3 = ml->get_tensor(layers_i + ".ffn_up.weight",   {n_embd,   n_ff}, backend_split);
 
             if (backend == GGML_BACKEND_GPU) {
                 vram_weights +=
@@ -2109,6 +2112,109 @@ static bool llama_eval_internal(
 // tokenizer
 //
 
+static std::string llama_vocab_type(const llama_vocab& vocab) {
+    return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
+}
+
+static bool llama_is_normal_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token >= 259;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return token >= 95;
+    else
+        return false;
+}
+
+static bool llama_is_unknown_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 0;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_control_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 1 || token == 2;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_bos_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 1;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_eos_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return token == 2;
+    else
+        // TODO: improve?
+        return false;
+}
+
+static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
+    UNUSED(vocab);
+    UNUSED(token);
+    // TODO: improve?
+    return false;
+}
+
+static bool llama_is_unused_token(const llama_vocab& vocab, llama_token token) {
+    UNUSED(vocab);
+    UNUSED(token);
+    // TODO: improve?
+    return false;
+}
+
+static bool llama_is_byte_token(const llama_vocab& vocab, llama_token token) {
+    if(llama_vocab_type(vocab) == "spm")
+        return 3 <= token && token < 259;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return 1 <= token && token < 95;
+    else
+        return false;
+}
+
+static uint8_t llama_byte_to_char(const llama_vocab& vocab, uint8_t byte) {
+    if(llama_vocab_type(vocab) == "spm")
+        return byte + 3;
+    else if(llama_vocab_type(vocab) == "bpe")
+        return byte + 32;
+    else
+        return false;
+}
+
+static std::string llama_escape_whitespace(const std::string& text) {
+    std::string result;
+    bool escaping = false;
+    result += "\xe2\x96\x81";
+    for (size_t offs = 0; offs < text.length(); ++offs) {
+        if (text[offs] == ' ') {
+            if (!escaping) {
+                result += "\xe2\x96\x81";
+                escaping = true;
+            }
+        }
+        else {
+            escaping = false;
+            result += text[offs];
+        }
+    }
+    return result;
+}
+
+static std::string llama_unescape_whitespace(const std::string& word) {
+    if (word.length() >= 3 && word.substr(0, 3) == "\xe2\x96\x81") {
+        return std::string(" ") + word.substr(3);
+    }
+    return word;
+}
+
 static size_t utf8_len(char src) {
     const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
     uint8_t highbits = static_cast<uint8_t>(src) >> 4;
@@ -2150,10 +2256,11 @@ struct llama_tokenizer {
         size_t offs = 0;
         while (offs < text.size()) {
             llama_sp_symbol sym;
-            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
+            size_t len = utf8_len(text[offs]);
+            GGML_ASSERT(offs + len <= text.size());
             sym.text = text.c_str() + offs;
-            sym.n = char_len;
-            offs += char_len;
+            sym.n = len;
+            offs += len;
             sym.prev = index - 1;
             sym.next = offs == text.size() ? -1 : index + 1;
             index++;
@@ -2198,23 +2305,36 @@ struct llama_tokenizer {
 
         for (int i = 0; i != -1; i = symbols_[i].next) {
             auto & symbol = symbols_[i];
-            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
-
-            if (token == vocab_.token_to_id.end()) {
-                // output any symbols that did not form tokens as bytes.
-                for (int j = 0; j < (int) symbol.n; ++j) {
-                    // NOTE: old version, before #2420 - not sure what are the implications of this
-                    //llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
-                    llama_vocab::id token_id = vocab_.token_to_id.at(std::string(1, symbol.text[j]));
-                    output.push_back(token_id);
-                }
-            } else {
-                output.push_back((*token).second);
-            }
+            resegment(symbol, output);
         }
     }
 
 private:
+    void resegment(llama_sp_symbol &symbol, std::vector<llama_vocab::id> &output) {
+        auto text = std::string(symbol.text, symbol.n);
+        auto token = vocab_.token_to_id.find(text);
+
+        // Do we need to support is_unused?
+        if (token != vocab_.token_to_id.end()) {
+            output.push_back((*token).second);
+            return;
+        }
+
+        const auto p = rev_merge.find(text);
+
+        if (p == rev_merge.end()) {
+            // output any symbols that did not form tokens as bytes.
+            for (int j = 0; j < (int)symbol.n; ++j) {
+                llama_vocab::id token_id = llama_byte_to_char(vocab_, symbol.text[j]);
+                output.push_back(token_id);
+            }
+            return;
+        }
+
+        resegment(symbols_[p->second.first], output);
+        resegment(symbols_[p->second.second], output);
+    }
+
     void try_add_bigram(int left, int right) {
         if (left == -1 || right == -1) {
             return;
@@ -2239,18 +2359,22 @@ struct llama_tokenizer {
         bigram.score = tok_score.score;
         bigram.size = text.size();
         work_queue_.push(bigram);
+
+        // Do we need to support is_unused?
+        rev_merge[text] = std::make_pair(left, right);
     }
 
     const llama_vocab & vocab_;
     std::vector<llama_sp_symbol> symbols_;
     llama_sp_bigram::queue work_queue_;
+    std::map<std::string, std::pair<int, int> > rev_merge;
 };
 
-static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
+static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & raw_text, bool bos, bool escape) {
     llama_tokenizer tokenizer(vocab);
     std::vector<llama_vocab::id> output;
 
-    if (text.empty()) {
+    if (raw_text.empty()) {
         return output;
     }
 
@@ -2258,6 +2382,13 @@ static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, co
         output.push_back(llama_token_bos());
     }
 
+    std::string text;
+    if (escape) {
+        text = llama_escape_whitespace(raw_text);
+    } else {
+        text = raw_text;
+    }
+
     tokenizer.tokenize(text, output);
     return output;
 }
@@ -2839,15 +2970,15 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
 
     for (size_t i = 0; i < candidates->size; ++i) {
         const llama_token id  = candidates->data[i].id;
-        const char *      str = llama_token_to_str(ctx, id);
+        std::string       str = llama_token_to_str(ctx, id);
         if (id == eos) {
             if (!allow_eos) {
                 candidates->data[i].logit = -INFINITY;
             }
-        } else if (*str == 0) {
+        } else if (str.empty()) {
             candidates->data[i].logit = -INFINITY;
         } else {
-            candidates_decoded.push_back(decode_utf8(str));
+            candidates_decoded.push_back(decode_utf8(str.c_str()));
             candidates_grammar.push_back({ i, candidates_decoded.back().data() });
         }
     }
@@ -3048,9 +3179,9 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
         GGML_ASSERT(false);
     }
 
-    const char * str = llama_token_to_str(ctx, token);
+    std::string str = llama_token_to_str(ctx, token);
     // Note terminating 0 in decoded string
-    auto code_points = decode_utf8(str);
+    auto code_points = decode_utf8(str.c_str());
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
     }
@@ -3507,6 +3638,12 @@ struct llama_context * llama_new_context_with_model(
         // this allocates all Metal resources and memory buffers
         ctx->ctx_metal = ggml_metal_init(1);
 
+        if (!ctx->ctx_metal) {
+            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
+            llama_free(ctx);
+            return NULL;
+        }
+
         void * data_ptr  = NULL;
         size_t data_size = 0;
 
@@ -4281,7 +4418,8 @@ int llama_tokenize_with_model(
                  llama_token * tokens,
                          int   n_max_tokens,
                         bool   add_bos) {
-    auto res = llama_tokenize(model->vocab, text, add_bos);
+    auto escape = llama_vocab_type(model->vocab) == "spm";
+    auto res = llama_tokenize(model->vocab, text, add_bos, escape);
 
     if (n_max_tokens < (int) res.size()) {
         LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
@@ -4304,6 +4442,62 @@ int llama_tokenize(
     return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
 }
 
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int length = text.length() + add_bos;
+    std::vector<llama_token> result(length);
+    length = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
+}
+
+int llama_tokenize_bpe(
+        struct llama_context * ctx,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
+                        bool   add_bos) {
+    auto res = llama_tokenize(ctx->model.vocab, text, add_bos, false);
+
+    if (n_max_tokens < (int) res.size()) {
+        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
+        return -((int) res.size());
+    }
+
+    for (size_t i = 0; i < res.size(); i++) {
+        tokens[i] = res[i];
+    }
+
+    return res.size();
+}
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int length = text.length() + add_bos;
+    std::vector<llama_token> result(length);
+    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
+}
+
 int llama_n_vocab_from_model(const struct llama_model * model) {
     return model->vocab.id_to_token.size();
 }
@@ -4357,16 +4551,80 @@ float * llama_get_embeddings(struct llama_context * ctx) {
     return ctx->embedding.data();
 }
 
-const char * llama_token_to_str_with_model(const struct llama_model * model, llama_token token) {
-    if (token >= llama_n_vocab_from_model(model)) {
-        return nullptr;
+int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * str, int length) {
+    if (0 <= token && token < llama_n_vocab_from_model(model)) {
+        if (llama_is_normal_token(model->vocab, token)) {
+            std::string result = model->vocab.id_to_token[token].tok;
+            if(llama_vocab_type(model->vocab) == "spm") {
+                result = llama_unescape_whitespace(result);
+            }
+            if (length < (int) result.length()) {
+                return -result.length();
+            }
+            strncpy(str, result.c_str(), result.length());
+            return result.length();
+        } else if (llama_is_unknown_token(model->vocab, token)) {
+            if (length < 3) {
+                return -3;
+            }
+            strncpy(str, "\xe2\x96\x85", 3);
+            return 3;
+        } else if (llama_is_control_token(model->vocab, token)) {
+            ;
+        } else if (llama_is_byte_token(model->vocab, token)) {
+            if (length < 1) {
+                return -1;
+            }
+            str[0] = llama_byte_to_char(model->vocab, token);
+            str[1] = 0x00;
+            return 1;
+        }
     }
+    return 0;
+}
+
+int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * str, int length) {
+    return llama_token_to_str_with_model(&ctx->model, token, str, length);
+}
+
+std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_token_to_str(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
+    } else {
+        result.resize(length);
+    }
+
+    return std::string(result.data(), result.size());
+}
 
-    return model->vocab.id_to_token[token].tok.c_str();
+int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * str, int length) {
+    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
+        std::string result = ctx->model.vocab.id_to_token[token].tok;
+        if (length < (int) result.length()) {
+            return -result.length();
+        }
+        strncpy(str, result.c_str(), result.length());
+        return result.length();
+    }
+    return 0;
 }
 
-const char * llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    return llama_token_to_str_with_model(&ctx->model, token);
+std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+    if (length < 0) {
+        result.resize(-length);
+        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
+    } else {
+        result.resize(length);
+    }
+
+    return std::string(result.data(), result.size());
 }
 
 llama_token llama_token_bos() {
diff --git a/gguf-llama.h b/gguf-llama.h
index a8ed69d918048..f342a534c3ccb 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -311,6 +311,13 @@ extern "C" {
                              int   n_max_tokens,
                             bool   add_bos);
 
+    LLAMA_API int llama_tokenize_bpe(
+            struct llama_context * ctx,
+                      const char * text,
+                     llama_token * tokens,
+                             int   n_max_tokens,
+                            bool   add_bos);
+
     LLAMA_API int llama_tokenize_with_model(
         const struct llama_model * model,
                       const char * text,
@@ -352,14 +359,23 @@ extern "C" {
     LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
 
     // Token Id -> String. Uses the vocabulary in the provided context
-    LLAMA_API const char * llama_token_to_str(
+    LLAMA_API int llama_token_to_str(
             const struct llama_context * ctx,
-                           llama_token   token);
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
 
-    LLAMA_API const char * llama_token_to_str_with_model(
-              const struct llama_model * model,
-                           llama_token   token);
+    LLAMA_API int llama_token_to_str_bpe(
+            const struct llama_context * ctx,
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
 
+    LLAMA_API int llama_token_to_str_with_model(
+              const struct llama_model * model,
+                           llama_token   token,
+                                  char * str,
+                                  int    length);
     // Special tokens
     LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
     LLAMA_API llama_token llama_token_eos();  // end-of-sentence
@@ -447,15 +463,43 @@ extern "C" {
 }
 #endif
 
-// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
-#ifdef LLAMA_API_INTERNAL
+// C++ API, will be moving to common.h soon (TM)
+#ifdef LLAMA_API_CPP
 
 #include <vector>
 #include <string>
+
+//
+// Vocab utils
+//
+
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::string llama_token_to_str(
+        const struct llama_context * ctx,
+                       llama_token   token);
+
+std::string llama_token_to_str_bpe(
+    const struct llama_context * ctx,
+                   llama_token   token);
+
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
+
 struct ggml_tensor;
 
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
 
-#endif
+#endif // LLAMA_API_CPP
+
+#endif // LLAMA_API_INTERNAL
 
 #endif // LLAMA_H
diff --git a/gguf_namemap.py b/gguf_namemap.py
index 7546630ed26c8..06cd0132d9ea4 100644
--- a/gguf_namemap.py
+++ b/gguf_namemap.py
@@ -4,92 +4,92 @@ def get_tensor_namemap( n_blocks : int):
     tensor_map = {}
     # Token embeddings
     mapped_to = "token_embd"
-    tensor_map["gpt_neox.embed_in"] = mapped_to           # gptneox
-    tensor_map["transformer.wte"] = mapped_to             # gpt2 mpt
+    tensor_map["gpt_neox.embed_in"]           = mapped_to # gptneox
+    tensor_map["transformer.wte"]             = mapped_to # gpt2 mpt
     tensor_map["transformer.word_embeddings"] = mapped_to # falcon
-    tensor_map["model.embed_tokens"] = mapped_to          # llama-hf
-    tensor_map["tok_embeddings"] = mapped_to              # llama-pth
+    tensor_map["model.embed_tokens"]          = mapped_to # llama-hf
+    tensor_map["tok_embeddings"]              = mapped_to # llama-pth
     # Position embeddings
     mapped_to = "pos_embd"
     tensor_map["transformer.wpe"] = mapped_to # gpt2
     # Output norm
     mapped_to = "output_norm"
     tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
-    tensor_map["transformer.ln_f"] = mapped_to          # gpt2 falcon
-    tensor_map["transformer.norm_f"] = mapped_to        # mpt
-    tensor_map["model.norm"] = mapped_to                # llama-hf
-    tensor_map["norm"] = mapped_to                      # llama-pth
+    tensor_map["transformer.ln_f"]          = mapped_to # gpt2 falcon
+    tensor_map["transformer.norm_f"]        = mapped_to # mpt
+    tensor_map["model.norm"]                = mapped_to # llama-hf
+    tensor_map["norm"]                      = mapped_to # llama-pth
     # Output
     mapped_to = "output"
     tensor_map["embed_out"] = mapped_to # gptneox
-    tensor_map["lm_head"] = mapped_to   # gpt2 mpt falcon llama-hf
-    tensor_map["output"] = mapped_to    # llama-pth
+    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
+    tensor_map["output"]    = mapped_to # llama-pth
     # Attention and fee-forward layer blocks
     for i in range(0,n_blocks):
         # Attention norm
         mapped_to = "blk."+str(i)+".attn_norm"
         tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to              # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to       # mpt
-        tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to   # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to           # falcon40b
-        tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to    # llama-hf
-        tensor_map["layers."+str(i)+".attention_norm"] = mapped_to           # llama-pth
+        tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to # llama-pth
         # Attention norm 2
         mapped_to = "blk."+str(i)+".attn_norm_2"
         tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
         # Attention query-key-value
         mapped_to = "blk."+str(i)+".attn_qkv"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"] = mapped_to     # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_attn"] = mapped_to                     # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"] = mapped_to                  # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to  # falcon
+        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon
         # Attention query
         mapped_to = "blk."+str(i)+".attn_q"
         tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wq"] = mapped_to           # llama-pth
+        tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to # llama-pth
         # Attention key
         mapped_to = "blk."+str(i)+".attn_k"
         tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wk"] = mapped_to           # llama-pth
+        tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to # llama-pth
         # Attention value
         mapped_to = "blk."+str(i)+".attn_v"
         tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wv"] = mapped_to           # llama-pth
+        tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to # llama-pth
         # Attention output
         mapped_to = "blk."+str(i)+".attn_output"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"] = mapped_to    # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_proj"] = mapped_to          # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"] = mapped_to   # mpt
+        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to # mpt
         tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".self_attn.o_proj"] = mapped_to      # llama-hf
-        tensor_map["layers."+str(i)+".attention.wo"] = mapped_to                # llama-pth
+        tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to # llama-pth
         # Feed-forward norm
         mapped_to = "blk."+str(i)+".ffn_norm"
         tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to                       # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_2"] = mapped_to               # mpt
-        tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to    # llama-hf
-        tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to                          # llama-pth
+        tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to # mpt
+        tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to # llama-pth
         # Feed-forward up
         mapped_to = "blk."+str(i)+".ffn_up"
         tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_fc"] = mapped_to            # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"] = mapped_to    # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"] = mapped_to   # falcon
-        tensor_map["model.layers."+str(i)+".mlp.up_proj"] = mapped_to          # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w3"] = mapped_to            # llama-pth
+        tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to # llama-pth
         # Feed-forward gate
         mapped_to = "blk."+str(i)+".ffn_gate"
         tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w1"] = mapped_to     # llama-pth
+        tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to # llama-pth
         # Feed-forward down
         mapped_to = "blk."+str(i)+".ffn_down"
         tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_proj"] = mapped_to          # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"] = mapped_to  # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"] = mapped_to   # falcon
-        tensor_map["model.layers."+str(i)+".mlp.down_proj"] = mapped_to        # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w2"] = mapped_to            # llama-pth
+        tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"]   = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".mlp.down_proj"]        = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w2"]            = mapped_to # llama-pth
 
     return tensor_map
diff --git a/llama.cpp b/llama.cpp
index 5504f14834c35..2ac907b966e4b 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -72,6 +72,7 @@ static void llama_log_callback_default(llama_log_level level, const char * text,
 #define LLAMA_MAX_SCRATCH_BUFFERS 16
 #endif
 
+#define UNUSED GGML_UNUSED
 
 // available llama models
 enum e_model {
@@ -1989,11 +1990,15 @@ static bool llama_is_eos_token(const llama_vocab& vocab, llama_token token) {
 }
 
 static bool llama_is_user_defined_token(const llama_vocab& vocab, llama_token token) {
+    UNUSED(vocab);
+    UNUSED(token);
     // TODO: improve?
     return false;
 }
 
 static bool llama_is_unused_token(const llama_vocab& vocab, llama_token token) {
+    UNUSED(vocab);
+    UNUSED(token);
     // TODO: improve?
     return false;
 }
@@ -4399,21 +4404,21 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
             if(llama_vocab_type(model->vocab) == "spm") {
                 result = llama_unescape_whitespace(result);
             }
-            if(result.length() > length) {
-                return - result.length();
+            if (length < (int) result.length()) {
+                return -result.length();
             }
-            strcpy(str, result.c_str());
+            strncpy(str, result.c_str(), result.length());
             return result.length();
         } else if (llama_is_unknown_token(model->vocab, token)) {
-            if(3 > length) {
+            if (length < 3) {
                 return -3;
             }
-            strcpy(str, "\xe2\x96\x85");
+            strncpy(str, "\xe2\x96\x85", 3);
             return 3;
         } else if (llama_is_control_token(model->vocab, token)) {
             ;
         } else if (llama_is_byte_token(model->vocab, token)) {
-            if(1 > length) {
+            if (length < 1) {
                 return -1;
             }
             str[0] = llama_byte_to_char(model->vocab, token);
@@ -4428,52 +4433,44 @@ int llama_token_to_str(const struct llama_context * ctx, llama_token token, char
     return llama_token_to_str_with_model(&ctx->model, token, str, length);
 }
 
-std::string llama_token_to_str(
-        const struct llama_context * ctx,
-                       llama_token   token) {
-    std::string result;
-    int length = 8;
-    result.resize(length);
-    length = llama_token_to_str(ctx, token, (char *)result.data(), result.length());
+std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
     if (length < 0) {
         result.resize(-length);
-        int check = llama_token_to_str(ctx, token, (char *)result.data(), result.length());
-        assert(check == -length);
-        GGML_UNUSED(check);
+        int check = llama_token_to_str(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
     } else {
         result.resize(length);
     }
-    return result;
+
+    return std::string(result.data(), result.size());
 }
 
 int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * str, int length) {
     if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
         std::string result = ctx->model.vocab.id_to_token[token].tok;
-        if (result.length() > length) {
-            return - result.length();
+        if (length < (int) result.length()) {
+            return -result.length();
         }
-        strcpy(str, result.c_str());
+        strncpy(str, result.c_str(), result.length());
         return result.length();
     }
     return 0;
 }
 
-std::string llama_token_to_str_bpe(
-    const struct llama_context * ctx,
-                   llama_token   token) {
-    std::string result;
-    int length = 8;
-    result.resize(length);
-    length = llama_token_to_str_bpe(ctx, token, (char*)result.data(), result.length());
+std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
     if (length < 0) {
         result.resize(-length);
-        int check = llama_token_to_str_bpe(ctx, token, (char*)result.data(), result.length());
-        assert(check == -length);
-        GGML_UNUSED(check);
+        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
     } else {
         result.resize(length);
     }
-    return result;
+
+    return std::string(result.data(), result.size());
 }
 
 llama_token llama_token_bos() {
diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp
index a523c320c2aeb..f973271a34eb0 100644
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -8,14 +8,13 @@
 
 static std::string unescape_whitespace(llama_context* ctx, const std::vector<llama_token>& tokens) {
     std::string result;
-    for (int i = 0; i < tokens.size(); ++i) {
+    for (size_t i = 0; i < tokens.size(); ++i) {
         result += llama_token_to_str(ctx, tokens[i]);
     }
     return result;
 }
 
-static const std::map<std::string, std::vector<llama_token>> & k_tests()
-{
+static const std::map<std::string, std::vector<llama_token>> & k_tests() {
     static std::map<std::string, std::vector<llama_token>> _k_tests = {
         { " ",                      {1,    259, }, },
         { "\t",                     { 1,    29871,   12, }, },
@@ -29,17 +28,18 @@ static const std::map<std::string, std::vector<llama_token>> & k_tests()
         { " this is 🦙.cpp",        { 1,  29871,    445,    338,  29871,    243,    162,    169,    156,  29889,   8223, }, },
         { "w048 7tuijk dsdfhu",     { 1,    281,  29900,  29946,  29947,  29871,  29955,   9161,  13535,  18031,   2176,   6905, }, },
         { "нещо на Български",      { 1,   1538,   4851,    665,   1386,  29713,   1305, }, },
-        { "កាន់តែពិសេសអាចខលចេញ",   { 1,  29871,  31849,  31324,  31934,    228,    162,    142,    228,    161,    
-                                        146,    228,    162,    133,    228,    161,    153,    228,    161,    186,  
-                                        31708,    228,    162,    132,  31708,    228,    161,    165,  31324,    228,    
-                                        161,    136,    228,    161,    132,    228,    161,    158,    228,    161,    
-                                        136,    228,    162,    132,    228,    161,    140, }, },
+        { "កាន់តែពិសេសអាចខលចេញ",   { 1,  29871,  31849,  31324,  31934,    228,    162,    142,    228,    161,
+                                     146,    228,    162,    133,    228,    161,    153,    228,    161,    186,
+                                     31708,    228,    162,    132,  31708,    228,    161,    165,  31324,    228,
+                                     161,    136,    228,    161,    132,    228,    161,    158,    228,    161,
+                                     136,    228,    162,    132,    228,    161,    140, }, },
         { "🚀 (normal) 😶‍🌫️ (multiple emojis concatenated) ✅ (only emoji that has its own token)",
-            { 1,  29871,    243,    162,    157,    131,    313,   8945,  29897,  29871,    
-              243,    162,    155,    185,  30722,    243,    162,    143,    174,  30598,    
-              313,  20787,    953,   3848,    275,  16125,    630,  29897,  29871,  31681,    
-              313,   6194,    953,  29877,   2397,    393,    756,    967,   1914,   5993,  29897, }, },
-     };
+            { 1,  29871,    243,    162,    157,    131,    313,   8945,  29897,  29871,
+                243,    162,    155,    185,  30722,    243,    162,    143,    174,  30598,
+                313,  20787,    953,   3848,    275,  16125,    630,  29897,  29871,  31681,
+                313,   6194,    953,  29877,   2397,    393,    756,    967,   1914,   5993,  29897, }, },
+    };
+
     return _k_tests;
 };
 
@@ -90,8 +90,8 @@ int main(int argc, char **argv) {
     }
 
     for (const auto & test_kv : k_tests()) {
-        std::vector<llama_token> res = llama_tokenize(ctx, test_kv.first.c_str(), true);
-        fprintf(stderr, "%s : '%s' tokenized to '%s'\n", 
+        std::vector<llama_token> res = llama_tokenize(ctx, test_kv.first, true);
+        fprintf(stderr, "%s : '%s' tokenized to '%s'\n",
             __func__, test_kv.first.c_str(), unescape_whitespace(ctx, res).c_str());
 
         bool correct = res.size() == test_kv.second.size();
diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
index 122e5168424ae..a44f38cd714ad 100644
--- a/tests/test-tokenizer-1.cpp
+++ b/tests/test-tokenizer-1.cpp
@@ -8,8 +8,9 @@
 #include <codecvt>
 #include <map>
 #include <vector>
+#include <locale>
 
-static std::string vocab_type(llama_context* ctx) {
+static std::string vocab_type(llama_context * ctx) {
     return llama_n_vocab(ctx) == 32000 ? "spm": "bpe";
 }
 
@@ -32,9 +33,9 @@ static std::string escape_whitespace(const std::string& text) {
     return result;
 }
 
-static std::string unescape_whitespace(llama_context* ctx, const std::vector<llama_token>& tokens) {
+static std::string unescape_whitespace(llama_context * ctx, const std::vector<llama_token> & tokens) {
     std::string result;
-    for (int i = 0; i < tokens.size(); ++i) {
+    for (size_t i = 0; i < tokens.size(); ++i) {
         result += llama_token_to_str(ctx, tokens[i]);
     }
     return result;
@@ -85,17 +86,17 @@ int main(int argc, char **argv) {
         if (tokens.size() == 1) {
             if (i != tokens[0]) {
                 std::string backward = llama_token_to_str(ctx, tokens[0]);
-                fprintf(stderr, "%s : error: token %d is string %s but bpe returns token %d %s\n", 
+                fprintf(stderr, "%s : error: token %d is string %s but bpe returns token %d %s\n",
                     __func__, i, llama_token_to_str(ctx, i).c_str(), tokens[0], backward.c_str());
                 return 2;
             }
         } else {
-            if ((vocab_type(ctx) == "spm" && i <= 258) || 
+            if ((vocab_type(ctx) == "spm" && i <= 258) ||
                 (vocab_type(ctx) == "bpe" && (i == 0 || i >= 100000))) {
-                fprintf(stderr, "%s : info: token %d is string %s and bpe returns tokens %s\n", 
+                fprintf(stderr, "%s : info: token %d is string %s and bpe returns tokens %s\n",
                     __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
             } else {
-                fprintf(stderr, "%s : error: token %d is string %s but bpe returns tokens %s\n", 
+                fprintf(stderr, "%s : error: token %d is string %s but bpe returns tokens %s\n",
                     __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
                 return 2;
             }
@@ -105,10 +106,15 @@ int main(int argc, char **argv) {
     std::wstring_convert<typename std::codecvt_utf8<wchar_t>, wchar_t> converter;
     for (wchar_t ch = 0x0000; ch < 0xffff; ++ch) {
         std::wstring wstr(1, ch);
-        std::string str = converter.to_bytes(wstr);
-        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
+        std::string str;
+        try {
+            str = converter.to_bytes(wstr);
+        } catch (std::exception & e) {
+            continue;
+        }
+        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str), false);
         if (tokens.size() == 1) {
-            fprintf(stderr, "%s : info: %s tokenized to %d \n", 
+            fprintf(stderr, "%s : info: %s tokenized to %d \n",
                 __func__, str.c_str(), tokens[0]);
         }
     }

From 6c63550f63657cd00c5e6166c533ff190d1e0f3a Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 14 Aug 2023 22:10:19 +0300
Subject: [PATCH 149/242] llama : update tokenizer style

---
 gguf-llama.cpp | 87 +++++++++++++++++++++++++++---------------------
 llama.cpp      | 89 +++++++++++++++++++++++++++++---------------------
 2 files changed, 101 insertions(+), 75 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 8275c7357ec4c..1c1d6718e5740 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -2112,49 +2112,56 @@ static bool llama_eval_internal(
 // tokenizer
 //
 
-static std::string llama_vocab_type(const llama_vocab& vocab) {
+static std::string llama_vocab_type(const llama_vocab & vocab) {
     return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
 }
 
-static bool llama_is_normal_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_normal_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token >= 259;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return token >= 95;
-    else
-        return false;
+    }
+
+    return false;
 }
 
-static bool llama_is_unknown_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 0;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_control_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_control_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 1 || token == 2;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_bos_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 1;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_eos_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 2;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
 static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
@@ -2164,29 +2171,35 @@ static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token t
     return false;
 }
 
-static bool llama_is_unused_token(const llama_vocab& vocab, llama_token token) {
+static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
     UNUSED(vocab);
     UNUSED(token);
     // TODO: improve?
     return false;
 }
 
-static bool llama_is_byte_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return 3 <= token && token < 259;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return 1 <= token && token < 95;
-    else
-        return false;
+    }
+
+    return false;
 }
 
-static uint8_t llama_byte_to_char(const llama_vocab& vocab, uint8_t byte) {
-    if(llama_vocab_type(vocab) == "spm")
+static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
+    if (llama_vocab_type(vocab) == "spm") {
         return byte + 3;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return byte + 32;
-    else
-        return false;
+    }
+
+    return false;
 }
 
 static std::string llama_escape_whitespace(const std::string& text) {
diff --git a/llama.cpp b/llama.cpp
index 2ac907b966e4b..fb2207f84bad8 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1944,81 +1944,94 @@ static bool llama_eval_internal(
 // tokenizer
 //
 
-static std::string llama_vocab_type(const llama_vocab& vocab) {
+static std::string llama_vocab_type(const llama_vocab & vocab) {
     return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
 }
 
-static bool llama_is_normal_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_normal_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token >= 259;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return token >= 95;
-    else
-        return false;
+    }
+
+    return false;
 }
 
-static bool llama_is_unknown_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 0;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_control_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_control_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 1 || token == 2;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_bos_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 1;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_eos_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return token == 2;
-    else
-        // TODO: improve?
-        return false;
+    }
+
+    // TODO: improve?
+    return false;
 }
 
-static bool llama_is_user_defined_token(const llama_vocab& vocab, llama_token token) {
+static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
     UNUSED(vocab);
     UNUSED(token);
     // TODO: improve?
     return false;
 }
 
-static bool llama_is_unused_token(const llama_vocab& vocab, llama_token token) {
+static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
     UNUSED(vocab);
     UNUSED(token);
     // TODO: improve?
     return false;
 }
 
-static bool llama_is_byte_token(const llama_vocab& vocab, llama_token token) {
-    if(llama_vocab_type(vocab) == "spm")
+static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_type(vocab) == "spm") {
         return 3 <= token && token < 259;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return 1 <= token && token < 95;
-    else
-        return false;
+    }
+
+    return false;
 }
 
-static uint8_t llama_byte_to_char(const llama_vocab& vocab, uint8_t byte) {
-    if(llama_vocab_type(vocab) == "spm")
+static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
+    if (llama_vocab_type(vocab) == "spm") {
         return byte + 3;
-    else if(llama_vocab_type(vocab) == "bpe")
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
         return byte + 32;
-    else
-        return false;
+    }
+
+    return false;
 }
 
 static std::string llama_escape_whitespace(const std::string& text) {

From 7ec125b1dc236130eceb0109c36ed5a7317eb80f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 22:06:33 +0200
Subject: [PATCH 150/242] convert-llama-h5-to-gguf.py : add token types

---
 convert-llama-h5-to-gguf.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 9d91b433bf29e..1a805cff83c80 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -111,12 +111,13 @@ def count_model_parts(dir_model: str) -> int:
 
 print("gguf: get tokenizer metadata")
 
-tokens: List[str] = []
+tokens: List[bytes] = []
 scores: List[float] = []
+toktypes: List[int] = []
 
 if Path(dir_model + "/tokenizer.model").is_file():
     # vocab type sentencepiece
-    print("gguf: get sentencepiece tokenizer vocab and scores")
+    print("gguf: get sentencepiece tokenizer vocab, scores and token types")
 
     tokenizer = SentencePieceProcessor(dir_model + "/tokenizer.model")
 
@@ -128,12 +129,25 @@ def count_model_parts(dir_model: str) -> int:
         text  = piece.encode("utf-8")
         score = tokenizer.get_score(i)
 
+        toktype = 1 # defualt to normal token type
+        if tokenizer.is_unknown(i): toktype = 2
+        if tokenizer.is_control(i): toktype = 3
+ 
+        # TODO: How to determinate if a token is user defined?
+        # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
+        # if tokenizer.is_user_defined(i): toktype = 4
+
+        if tokenizer.is_unused(i):  toktype = 5
+        if tokenizer.is_byte(i):    toktype = 6
+
         tokens.append(text)
         scores.append(score)
+        toktypes.append(toktype)
 
     gguf_writer.add_tokenizer_model("llama")
     gguf_writer.add_token_list(tokens)
     gguf_writer.add_token_scores(scores)
+    gguf_writer.add_token_types(toktypes)
 
 if Path(dir_model + "/tokenizer.json").is_file():
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:

From 5d518d421f46cb87214376e7195bc3c610106fdd Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 22:07:53 +0200
Subject: [PATCH 151/242] constants.py : add token types

---
 constants.py | 28 ++++++++++++----------------
 1 file changed, 12 insertions(+), 16 deletions(-)

diff --git a/constants.py b/constants.py
index 87d6b7a0687aa..2f96fb9604e0c 100644
--- a/constants.py
+++ b/constants.py
@@ -36,19 +36,15 @@
 KEY_ROPE_SCALE                   = "{llm}.rope.scale"
 
 # tokenization
-KEY_TOKENIZER_MODEL   = "tokenizer.ggml.model"
-KEY_TOKENIZER_LIST    = "tokenizer.ggml.tokens"
-KEY_TOKENIZER_SCORES  = "tokenizer.ggml.scores"
-KEY_TOKENIZER_MERGES  = "tokenizer.ggml.merges"
-KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
-KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
-KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
-KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.seperator_token_id"
-KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
-KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json"
-KEY_TOKENIZER_RWKV    = "tokenizer.rwkv.world"
-KEY_TOKENIZER_BOS_ID  = "tokenizer.ggml.bos_token_id"
-KEY_TOKENIZER_EOS_ID  = "tokenizer.ggml.eos_token_id"
-KEY_TOKENIZER_UNK_ID  = "tokenizer.ggml.unknown_token_id"
-KEY_TOKENIZER_SEP_ID  = "tokenizer.ggml.separator_token_id"
-KEY_TOKENIZER_PAD_ID  = "tokenizer.ggml.padding_token_id"
+KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
+KEY_TOKENIZER_LIST       = "tokenizer.ggml.tokens"
+KEY_TOKENIZER_TOKEN_TYPE = "tokenizer.ggml.token_type"
+KEY_TOKENIZER_SCORES     = "tokenizer.ggml.scores"
+KEY_TOKENIZER_MERGES     = "tokenizer.ggml.merges"
+KEY_TOKENIZER_BOS_ID     = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID     = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID     = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID     = "tokenizer.ggml.seperator_token_id"
+KEY_TOKENIZER_PAD_ID     = "tokenizer.ggml.padding_token_id"
+KEY_TOKENIZER_HF_JSON    = "tokenizer.huggingface.json"
+KEY_TOKENIZER_RWKV       = "tokenizer.rwkv.world"

From cedb4870c660875d7d0b6fd98015960a27dd302e Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 22:08:40 +0200
Subject: [PATCH 152/242] gguf.py : add token types

---
 gguf.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/gguf.py b/gguf.py
index 93591d5c76fa9..d3b95963afef0 100644
--- a/gguf.py
+++ b/gguf.py
@@ -297,6 +297,9 @@ def add_token_list(self, tokens: List):
     def add_token_merges(self, merges: List):
         self.add_array(constants.KEY_TOKENIZER_MERGES, merges)
 
+    def add_token_types(self, types: List[int]):
+        self.add_array(constants.KEY_TOKENIZER_TOKEN_TYPE, types)
+
     def add_token_scores(self, scores: List[float]):
         self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
 

From ab2cbd03cafc160c52c0c0b77960ddd7a599c618 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 22:10:50 +0200
Subject: [PATCH 153/242] convert-llama-7b-pth-to-gguf.py : add token types

---
 convert-llama-7b-pth-to-gguf.py | 34 +++++++++++++++++++--------------
 1 file changed, 20 insertions(+), 14 deletions(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index d763503564c1d..96d2e08003ac0 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -110,8 +110,9 @@ def count_model_parts(dir_model: str) -> int:
 
 print("gguf: get tokenizer metadata")
 
-tokens: List[str] = []
+tokens: List[bytes] = []
 scores: List[float] = []
+toktypes: List[int] = []
 
 if Path(dir_model + "/tokenizer.model").is_file():
     # vocab type sentencepiece
@@ -121,26 +122,31 @@ def count_model_parts(dir_model: str) -> int:
 
     for i in range(tokenizer.vocab_size()):
         text: bytes
-        if tokenizer.is_unknown(i):
-            text = " \u2047 ".encode("utf-8")
-        elif tokenizer.is_control(i):
-            text = b""
-        if tokenizer.is_byte(i):
-            piece = tokenizer.id_to_piece(i)
-            if len(piece) != 6:
-                raise Exception(f"Invalid token: {piece}")
-            byte_value = int(piece[3:-1], 16)
-            text = struct.pack("B", byte_value)
-        else:
-            text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
-        score: float = tokenizer.get_score(i)
+        score: float
+
+        piece = tokenizer.id_to_piece(i)
+        text  = piece.encode("utf-8")
+        score = tokenizer.get_score(i)
+
+        toktype = 1 # defualt to normal token type
+        if tokenizer.is_unknown(i): toktype = 2
+        if tokenizer.is_control(i): toktype = 3
+ 
+        # TODO: How to determinate if a token is user defined?
+        # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
+        # if tokenizer.is_user_defined(i): toktype = 4
+
+        if tokenizer.is_unused(i):  toktype = 5
+        if tokenizer.is_byte(i):    toktype = 6
 
         tokens.append(text)
         scores.append(score)
+        toktypes.append(toktype)
 
     gguf_writer.add_tokenizer_model("llama")
     gguf_writer.add_token_list(tokens)
     gguf_writer.add_token_scores(scores)
+    gguf_writer.add_token_types(toktypes)
 
 if Path(dir_model + "/tokenizer.json").is_file():
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:

From ca4758290c9202a89de2dcd7f13f4d9c212cc8eb Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 14 Aug 2023 23:18:41 +0200
Subject: [PATCH 154/242] gguf-llama.cpp :  fix n_head_kv

---
 gguf-llama.cpp | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index 1c1d6718e5740..e36d8e77cfc4d 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -630,8 +630,9 @@ struct gguf_file_loader {
         hparams.n_rot          = read_u32("llama.rope.dimension_count");
         hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
 
-        // LLaMAv2
-        // hparams.n_head_kv = read_u32("llama.attention.head_count_kv");
+        // n_head_kv default to n_head
+        hparams.n_head_kv = gguf_find_key(gguf_ctx, "llama.attention.head_count_kv") == -1 ?  hparams.n_head : read_u32("llama.attention.head_count_kv");
+
     }
 
     void read_vocab() {

From 2dd5d2c92c52e19f30f1545d922689ddd216bbbb Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 15 Aug 2023 00:43:10 +0200
Subject: [PATCH 155/242] convert-llama-h5-to-gguf.py : add 70b gqa support

---
 convert-llama-h5-to-gguf.py | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 1a805cff83c80..22405673f8a8d 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -1,4 +1,4 @@
-# HF llama --> gguf conversion, GQA/70b not supported
+# HF llama --> gguf conversion
 
 import gguf
 import gguf_namemap as tmap
@@ -10,7 +10,7 @@
 import numpy as np
 import torch
 
-from typing import Any, List
+from typing import Any, List, Optional
 from pathlib import Path
 from sentencepiece import SentencePieceProcessor
 
@@ -18,11 +18,11 @@
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
-
-def permute(weights: NDArray, n_head: int) -> NDArray:
+def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
+    if n_kv_head is not None and n_head != n_kv_head: n_head //= n_kv_head
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                   .swapaxes(1, 2)
-                   .reshape(weights.shape))
+                .swapaxes(1, 2)
+                .reshape(weights.shape))
 
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
@@ -220,7 +220,7 @@ def count_model_parts(dir_model: str) -> int:
 
         # permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data,head_count)
+            data = permute(data, head_count, head_count_kv)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:
@@ -289,7 +289,7 @@ def count_model_parts(dir_model: str) -> int:
 
         # permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data, head_count)
+            data = permute(data, head_count, head_count_kv)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:

From b6056c3db86ce55e6b05f1fc988fdbb02004c359 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 15 Aug 2023 19:53:44 +0200
Subject: [PATCH 156/242] gguf.py : add tensor data layout

---
 gguf.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/gguf.py b/gguf.py
index d3b95963afef0..1b15554f3ab76 100644
--- a/gguf.py
+++ b/gguf.py
@@ -207,6 +207,9 @@ def add_architecture(self, architecture: str):
     def add_author(self, author: str):
         self.add_string(constants.KEY_GENERAL_AUTHOR, author)
 
+    def add_tensor_data_layout(self, layout: str):
+        self.add_string(constants.KEY_LLM_TENSOR_DATA_LAYOUT , layout)
+
     def add_url(self, url: str):
         self.add_string(constants.KEY_GENERAL_URL, url)
 

From 66756c82af57a661fe7003ba126d5cb041a945f1 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 15 Aug 2023 19:54:33 +0200
Subject: [PATCH 157/242] convert-llama-h5-to-gguf.py : add tensor data layout

---
 convert-llama-h5-to-gguf.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 22405673f8a8d..fd3b9d55ba4fc 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -97,6 +97,7 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_name(last_dir)
 gguf_writer.add_file_type("All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
+gguf_writer.add_tensor_data_layout(llm_arch, "Meta AI original pth")
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)

From 2ae0e985b3ec9e4e12f9ae793107a9cca74af957 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 15 Aug 2023 19:55:13 +0200
Subject: [PATCH 158/242] convert-llama-7b-pth-to-gguf.py : add tensor data
 layout

---
 convert-llama-7b-pth-to-gguf.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 96d2e08003ac0..53ab5a3ed0c13 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -96,6 +96,7 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_name(last_dir)
 gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
+gguf_writer.add_tensor_data_layout(llm_arch, "Meta AI original pth")
 gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)

From 4a1741aa2dd589eabaa7820aea567e1fe55bba1d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Tue, 15 Aug 2023 19:56:19 +0200
Subject: [PATCH 159/242] gptneox-main.cpp : add tensor data layout

---
 gptneox-main.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/gptneox-main.cpp b/gptneox-main.cpp
index f336ee88bfaf9..aca8a5b0a3bd5 100644
--- a/gptneox-main.cpp
+++ b/gptneox-main.cpp
@@ -381,6 +381,8 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         if (keyidx != -1) { fprintf(stdout, "%s: model architecture   = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.file_type");
         if (keyidx != -1) { fprintf(stdout, "%s: model file type      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "gptneox.tensor_data_layout");
+        if (keyidx != -1) { fprintf(stdout, "%s: model data layout    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
         keyidx = gguf_find_key(ggufctx, "general.source.hugginface.repository");
         if (keyidx != -1) { fprintf(stdout, "%s: model source HF repo = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
     }

From ea5615a03a0e3b22817e2ac6e8a12ee66a8e390f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Wed, 16 Aug 2023 11:23:15 +0200
Subject: [PATCH 160/242] convert-llama-h5-to-gguf.py : clarify the reverse
 permute

---
 convert-llama-h5-to-gguf.py | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index fd3b9d55ba4fc..18b0f35f0c3a5 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -18,7 +18,9 @@
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
-def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
+# reverse HF permute back to original pth layout
+# https://github.com/huggingface/transformers/blob/main/src/transformers/models/llama/convert_llama_weights_to_hf.py
+def reverse_hf_permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
     if n_kv_head is not None and n_head != n_kv_head: n_head //= n_kv_head
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
                 .swapaxes(1, 2)
@@ -219,9 +221,9 @@ def count_model_parts(dir_model: str) -> int:
 
         data = data.squeeze().numpy()
 
-        # permute these
+        # reverse permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data, head_count, head_count_kv)
+            data = reverse_hf_permute(data, head_count, head_count_kv)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:
@@ -288,9 +290,9 @@ def count_model_parts(dir_model: str) -> int:
 
         data = data.squeeze().numpy()
 
-        # permute these
+        # reverse permute these
         if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = permute(data, head_count, head_count_kv)
+            data = reverse_hf_permute(data, head_count, head_count_kv)
 
         # map tensor names
         if name.endswith(".weight") and name[:-7] in tensor_map:

From 758ff1bbb59de40dfe8e1aae2cf5dfd3af51ae76 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 16 Aug 2023 14:34:03 +0300
Subject: [PATCH 161/242] llama : refactor model loading code (#2620)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* llama : style formatting + remove helper methods

* llama : fix quantization using gguf tool

* llama : simplify gguf_file_saver

* llama : fix method names

* llama : simplify write_header()

* llama : no need to pass full file loader to the file saver

just gguf_ctx

* llama : gguf_file_saver write I32

* llama : refactor tensor names (#2622)

* gguf: update tensor names searched in quantization

* gguf : define tensor names as constants

* gguf : initial write API (not tested yet)

* gguf : write to file API (not tested)

* gguf : initial write API ready + example

* gguf : fix header write

* gguf : fixes + simplify example + add ggml_nbytes_pad()

* gguf : minor

* llama : replace gguf_file_saver with new gguf write API

* gguf : streaming support when writing files

* gguf : remove oboslete write methods

* gguf : remove obosolete gguf_get_arr_xxx API

* llama : simplify gguf_file_loader

* llama : move hparams and vocab from gguf_file_loader to llama_model_loader

* llama : merge gguf-util.h in llama.cpp

* llama : reorder definitions in .cpp to match .h

* llama : minor simplifications

* llama : refactor llama_model_loader (WIP)

wip : remove ggml_ctx from llama_model_loader

wip : merge gguf_file_loader in llama_model_loader

* llama : fix shape prints

* llama : fix Windows build + fix norm_rms_eps key

* llama : throw error on missing KV paris in model meta data

* llama : improve printing + log meta data

* llama : switch print order of meta data

---------

Co-authored-by: M. Yusuf Sarıgöz <yusufsarigoz@gmail.com>
---
 Makefile                            |    2 +-
 convert-llama-h5-to-gguf.py         |    2 +-
 examples/gguf/gguf-llama-simple.cpp |    7 +-
 examples/gguf/gguf.cpp              |  248 +--
 ggml.c                              |  597 +++++-
 ggml.h                              |   73 +-
 gguf-llama.cpp                      | 2614 +++++++++++++++------------
 gguf-llama.h                        |   30 +-
 gguf-util.h                         |  470 -----
 9 files changed, 2049 insertions(+), 1994 deletions(-)
 delete mode 100644 gguf-util.h

diff --git a/Makefile b/Makefile
index 5e50f46919b4d..ce36fcf6b730c 100644
--- a/Makefile
+++ b/Makefile
@@ -332,7 +332,7 @@ OBJS += ggml-alloc.o
 llama.o: llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h llama.h llama-util.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
-gguf-llama.o: gguf-llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h gguf-llama.h gguf-util.h
+gguf-llama.o: gguf-llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h gguf-llama.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
 common.o: examples/common.cpp examples/common.h
diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 18b0f35f0c3a5..d7706f6187a58 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -135,7 +135,7 @@ def count_model_parts(dir_model: str) -> int:
         toktype = 1 # defualt to normal token type
         if tokenizer.is_unknown(i): toktype = 2
         if tokenizer.is_control(i): toktype = 3
- 
+
         # TODO: How to determinate if a token is user defined?
         # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
         # if tokenizer.is_user_defined(i): toktype = 4
diff --git a/examples/gguf/gguf-llama-simple.cpp b/examples/gguf/gguf-llama-simple.cpp
index 0679240d31317..e59d1cfc19d74 100644
--- a/examples/gguf/gguf-llama-simple.cpp
+++ b/examples/gguf/gguf-llama-simple.cpp
@@ -74,7 +74,9 @@ int main(int argc, char ** argv) {
     // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
     // example, we will just stop the loop once this cache is full or once an end of stream is detected.
 
-    while (llama_get_kv_cache_token_count(ctx) < max_context_size) {
+    const int n_gen = std::min(32, max_context_size);
+
+    while (llama_get_kv_cache_token_count(ctx) < n_gen) {
         // evaluate the transformer
 
         if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
@@ -114,7 +116,6 @@ int main(int argc, char ** argv) {
 
         // push this new token for next evaluation
         tokens_list.push_back(new_token_id);
-
     }
 
     llama_free(ctx);
@@ -122,5 +123,7 @@ int main(int argc, char ** argv) {
 
     llama_backend_free();
 
+    fprintf(stderr, "\n\n");
+
     return 0;
 }
diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index b32367f301ee9..ad212d75280a9 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,5 +1,4 @@
 #include "ggml.h"
-#include "gguf-util.h"
 #include "gguf-llama.h"
 
 #include <cstdio>
@@ -21,133 +20,22 @@ static std::string to_string(const T & val) {
     return ss.str();
 }
 
-void gguf_ex_write_str(std::ofstream & fout, const std::string & val) {
-    const int32_t n = val.size();
-    fout.write((const char *) &n, sizeof(n));
-    fout.write(val.c_str(), n);
-}
-
-void gguf_ex_write_i32(std::ofstream & fout, int32_t val) {
-    fout.write((const char *) &val, sizeof(val));
-}
-
-void gguf_ex_write_u64(std::ofstream & fout, size_t val) {
-    fout.write((const char *) &val, sizeof(val));
-}
-
-template<typename T>
-void gguf_ex_write_val(std::ofstream & fout, const std::string & key, enum gguf_type type, const T & val) {
-    gguf_ex_write_str(fout, key);
-    fout.write((const char *) &type, sizeof(type));
-    fout.write((const char *) &val,  sizeof(val));
-
-    fprintf(stdout, "%s: write param: %s = %s\n", __func__, key.c_str(), to_string(val).c_str());
-}
-
-template<>
-void gguf_ex_write_val<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::string & val) {
-    gguf_ex_write_str(fout, key);
-    fout.write((const char *) &type, sizeof(type));
-
-    const int32_t n = val.size();
-    fout.write((const char *) &n, sizeof(n));
-    fout.write(val.c_str(), n);
-
-    fprintf(stdout, "%s: write param: %s = %s\n", __func__, key.c_str(), val.c_str());
-}
-
-template<typename T>
-void gguf_ex_write_arr(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::vector<T> & val) {
-    gguf_ex_write_str(fout, key);
-    {
-        const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-        fout.write((const char *) &tarr, sizeof(tarr));
-    }
-
-    const int32_t n = val.size();
-    fout.write((const char *) &type, sizeof(type));
-    fout.write((const char *) &n,    sizeof(n));
-    fout.write((const char *) val.data(), n * sizeof(T));
-
-    fprintf(stdout, "%s: write param: %s = [", __func__, key.c_str());
-    for (int i = 0; i < n; ++i) {
-        fprintf(stdout, "%s", to_string(val[i]).c_str());
-        if (i < n - 1) {
-            fprintf(stdout, ", ");
-        }
-    }
-    fprintf(stdout, "]\n");
-}
-
-template<>
-void gguf_ex_write_arr<std::string>(std::ofstream & fout, const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
-    gguf_ex_write_str(fout, key);
-    {
-        const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-        fout.write((const char *) &tarr, sizeof(tarr));
-    }
-
-    const int32_t n = val.size();
-    fout.write((const char *) &type, sizeof(type));
-    fout.write((const char *) &n,    sizeof(n));
-    for (int i = 0; i < n; ++i) {
-        const int32_t nstr = val[i].size();
-        fout.write((const char *) &nstr, sizeof(nstr));
-        fout.write(val[i].c_str(), nstr);
-    }
-
-    fprintf(stdout, "%s: write param: %s = [", __func__, key.c_str());
-    for (int i = 0; i < n; ++i) {
-        fprintf(stdout, "%s", val[i].c_str());
-        if (i < n - 1) {
-            fprintf(stdout, ", ");
-        }
-    }
-    fprintf(stdout, "]\n");
-}
-
 bool gguf_ex_write(const std::string & fname) {
-    std::ofstream fout(fname.c_str(), std::ios::binary);
-
-    {
-        const int32_t magic = GGUF_MAGIC;
-        fout.write((const char *) &magic, sizeof(magic));
-    }
-
-    {
-        const int32_t version = GGUF_VERSION;
-        fout.write((const char *) &version, sizeof(version));
-    }
-
-    // NOTE: these have to match the output below!
-    const int n_tensors = 10;
-    const int n_kv      = 12;
-
-    fout.write((const char*) &n_tensors, sizeof(n_tensors));
-    fout.write((const char*) &n_kv, sizeof(n_kv));
-
-    fprintf(stdout, "%s: write header\n", __func__);
-
-    // kv data
-    {
-        gguf_ex_write_val< uint8_t>(fout, "some.parameter.uint8",   GGUF_TYPE_UINT8,   0x12);
-        gguf_ex_write_val<  int8_t>(fout, "some.parameter.int8",    GGUF_TYPE_INT8,   -0x13);
-        gguf_ex_write_val<uint16_t>(fout, "some.parameter.uint16",  GGUF_TYPE_UINT16,  0x1234);
-        gguf_ex_write_val< int16_t>(fout, "some.parameter.int16",   GGUF_TYPE_INT16,  -0x1235);
-        gguf_ex_write_val<uint32_t>(fout, "some.parameter.uint32",  GGUF_TYPE_UINT32,  0x12345678);
-        gguf_ex_write_val< int32_t>(fout, "some.parameter.int32",   GGUF_TYPE_INT32,  -0x12345679);
-
-        gguf_ex_write_val<float>   (fout, "some.parameter.float32", GGUF_TYPE_FLOAT32, 0.123456789f);
-        gguf_ex_write_val<bool>    (fout, "some.parameter.bool",    GGUF_TYPE_BOOL,    true);
-
-        gguf_ex_write_val<std::string>(fout, "some.parameter.string",  GGUF_TYPE_STRING,  "hello world");
-
-        gguf_ex_write_arr<int16_t>    (fout, "some.parameter.arr.i16", GGUF_TYPE_INT16,   { 1, 2, 3, 4, });
-        gguf_ex_write_arr<float>      (fout, "some.parameter.arr.f32", GGUF_TYPE_FLOAT32, { 3.145f, 2.718f, 1.414f, });
-        gguf_ex_write_arr<std::string>(fout, "some.parameter.arr.str", GGUF_TYPE_STRING,  { "hello", "world", "!" });
-    }
-
-    uint64_t offset_tensor = 0;
+    struct gguf_context * ctx = gguf_init_empty();
+
+    gguf_set_val_u8  (ctx, "some.parameter.uint8",    0x12);
+    gguf_set_val_i8  (ctx, "some.parameter.int8",    -0x13);
+    gguf_set_val_u16 (ctx, "some.parameter.uint16",   0x1234);
+    gguf_set_val_i16 (ctx, "some.parameter.int16",   -0x1235);
+    gguf_set_val_u32 (ctx, "some.parameter.uint32",   0x12345678);
+    gguf_set_val_i32 (ctx, "some.parameter.int32",   -0x12345679);
+    gguf_set_val_f32 (ctx, "some.parameter.float32",  0.123456789f);
+    gguf_set_val_bool(ctx, "some.parameter.bool",     true);
+    gguf_set_val_str (ctx, "some.parameter.string",   "hello world");
+
+    gguf_set_arr_data(ctx, "some.parameter.arr.i16", GGUF_TYPE_INT16,   std::vector<int16_t>{ 1, 2, 3, 4, }.data(), 4);
+    gguf_set_arr_data(ctx, "some.parameter.arr.f32", GGUF_TYPE_FLOAT32, std::vector<float>{ 3.145f, 2.718f, 1.414f, }.data(), 3);
+    gguf_set_arr_str (ctx, "some.parameter.arr.str",                    std::vector<const char *>{ "hello", "world", "!" }.data(), 3);
 
     struct ggml_init_params params = {
         /*.mem_size   =*/ 128ull*1024ull*1024ull,
@@ -157,6 +45,8 @@ bool gguf_ex_write(const std::string & fname) {
 
     struct ggml_context * ctx_data = ggml_init(params);
 
+    const int n_tensors = 10;
+
     // tensor infos
     for (int i = 0; i < n_tensors; ++i) {
         const std::string name = "tensor_" + to_string(i);
@@ -178,58 +68,15 @@ bool gguf_ex_write(const std::string & fname) {
             }
         }
 
-        fprintf(stdout, "%s: tensor: %s, %d dims, ne = [", __func__, name.c_str(), n_dims);
-        for (int j = 0; j < 4; ++j) {
-            fprintf(stdout, "%s%3d", j == 0 ? "" : ", ", (int) cur->ne[j]);
-        }
-        fprintf(stdout, "], offset_tensor = %6" PRIu64 "\n", offset_tensor);
-
-        gguf_ex_write_str(fout, name);
-        gguf_ex_write_i32(fout, n_dims);
-        for (int j = 0; j < n_dims; ++j) {
-            gguf_ex_write_i32(fout, cur->ne[j]);
-        }
-        gguf_ex_write_i32(fout, cur->type);
-        gguf_ex_write_u64(fout, offset_tensor);
-
-        offset_tensor += GGML_PAD(ggml_nbytes(cur), GGUF_DEFAULT_ALIGNMENT);
-    }
-
-    const uint64_t offset_data = GGML_PAD((uint64_t) fout.tellp(), GGUF_DEFAULT_ALIGNMENT);
-
-    fprintf(stdout, "%s: data offset = %" PRIu64 "\n", __func__, offset_data);
-
-    {
-        const size_t pad = offset_data - fout.tellp();
-
-        for (size_t j = 0; j < pad; ++j) {
-            fout.put(0);
-        }
-    }
-
-    for (int i = 0; i < n_tensors; ++i) {
-        fprintf(stdout, "%s: writing tensor %d data\n", __func__, i);
-
-        const std::string name = "tensor_" + to_string(i);
-
-        struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name.c_str());
-
-        fout.write((const char *) cur->data, ggml_nbytes(cur));
-
-        {
-            const size_t pad = GGML_PAD(ggml_nbytes(cur), GGUF_DEFAULT_ALIGNMENT) - ggml_nbytes(cur);
-
-            for (size_t j = 0; j < pad; ++j) {
-                fout.put(0);
-            }
-        }
+        gguf_add_tensor(ctx, cur);
     }
 
-    fout.close();
+    gguf_write_to_file(ctx, fname.c_str(), false);
 
     fprintf(stdout, "%s: wrote file '%s;\n", __func__, fname.c_str());
 
     ggml_free(ctx_data);
+    gguf_free(ctx);
 
     return true;
 }
@@ -345,8 +192,16 @@ bool gguf_ex_read_1(const std::string & fname) {
 
             struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
 
-            fprintf(stdout, "%s: tensor[%d]: n_dims = %d, name = %s, data = %p\n",
-                    __func__, i, cur->n_dims, cur->name, cur->data);
+            fprintf(stdout, "%s: tensor[%d]: n_dims = %d, name = %s, data = %p\n", __func__, i, cur->n_dims, cur->name, cur->data);
+
+            // print first 10 elements
+            const float * data = (const float *) cur->data;
+
+            printf("%s data[:10] : ", name);
+            for (int j = 0; j < MIN(10, ggml_nelements(cur)); ++j) {
+                printf("%f ", data[j]);
+            }
+            printf("\n\n");
 
             // check data
             {
@@ -369,48 +224,6 @@ bool gguf_ex_read_1(const std::string & fname) {
     return true;
 }
 
-// read just the tensor info and mmap the data in user code
-bool gguf_ex_read_2(const std::string & fname) {
-    struct ggml_context * ctx_data = NULL;
-
-    struct gguf_init_params params = {
-        /*.no_alloc = */ true,
-        /*.ctx      = */ &ctx_data,
-    };
-
-    struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);
-
-    struct gguf_file file(fname.c_str(), "rb");
-    gguf_mmap data_mmap(&file, 0, false);
-
-    const int n_tensors = gguf_get_n_tensors(ctx);
-
-    for (int i = 0; i < n_tensors; ++i) {
-        const char * name   = gguf_get_tensor_name(ctx, i);
-        const size_t offset = gguf_get_data_offset(ctx) + gguf_get_tensor_offset(ctx, i);
-
-        struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
-
-        cur->data = static_cast<char *>(data_mmap.addr) + offset;
-
-        // print first 10 elements
-        const float * data = (const float *) cur->data;
-
-        printf("%s data[:10] : ", name);
-        for (int j = 0; j < MIN(10, ggml_nelements(cur)); ++j) {
-            printf("%f ", data[j]);
-        }
-        printf("\n\n");
-    }
-
-    fprintf(stdout, "%s: ctx_data size: %zu\n", __func__, ggml_get_mem_size(ctx_data));
-
-    ggml_free(ctx_data);
-    gguf_free(ctx);
-
-    return true;
-}
-
 int main(int argc, char ** argv) {
     if (argc < 3) {
         fprintf(stdout, "usage: %s data.gguf r|w\n", argv[0]);
@@ -427,7 +240,6 @@ int main(int argc, char ** argv) {
     } else if (mode == "r") {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
-        GGML_ASSERT(gguf_ex_read_2(fname) && "failed to read gguf file");
     } else if (mode == "q") {
         llama_model_quantize_params params = llama_model_quantize_default_params();
         llama_model_quantize(fname.c_str(), "quant.gguf", &params);
diff --git a/ggml.c b/ggml.c
index c8fa60328e2d7..77f57a3fdaa41 100644
--- a/ggml.c
+++ b/ggml.c
@@ -213,10 +213,10 @@ inline static void * ggml_aligned_malloc(size_t size) {
                 error_desc = "insufficient memory";
                 break;
         }
-        GGML_PRINT("%s: %s (attempted to allocate %6.2f MB)\n",
-            __func__, error_desc, size/(1024.0*1024.0));
+        GGML_PRINT("%s: %s (attempted to allocate %6.2f MB)\n", __func__, error_desc, size/(1024.0*1024.0));
         return NULL;
     }
+
     return aligned_memory;
 }
 #define GGML_ALIGNED_MALLOC(size)  ggml_aligned_malloc(size)
@@ -4109,7 +4109,11 @@ size_t ggml_nbytes(const struct ggml_tensor * tensor) {
     //
     // is enough, but just in case, adding the second part
 
-    return GGML_PAD(MAX(tensor->ne[3]*tensor->nb[3], (ggml_nelements(tensor)*GGML_TYPE_SIZE[tensor->type])/GGML_BLCK_SIZE[tensor->type]), GGML_MEM_ALIGN);
+    return MAX(tensor->ne[3]*tensor->nb[3], (ggml_nelements(tensor)*GGML_TYPE_SIZE[tensor->type])/GGML_BLCK_SIZE[tensor->type]);
+}
+
+size_t ggml_nbytes_pad(const struct ggml_tensor * tensor) {
+    return GGML_PAD(ggml_nbytes(tensor), GGML_MEM_ALIGN);
 }
 
 size_t ggml_nbytes_split(const struct ggml_tensor * tensor, int nrows_split) {
@@ -16899,7 +16903,7 @@ void ggml_graph_export(const struct ggml_cgraph * cgraph, const char * fname) {
     // compute size of intermediate results
     // TODO: does not take into account scratch buffers !!!!
     for (int i = 0; i < cgraph->n_nodes; ++i) {
-        size_eval += ggml_nbytes(cgraph->nodes[i]);
+        size_eval += ggml_nbytes_pad(cgraph->nodes[i]);
     }
 
     // print
@@ -18579,6 +18583,19 @@ static const size_t GGUF_TYPE_SIZE[GGUF_TYPE_COUNT] = {
 };
 static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
 
+static const char * GGUF_TYPE_NAME[GGUF_TYPE_COUNT] = {
+    [GGUF_TYPE_UINT8]   = "uint8",
+    [GGUF_TYPE_INT8]    = "int8",
+    [GGUF_TYPE_UINT16]  = "uint16",
+    [GGUF_TYPE_INT16]   = "int16",
+    [GGUF_TYPE_UINT32]  = "uint32",
+    [GGUF_TYPE_INT32]   = "int32",
+    [GGUF_TYPE_FLOAT32] = "float32",
+    [GGUF_TYPE_BOOL]    = "bool",
+    [GGUF_TYPE_STRING]  = "string",
+    [GGUF_TYPE_ARRAY]   = "array",
+};
+
 union gguf_value {
     uint8_t  uint8;
     int8_t   int8;
@@ -18613,8 +18630,6 @@ struct gguf_header {
     uint32_t version;
     uint32_t n_tensors;
     uint32_t n_kv;
-
-    struct gguf_kv * kv;
 };
 
 struct gguf_tensor_info {
@@ -18622,44 +18637,69 @@ struct gguf_tensor_info {
 
     uint32_t n_dims;
     uint32_t ne[GGML_MAX_DIMS];
-    uint32_t n_elms; // TODO: is this needed?
 
     enum ggml_type type;
 
     uint64_t offset; // offset from start of `data`, must be a multiple of `ALIGNMENT`
+
+    // for writing API
+    const void * data;
+    size_t size;
 };
 
 struct gguf_context {
-    struct gguf_header        header;
+    struct gguf_header header;
+
+    struct gguf_kv          * kv;
     struct gguf_tensor_info * infos;
 
     size_t alignment;
     size_t offset;    // offset of `data` from beginning of file
-    size_t size_data; // size of `data` in bytes
+    size_t size;      // size of `data` in bytes
 
     //uint8_t * padding;
-    uint8_t * data;
+    void * data;
 };
 
-static bool gguf_fread_el(void * dst, size_t size, FILE * file, size_t * offset) {
+static bool gguf_fread_el(FILE * file, void * dst, size_t size, size_t * offset) {
     const size_t n = fread(dst, 1, size, file);
     *offset += n;
     return n == size;
 }
 
-static bool gguf_fread_str(struct gguf_str * p, FILE * file, size_t * offset) {
+static bool gguf_fread_str(FILE * file, struct gguf_str * p, size_t * offset) {
     p->n    = 0;
     p->data = NULL;
 
     bool ok = true;
 
     // TODO: how to avoid mallocs for strings?
-    ok = ok && gguf_fread_el(&p->n,    sizeof(p->n), file, offset); p->data = calloc(p->n + 1, 1);
-    ok = ok && gguf_fread_el( p->data, p->n,         file, offset);
+    ok = ok && gguf_fread_el(file, &p->n,    sizeof(p->n), offset); p->data = calloc(p->n + 1, 1);
+    ok = ok && gguf_fread_el(file,  p->data, p->n,         offset);
 
     return ok;
 }
 
+struct gguf_context * gguf_init_empty(void) {
+    struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
+
+    ctx->header.magic     = GGUF_MAGIC;
+    ctx->header.version   = GGUF_VERSION;
+    ctx->header.n_tensors = 0;
+    ctx->header.n_kv      = 0;
+
+    ctx->kv    = NULL;
+    ctx->infos = NULL;
+
+    ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
+    ctx->offset    = 0;
+    ctx->size      = 0;
+
+    ctx->data = NULL;
+
+    return ctx;
+}
+
 struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_params params) {
     FILE * file = fopen(fname, "rb");
     if (!file) {
@@ -18673,7 +18713,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     // check the magic before making allocations
     {
-        gguf_fread_el(&magic, sizeof(magic), file, &offset);
+        gguf_fread_el(file, &magic, sizeof(magic), &offset);
 
         if (magic != GGUF_MAGIC) {
             fprintf(stderr, "%s: invalid magic number %08x\n", __func__, magic);
@@ -18689,14 +18729,14 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
     // read the header
     {
         ctx->header.magic = magic;
-        ctx->header.kv    = NULL;
 
+        ctx->kv    = NULL;
         ctx->infos = NULL;
         ctx->data  = NULL;
 
-        ok = ok && gguf_fread_el(&ctx->header.version,   sizeof(ctx->header.version),   file, &offset);
-        ok = ok && gguf_fread_el(&ctx->header.n_tensors, sizeof(ctx->header.n_tensors), file, &offset);
-        ok = ok && gguf_fread_el(&ctx->header.n_kv,      sizeof(ctx->header.n_kv),      file, &offset);
+        ok = ok && gguf_fread_el(file, &ctx->header.version,   sizeof(ctx->header.version),   &offset);
+        ok = ok && gguf_fread_el(file, &ctx->header.n_tensors, sizeof(ctx->header.n_tensors), &offset);
+        ok = ok && gguf_fread_el(file, &ctx->header.n_kv,      sizeof(ctx->header.n_kv),      &offset);
 
         if (!ok) {
             fprintf(stderr, "%s: failed to read header\n", __func__);
@@ -18708,33 +18748,33 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     // read the kv pairs
     {
-        ctx->header.kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
+        ctx->kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
 
         for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
-            struct gguf_kv * kv = &ctx->header.kv[i];
+            struct gguf_kv * kv = &ctx->kv[i];
 
             //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
 
-            ok = ok && gguf_fread_str(&kv->key,                          file, &offset);
-          //ok = ok && gguf_fread_el (&kv->n_bytes, sizeof(kv->n_bytes), file, &offset);
-            ok = ok && gguf_fread_el (&kv->type,    sizeof(kv->type),    file, &offset);
+            ok = ok && gguf_fread_str(file, &kv->key,                          &offset);
+          //ok = ok && gguf_fread_el (file, &kv->n_bytes, sizeof(kv->n_bytes), &offset);
+            ok = ok && gguf_fread_el (file, &kv->type,    sizeof(kv->type),    &offset);
 
             //fprintf(stderr, "%s: reading kv with key %s\n", __func__, kv->key.data);
 
             switch (kv->type) {
-                case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (&kv->value.uint8,   sizeof(kv->value.uint8),   file, &offset); break;
-                case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (&kv->value.int8,    sizeof(kv->value.int8),    file, &offset); break;
-                case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (&kv->value.uint16,  sizeof(kv->value.uint16),  file, &offset); break;
-                case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (&kv->value.int16,   sizeof(kv->value.int16),   file, &offset); break;
-                case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (&kv->value.uint32,  sizeof(kv->value.uint32),  file, &offset); break;
-                case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (&kv->value.int32,   sizeof(kv->value.int32),   file, &offset); break;
-                case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (&kv->value.float32, sizeof(kv->value.float32), file, &offset); break;
-                case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (&kv->value.bool_,   sizeof(kv->value.bool_),   file, &offset); break;
-                case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(&kv->value.str,                                file, &offset); break;
+                case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (file, &kv->value.uint8,   sizeof(kv->value.uint8),   &offset); break;
+                case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (file, &kv->value.int8,    sizeof(kv->value.int8),    &offset); break;
+                case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (file, &kv->value.uint16,  sizeof(kv->value.uint16),  &offset); break;
+                case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (file, &kv->value.int16,   sizeof(kv->value.int16),   &offset); break;
+                case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (file, &kv->value.uint32,  sizeof(kv->value.uint32),  &offset); break;
+                case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (file, &kv->value.int32,   sizeof(kv->value.int32),   &offset); break;
+                case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (file, &kv->value.float32, sizeof(kv->value.float32), &offset); break;
+                case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (file, &kv->value.bool_,   sizeof(kv->value.bool_),   &offset); break;
+                case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(file, &kv->value.str,                                &offset); break;
                 case GGUF_TYPE_ARRAY:
                     {
-                        ok = ok && gguf_fread_el(&kv->value.arr.type, sizeof(kv->value.arr.type), file, &offset);
-                        ok = ok && gguf_fread_el(&kv->value.arr.n,    sizeof(kv->value.arr.n),    file, &offset);
+                        ok = ok && gguf_fread_el(file, &kv->value.arr.type, sizeof(kv->value.arr.type), &offset);
+                        ok = ok && gguf_fread_el(file, &kv->value.arr.n,    sizeof(kv->value.arr.n),    &offset);
 
                         switch (kv->value.arr.type) {
                             case GGUF_TYPE_UINT8:
@@ -18747,17 +18787,17 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
                             case GGUF_TYPE_BOOL:
                                 {
                                     kv->value.arr.data = malloc(kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type]);
-                                    ok = ok && gguf_fread_el(kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type], file, &offset);
+                                    ok = ok && gguf_fread_el(file, kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type], &offset);
                                 } break;
                             case GGUF_TYPE_STRING:
                                 {
                                     kv->value.arr.data = malloc(kv->value.arr.n * sizeof(struct gguf_str));
                                     for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
-                                        ok = ok && gguf_fread_str(&((struct gguf_str *) kv->value.arr.data)[j], file, &offset);
+                                        ok = ok && gguf_fread_str(file, &((struct gguf_str *) kv->value.arr.data)[j], &offset);
                                     }
                                 } break;
                             case GGUF_TYPE_ARRAY:
-                            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
+                            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type"); break;
                         };
                     } break;
                 case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
@@ -18787,14 +18827,13 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
                 info->ne[j] = 1;
             }
 
-            ok = ok && gguf_fread_str(&info->name,                          file, &offset);
-            ok = ok && gguf_fread_el (&info->n_dims, sizeof(info->n_dims),  file, &offset);
+            ok = ok && gguf_fread_str(file, &info->name,                          &offset);
+            ok = ok && gguf_fread_el (file, &info->n_dims, sizeof(info->n_dims),  &offset);
             for (uint32_t j = 0; j < info->n_dims; ++j) {
-                ok = ok && gguf_fread_el(&info->ne[j], sizeof(info->ne[j]), file, &offset);
+                ok = ok && gguf_fread_el(file, &info->ne[j], sizeof(info->ne[j]), &offset);
             }
-          //ok = ok && gguf_fread_el (&info->n_elms, sizeof(info->n_elms),  file, &offset);
-            ok = ok && gguf_fread_el (&info->type,   sizeof(info->type),    file, &offset);
-            ok = ok && gguf_fread_el (&info->offset, sizeof(info->offset),  file, &offset);
+            ok = ok && gguf_fread_el (file, &info->type,   sizeof(info->type),    &offset);
+            ok = ok && gguf_fread_el (file, &info->offset, sizeof(info->offset),  &offset);
 
             if (!ok) {
                 fprintf(stderr, "%s: failed to read tensor info\n", __func__);
@@ -18827,8 +18866,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
     // compute the total size of the data section, taking into account the alignment
     {
-
-        ctx->size_data = 0;
+        ctx->size = 0;
         for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
             struct gguf_tensor_info * info = &ctx->infos[i];
 
@@ -18848,7 +18886,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
 
             const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
 
-            ctx->size_data += GGML_PAD(size_cur, ctx->alignment);
+            ctx->size += GGML_PAD(size_cur, ctx->alignment);
         }
     }
 
@@ -18862,7 +18900,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         const size_t mem_size =
             params.no_alloc ?
             (ctx->header.n_tensors    )*ggml_tensor_overhead() :
-            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() + ctx->size_data;
+            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() + ctx->size;
 
         struct ggml_init_params pdata = {
             .mem_size   = mem_size,
@@ -18877,12 +18915,12 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
         struct ggml_tensor * data = NULL;
 
         if (params.no_alloc == false) {
-            data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size_data);
+            data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size);
 
             ok = ok && data != NULL;
 
             // read the binary blob with the tensor data
-            ok = ok && gguf_fread_el(data->data, ctx->size_data, file, &offset);
+            ok = ok && gguf_fread_el(file, data->data, ctx->size, &offset);
 
             if (!ok) {
                 fprintf(stderr, "%s: failed to read tensor data\n", __func__);
@@ -18944,10 +18982,10 @@ void gguf_free(struct gguf_context * ctx) {
         return;
     }
 
-    if (ctx->header.kv) {
+    if (ctx->kv) {
         // free string memory - not great..
         for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
-            struct gguf_kv * kv = &ctx->header.kv[i];
+            struct gguf_kv * kv = &ctx->kv[i];
 
             if (kv->key.data) {
                 free(kv->key.data);
@@ -18974,7 +19012,7 @@ void gguf_free(struct gguf_context * ctx) {
             }
         }
 
-        GGML_ALIGNED_FREE(ctx->header.kv);
+        GGML_ALIGNED_FREE(ctx->kv);
     }
 
     if (ctx->infos) {
@@ -18992,6 +19030,10 @@ void gguf_free(struct gguf_context * ctx) {
     GGML_ALIGNED_FREE(ctx);
 }
 
+const char * gguf_type_name(enum gguf_type type) {
+    return GGUF_TYPE_NAME[type];
+}
+
 int gguf_get_version(struct gguf_context * ctx) {
     return ctx->header.version;
 }
@@ -19014,8 +19056,9 @@ int gguf_get_n_kv(struct gguf_context * ctx) {
 
 int gguf_find_key(struct gguf_context * ctx, const char * key) {
     // return -1 if key not found
+    int keyfound = -1;
+
     const int n_kv = gguf_get_n_kv(ctx);
-    int keyfound   = -1;
 
     for (int i = 0; i < n_kv; ++i) {
         if (strcmp(key, gguf_get_key(ctx, i)) == 0) {
@@ -19028,71 +19071,87 @@ int gguf_find_key(struct gguf_context * ctx, const char * key) {
 }
 
 const char * gguf_get_key(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].key.data;
+    return ctx->kv[i].key.data;
 }
 
 enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].type;
+    return ctx->kv[i].type;
 }
 
 enum gguf_type gguf_get_arr_type(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.arr.type;
+    return ctx->kv[i].value.arr.type;
+}
+
+const void * gguf_get_arr_data(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.arr.data;
 }
 
 const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
-    struct gguf_kv * kv = &ctx->header.kv[key_id];
+    struct gguf_kv * kv = &ctx->kv[key_id];
     struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];
     return str->data;
 }
 
-float gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i) {
-    return ((float *) ctx->header.kv[key_id].value.arr.data)[i];
-}
-
 int gguf_get_arr_n(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.arr.n;
+    return ctx->kv[i].value.arr.n;
 }
 
 uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.uint8;
+    return ctx->kv[i].value.uint8;
 }
 
 int8_t gguf_get_val_i8(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.int8;
+    return ctx->kv[i].value.int8;
 }
 
 uint16_t gguf_get_val_u16(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.uint16;
+    return ctx->kv[i].value.uint16;
 }
 
 int16_t gguf_get_val_i16(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.int16;
+    return ctx->kv[i].value.int16;
 }
 
 uint32_t gguf_get_val_u32(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.uint32;
+    return ctx->kv[i].value.uint32;
 }
 
 int32_t gguf_get_val_i32(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.int32;
+    return ctx->kv[i].value.int32;
 }
 
 float gguf_get_val_f32(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.float32;
+    return ctx->kv[i].value.float32;
 }
 
 bool gguf_get_val_bool(struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.bool_;
+    return ctx->kv[i].value.bool_;
 }
 
 const char * gguf_get_val_str (struct gguf_context * ctx, int i) {
-    return ctx->header.kv[i].value.str.data;
+    return ctx->kv[i].value.str.data;
 }
 
 int gguf_get_n_tensors(struct gguf_context * ctx) {
     return ctx->header.n_tensors;
 }
 
+int gguf_find_tensor(struct gguf_context * ctx, const char * name) {
+    // return -1 if tensor not found
+    int tensorfound = -1;
+
+    const int n_tensors = gguf_get_n_tensors(ctx);
+
+    for (int i = 0; i < n_tensors; ++i) {
+        if (strcmp(name, gguf_get_tensor_name(ctx, i)) == 0) {
+            tensorfound = i;
+            break;
+        }
+    }
+
+    return tensorfound;
+}
+
 size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i) {
     return ctx->infos[i].offset;
 }
@@ -19101,6 +19160,400 @@ char * gguf_get_tensor_name(struct gguf_context * ctx, int i) {
     return ctx->infos[i].name.data;
 }
 
+// returns the index
+static int gguf_get_or_add_key(struct gguf_context * ctx, const char * key) {
+    const int idx = gguf_find_key(ctx, key);
+    if (idx >= 0) {
+        return idx;
+    }
+
+    const int n_kv = gguf_get_n_kv(ctx);
+
+    ctx->kv = realloc(ctx->kv, (n_kv + 1) * sizeof(struct gguf_kv));
+    ctx->kv[n_kv].key.n    = strlen(key) + 1;
+    ctx->kv[n_kv].key.data = strdup(key);
+    ctx->header.n_kv++;
+
+    return n_kv;
+}
+
+void gguf_set_val_u8(struct gguf_context * ctx, const char * key, uint8_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_UINT8;
+    ctx->kv[idx].value.uint8 = val;
+}
+
+void gguf_set_val_i8(struct gguf_context * ctx, const char * key, int8_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type       = GGUF_TYPE_INT8;
+    ctx->kv[idx].value.int8 = val;
+}
+
+void gguf_set_val_u16(struct gguf_context * ctx, const char * key, uint16_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type         = GGUF_TYPE_UINT16;
+    ctx->kv[idx].value.uint16 = val;
+}
+
+void gguf_set_val_i16(struct gguf_context * ctx, const char * key, int16_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_INT16;
+    ctx->kv[idx].value.int16 = val;
+}
+
+void gguf_set_val_u32(struct gguf_context * ctx, const char * key, uint32_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type         = GGUF_TYPE_UINT32;
+    ctx->kv[idx].value.uint32 = val;
+}
+
+void gguf_set_val_i32(struct gguf_context * ctx, const char * key, int32_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_INT32;
+    ctx->kv[idx].value.int32 = val;
+}
+
+void gguf_set_val_f32(struct gguf_context * ctx, const char * key, float val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type          = GGUF_TYPE_FLOAT32;
+    ctx->kv[idx].value.float32 = val;
+}
+
+void gguf_set_val_bool(struct gguf_context * ctx, const char * key, bool val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_BOOL;
+    ctx->kv[idx].value.bool_ = val;
+}
+
+void gguf_set_val_str(struct gguf_context * ctx, const char * key, const char * val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_STRING;
+    ctx->kv[idx].value.str.n    = strlen(val) + 1;
+    ctx->kv[idx].value.str.data = strdup(val);
+}
+
+void gguf_set_arr_data(struct gguf_context * ctx, const char * key, enum gguf_type type, const void * data, int n) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_ARRAY;
+    ctx->kv[idx].value.arr.type = type;
+    ctx->kv[idx].value.arr.n    = n;
+    ctx->kv[idx].value.arr.data = malloc(n*GGUF_TYPE_SIZE[type]);
+    memcpy(ctx->kv[idx].value.arr.data, data, n*GGUF_TYPE_SIZE[type]);
+}
+
+void gguf_set_arr_str(struct gguf_context * ctx, const char * key, const char ** data, int n) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_ARRAY;
+    ctx->kv[idx].value.arr.type = GGUF_TYPE_STRING;
+    ctx->kv[idx].value.arr.n    = n;
+    ctx->kv[idx].value.arr.data = malloc(n*sizeof(struct gguf_str));
+    for (int i = 0; i < n; i++) {
+        struct gguf_str * str = &((struct gguf_str *)ctx->kv[idx].value.arr.data)[i];
+        str->n    = strlen(data[i]) + 1;
+        str->data = strdup(data[i]);
+    }
+}
+
+// set or add KV pairs from another context
+void gguf_set_kv(struct gguf_context * ctx, struct gguf_context * src) {
+    for (uint32_t i = 0; i < src->header.n_kv; i++) {
+        switch (src->kv[i].type) {
+            case GGUF_TYPE_UINT8:   gguf_set_val_u8  (ctx, src->kv[i].key.data, src->kv[i].value.uint8);    break;
+            case GGUF_TYPE_INT8:    gguf_set_val_i8  (ctx, src->kv[i].key.data, src->kv[i].value.int8);     break;
+            case GGUF_TYPE_UINT16:  gguf_set_val_u16 (ctx, src->kv[i].key.data, src->kv[i].value.uint16);   break;
+            case GGUF_TYPE_INT16:   gguf_set_val_i16 (ctx, src->kv[i].key.data, src->kv[i].value.int16);    break;
+            case GGUF_TYPE_UINT32:  gguf_set_val_u32 (ctx, src->kv[i].key.data, src->kv[i].value.uint32);   break;
+            case GGUF_TYPE_INT32:   gguf_set_val_i32 (ctx, src->kv[i].key.data, src->kv[i].value.int32);    break;
+            case GGUF_TYPE_FLOAT32: gguf_set_val_f32 (ctx, src->kv[i].key.data, src->kv[i].value.float32);  break;
+            case GGUF_TYPE_BOOL:    gguf_set_val_bool(ctx, src->kv[i].key.data, src->kv[i].value.bool_);    break;
+            case GGUF_TYPE_STRING:  gguf_set_val_str (ctx, src->kv[i].key.data, src->kv[i].value.str.data); break;
+            case GGUF_TYPE_ARRAY:
+                {
+                    if (src->kv[i].value.arr.type == GGUF_TYPE_STRING) {
+                        const char ** data = malloc(src->kv[i].value.arr.n*sizeof(char *));
+                        for (uint32_t j = 0; j < src->kv[i].value.arr.n; j++) {
+                            data[j] = ((struct gguf_str *)src->kv[i].value.arr.data)[j].data;
+                        }
+                        gguf_set_arr_str(ctx, src->kv[i].key.data, data, src->kv[i].value.arr.n);
+                        free(data);
+                    } if (src->kv[i].value.arr.type == GGUF_TYPE_ARRAY) {
+                        GGML_ASSERT(false && "nested arrays not supported");
+                    } else {
+                        gguf_set_arr_data(ctx, src->kv[i].key.data, src->kv[i].value.arr.type, src->kv[i].value.arr.data, src->kv[i].value.arr.n);
+                    }
+                } break;
+            case GGUF_TYPE_COUNT:  GGML_ASSERT(false && "invalid type"); break;
+        }
+    }
+}
+
+void gguf_add_tensor(
+             struct gguf_context * ctx,
+        const struct ggml_tensor * tensor) {
+    const int idx = ctx->header.n_tensors;
+    ctx->infos = realloc(ctx->infos, (idx + 1)*sizeof(struct gguf_tensor_info));
+
+    ctx->infos[idx].name.n    = strlen(tensor->name) + 1;
+    ctx->infos[idx].name.data = strdup(tensor->name);
+
+    for (int i = 0; i < GGML_MAX_DIMS; ++i) {
+        ctx->infos[idx].ne[i] = 1;
+    }
+
+    ctx->infos[idx].n_dims = tensor->n_dims;
+    for (int i = 0; i < tensor->n_dims; i++) {
+        ctx->infos[idx].ne[i] = tensor->ne[i];
+    }
+
+    ctx->infos[idx].type   = tensor->type;
+    ctx->infos[idx].offset = 0;
+    ctx->infos[idx].data   = tensor->data;
+    ctx->infos[idx].size   = ggml_nbytes(tensor);
+
+    if (ctx->header.n_tensors > 0) {
+        ctx->infos[idx].offset = ctx->infos[idx - 1].offset + GGML_PAD(ctx->infos[idx - 1].size, ctx->alignment);
+    }
+
+    ctx->header.n_tensors++;
+}
+
+void gguf_set_tensor_type(struct gguf_context * ctx, const char * name, enum ggml_type type) {
+    const int idx = gguf_find_tensor(ctx, name);
+    if (idx < 0) {
+        GGML_ASSERT(false && "tensor not found");
+    }
+
+    ctx->infos[idx].type = type;
+}
+
+void gguf_set_tensor_data(struct gguf_context * ctx, const char * name, const void * data, size_t size) {
+    const int idx = gguf_find_tensor(ctx, name);
+    if (idx < 0) {
+        GGML_ASSERT(false && "tensor not found");
+    }
+
+    ctx->infos[idx].data = data;
+    ctx->infos[idx].size = size;
+
+    // update offsets
+    for (uint32_t i = idx + 1; i < ctx->header.n_tensors; ++i) {
+        ctx->infos[i].offset = ctx->infos[i - 1].offset + GGML_PAD(ctx->infos[i - 1].size, ctx->alignment);
+    }
+}
+
+//static void gguf_fwrite_str(FILE * file, const struct gguf_str * val) {
+//    fwrite(&val->n,   sizeof(val->n),    1, file);
+//    fwrite(val->data, sizeof(char), val->n, file);
+//}
+//
+//static void gguf_fwrite_el(FILE * file, const void * val, size_t size) {
+//    fwrite(val, sizeof(char), size, file);
+//}
+
+struct gguf_buf {
+    void * data;
+    size_t size;
+    size_t offset;
+};
+
+static struct gguf_buf gguf_buf_init(size_t size) {
+    struct gguf_buf buf = {
+        /*buf.data   =*/ size == 0 ? NULL : malloc(size),
+        /*buf.size   =*/ size,
+        /*buf.offset =*/ 0,
+    };
+
+    return buf;
+}
+
+static void gguf_buf_free(struct gguf_buf buf) {
+    if (buf.data) {
+        free(buf.data);
+    }
+}
+
+static void gguf_buf_grow(struct gguf_buf * buf, size_t size) {
+    if (buf->offset + size > buf->size) {
+        buf->size = 1.5*(buf->offset + size);
+        if (buf->data) {
+            buf->data = realloc(buf->data, buf->size);
+        }
+    }
+}
+
+static void gguf_bwrite_str(struct gguf_buf * buf, const struct gguf_str * val) {
+    gguf_buf_grow(buf, sizeof(val->n) + val->n);
+
+    buf->data && memcpy((char *) buf->data + buf->offset, &val->n, sizeof(val->n));
+    buf->offset += sizeof(val->n);
+
+    buf->data && memcpy((char *) buf->data + buf->offset, val->data, val->n);
+    buf->offset += val->n;
+}
+
+static void gguf_bwrite_el(struct gguf_buf * buf, const void * val, size_t el_size) {
+    gguf_buf_grow(buf, el_size);
+
+    buf->data && memcpy((char *) buf->data + buf->offset, val, el_size);
+    buf->offset += el_size;
+}
+
+static void gguf_write_to_buf(struct gguf_context * ctx, struct gguf_buf * buf, bool only_meta) {
+    // write header
+    gguf_bwrite_el(buf, &ctx->header.magic,     sizeof(ctx->header.magic));
+    gguf_bwrite_el(buf, &ctx->header.version,   sizeof(ctx->header.version));
+    gguf_bwrite_el(buf, &ctx->header.n_tensors, sizeof(ctx->header.n_tensors));
+    gguf_bwrite_el(buf, &ctx->header.n_kv,      sizeof(ctx->header.n_kv));
+
+    // write key-value pairs
+    for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+        struct gguf_kv * kv = &ctx->kv[i];
+
+        gguf_bwrite_str(buf, &kv->key);
+        gguf_bwrite_el (buf, &kv->type, sizeof(kv->type));
+
+        switch (kv->type) {
+            case GGUF_TYPE_UINT8:   gguf_bwrite_el( buf, &kv->value.uint8,   sizeof(kv->value.uint8)  ); break;
+            case GGUF_TYPE_INT8:    gguf_bwrite_el (buf, &kv->value.int8,    sizeof(kv->value.int8)   ); break;
+            case GGUF_TYPE_UINT16:  gguf_bwrite_el (buf, &kv->value.uint16,  sizeof(kv->value.uint16) ); break;
+            case GGUF_TYPE_INT16:   gguf_bwrite_el (buf, &kv->value.int16,   sizeof(kv->value.int16)  ); break;
+            case GGUF_TYPE_UINT32:  gguf_bwrite_el (buf, &kv->value.uint32,  sizeof(kv->value.uint32) ); break;
+            case GGUF_TYPE_INT32:   gguf_bwrite_el (buf, &kv->value.int32,   sizeof(kv->value.int32)  ); break;
+            case GGUF_TYPE_FLOAT32: gguf_bwrite_el (buf, &kv->value.float32, sizeof(kv->value.float32)); break;
+            case GGUF_TYPE_BOOL:    gguf_bwrite_el (buf, &kv->value.bool_,   sizeof(kv->value.bool_)  ); break;
+            case GGUF_TYPE_STRING:  gguf_bwrite_str(buf, &kv->value.str                               ); break;
+            case GGUF_TYPE_ARRAY:
+                {
+                    gguf_bwrite_el(buf, &kv->value.arr.type, sizeof(kv->value.arr.type));
+                    gguf_bwrite_el(buf, &kv->value.arr.n,    sizeof(kv->value.arr.n)   );
+
+                    switch (kv->value.arr.type) {
+                        case GGUF_TYPE_UINT8:
+                        case GGUF_TYPE_INT8:
+                        case GGUF_TYPE_UINT16:
+                        case GGUF_TYPE_INT16:
+                        case GGUF_TYPE_UINT32:
+                        case GGUF_TYPE_INT32:
+                        case GGUF_TYPE_FLOAT32:
+                        case GGUF_TYPE_BOOL:
+                            {
+                                gguf_bwrite_el(buf, kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type]);
+                            } break;
+                        case GGUF_TYPE_STRING:
+                            {
+                                for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                                    gguf_bwrite_str(buf, &((struct gguf_str *) kv->value.arr.data)[j]);
+                                }
+                            } break;
+                        case GGUF_TYPE_ARRAY:
+                        case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type"); break;
+                    };
+                } break;
+            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
+        };
+    }
+
+    // write tensor infos
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        gguf_bwrite_str(buf, &info->name);
+        gguf_bwrite_el (buf, &info->n_dims, sizeof(info->n_dims));
+        for (uint32_t j = 0; j < info->n_dims; ++j) {
+            gguf_bwrite_el(buf, &info->ne[j], sizeof(info->ne[j]));
+        }
+        gguf_bwrite_el(buf, &info->type,   sizeof(info->type));
+        gguf_bwrite_el(buf, &info->offset, sizeof(info->offset));
+    }
+
+    // we require the data section to be aligned, so take into account any padding
+    {
+        const size_t offset     = buf->offset;
+        const size_t offset_pad = GGML_PAD(offset, ctx->alignment);
+
+        if (offset_pad != offset) {
+            uint8_t pad = 0;
+            for (size_t i = 0; i < offset_pad - offset; ++i) {
+                gguf_bwrite_el(buf, &pad, sizeof(pad));
+            }
+        }
+    }
+
+    if (only_meta) {
+        return;
+    }
+
+    size_t offset = 0;
+
+    // write tensor data
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        const size_t size     = info->size;
+        const size_t size_pad = GGML_PAD(size, ctx->alignment);
+
+        gguf_bwrite_el(buf, info->data, size);
+
+        if (size_pad != size) {
+            uint8_t pad = 0;
+            for (size_t j = 0; j < size_pad - size; ++j) {
+                gguf_bwrite_el(buf, &pad, sizeof(pad));
+            }
+        }
+
+        GGML_ASSERT(offset == info->offset);
+
+        offset += size_pad;
+    }
+}
+
+void gguf_write_to_file(struct gguf_context * ctx, const char * fname, bool only_meta) {
+    FILE * file = fopen(fname, "wb");
+    if (!file) {
+        GGML_ASSERT(false && "failed to open file for writing");
+    }
+
+    struct gguf_buf buf = gguf_buf_init(16*1024);
+
+    gguf_write_to_buf(ctx, &buf, only_meta);
+
+    fwrite(buf.data, 1, buf.offset, file);
+
+    gguf_buf_free(buf);
+
+    fclose(file);
+}
+
+size_t gguf_get_meta_size(struct gguf_context * ctx) {
+    // no allocs - only compute size
+    struct gguf_buf buf = gguf_buf_init(0);
+
+    gguf_write_to_buf(ctx, &buf, true);
+
+    return buf.offset;
+}
+
+void gguf_get_meta_data(struct gguf_context * ctx, void * data) {
+    struct gguf_buf buf = gguf_buf_init(16*1024);
+
+    gguf_write_to_buf(ctx, &buf, true);
+
+    memcpy(data, buf.data, buf.offset);
+
+    gguf_buf_free(buf);
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 
 int ggml_cpu_has_avx(void) {
diff --git a/ggml.h b/ggml.h
index fb3db10e2cedb..ad12c133ee12d 100644
--- a/ggml.h
+++ b/ggml.h
@@ -566,6 +566,7 @@ extern "C" {
     GGML_API int64_t ggml_nelements   (const struct ggml_tensor * tensor);
     GGML_API int64_t ggml_nrows       (const struct ggml_tensor * tensor);
     GGML_API size_t  ggml_nbytes      (const struct ggml_tensor * tensor);
+    GGML_API size_t  ggml_nbytes_pad  (const struct ggml_tensor * tensor); // same as ggml_nbytes() but padded to GGML_MEM_ALIGN
     GGML_API size_t  ggml_nbytes_split(const struct ggml_tensor * tensor, int nrows_split);
 
     GGML_API int     ggml_blck_size (enum ggml_type type);
@@ -1498,7 +1499,6 @@ extern "C" {
             struct ggml_context * ctx,
             struct ggml_tensor  * tensor);
 
-
     GGML_API void ggml_build_forward_expand(struct ggml_cgraph * cgraph, struct ggml_tensor * tensor);
 
     GGML_API struct ggml_cgraph ggml_build_forward (struct ggml_tensor * tensor);
@@ -1711,7 +1711,6 @@ extern "C" {
     // gguf
     //
 
-    // TODO: can be removed if the API is extended for writing
     enum gguf_type {
         GGUF_TYPE_UINT8   = 0,
         GGUF_TYPE_INT8    = 1,
@@ -1735,25 +1734,27 @@ extern "C" {
         struct ggml_context ** ctx;
     };
 
+    GGML_API struct gguf_context * gguf_init_empty(void);
     GGML_API struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_params params);
     //GGML_API struct gguf_context * gguf_init_from_buffer(..);
-    GGML_API void                  gguf_free(struct gguf_context * ctx);
+
+    GGML_API void gguf_free(struct gguf_context * ctx);
+
+    GGML_API const char * gguf_type_name(enum gguf_type type);
 
     GGML_API int    gguf_get_version    (struct gguf_context * ctx);
     GGML_API size_t gguf_get_alignment  (struct gguf_context * ctx);
     GGML_API size_t gguf_get_data_offset(struct gguf_context * ctx);
     GGML_API void * gguf_get_data       (struct gguf_context * ctx);
 
-    GGML_API int            gguf_get_n_kv(struct gguf_context * ctx);
-    GGML_API int            gguf_find_key(struct gguf_context * ctx, const char * key);
-    GGML_API const char *   gguf_get_key (struct gguf_context * ctx, int i);
-    GGML_API enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
-    GGML_API enum gguf_type gguf_get_arr_type (struct gguf_context * ctx, int i);
-    GGML_API void           gguf_get_val (struct gguf_context * ctx, int i, void * val);
+    GGML_API int          gguf_get_n_kv(struct gguf_context * ctx);
+    GGML_API int          gguf_find_key(struct gguf_context * ctx, const char * key);
+    GGML_API const char * gguf_get_key (struct gguf_context * ctx, int i);
 
-    GGML_API const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i);
-    GGML_API float        gguf_get_arr_f32(struct gguf_context * ctx, int key_id, int i);
+    GGML_API enum gguf_type gguf_get_kv_type (struct gguf_context * ctx, int i);
+    GGML_API enum gguf_type gguf_get_arr_type(struct gguf_context * ctx, int i);
 
+    // results are undefined if the wrong type is used for the key
     GGML_API uint8_t      gguf_get_val_u8  (struct gguf_context * ctx, int i);
     GGML_API int8_t       gguf_get_val_i8  (struct gguf_context * ctx, int i);
     GGML_API uint16_t     gguf_get_val_u16 (struct gguf_context * ctx, int i);
@@ -1764,12 +1765,60 @@ extern "C" {
     GGML_API bool         gguf_get_val_bool(struct gguf_context * ctx, int i);
     GGML_API const char * gguf_get_val_str (struct gguf_context * ctx, int i);
     GGML_API int          gguf_get_arr_n   (struct gguf_context * ctx, int i);
-    GGML_API void         gguf_get_arr_data(struct gguf_context * ctx, int i, void * data);
+    GGML_API const void * gguf_get_arr_data(struct gguf_context * ctx, int i);
+    GGML_API const char * gguf_get_arr_str (struct gguf_context * ctx, int key_id, int i);
 
     GGML_API int    gguf_get_n_tensors    (struct gguf_context * ctx);
+    GGML_API int    gguf_find_tensor      (struct gguf_context * ctx, const char * name);
     GGML_API size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i);
     GGML_API char * gguf_get_tensor_name  (struct gguf_context * ctx, int i);
 
+    // overrides existing values or adds a new one
+    GGML_API void gguf_set_val_u8  (struct gguf_context * ctx, const char * key, uint8_t  val);
+    GGML_API void gguf_set_val_i8  (struct gguf_context * ctx, const char * key, int8_t   val);
+    GGML_API void gguf_set_val_u16 (struct gguf_context * ctx, const char * key, uint16_t val);
+    GGML_API void gguf_set_val_i16 (struct gguf_context * ctx, const char * key, int16_t  val);
+    GGML_API void gguf_set_val_u32 (struct gguf_context * ctx, const char * key, uint32_t val);
+    GGML_API void gguf_set_val_i32 (struct gguf_context * ctx, const char * key, int32_t  val);
+    GGML_API void gguf_set_val_f32 (struct gguf_context * ctx, const char * key, float    val);
+    GGML_API void gguf_set_val_bool(struct gguf_context * ctx, const char * key, bool     val);
+    GGML_API void gguf_set_val_str (struct gguf_context * ctx, const char * key, const char * val);
+    GGML_API void gguf_set_arr_data(struct gguf_context * ctx, const char * key, enum gguf_type type, const void * data, int n);
+    GGML_API void gguf_set_arr_str (struct gguf_context * ctx, const char * key, const char ** data, int n);
+
+    // set or add KV pairs from another context
+    GGML_API void gguf_set_kv(struct gguf_context * ctx, struct gguf_context * src);
+
+    // manage tensor info
+    GGML_API void gguf_add_tensor(struct gguf_context * ctx, const struct ggml_tensor * tensor);
+    GGML_API void gguf_set_tensor_type(struct gguf_context * ctx, const char * name, enum ggml_type type);
+    GGML_API void gguf_set_tensor_data(struct gguf_context * ctx, const char * name, const void * data, size_t size);
+
+    // writing gguf files can be done in 2 ways:
+    //
+    // - write the entire gguf_context to a binary file in a single pass:
+    //
+    //   gguf_write_to_file(ctx, fname);
+    //
+    // - first prepare a file with a placeholder for the meta data, write the tensor data, then write the meta data:
+    //
+    //   FILE * f = fopen(fname, "wb");
+    //   fseek(f, gguf_get_meta_size(ctx), SEEK_SET);
+    //   fwrite(f, ...);
+    //   void * data = gguf_meta_get_meta_data(ctx);
+    //   fseek(f, 0, SEEK_SET);
+    //   fwrite(f, data, gguf_get_meta_size(ctx));
+    //   free(data);
+    //   fclose(f);
+    //
+
+    // write the entire context to a binary file
+    GGML_API void gguf_write_to_file(struct gguf_context * ctx, const char * fname, bool only_meta);
+
+    // get the size in bytes of the meta data (header, kv pairs, tensor info) including padding
+    GGML_API size_t gguf_get_meta_size(struct gguf_context * ctx);
+    GGML_API void   gguf_get_meta_data(struct gguf_context * ctx, void * data);
+
     //
     // system info
     //
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
index e36d8e77cfc4d..d99d752ecf291 100644
--- a/gguf-llama.cpp
+++ b/gguf-llama.cpp
@@ -6,62 +6,126 @@
 #include <cstdio>
 #endif
 
-#include "gguf-util.h"
 #define LLAMA_API_CPP // TODO: eliminate me
 #include "gguf-llama.h"
 
 #include "ggml.h"
+
+#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
+#  include "ggml-alloc.h"
+#  define LLAMA_USE_ALLOCATOR
+#else
+#  define LLAMA_USE_SCRATCH
+#  define LLAMA_MAX_SCRATCH_BUFFERS 16
+#endif
+
 #ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
+#  include "ggml-cuda.h"
 #elif defined(GGML_USE_CLBLAST)
-#include "ggml-opencl.h"
+#  include "ggml-opencl.h"
 #endif
 
 #ifdef GGML_USE_METAL
-#include "ggml-metal.h"
+#  include "ggml-metal.h"
 #endif
 #ifdef GGML_USE_MPI
-#include "ggml-mpi.h"
+#  include "ggml-mpi.h"
 #endif
 #ifdef GGML_USE_K_QUANTS
-#ifndef QK_K
-#ifdef GGML_QKK_64
-#define QK_K 64
-#else
-#define QK_K 256
+#  ifndef QK_K
+#    ifdef GGML_QKK_64
+#      define QK_K 64
+#    else
+#      define QK_K 256
+#    endif
+#  endif
 #endif
+
+#ifdef __has_include
+    #if __has_include(<unistd.h>)
+        #include <unistd.h>
+        #if defined(_POSIX_MAPPED_FILES)
+            #include <sys/mman.h>
+        #endif
+        #if defined(_POSIX_MEMLOCK_RANGE)
+            #include <sys/resource.h>
+        #endif
+    #endif
 #endif
+
+#if defined(_WIN32)
+    #define WIN32_LEAN_AND_MEAN
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <windows.h>
+    #include <io.h>
+    #include <stdio.h> // for _fseeki64
 #endif
 
+#include <algorithm>
 #include <array>
-#include <ctime>
+#include <cassert>
 #include <cinttypes>
+#include <climits>
+#include <cstdarg>
+#include <cstring>
+#include <ctime>
 #include <fstream>
-#include <random>
+#include <initializer_list>
 #include <map>
-#include <unordered_map>
-#include <queue>
-#include <cassert>
-#include <cstring>
-#include <climits>
 #include <memory>
-#include <algorithm>
-#include <initializer_list>
-#include <thread>
 #include <mutex>
-#include <sstream>
 #include <numeric>
+#include <queue>
+#include <random>
+#include <sstream>
+#include <thread>
+#include <unordered_map>
 
 #if defined(_MSC_VER)
 #pragma warning(disable: 4244 4267) // possible loss of data
 #endif
 
-static void llama_log_internal(llama_log_level level, const char* format, ...);
+// tensor names
+#define TN_TOKEN_EMBD  "token_embd.weight"
+#define TN_OUTPUT_NORM "output_norm.weight"
+#define TN_OUTPUT      "output.weight"
+#define TN_ATTN_NORM   "blk.%d.attn_norm.weight"
+#define TN_ATTN_Q      "blk.%d.attn_q.weight"
+#define TN_ATTN_K      "blk.%d.attn_k.weight"
+#define TN_ATTN_V      "blk.%d.attn_v.weight"
+#define TN_ATTN_OUTPUT "blk.%d.attn_output.weight"
+#define TN_FFN_NORM    "blk.%d.ffn_norm.weight"
+#define TN_FFN_GATE    "blk.%d.ffn_gate.weight"
+#define TN_FFN_DOWN    "blk.%d.ffn_down.weight"
+#define TN_FFN_UP      "blk.%d.ffn_up.weight"
+
+#ifdef __GNUC__
+#ifdef __MINGW32__
+#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(gnu_printf, __VA_ARGS__)))
+#else
+#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
+#endif
+#else
+#define LLAMA_ATTRIBUTE_FORMAT(...)
+#endif
+
+//
+// logging
+//
+LLAMA_ATTRIBUTE_FORMAT(2, 3)
+static void llama_log_internal        (llama_log_level level, const char* format, ...);
 static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data);
+
 #define LLAMA_LOG_INFO(...)  llama_log_internal(LLAMA_LOG_LEVEL_INFO , __VA_ARGS__)
 #define LLAMA_LOG_WARN(...)  llama_log_internal(LLAMA_LOG_LEVEL_WARN , __VA_ARGS__)
 #define LLAMA_LOG_ERROR(...) llama_log_internal(LLAMA_LOG_LEVEL_ERROR, __VA_ARGS__)
 
+//
+// helpers
+//
+
 template<typename T>
 static std::string to_string(const T & val) {
     std::stringstream ss;
@@ -69,25 +133,71 @@ static std::string to_string(const T & val) {
     return ss.str();
 }
 
-#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
-#include "ggml-alloc.h"
-#define LLAMA_USE_ALLOCATOR
-#else
-#define LLAMA_USE_SCRATCH
-#define LLAMA_MAX_SCRATCH_BUFFERS 16
-#endif
+static void zeros(std::ofstream & file, size_t n) {
+    char zero = 0;
+    for (size_t i = 0; i < n; ++i) {
+        file.write(&zero, 1);
+    }
+}
+
+LLAMA_ATTRIBUTE_FORMAT(1, 2)
+static std::string format(const char * fmt, ...) {
+    va_list ap;
+    va_list ap2;
+    va_start(ap, fmt);
+    va_copy(ap2, ap);
+    int size = vsnprintf(NULL, 0, fmt, ap);
+    GGML_ASSERT(size >= 0 && size < INT_MAX); // NOLINT
+    std::vector<char> buf(size + 1);
+    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
+    GGML_ASSERT(size2 == size);
+    va_end(ap2);
+    va_end(ap);
+    return std::string(buf.data(), size);
+}
 
-#define UNUSED GGML_UNUSED
+//
+// ggml helpers
+//
+
+static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
+    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
+
+    if (plan.work_size > 0) {
+        buf.resize(plan.work_size);
+        plan.work_data = buf.data();
+    }
+
+    ggml_graph_compute(graph, &plan);
+}
+
+//
+// llama helpers
+//
 
 #ifdef GGML_USE_CUBLAS
-#define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
-#define llama_host_free(data) ggml_cuda_host_free(data)
+#   define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
+#   define llama_host_free(data) ggml_cuda_host_free(data)
 #elif GGML_USE_METAL
-#define llama_host_malloc(n)  ggml_metal_host_malloc(n)
-#define llama_host_free(data) ggml_metal_host_free(data)
+#   define llama_host_malloc(n)  ggml_metal_host_malloc(n)
+#   define llama_host_free(data) ggml_metal_host_free(data)
 #else
-#define llama_host_malloc(n)  malloc(n)
-#define llama_host_free(data) free(data)
+#   define llama_host_malloc(n)  malloc(n)
+#   define llama_host_free(data) free(data)
+#endif
+
+#if defined(_WIN32)
+static std::string llama_format_win_err(DWORD err) {
+    LPSTR buf;
+    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
+                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
+    if (!size) {
+        return "FormatMessageA failed";
+    }
+    std::string ret(buf, size);
+    LocalFree(buf);
+    return ret;
+}
 #endif
 
 struct llama_buffer {
@@ -126,6 +236,324 @@ struct llama_buffer {
     }
 };
 
+struct llama_file {
+    // use FILE * so we don't have to re-open the file to mmap
+    FILE * fp;
+    size_t size;
+
+    llama_file(const char * fname, const char * mode) {
+        fp = std::fopen(fname, mode);
+        if (fp == NULL) {
+            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
+        }
+        seek(0, SEEK_END);
+        size = tell();
+        seek(0, SEEK_SET);
+    }
+
+    size_t tell() const {
+#ifdef _WIN32
+        __int64 ret = _ftelli64(fp);
+#else
+        long ret = std::ftell(fp);
+#endif
+        GGML_ASSERT(ret != -1); // this really shouldn't fail
+        return (size_t) ret;
+    }
+
+    void seek(size_t offset, int whence) const {
+#ifdef _WIN32
+        int ret = _fseeki64(fp, (__int64) offset, whence);
+#else
+        int ret = std::fseek(fp, (long) offset, whence);
+#endif
+        GGML_ASSERT(ret == 0); // same
+    }
+
+    void read_raw(void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        std::size_t ret = std::fread(ptr, len, 1, fp);
+        if (ferror(fp)) {
+            throw std::runtime_error(format("read error: %s", strerror(errno)));
+        }
+        if (ret != 1) {
+            throw std::runtime_error(std::string("unexpectedly reached end of file"));
+        }
+    }
+
+    void write_raw(const void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        size_t ret = std::fwrite(ptr, len, 1, fp);
+        if (ret != 1) {
+            throw std::runtime_error(format("write error: %s", strerror(errno)));
+        }
+    }
+
+    ~llama_file() {
+        if (fp) {
+            std::fclose(fp);
+        }
+    }
+};
+
+struct llama_mmap {
+    void * addr;
+    size_t size;
+
+    llama_mmap(const llama_mmap &) = delete;
+
+#ifdef _POSIX_MAPPED_FILES
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
+        size = file->size;
+        int fd = fileno(file->fp);
+        int flags = MAP_SHARED;
+        // prefetch/readahead impairs performance on NUMA systems
+        if (numa) { prefetch = 0; }
+#ifdef __linux__
+        if (prefetch) { flags |= MAP_POPULATE; }
+#endif
+        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
+        if (addr == MAP_FAILED) {
+            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
+        }
+
+        if (prefetch > 0) {
+            // Advise the kernel to preload the mapped memory
+            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
+                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+        if (numa) {
+            // advise the kernel not to use readahead
+            // (because the next page might not belong on the same node)
+            if (madvise(addr, file->size, MADV_RANDOM)) {
+                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+    }
+
+    ~llama_mmap() {
+        munmap(addr, size);
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) numa;
+
+        size = file->size;
+
+        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
+
+        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
+        DWORD error = GetLastError();
+
+        if (hMapping == NULL) {
+            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
+        error = GetLastError();
+        CloseHandle(hMapping);
+
+        if (addr == NULL) {
+            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+        if (prefetch) {
+            // Advise the kernel to preload the mapped memory
+            WIN32_MEMORY_RANGE_ENTRY range;
+            range.VirtualAddress = addr;
+            range.NumberOfBytes = (SIZE_T)size;
+            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
+                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+            }
+        }
+        #else
+        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
+        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
+    }
+
+    ~llama_mmap() {
+        if (!UnmapViewOfFile(addr)) {
+            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) file;
+        (void) prefetch;
+        (void) numa;
+
+        throw std::runtime_error(std::string("mmap not supported"));
+    }
+#endif
+};
+
+// Represents some region of memory being locked using mlock or VirtualLock;
+// will automatically unlock on destruction.
+struct llama_mlock {
+    void * addr = NULL;
+    size_t size = 0;
+
+    bool failed_already = false;
+
+    llama_mlock() {}
+    llama_mlock(const llama_mlock &) = delete;
+
+    ~llama_mlock() {
+        if (size) {
+            raw_unlock(addr, size);
+        }
+    }
+
+    void init(void * ptr) {
+        GGML_ASSERT(addr == NULL && size == 0); // NOLINT
+        addr = ptr;
+    }
+
+    void grow_to(size_t target_size) {
+        GGML_ASSERT(addr);
+        if (failed_already) {
+            return;
+        }
+        size_t granularity = lock_granularity();
+        target_size = (target_size + granularity - 1) & ~(granularity - 1);
+        if (target_size > size) {
+            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
+                size = target_size;
+            } else {
+                failed_already = true;
+            }
+        }
+    }
+
+#ifdef _POSIX_MEMLOCK_RANGE
+    static constexpr bool SUPPORTED = true;
+
+    static size_t lock_granularity() {
+        return (size_t) sysconf(_SC_PAGESIZE);
+    }
+
+    #ifdef __APPLE__
+        #define MLOCK_SUGGESTION \
+            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
+            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
+    #else
+        #define MLOCK_SUGGESTION \
+            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
+    #endif
+
+    bool raw_lock(const void * addr, size_t size) const {
+        if (!mlock(addr, size)) {
+            return true;
+        }
+
+        char* errmsg = std::strerror(errno);
+        bool suggest = (errno == ENOMEM);
+
+        // Check if the resource limit is fine after all
+        struct rlimit lock_limit;
+        if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit)) {
+            suggest = false;
+        }
+        if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size)) {
+            suggest = false;
+        }
+
+        fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
+                size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
+        return false;
+    }
+
+    #undef MLOCK_SUGGESTION
+
+    static void raw_unlock(void * addr, size_t size) {
+        if (munlock(addr, size)) {
+            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
+        }
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    static size_t lock_granularity() {
+        SYSTEM_INFO si;
+        GetSystemInfo(&si);
+        return (size_t) si.dwPageSize;
+    }
+
+    bool raw_lock(void * ptr, size_t len) const {
+        for (int tries = 1; ; tries++) {
+            if (VirtualLock(ptr, len)) {
+                return true;
+            }
+            if (tries == 2) {
+                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
+                    len, size, llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+
+            // It failed but this was only the first try; increase the working
+            // set size and try again.
+            SIZE_T min_ws_size, max_ws_size;
+            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
+                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+            // Per MSDN: "The maximum number of pages that a process can lock
+            // is equal to the number of pages in its minimum working set minus
+            // a small overhead."
+            // Hopefully a megabyte is enough overhead:
+            size_t increment = len + 1048576;
+            // The minimum must be <= the maximum, so we need to increase both:
+            min_ws_size += increment;
+            max_ws_size += increment;
+            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
+                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+        }
+    }
+
+    static void raw_unlock(void * ptr, size_t len) {
+        if (!VirtualUnlock(ptr, len)) {
+            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    static size_t lock_granularity() {
+        return (size_t) 65536;
+    }
+
+    bool raw_lock(const void * addr, size_t len) const {
+        fprintf(stderr, "warning: mlock not supported on this system\n");
+        return false;
+    }
+
+    static void raw_unlock(const void * addr, size_t len) {}
+#endif
+};
+
 typedef void (*offload_func_t)(struct ggml_tensor * tensor);
 
 void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
@@ -133,19 +561,16 @@ void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
 }
 
 //
-// ggml helpers
+// globals
 //
 
-static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
-    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
-
-    if (plan.work_size > 0) {
-        buf.resize(plan.work_size);
-        plan.work_data = buf.data();
-    }
+struct llama_state {
+    // We save the log callback globally
+    llama_log_callback log_callback = llama_log_callback_default;
+    void * log_callback_user_data = nullptr;
+};
 
-    ggml_graph_compute(graph, &plan);
-}
+static llama_state g_state;
 
 //
 // memory sizes (calculated for n_batch == 512)
@@ -250,7 +675,7 @@ struct llama_hparams {
     uint32_t n_rot     = 64;
     uint32_t n_ff      = 11008;
 
-    float f_rms_norm_eps = 1e-5;
+    float f_norm_rms_eps = 1e-5;
 
     float rope_freq_base  = 10000.0f;
     float rope_freq_scale = 1.0f;
@@ -325,10 +750,10 @@ struct llama_kv_cache {
 };
 
 struct llama_vocab {
-    // TODO: convert to this gguf_vocab
-    // add a vector of merges
-    // add members for bos/eos/pad/sep tokens
-    // so that we can pass it to different types of tokenizers with a common interface
+    // TODO:
+    // - add a vector of merges
+    // - add members for bos/eos/pad/sep tokens
+    //   so that we can pass it to different types of tokenizers with a common interface
 
     using id    = int32_t;
     using token = std::string;
@@ -346,6 +771,7 @@ struct llama_model {
     e_model type = MODEL_UNKNOWN;
 
     llama_hparams hparams;
+    llama_vocab   vocab;
 
     struct ggml_tensor * tok_embeddings;
 
@@ -362,11 +788,11 @@ struct llama_model {
     llama_buffer buf;
 
     // model memory mapped file
-    std::unique_ptr<gguf_mmap> mapping;
+    std::unique_ptr<llama_mmap> mapping;
 
     // objects representing data potentially being locked in memory
-    gguf_mlock mlock_buf;
-    gguf_mlock mlock_mmap;
+    llama_mlock mlock_buf;
+    llama_mlock mlock_mmap;
 
     // for quantize-stats only
     std::vector<std::pair<std::string, struct ggml_tensor *>> tensors_by_name;
@@ -374,8 +800,6 @@ struct llama_model {
     int64_t t_load_us = 0;
     int64_t t_start_us = 0;
 
-    llama_vocab vocab;
-
     ~llama_model() {
         if (ctx) {
             ggml_free(ctx);
@@ -502,525 +926,294 @@ struct llama_context {
     }
 };
 
-struct llama_state {
-    // We save the log callback globally
-    llama_log_callback log_callback = llama_log_callback_default;
-    void * log_callback_user_data = nullptr;
-};
-// global state
-static llama_state g_state;
-
-template <typename T>
-static T checked_mul(T a, T b) {
-    T ret = a * b;
-    if (a != 0 && ret / a != b) {
-        throw std::runtime_error(format("overflow multiplying %llu * %llu",
-                     (unsigned long long) a, (unsigned long long) b));
-    }
-    return ret;
-}
+//
+// kv cache helpers
+//
 
-static size_t checked_div(size_t a, size_t b) {
-    if (b == 0 || a % b != 0) {
-        throw std::runtime_error(format("error dividing %zu / %zu", a, b));
-    }
-    return a / b;
-}
+static bool llama_kv_cache_init(
+        const struct llama_hparams & hparams,
+             struct llama_kv_cache & cache,
+                         ggml_type   wtype,
+                               int   n_ctx,
+                               int   n_gpu_layers) {
+    const int n_embd  = hparams.n_embd_gqa();
+    const int n_layer = hparams.n_layer;
 
-static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
-    char buf[256];
-    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
-    for (size_t i = 1; i < ne.size(); i++) {
-        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
-    }
-    return buf;
-}
+    const int64_t n_mem      = n_layer*n_ctx;
+    const int64_t n_elements = n_embd*n_mem;
 
-static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
-    size_t size = ggml_type_size(type);
-    for (uint32_t dim : ne) {
-        size = checked_mul<size_t>(size, dim);
-    }
-    return size / ggml_blck_size(type);
-}
+    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
+    cache.n = 0;
 
-struct gguf_load_tensor {
-    std::string name;
-    enum ggml_type type = GGML_TYPE_F32;
-    std::vector<uint32_t> ne;
-    size_t file_off;
-    size_t size;
-    struct ggml_tensor * ggml_tensor = NULL;
-    uint8_t * data;
-};
+    struct ggml_init_params params;
+    params.mem_size   = cache.buf.size;
+    params.mem_buffer = cache.buf.data;
+    params.no_alloc   = false;
 
-struct gguf_load_tensors_map {
-    // tensors is kept in a separate vector to preserve file order
-    std::vector<gguf_load_tensor> tensors;
-    std::unordered_map<std::string, size_t> name_to_idx;
-};
+    cache.ctx = ggml_init(params);
 
-enum gguf_file_version {
-    GGUF_FILE_VERSION_V1 = 1,
-};
+    if (!cache.ctx) {
+        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
+        return false;
+    }
 
-struct gguf_file_loader {
-    gguf_file file;
-    gguf_context * gguf_ctx;
-    gguf_file_version file_version;
-    llama_hparams hparams;
-    llama_vocab vocab;
-
-    struct ggml_context * ctx_data = NULL;
-
-    gguf_file_loader(const char * fname, gguf_load_tensors_map & tensors_map)
-        : file(fname, "rb") {
-        fprintf(stderr, "llama.cpp: loading model from %s\n", fname);
-
-    struct gguf_init_params params = {
-        /*.no_alloc = */ true,
-        /*.ctx      = */ &ctx_data,
-    };
-
-    gguf_ctx = gguf_init_from_file(fname, params);
-    file_version = (enum gguf_file_version) gguf_get_version(gguf_ctx);
+    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    ggml_set_name(cache.k, "cache_k");
+    ggml_set_name(cache.v, "cache_v");
 
-    read_hparams();
-    read_vocab();
-        read_tensor_metadata(tensors_map);
+    (void) n_gpu_layers;
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer + 1) {
+        ggml_cuda_assign_buffers_no_scratch(cache.v);
     }
-
-    uint32_t read_u32(const char * key) const {
-        int i = gguf_find_key(gguf_ctx, key);
-        if (i == -1) {
-            throw std::runtime_error(format("cannot find param with key %s\n", key));
-        }
-
-        return gguf_get_val_u32(gguf_ctx, i);
+    if (n_gpu_layers > n_layer + 2) {
+        ggml_cuda_assign_buffers_no_scratch(cache.k);
     }
+#endif // GGML_USE_CUBLAS
 
-    float read_f32(const char * key) const {
-        int i = gguf_find_key(gguf_ctx, key);
-        if (i == -1) {
-            throw std::runtime_error(format("cannot find param with key %s\n", key));
-        }
+    return true;
+}
 
-        return gguf_get_val_f32(gguf_ctx, i);
-    }
+//
+// model loading and saving
+//
 
-    int read_n_vocab() const {
-        int i = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
-        if (i == -1) {
-            throw std::runtime_error("cannot find token list in GGUF file\n");
-        }
+enum llama_file_version {
+    GGUF_FILE_VERSION_V1 = 1,
+};
 
-        return gguf_get_arr_n(gguf_ctx, i);
+static const char * llama_file_version_name(llama_file_version version) {
+    switch (version) {
+        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
     }
 
-    void read_hparams() {
-        // TODO define keys as constants in header
-        // TODO: read all hparams from file
-
-        hparams.n_vocab        = read_n_vocab();
-        hparams.n_ctx          = read_u32("llama.context_length");
-        hparams.n_embd         = read_u32("llama.embedding_length");
-        hparams.n_ff           = read_u32("llama.feed_forward_length");
-        hparams.n_head         = read_u32("llama.attention.head_count");
-        hparams.n_layer        = read_u32("llama.block_count");
-        hparams.n_rot          = read_u32("llama.rope.dimension_count");
-        hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
-
-        // n_head_kv default to n_head
-        hparams.n_head_kv = gguf_find_key(gguf_ctx, "llama.attention.head_count_kv") == -1 ?  hparams.n_head : read_u32("llama.attention.head_count_kv");
+    return "unknown";
+}
 
+static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
+    char buf[256];
+    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
+    for (size_t i = 1; i < ne.size(); i++) {
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5u", ne.at(i));
     }
+    return buf;
+}
 
-    void read_vocab() {
-        vocab.id_to_token.resize(hparams.n_vocab);
-        int token_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.tokens");
-        if (token_idx == -1) {
-            throw std::runtime_error("cannot find token list in GGUF file\n");
-        }
+static std::string llama_format_tensor_shape(const struct ggml_tensor * t) {
+    char buf[256];
+    snprintf(buf, sizeof(buf), "%5" PRId64, t->ne[0]);
+    for (int i = 1; i < GGML_MAX_DIMS; i++) {
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5" PRId64, t->ne[i]);
+    }
+    return buf;
+}
 
-        int score_idx = gguf_find_key(gguf_ctx, "tokenizer.ggml.scores");
-        if (score_idx == -1) {
-            throw std::runtime_error("cannot find token scores list in GGUF file\n");
-        }
+struct llama_model_loader {
+    int n_kv      = 0;
+    int n_tensors = 0;
+    int n_created = 0;
 
-        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
+    bool use_mmap = false;
 
-            std::string word = gguf_get_arr_str(gguf_ctx, token_idx, i);
+    llama_file file;
+    llama_file_version file_version;
 
-            vocab.token_to_id[word] = i;
+    std::unique_ptr<llama_mmap> mapping;
 
-            auto & tok_score = vocab.id_to_token[i];
-            tok_score.tok = std::move(word);
-            tok_score.score = gguf_get_arr_f32(gguf_ctx, score_idx, i);
-        }
-    }
+    struct gguf_context * ctx_gguf = NULL;
+    struct ggml_context * ctx_meta = NULL;
 
-    void read_tensor_metadata(gguf_load_tensors_map & tensors_map) const {
-        const int n_tensors = gguf_get_n_tensors(gguf_ctx);
+    llama_model_loader(const std::string & fname, bool use_mmap) : file(fname.c_str(), "rb") {
+        struct gguf_init_params params = {
+            /*.no_alloc = */ true,
+            /*.ctx      = */ &ctx_meta,
+        };
 
-        for (int i = 0; i < n_tensors; ++i) {
-            gguf_load_tensor tensor;
-            const char * name = gguf_get_tensor_name(gguf_ctx, i);
+        ctx_gguf = gguf_init_from_file(fname.c_str(), params);
 
-            struct ggml_tensor * cur = ggml_get_tensor(ctx_data, name);
+        n_kv      = gguf_get_n_kv(ctx_gguf);
+        n_tensors = gguf_get_n_tensors(ctx_gguf);
 
-            const uint32_t n_dims = cur->n_dims;
-            tensor.type = cur->type;
-            tensor.ne.resize(n_dims);
+        file_version = (enum llama_file_version) gguf_get_version(ctx_gguf);
 
-            for (uint32_t j = 0; j < n_dims; ++j) {
-                tensor.ne[j] = cur->ne[j];
-            }
+        // print meta data
+        // TODO: make optional
+        {
+            LLAMA_LOG_INFO("%s: loaded meta data with %d key-value paris and %d tensors from %s (version %s)\n",
+                    __func__, n_kv, n_tensors, fname.c_str(), llama_file_version_name(file_version));
 
-            if (n_dims < 1 || n_dims > 2) {
-                throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name, n_dims));
-            }
+            for (int i = 0; i < n_tensors; i++) {
+                const char * name = gguf_get_tensor_name(ctx_gguf, i);
+                struct ggml_tensor * meta = ggml_get_tensor(ctx_meta, name);
 
-            switch (tensor.type) {
-                case GGML_TYPE_F32:
-                case GGML_TYPE_F16:
-                case GGML_TYPE_Q4_0:
-                case GGML_TYPE_Q4_1:
-                case GGML_TYPE_Q5_0:
-                case GGML_TYPE_Q5_1:
-                case GGML_TYPE_Q8_0:
-                case GGML_TYPE_Q2_K:
-                case GGML_TYPE_Q3_K:
-                case GGML_TYPE_Q4_K:
-                case GGML_TYPE_Q5_K:
-                case GGML_TYPE_Q6_K:
-                    break;
-                default: {
-                    throw std::runtime_error(format("unrecognized tensor type %u\n", tensor.type));
-                }
+                LLAMA_LOG_INFO("%s: - tensor %3d: %32s %-8s [ %s ]\n", __func__, i, name, ggml_type_name(meta->type), llama_format_tensor_shape(meta).c_str());
             }
 
-            tensor.file_off = gguf_get_data_offset(gguf_ctx) + gguf_get_tensor_offset(gguf_ctx, i);
-
-            tensor.name = name;
-            tensor.size = llama_calc_tensor_size(tensor.ne, tensor.type);
+            for (int i = 0; i < n_kv; i++) {
+                const char * name         = gguf_get_key(ctx_gguf, i);
+                const enum gguf_type type = gguf_get_kv_type(ctx_gguf, i);
 
-            tensors_map.tensors.push_back(tensor);
-            tensors_map.name_to_idx[name] = tensors_map.tensors.size() - 1;
+                LLAMA_LOG_INFO("%s: - kv %3d: %42s %-8s\n", __func__, i, name, gguf_type_name(type));
+            }
         }
-    }
-};
-
-struct gguf_file_saver {
-    // TODO
-    // this implementation now assumes that the data section is of the same length as the unquantized model.
-    // this is needed to write tensor metadata and weights in a single pass by seeking to appropriate positions in the file.
-    // this may not be true when we add quantization version and change ftype description (currently it's string according to the specs,
-    // but better to have it as uint32).
-    // we need to calculate the delta in number of bytes written with a counter as a struct member.
 
-    gguf_file file;
-    gguf_file_loader * fl;
-    size_t info_offset;
-    size_t tensor_offset = 0;
-
-    gguf_file_saver(const char * fname, gguf_file_loader * fl, enum llama_ftype new_ftype)
-        : file(fname, "wb"), fl(fl) {
-            fprintf(stderr, "llama.cpp: saving model to %s\n", fname);
-            write_header();
-            write_hparams(new_ftype);
+        if (!llama_mmap::SUPPORTED) {
+            LLAMA_LOG_WARN("%s: mmap is not supported on this platform\n", __func__);
+            use_mmap = false;
         }
 
-    void write_header() {
-        const int32_t magic = GGUF_MAGIC;
-        file.write_i32(magic);
-
-        const int32_t version = GGUF_VERSION;
-        file.write_i32(version);
-
-        const int32_t n_tensors = gguf_get_n_tensors(fl->gguf_ctx);
-        file.write_i32(n_tensors);
-
-        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
-        file.write_i32(n_kv);
+        this->use_mmap = use_mmap;
     }
 
-    void write_hparam_arr_str(const std::string & key, enum gguf_type type, int i, int n_arr) {
-        std::vector<std::string> data(n_arr);
-
-        for (int j = 0; j < n_arr; ++j) {
-            std::string val = gguf_get_arr_str(fl->gguf_ctx, i, j);
-            data[j] = val;
-        }
+    const char * get_tensor_name(int i) const {
+        return gguf_get_tensor_name(ctx_gguf, i);
+    }
 
-        file.write_arr<std::string>(key, type, data);
+    struct ggml_tensor * get_tensor_meta(int i) const {
+        return ggml_get_tensor(ctx_meta, get_tensor_name(i));
     }
 
-    void write_hparam_arr_f32(const std::string & key, enum gguf_type type, int i, int n_arr) {
-        std::vector<float> data(n_arr);
+    void calc_sizes(size_t & ctx_size_p, size_t & mmapped_size_p) const {
+        ctx_size_p     = 0;
+        mmapped_size_p = 0;
 
-        for (int j = 0; j < n_arr; ++j) {
-            float val = gguf_get_arr_f32(fl->gguf_ctx, i, j);
-            data[j] = val;
+        for (int i = 0; i < n_tensors; i++) {
+            struct ggml_tensor * meta = get_tensor_meta(i);
+            ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
+            (use_mmap ? mmapped_size_p : ctx_size_p) += ggml_nbytes_pad(meta);
         }
-
-        file.write_arr<float>(key, type, data);
     }
 
-    void write_hparams(enum llama_ftype new_ftype) {
-        const int32_t n_kv = gguf_get_n_kv(fl->gguf_ctx);
-        for (int i = 0; i < n_kv; ++i) {
-            const char * key = gguf_get_key(fl->gguf_ctx, i);
-            if (strcmp(key, "general.quantization_version") == 0) {
-                file.write_val<uint32_t>("general.quantization_version", GGUF_TYPE_UINT32, new_ftype);
-            } else {
-                const gguf_type vtype = gguf_get_kv_type(fl->gguf_ctx, i);
-
-                bool bool_val;
-                float f32_val;
-                int16_t i16_val;
-                int32_t i32_val;
-                int8_t i8_val;
-                std::string str_val;
-                uint16_t u16_val;
-                uint32_t u32_val;
-                uint8_t u8_val;
-                gguf_type arr_type;
-                int n_arr;
-
-                switch(vtype) {
-                    case GGUF_TYPE_BOOL:
-                        bool_val = gguf_get_val_bool(fl->gguf_ctx, i);
-                        file.write_val<bool>(key, GGUF_TYPE_BOOL, bool_val);
-                        break;
-                    case GGUF_TYPE_FLOAT32:
-                        f32_val = gguf_get_val_f32(fl->gguf_ctx, i);
-                        file.write_val<float>(key, GGUF_TYPE_FLOAT32, f32_val);
-                        break;
-                    case GGUF_TYPE_INT16:
-                        i16_val = gguf_get_val_i16(fl->gguf_ctx, i);
-                        file.write_val<int16_t>(key, GGUF_TYPE_INT16, i16_val);
-                        break;
-                    case GGUF_TYPE_INT32:
-                        i32_val = gguf_get_val_i32(fl->gguf_ctx, i);
-                        file.write_val<int32_t>(key, GGUF_TYPE_INT32, i32_val);
-                        break;
-                    case GGUF_TYPE_INT8:
-                        i8_val = gguf_get_val_i8(fl->gguf_ctx, i);
-                        file.write_val<int8_t>(key, GGUF_TYPE_INT8, i8_val);
-                        break;
-                    case GGUF_TYPE_STRING:
-                        str_val = gguf_get_val_str(fl->gguf_ctx, i);
-                        file.write_val<std::string>(key, GGUF_TYPE_STRING, str_val);
-                        break;
-                    case GGUF_TYPE_UINT16:
-                        u16_val = gguf_get_val_u16(fl->gguf_ctx, i);
-                        file.write_val<uint16_t>(key, GGUF_TYPE_UINT16, u16_val);
-                        break;
-                    case GGUF_TYPE_UINT32:
-                        u32_val = gguf_get_val_u32(fl->gguf_ctx, i);
-                        file.write_val<uint32_t>(key, GGUF_TYPE_UINT32, u32_val);
-                        break;
-                    case GGUF_TYPE_UINT8:
-                        u8_val = gguf_get_val_u8(fl->gguf_ctx, i);
-                        file.write_val<uint8_t>(key, GGUF_TYPE_UINT8, u8_val);
-                        break;
-                    case GGUF_TYPE_ARRAY:
-                        arr_type = gguf_get_arr_type(fl->gguf_ctx, i);
-                        n_arr    = gguf_get_arr_n(fl->gguf_ctx, i);
-                        if (arr_type == GGUF_TYPE_FLOAT32) {
-                            write_hparam_arr_f32(key, arr_type, i, n_arr);
-                        } else if (arr_type == GGUF_TYPE_STRING) {
-                            write_hparam_arr_str(key, GGUF_TYPE_STRING, i, n_arr);
-                        } else {
-                            throw std::runtime_error("not implemented");
-                        }
-                        break;
-                    default:
-                        throw std::runtime_error(format("cannot recognize value type for key %s\n", key));
-                }
-            }
+    struct ggml_tensor * create_tensor_for(struct ggml_context * ctx, struct ggml_tensor * meta, ggml_backend backend) {
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ctx, true);
         }
 
-        info_offset = file.tell();
-
-        GGML_ASSERT(gguf_get_data_offset(fl->gguf_ctx) >= info_offset);
-
-        size_t count = gguf_get_data_offset(fl->gguf_ctx) - info_offset;
-        file.write_zeros(count);
-        file.seek(info_offset, SEEK_SET);
-        GGML_ASSERT(info_offset == file.tell());
-    }
-
-    size_t write_tensor_info(gguf_load_tensor & tensor, enum ggml_type type) {
-        size_t total_written = 0;
-        file.seek(info_offset, SEEK_SET);
-        GGML_ASSERT(info_offset == file.tell());
-        total_written += file.write_str(tensor.name);
+        struct ggml_tensor * tensor = ggml_dup_tensor(ctx, meta);
+        tensor->backend = backend; // TODO: ggml_set_backend
+        ggml_set_name(tensor, ggml_get_name(meta));
 
-        int32_t n_dims = tensor.ne.size();
-        total_written += file.write_i32(n_dims);
-        for (int32_t i = 0; i < n_dims; ++i) {
-            total_written += file.write_i32(tensor.ne[i]);
+        if (backend != GGML_BACKEND_CPU) {
+            ggml_set_no_alloc(ctx, use_mmap);
         }
 
-        total_written += file.write_i32(type);
-        total_written += file.write_u64(tensor_offset);
-        info_offset   += total_written; // position to write info of the next tensor
+        n_created++;
 
-        file.seek(0, SEEK_END);
-
-        return total_written;
+        return tensor;
     }
 
-    void write_tensor(gguf_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
-        switch (new_type) {
-            case GGML_TYPE_F32:
-            case GGML_TYPE_F16:
-            case GGML_TYPE_Q4_0:
-            case GGML_TYPE_Q4_1:
-            case GGML_TYPE_Q5_0:
-            case GGML_TYPE_Q5_1:
-            case GGML_TYPE_Q8_0:
-            case GGML_TYPE_Q2_K:
-            case GGML_TYPE_Q3_K:
-            case GGML_TYPE_Q4_K:
-            case GGML_TYPE_Q5_K:
-            case GGML_TYPE_Q6_K:
-                break;
-            default: GGML_ASSERT(false);
+    struct ggml_tensor * create_tensor(struct ggml_context * ctx, const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
+        struct ggml_tensor * cur = ggml_get_tensor(ctx_meta, name.c_str());
+
+        {
+            bool is_ok = true;
+            for (size_t i = 0; i < ne.size(); ++i) {
+                if (ne[i] != cur->ne[i]) {
+                    is_ok = false;
+                    break;
+                }
+            }
+            if (!is_ok) {
+                throw std::runtime_error(
+                        format("%s: tensor '%s' has wrong shape; expected %s, got %s",
+                            __func__, name.c_str(),
+                            llama_format_tensor_shape(ne).c_str(),
+                            llama_format_tensor_shape(cur).c_str()));
+            }
         }
 
-        write_tensor_info(tensor, new_type);
-        file.write_raw(new_data, new_size);
-        size_t padded_size = GGML_PAD(new_size, GGUF_DEFAULT_ALIGNMENT); // TODO: handle custom alignment
-        size_t pad = padded_size - new_size;
-        file.write_zeros(pad);
-        tensor_offset += padded_size; // offset of the next tensor
+        return create_tensor_for(ctx, cur, backend);
     }
-};
 
-struct llama_model_loader {
-    std::unique_ptr<gguf_file_loader> file_loader;
-    gguf_load_tensors_map tensors_map;
-    bool use_mmap;
-    size_t num_ggml_tensors_created = 0;
-    struct ggml_context * ggml_ctx = NULL;
-    std::unique_ptr<gguf_mmap> mapping;
-
-    llama_model_loader(const std::string & fname_base, bool use_mmap) {
-        file_loader = std::unique_ptr<gguf_file_loader>(new gguf_file_loader(fname_base.c_str(), tensors_map));
-        if (!gguf_mmap::SUPPORTED) {
-            use_mmap = false;
+    void done_getting_tensors() const {
+        if (n_created != n_tensors) {
+            throw std::runtime_error(format("%s: wrong number of tensors; expected %d, got %d", __func__, n_tensors, n_created));
         }
-        this->use_mmap = use_mmap;
     }
 
-    void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
-        *ctx_size_p = *mmapped_size_p = 0;
-        for (const gguf_load_tensor & lt : tensors_map.tensors) {
-            *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
-            *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
-        }
-    }
+    size_t file_offset(const char * name) const {
+        const int idx = gguf_find_tensor(ctx_gguf, name);
 
-    struct ggml_tensor * get_tensor(const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
-        auto it = tensors_map.name_to_idx.find(name);
-        if (it == tensors_map.name_to_idx.end()) {
-            throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
-        }
-        gguf_load_tensor & lt = tensors_map.tensors.at(it->second);
-        if (lt.ne != ne) {
-            throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
-                         name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
+        if (idx < 0) {
+            throw std::runtime_error(format("%s: tensor '%s' not found in the file", __func__, name));
         }
 
-        return get_tensor_for(lt, backend);
+        return gguf_get_data_offset(ctx_gguf) + gguf_get_tensor_offset(ctx_gguf, idx);
     }
 
-    struct ggml_tensor * get_tensor_for(gguf_load_tensor & lt, ggml_backend backend) {
-        struct ggml_tensor * tensor;
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, true);
-        }
-        if (lt.ne.size() == 2) {
-            tensor = ggml_new_tensor_2d(ggml_ctx, lt.type, lt.ne.at(0), lt.ne.at(1));
-        } else {
-            GGML_ASSERT(lt.ne.size() == 1);
-            tensor = ggml_new_tensor_1d(ggml_ctx, lt.type, lt.ne.at(0));
-        }
-        ggml_set_name(tensor, lt.name.c_str());
-        GGML_ASSERT(lt.ggml_tensor == NULL); // if this fails, we called get_tensor twice on the same tensor
+    void load_data_for(struct ggml_tensor * cur) const {
+        const size_t offs = file_offset(ggml_get_name(cur));
 
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, use_mmap);
+        if (use_mmap) {
+            cur->data = (uint8_t *) mapping->addr + offs;
+        } else {
+            file.seek(offs, SEEK_SET);
+            file.read_raw(cur->data, ggml_nbytes(cur));
         }
-        tensor->backend = backend;
-        lt.ggml_tensor = tensor;
-        num_ggml_tensors_created++;
-        return tensor;
     }
 
-    void done_getting_tensors() const {
-        if (num_ggml_tensors_created != tensors_map.tensors.size()) {
-            throw std::runtime_error(std::string("llama.cpp: file contained more tensors than expected"));
-        }
-    }
+    void load_all_data(struct ggml_context * ctx, llama_progress_callback progress_callback, void * progress_callback_user_data, llama_mlock * lmlock) {
+        size_t size_data = 0;
+        size_t size_lock = 0;
+        size_t size_pref = 0; // prefetch
 
-    void load_all_data(llama_progress_callback progress_callback, void *  progress_callback_user_data, gguf_mlock * lmlock) {
-        size_t data_size = 0;
-        size_t prefetch_size = 0;
-        size_t lock_size = 0;
-        for (const gguf_load_tensor & lt : tensors_map.tensors) {
-            data_size += lt.size;
-            if (lt.ggml_tensor->backend == GGML_BACKEND_CPU) {
-                prefetch_size += lt.size;
+        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
+            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
+            size_data += ggml_nbytes(cur);
+            if (cur->backend == GGML_BACKEND_CPU) {
+                size_pref += ggml_nbytes(cur);
             }
         }
 
         if (use_mmap) {
-            mapping.reset(new gguf_mmap(&file_loader->file, prefetch_size, ggml_is_numa()));
+            mapping.reset(new llama_mmap(&file, size_pref, ggml_is_numa()));
             if (lmlock) {
                 lmlock->init(mapping->addr);
             }
         }
 
         size_t done_size = 0;
-        for (gguf_load_tensor & lt : tensors_map.tensors) {
+        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
+            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
+            GGML_ASSERT(cur); // unused tensors should have been caught by load_data already
+
             if (progress_callback) {
-                progress_callback((float) done_size / data_size, progress_callback_user_data);
+                progress_callback((float) done_size / size_data, progress_callback_user_data);
             }
-            GGML_ASSERT(lt.ggml_tensor); // unused tensors should have been caught by load_data already
-            lt.data = (uint8_t *) lt.ggml_tensor->data;
 
             // allocate temp buffer if not using mmap
-            if (!use_mmap && lt.data == NULL) {
-                GGML_ASSERT(lt.ggml_tensor->backend != GGML_BACKEND_CPU);
-                lt.data = (uint8_t*)malloc(ggml_nbytes(lt.ggml_tensor));
+            if (!use_mmap && cur->data == NULL) {
+                GGML_ASSERT(cur->backend != GGML_BACKEND_CPU);
+                cur->data = malloc(ggml_nbytes(cur));
             }
 
-            load_data_for(lt);
+            load_data_for(cur);
 
-            switch(lt.ggml_tensor->backend) {
+            switch (cur->backend) {
                 case GGML_BACKEND_CPU:
-                    lt.ggml_tensor->data = lt.data;
                     if (use_mmap && lmlock) {
-                        lock_size += lt.size;
-                        lmlock->grow_to(lock_size);
+                        size_lock += ggml_nbytes(cur);
+                        lmlock->grow_to(size_lock);
                     }
                     break;
 #if defined(GGML_USE_CUBLAS)
                 case GGML_BACKEND_GPU:
                 case GGML_BACKEND_GPU_SPLIT:
-                    ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+                    // old code:
+                    //ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+
+                    // TODO: test if this works !!
+                    ggml_cuda_transform_tensor(cur->data, cur);
                     if (!use_mmap) {
-                        free(lt.data);
+                        free(cur->data);
                     }
                     break;
 #elif defined(GGML_USE_CLBLAST)
                 case GGML_BACKEND_GPU:
-                    ggml_cl_transform_tensor(lt.data, lt.ggml_tensor);
+                    ggml_cl_transform_tensor(cur->data, cur);
                     if (!use_mmap) {
-                        free(lt.data);
+                        free(cur->data);
                     }
                     break;
 #endif
@@ -1028,277 +1221,175 @@ struct llama_model_loader {
                     continue;
             }
 
-            done_size += lt.size;
+            done_size += ggml_nbytes(cur);
         }
     }
+};
 
-    void load_data_for(gguf_load_tensor & lt) {
-        if (use_mmap) {
-            lt.data = (uint8_t *) mapping->addr + lt.file_off;
-        } else {
-            gguf_file & file = file_loader->file;
-            file.seek(lt.file_off, SEEK_SET);
-            file.read_raw(lt.data, lt.size);
-        }
+//
+// load LLaMA models
+//
 
-        if (0) {
-            print_checksum(lt);
-        }
+static const char * llama_ftype_name(enum llama_ftype ftype) {
+    switch (ftype) {
+        case LLAMA_FTYPE_ALL_F32:     return "all F32";
+        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
+        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
+        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
+        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
+                                      return "mostly Q4_1, some F16";
+        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
+        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
+        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
+
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K:   return "mostly Q2_K";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q6_K:   return "mostly Q6_K";
+
+        default: return "unknown, may not work";
     }
+}
 
-    static void print_checksum(gguf_load_tensor & lt) {
-        uint32_t sum = 0;
-        for (size_t i = 0; i < lt.size; i++) {
-            uint8_t byte = lt.data[i];
-            sum = byte + (sum << 6) + (sum << 16) - sum; // sdbm hash
-        }
-        fprintf(stderr, "%s checksum: %#08x (%s, size %zu)\n", lt.name.c_str(), sum,
-                llama_format_tensor_shape(lt.ne).c_str(), lt.size);
+static const char * llama_model_type_name(e_model type) {
+    switch (type) {
+        case MODEL_3B:  return "3B";
+        case MODEL_7B:  return "7B";
+        case MODEL_13B: return "13B";
+        case MODEL_30B: return "30B";
+        case MODEL_65B: return "65B";
+        case MODEL_70B: return "70B";
+        default: GGML_ASSERT(false);
     }
+}
 
-};
+static void llama_model_load_internal(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        const bool mul_mat_q,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void * progress_callback_user_data) {
+    model.t_start_us = ggml_time_us();
 
-//
-// kv cache
-//
+    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
 
-static bool kv_cache_init(
-        const struct llama_hparams & hparams,
-             struct llama_kv_cache & cache,
-                         ggml_type   wtype,
-                               int   n_ctx,
-                               int   n_gpu_layers) {
-    const int n_embd  = hparams.n_embd_gqa();
-    const int n_layer = hparams.n_layer;
+    model.n_gpu_layers = n_gpu_layers;
 
-    const int64_t n_mem      = n_layer*n_ctx;
-    const int64_t n_elements = n_embd*n_mem;
+    auto & hparams = model.hparams;
 
-    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
-    cache.n = 0;
+    // read hparams
+    {
+        struct gguf_context * ctx = ml->ctx_gguf;
+
+#define GGUF_GET(dst, func, type, req, key) \
+        { \
+            const int kid = gguf_find_key(ctx, key); \
+            if (kid >= 0) { \
+                enum gguf_type ktype = gguf_get_kv_type(ctx, kid); \
+                if (ktype != (type)) { \
+                    throw std::runtime_error(format("key %s has wrong type: %s", key, gguf_type_name(ktype))); \
+                } \
+                (dst) = func(ctx, kid); \
+            } else if (req) { \
+                throw std::runtime_error(format("key not found in model: %s", key)); \
+            } \
+        }
 
-    struct ggml_init_params params;
-    params.mem_size   = cache.buf.size;
-    params.mem_buffer = cache.buf.data;
-    params.no_alloc   = false;
+        GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
+        GGUF_GET(hparams.n_ctx,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
+        GGUF_GET(hparams.n_embd,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.embedding_length");
+        GGUF_GET(hparams.n_ff,           gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.feed_forward_length");
+        GGUF_GET(hparams.n_head,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.attention.head_count");
+        GGUF_GET(hparams.n_layer,        gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.block_count");
+        GGUF_GET(hparams.n_rot,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.rope.dimension_count");
+        GGUF_GET(hparams.f_norm_rms_eps, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, "llama.attention.layer_norm_rms_epsilon");
 
-    cache.ctx = ggml_init(params);
+        // n_head_kv is optional, default to n_head
+        hparams.n_head_kv = hparams.n_head;
+        GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
+#undef GGUF_GET
 
-    if (!cache.ctx) {
-        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
-        return false;
-    }
+        switch (hparams.n_layer) {
+            case 26: model.type = e_model::MODEL_3B; break;
+            case 32: model.type = e_model::MODEL_7B; break;
+            case 40: model.type = e_model::MODEL_13B; break;
+            case 60: model.type = e_model::MODEL_30B; break;
+            case 80: model.type = e_model::MODEL_65B; break;
+            default:
+                {
+                    if (hparams.n_layer < 32) {
+                        model.type = e_model::MODEL_7B;
+                    }
+                } break;
+        }
 
-    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    ggml_set_name(cache.k, "cache_k");
-    ggml_set_name(cache.v, "cache_v");
+        hparams.n_ctx = n_ctx;
 
-    (void) n_gpu_layers;
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer + 1) {
-        ggml_cuda_assign_buffers_no_scratch(cache.v);
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        ggml_cuda_assign_buffers_no_scratch(cache.k);
+        // LLaMAv2
+        // TODO: probably not needed
+        {
+            const auto n_gqa = hparams.n_gqa();
+
+            if (model.type == e_model::MODEL_65B && n_gqa == 8) {
+                LLAMA_LOG_WARN("%s: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
+                model.type = e_model::MODEL_70B;
+            }
+        }
+
+        hparams.rope_freq_base  = rope_freq_base;
+        hparams.rope_freq_scale = rope_freq_scale;
     }
-#endif // GGML_USE_CUBLAS
 
-    return true;
-}
+    // read vocab
+    {
+        struct gguf_context * ctx = ml->ctx_gguf;
 
-struct llama_context_params llama_context_default_params() {
-    struct llama_context_params result = {
-        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
-        /*.n_ctx                       =*/ 512,
-        /*.n_batch                     =*/ 512,
-        /*.gpu_layers                  =*/ 0,
-        /*.main_gpu                    =*/ 0,
-        /*.tensor_split                =*/ nullptr,
-        /*.rope_freq_base              =*/ 10000.0f,
-        /*.rope_freq_scale             =*/ 1.0f,
-        /*.progress_callback           =*/ nullptr,
-        /*.progress_callback_user_data =*/ nullptr,
-        /*.low_vram                    =*/ false,
-        /*.mul_mat_q                   =*/ false,
-        /*.f16_kv                      =*/ true,
-        /*.logits_all                  =*/ false,
-        /*.vocab_only                  =*/ false,
-        /*.use_mmap                    =*/ true,
-        /*.use_mlock                   =*/ false,
-        /*.embedding                   =*/ false,
-    };
-
-    return result;
-}
-
-struct llama_model_quantize_params llama_model_quantize_default_params() {
-    struct llama_model_quantize_params result = {
-        /*.nthread                     =*/ 0,
-        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
-        /*.allow_requantize            =*/ false,
-        /*.quantize_output_tensor      =*/ true,
-    };
-
-    return result;
-}
-
-int llama_max_devices() {
-    return LLAMA_MAX_DEVICES;
-}
-
-bool llama_mmap_supported() {
-    return gguf_mmap::SUPPORTED;
-}
-
-bool llama_mlock_supported() {
-    return gguf_mlock::SUPPORTED;
-}
-
-void llama_backend_init(bool numa) {
-    ggml_time_init();
-
-    // needed to initialize f16 tables
-    {
-        struct ggml_init_params params = { 0, NULL, false };
-        struct ggml_context * ctx = ggml_init(params);
-        ggml_free(ctx);
-    }
-
-    if (numa) {
-        ggml_numa_init();
-    }
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_init();
-#endif
-}
-
-void llama_backend_free() {
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_free();
-#endif
-}
-
-int64_t llama_time_us() {
-    return ggml_time_us();
-}
-
-//
-// model loading
-//
+        vocab.id_to_token.resize(hparams.n_vocab);
 
-static const char *gguf_file_version_name(gguf_file_version version) {
-    switch (version) {
-        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
+        const int token_idx = gguf_find_key(ctx, "tokenizer.ggml.tokens");
+        if (token_idx == -1) {
+            throw std::runtime_error("cannot find token list in GGUF file\n");
         }
 
-    return "unknown";
-}
-
-static const char *llama_ftype_name(enum llama_ftype ftype) {
-    switch (ftype) {
-        case LLAMA_FTYPE_ALL_F32:     return "all F32";
-        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
-        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
-        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
-        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
-                                      return "mostly Q4_1, some F16";
-        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
-        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
-        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K: return "mostly Q2_K";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q6_K: return "mostly Q6_K";
-        default:                      return "unknown, may not work";
-    }
-}
-
-static const char *llama_model_type_name(e_model type) {
-    switch (type) {
-        case MODEL_3B: return "3B";
-        case MODEL_7B: return "7B";
-        case MODEL_13B: return "13B";
-        case MODEL_30B: return "30B";
-        case MODEL_65B: return "65B";
-        case MODEL_70B: return "70B";
-        default: GGML_ASSERT(false);
-    }
-}
-
-static void llama_model_load_internal(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        const bool mul_mat_q,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void * progress_callback_user_data) {
-    model.t_start_us = ggml_time_us();
-
-    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
-
-    vocab = std::move(ml->file_loader->vocab);
-    model.hparams = ml->file_loader->hparams;
-    model.n_gpu_layers = n_gpu_layers;
-    gguf_file_version file_version = ml->file_loader->file_version;
-
-    auto & hparams = model.hparams;
-
-    {
-        switch (hparams.n_layer) {
-            case 26: model.type = e_model::MODEL_3B; break;
-            case 32: model.type = e_model::MODEL_7B; break;
-            case 40: model.type = e_model::MODEL_13B; break;
-            case 60: model.type = e_model::MODEL_30B; break;
-            case 80: model.type = e_model::MODEL_65B; break;
-            default:
-                {
-                    if (hparams.n_layer < 32) {
-                        model.type = e_model::MODEL_7B;
-                    }
-                } break;
+        const int score_idx = gguf_find_key(ctx, "tokenizer.ggml.scores");
+        if (score_idx == -1) {
+            throw std::runtime_error("cannot find token scores list in GGUF file\n");
         }
 
-        hparams.n_ctx = n_ctx;
+        const float * scores = (const float * ) gguf_get_arr_data(ctx, score_idx);
 
-        // LLaMAv2
-        // TODO: probably not needed
-        {
-            const auto n_gqa = hparams.n_gqa();
+        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
+            std::string word = gguf_get_arr_str(ctx, token_idx, i);
 
-            if (model.type == e_model::MODEL_65B && n_gqa == 8) {
-                fprintf(stderr, "%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
-                model.type = e_model::MODEL_70B;
-            }
-        }
+            vocab.token_to_id[word] = i;
 
-        hparams.rope_freq_base  = rope_freq_base;
-        hparams.rope_freq_scale = rope_freq_scale;
+            auto & tok_score = vocab.id_to_token[i];
+            tok_score.tok = std::move(word);
+            tok_score.score = scores[i];
+        }
     }
 
-    const uint32_t n_ff = hparams.n_ff;
-
     {
-        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, gguf_file_version_name(file_version));
+        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, llama_file_version_name(ml->file_version));
         LLAMA_LOG_INFO("%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
         LLAMA_LOG_INFO("%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
         LLAMA_LOG_INFO("%s: n_embd     = %u\n",   __func__, hparams.n_embd);
@@ -1307,8 +1398,8 @@ static void llama_model_load_internal(
         LLAMA_LOG_INFO("%s: n_layer    = %u\n",   __func__, hparams.n_layer);
         LLAMA_LOG_INFO("%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
         LLAMA_LOG_INFO("%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
-        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, n_ff);
+        LLAMA_LOG_INFO("%s: f_norm_eps = %.1e\n", __func__, hparams.f_norm_rms_eps);
+        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, hparams.n_ff);
         LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
         LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
         LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
@@ -1323,7 +1414,9 @@ static void llama_model_load_internal(
 
     size_t ctx_size;
     size_t mmapped_size;
-    ml->calc_sizes(&ctx_size, &mmapped_size);
+
+    ml->calc_sizes(ctx_size, mmapped_size);
+
     LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
 
     // create the ggml context
@@ -1372,9 +1465,7 @@ static void llama_model_load_internal(
         const uint32_t n_layer    = hparams.n_layer;
         const uint32_t n_vocab    = hparams.n_vocab;
 
-        ml->ggml_ctx = ctx;
-
-        model.tok_embeddings = ml->get_tensor("token_embd.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
+        model.tok_embeddings = ml->create_tensor(ctx, TN_TOKEN_EMBD, {n_embd, n_vocab}, GGML_BACKEND_CPU);
 
         // "output" tensor
         {
@@ -1395,8 +1486,8 @@ static void llama_model_load_internal(
                 backend_output = GGML_BACKEND_CPU;
             }
 
-            model.norm   = ml->get_tensor("output_norm.weight", {n_embd},          backend_norm);
-            model.output = ml->get_tensor("output.weight",      {n_embd, n_vocab}, backend_output);
+            model.norm   = ml->create_tensor(ctx, TN_OUTPUT_NORM, {n_embd},          backend_norm);
+            model.output = ml->create_tensor(ctx, TN_OUTPUT,      {n_embd, n_vocab}, backend_output);
             if (backend_norm == GGML_BACKEND_GPU) {
                 vram_weights += ggml_nbytes(model.norm);
             }
@@ -1405,6 +1496,8 @@ static void llama_model_load_internal(
             }
         }
 
+        const uint32_t n_ff = hparams.n_ff;
+
         const int i_gpu_start = n_layer - n_gpu_layers;
 
         model.layers.resize(n_layer);
@@ -1413,21 +1506,18 @@ static void llama_model_load_internal(
             const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
 
             auto & layer = model.layers[i];
+            layer.attention_norm = ml->create_tensor(ctx, format(TN_ATTN_NORM, i), {n_embd}, backend);
 
-            std::string layers_i = "blk." + std::to_string(i);
-
-            layer.attention_norm = ml->get_tensor(layers_i + ".attn_norm.weight", {n_embd}, backend);
-
-            layer.wq = ml->get_tensor(layers_i + ".attn_q.weight",      {n_embd, n_embd},     backend_split);
-            layer.wk = ml->get_tensor(layers_i + ".attn_k.weight",      {n_embd, n_embd_gqa}, backend_split);
-            layer.wv = ml->get_tensor(layers_i + ".attn_v.weight",      {n_embd, n_embd_gqa}, backend_split);
-            layer.wo = ml->get_tensor(layers_i + ".attn_output.weight", {n_embd, n_embd},     backend_split);
+            layer.wq = ml->create_tensor(ctx, format(TN_ATTN_Q, i),      {n_embd, n_embd},     backend_split);
+            layer.wk = ml->create_tensor(ctx, format(TN_ATTN_K, i),      {n_embd, n_embd_gqa}, backend_split);
+            layer.wv = ml->create_tensor(ctx, format(TN_ATTN_V, i),      {n_embd, n_embd_gqa}, backend_split);
+            layer.wo = ml->create_tensor(ctx, format(TN_ATTN_OUTPUT, i), {n_embd, n_embd},     backend_split);
 
-            layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
+            layer.ffn_norm = ml->create_tensor(ctx, format(TN_FFN_NORM, i), {n_embd}, backend);
 
-            layer.w1 = ml->get_tensor(layers_i + ".ffn_gate.weight", {n_embd,   n_ff}, backend_split);
-            layer.w2 = ml->get_tensor(layers_i + ".ffn_down.weight", {  n_ff, n_embd}, backend_split);
-            layer.w3 = ml->get_tensor(layers_i + ".ffn_up.weight",   {n_embd,   n_ff}, backend_split);
+            layer.w1 = ml->create_tensor(ctx, format(TN_FFN_GATE, i), {n_embd,   n_ff}, backend_split);
+            layer.w2 = ml->create_tensor(ctx, format(TN_FFN_DOWN, i), {  n_ff, n_embd}, backend_split);
+            layer.w3 = ml->create_tensor(ctx, format(TN_FFN_UP, i),   {n_embd,   n_ff}, backend_split);
 
             if (backend == GGML_BACKEND_GPU) {
                 vram_weights +=
@@ -1525,8 +1615,9 @@ static void llama_model_load_internal(
     }
 
     // populate `tensors_by_name`
-    for (gguf_load_tensor & lt : ml->tensors_map.tensors) {
-        model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
+    for (int i = 0; i < ml->n_tensors; ++i) {
+        struct ggml_tensor * cur = ggml_get_tensor(ctx, ml->get_tensor_name(i));
+        model.tensors_by_name.emplace_back(ggml_get_name(cur), cur);
     }
 
     (void) tensor_split;
@@ -1536,7 +1627,7 @@ static void llama_model_load_internal(
     }
 #endif
 
-    ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
+    ml->load_all_data(ctx, progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
 
     if (progress_callback) {
         progress_callback(1.0f, progress_callback_user_data);
@@ -1586,7 +1677,7 @@ static struct ggml_cgraph * llama_build_graph(
                    int   n_tokens,
                    int   n_past) {
 
-    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
 
     const int N = n_tokens;
 
@@ -1605,19 +1696,17 @@ static struct ggml_cgraph * llama_build_graph(
     const int64_t n_embd_head = hparams.n_embd_head();
     const int64_t n_embd_gqa  = hparams.n_embd_gqa();
 
-
     GGML_ASSERT(n_embd_head == hparams.n_rot);
 
-    const float freq_base  = hparams.rope_freq_base;
-    const float freq_scale = hparams.rope_freq_scale;
-    const float rms_norm_eps = hparams.f_rms_norm_eps;
+    const float freq_base    = hparams.rope_freq_base;
+    const float freq_scale   = hparams.rope_freq_scale;
+    const float norm_rms_eps = hparams.f_norm_rms_eps;
 
     const int n_gpu_layers = model.n_gpu_layers;
 
     auto & mem_per_token = lctx.mem_per_token;
     auto & buf_compute   = lctx.buf_compute;
 
-
     struct ggml_init_params params = {
         /*.mem_size   =*/ buf_compute.size,
         /*.mem_buffer =*/ buf_compute.data,
@@ -1714,11 +1803,11 @@ static struct ggml_cgraph * llama_build_graph(
 
         struct ggml_tensor * inpSA = inpL;
 
-        lctx.use_buf(ctx0, 0);
+        llama_context::use_buf(ctx0, 0);
 
         // norm
         {
-            cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+            cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
             offload_func(cur);
             ggml_set_name(cur, "rms_norm_0");
 
@@ -1853,7 +1942,7 @@ static struct ggml_cgraph * llama_build_graph(
             ggml_set_name(cur, "result_wo");
         }
 
-        lctx.use_buf(ctx0, 1);
+        llama_context::use_buf(ctx0, 1);
 
         struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpSA);
         offload_func(inpFF);
@@ -1863,7 +1952,7 @@ static struct ggml_cgraph * llama_build_graph(
         {
             // norm
             {
-                cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps);
+                cur = ggml_rms_norm(ctx0, inpFF, norm_rms_eps);
                 offload_func(cur);
                 ggml_set_name(cur, "rms_norm_1");
 
@@ -1909,11 +1998,11 @@ static struct ggml_cgraph * llama_build_graph(
         inpL = cur;
     }
 
-    lctx.use_buf(ctx0, 0);
+    llama_context::use_buf(ctx0, 0);
 
     // norm
     {
-        cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+        cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
         offload_func_nr(cur);
         ggml_set_name(cur, "rms_norm_2");
 
@@ -1927,7 +2016,7 @@ static struct ggml_cgraph * llama_build_graph(
     cur = ggml_mul_mat(ctx0, model.output, cur);
     ggml_set_name(cur, "result_output");
 
-    lctx.use_buf(ctx0, -1);
+    llama_context::use_buf(ctx0, -1);
 
     // logits -> probs
     //cur = ggml_soft_max_inplace(ctx0, cur);
@@ -1970,7 +2059,7 @@ static bool llama_eval_internal(
                    int   n_threads,
             const char * cgraph_fname) {
 
-    GGML_ASSERT((!tokens && embd) || (tokens && !embd));
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
 
     const int64_t t_start_us = ggml_time_us();
 
@@ -2166,15 +2255,15 @@ static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
 }
 
 static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
-    UNUSED(vocab);
-    UNUSED(token);
+    GGML_UNUSED(vocab);
+    GGML_UNUSED(token);
     // TODO: improve?
     return false;
 }
 
 static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
-    UNUSED(vocab);
-    UNUSED(token);
+    GGML_UNUSED(vocab);
+    GGML_UNUSED(token);
     // TODO: improve?
     return false;
 }
@@ -2447,8 +2536,8 @@ std::vector<uint32_t> decode_utf8(const char * src) {
 // returns true iff pos points to the end of one of the definitions of a rule
 static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
     switch (pos->type) {
-        case LLAMA_GRETYPE_END: return true;
-        case LLAMA_GRETYPE_ALT: return true;
+        case LLAMA_GRETYPE_END: return true;  // NOLINT
+        case LLAMA_GRETYPE_ALT: return true;  // NOLINT
         default:                return false;
     }
 }
@@ -2461,7 +2550,7 @@ static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
 
     bool found            = false;
     bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
-    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT);
+    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT); // NOLINT
 
     do {
         if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
@@ -2596,7 +2685,7 @@ static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_
         }
     }
 
-    auto stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
+    const auto * stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
 
     // update top of stack to next element, if any
     std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
@@ -2829,7 +2918,6 @@ void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array *
     }
 }
 
-
 void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
     // Reference implementation:
     // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
@@ -2997,9 +3085,8 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
         }
     }
 
-    const auto rejects =
-        llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
-    for (auto & reject : rejects) {
+    const auto rejects = llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
+    for (const auto & reject : rejects) {
         candidates->data[reject.index].logit = -INFINITY;
     }
 
@@ -3208,35 +3295,35 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
 // quantization
 //
 
-static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, std::vector<float> & output, const size_t nelements, const int nthread) {
+static void llama_convert_tensor_internal(struct ggml_tensor * tensor, std::vector<float> & output, const size_t nelements, const int nthread) {
     if (output.size() < nelements) {
         output.resize(nelements);
     }
     float * f32_output = (float *) output.data();
 
     ggml_type_traits_t qtype;
-    if (ggml_is_quantized(tensor.type)) {
-        qtype = ggml_internal_get_type_traits(tensor.type);
+    if (ggml_is_quantized(tensor->type)) {
+        qtype = ggml_internal_get_type_traits(tensor->type);
         if (qtype.to_float == NULL) {
-            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor.type)));
+            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor->type)));
         }
-    } else if (tensor.type != GGML_TYPE_F16) {
-        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor.type)));
+    } else if (tensor->type != GGML_TYPE_F16) {
+        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor->type)));
     }
 
     if (nthread < 2) {
-        if (tensor.type == GGML_TYPE_F16) {
-            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor.data, f32_output, nelements);
-        } else if (ggml_is_quantized(tensor.type)) {
-            qtype.to_float(tensor.data, f32_output, nelements);
+        if (tensor->type == GGML_TYPE_F16) {
+            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor->data, f32_output, nelements);
+        } else if (ggml_is_quantized(tensor->type)) {
+            qtype.to_float(tensor->data, f32_output, nelements);
         } else {
             GGML_ASSERT(false); // unreachable
         }
         return;
     }
 
-    auto block_size = tensor.type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor.type);
-    auto block_size_bytes = ggml_type_size(tensor.type);
+    auto block_size = tensor->type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor->type);
+    auto block_size_bytes = ggml_type_size(tensor->type);
 
     GGML_ASSERT(nelements % block_size == 0);
     auto nblocks = nelements / block_size;
@@ -3256,14 +3343,13 @@ static void llama_convert_tensor_internal(const gguf_load_tensor & tensor, std::
                 qtype.to_float(inbuf, outbuf, nels);
             }
         };
-        workers.push_back(std::thread(compute, tensor.type, tensor.data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
+        workers.push_back(std::thread(compute, tensor->type, (uint8_t *) tensor->data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
         in_buff_offs += thr_block_bytes;
         out_buff_offs += thr_elems;
     }
     for (auto & worker : workers) {
         worker.join();
     }
-
 }
 
 static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
@@ -3300,16 +3386,27 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     }
 
     std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
-    gguf_file_saver file_saver(fname_out.c_str(), model_loader->file_loader.get(), params->ftype);
+
+    const size_t align = GGUF_DEFAULT_ALIGNMENT;
+    struct gguf_context * ctx_out = gguf_init_empty();
+
+    // copy the KV pairs from the input file
+    gguf_set_kv     (ctx_out, model_loader->ctx_gguf);
+    gguf_set_val_u32(ctx_out, "general.quantization_version", GGML_QNT_VERSION);
 
 #ifdef GGML_USE_K_QUANTS
     int n_attention_wv    = 0;
     int n_feed_forward_w2 = 0;
-    for (auto& tensor : model_loader->tensors_map.tensors) {
-        if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
+
+        const std::string name = ggml_get_name(meta);
+
+        if (name.find("attn_v.weight") != std::string::npos) {
             ++n_attention_wv;
         }
-        else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+        else if (name.find("ffn_down.weight") != std::string::npos) {
             ++n_feed_forward_w2;
         }
     }
@@ -3329,48 +3426,68 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
     };
 
-    size_t idx = 0;
+    int idx = 0;
 
     std::vector<uint8_t> read_data;
     std::vector<uint8_t> work;
 
-    for (gguf_load_tensor & tensor : model_loader->tensors_map.tensors) {
-        read_data.resize(tensor.size);
-        tensor.data = read_data.data();
+    // populate the original tensors so we get an initial meta data
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
+        gguf_add_tensor(ctx_out, meta);
+    }
+
+    std::ofstream fout(fname_out, std::ios::binary);
+
+    const size_t meta_size = gguf_get_meta_size(ctx_out);
+
+    LLAMA_LOG_INFO("%s: meta size = %zu bytes\n", __func__, meta_size);
+
+    // placeholder for the meta data
+    ::zeros(fout, meta_size);
+
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * tensor = model_loader->get_tensor_meta(i);
+
+        const std::string name = ggml_get_name(tensor);
+
+        read_data.resize(ggml_nbytes(tensor));
+        tensor->data = read_data.data();
         model_loader->load_data_for(tensor);
 
-        LLAMA_LOG_INFO("[%4zu/%4zu] %36s - %16s, type = %6s, ",
-               ++idx, model_loader->tensors_map.tensors.size(),
-               tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
-               ggml_type_name(tensor.type));
+        LLAMA_LOG_INFO("[%4d/%4d] %36s - [%s], type = %6s, ",
+               ++idx, model_loader->n_tensors,
+               ggml_get_name(tensor),
+               llama_format_tensor_shape(tensor).c_str(),
+               ggml_type_name(tensor->type));
 
         // This used to be a regex, but <regex> has an extreme cost to compile times.
-        bool quantize = tensor.name.rfind("weight") == tensor.name.size() - 6; // ends with 'weight'?
+        bool quantize = name.rfind("weight") == name.size() - 6; // ends with 'weight'?
 
         // quantize only 2D tensors
-        quantize &= (tensor.ne.size() == 2);
-        quantize &= params->quantize_output_tensor || tensor.name != "output.weight";
-        quantize &= quantized_type != tensor.type;
+        quantize &= (tensor->n_dims == 2);
+        quantize &= params->quantize_output_tensor || name != "output.weight";
+        quantize &= quantized_type != tensor->type;
 
         enum ggml_type new_type;
         void * new_data;
         size_t new_size;
 
         if (!quantize) {
-            new_type = tensor.type;
-            new_data = tensor.data;
-            new_size = tensor.size;
-            LLAMA_LOG_INFO("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
+            new_type = tensor->type;
+            new_data = tensor->data;
+            new_size = ggml_nbytes(tensor);
+            LLAMA_LOG_INFO("size = %8.3f MB\n", ggml_nbytes(tensor)/1024.0/1024.0);
         } else {
             new_type = quantized_type;
 #ifdef GGML_USE_K_QUANTS
-            if (tensor.name == "output.weight") {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
+            if (name == TN_OUTPUT) {
+                int nx = tensor->ne[0];
+                int ny = tensor->ne[1];
                 if (nx % QK_K == 0 && ny % QK_K == 0) {
                     new_type = GGML_TYPE_Q6_K;
                 }
-            } else if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+            } else if (name.find("attn_v.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
                 else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
@@ -3378,32 +3495,32 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
                         (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
                 ++i_attention_wv;
-            } else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+            } else if (name.find("feed_forward.w2.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
                 else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
                          use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
                 //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
                 ++i_feed_forward_w2;
-            } else if (tensor.name.find("attention.wo.weight") != std::string::npos) {
+            } else if (name.find("attn_output.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
             }
             bool convert_incompatible_tensor = false;
             if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
                 new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
+                int nx = tensor->ne[0];
+                int ny = tensor->ne[1];
                 if (nx % QK_K != 0 || ny % QK_K != 0) {
                     LLAMA_LOG_INFO("\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
                     convert_incompatible_tensor = true;
                 }
             }
             if (convert_incompatible_tensor) {
-                if (tensor.name == "output.weight") {
+                if (name == TN_OUTPUT) {
                     new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
                     LLAMA_LOG_WARN("F16 will be used for this tensor instead.\n");
-                } else if (tensor.name == "tok_embeddings.weight") {
+                } else if (name == TN_TOKEN_EMBD) {
                     new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
                     LLAMA_LOG_WARN("Q4_0 will be used for this tensor instead.\n");
                 } else {
@@ -3412,15 +3529,15 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             }
 #endif
 
-            const size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
+            const size_t nelements = ggml_nelements(tensor);
 
             float * f32_data;
             std::vector<float> f32_conv_buf;
 
-            if (tensor.type == GGML_TYPE_F32) {
-                f32_data = (float *) tensor.data;
-            } else if (ggml_is_quantized(tensor.type) && !params->allow_requantize) {
-                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
+            if (tensor->type == GGML_TYPE_F32) {
+                f32_data = (float *) tensor->data;
+            } else if (ggml_is_quantized(tensor->type) && !params->allow_requantize) {
+                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor->type)));
             } else {
                 llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
                 f32_data = (float *) f32_conv_buf.data();
@@ -3476,7 +3593,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 }
             }
 
-            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
+            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", ggml_nbytes(tensor)/1024.0/1024.0, new_size/1024.0/1024.0);
             int64_t tot_count = 0;
             for (size_t i = 0; i < hist_cur.size(); i++) {
                 hist_all[i] += hist_cur[i];
@@ -3490,14 +3607,34 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             }
             LLAMA_LOG_INFO("\n");
         }
-        total_size_org += tensor.size;
+        total_size_org += ggml_nbytes(tensor);
         total_size_new += new_size;
-        file_saver.write_tensor(tensor, new_type, new_data, new_size);
+
+        // update the gguf meta data as we go
+        gguf_set_tensor_type(ctx_out, name.c_str(), new_type);
+        gguf_set_tensor_data(ctx_out, name.c_str(), new_data, new_size);
+
+        // write tensor data + padding
+        fout.write((const char *) new_data, new_size);
+        zeros(fout, GGML_PAD(new_size, align) - new_size);
+    }
+
+    // go back to beginning of file and write the updated meta data
+    {
+        fout.seekp(0);
+        std::vector<uint8_t> data(gguf_get_meta_size(ctx_out));
+        gguf_get_meta_data(ctx_out, data.data());
+        fout.write((const char *) data.data(), data.size());
     }
 
+    fout.close();
+
+    gguf_free(ctx_out);
+
     LLAMA_LOG_INFO("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
     LLAMA_LOG_INFO("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
 
+    // print histogram for all tensors
     {
         int64_t sum_all = 0;
         for (size_t i = 0; i < hist_all.size(); i++) {
@@ -3514,500 +3651,574 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     }
 }
 
-//
-// interface implementation
-//
+int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
+    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
 
-struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-    ggml_time_init();
+    const int64_t t_start_lora_us = ggml_time_us();
 
-    llama_model * model = new llama_model;
+    auto fin = std::ifstream(path_lora, std::ios::binary);
+    if (!fin) {
+        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
+        return 1;
+    }
 
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+    // verify magic and version
+    {
+        uint32_t magic;
+        fin.read((char *) &magic, sizeof(magic));
+        uint32_t format_version;
+        fin.read((char *) &format_version, sizeof(format_version));
 
-    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gpu_layers,
-                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,
-                params.low_vram, memory_type, params.use_mmap, params.use_mlock, params.vocab_only,
-                params.progress_callback, params.progress_callback_user_data)) {
-        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
-        delete model;
-        return nullptr;
+        if (format_version != 1) {
+            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
+            return 1;
+        }
     }
 
-    return model;
-}
+    int32_t lora_r;
+    int32_t lora_alpha;
+    fin.read((char *) &lora_r, sizeof(lora_r));
+    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
+    float scaling = (float)lora_alpha / (float)lora_r;
 
-void llama_free_model(struct llama_model * model) {
-    delete model;
-}
+    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
 
-struct llama_context * llama_new_context_with_model(
-                 struct llama_model * model,
-        struct llama_context_params   params) {
+    // create a temporary ggml context to store the lora tensors
+    // todo: calculate size from biggest possible tensor
+    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
+    struct ggml_init_params params;
+    params.mem_size   = lora_buf.size();
+    params.mem_buffer = lora_buf.data();
+    params.no_alloc   = false;
 
-    if (!model) {
-        return nullptr;
+    ggml_context * lora_ctx = ggml_init(params);
+    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+
+    // create a name -> tensor map of the model to accelerate lookups
+    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
+    for (const auto & kv : model.tensors_by_name) {
+        model_tensors.insert(kv);
     }
 
-    llama_context * ctx = new llama_context(*model);
+    // load base model
+    std::unique_ptr<llama_model_loader> model_loader;
+    ggml_context * base_ctx = NULL;
+    std::vector<uint8_t> base_buf;
+    if (path_base_model) {
+        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
+        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
 
-    if (params.seed == LLAMA_DEFAULT_SEED) {
-        params.seed = time(NULL);
-    }
+        size_t ctx_size;
+        size_t mmapped_size;
+        model_loader->calc_sizes(ctx_size, mmapped_size);
+        base_buf.resize(ctx_size);
 
-    unsigned cur_percentage = 0;
-    if (params.progress_callback == NULL) {
-        params.progress_callback_user_data = &cur_percentage;
-        params.progress_callback = [](float progress, void * ctx) {
-            unsigned * cur_percentage_p = (unsigned *) ctx;
-            unsigned percentage = (unsigned) (100 * progress);
-            while (percentage > *cur_percentage_p) {
-                *cur_percentage_p = percentage;
-                LLAMA_LOG_INFO(".");
-                if (percentage >= 100) {
-                    LLAMA_LOG_INFO("\n");
-                }
-            }
-        };
+        ggml_init_params base_params;
+        base_params.mem_size   = base_buf.size();
+        base_params.mem_buffer = base_buf.data();
+        base_params.no_alloc   = model_loader->use_mmap;
+
+        base_ctx = ggml_init(base_params);
+
+        // maybe this should in llama_model_loader
+        if (model_loader->use_mmap) {
+            model_loader->mapping.reset(new llama_mmap(&model_loader->file, /* prefetch */ 0, ggml_is_numa()));
+        }
     }
 
-    ctx->rng = std::mt19937(params.seed);
-    ctx->logits_all = params.logits_all;
+    // read tensors and apply
+    bool warned = false;
+    int n_tensors = 0;
 
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+    std::vector<uint8_t> work_buffer;
 
-    // reserve memory for context buffers
-    if (!params.vocab_only) {
-        if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
-            LLAMA_LOG_ERROR("%s: kv_cache_init() failed for self-attention cache\n", __func__);
-            llama_free(ctx);
-            return nullptr;
+    while (true) {
+        int32_t n_dims;
+        int32_t length;
+        int32_t ftype;
+
+        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
+        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
+        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
+        if (fin.eof()) {
+            break;
         }
 
+        int32_t ne[2] = { 1, 1 };
+        for (int i = 0; i < n_dims; ++i) {
+            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
+        }
+
+        std::string name;
         {
-            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
-            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
+            char buf[1024];
+            fin.read(buf, length);
+            name = std::string(buf, length);
         }
 
-        const auto & hparams = ctx->model.hparams;
+        // check for lora suffix and get the type of tensor
+        const std::string lora_suffix = ".lora";
+        size_t pos = name.rfind(lora_suffix);
+        if (pos == std::string::npos) {
+            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
+            return 1;
+        }
 
-        // resized during inference
-        if (params.logits_all) {
-            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
-        } else {
-            ctx->logits.reserve(hparams.n_vocab);
+        std::string lora_type = name.substr(pos + lora_suffix.length());
+        std::string base_name = name;
+        base_name.erase(pos);
+        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+
+        if (model_tensors.find(base_name) == model_tensors.end()) {
+            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
+            return 1;
         }
 
-        if (params.embedding){
-            ctx->embedding.resize(hparams.n_embd);
+        // create ggml tensor
+        ggml_type wtype;
+        switch (ftype) {
+            case 0: wtype = GGML_TYPE_F32;  break;
+            case 1: wtype = GGML_TYPE_F16;  break;
+            default:
+                    {
+                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
+                                __func__, ftype);
+                        return false;
+                    }
+        }
+        ggml_tensor * lora_tensor;
+        if (n_dims == 2) {
+            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
+        }
+        else {
+            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
+            return 1;
         }
+        ggml_set_name(lora_tensor, "lora_tensor");
 
-#ifdef LLAMA_USE_ALLOCATOR
-        {
-            static const size_t tensor_alignment = 32;
-            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
-            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
+        // load tensor data
+        size_t offset = fin.tellg();
+        size_t tensor_data_size = ggml_nbytes(lora_tensor);
+        offset = (offset + 31) & -32;
+        fin.seekg(offset);
+        fin.read((char*)lora_tensor->data, tensor_data_size);
 
-            // create measure allocator
-            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
+        lora_tensors[name] = lora_tensor;
 
-            // build worst-case graph
-            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
-            int n_past = hparams.n_ctx - n_tokens;
-            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
-            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
+        // check if we have both A and B tensors and apply
+        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
+            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
 
-            // measure memory requirements for the graph
-            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
+            ggml_tensor * dest_t = model_tensors[base_name];
 
-            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
+            offload_func_t offload_func = llama_nop;
+            offload_func_t offload_func_force_inplace = llama_nop;
 
-            // debug - for comparison with scratch buffer
-            //size_t prev_req =
-            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
-            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
-            //    MEM_REQ_EVAL().at(ctx->model.type);
-            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
+#ifdef GGML_USE_CUBLAS
+            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
+                if (dest_t->type != GGML_TYPE_F16) {
+                    throw std::runtime_error(format(
+                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
+                }
+                offload_func = ggml_cuda_assign_buffers;
+                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
+            }
+#endif // GGML_USE_CUBLAS
 
-            // recreate allocator with exact memory requirements
-            ggml_allocr_free(ctx->alloc);
+            ggml_tensor * base_t;
+            if (model_loader) {
+                struct gguf_context * ctx_gguf = model_loader->ctx_gguf;
 
-            ctx->buf_alloc.resize(alloc_size);
-            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
-        }
-#else
-        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
-#endif
+                // load from base model
+                if (gguf_find_tensor(ctx_gguf, base_name.c_str()) < 0) {
+                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
+                    return 1;
+                }
 
-#ifdef LLAMA_USE_SCRATCH
-        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
-        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
-#endif
-    }
+                // TODO: not tested!! maybe not working!
+                base_t = model_loader->create_tensor(base_ctx, base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
+                model_loader->load_data_for(base_t);
+            } else {
+                base_t = dest_t;
+            }
 
-#ifdef GGML_USE_METAL
-    if (params.n_gpu_layers > 0) {
-        // this allocates all Metal resources and memory buffers
-        ctx->ctx_metal = ggml_metal_init(1);
+            if (ggml_is_quantized(base_t->type)) {
+                if (!warned) {
+                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
+                                   "use a f16 or f32 base model with --lora-base\n", __func__);
+                    warned = true;
+                }
+            }
 
-        if (!ctx->ctx_metal) {
-            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
-            llama_free(ctx);
-            return NULL;
-        }
+            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
+            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
+            ggml_set_name(loraA, "loraA");
 
-        void * data_ptr  = NULL;
-        size_t data_size = 0;
+            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
+            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
+            ggml_set_name(loraB, "loraB");
 
-        if (params.use_mmap) {
-            data_ptr  = ctx->model.mapping->addr;
-            data_size = ctx->model.mapping->size;
-        } else {
-            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
-            data_size = ggml_get_mem_size  (ctx->model.ctx);
-        }
+            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
+                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
+                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
+                return 1;
+            }
 
-        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
+            // w = w + BA*s
+            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
+            offload_func(BA);
+            ggml_set_name(BA, "BA");
 
-        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
+            if (scaling != 1.0f) {
+                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
+                ggml_set_name(scale_tensor, "scale_tensor");
 
-#define LLAMA_METAL_CHECK_BUF(result)                            \
-    if (!(result)) {                                             \
-        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
-        llama_free(ctx);                                         \
-        return NULL;                                             \
-    }
+                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
+                offload_func(BA);
+                ggml_set_name(BA, "BA_scaled");
+            }
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
+            ggml_tensor * r;
+            if (base_t == dest_t) {
+                r = ggml_add_inplace(lora_ctx, dest_t, BA);
+                offload_func_force_inplace(r);
+                ggml_set_name(r, "r_add_inplace");
+            }
+            else {
+                r = ggml_add(lora_ctx, base_t, BA);
+                offload_func(r);
+                ggml_set_name(r, "r_add");
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
+                r = ggml_cpy(lora_ctx, r, dest_t);
+                offload_func(r);
+                ggml_set_name(r, "r_cpy");
+            }
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
-#undef LLAMA_METAL_CHECK_BUF
-    }
-#endif
+            struct ggml_cgraph gf = ggml_build_forward(r);
 
-#ifdef GGML_USE_MPI
-    ctx->ctx_mpi = ggml_mpi_init();
+            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
 
-    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
-        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
-        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
-        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
-        llama_backend_free();
-        exit(1);
+            // we won't need these tensors again, reset the context to save memory
+            ggml_free(lora_ctx);
+            lora_ctx = ggml_init(params);
+            lora_tensors.clear();
+
+            n_tensors++;
+            if (n_tensors % 4 == 0) {
+                LLAMA_LOG_INFO(".");
+            }
+        }
     }
-#endif
 
-    return ctx;
+    // TODO: this should be in a destructor, it will leak on failure
+    ggml_free(lora_ctx);
+    if (base_ctx) {
+        ggml_free(base_ctx);
+    }
+
+    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
+    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
+
+    return 0;
 }
 
-struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
+//
+// interface implementation
+//
 
-    struct llama_model * model = llama_load_model_from_file(path_model, params);
-    if (!model) {
-        return nullptr;
-    }
-    struct llama_context * ctx = llama_new_context_with_model(model, params);
-    ctx->model_owner = true;
-    return ctx;
+struct llama_context_params llama_context_default_params() {
+    struct llama_context_params result = {
+        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
+        /*.n_ctx                       =*/ 512,
+        /*.n_batch                     =*/ 512,
+        /*.gpu_layers                  =*/ 0,
+        /*.main_gpu                    =*/ 0,
+        /*.tensor_split                =*/ nullptr,
+        /*.rope_freq_base              =*/ 10000.0f,
+        /*.rope_freq_scale             =*/ 1.0f,
+        /*.progress_callback           =*/ nullptr,
+        /*.progress_callback_user_data =*/ nullptr,
+        /*.low_vram                    =*/ false,
+        /*.mul_mat_q                   =*/ false,
+        /*.f16_kv                      =*/ true,
+        /*.logits_all                  =*/ false,
+        /*.vocab_only                  =*/ false,
+        /*.use_mmap                    =*/ true,
+        /*.use_mlock                   =*/ false,
+        /*.embedding                   =*/ false,
+    };
+
+    return result;
 }
 
-void llama_free(struct llama_context * ctx) {
-    delete ctx;
+struct llama_model_quantize_params llama_model_quantize_default_params() {
+    struct llama_model_quantize_params result = {
+        /*.nthread                     =*/ 0,
+        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
+        /*.allow_requantize            =*/ false,
+        /*.quantize_output_tensor      =*/ true,
+    };
+
+    return result;
 }
 
-int llama_model_quantize(
-        const char * fname_inp,
-        const char * fname_out,
-        const llama_model_quantize_params *params) {
-    try {
-        llama_model_quantize_internal(fname_inp, fname_out, params);
-        return 0;
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
-        return 1;
-    }
+int llama_max_devices(void) {
+    return LLAMA_MAX_DEVICES;
 }
 
-int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
-    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
+bool llama_mmap_supported(void) {
+    return llama_mmap::SUPPORTED;
+}
 
-    const int64_t t_start_lora_us = ggml_time_us();
+bool llama_mlock_supported(void) {
+    return llama_mlock::SUPPORTED;
+}
 
-    auto fin = std::ifstream(path_lora, std::ios::binary);
-    if (!fin) {
-        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
-        return 1;
-    }
+void llama_backend_init(bool numa) {
+    ggml_time_init();
 
-    // verify magic and version
+    // needed to initialize f16 tables
     {
-        uint32_t magic;
-        fin.read((char *) &magic, sizeof(magic));
-        uint32_t format_version;
-        fin.read((char *) &format_version, sizeof(format_version));
+        struct ggml_init_params params = { 0, NULL, false };
+        struct ggml_context * ctx = ggml_init(params);
+        ggml_free(ctx);
+    }
 
-        if (format_version != 1) {
-            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
-            return 1;
-        }
+    if (numa) {
+        ggml_numa_init();
     }
 
-    int32_t lora_r;
-    int32_t lora_alpha;
-    fin.read((char *) &lora_r, sizeof(lora_r));
-    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
-    float scaling = (float)lora_alpha / (float)lora_r;
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_init();
+#endif
+}
 
-    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
+void llama_backend_free(void) {
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_free();
+#endif
+}
 
-    // create a temporary ggml context to store the lora tensors
-    // todo: calculate size from biggest possible tensor
-    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
-    struct ggml_init_params params;
-    params.mem_size   = lora_buf.size();
-    params.mem_buffer = lora_buf.data();
-    params.no_alloc   = false;
+int64_t llama_time_us(void) {
+    return ggml_time_us();
+}
 
-    ggml_context * lora_ctx = ggml_init(params);
-    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+    ggml_time_init();
 
-    // create a name -> tensor map of the model to accelerate lookups
-    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
-    for (const auto & kv: model.tensors_by_name) {
-        model_tensors.insert(kv);
-    }
+    llama_model * model = new llama_model;
 
-    // load base model
-    std::unique_ptr<llama_model_loader> model_loader;
-    ggml_context * base_ctx = NULL;
-    std::vector<uint8_t> base_buf;
-    if (path_base_model) {
-        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
-        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
 
-        size_t ctx_size;
-        size_t mmapped_size;
-        model_loader->calc_sizes(&ctx_size, &mmapped_size);
-        base_buf.resize(ctx_size);
+    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gpu_layers,
+                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,
+                params.low_vram, memory_type, params.use_mmap, params.use_mlock, params.vocab_only,
+                params.progress_callback, params.progress_callback_user_data)) {
+        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
+        delete model;
+        return nullptr;
+    }
 
-        ggml_init_params base_params;
-        base_params.mem_size   = base_buf.size();
-        base_params.mem_buffer = base_buf.data();
-        base_params.no_alloc   = model_loader->use_mmap;
+    return model;
+}
 
-        base_ctx = ggml_init(base_params);
+void llama_free_model(struct llama_model * model) {
+    delete model;
+}
 
-        model_loader->ggml_ctx = base_ctx;
+struct llama_context * llama_new_context_with_model(
+                 struct llama_model * model,
+        struct llama_context_params   params) {
 
-        // maybe this should in llama_model_loader
-        if (model_loader->use_mmap) {
-            model_loader->mapping.reset(new gguf_mmap(&model_loader->file_loader->file, /* prefetch */ 0, ggml_is_numa()));
-        }
+    if (!model) {
+        return nullptr;
     }
 
-    // read tensors and apply
-    bool warned = false;
-    int n_tensors = 0;
+    llama_context * ctx = new llama_context(*model);
 
-    std::vector<uint8_t> work_buffer;
+    if (params.seed == LLAMA_DEFAULT_SEED) {
+        params.seed = time(NULL);
+    }
 
-    while (true) {
-        int32_t n_dims;
-        int32_t length;
-        int32_t ftype;
+    unsigned cur_percentage = 0;
+    if (params.progress_callback == NULL) {
+        params.progress_callback_user_data = &cur_percentage;
+        params.progress_callback = [](float progress, void * ctx) {
+            unsigned * cur_percentage_p = (unsigned *) ctx;
+            unsigned percentage = (unsigned) (100 * progress);
+            while (percentage > *cur_percentage_p) {
+                *cur_percentage_p = percentage;
+                LLAMA_LOG_INFO(".");
+                if (percentage >= 100) {
+                    LLAMA_LOG_INFO("\n");
+                }
+            }
+        };
+    }
 
-        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
-        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
-        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
-        if (fin.eof()) {
-            break;
-        }
+    ctx->rng = std::mt19937(params.seed);
+    ctx->logits_all = params.logits_all;
 
-        int32_t ne[2] = { 1, 1 };
-        for (int i = 0; i < n_dims; ++i) {
-            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
-        }
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
 
-        std::string name;
-        {
-            char buf[1024];
-            fin.read(buf, length);
-            name = std::string(buf, length);
+    // reserve memory for context buffers
+    if (!params.vocab_only) {
+        if (!llama_kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
+            LLAMA_LOG_ERROR("%s: llama_kv_cache_init() failed for self-attention cache\n", __func__);
+            llama_free(ctx);
+            return nullptr;
         }
 
-        // check for lora suffix and get the type of tensor
-        const std::string lora_suffix = ".lora";
-        size_t pos = name.rfind(lora_suffix);
-        if (pos == std::string::npos) {
-            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
-            return 1;
+        {
+            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
+            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
         }
 
-        std::string lora_type = name.substr(pos + lora_suffix.length());
-        std::string base_name = name;
-        base_name.erase(pos);
-        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+        const auto & hparams = ctx->model.hparams;
 
-        if (model_tensors.find(base_name) == model_tensors.end()) {
-            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
-            return 1;
+        // resized during inference
+        if (params.logits_all) {
+            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
+        } else {
+            ctx->logits.reserve(hparams.n_vocab);
         }
 
-        // create ggml tensor
-        ggml_type wtype;
-        switch (ftype) {
-            case 0: wtype = GGML_TYPE_F32;  break;
-            case 1: wtype = GGML_TYPE_F16;  break;
-            default:
-                    {
-                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
-                                __func__, ftype);
-                        return false;
-                    }
-        }
-        ggml_tensor * lora_tensor;
-        if (n_dims == 2) {
-            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
-        }
-        else {
-            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
-            return 1;
+        if (params.embedding){
+            ctx->embedding.resize(hparams.n_embd);
         }
-        ggml_set_name(lora_tensor, "lora_tensor");
 
-        // load tensor data
-        size_t offset = fin.tellg();
-        size_t tensor_data_size = ggml_nbytes(lora_tensor);
-        offset = (offset + 31) & -32;
-        fin.seekg(offset);
-        fin.read((char*)lora_tensor->data, tensor_data_size);
+#ifdef LLAMA_USE_ALLOCATOR
+        {
+            static const size_t tensor_alignment = 32;
+            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
+            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
 
-        lora_tensors[name] = lora_tensor;
+            // create measure allocator
+            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
 
-        // check if we have both A and B tensors and apply
-        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
-            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
+            // build worst-case graph
+            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
+            int n_past = hparams.n_ctx - n_tokens;
+            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
+            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
 
-            ggml_tensor * dest_t = model_tensors[base_name];
+            // measure memory requirements for the graph
+            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
 
-            offload_func_t offload_func = llama_nop;
-            offload_func_t offload_func_force_inplace = llama_nop;
+            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
 
-#ifdef GGML_USE_CUBLAS
-            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
-                if (dest_t->type != GGML_TYPE_F16) {
-                    throw std::runtime_error(format(
-                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
-                }
-                offload_func = ggml_cuda_assign_buffers;
-                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
-            }
-#endif // GGML_USE_CUBLAS
+            // debug - for comparison with scratch buffer
+            //size_t prev_req =
+            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
+            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
+            //    MEM_REQ_EVAL().at(ctx->model.type);
+            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
 
-            ggml_tensor * base_t;
-            if (model_loader) {
-                // load from base model
-                if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
-                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
-                    return 1;
-                }
-                size_t idx = model_loader->tensors_map.name_to_idx[base_name];
-                gguf_load_tensor & lt = model_loader->tensors_map.tensors[idx];
-                base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
-                lt.data = (uint8_t *) lt.ggml_tensor->data;
-                model_loader->load_data_for(lt);
-                lt.ggml_tensor->data = lt.data;
-            }
-            else {
-                base_t = dest_t;
-            }
+            // recreate allocator with exact memory requirements
+            ggml_allocr_free(ctx->alloc);
 
-            if (ggml_is_quantized(base_t->type)) {
-                if (!warned) {
-                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
-                                   "use a f16 or f32 base model with --lora-base\n", __func__);
-                    warned = true;
-                }
-            }
+            ctx->buf_alloc.resize(alloc_size);
+            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
+        }
+#else
+        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
+#endif
 
-            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
-            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
-            ggml_set_name(loraA, "loraA");
+#ifdef LLAMA_USE_SCRATCH
+        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
+        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
+#endif
+    }
 
-            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
-            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
-            ggml_set_name(loraB, "loraB");
+#ifdef GGML_USE_METAL
+    if (params.n_gpu_layers > 0) {
+        // this allocates all Metal resources and memory buffers
+        ctx->ctx_metal = ggml_metal_init(1);
 
-            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
-                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
-                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
-                return 1;
-            }
+        if (!ctx->ctx_metal) {
+            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
+            llama_free(ctx);
+            return NULL;
+        }
 
-            // w = w + BA*s
-            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
-            offload_func(BA);
-            ggml_set_name(BA, "BA");
+        void * data_ptr  = NULL;
+        size_t data_size = 0;
 
-            if (scaling != 1.0f) {
-                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
-                ggml_set_name(scale_tensor, "scale_tensor");
+        if (params.use_mmap) {
+            data_ptr  = ctx->model.mapping->addr;
+            data_size = ctx->model.mapping->size;
+        } else {
+            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
+            data_size = ggml_get_mem_size  (ctx->model.ctx);
+        }
 
-                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
-                offload_func(BA);
-                ggml_set_name(BA, "BA_scaled");
-            }
+        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
 
-            ggml_tensor * r;
-            if (base_t == dest_t) {
-                r = ggml_add_inplace(lora_ctx, dest_t, BA);
-                offload_func_force_inplace(r);
-                ggml_set_name(r, "r_add_inplace");
-            }
-            else {
-                r = ggml_add(lora_ctx, base_t, BA);
-                offload_func(r);
-                ggml_set_name(r, "r_add");
+        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
 
-                r = ggml_cpy(lora_ctx, r, dest_t);
-                offload_func(r);
-                ggml_set_name(r, "r_cpy");
-            }
+#define LLAMA_METAL_CHECK_BUF(result)                            \
+    if (!(result)) {                                             \
+        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
+        llama_free(ctx);                                         \
+        return NULL;                                             \
+    }
 
-            struct ggml_cgraph gf = ggml_build_forward(r);
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
 
-            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
 
-            // we won't need these tensors again, reset the context to save memory
-            ggml_free(lora_ctx);
-            lora_ctx = ggml_init(params);
-            lora_tensors.clear();
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
+#undef LLAMA_METAL_CHECK_BUF
+    }
+#endif
 
-            n_tensors++;
-            if (n_tensors % 4 == 0) {
-                LLAMA_LOG_INFO(".");
-            }
-        }
+#ifdef GGML_USE_MPI
+    ctx->ctx_mpi = ggml_mpi_init();
+
+    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
+        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
+        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
+        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
+        llama_backend_free();
+        exit(1);
     }
+#endif
 
-    // TODO: this should be in a destructor, it will leak on failure
-    ggml_free(lora_ctx);
-    if (base_ctx) {
-        ggml_free(base_ctx);
+    return ctx;
+}
+
+struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+
+    struct llama_model * model = llama_load_model_from_file(path_model, params);
+    if (!model) {
+        return nullptr;
     }
+    struct llama_context * ctx = llama_new_context_with_model(model, params);
+    ctx->model_owner = true;
+    return ctx;
+}
 
-    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
-    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
+void llama_free(struct llama_context * ctx) {
+    delete ctx;
+}
 
-    return 0;
+int llama_model_quantize(
+        const char * fname_inp,
+        const char * fname_out,
+        const llama_model_quantize_params * params) {
+    try {
+        llama_model_quantize_internal(fname_inp, fname_out, params);
+        return 0;
+    } catch (const std::exception & err) {
+        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
+        return 1;
+    }
 }
 
 int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
@@ -4337,15 +4548,14 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
 }
 
 static bool llama_load_session_file_internal(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
-    gguf_file file(path_session, "rb");
+    llama_file file(path_session, "rb");
     GGML_UNUSED(ctx);
     GGML_UNUSED(path_session);
     GGML_UNUSED(tokens_out);
     GGML_UNUSED(n_token_capacity);
     GGML_UNUSED(n_token_count_out);
 
-
-// TODO: implement with GGUF format
+    // TODO: implement with GGUF format
     return true;
 }
 
@@ -4359,7 +4569,7 @@ bool llama_load_session_file(struct llama_context * ctx, const char * path_sessi
 }
 
 bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
-    gguf_file file(path_session, "wb");
+    llama_file file(path_session, "wb");
     GGML_UNUSED(ctx);
     GGML_UNUSED(tokens);
     GGML_UNUSED(n_token_count);
@@ -4390,7 +4600,6 @@ int llama_eval(
     return 0;
 }
 
-
 int llama_eval_embd(
             struct llama_context * ctx,
                      const float * embd,
@@ -4577,7 +4786,7 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
             }
             strncpy(str, result.c_str(), result.length());
             return result.length();
-        } else if (llama_is_unknown_token(model->vocab, token)) {
+        } else if (llama_is_unknown_token(model->vocab, token)) { // NOLINT
             if (length < 3) {
                 return -3;
             }
@@ -4641,15 +4850,15 @@ std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token
     return std::string(result.data(), result.size());
 }
 
-llama_token llama_token_bos() {
+llama_token llama_token_bos(void) {
     return 1;
 }
 
-llama_token llama_token_eos() {
+llama_token llama_token_eos(void) {
     return 2;
 }
 
-llama_token llama_token_nl() {
+llama_token llama_token_nl(void) {
     return 13;
 }
 
@@ -4718,7 +4927,6 @@ const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_
     return ctx->model.tensors_by_name;
 }
 
-
 void llama_log_set(llama_log_callback log_callback, void * user_data) {
     g_state.log_callback = log_callback ? log_callback : llama_log_callback_default;
     g_state.log_callback_user_data = user_data;
diff --git a/gguf-llama.h b/gguf-llama.h
index f342a534c3ccb..d81a1b5defd89 100644
--- a/gguf-llama.h
+++ b/gguf-llama.h
@@ -111,6 +111,7 @@ extern "C" {
         bool use_mlock;  // force system to keep model in RAM
         bool embedding;  // embedding mode only
     };
+
     // model file types
     enum llama_ftype {
         LLAMA_FTYPE_ALL_F32              = 0,
@@ -190,17 +191,12 @@ extern "C" {
         int32_t n_eval;
     };
 
-    // Set callback for all future logging events.
-    // If this is not called, or NULL is supplied, everything is output on stderr.
-    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
-
-    LLAMA_API int llama_max_devices();
+    LLAMA_API struct llama_context_params llama_context_default_params(void);
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
 
-    LLAMA_API struct llama_context_params llama_context_default_params();
-    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
-
-    LLAMA_API bool llama_mmap_supported();
-    LLAMA_API bool llama_mlock_supported();
+    LLAMA_API int  llama_max_devices(void);
+    LLAMA_API bool llama_mmap_supported(void);
+    LLAMA_API bool llama_mlock_supported(void);
 
     // TODO: not great API - very likely to change
     // Initialize the llama + ggml backend
@@ -208,9 +204,9 @@ extern "C" {
     // Call once at the start of the program
     LLAMA_API void llama_backend_init(bool numa);
     // Call once at the end of the program - currently only used for MPI
-    LLAMA_API void llama_backend_free();
+    LLAMA_API void llama_backend_free(void);
 
-    LLAMA_API int64_t llama_time_us();
+    LLAMA_API int64_t llama_time_us(void);
 
     LLAMA_API struct llama_model * llama_load_model_from_file(
                              const char * path_model,
@@ -377,9 +373,9 @@ extern "C" {
                                   char * str,
                                   int    length);
     // Special tokens
-    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl();   // next-line
+    LLAMA_API llama_token llama_token_bos(void);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(void);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl(void);   // next-line
 
     // Grammar
     //
@@ -459,6 +455,10 @@ extern "C" {
     // Print system information
     LLAMA_API const char * llama_print_system_info(void);
 
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/gguf-util.h b/gguf-util.h
deleted file mode 100644
index d8557d94f114d..0000000000000
--- a/gguf-util.h
+++ /dev/null
@@ -1,470 +0,0 @@
-// GGUF counterpart of llama-util.h.
-// we may consider making it a part of ggml.c once GGUF work is complete.
-// this will require extra work to migrate this to pure C.
-// Contains wrappers around OS interfaces.
-
-#ifndef GGUF_UTIL_H
-#define GGUF_UTIL_H
-
-#include "ggml.h"
-
-#include <cstdio>
-#include <cstdint>
-#include <cerrno>
-#include <cstring>
-#include <cstdarg>
-#include <cstdlib>
-#include <climits>
-
-#include <string>
-#include <sstream>
-#include <vector>
-#include <stdexcept>
-
-#ifdef __has_include
-    #if __has_include(<unistd.h>)
-        #include <unistd.h>
-        #if defined(_POSIX_MAPPED_FILES)
-            #include <sys/mman.h>
-        #endif
-        #if defined(_POSIX_MEMLOCK_RANGE)
-            #include <sys/resource.h>
-        #endif
-    #endif
-#endif
-
-#if defined(_WIN32)
-    #define WIN32_LEAN_AND_MEAN
-    #ifndef NOMINMAX
-        #define NOMINMAX
-    #endif
-    #include <windows.h>
-    #include <io.h>
-    #include <stdio.h> // for _fseeki64
-#endif
-
-#ifdef __GNUC__
-#ifdef __MINGW32__
-__attribute__((format(gnu_printf, 1, 2)))
-#else
-__attribute__((format(printf, 1, 2)))
-#endif
-#endif
-static std::string format(const char * fmt, ...) {
-    va_list ap, ap2;
-    va_start(ap, fmt);
-    va_copy(ap2, ap);
-    int size = vsnprintf(NULL, 0, fmt, ap);
-    GGML_ASSERT(size >= 0 && size < INT_MAX);
-    std::vector<char> buf(size + 1);
-    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
-    GGML_ASSERT(size2 == size);
-    va_end(ap2);
-    va_end(ap);
-    return std::string(buf.data(), size);
-}
-
-// TODO: can we merge this one and gguf_context?
-struct gguf_file {
-    // use FILE * so we don't have to re-open the file to mmap
-    FILE * fp;
-    size_t size;
-
-    gguf_file(const char * fname, const char * mode) {
-        fp = std::fopen(fname, mode);
-        if (fp == NULL) {
-            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
-        }
-        seek(0, SEEK_END);
-        size = tell();
-        seek(0, SEEK_SET);
-    }
-
-    size_t tell() const {
-#ifdef _WIN32
-        __int64 ret = _ftelli64(fp);
-#else
-        long ret = std::ftell(fp);
-#endif
-        GGML_ASSERT(ret != -1); // this really shouldn't fail
-        return (size_t) ret;
-    }
-
-    void seek(size_t offset, int whence) {
-#ifdef _WIN32
-        int ret = _fseeki64(fp, (__int64) offset, whence);
-#else
-        int ret = std::fseek(fp, (long) offset, whence);
-#endif
-        GGML_ASSERT(ret == 0); // same
-    }
-
-    size_t write_str(const std::string & val) {
-        size_t total_written = 0;
-        const int32_t n = val.size();
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        total_written += sizeof(n);
-        fwrite(val.c_str(), n, 1, fp);
-        total_written += n;
-
-        return total_written;
-    }
-
-    size_t write_i32(int32_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-        return sizeof(val);
-    }
-
-    size_t write_u64(size_t val) {
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-        return sizeof(val);
-    }
-
-    template<typename T>
-    void write_val(const std::string & key, enum gguf_type type, const T & val) {
-        write_str(key);
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &val, sizeof(val), 1, fp);
-    }
-
-    template<typename T>
-    void write_arr(const std::string & key, enum gguf_type type, const std::vector<T> & val) {
-        write_str(key);
-        {
-            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
-        }
-
-        const int32_t n = val.size();
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        fwrite(val.data(), sizeof(T), n, fp);
-    }
-
-    void write_str(const std::string & key, enum gguf_type type, const std::string & val) {
-        write_str(key);
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-
-        const int32_t n = val.size();
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        fwrite(val.c_str(), n, 1, fp);
-    }
-
-    void write_str(const std::string & key, enum gguf_type type, const std::vector<std::string> & val) {
-        write_str(key);
-        {
-            const enum gguf_type tarr = GGUF_TYPE_ARRAY;
-            fwrite((const char *) &tarr, sizeof(tarr), 1, fp);
-        }
-
-        const int32_t n = val.size();
-        fwrite((const char *) &type, sizeof(type), 1, fp);
-        fwrite((const char *) &n, sizeof(n), 1, fp);
-        for (int i = 0; i < n; ++i) {
-            const int32_t nstr = val[i].size();
-            fwrite((const char *) &nstr, sizeof(nstr), 1, fp);
-            fwrite(val[i].c_str(), nstr, 1, fp);
-        }
-    }
-
-    void write_zeros(size_t count) {
-        for (size_t i = 0; i < count; ++i) {
-            fputc(0, fp);
-        }
-    }
-
-    void read_raw(void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        std::size_t ret = std::fread(ptr, len, 1, fp);
-        if (ferror(fp)) {
-            throw std::runtime_error(format("read error: %s", strerror(errno)));
-        }
-        if (ret != 1) {
-            throw std::runtime_error(std::string("unexpectedly reached end of file"));
-        }
-    }
-
-    void write_raw(const void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        size_t ret = std::fwrite(ptr, len, 1, fp);
-        if (ret != 1) {
-            throw std::runtime_error(format("write error: %s", strerror(errno)));
-        }
-    }
-
-    ~gguf_file() {
-        if (fp) {
-            std::fclose(fp);
-        }
-    }
-};
-
-#if defined(_WIN32)
-static std::string gguf_format_win_err(DWORD err) {
-    LPSTR buf;
-    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
-                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
-    if (!size) {
-        return "FormatMessageA failed";
-    }
-    std::string ret(buf, size);
-    LocalFree(buf);
-    return ret;
-}
-#endif
-
-struct gguf_mmap {
-    void * addr;
-    size_t size;
-
-    gguf_mmap(const gguf_mmap &) = delete;
-
-#ifdef _POSIX_MAPPED_FILES
-    static constexpr bool SUPPORTED = true;
-
-    gguf_mmap(struct gguf_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
-        size = file->size;
-        int fd = fileno(file->fp);
-        int flags = MAP_SHARED;
-        // prefetch/readahead impairs performance on NUMA systems
-        if (numa) { prefetch = 0; }
-#ifdef __linux__
-        if (prefetch) { flags |= MAP_POPULATE; }
-#endif
-        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
-        if (addr == MAP_FAILED) {
-            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
-        }
-
-        if (prefetch > 0) {
-            // Advise the kernel to preload the mapped memory
-            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
-                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-        if (numa) {
-            // advise the kernel not to use readahead
-            // (because the next page might not belong on the same node)
-            if (madvise(addr, file->size, MADV_RANDOM)) {
-                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-    }
-
-    ~gguf_mmap() {
-        munmap(addr, size);
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    gguf_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
-        (void) numa;
-
-        size = file->size;
-
-        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
-
-        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
-        DWORD error = GetLastError();
-
-        if (hMapping == NULL) {
-            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
-        error = GetLastError();
-        CloseHandle(hMapping);
-
-        if (addr == NULL) {
-            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
-        if (prefetch) {
-            // Advise the kernel to preload the mapped memory
-            WIN32_MEMORY_RANGE_ENTRY range;
-            range.VirtualAddress = addr;
-            range.NumberOfBytes = (SIZE_T)size;
-            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
-                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-            }
-        }
-        #else
-        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
-        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
-    }
-
-    ~gguf_mmap() {
-        if (!UnmapViewOfFile(addr)) {
-            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    gguf_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
-        (void) prefetch;
-        (void) numa;
-
-        throw std::runtime_error(std::string("mmap not supported"));
-    }
-#endif
-};
-
-// Represents some region of memory being locked using mlock or VirtualLock;
-// will automatically unlock on destruction.
-struct gguf_mlock {
-    void * addr = NULL;
-    size_t size = 0;
-    bool failed_already = false;
-
-    gguf_mlock() {}
-    gguf_mlock(const gguf_mlock &) = delete;
-
-    ~gguf_mlock() {
-        if (size) {
-            raw_unlock(addr, size);
-        }
-    }
-
-    void init(void * ptr) {
-        GGML_ASSERT(addr == NULL && size == 0);
-        addr = ptr;
-    }
-
-    void grow_to(size_t target_size) {
-        GGML_ASSERT(addr);
-        if (failed_already) {
-            return;
-        }
-        size_t granularity = lock_granularity();
-        target_size = (target_size + granularity - 1) & ~(granularity - 1);
-        if (target_size > size) {
-            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
-                size = target_size;
-            } else {
-                failed_already = true;
-            }
-        }
-    }
-
-#ifdef _POSIX_MEMLOCK_RANGE
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        return (size_t) sysconf(_SC_PAGESIZE);
-    }
-
-    #ifdef __APPLE__
-        #define MLOCK_SUGGESTION \
-            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
-            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
-    #else
-        #define MLOCK_SUGGESTION \
-            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
-    #endif
-
-    bool raw_lock(const void * addr, size_t size) {
-        if (!mlock(addr, size)) {
-            return true;
-        } else {
-            char* errmsg = std::strerror(errno);
-            bool suggest = (errno == ENOMEM);
-
-            // Check if the resource limit is fine after all
-            struct rlimit lock_limit;
-            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
-                suggest = false;
-            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
-                suggest = false;
-
-            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
-                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
-            return false;
-        }
-    }
-
-    #undef MLOCK_SUGGESTION
-
-    void raw_unlock(void * addr, size_t size) {
-        if (munlock(addr, size)) {
-            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
-        }
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        SYSTEM_INFO si;
-        GetSystemInfo(&si);
-        return (size_t) si.dwPageSize;
-    }
-
-    bool raw_lock(void * ptr, size_t len) {
-        for (int tries = 1; ; tries++) {
-            if (VirtualLock(ptr, len)) {
-                return true;
-            }
-            if (tries == 2) {
-                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
-                    len, size, llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-
-            // It failed but this was only the first try; increase the working
-            // set size and try again.
-            SIZE_T min_ws_size, max_ws_size;
-            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
-                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-            // Per MSDN: "The maximum number of pages that a process can lock
-            // is equal to the number of pages in its minimum working set minus
-            // a small overhead."
-            // Hopefully a megabyte is enough overhead:
-            size_t increment = len + 1048576;
-            // The minimum must be <= the maximum, so we need to increase both:
-            min_ws_size += increment;
-            max_ws_size += increment;
-            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
-                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
-                        gguf_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-        }
-    }
-
-    void raw_unlock(void * ptr, size_t len) {
-        if (!VirtualUnlock(ptr, len)) {
-            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
-                    gguf_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    size_t lock_granularity() {
-        return (size_t) 65536;
-    }
-
-    bool raw_lock(const void * addr, size_t len) {
-        fprintf(stderr, "warning: mlock not supported on this system\n");
-        return false;
-    }
-
-    void raw_unlock(const void * addr, size_t len) {}
-#endif
-};
-
-#endif

From 88b57694876d1757d77db4766b77eb7889001ff3 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 16 Aug 2023 19:25:29 +0300
Subject: [PATCH 162/242] gguf : deduplicate (#2629)

* gguf : better type names

* dedup : CPU + Metal is working

* ggml : fix warnings about unused results

* llama.cpp : fix line feed and compiler warning

* llama : fix strncpy warning + note token_to_str does not write null

* llama : restore the original load/save session implementation

Will migrate this to GGUF in the future

* convert-llama-h5-to-gguf.py : support alt ctx param name

* ggml : assert when using ggml_mul with non-F32 src1

* examples : dedup simple

---------

Co-authored-by: klosax <131523366+klosax@users.noreply.github.com>
---
 CMakeLists.txt                                |    1 -
 Makefile                                      |   12 +-
 convert-llama-7b-pth-to-gguf.py               |    7 +-
 convert-llama-h5-to-gguf.py                   |   13 +-
 examples/common.cpp                           |   16 -
 examples/common.h                             |    2 -
 .../convert-llama2c-to-ggml.cpp               |  144 +-
 examples/gguf/gguf-llama-simple.cpp           |  129 -
 examples/gguf/gguf.cpp                        |    2 +-
 examples/main/main.cpp                        |    6 +-
 examples/quantize/quantize.cpp                |    8 +-
 examples/save-load-state/save-load-state.cpp  |    1 -
 examples/server/server.cpp                    |   19 -
 examples/simple/simple.cpp                    |   18 +-
 .../train-text-from-scratch.cpp               |   77 +-
 ggml.c                                        |   33 +-
 gguf-llama.cpp                                | 4968 -----------------
 gguf-llama.h                                  |  505 --
 llama-util.h                                  |  553 --
 llama.cpp                                     | 2739 +++++----
 llama.h                                       |   67 +-
 21 files changed, 1776 insertions(+), 7544 deletions(-)
 delete mode 100644 examples/gguf/gguf-llama-simple.cpp
 delete mode 100644 gguf-llama.cpp
 delete mode 100644 gguf-llama.h
 delete mode 100644 llama-util.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index dff4942cdb9b2..c6ed458b300aa 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -529,7 +529,6 @@ endif()
 add_library(llama
             llama.cpp
             llama.h
-            llama-util.h
             )
 
 target_include_directories(llama PUBLIC .)
diff --git a/Makefile b/Makefile
index ce36fcf6b730c..4d2a8cfa96d52 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple server embd-input-test gguf gguf-llama-simple gptneox-main
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple server embd-input-test gguf gptneox-main
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-grammar-parser tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0
@@ -329,10 +329,7 @@ ggml-alloc.o: ggml-alloc.c ggml.h ggml-alloc.h
 
 OBJS += ggml-alloc.o
 
-llama.o: llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h llama.h llama-util.h
-	$(CXX) $(CXXFLAGS) -c $< -o $@
-
-gguf-llama.o: gguf-llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h gguf-llama.h
+llama.o: llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h llama.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
 common.o: examples/common.cpp examples/common.h
@@ -388,10 +385,7 @@ $(LIB_PRE)embdinput$(DSO_EXT): examples/embd-input/embd-input.h examples/embd-in
 embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-test.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %$(DSO_EXT),$(filter-out %.h,$(filter-out %.hpp,$^))) -o $@ $(LDFLAGS) -L. -lembdinput
 
-gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o gguf-llama.o $(OBJS)
-	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
-
-gguf-llama-simple: examples/gguf/gguf-llama-simple.cpp                            build-info.h ggml.o gguf-llama.o common.o $(OBJS)
+gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
 gptneox-main: gptneox-main.cpp ggml.o $(OBJS)
diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 53ab5a3ed0c13..27841939dd4dd 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -132,7 +132,7 @@ def count_model_parts(dir_model: str) -> int:
         toktype = 1 # defualt to normal token type
         if tokenizer.is_unknown(i): toktype = 2
         if tokenizer.is_control(i): toktype = 3
- 
+
         # TODO: How to determinate if a token is user defined?
         # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
         # if tokenizer.is_user_defined(i): toktype = 4
@@ -223,7 +223,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -261,7 +261,6 @@ def count_model_parts(dir_model: str) -> int:
     for name in model_part.keys():
         data = model_part[name]
 
-    
         old_dtype = data.dtype
 
         # we don't need these
@@ -284,7 +283,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index d7706f6187a58..35f85bb978959 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -95,12 +95,21 @@ def count_model_parts(dir_model: str) -> int:
 else:
     hf_repo=""
 
+if "max_sequence_length" in hparams:
+    ctx_length = hparams["max_sequence_length"]
+elif "max_position_embeddings" in hparams:
+    ctx_length = hparams["max_position_embeddings"]
+else:
+    print("gguf: can not find ctx length parameter.")
+    sys.exit()
+
+
 gguf_writer.add_architecture(llm_arch)
 gguf_writer.add_name(last_dir)
 gguf_writer.add_file_type("All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_tensor_data_layout(llm_arch, "Meta AI original pth")
-gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
+gguf_writer.add_context_length(llm_arch, ctx_length)
 gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
 gguf_writer.add_block_count(llm_arch, block_count)
 gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
@@ -318,7 +327,7 @@ def count_model_parts(dir_model: str) -> int:
         if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
             data = data.astype(np.float16)
 
-        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
+        print(name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
         gguf_writer.write_tensor_to_file(data)
 
diff --git a/examples/common.cpp b/examples/common.cpp
index 18de8ef43d403..8beb63f364abd 100644
--- a/examples/common.cpp
+++ b/examples/common.cpp
@@ -170,18 +170,6 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
                 break;
             }
             params.n_ctx = std::stoi(argv[i]);
-        } else if (arg == "-gqa" || arg == "--gqa") {
-            if (++i >= argc) {
-                invalid_param = true;
-                break;
-            }
-            params.n_gqa = std::stoi(argv[i]);
-        } else if (arg == "-eps" || arg == "--rms-norm-eps") {
-            if (++i >= argc) {
-                invalid_param = true;
-                break;
-            }
-            params.rms_norm_eps = std::stof(argv[i]);
         } else if (arg == "--rope-freq-base") {
             if (++i >= argc) {
                 invalid_param = true;
@@ -546,8 +534,6 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
     fprintf(stdout, "  -n N, --n-predict N   number of tokens to predict (default: %d, -1 = infinity, -2 = until context filled)\n", params.n_predict);
     fprintf(stdout, "  -c N, --ctx-size N    size of the prompt context (default: %d)\n", params.n_ctx);
     fprintf(stdout, "  -b N, --batch-size N  batch size for prompt processing (default: %d)\n", params.n_batch);
-    fprintf(stdout, "  -gqa N, --gqa N       grouped-query attention factor (TEMP!!! use 8 for LLaMAv2 70B) (default: %d)\n", params.n_gqa);
-    fprintf(stdout, "  -eps N, --rms-norm-eps N rms norm eps (TEMP!!! use 1e-5 for LLaMAv2) (default: %.1e)\n", params.rms_norm_eps);
     fprintf(stdout, "  --top-k N             top-k sampling (default: %d, 0 = disabled)\n", params.top_k);
     fprintf(stdout, "  --top-p N             top-p sampling (default: %.1f, 1.0 = disabled)\n", (double)params.top_p);
     fprintf(stdout, "  --tfs N               tail free sampling, parameter z (default: %.1f, 1.0 = disabled)\n", (double)params.tfs_z);
@@ -638,8 +624,6 @@ struct llama_context_params llama_context_params_from_gpt_params(const gpt_param
 
     lparams.n_ctx           = params.n_ctx;
     lparams.n_batch         = params.n_batch;
-    lparams.n_gqa           = params.n_gqa;
-    lparams.rms_norm_eps    = params.rms_norm_eps;
     lparams.n_gpu_layers    = params.n_gpu_layers;
     lparams.main_gpu        = params.main_gpu;
     lparams.tensor_split    = params.tensor_split;
diff --git a/examples/common.h b/examples/common.h
index 0e520ed4e7775..b8510dfe453e0 100644
--- a/examples/common.h
+++ b/examples/common.h
@@ -23,14 +23,12 @@ struct gpt_params {
     int32_t n_predict                       = -1;   // new tokens to predict
     int32_t n_ctx                           = 512;  // context size
     int32_t n_batch                         = 512;  // batch size for prompt processing (must be >=32 to use BLAS)
-    int32_t n_gqa                           = 1;    // grouped-query attention factor (TODO: move to hparams)
     int32_t n_keep                          = 0;    // number of tokens to keep from initial prompt
     int32_t n_chunks                        = -1;   // max number of chunks to process (-1 = unlimited)
     int32_t n_gpu_layers                    = 0;    // number of layers to store in VRAM
     int32_t main_gpu                        = 0;    // the GPU that is used for scratch and small tensors
     float   tensor_split[LLAMA_MAX_DEVICES] = {0};  // how split tensors should be distributed across GPUs
     int32_t n_probs                         = 0;    // if greater than 0, output the probabilities of top n_probs tokens.
-    float   rms_norm_eps                    = LLAMA_DEFAULT_RMS_EPS; // rms norm epsilon
     float   rope_freq_base                  = 10000.0f; // RoPE base frequency
     float   rope_freq_scale                 = 1.0f;     // RoPE frequency scaling factor
 
diff --git a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
index 1a238c4dd945a..af493e15bfc88 100644
--- a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
+++ b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
@@ -1,5 +1,6 @@
 #include "ggml.h"
 #include "llama.h"
+
 #include <unordered_map>
 #include <vector>
 #include <cassert>
@@ -502,7 +503,7 @@ bool is_ggml_file(const char *filename) {
         return false;
     }
     uint32_t magic = file.read_u32();
-    return magic == LLAMA_FILE_MAGIC;
+    return magic == GGUF_MAGIC;
 }
 
 void load_vocab(const char *filename, Config *config, struct llama_vocab *vocab) {
@@ -590,75 +591,80 @@ void save_as_llama_model(struct llama_vocab * vocab, struct my_llama_model * mod
     if (file.fp == NULL) {
         return;
     }
-    // write_magic
-    file.write_u32(LLAMA_FILE_MAGIC);   // magic
-    file.write_u32(LLAMA_FILE_VERSION); // version
-    // write_hparams
-    file.write_u32(model->hparams.n_vocab);
-    file.write_u32(model->hparams.n_embd);
-    file.write_u32(model->hparams.n_mult);
-    file.write_u32(model->hparams.n_head);
-    file.write_u32(model->hparams.n_layer);
-    file.write_u32(model->hparams.n_rot);
-    file.write_u32(LLAMA_FTYPE_ALL_F32);
-
-    // write_vocab - for now we are just writing the existing BPE voc. assuming karpathy's vocabulary is the same. idk.
-    uint32_t n_vocab = model->hparams.n_vocab;
-    for (uint32_t i = 0; i < n_vocab; i++) {
-        const auto & token_score = vocab->id_to_token.at(i);
-        file.write_u32((uint32_t) token_score.tok.size());
-        file.write_raw(token_score.tok.data(), token_score.tok.size());
-        file.write_raw(&token_score.score, sizeof(token_score.score));
-    }
-
-    // stuff AK weights into GG weights one by one.
-    // w->token_embedding_table -> model->tok_embeddings
-    // float*                   -> struct ggml_tensor
-    stuff_karpathy_weights_into_gg(model->tok_embeddings, w->token_embedding_table);
-    stuff_karpathy_weights_into_gg(model->output, w->token_embedding_table);
-
-    stuff_karpathy_weights_into_gg(model->norm, w->rms_final_weight);
-    //print_row(model->norm, 0);
-
-    // for rms-att-weight
-    int row_length = model->hparams.n_embd;
-    const auto & hparams = model->hparams;
-    //int n_ff = model->hparams.n_embd;
-    int n_ff = get_n_ff(&hparams);
 
-    for (uint32_t i = 0; i < model->hparams.n_layer; ++i){
-        auto & layer = model->layers[i];
-        // 1d
-        stuff_karpathy_weights_into_gg(layer.attention_norm, &w->rms_att_weight[i*row_length]);
-        stuff_karpathy_weights_into_gg(layer.ffn_norm      , &w->rms_ffn_weight[i*row_length]);
-
-        // from 3d matrix layer x dim x dim to 2d matrix dim x dim
-        stuff_karpathy_weights_into_gg(layer.wq            , &w->wq[i*row_length*row_length]);
-        stuff_karpathy_weights_into_gg(layer.wk            , &w->wk[i*row_length*row_length]);
-        stuff_karpathy_weights_into_gg(layer.wv            , &w->wv[i*row_length*row_length]);
-        stuff_karpathy_weights_into_gg(layer.wo            , &w->wo[i*row_length*row_length]);
-
-        stuff_karpathy_weights_into_gg(layer.w1            , &w->w1[i*row_length*n_ff]);
-        stuff_karpathy_weights_into_gg(layer.w2            , &w->w2[i*n_ff*row_length]);
-        stuff_karpathy_weights_into_gg(layer.w3            , &w->w3[i*row_length*n_ff]);
-    }
-    // write tensors
-    write_tensor(&file, model->tok_embeddings);
-    write_tensor(&file, model->norm);
-    write_tensor(&file, model->output); // ?
-    for (uint32_t i = 0; i < model->hparams.n_layer; ++i) {
-        auto & layer = model->layers[i];
-
-        write_tensor(&file, layer.attention_norm);
-        write_tensor(&file, layer.wq);
-        write_tensor(&file, layer.wk);
-        write_tensor(&file, layer.wv);
-        write_tensor(&file, layer.wo);
-        write_tensor(&file, layer.ffn_norm);
-        write_tensor(&file, layer.w1);
-        write_tensor(&file, layer.w2);
-        write_tensor(&file, layer.w3);
-    }
+#pragma message("TODO: implement file saving using gguf")
+    (void) vocab;
+    (void) model;
+    (void) w;
+//    // write_magic
+//    file.write_u32(LLAMA_FILE_MAGIC);   // magic
+//    file.write_u32(LLAMA_FILE_VERSION); // version
+//    // write_hparams
+//    file.write_u32(model->hparams.n_vocab);
+//    file.write_u32(model->hparams.n_embd);
+//    file.write_u32(model->hparams.n_mult);
+//    file.write_u32(model->hparams.n_head);
+//    file.write_u32(model->hparams.n_layer);
+//    file.write_u32(model->hparams.n_rot);
+//    file.write_u32(LLAMA_FTYPE_ALL_F32);
+//
+//    // write_vocab - for now we are just writing the existing BPE voc. assuming karpathy's vocabulary is the same. idk.
+//    uint32_t n_vocab = model->hparams.n_vocab;
+//    for (uint32_t i = 0; i < n_vocab; i++) {
+//        const auto & token_score = vocab->id_to_token.at(i);
+//        file.write_u32((uint32_t) token_score.tok.size());
+//        file.write_raw(token_score.tok.data(), token_score.tok.size());
+//        file.write_raw(&token_score.score, sizeof(token_score.score));
+//    }
+//
+//    // stuff AK weights into GG weights one by one.
+//    // w->token_embedding_table -> model->tok_embeddings
+//    // float*                   -> struct ggml_tensor
+//    stuff_karpathy_weights_into_gg(model->tok_embeddings, w->token_embedding_table);
+//    stuff_karpathy_weights_into_gg(model->output, w->token_embedding_table);
+//
+//    stuff_karpathy_weights_into_gg(model->norm, w->rms_final_weight);
+//    //print_row(model->norm, 0);
+//
+//    // for rms-att-weight
+//    int row_length = model->hparams.n_embd;
+//    const auto & hparams = model->hparams;
+//    //int n_ff = model->hparams.n_embd;
+//    int n_ff = get_n_ff(&hparams);
+//
+//    for (uint32_t i = 0; i < model->hparams.n_layer; ++i){
+//        auto & layer = model->layers[i];
+//        // 1d
+//        stuff_karpathy_weights_into_gg(layer.attention_norm, &w->rms_att_weight[i*row_length]);
+//        stuff_karpathy_weights_into_gg(layer.ffn_norm      , &w->rms_ffn_weight[i*row_length]);
+//
+//        // from 3d matrix layer x dim x dim to 2d matrix dim x dim
+//        stuff_karpathy_weights_into_gg(layer.wq            , &w->wq[i*row_length*row_length]);
+//        stuff_karpathy_weights_into_gg(layer.wk            , &w->wk[i*row_length*row_length]);
+//        stuff_karpathy_weights_into_gg(layer.wv            , &w->wv[i*row_length*row_length]);
+//        stuff_karpathy_weights_into_gg(layer.wo            , &w->wo[i*row_length*row_length]);
+//
+//        stuff_karpathy_weights_into_gg(layer.w1            , &w->w1[i*row_length*n_ff]);
+//        stuff_karpathy_weights_into_gg(layer.w2            , &w->w2[i*n_ff*row_length]);
+//        stuff_karpathy_weights_into_gg(layer.w3            , &w->w3[i*row_length*n_ff]);
+//    }
+//    // write tensors
+//    write_tensor(&file, model->tok_embeddings);
+//    write_tensor(&file, model->norm);
+//    write_tensor(&file, model->output); // ?
+//    for (uint32_t i = 0; i < model->hparams.n_layer; ++i) {
+//        auto & layer = model->layers[i];
+//
+//        write_tensor(&file, layer.attention_norm);
+//        write_tensor(&file, layer.wq);
+//        write_tensor(&file, layer.wk);
+//        write_tensor(&file, layer.wv);
+//        write_tensor(&file, layer.wo);
+//        write_tensor(&file, layer.ffn_norm);
+//        write_tensor(&file, layer.w1);
+//        write_tensor(&file, layer.w2);
+//        write_tensor(&file, layer.w3);
+//    }
 }
 
 struct train_params get_default_train_params() {
diff --git a/examples/gguf/gguf-llama-simple.cpp b/examples/gguf/gguf-llama-simple.cpp
deleted file mode 100644
index e59d1cfc19d74..0000000000000
--- a/examples/gguf/gguf-llama-simple.cpp
+++ /dev/null
@@ -1,129 +0,0 @@
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#endif
-
-#include "common.h"
-#include "gguf-llama.h"
-#include "build-info.h"
-
-#include <cmath>
-#include <cstdio>
-#include <string>
-#include <vector>
-
-int main(int argc, char ** argv) {
-    gpt_params params;
-
-    if (argc == 1 || argv[1][0] == '-') {
-        printf("usage: %s MODEL_PATH [PROMPT]\n" , argv[0]);
-        return 1 ;
-    }
-
-    if (argc >= 2) {
-        params.model = argv[1];
-    }
-
-    if (argc >= 3) {
-        params.prompt = argv[2];
-    }
-
-    if (params.prompt.empty()) {
-        params.prompt = "Hello my name is";
-    }
-
-    // init LLM
-
-    llama_backend_init(params.numa);
-
-    llama_context_params ctx_params = llama_context_default_params();
-
-    llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
-
-    if (model == NULL) {
-        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
-        return 1;
-    }
-
-    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
-
-    // tokenize the prompt
-
-    std::vector<llama_token> tokens_list;
-    tokens_list = ::llama_tokenize(ctx, params.prompt, true);
-
-    const int max_context_size     = llama_n_ctx(ctx);
-    const int max_tokens_list_size = max_context_size - 4;
-
-    if ((int) tokens_list.size() > max_tokens_list_size) {
-        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
-        return 1;
-    }
-
-    fprintf(stderr, "\n\n");
-
-    for (auto id : tokens_list) {
-        fprintf(stderr, "%s", llama_token_to_str(ctx, id).c_str());
-    }
-
-    fflush(stderr);
-
-    // main loop
-
-    // The LLM keeps a contextual cache memory of previous token evaluation.
-    // Usually, once this cache is full, it is required to recompute a compressed context based on previous
-    // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
-    // example, we will just stop the loop once this cache is full or once an end of stream is detected.
-
-    const int n_gen = std::min(32, max_context_size);
-
-    while (llama_get_kv_cache_token_count(ctx) < n_gen) {
-        // evaluate the transformer
-
-        if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
-            fprintf(stderr, "%s : failed to eval\n", __func__);
-            return 1;
-        }
-
-        tokens_list.clear();
-
-        // sample the next token
-
-        llama_token new_token_id = 0;
-
-        auto logits  = llama_get_logits(ctx);
-        auto n_vocab = llama_n_vocab(ctx);
-
-        std::vector<llama_token_data> candidates;
-        candidates.reserve(n_vocab);
-
-        for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
-            candidates.emplace_back(llama_token_data{ token_id, logits[token_id], 0.0f });
-        }
-
-        llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
-
-        new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
-
-        // is it an end of stream ?
-        if (new_token_id == llama_token_eos()) {
-            fprintf(stderr, " [end of text]\n");
-            break;
-        }
-
-        // print the new token :
-        printf("%s", llama_token_to_str(ctx, new_token_id).c_str());
-        fflush(stdout);
-
-        // push this new token for next evaluation
-        tokens_list.push_back(new_token_id);
-    }
-
-    llama_free(ctx);
-    llama_free_model(model);
-
-    llama_backend_free();
-
-    fprintf(stderr, "\n\n");
-
-    return 0;
-}
diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index ad212d75280a9..d742dce178523 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -1,5 +1,5 @@
 #include "ggml.h"
-#include "gguf-llama.h"
+#include "llama.h"
 
 #include <cstdio>
 #include <cinttypes>
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index 9c15febb51f4b..5c2f64883cc2c 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -266,9 +266,6 @@ int main(int argc, char ** argv) {
         params.interactive = true;
     }
 
-    // determine newline token
-    auto llama_token_newline = ::llama_tokenize(ctx, "\n", false);
-
     if (params.verbose_prompt) {
         fprintf(stderr, "\n");
         fprintf(stderr, "%s: prompt: '%s'\n", __func__, params.prompt.c_str());
@@ -778,8 +775,7 @@ int main(int argc, char ** argv) {
                     if (grammar != NULL) {
                         llama_grammar_free(grammar);
 
-                        std::vector<const llama_grammar_element *> grammar_rules(
-                            parsed_grammar.c_rules());
+                        std::vector<const llama_grammar_element *> grammar_rules( parsed_grammar.c_rules());
                         grammar = llama_grammar_init(
                             grammar_rules.data(), grammar_rules.size(),
                             parsed_grammar.symbol_ids.at("root"));
diff --git a/examples/quantize/quantize.cpp b/examples/quantize/quantize.cpp
index 744f549c5bdf7..f628d06424ace 100644
--- a/examples/quantize/quantize.cpp
+++ b/examples/quantize/quantize.cpp
@@ -68,10 +68,10 @@ bool try_parse_ftype(const std::string & ftype_str_in, llama_ftype & ftype, std:
 }
 
 // usage:
-//  ./quantize [--allow-requantize] [--leave-output-tensor] models/llama/ggml-model.bin [models/llama/ggml-model-quant.bin] type [nthreads]
+//  ./quantize [--allow-requantize] [--leave-output-tensor] models/llama/ggml-model.gguf [models/llama/ggml-model-quant.gguf] type [nthreads]
 //
 void usage(const char * executable) {
-    fprintf(stderr, "usage: %s [--help] [--allow-requantize] [--leave-output-tensor] model-f32.bin [model-quant.bin] type [nthreads]\n\n", executable);
+    fprintf(stderr, "usage: %s [--help] [--allow-requantize] [--leave-output-tensor] model-f32.gguf [model-quant.gguf] type [nthreads]\n\n", executable);
     fprintf(stderr, "  --allow-requantize: Allows requantizing tensors that have already been quantized. Warning: This can severely reduce quality compared to quantizing from 16bit or 32bit\n");
     fprintf(stderr, "  --leave-output-tensor: Will leave output.weight un(re)quantized. Increases model size but may also increase quality, especially when requantizing\n");
     fprintf(stderr, "\nAllowed quantization types:\n");
@@ -118,8 +118,8 @@ int main(int argc, char ** argv) {
         if (pos != std::string::npos) {
             fpath = fname_inp.substr(0, pos + 1);
         }
-        // export as [inp path]/ggml-model-[ftype].bin
-        fname_out = fpath + "ggml-model-" + ftype_str + ".bin";
+        // export as [inp path]/ggml-model-[ftype].gguf
+        fname_out = fpath + "ggml-model-" + ftype_str + ".gguf";
         arg_idx++;
     }
     else {
diff --git a/examples/save-load-state/save-load-state.cpp b/examples/save-load-state/save-load-state.cpp
index d5a81978e97af..3db61b7541171 100644
--- a/examples/save-load-state/save-load-state.cpp
+++ b/examples/save-load-state/save-load-state.cpp
@@ -26,7 +26,6 @@ int main(int argc, char ** argv) {
     auto lparams = llama_context_default_params();
 
     lparams.n_ctx     = params.n_ctx;
-    lparams.n_gqa     = params.n_gqa;
     lparams.seed      = params.seed;
     lparams.f16_kv    = params.memory_f16;
     lparams.use_mmap  = params.use_mmap;
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 222dbcb430f5e..14437206143fb 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -651,8 +651,6 @@ static void server_print_usage(const char *argv0, const gpt_params &params,
     fprintf(stdout, "  -v, --verbose         verbose output (default: %s)\n", server_verbose ? "enabled" : "disabled");
     fprintf(stdout, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
     fprintf(stdout, "  -c N, --ctx-size N    size of the prompt context (default: %d)\n", params.n_ctx);
-    fprintf(stdout, "  -gqa N, --gqa N       grouped-query attention factor (TEMP!!! use 8 for LLaMAv2 70B) (default: %d)\n", params.n_gqa);
-    fprintf(stdout, "  -eps N, --rms-norm-eps N rms norm eps (TEMP!!! use 1e-5 for LLaMAv2) (default: %.1e)\n", params.rms_norm_eps);
     fprintf(stdout, "  --rope-freq-base N    RoPE base frequency (default: %.1f)\n", params.rope_freq_base);
     fprintf(stdout, "  --rope-freq-scale N   RoPE frequency scaling factor (default: %g)\n", params.rope_freq_scale);
     fprintf(stdout, "  -b N, --batch-size N  batch size for prompt processing (default: %d)\n", params.n_batch);
@@ -773,23 +771,6 @@ static void server_params_parse(int argc, char **argv, server_params &sparams,
             }
             params.n_ctx = std::stoi(argv[i]);
         }
-        else if (arg == "-gqa" || arg == "--gqa")
-        {
-            if (++i >= argc)
-            {
-                invalid_param = true;
-                break;
-            }
-            params.n_gqa = std::stoi(argv[i]);
-        }
-        else if (arg == "-eps" || arg == "--rms-norm-eps") {
-            if (++i >= argc)
-            {
-                invalid_param = true;
-                break;
-            }
-            params.rms_norm_eps = std::stof(argv[i]);
-        }
         else if (arg == "--rope-freq-base")
         {
             if (++i >= argc)
diff --git a/examples/simple/simple.cpp b/examples/simple/simple.cpp
index 0dfcd7a448826..476b31b2e9754 100644
--- a/examples/simple/simple.cpp
+++ b/examples/simple/simple.cpp
@@ -36,16 +36,17 @@ int main(int argc, char ** argv) {
 
     llama_backend_init(params.numa);
 
-    llama_model * model;
-    llama_context * ctx;
+    llama_context_params ctx_params = llama_context_default_params();
 
-    std::tie(model, ctx) = llama_init_from_gpt_params(params);
+    llama_model * model = llama_load_model_from_file(params.model.c_str(), ctx_params);
 
     if (model == NULL) {
-        fprintf(stderr, "%s: error: unable to load model\n", __func__);
+        fprintf(stderr , "%s: error: unable to load model\n" , __func__);
         return 1;
     }
 
+    llama_context * ctx = llama_new_context_with_model(model, ctx_params);
+
     // tokenize the prompt
 
     std::vector<llama_token> tokens_list;
@@ -54,7 +55,7 @@ int main(int argc, char ** argv) {
     const int max_context_size     = llama_n_ctx(ctx);
     const int max_tokens_list_size = max_context_size - 4;
 
-    if ((int)tokens_list.size() > max_tokens_list_size) {
+    if ((int) tokens_list.size() > max_tokens_list_size) {
         fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) tokens_list.size(), max_tokens_list_size);
         return 1;
     }
@@ -74,7 +75,9 @@ int main(int argc, char ** argv) {
     // tokens (see "infinite text generation via context swapping" in the main example), but in this minimalist
     // example, we will just stop the loop once this cache is full or once an end of stream is detected.
 
-    while (llama_get_kv_cache_token_count( ctx ) < max_context_size) {
+    const int n_gen = std::min(32, max_context_size);
+
+    while (llama_get_kv_cache_token_count(ctx) < n_gen) {
         // evaluate the transformer
 
         if (llama_eval(ctx, tokens_list.data(), int(tokens_list.size()), llama_get_kv_cache_token_count(ctx), params.n_threads)) {
@@ -114,7 +117,6 @@ int main(int argc, char ** argv) {
 
         // push this new token for next evaluation
         tokens_list.push_back(new_token_id);
-
     }
 
     llama_free(ctx);
@@ -122,5 +124,7 @@ int main(int argc, char ** argv) {
 
     llama_backend_free();
 
+    fprintf(stderr, "\n\n");
+
     return 0;
 }
diff --git a/examples/train-text-from-scratch/train-text-from-scratch.cpp b/examples/train-text-from-scratch/train-text-from-scratch.cpp
index d446c6ea5cfb2..79599951cb174 100644
--- a/examples/train-text-from-scratch/train-text-from-scratch.cpp
+++ b/examples/train-text-from-scratch/train-text-from-scratch.cpp
@@ -17,7 +17,7 @@
 #pragma warning(disable: 4244 4267) // possible loss of data
 #endif
 
-static const float rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
+static const float rms_norm_eps = 1e-5f;
 
 struct random_normal_distribution {
     std::mt19937 gen;
@@ -2612,42 +2612,45 @@ void save_as_llama_model(struct llama_vocab * vocab, struct my_llama_model * mod
         return;
     }
 
-    // write_magic
-    file.write_u32(LLAMA_FILE_MAGIC);   // magic
-    file.write_u32(LLAMA_FILE_VERSION); // version
-    // write_hparams
-    file.write_u32(model->hparams.n_vocab);
-    file.write_u32(model->hparams.n_embd);
-    file.write_u32(model->hparams.n_mult);
-    file.write_u32(model->hparams.n_head);
-    file.write_u32(model->hparams.n_layer);
-    file.write_u32(model->hparams.n_rot);
-    file.write_u32(LLAMA_FTYPE_ALL_F32);
-    // write_vocab
-    uint32_t n_vocab = model->hparams.n_vocab;
-    for (uint32_t i = 0; i < n_vocab; i++) {
-        const auto & token_score = vocab->id_to_token.at(i);
-        file.write_u32((uint32_t) token_score.tok.size());
-        file.write_raw(token_score.tok.data(), token_score.tok.size());
-        file.write_raw(&token_score.score, sizeof(token_score.score));
-    }
-    // write tensors
-    write_tensor(&file, model->tok_embeddings);
-    write_tensor(&file, model->norm);
-    write_tensor(&file, model->output);
-    for (uint32_t i = 0; i < model->hparams.n_layer; ++i) {
-        auto & layer = model->layers[i];
-
-        write_tensor(&file, layer.attention_norm);
-        write_tensor(&file, layer.wq);
-        write_tensor(&file, layer.wk);
-        write_tensor(&file, layer.wv);
-        write_tensor(&file, layer.wo);
-        write_tensor(&file, layer.ffn_norm);
-        write_tensor(&file, layer.w1);
-        write_tensor(&file, layer.w2);
-        write_tensor(&file, layer.w3);
-    }
+#pragma message("TODO: implement file saving using gguf")
+    (void) vocab;
+    (void) model;
+//    // write_magic
+//    file.write_u32(LLAMA_FILE_MAGIC);   // magic
+//    file.write_u32(LLAMA_FILE_VERSION); // version
+//    // write_hparams
+//    file.write_u32(model->hparams.n_vocab);
+//    file.write_u32(model->hparams.n_embd);
+//    file.write_u32(model->hparams.n_mult);
+//    file.write_u32(model->hparams.n_head);
+//    file.write_u32(model->hparams.n_layer);
+//    file.write_u32(model->hparams.n_rot);
+//    file.write_u32(LLAMA_FTYPE_ALL_F32);
+//    // write_vocab
+//    uint32_t n_vocab = model->hparams.n_vocab;
+//    for (uint32_t i = 0; i < n_vocab; i++) {
+//        const auto & token_score = vocab->id_to_token.at(i);
+//        file.write_u32((uint32_t) token_score.tok.size());
+//        file.write_raw(token_score.tok.data(), token_score.tok.size());
+//        file.write_raw(&token_score.score, sizeof(token_score.score));
+//    }
+//    // write tensors
+//    write_tensor(&file, model->tok_embeddings);
+//    write_tensor(&file, model->norm);
+//    write_tensor(&file, model->output);
+//    for (uint32_t i = 0; i < model->hparams.n_layer; ++i) {
+//        auto & layer = model->layers[i];
+//
+//        write_tensor(&file, layer.attention_norm);
+//        write_tensor(&file, layer.wq);
+//        write_tensor(&file, layer.wk);
+//        write_tensor(&file, layer.wv);
+//        write_tensor(&file, layer.wo);
+//        write_tensor(&file, layer.ffn_norm);
+//        write_tensor(&file, layer.w1);
+//        write_tensor(&file, layer.w2);
+//        write_tensor(&file, layer.w3);
+//    }
 }
 
 float cosine_decay(const int decay_steps, const float alpha, int step) {
diff --git a/ggml.c b/ggml.c
index 77f57a3fdaa41..f7e72ed84ddc6 100644
--- a/ggml.c
+++ b/ggml.c
@@ -9140,6 +9140,8 @@ static void ggml_compute_forward_mul(
         const struct ggml_tensor * src0,
         const struct ggml_tensor * src1,
         struct ggml_tensor * dst) {
+    GGML_ASSERT(src1->type == GGML_TYPE_F32 && "only f32 src1 supported for now");
+
     switch (src0->type) {
         case GGML_TYPE_F32:
             {
@@ -18584,17 +18586,18 @@ static const size_t GGUF_TYPE_SIZE[GGUF_TYPE_COUNT] = {
 static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
 
 static const char * GGUF_TYPE_NAME[GGUF_TYPE_COUNT] = {
-    [GGUF_TYPE_UINT8]   = "uint8",
-    [GGUF_TYPE_INT8]    = "int8",
-    [GGUF_TYPE_UINT16]  = "uint16",
-    [GGUF_TYPE_INT16]   = "int16",
-    [GGUF_TYPE_UINT32]  = "uint32",
-    [GGUF_TYPE_INT32]   = "int32",
-    [GGUF_TYPE_FLOAT32] = "float32",
+    [GGUF_TYPE_UINT8]   = "u8",
+    [GGUF_TYPE_INT8]    = "i8",
+    [GGUF_TYPE_UINT16]  = "u16",
+    [GGUF_TYPE_INT16]   = "i16",
+    [GGUF_TYPE_UINT32]  = "u32",
+    [GGUF_TYPE_INT32]   = "i32",
+    [GGUF_TYPE_FLOAT32] = "f32",
     [GGUF_TYPE_BOOL]    = "bool",
-    [GGUF_TYPE_STRING]  = "string",
-    [GGUF_TYPE_ARRAY]   = "array",
+    [GGUF_TYPE_STRING]  = "str",
+    [GGUF_TYPE_ARRAY]   = "arr",
 };
+static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
 
 union gguf_value {
     uint8_t  uint8;
@@ -19395,17 +19398,23 @@ static void gguf_buf_grow(struct gguf_buf * buf, size_t size) {
 static void gguf_bwrite_str(struct gguf_buf * buf, const struct gguf_str * val) {
     gguf_buf_grow(buf, sizeof(val->n) + val->n);
 
-    buf->data && memcpy((char *) buf->data + buf->offset, &val->n, sizeof(val->n));
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, &val->n, sizeof(val->n));
+    }
     buf->offset += sizeof(val->n);
 
-    buf->data && memcpy((char *) buf->data + buf->offset, val->data, val->n);
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, val->data, val->n);
+    }
     buf->offset += val->n;
 }
 
 static void gguf_bwrite_el(struct gguf_buf * buf, const void * val, size_t el_size) {
     gguf_buf_grow(buf, el_size);
 
-    buf->data && memcpy((char *) buf->data + buf->offset, val, el_size);
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, val, el_size);
+    }
     buf->offset += el_size;
 }
 
diff --git a/gguf-llama.cpp b/gguf-llama.cpp
deleted file mode 100644
index d99d752ecf291..0000000000000
--- a/gguf-llama.cpp
+++ /dev/null
@@ -1,4968 +0,0 @@
-// Defines fileno on msys:
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#include <cstddef>
-#include <cstdint>
-#include <cstdio>
-#endif
-
-#define LLAMA_API_CPP // TODO: eliminate me
-#include "gguf-llama.h"
-
-#include "ggml.h"
-
-#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
-#  include "ggml-alloc.h"
-#  define LLAMA_USE_ALLOCATOR
-#else
-#  define LLAMA_USE_SCRATCH
-#  define LLAMA_MAX_SCRATCH_BUFFERS 16
-#endif
-
-#ifdef GGML_USE_CUBLAS
-#  include "ggml-cuda.h"
-#elif defined(GGML_USE_CLBLAST)
-#  include "ggml-opencl.h"
-#endif
-
-#ifdef GGML_USE_METAL
-#  include "ggml-metal.h"
-#endif
-#ifdef GGML_USE_MPI
-#  include "ggml-mpi.h"
-#endif
-#ifdef GGML_USE_K_QUANTS
-#  ifndef QK_K
-#    ifdef GGML_QKK_64
-#      define QK_K 64
-#    else
-#      define QK_K 256
-#    endif
-#  endif
-#endif
-
-#ifdef __has_include
-    #if __has_include(<unistd.h>)
-        #include <unistd.h>
-        #if defined(_POSIX_MAPPED_FILES)
-            #include <sys/mman.h>
-        #endif
-        #if defined(_POSIX_MEMLOCK_RANGE)
-            #include <sys/resource.h>
-        #endif
-    #endif
-#endif
-
-#if defined(_WIN32)
-    #define WIN32_LEAN_AND_MEAN
-    #ifndef NOMINMAX
-        #define NOMINMAX
-    #endif
-    #include <windows.h>
-    #include <io.h>
-    #include <stdio.h> // for _fseeki64
-#endif
-
-#include <algorithm>
-#include <array>
-#include <cassert>
-#include <cinttypes>
-#include <climits>
-#include <cstdarg>
-#include <cstring>
-#include <ctime>
-#include <fstream>
-#include <initializer_list>
-#include <map>
-#include <memory>
-#include <mutex>
-#include <numeric>
-#include <queue>
-#include <random>
-#include <sstream>
-#include <thread>
-#include <unordered_map>
-
-#if defined(_MSC_VER)
-#pragma warning(disable: 4244 4267) // possible loss of data
-#endif
-
-// tensor names
-#define TN_TOKEN_EMBD  "token_embd.weight"
-#define TN_OUTPUT_NORM "output_norm.weight"
-#define TN_OUTPUT      "output.weight"
-#define TN_ATTN_NORM   "blk.%d.attn_norm.weight"
-#define TN_ATTN_Q      "blk.%d.attn_q.weight"
-#define TN_ATTN_K      "blk.%d.attn_k.weight"
-#define TN_ATTN_V      "blk.%d.attn_v.weight"
-#define TN_ATTN_OUTPUT "blk.%d.attn_output.weight"
-#define TN_FFN_NORM    "blk.%d.ffn_norm.weight"
-#define TN_FFN_GATE    "blk.%d.ffn_gate.weight"
-#define TN_FFN_DOWN    "blk.%d.ffn_down.weight"
-#define TN_FFN_UP      "blk.%d.ffn_up.weight"
-
-#ifdef __GNUC__
-#ifdef __MINGW32__
-#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(gnu_printf, __VA_ARGS__)))
-#else
-#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
-#endif
-#else
-#define LLAMA_ATTRIBUTE_FORMAT(...)
-#endif
-
-//
-// logging
-//
-LLAMA_ATTRIBUTE_FORMAT(2, 3)
-static void llama_log_internal        (llama_log_level level, const char* format, ...);
-static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data);
-
-#define LLAMA_LOG_INFO(...)  llama_log_internal(LLAMA_LOG_LEVEL_INFO , __VA_ARGS__)
-#define LLAMA_LOG_WARN(...)  llama_log_internal(LLAMA_LOG_LEVEL_WARN , __VA_ARGS__)
-#define LLAMA_LOG_ERROR(...) llama_log_internal(LLAMA_LOG_LEVEL_ERROR, __VA_ARGS__)
-
-//
-// helpers
-//
-
-template<typename T>
-static std::string to_string(const T & val) {
-    std::stringstream ss;
-    ss << val;
-    return ss.str();
-}
-
-static void zeros(std::ofstream & file, size_t n) {
-    char zero = 0;
-    for (size_t i = 0; i < n; ++i) {
-        file.write(&zero, 1);
-    }
-}
-
-LLAMA_ATTRIBUTE_FORMAT(1, 2)
-static std::string format(const char * fmt, ...) {
-    va_list ap;
-    va_list ap2;
-    va_start(ap, fmt);
-    va_copy(ap2, ap);
-    int size = vsnprintf(NULL, 0, fmt, ap);
-    GGML_ASSERT(size >= 0 && size < INT_MAX); // NOLINT
-    std::vector<char> buf(size + 1);
-    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
-    GGML_ASSERT(size2 == size);
-    va_end(ap2);
-    va_end(ap);
-    return std::string(buf.data(), size);
-}
-
-//
-// ggml helpers
-//
-
-static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph * graph, int n_threads) {
-    struct ggml_cplan plan = ggml_graph_plan(graph, n_threads);
-
-    if (plan.work_size > 0) {
-        buf.resize(plan.work_size);
-        plan.work_data = buf.data();
-    }
-
-    ggml_graph_compute(graph, &plan);
-}
-
-//
-// llama helpers
-//
-
-#ifdef GGML_USE_CUBLAS
-#   define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
-#   define llama_host_free(data) ggml_cuda_host_free(data)
-#elif GGML_USE_METAL
-#   define llama_host_malloc(n)  ggml_metal_host_malloc(n)
-#   define llama_host_free(data) ggml_metal_host_free(data)
-#else
-#   define llama_host_malloc(n)  malloc(n)
-#   define llama_host_free(data) free(data)
-#endif
-
-#if defined(_WIN32)
-static std::string llama_format_win_err(DWORD err) {
-    LPSTR buf;
-    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
-                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
-    if (!size) {
-        return "FormatMessageA failed";
-    }
-    std::string ret(buf, size);
-    LocalFree(buf);
-    return ret;
-}
-#endif
-
-struct llama_buffer {
-    void * data = NULL;
-    size_t size = 0;
-
-    // fallback to malloc / free
-    // useful in cases where CUDA can try to allocate PINNED memory
-    bool fallback = false;
-
-    void resize(size_t n) {
-        llama_host_free(data);
-
-        data = llama_host_malloc(n);
-        if (!data) {
-            fallback = true;
-            data = malloc(n);
-        } else {
-            fallback = false;
-        }
-
-        GGML_ASSERT(data);
-        size = n;
-    }
-
-    ~llama_buffer() {
-        if (data) {
-            if (fallback) { // NOLINT
-                free(data);
-            } else {
-                llama_host_free(data);
-            }
-        }
-
-        data = NULL;
-    }
-};
-
-struct llama_file {
-    // use FILE * so we don't have to re-open the file to mmap
-    FILE * fp;
-    size_t size;
-
-    llama_file(const char * fname, const char * mode) {
-        fp = std::fopen(fname, mode);
-        if (fp == NULL) {
-            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
-        }
-        seek(0, SEEK_END);
-        size = tell();
-        seek(0, SEEK_SET);
-    }
-
-    size_t tell() const {
-#ifdef _WIN32
-        __int64 ret = _ftelli64(fp);
-#else
-        long ret = std::ftell(fp);
-#endif
-        GGML_ASSERT(ret != -1); // this really shouldn't fail
-        return (size_t) ret;
-    }
-
-    void seek(size_t offset, int whence) const {
-#ifdef _WIN32
-        int ret = _fseeki64(fp, (__int64) offset, whence);
-#else
-        int ret = std::fseek(fp, (long) offset, whence);
-#endif
-        GGML_ASSERT(ret == 0); // same
-    }
-
-    void read_raw(void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        std::size_t ret = std::fread(ptr, len, 1, fp);
-        if (ferror(fp)) {
-            throw std::runtime_error(format("read error: %s", strerror(errno)));
-        }
-        if (ret != 1) {
-            throw std::runtime_error(std::string("unexpectedly reached end of file"));
-        }
-    }
-
-    void write_raw(const void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        size_t ret = std::fwrite(ptr, len, 1, fp);
-        if (ret != 1) {
-            throw std::runtime_error(format("write error: %s", strerror(errno)));
-        }
-    }
-
-    ~llama_file() {
-        if (fp) {
-            std::fclose(fp);
-        }
-    }
-};
-
-struct llama_mmap {
-    void * addr;
-    size_t size;
-
-    llama_mmap(const llama_mmap &) = delete;
-
-#ifdef _POSIX_MAPPED_FILES
-    static constexpr bool SUPPORTED = true;
-
-    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
-        size = file->size;
-        int fd = fileno(file->fp);
-        int flags = MAP_SHARED;
-        // prefetch/readahead impairs performance on NUMA systems
-        if (numa) { prefetch = 0; }
-#ifdef __linux__
-        if (prefetch) { flags |= MAP_POPULATE; }
-#endif
-        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
-        if (addr == MAP_FAILED) {
-            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
-        }
-
-        if (prefetch > 0) {
-            // Advise the kernel to preload the mapped memory
-            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
-                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-        if (numa) {
-            // advise the kernel not to use readahead
-            // (because the next page might not belong on the same node)
-            if (madvise(addr, file->size, MADV_RANDOM)) {
-                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-    }
-
-    ~llama_mmap() {
-        munmap(addr, size);
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
-        (void) numa;
-
-        size = file->size;
-
-        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
-
-        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
-        DWORD error = GetLastError();
-
-        if (hMapping == NULL) {
-            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
-        error = GetLastError();
-        CloseHandle(hMapping);
-
-        if (addr == NULL) {
-            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
-        if (prefetch) {
-            // Advise the kernel to preload the mapped memory
-            WIN32_MEMORY_RANGE_ENTRY range;
-            range.VirtualAddress = addr;
-            range.NumberOfBytes = (SIZE_T)size;
-            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
-                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
-                        llama_format_win_err(GetLastError()).c_str());
-            }
-        }
-        #else
-        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
-        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
-    }
-
-    ~llama_mmap() {
-        if (!UnmapViewOfFile(addr)) {
-            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
-        (void) file;
-        (void) prefetch;
-        (void) numa;
-
-        throw std::runtime_error(std::string("mmap not supported"));
-    }
-#endif
-};
-
-// Represents some region of memory being locked using mlock or VirtualLock;
-// will automatically unlock on destruction.
-struct llama_mlock {
-    void * addr = NULL;
-    size_t size = 0;
-
-    bool failed_already = false;
-
-    llama_mlock() {}
-    llama_mlock(const llama_mlock &) = delete;
-
-    ~llama_mlock() {
-        if (size) {
-            raw_unlock(addr, size);
-        }
-    }
-
-    void init(void * ptr) {
-        GGML_ASSERT(addr == NULL && size == 0); // NOLINT
-        addr = ptr;
-    }
-
-    void grow_to(size_t target_size) {
-        GGML_ASSERT(addr);
-        if (failed_already) {
-            return;
-        }
-        size_t granularity = lock_granularity();
-        target_size = (target_size + granularity - 1) & ~(granularity - 1);
-        if (target_size > size) {
-            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
-                size = target_size;
-            } else {
-                failed_already = true;
-            }
-        }
-    }
-
-#ifdef _POSIX_MEMLOCK_RANGE
-    static constexpr bool SUPPORTED = true;
-
-    static size_t lock_granularity() {
-        return (size_t) sysconf(_SC_PAGESIZE);
-    }
-
-    #ifdef __APPLE__
-        #define MLOCK_SUGGESTION \
-            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
-            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
-    #else
-        #define MLOCK_SUGGESTION \
-            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
-    #endif
-
-    bool raw_lock(const void * addr, size_t size) const {
-        if (!mlock(addr, size)) {
-            return true;
-        }
-
-        char* errmsg = std::strerror(errno);
-        bool suggest = (errno == ENOMEM);
-
-        // Check if the resource limit is fine after all
-        struct rlimit lock_limit;
-        if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit)) {
-            suggest = false;
-        }
-        if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size)) {
-            suggest = false;
-        }
-
-        fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
-                size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
-        return false;
-    }
-
-    #undef MLOCK_SUGGESTION
-
-    static void raw_unlock(void * addr, size_t size) {
-        if (munlock(addr, size)) {
-            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
-        }
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    static size_t lock_granularity() {
-        SYSTEM_INFO si;
-        GetSystemInfo(&si);
-        return (size_t) si.dwPageSize;
-    }
-
-    bool raw_lock(void * ptr, size_t len) const {
-        for (int tries = 1; ; tries++) {
-            if (VirtualLock(ptr, len)) {
-                return true;
-            }
-            if (tries == 2) {
-                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
-                    len, size, llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-
-            // It failed but this was only the first try; increase the working
-            // set size and try again.
-            SIZE_T min_ws_size, max_ws_size;
-            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
-                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
-                        llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-            // Per MSDN: "The maximum number of pages that a process can lock
-            // is equal to the number of pages in its minimum working set minus
-            // a small overhead."
-            // Hopefully a megabyte is enough overhead:
-            size_t increment = len + 1048576;
-            // The minimum must be <= the maximum, so we need to increase both:
-            min_ws_size += increment;
-            max_ws_size += increment;
-            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
-                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
-                        llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-        }
-    }
-
-    static void raw_unlock(void * ptr, size_t len) {
-        if (!VirtualUnlock(ptr, len)) {
-            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    static size_t lock_granularity() {
-        return (size_t) 65536;
-    }
-
-    bool raw_lock(const void * addr, size_t len) const {
-        fprintf(stderr, "warning: mlock not supported on this system\n");
-        return false;
-    }
-
-    static void raw_unlock(const void * addr, size_t len) {}
-#endif
-};
-
-typedef void (*offload_func_t)(struct ggml_tensor * tensor);
-
-void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
-    (void) tensor;
-}
-
-//
-// globals
-//
-
-struct llama_state {
-    // We save the log callback globally
-    llama_log_callback log_callback = llama_log_callback_default;
-    void * log_callback_user_data = nullptr;
-};
-
-static llama_state g_state;
-
-//
-// memory sizes (calculated for n_batch == 512)
-//
-
-// computed for n_ctx == 2048
-// TODO: dynamically determine these sizes
-//       needs modifications in ggml
-
-// available llama models
-enum e_model {
-    MODEL_UNKNOWN,
-    MODEL_3B,
-    MODEL_7B,
-    MODEL_13B,
-    MODEL_30B,
-    MODEL_65B,
-    MODEL_70B,
-};
-
-static const size_t kB = 1024;
-static const size_t MB = 1024*1024;
-
-static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   ((size_t) n_ctx / 16ull +  92ull) * MB },
-        { MODEL_7B,   ((size_t) n_ctx / 16ull + 100ull) * MB },
-        { MODEL_13B,  ((size_t) n_ctx / 12ull + 120ull) * MB },
-        { MODEL_30B,  ((size_t) n_ctx /  9ull + 160ull) * MB },
-        { MODEL_65B,  ((size_t) n_ctx /  6ull + 256ull) * MB }, // guess
-        { MODEL_70B,  ((size_t) n_ctx /  7ull + 164ull) * MB },
-    };
-    return k_sizes;
-}
-
-static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,  128ull * MB },
-        { MODEL_7B,  160ull * MB },
-        { MODEL_13B, 192ull * MB },
-        { MODEL_30B, 256ull * MB },
-        { MODEL_65B, 384ull * MB }, // guess
-        { MODEL_70B, 304ull * MB },
-    };
-    return k_sizes;
-}
-
-// used to store the compute graph tensors + non-scratch data
-static const std::map<e_model, size_t> & MEM_REQ_EVAL()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   8ull * MB },
-        { MODEL_7B,  10ull * MB },
-        { MODEL_13B, 12ull * MB },
-        { MODEL_30B, 16ull * MB },
-        { MODEL_65B, 24ull * MB }, // guess
-        { MODEL_70B, 24ull * MB },
-    };
-    return k_sizes;
-}
-
-// amount of VRAM needed per batch size to hold temporary results
-// the values for 3b are not derived from testing but instead chosen conservatively
-static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_BASE()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,   512ull * kB },
-        { MODEL_7B,   512ull * kB },
-        { MODEL_13B,  640ull * kB },
-        { MODEL_30B,  768ull * kB },
-        { MODEL_65B, 1280ull * kB },
-        { MODEL_70B, 1280ull * kB },
-    };
-    return k_sizes;
-}
-
-// amount of VRAM needed per batch size and context to hold temporary results
-// the values for 3b are not derived from testing but instead chosen conservatively
-static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
-{
-    static std::map<e_model, size_t> k_sizes = {
-        { MODEL_3B,  128ull },
-        { MODEL_7B,  128ull },
-        { MODEL_13B, 160ull },
-        { MODEL_30B, 208ull },
-        { MODEL_65B, 256ull },
-        { MODEL_70B, 256ull },
-    };
-    return k_sizes;
-}
-
-// default hparams (LLaMA 7B)
-struct llama_hparams {
-    uint32_t n_vocab   = 32000;
-    uint32_t n_ctx     = 512;
-    uint32_t n_embd    = 4096;
-    uint32_t n_head    = 32;
-    uint32_t n_head_kv = 32;
-    uint32_t n_layer   = 32;
-    uint32_t n_rot     = 64;
-    uint32_t n_ff      = 11008;
-
-    float f_norm_rms_eps = 1e-5;
-
-    float rope_freq_base  = 10000.0f;
-    float rope_freq_scale = 1.0f;
-
-    enum llama_ftype ftype = LLAMA_FTYPE_MOSTLY_F16;
-
-    bool operator!=(const llama_hparams & other) const {
-        return static_cast<bool>(memcmp(this, &other, sizeof(llama_hparams))); // NOLINT
-    }
-
-    uint32_t n_gqa() const {
-        return n_head/n_head_kv;
-    }
-
-    uint32_t n_embd_head() const {
-        return n_embd/n_head;
-    }
-
-    uint32_t n_embd_gqa() const {
-        return n_embd/n_gqa();
-    }
-
-    size_t kv_size() const {
-        size_t result = 2ull;
-        result *= (size_t) n_embd_gqa();
-        result *= (size_t) n_ctx;
-        result *= (size_t) n_layer;
-        result *= sizeof(ggml_fp16_t);
-        return result;
-    }
-};
-
-struct llama_layer {
-    // normalization
-    struct ggml_tensor * attention_norm;
-
-    // attention
-    struct ggml_tensor * wq;
-    struct ggml_tensor * wk;
-    struct ggml_tensor * wv;
-    struct ggml_tensor * wo;
-
-    // normalization
-    struct ggml_tensor * ffn_norm;
-
-    // ff
-    struct ggml_tensor * w1;
-    struct ggml_tensor * w2;
-    struct ggml_tensor * w3;
-};
-
-struct llama_kv_cache {
-    struct ggml_tensor * k = NULL;
-    struct ggml_tensor * v = NULL;
-
-    struct ggml_context * ctx = NULL;
-
-    llama_buffer buf;
-
-    int n; // number of tokens currently in the cache
-
-    ~llama_kv_cache() {
-        if (ctx) {
-            ggml_free(ctx);
-        }
-
-#ifdef GGML_USE_CUBLAS
-        ggml_cuda_free_data(k);
-        ggml_cuda_free_data(v);
-#endif // GGML_USE_CUBLAS
-    }
-};
-
-struct llama_vocab {
-    // TODO:
-    // - add a vector of merges
-    // - add members for bos/eos/pad/sep tokens
-    //   so that we can pass it to different types of tokenizers with a common interface
-
-    using id    = int32_t;
-    using token = std::string;
-
-    struct token_score {
-        token tok;
-        float score;
-    };
-
-    std::unordered_map<token, id> token_to_id;
-    std::vector<token_score> id_to_token;
-};
-
-struct llama_model {
-    e_model type = MODEL_UNKNOWN;
-
-    llama_hparams hparams;
-    llama_vocab   vocab;
-
-    struct ggml_tensor * tok_embeddings;
-
-    struct ggml_tensor * norm;
-    struct ggml_tensor * output;
-
-    std::vector<llama_layer> layers;
-    int n_gpu_layers;
-
-    // context
-    struct ggml_context * ctx = NULL;
-
-    // the model memory buffer
-    llama_buffer buf;
-
-    // model memory mapped file
-    std::unique_ptr<llama_mmap> mapping;
-
-    // objects representing data potentially being locked in memory
-    llama_mlock mlock_buf;
-    llama_mlock mlock_mmap;
-
-    // for quantize-stats only
-    std::vector<std::pair<std::string, struct ggml_tensor *>> tensors_by_name;
-
-    int64_t t_load_us = 0;
-    int64_t t_start_us = 0;
-
-    ~llama_model() {
-        if (ctx) {
-            ggml_free(ctx);
-        }
-
-#ifdef GGML_USE_CUBLAS
-        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
-            ggml_cuda_free_data(tensors_by_name[i].second);
-        }
-        ggml_cuda_free_scratch();
-#elif defined(GGML_USE_CLBLAST)
-        for (size_t i = 0; i < tensors_by_name.size(); ++i) {
-            ggml_cl_free_data(tensors_by_name[i].second);
-        }
-#endif
-    }
-};
-
-struct llama_context {
-    llama_context(const llama_model & model) : model(model), t_load_us(model.t_load_us), t_start_us(model.t_start_us) {}
-    ~llama_context() {
-        if (model_owner) {
-            delete &model;
-        }
-#ifdef GGML_USE_METAL
-        if (ctx_metal) {
-            ggml_metal_free(ctx_metal);
-        }
-#endif
-#ifdef LLAMA_USE_ALLOCATOR
-        if (alloc) {
-            ggml_allocr_free(alloc);
-        }
-#endif
-    }
-
-    std::mt19937 rng;
-
-    bool has_evaluated_once = false;
-
-    int64_t t_sample_us = 0;
-    int64_t t_eval_us   = 0;
-    int64_t t_p_eval_us = 0;
-
-    int32_t n_sample = 0; // number of tokens sampled
-    int32_t n_eval   = 0; // number of eval calls
-    int32_t n_p_eval = 0; // number of tokens in eval calls for the prompt (with batch size > 1)
-
-    const llama_model & model;
-
-    bool model_owner = false;
-
-    int64_t t_load_us;
-    int64_t t_start_us;
-
-    // key + value cache for the self attention
-    struct llama_kv_cache kv_self;
-
-    size_t mem_per_token = 0;
-
-    // decode output (2-dimensional array: [n_tokens][n_vocab])
-    std::vector<float> logits;
-    bool logits_all = false;
-
-    // input embedding (1-dimensional array: [n_embd])
-    std::vector<float> embedding;
-
-    // reusable buffer for `struct ggml_graph_plan.work_data`
-    std::vector<uint8_t> work_buffer;
-
-    // memory buffers used to evaluate the model
-    // TODO: move in llama_state
-    llama_buffer buf_compute;
-
-#ifdef LLAMA_USE_ALLOCATOR
-    llama_buffer buf_alloc;
-    ggml_allocr * alloc = NULL;
-#endif
-
-#ifdef LLAMA_USE_SCRATCH
-    llama_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
-
-    int    buf_last = 0;
-    size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
-#endif
-
-#ifdef GGML_USE_METAL
-    ggml_metal_context * ctx_metal = NULL;
-#endif
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_context * ctx_mpi = NULL;
-#endif
-
-    static void use_buf(struct ggml_context * ctx, int i) {
-#if defined(LLAMA_USE_SCRATCH)
-        size_t last_size = 0;
-
-        if (i == -1) {
-            last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
-        } else {
-            auto & buf = buf_scratch[i];
-            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.data, });
-        }
-
-        if (buf_last >= 0) {
-            buf_max_size[buf_last] = std::max(buf_max_size[buf_last], last_size);
-        }
-
-        buf_last = i;
-#else
-        (void) i;
-        (void) ctx;
-#endif
-    }
-
-    static size_t get_buf_max_mem(int i) {
-#if defined(LLAMA_USE_SCRATCH)
-        return buf_max_size[i];
-#else
-        (void) i;
-        return 0;
-#endif
-    }
-};
-
-//
-// kv cache helpers
-//
-
-static bool llama_kv_cache_init(
-        const struct llama_hparams & hparams,
-             struct llama_kv_cache & cache,
-                         ggml_type   wtype,
-                               int   n_ctx,
-                               int   n_gpu_layers) {
-    const int n_embd  = hparams.n_embd_gqa();
-    const int n_layer = hparams.n_layer;
-
-    const int64_t n_mem      = n_layer*n_ctx;
-    const int64_t n_elements = n_embd*n_mem;
-
-    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
-    cache.n = 0;
-
-    struct ggml_init_params params;
-    params.mem_size   = cache.buf.size;
-    params.mem_buffer = cache.buf.data;
-    params.no_alloc   = false;
-
-    cache.ctx = ggml_init(params);
-
-    if (!cache.ctx) {
-        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
-        return false;
-    }
-
-    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    ggml_set_name(cache.k, "cache_k");
-    ggml_set_name(cache.v, "cache_v");
-
-    (void) n_gpu_layers;
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer + 1) {
-        ggml_cuda_assign_buffers_no_scratch(cache.v);
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        ggml_cuda_assign_buffers_no_scratch(cache.k);
-    }
-#endif // GGML_USE_CUBLAS
-
-    return true;
-}
-
-//
-// model loading and saving
-//
-
-enum llama_file_version {
-    GGUF_FILE_VERSION_V1 = 1,
-};
-
-static const char * llama_file_version_name(llama_file_version version) {
-    switch (version) {
-        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
-    }
-
-    return "unknown";
-}
-
-static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
-    char buf[256];
-    snprintf(buf, sizeof(buf), "%5u", ne.at(0));
-    for (size_t i = 1; i < ne.size(); i++) {
-        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5u", ne.at(i));
-    }
-    return buf;
-}
-
-static std::string llama_format_tensor_shape(const struct ggml_tensor * t) {
-    char buf[256];
-    snprintf(buf, sizeof(buf), "%5" PRId64, t->ne[0]);
-    for (int i = 1; i < GGML_MAX_DIMS; i++) {
-        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5" PRId64, t->ne[i]);
-    }
-    return buf;
-}
-
-struct llama_model_loader {
-    int n_kv      = 0;
-    int n_tensors = 0;
-    int n_created = 0;
-
-    bool use_mmap = false;
-
-    llama_file file;
-    llama_file_version file_version;
-
-    std::unique_ptr<llama_mmap> mapping;
-
-    struct gguf_context * ctx_gguf = NULL;
-    struct ggml_context * ctx_meta = NULL;
-
-    llama_model_loader(const std::string & fname, bool use_mmap) : file(fname.c_str(), "rb") {
-        struct gguf_init_params params = {
-            /*.no_alloc = */ true,
-            /*.ctx      = */ &ctx_meta,
-        };
-
-        ctx_gguf = gguf_init_from_file(fname.c_str(), params);
-
-        n_kv      = gguf_get_n_kv(ctx_gguf);
-        n_tensors = gguf_get_n_tensors(ctx_gguf);
-
-        file_version = (enum llama_file_version) gguf_get_version(ctx_gguf);
-
-        // print meta data
-        // TODO: make optional
-        {
-            LLAMA_LOG_INFO("%s: loaded meta data with %d key-value paris and %d tensors from %s (version %s)\n",
-                    __func__, n_kv, n_tensors, fname.c_str(), llama_file_version_name(file_version));
-
-            for (int i = 0; i < n_tensors; i++) {
-                const char * name = gguf_get_tensor_name(ctx_gguf, i);
-                struct ggml_tensor * meta = ggml_get_tensor(ctx_meta, name);
-
-                LLAMA_LOG_INFO("%s: - tensor %3d: %32s %-8s [ %s ]\n", __func__, i, name, ggml_type_name(meta->type), llama_format_tensor_shape(meta).c_str());
-            }
-
-            for (int i = 0; i < n_kv; i++) {
-                const char * name         = gguf_get_key(ctx_gguf, i);
-                const enum gguf_type type = gguf_get_kv_type(ctx_gguf, i);
-
-                LLAMA_LOG_INFO("%s: - kv %3d: %42s %-8s\n", __func__, i, name, gguf_type_name(type));
-            }
-        }
-
-        if (!llama_mmap::SUPPORTED) {
-            LLAMA_LOG_WARN("%s: mmap is not supported on this platform\n", __func__);
-            use_mmap = false;
-        }
-
-        this->use_mmap = use_mmap;
-    }
-
-    const char * get_tensor_name(int i) const {
-        return gguf_get_tensor_name(ctx_gguf, i);
-    }
-
-    struct ggml_tensor * get_tensor_meta(int i) const {
-        return ggml_get_tensor(ctx_meta, get_tensor_name(i));
-    }
-
-    void calc_sizes(size_t & ctx_size_p, size_t & mmapped_size_p) const {
-        ctx_size_p     = 0;
-        mmapped_size_p = 0;
-
-        for (int i = 0; i < n_tensors; i++) {
-            struct ggml_tensor * meta = get_tensor_meta(i);
-            ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
-            (use_mmap ? mmapped_size_p : ctx_size_p) += ggml_nbytes_pad(meta);
-        }
-    }
-
-    struct ggml_tensor * create_tensor_for(struct ggml_context * ctx, struct ggml_tensor * meta, ggml_backend backend) {
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ctx, true);
-        }
-
-        struct ggml_tensor * tensor = ggml_dup_tensor(ctx, meta);
-        tensor->backend = backend; // TODO: ggml_set_backend
-        ggml_set_name(tensor, ggml_get_name(meta));
-
-        if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ctx, use_mmap);
-        }
-
-        n_created++;
-
-        return tensor;
-    }
-
-    struct ggml_tensor * create_tensor(struct ggml_context * ctx, const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
-        struct ggml_tensor * cur = ggml_get_tensor(ctx_meta, name.c_str());
-
-        {
-            bool is_ok = true;
-            for (size_t i = 0; i < ne.size(); ++i) {
-                if (ne[i] != cur->ne[i]) {
-                    is_ok = false;
-                    break;
-                }
-            }
-            if (!is_ok) {
-                throw std::runtime_error(
-                        format("%s: tensor '%s' has wrong shape; expected %s, got %s",
-                            __func__, name.c_str(),
-                            llama_format_tensor_shape(ne).c_str(),
-                            llama_format_tensor_shape(cur).c_str()));
-            }
-        }
-
-        return create_tensor_for(ctx, cur, backend);
-    }
-
-    void done_getting_tensors() const {
-        if (n_created != n_tensors) {
-            throw std::runtime_error(format("%s: wrong number of tensors; expected %d, got %d", __func__, n_tensors, n_created));
-        }
-    }
-
-    size_t file_offset(const char * name) const {
-        const int idx = gguf_find_tensor(ctx_gguf, name);
-
-        if (idx < 0) {
-            throw std::runtime_error(format("%s: tensor '%s' not found in the file", __func__, name));
-        }
-
-        return gguf_get_data_offset(ctx_gguf) + gguf_get_tensor_offset(ctx_gguf, idx);
-    }
-
-    void load_data_for(struct ggml_tensor * cur) const {
-        const size_t offs = file_offset(ggml_get_name(cur));
-
-        if (use_mmap) {
-            cur->data = (uint8_t *) mapping->addr + offs;
-        } else {
-            file.seek(offs, SEEK_SET);
-            file.read_raw(cur->data, ggml_nbytes(cur));
-        }
-    }
-
-    void load_all_data(struct ggml_context * ctx, llama_progress_callback progress_callback, void * progress_callback_user_data, llama_mlock * lmlock) {
-        size_t size_data = 0;
-        size_t size_lock = 0;
-        size_t size_pref = 0; // prefetch
-
-        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
-            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
-            size_data += ggml_nbytes(cur);
-            if (cur->backend == GGML_BACKEND_CPU) {
-                size_pref += ggml_nbytes(cur);
-            }
-        }
-
-        if (use_mmap) {
-            mapping.reset(new llama_mmap(&file, size_pref, ggml_is_numa()));
-            if (lmlock) {
-                lmlock->init(mapping->addr);
-            }
-        }
-
-        size_t done_size = 0;
-        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
-            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
-            GGML_ASSERT(cur); // unused tensors should have been caught by load_data already
-
-            if (progress_callback) {
-                progress_callback((float) done_size / size_data, progress_callback_user_data);
-            }
-
-            // allocate temp buffer if not using mmap
-            if (!use_mmap && cur->data == NULL) {
-                GGML_ASSERT(cur->backend != GGML_BACKEND_CPU);
-                cur->data = malloc(ggml_nbytes(cur));
-            }
-
-            load_data_for(cur);
-
-            switch (cur->backend) {
-                case GGML_BACKEND_CPU:
-                    if (use_mmap && lmlock) {
-                        size_lock += ggml_nbytes(cur);
-                        lmlock->grow_to(size_lock);
-                    }
-                    break;
-#if defined(GGML_USE_CUBLAS)
-                case GGML_BACKEND_GPU:
-                case GGML_BACKEND_GPU_SPLIT:
-                    // old code:
-                    //ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
-
-                    // TODO: test if this works !!
-                    ggml_cuda_transform_tensor(cur->data, cur);
-                    if (!use_mmap) {
-                        free(cur->data);
-                    }
-                    break;
-#elif defined(GGML_USE_CLBLAST)
-                case GGML_BACKEND_GPU:
-                    ggml_cl_transform_tensor(cur->data, cur);
-                    if (!use_mmap) {
-                        free(cur->data);
-                    }
-                    break;
-#endif
-                default:
-                    continue;
-            }
-
-            done_size += ggml_nbytes(cur);
-        }
-    }
-};
-
-//
-// load LLaMA models
-//
-
-static const char * llama_ftype_name(enum llama_ftype ftype) {
-    switch (ftype) {
-        case LLAMA_FTYPE_ALL_F32:     return "all F32";
-        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
-        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
-        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
-        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
-                                      return "mostly Q4_1, some F16";
-        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
-        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
-        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
-
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K:   return "mostly Q2_K";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q6_K:   return "mostly Q6_K";
-
-        default: return "unknown, may not work";
-    }
-}
-
-static const char * llama_model_type_name(e_model type) {
-    switch (type) {
-        case MODEL_3B:  return "3B";
-        case MODEL_7B:  return "7B";
-        case MODEL_13B: return "13B";
-        case MODEL_30B: return "30B";
-        case MODEL_65B: return "65B";
-        case MODEL_70B: return "70B";
-        default: GGML_ASSERT(false);
-    }
-}
-
-static void llama_model_load_internal(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        const bool mul_mat_q,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void * progress_callback_user_data) {
-    model.t_start_us = ggml_time_us();
-
-    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
-
-    model.n_gpu_layers = n_gpu_layers;
-
-    auto & hparams = model.hparams;
-
-    // read hparams
-    {
-        struct gguf_context * ctx = ml->ctx_gguf;
-
-#define GGUF_GET(dst, func, type, req, key) \
-        { \
-            const int kid = gguf_find_key(ctx, key); \
-            if (kid >= 0) { \
-                enum gguf_type ktype = gguf_get_kv_type(ctx, kid); \
-                if (ktype != (type)) { \
-                    throw std::runtime_error(format("key %s has wrong type: %s", key, gguf_type_name(ktype))); \
-                } \
-                (dst) = func(ctx, kid); \
-            } else if (req) { \
-                throw std::runtime_error(format("key not found in model: %s", key)); \
-            } \
-        }
-
-        GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
-        GGUF_GET(hparams.n_ctx,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
-        GGUF_GET(hparams.n_embd,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.embedding_length");
-        GGUF_GET(hparams.n_ff,           gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.feed_forward_length");
-        GGUF_GET(hparams.n_head,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.attention.head_count");
-        GGUF_GET(hparams.n_layer,        gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.block_count");
-        GGUF_GET(hparams.n_rot,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.rope.dimension_count");
-        GGUF_GET(hparams.f_norm_rms_eps, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, "llama.attention.layer_norm_rms_epsilon");
-
-        // n_head_kv is optional, default to n_head
-        hparams.n_head_kv = hparams.n_head;
-        GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
-#undef GGUF_GET
-
-        switch (hparams.n_layer) {
-            case 26: model.type = e_model::MODEL_3B; break;
-            case 32: model.type = e_model::MODEL_7B; break;
-            case 40: model.type = e_model::MODEL_13B; break;
-            case 60: model.type = e_model::MODEL_30B; break;
-            case 80: model.type = e_model::MODEL_65B; break;
-            default:
-                {
-                    if (hparams.n_layer < 32) {
-                        model.type = e_model::MODEL_7B;
-                    }
-                } break;
-        }
-
-        hparams.n_ctx = n_ctx;
-
-        // LLaMAv2
-        // TODO: probably not needed
-        {
-            const auto n_gqa = hparams.n_gqa();
-
-            if (model.type == e_model::MODEL_65B && n_gqa == 8) {
-                LLAMA_LOG_WARN("%s: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
-                model.type = e_model::MODEL_70B;
-            }
-        }
-
-        hparams.rope_freq_base  = rope_freq_base;
-        hparams.rope_freq_scale = rope_freq_scale;
-    }
-
-    // read vocab
-    {
-        struct gguf_context * ctx = ml->ctx_gguf;
-
-        vocab.id_to_token.resize(hparams.n_vocab);
-
-        const int token_idx = gguf_find_key(ctx, "tokenizer.ggml.tokens");
-        if (token_idx == -1) {
-            throw std::runtime_error("cannot find token list in GGUF file\n");
-        }
-
-        const int score_idx = gguf_find_key(ctx, "tokenizer.ggml.scores");
-        if (score_idx == -1) {
-            throw std::runtime_error("cannot find token scores list in GGUF file\n");
-        }
-
-        const float * scores = (const float * ) gguf_get_arr_data(ctx, score_idx);
-
-        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
-            std::string word = gguf_get_arr_str(ctx, token_idx, i);
-
-            vocab.token_to_id[word] = i;
-
-            auto & tok_score = vocab.id_to_token[i];
-            tok_score.tok = std::move(word);
-            tok_score.score = scores[i];
-        }
-    }
-
-    {
-        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, llama_file_version_name(ml->file_version));
-        LLAMA_LOG_INFO("%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
-        LLAMA_LOG_INFO("%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
-        LLAMA_LOG_INFO("%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        LLAMA_LOG_INFO("%s: n_head     = %u\n",   __func__, hparams.n_head);
-        LLAMA_LOG_INFO("%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
-        LLAMA_LOG_INFO("%s: n_layer    = %u\n",   __func__, hparams.n_layer);
-        LLAMA_LOG_INFO("%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        LLAMA_LOG_INFO("%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: f_norm_eps = %.1e\n", __func__, hparams.f_norm_rms_eps);
-        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, hparams.n_ff);
-        LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
-        LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
-        LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
-        LLAMA_LOG_INFO("%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
-    }
-
-    if (vocab_only) {
-        return;
-    }
-
-    auto & ctx = model.ctx;
-
-    size_t ctx_size;
-    size_t mmapped_size;
-
-    ml->calc_sizes(ctx_size, mmapped_size);
-
-    LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
-
-    // create the ggml context
-    {
-        model.buf.resize(ctx_size);
-        if (use_mlock) {
-            model.mlock_buf.init   (model.buf.data);
-            model.mlock_buf.grow_to(model.buf.size);
-        }
-
-        struct ggml_init_params params = {
-            /*.mem_size   =*/ model.buf.size,
-            /*.mem_buffer =*/ model.buf.data,
-            /*.no_alloc   =*/ ml->use_mmap,
-        };
-
-        model.ctx = ggml_init(params);
-        if (!model.ctx) {
-            throw std::runtime_error(format("ggml_init() failed"));
-        }
-    }
-
-    (void) main_gpu;
-    (void) mul_mat_q;
-#if defined(GGML_USE_CUBLAS)
-    LLAMA_LOG_INFO("%s: using CUDA for GPU acceleration\n", __func__);
-    ggml_cuda_set_main_device(main_gpu);
-    ggml_cuda_set_mul_mat_q(mul_mat_q);
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU_SPLIT
-#elif defined(GGML_USE_CLBLAST)
-    LLAMA_LOG_INFO("%s: using OpenCL for GPU acceleration\n", __func__);
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_GPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_GPU
-#else
-#define LLAMA_BACKEND_OFFLOAD       GGML_BACKEND_CPU
-#define LLAMA_BACKEND_OFFLOAD_SPLIT GGML_BACKEND_CPU
-#endif
-
-    // prepare memory for the weights
-    size_t vram_weights = 0;
-    size_t vram_scratch = 0;
-    {
-        const uint32_t n_embd     = hparams.n_embd;
-        const uint32_t n_embd_gqa = hparams.n_embd_gqa();
-        const uint32_t n_layer    = hparams.n_layer;
-        const uint32_t n_vocab    = hparams.n_vocab;
-
-        model.tok_embeddings = ml->create_tensor(ctx, TN_TOKEN_EMBD, {n_embd, n_vocab}, GGML_BACKEND_CPU);
-
-        // "output" tensor
-        {
-            ggml_backend backend_norm;
-            ggml_backend backend_output;
-            if (n_gpu_layers > int(n_layer)) { // NOLINT
-                // norm is not performance relevant on its own but keeping it in VRAM reduces data copying
-                // on Windows however this is detrimental unless everything is on the GPU
-#ifndef _WIN32
-                backend_norm = low_vram ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
-#else
-                backend_norm = low_vram || n_gpu_layers <= (int) n_layer + 2 ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD;
-#endif // _WIN32
-
-                backend_output = LLAMA_BACKEND_OFFLOAD_SPLIT;
-            } else {
-                backend_norm = GGML_BACKEND_CPU;
-                backend_output = GGML_BACKEND_CPU;
-            }
-
-            model.norm   = ml->create_tensor(ctx, TN_OUTPUT_NORM, {n_embd},          backend_norm);
-            model.output = ml->create_tensor(ctx, TN_OUTPUT,      {n_embd, n_vocab}, backend_output);
-            if (backend_norm == GGML_BACKEND_GPU) {
-                vram_weights += ggml_nbytes(model.norm);
-            }
-            if (backend_output == GGML_BACKEND_GPU_SPLIT) {
-                vram_weights += ggml_nbytes(model.output);
-            }
-        }
-
-        const uint32_t n_ff = hparams.n_ff;
-
-        const int i_gpu_start = n_layer - n_gpu_layers;
-
-        model.layers.resize(n_layer);
-        for (uint32_t i = 0; i < n_layer; ++i) {
-            const ggml_backend backend = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD; // NOLINT
-            const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
-
-            auto & layer = model.layers[i];
-            layer.attention_norm = ml->create_tensor(ctx, format(TN_ATTN_NORM, i), {n_embd}, backend);
-
-            layer.wq = ml->create_tensor(ctx, format(TN_ATTN_Q, i),      {n_embd, n_embd},     backend_split);
-            layer.wk = ml->create_tensor(ctx, format(TN_ATTN_K, i),      {n_embd, n_embd_gqa}, backend_split);
-            layer.wv = ml->create_tensor(ctx, format(TN_ATTN_V, i),      {n_embd, n_embd_gqa}, backend_split);
-            layer.wo = ml->create_tensor(ctx, format(TN_ATTN_OUTPUT, i), {n_embd, n_embd},     backend_split);
-
-            layer.ffn_norm = ml->create_tensor(ctx, format(TN_FFN_NORM, i), {n_embd}, backend);
-
-            layer.w1 = ml->create_tensor(ctx, format(TN_FFN_GATE, i), {n_embd,   n_ff}, backend_split);
-            layer.w2 = ml->create_tensor(ctx, format(TN_FFN_DOWN, i), {  n_ff, n_embd}, backend_split);
-            layer.w3 = ml->create_tensor(ctx, format(TN_FFN_UP, i),   {n_embd,   n_ff}, backend_split);
-
-            if (backend == GGML_BACKEND_GPU) {
-                vram_weights +=
-                    ggml_nbytes(layer.attention_norm) + ggml_nbytes(layer.wq) + ggml_nbytes(layer.wk)             +
-                    ggml_nbytes(layer.wv)             + ggml_nbytes(layer.wo) + ggml_nbytes(layer.ffn_norm) +
-                    ggml_nbytes(layer.w1)             + ggml_nbytes(layer.w2) + ggml_nbytes(layer.w3);
-            }
-        }
-    }
-
-    ml->done_getting_tensors();
-
-    // print memory requirements
-    {
-        const size_t scale = memory_type == GGML_TYPE_F32 ? 2 : 1;
-
-        // this is the total memory required to run the inference
-        size_t mem_required =
-            ctx_size +
-            mmapped_size - vram_weights; // weights in VRAM not in memory
-
-#ifndef LLAMA_USE_ALLOCATOR
-        mem_required +=
-            MEM_REQ_SCRATCH0(hparams.n_ctx).at(model.type) +
-            MEM_REQ_SCRATCH1().at(model.type) +
-            MEM_REQ_EVAL().at(model.type);
-#endif
-
-        // this is the memory required by one llama_state
-        const size_t mem_required_state =
-            scale*hparams.kv_size();
-
-        LLAMA_LOG_INFO("%s: mem required  = %7.2f MB (+ %7.2f MB per state)\n", __func__,
-                mem_required / 1024.0 / 1024.0, mem_required_state / 1024.0 / 1024.0);
-
-        (void) vram_scratch;
-        (void) n_batch;
-#ifdef GGML_USE_CUBLAS
-        if (low_vram) {
-            LLAMA_LOG_INFO("%s: not allocating a VRAM scratch buffer due to low VRAM option\n", __func__);
-            ggml_cuda_set_scratch_size(0); // disable scratch
-        } else {
-            const size_t vram_scratch_base = VRAM_REQ_SCRATCH_BASE().at(model.type);
-            const size_t vram_scratch_per_context = VRAM_REQ_SCRATCH_PER_CONTEXT().at(model.type);
-            vram_scratch = n_batch * (vram_scratch_base + n_ctx * vram_scratch_per_context);
-            ggml_cuda_set_scratch_size(vram_scratch);
-            if (n_gpu_layers > 0) {
-                LLAMA_LOG_INFO("%s: allocating batch_size x (%zd kB + n_ctx x %zd B) = %zd MB VRAM for the scratch buffer\n",
-                        __func__, vram_scratch_base / kB, vram_scratch_per_context,
-                        (vram_scratch + MB - 1) / MB); // round up
-            }
-        }
-#endif // GGML_USE_CUBLAS
-
-#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
-        const int n_gpu = std::min(n_gpu_layers, int(hparams.n_layer));
-
-        LLAMA_LOG_INFO("%s: offloading %d repeating layers to GPU\n", __func__, n_gpu);
-        if (n_gpu_layers > (int) hparams.n_layer) {
-            LLAMA_LOG_INFO("%s: offloading non-repeating layers to GPU\n", __func__);
-        }
-        size_t vram_kv_cache = 0;
-
-#ifdef GGML_USE_CUBLAS
-        const int max_backend_supported_layers = hparams.n_layer + 3;
-        const int max_offloadable_layers = low_vram ? hparams.n_layer + 1 : hparams.n_layer + 3;
-        if (n_gpu_layers > (int) hparams.n_layer + 1) {
-            if (low_vram) {
-                LLAMA_LOG_INFO("%s: cannot offload v cache to GPU due to low VRAM option\n", __func__);
-            } else {
-                LLAMA_LOG_INFO("%s: offloading v cache to GPU\n", __func__);
-                vram_kv_cache += hparams.kv_size() / 2;
-            }
-        }
-        if (n_gpu_layers > (int) hparams.n_layer + 2) {
-            if (low_vram) {
-                LLAMA_LOG_WARN("%s: cannot offload k cache to GPU due to low VRAM option\n", __func__);
-            } else {
-                LLAMA_LOG_INFO("%s: offloading k cache to GPU\n", __func__);
-                vram_kv_cache += hparams.kv_size() / 2;
-            }
-        }
-#elif defined(GGML_USE_CLBLAST)
-        const int max_backend_supported_layers = hparams.n_layer + 1;
-        const int max_offloadable_layers = hparams.n_layer + 1;
-#endif // GGML_USE_CUBLAS
-
-        LLAMA_LOG_INFO("%s: offloaded %d/%d layers to GPU\n",
-                __func__, std::min(n_gpu_layers, max_offloadable_layers), max_backend_supported_layers);
-        LLAMA_LOG_INFO("%s: total VRAM used: %zu MB\n",
-                __func__, (vram_weights + vram_scratch + vram_kv_cache + MB - 1) / MB); // round up
-#else
-        (void) n_gpu_layers;
-#endif // defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST)
-    }
-
-    // populate `tensors_by_name`
-    for (int i = 0; i < ml->n_tensors; ++i) {
-        struct ggml_tensor * cur = ggml_get_tensor(ctx, ml->get_tensor_name(i));
-        model.tensors_by_name.emplace_back(ggml_get_name(cur), cur);
-    }
-
-    (void) tensor_split;
-#if defined(GGML_USE_CUBLAS)
-    {
-        ggml_cuda_set_tensor_split(tensor_split);
-    }
-#endif
-
-    ml->load_all_data(ctx, progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
-
-    if (progress_callback) {
-        progress_callback(1.0f, progress_callback_user_data);
-    }
-
-    model.mapping = std::move(ml->mapping);
-
-    // loading time will be recalculate after the first eval, so
-    // we take page faults deferred by mmap() into consideration
-    model.t_load_us = ggml_time_us() - model.t_start_us;
-}
-
-static bool llama_model_load(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        const bool mul_mat_q,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void *progress_callback_user_data) {
-    try {
-        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gpu_layers,
-                                  main_gpu, tensor_split, mul_mat_q, rope_freq_base, rope_freq_scale, low_vram, memory_type,
-                                  use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
-        return true;
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("error loading model: %s\n", err.what());
-        return false;
-    }
-}
-
-static struct ggml_cgraph * llama_build_graph(
-         llama_context & lctx,
-     const llama_token * tokens,
-           const float * embd,
-                   int   n_tokens,
-                   int   n_past) {
-
-    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
-
-    const int N = n_tokens;
-
-    const auto & model   = lctx.model;
-    const auto & hparams = model.hparams;
-
-    const auto & kv_self = lctx.kv_self;
-
-    GGML_ASSERT(!!kv_self.ctx);
-
-    const int64_t n_embd      = hparams.n_embd;
-    const int64_t n_layer     = hparams.n_layer;
-    const int64_t n_ctx       = hparams.n_ctx;
-    const int64_t n_head      = hparams.n_head;
-    const int64_t n_head_kv   = hparams.n_head_kv;
-    const int64_t n_embd_head = hparams.n_embd_head();
-    const int64_t n_embd_gqa  = hparams.n_embd_gqa();
-
-    GGML_ASSERT(n_embd_head == hparams.n_rot);
-
-    const float freq_base    = hparams.rope_freq_base;
-    const float freq_scale   = hparams.rope_freq_scale;
-    const float norm_rms_eps = hparams.f_norm_rms_eps;
-
-    const int n_gpu_layers = model.n_gpu_layers;
-
-    auto & mem_per_token = lctx.mem_per_token;
-    auto & buf_compute   = lctx.buf_compute;
-
-    struct ggml_init_params params = {
-        /*.mem_size   =*/ buf_compute.size,
-        /*.mem_buffer =*/ buf_compute.data,
-        /*.no_alloc   =*/ false,
-    };
-
-#ifdef LLAMA_USE_ALLOCATOR
-    params.no_alloc = true;
-#endif
-
-    struct ggml_context * ctx0 = ggml_init(params);
-
-    ggml_cgraph * gf = ggml_new_graph(ctx0);
-
-    struct ggml_tensor * cur;
-    struct ggml_tensor * inpL;
-
-    if (tokens) {
-        struct ggml_tensor * inp_tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
-
-#ifdef LLAMA_USE_ALLOCATOR
-        ggml_allocr_alloc(lctx.alloc, inp_tokens);
-        if (!ggml_allocr_is_measure(lctx.alloc)) {
-            memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
-        }
-#else
-        memcpy(inp_tokens->data, tokens, N*ggml_element_size(inp_tokens));
-#endif
-        ggml_set_name(inp_tokens, "inp_tokens");
-
-        inpL = ggml_get_rows(ctx0, model.tok_embeddings, inp_tokens);
-    } else {
-#ifdef GGML_USE_MPI
-        GGML_ASSERT(false && "not implemented");
-#endif
-
-        inpL = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N);
-
-#ifdef LLAMA_USE_ALLOCATOR
-        ggml_allocr_alloc(lctx.alloc, inpL);
-        if (!ggml_allocr_is_measure(lctx.alloc)) {
-            memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
-        }
-#else
-        memcpy(inpL->data, embd, N * n_embd * ggml_element_size(inpL));
-#endif
-    }
-
-    const int i_gpu_start = n_layer - n_gpu_layers;
-    (void) i_gpu_start;
-
-    // offload functions set the tensor output backend to GPU
-    // tensors are GPU-accelerated if any input or the output has been offloaded
-    //
-    // with the low VRAM option VRAM scratch is disabled in llama_load_model_internal
-    // in that case ggml_cuda_assign_buffers has no effect
-    offload_func_t offload_func_nr = llama_nop; // nr = non-repeating
-    offload_func_t offload_func_kq = llama_nop;
-    offload_func_t offload_func_v  = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer) {
-        offload_func_nr = ggml_cuda_assign_buffers;
-    }
-    if (n_gpu_layers > n_layer + 1) {
-        offload_func_v  = ggml_cuda_assign_buffers;
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        offload_func_kq = ggml_cuda_assign_buffers;
-    }
-#endif // GGML_USE_CUBLAS
-
-    struct ggml_tensor * KQ_scale = ggml_new_tensor_1d(ctx0, GGML_TYPE_F32, 1);
-#ifdef LLAMA_USE_ALLOCATOR
-    ggml_allocr_alloc(lctx.alloc, KQ_scale);
-    if (!ggml_allocr_is_measure(lctx.alloc)) {
-        ggml_set_f32(KQ_scale, 1.0f/sqrtf(float(n_embd)/n_head));
-    }
-#else
-    ggml_set_f32(KQ_scale, 1.0f/sqrtf(float(n_embd)/n_head));
-#endif
-    ggml_set_name(KQ_scale, "1/sqrt(n_embd_head)");
-
-    for (int il = 0; il < n_layer; ++il) {
-        ggml_format_name(inpL, "layer_inp_%d", il);
-
-        offload_func_t offload_func = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-        if (il >= i_gpu_start) {
-            offload_func = ggml_cuda_assign_buffers;
-        }
-#endif // GGML_USE_CUBLAS
-
-        struct ggml_tensor * inpSA = inpL;
-
-        llama_context::use_buf(ctx0, 0);
-
-        // norm
-        {
-            cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
-            offload_func(cur);
-            ggml_set_name(cur, "rms_norm_0");
-
-            // cur = cur*attention_norm(broadcasted)
-            cur = ggml_mul(ctx0, cur, model.layers[il].attention_norm);
-            offload_func(cur);
-            ggml_set_name(cur, "attention_norm_0");
-        }
-
-        // self-attention
-        {
-            // compute Q and K and RoPE them
-            struct ggml_tensor * tmpk = ggml_mul_mat(ctx0, model.layers[il].wk, cur);
-            offload_func_kq(tmpk);
-            ggml_set_name(tmpk, "tmpk");
-
-            struct ggml_tensor * tmpq = ggml_mul_mat(ctx0, model.layers[il].wq, cur);
-            offload_func_kq(tmpq);
-            ggml_set_name(tmpq, "tmpq");
-
-            struct ggml_tensor * Kcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpk, n_embd_head, n_head_kv, N), n_past, n_embd_head, 0, 0, freq_base, freq_scale);
-            offload_func_kq(Kcur);
-            ggml_set_name(Kcur, "Kcur");
-
-            struct ggml_tensor * Qcur = ggml_rope_custom_inplace(ctx0, ggml_reshape_3d(ctx0, tmpq, n_embd_head, n_head, N),    n_past, n_embd_head, 0, 0, freq_base, freq_scale);
-            offload_func_kq(Qcur);
-            ggml_set_name(Qcur, "Qcur");
-
-            // store key and value to memory
-            {
-                // compute the transposed [N, n_embd] V matrix
-
-                struct ggml_tensor * tmpv = ggml_mul_mat(ctx0, model.layers[il].wv, cur);
-                offload_func_v(tmpv);
-                ggml_set_name(tmpv, "tmpv");
-
-                struct ggml_tensor * Vcur = ggml_transpose(ctx0, ggml_reshape_2d(ctx0, tmpv, n_embd_gqa, N));
-                offload_func_v(Vcur);
-                ggml_set_name(Vcur, "Vcur");
-
-                struct ggml_tensor * k = ggml_view_1d(ctx0, kv_self.k, N*n_embd_gqa, (ggml_element_size(kv_self.k)*n_embd_gqa)*(il*n_ctx + n_past));
-                offload_func_kq(k);
-                ggml_set_name(k, "k");
-
-                struct ggml_tensor * v = ggml_view_2d(ctx0, kv_self.v, N, n_embd_gqa,
-                        (   n_ctx)*ggml_element_size(kv_self.v),
-                        (il*n_ctx)*ggml_element_size(kv_self.v)*n_embd_gqa + n_past*ggml_element_size(kv_self.v));
-                offload_func_v(v);
-                ggml_set_name(v, "v");
-
-                // important: storing RoPE-ed version of K in the KV cache!
-                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Kcur, k));
-                ggml_build_forward_expand(gf, ggml_cpy(ctx0, Vcur, v));
-            }
-
-            struct ggml_tensor * Q =
-                ggml_permute(ctx0,
-                        Qcur,
-                        0, 2, 1, 3);
-            offload_func_kq(Q);
-            ggml_set_name(Q, "Q");
-
-            struct ggml_tensor * K =
-                ggml_permute(ctx0,
-                        ggml_reshape_3d(ctx0,
-                            ggml_view_1d(ctx0, kv_self.k, (n_past + N)*n_embd_gqa, il*n_ctx*ggml_element_size(kv_self.k)*n_embd_gqa),
-                            n_embd_head, n_head_kv, n_past + N),
-                        0, 2, 1, 3);
-            offload_func_kq(K);
-            ggml_set_name(K, "K");
-
-            // K * Q
-            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
-            offload_func_kq(KQ);
-            ggml_set_name(KQ, "KQ");
-
-            // KQ_scaled = KQ / sqrt(n_embd_head)
-            // KQ_scaled shape [n_past + N, N, n_head, 1]
-            struct ggml_tensor * KQ_scaled = ggml_scale_inplace(ctx0, KQ, KQ_scale);
-            offload_func_kq(KQ_scaled);
-            ggml_set_name(KQ_scaled, "KQ_scaled");
-
-            // KQ_masked = mask_past(KQ_scaled)
-            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
-            offload_func_kq(KQ_masked);
-            ggml_set_name(KQ_masked, "KQ_masked");
-
-            // KQ = soft_max(KQ_masked)
-            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
-            offload_func_v(KQ_soft_max);
-            ggml_set_name(KQ_soft_max, "KQ_soft_max");
-
-            // split cached V into n_head heads
-            struct ggml_tensor * V =
-                ggml_view_3d(ctx0, kv_self.v,
-                        n_past + N, n_embd_head, n_head_kv,
-                        n_ctx*ggml_element_size(kv_self.v),
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_head,
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_gqa*il);
-            offload_func_v(V);
-            ggml_set_name(V, "V");
-
-#if 1
-            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
-            offload_func_v(KQV);
-            ggml_set_name(KQV, "KQV");
-#else
-            // make V contiguous in memory to speed up the matmul, however we waste time on the copy
-            // on M1 this is faster for the perplexity computation, but ~5% slower for the single-token generation
-            // is there a better way?
-            struct ggml_tensor * V_cont = ggml_cpy(ctx0, V, ggml_new_tensor_3d(ctx0, kv_self.v->type, n_past + N, n_embd_head, n_head));
-            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V_cont, KQ_soft_max);
-#endif
-
-            // KQV_merged = KQV.permute(0, 2, 1, 3)
-            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
-            offload_func_v(KQV_merged);
-            ggml_set_name(KQV_merged, "KQV_merged");
-
-            // cur = KQV_merged.contiguous().view(n_embd, N)
-            cur = ggml_cpy(ctx0,
-                    KQV_merged,
-                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
-            offload_func_v(cur);
-            ggml_set_name(cur, "KQV_merged_contiguous");
-
-            // projection (no bias)
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].wo,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_wo");
-        }
-
-        llama_context::use_buf(ctx0, 1);
-
-        struct ggml_tensor * inpFF = ggml_add(ctx0, cur, inpSA);
-        offload_func(inpFF);
-        ggml_set_name(inpFF, "inpFF");
-
-        // feed-forward network
-        {
-            // norm
-            {
-                cur = ggml_rms_norm(ctx0, inpFF, norm_rms_eps);
-                offload_func(cur);
-                ggml_set_name(cur, "rms_norm_1");
-
-                // cur = cur*ffn_norm(broadcasted)
-                cur = ggml_mul(ctx0, cur, model.layers[il].ffn_norm);
-                offload_func(cur);
-                ggml_set_name(cur, "ffn_norm");
-            }
-
-            struct ggml_tensor * tmp = ggml_mul_mat(ctx0,
-                    model.layers[il].w3,
-                    cur);
-            offload_func(tmp);
-            ggml_set_name(tmp, "result_w3");
-
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].w1,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_w1");
-
-            // SILU activation
-            cur = ggml_silu(ctx0, cur);
-            offload_func(cur);
-            ggml_set_name(cur, "silu");
-
-            cur = ggml_mul(ctx0, cur, tmp);
-            offload_func(cur);
-            ggml_set_name(cur, "silu_x_result_w3");
-
-            cur = ggml_mul_mat(ctx0,
-                    model.layers[il].w2,
-                    cur);
-            offload_func(cur);
-            ggml_set_name(cur, "result_w2");
-        }
-
-        cur = ggml_add(ctx0, cur, inpFF);
-        offload_func(cur);
-        ggml_set_name(cur, "inpFF_+_result_w2");
-
-        // input for next layer
-        inpL = cur;
-    }
-
-    llama_context::use_buf(ctx0, 0);
-
-    // norm
-    {
-        cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
-        offload_func_nr(cur);
-        ggml_set_name(cur, "rms_norm_2");
-
-        // cur = cur*norm(broadcasted)
-        cur = ggml_mul(ctx0, cur, model.norm);
-        // offload_func_nr(cur); // TODO CPU + GPU mirrored backend
-        ggml_set_name(cur, "result_norm");
-    }
-
-    // lm_head
-    cur = ggml_mul_mat(ctx0, model.output, cur);
-    ggml_set_name(cur, "result_output");
-
-    llama_context::use_buf(ctx0, -1);
-
-    // logits -> probs
-    //cur = ggml_soft_max_inplace(ctx0, cur);
-
-    ggml_build_forward_expand(gf, cur);
-
-    if (mem_per_token == 0) {
-        mem_per_token = ggml_used_mem(ctx0)/N;
-    }
-
-#if 0
-    LLAMA_LOG_INFO("\n%s: used_mem: eval ctx %.3f MB, scratch %.3f MB %.3f MB, work buf %.3f MB, n_past = %d, N = %d\n", __func__,
-            ggml_used_mem(ctx0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(0)/1024.0/1024.0,
-            lctx.get_buf_max_mem(1)/1024.0/1024.0,
-            lctx.work_buffer.size()/1024.0/1024.0,
-            n_past, N);
-#endif
-
-    ggml_free(ctx0);
-
-    return gf;
-}
-
-// evaluate the transformer
-//
-//   - lctx:      llama context
-//   - tokens:    new batch of tokens to process
-//   - embd       embeddings input
-//   - n_tokens   number of tokens
-//   - n_past:    the context size so far
-//   - n_threads: number of threads to use
-//
-static bool llama_eval_internal(
-         llama_context & lctx,
-     const llama_token * tokens,
-           const float * embd,
-                   int   n_tokens,
-                   int   n_past,
-                   int   n_threads,
-            const char * cgraph_fname) {
-
-    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
-
-    const int64_t t_start_us = ggml_time_us();
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_eval_init(lctx.ctx_mpi, &n_tokens, &n_past, &n_threads);
-#endif
-
-    const int N = n_tokens;
-
-    const auto & model   = lctx.model;
-    const auto & hparams = model.hparams;
-
-    const auto & kv_self = lctx.kv_self;
-
-    GGML_ASSERT(!!kv_self.ctx);
-
-    const int64_t n_embd      = hparams.n_embd;
-    const int64_t n_vocab     = hparams.n_vocab;
-
-#ifdef LLAMA_USE_ALLOCATOR
-    ggml_allocr_reset(lctx.alloc);
-#endif
-
-    ggml_cgraph * gf = llama_build_graph(lctx, tokens, embd, n_tokens, n_past);
-
-#ifdef LLAMA_USE_ALLOCATOR
-    ggml_allocr_alloc_graph(lctx.alloc, gf);
-#endif
-
-    // LLAMA_LOG_INFO("graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf->n_nodes, gf->n_leafs);
-
-    // for big prompts, if BLAS is enabled, it is better to use only one thread
-    // otherwise, the threads are spin-lock waiting for the BLAS calls and are degrading the performance
-    n_threads = N >= 32 && ggml_cpu_has_blas() && !ggml_cpu_has_gpublas() ? 1 : n_threads;
-
-    struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
-    struct ggml_tensor * embeddings = gf->nodes[gf->n_nodes - 2];
-
-    GGML_ASSERT(strcmp(res->name, "result_output") == 0);
-    GGML_ASSERT(strcmp(embeddings->name, "result_norm") == 0);
-
-#if GGML_USE_MPI
-    const int64_t n_layer = hparams.n_layer;
-    ggml_mpi_graph_compute_pre(lctx.ctx_mpi, gf, n_layer);
-#endif
-
-#ifdef GGML_USE_METAL
-    if (lctx.ctx_metal && N == 1) {
-        // TODO: disabled until #2413 is resolved
-        //if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
-        //    ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
-        //}
-        ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
-        ggml_metal_graph_compute(lctx.ctx_metal, gf);
-        ggml_metal_get_tensor   (lctx.ctx_metal, res);
-        if (!lctx.embedding.empty()) {
-            ggml_metal_get_tensor(lctx.ctx_metal, embeddings);
-        }
-    } else {
-        // IMPORTANT:
-        // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
-        // ggml_graph_compute(). It uses Apple's Accelerate CBLAS API which takes advantage of the ANE or the AMX
-        // coprocessor.
-        //
-        // When we implement Matrix x Matrix Metal multiplication, we can avoid this branch.
-        // But for now, we have focused only on Matrix x Vector Metal multiplication.
-        //
-        // TODO: avoid these syncs via shared memory (ref #1696)
-        //
-        if (lctx.ctx_metal) {
-            // We need to sync the GPU KV cache with the CPU KV cache
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.k);
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.v);
-        }
-
-        ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
-    }
-#else
-    ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
-#endif
-
-#if GGML_USE_MPI
-    ggml_mpi_graph_compute_post(lctx.ctx_mpi, gf, n_layer);
-#endif
-
-    // update kv token count
-    lctx.kv_self.n = n_past + N;
-
-    if (cgraph_fname) {
-        ggml_graph_export(gf, cgraph_fname);
-    }
-
-#ifdef GGML_PERF
-    // print timing information per ggml operation (for debugging purposes)
-    // requires GGML_PERF to be defined
-    ggml_graph_print(gf);
-#endif
-
-    // plot the computation graph in dot format (for debugging purposes)
-    //if (n_past%100 == 0) {
-    //    ggml_graph_dump_dot(gf, NULL, "llama.dot");
-    //}
-
-    // extract logits
-    {
-        auto & logits_out = lctx.logits;
-
-        if (lctx.logits_all) {
-            logits_out.resize(n_vocab * N);
-            memcpy(logits_out.data(), (float *) ggml_get_data(res), sizeof(float)*n_vocab*N);
-        } else {
-            // return result for just the last token
-            logits_out.resize(n_vocab);
-            memcpy(logits_out.data(), (float *) ggml_get_data(res) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
-        }
-    }
-
-    // extract embeddings
-    if (!lctx.embedding.empty()) {
-        auto & embedding_out = lctx.embedding;
-
-        embedding_out.resize(n_embd);
-        memcpy(embedding_out.data(), (float *) ggml_get_data(embeddings) + (n_embd*(N - 1)), sizeof(float)*n_embd);
-    }
-
-    // measure the performance only for the single-token evals
-    if (N == 1) {
-        lctx.t_eval_us += ggml_time_us() - t_start_us;
-        lctx.n_eval++;
-    }
-    else if (N > 1) {
-        lctx.t_p_eval_us += ggml_time_us() - t_start_us;
-        lctx.n_p_eval += N;
-    }
-
-    return true;
-}
-
-//
-// tokenizer
-//
-
-static std::string llama_vocab_type(const llama_vocab & vocab) {
-    return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
-}
-
-static bool llama_is_normal_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token >= 259;
-    }
-
-    if (llama_vocab_type(vocab) == "bpe") {
-        return token >= 95;
-    }
-
-    return false;
-}
-
-static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 0;
-    }
-
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_control_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 1 || token == 2;
-    }
-
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 1;
-    }
-
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 2;
-    }
-
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
-    GGML_UNUSED(vocab);
-    GGML_UNUSED(token);
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
-    GGML_UNUSED(vocab);
-    GGML_UNUSED(token);
-    // TODO: improve?
-    return false;
-}
-
-static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return 3 <= token && token < 259;
-    }
-
-    if (llama_vocab_type(vocab) == "bpe") {
-        return 1 <= token && token < 95;
-    }
-
-    return false;
-}
-
-static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return byte + 3;
-    }
-
-    if (llama_vocab_type(vocab) == "bpe") {
-        return byte + 32;
-    }
-
-    return false;
-}
-
-static std::string llama_escape_whitespace(const std::string& text) {
-    std::string result;
-    bool escaping = false;
-    result += "\xe2\x96\x81";
-    for (size_t offs = 0; offs < text.length(); ++offs) {
-        if (text[offs] == ' ') {
-            if (!escaping) {
-                result += "\xe2\x96\x81";
-                escaping = true;
-            }
-        }
-        else {
-            escaping = false;
-            result += text[offs];
-        }
-    }
-    return result;
-}
-
-static std::string llama_unescape_whitespace(const std::string& word) {
-    if (word.length() >= 3 && word.substr(0, 3) == "\xe2\x96\x81") {
-        return std::string(" ") + word.substr(3);
-    }
-    return word;
-}
-
-static size_t utf8_len(char src) {
-    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
-    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
-    return lookup[highbits];
-}
-
-struct llama_sp_symbol {
-    using index = int;
-    index prev;
-    index next;
-    const char * text;
-    size_t n;
-};
-
-static_assert(std::is_trivially_copyable<llama_sp_symbol>::value, "llama_sp_symbol is not trivially copyable");
-
-struct llama_sp_bigram {
-    struct comparator {
-        bool operator()(llama_sp_bigram & l, llama_sp_bigram & r) {
-            return (l.score < r.score) || (l.score == r.score && l.left > r.left);
-        }
-    };
-    using queue_storage = std::vector<llama_sp_bigram>;
-    using queue = std::priority_queue<llama_sp_bigram, queue_storage, comparator>;
-    llama_sp_symbol::index left;
-    llama_sp_symbol::index right;
-    float score;
-    size_t size;
-};
-
-// original implementation:
-// https://github.com/ggerganov/llama.cpp/commit/074bea2eb1f1349a0118239c4152914aecaa1be4
-struct llama_tokenizer {
-    llama_tokenizer(const llama_vocab & vocab): vocab_(vocab) {}
-
-    void tokenize(const std::string & text, std::vector<llama_vocab::id> & output) {
-        // split string into utf8 chars
-        int index = 0;
-        size_t offs = 0;
-        while (offs < text.size()) {
-            llama_sp_symbol sym;
-            size_t len = utf8_len(text[offs]);
-            GGML_ASSERT(offs + len <= text.size());
-            sym.text = text.c_str() + offs;
-            sym.n = len;
-            offs += len;
-            sym.prev = index - 1;
-            sym.next = offs == text.size() ? -1 : index + 1;
-            index++;
-            symbols_.emplace_back(sym);
-        }
-
-        // seed the work queue with all possible 2-character tokens.
-        for (size_t i = 1; i < symbols_.size(); ++i) {
-            try_add_bigram(i - 1, i);
-        }
-
-        // keep substituting the highest frequency pairs for as long as we can.
-        while (!work_queue_.empty()) {
-            auto bigram = work_queue_.top();
-            work_queue_.pop();
-
-            auto & left_sym = symbols_[bigram.left];
-            auto & right_sym = symbols_[bigram.right];
-
-            // if one of the symbols already got merged, skip it.
-            if (left_sym.n == 0 || right_sym.n == 0 ||
-                left_sym.n + right_sym.n != bigram.size) {
-                continue;
-            }
-
-            // merge the right sym into the left one
-            left_sym.n += right_sym.n;
-            right_sym.n = 0;
-
-            //LLAMA_LOG_INFO("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
-
-            // remove the right sym from the chain
-            left_sym.next = right_sym.next;
-            if (right_sym.next >= 0) {
-                symbols_[right_sym.next].prev = bigram.left;
-            }
-
-            // find more substitutions
-            try_add_bigram(left_sym.prev, bigram.left);
-            try_add_bigram(bigram.left, left_sym.next);
-        }
-
-        for (int i = 0; i != -1; i = symbols_[i].next) {
-            auto & symbol = symbols_[i];
-            resegment(symbol, output);
-        }
-    }
-
-private:
-    void resegment(llama_sp_symbol &symbol, std::vector<llama_vocab::id> &output) {
-        auto text = std::string(symbol.text, symbol.n);
-        auto token = vocab_.token_to_id.find(text);
-
-        // Do we need to support is_unused?
-        if (token != vocab_.token_to_id.end()) {
-            output.push_back((*token).second);
-            return;
-        }
-
-        const auto p = rev_merge.find(text);
-
-        if (p == rev_merge.end()) {
-            // output any symbols that did not form tokens as bytes.
-            for (int j = 0; j < (int)symbol.n; ++j) {
-                llama_vocab::id token_id = llama_byte_to_char(vocab_, symbol.text[j]);
-                output.push_back(token_id);
-            }
-            return;
-        }
-
-        resegment(symbols_[p->second.first], output);
-        resegment(symbols_[p->second.second], output);
-    }
-
-    void try_add_bigram(int left, int right) {
-        if (left == -1 || right == -1) {
-            return;
-        }
-
-        const std::string text = std::string(symbols_[left].text, symbols_[left].n + symbols_[right].n);
-        auto token = vocab_.token_to_id.find(text);
-
-        if (token == vocab_.token_to_id.end()) {
-            return;
-        }
-
-        if (static_cast<size_t>((*token).second) >= vocab_.id_to_token.size()) {
-            return;
-        }
-
-        const auto &tok_score = vocab_.id_to_token[(*token).second];
-
-        llama_sp_bigram bigram;
-        bigram.left = left;
-        bigram.right = right;
-        bigram.score = tok_score.score;
-        bigram.size = text.size();
-        work_queue_.push(bigram);
-
-        // Do we need to support is_unused?
-        rev_merge[text] = std::make_pair(left, right);
-    }
-
-    const llama_vocab & vocab_;
-    std::vector<llama_sp_symbol> symbols_;
-    llama_sp_bigram::queue work_queue_;
-    std::map<std::string, std::pair<int, int> > rev_merge;
-};
-
-static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & raw_text, bool bos, bool escape) {
-    llama_tokenizer tokenizer(vocab);
-    std::vector<llama_vocab::id> output;
-
-    if (raw_text.empty()) {
-        return output;
-    }
-
-    if (bos) {
-        output.push_back(llama_token_bos());
-    }
-
-    std::string text;
-    if (escape) {
-        text = llama_escape_whitespace(raw_text);
-    } else {
-        text = raw_text;
-    }
-
-    tokenizer.tokenize(text, output);
-    return output;
-}
-
-//
-// grammar - internal
-//
-
-struct llama_grammar {
-    const std::vector<std::vector<llama_grammar_element>>   rules;
-    std::vector<std::vector<const llama_grammar_element *>> stacks;
-};
-
-struct llama_grammar_candidate {
-    size_t           index;
-    const uint32_t * code_points;
-};
-
-// NOTE: assumes valid utf8 (but checks for overrun)
-// adds a terminating 0 for use as pointer
-std::vector<uint32_t> decode_utf8(const char * src) {
-    static const int      lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
-    const char          * pos      = src;
-    std::vector<uint32_t> code_points;
-    while (*pos != 0) {
-        uint8_t  first_byte = static_cast<uint8_t>(*pos);
-        uint8_t  highbits   = first_byte >> 4;
-        int      len        = lookup[highbits];
-        uint8_t  mask       = (1 << (8 - len)) - 1;
-        uint32_t value      = first_byte & mask;
-        const char * end    = pos + len; // may overrun!
-        ++pos;
-        for ( ; pos < end && *pos != 0; ++pos) {
-            value = (value << 6) + (static_cast<uint8_t>(*pos) & 0x3F);
-        }
-        code_points.push_back(value);
-    }
-    code_points.push_back(0);
-    return code_points;
-}
-
-// returns true iff pos points to the end of one of the definitions of a rule
-static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
-    switch (pos->type) {
-        case LLAMA_GRETYPE_END: return true;  // NOLINT
-        case LLAMA_GRETYPE_ALT: return true;  // NOLINT
-        default:                return false;
-    }
-}
-
-// returns true iff chr satisfies the char range at pos (regular or inverse range)
-// asserts that pos is pointing to a char range element
-static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
-        const llama_grammar_element * pos,
-        const uint32_t                chr) {
-
-    bool found            = false;
-    bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
-    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT); // NOLINT
-
-    do {
-        if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
-            // inclusive range, e.g. [a-z]
-            found = found || (pos->value <= chr && chr <= pos[1].value);
-            pos += 2;
-        } else {
-            // exact char match, e.g. [a] or "a"
-            found = found || pos->value == chr;
-            pos += 1;
-        }
-    } while (pos->type == LLAMA_GRETYPE_CHAR_ALT);
-
-    return std::make_pair(found == is_positive_char, pos);
-}
-
-// transforms a grammar pushdown stack into N possible stacks, all ending
-// at a character range (terminal element)
-static void llama_grammar_advance_stack(
-        const std::vector<std::vector<llama_grammar_element>>   & rules,
-        const std::vector<const llama_grammar_element *>        & stack,
-        std::vector<std::vector<const llama_grammar_element *>> & new_stacks) {
-
-    if (stack.empty()) {
-        new_stacks.push_back(stack);
-        return;
-    }
-
-    const llama_grammar_element * pos = stack.back();
-
-    switch (pos->type) {
-        case LLAMA_GRETYPE_RULE_REF: {
-            const size_t                  rule_id = static_cast<size_t>(pos->value);
-            const llama_grammar_element * subpos  = rules[rule_id].data();
-            do {
-                // init new stack without the top (pos)
-                std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
-                if (!llama_grammar_is_end_of_sequence(pos + 1)) {
-                    // if this rule ref is followed by another element, add that to stack
-                    new_stack.push_back(pos + 1);
-                }
-                if (!llama_grammar_is_end_of_sequence(subpos)) {
-                    // if alternate is nonempty, add to stack
-                    new_stack.push_back(subpos);
-                }
-                llama_grammar_advance_stack(rules, new_stack, new_stacks);
-                while (!llama_grammar_is_end_of_sequence(subpos)) {
-                    // scan to end of alternate def
-                    subpos++;
-                }
-                if (subpos->type == LLAMA_GRETYPE_ALT) {
-                    // there's another alternate def of this rule to process
-                    subpos++;
-                } else {
-                    break;
-                }
-            } while (true);
-            break;
-        }
-        case LLAMA_GRETYPE_CHAR:
-        case LLAMA_GRETYPE_CHAR_NOT:
-            new_stacks.push_back(stack);
-            break;
-        default:
-            // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range
-            // (LLAMA_GRETYPE_CHAR_ALT, LLAMA_GRETYPE_CHAR_RNG_UPPER); stack should never be left on
-            // those
-            GGML_ASSERT(false);
-    }
-}
-
-// takes a set of possible pushdown stacks on a grammar, which are required to
-// be positioned at a character range (see `llama_grammar_advance_stack`), and
-// produces the N possible stacks if the given char is accepted at those
-// positions
-static std::vector<std::vector<const llama_grammar_element *>> llama_grammar_accept(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const uint32_t                                                  chr) {
-
-    std::vector<std::vector<const llama_grammar_element *>> new_stacks;
-
-    for (const auto & stack : stacks) {
-        if (stack.empty()) {
-            continue;
-        }
-
-        auto match = llama_grammar_match_char(stack.back(), chr);
-        if (match.first) {
-            const llama_grammar_element * pos = match.second;
-
-            // update top of stack to next element, if any
-            std::vector<const llama_grammar_element *> new_stack(stack.begin(), stack.end() - 1);
-            if (!llama_grammar_is_end_of_sequence(pos)) {
-                new_stack.push_back(pos);
-            }
-            llama_grammar_advance_stack(rules, new_stack, new_stacks);
-        }
-    }
-
-    return new_stacks;
-}
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const std::vector<llama_grammar_candidate>                    & candidates);
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_stack(
-        const std::vector<std::vector<llama_grammar_element>> & rules,
-        const std::vector<const llama_grammar_element *>      & stack,
-        const std::vector<llama_grammar_candidate>            & candidates) {
-
-    std::vector<llama_grammar_candidate> rejects;
-
-    if (stack.empty()) {
-        // accept nothing; EOS is handled elsewhere
-        rejects.insert(rejects.end(), candidates.begin(), candidates.end());
-        return rejects;
-    }
-
-    const llama_grammar_element * stack_pos = stack.back();
-
-    std::vector<llama_grammar_candidate> next_candidates;
-    for (auto tok : candidates) {
-        if (llama_grammar_match_char(stack_pos, tok.code_points[0]).first) {
-            if (tok.code_points[1] != 0) {
-                next_candidates.push_back({ tok.index, tok.code_points + 1 });
-            }
-        } else {
-            rejects.push_back(tok);
-        }
-    }
-
-    const auto * stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
-
-    // update top of stack to next element, if any
-    std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
-    if (!llama_grammar_is_end_of_sequence(stack_pos_after)) {
-        stack_after.push_back(stack_pos_after);
-    }
-    std::vector<std::vector<const llama_grammar_element *>> next_stacks;
-    llama_grammar_advance_stack(rules, stack_after, next_stacks);
-
-    auto next_rejects = llama_grammar_reject_candidates(rules, next_stacks, next_candidates);
-    for (auto tok : next_rejects) {
-        rejects.push_back({ tok.index, tok.code_points - 1 });
-    }
-
-    return rejects;
-}
-
-static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
-        const std::vector<std::vector<llama_grammar_element>>         & rules,
-        const std::vector<std::vector<const llama_grammar_element *>> & stacks,
-        const std::vector<llama_grammar_candidate>                    & candidates) {
-    GGML_ASSERT(!stacks.empty()); // REVIEW
-
-    if (candidates.empty()) {
-        return std::vector<llama_grammar_candidate>();
-    }
-
-    auto rejects = llama_grammar_reject_candidates_for_stack(rules, stacks.front(), candidates);
-
-    for (size_t i = 1, size = stacks.size(); i < size; ++i) {
-        rejects = llama_grammar_reject_candidates_for_stack(rules, stacks[i], rejects);
-    }
-    return rejects;
-}
-
-//
-// grammar - external
-//
-
-struct llama_grammar * llama_grammar_init(
-            const llama_grammar_element ** rules,
-                                 size_t    n_rules,
-                                 size_t    start_rule_index) {
-    const llama_grammar_element * pos;
-
-    // copy rule definitions into vectors
-    std::vector<std::vector<llama_grammar_element>> vec_rules(n_rules);
-    for (size_t i = 0; i < n_rules; i++) {
-        for (pos = rules[i]; pos->type != LLAMA_GRETYPE_END; pos++) {
-            vec_rules[i].push_back(*pos);
-        }
-        vec_rules[i].push_back({LLAMA_GRETYPE_END, 0});
-    }
-
-    // loop over alternates of start rule to build initial stacks
-    std::vector<std::vector<const llama_grammar_element *>> stacks;
-    pos = rules[start_rule_index];
-    do {
-        std::vector<const llama_grammar_element *> stack;
-        if (!llama_grammar_is_end_of_sequence(pos)) {
-            // if alternate is nonempty, add to stack
-            stack.push_back(pos);
-        }
-        llama_grammar_advance_stack(vec_rules, stack, stacks);
-        while (!llama_grammar_is_end_of_sequence(pos)) {
-            // scan to end of alternate def
-            pos++;
-        }
-        if (pos->type == LLAMA_GRETYPE_ALT) {
-            // there's another alternate def of this rule to process
-            pos++;
-        } else {
-            break;
-        }
-    } while (true);
-
-    return new llama_grammar{ std::move(vec_rules), std::move(stacks) };
-}
-
-void llama_grammar_free(struct llama_grammar * grammar) {
-    delete grammar;
-}
-
-//
-// sampling
-//
-
-void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(candidates->size > 0);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Sort the logits in descending order
-    if (!candidates->sorted) {
-        std::sort(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
-            return a.logit > b.logit;
-        });
-        candidates->sorted = true;
-    }
-
-    float max_l = candidates->data[0].logit;
-    float cum_sum = 0.0f;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        float p = expf(candidates->data[i].logit - max_l);
-        candidates->data[i].p = p;
-        cum_sum += p;
-    }
-    for (size_t i = 0; i < candidates->size; ++i) {
-        candidates->data[i].p /= cum_sum;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    k = std::max(k, (int) min_keep);
-    k = std::min(k, (int) candidates->size);
-
-    // Sort scores in descending order
-    if (!candidates->sorted) {
-        auto comp = [](const llama_token_data & a, const llama_token_data & b) {
-            return a.logit > b.logit;
-        };
-        if (k == (int) candidates->size) {
-            std::sort(candidates->data, candidates->data + candidates->size, comp);
-        } else {
-            std::partial_sort(candidates->data, candidates->data + k, candidates->data + candidates->size, comp);
-        }
-        candidates->sorted = true;
-    }
-    candidates->size = k;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
-    if (p >= 1.0f) {
-        return;
-    }
-
-    llama_sample_softmax(ctx, candidates);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Compute the cumulative probabilities
-    float cum_sum = 0.0f;
-    size_t last_idx = candidates->size;
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        cum_sum += candidates->data[i].p;
-
-        // Check if the running sum is at least p or if we have kept at least min_keep tokens
-        // we set the last index to i+1 to indicate that the current iterate should be included in the set
-        if (cum_sum >= p && i + 1 >= min_keep) {
-            last_idx = i + 1;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the top-p tokens
-    candidates->size = last_idx;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep) {
-    if (z >= 1.0f || candidates->size <= 2) {
-        return;
-    }
-
-    llama_sample_softmax(nullptr, candidates);
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Compute the first and second derivatives
-    std::vector<float> first_derivatives(candidates->size - 1);
-    std::vector<float> second_derivatives(candidates->size - 2);
-
-    for (size_t i = 0; i < first_derivatives.size(); ++i) {
-        first_derivatives[i] = candidates->data[i].p - candidates->data[i + 1].p;
-    }
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        second_derivatives[i] = first_derivatives[i] - first_derivatives[i + 1];
-    }
-
-    // Calculate absolute value of second derivatives
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        second_derivatives[i] = abs(second_derivatives[i]);
-    }
-
-    // Normalize the second derivatives
-    {
-        const float second_derivatives_sum = std::accumulate(second_derivatives.begin(), second_derivatives.end(), 0.0f);
-
-        if (second_derivatives_sum > 1e-6f) {
-            for (float & value : second_derivatives) {
-                value /= second_derivatives_sum;
-            }
-        } else {
-            for (float & value : second_derivatives) {
-                value = 1.0f / second_derivatives.size();
-            }
-        }
-    }
-
-    float cum_sum = 0.0f;
-    size_t last_idx = candidates->size;
-    for (size_t i = 0; i < second_derivatives.size(); ++i) {
-        cum_sum += second_derivatives[i];
-
-        // Check if the running sum is greater than z or if we have kept at least min_keep tokens
-        if (cum_sum > z && i >= min_keep) {
-            last_idx = i;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the tokens above the tail location
-    candidates->size = last_idx;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
-    // Reference implementation:
-    // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
-    if (p >= 1.0f) {
-        return;
-    }
-
-    // Compute the softmax of logits and calculate entropy
-    llama_sample_softmax(nullptr, candidates);
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    float entropy = 0.0f;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        entropy += -candidates->data[i].p * logf(candidates->data[i].p);
-    }
-
-    // Compute the absolute difference between negative log probability and entropy for each candidate
-    std::vector<float> shifted_scores;
-    for (size_t i = 0; i < candidates->size; ++i) {
-        float shifted_score = fabsf(-logf(candidates->data[i].p) - entropy);
-        shifted_scores.push_back(shifted_score);
-    }
-
-    // Sort tokens based on the shifted_scores and their corresponding indices
-    std::vector<size_t> indices(candidates->size);
-    std::iota(indices.begin(), indices.end(), 0);
-
-    std::sort(indices.begin(), indices.end(), [&](size_t a, size_t b) {
-        return shifted_scores[a] < shifted_scores[b];
-    });
-
-    // Compute the cumulative probabilities
-    float cum_sum = 0.0f;
-    size_t last_idx = indices.size();
-
-    for (size_t i = 0; i < indices.size(); ++i) {
-        size_t idx = indices[i];
-        cum_sum += candidates->data[idx].p;
-
-        // Check if the running sum is greater than typical or if we have kept at least min_keep tokens
-        if (cum_sum > p && i >= min_keep - 1) {
-            last_idx = i + 1;
-            break;
-        }
-    }
-
-    // Resize the output vector to keep only the locally typical tokens
-    std::vector<llama_token_data> new_candidates;
-    for (size_t i = 0; i < last_idx; ++i) {
-        size_t idx = indices[i];
-        new_candidates.push_back(candidates->data[idx]);
-    }
-
-    // Replace the data in candidates with the new_candidates data
-    std::copy(new_candidates.begin(), new_candidates.end(), candidates->data);
-    candidates->size = new_candidates.size();
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates_p, float temp) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    for (size_t i = 0; i < candidates_p->size; ++i) {
-        candidates_p->data[i].logit /= temp;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty) {
-    if (last_tokens_size == 0 || penalty == 1.0f) {
-        return;
-    }
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        const auto * token_iter = std::find(last_tokens, last_tokens + last_tokens_size, candidates->data[i].id);
-        if (token_iter == last_tokens + last_tokens_size) {
-            continue;
-        }
-
-        // The academic publication that described this technique actually just only divided, but that would cause tokens with negative logits to become more likely, which is obviously wrong.
-        // This is common fix for this problem, which is to multiply by the penalty instead of dividing.
-        if (candidates->data[i].logit <= 0) {
-            candidates->data[i].logit *= penalty;
-        } else {
-            candidates->data[i].logit /= penalty;
-        }
-    }
-
-    candidates->sorted = false;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens_p, size_t last_tokens_size, float alpha_frequency, float alpha_presence) {
-    if (last_tokens_size == 0 || (alpha_frequency == 0.0f && alpha_presence == 0.0f)) {
-        return;
-    }
-
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Create a frequency map to count occurrences of each token in last_tokens
-    std::unordered_map<llama_token, int> token_count;
-    for (size_t i = 0; i < last_tokens_size; ++i) {
-        token_count[last_tokens_p[i]]++;
-    }
-
-    // Apply frequency and presence penalties to the candidates
-    for (size_t i = 0; i < candidates->size; ++i) {
-        auto token_iter = token_count.find(candidates->data[i].id);
-        if (token_iter == token_count.end()) {
-            continue;
-        }
-
-        int count = token_iter->second;
-        candidates->data[i].logit -= float(count) * alpha_frequency + float(count > 0) * alpha_presence;
-    }
-
-    candidates->sorted = false;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
-    assert(ctx);
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    bool allow_eos = false;
-    for (const auto & stack : grammar->stacks) {
-        if (stack.empty()) {
-            allow_eos = true;
-            break;
-        }
-    }
-
-    const llama_token eos = llama_token_eos();
-
-    std::vector<std::vector<uint32_t>>   candidates_decoded;
-    std::vector<llama_grammar_candidate> candidates_grammar;
-
-    for (size_t i = 0; i < candidates->size; ++i) {
-        const llama_token id  = candidates->data[i].id;
-        std::string       str = llama_token_to_str(ctx, id);
-        if (id == eos) {
-            if (!allow_eos) {
-                candidates->data[i].logit = -INFINITY;
-            }
-        } else if (str.empty()) {
-            candidates->data[i].logit = -INFINITY;
-        } else {
-            candidates_decoded.push_back(decode_utf8(str.c_str()));
-            candidates_grammar.push_back({ i, candidates_decoded.back().data() });
-        }
-    }
-
-    const auto rejects = llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
-    for (const auto & reject : rejects) {
-        candidates->data[reject.index].logit = -INFINITY;
-    }
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-}
-
-static void llama_log_softmax(float * array, size_t size) {
-    float max_l = *std::max_element(array, array + size);
-    float sum = 0.f;
-    for (size_t i = 0; i < size; ++i) {
-        float p = expf(array[i] - max_l);
-        sum += p;
-        array[i] = p;
-    }
-
-    for (size_t i = 0; i < size; ++i) {
-        array[i] = logf(array[i] / sum);
-    }
-}
-
-void llama_sample_classifier_free_guidance(
-          struct llama_context * ctx,
-        llama_token_data_array * candidates,
-          struct llama_context * guidance_ctx,
-                         float   scale) {
-    int64_t t_start_sample_us = ggml_time_us();
-
-    assert(ctx);
-    auto n_vocab = llama_n_vocab(ctx);
-    assert(n_vocab == (int)candidates->size);
-    assert(!candidates->sorted);
-
-    std::vector<float> logits_base;
-    logits_base.reserve(candidates->size);
-    for (size_t i = 0; i < candidates->size; ++i) {
-        logits_base.push_back(candidates->data[i].logit);
-    }
-    llama_log_softmax(logits_base.data(), candidates->size);
-
-    float* logits_guidance = llama_get_logits(guidance_ctx);
-    llama_log_softmax(logits_guidance, n_vocab);
-
-    for (int i = 0; i < n_vocab; ++i) {
-        float logit_guidance = logits_guidance[i];
-        float logit_base = logits_base[i];
-        candidates->data[i].logit = scale * (logit_base - logit_guidance) + logit_guidance;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-}
-
-llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu) {
-    assert(ctx);
-    auto N = float(llama_n_vocab(ctx));
-    int64_t t_start_sample_us;
-    t_start_sample_us = ggml_time_us();
-
-    llama_sample_softmax(nullptr, candidates);
-
-    // Estimate s_hat using the most probable m tokens
-    float s_hat = 0.0;
-    float sum_ti_bi = 0.0;
-    float sum_ti_sq = 0.0;
-    for (size_t i = 0; i < size_t(m - 1) && i < candidates->size - 1; ++i) {
-        float t_i = logf(float(i + 2) / float(i + 1));
-        float b_i = logf(candidates->data[i].p / candidates->data[i + 1].p);
-        sum_ti_bi += t_i * b_i;
-        sum_ti_sq += t_i * t_i;
-    }
-    s_hat = sum_ti_bi / sum_ti_sq;
-
-    // Compute k from the estimated s_hat and target surprise value
-    float epsilon_hat = s_hat - 1;
-    float k = powf((epsilon_hat * powf(2, *mu)) / (1 - powf(N, -epsilon_hat)), 1 / s_hat);
-
-    // Sample the next word X using top-k sampling
-    llama_sample_top_k(nullptr, candidates, int(k), 1);
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    llama_token X = llama_sample_token(ctx, candidates);
-    t_start_sample_us = ggml_time_us();
-
-    // Compute error as the difference between observed surprise and target surprise value
-    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return candidate.id == X;
-    }));
-    float observed_surprise = -log2f(candidates->data[X_idx].p);
-    float e = observed_surprise - tau;
-
-    // Update mu using the learning rate and error
-    *mu = *mu - eta * e;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    return X;
-}
-
-llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu) {
-    int64_t t_start_sample_us;
-    t_start_sample_us = ggml_time_us();
-
-    llama_sample_softmax(ctx, candidates);
-
-    // Truncate the words with surprise values greater than mu
-    candidates->size = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return -log2f(candidate.p) > *mu;
-    }));
-
-    if (candidates->size == 0) {
-        candidates->size = 1;
-    }
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-
-    // Normalize the probabilities of the remaining words
-    llama_sample_softmax(ctx, candidates);
-
-    // Sample the next word X from the remaining words
-    llama_token X = llama_sample_token(ctx, candidates);
-    t_start_sample_us = ggml_time_us();
-
-    // Compute error as the difference between observed surprise and target surprise value
-    size_t X_idx = std::distance(candidates->data, std::find_if(candidates->data, candidates->data + candidates->size, [&](const llama_token_data & candidate) {
-        return candidate.id == X;
-    }));
-    float observed_surprise = -log2f(candidates->data[X_idx].p);
-    float e = observed_surprise - tau;
-
-    // Update mu using the learning rate and error
-    *mu = *mu - eta * e;
-
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    }
-    return X;
-}
-
-llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    // Find max element
-    auto * max_iter = std::max_element(candidates->data, candidates->data + candidates->size, [](const llama_token_data & a, const llama_token_data & b) {
-        return a.logit < b.logit;
-    });
-
-    llama_token result = max_iter->id;
-    if (ctx) {
-        ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-        ctx->n_sample++;
-    }
-    return result;
-}
-
-llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(ctx);
-    const int64_t t_start_sample_us = ggml_time_us();
-    llama_sample_softmax(nullptr, candidates);
-
-    std::vector<float> probs;
-    probs.reserve(candidates->size);
-    for (size_t i = 0; i < candidates->size; ++i) {
-        probs.push_back(candidates->data[i].p);
-    }
-
-    std::discrete_distribution<> dist(probs.begin(), probs.end());
-    auto & rng = ctx->rng;
-    int idx = dist(rng);
-
-    llama_token result = candidates->data[idx].id;
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-    ctx->n_sample++;
-    return result;
-}
-
-void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
-    const int64_t t_start_sample_us = ggml_time_us();
-
-    if (token == llama_token_eos()) {
-        for (const auto & stack : grammar->stacks) {
-            if (stack.empty()) {
-                return;
-            }
-        }
-        GGML_ASSERT(false);
-    }
-
-    std::string str = llama_token_to_str(ctx, token);
-    // Note terminating 0 in decoded string
-    auto code_points = decode_utf8(str.c_str());
-    for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
-        grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
-    }
-    GGML_ASSERT(!grammar->stacks.empty());
-
-    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
-}
-
-//
-// quantization
-//
-
-static void llama_convert_tensor_internal(struct ggml_tensor * tensor, std::vector<float> & output, const size_t nelements, const int nthread) {
-    if (output.size() < nelements) {
-        output.resize(nelements);
-    }
-    float * f32_output = (float *) output.data();
-
-    ggml_type_traits_t qtype;
-    if (ggml_is_quantized(tensor->type)) {
-        qtype = ggml_internal_get_type_traits(tensor->type);
-        if (qtype.to_float == NULL) {
-            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor->type)));
-        }
-    } else if (tensor->type != GGML_TYPE_F16) {
-        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor->type)));
-    }
-
-    if (nthread < 2) {
-        if (tensor->type == GGML_TYPE_F16) {
-            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor->data, f32_output, nelements);
-        } else if (ggml_is_quantized(tensor->type)) {
-            qtype.to_float(tensor->data, f32_output, nelements);
-        } else {
-            GGML_ASSERT(false); // unreachable
-        }
-        return;
-    }
-
-    auto block_size = tensor->type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor->type);
-    auto block_size_bytes = ggml_type_size(tensor->type);
-
-    GGML_ASSERT(nelements % block_size == 0);
-    auto nblocks = nelements / block_size;
-    auto blocks_per_thread = nblocks / nthread;
-    auto spare_blocks = nblocks - (blocks_per_thread * nthread); // if blocks aren't divisible by thread count
-
-    std::vector<std::thread> workers;
-    for (auto tnum = 0, in_buff_offs = 0, out_buff_offs = 0; tnum < nthread; tnum++) {
-        auto thr_blocks = blocks_per_thread + (tnum == nthread - 1 ? spare_blocks : 0); // num blocks for this thread
-        auto thr_elems = thr_blocks * block_size; // number of elements for this thread
-        auto thr_block_bytes = thr_blocks * block_size_bytes; // number of input bytes for this thread
-
-        auto compute = [qtype] (ggml_type typ, uint8_t * inbuf, float * outbuf, int nels) {
-            if (typ == GGML_TYPE_F16) {
-                ggml_fp16_to_fp32_row((ggml_fp16_t *)inbuf, outbuf, nels);
-            } else {
-                qtype.to_float(inbuf, outbuf, nels);
-            }
-        };
-        workers.push_back(std::thread(compute, tensor->type, (uint8_t *) tensor->data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
-        in_buff_offs += thr_block_bytes;
-        out_buff_offs += thr_elems;
-    }
-    for (auto & worker : workers) {
-        worker.join();
-    }
-}
-
-static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
-    ggml_type quantized_type;
-    llama_ftype ftype = params->ftype;
-    int nthread = params->nthread;
-
-    switch (params->ftype) {
-        case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
-        case LLAMA_FTYPE_MOSTLY_Q4_1: quantized_type = GGML_TYPE_Q4_1; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_0: quantized_type = GGML_TYPE_Q5_0; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_1: quantized_type = GGML_TYPE_Q5_1; break;
-        case LLAMA_FTYPE_MOSTLY_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;
-        case LLAMA_FTYPE_MOSTLY_F16:  quantized_type = GGML_TYPE_F16;  break;
-        case LLAMA_FTYPE_ALL_F32:     quantized_type = GGML_TYPE_F32;  break;
-
-#ifdef GGML_USE_K_QUANTS
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K:   quantized_type = GGML_TYPE_Q2_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M:
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: quantized_type = GGML_TYPE_Q3_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: quantized_type = GGML_TYPE_Q4_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S:
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break;
-        case LLAMA_FTYPE_MOSTLY_Q6_K:   quantized_type = GGML_TYPE_Q6_K; break;
-#endif
-        default: throw std::runtime_error(format("invalid output file type %d\n", ftype));
-    }
-
-    if (nthread <= 0) {
-        nthread = std::thread::hardware_concurrency();
-    }
-
-    std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
-
-    const size_t align = GGUF_DEFAULT_ALIGNMENT;
-    struct gguf_context * ctx_out = gguf_init_empty();
-
-    // copy the KV pairs from the input file
-    gguf_set_kv     (ctx_out, model_loader->ctx_gguf);
-    gguf_set_val_u32(ctx_out, "general.quantization_version", GGML_QNT_VERSION);
-
-#ifdef GGML_USE_K_QUANTS
-    int n_attention_wv    = 0;
-    int n_feed_forward_w2 = 0;
-
-    for (int i = 0; i < model_loader->n_tensors; ++i) {
-        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
-
-        const std::string name = ggml_get_name(meta);
-
-        if (name.find("attn_v.weight") != std::string::npos) {
-            ++n_attention_wv;
-        }
-        else if (name.find("ffn_down.weight") != std::string::npos) {
-            ++n_feed_forward_w2;
-        }
-    }
-
-    int i_attention_wv = 0;
-    int i_feed_forward_w2 = 0;
-#endif
-
-    size_t total_size_org = 0;
-    size_t total_size_new = 0;
-    std::vector<int64_t> hist_all(1 << 4, 0);
-
-    std::vector<std::thread> workers;
-    std::mutex mutex;
-
-    auto use_more_bits = [] (int i_layer, int num_layers) -> bool {
-        return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
-    };
-
-    int idx = 0;
-
-    std::vector<uint8_t> read_data;
-    std::vector<uint8_t> work;
-
-    // populate the original tensors so we get an initial meta data
-    for (int i = 0; i < model_loader->n_tensors; ++i) {
-        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
-        gguf_add_tensor(ctx_out, meta);
-    }
-
-    std::ofstream fout(fname_out, std::ios::binary);
-
-    const size_t meta_size = gguf_get_meta_size(ctx_out);
-
-    LLAMA_LOG_INFO("%s: meta size = %zu bytes\n", __func__, meta_size);
-
-    // placeholder for the meta data
-    ::zeros(fout, meta_size);
-
-    for (int i = 0; i < model_loader->n_tensors; ++i) {
-        struct ggml_tensor * tensor = model_loader->get_tensor_meta(i);
-
-        const std::string name = ggml_get_name(tensor);
-
-        read_data.resize(ggml_nbytes(tensor));
-        tensor->data = read_data.data();
-        model_loader->load_data_for(tensor);
-
-        LLAMA_LOG_INFO("[%4d/%4d] %36s - [%s], type = %6s, ",
-               ++idx, model_loader->n_tensors,
-               ggml_get_name(tensor),
-               llama_format_tensor_shape(tensor).c_str(),
-               ggml_type_name(tensor->type));
-
-        // This used to be a regex, but <regex> has an extreme cost to compile times.
-        bool quantize = name.rfind("weight") == name.size() - 6; // ends with 'weight'?
-
-        // quantize only 2D tensors
-        quantize &= (tensor->n_dims == 2);
-        quantize &= params->quantize_output_tensor || name != "output.weight";
-        quantize &= quantized_type != tensor->type;
-
-        enum ggml_type new_type;
-        void * new_data;
-        size_t new_size;
-
-        if (!quantize) {
-            new_type = tensor->type;
-            new_data = tensor->data;
-            new_size = ggml_nbytes(tensor);
-            LLAMA_LOG_INFO("size = %8.3f MB\n", ggml_nbytes(tensor)/1024.0/1024.0);
-        } else {
-            new_type = quantized_type;
-#ifdef GGML_USE_K_QUANTS
-            if (name == TN_OUTPUT) {
-                int nx = tensor->ne[0];
-                int ny = tensor->ne[1];
-                if (nx % QK_K == 0 && ny % QK_K == 0) {
-                    new_type = GGML_TYPE_Q6_K;
-                }
-            } else if (name.find("attn_v.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
-                        use_more_bits(i_attention_wv, n_attention_wv)) new_type = GGML_TYPE_Q6_K;
-                else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
-                        (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
-                ++i_attention_wv;
-            } else if (name.find("feed_forward.w2.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-                else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
-                         use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
-                //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
-                ++i_feed_forward_w2;
-            } else if (name.find("attn_output.weight") != std::string::npos) {
-                if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
-                else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
-            }
-            bool convert_incompatible_tensor = false;
-            if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
-                new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
-                int nx = tensor->ne[0];
-                int ny = tensor->ne[1];
-                if (nx % QK_K != 0 || ny % QK_K != 0) {
-                    LLAMA_LOG_INFO("\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
-                    convert_incompatible_tensor = true;
-                }
-            }
-            if (convert_incompatible_tensor) {
-                if (name == TN_OUTPUT) {
-                    new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
-                    LLAMA_LOG_WARN("F16 will be used for this tensor instead.\n");
-                } else if (name == TN_TOKEN_EMBD) {
-                    new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
-                    LLAMA_LOG_WARN("Q4_0 will be used for this tensor instead.\n");
-                } else {
-                    throw std::runtime_error("Unsupported tensor size encountered\n");
-                }
-            }
-#endif
-
-            const size_t nelements = ggml_nelements(tensor);
-
-            float * f32_data;
-            std::vector<float> f32_conv_buf;
-
-            if (tensor->type == GGML_TYPE_F32) {
-                f32_data = (float *) tensor->data;
-            } else if (ggml_is_quantized(tensor->type) && !params->allow_requantize) {
-                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor->type)));
-            } else {
-                llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
-                f32_data = (float *) f32_conv_buf.data();
-            }
-
-            LLAMA_LOG_INFO("quantizing to %s .. ", ggml_type_name(new_type));
-            fflush(stdout);
-
-            work.resize(nelements * 4); // upper bound on size
-            new_data = work.data();
-            std::vector<int64_t> hist_cur(1 << 4, 0);
-
-            const int chunk_size = 32 * 512;
-            const int nchunk = (nelements + chunk_size - 1)/chunk_size;
-            const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
-            if (nthread_use < 2) {
-                new_size = ggml_quantize_chunk(new_type, f32_data, new_data, 0, nelements, hist_cur.data());
-            } else {
-                size_t counter = 0;
-                new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements] () {
-                    std::vector<int64_t> local_hist;
-                    size_t local_size = 0;
-                    while (true) {
-                        std::unique_lock<std::mutex> lock(mutex);
-                        size_t first = counter; counter += chunk_size;
-                        if (first >= nelements) {
-                            if (!local_hist.empty()) {
-                                for (int j=0; j<int(local_hist.size()); ++j) {
-                                    hist_cur[j] += local_hist[j];
-                                }
-                                new_size += local_size;
-                            }
-                            break;
-                        }
-                        lock.unlock();
-                        size_t last = std::min(nelements, first + chunk_size);
-                        if (local_hist.empty()) {
-                            local_hist.resize(hist_cur.size(), 0);
-                        }
-                        local_size += ggml_quantize_chunk(new_type, f32_data, new_data, first, last - first, local_hist.data());
-                    }
-                };
-                if ((int) workers.size() < nthread_use - 1) {
-                    workers.resize(nthread_use - 1);
-                }
-                for (int it = 0; it < nthread_use - 1; ++it) {
-                    workers[it] = std::thread(compute);
-                }
-                compute();
-                for (int it = 0; it < nthread_use - 1; ++it) {
-                    workers[it].join();
-                }
-            }
-
-            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", ggml_nbytes(tensor)/1024.0/1024.0, new_size/1024.0/1024.0);
-            int64_t tot_count = 0;
-            for (size_t i = 0; i < hist_cur.size(); i++) {
-                hist_all[i] += hist_cur[i];
-                tot_count += hist_cur[i];
-            }
-
-            if (tot_count > 0) {
-                for (size_t i = 0; i < hist_cur.size(); i++) {
-                    LLAMA_LOG_INFO("%5.3f ", hist_cur[i] / float(nelements));
-                }
-            }
-            LLAMA_LOG_INFO("\n");
-        }
-        total_size_org += ggml_nbytes(tensor);
-        total_size_new += new_size;
-
-        // update the gguf meta data as we go
-        gguf_set_tensor_type(ctx_out, name.c_str(), new_type);
-        gguf_set_tensor_data(ctx_out, name.c_str(), new_data, new_size);
-
-        // write tensor data + padding
-        fout.write((const char *) new_data, new_size);
-        zeros(fout, GGML_PAD(new_size, align) - new_size);
-    }
-
-    // go back to beginning of file and write the updated meta data
-    {
-        fout.seekp(0);
-        std::vector<uint8_t> data(gguf_get_meta_size(ctx_out));
-        gguf_get_meta_data(ctx_out, data.data());
-        fout.write((const char *) data.data(), data.size());
-    }
-
-    fout.close();
-
-    gguf_free(ctx_out);
-
-    LLAMA_LOG_INFO("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
-    LLAMA_LOG_INFO("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
-
-    // print histogram for all tensors
-    {
-        int64_t sum_all = 0;
-        for (size_t i = 0; i < hist_all.size(); i++) {
-            sum_all += hist_all[i];
-        }
-
-        if (sum_all > 0) {
-            LLAMA_LOG_INFO("%s: hist: ", __func__);
-            for (size_t i = 0; i < hist_all.size(); i++) {
-                LLAMA_LOG_INFO("%5.3f ", hist_all[i] / float(sum_all));
-            }
-            LLAMA_LOG_INFO("\n");
-        }
-    }
-}
-
-int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
-    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
-
-    const int64_t t_start_lora_us = ggml_time_us();
-
-    auto fin = std::ifstream(path_lora, std::ios::binary);
-    if (!fin) {
-        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
-        return 1;
-    }
-
-    // verify magic and version
-    {
-        uint32_t magic;
-        fin.read((char *) &magic, sizeof(magic));
-        uint32_t format_version;
-        fin.read((char *) &format_version, sizeof(format_version));
-
-        if (format_version != 1) {
-            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
-            return 1;
-        }
-    }
-
-    int32_t lora_r;
-    int32_t lora_alpha;
-    fin.read((char *) &lora_r, sizeof(lora_r));
-    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
-    float scaling = (float)lora_alpha / (float)lora_r;
-
-    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
-
-    // create a temporary ggml context to store the lora tensors
-    // todo: calculate size from biggest possible tensor
-    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
-    struct ggml_init_params params;
-    params.mem_size   = lora_buf.size();
-    params.mem_buffer = lora_buf.data();
-    params.no_alloc   = false;
-
-    ggml_context * lora_ctx = ggml_init(params);
-    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
-
-    // create a name -> tensor map of the model to accelerate lookups
-    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
-    for (const auto & kv : model.tensors_by_name) {
-        model_tensors.insert(kv);
-    }
-
-    // load base model
-    std::unique_ptr<llama_model_loader> model_loader;
-    ggml_context * base_ctx = NULL;
-    std::vector<uint8_t> base_buf;
-    if (path_base_model) {
-        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
-        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
-
-        size_t ctx_size;
-        size_t mmapped_size;
-        model_loader->calc_sizes(ctx_size, mmapped_size);
-        base_buf.resize(ctx_size);
-
-        ggml_init_params base_params;
-        base_params.mem_size   = base_buf.size();
-        base_params.mem_buffer = base_buf.data();
-        base_params.no_alloc   = model_loader->use_mmap;
-
-        base_ctx = ggml_init(base_params);
-
-        // maybe this should in llama_model_loader
-        if (model_loader->use_mmap) {
-            model_loader->mapping.reset(new llama_mmap(&model_loader->file, /* prefetch */ 0, ggml_is_numa()));
-        }
-    }
-
-    // read tensors and apply
-    bool warned = false;
-    int n_tensors = 0;
-
-    std::vector<uint8_t> work_buffer;
-
-    while (true) {
-        int32_t n_dims;
-        int32_t length;
-        int32_t ftype;
-
-        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
-        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
-        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
-        if (fin.eof()) {
-            break;
-        }
-
-        int32_t ne[2] = { 1, 1 };
-        for (int i = 0; i < n_dims; ++i) {
-            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
-        }
-
-        std::string name;
-        {
-            char buf[1024];
-            fin.read(buf, length);
-            name = std::string(buf, length);
-        }
-
-        // check for lora suffix and get the type of tensor
-        const std::string lora_suffix = ".lora";
-        size_t pos = name.rfind(lora_suffix);
-        if (pos == std::string::npos) {
-            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
-            return 1;
-        }
-
-        std::string lora_type = name.substr(pos + lora_suffix.length());
-        std::string base_name = name;
-        base_name.erase(pos);
-        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
-
-        if (model_tensors.find(base_name) == model_tensors.end()) {
-            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
-            return 1;
-        }
-
-        // create ggml tensor
-        ggml_type wtype;
-        switch (ftype) {
-            case 0: wtype = GGML_TYPE_F32;  break;
-            case 1: wtype = GGML_TYPE_F16;  break;
-            default:
-                    {
-                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
-                                __func__, ftype);
-                        return false;
-                    }
-        }
-        ggml_tensor * lora_tensor;
-        if (n_dims == 2) {
-            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
-        }
-        else {
-            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
-            return 1;
-        }
-        ggml_set_name(lora_tensor, "lora_tensor");
-
-        // load tensor data
-        size_t offset = fin.tellg();
-        size_t tensor_data_size = ggml_nbytes(lora_tensor);
-        offset = (offset + 31) & -32;
-        fin.seekg(offset);
-        fin.read((char*)lora_tensor->data, tensor_data_size);
-
-        lora_tensors[name] = lora_tensor;
-
-        // check if we have both A and B tensors and apply
-        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
-            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
-
-            ggml_tensor * dest_t = model_tensors[base_name];
-
-            offload_func_t offload_func = llama_nop;
-            offload_func_t offload_func_force_inplace = llama_nop;
-
-#ifdef GGML_USE_CUBLAS
-            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
-                if (dest_t->type != GGML_TYPE_F16) {
-                    throw std::runtime_error(format(
-                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
-                }
-                offload_func = ggml_cuda_assign_buffers;
-                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
-            }
-#endif // GGML_USE_CUBLAS
-
-            ggml_tensor * base_t;
-            if (model_loader) {
-                struct gguf_context * ctx_gguf = model_loader->ctx_gguf;
-
-                // load from base model
-                if (gguf_find_tensor(ctx_gguf, base_name.c_str()) < 0) {
-                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
-                    return 1;
-                }
-
-                // TODO: not tested!! maybe not working!
-                base_t = model_loader->create_tensor(base_ctx, base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
-                model_loader->load_data_for(base_t);
-            } else {
-                base_t = dest_t;
-            }
-
-            if (ggml_is_quantized(base_t->type)) {
-                if (!warned) {
-                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
-                                   "use a f16 or f32 base model with --lora-base\n", __func__);
-                    warned = true;
-                }
-            }
-
-            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
-            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
-            ggml_set_name(loraA, "loraA");
-
-            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
-            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
-            ggml_set_name(loraB, "loraB");
-
-            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
-                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
-                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
-                return 1;
-            }
-
-            // w = w + BA*s
-            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
-            offload_func(BA);
-            ggml_set_name(BA, "BA");
-
-            if (scaling != 1.0f) {
-                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
-                ggml_set_name(scale_tensor, "scale_tensor");
-
-                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
-                offload_func(BA);
-                ggml_set_name(BA, "BA_scaled");
-            }
-
-            ggml_tensor * r;
-            if (base_t == dest_t) {
-                r = ggml_add_inplace(lora_ctx, dest_t, BA);
-                offload_func_force_inplace(r);
-                ggml_set_name(r, "r_add_inplace");
-            }
-            else {
-                r = ggml_add(lora_ctx, base_t, BA);
-                offload_func(r);
-                ggml_set_name(r, "r_add");
-
-                r = ggml_cpy(lora_ctx, r, dest_t);
-                offload_func(r);
-                ggml_set_name(r, "r_cpy");
-            }
-
-            struct ggml_cgraph gf = ggml_build_forward(r);
-
-            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
-
-            // we won't need these tensors again, reset the context to save memory
-            ggml_free(lora_ctx);
-            lora_ctx = ggml_init(params);
-            lora_tensors.clear();
-
-            n_tensors++;
-            if (n_tensors % 4 == 0) {
-                LLAMA_LOG_INFO(".");
-            }
-        }
-    }
-
-    // TODO: this should be in a destructor, it will leak on failure
-    ggml_free(lora_ctx);
-    if (base_ctx) {
-        ggml_free(base_ctx);
-    }
-
-    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
-    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
-
-    return 0;
-}
-
-//
-// interface implementation
-//
-
-struct llama_context_params llama_context_default_params() {
-    struct llama_context_params result = {
-        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
-        /*.n_ctx                       =*/ 512,
-        /*.n_batch                     =*/ 512,
-        /*.gpu_layers                  =*/ 0,
-        /*.main_gpu                    =*/ 0,
-        /*.tensor_split                =*/ nullptr,
-        /*.rope_freq_base              =*/ 10000.0f,
-        /*.rope_freq_scale             =*/ 1.0f,
-        /*.progress_callback           =*/ nullptr,
-        /*.progress_callback_user_data =*/ nullptr,
-        /*.low_vram                    =*/ false,
-        /*.mul_mat_q                   =*/ false,
-        /*.f16_kv                      =*/ true,
-        /*.logits_all                  =*/ false,
-        /*.vocab_only                  =*/ false,
-        /*.use_mmap                    =*/ true,
-        /*.use_mlock                   =*/ false,
-        /*.embedding                   =*/ false,
-    };
-
-    return result;
-}
-
-struct llama_model_quantize_params llama_model_quantize_default_params() {
-    struct llama_model_quantize_params result = {
-        /*.nthread                     =*/ 0,
-        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
-        /*.allow_requantize            =*/ false,
-        /*.quantize_output_tensor      =*/ true,
-    };
-
-    return result;
-}
-
-int llama_max_devices(void) {
-    return LLAMA_MAX_DEVICES;
-}
-
-bool llama_mmap_supported(void) {
-    return llama_mmap::SUPPORTED;
-}
-
-bool llama_mlock_supported(void) {
-    return llama_mlock::SUPPORTED;
-}
-
-void llama_backend_init(bool numa) {
-    ggml_time_init();
-
-    // needed to initialize f16 tables
-    {
-        struct ggml_init_params params = { 0, NULL, false };
-        struct ggml_context * ctx = ggml_init(params);
-        ggml_free(ctx);
-    }
-
-    if (numa) {
-        ggml_numa_init();
-    }
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_init();
-#endif
-}
-
-void llama_backend_free(void) {
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_free();
-#endif
-}
-
-int64_t llama_time_us(void) {
-    return ggml_time_us();
-}
-
-struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-    ggml_time_init();
-
-    llama_model * model = new llama_model;
-
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
-
-    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gpu_layers,
-                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,
-                params.low_vram, memory_type, params.use_mmap, params.use_mlock, params.vocab_only,
-                params.progress_callback, params.progress_callback_user_data)) {
-        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
-        delete model;
-        return nullptr;
-    }
-
-    return model;
-}
-
-void llama_free_model(struct llama_model * model) {
-    delete model;
-}
-
-struct llama_context * llama_new_context_with_model(
-                 struct llama_model * model,
-        struct llama_context_params   params) {
-
-    if (!model) {
-        return nullptr;
-    }
-
-    llama_context * ctx = new llama_context(*model);
-
-    if (params.seed == LLAMA_DEFAULT_SEED) {
-        params.seed = time(NULL);
-    }
-
-    unsigned cur_percentage = 0;
-    if (params.progress_callback == NULL) {
-        params.progress_callback_user_data = &cur_percentage;
-        params.progress_callback = [](float progress, void * ctx) {
-            unsigned * cur_percentage_p = (unsigned *) ctx;
-            unsigned percentage = (unsigned) (100 * progress);
-            while (percentage > *cur_percentage_p) {
-                *cur_percentage_p = percentage;
-                LLAMA_LOG_INFO(".");
-                if (percentage >= 100) {
-                    LLAMA_LOG_INFO("\n");
-                }
-            }
-        };
-    }
-
-    ctx->rng = std::mt19937(params.seed);
-    ctx->logits_all = params.logits_all;
-
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
-
-    // reserve memory for context buffers
-    if (!params.vocab_only) {
-        if (!llama_kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
-            LLAMA_LOG_ERROR("%s: llama_kv_cache_init() failed for self-attention cache\n", __func__);
-            llama_free(ctx);
-            return nullptr;
-        }
-
-        {
-            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
-            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
-        }
-
-        const auto & hparams = ctx->model.hparams;
-
-        // resized during inference
-        if (params.logits_all) {
-            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
-        } else {
-            ctx->logits.reserve(hparams.n_vocab);
-        }
-
-        if (params.embedding){
-            ctx->embedding.resize(hparams.n_embd);
-        }
-
-#ifdef LLAMA_USE_ALLOCATOR
-        {
-            static const size_t tensor_alignment = 32;
-            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
-            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
-
-            // create measure allocator
-            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
-
-            // build worst-case graph
-            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
-            int n_past = hparams.n_ctx - n_tokens;
-            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
-            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
-
-            // measure memory requirements for the graph
-            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
-
-            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
-
-            // debug - for comparison with scratch buffer
-            //size_t prev_req =
-            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
-            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
-            //    MEM_REQ_EVAL().at(ctx->model.type);
-            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
-
-            // recreate allocator with exact memory requirements
-            ggml_allocr_free(ctx->alloc);
-
-            ctx->buf_alloc.resize(alloc_size);
-            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
-        }
-#else
-        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
-#endif
-
-#ifdef LLAMA_USE_SCRATCH
-        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
-        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
-#endif
-    }
-
-#ifdef GGML_USE_METAL
-    if (params.n_gpu_layers > 0) {
-        // this allocates all Metal resources and memory buffers
-        ctx->ctx_metal = ggml_metal_init(1);
-
-        if (!ctx->ctx_metal) {
-            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
-            llama_free(ctx);
-            return NULL;
-        }
-
-        void * data_ptr  = NULL;
-        size_t data_size = 0;
-
-        if (params.use_mmap) {
-            data_ptr  = ctx->model.mapping->addr;
-            data_size = ctx->model.mapping->size;
-        } else {
-            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
-            data_size = ggml_get_mem_size  (ctx->model.ctx);
-        }
-
-        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
-
-        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
-
-#define LLAMA_METAL_CHECK_BUF(result)                            \
-    if (!(result)) {                                             \
-        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
-        llama_free(ctx);                                         \
-        return NULL;                                             \
-    }
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
-
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
-#undef LLAMA_METAL_CHECK_BUF
-    }
-#endif
-
-#ifdef GGML_USE_MPI
-    ctx->ctx_mpi = ggml_mpi_init();
-
-    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
-        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
-        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
-        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
-        llama_backend_free();
-        exit(1);
-    }
-#endif
-
-    return ctx;
-}
-
-struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-
-    struct llama_model * model = llama_load_model_from_file(path_model, params);
-    if (!model) {
-        return nullptr;
-    }
-    struct llama_context * ctx = llama_new_context_with_model(model, params);
-    ctx->model_owner = true;
-    return ctx;
-}
-
-void llama_free(struct llama_context * ctx) {
-    delete ctx;
-}
-
-int llama_model_quantize(
-        const char * fname_inp,
-        const char * fname_out,
-        const llama_model_quantize_params * params) {
-    try {
-        llama_model_quantize_internal(fname_inp, fname_out, params);
-        return 0;
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
-    try {
-        return llama_apply_lora_from_file_internal(ctx->model, path_lora, path_base_model, n_threads);
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("%s: failed to apply lora adapter: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_model_apply_lora_from_file(const struct llama_model * model, const char * path_lora, const char * path_base_model, int n_threads) {
-    try {
-        return llama_apply_lora_from_file_internal(*model, path_lora, path_base_model, n_threads);
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("%s: failed to apply lora adapter: %s\n", __func__, err.what());
-        return 1;
-    }
-}
-
-int llama_get_kv_cache_token_count(const struct llama_context * ctx) {
-    return ctx->kv_self.n;
-}
-
-#define LLAMA_MAX_RNG_STATE (64*1024)
-
-void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed) {
-    if (seed == LLAMA_DEFAULT_SEED) {
-        seed = time(NULL);
-    }
-    ctx->rng.seed(seed);
-}
-
-// Returns the *maximum* size of the state
-size_t llama_get_state_size(const struct llama_context * ctx) {
-    // we don't know size of rng until we actually serialize it. so reserve more than enough memory for its serialized state.
-    // for reference, std::mt19937(1337) serializes to 6701 bytes.
-    const size_t s_rng_size        = sizeof(size_t);
-    const size_t s_rng             = LLAMA_MAX_RNG_STATE;
-    const size_t s_logits_capacity = sizeof(size_t);
-    const size_t s_logits_size     = sizeof(size_t);
-    const size_t s_logits          = ctx->logits.capacity() * sizeof(float);
-    const size_t s_embedding_size  = sizeof(size_t);
-    const size_t s_embedding       = ctx->embedding.size() * sizeof(float);
-    const size_t s_kv_size         = sizeof(size_t);
-    const size_t s_kv_ntok         = sizeof(int);
-    const size_t s_kv              = ctx->kv_self.buf.size;
-
-    const size_t s_total = (
-        + s_rng_size
-        + s_rng
-        + s_logits_capacity
-        + s_logits_size
-        + s_logits
-        + s_embedding_size
-        + s_embedding
-        + s_kv_size
-        + s_kv_ntok
-        + s_kv
-    );
-
-    return s_total;
-}
-
-// llama_context_data
-struct llama_data_context {
-    virtual void write(const void * src, size_t size) = 0;
-    virtual size_t get_size_written() = 0;
-    virtual ~llama_data_context() = default;
-};
-
-struct llama_data_buffer_context : llama_data_context {
-    uint8_t * ptr;
-    size_t size_written = 0;
-
-    llama_data_buffer_context(uint8_t * p) : ptr(p) {}
-
-    void write(const void * src, size_t size) override {
-        memcpy(ptr, src, size);
-        ptr += size;
-        size_written += size;
-    }
-
-    size_t get_size_written() override {
-        return size_written;
-    }
-};
-
-struct llama_data_file_context : llama_data_context {
-    FILE * file;
-    size_t size_written = 0;
-
-    llama_data_file_context(FILE * f) : file(f) {}
-
-    void write(const void * src, size_t size) override {
-        fwrite(src, size, 1, file);
-        size_written += size;
-    }
-
-    size_t get_size_written() override {
-        return size_written;
-    }
-};
-
-/** copy state data into either a buffer or file depending on the passed in context
- *
- * file context:
- * llama_file file("/path", "wb");
- * llama_data_file_context data_ctx(&file);
- * llama_copy_state_data(ctx, &data_ctx);
- *
- * buffer context:
- * std::vector<uint8_t> buf(max_size, 0);
- * llama_data_buffer_context data_ctx(&buf.data());
- * llama_copy_state_data(ctx, &data_ctx);
- *
-*/
-void llama_copy_state_data_internal(struct llama_context * ctx, llama_data_context * data_ctx) {
-    // copy rng
-    {
-        std::stringstream rng_ss;
-        rng_ss << ctx->rng;
-
-        const size_t rng_size = rng_ss.str().size();
-        char rng_buf[LLAMA_MAX_RNG_STATE];
-
-        memset(&rng_buf[0], 0, LLAMA_MAX_RNG_STATE);
-        memcpy(&rng_buf[0], rng_ss.str().data(), rng_ss.str().size());
-
-        data_ctx->write(&rng_size,   sizeof(rng_size));
-        data_ctx->write(&rng_buf[0], LLAMA_MAX_RNG_STATE);
-    }
-
-    // copy logits
-    {
-        const size_t logits_cap  = ctx->logits.capacity();
-        const size_t logits_size = ctx->logits.size();
-
-        data_ctx->write(&logits_cap,  sizeof(logits_cap));
-        data_ctx->write(&logits_size, sizeof(logits_size));
-
-        if (logits_size) {
-            data_ctx->write(ctx->logits.data(), logits_size * sizeof(float));
-        }
-
-        // If there is a gap between the size and the capacity, write padding
-        size_t padding_size = (logits_cap - logits_size) * sizeof(float);
-        if (padding_size > 0) {
-            std::vector<uint8_t> padding(padding_size, 0); // Create a buffer filled with zeros
-            data_ctx->write(padding.data(), padding_size);
-        }
-    }
-
-    // copy embeddings
-    {
-        const size_t embedding_size = ctx->embedding.size();
-
-        data_ctx->write(&embedding_size, sizeof(embedding_size));
-
-        if (embedding_size) {
-            data_ctx->write(ctx->embedding.data(), embedding_size * sizeof(float));
-        }
-    }
-
-    // copy kv cache
-    {
-        const auto & kv_self = ctx->kv_self;
-        const auto & hparams = ctx->model.hparams;
-        const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd_gqa();
-        const int    n_ctx   = hparams.n_ctx;
-
-        const size_t kv_size = kv_self.buf.size;
-        const int    kv_ntok = llama_get_kv_cache_token_count(ctx);
-
-        data_ctx->write(&kv_size, sizeof(kv_size));
-        data_ctx->write(&kv_ntok, sizeof(kv_ntok));
-
-        if (kv_size) {
-            const size_t elt_size = ggml_element_size(kv_self.k);
-
-            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
-            ggml_cgraph gf{};
-
-            ggml_tensor * kout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
-            std::vector<uint8_t> kout3d_data(ggml_nbytes(kout3d), 0);
-            kout3d->data = kout3d_data.data();
-
-            ggml_tensor * vout3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
-            std::vector<uint8_t> vout3d_data(ggml_nbytes(vout3d), 0);
-            vout3d->data = vout3d_data.data();
-
-            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
-                n_embd, kv_ntok, n_layer,
-                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
-
-            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
-                kv_ntok, n_embd, n_layer,
-                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
-
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, k3d, kout3d));
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, v3d, vout3d));
-            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
-
-            ggml_free(cpy_ctx);
-
-            // our data is now in the kout3d_data and vout3d_data buffers
-            // write them to file
-            data_ctx->write(kout3d_data.data(), kout3d_data.size());
-            data_ctx->write(vout3d_data.data(), vout3d_data.size());
-        }
-    }
-}
-
-size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst) {
-    llama_data_buffer_context data_ctx(dst);
-    llama_copy_state_data_internal(ctx, &data_ctx);
-
-    return data_ctx.get_size_written();
-}
-
-// Sets the state reading from the specified source address
-size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
-    uint8_t * inp = src;
-
-    // set rng
-    {
-        size_t rng_size;
-        char   rng_buf[LLAMA_MAX_RNG_STATE];
-
-        memcpy(&rng_size,   inp, sizeof(rng_size));    inp += sizeof(rng_size);
-        memcpy(&rng_buf[0], inp, LLAMA_MAX_RNG_STATE); inp += LLAMA_MAX_RNG_STATE;
-
-        std::stringstream rng_ss;
-        rng_ss.str(std::string(&rng_buf[0], rng_size));
-        rng_ss >> ctx->rng;
-
-        GGML_ASSERT(rng_ss.fail() == false);
-    }
-
-    // set logits
-    {
-        size_t logits_cap;
-        size_t logits_size;
-
-        memcpy(&logits_cap,  inp, sizeof(logits_cap));  inp += sizeof(logits_cap);
-        memcpy(&logits_size, inp, sizeof(logits_size)); inp += sizeof(logits_size);
-
-        GGML_ASSERT(ctx->logits.capacity() == logits_cap);
-
-        if (logits_size) {
-            ctx->logits.resize(logits_size);
-            memcpy(ctx->logits.data(), inp, logits_size * sizeof(float));
-        }
-
-        inp += logits_cap * sizeof(float);
-    }
-
-    // set embeddings
-    {
-        size_t embedding_size;
-
-        memcpy(&embedding_size, inp, sizeof(embedding_size)); inp += sizeof(embedding_size);
-
-        GGML_ASSERT(ctx->embedding.capacity() == embedding_size);
-
-        if (embedding_size) {
-            memcpy(ctx->embedding.data(), inp, embedding_size * sizeof(float));
-            inp += embedding_size * sizeof(float);
-        }
-    }
-
-    // set kv cache
-    {
-        const auto & kv_self = ctx->kv_self;
-        const auto & hparams = ctx->model.hparams;
-        const int    n_layer = hparams.n_layer;
-        const int    n_embd  = hparams.n_embd_gqa();
-        const int    n_ctx   = hparams.n_ctx;
-
-        size_t kv_size;
-        int kv_ntok;
-
-        memcpy(&kv_size, inp, sizeof(kv_size)); inp += sizeof(kv_size);
-        memcpy(&kv_ntok, inp, sizeof(kv_ntok)); inp += sizeof(kv_ntok);
-
-        if (kv_size) {
-            GGML_ASSERT(kv_self.buf.size == kv_size);
-
-            const size_t elt_size = ggml_element_size(kv_self.k);
-
-            ggml_context * cpy_ctx = ggml_init({ 4096, NULL, /* no_alloc */ true });
-            ggml_cgraph gf{};
-
-            ggml_tensor * kin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.k->type, n_embd, kv_ntok, n_layer);
-            kin3d->data = (void *) inp;
-            inp += ggml_nbytes(kin3d);
-
-            ggml_tensor * vin3d = ggml_new_tensor_3d(cpy_ctx, kv_self.v->type, kv_ntok, n_embd, n_layer);
-            vin3d->data = (void *) inp;
-            inp += ggml_nbytes(vin3d);
-
-            ggml_tensor * k3d = ggml_view_3d(cpy_ctx, kv_self.k,
-                n_embd, kv_ntok, n_layer,
-                elt_size*n_embd, elt_size*n_embd*n_ctx, 0);
-
-            ggml_tensor * v3d = ggml_view_3d(cpy_ctx, kv_self.v,
-                kv_ntok, n_embd, n_layer,
-                elt_size*n_ctx, elt_size*n_ctx*n_embd, 0);
-
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, kin3d, k3d));
-            ggml_build_forward_expand(&gf, ggml_cpy(cpy_ctx, vin3d, v3d));
-            ggml_graph_compute_helper(ctx->work_buffer, &gf, /*n_threads*/ 1);
-
-            ggml_free(cpy_ctx);
-        }
-
-        ctx->kv_self.n = kv_ntok;
-    }
-
-    const size_t nread    = inp - src;
-    const size_t max_size = llama_get_state_size(ctx);
-
-    GGML_ASSERT(nread <= max_size);
-
-    return nread;
-}
-
-static bool llama_load_session_file_internal(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
-    llama_file file(path_session, "rb");
-    GGML_UNUSED(ctx);
-    GGML_UNUSED(path_session);
-    GGML_UNUSED(tokens_out);
-    GGML_UNUSED(n_token_capacity);
-    GGML_UNUSED(n_token_count_out);
-
-    // TODO: implement with GGUF format
-    return true;
-}
-
-bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out) {
-    try {
-        return llama_load_session_file_internal(ctx, path_session, tokens_out, n_token_capacity, n_token_count_out);
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("error loading session file: %s\n", err.what());
-        return false;
-    }
-}
-
-bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count) {
-    llama_file file(path_session, "wb");
-    GGML_UNUSED(ctx);
-    GGML_UNUSED(tokens);
-    GGML_UNUSED(n_token_count);
-
-    // TODO: implement with GGUF format
-
-    return true;
-}
-
-int llama_eval(
-        struct llama_context * ctx,
-           const llama_token * tokens,
-                         int   n_tokens,
-                         int   n_past,
-                         int   n_threads) {
-    if (!llama_eval_internal(*ctx, tokens, nullptr, n_tokens, n_past, n_threads, nullptr)) {
-        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    // get a more accurate load time, upon first eval
-    // TODO: fix this
-    if (!ctx->has_evaluated_once) {
-        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
-        ctx->has_evaluated_once = true;
-    }
-
-    return 0;
-}
-
-int llama_eval_embd(
-            struct llama_context * ctx,
-                     const float * embd,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads) {
-    if (!llama_eval_internal(*ctx, nullptr, embd, n_tokens, n_past, n_threads, nullptr)) {
-        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    // get a more accurate load time, upon first eval
-    // TODO: fix this
-    if (!ctx->has_evaluated_once) {
-        ctx->t_load_us = ggml_time_us() - ctx->t_start_us;
-        ctx->has_evaluated_once = true;
-    }
-
-    return 0;
-}
-
-int llama_eval_export(struct llama_context * ctx, const char * fname) {
-    const int n_batch = 1;
-    const int n_ctx   = 512 - n_batch;
-
-    const std::vector<llama_token> tmp(n_batch, llama_token_bos());
-
-    if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
-        LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
-        return 1;
-    }
-
-    return 0;
-}
-
-int llama_tokenize_with_model(
-    const struct llama_model * model,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    auto escape = llama_vocab_type(model->vocab) == "spm";
-    auto res = llama_tokenize(model->vocab, text, add_bos, escape);
-
-    if (n_max_tokens < (int) res.size()) {
-        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
-        return -((int) res.size());
-    }
-
-    for (size_t i = 0; i < res.size(); i++) {
-        tokens[i] = res[i];
-    }
-
-    return res.size();
-}
-
-int llama_tokenize(
-        struct llama_context * ctx,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
-}
-
-std::vector<llama_token> llama_tokenize(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos) {
-    int length = text.length() + add_bos;
-    std::vector<llama_token> result(length);
-    length = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -length);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(length);
-    }
-    return result;
-}
-
-int llama_tokenize_bpe(
-        struct llama_context * ctx,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    auto res = llama_tokenize(ctx->model.vocab, text, add_bos, false);
-
-    if (n_max_tokens < (int) res.size()) {
-        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
-        return -((int) res.size());
-    }
-
-    for (size_t i = 0; i < res.size(); i++) {
-        tokens[i] = res[i];
-    }
-
-    return res.size();
-}
-
-std::vector<llama_token> llama_tokenize_bpe(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos) {
-    int length = text.length() + add_bos;
-    std::vector<llama_token> result(length);
-    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -length);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(length);
-    }
-    return result;
-}
-
-int llama_n_vocab_from_model(const struct llama_model * model) {
-    return model->vocab.id_to_token.size();
-}
-
-int llama_n_ctx_from_model(const struct llama_model * model) {
-    return model->hparams.n_ctx;
-}
-
-int llama_n_embd_from_model(const struct llama_model * model) {
-    return model->hparams.n_embd;
-}
-
-int llama_n_vocab(const struct llama_context * ctx) {
-    return ctx->model.vocab.id_to_token.size();
-}
-
-int llama_n_ctx(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_ctx;
-}
-
-int llama_n_embd(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_embd;
-}
-
-int llama_get_vocab_from_model(
-        const struct llama_model * model,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
-    for (int i = 0; i<n; ++i) {
-        strings[i] = model->vocab.id_to_token[i].tok.c_str();
-        scores[i]  = model->vocab.id_to_token[i].score;
-    }
-    return n;
-}
-
-int llama_get_vocab(
-        const struct llama_context * ctx,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
-}
-
-float * llama_get_logits(struct llama_context * ctx) {
-    return ctx->logits.data();
-}
-
-float * llama_get_embeddings(struct llama_context * ctx) {
-    return ctx->embedding.data();
-}
-
-int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * str, int length) {
-    if (0 <= token && token < llama_n_vocab_from_model(model)) {
-        if (llama_is_normal_token(model->vocab, token)) {
-            std::string result = model->vocab.id_to_token[token].tok;
-            if(llama_vocab_type(model->vocab) == "spm") {
-                result = llama_unescape_whitespace(result);
-            }
-            if (length < (int) result.length()) {
-                return -result.length();
-            }
-            strncpy(str, result.c_str(), result.length());
-            return result.length();
-        } else if (llama_is_unknown_token(model->vocab, token)) { // NOLINT
-            if (length < 3) {
-                return -3;
-            }
-            strncpy(str, "\xe2\x96\x85", 3);
-            return 3;
-        } else if (llama_is_control_token(model->vocab, token)) {
-            ;
-        } else if (llama_is_byte_token(model->vocab, token)) {
-            if (length < 1) {
-                return -1;
-            }
-            str[0] = llama_byte_to_char(model->vocab, token);
-            str[1] = 0x00;
-            return 1;
-        }
-    }
-    return 0;
-}
-
-int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * str, int length) {
-    return llama_token_to_str_with_model(&ctx->model, token, str, length);
-}
-
-std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_token_to_str(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
-int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * str, int length) {
-    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
-        std::string result = ctx->model.vocab.id_to_token[token].tok;
-        if (length < (int) result.length()) {
-            return -result.length();
-        }
-        strncpy(str, result.c_str(), result.length());
-        return result.length();
-    }
-    return 0;
-}
-
-std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
-llama_token llama_token_bos(void) {
-    return 1;
-}
-
-llama_token llama_token_eos(void) {
-    return 2;
-}
-
-llama_token llama_token_nl(void) {
-    return 13;
-}
-
-struct llama_timings llama_get_timings(struct llama_context * ctx) {
-    struct llama_timings result = {
-        /*.t_start_ms  =*/ 1e-3 * ctx->t_start_us,
-        /*.t_end_ms    =*/ 1.00 * ggml_time_ms(),
-        /*.t_load_ms   =*/ 1e-3 * ctx->t_load_us,
-        /*.t_sample_ms =*/ 1e-3 * ctx->t_sample_us,
-        /*.t_p_eval_ms =*/ 1e-3 * ctx->t_p_eval_us,
-        /*.t_eval_ms   =*/ 1e-3 * ctx->t_eval_us,
-
-        /*.n_sample =*/ std::max(1, ctx->n_sample),
-        /*.n_p_eval =*/ std::max(1, ctx->n_p_eval),
-        /*.n_eval   =*/ std::max(1, ctx->n_eval),
-    };
-
-    return result;
-}
-
-void llama_print_timings(struct llama_context * ctx) {
-    const llama_timings timings = llama_get_timings(ctx);
-
-    LLAMA_LOG_INFO("\n");
-    LLAMA_LOG_INFO("%s:        load time = %8.2f ms\n", __func__, timings.t_load_ms);
-    LLAMA_LOG_INFO("%s:      sample time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_sample_ms, timings.n_sample, timings.t_sample_ms / timings.n_sample, 1e3 / timings.t_sample_ms * timings.n_sample);
-    LLAMA_LOG_INFO("%s: prompt eval time = %8.2f ms / %5d tokens (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_p_eval_ms, timings.n_p_eval, timings.t_p_eval_ms / timings.n_p_eval, 1e3 / timings.t_p_eval_ms * timings.n_p_eval);
-    LLAMA_LOG_INFO("%s:        eval time = %8.2f ms / %5d runs   (%8.2f ms per token, %8.2f tokens per second)\n",
-            __func__, timings.t_eval_ms, timings.n_eval, timings.t_eval_ms / timings.n_eval, 1e3 / timings.t_eval_ms * timings.n_eval);
-    LLAMA_LOG_INFO("%s:       total time = %8.2f ms\n", __func__, (timings.t_end_ms - timings.t_start_ms));
-}
-
-void llama_reset_timings(struct llama_context * ctx) {
-    ctx->t_start_us = ggml_time_us();
-    ctx->t_sample_us = ctx->n_sample = 0;
-    ctx->t_eval_us   = ctx->n_eval   = 0;
-    ctx->t_p_eval_us = ctx->n_p_eval = 0;
-}
-
-const char * llama_print_system_info(void) {
-    static std::string s;
-
-    s  = "";
-    s += "AVX = "         + std::to_string(ggml_cpu_has_avx())         + " | ";
-    s += "AVX2 = "        + std::to_string(ggml_cpu_has_avx2())        + " | ";
-    s += "AVX512 = "      + std::to_string(ggml_cpu_has_avx512())      + " | ";
-    s += "AVX512_VBMI = " + std::to_string(ggml_cpu_has_avx512_vbmi()) + " | ";
-    s += "AVX512_VNNI = " + std::to_string(ggml_cpu_has_avx512_vnni()) + " | ";
-    s += "FMA = "         + std::to_string(ggml_cpu_has_fma())         + " | ";
-    s += "NEON = "        + std::to_string(ggml_cpu_has_neon())        + " | ";
-    s += "ARM_FMA = "     + std::to_string(ggml_cpu_has_arm_fma())     + " | ";
-    s += "F16C = "        + std::to_string(ggml_cpu_has_f16c())        + " | ";
-    s += "FP16_VA = "     + std::to_string(ggml_cpu_has_fp16_va())     + " | ";
-    s += "WASM_SIMD = "   + std::to_string(ggml_cpu_has_wasm_simd())   + " | ";
-    s += "BLAS = "        + std::to_string(ggml_cpu_has_blas())        + " | ";
-    s += "SSE3 = "        + std::to_string(ggml_cpu_has_sse3())        + " | ";
-    s += "VSX = "         + std::to_string(ggml_cpu_has_vsx())         + " | ";
-
-    return s.c_str();
-}
-
-// For internal test use
-const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
-    return ctx->model.tensors_by_name;
-}
-
-void llama_log_set(llama_log_callback log_callback, void * user_data) {
-    g_state.log_callback = log_callback ? log_callback : llama_log_callback_default;
-    g_state.log_callback_user_data = user_data;
-}
-
-#if defined(_MSC_VER) && !defined(vsnprintf)
-#define vsnprintf _vsnprintf
-#endif
-
-static void llama_log_internal_v(llama_log_level level, const char * format, va_list args) {
-    va_list args_copy;
-    va_copy(args_copy, args);
-    char buffer[128];
-    int len = vsnprintf(buffer, 128, format, args);
-    if (len < 128) {
-        g_state.log_callback(level, buffer, g_state.log_callback_user_data);
-    } else {
-        char* buffer2 = new char[len+1];
-        vsnprintf(buffer2, len+1, format, args_copy);
-        buffer2[len] = 0;
-        g_state.log_callback(level, buffer2, g_state.log_callback_user_data);
-        delete[] buffer2;
-    }
-    va_end(args_copy);
-}
-
-static void llama_log_internal(llama_log_level level, const char * format, ...) {
-    va_list args;
-    va_start(args, format);
-    llama_log_internal_v(level, format, args);
-    va_end(args);
-}
-
-static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data) {
-    (void) level;
-    (void) user_data;
-    fputs(text, stderr);
-    fflush(stderr);
-}
diff --git a/gguf-llama.h b/gguf-llama.h
deleted file mode 100644
index d81a1b5defd89..0000000000000
--- a/gguf-llama.h
+++ /dev/null
@@ -1,505 +0,0 @@
-#ifndef LLAMA_H
-#define LLAMA_H
-
-#include "ggml.h"
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-#define LLAMA_MAX_DEVICES GGML_CUDA_MAX_DEVICES
-#else
-#define LLAMA_MAX_DEVICES 1
-#endif // GGML_USE_CUBLAS
-#include <stddef.h>
-#include <stdint.h>
-#include <stdbool.h>
-
-#ifdef LLAMA_SHARED
-#    if defined(_WIN32) && !defined(__MINGW32__)
-#        ifdef LLAMA_BUILD
-#            define LLAMA_API __declspec(dllexport)
-#        else
-#            define LLAMA_API __declspec(dllimport)
-#        endif
-#    else
-#        define LLAMA_API __attribute__ ((visibility ("default")))
-#    endif
-#else
-#    define LLAMA_API
-#endif
-
-#ifdef __GNUC__
-#    define DEPRECATED(func, hint) func __attribute__((deprecated(hint)))
-#elif defined(_MSC_VER)
-#    define DEPRECATED(func, hint) __declspec(deprecated(hint)) func
-#else
-#    define DEPRECATED(func, hint) func
-#endif
-
-#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
-
-#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
-// Defined when llama.cpp is compiled with support for offloading model layers to GPU.
-#define LLAMA_SUPPORTS_GPU_OFFLOAD
-#endif
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-    //
-    // C interface
-    //
-    // TODO: show sample usage
-    //
-
-    struct llama_model;
-    struct llama_context;
-
-    typedef int llama_token;
-
-    typedef struct llama_token_data {
-        llama_token id; // token id
-        float logit;    // log-odds of the token
-        float p;        // probability of the token
-    } llama_token_data;
-
-    typedef struct llama_token_data_array {
-        llama_token_data * data;
-        size_t size;
-        bool sorted;
-    } llama_token_data_array;
-
-    typedef void (*llama_progress_callback)(float progress, void *ctx);
-
-    enum llama_log_level {
-        LLAMA_LOG_LEVEL_ERROR = 2,
-        LLAMA_LOG_LEVEL_WARN  = 3,
-        LLAMA_LOG_LEVEL_INFO  = 4
-    };
-
-    // Signature for logging events
-    // Note that text includes the new line character at the end for most events.
-    // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
-    // if it exists.
-    // It might not exist for progress report where '.' is output repeatedly.
-    typedef void (*llama_log_callback)(enum llama_log_level level, const char * text, void * user_data);
-
-    struct llama_context_params {
-        uint32_t seed;         // RNG seed, -1 for random
-        int32_t  n_ctx;        // text context
-        int32_t  n_batch;      // prompt processing batch size
-        int32_t  n_gpu_layers; // number of layers to store in VRAM
-        int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
-
-        const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)
-
-        // ref: https://github.com/ggerganov/llama.cpp/pull/2054
-        float    rope_freq_base;  // RoPE base frequency
-        float    rope_freq_scale; // RoPE frequency scaling factor
-
-        // called with a progress value between 0 and 1, pass NULL to disable
-        llama_progress_callback progress_callback;
-        // context pointer passed to the progress callback
-        void * progress_callback_user_data;
-
-        // Keep the booleans together to avoid misalignment during copy-by-value.
-        bool low_vram;   // if true, reduce VRAM usage at the cost of performance
-        bool mul_mat_q;  // if true, use experimental mul_mat_q kernels
-        bool f16_kv;     // use fp16 for KV cache
-        bool logits_all; // the llama_eval() call computes all logits, not just the last one
-        bool vocab_only; // only load the vocabulary, no weights
-        bool use_mmap;   // use mmap if possible
-        bool use_mlock;  // force system to keep model in RAM
-        bool embedding;  // embedding mode only
-    };
-
-    // model file types
-    enum llama_ftype {
-        LLAMA_FTYPE_ALL_F32              = 0,
-        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
-        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
-        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
-        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
-    };
-
-    // model quantization parameters
-    typedef struct llama_model_quantize_params {
-        int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
-        enum llama_ftype ftype;      // quantize to this llama_ftype
-        bool allow_requantize;       // allow quantizing non-f32/f16 tensors
-        bool quantize_output_tensor; // quantize output.weight
-    } llama_model_quantize_params;
-
-    // grammar types
-    struct llama_grammar;
-
-    // grammar element type
-    enum llama_gretype {
-        // end of rule definition
-        LLAMA_GRETYPE_END            = 0,
-
-        // start of alternate definition for rule
-        LLAMA_GRETYPE_ALT            = 1,
-
-        // non-terminal element: reference to rule
-        LLAMA_GRETYPE_RULE_REF       = 2,
-
-        // terminal element: character (code point)
-        LLAMA_GRETYPE_CHAR           = 3,
-
-        // inverse char(s) ([^a], [^a-b] [^abc])
-        LLAMA_GRETYPE_CHAR_NOT       = 4,
-
-        // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
-        // be an inclusive range ([a-z])
-        LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,
-
-        // modifies a preceding LLAMA_GRETYPE_CHAR or
-        // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
-        LLAMA_GRETYPE_CHAR_ALT       = 6,
-    };
-
-    typedef struct llama_grammar_element {
-        enum llama_gretype type;
-        uint32_t           value; // Unicode code point or rule ID
-    } llama_grammar_element;
-
-    // performance timing information
-    struct llama_timings {
-        double t_start_ms;
-        double t_end_ms;
-        double t_load_ms;
-        double t_sample_ms;
-        double t_p_eval_ms;
-        double t_eval_ms;
-
-        int32_t n_sample;
-        int32_t n_p_eval;
-        int32_t n_eval;
-    };
-
-    LLAMA_API struct llama_context_params llama_context_default_params(void);
-    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
-
-    LLAMA_API int  llama_max_devices(void);
-    LLAMA_API bool llama_mmap_supported(void);
-    LLAMA_API bool llama_mlock_supported(void);
-
-    // TODO: not great API - very likely to change
-    // Initialize the llama + ggml backend
-    // If numa is true, use NUMA optimizations
-    // Call once at the start of the program
-    LLAMA_API void llama_backend_init(bool numa);
-    // Call once at the end of the program - currently only used for MPI
-    LLAMA_API void llama_backend_free(void);
-
-    LLAMA_API int64_t llama_time_us(void);
-
-    LLAMA_API struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params);
-
-    LLAMA_API void llama_free_model(struct llama_model * model);
-
-    LLAMA_API struct llama_context * llama_new_context_with_model(
-                     struct llama_model * model,
-            struct llama_context_params   params);
-
-
-    // Frees all allocated memory
-    LLAMA_API void llama_free(struct llama_context * ctx);
-
-    // Returns 0 on success
-    LLAMA_API int llama_model_quantize(
-            const char * fname_inp,
-            const char * fname_out,
-            const llama_model_quantize_params * params);
-
-    // Apply a LoRA adapter to a loaded model
-    // path_base_model is the path to a higher quality model to use as a base for
-    // the layers modified by the adapter. Can be NULL to use the current loaded model.
-    // The model needs to be reloaded before applying a new adapter, otherwise the adapter
-    // will be applied on top of the previous one
-    // Returns 0 on success
-    LLAMA_API DEPRECATED(int llama_apply_lora_from_file(
-            struct llama_context * ctx,
-                      const char * path_lora,
-                      const char * path_base_model,
-                             int   n_threads),
-            "please use llama_model_apply_lora_from_file instead");
-
-    LLAMA_API int llama_model_apply_lora_from_file(
-            const struct llama_model * model,
-                      const char * path_lora,
-                      const char * path_base_model,
-                             int   n_threads);
-
-    // Returns the number of tokens in the KV cache
-    LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
-
-    // Sets the current rng seed.
-    LLAMA_API void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed);
-
-    // Returns the maximum size in bytes of the state (rng, logits, embedding
-    // and kv_cache) - will often be smaller after compacting tokens
-    LLAMA_API size_t llama_get_state_size(const struct llama_context * ctx);
-
-    // Copies the state to the specified destination address.
-    // Destination needs to have allocated enough memory.
-    // Returns the number of bytes copied
-    LLAMA_API size_t llama_copy_state_data(struct llama_context * ctx, uint8_t * dst);
-
-    // Set the state reading from the specified address
-    // Returns the number of bytes read
-    LLAMA_API size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src);
-
-    // Save/load session file
-    LLAMA_API bool llama_load_session_file(struct llama_context * ctx, const char * path_session, llama_token * tokens_out, size_t n_token_capacity, size_t * n_token_count_out);
-    LLAMA_API bool llama_save_session_file(struct llama_context * ctx, const char * path_session, const llama_token * tokens, size_t n_token_count);
-
-    // Run the llama inference to obtain the logits and probabilities for the next token.
-    // tokens + n_tokens is the provided batch of new tokens to process
-    // n_past is the number of tokens to use from previous eval calls
-    // Returns 0 on success
-    LLAMA_API int llama_eval(
-            struct llama_context * ctx,
-               const llama_token * tokens,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads);
-
-    // Same as llama_eval, but use float matrix input directly.
-    LLAMA_API int llama_eval_embd(
-            struct llama_context * ctx,
-                     const float * embd,
-                             int   n_tokens,
-                             int   n_past,
-                             int   n_threads);
-
-    // Export a static computation graph for context of 511 and batch size of 1
-    // NOTE: since this functionality is mostly for debugging and demonstration purposes, we hardcode these
-    //       parameters here to keep things simple
-    // IMPORTANT: do not use for anything else other than debugging and testing!
-    LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
-
-    // Convert the provided text into tokens.
-    // The tokens pointer must be large enough to hold the resulting tokens.
-    // Returns the number of tokens on success, no more than n_max_tokens
-    // Returns a negative number on failure - the number of tokens that would have been returned
-    // TODO: not sure if correct
-    LLAMA_API int llama_tokenize(
-            struct llama_context * ctx,
-                      const char * text,
-                     llama_token * tokens,
-                             int   n_max_tokens,
-                            bool   add_bos);
-
-    LLAMA_API int llama_tokenize_bpe(
-            struct llama_context * ctx,
-                      const char * text,
-                     llama_token * tokens,
-                             int   n_max_tokens,
-                            bool   add_bos);
-
-    LLAMA_API int llama_tokenize_with_model(
-        const struct llama_model * model,
-                      const char * text,
-                     llama_token * tokens,
-                             int   n_max_tokens,
-                            bool   add_bos);
-
-    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
-    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
-    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
-
-    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
-    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
-    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
-
-    // Get the vocabulary as output parameters.
-    // Returns number of results.
-    LLAMA_API int llama_get_vocab(
-            const struct llama_context * ctx,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    LLAMA_API int llama_get_vocab_from_model(
-              const struct llama_model * model,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    // Token logits obtained from the last call to llama_eval()
-    // The logits for the last token are stored in the last row
-    // Can be mutated in order to change the probabilities of the next token
-    // Rows: n_tokens
-    // Cols: n_vocab
-    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
-
-    // Get the embeddings for the input
-    // shape: [n_embd] (1-dimensional)
-    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
-
-    // Token Id -> String. Uses the vocabulary in the provided context
-    LLAMA_API int llama_token_to_str(
-            const struct llama_context * ctx,
-                           llama_token   token,
-                                  char * str,
-                                  int    length);
-
-    LLAMA_API int llama_token_to_str_bpe(
-            const struct llama_context * ctx,
-                           llama_token   token,
-                                  char * str,
-                                  int    length);
-
-    LLAMA_API int llama_token_to_str_with_model(
-              const struct llama_model * model,
-                           llama_token   token,
-                                  char * str,
-                                  int    length);
-    // Special tokens
-    LLAMA_API llama_token llama_token_bos(void);  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos(void);  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl(void);   // next-line
-
-    // Grammar
-    //
-    LLAMA_API struct llama_grammar * llama_grammar_init(
-            const llama_grammar_element ** rules,
-                                 size_t    n_rules,
-                                 size_t    start_rule_index);
-
-    LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
-
-    // Sampling functions
-
-    /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
-    LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);
-
-    /// @details Frequency and presence penalties described in OpenAI API https://platform.openai.com/docs/api-reference/parameter-details.
-    LLAMA_API void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float alpha_frequency, float alpha_presence);
-
-    /// @details Apply classifier-free guidance to the logits as described in academic paper "Stay on topic with Classifier-Free Guidance" https://arxiv.org/abs/2306.17806
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, the logits must be directly extracted from the original generation context without being sorted.
-    /// @params guidance_ctx A separate context from the same model. Other than a negative prompt at the beginning, it should have all generated and user input tokens copied from the main context.
-    /// @params scale Guidance strength. 1.0f means no guidance. Higher values mean stronger guidance.
-    LLAMA_API void llama_sample_classifier_free_guidance(
-              struct llama_context * ctx,
-            llama_token_data_array * candidates,
-              struct llama_context * guidance_ctx,
-                             float   scale);
-
-    /// @details Sorts candidate tokens by their logits in descending order and calculate probabilities based on logits.
-    LLAMA_API void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Top-K sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
-    LLAMA_API void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int k, size_t min_keep);
-
-    /// @details Nucleus sampling described in academic paper "The Curious Case of Neural Text Degeneration" https://arxiv.org/abs/1904.09751
-    LLAMA_API void llama_sample_top_p(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
-
-    /// @details Tail Free Sampling described in https://www.trentonbricken.com/Tail-Free-Sampling/.
-    LLAMA_API void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array * candidates, float z, size_t min_keep);
-
-    /// @details Locally Typical Sampling implementation described in the paper https://arxiv.org/abs/2202.00666.
-    LLAMA_API void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep);
-    LLAMA_API void llama_sample_temperature(struct llama_context * ctx, llama_token_data_array * candidates, float temp);
-
-    /// @details Apply constraints from grammar
-    LLAMA_API void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar);
-
-    /// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
-    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
-    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
-    /// @param m The number of tokens considered in the estimation of `s_hat`. This is an arbitrary value that is used to calculate `s_hat`, which in turn helps to calculate the value of `k`. In the paper, they use `m = 100`, but you can experiment with different values to see how it affects the performance of the algorithm.
-    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
-    LLAMA_API llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu);
-
-    /// @details Mirostat 2.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.
-    /// @param candidates A vector of `llama_token_data` containing the candidate tokens, their probabilities (p), and log-odds (logit) for the current position in the generated text.
-    /// @param tau  The target cross-entropy (or surprise) value you want to achieve for the generated text. A higher value corresponds to more surprising or less predictable text, while a lower value corresponds to less surprising or more predictable text.
-    /// @param eta The learning rate used to update `mu` based on the error between the target and observed surprisal of the sampled word. A larger learning rate will cause `mu` to be updated more quickly, while a smaller learning rate will result in slower updates.
-    /// @param mu Maximum cross-entropy. This value is initialized to be twice the target cross-entropy (`2 * tau`) and is updated in the algorithm based on the error between the target and observed surprisal.
-    LLAMA_API llama_token llama_sample_token_mirostat_v2(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, float * mu);
-
-    /// @details Selects the token with the highest probability.
-    LLAMA_API llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Randomly selects a token from the candidates based on their probabilities.
-    LLAMA_API llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates);
-
-    /// @details Accepts the sampled token into the grammar
-    LLAMA_API void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token);
-
-    // Performance information
-    LLAMA_API struct llama_timings llama_get_timings(struct llama_context * ctx);
-    LLAMA_API void llama_print_timings(struct llama_context * ctx);
-    LLAMA_API void llama_reset_timings(struct llama_context * ctx);
-
-    // Print system information
-    LLAMA_API const char * llama_print_system_info(void);
-
-    // Set callback for all future logging events.
-    // If this is not called, or NULL is supplied, everything is output on stderr.
-    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
-
-#ifdef __cplusplus
-}
-#endif
-
-// C++ API, will be moving to common.h soon (TM)
-#ifdef LLAMA_API_CPP
-
-#include <vector>
-#include <string>
-
-//
-// Vocab utils
-//
-
-std::vector<llama_token> llama_tokenize(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos);
-
-std::vector<llama_token> llama_tokenize_bpe(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos);
-
-std::string llama_token_to_str(
-        const struct llama_context * ctx,
-                       llama_token   token);
-
-std::string llama_token_to_str_bpe(
-    const struct llama_context * ctx,
-                   llama_token   token);
-
-// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
-#ifdef LLAMA_API_INTERNAL
-
-struct ggml_tensor;
-
-const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
-
-#endif // LLAMA_API_CPP
-
-#endif // LLAMA_API_INTERNAL
-
-#endif // LLAMA_H
diff --git a/llama-util.h b/llama-util.h
deleted file mode 100644
index 75e19c50c4416..0000000000000
--- a/llama-util.h
+++ /dev/null
@@ -1,553 +0,0 @@
-// Internal header to be included only by llama.cpp.
-// Contains wrappers around OS interfaces.
-
-#ifndef LLAMA_UTIL_H
-#define LLAMA_UTIL_H
-
-#include <cstdio>
-#include <cstdint>
-#include <cerrno>
-#include <cstring>
-#include <cstdarg>
-#include <cstdlib>
-#include <climits>
-
-#include <string>
-#include <vector>
-#include <stdexcept>
-
-#ifdef __has_include
-    #if __has_include(<unistd.h>)
-        #include <unistd.h>
-        #if defined(_POSIX_MAPPED_FILES)
-            #include <sys/mman.h>
-        #endif
-        #if defined(_POSIX_MEMLOCK_RANGE)
-            #include <sys/resource.h>
-        #endif
-    #endif
-#endif
-
-#if defined(_WIN32)
-    #define WIN32_LEAN_AND_MEAN
-    #ifndef NOMINMAX
-        #define NOMINMAX
-    #endif
-    #include <windows.h>
-    #include <io.h>
-    #include <stdio.h> // for _fseeki64
-#endif
-
-#define LLAMA_ASSERT(x) \
-    do { \
-        if (!(x)) { \
-            fprintf(stderr, "LLAMA_ASSERT: %s:%d: %s\n", __FILE__, __LINE__, #x); \
-            abort(); \
-        } \
-    } while (0)
-
-#ifdef __GNUC__
-#ifdef __MINGW32__
-__attribute__((format(gnu_printf, 1, 2)))
-#else
-__attribute__((format(printf, 1, 2)))
-#endif
-#endif
-static std::string format(const char * fmt, ...) {
-    va_list ap, ap2;
-    va_start(ap, fmt);
-    va_copy(ap2, ap);
-    int size = vsnprintf(NULL, 0, fmt, ap);
-    LLAMA_ASSERT(size >= 0 && size < INT_MAX);
-    std::vector<char> buf(size + 1);
-    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
-    LLAMA_ASSERT(size2 == size);
-    va_end(ap2);
-    va_end(ap);
-    return std::string(buf.data(), size);
-}
-
-struct llama_file {
-    // use FILE * so we don't have to re-open the file to mmap
-    FILE * fp;
-    size_t size;
-
-    llama_file(const char * fname, const char * mode) {
-        fp = std::fopen(fname, mode);
-        if (fp == NULL) {
-            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
-        }
-        seek(0, SEEK_END);
-        size = tell();
-        seek(0, SEEK_SET);
-    }
-
-    size_t tell() const {
-#ifdef _WIN32
-        __int64 ret = _ftelli64(fp);
-#else
-        long ret = std::ftell(fp);
-#endif
-        LLAMA_ASSERT(ret != -1); // this really shouldn't fail
-        return (size_t) ret;
-    }
-
-    void seek(size_t offset, int whence) {
-#ifdef _WIN32
-        int ret = _fseeki64(fp, (__int64) offset, whence);
-#else
-        int ret = std::fseek(fp, (long) offset, whence);
-#endif
-        LLAMA_ASSERT(ret == 0); // same
-    }
-
-    void read_raw(void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        std::size_t ret = std::fread(ptr, len, 1, fp);
-        if (ferror(fp)) {
-            throw std::runtime_error(format("read error: %s", strerror(errno)));
-        }
-        if (ret != 1) {
-            throw std::runtime_error(std::string("unexpectedly reached end of file"));
-        }
-    }
-
-    std::uint32_t read_u32() {
-        std::uint32_t ret;
-        read_raw(&ret, sizeof(ret));
-        return ret;
-    }
-
-    std::string read_string(std::uint32_t len) {
-        std::vector<char> chars(len);
-        read_raw(chars.data(), len);
-        return std::string(chars.data(), len);
-    }
-
-    void write_raw(const void * ptr, size_t len) const {
-        if (len == 0) {
-            return;
-        }
-        errno = 0;
-        size_t ret = std::fwrite(ptr, len, 1, fp);
-        if (ret != 1) {
-            throw std::runtime_error(format("write error: %s", strerror(errno)));
-        }
-    }
-
-    void write_u32(std::uint32_t val) {
-        write_raw(&val, sizeof(val));
-    }
-
-    ~llama_file() {
-        if (fp) {
-            std::fclose(fp);
-        }
-    }
-};
-
-// llama_context_data
-struct llama_data_context {
-    virtual void write(const void * src, size_t size) = 0;
-    virtual size_t get_size_written() = 0;
-    virtual ~llama_data_context() = default;
-};
-
-struct llama_data_buffer_context : llama_data_context {
-    uint8_t* ptr;
-    size_t size_written = 0;
-
-    llama_data_buffer_context(uint8_t * p) : ptr(p) {}
-
-    void write(const void * src, size_t size) override {
-        memcpy(ptr, src, size);
-        ptr += size;
-        size_written += size;
-    }
-
-    size_t get_size_written() override {
-        return size_written;
-    }
-};
-
-struct llama_data_file_context : llama_data_context {
-    llama_file* file;
-    size_t size_written = 0;
-
-    llama_data_file_context(llama_file * f) : file(f) {}
-
-    void write(const void * src, size_t size) override {
-        file->write_raw(src, size);
-        size_written += size;
-    }
-
-    size_t get_size_written() override {
-        return size_written;
-    }
-};
-
-#if defined(_WIN32)
-static std::string llama_format_win_err(DWORD err) {
-    LPSTR buf;
-    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
-                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
-    if (!size) {
-        return "FormatMessageA failed";
-    }
-    std::string ret(buf, size);
-    LocalFree(buf);
-    return ret;
-}
-#endif
-
-struct llama_mmap {
-    void * addr;
-    size_t size;
-
-    llama_mmap(const llama_mmap &) = delete;
-
-#ifdef _POSIX_MAPPED_FILES
-    static constexpr bool SUPPORTED = true;
-
-    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
-        size = file->size;
-        int fd = fileno(file->fp);
-        int flags = MAP_SHARED;
-        // prefetch/readahead impairs performance on NUMA systems
-        if (numa) { prefetch = 0; }
-#ifdef __linux__
-        if (prefetch >= file->size) { flags |= MAP_POPULATE; }
-#endif
-        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
-        if (addr == MAP_FAILED) {
-            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
-        }
-
-        if (prefetch > 0) {
-            // Advise the kernel to preload the mapped memory
-            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
-                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-        if (numa) {
-            // advise the kernel not to use readahead
-            // (because the next page might not belong on the same node)
-            if (madvise(addr, file->size, MADV_RANDOM)) {
-                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
-                        strerror(errno));
-            }
-        }
-    }
-
-    ~llama_mmap() {
-        munmap(addr, size);
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
-        (void) numa;
-
-        size = file->size;
-
-        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
-
-        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
-        DWORD error = GetLastError();
-
-        if (hMapping == NULL) {
-            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
-        error = GetLastError();
-        CloseHandle(hMapping);
-
-        if (addr == NULL) {
-            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
-        }
-
-        if (prefetch) {
-            // The PrefetchVirtualMemory API is only present on Windows 8 and above, so we
-            // will dynamically load it using GetProcAddress.
-            BOOL (WINAPI *pPrefetchVirtualMemory) (HANDLE, ULONG_PTR, PWIN32_MEMORY_RANGE_ENTRY, ULONG);
-            HMODULE hKernel32;
-
-            // This call is guaranteed to succeed.
-            hKernel32 = GetModuleHandleW(L"kernel32.dll");
-
-            // This call may fail if on a pre-Win8 system.
-            pPrefetchVirtualMemory = reinterpret_cast<decltype(pPrefetchVirtualMemory)> (GetProcAddress(hKernel32, "PrefetchVirtualMemory"));
-
-            if (pPrefetchVirtualMemory) {
-                // Advise the kernel to preload the mapped memory.
-                WIN32_MEMORY_RANGE_ENTRY range;
-                range.VirtualAddress = addr;
-                range.NumberOfBytes = (SIZE_T)size;
-                if (!pPrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
-                    fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
-                            llama_format_win_err(GetLastError()).c_str());
-                }
-            }
-        }
-    }
-
-    ~llama_mmap() {
-        if (!UnmapViewOfFile(addr)) {
-            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    llama_mmap(struct llama_file *, bool prefetch = true, bool numa = false) {
-        (void) prefetch;
-        (void) numa;
-
-        throw std::runtime_error(std::string("mmap not supported"));
-    }
-#endif
-};
-
-// Represents some region of memory being locked using mlock or VirtualLock;
-// will automatically unlock on destruction.
-struct llama_mlock {
-    void * addr = NULL;
-    size_t size = 0;
-    bool failed_already = false;
-
-    llama_mlock() {}
-    llama_mlock(const llama_mlock &) = delete;
-
-    ~llama_mlock() {
-        if (size) {
-            raw_unlock(addr, size);
-        }
-    }
-
-    void init(void * ptr) {
-        LLAMA_ASSERT(addr == NULL && size == 0);
-        addr = ptr;
-    }
-
-    void grow_to(size_t target_size) {
-        LLAMA_ASSERT(addr);
-        if (failed_already) {
-            return;
-        }
-        size_t granularity = lock_granularity();
-        target_size = (target_size + granularity - 1) & ~(granularity - 1);
-        if (target_size > size) {
-            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
-                size = target_size;
-            } else {
-                failed_already = true;
-            }
-        }
-    }
-
-#ifdef _POSIX_MEMLOCK_RANGE
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        return (size_t) sysconf(_SC_PAGESIZE);
-    }
-
-    #ifdef __APPLE__
-        #define MLOCK_SUGGESTION \
-            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
-            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
-    #else
-        #define MLOCK_SUGGESTION \
-            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
-    #endif
-
-    bool raw_lock(const void * addr, size_t size) {
-        if (!mlock(addr, size)) {
-            return true;
-        } else {
-            char* errmsg = std::strerror(errno);
-            bool suggest = (errno == ENOMEM);
-
-            // Check if the resource limit is fine after all
-            struct rlimit lock_limit;
-            if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit))
-                suggest = false;
-            if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size))
-                suggest = false;
-
-            fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
-                    size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
-            return false;
-        }
-    }
-
-    #undef MLOCK_SUGGESTION
-
-    void raw_unlock(void * addr, size_t size) {
-        if (munlock(addr, size)) {
-            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
-        }
-    }
-#elif defined(_WIN32)
-    static constexpr bool SUPPORTED = true;
-
-    size_t lock_granularity() {
-        SYSTEM_INFO si;
-        GetSystemInfo(&si);
-        return (size_t) si.dwPageSize;
-    }
-
-    bool raw_lock(void * ptr, size_t len) {
-        for (int tries = 1; ; tries++) {
-            if (VirtualLock(ptr, len)) {
-                return true;
-            }
-            if (tries == 2) {
-                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
-                    len, size, llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-
-            // It failed but this was only the first try; increase the working
-            // set size and try again.
-            SIZE_T min_ws_size, max_ws_size;
-            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
-                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
-                        llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-            // Per MSDN: "The maximum number of pages that a process can lock
-            // is equal to the number of pages in its minimum working set minus
-            // a small overhead."
-            // Hopefully a megabyte is enough overhead:
-            size_t increment = len + 1048576;
-            // The minimum must be <= the maximum, so we need to increase both:
-            min_ws_size += increment;
-            max_ws_size += increment;
-            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
-                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
-                        llama_format_win_err(GetLastError()).c_str());
-                return false;
-            }
-        }
-    }
-
-    void raw_unlock(void * ptr, size_t len) {
-        if (!VirtualUnlock(ptr, len)) {
-            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
-                    llama_format_win_err(GetLastError()).c_str());
-        }
-    }
-#else
-    static constexpr bool SUPPORTED = false;
-
-    size_t lock_granularity() {
-        return (size_t) 65536;
-    }
-
-    bool raw_lock(const void * addr, size_t len) {
-        fprintf(stderr, "warning: mlock not supported on this system\n");
-        return false;
-    }
-
-    void raw_unlock(const void * addr, size_t len) {}
-#endif
-};
-
-// Replacement for std::vector<uint8_t> that doesn't require zero-initialization.
-struct llama_buffer {
-    uint8_t * addr = NULL;
-    size_t size = 0;
-
-    llama_buffer() = default;
-
-    void resize(size_t len) {
-#ifdef GGML_USE_METAL
-        free(addr);
-        int result = posix_memalign((void **) &addr, getpagesize(), len);
-        if (result == 0) {
-            memset(addr, 0, len);
-        }
-        else {
-            addr = NULL;
-        }
-#else
-        delete[] addr;
-        addr = new uint8_t[len];
-#endif
-        size = len;
-    }
-
-    ~llama_buffer() {
-#ifdef GGML_USE_METAL
-        free(addr);
-#else
-        delete[] addr;
-#endif
-        addr = NULL;
-    }
-
-    // disable copy and move
-    llama_buffer(const llama_buffer&) = delete;
-    llama_buffer(llama_buffer&&) = delete;
-    llama_buffer& operator=(const llama_buffer&) = delete;
-    llama_buffer& operator=(llama_buffer&&) = delete;
-};
-
-#ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
-struct llama_ctx_buffer {
-    uint8_t * addr = NULL;
-    bool is_cuda;
-    size_t size = 0;
-
-    llama_ctx_buffer() = default;
-
-    void resize(size_t size) {
-        free();
-
-        addr = (uint8_t *) ggml_cuda_host_malloc(size);
-        if (addr) {
-            is_cuda = true;
-        }
-        else {
-            // fall back to pageable memory
-            addr = new uint8_t[size];
-            is_cuda = false;
-        }
-        this->size = size;
-    }
-
-    void free() {
-        if (addr) {
-            if (is_cuda) {
-                ggml_cuda_host_free(addr);
-            }
-            else {
-                delete[] addr;
-            }
-        }
-        addr = NULL;
-    }
-
-    ~llama_ctx_buffer() {
-        free();
-    }
-
-    // disable copy and move
-    llama_ctx_buffer(const llama_ctx_buffer&) = delete;
-    llama_ctx_buffer(llama_ctx_buffer&&) = delete;
-    llama_ctx_buffer& operator=(const llama_ctx_buffer&) = delete;
-    llama_ctx_buffer& operator=(llama_ctx_buffer&&) = delete;
-};
-#else
-typedef llama_buffer llama_ctx_buffer;
-#endif
-
-#endif
diff --git a/llama.cpp b/llama.cpp
index fb2207f84bad8..38a2d5ba89f3a 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -6,96 +6,154 @@
 #include <cstdio>
 #endif
 
-#include "llama-util.h"
 #define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include "ggml.h"
+
+#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
+#  include "ggml-alloc.h"
+#  define LLAMA_USE_ALLOCATOR
+#else
+#  define LLAMA_USE_SCRATCH
+#  define LLAMA_MAX_SCRATCH_BUFFERS 16
+#endif
+
 #ifdef GGML_USE_CUBLAS
-#include "ggml-cuda.h"
+#  include "ggml-cuda.h"
 #elif defined(GGML_USE_CLBLAST)
-#include "ggml-opencl.h"
+#  include "ggml-opencl.h"
 #endif
 
 #ifdef GGML_USE_METAL
-#include "ggml-metal.h"
+#  include "ggml-metal.h"
 #endif
 #ifdef GGML_USE_MPI
-#include "ggml-mpi.h"
+#  include "ggml-mpi.h"
 #endif
 #ifdef GGML_USE_K_QUANTS
-#ifndef QK_K
-#ifdef GGML_QKK_64
-#define QK_K 64
-#else
-#define QK_K 256
+#  ifndef QK_K
+#    ifdef GGML_QKK_64
+#      define QK_K 64
+#    else
+#      define QK_K 256
+#    endif
+#  endif
 #endif
+
+#ifdef __has_include
+    #if __has_include(<unistd.h>)
+        #include <unistd.h>
+        #if defined(_POSIX_MAPPED_FILES)
+            #include <sys/mman.h>
+        #endif
+        #if defined(_POSIX_MEMLOCK_RANGE)
+            #include <sys/resource.h>
+        #endif
+    #endif
 #endif
+
+#if defined(_WIN32)
+    #define WIN32_LEAN_AND_MEAN
+    #ifndef NOMINMAX
+        #define NOMINMAX
+    #endif
+    #include <windows.h>
+    #include <io.h>
+    #include <stdio.h> // for _fseeki64
 #endif
 
+#include <algorithm>
 #include <array>
-#include <ctime>
+#include <cassert>
 #include <cinttypes>
+#include <climits>
+#include <cstdarg>
+#include <cstring>
+#include <ctime>
 #include <fstream>
-#include <random>
+#include <initializer_list>
 #include <map>
-#include <unordered_map>
-#include <queue>
-#include <cassert>
-#include <cstring>
-#include <climits>
 #include <memory>
-#include <algorithm>
-#include <initializer_list>
-#include <thread>
-#include <atomic>
 #include <mutex>
-#include <sstream>
 #include <numeric>
+#include <queue>
+#include <random>
+#include <sstream>
+#include <thread>
+#include <unordered_map>
 
 #if defined(_MSC_VER)
 #pragma warning(disable: 4244 4267) // possible loss of data
 #endif
 
-static void llama_log_internal(llama_log_level level, const char* format, ...);
-static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data);
-#define LLAMA_LOG_INFO(...)  llama_log_internal(LLAMA_LOG_LEVEL_INFO , __VA_ARGS__)
-#define LLAMA_LOG_WARN(...)  llama_log_internal(LLAMA_LOG_LEVEL_WARN , __VA_ARGS__)
-#define LLAMA_LOG_ERROR(...) llama_log_internal(LLAMA_LOG_LEVEL_ERROR, __VA_ARGS__)
-
-
-#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
-#include "ggml-alloc.h"
-#define LLAMA_USE_ALLOCATOR
+// tensor names
+#define TN_TOKEN_EMBD  "token_embd.weight"
+#define TN_OUTPUT_NORM "output_norm.weight"
+#define TN_OUTPUT      "output.weight"
+#define TN_ATTN_NORM   "blk.%d.attn_norm.weight"
+#define TN_ATTN_Q      "blk.%d.attn_q.weight"
+#define TN_ATTN_K      "blk.%d.attn_k.weight"
+#define TN_ATTN_V      "blk.%d.attn_v.weight"
+#define TN_ATTN_OUTPUT "blk.%d.attn_output.weight"
+#define TN_FFN_NORM    "blk.%d.ffn_norm.weight"
+#define TN_FFN_GATE    "blk.%d.ffn_gate.weight"
+#define TN_FFN_DOWN    "blk.%d.ffn_down.weight"
+#define TN_FFN_UP      "blk.%d.ffn_up.weight"
+
+#ifdef __GNUC__
+#ifdef __MINGW32__
+#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(gnu_printf, __VA_ARGS__)))
+#else
+#define LLAMA_ATTRIBUTE_FORMAT(...) __attribute__((format(printf, __VA_ARGS__)))
+#endif
 #else
-#define LLAMA_USE_SCRATCH
-#define LLAMA_MAX_SCRATCH_BUFFERS 16
+#define LLAMA_ATTRIBUTE_FORMAT(...)
 #endif
 
-#define UNUSED GGML_UNUSED
+//
+// logging
+//
+LLAMA_ATTRIBUTE_FORMAT(2, 3)
+static void llama_log_internal        (llama_log_level level, const char* format, ...);
+static void llama_log_callback_default(llama_log_level level, const char * text, void * user_data);
 
-// available llama models
-enum e_model {
-    MODEL_UNKNOWN,
-    MODEL_3B,
-    MODEL_7B,
-    MODEL_13B,
-    MODEL_30B,
-    MODEL_65B,
-    MODEL_70B,
-};
+#define LLAMA_LOG_INFO(...)  llama_log_internal(LLAMA_LOG_LEVEL_INFO , __VA_ARGS__)
+#define LLAMA_LOG_WARN(...)  llama_log_internal(LLAMA_LOG_LEVEL_WARN , __VA_ARGS__)
+#define LLAMA_LOG_ERROR(...) llama_log_internal(LLAMA_LOG_LEVEL_ERROR, __VA_ARGS__)
 
-static const size_t kB = 1024;
-static const size_t MB = 1024*1024;
+//
+// helpers
+//
 
-// computed for n_ctx == 2048
-// TODO: dynamically determine these sizes
-//       needs modifications in ggml
+template<typename T>
+static std::string to_string(const T & val) {
+    std::stringstream ss;
+    ss << val;
+    return ss.str();
+}
 
-typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+static void zeros(std::ofstream & file, size_t n) {
+    char zero = 0;
+    for (size_t i = 0; i < n; ++i) {
+        file.write(&zero, 1);
+    }
+}
 
-void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
-    (void) tensor;
+LLAMA_ATTRIBUTE_FORMAT(1, 2)
+static std::string format(const char * fmt, ...) {
+    va_list ap;
+    va_list ap2;
+    va_start(ap, fmt);
+    va_copy(ap2, ap);
+    int size = vsnprintf(NULL, 0, fmt, ap);
+    GGML_ASSERT(size >= 0 && size < INT_MAX); // NOLINT
+    std::vector<char> buf(size + 1);
+    int size2 = vsnprintf(buf.data(), size + 1, fmt, ap2);
+    GGML_ASSERT(size2 == size);
+    va_end(ap2);
+    va_end(ap);
+    return std::string(buf.data(), size);
 }
 
 //
@@ -113,10 +171,439 @@ static void ggml_graph_compute_helper(std::vector<uint8_t> & buf, ggml_cgraph *
     ggml_graph_compute(graph, &plan);
 }
 
+//
+// llama helpers
+//
+
+#ifdef GGML_USE_CUBLAS
+#   define llama_host_malloc(n)  ggml_cuda_host_malloc(n)
+#   define llama_host_free(data) ggml_cuda_host_free(data)
+#elif GGML_USE_METAL
+#   define llama_host_malloc(n)  ggml_metal_host_malloc(n)
+#   define llama_host_free(data) ggml_metal_host_free(data)
+#else
+#   define llama_host_malloc(n)  malloc(n)
+#   define llama_host_free(data) free(data)
+#endif
+
+#if defined(_WIN32)
+static std::string llama_format_win_err(DWORD err) {
+    LPSTR buf;
+    size_t size = FormatMessageA(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS,
+                                 NULL, err, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), (LPSTR)&buf, 0, NULL);
+    if (!size) {
+        return "FormatMessageA failed";
+    }
+    std::string ret(buf, size);
+    LocalFree(buf);
+    return ret;
+}
+#endif
+
+struct llama_buffer {
+    void * data = NULL;
+    size_t size = 0;
+
+    // fallback to malloc / free
+    // useful in cases where CUDA can try to allocate PINNED memory
+    bool fallback = false;
+
+    void resize(size_t n) {
+        llama_host_free(data);
+
+        data = llama_host_malloc(n);
+        if (!data) {
+            fallback = true;
+            data = malloc(n);
+        } else {
+            fallback = false;
+        }
+
+        GGML_ASSERT(data);
+        size = n;
+    }
+
+    ~llama_buffer() {
+        if (data) {
+            if (fallback) { // NOLINT
+                free(data);
+            } else {
+                llama_host_free(data);
+            }
+        }
+
+        data = NULL;
+    }
+};
+
+struct llama_file {
+    // use FILE * so we don't have to re-open the file to mmap
+    FILE * fp;
+    size_t size;
+
+    llama_file(const char * fname, const char * mode) {
+        fp = std::fopen(fname, mode);
+        if (fp == NULL) {
+            throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
+        }
+        seek(0, SEEK_END);
+        size = tell();
+        seek(0, SEEK_SET);
+    }
+
+    size_t tell() const {
+#ifdef _WIN32
+        __int64 ret = _ftelli64(fp);
+#else
+        long ret = std::ftell(fp);
+#endif
+        GGML_ASSERT(ret != -1); // this really shouldn't fail
+        return (size_t) ret;
+    }
+
+    void seek(size_t offset, int whence) const {
+#ifdef _WIN32
+        int ret = _fseeki64(fp, (__int64) offset, whence);
+#else
+        int ret = std::fseek(fp, (long) offset, whence);
+#endif
+        GGML_ASSERT(ret == 0); // same
+    }
+
+    void read_raw(void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        std::size_t ret = std::fread(ptr, len, 1, fp);
+        if (ferror(fp)) {
+            throw std::runtime_error(format("read error: %s", strerror(errno)));
+        }
+        if (ret != 1) {
+            throw std::runtime_error(std::string("unexpectedly reached end of file"));
+        }
+    }
+
+    uint32_t read_u32() {
+        uint32_t ret;
+        read_raw(&ret, sizeof(ret));
+        return ret;
+    }
+
+    void write_raw(const void * ptr, size_t len) const {
+        if (len == 0) {
+            return;
+        }
+        errno = 0;
+        size_t ret = std::fwrite(ptr, len, 1, fp);
+        if (ret != 1) {
+            throw std::runtime_error(format("write error: %s", strerror(errno)));
+        }
+    }
+
+    void write_u32(std::uint32_t val) const {
+        write_raw(&val, sizeof(val));
+    }
+
+    ~llama_file() {
+        if (fp) {
+            std::fclose(fp);
+        }
+    }
+};
+
+struct llama_mmap {
+    void * addr;
+    size_t size;
+
+    llama_mmap(const llama_mmap &) = delete;
+
+#ifdef _POSIX_MAPPED_FILES
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, size_t prefetch = (size_t) -1 /* -1 = max value */, bool numa = false) {
+        size = file->size;
+        int fd = fileno(file->fp);
+        int flags = MAP_SHARED;
+        // prefetch/readahead impairs performance on NUMA systems
+        if (numa) { prefetch = 0; }
+#ifdef __linux__
+        if (prefetch) { flags |= MAP_POPULATE; }
+#endif
+        addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
+        if (addr == MAP_FAILED) {
+            throw std::runtime_error(format("mmap failed: %s", strerror(errno)));
+        }
+
+        if (prefetch > 0) {
+            // Advise the kernel to preload the mapped memory
+            if (madvise(addr, std::min(file->size, prefetch), MADV_WILLNEED)) {
+                fprintf(stderr, "warning: madvise(.., MADV_WILLNEED) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+        if (numa) {
+            // advise the kernel not to use readahead
+            // (because the next page might not belong on the same node)
+            if (madvise(addr, file->size, MADV_RANDOM)) {
+                fprintf(stderr, "warning: madvise(.., MADV_RANDOM) failed: %s\n",
+                        strerror(errno));
+            }
+        }
+    }
+
+    ~llama_mmap() {
+        munmap(addr, size);
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) numa;
+
+        size = file->size;
+
+        HANDLE hFile = (HANDLE) _get_osfhandle(_fileno(file->fp));
+
+        HANDLE hMapping = CreateFileMappingA(hFile, NULL, PAGE_READONLY, 0, 0, NULL);
+        DWORD error = GetLastError();
+
+        if (hMapping == NULL) {
+            throw std::runtime_error(format("CreateFileMappingA failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        addr = MapViewOfFile(hMapping, FILE_MAP_READ, 0, 0, 0);
+        error = GetLastError();
+        CloseHandle(hMapping);
+
+        if (addr == NULL) {
+            throw std::runtime_error(format("MapViewOfFile failed: %s", llama_format_win_err(error).c_str()));
+        }
+
+        #if _WIN32_WINNT >= _WIN32_WINNT_WIN8
+        if (prefetch) {
+            // Advise the kernel to preload the mapped memory
+            WIN32_MEMORY_RANGE_ENTRY range;
+            range.VirtualAddress = addr;
+            range.NumberOfBytes = (SIZE_T)size;
+            if (!PrefetchVirtualMemory(GetCurrentProcess(), 1, &range, 0)) {
+                fprintf(stderr, "warning: PrefetchVirtualMemory failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+            }
+        }
+        #else
+        #pragma message("warning: You are building for pre-Windows 8; prefetch not supported")
+        #endif // _WIN32_WINNT >= _WIN32_WINNT_WIN8
+    }
+
+    ~llama_mmap() {
+        if (!UnmapViewOfFile(addr)) {
+            fprintf(stderr, "warning: UnmapViewOfFile failed: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    llama_mmap(struct llama_file * file, bool prefetch = true, bool numa = false) {
+        (void) file;
+        (void) prefetch;
+        (void) numa;
+
+        throw std::runtime_error(std::string("mmap not supported"));
+    }
+#endif
+};
+
+// Represents some region of memory being locked using mlock or VirtualLock;
+// will automatically unlock on destruction.
+struct llama_mlock {
+    void * addr = NULL;
+    size_t size = 0;
+
+    bool failed_already = false;
+
+    llama_mlock() {}
+    llama_mlock(const llama_mlock &) = delete;
+
+    ~llama_mlock() {
+        if (size) {
+            raw_unlock(addr, size);
+        }
+    }
+
+    void init(void * ptr) {
+        GGML_ASSERT(addr == NULL && size == 0); // NOLINT
+        addr = ptr;
+    }
+
+    void grow_to(size_t target_size) {
+        GGML_ASSERT(addr);
+        if (failed_already) {
+            return;
+        }
+        size_t granularity = lock_granularity();
+        target_size = (target_size + granularity - 1) & ~(granularity - 1);
+        if (target_size > size) {
+            if (raw_lock((uint8_t *) addr + size, target_size - size)) {
+                size = target_size;
+            } else {
+                failed_already = true;
+            }
+        }
+    }
+
+#ifdef _POSIX_MEMLOCK_RANGE
+    static constexpr bool SUPPORTED = true;
+
+    static size_t lock_granularity() {
+        return (size_t) sysconf(_SC_PAGESIZE);
+    }
+
+    #ifdef __APPLE__
+        #define MLOCK_SUGGESTION \
+            "Try increasing the sysctl values 'vm.user_wire_limit' and 'vm.global_user_wire_limit' and/or " \
+            "decreasing 'vm.global_no_user_wire_amount'.  Also try increasing RLIMIT_MLOCK (ulimit -l).\n"
+    #else
+        #define MLOCK_SUGGESTION \
+            "Try increasing RLIMIT_MLOCK ('ulimit -l' as root).\n"
+    #endif
+
+    bool raw_lock(const void * addr, size_t size) const {
+        if (!mlock(addr, size)) {
+            return true;
+        }
+
+        char* errmsg = std::strerror(errno);
+        bool suggest = (errno == ENOMEM);
+
+        // Check if the resource limit is fine after all
+        struct rlimit lock_limit;
+        if (suggest && getrlimit(RLIMIT_MEMLOCK, &lock_limit)) {
+            suggest = false;
+        }
+        if (suggest && (lock_limit.rlim_max > lock_limit.rlim_cur + size)) {
+            suggest = false;
+        }
+
+        fprintf(stderr, "warning: failed to mlock %zu-byte buffer (after previously locking %zu bytes): %s\n%s",
+                size, this->size, errmsg, suggest ? MLOCK_SUGGESTION : "");
+        return false;
+    }
+
+    #undef MLOCK_SUGGESTION
+
+    static void raw_unlock(void * addr, size_t size) {
+        if (munlock(addr, size)) {
+            fprintf(stderr, "warning: failed to munlock buffer: %s\n", std::strerror(errno));
+        }
+    }
+#elif defined(_WIN32)
+    static constexpr bool SUPPORTED = true;
+
+    static size_t lock_granularity() {
+        SYSTEM_INFO si;
+        GetSystemInfo(&si);
+        return (size_t) si.dwPageSize;
+    }
+
+    bool raw_lock(void * ptr, size_t len) const {
+        for (int tries = 1; ; tries++) {
+            if (VirtualLock(ptr, len)) {
+                return true;
+            }
+            if (tries == 2) {
+                fprintf(stderr, "warning: failed to VirtualLock %zu-byte buffer (after previously locking %zu bytes): %s\n",
+                    len, size, llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+
+            // It failed but this was only the first try; increase the working
+            // set size and try again.
+            SIZE_T min_ws_size, max_ws_size;
+            if (!GetProcessWorkingSetSize(GetCurrentProcess(), &min_ws_size, &max_ws_size)) {
+                fprintf(stderr, "warning: GetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+            // Per MSDN: "The maximum number of pages that a process can lock
+            // is equal to the number of pages in its minimum working set minus
+            // a small overhead."
+            // Hopefully a megabyte is enough overhead:
+            size_t increment = len + 1048576;
+            // The minimum must be <= the maximum, so we need to increase both:
+            min_ws_size += increment;
+            max_ws_size += increment;
+            if (!SetProcessWorkingSetSize(GetCurrentProcess(), min_ws_size, max_ws_size)) {
+                fprintf(stderr, "warning: SetProcessWorkingSetSize failed: %s\n",
+                        llama_format_win_err(GetLastError()).c_str());
+                return false;
+            }
+        }
+    }
+
+    static void raw_unlock(void * ptr, size_t len) {
+        if (!VirtualUnlock(ptr, len)) {
+            fprintf(stderr, "warning: failed to VirtualUnlock buffer: %s\n",
+                    llama_format_win_err(GetLastError()).c_str());
+        }
+    }
+#else
+    static constexpr bool SUPPORTED = false;
+
+    static size_t lock_granularity() {
+        return (size_t) 65536;
+    }
+
+    bool raw_lock(const void * addr, size_t len) const {
+        fprintf(stderr, "warning: mlock not supported on this system\n");
+        return false;
+    }
+
+    static void raw_unlock(const void * addr, size_t len) {}
+#endif
+};
+
+typedef void (*offload_func_t)(struct ggml_tensor * tensor);
+
+void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
+    (void) tensor;
+}
+
+//
+// globals
+//
+
+struct llama_state {
+    // We save the log callback globally
+    llama_log_callback log_callback = llama_log_callback_default;
+    void * log_callback_user_data = nullptr;
+};
+
+static llama_state g_state;
+
 //
 // memory sizes (calculated for n_batch == 512)
 //
 
+// computed for n_ctx == 2048
+// TODO: dynamically determine these sizes
+//       needs modifications in ggml
+
+// available llama models
+enum e_model {
+    MODEL_UNKNOWN,
+    MODEL_3B,
+    MODEL_7B,
+    MODEL_13B,
+    MODEL_30B,
+    MODEL_65B,
+    MODEL_70B,
+};
+
+static const size_t kB = 1024;
+static const size_t MB = 1024*1024;
+
 static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0(int n_ctx)
 {
     static std::map<e_model, size_t> k_sizes = {
@@ -190,18 +677,15 @@ static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
 // default hparams (LLaMA 7B)
 struct llama_hparams {
     uint32_t n_vocab   = 32000;
-    uint32_t n_ctx     = 512;   // this is provided as user input?
+    uint32_t n_ctx     = 512;
     uint32_t n_embd    = 4096;
-    uint32_t n_mult    = 256;
     uint32_t n_head    = 32;
     uint32_t n_head_kv = 32;
     uint32_t n_layer   = 32;
     uint32_t n_rot     = 64;
+    uint32_t n_ff      = 11008;
 
-    // LLaMAv2
-    // TODO: load from model data hparams
-    float f_ffn_mult = 1.0f;
-    float f_rms_norm_eps = LLAMA_DEFAULT_RMS_EPS;
+    float f_norm_rms_eps = 1e-5;
 
     float rope_freq_base  = 10000.0f;
     float rope_freq_scale = 1.0f;
@@ -259,7 +743,7 @@ struct llama_kv_cache {
 
     struct ggml_context * ctx = NULL;
 
-    llama_ctx_buffer buf;
+    llama_buffer buf;
 
     int n; // number of tokens currently in the cache
 
@@ -276,6 +760,11 @@ struct llama_kv_cache {
 };
 
 struct llama_vocab {
+    // TODO:
+    // - add a vector of merges
+    // - add members for bos/eos/pad/sep tokens
+    //   so that we can pass it to different types of tokenizers with a common interface
+
     using id    = int32_t;
     using token = std::string;
 
@@ -292,6 +781,7 @@ struct llama_model {
     e_model type = MODEL_UNKNOWN;
 
     llama_hparams hparams;
+    llama_vocab   vocab;
 
     struct ggml_tensor * tok_embeddings;
 
@@ -305,7 +795,7 @@ struct llama_model {
     struct ggml_context * ctx = NULL;
 
     // the model memory buffer
-    llama_ctx_buffer buf;
+    llama_buffer buf;
 
     // model memory mapped file
     std::unique_ptr<llama_mmap> mapping;
@@ -320,8 +810,6 @@ struct llama_model {
     int64_t t_load_us = 0;
     int64_t t_start_us = 0;
 
-    llama_vocab vocab;
-
     ~llama_model() {
         if (ctx) {
             ggml_free(ctx);
@@ -394,15 +882,16 @@ struct llama_context {
 
     // memory buffers used to evaluate the model
     // TODO: move in llama_state
-    llama_ctx_buffer buf_compute;
+    llama_buffer buf_compute;
 
 #ifdef LLAMA_USE_ALLOCATOR
-    llama_ctx_buffer buf_alloc;
+    llama_buffer buf_alloc;
     ggml_allocr * alloc = NULL;
 #endif
 
 #ifdef LLAMA_USE_SCRATCH
-    llama_ctx_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+    llama_buffer buf_scratch[LLAMA_MAX_SCRATCH_BUFFERS];
+
     int    buf_last = 0;
     size_t buf_max_size[LLAMA_MAX_SCRATCH_BUFFERS] = { 0 };
 #endif
@@ -415,7 +904,7 @@ struct llama_context {
     ggml_mpi_context * ctx_mpi = NULL;
 #endif
 
-    void use_buf(struct ggml_context * ctx, int i) {
+    void use_buf(struct ggml_context * ctx, int i) { // NOLINT
 #if defined(LLAMA_USE_SCRATCH)
         size_t last_size = 0;
 
@@ -423,7 +912,7 @@ struct llama_context {
             last_size = ggml_set_scratch(ctx, { 0, 0, nullptr, });
         } else {
             auto & buf = buf_scratch[i];
-            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.addr, });
+            last_size = ggml_set_scratch(ctx, { 0, buf.size, buf.data, });
         }
 
         if (buf_last >= 0) {
@@ -437,7 +926,7 @@ struct llama_context {
 #endif
     }
 
-    size_t get_buf_max_mem(int i) const {
+    size_t get_buf_max_mem(int i) { // NOLINT
 #if defined(LLAMA_USE_SCRATCH)
         return buf_max_size[i];
 #else
@@ -447,377 +936,294 @@ struct llama_context {
     }
 };
 
-struct llama_state {
-    // We save the log callback globally
-    llama_log_callback log_callback = llama_log_callback_default;
-    void * log_callback_user_data = nullptr;
-};
-// global state
-static llama_state g_state;
+//
+// kv cache helpers
+//
+
+static bool llama_kv_cache_init(
+        const struct llama_hparams & hparams,
+             struct llama_kv_cache & cache,
+                         ggml_type   wtype,
+                               int   n_ctx,
+                               int   n_gpu_layers) {
+    const int n_embd  = hparams.n_embd_gqa();
+    const int n_layer = hparams.n_layer;
 
-template <typename T>
-static T checked_mul(T a, T b) {
-    T ret = a * b;
-    if (a != 0 && ret / a != b) {
-        throw std::runtime_error(format("overflow multiplying %llu * %llu",
-                     (unsigned long long) a, (unsigned long long) b));
+    const int64_t n_mem      = n_layer*n_ctx;
+    const int64_t n_elements = n_embd*n_mem;
+
+    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
+    cache.n = 0;
+
+    struct ggml_init_params params;
+    params.mem_size   = cache.buf.size;
+    params.mem_buffer = cache.buf.data;
+    params.no_alloc   = false;
+
+    cache.ctx = ggml_init(params);
+
+    if (!cache.ctx) {
+        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
+        return false;
     }
-    return ret;
+
+    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
+    ggml_set_name(cache.k, "cache_k");
+    ggml_set_name(cache.v, "cache_v");
+
+    (void) n_gpu_layers;
+#ifdef GGML_USE_CUBLAS
+    if (n_gpu_layers > n_layer + 1) {
+        ggml_cuda_assign_buffers_no_scratch(cache.v);
+    }
+    if (n_gpu_layers > n_layer + 2) {
+        ggml_cuda_assign_buffers_no_scratch(cache.k);
+    }
+#endif // GGML_USE_CUBLAS
+
+    return true;
 }
 
-static size_t checked_div(size_t a, size_t b) {
-    if (b == 0 || a % b != 0) {
-        throw std::runtime_error(format("error dividing %zu / %zu", a, b));
+//
+// model loading and saving
+//
+
+enum llama_file_version {
+    GGUF_FILE_VERSION_V1 = 1,
+};
+
+static const char * llama_file_version_name(llama_file_version version) {
+    switch (version) {
+        case GGUF_FILE_VERSION_V1: return "GGUF V1 (latest)";
     }
-    return a / b;
+
+    return "unknown";
 }
 
 static std::string llama_format_tensor_shape(const std::vector<uint32_t> & ne) {
     char buf[256];
     snprintf(buf, sizeof(buf), "%5u", ne.at(0));
     for (size_t i = 1; i < ne.size(); i++) {
-        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " x %5u", ne.at(i));
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5u", ne.at(i));
     }
     return buf;
 }
 
-static size_t llama_calc_tensor_size(const std::vector<uint32_t> & ne, enum ggml_type type) {
-    size_t size = ggml_type_size(type);
-    for (uint32_t dim : ne) {
-        size = checked_mul<size_t>(size, dim);
+static std::string llama_format_tensor_shape(const struct ggml_tensor * t) {
+    char buf[256];
+    snprintf(buf, sizeof(buf), "%5" PRId64, t->ne[0]);
+    for (int i = 1; i < GGML_MAX_DIMS; i++) {
+        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), ", %5" PRId64, t->ne[i]);
     }
-    return size / ggml_blck_size(type);
+    return buf;
 }
 
-struct llama_load_tensor {
-    std::string name;
-    enum ggml_type type = GGML_TYPE_F32;
-    std::vector<uint32_t> ne;
-    size_t file_off;
-    size_t size;
-    struct ggml_tensor * ggml_tensor = NULL;
-    uint8_t * data;
-};
-
-struct llama_load_tensors_map {
-    // tensors is kept in a separate vector to preserve file order
-    std::vector<llama_load_tensor> tensors;
-    std::unordered_map<std::string, size_t> name_to_idx;
-};
+struct llama_model_loader {
+    int n_kv      = 0;
+    int n_tensors = 0;
+    int n_created = 0;
 
-enum llama_file_version {
-    LLAMA_FILE_VERSION_GGML,
-    LLAMA_FILE_VERSION_GGMF_V1, // added version field and scores in vocab
-    LLAMA_FILE_VERSION_GGJT_V1, // added padding
-    LLAMA_FILE_VERSION_GGJT_V2, // changed quantization format
-    LLAMA_FILE_VERSION_GGJT_V3, // changed Q4 and Q8 quantization format
-};
+    bool use_mmap = false;
 
-struct llama_file_loader {
     llama_file file;
     llama_file_version file_version;
-    llama_hparams hparams;
-    llama_vocab vocab;
-
-    llama_file_loader(const char * fname, llama_load_tensors_map & tensors_map)
-        : file(fname, "rb") {
-        LLAMA_LOG_INFO("llama.cpp: loading model from %s\n", fname);
-        read_magic();
-        read_hparams();
-        read_vocab();
-        read_tensor_metadata(tensors_map);
-    }
-    void read_magic() {
-        uint32_t magic = file.read_u32();
-
-        if (magic == LLAMA_FILE_MAGIC_GGML) {
-            file_version = LLAMA_FILE_VERSION_GGML;
-            return;
-        }
 
-        uint32_t version = file.read_u32();
+    std::unique_ptr<llama_mmap> mapping;
 
-        switch (magic) {
-            case LLAMA_FILE_MAGIC_GGMF:
-                switch (version) {
-                    case 1: file_version = LLAMA_FILE_VERSION_GGMF_V1; return;
-                }
-                break;
-            case LLAMA_FILE_MAGIC_GGJT:
-                switch (version) {
-                    case 1: file_version = LLAMA_FILE_VERSION_GGJT_V1; return;
-                    case 2: file_version = LLAMA_FILE_VERSION_GGJT_V2; return;
-                    case 3: file_version = LLAMA_FILE_VERSION_GGJT_V3; return;
-                }
-        }
+    struct gguf_context * ctx_gguf = NULL;
+    struct ggml_context * ctx_meta = NULL;
 
-        throw std::runtime_error(format("unknown (magic, version) combination: %08x, %08x; is this really a GGML file?",
-                     magic, version));
-    }
-    void read_hparams() {
-        hparams.n_vocab = file.read_u32();
-        hparams.n_embd  = file.read_u32();
-        hparams.n_mult  = file.read_u32();
-        hparams.n_head  = file.read_u32();
-        hparams.n_layer = file.read_u32();
-        hparams.n_rot   = file.read_u32();
-        hparams.ftype   = (enum llama_ftype) file.read_u32();
+    llama_model_loader(const std::string & fname, bool use_mmap) : file(fname.c_str(), "rb") {
+        struct gguf_init_params params = {
+            /*.no_alloc = */ true,
+            /*.ctx      = */ &ctx_meta,
+        };
 
-        // LLaMAv2
-        // TODO: read from header
-        hparams.n_head_kv = hparams.n_head;
-    }
-    void read_vocab() {
-        vocab.id_to_token.resize(hparams.n_vocab);
+        ctx_gguf = gguf_init_from_file(fname.c_str(), params);
 
-        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
-            uint32_t len = file.read_u32();
-            std::string word = file.read_string(len);
+        n_kv      = gguf_get_n_kv(ctx_gguf);
+        n_tensors = gguf_get_n_tensors(ctx_gguf);
 
-            float score = 0.0f;
-            file.read_raw(&score, sizeof(score));
+        file_version = (enum llama_file_version) gguf_get_version(ctx_gguf);
 
-            GGML_ASSERT(vocab.token_to_id.find(word) == vocab.token_to_id.end());
-            vocab.token_to_id[word] = i;
+        // print meta data
+        // TODO: make optional
+        {
+            LLAMA_LOG_INFO("%s: loaded meta data with %d key-value paris and %d tensors from %s (version %s)\n",
+                    __func__, n_kv, n_tensors, fname.c_str(), llama_file_version_name(file_version));
 
-            auto & tok_score = vocab.id_to_token[i];
-            tok_score.tok = std::move(word);
-            tok_score.score = score;
-        }
-    }
-    void read_tensor_metadata(llama_load_tensors_map & tensors_map) {
-        while (file.tell() < file.size) {
-            llama_load_tensor tensor;
-            uint32_t n_dims = file.read_u32();
-            uint32_t name_len = file.read_u32();
-            tensor.type = (enum ggml_type) file.read_u32();
-            tensor.ne.resize(n_dims);
-            file.read_raw(tensor.ne.data(), sizeof(tensor.ne[0]) * n_dims);
-            std::string name = file.read_string(name_len);
-            if (n_dims < 1 || n_dims > 2) {
-                throw std::runtime_error(format("llama.cpp: tensor '%s' should not be %u-dimensional", name.c_str(), n_dims));
-            }
-            switch (tensor.type) {
-                case GGML_TYPE_F32:
-                case GGML_TYPE_F16:
-                case GGML_TYPE_Q4_0:
-                case GGML_TYPE_Q4_1:
-                case GGML_TYPE_Q5_0:
-                case GGML_TYPE_Q5_1:
-                case GGML_TYPE_Q8_0:
-                case GGML_TYPE_Q2_K:
-                case GGML_TYPE_Q3_K:
-                case GGML_TYPE_Q4_K:
-                case GGML_TYPE_Q5_K:
-                case GGML_TYPE_Q6_K:
-                    break;
-                default: {
-                    throw std::runtime_error(format("unrecognized tensor type %u\n", tensor.type));
-                }
-            }
+            for (int i = 0; i < n_tensors; i++) {
+                const char * name = gguf_get_tensor_name(ctx_gguf, i);
+                struct ggml_tensor * meta = ggml_get_tensor(ctx_meta, name);
 
-            // skip to the next multiple of 32 bytes
-            if (file_version >= LLAMA_FILE_VERSION_GGJT_V1) {
-                file.seek(-static_cast<ptrdiff_t>(file.tell()) & 31, SEEK_CUR);
+                LLAMA_LOG_INFO("%s: - tensor %4d: %32s %-8s [ %s ]\n", __func__, i, name, ggml_type_name(meta->type), llama_format_tensor_shape(meta).c_str());
             }
 
-            tensor.file_off = file.tell();
-            tensor.name = name;
-            tensor.size = llama_calc_tensor_size(tensor.ne, tensor.type);
-            file.seek(tensor.size, SEEK_CUR);
+            for (int i = 0; i < n_kv; i++) {
+                const char * name         = gguf_get_key(ctx_gguf, i);
+                const enum gguf_type type = gguf_get_kv_type(ctx_gguf, i);
 
-            tensors_map.tensors.push_back(tensor);
-            tensors_map.name_to_idx[name] = tensors_map.tensors.size() - 1;
+                LLAMA_LOG_INFO("%s: - kv %3d: %42s %-8s\n", __func__, i, name, gguf_type_name(type));
+            }
         }
-    }
-};
 
-struct llama_file_saver {
-    llama_file file;
-    llama_file_loader * any_file_loader;
-    llama_file_saver(const char * fname, llama_file_loader * any_file_loader, enum llama_ftype new_ftype)
-        : file(fname, "wb"), any_file_loader(any_file_loader) {
-        LLAMA_LOG_INFO("llama.cpp: saving model to %s\n", fname);
-        write_magic();
-        write_hparams(new_ftype);
-        write_vocab();
-    }
-    void write_magic() {
-        file.write_u32(LLAMA_FILE_MAGIC);   // magic
-        file.write_u32(LLAMA_FILE_VERSION); // version
-    }
-    void write_hparams(enum llama_ftype new_ftype) {
-        const llama_hparams & hparams = any_file_loader->hparams;
-        file.write_u32(hparams.n_vocab);
-        file.write_u32(hparams.n_embd);
-        file.write_u32(hparams.n_mult);
-        file.write_u32(hparams.n_head);
-        file.write_u32(hparams.n_layer);
-        file.write_u32(hparams.n_rot);
-        file.write_u32(new_ftype);
-    }
-    void write_vocab() {
-        if (any_file_loader->file_version == LLAMA_FILE_VERSION_GGML) {
-            LLAMA_LOG_WARN("llama.cpp: WARNING: input is an old file that doesn't have scores; will add dummy scores\n");
-        }
-        uint32_t n_vocab = any_file_loader->hparams.n_vocab;
-        for (uint32_t i = 0; i < n_vocab; i++) {
-            const auto & token_score = any_file_loader->vocab.id_to_token.at(i);
-            file.write_u32((uint32_t) token_score.tok.size());
-            file.write_raw(token_score.tok.data(), token_score.tok.size());
-            file.write_raw(&token_score.score, sizeof(token_score.score));
-        }
-    }
-    void write_tensor(llama_load_tensor & tensor, enum ggml_type new_type, const void * new_data, size_t new_size) {
-        switch (new_type) {
-            case GGML_TYPE_F32:
-            case GGML_TYPE_F16:
-            case GGML_TYPE_Q4_0:
-            case GGML_TYPE_Q4_1:
-            case GGML_TYPE_Q5_0:
-            case GGML_TYPE_Q5_1:
-            case GGML_TYPE_Q8_0:
-            case GGML_TYPE_Q2_K:
-            case GGML_TYPE_Q3_K:
-            case GGML_TYPE_Q4_K:
-            case GGML_TYPE_Q5_K:
-            case GGML_TYPE_Q6_K:
-                break;
-            default: LLAMA_ASSERT(false);
-        }
-        file.write_u32((uint32_t) tensor.ne.size());
-        file.write_u32((uint32_t) tensor.name.size());
-        file.write_u32(new_type);
-        file.write_raw(tensor.ne.data(), sizeof(tensor.ne[0]) * tensor.ne.size());
-        file.write_raw(tensor.name.data(), tensor.name.size());
-        file.seek(-static_cast<ptrdiff_t>(file.tell()) & 31, SEEK_CUR);
-        LLAMA_ASSERT(new_size == llama_calc_tensor_size(tensor.ne, new_type));
-        file.write_raw(new_data, new_size);
-    }
-};
-
-struct llama_model_loader {
-    std::unique_ptr<llama_file_loader> file_loader;
-    llama_load_tensors_map tensors_map;
-    bool use_mmap;
-    size_t num_ggml_tensors_created = 0;
-    struct ggml_context * ggml_ctx = NULL;
-    std::unique_ptr<llama_mmap> mapping;
-
-    llama_model_loader(const std::string & fname_base, bool use_mmap) {
-        file_loader = std::unique_ptr<llama_file_loader>(new llama_file_loader(fname_base.c_str(), tensors_map));
         if (!llama_mmap::SUPPORTED) {
+            LLAMA_LOG_WARN("%s: mmap is not supported on this platform\n", __func__);
             use_mmap = false;
         }
+
         this->use_mmap = use_mmap;
     }
 
-    void calc_sizes(size_t * ctx_size_p, size_t * mmapped_size_p) const {
-        *ctx_size_p = *mmapped_size_p = 0;
-        for (const llama_load_tensor & lt : tensors_map.tensors) {
-            *ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
-            *(use_mmap ? mmapped_size_p : ctx_size_p) += lt.size + 16;
-        }
+    const char * get_tensor_name(int i) const {
+        return gguf_get_tensor_name(ctx_gguf, i);
     }
 
-    struct ggml_tensor * get_tensor(const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
-        auto it = tensors_map.name_to_idx.find(name);
-        if (it == tensors_map.name_to_idx.end()) {
-            throw std::runtime_error(std::runtime_error(format("llama.cpp: tensor '%s' is missing from model", name.c_str())));
-        }
-        llama_load_tensor & lt = tensors_map.tensors.at(it->second);
-        if (lt.ne != ne) {
-            throw std::runtime_error(format("llama.cpp: tensor '%s' has wrong shape; expected %s, got %s",
-                         name.c_str(), llama_format_tensor_shape(ne).c_str(), llama_format_tensor_shape(lt.ne).c_str()));
-        }
+    struct ggml_tensor * get_tensor_meta(int i) const {
+        return ggml_get_tensor(ctx_meta, get_tensor_name(i));
+    }
 
-        return get_tensor_for(lt, backend);
+    void calc_sizes(size_t & ctx_size_p, size_t & mmapped_size_p) const {
+        ctx_size_p     = 0;
+        mmapped_size_p = 0;
+
+        for (int i = 0; i < n_tensors; i++) {
+            struct ggml_tensor * meta = get_tensor_meta(i);
+            ctx_size_p += sizeof(struct ggml_tensor) + GGML_OBJECT_SIZE;
+            (use_mmap ? mmapped_size_p : ctx_size_p) += ggml_nbytes_pad(meta);
+        }
     }
 
-    struct ggml_tensor * get_tensor_for(llama_load_tensor & lt, ggml_backend backend) {
-        struct ggml_tensor * tensor;
+    struct ggml_tensor * create_tensor_for(struct ggml_context * ctx, struct ggml_tensor * meta, ggml_backend backend) {
         if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, true);
-        }
-        if (lt.ne.size() == 2) {
-            tensor = ggml_new_tensor_2d(ggml_ctx, lt.type, lt.ne.at(0), lt.ne.at(1));
-        } else {
-            LLAMA_ASSERT(lt.ne.size() == 1);
-            tensor = ggml_new_tensor_1d(ggml_ctx, lt.type, lt.ne.at(0));
+            ggml_set_no_alloc(ctx, true);
         }
-        ggml_set_name(tensor, lt.name.c_str());
-        LLAMA_ASSERT(lt.ggml_tensor == NULL); // if this fails, we called get_tensor twice on the same tensor
+
+        struct ggml_tensor * tensor = ggml_dup_tensor(ctx, meta);
+        tensor->backend = backend; // TODO: ggml_set_backend
+        ggml_set_name(tensor, ggml_get_name(meta));
 
         if (backend != GGML_BACKEND_CPU) {
-            ggml_set_no_alloc(ggml_ctx, use_mmap);
+            ggml_set_no_alloc(ctx, use_mmap);
         }
-        tensor->backend = backend;
-        lt.ggml_tensor = tensor;
-        num_ggml_tensors_created++;
+
+        n_created++;
+
         return tensor;
     }
 
+    struct ggml_tensor * create_tensor(struct ggml_context * ctx, const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
+        struct ggml_tensor * cur = ggml_get_tensor(ctx_meta, name.c_str());
+
+        {
+            bool is_ok = true;
+            for (size_t i = 0; i < ne.size(); ++i) {
+                if (ne[i] != cur->ne[i]) {
+                    is_ok = false;
+                    break;
+                }
+            }
+            if (!is_ok) {
+                throw std::runtime_error(
+                        format("%s: tensor '%s' has wrong shape; expected %s, got %s",
+                            __func__, name.c_str(),
+                            llama_format_tensor_shape(ne).c_str(),
+                            llama_format_tensor_shape(cur).c_str()));
+            }
+        }
+
+        return create_tensor_for(ctx, cur, backend);
+    }
+
     void done_getting_tensors() const {
-        if (num_ggml_tensors_created != tensors_map.tensors.size()) {
-            throw std::runtime_error(std::string("llama.cpp: file contained more tensors than expected"));
+        if (n_created != n_tensors) {
+            throw std::runtime_error(format("%s: wrong number of tensors; expected %d, got %d", __func__, n_tensors, n_created));
         }
     }
 
-    void load_all_data(llama_progress_callback progress_callback, void *  progress_callback_user_data, llama_mlock * lmlock) {
-        size_t data_size = 0;
-        size_t prefetch_size = file_loader->file.size;
-        size_t lock_size = 0;
-        for (const llama_load_tensor & lt : tensors_map.tensors) {
-            data_size += lt.size;
-            if (lt.ggml_tensor->backend != GGML_BACKEND_CPU) {
-                prefetch_size -= lt.size;
+    size_t file_offset(const char * name) const {
+        const int idx = gguf_find_tensor(ctx_gguf, name);
+
+        if (idx < 0) {
+            throw std::runtime_error(format("%s: tensor '%s' not found in the file", __func__, name));
+        }
+
+        return gguf_get_data_offset(ctx_gguf) + gguf_get_tensor_offset(ctx_gguf, idx);
+    }
+
+    void load_data_for(struct ggml_tensor * cur) const {
+        const size_t offs = file_offset(ggml_get_name(cur));
+
+        if (use_mmap) {
+            cur->data = (uint8_t *) mapping->addr + offs;
+        } else {
+            file.seek(offs, SEEK_SET);
+            file.read_raw(cur->data, ggml_nbytes(cur));
+        }
+    }
+
+    void load_all_data(struct ggml_context * ctx, llama_progress_callback progress_callback, void * progress_callback_user_data, llama_mlock * lmlock) {
+        size_t size_data = 0;
+        size_t size_lock = 0;
+        size_t size_pref = 0; // prefetch
+
+        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
+            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
+            size_data += ggml_nbytes(cur);
+            if (cur->backend == GGML_BACKEND_CPU) {
+                size_pref += ggml_nbytes(cur);
             }
         }
 
         if (use_mmap) {
-            mapping.reset(new llama_mmap(&file_loader->file, prefetch_size, ggml_is_numa()));
+            mapping.reset(new llama_mmap(&file, size_pref, ggml_is_numa()));
             if (lmlock) {
                 lmlock->init(mapping->addr);
             }
         }
 
         size_t done_size = 0;
-        for (llama_load_tensor & lt : tensors_map.tensors) {
+        for (int i = 0; i < gguf_get_n_tensors(ctx_gguf); i++) {
+            struct ggml_tensor * cur = ggml_get_tensor(ctx, gguf_get_tensor_name(ctx_gguf, i));
+            GGML_ASSERT(cur); // unused tensors should have been caught by load_data already
+
             if (progress_callback) {
-                progress_callback((float) done_size / data_size, progress_callback_user_data);
+                progress_callback((float) done_size / size_data, progress_callback_user_data);
             }
-            LLAMA_ASSERT(lt.ggml_tensor); // unused tensors should have been caught by load_data already
-            lt.data = (uint8_t *) lt.ggml_tensor->data;
 
             // allocate temp buffer if not using mmap
-            if (!use_mmap && lt.data == NULL) {
-                GGML_ASSERT(lt.ggml_tensor->backend != GGML_BACKEND_CPU);
-                lt.data = (uint8_t*)malloc(ggml_nbytes(lt.ggml_tensor));
+            if (!use_mmap && cur->data == NULL) {
+                GGML_ASSERT(cur->backend != GGML_BACKEND_CPU);
+                cur->data = malloc(ggml_nbytes(cur));
             }
 
-            load_data_for(lt);
+            load_data_for(cur);
 
-            switch(lt.ggml_tensor->backend) {
+            switch (cur->backend) {
                 case GGML_BACKEND_CPU:
-                    lt.ggml_tensor->data = lt.data;
                     if (use_mmap && lmlock) {
-                        lock_size += lt.size;
-                        lmlock->grow_to(lock_size);
+                        size_lock += ggml_nbytes(cur);
+                        lmlock->grow_to(size_lock);
                     }
                     break;
 #if defined(GGML_USE_CUBLAS)
                 case GGML_BACKEND_GPU:
                 case GGML_BACKEND_GPU_SPLIT:
-                    ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+                    // old code:
+                    //ggml_cuda_transform_tensor(lt.data, lt.ggml_tensor);
+
+                    // TODO: test if this works !!
+                    ggml_cuda_transform_tensor(cur->data, cur);
                     if (!use_mmap) {
-                        free(lt.data);
+                        free(cur->data);
                     }
                     break;
 #elif defined(GGML_USE_CLBLAST)
                 case GGML_BACKEND_GPU:
-                    ggml_cl_transform_tensor(lt.data, lt.ggml_tensor);
+                    ggml_cl_transform_tensor(cur->data, cur);
                     if (!use_mmap) {
-                        free(lt.data);
+                        free(cur->data);
                     }
                     break;
 #endif
@@ -825,348 +1231,215 @@ struct llama_model_loader {
                     continue;
             }
 
-            done_size += lt.size;
+            done_size += ggml_nbytes(cur);
         }
     }
+};
 
-    void load_data_for(llama_load_tensor & lt) {
-        if (use_mmap) {
-            lt.data = (uint8_t *) mapping->addr + lt.file_off;
-        } else {
-            llama_file & file = file_loader->file;
-            file.seek(lt.file_off, SEEK_SET);
-            file.read_raw(lt.data, lt.size);
-        }
+//
+// load LLaMA models
+//
 
-        if (0) {
-            print_checksum(lt);
-        }
+static const char * llama_ftype_name(enum llama_ftype ftype) {
+    switch (ftype) {
+        case LLAMA_FTYPE_ALL_F32:     return "all F32";
+        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
+        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
+        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
+        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
+                                      return "mostly Q4_1, some F16";
+        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
+        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
+        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
+
+        // K-quants
+        case LLAMA_FTYPE_MOSTLY_Q2_K:   return "mostly Q2_K";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
+        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
+        case LLAMA_FTYPE_MOSTLY_Q6_K:   return "mostly Q6_K";
+
+        default: return "unknown, may not work";
     }
+}
 
-    static void print_checksum(llama_load_tensor & lt) {
-        uint32_t sum = 0;
-        for (size_t i = 0; i < lt.size; i++) {
-            uint8_t byte = lt.data[i];
-            sum = byte + (sum << 6) + (sum << 16) - sum; // sdbm hash
-        }
-        LLAMA_LOG_INFO("%s checksum: %#08x (%s, size %zu)\n", lt.name.c_str(), sum,
-                llama_format_tensor_shape(lt.ne).c_str(), lt.size);
+static const char * llama_model_type_name(e_model type) {
+    switch (type) {
+        case MODEL_3B:  return "3B";
+        case MODEL_7B:  return "7B";
+        case MODEL_13B: return "13B";
+        case MODEL_30B: return "30B";
+        case MODEL_65B: return "65B";
+        case MODEL_70B: return "70B";
+        default: GGML_ASSERT(false);
     }
+}
 
-};
+static void llama_model_load_internal(
+        const std::string & fname,
+        llama_model & model,
+        llama_vocab & vocab,
+        int n_ctx,
+        int n_batch,
+        int n_gpu_layers,
+        int main_gpu,
+        const float * tensor_split,
+        const bool mul_mat_q,
+        float rope_freq_base,
+        float rope_freq_scale,
+        bool low_vram,
+        ggml_type memory_type,
+        bool use_mmap,
+        bool use_mlock,
+        bool vocab_only,
+        llama_progress_callback progress_callback,
+        void * progress_callback_user_data) {
+    model.t_start_us = ggml_time_us();
 
-//
-// kv cache
-//
+    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
 
-static bool kv_cache_init(
-        const struct llama_hparams & hparams,
-             struct llama_kv_cache & cache,
-                         ggml_type   wtype,
-                               int   n_ctx,
-                               int   n_gpu_layers) {
-    const int n_embd  = hparams.n_embd_gqa();
-    const int n_layer = hparams.n_layer;
+    model.n_gpu_layers = n_gpu_layers;
 
-    const int64_t n_mem      = n_layer*n_ctx;
-    const int64_t n_elements = n_embd*n_mem;
+    auto & hparams = model.hparams;
 
-    cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB);
-    cache.n = 0;
+    // read hparams
+    {
+        struct gguf_context * ctx = ml->ctx_gguf;
+
+#define GGUF_GET(dst, func, type, req, key) \
+        { \
+            const int kid = gguf_find_key(ctx, key); \
+            if (kid >= 0) { \
+                enum gguf_type ktype = gguf_get_kv_type(ctx, kid); \
+                if (ktype != (type)) { \
+                    throw std::runtime_error(format("key %s has wrong type: %s", key, gguf_type_name(ktype))); \
+                } \
+                (dst) = func(ctx, kid); \
+            } else if (req) { \
+                throw std::runtime_error(format("key not found in model: %s", key)); \
+            } \
+        }
 
-    struct ggml_init_params params;
-    params.mem_size   = cache.buf.size;
-    params.mem_buffer = cache.buf.addr;
-    params.no_alloc   = false;
+        GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
+        GGUF_GET(hparams.n_ctx,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
+        GGUF_GET(hparams.n_embd,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.embedding_length");
+        GGUF_GET(hparams.n_ff,           gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.feed_forward_length");
+        GGUF_GET(hparams.n_head,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.attention.head_count");
+        GGUF_GET(hparams.n_layer,        gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.block_count");
+        GGUF_GET(hparams.n_rot,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.rope.dimension_count");
+        GGUF_GET(hparams.f_norm_rms_eps, gguf_get_val_f32, GGUF_TYPE_FLOAT32, true, "llama.attention.layer_norm_rms_epsilon");
 
-    cache.ctx = ggml_init(params);
+        // n_head_kv is optional, default to n_head
+        hparams.n_head_kv = hparams.n_head;
+        GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
+#undef GGUF_GET
 
-    if (!cache.ctx) {
-        LLAMA_LOG_ERROR("%s: failed to allocate memory for kv cache\n", __func__);
-        return false;
-    }
+        switch (hparams.n_layer) {
+            case 26: model.type = e_model::MODEL_3B; break;
+            case 32: model.type = e_model::MODEL_7B; break;
+            case 40: model.type = e_model::MODEL_13B; break;
+            case 60: model.type = e_model::MODEL_30B; break;
+            case 80: model.type = e_model::MODEL_65B; break;
+            default:
+                {
+                    if (hparams.n_layer < 32) {
+                        model.type = e_model::MODEL_7B;
+                    }
+                } break;
+        }
 
-    cache.k = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    cache.v = ggml_new_tensor_1d(cache.ctx, wtype, n_elements);
-    ggml_set_name(cache.k, "cache_k");
-    ggml_set_name(cache.v, "cache_v");
+        hparams.n_ctx = n_ctx;
 
-    (void) n_gpu_layers;
-#ifdef GGML_USE_CUBLAS
-    if (n_gpu_layers > n_layer + 1) {
-        ggml_cuda_assign_buffers_no_scratch(cache.v);
-    }
-    if (n_gpu_layers > n_layer + 2) {
-        ggml_cuda_assign_buffers_no_scratch(cache.k);
-    }
-#endif // GGML_USE_CUBLAS
+        // LLaMAv2
+        // TODO: probably not needed
+        {
+            const auto n_gqa = hparams.n_gqa();
 
-    return true;
-}
+            if (model.type == e_model::MODEL_65B && n_gqa == 8) {
+                LLAMA_LOG_WARN("%s: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
+                model.type = e_model::MODEL_70B;
+            }
+        }
 
-struct llama_context_params llama_context_default_params() {
-    struct llama_context_params result = {
-        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
-        /*.n_ctx                       =*/ 512,
-        /*.n_batch                     =*/ 512,
-        /*.n_gqa                       =*/ 1,
-        /*.rms_norm_eps                =*/ LLAMA_DEFAULT_RMS_EPS,
-        /*.gpu_layers                  =*/ 0,
-        /*.main_gpu                    =*/ 0,
-        /*.tensor_split                =*/ nullptr,
-        /*.rope_freq_base              =*/ 10000.0f,
-        /*.rope_freq_scale             =*/ 1.0f,
-        /*.progress_callback           =*/ nullptr,
-        /*.progress_callback_user_data =*/ nullptr,
-        /*.low_vram                    =*/ false,
-        /*.mul_mat_q                   =*/ false,
-        /*.f16_kv                      =*/ true,
-        /*.logits_all                  =*/ false,
-        /*.vocab_only                  =*/ false,
-        /*.use_mmap                    =*/ true,
-        /*.use_mlock                   =*/ false,
-        /*.embedding                   =*/ false,
-    };
-
-    return result;
-}
-
-struct llama_model_quantize_params llama_model_quantize_default_params() {
-    struct llama_model_quantize_params result = {
-        /*.nthread                     =*/ 0,
-        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
-        /*.allow_requantize            =*/ false,
-        /*.quantize_output_tensor      =*/ true,
-    };
-
-    return result;
-}
-
-int llama_max_devices() {
-    return LLAMA_MAX_DEVICES;
-}
-
-bool llama_mmap_supported() {
-    return llama_mmap::SUPPORTED;
-}
-
-bool llama_mlock_supported() {
-    return llama_mlock::SUPPORTED;
-}
-
-void llama_backend_init(bool numa) {
-    ggml_time_init();
-
-    // needed to initialize f16 tables
-    {
-        struct ggml_init_params params = { 0, NULL, false };
-        struct ggml_context * ctx = ggml_init(params);
-        ggml_free(ctx);
-    }
-
-    if (numa) {
-        ggml_numa_init();
-    }
-
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_init();
-#endif
-}
-
-void llama_backend_free() {
-#ifdef GGML_USE_MPI
-    ggml_mpi_backend_free();
-#endif
-}
-
-int64_t llama_time_us() {
-    return ggml_time_us();
-}
-
-//
-// model loading
-//
-
-static const char *llama_file_version_name(llama_file_version version) {
-    switch (version) {
-        case LLAMA_FILE_VERSION_GGML: return "'ggml' (old version with low tokenizer quality and no mmap support)";
-        case LLAMA_FILE_VERSION_GGMF_V1: return "ggmf v1 (old version with no mmap support)";
-        case LLAMA_FILE_VERSION_GGJT_V1: return "ggjt v1 (pre #1405)";
-        case LLAMA_FILE_VERSION_GGJT_V2: return "ggjt v2 (pre #1508)";
-        case LLAMA_FILE_VERSION_GGJT_V3: return "ggjt v3 (latest)";
-    }
-
-    return "unknown";
-}
-
-static const char *llama_ftype_name(enum llama_ftype ftype) {
-    switch (ftype) {
-        case LLAMA_FTYPE_ALL_F32:     return "all F32";
-        case LLAMA_FTYPE_MOSTLY_F16:  return "mostly F16";
-        case LLAMA_FTYPE_MOSTLY_Q4_0: return "mostly Q4_0";
-        case LLAMA_FTYPE_MOSTLY_Q4_1: return "mostly Q4_1";
-        case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
-                                      return "mostly Q4_1, some F16";
-        case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
-        case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
-        case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
-        // K-quants
-        case LLAMA_FTYPE_MOSTLY_Q2_K: return "mostly Q2_K";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_S: return "mostly Q3_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_M: return "mostly Q3_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q3_K_L: return "mostly Q3_K - Large";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_S: return "mostly Q4_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q4_K_M: return "mostly Q4_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "mostly Q5_K - Small";
-        case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "mostly Q5_K - Medium";
-        case LLAMA_FTYPE_MOSTLY_Q6_K: return "mostly Q6_K";
-        default:                      return "unknown, may not work";
-    }
-}
-
-static const char *llama_model_type_name(e_model type) {
-    switch (type) {
-        case MODEL_3B: return "3B";
-        case MODEL_7B: return "7B";
-        case MODEL_13B: return "13B";
-        case MODEL_30B: return "30B";
-        case MODEL_65B: return "65B";
-        case MODEL_70B: return "70B";
-        default: LLAMA_ASSERT(false);
+        hparams.rope_freq_base  = rope_freq_base;
+        hparams.rope_freq_scale = rope_freq_scale;
     }
-}
-
-static void llama_model_load_internal(
-        const std::string & fname,
-        llama_model & model,
-        llama_vocab & vocab,
-        int n_ctx,
-        int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
-        int n_gpu_layers,
-        int main_gpu,
-        const float * tensor_split,
-        const bool mul_mat_q,
-        float rope_freq_base,
-        float rope_freq_scale,
-        bool low_vram,
-        ggml_type memory_type,
-        bool use_mmap,
-        bool use_mlock,
-        bool vocab_only,
-        llama_progress_callback progress_callback,
-        void * progress_callback_user_data) {
-
-    model.t_start_us = ggml_time_us();
 
-    std::unique_ptr<llama_model_loader> ml(new llama_model_loader(fname, use_mmap));
-
-    vocab = std::move(ml->file_loader->vocab);
+    // read vocab
+    {
+        struct gguf_context * ctx = ml->ctx_gguf;
 
-    if (vocab_only) {
-        return;
-    }
+        vocab.id_to_token.resize(hparams.n_vocab);
 
-    model.hparams = ml->file_loader->hparams;
-    model.n_gpu_layers = n_gpu_layers;
-    llama_file_version file_version = ml->file_loader->file_version;
+        const int token_idx = gguf_find_key(ctx, "tokenizer.ggml.tokens");
+        if (token_idx == -1) {
+            throw std::runtime_error("cannot find token list in GGUF file\n");
+        }
 
-    auto & hparams = model.hparams;
+        const int score_idx = gguf_find_key(ctx, "tokenizer.ggml.scores");
+        if (score_idx == -1) {
+            throw std::runtime_error("cannot find token scores list in GGUF file\n");
+        }
 
-    // TODO: read from file
-    hparams.f_rms_norm_eps = rms_norm_eps;
+        const float * scores = (const float * ) gguf_get_arr_data(ctx, score_idx);
 
-    {
-        switch (hparams.n_layer) {
-            case 26: model.type = e_model::MODEL_3B; break;
-            case 32: model.type = e_model::MODEL_7B; break;
-            case 40: model.type = e_model::MODEL_13B; break;
-            case 60: model.type = e_model::MODEL_30B; break;
-            case 80: model.type = e_model::MODEL_65B; break;
-            default:
-                {
-                    if (hparams.n_layer < 32) {
-                        model.type = e_model::MODEL_7B;
-                    }
-                } break;
-        }
+        for (uint32_t i = 0; i < hparams.n_vocab; i++) {
+            std::string word = gguf_get_arr_str(ctx, token_idx, i);
 
-        hparams.n_ctx = n_ctx;
+            vocab.token_to_id[word] = i;
 
-        // LLaMAv2
-        // TODO: temporary until GGUF
-        LLAMA_ASSERT(hparams.n_head % n_gqa == 0);
-        hparams.n_head_kv = hparams.n_head / n_gqa;
-        if (model.type == e_model::MODEL_65B && n_gqa == 8) {
-            LLAMA_LOG_WARN("%s: warning: assuming 70B model based on GQA == %d\n", __func__, n_gqa);
-            model.type = e_model::MODEL_70B;
-            hparams.f_ffn_mult = 1.3f; // from the params.json of the 70B model
+            auto & tok_score = vocab.id_to_token[i];
+            tok_score.tok = std::move(word);
+            tok_score.score = scores[i];
         }
-
-        hparams.rope_freq_base  = rope_freq_base;
-        hparams.rope_freq_scale = rope_freq_scale;
     }
 
-    // ref: https://github.com/facebookresearch/llama/blob/6c7fe276574e78057f917549435a2554000a876d/llama/model.py#L194-L199
-    const uint32_t n_ff_raw  = 2*(4*hparams.n_embd)/3;
-    const uint32_t n_ff_mult = hparams.f_ffn_mult*n_ff_raw;
-    const uint32_t n_ff      = ((n_ff_mult + hparams.n_mult - 1)/hparams.n_mult)*hparams.n_mult;
-    //const uint32_t n_ff = 28672;
-
     {
-        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, llama_file_version_name(file_version));
+        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, llama_file_version_name(ml->file_version));
         LLAMA_LOG_INFO("%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
         LLAMA_LOG_INFO("%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
         LLAMA_LOG_INFO("%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        LLAMA_LOG_INFO("%s: n_mult     = %u\n",   __func__, hparams.n_mult);
         LLAMA_LOG_INFO("%s: n_head     = %u\n",   __func__, hparams.n_head);
         LLAMA_LOG_INFO("%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
         LLAMA_LOG_INFO("%s: n_layer    = %u\n",   __func__, hparams.n_layer);
         LLAMA_LOG_INFO("%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
         LLAMA_LOG_INFO("%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: rnorm_eps  = %.1e\n", __func__, hparams.f_rms_norm_eps);
-        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, n_ff);
+        LLAMA_LOG_INFO("%s: f_norm_eps = %.1e\n", __func__, hparams.f_norm_rms_eps);
+        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, hparams.n_ff);
         LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
         LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
         LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
         LLAMA_LOG_INFO("%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
     }
 
-    if (file_version < LLAMA_FILE_VERSION_GGJT_V2) {
-        if (hparams.ftype != LLAMA_FTYPE_ALL_F32     &&
-            hparams.ftype != LLAMA_FTYPE_MOSTLY_F16  &&
-            hparams.ftype != LLAMA_FTYPE_MOSTLY_Q8_0) {
-            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1405)"));
-        }
-    }
-
-    if (file_version < LLAMA_FILE_VERSION_GGJT_V3) {
-        if (hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_0 ||
-            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q4_1 ||
-            hparams.ftype == LLAMA_FTYPE_MOSTLY_Q8_0) {
-            throw std::runtime_error(format("this format is no longer supported (see https://github.com/ggerganov/llama.cpp/pull/1508)"));
-        }
+    if (vocab_only) {
+        return;
     }
 
     auto & ctx = model.ctx;
 
     size_t ctx_size;
     size_t mmapped_size;
-    ml->calc_sizes(&ctx_size, &mmapped_size);
+
+    ml->calc_sizes(ctx_size, mmapped_size);
+
     LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MB\n", __func__, ctx_size/1024.0/1024.0);
 
     // create the ggml context
     {
         model.buf.resize(ctx_size);
         if (use_mlock) {
-            model.mlock_buf.init   (model.buf.addr);
+            model.mlock_buf.init   (model.buf.data);
             model.mlock_buf.grow_to(model.buf.size);
         }
 
         struct ggml_init_params params = {
             /*.mem_size   =*/ model.buf.size,
-            /*.mem_buffer =*/ model.buf.addr,
+            /*.mem_buffer =*/ model.buf.data,
             /*.no_alloc   =*/ ml->use_mmap,
         };
 
@@ -1202,9 +1475,7 @@ static void llama_model_load_internal(
         const uint32_t n_layer    = hparams.n_layer;
         const uint32_t n_vocab    = hparams.n_vocab;
 
-        ml->ggml_ctx = ctx;
-
-        model.tok_embeddings = ml->get_tensor("tok_embeddings.weight", {n_embd, n_vocab}, GGML_BACKEND_CPU);
+        model.tok_embeddings = ml->create_tensor(ctx, TN_TOKEN_EMBD, {n_embd, n_vocab}, GGML_BACKEND_CPU);
 
         // "output" tensor
         {
@@ -1225,8 +1496,8 @@ static void llama_model_load_internal(
                 backend_output = GGML_BACKEND_CPU;
             }
 
-            model.norm   = ml->get_tensor("norm.weight",   {n_embd},          backend_norm);
-            model.output = ml->get_tensor("output.weight", {n_embd, n_vocab}, backend_output);
+            model.norm   = ml->create_tensor(ctx, TN_OUTPUT_NORM, {n_embd},          backend_norm);
+            model.output = ml->create_tensor(ctx, TN_OUTPUT,      {n_embd, n_vocab}, backend_output);
             if (backend_norm == GGML_BACKEND_GPU) {
                 vram_weights += ggml_nbytes(model.norm);
             }
@@ -1235,6 +1506,8 @@ static void llama_model_load_internal(
             }
         }
 
+        const uint32_t n_ff = hparams.n_ff;
+
         const int i_gpu_start = n_layer - n_gpu_layers;
 
         model.layers.resize(n_layer);
@@ -1243,21 +1516,18 @@ static void llama_model_load_internal(
             const ggml_backend backend_split = int(i) < i_gpu_start ? GGML_BACKEND_CPU : LLAMA_BACKEND_OFFLOAD_SPLIT; // NOLINT
 
             auto & layer = model.layers[i];
+            layer.attention_norm = ml->create_tensor(ctx, format(TN_ATTN_NORM, i), {n_embd}, backend);
 
-            std::string layers_i = "layers." + std::to_string(i);
-
-            layer.attention_norm = ml->get_tensor(layers_i + ".attention_norm.weight", {n_embd}, backend);
+            layer.wq = ml->create_tensor(ctx, format(TN_ATTN_Q, i),      {n_embd, n_embd},     backend_split);
+            layer.wk = ml->create_tensor(ctx, format(TN_ATTN_K, i),      {n_embd, n_embd_gqa}, backend_split);
+            layer.wv = ml->create_tensor(ctx, format(TN_ATTN_V, i),      {n_embd, n_embd_gqa}, backend_split);
+            layer.wo = ml->create_tensor(ctx, format(TN_ATTN_OUTPUT, i), {n_embd, n_embd},     backend_split);
 
-            layer.wq = ml->get_tensor(layers_i + ".attention.wq.weight", {n_embd, n_embd},     backend_split);
-            layer.wk = ml->get_tensor(layers_i + ".attention.wk.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wv = ml->get_tensor(layers_i + ".attention.wv.weight", {n_embd, n_embd_gqa}, backend_split);
-            layer.wo = ml->get_tensor(layers_i + ".attention.wo.weight", {n_embd, n_embd},     backend_split);
+            layer.ffn_norm = ml->create_tensor(ctx, format(TN_FFN_NORM, i), {n_embd}, backend);
 
-            layer.ffn_norm = ml->get_tensor(layers_i + ".ffn_norm.weight", {n_embd}, backend);
-
-            layer.w1 = ml->get_tensor(layers_i + ".feed_forward.w1.weight", {n_embd,   n_ff}, backend_split);
-            layer.w2 = ml->get_tensor(layers_i + ".feed_forward.w2.weight", {  n_ff, n_embd}, backend_split);
-            layer.w3 = ml->get_tensor(layers_i + ".feed_forward.w3.weight", {n_embd,   n_ff}, backend_split);
+            layer.w1 = ml->create_tensor(ctx, format(TN_FFN_GATE, i), {n_embd,   n_ff}, backend_split);
+            layer.w2 = ml->create_tensor(ctx, format(TN_FFN_DOWN, i), {  n_ff, n_embd}, backend_split);
+            layer.w3 = ml->create_tensor(ctx, format(TN_FFN_UP, i),   {n_embd,   n_ff}, backend_split);
 
             if (backend == GGML_BACKEND_GPU) {
                 vram_weights +=
@@ -1355,8 +1625,9 @@ static void llama_model_load_internal(
     }
 
     // populate `tensors_by_name`
-    for (llama_load_tensor & lt : ml->tensors_map.tensors) {
-        model.tensors_by_name.emplace_back(lt.name, lt.ggml_tensor);
+    for (int i = 0; i < ml->n_tensors; ++i) {
+        struct ggml_tensor * cur = ggml_get_tensor(ctx, ml->get_tensor_name(i));
+        model.tensors_by_name.emplace_back(ggml_get_name(cur), cur);
     }
 
     (void) tensor_split;
@@ -1366,7 +1637,7 @@ static void llama_model_load_internal(
     }
 #endif
 
-    ml->load_all_data(progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
+    ml->load_all_data(ctx, progress_callback, progress_callback_user_data, use_mlock ? &model.mlock_mmap : NULL);
 
     if (progress_callback) {
         progress_callback(1.0f, progress_callback_user_data);
@@ -1385,8 +1656,6 @@ static bool llama_model_load(
         llama_vocab & vocab,
         int n_ctx,
         int n_batch,
-        int n_gqa,
-        float rms_norm_eps,
         int n_gpu_layers,
         int main_gpu,
         const float * tensor_split,
@@ -1401,7 +1670,7 @@ static bool llama_model_load(
         llama_progress_callback progress_callback,
         void *progress_callback_user_data) {
     try {
-        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gqa, rms_norm_eps, n_gpu_layers,
+        llama_model_load_internal(fname, model, vocab, n_ctx, n_batch, n_gpu_layers,
                                   main_gpu, tensor_split, mul_mat_q, rope_freq_base, rope_freq_scale, low_vram, memory_type,
                                   use_mmap, use_mlock, vocab_only, progress_callback, progress_callback_user_data);
         return true;
@@ -1418,7 +1687,7 @@ static struct ggml_cgraph * llama_build_graph(
                    int   n_tokens,
                    int   n_past) {
 
-    LLAMA_ASSERT((!tokens && embd) || (tokens && !embd));
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
 
     const int N = n_tokens;
 
@@ -1427,7 +1696,7 @@ static struct ggml_cgraph * llama_build_graph(
 
     const auto & kv_self = lctx.kv_self;
 
-    LLAMA_ASSERT(!!kv_self.ctx);
+    GGML_ASSERT(!!kv_self.ctx);
 
     const int64_t n_embd      = hparams.n_embd;
     const int64_t n_layer     = hparams.n_layer;
@@ -1437,21 +1706,20 @@ static struct ggml_cgraph * llama_build_graph(
     const int64_t n_embd_head = hparams.n_embd_head();
     const int64_t n_embd_gqa  = hparams.n_embd_gqa();
 
-    LLAMA_ASSERT(n_embd_head == hparams.n_rot);
+    GGML_ASSERT(n_embd_head == hparams.n_rot);
 
-    const float freq_base  = hparams.rope_freq_base;
-    const float freq_scale = hparams.rope_freq_scale;
-    const float rms_norm_eps = hparams.f_rms_norm_eps;
+    const float freq_base    = hparams.rope_freq_base;
+    const float freq_scale   = hparams.rope_freq_scale;
+    const float norm_rms_eps = hparams.f_norm_rms_eps;
 
     const int n_gpu_layers = model.n_gpu_layers;
 
     auto & mem_per_token = lctx.mem_per_token;
     auto & buf_compute   = lctx.buf_compute;
 
-
     struct ggml_init_params params = {
         /*.mem_size   =*/ buf_compute.size,
-        /*.mem_buffer =*/ buf_compute.addr,
+        /*.mem_buffer =*/ buf_compute.data,
         /*.no_alloc   =*/ false,
     };
 
@@ -1549,7 +1817,7 @@ static struct ggml_cgraph * llama_build_graph(
 
         // norm
         {
-            cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+            cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
             offload_func(cur);
             ggml_set_name(cur, "rms_norm_0");
 
@@ -1694,7 +1962,7 @@ static struct ggml_cgraph * llama_build_graph(
         {
             // norm
             {
-                cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps);
+                cur = ggml_rms_norm(ctx0, inpFF, norm_rms_eps);
                 offload_func(cur);
                 ggml_set_name(cur, "rms_norm_1");
 
@@ -1744,7 +2012,7 @@ static struct ggml_cgraph * llama_build_graph(
 
     // norm
     {
-        cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps);
+        cur = ggml_rms_norm(ctx0, inpL, norm_rms_eps);
         offload_func_nr(cur);
         ggml_set_name(cur, "rms_norm_2");
 
@@ -1801,7 +2069,7 @@ static bool llama_eval_internal(
                    int   n_threads,
             const char * cgraph_fname) {
 
-    LLAMA_ASSERT((!tokens && embd) || (tokens && !embd));
+    GGML_ASSERT((!tokens && embd) || (tokens && !embd)); // NOLINT
 
     const int64_t t_start_us = ggml_time_us();
 
@@ -1816,7 +2084,7 @@ static bool llama_eval_internal(
 
     const auto & kv_self = lctx.kv_self;
 
-    LLAMA_ASSERT(!!kv_self.ctx);
+    GGML_ASSERT(!!kv_self.ctx);
 
     const int64_t n_embd      = hparams.n_embd;
     const int64_t n_vocab     = hparams.n_vocab;
@@ -1840,8 +2108,8 @@ static bool llama_eval_internal(
     struct ggml_tensor * res = gf->nodes[gf->n_nodes - 1];
     struct ggml_tensor * embeddings = gf->nodes[gf->n_nodes - 2];
 
-    LLAMA_ASSERT(strcmp(res->name, "result_output") == 0);
-    LLAMA_ASSERT(strcmp(embeddings->name, "result_norm") == 0);
+    GGML_ASSERT(strcmp(res->name, "result_output") == 0);
+    GGML_ASSERT(strcmp(embeddings->name, "result_norm") == 0);
 
 #if GGML_USE_MPI
     const int64_t n_layer = hparams.n_layer;
@@ -1997,15 +2265,15 @@ static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
 }
 
 static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
-    UNUSED(vocab);
-    UNUSED(token);
+    GGML_UNUSED(vocab);
+    GGML_UNUSED(token);
     // TODO: improve?
     return false;
 }
 
 static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
-    UNUSED(vocab);
-    UNUSED(token);
+    GGML_UNUSED(vocab);
+    GGML_UNUSED(token);
     // TODO: improve?
     return false;
 }
@@ -2024,7 +2292,7 @@ static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
 
 static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
     if (llama_vocab_type(vocab) == "spm") {
-        return byte + 3;
+        return byte - 3;
     }
 
     if (llama_vocab_type(vocab) == "bpe") {
@@ -2278,8 +2546,8 @@ std::vector<uint32_t> decode_utf8(const char * src) {
 // returns true iff pos points to the end of one of the definitions of a rule
 static bool llama_grammar_is_end_of_sequence(const llama_grammar_element * pos) {
     switch (pos->type) {
-        case LLAMA_GRETYPE_END: return true;
-        case LLAMA_GRETYPE_ALT: return true;
+        case LLAMA_GRETYPE_END: return true;  // NOLINT
+        case LLAMA_GRETYPE_ALT: return true;  // NOLINT
         default:                return false;
     }
 }
@@ -2292,7 +2560,7 @@ static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
 
     bool found            = false;
     bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
-    LLAMA_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT);
+    GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT); // NOLINT
 
     do {
         if (pos[1].type == LLAMA_GRETYPE_CHAR_RNG_UPPER) {
@@ -2360,7 +2628,7 @@ static void llama_grammar_advance_stack(
             // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range
             // (LLAMA_GRETYPE_CHAR_ALT, LLAMA_GRETYPE_CHAR_RNG_UPPER); stack should never be left on
             // those
-            LLAMA_ASSERT(false);
+            GGML_ASSERT(false);
     }
 }
 
@@ -2427,7 +2695,7 @@ static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates_for_
         }
     }
 
-    auto stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
+    const auto * stack_pos_after = llama_grammar_match_char(stack_pos, 0).second;
 
     // update top of stack to next element, if any
     std::vector<const llama_grammar_element *> stack_after(stack.begin(), stack.end() - 1);
@@ -2449,7 +2717,7 @@ static std::vector<llama_grammar_candidate> llama_grammar_reject_candidates(
         const std::vector<std::vector<llama_grammar_element>>         & rules,
         const std::vector<std::vector<const llama_grammar_element *>> & stacks,
         const std::vector<llama_grammar_candidate>                    & candidates) {
-    LLAMA_ASSERT(!stacks.empty()); // REVIEW
+    GGML_ASSERT(!stacks.empty()); // REVIEW
 
     if (candidates.empty()) {
         return std::vector<llama_grammar_candidate>();
@@ -2660,7 +2928,6 @@ void llama_sample_tail_free(struct llama_context * ctx, llama_token_data_array *
     }
 }
 
-
 void llama_sample_typical(struct llama_context * ctx, llama_token_data_array * candidates, float p, size_t min_keep) {
     // Reference implementation:
     // https://github.com/huggingface/transformers/compare/main...cimeister:typical-sampling:typical-pr
@@ -2828,9 +3095,8 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
         }
     }
 
-    const auto rejects =
-        llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
-    for (auto & reject : rejects) {
+    const auto rejects = llama_grammar_reject_candidates(grammar->rules, grammar->stacks, candidates_grammar);
+    for (const auto & reject : rejects) {
         candidates->data[reject.index].logit = -INFINITY;
     }
 
@@ -3021,7 +3287,7 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
                 return;
             }
         }
-        LLAMA_ASSERT(false);
+        GGML_ASSERT(false);
     }
 
     std::string str = llama_token_to_str(ctx, token);
@@ -3030,7 +3296,7 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
     }
-    LLAMA_ASSERT(!grammar->stacks.empty());
+    GGML_ASSERT(!grammar->stacks.empty());
 
     ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
 }
@@ -3039,37 +3305,37 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
 // quantization
 //
 
-static void llama_convert_tensor_internal(const llama_load_tensor & tensor, llama_buffer & output, const int nelements, const int nthread) {
-    if (output.size < nelements * sizeof(float)) {
-        output.resize(nelements * sizeof(float));
+static void llama_convert_tensor_internal(struct ggml_tensor * tensor, std::vector<float> & output, const size_t nelements, const int nthread) {
+    if (output.size() < nelements) {
+        output.resize(nelements);
     }
-    float * f32_output = (float *) output.addr;
+    float * f32_output = (float *) output.data();
 
     ggml_type_traits_t qtype;
-    if (ggml_is_quantized(tensor.type)) {
-        qtype = ggml_internal_get_type_traits(tensor.type);
+    if (ggml_is_quantized(tensor->type)) {
+        qtype = ggml_internal_get_type_traits(tensor->type);
         if (qtype.to_float == NULL) {
-            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor.type)));
+            throw std::runtime_error(format("type %s unsupported for integer quantization: no dequantization available", ggml_type_name(tensor->type)));
         }
-    } else if (tensor.type != GGML_TYPE_F16) {
-        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor.type)));
+    } else if (tensor->type != GGML_TYPE_F16) {
+        throw std::runtime_error(format("cannot dequantize/convert tensor type %s", ggml_type_name(tensor->type)));
     }
 
     if (nthread < 2) {
-        if (tensor.type == GGML_TYPE_F16) {
-            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor.data, f32_output, nelements);
-        } else if (ggml_is_quantized(tensor.type)) {
-            qtype.to_float(tensor.data, f32_output, nelements);
+        if (tensor->type == GGML_TYPE_F16) {
+            ggml_fp16_to_fp32_row((ggml_fp16_t *)tensor->data, f32_output, nelements);
+        } else if (ggml_is_quantized(tensor->type)) {
+            qtype.to_float(tensor->data, f32_output, nelements);
         } else {
-            LLAMA_ASSERT(false); // unreachable
+            GGML_ASSERT(false); // unreachable
         }
         return;
     }
 
-    auto block_size = tensor.type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor.type);
-    auto block_size_bytes = ggml_type_size(tensor.type);
+    auto block_size = tensor->type == GGML_TYPE_F16 ? 1 : (size_t)ggml_blck_size(tensor->type);
+    auto block_size_bytes = ggml_type_size(tensor->type);
 
-    LLAMA_ASSERT(nelements % block_size == 0);
+    GGML_ASSERT(nelements % block_size == 0);
     auto nblocks = nelements / block_size;
     auto blocks_per_thread = nblocks / nthread;
     auto spare_blocks = nblocks - (blocks_per_thread * nthread); // if blocks aren't divisible by thread count
@@ -3087,14 +3353,13 @@ static void llama_convert_tensor_internal(const llama_load_tensor & tensor, llam
                 qtype.to_float(inbuf, outbuf, nels);
             }
         };
-        workers.push_back(std::thread(compute, tensor.type, tensor.data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
+        workers.push_back(std::thread(compute, tensor->type, (uint8_t *) tensor->data + in_buff_offs, f32_output + out_buff_offs, thr_elems));
         in_buff_offs += thr_block_bytes;
         out_buff_offs += thr_elems;
     }
     for (auto & worker : workers) {
         worker.join();
     }
-
 }
 
 static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
@@ -3131,16 +3396,27 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     }
 
     std::unique_ptr<llama_model_loader> model_loader(new llama_model_loader(fname_inp, /*use_mmap*/ false));
-    llama_file_saver file_saver(fname_out.c_str(), model_loader->file_loader.get(), params->ftype);
+
+    const size_t align = GGUF_DEFAULT_ALIGNMENT;
+    struct gguf_context * ctx_out = gguf_init_empty();
+
+    // copy the KV pairs from the input file
+    gguf_set_kv     (ctx_out, model_loader->ctx_gguf);
+    gguf_set_val_u32(ctx_out, "general.quantization_version", GGML_QNT_VERSION);
 
 #ifdef GGML_USE_K_QUANTS
     int n_attention_wv    = 0;
     int n_feed_forward_w2 = 0;
-    for (auto& tensor : model_loader->tensors_map.tensors) {
-        if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
+
+        const std::string name = ggml_get_name(meta);
+
+        if (name.find("attn_v.weight") != std::string::npos) {
             ++n_attention_wv;
         }
-        else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+        else if (name.find("ffn_down.weight") != std::string::npos) {
             ++n_feed_forward_w2;
         }
     }
@@ -3160,46 +3436,68 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         return i_layer < num_layers/8 || i_layer >= 7*num_layers/8 || (i_layer - num_layers/8)%3 == 2;
     };
 
-    size_t idx = 0;
-    for (llama_load_tensor & tensor : model_loader->tensors_map.tensors) {
-        llama_buffer read_data;
-        read_data.resize(tensor.size);
-        tensor.data = read_data.addr;
+    int idx = 0;
+
+    std::vector<uint8_t> read_data;
+    std::vector<uint8_t> work;
+
+    // populate the original tensors so we get an initial meta data
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * meta = model_loader->get_tensor_meta(i);
+        gguf_add_tensor(ctx_out, meta);
+    }
+
+    std::ofstream fout(fname_out, std::ios::binary);
+
+    const size_t meta_size = gguf_get_meta_size(ctx_out);
+
+    LLAMA_LOG_INFO("%s: meta size = %zu bytes\n", __func__, meta_size);
+
+    // placeholder for the meta data
+    ::zeros(fout, meta_size);
+
+    for (int i = 0; i < model_loader->n_tensors; ++i) {
+        struct ggml_tensor * tensor = model_loader->get_tensor_meta(i);
+
+        const std::string name = ggml_get_name(tensor);
+
+        read_data.resize(ggml_nbytes(tensor));
+        tensor->data = read_data.data();
         model_loader->load_data_for(tensor);
 
-        LLAMA_LOG_INFO("[%4zu/%4zu] %36s - %16s, type = %6s, ",
-               ++idx, model_loader->tensors_map.tensors.size(),
-               tensor.name.c_str(), llama_format_tensor_shape(tensor.ne).c_str(),
-               ggml_type_name(tensor.type));
+        LLAMA_LOG_INFO("[%4d/%4d] %36s - [%s], type = %6s, ",
+               ++idx, model_loader->n_tensors,
+               ggml_get_name(tensor),
+               llama_format_tensor_shape(tensor).c_str(),
+               ggml_type_name(tensor->type));
 
         // This used to be a regex, but <regex> has an extreme cost to compile times.
-        bool quantize = tensor.name.rfind("weight") == tensor.name.size() - 6; // ends with 'weight'?
+        bool quantize = name.rfind("weight") == name.size() - 6; // ends with 'weight'?
 
         // quantize only 2D tensors
-        quantize &= (tensor.ne.size() == 2);
-        quantize &= params->quantize_output_tensor || tensor.name != "output.weight";
-        quantize &= quantized_type != tensor.type;
+        quantize &= (tensor->n_dims == 2);
+        quantize &= params->quantize_output_tensor || name != "output.weight";
+        quantize &= quantized_type != tensor->type;
 
         enum ggml_type new_type;
         void * new_data;
         size_t new_size;
-        llama_buffer work;
 
         if (!quantize) {
-            new_type = tensor.type;
-            new_data = tensor.data;
-            new_size = tensor.size;
-            LLAMA_LOG_INFO("size = %8.3f MB\n", tensor.size/1024.0/1024.0);
+            new_type = tensor->type;
+            new_data = tensor->data;
+            new_size = ggml_nbytes(tensor);
+            LLAMA_LOG_INFO("size = %8.3f MB\n", ggml_nbytes(tensor)/1024.0/1024.0);
         } else {
             new_type = quantized_type;
 #ifdef GGML_USE_K_QUANTS
-            if (tensor.name == "output.weight") {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
+            if (name == TN_OUTPUT) {
+                int nx = tensor->ne[0];
+                int ny = tensor->ne[1];
                 if (nx % QK_K == 0 && ny % QK_K == 0) {
                     new_type = GGML_TYPE_Q6_K;
                 }
-            } else if (tensor.name.find("attention.wv.weight") != std::string::npos) {
+            } else if (name.find("attn_v.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
                 else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
@@ -3207,32 +3505,32 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
                         (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
                 ++i_attention_wv;
-            } else if (tensor.name.find("feed_forward.w2.weight") != std::string::npos) {
+            } else if (name.find("feed_forward.w2.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
                 else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&
                          use_more_bits(i_feed_forward_w2, n_feed_forward_w2)) new_type = GGML_TYPE_Q6_K;
                 //else if (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S && i_feed_forward_w2 < n_feed_forward_w2/8) new_type = GGML_TYPE_Q6_K;
                 ++i_feed_forward_w2;
-            } else if (tensor.name.find("attention.wo.weight") != std::string::npos) {
+            } else if (name.find("attn_output.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
             }
             bool convert_incompatible_tensor = false;
             if (new_type == GGML_TYPE_Q2_K || new_type == GGML_TYPE_Q3_K || new_type == GGML_TYPE_Q4_K ||
                 new_type == GGML_TYPE_Q5_K || new_type == GGML_TYPE_Q6_K) {
-                int nx = tensor.ne.at(0);
-                int ny = tensor.ne.at(1);
+                int nx = tensor->ne[0];
+                int ny = tensor->ne[1];
                 if (nx % QK_K != 0 || ny % QK_K != 0) {
                     LLAMA_LOG_INFO("\n\nTensor sizes %d x %d are not divisible by %d, required for k-quants.\n",nx,ny,QK_K);
                     convert_incompatible_tensor = true;
                 }
             }
             if (convert_incompatible_tensor) {
-                if (tensor.name == "output.weight") {
+                if (name == TN_OUTPUT) {
                     new_type = GGML_TYPE_F16; //fall back to F16 instead of just failing.
                     LLAMA_LOG_WARN("F16 will be used for this tensor instead.\n");
-                } else if (tensor.name == "tok_embeddings.weight") {
+                } else if (name == TN_TOKEN_EMBD) {
                     new_type = GGML_TYPE_Q4_0; //fall back to Q4_0 instead of just failing.
                     LLAMA_LOG_WARN("Q4_0 will be used for this tensor instead.\n");
                 } else {
@@ -3241,27 +3539,28 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             }
 #endif
 
+            const size_t nelements = ggml_nelements(tensor);
+
             float * f32_data;
-            size_t nelements = tensor.ne.at(0) * tensor.ne.at(1);
-            llama_buffer f32_conv_buf;
+            std::vector<float> f32_conv_buf;
 
-            if (tensor.type == GGML_TYPE_F32) {
-                f32_data = (float *) tensor.data;
-            } else if (ggml_is_quantized(tensor.type) && !params->allow_requantize) {
-                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor.type)));
+            if (tensor->type == GGML_TYPE_F32) {
+                f32_data = (float *) tensor->data;
+            } else if (ggml_is_quantized(tensor->type) && !params->allow_requantize) {
+                throw std::runtime_error(format("requantizing from type %s is disabled", ggml_type_name(tensor->type)));
             } else {
                 llama_convert_tensor_internal(tensor, f32_conv_buf, nelements, nthread);
-                f32_data = (float *) f32_conv_buf.addr;
+                f32_data = (float *) f32_conv_buf.data();
             }
 
             LLAMA_LOG_INFO("quantizing to %s .. ", ggml_type_name(new_type));
             fflush(stdout);
 
             work.resize(nelements * 4); // upper bound on size
-            new_data = work.addr;
+            new_data = work.data();
             std::vector<int64_t> hist_cur(1 << 4, 0);
 
-            int chunk_size = 32 * 512;
+            const int chunk_size = 32 * 512;
             const int nchunk = (nelements + chunk_size - 1)/chunk_size;
             const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
             if (nthread_use < 2) {
@@ -3269,7 +3568,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             } else {
                 size_t counter = 0;
                 new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size] () {
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements] () {
                     std::vector<int64_t> local_hist;
                     size_t local_size = 0;
                     while (true) {
@@ -3304,7 +3603,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 }
             }
 
-            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", tensor.size/1024.0/1024.0, new_size/1024.0/1024.0);
+            LLAMA_LOG_INFO("size = %8.2f MB -> %8.2f MB | hist: ", ggml_nbytes(tensor)/1024.0/1024.0, new_size/1024.0/1024.0);
             int64_t tot_count = 0;
             for (size_t i = 0; i < hist_cur.size(); i++) {
                 hist_all[i] += hist_cur[i];
@@ -3318,14 +3617,34 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             }
             LLAMA_LOG_INFO("\n");
         }
-        total_size_org += tensor.size;
+        total_size_org += ggml_nbytes(tensor);
         total_size_new += new_size;
-        file_saver.write_tensor(tensor, new_type, new_data, new_size);
+
+        // update the gguf meta data as we go
+        gguf_set_tensor_type(ctx_out, name.c_str(), new_type);
+        gguf_set_tensor_data(ctx_out, name.c_str(), new_data, new_size);
+
+        // write tensor data + padding
+        fout.write((const char *) new_data, new_size);
+        zeros(fout, GGML_PAD(new_size, align) - new_size);
     }
 
+    // go back to beginning of file and write the updated meta data
+    {
+        fout.seekp(0);
+        std::vector<uint8_t> data(gguf_get_meta_size(ctx_out));
+        gguf_get_meta_data(ctx_out, data.data());
+        fout.write((const char *) data.data(), data.size());
+    }
+
+    fout.close();
+
+    gguf_free(ctx_out);
+
     LLAMA_LOG_INFO("%s: model size  = %8.2f MB\n", __func__, total_size_org/1024.0/1024.0);
     LLAMA_LOG_INFO("%s: quant size  = %8.2f MB\n", __func__, total_size_new/1024.0/1024.0);
 
+    // print histogram for all tensors
     {
         int64_t sum_all = 0;
         for (size_t i = 0; i < hist_all.size(); i++) {
@@ -3342,508 +3661,574 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     }
 }
 
+int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
+    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
 
+    const int64_t t_start_lora_us = ggml_time_us();
 
-//
-// interface implementation
-//
-
-struct llama_model * llama_load_model_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
-    ggml_time_init();
-
-    llama_model * model = new llama_model;
-
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
-
-    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gqa, params.rms_norm_eps, params.n_gpu_layers,
-                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,params.low_vram,
-                memory_type, params.use_mmap, params.use_mlock, params.vocab_only, params.progress_callback,
-                params.progress_callback_user_data)) {
-        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
-        delete model;
-        return nullptr;
+    auto fin = std::ifstream(path_lora, std::ios::binary);
+    if (!fin) {
+        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
+        return 1;
     }
 
-    return model;
-}
-
-void llama_free_model(struct llama_model * model) {
-    delete model;
-}
-
-struct llama_context * llama_new_context_with_model(
-                 struct llama_model * model,
-        struct llama_context_params   params) {
+    // verify magic and version
+    {
+        uint32_t magic;
+        fin.read((char *) &magic, sizeof(magic));
+        uint32_t format_version;
+        fin.read((char *) &format_version, sizeof(format_version));
 
-    if (!model) {
-        return nullptr;
+        if (format_version != 1) {
+            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
+            return 1;
+        }
     }
 
-    llama_context * ctx = new llama_context(*model);
+    int32_t lora_r;
+    int32_t lora_alpha;
+    fin.read((char *) &lora_r, sizeof(lora_r));
+    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
+    float scaling = (float)lora_alpha / (float)lora_r;
 
-    if (params.seed == LLAMA_DEFAULT_SEED) {
-        params.seed = time(NULL);
+    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
+
+    // create a temporary ggml context to store the lora tensors
+    // todo: calculate size from biggest possible tensor
+    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
+    struct ggml_init_params params;
+    params.mem_size   = lora_buf.size();
+    params.mem_buffer = lora_buf.data();
+    params.no_alloc   = false;
+
+    ggml_context * lora_ctx = ggml_init(params);
+    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+
+    // create a name -> tensor map of the model to accelerate lookups
+    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
+    for (const auto & kv : model.tensors_by_name) {
+        model_tensors.insert(kv);
     }
 
-    unsigned cur_percentage = 0;
-    if (params.progress_callback == NULL) {
-        params.progress_callback_user_data = &cur_percentage;
-        params.progress_callback = [](float progress, void * ctx) {
-            unsigned * cur_percentage_p = (unsigned *) ctx;
-            unsigned percentage = (unsigned) (100 * progress);
-            while (percentage > *cur_percentage_p) {
-                *cur_percentage_p = percentage;
-                LLAMA_LOG_INFO(".");
-                if (percentage >= 100) {
-                    LLAMA_LOG_INFO("\n");
-                }
-            }
-        };
+    // load base model
+    std::unique_ptr<llama_model_loader> model_loader;
+    ggml_context * base_ctx = NULL;
+    std::vector<uint8_t> base_buf;
+    if (path_base_model) {
+        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
+        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
+
+        size_t ctx_size;
+        size_t mmapped_size;
+        model_loader->calc_sizes(ctx_size, mmapped_size);
+        base_buf.resize(ctx_size);
+
+        ggml_init_params base_params;
+        base_params.mem_size   = base_buf.size();
+        base_params.mem_buffer = base_buf.data();
+        base_params.no_alloc   = model_loader->use_mmap;
+
+        base_ctx = ggml_init(base_params);
+
+        // maybe this should in llama_model_loader
+        if (model_loader->use_mmap) {
+            model_loader->mapping.reset(new llama_mmap(&model_loader->file, /* prefetch */ 0, ggml_is_numa()));
+        }
     }
 
-    ctx->rng = std::mt19937(params.seed);
-    ctx->logits_all = params.logits_all;
+    // read tensors and apply
+    bool warned = false;
+    int n_tensors = 0;
 
-    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
+    std::vector<uint8_t> work_buffer;
 
-    // reserve memory for context buffers
-    if (!params.vocab_only) {
-        if (!kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
-            LLAMA_LOG_ERROR("%s: kv_cache_init() failed for self-attention cache\n", __func__);
-            llama_free(ctx);
-            return nullptr;
+    while (true) {
+        int32_t n_dims;
+        int32_t length;
+        int32_t ftype;
+
+        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
+        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
+        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
+        if (fin.eof()) {
+            break;
+        }
+
+        int32_t ne[2] = { 1, 1 };
+        for (int i = 0; i < n_dims; ++i) {
+            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
         }
 
+        std::string name;
         {
-            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
-            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
+            char buf[1024];
+            fin.read(buf, length);
+            name = std::string(buf, length);
         }
 
-        const auto & hparams = ctx->model.hparams;
+        // check for lora suffix and get the type of tensor
+        const std::string lora_suffix = ".lora";
+        size_t pos = name.rfind(lora_suffix);
+        if (pos == std::string::npos) {
+            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
+            return 1;
+        }
 
-        // resized during inference
-        if (params.logits_all) {
-            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
-        } else {
-            ctx->logits.reserve(hparams.n_vocab);
+        std::string lora_type = name.substr(pos + lora_suffix.length());
+        std::string base_name = name;
+        base_name.erase(pos);
+        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+
+        if (model_tensors.find(base_name) == model_tensors.end()) {
+            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
+            return 1;
         }
 
-        if (params.embedding){
-            ctx->embedding.resize(hparams.n_embd);
+        // create ggml tensor
+        ggml_type wtype;
+        switch (ftype) {
+            case 0: wtype = GGML_TYPE_F32;  break;
+            case 1: wtype = GGML_TYPE_F16;  break;
+            default:
+                    {
+                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
+                                __func__, ftype);
+                        return false;
+                    }
         }
+        ggml_tensor * lora_tensor;
+        if (n_dims == 2) {
+            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
+        }
+        else {
+            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
+            return 1;
+        }
+        ggml_set_name(lora_tensor, "lora_tensor");
 
-#ifdef LLAMA_USE_ALLOCATOR
-        {
-            static const size_t tensor_alignment = 32;
-            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
-            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
+        // load tensor data
+        size_t offset = fin.tellg();
+        size_t tensor_data_size = ggml_nbytes(lora_tensor);
+        offset = (offset + 31) & -32;
+        fin.seekg(offset);
+        fin.read((char*)lora_tensor->data, tensor_data_size);
 
-            // create measure allocator
-            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
+        lora_tensors[name] = lora_tensor;
 
-            // build worst-case graph
-            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
-            int n_past = hparams.n_ctx - n_tokens;
-            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
-            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
+        // check if we have both A and B tensors and apply
+        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
+            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
 
-            // measure memory requirements for the graph
-            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
+            ggml_tensor * dest_t = model_tensors[base_name];
 
-            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
+            offload_func_t offload_func = llama_nop;
+            offload_func_t offload_func_force_inplace = llama_nop;
 
-            // debug - for comparison with scratch buffer
-            //size_t prev_req =
-            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
-            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
-            //    MEM_REQ_EVAL().at(ctx->model.type);
-            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
+#ifdef GGML_USE_CUBLAS
+            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
+                if (dest_t->type != GGML_TYPE_F16) {
+                    throw std::runtime_error(format(
+                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
+                }
+                offload_func = ggml_cuda_assign_buffers;
+                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
+            }
+#endif // GGML_USE_CUBLAS
 
-            // recreate allocator with exact memory requirements
-            ggml_allocr_free(ctx->alloc);
+            ggml_tensor * base_t;
+            if (model_loader) {
+                struct gguf_context * ctx_gguf = model_loader->ctx_gguf;
 
-            ctx->buf_alloc.resize(alloc_size);
-            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.addr, ctx->buf_alloc.size, tensor_alignment);
-        }
-#else
-        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
-#endif
+                // load from base model
+                if (gguf_find_tensor(ctx_gguf, base_name.c_str()) < 0) {
+                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
+                    return 1;
+                }
 
-#ifdef LLAMA_USE_SCRATCH
-        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
-        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
-#endif
-    }
+                // TODO: not tested!! maybe not working!
+                base_t = model_loader->create_tensor(base_ctx, base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
+                model_loader->load_data_for(base_t);
+            } else {
+                base_t = dest_t;
+            }
 
-#ifdef GGML_USE_METAL
-    if (params.n_gpu_layers > 0) {
-        // this allocates all Metal resources and memory buffers
-        ctx->ctx_metal = ggml_metal_init(1);
+            if (ggml_is_quantized(base_t->type)) {
+                if (!warned) {
+                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
+                                   "use a f16 or f32 base model with --lora-base\n", __func__);
+                    warned = true;
+                }
+            }
 
-        if (!ctx->ctx_metal) {
-            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
-            llama_free(ctx);
-            return NULL;
-        }
+            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
+            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
+            ggml_set_name(loraA, "loraA");
 
-        void * data_ptr  = NULL;
-        size_t data_size = 0;
+            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
+            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
+            ggml_set_name(loraB, "loraB");
 
-        if (params.use_mmap) {
-            data_ptr  = ctx->model.mapping->addr;
-            data_size = ctx->model.mapping->size;
-        } else {
-            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
-            data_size = ggml_get_mem_size  (ctx->model.ctx);
-        }
+            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
+                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
+                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
+                return 1;
+            }
 
-        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
+            // w = w + BA*s
+            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
+            offload_func(BA);
+            ggml_set_name(BA, "BA");
 
-        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
+            if (scaling != 1.0f) {
+                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
+                ggml_set_name(scale_tensor, "scale_tensor");
 
-#define LLAMA_METAL_CHECK_BUF(result)                            \
-    if (!(result)) {                                             \
-        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
-        llama_free(ctx);                                         \
-        return NULL;                                             \
-    }
+                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
+                offload_func(BA);
+                ggml_set_name(BA, "BA_scaled");
+            }
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
+            ggml_tensor * r;
+            if (base_t == dest_t) {
+                r = ggml_add_inplace(lora_ctx, dest_t, BA);
+                offload_func_force_inplace(r);
+                ggml_set_name(r, "r_add_inplace");
+            }
+            else {
+                r = ggml_add(lora_ctx, base_t, BA);
+                offload_func(r);
+                ggml_set_name(r, "r_add");
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.addr, ctx->buf_compute.size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.addr, ctx->kv_self.buf.size, 0));
+                r = ggml_cpy(lora_ctx, r, dest_t);
+                offload_func(r);
+                ggml_set_name(r, "r_cpy");
+            }
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].addr, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].addr, ctx->buf_scratch[1].size, 0));
-#undef LLAMA_METAL_CHECK_BUF
-    }
-#endif
+            struct ggml_cgraph gf = ggml_build_forward(r);
 
-#ifdef GGML_USE_MPI
-    ctx->ctx_mpi = ggml_mpi_init();
+            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
 
-    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
-        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
-        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
-        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
-        llama_backend_free();
-        exit(1);
+            // we won't need these tensors again, reset the context to save memory
+            ggml_free(lora_ctx);
+            lora_ctx = ggml_init(params);
+            lora_tensors.clear();
+
+            n_tensors++;
+            if (n_tensors % 4 == 0) {
+                LLAMA_LOG_INFO(".");
+            }
+        }
     }
-#endif
 
-    return ctx;
+    // TODO: this should be in a destructor, it will leak on failure
+    ggml_free(lora_ctx);
+    if (base_ctx) {
+        ggml_free(base_ctx);
+    }
+
+    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
+    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
+
+    return 0;
 }
 
-struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params) {
+//
+// interface implementation
+//
 
-    struct llama_model * model = llama_load_model_from_file(path_model, params);
-    if (!model) {
-        return nullptr;
-    }
-    struct llama_context * ctx = llama_new_context_with_model(model, params);
-    ctx->model_owner = true;
-    return ctx;
+struct llama_context_params llama_context_default_params() {
+    struct llama_context_params result = {
+        /*.seed                        =*/ LLAMA_DEFAULT_SEED,
+        /*.n_ctx                       =*/ 512,
+        /*.n_batch                     =*/ 512,
+        /*.gpu_layers                  =*/ 0,
+        /*.main_gpu                    =*/ 0,
+        /*.tensor_split                =*/ nullptr,
+        /*.rope_freq_base              =*/ 10000.0f,
+        /*.rope_freq_scale             =*/ 1.0f,
+        /*.progress_callback           =*/ nullptr,
+        /*.progress_callback_user_data =*/ nullptr,
+        /*.low_vram                    =*/ false,
+        /*.mul_mat_q                   =*/ false,
+        /*.f16_kv                      =*/ true,
+        /*.logits_all                  =*/ false,
+        /*.vocab_only                  =*/ false,
+        /*.use_mmap                    =*/ true,
+        /*.use_mlock                   =*/ false,
+        /*.embedding                   =*/ false,
+    };
+
+    return result;
 }
 
-void llama_free(struct llama_context * ctx) {
-    delete ctx;
+struct llama_model_quantize_params llama_model_quantize_default_params() {
+    struct llama_model_quantize_params result = {
+        /*.nthread                     =*/ 0,
+        /*.ftype                       =*/ LLAMA_FTYPE_MOSTLY_Q5_1,
+        /*.allow_requantize            =*/ false,
+        /*.quantize_output_tensor      =*/ true,
+    };
+
+    return result;
 }
 
-int llama_model_quantize(
-        const char * fname_inp,
-        const char * fname_out,
-        const llama_model_quantize_params *params) {
-    try {
-        llama_model_quantize_internal(fname_inp, fname_out, params);
-        return 0;
-    } catch (const std::exception & err) {
-        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
-        return 1;
-    }
+int llama_max_devices(void) {
+    return LLAMA_MAX_DEVICES;
 }
 
-int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
-    LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
+bool llama_mmap_supported(void) {
+    return llama_mmap::SUPPORTED;
+}
 
-    const int64_t t_start_lora_us = ggml_time_us();
+bool llama_mlock_supported(void) {
+    return llama_mlock::SUPPORTED;
+}
 
-    auto fin = std::ifstream(path_lora, std::ios::binary);
-    if (!fin) {
-        LLAMA_LOG_ERROR("%s: failed to open '%s'\n", __func__, path_lora);
-        return 1;
-    }
+void llama_backend_init(bool numa) {
+    ggml_time_init();
 
-    // verify magic and version
+    // needed to initialize f16 tables
     {
-        uint32_t magic;
-        fin.read((char *) &magic, sizeof(magic));
-        if (magic != LLAMA_FILE_MAGIC_GGLA) {
-            LLAMA_LOG_ERROR("%s: bad file magic\n", __func__);
-            return 1;
-        }
-        uint32_t format_version;
-        fin.read((char *) &format_version, sizeof(format_version));
-
-        if (format_version != 1) {
-            LLAMA_LOG_ERROR("%s: unsupported file version\n", __func__ );
-            return 1;
-        }
+        struct ggml_init_params params = { 0, NULL, false };
+        struct ggml_context * ctx = ggml_init(params);
+        ggml_free(ctx);
     }
 
-    int32_t lora_r;
-    int32_t lora_alpha;
-    fin.read((char *) &lora_r, sizeof(lora_r));
-    fin.read((char *) &lora_alpha, sizeof(lora_alpha));
-    float scaling = (float)lora_alpha / (float)lora_r;
-
-    LLAMA_LOG_INFO("%s: r = %d, alpha = %d, scaling = %.2f\n", __func__, lora_r, lora_alpha, scaling);
+    if (numa) {
+        ggml_numa_init();
+    }
 
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_init();
+#endif
+}
 
-    // create a temporary ggml context to store the lora tensors
-    // todo: calculate size from biggest possible tensor
-    std::vector<uint8_t> lora_buf(1024ull * 1024ull * 1024ull);
-    struct ggml_init_params params;
-    params.mem_size   = lora_buf.size();
-    params.mem_buffer = lora_buf.data();
-    params.no_alloc   = false;
+void llama_backend_free(void) {
+#ifdef GGML_USE_MPI
+    ggml_mpi_backend_free();
+#endif
+}
 
-    ggml_context * lora_ctx = ggml_init(params);
-    std::unordered_map<std::string, struct ggml_tensor *> lora_tensors;
+int64_t llama_time_us(void) {
+    return ggml_time_us();
+}
 
-    // create a name -> tensor map of the model to accelerate lookups
-    std::unordered_map<std::string, struct ggml_tensor*> model_tensors;
-    for (const auto & kv: model.tensors_by_name) {
-        model_tensors.insert(kv);
-    }
+struct llama_model * llama_load_model_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+    ggml_time_init();
 
+    llama_model * model = new llama_model;
 
-    // load base model
-    std::unique_ptr<llama_model_loader> model_loader;
-    ggml_context * base_ctx = NULL;
-    llama_buffer base_buf;
-    if (path_base_model) {
-        LLAMA_LOG_INFO("%s: loading base model from '%s'\n", __func__, path_base_model);
-        model_loader.reset(new llama_model_loader(path_base_model, /*use_mmap*/ true));
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
 
-        size_t ctx_size;
-        size_t mmapped_size;
-        model_loader->calc_sizes(&ctx_size, &mmapped_size);
-        base_buf.resize(ctx_size);
+    if (!llama_model_load(path_model, *model, model->vocab, params.n_ctx, params.n_batch, params.n_gpu_layers,
+                params.main_gpu, params.tensor_split, params.mul_mat_q, params.rope_freq_base, params.rope_freq_scale,
+                params.low_vram, memory_type, params.use_mmap, params.use_mlock, params.vocab_only,
+                params.progress_callback, params.progress_callback_user_data)) {
+        LLAMA_LOG_ERROR("%s: failed to load model\n", __func__);
+        delete model;
+        return nullptr;
+    }
 
-        ggml_init_params base_params;
-        base_params.mem_size   = base_buf.size;
-        base_params.mem_buffer = base_buf.addr;
-        base_params.no_alloc   = model_loader->use_mmap;
+    return model;
+}
 
-        base_ctx = ggml_init(base_params);
+void llama_free_model(struct llama_model * model) {
+    delete model;
+}
 
-        model_loader->ggml_ctx = base_ctx;
+struct llama_context * llama_new_context_with_model(
+                 struct llama_model * model,
+        struct llama_context_params   params) {
 
-        // maybe this should in llama_model_loader
-        if (model_loader->use_mmap) {
-            model_loader->mapping.reset(new llama_mmap(&model_loader->file_loader->file, /* prefetch */ 0, ggml_is_numa()));
-        }
+    if (!model) {
+        return nullptr;
     }
 
-    // read tensors and apply
-    bool warned = false;
-    int n_tensors = 0;
+    llama_context * ctx = new llama_context(*model);
 
-    std::vector<uint8_t> work_buffer;
+    if (params.seed == LLAMA_DEFAULT_SEED) {
+        params.seed = time(NULL);
+    }
 
-    while (true) {
-        int32_t n_dims;
-        int32_t length;
-        int32_t ftype;
+    unsigned cur_percentage = 0;
+    if (params.progress_callback == NULL) {
+        params.progress_callback_user_data = &cur_percentage;
+        params.progress_callback = [](float progress, void * ctx) {
+            unsigned * cur_percentage_p = (unsigned *) ctx;
+            unsigned percentage = (unsigned) (100 * progress);
+            while (percentage > *cur_percentage_p) {
+                *cur_percentage_p = percentage;
+                LLAMA_LOG_INFO(".");
+                if (percentage >= 100) {
+                    LLAMA_LOG_INFO("\n");
+                }
+            }
+        };
+    }
 
-        fin.read(reinterpret_cast<char *>(&n_dims), sizeof(n_dims));
-        fin.read(reinterpret_cast<char *>(&length), sizeof(length));
-        fin.read(reinterpret_cast<char *>(&ftype),  sizeof(ftype));
-        if (fin.eof()) {
-            break;
-        }
+    ctx->rng = std::mt19937(params.seed);
+    ctx->logits_all = params.logits_all;
 
-        int32_t ne[2] = { 1, 1 };
-        for (int i = 0; i < n_dims; ++i) {
-            fin.read(reinterpret_cast<char *>(&ne[i]), sizeof(ne[i]));
-        }
+    ggml_type memory_type = params.f16_kv ? GGML_TYPE_F16 : GGML_TYPE_F32;
 
-        std::string name;
-        {
-            char buf[1024];
-            fin.read(buf, length);
-            name = std::string(buf, length);
+    // reserve memory for context buffers
+    if (!params.vocab_only) {
+        if (!llama_kv_cache_init(ctx->model.hparams, ctx->kv_self, memory_type, ctx->model.hparams.n_ctx, params.n_gpu_layers)) {
+            LLAMA_LOG_ERROR("%s: llama_kv_cache_init() failed for self-attention cache\n", __func__);
+            llama_free(ctx);
+            return nullptr;
         }
 
-        // check for lora suffix and get the type of tensor
-        const std::string lora_suffix = ".lora";
-        size_t pos = name.rfind(lora_suffix);
-        if (pos == std::string::npos) {
-            LLAMA_LOG_ERROR("%s: error: '%s' is not a lora tensor\n", __func__, name.c_str());
-            return 1;
+        {
+            const size_t memory_size = ggml_nbytes(ctx->kv_self.k) + ggml_nbytes(ctx->kv_self.v);
+            LLAMA_LOG_INFO("%s: kv self size  = %7.2f MB\n", __func__, memory_size / 1024.0 / 1024.0);
         }
 
-        std::string lora_type = name.substr(pos + lora_suffix.length());
-        std::string base_name = name;
-        base_name.erase(pos);
-        // LLAMA_LOG_INFO("%s: %s => %s (lora type %s) \n", __func__, name.c_str(),base_name.c_str(), lora_type.c_str());
+        const auto & hparams = ctx->model.hparams;
 
-        if (model_tensors.find(base_name) == model_tensors.end()) {
-            LLAMA_LOG_ERROR("%s: unknown tensor '%s' in lora adapter\n", __func__, name.data());
-            return 1;
+        // resized during inference
+        if (params.logits_all) {
+            ctx->logits.reserve(hparams.n_ctx*hparams.n_vocab);
+        } else {
+            ctx->logits.reserve(hparams.n_vocab);
         }
 
-        // create ggml tensor
-        ggml_type wtype;
-        switch (ftype) {
-            case 0: wtype = GGML_TYPE_F32;  break;
-            case 1: wtype = GGML_TYPE_F16;  break;
-            default:
-                    {
-                        LLAMA_LOG_ERROR("%s: invalid tensor data type '%d'\n",
-                                __func__, ftype);
-                        return false;
-                    }
-        }
-        ggml_tensor * lora_tensor;
-        if (n_dims == 2) {
-            lora_tensor = ggml_new_tensor_2d(lora_ctx, wtype, ne[0], ne[1]);
-        }
-        else {
-            LLAMA_LOG_ERROR("%s: unsupported tensor dimension %d\n", __func__, n_dims);
-            return 1;
+        if (params.embedding){
+            ctx->embedding.resize(hparams.n_embd);
         }
-        ggml_set_name(lora_tensor, "lora_tensor");
 
-        // load tensor data
-        size_t offset = fin.tellg();
-        size_t tensor_data_size = ggml_nbytes(lora_tensor);
-        offset = (offset + 31) & -32;
-        fin.seekg(offset);
-        fin.read((char*)lora_tensor->data, tensor_data_size);
+#ifdef LLAMA_USE_ALLOCATOR
+        {
+            static const size_t tensor_alignment = 32;
+            // the compute buffer is used to store the tensor and graph structs, while the allocator buffer is used for the tensor data
+            ctx->buf_compute.resize(ggml_tensor_overhead()*GGML_MAX_NODES + ggml_graph_overhead());
 
-        lora_tensors[name] = lora_tensor;
+            // create measure allocator
+            ctx->alloc = ggml_allocr_new_measure(tensor_alignment);
 
-        // check if we have both A and B tensors and apply
-        if (lora_tensors.find(base_name + ".loraA") != lora_tensors.end() &&
-            lora_tensors.find(base_name + ".loraB") != lora_tensors.end()) {
+            // build worst-case graph
+            int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
+            int n_past = hparams.n_ctx - n_tokens;
+            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
+            ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
 
-            ggml_tensor * dest_t = model_tensors[base_name];
+            // measure memory requirements for the graph
+            size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
 
-            offload_func_t offload_func = llama_nop;
-            offload_func_t offload_func_force_inplace = llama_nop;
+            LLAMA_LOG_INFO("%s: compute buffer total size = %7.2f MB\n", __func__, (ctx->buf_compute.size + alloc_size) / 1024.0 / 1024.0);
 
-#ifdef GGML_USE_CUBLAS
-            if (dest_t->backend == GGML_BACKEND_GPU || dest_t->backend == GGML_BACKEND_GPU_SPLIT) {
-                if (dest_t->type != GGML_TYPE_F16) {
-                    throw std::runtime_error(format(
-                        "%s: error: the simultaneous use of LoRAs and GPU acceleration is only supported for f16 models", __func__));
-                }
-                offload_func = ggml_cuda_assign_buffers;
-                offload_func_force_inplace = ggml_cuda_assign_buffers_force_inplace;
-            }
-#endif // GGML_USE_CUBLAS
+            // debug - for comparison with scratch buffer
+            //size_t prev_req =
+            //    MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type) +
+            //    MEM_REQ_SCRATCH1().at(ctx->model.type) +
+            //    MEM_REQ_EVAL().at(ctx->model.type);
+            //LLAMA_LOG_INFO("%s: (debug) equivalent with scratch buffer = %7.2f MB\n", __func__, prev_req / 1024.0 / 1024.0);
 
-            ggml_tensor * base_t;
-            if (model_loader) {
-                // load from base model
-                if (model_loader->tensors_map.name_to_idx.find(base_name) == model_loader->tensors_map.name_to_idx.end()) {
-                    LLAMA_LOG_ERROR("%s: error: tensor '%s' not found in base model\n", __func__, base_name.c_str());
-                    return 1;
-                }
-                size_t idx = model_loader->tensors_map.name_to_idx[base_name];
-                llama_load_tensor & lt = model_loader->tensors_map.tensors[idx];
-                base_t = model_loader->get_tensor(base_name, { (uint32_t)dest_t->ne[0], (uint32_t)dest_t->ne[1] }, GGML_BACKEND_CPU);
-                lt.data = (uint8_t *) lt.ggml_tensor->data;
-                model_loader->load_data_for(lt);
-                lt.ggml_tensor->data = lt.data;
-            }
-            else {
-                base_t = dest_t;
-            }
+            // recreate allocator with exact memory requirements
+            ggml_allocr_free(ctx->alloc);
 
-            if (ggml_is_quantized(base_t->type)) {
-                if (!warned) {
-                    LLAMA_LOG_WARN("%s: warning: using a lora adapter with a quantized model may result in poor quality, "
-                                   "use a f16 or f32 base model with --lora-base\n", __func__);
-                    warned = true;
-                }
-            }
+            ctx->buf_alloc.resize(alloc_size);
+            ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
+        }
+#else
+        ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
+#endif
 
-            ggml_tensor * loraA = lora_tensors[base_name + ".loraA"];
-            GGML_ASSERT(loraA->type == GGML_TYPE_F32);
-            ggml_set_name(loraA, "loraA");
+#ifdef LLAMA_USE_SCRATCH
+        ctx->buf_scratch[0].resize(MEM_REQ_SCRATCH0(hparams.n_ctx).at(ctx->model.type));
+        ctx->buf_scratch[1].resize(MEM_REQ_SCRATCH1().at(ctx->model.type));
+#endif
+    }
 
-            ggml_tensor * loraB = lora_tensors[base_name + ".loraB"];
-            GGML_ASSERT(loraB->type == GGML_TYPE_F32);
-            ggml_set_name(loraB, "loraB");
+#ifdef GGML_USE_METAL
+    if (params.n_gpu_layers > 0) {
+        // this allocates all Metal resources and memory buffers
+        ctx->ctx_metal = ggml_metal_init(1);
 
-            if (base_t->ne[0] != loraA->ne[1] || base_t->ne[1] != loraB->ne[1]) {
-                LLAMA_LOG_ERROR("%s: incompatible tensor dimensions (%" PRId64 " and %" PRId64 ");"
-                                " are you sure that this adapter is for this model?\n", __func__, base_t->ne[0], loraA->ne[1]);
-                return 1;
-            }
+        if (!ctx->ctx_metal) {
+            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
+            llama_free(ctx);
+            return NULL;
+        }
 
-            // w = w + BA*s
-            ggml_tensor * BA = ggml_mul_mat(lora_ctx, loraA, loraB);
-            offload_func(BA);
-            ggml_set_name(BA, "BA");
+        void * data_ptr  = NULL;
+        size_t data_size = 0;
 
-            if (scaling != 1.0f) {
-                ggml_tensor * scale_tensor = ggml_new_f32(lora_ctx, scaling);
-                ggml_set_name(scale_tensor, "scale_tensor");
+        if (params.use_mmap) {
+            data_ptr  = ctx->model.mapping->addr;
+            data_size = ctx->model.mapping->size;
+        } else {
+            data_ptr  = ggml_get_mem_buffer(ctx->model.ctx);
+            data_size = ggml_get_mem_size  (ctx->model.ctx);
+        }
 
-                BA = ggml_scale_inplace(lora_ctx, BA, scale_tensor);
-                offload_func(BA);
-                ggml_set_name(BA, "BA_scaled");
-            }
+        const size_t max_size = ggml_get_max_tensor_size(ctx->model.ctx);
 
-            ggml_tensor * r;
-            if (base_t == dest_t) {
-                r = ggml_add_inplace(lora_ctx, dest_t, BA);
-                offload_func_force_inplace(r);
-                ggml_set_name(r, "r_add_inplace");
-            }
-            else {
-                r = ggml_add(lora_ctx, base_t, BA);
-                offload_func(r);
-                ggml_set_name(r, "r_add");
+        LLAMA_LOG_INFO("%s: max tensor size = %8.2f MB\n", __func__, max_size/1024.0/1024.0);
 
-                r = ggml_cpy(lora_ctx, r, dest_t);
-                offload_func(r);
-                ggml_set_name(r, "r_cpy");
-            }
+#define LLAMA_METAL_CHECK_BUF(result)                            \
+    if (!(result)) {                                             \
+        LLAMA_LOG_ERROR("%s: failed to add buffer\n", __func__); \
+        llama_free(ctx);                                         \
+        return NULL;                                             \
+    }
 
-            struct ggml_cgraph gf = ggml_build_forward(r);
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "data", data_ptr, data_size, max_size));
 
-            ggml_graph_compute_helper(work_buffer, &gf, n_threads);
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
 
-            // we won't need these tensors again, reset the context to save memory
-            ggml_free(lora_ctx);
-            lora_ctx = ggml_init(params);
-            lora_tensors.clear();
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
+#undef LLAMA_METAL_CHECK_BUF
+    }
+#endif
 
-            n_tensors++;
-            if (n_tensors % 4 == 0) {
-                LLAMA_LOG_INFO(".");
-            }
-        }
+#ifdef GGML_USE_MPI
+    ctx->ctx_mpi = ggml_mpi_init();
+
+    if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
+        // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
+        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
+        while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
+        llama_backend_free();
+        exit(1);
     }
+#endif
 
-    // TODO: this should be in a destructor, it will leak on failure
-    ggml_free(lora_ctx);
-    if (base_ctx) {
-        ggml_free(base_ctx);
+    return ctx;
+}
+
+struct llama_context * llama_init_from_file(
+                             const char * path_model,
+            struct llama_context_params   params) {
+
+    struct llama_model * model = llama_load_model_from_file(path_model, params);
+    if (!model) {
+        return nullptr;
     }
+    struct llama_context * ctx = llama_new_context_with_model(model, params);
+    ctx->model_owner = true;
+    return ctx;
+}
 
-    const int64_t t_lora_us = ggml_time_us() - t_start_lora_us;
-    LLAMA_LOG_INFO(" done (%.2f ms)\n", t_lora_us / 1000.0);
+void llama_free(struct llama_context * ctx) {
+    delete ctx;
+}
 
-    return 0;
+int llama_model_quantize(
+        const char * fname_inp,
+        const char * fname_out,
+        const llama_model_quantize_params * params) {
+    try {
+        llama_model_quantize_internal(fname_inp, fname_out, params);
+        return 0;
+    } catch (const std::exception & err) {
+        LLAMA_LOG_ERROR("%s: failed to quantize: %s\n", __func__, err.what());
+        return 1;
+    }
 }
 
 int llama_apply_lora_from_file(struct llama_context * ctx, const char * path_lora, const char * path_base_model, int n_threads) {
@@ -3908,6 +4293,46 @@ size_t llama_get_state_size(const struct llama_context * ctx) {
     return s_total;
 }
 
+// llama_context_data
+struct llama_data_context {
+    virtual void write(const void * src, size_t size) = 0;
+    virtual size_t get_size_written() = 0;
+    virtual ~llama_data_context() = default;
+};
+
+struct llama_data_buffer_context : llama_data_context {
+    uint8_t * ptr;
+    size_t size_written = 0;
+
+    llama_data_buffer_context(uint8_t * p) : ptr(p) {}
+
+    void write(const void * src, size_t size) override {
+        memcpy(ptr, src, size);
+        ptr += size;
+        size_written += size;
+    }
+
+    size_t get_size_written() override {
+        return size_written;
+    }
+};
+
+struct llama_data_file_context : llama_data_context {
+    llama_file * file;
+    size_t size_written = 0;
+
+    llama_data_file_context(llama_file * f) : file(f) {}
+
+    void write(const void * src, size_t size) override {
+        file->write_raw(src, size);
+        size_written += size;
+    }
+
+    size_t get_size_written() override {
+        return size_written;
+    }
+};
+
 /** copy state data into either a buffer or file depending on the passed in context
  *
  * file context:
@@ -4041,7 +4466,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
         rng_ss.str(std::string(&rng_buf[0], rng_size));
         rng_ss >> ctx->rng;
 
-        LLAMA_ASSERT(rng_ss.fail() == false);
+        GGML_ASSERT(rng_ss.fail() == false);
     }
 
     // set logits
@@ -4052,7 +4477,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
         memcpy(&logits_cap,  inp, sizeof(logits_cap));  inp += sizeof(logits_cap);
         memcpy(&logits_size, inp, sizeof(logits_size)); inp += sizeof(logits_size);
 
-        LLAMA_ASSERT(ctx->logits.capacity() == logits_cap);
+        GGML_ASSERT(ctx->logits.capacity() == logits_cap);
 
         if (logits_size) {
             ctx->logits.resize(logits_size);
@@ -4068,7 +4493,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
 
         memcpy(&embedding_size, inp, sizeof(embedding_size)); inp += sizeof(embedding_size);
 
-        LLAMA_ASSERT(ctx->embedding.capacity() == embedding_size);
+        GGML_ASSERT(ctx->embedding.capacity() == embedding_size);
 
         if (embedding_size) {
             memcpy(ctx->embedding.data(), inp, embedding_size * sizeof(float));
@@ -4091,7 +4516,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
         memcpy(&kv_ntok, inp, sizeof(kv_ntok)); inp += sizeof(kv_ntok);
 
         if (kv_size) {
-            LLAMA_ASSERT(kv_self.buf.size == kv_size);
+            GGML_ASSERT(kv_self.buf.size == kv_size);
 
             const size_t elt_size = ggml_element_size(kv_self.k);
 
@@ -4127,7 +4552,7 @@ size_t llama_set_state_data(struct llama_context * ctx, uint8_t * src) {
     const size_t nread    = inp - src;
     const size_t max_size = llama_get_state_size(ctx);
 
-    LLAMA_ASSERT(nread <= max_size);
+    GGML_ASSERT(nread <= max_size);
 
     return nread;
 }
@@ -4235,7 +4660,6 @@ int llama_eval(
     return 0;
 }
 
-
 int llama_eval_embd(
             struct llama_context * ctx,
                      const float * embd,
@@ -4410,23 +4834,26 @@ float * llama_get_embeddings(struct llama_context * ctx) {
     return ctx->embedding.data();
 }
 
-int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * str, int length) {
+// does not write null-terminator to str
+int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * buf, int length) {
     if (0 <= token && token < llama_n_vocab_from_model(model)) {
         if (llama_is_normal_token(model->vocab, token)) {
             std::string result = model->vocab.id_to_token[token].tok;
-            if(llama_vocab_type(model->vocab) == "spm") {
+            if (llama_vocab_type(model->vocab) == "spm") {
                 result = llama_unescape_whitespace(result);
             }
             if (length < (int) result.length()) {
                 return -result.length();
             }
-            strncpy(str, result.c_str(), result.length());
+            memcpy(buf, result.c_str(), result.length());
             return result.length();
-        } else if (llama_is_unknown_token(model->vocab, token)) {
+        } else if (llama_is_unknown_token(model->vocab, token)) { // NOLINT
             if (length < 3) {
                 return -3;
             }
-            strncpy(str, "\xe2\x96\x85", 3);
+            buf[0] = '\xe2';
+            buf[1] = '\x96';
+            buf[2] = '\x85';
             return 3;
         } else if (llama_is_control_token(model->vocab, token)) {
             ;
@@ -4434,8 +4861,7 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
             if (length < 1) {
                 return -1;
             }
-            str[0] = llama_byte_to_char(model->vocab, token);
-            str[1] = 0x00;
+            buf[0] = llama_byte_to_char(model->vocab, token);
             return 1;
         }
     }
@@ -4466,7 +4892,7 @@ int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token,
         if (length < (int) result.length()) {
             return -result.length();
         }
-        strncpy(str, result.c_str(), result.length());
+        memcpy(str, result.c_str(), result.length());
         return result.length();
     }
     return 0;
@@ -4486,15 +4912,15 @@ std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token
     return std::string(result.data(), result.size());
 }
 
-llama_token llama_token_bos() {
+llama_token llama_token_bos(void) {
     return 1;
 }
 
-llama_token llama_token_eos() {
+llama_token llama_token_eos(void) {
     return 2;
 }
 
-llama_token llama_token_nl() {
+llama_token llama_token_nl(void) {
     return 13;
 }
 
@@ -4563,7 +4989,6 @@ const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_
     return ctx->model.tensors_by_name;
 }
 
-
 void llama_log_set(llama_log_callback log_callback, void * user_data) {
     g_state.log_callback = log_callback ? log_callback : llama_log_callback_default;
     g_state.log_callback_user_data = user_data;
diff --git a/llama.h b/llama.h
index 6ed1c4b916c68..2e407a1db3ca3 100644
--- a/llama.h
+++ b/llama.h
@@ -34,29 +34,18 @@
 #    define DEPRECATED(func, hint) func
 #endif
 
-#define LLAMA_FILE_MAGIC_GGJT        0x67676a74u // 'ggjt'
-#define LLAMA_FILE_MAGIC_GGLA        0x67676c61u // 'ggla'
-#define LLAMA_FILE_MAGIC_GGMF        0x67676d66u // 'ggmf'
-#define LLAMA_FILE_MAGIC_GGML        0x67676d6cu // 'ggml'
-#define LLAMA_FILE_MAGIC_GGSN        0x6767736eu // 'ggsn'
+#define LLAMA_DEFAULT_SEED 0xFFFFFFFF
 
-#define LLAMA_FILE_VERSION           3
-#define LLAMA_FILE_MAGIC             LLAMA_FILE_MAGIC_GGJT
-#define LLAMA_FILE_MAGIC_UNVERSIONED LLAMA_FILE_MAGIC_GGML
-#define LLAMA_SESSION_MAGIC          LLAMA_FILE_MAGIC_GGSN
-#define LLAMA_SESSION_VERSION        1
+#define LLAMA_FILE_MAGIC_GGSN 0x6767736eu // 'ggsn'
 
-#define LLAMA_DEFAULT_SEED           0xFFFFFFFF
+#define LLAMA_SESSION_MAGIC   LLAMA_FILE_MAGIC_GGSN
+#define LLAMA_SESSION_VERSION 1
 
 #if defined(GGML_USE_CUBLAS) || defined(GGML_USE_CLBLAST) || defined(GGML_USE_METAL)
 // Defined when llama.cpp is compiled with support for offloading model layers to GPU.
 #define LLAMA_SUPPORTS_GPU_OFFLOAD
 #endif
 
-#ifndef LLAMA_DEFAULT_RMS_EPS
-#define LLAMA_DEFAULT_RMS_EPS 5e-6f
-#endif
-
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -103,8 +92,6 @@ extern "C" {
         uint32_t seed;         // RNG seed, -1 for random
         int32_t  n_ctx;        // text context
         int32_t  n_batch;      // prompt processing batch size
-        int32_t  n_gqa;        // grouped-query attention (TEMP - will be moved to model hparams)
-        float    rms_norm_eps; // rms norm epsilon (TEMP - will be moved to model hparams)
         int32_t  n_gpu_layers; // number of layers to store in VRAM
         int32_t  main_gpu;     // the GPU that is used for scratch and small tensors
 
@@ -129,6 +116,7 @@ extern "C" {
         bool use_mlock;  // force system to keep model in RAM
         bool embedding;  // embedding mode only
     };
+
     // model file types
     enum llama_ftype {
         LLAMA_FTYPE_ALL_F32              = 0,
@@ -155,7 +143,7 @@ extern "C" {
     // model quantization parameters
     typedef struct llama_model_quantize_params {
         int nthread;                 // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
-        enum llama_ftype   ftype;    // quantize to this llama_ftype
+        enum llama_ftype ftype;      // quantize to this llama_ftype
         bool allow_requantize;       // allow quantizing non-f32/f16 tensors
         bool quantize_output_tensor; // quantize output.weight
     } llama_model_quantize_params;
@@ -208,17 +196,12 @@ extern "C" {
         int32_t n_eval;
     };
 
-    // Set callback for all future logging events.
-    // If this is not called, or NULL is supplied, everything is output on stderr.
-    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
-
-    LLAMA_API int llama_max_devices();
+    LLAMA_API struct llama_context_params llama_context_default_params(void);
+    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
 
-    LLAMA_API struct llama_context_params llama_context_default_params();
-    LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params();
-
-    LLAMA_API bool llama_mmap_supported();
-    LLAMA_API bool llama_mlock_supported();
+    LLAMA_API int  llama_max_devices(void);
+    LLAMA_API bool llama_mmap_supported(void);
+    LLAMA_API bool llama_mlock_supported(void);
 
     // TODO: not great API - very likely to change
     // Initialize the llama + ggml backend
@@ -226,9 +209,9 @@ extern "C" {
     // Call once at the start of the program
     LLAMA_API void llama_backend_init(bool numa);
     // Call once at the end of the program - currently only used for MPI
-    LLAMA_API void llama_backend_free();
+    LLAMA_API void llama_backend_free(void);
 
-    LLAMA_API int64_t llama_time_us();
+    LLAMA_API int64_t llama_time_us(void);
 
     LLAMA_API struct llama_model * llama_load_model_from_file(
                              const char * path_model,
@@ -240,13 +223,6 @@ extern "C" {
                      struct llama_model * model,
             struct llama_context_params   params);
 
-    // Various functions for loading a ggml llama model.
-    // Allocate (almost) all memory needed for the model.
-    // Return NULL on failure
-    LLAMA_API DEPRECATED(struct llama_context * llama_init_from_file(
-                             const char * path_model,
-            struct llama_context_params   params),
-            "please use llama_load_model_from_file combined with llama_new_context_with_model instead");
 
     // Frees all allocated memory
     LLAMA_API void llama_free(struct llama_context * ctx);
@@ -384,27 +360,28 @@ extern "C" {
     LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
 
     // Token Id -> String. Uses the vocabulary in the provided context
+    // Does not write null terminator to the buffer
     LLAMA_API int llama_token_to_str(
             const struct llama_context * ctx,
                            llama_token   token,
-                                  char * str,
+                                  char * buf,
                                   int    length);
 
     LLAMA_API int llama_token_to_str_bpe(
             const struct llama_context * ctx,
                            llama_token   token,
-                                  char * str,
+                                  char * buf,
                                   int    length);
 
     LLAMA_API int llama_token_to_str_with_model(
               const struct llama_model * model,
                            llama_token   token,
-                                  char * str,
+                                  char * buf,
                                   int    length);
     // Special tokens
-    LLAMA_API llama_token llama_token_bos();  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos();  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl();   // next-line
+    LLAMA_API llama_token llama_token_bos(void);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(void);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl(void);   // next-line
 
     // Grammar
     //
@@ -484,6 +461,10 @@ extern "C" {
     // Print system information
     LLAMA_API const char * llama_print_system_info(void);
 
+    // Set callback for all future logging events.
+    // If this is not called, or NULL is supplied, everything is output on stderr.
+    LLAMA_API void llama_log_set(llama_log_callback log_callback, void * user_data);
+
 #ifdef __cplusplus
 }
 #endif

From c8ee87f1411dfc48aa9c0d0d094b38b15b2a7827 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 16 Aug 2023 19:55:49 +0300
Subject: [PATCH 163/242] gguf.py : merge all files in gguf.py

---
 constants.py                    |  50 -------
 convert-gptneox-h5-to-gguf.py   |  10 +-
 convert-llama-7b-pth-to-gguf.py |   5 +-
 convert-llama-h5-to-gguf.py     |   4 +-
 gguf.py                         | 258 +++++++++++++++++++++++++-------
 gguf_namemap.py                 |  95 ------------
 6 files changed, 214 insertions(+), 208 deletions(-)
 delete mode 100644 constants.py
 delete mode 100644 gguf_namemap.py

diff --git a/constants.py b/constants.py
deleted file mode 100644
index 2f96fb9604e0c..0000000000000
--- a/constants.py
+++ /dev/null
@@ -1,50 +0,0 @@
-GGUF_MAGIC             = 0x47475546
-GGUF_VERSION           = 1
-GGUF_DEFAULT_ALIGNMENT = 32
-
-# general
-KEY_GENERAL_ARCHITECTURE         = "general.architecture"
-KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
-KEY_GENERAL_ALIGNMENT            = "general.alignment"
-KEY_GENERAL_NAME                 = "general.name"
-KEY_GENERAL_AUTHOR               = "general.author"
-KEY_GENERAL_URL                  = "general.url"
-KEY_GENERAL_DESCRIPTION          = "general.description"
-KEY_GENERAL_FILE_TYPE            = "general.file_type"
-KEY_GENERAL_LICENSE              = "general.license"
-KEY_GENERAL_SOURCE_URL           = "general.source.url"
-KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
-
-# LLM
-KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
-KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
-KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
-KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
-KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
-KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
-
-# attention
-KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
-KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
-KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
-KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
-KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
-KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
-
-# RoPE
-KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
-KEY_ROPE_SCALE                   = "{llm}.rope.scale"
-
-# tokenization
-KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
-KEY_TOKENIZER_LIST       = "tokenizer.ggml.tokens"
-KEY_TOKENIZER_TOKEN_TYPE = "tokenizer.ggml.token_type"
-KEY_TOKENIZER_SCORES     = "tokenizer.ggml.scores"
-KEY_TOKENIZER_MERGES     = "tokenizer.ggml.merges"
-KEY_TOKENIZER_BOS_ID     = "tokenizer.ggml.bos_token_id"
-KEY_TOKENIZER_EOS_ID     = "tokenizer.ggml.eos_token_id"
-KEY_TOKENIZER_UNK_ID     = "tokenizer.ggml.unknown_token_id"
-KEY_TOKENIZER_SEP_ID     = "tokenizer.ggml.seperator_token_id"
-KEY_TOKENIZER_PAD_ID     = "tokenizer.ggml.padding_token_id"
-KEY_TOKENIZER_HF_JSON    = "tokenizer.huggingface.json"
-KEY_TOKENIZER_RWKV       = "tokenizer.rwkv.world"
diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-h5-to-gguf.py
index a3d78c9ef2aa0..79876eee3e188 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-h5-to-gguf.py
@@ -1,15 +1,15 @@
 # HF gptneox--> gguf conversion
 
 import gguf
-import gguf_namemap as tmap
 import os
 import sys
 import struct
 import json
 import numpy as np
+import torch
+
 from typing import Any, List
 from pathlib import Path
-import torch
 from transformers import AutoTokenizer
 
 # ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
@@ -188,7 +188,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = tmap.get_tensor_namemap(block_count)
+tensor_map = gguf.get_tensor_name_map(block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
@@ -227,7 +227,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -292,7 +292,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 27841939dd4dd..c4e425ee32fa7 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -3,18 +3,17 @@
 # HF files required in the model dir: config.json tokenizer_config.json tokenizer.json tokenizer.model
 
 import gguf
-import gguf_namemap as tmap
 import os
 import sys
 import struct
 import json
 import numpy as np
 import torch
+
 from typing import Any, List
 from pathlib import Path
 from sentencepiece import SentencePieceProcessor
 
-
 #NDArray = np.ndarray[Any, Any]
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
@@ -189,7 +188,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = tmap.get_tensor_namemap(block_count)
+tensor_map = gguf.get_tensor_name_map(block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-h5-to-gguf.py
index 35f85bb978959..a2b3f9a3061fb 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-h5-to-gguf.py
@@ -1,8 +1,6 @@
 # HF llama --> gguf conversion
 
 import gguf
-import gguf_namemap as tmap
-
 import os
 import sys
 import struct
@@ -201,7 +199,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = tmap.get_tensor_namemap(block_count)
+tensor_map = gguf.get_tensor_name_map(block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
diff --git a/gguf.py b/gguf.py
index 1b15554f3ab76..e7f6f0ac89b96 100644
--- a/gguf.py
+++ b/gguf.py
@@ -4,14 +4,169 @@
 3. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
 """
 
+import sys
 import struct
-import constants
+import numpy as np
+
 from enum import IntEnum
 from typing import Any, IO, List
 
-import numpy as np
-import sys
-
+#
+# constants
+#
+
+GGUF_MAGIC             = 0x47475546
+GGUF_VERSION           = 1
+GGUF_DEFAULT_ALIGNMENT = 32
+
+# general
+KEY_GENERAL_ARCHITECTURE         = "general.architecture"
+KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version"
+KEY_GENERAL_ALIGNMENT            = "general.alignment"
+KEY_GENERAL_NAME                 = "general.name"
+KEY_GENERAL_AUTHOR               = "general.author"
+KEY_GENERAL_URL                  = "general.url"
+KEY_GENERAL_DESCRIPTION          = "general.description"
+KEY_GENERAL_FILE_TYPE            = "general.file_type"
+KEY_GENERAL_LICENSE              = "general.license"
+KEY_GENERAL_SOURCE_URL           = "general.source.url"
+KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
+
+# LLM
+KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
+KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
+KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
+KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
+
+# attention
+KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
+KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
+KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
+
+# RoPE
+KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
+KEY_ROPE_SCALE                   = "{llm}.rope.scale"
+
+# tokenization
+KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
+KEY_TOKENIZER_LIST       = "tokenizer.ggml.tokens"
+KEY_TOKENIZER_TOKEN_TYPE = "tokenizer.ggml.token_type"
+KEY_TOKENIZER_SCORES     = "tokenizer.ggml.scores"
+KEY_TOKENIZER_MERGES     = "tokenizer.ggml.merges"
+KEY_TOKENIZER_BOS_ID     = "tokenizer.ggml.bos_token_id"
+KEY_TOKENIZER_EOS_ID     = "tokenizer.ggml.eos_token_id"
+KEY_TOKENIZER_UNK_ID     = "tokenizer.ggml.unknown_token_id"
+KEY_TOKENIZER_SEP_ID     = "tokenizer.ggml.seperator_token_id"
+KEY_TOKENIZER_PAD_ID     = "tokenizer.ggml.padding_token_id"
+KEY_TOKENIZER_HF_JSON    = "tokenizer.huggingface.json"
+KEY_TOKENIZER_RWKV       = "tokenizer.rwkv.world"
+
+#
+# recommended mapping of model tensor names for storage in gguf
+#
+
+def get_tensor_name_map(n_blocks : int):
+    tensor_map = {}
+    # Token embeddings
+    mapped_to = "token_embd"
+    tensor_map["gpt_neox.embed_in"]           = mapped_to # gptneox
+    tensor_map["transformer.wte"]             = mapped_to # gpt2 mpt
+    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
+    tensor_map["model.embed_tokens"]          = mapped_to # llama-hf
+    tensor_map["tok_embeddings"]              = mapped_to # llama-pth
+    # Position embeddings
+    mapped_to = "pos_embd"
+    tensor_map["transformer.wpe"] = mapped_to # gpt2
+    # Output norm
+    mapped_to = "output_norm"
+    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
+    tensor_map["transformer.ln_f"]          = mapped_to # gpt2 falcon
+    tensor_map["transformer.norm_f"]        = mapped_to # mpt
+    tensor_map["model.norm"]                = mapped_to # llama-hf
+    tensor_map["norm"]                      = mapped_to # llama-pth
+    # Output
+    mapped_to = "output"
+    tensor_map["embed_out"] = mapped_to # gptneox
+    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
+    tensor_map["output"]    = mapped_to # llama-pth
+    # Attention and fee-forward layer blocks
+    for i in range(0,n_blocks):
+        # Attention norm
+        mapped_to = "blk."+str(i)+".attn_norm"
+        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to # llama-pth
+        # Attention norm 2
+        mapped_to = "blk."+str(i)+".attn_norm_2"
+        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
+        # Attention query-key-value
+        mapped_to = "blk."+str(i)+".attn_qkv"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon
+        # Attention query
+        mapped_to = "blk."+str(i)+".attn_q"
+        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to # llama-pth
+        # Attention key
+        mapped_to = "blk."+str(i)+".attn_k"
+        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to # llama-pth
+        # Attention value
+        mapped_to = "blk."+str(i)+".attn_v"
+        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to # llama-pth
+        # Attention output
+        mapped_to = "blk."+str(i)+".attn_output"
+        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to # llama-pth
+        # Feed-forward norm
+        mapped_to = "blk."+str(i)+".ffn_norm"
+        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to # mpt
+        tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to # llama-pth
+        # Feed-forward up
+        mapped_to = "blk."+str(i)+".ffn_up"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to # llama-pth
+        # Feed-forward gate
+        mapped_to = "blk."+str(i)+".ffn_gate"
+        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to # llama-pth
+        # Feed-forward down
+        mapped_to = "blk."+str(i)+".ffn_down"
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"]   = mapped_to # falcon
+        tensor_map["model.layers."+str(i)+".mlp.down_proj"]        = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w2"]            = mapped_to # llama-pth
+
+    return tensor_map
+
+#
+# implementation
+#
 
 class GGMLQuantizationType(IntEnum):
     F32 = 0
@@ -19,16 +174,16 @@ class GGMLQuantizationType(IntEnum):
 
 
 class GGUFValueType(IntEnum):
-    UINT8 = 0
-    INT8 = 1
-    UINT16 = 2
-    INT16 = 3
-    UINT32 = 4
-    INT32 = 5
+    UINT8   = 0
+    INT8    = 1
+    UINT16  = 2
+    INT16   = 3
+    UINT32  = 4
+    INT32   = 5
     FLOAT32 = 6
-    BOOL = 7
-    STRING = 8
-    ARRAY = 9
+    BOOL    = 7
+    STRING  = 8
+    ARRAY   = 9
 
     @staticmethod
     def get_type(val):
@@ -51,15 +206,15 @@ class GGUFWriter:
     def __init__(self, fout: IO):
         self.fout = fout
         self.offset_tensor = 0
-        self.data_alignment = constants.GGUF_DEFAULT_ALIGNMENT
+        self.data_alignment = GGUF_DEFAULT_ALIGNMENT
         self.kv_data = b""
         self.kv_data_count = 0
         self.ti_data = b""
         self.ti_data_count = 0
 
     def write_header_to_file(self):
-        self.fout.write(struct.pack("<I", constants.GGUF_MAGIC))
-        self.fout.write(struct.pack("<I", constants.GGUF_VERSION))
+        self.fout.write(struct.pack("<I", GGUF_MAGIC))
+        self.fout.write(struct.pack("<I", GGUF_VERSION))
         self.fout.write(struct.pack("<I", self.ti_data_count))
         self.fout.write(struct.pack("<I", self.kv_data_count))
         self.flush()
@@ -201,126 +356,125 @@ def close(self):
         self.fout.close()
 
     def add_architecture(self, architecture: str):
-        self.add_string(constants.KEY_GENERAL_ARCHITECTURE,
+        self.add_string(KEY_GENERAL_ARCHITECTURE,
                         architecture)
 
     def add_author(self, author: str):
-        self.add_string(constants.KEY_GENERAL_AUTHOR, author)
+        self.add_string(KEY_GENERAL_AUTHOR, author)
 
     def add_tensor_data_layout(self, layout: str):
-        self.add_string(constants.KEY_LLM_TENSOR_DATA_LAYOUT , layout)
+        self.add_string(KEY_LLM_TENSOR_DATA_LAYOUT , layout)
 
     def add_url(self, url: str):
-        self.add_string(constants.KEY_GENERAL_URL, url)
+        self.add_string(KEY_GENERAL_URL, url)
 
     def add_description(self, description: str):
-        self.add_string(constants.KEY_GENERAL_DESCRIPTION, description)
+        self.add_string(KEY_GENERAL_DESCRIPTION, description)
 
     def add_file_type(self, file_type: str):
-        self.add_string(constants.KEY_GENERAL_FILE_TYPE, file_type)
+        self.add_string(KEY_GENERAL_FILE_TYPE, file_type)
 
     def add_source_url(self, url: str):
-        self.add_string(constants.KEY_GENERAL_SOURCE_URL, url)
+        self.add_string(KEY_GENERAL_SOURCE_URL, url)
 
     def add_source_hf_repo(self, repo: str):
-        self.add_string(constants.KEY_GENERAL_SOURCE_HF_REPO, repo)
+        self.add_string(KEY_GENERAL_SOURCE_HF_REPO, repo)
 
     def add_name(self, name: str):
-        self.add_string(constants.KEY_GENERAL_NAME, name)
+        self.add_string(KEY_GENERAL_NAME, name)
 
     def add_quantization_version(self, quantization_version: GGMLQuantizationType):
         self.add_uint32(
-            constants.KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
+            KEY_GENERAL_QUANTIZATION_VERSION, quantization_version)
 
     def add_custom_alignment(self, alignment: int):
         self.data_alignment = alignment
-        self.add_uint32(constants.KEY_GENERAL_ALIGNMENT, alignment)
+        self.add_uint32(KEY_GENERAL_ALIGNMENT, alignment)
 
     def add_context_length(self, llm: str, length: int):
         self.add_uint32(
-            constants.KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
+            KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
 
     def add_embedding_length(self, llm: str, length: int):
         self.add_uint32(
-            constants.KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
+            KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
 
     def add_block_count(self, llm: str, length: int):
         self.add_uint32(
-            constants.KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
+            KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
 
     def add_feed_forward_length(self, llm: str, length: int):
         self.add_uint32(
-            constants.KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
+            KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
 
     def add_parallel_residual(self, llm: str, use: bool):
         self.add_bool(
-            constants.KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
+            KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
 
     def add_tensor_data_layout(self, llm: str, layout: str):
         self.add_string(
-            constants.KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
+            KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
 
     def add_head_count(self, llm: str, count: int):
         self.add_uint32(
-            constants.KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
+            KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
 
     def add_head_count_kv(self, llm: str, count: int):
         self.add_uint32(
-            constants.KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
+            KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
 
     def add_max_alibi_bias(self, llm: str, bias: float):
         self.add_float32(
-            constants.KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
+            KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
 
     def add_clamp_kqv(self, llm: str, value: float):
         self.add_float32(
-            constants.KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
+            KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
 
     def add_layer_norm_eps(self, llm: str, value: float):
         self.add_float32(
-            constants.KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
+            KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
 
     def add_layer_norm_rms_eps(self, llm: str, value: float):
         self.add_float32(
-            constants.KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
+            KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
 
     def add_rope_dimension_count(self, llm: str, count: int):
         self.add_uint32(
-            constants.KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
+            KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
 
     def add_rope_scale(self, llm: str, value:  float):
-        self.add_float32(constants.KEY_ROPE_SCALE.format(llm=llm), value)
+        self.add_float32(KEY_ROPE_SCALE.format(llm=llm), value)
 
     def add_tokenizer_model(self, model: str):
-        self.add_string(constants.KEY_TOKENIZER_MODEL, model)
+        self.add_string(KEY_TOKENIZER_MODEL, model)
 
     def add_token_list(self, tokens: List):
-        self.add_array(constants.KEY_TOKENIZER_LIST, tokens)
+        self.add_array(KEY_TOKENIZER_LIST, tokens)
 
     def add_token_merges(self, merges: List):
-        self.add_array(constants.KEY_TOKENIZER_MERGES, merges)
+        self.add_array(KEY_TOKENIZER_MERGES, merges)
 
     def add_token_types(self, types: List[int]):
-        self.add_array(constants.KEY_TOKENIZER_TOKEN_TYPE, types)
+        self.add_array(KEY_TOKENIZER_TOKEN_TYPE, types)
 
     def add_token_scores(self, scores: List[float]):
-        self.add_array(constants.KEY_TOKENIZER_SCORES, scores)
+        self.add_array(KEY_TOKENIZER_SCORES, scores)
 
     def add_bos_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_BOS_ID, id)
+        self.add_uint32(KEY_TOKENIZER_BOS_ID, id)
 
     def add_eos_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_EOS_ID, id)
+        self.add_uint32(KEY_TOKENIZER_EOS_ID, id)
 
     def add_unk_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_UNK_ID, id)
+        self.add_uint32(KEY_TOKENIZER_UNK_ID, id)
 
     def add_sep_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_SEP_ID, id)
+        self.add_uint32(KEY_TOKENIZER_SEP_ID, id)
 
     def add_pad_token_id(self, id: int):
-        self.add_uint32(constants.KEY_TOKENIZER_PAD_ID, id)
-
+        self.add_uint32(KEY_TOKENIZER_PAD_ID, id)
 
 # Example usage:
 if __name__ == "__main__":
diff --git a/gguf_namemap.py b/gguf_namemap.py
deleted file mode 100644
index 06cd0132d9ea4..0000000000000
--- a/gguf_namemap.py
+++ /dev/null
@@ -1,95 +0,0 @@
-# Recommended mapping of model tensor names for storage in gguf
-
-def get_tensor_namemap( n_blocks : int):
-    tensor_map = {}
-    # Token embeddings
-    mapped_to = "token_embd"
-    tensor_map["gpt_neox.embed_in"]           = mapped_to # gptneox
-    tensor_map["transformer.wte"]             = mapped_to # gpt2 mpt
-    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
-    tensor_map["model.embed_tokens"]          = mapped_to # llama-hf
-    tensor_map["tok_embeddings"]              = mapped_to # llama-pth
-    # Position embeddings
-    mapped_to = "pos_embd"
-    tensor_map["transformer.wpe"] = mapped_to # gpt2
-    # Output norm
-    mapped_to = "output_norm"
-    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
-    tensor_map["transformer.ln_f"]          = mapped_to # gpt2 falcon
-    tensor_map["transformer.norm_f"]        = mapped_to # mpt
-    tensor_map["model.norm"]                = mapped_to # llama-hf
-    tensor_map["norm"]                      = mapped_to # llama-pth
-    # Output
-    mapped_to = "output"
-    tensor_map["embed_out"] = mapped_to # gptneox
-    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
-    tensor_map["output"]    = mapped_to # llama-pth
-    # Attention and fee-forward layer blocks
-    for i in range(0,n_blocks):
-        # Attention norm
-        mapped_to = "blk."+str(i)+".attn_norm"
-        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to # falcon40b
-        tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to # llama-pth
-        # Attention norm 2
-        mapped_to = "blk."+str(i)+".attn_norm_2"
-        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
-        # Attention query-key-value
-        mapped_to = "blk."+str(i)+".attn_qkv"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon
-        # Attention query
-        mapped_to = "blk."+str(i)+".attn_q"
-        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to # llama-pth
-        # Attention key
-        mapped_to = "blk."+str(i)+".attn_k"
-        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to # llama-pth
-        # Attention value
-        mapped_to = "blk."+str(i)+".attn_v"
-        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to # llama-pth
-        # Attention output
-        mapped_to = "blk."+str(i)+".attn_output"
-        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to # llama-pth
-        # Feed-forward norm
-        mapped_to = "blk."+str(i)+".ffn_norm"
-        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to # mpt
-        tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to # llama-pth
-        # Feed-forward up
-        mapped_to = "blk."+str(i)+".ffn_up"
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to # llama-pth
-        # Feed-forward gate
-        mapped_to = "blk."+str(i)+".ffn_gate"
-        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to # llama-pth
-        # Feed-forward down
-        mapped_to = "blk."+str(i)+".ffn_down"
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"]   = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".mlp.down_proj"]        = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w2"]            = mapped_to # llama-pth
-
-    return tensor_map

From 5ec18934ad776a4b5ed6881a08d3918fbff29d93 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Wed, 16 Aug 2023 20:16:15 +0300
Subject: [PATCH 164/242] convert-new.py : pick #2427 for HF 70B support

---
 convert-new.py | 87 +++++++++++++++++++++++++++-----------------------
 1 file changed, 47 insertions(+), 40 deletions(-)

diff --git a/convert-new.py b/convert-new.py
index 75b07ebddd609..2c02ee73c817d 100755
--- a/convert-new.py
+++ b/convert-new.py
@@ -104,7 +104,7 @@ def make_tensors_list() -> List[str]:
 
 def find_n_mult(n_ff: int, n_embd: int) -> int:
     # hardcoded magic range
-    for n_mult in range(256, 1, -1):
+    for n_mult in range(8192, 1, -1):
         calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
         if calc_ff == n_ff:
             return n_mult
@@ -113,11 +113,12 @@ def find_n_mult(n_ff: int, n_embd: int) -> int:
 
 @dataclass
 class Params:
-    n_vocab: int
-    n_embd:  int
-    n_mult:  int
-    n_head:  int
-    n_layer: int
+    n_vocab:   int
+    n_embd:    int
+    n_mult:    int
+    n_head:    int
+    n_layer:   int
+    n_kv_head: Optional[int]  # This parameter is only used for Llama 2
 
     @staticmethod
     def guessed(model: 'LazyModel') -> 'Params':
@@ -139,31 +140,34 @@ def guessed(model: 'LazyModel') -> 'Params':
         n_head=n_embd // 128 # guessed
 
         return Params(
-            n_vocab = n_vocab,
-            n_embd  = n_embd,
-            n_mult  = 256,
-            n_head  = n_head,
-            n_layer = n_layer,
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = 256,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = None,
         )
 
     @staticmethod
     def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab = config["vocab_size"];
-        n_embd  = config["hidden_size"];
-        n_head  = config["num_attention_heads"];
-        n_layer = config["num_hidden_layers"];
-        n_ff    = config["intermediate_size"];
+        n_vocab   = config["vocab_size"];
+        n_embd    = config["hidden_size"];
+        n_head    = config["num_attention_heads"];
+        n_layer   = config["num_hidden_layers"];
+        n_ff      = config["intermediate_size"];
+        n_kv_head = config.get("num_key_value_heads")
 
         n_mult = find_n_mult(n_ff, n_embd);
 
         return Params(
-            n_vocab = n_vocab,
-            n_embd  = n_embd,
-            n_mult  = n_mult,
-            n_head  = n_head,
-            n_layer = n_layer,
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = n_mult,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = n_kv_head,
         )
 
     # LLaMA v2 70B params.json
@@ -182,11 +186,12 @@ def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
             n_vocab = model["tok_embeddings.weight"].shape[0]
 
         return Params(
-            n_vocab = n_vocab,
-            n_embd  = n_embd,
-            n_mult  = n_mult,
-            n_head  = n_head,
-            n_layer = n_layer,
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = n_mult,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = None,
         )
 
     @staticmethod
@@ -293,10 +298,12 @@ def __repr__(self) -> str:
 Vocab = Union[BpeVocab, SentencePieceVocab]
 
 
-def permute(weights: NDArray, n_head: int) -> NDArray:
+def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
+    if n_kv_head is not None and n_head != n_kv_head:
+        n_head //= n_kv_head
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                   .swapaxes(1, 2)
-                   .reshape(weights.shape))
+                .swapaxes(1, 2)
+                .reshape(weights.shape))
 
 
 class Tensor(metaclass=ABCMeta):
@@ -305,7 +312,7 @@ class Tensor(metaclass=ABCMeta):
     @abstractmethod
     def astype(self, data_type: DataType) -> 'Tensor': ...
     @abstractmethod
-    def permute(self, n_head: int) -> 'Tensor': ...
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'Tensor': ...
     @abstractmethod
     def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
     @abstractmethod
@@ -343,8 +350,8 @@ def part(self, n_part: int) -> 'UnquantizedTensor':
         r = self.ndarray.shape[0] // 3
         return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
 
-    def permute(self, n_head: int) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head))
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head, n_kv_head))
 
 
 def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
@@ -367,18 +374,18 @@ def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, conv
 
 
 class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int) -> None:
+    def __init__(self, base: Tensor, n_head: int, n_kv_head: Optional[int] = None) -> None:
         self.base = base
         self.n_head = n_head
         self.data_type = self.base.data_type
 
     def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head)
+        return self.base.astype(data_type).permute(self.n_head, self.n_kv_head)
 
     def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head)
+        return self.base.to_ggml().permute(self.n_head, self.n_kv_head)
 
-    def permute(self, n_head: int) -> Tensor:
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
         raise Exception("shouldn't permute twice")
 
 
@@ -474,10 +481,10 @@ def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
     return ModelPlus(model, paths, format, vocab)
 
 
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int) -> LazyTensor:
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_kv_head: Optional[int] = None) -> LazyTensor:
     def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
+        return lazy_tensor.load().permute(n_head, n_kv_head)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_kv_head}) ' + lazy_tensor.description)
 
 def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
     def load() -> Tensor:
@@ -502,7 +509,7 @@ def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
     for i in itertools.count():
         if f"model.layers.{i}.self_attn.q_proj.weight" in model:
             out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head)
+            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_kv_head)
             out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
         elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
             out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)

From 42f8fe19272554c2aafe1be5ab2366d0e136ce3c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 17 Aug 2023 08:56:42 +0300
Subject: [PATCH 165/242] examples/gguf : no need to keep q option for
 quantization any more

---
 examples/gguf/gguf.cpp | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
index d742dce178523..dee00df874dbb 100644
--- a/examples/gguf/gguf.cpp
+++ b/examples/gguf/gguf.cpp
@@ -233,16 +233,13 @@ int main(int argc, char ** argv) {
     const std::string fname(argv[1]);
     const std::string mode (argv[2]);
 
-    GGML_ASSERT((mode == "r" || mode == "w" || mode == "q") && "mode must be r, w or q");
+    GGML_ASSERT((mode == "r" || mode == "w") && "mode must be r or w");
 
     if (mode == "w") {
         GGML_ASSERT(gguf_ex_write(fname) && "failed to write gguf file");
     } else if (mode == "r") {
         GGML_ASSERT(gguf_ex_read_0(fname) && "failed to read gguf file");
         GGML_ASSERT(gguf_ex_read_1(fname) && "failed to read gguf file");
-    } else if (mode == "q") {
-        llama_model_quantize_params params = llama_model_quantize_default_params();
-        llama_model_quantize(fname.c_str(), "quant.gguf", &params);
     }
 
     return 0;

From 5a0a2c5685544dc41304779fb3f05f2231e300bd Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 15:18:16 +0200
Subject: [PATCH 166/242] llama.cpp : print actual model size

---
 llama.cpp | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/llama.cpp b/llama.cpp
index 38a2d5ba89f3a..5a15016514124 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1023,6 +1023,7 @@ struct llama_model_loader {
     int n_kv      = 0;
     int n_tensors = 0;
     int n_created = 0;
+    size_t n_tot_elements = 0;
 
     bool use_mmap = false;
 
@@ -1047,6 +1048,16 @@ struct llama_model_loader {
 
         file_version = (enum llama_file_version) gguf_get_version(ctx_gguf);
 
+        for (int i = 0; i < n_tensors; i++) {
+            const char * name = gguf_get_tensor_name(ctx_gguf, i);
+            struct ggml_tensor * t = ggml_get_tensor(ctx_meta, name);
+            size_t elem = 1;
+            for (int j = 0; j < t->n_dims; j++) {
+                elem *= t->ne[j];
+            }
+            n_tot_elements += elem;
+        }
+        
         // print meta data
         // TODO: make optional
         {
@@ -1413,7 +1424,8 @@ static void llama_model_load_internal(
         LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
         LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
         LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
-        LLAMA_LOG_INFO("%s: model size = %s\n",   __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: model size = %.2f B\n",   __func__, ml->n_tot_elements*1e-9);
+
     }
 
     if (vocab_only) {

From d6fd53afd64417203d77e1530f2f7bf182ffa96e Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 15:24:35 +0200
Subject: [PATCH 167/242] llama.cpp : use ggml_elements()

---
 llama.cpp | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 5a15016514124..b7ca6db3c2fa5 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1051,11 +1051,7 @@ struct llama_model_loader {
         for (int i = 0; i < n_tensors; i++) {
             const char * name = gguf_get_tensor_name(ctx_gguf, i);
             struct ggml_tensor * t = ggml_get_tensor(ctx_meta, name);
-            size_t elem = 1;
-            for (int j = 0; j < t->n_dims; j++) {
-                elem *= t->ne[j];
-            }
-            n_tot_elements += elem;
+            n_tot_elements += ggml_nelements(t);
         }
         
         // print meta data

From e0429d38e416c10a70a941faa423b1f2e80a04c7 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 17:19:52 +0300
Subject: [PATCH 168/242] convert-new.py : output gguf (#2635)

* convert-new.py : output gguf (WIP)

* convert-new.py : add gguf key-value pairs

* llama : add hparams.ctx_train + no longer print ftype

* convert-new.py : minor fixes

* convert-new.py : vocab-only option should work now

* llama : fix tokenizer to use llama_char_to_byte

* tests : add new ggml-vocab-llama.gguf

* convert-new.py : tensor name mapping

* convert-new.py : add map for skipping tensor serialization

* convert-new.py : convert script now works

* gguf.py : pick some of the refactoring from #2644

* convert-new.py : minor fixes
---
 convert-llama-7b-pth-to-gguf.py |   2 +-
 convert-new.py                  | 441 +++++++++++++++++---------------
 gguf.py                         | 301 ++++++++++++++++------
 llama.cpp                       |  92 ++++---
 models/ggml-vocab-aquila.bin    | Bin 1963875 -> 0 bytes
 models/ggml-vocab-llama.bin     | Bin 466955 -> 0 bytes
 models/ggml-vocab-llama.gguf    | Bin 0 -> 467382 bytes
 tests/CMakeLists.txt            |   6 +-
 tests/test-tokenizer-0.cpp      |  11 +-
 9 files changed, 526 insertions(+), 327 deletions(-)
 delete mode 100644 models/ggml-vocab-aquila.bin
 delete mode 100644 models/ggml-vocab-llama.bin
 create mode 100644 models/ggml-vocab-llama.gguf

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index c4e425ee32fa7..9afea8a7e4bf1 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -298,7 +298,7 @@ def count_model_parts(dir_model: str) -> int:
 
         print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-        gguf_writer.write_tensor_to_file(data)
+        gguf_writer.write_tensor_data(data)
 
 gguf_writer.close()
 
diff --git a/convert-new.py b/convert-new.py
index 2c02ee73c817d..b243356f0db28 100755
--- a/convert-new.py
+++ b/convert-new.py
@@ -1,5 +1,6 @@
 #!/usr/bin/env python
 
+import gguf
 import argparse
 import concurrent.futures
 import copy
@@ -33,6 +34,13 @@
 
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
+ARCH=gguf.MODEL_ARCH.LLAMA
+NAMES=gguf.MODEL_TENSOR_NAMES[ARCH]
+
+#
+# data types
+#
+
 @dataclass(frozen=True)
 class UnquantizedDataType:
     name: str
@@ -44,14 +52,6 @@ class UnquantizedDataType:
 
 DataType = Union[UnquantizedDataType]
 
-DATA_TYPE_TO_FTYPE: Dict[DataType, int] = {
-    DT_F32: 0,
-    DT_F16: 1,
-}
-
-FTYPE_TO_DATA_TYPE: Dict[int, DataType] = \
-    {ftype: dtype for (dtype, ftype) in DATA_TYPE_TO_FTYPE.items()}
-
 DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
     DT_BF16: np.dtype(np.uint16),
     DT_F16:  np.dtype(np.float16),
@@ -62,6 +62,13 @@ class UnquantizedDataType:
 NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
     {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
 
+SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
+    'BF16': DT_BF16,
+    'F16': DT_F16,
+    'F32': DT_F32,
+    'I32': DT_I32,
+}
+
 class GGMLFileType(enum.Enum):
     AllF32    = 0
     MostlyF16 = 1  # except 1d tensors
@@ -77,48 +84,31 @@ def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
         else:
             raise ValueError(self)
 
-# TODO: this is LLaMA specific
-def make_tensors_list() -> List[str]:
-    ret = [
-        'tok_embeddings.weight',
-        'norm.weight',
-        'output.weight',
-    ]
-    for i in range(80):  # maximum number of layer
-        ret += [
-            f'layers.{i}.attention.wq.weight',
-            f'layers.{i}.attention.wk.weight',
-            f'layers.{i}.attention.wv.weight',
-            f'layers.{i}.attention.wo.weight',
-            f'layers.{i}.attention_norm.weight',
-            f'layers.{i}.feed_forward.w1.weight',
-            f'layers.{i}.feed_forward.w2.weight',
-            f'layers.{i}.feed_forward.w3.weight',
-            f'layers.{i}.ffn_norm.weight',
-        ]
-    return ret
-
-# TODO: this should be generalized for non-LLaMA models
-TENSORS_LIST = make_tensors_list()
-TENSORS_SET = set(TENSORS_LIST)
-
-def find_n_mult(n_ff: int, n_embd: int) -> int:
-    # hardcoded magic range
-    for n_mult in range(8192, 1, -1):
-        calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
-        if calc_ff == n_ff:
-            return n_mult
-    raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
 
+#
+# hparams loading
+#
 
 @dataclass
 class Params:
-    n_vocab:   int
-    n_embd:    int
-    n_mult:    int
-    n_head:    int
-    n_layer:   int
-    n_kv_head: Optional[int]  # This parameter is only used for Llama 2
+    n_vocab:    int
+    n_embd:     int
+    n_mult:     int
+    n_layer:    int
+    n_ctx:      int
+    n_ff:       int
+    n_head:     int
+    n_head_kv:  int
+    f_norm_eps: float
+
+    @staticmethod
+    def find_n_mult(n_ff: int, n_embd: int) -> int:
+        # hardcoded magic range
+        for n_mult in range(8192, 1, -1):
+            calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
+            if calc_ff == n_ff:
+                return n_mult
+        raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
 
     @staticmethod
     def guessed(model: 'LazyModel') -> 'Params':
@@ -137,37 +127,57 @@ def guessed(model: 'LazyModel') -> 'Params':
             raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
                             "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
 
-        n_head=n_embd // 128 # guessed
+        n_head = n_embd // 128 # guessed
+        n_mult = 256           # guessed
+
+        # TODO: verify this
+        n_ff = int(2 * (4 * n_embd) / 3)
+        n_ff = n_mult * ((n_ff + n_mult - 1) // n_mult)
 
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = 256,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = -1,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head,
+            f_norm_eps = 1e-5,
         )
 
     @staticmethod
     def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab   = config["vocab_size"];
-        n_embd    = config["hidden_size"];
-        n_head    = config["num_attention_heads"];
-        n_layer   = config["num_hidden_layers"];
-        n_ff      = config["intermediate_size"];
-        n_kv_head = config.get("num_key_value_heads")
+        n_vocab    = config["vocab_size"];
+        n_embd     = config["hidden_size"];
+        n_layer    = config["num_hidden_layers"];
+        n_ff       = config["intermediate_size"];
+        n_head     = config["num_attention_heads"];
+        n_head_kv  = config["num_key_value_heads"];
+        f_norm_eps = config["rms_norm_eps"];
+
+        n_mult = Params.find_n_mult(n_ff, n_embd);
 
-        n_mult = find_n_mult(n_ff, n_embd);
+        if "max_sequence_length" in config:
+            n_ctx = config["max_sequence_length"]
+        elif "max_position_embeddings" in config:
+            n_ctx = config["max_position_embeddings"]
+        else:
+            raise Exception("failed to guess 'n_ctx'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
 
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = n_kv_head,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
         )
 
     # LLaMA v2 70B params.json
@@ -176,22 +186,32 @@ def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
     def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab = config["vocab_size"];
-        n_embd  = config["dim"];
-        n_head  = config["n_heads"];
-        n_layer = config["n_layers"];
-        n_mult  = config["multiple_of"];
+        n_vocab    = config["vocab_size"];
+        n_embd     = config["dim"];
+        n_layer    = config["n_layers"];
+        n_mult     = config["multiple_of"];
+        n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
+        n_ff       = -1;
+        n_head     = config["n_heads"];
+        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head;
+        f_norm_eps = config["norm_eps"];
 
         if n_vocab == -1:
             n_vocab = model["tok_embeddings.weight"].shape[0]
 
+        if n_ff == -1:
+            n_ff = model["layers.0.feed_forward.w1.weight"].shape[0]
+
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
         )
 
     @staticmethod
@@ -206,10 +226,13 @@ def load(model_plus: 'ModelPlus') -> 'Params':
         else:
             params = Params.guessed(model_plus.model)
 
-        print(f'params: n_vocab:{params.n_vocab} n_embd:{params.n_embd} n_mult:{params.n_mult} n_head:{params.n_head} n_layer:{params.n_layer}')
         return params
 
 
+#
+# vocab
+#
+
 class BpeVocab:
     def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
         self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
@@ -294,13 +317,17 @@ def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
     def __repr__(self) -> str:
         return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
 
-
 Vocab = Union[BpeVocab, SentencePieceVocab]
 
 
-def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
-    if n_kv_head is not None and n_head != n_kv_head:
-        n_head //= n_kv_head
+#
+# data loading
+# TODO: reuse (probably move to gguf.py?)
+#
+
+def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
+    if n_head_kv is not None and n_head != n_head_kv:
+        n_head //= n_head_kv
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
                 .swapaxes(1, 2)
                 .reshape(weights.shape))
@@ -312,7 +339,7 @@ class Tensor(metaclass=ABCMeta):
     @abstractmethod
     def astype(self, data_type: DataType) -> 'Tensor': ...
     @abstractmethod
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'Tensor': ...
+    def permute(self, n_head: int, n_head_kv: int) -> 'Tensor': ...
     @abstractmethod
     def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
     @abstractmethod
@@ -350,8 +377,8 @@ def part(self, n_part: int) -> 'UnquantizedTensor':
         r = self.ndarray.shape[0] // 3
         return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
 
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head, n_kv_head))
+    def permute(self, n_head: int, n_head_kv: int) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head, n_head_kv))
 
 
 def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
@@ -374,18 +401,18 @@ def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, conv
 
 
 class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int, n_kv_head: Optional[int] = None) -> None:
+    def __init__(self, base: Tensor, n_head: int, n_head_kv: int) -> None:
         self.base = base
         self.n_head = n_head
         self.data_type = self.base.data_type
 
     def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head, self.n_kv_head)
+        return self.base.astype(data_type).permute(self.n_head, self.n_head_kv)
 
     def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head, self.n_kv_head)
+        return self.base.to_ggml().permute(self.n_head, self.n_head_kv)
 
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
+    def permute(self, n_head: int, n_head_kv: int) -> Tensor:
         raise Exception("shouldn't permute twice")
 
 
@@ -481,10 +508,10 @@ def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
     return ModelPlus(model, paths, format, vocab)
 
 
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_kv_head: Optional[int] = None) -> LazyTensor:
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_head_kv: int) -> LazyTensor:
     def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head, n_kv_head)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_kv_head}) ' + lazy_tensor.description)
+        return lazy_tensor.load().permute(n_head, n_head_kv)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_head_kv}) ' + lazy_tensor.description)
 
 def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
     def load() -> Tensor:
@@ -500,34 +527,6 @@ def load() -> Tensor:
     s[0] = s[0] // 3
     return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
 
-def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
-    out: LazyModel = {}
-    out["tok_embeddings.weight"] = model["model.embed_tokens.weight"]
-    out["norm.weight"]           = model["model.norm.weight"]
-    out["output.weight"]         = model["lm_head.weight"]
-
-    for i in itertools.count():
-        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_kv_head)
-            out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
-        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head)
-            out[f"layers.{i}.attention.wv.weight"] = part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
-        else:
-            break
-
-        out[f"layers.{i}.attention.wo.weight"] = model[f"model.layers.{i}.self_attn.o_proj.weight"]
-
-        out[f"layers.{i}.feed_forward.w1.weight"] = model[f"model.layers.{i}.mlp.gate_proj.weight"]
-        out[f"layers.{i}.feed_forward.w2.weight"] = model[f"model.layers.{i}.mlp.down_proj.weight"]
-        out[f"layers.{i}.feed_forward.w3.weight"] = model[f"model.layers.{i}.mlp.up_proj.weight"]
-
-        out[f"layers.{i}.attention_norm.weight"] = model[f"model.layers.{i}.input_layernorm.weight"]
-        out[f"layers.{i}.ffn_norm.weight"]       = model[f"model.layers.{i}.post_attention_layernorm.weight"]
-    return out
-
 
 # Functionality that simulates `torch.load` but where individual tensors are
 # only loaded into memory on demand, not all at once.
@@ -621,14 +620,6 @@ def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
     return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
 
 
-SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
-    'BF16': DT_BF16,
-    'F16': DT_F16,
-    'F32': DT_F32,
-    'I32': DT_I32,
-}
-
-
 def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
     header_size, = struct.unpack('<Q', fp.read(8))
     header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
@@ -678,7 +669,6 @@ def lazy_load_file(path: Path) -> ModelPlus:
 In = TypeVar('In')
 Out = TypeVar('Out')
 
-
 def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
     '''Parallel map, but with backpressure.  If the caller doesn't call `next`
     fast enough, this will stop calling `func` at some point rather than
@@ -715,88 +705,133 @@ def check_vocab_size(params: Params, vocab: Vocab) -> None:
 
 class OutputFile:
     def __init__(self, fname_out: Path) -> None:
-        self.fout = open(fname_out, "wb")
-
-    def write_file_header(self, params: Params, file_type: GGMLFileType) -> None:
-        self.fout.write(b"ggjt"[::-1])  # magic
-        values = [
-            1,  # file version
-            params.n_vocab,
-            params.n_embd,
-            params.n_mult,
-            params.n_head,
-            params.n_layer,
-            params.n_embd // params.n_head,  # rot (obsolete)
-            file_type.value,
-        ]
-        self.fout.write(struct.pack("i" * len(values), *values))
-
-    def write_tensor_header(self, name: str, shape: Sequence[int], data_type: DataType) -> None:
-        sname = name.encode('utf-8')
-        self.fout.write(struct.pack("iii", len(shape), len(sname), DATA_TYPE_TO_FTYPE[data_type]))
-        self.fout.write(struct.pack("i" * len(shape), *shape[::-1]))
-        self.fout.write(sname)
-        self.fout.seek((self.fout.tell() + 31) & -32)
-
-    def write_vocab(self, vocab: Vocab) -> None:
+        self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
+
+    def add_meta_arch(self, params: Params) -> None:
+        self.gguf.add_context_length      (params.n_ctx)
+        self.gguf.add_embedding_length    (params.n_embd)
+        self.gguf.add_block_count         (params.n_layer)
+        self.gguf.add_feed_forward_length (params.n_ff)
+        self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
+        self.gguf.add_head_count          (params.n_head)
+        self.gguf.add_head_count_kv       (params.n_head_kv)
+        self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
+
+    def add_meta_vocab(self, vocab: Vocab) -> None:
+        tokens = []
+        scores = []
         for text, score in vocab.all_tokens():
-            self.fout.write(struct.pack("i", len(text)))
-            self.fout.write(text)
-            self.fout.write(struct.pack("f", score))
+            tokens.append(text)
+            scores.append(score)
+
+        self.gguf.add_tokenizer_model("llama")
+        self.gguf.add_token_list(tokens)
+        self.gguf.add_token_scores(scores)
+        #self.gguf.add_token_types(toktypes) # TODO: add this
+
+        # TODO: added / special tokens
+
+    def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
+        n_elements = 1
+        for dim in tensor.shape:
+            n_elements *= dim
+        data_type = DATA_TYPE_TO_NUMPY[tensor.data_type]
+        data_nbytes = n_elements * data_type.itemsize
+        self.gguf.add_tensor_info(name, tensor.shape, data_type, data_nbytes)
+
+    def write_meta(self) -> None:
+        self.gguf.write_header_to_file()
+        self.gguf.write_kv_data_to_file()
+
+    def write_tensor_info(self) -> None:
+        self.gguf.write_ti_data_to_file()
+
+    def close(self) -> None:
+        self.gguf.close()
 
     @staticmethod
-    def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
-        of = OutputFile(fname_out)
-        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0)
+    def write_vocab_only(fname_out: Path, params: Params, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+
         of = OutputFile(fname_out)
-        of.write_file_header(params, file_type=GGMLFileType.AllF32)
-        of.write_vocab(vocab)
-        of.fout.close()
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+        of.write_meta()
+
+        of.close()
 
     @staticmethod
-    def write_all(fname_out: Path, params: Params, file_type: GGMLFileType, model: LazyModel, vocab: Vocab) -> None:
+    def write_all(fname_out: Path, params: Params, model: LazyModel, vocab: Vocab) -> None:
         check_vocab_size(params, vocab)
+
         of = OutputFile(fname_out)
-        of.write_file_header(params, file_type)
-        print("Writing vocab...")
-        of.write_vocab(vocab)
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+
+        # tensor info
+        for name, lazy_tensor in model.items():
+            of.add_tensor_info(name, lazy_tensor)
+
+        of.write_meta()
+        of.write_tensor_info()
 
         def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
             name, lazy_tensor = item
             return lazy_tensor.load().to_ggml().ndarray
 
+        # tensor data
         ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8)
         for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
             size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
             padi = len(str(len(model)))
             print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
-            of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
-            ndarray.tofile(of.fout)
-        of.fout.close()
+            of.gguf.write_tensor_data(ndarray)
 
+        of.close()
 
 def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
-    wq_type = model["layers.0.attention.wq.weight"].data_type
-    if output_type_str == "f32" or (output_type_str is None and wq_type in (DT_F32, DT_BF16)):
+    wq_type = model[NAMES[gguf.MODEL_TENSOR.ATTN_Q].format(bid=0)+".weight"].data_type
+
+    if output_type_str == "f32" or (output_type_str is None and wq_type == DT_F32):
         return GGMLFileType.AllF32
-    if output_type_str == "f16" or (output_type_str is None and wq_type == DT_F16):
+    if output_type_str == "f16" or (output_type_str is None and wq_type in (DT_F16, DT_BF16)):
         return GGMLFileType.MostlyF16
-    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
-    raise Exception(f"Unexpected combination of types: {name_to_type}")
-
-
-def do_necessary_conversions(model: LazyModel, params: Params) -> LazyModel:
-    if "lm_head.weight" in model:
-        model = convert_transformers_to_orig(model, params)
-    model = filter_and_sort_tensors(model)
 
-    return model
+    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
 
+    raise Exception(f"Unexpected combination of types: {name_to_type}")
 
 def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
     return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
             for (name, tensor) in model.items()}
 
+def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
+    tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
+
+    out: LazyModel = {}
+    for name, lazy_tensor in model.items():
+        name_new = name
+
+        if name in tmap:
+            name_new = tmap[name]
+        elif name.endswith(".weight") and name[:-7] in tmap:
+            name_new = tmap[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tmap:
+            name_new = tmap[name[:-5]] + ".bias"
+        else:
+            raise Exception(f"Unexpected tensor name: {name}")
+
+        if gguf.should_skip_tensor(ARCH, params.n_layer, name_new):
+            print(f"skipping tensor {name_new}")
+        else:
+            print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
+            out[name_new] = lazy_tensor
+
+    return out
 
 def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
     '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
@@ -847,11 +882,6 @@ def load_some_model(path: Path) -> ModelPlus:
             # Try the PyTorch patterns too, with lower priority
             globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
             files = [file for glob in globs for file in path.glob(glob)]
-        if not files:
-            # Try GGML too, but with lower priority, since if both a non-GGML
-            # model and a GGML model exist in the same directory, we assume the
-            # latter was converted from the former.
-            files = list(path.glob("ggml-model*.bin*"))
         if not files:
             raise Exception(f"Can't find model in directory {path}")
         if len(files) > 1:
@@ -868,12 +898,7 @@ def load_some_model(path: Path) -> ModelPlus:
     return model_plus
 
 
-def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
-    return {name: model[name] for name in TENSORS_LIST if name in model}
-
-
 def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
-    print(f"vocabtype: {vocabtype}")
     # Be extra-friendly and accept either a file or a directory.  Also, if it's
     # a directory, it might be the model directory, and tokenizer.model might
     # be in the parent of that.
@@ -892,8 +917,10 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, Sentence
             raise FileNotFoundError(
                 f"Could not find tokenizer.model in {path} or its parent; "
                 "if it's in another directory, pass the directory as --vocab-dir")
+
+    print(f"Loading vocab file '{path}', type '{vocabtype}'")
+
     added_tokens_path = path.parent / "added_tokens.json"
-    print(f"Loading vocab file {path}")
     if vocabtype == "bpe":
         return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
     elif vocabtype == "spm":
@@ -933,38 +960,52 @@ def main(args_in: Optional[List[str]] = None) -> None:
     parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
     parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
     parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)")
+    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)", default="spm")
+    parser.add_argument("--ctx",         type=int,               help="model training context (default: based on input)")
     args = parser.parse_args(args_in)
 
-    vocab: Vocab
     if args.dump_single:
         model_plus = lazy_load_file(args.model)
         do_dump_model(model_plus)
-    elif args.vocab_only:
+
+    model_plus = load_some_model(args.model)
+
+    params = Params.load(model_plus)
+    if params.n_ctx == -1:
+        if args.ctx is None:
+            raise Exception("The model doesn't have a context size, and you didn't specify one with --ctx\n"
+                            "Please specify one with --ctx:\n"
+                            " - LLaMA v1: --ctx 2048\n"
+                            " - LLaMA v2: --ctx 4096\n")
+        params.n_ctx = args.ctx
+
+    print(f"params = {params}")
+
+    vocab: Vocab
+    if args.vocab_only:
         vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
         assert args.outfile, "need --outfile if using --vocab-only"
         outfile = args.outfile
-        OutputFile.write_vocab_only(outfile, vocab)
+        OutputFile.write_vocab_only(outfile, params, vocab)
         print(f"Wrote {outfile}")
     else:
-        model_plus = load_some_model(args.model)
         if args.dump:
             do_dump_model(model_plus)
             return
+
         if model_plus.vocab is not None and args.vocab_dir is None:
             vocab = model_plus.vocab
         else:
             vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
             vocab = load_vocab(vocab_dir, args.vocabtype)
 
-        params      = Params.load(model_plus)
         model       = model_plus.model
-        model       = do_necessary_conversions(model, params)
+        model       = convert_model_names(model, params)
         output_type = pick_output_type(model, args.outtype)
         model       = convert_to_output_type(model, output_type)
         outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
 
-        OutputFile.write_all(outfile, params, output_type, model, vocab)
+        OutputFile.write_all(outfile, params, model, vocab)
         print(f"Wrote {outfile}")
 
 
diff --git a/gguf.py b/gguf.py
index e7f6f0ac89b96..a4dd108722d6c 100644
--- a/gguf.py
+++ b/gguf.py
@@ -8,7 +8,7 @@
 import struct
 import numpy as np
 
-from enum import IntEnum
+from enum import IntEnum, auto
 from typing import Any, IO, List
 
 #
@@ -33,24 +33,24 @@
 KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
 
 # LLM
-KEY_LLM_CONTEXT_LENGTH           = "{llm}.context_length"
-KEY_LLM_EMBEDDING_LENGTH         = "{llm}.embedding_length"
-KEY_LLM_BLOCK_COUNT              = "{llm}.block_count"
-KEY_LLM_FEED_FORWARD_LENGTH      = "{llm}.feed_forward_length"
-KEY_LLM_USE_PARALLEL_RESIDUAL    = "{llm}.use_parallel_residual"
-KEY_LLM_TENSOR_DATA_LAYOUT       = "{llm}.tensor_data_layout"
+KEY_LLM_CONTEXT_LENGTH           = "{arch}.context_length"
+KEY_LLM_EMBEDDING_LENGTH         = "{arch}.embedding_length"
+KEY_LLM_BLOCK_COUNT              = "{arch}.block_count"
+KEY_LLM_FEED_FORWARD_LENGTH      = "{arch}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL    = "{arch}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT       = "{arch}.tensor_data_layout"
 
 # attention
-KEY_ATTENTION_HEAD_COUNT         = "{llm}.attention.head_count"
-KEY_ATTENTION_HEAD_COUNT_KV      = "{llm}.attention.head_count_kv"
-KEY_ATTENTION_MAX_ALIBI_BIAS     = "{llm}.attention.max_alibi_bias"
-KEY_ATTENTION_CLAMP_KQV          = "{llm}.attention.clamp_kqv"
-KEY_ATTENTION_LAYERNORM_EPS      = "{llm}.attention.layer_norm_epsilon"
-KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{llm}.attention.layer_norm_rms_epsilon"
+KEY_ATTENTION_HEAD_COUNT         = "{arch}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV      = "{arch}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS     = "{arch}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV          = "{arch}.attention.clamp_kqv"
+KEY_ATTENTION_LAYERNORM_EPS      = "{arch}.attention.layer_norm_epsilon"
+KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{arch}.attention.layer_norm_rms_epsilon"
 
 # RoPE
-KEY_ROPE_DIMENSION_COUNT         = "{llm}.rope.dimension_count"
-KEY_ROPE_SCALE                   = "{llm}.rope.scale"
+KEY_ROPE_DIMENSION_COUNT         = "{arch}.rope.dimension_count"
+KEY_ROPE_SCALE                   = "{arch}.rope.scale"
 
 # tokenization
 KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
@@ -70,34 +70,137 @@
 # recommended mapping of model tensor names for storage in gguf
 #
 
-def get_tensor_name_map(n_blocks : int):
+class MODEL_ARCH(IntEnum):
+    LLAMA   = auto()
+    FALCON  = auto()
+    GPT2    = auto()
+    GPTJ    = auto()
+    GPTNEOX = auto()
+    MPT     = auto()
+
+class MODEL_TENSOR(IntEnum):
+    TOKEN_EMBD        = auto()
+    POS_EMBD          = auto()
+    OUTPUT            = auto()
+    OUTPUT_NORM       = auto()
+    ROPE_FREQS        = auto()
+    ATTN_Q            = auto()
+    ATTN_K            = auto()
+    ATTN_V            = auto()
+    ATTN_QKV          = auto()
+    ATTN_OUT          = auto()
+    ATTN_NORM         = auto()
+    ATTN_NORM_2       = auto()
+    ATTN_ROT_EMBD     = auto()
+    FFN_GATE          = auto()
+    FFN_DOWN          = auto()
+    FFN_UP            = auto()
+    FFN_NORM          = auto()
+
+MODEL_ARCH_NAMES = {
+    MODEL_ARCH.LLAMA   : "llama",
+    MODEL_ARCH.FALCON  : "falcon",
+    MODEL_ARCH.GPT2    : "gpt2",
+    MODEL_ARCH.GPTJ    : "gptj",
+    MODEL_ARCH.GPTNEOX : "gptneox",
+    MODEL_ARCH.MPT     : "mpt",
+    }
+
+MODEL_TENSOR_NAMES = {
+    MODEL_ARCH.LLAMA  : {
+        MODEL_TENSOR.TOKEN_EMBD        : "token_embd",
+        MODEL_TENSOR.OUTPUT_NORM       : "output_norm",
+        MODEL_TENSOR.OUTPUT            : "output",
+        MODEL_TENSOR.ROPE_FREQS        : "rope_freqs",
+        MODEL_TENSOR.ATTN_NORM         : "blk.{bid}.attn_norm",
+        MODEL_TENSOR.ATTN_Q            : "blk.{bid}.attn_q",
+        MODEL_TENSOR.ATTN_K            : "blk.{bid}.attn_k",
+        MODEL_TENSOR.ATTN_V            : "blk.{bid}.attn_v",
+        MODEL_TENSOR.ATTN_OUT          : "blk.{bid}.attn_output",
+        MODEL_TENSOR.ATTN_ROT_EMBD     : "blk.{bid}.attn_rot_embd",
+        MODEL_TENSOR.FFN_NORM          : "blk.{bid}.ffn_norm",
+        MODEL_TENSOR.FFN_GATE          : "blk.{bid}.ffn_gate",
+        MODEL_TENSOR.FFN_DOWN          : "blk.{bid}.ffn_down",
+        MODEL_TENSOR.FFN_UP            : "blk.{bid}.ffn_up",
+        },
+    MODEL_ARCH.FALCON : {
+        MODEL_TENSOR.TOKEN_EMBD  : "token_embd",
+        MODEL_TENSOR.OUTPUT_NORM : "output_norm",
+        MODEL_TENSOR.OUTPUT      : "output",
+        MODEL_TENSOR.ATTN_NORM   : "blk.{bid}.attn_norm",
+        MODEL_TENSOR.ATTN_NORM_2 : "blk.{bid}.attn_norm_2",
+        MODEL_TENSOR.ATTN_QKV    : "blk.{bid}.attn_qkv",
+        MODEL_TENSOR.ATTN_OUT    : "blk.{bid}.attn_output",
+        MODEL_TENSOR.FFN_DOWN    : "blk.{bid}.ffn_down",
+        MODEL_TENSOR.FFN_UP      : "blk.{bid}.ffn_up",
+        },
+    MODEL_ARCH.GPT2 : {
+        # TODO
+        },
+    # TODO
+    }
+
+# tensors that will not be serialized
+MODEL_TENSOR_SKIP = {
+    MODEL_ARCH.LLAMA : [
+        MODEL_TENSOR.ROPE_FREQS,
+        MODEL_TENSOR.ATTN_ROT_EMBD,
+        ],
+    }
+
+def should_skip_tensor(arch : MODEL_ARCH, n_blocks : int, name : str) -> bool:
+    for skip in MODEL_TENSOR_SKIP.get(arch, []):
+        for i in range(n_blocks):
+            if name == MODEL_TENSOR_NAMES[arch][skip].format(bid=i):
+                return True
+
+    return False
+
+def get_tensor_name_map(arch : MODEL_ARCH, n_blocks : int) -> dict:
     tensor_map = {}
+
     # Token embeddings
-    mapped_to = "token_embd"
+    mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.TOKEN_EMBD, None)
+
     tensor_map["gpt_neox.embed_in"]           = mapped_to # gptneox
     tensor_map["transformer.wte"]             = mapped_to # gpt2 mpt
     tensor_map["transformer.word_embeddings"] = mapped_to # falcon
     tensor_map["model.embed_tokens"]          = mapped_to # llama-hf
     tensor_map["tok_embeddings"]              = mapped_to # llama-pth
+
     # Position embeddings
-    mapped_to = "pos_embd"
+    mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.POS_EMBD, None)
+
     tensor_map["transformer.wpe"] = mapped_to # gpt2
+
+    # Output
+    mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT, None)
+
+    tensor_map["embed_out"] = mapped_to # gptneox
+    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
+    tensor_map["output"]    = mapped_to # llama-pth
+
     # Output norm
-    mapped_to = "output_norm"
+    mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT_NORM, None)
+
     tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
     tensor_map["transformer.ln_f"]          = mapped_to # gpt2 falcon
     tensor_map["transformer.norm_f"]        = mapped_to # mpt
     tensor_map["model.norm"]                = mapped_to # llama-hf
     tensor_map["norm"]                      = mapped_to # llama-pth
-    # Output
-    mapped_to = "output"
-    tensor_map["embed_out"] = mapped_to # gptneox
-    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
-    tensor_map["output"]    = mapped_to # llama-pth
-    # Attention and fee-forward layer blocks
+
+    # Rope frequencies
+    mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ROPE_FREQS, None)
+
+    tensor_map["rope.freqs"] = mapped_to # llama-pth
+
+    # Attention and feed-forward blocks
     for i in range(0,n_blocks):
         # Attention norm
-        mapped_to = "blk."+str(i)+".attn_norm"
+        # TODO: is there are simpler way to write these 2 lines in Python?
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to else None
+
         tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to # mpt
@@ -105,56 +208,93 @@ def get_tensor_name_map(n_blocks : int):
         tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to # falcon40b
         tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to # llama-pth
+
         # Attention norm 2
-        mapped_to = "blk."+str(i)+".attn_norm_2"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM_2, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
+
         # Attention query-key-value
-        mapped_to = "blk."+str(i)+".attn_qkv"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_QKV, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to # mpt
         tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon
+
         # Attention query
-        mapped_to = "blk."+str(i)+".attn_q"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_Q, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to # llama-pth
+
         # Attention key
-        mapped_to = "blk."+str(i)+".attn_k"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_K, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to # llama-pth
+
         # Attention value
-        mapped_to = "blk."+str(i)+".attn_v"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_V, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to # llama-pth
+
         # Attention output
-        mapped_to = "blk."+str(i)+".attn_output"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_OUT, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to # mpt
         tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
         tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to # llama-pth
+
+        # Rotary embeddings
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_ROT_EMBD, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
+        tensor_map["model.layers."+str(i)+".self_attn.rotary_emb.inv_freq"]  = mapped_to # llama-hf
+        tensor_map["layers."+str(i)+".attention.inner_attention.rope.freqs"] = mapped_to # llama-pth
+
         # Feed-forward norm
-        mapped_to = "blk."+str(i)+".ffn_norm"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_NORM, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to # mpt
         tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to # llama-pth
+
         # Feed-forward up
-        mapped_to = "blk."+str(i)+".ffn_up"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_UP, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to # mpt
         tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to # falcon
         tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to # llama-pth
+
         # Feed-forward gate
-        mapped_to = "blk."+str(i)+".ffn_gate"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_GATE, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
         tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to # llama-pth
+
         # Feed-forward down
-        mapped_to = "blk."+str(i)+".ffn_down"
+        mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_DOWN, None)
+        mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
+
         tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
         tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to # gpt2
         tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to # mpt
@@ -203,14 +343,16 @@ def get_type(val):
 
 
 class GGUFWriter:
-    def __init__(self, fout: IO):
-        self.fout = fout
+    def __init__(self, path: str, arch: str):
+        self.fout = open(path, "wb")
+        self.arch = arch
         self.offset_tensor = 0
         self.data_alignment = GGUF_DEFAULT_ALIGNMENT
         self.kv_data = b""
         self.kv_data_count = 0
         self.ti_data = b""
         self.ti_data_count = 0
+        self.add_architecture()
 
     def write_header_to_file(self):
         self.fout.write(struct.pack("<I", GGUF_MAGIC))
@@ -228,11 +370,6 @@ def write_ti_data_to_file(self):
         self.fout.write(self.ti_data)
         self.flush()
 
-    @classmethod
-    def open(cls, path: str) -> "GGUFWriter":
-        f = open(path, "wb")
-        return cls(f)
-
     def add_key(self, key: str):
         self.add_val(key, GGUFValueType.STRING, add_vtype=False)
 
@@ -269,7 +406,8 @@ def add_bool(self, key: str, val: bool):
         self.add_val(val, GGUFValueType.BOOL)
 
     def add_string(self, key: str, val: str):
-        if len(val) == 0: return
+        if len(val) == 0:
+            return
         self.add_key(key)
         self.add_val(val, GGUFValueType.STRING)
 
@@ -323,6 +461,8 @@ def ggml_pad(x: int, n: int) -> int:
         return ((x + n - 1) // n) * n
 
     def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int):
+        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
+
         encoded_name = name.encode("utf8")
         self.ti_data += struct.pack("<I", len(encoded_name))
         self.ti_data += encoded_name
@@ -331,14 +471,13 @@ def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.
         for i in range(n_dims):
             self.ti_data += struct.pack("<I", tensor_shape[n_dims - 1 - i])
 
-        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
         dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
         self.ti_data += struct.pack("<I", dtype)
         self.ti_data += struct.pack("<Q", self.offset_tensor)
         self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
         self.ti_data_count += 1
 
-    def write_tensor_to_file(self, tensor: np.ndarray):
+    def write_tensor_data(self, tensor: np.ndarray):
         pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
         if pad != 0:
             self.fout.write(bytes([0] * pad))
@@ -355,15 +494,14 @@ def flush(self):
     def close(self):
         self.fout.close()
 
-    def add_architecture(self, architecture: str):
-        self.add_string(KEY_GENERAL_ARCHITECTURE,
-                        architecture)
+    def add_architecture(self):
+        self.add_string(KEY_GENERAL_ARCHITECTURE, self.arch)
 
     def add_author(self, author: str):
         self.add_string(KEY_GENERAL_AUTHOR, author)
 
     def add_tensor_data_layout(self, layout: str):
-        self.add_string(KEY_LLM_TENSOR_DATA_LAYOUT , layout)
+        self.add_string(KEY_LLM_TENSOR_DATA_LAYOUT.format(arch=self.arch), layout)
 
     def add_url(self, url: str):
         self.add_string(KEY_GENERAL_URL, url)
@@ -391,60 +529,60 @@ def add_custom_alignment(self, alignment: int):
         self.data_alignment = alignment
         self.add_uint32(KEY_GENERAL_ALIGNMENT, alignment)
 
-    def add_context_length(self, llm: str, length: int):
+    def add_context_length(self, length: int):
         self.add_uint32(
-            KEY_LLM_CONTEXT_LENGTH.format(llm=llm), length)
+            KEY_LLM_CONTEXT_LENGTH.format(arch=self.arch), length)
 
-    def add_embedding_length(self, llm: str, length: int):
+    def add_embedding_length(self, length: int):
         self.add_uint32(
-            KEY_LLM_EMBEDDING_LENGTH.format(llm=llm), length)
+            KEY_LLM_EMBEDDING_LENGTH.format(arch=self.arch), length)
 
-    def add_block_count(self, llm: str, length: int):
+    def add_block_count(self, length: int):
         self.add_uint32(
-            KEY_LLM_BLOCK_COUNT.format(llm=llm), length)
+            KEY_LLM_BLOCK_COUNT.format(arch=self.arch), length)
 
-    def add_feed_forward_length(self, llm: str, length: int):
+    def add_feed_forward_length(self, length: int):
         self.add_uint32(
-            KEY_LLM_FEED_FORWARD_LENGTH.format(llm=llm), length)
+            KEY_LLM_FEED_FORWARD_LENGTH.format(arch=self.arch), length)
 
-    def add_parallel_residual(self, llm: str, use: bool):
+    def add_parallel_residual(self, use: bool):
         self.add_bool(
-            KEY_LLM_USE_PARALLEL_RESIDUAL.format(llm=llm), use)
+            KEY_LLM_USE_PARALLEL_RESIDUAL.format(arch=self.arch), use)
 
-    def add_tensor_data_layout(self, llm: str, layout: str):
+    def add_tensor_data_layout(self, layout: str):
         self.add_string(
-            KEY_LLM_TENSOR_DATA_LAYOUT.format(llm=llm), layout)
+            KEY_LLM_TENSOR_DATA_LAYOUT.format(arch=self.arch), layout)
 
-    def add_head_count(self, llm: str, count: int):
+    def add_head_count(self, count: int):
         self.add_uint32(
-            KEY_ATTENTION_HEAD_COUNT.format(llm=llm), count)
+            KEY_ATTENTION_HEAD_COUNT.format(arch=self.arch), count)
 
-    def add_head_count_kv(self, llm: str, count: int):
+    def add_head_count_kv(self, count: int):
         self.add_uint32(
-            KEY_ATTENTION_HEAD_COUNT_KV.format(llm=llm), count)
+            KEY_ATTENTION_HEAD_COUNT_KV.format(arch=self.arch), count)
 
-    def add_max_alibi_bias(self, llm: str, bias: float):
+    def add_max_alibi_bias(self, bias: float):
         self.add_float32(
-            KEY_ATTENTION_MAX_ALIBI_BIAS.format(llm=llm), bias)
+            KEY_ATTENTION_MAX_ALIBI_BIAS.format(arch=self.arch), bias)
 
-    def add_clamp_kqv(self, llm: str, value: float):
+    def add_clamp_kqv(self, value: float):
         self.add_float32(
-            KEY_ATTENTION_CLAMP_KQV.format(llm=llm), value)
+            KEY_ATTENTION_CLAMP_KQV.format(arch=self.arch), value)
 
-    def add_layer_norm_eps(self, llm: str, value: float):
+    def add_layer_norm_eps(self, value: float):
         self.add_float32(
-            KEY_ATTENTION_LAYERNORM_EPS.format(llm=llm), value)
+            KEY_ATTENTION_LAYERNORM_EPS.format(arch=self.arch), value)
 
-    def add_layer_norm_rms_eps(self, llm: str, value: float):
+    def add_layer_norm_rms_eps(self, value: float):
         self.add_float32(
-            KEY_ATTENTION_LAYERNORM_RMS_EPS.format(llm=llm), value)
+            KEY_ATTENTION_LAYERNORM_RMS_EPS.format(arch=self.arch), value)
 
-    def add_rope_dimension_count(self, llm: str, count: int):
+    def add_rope_dimension_count(self, count: int):
         self.add_uint32(
-            KEY_ROPE_DIMENSION_COUNT.format(llm=llm), count)
+            KEY_ROPE_DIMENSION_COUNT.format(arch=self.arch), count)
 
-    def add_rope_scale(self, llm: str, value:  float):
-        self.add_float32(KEY_ROPE_SCALE.format(llm=llm), value)
+    def add_rope_scale(self, value:  float):
+        self.add_float32(KEY_ROPE_SCALE.format(arch=self.arch), value)
 
     def add_tokenizer_model(self, model: str):
         self.add_string(KEY_TOKENIZER_MODEL, model)
@@ -479,9 +617,8 @@ def add_pad_token_id(self, id: int):
 # Example usage:
 if __name__ == "__main__":
     # Example usage with a file
-    gguf_writer = GGUFWriter.open("example.gguf")
+    gguf_writer = GGUFWriter("example.gguf", "llama")
 
-    gguf_writer.add_architecture("llama")
     gguf_writer.add_uint32("answer", 42)  # Write a 32-bit integer
     gguf_writer.add_float32("answer_in_float", 42.0)  # Write a 32-bit float
     gguf_writer.add_custom_alignment(64)
@@ -493,7 +630,7 @@ def add_pad_token_id(self, id: int):
     gguf_writer.write_header_to_file()
     gguf_writer.write_kv_data_to_file()
     gguf_writer.write_ti_data_to_file()
-    gguf_writer.write_tensor_to_file(tensor1)
-    gguf_writer.write_tensor_to_file(tensor2)
+    gguf_writer.write_tensor_data(tensor1)
+    gguf_writer.write_tensor_data(tensor2)
 
     gguf_writer.close()
diff --git a/llama.cpp b/llama.cpp
index b7ca6db3c2fa5..0c3d412e941eb 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -676,22 +676,21 @@ static const std::map<e_model, size_t> & VRAM_REQ_SCRATCH_PER_CONTEXT()
 
 // default hparams (LLaMA 7B)
 struct llama_hparams {
-    uint32_t n_vocab   = 32000;
-    uint32_t n_ctx     = 512;
-    uint32_t n_embd    = 4096;
-    uint32_t n_head    = 32;
-    uint32_t n_head_kv = 32;
-    uint32_t n_layer   = 32;
-    uint32_t n_rot     = 64;
-    uint32_t n_ff      = 11008;
+    uint32_t n_vocab     = 32000;
+    uint32_t n_ctx_train = 2048;  // the context size used during training
+    uint32_t n_ctx       = 512;   // the context size used during inference
+    uint32_t n_embd      = 4096;
+    uint32_t n_head      = 32;
+    uint32_t n_head_kv   = 32;
+    uint32_t n_layer     = 32;
+    uint32_t n_rot       = 64;
+    uint32_t n_ff        = 11008;
 
     float f_norm_rms_eps = 1e-5;
 
     float rope_freq_base  = 10000.0f;
     float rope_freq_scale = 1.0f;
 
-    enum llama_ftype ftype = LLAMA_FTYPE_MOSTLY_F16;
-
     bool operator!=(const llama_hparams & other) const {
         return static_cast<bool>(memcmp(this, &other, sizeof(llama_hparams))); // NOLINT
     }
@@ -1023,7 +1022,8 @@ struct llama_model_loader {
     int n_kv      = 0;
     int n_tensors = 0;
     int n_created = 0;
-    size_t n_tot_elements = 0;
+
+    int64_t n_elements = 0;
 
     bool use_mmap = false;
 
@@ -1051,9 +1051,9 @@ struct llama_model_loader {
         for (int i = 0; i < n_tensors; i++) {
             const char * name = gguf_get_tensor_name(ctx_gguf, i);
             struct ggml_tensor * t = ggml_get_tensor(ctx_meta, name);
-            n_tot_elements += ggml_nelements(t);
+            n_elements += ggml_nelements(t);
         }
-        
+
         // print meta data
         // TODO: make optional
         {
@@ -1123,6 +1123,10 @@ struct llama_model_loader {
     struct ggml_tensor * create_tensor(struct ggml_context * ctx, const std::string & name, const std::vector<uint32_t> & ne, ggml_backend backend) {
         struct ggml_tensor * cur = ggml_get_tensor(ctx_meta, name.c_str());
 
+        if (cur == NULL) {
+            throw std::runtime_error(format("%s: tensor '%s' not found", __func__, name.c_str()));
+        }
+
         {
             bool is_ok = true;
             for (size_t i = 0; i < ne.size(); ++i) {
@@ -1332,7 +1336,7 @@ static void llama_model_load_internal(
         }
 
         GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
-        GGUF_GET(hparams.n_ctx,          gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
+        GGUF_GET(hparams.n_ctx_train,    gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
         GGUF_GET(hparams.n_embd,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.embedding_length");
         GGUF_GET(hparams.n_ff,           gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.feed_forward_length");
         GGUF_GET(hparams.n_head,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.attention.head_count");
@@ -1406,22 +1410,24 @@ static void llama_model_load_internal(
     }
 
     {
-        LLAMA_LOG_INFO("%s: format     = %s\n",   __func__, llama_file_version_name(ml->file_version));
-        LLAMA_LOG_INFO("%s: n_vocab    = %u\n",   __func__, hparams.n_vocab);
-        LLAMA_LOG_INFO("%s: n_ctx      = %u\n",   __func__, hparams.n_ctx);
-        LLAMA_LOG_INFO("%s: n_embd     = %u\n",   __func__, hparams.n_embd);
-        LLAMA_LOG_INFO("%s: n_head     = %u\n",   __func__, hparams.n_head);
-        LLAMA_LOG_INFO("%s: n_head_kv  = %u\n",   __func__, hparams.n_head_kv);
-        LLAMA_LOG_INFO("%s: n_layer    = %u\n",   __func__, hparams.n_layer);
-        LLAMA_LOG_INFO("%s: n_rot      = %u\n",   __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        LLAMA_LOG_INFO("%s: n_gqa      = %u\n",   __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: f_norm_eps = %.1e\n", __func__, hparams.f_norm_rms_eps);
-        LLAMA_LOG_INFO("%s: n_ff       = %u\n",   __func__, hparams.n_ff);
-        LLAMA_LOG_INFO("%s: freq_base  = %.1f\n", __func__, hparams.rope_freq_base);
-        LLAMA_LOG_INFO("%s: freq_scale = %g\n",   __func__, hparams.rope_freq_scale);
-        LLAMA_LOG_INFO("%s: ftype      = %u (%s)\n", __func__, hparams.ftype, llama_ftype_name(hparams.ftype));
-        LLAMA_LOG_INFO("%s: model size = %.2f B\n",   __func__, ml->n_tot_elements*1e-9);
-
+        LLAMA_LOG_INFO("%s: format      = %s\n",    __func__, llama_file_version_name(ml->file_version));
+        LLAMA_LOG_INFO("%s: n_vocab     = %u\n",    __func__, hparams.n_vocab);
+        LLAMA_LOG_INFO("%s: n_ctx_train = %u\n",    __func__, hparams.n_ctx_train);
+        LLAMA_LOG_INFO("%s: n_ctx       = %u\n",    __func__, hparams.n_ctx);
+        LLAMA_LOG_INFO("%s: n_embd      = %u\n",    __func__, hparams.n_embd);
+        LLAMA_LOG_INFO("%s: n_head      = %u\n",    __func__, hparams.n_head);
+        LLAMA_LOG_INFO("%s: n_head_kv   = %u\n",    __func__, hparams.n_head_kv);
+        LLAMA_LOG_INFO("%s: n_layer     = %u\n",    __func__, hparams.n_layer);
+        LLAMA_LOG_INFO("%s: n_rot       = %u\n",    __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        LLAMA_LOG_INFO("%s: n_gqa       = %u\n",    __func__, hparams.n_gqa());
+        LLAMA_LOG_INFO("%s: f_norm_eps  = %.1e\n",  __func__, hparams.f_norm_rms_eps);
+        LLAMA_LOG_INFO("%s: n_ff        = %u\n",    __func__, hparams.n_ff);
+        LLAMA_LOG_INFO("%s: freq_base   = %.1f\n",  __func__, hparams.rope_freq_base);
+        LLAMA_LOG_INFO("%s: freq_scale  = %g\n",    __func__, hparams.rope_freq_scale);
+        LLAMA_LOG_INFO("%s: model type  = %s\n",    __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: model size  = %.2fB\n", __func__, ml->n_elements*1e-9);
+
+        // TODO: print number of tensors for each quantization
     }
 
     if (vocab_only) {
@@ -2310,6 +2316,18 @@ static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
     return false;
 }
 
+static uint8_t llama_char_to_byte(const llama_vocab & vocab, uint8_t ch) {
+    if (llama_vocab_type(vocab) == "spm") {
+        return ch + 3;
+    }
+
+    if (llama_vocab_type(vocab) == "bpe") {
+        return ch - 32;
+    }
+
+    return false;
+}
+
 static std::string llama_escape_whitespace(const std::string& text) {
     std::string result;
     bool escaping = false;
@@ -2446,7 +2464,7 @@ struct llama_tokenizer {
         if (p == rev_merge.end()) {
             // output any symbols that did not form tokens as bytes.
             for (int j = 0; j < (int)symbol.n; ++j) {
-                llama_vocab::id token_id = llama_byte_to_char(vocab_, symbol.text[j]);
+                llama_vocab::id token_id = llama_char_to_byte(vocab_, symbol.text[j]);
                 output.push_back(token_id);
             }
             return;
@@ -3373,7 +3391,6 @@ static void llama_convert_tensor_internal(struct ggml_tensor * tensor, std::vect
 static void llama_model_quantize_internal(const std::string & fname_inp, const std::string & fname_out, const llama_model_quantize_params * params) {
     ggml_type quantized_type;
     llama_ftype ftype = params->ftype;
-    int nthread = params->nthread;
 
     switch (params->ftype) {
         case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
@@ -3399,6 +3416,8 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         default: throw std::runtime_error(format("invalid output file type %d\n", ftype));
     }
 
+    int nthread = params->nthread;
+
     if (nthread <= 0) {
         nthread = std::thread::hardware_concurrency();
     }
@@ -3669,6 +3688,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
     }
 }
 
+// TODO: after the GGUF PR, this likely won't work and needs to be updated
 int llama_apply_lora_from_file_internal(const struct llama_model & model, const char * path_lora, const char * path_base_model, int n_threads) {
     LLAMA_LOG_INFO("%s: applying lora adapter from '%s' - please wait ...\n", __func__, path_lora);
 
@@ -4876,8 +4896,8 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
     return 0;
 }
 
-int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * str, int length) {
-    return llama_token_to_str_with_model(&ctx->model, token, str, length);
+int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * buf, int length) {
+    return llama_token_to_str_with_model(&ctx->model, token, buf, length);
 }
 
 std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
@@ -4894,13 +4914,13 @@ std::string llama_token_to_str(const struct llama_context * ctx, llama_token tok
     return std::string(result.data(), result.size());
 }
 
-int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * str, int length) {
+int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * buf, int length) {
     if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
         std::string result = ctx->model.vocab.id_to_token[token].tok;
         if (length < (int) result.length()) {
             return -result.length();
         }
-        memcpy(str, result.c_str(), result.length());
+        memcpy(buf, result.c_str(), result.length());
         return result.length();
     }
     return 0;
diff --git a/models/ggml-vocab-aquila.bin b/models/ggml-vocab-aquila.bin
deleted file mode 100644
index e06b39b5a31c1ced47e4891fc892bc956e97bcb5..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 1963875
zcmc${d7Pa^mH2%G1X0m;$NkD4WVwW}gFNgY+ZDnh>%HmjBn_SPolZhPWaa_fN1b*V
zmvMVY(n}yDHwh#_2zl(>cA3$cc3)?_qmDW{YLDBD>%8By)DwJv^T+S~>kS{0Q%_Z$
zI(6#QsZ*y;)m?eZk|j?Ok*T-n-~MM0<mm(W=kzxYE*}_KymIjNm2Z5Fe%3;-63KTd
za1j5O3LGqw8E^<$1{^9<7;qRx3^-h*XTT9u_@vgFC(>uYks>t%UM*5L;3yCpaI{Fn
zfMZ0O1{^EWHlR*h1{^2yHUs91j2Upe$cGGA02v0HKwk|wQKVzQNg|IJa55AcaEi!h
z3^-L})_~JQzHY$j5O2U4B0n+!Zp;~QrpSLAkkO@S?blf%`xtPx$kPlsN936XoJ*Gt
zI8WsH2AnT)paB<%yv%?L;f4Vhi5z0UYekMQ;9`-Z47dbF8E~n{0s}4+IoW{A88rhM
z@X&zQiJWV|6(ScJaHYs41}ucV23#ew(15E&t~KBqksA!S7S0>+dXWJGt`k{m!1W@-
z2He1)7;vM=N&{{Zx!r)9Mc!yYlkqa3C324ei$vaJK)=Xa3>aW24Hy)8hXIR4-f6%R
zk#`%glu<TdNaVc++#>RR18x=hfC0k@fC0-y{?35qA|EwiMC9WJw2=)1{z7Dp0iz=8
z3|JvDVZchn#(-5K_Zx7V$OZ#ei)=LDb|lJxJ4Ch^@CK2G47gKdn*nb`*bKN!<S_&O
zQsnXJy;NlG6M4d+HAtdEYx(ZbI+3nJ<07AOXaYfXXj0_w9hwsPoI~qH{?VcPkz<D*
z5c#4*8$`bB&_<E3I<yHfcc>%sO@}s%eA}TdBLCvhR%U@i4~cx&p@&7j@6a}pA2_s~
zN#f8WB0qNMQIY>}=rNI>Ikba0<j~_Hzi?=$$gdoFLgY6N?P7`<v^y91okMqv{K27n
zME>N^y|8n)@^tr`&{Ph+S!6GV-XgNMLvLk*I`lS?r#kd@k$oL{hsb^o{WbH~p?8Yx
z@6fwMp6<}QMV{f%-!Q!$8WTCdq4$V9%c1v*Jlmo7G5Z~QzsPeO`dg9bI`jdN=Q;F2
zl!HSb5_y3`9~OC`Lw_goB8NVL?r`X%A}@C6V<Inc=;I<Ub?6hQ7KaLvmpg>iy}}`+
z?v)N9bv1{Ox>q@b)E(pyQg^UJNZlb0A$5m3gw!485K?!zLrC2b4k2~(9C}dXNQaQW
zS389K9pw=6ceF#u-!Tp$f5$q6{M8*o{*H49`J3+$@^`#L$ln5okiQcgLjF#42>Cn7
zA>{95hmgNh976t1bqM)8%^~FPbc4FN$QceHf3I-}`8(4g<S%mw`8&%Y<nL^UkiT;r
zLjKNm2>CnDA>{9ThmgMu976ssbO`yo$RXtKwGJVF7dwRfUE&b(cd0|j-(?OVf0sLi
z{52dx{$A%0@^^(p$lsL?A%6=ULjJCD2>H9(A>{8GhmgN(9YX$I?-25LokPgq^$sC_
zH#mg+-RKbVcauZN-^~soe@%ywzm`MD-y(;QzkY|1zX6Akzd?tPzr_wAe@h%f{+2p~
z{0%vT{N3UZ@^`C4$ltI-$lo%DkiX>)A%7zdA%AU$kiWlh2>Baz2>Dy#5c0RuA>?nB
zL&)E44k3T59YX$YcL@2r!y)AF4GtlHcRGapz0sf<<nJzrkiWll2>JUfhmgN~_dddk
zQFIH*9p1fe8Y=GIy|;Z|Df!)2WcQ|N8og)tzI;}=yHWCcqEJiZUc<O_P;!Tcma|rS
z#kJivk_SB51<9Ko32=8p@;6(}(Lt7UgGQE1{uZyfZkoQob@x-95=H*j-A@a0jbwS7
zRT&(W+@aP=k+%m`D<prrXG6<7;^T<q@9^d!tBctDwKp}<eM0hp$GZ&XJEOvC$={i%
zf&cG{&;63W%Ri5I5wCY!Lqjwf5On;FcL&-W8XUk7ae^o}7PY6L;XU@XMaSQhz9MYz
z^{>N{zc(6fi@Yy2dcWiWPfau4?~fvjC4avcSvQSn{B3+*B>CUk=WanW@&Wriw0w!=
zA4muh+YdTC+1)02z!P1D;6qV$1neJjszyaV9A6pg{KNKjWEIxa9`3G{{O=s&INX(w
z1T6!SCvX`?Rw}VU@_^SMF!C`cQKjKyUSx6_{(syW8tRw)<My>%tQ-;fgqN7;!j(^0
ziDBfaFzD!`C3mQ`Lgc=Xrq*)ExX-Jtn`W}Dad>KaljLi>4@?B5WUc3n&)h3{z!P1j
z$T}|thqTT*Yitx^#&`EUNk&ypeR_=to@tcEL~3oL<P*V{r6QA2q`On{fG4^TH)Y@H
z>r%-bW|lIO)?3xV<;>Fc2|LB_H(L7ndjFoDptK%H6hF~@K=Md2Hq)rZ4XOTi$pfC~
zB3KU^u0e1GG>jb^<S=RTjX^z9wJA_^VMl^hmvjup3Q#yS!mw}lvJ*<k=0xOUI^VZ=
z&id|p$+rY@+T5BdKwrS)%7Tad`;;n~hblx}di1b0JH-5XIM5&y+r0QVE#(yqYqkd(
ztm{1BaYgb-Dvx?dFqFxo_PsmKB+DQ5hM3-udFk~^Ou*wPO4*SJdqjn3hnJd|K@&gj
z@T87-z*EyGl%0li1fDrG1kaxEvQyI!NgnWcmz8Cgx6fqW<?SG<)2x1FbeoLQirfYR
zJ{gU7nQx!8udC2i)9EXOPglh1aO@70-G@};M+yxw!-CR{1}>63feS@u9M|~txa1B`
zPjp$TKV{#$g=)!9dG{FCPdol8E0cfPzPIS?XQJKdb+9A<jN_c_GIu^3tsWuyXYKRR
z{9#|?MW&`7mi+Gx&F~!}|KJGwC3k2QI^c7GOc8(1^g(Oo5Q;`V?-}FMn<f9eQ@EPV
zk$vqJi;(tzbaV{$KN++N?fy@pQWlS(6lWd9gsRZl#K;}6LcWlY?~?or(S<A7X8Gs!
zO!NGUUV$k;jdAcL&zaJR@TG+6k!j48FI!600qb4vsoiKz)v{knIPULmmHaD)qg!b2
zzFOhXvHz;m&-bqxRIf|^HE(f3HPqJ=U6A{Auc<Ad5x(I$Ci($SGL7Y%3I9%AOTQUi
zJWJ$T2`vN#Jgx)zZLcupv*_EYLA3loN6z%F?heT#g~n6s;a{v=w@`iJ@a_p_m3${P
z{|Ke>?*x%BL3&2YZE7Vr*e%Wx`ByJ7+1(;}!0U7+{jLx39g=@HO6WBBo?}PwQI`4l
z4C4^$SibKW*bK_=?>kqK>wojg6Ld8Hw<t5cNmrU5SebsL=?Cfib`_8xMva+V@*f8G
zj$>!+pVv<_H3Od1{{Fk68DdHP_tX?fer#V?fc(dau={m5e-a4k)=xZRYWhA^P5+T-
zcvRKOe*~&&tbw0ePFlTEYPI$HXWlp*{h8O8RN4Ev!QDbFwVyj)mSh+#bJ0q7rz*@y
zvA$@(|1-78vQfd9$-jts)vq?dFA{>yEVlAXXCbTMFa0~CsM`8hK>#a#{;TNmK_b6S
zmDj7<|FyR_rPkkXcGv!wwK}cY>0ilzW5`w_Lh@fhuByZTa>Qy5Bboo5awa4XcxoE9
z{MM))!Ql99G_9uD?;IZQqOX4EcrYteKz?sIC`u&!_nxzEddKtzJQ$urC7=fKNR3Z-
zP+0Ou&l`u~Na!D<_m|@ANMt=I`JaNPZ8cQ>C)L=XGM!S>>$ES@-NzWy8kS7DdjYpt
zMgm$SrF3m0fbGf3B~P&ugUcj)iq+da&ia!(+$~lQN}0avvay%fRl}nrS){P_b*k)T
zm{u*5axcd;IgOC+ZC~4M$@cE<ZFJox`QCwMdL!Dh+&c(f&79oFijNLSwvQEWtwbRA
zsfshf%1E)G=<Imvp5CmN?5W+SS{tiXj8ftbjN9_5hIoD*-%rrdkZfNkbp14@Ou*x;
zcjdmRn_E=b?dK^<VNqxM1)J3Ti5#p@m{aZ-h!4S+WUc+hD#@M}og1JZ4)0zMn(}F>
zi48gm`#aJJCUUmF=cozGu<vh3M+a9(xxXPDVHtdSDzXz1%${!F+ce<NXiLhcCxnRC
zGrTlLIJ=|l87YUA<Qay1DRZNAuv?ssgDXgz)^#F#reQzh49TDAH7A*UWt5rdZb!|_
z0bXN0T49Z32gD#ORAq8NR6<V61FX3r)W19{wXqSA&7Ng#EM?L>%ZdzAq>LgH(_6dj
zvY%~uy2aq_l07>rcPCZ)J!kg>ohzSh)mvEA(iat2XS2RN73j6+V1_>-Wx$h2fIKJY
z?#If_p5t_{?~Y6U9K+fzE|jwRycTDkV`YM~r^q6O0jJB-b5pl>O6f2wIV$LRUWa7>
z`N<;11X6?MdB%m|0V(4<>OuFj&yVIuC3}AA$QIRM&+k4BX;&Lc|413R6WujxhrS^7
zV7p{5Fx0vSc|pRk5u(c~wT1ncywLMh@|fK(3|<aM{z6AJ!Iq$Om_>XL=VVZJgJdrX
zyy#O_JI@##MQUFZWT^A|Krf`qk_HYmqK4212U;0$GEoooX4MPv;#B+zG-&o>%NSh2
z%yzI_oGayvW6)aahImQPbG~FRG2|otlE1{Uu+4yHWu(UGBa6sOqv1t&NcK{zJTyEk
zxqse0HN9O*|2(QYpO;yQ5n9S#=19k<pHS;BQZSB<Zuv6r@wje!Uv6aFN{yF$4FpPs
z*1wOmrF^-Uo>cYvilF}*$s8P7$*TW~)YdM9u}mq19|P%?UVBRU`O4J7dfmoVDV@b-
zl$%gxQ}b@DS}s{FP<J=5qsdbWah#B{N=+!IUKM0wMrH|~UMIQ37!D|%SH)N?XY(Tm
zCDc=RS+awoXY(ab;1yCH6i9V}JlM<Q&0(d_4z`?Swm)`w_c(%89vpZs5jn&w=(y5A
zq)<+*!}%dbNx$t4%0s-BDYeWG4Hzw-9U2s}d|~pZ6nte(I@H=3(rwpa268wm*<q>1
zW>iuhDWn-Yzl_urYZ`{b;i)$dKy_B7Hmhzp++^a&Bc(jtsOc8VY37K)-Mydovm-2L
zXo>2?BcjjD@F=7!?t}Q*BjYm3<|WE@sGZ~)Bco{1dEVg(L_&^?@@GhPWNPgJm6b@b
zU^7e0Na0UmIg(d<N!Uu0S)?$1;9wr9_1*25-({q94>XRO-|Mld>rP1Kscx}YCsKZt
zai9%(QEH-VzPO`}z|lcieYDd(2{ZGf9Tgib2r46mpuk!=#;UaFYIclQ8An9(W2}Y4
z(Lcwe%Ij5^9&5!1ndI5ADPxW9IwGZ35v%jDj%xihYQ7#EWP)aOtE1aD-3ogO!Q7y1
zixkegnR__1@LE%<<^$FVn;+MGCOR7{i;V+IkZ!oj<D3#!Gac3Wp3-dzT5rB}WYvJ=
z^9>_Pa;cOKV@Y7|9q%oT>(U$W1R|6>+%1^r<?(?Nb%mZ^5SZ9eXA2C!T3HKHnRU~M
zbEMc;>BjMdkeu@*OW*~PpO7fk9+xLL9TRF0oai}|)02o-7Afs5!j|&bP@s<^#l}^I
z?Ife7&FGxu7}z^89{EYu{`~rJQaU&~$og=ym%(a6)n+GKnISBtJTk_+TXir`cEYA$
zr<@X9?_VmJ!y~I$adU@ZD7x|#BjT8YI37wJG@W#+H-KH8pXw+l5$!Sx>Hd<f&uLag
z^&z?=JIzPi+>z>?SKI$Iue%<lQ+jC@8l8A@di1BOwoDc&CI-_fKi!C0j6GhJvavkF
zuwf<-NOp!1*wPj33`eE=epV#U7{TQsuZb=(7MPB&vD)1NpKI>$?n#y9C^gyLrE}~|
zufs-BRY0Vef6R#d%;1E|%$eR>-8gQgx3=%eyoGhlq%2E#c4A26p0j(B1+>hJ@>RNq
zpXCV0F=4W^yu!pxu6~KLoc}1xGBR*p(5{>ved=!1r6W?hvtz@0wxe6G^DJ^yl`#v>
z3FOEjR`)r^-w}37=Xi|?<-<8nms+gn2JzEdRoEh>wjW!cNU6PsnR~7ym{fs2FJQG0
z&x;19H|cs4DHV+eq>L0hd#nRF--ua)0G%K8nPB<(MohQ3Rm$@n)%xyQMo}*C)cEv9
z)ixJ+l__0nB8OS5y{XEXlj=fer_O{6t-T?H;KGFJ5w-Oqg`ZsaK^H|HjbgJ$iYcJV
zCQ@j7wV5t*63ln_+JteP^5C_h{<=5=Ohgdv*BT*h-Gw;V8Xe%kv->RD%5}pFum4JR
zal*D;J$)Bj%~n4gsWKi_v*;2dV5ndBz?XOxwkjA&d8Bj&WvX9d9l!Gi&QiUUt}z=W
zyENFYrxF!R$1V->)k}-ynk)4(uZizhW&E-jq{}6bjLGQ@*f-^6Uez1`rv6!tE)RpL
zpOc;Ja>t`u5(Dvar(iv{O&K{;(>v8S-0*U^&{az`tOxUsk-QNtV^<`Q-R|pB=||{G
z7AclnU5g@xS6X*xuX7=I1E<6lB`oV%q}0x0u&yw=y2UbPVdPBELA2MEUQ4%`XqGF3
zgE;#0$iZu_!f|ELc?(C+*5s<TWDC8rst#1cLfhDM-{E5~<w754+LWuJ;Pe*S%&xMG
z5&YnJfJ)a@scqa>R|kw5%&ztxE|UCe$AXT~jeDe6gVg7EO|-xmGoIIY)6B*En&^59
zC(1Q3a;w?n%C&~~xVmHxt)lR?R&PLs`dTMpg3in99XW%iswz_0{;J|5g&=iVR$iYv
zft+0DwbajWk7U<*r`3xZ8JxwSnafDwM^deLy&-5VLbqqvN3%4<YU(*yF%ai4vW&@g
zgCXt~$4hoYq7xQL9w{t4+AnW#%5eU2#P8)e@_}{Pjh?ffqm4XrCb|>K;wP2EDRh(7
zQumVXrEW6z^|NaE_ucsS@M4ycLTvQd=4Qj)Ez~%;Ik>NGk^E+(det!2q=Q5FTBI2r
z=x%|TSu-kMgg-LDG+0I{tV~sKEw94~fXaQ#@R}3D!EW(dDIIL-im)g((a{wlQo4V^
z$cfY>lX<I@iz=<h!pIqgIyCy7SL480e`<I`ca7wcGsRGq4zE-fjSLt;t>sv415p7P
zT_<^@)PIOh^HjIMjnO7%B3PwXK{n_Z)W<yN7<2_?{f*QFq_Njq?8P{R%oay6CN*xs
zNU2qU@v_)^Fh!0mNxebGJKQY>MkHV2_4ULa0bSyV)%m_O)S-GIv47N|oh=Po&A49b
zHPtl2w1|`*Bw&&cnZ&eim2AjMVJxdjF=SM5P(C1K<fw;=#r~Ga;V>?{#nnJt@>_x*
zY!ke|ZuQQ}tyXwsm2UZNwZg}sUvKrM)TU)fBc&a~tQ<~cY*Bl6*w7A9Js&nyXYfbL
zVW*lwS!Ts`i11Wpo?<n8Or1K*tn$d{GASbiK~VS9a?e<`M6v*}gmMRmmSE(k0yqdq
ztf>(gG-3o`)uCWUoJuu09+on4Y|qh7^lwsej1)GLietOGpZipo4q+>lk*VetBk>nr
zaXog+qmo5RO+WU>e-XpE2=!d1f=q3yje0e9rMlXVTF<)0Rgy=Bdc@XAS*3Ppr4`<l
zDc$I-aPcvDfOB?mw8D{U>)5=hGy@_ljUB5NNw(7K*-jUwr$-O@N@pw~73E4JdC1Qc
zS0y5!&_P{gt+OV*L-NSsC`@%*q$WA<Waocds<i=@XH}}B18`e#WQbueZ!@CW%!SpC
zx?oR}ISgrRuUA`%{(dP}ceDS)7<#f{^nV8vgXAn6pUEW)cuH-Y+hYW-m-2R}UzdA$
zeurfYapaZV5yY>?kD1@$rN-ePlOa+Q-Hlr34Uxxj1giH9(YY42QVz5Aeku2)*rmyx
z(JqTHe2Wyugl;lY3dM|*_fDrtt${Zh*c!!s_r|1SvA5Wgn?87B!nqxtTt-UG5Vp;C
zB`SBSJVi=XKh|u@>*}&t-sMbNucr22dM%YPyfl9qIhbYnU)o>|;?b+hJc10%U)gv3
zc-dbC16!Pi<bM?$zf;QeeW^yd<levpn}p3$%dPUf(<IBiduE@arF2)^%md+oIwPuz
z>X|8Wr)-LOQ$~(%B)3a1QmR*>?(QJt)l$3LTUPsx<?L=lzyiBMnvtU(34CjJcl(e<
z_c`zRTZ$-!+ekM-cL!~1oAoNxkLfY(J=Up#RZ_brI?IUMC)qvT#R)a-aOZ2JOY@#U
zvlNw{-{YN{U|}pAiS7j*ns=1+0-bKdu;cEv(u2#9-I~MQV#$bP_geGEog&S9y?OP;
zuvd!I`tA+{qVQ6S7fX4sv1kza?FGmREN`-}I+1H{a=Ot%Gxs8SdQO_X$xvHx0t|eU
zH96Fh+`rSiM<`XK6a=%LzA1HYhxEM08TcUH>_n<#YKzp~>}AH)G>sfK>?qM@Rc@2I
z%-$U1x<HD^(A^>amPirjf+^SAlY-gumKcui8Wd<PQYvv%lD)+m?iQSBH6sU%>T7+A
zb*^=r6b_Bz#P`y>Cm^%;79(Xu&kNpaU`t2jt=;FUMhg>vkN%SEt=<wj(tNAozWGkV
z2NN6Cuof1MXb97`d}|UkWTEHZ_1Ib7<`r;@q3mj&!s1l_einI?cJ9)AoAdM@7?Quu
zC|rcORU`yh(`?XF74v%ZB87iY?Z&qU3A%-?y*-$S6f=$9Zn#@2gw40R;LRtvF?D8>
z&Q&jis~a;UatJ-5cJJ-ZI@FuIBf5h*GbOcm#AvKQo+C$9*Q3((9873TiGU{%kK!FM
zxv0M*8Q4|LjTqRyDz#0QvA=f0hl#|g{dIJyyFopYDTO0sN}7=xX9K`e8Y#S<%&5X~
z;DJObdX9Qd2k($x!eJX9d1v%y;Zmu+Gx~{T%Xx03R9SVTnNrZPj@V9P&L9L9DFZVe
zeO;v<f+4*sHAy9TmyH2ztG&w`!u_Pj2a&>_Knj}gGJ-}{Em=yeXzKiS{5gf^pn$PS
z-(>_0v4!a+WHa|Lk>%adpP|K4dv`R0alnS}-BzWAsnLuKw6ZS1RcbT)-y)@0L21X)
zVI*&sUdmy8W2OC@U=}B}Tc!3lhP+#xAX&<o$uYZ|p3%a3%Torfgw0Y!iV!yXR{l*A
z5!^#~Le<&ci>m6F8#C4+i?y-D5<NV~{5yND=9nX}7&eSD-7Mx&mK2L*F&3N|yiLjk
zuV_inaolwm0kzSIi>3CSU>yhjG!rRZ#db*ZJ<*f~P@qjMj?=3S;ma=G6E-W-HpC|P
zJ=QEn@PzaN&jj`jYteht`l3_!z1G8HkCp7b-n6b|+ok#5pn4gHab5-`QutRawMi-c
z>v6|PFKSLsZ-z{HpVcHnZH3g{m$<f8yAvrjUDr#q%6kAKBllX&(j8K~&-gL28vEW0
z;d;>%$lm*m*u&X}zJE_=$*8^G+QdQ5*5duq528-CO4D=LfU?@9oarZ+?M0+G$v|0_
zUaqa@M?HrLkmLOGZ=(my&2g!DiVe9PzN9=t0Pd4!r0_wi*_%?#Vpi@*aRSS5l%DDq
z1Zl9qMjk6Xi=BKR_&z)&HHV1=!;tzwH9BB#Mrsm;%fc5aJ(PocA8_vBPb?!tJ+^D4
z_kkGqWg1icLB}`VRdrmeQaja3{-Dvogh%K;xJSO%?~a51585!?zEX;)H-&qYb@oZA
zZPH6A_KvuBKNNgKk8PLQhrE|>@4j8KfD0manhvvu;(+;(w`|1g%tB=pA4;nZgI-oB
zJ1FH|q&Usc<~|(FO>d)G?ZeIz-5WaC;&i$hWz>3siIGw>b4rR2$1p4(k@CYKYa`~8
z=mj#)`Q`5t6E%#p<|&Q5)J;v~O_+DM`FDm7oy9p_sx@;r3uzH4-PPgyh!jyJ3`8%L
z>`tkJ_aojKK6aE$EmGqsSzP&%Vk^%;G(Y03Q+sk#9M$C?F%$OxZG1@a5gW|KY|cL7
z!#>UqruPx!*dXo@`KW;{B9m(l-ZqFi@lhv+xE^L!Gg5?<pnLO=M)P_`WzECTD4}zo
zQ;CnJ{k5G$^*$QhUV_KxW6?`Y5H<smnljF1AB%D;+S2@3>LdOmwVY8fd6ecDLxrWP
z`w;6!87X?l81x(yiVR1_$Gt^We7mIfamS0X*nOP!XBVm}ixPxbqx_pGg?eW+Qwklu
zS&B&M=?xRT%2O9p@8gNTESI0~HjzEtSG7;r7!Y`teIk0Orh0-`5YX`nr^@!*Oa=|$
zD?VWZGdjpr@S<qXj`SSat%zq~XzxG|7v3E1ScuG03Kpa33q#W_u9m!r>9u<4h!jPT
zawVeSF!Pd$=H*w7qD1bqs>9rGsNI)Zco-V8`<z^hI6(sRD6#2<TO7dVsW1fWlEO2(
z#pzO385`KB_3ktJk2`_u4vAeG*c#U&rJ4cTEK&sOuqriENlt$VR!eyt<~%G#l{b!&
zR7Ofq7&b_6O?17r9K&<1H^)vAl(n@!Zj*ZS5IJ}~7`JAm^jHGlPnFt5tzxaSK#6>W
z5z~XgGD=`V(~aI*<DN!d$vQ7=p=GLNStV-gVve*H>3%X>XEb5WFGagFJ;k!N4vEYo
zrN<~}lyye$=pa#GW#p)#yh(bVgJK`&-iJ474lm?f&#h|X(MD@njqE5vaAbEKLCbm`
z+4K@v73<Y!kjk$d!ZA^dJD)W6r5v}u%_mH|7gf{+LA2<Ei_FMssZGS7vPr}H5h+}a
zP}+<XF}LL9k<!_?Rf-8?%x=38b%c}?-KYPVpGbNMF$N<$mC2CfLvE1Tq)F8EnPihj
z$KdE~(wuZo*wGdut_SP+q|?S`j+L|UYO7fb;yZzux_$8PIM`*%aS;DB{TSTTc$nIh
z^Fu?4vZ?56cMVHhbIKZTt!!~W#_7O4TvTQ5XBjWOHiz6Rq~|#nK~G!j6RypSeJxUi
zMbKupKKg>xVIvZU!WJUF*-<Hz(E<o798-*Hu|8Ntm~H7fMDA$>dT+fk?D#sjbW*E&
zDXvzfHZt0orwG#+K@OUcr)N4i^ioOI)p02zg^y|GUMcUlc83<DKzjE(<vPYYm@js7
z;DPiF>7+NRMQaZ@9V|S|q3nU6d}NU{BZrtN_?JgY=frj?BBgT6D*HeT{xa!Rl{T|v
z$Oh}~4L3?{gB9r(hfB60dXBzlsoM}O-pa8`<fvWEo;FgfBic7F$J-KcaYHDVA-22n
zpkW?WrTJju>3UZ5EK)dQ;A!*0sI$l}GdUwHWQZYDbEAmj8lk}Q_Mnlqtc4);B8T8F
zUEVfYdo5iC9Mo$V*+%QBvTb8Dv_eByH>SRBREJQN+RQ8{Q)=d3ES26yYmfs2OoUBo
zOf>YRwkbHYnDwm6S;J^GH>IXDa5u6vl4!jYRZ1^sl#!amFSS#86%xH=B^_@}6%s0|
z<D6k3ow>)vh8<_e2*4=7{_j}t`}NQ)Dyqr*s1%;jEe?~?K?GiU2{|(u&*x@GZdX4+
zt{1*)o4t88WblSYiiJ_P0G=WgemU3t^3B$C8;4Ed;OO9DDK{H&L!+Fg^`f>~9-Jp_
zam<sL@sCO^QpEG9D{4#dl)FaFE#4(e;5>cS{oj^gZVUHi8MRpyIV0+&R97!_$yTf1
zEl!l$*5DL_oo)5z$MH|#Q;8I3C>Wx-m%Byn!6NF6tCHFpZ4RiGN(k5na(??z%2JQO
zL($~xC40!5VP~%#ixl@xG#dCJ=gsmZE2M}V<ug$kj$nw0pI$<|M8lyT4x;hG!?TC2
zK01@I;fGzo(eDq37SW44YL2pU=MNh}qeF|N2o$QWki1B#`G*DKDf8g4N^i4Dbr!YS
zwiu5UY|XaCxZ$_ISDKN+8Ko<imm4<5rE+LsR>*|mBIID3(R$Y*T<mjPy4}I^xjjvL
zE?VH7ut@VLq2~_xiX)}5iP!;AUL|#so|;!YN^mNRORTKQ-_Ln<Z@YD92!WADs!p(-
zs6|SH$*`~<vCb~mPBc9SRo@yK;EtgoSjw;zQAJm5PDdVb%JqzVQ&<;0M}y5xz~7@`
z+%Qwr3-PG6z>32;@1w>HwgSyZV{j1xSQ#mfv2^-Iim8JQTR!T;iB-{y3?xb~8$Oo!
z`MAnQm3kE3Ws%a=nmtpbCUF+xMtsbOIPOF#s_$DMqKuTDwQ#QRm^C{xJiwhvZ%94q
zEK)nHKmEh21|*9dWem$lD#voj0m_b;BqNI#OR>Y-JYGt#j+J%4^iq-KgIw-=-0<#R
zk20-29%5|^rdn5=V#YVSl7|53LfG^=-Qp}3tjIT2M!tp$7LNy)QHwa<Qi|E|p!6Ph
z0a1s+&QxqO8MU33(Jk(hEHX66=OJnCjDgp~QHENDF5elwT)}B%vD3Q1#R9f1DWk=C
zU(ZqYBTMpxH@o~+sXbxM60JAHmOV?U)w+f?pKykq%>7a)PNe}4NTf7IhrMQ`beoS8
z<cXkV3CpVNa$?omz(MRb@UGO)wOGknq}1Dn^&2U|7+DYVD(`+}W2E#b2{!JEW=03;
zS!A(IV40UPXl31FgIlH6p$@Z1p<C6@obsl(Y2;GrOj(HvjtsRFh#83tVq|+|k_ik@
zGehs=3T=?GlsJG|O02|~%|2=TTh`*~DT8j}%;A%XB|E2gF@Ab~u=u1=tk*JeJeQx0
z0b~vDdA&CJV>%`@0R%dsr=!u922@Pjxd*#b+!JhB*n@&uUsGsK`v@-J8n0nxcgRXM
z9qq9-A|hZqQG}O`HC;W4z4%7VXm<<S8NH+=yPdBZYQhYxQtQ-AjFg7zVGvh&^p8uz
zj$fxVs!A6LuHqYA8e_NHxZWDQm3x`dr)49EXxB%BBDHROS8ra{^)Rktw+4(LGF`_$
zIfIuxZJWAP|L~C(sro2xmD01!pTXQuS<6~%N@uJ^d{9zza0tDV%_Qo%9YZyZa%|27
zCx|paZoKd+!UBsK>(&W%DQAq`LnG|QdkKXGO52W-lj2XsATPq}^C?HiWJa@RDd(J}
z(u^F`i8At2MxaitA~FaTKv|TZ3L5otRg8SELb?Hp<<r3t&e>t8r;LL{)qOhHyNIdO
zj4YOFW?BAeN317{kFr0|OH)OZm|!P%pOlf(89pw(Pg}264Rfw3pD}99OZ}O|uMM~(
zvd<)ra;OPA^txWt3%3R^xf~qCPh5N^C|j)I=79t+0aidSQo0;qvVJx-u!;Rx?XxaA
z9JFMgO&#FMggFWD?Z!7Xd^YV4H~(ji@ZD3mAc{byE(mxVIcvLXrT1B*hXYtFS^4`I
zHS`gOXMgVvnLi6L)Bu?5?}Pj{*Xx>IoOOLAYgQh`Rhbd<X)3Kq#WlvKi~<nJ`L9pr
zUbM(0!@`n(2nphvW$holbuI&-XF;b^GQQbCv^B9=nqH@^hDQDmCaX8!EQLeF0G0pX
zebSB1r1U(68(q(-<#RD?ddjog?V;M|ygYkj_1%0fRMcWKzMGM$;T%{Ok%Fu0lKNb*
zW^|BNQ@wt7Rwkq@7rYknd2iMtSYxzmpSP}$>VqYj!{o5`f8GXuaCkBI^;0$8?9iPB
z94kH_jSlEG;qx(kY_>4qy=J#KiTnL73X@nVHHS5(O0s`Uyt%KtMw<T^xNg(P{Zy>Y
z3GY7!6IPCDP+b|dbZ+AW^HjIEWtGPI{gY!<57wmA{wc<j;GHaT)aZCbnw~=!iGYk8
z^-3eE|76NgBf<2Kl%9)_t>Rp+SCzF(Ewj;KY*(CLk<ydGNy%oNTj3sKV%#sy+2~<w
zV1Ty^d@`erizw?Rj!R}4*e1FUG8=odCRxY9%P;s$#ImY=A<zvizfH0)I4#->TKa->
ze2S$o|3Z>7y~^VAs7R#=MP%jo(rh@<tPZzUs7>?5Cy%)LV(_AT(c0mLkRCN<o<bUS
zBCX9xsSgL2`WJ%+He<yXjsAlVmNLpf8Drjy7Tk}e{pUjDmr_qQs&D&CRul2czO;v$
zhiOq`$(moP<d5n1CFdwhT~Xn9Oik^`VB6mdB+4eX8>T7c%Ly^pgVTKYa>yL6OE#_9
zmz^e@)!=JJif~j6=e)|>#4=x`iY$8A1?88WCUtRmIS%4999F(!8G1{+_LUeQT=h=V
zqmq3k*t}4KESq03R$@`&ee)cajxFqYi&T#0Q0QMVcA!m}#=R)0cOTikeAQ5lSR~9>
zqjN2D^<-a-&T&X{w=}ENMh(@B6b?ZSB#M;Bc_9(3Uk&E#eV<;!IWV*sCyIP6`lJss
zn6h9IiErivwUv^U>ucUNN4|P;7b(utbaDP#;O`&74oHcmY?#U@r-vJaUwkdNa}Di&
z-NXe!u6^CUGM-$5__|Z0r;VHlvJFKAd_A@NfV$qi^pKv$7ExMnu5%P+PQdc(K}!o;
zwii|Ptcq3R8<91N6;b<!wLhP;{cm`ix+{MGDs&n*zmeLU)H@E}FcIh$Crc3->L9{8
z{)UmhVkPTP&tVYY+xez<NpGgszUioRNrjD`vszE_nvtW1472~6PKKIxyW&-`BEGo|
z;ujRkZ@QqeQ1((YtTFPf=-dEX=i0Zdj7nPeE$io*2TAi=MhFU0gIn`&8Hppq+|Vh!
z_y`tA>4gx4U2Lv;3F+YNZQdw}PR(#Y81Mv=oqgLosJ9+i96YC65WSyQIU6xsif<cH
z!_2Aj+um1ua)4phSC@KGSdW2m<^OYZr_Elh_RmpVqg&W{Xgpk&3XH=0=07JAo`9g-
zQ{3xWjtCZ!r`{?wXr$Dohb0v$ja%Yc*1tFw4ZUIW<SEqxBWUe^aT4tKg!2lGuWSB`
zvG(+1d3?pF9U9i_#f5*qk<){O5cR``K2quu;4J<-RvELi_8sTzgf2z_k1P7`boaNj
z2b{DfFL`86LTm9Irzd%|?eficoPIqI!m$}CT(}yY*$Wox-5Jx#YoylmhS#fazZd*o
z!Q5!}43S==WpT+p#V|d>uBu1{+syT{XPw}vp%?hwuvD{WBhSiM?huOeU%ho*eV)+P
z@pWd^H#J3lYt;N#6Okb`(f&0CV|u$TOaE%5j;O!0j6x{wURCG`q*cBfebnt=S{}Y@
z{X)~}Uh%tzW#t|0E}P$V;%?_{8*8*%U=4mZNr0ZrmQ_lld3u3eIsRB+m+$$w52!Y(
zeJ@qkNYm_lPbNm$!rZ7zbo6ldd(QrGlztI88upFnvnsa>o!Ik6R(TBd_q`F~r8x7e
zecv0I(7*>KhTROVzUlTg&Jw?GO<@aQumuJUsM#vT_oMTtNcnx^$Xz!`&!G|ah5wdD
zLGM%6A~lKD=UT$QdE<85X}3D#Ff6<TVG4SOERPZvaDaYRcX5$Q^y{7BGD^5$xJRHD
z-+bkFBTIhZbkG*lwiYQAhZ_4;ie0uee-MJKoBjL;(Yqlje&E!qgMK?~(VMg753I+$
zfYFOwojT)4y51C)AEp|-8bpo9bgT2jJ^WZFS)`^|3^-Pd6uvke-5(kqBg4b0gMJwN
zzu|f*1Fh;1_LUZ}-}Azw79b}-GOCU`NNPVa@<)06Ci_t$hBGGiMEX=q^GAl2;5T~e
zmF6#y;z!{mz<9AcJU<GF!JmV3!^_RPlSh-RnPZNU+P@o~<+i2x_h^S9Uc->7qptby
zjzPC-MAk)09bo98NU3tb9hOq89uG?|aM=qDFjXJpksmvLdkZNvLG8!EomRhO4)+go
zm%aI8Yga9L4e<7q_FX4T%43)Zr2MflYUmB9gdZE<NA0G+{3IyHvGah`eq!wm*#)l1
zVK$>)nyDO?Grl)3huw_3=qFCEIvXFM4|?Ui{E4@QOQrV{m!Ks)9_3AGTpkvbryyOe
zpe*v#v4Au9KdjDT!uR95Mt5S6{ztIkD(u4qufVW~Quv-ARel=sy@Jq;+E2YvuB)kY
z;iraRw+4DPBS%koIFb0NQM`DFwZ5v4L#h0!_fAJ>XXvzEg>wf)$<Mq69X)*9k<z^q
zi`CD(J&k#3{>)n7Y@cVaLI8)Bn~%EiVz}bzS0K>Y^nmmt#T-}l{c~>$ZNX-*_VeiH
zavng+er_z|+^hNX?t%Ynm~*$#|9ET5ktj>?o2C?UOgv_!boau3^yfyg3WLn0PTCRU
zoHa&(U^W*cvU^OLsSK18ju9!1@4zP<DeV=_R3$et+Pz5ORY!E>Kcl6YyU|>=l)_5q
zB;-GXZ^(>Zp>O`D(a}$gq?h9Ym_GGWRJC++mQ`xAwpNwY^MCn;@lYTAz|8(d>f~1D
zZT5>jBR_@m){yjOlwdq{7DP&wGM=wrxSQ3~F_NMYt$qYhR(SPwi=H>BUIW}Fzl<@w
z4IfJFm)<_n^<d9_Y24Q{>*g<=6g?Y6wOMwazNyl%@DQS(9RYI+AC8oPj?0i3;=Nyn
zt+aA!DA8YeYn<yVqkm<sEz_er2fM`^rRilf{u8$5zcO|c8d>}*O);F><*!TzaKEyh
z@FFJ@u<+}|%xy?+?bn8Wl}5t<ItH;lXvRu26+|?M1N^nmm~m!k5tY;&t_rP1cl11?
zTO7liY}O3xwz|4&o`U;pGz`j9_9P2T{7tmG*jJqVH_lX*b*z<Xy}(iY#(H?rQBwZK
z=peF}O+=KmJF`fFU4Z{D@1EK^#F|7(N1b@J|1vD|=S$O}1sM7cEyLkl)x2+dT*^pk
zln)-~Dz#mw`+vu<v)FEp6|MH)M%U;nHP-$+SUjz|yBVcfHEm;P!SVl{cE?LvylzxP
z1>BFghssnP2_huV5f3d_XUT6}^fZLnL3J2qzx{J^^d@>UzNz*02qfC!U3@cxl?75y
z0E*u_**Zw&Z;d@C5}WkfBqR24!S7-ySF7Dq`<>Mo)G(N=%F$=Gnx0}FJTODF_wS;M
zt5)dhTtudtQX8fGT_Tx#NJxd=@98;W4wW(9es83-ZkO8cos<cJznFU|g^uJ9;rAx(
z9)Frg76v?yvcj`gEz%&o^7lqEPfFl>`n|WTm#Vg5cj_$x`Gd2yAGZF%P_A+(<sXcS
z^RAHQADjve>*VO*57qXj-%n=#5Pj_*wkU%r&4!%4d+!h4qN>;_TGXcj<d25=pu42@
zN5`zvzD2S>rn2h))a#`fXYox}@NL)~`odWL$C&4X?4^r8ChdENZRg8BT6c$*<KyfF
zdID^9Yx*Z|OuZvprS_+w+AWUkPe#-H<E8m0qr%svv^xFCkTM^~rEoMF&w=(S1F15e
zQA~NmBZL*p|9FEMu)=JN6e~6hQT9KEk*I_v()^!pjkR&l6|(;w;(0^okKKz$gw<PE
z1XlHH5WbjFoPA(3{*R5Udh$6a+Lym}Yu~p>pUgZ{Q=pj1(-e+xcHNO2b*I*4W-rQ(
z+%B0z_(U6liQAH_ea#9J;f<WR=`N`4CA!7vAa|s4nb`*%LxZf0W2v;Q7#g0tL5lds
zT-8fs(;C{>N#!vAu<8*GT#V5!@#pMjXPura(d{zxRPgg29+Pk`D%ue<>zrQ8kf(Up
z)fKr_`kvxl_x-oaSnVk@ez6yi!N4l1rzEi?I?bPA4R95f3DkIs^=5?o`qF%gac4!p
zwEZib#2w_t2kphKv1nWEyh$l4N;=qfok*#AV(S=m*_C_kGWvTqN^swj<ECCJkDq0m
z%td)saT{n@4^CyTnZ04f;#I@ax0iJquH1$%skWC<-7T)qq+aFdHJB`=`j@iQHyptT
z*OZ!jdFR^F-pe_*$EnDA7}eg<)nNUxD4~8X-FxV*{$j5*EZhj_qyiR?rOIrbl&OIF
zdS(+DOELeY=j5(#aa&=|VRi^C0DD{ihB$BT+uH|~lQ1<CJVURO)+0lyz|u`AX6GZ)
z@M6QDYVK{_TgBMz?ee*d(aracd7;|{6p%h}GZrWayXfMdt=iSs0y-(rY{YW!nP-?J
znuFG^oy-P8)A49}y@B8Jrn%xZ7u7X#fGaZlSnrPEhJD{YNk|D&SEUCXwS7{L^<lGm
zlu#cvhi4WXmc=*hv2KT9(0BbB`&d8c9U{#q)!!)XD68V5=ZP8umZuUd4x41m$r&8w
zcxWFNA4Xd`0T)*eO1V#r@oG+@XFY=(hXc}c7^7sJ%<XB3bAB^xc&pU=qK@-P>3eES
zKkA}UYftsg`y4jALF(}hgvJoP_K{Uc9z*jrJf#oaARf)924!5yk#-cMuc+ZXQ0dzs
z+f27i$k<bjOE=yrh37u`LWFdpzAk9Qf~98hjPI1$z^0D?JRYa}z0@DI67G<>sGz%W
zR@;43*Eb=!efvgVHNM%*Wr7#I1V<4gSPu0lf$ZSSh-U)!joxV_bz@%>8_q(cxvw$O
z?%J~YwNtTW7+HC=tGfb(df%C+xzcn8Tj5BSpw&A28ZVBxfyZAB^DvIZ*`UHaRjh(~
zDXEuwf@5Zb%vE&qZVg^ReTq=_o2h9l?|AZB`hUXGw_i*<@Tyg^iB-0?pYaJBUvK5q
z_lq76aG^De;uapL?g^A*W53XF6mBMT`j#UjtPjkz9Uby#q2v2mABl|_i?a6a2Dy6V
zRU};a2|D{34>SOw3>1tb*7R)E2v3dZ1kw#al+bI}Xz0{~0ll*(PxEnCAwuMz7CqM6
ztxOZWXkB|+bYWmf>Rwu3q1Lmplo}Y(C8(;aEn954Gy}VyJ$9u1G$$HIZytqp-c3gT
z#-cD6M$p4WLdUZ57~E8sey1v-*VN0@yovLyBX8?%_9({98mzjxNa19J;j+K?S1q;&
zrEh<uXp|>TYx^7TId!h@pN5_L*sxF^+0OQlf$rDkXn*4f^0rQzfsbdP*s8ZZ#ctK!
zc#AysnQWD@s^l)+twl;*IQL7(>1e4p+QFg4gEH%7Fd*~={oejDHx><=CuuHWU@t@j
z>(diW6VmtekkJ(_BHU|F_wEpJGR@Z4p02BV>DEd%)mhY>!uV`>O-BA9wmi**hp_&Y
z(tf(}s>OB9JW#S*(TEFsq+=|qsyZP^|LMlfi!PInqsN2Gt@Eg-8t*aHx!5JMj`ldc
ze39Dg=MY-vqOjTK2!P!xdxrJWPC5IYVf5U1#KBT~h7T$SEhv%tGpy|!kK!ri)OFr2
zL?rd)u||}|*P(V|q&Q?_ZD~7t{7i#9y&QE-wY%aNtJ-;(qxK?7=<HmBYEiqh6D9Ow
z9un%5%s*LXt9JA<bI;q^J%-+&iyZZOqne(X6bdehzGpfKsv9>OyKn?P(>TC=nE|Ok
z(>gl;B;J5F=!kjJc&4+0Eh6Zeo?^eRaWHMK%R{^aycHd|+0F3c5vC8zP1NPJP);if
zFTHz;CAbqg<1E2E9Z}WgLMP_PIBvw*M0mg65=}&~IE>5OGedAWCbf%L2c*&3#MY!Q
zQfi72s(FBu6{qw@Rz1F{GE>E(II{z+!vrEU4lps$rM?*jb+9&?QrRYLN00ciA?2~_
zm>C+$D`U~LM%*B>2N-8tqnMwa17hjdXdV11soIK_oF2|rH0mxz!x|Y;r(zV<HWygY
zs0f5S%R8ipbnBERbWmS>v)h|=NMG=(Ju5L~Gb2%dmeZnsw(MEKz6ddEL~T75AgaSG
z?B=sfZ1fO&g|wfgZe4aB|Ns8SQ{lnHl}q()i7}^<kzoBQyn~}dxaB*kgL+R{9Rr?+
zZ)JtdI$QL$PPN|>#>MvZ(_HXOS3XW8pY5~ZMCp6BYbpZEYR``8z~RV4Qh#>Lu4T-C
z%uDG}K;zlQZem8+U_3iXpk5@?bDizd_M#*GLp)|464cU-{8*r8j;MR1LXVVm94C9v
z$9eTUCM{}oHc`aNhtXYCLBr2QmP!fH;C)tU@hGPleSO{y4a-Icq!dQ>R;l+{Z`GpA
z`hw$ai}q+lISoWuE6qOR&mGI8?cepKHfBv8r8J~;i;Q`zttBwOh#W#KbYJN8xdBl2
zC7F)fPxxPF6Bal@a9?VH+uK;0juTc8g7%ykf))#P-*bXqG%of;?KwVf%a=?2IYul|
z;MsG0Droph<2k8eUL%Xi)O=2wI2&L_+e;tGA6|v`(jEX|+e8`j<_8f;Tr^Qtt=Bcu
ziQ=Y+SZ_3PymX3Ip)6*jFm6p0V$a**y_!`rm%2oh8gI@!IcnIT+@5RQIHfMN=UO*V
z`5mc0*Lc9qWaKT23Or?w;ob0J{pvJmJ~#E&4~W$r*M4qrY|)_l@A4>u?`B%>6|1Q-
z76{d^z{W6A`ijPS>3E8fx0m0ewl841{T-KCr+7YbJx(RI3_-ke{+U-H<asejw@Tmh
zVlFJxSkv0`s_8yNEPVZWG0-d2SK%eq`oM5&JkL93`%c?k;5Br#H}leG`*|~dN0LZY
ze;_=sDrn@XZCYh4zS)DUEJ83`isu=-M;1#b8r0n;j?F}i+A*_H(O&+eqM9qc_@-Oh
zaa%dqIe4Ykt;%S5zV*6Wyg~Y&p9F3bMtRLMZsZnT-81wJBZMn^e)NBsC+-_5Yw&ix
z6ZL!-=@rbe_Va^lM8v?TJdktDz?2(%eqxkHlPhuCq=*uj)W~~hPYDj^Ss~SBlUiOz
zB+G17K%e~8YX$f|<C{Jvad$*Y%%yVEOccgM-@B9-L{F!;pg;N|#cWa6Mx-V&&(T3I
zFvcJK8h*$iRx6I#^pT#%3#_mDqt(*%Qm33EZHEqe4L^4fGK3qrEn}(0nS0ox7cWQ?
zfH!ODoc^Lsr~0Os2Gjs0t|DrwP5+q8MhU%@^dLRfUkK@WdzW7(bCGLrzQR&{(NA6&
zV}PTEsJy-xCa!DX)C5PFcx7Mctk9)qQtFY?v)TLj9R`z)k%2p;;Td|chCXY)(C7Rl
z5nSzrbecCF*h}imH+lM|o^~kJv*0nW(^|oLP`uDYsD;%uC>?L5TX5RqWpJV)%|VI=
zG2!QZ;mkq*7dzW-A#*RZ5j&pi>Mt_r4xDo@GJ5SUZS6(YDUIH#zbGVmrT*&6i<~*C
zGtynD#*1PMm+8SrGZjCFyW;IXEACGL<nc{SqWkE!UhW#J%Befv6p|E<RvDuvXhOS~
zw;f7P;LE~NpLJ68DdC=j>R6bI+N%9|IqN{{&H(FO-+|r(4F;TKc-2a)9q1!9!R)A4
z`TXPtr04_iS@q4m{pyx##5bLmNGApcQK<(erN_&yZEv$%oXYQK#CP3tj~(c;q|x%@
zQg|u#=yp;o%)UoW9+=F3_6?TN@@ybb+2;KuM=*$%noG@a1i+7T*tqh!@5SDrx~LK5
zNR4;bGI#1;>c(UE`3~bjx7dA;G+vzgfkE5!BHE+$`Fwq0FeR2^&y9IymUzxpqX9j}
zd009x4tDD^Uvx@;$D(wML~YJi#B5<+^&F)}t>80HS<}RV@XH}KfGvC@eJ`<&oy4O0
z5+AlPf<x*K<K1uZ+afWcS1;8U$Qmz6{eJ@W*mO)oL_)M*;<T{|H-_swV!30VH&4bA
z#<Q?jUt$d5!m^H&W4Q6oyZE&dM@<wE{hIX*4P);)*iZ2}$9UF4e$B+`RS#X?OQR;;
zBM{=HNho<47^7ZqGS#E>q=sU=GzMgp_X`_QSl6zHZT*~(=1Yw!i*c_7VonmUHF8HX
zxLhsp$WaFskv3jq5vt{-!7j8PmQLvzyg)D;4YA_tq3?`76VgkR=)IJ=ssJa*%oTgK
z{be>%-6E5|m(6&@A&4!;y!JBhrdmgn+D%rHdVI6BDYjEteDl)~zgE(CnX!QDBI-zu
z0?q=R2=`0d3vhRc0(lhha|>Oi9+9z^CH8Yo>}AG&Jq%(e-SLz$o_HMZ!4fVK&Zc5F
zT+bOr?`6j1yY7@ZZ;NAM_5Z&-34xxb^hHXoF`_<SZrm6oj-~$ciW>xoW8h^kH+1t)
zkw(HW{TTDR87W<z9+&pZLp1DA%r?RNI*qreJB5ynrP|oB_el{cPG`DnrITvwOQ=jg
z`yq%Zp+~~(+?|R=Y};S%9A2z9m*<@D5u$(P6-H-^srL%&jm0~^!i0updxd(Z>aQ@~
zvg~2{cov3`J?qhUg?GTO*Ye04OJDO9#tXelU~XY)zruTjHv!SlBefpi7G7b;&3=$Q
zN#y7Y)a#@ZDLvug=;jqMNqE^oW}{ULB*Rsw!DYSb8*ffy`sqRLTtdP2XB|J+<07Kr
z0DZ5FVMUMO{*IJhKH;3h%Uxry6TC8-J&P@B!;4vH4L87EX`EZKMA{A=e>}hY6SIxb
z5<;%V0xyp=qnwLK;eb=OU@EHjoVd*R%8&$BE(@(^<(iFe%G{(lda2qlo_-t>DT()j
zpc5|S>EY>u6Qr-^_|&L*9NExGRI4Rf-lRK~nvJ)f)n!%2PBI!%)gl{NMC~SDeDej7
zU>+94c0z#1DG+Ex$a)!j5;_d`Dz%o?wo?m+F2OHb*217%F=%Hvvo#-~aR$5>)l{L~
zFLPCD2h&Dg<(<<APmEj)E=<_?ruKz)SYO4jy~?`S;wkR>s|@uDz4Mg4YUa6ciHbz-
zA}R!ds9D17Y<Q{RL0+M1dgX3$th8U11Q>Smt4lr|^eKn2$k8wb?Dv$yLx}wztq<eJ
z50DM>Y@%H^-Y6yg0f%0E)8nUy;J7|lJeS(Evj#aR*mem<Ti-!xY<W0c=RKB9?VvQa
z8g@{R5_+b~WlvjjvV)8ni`8(CY~_{x5CmSXre}AH+ofG)^Yb&1qfa2^2Ze$gP`P7o
zI~Ilf=Bq}^;#PDV{Tv2&f*uXI;<@XTVj|urvxz#@Vo;}nj=d<sxg^mla|aoVTcc=u
zIXJq0wt5u~j*(!)z`Q-!rke&djBx5zk4%kHAZENO+QBE^IM{i{Z*ZBrSbwasd9b%K
zjQ6m8uyJPulOp%es5opN|GZKy_Q0S_!zFC5sdb_au0tZ*<-x%t{jE=ZDrvTAR5xhu
z&g)fZbPMATJ<fR-hfLebAxYSF;AZYS#Aw{Dm+EVWq;c0bb^Lk1y5|qGSV<N&)vnzk
z4bS1aJ*RY0Qzc$so|r_y+Flxe!3erHKg9a4&a<(q8Fe3PxHtwypw@9_qZTDRK*l0E
zq&y^sslhgDHWgP}KzgY-oYO-Wy<amI{TgR-%b{sdA64V&Q0p-lg7G^XTDh0)LB#r@
z(TGNQWKoR=DCkbZQ$+mPQ%a5mha{Hmp{5kixq_b=Gw8f4W$aLAr8>s$H<BKZBJfRd
zl!kszCD;c(6k}J$H>_XXpfWQ%GMmc5)Tx+^y+fUu7#VYamhCTR$YIf|0grm>J1lxN
zfB{%LEGT37-v~Z^lcs)H5{V5Q@?}*@BY_)-8M`<JAUdSV(V(gJVX+utv#vs{Q<B@G
zEEQu>7T**Wb`h<reqW1l>kr{{suHST%1Cj!5=A+iNM43M>K$g?U4<CURe1CjoCpy)
z+^N^wW~{s^g-ePEgTsS*UNEj7ZliPAS^Q+1^T6-vyN+xeZjB6~zM5X}`1#U~g8tf-
zT7txiMFToK*cVX(KPd#c7p?I6haM%Jsv&Lw;&IU56D&RVLN+jtiTJod4kLOM;rGzW
z+~F~1%TWz-gwyM8dp%5kNcxVjp7C=SgtI!_EiPIt^&^t7Z$kxTM_8{0N0&(Bh`?x(
zeMcBC&bmU{UU#uRO_ir2E7<Ff1tNkA(VvB9wd__=C$d;u*t3=1MvI8c*(zfb8}pu{
zrgsdS*XOpE8ol$9403#dqS5y>`sNv>E%Z!no(a~_8{E9B&kMDuq4)Z1Y34;&u&Zo%
zQD)^zefzl?h{LgGHxt_PVm2U3>bcG5C1KYqVq;Ye9xbuyfPFtNIHh3%9d8Wu{M??e
z>wfRbb|4{wvsH&UCIpK<$JTQ!t1M`0F43tsP1s)_nFNrx!k7;F5N+R)P8ShgOuO2V
zF}X%Yb%g6j8h1|B=WCC&fgf#gKe=&aum=C1+QZTe#P(h^+Fy^0+toK*yxK)JPWh2G
z{v$*SRL$BIoItKByW?eAZ{SA*om^j<)E;J$t+4VuFIY83vlrh~Z~MKLhh;8(!(*dY
zO!R$1d3DTK?|7tCt_c}<bqrVALp|%MEH>ddKT+tE?Ag9PA&n@Zr+aK>qeQYB^cl4F
zt9`Dl;Djr$@;S8)7tNSs81?|SBH?2b_PBIjy~mi_sb-wHIMv-<MmdcT#B~^Wy@Xj>
z(u;5Y^q@cMHW#&YgGLPgQK6rXJ6ZaUvN1X6RH+@6=7$BJX;8A%y(IIUP0~>z94nS-
ztUx0nX}6^5NRB^I+6f7R!7^(Pdgn(4<FQhnkg-VdxGrw4B2sK05Ri`J<o*VGqsU?3
zM7dc{4fyY9^rDr-7_}qjQhC$^k+DZxx9(UjeMft@^aqX7ExTIvjaQzSO>C^Besr1%
z8(6)wqm4y~*kjT-+GggE2Sj->!fEspe>)J_lkiBs{%B?HRq^zp&W?_g0WNX+S!a=I
zz~6N?FYb{}G&ZGoCxQ%}o||O$=+IC7(u)i(Ng}Xwi41&MTueR2dV2@~Z^szj!+Q8w
zJ0=DcXNX2T>CcJR<C}V(`ArKBA&6%`CN$551Q0d?)lt%PRNZ2JUE2N?HJBfhX3;||
zD`Wos@MTg2Vtc9$#QG#e$1&VCM6_$^@b34p3z|J9B#A{EuR<>+^{|?gn^a3rF^;v?
zSgbU7sqa`LRA1p+B(-Bx@3<mJ@AP*<>c<+pMsSd1#~MqHx<ML_Lk~_Eou+?pt(JDe
zO|Uinic)GBJ}hI7==J=}cTr{RLV7!1eSjnOs=8XKvq6pBvuDEUdsw~rX73p>)V2rk
zF0k8Xv6Fi6fv|Ztj(sV)gc}^#O!W{*j>@o>>Z!Ykr^*dRWmdO75;D_>vg~W&doxmc
zjJieIks_v6b#UOZUtR={T6<&h&4bc(W6r+5?i$Ff9kb{+;w#nP2<g;)YG?p@Sz*%)
zr>b2Zlv$@rFP~RE&|ejutEkc?P>%E7sqxN+AW{=JQ#hhJE(EfljZ6JFZ*DDulO30s
z!0!bjWSk;OBYiV(lN)}`3i(b}wf1qL+=rIgN}fjzl^L$sTQtL1e6t8)H5JwR2rO7&
zI`XL<n@w0JsNeE9lebaEayAO{61on3s*1;k$y$-K3xZ|7cS5%e1o_%=PhWhqsBILI
z{)kX5zS+xpkf~-|eSXXb9%kT5c7=&`#}@P{Y0P($FyNY1ww~6w+1hpl4CtgjnUK)&
zz+k_O&5w1xrMF3#6Gc^cEi<Q+=)oPLZ!Kv|Z&{UijNSI^d>^VYQ1=pQI~ksfavITu
zr{j1l$59v&^v4^$L!<pt^Ut8(B=zH?qsuiSEIU3}riYAlRDZOv5#O+gkOej8wo3DO
zn-HyAr0ux61q)a1UoXEx#u6Sozw-CJ3U5iL6cJ>dpkWH{Yw7h4KT2ku$kwt&(mOs`
z3Zo;75YIU;#7kPTAm%Hq<sf}Q8fBga4)s)v5^6`##RbL%UIQRxW<jhI`Z9q%*xK-1
zmJOV8UQd5h70uQTEIc!eQz2hqI)q1;Eaq-(fsuDk@*WfwQJb`PURY1a^>wQ<Xw-vu
zJB();on7E6l;7L#MO{^ikII~93@t`)o#1`57*K|T?YH!uV5~ps1gV|ie9(icT~a^6
z7ta;?iecsneAqnc)Tqj7q+~N?^>cLCKEa6SUrM;`3D&>XB7a(7EEOBjm>e&r0Z$z-
zrava5N9JY1jg%7V;j|6jDm{meUm$Y{k?MFkF?EP1<aD7xQTCmfy3s)w)J_bQw3y@Y
zdKA~W%Wf9+z>iXSaRlymeon9vmGvoVF4snR`)wDdo`$`)<C{IJWwZN1e#hAQdQ3bx
zG8RZwtrEOf(Ww3_H4pHWJ5?+nSnH)1$5yj5#)Mxzjkfr6j2_>xl)ICv`_VFUCmO5m
z_tj5wF4)6`khC3hvoF43X+<;`9@mphaG97K5kx*-Zw_4i4TJ2YG)35hp@$k#Mh`Z(
zS>Jf_>m*;w2lb)ewkIs;j~`}^X~Z=Fj#*QdI)aOnOva8sSvpPvZp>r(`Q|k5I%->Z
zN+YA-Nw3PgACqz}rRW;rN+)}F?bj8|#(rG-PENS=LET!U^!ENeQco!?Zw^>fZRSpg
zj%%29<7A`n%7xN&h|BKnlZ{#3LgRZ9%$w_DUSjB0DI7ZDXz8RDh|uH5%ucqEQQzq7
z$)TBXD6iCzy`E!2sjrv0sHK;qxcqR6_dt)fINCWS`e-ND{<dc=z9Ci?)2dIW*P|A%
znV|Txs)V~r)wF7y5)+_JxMVYm<5XgLwIjvl2^f<{Wm~Y4f#D^(1=nDQq=*t~k#=HT
z>YQS%J8=QOm2J?er^;+13J(ar)KiQdssZP`$g7API5kaNKit<xEBa1#4rsu{gw#&8
zYUgwDyB-<3Qy!P>)acM<dKb$}wYcuuJT;86C4{7L3{qixfJ(WS)1Pr1a}a$ag@fJV
zb<#Q2s6X#Ae#ASGmLha8jpCa*e`4wHB{~KOewj;+V+zxToMys<;YAGQX=x~l(?VSJ
z$F^&ySywf>9k$d@vmUX}AjTj&Elt8vn+c7mZVxBWwEbp)S6A7`E@?*zJt5Kfb$xj#
zcLJEC_|~GdMnT}NIL$h`dbM<_db;IvS9h7dQP%E;%-~&fTpZUEjTX2jI4)ye2I2m)
z-s#>?^J<yV&XArSV@24V)K2&D)tx)5sAuUbKD-qiSu8y)u#MA$1ug7pxY<lt@b*jl
zbff>Gh5QbBFp<f%SjJK%4Tz!Fg;yF{xst`c6ZqGg=~4!zX0D-;)gqjYZ}ui0vaR>X
zdx>Bc#0i=66y8ev1=ce{JuD^8;|vqk!!@uleO`?h<_zzzKFY(@)iaFlQ%;aZ^&O*E
znrArMxH4g(*ZLC{?K6UHS1ujo_uIXInfJs<>w_I*slW(Pt%bMQTB+g0-b(!he&{__
z<1B~`y*}JM8!hV!KQ28dd+AVH=Ay9f%O7S1(Z?O-HI6`coDb@F(r({ttfwt}N43`&
z16Z|KOkNWsF|3yfvXsJ=Pb8(MkS~@IX$DdcBlO*7+cBA^Jde_PNJM~PDs6jiC9Wey
zRYIR9?o@5r#aqXOe{Wc3qpnJuhK}pgj=c(>O1v)59hrMg@OtIQh{&0)tML5vojG%W
z#WoKu-*a|cJ97_D^jfvvIjo=Q!>n)U6Fw0olwX^f|N6T#jWdl^cdXE7)*@dwe8dDX
zfgh3fnWnk8uxqAZUKLqmp$po`n4=&(i(t|-O_Gj3Njm<urE$)slRm_o6SH1m5zCDi
zVCueG=FYT^;%-OF@Ib2GhQ%8)$S+OyW!CK>Vr^<!uwxO@T=!DSDR`Y_)(v|tu91bZ
z9ke&Ro0*%z{35D{f=fFUSZ2?e<c=2WmNT@m%(-ZChG3}krEn6s_bZ*0t;fWsY>BJ%
z>=8Dnvytrz(T)hD=cHjD6Dblzt5+9)(pf$_`ZUK5={w6?b{(a`CPZY#H`B}rif)hU
z@y+iU`<vz2S;nLhWTWBWAP-eE&kBC&YvJu^Om8FN`i~Ud#3~r`l-W1SrKn1-WnOkt
zNzSvdAEI%4T#f!~u=K3A*y8!KUPZ4zG&ji+*xXqr5(7j9pB)2BKm>7iXQ!PZcA~1S
z+Sy5f)k5&=2=%k03rqFAk|==us_@y{?TxczDx^)0oe6kDC=2wPw4<6*t5&Q12wRlU
zZQn*2ixL|A%P}j1!$$pVV`+P7OFEIQ+IXE=$=v=pJA{nqZ`t+Crmop5SUpeSrPtMd
zF51F&P(SH8X-e*71JoBO;vU#a)y}atnbT5tP#?_Csxmv#6AjPM$2ywl1PA?1BF6X}
zn+4tC8ZLP_ylNCyk5w2n^sR`L-Y~-}d`=QOJMQgTWZF5wb^Y~&*(i?mv3U2QxL&E;
z7*{IiyaZCFF)(tjjrTG}_+0O*ZVLDnhI6fR-Qqf_dj?a2TR4(MF=ArSc?~b7WH--s
zcahErRf!mtq1W2y#w^tneZ)|Iza~#em+AXVV}Vo^^SBh}+Gx#-2LL;%Is>XVc-f#y
zd2aA-;i$}daUKz{C&+r9HMA0%<^r$A3vjRHJnx|fR<J#f6gT2nk80;7appIo(WV-Q
zR6j4Q2er8g&&*O<&%sGI0)yW3+9J(_kKkK8PJf0^mn?p8?K~rz$K~&pv4E$r;fnKO
z0QG`!Cn~03-$!tLm*<7}uC(B!S+8sN8G9*<H?uWv<~$qDQ%)8+-x^alj`W?MI>-J1
z4(k1o+WFP=W^JuU30*Dq!kzwZRTkgki3EoAA!(c+V!;o#Is(RX6Oz{U=7@B|+Y&ka
zmKVg21sc1z&wMn$+WFRF0?JV0ovQd+1h4!vTGb`Pna}yws}oO<UX<179lUDi#{dls
z+EJ!l5Ddna?YqGG$1#@9>e>a)0rzFvO)aTkV0{_kNv7<AxNhz*$7qd4)JSe}{o%^y
z1xCk$lcnt~4d^2+x#MYpXUrQl=_eLEP^j6)5xT{Fcf7E@Kpfc`@Su0ZxS1uh7sN>O
z)Cr4gFHo~s&0P>eGK}L|E=&We$3i-5xEFe1ObN`W+J!FZ8iPV~u;&cwEX^W^z%7h{
zMx^u@jrTz=v?drs+$n9(>J|(2w^j30%{(1rUQmBC0Jpzm=UG)Y<rfC|9F%a9P<j^k
z+&H4W@X6j}-C=NGY0n$<qBC|p!PL{AmYciKIDOV7A{QA-*85S8U!?CM8?<5GtE*k)
z7}fi<!-AQ&CqwEYm;GB|MRt+1ifbGeQw2+;agnn}pKjqt#v-Ls!EKc&zvqXh@UUGJ
zyw^8Q#~e8uejc=q!i<1TEdAAsPD0SNPys@@P*)dOcNXC!cIb$Mq*oEfPx>+CbUQW|
zq^QvG+Qw_GSznj>JfT<7mP+llA^4m{s|eO#8y#Z2i4c0NciHZrs$ije8eSchud4dB
z#vgO6wxh5;7G`8@lHBuKi`2w&)SSQQJ;)-V#*54E0-)-Mr%05w1_*si2|VpG8>MX(
zH!)=^6r^HBb51cADLuI*VESU~#;9AbeHVu?we5+J+QmLyIUlNDZ1me(_L;+6QsN-s
zVq^31$4T?zG~e-~qjdD7qJ43U;s`&6k$Y2SwlZ%$MfZ2H87(5E>%>FSaikb)Wi+Li
zD`qbacdh-tiaH#7(WI_#PZ%bmDJo3cIpUB@k|^?f@aRpAR_ePXM)E8k=E4zHyTnHl
zOOc&Zy-IP~Az9Q{rKqR<8(@CJOX%Hx?$tLh2_0}Q9@I;MearRj1j#Q6&3&zmUE-7L
zGASJ5MdXg-K_lxdD-)g-x5{kLpzb813lc?4I<`C&MLL~5U`Z~u9;kAFD}9$n{|MmF
z{d4V7qmZ&3XkF_1#)4$COO0GT+0$DQ4aY*cow~M0O2f){6vuINiz_jl90%e_pxoG{
zp~f}p($9&Ds^%I!j*OI!EYW+Z?rM#_z0?QCOmvuQ&z*X43^omYo{JLhHM7HHk;{Cz
zbdnLPd70x-g{*He>v>J>GVi1r&GpM{5LW5$%{q7+Dyk7k^fZD%f=DTnUDB=^P-k9#
zS&Z*xBQh3^*)Og!6?K6sqO|&kx!CO~HnF;(l~La9Vh;asZ_Zv8Qp_9gajqVvv_l%E
zrO(~YRW&<Ajp*wcxK1zkj_VIz5<_yi(Zvrpalzs8B<1{SEK^U92I`l`)N3J7Sygp2
zLeRL}STeMDu{19?xtzy!x+tO-)Ol^wQyK!N`q$gyapn0^ILuA9W26)9T7)>tMjux!
zJ)aeg1t`(57;j%M8sNulnM8Au(&I$j6^%5CW(r#5NMFOnb`5J&tr1=4$~o~7jp*8K
z?8&o~MFaw?Yap_{IegRD`eq{}U_{!HZ9AAy(|IE#Sbylx75`YnxN^=yDH^eZ&>MOB
z3UJ4p)0@WJ2riw*ZgCDiiy%aw&mca=b9g|H)2dYJa*Y4ic~4mriIMRr_rCaMKLKIG
zw_a+mvmWck12EKI=UVm@$^1LCKZ^Elyv}(v$r99z9KBC7CGC_lZ&4mCaHwSO%<3;-
zjlJ&AtOGrGU<maZQxT<gi5*AG`ME3Uyv}&uzjOpgcp1nw?w+R^Qi_cuPu@jJ?SB@V
zL<X~eyT}#Z`FMxLt|9haVO`)7AU0Iw+kquBL2rT9uSim_@xGatwx6nQTwzQ(?{;ZA
zG{7;PLkkv2?j>*LkBnWBmbGCWkHS&v$sWPkohy=?f7PhHdQ_%t{57j&Hrk(@xd(OA
zbNnZtEOY5Q{&X&mU6}+_13vq%OoPmf$4t_@&b9Q7S*{1r8V6R7Z=~=pvk^c)WPwR_
z26td9yxYOij1-<JrcnFJm<EgR0_9hlkn8X5j0ILV(scOPtXJRkO0?Z}kxqQGD30(J
zlvlb?JjqxL18=q}t-nj7hj3e^7vI!&^5d9|G8f;p#QoIOLv>l`+))kpg!Dy<*C7$%
z+Ct}*hKp&ujQ%E7eWA%jdzh!7vxR&1@b)7rji|2S9vnU|jKxk}CfH7G&(*QTr;{&?
zf#<G@KH!#;JQs#~E*83C&}kh=_1Xq!&_R<<0DX!|A5@)PXx#QAf}YpJ$1iiKCZ;Th
z>{oej)Y8yjY|;0~`>u-KV=1na+Evyl*A%3Fl?m_({1(7fKD+FpgGS&`3E=Ht_Atbk
z$2Z+9Zbe1t55%?O8&(46r+NVuGtv>YhSe=I=1nbSD^nyI@DXg2PE}LiMfYR0(rfCq
zWZZOD*#Hn_%A<=FIqHiYXBu#J5a{%J4Htl~_AcSu=({?GLx1Lq<L27cF}m;&epuvF
z{c7v>T}McEbx5%O+Favm*JbPR^E9KHilMSVSM|2%bDT9O`PIf!wahVi#ymwi`$go0
zbP>M;)k*D}bh+cIbaGn+%Faej-69h|l4`cOlrUG-)Hi3<w}L<Nnl!rlJgi60^<5LB
zF!ivtJOwF_s5c{0`p!QGeAk39a$g0nRpXlIyuL-$yvBHP4z~xY&x}KUO^m~x2g_J$
z2-Xft;VFKyN#7vtRD|&wI*seIt7U3#*qpet39s%FdqEg?^$>@-Yhndt1I(}cspTTq
zMo%ri$35YF*Us<)EpF%kZ~sZ{+K`Djg|1)g^MN&*P?RiMQd`&V&UDvEBfe>9<;KuI
z&8pTzaJC&OgoRm`M@rwO<n;Pl6URaQO@wQWe@jM*ALs;9yA#9I+{}oVsUdUttHR`I
ztoVkVqAmnTU<VZq=#<0<a;=TrkiA?kuXj$xD=jnnz;fT~U6kJgrM1_`F!#5(I^-Ec
z`gnL275&l>va9AxeA7Uu^e$dAzNvA#LEU6J^<QsYwA&W15ADL-!ciM7uT`I3<Y?$N
zTSl+Kn~+kbDkv##9i%p~C{)n(Ct!Q&n{o*rXuR57d~+W#0<OEq>t^~83I8diJ%3Bz
zbw=M(J(IZ3`J<XuFRtiin|gf1rKFk*zbUS->ud(hJ5?HyYsLtTt6#8rUC0b(6AtT0
z>FY)4ysErDVy8m{AFi_zP#={P4if~)>mAqG9Gth?-}xxL(GyRU*+9Z2lI}xh2l0^n
zbs@O?ev;~$IY-1RH5?LO@7+@6d$)0o1Geic!G^5b^)cP?=IC)uRN#6SXGf~RcSx2h
zbax>&jq8nRoFqvzWpJ84Dedb+gm_+3^2pLyed2nrcS*Bfyp9%hjd)mFU@z10e0HOR
z=Oltw4r+|g?Df$*4&U_2nqF0%SJ7D%^y1`P^-WO{bHjv{8$vaNKS|xkeK$C}mlL8{
zyCJAuh61angl;gi8)7LZ<kdFFjT@}T_PZE?mD!*hZ43oYQg3*&g=%&v-I_og@*9%S
zcc5}CQtGc~LsrqJKUomZL3j!MWf2~jC{v32#*1XuiD+>c+e;ZsRmIL#Xm{d$kQ=Rw
z-Qo&DVsA8AT}C|YjcLf&Fn#McI&1tku{~}nnU~;qDOyVi8I2OeaOl3y-p^>pHx2C4
zV;u`r@>&Gv=npXFHwH`3VTm1!;^SSvs;fV&UZjQ@bUpdk^Zicxrsp|KBaQbegIw<4
zBRsP=8vFJ4;CfMbk_`d}&iKxG5ih=}m2ba6fMqK;c@J%miOr}7vwiW6Qg#PYACRua
zH(iSO&33!^S-&X_+BzD~5+?fr%_-C$PQshg3Tcl^G`)ge5N6G4-(*Z*gjE++?e;6Z
z<rk!6%yUmUNs60%O5DqKp%YE2+PP24gwO8d&fa9bzlo=0dR}GdHf}M`B^r1R8ds~_
zoI1?US(_Qa*8Ap|sQ8X<4yjl)BK4b%z@_|dT~tatdiW%b_@=ImT?~hw{57j`>{M;o
zxSoEqb7>N#ODKrr9J)ou;=8VsoVZ469Pil{=|oBo$*^8;4ry7588aJAXxys?u;>r=
z^c?Ax=gFM+MBn-37dHJ(&5=cRnBUh7Iymg(oV6J^)oQ9omfmvUCx@HXZKg#dzO%M6
zn9Zhh&i((cj@t1}ZFh}!(ch-bn=$BYV^FMPe^$-pM~e97w{diapvjxT`6CHNtI+am
zAgBledSx~d1Cbk~ml|2dxe`y;rbaYc2|Dyjv$VX!>YvyqeJvZ~Rk&1Zk--$ASiNPO
z;2jO{WUZM4S^p4l<*quD^wtgMNc_E-p=dZ3yt>k?uxti>J5v0F;tWsMwv2^$V9Hjd
zwnKH%N`{%mQ0vKUr)Aw9>gQ@?=`d~?)-P{K6+|z6K8y2o$Gm{w1zePL6Ook=th?#H
z>e~}`@T)K9)Z&}Pm9bu{53Ih(S)+P^U>HwfciN%NB4ZSHl%!dqV-tsRK-#3eC<KQ`
zCM91KGDhT--jp44OvD`!c_76iSNBhHUF*V%H>y*W!}gRisL@bd%$G$WK0IBBGq>mD
z=*`i&iX43bPn|2ANA@Q!=u;IchJ+vX2Z?sSuHU<=Iuv(l6xUc6`=KLjuPnZK2!sFR
zd!wq=(al=Fk)**DdfwZ1{QNFEhJNmO>`$3VT<h=mJ{xt6${tFKKsnjf9|!J4Ipv4a
zy~EC+^c{_|KSghZT`sdhpqAA&Mqixi{aL@&pu_BsnuWemXYnmE5S-wUUit={6KYUl
z%?vo>I9M|R^i9tCKny3p6sCq&Heg~r?+9rOm{91?dpDz^>N5>x(hD~2_@?T}ZGLR{
z)b<K~c4*A&xA+m}B7I-PkaoO<lZlW@pNCn!W`k5NU*Wm#Rix?*vz!C79?cCTB{MiY
zL<G&C^@a<}_LiXicy8ZdvWMDj{o#Syp!Gq2KBk_sMi$*-Kj<}Rtie0FiZ>e?o}zEt
zP=`%N%~)<TKM7%d(jE-MN6V`vnWsXwUaLEdSG*EZtdhl2L<tKB)@guTby6*LB-1f<
zKbNUo*Dhz6r9K-3>Ef{Y{~_<qg5$i7e9xSh2|Uf<VPXO?F)@ML*Byp?QI}=ehNo|L
zK$c}gH*BlMvb!<S?rwl2z=A~*C4vx3w&gBMUSQdl;T_&fT#JjyB1MWtQY2vswBa4z
z;WhAxM?B(je*eroAL)B<--wBM94dsU@B7Y@=bSuw^5klpr35^c@(e14{C3rnd>C}3
zAi|pQ@#3sugELNKuzB5Q!!rrHycNJ<tIhZ-zv&G&Hsk0bM8;HXCJ`h8_;?iE$!Ky4
zMNTtnz5PLJ&lJ4s!DKAOdy>6&qEko?@eUFIr7K*i%98`q*7Hdha^fP+__7j+z|AkW
z^COf2s%m@vWAOTD3gYpv($0H$Z!WjXg5b~h23T_O!!bIaBvRn)-s@8{OF;L@`~Y(Y
z-__h(b_L0yp!xC-YUjJS&_)pdrF<8LWpQ`v?9BtLj^*E8pHt#|tW;njK!JFyXH9Xx
zCByh;zH>`JR+O{Hw)gs4?DqIM*2sPQDx&iYmIq8VY`Bq=PZ*Jxtk!T4&?6HYwh=?o
z>8gO1KJJE(dx4y4lcg1WT&>y2x!Wr~4R?X88+y1v)}`v*MdW&6ZP$rL$a}#--)p3(
z@s8v<i>)M(O7wHd*2@!uFtE8Az<Dl`+Xbq*T|_LI%?W+&eKt57qY0igB0LYz=B!|}
zVznsDv4CJ^)`$6O`B#=U_F!+FZRl7J_fcX~=R1v2Of~7%tvwraiegp>J)4wfY8R7Z
zovh#<t6l-@E=3MciEOT9eNamJp{x)xa@K(h;AVNpgm97p4|>i3TIA2(mji!fw++58
z2PxCQ;r9h(ng!gL@=58SPuf^1c)uc>mWTc1oSZDp?V7b_{Ek!@?{g%5*MqhcZ=Fnb
zvQ_w5pzpmeF~b!p=x^UwHb{i4*R^gvYjCI{>f6i`+~tu4NK1%vs6bHPXZENsj#4=d
zK3WzPbr>k>#@P9;q!QC!sO399-f2=f1jiorIhs{Cs`{fz(iB=lF~lb0W3vfbHme4)
zY}DprRP9IOVEY?edbG@_;?Q>jwpX*xy(sd~TWsagL;(DTAoc56Lzh!@xlyA59o^0k
zM1GZ@^uB1ASkJF4*<e2DI>y<%&tYfbQBTJg^iNXtj_vcwIV5pn{G5Y)PiAf59BP87
z&0%%e+WUOu@^wnjrKp9qx^?y?+-PzU67{kX1suXJxHR&*nyp8jk=OjKZB~VjA?V&N
zu;F*xFxy|Y#W~pq?e{5sH4d`-9p+iolC}N05F}A{W$jpb!5Qd=KK1>v<RNAWQa2e5
zBM6p@YjeL(PE+|^+u0?i+3b*|=<<zR3G+QT3dhuo_YXZ_D^Yj1;+wA*aB)7rW}EpW
zFDz_C@`;S4sP^nY!D*#q7Jt>Z!2>?L``L5VS1i*JnMfX3I}lTWD+m}rb|5vtm{^Af
zxSrq9y<(6wd7!A&Noo>aM$IgsQ{b$%vkO-kK~t1!Z>j3`Eb;R~q}DlbxBLVJ0pQ9&
z1)J3hpc|Z`D+iKhJce7vdVG?CljCmlK$4cXa+iqTC|bA{&y{^}MDYybNuqga@#f0>
z$ec6;HnlmQA4FSYbH4l@1Zyt)y<hPhvR0VBxkIxxqwH$uKUzCKOL{A*tPC-h<{Y@`
zeVFog3J5b(gm36&X@W@QW&6CX<hz))5OJ)RVIWn7o5&|6IVIwH=@CxjIfjAa#jKg+
zO<`Cu0vc@Q!ouwJY`9sF70Toiw2M}2-dXJVNCDY&s7V&WjX8e*NB-Gl)|Pr-sv{4q
z)66}I1QNY#JGxsXZs;BaL0Q4Ao~fWriJbyqW9o6;&Uy(T!XhZ|O8JUX87n<okM5B?
z5EC@In%kXGEUK<#NoJ4v^1bR+Hu#uBKed1Vlnuvkj6x88JeK2xddzdcavON8pn;Sn
zx^f`kc`WB|YM)!EO_ml=;x@B<XvSq5b<x_7#r$IU#o50UPx4qp)zT^7;GyA-OK*Nt
zlT3)=K9=M~ay}wH*m_nW@&LPXGpop5P%-Z$XWlOFxQTfVFUYSQEXyjPg~GudvB7+&
z`;5Y0h&>-Zn2UvbJvdk^kMl0;96MNcOpha~9#3*Xhsm4F*2EYpgRt)5%`6Yo3ta7k
zK{O%sOYuP_Ch#$xswy7PekO8Rtg60*uzV#;#0eQ!-t~jI^X54GHxK5bu(RbkxLrDs
zp2}?A7aSqOYi)4e#~`baeKR&ZU*;uDZ=`gpMF|Ly^6-t7@-pYK`h50-EZCD6m9I0?
zXsw&`3DRLeXlK)*!WJiu4SoUS-P{$Tu)+ee*-M|y8zRnplKt<Cwq9B#aD;8n=K?7~
zH{&e*>qcO9C`Ka^s}%4k9&PYY!gL5x_HY)E=w6Z4<bP5t1++WI4*41>vqAlkPueR_
z-+tOA55>{&h&2x-wIHA(OhrCPGKTyzpTrcgjiX&$pZ8#0AB`Q9qHiUB!?C}~)}spC
zw08Pt)!cKiJKF(^qNm$JwM^)gf^rTnIDE5O?%{>pxzmV0{Tfxv8lpj6c=C&gD#!9&
zSkAbp+Ug65(lBipjAaQK-6&PC99f&u&nq3I^zHJU@#@Yg!Ol{ACsh`hyx{XR`UdM|
zyHcByIW6Abhm0%VA1CbyiQ5-)adx2<*)D_N2qQia!?MQ<>D<rux53AKapsy&*zn^q
zA(+B)RO+U~LRRk-<xYCU%QhBO7MR?6`Hc0yqSGSbv&qr{o)<*tYCfKOgWiyIWd$**
z=b;(yL=h!uU0g)x@jT@T)tc6By;6V^9qJX`E3Eo@w&r$VIdz+*`&|q2D&MX;jEjjh
zc6K<{=Qy%~3pQ9i$v$ZKaM{*xwA$gEs;S3yc#a+Rg=o%NJ*&vogaxjiE^|CK59cgu
zF*qI-(5HRamt+b%YT_9l4(>z5QEAm+{gWB!9QL)4eLydM%0_J^n@6GseCydf4c~s6
zdKf9#X23kVm$PR(eox7j!CGaUm!T7bEpHw3IvZ{k7_lT`7354>s|xp!Y>u@OiT1$y
z)LXf0Iq4^}6~f6hvlZP<WU?h~YCGR~o*ckKN@q*0oW{9>iV53^)-X823?Vy(lF{}`
z1p=0jw%)q^zy7;wb=%BV_CY|mqXVfah+b><L}Hwxu-V`f2|UDP_z9naSwdjdq7XPM
z)5fuUl94)L{YnE<o))DlDfh`{KFONw1#3SMH0%Ledg5+sA}I{lpy-;wfiN6Va%#a=
z;ukKK>j92}E^a;%NFiBS#y5vik{kDS)v#>1)iBc9_xp(<H&bOsv%&Y5oht?_g1HU9
zKNkNKI}LU#pOh+|9sK^>WA4P^(Yfl`g!I49r?<;wzEh_%w)EpS8<MXhaXPbvT&hCb
zOE|G8vD$fbnpk_CKWRf6g`V}YjT0aaAsw>>)I(uZa0b9}wW_5(64rp5@ArlDq#N4>
z3JxWVnLiN2*Qt0h5pnPX1@~nVG@K<ubslU<?E}7c`-yUxUnQAW%t5)%*5fN@>STVE
zz!Y`}+Izy9A8=SWZr)(+=n3xYpyJVfAhx>aTJ01d(Do4@Au7!tgbJ^eE>kD16qD<v
zp?Oc7o^`vj;kNB8FZ;NofT&;z_Q4po@<5z)dX3JpdZ$2V;e|My??j$2u_qOgs`f!2
z{6mk}*avfW60-yDv|cqJMS!ZX=RfG9UG3n<*o@x<%Op7V2a`w~#NT_VJj~;8*U45w
z<t|*o;JLJdVsRyYPrXkU%nv3vI+I?=n|DCa;v>_?3FCvYKg5M(+<X?=vtvG4W`Q7@
z>=`<<!6zNCeN*xw9ey$gsNC<!tmB*R3dF2=(nkhwj-d8QCk_YkX_<_lQ%KyKPv$ma
z87^5no<criOHu5}e<8tOR)(4GMH!d8BIt=HEV7h3c0Kx=RveYh1iU2D5arp9@3RUu
zWFIOEGv|B!L%9fV!*=9DiQI6Lh;=?x%>dg}TkO1z<-1Tao_bln81-oI&(p?ZG8%MG
z!{&#4!ls)_z1z;ZWv+#rOCL%em53us)Y(Z0<Vl6y+=qPXb|FJr$qzF7L6aVR%!(b|
z%w`qgLJ@l~d$H{@5N_X`%l{U`(t$!6wttHy*NsOO>?~|9`S8ERfIWvhh<{ZCF6t_M
zFRM28Z;sv%zQyYQmVnFfoLQZWFK@cfn(^zQ8Ecm^@1d?#t^ejzjiWOhXZ)VtD{Hds
zSCTWiRVYruN`V>1m@_M%IMsD^EKeSlbGs_(@s-;#29Ioz;~FUMFaSN~dk=kUJD2_b
zJvP)XH0?ALYPjuVtleuPS<emWv#naxBSdb)M%!`haq^F4QJu_m0Tc0I8!zu%it1tj
ztH(PJAOjcX{+SceJ^l?!nLSya_KW{+BxonttuI)!d?RlAw6)3*<h&K2a8I(u>^3xK
zv6+{m;v9x)gmzqR7skOub9UN}q};~#JYn6^5N?Xuw~pkK)^8WV0zs)}32|#H*%6`*
z8IJ+sWm~K2TtZ5<o=-C5xndhx-giw-=VmEE*s`Z=E1ytxO2tsxmpwDfCWDfle5XtM
zDQK=3bNg`FwIH(%d^on;ex-RC{IIWba}ELHP!=Qb6q?rI4+oyOR5xAkN2+&n^@=!X
zwP;SK(5#I{bE}<qAa2OQahrE|Y&^b`llC5~XLs60*KOj%zFm|6O8;y!zIu?mUDo)p
zBhqdzb<Gbu_F>=OieIB|wDyN%#|F`H<CfIMEe3RCP3P=eidJT^F<-8xq=#ujr;OnM
zdb4hPlGC5kfn`B*U~!A<M+*xxGqw_SP;z4mvsUWhipK5W!-+nJW)&Q0qkNnrHMnh-
zC#*(y*;Y<cszv+`yIr7}fvoLh|2SQ64A@h#k_5S7(+@nA<S1k#eku)S35gZY+R#%;
zgr@P<Wx0n-eVB>!Hu6*+fG!Y$Qr7ab<ZUn7=u?FPK-Zpfq;VH38_!}0Fu}=sLJ>?*
z__j_w6(r(5n|vzg0;4;&WFx9-XG^pn#h(0EYh{x%qC0NwY*KCqy4j2WTa0(2vmBto
za4$U->n^@Z${)p{iv7G?0VRqV*S${kfsR>mAG+Bif+sMbSHA652)6!Hj&5$6mc@jn
z7i_Ie!#*Mwtw$+P)HbS_oaK~a**44Ey?LK)-8Gu9=OlYUd}kh-wiB~3{-D{@v7kPQ
zj7Kunfv4}t%-n2t@aez-oq(+H(>~L0c%u#Hdmx4B8Hx9M5d=P+%lMv0Y&1Ivy9N(_
zXk%GVW;Um7ymaTu%E5_xR-swvv}K?%5sfRD%+o$oN*UC6I#Hxa9KdEgfd4Q!t*VDB
z9L?>h{CDrQ#cW!v_gQF_drB8*|I^>tax^}NZDgkm+QX%GqlqU8@?5~>k}qkA^`p)d
z;;@wh&_e*aRyA-{wn)(c3Yd-3-{TluR<m;*gUj$&-^zCi=X*rw7Dw}T)Ikui$9A$^
zB!!5U>?1zcv(jG-e8g8~YSsonl6!#gAb6q=eWYw#*)xN?R2=?DA<;eb2_yN0i3TiC
zEuM%ypa?Ld<(pq|$MQ+$D$m<^J`smix5)s8yTa>j;v;bq6WV7oUQa!W=(3R&&^^n4
z{Uf=84?t&IrDb>sRH`3ILV}YOsWojep50A)7Q61HD81UjvTE9vv#wG`z{$jTS35)>
zLLK2Dow@<UYUk}Jwo2W3<;&{F7Yqo9%$M!ROX`NL<&!u#l$NE|KQ<W1MtMSCI&GU}
z!lyjv^;Y?Kc$(mM+d2KxU1>#+p?lr=NS@ueC@6m=0``xV<4>d|lnnzPjYWb;p%ohZ
zXsk7gE*tu2uJ8jMe|`9)KHE?J>j!Kk+mU7(!!}greXC_Vnx6AE8trfb<6D!ZbpsDe
zp}vhrX)QnaUH@p_pj|ZNvpw<AT(+rQM8wIqbqJocMyb0?S<agIO``w4wW43blfxZF
z6T6(-Tr42Fsi>Ry*;2lf13({~mG$Ow)zvkQfliEhL8P_xx&U8cy#gLWMe9c?Ibg5E
zdn}!KSo^4NmXTN6dbTQZ1WS}pVr0(SW<JSh{(0Lf-EcEUcwDBAeSs?oJKzP|$#>E%
zN!)Z@PMaS~OnyXRcIR#2W8U+JK4gO*%Xxg?0xWO5ZsB78v9dKUVoWmfvA_$tzJ|v-
z!ME~VM8Lw<5ry99#~eF$VP<S&@q4v%@9S+mT7z9$kaa`0jYb!KZX(KoLabw{^GMSg
z1u&fbtobp=Eks(@%AcGe?FvYlu`eb-+*}aLgt6~ZzSCJLyHiPfmp@i?L<)q-#nC!t
zcBXNz#M!G`N^r_W@PXdP64klYV#D6gW=}yT6qrQjtSkABMMG4m^QdQQWo%fGt=|EK
z@7jUqY$LyG+i&LgST^4awiRRD$0pg1pQy!br<%WWEVq3;*2Y~#px@WHV+>U9t^~1#
zbPVRZI0Ym7{7_X6wV1}Ci%!1tY#e?qx^5#McWjWxg)|klk0;8!1Kb|H(*$O2nrGbA
zt9MdON34MH(kyp1IkPDeNj(~3iV@jO!~^yOuxyiATI~tp#0xb(o;W-o?w`%li(J!^
z18RLdkNwxa&e{dc!wCNttL_Ov>`caUwv_L5wd^_TeBG8mo(nyN+0$O@lz}%J=pVcB
z1UyG|4ga{~<UYXd$4}H2wh|B=R933BQhV<d%eG!Xa8q;XCtk3PsOXWf<ELa1Tk%u0
zal16fA-EH-WshZxu{~ubcG(6NeK+4<x50cbn~Re+6i*Vq%%CVMPedczXCsR_n<-tX
zm6}iu$aA#FxjmNeMT?Hxcs}VE!7a43fk%iK?TN)gCRi_oLpW^NWO;{yLW$DR$amcA
zZ$Ir<M07NZg)Jw0)GQ>bm4ze}lb4%k+A4*}ROcRNi;KSA>`Bac%lHn-k8imYo!S>y
z*v?{*5F9MJi;lo#r;(^QAFv(WWBtWk{_wzFDNXLy0<IN^9N}aWPBe%zJxtq0+bAU<
z<HD$y5V&sURL<E}mL-Y=v|&5?WlzCM@8l;rqn<STL=s=w4cNdZa-|4rDp~sAC(4G`
z<_?AN&?kIwPk=nbpGfjHk8Oa56CNoQxUuW(=daaDqn=7--bSNK>@75TW1sK|!aJ1)
z$Kw;(E_9u&M^6&xO;qjPf_k{X)+RsUsQAx+YmHC%(qO)U*<}t$Nu{n^D_Td|!SuJY
z5iy>U!pH`GG5UD&Rj;z80_XGr8A;i4v<jido3y~}K^nSIN%GKM1sF$+J3b$?e!h!{
zuWrw3D+!jA>Cf9*G|+6!+Indjjo_4Rl&3us7;Kh?FqcJ34qCrtTiKBE72LAzfbo@Q
zxsmLAA`xo8OJD67U#GW<Q+~$h$JOV9&m=|N#c4G3OssEMh{VlqIN#~05j{i%Z6o>4
z?LCVeDjMRd$kdKj1y0?TE<=akSoAu(hYd6SjIR~G1@(CQ%F~dwi2!jwa=pm{)_&b|
zqKpR=Z9a47dgi9ct*R>-*a;P$vi389#k#T`wZ&(0B6j0nfm$wlnZk8>xvEQw2*z;+
zIZP)&o8D#JQjc6LkeFW8=oN0J{j4fA%2BtdMZr;khRoSo76=Y5E;8#m*b{8djcA=D
zGoB`9v($YTAZ``7++w-=@=>XWqArfJ?Ad2>U#9tleX{J!@VvnF+6F%9>pji+Kln*s
z?>$qP2@QQRrv@@j%Ail?>c5sC8Y9t)9H3>?>Q~8HDex#K!ssV`)d+t!_Q{;mxfz+>
zjej!X!9<$jSUo<^?OSk7^h7-F&+u=P2?D|~aK9075=zXP1vbpKk60^ut8tv>@PvHw
zsPC~%ivbu0hAhoeRMcsM%wst!zG2ilQHYcoU{SYB5`@mx@30HKd?#6oY#w^dNk1nJ
zxx`9#;#08}&EALIL3Ahg1_<{?yx$EUReC_gaSc>dLW8YnO#z6vqu~5wwi64&NzKLQ
zQ#p5pRnoaX@TsyuM<CUM@q~^IF^mo6x8qQai#A+J5;OOhjeIJXvPoC9QjLfGHu|Yt
zEsXs!T^svU!4~Zp*Ob|Ksc=ZfZS`y!qQY*PDAh3n;p#A1bqfz_jciO@qr7S)=W2ec
zh>2ceZ#ZLCn4|*aKKXWMRaYoeOlAm+pK?^)ubcPMrySkxyVsU0h?S30oCmkuC`rnB
zxT69uW{>BrSK4)5Wmv!V^PL+|I!H%sr2+;4%37HKO#HD~twwMi&p+42<_I>*WK6eY
zf4W)Pknj6eSxO0{vCP~qZF>HPo$^G&DVF`yzCCsI82EHq4|z36tjxMs?;_9b4Wtx9
zpY|12P+@Egvp}Tv3enZbryZ3fLs!-pt9`mKHVfglF`AVm8kd<BszV!#R*^JvGK_!P
zr~I!;U-D^3LkS?colnFQjMZgMnaL>i&_mWJrIb>|nxFQ?TkSl2&{|moCIRNH9bXWQ
z+-Hl`&=9w9(^ssXr2@nQAlPyYLl<zSXWBZY6QW*DTQ`~_PBm5qpDu_kGg0eTeLnv(
z7Q6YRYv^fPtLEz>45H^ovGoe7du^k1LLjhd+pJo85mLGp<E;_%Bx=ZG!gi{0;Pl|T
z{es!Ev6Og)LSx)?!v>zs4Ng{3`H2iZ8!NwSpA9|h)BUgSwc%%TsZd?IwcN<FIsb4N
zgecB-<%Xq{2`_Shj6Rzi{s0#6V^slIJPHtd*T$cXZ77KX#O*e|%ZbGGXMLewIf+<h
zqJTFQt`n2dF}$&2>NK)l8QySrq9t_V&sMW8$xSN;F*RfDEPRr#<msfKdW!+21ScH5
zQS=ROvgP<iNU2WMgIwXYTd!HS%*Ome0M$#Npfaeb`}xFXM)|f<2H;s-=WVUb%hW77
z=WJcV&Z8nv(3p*UCyN{$QT5k0O9u!OuGm%?Awjx0-nQc-`Y@?>cXC=qu+Xxv-cfSe
zXL8M;x{upHc_M-oaX~&4#O!r8^qIgBaYgcEL$zbWQ5FE7=MFs*W#9F%)jpF`O-gxo
z#%LCUkrqQ$J@sG&$3Np}HMM}TR@9%JwTXDC3`>)rNpwMAZjI94%a=HUn+Z|z*i!th
zR+b>CHt@ASQz-f!Q>I`K7pu`+Mc$_6Be74_-c{+T8DZa+vpxBe5XQY~T{gSj&p0~H
z%KEw&zuD>Y)-R(cMk&N;r93w02<&ESpUEx5p}$-C64w(*u19V^jBARM?&G$R0EzBl
zs^#Lu%?e0mN`b_K?OUY>?AfNUsS7As_bKwu?qmgB7r6Ah=ejNVq|d?g48+gX&fnYM
zQX%MW|Lv9vLrdjYKL;rvUh)w`4_{{^`31K^X1bP7AS_#LG@oR&imtzO@gHnF8X%Yj
zabv2cUx1WOEIFd>!k%Txk!Ln@m^2b9C;JRo{!PsaQp~^o-3@EyyRZrfd2M?skw?PU
z#nK<TBcf;pBvabo%K_j5tJ5jtz>ejjb@Kz%ycgwNoYRlrpa_0^QGs}0b1-XbF^V^i
z+Is1F53e`k^@H!W&1|1k6ezH^s@|lv+pb0`{1cTpYdcGMH;qshC<g7bzIL;mBLkoH
z;c<6Cv_AOR+;ay!Y|PMSV=?6jti;iT2e9GK#$w7W@I@Pm+PYNkwpxBVftU#zos~UU
z86*{FLmT^So(C+RWWnQELj?Lot4Cu~E+CjFeM^<iKolfnGT#+WM$6dvtfLb?wa6>u
zJDLHHYQbr?$6DD1?jM{B?E(M+VPyugSl}mCn=O?GWYe}>Km+}bTPL2Bn38qu7C7Xz
zrF)my>8c+rtgK%qa=xjcV=Gxf%qteAZvM4ay(7pB)Wb6x$6!cV+q8{n5LRGH=F6Y;
zHPn}kCb1Q3xG;}Ld+CkiQ&*%iC2z*S=5w)x@?{}2DN-G@@}G+%aEU!P_&MJ)4|4(j
zTvBjsQZXosCqLVJHuAaL16+=*mY=xIb9#(^E|H6B2!xpNvGSdWqo-{Ab51`0>##vq
zFO4cT#zerh+WEV`v&n=x%mISdHu9Zpvxzd!lq0-~I(`#rWv7ZXgBxay>dAkx#XI25
zqak4NmP$)-Uy8*d3;X4$jk>l0bKWU)CRr1xCv9xEY9D%Jy?9y(KdOFy1Je<&uH=&p
zcTU;b=X|H>3b`IYz3Fwfkqgk}B-<>Vi6dp3AnLcRXabH&^1bacghO-rW+%H)vd8Dl
zK3@*iFpTVR3=QNvlH+-5Zv~tf{Jbx9?Tt3{`Pjo!6hV+b?;sxJ^c;x~gt}KW<XSwj
z=31fAsNUof6~FF(hmB`9lE*0Nr>IRm-#NnsU3Dey=N*3zxejSEd)KGy)`(A2s4<aP
zWy^1|R{VPV+pQhN_k}(DVpb+?9h+|H^Nz^`IcwT-3}Y`C*!g^tF<kt#T@#`bc~DUH
zstH0-1LVq0-j6<?e2uLHP#Co7u#E>SDj@UgSx@>j#o4*Soo}PSgszJP{k+pS_xP9S
zZY#d7b`EQAZU>aR@B!Qj;IUZ8iunr(F#)<*@_{ep`3__12^$B$;1h@qA1A@k7YbRH
zf=ph@kmV@wxEo%MWC6wR4?Y0j!7l_GBeDBcwza#sxEaT2mX>H0FJs-0Z0rj@v9r=4
zkAESF_Wmhu;;>j%dDPo<FQw*0z`t-HeOkVdv}E5LD$+(79NJt(>HC7?;r-}+vn3t3
z#hRj>J&V#TmYzL5BRq683-njfV=w2E%q6Z_C!chwI%(YkqnvbEUR}+5Spo$#XKp|G
zA@><EfdZn1maTok_rhxDHMSnbSvTE^H?lzmeo#)+v)unTOP3riFWOc#@}_%jJIi|3
zJa9xN_dEIS1OhE?F-S3=H~V5vELsogA_l(b6SFX5gI{zo<_<hzLtl)&s&&D}H;No?
z*hqO#*lU~&zv%15OdPe*FD6CC>IbQZC~}sSAo!Ave=*g%%{c^E9)GwV&A$GPHc=`|
zyyiT?WPW>tHNIE~jnqT#8EwrkW`+qe{7mngYlacV7af~-Pd8l#*3NpcOk8u~3|h>0
zQl28#05Foq#2{u7j@xp9L*B>a1!7a@Amoz@J&q?B4<SnLi+MiH9(bSi?{pX5f>IM^
z(+Y{a7Qb<JV>@S~I+9P@MrmehX2v#44<5#Ks}$sBhe>C4A)a6U54yPR#611^Ys|jn
zi?Q1I;0JBsOG#=$OEOA*DQ2*2HCaLoRqxnH+D_25;jASU9JW*WbP~k`@iF5`B7y^t
z+vt}H4iOZFi@;d6MtB|^7l|>29pHG?+7<jX>hT2LE5P_fmY1Z8TtQ9{Ccl)#Xun6A
zX?!Uthks-*iGGBSX)_y!l7oY-s+;HOv|SoONJ+-<#j2kWrGo{wl<zP>@>#se=$60a
zd*D5&hqJkpj?~>R<qG2tV7>A%b>M*YvlZc$^6Mfcu$4an89!WZn>a7=Wi3mGE2%Oj
zT(tFAuXjq@v=P72!7SKjK)^*`=ik<sk|rw*y6E6`0;pJlCu}DhF8TvZBF(-Wi_fi#
zb9&&*Iju89dBdOK%LNxAm$<?Y4t?22zjw-pzw8uXeqoo5eEH4=6n7$BzSuAvKURzO
zNfoApX7sKGWdbkXg|U2hXYG16{^i?a_?(ci=l|0`tAE+i96x4+WI56kY-a-Y<&&7P
zi`FQ!!1;37nlZ3FBot_UIoDKHKJ6$4TFa|v@yj{cx&3V^CJ!^*L$hqo<<cMCQs^zd
zoSWdlw1>dxR{fLh%z8P0E^cF96a+V}fc<J_6@JoVs^k_}iLtGAM*hjxsyXx^hHQ2+
zWE9Z%kjru-3&<Z-Lgky#_RX@`SZ-mKu@$4924A<!GQlmJazJ{vQ@xXmncH@lE$8gK
z3SWYO<=g*@{DhwQZ}8EB%Smn?n6sf}N1*#&Ys1TlZEkyFBg=tp)zt}#Rf`@xqP~rm
z@_VM`z&{oZjlRjom-9G@i1|qW>dTJS8R2LmS`h>HgiTi6oPj+GjZ`Cwb7*4AW6dlP
zn^M*_$E;O`Dmvee;<Le9FD_S}a*=NpE+EFcUpB?dQS%8@Avk0#`<{6{`kZcn|Ipj4
zm%WRPziRyoKFEc59T>Ki(mTwPtrhS9Oj*8M+Z+Y3Dn$mnkp<9QD&dav>zkz#<V2!R
z!u#1)d5648_$Di+?R=*L!<SyiMnHPctIfVr2*OK{tH=g8@Rh{uT{vw+o(8|-YpnnY
zHdM+>YA!^A4;%i9FO6awA)z0M(sNjuS*?_oSs^g$x;gq4$Cll&^j}G28~LX(8~;in
zI5NZ3C7}qD>e-0461hkH?le&vjyw!LW0R%1>DF#*WPva>*Hz&;YX)Tdr}xZSE1;<1
zW%QNY^%08}aZ3!x`6W_>S5H@HVz~_UprT21qCw<Uhpk(_$bis#@q{Q@M6#c4h>Jje
z8c$Okg-p>}0_wMc+=c7aJ7um0c&PNoR~*k4q?XRlAqovW`c_V#bDgY<dv|g1#}YSy
z#L8jbcCw}}wDSBA6K`MjhX46B_t?N!eL)nGmsKwh(nkz^H4c@N2&D1$`>JnqcV@dW
zVk7Z!wez24$y)oWV+;H6jE#P^5CAtO0|Rl9`)Xkx@Su<iB%79s;H1@~*WlqMvbxw%
zg(6a9@5!q84T?7c5`<VBp3~MW71;HPB72I{s1@TQ3YN9w=Leu|i&Zmc;7*tFNmr{E
zY`H)vUKSU^PE>p7;dfg1tB$<;NoZ7h#s!C-`!VW2dy;8fi9!#*$JP=aNo(X3qWNAg
zbD|)$%#uP2Zj^T}_1DdWFg0Z?K6r&w=-O7iTkYUa0}Dzp56p(5H{xdRWQ|af>n{$C
zU&|F>;c%gVQ9)Ck?~*l!N{}e)Ye~8HcskUfuf=*})TQ<QnlGtmA03I`NK#*GwP?t>
z!bmNFbxCOm4_n;=zLvX7bGeUe&-m9I#}p@*q!{tQ6rux~sKCM~*hveq%?kKZut<=M
znQS9r)2VXRnpr~r7h(}5?QMN6_6k_s{+cgX?M=2=07EN|+<Pg?DB#%g*AgSiFeP{7
zPC~^cdAD_=+`oK)OoOFk+_;~zem-gcv8gLSpGP^Roq5gHq8&8!xDT&a1?2t*%YwjU
zBj%Fg#LJ6*Ry&inRn3`##)G%p1@t|0)AK08qGwcv(@dy+J)y@99Z`o3d_AeEoW=&f
z9_y%t^IaSIdY)?;S&j7>{<_Z(!K_BU?u*PCA%yvQo(hmxPOP}Ge?5=nv_xlZqo)NC
zG8@k-iVClI@se=Wt9P>52nWxJ=w|iPtIybE{_fm+pEbUoh%!Ah!y%7YroiAfo}r~}
zl|u3>K(J=*9G|53*K9GX;EU^;9uMJ>?>y)cBthBVm&;Hh@~G~4Uysu#4#|)rIa6)j
zn4s5__bOWwaTU+shl%Ka-7yVYWm_p7;cXkf8Yn>%Jnettie(A6NCK3Tvj2?|tDOfI
zh#9b1E!PG1=~e<2d~|r&y6)b#vm8QiOiOo4yYiHayOP;A@=QJ?Cy#+|#CnEDq^Lg*
zt);|%w{*x7kwHr(*!JH@njrb_K^ysok9GDSel*{3teJkyM!(@$139SKShi8<isCVh
zM=zlG5n8O57V+rCx^$xYh&PW-MpH1G&sd`Z_&RC7#5fbDHolWtuWhaT9@=<T1VY~~
z-HV`6N6w3j-^k6en_C-F)}`zbt{I}Y94)?E;h;MKS|AoC-rvYw^aNY1S0F<|-v36i
zPe)1?dFNKjOcK73n|?f@N|zu0dKodMoQPvL3OtAAuolOH<r|eCd!7pkDS5X_`>-Pl
zj{<UTSA)e<0mM^Cvz;<nuuUd9O5|bR%w5ZMUXg>PdmZ@ZpU~fy4Sq9c1iiZiy+hwj
z3bA_XHf|c(1m`?P)c)5-zUfmjg--CW)xLT6y<C0BMoky%(Qi5?;OB|CQU!@i>*W4o
z;{_7DAY8gpj|$k69I}Z5)qWy{!|hLgGjMI!L2Hz@7-_NnZI-sM6Wc8bXDiARn917Z
zn@ge)W-Z3I2A+{i)nHG;i7dy!VBer$oicJMoY<TSTGh>wd#ur376_EnKEx>Po4I^t
zv1}!KP`H6(wpPKe)Fk3U@#`q?DVMG&=SCLNc_y1popGDli~^EewXFi^gOUK`CyB2v
z+fG*27EmTt1q`-t`I5S8;fxJ@E3iUVy;nGR&LZ#_{Fd)uCT{3kcdsX1yC(NCeB<!9
zat9y`wvqDi4or7y-}044$F#>rqb~^lWhmBIKIyo*ZsXrdRP$uAoc7V{0}t3l!sJw)
z$(H%q<hLqJJ+Mdl?izPqAgJV?*UZ+H*05`>Z`~gGcW$Tu^?h}<W9<Sdcm!bxSRTSB
z`CH{&^T@2Ew_c9HVg;*{p#oL@EzVe5w*Z1F<hb?9FlCB|jHO>DT-*v1evR%{@*RGS
zIF{;^!DAp=S?wqcwXK&{gtpunHp=M9+v1Xz&48HLom<&G+$~(jcB}>wai?sjG})Zx
z{%@Vwk@yJNKqpTCM9H$rS02j2POcBM=}-eEzvEc(_y5U;v$Sv~%wKNVNWK&OIi5U0
zE#JxRQ=to=7afRR56%(Vd@O$No@OG)vn+A4xZn{02|ZrDYE!wOTvcKdojkc_6bx!I
zU?SdEqm!o*X7ijz%_v5vIx1Torx$ZGGuF-zIxlb7VkrhoK$_A{;+O)C&!TNwj^^L;
zpmoYnV3ftG%a5sBx;%`jG?%*QWR?gm{iyfGH`q#cplP{kYn|LQ4*=qN=?#wQ&@{f>
z8(H=wfiTe7<P%~wHu6^Ki;Efu!*=v_5UaMGDCbE!aR+!V7Sf$wwBaHA1JC)wHK%Ow
zxw52FPPB%ea}>bZmpI$c1*KPTkmEM;+}#^L1gvpAs6}h<NBb7F7v!igmbK+ub;ZWx
zi~N7B{#>4jk*885Z6ZF;D#!d}0t~%k-KBp!Z;fb7F{#mqH=pyxBSe<9s?kZ{tGJ<r
zYtMI{NEG1*hmI|lv2bC+B%&%IciM=VxEvs$3P-A2X5~P{@9Ji4-EOW~uj*5G3_?9&
z(%8=uCtp5_i|WdAzN%=NF}hnTu+L+QYwH2~YUj^wBilJiLXV@ggMEo@#$0mpZ&iJs
zhN5o=bgP{@9?;RmyWke=NTSc%uwWWEa_2bb;PcD;;E~)Ms2DlBvxFogBu9m`KWD>7
z5(DO)1F?}TBg!S_;TZT*JL33q|J!V|boelGg|RHEhGZO|<d5S=e8;$o&($pTs5^R}
zO+;OoAV8Un*ROw_HSX$A@=bJp){Itg3PZyqoonkz+01jKK(qD{M^jYI2W+u?I^-Ul
zOW9not{TNzTRsxp!9m1trSz_Sb67NY%jgO0flSwWWqj`=??ykW5eVK^0^IlhtF4t8
z-AyzhThHEfc|P`v{^IEoCD?0NT-jzR|9~tox3auj9#VpQ6t*3MS&)IwPSzz(IA82&
z>_D!J2ptBF`Y387<{Ug)kqvAjvXnAyatt30Ebx<u^-~D$ky3m&(uvwppS#u0oAJ=f
z!of;JaFVg|4h{jUL>P(_Hh$C*;7zZ!dVUS69zl>M0<bwO*(a;YFSuup67x0U^Xvg^
zUrOK5`c||#zhLe7HTpVREY;o4fZR`*=qw#|1a*^evFOWL#+CLWXxyedQ3i(@nP;-B
z(a&s)-@WQvcvkk0-c87uCO=?x{aiVkC%e1kirro-@Iu1ZkKW-oq>{9a3J@Hg0OUMZ
ztIcWvJ+_3h{~WQcSRV|2_JX?GF&hVGTeg#s#8L?oH}QfD*s<KUGs=ZE5KlS-X>RaX
zIho-Z{hoBrhK}W$zMfQ@@tYHuh0mH@G!TRuI+j?oN0BMA8M?jwG}Pxjn|JJ3nHtGP
zJRm0Hl3IHFm@k5drP0{yrKz`Ks5KG2a4(d_^<?Q2*%+2j$CB8?d1D@oEY+IXv&T*n
zhs1zeQAsLv2z<M0;xZ@Q;vGn#))I<A!J~v?z~b1xT!zmH-3drx(s2ZW&~&qQ2zCRV
zdr^K)!ukF&-&ML^t{ihLfB4<DmX+L#oQY5tko5u-w--b)`J^L<JWS==%a`GkwqhF5
zgZvko5ZlgjO1yK+c8(=RantnRKX%*~Xx6n$1INq3t)w;JIS1EZzJpOja}kl-hmJe0
zh<k^=4IfV&o;vVEG&&OBHE|;4)%V?p2+@Y6!)Vne#3K2$v3$qT=}O59T$ji5on#Xi
zIXsjat$y6`?%)(B+eGPygX}q*EUiMiWd1>R3Ts5^`NwdRZI;s5g5I=N^+ebjYajO|
zIehrAEyj%CID*&1QdR0axV@Y`h?i8XFvSz>l+o=+KQ}8|$ZiD((i7k~23hS@?;?kp
z?!x`!dB6~;iY>QNig<Rph~cpoV9m;5w|ov`0k<yMNa%nE{Mal{uzA=@709^-B4ICs
zdW0}#JEh~t7L+~3PUOx<Ry%D2@x<AB6#6uHB4NaQ8A15aiNc)9n~#gc@QJ_}Nyc?)
zI4`yQiP+L|?K|z`e~B$8XWm+tCWigXM0?}=RU6Iz<h*>&#!mQtxaT!Ce!^EoYVCRf
zV$Za^O(zNvgg^HfKa)}E^+@O{;CrA-vvdbpkOSx$?fMGfqwd|KC2W`W@pqe^vBeXP
zVJIlksa1VRluhT$0ha8aI@uJuRF|B~<-8kBDH*y-_fGikn({Ce{b(1X4Z;!?&|KLG
zRprE6OW>tuxoqnv%C2g{o6~}AL}N;}MSR+9Kw4E!0?!t=6$SP>6SN(@5%B@T$1;j}
zR|vwyoh-|SeF>NTry=qKCo6O$PO=T2%%zqXO!AMR@_iriRfzn0(pT#BM4&oZSO)nJ
z*Ge&+i^$sJ<7($0iGE{aCzJC-jLn_V#!m*SNX1OF50vBzlB&pV^^?8`<O{%&H=%XG
zVt${@?q6$->|PSpOV*6{_xzK!vR1KwAkl}l^PRGJd$?NAbusI695V|`CmrqP6oD#0
zz(o!cpM?<7LM^)4j6$j_rhzC(FG@(AhQ^~`28yqu4zHCm1fu##wBAF@vKH-Qs>403
z%-a1p&mFXl3eI$~V54lFbX<brcmk#_>Uwyj?VLovuR23moDX-R87V(tVI__xM_)I0
z`|f>nSUGon70Glr*v(xV<{~IJ`#IG^-CQ(sqKQKpjuw0r1Y?LNMYralt#y6b7&K8*
zy5%%c9w@SnAwe0-cigbKfW!@ZJlj)vN&?koiNc`}D%K_n%*6D<*{t->f}OHPRUTPl
zvJTCB=d37nM*L(e-$k6XkS?^Vmil1sV%L}cp?AK`ma?#W)j;SK)oZz%8$u3+5{GoE
zLJBPwOiMRg6ul$UT0S8)z!0wAEgKA>DWP3s>fd;;t=$D~?ByG_9#yfi5gpmc#i7}Z
zpm4;w7;fm%%KZmzt6PLb9zh?$#C8m58mawG1*-&jx;MRGb}H93Zag+{s;t-9%93XA
zRP60jxNKdsp;I{p`=@1PfXCS>$4Qt63E#6UHa&18wNpNuvyXB|uNw9+_1q5#x|)SC
z^<`g<ml7TURXg##)$^SPo`*Il9M(hud1{}#15QRKtDVQJ5unI9jj-vb0+|*4;=>I0
z(#swfL-fDu_cFV^eJadHGE3CmLOEMJ<)}Qn9~bVW_>Fu6c}%ul)Yx~jPIU@&ho{|5
zpc{QTQ}&GYs-ABlS?LG3GMgvHP7c*Qg25yN3$~W;q?IW8t@Ug|=O^C6r+n{@-fNqu
zV(Fh*{Y3m{Tc;A&X55@_I~xlt%?MPzQ|5P77Q1pEA{Y7TL<wx@2x?#hr*px%xw>uF
z;OW2z5zt#UR6cug$lUO0$A@N9c3>l=F#>f>!}Zlp2S#wjj-C!wi%lL|XRH)&qTWTh
zHhwyXpvVER|53#ixgJZOd7H>)N)*8}NR}t_of3G<eODrmMu3HRnMAau5}Tk130tRq
zDj#WBJAiyr`u8lT>w*+K@V~Zn+K~eLr@ye}`2BX>Ksp6@DMUCgx@Dda&ft9-B|*d!
z){ieT(1d?l$vza4(Jva}0}6m#?_OX8N~g1)EgV6rypi8!<*tYeikG|@leK`xMD${-
zfU|$<;G>Y@?E((0#Sz=7I=I1Mq-!-fy3hE^5gG(V@W7dzdMuk^RtC?+qU$nv!iLVo
z+CfBIb}U1s;WIveZ@_aYzsiK@mh)VM=gfDhvt+0**yx#@$afHNDQX`=09&<v5|)bp
z*$DMBfs={cHW8(vK9gr0p{8bLtdSs|z?BSJx)=;cC0E|onWT%PL&8iBv}qT3Tn8Iw
zNsG|{;vdOqmP&I>anF{cJWED`3#{GQ8rG>GB}XU7aUG3FH@ZFS_E5b5^WjIVA44T-
z?gd-PYR+sh^~X5f*0M8QGs)#u0e6es1?W22XB$-srrb8m>O8^Cb+2tzgT-!lyOg~B
zG}?`w@&tEIj@q+vHcJ68aMl;>HGg4)@pj+b6u#tV%L#BqXX$NXq}cFTUwmbW6^%ce
zm?7=~C4BYhMz*8n5i<-fTw_sH7dMO%#<Ov@BxhJ;9j{mK6gDtAoG9-YLUB%=EDg`?
zRwjW)77@qLXICj2ck`?x@M`C6tY52?-?dM{pW3B7R{-v{=z)j;5x^l#kTiIybn3jk
z$kOw+obNz%86V;~{<L*s-v08v)-Ch$e`EG)y|a#ru*TL8KxRDw!AfZt6@c5@GZAY!
zEsBVUk!?IZf;T6Nxsl~v#>DEMfoIw*BZCvQt@5A*_S>cPg!(#Sik(sbQegIcP6~E_
zHt@W|sLTL^&nKP7I$b)3q35eT&75mOFWd0*#S&vy-s7@99m!&1LzJTMI!AJ?H09jf
zH5+}tV2CK8jTH#?(Cv6B%a(PT{=8%1>)&J(*|Nuv@dF!Wd#PHdokl5ufUekMH)BBZ
z18qf1Z+w%rOaFVOp~beC^&+-<+S3;>(Aa6xm!D4x;rB~a1lGx^;_lRaK8W8EI;kb=
zmF8A4AqQb)M&6IYN*T6Neb)UV^rUO~E`ok&>M*m&cMt&N`C&}8ksTdJWk&emc$z(Q
zNDyvS?aCP{{N%Q)@-i~t$r57iSV2S}$IZ_9f-9ivz`1g;N8pRWbL9YzoU`X-a395S
zO}624PAVQyXzY>ry<2e9&IMl;<{oZA!*<dMFnTUI1%w&1_#W8UIbYj9<0gNuAR)HR
z_3T)8PJ-dMK42o>$)rQ)k;DR%(d)f6j8+@x5~WCA0QqiK#TEI1xF+1bTBTWu4%s)H
z{nkF0lxvS%xE8ZSupANup0cIrWS0W-EywRyK8uZNr_9z&7!Gw~W?ly@z4D1$9XEOF
z$KV#+>UO0XDJEKLF#&%&V(Wi0YoSmMCT}C>mb8YpnE<+>1j8cpn{Abmhlkj9RU{mo
z#F6YI++B`JJMSX~gSA)#@x*K>ul3;hSn;$kmTjyJ<vV#gyQM5tZ}@!KzqnUU+ekJ7
z4~;UkcHS}SA2>lq&nHGb(7@1?jeb5?8J)C^pO2&FAkmxZ=ksET*i9joP9j1Mj&Gt=
zIlu;=j4IqC(W915VE{-Jnoy#raUMzPh1U6U*&$0C(YFYnSe1$(cnUAHSj~r&^-D1_
zxiH8aa{0U?=mX={DH8=M9kXscaix*dX0O0FgG{7f3IYi#w`gGDQCrLRvK~Hd>($dO
z)?_2bbx2Nfn+anV`+pHJF%4VibK}56<1qkHFFo#;<fd97u^D0WWWB^|tU6J_^~#-W
z_J9XI8SLdkkgwf_qAZ*0g4C<MT-zDK%CNJ0ff91mkTs*wsn*M7mG%2x)HPWKfiy;g
zag`qH#fpX$MNCXFdKu5s+GNt?$xftRT9nTfy-sAAu!hub$1x(ol+cm*G5I7x*ad5r
z0V=tlZ0TF2FsiH**1iKE+^}TUs9t-HrcEv#OVww5T9V!@R}B!!z&bt0)hEQ>bW2|c
zr*rwUy2z>sHCZ66pVdVK$t2ZtG<xHGwidts<-N8ZLw@Bc+`9PRHhSp{hT%0?bc09V
zjAqa$OtVY2qNBS#Pi#9Nd*w3>bEiPonBDJ*rtCtTOLCni=H7)u&u_*uAzvD--d%H>
zJ1JF$@|}kQPx;R91)r@c82^zAF__m`Endz_j+1qigY7yvh_jlFMep3kB<&i%;8VB@
z;l9=5_iE>@`)nd%l)M~6*$N&Zhf9pD5gqO&ruc>2(P`rj6Bo6ShlyitUvRXZA{IZy
z{6b<WA}QT>mdZ2U0P+M~zL1FCB%~L|dDK<VhL&|>WH7AOi}xOM&iZATM3j;^M>*tI
z$^ba0F4|ho&TDNwDk^>`26I(Kh#M8*3T!i{52k7B!fpE>a8dbh$Qb_r^v<>m%1l22
zDcp&<y#Ic)i-Ed#rFyV|i;jnHxYq{rtJ^%l7)q``bTLq$h&pfz!}%m$;;4;W^pz*c
zhF^qh7YnHh!!@y6Az&UY4MU_}fcQy2F;=}(w2&|m9>196^nQE{>K7f+aU+?tiHniu
zUe+ErX#jt2{)VVPsiP+^`nEz^XpI07b?gDcT{p8A*RLf>YUR6NtYn*Z+}ag1+!7rj
zkJ)0hG>iQe7p$cmIuw8tez}x~Yz8^MX`KKV;cvsbWh9({){A-KitIGCAHW_Yxb4ct
z+*`>;tX<5FIbUM8truY3isU5M#7nl3@7x2JDVEXiW_13~e}(VNR_Tuo3WL2}L9W<=
zJ7s=2{in=c@JaS4ZUZme&2R?qrb#BF)eE=(C)w6@oK!;u<{Q|>4ZV<O;~}M`8IG1X
z?f2No3xQ@a>s{8H<SF?M2T8d|atz+E(bB>LI;_T`1<Xo`Xcu3gTrl8P&nLJmq@5Gl
z5=w2yCP~L8Ur5ZE*|%Vgs`yKsaLv-mj7VmyfF|!}(2neh){e^X|FSo!Dj$VITY4cE
z2(K3fTUxHZzQlIuyx?2sRex>Wd@q)Z&<uYvlfh)l%!=T41<psfB(7Ajkhk3#j<dE_
zpgpXZH|yo8xv*=WZImgL)+SH<%?hAPOw?9P;qC@0=C`ZT7vVVE4&Qc4dq@<}OkMKD
zZyef>km6FoL`hwcLtV-xgu9Rrnj#$xT`IKT5(1CmctUqO&5<#3DNg|mZ;%4ivZxqv
z1W=cJdrsqnJC^OboL{6sMn#u$@#aYhRF8*!ep{TlRQTBl=ZV1xb;!?Uli7}RawI<2
zrQ8~23>dpf6w!>Puqor1YTW^MHz!;*6dJI_Y)qU7CYoHAj@wdHo<5*6aJdvfKzt6Y
zK~QX+(gb~Xt$V53A}&M4SdrtZ7Xw#(`F;*1sSfu9+2&v6R=-jWMcVja4cB66nD|q+
ze(CoA;k)m@cZAtSnZ!E0)n-&*?fl)}+t#IITDj3dn~+;&e{o56hf&)p6_EJYWyk8#
z*V@44vX3Rp@e8yK<~vCwlmzJvQnjH{4z5L+W3}eP`A*JS#~BsT#cbqqf<ojQmk!pd
zLKLH-5<rYzcC7qIg}EQg?_o<4*U$0GNh8T2JIyq6>nYts36=Z;?=YK)Ry<VB<mJFy
zlqIlgjTqj;@33Z+!Bh%H_j2O@K0?!?J#A;5ldE#>S-hOstsFd@1=(THNX^SBTP{Fg
z`$6&>cTVd>SNFf!x}|xkv!txP>^tHSax3<$iipa9o|D#A@}0~cxVf(?Wx`rC_Ru?Q
zJ-^CSL#lnn_u43}xUeM5B{rkR|2bk?mwktEPJ_(b)zDt#B-<%#LR=UBY%e-!v$zQl
zyjX|=halD#gD(a?M1-$!zZ!b6Om2@;eK?;mq||yxQ&W4UtoEX>>T2iyciL$Dma4+W
zqR1ig^^KR-B`;uDvPrJL=zvQMjDvSP%)Vcvn5;_6zeZ}OtJa9d7T9^s7afUPxOTpn
zYlqJ~v8LJqAt`(2B(f?`q}stBpo|yy5UFIA%h$(cz1+FeqdWQc;TyUE5|-iIa(mSn
zkxWMaN-J4aBPE|$3y8~FPLos-Tw+VN3IRxT<BhCw$LjE`ZDx&=D6EJ&TmaK0$F_6D
z_RZQ(G!+&jOoLtV)j?B-o0|<>$-5DZx~#P{N*jzK9JSXGK@a5<c*wCje8q8Lm(KJn
zxBq9SLNaUj{Ot!0(LwEsudA#sY&7c%G5R)kC6LA0J)LfFjN|#vV-rFYB-E@&3rL?3
zO=b(qzKwb*ONSIkgb{1Qfasok{>GZ|bB^o<t*FPPqk&+)eZ|-QZ~np-uOz|2<j><m
zEoGCQ*PhL$6fnyLC~V<nb&idJb*|*Wb0^=bFj69m(iJ*}lh(U(2e}?4VqQG5SHcn@
zhDMuQXsTXZnh9TX($@2x=bBDlc_ZJEMk@6JI0HsIhZS9StMn+dcRcxX#w1okKvjSv
zYI)PFpIeg+h0RI7S`R-l-Q+(@C}aXb<L{J^S(rN!4fk_^v-_1Ub);WOKq4r3E$b*s
zdn5rMgnYgec8gC@EP}DVW6~b<<Ky`mt1<7!O;7?5ZrhJ*qxTCphah{h6i3jAsTqg!
z32XEn>zEH519J?rYq@BxzOVgX{FSw%5&6Zk&$AJ!J)Q#@Z$MC&qLDeS(zYDK1WA11
zI%QsPuz$(AWey1Cget6;ZJpvaBSRXJcUwOyV5}oS)=H_c%k3T4%A_G3R^Sd>FGFH#
z+-eI<+NiozUe3)@0x_+)WdwP)RlOrjGSe#A+;+YT^Z&SQ?ZmiNI}g3n?5a-``H%ZH
za5V_~Jh6SQ24;wCK4(K$12eQ<@^z4L)o>JE*w32hXBm;|rgo0&MlC+y578XGni%4T
zx{XDTo*{kwYMvkRX2hwkUbS=zX^mX+dNz@u=!hq$@zrvNJgvA+lAKabQ3V=VHw0)A
zJGYtdq`{GGtz>+ytBE_~Tn5`$lUG1NLCk*+-o-m0NFVD|Y3Zut%zyYNTMppmb7P$t
z;v9!mw^Y(mD++-at(S10;4+HmS3Yqy1wB^^42Q61Sj$fe{)fu3dOCx4cq95FWL3j9
zOQoqj@3XC{7%Yu#M=>G<V^COqd4coMuKCizgrhjZDo6x<Al|Wo#TrM_tHEozNQdV3
z;IcJztx$%()Q<Ud0+hOGBk`oPV;r-!Yrcf=QE|y1y_VDa4)A;|dy04_fqmnpCmi=A
zqPi9fs<Z6*G?>!wvo>+f(Q$4c%H?dwBXQ&{y%8Pj(u{DJZC;)aV@FBftx^ZqycgUx
zv2X2YavzSUi`SA+=?<*m#7hCVjyzIC#29fnnLR+<h8({FTi>#7KAnUn60j?~@r(|R
zV#lq2&GBUq_xI@V$tT}yYgtWUm9Qo*(=<IZV;lKN;czb5W(*`(i(vEBB;Km53%BDP
zLX17(NDqnHDRANA^OV{3T=?c(Q{r>s_Q3TVv*O>%GGZ_a0XBDJC&X86=z1>T<FivX
zT-9*@Aolw8WYgY3D9!7MByOELdfk^xKJa7LV@*X0p5qY3(+R#t!r<|!{tt@2UC+*u
zS$Mq33*1j9vPmL9GD(wF2TBdyxSl9F?}*d9p0wit-eOws0@-`tYwhdF$4no%<B76(
zJ$0z;UXCisN|&O})B%p><pf7ETy_P=<!S4b;bTJAn1ea&mdV+LM57nA;cPG?j}LqB
zD4nvEG72f0ZLI)`b^Bp%4(rvV_c^cRu}M&`Xv&$e&AYxUiW^B+ZL0$AiqgAt=RRiJ
z1-K@wEBVMy7E%nG=aSmlOOA?f;I8&kA>xYb5;<lDU-A{i&|-EU`iXd=<~2oed&yTC
zwrkWzUP_!G{fgCIiemzc`O$3CjS65874K>6rNR&#Xb?bfJc}Ghk}Crl`5e`=BbiwB
zZK5>(4mt8h<9|=KyGAsGqQTkFW>&(N0m(|Mj0QeQ_vGtLQaejPebJ{a#uKsqWDB_z
zqw(Co%N0oE_4EZeX`Ps-`h(WZuMu8XszLH(XUme4AGLn8g%3XIqhCraoIT{4fVHZ&
zTgkhmVEwNCE-<?aHn|bwR8rW@(%ikTwXFo>5!=o#Stw}2PBpX_IsENLE*ms>!3N@q
zV+VDE4c^G<N5cb#58d$ff9qRq_=c}1^8v4ZBX9vW8)Li(1YV2pXbdqFzTxYQ!#CJG
z7ESykSwW)6VObB?v$(d-MCnuKA~vDrk)U`;SsK}~awj*-CnXwMuvUUDBe<*9uAsUO
z{(9niTg-PSF+h<GtYQQ#<vZQ35$(Wrz@ukN!i27M@=1<Uh+e98#h(OU)Qb}6Ok_pd
zFXfp+6b(0AnEdZaQ*12;^vcun&c2cBzu%pfZ6jNfr0A$^-bjMW^_YB&Fy|F0oN=&*
zr!g1XuHGrYi}S>`6V1qgnOb%;mMeH@myFp!zEe06_cMly4BiZaj2uKlmz$0W5C<1E
z4c`o+T;xkVY$n9#atptiyZfC)JgD8w0m;e`R2eOOD|0nlTuQgGn@LF(5n$S*{*Pxd
zF^p0+Dy~~S0N{k4mSk@tx{>!Xc`|R7nLz@rKv9kIVjg>CVjozu)ZXPu_*ywOISqHM
zeKXkBXE~1-OF{0Kkf){iL@pJ$_NYX7Os@9U$r;cU0r`QOG>IoXYu$t%Coatj@*1+F
zmk}XFCZe6Jm8@TNMNF9ewsO-q7glp?rK6^DFt3+6!*!ghOa}SfZU#5QL~K@{72h;O
z%v)s*_7K*2JIAWr3}_qkN$$>sU%FM+_$EBkz^&Zv@G0;>gST=eACP~@Er)Jy&W5uN
z<cnn)$)_$1HM#n31#XEfDO#5DBah}gT`SJA<^(>qv0Fal2NgnQJX;Vm+qZgw2xzei
zzLofc^i5&cCrjTXDm`kAY*~XA*1iOBZC3CcWkjtQ0&%Z7RonULIC%li*kV<XtEVmH
zlMc{hwp_a1HH|A3Yn>gPWC+qSDRWRa-^m4BhC7l>_HM<p?B)os8a$2x&PwSFONlGC
z7NbWfgy;ROyfd(+cM(AcrQ@x#wQwbZoFd3)Y?8@s-byqV`=rEFTNMmvxwioc$KiH0
zDWt&W-8Eg*qeFszA%<V^sbB59YQzTOH-r}MZ6%L^eIgP=E56)MG;O%NZzxO4$cp0x
zzJ~~SYVm-S4UTuTl*7yt?MivttB5V*E4hprW~9F2h~^=@UK+890OQ`;y)~PR-()Dl
zIw%@p!j(0oxsvn3CcDHt%}&_TwDOC@zS?k7@r4~qyVMCChHmrBrG&$a(GLMIInJ{m
zIqMLZql~JK9KBOOfP*7bMLvmWawn^vx)^@-qrq7@-ma7eB}A~b6~~GvxOJ_s6xy3c
z!5mZ@QSop1XPZ$m1lG0{Pv|5SvUfWs4U2dp?sGGXYFI@=(!U+e>NHaLZdsWReB0-E
zmgEVpVfl9KbR8ux+R(Rsc^mNLRaI<-sVa~Dk$mT=sF<%?x`Kb(2SMy68;zgSoYiCB
zc4T^7&Sv9T%Oz*jlCGs*dgUM`183<AE<~;}9(W^KaQ#c;+qoBr+CPVA27|Ep>IAp-
z?TW~wK!js%e><@XUu|6;7pvw)Xd+PcQZ*C>Tn&Hl<qFW3kPDnxC5@(a0#NB3!TxT7
zjr@k&cCTue*x{g;!makJfe~-qR${KjBzd~k(ipRalYRj+{ZThcbKq&oB(oXgd}SHY
z!`8P0&+m$y&uog#+y1t1K5~dP;LpDuO6rK`q)C^uN;;h|BD9waDV9wJ*CuGgz{`OI
znBQ<c8q5+AF<rXX8#a_BFsad1{bfg!hu&o)`JUShC|j%Axdq7`y{jk1>yV@R*vmNx
zc8WCCiU=rQ8G<)tX<5L^IYF~oFOYDg5MrD_g<RfV_Q8raab?c}Qo>$f8L!Jd=w-(}
z()My)e>rgjojmz!QLrTBZ?|^U--~#pisfMQzwF2?)7hoyM%M!J*2e=~?wztu6ykPk
z-7=aq>Lf)Zv#jjVz5L?jCO9CCmdj~B=18e%IYCxlPSjSGdJl%X77&s)eEsD-c6Ik8
zp?w0Z^Gx!wbEvmAOJk27;I6w>8q=A9@E0Xp%msT2m3uCZtg+9Sy^@U|ZnzKVD?Y35
z_OSGWulVxLb6I;Ow-SljreWEKUnz7dLI=uzY~&T6Z$iSs*S!*mgp3JNGn!Ax^B{y{
zQCCz8RxG?F$Fqc{scZFYT?`}U*RW2U^ie_r7<?cn1G3jVX^mG3B1wfAeg?s2HDf6B
z!%eSM0esz0V{)BYyL3wY;0v~xB|L-&I7S>9TPlD!-BeI!f^23jl2B8kQ`+D*dda$3
zUL~cCgdeb8nc(S`1V8<lB@(2t=viE}kYE&4!ucn@V}QBG)=O7x5}bbStbxl{HsdRa
zc;z#*m2Jsy?L|&Mr3l-O64Nj@lGs!wpsZ{|#SFJqNAw1vVh2`ZFaLuL#;*g!Q(aBu
zh+Q_k8mk^xexxR>)yQgfZpf@6OZen40fKup-$^Vg-?bay>)2{C8E-)VpWivLahi4#
zW%bqE4pV#5(SM?Vg53lX^vSH`>|)aGMhvEfC4RGfW==)$h%0QZRYz^2z#=qi-vJ7m
z&4~!dwphK3#Dbc_r2zDmXIYx%@*S#=jL+lTcH;XJ2k;p0=Jzg`AsrxTIjvVs&kZK4
zznaK;htXiEq;Xiua&qTAV{2JXQ+5XE0v?rHuO<#x(d^F^)wH8K!)D9@41sNBlTZi-
zw;hEeI($0o?_>!%Q`|DUovRDCtOzluC8WRYNHGQbciUI&o_p^3GaI^{tCN9>pa-`d
z3$)kJJdWIUtl}K9TKO>Dn%-@r*@cuar&%?nr5!7c$;HMKmyAap#;r4^o=?bGIEN>0
zJ0bel{AZID&=41pM|Loc=>Ex9{gpLu`<gRq*wI#&75$YDKY^vKo$n$rqy~*Fr-JeV
zhl$Q6;)2{Rhb}e}{?2_mkOM78Zv>#TPV_drXOCh5WGiuv4vVs0jztC{U_;e6z8=0=
zuq?iv2Ne$baAw@~$^;U9hTA~Zz%40hvVq`hql?Yk#dl*`>3no*ZIv11s&K}(V<F)S
z7i_0$z!Q0Kxj%3Aoj?Qe4G>gj^q38N$Cno&v7}pr@uaH<GdNVri`wxHX^IU;3G@JS
zTn$IQ<EZl%B^<3)r7!yJ17c;P-^nq9wR?!%Ze!mmlU>FC*Ct?T#=n#2kYqntmejwK
zb}pRjn7;4I45JC4a%i2f$?x3$-~PJ>E7m9j;;M!pd4S7l2+AZRDaP=q`I0T4wRTnK
zC_<dY(jy`}%(TAa_)pd_TaFK_od;xg-N|BIg!#`~_d7w5JV$OX>L?>;KT1t2%h9y0
zl%}ae%+8#@;K_n^tpDkBIc*7b=|<@fj{mf6W?f<QSrJ-ra4Wi<rlalr9Epi_H$Q4S
zRRbrPefw_i^0$$|zy`kSqx`ei{H+asH_7?E_u5btf<9hkGKasLq~e{f6yO*d`EH&m
zvj-HuxK`Q(Q;8Gv-CSRIgWaIoSTrnV-^NQ*ZbzuIwEo>dRd<$x;3}B@L}~tj2L_wW
zB4vx5gQzvCYAmGY0dB0BQ0rJcZLRP6$|!&sTD$TIf_~B#v$D&@=u@Etm%dx@Rm4Hz
zRwPhVsG#LCo*6k0A}Gj`LD;1w*NJ$QF1uB`kRZ05gyOvdsGJ3p!1w>81wYw*DSNh(
za1wzO<XX$RF7Qd&-}-l*nm<BR$1>|P@?_b}n#Wg_PJXK_=Hqj=T|sqGf;P8^+R4(G
z5g@67K>d%tyWTBZhk><R5#8M=FD=X9n)gZq-Jvx{6ORWmyq0LN%RN>`vSMW65rw{F
zwP>12z5=H;nokN(cgDu@3CC2n-+0yz5mKB2_2>*BuG&PF7EOE3ChsauFinof<JPE3
zoP#ZDM%~>oyVlCm9_zvB;Sp=ccYJ~W+7{Q6;6|ne`57-oW3DX3y(pVzB|yH;T4MMt
z(pH-9#_as%Us`W1&#HO2v3~X%(j(j^*-I;HxtD_2u>sZs4uZKNI#{c&zuf5AXdBst
z@Jlv{iW#_B3X!1JO$)ZF(ZRB@XJxFm9UwjU4%;apy#=1zzE^hJ6_>AI$7uuK%Z-0u
z%?7`hm@o}jZ$nv{rMykVClMtZj?zltZX@w~O1HCGmW7}616P5!Mx!-9P`ERcw%u8E
z{CkNz%7%&_|9d`hGqclpQ%qz%-HXuaeJ^KQ)_D-%MifOvi{P|b1_Z;SaQ;fT(<%e{
zzlao;-!Y2ftn$CM_&rDF2i|T=70~CL;TCtcT$-0rT_+2O+-mP|WwvfKuZWA*%kRkG
zQHzL?J!}0k&N>1#Tgegx4FLernwz#3V~3;nOTc>ch=K!NVWW&zni_6ln|C148bBWj
z+g#?A<PS$}`+KRnjQCX^_1|_Xz%a*P+_JE+@B88crns<y?<cA@5rGeW-*JE|6ZTr)
zPv!wpql}l%!&eS}-_HqIy^ef87mhn3TzV}V)rLK5qgmGtw5fX&mV4~`xqdSd4|qJw
zKtCnjbJn?D`b^=Lj94eW@5}U_ciUus)hdvF!|zy)Qp01x2<fj_Ge)u6dDW||Rfe|*
zAlp^pqY{95xShp(Cl$KvVQ^lwrRp82I(;V*IHZghA%Twnfa;%LW8ECNBEt8r_x(6O
zg3A!3isq*D2>om{qBdf`#MoByo%Xftw58)*3+U%(iHo_O-{rKU{DzQJ+o)z5-+bFF
zt%BH+ghDoLE8dAyuFUne6RvVv2`-?6mDq`ONmV~W;~&HtV<bE9gCqlor-&6a7$0`c
zwQT4Iz9#n*D<X<W%z-?6<Ohi}9{q~(Aj_@gJI#eebuLD#-YM=;%*9x=<ZK^tBjSk>
zJ&(ko{)5CFBFyeZyg!lk!_ZiotaCP5wHa~g2+?JYJMWOM#5y$NU5NwWq_dSn;{a`E
z$6ajp&>ZBG#89VgsQ^Vlfg85`gS(enL_rCz&VlNk?A8GT&q$nO-I%g@LPh={3GMsl
zte+(#nVPQJnCJ#D_aPNuD-D5L@=KAO;(7w1u8<`BAdf7!qxHdTmP+U_b>kHCvh{;J
zQKryp>t45AUd$~V^22NA2YDhXJSE<3`^<jmi~Mli27Z{|;Oe2%GVT0fEVPV9)rZh-
z=!ZTockLSfVbB<%*i{?J(t!dJy?3pa@1(Z81i?82#U9OCCkRnN7UV1uH_UKD9RFcb
zt9Oq-X!Re)0z@o*H%wrTq?VhAhH#44p)&cy9K3G_h;bwPImt18&YEQ`Kq7k|j7qFk
z`uWcYoZ8j!vG|*l8{6WY;X!$q;%T+x!6BC8$!{>z)~SX`el9RjqS<cNR-~+}*2@yH
zGsAGd|HHyrD1`*4M|nnKEeX+D)^WWY`v<CP>jlagJobr(fOIiSMF1tKyH46>X=z>v
zL$uhc_j8^N^fcseJKk~SQ}PUiHgNAd70k-(Y(Gl)c5$;op)~NL!cWL?Tg!aj27i<X
z1P0JjD*h-iLLxx?le2<QT4W2GJR8XpOUQ+ntd=Eow|~l;<XHJpEcjjI5TYnG#@LUX
zXerLvcqs(+kyaeh4d^8+)OrQNhoMUo<p~}i>QTDW3Q<ez)QA#XM`w2<ch2%kU3A1+
zSso)eg5ZVLs$EL3Te)3i32_6bXi=(%rK(n!GY`)a-7B(|!VtV@orL~RMg7tUcC(x$
zcBifPqui28x9nT8pEbq&D7vuQ{D(lQ^q}NmwpMC#FMAxem7s+AU2%o3+eTS8Qlx|Y
zn?Ld$N6aqn1Y22Bfz)r<cGd1>g(=CgNu;kFwt~`roCnD6DO8vP@x*N1;FdJ_<HBE@
zmmw-A{v{jAcM1UGet7T_!#~a~kG#r8%EK<QS=Fl23UnFF&FGJv2+i$PB<cJvc28M-
z#MX`fI8aFTnX-`z+hS>4>RH<DD44&~jA^2@0IwkdiY%Kat5HaO+$aFOg@8!SA3MeT
zvwN)d<GU!P2j6d(H+zWX<aBzmGz1wSwR+VR*K^@JH`(PfeF#LRtdk{lp0H;;=tB3$
z<?4kR1Lu!yQua}|tRHi++Ichfgi){<Tdq}oUt~M4=aW2qiBS6E+}s=uy4G)IzsD8D
zM(jT<{Z{FB|H3>X*zKyhBS?gI;t9@NcAEAR-<k*EzXpEdi^Ca&#lcU?@{1RX2v0*l
zx&3BcK>v&Ar>mX+xBuC2X&pv63=xQEdDlB4GC9K40;1K<y|1&;pCoEPz7*m>>2m^U
z#!9LBl#Tx+XJ>i_`xOKiIXnUjOk@M1<e6U~8<?!RzDcNFkKbjDpOjOH#l=hCn$dEV
z?W0wCn#bfGbZ!?Y?j5zopX9nX8wz5#RJD(kEc;ykN$$Nva?t2x2`9A-UWr+{KZyf=
zm-V8#gGgdz?%9tIvQmp*57lk0G`C+7>es6d<#+3tVH-d3jXHa1fxGZ#Dd3({@VDF<
zx5_)+UJ}afXcly4)@&!Av}1{5w(jV1aH{!)4Xg*cNX{bpiMZ#%_)c^z%+=Q8c%c;(
zK*L#Dbic%m8(DYo4$tnv2(6aIyC^C+SJrb)J;LHX8(Vjjeb+nhxAEwMoY6!*DZimT
z6H(;JSKVWi*_k`wuyZ|cdj+msiR0LBuIHNVnN`S#R_WBE8)LGvUKlXh!YMh@Vi^<4
z)N{6Ur%BOEjMs1t`;&17N_W<MRp0T5bt}Nma3aKYr?*~^{HlVt#+G)cg?^0a!FRsf
zR?0x{-)n2p+&%wl>ji%7D6sI|i1~br8?J7~0~$~S9$Nu7GLt!wX}gSbo@*{10z1**
zd;h}hr#T^q31mT}lb`x3>^?XfN!bQ}>I=p<D9q4L9T!$R5AU+!pO&?jmD5OkS<v}Z
z`)ST0;r=5&$mmZCXQDW9E@QK?pT-=B$ST||VZ1ntJz8~uu-bt+a;aqWM05u^8BAoq
zoPH8Lm2b^C_^3vVg^YIY4A3kfl78#FwGs$j`#t44wutQtFe3ZXy_{?pW5f}JbLpqP
zjt>zBp(^T0)K}phtP@3volEE40?NEX5cjfp1g_Q=z_7ePpm^SnquevfV3jHt7H_Sz
zf+HbO%YW*7U~12tZRC8#oi@(QxZ7>U+|LmX!?p^P>8g1eUSd1niOsuAw>oEc3Md|s
zOPOkkarCoTCDOm*CaYWBO&j=`W5?Wlb5<64gFo{{PpiS9tiX}X=@C?VINyo=!~%hs
zftPILXNfjkd1|E&mR&5nNP~?=tAKW$r?D)dx;i109&<dvP)Ip409pNKzAC$BY~p9h
zmngU3<j<0rI|V_U*Z31)bXR-Pn$elt`0Tb;X_z!bm;$s5bk2avo2YM#RU`6<4W3}B
zddKA*ydjl@EypO)Ga;7#S<<pS9&_EgIYNcPxM{rt!UG=P_GfoUjttv(+e-O9kI$y9
z{mgNItAedZ7gMviDQ#4sD(RA@`=)JXH&PW`v#pq=Kl@wT{#op`Fq%-=qHNpARsv1=
z0*`Ay_u0h|e;|I%z$pAY&oK`IH}vzuHpqcTHy>h0loBvnkK0H*F*P38ET41_=-X&M
zVHOB-jQu>PiXE)@Y2)Q9J_Hx6{`1@a`+ry0wA!(Ws4<NcRF0z+Y_hBs1JScazLVlt
z@kn}bzRjQKogMQPWJ$9#GuFytsbLgbYVDuL)QM_@Ls3aox|pSPOiF|4>BX~jY|`j1
zM77Hm2+9~3+N93UeP7JVxwM<Z=IRMk(EE83W!()BP{caC=Uujv%}6jWGauQBto_{g
zfIHIIdKm-!>`nHUZuA@3w)WSHwi#`|;x0S*4oBN{{7g&RY-bteBbBzD3TzViK&7uh
zNbMK7Vh5+?FKq+ANZgoFLaxDIBo(1EIi-*a)`ouJ81?2SZ1@*G<U<X@RE=awPh6`h
zkw7Ba;+(t!29K7WkS_0msE(DM_QJi|czN1A2X9i(){YY<wD-y>G>e!tWo!6c1I1+Y
zz1n&5udMNloM6~Kg%Y?6XwTxnU7+=gL=fzWoFjug{35B1Yz`NHk<{k!!3Zs}RN$G~
zjXUvjfd_Ff?tCt<?-U5InC>QQWQ>9;^|Ey2s*js*zm(oTYb&Ma-LP3}r6Z|mZM_Ud
zGSS_(ku4P;T+b9;!6MhfV2-w8O&)%`ZRc0_zXwu@zjjI&gr9ue?3YQv>qr@X>8tI8
znQZgFEW5~sQu5pyio%U4BJB_VG6{v-amCRy@=HgCfB6fmWmDKKp!yh%dPI%Go%ENv
zJg9s*TE|QOQ%$R9r7pKq+zX?zSHH$4vn$OVwvKLq(kS&Fc#AcEnPhebS8z`=()wjN
zJFvQKXS15N<F*)22w&u9wN#!6@It<@>`3`d<o<e{J7Om~+J}_Fx&b@qrL*z9EMCGy
zNz2pEcOn`hDw5Ey6lm}wvbCxb*JTRG)_FY&aEC|sx=|(wH(|EIW_g5bnSI>0ewjNI
zt-fgD_AmVifk(HUs%xFk;+CH?`;|`w?Xr9WzseQ$Q1uE1V}rjcmX8E}2O5zcr(jO?
zkU9R!n<oIK4gab-Cz!aAXc!)p+e+<MNf;yopr*uAqop`rSY(L)RoRrJXhiy4D)_x$
z>w3)U(IOE$!LEr?UZxz_EN1zjClUCkQFWlPHGh?O3_sV3%5m9ZYnOtDr^vaw7*CA#
z3e4?NHq7yf&vwL`TP{OWP?=5^a60o!B);hu=-qDm6b;?7-mj8m99%fyfk*mT2rdWV
z$Z;gN#I^lZB5Jeo1lM}FuC?e3jAHYz#~5Z&+S^7<JoL#nV>lu3BnDb(WKID`rM=w#
zRpLYh9JifnP#_nGAkN!<9f&ZykIUo0uYIzhy9aFW*Rgh51S!)!z3i`ZsbSI|wc&W8
zp$p>ps`4^@ii>A08doBnIU6kxnE%>Xmd@<}0Vj<4uY<*#vih&f72R*(Ec>s0B$Wtf
zvJ{Ab(R-{>UcfDS-pyw8#+0!?TTva$W101`^1n`mN0xxSl#KBgb1-Z?Ou&|YT^1zl
zD{(gbwIlqaTKY1OS<t4N13ATQMj5*L%xSLxLu5u`N)st2=9SQ%2W_RACoG9r{~ItX
zYt_5+t`?TSY&|+8fP`)QIyVVNB0)em3#c!n)3&VwO#_#J?NXAUY#c~C<=ZX=f!PQu
zu2}FFY+%Ee8kb&j2yf(4!_~<RQGDf4zEgmvr%_o-lQtZ66eQS2^1Gz0V%aYsp{s2;
zE_noU8;!!)Y|PkLsrc|)Z9EE*7>&uOmk;;vvWfV>=?wju+$e_(EFxlHCb5z5qv_?!
z(A>xsnwQu^viVj(#2qs7s)`D>OQ}8BXzW&Wi^UCJM{(Ip8+j#nk)#wDu5JaFmje_s
z%92vFv*BpjB=%iaV1|54oQG$MD^RbD<DEo*2{;)brs}p5{pmu!wsB`WDiY&*H2lH$
z+D1;MLeo5r)WQ`8`A)P~6RenKTV>KAk67QaGmqGIzSE7Q#97~oPTz({gZ(B3Nh5sb
z0#@mK2YwT)E`FV$2dhU};lbbd?7&sq&~I|DA0SMh4gV&|=!5to#upIf7I<3wO(Lfv
zvg2SH1#v=sqmBJ02TY_1L>D$*iXNVpV?#Xw@N<FIB)^%6>V!)M2Lc!!z>WAhwHI|j
z_O4T#9HFHPutwVPb`(U{W{c(1JVb(RyHt9IgW_OV{!P*tj7YfWvd#&PB+znW5XN@{
zoPYiR{9`XW?t(5Ct$!ErAr(4pE5AwZ;vd;mYgs@Cgy{SYZWilhPWKR4-ZlyZaB(h6
z+blyv_j<{;N`G93pSSIRWH(nH+bPYw88+1Hw>itAb@;dq{MMm>OF%$5_}f@y{1sg6
zgvFo@{Wb`Ju4XUU@Ne^?2sT}YmdpG~fj6M&QP+*k6(v?KF16Ag+?}jXufcMUmUmpd
z6q@`R$KF^~cM5a7Y4{8d_uo2l5$xCMzx6%!DCvlPo9M$5Ad8(WZPC7BjVsejqZAR>
zjho?b9TlhCB>lHZ1)R>yP_td$Oow$2OnVGhSvJJjMbH2IFSeZR>?K!xC(20khLujY
zbO`YUr+QK4Z8)682;_CSl7Pk~G1%O-d?&Dp_R5kItGcwjA7@9~$f;*jL&G-9x3`Fi
z+zNOailDz;M!~r7w$z&cHVNXRWYLV^ME1MHV6?+hPujrmVzJ$<n*Ir<XM?}<nSSIE
z8~R-=y8^g(ZTNS-7_&GcjO2$d;`5tU%SIvtM3{_@MhSu|$UklDcLigl57F)|CJp0R
z7tOKMHi$}5@BA(om=voZ<U|&UB+JtI5bLoNA)_S*evRM7F(S;dW&(E{Q}J`wDo<D*
za8dtVV!i@Ha+)muF6XGypW0GXm?gX+$H8*6tQ=5}Tc;Xm<g=G!U$?yDE*Xcx5$hG`
z6qX;)<1A959>VR|O1_hUScI{}%CxMGmQX3$*GoGD@|{AMQ-%WXL$JqK4)jyD6^(I@
z>0Bc&6U1u!uEx-c;+a%l;3WuJbGrY&kn3skMh*PlSN+d1D~<P@Q+*iCq2DKk+6x53
zzt7FHYXR<0^E2{$M~XQxwH763$qb7(+V7K!C=N0r0jV>`N{92(#*AkPnY_vxqtC3=
ze_!xK?v-45{0<@BW+M8TC5Y@~Ksq;Vjcfweku%EKV{86CSJ)**xExi0+(5@Hqai|s
zWH+%OfsUc1Tg=k3PSTZ%6y3J;``~+!7u$05{qA>JCrid%+~Z$Zx9Urlqk`%w>y>wY
zA&RID{py{pPJ<s<`F(E9xOsBB%R(}+Wd^za9kcb{=g#=YIYs*32*~F4A(Y(A;vv74
z^c<Vgs)_S{&`YzkovSPM;JocrB`}Q#k{is9{lOs#VyMf+z#qzqfMKcD5)sVc9~@ow
zP1#U>b=MY71LbfY{zIbHt~t1-k$8g6D_>=;l=qvlOhK?FDeoAi)Nxmkl#czur=6Sr
z1sg9Ah^OHUk+X9B54q69WKeE{i7cZG*eATq3pSZ;Jc|QqBcTxdI3)iegUR|m)2(7y
zZq7HSPS0EG4~h2g;)rSgAvq#1>{1%$N+|^wv(vbg1nFC<fR!g<D3q3~cap<0KzQFy
z1ss`7s_#A4A<l+ge8&WcW8@EoBu*)R@=AG}@}P8Uv0?;xlO;cl@A@B-tl_dLilAFy
z_8^t?6`(o;^qXaw9#E{Ktyt2(dbMq5StSj7DU*inL|M0A#q#WrzU`mjU>=B{3-G;z
ze+)!%A5!1G(2sbhsBE~zMQE?#KgPL&Oc+`|^2Z$8jLv<l{jsp=<WlBhMo76o=1Nbu
zJT%@|d6qDfO*vkkp_?;!ztj^9Ha5I*XjLZu=<BJti%^g|1<=p^Kb*a9aNl=z=X;Vk
zL?lk4fI~!LKuLYKg6mQNF4cD3u1uU%um&{Py=bQ{m2AnDiEL>tIgzR9sIBje=8pHS
zcg7ENXZ+xYW5;r$Bub(rO5!9MqImSc5A?tf^uV(SFd)Ey00ROH+}Hba&iUHe&fM+X
zw0QLUd_RBA=im99|3~-#Y3ub)QjBK<)m%LG=@jT`xPp5M(t0Fo$ZU)AAo=#_|0stR
z-|Jm=G-^M=)AoXmRb!HmqO1)eETV?g7t=PL_0g;#{{J7<%S31gbXvS4-Z=rDcH2y>
zIu~&6f0V0;^lCoFa0;Vne{8`UZR;i9^rHX0M>$cKd_nMbhjL#|u#uOnL??u*V)fi!
z%)SvDTLMdhk>Dcgz7YtEc)(6LexvMTVUIa*4c~|xw3+9i@f*J5XJ!=uw<&%pnhD-(
z*^x{wbW~fuQ5Nxla8GND>~6PyG1`2f*&1Oc5QDX6`;sS-b?Wf%{ku^|_a3y)1Y`%_
zdcAK>#hW&eA3lv}WY|Ao8HS;Gz=x|1G{fjO%KEmGoH7R<96E@sYAmbGDuH$_!6-53
z8EQO04@na|0KNs@7dqwXYDvQm3mcr7Z^Y4uqs6k1ZJw)YdHzHNs+sp;AxREkEwJzn
z$4!BOI7wpt^q4N@^tPfDTB?ANB2|>y<F@P-(7$pXR=(lX-OetvMSs(km};@i{UJF1
z#ki$gu{G(nhQ+|}1fGJ|TlDiG+e_qyXvDr|P2Th?U$N%JxD=oMw6)~l523sFKETx{
zG1F>`KFa$79e)-&zC8!^oWUK7iHAKM$E(DY^4GI78Y`@aEUF|c?e(GFIx0t#_^!Jy
zn*?bQh8N?25>?(t7Gvqj1Wts%-Mu!t80WidhrIRiFG$co{z)72CE7E%(`r5ooK_@K
z7>|KYqoFwsYLS=AWR#V+4&daWJRNhFRuuhi{L-vX+FblnT#ILH-haVU$w_M=S7fk9
ziJ2DjC)qO4Z7w;8F3wGGiKo1heeA*y)9jmm89`XQqjlfR*S#bPDYi=W-;5pGiW72t
zFIk)daW#H3F!HA%R0%-Y^vzhxt)c_Hv0^e=i+_Lxm0~7a6(JX@nQxYJwjImt_6i7@
z82O%deABn&zowCqbvoc+SJ~P3MjK%f>;<&*Ht@}SONuxCaKa^-)yOx)YpnzXC@B;r
z-O1@e8It{qUN9<;GENIZ6DMBBd{+A<C9P$*kM<DdJw6=3Y;Lleh8*H0ZF#|_6X?BC
zT+SqbT?d9>6z5`&wf+GGm?yaAd<755U4?$>VxfZPH3B60p<0YhJcUi<v@LnJk;=iA
zD>%U7OiHP3R;o_U!q;o>aV`5+;`MM6rmXH;v4r?m;<8`=t#TAlT5##V6<0#e{3RXb
zw_@Knk?^O<J1aW#VuQ%)Hh(M71o6R%GvvSI7lo9*HGbh_!wlMdoB^eeXb-SF(#07(
zp3&iaD_Db~ot4dEm8&eF+IoORgkkKBt_tg)5PpgiFzi3vEnwu@+za#Yh73Woj2BL5
zL~a5{zg1A+CgeZgdb=Jq^GE)t|Ew0ErS<c*oCo8%mY-vDdu=j-=^|Y1bb^h};efpE
zW^#3k8OU6KB4Ss;mL%$$&((aaX9t+yr%6yv;)_LFLNw?!_&E5Df)keWhrDA}a)M|Y
zi;ho<oo~k`ij%JUw&Mme9KbBpe>-u5XMU2Du_0=RJA2+5za6J#+x`R4kWJq%SGpHA
zyxHp^O2bIVTB@ENhaW83A#441kT_V1fGlLJ-j>~w>Ks(k{_S%3_B={@rw;EAfy;g{
zT4(kLLRVTV*?;tYJC8_}3@lK|K-QM8E%!lgmBZhT!z^xF8tRc8;4{K5qq!<_qBt6J
zK7&Q&OE#7*(PowuzqM?+zX#i+@oxumRCFZrOigA_P22z0rUMdTQwjf{gL0UOGQlP2
zh^}RG-pKFZe6C6tP8GNj`XVF0v_)Sa<Y0K7``^TtCGtc7Eb<Loy=wXrx0Os_W|zvY
zBcNcN>n@dT`)bp6tG|?64i@Iz@m5+xRrUf7<g$sJv_@~T2l~6I`dorRz@rh1<WejG
zOCYgZi&s|!kJhTX0>UUP%7nE=ncipog%BnZk0Q$Ca3PZAO)QN%9BOohpyAG_AWIB4
zb09SIB|pDHi)U;gex1g)iye0ul&^Lq`<785#NczR(x_Kf(hvxyqgCg!&UTxIvFJ`J
zzGtkKLjl<nHh!rbRV*$E4jgSEgs^rNr~Qge$9IK6n2GN}t-uK5IqKh;8E-)KT~2P!
z?k-#^Xy~(>Y%$^7ssK?-{;LGGW#4nz2^S^$aph8wH9WbH)-0NRCut}tSPOT`j!NHL
z_npAi(9O7;e<v=d)TU^pK;{kElv^NA68uE~j`(KQ_?<F&BE>-qd?)VJfkB+<^#)|W
z-O@U@cvYnRpS0HRlsRqY-g})c7rM3kFIEmsgLQl-@YClZcsjlLY>mL)z2EVp>X9E@
zSTj&nhFBW@PRu^uMbvotS-H4JqrJ#w5fINog&zHm6AACIYqqgm4A`hJYpZ2Xdoec{
zuX<91h2kD!lhq{N<n^7dYMc!*)C)EfWs&q6+K-^{QC1B6*%&dJ^WQ1AQz^a|z7y99
zZdwYV#i;Jyd%8Unw&c}!3->3Go3W2w@gE44IimL6yyj9}xZt8L$^_3V9C<ko)PL8H
zi++i*K+0dP;wfwRZh`8{{n+b#H@4jUt-%OY@A7NX0e;S!zZ;wJ<pbTji&cF~mJrXa
z3yNXCwPr^TAK<-&5@?H#N`A${sUseaGJ+m)cV#hwtMPZ^jBLgbrSrRv14JkhCj7gx
zWJSr1GT^)A(EXSHV#8jq5G>esB+SJj&V_fQ(G&PZzJ}k8E1^Uz&_lf0W6?;AsovAG
z$nR_4jVE!(e&u)__wlxFw#o07TQXbwWV22C>J1>hok=j-L^01*V7MKoZ0+U)4A?fP
z(*<vew8R!;cmB2fYAL>G9=NYu=N!MBbHZ{G0ibvGB{Bxt_X2Kwub5ih_ws}g&;c}E
zoXWGxPK8{xhVS|FJV>mg#_#z-;S3@+D9S7LMxxuM=*SHt_?IWIIa(r}vX=NRPwh9L
z4NtJtt=}t~ggbEC_sY)p7tf6L@0Fe9<&)D$hj;SU|BXvk=l9|W?kkxgaU4tVAq^nK
z%gPNzSq8;cq_g2DJ5wGJPZ2xvy>hhmZjJh%A0r0dQ6CN~*IpY7_(6_V%aOP??ooBZ
z#=n=+o1#UQRAVyNj?l*J^K`V5_*=>lkybM~34*dfTOb4W$%{H7E|E`Oe%xU8g;;No
z!cb67toNeV!w#OT;I@==Ml7mzI1PQc?A<D?8YE#<c!E!|_2FM5c*#vrx;p^7ZYi!0
z3P*xWS^bh<E>Y!3_m+a>f}T{~68YZvt~W0E@sZFM%Mry=u%@MQUZB7GtU12Rs#xB3
z$~NB;Wx7qW&N>C+Y>n}-li(F?iyEh97233z@mu>+taY(%ESROE0upJkb><Yh_8k=c
z)VmZ1y>Kx$kbMH9Uc}c?K@G>einEK46Z0C0+C`D6^+bX->cin#E7P{4KFitK{eNj=
z`S-{F+-ghZOn-i}jeF@^teH(#qfZxq5S_{C1dtg)*v&*0m#|AO`-08+4jmwJ`Mm!*
zTl){TP%YMT5&(HOT{|P_Ktf)kyn^!4Z7k>LusWwf`H#Z{t>khcXT+h<?EBY|a=bxx
z-!F$_w)P*b{`>I)!|9&k5G&Gb!}kLf6MoPd^M?X&HGMxeN^bT=Rb|cJ_tS)3KLlFK
z_sjVh)cJ0Wu0<osV@zHoZBbUpeZ2;3ZhQ8Ci}*9vk*#p3mHtMaVVw#8gXjv8<MjF{
zv$ZW<HgK(ekqZf54g2>l34uoZZ(d9|f*idlyrp3w28oYOvTnr|82i5O@Pk$hsHEC{
z(Z+KWmPzhPxF6eOlu5dvm>tsCPhTq&oi0N%o2f=6XN+dVI?a7Q4)7Oy32bTeRrk^*
zx*}`g``7C?JO?e;)Ag<p35!*?(%wq|O+uM0gZt%`hb$%QfN!8|SCJ8ZKalLEy;u-R
zQuu?|Yo+w4`$0hU*}_xQ|DbGbgr#rzLA)k~f{^9R<JS0ta*!THGTrn8$0Nx?%h%8G
zNb?W!^kIR`JJaGrB7tDRTccMkvz*2WL27L|(ya1JxT8(`56Z#nKLD@Okzk^#7eN`-
zi@=ffRpO0a2fbirn23ZxR)E-j4OcBraK>ChGZHNn7xqaTtw6w9w)FCp9gQ-HH|4|f
z2KaC+d+y%{PVceW4<asvthOKxq|LSQ=$a>$eVg=Ov&6K9Tc3_^lDiV3pc>-M;$*4A
zJeOTw%MF1SX;j1mpcn<c4y&|~9h09NN5ZP(6L1Dg@k`Mm5F1BpiI9|(TolSmlu1sZ
zY{MtyDe^<#@?vEhw#{`>mb!eYh3l(&I7onl6rH#s%7pB66XZDD7-i+~a5}`ZH2ts~
zGLdjtSTt8XDOO4$W3A<fac0GfYt34#uE{=FkjRtP7G*BAm5OZ2+OxCb!zt4GgmwI|
zoTIKsxD1`~eH<ToQO42%^!~7%=zavs0}dHbJP*w_>>m#7N16D;xF%cSO0Xi#VTu-s
znM(ecvl^0F<yiD|8gBBW)uMh$Im0SQl0E*zxFb*vLtwbcAI7!(98%ZmAI8-Z1;+uG
zPThx@sX%-hIoDhT$b>ZZMF=(j!-CKrLe=bD65<$yv*@!SCxI<_IR_Qqa`}fr;tv$}
zL}s^=(9)gI%o`j@yBw$hiwUK=6PByHoIuKSO}_HVut$IxtG^ukM^LCiy@n0ZTw;uL
z25ZbNwS3Hu$Zpnjx$MP<ar<k&9H&GXA*|(cnbd(j)*8*6yYL{SZ8@Hz#$w7~?OwfU
zoppFO;*8kq&dX(yNzIKlP;UYN4J%p2Q^+a?5>S4<RL=ENHtd~!_%Cgw`YMo5ae>Ee
z)CVA?lU~51Q9~MiPZmCDV^Jm{k&Gq?m20)jai9pe3xPJClP%oGWHpRTR@||0I?6b+
zMZ_b^t(nW$zZc^6cyyJ9E-v_74x)fzJ8Zt{LO%HT!*kykVtHf`Hg1cT%T>Yhv(J{Y
z@eb*=Y}r3`i_2JvjTb`1wzWhkEO43q$d5_6c|aj2mbxf&6}%ozED@@D#Y`UC0eI7@
zY`RD)WVYP+qXJMd2y6OL94c5eo_R3GkDLxUg`fy0gX$Nj(^|8#5~j)0pe=gWErJ1s
zz$H7WH^Kc7`G9r!Sdi*H*7>7++C<<Lk~N;TUav*wi(VUuAx~o;2MLlReElWl2`+~o
ziLzp5$Vw@~*65GQQEn$LQharU>4s!uSq7=j&F8+HvD%LUH4Kv4V>|{W-3D^t$((}1
zP1tlamQ9Eb4w+JIChAKnIBs(lKv=EEl;(dFXR&`5Op7fze4G1iv4X|PIlWCw*&P=I
zQZylh4YrkBbMj4Ng@>`ykIQZT*q=39-H&~*PQ$>SvictfHj$74v~^P3|Km7&Ten%`
zkINE$<R7i+$C13QIQ|vMt*%+~j{|8HjhYmFEzzYjr0yir>VqOhlSh1;4*&v0M(tk^
zo{LHA@Zok56#vJ8l+h?kF4kN1N-!8s%~LGsKuksP)K1y3xBbdN8_BkVT_ptfXtdSS
zr+AyvDjxlDEcI@zCa_HUacnGcRIK)6KXccK_hN&_y&BG_w3&?I&+vwmzAQV$K8zDN
z6Jsi@k4*nx5H;ZtZfmpo3ZiK#?0^drB3tONgw$s-R~M}~nhHFTmK*}f`+IHKxApT7
zKP%B%k_O>$*#+58a&tWahJxzU{Uq;G6;++_pMxi=|4E?vUBWC4(H1gV>y1w(a>}f+
zLzF;M_Fjk<cS38f%9NK5a)#U3@)Lgrw^-{>V$Hit=-IZPlr@KsdtsloCtSs*RTxwP
zA%$(~$WcFYfO&TQByf*N7NL{gYDlst#5N85q@3WM{h~Lsfs9#oVn?z8%%_yu#PzY!
zgzRD5J~32xkaz5`F^5rp(TFOdrCS&)iyDIKtkA|aoAhxxumr=4${yQTblEF5<Nbd6
zVVm<w_3tEOJ%;H&iAz+rS+Rf@qBT^aUI><-c-dKNA9{d7OE2Zp3@BvjvX8z~O6rwr
zl1~;gjX>%~6xACMOKidIWI48at8!)5FUPigrkMCPEXS!qaUj-e0_);NZ~CbRttq}1
zZh;*tVf3TSZS6#8A&l@LXnz9a*t(puzlwKlTaF8kh18QG7=4Jp`J8nmbRTd2*e9)X
zxttP&K2Y<$*$AX2vx|Wkg+MH5(}tJ*Y<v7cr8O3XD%Iaeb}6SO8_izGpv&`NM+1Jn
zm7Iw39FIj=(J_G-pGF2-b5IZsVU{|cKYE6C+GO^Rxdh=z0;)5t;F~t%6YJ}@xoRA}
zwZ&P@=5z7eaO+1o6-|k117EP1fZ~KGv!$iwa;_1>V4+%#zpvL;a{LFSs<)q(?d=*;
zzy+)OX{<e(I{G1axz+zPu${>M2CvN_dyQ>q{HgE0JHNZ^xhcz7QISm;wpsH}3n)9g
z>7m7YAht;J$3AYYIi3=FxJgo5bRgc{T_)bJcJDxzu-4&!&({9kzq8JtmeW{iqZ8Ck
z7`+bsG%i1^9m(`CPyK0J9BCo<?1PjXNmxipN%*R4K;A#63V)i<3dzXnaV#b;g5V{q
zIUM2xY}^MfQYo9vYOqbXBE+vVvhlK+n4F76U8pwa9Sx#VxA|)F61q9#wcwLX>%zsX
zf`kKye&~afwv-cLQ8(CfRRdq&&_S;JwA|;<v*ABW)Ga=^&VO-em8GTrXJw<_`fogu
z4f%zGPVA7M`Bp2E_B0n&Hm^<D(_>u+aXyF|qR3yNhTNZ8eilo<7qhhkdbC@81nj3t
zit@8UXdKv&%RvlAbY<S^j;gDt%Nde)O{Z5Do7MZXaw5qgcM#?;S|Y$?z~P_yG!Z9x
zIT=JtXtL<>)U1qN{GK2ES!}NYXM!Kb@^^PXN{1e+Mb9`iVH-5=<D+P{$>_{|^xYw7
zIx5S-8%a_Q$J$AYEBn<shnky<qHaE+Mi@I~3sqy@UR(4z?BM<O#oDzW^OfbFB@Kc|
z9-RHNAUlek>~&0p+0V-!_Q=$#?&oENd%Ak9{^!1d(ocvteHNOw;pe`PMIzGZ?PP6Y
zO;K4UQ^j|e_o(^jac+h%LbMkDh2~4y%Ub;x*Znj|tSws0vU{A9C3EU_9}34eP*X?M
zJt<_%Q8{6qKabt$-%(5idSe<ABMx|D)WyUgAI^#J5-4V<0v?aV_#z`obl|f-00(<C
zD!b@UG4`Ifu_!CF9iyYDx7yF+po((BumzNzU2-vSGx&KqSGy0|boMAhp2=y)X0kQ*
znf}<$d23Y6_=wtk^aT!>ecTpurb5(c9v0)PT)if3$>&;h8Md6=$h*N-et!L)OTVBX
z{dkU<T`A!G^}|+o#kUzT6%uxx3y8HS3!^whj%X9Elv6TW``eFL<CSts6pyayO5lIY
zB0UqPHD8H?hi8K}K04FzGAU4I-`4Ex)7{o~CD1M?vqVR0zixC1Z!~(xm3UK#lV_b#
z;abX40*FPK)DtqzdK|8Dpepk`bqdQi>?1=Kp0*MHTj30CG|Cl){%B4>0i~ecuat8P
zC83bRwdhp#*U!A=nNr8SMtP$C%qDY5ctj4cyU=#i(S-t6ylOMmucr~Q%|(Y=dbt@X
z4flN1kluX47NSPM?m`YM$YR#uX(2A9|KhywgmqkwZ^}G#(pKV^4*prQU-)EXm`5^$
zUnISIV1F^ss{chCYmW{r!-8MrlMNfB=NZh@_>0($2MOWS6cwZl(EF{By3H{PssZ3{
zi5e10N&ur+maS1k!3s`V+b`nCJc_LZ!N9!#1G^MFs3ZGA;_<9?#xGt%&QWi|kLNlT
ze7F<-qMXbR6M=pB7v+e58N-5V3ZxR#3N!Q3U*y%emNLwR+DCs82Ypk!jpan~V6<8c
zEen`eY&=26{uBf_>8-lRy>8P{Q4H@HoAF<`=3>ihbKVRMZ3%ERpA+C_7X{UVk0^sf
zE_wpa*2!u64N&k>lquxG^DO2`NqwsYR?I@t^QUaZSK_N*U1#>o*e>K@R`*Lk0Fa*a
zR{zUD@@V>$pa*rbHT*IVYcF=1oPfq?AxRuk-T0*^k&X6Wdxmh|NB$qZ<yBW4-qv5n
ziBgE}l1Q}emt|wNK4|U!*CQbK3JzIZJG<BU%QzIh2c^R3&7TPV&|VGvvg~yKUg)&^
zLI5lj4<pf+^p?t~hl`GlCKQ{}g!t$${QwknUhv*)g)M%@#<IJC5=Fe0lPac89)^Cr
zDk~y35im9x9hOrhib1zQ;UjJOm*qrZ5rjcO?!_+I6J$3qi;u_4G@q-#yRRfZUx+#q
zvMEJ}c4#s7K<PN2w56(syY4B;_Oh?+{rB5Se9ib=vg)^A`K}jhF%ZfGv{_x0iA_;b
zPsJy#k1~mKaqTIVp$)%^>+$h_Z;kmK0io2>vVd>;RXJaMa1^}r&A$o+1HD{AcC`3s
z$hkps#kKw_P9~Xotu1;J3M36*2`k<HtFp@jc)xeVXW>7xU==g+&Z>#hBFUx~|I%N@
z@!fim#7MXtI>0F7$WiRw5xc=geif%{M-S${HtKNkJbsPP*GKas0XMsL+SsoWSs)r?
zYw@UOx3C?vS^@>-qSRd6N*j+d7jF2ioy>^}S4{sZt^|Ib`-|3lCcD8pUk)g5a<%4i
zZX93*?!h(4=5zJ-AK+zOsA_nY`Ztj^Eavw7#b4S|1&9(LmAve>>^<y3+KY+OO3oTI
zf^*r59}7)Tfx%_$QMVG?jdZAfB~S+&ebO4DwkRdpnkn~iqu0iO?9Z&J`u;3(UVnX>
z{d@mT%phB$H;z89V5|QsPv|GCEl1sYz}o!_T3$)&I#$YoAb5s#dZi*R>W#s)+KTKC
z6>S46<#6@mB7mSK`yV6~ppC3tzoCLn8>(QnqZL%oLuVh2dSw%JDL}){3yiJ2-Hf;+
z&EIGLjj=VYRso9klfFv!(`w9VsLjc0-8_=I?3|~4);oKNK<cwD`PSzAZ@4c=hLt$>
zdk>aG$O}0T#=>&#mRsIp)*)P$7ISGO4kGdn52m{81Icb~#g`BD9casbo!D|pQ4)CS
zl{!yx=<0r5u=!WItp3;KctFD9^k@ygE(hnW!(iV={{rzVTCVBW@#aK?J58`T%0%MG
ziBNV(El~#dECvL}T9RC385rqt$|M48%QBDEE1MQ;|8*dit=zYc1ne^hc+NTl7)eO*
zJd9ro_`#9+bs~j5vI!slbvbOGBZ1tA|M^+GxkqC>w}AKTvPYw=LQcGtV>#=$N5I*&
zSV0#)7OEDEDdop_w3VRw&EnCAhnWxeAP@GmPsRP+XM9zO+r{gV^Zq*VbUB=2OXg#g
z>&#KW?-H*rR8Z?!$(;E$Tl{sL8O05OI9d92l7svEcj-Z2&XpVJl{HdK2&pqe*>Cb;
z$6#c)A|KZMrffNZ{4hoOjpGXG_+PVz_$6|Mq!+)5Vad@T8B)`4;yGed{T`mM=HJ8y
zV$EPJzlm+xRLqlFqc_p8=!bn&UjA-tuUd6MQ&b?FWtBS<1S0D_X}vxqXZ1h{uOso_
zKu!pj&aM83y$#ywBHHy|a&5!1XEb4Y5MT5zJDMNAD07stoFP&-vDLM#Tm+Oh?jXa2
zkPb260u^8nGo9VE6FSIdVqpqB#A#Q)lDViaq5m11|4lieAO43wvW1*`Ka!Hg><2y&
zMR5KqIVP5)SIATDvWQfegamNco?+I}ep_}A5x4we>ikzx<PTW=Zygo5v#Q<-IR!~z
z(eT^2*l;;|#gsd%F={(iB=JDRpR%UkmeYcw9TneigNwwMGSYao<e*5XnyvM>aU2l5
zSzC;wv84l*47S~?bERH~%jx)SoXrvqM5_Ew?|{H=h!c9>0hFw*6v$FS-htm<y8%0;
zsTlrkSzv;{*oXt%zaJaJQ6CQLE%|F6eMf)7iwc><!q`}VDm!zfL6)Ud?Y9MEKDf!o
zs|Azmw6u8`k$9O-+0I8I|LOE^<NWWHm6Xj~w={olb3U^@C27!nwKT=+zTkv~oD+0&
zi3_m!+qi!+%2UYEXe5^Q@seQ}mLePVhVpDB;X*IcyN_7(h&JqM5C>Zyv%0Ik`65(G
z`asyT`m1?~q2C&=mX-V@`4}3bZdwW|)nHSW6*f|9hy*Fi!Z4zQ8jUiUMPo;TaOyQ{
zz3LoAyS2qv$FefFLKb{Ar`tp57#Paz0m2HGd5{hn(XtNCQpJBQ>J&K2%{yxYQ6~CS
z$qXS}Y}jYmrKIXM5*4t!=llz~!+2Z`JSYasj^;4Cx(Yoy=3j}5N<>hrMO`tNq`#`b
zRH}MsyeBKLw7u>`Je^<_V>}r-owAuI6M-)$2hqiIzW#soM}K7V2}XAh0Vg0N6G-Mn
z4ASgSLF+h`MIC%KE+}kE36y`(mam3E6bO&psBUw;a`k$`s@O4xv6}rZkUVcTYID(P
zR`<JdB$SuC-v8YGcOSKesHuQN_~Xywlw^%j24x3tQ2w>1D3e7KTrZ?Eii6+za?2X~
zYu56+xHpKB2e4p|j}kh*W^K`2#-cFN<sh|J!#pQX01hE`MZa@A^%cYmoxcn8Kt7A&
z0^M76sZ|=NAQHhXYUgDe_8JmJqEj9DUAYmcSda>hdUfo@@Wa9QRX(amqYI)P6j_B#
z3da&EF@J$;Yrpe_76n;i+H&c9tHymL59}bH((lT>`0N&&PACZ3K!B56m)%8UK9~JN
zXr)eTJ|?DI&m*=FHFO;^p}=AVsmE0n`LUE8_w1AJ!LrY`hgTjQa@1F7+ouY8B*EtI
z{ZL3?pb!^t09@Vg%X)89kd^x1#|FwRbkG|76H)}E64A6gJu;wxS!w!xnQu2a4Sruv
z*1#YM^nUM00<FH}PSuD+aAJq`q_#!3iUHmJ`#@QnVI4c7GMq0jWzn@-XM%)Z7jIGT
z@8d)eq=q|<1S$*u+I@3uIK~(4Lz2Yb$ICQ$V29)(aGq63iIu^dbM*IliE!p}!Y=l9
z?Dt8qDG_3~)f{rdd4Ivieabs=o|*I^kumk|8L;VuxV)@3lPe&>YLJa{+3#bVOdhoP
zfLSJ&kdD81`y<ITBG@J9PYFQ?eTv3v(MR6LkzR6u`rz<vIR`?PfNY8P5UEtcGDs*4
z%$NUA_U7|DzAR;X-5>HaK%&6}kVx6;|Bx@ahtEME8MTyeTS>$uVQKtB+3v0(BH8_+
z93PGxWW^ui_$Zz@n#5=+(V46~PC%cu{vlSB=x{u)*5-BmPAYu7&cq*LkCCbODd=BE
z0?G>_shN&`XO2}O8(8ljd=J6B5a$xEpDt&9!Ul3gQnWuJ3gZvu#6b)`-fttmAXGhz
z2w*hE5iOzg3v%Z=n#~co6MVwPa$F7sba$;9SJK#`VNJY{6$cSS;SL>ryNn-215%@D
z?^}A+;vF^<ojHZX9m7ZpSl<2@o6F9KaRPNU?~4xe@`^3^-#Y!a7#)=EStvc61!LKg
z&w94j-Lu1%ugw}teI+2{eN_lE9yIh7cJ5ZO+clKxqG1MYm7Tk#l=KvD@;W=`Nf-Z<
z2)xyHZZ#j-<2NwRlPws=icwZ3z{<$7ZcUU)K`SNYgcKci?)K8LMk0DMj&g&xHY#eh
zq+TdSYj*BE8pQn&T$bj<=(?)*D>BzCN!**FjHjpAzbeSXdM_h33{i4-yC*<%)GLN|
zES8Ad4R-EkMnZMQ+*+#6u~K4&udy8)?c6FVJfiQdz7is6xp159+-+qxJ4HGc%c3?r
zCrgd5{-tg4W^CK5)*js!?OTu*+j1y&9!Fl5j68A`d_`9bwfB;3&oU@jA=Q@&;L#ai
z=+vCFZlA$y?K5~+_eOOw%UVoqTCFe2*qzFXY9PDCbxZFa@)sMjbMG$mDSM;q6T6^0
z;sJs(M!_LJA$UU-B7G!!lg*Swb-eI)I2Pc)BODox?;f&1aJ*_qe5pSCCwA1%$?N7n
zLj)YlCG8UNRD4>-@^}`zA(#$39=#RA)d<O4t00gwlJ3N~os2SRCA1bVL9mYJS%&rb
zm`x;H;6p`W6SA3%#U*45OygAci+)9?Sen}Dgh)B0+4g;Qw(5t7;>=%$<uln5O`>qA
zT-avw{9sUi#^$0%;pWOS`-+{<tEWg~L_)K3b!AJ2u&;#Myihd}xdM&P{ax@TJv_*`
zN!vwV%^p^BF{i={e%da@h8%?#Tk?K{B5+Q1HR<I%M_tGbmSdpe0L)ELq{1tCKBS{B
zIo@Na;Q_l^0VYkPvs-q<J1_B5Qol>QWvgzu9*2mAe#PpdOp-u3zD&ZntiGY#J%n&4
ztp0{_O3-52n&_BeM6tT=wb0THUg;|w-nH2uBVr&~X!ph}bA2iDlNW5=4F$V#KAwk$
zXu6?X;H^7teZt5h6EYnk-wW1!LqUxFJ6-j-;fA;{>`meRTC$Do0_jn%MjLM^2!?<|
zY))%b5bp7?e~;Vdtbn@O+M<HO<BAwp+|n&sfzW-{?n@%J#<t~$?tVlv4*Bi_w*7`c
z=#UByXy<jqDuO$7t#x~?66Y6HzDzl?yFNnR_4!Kf>Bd>;hI5_e5dK>wvH$<-*^sZU
zLS`vb%J2<w_7TnMAv~17F<%imIg;S^B6U9Ov;W9PZPWoLQnek4X-AIf@(v&M?VhbY
zEJKcC{);#*A{WQJ$G`5vW5SMSW6-da_qxBeD04T@V!Lf8<1j{2E~GGtB*(owg`wPM
z6M0s;dv<k6<vDr7xdv9X2qj<o*Z<m-w?;(3V?P~-^fZPw({?t&ka2i<$FNb()fp7t
z?6%pg!%>%FSZ1YjzBDLa?0kIo=r(RGHeZ!Zi0R|TJZ~3bV4dY>Y$10DHzd~UVtf=K
z_RqE$&Gqy=&xle*U-C(OzKGP9e0YLY+U2M#uOPkRQxFWxH^g%)+eUv^u6T1qiCtvn
z$-(#X>~!1J{E6I>(_uF{O7S$6&aSPxF|OVolu~s!#y!VK<4Ie6V~{xe=!^(%(Ha?2
zA&)UPwlz1#?Zz90aF(!0H&*X`@RYXpMn_Qp9c0y*T|#^!;9i$M_lb(KrWhWUm{)y$
zj0ds=T?!k#WX%afp@aM}+HhlBWyL*Zo)FqT;%sgEM%(Cr!vPW=-uqWvGkD?63A?Mt
zu>^_X6}0<#+fohxTnUFlWE^YH;R`7$74}oM?Z$$0KZcaBBR=y4f)2@?j0|9VHi15n
zZd@iaoemQ$NS}2l^bslz@+1cqwr1Y)WbE@6dv<WA2XfJN^ecksP_==?6B&*vNWT7r
z9lFu452qXZcO7G~bH&xCkRKyCVbK38b~wfzXKklI=3;+Fvvd6G>_{${Ld+<D&e3Qu
z58f#|<}fNPu#MdqcnRM}oxJ0|NFQFe-fG$97dW6NZ}jI+!rT)471<0s4H_BGwS4=Z
zfSTy{oSuoOEkuJ@iA^Z~a%(63%WUmW{-^(JDq%#&J!#YaYqrFVva{J=A#Fs#WgPpN
zgn#Fr{p9(x*<5qY8{cc6&Q2uV=Wi^J;itCRyw9J^WIcAlOaHfz+Ja9P3J9UW#Q;X$
zGbd~@fT;}<cEK*?y8P`&K4MEbUSEGdy2{JdA$nOnGYdqp&IEBsPcfOjk_#p(ifw)p
zb=6ANA!N?}J;Qc2#)rm4s$w@K3xkqlx2?J<?vX@nm~#>L^`;;}xDc##VRctWSuu|0
zJQEJW>Z>x@2)I+knwt`p2v0QlFN}+oXJPG4i6jo}m*q2ux-qLmji0r3Hx-nzzn|x)
zDLU7t%Rv?%&eq=)Xh)8FbYRUl6)dLs75i;Nv?x(OIQB0<X}4T!fjAz4-^Qrqc2q_8
zWv$+e7`j2*T=k-OUE<+hwYFTb0W9WwY|Bk$@3(EU_M6Iicl7~gY1^_fd^A9dP_YiL
z{YaN>&#$~Y?gM(-I(=;3+@2lQ9Tf^a&YVv}i}c0{onaf^u)eAWIado^jPl^7a}PKT
z@{R~4|C-XUAz$Bqa=O~E|4oYf9d@V!Oh`hS^=E9P8cC)x{+=Jsb->~p!NzF9t`y=&
z0!~GOMB?n9`U~lI%!fiYLUfGyP|_G5ujWS7d8_4KeSYUqzw{d?bGThdt?^>Z;lM*S
zaZ^FR3QIHT<$Lz)ZJ&z%i)HNeO^#*?YGFNb<PE9B**uC7R)Ll?G2yV5g)ex@W_|A-
z>ae+-9ytDMc0PWIogsIx|8f*$7xF&AUArmS!gb3lD<j1{xae?7(=uR-InljPKYQ$w
z_lL0x@fB<-M+R5EV3*_96ZqvnWy`(_3a5D``{2z-OS=;FPR%}!H27+M+zC000%cWM
zJi;N{D*uz9eO9;1ue~T9rM!IxzGiil6%!Lc=Si{pRe|KCMP=dEtO^;(&VD*<Se4X6
z@j+U<s@#=7{VQwCuKW=#{=4h^Yxgc|T2)<bE-FR>>vI&D1afhkSNUE#W22`{);LnC
zZitTER0$_o5Pr+5Kx&e4ACUFSMxQA3tF>M?Q9Mf#$h~QstMS<aEWoX8Rl!BDfnByG
zKX&(!p@}KCM-P#9{%u8T+E(?746T4hgt{n`h)2N?<kz}A+w%H6&c%9@W$yIBptXCf
zJKKOg6t4j1T<c8$uo)_P?!K&4p7l^7Mp+RPl^c7)hP-UHwz)^L?1D6hqn<ll3!#6_
z4&@#`wr5w5jrcG2`whf%hkfO?_F%4B4Lea{ZHS3G5|y=%gvMJHbmfi%2kn?|?}HE7
zSe}ZW;hz0F?RfSBo>Z1+g%zqrnd}G=Nk~g(Cu7s4w~_V+W|)b+ipkG-u834x+OG*e
zFGM;QeTBnw4gpz%cXJ5^F%@O5rqg+O)~2&wkxM(CJDboB3?7u*n=e3f^T%zrsx2{;
zz&dSnKFjTYZ0EBbFDXL$d0&|jK3wo#_QRwvTsxg5&fK6~tojviTlghg^nRO~>{8wZ
zoRv|7*-|wRf^z`HNr<G&Ui&W|uw~x@C(N(pP_o@90;!c~tm#Y6iJH%@dgFneW;e$@
z+|>hf#UxhUT<*aal`^94=IT~S%E1$``sRGnOH4O;5Y~HxxBeeAQ){vTLQP%_P*20n
zL0ib~fW1VnX=|fQ`|AnuUbV)X;~W9Bt@9teCiLHwwGlYE3hIol_b~@BinZpO109t;
z*pR<uGG#3{I}!=2MA#iUzT;YD8>3e-a}r=9s<&3}wZzljoc)wM3T^&HxgTxI%>~0L
zlU&vztm1ClR?YN{B3hKQa!0~m4qKr-ZTroEFndW}MGS$eiqg^PuuodIcMMoBK>_wA
zsPgIJJ(Ehauc`stWCJ$`hF0=m&dbov1>ZL}f5L{nu@nvE=t*>YC?UWkz3UMh@zvh6
zOK9_ObSyC_d_i{H)lDduAv;n5`$kb4Dm}^3SYwx2OQ0h2dx;N^MPKr{!QR!^kenrt
z*zxG62-7(J$`)A5llTykofK;HWEKqW<%R>>c#ba7W8ohrY$72n@kk*KqK*_t$cTA+
z9=EAHTk>R@&dH<rgZn<~-xTQui-!cu9iEYPowiw@M2Yvc)8=wGAnGS&!ue}cg#4Y4
zUs{SOyHNE8#?`(d7+CN{D4x+5eU}JxW{U|F*2_El>{2yRMN{+xyOgUTb7>j>T#icu
zABF{$_q#_{Kv%q9Oi*m4T8)BZ%Nu;vt3Ij-RQ9fNmfOEztKQ`|1uWr&CU23O+tgKG
zJQ3WrkbkS+m3RA9PJ{OHWvhQzk^@-!?YA}W3iQ%5jFCNPx*^)YxS(|}7QeRkUFGg8
z+j);QzN_FLshMnDRCax^AF3w*{;;eq*T2h|4*`9~nsclo>*Lkl5T8#Wil4HU=!{U)
z3L1eP|7F`4bMq8Rg$L(cy#~?K8g1n_^@eS(08!{{=zNqRZB>8LAA!qw8@I$vWN>Tk
z?+WbM-4hW|x4kRw>0n>cCUwN{q4<I^l~a~f{M)1R;&V;sucI;QtOgXp;cAlZ7_d*0
zvEJ++B}`xUKI@A)h*KjL#k&epz~az`a$zLTvtjQ8^$@Xe4pnU+pKT<Dl%%8h$e<-U
zT$M@9hPIGnJz9aQu&Y`}lp0Z1Tu}>s^sF8Ap-RR;JC^%}6JGIa9IM8D8hDTAXC%rc
zWPdFSU|V-S!dW_*BS2NWs7I8$KAx3|mK%Hi3GY-0u>TR8^hM~}*Kbq#VJo@7Y&zS?
z?0-L<pUr7vtnw18{ggavsxl%>Fr}ASj@c|D(epo3HV$bs7iEf|@{E=Py5oFRuc)bE
zH=eio95mF8cA=_N49#T-YztY2^hS?ALB;K&uOsgVUTBNiivsdV)^y2d7O{$!szWB7
zqT}cC9@^zRFL-SC@boXoNCNmoVC5^(x?Da@<?VlAlEV8d?+Q{t4x)(pxNukV@+-_I
z*yWZu?j%T;;nJ#G;-y9jea7l;iR(cEP;7r`zWNsD?)MXO%<6Bcq!v(ow&s?CskUyi
zhFg+tDF!tKNw2-7APoZM4_RY;7w&(8jgwM#o%gH-B08y8CW-JPwmvI!6Qp8)63VLC
zE6>)pqNv~CFtZwpe$kSjHly0!7@rg;`DJVMUT15c_?T_>zu`lWHs+*v_k*Xm<gd4+
zdTNi+OT1r(kTKD=D3g9r9FJrI9k&D)EgOaUeS5AQerf~78K^Vi<j9?{?rK`kLEVV*
zd(L{J45EVFbuo8e&SxLOjDcGUf_N0Vf9RG_P8VN(I7~&_8cx{v_9OZ_<oiR!jN*DW
zl0AKa=7)W=y7$;<3`%$y+=#>wy>3T*b?|sVw3w?#K;C0fSK;?W_A4pkn1APC*lWlA
zdm&}^#jv&PA2eOO7eG8GvkZZ<++AK|^5eOJo{D(TCStu#q1v3b$y_}!*ON9C9Vj=j
z<Y7AP(@|E8wYY`yeL9=xdn-~m=&G6g-PQeN$xCK^8NT$t+uSXAS7a0czYgV)tNEbP
zk<3T;9<9T@94=%TRG<j==R`kbM;?+_2kc^0)A4zp=?V)jX6Ml02kcV9B{3Q|WXTs5
zm)A$^vcvfKUACMvM1rb-GtWbaUdeN|8K&I1iC8JU_RrhZXd{~ut<(#wl-0$>9Xf!+
zx~-}!r|NIEY_U53bEv1sR@Vi&zw;pGSJ`44F2txb^C!t*$*b0oRd;h8*Vg&vmqYCo
zlYJIezOgRQ49^~z#@0n0y^vt(f@z!T3WAVw%huNg8rs#vEpGN-)EhyPx}oYH_gy9=
zDP&8XlSXGy(7eL?{CXAOa$}C%KTs|~YqW+#E2x@(ZFBT5k6c`k6E;X*`3Op$v865$
zVOL+cEw|UjeIM-YC!;mIMGmBpZ*o$~Ze$@MIt}t*d#;n52Rf^HD6}Y6oe<#N30)z*
z3M#kWT<cB5?cHg8QANCc!JU{}4`jc+T{~<j0pW2G%R5~4Il=2ED};ZtLv?{X+$KOW
z2%IBx(FX@GK|LHxT8JH%K|uwCd|$KCx^s;PH~;c~@t+-uft6%e{%Sm`b~J%OnN6fI
z=Ru+j^AFIeW5ThV1k$WMNd1r3mAB;~Vm9Vhz#8luRQRHkv3>Y6V>LZqZRS&`%O||u
zXWMNuml1Ir0j{U&;<`&>=4$8ZfWu><xSr!|_Ki07q|L+;mHXywtdF2R;cr+FhI2W`
z5*3d}0=fM8=tsdXp0;_1>f!x%A?n6%mKSWH8uN8%;)@A)kK7Y%vFiLeMbyXidD1TB
zl*Pi@QZCTvcuNaHxE!FCyTSs?PSlrvIeNR!6~(Uj(tkAmvlZ_>TsE)1<5)ZqnT8Zz
z)!yy*z9?_?j<8^>-tE-?IOgteSlzqJ4KJ$f)$v6JmAIM5uP30aU$8arF8HXc2hrub
zU2i7Piw3T&^)>!xZ8YSXvdWF`4m9&<e;>S!t$Vkl1yDBaWDWxlREYPccL#F&EckEz
zyUXF~>au3<j@-WT6xa|m60zk8s7_eRyHnnxFh9t(Y@@e9t0YlKtN$X9kL<WNM-O`8
zble|Du-KM0;Gdv2sv6R%cubgfhv)A<YTLYS5xwF1?setcO;{d=r~UQ{o_;_^1Z16A
zA~j|a33SI8#cU8Ll!3fAmd0b)xZ#@K{&DN8##FGofp^DKLB4dj(INlH*m@F&b0{)E
z4v^Bi&kn^<rzNW|{N`yJ@ok^2eR``Mj_+Dh1(K8L-DtKcx-ES<Ye%xpCq8aR-yNuS
zpr15GxL+Le_4yRaP2a}HdZk4^UiBkwEz(N*u6aM$g}{qcXNi<OEJGpk!`SK>NVsvY
zi;2{Jn}|h#S%A}=j9z3`IptqwYel;_9b<|VRG-L{?X1_v^>KsE#CMOMtoI*w#J|Jb
zqrQxBh?0hC?0nSL6)jRX;fg%dI|+slEpZ_sV?@G#M2nQhN<w=v`^4+{Wn1*iMBWd(
zRCO=wT^1TjY{^%!q~o^Bxt~LLG8GfTW&iAnovuWC7YX@ATJgyguZOF#z&b}IA|bXV
ze(8kNNlAXIV*|0OFWRbG%PrsBv&-tL4^ILA>RaQAiOnlqXT4YLMyIvrR>v}mrzOjk
zSD;oKqD+RxQt|QF+FEb274o6+*1!UitrU62<ADX%#ke@cf}Wag4eUX7I*G5>$5#Z^
zfZH6uWGsdqX7pJ^^`K#bOfXgXFSZ0Q{>90}jo$X7AN^DQ-&$`C+{2Z!&9|1r?I{(8
zu(@ihCdkyVzI+<-;ufFG4tTEiTMG^)v%+85Ht!a1Ea93C?{<IjM%eDvNEbg~ozWx6
zT3Qf9)6>@Nl@*8pGlyzOxm*^Vob}xr?_+TgeRL0wj{`C1j0*cINdx>440&(FJ|al)
zu+K<=%^gb|$^z6KPH0q1iHWHlj?r~Fq%D`?VbmwwHHZ>3Ps0~_Lcn*_KYkY1|6>7?
zTNTT5<7;|~d5y-RS7Is`Qw!Nc9ly2Uc}!7izC8VMbv$`%&_bTf8a4{i!gx-<P#NN#
zZNi~eTne^nGUlLMGbe58*1(c7ts~G&3@%qv`L8Sb!P(p_;x0d8Gtr!PO*%~n?BOjo
zck8)3i{cnru-h#bs`dPJOIXnUd`!I9M6+lL7t4i4=5-+<SH2=AY!-4YcmN;Q%M7J?
zF(Lk<BGubsZUT-aMFPRbTuNAqQJ3ETqB>qmSQNNkOL5tEe-D%tQA2Z@yA*cjim&rm
zzw{5=ZN<Mh)o51}%yPa;BrLnF+zXiw9kf-q73{FHcaPQG7Wkhoiym;bSLEFDc-Q+6
zx4>C*n`0Pp0MA>)ZH_D4<kEu)t@VDpkfSx;R*u6*{>0XKv)S5i*%CD6cyi1@ZDi}C
z%?`4p_@6t#rW<Z6H*v2-A1${9$^v!|cd+rcWU+c>tk{~MASM*GDF3z1Sw+UfT>JzA
zxx?$WC1xTleZty(Ea_`(TXZFftL_V|pmjuT3`3arcJJ6JHe8kbBR}Qh*B#A+t}oc2
zH^ImL3Nb$1OkXvKEOCUsAq5gZjFDSx$U(y4WS0$Rd!976`#O~Kg?E?bsQB}d0P+d=
zUsRaadTVa6(c4lE)kVOxZab1|gLH%Icr?4?VSJV4lX>8=JO$#H$9yk$_1p1;Wv}u9
zDzrhZnhP=(@E<Ys?PTt8c=+0QHbdyX2PZbPm)_-9|AS55mSm(9WXU<olWr<uK*nOz
zF%HBYjH44%XS2dqc6cTy#a8UvWwQwbcYU9{UgmsB;qo4_^SL11UCQWS^SNBR`ysO~
z#8A>$@T3(dgoW%*$>XJ*xtLSJ)sawUtm@)ziR$F_&t8dfy_9PUDneLXyx*5%tC$KB
zjmtUqTmQnZ<y;LB(!#cdmQVnIE4L+2RpK2g2j5BpN+4CGik8mls&|Jcj`a4nx?J>+
zu3KlTR_8^hy%ffEtD{3+4onxM{93)b+`8G?-~YYUuZ}~r6`Nb^9M)v}a@4_04XYgi
zpgR-?3QO{Ytz8{N2=bw4tTAf%Gf_A{TNh;+-cgK9)9Sd+vU};a^;sn%qCF!Q)zyAV
zoXZ-Q`EKmYhSdc-NSzak=a!tsql325|JKE`RtIjjwxbUn;pWvrC_wgFTdoEBCX*C)
zZj0AOkt;n{yDwxPai6#4!Vor(q;=Ne9f6?vu{}R-ao@vE|0o}%LF<l5aC#As6zx*H
z@x8fHILJKB#FXcwFTsV@{r5JIi=x2ZthQnl4aI7Mq`33LIiX$HjvvZ3z%y?onu_?B
zSX(UfVPCb`+Cc*Gl|UAw-a;W8?MQU2x`-8g?)^kWKIVUJ>c+6C0_geT<%OKZj#p*R
zU{NFI&{I~6zK8<|p?5NVNew?`<59n4Kb15Kl6oS_^myxe#|YOZqpUo_x)nU@Q#suS
zF)EzS!ztq#iQdmTppU?r%=nNTkOAJi**t4F@FA9&%h5p9qNh>&+WD-4GLd^SA61^1
zB_@|J9FmU<IVqO(v@Jx9(`eQtAd>>?VnWovqa=c~MTba<XfMS%NQ+;OCnU-1Qht$=
zn*|ZBlf}#_yPOAWpQ~q<qh^|RloUq=Rb2_#&U{^w(O0rv8yXHwk*`)G2*F%usBU(9
zx%`jG@L|>M`M65~TMmxZ-5$>{Se-|euobJLmQ&-dH>}T=5<9Y5#1yeLw<nuR*lD8I
z-|k34ipbY&t^ZQ`5Txpj(Y*ML#Zs=jz1&o?mBQ0CRgIyPZT;;9zX{bf#|On5Ic*z!
zGJB-7Z@JxZ2lCLj|CuR^5_YV<(QD1twn0Qht$4aoo}3BU>E`HpoJ}Eynh*J5tXH?(
z9w<^M5fIwFJ4lQv+ve5w4&sdw?ZvIVVcWe@uY$L9`j^?-Tgu<meS08eL}s|+$RDmZ
zNAPT;)^~fMsy6MgfdKbZd9*bt1$frU8@fFZJ#oWbkdXau+R6GCGV_ox>1?eH%sx_W
zqhd_3+3d>UfLXtuw^5(Wwuio8N4zCyv4@<Hx!H&s<U%gn(PLRwnE#SZ!p3qMczWuc
zIquW=!l$j~wbP(<9}kh8%n9&x7ZPzi#+2IBCh{;Lfg#cPWDF-e?z3fcpS7u~1BoNq
znCYmbWs-_OQuDJ}#wiqEBZbCH?#EEq;3GC$of`?6rRW{uK+D8+&esFuLOUM|rHK+a
zDwmp&D?Boib6vRo+*&am{~xxhwvezuAHY6;UAby6`WU=wwwTBC6Mtrx{O^za-9LQQ
zmU5nhEUVtE%K?mXwUM_bFRX_lf?&9kYqx`C!HXid1zQ6&gLXB4cPUmQNs{k~d)qVI
z%fq+o4o3<?2Qur#)7t9ps5sef1a`UwtM7>Ow*`xMHqh$5o5I5E>9RG^4VL?|2jrri
zu!bm;+(>Ace{Jm@1x?J>9+k?zF<TYFs5qS3x;x5k`1}{GDcj55pL?L}oa?<ky-o65
zG$&N>`Q-|1xZ|9BIk|byTT4!ql#v{<jduiTaF8FZ(KWaVIi;(GHs9eG1+fz6zE~n~
zJ|u2yyCYD7oGsY>E#6vwOeM_%iNme^j)JsUWZM=n65|Lus3R9MXjT?;do{61@$bBl
z)7F_|?A_Dd4PDcn3#IfR*6Y963T<hh&k_>i6E+YtuJN1<<#PCqMB123Af3QYJLF67
zP)VOV;_KYrZig#yUM|^5<bgVxb3jgmb!4?%#fe>tAUm4#;^iQEq$IP)s%pd@XI?T9
z9!r?^qL<QfKAr`|sg~K)a$_)vvXcpma!JvJF7SAi6{pX@hZ9jIF}XYcJ!O+oCRSD<
zbYy2RRrMz>vKZXjbhd@FoB(l%(>j|g#ccX_*-X_{QN$o0d(mcNNhFKIUNUPy_Ptpi
z(+N8tWg-Q!(T2{#ddg?V%$e$is)8JKU6#Ax`|#l}*+u{BZ?@QC)RUA=LU6Li*d;$;
zU;2k8Te>42Z8=I6YuL-V85}ZJ`2dG(Ilz_4J+7V6ql85~fymuX*h-XXRZ8q{5Y*LN
z2kBACrg>-e?8rT#+g9D_=z<3seX1kax;y>p7uAB})t!EgN_0)6aVOYEf5Ymdo<#O%
zY)$-<<LoQeaA#m81?R)bB8Jh;6sdxA)Ec8q_+E>LAD*qdv)uQu{yox?CjWbz49C|e
z+{GRatk-;J!77g&v<(5DlLW<0pgfPST1z%k7Is;UZp;EqQn2R$ZDd>caoh>s9ucma
zC4!1=_Swis1QBIht`hn+IeAW^#MqK$(rXqBNu1_8U034=#<RZ_mk$unYhciAKBd{(
z0}oh7Om9DI&h|S4m+gd+?@TC3(9$Oi-yI{##Y+UcT*Q0tOlbjLkpzbC%Q~)3E((qT
z?-d~c*L0|A(1%F_uHwEyCFU|e59R0Xo|1fO#G&irrycgS{^)iaP54QyaU^$`2zMxZ
zj|KoDhvnGyJYwu)Q6|-%d>qh8+n9IJzZZAU<2fkjUNF|GNhmnAQ-3F8i=3N8%!f$W
zkKg#61w*$GLun!h580i}Qzw0(HsvavinI9yP9!I7I*0A`a0zFtVV|aRZ0?|0XR-yS
z`J~O>dG3xPlawq*FK8jz=5jSixpY1^4KJ3mY4iTE1Rc3hb@mFh*Me^g9WmvLcgEF6
zZ0a^Divf*9`54lI&A=5Egfbw}Z9i>G-p%09L0*^3*+^n0DoEVAoEy}&L*YfQWEHO4
z<DB6Vfh5a#%VqbhNDf!;3}PY0Kwypfaw}235Ne{ntonaK%w6ZjdxkJzTV3yX#N*eC
zcq1u$eb5GjC{k=q)IbF&zbvOH>f_w+@84l-z3185m#i^ce)#~>t9A8(w00;@IWRPN
z%`f&IwDsPr%x%iLOB&nkttE&gK=y|0Yj+Q<WJ^E+4_Np_1<l*&^VlzKYimNuBQEnE
zCBNC6WsuI396sf9Yx`Gar)*1o;5_tVFmdfZFuO5{&}CbK(TB0Tb<{gj9cL@0Z}10B
z=@z%=WXb-$XD>q5PHz}7lDl&Z5#i`9Vhm(f$N>6$B5*2WHjw>@DJck`KKUTDv@l%5
z0fOUA&#;FX7p{1Ra&Cp)b{o`@99Kq8gLc?Q!q}d~a9N!#eS*0tG2xLYi#&ChX+vQ6
zW=rWeWXG~XahWOXo{dGBo5UbMz)a(Klobz9)?QiUYI$n59@t?gV+CY{@Vbr1q{QaR
za^(aXGO^DmZKA6894dmz?BY|L-KqS6GQ_46u%f<oUy8F8V9KcAEc8r$b#WhsX=g8H
zvzEjSaZKiX*+i&I;Bh`zQ6?AIQ@wKw;E5%e;Aq-~`trPO-ee2e<lvqjn9++dh}cwR
zy;mI2#h98bn_jd_{%f}OpzQsZ@)$hYgN9g`B0smu$$9x55J;LMdAgK%S8{%X^3j1%
zwUs<Td-j!Uc{Num$PH=E?XH3;@b?E(+~p`i+}IPvp^4P(ca;mzckb!xy8=HvrpzSu
z{t3Y!!Lruel~<l9*1b?#4cS?F0AvsMjIGTUO|bcm`9nd+*WDG@3kn7?!gWfqCLdE8
zDX63M(W~zHvuvzHxXn3IInMsnY>2X8s)`J^<*qmY$Q8)I$69ZUz9mV0+FJA1?30YN
zxf<z7yo#kie8Jl83VflQthj9Cau7^mzqRLl2tsMwDyUzE&g+OpMx<@qv)U#A=yZ_2
z(4#!eFg6G?{WHiBJyCDg-367=cUPc;?jG!V$u?eri#?(ZC1?b2!kc6`KVlGe#t!92
zbmYkCM-n)e<wZLjOYb%?yhNZ`U&PV7;t`e4m(;R|eF*vB3!@azp8Y|(j=P+UlB^M-
zH4k#J^O5*mn_|a&#M#<`Av8~8)zUtPas#3jHsN@lpHKDdRCF+_Ws^tbws-QbbN{Zq
z=d=HYP{lSLQ#uVGFYzLi0a};{>@<;!C}UGpf0MZsME!Z%rmCeV=)DA#llTSC?*Li1
zSZ}eI(z8)psyeZ-JmG;%XL1xG)ajhf-c_F9ehAY!|Fa8?>-jttytVUp74*^mpk2s?
zD8A$D`$8UqK@1fyCL9<IKE-I{46?<nAn(gdxom?Zd9|ek1ky#i&CAuwO9SG&yj*qp
z43xf$0ItN;bbV0$$*Cb~uT)E<tcurP)T>oL3I|&9#+$t-5XM7F$G*ycL0Z3#P^a!a
z1@mxR@&Af1BIw_=`u7Ab5hCfrBWlfi96Jb{X#gQmhByZFy$xAa8AG(xYv1EIEbWFw
z)iy?%8v%H%#B~W-yHwjv?<pI!A4InPJ-!p0tT~!VQv5Z>lY`WT_rxXKCFBlu@}8iF
zK1(>2ja3628aX5iInnCfZC`Jj{jb3;YfGS!LZPL{peh<l3NCK*Gy?l}ul|Kiw(UKE
zb$GjBfjWFhMVf^V$?-RRZnJgfoQ9-X>h|>|Xcb9BvkT}5aC2W)#1pRD1}d;8F|rd6
z`GO7k=);&s<ksLqHg1O!(h|&3YNJUT@quu)-ff596SSyXIx0T$Xw^ifY_1D9;_Kay
zQ{K^Pbg2auY7RAN!ndyn!OmFJkf=zvnJhbY+~L~`3*svP+T(pb83)C+97UbO#tHE8
zY8p>Lo=@ZeL&DsX@lCAq%Qod*e-`z}dy>oV`mz%L;JtATlwt!;f_a<CBQXGOWjSZ7
z3CT26qIx_xsxp!@v&2e<em=Ku0A2fhRB$h%lG)xaRHq(|gXUc!=7n74^3*=I2XuKc
zI&j`x<UHouIcJ#QC+t$rd9VxL+@<USaW-*4bagKK#2*>5W&i7M|JbhN5{Mr!&P6LZ
z5M+;Xb4t^A)$4y@o7sEIMS_8`RqwrCaT~l+-FwTW?;<3%t$uGvEsC6|{=LDE6k-e;
z5j1gY-dpbC=eAmd_uKU0kJ?)AVz$;oC|<A7jc=z{K-FnY(UXWt5l3RJ*1tFItwJh;
zRGafRyt6|6ZSao%?k}w6y^d>6BV&sw@qzj_R{biYy{tEh=$>=HSQ6`q_r~d$5NQ9N
z9UR^^hpQju;g;wbS007FL%x8m6mfY7|D3%5>kGCm#&O16_wjY>@JVyZ5nlS==tIt0
zXS9R}g-Gs>5sK}C&s;Z(L)!b^K&iMyA>||Zly?GGPco8h1_iAJLpvE$9&+eE%(4&r
zNVBz`ZdCh+{1-8bc0;0$1jNG0XLR9Unh)m+cQXdNJsOZI<r%h;RSOR-$-j2=y@BU2
z%tKasETB?%95m4HB(O=7IF?;QB^}ti&yMFJaM#d`R$VK;Hcp@ro+o3r3NAAq-(?Yq
z-WD4Ed7G%p5RK7Bn>qR3g4z$1i$4{>6d5=(B*{F$l;j52z$|e>-NPKUr=5*XC5Oue
zoryB-s`z)sB%BTKPjTkY*j&^IlMU|189SdVGk8$ZhUa58kMO2ms0I<q;yT;~?-0*B
z*??X2)#A9k$p2oq#az7|{X;`UNV(*_LEquRkpn56HGyM%3D6ZFuv$*QL=bVT6wdQX
zlsTPHY%Fc%x-toOp$P42LWItQlzeOQeedck7MrWq6s)nk=f7Cp8b<@dVA7f?s=(?s
z1!?qy?W}%{BMqH|Cu~jpI?YzRVGV1_j!7f3*Va}YO|hj4n|Rt9qxuuPu#>jVyY0e&
z%9^U$<R1oT8HcZrvf_{zQl{o8Q-4$CEqc>7tf}5-!tq;6&Tf#mMI`;kstwqk2SGLl
ztqv-IW-<2M?0{fV^@z3k-!OiV1u>tZ|7myF_u%A4O4w~{0-yB^V|xZskYm6Rc^cnm
zZ2OvczK99q7M7iB61P=SQ7*bX|GKM{d6xIL*MT79<^k)Ao+HpS80w@AL|Hn}N`NpF
zeG@;E)9$C&3C!UfHH~ki<|<hEkdKO8axVvQB<FycELR;;@M$|7aFItCCTG;o$ydMH
zWJkQR#7O%xRpe0zn;_pkcFg~tt^NCE8_T+1yk@d`+}C!t24`$F{{eFX%YQPL_EBU2
zFk#~XmW*ggKv#rG6W7Yht(yY*PUd7vXuT;1l|(f*9hI+J89N*ELd?qQ&#Wo$+?Vmt
za7YjVq2-#3g%fA?f*;B_KIgsnJxZaO!+xa)$;gFjESd70LX7jAEkv0|q!BqZX&0-q
zGw?1>OE2a*-rIF>2N9*Rt72n?kR8f=X-&N8i1pC;xvA!5hgr%_lpfiN=TkJW5aE^o
z=>SyDC$^GnT_!weS94c(U_E4a=T(0e<<F|SLpXu%;uX-g@*ULOU2Z`t_HFguL5&m3
z&FZ5$kn_Z@@v3sw6fa~AcNd)UAn_(_Z8qyUh^a(2<HUW<*4<q&lcXa})rYgtlk4w}
ztK^DPNy_8RYz{yr`jV%mym>eH=p1x+p>DZ5t|qP=J%Y(LMyo=(fv%o~N^QNndbCL)
zhRA+%RMLJaDbZ<+ecF5?+a#mgQVos7lkYMfY7am`rC6q1?*zH+y50yV%5!+yI_?f!
zMq&%j{`Q!&KE7_9RadU529cGc^Q?7ySBS94cnSL5heEpd?l>#`#gA7n%0cVPh3Exs
zF~C4VjA~uayJ)qcYN205Ejye`g(!I+sZ3C!`gUQ2!W|j$>F<KzvcuW=;K6-*=|}zB
zZ0(`_cElS!NZ?Jv)W@iDgIA6<H0E}!8kGqDHs<@jv!~CFS8Jg7(6#)9F&Xi1eS`wm
zcy`Lm^;~({PuWDS%dP_?3PFU1PHQrHAjLcc+*I{T=C#xQi*up~8fP8mgL`+t&duB%
zv?Uwu2@7mC+F~V)8eq;r{$r8{o{tZD7GJmdY`Fu=m3_rI;lkYo;Xns21k^fG&jPg!
zi!N4B5Tn}`vonn7UbjmQQx^tqkPF#!(Li_00pW7B3e#Bn$Tw=*mZQv#N&HQ}5@lj`
zA%xH+PTPvlpbLT0)!e$K?Pl*QHy>2r)rYlGc@0;287Ia=jlCgPt@k;;5Tbs?R!0rJ
z9A5)pKyzEY*C5zsF9gk+sDaTOmsg?(w4m?vYwa0NJid#+Vr$<QFRuHQTjTrUa_&Vh
zD~z%(x)W(9l@FGw))Zx8UFG*FyTA3>5mDd<NygF~b&8-VaL9%PjTWMz#3Zx?Xobe3
zFD1qjGJoARzGLKKd}ggxuV-K$^&FqE&F_mNgc*(G+t592xgxk@AkciDb8ld~_V)!M
z*og?M&$eaLVx3u<cEkv7s0ssu6ny*p0s(J@t?2X$n7q4vf_TjC>=p-NsJUKWi?>MY
z!2eEKUqUSdhcdfK8%Pl0TX5!jU%}hp0*ouF+EV@CAO!+Dlo0$4KK&#9L#EN)cG!RH
z0qqRhXf`e34t<A@M43X5m&`)YfJfs@!B|K=>c2=rqC6yHQ6DOZCD!4dvf~Nt=B}YG
ztND;J*FiRsQ^zP6G%%j$M<O6?#Y8r3CwQ$*#%>qT;+VS#bIMnMK-B0y0t6AX#YCSK
zIP07Ez*pDXOa-Bwn;g5&<^uKh57=C8>)?UiyFn~0)A<Awi&fsG`MeO&J&GcthsBrc
zLUb;de+ar_YqAh!1OerAmK2tDF(;~fg7u6oI)GpO>fhTX-~Z<L&z7=J?h6!8;mt04
zndHqp!=8k5IqF4fDQ5qYU5PTkYvMbmY^AEFBy7HZSF5s<5}~4pBS8N>1$DsON;SXg
zp19gFkQw9&sJkaAeBeiLaZivU5?pXf<tJ8u&o$P+n>_NY<eF^N*R_LA8lsinQee@8
zXKS;{=ZnsyF~QhDf+AZN4LDh#NO|kMU`<h`U#C#fD16BJ1cx%koO)crChjTMq<<gp
z<%R@pFS(E4|8Y=i@qt+37x2E_=#{bABHl!FCd>kZ16?z1o9`*N8YKV+qAf=&-qo_H
ze$BR2Ww>9)@o3LQBs2mPPPc7~N>bgt$YDIgx_3mGD>D_)Y23CuG_$o$NHp&$2uV>t
zyZujuRo0u!^i~aKr0<@frJx~)3hOsejpQ*W<*hrED=qYmR5QCMrn86c39Qd_CG&Zu
zsK1m<c_h0+1pYPF7Q-|<9E+zYxo2!NJJz+{VMiSLoeE5H)Ir{~?IAmMPkC^6l8-ri
zEL;vzb)hzndquX#Y5_`mB;)T`@F28CG?ef`l(kZ9j7M37k3b~#tW7us=#vOr6`(43
z?-`rQErEKMRfJ9FDhwXLYVK@y#x_VT^Ew+eb5Fc_q<H8fr`v4QDN<s9P<Z#b+yGX#
zc+1)OEFrB74-K0?pMXHBpe&7Eil-sxMf&-LT*t2NE@T`RtAS1<YSH6Nrj`o+F8AEM
zR2`LVwv-?HOF9;W7x#qr3X2$C3*?sqWZ8$25Rf=3b|vS`gYz7RQqlfbd^h&&gAA#F
zDw(FNA<#q#aJH(U+%e3`u&l^0m@p8Aq9Kl?0{9i=Uf<v+%%c{G$j^JSCceP0l}G~a
zhAIhKLqnn=ZtXrW$l8X2cM8udR4Ai>#)d#mo4d&NVe1;oIU698j6*kD+qluzH^j5t
z!<*aO5L-Hk2OHwV4f%01k5Eg)xd;OLztFSU#vB+84Fk72Y#bX%Ck_Ug@@#XIIg#eC
zMw^3$<>bd~i;s-QH<N3R%8)ljYAd7XZ8^q{634hB7la_nMSr_J){Yy?3w-^V?98EM
zq^27UINecO4B=}aU`4g|v)AjZwzJFnDk$Wy?;sB}6eRSOgL`erx4F>CXKc7393R6H
zSUb(jekfP|u`U~l26Dwf)f&Hq{BRhJv7}jgQvx-#&PSrmPlzlip+Ty8QpR!BsdwzU
zGU*<5ZpW(mNy$)N`<TJ;sO==CNd2d+mdD^h*9o1>I)(IlqL?9%S7*}$k4QUJ+*WO(
z8ea+PWu!|K?d(Bm9vHt^f9Abl!~fi-tJQcFt@zofuYD_0W3a%?wLZ%GhWglM^EN$n
zU_Tb^P?mGPUk?u1c|Rc~&Yq|h->N>Wh%V&FkKin73)yeEWlj!XjM?cahC1R|_Gk?g
zwwUK`7Xh^Q+a>Q+!Ge~eAB9%d8Y`dm<*IK*A(iRexGj4V+zsqXOj=$)gtv;y$g$u4
z7q;g~-ruX&bt;FmH<6jx`{VM<sH+FgdDZ(1wt1|dFf4WNFL<ah9;6F;e;_&$#n9DG
z+1J0noLX`5YrJW{$j*lM7aYRz=kd~{c}3Q~KX3@zFrJ{ssG>Q(6#KaD{ROXp5}T?H
zCeX$e>rz|){=jZwMP6x6toi+cY#x?<L$7Uke_$dkuB;`;C59R?oj3Xz?VGLDVgCxW
z*5>R$S}T^JtpX0QC0AZCrnf|yF3roVsW_4Lsvgls;u>tr@xDkt3p~txS%~ql#kS|H
ziieQzYiCx+Bybl-oiWLA4$4!Y6j$F!&(`bP!8LdTYN_x2f&66=FWY?^i1k;@<e?Y{
zs<9XbDzwpXl<9iPhlXr6b|?o|xK*-uR@0H68Ut~E4p+@yfS`1*{ZTIyv8?EJ#VPcN
z58p*5X*(KiuHCswJC;M?p1_MT=Cdh5?*FSDk6F88;WZH=MqSnLG8D(7OnzwDJw)hB
ztorMiT#Q%s6nxIzt|nqd6d7YOcNnh{39+W~<jehsSI(yM%=|U7^0NV5?K318wHfaR
z&hmiG#yC!2ORjLr=DcaiPH5+2Q0X4$vm1=f$@x=(NYYnMypokKWDUGP2&|hxxgL|m
zKmM5;^f)BYrJQ$3wt>MUWa3hGiVj1HJ`@Xf*+GD|{j4n~yl?*#n81zau0*fW5JQOM
z%HR>x19rRWtsm<}nP~UAutNb7p0ZW<`mL4~FYafcr6;YfDwFwUN!nqnqs)c*#g@_P
z?=6T01pRSab8lQvq<GeFZ*o1*%ssX?Mns~Ckgl=n{j?%l6=D6lEGt>#B>x<@ri2sz
z@U*Ru8b~w>_InBfy4h<$)ev^jHbi~+S*Z3FhevXF^eFL3@7I&on$U?DAl`oVg0uq1
z(|a8u=@f{pyXMfnC7X50sL9%W@?Y+Hlw3dA1;jgfRufe0*9dZ>*TJ?&w}md?#+{G}
z4ACymSm(V-R6Vw52t3yvtp)3ef-FE2KGwt1PW8oiH?83*fV>^Z@dwEFXG7Ve{Dfg9
zhOhO=(v|dqcF23|SG2Q{dkemwt^K1NzPBLjFZbK%I{=mmF$ySQ89Q=sU@1Jc_u&_^
z(~jO794z7*NJ&X+j(PtGQ1{wc!b30#h3Gnd@40%nu<qtUaZgKr#(h=`I3xZ&X)B+#
zld*HsHIbVm4;e`^ZNg#QvDYSj0-wVBZ0g>4{FF-KQ3jjNrGe({wzJV4Cm-@)#(~_g
z5E!$$y90ZAq(I4L#39^k=W{lH@>e#WBWw*p*oEAX;>}2;7Ea$nuF(JsRLJ;?QOnV%
z<FdsZge>G}Q?G3ksHC`>+ENaT70~7C`>P5jgp?5ZXpSvrkTz4iUstL!@=s7tE;1`#
zCi<W7j6?v3V88044>+4;_m%s5U{Jfa>b~GscFQTU?!H73vR0E){iLnF@A`MUyec4a
zf^=DZ)Yc__jgu~A?;7uq8cNO3a9=rZ%4f1RD;FVtAsehQ2QM)e*WFi6<mdknYx2Jy
z+-mEimE-}4{)IEF`My9o+sezeA%7yT1x>eHHw9FT#eR6B_xHD7vex*@mD4MV-Qh{w
zd|$zBc%Na3;cGKnlehVnm^n8Os!z92p_Jw4*?wO+dk+mOU4Z0r+wODQM%TjW-4}6t
zEf-`*&U91XV83mzU__$LtRzI}jIu%+(pJ$gx??d3s|o4co2_7|MPlO<X??!HifcCz
zmG#)1w4wXr>cVeh_11&6=Y4^FMC<bVP&FX(Aq|KqY$VD`Tnx%2iazW^Z0$i;5<B#H
zJ94ceOqHA19<9nWdtazyzP56ywK4B#KRyP>6BarCKrz>H*^!gm$#)#Mq6q+j5{Ssg
zV~VGVmq_^S_@%@?SYo`RmE_+NW<CcKGnEr%O<e|g#-<&X2MA$*Ha<H8>Qvo~cddja
zLKq3%ChqQ>|L7kgg1Vj0vxoE!8hYMm%gepfF5DNG5_%<c{0qK(f7)ahv)%5ZYFf;n
z!m21$6C#;~aXe%kwiJ_9u$MROvJdpuf8)7aPUv<btG$v<(FQ9@0~DF9<YKpzs018-
z)$5>#Q%KIiVzg~*%N_W<w2Z6P#`}(SsMYxoR6=}Nz4jV<7^ins@KzsX<s~mD<5gRe
z^`N|8P(A3B1|MOzw&Blgt^fULF-whW<2HA76IaRBc~i76g-JnMgi`e#LpVZRzt%~c
z@vol%y*I}%RH9I28`j3dUhZuX=*C%YSsQPL3|)2KH+n;?U_?Z=t}W<l``=-RnLUDT
z(Vr#+ND9hJ&9U0DHg4Ds0tjP!jXET13GoXbD*5HM=tt58GK)qHxi!2jjDo;zyDvrw
zn%C)ngHdoT&211z)|*{pc8(`>F{$j!GD)#PxXM8^5M_lwq~17m=THs`N|v-%2yD1&
z>yduNq8`e6CD;OY8Z<JJ{n1uO@l8YxhjZd2u<ODg=%9_pf~VH$S^|<IIf(=NcG=M!
zfD0?-r@VBI`HWE%67(*QLXigBac|q-C&6%SZFw9%(qt#Ao%tF9{c+ziV*A)cc8|HC
z8<(81$sB(tv1q5N3LcFbUm=^0-ViE7BA<;KqR1qEmR;4%J6aX;$Y!&Q)4{_F*@|O&
zmMM)wEc5v&)8(LEq2T5#_z6m87qV5^97SY1WeaQ1-BqY2B=P&<CBUADvJeYjjLLew
z*zsZ*wiso+bes%H<e#=n)uDS6^XH}brTB3q_3;JA=H<_9IokSxRj()QO7<t6M}WLi
zEz)b8v8#E#7z^f}(<e)AIhol996#tic$Fm(kG<*x$=rMJxw;P|Pcp$-T3xkpeJBDq
ztN%dR>OcMqTjO<=cM!DL5H01g`?{^oZ{+00AfxKXtq#t@`mXyx!DGbKfLd&d3J^(9
z#`0!GC*lM3l1sh%It^>2MVzz^*LC@#L?L1|qAU)2F<jZ`!$3^&RJ0~kp9Qhn<_{FC
zN?h(wTH6N#&wUz>pEN(!r$4v$YINd31DjZ<ZMkfp9xNVg9bW(MgqyZUeMh;TGp#eq
zw3WK3;zYVX5E!Z(J!1)E*z2$k6raDoYBCdGX&v^LY#_>@F3{|pB9Hy+HWcfAMtqVD
z=M2#rVdCjvAp(2MMiMNHZ@~Jv=Md|rYWot*%V@TRk8HLhxgw%e_uJ7A6y!_PVLO)f
z`nrh!GZy_ipH`OMj(cTz3aj~?NS=hT!O3b)(uw<1F<!wSllcjU7_~aNKclwD9bI>M
z>P=;Lo9J%3s-R<~YR}o(1Sh0~^75F;GRPEJ$0Pi)*$)H(AlGkU^lqEWq8&XXS1JiJ
z&PPj6@GHDCVe`KC2sS*u(1iqP0OyW{s(}&|7Cok2^f|$+_S#~C5{cWT^}LkpNkC`(
zJ)#D_g%GsPm0yms)1W9FLFEuxjx`?oie1T*R4fU_H>_lrAgCU@nx~Z<A`qNrAB=;K
zoyjWyTb?Rbm*wrBwbdUC#IOtBWR&*xA57#?B-n)kUGu>pMWC;&;e-CHOPuJguF^H@
zHCy|^SZegJ!KyWSOT45ZAlBtjo6+*@+;2@WlyYC_AtFM&t^Z&!LW4xkux9@S3q1+t
zy&*;}qcP^B#Jb2HPFagj2tw?nZHzXLBiDMuTB|XL=@xKq_Sy$zPu=#xa@@ONS+;yI
zsFFdlLCD#nJ%I1-29#~lk^8{A%S?wiM&)DM<8#sV#W<bb9%Fn&b=?&pa=B4>5~K_t
z49q~7?EXQS<n&dw1g;k5NgK!|dFx-WdKvoQxpn_W?<yO1_^_I_L$NAPz@5HgBe|hF
zNkw<K8k3aZ<+8qEqq)fv+H$-V7RNXFNJ4`_H;(DiEt!H~1&&3lLXxs+qMM%pi$8B;
zx%CC@Vg{bQ>>zTS)$;33&IxOEGIs(g`7Y7Q<I%DBF#tgwB=fk}AloJa0Ny|Nu*qsF
zZejf_tngG01-AR7O((#-tAy{*Z=XgQZ!_5%uAVoy@T{{58oQ{#fphsALlJG*`Dmjh
zl2r#@rZ)e<cw=zF>mit4RYl=bivO)G_|81=R~zkO&MhQy%CD`bZ`>B6VafLtWtXC?
zFi4;{$<&u}H!)nZ%TXaEUh=VC&S?^}5MTAG3q|$xU0um8kg?d+sF2UA^bYo+KmqNL
z!K*&xC_s}IKECD`Q1_u^O%&`;DB>AgU6r9M;p8h6Mg50ji7)^rl;E1Eq7G%}p^zyJ
zA1cTTRC&PG=IC3x9<#;|1;*-Dcv(^eeW+aFKW+LGYx+=JVZ}C;cl-M2cpQ<p8w8)S
z=B$MH4{L=%+Ypt~c2_2bEeW8EJ@(qhY7TPZ)g_v;)~q692BPWZPu$xcMfq`E3mF8n
z7PjR>L3AN{K4b0vYtOFYLB1_#h1yJyREI<9%5TXwwnt-yp?}#reKbxX->|swkFwu_
znd?nZ_V?_@KB+HSiupNZ1KAIR|5G-U4S6!K((<Jw#ef}(Jr>*OF;J%Ctc=9WxPbC<
zg1dvj$Pqt<HS{Akn)~%R)Bs05bnf<ov2dZ2t#paOax}Up6&-TUl9aXl`-UCMw!KJJ
z#{yW=(L)p*&r!Mwib~*3KOt}ZpAc#%y_J$*lQz_U6yIT+@E@4{A(@@bU6HRXSNwJ8
z&#4a;LibafY})^Nu-(o^J%!&baknABW}?hby||6DHtQ|^>aT3hm#&LMLgy3A-9+oc
z5j~f9CuutELhiGozv5uA5DOuTb48?Z{idYm7rh-1xnBB3pN6apy6sYcBKf0YWIhQ+
zwUlKn)tgLBk=8Cdcn|%xEyvg54$BRDl{0iD%An6gjDP)QFz8B@N$B`$!8uRc)$AEs
zM)4Gsp3)lQ0wYWACa+JU-+Rv(u6zHA)is7}V3t+1)s22@|1ZLT@qbp|=r|-vTq27D
zf~dGPjb%S}lVUu2EmS|JPtiKoMmtx^IqmH@tg*2mj3NzRmo19`N(!CFt*J3k*kkBq
z)_VoXXG;vAehyo6HY(x;X<cDId=j&@|19srmd1Fnxn{DTK@r**7-v@x!PwZR)*Pj~
zf2VEE9(l-LV>PABZ>yH&Md*_)u`Go6LKnQ(+T*JPyl>dH{OX2T){$+w$xqt$=-maQ
zB8#83PG1C6+Y{F93;F3y*6To#DHD9zms{7>r#Qm{6^!V&*#ZTT8p@vGW^rdMy20TX
zn@61se<-%0EIl>!YIz#;5r;qqMs~O{C_=<^K}C-?7Hlm8mLm?;!}z#ifEC?8F*^nO
zdn^EOO5zL*&>J?EE89oV@#EPu#_!Ufa_egeAkofnBg-CyE}52Io$=hhFMtcW_M`im
zh-qk2dcH+<`Qtm8tMr+Y)OP9}^DS1sh?X$EvT3jQ7U`wm{--88o8!W|$%E)kZo-Uj
z#KW6x)+-Znw_w3JuME?qI03G=IPdF+s20_yH+*!DUC7DrCdukTjG_2(Qkh9Ua?zX3
z*8aTN7JVgVYhO^_f=mA86Jqw3qGd7q0Xnhpm$U0OOfi<TLc|uBgvPq!K=0Y@_E9VE
zSc=ncSAP3;H5#9R3=-R_HMjd63y>oNEU@Z+#}h7zk#)~$tGhoAi83D^DB+n`-(Rlk
zRt1%``e-9D+Vjj-#)xa8><QLak;)V<u;Ko4|B1XslA!qRA{7{B+B8O4(U6qOIAQDV
zFSrJcVpDuotghFA{S*w_`uhvc`qTK&n(t4l1#u8`gHH`D1vV|#lG7j_5L~5R<&EAu
zF;`ptug*@}>{A2nATMdlX%uS$$E;f_&`*@G0ZKa4?o;D&!#4B&z&PBf60W=>p(3UL
z*L%A|CDIvzOXvNLW_3tp2~lFL6>hCN#wj-*QcW`I-Y65Ma#iR{)>oC4pgz*3*g$S7
zR}INTjFGrN{xvRC9j@jAO~uwpk#Q(jjR4I%Y~=oe4)!3$KOB{%WYV4jtc~993LUXk
zy2V^F3B!2wI6<(BM-mX?MoigJU)!D%L-SaUjpGP;dwCU!uNT#&2!6qiR|6I&wjzwG
z-5<ot!zASt>8Eh5C;c2C0K@!^-{aAZk}ADu6S@A~{Reg+&G5Y%?8gpesv1uQ=hGFi
zCy?Nr^<VNc!c1r;VNtYSn5Ji`Kbx&3$+0=_x4jGL!uj|jHu99sM_XO07s0|xrFy{+
z+<*LITgVwfX%BEJ_bB-IMPGKzjBN28yO$?Ki+jmOL<4{idO}Mg7DDppFY;_YV#|3t
ziGg?J9lc3bBy$&q<XiF0*|U%2+;-J};2`-DKBf<piwg^kC;9{B=A(gu?5}$ujP`g;
z_V*L933LG6xB3A`BAilmutjQQ_1?$+19--+@xRf!_FKaP<$Nhu%Gw9~D!|sztO%tz
zMG{;zJ`nd>X(&;CM8$%<sR)uOtjuX^dLX0|%4cHhz3)Ab_FHp+bd4=LVH=`MNyE_W
zL2^B3Ez$M$0oOd<&c>*n@tG8duGNR*F+T$-viX7Fdv+9Iw6#TFr@#JGh@7RQR-mRW
zUelQa__g+G!CoyQMsnxcwg-Zg*@M)t7=3rdXtJLmmsl()F~O`eXU7+pv*Siql}+*n
zN&Gxzz1~9s54Uyi0U;-D>%LU~_>cc+L%F(;b3(K-RP=2a&Mpg=U&8v@p`0|nU`n|(
z^8XU|{;*wMS-<ZpC@3h<6$QFNSKN$~7~&)*aS|tS5(5TIV!#2<JG>gtxfpD)CkFfV
zpTx`%$!A`U#F5ZBz7o>ql_(r=f(cG=5>H|hhoFfQQBY7&P*Av_prD|jprE*)?_O&?
zkSUJzHuuhG82_GUKYOpe_u6Z(|0{sRjS}~S<NgH)bM%1HYO40_2_I^4@=tcsk!jfL
z3}){#696!JaXaN3|L4$Foz5oWSXoPzGCPy2QoNxjpDOpojx9D7ZCNd`NLm@2j%VT{
zBqb_``An`exFwCju+4gpq9#b=bKYY&Wr=;g4h|`tEP9d@e-;#1*{ZAYEOF;+bjypm
zn36_VOcaK3Ag%@6ioQH+*K-*+4i66;=(8KXGNhcw(d?#!a1>$0*KQXl^!8IJN-Jm=
zm!73+a`JqW3)=;|lVCFZ*Pto8X?H7-vaWfS!?~A0P)$<)5ci{{-!@Q+qILP4MSJkL
zrTA=nn1lcw^<knDqwvThUonzb|BKo0g0ny+A{<rju4TV--a~`}KI3<Vv`}LhHho1j
zFE%{r?PY?}R#wjxD<U7zE4J!)!JPD{=2vktSRKuDQWZ9*OloW5S#bxTJBjFPe^<!b
z&&QEHYJL~@SQUvO=~Kb^G7+A&+8i2>4*3kqW?ARZOU{5O)kh078AznUH&T56F6b7;
zRw?Az_`AI8#kF%~7P3RUl{Hl#>mK&V!RDNqJN`?>&=SpDijx}VD{GBs+9bIXD8XCX
z?+Q^RuF%?jQOOoH2y5aq=Eq^%>ur=pt}`DKG`;V4<pBMD(7N+!?Nip91LM3R*;Pz)
zf6T_YiYLT98}u&;N<CsDKFs>{>ufC6*srk256aa;htS?DcFddZhp!%wrha1<gCL_4
z{mZ`!YdxN8Go+-5u<G4*B5+ALBQMy=09gA#TE4>RPW&$Dvyu?vGJDFGreyB5)79R{
z8-=<6?1L02|88fhxqmOC1q6J`CgYhr0i`mO9&_q<aUf9^_bW@sbaV-gV7py*aY8!d
zi+o_vW_`}}^{?1mwLD(H_Y&sOCzi_Q9efhocTpwafGq?J+W0GUkCVM!^^RHW#e_z6
zB@V%sUW?|E;iYJY7})h(k!=G5$~|%;PGAX6x>?OprZQI*-pWpj+$sN^PwjR*!?szH
zL+H3ICDygDnRjAm(8?+7Jyu1@b;2;iG;xKod(jc%tWrJ|sp)=pbck#6LGHS|8(Cpk
zjfeSuKYE=<-T|4cRT|QzHLD4ZU_aF^Y*~#HA+f0EVlS8HGc+<1)>Oi6MNK&cFVxz~
znsSunV!f)S&?m1Bv)Wti*=}p%duUX#38@yy+M48;;8tpaw7?}|?S=fEg?1?!zV_@#
ztIh9`(NQoJa#L61l)}%J3%|v3Vf8*A9Kb~-8)_<jhwy@|O^sD=x|nowD50Kfum{&R
zYxZA^F5jsZhjg8^v#tKOa-CRPO&peO@E(`|+G~Q84IGe7q9Xt-25t85an0$yF<mc{
z>p_mLI};;Fae){fTglIVvwbzyEygj%^wpiaf#@usdc{!pCj7*{eqjB+4U3aIaDp1F
zz{0_dn3DKfnpA}$Ou;;kWs{Fe%DrPbF_e4CkUH)giMi{zPk*Sd8y7Ag1pfB|S^9i@
zwAotoiR_?Y83jiY9^xDAbO3NZh`Z|Xok;+&o&f)(x5o>T6F24UqokwD@6*wT)E6Qy
zSbc1!TJN)kETCQzs>-C2C{Vmcdk!77xmaP$v`E$S)t96}5J4%~3<w0Xg@lVd1Eh_-
znm>0BkbcY-6K!!0n<~d&?hDd26A2F+ajlxE+{IjizwU=+sOv9wBVPMXgv{fwNBa-j
zt$<NEx~N1FjXg@DI5gXm@5V?km&l!5azYZNKfYTNmwwi(B33@TKnLnUG+1V15CsOJ
zC5c}`#w+$P5ii`5?2+TaThE$3T@HV1hb?<LhzDuGdTsgBm3&fsmA!)PS)8_3JRM}E
z=!nJo;`RA-92oSk$T_Q?4n#+ZB5yKVoqdyZf5g^A4|$$MSl2%N>w5I^)#>`D3xWOr
zXNT;XrwiHH)Atvum_410kzYq#0d)xyfw{Qh)jI_2tu*BgPygy!o*$P4hU7N-T5KYH
z_|wTDDT%oRVB%TuDjSCzQY2?{y!R01C!+nq8achRJY5KBKjx3trwbwN$EZtUo`6D3
z#W&XO;~}wi_;Xx(B=YTz^>K#4ZI+!rTcv}recq+2ksY*de?c%xi}h9uD;<I>CHmtT
zxQX<uWAeQXKArRqodTEEh_6)_8ZQM_j8$tTmziS@<<<@xkFKy57dxy}yvM8VuW)h7
z3`))91c@wrC!;R{F$pc0$W5WzWj9p^0?8_0Bd9o%5O@3O=vE@f6pJq1<{9r6YYXbt
zc#ma4Y^xX_n~G<uIHSuJ{|%dt5&Uiucz;J5G!AK=@o5bowArWQUcm?=;fu1TPlu>q
zlvtjv2K<=!b?+l8l#G742^&W+)>bn<Y>T;i#Y^y7extf(*K^%TCa0K%8`1Pr812vP
zrXvlM+HPg%qer^zb}k_@$v3tXGa~gYH>}hLnTF?%x8tsmAIsgGmGg|u>Rv8BwsL6R
zeTP{UVC+FGowQwYdSOxRp~FBB87TQEAs_A;lM}x^Qw|>}*lWw4iPL_F;(xj{mOoRm
zAzY>_o=HtXPf0zn@)_qW4ya_>s_cgZrmLSxl2b)rvx{Pu);!}Rr|_e?C&7lb*)2y{
ziEC?~iCyd&<7!oA!`k@DE!Hju)g@?c^6jjD#*g_EoF^zu?FYtPfU{}v9^kQvus#AK
z4LNSCpu|IjrTVJq^Axh!l<Uw<R$diaX-*K?Z22p;1VEBygrLO$lK?3fm9;%nsPVIJ
zTDz|j8Ps}6qLDKJ&lSP_6A@E;tIkyG2SWwiduPBV{UT&bzJLrPm&ScQ4Df!zukL4(
zT%zE{;>1SuMt3r^OU75Z-2V6or$6rA!E8}ngiu=2(a1AF_q#qK7o?3P(A~`TnE$|A
zk%^7R$Ayb#26B!(?hOX0Y4VYs$aSD%C|248u&Y}|opp95s^Lo!E>hloVW$E>MMKWm
z=^ULf96J+Bp_!2H00P3jUG!^{)vR!sw5jNpN)KG6USfPYp1F68^j+WDj3a6(3OMUM
zj1qRn)aG*AC7umI)#l@KQ5z|v6qT{yupJuE`MBzR!Fu)EVvY~<iah~V`J!EmXA;cj
zfl11x>(9iEg?{WH$|WRZcI`&Y1+5*V;bsoZ9!j8-RDR0`CYXZq61Q`(BJ8`GY$<>_
zLOIDhiLjBN>#lDv!UYAmkJ>$NN}<{=yYC|(RQl-=dyq?ikPwN72^8|;tUbzCstjPy
zmU|Ivmo0m?kO6Gu%wzepalj$Xilb8q-il`nSs^i$=kBqU&jwjR3GsogicdTUTA3H;
zZS}K(j>=<8cuD?bp(#AKHolN>Dklgj32UAWs*x41tTsOa>6lraf3!G>+mzJ62f*xl
z2fetUT0`{eHVsFy#(c&Ymo*ZiU{f{vWf0Q+3Y&d7KCbJu7M~m^?*|sG)n8E3)7lb3
z^mSgDy4{CY+HbTR32yJmh;sZUXzx<PxHI~cIfY;n?~`)x)*Y~v%r|JtK7mK<tziBN
z%WH|oxBh3HBFF8?je|~r*-#$!*<V$y3D13sqgG!k*js~rV`B+5w-5Q=L=JNI+I2gP
z|E6b6`nwWLvtY-w=ik3)C-NK8`I3n%=XMxV(nL-HeKUR^sKvjvQ;soSj9F-PIw31N
zE^*)sJL4mlDAnJjDzM3D9cH{mcGRXE0(35-V$;#W-I>ImP1{U7(;cEDD&kaT9nzv=
zwYda)_fDKTs@4kK7lkJ8PCP5pFqk2w`PJ%~+rg2FZ84sCs$Y$SV3OP4jxu-HuIGaF
zjzThT_^+RFK7^{=^!<O~PqlWdg2FSMDn{(~v*nsbYIDJCDLz(w?j^h9A481oZnd?&
zJ!Dm5neXMOC6V_1XH#DBIvwo&gXl{)_viL7kQ>K`M?NSzx+CPRr2I&29C6&^Ua@7h
zg<AXtp=i1P**8o*vuuN?K4L3topOjuW>uqA@l5wDmS3*uZ)|mKkOrtO*TWj`4|c`Y
z)~2w6NlM|}HMQjuZrN?Mk2^x4?UzHHf4OB(ht<~>61F&bkf7IwYzu8w9E@_uG}Z=n
zK@KAC1QfU_TR|D<5iWyF-?il|4JyH0i}(K0EA`fz5Rc%}`GK|hV9K_7)Y|hMYDc8(
zj_8$hguv_`&U<~CIImZ5dF+f9MJA9Cga_-seD;qLkk7gP&bn*;#4APZWr)%DaF-t5
z(Ce%6`yJLFZ3<^0RsoVNerdr59s0rEK^w_kI@pKD+E@&wutpR$fP~XVZG|V;uZmmh
zcr<g~R&M!!v*R%j&Kz0qPSpA}D7T&J^xDb%t|+v@F;37#ZLpntlv#x`B+-w;Y!>Wv
z&RvFE$cGZiobeg<kJzODRrkD2`CprN*>sFNsakyau7LQc<Yp2DhVgBf%^x42K%2|G
zmktR|fvSfUlpfDck_21GXYzCbj&pE!SL2zy(Z1mDp2g<An2U9I0B`0ZrCsw0VutK3
zR+{TMC017=kx0gkd{$gy#M?m!s%KDzGUOGzRpFC3RlA)>0u=;>laAO@d_$ke2Z&Z|
z%$@2Pd5S?g^?f%`Stgo$*@^Nsii5kK4^(|uEgnP<$}ovu91iGvdzc6aKj!g38M5~%
zkV|=h*9DnC?>TDA)+OgJM{gJkTfXked&{|duU4~ST^x3-a32V$mFvpkBV7k9#X6@Y
z4v(q@5TdrO5Q9Jc(bmLUUG&Osp;XUn9gsp}zO|a)42wTWA@Ne`)mGmsPgGHHT3tMI
z8>{rdGge=<L#skNmECTro_&c)%Hy{i*A>dBO2|#?!bhOHXBeM84?S(pZwi@_IPi_N
zRO2Z_jl|<i*6RJ#uB)>)UkHqFUDh5=iz^G0lWMKQn{Ify-uC*Mtb58j{i6<fx$N@>
zJ#fI?@x5CNit^ifeW?zzMVy8HSSo3SMX_8`Avv}UR&S&s7Ipa|DB4KXE7|j*Ys#TN
zRz3TvI7><CdMxol-gHA^1jik;&D>tc69GQLpF%gmiReKZ=d+CT1zLlXKCRKB0rkIs
z{u}busl+>&aMYjcg8Xx1+L@}G!UVyN{%Diyf&?kk0F$2bb=%Ww(+LF~DUF}aWK-<Q
zsD^DeR-07ybL%QPi4)C`YC+BCUcLOXEhGve=)gBr5F&TkVt!i5uXZiqgl#Uyt&%0&
zKO#%ECVV4*QH^idZbo}9G};#?$8N=vOiG)XfFt`b6I=3kB}(B=0@I`9<9B0YU5_Sx
z?%&iOg!kUM(6$w)b5(LU&qjcDpdL4d-Oq*o5GDA7Tsl0PDWHs1q*{3|A3`9AJz7^K
zSW*agI*Zx*LO0;j2+6T!>*M6m4_bfN9!THv^<`7>8elRj*2e~7Pb|(~>qAT_5tq}&
z?{QT;<35K+aB97JeQX(?l3Z+B)ivuw@)@+X>w^{H)<cV4vp&dI$&~gNwzoDvZ!gzk
zA6mY;fLu;;pIH6+pkipEl$O#OvP;Zg<lt<~_o`~F(VKkqEqnG@bHFQGvxH0KbZCiZ
z_({p7NoHSbO&}0_dWH@vqNmN*W^wWl*y-E-$CjeE=*TzRM7B3am$$rRl+Ns`tN)P7
zVf0e!EC3qA^!gZ}!of*@+MCZf{;Huw=``#2B_OASvTO|cjHz~XNYOkaxrhge_utRe
zY-8&S;XaJyk&DwkG>RYZc=nA6Oh!j(I*!M)lB0?RP!#HkSaOWiFdiq@2V+G0&zv9y
z_D=ZRi86wJJCzFoKW?WJeTV~>#_f!cBKbr`Sc+HPWI|K4%c?xFU{enI<{dU2Z$$=)
zH*|rc`r^kTs3|?n?D}6_9NS6-#C$Jb0Oq2||M_@UZs{`T&ut;Msl<CrEl~|diiffY
zXKgV*1c1w&o3lta*YZO-Zz#3Wbsrrk!Y;d!9~38J-C$F8GuMIGZqbNaKEEwn?6xl#
z1_zdBDL!>q1QkC$XLoX9n_KO!|BW^R=Hf9jz<AwvY}&Z9!ye=YZ0zo~hd$7e{q`vP
z`T0N0o(tko3MCzu&RuqvWzWS9WB0*a`HvYc85if5KNn=cYhc)l=bVhlmN0KCpGzGW
z#TUt$<uCE1Q&x}F&&A;$#_ZHP0>`+<-}em`pWC(h@qwc%*;bPu!$A_`{#?p7^57Az
zq_#jjllns305w{zKEK-e`?fY~h!!Qsq9P*rsxfCknG?BXHsu$a6@k~BK#-4x8MI^v
z-22nqvkF{n^$);`d8CQ9=gQe?@38jg3b~ZK@KNi?ez*|XmA&4oZKHKoJ;FE^pue<z
z&lQy2txSU5{vQAL3~r`wa;{w0!Q>edXo$=q8}yM$ScRL_NH$iABe_M|SO82)BkWVh
z0sxUasmsaJY2)4jR?jKSq0t;x7S{Ge_I(6v)ycpM7j!(=iD&+vc&^a&&AWEmspo<s
zqwj!eCK*;fPy~%g@R<MsUoYjQEQZd>3aoP3$?e;ws%LWl_8^4m=klg14pT0m&qTwt
zH!JhwtS{=~<WqmNxqussR1pflhx2rNo`1Yg{cuMM)lX_-AGWJG7*oLu(me|nV=zgW
zN(`dh0N3J~C+u+#rR(ubx=(Ni(fdogQ9b*#_zXzfa5J8joMqsc{NHZnGf3<Ov{I^b
zdE0TQWsfc8@kFQ>d!X!Wp6&ilwLv9WabY5~Swy)f#*h_#(C+0t(d!Y7bw9RT{Kl8|
zAomMOgex4y*aMo(myLI)JxaXTgI9pn`RNpQ4Jg0JsI;suF8-c_NQlL9vOJnardr8)
zwxX__8ocTyfz$>4kdh1AslQY9i7{JUmy}0oM7wQGHD<9>mt|X97pOO^0lB7YvJqYd
z&~ZE)uUKt7ljQ#i2LOGW)x|UYLL<qgch>6TnSwokfOz0eZw(0qy_dL?1eok%-F?<n
zec~lEk>#wpt{m<jt#eCGsN*BbD_6k6=XUY7tu6@UK}tK7z})ue&y|jBB#d`QJiCCN
z8>$RV`@;5C<BBu&a_^mWPMKkDIL7;YWR=w*bw-_&_t0gsDvD{w&(C_JBmKe!L52Y9
zuUdP~7|+)%W_2*$KtzgUBKPpJf;N&96+sw_PTf`tw+q{LEIxIKub?O(T!Pma&m|_f
zSHZr=y(?TY`Eeq<La6wRv+^Yj(#afa5dGUkt|2Po4@=ssQ@P9o!b8a6Y2OgC4m8-A
zx}YWqAN!TuPqcf|S8j1~ClxAfD%NqD1ZYz>{dlEiBY{c$#AdQRo`|@a*A+^L^PbH`
z<IKz~ZRUJ}vxS5RGS)3bBPj$gLY3rWakZ{oDj)ROV)a()bNS1CYuDnLOvQ4Qk-_(R
zUCAVdX!f@!|7SOHHHSxw9C9;Zr5MJb-AeTR^Jq86?{?L(Do|=CzqO@!7NxeZCx2~s
z^0RGd;Ro$*Ho@EOGK5H6>%Hn3#S7TkuQ+7)bK&<KwFmK@Bzoe^O8MM}*)3Ne%E(9F
z{O4(mYxaCukj`Va?0HA)^F#_P*z)HqwxEb)I?F4b4_S0GUOp?IckRIN0b3PqWfK>e
z3UOH-y~#uNYg<$G=68@H*w^M)#hF{Sl^?C<c|X!r<+9p*FNTzf;5uq`&j+bOTiFdO
zQy*<u(V~t*onsBp2j(Fxl6KASu||iMyh_&ee2_BG7U`Vi!1(~oKBRQXEe<=kzI39k
zF%x-reP?aeycp5i6GnDaUYx2E(cweYZ$O7py`GVZa}k<io$)O5Wqd`a`?4KV#dQGU
z@U3;nvq)`;^5RSDjp_gQ#qwY4f8NQBa}B};MIcKjH|Sf1+c6&GBhi(|y*RNOtA@nA
zwW!6uwqwsHAL;q_5o2Ni2?dI?Kn!3S7unzA)%THL27jOvu{HpLQtM>?;x3)H2?q$p
zI6Apg6+lXSUz}0xbUfqabM{J!cRS<AIof5D`L^&G#kOrz`K$!5tCZTOHl5o;no{Ca
z5?xd;hJ~1|)~lq6VLHiu&qZH8QTZEvVe`-DJ&SO!KnzHVH(Q*<Y!}j%T}|}i#xMA^
zm~fSuxGC$J&*YDtcK!Ku2bhPqcAviD{akLBX}gi#5e$suZnxdc2i<*#u+@<*IJc?7
z?%Uq`z+Y_156~-$e7h5~byE(uXCyhNA$J49Vj^c;`Ci}i4I-9I@hy(S2L^iWL9~^L
z8Ltfwne|bC9@DF@xLicBnAr<XMC2g$nJs%E5Q^B*p0a?;UvRP#?A#(gKDHvBITq{R
z8C&^6keWS^npOE@cMmCoUx?#N701Gfta+gjqFos8)<#cAgujeKX62ky6VH5Sm1_Q+
z)yA{Wet|C+LZhXud!djXane@re=knrwA~Ohl5;s}H*!XsDR5D&VNEXtxlvef;bYmN
zW{0VrIJg%7^Qjl#w$>L46UJwt3VvEXtx@DsGH>`hvX~T)`;Hfqk&t}t?1b;#-dqd>
zOY3ytFcMRS$MyxNM8ZPByAxD*8DNaw7YeOk_m1^@Cvupi!fK2qla8}(8J;qdYt5-e
zZ-wtgE@pAYbT6u7)s{~|5yx{2P^v-lSm1MZ{E26>iHp*o@MSqb0M*Id4NjesOUXNP
z!gqj>Rj7pnd!!WjJROZCsLHlA$L(?^!8|gGe2Z_{q&HfejQC1<`cLIjbtwJXbgaFu
z0Gsh29rAIu+3ZYtLl5KDKNriBd0JJ{N7}3JY`$8!Qf(M+_H&R)_tb*#X3^A5J!My`
z1yEgZ<U=mrMIRK4FY@}eTmqqI(SR)2_3D{>QGdfBz2Q4}w99U0mq4Nz=F7ly>xH--
z_e<%FCq`mGDa<&4FY}Ur-%i$W|Led(yBqr_VgK)T&->{6OQ??SSBOa17~?!9A`cS2
zl9N<QTMR4tjFRQJ#6w^&*`r+epC`Bn?8Ojh<Vj2m1-LJJ(TN68bFit)vjw+^vfV7;
z*|H*9Alu0iG+JOeKuS|cA{Er4g#tOgwAC*Lkw8R3P{tJd;;*i|s;#xPFNQ3%X+Zu&
zH7}M`1G+2x?`x~A#`=bS>*AMek6&ATjI5hb&|gFVXh>+}_`^{J=dJO@LX&p)pcVDk
zrOuH2V}y&%RX5ITtAJ9Af3o4Fm#y{1Lj1(ULM0sXzJ81W?GE{WA#a58By~hLg<$~m
z<T$(c#fl{;u`rTj-Ne)B9gQl%^uAmI6`UQiZXa9DGrhhVB};%*9R2<ZKW!U)Vz?|<
zGm;Cf3_@~c8FSbbBSEc!oM#aT-NZW{&vf@_xfG3f+(Bbn7wtrDB%JGvoveCLo@1dX
zw82{}PX7Eqh}JokOMSpjM`OQ8#3WF>B%V<Rs+jr|^E(;OBqCkln$U$a<#2ycg7~K$
zo(-?R{<_Unok|o_<1cJBQ3RwQn?}__A;fZYxoq>*EJ%K53%N@B`${1ryBeE1giy2?
zpUZ5m;HVi~f3Lk*DD_C6UC*Jwd=hk3yQ3=f7*6sxP6g<^?N;o*dxhz;%Y=ZptA4%$
zyrmdZ29ohYrK!N|FmX5G4*p_XYVs*4rGg>;_k4-C2&CA^{!nxMXb&nlignN)=7bL@
z%_x9A@?GkqqL|ee?Ea{?4+ef&eaO;eV!mL@>*LsND=x#tz1EjwyGP91%6i9ax7&Nv
zhwt60`d~A<v(m!q`k*qL6KNOGhS{3<KCOX|>pZTlTFYiGE_vE&95AjB989d%;edNR
zY<1bX=!i-<l-2utJe6LuhWwI{zaDFh35qUJ3$6NAsZZRWp}D@Y<CJJ+R;#7{S5!YZ
zKK+0DUu(@JE=mf4r!aC!m~HhgH<PZzr~30GguA0Y`~@+BMwwHXpS|9qb{#p`I%C42
z{y~HyVMi>9G%P<_cYRRft(3EZN6FPgu}wg2zC1FZ#T<;@p7gXFsg_i(=a4#R$XI^%
zXIvpi>{#{*egEFZV=vK2*ztUWu5Hjx<Yzb&5TJgtz7X+^c+^Z(xTBl?NkQvWe$#VA
zC0I_^#}&-AgUDcK^7S@UYe#I-$3}Lb;+mfWFpT3g?d!XHkIi@!Jmx8olyifrK=3M-
z`nhUFuSizJ$Ny8C_iyk@=(U9a)p<s|OZl=SE3vD2x;CpKAj`QJU8TB1K?~P%NZIG$
zL==c{-65sUBlppb1bZX==grtkgihG%TX_y;%{wCZ+}qKcYOTRk%8kyJ5@`m84q@wX
zM#JvpPLS;b`V-%|7p0$*d-X0j{U!{`-6V>l=6?OJ^&Fb#|C9f*2f1aWgw_rAFekVJ
zTz(XTx=pL7@?2ix-^<+EwxJNH?Q(2gwxN)rvP@nTbNL1*Ll^#W4#Fb4!1uNyo=L{c
z409{<8TZ@eqIsORRT~PK#25JAyKMD_I3z=OeGl0he@CGZNa)%PK^NtM##YPM$W7ig
z8wvt%c&X89{R@>J=&`ySuEe9eF0wv)Nj*C(2~(7xVZ(+X(Z!#u(fj=P<G)&yzwnAy
z)|?<>td$PMU$%7}-rwD2t<k&mG}2@$*viG3wheJ$WT%EvYp-CFp43xfbY$10<j0v}
zZ#0oz>KuoWhSr&j@Q;!%_z6wOzG&_7W2!fg1Da1_{rdIR>l-MHwf>mt1>pP22CEO@
z8901Duz4fVLMqEEHkPn+N1`#YWARpjH#0WAAq9PjRWk5i;@`)&&hZU}q;AGLB>NHH
zPOgiS-jB)xfPxdbB9LR;+L-P)cFHF(fST;IPXG%O%#L@I&SN&26T)wbItM=BcHGST
zn|>;9+e|L*2S?=TGwZJwC*Q-!>qxVw?-0J&^Zr6r4s9W(rmI*t1nh=05srR#H9*$&
zjE05ddEOR%9OUK^4w_xt@azAsEABTwc28b+c0D?gF8XWk@Sl*YZalFTguF8_iHvgs
z2ZoL!^xTRj&^&2<$%8}5*lH`}MhF9d$}9A0rAFbBckz_VYj(%yeLyQ>cQ?fKO!C}A
zM=(`k7mguZ#jbiL-`IU$9%W~@2mZd)gv=27hw)AElmwe*VS69N_MB&Kqz!thoHH1%
z0nX1d{{@(HaQe$%Dr5$h_y4gKFU5K1s{4}nj}thp^j5INWI}(*NsrTZSz5lb)%i>j
z*L;bOxUGq2u0hp?E!f(Z5_91>fz|7B=sfD?o41<y2K7mS;73VKF9n4{{sx8X928Wr
zC}{&=53`aUX1Wk4QjNV7TAuwdJ&oA|cQG`s>7^hH2YM+~-)qe;6=K%TtCp97O+?uL
zP-f581c=l9rGVu4wPhnR1(ZhZFa28kpJ1x2!>4fcFg)wt3Q#bj^QFXy9#6#8ZTk}T
zgI#!@Sa-mKvo%Lw!s12i%`RFHjQXPi{s%~%xD~S#gFZpkkRFNGTztPM&ix{n@K`+4
z><XDjSU;Ai0CACVO}R>3q8g9ZVPb>qerm_7UUdr857~xZDn#fIr6Kc+qJr==n-jT)
z2Nbf*eSXTZl1ijq1YMmD90g1fpX>l$Kpy%fsz2NBQR%L!><$#CI7GKi`)jy++3;o@
zmXDve*_>Co4@<!yX7?S^-dr_^TbkvOKcD+Sra*W|IRh=c6cRy|miKBx@&~0LS<KBi
zg16o^-#na;RLIY+`#y{g5&2Se18acu7Kuar!>xTiq=LQWz?85JmZ!Mi-FB=Y$?G9o
z%2A-|b|-&yK(z11G<2&ddTN1FbuSTq7+w^e(EaQk<!P!6A0$A7T}0FkA*nr#o^%<0
z%_4~%e3Z|M7l5w3DYKXJ7Qe{hSoU%puEPVE=5R=RIe7@LMMUmqprR_Yt%x_0D5dcr
z18tz#%IpIREvW!(RgCZ%J~h~XUUvH9R{#~XLDw}e$GQE8Q=)UbHX4Z>>WUTcH7^Gt
z874O~?kV1@%6me2>*51>P?AXpsy1!)-ePf*ne<!36XSy9(8w#+_;QFTL#o(=yIfJf
zHYNPU4m51d+0OxminY94C@wl)$cz6?q4P0o^H!t#M@hG8?Jt+()3(Pt{5cxQb5PK|
z0fkbliPOO$#5!|4hz5#c+gCxlfZVTJ?i1^NIS3*?b=I59#k@wPrs;pVkia35aSv9j
zMBY~{JkggNcgey)QXGUJhwaZ@cFeyUz~Nxr2OuYL&zKz#n9;(a+d7f(VKM`^>}2#Q
zMUBV>6_#xx;rjzVk*8jcLnd{0QN^FGrlfmcmjz_u@iTPo@_72hCIhJ)Irz6^r;<P1
zR1R=pfTUvy5k7u$OqhwT#Ei?qU-HUqe$Ym0mbnC3zQ`c3R3`H=$!U(k6<erU$j@0<
z+vj%m@iSRe6>)BhzU7OPo4f3qFXXyE+Vvc42lCsE_)M}7?$$^m7wl#ZhEDQxyOo=?
z=df;t+x~KKa+A!8OR)w@^bIScXoYesPTQT|td84;6mBqWcVkE_8^zqH6zlgKx@Z4n
z_p|>}OyxoDj_#;(+8$P4D<xj>LbOM}8B51yLF9;qwT3`N@<!UShC)3SCkdHpu;u;>
zIxvqxdPV$F_VLTMvcb>2Tn@gXEBQ*V%C9#MbW=^C`g(yWt!XGYt%`1I8_JQR^7db?
zCb~g5DbY1r+}eal$(ri2PYwnwMl_e3&3E8Gq{JE;oTj-T4EF;6QV`A>qxE^OH1Q8~
zgVt0HNVH5jjbB@H^dLn3%)iyYAazT&miW!y%m1L&!6#F|khN7KklU#ogmc#J?+0aa
z>By<H5Hh&8A@~V-aC_W$XZD}dXQlSVGj3PiuS)N1-T4f<d=WZB>L2U<&Ai0`oCj(}
zrS&%iX+^RC9|s%a5*R*=KWndzL}x4qOo>}y$8rEp6wD^(oKNjoJkxMW8m{b=<4-(O
z-AxHl$Fu*DUQiGh6W{7SM7Wa;&SA=^C@M`cXFmGcv!f9PJP~s!ax8O2e^*Wz<sftA
zK7fRtPUzTh1r6e!Z)bAU5?V<RfBxRz(+js~Qw@b=KhtT``L+926|2llwdE>R_@rHS
zHfJd7JYz{`KbL(E<5I}B&Sx9Quvqgc$nZin3hr#8IG6{n<`Llf|Hc-5N`vI(=&@_r
zyku+f;vd(exvIfq1IiYv8bOD054~9p467)7<0Rg4e1n8-wA(%evIS5sRULSU&<uNe
zCuc(V0wu2PZq*@gaTFo{q}@yO`TY?YG459<kju#)_-NFUM}rdZP6HwJV}FB>QiaH%
zDRr;C60`!c^C`<E_m*X^l<jHXWy@a)ZegT{QnDPX6|WRhv2kaMt$fALHUtaB3iTvG
zPN~PHY*oHtY<b!!Ued#8R=-ln3)0aKw&oS*hR(4?U)WmzwK&->r^*_K;Dz;8`$~{N
ze0QxbfK{Y+@#Wyd`fMOwV#6zOoVN9#*+TQ1@DCH*M8aQd${wUt16+9*HOI5_h*BsP
z@l>Lqy#vrqM)mBHjD;!+(3U+&YUM7)h0)p*hJ&yS9nnA@on&w<bZ~FNLDr2!vVC?&
zqe2v7EB3`R0$$2i%p$ROSM8Ld-i>*^Ioc*Rq(8yiisclaz(M~&HSAy@x=O0nksMti
z<zh9*s*#k{&22)*V$)Lj^hx0!<byFDt>sjtbkf*h?D#8D+Yzm^vfihc`-$8}LTd1+
zjV*?(U=W;mVlMLexJW9$Q#qH7WZF3G*nyUouzr6bq=SCElE#IBkB;;YK*y${uM)xl
zjZpQ$>Fi5v?MyY8d>%^>jLil(aK|t}a|vp<if7K}Bq1u|B;@zH5Xio}!>-2bVrt-+
z{3PqK<H#TXWY?l&g%(}5>#qcfCxWw&oI(CGV<c|o;V9;N8OYAF7P7$IO8D7HyZy>z
z`JZSd@$L-kwf~f^d@1++Fji*p;7+Wa>agC;d2-)Ud^&;S&O%2LC-4*4{09MwBnnp^
z;c&+x_}U)&H<GG`2knusL+xKX&0h6GOG*rYSmwW!mJDLFJl?6K;f$?#)d`E6dc=xQ
zE-s!G8=_CgR=rwC!T@=NZFRgCK&rZ=q+@GdEyQU1CR_Vzkb*t9nbbs6b?geri&mS!
zK~YGfT>bhUdv*P*K^eNbOQ1|c1<Dl;n|P#IYs@dmX7-IW`EV>K7}y-Iq{aW*T3!uq
z0qj@GHVAsFH&A+gsBha7(3Rp7pl5ru;H+TH=<pV-up%W&WP<H|H4%rBP5suHqYzBd
zKX6p1Vqbs(1xInvoeT0&50Z+%!VP7RsxtlA)nW25!Da+Jiii3;2cA^4Hj+(#4`+`b
zz3&4$%&+WNG%u_g*r~L<<FCdU3{TTxJN|093x+A_c)|xKXL*rYysLM0Cr$X@f7aip
z{BJ}l>OuO)|FUJbok_@%V0&ybCpFZw-==cVQsbJsCN^Dx`Mu5f-vkPMXtVxz$;&b4
zT^t-@cWgd4UOFh|x$x?*{~IbO2s4Q`ouSbc1#@>bM;;#Ownay~_uK4ReBenoA(r?P
z`+>dgbJ$2M#tk2SP=4Mw1H}KJ&s#aSm+CjXZntyE!2yi<ORvW5x&xz87khXoVMXGF
zPq4fByN8SrVBtLns-KJB`z^&QzPATCO3xqy`a}QK$6bR8<58j__tlpGuClfEnqTnP
zeqZy|a0Xrb{dZPF3drTJ#c3lAxvhAu9MaJe8~0ksA;Z1M0ITxXM+o({`ZYiFo-S1X
zlG7)isoP6*PE~eW8_$pkg43V1n%4?C^>F9bdZ+BpC6qFC(O9=W8VZc_C>-Kh=IT<;
z+=l8aglTJhEf!b0Tt$vG`J@k1=`lWbpAFy3mRN`|XmEGW2`!+Xt*-^*%Ym_kKDOoH
zL<jzj%SXDM_H2c=f>mPkI(*GIQ1iCezgSdVUZOrrMx@TyoWgUhX8159QHeuUJ?#hE
z=ij%=2-zJ|L!-ykdVOn%j=?0~|60%@ii-6r=<cWu`bv_I+(r^C1wR$SH&y}fb$E~;
z>{#wKk;OKipOi2~ISw56mD_`LHa=z-;UJ|lnzoZwhj=YB1DQD{yhAQB&cmr_AVZR~
zgHv|;wN!q=JV5Kt#PaI~mGG$4KfW=OJ~&Q7Q~3?mf|T}hI(Jz9W}?9}3Awa)T<REa
zE7(Agf;ty$&R1s-v|zs4f-1;_2^6g+u)+Yu=Y@&mO-V)?&r0F2z{bUZzOc_aKS&u)
zQ?6A#V0@y784kqt1j;!8nNRIT)vH&h<fz!qT$9u`7pA>%3V{yi{~<T;+u0x17Nz2p
zwPh*$Lu!yR3Rfw+ljo!O_1}#avITG@fW-KpI?DCwnBRXbE+cv44B3PDSX21S9%fg3
zMzl=Tlu+Gnv&QOn<l^E+AVX=RlNr$fG47X&O+$Rl^2T5&(9g};ifE@;7U>orOIIu7
zjZVvKp(5z?n8<gwsxcT1e(A2o$H-PUhPF{QMcudXCpe%*OF@B`AX|LV*82Fxd-Jf>
zG#0|RW0%!>TcUHw2!@G#7*B}0MnAFAMTnA#vy?$)fdwX5!RnvrU)nnHkPZLzOvI}(
zo;g~%>u*zIp?<wE56z8**pgp*m$fwh`hV!Q5?iKlEcqZft(NWk%g_+C`WK^=g0Z&f
zTn12-(fY#L^BHX9w_u?(109c_mAD33()W5lN2DU^^uJLolRmjI`R|f?U_U`N)w1<c
z7NxiPHA)C`|Krwfo3X*hJnbY_LjD->-#`D~n80xPv#|;m1ZiMa;MlQ1j8EBC;%0J$
z(*Ag~SKcheh^XSniN;@*SMo{o<^TF`I~gM=YtFPy_%CduDqwLcMsO~n82IdTLiCTX
z(I1=%WGLz8fI;Cy{<ojDNk^q^Z8jCC>-#1IhG8v|=X7H^V<l|}tjbKa3_mJo3`cj$
zW^*wIN4a}MjpqUkNdu@>^RN}Rj0A63hP?sajw%&*HIG7Mf_KIHqNCdf5W#D?Ie#9-
zTdLcxS0fTc(<2t*Mz!2CC0@8B8H#7;L8eN8sUZ7~b9p<@a=A;DAwt5VZq%j5U@bP`
zj&sL<>@PJP?k4D)qzt>4<No6-NS{m6_wyOn2Z^q7FMklvSX9`Ta#ZYLz=Ad&+wG%l
z!7nZrprBsu>p|d@v;br8>*dt#SFWMuuP0FwW~_KUu-}9HA!jS|$>6|8gv=9{y6W|E
zeBZ#EVYPpZo%nx3Dc8L2)JaiNI<9D5Y^@K4PdJ(m|8T>r^;Vm`4#~k}h;iy(kA>PO
z{c8403KLx7vtM+J=<Q0Ox5fZlim}hFDOwcQdJdH`+&4!Hf(`JRZFxOdq|Gwf5$fOi
zdYn_0mV<?9d%cjuzFupO&K?&|`_?)Vy4`r>?TrSdNFDP;Ibxj&0gT%m5Rh-n_BjN^
zOv=%_`*pt;+z~(--bgK4gyG)!*7c53%}^62o{4^Q?9o*crRP9w7oE2eUs%*_yKF3%
zi=vvipXF5H9F(Nd#{D%}L;Csfcs4#HzWjv$VuXvRf=&jEg`NYQYEC&>oyc~WNWZsJ
zIoonOC{y8Ab~*vt@^5w~@eNi?I*okQb<n1AB@g4zHT`;CpU6y@WobK;`?7y@MCs>e
zy}e8_LpGP~OL^8}1DcQb#lX7|*rrSUYQPAWR!$VY+qPKE;uH4lTD8VX8{+!F>yD#G
zaKyinla!3l@e>-|%x5SV<epAJD!Y{f6-}*c$!@>y*XNVYV}&5FoZm3?QZ7)3l5pA`
zhijNfRP<VR^KkTx={|<7ycgZ+?k!c$Fkn4-Dc*Oe`{05f<od}_A?0W_N#espd-N!l
zL8vxo_J$wZ*z|v#XHLuBD5rxPXqNx|Xv^Oyr(@&4*otUfqFv~Z9F(^5ji4Io3umj|
zh>a~w3FnJ+$!|EJ(J=G2CI<~`pOUk#^%kTalg+#)f9>qSPpI~dAP@UV#!>f19A+$s
zMZ-{E_4XxJAYDpR*6@a(H=UqMJd^s*d6&lgyzdBsp5#Yq%Fh3lJE!@LLL7_js3l;=
z`&an_ViYag#lrN?T3a=%DP+C&H<D}OWcyZwBudxsoKHuzma*f4?T!A#+aYHc6;Nlw
zhF6L#$F?u0<_u!tak}3KcWG`(sttLtYp-w7voBbG%-rF85>EybTH+<BK^I+Xg`^&k
z3Tn(-Y~(DWq&ij&{tY~;8$8CVXA-KETo)B-j5xjc4J^3CfiIY`d^t~KYeh}47E(f=
z%;nyJpTb0Zny#0m#-$lyA9m`Ea?cEL?oVeg^1uSpukf!v<GmmqVbhw71~O}lWmf&_
zDQ}<<Y#d&_vp)1iGqDJqR0!>C0;9x0bH4XDk)uqh##BL4Pj9vG#$!Q2{6^K&ITm3t
zJ}6=##8ZhBujO|Bfir*ug<W?zNFHW4^3)Wk5fnIfGlzu8%B;eo+pX+DA%GOhw%geV
zo=GG~>9m#{5KfNrbKHp*;@TA<Bys0%LQY^DB<5budgrbVyPwk_Wci{!@HN0X(P9tt
zZ+u7|#YU+_*yZ22-`Sg?C4qlFV$0qPVsL~kYrSNNvE^@;v%9r#h-F=ouT`tFPd+Xy
z-wdH<Q95@Wjh}4Qo3YS`2R`gW*Ah)!11fwyYioQ|7WxYa^V&D_I>$$goveAY5UibK
zOn$SVQP*If)nzwj>7@Q3&qzEgR_M5p8?v2PF0e@{yc!)MX+aU=t*L4vR}x|TtTpG<
z;C6ZjmBOecS}Bp^94A8Blh%Nn0!Scfo5Q(-tRvo5dXpm*ZYyf$j_P}*HzQF1R{p)d
znq3FQ1$0IO-7UncXkNB2o)rv)PsgHd-To0=FKg3V^(yO~j?9$xzgegrmG!Wzcw@4!
zLxxA*bTV2z0zf7&a;N-4W4Y}Z>0h>ExgCY=9e)C@?;#Q}F?QTH0wunk$n6~*ke%sd
zv@WT5nC2Dq&J!_Ba0C%08hFf#R(sk`=Z-@s<)nNj+Q_Zv`!cIpn~Z1YNz^_=$CARQ
z@|hy13V-~QO*`=G*Z=9SAJ^GTw1{+K(B&&Oo67+1DTGrMnuLclHkT8}j7_adpQQ4q
zFT@Lp=`(gUr;JsYQ?i)QZ!Yzg?An{bFo+`rQl)nG6#NKLa4y%YJ=wn9Zuk$3Q5@x)
zzH<XdIZC&35Zw9kaJU^1XyV`5QvBk&L6EppEtg!voLJv=1b(afkKKz_7=rOE^s+(s
z<5@BzO2z%aA%e!iQ$DPQmBYv*hp83uK{{z`3V8!OIcv+BeA<<gEn|k6yiyr2Z;E5R
zA8DJfS2Q^`x_l`*PX_D~)VH#!9P=0c*leqs3NEA1z>~cx7?ELov9w;OMw(Jg`J}jL
zC^^E~Y=-^^A}dd@nrK$24acYi@-_wIA>S|Af$DrVI4Jg0yrjPRMCuY3Y8!Giav^>L
zxxU0|HRh~_|7}z{ho&4oa%h3^%|3^p{~IcUmYB3d;tE60huYdysNCjl))vbuV)eDP
z$Dle;7a^$%5$k9Qx=d6N>Cq7b_eQ_+5tcjWWAj{R^ro?<Y+qAFj7!oH2aTe{L)IGu
zi7<ax4#ZjOkB`wcFxSCcp*@ruwUKHy9L=YotYguhayOilW4Sgh5Gxj9-1l`49R2Z}
zK{xvMlFRNybf@s|??6wL-8&i2NO&ofEqUc9d<L=)=@LHWGeCPyI%QwN9oS|94juc?
zY_h3bTBHSs2u}G2-9r>dpY|6Uhlb!gW)c9o)H2u=Zj{-aqiUkYjWQQY?G6Mgrc!JV
z=vL-^enSI?Y$2EGL!~RZ+VrcJ^MJ)YQYwxJ;TC;#(2v!>mgv~u2V0Cw>-9vWqR7M|
zj@wwaVe%y@!Oa|nr1TWJjs6G<AGX`M_U%0bXftgoW>Od~9A>yt?2bPJ<MD0$5T8jd
z!@Jq#ASuf2UbQAy<ZL5v7rP%gbEJE~9#rV8<bqF#n)c9vStLZM%N|8DKk71*epOE4
ztl3*Buy~}hyfT-?GqDfK8Sxd;?DDsq)Nl%rI9B*CZZ&RjTbX^JO;8=FPl+d570>*f
z$vNV(t*+WlF+MkjBKzKo^Pm_K0%EcWI3R|Y%so~ULo44ZiHhDT1PNXUit$#E5Q^+6
zXv^x~auVp$zmh|G^_x_9HslBJ1rX2?YmD(^7@cLjA5ic#B}gS{gEjk4jDQ4Q5#N_n
zKiJ<hX07qPlxp%#kWX`4J_{LEMQ7XJ3ff!h$zbHkHgcRPxDGRt<9<jTj04sgO}O%k
z^AQxq_W85I*C=zT{L#9jom=CfS?pHtTS1@j-1sxcn`pcQ04;(_st(4Yvcp_pBX9Xt
zApYe7JBU{fK*?!)ET;=&{h=L85RtA?w2k}Al2R6PkM9gAgc|Kc1(b3PyZY*+w<bna
z`o)O=fV3p2BTwbVK*Vv*B{U{y(0dfSv<}o6AG???e|o_teKdKB*;I~(KZisn`65qe
z2THIhW4p~H@c2CA^E3NaC02JUr#VB+ML%MS@Bu6SuJhS9+b~yHoNsL5t#Utd&h2Wh
z>wZ#b^p4nKG?%R$gH(yTx|VCGkVd;66PNn<yxpkU>&7UFf9z&HE1Z$|qglJ<L&A#K
zZU6I6<m6q-N$%&)geB;<J8u<I%Ob)H-t{$-0o(4qRj#*9+wFdAw#<bJZYp_PIBB!?
zAa`M?2L~W|s3+tH(d^lygbO{x1(xR%IB0u21P)TTNV5A9fn(X*PAhPa#ffhD+vS9>
zQ^|4k?JN8xjz**0NGtPmWrv1O`4*C}>h0K=&Alj=DL!(<R{KX#2vT*~n)paY>C3=T
z%2&Q!=1pxI^fWcmKn`~DO{bcQ)xQ1f3;!uSzpMaOmp#Ems;sUY#Ore?DtlT(%vH4J
zk~O~Vi`-^S{#=KTU1@$hjz1CAN|)V|UA&0mxApDd!;1Gx@evA2-<F?YpH&opyDz26
z&8WIvN40ofX@<x)+xvDQk94qLo$=m%?0^rtyf5HIZ^VV!9WA(w!p=Euz41&;qg<-x
zsofvX6ty)|m@)D&2iPKgnTJ_i10Et(>#q=`T;v-|(0A?cf|We>b`Yo%xT$c{@%#u;
z4)#MXly>~>a?Zq=pUBtfNb({)8P7`EBxWp=)kGrCcCsUo{G03PR6fEfiQ*RO?eyF3
zhW1YpY<A}D;1g2$rvJCe7^g&Af&^dVbUBsqppnuYQyj?>9@uet^vb1m#^+51=3+{m
ztv-{e>o(Up-;$C7mf|o~i|<f%N&kT#YCQXl`(LRIzrZW*YCPkD#X7Dqq|a?J@ddI!
z+;hOL`TBjluEwtCBA^FFo)-Iaqv}E0Og~09vj;VUS-X|nCJh6*&;q+88ZY_Zyx4Dd
z0vDc0ei_W#T|W^!w%fh8Qz{d#xj9Q(ab17PDRw`HLeI_x@W3JIBch`eCx2M2%a<&;
zKL5@h`7u(Z8_dV=1m(j~(w4mwa*3Q(p)2I-u;uTRqw%jjUA?yAokF5Y7U!e3GFnrZ
zM4IRlnz$;SDb0eMHKdze?O)N~95VZws>N)XkOD>4<}=i$3hCirtBGfXP@~EdjU(Ut
zI|Z|ckl*Xx@pbWVuX1BBJy~bOA$?}`KCj{)Vh!&U0#o~pHF~e`j*8x=<Q9s#rWmP6
zpgOu-FwO6zeiz>cZbEDEp;cb3^_^h84wclrZN5N@lY3gM{T(MtQYBJ<n&lSa$^O;x
zPLO9Yt)8TI?>j->dVBf@Im?~ZIw2L>zIQ^yS_MPVq&xa|j?`^U#hPQi36$>SL)Py=
z^-+?{24ks;gAm~S1mW?&olK47>MP=c(?wk>8_W5l{>O)wY|jqt&}dlq$E!sncd_OO
zi+#N6=n`zTdpw_TK(MUf10V0TPeR785U(eye!edyWo7L=<=?<@;koAf_z|u)J$5D+
zkz*>YE7HTHFB%Mc&;AjciZ5{zwCQR&@iDcTgbhN8ItV5OjY>RIY%_QRJX1Co&y=I^
z3dkUB@qB&@ii@+Kg}q+L^+vbL)UNu#KmR*NaxvGg7a2&oMXvcAa4|rrxb8o;AC~jl
zjdy~elUCU?!rn}zI7~eskn>h{hL(ncY)9>O4y-gky|$Ed;IJ-m;O5{y@5D2Rli-D+
z{><*?v)FzjjO|{IHQa;J<dEGD>_Vx*;njMC--SJh=_&5_VXiowJv!NHZKkz<_@$pS
zYYy&4N;giyMUMTl=1{tz2RmoWn*;TcUsU#LMRTD-m=SP(S=n4Dj$A=)RdYG<8(w<F
zR#&}Vc0VwP#2Wuj$%c>E+U9b~@#Cmzj%8I`IFah4R?GqSW3YjduWK%67%gXU)2eUw
za|`VdKc+ho{8iEJhUTO*iokkOD^|{+HG0QK2C&y*&9)|==i=l>2|CT$ue?*}QPxcV
zt|i;;SL|JD0vuWPNDA*WQ<Av>YmaVRM<wA<G2Ml%_*&ujLzV!a6SeL7hT!))ed#C$
zZ~J^Ozz2D)I~Gz-0!n#s4pF8zhda{sq4h_D^YGeYFMs3=3`Q4<#rf7oaxcIzg}uqo
zW-MofGIoO<^A5*~OZ<33ro{P(JULsE8^~t~ZEUV`!C0KVpeYh3u;CQV(a9KH(F@Rw
zYIH>|74h9pMGNwy(c*NpC{jRC3eDJ=Sj42A=aDpu3DG8f58adna!fS`t#%g#X_=;T
z2B>=BL}wh;HgC1rd_lSsnEkn&2(}lg%YWh?mG)I7FqGwOKEQ;KKxG#ypudN1TumH>
z-}(?Y-^FSY5FC2Z!e48SdrA3WMM17t9e$1saU(w|nzZ6^X*XluQo72Hf1zk6l(+Mi
z?~#fSpm|9QAhE42Ia(D9+#R1QN#O6sSdzgAj`8PP`|ZAeN(o?ujt9{QtpL2nL+=QO
zuYa>gfslx?w$Z5p`Ok<AShleo@{fA<+wzS;2(;)cHvU?ihn1g|%oxxSQiT+Mg_Yh6
zi#=IuR&7jS>MKcR#oDks29QaS;1?g@1;mE6*?3XhKW>5uIW-##;n>|~wHt$QkOhi_
zax6=oe>yz+q18v9(<KP7!Cw&8ZH*g)njIb;R=WA7YzT)oZ_U{!_P3H$(^>)sF`?<D
z)0(ZIPbg{F{x+Xu7wN2#VPYwyYS6VLQPP*9BRUhAUtmIVHs8B3nT;__@P~W2M>?a~
z1wNId+AM0qeH(wRDbCp?S(zk`AJu{MwQQbRJ<fp8ox>DK!c{)K8v_})%8JGy{n2Qe
z4U;Ysv$G8b<R$K<oC4wU_voZXa@i!@!D5W%Z+K|s={Q#Hp)R%Yjlp#t>>s5L)bWkw
zlKA;=9HSGyQgC2sQ&0N7>>xAw#BX4hYV@p~dZN=G?R2zo({IUlXlLS?`x+J%mrYhZ
zDJi+jEmIr)W?&bw;%NCrvlh_hi9j#dbWB{oq~!Xsc;!ginyE&>4UkoqEqOLtYXfxy
z$+GQ~cdi;6a|8=0vOgcQ&`G>x3qEB@#aE*-`d!f5udqc&?`~3%<U(!g>9^~-ds0G6
z8GoZ1NW8nJ^}FfIj8HRRw{j;PEl^r0eoIA7Zs*>RpNFgmOFrPvEq2Et>!E_hUGItf
zydT=V+-dG-yie?Yf&y1p3Y$GhP$WmQ8ec;mAI3AC8ftnW>YlSlKGIPr%BCa&=rte$
z^66d{&!m}?N0ee#mv3^KQpiL>QKgoUo9~Kft^Oq%DoAkUrsP`wh9P{_rb4Hls(Z>-
zZ~D#V7YAA=P-1srOg0s4C5wTsJX^b|U@LfyRVSfB69itMhnct9guc{ml7@ynjNXOH
z;Ps=nsE;O6EQ<`Pv|2+pQJq-g?>40hs4DK+R80y$A*Neo8wUoy74z_>Pyp>8CF(zB
zOEw@l+_v~7U%F4NJvst5_&nj~H`d`D!F`>zy_<ptmbz?k9|?<Mbg&=5z3<Dh{sq6Q
zZvPhD6xQH;(LVt5kw4xAyg_e8T0-h!`rnI_wg0D$J+T-{lBoQu$8tm*^GAnECdu*G
zY^R#~_mkM{;|cDNRDcKVL`+ud+p~5ux-UgKST$)zCn{9<uNciwZOV&+@6m}9PM1%_
z)RZ8lGubC`6=!X7Q@Lb5e$J+<J)!alz$;7pw7)$(U^6-Y-KgYdtES?}{p`;L>V=&F
zeG@0-2@jw7_{J%@WBo!bg!A;S`9T?gpnKH^9z0}=K5!QW1fUeriO80Ggh@<l*K>vk
z$rD>r_}=hOHd8K$%xHc(Ho*Md^5>)g7~OBTbEyvyK!E_Xl)qK1oISqd-wl8WcYTXT
z<+yOqAJ^1$+Wi2gvJygFAH?D+;`XdPj9;X@;#fS2*~SX}@?Q%n)ET$UDVeAs_xWG`
zh6K87bHUt!!K1E=UcT8iBGeY5%C^5;oRr4=06LH80IzAnaODs?XRG3w2t8qgVqtCd
zW+xoVD)X(a@h<Vt<`}Gv=F${OqEMRinyOnB!E<_4>s>5PwxQ#)x@e&}iDJ#l1AcQM
zGNi~ZHMSZacMJ(s!|$!pM_!zKm*s8pA0^+CHT%a3IwU!twIslWBWsOcpAqu#jkRt5
zwNXmoCu&3fubsVr`_|e6FnOH*1QyF%wPSPe4pfyY%Ei6jNP@g9IfD{NJ0H&o0`9JJ
z`{J4Ar}Qim(7SUko0WXidZP)ruTum4x@`S1BT0Ba*`Tk)#*XdOlo-i1IzYnNF>gou
zijkfkJ5~X}ZNsV~fZ2Ej!1u+tiqh4N$1|b8Sw@t0;Y0-mG+z6R=J;ej<Ip3yaCx4$
z2_K$R+RXS=E<B6$jh(KVgK0D;f2^H}XHrn0v;YIYv&qdtjo$4-N1#N9Q#mJ?KbwwE
z#MLRQgWL^fvg^f5WsNR@rnC7BQA0s0Uvuc^e4o6Yg<N;$H~(6ga{1xNYvU}RYuXF8
zkYnzFa$SwS-0#Sl^2K;YbAhf*tai;OjGT)>*{(<D<plUG&Z=Xtup1R@L_sm2^i4PY
zTgCXJp}V!Y+|Dh;<lW9wD@w(YTgu<k&w?O#a!~UB9<sasVH@`Edx=P>y||?A$DqZl
z3B*-I(t`w~%-`|vVIuRNai5h#*`sQva*9$towH^wDf65sGte^swK&-->;CeVxP}P^
zu@x<4BAc<3t!#06g=$j%kzC)9o0h^$P#Y4DR{y(en55!syhjd`Cbc%cWTgr}sfd}H
zc;@;IX9jA0uyrriT3t&VNrJULloVL+?GfG-XDjCwV*VY{fwFRqEy0r?>KjtHfAmlM
zG?KeGg)7$V{Y#W;i3N}z7$>xBjMnjMZ752jHgCLZH@;}m_#B3|1?z~$@+Sr93TfNh
z5<&5mg47eqcX}sB$zZTAK6AZEvD{gA^-R`E?caIpZ3)KdAPHHyrI=@bj884$U;bu;
z@#|BJ2zagLLaA7-7JJNJlm1TZ@39!6Bx_>R<Q_SmpHR@Y6j49klE(-;C1>Hwq94*7
zb)o`L(Ocyfx05lxECm;B!uLQW3du=Sb;O~eIc23em8x%Wx`JvHR5=sv@@UQZ5v1K@
z1+UVqpk)9(r{b9h7&xj-JF*p5>KX4FR$+*GO0!jKbR5iB{L)-JE3g2OTyf`f6_A6m
z%;9h1dquplUCnox+KT~d(Pw$2sDZD=0$|_b#=RcDM35<jm4|1TqH8xD$fw?JdeLrq
z?@z6J!EWa&AvR25ileF&Yb8-?Dd3a?hBE2NSL9B%E{Z@W>LOa7+ufXL@ltU0(Y@G?
zE{Z1FeIK8KKYQ#!G?tHnJnFOr4;`j0|2Arm{7)Pr&9?YC7hCb~1-#3)Y>U$iad1D`
z^6J+-x95tg{ohLle*U$UTM7*!;WXFzs%TK4W>)x2+UhNFOORZ|&E9LgQ&L|lYvEe&
zfV>+Ak6O)^$M`9IIrl^G*|qWJJhwTROU8{KK^kFhoezp}CCYTcFx6L(3eR=)5nHwd
zp}{{_?oigaC2So<SdbMS8rqcLx~~-qg61tDW1#vz!0(pmTW91leN(#6S_3+9x6-Z1
zNGBE>Klz;O4yn=Gw-hSM%s#h{s-G{p<^9C%_0}KBgO~&&IScL`)&Z62_qNaBD>bV<
z;;}pXf$5vpE^_dDeNl^L)cSJ*%5I0Id$3wcrG$9G|J+7=c%JFbA9G+;T-lDrn`wM+
zzO(UMq^+6E@A#H-dkkPZIT7$k?<~z98c#bJ&-_?AFEFvCknynI+No%wW%~)V6@z-(
zM+NRt>QbF?corvjZ?VZOK|4{O783?}o^zTbA3;j-e6Kp36Ly1)KiNz?lgEe1@C@wE
zY&^qMTb$43TruYZ7vCMy4MhvN46(29Gq{yhy`!fVD*_f=^%=po69I{PN&YNpTkM)I
z8upV>%F{|acKtClV3JnlQ@atH<2R_DNe3xBznQIBQMuPCVZ*K1mQOKPp0nH0LbxQK
zpzqnLr7cAO{Oe!sPKBc~#eK%xasfhk?#6endvYDbz35r_{C}|f-pl6g_P{rqq~rF`
z|Ln&%@KLlr|BrL*wrwpm1e1VlWy`k4`RN-T9_9Qj-<niMb=Znw&{kB>6p`U^RV(A!
z1x~zJ&ccOc&-=UfAGOu_3*G-qwq|RgL<2Hr*joRwI7x*btJzvekUKMRdGuOs_Fam1
z1s&O1R*H0#_1W*AM-LN)7j4Um1Fy2VHg0uvhtxEAV=xx|qcwXcn+L2VA!y^8Y28|I
zeYe;d#oWa=D90fL?G7HuCUYZ-SL@hX$OJKN9k$np>V-CSdgt2)Y@c`DRpjCB_;H9_
z0oI$V&5cjKK<nR{N?CjsmD_<jG}VNzKzLlkKa$TNl`<ZmWs+k)Tzjh>i?>qapB3AH
z@@PEffJYL_^y5_vS#v$N-wB6YnR-vgYegt|zUx_=aIj01pq;83C<(C)py%y$PG$H5
zxO8|4|A0~lxExRB9+47Zz^1CtFUctY67q>nZ_T^JvusLV>N`DC0ZU~QTEk!2tOMJP
zGv8di_8Zu_kokCq(--i$V6l+!9qA%<$47QGLHc<DhI!GUg<1{lw`-5#!~XLb64;bo
zk7q^Y31pJ;Zsf+dY_XgEHv!b=?N-&6@|Lu{3}1b_0!guP(^#)A+R_u_l-w5RCGI#N
zlyyf!yBiH8%19JcV8Fe2_9-A}@6htw{lrPdvB7R0vIo&hNi5;`im>_67fpGc9%U!+
z%8L5BfZL00^OKJES)n*8=DjSQ$(}7~Oio(M{TXyqH7te;C%j@?In6j%g7_=9#nr6}
zx1;z2uG$s{q;{RH-d4`y?w&zgv#k&wj)i<?*5*){5>a`gJp9&byju#3aS_%gd}7JC
zuch0q%LW}JAgbTyC?rhCeAdOdXARqat@}@Us9B?Tt>oh$Skty(K8RQt#THg9mdy@7
z%knL<N=twjga==gOV+wA%N>^(c3h;2dr(_0&;hEx%GIqsr-o`ORXZK|OsBh?@$<HK
zTabrdIF3@)*g69UO_~fP+k$#;9qLoSS$8Z+SuMxpI90vTz4QxHV1T@=`lBy7Ly_ry
zTL^B4$+1EnME?NYOW$xNJ|4@7?1}oP$GiuQk<w0#`@XB1jve3T%J3&Vq`A@2_HdKR
zdGW-yLeCHP*~#3~9T>+a5=^4ZgzY*3r}9}zbLTsJIw666;c`m)62~+7&QDg0KnRgo
zK+!xHj9HtC311+0*57S9ekt4AjLqcq51_duz&8giYJA=I=WVWfMi$rt9BG%UXBSX<
zczU>nc*bocJqkSjtX)l@ibrs<MTk@*;4VgMc?KzeAI@rat?J=xN$T(~pV{?o!K(KW
z4ppv-8&#_rlzW=!MZ1}kV$?aiRkcvwGTG@b+HIexJVg|Bu#_k?LQpR%gFA_HUE~|*
z#v?-hZuF9IZkLfg?)}p<lH=KZ-_sBJ?Llr&H<|JkNB=O<fwNOoHAMm8(HTU`*dyP*
z9pJFF`mugO%nTe})>>!@+Cf|1>cmTvh951$|BCp~C3~j|MXhZsqqPSKdZO`Ft#MX+
zj^G<Ha71*Cj#szFi9#q>pzE4yAYHxgzqGcs;O_6YSWRm=GZC3&wf^G0Emqf>9PkiP
z6UB42-rFiKpdyYMT4Os&BINb38e0odRsJv}ooqc)gpKBEZl2l3&5M$$TPgq*`z@Yj
z!CG5`0KeK`ZSiG^UWFvMr?ONA?X9VNItMr35$#Zbizby)B?=A1j8m<i)NIZYb|&=U
zAaT_8MJvS?aCQ75hr4cv1A{F3lz5-iS~ivboFx8l=m!ULAb99$KoH9|lHI~<pJ(ur
z1U^5bZZo)btTjmu>GdK6RMrW_HK{<|c+5<Ye^MCvZzU4Y`SIvnT0DuZDsFM2TCZYe
z6a@<OJ6W}sl@kv~+yo|C<L-lM^!F&$QZ7u1k(WN>v@bJCbp_L&N!W!2-8ShMvvspg
zIpU$|@GMc&F<K;qMe8tYGyZIGvQ!E3%Vjp&c^)q3hvstsH}AIj3TkCbmdJ9>7X1BD
zdHi2ZgvX?)Ac)2M_V+vOTC4@D1v?OzN>MjmPh8!^iM^3C9HFF+-Skr`=e}Dpw)98S
zjI994+pXnlE$Q7kU>rK6kR=}!FJt`fs|iu_j%Aib>TWy}gD3G*{L8(Zn7q~tf&bR-
z`*w6wXp_tgsO%o(>^6J-golCipMl?>*dw1sDZyj5y-=57rm$>#s#(5;+gcvKC{Tey
ze8u)Sx5PbvG%y6kT<ISU4ENe9Z;G&R^Z>MEdvX!@jS#6(Sf@47n_S7KZ0&Xzi*N%u
zU^O}HFaM78Si3#$V+^-cAg$Z(6wj@1T%w%w`grEzs0RZ!L?;yLW#JmD7EU+Fkz-BU
z3v~d`x~w@`E2QqCwN&5g5^D8d7@J3X`mHU#`T!hhcc=*w=sIK_-t0h6LFm2JfS2Jk
zJ?BAZ&g9)v26|sKK@i3WCvlkiSGlT0^$>dt6WQzI;WW3$`r`{(d#=za|Jq={ihzJJ
zcEnr%{NGs9F^7(73asg|?S-Dh^#2`R{j7~g!{b~!$17M>2v<h>OLiijRr+v*sgv6a
zz2CCSCMuwI+o|en*QQf5z)t5(Nhn05$eCOxr3{&j_F?^i1;O_{p2><SldAZEX&>zn
zDFJ-6-{Zz@vyNm3dnia#4Gaw{l<cg{=SUw@ZfwDSphP4`+tnvpi;XTUN3J%y>Vc0$
zq^@oEOIHdZ(Rc-RmpE3t?w!ePwcBpwFe4aoZ&tvUO7RMXvs>QaAX$*`agPS_vXKt$
zTUh+1c;=4PTt{%Rcj8&bT}kbK*E>3d(@&S(OL!w%aM<n#yhUUzI3$Wp=}~*&i-x6+
zkYC??LT>wxD1PCQFQWV!?A^F*yU+zKd)LoaxfOi&w)|aR5n<c+kQ}*sx)8GB-Ew~Z
z6}yhDj4#FHNmxaeu~qM;JQR1r>URtNyx3`L-Yqod<BvbKweQC99{EUim74rXW)<AU
zYQ5ohJnZUX2oGj0es5O)u3vfL=+2Aw1&e!ut4awT8sZy(mnK%1xixy1-J=JrDIrAc
zM9X{Fn*H4eT*56e@1hzkIDk>SsJL1yK&VNHp<D>uRt<EeSR+xET6^|RU}y1l?szve
zPQ<JH#rFDGUBft_<qJv#_uIaAgJ)8r0<Tun?F~v$8CCIOFgQb)KIdlpeQlq8)&>J`
z-43x1apjD7FTy}5z}UON;kEU29U8@+8@&eOBK5wF=WhI->)?<bPh=UP_Qi>4Uzj5h
zLmK>(KEl9(qY8eTsCv}3r)m_J>{P%&9eAwgr=x|$`30b)5}2LIL3Xeqli7glPp)5|
z6Dm5DfJ^G<ZkrAy5ci9O6HXr<4TEvPX5NjP3eR{~M$h_U?P#$%$Kjv<2cXTzSi*i(
zesFG6wvgar-1^?GKAupa*tp`_Vmwo6du4f-IC{;ub|YCyufJO?asSsp{l{+PEMWlb
zrnmnyxBji%d?Af$AY!p~7|5cg_|l~quW;{WMX<^EEOZaq-P|%6G7j3kXs+Y{3S0Sv
z>;8Ur*w*(!k3EPbn2-e~T1%XeRYoOPA7u|k>~t%Iz2}5NS1s0Xq{7c^*?WQAWHIZt
z<<X*8s96?~q!r$xK=GZe%s$xE&&zT{GUA!Gpty14e!V)L>2CNDEc$}6*1Q+E%Mp={
zcx|*1yGvh%HLHoyJ|m8fAm8|<xbv^9E`CAIu(Q`!qy8ix44AwN*6?1CCFm}<N23o|
z9Drn5Y<jN{FtiQUoa3-&=dI;ECzn10)iBN=^VVpmTBYZ$?Y*G8TX*fY_WbcL%G#bk
z_TZbpH$Kt<`OZ4u`_+TaNsT0r*+WJA-B(S%RAnfY3%V<y=HYXDt0k1@<1`ce-1>6~
z<@L|SIv9ZaO(G{)&JymIU)zY!6|ZU=doM`(VU&q|T!hEoi?dZ!BXYVP&wk<c`snv~
zeo~^*_~e9lx1pilPQI6FUlm&{IaMaA$$0d*(zTz;F0t{b2;1rR%EhBFTsz}`_jEJ#
zWK3Fne8Hx2VBFYPr(v7+Eq%JwbvtJ>zRs$qM4`J11%!AZJ;bxm&?REIEiRJt6?jVH
z>iWfn1dpV=au>Rq-?=meCxM<do)t!rdqn)_HHSx;@ujo69y5^=`G)_!IQiNucJsY*
zYaTdax2nPA2kV#7ZEyMWX|Q@JXN)5$o(_X{Cy^LmQjqv=zL)+|y2pF|@%r_(&)a=p
z4kh42T*z%zKSh9enBV>Z@kp7>9#wsV3nON2aU%D~O#pjJn^OqvfjZqK25Na*a`MB&
zG+EK+M<s&8(R>Qyq_tleAEz+`F9=&zJ;S?`4kZq)ZYzg~1In_mi3XwpasvGkW!Bmp
z5m!^}gEeghaiI3Z%lY43a!RiYNZiUO-f)V`NLh61eFT)ghxc1UTVQ`z_lLzVt1;i=
zCuvR5iL8h48e$8epb8*qj-ix)sU@H+-50jvJ8O++q;NnlsNpYJTRfAbe*uJ)r$u|4
zb2|vHFk$7kEQ${DNEXJVqs=8ZJOuappY`>xT4!6(<!uTG*w<FDY3p|D_7RjEMS(y?
zA=~S1;go(rG1%{LbFB>6V8A0CoUXZxa8e`rep8=~`GB(A9LuJKhj!iNcr_gv{T$rm
zxts^^%u)t{6W&J`sh>|K(A}dLNhh+U3$(*_sv5#w4XH_Gryn=MsdnI~ory+LD1Iu3
zMx@rsL<rbjaAqnukl0?E&i2Lp<NH6;mb`;j=n&@6S#PyC37#t_sraHf?}kEN{Fu)n
zNR35myoDU11?TRoF@%zvod-LVacVJ#h~)a+b}e>+dzIVidUl}EZ)9WSkRi(Ob=%Df
z?MOdWxSv!O+^Rlg-*5X~{ro?<(U;;AWi?jhvpfNHU(MK^oLtFjf~9W|t!K_>chT3~
z%b)B8Hp=_HXMG%x2f1hH@d)(K?NAaOt@Wb>hY)<T9YG2Pl}o{v?I<`3LA?x{wR}f8
zrlfs%-Bx5nnWsOnl{?Cb?<<<bRXdUkAsUT&NUOU#+e5jK&DQKFN20#|HCyYSZW|ha
z<Eq(Fj`GF>(Co5nwGM%lj=1>NWiS1u;A6c5v{TuL8+HUnz<Q$6WltKtFFwUFQW((W
zBf)^--rnq8ZRvNlM@xR*-9rkAem1amM^Mhe(V}>2i&?ocqx@^_IRs@5x+p1Z9SQD-
zT?Y=>-dw)IrFQOcx|%onjP869CEJ%n^&LJuiv77ee;-7KV>5eWXxCHeUx+|F15FAO
zF9Y77Pmm;ZP_&U9g`ynnvatlJRsM>{eChYYB8_{Kk&o<n_J;98awHKub|T=_iTfNx
zlp4V)#L0w@K)??eY9f~b^pqQoor+H-6Gl`mN^m$nI~_}+s7o9}6S9LiXWL{h7hWh7
z!mavHCP7^R?r|_}(-p81CZx)}U^DTI46;msR2T4avk4}?hJ_sX@^EE<_VY0qDm!km
zg&3^FY>UreZ+xy-tHCZ|{%{s{G1qwqR;6oIgRg!Wr-3fO>)GI7*M7UP<JWPlec`I|
z|J(oB%>=ZBL%E6VmLtPcALDM8E2uk!c}P@V+7YVja%;M&N{Ap$x0L&Le4dzLIB$28
zJPaPSd%4t#YQ7&ma|0Kp_jgdw2k{K=Nbc5fNp)cUL?VVq)qZ%6a_2#m1J-s1g&Kwd
zBK&`6AtX3t+wz@$zC<!TafHMd*&9~ujH?!f@`$bUU&YpN(N^sYUbjfCkP19Qb_PEo
z(<0drY|YL>&=i4R+#rg5IG>gHA9yh-N&J~7t_gOwHsL)4jAK@peQUTw;9-5fAJ{KL
zc|(G;k=(vR3Qx4gfS1S)wxnrip|m@odHvQL1L9i2UEy8;E#7>%un*Sycu|6Rilc08
zkKYuT?y-kMd-SSe+;U-`!?&O#o_U<LE1LH1be^T%I&)TvaUVF0wZ(zJ&0h>eclEI}
zg2d|M0@dp+O5H<hR74B8sio82pl|I~R4*d|tK2lXpOKbB-EU*w4RPwqT5)V=xuGy=
z4cmD9{Q-X;uV(bMoIR<1dd^Pxm)!Pqn3qoG3_QXEB}wj!Dz?DcB!G-9HdplJbhTH9
z6?Lj~)n}qp@D530a%Ws2+_RKtv8nu^yXQckIDMF#oEh=_UvdsS+(&^@Gd>~)ZW1-Z
zyQ}W{Y|eIn3ET~s9N{IxoU{3OCO@o;%uPul7h())8L(TGh;9*e6ekJbfv(Y#)2RAd
z1}Bf@yOs-qG7kUp>yCdozoT2f;cNW!f54;MbZq#5pu6a$<OmY}V2{0>pAB~%uqB@c
zm4JH?s_ytc%i;HK!lOB3*}WG%N#ijESyoQT`;U7sI1e9g4{{ZYvxeUlPQD*jGg9sv
zfA*;AL7A8o<Vkj;U2*1RFqGuEY*#QlJ2u+#U9peDlHd;6iu{d;C40llUC!{xX-;Nr
zspwYia@I+U_>--UUxaItCu7a7a)vi!N%+9lRv-VwjyZ%i3C+QQ10)aZwpwqbV&#WM
z5t~JU>bw`k08U7~4}gQ1q`Zb*$;M%d$5%CBLFF=SP5#|S&`(N9_>VvIQX1J>@-rpD
zCTYm7AdUDnjmkJQY;6gL^zB`qVy8X7Is`{a)y#xltT1?Pdvmp9J0UK((;JlWOR|rB
z`GpdkXnnhLwV)I;*6Ta*GogppAB*K$Z}*NL%yk&Rk$oh8$WAtvKiYZ@*|A;arZ_CV
zCwh@jwS3qVVR$^DgmF{w(ur(?_r*8jemKu3cNKznsLv+itsgy4W`8P2!<EiX`wt>R
zaW*;Qt<e+?4IJp{#?(f3cb^qh3%an&Z!R}5JeibjI^Os=xSZlHDu3_KZN{fF$^n|q
z4PbeXlH1IOK3r~yd0$Gb`Z)C2LV%st4XP&C)mUtTb2wd#)vxntc&_b=cm<U<bZclq
zz3b7B?uJyq-w05PhX@qX`883Qo6-7Hcq36?%KqA|$IoE3F2Hll+3k3y$s#vEp+2*v
zJcg8;y_5LpDT{QM-}O}>fsEbD9-W4wl_Y^;Jkzx@Q#Kk}k<U;_XKeDrgdI38N#^2y
zktzBIdlc(24L?LIUV9J%-14OKUDoagUmJ(`{0~{^mbaI~O_a=b^1inhvJx3eF}ba*
zzBmt8i>`n<u4+%lMGgS?kpY9#fz{Dk!nz0xP6M{4y`0K!FsG>Ia0*%59>*PnFS}gR
z?xaVN)}LE#{Nmb$Qr1;tiAnO67L)Xy)qA@Ecdu&jSrDhhiEZ@$J+a%5*3@23(gEbP
z=JrA*TTtX#i~rp{0G((}z%h16d-OTl_cmWV;_hAB-`-wM)6SjNk&7aA3Cs8i7xLcr
zkmo+^BfGM=tIn8+bfDkazT66&Rfb6&(w%E?h*0|;>&;zw9}`M{w(VEk6+WyLbfu6w
zi~Eg@WE;H6bUw#oG&c2@f45`p!3alesfWiAu>$I<F3z~OMjwT&XvgE*Y1}Gj?1ZmN
zDd~06@c{$8iiLCZClYMP4tnKN-ga^Fxwq_e{@u`EXZ&xt=^>kR1i-(SB*-4~H|0CB
zg%dHIb3@rG2H-5QCo|boOP|eF?<MJpu2CU3kfI<RyQQK^LW+5tPw*&5C;Ab~AfXfY
zl~4^9Z!=CcsmgpyJO65n?UhQLX)1A*uB>Z0ef$<-tKcT8mHI&rthwh}4BU>6+RX~G
zALVji@WW$ZZpA$CG=L2tF$M~~+rH^tAE+e8QgqEJ0J1v{*|tq~H$E&msLGj|w|n_a
zi8ypB*z)_{8K8qf53+$w7!VRNli9-x0##luMoxPa3*<tQj+asaJofdm_k&XqkKo78
zmc5_c1-Uk;6tn#OP`V(eU9c7J7joh{iItC=f~2Goc|Yx}r4TX)haB^G=zU7ps_DJj
zyTU~RcRE}1emSi5%5}3g0CD+C5vG@|CI+TdGzPTVXd!AS`5X6MtIP4F<45GHk2bh8
z7q*>bCf4wNQZuE2Pi|9a*7$xo@Eg&@RDDaSpk*QteD>YZJD_amEzw5zg}kzLNw-Gd
zeo+=DE=Eu+L)-hozd2^%1LEt3z{Z8&k?XKcnGN>7ALMd)U<jL0XLOqN+T_+PJpd^b
zt2b3yRd$opVch|oxVs<AVv~>P{h(NIJ*WWbxxZ>f2oZp)vf^MhEu}XsNN6KfyGyzt
zu_?{lSWdOzyp~9+w9mnaW6|2P%!%9l-o|}tw{5fIv5CUbIXe-*@C`yyeKNn~V)*;d
zN~J$#6S;6?))gtWQwbo<(G)<F1H?|pGvfBSt(9{6ik)$6B5JJ5CIi?~rX?`YZ?LIc
zYho8zk?H72rXZY2(M11XGmoF44+e4JmuI6xg<mU**!gOuES`(jGI7q*gOXRz$1^;N
zxVVLkgxKVkgZhW;YSo{V7jD0B+PCN{hXUl9_fqV+s#q+m+4bldItrB$OYoiDct6-;
zvdjfrd^4I&vq9o1RN?SeEd4a*Z=!2*nPCo*{HgDqpJ1MxDY$Xj?!+^B`pL)f2guOf
z_rpnxq*w#oA%%|Ki-yH&3Rw6u+5Pv+tvq%Z_txLwPVrp_D1q_NPX){g+Ji^Y3;J2C
zp%==++wK^S@DIgCeJF*<vfZf=@yrpne7B!*H_ViBY)pa6+g4;#Pr`@y+3q-O=ywm{
zfwsHQfNfN#$QNDvl^=c0Zl{cL!;$j@W=2~Z!+Z*LXRkQ}LP!l$6Rpp~Ur4MY4#jGt
zwWnP0D%IY)Zg(Mk0}_qu{qM!eKaW~N0_(Z?hOM#cmRtp_tGL&uc&2PQ=XBk(Db4v|
z&k?+m;Z0hqM)JqhIsM66cL!aE>qn}zw){wP5=da}@sV5x@Wa6VjGL=}gg@9U{$NwK
zH&^LMQEGJV4jL1tE(}MzlifrS+r8Vb4ry-44V~XoGmyKsKPN}X%1>aIyd(zwYgBGj
z02qk|x;y?ZxgCl><~=S>?$R`mC3c|Zk^aqd&1b)HJD1Wn`B|@Wi=K!U>EBEROFLN&
zrn9Z3{@f<A9i|V@{6b>IoZo)*4X2}9Wpn-Sb|&YC`;>xu@@uXNndX!Wcq+gCL(hJj
z_N|BSpkS8wgi1qtkJ$uQ!SFU09m$(hPP(9-&1VZ4#BCve!7GwCpc=Sv>EOw<E$;T4
zxg^8}A<#I{Q%OH0^P64s9w1iVprE{-{SlFYl{tUOrfWAHYBVK1b}LZ4pfx*G1rXTn
zfI&JQ(pbTnU9hDJu1_UB;0Qw^`dqU^1F8ggH^2a6;koZ6hC(ypM7r$0_pms*O&416
zLAM9Kr1&{N#U92GxOt&oL;=a+c;qNdnOnDJ+nz$0n3^ryQ}B}jntt58mhTB<mRg_)
z{wwyxk@={rzbIDx)0e4zx61$AzNf=h?+H?YI4pTd@e6C>S;?#iazos0?Vdtwe&1p>
zdwe(0k&!lB8Jv~NQ%;1Q+q!m75UFm$YYtnTcLp~vYaXG;$OHA&0Mpz^x;wtGhCSu%
zf#Mvm#yzP)B1Z5dC{0s#N=bU&H)p2@hdeI<{8P&wr=!kt>&nh-ajo%25raxu@z~9~
zZBHR)IG(myd#)#LCuE5Z5XF2dDnmj1dlPIp+pky`RjcZ(MwR$eZmWVZpV&U15DwgE
zPr9@7Vd@oFZ*~ONAdfGxgZ<tD@cxc5V9<AHL;Wi@l5;J_SWkI379EsGWKczZfX8yp
zSd*S^$|a3g)2_VqX)20;+<PZ!l%4Q}WovL29I=zJxlZwQQ$stz#~BXDcL?MY)jp6B
zV#rSU91$v!Y){7w$a{dd;F%abkJntuF)ua1WB|`CT2#*n6*d(e6*+D?IwCqxj_}c3
zXrdJsNt~@(d=06aOQ71Z!XCBxe7Pt-ZDCI;-io8RRAN`N{U(A1kJw^B`b3(rf|Ij$
zErynl&?Q!27P;v9<7XvQcmebFjXY{f7&-KT-OS%Ryf)Y^he%0IP*UEmCUlvLNw@R7
zEmbgIKp%Cd8so>J<Y9cf>nnt6{Lp~iiv!&8uXaDTefKCNf^B#ZAD{n4!5yM&)ArEY
zkMxY$qa+k$E>y%e$p<?8;Hcs^iXBYM*cIS#NJ-yO4*ibJwmjOnpFs({rlsPnR&>M>
z+Cy|;hZ7TR)mQYPN;9iEf-sfD#PTa!onO5Aimiz+ih`;fABqI~Gi=Rdf+&hIlBPvF
zUGEgp<HUSowH-kuMvus43MH<03_H+e^&N4Q_Z-3y^#8K={;^%3S$^-)3awCuR%nG*
zXoXgE%iZfqnOSFs%&=Q_mjowqrY7SHPU0j^;vZid8;tGPUxOhg<7BuMcg5Y(3RSd1
zl|xK0iIX@9Ax`1|nm7@vP=zW~p$b)~LKUh|g(_5`3RU;@e$F``oV{1NYX7=_+!+ny
z@AG+noafhh&hL{yp6ASNj;U~0b+fTgDDDcPPI8dqW#Ct3Yg@sx#K-T)!l0?Dv493u
z77m!s=XAC`V=W1iEE^S)t+lNnLqzahazt<QpC!2#0<3ml-wEU~(Xn`wk3l!Q<JRdN
z<GKa@c-S%F*y*X?;aZx0LhssOG)G86xGr@vXCrOpsTdyXvU=6Kd}fPG)FyH&3b@WO
z5rHyU)pM=iyRarxQHHxPxIoO}X*(QC=y%(*fXL9}tR0EkdY;9tE8F(bgdLeRg43EE
zYYTTQgyy}Y5FAd3A8$)u0D1Tz&sg0~B#;oTthfHPlfEq9qPMc?07OKhSR0&r%GI5<
znQB)E-kRJ0EPA84{Lr)4<`dc?K`5R<cFvb=aA1!uWYxzWv&HNKQHouv+Lege1*a}$
z84@r+MfeH+Nbu=jH%op!SA63NvGTQDjb93u^QB$OO&y+q(Y~I)$O>Jz8x9x2FzjZ2
z_?6@ax7tEFG%S?7F9d?OW9YM76;Vq{XLu(D!^;|zguA&dGINxb&Aq&-5E~pZO73R~
zBx(;d%a}dLp^0CCLELkG-_vhNu~Sw&SMWi%h`*K3IcDKNe)k&z5hez(HqQldB2Rma
z99Bn9j%h>d<2@}+Z+OGj`riY)iG;W=>cTK7<x=7B*GHKiWqBBj(cbV}h*s;?@LXVn
zqU-S=g&wiS08t2Uz6k2l(ri-&k|>F@phYrOL~T8ZpXj7X=f#&r+wxqXxo6-=F_A<Y
zlwfkT6x>6$3{B4!Y_iKE<264Q2yDOX4SASao(qPtSkse;LEyGVKSjRHXdqu}tNJF+
zC4I9L*8W`3UE^cKwYQF_qPRACWlB2WT=kKGUd8;aIXGa1IW%eg{%e5fi5#mzZ!axb
z&+x9HfrFff5ns_~+M2B%-8=o~A%7-v!+J<n;hhzC3=C$YwpI+?BSO3n>~NI1t6s!R
z^H$@dMe}wv!6lK06cop@PmIQ)Jt_!3MK~+I{R2CZ3s@2c(FhHZoy^Vu@sIz~rk^WM
zT1%JB<c>l6)Q9ouvstg$TT@WVIiKr5ZP?}$4rQXD!*kxpvrpMV1<@dBJr@%qc)m<}
zb1BNsFlULVING-4U5~>AUal5Fq-lAQ?23aijL9<TCbGTch@3)sHC*%d-F&|8pzM6k
zZshjM8xE03o>Z;}4+?$>^Q@5~zuiihAw^O8cn58yx`~k#XG8)Mr2)9>of6i#n91GC
zDoQv_r}L0v_oGaYj-+azK>Iz2GG`Gmb)hSJz8udI0h!~q;`y=<fAaTi<@1h2B%4gF
z@+z<F#vdDPbyRr|(1_k7GTzpB6;%4zN!4xb^U2Kjgf;KFXjBZaKod$^zy5jW>GkVt
z+wgo~g-3r%KJe!qSD}X}S6eP^Y!(-5qc`HI9Uk6ejrj{E3rjig0*52!w7CKkoB)_5
zE!Hw9v^=;$+t%m5Ur|C9Jkw@P@jb}oB3NpEz92{B7V=}8u@;}I!bcW_kPr}w1S>hO
zta~bNI%}`yQHid(-^w%E5km=MDrBfwnoi$__I9f!P*3i#{;0{G7DE?#Z4c%|Ffx(|
zS+w0qF5u%s*vQ4e%VHpF0U;|Uo{xvUIJ~0Q9<#|T!qTHgI0MMgsqCz)w$~2l?vCwO
z3h5)y7gXK~MRnBS#7rK$l3YAk)J!;Wyqcw~a9x{xB1Z!)U|FI?JXuw9IRhDGZMrH`
z;3LT&v7D=ha9poITC-919xM@LuKu;Tn7!|#QzCBqH=idHB{IRHgg<i2`XF<#^<T)}
z{o_d7k%;6<JPjVd6oaX=cR&DV;RTlB80(0gv&#uK@1c+@47pwLF0jT%@pCmRfFot2
zAy1ZTQFeyQDP@o%EMKqcy@Q&*h-vMHul&>CnwvRAse?f#3o-#Ow9hTyEoOtDW7KX(
zKdBWGn{X#U)~e3h-CPs0CiEiA9J71TMy!LBoS(v5-}gzlwEIDxh^Gd6(L2}+53z>S
z_^1wSMV84G9FmskMYi&Vf_A|1CGz8{7vjoeS!JtV2%IrSB$HxXwdRF_A2^Pfi`iPQ
zyc089WUAgduHE=7!D-1BYkdwF(O^ZjZ}0(0cv5^)8(s)Pr?`d?Pd=Iy1qN%Z;98AQ
z<^p;*N!^ryDi-eM7ZPERU?MwUdLVWr2%p+kuPk?&9&38R<uj5+NR=k_s@6i6RyeRZ
zA(BlTqKTGj^rD;MdY`h^C==>FM<3e3HU|O+|I^m)>x6iJ99z?ls<zxGMdr;~=L<pG
z>>wnfOj&ACNnlC6FSS^Il<5~jTZQSi!57NIuzSEpVj>Q_bjNQOP|n8@uKTP(Al&DP
z=&F!|9Lq0||4zQ(C|&lIBFS-Eivv5AJC9YO%$djSa0Ln(Oc+Mi0Y_daSf?Ka+R^NO
zNm+{&sp&&I=8M4MDmDA@7|apCav#zYIT%58#^H8$*~#b)^Mn#8^L-G=bk-UfCm090
z2W+OQ1+E38i>%mejCqQ{Lm-_PQzYp7z;O{K$ZzC)1+O$?<%^y3=^TIuUhvNdB(QAU
zSIp;Z(f9YYH{R&5OJ231<7Hc_j*ImDqjovB9+F_puH;u_=rZECTKy&rD&Y{X<uTTE
zxn2RKGwHF#Zd7|DA(iKGDzWM8W)9S?l%BWpJiRc$e&cg++dqp^rt3Sodi#_GQhJuV
zeh;CBiYxv-|G7jo<#FJ?AL}U1*C+Pig`m4hp<ALU%bL*IgPzu{+G8u);}Hv92t2;B
zJ;fbgJ3>TYyQ<wWh|6GCxBFex0ViMwxbIW$yi6U$v0~U3Jt&y3woX<<Yuf{3xIo9&
zwfiZS&A0r<eXA8AX}ym{*{2w>Z%6=<hz1kb5EW)Q&X}ttFz}im0E~!~h@vH_iZ%Mc
ze=q)Ho1(RZ!tdGU02NZdfGQc^miB^8J?j6~3K~V=YibXawdrTp98L9HiELCH{+8?x
zAEpx87yw$m2PCG7C*GE8gNKNW5|sC}J<9a>C=wkyRO@(HRtQjq-|mbue3M{2<)L)e
zYV8HF{OOOazv|gz6yXtRgY9tymbo3;NPEhzi17q*MfTvT<3%KM?)o$l8?<Z_p*zTo
zYLii0?v@LM<NwU2+6z*7_3yuGhkcERZvP8Al5>~8qa95MW%R;Uk=oaed2Oh|FYI`p
z2?)Eza=oRb5e;)9S~?yq5_~)98(a7qsqLn#I#`9#PVxL^y!S5dmdz&EjE+wLXyr?r
z%Qg`BdQ}uHY~ELNJ6fJxsQq1aTtYx`3t7T-ErM5Dj2=A7ng?#Xl(X&|=ZY;=3n|m3
z_r9z0)!5~{i%%j7>9;F+)Cp@CTHLGcAqCk@(&n)mdnGFGn)lz;ji8ATCH@Q2<pP4?
z8vzfZuYz82zP6jbIWRkRD?5~WuH-~ihqry11_r?}cYXk_^FJbAzOlR2RL>)!aHEQQ
zd95)3)ti1lfA7UL61@7r?^4$;ney0+L8fentz7Y9kUEIN*^Ux8Vr7&mmWQ}|(o5N@
z7vr?bK!_w)*wVh}?2d@8k8I70DPOBe$;b0=YhNt46sN8(TbIKWPSsD-djGyO{U?86
z8}jeRN30>?8$l?#u^Kl58(QOwfx@~K1cac`FQ&>!DIQ%V$`O)s4Px{Q+x%h>3wwHd
zOW^k{FP2N9aQRy+u#|cQL<*8^$`L(G9W!lyv0U!I`YUUBG0-+LVa`-*Ru$SN+gGo(
zIqWPF?b>r9KO;J2M}mQ3q$rg;9SnH~)N)EtbyD{A`)V9I#A`Vi)001@-ZiDuU2c6o
zlKno_%T7aB)?aj#?z93YN7AuE!cJsW1+_*cW0Ns`aaD>QAcKyns!TkIyFDGQmWi9m
zvJw&3jyU)~`il-bn&8V)7*_RIuF^<9!Vfzh=U|A-b;4V<^5~v?@%ul5QH?lDZt3AL
zySWfT7cSDYpEBj9Wx;2>tL=&?Je#Xq??>=jLG8ljPmod1=XCKr92bdju9`*(OOM#%
zBU{LWx9JzQ7$;ij<CI;hYRhb{Jg>Ht&|n<0&o1Y@p-&Wg<4UxUq$R1)tGO2L!*|V_
zJc+w@iRFGh2Y~_m5|l3n>_$~~3O}X7P}|MeRDYX_&ky^h_$IgFi%TM$k&`BzaMIZi
z7ZksLMz~;id;&#yMevgZj(y|g+{=>#-LE%XYF8X*0qMRs8zpTXAsHU{nqjsKtMpR2
zjRSps*rg*zTk%r4^tf@2*~*s!XJAO}QutLbCC>N=(Zg!5%VrYl@g+aC&Qd_a@dP`y
zxz^i|*qm!)>+)BhnA!C&B^fn>f17QH?iJhQBWs8*W!WaRM{!~;Eak?R9O+?zQt5FU
zUn*E(2Pw5T#pmLA%iV#>waw8r8dMB6wp0Tt2*=i}kIYIuvaFw)5;BZt@EmW>257<;
zt;K&qA?q<{jpi~MpJP`S;FjB>Oz2F8W<^7s^|;KKWbuW7200A82EpDsy~@&bH@3LL
zyGlUA+Dn1%Yj_{@`*vX^BKJQGZv7*K>_DH5#DaNhM^BXJ1XVr0$Vycw<3sE=3NvjJ
zF|qg1EJKszj`xwKrsP|kdMU84to5O$?Xa)&lZ4?vl5KG#BrncO1<Q8)T{~7yUQ$2z
z^*<hE;@8ANiD^6$P?cP=P(TWG>}Td=)P8TtgF-7i-gHhKX9wOAn~Bz9$Fyti#4Gd1
zSzi<k$2s70IVD1$!`CCKm=Ca}OedD3@4z^23{vecyyUMDTgK&;Ui)((dl(OYu^RQE
zVXjNiF6E(MGsbKw>PaOn>tBLK1EU09mvco+gj^uJQcd<#P<=6Uw5w63msW;$Owq1Y
zGr_6}O9)a4$4q2#LZYo|iDj)KDal86^9Ox<V7ptsQDekI2p}DK3m?g3PeuTDa;$Cy
z{Dm#OTTQ$Q));l~C9FlzA)m1ecHhS!G2VmxS{z<I<gmS*M2ljxJGZmq<&*=C;LB1(
zVJlxQxM*B?x>vpIm`BXgXSUjZElm&g@~W(PIk1SRu^w9+y(`)g=2x)Br)=HJ<r0yN
z>K(AovcAv+#V;Y<FRUScNpSa(ZH!+m4U|P=b}YYOHsusH51U>t_ronDL<DZG=2n6b
z61x!Hag+%+2$Ny;Wm}_6zMfyO1|LJ}H2H7}>o-a?>X!>f`_Zd!Sj)@hZZ*7Nt=YZg
z>ktx)=vCW%orb0GYxjvRO%vwMI&w{kJlRjYS#C||%Z~4P=yl+&7X6{2m0Mgf@$HW?
z=^a0Sg!mA0VKB-v7Kl#rh>yE8{p~-o`pcoV9aXRq7J}fuBN+GRjCP}dw8;dniEB9Z
zvcDL9LyG-G9H_%N7Kjpzz_Gk&M-on+0mvwXmG9b7AHPrc_L%=IBM&=X^^E;nxqEU+
zk>Ke>^y1PGbQCb7CzqAEiS)-d?SKy9Cxz3{%MWE{(w4{;D#$XppVKy*b5!6&oAX~}
zwUc0QJ~|fxCp4g)Kj$?-!|?jHkb5Lmn!DmJMkA-DBol+0i837_tXS#jGh3?aiIWf#
z_{=URFiLV}SMnG8KWtZhjWDz#YC`Va_MLXUS}zQ|082t&Ss7sGAE4^pn6sNv&qI5-
zld`8pyyfeKjUM>ywx3%vkl39ZWz(Z}H~%In=DmQt3XBW6bAQ<kr3?{c`5+HWzZA+H
z1sy!0L<cK6f+Mf>?iC9LebQ?yI|`1#;<RfYX@fh;ooRi>R(Ax)QRKK|w#GZ!hUn4O
zb|eLYD;}2n9R*K3^2jT15Sh(?EKTo0nX|zg;s3hB8af<*u^C`zJ!vzxF$Vz&;mTl*
z9f{9KV2N!!Z11M*ua10qbMzr|8Kq|V5S7W6DAPlsjkw6c-<ti(l@$)9DdE`ptTk8l
zF);^@kGT{r`Tfv8u-5!uY#&@^-xg)Mc6bdS_5IS?y^X?ySx2^!3<?;&ka@l@EQjJ0
zhoY8CwzCH{r1fWq%56piC0`oMDkSaWM2z^)osl}I-r*NY<TB!C%-^N_l);Vs1j%R-
zJt~MYA|a-Q^+~SXqmMP(R928}8x?d54@a5g<R6CU`$&gl&_^2VXf7jJ9mGo<^C@A<
zfRmsdkH*4+_zx65@<g7dQQ7L8%<-Q9uTE!YGT#&_(vie3d`mp`<)YfGuk5#nr3K0k
zUm#kj&1WA_GqRICmtT?i{FyB{<p0Z$Hris2EQ0w`HOsG&_ql`4Qm%#&r7#oX=gS>I
zx<j@Rri4AXk~c<%Qxa_JHMr`7a@(L#>{@j69$HSBd+PXIugVm##Z{CyqD&I5FR7>S
zVmEU-D3yqyWVfP@c2E8XsNwClm!bA2n0jc9?^F<6#8~fcM^KvzN!{Ch!0!1v$OZa-
zEOJpkmuc!L#ka~H<QI`HuGQm<6j}C4K^a_)8C&s6!2;i&fXMmhoU)a#I7Pv=0RC04
z_?0H56AErccK^gyzfz9RIA$elUMXl|J5P<g71q8IH2E$PaIbsi;rk!~hWk{?#K~!?
z_0hGgGn7~e!xY=#UAOHYwg!i>MH=Icuf&`9P`nkLG*+D`NJq&XZJS~S*j^#r+Z;9A
zI<tg#w=GcvjsX0<)d&6KLqD>nSK@pqIs$e<gVvk_An_}511?=lfTO%6A`Wn?VSeIT
zTeD-zu%*v>=#zv3<kRkRZ}(nN#P?$IRxi@_j#mORLMjaGl0}Ji#^8u-;SSvGrS?h+
zUH2mG*yqt>`g7rXi1#rVK)^1d=NidwAy|vIUd`$OmxQnKhjoK21QRxqosSF?btWX@
zli33jiO9`56@Y11b=)LHJ?z6WCwMkH5;Y3l%f?7&cQm1FLzZ{U_jjn9%E$e~K*?)<
z!ar=2qW<J7<+0k(VABZ%vgYBzeR?-$mI0xVE3yWgt;*!eDhs_YY_8grb8uJlxy56>
zB`OanCfeXURJQ2?cXA=iBotzaBr9C>!wwDEI|vPTDXU2D#qdJ#OSuNZ>2}$#78*Yy
z{@{RdCBKdh<7Ri&CykLL#MHHbQBjZIv+KD~+xc|Ee~{}CIl)a|oTcd(n(fvrffb0)
zUYGgO?d+c&*kgBc^qsx3g0Q;@Z(-s59Nvqv60Dm!d=7HBAE3x>XU-nva$%fC5C}|B
z?bU)3Ff|=V?pjoVw&K--P!x^FR=ygf!`@-scWhPu+{w1#f9$QLwL(MT%78Vm#;xn?
z+K1T^B+J@YA0i>Rlx%|}?>#mZfvBy2HHeqT`xN@ZHbe_#NW9DqRf|(a@=o{`+Zbg+
z*2*_6#dBkh%q1Kk;@g;Qs#<ESV)t&Ft6KHCjPYx>B}T(X9X-I-oMH*uZB6k}90EBn
zUX4ev*nkx2ALmKIkXvFXnbAoeAn&|ZZz~@#LTt5FAWPxn&suv<TmdKvIMCrs|Ls4f
z$IhtlNrzyMA)aap90nsOW&2+(H)`8Iw7~$ZoCK#9078*JMxq<&H{?j7pWd_jt1d_S
z)pnckKet0BPF9`E*}|q0wlPdJ;JOay+{+aaCz-}dM{)@GR%v0irAK2Xr%}uu%XRn_
z5en^i4uh-$?%_l%usmyFR2&>=U7L<xrTCJ817^xL6P580;eyU)zhk|<du+}Zqp8K_
z6VhPEWelMsg1_IZ<xN9L!(CkPv7R8z?P9go%Bec55$%$Hei9aADLy-fRPL?wcKOv1
zqVOJw_3|+JZc&YB(5~c;K;V5=36LdsxavS)u%x)szD_N&thye<NFgQ7q{0f`@L6qa
zu$vVSU-LS;x!|p;thmSGkn*wJe)W5gA6kg?(J_~5-dR>tW^~GRaM#Di>GD^0&u4q+
z(5rU;)j&mU`{fXhW~OQ@C9NXxUNn2nZ-So6Gld{nq}R`(GFH5njDrOB{+z6gGCd&p
zib8drvQ<?*r85VGK;W*fdKH0@dJ;v8Tobi(ZJ=Ro?Q4!$#KV7N>-^VFOd~l~>*Je*
zF_0p!m3<$B9wglKYvtDLY_W~6Ii@?n^=z!bQKo6P7v2<Qa!8Urn_h*@Q6`qPIBFxD
z*z#KPL2zHl6w!@s^{$tucRp!Nua)ZtC1A}C7I^@q7HX;bQq~{HQ~Gc9%7@Tdw7nL{
zw;x1GEa}(EzDU78;zWi|WN_R%eSTQ8uwAtr`pM_4Kf559gM5f%KsWl=q5aB6e869z
z|Eb4kZLw&HQ^<iQq6`ryT0-HEMVovra6E{@mwU<qfKNq%a=04hj3k2PHXU)`We)3(
zTt~eHVODzg+A&|2LN(Xycy$WCDeRdR$K{GAqGtsNmVd<pYVDIz<{RQ6L8o7f2L-NK
zTI-qD9^MH`y2p!hbT-Cuq*;hQ`jjMnF8b7k*FwlIcs>B;N?^g3YmEAD!M9-_c569o
z$+`zs;Z2ZrJ@d3J<*Fz`1pJq*)Go&<m7~x5b|qIq_QrO#npe?r$OFf&c?}YcaU|@z
z@7xQBJ#P3&gS`X>vzyDH=LQxZ^INY4KkAmN+*&<|XNzqNo(*=p8tsEZMv6wd^RORx
zE1>MIA1D%D6ei(bR7rO!#lpHDlTs|Kv-TirlpOU14S#6X8Rti2Y!T_L=!~P@Tpop$
zorztlsQ<+P;<T;mOzcvwX^}Rs?hH<4M{l34=`2WLX?la~c-D62L6E~RvJd~tq9}*Y
z`sfUiAjF~sK$x#}7M#$nsLu_ZfeA<8UJ=3A#%LfDmTy23a^`HzA><p*rQQ@nNOnq?
zg3faHcamm%OOE|pa-7=M&Vq%WBQ`;FTC!pDYU_1r&N8=N<0&|AE#B1#W3<NS;t(S2
z%js!zz)OaMcK;hLW5GJ2uFkS>Dp=V%vwK-?apvbRuc&2N5zNSb>#X(X+=eGg{**!Q
z7Zn=<kJuwtqG;tm7xPeb9`)P;yua!;k&D=aagjt=lUdM3#9~Deoyz(U&%<^&;7A_t
zGwA#yF|LG8N3%Hzaf)E$San#y&rHw*IqM`I&#uZrcg^LAC}YydF@yqf)k^@dw5r>5
z9;ZGYi<zuW3<rr}Fd((rtRU1i7o)p~x9FX9xxwV~*_5;PWjXmD*g0?7J+7#s3sqA$
zGtsNB6+CN;aa5$mx?q?57xa?M8<+eTJW8I)%YKHQLaiHPF&{Qm`1^gkns=(C>#=M3
z6W2(T4cBvPi^&%O=<{=ZbR25$rjLnrqqwtMKIRykb-NnrD`ur9A4HK`f;C-{0^SYy
z<xBsu-OKgT(Y+t-{W!X%+XFAdv=kj5ZgLvZy5jDYj4<fpS9Fz|q#Va9yUHC!$}?)K
zx*Q=CF-f_@i?+He-eX80IwOvmt;uRZu1RCOw#%g@$G~{UkPWS?>U@q&-rueDU4c%X
z8tp|u+7)POTh}NcHFSNy9`k@l{$KuM8y()I=}muTjnyEZN==C{@Jrj2gKWp1!#3w2
zm;`WLwsa*fOnPb&YDt!&Te|}9^$hn}lTQXW<Dnt(kImjEXJG*mS4)iQkDN;rTYW*F
zU=iDLvE?eQ)o=H%M+PNF?+6IwRfGYa1pMVV&s%3#L1Jy&tX56&=hh#C$u>vUKN3a^
zR?|KOFE6Ta(MEEjT{TShte#!8jN61)FHY0gofUEBWLLf#L_0#uIcif5r8A?4a|BT}
zTtR6Mj^w;rM~8Py@_RHZaSXmLh>xHpu~P7V+}h(&L&_S+Qu%n>iKu}Y5lq<0s*g_#
zVIoV}>4g2;Ih%>^BC@6Rf(Wu%|Bg`ZjLrGFA3y>yA62D0Rniz4x17sW8AGU!vbNfO
zavHGga?9{rvsl4_e?H^Ls_;^j5mttJ5}BN~rRYH_UJ)}q?S;%aWtX!y(q)&$pSCMi
zuablmY-(4dj5~`AUUce*cCE_^3jNZvLRJrYExrW2qhH(gTr)@;^6(`L1?v_EOb&tk
z25v^*sPd8QiirEhZh6_#G<Lu!frk_`^S19baZNd%ccKD;8}h$)Hvp0g=qF|BdfCpi
z&C;_h+P%CWgI(j|weMGhxN-$4p*?V*C79)xb(g!}igP#C$Sb-X(TGft*i$a&E4vFm
zKn*oI485?*t3-yp@_)9vyIfMGz1wGNx&x{84nZxzrmXc2{>mMFY+ZLjY{RmVTAy_Z
zTV7Yj$qfO3l!Xcc4hYuZAR@$4#_)~)i&cXXu}1&JA}j*xrf6Lhf}9{lCEr~2Sn~aG
zKF-*dZr|63mL8{KUtNB(HT&V3$>vgiL`^<NZY%K#o1+6Ec;N{}t<qzwZZ7Z|v_uE`
zBFjBe;#zBzk<_~QH+_h(s;%4cUQ*|7G$n<9dv~CBvI*|(RbDdd@VP4J5Tr?GcR@=}
zv|7!}k#FImZ~Y194wiG!EB#f2jdX_;c7(7fJQ$q016J=2Yy+u1UKEWJIdf#J#U}u=
zd9u4aO22xp&8EC_sr88wo10fVFoc;&j~(%fh=}cIP8NKlBPwUEW7&6+=Xu%jC?nib
z(ZGESo;l(3VcBG6c`{*?8Q|V-pq}mydKdI4Gq;(nfm=Rydz^~dSXLPhT`b#kfx}GB
zDx1r;@N_conD+^da6Pa~i)Om+BAXNy#l?;ber9otIB$y&r@^kf%;l1IyHimg2l@`!
zQf^6c?YjK%w8$QfV<{B%P>6k{>JcN|81ZT@E(<$4j5Wly+*kI=u6KXmD4RV;RvAv{
z0|~$INtUkNjh^+IQir7#@&ZLwBjEs8YAxE$YO!Q_suP=wbt_?j1+I6YFT0%u(gEl)
z-iiLi2AvY2#1h~2wUTwN-OF{ry#~!s)W&j1D*zeynLVgB`%|notfw664L`CKJq1e~
z5H+~6CpMf|>54SEswXf^t(TZ7oW#{V59y<V_qkeYdJ1A=n2*5>YkT54jY=$CIOBEM
z%P)ufhv0l{eYB8i#8<W<e_<bue6}YfVJweqawG$j3Ckx4sbp}qF>1>v<r8ZRcwEH=
zk%d}gQ%{gpin|N}<PAy$G3cM1Vc_Ad{u_#IzEEvV*<ny8Gl<xlqfAcAazrIiv$X^m
zLY#}%+T+;8pP~{ji8tdL<bG`q%qY=7q&97@x{@Lg_6dW}j)Zpz`$9B32_ag+QAp%k
z^j%E$S!CJ1`(5de&NzYP@)bR@@9X1xZN&Q-mJ*?hxaakr@^oNp%??aNQ>nR%I1{>i
zGRoW((PP(4Rq%?r*FL;whXbG@)dC;|YjhW$BR$Ko91Jp!J~^41M{qk9%QlM`;i4UP
zkh%^^lXt?;A-5XcvHuMvSF>qvvNVmFx6x)2aCiZ!R%F33>$S0$eBS2#?{5!7Cd@}&
z=_Ws7u@xchTrTzqmQAN^A?E^5$1KDyan8=Oaf?-BNo1u$nz2h!R(x%!EibtxU-19!
zIlJsb6CrTWu2f*o$?yr5c+sv#nc^-@aXb`L>snq7-E)cCu2&6(W5PT(%5<V+Ji)(q
zGs>hV)_EWfncb?&6v0vN<DA`&l^g-<-igWPW2^u$1^3$BT;o0kA-tF4At*qJ3DG)l
z_kANvs13<39`ux_N}?#Ml|SqKdu>H64uD6=l!ZynZ-&d$g!4q+>A6`|bCJc<wz^g@
zP{+#z`dU*94ykJbzoNCZKsO?LnUR9ktn=CfQaG)zCE4g<h2*2_ZZU4gy&>U|H*c4`
z*Bfeqo~j_Vw$X=VKR-o8-ss(vsA7k0N?=+@<YJqBp5q4(+LoA=EGdgQ9@xX*r>(Vk
z{UwYdZed;TcT<AVlRllzF)^7{K~feiX^W3c#G5f|tp%MY%CW5$SeX#tiZjq&t%_{;
zC6W8sI-)G!DFP~3r+1>fGX%QNMjRihl%zZ8ugYMcfI>FZg9+=%FszcGAIWYcr>W=X
zotcDcq88{D1joUcj6Ox!>PAWQHx*?%lBjgJwBOj_T-&}uOcjn~w>X;-8-ad~#?~%N
zyIiiY9$47{AFKKl-Q$t+k0+$hfO<|;72HktBeVo3t9^TDi|VY-bQ}U+*6)5}Gu5dr
zCO_bh&$)Nx%eL7(9@|=N&Szfh?E`<+Y~By_kN&=$i=kywfn@-^-}|-@Wy*uCb@_&M
zS**qLN>I@;5TsqI;FH_IGG@q@a=FM~jMtz;0Y?0}UCH&A9pQvs&BNmM19mNcV-f&v
zwd>hY@sqkyP4p8uJm=>8Y~0Ltl8dmMw-VNot^-A1aQkn;>Ni`m!#lYRBrvzTerx*3
zOM5RN-7k{R?)xM*5%kp_I8?<t2v(^#h;;~;cQ~s`-{p$Y74bz&vrx_h<RhmF=Bc-!
zldfS~)m!ck4huDu%DqmUC?Y9(pWfiQe*3=5-$+5QHo7DjCc>_Q?dVwH71#9!uGkB?
z-!m>9@OtkXB`l`p8=^PKrR0wIE=aziH!vV!0H^g|+nAGyQ-pjRkQIunvA5jo$Dytf
zN(|Yi-Z&0y6%JN)+w8z&Rv><3OMdPf94~R8w<a)8Kv^^eFp7*q#7y3^Ppr8r`vOy>
zV%}~oQFa>S&RbVf|F!x!C^AGFv}Fwryo8j*mbCWXf}+Rwax**f`_kF(-exMYq3iVT
zMUGO-QGfeIQ7vk!{R#U%EI!3J3}!)JSC4dDHWC}f8H6*b=Q0*+8`xc&h+g&lY0nj@
zeX<(q<05#L`^=O>0{%wHTtTW-pz>QjZAW~Z;gLd`9<6G8%xZXegJZqrJ$wA=C+&EA
zmR%icSB~S1orp5Ww4pT$;7P_tB&})RSM*Qrv_6x)loQ~xq*(_O<UoR;tly(Z_#`aO
zwE+cl*YF4I4HS#9F0bN3oG9mP3ZcfimCf&Bbv-H{D9*CXk1hoa#Y2}-Az^Ik;Y>I(
z`tq(_jxx~|TINr<HCL)KkwUs5AKTTKu{?*+BxMa4cp+mRHM^cbk03zUYd7++l9SnP
z=8Q3*6j^*Lx0@(c1Kq<lGWKMR;!!nbcXC@F`?=lq4cbfSsC!;4;}E-_1A4}Y2UVZl
zu<Me2T3=ENRkqf0kQ4?_Z1ak~xPNk(C5ncv?2Ct20u^Y&Rej|a6e4MLpW`UKX7BK(
zeOmJBevHTQnrzjF0E69In>GEd=jxx1$7Effqa$u6+E9cQ5OM4K0)Y(o^pFmd`?$eJ
zU77|@HuNR01*>5j`wGr$-f4|qUhLorqtX}W1q*51Z8rCn11){lmV~W@n;){R-UBN8
zroQk05JorAmLsp_|5|fj!K+wtk;$PYK$if>jbUc3H75EVh{O3?(Rzw@(iSUUo;}?o
zUN`q)(dj&6?a^|Xn;%PBqO9b^cx8L7Gr?DkGv3QuUtnlN^Ti#JOvecr#@+G(SK7?M
zgd7t}8;K<(%R2;k-PdNUUbBhpyjw1$LpB*bDS{8&r#MlY$|HhpG0zL4ne4i$7qN`*
zRL3hn`iUKlVPyLKfgMYz*!_38LYU;+@tj(pIO`Mn>uF8wq=Pkt&)sx174rov3=a9)
zW~%N!g)Q^fYSx<on9ap!Sv*ONrgWn7c_w~II5I5Ta^*6Za<45^onm1f6yjpl(-*}W
zUo7S&KQnEQ+fsZ504M8Nq02rGJ*DU(uH>kY!W`wRvCo27LPtd~^SgH~r@vEbnCrff
zkSLGYjocPwwF<m?Gx{hR6|DzX#|h|L-Uq)F#`bph$TjM>I}b+&RN1?{!^jFEb~iyJ
zo-_)>d%45pxGhm}?q@&5V%cR6@)sv=x7n`5C{h!aebB#UF+>Ti*p>W?#NQ%`-?f!d
zrr-=9f+dv4wrZCn6SqgvJ99oZVD+-v+Wi?eV9l;_?MM4<ZL}-)U&3XD4#AYf)<rvs
zW|dZpbXm53SKv1{j03YoC4nsK6TRdOQ6_Q2X*EQmvT;{IAf!aH#$AqpBy)JzHbrH<
zKIsUxIm&{WhAR5lw(JU|HiE3G<e1sID^Qfotp}~iyCt54*#GA27PeOL21LuWWEY5O
zE+XtBs!9$d$0MLuETOhtDZo;=FC3B`^2SFWwT@kZzP3XpcUEl+=To#QHLv@VEq`kL
zG0WMd&)`%CbKT$}Wd?2|2`vY7v(>A<T$QE7Koe0WB!=_eZ%H*`li6yg98IR8hU0#x
zZV%@iiPOngJXuGwO%dWBt=gQ|0V%Fb$D)h}-3d!!mK+_Gnd(Urf5Lkr4G(%iJDH1t
zQ>jh+C(a^<oi-C^+HaP7@^L`e2xkmtqw`{S#pcK*XU@x{a>YibM4q?#UH(`~ihs`k
z-aCv%??SXkPReP8Pvw1B%+mm~fv?J?ob?WfZd>x>{`BK^c~|EAkqtIpjWWy!tz4<P
zlvw{g&I>_gqpVyRK>w2W@>*UqUTcC`u{Sd@sq4FfZSRvQtN0s)I5(Dc3(k<T$!=CR
zPLz)5D69$fp4qLat>X$QiSs%*$S2Z9BzBnkJ5_&QabN)qP3GMwBZ&`NCBERi-K+NF
zBNTl1{aB86y>1U;@H5CcxvwSgm`Kx)&H598D3-FL1J8uFBECp@C4mxou4B$C`^yF$
zWEE|d|J|)1GpqaK4&jj{$B#ArNu^-e!Ia6=W^4O{4uRCWXzThN4Z1W#Xz+*HyY>AA
zVf}*++u-9soj-03(QlC|G8>|q*v2SRPYB1ceE85B`zur-DsGc^Hco)Js4P~-^#dQ;
zmSr6%aR6yYY-^OsiJZwJvH!}N`eVcKhm`fv=m7eS=9rZny2!U)b)^{iA{VfybNE62
zeb&|=2nCI?wR_V6$W??b9oZDIp$-~!vvqoPg3gf*u$KEJ-j!`<ru-{;V!t=lRZ_zC
z!fFrZ1in)Y4|X2`^hlJ6M3;L5p{1?vKwt}s?A<1^6<C1%09$-ulQFI^zO3`*gD*?O
z&uuElE#eGMn|XLy?GLGUk^Ofha3uQH-U)IURUVDn1tkCx$%mr$u{=>&KH<q?$9*Y?
zwca&iC!*QI!!uJ}()aCTe|Shr1qlxbn|u%gU^>SpdgE6hEtz`G<Xpjml4FRKne}s1
z4D=wd=CT<e<HiM9_L<G+f_2Md>0H#1q*%%<EURsy>KKn|T{#ZMqL(S6vb^YaNl137
zKY4e8TFRzxDH?LFxvQX2F;~mh6B*Dl$@!JM7mqz^SM%?kgud>tL=dB(B-w&q&ykQB
zfoaLp?2{-2HU7(aTZG9F+f-oaaT3RP(QapzF=^54j*s{4VUW$;>>P)uF<H9YO8_L0
za3|yY(IpE6|L@Yg2RRI`O`*^;8z}d7d*oAEF;K3H66w`#<v_tbI0c|&8VF7bhl<mz
ziqq|qPp<YR!{k>cJ;y+-@L2Dzax2yjI1ZA8@U*Rq?g<u)w%dmW%_t<YZ5T)lCihm?
zPc~F#UxE?Of=i)us<L82#eKYBjWNnu;JRp=qK1TlB@_*~aPvSCfpS^b&61mO37lI*
z=UbwUKcZreFt(}6WP#x+eVTj*NPBpDng<HW^~|$RT1%D_hQ?Z>VTsAhFyi{#a<;@e
z8ymOw{E5j2Mv5Ii0TL}K#&Bo;+PmY~$F1huFggJH-0wZhd4o{qXdMxG5Bks^gJ{H?
zA{3SbQaw8o>tz!IDONAhiL}L&0V&~@iXA`sgKVl=Dwj`uEJG2v0BLd*f$eba@wR>>
zVnk*?;-heIKVXlKMwdn6$4bix|CrYxkBFRQjt`^)0)gr-No7yuYRI~nK$Sk@nh1;O
zT-a@(p_vL?od;Qyd}^~%R^s24xAT<E4aCFIO8^j=VMsAP@4MB7^5k4hQBQfX!Y*&-
z!+d@V1M#|&e;Ki#EmmX5B+ykHmvR|KBo}5MmZFl#Nf#b~Onj<cdq=`|+^)o8VJ^s1
zbv3S4(R;Cgm|q7vzBW*v<EQXremH5aGFK0&(2c6>U5=(SY-jAIxA^l8yA@yM)GfrX
zyvMhF9l<v&>YZHFCrBB5H()?w#KTo&Wt>?e586Fnu}2^Msol@X@#YhKfFu2$J&1#;
z&~&!juL4B#irsO{l<XAI2IH;V?ROi5UaX05?y^-?S)r4P8R_cXj%6Z=4`)ZpcVU?|
zQM)(*M%miku}jd)@-<zzyWko!Ta4mHw?10LD_V>!H$<7j=15d1zqf{L)plS6SH;~0
z=aodyjX4q2M3FaO)9!*FM~3%vjyG2@DX0b_J~n2HS6G_<B^hP5?oM8g>^nO7O}oP=
z-AV8gu#04ezIH_4v6gJ%?z05bk1-UkT7j@+N8GkM?#~!nOa$I_YmZum&SZB)ZFWQ%
z^M|q!Bm0hZ?shC#NG?{<Bep0)oyXy+RWlNP7uE8i^(RPeksNj~m%aEeVioL5wlw`@
zyVdjWE)4p7T8}nvw#n>c1nITz%2W<i2%~aN?QoXq{T6VX+9SR>;2+6sj|TYV;xo-+
za8QT|P4X){R!vzs{Z8YHcg~JSnV5?+pd*QzPE_@zlNTZRg`G?gQK+HX0`cTd%BXxM
zmz?}AgN4AE&5~{`#gKaR+FaFZy3Zrf$<g)5XhEgj_B9$FI<VIka*!U}pBDW?iM4fU
zce(y8+iWTBM42Kcs&5rri+A9yFMFLIH`tZjlII6T6;|YG{(go?bl}%(0fy_23e{)V
zbLdfUz>TPoB)tM*-}JQ~7%Ab&Zh7^d{!tM0?W#Z0LozMdJKo6!d`gaONWkyrS$OQH
zb}x2L{hwv$U>WoXSN!Vz1ZEr9;=%6VCB{njC)*RZh(`=T7{Q$OIFV9f%B|d!WW!el
zHIt^<R_$>#A_bHPorlS9R!42(U2`cUOhHn!r=YGETc5PGdkO~Gc>tL(Y1^an8N?$}
zI(@|5Tfe8k2B%+`+6~@-F=V&UuqQCzaJNiPB_rIpr|jcv8?7+`Wd}t;$!KL$G#6v}
zfo;x-Dmtl{x-HR!1?EL@M7TBP=8Q4dRCIDdkxlXC3=D__3TNR}i;225`!K^`H6F8;
zgc9DohNy@0(dq+1R-jBqx=#=$YfET^tCX0c-IrZzwF&D8(8&5$XnD5OIv;|otq{pz
zwd}dLFS{R5|DM2Kz3CY=SdAr%F(}%RtWr>~6p?jb_yMHl3MDv^E$}aZdPJq|b6lE+
zJ>G9q**?5hkO3$3V`Ppxvd8gYpb*(CIsPtaXU>jRvm@v%OZh2?<XG&LtX602cr3Pv
zbsXeMdOsIco^U`}OQwPlCx7BoPrjCEKWslb1Wwsgo|>iUU;NZ&9Xchls@q&PsiRRu
z)8yZt#L)48oy#=_NxJxc(H5di7A!Lb)6dwVH^LH8d{0gX%H#tk`XOedOHrn?BT{Ms
zp5t<^(d**_cEuqb$5r}jmNyZG>{`qm>RQYehpn)Jh+VGdp6nzXZ7*EVjhx6cV}qo{
zy16H1785<fO1E;8!eSDO-z<@6ZTLVhluFJQZc)6jyngRyEqOU~b#of-1xzwvAxehc
zEzrx?F!cAUU7MF@QeF{V7~VXYibY6;K}RFPvYO8Y2&xrP2HDGV;$-~F!Jsh++E}+$
z{;TLDKeyF`iFTlQh2ECZjAl*rgdzqZh>y7T;fSC?L5`>NV60HYCte?Q^inH`81J=h
z7<?$l;qfZAVQ$P;UI=SQs1cuwG?aDq#)P_i7`0I2V8Nd}*xaj4{-d{q3EP}-7U4Fu
z)^}63Ww0F6;gNn^9Amyxrhi~fQQH$H7M{?Wa|T_b-H7U~Ww0REeYh^SdiAAg@VI;A
z5ty<9mMN4iyJ+n`hzwkbdepIOx-#E)x5<an`_3GYyv>SGTJ!mkwwc?~KbW#ebVUWN
zIockKPDL6k&DY08O!yH`v@5hma=lQBVwC4yA~N9Vo5&?9s(EY*5MxdbmKO?RwKM$h
z-C>*Z0>VSeEF0eAurH#5wb_yU81_zfG*(rfA&Yh_e(C%mCH96()5!cz45p}shmfE-
z@dlk740MVFW@v~!e>NS94ka*?;1;DXYHd!?Y_;g;i-eLW7&ezcVb)G6A`E4635e&0
zayyzR`I@|bB{yA2(3-|=aj-mdLj!gxp(#eO_~hG?*ZcOLLmpqwVTu-=g}D-UMmd8~
zzCfNy$t}nCtFcEe;w~yk?x09MrQSJh*AhS?x)0j*=qJzOqNt?uEZ)e|E>E(X)fH9X
z_u?msLyj_4-|}+=Mj0;tbhi^y*@BB?z2l4eXyec9Zmfo1aW@jYmn#HXRQ_N7wfloX
z$UHtGF6cp3<$M@apE29(IHSPI@vcHu*^0dd(`c93j+J`@GwAX_o2=R!L=kQ-7|{^e
ze{YPX#73jG##`&cz_PFPUuw&lSeJb%(g%8m^#OxtnN&7c3BJX3ZSYPX5B8;DuOmrO
zan2qr*v2S>h8L1t`UE1Wv8qS@JkZ?*kvdT()~?9V5k_uyK)Qx{Sz`{vmc0el@CFhC
zH-YRXg#>R{)82wex_c43#N4EzB@Dw}$A@AHAd(7cDN9f}Yf@sZxiC07aW~rXCpt{f
zUwia*7OLz$>&QXT8bfJVrw`iP3tn;9U}<qG*c%8G*53wW{37Y#Y!U*bj7+w~vW6o+
z++lFddA%>NA$1>nX&P-AUe~ceNU-%FL-M_Elese(a40G@@(#ZPHkEB3Cvv_W&RHO>
z#C0;i{Sr$RJGwWBQppD8jY(8FpVqD#p<Aj^Ngq{SM2u(lmdm_zhn@7V1!eciLk9nq
z>Fj;%b<p!nt{7rU<!GCY{-idR6i=sUE@q+YA~RU>6vUT#2V{7t%g$xPB5c>4e8(32
z^P`QMY%xC5M|mv?KWUd@)(SuIl`TaLNZukFA_Rn8_L)4j!>;_Gy$FB2CHIDu=oz`@
zuryLrUGu$_hP!Up6DZd83+|F^+iy5fO-SZ$27D+kZ#gXTA;xg(cHYiyJUjC2xs!`l
z7qYsWXRyX|an|ltqj-{<Zz#K;M@&x14`P`J|7Jt6n-XAin}K-6P`Ne49cZ?dL&*(&
zBy=iz&sGhUn==4htA`xh=zd73C!K?vI;|Nh*rrfEAKTjag%}70^17jdS}?DZYrU->
zN;0lSFgyge)tE9*EFOT?FcddrC)3?njVg()URw#J6e_hb$01JKZ){W4bIGCf6~%`Z
zDsFQ%1Zak~IAz!thiYkh8-|Lc9*Zh$EjH&O;4WH|cfT~fzn3?xIcpRE#3d0{&O1@`
z6LD*;xw^`w#T!deTY#uERQMtJTBF*Yw~p)&a#P;@(y?~>TK_Qt&T3v287B5F{x_PH
zx(#}j-l1_Dsd{i#XECbt&p#zS(?l+`)Ky4%CUeH5e}a{m%GQvqx>oW|J3Qp?@LoIO
zf9>3EN28uUp)$&ww`18BXI_|5{cFd4IFF{xKA))CNt#kr;C6B-k!|tXf?&D8G5WOF
zX~?2sI?meOE}O}<9Fbb6*hJ6f1mxvPbnN`OUp81aALB{Ntc_Q|w{yO5B_YznP&|2<
zZ1wBlFXkkmZ9W1F<(axP^!-m|Ayv#e|406{rR=RdWS-&ZvTuH0o#^Pjc4a6Ct$r9%
zcxM<RU*yS2yB5bp913Q2RbL-UqzW!PZ#Qx`VY_=c6F2=gxPO7;G-J1-jckt<;9BbH
z+e6>m@BGiq=6_`7|9{>8udjgJ37nyn+J4ZMrVmPoeJ>7;$X>{f@8nu_KWmUf6CaQ`
zm>9Rf2-<Mq<(}bwTQMBG?AQS)p_Rks83CE@vQ@+Wj1-?y2;|=&@mM_^ZwF~Uq!(3!
z9(_M$YqHZil9#p7RN`<%?1&`J+PdNLVo3f3U4?<R56s0tykZ-CVB|{8)-W7YERql?
z)s0b6**h;<V}2a#8AepUDL#sHJ7b%3+(KZ<z5|P<C@XSTw8_HwTeA!nx3K53GZ-!}
z<oGzziQ}_dZ#ezZ;^UTpdMr+>*FFdaYm3?-E7*ymYqj>A9B=Gt>+oN@hKF~-0d)FQ
zJ09tH#A-P=0D0f~z0+n^XRv}$SydI-;XNCP#V-JXi;056+Df5c&usz!qnP0Yr03a$
z55Vi!HHx8R^jZ?D&;wMsP5HjylZAYSBXBrClAvMHj%52H3_UuWN{5l&t`acdSk#s{
zdBKin?LEpIbYi%?LoL6ulUXja0-Mg|6=BHSn2F7J-KU?l+59S_DQW*93g#-fWil`I
z&ncVt7UTVPE@uFrElv1>A4T{ah_1y1d3#Cy!G2xJqxQ!Qwv=B73iSs@z8pg#rIDpC
z&l<Lo!yzeL7Oj7#=)j7!&8|8q&o<e$Xsw4!2G~;iUQh5O#^eRLF&q@9MA_qe-Msx~
z4(p}{T(<Y@RyO3_mx<Z!Y88qLHstG<cE^uQ<D-w+-MqW_fZX#Xfb)jtx$mn#fMW!|
z9z-vm{soUDWY#uP?j{EQ(4wC~O^!Grcm|pPJIIlT9G@OCxsb0KiIYImLPZNB=Zd#M
z<SDV<nvuXU&te$6HoBAk>OET*zhtJhVC$>S#U^@)uni*x3t<#M2+{a1Pd@4K6=8Ja
zh+`p_1eDyMTm+>yX!KzffVJB;B@D37wmE<F5sYNZNMM+;5(#i?_E!!h#$JWg0lQgK
zjPBAu><U0K%_9X-48b(FjD%<j+EM<}0}%zR)dwSlFwv0Pq7Oe&l1QGnb_bwjMj$a!
z$4F2$za+$kavyXCL_(lC@c4T%LwV2Cd~F8$*ya9eT|b9D9;{$>zY>w6Bhin}qr5`p
zrBlRbR<GJBAIU_nP75NYNgo2>>Nb^~77zWxmfB%ofRS3C9f{8p28is(<HL?-8H~qi
zuD4R~9LohC7#bShXUC(;8LYlU$_s~`h%yhQqkpk03Lv3Z5kgL94f3mt78R{IQ`I;J
z{+X?YzJNfQkhHJ}bFl<r#8c>WMg-%}kHkeBB@79iez8?K=N*#fSv>VZHg<-zXpCK^
zyy*3I5?UL<%OzjGp4u?mvgA-GECLT55(YN(a;zCo$CUt5mPREkDah<<u0K?zUGpE<
zS8`OZ2L#A8U}*|P$kb&HeZwJ8@HE8pH?!B`p&+BF({?Ls{8B-~Z~L}l)-_~z{K7vz
zs3-@!`t5Fv;-XdhM;9XZ`1kEz^vmhh(*j1r(9w4jZ8nN$sEerL!VlRtBhq@kV2~Q#
z)NqF@UN6{SM=^6*`FgO9`xPdXNIw2?X?h<4Rad`W-i)WW*_zkmP9S&UO4Mxa>yCz8
zi7WX&j$g0mMebL+)b;rj^Iv-2@OrtakN))M*6@0wjE7+vWmI?GHaeuz-hv$)vtB_3
zC9@8bR)+UYF}#)q%ML_5pV{Ud4@Mjuv*q=Ix8#IoTYVyrzi3T90J!ZO2evt<B_~XY
z6kEIzo{6a7d@0Z_69}#?JC~gw`E6r-*Y-JQ)R9ZVwn*S4cgxOd1`DWWe6gH_sl^v@
ze;*Zzrx2=yGySjUyFOk+NRAY4u<994UJ*4I9*!jFU3<lCV_D>&kM)k@laU{H2MU|?
z#ed|HjW(4#j^<VdOosyyH_Aot$$~l(IO1!?Ss2(iY)8EdTG_E`R-ze-3MjbU<F5xd
zst2HOj2@BlJm^S{+b61Zlie=6hfL|DcRbX)%cfrsB+xrSzz%t<z*Nro#N-uhv(dhU
z0HDtcti)XQi4mPb{YcI`?`6Dy*j3AR9^&Ji_tS;xku6k{{s?AI)ab`W4Dzlmz8>P7
zS{EU2N><oQxtiNBsFXY7QtnV2$uCNh@XNlap*<xqaliyB#t=o1;=6X$=ZPDlUGqO5
z--#`epE90E&gl&=AK1IgZsvj8UW6&PDzKC?O~DiGc7oM6u&X!++=)RYOZu2eUF3bg
zTLFR_uBOQ1+>1LPtCSDyzAwvAFUBqpa>tn${-mQxYJ6Uf-B)0s6{CK2-Cn0)Hm0W_
zsVhfg??}L*NZ)1eR*e=6vbV)nM`sdg>vo6%Uo+~bMGqFPGcu9^og!!jc%)XgHrsGc
zls;A`XkE04$5^zPYO4(x<V#y0ZAvm>+FXQ0*${0cZ}G58EDuOo2p@c64G!h2Kemnj
zSKE`;IGSWnvuzp;BOx@^GJEbDS8$rm)$+)KLfIAHu`RJYexmd^$x1+SpsijTo=SXF
zojh1b(0#O@{MeeaFHY~f;4>WOtR-8NAjtk8wnqO>LeM|+b`qj(wBUF*ue0{if^5NK
z)=_o+A+kOs);!*wqvZ*}43?C8wW`95j)>ANT7Q(uQbRTiIH22Lu1F7Zve7{Gy?eWP
z4eD8)ZJRHY%_lbD05M@SI5z3O3Jt$tQ`ISYm~Yh?iNm9T{E+61_Xs5)@xEo*T_>DM
z(Y+k4+7qmSO_Z>%AW}Q#6K}-C>3CF1*H`EoVI*jc6WJy9QpIKyEORnv_e`rz$8^Hz
z4MO=Hn~5?BOELb2t3z$)`}Y?OHs{mhZp!juel*mA#0Vv>1utfaW^^uB<lDc&N^-#=
zedd{GY|;Ph>h88n{^yghFH2sXXxwtixg6sw=+Fn{tY5GzzKu)MPw?Dc^$9$S542tL
zUUu~Ewd-Eo&uectvi{h}sNM8BkN)g2yXAGhJ;5&D9u54rkIZ0xiSJZX*EN+r`#W|w
znj9lw!aeWl@BaA5c0aaR=kqIjkf#K@RBSwL%<+jXoutmtzqTUET#c!?n<Pvai>uds
zz*da~eIg46kofAcU}d(0#n<FVg+Bl<udN2qjr6@<Hx?q5;XTmy>&FT{YCdQivMy#D
z9|dk{7z?_j&>QJhhPRz<{J|G@t>?jnA8X8hTu52OBAcR2-t!-FFy9BqY>qPUgzhZr
zqjR<;d#w_>$reOW_qXOW$B6&l6xE8ilWs6bMMS^_YtBv!E$n1^OVk!gu2&$k2r1l7
zF2$Gx0&V__3;zWs#_eN4;3<r^u3g8N<L2dfwwODJR>3K#GuIyLo*uk(YGZ*BkZ~c0
z?a!aTl%(7SV=(!2T(ps~IAgL`^t{jYn28RKoQBTB?@c7U*eW1vf4HciNY^Az<pTPK
zfd}LoC0onGQ5!C|cq=Kk){(J-6q~p#N2?`OW@?!XDKZrVwH*scrS6-x<FN|<LP-av
zus;uN9Zn>OU1Qh|o~(MUJid7xr9+tZBR#MSH;$QHnx0`|-`i}S_FoR$od3;@-)HmH
zid-xQRDn3o<p3^UXXi2;*FtW`b}YJyhrAfA-C6YXcT<S&FO4N(Nx*^q%5=2k*I{gz
zUG|YNLm#y(u>^W>aam9J9@}Ksd<7xs?RpNNP$jG&v4ZSIt{v~}g5C6nL&IR#TRCT_
zJfgea_I?J&y6jGV=_AY~ITd1tg`te4)<r}omr!BpLSUWT-$FX*q5Z}lR4XQVPjPfL
ztH<?6s{e_ts5=J22M{AYe3Mwyl~JZ`<tY4?q%^jwUQWtzHwwCXIoHKP28X3JRS$%-
z;>aij=-McAk7rkOt&1{oMj|j3XnK8B_KkFXXcgvdLp`oS6OP*W(4jYLsH&Bhrr?8(
z)ev$I(i=2qja9u*$`R+_+omY<$rM~;n|%gLN{vF!j4kzoL5p*pZH@X`ZbZo319WtC
z$AZuxD8`rlYt6na&vJoU>M3#>#~46Nb!)w#)}|+{tzPi=HcpwfM`z3T({E_ULr{y2
zoZQwsAC?ImsMRm6mLT$UpJUZzJ=I^2SD~cH#o;1%xm*1INVF*I6us#5)T7J;fr+Pn
z*CqnCg2;W3IC1ihm|O<O9Ahdp^HeS)VG+<m9<Dl6J`-p0kL032k@pv?>Z875dq9E5
z97fK}Iqu@|df;0EGNU!dk=++<Y5K8W*vV=-^5B{Fg*`Z6GdU*o*gOEs(PlQvB*2q?
z8LvK@i=m{r`qbvD8j?|~?5v%OGR}!oII=`%Y{AQxrrp%92w)cdu#g>v_~w@$HsYof
zb5~o6GIwiLe6kC6IhR2)RY)S(p{ivez(J$L^@`Q^UGkullIeYJ*L=}hcl^q(=Q5%G
zCAtJ0QtrYI2rI0I^7?EW9yx&7G>$vBVl1KTZ(=`hSEtYetE1_TFL$E0%M=xGo0JSc
zgskKrm!RPTyXOllb`l2QzHb@c%$w{%0*Ye0I2PD=ARa#d+*bH6v9uh8mE(?v)T#20
z3AL{pj{`A`gN&@*S9=rT39(Ai#A#jm4xF~N<3T5V`@h?|@wgF?`~x`dug`D9Jl`<x
z?92bTJW7Q%WRJyjOinCB#;D3>{8>aK(>Px6MafoUn-WU7$%#zZ?Bk3>oZFW1f;CCM
zU|Yw78bBaN6q=^|NqVJG%!OiVNPj?qE&fY^u^{kTb0S^!;`iJZAf*|P%yill3QR(X
z_D*b_tOQ$$X9ei-qIG(=*z(BaqE^l9qosEtzdmEh`=iWrWeeLZ+Mq*P(B8!#*~ob2
zp02n>e^|uTp^|WkO}&DrU!IzSHc{;zc8oTe&_PllmxWe2Wm5?cOYm8Ab2wLNfV@&L
zJ~@FQr9L`39%_$baHps<V1VB$@$*^8fQxo4HhEeOTrw9glFsAV2RV^nw-Y&lJVP;r
z9<M~MXhXQ~C2pW^Hz85r>t?b!fki*CS>IpM-q>8$lFjXWH5aj1P8OewG6he+2&z&%
zvjuM?3!7eBbXdw|VWdk@--#=iC@ekfgVR{@f7s<b9)mcLjt$zCs*W5@-J0@hZbwg-
z8wy?<|NiIyQzi&fL)&!+yPwYLb|dfAV?VQ-xjKU=0PL3k63qhfb9>w$Q3a|%yoXly
zV}0T9-tiuWkRjOJoDJEe2YUD0z44$1Ug)hsHrL=-?#BdWAcbUrAdAE#%}JO1c|~Mr
zwy!+DEjw(*zQ82ANQEwYvz7ZC-MECW*yGu%effTqZn?-0S7$9~6-W%=MvuI&+-L+4
zwsxOS=!cSfsd?7zE0|7dF&rG%`!66R!i{b4UqZ|fmJR+3pGu*QjZsrl(R0?AP!IKv
z<IqChsC@;6bODMSr8YY#ue|yb+u~DwebBZB%*q!jK?`<l)>M_vREpQ;eSs;Ve_G<Z
zGiSwG+gf9C{$><GQ{2h6eFgjNFCmTFqiLb(xp(IvDLbNRk;>9&zP!v*4t4HJB1JK8
z_F63g6O|=G29w%;pUU>_Ht5rRf&e+=?AeHac;-o~=faTLqD$T)6V(Jng6dq5Pus!7
zyqOA9KHKMI+jO|93=dFvBV4$08=jTH!qEg1<L53S&mHsD1472f_r-ZE9w&IV;TRh8
zU9JrCkO?Op<eqMu_Ad6zL~14g5K=BO*6)5}vk4S*jB=xtY%dN5#<Q}?!Z$Qd$7yug
z-`F|deP|_HNa!DzCHP`a{0VrHOa8;X<(G2ZU@}RZhSn!W72%>dN|GjCNl4IreZ=x8
zQs&hd`7E-pdAsI){;E9c*JEuE4BL$g!1tk2Z)QjEpxDwCJJ(xLChau3H-x~a?6#Mc
znEF^S4b`C|?&J=T08G}vclUk2>3<Sk*}d#*v{x3Cc0XX4l^=FFEF`=2DQ#<e2I9c>
zm)pyykGRV#_B*O5!Z<+w8tavnQRW(@BKWmc`%?@<SjE*|fjJ@9>~`beHTyj-)(>co
z<yJkmHoMq~<;c4I<${uPwb9lm46sf@o!XNP2_-0W)*AN5%^DjaIw=`pqJ`j<G#I;Q
zYusPS7>cnQr+HIUD-rGDes9h)IAk|*TeK}%CN4qU@w{%`Uohdoo&jsxAE*<QCM5;*
z%znp>aQ%oQwNAjTSV`6rUtEqVW>Am4f~vIojK<vot1X6<-1vQZP=1*9{ej1ctU~NR
zs2qo(n{Z2=-Wu6r|5%^Za&>rYzq0-qGbttl{4)t*q{026oe^UNUj4rrM(=kPZkyE;
z?uYDanb*8c_*hTqc#%dX;1-v7@mmNen6s&z3pOK?3~2cdM}K+(#kmk~p8g0%X-D=4
zrARU*@ZQn<DO?28<U0-X{tf@TV8^|0c-aBW3J3~<Y~h5T9ik!cvXi;%P{9#HemcOB
zL%lpBMTyMhC}VpEP`k|Ljv@~4CiDSOFMM3NJQY4@KKrfVSqcpcB0T4d_2)0!LX1$<
zQCu4FmWxqFOdSx5<P+!ZQfw3Mt;pb)vK6L7Qlc&Lm|U(7_BcwEx?S<9DFC=#^}io~
zqUA}uR?SJ)5n9gA?0Oyo+zh|48`+cmrkUc+oT8-Y1bcv9@%qT;(P7!`I2vx1-~!HN
z5lQS$j3{Ns=X$(Y#k*0K+xZQ4miK&Bo*cCM)mMd)7T+>&4-zU}Nw&)-3i9cy6<%V+
zgd-oh!zmM#Oj}k?6vQCyw5{^mFr&jrxK>AP1*R!MUlCrfnFu09$DJ{W48y+&_H1o@
zah>`y^^C3a5n3c;TpxY9Z?`gcU<@}A2eeKU;xTK8Dq>Y7t(C>j#vHb|^k_LUgiE7$
z{<HwzG*M6p`efUjos<Nrr}1Wd*S1viImey~M}2BrvmSHA7XVK5Gi#a%)Qh!)HT#nA
z=^JoLOTtm~VG=Yb@5U!AV{0{i>2pdF2@a)nA#K?MGP_T$JqFY_wN*NkjwmDX3fm&F
z{a4nRThZTD8?u^%M4%b;$^9{u$MA||0?1@Px50@DCFAx`LI;gxKV#$Fb*txek(f(<
z{}otpqAJs7I=?mPTLO<$5Wv(#AmrX5{H<i2a5xtN8yF!!Pm86=9EnY+57^O(%3cUd
zF!?dl9g>}66Nz`_%*fq1UM<R31-;AG{X{IABIuI?B5L^U;JpX+Pe&OKW>HHM!O&)6
zbl8R2_$=#uoZMK|i#9h=F=TG?Qj|%)xKJ_BFy|%;hJNZ-wooneXFOcGUSg4CKd|U^
z2mx`a8snqlrh@;mE#>hU+jGD!`|(u#IJ@G_4i4MZTmW_n-*wbUF<GIuZ|wR+aH8nI
zLSMm&BAaB4f1?^z%(%24Gj`Jl>Q(;3TL}S%6?Qv!3bB+pBn3CqV$9o}+|>xIg8RC6
z!b*(Ty*xM|@I7`v*P39&dSI1v^?}#kt?(swATY~!Z^41<5jtfn4#d5F8t-_FZ4boZ
z)XO3<1-hpLajrZ=GS9;5><5ZY;!<V7U30(*C_Sg|+S&tUgC2R+)@3<oaNgE?%O$y5
z_JDwe{We71KedJfK|mBeh8)Een+hj++Zeqlpxigs7~>1GsRm=(rYvLa#C;2IY(5Z2
z7(1$ACEea~ASos?oHLQF2MR7a^rH@IiuPI+$-ggxEt@NNU^nR#B&X}h#cw%K@G<uB
z?^|n*0)0|m-?j=A(HCKSdmzrV95|%P6v5UJ73E{deit!iXExT`Qub)hY6*7t0GW}&
zEdf_xcFxIpc0s9s2jh!ofy9<eg((Ps&)7)zp>spb^m^_p@k49fikmm#`|x~|O$H2N
zRzygG@N6n}Kr;OEb~xrg&1-Wc=Z{4OI(>q#`>H<L_zOFhGeldVx9lsJz2nskKZPGC
z<g}eg=)gZ?V9%4@|I#$hDARsIi<X|dKa*7ve3m3K(VZ?$iZka9#N*Hcr#oMb2YtYc
zBBO$Hc`CWE=WQWscrtnq@wS-5z@6aHYL}u8?lf?aE+<@aD8b=fcDY)4$#$2E#jaGn
zOXcZ|>Q!$E%}>Ud9EsFvY=&J=;Jfg!y5Vd2v&S0krti<w<95qiV;T+Sh~5d?Q}$W`
zLGR@3c)rfsUH|oj$LGG6qw(}8BFFt@7ap6813ViKVk)lflXO;=D!71|9bA4vMb636
zh}b~3;$XS{SQ-uEShdoBab*@H(>>^D2f|R9#pvmrt&SQfu`thE6E!%*Y?1T2YY&zy
z*3*l8dtLSbayh3P#VfJ?V7a^a@Ysg<EODT8qC_CDhJ)n@jF%vU8?!<YV*C5h7!YtM
zIFy?X1{!SX-HVUV=9s8FpS{NnP`btY<*i_=wkF(M!3Z^PO%6BGondQU2DkVbCDm=o
zGA`Hq#T4~RYt2qux9zaD01U<(9MtaBQAdQE8)&Sf>f)o~Q3zXQoz;?~6tS8E(K-w@
z*B^6n?=^yq7uG}30tXKUvX@%V6ZVcA43xQDAvhST?)|c{;{>psh><;MuTy7}-UK^|
z;a#Q4lsDOl!86H>4@VP6Ngh)9NOrnSnw_J*laD_7XLjrd2dvn#ga-3?E)mHO?SxMc
zGm{ZJ8U2@RE}S?}o=tm?n5o&!!*yYsXMfPY-%??8GT}W0vzYVaT4wX!wzYRyUef0r
z0uU7Py@fn43W(Zoi(Yf@a0v}|>EJ^l_2(RC+2s_#M9-qJSaR@2$ozFVz&iua`iWhs
zfZ(Krd&||FALr#0oqd$i*Blfiv0$d_0T=ed`|L*aCmQ9V-K>7ifT(YIPck9u$11!K
zjklwB7($WTg<HQ<!8DJ!`>y|j*X$kJWB00G^6kE#u<>+>*Iz_M^PnnIC@Xiwu{V;X
znJ->_I-)DyaDqW9r{CDhsz#)|!H-~7lvP%aaK5YG2x8;~QUMHf?IoM$8}YL8u9hUO
zB3{>e=kQ3VsBB&S8W<wC<N6#b7F6%~hA1Qaj);3{D6F9x{sK(B%iTA=QRtc<O<H5s
zLWHZxyLsF6MnPF){kR)#e#6ldNi(**;aD^ADNJyA_CB<&)ewc~K+5o?HAR^p8`o1d
zzY(`m?i$i0wM3T%mC>0@F+5ypYczKnTpIpaYje1krvL2Eto;o?o68Maim?sbM>Z$x
z$a%pn$P7-Z)6O?s{UePv=sAoHCB%_{><iBJ87Pcez$D$Y%zYH=1e^nf)Sv7Bl*9~!
z`3n=!2^+~@`v@Fi_57v8h&E9zPf-vSUU2e_z*z%ByA{d^vUVzqBCZ=)>+l;v=%KHZ
zHuOlqR^A4nn&*x?<#E)<P`U+$sVE5S*c&N~A%%L~j^~mV)G1c(8#|GJZX<-Aos8lB
z+Et4-o%@9~DdfYL%|sW_4aNUokLm0W_OLjHamsjZZ7yoN)I_p~Pi#JGFB7h@?_BG1
zQCo7C4|GnU0~h@CJ-K72EheN`Rlp17T5+4@kOCFDlxsyg1sTR)_S5q1G>}}0zJ=+f
zSNa;Ha@EV0rg!v?jFSUX$w#hbz4BfPuCMHR1sBOpz<dNqu^TZ9$y5}bNahwUHoqC8
z%1r21E?-xTOdC)nx1*~w96eWWeQJ07D*nk|*xjtYXP}Pn%#huS>UifB131N<zh9Nf
z$XT}o-YCjM$PicmJGtT6WLz~-vzR{Hipe-iEEvhAZRMojW?eejOeqha@FsN5WYR!G
zAb{1Aj`P&H(7ys7uZc4KdWY#NH2K;nlXvll)RVi)x*W5AK(_C;ezG7I)Q^POnJkCX
z31tK^55Zv=s9Qq4xZV23$%27MMq!N=EDF``4ilSlT-hK~CNsp%lR-D{<63Q*jGKmK
zTahrpFHHva9hKTqk57}&`Pp#{V3EW4FP2UzuUfLVC&9X{%RrDViQd6BuMU=(fuU}X
z8iilc95_QAK8Ie?YIep%9pyR75{@L4S9GJvzhnLWi|m1l2{0I)YQ;T>&bu}eW!{tP
zjq4RCGEthCENK2OH`?S6>O#Fh>rO@4dk_xi+57XL$io30vS0M!E(g}X7wm{np$p@{
zqtVa}wG|T<hxgoSh=0jCkNct_(`5xuR1JzrUs3Fx%rY*S7+r!E*mSNmsw|s{3hoW)
z8#^0iT6YB}M!sNkxxb{KMb{*^*7?bJOCk2K`FGofb3TC`z4!~{I6KjhE&8EEYf9>=
zJQ&LRg`qE4rCLx#GIrVfc#=6?N%+M#a1j-k<!Y31zl&9`(Bn04^5`!b?RvD{vBPfo
z>BkMP2!d`#L-FQZL$+tmZdG&rO2h`-z*)QP&`2=|#=A2a<_yK*L4m!MRpaaLe#5D&
zUNyU$I|H7SbMU<%96$&xHrgX*-B0j25K?JM?)4yBBY778Xouoz5o^{~_z#6=++`~d
zIYLp&WkvmZ7*%`Kp(LAx+4Zh|W2+C9E4U4cos?V(!?q@B>0$a>1Q!Ql?I9<wz^0IJ
z>!OB4vRDLiWIng`QA3uY=WIju3*#tjI21ULfI~F}QQdebaG)3k^e&D0yH9R6D5(z>
zRMXEYZBA&)S}1$KDcf?Wpe)gAh!<^ZwkD#HbfZnnrYYf<xTNMo1>-4sm9=<tgaN(#
zNdeG$$SD}TGN+5n&?0N2fcj3OwPkgb6s&c-Ppu2Osh`;D{>#nHtTTay%+W(6fm<z`
zkB#rb`KsU7TN0~r;^whmgWd%Mu`D4*V&%?Ypg3c7pPFPWMSVNr3nM;gGAn}NF?`86
z^P;`WrXe@9!%>5DK0K924iy~qt7bdu5NXS`e8=Lvl$iGgXsN#?{W<Q(YFmG=Je^|%
z4itJgCw5L&WeRoT(XyuVU_vU}O#T?-)tyZk2}UL{5SPQ|a_=zk!couW4?UD?A&+oL
z^to*K+w+)Yf~wJoEF4O@4HFyXcE$E3s)-CZ%OmhH*XEM<kG+%!uU~S2P;&7MUQY0l
z_?0&tj_ykC#$E-mxmukKr4;eFQPN*%vU+T<`F<Qaw9&3dXD*u8Ykt;l_-OLG!+SA$
zmhSfduv`AA{{V5w>~>BNd4jwU(1F;UC=<!7m*zvc9aV=13(W&MJoj=6D7qAj`5m6Y
z``KH0i6NHQ{Ra+c53Y>%W|9?X!)(Qy@dT3-)osRCzFDrynCwhfz3I5bO`;-fGm)p}
zZS|W4|Bx;Z-SV0@<AN)ZnJl%|M!RxMI4oh>jmml5n~sD0u>^x)$+bR5am^CfMg)Yo
z=ndI1fl>B=_^ctCYZKTqM7kGkV@_xUPjhR`8m@;Gb77m_416OC8BpBjsKbj}1lJ<y
zwq#jREO;i6t-cgw!i6$PAdA_7tMc3!x;cSFZ(h<Pw^UszEi86$NC;cA5++HC(PwRM
z792M~sEhWfU$m&4vNMPrI^HbT_30hf>0jX}hS9!4ezEPfXyK}HwfMmLvy5M465-Nc
zw80#;grO69@wMWlAP%y+Z{pH)H#GM|v`?D`q7vDpF94_gJ)5d34-DJkH$%%Kn>RQ}
zWY8rU&5_)T9m7Z?jz*UPQ05zoX>crSv`9a2+;`^Nzu}?#$B3s+RI99POw!nku~NRE
zlfEE35#Ux6#k`f}6^k>Iy><^^u`(MqLfppa=WQ;_3X=eJ%ricp@F_?w3opU+xrBgo
zfEb)(s4=7gT6i-b3dIT115faTWf3YCeX*9NJKnHM@%LXZ9krzx#jk<bs#&}Iw?>v7
z!IeCvT|)=<W21Jp8gKc-^Z|&7dk%WO<{Lt$cm%uGV`F7Lb-`|U_wt)YGc8dZad=`#
z9MR;l2G0B1ZoL`kmQX#z7&y8NC%V$ZK4W)$6tuLwUUz*Ho*)vbS0XgOv}yNqIv`z9
z7Lv<9NX&w!$#t0a>w;POg;`eo+A)g@F|-}hIh+EfmA`fjA@Qza(k|liv?|I<1O)0y
zPrmxsg$lqdw@?9VeqHX<qm4hewZAU731KDV>$(`Ixbd)`(1Nj_>m3{z&o<le>%b5-
z85+X|{W>Ki{avt(qqgza<v#B3vPK`MiACS^>p-PD<;=V}0f%H-1V~a^+LkEugh)E8
z3$``NWRCfzUQ!TUlTT)8`d58&gl<l7`Y`J}Pjt~}BAfYd*#1v=+FG;8zFw?#t?k!-
zI$fONqDcRn`+@MW-Df+%qiP-5z2flFwzKM9o>eZmsre!@{vuS1PZT_W7TpG`B~`S(
z*un2`;;|9$2<wN@;XcJ0V*~2ZxhT<hSQaTRC%p4kr6;k;tRN#2y8B$nCd=3_-g1%S
zcKFxvRPlN{VQ|Ff!Fj6p3oRZE!2CD`8(!}2F&_}=9jx?mUkbTY*$H1o-WJTP#_VJ@
zu0q}G`TEeNy)BbMvpVB#$*Q-}X0tMg<D#Bza=*;^_mVJr-v5>fr=5!x<UF&23x1lO
z*lvr_NAYjrw4dR$Uy8Djt;5%SW=lC1yq4r-vWC6vuoGYf<;wxP5>>?Wf2~UhS$;Lj
zTtHGHnAtT4YiW90i(PlH<VgvNxZz(w<z03&YmK5Hie4nJK`6kRipiZ{ga4K6f_L(>
zL|M9G(h^0Pc3t90q8Zw~+(6{2&I{g;LtDOcHe`&T3-J|)P*JrXI1qiqcsJRU^B_|3
zoU`BYUrW>5CGuY}71x(@N3<FHUH^);rTo8FPC5JCV1H*Sg}XA#jJUF^rsC$XjRIe3
z0=T_Jvdp_>zc&@EmRu)R2Nc+hAe{dFDaTduz!e$8nyHZXb&Ws{*&jp|C>Guer_9$T
zxHSba9kM^1a@<9Pn|0Z`d0d_tf0WbYbv$M36OyssUG~QT$r+-LoVE=WB=d!Y{D?F2
zC)xPP8nM6|a_+J^9jK2NT>r?FqhTk<^g=1Bk_>(~X7!q~)?ksR3-%KS^zm)~&>C|<
zwnIpOe8u)t|NN^RO|~gM6Robql>Y>He(s;U4))sS0A4bU5ADyrGNz7^9ToG<EmJ|@
zqiYwaP?dj~^Vv0uRrl8DLQ!(h*c0COPd7DMQ&uC7(wIG!3yORRlB78%?Eap`YCE2T
zU-~I&+HNfYdz?hc!0g#-J|4pGE9TScV<I`Q9XSewmb^f{uMKb5^HWJ*_bR|Jm#n>l
zz+?A<5d(skrsC1rgTPPA-QnHr+-tA;UmIVuPXFuBpIUd$vt*{MIWVo@n!bb;$Q1wp
zLdE){tlS<Bxzwb)r^<U=?7|1D?;eL|24ZZeI&R~AHsbw1@<@Y?X8HCWY#I{^lK0hZ
zU!EEzb1KgD%y=U9hHSY9W0;W4@Q`C=ojs6ve2=}EJs=xE@JE;ldYH;8c%qHk+qq7?
z6Syez`EcH+E+>Zmkyqzbz6))5#D^ayYU^KTMJ(g+|FEOgN-67z1b(yjH@+$a3wz3r
z<+yz?%l1!mji7Q|NArapPvlYjhy6S)<D~7MC9EWyEFQoo@?2vbMW|Vwh<}lbg4kDN
zsFG)#%(1}t7eO0}2mdd<jBA2-!6*+F{(;lgX;NT3m-@Q|{8W4q5&f<(a3M;>*U$JW
zRCyiW%dwygC0d4MwApCwu_%gIkW(sB`~9db%g+zW6G_TN@9|rVN%`NReU$6WF&6bS
zA8iDMgxuKJ*r%~c5*y$;=c`WOo-RA@N4%-k76J;zygUzNpBLXVy(M4e&OD{~#fy3U
z`Z4Dkv<vz3SrU9<YmxJlCZAbc;069|HToAI)}`F@-Cg#5wW307Y1t?O{c^RUUqdG8
zF_d^oo{j%12Sm?AM1U**4~Js8*x+9$@Pmq_bTw)(i!VyIU!%%kxlwoho2l<NV7RVd
z@EszzL;#2k(18_y^w7NPT28eH9tmI1QvKW9$`UeqIY)Cn52qru4{#Fy-NSuCdRZ>1
z-N-$F_LPX>Vf6Ta@3WQ=Phn1O24<SH|B(G6)ltkayXEW4n=@$tF~NZmQ`nv2RV)pf
z-Hy=}3Q($E#XtE^d0nt2EQ~YaR=>|<7>)nu#6*3(7<YY;H~EkKm*@&1r^Kqf>~nU{
zzZVI>f6Wcs-gN+X<oi(_TTq=4tPj7o{}ymcb5P`<l_}nXY`h=i>bm`Rho^OH!0fF;
zX}q$*e&?;A0_q?&#4K+G4S~wie)p}U8pL91Me&!;GUy&bKw3ciJ8yme``Y-kFzXk}
zzv7XU<Z9Jh->={c8TUr*_ug{yWx0(pHgu4$PROs$q1(vR^!xsU7nSE_&0B@+cyh;+
z_6O04PZr@LoQcHWUqS6nL65KXs)SCmKdk!eC3Q@IEjt1hiH*Bxe-vP|wW8-;4g-4p
z#MXN+kY)DAZw2cF>50WnG!eCUih(^2o|x`Gc`M#T!u1Z+*yV<03#EG?;`k9?4&K2=
z+Zf-SqLGmM1N%u1RvU+mX!P2o8rWk$je%2B@GuC(rW~gT^_}DXc~vICx8_F#3h(XP
zzu`gsb03F=Cqj*Fc`Jz012E699e){L&k$I2&bDS>ASH41L^QG|{P?`kY)$zUlK2As
zy#pS2D*Nlh#1kcSbJQu0)Ig)mJip9+EL5y!rHfRoNQtZQv9;u}#NC0=%Fn)45m~+X
zd=B*-u-2RdVyjPVN9-7x8QXH-hdBh#zvX0&G}Pt4)}FJ2m7lkl0yc>wuwZ3d-?0u~
zDMSWi_G-c*!NZVs`Y;kPbVom&M9_5Yt?&PUgYJ6prN0&I*!rr&;n}LJzgjlA^@$TX
zW4m)Gi32bfsE*ph$j;?ivLRo6uJK5oiefk#yZ~&;7dBcA@QL^X*7sAZ=OPyTR5bqk
zsss5mo2VxAF&xc7|Ftyz=$~zFw8?4<i9gOp$u|GZoK)A~uHJ60%2Yz59W3_|^-|$#
z->#s+XRe%pIXfI>e)}}DQ}&OdEJG=v8MY&_gVW9AK^{;T(Z8;G6T&O%K0E50{`$D$
zLjO%(oR?p*V^x<5FD3J!ckQ1hpk>*X<zBPnIVbUF130AovuIwlos6Z}YbSF2Kmr6>
zv47#?W8n%Ze=_=a&DL4m$};!Gzs#mQh!?<-pK=+eqs*yz>9Nk(DKF#d<JB%FBM{1r
z5An>9QbfF$dkIHPXhLDh+2}{!3(^vN0JZXd?kOV8&ulJGz%+{6kNj_pA+Z##PTu$^
zMns=x-(7vsL?wX7xvT-*N`SfZxxhnYOsl4Z;<~(Gv-VYDfRdMQ%oejp0vqE=C#~Xz
zTxoG^62)JNMN`sBkt@m#@8$}>jS)Ql3@j{|aw$i^wQ{&;)V|N-S~P2ytG+6=CXk7w
z{8#Zs8VGr5xWA^Hg093UIkMs;{Of9Nl4h3JWOlXMT=Z4oL!i;D!6pADYP&hDA~Kw^
zYq8|V!2bU>H-nevyj_nkvtW;Zm$M;qsj$wC9Jfey9rAylGbF$Z;hS<ph$6$&{vY0Y
zn2E;Y!@-F{&SXERLnW}#wf&EY9`|A5b2|s#N!n^uQ~xO&K#NN;_zg;lJK10iw)8*e
zw(akd%}LGfIw1SODGDj`UmQU``smN?UiMw|UPJa@tA#7sh9L%J(kd&o`+1z68?*nG
zt+tba1VZ3J&dmK^4~9j?f6vYc3e4-O_`B?;!u;6V!N_8cHa=K`@c5m#%M&iaKi6c%
z+fLLd<dn!18S(#alnEDkJdCrp^6f$ik&lh#{yY8yQ53$=tKN2U$Jr-;yEp}$m*0Ea
zNt}Wpi~LEct7P9IdG3O(emjXRSy$Wd$FSw`=Xgmjw<fE=pmWpyAb+9}suAMq?LxOb
zJ!pUUcA=~Y7_!;cML)VA(xfZB%OClqQJE;1<@&ede2`0s6Q*E_G6?wNoYvz#&^FZ%
z41xc|2O>#r!`ngFVE&2OcHJI%+Z95?6Snc~pd1)PlD(hA%KU(p14|M{BEG<Q7O^0~
z^w}>FPCxbT4gk(3hZ!<L?l3>k85FVk=A2_OMsv%%KY#oC-_xUuJ0KS1a;H^{ur0na
zdgSacy$=#dio4qyeH0;3oV_P<)*}NW)>I9m7fjUntUcx3VUk5+&|C|Wjg6sgvtQ<6
zL$ix;moZx62)G$zp=^C;neI1`Q_tpjPhz0i>f=ca-HV?_38b?lV4$f21arPWu(n(Z
z!XAv-^9cnpUrQ>;_8fDW1KR(^+S`D7S#A5jgNTTT=yoF{BD!9W$L%q%x;<{U$1C}g
z+dOKbZg;!w_O@PK%a&`cvtE4LZt0krnVFfHc_UW7EMGEPnVBuk!qygvkcbG8h=_;?
ziHL}Zi2M1@F~*~vd%5=y=eF6he$O+XImaAx%sI#V&=q==k@pgna{py02{O=@Sd^HP
z@`8-J9+$B=hZ!@9Y8MaWFGb{At^X?CmTy^oZUG#TdKrbJWzN1?V>Le{t)^Uc&d1}-
zMKR#M>Px-5;>uBPEskS?xEriB29ox6zqNaJN{Lc$9XSuOXM4JkuG}87)pXkOz@2lF
zaAsVkE903Y(PMCU#OT^;9}j<}leQ)SZe*5gqwxXh7RrH1=ARscG<FS*wr*XpqNvF*
z9LQIE{kqWBlJJki+4t6lT*s0E!;PyO<C&ypIf<%F=cXK=8X~sYH$wwEa7#X4K^*_q
zY7$B}<g)xWM-+6UMC;qv6+`F9|MH*h@NT>y@6HOQmr&N|cz?@w`Sjd(YnVNs!*;PB
zUPy%3)RnBvl#$OT=qpNML))F`Y3%6cM%f$7w;#pyaqIN~ac5c8u8O<5`43dIQ=&9k
z=3X6=T!V*vY)m;Xw%TF;p}_kFJCgH9$dUF)d3yRBSASnL!j4w+I#i0=V|cS;ISh(j
z-3^#5in!Wq$Me)ucodi3iR>IUyhMCJcm27f7>E2^fdf0`Ahi)O+hC`?7uj^0OHD_g
zI7RQY!OkRBFe<_t5BN)Id7ziG4*BE-lkHqUse~IIvYLTi?<MM#&s-kAaAxlZq@@M{
z`$^K)`CPEcbMLVa;_Oo_(TCAkMkyU;@+<r(m#>T3%Bn?aA3H2Eyj}=c<W!^zl=$?q
zFiKi-F;5&%%u607|7z+z3+&UxvrfxG(@M$tsxrViFatf)?zPVo$H?jjO0PJqIZ}ui
zA^oy_nft01MuO~<WES|!M`<FX(ANn;2?qrImvRxix|gBdux|nk7pM23Q0;qHg(zOL
zZxcZcc=~-8a1?n1IF(85`wBKfM1hSAoAyH@%55(!+K;)SrLuwjlpk@iqYKr)_H%^_
zv7OFb`d<AdyTkLM37a|IEe|^zqGR6Cw$HA5_-xUm_}vkVC`5VMu70@Oxqto3U3SgG
zg_}WNp^)lpqZ`TnoUrh-`|P@Cu?J`4*X;U-FI$L<v+51;iF*mbEg&x5_^^8h23ojB
zxSp?2!|C~9yD7U;Zi%9pdAKkkOIFE#^l<P3b<G{BGB_xPDSoEX8?YyD_4arkOtRY^
z4o%Yx&b`~c@jwp_w$OXA$KK)780cA|{5pfP`MsnP9rAGSQLvzJze67mmT!JTT{p)s
z%9p@oiVRBg=oioM#BtGIAs83836eS4h#cW=x*VB<kCxxisE6as?{sS%n`KN)PwvY~
z%!3brjrHl_RoZFe5*_sowc@PDS6IiCfI$AAj5xvDBWHHl#Kb@&l^^6w%9%bXfe>qB
zcLn4JVCMG}GL=?RsUoK2Sc+pNS=6*@_Apowxyai6z@}GF94>6Ll2pvdXO0mBnxi%|
zdvr&Bo0ThHQXn(^IS;2YPoWV-)?pyoCB+Q%+1y`&(EAT%&U7uryr528B#=fHBxa{J
z$ukGbiY<(;B^&H561$Ee4BT}61{eB$xlUtQnfn8HH9i_6EJ|P-R(jT^#fiuIHrS^J
za=tnd0s3wHhia>5MISQ&Q*ZT-0|l-Y+fHLlNglB4fJIV_@9rkwogCA)G#ZtXnev8{
zlC;PrE%~Oz;<=&1CA-$dm6WNj#|&a^ufQr-rEW6oaH6ScB^_l~yeBkODK9yMPTBID
zeYrKbdsgNHQY)3(cdH$NurAQ$8t?Ai^{_%~6Hkav?e>swGTcWC1HLZS4W~LZfa_zC
zt~5_xsYu!Ia1m5yPqB^uUFgI&<<6ap4=i@An1GxB*cD>YY)cHZ2M*%2ZFSsI0Nl1k
zV;zhK$^uZ1)yG*d-*#lrbxoz<jrZdDmu**)8TKeNc}lzUyic6qpmFv>tbmsVksv14
zS?-%Xxe>VLC1~%?6_Hc}@w^40?5!pNv(0zX<o3ohxyS10c^MC-1NkkICmF-hKkl<b
z4h7bKvW{GSufExi$L&b0Ri}#9_FZmJoK22^pQE|DU2W*bXml*M2jYYs4_tB=aX+1?
zzK0!R09;3`Kc=n?)DJyN6)uJPVLMgrLwQA#{sC38e7t2_Q6BlzInb0g5?4pp;whc8
z?6ou1vwlqLMGa;H(W#_wdFRU5R}3hCql@UQ<AaEkPCFOC3sbM!d)^xk$&l?5xNI=?
zezZ{flwLdU0LcW3v8tMvW~6z!+P~jEOj?kyat+dbr+pNVD)2}vsF0nH6Lo0rPuPWM
zA=eK@;wZ!IMQ<^$p;Lu6Kgrp02fsldd!d3)^DFvzGtl-~JWEDJd-ikhsjZH}f|TJ7
z&|NK{BPN_p>`UJacwE?5)%>JhD8$473IhE)0GD?0q+N<n@X`1tK9O8>0(|!tevQy!
z`?i8pY9C-FAnv>BnG8L$#=T+R$1^3b-S4YsKlqAa(->nv#`~h<FY?}T`ziOON;<UJ
z&$*!xmX+1zm+F(Z7{<?2Tc36(@lpQlDt|W61Fq2*Vm)1DSFcZg$yv`s?$gKOU9;YG
z6j*+)U0-;VcX#smy7j@MU_y{{-}N~VZU*uV5jS|l`odqa+0lL4jnRVI5+uHH>&i{}
zRS9s_6@1ce_ED($d9U3P??v6hW<Rt!Xno;WRG~(W0Jm0<!(|mjraQ}R{;|j$*CUcG
zLgu$ehrY@sp4skjFcs8F!1dthuK=k2pn2zahO94a77PUXnW2G}=4KmKO+v=A!g1Pg
z2XZ0&iyY!c<PZy$oN%Q7D!`J8BHxitqRYnkuc>8rHg>)923I)TL7-y6kaCbX$E^>w
z&14e)xXTQSIo<&s=$T3BnF;HIX~lOy336>>jE-+EEXyRH$6tT*Z+2JAgT=!V>>U3T
zpNIQ(P-h`~u0z6~o`4fnw#R0y55}&oRw9RL9?#6yD^{$qSuvGd$`eIaE$1Nw@ruo<
z8gtRkP3j`gvKD#X`f>+Oqbywj>D@?rd(#$1*Brw|gDgV0!Zd34`5gb{Uw&ivSJWal
zprl~a)dMOQ<!ta@Xt2dGj0Vstu3R7w#4s2ziq8Z&jaUY<EeCznCSp1CUbT83bkyjv
z)|doRgaNnOHC2G%7*V<@mjNfp(xm@sQ?13(pNjp~gbK?BELzzP?dwBh5|Na}$Ezdh
zLJl7~noC)euJvc-Rq+38S^Kr6&6dYx@F&3dyfOwBclrjnRS~Y$Il(cuW_`Mb7e$V8
zrWP>P<}xBnENAyU6q9z}j#6Mgj`DTc**IHYja7<;%B#UPRL@>3*n%gtjRB!{LceW_
z7W+AI_kd5Tf7s@DCKI!`>9@J$xA;KKRF2x3(9LdG*<9CV+mZqr#GY(-dc<>%{ky~I
zuYRTNj6N0gw3mfP-D0~EIwBPC!bA)`WzSdb-r`arLj?E03(;jA9GvxJ15rO89U))y
zZU>IATHBkbg148D^DxS)a@$yMe3%KL#ONLHsWeKtA#25<Xro{>+<VGdao9Vb(a>T?
zs@FKCGxkXif<7m;fu03TM9SDvr$SYdvt#}@Di_!wp92R8Jm-o02E74teSa+CsNdVE
z7@>r<a?riW{yFX4Q22^yKgaddlC#^+B<AN*#=r)ut$~?|Ws^p<Xn*Z&wp#*!X6O7D
zfGAtmd%2qT)+(yo-p{tR%~Y#9@4rCBY3BMtKvx`(L8}URei+Z>haoMR^2~jdXBtVh
zq8{MK6>zSYQ0`*CMlR%(S+U|N05@_mo~6c=dRF$y`n0N$4x=bNKh2>B_Hgh@-S?7x
z7UMW?qnlAzi!uOyzTOoKE~z9NSxGH%7=2NVtYE;BY{R~cXK{AIH@;$DRqar77jEoT
z`?{K=>fT4<;bK_76s>h1X-6H0E&3+U8u@*ZQ9(Wh4fd_?m?_ijJ71V^6r*TTE=(s%
zcl#kvJIaRsm{*lN1mwl_lM~c$M*hWqUVrv~=IjE(lhQqV&S?B=*VWlC)fAMcmZJe{
ztv%wZ9n>~}c-14ptdQtfAt+ZrQWzP0X*%tiN1XE!C$4&3CBCsJ-L8Ek7D}-U?I=&}
zx<^7;f}bur-RpgjHfaDm?S>c<Z9DwxjgN#*W-jdFO+H>5=>qNMN1P)QXCcW!nPzSY
z7({!D)97PYI<Gb;T6<<~h4t;TTYaE`o+-+?blW3wzfXl(#|U7z2OL!0K<c?8Iv}}P
z$*&<Eb#MiO(o%TFjv)!}p9u~anlqo@7WJWr<?CfgD8qd^EvtyV8j<}=27@AqgRqef
zfwcFWoui@!&hk)`IKxvm##^*{%=g#|9#OfR1NYmwM0JgPz-)ZgM1`Of$+F)jI2awx
z)V7^iy;lk+HwR3ro{{h*w#8lXOxh|P-!CJbO$qFmZ480ERXAm7>82$f2xcZT1iI4c
zkHk%<(Ed&-@BF5l@kqIa=XbY<53Fo>Gjk4&H4<xP<-+`99p>{n2?;_D9=nLTFWcP2
z@;ox5%*%x(KTL68Ti{;|^wcUDx3uYKc%m0k?=ctu?7mo@hK6>#zZ&IWIZID+aW9Hz
zMdQUcN_xKdku=5N_2E^y9;iO$3_|AIgO0S;r!asA2|R!OBdNJMiGJHvt%U2ARVHB4
z8iVZXF}F4)E@m{ef|{ij4B`OGfn+W5tdwJq^=kDl271PgvGzxbBy!KxyR9Q(gA-A3
z3`xeUD>u2DFm3rGVW=%jRWb>Q99IU=t}SM)qE?c2&xO_1K1FH&*zs$8{CA(ie7QCk
z1d&<{l|7V%Q)C`ZLf#fqBd$wyF6&k<L|w`2y{9>Ho7|B5y?D(yQ{_Q08+{TSmap^w
zleVcEvje>ickE_ILM`suTf9fispOhYO3kEuw#`w5y$EF>GQO;9+XHO{VVpgx!@eUI
zMyxi~GrUvOqz$$!!9<_1x6Dk|A6c58ukgPY-x5!-_d+~V@-sQ)C`P#_o^hBJfl~UM
z-3e-OB9($+Z%&deg(|DldSj=EyMDzE<ia+$5(ZRnhZ167Bysg&hj^f8&fRt-#*=s_
zH%qxS_dQZ%hbi;y=oN2eC_n_TV-<wB7?=BRza7s3i+!TrP9)XWs<2LfV8zGq44|%k
zJXHavOVCdHQYq8;JNc$$)3q}(sH=JRp(lRX267N(z_YW_LKz-Xfp*R(`tBAOzV{M|
z3!9aLjqGvnCoyt1%6*Vn8dezRMmz7{HMf#0=Yznk>KOIfhY8k#CZ)IkD8b?e^W=_)
z?c;bxVRSlFUgA@BAv!@efcw3eUzJoTD3)3GPhyifr=hD(xyz&$bNcx-dXSFmpnc{&
z$ac>XH*<n^+UL208dhS7C?>lvVj!tZh`o}(t9_X#=3LcrbcBzF|NOcF3cs+BF)sP*
zQKQG(H#wrRA$2s6@r9(A+0`-@?7NBtmS9C8=sH)YzJtt%KO`gt%i8j9+K-8UaPI0j
zEQLjl{8LVay;*z)s4eT)MFeYNJ`~IJOY|y_j6G~B1@zm7;6aEu@tfwV4TZD8mLn<U
z>J8;ggdetRHWWIaTVvO5$aDF!_n;&J?P0P;+jSe91u3=DSP>Et+x6KkDQa5mh7G~!
z;4@0yv>W3CS5YHw;U;=hJabk|^hRx!4Izul)J&!u)~(KN$&b;M$-`Jx*#>P0hD=d?
zcoN+jZIs3m4nY6fZ8-o}_33iP@3GrA1XtD8S!Z`_2+jwq5JJEz8ysKs4iG&I0#D<d
zH^dPnFN~p%pvlv1nE%z(WW#f!4fV?AIl{3}EC+Ac$PI-}q|Rx*jY|B?)>S?xVS`tZ
zT1~rRtdDeWW224R;5=n72d8$S+<oQPcF@Mhhv^j|Xa5QDOc$seu$07qq7R4nst(9W
z)x_kL<j2}wNg2~JihqiOi>G6YO{)OZ&E}W)bbrs<tg;!7NMbLz<YqeLgfZEyfV^;@
zg<M3%ip!f5L?wfCNwEPRR=5@*meb$dT#cF7n-#h>FEKW&1`=A3EBEJn|7Z&X&4?U0
z2;7&jqL5OmWOR>qe~hN1<D}1+H#ng{K;qdJB}U+-WEO(CwZ*xXb%dvr?I)W^1<%JA
zhpjdj2De-Jsf&AXsk%O=#}1LiT#EPl!a;w%)|kCXyZ{KIFs#YRh{DrL9ktDkV6&xa
z$(acXDLQAZi5&b~3NP56u#&_YZ-kEgiOCW&>`K)A>7E5MY<YATaTvwbWTg+iyv|k!
zE*0ZVg&beXu9}s$HplF4m+q5@s%R-g%8RxxesRYl>8|(w=Kt9?RN$il<wAT7K6qoo
zBZH4^3RL~ysGG?u(e&o%LARSkPr2r7N!t8ZGBjRZ+rluZ^s3x=w6-!AY)fQxwGlyO
z+x@I4RxdfF?Wm^l8g}rVv7%yTlpI|t;Qfx-6>n4uQ}!in{QQe%$eypJejLT$3l)0h
z$EwQ(N8$vbv7v!f<-4o->zdq~LxOP;*5wrZy@_)67)kxfO<naZW2h??4rQZd@X*@b
zhkb(Zdwu-B-;M-$iU5OsNs`5V4JS64#Dr(s9jzu<_Bg~Wrluq`h+!4hxu1iOG97k2
zp6LqEo_?Dn_C#(<Jn*bPmuUVU?Nl|zVlCAU;7nWx*_?Jl`ORPLod8n^c+<{A_gq^g
z4CNSS??62BfY36qoxKc3QI;z#>|A1-f^dWbyqA{)I&e7vy`R{hi|}se^W38`1@qt*
zKZq{HE$#zM$;cmG?gQm%y?qqGD!(i4!|_Whls$H#ns|*}+>mlmRR-;oid-)T)yblX
zUk|6m3H!9V^omI6@qM2ql+Mz!`#ujS_n~$<#fc{S(iat)ixp987sUHz5{bg~<x%ui
zG!+)ZzyUG5&35@Z!JoX`F8LsEQi|Vo=Z|lyf$^(_bW~j6+eA-0w=^jHu4*F?WMO#O
zzK>@rx1!hp1-<=Hp%LP=AEO201Z!Rx0sAQztE-zJtDmEVcxB;B9QQBrj8OMN&=9HF
zqlJwjdfKjfG?)`MqVh_Rxx%h~)bCTd21^%sgnj3m{F-><vaHj#U&q(xHz<r!IMuPe
z->$2^AsZGufjCR{$m=MFuYa_#7o9}Lb=eJ%h6)KR@3k8rO`jYu@^n*l!iiFO;X00P
z_U{z#K{>ly9t{R=c>~6DakhgVO%E9M|BB+1cdJi-Hsa}RF@4t<YP%n@+aEov(61V*
zx7r=qAIU3m(4oJ<@kL1y!F1%gJtUr?cm+BISG{dReR9fa&}zf{2Uj(;H9X%;p7UoL
zQ30TI2;4dUc-TfhdUkyA4*s=~;D7X&je0cLNz@i3@*9&Fnu;K8V{=C0;)vwO;U_S^
zpCj*(jZc6e1w}UPhg+TSXyJALu)roJHdkQpZRkLCVUv8cHd16rX?It(Qu|e*{vQ-@
z4dR&+J-a!=FRy9d^RU0zbnm7^KD}_-Gae0YbE)_nSg4u#bvcruaNI6uB_?Mz*V>!}
z4PPoQ18Icj=A&9UH^TXL+Pp*tmqE8Ja7s`T=Y_uN<SJ>i`(i8!?t55cxf|SHk+E{q
zpvWki=%V29);M!Fs2)SHKrRlPO1rk(9*F+RCSq@s)6r@ZHqxKq(I~dOzMA5zvNy4N
zUbM!XVkhikQ}n8YizjSpPKEFVqONjXTcstFKSirQ0sAM}+Enkc_834A;20qtrFFzJ
z-vO@b>#EjJ1t&AsM5oE}Y+d3qin-2KIuciq_-(aAi|Pc9aE<?{Sw#U%TN@u%z`%7r
zlqetnueL71Cx-xjFkoYSz8yEtHpJH~2pqX{z#DzozyHf3+f+eS>LD|)UfUdVaDhqu
z=vzUDTM}=rDj>5p2cC}#s#)ydHXj&g-nVRf^o-I-3^<Iy3ES~#Sb<v`h!U)A>u6Gy
zl%26Q+C_V9SA6FlB7S&2?=46_tvIL`Dwt(<5!3&w^<*C$y#2P@;h5Wm)8*b=5ky?f
zR~Sg8PCbQEq1X3~`*Pr;Ts71=dxh4`4&|wU)|y!!e$?$63VkhXRdL2pH6*A5AwWi?
zm+VM2e|+Ph7-fm+bGXPRx5|!I?~j!0<1qKsv3MqI^_JfH?Rd;g?uM`0iQL>oFPCgv
z{dwHi<3&lZ&M6;)olm?~E>QRvuCjN00A=`-4j#o}r=7{|#BqVkk<6b16>NfCdiT?I
z){%hbW4SiXxx|TFq#Eo!KaJ|^?EM^+&Dm1&z@2~8rF>yWZaG!-I9QS1>w_2#&s@$2
z-MxMGp-%%-0iHVkH@e&f_Hkl}R`Mcx1J3$Fj@YJ%5N&{q(Mk?X?%hqkyF3o+xtu<Y
z1`d+vbNS30u*Pjm>~r5DqelPMzVHQ5)?}1wU&fd6UK0bN^lV=xP(>pzmpEd>y@Pgc
ztxHK*;wqT)+i)pf5c!+z3{{NGDawicZ7zp2OvJOGZ~4xfk(Zv-D50hO-us`{P-j2*
zpKY~E?Z?a1P^_0f1@ZLa8~wWd9Dw*CR<x>`0j$|CIVib`>K>HJ-fUyxbX3Ohs*Q1@
zbT*)J;_knCV_`qaMP}D*tn5S_jbfno*|qUZhBx>i*z!GgT|5gHkzz-_eq-`WZ|TZI
z^r0kQ*_=02y;2z<p53^yuoonvnr}C4ypqjXWjAk3W`!%vFNIq+rfCgE%m(?#<f*H%
zTdQF?2LM?fV8ubgZp*34xS;rx+pC$pENKMi@e_83Pao47Wp@WhUoO{qS|p#sknBq;
zdXeQ&2VwBv$<H(_`bDSeIzkwq;ni45b;T;d(Tvy_Y+0>mva7L?J}neUDi%6wV_`6+
zF0?Tli%zS)tqyZxYgY+Xv9TM&wm@(zHB04TA=|<@?_;0`kC@kOyw3;b0WM^M|8V2z
z#BA51(iD>t$1;18k&RqAcLk26p*vzza#E;X<b6SkSexdP=9D1c&Bn9q{s;X!2L4Mt
z_5Z)WZAPM+I|o$`jQUG9Goj~H+Uvw<v*KCNJ#yakqfME!F_=VhreH9d>k}jcBJMwq
z4lfU33xYJ1Uc-fI*BTwSg^pxwa&=91U(!iK>k_KG)Z6_@Qn=mzC*&#rODCyC71guC
z7<Lv{AXE2-md8Q(0}eyUcwTR{P9OK*ZS@H@E>SD$tua4Epg6>*yOD01s)601foN9V
zoaTU7nyuk~l!irl&{}-r13iEFi?#YNe;sV?{<K>&>+ru<>DKIu9#Gaudx4CO-?D=`
z*>l^kbfj?RW71ijGwocptQI}tn)qC2mtwl{Y_hc*Lw#JVQ+}#>1v}R+Jam~}3u_E&
z*ZGKaaEz$#_0g%sHEH}+0ANE7Tua_>Zl8_0i10x#a!MSPnv!_rW46g>hhR#&>+Ff5
z+QlCp;eAVjhngI9sp}^2#z~xRjc+6gp?)eV_-#2l3E=7wf3{cC)^(%!oHuPpeg`ju
zz=$7gzwL}?)Zj(6;N`G(`3X)EOncrzz4t**j2EiUbsgzCK4Crno+F~(c31DEKUDzq
zY1`}X$vd>fdgFcXz|ns^#PD+OJ5YV2ENOf}q{E>=T#;opj$C9B3rKhA#M0SnN2>26
zGP$my&&OKD!rReA2Qf5Iyd8@^eXZR;*>}eActdhjZRmeo7M@PT7+zz`Pu2e13#E>W
zok~I}p%0x>be@j>%Cay^1rnB>Xzw_gqC??CJ5x=7l$GxNWCQU`7oB3Nl<47XK0^hq
zaNSohte>mCJ;XW|Z%KQvdWPeG5D72zeqI3>_g(FJK3btJmn?;9d9QttQ);Br0g+&N
zsK68!lj=u7`Mqa|RJM<O3I3OVwhK98k-#M&yl5BmV>UoH8G`K-|G3;2C+*YdN?ss+
z_E}=S>_W&NCaxIQ?ehea!$rEl6R?qARKvY0mWSO;c)AnnsL{W(uW}%Ya@N?_xd_}R
z5ds)I<E89|ohYS=;{3mfXY$)UKs!8*>|3Abzr#dDU%`HG{t%ra6_<Tq_0@;Zpw?lx
z{g4!)GU?68`{dz?S??<ePJePLSlQeFFaLA)z)n9v5BUGvFFq;hFzm5%VQ{~~JGa<X
zk2%961H~RtCClQ~@yu6LQp!=g#y=yps)MWycI{(++Lc_6zQr{Gd6{t0zG2r@-ICC=
zoRBZu^^X-Mu4_e`-SAjp6l&}2#>aw<8uwef$;W`T<d|r;n;&zQNyg$=>=yszZ-Z@+
ze}dNcO}q86!u)YrL1wo-<_w%~NG-AWZAoOe`_N<l^<V7{hpZ_3GK_4n_dV?%8{&W6
zJH>|j=mR}dTe@p)ST!G|-jvd#&xXe{eicK%%x@#AXD6{hlU5CpV&r4NvDT11xE_XQ
zluuyfVxsBB_}>)OS!HA6V=q7|+Sp^`5{wQ-501|kVq%$_j=Kr*%*{%u@wm$-1|~e7
zKr&!f5}R}xM`a?FcGm9tFMTW1$dt!iMkW#W<^S0<hnZ9|a8bShq*}waIK#Uj+keqP
zcQ)bfH8Tb*CnLwhe-_6&cB{>bZqPv4oX3K*oPn2qi_+1~^|$Y?$1iPOfF|9(u=+L!
z(SpQk9jWhR;aHf|gLbf?#qRsnMBQgu=l)^4KgW{9S-eaZRsAbRi&`DG#ZEGR|J$AR
zz+=JAqxF@uC`(nFGhNb+T;zGn>mLg#V6qonZ}b^-C?{uALaUH@soR#uklHT2W$z!f
zmc->W>WsGr{LXQS804~}bjIzEg}qp=gUH#1iSF>h>PQr8UGb@q9l<ogvMrA-)gx~c
zTbZ;qTMYGT?*W3JZ)>s#_K+eq;5b2x?6qqX%d=Id%pQtqDCdhTii%@gSABfB`l%~<
z*C!&;E0MXe1s-UFqj@yP?Z(6@<~|aXmuYQ^{!4;3&XG4!UTw}##@d!dc|+@JDka<2
zm>+4_adq34hZNdR%%L3oVtn65H=_I)2W@+FuKl5Cc)7>zh-XDiWae<+wlkmcm0aRz
zZ@ZF+rjjR^<O{9#e6F6%nJ)w&VuAKsPjsOB2<<R&F>l#!KWWz7X?wGKP8G3BvNQJj
zV%#m~;RDszCt&xK=Kd8s6dRN13q(x3{aTkmm4~aAgf<3acqDt5EUXnm#_G#<MGYlU
z|1~?B6Cy<psrEtUv7|G(d#K!y%%R6~mM!RaPWT)iTw(puqv#YRO$h@&b;Zbq%|xs@
z?O)8DZSUmLspveK;xkoil`m62{WehbN^u?*NN}D_0>r}{Mew;KDKhhsSVWTTd-0VN
zo>G{;CblJ>=~CE7?|snG`Rs{+S45`|@@H+Es`q`E@X%In5v7WOH|gkh;`}(8lpCK(
z1GrttJ%#hzZo8P+nhd}ANzlY@8lJLGD=4LHkYLkqpZTgccTn2n^CVssIbi%!^`t2C
zOYJZFBEFZelk^m-tMp~!6akw`bM}=Zhii_ED=Z%*@^!+D%7fwog&Zz9luEJNVc#Uk
zMCaSL*&2}FE;96f`!1%jpX>f@`~LEfgrP079~?z9xanHx?MMIlt{kEE{>6UE&L?-Z
z*w4O?+bC)Ci%*=B8wO=l9HXe8CDLBC$vG5h@ly|ZwYMoAEL!qwHYKZYObVa)Lb^in
ztsEY9-KN4Q;9EoU>-`6zJ`}OMVN+-&a97umQMdiZO@%S(RKoI`ynAUXm$li=`KyDY
z-)@P)q+md7g9|-qgZx|kx>wn)n+mJ5c7)xwsqjA-bL{p6fbQ*%%M&avfC{T09M8mC
z;KE)+ts#k+1=LF;RpQY6-AJu^MjGa$QX#X=h9^{7V4xBoGQ!~wUG2zC&cR(72<f%y
zs0xMXIck$LQ?mUkJmXcEd$AaMY)tg+9+;BzPutkU$vkEC9p}@N+a6BH(cHxrr58JD
z6B2Yv8>2e5iT*jmD4umvb_R?0hTWCFrYd}4%BI3hEkpTD5*CErX#uCl#V83Qeta@6
z*>rD8aVH{CmyrB=Mm|u!8QtPDeQo}Eh|Ss*SN!Z|G!mUSaNC@Oo2v~z8ouf!o4YBr
zE%%lzHuD^`Qe6jUo&11v{0(-j!hII{nv`02_iYLuhv;U{D|vrDxkrH^vX?E&0pxv3
zHsky$b$Ww6kZ8lXj+`G>>%G7GFt=`f{QVRXbfZs{+NCQ|%+RnY2YPoWJK{O?B})_I
z1ff%qv;tkh0HxfZwN}rLlqd?N(`?Uv>RL(-%uegbR>chOl`FcEx~X$#%X11-QABmv
z${;%_FZXd9%OgPl5}Q}_@2e9o5)`bnHQD=_-CUf`UaXA;*H=V0BDd&3e}#OS4@GOw
zf2OUt&(`^-8$ITCw!Z3&yM;kzn%|J0a7!G7Zpf5p8~vGfduy<Ww#mEcZYOtseB#D)
zxfdO@E!nkv8-Pg?;MREN4i@F~vTXso`%uNa<xgeXefe-?>aZP&vJx9&JFCtmUTRvs
zwkyyWT(!L4fL``Sb^5YB@9;JgaQH$JL&?;Kjitwj`upD(|IKz+K&p6-Jj>9p<ak&R
zP@I;Vr8l05Ey83ay@nmgZxL~&jaG&EL%H;XqS)bBOAa4EI#M-%8*Z|&Le}R5%|4go
z_9;8+pv`K=_wAT}4L{n3FnBzHVmBTHD3w@0;giJ5V*NQuiH<trPbH>BRRRv@hvIu|
zr#BUg+5dYypS=@ETKu<Rb|$GF14x_f&jbVd;rQ!AXD=K6Y{Cf#iZEPw%5(87a(u%6
z;fkC9LuB~mG}!w-;Jx?GUts5x0A|%-m$VNY6C9`VGW{@mbLtTpAF_{f*A&*`<NTZ=
zk?4Uh1hDymYAzPVr(Mj6A#=49N$8UV3;TVsK7X1`!lF-kCI@t&XH|oJp3AMg8DCUW
zSN3R~!Ef1@@vOubv21<Zj$h?j+uq$lRU8>MzRpjBkX5Acl7m9Q+{i}qO*VqU#oa0b
z<J$zUR;#Y1`7T>=i=m{%DZAgkPcZ9hmfH^rQVC+{#n0_+`!Sx$=}XxZU*!<}DLaAz
zls$}GpZ2q3mXc-<ss#EkNe0E3R;*3j)v7pQkH<w(5-QqNj~8ZQz6xAi{dnO-rc=ka
z!LE5cbOS4DI*Rkawf^C>29JKZ?(xDium$Y;$DI%Iib`7dm+gjlCYu3LjcmYn<Ku;Y
zLUQmBikqTMT<*m+<>tqOev8)D3-L+p-DkHX;CGj#I)l73IHiUaGBDlxcww~gqZa>v
zTXdai3kV;u+p}u`R6-zm-j!3z?sz=z*m^RNE2|;C$JyOfrA;03_}Rh5Vdl>NkN<3F
zb_JtbkND5J4+~garO%NH*FQX=S4tp+?hzHb;A6zoAl!}2HaJJK3v5(^1k>DPWBd!h
z3rNK%d(~J+(a4gq?4*tJ51<etwQPLi?;he#R@sEW276C@9H*xoEhQyS%#|ly<`QYI
zCRM<lfT5RS?WEmx`SWtXL#v>agWje*o(j0j$=9{POIQ&8KhqM9PAp@nX`*x8a8$<T
z#NKB!qI1bw2LazBsxU&@%xGPtFl`4U&RL0<nGN`A&&dU-sjru_410KPe3Vy`%1_RV
zA>!FT{<9=mk(zZuypdK}`q9hkjfIH^PWl%CRbD&yRp2YEwIqhJ`>P%l_$MGU`eIS6
zh$g$o7RN8lG<bA8kUhaD%63939jo=-+0oh9Y4!d$_Wfzrm{Xprpn)bQ6wG8zrR7q`
zFr~2d$Xa}-j>bse`gk&W-6bt#y|w$xyQeO&j$C)5cw43D>T-xgVx)#?DN^=wKO+Wu
zR^Sf1(m$^4TxF}1SeGy)G_xjgP&V6@x7^y8XNfw8X5`EEP;|suy)HXytXYn$g{{v4
z8(Z3JLwxHOhOYeGwlSXR98|QS;-@z`u2yK<Yz{1AKt|!QC0ZB#HEZ}PYq&LFEk=Ge
zU*(u<+p4~`lD@jzy>F@nkOMh^n~S)%(_g&%EGEKT@u3)2sdnVw`@B!GZ6(H~7*QtA
zm!J@igo24Ttj9ZP6y&?36EElHby4?vC!<FH-g+Oev~6BxLr1`YT;XNXf+_XEp_~I-
z3(K5z?{?U6qM}Q9799x;DKY&U))(y+`GYa%%rPkSXgn*HUEn2cmL2oC*DO(v#pBf=
zI$OQLi3A1OX)aZK`g7|q!>5Ma$AQUZ_)(mW4&=XsnDmczT*vQJkYH=@Bg)SBH*<B#
z4OCDaFU}9jD|5E$$Bzz)tmmrrt*nZGC{FmjB(~yTOOSchqZ_i6Gwys&15X`%f#7pK
z$ldqulko8$R=rmqOhG9h#YqU(5K+aI%Tc>TD!`kcwvTi4DGew>8tg)hzMo1wqK9M9
z=EZo%btM~C;o3fl5#_1vClq|AQqtL{3E3Q7;h$Ah#+l9GPF-F8eTmci^9p7+nyZY1
zhpc|#^mXsNyY0&)3k1yl_LaXN<jKDF7qa<4ahI|gHnBJDo9gV5McTftULND<|IYis
z2g|;HyeJ=5w%ZT>@b8Nk*^kv^U22f5{L|xSC;T^jRR?<PXGb%~q31vP#Ye%IbGd+&
zZH|MnrhX}s4mz=`HW$WdmQKd2H~W>QeJuVC|9c((g&_m+cFpGCJ-Be{?Apy`YX5Yn
zUFXkPw%2j{xPG&<Hv8fHUIN|j%R@wl-4JiYlkA3Bx||jL#?6H#Beyu`{7st+pEb}^
za^-li!Oi}R-62!1?wU93md%AVBiDlsir2lUnhx<AJ+0kZJu9W+5DX96ZJUDwiW-8L
zc5inC4D?L+jolFl2++y|u2jhk_7PXETxmmME?2}+Fjw8(LwzpjvAC0m1rP@*eI+Y{
z6g$In5meqvWrJ*lBao%j9Wye}$z(wKsOUjHgYFgfnvIEPFj+7YB~7G_&3Qby%*IvU
z;I3aBuJ+mZ&4nReQPWv$=@WbqSs@|ri3t*cexAsAQetXZjjAnn*W=ye3r>EcDLH;o
zrx(V>ruo|`Q*3&^g#{GEtd`B#96Cj6Impd$W<rQ$s`wWvUuWerzB_`Noqlaj0@mCr
z!;iR}K)kb=SNhb>^C{sF#xZ9L9G<@}vV}hLfu70i<0W=q(q~wxOXvV(b=2-pz)_9p
zHao$wx5z(6?|>uOgSI&Ps%dF&s<8*+3%@Wp&|tZ0ec;J;Rv&Nu8h6)>MyC$UbuABC
zQ_!34M`=>TZ(W%IXzAwgkd<b%Bx#n?tfd-M_7nx4^;v5?Lj=NmM6M8h)*jD_K*UCd
z?X`}eEt%T;t*iRVjYn^zF<)L$UoXeu%4~rpiy#$DAE#|~wxGwiY)v&vk+(5uoUpaN
zPv*fpJ(T-I9IulqAnQ_Ew>c!Q$=HsiNL*iyMZ^w`l&*F|(v5Tw$Y2|@J5GQ@Wr?J>
z;tsS;)ymW_$HjZIgD~~p*|x>u?NpfV65CqAtMEpb;<n}98U7pFo+H*tUDjzk;(%0@
z*qshCT4z)_2)nkc>P!-`G|>C)`FIwqF0S5Q*j#uR>X9fdM10&+HB@R#9ki!ycg#=b
zjw2kwr5;K-R`v$gxnH?2dvmr5Ho{vue^MwM3OTg7q<i?^{>y)MIEimED&HgdjkaXH
z^*Py2Tx3VH)q`Z9IaVQ@(@hBHa+MRy5j!5>OQFosR#?mvd1}B}7s0T>`W>HWi(MK%
zRn1yO+l%tsPAAS$A`sAO@AzDG1E9b><4ZXk(S0EM7pEn)#40=M>x1~(W#^)A_j@Xq
zU3<^Ju0{HMKPM(};e7s3rKb;~lTJGP(Eo0eN$Mm2yQF*mIDsNS=C^jiNuiRH?P4J6
z3V~bH1r2`^P^I&b^8J09kd`={;^$$X`Cu^Ca4r>=h;e{BRhrf>982iwRJ!an`*QQy
zY3@I{Y2d%<A8B7XmPQj~{`Kaw|5M>auE?ohE{RLIVcV7VoLl3YoHBb`2farPOSt^j
zr-FyrJo_%E$|0vKP*?Z&@k~zs1W6Mc`>OpA&s1myiwOZAC+x>;=VvvRw%6>ZiX!C?
z=^jSzxBPP;u!n2&7oSiUX)ElB!dwv9L>-%}{0EoN64D1g;Z`dJ-@R?uJW*Jofu8A;
z?OJb!c8pzqUB0+ys$Ks?VLF;P4Q_bC86A1J$j`^q!J=lc8=okw(X459lYh#M)M__B
z5q2~fD<z1Mj_sC&fUghPplIPWP`y&+txv?w+1jv#EJb!(&SqKjJi9$UND$<`eTR36
zch<w~(!o^=)vHpDp8YoDiA=?`UrsDef(VoxINV^F>|m((BbzR3JuG`Ac(K!lCs1C|
z)kZu~m<w#-rSZtXjI)5HYDyXAbsH7n=x}^F4vjI<+OxV!-Ti7gm-}pN1w1Srr{O{D
zyW^gSYXi}-xTcS<mccXQmpIXK0ZwpKc{QbpSA5r?!~>IjB^NX-v%7qimZ_k_lvt5H
z<nKIU)1q_V)q)x%a#EkN=~v7|@(%f`Y({)5j%c^d%n`L;B*@KjEd8b!V_&v8xsofy
z*5r=`NJ#S>o;H~;7vu{p$hOdX5ci<4_xl0{m*{osQR*R{c{&DJJ@(k5cqZH)1&;@~
zw--O*%)axtt{8Jgl*?ET<hWeUR_i}VvOtF8dI!QIv^uTviIgcQ2HvDx{Y@1JO7G=q
zUzaB2oh2_pi_cJ5GOX4Ap4@`W)$Rjvt1ZK{mETaJ6OOJcXG;YU@dqkCzdUAfh5m{+
zmsz@%Nw%|P(O8{5z*}Y*!J2G=;itjY=C6{?^r6HXq9F>p;>oryMl5kXj3u7!N*Q6j
z_b{u5{70nBNi24EwAe<U=qfUg+NLCq2ccu}C7UbQPRTzQz8I@kem%1d*BYnSyYE|I
z+j9Hi=7nCuw#T<pE6Sgge{Dy#EsFMBS|OI=MVOJD`L#lAA-!EWN@GKVJ?}6M^o$#C
zFGRPXn*&Q>ZoSr%t4+NUl7jO2?(7a>Mo8Y9znfO+I=9}ODFtj$v%;+Wit>uFTlyC{
z>VnM!cIb%^(lUKjy&ZO>7HRSoJK~sFiY>6On)zO4PHY~OeKcksc@oa-E!Tvi?vD9P
z{{zzsjx>Hq|F0A5M0TQ#5Zp#`{O(Vr)UnS_`4m*}T)9Y5<HQ#n3w)t%^}4;22uB9p
z&xFWMSdOs+b|&#i|5{qf!UNg=LJDG>&FO*Yx9wcg;B?MjdoMQusey^&e?NbAHK5h9
z^NCz(yX6G+fn%RAHkGq%un&D~N-&f%S@^2e*+)5FPZWU`ez$$>{ZjC-oZuI#$-t_Y
z6Hii_#OjMqAk$^o`XncYY^nqw3J(7?J7==;eJ2m3edY@`8k+jN+CG&VSi#6){#D6{
zFJfR<SQlR_hL?f%%W5!%OzFTqZePW-;_p;?*kfPEGj1~0q5Nx?9B6LDWA;tZr@Z+D
z@V8C}g9qDp@uhR_h3f75T#VXg*8PWAr~`;ONA1UGFHMQ!Eo6Q9$w$Ia*lIt=`<P4F
zKfmPen1YWJ_o{7ira)tg6H;pRtF|~>BID)>yLwBx@DxT}vVqw(@d<`#qTYm}e!F%{
zn7jT!28Shfo%bMZHp_H<egK_Hk*0lo!xm>=Bu>CZ6-sdZ#f{Ng{G3uxC_Da54g+2f
z-L4<M*$2Q3r7eC-eo`!Ta!WaEgW{QdX=U(`m&2|8Rnai6wA-SErVfuQ?V;TsU};Mg
zo=TBJMgF;COQ?a-$H||nXj%v7z~PtG2Z0W8Sc`vga~n+j&}^jho7?XVPS#-we>n<<
zZwVfY98Rr>03-a1Vsuh+LOD{6j9$bENX3LdmW|4xNWNQ)hhw%B?(jE%|C^1i+Dog;
z{lq>PmyiI1Oruz1Y`jAv)(cC>goKOjOd+f4mrjC7TjHXv$2kzaO`Y8p-|8+pW>dBl
z{&n`<a~9aN3XdMsCf7rmVvpMN>_uD&Wnn2azNK)hfBmb?%=t_q38FaPS=qR06)uU&
z4mKyIKtML`YwoE#cS|tXlQB%qiw4rbvc*ak^aYOJnq?Sj#7r(sYDLpbiL?7+8VWl)
zZTCk9`V{?R0eX?+75{BpoH*o0le0TapFQ9JFdIsP)J6l5)nUj=iJR)LcqRr;cE?6X
z7eS0Q)|4}tPk7AI00%A!2D;@k-67L9*LF(@)tV5)io9y=@k{koc3VdR2u;4s!uPQm
zy7Cz(8BTv^j-v5hp2(g|+`2ruS7syBPs%i|h=tX$j#s!(gLL@;uZh-Jhl(_z+!t#D
z5H!?S+8*)=a+Uwi)@_M9m17%?qZVO(bnSaz>>|OBVy$e0FW0^J-V_hJjWHDNxh0Q<
za+7X~X9smgfl);#Zr<W@1&e|#p%g;SrX4OIE2oZRzb(1=nA-qjt5e?}@3C#U2&$K~
zJ*J{`ps(7F>Qz&R?aZO&fS{8FWbUdaBaLV|isRgP{)+X>LTC0u&PX=^sD6bmdg7Uz
zspP=C&vyH`*yK<??9Gq3*$y#={1$q1gVbO;EIGdq1SDmZfUa`EwnMq#&N>k~UM((i
z8y(8;R-?aVN8(Mnp?GsS=C7~%M$f!tv6NVSb~NFqCbR;`_}7kYaV8(FI6kiauWl6%
zF#T)C9n0j@Z^h&FMBw}Ybo`R_$4DN;THp$lp30G=pbm08ox}!1|4#nBY_^^8)rNmB
z1#Jc@@HJ&OF}VKjY`iI4cB~+&8_<QEOCYDnLGQg7yrfnxt54<5AUw9b)Kfws|KcPC
zaH!<;H#_Eh_2mh<nsNFbw-2f*zrmD0%!zT2m8^5ztRGc(LpPa!>+R#JmsbW<O^&>w
zcENkWd6hCAb}`p^MoD40%0BVaOEpeEP1FosWS{w0a%QHQ@#i^C9qFSGvA^(^pnSJ|
z8Gj!@XZzJ<JTbpg2n_<p$I8C0-l*Q0Bsyf4a(rG2n7Jpo>|L+$O*N0>sAXVRk*&XV
zh^YLn*x&CG8z{#W7p*P+y|+-B9&D~ZWFHt;lt<wuB<3IE*?we=BV}tHwx9efQoP~d
zYd`0XoJMNymbP|Tc<h(>tV9^Fto<D8_N4Ph#Vf8X;wyI5lVO9x{RIV1{+iYX%C33R
zwGafx>wvq~UsCFl<cxORlfgew>9SZk<gb5yjNm7mdyL-TpUG8-J#nM=38O=DicY)9
zKTz=fGO<D$=jP}{R!fyx5+8HRljSb{Q(MP_Ht5OVE;^gZr_f=yMtfaQ9{qXPZu9nC
z1jpL#Ii&Ovq|V9Dmv^J^=&-@wR4GzyNY#}BD5QzqZ$qC91<CTZmTrttHtfl>rKs4i
z?+_@#hWk`GMqWm7G{QR`=s_AC`DChy<QEc<LQ(RlQ+91PW{isAWt$R(u+MBv^x1oc
zaxpeG`@D0Jje9cq7p0+Y)LtGR-<R5makxyVpcUU#e4<1}Fzyq*XMD~{FJhDY*Feu)
z$&`1+SG4CkDtg!GZ%U%0uB{vPuHL5Q19%T2pXzOTeCwVSh2aI2N~|;^$9Q)OC&5gI
zmf$Km1*&e$Dx39WVSPGkR-;9p6Vq`T#(aV>IyV=ysVM5_JsAfwaO&<VDHahX6!9rZ
zX+gl^B7+C_i7nb=3muc#;t$$=6^?X@`s%nqbbnx90T$?eVl>C1+P%#B9c2CZ8$<Qt
z=<{L3*9R&LdZIOMuE1bzOiNm(!bQ?^ed2;#JruGni76Tru{iZVsPY3%i9WdMPFot?
z;QCM)H)UsRNs>?v%W`<OI!K5u4c49=E-6ceLMUL=jiI<I&EC(xF8^$xXX+|jp0J76
zky3(lcV#ZaRMJki*y_uZEf<E$h^zR*E4IcLRH>r7Y;A%^FPIM5d=CXiy>>8yJRcVE
zznpY$+q&GW@^pa0*FRYVf+cM04f(3RLE&k(F<>bE#7su&V;6Nc1*Q~O{t9oz!EVk8
z$YT7UZAoO}%c63i)XKK4)n|my(UpYTZMg&}587>e>}qvM7MhYrvX`TlRJ}M9OCXkL
zxFh;eiXok6if-R|8DYh{gT>P>gYDRr2%C*`@O;iGY^#@Z489PZN@F5fUoqi5)g-(S
zv)p%gS0r0f)kPvOcRG6&y~*Aui?A|hs`WZ*7fzmK2NF5|4%TbzkiX%wvBNn8@+YdD
zq76Ob5Tgt@*lK;zK$}A;p%pW5G?#r^_(|Haoa6#hy&bQnLV;73N6FbvI9S7mTmO?`
zv(*93&E|K+DPM<9Qe_~JpN<~(4eU|keDrlk?45Y#_FpM9_S>02C#NlEVM(}LuztV>
ze!U*Bck<%bdYpYSIQ_^=#sPOOyB>fYfG`kV?LEiQXrg}KPiSPj1xL!}aNg&$0^jGP
zAOd$VjzwGNZk?0(D3LP{z2?W&SaP;;@$5q4Q||NxQnc8`sv8w~7wuyv%SEh3Y4M<a
znjMuiIy#ki+h;j9@CrjI3H5WI#|qRfU&I$)VIq`94!x+D&HYuh2o4c^9k8#X1%60)
zg<VQekxX$;<BI(zx9wc=A*ez`9!l7UyO(jUeOCb`8z)xJ2Kzq0sl_F%q0xTGUgaXT
zL_EWf{-iP=-f2H2o<UHp{hW|;L&!B$k^jHMGu?60sPq#bY+K8Az*q9Bt>vU?f@isU
zYq^xV%MEqS*1Y|-=N0uqCNg?GZP#uMoe>G1$qs#8{+!y>zzK8xR#!qv%e2RC*m|Yy
z3PBDMy3wD(3yH|3kbpnhTKFE0s7||i>)E^9KKb9WPTDO#$GHn_P{6bgZSg_7)qf52
zOi`V#+XBK;GlcUD-!i-1ThxHk_9nZdnvvqV2S<CDONBI}IeyiK<SaQSln;PPZ#L9N
zA3D^AZ4LfK!9Q)v6+GN=j-J{^RG-Qr&utzf{dFB#8Amx@$oBqc8{?1v)KF_<^DBf&
z&whT~#`)@zy$!?5_-LTLeS~$Riil0vnwFarl5;pc6js_-R#$#B6CL7z`X`%|T@_`K
zuWyR~Q-Usn(_nw33U->qI%o3iyKVZ`I88WL!+<}d0!jK6Pf$EFmx3A(Fhz+d%+zo3
zf82QDn_dK!;`YeZG&i1!n3U{M32f(W4c>Jc`fJ6mEr>>Lg5aJ*=3eOVQ6sp)?yG>&
zK_V<2u=}?Lzl{#<L5{ygj*q+N+G77}&L8ao{|?&q5*f0<)NTz4p<NNN$P9P_Tm7%l
zBqr?`!zeVrF^2YpQTvKIpecuLqLdhi$I@uyX?$sk$l2oHqu=YaR$n0E-SA^<|J6{k
zWhl<(FiPc)+?us=_pq*t0OZ?W0{S?%tE@t{+(*DaqSjVMr{R3SWiA$831C;}($*{N
z6q#w2t?>_M*KpFWO>A72bf|<CTWBAb1Rgwz#K~f_-2SjWSCH)YCDe06&|oJHakep_
zkw(LvbmciBbLyt-nH%_Z+Z;1c%)pik&$6{R?zdK>w3MRawk<jqU#mt{`fb~uq&cgt
z?Lpg-3p9CI&1%l!o%t3H|8?7ygo}=ux)RXN^NIQPHtNRM3%Q;o_EEIjZkXtv>{1`0
zCfZ%iQLBTvgaGAy-QEhwk`<-6RF%Yr)?4-K+Iar81Aa{Wao#LDlwB5|ZJqy;jFg1^
zup^3y$vQi-HREVKF`ITeE>nGpD?He0<Ue*adXsDB5%)n30FPB`PKCu3Z;waw!f0zg
zhz~CQod~F1(<fK+e(TTCYU)K*4R$I%cP}*8>pMD5SCf{GrWfvGzrEwsGSKs<$@A<?
z)xQK~E^tJjfuIcXlG<7Sdz=b1pUZ_*J_vg+X`5^<QB(8%goC?O(F)3QRlKb1d<^58
z8><ys@((K5IN(@lovR=I3LGgj`|P6}c9!%ZAN%>zw9GD4eUe`l?8+{AF}5wIROMbP
zNk$1ZPII7ql6^ghUDiHLcsbKNPF$)sad@94Z0H+gGLqQxd9>DH@JiVveWmpQ`yx?s
zH}v4l=$AQ=GWgk7-W~RmHr3`So;_b*247l6nCCvbl+Q{evKR^6pmVv#{Oc(Dwt`Ey
zr+DpG?7Mgt>~*Qe^L;|pDYl`pnQiui_k*7)HuxW%@Mn_@{-^ldFDdCSU$vioOE!o0
z<(Hfc*FlHbQ-$@xeH+f~s;A;W54#nYHq6meg+m$W8FHsx^HlKac$c;m>x5+bYoBsn
zM}gO3VRbxR_f+sov*GVr?E0sYIXZ@t{)VUG)|%h$SpaW*%6Ts7AT#fUd#~ISA1eCd
zwA~yd!HpG{r^Bq|E%8jcfl`r_XM>(9yxsgRlt)jM3;8nRc&FVKUmWn@Ccf*p+vAyh
zIbLB7DCF&qfJ$p2Jpi>%d80Uz!8vB^4V7{j63=wn^pzcO$c83D;g0JXh`dk&h+)yK
zlr$_mB8@B@vJnh-uu0xzBfPhP9yygYbF6dQjC5RJ4Mc-rqn-+`2D^uLW~t^a-9C`Y
z4|>c~sYOG}iU>b8aa{PHA^~fyuzQSqD)_WcJlqLJZL{&YdXtw(W|$D4I#mfBsNsr>
z-9#r20<;g>B!|8ms=6y+LpRNuXy<E>sic>498;12knRxtm)W$KT1S_ZQ`4O&iQ(hM
znc>}%F~nwGCO^36E>a_iKxaKwSo$IVht2V|7L$^>Oy|a`kZz58PDk#`HZO^h^WdZ{
zsMb>Xq_9$Ol`PEp6SD-PabLBxC5i=t*>Cp;PK1qDK)){nZppT&LbQTKeA@23wmAB!
z@Tl;vs&z4w;y9E3WVP8Ey_k;FlaOS60#?&q&pI{c&lPNDif`q_&LFqiQvXFvJ$5*2
ziFa7cIXYW?-Ei6&ZS7CF>av&X&GUx5!eOK!WQVh;jjY4hggo~IV|4{+zJnAWbl8^X
znvw;#IOwd5nYkKOm!}ks@G-VJ!JgN)3aMa?1A{k4o2^Z%RSeRq7JDdrB}H-(&(<YY
zipE47#YtNqBN3`nZ1c)!USPKk-gS)(XB)FC1cf^Bxtn}SlWXO8v^n3-MUYhn*DXE}
zI{0=Dde@L|&1dZA<;|XoX<G$QDcT~J726)qWcz*zWXj)qhff-T*3&jSSudDzx9!TI
zJ7Ej8-=5D6DfX2#nM>@2#3dX2W#&dMIBg{kuAba#MJjNyM*7qvGLXYx>`q*l{MNQN
z$7)z96B9MKa?dENyG|8-?Lf{5h6@85<bKQ!B{8+sz%Cqq>Z~fva%?$iM*LL8T<Zi%
z>F7ut5nCY1zUVpqHKnCK8qdNDO=>+mmQ*072sZe59tg|e=1(Lj^&|$d{^<TiaB(Wn
z2U6>}GD)~Qo!BH=W=XM~yc0u`>P|aLYI-}9Lkt<V$OZ!4MPI}D1$_s)&sNZ8!Ku@B
z&Y^WGLB@J72b|mm$ts8Z`wrwl&+JY+?+ZV9>fQE1)mh;Y#pBzDxj^LH=x(WRwU6Tb
zLm>Y+J656Kl5X}w;$lh*ZfzI6>F;6hK8db-fn9zfN_q0BzwZ2x7W>Spf@`H)>ZQ;9
z2<={KUu1{HU+K#nu(MsN8d`o8n33sTMp4Q^aTO>V{n!3w;g9#*rTBG#IG)$-n}9}#
zB?%#lf6BM<4E-I~i5zVY*msGi`BUwC|C_6>&3>p5<<(tY=HUF%VeW$I@s_o#5I=wR
zwWw)b(v1gcr~Tq{9y5BZ*|xBR>1eU5wmH)x*Ha7&#kJ(>cqVhW3{9xZ?V4>V<l@f(
zQ5AOY+HHRKUJ<odd<9iEPd*UWZ42&aIsRdAKidim#i2pDiEU>m6>stXjtjcoxGgUC
zy0$jCdfv3n87SXaekZxx!_C=JKIkaeZ`l?K3M@I1A#2dK!YwJ$->r!igbNAN%F%S&
zwqOqkc}Jz!T4T3Yt)ylx<;Cod#O}RKTug&~##kw^xkc+t?FaW@L;cx651eLA2Z@(B
zaECc$nDDA1j>BVGNE*1Xx#yz?!;GjpEj|rp`9@Y?5pKb$tBBT7+k#{5YN%^!Et!R(
ziZRizj!-$Gp&8^~Wv(2X)2M4>;l@ReI>9}g6m=@Rg=io5y9v?Le(c62Vpq~jv2joI
zc0`X+?!qQTPn?UKjFLs)w!5Myq#e$|DOC%(_!n!iP1}|dYHAEiii_zv(Q@D6QNu)M
zRH!OabTN@>d%k8fw*|Ld(y4Id&+^5o=?3X@;xj2+2=oVMaKHN7<|g4_Rc&ruW%FXt
zl7yhJtw(G@wrQ^^x#|}tsmKzG#Shjg<|wb5gLZ%RAsYbWs07%eYAs}i5r6nHYaz?%
z;@m}~i-L|GsAgS?W5_ILwcCpLvaE9nw^DtQHc3a|m^FN3&K=u^Dt%g0wqe&`R@LRQ
zH2Rfbs6^&{=-OIxV#@!@mDn0>oc9%i$ZD9&M@rZO*0HT{q1~&k%h8KIpv#s!d;>k&
zdMmG(sgz1O%ls@{?Oibq^g_0#!jb|pi@ws<R+D#YB#Cb9A;%BL%2G~eTj!k&^gP%=
z;fD3TssH|)f3pqwEm|Js`cZ7mMhADsJlm9?p+4-e&Hk%cdOK`OK%9)A!cdD1*0y>(
zez8`zE&CuhcBgIkU(KzBb=lz~O}W>0W^YJHuX1d>uG|We!E9H&QQt1hkWjB=8G8O#
zh<BIK7ZR!UZR7?@lB>Z@fxu(@tK$>Q8n7>g|F*qVH*b1A3%JJKm;!!3C>suVi*h2B
zyA=lUPzBZ-gaOFz2-xLTIP7bTh8Hzb^dqDEQR_<rlb2cxmYFU)>hRWK4tdayB}I~=
zS`~Qhc#_(DkGDG!-QlOoj_D7;C4DON<Sjrwm0fTnqea0y7L62^vrnR+6hrx8h<m5F
zGS5`kmN;Y?HA}MX;-mH=%CP|_sJ4<U5HIU$oKog&nVpM1IlfSy!`Gu+e$VNQ<eK(=
zyxvdTiDDm5AYGjg6l%2vf5}W%EPx;QhlJeOhmOt3{AV9!Cz$X=tP;^a_V+mEjkXI3
zl<qxk@ry|fwcKByR1iwn&JD;J@u?#ek8T3@>@#mMpv-u<H>{Kz_w)D)^n#ai1lbok
z&SWK*<aGJ+3hI^bhRo064Dm3^CC5*Om`efmb$-o$#_zn-F6CLqZmn&wZ*rCdI1_$)
z=5@T7zpYRuPxz8Fh3!SMRr@aANUa8gCHEpfIKKC})q0|U9~`A5q_rP&06IXOTDtiu
z@4~t^GNPnw{?Ca&g?|2$KWY&}?CD@Krpb6yZ&y7XTn#rGI-RSZ4!)qJrn?oHs@|@7
zI(RC)Hs!sj5Q1x;jw`G_Dl523?YeAQ^7){L00GxOolJ}Z>7_)|T_g<O5Ur)<$nw`W
z=J3R`H`J3^&u)6!xiTHnit;YJX4w+btsStNpN@-gMMDd#PbS0|&@Czo@l6HApy<kF
zSDhhJ{Yd(k)`r^bPX{jvPefLsrwao$8=T!9kczY5hU0WRXm?b@=ysLx7HV!A{B$sv
zrFL~F=!WFPyl@~XDu*WElxbbTJ!ZqIl|g^TQVf5(T!RBW|BkRRB44&5AJw8Zdb-?9
zcTcua0UREpC_;6lykTP!IOZkicqLeO+8P_Jb?zm-jeGj6N|bTx;Ea^OlM=BFN39}S
zO~{W1_LSAb$IT}CGdWFGmXk@*u3V)cRcV6n@>$?X&22x$XMr{B^t(z-IxRkc353N`
zu=_rn?lU9d@r*<Od0=g30#>l<W;mPWkPr0SJ<aC$>ai_OLw0j3kkbOc-{$4|VhL@v
z1<}G&@hbg1c@lF1B3&>>(LK8_*J&!XwRD@_?}RZHW$&WgJoBjC)MbkuoYB7@Ki(e5
zS2gubFsHTu1s)1sT79mmvMkc3F{j9p!*4=Y@U?47phy4yccW}+1)zMU%EDSpqG?Lo
z3Tv&#@dO~s#n2wl+<Yi++CJ+D3SmQWyY&8p(i1@;T|S8pIr}aT${-)9t@OXym&<H*
z;-siiI3(7@5UOk?Aqx@Q)>gy1*NJ@pAId31USiYMRf|TR2V3t$u$-N?A-<OHu=o>2
z7H-T@;M^sm5G`e!5<hq+x7z0X)!u?RdW)|Cxf>dEr*DnVb%?wIkO#RPw|SGP-C!Je
zw)@(W3(I!ojx9V0I&ItOv-bR3stC6$km#ILDI^9Gu|eDOx%v&uYH9dFj9hg09Aa3$
ztmo-U?e0<H#f@^eQ-4E~?Y#n8;s+5Y4p?usYF;Gy6ue7u8?*yS1dXi5p&T9M2t4HB
zTrId0azw-2B%W7yTVFNJOsao0arkGcVvb!NlJnb7KtxFNsmFcDRjue#?L<K946qjt
z>i1!A!RfG5{`b4<Q5>9(E@kbO#_Mvd>N^#l6r$;7-!uN@Y-K|p@a4e@X=igl4sc~_
z!Z*gwCFHY|LFK*t#Zhz6-jDWPdrh7&hwXd-0*w_}QLg9ZTCfkQ-o?AhKJc1-n3y6$
zP`Xe8<MRnA1UtLM6@KiPS9T9&oxc#>6f+P6OZj2D=+9&X;Ba+y<0lnx(owiG->30R
zC$+@XQ(T#!#WOhqc~+W3_Ia+Zl&Q@GjDFz=YwK!TX<s@HrmKn{HP&<YwV?Y%iS~7L
zS4ufUVtZJeOVxVIOWG^K7OSOwlRzpbDwC0C>$g7Eco`qR%lBw~R927bdG@`(ALv=M
z#(t=JzI>Cr5&p*{-g}lUv!9Z9$(Y;SZa@1Brco~wh<~XNAp50nB-<V=OQ;7id2LUY
zrNnzG(T-ic-4)^@D(-7e6|F12+K}0avhGs6UK3w>qIbVCuifq(jI33!F)vj@_Y}m}
z<v7e$G<XCV?E30iseZ!M3#Sv$q;XJur|hUVMz6c^Ykm7&u4y3A->2=S?V*RFofvYt
z{gpCCz}wIJo3|InWF8y+mhHj3)exN3WrJdzqCf{m>YLlGi5*nIT%8<7w{3T`9h43^
z-d@2=r8UNRg`IuJ_QDVh{`+5t+29=F-i{@lfkU<zri8@7Fy=$6Z;zH}Fl7oG=Fru1
zf)3vvj1WV&X6B0#F>gtQQbwu1vx?0fw~^VK>{rZHIaEgZ=t|Pm4d`R4Ch|ZjsnTrh
z_QKG@pv%XV0;JJKM-_Z^v1pC2p2_1vTfEOE_)Hq8MzX{v<{HcN02U4x+pT;6Cpj(#
zdM5qvc9(yGw{^{fHYH&}X&P1)o3=glaLRbh(aRD|4{RenVNaP6AITOjU5PG&nK8D9
zu{cJY@@D1O^P6o>;_b2)zEIKun_In+#aT8$ao9>OGq0Mqte54gw*}h^<2YloE%cRY
zt*x>9a$*Zv;``$(&lv2~x+tD0%{w#+wNjcQDdP_?fUWjGfWa{TQ25DF7~*0ov05MK
z4|gxHdjFHy4r|<A7<X)3kr%ruXZ!a>w$v9DGuJXQ40T#d{w@htQQahrux;}E)!uCn
zOq2Uh3{Hg=%&<UA1$n4V>`+&HD=vjIpyan(UbT*MrQ}ePqI6}nPI3~Zu69T%6i3na
z_%e)ZSGcpRb)>r6-9t$k^P9<XTcbqj>l^{N9FX!JFH2kR-L}hXe1lJhbWNQ^=55UB
zV*^pe#Z58j6={;#Fp6N_?2A%^z^Wa(#k-jgj$s7028QG&^|o!RI&H&Y4c;eDiq0Ap
zT3}b~@YVVw`Qdgt+1&f@w#%obgr0Z?Ra-o?a~kvoZ&5?#`Wox0@K=~Fw%&`jJD$m+
z^eTXpCC~OcG!!#+$a|xWD`+I^9<l@RtOyElzX%O>C`ni5@Wa($@-0G%#kGAT8+5h7
z?NlSn(&KA?G?3|xi1v?rjr7^Eq>A7{o!t@*9gluAg!1vdg?Ge>3V*Wq34i^LZ=n<R
z{#4TU{MmN8!rNu3-QLOJBx3`;)Z?y7JLAKNXMmZ76M~x^$nCd+lnt$RHX%}2CG`FV
z#BeT0K|cW*Qnd-7-iwAptXzN&p^JIHx^mbe7(Mn#iN|~GVAan%AK$o4T@EVyz$e#2
zwgZUyL;r<euUtYuiuO^ui8J`9eVjB)k2vlYrIEYZU#LI_XppJ6(2I%8$@Oj*vrn7?
z2mjSR^%cd41b_NjMbob^<IiId*W~Qyet%h~1ESp*Nd_x`@5}5R#9-m$$ybScxq|4Z
z<P0kZpnaWlm)@N_=LiOZOV!p<DX=B>O|%gS$?T;Z<lpA{{29F~38C{RHd>OKe($sS
z<6Qe8`!7sHadNjG6C?0>(*3De!B15?Z5mg8{G88%fmNN+U*ee#EM?<5$@<ze<<S4r
ze7owIP?tcF#nIyGXPj5^Bt)*7y5^Z+EB>Tg)vkS}FdToJwZN`>CR8NYu*n7eOz4t8
zBdcULJmZ{_h)O=W`yKn@cCsa^vo_LW-1JQFY~-M+p<oz(($9oABsULNV?kE81au&l
z%UtKhaT}CVsPCvD<a~+U>hRrj_k6o8X9Lftuw&WiZ-1t+Ay+sH-Fc_o5e=2{R8{FI
zXK#a_DQwR`&)`Kiq?(HMqPrLCXuuUZ-UuYjSC@S%Y*_S(&7qi$$joHJ^BH?et8of3
zWkmF#L+v%5VV#I8Y-F^Sz#*;FG0f|u;#pxifkc_5##HT;@I&^&12#5$Ej|<nVI`61
zbsgEbXbr2yg(pqp9vhzvNdn9+n~*J#GUd@J)9=LU*~{fLB4e{n%4I96#u@@P?|P>2
zgIr;FIZmm-QtYL;s5fm|K670+7Na+9dQ!u3^1z|)pW%pq_y2@vndyr;ni^@d;#19n
zlL({29-EV&5<X14xXn#s=v>xTEEDr`5$Dgf1<^rK8Eea6kz1IwT`c={pTj@UQ_Hn_
ze|ApE-wssBiyTQ(|H6DOPLvfR{R0(&9Ac@8e_Wh2wbhJIz}b7`P<{5SveS@kV?qJm
z2oh*|rgAW^YiLIsgwwS5Qj&oxJ*IpDTe4@}nBZ5YiB^Y-u_07zPjn#ep{XHQEIui$
z4<{Na4y`NaJoVl`+VY$g1)Nsc%4b5ciy5sOSQ%E}t`4N`MdVu(3p8eot<6~>c`YS{
z!9x`Z%4NrgRPf6oTbHY{=uTUoU(s_1Jj8|o5!u6IWm%<-33fR{bP4XUP1(oqerKDj
zZ(N_Oj5%ALDf#|n1R$koc#=UPTXTj@I)}H#m)gF5Joei5Tnz=Kt+E{nAl7xM`FAE#
znI>NG4Yn(LESZU)uPEWL;)>af>FWGKeuxUGR8H=xz?S~T*^k}T=U%i!qLgbM_vSFu
zWPRw3ZzDhx#2mK+ey%s7^6|fxFSEnGK5b;wIpXsoQ+OMtU#u^BC`NR)gF@Pl=G5=S
zv%B4nRS=W5Ujf6%6RE`k9z9?K&O{u{Cmc2i{}^!eJ4Mx@SHRc#nbg?30H$kGK%dk_
zhAZ9a0OKH6lVYxss_mTwqpo9BdzYQ@CbHX6#!PXQoi>oznA^e{o%LBZQU>>2)gLS{
zBT4dkFTj&c{)D}spBC4+^Lf*VPB?&wG=lm5AbKet1HOv(Va!CAz9PGeBO0`*`cHT?
z+eZ$>GCa*b&K(SmENQK2<!HQ+bGlnzS#~i7z#Lx`D@x<`NupcT)4G`Cr$G!FQMtA}
zGFC}`KJ(t1xf?|dpXZe3AZpqdj(>DcR1Equ`Y3!4%M8P!p5RcwdgkoJ|E8OYeVy2&
z7;h8SWj~oNRoHWXIre~k<NI|g22=Ys-b+1LqUh1)epfwv30*e(k<0G;ykKO=D@jIv
za2zjaSZP1za-gU>q7S&?bje8Jtb)uxMTg>iq}IVV(SA<A;6at_>@Du(U#?iII_@pV
z*><>2WFL8YUb3rp1Yc7FD}p2I)jOO=V$0x6;4XOA>~JdAHp23w3@lKMa$<;Cc-5}m
z5j<3P$$EEPbm4)fTG(TD{f@$wjr=#eA%@@ok99a5-{`*vdKSpGdDD*I*f^YEHXH2b
z9l@r|sIgmel%}>;q8tXrC?1O-H?5cK)_7L9QeA?e`L^nra(%ftfBTNOgKF{0#n}R^
z+~FgQ9)osoaDs79<Wm|FeTdT_{-*dP4z0#7I$b#~+przM6qb!sRL88HDqs$e*5b1X
z!GLKzY9kUdEE2DS$``TajEo-CLe^u&)9S+AXQSdx@ro*>POcT`;*gDT(C_)fj0H9}
z`Xsm!<%L_H$K?`}%0@iy_~^+mEk6WYdoiK<TB;%UznYl1B5088m>r}wob;;>_CcZ$
z=NaYLzsr}i1;L-oWXg`hOgD3IO-mTMs4sx|Xu7{zK;+{L2Oq|`%Vq`uLUEovvywz8
z>q!|5ZxQBiTyBZboEu+3QbnLqQr>w9c8!9&ifM0w4+O{2U<-EyQ_hU)sRV<g&yFxb
zcDJsu`@QKjrfiEG)H(Mqu*I<vih~quaIBJDi))uX5bs5Q=`b!{R?hcI&|I6N3Z{}4
zyFS)VX&V|7M^ilkeN%<z;~-cY4&_eq-dI|#(;ls|BqD2xXU+f;42w9>>b<{v7*4!B
z-b=3afT@1#sE~v@eK}_>p=RE!xRJ;ON%0ZOt507kg!%?I+sZ^21(R%bjPoSI=bGH6
zvx=Q?tz+*ue;;BGRS+t7g_x3lTUSxFPB=-*B*ZSNtM&25;brHE&EkN6)i&e^T{SDY
zM>gheReg>INMut$)5E#5If;k30E}b-flg8}l`p|+ZS@vtA0C27*yi6Bh0=BheKtnG
z9o53ad-}%UUr%Uv3PEXSH8tWDc_SU&t{j@Yi7obg3`1ou27Mt3fvBb>4Z?d*wb;tv
zm3biKMYh{9{f7qS0@<6LHZ}9Bx0(;`u|Z=8;+fYwFCqPQ$e*EpYr<6GTDrsD?bN1b
zJR^<-bPx}>SznSxCwgY*fsQ&ZY96eyW7Wc_Zuu3F>UP`@`8o6KM1ncH1_`)7`@@{y
z?27qQ)uVR2buiJ8Vle=h+mk$PUK6hr&v0^O`TdYO^D9hAX(I#Y5d^(~gbe9Nn$pXa
zme1yJQE`$(<6L$FYle;i9j>w~sfg`+`KCyh$y#t7+WWafXTtp9&wSpeDV-+F_Xm!n
z(PPHihe?wp1SsXpKB@=^!wD2lF*N{`Db6EjKCULOL>Fbcb|EJ(lN=1C$`f65`W)!_
z{doH%kvt#5w@>rUOcX`cm-6^<w0-XDjY%8n?~9nC7z^q2PD=@<P*MAGN9ZQf2eD{Q
zzKRYDWpnWLqYeH#TPn!EV;SDKV%9GCFssm+5K8YqsE>&T<=aG_9A?qteOIBmGJvuG
zC<^TR7_DeRxmhS>YClvE6txo&?6)6xgwdrzK{$2p!0}Tw^tdF~i~bzXgu`QCU&7=s
zj=g$zC)8li7N$tCJZ#3Ro;_O<=)~Gz!RlQ7Y?|9t%vDCJUc2Vm;96>0y&n9v(Wvlw
zLO+byb=9*t{IHN*AI~yDin2o7@ND6D(ABN78=no{Y<6=!F2;6KwD(k_<zKrwpJCZ6
zbsb@OZm9rjN4d=rImkQ3B27xh?35^ABm%q5XU^Hs(!IoPufTLPb9dPtIRuA;>U$i=
z6=CqR!SXGja&DIm$xiQ9Nw@|Z8lUL!R%)^%?B!w4hOO_4_$_hlN*Q;Eqj&hTg+*L2
z%|>|V#ZtIun2n4N-QUmkQ^ohE?4Va{R1QwP%Wi73jLE?z2++C8VvUVwMNL))JYeH|
znBr5-9L7fzZA&SOr9qjHn4I6<P)eRpe73MOqklWzCOumi29$yL*WTqAE?KFjL`Rp4
zV)ofIpWLFwHa%!du3+p<5WpEwZoC=368w4tZhqz!5K9*;XYg4*dd=$9HYZ+tsEnW`
zz&@c>ba@jVGV=m+SH$Q+jxw_9*nZU(<SI5StHtqjVFC`ylo|y**nR%&o=ERsF2?&~
z3R>O6fP1;%%cAOys}ki}B$Y0{(9w6f_O&NqiXVt)uDdRPS#5$ud?wZJ>T^mY!)2f1
zI27OP#<B8stjPRpP4P^maTHYT=5$(`WKG88Mr-l48tC~0nFd?^mGU!jZnpc%&6c-w
zhj%11m4XY(WV;+7tV)H6T%J7<-eN0bx;kNHh!poy(oxvz1PRiXmQnovntWD7-CokA
z?X$Jd1|K?=5(UaPVh=^DUP=oRF^`JnxUCDyAUz)5l<Th;Rl!-8h29Mg9NDT6(>Eq2
z>qu6$$~NWP&?d4^%G@?PJt-sc7AMfvO|~_<51ke#@-YVCPPT2{BhKtajj+9fRAIBy
z(vwQaySaP1?R1zsJb&`8m;s7Bq@3sDOYH$kK5}?^;n_+i!<bxhIQ2v`?MT_qj$oeI
zoy|x+-;E-Z(|WJ3Am<m+VDX7alNG^^(Q5}{7ZjZ~XGSCpDAo&|4TrMJnr>J~DTC~=
zPvb9x?TB}Xya7FSqLN_1z8C~ovJzw>E7{Q`RF839Wyc&UOi@%3^VzGs1%|m3{^s2+
z?DBs9jLEbR5Q)z2)UU8#L|9M@V`8UsJ-a(<F<rkCC()SS+Zl(pzQqQr=_?6=ms>m=
z&ve(z3G^sh>2v-uCCcr+oW0~bDYfu}&iP5$&>@Td=X{RN*!7fd_a?*zlrlYqN%%0E
zkbw(F@sE6(co1~j$5oGqw30C1`|U!I1R^yX;bQ!fXIY<pl1F-N1B<fCK6S!IMxl<$
zXEFAs%LQfa^JsuY4zHlVp{{R<UY@aC(J^2C3cCnS#e+|q^Q&B<84?t}&hb|?)Zm+F
zmmFXB%$a82__PSYA~jHAL2wmYF4=(+3$hm0+xM|0<zQ?-_>X~}mCIV}$ACpdDvz3z
zQret3$^HFPc7YieUFOe$JneqvP9krW{qn1cXz!^Qsp2JVXL2Z>+(|l+tKyk-5pwxZ
zIm@ee#>vAeNKRx}w`+E~V(0+3(H^^Yr}IEX;)HC-Wm4|jH|)BdAt?`8^f$YHXYeSH
zc3XR=-4LJYf>uc-;)M8E0mnD)EG$DugCC?f1x(0d#pnMx%X9P2!U8ll+burETy#AR
zHmI7?9+X|)+pX2JeXzKy!`f%J<&?rtm&}SggEhk)kdvsx?#L!Jt*zK9;rs^2thHI*
zvLOy6d>~4JdK>Bhwbj_Lot4?*Kf0TXrC_*!S=&aG<T4wP0Kr{~rS9jj8tE;F=ava^
zRG@~{V#ZfubWB1eHEs7Y;2gWtUATlm<$08)_VR%&OPb>C$3>5Mrd3L48=n&=Bv)n@
zo8X|dmdfciu>zI?ZtM`%hMMG4*CB<W%<jr2a8XJ~1tVuu5(y<?PSKl9^RK4L;|e{`
z^!Q2^Ea@ucVl|_hw_L(>49RZHy=XIc7GdL8=E^&5R`d~J7{y^~PSV`y(W7i`{%CJ#
zv3Uvhlv3AtK~9|D+G4L>nDdaOnsFso-j|KYd$UXx3h&PaK)>xi8H;v0w<=Y^VXomy
z+ZoK-;tKW3)1(kW`8X-f&I5^pVnpZk;liuUT|&_!u1~A4+Q=k}vog19qpuM569p4C
z?K~?-brpNHrT#a**5j-tzH}~Cer+eLHKzmluy_<d&FM_$JX|H=x+76G9YXB#{_wVM
zwdK)Y${=J+x0UhCrEWzqke9l8Ch~?HP(rVXX9u}cbTBDb$=bkM5~~zpZYPqb-hO*1
zL2Fp4D~}U$U4=<0QZ*yIy=7{)_3=h(%o1(KIVnBk5!;Z^l>7`Z_277;&#JSV@_?Ii
zx^uDkZjQ#jmprxF7H`Z6fU8gpQy5;+XYuKhiE&%?rDrb^1F}7VD&7-Usy}w@EV982
zw(rjBL%)G=hPGXPmi%G9J?|^s?J4A6*co=n63`=)N4ZUVVw_;FL!}`U@Ah#<|8|V+
z&6S+;XX~v1E0&Z{L5nvzkl2Ekle3LnCk`detU&LobZUq5tWrpk9m!uaR8OEU7FE)d
z^k$0C^nI+8=a3!E0m`7|M7LuV3f#F?W|HIC>xw35@<alBg|@U9gRDP+qhLvwG)SkS
zX$jTk=#)altzW0BPNao)zw{!u+dBaO&P6CmbveFmXA;n&dJ$XxnhhjHc6L(;ReSHO
zgNfd{jha#Ca_^GKuL-U8d(nsdg$d*UC9m20@k~*)%Ag5PW9PHqWn^|kD28eBg=;|J
z_F*pEA7<G{`7!SFsIs`rkK4yFEg9eBnykP@-xL=TH<;0oUpnn#9)K7RF>8O4q>nw@
zK8>#B6DSRjQjLF>?Yr<!A)lg8mQC6R7yd;xS<2tBIls&mn<o#1ul!Sv8GK;s?Cb2Q
zZWVpmrEJ?Iy}~y>v=sHk8@Md>ZOqCK2agcrSYTnk3qYk{QYBJ}jNd2l14^k>m^J&M
z8beuEeoPdX+z>*>Px%d|R(TVl@3Eg_hNxZ<Xn#qd|AdfCjM#Ib$moDAS!P!~7u=If
ztI)Vu->c)9=)RO7grmCVxo{4ejj`LV^<Th9f>66I+Lt&6ZfZG)U;kV=&IfvaH{Ncj
z-m82r<taE-ZuIxa1V`+q=Yqewn=t>IpG&QY{2)YMP=3*%$eHG4yXCpUe-UPMlK-Eu
zK>^Bs0`oA4<*-B6*lpD`M7Q#&IAFKu?CYASeQ0-N6XIr0=%|&a&~wfQ`kFZhSy*Zt
z5<`(nk=V?k33W?btqn``bYRrB;nA%dOiD*KqU!b~)DfiwPu#2{6XthcvQdGS9&Vd4
z(Thk#k&9v=#`>TQC7s^5Y8Cd<e0&AnYp5C~JQs|h(z>Y}(?svNyPe3$N&YuE)nI#a
z8q;_&C4ejnPf>&PwMADmEhk7xmj;_2pkeLi?kNA-jCdx_NkPy$|7JcHOr*+Sxd=Wh
z)>r3<n4|u3eq2s}GAFyi`5+6}%#CNBkU;$KL7VqnxqSY4#6R1D{KSj37W?bM9C=wY
z>i;7A-xtkv<m&e3P`CR7X(*^U))zT-pt3KOKo%#lAOhmZ^FVglO5Sv`L|{a%b>zX~
zlh&xdT47l<g6-3#X^l>^13fan@mW)>#WLiFr2!SLTT+iHRMc8>LVV5X3Wv~dt^Um{
zg0!u@daoD=@Bw9XJQv~(<(kQtffcLEfh*<@TOJct?ZFpqWxV1Xqh^?6a{hvOwkA6+
zUaHcf9kI3ftY~pyk<iFI<e#Xr@4D!!T{`*o&lP#*k5g?!d?n%)E3^B|Dccy&qy>SX
zvGN|WP0wYtDK}p!K2n&0&51GPF=<|ATPh$#HT${Sj@Z_~w>pX(TCA#gu5C%IxC(XH
z_UhZcMQMe8$9DMA{<RdU+iN?kCc719&2e+mc2yvFLh9$EjZOp!dM8mpzK{bEKC#4l
zsum^W2u|Uk?T%-WzEq;G?ak4U7L+<quIRmi#*TVa9S5>O;i#2Y?NvMEsBLSu!_~am
zWaU0S3OK5mawOlt_b8ZJSBtyh_Br*Ct+&OFCZ$a2CQ?ZE|FQT^7C!G$8S&*=d;GZ~
zuoSh2=Z|A~>_l#iMtoA5tUu2hR~b=FEa<z7`{s1Mq{JP%`FeXN1}h1mb*#y*AbPVi
zIc@kO7#Qh`ZNM>%lZ`y*&nD<-C*6E|F20d%lN5{Kk_h+QsFS(y>WdcheTN8%;`eqw
zp<j+6lJw8nCn+)1w$naLkSRliw>Pn1`9*tY1Fmo%S8!tEF59oyE>zUws=SL;8~J<{
zwZDCmu(GYCBv7RIr~V9okW6>7%07#B&Y3I4$IJG)PoSZd#HC+kKkUz=B6%DUzD!E|
zg9^rf6)lRd87&mp|8?Nq?={@+irr0kzZ9)y5R=!9aQIF2OahIM@ommgRRsu1`p(gU
zGeM!E@ADIR1wubW!PpND7WW;?Y(FNDWcsPMp8_DIVv~S;9Le0d)}MW{XuFTwFF9zb
z>x1EHmn&`LQTU==6~AzjMex3QS7ADu>k$dB*;S6ByBFHEyRN*@rByA4ESQ?>{A(OU
zyQx)u{jSjLv?~({nw?#RQ|crQi7aL}?kX3~LP+MOU2)dVNAJwJM}>@C!NQFG?I^p&
zTaNziI2*Lf*(+`vcBF@M-Wty&nxhQXzjoXIMcw;{cwKM%{)>o+h=_=Yi0FDfUGwYq
zsJY$db^Fbq+dO6-b-Ozrx5xSEjEysM=ZC4g!t3rvH$uD-T`99^9LI4S$8j=$WfnF`
zL_|bHL_|bHL_|bHM8x;;e66+Kqvw0?<@@h<pVQ#y{eHb)uk~85^;)m>T0g6H6mbW4
zuOeXFme1bWRlK|uaesUDE$+r(RSbGUj@?wV#U{pNtJ{jsV%B=p?#N~}%wSUVa~QW7
zPnSzFL&sQCr%mw*+<E7nHr2n{-m(?GG%fK(LbL~M#&aR56>cNf9VOeS3<EYZVdW|(
zSO+;_XT^FcMswI^`>*@EyYL6IInmxFgHm?M=Ek$)>JF|HaW^lfsCK%UQ+_(Q)jiLJ
z(6fb<i&8Ex@F}%-+oI@5c#%t&M2ICmx3;2tU+T28=3iDWwq?<{a9&1`mScG=>fO_9
zg@avNau(W3{}9&lds}rmsi6vy5uZ>bTN7`J*g=;HTyn{3t$*LGycQ1xdL)jVvvp2@
zSounX&-%m#I95tI-E=o3B|g;L)g+&J+vp!r<5w19tBsEehb`ltw0fUPH~PNDcwNkG
zyjGN7Q+~qfqw->>5GOb1*qpVfxsc5cSW6NV=_;+YYCuc^_~puK?Nvvz%0twMF&(+d
zN^T=mGl3V0x8)+qiReeq)=S<<U3{EnmKx@iZHskN$g}N;LDc<e1}1-{J)DEAEjbA7
zQ740?3vFk<na{@M0<g=G-=TByu>==N6^Y0-;_(1Nd`Y1KqGZ2^KjBLS?`XpW`(%Dw
z6apf$bM{m`a~C+L-QE6-^0r-Ds8<2+pzNu`K$O^adpg&%9*t}9d4DF+P`E0)rtnH7
zf_OIGXf#E=sv41Gp>wu3zmXD?)l%HPb_=yEa)*?lF_`eaB$u|jI%4tm$CtX4A7>$?
zy7Vy1p2VCCMv!W6%&wm{)Llhmb;A1MYd>PNjK%E<i2CD=--vu)IRy_S<~Zgg6soA#
zVAc0260UNRu%<%^8HB<*vw@tXPTO#_2HL{QMv&`As%H{E-Q{{Tr^@Y|3(*PR1<h0q
zIq56VP~XsKr(+zSyOdEU*aJ!|ch;$B8}g0!u8^&`>Eq4geLU3A@SwesoDrOZe<e=m
z?>J|?5M$`7q}#2hF_8o}mNSQG=sGUHs*BZJilUB<aNOQ@;JJ6$J3cTTGJmjleTwKS
z6q9T3RdZ1s{UxQ=?ER!(d~dt#gZ#A+m9l*p;|a%7DM+{Tk*_{C?$?O8KWZOWZKO7E
z3&kf6$h|n|+owq_Ej15q=e+qW=f(9yci9p8_&lD4!(h2Ee&O>{ZtO3;JNW7!?5kWl
z?sCYhU;8U*W$c@nqn{AY1$2H6*|z~RQ7LaKGUdCRWoKJmvwdGdB>I&te#Cyr7Vv}Z
z_G7ltJ769_HS$vhGx67aHN-jnbIuiqAQWi!i$k|Xdc0r#-_4uB!Ee>x+t$>KUzpkR
zMd0vsARzt<|4q_KQXKHXmCuK&hV_J%U-i7p5;CXA)GP0qtDi4I2}y12ns_aBr`*fM
z2(NwqV)5%KiZGT^s%AgVuFJ7HxsxL7T>pF#tBCbixc?2$$3aR#uTuHw#^>W6u<(-j
z3%e<X)P7d&2I*&RPB2g&D0jEi0Jr2sO5n8JnoS6s7U~I&x7(f%O@W-N{Jgu}n_)~U
zHHo<&cG`sa6kl%cHgftrY7?I?Qr}q5z4zH2&ljoh-Un<_v`lyDS2)8b=cFG*X2X;+
z#V1Yavxe4go9g{E@;{rFecr##W+XCJG;~pb^Ff<gVYN7Y6lJ|LvsvEZn|on8du?_O
zCnF09B<4IHLSh|xa_3g?oh#I(3-Y|@L$1U=>0I`n7|J=Sk`jA}`$=1nP*Jw2O|s6S
zL^A$v5YLhvyq0h$WV)pe8u3nYAy}4;J5`<AmgjW;Fw<7#O5i)m2_=8nl@-=qNhll4
zs_cQ%k+{;Vi6#(%ymmPT58B#zCSPHZf~>F)IKeJzXt#9^5tj#CL2P}6)Z(^Y3PQRd
zupyADI=iE`(dQ*DQ+%OoeGqtYg90r|?u+__t)&o6W59u19S4&B9<Zir#<Og5yf0B`
zI&NTiEeSiDt;CGtH_;l;oH6njhTZP{bjh2)!vP)ZSuJ0rF8@W19WBS!1gVz&)MVRy
z7E&`Q=*D(DAJ#VzN5X+}mLEcFm&242y*-?iLH+qwqT@R3(E#QU(KlQ$OBLFk&;Rdg
z+JSmbS1;PKD;Ky`mRE%X*kegd*aH+bVUK6;2+GLr*t>@9i6q|z@UAC)-gi!(^gDa%
zaw7Fe#4*xI*>2yC+qc`E{P<6>s;B+qRr9Ii`%I36epUul<@<iNq6KyZyIk=X<vg~%
zxdg~LI#~&$w&$FrsFPv)avr#yP@AK{_E+s?P%a5<tS9K{knUC1n}FfeYJD-VJJ}YA
zmEnic*Z1ce*sdHp29ZWXfXP)SLZ{0H6S>W;g~=mh4&?}onEG%6MBL+9`cWpzkqQNh
zo|bdit2UbL5PnVu8$ID@BXSQ9o1`bn(U_K=56@7pqu6(G>0wpRCN{QU`PZ&LSK(ER
zD>bmlcy_)5LE33o(Z7)(Ks%?wzp5l%s3zOp(r9Bovjr_}xW8P?E)Y=iZu54aTDFXS
zdnXSO8UE0ix?1+#m{mItpzme(guxTGH9kkh*nbevE1${%_m%X&_F)XII7!K(uj8-(
zQT2@EIxONjLTf+H8KDDkiFnjL$<_G_k!PP)A1Ui3B!`mVGbgV{)>F?k@uzA7EI*l&
zLbm)OdoCGdTps!|_iPg;#bTQO%BQ!$FOFYVgSihqjt}-tj3CLc#7z|&pYEUECNdgZ
zq;~x-LDAhFIvP*k{e4ad8<G7GW5~r)Tw<6z<B!=>9p^@!{p5Re>2muy=|$_KyllV3
z*8_MY%avddY3JAMUS(T;i}jTX=bYI-mo~&f<%UF&h1erk>?>kQGl%Du`&^3g+kx-M
ztKymGoGdm^yE;0P;#ATO#fl3LIBM5;Xa8{LRJ(RxsB9=z)`t1-x)_IzcKtXotdh7#
zB+0iM_7&lVswt`>!#=rjU#gTS5okC0FBH16uid;aHWo6N96riA-V*Q<a$DSx&Y~{9
zbzeE$OT>`fmT(aQth4!c@AJ>?q|LwsXI}^`By*h@U<!k;vLz+jIbXTMTWvyDW|Q_^
zEa}@_dmuSZZl6^>q|4TSvVXUw0fo(!=w07Q#ij(d&Pb=moCg^96`Qs%q%ZOn*3@$m
z?6Mh-%EsnxHq-w({jR%gR?J+*(ND3{PC~P@_l20crDhXG(m7QVpOM1!=T`V=C~ijc
za={@;$*SiRN)@n*-2-d+>ld#D@g{{JPvnVi#TH#YKMWz6>2Ars2pF!391}7EkxqE2
zkF0PX1PIJH$YskM7bK*?;%&<VHmqMIUVadB!HR@eL0{SkaOjneXYv*|*s6G6^h$7W
zsj1dDnQ_&p)|O<~wTUM@K{zQM2-rj2<sx)Z>a25kwvfaV;cI;bk4V@xu^YUF<Q=r)
z8~<GkmF=!|=<e6Wp}vAn2dQ*3uUTU};}As1(AGX_P5X-Q{I^NAIcP}K+7d1M`A&zi
zaul@YdM@y~u<f~a#j`>9QM8BRojUw$S;eUx<>9zpiEnAPY-^&Wt~l9lOTf_Qa7Mw{
zUbP**w^|x7KH9_4Ae1rq=%2AiD>@kAV7t^QVrR4#)yo-F=D1xsVoS+^{aAvCk)c^C
z8Cb&O(V;%m-Oumo$05@vE*V}@iDcm?bNYn<<=?3Al@naRp7NpE<aULax7+c$qoJeC
z_T--9xbE&KIc1)%fPJ0wrHG^UjL*Gmdpk$qvk58|KAfHrZ@1Xq_=r8Ov^RM54cT+P
zl8tq?uj<mhd{ppnzvCZ1Gm1dn6Tn~UGw+Ir-uRZpQ1TE$opZ+eD#W0!V3NIW$+Rwq
z(*9_j@d_&FIN+RvGT1g4gHlEgP0>(dOuklPdnjeyJ7str;35ycju<%--%8gsWTQ!Z
z;n02}2S(%+DYII3vU=v`7iIf8oy*?d-GYdF))A}ACjK_QF!=7SvGcick+nM-NPY4~
z0Gs?rMl$V~!*(G*yKj+=W%mf^9h>Z85;?krW_vqlfSKx;y|XXwd39tRCHVs$$i85I
zT50cjm$DSw`!Neq!mH4lEOH-Q28{d&&Vo6GNPn1k;R-F?!$<kM)ib(%>=RzltpK=B
z5<KL`P4;QhV$xNeeHIYQ31HMd_ix8~Rv{&Qu`d<>2c&RAR1r@c-y&c76bOiFu&-hY
za``JFCsgrucDA`~yM2=@C4~i)puDo*#xv#fJHxZ%cy@f39dUh=5<{5)zE7~X7oV<r
zVEZ9plg&$BsYjsCaHm7)BWYnGWi1AL6s|uxg|6mK_H#nQr|9c|=Y;)|Xj;>Napl**
z33?8Epni*Az0{50W%hy#KVs+P$ANEUYT_$iDB=a0;s)+*SH9r#PtnwZJOvzG^+I91
zwe5EG3vrfiBGR|ZuBjTIWt$WpX4mF3$2ce&u<PO(X&$vxd2`0De<99v=e!a_Zg5DT
z7w(L9;|sowF2|OPtGpQxmbvMLBBrb;9?3B~fuPEKIKlZ}yCuN^y~+zDqvowSG(rgF
zCpGkKFNCBfhg^l9zfdH?`D>Qigq-t298#!0;cseciK==b)KHBC@#zXMDN)&sIA)V$
z0>unguyYQcCZSwj4wQTM37{FBX;WhYzK2166{s^UhsU3r(1jT<gpy{pB%hd^h=FrO
zcQ%`q&&J5$1Fm(}&yHuFF~~jI=J-6i@O+<}fXuqv<^>={<H+aZg=FrjFo{npm{eZU
z1=TZ&Hf~d0blFr&hJD`XcD7`^wY1)>#i%W<o)xEr!m5|~l+%NMc{CwSZ?U`MjAJVt
z`sp)&Z!4?UBgL+J7JbC3c!m{H*I2X!1Xsl~pmd2F#H{t1V^M9j2eKciN5Ml%q_r-Q
zzgRJqwmw$$Qgh3pF=ab;8|H>woar;j!kpjMw<)PktwX#L)p322$?`TtzQ!cIcCM+o
z$2J8VEFH{&`vn5(=G+;$(H*ju!0@FaHd=fH&er(WU06t_#97u}JtK`TSXaWPjvSLJ
zU!7<gyP}O7<h5F-Z0lv~=qg$W6G3cSApfpE*p2{<a`)UtAATVYj>ZmR7_h|HqtzCH
zpJ}7*%qBJUn~7B9Y}plKqsd{uxLvS*k2!qH5lnSl+2euuxG1omx<~7p;XC>XU&k8k
z9rom9a7&1gW$bkoH4V7`@O#@Gjfx~u$WyWjVam3r0!N~iOBzr6x=I^{zfLsJbwxRQ
zspV(S=E5&!-|wveAcr?1qSxb7Le%FH%^aAD4_CC)z9bWB+dvKbtKN^7d>#^pdvbr0
z+q>82FatY&U$EtF>#tx|sAQ4YZQz9}CDE2G1Yo2e^I)`<*-vGQ*$y^TeXkIDOk7Cl
z!xbQYr;sc*l8m!tY3#yX+(rZQ()pc;Pq^$CyO9pGlTIL8vG<^BIvp((+@Vl81;d>6
z9#q%S&Q*gt-%(uB`Ae26bq;<7eIwo!Pj@I=7Dj(_A^RzwnNSUTW6Z}P;ZD6>bfi<0
zs`$yh?aS9xQgy%MlcWH0RW|6|+<pjer|i8X;Sy-w%(ULGh8rDI$v>r?`XDJ^aa_&q
zBqoeb<o5cCedN%n?xlSkZ<Jbj#6HQr$r;kwP-ma|GOwRtpXD%O>X1+I9`?5@+w6-9
z8*&Gi%yky(l-2IbT(%|LC()nQwb@rmu#+cGvaj<!&O}9P-D%&%P{k#Lt#%kn{k9q&
z_xTWozDryXa)VzndgbqZVkEC=upj&jl;0FMr|KODvmdMFl(C?AmoXJN%lwo(VKrG8
zy6oo|PCHB?o1FaiOMF~nAVfdR)SZHVbsW~z6|3}b6{DskZuwT(jK|aU*#08NWLo_z
z_J;`6`4G;0SMGNasNC*@psw;3><y(szS=+Mmj3475wov}m)LTk=WDCak+On;T(|#X
z7`l{$NU>P58<Q>9uHPRlv!<3TI*1W=L-d9V2G{r-qc_elU1*^3Gj>z-=E6d5WaUM_
zIRPWbJzd?hzeq7NW=*qO_ZPW=6mS^FZ;K9`o%&Jl8|QZaVm|nv5OC;}8McZ3qNaPB
z-LXH!9@5^I7|BTiqaTH`boSfic&4nq9*=Xvro=P1^y<PkWK$h3BzH`c)3Rsf8YNQG
zW+dK9xSPCBG1SQ9VKeuqkk2j<wHHbl&oD8|`&FUE?KXRV5f!o3&a^qzhw|v~xqCp2
zyO+%MX;T21;^j84LO*8h5W4S)XYvq)gW?!)%og}pt1;aZpRp+LPHfbY{UPQqYS_|7
z<S&Yrr8(seN^va90WjBS*D9c7dC~;ATP2`2*ovx^SCf?#yE2ib9KOnjxGF#ECbt-M
z2e5j16Y96%6D-fjwb=x%9h?A>^?|Bqon}R?W$W@8-VVA-o@PDQ$1_>ExZiM|5<?yo
zBHv)*tP-B)*3IS8SnI?nj7TM1@2%TV-J-T>bQCpj?zSfX=jwl6ZkrP|%ZZ$536i4%
zRBMbLw9d|X+1e{8dr?igR=uN|Uk9vCUVL3aLE<7MNQP6@8T3{sibtkSoM_u}N1?7(
zN`>*t?>nN^Bab{{4?959XU?=oD~g(CJNK9C(O*{CuAnUjE3&spQ`;emQ`uvQBJ`}d
zGIZJFF^9<S#@+0oJ>gxC^{kk0PgZk88iev+XZJr9&om>+sU*9$-SJFrrs9>SkX`pA
zFTiCNE!)#o6Gd<M9pIT9>JN9>v$@rj7#!xLsBBkx?5$Af1$R}v*tx}X{yl&lw0#w*
za(6DWy6ul=S~lImrE}?VuojY2Nuf-=x#IVso9)X!Ai4AO(Fd9LSI<1ry_m~DK7%Qq
z2E&T&9L#4$yQh%mQ5%Y9kyTh*%!Zu|mp8O-!mK;uztI0FZN_M{mm#~T^I<e6;+cC$
zy8_^3^-Qkxemb5GEcvOS#pw-peYWcLcro$~ab%oJLS;9UT9|g;A;Et{?jLXD;&pa)
zW4yZHFALiHZ0vurRIcf}=nHZ;|FgIK-}j?7w08o8lD3{+Uf~dXH-{kT4%YBq41p1b
z3&Q*UG09k3?Sr6d&Ak`ODK0XYn3O9L<A;4%fm6Z)MZ`fLA60!~;IWS@JiG!6@l4L2
z#50-3NY2L%_Mm+l&x)B3Dmcrt&*E9p#nZcd(m#)J{TTGKOj_}m?Tdtn{-mH8W;d3@
z$-Ydy*KzeKH~+8l2^x1*^sBe8D`daSW#^j;>QX!pZweGZi8^I|CA2C&P?EQE_I-9J
z555vG{zD+pmDy6i%QwuurL^OJ+#f*(_z;t1oj<Kz=pcRy7KxJmXCIK_pCtj<FNqBW
zv$7&G!T%bq$wl&81#$`0DqfTJVhS}fUzGFGu6Qxd&>|&E9y!k&rJ57ddd04MF~p+c
zmwwfY<q##OV2fS-;w3SX9Ss=$J7vAN=EV?mid9&yS?9#GxWHZ;Z?sp$$cWrzC~U`e
zop+1*s?Dx{F%C=jIa0ve4G9Cj0OTKhF*TE=(k$oKO*tf*S|>>9!=?1N*}r7r4{$o&
z@?z*sR^bhDYj(Q&A%?ikf5{1o98$MOht6JoH|tm&v<cN$b=dMJI^4vqZ?`+*n-s_t
z>n0W3t2XJyqE~uz50>*iaFfaZ3zQcjz2Icnro=L(+wf`B_fvD+dt0%~OpCGPiXbDg
z{6Cx<&v-GEb2z-SMl=00(0mj&IxAl;Y{KYgvtKNd;Um9WZ*%-jaiJ#P|6E5DY2?rg
z%<~^ylG^I*o`eIsCYfM`C%2}K7oS|@eQ!bCxg@EM05%-}1NiSQeR14mNlXn949G;w
zvLSkF8Y1W&wdK_WPT^X<A}1g>nk}|6K9HtE!3sF`*s2%HLEO}0YXUegU#rN_owT7^
zrLB$cMD0||BsUG==dmchCw&nmot_Lvvo3){`vN<`LeOXH6P(rUiq7BQ>x>xNAQvXv
z7_eiSX3$!1++2$N5{Xq&F;gjKzs5vuXC0cTrW~)_;ar~HoGmEfP;V_UKM`3(5QAMA
z$y&W#TVtc-i}w5q54;xZsOC=bM%qim)m1%{4`$Ih+1A`EolPCKEgA^XuNI?EukFam
z-wQ1@*~1BMtTHzjd(;u)6^>yIJ9E2m!P-@gsXWMjkA2MFkM$s;Kkn<eb-O*`@GD>p
z@!-h>h1&qeXuNBV*i-QguM4z#<*)62F)k=X#gbvO-u6TzF$Ur&naO#3+Nl&Hl|AEQ
z--oM;JsVA3Z&|3o_9j~*Mo4wqQDZ&lU-Q8B#TU}lk#<M=f7|apmI9GI)%(E`fgqpE
z-f9%-8l`o^bd^*AQ7AxpllAAgZK{7$ie3!FhfbbuO&d&Nz!NBPnSxaIqYdRS>?gTx
z$+$CIfpZ>vsrd1*ja1Jh&-evvwA$awVxjGlv6d&Ib&2NX^kHk;$t0PsCK5E~T2iv(
zY?8`?`|n@9#?CoeM0PqVjV6S}D^y57Xm7;Go(9kLX9tl5iNLB_E5D*R^ido0!Af0<
zi#gaPvYL`o&|ej{`A&Q)W_y+?p5*TNZekp*hX?t;H{J@i>Zj!UPEBoFYdTc~xWhhh
zqAcnRg0FIm)wglU`6xOF^NCc1_HhNZdpR5eU!Nq3FsXf-VBi_)=>R`V0>;fAF{RT!
zcXFG4*WLC-g&mG)KzzL@vt@ewGS^@^C2qd*E=njAidSN;63!(u=zWu<rF?sSH~iMe
zxPRe4*>?eP$)XCL<>vjpLyV<^RDo4r^lv{_^HX{vcP{!Vo)yxD@^nZ3Iah~kFu9_C
z@lHzCiB9{qIw(gtC`kuck)O2n<+{SEokv=-o;208)k{LSqUWMAb6{QTv@8AJIB|5@
zRXriSENp9Sw5xku{*eGFR}(J(c1_R!TmJFu8renI7T3l%VYDfMGS^l8a0dq>2{+fr
zGvyP1JvPJ*J>`IHs@ZBc<|l9;u{)*4xvA<}J5Cm{VY|5}gc~y2qb_A?w<LhYZ$WDL
zaCk3=`eC~@dXXaRl=dxB_HB-W1}T$ocObgD+HHbkgVglYL$`@N!6B7&uoDgD9sXL^
zXe5qFj$#(#HU7gfxhJ#V<h}NMLE+_bFH?Goh~*4&YLAP8+N83`$sCVuor~JE7)lXW
z$8CmBVXWt&@oG#neG1KWBqpAfQ>3-dUY&8XeOl9I++}lm+%#ElbE|GuO><uL>of~}
zPlauH3MdZ%xpQ*hnpzN0DiqtIo{*qgnrpc=E%9+4`M349)IlO&z;s&{y@dqEdY-W5
zu~JM|?$WG82|LiF<)>pSa&0Lp)zY?|&R6Edlv+h!0Mn{Vdd@rc8lU0nzu4O9t;X<O
z`9L+U-_4y@``ukB{nq){QzzQ`o{+60850uX2A|)WwszYX?f%hf9S22Y#2Hjy4f;B?
z?Sf`wcBM*}Wc5MA)^rI$!LS-^vlGKu&-59;x0YB9jE3^g@mGK8AcecNCt~OR(K@_Y
zw|Ci<`@CG$I&EwAxT0;VZSx<~OE|!eTx>Lz6p09krA?QwsQg?W&Av!5E*sm<T#kyV
z?~1S7M5^2Lh&|R5l4?uCrsmG5ru(=fjbJu=q9;r&Qt@|?V>`5LPgZ@U3tXQ)l@P3{
z*{;+MyB%h!#yK|j_;gUwvFo39l&rq@{uTC2)fLy6K>xF`wd4hizZASf=d(_Lz0q3k
z)N&`r*k{k>z@_+4r|ru&7e<hU++@4IdR8*k!)DRbB)aR+QubCG159*TU$l3%Rmq5B
z{n-RZcYHGk0!Xjz10a&Xi&w;8Hf(I5e3T7EU${`}F3X%w+HgEWn3A_XV(=?A5*>==
zD%p&TqwX&?8m;B@r_^HyP=uez1?Gm)y=fCZa3`yAq%(4L;pqyKazB?^Ucs%x<k`g9
zcFG~wQ$VX~iGr5%Or|Dpc|J!fUd-HP?2T$9S@ESqJ8Ksln;h^QpJNqUm-mm~yTu{p
zLql;6yXXYa-lhr@6)IpPuy|@S*gJ_^)GxAYNcZhK^WDTPYJMo`z4%y{dMRx&hTHqq
z$5kHm`1pgIbB8KlAlX!4j%QEr$0_{`LsBlqK8l`%ht~l8)xzk{a8v#`-guszlJ+ph
z_$1y)N>I6iagWGPb3;&(c#D0OkYg!T$iLifKKB+J0_PCkzNqHm9u*YiWd#*4{$-+!
zZ&mO?HoLFlnGDz{fJ<`xucJSa=o$MaN2GV1(g#qHeVfc_RgHaDL5?l}$h91<(fmFk
zXKg9Al}muFm?QYd1crM7Y`(<m{^ZPd#Y+1*JLQ1SoE$DJ{gR8Zf)u)JlV6?fOrJ5`
ze#?)FmPH3_r1Ua-$%PzDHYLrU3I<;BQaP2;(7<)Bd@16jwzSoIIS*L{u8Pmf{aj=X
znOq&uyn9_I4%#&@g+$fTOlFX4tFKB<F%)>`?YfsP7CV>dROt!*AX0SV^)H2RQ>p{$
zc5}l^m4A<1uQ*1denpVVjnThiN+G$jWH%)QT{Z3GA*^6jCJB+@3A^Q`kZed8TTC9e
zz7%IVVjyDhZSjq60P@q4rs($ZehS-^f_B17MXID8^je!3vvSQq!H>wDv*wQRFS#Tg
zf~e2iq?i|%nZly5l$2YcOhvL=Ope}M$*fy@Im}@?Q*w4Vi`U!KXjha^Kw4#PvuV{E
zPt<^V1cEHZxw3~w@s61<g``}gv~$Q(`MZ-49Za#a6E!P~Wp++XMkk86u2PlE%|+7G
zn6$*WOtk+hXqxBLMgB0(@_TZPOZrV29Twz>wGB-*L}V;V$lb4~_;I-Cvm_y><Yuca
z{db^n(Alz=im11>!<PG7calaiAXc{`=klixg}APKsfgeAE?;h|yk}UxGX1Yf2%u07
z3nhDswOpGJV1C9MsDeP%gFL~~x<H1ng^T0TDk)>5Xl8x%9llfCArzX$#o0Dw-)K_r
zvW?Ns>8a>iQLHGlPHl_|gDo!fui&{?AI}s6e;{rgjfvNyvL-uW%#lKG2w0o*EgtWb
zAGa3A4bC(ro}e|i2!_KFcheptl=!OxF*Qtc{)N$LZymKR-xJjPMH<T)Vb|}pZ7&sl
z@a2gke`h-qkiRsy$P@hG023D^(Dmp`A(4h_0IRsuN1y~+gYEK<yK8YHZsOfzzNCLf
zxO}|YV)C_jHX}x*i-z*wJ&_Ya(}NNZhIYuFOjx#Jeq;hq<phiT<j3c3Z^NBOm(X*z
zr(y!__H+_h4YjlD?U|fEJ`>&cY@!t52Ngks?M*V{3?hO6v!dh+olegs99x?4m*}v4
z6&(0J3w~S(_UAL*3V<KKH|ufoU$EGE9V+f|NMeqZzy58i^;aLgR-^(cxd!qX+(U=j
zAue-+{z)rJf}xy~JKYX@$CMrFw2|t|VwdJ{f7wRknIfV+#n5YZBD=scq0GwFno@ba
znF2SIXLs5l(13J)ZQGhT0Ceb_bujM4`g*P!OF0Am{5c=bBIc%ex4jX|vl~6yZaBh)
zga^fz9DU_HKbGr4#e*ihn7z2z0u_(i+wsg*T5j}zCplM59clKIOYhxi74G8fy_5D{
zknaF4f4l5`$5vxEK>{BHL5e{NGb*0Kxm6ro@ISK;olUMJ!eEPi6kQh?lnt%m#g8jM
zrKu~fZT3lmfQh%HOQNGslOpeLvCrb)s$}!I|C=lwo%Ti5zkA#2{g+i=#~_Zv;q9vc
zS`xLpZ!5!v<eFmju&<MbFLAzb8vZ62O35PFY2POB5V&%)D(lO4)iWuD++g>8?vhmv
zo!D>4&76=jf@E^WwjYzAn#l-E_*bsyl2!NLXFtaZ9MGw0zf^qmvKTBj(67~W2TB|e
z6SUuQ<K0I{F%p&ay0FB-CPxsiLnrKtcvgxP0+8;GSN0aEhl_>0$gb**GnG1x%DQ%S
z)eF)`nL7ywdxNoQi*|6^ueWP^UD%N{DzO6>n8L~>(o43zuIffytcV77eQ!C#+8(O0
z8{##xzO-TB`H0=v>#~+gW|Enx7~7nU-4qZjlXsupoZrezLtSG`xurLxq)Acu{nngT
zT^rV&F1sy%t0XGOy1h59!epaCbnq=aA$w7ta*1>kvlU@x;8?XMHrX9MjR(0mPD;dV
z!TRdStS09`Tc8Mv{!Qr(IT3$=re>9WpPHj^0p|c~uxT+NH)iW-I%hMyo67TLW*|rR
zTTk*qb>9-bG|R!cYsNI2U15+*C$C?_iQWBe=j50Y;uRS)xA&ryjIvkRyg;v9ReSB8
z=tL!ga7yK7nD>wc0jKU9FR{+U+?f_tu#=rXo-L``ajOtjaDpxM@!tGzXs?#VduN4y
zt6%Q3_~V5CXe%mkUPVS<nUG^bZIcdgRfP#9Mn>3>z3{DFV~9dNMqzhry}!DyZT3J8
z$MjU-m86)qE<w~07KhaO7<T}^J8B#9D~@*L_@emU=q(8D<;b+!T>K~`&90~q*vbh<
zx3N};<c;IgsP9&tiKbjD`lqM^eD-)O=;q$I(-3(|Y%%GzC3&_~5EBR!En9P`I0xnK
ztGp&o6z$%c5IZU)Q+!h9_YQ}CtY;xgdF#qi@EP3F&H+X^b-QhiQ8HVzF4)_0<l>yd
zk`nB&qiQ$8n69XOINQMtNR!5ujX*J9lChpe_u0-w0Y;`$24GhOzODlJU~&vP_>c9*
z9cNJu>Rfv~za#>QT349SsxR@t5rx6<obNoDU#{-rPGC>vZfqi+3pTJjVdM^W4&EVy
z{ho@I$uG0qXu+kjp^C&YtW|w@c+xXIC(izQdp6&7V#b63_QtzztccGgkVO$xlogT%
zIP~`QmMd&?s|t(m?=9DW?VGG8fy1xoD8OM#v)%yCV`@ruV%~T%aH)>_d_aPpsA?MD
zm&1vPD7n-?JS&!{G9l_w#IusBkR7D>_aO)AzMAc|4K|#VvIsq(jpQ2MeV2_ou=D@P
zPE?(|QkLKpz@Cg9#HNHNpUxKLCakEUK|32e$OCiFRTxrT6}MFQfqXvRNc=s^p2b}k
z(vWNS8;Nl-dg9J6_(tEjaifjp+7oG9({302%_@9&-*!++;t}EvNma1P-U(pw^hZ1R
zt~b8(&MEd@?&~ge+g`5t{c2)h-ag22<P%(PACALXv1S+BKl0Dols&f7J`Q{oCp4C4
z1lz$U-U4?m?Y>W=MbRrlgHoY?Ry||6wWx^Pp9fKh@Jlor`m(rHO95qH1VrJNddZFa
z%UnxtZC^#to|ar3ZqUAtXOj3{5w)`azlmqsV7*8L@)r2k>EPey*>_Gjc$zcC@1v<-
z0cFkYwI2d^GC>`;AIAqMBCe_o{8Y7|0-^ov9gg*MQ&r%XXs*lqK?wg8Qv0v*tfbsw
z-V(BYOAhf*EG=my%?`LQ;f3xT^di986$c7$X~y?NV*8cR&~0LxvBVi|ovRL{JSFpl
zHr0q-9nUavNVB5*zFp%Ta(i!ULk7GyI+WpDb`UCPf=DS>u8WVAf)Gnn@eR9vd=SY=
z66@`TY=<Bx*;!?hZj5JgURJprB?Z6fKw%>?N!(>O9|$>aIW7V)gIf-i1HPt<%iFE~
z?^P@2ciL^<t{IC^M>Cr6+Yc0>Y{vBIHX(5ad@^p0B3_|n;JQCC=Y*@5BATfc75wRr
z11?O;i|(*ZsvsA8^$k1uK!|4DWz&=GFdLPNs^_MfdLS&9agxcVDbHNkIslxtX*s*v
z?h->hBl`%A3uP*8rlX^_1)q7F<#17Kf?UEjJ2ADqn}GTJ77r_QTt$eQ>ll(Zgv~op
z81f%~XZPgF;uC`&d_jdD`6$Ua`Vv4Yt?{DhQa&Zp<&#g<mgF!vqak(_1I*HR7I~Z@
zw%2W0PMy_G4RYCd;9jJ?Jg1IsgTfiMBKpu4mFL-csAFY923-_C&$s>_6I*p4<lG{>
zHn@x68t;M|Y&~fvyYQVRTHY~Gt3rbhM0f5icB!p;-GPv1J;5~GZ+)~acyt05QXuyY
zx$qt<Lqca?lP#WYtJcxHc#GBh&;$dow#Jy9d$`I1GGI;l{;yMPbFL&3xwXXDLWMks
zUGu0YgjQcU*W2MX6gXDHUNEC1bFS|4n@NX{&_JDc>k0^Tr%_!rNx55-5EhrrklQMx
zBFmMl14!GE-yzB=4Y0m@*iq2hna*gB=3?R(iP_Y4UNQrjVJfAh?TR;YypTVb{)TW=
z5zHxgqFY<U8H=|d*89g3MN8VY*%JYkqew^3s681STymjjXgINsV1(nRF8M~PT%=gr
z?HIaK&gFZ2033Z0U!KmzS-jkyNi>m+hdb7@-p-2x?~OqvbRetBA&wQ3fy&ic&mD+6
zW+_$GDb~EtSyS`oR@<L!hAnZ%dXfs`PJq{Sfb8Kt=yOnZ^*Mp9Y_|UFo3O}D-G$r-
z64d!6LB?RVXdqO*#G%`ePmGHl#ox1atpbj0B$*2tottbl=^Qs#9JE^P#3hT`3;q<m
z>l%oY3Fce>l^xnI%Z#0lXQ=K=7^(`Sob~bUEXlJ*?Oa6*QlGi};77{&_?X)=%C>Tw
zwl@;GE-}Ij<1LU9I6lxOjQP~i47J$Bc<%(Fzenxu#Kt0JY-UOg_Kr6z#d}&i6<^wA
z@8&byts#XXRlVm(U(janN5{G?Vp-zv8T%k`s@w`6=KEsR(z$lVK8j}o!!I<i+sE<D
z^I`Pk;CReFsW7YzdqTwlrbfiOed^;+{i}WE)U$H&D*HTF2p2-m#(Mk0x8eNyu&I0*
z2vLIgubj9_)kd<_aNGFW-`2_m@l6HIs4Te`Ug6&Pt&g;x62#wCBX!q;kS_avd`6@|
zlE84>en|2~on5m-*!?kDNn%2+slWD9vI;o2a?<?lGuz%~zvTCwnGX5az=t#8a-_1k
zB*grdJ<CBz&k)S5uZRjoEBrr@(5~n!f{v<42|Xo&_m%#P&53Rs!jgUTs(7XlsN>B2
zRbt$(?u(mReFMr-yQa^@frDeOAUR%}9YY{cN4Lc8>-vgZLKqV%p8G<8`Qsn|Xg9>C
zM5M98H#&@7EDE*MO?{~xQowvMo-rFGCAc}DQW_9t3X@CDEqz6(sYf|^YhQ>TYf+Mm
ztK617*LUk`Y`4ev<^ITgl~i#;0`^aMcTUU!;8qfTR0;Twcy@@u4k@(t(WGow@{#yW
zZ?ey}xIs+ma~a2X8rQqR0xCnjO^x@LNb(?F>Bvau&lPQ2qPn4_9_~8B=Qv~bG@F@k
z>TtAduvz24p(I!E&ncT7&(Oq5ajvVp&B@Oi@%Y8z$YEVgsFlqNbmPk^vbk`ooIZ%7
zc29KVekC$1pR)yh!Mnu>l;p0{7Wrzb(j_=wlK7l75!WLJeEzC`v1Jay{l#0U)0W4m
zVg>_<KZPNdMP5-&Nwq*ZrjWZ=Mr$d5WE4>pf}fD9l9V7@)Iz@CD#au<hpo+NQ(=>|
z+7DC{(nYp#9$OdBI9%V_1!qx!-g=)yOO2eJHdG(U*~zuR8xsq-8%Rt5rL{TKgAEO^
zmin9pX)<=Qw#08*W0GTQgM5#yDcb04lrs#rV%zL1Lt=#))>46{8=C8;TOG7DrQS~a
zB@N_rMKr&4RL{=QY4P*2u3RBZc(ZM5q8?I{C&+0~TCSJ2ExtYUmO{z=O1Q(9kW>@)
zaDE7q`n57#4SW?IjW<Q;;29a?<IT>T^ZeHO4g%wLCFYbrvZ18rc`UbWLn~QJK;`2;
z<P_wOC!%BBt>hmpb+;!zeKOuSAJi69WfUcHe=26-UO7c$WV;hUBpjtU7Q%}u$oAyA
zLA~sTr>l>>{)dW6Jd@ALXB_XaqG$725#`H1x4lXF${qZi_q(K<6roVjK7WT#Iq&xS
zJEf?Ug0?4s6R0v~6g4`TrC+k%;3@53r7uP-=Q`Mb4L04MBTk<--3I)J!g@)Tm=i7@
zHh#%);vPzLlP<HQ2OUlf5Vs1G9Lbq!zZb$dY@<n)Bn&>0D~fK9qy=`;S8MvL-`i=Y
zGdztIFn>0ITL4=)=Yz;`L)VV;Ns$z#>$EpwNY@e<vC%H{g=!Qb5g$jCGumZiF)i0v
z7qqjQ=oIaugE4K!40}6itdtfkSO0ga?@mFw`XcivzMBk$h3vQY0ut}p5%#|KOuVpt
zP_4lsRN%VSAGQy3nv1u!+egtr!5Oku$<*_4U);-NSJ+hU5LDVq1hlGB(5HRwr0Uu)
z=eJ4NifnSO^I0_slDe@nic9gc=kpl$Fm6()@Y%kj_C-AN@Ev!C{nBB@O$k-OR|x{@
zwO8%yTvnnRRWjfke@n7!l`;7?-y%mMDB5>UL<Am6#m4%6AEPObtPg&!5}ZE-ygm~h
zMBohF(}4Y0wU(Z~9LM%kJQMWNbm1m%KljBQl+&h>(*-S2U%9b2)!VO$_l0C9{4G94
zK153No4OqgHOalL2=-le#X(o{z-U2|Cz!tS;6>R+#(J<#Uv+TY^hLB9$rkKte~sg1
zja_rl<sGfXX}k7d5j*CS;pw`AA$H(ONy@2OyZ&H6ExmEE0NilUMH(3|l}TU8;+%8d
z7(nTQAnAhjw3`lIyd9w4+{J<PfB4UCc8H4LwAF4om}WXT7U<kO%pIMeMDTkI?dh#P
z5}JUPO<2OZnlO3a=CdGx)NW5qOr2^I5_6cnmA-#s0*PW*=0^FP-BC?Rx+57ik{Bn&
zo69rJsHDPVAC9y~HYK_&^edi+#&~LcSaQ$$c_(=UN6=}}8h1EF`oQeZ*o=cAQ7$Hs
zW#++<O{t*<51SS3%f4~x!e&=ZaIdPdIk}uTxty}OF|_Y{{gni-g#Yt$?l4k4)1B>}
zq!gtK-d@{Qzugx2)HZR=TI64#=tTrUCnbZ-lEfC%N%gl=>C&8<r~5(aJ8H`g79wA=
z$d*@c4~k>S71dV6v)ooF&}?O{-b$Q?lKd1Via%?Et%=c-rO4A-(G7=fZ6XCHU{2=;
zd={Mz+iYF-LM&Y~^X#<s*>+8v(t&LV1fny7{cg-n&`x?asuWsnHDl?z+&#xhzTO8x
z7O}<z`(C7>n3fy}jzMH9@wC~8!~}=xs>NE8;Fl_eTWjJ59Z&H-Ve7Q#GbzVIg6T-O
z@8uHKU|l)ppC7cX-eE&0RR-|{st_yRR7s|XY)7IFMW|er&v1Nr4&R4k8o!@qkH&|h
z8<q9Yt!8H;WMxA~Qw<W~uINc3S=KRyC!&OTEVrAo3&WTn_wJ}^Mxn&6CTejs*b^~?
zh~^mRgb?k?cqY!OWIZIsv8QrZtVX@LyZXq}_YEN;?@1ggADum&Lu1vJR{xm*(p?NC
zkfc}fv(cCD5N*p*+nar%dER2r<<ML4eb|?;IvePAe?nRU^!<G5NqnG=(zO%E?>I1f
zgOiC{N-c+XCU2CGK&k}{Vcs|EaE@R7zFGeGHyiL#l;ppal`fat!33&~{G&th;omO|
zfYRZ_j@&28a+5DQfRd>p{0M_lj~)zpM7tWxI>#PH#}m2AgsqnBJSPvvVZShh79EbK
z9SAbqtE5-b->N1%7a)l5OARj-!}$PGq0yuEMlMh(7tm!F4qkkNvtg{~|KI;-W7W(L
z<HhJPZ5NZRDyFTwM2o+jWF*78y^||e+*AE-^KLwI>kCIK5o+({Gdxz%UX=a#emslv
zkSND?*$43qLMlh*A<+L}JUhhgSw2OtKw=*y5iLdn`Pf$kn+k>6q3uuN1NWIl@<DF@
zG?^zlHC)M%BtFYkUr=M8#~0d4L|r1u_uCi#ekqYe_N7m{)D)*|YNDX7t)|Pq&YorM
zFBR0jNeWT<P^>)PI$u(B_;<<q?yK31K>mH=SFV#C9WuK8kS&_plmzg{fJ{L^3c|nx
z$$rZIJKMQ*cG}OeMdY<}1mqsAd=!3(H=eglYH?+HlH%sq>=&fU@F5${ZwDg|zewq%
znQ}qW`dvyZ_cj=vA|kHnzc@ie|0MtSe<XanvcJf4k8Jqc@9e7RsTfy5GnLZq>i!VT
zx?5YCA5^x)Yob+u=q3*k0()&fi<FJ8AwFH#pCX(W2X+I5O#IjPr?RJn`-$2l&>$`F
z7qRAs><yloUJo}WY_K(jL0~7eo3dTGvK5ACH}@BzQz9Ps##{P}T=4J)yR{lq<`9>9
zb*m}=W}deNw8EmiH!&p#AG^K3h)vW^U@a#&y3j|U`|_d6t)|!R=nvs}rLsff{!{^c
ztlSD!g?h4oP2BU{HYMBFqy7M`Q=`4iWXg1Ro?S7`+r0Td(4*|(u%6-5>Dp>DqqXid
zx;Dw#bXI?nu>TWg`q}Y1&K_M%ZBBpM8e6+N3A?hn(Fk7wG+6q3(&oi8KRpU8c29qZ
z`ARgTa-+C<`$SgCB)F(QBuR>Ua5Y_$Siw8_v@IP6gj<dbN-x>6c!oLwt%ro(v$niH
zqN~V}Mp+$Ixhw;%$mPTnm@qp3r8EL;)>S!hd$Vj)YkZx`CbzW##s5Ue{-5?h_EM+f
z0PDQrSP$h!Y<>Qq+>mVuB5-vM2F9Y6#{0-N<{Mc=As^i_{q(8LzPqHTsIQ2=6jdvR
zBfr!(=0Lh1G$rV@6g{Qxs-u%yfrxAYxk&VkwI-D;QMho2wdZg0y2-zyBXH?Oa$Q{P
z@{w?)Cf0AOcfCn<QMUQN$9k47wjKWO`&%Kshoci$sP@B#Ua?1W(J5=)*-ijPhwb!!
zar__#a91_3oSbB}f7u=j(7Eb!LwMYY^&cL!C!(!(H#w4oLVGe1Dyv%Tet+|nLq=wl
z8ryxzC^CuRWN&-o8A2TH2t`q2Py07@#H!dcN#`XW3)6cxKJcAI+8U&Uz12ENVv-tI
zmARCC^SNB7H7M}*^=B-sZU8<a=VrOTaaCe;rQF*e^Oht3aqCHfSO(^-H*r3jg7E=}
z8>!ra-JioPTztO`Bw4U@XcPu>ChZL<f$MSM_8s2JZDTmGfKz~CPzejMk>GJsG?Km;
zABBV5fW{q@N0XQ11VZ>CNArp3MVHDf@pUqRSVM@@gLXQBmN^F;or2w;4eV(_Jzhfc
zs6t52`E*eIz#z|8NLJzs_ki{_ej^|iE+e2qN*F0gl{42a<e<eGtpsF8Y|Kf&*q)Ta
z{$kaJLh{hYaTx5|{ULJ_xZaKCUOa?^Nc|yVMyAq|3+>$i4(AL`WqpoOt*D>gOESjT
zGGy;3I$BVYH`xb4oeB&1@ZX&&MGc9U$|+_c8vm%mu9pLp%hJcyGbKPR=Zt+4NULo?
z+Wj;d_;Kzg_|M{*yxt{b<1}HP2b78eIbvT_Ewm}5AEKOm%&<7qfPS1+v|TuWzlzo+
z4uxm0BC>rQ{T;k8$X5O)e{m2Vv2T-z@1Abo`7$gClYxEjE!|tfe#j-~ilDr~%B$jn
z#E;$wH?}VODaP0FBo%?Ao1d#0;QIx%ieC67dUMhdwJH4#wR!B<<b&9$NK9zI#n+-w
zC=x})pxL1~n6`8I?X)ZMr>-MicI6=#o65bJg^<*B)uA}UMLzK2s}IG=2m2$WMP<6K
zITS`b_SJ}8d+1^bffI_AP%p~4vi8?S_c+KF%K)x=cKx9^fpKUf`){k=aHt4hWU;Kb
z8>1_kx}~wv4b}a(ZaP#15Evp^pl`1FMF{pqwOituOD(SLzcpv8niIHqM-%Xh_#-XP
zD|UN=11D%kvu+c7Zb-JWp-s%0!gUYY9S#JdC)(FZ)vM!}Q77k6jp%D}VLB8NqrAjR
z4vnda#!f77_0*`Sw`l=S*(lm~P~{9Cg4-xc@|pP!sE@=hvm9G^Ahg)*YCekTD*8d2
z<8Ult+pAF6+-M@j=<9A3wRzcVO)beRN~*kjqM2B^qWTmLy`WkP!Zp|*<D%)Joa_8f
zlA_xZ2kFh-h)zobDKcH1uw~v5Nf1s;Q8RByN`bp=MFK%_xdw`wVdGwzbX3F(%%8>e
zgM9I}D!@|(8yx!I+R5Ryrh2BryYfOmZ)*>w&b$r{Ry!GF!14nX3QD4axBe@0TNlq<
zRpFRhpLCD!w(QUws?I#KVo?li%u(5f@M+GNSFJYJx{VUI#ei6Us0g8dCjCoeHPQ)c
zn2IO$T2nj|PK)A)DQ@;pAUkVG5>TG%)*MiAo6rb3Zmm6^@%=DI99j9TBcH*;b(km(
zOIJLLLx-@NBepe=EIrkM!u3iZ483lP*6txw{3&fm;$#hH@h0*fIg~Z6JM2;a=k!1P
z!FE<-ACqq+44}_;#W<QHo>xUX?-KfB)i_5>><n(7_P9@(6xxu<6FL7?-PCc7)^fsC
z3SQ;Pe#(jLFK`T8cXxa9?qccLlK`VrZ*L~EKzxqd6!9kl?PYuB(8W@_$0Y<=k1lhp
z0(s+=6P~R$!FcXeKiBpKr5}Q~k>~o5-{y!_;G!e{N6AG}`{_bEV*9Fn@j8pJKbg)$
z$_RlmyeGj%ou(^zZ-9-m0*oHw95`fsxpF)!KD}Jq3bVET{2I$aBYbWkTDpjJNu03_
z=5$KB{jwgMq*6Q%c^}eG6wAzTg?9uGXzUC^-H0zana*uA<{P^d+X!~nw{}UPh!SAH
zPUIR+pE=D=dS4z7M>#tcu{oVQg!*<J-F7zF0oIT$WDNrc=dwk6vnt7*Pv)^=p}mn?
zlk*p+eQrVs;}<T$0>QpE79gtR#7-9DtRx(+^KRZgY8P`6YHQH8!0X@kHNvW++sQlm
zdVUFz+ivexSbbSG5?GP^|D}a_FCpzH2Kx5{b*>E*H667Nyzezlq`E?8%DJ>R+eeA1
z)tFO0&YqCrq25n&O1QCMC9+Qwrbe_p_2B3;?`SnSZtU|I79I}SNj^e9l%3B<h5cm(
zm&7Ca03EWgl6S3;mQjV%zm8U}13+>{i25d5$xpo)F2232FOezbTz#-wltkO@dxz{l
z{OeyQ*$+7@R9aXCe~fRv7KppX{1nf0;X4glaa*^ay^A-WfXDvgeWMFSi1{@^<cdX%
z1|su8`z^<)BddC+*+2+3D4G<;F$(!!F;JuyDwB0@!L55B#2hY&V)s}1+dt#=4Pvj3
z@1?HlgKU(3_?md8)O(zzyg6Xk`ZHN=Fw@p->SpZgyr;1qZ0hKF{1^B})owRrADqYN
z7-5UUcH=;5f-2t&T912x-xS|CEA>dpo1<5Z3P5qoKp3_D3A(>^AVd#>$CO!<bdLie
z1EN=e&Tk*MDC;HV3Q*+?xa{VBEE0kh5GJ{O;y}?FwQaULDwqc)j<MFGHYvVy(WTr!
zB{snLC&wG;KJ`jEW;k@Gc#n_#t4(zX#(L(>vuOh%UUI^8Hc0HU8JBhL4jFDenHiwy
zS|?{U8OmqHGcA?m=+_{{+3^g!2NU8i*Qq(t>tUiq4)AQ;Z)$F|Mv==IEi3dPo0sH)
zHLQ!P<~<Jlq<^yo6$~Vo6B)pp7deKvHZun<+e@;E=oqPM2-#$54DTij`OIbtlVv_s
zQLh#;DHjl%0roYR<%)qKPqdS1Ze=VyLJ)V7RWXpqgLpO3H3K1F19`nt?W}cdqulDZ
z2P)L6tfUZ_P^`-}L36e~DY&b<wbeHG8#eD2+n8_g<?FKA07gh3vikU?OIDvX256eE
z6u!!A+*E<95NipW$8B@AgZ5r3Y<j?25|-JstTo;vCKQ9u5S*bs-@o~ob;NrYP>ZNz
zU9qF2?9eqz#gJutQb2~TjazdTYdX3cY+Jys9evt%<YxKjZt1cf&cz^gfb8Gf?a{>V
z@}kVgMLTfni*V@;ewP!?fBw(*SU?Hk7HQ)gX3EDCia&8>uqSd{B?xV?Cw<A40`{q#
z?WF%?yQ3coc+%RcGVUI4Ivppir~P#?B69{RaMhlPPjq5QeWduHXA>OH2~zH!djn~T
z3Y9kKb?p-BOe*L^Ns6DV2>4V{KI8BitYTk~C#UrO<V5#zeDzc#Yp)gdWxa{L4rSy3
z`+d=^U-e6R5`a{q&i?!ySzD&v^T=)kj?p!1Z7@5+g}+363=LeouGqMuR<oPhh6h}E
z^}i8R&|51obfhBP*I>w_xiBkL+x$dD0UVJ<yg%ua#3vH{)ajh@U$C>B^+hbn1>m&j
z5=N@{%KB&LtH~;Too=85!JXcS$+D!V0wpw3%q&*ql32q+z*1<*pL8sF6v4=yF!PJi
zhda7T;=xnf=R)ms9dhsF7i3+PCjVVW=iL9>-m6wuNp@W_)ZY<%zk-Y!IkDm&B(1>M
z{Tifux+i}^y=fnMzfBL>M@iP?-2)>ZJGLYTHP|Q7Gy40tcCk%$pnjSVHu?qpv&6pY
zI~4ca&vTZX0WMzm*%txNC6`8Ubgnq<9mg8=Wgaqx$x<i+FK$N+>G(1G+Sf6Covt2@
zC4cb|EI8kc1JK!C!uh_<Wr6ie)~|{ovy?dN+IRUyqaw%aIS;<~4T84Oen?1*S*L_H
z(?!Y9h*S3|`!PFba0Cw|e=1HF<B;E;eoFkj`FA{jcG=H~U*u>!ntrL4@ltz|0$zSi
zsJJyT4bFw%D)bL492a>Ib<yD>8?01B{1u0b1h9CkU3s{iKdTq7v8xWdU?pDy#o!;}
znt%1-xOUf3_lf!#$oHLg&EX<+lmLak-`TZ?Lqu7GMg6+NE_FziCudz5@m=b;-n$nA
zBlE1+YQ5obk=q(6>Hscp3;;a&EUXEP-*mXha%EbwoZlS%NA)kXkrMjlp?-_ckI4J&
zcB{{-8%}jw4okI3LNxfV)o%A;F`~&kYeEbQ7sD%Cv6-*h#CYZ*ghhXDcN`8bRI?Q}
zUQ<gL9YalumO6?`q#*Lh<U~RdltlgMH#z073xw&>)K_d_RV<H9tzbEWo==s=$uW62
zPkoRk33i4fTnX7bZDut-8b#>sAiB<3hePm^a;)TOw%O6lBPnzcoV7XeO#E3&MKrcH
zHwUYYd&s<h*GTDS?s5D<BMLxU;K-hS*GyX!jdfQp2NLSDC5J<3r6i2sLzX)Ia3?9z
z?aL}`p`$EnaySm+^QbM4amZl7MPWsPR!?$TW$j)W$h=&oIF8I!73insN>Y4**TgeY
zWN;J~-I=ZRXZWiZHE4<Zc)+P?tY`UlTjw-bY|Yr4ZGAPa{02P?;L<!&8xs8;4Mjl+
zX>W`!MEtd!CbbC~XLfFo`kXELo5R+aU$nP9Bz0v|d=XTN{`9nMPHN&}eq5(8w)7C#
zi=f|<J?chlt<}_z;WSMc1>C$naL)z5Y$VCl^dqg4FWc5;yey?3>$EQK4fN=|*_x9?
zlr4fZ_P1OhdBng_ZAZZ8C!g~=#5nHk#mHn2dyfd=T^(EOQAhckyMSwF!s@a-Gdpd&
zDx3{gSDD9hNUGH~Zz9F&<2j*XGZ#ww95MEUQ(ar+`*_l+mSo)aRCMer1XsZC4j!ih
z=);q2Pt`<ZBN6SOJ)I;nf1y3&%^@MO;!r!*JMScm@7~L)yBPIcxAUBD<JDc;aR=EK
zOC>Lp1GYa|>U63d^~BhRkmV#(9D~F3CNJ_V-jRQ#uOdvO2zDp6*@1Oj<<*}UESHyJ
zy|96t#q8-e7%R|&&T#0`8P(hHC3sb*;~DZtd<o`PIg&;l_&=8{DkpqiV?ArAyK5)&
z*Am%lPbX5bE6N97>Fdr`=o<GTMX-gBoO9qPHPd0|9l!UlSZr?u;uJbWYD$M%$)|B(
zw6P=)@*q*51vx5)bl;qEk+rw85rR{l?0WC`=~%zT-p%GHSY-$VmG)jdlVM-k7yG#X
zzMt4(Bg*SlDU%A0>;s3ihP*c)dbi8}#Xd?j;FN}DD1T8UYP)@sLn4`bpz#^|G@i-Q
zB=;pbhkWL}t@x9Do?n0gS)-s6`ywcTtUmymx-ssvaTtFY-$+q90I0H0d=--@nf=OS
zm25`A>t82^8ZpXC*89en*mGZh>kIRcZY1AjOJX^26|T4MtK~U~2H}SQ>rxf93v-HP
ze%XEuuw;prO0OS-^-lqouqtitxV`u13SimTC0LV8MOIe(B}fn*3E3udu6PH+Yr)bP
z`z=0YCt}{eR5@aUE(K{g!i6OT+ZFLF^{pfD$1CF*Tn(_wU%P5Bq@^v=Tk@Y6<<)~B
z9xX?sea&FF?=)=MM7n^%BHXMZyG?fod0ws?OmqCd3vC2Kd1T0=^N?LXSbRy8E%?&^
z*$snn&_?cx7Q1n<h<r=ojCRvt5!&R{g<jfj9t;ryy<1EDL$bc#5;GEE%jQ(7%Q}Re
zL^PtU_%C+bV8|lUOUl{p_QBL}p$7`e_ag`*<<Oasvj-P8aqyy>6H7c0ii1i~az_PD
zb62a%cv62hky+os#-PZQ#J|bW!=(gNWHg%+(94jdD<#>>ZK^*b6AnbX-9vk)MY~vE
zbg6?jBc92$MZpeK9>;BF_J=f2sDHU9&x+A6b!1mRM{V|Hy@poGpW(HOc3}sx3K&4<
zCOS5;)#o{KiG4?`X7k+Rzj1|9#=cXwAciZ+;2<&iO-s3T&=$oT2FGBazmh|w+gmai
zLh8zf*74Bur8z-tX}bC=lxSHlCEiHTDG_*1Eo9jt17!>#%>wpR&7Si_=^$46XA2SH
zvFri$Dko`54Xf(Snw;QfY|v}Rv4UjH4URkK1G&-*YqpSuq8{z_y1{Y>QKriE@rA<l
zCC#2fu-{MtPgWs%LDyj$vzLZ#<lHaydTPCw^;6bceeQ(L?f5h(2(dBu-NNFY<W1+-
z)!F8xGHK-06QYZ?Borz#QMAwptTmxv8Kih3*@Ly`I1BlTE!1Hhj>B4|uIq{~q{>lB
zALSd`>hH0a?67SP#q^m}o~=4pHK}kXuhv+RY{ff2*7(t+8{A?Q9=FpcC#M=zth=Hw
zWrkPEt8>_H9~&(9y?ImZ@#-z9u$;-;6FH4~_P9d6?MY`AV?CsR#831oAAt2Zh>C6Z
zW%Cm6JZO93S;UOOgb&!$@eH3)h7ush?U{;Qs9+}Xli>H#^aIa2PL|xi%J$}xJ){Eb
z&*d+Cptjh)=-L%uMGI*At7p`sW;j)8=t(k>`3$zgTzac!)F;)x=AO|P^Fz7G+0dWp
zMX!RiVFR%calmj^W|5-7L>#2=*)@iutt-32wYY4|;p)u@*wAK9_3TJ)M64TlAa>el
z0?=6K(N1Io%4y;Mzuive@@VU}+v(_9$vQ%?FP!+iopp4sMyNSgt;Vt9J<q<j^D%Eu
zuebhgZ&VOUy)QX^7jQ}kej&asE?rQ8Vj#yH23#an-@z^>NNW(VTkP%V8GjWh<{by4
z7A@ktK>$*JNbo35{794i{97ui_p0|VV?dY1Q5t4>K)ERLegG+X0ZSJGkt8jhG#>=}
z!r8+<%w8_Z_IPP8A328Mvgi2^uccf_=mhr1M*gHC6GFaq5r<=bnwxjgD*Md;-LW0p
z-RJo=x|fEc3;V*q9_zu8Q-*IT3cif({`=YXRZPP1>GvYn0eziIy^`Xlghtvo(bTOm
zxJ%)^D^=L5_H8Z-%vo2GS6I;R@)>!(%DI3J`TN{CYqXDkNLWk0T*)Ut#y4&v!Ht*2
z{VBQPqDu+~{~W8x+5Ag<Be5i%41V?3?Hp{s#cSn@Vv(_09ydGUa+Q`;v5HC*dBu^i
zUNvHj*lt%INq3}<HU$~N+O9g{5<|p!DIP-Pz}bFvyeUDuJd+cQZZzlZnj_^nX@c6W
zJyHaqJMa9*Np_w8yRL=Rz5YmuG05c|vZLJ)y(uqoA8ZyrP{{7I-FPI7iOWz9G;Ft<
zj)XYZ*wA7(C&bce5vv0_Z;5%z`QR`o=id=)Z#@!H-xlI`?6zu7sqg5AxXG_=w|l$4
z{_PR|-zG#)#TEuMvUE+%o=_rm*&S62Z6*)*p5)UkxtS-&`@<zADa;a5pOW)g+}hBw
z3BT-w2Jty_NV_O8E#G)7eo>f6^Nc4`g+B6W)^XI$%t?bd8Rdqt63seN&YU~{&1P3%
z%7KLiBSFiYTzY~CV0m-%=X^YJkc?tz-?NuyTe>H~#|xysxy=^%V%I#>YKxA9{M&{e
zr?`wPiGkym!>#wAEzMaWw@C}D%F>?6e%TQhWsA&TB>%&rUFpn={&u<d?@FB_jjo8k
zB#kO(m1t0Vnxss=3OHSpv@$^^K12ZqYvL;%(CAs^A1+3Rz_-@n729ZPMuw1*>w#Qx
z48X0nu0o{(vC1i8>+_lBS9FHRO!3UsQo7|Fv5oOeJ5b7Pg~r!bh>kpGMJP}jjd~v!
z+)xJG8gs?Ba3pb0wkBWQ&aJlDm+n6l5Bt}xC9o*gfLjiGRa$Np_mDo~u(c*OxS|s|
zZ0!M+G*CEz!50o%M};A2PF?zBhj6j$a=ec9O#D~d8ed|^;&{;CQQH<@DxhCUWRw=f
zC9fTcNAyc0_OSQ!$fUomw@1AnssG5@f&d7E*cr2M)o60)s+g?pN~rOkQJMkDPuOFL
zBjS(IY@&;DRzZL}eBp`Y8~@io{?4BCAIJ<a)29-$tx5>HJMb=lsPnex$i>N6zFhDF
zgUW>%(cr~CqYX0ZG$&lMH=oXNRNeZS3XunjeSbVg<=F)GFY{Nen!jR=?e&d>8qJ=I
z$sgook;|Y`|M?<_bVy<8D_-(70P+67#Q=6|c`qqtRL_yPh7_v|t_elc+na08v7$^P
zeetCnC?bcK^#{oCAa;iiUh!l6*KMGBBOlmu@8Ylc(O|rBJqK@!9v{)dhN^G49F;wN
zoE@inSi{xWj0Dt@b4Fs~nH90bM<A%kbNV<M$Wi}B>dg|tp&RK5pA$7D{f2t7>Q}y%
zhX{~<)lTQU>$Y?JblO>OQxc?f!9W!_c+S@yHzP%dollx*k#O}!(1hGm4%!7@sX7we
zj8%hqsum>;zZlPa5=s(s&fd;;2=L05Dx3a0$#RLa!twjvXrjHUJI_vTEAM5K)m@~S
zw)dk6E^9@E)iL}*AXv*mf=D*PU?}fDtU5eUWENTAKT5KwsiP8)(n1J@A78deew!_9
z(LTvOE+`^X$NMYxX|#6Im##tS1xOcTpG9l23YlVbAO1XHM&On@OHAYo=UuB?>`P|@
zc=MdGuk!dHkPclTLc!OuB{q@zDYwL!sKu?ex}8dQgl5LRjbWs?I;>})<~#p<3x1#9
zS2I+eTV11+wM5FCA999l-f;c19|NlY#yR;@62bx<B|k^&;+)1XQq=#FLpx|}r9<{>
z5*)gontB8>@3FbFq19}tQgA8j0S6p)RfdXy!ZtX{Mc~S4A%=1o+>$lOt{RHNmB3yv
zigvYsqbi}uz}E~Fp=&x>YOWn}nN3z3ov>m(uC};tsEA{KhMMep|MyrAwa9O%KvEr3
zkMzDVL1L1x+f73uDU!qu7qpv)iV)`B0k;f=ga%7d8jD+_H#Z}?CF!=I(9F$mSK=q9
zyW6XI9u;>)y?(?d#52UVl1xJwo|pp^2g**nV<_ZQm>P;Z#gI=LN;Q~rh-zV!Grtry
zd`q=3CTACD(&Vo`C5K_RdA9DeHZ>8trDn?(n^pk_R|hy1nV8`?<D%bXGZQ~!B*wQ{
zf%8K2V4KT8Q08p!M+v@Uo4_Y~yUp=$(SOKkZ*Gpn=5Sw*5u2B=Ourio#63PPiC3Cz
z!B7$FR{qHr#hj&@NQa|5DHX4`B;LEJ&BJx%7$o0h@MBA3Tu%kyr`NK0rWWp7e9D%`
zvk(S#bG8*p56j`ZEw(bjy$_|Ut#WV?foIs7YLLT4GCj)2Ykdk6CfNf{wKcVMwr<FE
zuR6<NVZ}~^R9#MYRY&yAw%)r&*VSzsqHFh913DzQjSdKzUtl3t>)o%wxL)rRJJz!R
za2pdSSU|xzE7;^;!nTOivCTPWG${%{6W?zcic3ppcgI#88La_wD9dQ1;P?~p=2VZ$
zB+{O^S*OcKhl50vk#z;lD5bQn;_^awrnxmnXYUs3B!<Ygc{ginIyND-?Z{3yk)_@q
z&dGIBhvU%-E#h*{3U}uAi+fTz@peT+VOBizEEN8j_jIqShTG%0&vZ*8Hxab@M1`d@
zu&pO^S3Oiys;WGdaJ1EKw%z&TKJr2DiSG^~Rh+Y@lX|w)P{PBW2|#@FxE|u!{2JxC
z?Y(?=sd^I>BAM^G>}j#rKZ%}P;ALXqe}DEg=^yU2o@h{1e9T#LO|R3`@&@XN_T_sN
zwaRNIYkPk@6JT;(IfmRmke$|HR2cLIY;hP{Dxz}p!NWs$0w-6-k|I9ZaJ<Js2f>Jo
zgoGRxBCbR>;y@9ad&EWq05-b#Bp%^s?L?xUxM7L6C-XP{P}0zy&fjEwl{WuuA`0S#
zU!L<V_?N|YK5*+Ub|HOmfjnSuRBz-dQ|>DODv$OH0gx`5;wHl4SW+bXV2fRJ@W*=Q
z%(1ttH@L9aJBhi~nBm{8Q0pNW+Gg+Nv%&(Tv_50+2Pl525fhWvNiM1%#2aUP+KCF$
zkaF$A3TvwA<|6DzNrq(+KaLiD$v7mXgME_ctO_^PE33(;-b(5Gl>@j)OrOQ-93sr=
z^Gnc#5&)>9wv~O6GyolPj$i0!n-@Q(((0^znPWYKdrk4*{>uBUZNYo*>s-N3%B!J3
z{3h`tW~6<M#Pw}FbAw31i+z{qZfv1m?Dq*hJD-D3MgsdGu)u{7dG1Fahnj9&zMzz!
zlFcowX_pPney%2Xthlraokh8AzxY~~cxn4JV3F$`+Y>f=)P75t5y%cBaml~gj=KEA
zg_E5j${(;Rju!EvVH@QTkiM@xT7;Qql002?v>bPf7TMKDi||s9Ba~g^4QknZ*G6}e
z2Bnasf{9&sG(CmzziHJyNwLh=`$wp3?S`W+BDulI_n@=h8@;=+9{F%<t?^*H>8ML*
zn%QByIR=yrsItyS5H)XcDDSJYTl0H-53mCw=+xP5M??5QSxfZ=&wYM-bR!M8tj4dg
zuO>KzEe%_3;?X#48u4|~2EXHI5xG#UFqKL9G43le49UH3^3fuOiNlwSU{fkMQS*Zs
zC8wY2I4l9G)8c)(!a+|&2iT0GAz2YY*iui5#(JCSOM|ag@z9&)xWIrSVPy7E7rS&o
z$^pg`xPwX7UpnDAISbs^$}Hx_pvn|OggZR^jLoY6ajUDC_B}qCV!IRjTHu%gcebc%
ze!6%%%L8vo^-OWp&ZL&c8W+I<9LX8O*LYd9mcP9W@^BJco{(}ULVML>EAl}d+d#LZ
zmGPAy9daNWv{n90#p7xS;N<FvdMPo4qT{5ktss*?CH3YJd%#zrQ8L6jr<}F-+4_Lj
z)e9;(t81|s@rHO4Ih*K1`$h>YH&(s78bk`V+FY>3r7mB6VsaCE8PaJyS}qdkYtXIf
zuQfRcO|7;$29mxii@!U{S{!8DvF2N=_W%iyW>*~1+I`A+_kla>h+aJF2;LHiMczwS
zeu*ENYX4CAEheBs7)OitY0$PM#7c<7wReY)y?Djy#rAOQ5KNViI@pi=)pq*U80%@o
zO><ZD=u(wuw|gu{A%BXF**<QJkNf8vrq~nJ7xK>#xsb^zVQ6eA9?(xEQmHvskCG@m
z)OrRF*yErnDZ<kZ&_m^7|4g7m9ZShl#ctTM-i){~v^aY&X?FB7a(ylaW0Rp*kx(;e
z`^KNij>X1%-S#^G_aPqkIPt=MP;tp6w>PPoATW{%;u~H~x>UgBX5Vl9)w3Z~hlfhK
zt$}!Um=v!=+Tp0v2YtE8cdPu#L)o=BkGdXCT%o6d!bcL7e`;x~Wn+!{ax&`{JCW}?
zbuntOlQ}I+7AQeD%}4FDw_n}{6FghJcS#8^VLO-HRcUKlF%qbn$N6X|*#x>O`l~nM
znfurHaeu+5qQkMf)yA@W99oZ*%{pKg<CzN>!v0x%J9oqCR(r>Lkj}WS#oo=iH*wqR
zviE#RrV{w^evV#j`Mw`M2=-aRdqA#&l(5z^bzjjBbHB&|qeL8hR6#gg%x_BU{c%Dl
zRRYuviP<NKSoWxn{7<8e>XRKUJK(f^mJ5dq3Y(EIx2kg-y9=dK88@B{5UwE0xU9Ng
z=6;(r>38;3&Y$bvAMESw6FX@+slKTOF4WHz_q2VR3<5t0(Ude-9%}ZTZ_4^s`#!s+
zn-2RSN9=5Awjcc!vp5Vh{V4`Ggg8M47{uuT``MeLPOigN=8M%`_uzK>)n9aWbBx+=
zNjNoi|D{{5z2fpVoZ1;rlOuiW6~itql;aA#NQ>(WcxAk)h=ZBPPI}U=8osDVFe2<;
zv#X;|0Vi=;e!kaK<9Ny`6}-JRp1EyS3i+4pI)5frrJ_0tjMonr@#yZm@3tG_6Std5
zkU4HQ4u|MK;=ht-^QPfY(9|^QwqQ38`>9>d5H`no2u^3W3^6*jZ;4q+#*9>_G2dHb
zI4nY#I;7q|X}9HY-KdphXuZAq+!>zgQac}>;DD2P-6lGI2*zJ!cMKQNhT@k9OOxVj
zDZEv6fuxA+335`~<ai^`s6#wc)Uu1bQ@jJESM0Q@IopngVi})SVHIa-U{>y^8QEsF
zn*e5d2TS2WZ8j@DP_BoJ!<p?L{Bgq6ssG6T+nhuMEJ~SrB*f0Go~gv5E)@rC-f*bx
z?v>ZBN>1IA-~j(XA(n%-AfCBJs_0p4ks}CMgq%n59;a?yZ?I{O+R}VRtuJk&GuU{S
z<!Go-k>@#ImpjJiuUcd){GV&Kx7*6<TRHB@y!WcD8ZIJuO*j7PoUUuUbzMW7tsQpp
zSJvcL?1AdrQD$B&d$!J7Pz>j8!phc1dx@oIY=gJ=umc2BZL*Eop6e|F%dl&H)ECww
zXjmDiDzDifz4|1WKh;>H!;0fphl;^9`KxAP$ZWH(B#9bPina`w%L5FgHEE|DzuH<E
zw$Pq9Yo}sE@dxcl=%~@qYF+WAL^15iQfjkhbPL|<LlIQYg?d|lRNvf=$dW|Nf(?Nr
z2{+C$9(K@C=a=jdj|MiO7CBr>;cGi{c55hIi$9Ui4vX}9dn}rJsv$hz(U?9yJZ`n;
zwj*x;ghNImCJwbH<C`Mki%>9QKb1ptl-yvutL?zefbaJN_~o=?%M2Ims`P75M`v(b
zE<cB`osQTu4q#_ZV}m_g4d#1aXVKoouqs8DIKSt-73-&rd0owojh(hHzLe!qtVO1j
z{ljr_TGi05(!F&xD7$(R(+ibBzBfNZs8e+F3C@?kc&2bv#Z??fJ<*?2+=4d<rmunE
zfpW76cX&>>%DiWTfdWrnCgWybF|eMuAxAEEJGAJ-F_aqwWJkk_<;W$GHlb;F6U9h6
z=cI@)H572mk<+T;W=`b#;p?Sq1iSWRE{_tLpv%eSoG@X6oejYJ(khTsx;__g6mB$3
z=g0BzDwjJu?<=$ziOt?{^l9gUste;QtzCz<6DwUwku(<V6l8kNE{>zQT$!Y*e>)d~
zYj}&jlW>8mVUVtD02-b0Zq*pF-(u-wB*|ri_FgVj@sDuD*!#Iqcq^%v)(4IYN)EMd
zLPq(p0;KrQA@Urwj}nkISOe_iYDr+gOp}S)Covi2t~qX>X8WaU?pto3MSHo5AWE0?
zKc71g+qMNEei3YR;Trog*eKaAzlxRBA(aaFl5B`cYT{peQ^lKnlS6hs1e5wUKD<QX
zCnk}(6zscfy<C}lzK_<SDrDyo%>)+lHIOfl{cy<yB>#~U$$rdd(8s~T;m_DlxexK6
zaIc%6^GVcdLRfN2QBpShHTsAQMkNK<Z?Ts0`IXS1UGAzVI~Hmk>Qx*C3i+K~5zk!2
zE8YNh<*|xgk*}hJ9bT1>s1HPq9}?Z!)zz1>;T&Wm57;%wLX24mkHXG$?Xi$^79zb%
z-GAM&B1dpZ?LymgtO#JW(w*FJEc}0JiRQE$j|ErZE<~bjcR9W3nA^{OKhthbaB4_-
z>KdF|j=2Qr?7t-DgHn!Q#=h0Rn}6R*yDc#Qw8~^At?2F9S30(l7a*rmTe2M6#OgD-
zBzl?PJ9288sqV>>0)P}W$E`S%eQw1;VA!S{3p-sa6>qz2sv}|XwuU;J7USq>RZPz7
znDu5vxA*~SYqI@a|2$wbb3jN;>Z$XEDI-1Etb|P_ca`Xw9S~q|<Tf%Vag9$|oz2Z9
z*+e|=ym1JM#3>`~J^73nWCcCSj=Uf{L`%EN7G;Nv$dhABj+G<5v(}d8b28Dk*s|!S
zFs~vaow4OUcypVSa4Qn+t!=f+55BVcK$};dJZEf`KO5^=)X=E>fQq_WQ?--cljMb{
z$<`)06qZ4?RUI=W`ul-+BZ~JMq|D|LIoBoH$b`iySZC`i+{+D3f@~V*HpGy5Vk-yW
z#++?g%UWO88gAHDe=OXlim0rCFE1C~=veqmb3NJLni2}agiG3?%`uidK~=<xGs;?W
z?Ju)kXzg2_ygZrb1Zz(`6sagsPkN2ltRpd5&zavMhcNuHbHZpmM_B4pZp;9mTl0;w
zr1|*+ropOh^NEi2{ONw%k+92<rv3hK1$#;2U*eY@^$)l$HQLS!j~PmG+;$xcfgIG;
zk@m3D9t-?k%8cDo(~soPXOAc0)JfRH1XlH`+fi|$wkI9O_b$J8&A-@FPNEH@mvrV(
zv!llLcrUt<AOTalt?K1mNmM2Bd&WPUI>(;Psn#}-(-;o6_t?cHB?5o!|Cny~|Jz%8
zuEKYvj1m`TJMK#|gEuzW{saRxAHm0)tS4VfPO;t!|75aZLFIVYS3Of9V&4?~z9ySE
z!fn8N&;iEs4rXhtCS;W4N_5bMl7*o_X`=kwcoPU)XU(fN;&`QMhK*)x60RZLoJfcp
zTiPBH(VfiIEBT;0?R2)6nh3L)_jK1?)9hRh<UYP{k#Wh+$1}{(#c$*oBGVi3Eb3rg
zkn8q^#0$O{?KYN7Uk6h+S#P%6#hA8=Dju`9EBu$jU`S%}K*Zs}-f?hcX4<On-i?-S
zdzWfK`b#Bxe9w{E)r83Neopefwvxs5gTw;YpyCw!VZw(<O5%Wouak6aEj1nR@Q<?x
zbOy!!IzNC@Q4J+`Ve^CbS^oRaYwhzug;ph1B3}f}Z`x#E=3YbJg4YMq+gJW_QA4eL
z9W&!wj-HI{A|bPV6TN!=Ue_Fao6E)~q@V8+h`GPB?-SJfnsLzl!AF=r`}g)^{B!0k
z`{`IX7Zgu-WWu6}mDjtjf|+{Z=R|^1AwuoH<VsTvRi=5|-t5;r&{ng*e~Sk0TOr%&
zNwX1`gOuJ$ftormu83#p+n|yaS61yxP7rh&y>`_|h#7duVY0czBjD<hBF&&K>9%X)
z%K^M2WKKQ?0bW}@<5(}U<Sx5zB#bj03B%=tQ@;L^ph~{RcKwLo>vbhLY&VRAT-FK8
zk!E+3-8fR@DQXIV@|#9NvLFT#Yvj!Vs+)}q^AO{P1Kbjzcok7;m(JR)F_aSO4cl!w
z)Dki~wLPSvZXb!<<DNwVv8Qm!2_wO+Nw1`$!3{RC8h4y^c|gP+*_Hxe+e(V}Ne(};
zuLuZ}11vFhZrYsO_+nSj$lr&7_EnT$Q*#aCrZ1mjT@6m#w2>krl6k;pjHD|Fv1PJ(
z%#5D26QtGAC41IL$g2oS4}!tj(X-P!0vPu*kDZ&7xWy5!ufyg#wlKL#=RGfHQPbUM
z_xKN%hz=Jd9&u{g%)!E=Mcz|!L*VC<{B|iu)1?)RXG#k4*PxqaJ``mR#VnU63O6;#
zmV=|$iueYhQGshnNyltug7;u6lKQHg<q}UKu_nLZ2G7kD|I3$cZ9L;9Swf;zSoVPg
zzF}MOwFcUCj^I0qAX}dZ-a^41xomFmnE)mYH|De!P{y@3Ixjap#D+dVtoLWd!LzP`
z@-pDDF}tjTx-{FST<M~BDqg9Ou$&e*I~Hpj>QIL@SWAo`+3F}GD0;6oM<`kAn~SNm
zJ*XUqE0(__Cy4M<!cyw3D>{--H(?}PO<%LE)ib3fD>o(E7SB}4MOq;$wc3uP9yo<0
z%kxOBp;B_*P>*rh9?ePW+#!X(?W|xrkI`#aPPrH%1jaE)P=Th8#TzN@WYJO)0rvrY
zJbKVlzGhD(i4gdQ#vHsp>8mz><-PV)j=s2*8E&=RNh-D7Sa<g1>(x>Ou_vF7dCKM?
z)vW5JJTp=R{Dlkb*~9{-+t&+bhHP(ATf4ipJQu%Jh~U4E{de1!)5UY@W!oRWl*4$)
zdi+;g%T|HiTj7%=lI1=v+n<uD^;J*}7r!jDDb}AmbZw1%GX|=UUr~s^%-DQfT%i-g
zAreECEE9e^Hk8xBwO4`7XTj#MgH&I$3BtCKXrYrvdUveTHd+CrW!F{WxShyJVv{>=
zC!>XAEkbZgtp4fCpeRJex3GRY8?6QHuut30`3g>-Ior-V_BT$kH=?0pS9H5k0`?2u
zu%WTh#{8e84zi00K#5@SL*VUrhPJ1K#;V5HJHBhB$H7pRv_OiWmhC$l>^*<W_C`VS
zzW09b+Qs%k1+av8X-Eg`!=!>mB`=MA<Omt-Ax49GAl2A%>2<f#Pm*l6G?Xx*PWv=k
z0dI-y*7o?!S9yMD?miDJdRlC;z>(N;U&IO==3wZzFa1|@7wr46eU-qi#O0xm3bxT!
z%C%SQn<S(^5wJ@cw}iaZ8Tl@MZ2=4ReRfnwPMzkh_JgktH=!SWHnnK1H*t;QD)dv#
z#%&`yg^t<J{!Et1Mu?W{PQCpSGJz;vlB7aZB~_v$VwWAXUq`}oW`(vW!entv`Yom)
z3&3f!<8k;=damUBxZ?Olui4htZC4(5q2uykO<kQ8N-%0y9e42v=L5t(sg_UL)yG4$
zm`~ChT$!`G``hi>Xg+`(MQ*MV&3IilU(g{9@b%dv%GXkc8F}M|Y(iZsPO%&FXM6KD
zXz!-@!WE^S9{lFxsT5L{A|+bl#Jr_yD0Qv)@BzE^xXWn5n!Euie5|d{ptl_lEfZN)
zTkQ5|SF~PCRFN>wuP0PJaN{aG=X6;E9G$CgR4mIyQ<os`sJ<yzSx_nN&tIFA*!s&F
zo1F8+Z-TnLQxbgDhbu#zvZ>WPaf^ZvV)$&bX~#oSSzKQ#a?gm-{MLaVcbUCxTr=a1
zE5c;4i5z7%D+epdrA2|S*z9;#?l!=y%<6OU83&CNKB(VqZZ%$UgT$E#+GM+(wt4YJ
zzXpr^q4Q9X+!KW6b{P*um2O$-L-9DV1;>lDwvs#Nq8L(cE$U<lfBGf9SS-wI{92kV
zU@ugVkiEv1T{czcp^oz9-ZjarZN>3WIW8pA?#il{V)88K>Z-&ywg6J{h^%lC@`i+W
zH+N<ptF^w4+oi&O;F6Behe*q0>yo;#vHjlGXNy8tI8SnfZ#W(je2I8%)IDiqezaV<
zachs4qjmiwRv*aItmLbOT<$~+KBU@jjPBhX6l)XmOjAyE6;`p$*?{Yl(uOOGVM{!N
zOXxDMJfhZ`AFU*+uss<E$vj)csXOukB2q~O)-~R(Y$Ym0W?Pe}rLXF;ZI1lu7{7No
z%3xA0_OK5$88g`wdo&@D0<kC+hiqp&lZ{oenv(N&jknWPP^wOQEFoU~&&%!cssZ7$
zjv;#@fmYhQ7IHN_>AQz}5n5<&%TL8e+5<=yMX9md7ebkqY8!1&PU%l5yvd|peV|k(
zE&@JNZ3zViI`tKnW6$PhrLdojMSHWeM%?u}?K#KKjA=7%U-VHnBUJrzIf^y${XzW$
zY^&qeb3ATYGN70A_r2A&Q}6>iY52G2)9G`Z<FV9b{Q<Y~9jo|`;*kcbZCMH>6oXPR
z-<M@LZiDenNd>EUaS0iUZ)B2^g$&QSBAIX~iM(vMnujcCct@0@K=T-h-bAO$z(GKe
zjYey!S`<)=R*G@tlX1dPbKm?2{vX!f2gK`o-S>Tnh=_<zHzFdU<MFm#ZkJbGEtjj^
zk@8>WHkVgjF0b3;ZvN?v&A&Z=OkE{zcPApc-cCfv8xfO##>q^laU92S98JebA|fIp
zA|fIpA|fIpBI3T@pXYhLWBboJ`*K!`CG-1!Ki|)v=kt7?&-4GWq3T0L8%QEMZo~0R
zntMe^pM|=P#53OjnO#rVXn;WJHpb#TRl;?}p|d<`=W>Hunzo74Jn!vi^Pi0;cHK)_
z>52TYl6aC!-s;gur_QuD{cqTbSL`is<Z%{n$8dd+WO?WuhTnL{R|U6k@A{wJD30Ds
ztf2TKgy8)EN{Iui4!k_wK1c$E&w%WeMehFaGKeD~0C8P)N7;*ylA`=y4K<DSabgZ;
z8;ZW5NctqlT`kM5eVT`*76;G<Xz^zWFjO3c^?=fT?#mtTS=en~<lpx-*_S>RA(4tP
z`6|J>wWh8c4aC>g9O1i}-65`H=fS^;F+7Gy^!}iIoAai0EP1kX-e`f9$r-Qb=uK`L
z;Xw*51Uc-7c;hBbx8FMX_%Y{7u=r0&eDP#!>V)`q+Ry%44y9%TaZlX0;*WO4KoIJ#
z63ihSxpJVeP!BbbvFxe==P)!CmBsa(a0;Bex_U$DG{m!)3KyWd)7K0H@6oxP;;YvV
zq)EVCl>2F2L<hD~X$r0zm?&xb!P<W<&8AaFMwXYu^#jSUsgw?+1e44S1HmtlF+-kI
zH~Q#p)W+&;v76!xi9PcC(LudA=hxoUPNt)T26fKt)>sr&gd-$cm5G*abD-cZM9N72
zq_#oOnv&y_kzBv0X7AjMciJ5}<5Ju1W$vcqHZ4IeCq<H9vOUxDqra7wWyU}-wemvJ
zJu@?Uk#~&zki@&p@(GRiZ1@|S6VrB1NIBvAgT&oA5Il54>t)&d=K63zRYh#)Memq>
zibh1aS?<aaRW=@S4Kmx^6>z9J0dQ2+!XY|8TK5smD6x-Jh;~oHbrG?t3kHI#UM?N5
zEgC5N5SN54@n&1gZM8J|K`|%x8SfidpZt_ev7DCW)x;<P4I=fkp(_UB&LF&!^r#r>
zSNZ^(DcQCv@b4E{HrWSxHy(xHugTp&ya;>9+T2o{ILQ@d>wJ!wNFiqHt2q{(Sqbmh
z5RmZ$sXPI%1!OkH8|g?&8QP-kRj|l|37^^Aj<q==rlM{<c<OVN9_xxFvd=bE;Fppb
z@E0FBv7e2Jn^^L;#Fsj-lG}yw&DEE>K}%f~HVN~VwdS}Bp*?I&djcDUKV+_>S}SQ9
z3h8~tx~dgHg4uw{Qo36QLTsTv5<HY=b=@{l_&idp?#QuL38x7Z^F1PCvF(g;Bwxsz
zZa=r(u4*A<Lx{7!JEw;!3Ioey)z*>+i>atq86q?%*9~@$=Q(<~p}w5WClcn%VV9qb
zg_L`ruaz*t;q}yjYv}()0ublM0fh9(dmQR5cxKttaiF047}}rlmA008?sy$Po5TJI
zpS$N0rsR`yuZgp^w>poQcI^51g~uA({(}D+?^!_|C3~?t&v@0<dHPD+ki1XDbrLGl
zT5=5Ri*baL3iDQul>>?EB&QL7#bw+Z&59ty$`m2qmta<%14^^`LUD7z!8)4DX9o2n
z;R=WBXo8slSjKL){%GWGk;O;V2I3jk%wmgEe#gOh=47i76dUs8$|sGJH|#^VqkOiJ
zoE1Vb8A3;ME#cb)6&(wFa1R1MA#hvxVu|NEV&}Z?rN|##6z5}f@sdiF@hX?vcmkW~
zBY2|x)zZfJm!iFjsENargTb4=v)D!3$g-Us7IlJWlX}~qDS1{EZSzjHQ}7%hm!H<V
z)iWJYkM(=chc7|a4fekO64ioz{2==+B|;#oOz^`5&}y6oYV4z5bSclX1NO1Q6ODa`
zeG>0IFEdUu2*01^H*gy&@v53PpH)!B@s;5z7q3%BogKOSoVCyMy_<8S^W)3PWzxQ=
z#!)JHsdmoVmx(e`+fYfwzKTY|s{yu4E=ynMu~~-a;5St-6s@9vsQ|x?6D(7}E24-f
z<y<5FU9|ROH${(Z-}~Ah5a;(p(jtn&C<yw;+-bZC4%<%&NL=0Fmwt}Uq;tVLqu566
zc;Q6Gdx&p@R2KMOaeSg=G`!4njY-L2;>zRB1F<@YQ>kZM0HO0ggf{r9<8J*DTlbP(
zeLSu+Sm365l}ftixM%(lG8fJ8W!j#KMWILM&ZPgj*5{`!EH|C%*IZXUdyOOL$L;zY
z3ogXj?Hj6AxW7Oam9Oc>U(8=$9*ZO(o(J%z<CPy3`l^!a^=4lZK~rMAZ%J@<x5^#3
zn1^l+a1|oLL{ZEH{k<(GwS}}k)Dc2~6_ZM@ATrL{lpLWJcY2$8yfCd~@EWo^qP2W3
zg>jh8Y})Z)6Yp=W?}n&N_x3Q2bvEO;t7p*n&8#L>s)&UekFXS*l}nI5G?E6)$xd;N
ztb;kZ^LVg1MCC(|=2r6{YXS_J)JyX!P-^3nM!8Rj;u-GBVaST1El~oHyM2CZ)~vMo
z#|!ViN{VZ{$Dhyn-7H%WLpkmhPfS~s&(7~FZl&@8TaxR8vB1g!A2h<mRDW=3w65Gs
zu>mj3^?5;LTOMshpWL2&$W}NYDRoE^Zl4{(6K;m9yfL9!tc2)Itz=C$m_7Tqwl;cm
z*#ZkQ-2A%Bo{5a;qFwJFt!Y$E`V9fUOd%khQ89MYSP{D$a|=jzfnDvvJd^aKfY%~@
zK%7olZ8jvED0@|3HI%+gvOb<mg?rj`+yxY6Pm$2!BNFje4jd9`(JLr31|{EoM{4lg
z-x4d)m{MB|!Lv8}>n2h<w8m@k9SRi{Bi`=Mh#e#08{rEgJ{>_CNIw90x^k9W0vaDT
zXz?U+0I&KRL1Ew_JKN$}Q56(m_S=r=!DA4Gf0@AW5ue=jozrb+G}2-rwx!~O>|!Tu
zS01|6YwlfRyM6D+d)i>ea%z67UHnHVo!@Pb=cZ#cw<jFzO*?8z1<)raM=G_yC$p$L
z6$|ghhx54Y$>o)-Z%+Q7&VH3u<(cClGr`()bTiPi)&7#k4`*;+-Te<Nw!P7FA4bGe
z_I&g#X(J*z?1fw<7k|CIm@C8jq>EpdjgalQ_4xY9wMj0?=oK2yrFbBF=#pnoZ~lTj
z^;=(_Dk27?qdQ!!O*mR$ZbxEFsf?63K@``ET^`MGkUWO0zxqXpkPXDILrBZ7*kHn^
z9)0VOPnFEQ)H3mA^=<gEH`quL?@s8V$7POsgYllaZcg=Bf_76_ug}GV`^NEb=R%Y%
z_Ix~(`9W@Tk~_u|J?>p#6S+tNrDR@}L~|+5vaIFDid>@D;K?LwZ~7LN)Ie<ETiykR
zbOM6gI@?qy>Fw;6Gpz_5;AHO{FCxcMbWriulVNgDf!KTTO!_hHw{FJwvmJ1g(1eY_
zKFD1kKN7ys@+3aMvOcWFJW^DIsQc`rSc<Gz1Dtj#`kY{t%s}>WETwP?o%V@u4OVhQ
zwoj}6s5TLx`z)~+-!YKu^Slw2>8=EUe&IbpCfeKW%WAGAiy+Lde7GpqXI~vJG8y9C
zrl!rLrzwoe*FNBQj~Hpx#yjA3zX{aAGRDq)>%Dva&hH$abvx|)9EX!2iyb7XkeXBW
zLvHoL2kb|mJK7D@&|UUZqWis`ui)qWy%2nBYcU%P=7GQpI3DTZuNZVrN@r~!(3}i~
zxUzcVxz1D()sgM0!MNZ%8*5ZA^6J6DQ%TH0YQJVMHAkb;Q*rwpvTNg+b7dtFlU+BM
z+A5xM%JIVil4lD`tLx=@?-j8~iTZCC3{9F&)s2G{+t%)mI5))ry=;imxAU*voUL%e
zhE)Rcw+trb;!>9e>DF9H7k;t$zTf6!vQDwiw+|LZ=V3)7Pf2KS1YhC+{i8x<<$7yV
z<4rMYvQsz)2AuAQH<wcu*RD<TafsBBfo8hT0*a-ysAp_OPTaXtuFRRa`acmCHY+C$
z2aRBBbMociLW}RrEtZX`9$inD&CLxdj%7mDmu+57U=<!ecGqAqQ^*Jno9*twh|DUQ
zJS9Jso%nDm77#f>@YtB|&{1_>yC(rlFm5S}v|w;zuHqp7-E1^lkChF;fef}NcN3!x
z+6JV?ep`~OfN@t@7rDPJjc1<AzN8JZW%(_aOwkO3e#_&T6Zb>Ku(l$$T4i?*+RCc;
zqilj?nLb-JSa{VrbAD@U5-)H9>a(>u1lm7&{6p<*UF<bOz<}ysw_6^z_1QJ|7;zfn
zj%-8p$Bl~_a^qz!#5u_s`$30px%+E#!*PD8K~r3x?VW&Q%pp@`Jd-!uarz*p*c#)R
zG_sgCmD$g>Bz`>HrY!wB1kKgdl}jGy!oqvWEx0we^-oP*+`L5mS54(_D4{f!?;Vp6
zLX+5)GpwA`U$w2iqQzLxY@7dGs)6l@FAu^99<WFJOZXg8)hVANxAx9KH(<`UUHKMI
zw~i(%=5aQ5dtb`&XOG3Wa^#jtX^=qi$NfD%lq>CtL1!?fq);*y=~WB7#o9q(-6X1$
ze9E58#mhI5s7iav&si<*7oD~z;aE=?o9ByuI)LN8k)F@NeaWAxz;UDuat2oW*=Vg?
z+OoA&evS5Ac3nc&O6r!q4x{m&CCa7qyszlO|HBRZLQqj<KuM0tOY@d<YcB>8&-$(P
zB=Cs->kdJBU(C!2ubgfV+JTta%i||N*rLt~g4SDYzoMkcS&YHguY|s&K~1d>+2Lxr
z10|+i8to%K=Njm>9gS|nG<japsO$x&s6V@zJ!_5)BvryKny&_PKVU|P7{<ptyAcNj
z1jRfQ?`#|PMU%<dMshBwuf+_>hkCSn<{DMS_K*4es25OU=lt_dBFE1UR{kNS<aW8#
z$CEzmVrfHfZ4+@q|L^ceyA)HKRF}x%jv6-hW{%BtrCL_j&bQ*3;(T@2OY8J@(BPt<
zp`Aon4?wycN8cHom?biM`R*5fnJ?U^{I7rauCI18&Pev2uXgrtX50ITOhu>UPA-M|
z#x8$Q?Y}hQ8bVqs-6tP90A#K#Ao?f|HM+3<_HiuGbwtvS9kEYxJBuw2HNmI3_tU3O
zx6fi>ZcO?&t{~Wzf1b+s^TCM)0?7Z*@BSsY?2DXVds}f%wlAwijKHbDRO6lT)!@YX
zxaR(6^n@Y)of;)TI(Zd-9ixg?6|Zj?n{VRTK`enqK0@2@ZET$%*P@64E8wA_Y-bR7
zzl#k4lejIucUZ#|lNITQT<LveXW86TV?QPVYVC%~mi_$c7iUeOfk-2;1h3i8-uUta
zk9H#YLq)c>*cB%nO8?87A~MPC_Ae3h$`hdvEXG7s3Q&J{W4-ExbD0NGO%b0eR)UG+
z&n$u=4d98e+F=0fCQpJ#LSJ(td00@D_NpOqZ9c=nKyo61TXx-v!j<Br&ZTz!iQr0=
z9j9<XH=Ia*6E?4u+KYMm{UZf!8bWT0-X}9hAGVuMxULf0Aw0t^xkh*fh%eam8j#qB
z^m(hd8}F%AS?}9?PW+<S`|bXUd<QV+Hl-R~9?h<ofGnYQoSGkXAcfy?BE0hP_>`fJ
z1Qd#Bo)%M-nFFIIZo9)aJ-~aJ-~ebublf3y#$_W4<Vwz4NR!Qs#gWt%m$6y71froI
zvN?%BE%-vbgtD(18REwGB;>QXxgA9pE?gY3c`=%dw=n5NeNd>B-BtCV#Gg7I$L#K^
zhcoDL++aLEo{5c8Nc0KZL+{CUMX@EVJgFP9KbRJjXEJiyq7%VIuij4f9a~cE7g@kL
zi(@n_S@UgaE)Tt#ax*Q9mXd~bPbkLMjfTq;EL_-~w&FyQM}WymTUqt)zMPY}msjP~
z;La5Jagr1xYqFo(?I^m|`ex%`0r$5qI|720#S&rz;~LG#V9{V3ve{+6qlJI6jrm4e
zMK))W^L;P@SGLGEqBdu8VGoq7{zP1_WLHu`fFf^i$|fjnwpwGJA^fgN8OAL!9@{Iu
z*Jb9D<`acszkkJQYxSYWdy21?-%0J+6W2BtA{2f(zoOBQYNX$~a&O2a+CT=St<~zp
zpe1!x*6MA(-NdmV19#dEpEdmR0oeCPaubS!KykRRok?mWd;5Z+8oN$R{0FWO<q|`T
zhuia#OYpVP;UOtV?XjFZk}F!84txB>#EpFSpZ}Hr?1?y)&eg@FFDpF!WOS|j_>eu7
zU_dh}rhX9R;~qyfOkN%Kbbz36dTg*fcp*Rd20oKt!YShNq4+9$HmP7kOD(ow(EGWh
zEemj#+?#-b><kK6h#AkvbY%5;#a{4TeDu*@*^9mzXwq=arRZi)^)dCo*?PH8?2C`7
z#f?_@K+Z^vo>MsM^$ur~qol9G9o6Km2n77!;p#)rpD%6rkrPGkz!q@S2PvsOaoBg@
zEM2_tZ*3r#yAmV=hpBE{fvbuzW~9@|UPCd4&W&(MjMK0K2R|wxHInNikHCqblcUj{
z=*}P=Ia?>Ups}3Pz4ctO=Mu{I5hH*8L@0W|G&oubQR5B;sX->P=~l8CTuQ3dPH;@O
zz3K1J|6>z=E0K=iF#74Vw|z*0;D{Bocf4(#0-fyL3V;eVL*a`A{oZBI<iMZ|Nbl#Y
z>uT^+_@DyGD0hm2=g7qw&!mKvN_>D^N*@Ij9CRhw0b_^~f1C%l6OR;mqE+B8QbEB3
zsh?l6Pkr0SYD-m^{OX~m`c}$|e4aBc76#8G`b9pY4=f+zdF;z-H{_({#16s+vV7&p
z(zwIEPN@9}GfE@Z?l*bxk>A@evWcVWv~S~(N~z0bRn9|jY+RFl_MHO?Bt8SD@qJ9j
zeQR{oj@u7ai~SsgAFCE<r$7|cA@)-oV`c06xmwNv>HozknGHFgqsv?h;e(1=94cG~
zE)cpNu8bBkP)nl5ByCs4GsUt=0j=3yo$WZ3o=l+MuBqC&(QOR*{@S774L6hL1CQeC
zhRU5b-ZSkUyMCz9?=4Wb8+;O^ic!+-8~qn`cjcXMQ*;Y{z)c7XPT0*u&gS`%E*^X=
zN0{jH=`G#^T+3mY;ai8CT`YD&+KHQWrQ~fxp=QJd#BQ%T9S}nUVZ;*0`c47|vK85s
zm<Xr=TjF}4sX6gAOYdK8cT_EuE5N<Urp4e+PGz&gD1|6JWYeRy9AXvotR(0&s>L#O
z+Mx%xnb|9)wbf2`?^!->)20%fHD@STkrl<M?aufBB`ZuLHhQ*xZani7RT6XByqKDE
zcyb+->5=#%yQ>=V6-17rWU{++sY?hIwfO;IX)=!5Jwxg1NFFy;RM~C|h6*($x?@q|
zKtns%V2druj<zVb&eFt)Ki+T4avdr^R%^?19Uz{VDJ95WkvrDXU1uw!for@ZaShq3
zoQiU=c!cYkYX1)?%?N_I%<lv)+S*)&qP1+@P_QT5*z!JT6;WKD{mG}0m{U|x8v;>^
zJ09~Imh-}8LdM1ldddooLZ6L&(3eT{WS!OK8fK%Wsm~uP3GCaHgDHtTJ`(NZa`*-Q
zh^6jjL+}Z;E%{|5W$v22|JlF)-8^fJRS9DK)}FsWoFVHN3aJH^S`#LZ2J6ahIta7e
z8jX8l>R+~P-YMS7Gi*o9LmZ|sc(U~LBSRtfP_v*&k2}4S@t)0G9JVW<Au(4hfC2_~
z=f;Jph_qPtSbnueC3qgs!>F`n6^u((B_#&8CwyvcwZQYq3f|<MFGqHtJylH?d8Zt%
zH1qArC64chKs_DbdKk8wp`OY9V8!Hrjb_=Njs7xkE8bkZk-On@*&6XqDyLI;1MRJP
zRY0vM)1W;+6k^sNTV=iJR+V=yuoq%F|D6lwzuSvW6T7i`^yI!Q;!fG;%U$`0Rcq{k
zca9BF6;jwlc>LZRP&7U8qc5JtCCXwB+u<SC^~=C5e^QFU0i2T17xd8aSMl8=i6ooI
zEYK}ywxc=u_Rj5fjpVi8T)egW6JNRA@r@&ojtx{ZIZ&h<)jt}%Y$h6mAX6L4!-k_s
zRpw<l8iKdQ5%w(n!HC0wD6p1J8_jJklD}>#uTgWv#^TdrQeX@@VV;X;%0dDIi6LOX
z&ga_j74WPL;|}xLbLZMbfXD9=H}SYR$1C%=Q_iI%Q>#i;-ka5=rIUpT2PxjFW+SQP
z5Fwl+_I5R!eMOe1wv@dSyDx{D6ZUSNq?VdGgqB8oFFsMy9bvIzPv1|NH?|RLgfnTU
zec&L1&wx687^ng-XdfjG{h__fKF-dN6y>3!*!WKZo_;&|O(Ys~*glN|Sj;Eo04bbR
z!6kzFDf=vXa}LFCrO)G;V_acg?TZ*eY62aPqPM=d3qkwK1mgqvlQh~_eqpSB*uJjl
zYvt<(6;+t0$9{a{8`gyn{I{`pcmBq{%Z;V@Pfd|^zmJ!{n`=MB%(OzW)k>lKqXS3j
zb4i1hkXnT9`RBZ`(4ruvnw<=yS-g?NCtYzeHAYHM#{I)RedWox1e@SZT?W0XYSv$_
z_z|w|t4{{cv4(Sf&B>shggbD6NOu)qV5h>cSQJ=x-O0G#Yg$`jkFL+34>u@t;0^Hw
zKo_Ot7*kH~q;oulR~56#X>!i!rpsm|Hz3vgzWHQ2(MeE&8_>w`=2D;yW0t?gm!Iq`
z<lxq-ms5=Cp?J6Xx{BMR-JXlYci*N|YmzKk)O9vBW~kdyL5}h*hUG$HxZ`Bp!XPmE
zNH=Xys|HkcX=hTVpA24d!xWowvhbNQgqMseGfx)oqh+hjsup*uD2bE^SsFH}rx?)Y
zWH0~m@9a)rsMtn%7C|rP#<<S=mWo6+@1*PL`Y=Qsu)Cs#QpAfZR0P4@-sb`ZuFjA5
zU|u=(N?mwQEMC-B03QJ?ez+Gz-y%IHY*Dppr9CYALtApP(qM9b$g^I4j7v`zUV0rG
zlx05ng}tEU@?6}~Hd_&$d46#nN0K~L1Bm$JN)M;FD$(mcWoffD)%;H(Gel;rc*fc1
z{wO@Ht@9aBsl?XD>p~RR)zf5D*>Ey$KeREb5euyn<=Pm#lWCbsOem<QYy34J`JnG>
zTj7N`FSWiP7_@q;&zWITi=3Ud$>G<;{R&hXs}`>oB;#M(QcX@aLB)g(TC<OWOQjr`
z>#a4BlfxwI?xf>Qd*by2#6@*Dq}u9q++n1mXG9d{@20@5K3{SiZMSXJ$cXbm4WClm
zQDK5=Ao~#p78n1?9*NdUJtY|PukB1~(a^eMy8_F1<pY^4<*B?oz5;rcvF<US^TVZX
z@#EDbqy=?V_Y;>5H3Z;A9G~=wlimJ=J(Z&rk7D4uC*h;mAsybhocir)|7^VH-X->o
zL$#@-9RxnlCY0M7Nq39mSPbjQHxaraj(c;c&Te$`aHr4v<Ow=LA@o8uc~7S-&-)kS
znNpj{NTrD6o<!m1Zn${c=S{|Y<g9~lzVsspPC6r`s|w7A9g^Z-`eyf+)$>=aHwk}7
zgD5(Lzb`hlxwFj<Co;^QHP?<f#aX@9jyeu9j-289bJbgwe40d|eomGzUSxxLVMw$f
zuQJ^a`L~oIp{|4tS8GGMkBu3ro}Iw}3PXapHd;NCV(uVp|5+Q0J*sYYQ6HG^gLW=w
zL`Lw3nBn<c<Vqq&a8zrx@v1*1LntYJY$Cs^QSe!pUGl{P72I{+jKK<j1RYUwb3yQk
zz2ysHXas3{JD0~ckpCf(zi3>_QPg3B$9Xr=XgvnYybr|g5`-iJg}oo2O64i*oV>!s
zm-N{OIYcRL^kJ?|$rW_d+eg)xO0`iY8CT=S)vCRkCVo)%NkyN;w`BS}X%jti>(lBR
zueU4jz|Z0t)?T3JLc-@c4|z>dkDbCV(aQNJS*l0vOUK*JZZ8@3Re~I6MyZg#p6s~T
zb5tn)n_MnA)9dWpSX?hs2s#deess5vA-{Z=xLQjr^Y@cu!wreqec679XX0ffBuixZ
zF*Z+LU1#m5Xd&;rp+eTj?B`q;F)V&SY&bMN(uGkI>WblHMM^FTc*{($apiDuRifs~
z#$Z<shkmN~r8yzGI-dCr;Y`dm(WjI+$AvPSe1CIov{o(xMP-Y}zHYd%PjzG=R*>xV
z!-fBnN^`s2P>uKskP!0q+l|A8uek7ExwdZd)nL6i!T(R$&ECGG98xkS(cD|SJy+CL
zyESK4oUfd?@GqRiZPhxYs&?vs`>=Boa=jR^DcJ&Iq|G}CHczek6Dur37QS-`(zxPB
zYtKL?x8z=%md}cdgeMuD9?$w{QSco}WJWwIv*3L1W2BkG&IA@cb+}$TTjoAYvvOa_
zGa|lVPRy)$xCvh*5Eq|^JEQj^VQ>!~gVC5f9DHs~dpmg@J8hmX9Y;L5#NCw>I!|sf
zmcAeH;O^nV(zN4oIN!I7B8UV7-4j6TL*5>=1^H_BoZs1^Sc5b{x-(^Oz05dhOLCZT
z&sQT?+NF*iiX$R2%9d3Cm&>qNM5PgspUiS!!YY!6`W3PwnuePL8WH(`DVlbrPmu(q
zt*Gi(`9d1;$XpY1r&qKfYomjr-VKnuxBEJuC?5WBN}F4YQ_%WqX2%NBE3JqRw;_T1
z2U)bC=o@1gX)n*%gSj6i%0?kOqgI<>TMYu&`vjXhshL*+R~Z-^E1W3h%Ve2ikB_=X
zBsa;H9Jr>Q@;RKX=A0wa%WgF-3QJM+)|f;w)3erH+^+3@a8}>HezA3US19#W#|gLd
zE`P}tidD{Mi*^aZxh<!@WWoKm!*?&OWRLhtRxQ5IcE*53s0iw?D=AJXVj$<--4$#~
zX;*@u`s}gckmG8>&4)HA#r)$w8D*!oC-NY)OZxA$Cv#87kHbAb6(Dolcj2!fD+*5X
zz1ou?Tj|x;?dbrSXq6X}mreKpT|MKAS$y9Td$yWpx!~D?A$!hW-@p1^+nZoVo>1Id
zKi9)SY*2mnd`zxa8)7f`GHR9j|Hb%5w-i1WAi8MiX*jAL2QYT$16{Vy0o>8xX2=71
zS_t-O*k--eYIW0y5QqB&SF%#O_r<*72qaB%D;=`K@l25*CxMFC;3F~39<H5!J6fTg
z@^QFxi(m02t%vo;)Sb2Qq(Qox2W`NoSIShs`YRiB5dF=R`8JeruW8+0tIWW|i3?n<
zT^LNr66%w{lR{$kXf^di$UP`8aY7iYmRpDeY)(|BW$EJC&s7uiGPn|_{r*4iz0o($
z6>>X)iyqGz+$%;AgPTnxbUTR$=(I~!cdB>it1T)!dow@f>fDUMugl)@ttFdax4k`F
zgx{v^_D(JW25H#YBzrgUcQqAe$?x%Aj3NXaa2AnC>+*R&zSh1<u|+8k`yj^w?LLez
zU51xQOWd<kVSZH2{Io2$V$5_3IP)LJQ6`?X!9MYcuDRbnP1yfQ9*>{pujMuD)aN;2
z?5&MR0bj%}$jXX0K1(|Tl9b9)Ak&C_=~E>Bu*SZs7AHqv-ECxJEV{w39U<6#QjzHz
z|0Y&j)=pCi>fh!>+B>i!ewV)}7Dg_K+^l+>3p(uwf7wau(%J_5u>zg^gGCIE*-yC+
zlF~Lb*W1t4Iwq5-P&`mGgl@lPb}G3c`F!XIowX}Y72f30M=29=<*CBgEQM#6!R4w`
zg_$4^rf!<6<6~|xt|e#juBo2MxuKVffNC_S3a<Rs-`jPk3ST7m511tHOi2^~%?<e>
z^YW;Hm+i(JZjqv=Z}MM}QS0VYg$KOwG)N~8_gnn!{R>NVuUmZ$YseL1w|UQ8bm`$+
zZ@0&2;+~X<QHIzl-UD1GxdCkIsnCJQ8@H(>PGbk|$XS;B(#5D{(_&r(MIok4Kjplr
zmM*<F)d|=e1!m7Um7Xb7J|+yYnD%EzhaOg1f>dpm!)xQOY|g1v_fh{!TY%%zDd(XM
z;gKc|`DJ&gg@u6l!&%PF4oNjD6O_un&C6Ls`Y@z3_c6QcRAGHqlSJcg-<6u$TD<+|
zM_UQ1sy(A?;GSwtrREbiKo%ic5VO&ZEwNsa3NCGUlD0)T8)T16Xt2ch2LFWZwlv<0
zW5DmdsBxB^DtsI+#c1aMk}dZSxVKwvg>O5XiHL)cw$hg|`?tTbRsJ1rFEt*;x+Xd=
zgp<kXwpbg_g#JP^l{i?JAms?T_bJyD{#PF0-)HON8>c?fbPm~uK)?USwEr6iX`}Di
zlEwC*Llft40xN3sQ%IJANC%OE>T{lyP~s$zPc(PAy}P*uofQYI@l?nyq*Wt(JalY}
zV_D%(NhF$sG~>MA>hKYwT6?~jJ?r<@5g3hSfV;jcc2Rt^9MQB`H&Ji(Q%)QV)9JKr
zF=_FPy5uQ!XFH-R(1NZWi54P^Y<n^M@ARGM#%F$4ZqnAa=597{xA#BZ!+n3W)gG(*
zS7Ku^2fg;V!|y--%ASa~s96esC->?nW9mAtsY!8>;i()7BD)o@i#_pms68+K6Ibff
z)ie3UtB+UgnS2J~dPbrFdp7q$`PHy_aeqD++l*XxVGneFZ|=fcU2D(BwEWuif~-aL
zX)jbK#-+!}Tl&S^S)yska>MrZ<dcOel?mT(`+m_ca;EID$g9WgKu()8SYn%CWqK0?
z^{|}Ks6I!5lD6Cq#|PqshS_dG^T;otS<Xc;uSb0s9@=XC(MFXm#Vn#IWeNmr*}$nZ
z@BOpzuHYV=2OV^rB%>P9i`ArcM=P01pAE+|cB5$b2VreS;u+e6;vI7wk#^Ld$(N06
z%q|R%`2xGz9s*Y9s);(kBswHm9kug$6c&+Zb=+6>*HdgFk6&{QY2dhrF8QlPf3P=m
zcRNYMVQ(ebA@9fA`H}MnP4<qX^y-+>yWYowKiYfMe4I~)RSDp}AItNypCzY^eUN*P
z-be}3F|FlDa4#5{I_#rU6Eoz^_vHA$J7K?loS2MLBBD+c_NGr_bi@UYm5%m)`?Q)X
zx)<oSGE~bEYRo>1)}A$?Fvj+Iwk~qGv(~N=`6BPiCfpYyyuQ=EjP_1x3U>>skj?a~
zoC~*;ZW5(K|2m#2rc<SsDB@w?_<nRYwCHC2w!(9+1yp<ARdeL_;10+8&%V!R<)+h7
zKVd&iKI8b4Dny=5@yzu)E=c^8?TSE7lrrwWKj)as;2Du<Y{dB`ul%U1;J95eQkl+_
zd-uwbP(xsRhvy>;*GOFAu7Z`YdG$!}QA?FE>KgA<b)j|Tl?)8m#y~FLN}@e$*Nr%9
z;r;=-nibt9d&>3E+66qxJBRIt5!XR=>YBW9q}=wunfqJ2$;ZY$KzGm0@ujOU3OMYR
zToCCM(5~G&;v7fm7A#ThrV|Zsew%m5GdV)to*%c6th&*rjAY=SeCM2g>&zc6hg2?$
zm>q1YPwx+Zq=a9Bret%UmcQ^4Ako%H=>4Sb=j6`F$wS<Vb2b;%%m7lcFf*V#j!O$O
zD?!T5=h4QL@2dcZO&=Um&Z;}JSMF@te1_1Q&Gk*K!$JT!ajBRWiw15o(B;G+=n87P
zJ62uLox@gD@@7H`bHg`^LpDDrBxL7kkTKNmiGKT#ZBN^RT>e7Nu`Q}5_8P36n57Y0
z63?8D>P{T7r6a*SG_|516$iB}N5XXYGIKbMyZ7=06nCU{2Ga=tN{_N4=5PoTWv{LD
zy;-z)m927=!Uei%v#qK6WK32>Rb*S6a45mAt+sBYh!WVH+ibn}GyC`C9L;4>hON~$
z27U=STzjb9sdD1lgWf6Wyqchh@Kn{nvK;8EBUYb>4mJSoRF`d<e9%_B4jV^;`>x;K
zse(d99@`Q_yW+P%FQn5*`4IE(Ky+`-VcAU;k`$`6$C`foJL|}<pdPXwD=a}a<l<q+
z#&!j`rA39x674HTREp2q);uw7b;@qD&Br7#Z@KNLJ|g~@=Hgj>j~>b8kT$Q5s5DZj
zZ?~O39kRvQuJ}?ytG@BX%*qI~J9f7RH=dX6u~?N1WJ9GI4+V{L7m1t@_B@dU%mXMz
zHlEBTR3;Wddn(tkS|RSXr-D_<lUuy0o{ne2)Dg&t=+-liP)(|ZY|kb_a^jp#!vZ2L
zvptuai4fE&7y7-?S7C~1r21LUS7R1uRCbBt#yQQ_2D}z)?ZwG9#dd*flJ$&C{BH`l
z;yfV&yE?ik^lEM2ND*9@blL%*Xam7ab=I4`l&zK-V4w9Rj69T3+ONYo>rJG%5zvk#
z+BY|JKV(O9Z&@zJ5Ku)ToX&5*x07O*(FUr~#Jwt(_K*!`PbANPzls*p1dgx*)QmRl
zQ{VY38*zNXT-|7+)py(ZH8wc|bcZ75I2Q;#?OfGkA@)U+dcJxlgS^Dg{Wcywl%OHb
z#4)GwtYN}OVm2gYuuIWShm}pmEl9?9JU{Hss#hiEkRpH3-pXCOcT=hK^LBjgjy+Pm
zXX?Ut676ccJ0Otn`l5e5-QIK9Z|Wc%Q6@-xKU;2Uu0h24pz4x1EGDiT0VSxxKCGrX
zdQpxf9+fYp;ztR*<w_XvaZKF9l-z6MlN>}QhllOck%>hm4&&d?x&2!$?X#GciyD_{
z^X>B-l){F1D}50SW$WHw4)obL)?dbqrIQ$8waT6GRl)(pQ|7YjqOz}}buoc}c4yH8
ze3PW<zE$>ZehKX4<AC(TW(!W=#g|U5gpMQjeG<yw+Yf$}maCkX{a8(&i>k$b$|IxP
z5=s*I^GIl6pzEC|p&&<GzINIfAqR$-BPIHLg?svn)4_|?%eU69Jnih7dtSO{-BsD-
zUefNi+SRASCMCOdxlNRk$o*^pvt4t#uxg7-+?ri`x-eC8$LO@{P6s<e$_iaf*Pjl~
z?~ko82~fowVtDjwML93>d*kVHC9bJ?sKIWEbx2nf*~TFR;*f9lamX8p-|a2YjZTkD
zP6{2o)u%Pyvw(tHcH1ui;ywnVXNT?f_|`pgdU5LAZ&RxNWWrLf$8759;2B7M)!r^@
zb%&1tVE4llOgmjj;X~av-52oaqtiCnjMIg!T)g@Lo9U02EWh_Yn-v29NAy8brMzr&
zaw2F{6;-~UMcf(B4#7m70{f052h5FOl`rFz&C6*O&j#tg?mAs@(^NDpO7**|uFKV0
z^fU7lJ`bSvvwQqk%?=py23wHtmbBsOwkYv|g|mrE0+hn&5*I3h1*`p%Ee%W&0b(2A
zIZqXWeWZZ!h%L)6A!5qmplVF+v$#Cr0U?zZtT>rF=u}&AW#SEHxHdF{wkoGo+a_P|
zH31TCXxx?7=IsB}(D0D0OGNqguYYapV;l?>%Fcj{cg!|az$x(p?8R~0m|OVhRC}=c
zMwN?XNqE_6y;F$l=0>Zp8pt2Hu>ZCxC)iojjc%$jCqbT0+Y;@a=a6-C$eJB{7Ob(>
z3gwmVPF$2?9i^pd_f9zV=-)e{lj6$FtfVsN@=j(<wXM;KA8Xdh>2i9uEhh^SNc*Y!
zSUcjGSRr{Gpv$*Ms!2*SQ_>FD&Ul6-z|NL`ZP)2=DI-r{8&1is^~7N|+wSa=#GE32
zc+T2m`MLP=0>~kIJWwV@;F2;j_-({+d*XDegqx8pKxY{^pL9Ufm7MHPRV`F%2=P!Y
z_W1rORKTA0zc;D+!ZZG5SDVtmKI;hf_kW&d&sD1(k$XHu;Gpdda6?l`{}31^7eo?3
zL~{{Vd6o}jnSL=q9`{cKO3LxN$H&2U$@V2dL~BX3mmQcqF;&h(nF$WXx#&$)Zj{$j
zUo}gKB2qb>w8P$cTXU-&$y3sRWVxA1AFbMJ50Ui1?EZM>qsU-)#s+d$Kt=5AU@WB&
zKE|OC4%?+{)<ctTpj$#q=}B=LwPA-A1z|alBZ>WW_?W>|RsAa=6dAXYLbV{JZt12h
zjh*vlba&zj<@>eq&%fSa;{o<wq=geU;d7_T3W=rNebFxYxQkVY!QQOEiIi4|j!4s_
zyYyQ*aNENa231)?d)r6DSMEN0r-Ii=VZUJb>|KA)b=z+5`B<tx&}r`nD7c!$6%wB<
z!lmo|gVPhs{z6=hb3687PV#|2*hdM>`i2I=2kqmShiqj^409Se$c2JW93M!R(LqWv
zvW0(ItxqC0nk)H91%ze5?YGZ!<JcIS1HZ`2r4}z#+^$hsCHjzx!JQ8K?5hgunczh_
z@t0u^okV<{y)-v%q9PFKFTU~Jp7G}m_HFK|Y)@QP(#3ptx+Ihs?}23hFa2lV$JE5X
zo)$tNZ3?TUal{yIKg8-K>K(HmeOmZ++D{dz#E*+GzG^@F_SZMmQXJPt3!g)cDg?7D
zM$0X~YVm5ja`bY}OJhkLFaOr7Mw8Vd2Lh)`1r)9x4K)|p={xP3(csNGNb+58*G5;G
z_K2pfBUcXWbyZhh$V@hx>%A*{xJYtmH$)43T)2t}az~QAvHGgSm=y=6n|v9lWI&~x
zvjLhy<vBQpg5{Rc2`@PGz+$^K-xF?zDH@(^zuh+K>MNHJe7(KU1tqMWbY8#RS4PZA
zaV42jbtLymsZdGj70)DJ0dZo%_S+q!g{PXo!KMW)%0<V1Q1aTh)25FG1B?+K{x@>K
z&4}I{Y9&>j%^VH1`nP0=fA4df<@5UAr`ep*;BZ$HwarC;r^6GkQ#8tRW70+8U+_*^
zSi-1no+C^{SDoDzGsS_MtH<TLyAwVjiA+#(1DqevT!s{s4%t1U!Ro0-uT+t+Zi#<f
zvV<L?ag>%t9P=Wdn$+DeasER-O%1j*v5@l4Ew(JNunVO_oh`2*PuhDhsl*5rrNfFG
zUAbB|V-KRzvv-Tx1}<`y|BxlJ&er&f=isEfHdhDJBG*jP0@}Lli`=4E*yLigKIid3
z6LC1Up@NfFqu2U3#xvI$DgoE9JvbWT3apK)n#f8iSv*X^RvX_Gjj>?MH}wfLtt&Rl
z^>v%PbqVuol_>+hztO?f-he+eX2&@DxZF4AB4lbKW`Z-^8f}Rr<j!c%2Dm9I^D&ZI
zN3{g$_6lEWUGYo@<7Dt{Tk{zT$RQ4Z>Yr|lXG-p^Y;p2>+u;bQ;<U|T#2$&I!GfW`
zh37_*S`LNnj5i_plmsIRrW~<d(J$8fq70S!aktZflAHXooC|!Q@<3vJv&S6_^=(b|
zM08x72ACEZYhuVJ<HM|13h!%ARh^YM&7uk0Q%$<mUMV&I{7!q?Pu^c`uxEV0`|n#}
z&qi~&Qutg2!)BP_a}_cuxwkYCQVYxfq%fSe*9VvJtfiraOx<kR^P@#bMYlx74T|sB
z3wb<NZI{cI1L6URr#;o&aT%a{YU<cNe@{u|dOJ`}%qjS!%C<MY6m;bHC6>EyG$hQ0
z7=>H#53|GGS6y5Cb~{piaaJ}$%$rB;Xgtd$Nfp?i6bU5U!bLTZ-=Xe!4f-;uP(be&
zre}u@#d|MqQK+vC2iRo9ki)AYJ|(9dDIh#*BiUz|%3elWKAOi>Wwqe(vI$b)XdJV1
z4#?GumoHwp#?I$qAQiAmRB{-`<0C0xl#W!c6%#&JTo!P54|o<!Ksl#mCFB}9WN*eB
z=e}K6$iL|M-b#2-wYr}3`gV+=cy$ReXHbs6<6DmVX>q+rPd2%6<pdP5d7Nl-doMnZ
z^oJ0Z{r0{iWn;H}kR8=2Y#u%Dhdz5HEwhh&`I<4i{qf`wu9M_oA~JImep0PP#o(0P
zm;(<iPS~eDxTsMHIuf(=S*{P4U<j0b?vvTtWnbiuc&(T(^J92tyl3&n*=}EDlZ9>a
zu>3mvg4e@@Xy4@TQYnG-2{E&xVF52nSY^2jz6-F3X^<5iM}7O=XYyZuZ9gQw{rS)K
zqu&aMX;eA*DF?XMo$Y_lqgK}jziMX+cSNmeU0GM0N&X3SHXA2ozw%5dl<HZM+y}2Z
zQ&=b&S#=*>eI}R#FBK&I5{=e1XX4V{q~!dacI{-(ptzT$yDpx&U$LxY)aUW<@TMBW
zrR9L#aK?GPq7z_7a+?lTE8Y}SgcCo}cDDVdGZRbyg2I8_93x$xR)LMdf#a4lg+nEX
z9YEZArf{fpX3w$PVrAtH(X1c{w>zY9T$FR|l!Pk!jS|B-)n7->#XHUfom!GsqG{2k
z*uPQmLX6?`c;;qp=WAx1Da<4Z{~By&0uPB=MlBUa@J*c+Z;JC6Z%SZviT|GCt5~&o
zvE7-I*i2Ygq1$u4^9RKK&im)^7ZYEK3E!2&B2f~Kd6_=<?t}qYL289+Nv9EX9HRGB
zy<#<|zhu7qZGpoYlT_0tTa@6Vd^NdUDGgnLLdHcojJ#}1eYA!0W^S-$XF|;Z|JvG#
zDn-1*^7yQH4@uZ26lO(ElIn&nw$gt=QI5e4t%~-&<4Q)NYw(n<Ipf?AmP%<VN<hSA
z8T@SRncyJtgjPf#T<f}O<$n9i2YtOi8}F&DX?>`pyS<BIs~e)7QZI0^z>@E`jg#LN
z34j=9d+@ST^E7Ip+T1yKPl=V31M;04?}0_y6g?=<xDt)*N5|2a&|PuwYTJ?!SctC4
znxnZy4iD!$X025hvU8~(5W2H?hEoT|MF=^^x5K+_0)vPfbi%<Ez)jO?TYalNwWP`&
z=l^V5Ac2dN#a~D`itxOnLJx()NM@9TEHd>-j4oW3YL_(qwlilAqb>)_A(*UP)y#29
zESKhMwmWvvC3_K~lOiY}H2<Q%e$2lS3qw!`cmr8~JTXw=FrtG`<Tk?;Y=XH)$@And
zW?P&&zz?YrpQ>hdN`_c*>9!|Y%TZH$j^Zp^T0fnDQ+4i6d*)0L+5UQ?JsaI*)huQD
zdM>-cD^oE~r^FkSNV&c71}?S;tf%oOF4WVWkA7JC#ocidU+}F11-P|dj8pLP#a$Qw
z$$Fx>{M$~^g_82_tA>?oEnP9ia~<&3g#jdua&b`YjUJqrC~<Vw7d>2_pS4gsI~@BX
z3Rn2tqGAvRl$-^2q?(Uc09Tf`qt%{PF$uA!u9EJLZ>kVz!qNufnLCeSr=Xv~TsbP+
z5gW=)MmMHC=41`~=Afh~j&CD*kQS1raI~73iuvmH>bJ3Yrlq~g=|hCFbJ?yK*L7<g
zv-AFp(}D*KzSa_p#uE{02~)t0-Vb%j!GFpB-mc5`%~-qk=Qw+T5|e>Y`GW}5DYUn8
zIGA;h{W)lF`#K9>+-mQ{UNFg`5m9uIn-SiPZ?T_>V@0-m+1~SM-G|G+>Y%?LP26Wt
zxu9RO50W3?bWpI|K8!ZS2E(+9GkqSZhHbL4)!WB8&jJ?5g-;UC$|)%d_i}zl?9=Ma
zW!244kQ}tnqMy<O>kzs9JX%+3t8xWNy!#@l-_|xr^_PCw)-JTKa>~$zQq2_x;A`J)
z+~k?=H@WrAI4pgeKjzMwW#2j45EGW3Y4&~KmDK*YhxzWo3{XiZF%D*U)PBlQg%m~B
zCO4_KzJCp|wKg^(<q<Bsuy13*s-dP(9R|B{tneqLE}dO9=FFX-fX)`mLXL*9VAHs7
zByC+2Z3sDl%XpdH9k6T1oCR|YfOI4Ki;QyIuJhI^$qlPXkOlbyudhZ?E@e?;>Q|2y
zCX=s*pvpIn#eG@y9BOf>;FdJ2@HTeSSg7d!Pt?|3cJr8fjm?{7w^Xx}w~w3zkJzo=
zx7aE;^V{+l#mYC^?fwh1XeShVig$#C+NMTF3QyyJaA6&^J7SrLU`YRRY?GN&K|9kb
zuy(ZJ`axoQpCP`IC1v4^>dTkpLIui=#-%iKEd71(f82^I9=;<st6D|Bv~b0TXHE_Q
zlTBO#9`?})!$;Bo>x!G}&qy9t(}Cil1k24!V5vNNm)$iMd}D2g)DzHxyQ|%E1BH_H
zx-2|DVb!Ufzh^At#LpB~!_($S9)(_ldBw}c7C1toZ~BLBTa=UTz>Ka~$0h!jB5pW8
zEsbsNg(okuw+R)Xs$;fnY+}`4kV~IbEZTBE5-8+rY(*^J!$U4tt*xxKLARmbgRA_@
zf9S>}yQTu10uWvD@O&z39g$XH^IqpLWYdT7rCaaoM@@&OcSB-DQFw(d2uA5u+n7LH
z%8oslC-4v3FnUnc+NaR1l4kY(H^pZ0C5!1Ef~Je-c@4{5V-7*?D6SVm-*yG!ZHYG?
z!t8E{E*v%Irjg-`81n{eO*oX~vW^<<iB(%k1CAjydQ<%gxj4F;g;269QL+wNiQ!dj
z-d3OWcuzM8@qIL*s;u7iwj*D*HrXRN3JFo6v^xV~bkB&lQ)H#P1It%u*mfmum)xgf
zxppT8FXkxPW7S>^()w{9Xf<NCJrPUfXi66jRqM4U<C*B$DUQ2>cc1c&so5;%ZBI0C
z20{BbXirxiDNl*4cKhs^c-D(gue6J*r0{IEgF#iq1v)W%E~Y0X^lP>^;jns@J)f<)
zDG|`5PPG>t#~P_J`C?+o=9=c3ZQHHKe`C@k?gF>)ebod?!Lbrya-e#4MA<Kj$3<^q
zOCY3z0*YnL`VuGHk8isjuC_u%10k_60TnZ0N2(dC-n5_Rqt&xBDC%H3ka_$4nd<E*
zw|!AS4fx7Bx3}7$uOy1fkoi96!3K0INziaUAOldT_&E}TIHSrP#&H^rXW(Vw{P)4(
zj8(Xxb8;?Pcm!opx!U=$kjki8g!iJ@jq$|C<qZ!5*NK2vI#<cw>CTdusu2rxxg<vH
z%{<iiVvT*P`V<@~mzzX(X$;=3hNbX?TWa1(ET*b`hrJtpPbLf5Z|@}-O1zOctUQpg
zbIO^mm_Sjme)}L6j_RGHZy$~oNuj;gK8kML2|z6Har-!Wk(!QdW#wOI@;;edpkfXS
zMt&O4BsCK0N1>8k_F2++o)yNNez?SwJ-5%Rc|!5^M&qCrfXNyQt&!x5QgT#2RbN(T
zWFOo1RW*wuY)dO|Uq_Epf}Nx<OwX<*uy1^=^Y5^4bFD=SO^gHN;Ja$Am8CsM<@XL;
zEDx{=3VZq?ck)kgP(N04kl%(@@T&ckb3h5><=hm0B(d)2#Gs`p4{;GY8@F+NTOE-V
zAoLa0lLk^hiXdKj)_Ehj`V81rXD7tX{fVF~yV~D~yK2~K*PJbE&<aY0UwhUWxx&Ih
zUC0Sy*PRWf2^~AM^!l@*HWSd;Kz9?OdZyiQHh4rh#;!&tdE?o_Bi3{%A<#|v9cB^f
zGC(c18i~$sjzK5+Wz=8QMEd283^<P2t=SI#d=!8w+0AY7Oj}<x2pojl{aF!ja5*YY
zCsSf{*G=QXT-n*D=DeV8;wn!UL?f)x?x^}MmaamQr=1P9i|Wk{9VM#(e*0%DwiC;%
zlo^Uw_3YQEENq5DP(E|=t)ChFa!?T6SO=2qtlW{7ZZx-Z5|X5+uX_mK-RWCY*WPV&
zW3WPNL?B@?ihE{ZEHO@cQ}}(mE0&g`u!58Q7P;GZX1r(ioI7p)*>aSZwb?z6BZRS`
z6pL9h0{SSoST3p{GEicn$@p$d0?XI|OruMyugKWR?Bo)-%zH)5L@O4p$vlm;zoP2z
zIIX3KeZ^K*Eu=hJm5{(Qh?t12wx$A96Zibue2rUQhmQHW_<BFatpm2+SGJ|wHU!kX
zu$pKsiMWs%V803nu(6tnoA;@EXb)!R*cGHTvs&NBOgm7Y2cy2QsGF*{_1(w|jnSiQ
z6gVuXSdMKuTV$e@ZKN@hD!bWd+t9p;NNa0#IH4P)*0r~U8%7a;mZYQVO%-Nj{XA@4
z@l3&`5wc=itBsxY8{6hTl=-X0cAU*yX*)U76^&(&`0!N!#o}!{69hON$o|j2wkw{A
zJ3U^wjia`^dgg@zUS-=KbEKOyd#*j6NQV_cE@}!e*KNArp78Zi6#hTild&I#_rK5w
z^QtiaV>&+N&U?zIR_ekv*q)pk8c=bGV)CC(2o)!2R97)BZjD77!2vAvWze4W$<6%L
zOnc6^w5rlzd*dsSUkPQC*-@X*87fgMD=7H`ido)`k1yozceX?B?8OQyUd2=O*m`n`
z4`cq~6z|K;S-D`f9f%bc1-SLbXwo21Ce6K`3hX2<j#*#K(=+e7yN0|&UbVvsRj~g^
zOu~H=WqTU1qdo~Gsx37)`g0OXWnQdiT{cTvbsLOll2(=SP}%r~;+eC#cu_O(a6FUS
zu>wX!=|&RC&|i_-L&C~vehNX9+o*W^F<(29wb<cvIUp9;Gtm0ikay1Ks#=!YxbNTW
zdB3xXyyT#zgv%g6T=IrgoO{UL3?O*~B8ViyRR(-3y8{Q_4p94AF#C=x`MT0%IBk5#
z0pyPK?<TbB5R~4lmQq3$O4aiBlSn|gOZd(QzBY=ZC{E|Y7)<(PF)&q}O8NVUedPVO
zk@t1GeH{JEkSMdV%y6H?GhIwNBH{%;jc0EA_weM;g6ucc+2;w}O_G_u2%ubMAuene
zb*>XV(OWQG`znXWfoPbsupe&s>-bt!Qdz7?bEM!NTn4|1)_@URgS)4G>m#!^n0Naw
zVfVK%pWi0}YXw@o2!0VLe(;YLHsZ9_Wj{K}et>kM)c~WA5Kx}W;pcd!<0@!y1kBE*
zW${?K3J)LyU2(2(9#Y-3+m+`kAEPQ8Cbq_N&c`V8hO>JrA);)=SD!2Q;Wk{ls;=dF
zP<(vs+H*liS5pU2{%|<TM>8dV+{M08ir22Mx-TvQc(6(E6ydrdI|Zb0cl<kGH=YYm
zvPOAa-gK@oWL>q*c5^f?pF)6DrTG?rO@0%*wHn3Ujg_CuG5$7h^e`vz_H$`4MM`RD
zwJFtf&!B^Gq@S7-K=Zua?l@Oz`|u5d>6@0Xmnnp6dUm*=0c}yM&4><Nli|K`GjkAJ
z4dp|J)Mm4)*}BJzLSlTjbF#;-hZ~wPi{BYv7kme9P$!$~+tS=(^UgWHb@}BX8AuqR
z3PpFF3y!roR@vQEXP%v1fv@uuL1xdMZTBRC)U{DR&-2c9+5%s*C=HNabS@a~!UZc1
zeMzh)j)r(;Hs}->xwINkv2tu02WeS6lcC`ZtCp+j@<b(k8yY$u7A;)ii(J0aR{HRz
z%Mn{uEvZd)Sk_d%4Tyz7(RtR^=BBOgYG`j{pVvhr4vQF4lrY1#-U0oqlHTXQG}};h
zGz<?>?iJgZ+oX_Vo|VK^aQg%HVAYx&5xjYctD!cY$xTuMyRM7+a}(}%_}71Jo8rAn
z^$>2VOT`*<1UN7WI9!&22wUQfZ@hT?)ZI468?8nf+G5=*@<XeSM%q#MrWg%tl6+9g
z;p&KI`tmfWs05B()eJ<+-I}2p_S@EkbZxV0y=}{%^=&)oe@7x}CxvO-HrXT5w@eYZ
zg%)GO&UhvfptnHjsO_rS4dBY!Tk3l3&UV;;6lx*j@L1Ijtr28W8TKFd_05_^g}*0q
z{Dn#u*kDg)`@*34VR|Z_2`#jzP;%Q7M<xhoAxE;<oCT7~Rr<7J8-<Ihxogkl{vmHX
z8@=gE`TGrE>3q(KOFJ^gUWbpEdzq1*_cwE9&9fJBYHY<A!^oNb#dub%hCD-2WIZ|9
zf`zs(q5V)R-o|zyX6?e1-=5;c27#8Wu)S4xvd&2seaZTKB}EJ)S$j?bELId^KQiMH
zAMnCIaRiTgYqFV0F7Ef)FYf5Dfg}(FAlaZlUyV^@$Ui{D#xP^Uz8$WhM8F^M;flkd
zvTW2~X&=oAaZ#SMu^bTZ!x}qRp`mUGE$~Pn!uiRGqm6i(2%<3?kJe?IwG~2LQ2dll
zBp`}8%c;<%>X}@n6vlbd-b{dENsva-jc+3|VPh0NWN!s%R9E!v_+9`X=6hW|?^J!t
z6zk)>>)#OA0*7=il4r;fP<vhk^~!OM70PJye!_P4oH_PEbm?BfD$8)zKFnd@1!akq
zlJlcPjrP_CG!-A`XYfkmKC!9TC)MhdRjjbbe2l)QK=!148sFgaST4R-Wx(^Tt=V2<
zpT|at<=t;z<Vou!dq2GOmpSRRkb$rKtIa(B+W+kCYO`;uY0Jk+ZXNsW+r(}vYkuc{
zQ()rv_I<VUiqI4!<$|aLBtPU$)7dSf9yiU8i8HkdhyAHK%8LD$YD8hJo@xeX1p7H}
zl{RAfh$`edwbz~3k!lMH$iHK3xQgap@p@rANK)0>waKo0y|UwRW4QnSRj(IT3Lw;U
z+ttyL40f*CK3lpQv}@v1QtAM2F;yq*+SdyogE7&rdp)jC3Yc`ZVRyFcy<eTYF1x{}
zfu=}A<i_Yr>U5=0z>n5$dc9mJ4P@Z3o4wyf4b65-_EJwY$gR~67|z>XcP2qpU!KHr
zWc17uo>uzyYQmIA=P0PWz?9bu=eMZIrpDL05uw+j%^*>+ZX)O$vpakWP3>(qEjof*
zFDfXR0H#+fQ1<IyTm}!>jDVdxjYel8W;Qe4RPHETa77TbS-$F~W;o+HuLt9)gqF%N
zOzG;^gHJ2@p?NU3+8^=9QVe+Jw8J*fo0jl1;(G}+^sZqL$&eM(l@F`}`U&p0`M-n@
z2aH5nzD?sjn-tKr;C1)blVde1i$w|j1z2a6c<*RV>TIe1J>D~QrY(CtdC@kq(~$cr
z0mU&^fG+kxTj3o&@CU_RuJj+oTj6N6Dq0tr7_jR4SmQf^2F2D!Tac*1*7@uxi`=2S
zU+ewnuOe7@L*iqpRVt+07@%|F>Xs;>9xhO?PptThn-i@zAwgzh#pcxKd^b^%XuEC7
zE#dke7V6<F)L8BAh~z`I<fU@*$;h6FunAMOGTAl9?kcfdYrbsnCTNPKw&ymY46e71
z*F&WVUUid}n!W6_t_sp#Lk;HIw#M)d8);HYItwi2+g_h2MRg~$EO@Sk|H;3$BS#tE
z51kiXI%$vOv%;gOEcHIynG;4*?Xq2oXo>*aop8d%P;?LryFKQhg&}M%mfFW1MOrAW
z@`S^MtLcb6>3^@O>3qnZ^1n$aN_L0WC$49||I<JF&z_F;i(ABY&BZ$;`xlb#GtujR
z1xnAxgw$#T998+R=W@n>%k=m9Y-*a}mU3N0{g9g7e^Jp%oAjc8qZ%d{=6rzJzn?eT
z_Qe1SfQEfTp6<5;NgG@(t?zsNdt#BG3w=IjTYbG9uIA_2s+0i!NS=nyhRrglcG}Tu
z5RwG~*0UFPTYvTCK4_`)ngc#D(ze}egYn);n-b)`WJ3X0aLa}(___Yxz2HXh@*asd
zx(1IHErTE@et5JRQ?^vEE;^Ra#Qg^V&b{u8M==g^T%;9uiA3V4LieS2I-g&%H#$Ir
zkmd0_9W6E6NN?V36V)=#(1{lFs$I$roIU6F_U7vo*A;rexuslF$<e*t-tzO?Ou-6)
z|LtnA1jsN+h2MB$;&&V*Qwfo@cVjGcJxo$4bKKtZy_!CKroEr@Wj$KG*6@LkgTjV*
z&u05Drw>Uk@hu-!1HYDOfWT6T#UCfOkr|-TKJg`C4@DFFX%eH=UiS907`S*?aaeSx
zeeNGDUbX5EtL%#;8t_I<HMO8DH0#S4+AmGfESQLLU49k)sWP9)yz1J0?bGpUp!Q9)
z;9b$;j6n*%%~_PoTUs%)FXSRNH@DT=_tB!%5n;nsVfcqc<_EA4pzy)%DOw?Kz?&4V
z)PBmoi~0v%5zg!9{F1xXvx?dIiXlRaAzUDT==snt;IK(q`zy~+to)1L4&yy`RSfOy
zg0yTycJ+DZ;~;Y|uzt?OC>*nE&KLHGe3C>SR$nUBxmUBf?tI+4#j6?at6G?3M2#_O
zH)NYy)F7k`zwvx<7Y|UBv8}$-ZaVJ<MRmE)ZcaGVH*G~lbj$g|JGIo=t=0H0sWA(?
z?R;D^gfG;4L7GmxJsKW_KYFR$XGd*H4tjql@;9hC)mMhOt<~;`4>Dc2j?1+36`u#o
z<9RIw88MEq;WoV*L=NB5Q&J)@XH)p5z}}hX(@m*30OLI=`|p1CT^8wojMF<Sx^pL5
z*G$dvDbD#V{&;sfj7WAyPF$Ott&w~70p?T4YV#7z^+Y{zAKZ05bd`(vad&*6BP1<{
zyj<r8*d+uWvwOS+1GH|2Cn~O53jzk7wvCIS2nI^{;d!1GMXxxLh!-yL{7d2+vFe)I
z7#yxN0vP2LOlC{t8>%RQArci&+cKZY=8m>*viB~J9#FcKW2Hy{$>oAF31@6YHL851
zJx+9GE_ODt46CXkC`DVQb<oyKHu$}*t$L6%h=j+Zwl4P$(>`RUwxbDYaD6n?AX>EY
zBW{SpqRI5ryE579V(7DtxdTe$1<)UimW9qiW|cYAiEVAp5I(&RY)}rq`ow$E*}?X1
znw+E5)bONdtT6{50_`CZX>N%D@+?>7$gMdiUIcd^wDo)un|}A(S=L^?-Y;`Gn{dK9
zDm>K%TIBjzU2$4r=CaOoc_)|}ZEJ$G9k;S=zJ*?%)^;Q?b)9r{``!OYwW8u)$Dk)|
zXEtn?GoX@h!zAwVBj40*yM46T^L}rS<=XFW0X;kI@jTG%6ge*OX7{$JJMGDAi*;8{
zrB78Ot5O;EE9?w=;+aeahuFX|Fze|!5#l!w*)#F$5K+gc?b)P}x-9`u;a^K!;&V9@
z6otk4Y_9{R4xxxDIo^x%+IDQV7owNq1<pB~+_4v9K1#?VF-;1Mo_Hqvf%KVpo!h>A
z#^osnp4b69kf_trR*#U>8#`IHy$m7@l~iB4r6Z}T`y5DB-pRw!aS?y`N~Y=~=O?Ce
z2?4TFAr5Ta2Tn-8F8h`rU#a?ELj`;^7oe&0Octenr(l#ir@GGhkTYu=8i_wpQkucU
zU*sX_eaDJ{m~s_16mN7tNmHw&Qtmb~T)mMu7-}lCzpjrRiDe6&OZNJZBlfYFNe9=_
zoF*K!+h=fak7rKb!38eubMXvd6^AGs%ItjN6)F_GG;BQDs0yn}_~0^c6M;37V@}wm
zn5~Ef^JLGG!+kSP4^f@=R&*qZUTOhSh1uIVfpUbYi-u3oI}Xp<rp=q}-GpbuHu(Wb
zAM;*Cd#l>8+9=p1bhX^`?<c0#JWMrew9p^qf+A1ThdC)SZQDoZL$|WJj-=Z%ihi6^
z0w#nyLGLjn#+0N&`ZV`Oo6oM1Vi4jc1>tAjibNpy+vnK}jGSB&m0$XcoEgagn(9gD
z{$)0jV%Rs(zKRWY`=r?HjNUCAF0GXN<m;FO!<U0}x*QWUp!Q83#f4h?w=se)w^IlQ
zFVo_?YV%Ih)AtF0l^EoTF5-s(up^2Giz)@VEWN5{^J9!122F?r%ArC`emWn_eK6Q1
z$*KJubBNGq86P-ph2dFPV^>^=8@!%0BUfGsuBo9_@hBJ?FNE@quqD};C#&~deIb~#
zCTcdLvbg3#Xit#?23YAUVgWcfk)<W>q-!rE!=r0emwl?HwAxqgx|~2grmXFluCCAN
zwKg@_4Hr@ohZ}p7^3L9PA-FmS+DmrR1?PB+1`n7J!fiL_H^9WLVYgf;S5y-f+-|*)
z#x~LbZnxW_w>~J*8M{4t6JeAIisNllq6K6U(RpgDrdSO@MW=pBgW|TlJG^`1AWItk
zX$i7>5nPbfr~5?ONHH+O|0MsN&8!xoOaCC+g;AT8U*fN!#L9DGTS_5dKr8<ar(Ac&
z8@R_3hcZc%=3a0mWb2L^n-^nf$`TWmbM&r+MJ>^7ILJp6$vd!Bvi#ZncqZml=@=!_
z-IF^;23-s(q=CpCTgnMr6w@gRTNa~?%bpWrN!6DU6M9XfrM_4AK9+<BFeuBai56d5
z8OCgR_3SKiXRvGW4C9oDyskG}nG>0fL1&c%cq^r=*Z7~ae>=z4=HwMMP!GDT%Wvuk
zQ(B*>BW{hlMfJ8J_d&>npVY<}wj6AxLftcaFrJB<R}h6rYOTY*y%7<;KF6fkpr@FR
z55)iJvVhQAW8j$>*dg1Jr;zg>$_3i#?rY9XThIU_V6D+GNrxxel8vmr>hLA@y2Ix{
ziUHD+<9}>jxv&+b;ID0UI5c(bXy~wQ)wif*ZHKSr!d`}d#HS8FPa$H`WJ{I0GhqTr
z5<9B`b-NOh7bpk8|4-O%?{l1*eB1($#X-_$YJs}^wT6BCD;>+@)w&eJRAeJi2Gfu|
zaUnxVny?rsp)Oq7lerq0C1s*`Di+X3lD1LX6P@)!o%Vw(@<nzt&C@weqOpiLX2YM!
z%`E!L!XnwT@l4a)Cw&0^#m_~bT)wDvTzTWU<n~5uy*V5d{P_!!xh#|2DoT!0(7Bmp
zl&bHr7qidejp+tNQMM!26K?=f<$>-7;C(R~t16rrw8#!rYoKl~z&#8C^j6O#+xbcE
z%Ws24lF!QF3UcV<?Z}0=UrMa7hqR+^i|>js4?HAn^(T^-Y*_$uAR4*h&pFS*c-E_!
zeGnRPz=o=4B&&&u4d=ITHnRN?gJmO*g?O>l*l5D7v$1AVmyJbp4I<Y<dEB211S=|8
zb^%renkJ#zrQq|9F{&?EvdN4mjuh7?|H`2>k&|)35uvlsE>+Kt70;kC=;E6RD@<@(
z8!*t8c;mOKmWq7T8Ky4ug^B-AlrZh8=!`GucK712cP=<%?59}>0J>sh)T9;DAg<>y
z04JHDy_+55<)m~|Duel+6S_qWE%tsK+W(4B@<D=&_>$H-`!EknDR)K9=scN4Vf%6R
z3m+zXyTUy_apHgx7$g3t(Lj@QLHUS%7FeUKkz@9Gz#f7Kwt4~ww;AddJ!D^0Lx{T&
zpWkm^`qby&b(eh=uR|jWkvwi+=O-vSTw*_N-^8E?Inamf+h9k;V&MIR1wG0RY5<fY
z&c=Ng->|{>DSls#B$D91ML$%CP<dAdR{nU|9OSL3r1(Ew2=SvhSSvQees;*$bP{!C
z<H5hw6Yxto8=;Bg6YgxXaEV>%e{aW-XjhFFzDC)+Ydh@fap%H(+nq2<8FbXHi8ta-
z(k2c@v&+|J4|x2x*mdL1cR3L0K|D&~&JZC`k6%BYoRd@_5?5cf8^&WXUi?drL^npW
z(g0k9IkuakYgZF1Bwra@Nbm-nl%DVA@nVs{olOY_+HrOEd*YUy;D!GeB)BztDZUYf
z_RFj<4#<z)ZL!eee56|ew%gy_J{}D0>^XC7%6KraTt*1S^)@xXV@b+gGZ_nZhmS8?
z2n2E3c;N(LVv15~dVIz~l8+LuK7xw$FEhr2?|qmQpWHPwa~W%tRb<w9;VozXZl2A_
zhV`r$rTU%YSpiu=KW)uzq^ss`nj0e(k4Yw|TjFJe%VC=rZ;F(hG^-?-y(_!!BnLQL
z`rX;GsKA6DqXl?pZNB3a8JXJc9^bt2o>~>IQ<i?rps}H8!yhFCcu@dGIf75vlKhH9
z@rp>2JHgV#jLA$z%c3K{!ToMtp2+=&mA1nFuECMlRz_QC2zA5D-*VM>5lHTBv^BXe
zcnFW$+Ny28*ie@LvaQQaM%7)j8J1)Hc*-6go~e)rp@;(R$>I(=iq39`uB0v$2_>Sy
zHYVgCB+Ac38ZW*f2c13WJGq^bIkhpP{qW)^tv)^xhb`v~35bq~n_>XhWEQKvHC6*~
z@zS;a_10S)<LVUx3!*kV0M_1btw|-6YDXc|)}BBp{Bt>C)={;hv@H$}_15Ln6eEk;
zV{1;PgcEDcBeu<-5qC!YhlXM;wH?t;p&XKO#4JAIo3)}F8_mvGfFyv+)F+VGUAYbS
zb~KS!rN(yqipG2J+2<a9EN5N93rpalJ)XT4%Z0n=mBVaLRIU5DLq(w|LLJXY9mU<!
zheF^rd&-Y{Lo4<X+v5#Xj+Izjf+;y?PdgZ}o!2AAJd=}D`4hPvJez%@a8tl29JW2@
z>&20z#rEdsE0y&6`J5G(MsaC_F1`@0Ymg;g^uOgJZ#}vE<xQ>K+iYKaN>N2HSHy!7
zK^3hc*r3<H?NnOIJ|A^4OMnqP9K?`R_u#Pqt1wk$F38wPJCbu;q8&P#m_QmnpaKZ`
zb0*ZvL2Wt^Gx3X~*dJ`ro3z!!#lhkad5;D6PuXzxzY+(38}WI9V>rYo<ylFRdQRS0
ze5X<y19mRIYa#N-&c|NM5mI@$C;-iIk)mcizEVnyBQ}vc!dZ21pi_3qPv1gB{5SJ{
z`Nb8dEFfOS?5#NbMIb7kbD%M@7x#mFJcUZI%y0W5Nc7Td@8p0m@Mpt4?cI3h{Hc5=
zj@f(FGg%hO0kQYJAEcjC_CdVY2(I=#Vjoubm{Qodz^fnw<pWXPz>l)eqPB59?c;cs
z8g_UZ`y`$@ebL(b?NbL^TSsdHrKvu15Y3wNTl+kBWfRd)qaF5zqZ#hlkoYg7Cn-~l
z076e+`6}8ehG}2t$4b`(asMXT>j=t&mf#ut)}M{{l&DZIHUC|OWEg@dhZuVbUyG#h
zLtgrIa*Jx)TGQz{0e<vJqk^}eVx06(LWO2OCn}R~qm4v`ZDto;!GnqgF5!xc!9XGC
zLFFKNS6(c)zmobl*i{#ui_s=33WzMAcJ;;Noe1;O`N0}iyie?!i{&OSNg5QUbZvYs
z*-7a?l%DN6Z-dCHyj%P2djCz*7QfVQxEMSVal07P5vJ`%pAq3{v+uN<E;`?JS+*y7
zR=A&MfxS6rECppb-G}X#>RI9G3Prxv!LU%d-CFE6Z=v*-U8TMVe8lbEqM;cUropCU
zA0*3A=?2uxHq~2r%1pZ>W+uWU7Wz<8fxE~pP298uwcMbT+S;Z^pVH-IqTd<0eu!(A
zXD6O{vE1Zctv2goXxBPhsI6>s{Kd`<cIU<5(&}quXQ{Wju}Q)?iA6`Mi2Zr7|B5O)
zW_RVJI=7PvMHbe(a|{y1W7U}-bCi2`NyCD;a!)l!9Lw9fwkuU>_M@sP?J8HaC>zz3
zVgjO~ODY(a5K560Tk6vp@4*4~AzKzLQnERVs$+QtrBg6BMY~`tVi9Drh9O>AVWb-d
zCBuucIby5gTM-l~wb5nS8vmALT}mOiHhV91LW`2m);UhLG@+lDPIP@P#kCmH$!v%r
z<wGpgehtCcxsZ*ygko{x$e?a|@M2}}Cj2hc{<XRC7C>+HzL$&1niU<C0uCa-n5`TA
z?V5GAB>}yvq{!t4Y0jw>jF%LXxYU|!gV>So5kYpu+T$7JNQ%5n?FZ|)IN>tWg5}ng
z=aIXw885o6`TK8CD-rFnEn(4CD<hq@b%%f6BG0Ku;t=R^mA2uPLbHp1jUobUXLd&x
zMYsdo71PlMA>Sa8587@&?@AR|Z;w@LP@)(aFAmz{x!)J%?JEvT?l1~9c*3WGO1HzF
z{8!H7WjR9ZsR|ORkcEH>FSMtcju!)0!0OZSOgC+b%;w}j6Ei^6DN-U*x2Pjm*t1oy
zuc54zQR<LAm+-BxwY~XXJTX*a2%7kOt_LAcao&|C-(E<>DWaap0U&uXx`2De7$6?H
z*Lr*in7A6-7w?PM3epQO2OQbDyF4pa?m``90bz&x{Ob~Kbh!Fj!Tv=%Vn=e*aqpqh
zh(ZmH`X^X@D4&?KgbSv+t?S+fq94d|u@@<Y#b7)`N`W2~lZFi?y&*v>rE7-$=R@Fv
z!dOOfHh_y{4<-5WsN=_)rY_Fr*u^3rPI>fi>|EXxD<~#)KD#ZU5jLLvlq{zT%7RVu
zPwU#-?UI8B<yYf&QqSi!sfUA-Lz+r^%Ud8x4xzGrJI^kYCW)1Iw1A>*?_Qi({ulQD
zvG*>9l2H5v#`bpLUvKXx$w%v5{G>nd2hb=TOb)<@F=g`E68ZU2LT5EQ_HiOWa|4!>
zdix|+ETF@~6C8BPK203wD67<U@gC-G8?w*hO-3A^K=<dWi_c@oLxt9Skx;s@hp)bj
zF*xebXE*16mDEh3c*Rrh>x-e>xVIimlgE}TH@@+{+{4bkb?j4YlYMv5m3kPEp%8A9
zBhXm4bP4IcJRJUe2ZAJof3P2X+whRJA7e%0tmS$-DT4l#xZp}Iq^vXcb3RKBoC9bR
z1rJv(USd~FgsN#p&DQO9W&WyZA|>Kg6O%Uq<tp7idG$oO)+u_88iSOk6S3dgN?hFG
z+i+XEHv2(~ILEH@;r?#=bi01S`7x1DN$Hb8Qa4N#Rt~RUu;IpNCativb(40ioAOQ3
z8mk+r9iiWLbJexXv10NLFaQQxyJf-|L$WWjnJ5K_1G;r0RbsMJh~t+%-xG%@oAPb(
z4LVQAc2O4F?YXK>JX>u_H4|=4Pz1-FxK*b{->z#cdNAF(Y?j-D?(n|H$)?-F`6$=+
zwD=|r&zyPL^iIP>NR#lAP4}^diYQfQ<m^M0{hHJqMWN?fx0yKt5}<@iDSKu`4~H<L
z4cZ*v6Ut)Pov~B_6GtkD*BP5TQP{YL9<q58g-`A%=}0hJ-WAIZg=}$Ab%)oxqaRr*
zlwFYv(&ooARDzgpitmBl<2#KLD1Pw^Ch|c1t9P&nT9M+E3Ve~HUKZk95`BxS|0j5@
z!(7=*eOYtn%(i8*UTK&P+VX%{@zV#U<mN2c;l^4jD^3Jcx=^WCI=0%%SZ86BSdCYv
z#Kvo@93q&#@S-A6dX29~6-c()+G@c?&s<EF>n4I_S-8;F`^ShBDvmyC8xo5Emo!O5
zbSYNDjn(o-N)<W<xIgGUHOnWt*8eW~SnIu4R2-YT1=URxm84W^^LAQe^d`Tfvlnn0
zfEC@6ON4}!j3%5LzsZ^djI(}kt^OejT!bu$x%ixRUvbUD%Kgxh^Hlb&a#qh;S3C=%
zi+XL;w)*akQx{7$__ig~X8-25wj<g{U_j9vKN8P+VN2xBsC!|jueq%o-OjGu2Ii#n
zvAgqhQmeVy9*f<VgXUp-JjY&EQ-|;E6Zz><<qqn!Cw;kpAW_CsfhR7F%j%+FZyc2%
z)-Qq3d%UNaB?|S3J?+ED{Te>w8K2&xQ>SjQXZ-~I=lS+rP6e)g$oA&>QOJN37S4sT
z6g@vN@mKPM`Zw6o%Z~y%6YT|`blc_{doh<MufP`T@o6BYHrhUKRfi(zKmb(dT-BCj
z8S70Tb9&A?J+?l7hJ4u$<0}s{T<`OkUZz^nW|d$pnM(ka?jj|8IO01ybNYYz^P_fj
za?s)`Lw<GZp9ozzayKj<^^uC!eW03)?scV$=(E9G*~(^R#2>198?F5KhXcGXqdb4j
zMxuq&Ud6qS*l2*<37j0J<4&=$Xzi5t;CWmX?OalV`#U>o;34gNV(OX|OKd#8kjstI
z1QTOr6Y<O~*Tr05m#Sw9k~@U!581FLLS2C2P2t+O5WMAEh2mzw-u7`w7vI%w@8sDm
zl36iRGTI@O5bt_3F&k6{sxTSugG2CsPOs<=HN64*Ai9$IPLYq4gJnVrSNbqo>zbAA
zjeqT<3RsG*6v)z@>|-BTexVetu0DN@W~g`3&wQFxr=8@hRD|7bpCv3YL)9y_%;(-m
z9U6g7`y$So29cYqAVqjy<<{|KjN$uMDp}fB6Q!0)nV3|v{{yRNf!6_ODn>4PKU}Dr
z<ODYed_9q7gMVgC<to{avb$t5v2P|ON_>!JcTgaL66AdUN8f$xScz_0c>7%dRXR(d
zw)h#pG{4Ucl>HSSjvr#pY#p=iBiqwU_TwZ?k`n>4sgj$>>xY!yXexh-ZzMGL(z~+H
z{_N)-0%!I{DA9;Bq9*nmp_b_;*Ecn>^Cub=Qte#zhVy6K)Lawpr>W&0v8&^a_-(~R
zpHWDB@d3K#4OjVWvTNT64IBn&`Lb{b?YcMO@@wg~>*M{Q^D3I_*(Pqt_Y3jo$}tMk
ziz(|ZTTBe)8-+U@?`cJDxjBa^y8++B8NcO?FwEhr+svW8HQK;0LD)(fLc1-xFKTF>
zO}bUz{zh=ItD|ObNy;}R+PjiWVAvP9Gqvie@+S>o+)<7DN|6DT7iwD0txJkal5V{b
zd^J=+40Caunc*Kob=eoivZ1V)8J#J@Sy$iy7tkz+cFp}Z#~}yfz18k~BS_2Q=B6EF
ze@AgYH$GNISeb4ViaPHX0QMD$aXMmmd1s3{Y96+`t1m^=#VhsN{CMU;$!;XO=Z)Z*
zh~3}@#-5lD+G@!JjVWW%8_s3RwN%OvIV)byr!I+JlgT=c+`TlO`Rrsr8M9^8PLw@>
z)7?*RBk+XF9l6m>qcgS@Zxq31!{7g%t@Li-(%PHQ&6b)ptNisGGKa0HdaRr`%luB+
z+8AG6xng4npt<Yb2!3i)8wEDBHS2R+(i3*thN`z>nZ7JX+s53OdR0|LP1#`&Rtr3Z
zADuh#)>hACIPh~-?^{wUAAn7c|1~8~PouxMpxR(t0>W8|MF9Z4MLj5!%WKx`ZO412
z{2$hut0)Q@lKoqIZkR&)<?+-J^YBwHIr<gra;Vj{)Z5m0Eh$~De2QY(mN<nN+Y8k^
zt(`8qj~&_j3PQ~Uv`2h0ZKOZ4o%vZ@a3%3(adzL8Q(pZi+nwY|B|uTIIT)H+T1fQ$
zxc^X^Ib8Ge1iK?*Ia4rp+0~N!?Wx=em>Q)OLr=aZfXCrLf^aWIowe|eu%~@=LN`(G
zky|2(&_nRA&nDpA!peOs#~ZvU3e`Mj&*h<S$6+7K<K7CgXN%7B07BUF)iW-3fc7d(
z`3u#N5d|(vMkr;tq#)0~=!jR}RQQJ;UtqWV0QN<1I$%l|A=kqLiKeQ|XuaP00TPK?
zpD(7dyM@IZ_D@|RZT85VBfjk$|NPf>)Z0Q7p}zgz$!2Np2Xc{6vaW`0HkdoU2Des<
z_}Wlz3#xXUV5Dz@Ef~&OZEo)FvXPiKsU9!xV#P|cGMby!)S3a}V{v#97tSRjx1&5N
zl=8gyC@(9H<9IFy8v)s=OBthyXaQ#eCSD5c<0@uM5qOnCK#aCGb8gF4|6zr_l?!af
zr-zxpoo~AuxUAoCP?Y)u?^bi_z^p1w{ChEmOmDJ%B1Uksl}_&cY%9ApnHB5pgG8-0
z4WxnhZlR%F_EBysJc%+#Dwz7?>Y2*nVeLeC_#``6%65L5KmJ5?$Y(JPai@xtBA&7=
zxZgg{%^@?o2SQ3u@P%J6Ev@#Y|Fv_veN`Qll2HeZgfbE+8zv|Z75+MABtR;Pgg5q$
zgNh~NUT@#VEX1)L<SP||EIrtFzOR4mXtwWjVX!bFFUP>(AEKL=OHQ#L6E8?dso;Gg
z5`K!0R91Vmh-CxzvqO-)Gru)haTyG#vW~;mrDSiEw8Uiucb&ZQlCv)8{fmE?B!R0g
z6{Z940;H#_FU1u=TnW}L_UxKVg|CwB3$R^#DflWc(|Dc#tZQyl5)QlGC$U(ObvMK$
z+z;_!*;o&kyD=u=<x;%(<4u?1X4gHB;K0B)U-Da3H>9{pB}wtWfhabKTXL0R7=XWg
zviwT9wHg4o0^l!m{B4)w7H{q5cA>)W?eV$5Q&ea$=kk5+w<*2~R3w{hYIG&W+GBA>
z?2dRQIglV`XkxETyHvQtNB_#E$6HNR-fF5!G~-gaupeO9nU{Sm$4%9%n3X_9P$V7u
z9EZ3pk{pveeS7X-b3X@lZhTz;Df~v>ns+I9LR`-j&vn;iZ%vyv+1;010JuDuN510%
zspzePY<_|Vfs3+<jdqU@OWA9-VnI&2h=XDl`fZUzX${%RZOP=<(sS3_QvaHC^%#_v
z#n*=}Qq7A|<zBhm*Vu-G7PrQVXi?(#pioL>wKBU{Q?t!hd2<M#T%Ffsb9u>f5s^mF
z*5(wr_f+H1*5!i+rCNrlt@jS=8g!gC1ai4BTe>JHCWD47OdG2qBv2K!vj=^}K)KCo
z<2}e#sJZ;3>*IY3A$^;2Q^*xcbuVj-_EN8aM<O7$#j#>Dhow22cM{OW<Xe5c<2{SI
zL5TK5Na>-nWj*H=?R8gl3~&psW8${0xpbr*E;F3l1J|{V5-|@S;F!6?yPhf~#v{>{
zv(klx*-js>=tzojFyN+Bqjl&|h6PW)!nV8Wqf{qFZ+g%k^D!R4c59EvdvW<rpkK8o
zd}AJc^zSKH`=lSFO||w^BEan5{KocF#9#!4gz)4LD^R2bc8&8`1QRSotIN>xOwNPd
z?DLlPY#xl2o!i0i=b{Ohm9#9%pl^HgvqGBP()N5l!xV^kBKh53sGcdFP>F<&*^AkI
zO)XVot;bir2#tm9tD2Lz1#BAS1RRKGQly+@cxC18jeU}mXMa(N^QK6jeMzm#O|HB4
zP`o*O+1~2nQ{iA2x{t)0VnNq2kvX^YcGNNNH?wA2e@sKBZ6!EhJvNY+zyk^eSG>z$
zE&#xnGzhHs|D)|~V7xA~e9seN3~P)rt}(_KW3$AI49x)2H0{u?AR^L;T%c*kVQs6Z
zDyp#Fre1g(+c1o4T;p;XV~jDzlu$zPA{2-S2xzdHC?X;vA|e`Nj4{R-V~jDz?C1NO
zbADC!+<W(CZ?-4bOa1=O>v_&|p7Vb8WGhHEkr8Nf_9hXpUjZ{%EMWT`3M2-uw*xVV
zun-ZG$MHEk6z4irVeJ%I4icnq&x=rUG99i)P@#(A3}plH&7;RKBcK*IQY}{=CmtC0
zdO~oZuPAVt9cU=a-_dH+D$dQ0#kagxq${@L4z+22q?C2;w4sL3`oagD^k1jrFn6jt
z66K0@ZHfc_w1We_eVLuft^J{uSpKus@>G`xuP!<Al*2mbOTkULyIDnc#B`mnj#gpE
zCATMiwxXr`BySDY5c@O<L4!y1+h^4*at|#eK=v1a@OjQ6SD1LtzX%{GcfPJ$aaxaW
zpcqa@<-hh-Vgh}&ygK{ZS0SF1GKaaOJGd6#=0%0EZ{o-Aaz#buD*KX5^vWFmy#sW@
zxEt(;oM3TN$6`vmlX0tBMsEpzm(Rpc@l6!GkEN-S($6uxE4uaELOr+%2hHAa_M$ko
z0|X@ymAw+Ipor%h8}>$;)Am8>gem*&f;WOYs>4@Ed3@}`H-aCMXGV(+k4{oaBH(h0
z+w7t@Trowe0yMW54;fZh7)jPHdE-bpd!Rgu-=(IM;#>jkm%R}NSi}ScAyH8c8KcuK
zug2<ytUI<`@rLt$LVEG2%7o)I>&iC@6S4{x;i_1Ye1UOGk72x^Y)J!Y*Hp_;K{st@
zzg_!Auuxb>TEs(-@OA%5j-ewH5SWNj{#+L`D7I^hjq<6-p^+Hv|C~Uo#4&G#wu|@L
z#(IZ=o;p$-*|_|xr3q}GkQ2)RM$h9!ALNdj#Wu<RIk&6BCi`;ctHq7Er$Y7gPHnO&
zapH2n7dne{dm?Aq*8bIQdSlQKT<SuIQMlR1udnO2X%$SA*H4Z#RDI5Q{wh;j;jwJS
z8-pB|VvuES{f`{i%z%*i#{)LY*FJ;tIX2s;{4?oi=H&1LyLqzpNbIn=)i<4RA?r3T
zzCrvU@&twA)V?6TNpo1DxWy&Kw^Bm}>=I{uS2dnz+9XM@-JS5SMe4If{+AeMn2CGx
zG-^>d)i-q7y*a1|5O6_QvfB72R(&J*gXC{@3762D?kv)e-&NwY*`C}>8uBvy-6CrY
zWR`(LjJ$F{E0F4tEsgODXU3dzsA+b%-9B};wFYp7;0LTd8cSy^wpUC^M|@*2Y+s%w
zsw<|Gg5zIlM-mnOOTs)`mZ$-RLlXCVYAa%hU?61Fv{ls*dgF@Hkt0%~r2F3}Y<e3d
z8pwII*dE9WLuPyk+Un}wmYhk%yn~`e_qWCYvXEc~doZsI`V4!hqF|L(3JjEqzeK6o
z!#*PqM_bVzJn}}Uf7{TzwxB>Q8resE7jk<gsf-fc*4f%>XM3fQMZ*n;^+rf&Q*oDb
z-N|FV8mXo_>~ZfynOxGsU?)_PvnOJl;@Zo0dQq`HnKTs90+T&=^{HspM~)@sOGKk?
zPX{<4YCME8D9FTdfd0?qQNZXcl&x%{r<#aT?7^<DufEB((Sr(m6C|D8wG{EQjnT&Q
zTInious%nl=K6Zul6yutR)8~$-4;kJ9RMf^*CZlBbXs{r$XNv;*kRj!`2#&vxa=AF
zf7YL?a@RmvIc!&aQ#7sO&vx=o?5+^WaWS;pVD6qcc8ODNb8|W`Yl-sX?oAwN$8xsc
zu`FG74*1@F^{Zdmp<D<YqjY+T`+3z-jbNqc>~LHmW!Twa0||+)wq|$`1av!+bf8^<
zV5q@g_ue!3pB=46mqJG}-zGcezas|J*zv>yu^UQ6ZzmE6XaSW)K<~!M#DZUJl_R!O
zd64rwJ^$&PkA6xVwgGy7CVGno;rOUz{DDJftJa>?Q;8GK#Wy$RyH?_Sej6ZGiM|Js
zOFzlJ;G{BkNdNe0-sQTL!jaFSNie!-*k7>Eee(l7)8^S1Ll!30_+UkoFQcb>(`+Is
z*%tfCx4IaEKQdpdef>sJHQdr+-^4gd0#-ac#MtsI_1d@IroGO-b8H~YhDY}ML=8Of
z?1%hg!OC{~F@|EvGKR3L&72y_Ci^LV5&Nf<O*}0@(SG(V)e+lcSbZ-pR()Ptd2W&s
z-i94@%~6T?v<v(%!dIo@zVK-962w;)A}~A#;Gy8Q$}GO$E{bpDGbvF>(zjiFG*~bc
zy4^+1dPz>rL#x{(dVT59!s{%<1^+Uim(x?M@8$l(_}}8Tezb61WE8y5uJnHsnA&Am
z9SxR_EJ{eMSNrf-tm>BHnO<!8*Tj@O6y?POQg*F(yFOedMpR8oqE9%%LpJhgvYfk$
zE{nWig@VY!F3NLV0tJ^FY)$nx%2!(lr5f!kT*3cr%#gA1oZ`Z~U}OFFHf72gmpefI
z=b$G91Y{oEY!ed#H`t^Y$KSvbMrM=an~qxy%}zLk8}nN+t*bWA3pS<trYqz+wwrt+
zvXh~4v77y`nWcP}P0P7(ua#sYN;TbEuvOif8AqLS6=Su9T}ePgL4^TfrZ2f^NsG;j
zVKjiGT|zKzw*NeJ&TTfw|M?G7{@&)sa9-G<ut`3s%{y8dCUHzFX)8Ky!O_BuioNW#
zg+2_$sqC)YCliX9+sZlK?dvWlO&TkQ{6NqA`F2mVcH66j3UVdnsj7&JHd;HR`wpv1
zz~~qOq_Uqh91WG;tu5rJMtf8(LY$`z;IG)y><4=%6Zsw<jpm#lk68J>yo@m4nqwiS
zsup&8E=R%yT6yaTKo!pe7O({lvXBAm$|+QQ8g<0K@%k*w`&UG<qFg!}i~(+|JPneH
zR>hD(0LxP+?rxLa?_kCRLwcnLjuuMxzh7^w{j;1d>TFG1lV2M9inB2zy<iXKj;0b0
zhi>hmXrD@V#T69OEypF!>*2f(dZ)<99?d<_Skmtn(O`B6al<nE=+TTwcOAa_UT%iA
z3M#@(MM)P26I&ZciShyNdR>Kba_t!shHsC>FGK4a3wfWl$Ne}|=GS8tpKt(A0~Si-
zh9~fo334T~EP~%taWLZZ6flY{&z`QnDehTUdcQqW9c@)vicN?m?n#(X$+D@<)>kL(
zGQ~@1rF#?13d`VFHpXDV7t1Od&uO1SL0%xl+HZ-mHt}eRhwX=H+vax!Yf_tS5165G
zW0w9n4Iw@r3?(i3CagDdc|OSNw<|`I2!Ry{0e!3Ou2v$S)S>RRd;H>#9`$S6JGdo|
zZeyuXv46<T$F{1BU8tp^p<C=F7Lvy*@AQg#do{t2%Yq~5;iE3#g#$K>{}$-Dsw+sg
z$XuoavC6`>@{3Y_xu?sKgcspfJM8snSE2%dC@ToK9*vddqb$Yrv0T|d{U<w~J&O$E
z>*wIX#1qj|q@qx9dY*LH)V8g(Qvr0j=!ka6DbwF~*ZQ80rWl0IRJ)LxvRE8qb!T%I
zw<x6aoFkMXwG@GRKJkkH30T8Vs+G!{i2YOw=ucy%t}$^{-DkPSxp&&<@w<oS5+Z|h
z$QOY@REmI6{W8}<-CGjBM|)~37bOs3T=sPmo$=!*+c(hwJ9N=tlR479^$F|R7I)fr
zzW4gNdi&ltF}<Ej0U*i`K1@^H5;6w<n8Qew4|);CW<OPH-BZqR!(ac+ey)g9N#<5w
z(4;nGxiEV(NIhW$%0xKq&0vgFNf-BXyWmY%A4x;nYZty5I+NO(S{wdm;b745H`zsR
z1{J2-R-_2N_|4$S29z$ty&Nuy(V&c2ATEs-8<DIe=LvN#i*H3D04wCIa`~H~R4S-0
z`%fw-uXxj08Zj_^MFc8kcv8P&S5`fxa2#?;?5dc@zj>X>t8?=uM+r3Nnry_q3Vo$r
z>syd5NIIx$3q_7E!AK($NHam;>)s4E%sX%=#`lci8f8}>^=9D^JMpj>9cwJF1v|!K
z$8}YZ*qHc5?jCTxh;|!+l5)k5t@Z`EpwD3T;BkH^3ufLr&nEaJi0(Eq041iHpgrDF
zyhw^ug^RV5DzLj#861R7elvL8rq-pX8-d4-F|FU0p(4$cY^$)1#fovg>CM8&D%%4D
z+-`QDkgkqwGShsKZHwC$+w=-0_QC7T@Fh<e_gkBppk?1Yju1Gi<m7=M{V42MDOhGF
zxEngUZO)tJO<yqE=4N;JIR$A+z?|pv4)pxxdRyS#N?|}-m>a{4UZ9Uv%}McH6{HK5
z;+c?xBLB9#V}cT8suSZ~TNK|=pF;KZknFL0d`Z*LRQXmLn(CYDa4pX(qa>y9fnZZz
zb}!Qu_dsi?_MsAc&c-z6mZziAUOJ?cc+#DuY-;w6wu|3ujo-x$$;<R5<kxlw^DNvL
zIx3`J*+TsYOnR|ne>em<$*m+}uw^+D&me|ROn%SWiiD4H_OU$L2e4Q9Zh4kvgSx-k
zMq%bKmv6QQyajIEG90gVghA3NxVFZ(KpsJIS4HQ%2x~~)@lZCORzf=-P6T}Izi>;|
zf=Z9%{I@A?`BDE1_(;pQwpz6+fjDGbmR>gqt)lXjW%aQ@S17MNp3_o1jhE$#3M4zA
zxn8yY$-HYwFIz;`oTD8O!pgJcspunxBJvZ;g%|DVL<V`Nx*yjwgDat#i$XosF7`qN
z)>o5nLqK(%L9efM0ffdj=I=}^A5VFh_W3uUyrhad-Qw7A3$Cf#{GZof|M#}N>Rnk`
ztv}}9jBDCP+Z9mEEa38F@KoZv9p`IXZI2JQl(%JX9tbZqDGy?JE-_a{izA!619$4Q
z+h^at;2-T!qWS##7JIdNsX9>jciQ3TujF9cAP>E~fCD*0tLzg;{I4dsRvy;ZeIer~
zlAwKXs3v^!>{w#_5a&8i^>^G`65xa96+TZV^7JtTpbI%!fk?XAKCb&tJC#6z10X_*
z$8QsSmpn#JSHDQR>gr@&m_v3Z_aYS|M=b}3v(Z|#7E?E3mOFs-*}0q*)mbT=1d%x(
z-{1oE(CIm}PvTqg8|T{|Wc*L7b{pgkUhF%cy_o_jdbGkp!EJp0W`+#u<<#*=8C^oe
ziYBWlhXbd+NO&SfkWa<F^mA)b1{C|s5t5qhwf422K`prXjeoBBXC$a^y**jM2ow1(
z52L_hGdf9G<o2QZ`rdorx?raL5Zgd!D+3dacG{0|OWQkF+E4!PrEMe){yDD}*>w~O
zMm`220*05E%!ZCqwQJb1;Cg=T|Fa8@Imc0~c-*Y*g^iWw@4{o@!h|*s?`y8`3pPAj
zdr-Yz2WjT|(E(l*tpRj#*^(;m;%Mz0jzd;%($Cr@#|n4VMuj^vut!&gWf#c<&hxTk
zh2x@zK)YRj%o#Ju`Wx(u_)AGb4%(Gf-}SO2NS9()RZAh1v*6`EyZTsYgjlkx?WJ5}
z1N0FGa7~T_mTq=_!LE(v53R-Eo~R>o4KR1xY~-=>63?YtV2;9rfouUI*>9tc75?Y8
zHdIS7puaaBq3K@OV|=@~=+xWT7$jO1)6ux>S*5$*DvX50PRL&h61fdbGU+y(7~?4a
zlaiJxu60r_FpQQ80-Icopvy|~JPGK!%ywgJlS~7MmQ(!92)QLl?WSX)yQ^QpRkEAC
z<v`CI!nAE#%pzNy%C7YzJWM|pR}W{~Cj4cKHDN|}du<))GBX!ao>s3TFNSZw&5F|l
zk>u2)WY7#tW_A9Hsmf+M__`E@KPSheJorkR8%vQ+Mb2kD@ixylGSD-%jw;I0z@c4a
zGh0|eX{QjRaHx0XKqxStWeEs9cK5Lm1!Sy}svghn*n{Yj6hkk`X^D4ngWZ#ZmQ;UI
zaqqQz9XD!FJJ#mKY-%MxC?=h{06roG$J~(kTliVo%_M3M#+Nqw=&&H{ni8t=-ddVN
zqE}cV{-`;ZAngv_m0_Z_+LuxrOZeYn?Y^w(x6iYVJZEV{8muc?%T6R){mTG<SweyI
zVdRbycef%T+>FM0RW2Q3FRs2)>d;?yru(Z^Nu%wRzaH=dmBGSR=U{BZ*7y%-a2DHx
z0qcSz46GW~a;Z9G59Ko1j(j}C2tAyw3+js)$(jSnw?_if8>A7iM;&TSROWI|&9%ou
zIJk{`0VLjQwRH}k@soaQk0rJda`l`&?tjUx0~z9p*tjrNiXty83}IyBGNV431Cs+C
z3oi%sR6@J8rWL~Xbagh{K_-XRGd}kX<0e{9tb5a6SBphTFpWo|=uNUu@$)8@vN1*y
zcPl?KRT1sW>7`9vqGQ`qT{EfD!;%h{EmN+f*liVZYeR)^0EF@g+3sUa$HpWWp&sbZ
z#i1dEf6Bk~#M|x|XULsnHWW9uC#i(qFA_EsDzVoWPqwA{mRfid6+7CW8*Hy9moBNK
zvSVjAp5&O0#TFgP3$Uz>l;v#j)!Z^B&EouOhmS?NOze&X0)xTf@-g6WzGLcKJCe(V
zGua9#lo8#D@9VjW#WgIq(~jnWf#h-ydmc{hSgx$T6$D=F2HoS)P-y6J>j&&a4z{4X
zLm>z!17@<rZ?sc>yp3eaA_1qrIDdD&oyp55^PbSI)6Q0i?gnnD?i|t2#jLUt${Z#x
z`Fz3%Z&Wnkq(g{CP*xsy{dcQ<>XR3fl6~guf_)O7_<5o%4kUP|e38Ae*UAA%u4i9X
z_n^2HlzTcv<!oO$sHf5W>p{2_#Z^Cgl5g_R&}ou2fNSz?H3@7ed&3pOzDq#V)FOnF
zL%hzu_i^j-^!OoqxZ1NQ1?<Owli(m}a$Dj4q&xp9T1!c`D;|xX6N4cV_%yv0dYVE>
z^=cGl8}J<V))BcFK?qyzg0~8nGj;9)R1_D!6;1)poGDC|H4cBvnJg(gp5sa>-PJ{J
z#j}FdjT9AiJYyF}Gp$wJsl4efd8@ER6r7??mR;)e-X|Bg%ifB|nT*BQ9WH+>cs`8N
zQg7H5xq7kR9Fi!qzkF9#^T_BfP1XUs%I9HL89cA{!Ei-vbxp}Nxr~L1lfL$?;5Xq5
zh&r<oRp(v=)sZp0-=Q+GUFSVyn<^m@qjJ1*g-1sZdE+09igjah6cWBk1ed8|Y|OJ6
zrs$xJ^S=sVTWJ&C3IfjEgV2FM6Z1>xY&vaH!f1ZU1=VSjbCPmblqMMSTz0F)`@c~+
z|EFZD@CH%p^P3U`R7O=rshhJ6<U{o*#9L2`6L67PrjbmcG(B6xHAxXi_IR5S-*g$}
zy|Wc{=gjyP{&cKzlg-M$oGXfQ9zO!w94}yY7TTP|3+O1CvboV5-<`r$iFcm&R^e9v
zGqTEpSR*(mZxJzC@dho7j>@FSYl67dYj@??(}?k|v%9P1DJeh2ICy6lz2!W(lYNiR
z9h$oLp6F8&2hpe4^b+LD>+vH|U_@<BbVr+#eb?n$V3Qy?$VbmvgKuI!7r)W}dBeC#
zwlrF5C-N$jA*tDKsH&rPSZj4~_=cnLXwR8=ao}+~5=BVrMYcxNN3kE|hr-xMFDhV_
zz%5A9vh2_aY2?D#is&FyjrcG5Vo6@ozxVMXuSyajM_-ag`15!FTSdIM?GAe&b|C3g
zx;Zf~t9^sxCr+?6Z-oMuASUYVqOI<h+3>+Q$9Qdv-$r4i?V)NtRZ$mi`Jgk_!wwyo
zbtu&%)#-ShEMzBp)FC9}nXc8^yzIrDVx6z7gLm;UM?#>`XpiS6;9ljhqG$h!>S)M$
zuLvY~{wMR~DGueiwx03<yX$J~=^O;Ho)<vA5cW)bb7imG6AoEVe3KSO27Gb6>yyA>
z(`>QcJY>if9>p80fnO?$B&D$G8v-V&(7Z5%EGy2EZHYd9QOZMN+Y+(vn0LEvj|O@U
zr99aQ1?bO-3PbN;*cIP0w~&JTcUOH66|<LW-0aE0Td)W1%?(h93f0s8Xzz(aeSrt!
zT*!fj)N;sMASz&(i(Cn><{*Dkreix?JrLLyx%ztw&njkpAYg$DYmuGtVFdhz(a=Wh
z$Xg-iVS-mdz!E3>dR_#|E)_&M8jDQ(4C1C8%U2JR1iXmS{v3A{m`W!xbci%3V)~(;
zMp7%{Lj$?!vy(Y}2}#B0KtV63Vm1L;R}B|OJ00JUp^zFp5avt*LH=j(QGV~H&$M&d
zm@Q_0x${ZE6$q_sUzkGW-S{NdD6s>jXsMefaZgDiWCQy&d&*|ca@Ok&Mqxn7nOtbm
ztFTAJ#?Nz6SR0^zUnD5;2_@i#gq+c*k5^nYUy}M)ao#$l9rksu2Tonu)q@aw!AvT4
zh5UZwBMkJYa5;uraqr*ej?247=}vK26;f;|DiShU#s&C3=YxmeYCq%>aiZ#!9Q0%S
zEQdqfoT{N!5V$-m_H#Z`?JY*$3U_cSY74#R8*JF|FwFt17wm#)0noXF7>h8RU3k2(
zUHFE~w&BNv{UJX%>eGv&xpW#EP+U`($u9N*71_=UeMz>!_@$RYh<<4fqUWsBF7pmt
zntG&-%a8j@CJ9_RJ28;*DHI2{0Uh@h*_uOZu`7=|v*-l4Xd*!caV=NH#Nvc>kNOaf
zu8wa=t_8r`0qHfz3nSUYJy>biMg#P2#s8q6vl($bSe(|TF2a#f*p5729`Zlj`j2*9
z%s<3WA|8)X)le!H2So=RM*Cc7E8zjgI9Q7sdu%ig+hPe|8FxIddo$c1_Ao7Qf_IuB
zW5mShL@=zP7+_rY+9Y2ZciJY$&!wt6lu>Hs8+`%fKU&#pQ*u%S$IXQ&*l9P#H|4Cr
zSx{1$-JIXB{%_|*bP=Z=PYW+Xa2wCi^k_u3h+_4U@_9yf<c6w5<Q4>$nE^1!Q;A7a
z(jNEbn3ZdrU(;Z-qnXTSh*(^AX|%=O&B^n}=cC=`9uIyVsZ({Y@Y|W^ARu1X7F3Jc
zP&~>7f`#5+><p-QSK^PZD+ELAy8DLR9ew>KxFUbi;Nlh60Ab05EB7QyFGsmS-hf=>
z_}@`quQqo^B)HW%<XXE$bW{Yl;drq1q(xZ9m2Ny<c#<(=Y-s|ayu?L{2Nb>Ht$aj$
z)F3FW(M)TU*>TXzwmt5Wm|N-dsh!};0AQ(va~;(L9^)e)W51$ZzIzB)i)8mwPi$Gz
zf!ZauVh}ndMnr*MtKu7KE@d1*wBBO(J6!$*{cBs^X%7tPCg&Kcd)Vqc!or!glDOE`
zL^GLzsL%)E?6U_Gq)_KBd#E}I7usZx7Ofyv=in4jFzNiCw@0dP?w_-pX&+4th99K_
zXsfNwdoB%XeFLst>xQljH6(}gnD6J-)+G`$?QuV=(Z9DR{1oMkvqWCwPdXO1@r*nb
zvrDn5#}uC-dphyOvrRtZO=Pu{cBYKo6I+yOUZ!V+ui`bg*Vb3tkz1FT{cYBplW>_u
zZjiWbV~$E*!Y0^N5s<#5Dz_4%jIT*?1>KUcB~lyj@Cqv^Kn7iT0qM}OV*3!NE2EF1
zxQm!&{k|VGJtg_mu4?TvJ9&(;ILUps+acDDO)n?Kvbx0fCiIJ{h(av3KX!zy#pU0K
z5650Rkh=twLTE*_9m;XA*S=t{`d^SUj@}L@1nNizqOEd-12Kl|>dHKfR^E=}T;&Ss
zH6ycmeDepS@QrpfzUifxBNPEGcFezVcHLxJC&NB-9FP6`Bj!1Ah0N>iM2-pvqEs&m
zUOAcHipR3l+Z*guA_&+Ba-Mc{8vnaVb|$Gw*YbKhJLE)OWYhb&=YCe_62G&4#rc>e
zo>b(VefCLWc-smxu9g`6PotSMtJLg8gd#C#k|+!ms*wA6)#i}sWpM`hBHLhiSAv-@
zE7~K4CMm4IeLz{L#a7sauLhR`RZ`}^=k4q2n_L1thTxm{R=L2UvH!Me=YC6ako+#b
zdEF8Hce8z;sL@v2x!8Wl7QAYDQ;4InAH78*`Bt$q{uC{Wn+W@o3jXIj2m(+zS$n&%
z3ls=L`!nqAu%4CI{erjS!LIVD!XRJxcKQ}M7zln67yw(*m)Y>Q;~}lBuj^>T5#yq_
zok#P4obvkE#W9F;ITCSqmfI(3SYf!ARO_G;qg@&!6m>uQ+<{<rS$r#ktxlC*msj89
z7wXx2uJG+OU>xkQE2{=Xm~n)JWZ6|Q6`E5h%+)zDE_#RTnzxU52q4MVu1#Qwo|PAb
zmuiHM&VA>Pk#7g<7S)FE+k$SX&d_x}&T?f<8x?E7b<S(kij8_ZZ9~*HF1jrnlY_yC
zZw4xQ&-@;YjRA^X0MN_-a9n&-(p@|Yi_^`7{HD^6@jg!U?G2Fq6P4qnXi<WG8C_+4
zCI<jE{dF~@;Kn@6uD0%yi*L%?sVyX6q~30NyD*Noqd&hn21COGYvQ3>(;Pggwx~)m
zJsQ9_@{Wml+GsQ4n@EI;fytd_=G$q7ZdMqt&GJF+Ebi;G6H>%F7TZCm%?U`!7g6H9
zJl`Esb7OQ7h;1whqm|8z)-pv1u6l61^#b3^oeQVh!V36uJn+iKqWSynu4>3v6afVr
zx7F^hFi2Hv7FFBa02$a!q|_n1$G5PcO=`e<qk$*jmF8-*)h5&ln?GcA34vA&MH||z
zAy19+!_C%M^<Q6nDaAf5%}G&YJTFQB9Gml7G4@LHxzSqVo1RlqFG?<1d!BI-=@1(`
za{8uLPOjd%;<8HBqO@?^IX+p9Gd<TbU!D>RHlw$CJERL`zbP8{Rd0u?iQlQCo=L~<
zuhysXB3^O(fw)g?%W;KSoeM*OR+62rsm^Xw@i2CVTiG@q4EQOF;y!yQMwTK6N{#rs
z#~vQsMA1SN?*YuSnK<&1>d3b`cfj$Ag1}cL%Ha%-0NB*oVrv~$mSE&u=i4RqWzCY-
zA|O7Nr$vO>BldU&j^;>2^+aBw=C+j_=acy-dJM{OJ{4=130~1_k`SLxKr}RUmD*j;
zyj?izzaMKoNtKifg4pe%eOn)6l&bBFfwzP8<`_ySCFAqPY=phu%d2g+K3{sfS6JDS
zcu=Rv!aCcQV1-Ljh86rxZF?Sxq6p)rJO|jQjT}cJXXj0}tD4OV0AN|P-4)1n%UruH
zX7UBwlZ$O@z;l_!?v25uir8!WW3YUE#j77kFhc3!%MazBEj1nJs9ud8o?JzqM1<1i
zIpI(6q7CFJw<<D*I#nHZBs#fd=fc|Sj-Zs-xM89l&2<Fjk=y053LwtGS3;p3_eX#l
z8Fh9d_9InKznx4#<D`uy)=uTPSie-0VgnT7bbQN_JVfN%8ApR<b#^vp9pa@zKS`J8
z691HtzQlB$_mN~Ftx;J52?n1eEV!ag_Gxai2BXYpu~d@x!jya-%kbm|a=+STUu5$+
zq__RDTDwvJ=r!-Pul%}9oIK9HPVn$9is6>9<Gyj<1b9^I=i5BJPGzmL?{b2f+?Vh3
zA~S;6Ecf0-KhM8qkNY7OP;6ALXt5vTn*=#A8nR{nlt8~lK<}`hD-cT4?v@gN7|4fO
zluS|-D6m07*|7K~2uM)V(O+;PO-DG!m9m-_Mk^h&j8CYbZ1{;v;ncxva8WintmHpF
z;l=UI%_vd?_uC~W3Y#XjmW)m(3SWZSrNb`E@!(0|t1mxsWVC#^{wBf|TJ+!eKfB^Y
z;rXz|tH8^ZxiK!H(ztMfc2#byi|LEj=IRsiMq^1PJG)&Iy#$CIyg#_Y*tI8|9g?z@
zWCdWcG8#%sACYquzHc*}^2;_dNBje&`pEQlU9>3=C?m=hU{s7K0|?GXde6&o9I(*|
zgsE6TTWw6$x$vl5qZe)Li9x<qv6Mxh8ZwKLbdPhG4)pwfkxi%;s0?k=t-fLt<C|C&
z<*`8BVv`(nEQM{`WdDnpZ8;g-m>XJ54I+ZXr<^D>`^MkfP1QDBAt;TJ3!OJ7prxcJ
z`5LEHZONO#Jo1+-=5U+t<NalX&2ZE}or&N0%oD*%PAAP&sY+|JVvK@Z9D$rl{9&0L
zdy><U*dSSo=OieVET+@uR-lk0bCHm3UiD4>Z_h!Kdu@R)XTc0x7(gM0n%q5iMO!ED
z!jG5j?jikkj+l{bQBF(XDg^SLXd;%-8N>B<Z+7DrD-Kr1pxXE*<A9i4S-I*G1YR;q
zQBn=rwOAiK!$M;=Lv7IIL90sxM?^d1qAVOK_g()*Ol!_AN<=|v1lZoz7}<qtIlpbN
z_WXuEO{wH$Z0PVqQaB0q^5R6o7zkZSTko<29@lS+t#F*|mN)LI1kikBavaA8(Tg~+
z7JDG(bYE~#7E-gi+RmWPQJzm8C%4(wB+Add-5!j#+J_`nIe;rVUxu(l5dMd9K5Aph
zU+iH&@9BTEM|>2@qPN<kIpUpNw$}eW4YtSDRcI{3YjG8bYxP*Hdl1Zr?D0fch$}uc
z#TCr6Y&;RIQN0xQ`yeZO(&r(?RGmHL|AuYow5PKzpwwPeeE6B_+pcmlB)hPlymU)S
z=~N_+^)Y|p6FB<8Nbuh3c=%=;<1b2*^9=SSgqF0C+l;)HF^aMP6N`lSATLsdsciFA
zc6N2w+4gAV8qG+MU5pC--fEy{#7NtfYg|r+^iq9n_lYW<33N5DHbx-UyDu0zjDN9T
zsb%EYqbRz)KCURY?RTua6&vtMI}nZGz<5phFJ26X;u~lI*YEnNS93AYsJd>worh0^
z%wO_I!><h_vXbK-ACn^|ikNsSafq+yfIOpeAeHX@Xs(hsLT(JZQ2!lEqBfJq;J72(
ztoe2#KTqSFP6q6B`K39K>#-ZFPvwH->MU#XY410OYU`bLrkW4$cs9BdAL6t5O)6aQ
zUkTvP<%u>oHAq!*K6kZTJfeM)a3GSWrnriHnw!%z#~=rIS3ZkxkO7nu<$<)%t8e(j
zaUCeNVPE8Eb8DLAlST^698EekK=nLg%vaU^Ik2w-|40=9{(X~!mXH(C+k^J4?|_ud
zC_)LF|1P^Xko)xeL_M^Fl4bQ)ipM!<KP2kiG24DjP_|R?QK0-OD1}I3de-{w^;qPe
z6G|;@ig2-a!t<!Mo4kj^-U-H`xaT}?7rc{h9wl-dS%Is{fjnEx886y}-W#OYjnBjI
zcY?!0;exvY{?8Y^<JUmedi4|~)gJmeR#{O+4iDMIIjj8eImk=ian{7IjR&!AD*Dw+
z-$^r2WZSyzox-kZxrcZC+Ah!COjix2D`Ffe!d1zAJtyoe&Xw;Rk>&_7qRFm$$N5K*
zQ)S&*4{5&I$C=$!5({4Qi}eVYg;m1${o1Op;tkN_A{~v0Z+>FDgXld+<_=4RUD&4>
zaVR-wTxK2>5O>9y`#-3_h0b2TkM<42#Vp4fV`HMZG>FPtD#cmxwqYYOc8q;z&|cL=
z<QkDu<9vk9?)D<JPskRn5Cd1qPK+JM8&B?$ihZ8s5UH2<alK8hTI`f+6EBE8c4MA?
z9T}7vh!TAMWoeD7JStS%Zi?k<%u-~fK(drW(Y3If{ao<vB4c=kTk`&xo+mFg$TFO)
zXC#=pcoZOmHtn&Q?-VBPmN_=d2kLHy{LQX_EhI$yR{Xe?OJPnPENCd*lyj`w?%Wt%
z1VMo$FGC0C<&0cKam@HC+XA0wf$XCT^XSm;>0R{DI}g=TahSW_DKf%9&*X9A?CzM_
z56j&yT)bEm(<tZgW@pUpo`fw*1f}96Dlc21Dw6$W3o=u!e^XYXa<+nEoiA`^S690=
zI2JaMc~C-?HO3fH%1Lq}bl;ZxOp%sA{sYZE!hBV}X?>@>6w_<S5AM4`i_eYi@HUNd
zMeE9KHltNsmKTz016{TvCR8RZ*~$dzRn@n>NTkZkvDNO6Z|+xC{%a4!H(fskOh~V^
zI!Ev9CLhU~#ANx+myA*NU_d$4(6CD48$knmD5t<7M$;<m!^3faB9t4TO;tKUewWJs
zk-<|arOE2-(Rb2jzG87R+24!T<JugWVH5+ih4*V+wR+qNS-k|3$11#(x4UfL;zp$v
zcsxhQ!v$BLh;J_X0tNQ}WHqWvGz%ur+Eck@4pn4|2>Fm(pRQW(D;^38)_tbBki~Hp
z%$ARk-dIP*o*b>P-2Mxz$DXp@@7jVptk;oZ{BOqFMxR3A7?qz-UtWv25LGNHP?hic
zmc*B-knC-Fq6>HtZF`Ua?QFgEdqe0CUbMRsc%j^ps8bBmwmW_)_{<zTd2`&FwkH7K
z#V+;s?6ker21UVzT<Zl$UfKOgag?-!=h+VA+U6<6&7o>A`CDwle|4k1>MfR|2ZpOY
z?CY&>rk0=$M01%!6!Wb*?nfMiciv{N=dT0Lb6i@*^L8{=7R+NITjjm5W7V3gdX0fN
z$Ne5+f@`xA`R2@!dot`;;wrw^VqxleZS9nI80fjP(@w`CHi7JV2&Ftf6H_@b{gn}T
zi1f?WJWgmvFq-bLbJaFVDkLO>7wmjB!d~#(!;Z_7Zl6SJZ8rot`?Nxq4TV<kFRYbf
zV?3DTvz!5MiWeb{1SQ$$zK7+Q3hj%)7!DWl^UHURTv8(D5W#T6R6;M#@v9i41Tz6R
zxk7)PSDRAyIGRz+?3?PLZO8l4zKu>P93AAize@-g-EO;mpMUfGDYN$v-bdb8xWUzx
zh~FP$9HBI>2Jm*IKp6a19-F-=<$iYL;ow*?g?)yxkGiEc?4(-&RV{O;U2xJ_7bQrR
zg|Q!b>B5uF+Te6*!%v3&XPI4e(s`=lFAs((=m@~WQCxhouuwIep%iqNc%OD&>q}3@
zgN5^fT^4iEiG|CCb;vG1nSuq3o0!ild@OM%`t!;h;JV*mXII4lGOS9}+KM^r>gwAD
z31NjETyrv&SLDcI^Vgn?L!3)8uNtCj$OKqpBOD|vYHXy>sWMx7uC9wAB^{Dm1bKR=
zjdG|B^vqDc(a`}40l$|cP!dz8c4MM%AErl{+ZB&CHpfGSF8WMXR~r|nAPdcYn~=M_
z4Fkc%7~F4dF`JY{Y_~~+1C<mnSm|wYe2Z9yLi}&^187^^Y*S8#Z`zP-?Mht3Blg$T
z!2hNk(gm90wRv-VlbngI3C7?RoA!%uiajQeR2~)9V@!oJ66zFgQ^3VcU&W~F|HWqc
zZb{OsfP>jFmdnDfVxHrTCrtc}&2{v<wHsK>JDIj7m3)vF=mH;VNi%g2nr-3WPztJI
z`Rnbj`1TwKF6R}>iP+sI2boyKu7O#La_G+HdaAha<(}LRrRRt~xz}g96IBDAw6%^c
zq+71@@Ay=AvCD=)cA|V61J)&@7Zb~XCaUL7Fr<!308MR%bix+JLS=YF1y<ZITYYd`
zqp1#N?fw^z8Zgb);eUbT&$G?0YMcF>5~2dMc3FHAJyf)gLM>PP0-c;_6hAOql^5Dy
zc=XeHh2-vdf1JjI3HCrv!(g)E$VYT_bj*uf3hxwBz2;<cGuR4AFIY4g4_2#EN(HSH
z3tsN5<kx&SAy>pKJyHDy?+%uS?0`KIJI5sua`R|4o}OBHMQpUSd4p=Ym)N=(F5kV1
zfew20*pLCFQ``aP`*`k?zxrt>=I{weh<4~gN7D*>GV$XMB|o&M^3S?@g468j96+{N
z;3!YcXB?FWddT?Qlc$gFxE+LApU~HxV?rmVa>?$^UkWt5_?;_0H&%e(F9%|lDqj}G
zllE18;QoA@{H`JhTVh@AsOhZhwxkO5MVk+g_$=pArINAjG1^9$4Cx2;U&oC87!7RY
z&6XN<gYEKNDg&|Y&Xa_?NT`tio+yaoE~SOplTZSYc9h3!kL``qeXcks?XPxH<xq~f
z5ByhFT|8&YI>;WCI53Ym(N|+oo;g^}!}-%KbN^@q`Nw=*N{>{iwE@h2-46-8WjD~5
zqj_j5go!+QtXiBa;*=F%%*-A;9vjou+W|0DDfxs022P3iz3Ld`@>p(;or=*m<G_I@
z0sr+<TGVjKdqn0dVz^Lgkb>h(!k$XA{|e_KH|(=MMU#phob!F*Qn0zh&gVf03DBD;
zuB7m*{QE_S_a{!zsActO?n(M~4qKK2$>Y!4XVruoifU69k<X)bVc}^l4Z0M18|{l|
zEwofNZ>1~z(xEF;WXEFrs_HAaNnB9?rCzbGqp#=-k;Dah-$ZMKH#iLy8j-ucq^NIw
z)9sBbQFv8jDs7xtu{5%MpAg`&B8?uhFj+cE*sYky4faE#f#OhpjA2|aF6qimkC+i1
zP}3IL&v_^wgW^%8_HN;DsF%&-H0)jHrt+?l0mUwOH`HUZ>)|afd^aAV4jh3f56ZJM
zJUU_ImLD<>4qNS_cY}jzYK68?|K;6y#zUM_K0+65OURU662A{!{&U#x-Ez6ZF7-JE
zdTQ$%31`AXquDNtd3EAS;UXi=<vy<r=KR0?c15)zrSotibY*;d4mLt|GO69Kif_dw
zOROD&;nluz?!rNs!fO%?_&~JVweKETP}u7KmeKP6`*#~rt(ls=;0=lb8~JW(5lf-P
zqNe$enz8HNjn@lb3)IS<baYfMd<jIsM!y?OF#dB2uu#?k*ZGg}b|irk<32V4!uyY=
z2juIsaqos<xTOZy8JpluI1>^T)OW$oG~2}Jrh6cgy9aC6q-<A;5}}cYqn-Tjkr9I5
z|NZ}F2DTexM4nFY---B?_=d(Bt)K}1b9U3a!HCZ(Ofc?V(MWPB368J<?zL&T6#0xt
zK>YNC586bSK)((HJ-6IB*Jf7p$UNoT*{pZNSLpV_9Y_zp1YvykVAF!M(oM|C74R6l
z9eA6~&HIDEgkPsPuz3mA=|sN5Am$<B7~Ir}4jB=^7RFG*MS0<h1-Yx5Ql1L_Fy5Wt
zij$e+`=Z>^k|ub>+%1JMCfmIUG?FQoyIdQ1rdL3$iSov{$^-tKnO#1qi`!aiR7wpf
z7Lor-s>dDH8md)*0gSQ}xAaC|1<}l0-KFmq39TN+#G2nN5?TvR=&jk02U~m%Ne!(%
zp;HnpSw}R{BOvcvm}Kk94K8TA&zAXcH;lW%R^)JnLvcQRRl>hy>c;_WvE83TGMEfL
zXcO&$*rA*NBN1YGC!goP(vz)@Yb3N9Or}jD4_lLy5||}Van);kFu}aIjpKx5J`~-{
z&%O&#TmG{TSBsHCQqS;SdnCTeMuyp-aJY~9o^gFf!?`ww)7qrAQkg^N0M|uVKTl_2
z9&?BmUop<&@roiSQL}t3ceCt%fbc{vdnVx~6eO}It7Q||&H3!*k#QgEr}79?GW}^k
zg5Qt0*`BGE4go1Icc1mVTSWP(%WS<PBMD|z`KH&A0>w<DZTtlo{7m}Z4OvRQ2Tjr_
zZK+-rXSn-#XET=vr7dimk2rO~t<!CLZU)V7$+q2}Pfkg~rzu>27#Ue?S1duSgd!po
ztF^m=l{DMB81nsA?vp)vB}(+HfKQbUu>pP@w%3<3muJ)V=SJo)ZtJ2dd7T}|=Ta^W
zXe|y6IX3Kn#qjfLuBWg<MXPCtb5mY^9J6NJvN{UV+mUKbGMuG-S9s-*?(2R|^>sCN
zGzKR%CR~V~g<W<muS!#EV^b~B$~f6aGf!zK2IvJlkx0SLl?!(;u=-@Kn2fcapkard
zs=n!p3%`5qbbf=;QFuTvjs<5DM$kyS!gw;yIt1JCQnYi~W8h$Me^F4t`TPcaWFN(I
z(mtsW<awDp%Y!eYrwR@#wD8mD<Ih#-*K%fbnD-TD#?KvLJ66~iNsI`g1{_}|)-Q!&
z|0?@U#qYSK7E1ng0<ctkE*i*ha*=YwV3GZ(Cce!@9Hw#+NTVF+-zB1zw2#S?D{|rc
zXsvJ9MigLVKSXP({G{+MY9Dcp_G6B%3td3F+kVP56q|>*r>*vLJ`3}e2%N&%{|m1(
z97KIbgS_;IoeIWcCUHB>kOJk6x!{zu5@H{eMqY)(1=fRh;i*F2|48_r4L?<w29m;H
zi6VT_hlZh!oWvJ<Ta5Hwc8UL+@?|x4>8ao#DGG(E!Y=dQW;ZpE0^*djsZOkTUvT5O
z;#8q3Q>NIJ*|Vd*SYWU66%6z&A)|bkT^-8?fJMLI>TbEmT@$|)3k1MYsu9PDYfq&X
zYD0Mz34Db{N<UKUjW#0TQ!jT=1(|XDBTu=aQfi`S?YdZVv4%oUdbwev;+wN!;%4^S
z=&GF}*L6DkY)pQWc8rUxP=c{l-)9TWmt*?4Q=zR?{K|yvRBn~fg(O-I3Jg&cAg{Mc
zvEfp=h_jOb?dF%sr-IhjHk0RGPu7hwT;k#m9*8M^Ciwd-x0`&ob@eLqVK<)&E{SvZ
z%ycLwF=!6L9^@@gRCz$2wdr|i+;-u~j6AgY)b6RbnYo#gZVKZJQq-(GE+n;`tbVY9
z=<EtMie~ooM02VU^lG_p*4+38ly#uoo@Z@dE`(@Xgl1b1ZB(S*?GoE;VUCaDX|JAd
zc%HkmoqYMY#Y)n2x5H$hr?IQ0*%swrfWOo3$-mmQzkB`UaJ}U@uZ^MH|K8vGI&V-*
z&X_K1$OgDo6rVfrp)sK0$|n(cg#h7UU`xF<8flo1&W7^%HOC0Zy5))53hU9DE28*f
zi?t^%O`Sd4I{Y+9%8XiJiFM_52+4LI#5E67?q$n{9CJw`A+rY%!Cx%qEmS9r7pron
zX>f!0`yc24`iNZ|vdZ0HrufL!)o#Q+lgyX(tVv`+B84nJm`$9qDB)obWfLAVQXoHE
zjV%8bUFB`|NPH_k41T@s(NiJrPHnDV!As5Cy!KS#S0&Z1i+z?O0t5FH&BHcM{;{BH
z%BuKyQZ;g3JdxN%&RdKz_M|VXJQ%F}sRYHX;Kb9hp7ec0#b(b`>lS9q=&{dw@~+mm
zaz2U-USG8;eg#Eg-<z%QiGl^fGi770g=%Z^2krCoXCM1`8n+~&md&B5z1_CO&p2~%
z`L<V$x1*!$uaHk!!Q_sbF>$;821L6{^qR!N-44fkTsZdlKgUn{o$ZaK7xN&8AaAt&
zF-8$*7=6$U=zw>W>f%s>Zt8-$_G+GX5wh%X?nPA_gfIg>%RrAbA)?n<yN<-n!2ye#
zvDb5)TS^_2qlpb%ge7fMqHnch(M$Xz*0}N<A3qi58o9{g3JK|h&puJLd>KC5GrFIw
z;5MYRSmEza<;>!DFihr<=qT$|5P~J6P|G2!s+fItwgQ&BRF=%2Wyo_rWOwW0Mlx5}
z`8YuWI(hp(sdg`Yzg+7mPZQUimk=(i)cY%(BmR+a!~5+sA3^D6h`Igz7bh*9tZYy6
zrv4&+k&amJX36$focF$rS>+*s#$3Ly_Eo;6gAN#-_H~{hAB!UMo9L<M2t@;BvG>@w
z2@Wt+Of1*YzVpl1-AeY*?|->M*q;|fRPZeL!SQHG{R;arX&XW#TDQ;-Mygxh=br*s
zCyuwDtJTX0E&tCK%-(ZOK~^#;ul7rjEk3=-pC}oJz2{5;{3#H=;5}z#N<a#&C0bti
zp0iaB6UTO0=(gMN>K6n#PU01kT9vR^yXd{p%n+c8YX0JA=yXYXi}2;JOJYLNU&;3}
zx?CEqU1$+N#X}y1rM@g$BZn7-t%CV4kJe({l}AY;4t;2S#V=-+j&gTU_$%KlyxD@~
zc2#aTOo{S_zPg%U)ky(Dk;0`Xz2?2(XBN~n*|h;VS^x0R&m9+6D;pVL7bVAco*=dU
zvNpZ+y}WJL#T=eSy(I5tCOO}XdM_ya?bJ0v)<CQr9g7qwT$TKSjmd@c2+2QK*__7u
ziUxX=;?BlJ3wO7W%yz&gWDCq;?KW``H+Y<jQ>8$M9>W!mlimxas}Z+E^f8NVa_$)w
z{Z_j%H-U_<01yKAU~XjZn&JR%USc;T$jDb?H@{c-aLOr9vuUwBA=C?kGKP=o?-d?<
zC25@r&4B5u+N8>CgUzhAwGES;PG_&p%5TN*Q&(t%&3>;ui}R6$=Hy}Mj<Yu@<i%Xd
z-3>ojo9ny6&!8la$iZNfbzEddvjus&DlAW3odjb8&c$|Dg)bg*iW?jFD&G0-TtB`{
zN~Wsh&5QDUL;DC9dXF!4<n?3i-UQdQwiQ;Jv;K28fyY*tv!W{djWxtkY$8F+W^44n
zrk40~Tk3y7?umc1X8(&~Omc&7eJ>sZtSA&Y#SujhP`j^`1T5G9c!3Bhv5uTxVhZ`V
z{P5&vWy@k?!WJoBlyYf>?_&I<iMA>r;`hDqwgvAoghofuk=t0z0`@?nQIRb@d4;3V
z>S$fE>vDIICvA;S(~U3ogT5ksC|m5I_go?EQ?jn2h=nm{b)-2g1Z0gpoZxM&Z?{Ks
z^uH>JhXkj7do+eAQOIam6^<Yz0d08trF&%Mw$_Kajp$@N^>a($l@4Bo)E-L!6msMW
zFsU7T?eX`HEd2jTQ9saQPXwlC9F)A+oCLXC_@qPZwI?wpJ>~0GPFoc)TqYN;r(-X1
zWC-rCc<+U*Ef>ZN<L+^s8t9pZCEeC1SSS4EH`ePv@&3T{+s1@MsS)8M-9@^-n2VG_
z#cyZ}Lh}}1-=beF`ju_-gPw51c-!vdcNb>AKjDp@ON2u?hj-;(h`R1*#buqa``t;s
ziVjO0)t>ChE2u<aWD&K!@lAq<UN1cW`vZy|JrDfAA0dHaBIKxeAoir)$g7k>3H-Oz
z+i?Om&;y(nV?lg10W*I5Z|!h3m7eE5MdR5(HkjJ6#E!(g!d~S%f~IqA4k`s3>KQ8R
zRO{bWC=*<*9nDL+w41n3I~GkuYlO23ggowmy#CMsVkaEFe;G4!q@9eWind_qn#WG%
zSu2Z2W~(~w4F`I7G!!hh1i|1;Aku%4+nJ9~Vq`lPV=ASFJIaz#K6d5GAW;kjZ^;}3
z>F3IQ;zP;(%|1=E!hhsG`>YyG*$!k1J7}NBW=8+ozNmgi^UILWK?lD~aF8<taC}u^
zUkR8!%)KZSeW!h$fWN)9!M+KU6%wHTdT-Ej(=Ef^chy9`4jB}`e=igYgcO#1xUIxo
z<6q_@kt9RzhCfyvmBLMl7hkcT{;OBCNPUF`{W+E_CXIk|*^*c?!PjP|gE^BzQrC+Y
zX4vUi>vEDviI87#x;%SxYL?lBr=7J^a?h<c{IoNIj$EQ_2S72|p=DLsVi(2e(pNtV
zG;tTSi%%Cm=|BC&F3Gu}lIUc#HoNq6Xv8QhB!S|xXp<RF$e(4G`x<!hRdV+V@3FYP
zlnCpzD^CZnbBC(kmx?Y|#iBPtJvQ3aF{BK4nHuhzY8uBOJ*wAM%ahGOnMB~oVk8wv
zdfrAx3*9QXY^iAMx@e&+Xd7+R>ELsT<PxJe7<N7Sbg+HP;Yr3sGk<855ku(%aL=@{
z6+9YYdhzPd0fCvA?!8#?MXvg6f-eTri|nh~uY~4IthRu2GSffLh?5-t|8Sj6_9dVy
zAQ|nARa0eaa+)<I0o_^Ow90NeT^Ox-w@#gIH~Szhci6OC;Qa0u6gxKE|CUA-J<feL
zBasR397;|$+sqsk`^I9MRe@09S^k{Ou9mWsZ^RygHFL5THayaEMT_t{02g1edA_8^
zc3a^8n%Zs)6UgI#JI?N^hE*h*1MTiWV3gEQ#Q!J=E%H5sYPZ@w)l&DAtONuT+r54>
z<|1m=#?K}D6gznZ_N6YqaX<d5w0C4Zw1(=emG-QB^NFpoYE8a7_N|IjOMM4zB!(dw
ze6`xFlccy$rKok*ruB44NVO#;WPAPz;Uzk_BYR6mWB=~UROD31+z#ss%%!+2n7%9)
zBR@9T#-)~Dkyi^Pn+j_|TvsK8=eKv*{n5vrOW+PLZ9b3(x4Z_E*XqQUnK(E=S=Pjy
zGC}UJ2Xju8Gh&hC-t~~TAb}MZ+#dGxAe5KvrU^(X03lt37|}->;79Wp$qnH04!w48
zO)uEmn6LNnD{uzuV!l#wlfL5HAFKB0sy8_E#}kg7XzZ{UMlXq0F8$)6pS$dAMFx7x
zSJ@`<<Z1sm9B#8clXroHyGSD)))PnK0<a@tTB60+t*@q%zf$o@wq8d$j=(m?&tf1q
z7Q*|A^~E>oCX`S@$w#-uuF>4U9&F3C&XCc4d;F%vJG-slug)JA+O9m<+gn?0cP_k6
z0k>S@J-(}2au?WMpOL4$rd^~&>A?2;IE%*Gft>A6_@COLYQ+S5v*MR{yk3pd@_d6s
zRosVj4dch(U;{Z&a}8zJkDLw$GCUZS8_8a;_WlZQyWjAm`3-uEbxgO#j>Vq20~nT$
zS8I~aO@iJQJCS##xEMHD9_q?-G66!sQ@sL^PbFeEc$M(ezI1{Z;2O^)0(I5bAYz=&
z4gXcO!$NTalh|+Ps-^VHA)du<xAT79H{5W8eG)%AiBYDV-S(;P1xIXUf&I*fE^Jlt
zeI;*LaTT%8qpzxLX10|t;+w=H@q?;o`(-T76WR_jn%Gx)8XY8D`8t<^%}pGIqEx?0
z7~q#9%k{T8Jica1+$Vpd?*jCS#5rW&I~q;-)vteTKb#K5fHEkc`lF0sg6$SL_1jPW
zCm=1X_|Mho@)^_1@sip5scOQ)SZ3U3!{Qs)6|z^vCcD7D0VZX|L-^!nz3~0=yv?|M
z_N_L&>QihF#h>k>_k*}j?-ryle!nn6c>PMneaZX&&X>>>z~M>Ql+fT(AGU=8dv=-s
zd)zquHZFfZ7@a!^VMLvEh5tr6J^bl?>;|&4!xmqao08F#Q<A^v)$bSXp^0a<%dV+b
zTV5H(2-~%}!@=STHo`kmfu2!E#)@6vEJ;qzC_C}QxXznFphzn@%9p}V`yq~_-w)2P
zlPED+amK_b$S{Qik}Hg>oW}Yn%01a(<8reIb~1cT$jQsI>r2pe?YD{5X7@{tTC9MR
z;+s>(Vkxl6KKwvWD=*uP?-!PGRhLbvy34Lvd}i&YY&v)5T)Vj%YkPUt`r%Tiy&vjK
z9wFBsyvU2JcstL}Fx`j8UOS^2UU_yT&{6{D{lXkCCF{zp_mBLq$l-d6;P!w8TPQcE
z#%3oRVG?)PoW!O1O$4mi-0DY8!R9$84D{UdFSZ~72h^msTbK)wnJ6HASM^P2s8sW7
z>~0@%pl7Kr;-Wk@f>R1Ze@}EN3^7o~<H_#LZ?MIvsEQ)DHmC0*%>tr)owph2K{Jm~
zOl3TL>m54to2)SbT|@Zk(rP$mYj*ym`TbzcG2l?Og8)Tqjb^SkFM1kl&&AIr*3LSj
zMRC1HDNh<T>k9lSp)4Hp7DQ&BVp&e1te=EGa<Nw=usfTfE4C`Zi?jayF}6fu6#074
z?SULYbf^qrtFswg{4S=GJio>_3oV3;eXv@iglMJAJ!lWb8p|`pQF%^HiTV#a#x%8b
z+9T1?9nU=b)}w!Or&Q3w&97^7+Z@qLpy;3z@H#()skho=KKKpeeq)cnA6hxg+hoFt
zhv11s254+ClRg=12;D%jW;_+&v~1<-R4MVNt8Xut<eONz?3wr`Rr(qrLq;&`@fA#%
zJlWP)Ka1to*}ZJNgR3H5g5(QD&f8eERDv2ez4SRAEUvGUv*s56>orN9TWni^Lf$1}
z+~x37^lP>~e?if--SIeZqrU<yHNxqm)Z&gcyQ<lde*v~K0qk}>X_M3To}dr{2M&3%
zNYY;-o!0`}>&P&@zLvm-{Rv>Spe13!0l!7lBUh*$%6oDfJmIU}x)ZLd83u~df!Q7B
z5qQK-8_4VM+8WUFh%c+2YOVHqqH^0ZJDP(ooI2Z%<-KjKrxr=Q9k13emuYOj=!(Ro
zzXV<?MD;{={^R^<b~4D_3_F!ep?m^)JeS$&3Qh;Xa}*z3gfrf^y{R7E$k}Kh&qY$%
zpaj@w=W>)Tyj*mW=kr6!Ow2?7N%X2x{==hvnp;5IEU&E3vXjz-G{a<n?whA#$x8de
zL8R=NNYh{D4X#(sj1IlLU&WsOWM5b4;ZnV8;ADvT=KUiRiWW8M?EMe_fA*~}1xx&W
z_FY0?2zq@Va}M#96G7ma{=x5xG{-+y8)h61_osZ7L~b~<zb&Etb9JWf;!~wZiD$vi
zg#Xat{gNgTk6yH4XHvT&)TDlLwDFhKZ3!2iaaKmORd3v&ZFP9|E;;gWbBf;1?m&gn
zLR?&ZJ5*8>$W!)`7!m4TeCMUN@_Towk2qJQo#b|K*_k8C*U(Pzja_~w7%X0zPP^hv
zVR%R?cVClTc_vuJ8B&uiLms*6j6dO?jq^VO>uMj2>-qv_oNIE6As@=At8CP+tpMV(
z051GC;+L&qV`bw}O8Surkk0OoWt2_3E?SCA$YrJERv~l(16<C}jGJwA1rXJxf=hu<
zC=SLpCgztxFAA^O*!boKO7Up@HZI$hz!BtEv{4hzR0$qiI~OZ^dU5QV7)?dBRR8TI
z1j0$V@D6!0b=hQZ12d~K?u~`NzA^fR<_-ROy-mq);>)0bGUeXnXVxu`_nYH)0aKyw
zl*hDbesVZ@bJM4LV*<y~3{}ul?ggJlGWa%|>8-j{J#JRC8X^`0F-6bMjx*m_ym<w&
z%)sNCgE<wHUnzz%)s=IZaIP;?xp$gtC?2#D5;-qgyN?6^E&6v^_7|K9t_8)KJM`JY
zGes6C1;|K~aaU~Hbt?XN-8}>pJVQK2{kEu@Lr+ifj?n2d$33}LPFH>tgZ`!W<~4u;
zBfv3$-O>UxmcNj`q~00=8j|=GAt?7~;8Z+#ckrAw#_#T2Ul?0K1AkywD=dxm%GFt6
zPI9_w&KYiRTV7+W(L%}!1>x@@hgf@_=iEP7N8ZYkh(|A1*O|hX*DkST4lq)pcgU-9
zMFlXq5W8J-RW6`N#i-(-SNHqyw<Dz417||^z&nI=N0ekvBvyhkn6NduzVa|D+-a*l
zc&0GtB=NV0hRi8>aX<3n!|_e#id|gTogBp@XM$@go}p+`i@E$!pS{FG6mQYBabks#
z0Z9)fkP3TUt^*^gm+g4b9`kifm@wHM_xqsSm6Pm=>eQ4kSEEP@Bh--v^2MI4@Is7<
zOBBN4etRlr@fsE-<D@;EGsA}|=1^?tGkyk(yRGNUkxR?_)=^)W<Bo3M+Il}5n0om6
z4(qK3A=L`U+>6e5V}8S+Bcr|?1N!`I;BfF!&nrS{V<oFaEx#>s2LiLww&&qZ4X@1p
z>`@ZC7lN}Z_F8ydpitrwi{*cJQa0322<*kQh*+~HR#oCLL_knq?2T^{NW<J^`zu0G
z<%p>GLRFMH;D_^<k-xS>iNh!`<X21mLI3gg>GSPyZekwG=%|A;^WSO~x7m@H!;i>~
ze2{8{V%a`++3VhP{ABW#R?9w6v@Sa!lE)lcG$v-$j^}=*g6xo>Q~-1$LCWH~>}0N{
zWhLxA`p{GPGXW=!b~=9sTf3X#k<K{Y)ncDtX=nZLnzV~^dH%QD`Nw$+?0hV56LvoN
z;mRrF6L0?7^W2$F^L`A8dW1dm$x-m*iD;kakWH{(b@qir6|7SMI?1tL#(@7Pcl4_`
zcd^n~(s&J5>z!t*uYDNo1f|nA2{^D&DrqIk{<gXx@++4BlB)ghvUM?eM&kDPhN+Md
zJOJc}_?B7-o^Jawz|#fi@snSmhIacoL5dcSb7?jEz*#Km5v5v^m&&jY3dLDS(v1r~
zC>%_ED<H483qJ^^g=Yt&Cu*MI@#T?$|7)P61+|O32m4SG`HOw1+h)wQOFjtV=ZTpp
z!0f;M`Okm0%RV?VNm_^U>Ir^0sus`xa(I$g<xoAu9f+37Y=m!(ayxvGHUqRx@L3-?
z_veF^RN%3(D}BcClW(xA{GZ~|t_~1*ngINAc@(bsAoVGKlTe<C^}Js@1QIpHlX^sq
zA``BnkEFyO8Q<U~#5JOi!N1(U!KV~+K*?@2D%SP88*KCk@lq3MgH$Q8VN5l$n>!%W
zHa6$ySkQ=Ih8Xuj7~{HNh9(3E{)StZa_WP^KEJk(7jhk}&!mLie4+)&mKAe}bCy9u
zv9&kGiAu-PXH#P6!VYi%uT&x1H$`86l8Y(ZZvLQf!&0-@v_yagQd{V4ot{gU6>Y``
zp$!MO2#bXNvh|tXUF9QeR&KeiSV`fSXJ;4pK-9IZvpEjSTjwpbxv^Bw?<(#rvuDin
zEs^X*x$G8HC`+h@E%d*LEL3LuyDET`7$LxdyLfz)u#xtfMGn`2o<B8p;W=dYRAVX#
z`rhbKlz||J1lHP|2ep5r)kO<;rvZJi8#UzmOKk`e;mbX++|WjEEkmiodyA9lQXlV*
z62M{2(bqL1!Fl4Uv&mX>-r4dcXwP-erw$!XQ61GGfqn8_?MmF7Xv_Sc6Mi?&R`@@O
zFR@j5E;v#(6Pj6P_xq1i=l{VT$RlnfZ^i0_0#^j{dT|9?<85nT1|R$&xE6>kD!y*G
zf5`kpjsnwG-d<-9XN!hr9{%oP6+i*y3m{`8io!yt9`!y=Qmn2G3?z||t;?;G@`dt%
zdDU1rPgE1BMIQItKu$)8W2ZfltDlXf>PcVei0iMnr}EPLp&k>UJstaU4obYuc6%ni
z3I04mC3vUDmx2XuhpqR2cPrk$H`+=Oq-eGEyeJ#1^P?Cl-^8zK<UZ?*)>2(4QK5oN
zwnXdFq7WaYhLT^%w!!mMGLTN&9+b@^)FJyNMVXj3N3H%G9%OOfydAbH;Xr_t?T)$9
z-Cbc6vflOCo)3;(C+_TjC}Xtkt#++!n^J)8vHgili?L`FuciahN(&WPRo34_*;5z<
z1n@JmSMx}2!9V9dJM7@aF{i->^3?GUoMcCGFT(cMV<g1}7Vig0>(Lykv%dbmPCHh;
z1SQcwnhY_qZmw1Ch2yyx4iEhu9w1xoL^Z6cSj(mIMLQYa3M!P1ykw`U%h%Y}-CSp<
zebq7!*%?PSu-eWJb}jqeg<NwkSBZ)jf`2~$#32<0+$WCOXcoG<#dkm~KlSmJ(fG3p
zw_Zj<7jC5z@pA`Nb3OX14*R0QD(M0+Pw|Ke!}{&Z#3~}Un1qZA6;cd(1h|O%b+!1H
zAYX-_v~Ti6=1zrM`!)wtxPzXu?_w}~2H7tz&9Xrhy7zt6ln3HImiR;No9a>o#Qj(S
zUP&ChWX4ZE84(6e^~>$&>NxczYc~Y{f<a2zAPt(GE$?F+f^t_qcYD~`;4P>^-BNEC
z_@DfWBUOuCSpA8itKNp6O=jwV&<XByvt4wyFjI?3K6`Qe3<}7~fd-sx>{S=;HZaED
z%r5nBL_M@ES1Pc2X1XleDUX<3rsc_V`Pon};hEZKS40ctH&crj@ea81tg|s(c3>#r
z?<+U0B%a~MUUfEjGAfDJqh++K&zARE>8I?P9MG{HCyJeRZG2M{p<J5}VrU+5HaL&l
zm)vF}eE?|OcDwFuu!|V$yq?6UvqhSoH_t}rexPECLFornV{$^V9mww(aBOU@a9S~`
z;yvUCR5FzzNZEvJuozbz#h*^hSyi#U<Xf6_Hh8MG+NHs~PtK{3Z1yod0-N2K;H>MQ
zLhF>=DM}d>Vd#O#U66AvB`Bz=VWr)C)>+Nsqfj&s#P1ebKh2MwFczDh<1T0;xM4<&
zi*#2^Yw~TMnX5r$kgGVnv(54WI#;&X?8L#>)*e2v#O6fnl9&bLeHL57+&s5iZ?SpV
zoO}<!V}Zkb{(ZJEuwkgDkh^Wj?d>JIE8#?zVziP{Dczkb)Aa*AaoVv(xr0dcrCjg#
z#Lx=L*$av(xvRT$xKi@o=%e5iIijgzb#3%<t)3nL#mOqsMNuAEokKv`6&n(S(@6Pc
zjRAFkZ;M|KQ_8JwX(9)~H~4B<b4<%U=f)MML2Hd~eN3oAsXO4s+I{?55)(roJE8@G
z7&HPd57(@&_*S+9;**KZmiYp&|H~*_;qb4mC&Fn}G*zBumBx|B;QhHUVr08pYV3it
z?v@>u3yZ*0NSw$W7%2$8Y^y8ua#Qf2fqP8?1Z700Js9mIPZkrn*NBkok+*wIN%(s0
zVPEhBT+tuN0j4P($fL1jK~$U>&JMQLuQi43S?{_SM|z5b_E^j(O)Tyv#Tlz;4easi
z@?mc6mhJP23Yc;(E~jiyCWij0JbO<i-YNItVthVuxXw@Yo~;yl`<Vo5*k!QbSx+oP
zK;Ftp$*;Xob6cMa!ED%Ei=1!0)fWhAG2(2@ZQ!J$-h-&umls|7RdG#Qa>X^Rt!*o9
zTOQ|Bilg!FY_HZPTZ||>WozP_;+y2ztJh^$puF&LKUky?8ow-D7!G`Q)u#mIDA3pT
z`2D^0pKY%r8SF6@x@ak*h?37h#_T{s1Nx~;MB+z&kwkweS}WRqk9La6=+%UW$V5pB
zb2wTRZ7P@a75J)w#1jgrblH(=eR88#gwM<Ndcp!;fRaphbV!@MXoIMcXvcizawW3k
zj)WcctL#Jt4o{S;TEdEp-3MrpEA3<qm1c;S;I~fYp~=sec7sK_PWwV2lKc37JMB#L
z-o){0(}z%Op7n(+*MoU3es?vU0u(mc`TPc>sf&Om*FK3E!*?7Gk3++Rvd2D+^%qqS
z-#mauj%-f#K1<+X!&_pX$2b!1v`=M2^5gl!K{D<31@@(PLCY-B|Eq*;6jJ&+Cj`Uf
z8MmM3!gY;BfoR{vH3jOhq2ESp@lo;;-dkRON-A+j2)wfI9I45z(kA8e_f>PRCgMr-
z2E&1VID2HIlr%1)mYS6eRArm-y#%8F=tEy$lC$rzpYlG52NR3;b0Q!0&)wo_9~MUI
z@9BXHiw*m*a0(p_cEN|i<xtR-qJbYeizNfF$|@fMyNAa&cM2D~z1uGOu&`Z}3A2my
z=c1V@$^g5>zbUX3P8Ugohw{=7{V{hsEK?Q!uDr#Y>@pv$X-SP;K4gBS786@=g|Dz?
zNl|fL87*YxLLdP7cG*?&tz_#1fV*LvuKutxWY)AZ{WU&Hai`s9*T!PF*RToU)yr;S
zw-O$RGrzTwF>-qLDS!WU-a<um(T$FBz~0hE+#C*uqpRg9vxTc7#^fM}DgVmyAOJSD
z`nCs^rfVw4IUEQqvI)8R4$4DwpcAX+LI_9yNgoDxc{_>Fscc}AbLohPkJybL7QPWy
z3QV7p<1KF^@l>nb<UfvG^iThEv)$}l80bL`hv}bI!A9PzL#$>tJvUeA91e6CU}nU(
zO%xl`<`Cd)W_**Ki<D&8x%+I^hr#;ZzsP3$?8<&YS-@!QtWqi0X>&jH_gfEU6nbSC
zfX@Vjuk^g^NjA6_xGQo%UJ%Dju^X=7!VjH64_C$V5czv5mh4%(%SRaKQIaX1RQ7Lo
z`{H1+dr4Lhun;-nd3E9hDcfuJB=nnF8;TRry&qOa!1|^5tf1Vfb-*?)aqWCv>}u#>
zu`3=ds2afQw}w2bb^-(OENM)@-LlY@Cg9MLTXSsR)zZby#ai<lZxpe=5ZLY3?gP9=
zJeqYRiXdNgSy#3$mJ}J(`fXY4xCDRrd4k6BL|Bmvqi=g-r>%<KI;8dGspzx&2S+UK
zHp-~{K-Eswpd=Ia+3M<G6on<#E#9R=7@akFj03w^^@C2W{``0LP&NL}%u{3!$2Vt=
zWw+dHk5qGp)eSxNquEz%cC2IVFWSk0j12AB4n9qZXGtV#k0soQ)?8+f=h@4Xmk@Y+
zBA1IP|0R1e0oC4!(w8#K*?CqCI^d`CS7cVI8QC+ABwe`Y>h|^I5znb#V(Y7`D6gEN
z(u;-n#v<hqlWNzE)i>P~0lMG%{P2<9m)n;39n})Pul_b}i*HoMQmg@HidSrV%<2rE
zvn#rto2@@v7ZszMg?ITk{48<UU*fLByM5NC`zVyY=fe<FTa`&frk%YpimsD5cyc}3
z{`lrxfS%j!b|9dNbO@ht$p31oZ$S*OSA7O};1)aV=<(~3Hc*YczE}k0^l&7pTT9JK
z_@~z${IC5x`ha!#vL5w;u;rnQ3ao_0h@~k`P$f4Lz?8=Lcpl^}jnL2&F<B{#7vMZu
zEx=t^l%VC5ll2MXe`BZpaBJ%1%5o;!dJP2kbU#~xUU7ltM#h1ki(jN?#}FMl4soC7
z159EZHrpq$2PdaR@&s`zPbT}+Ta&s0uQ~fHv8@(L{JH-`rW?f(eUW(D)GD#wzVxQB
zp#&d)6>E1t8h4rgI-jo^`{u(VGyjKGm2TPk@BP=l_31D`x73nJr!WHF<r28g(gN}T
z+4sIiWp1ytAL69f!L6@@u>KfulEHet{S@6PCal1-BBB3Woj=bI-{gER6QD9HQ1{mC
zT;Y|Vb+5n!4Les}w2n43ofr5osMwkiXD&Q<<TAIn$?-~bNi|A!R>RLZf0aHMdx?&}
zsQT8+nbQu`zxZ4zeLBgFdda!CLEY^bx5W-z8eMqz#0{f(+H05PFeT$2ECLFu%g=>+
z4dxEqzv5gxQ8P)%vQn&oUFj_#KIllh>?+^WK+mo7?P?!zpr?&^?YP+2#Lk8a);Ps-
zH?GYUPbEeCh;vo!h1entNW6w4&xO$sc&O--gl^zoXsjx)T^AdW=3SmC_*~hj>KnI_
zLsn_u(Qyj8lUr@fx#06kw0fP5^>zHItpknkxT-rM97p*A7i2<w^JhyO_IjHb-}Ex8
z<oqjeXp`cbGeBy+*(Uop1=@BIcUkc8#%ertO)Oj0R;L7*oY$2&Dt|V465NzC6{V;i
zzRh;?xnSfXL7GjAMxsL=Yusnkt7+VYpG+w>!}qGnS>%SwQ5H9<$ZeO+%EO;u13O}~
z|7J~k7@a?zlOUAA4k2LfkV95Fl3jucNuAf&ygcOFnrdx9tO(bb>iu6BT?I&0>?(SH
z(DQk_%Q36D34(pM_k`_&=?ctRl&eJ{fF}5!#2}X+{NcDaW(*A*{6eqQX1{4l(Peef
z#NC_Ze}<tirz2w`_p_W@W3+SimI4xX+R|9OcuY_eHz!4;O0m#<u82E#LIztM1f%)i
z7;CR!&PG=8CUiJ%ilKyrc2xt)%>yk;<$AEJTBa0##k0*;MC<a_fJlSRx2tm0Y4vRJ
z{`_lkH@1ied}+V=%{W^fgT{-EY_;Cj<cT%l_WEEn2!S8q^w~oRzgq|_xz8T<_Cyon
zdHjgaByS+RM;`T-?VN?MWNq$U)zaIcx9h6W726`mfPMCu4~Gy+kH?ezc>Rqf2*pn%
z>Y?aS!Nn(YZ?`pdG}}|THXL+O96p^h{$pGBg6`USdnN~DeJoOfPLHEmr%HgV_sOTj
zm(*Er&Qt@-faPJM&s2oUPV4ib2hF`(a_T=2#)w9$+M8?{E|+dk(AJU&qCW}e)Z1-W
zG>#zpBBJgN;-_bMKL@Vsfu6S9!F`CGd#VZ0*Ty5T*N=AoKhC%P`7@dc&knNQ4g|i)
zeMt5rx#4<b&7o+G3L8C=gv*2Ws_$3!Bp#5%{#PdjaB;UD@WoS7xYds22{b5(?)3_T
z<zhUPLqD1T;@<I&9P@6s+&O2)?YG%+AMmvvh}4N}I32xUqn*q@yV{n+1f9wwbo;W_
zb~?ca-H>K)A0+ZjK7sCG$w8mZJ(hH1&g`Dcwb#|OwAy(;pHhs;KFP<2n=V04MN)@k
z*|ty59q|G&l#IdyF(M&8i|s>*(fs<0Aqr1@WgmP#&an5{7k*IVCj8dE^qu4G(QaSG
zzm@;wYj3ik1Li2FYFsJ9II)3Ziof->qjTAeto2>uR%87N`@TAS91H-6^0)txU_=p5
z$jXnobCzFkKV@&+!0hL2`6tyYH2Wyt%UTaG9rls)DnkmC2>swyE{I=5A|tg7R7rM#
z3qNu$#@{E`;1B<(kjTILyMM5YJ}U27snS`9?ZqD*k<<2T=z;B$=u0&M@|L)x;H4iG
zHbv#gWE8wCmk2v0M(*-hh`aG_E~0pmP3($~f-|ZjC$9V`)LCSd!Rb3kny(}@*xzDu
z41uUNR3(#)=03S{+r!%0wI79#G){z_<a^Z<1=vR9vXu}GFrXc^k-n{_74^+_UA4wN
zMetQ3FBf%3eH5%*Ee>C!{g0YDk)ScYR*w8-l(=JatqpjJa1F-AjD&KcUKG}Cu?Zig
z!I8uqj9+3kN@#D%2w)Sd{mHKi2T}GnDZZ6F5J4a&`&_`6Jk*f(8?#?A`Ku~cuTAmY
z{RvZ*-Q>SZVbg(7OI0b1bJ0h}(qh!W3O}uaf^?iibv4s-hTw&SU^5&To~>|ZG$^)8
zy2^ihmgCgag}2)5K%q4--g9y+XCx7O_S;-<inF%O3m_?pFMK9;C^0d`MNB^PwjjF8
zgZm`_f~T7;^x@H>HQQaW3He0G<8~k06Fc#o-5ote2yk9U3%1u5CHl;->8NS6d#Y3H
zkeiF$>+5Q5y3cAq3ho<zt;-Fzb-5IYA1qQ_NP{nB>_7b<|6+|fc&+Lk*;2n)#V@Hv
zj8t<%0=Zpbm6G^d<6C)vp%xSgvi58zNul7<3)T@}a&Ct${q^7Y1uZ>&o>v52;_XbH
zw`Cs%Lr(^|dU<uJhR=%JD|ttByH<G*OgdcJ`#&lg=H+$vKsIZ^ksgx2I^iI_nfo`a
z@t(LOan&BoiCa2J5uOk#F5AWEVh<-&YU-B3)jSf{NJh9F_NdR+CZ{#UnXir0bXGj$
zZk3*QU97Ol9&?Zn^xQGc9{1VFO3x0TsNe>vD+Xw^I|*OJl&ugj)o*($zLCCzCyN~Y
z*hikuLD7LcQ{4-_`)<y53fAM(@m}x_uXkh_b^VRj8$F6V!HMX63-4eXW7%ES7aQ+G
zUg@_j31$pfXsK*lg4rYPMZ~rzWJwol{ex>NR?R_nb5~N;#T}fO44AtU4hUe_^Gk`h
zJwsNAn+ud9?JawwwUiWmD@y$RIRe*FrY$m++ktF{jUDL>aeJQ~if`zsieet@v{&Ps
z<ESp*i*`8a{M;r`YM|PS`*6zqr`h-0k@&^QimX)9qbX_9>j|$r331^8J{rVk{!BYo
zEp~f}k>I~}JO)yh7sWd&r<;fNoydbm@;@0pQR||glumZPo$`qmV2hE%xSftRih$!O
zk^haI@i7P}XtT5a@7s9T?OblNy?fOvJD&$bOt)vw`otGK{<jnC)0`TmyL19C;34~2
z{yg43&smG=PVUWH?TgrxtH%qeh`0Ys$CMf_B+8!XD#)uOl5#A6UBOiXjO18_&hQ@D
zH+ee=i~y&<jggA0$alE`m&6$R`)o~)aWM3Un98+R&w<$c>_^|k4dW);Px14=1Gs+&
zXV&32|8v{|Y0{rJJAXv1MolgCe21NPb&cri7Q5hlNZHLb%R9T#*<5%&jRS{dW=9z*
zYG0M3wc+RE;hs%mm5a`YCIz~J+5O`4&NaC?RhLvgUds1!$@!r2m7}d)ngcD6r&07Q
zq2WSOJfx(+dEPERAI!{L@}<HUuCyzB(k_~9K-9f5*KkWCVys<tK6FV4dZ>VlfPQt3
zUOc*0IV>~3Tob!RuKcUq;FZHC+>6U~zBXqsl&_9=bVQ7=O7ytR0GStUWPFoKLkbn#
z?CrV$GkhY}u~E@N1c7`;hYz8K8l5nq;3SoYbqU7g;(57byOHgAY<`OnCpkTi^T8E`
z#XX<ke^It%3G(2?*i!L@W-~I;xX#)c*h#r%L?b0kQw}LT`X!$@QvKxIvW}`eO*dwb
zVh};M4h@o&P;_jlqo!29SCxvHaWLQUO|b)s`YQM;{qxOvB)rau#YhqwP2AkH^TCSO
zx8kDRXwwr+ov2|DLQu)hNRX7+HCNZnjBkA`Op4D#$l0^b7w#LIrOi%k>uhhcIsUzY
zw2pJ*m}H^twRzR5iY1;?D5^)XNBIl)93?2+)yqu6<7x|i1VZ!VHhEW!N6si7)4QXW
zkU*(}T;nA!Z&9}6+YY-Y+Qr2Y#{)FP?A_~2h2OSXzr+JQBmOt5i`|y6ewL%~_=fl<
zQUvSCATL>Cd=nE?Jc6OBORKF^iEBlqP-HbnYd5Lrh__p7EL2ZA1y`!^MD(<wSFEGz
zrBX98RqVE|T<gjjTjpC>imG)*G{tC+_Ds=ZFW9QQzcnqj)S$UP2~uHWpoI@aA6J`6
z`TwG=PMqLXZ@~8!Hfv4n3P*?{Y7jRF2$gZ*!CcGq+wGwkOjZmPjhEN+!?{D8J9gV6
z(Lxq8xkO3n`Ka$eQs>%g8d3P-bOZPmLA7phADF=HF`pGZ9v<G_zlll(Pb7BNAd|K&
zu_w<LhOld~Jrxr_$HWKVs!%J~(|NPP|3jggil5IKe$O~o&m%1aN}gXV&CBoTaXMh@
zfANd-YSO<`IM90YS}a5Dv@y31AQYvp8m@i5^ttnHwJixdWCI1WZQvlb`Oo8j`#aly
zK6C}Qw6(M}Athqv>#+VjPU%pU0B2W>qLg`hSrnD9+eabUz3r)Xyoo2k9gRHfM($P{
zY;QuT@b>QXv_B!=ify>w4)}Qv^h}p`{h<VSXS<5U+N<6e6;DaHbhug$a}|W{wSnrJ
z(vjfs0vwL`67QJ0z+Mk%ifi6Um&&omj^;v&D$kBZLyX2yH<|s#4XAeU@dW#LN`jt<
z>5!>J1i7jE>|}g%!tMHnQ{D+dvyPnKNk5=Ouh^L!?~b-cI~#pV83`T{RcSq!P{7em
z$>h$TKXNk+!SDYk|7V{htSQm4%s!37E47i?iI=jjDizyj-kM8GqILT`Q3G%C!svgI
zySNjz=$Ekz&t6$br+wuOs8vXXv9I%;YHw<<SMa2L<9BIAgMI6#b?2Ph?7KXKA@&sr
z4|MQ*zoA%R8tn%knL;)I=f{LCW|)`krz9T>rrFQ2sEtYi%1xBH!9Grg!)t3&LBfW8
z9B<gNHas547w~a;h^C<ux$xsC)`oDqk0Ewo1#>Y$^fryw!(&kA_=Gh5c2Q1Lo+UBz
zTkYbHg8^$0gJPFN6Cow;*tnSNwM%{O+u`6Zi=Q`Pyx7kLm(SAW9~T1jFMqKsK2GjU
z;#e^@7o^>9SH>@*DM}Zj5;Cq+z3Ss&A5oy<df8-G`*3yl*)`dzI7n(E2bogHc<slb
zjUpXImyL*aZV>buhXTbe8(D3b_=a*f_uF;31=ueUcx56VHMr}->!>jI=<Lg}i`elf
z*qG{@<W{F0V{^RX*{4h>2W{NP&Mnu~t+Wa8y9ZIhE1~F{nBPKG!mDnR^2FPv_@ahg
zwt+{$P>N(nithNv7Z_vZW5Fiou;^|s>du>d=7FA>tL)}zO<F;C!Ez03+Q)_Y2K|X3
zo1P7LeiQ_viqkVb9yBp33lGS>(q>i--Qc|shIH1)!B)$=hb+AWw)xyGC@5{t$HA^E
zdv9IQR?f|nZbr8&P29W~RYqbZa;5yUEpS|$ddID{Fn-p-<1d1INb%z8+YZz!fRw!a
zcl(wGdTv1pyeJw3f|jhj_ry0{0`aIM8nt^rK5`f2>%V6m|DXBIYJDFQCjG|h0zHcd
z4G1T_nKRN2dE|B2jfvcDjD^b3q*_-hzO*#H$=5)(E)^JVj&FKkB@4WSrmfY{;I%%0
zx!&4iPF-VSQHquhQ62HTBcU~(tm9qPga>$8Txq;4&lN3FQ`-uJ<rVp5s>+EeZe~^9
zxWCE;nb);g^yHLre=Zkf!qE}t=Vf~!mxcrHzuD^OyAksWn`EnUCN7DcK6b$|NtT>`
z#dfOBo%9tCM$g<1?#le4;J}BXXLxoqw~9k-RQfBGZMKJ#$hm^}5r;u3i(-$)irr)`
zqPWf0CVin6MwhLtfGHJ)vxjcKe9XsPPEeMD@gI+6ltI{<!nj3<6c{ZL{)q&ehb6#8
z9b^Mi**qCNpF`=KOp7>2F>7L4p31v3oxsUfn3AXSn)39uleV69cG@!ue2&Gb^Cs)b
z?Lo0+ImQFS*2gzp2t7hnIksNEyd5|t*v7oPdS{hyuFm=r2T1MPR;<}ua$wL(^=!q7
zZ_8~UcuR0s?))K7lWorhVx$l<D?LVkBINBe=Gd+T{8SN?-PIJz2|~tb$lac5?Q$v5
z3$)4hR^N(OnglG{pCjRLc)$)si-M4WmWT585sc7?V(-=5A~pn~_z0#y9KA%;RmN8~
znt>#+Gv?V5|GnsqJ2?Rnwb!GME5`huAB_v3WM`Z0SdLm4P*jk}O~>P#)+dHh`HW8_
zU>Zqnbu$0B50%=fs<pyW1mG9!be<WdAkYaZK=@2d`)`L|go8L+-8<yP|A)5s5AnL*
z_J0=<5fRciA|fKXex>8?>-OkmZnwwNH^(2_+-8Hi-JXueZ!^w}Gd9l5nIER^ws<^!
zqZ{G%MD*=O#AK%9WNPX*GqbQsA|fIpA|lMp%*@P8A|meN`C4neXFQh+_m6wo>~P+n
z_xtr)uk~85^;*BzMJ~rrV>o4g6n!2R^H~y{Dq+cabkqkUzDb75V@~)hpnEO)c%Cox
zYkc`9LPAkeh`5h@yBuJsS>iZK5C7d}C-XR?7h(i18K;h~Qvs0PQ$au#yE&ae`gK_1
zhR)=gvG3vAA)n07=4%w&EO!b5%DJJ}8dU`yHS+w>NiCAfUNMXkzvv9*LLM)Qik1=x
zTIEkkAViQ-e2!D`b1tELDoaR~a{R^N80?#0`n8o>b-%h~(|MoS(YQO6s6ox>M!V!_
zkr>8L95>D`JsR?$)Mr?S>LOeg&pa(d$l7I>`!~!n)2=w`hKh+bcIDBKsQ!HCf?0R}
zX@QN1Z>2Niol{payXt5d;l|%O$*#^9;8@9QukjbuUudaado%^R(&vxUoq?X$#iuIg
zF6~G41Fk<BBISIQR<j#&syYtYjq$!42nuhon~oL%^0&XSkui_wVD-cLl$IiE$Ec&}
zUZt)WLh#K;-3BQy-XR<9f2rECV#O1?DxQW<Xtj-Th}7>u41KJ>P6QS`S!|r6PQJGq
z|7Q;lrwIXy^GI2oOT-YS?}`4B`iM@-mG*St^|i^q()!XRYf80JFVLqL<lAFVg>JtF
z!BHZ(ct2jXJF2$_ih)}Wwc147)D;ajJzmgF;%>{GHZ!NEYCRFnvyO)7rp)lA5fnN*
zzLQN++JZU{%t;EFPY(!TY;HbKRZ|7e)>Rv>zNLzU&<R%0s(ICT`xwt%y$j;mR$Mo#
zpCfg5p%W)HR(ovGQNP&3`9OS;T)i?<xXa+49FtQmp03iMi}M-AKG~h5C@ra;5r$r*
zzdd$u?q$0Ys2k!9DK648UV|JPV=d{c^r99LIGO{lk`WXvu?RsB9lEr&#Aw@4K=zl7
z`kO7UP(@RDdC}XBhNRoEl*nUj7L?XZn0h2-cUDj<?aUp@T@{>UyntVfm)4V5g6q8w
z78U90Su$V=lkLj*Mzkow2OY~+1;KW7*qQ_uo3=hD9>X3;3L1aw?`&;;59SZCLS;d%
z3yMX6S#9h6585UZ>+oP=t>^|teu<;_kgtxv$V9Y<{a`92S?RL&h|j{=D!nRU0ITiM
zq|-)Jg+|*@4e%US^Tr{BQqhR>_?WM$j!2Kk17rzF5C)q<5%B6+X>`VAzuTS+V22kD
zk*DT>JvFpwd4j>s%l34Rz_uRZ86G-&rh2a=Zx_pHUyfSDHIbm_dA(+voVZ(QS8AIb
z0Sbl1ck-NV8QMlwLN2OqacF@sN`qN{%%uJ|yC7n95+1l{CYP98<J^Aq$bTt%iVQ;X
zd~*Tcuh4=WJ`A!)6lDUjyQ)R(E~Z7bXx;7Dtf5HD-X!;8CeUp~FA5H$?8*{C_W7z7
zKR|1d{S}m3iV!Nb<3N%GN`(rm5o1|6{)=|dzigmeiyewDMKUrXORIb1qV-82P-25}
z<qrFoluGHaL8l8^*WjnIBQcI#f`szPqHS*{R4@uXqhOl76JwW1WsXG&<SJVDyK!ek
z7kGOw*6;U<$~mh=OlhuG?wI!-Rimi>`$1kexTVx=^+P`za<NeFz<<QODgBs$^MJ|)
z_M#2hC-IB`Tm(8duYa2OtIQv6FFatMCBvAdl^^wQC03HnkNAP?aqMV>&y;?(gbh;X
z1V{SuYDg7fLT@aN74hE_eypaEeMS{4pgb9;jzm*V%&EYIik-a<e6%=*K8K28oX#~t
zE6-|RMBg)hPuKpFolW%ADG#mN&L#YRAUeR#$G1{U6|Ew#@P&AmwS-mV?x#3KZEYR)
zb8-P9ze+O0FY!Sk69#z=V)-=(K?~l-_U|Ax#y&5?L-!iH<ns^_kOn1%Ui!I<QK{Iy
z&f&W(o~Z&83NI1XuiE9Ghqyv)GnF&0h*!Ay%cZBH^jB7&VX_98t#FYMpBIViw(&J~
zRnCO?vlV2@2Yhux)7e4D^lH22a~D{)Aio~4Yx4`<m2E@ewAWRWyjG$MrTt%@uuX5p
znQk}uoZU?wcB8+of&!)2ykgyM`aHx+LeACC{yrO7&Ep-a_rP#ReSSo)+`+!uyX<Bk
zu;$hYHaY-QqrrZ=<#WHX<7#AK16<7VL4i&N8509bJSvSTDV{d=^CBe<E2I)7Rczen
zMcTysitJqzLu06`9AT35Ww{iCoKOudUnpz_Pv^wy8{K`fg(z1;9wdd<?6*k?JWT&(
zo9xpmR;AmfBvQD1NEt>=zTFo+j*@_PILOQ_XN)#>4|GK?pw$Kxaxv}Y07$)^9<xXi
zdc|hO8^uosBUI0fSrys{8RD*gR<_v^NG9vrL7V+~M!_K$(U?s^zntzl)sj6<UY?t|
zJ{iTn6^T(7Zy-V;qPwMLUIjquP>eRc(-!#JT1b9hSbd_J)z2SRpQA<jNpZxw1@@kJ
z=9M=UYO})@$2#OV@rLb)tL3VINnrV25_l^S)@47uJi_yKuj8Ho2lzrmg$|LQ3?fno
zX+s9O<hE&yVcfPGH&Wpj%?T(*Fy&7vo1o?MG!>KiC$ue(&*kTrfwKgkme1Qh521`(
zllw$3x93*BC7YGV>feA!g-bg9K)k6U&t26V_|#>S>#0_%($&x@)Zi+~$~@tv<^<Kw
zeZvO(usH0qHI4`D)96nRR47*Q1~IckwzdL3Lx3S>VB2l$^3Zel;BQ(VAA=-u_p-hR
z6HI(i%i6h<9{T*q1Rd)CS4E62wrtwNLDo_ccG)AjVBM||?`v`?5kvH%J(_#s=tiM`
z!8XJWOBj^LSt%&uoVwYl(PSGFDGh4k`dE&@ZI1mM+uTljJf6u~w6m}PWW0cgmL%vw
zPktVL!Q%F5q~ePTDL&=9*3p1*;^~0dEr&9a?u0$aFaAsgpUSd&<#D%HqJ43F^i>e;
z#$H+C!E94J(_e~dmf?AG%z#(5NNop8D84k)Ee;3OzE@j+;*O&t11>UL5X*p*g9;Ef
z+4h{EL@Fu4cEEPzr--_*h~O0mRhF<T7AlX|?us-}@fvNfkBYIUiBjqLTQiKV&GzTY
z@NldXdxEtd$XDu@=|~)mIaSz1;Cf9P`Ku2lpSq|Nx4n@-mi0-tlkBMI$ip$4f|_5B
zH4o-8QQ?Xg6bbG~67Eu!X~R13wl9Gs=^4}QoeE}R!J(#?U|H{0&)lYniPYXp`hblQ
z312NknL<itlG&9=5&OVVhcdF%KCGaTXUh%gA0;#mt)%vlrtxv|4JFvS<n>8>CyvF+
z*wdZ%X*`ohR@Vk?sqC|W!mpC8P>sT4#b1tAD^R42AfpCDJS*&!?LNS{KOQ5Xk#G%p
zT(OhDiHpYevPf<dKbhxCRfFtQj2fp_4AD;K?8Sruk43pYldMa2V`$~9kAi1c6~E6_
z`1FtuRfbfA!s~Y4Z<kvqjI#^=&wBi*KP3z$qWkCE(D4%{*e{7svbhnSC8Yf70^2`~
zG&>do8iIfqc-m~2919WV4<*bUapBTqMXCTFo^q^o4CmXI#rNa}bGH35yZl((5rw|E
zg06^{0-ZvFU3sj?jB`3{L=29*sKjEuTBHj}6jvP!16ghDG`qTbjdCZt1PUWobj>x#
zLU>e4GLdEZ=hC*`20G#$3qNHqt#j;p|FIEAD;kd7aI6TB^QTcnp#o9{d5qdzu{Rwn
zVp?O<(l#6E{~kXH>)@z>MW+w<0MeKkg}fj)`&&?ZkB#=3%Z+l#Zt-8{+&OoKjmfoy
z$fFir3ai{^V-ug49w;wB()Ixx7vl(~bykzRgNd9D!q^NZIKE49;L>qv;<1o$^r+W`
zc_?g>qkQszwaGDy`+u~n+ii-&SG-NPJ6bwg2!;25jzS-;&6!I9S23^Y2`veY((L6|
zof&Jxbyno`orEzJn}E&o8D`P@)@D~<mPB-kH8v-n=?2M&;JMX%UVu#H!s`<I@>uCq
zAb7YqFfZSBqc?{9J+>gB?o^LN)oEYoWQF0Nk?Q4+lz;rMc29h)<F0lc&%sg`SI^w-
zB)+^Pw|H)=-J28Yf&vspa5cnsZAFlH-5M|2>cTL1PiZdA$3j<XY-*4_yv4yn)WmnY
z{FuxBy2o)Nl?deFp#^Pu#(Fy`S-?8leI56AT4&Nf8BknFT`|(|61Z~NWUJ<$TtdNS
z(Zy|Lg<*7Oc26R-)AuT09I+ntR8U(}p<=c@5aXy=%u59o$XEVskUpoX9^c(T_TxGq
zVX$wtOepJfc0_aC7~<6rI>}9^3fDvaG65aRvVAySF6A(Ic{fJ%M*`Hsq=zHQ9?hW#
zUxZaqaE{A&Lp;+JsjJ~t2!CUR>Q}igGNx=OnEJ8gLOnD&dOS`EHK*-~YOvR#ek>Z?
zluyPp-BePN)Dq^Y1O#1N^?&vt;XPepct~t!uhBW4N&GBTl3HJU>j@ntB*!-S8_c*F
ziEwjncOxAp8*NK9aLGg-W?JcwUp(vI?ZAF)269SbEN`{#F{Rh}-Gfo>knO0zP{^d)
zLU#=<t7w7}^>+J6GZdKBZF{RXU^FaTY1{h}+@6xiyMO2qP<5@z4mf_*?$ZwX$yL2X
zQZ}i^WrvdL=^WK6&EO4xz0|Wg9JA@-EeUPtf`k4$O~|Q0<z%I%eQ#IGp<!|Gj&}m9
z#OumI7=LY|j69KE{+Fn}q0Qc}2u49;Md{?fL;m*<oM1?k1>_I?4|M}2-Ye!+1j_4t
zBd6NOesD+rcl#vKg$gd>R9E{?D@b1~mdrhzkI(WM8o?n>nyBWeFLbbPWxX962H$38
z`OmenjwjBM<<ZUTgd=AB#PN1AXrv@Ia0${6Sp&ccPdN$(`~HNh+fK(6Wlw!)p^BnJ
z>A!Zy-|pyAB}F@%H0j<}u1lnA1S|@eCv$FZ>g_6_l;<lXs1HB^-Y(?abDEla?5F&T
z&eXKY%lF~7D(a=`dcWiXyLbbJ|Eqriocl2v*%xV(D(FK%d(JNTB19G{8xzIdZkNVO
z3d4Q2fboi5_C-h%T}{nhc6q!}4g-r<`u-L9tlU2(#?!9+BBYb<4(vfT;)@Kh@Y-(N
z7~om|X2TI6u8Q9YoZv<Yty<S3WGb!y>KyJ)$&Gf+7ezu5W!kmzX>pP>oZ_Bk`_7Qz
z>%MR)jvyF$Cy+vAFpy4ieWHYW;SflCj(QU}d{KliB3yy`#+-FoZ!3*BZu0kK_-gN1
z=`!WW7(-ho*`8cM8<qHg4=UYN?vk4w;$G72Z1m7K#ZHZ(!*2N^qn3KO6%pv@vlp-F
zWb2YpGoP_v6!~#ki;at4ix;0?pM$;D1S;H<<nfo{d%Fuwzit!a8^y1_&NCH&ndsZ=
z&0}m*AWMnN3e;8qaL<XET%p;;kD)?_yv0$R5@XA$A=?;D?(Oy=7%{x4{?H-2<BKAD
zj>9cf>o^(g`{$81J=f3C#GS59p7}+Q6iVKgOc}FsH%P6*;WoP(d^=clt!j=hhC70U
zb(<S+D4a=^GDWL(UxbujdWcGzn-_1n4I0Y;?0P{?1rw%u=fYUKD_DhMFS1E3a-zYm
zN#lcie1SE$O|r$l0+K9yi*IvDp0|c&95*Q8-axlU4r&dXL@y}i4T)}`mdv8=z-|$h
zFKV|nCcacX2b6K%n{&2WI{M&8Yw^EZSwm5Omd8Q~-a|4_Am46li)S1Q%<z5|+Y>10
zgZA=XBeID3ox`YO6F`yX&Q8lf)>UChfj6$z7yD>jGd+n=byM%QmB~{mK#ut;0VngB
zBW-IOon%W<>-&M&Qc>g{Tbr8&RI=KtBEh<NCXskQVw-B0tgj%2=74c$gbf_D2c76h
zs_wFf;!6pJ5~;UCvkwn_St3YSU)hoa_DDdYEHC;E$G5UmzK-0y8`|Cw-$wKZ7nM|W
zDQg=m7}W*H<9Qy-w~JY@tc4ZH`}_D8N3Ox7c98`n{;((VtD*{usGnuu)Trsn3L+0S
zlw<9wB$K79d#Ls+GI=@&C|VeY7k=os=`$6O#cEQnv_3~yy$aWF%Cke+QPPaD0&R{j
zl&JZNYNo=fw*)C@WVQnh*&|7u6K=A8AMrPT{mWt-i0{LDiu6b%SM|&bb&CD(s7^zv
zB{_641bOYMAoq)fKA*NbNs<fZkWLUft6$tov-z9tO{yb+yA9{>zQBu63du+giOt)e
z>uzrAaI5(NU&CM@@lgkpmMP2FV~1iCv0!D^DN@kI%{Kxbk(mm)ZNj|mQk46P59b<0
z=*&$1oed_89bL;{i+054Mev(wZ&!HQ0oN$U)ZX!vvw$)L)V@ltG_T`+d$0O(oV_1k
zK93DaX>!U0`XGp#XXxV}=8C&}8<rtXblOJ=o#G*Z4!ehaT&>9KU()Bpb3=8Pa#?*6
z-$*MEMzOrwr}2$09|c^f=~eouDYt2#1^#4#Q-GPGRUCguzo;q*R`REFOs}k|@P=J>
zEEl~%KFVBZF+>&<v7HD2JmyN<O!_pSiSgpVwUc?uN!Fn=a*j}9BK#^pooA|~Ll)M0
z#(6#&33fI)T|=o9cP@8ut~25Dc?GO&?JBv^7h(oau^_7l3(!wrL^usemB{c!zGC71
z*>~{G|C8PN<%=VuU=S7k{O9&B!=e%FSD%3XDUG-U911(0J~oBR|KIiOOOB_%1EcW@
zXz9}983cu+or8bbaTlumM)cTvugH6Oe%jcICcq|Kaoh#Ztk6J;n_W3Hza%8={m=5>
zy+z3q-H$k4WKK#oXoRbB$bao+Oi*$4ahD-##@jUkh$K;Hxo`&Als4J5F|V@Y<g!)U
zhwF}ql+~?>|Kd`=-a+|)N+z)zVi*xI?#AM_xiOxJ&yx<#fB$AT9e0s*8CB;-R#Ve?
zfT{8L+Nf$Aac2pjFWSxdo+_e1f3M=F8+|;ifAc7te2e2@u#c7Z*ce}SP0d6bTS2g?
z_zktVamR~nco(%s{a4hYR^?`v$e;;1ZavXs6XSDL5?8-3mGhhwd!q2s7Yl#l@<&~H
zuKatGO+H>EdUY6ACCn)f8r80q;eUG!D765#$>H8%clZZwD6NWJsP#WlP3V(CSHS8E
zj}C@BY-YkZ2PDqQU01g-32L)_#-i=KU~}?kS^F@V+uR%w{SsfF{6SXdn?mukyKG*r
z8DvXrkq>Es!$kfgiFi%6aA?fZ=-3t|+0(m8Ek*CiIm$Bq)?6IVBss1JSKFzyvBbY4
z=np3<1K%6(xRK2_x*_llVIhqBO(CWv9Yy+XV^B6_RqzG*e@l|DHTy=_&a{@}Ve)LH
zC>v=ks=NiJmM593raOHW5Tn{0wF_E^`Ly;JT5h7v?AA`?K~eD*>r4#L3XKEYWnBT~
zuwZgvQSE|KN!og<l}Tpsw1|~H7p5|{eU)#)GPM9)lW^f4$CoB0^?_t0@BpvJ^8%x&
z@6+1c8%1_dV<HA^Z$x)r@3X3E8-)?PozH{*0zt#*3J>|LC~@ufaBkvE{^W{&B>z^C
zD{>DXjU|XA6ep}++8Z3jcQ3My0brQ6O1P~~FBL@NyAs00iB_Eb<JB`Yh><1X1$)Ae
zZ5IVYo~)*l-@w)Or{Y;Ly|VW&p@lx};FDQHNi1KNf+p^?zSxeu_>`Pi?5}Z2Zb}Rj
zkQ&EkvoB@w-`kelTors2o%H)Or62h~fWgJAfH#;X4VV4IEu_E}TeTzc)<SJAl$Tw}
zoCwlxvfVzGOklP*x2~~ON$>k&Xb+q4+kAiZ%ulQ2i37(&H$x)xMid8q?X~cjMmu!;
z$UP-`<YFiC|95}c8^`0`T~SY(5+%--vFyVEsr%K6#AbuO*x!xAD{`bl+}{-V3Hn~J
zw-eK#ie~Rk_Ktr*h*6ilTOI5dh_iGN2yqV+|9cg!y^2&KElNVOER63v4NWJB{{zQ~
zBCCiFi^-8-3e(dlA0_r@5yRDHAN$lzYMJ{<yza>e*tU@+K25%_1dUR&L)kO3l6X%0
zXUPHFrUFSTH%6V!Ri9y}9SyduHUe}C%soVOfv0k79UlUbi%%-%UOQ1OR1wOOxHoG|
z6t|NOEgZYfPWhKaqY=?j62f}ybP^qQf=)XVW6Me}gA&FwJDVf4wA#4}cGbT3tdQZ!
zf#>6Q&5fbsE#iwnI03(`UG0@F#Q4fFQLLJLT|fEwZz>k)=W2d=Dv_h${l6rHb5#N0
z*LXu1D|vui@%E*Q3hmVUxy1iXgExe%FN>r!*r(WhEg6&JWnYHiMXfLO@45WT5D!$_
z{R;npg{cm6KVlDAl|Eo&6JD$l@w>Q%g2yCUUG=3$gFVNk|EgV`!xP!Aw1R7fzK{~c
zg8J>+Y8I+tFqbsc>*AS51pC%q|78e`^C-Ke8~=tcUCMkO%Np^X;86ZT8IIhBapRXR
zM8bGEIT}QD38;)=H~Au(70EL)SKNtbWK=ce)}n@}U#OzC8NRIU<}btSII~L#8tsd%
zqv#la-BJOq5N&xb$gA|xYn2&jWBm`V1?C->L&4H$N>lS?8tA&aJIF&u;+PP?$TN?|
z=341n>lL4YfsIyyF`vu}J2@>hCbLPgW^a2w%uryH6Me1iEmSG%u_^fgd8^d#z5UDJ
zk3?rSm$dmizKj^%(v?^l1_%I10#@q?oObv088$s3$L7>Ve;uf5W?};O5aCsrN1n4;
z@hmE<%dlXxzYO^Tij-v0+uGg2#?FaXv6m}&vshf_<~B{MZ?L)?AN5LFw@keA@)^4=
zACYSLEbx!WxF?BVVUDn(x6u~G+d4Pm8n{yIo_MBnD0hrZjEnslb%bfAN4Tl3wI#W8
zSiW$s-0L(Bg~J0OgN8T(y0Q0IV{AxjGRl9HrE<WU<2~x=!GT-yq?8T~bk0d=r{bk=
zOG+;3j@)yV*4iGwh{wshK`5w56*k>@5k4pujWM<>Cz*+q%@y0@APn{`YeR=$nN!yB
z$ExacodGmc%p_~RjQfjLnI%wn8OqWFiJ^H-D?8dscQacXAIl1%<?1A?i)T8AN=c&L
zoUPAE5O|c-@DhIc2fr-J*L1a4u7379$Etn&a5eO6x+02H`G`~FZ<hS!xAthF3hS8q
z(+ySU*$`iD!5$`iw=CKlb1T(vg($8#an38GGOH@o5J!2u0!Y`PXKX)_IHa({3VSl%
zki4y!8o6VhiWMQDK0P!+xd3H^80I*zXA+)9<b!VO%iSI3sV;^E+Z6N2Fo*`lzHhb7
zKF`$oGi*x&)m-1b)cPy16>?alavR7u?ovYC_Iv~9K5k}M&8xN}Cf^Fa6jn$#0RF35
zD7*Yy#TDUD+?^Qb;3-y7ag{EF?M-@_+PsW<uAFgDurHT5N6{j-KmY75D$;>K5HwS8
zuaGgXZi3)UTF1edOkxYE9-Mqn!+XgNRZ9?{V%l5njimP;mC_>Z(I*r6_Ht4O;}e%j
z302}89`Qra+_98+p|^8PygoQN#DVRdMBJNCa^1b_OtfT#l!Ws4a>{n)G<4hh$z=;i
z<%azrRv;SF6-W>;WGa8rhxut|TQ5;;_L1{r`4fZld>rs>)qZosaK7vlCw{n3mwlS7
z*h}syjO8<*pDT)db;a}_%~|GCgAWPfSOpJmd$vy%_BoxTX&%qn%JuJYQYWhY@p_TE
zbBD!1*va_a*_~^0aMEH2)R?CnS{$BTcDh18rDm9O2Z6|E5?$2XqJC4io%IhoC<$Zd
z;sa$4WBg>lUa|AOocac(kq{eyA>S#TpS`2}PXScvKo4z}>gu|H__yM+e@>J%cQrM^
z(th!SRCC(|`?Vr&Wm9<4k`uR`aOuKV>fRl!iLCLG6GevlgW6izr6-Ej<nKn?WhY!R
zQV<+<-A}mLN;W@TK=LYF;U8iyK#jaIUXU9^@u%YHBVq^zZNFq!1qA7hl-BI|=2xGH
zxKee;BG1$7j$d;kjcjOC#H?QHpf&Z{btkIOCVbb5PwQ4Cl<Q*xwz2p(B$WH&Z-`$+
z#2x_Y%Jp`;F~BC;TB0{B{gB<{VAG!d_mrI*8E?p-x(SqSL#Q6*Z<M;wU3PPh!5$Ug
zB6ic!@l3HkDwy~J#B$4tBIv5VhK;#sJKSp`Q}|{Z8(+$GE5o<^CFA@{Vr8&q#daas
zwA+MunS~*Gc<9l@6X|PiT}3YBq!S@_>X7gXT*H-%IL4DJ=+(y)85fQ@C1Jp@!<F*Z
zzu_CZy`qPgU?g`qM9B0sW<C+JAO$Ixwa~+hf)+^{^ATQV##-I5Iy_WsR&FttO*IIg
z9rJq5w%1*l<L}^<pou&B;#_~HG;r)~a06~#!dvVS5;%6*yc6YOt|6^<L4eLdWN{1Q
z<&55x7PBaxxz!=cStJ?2>N{<*W7V_#yKRa8S=T|13tZ}6r_I_{Ylx8qykwgJl#s@1
zv*iwSuZt6Ab5gQO0}nM@ZK>Ad_IMbQEg!azp&k}$RG6Zagtl0uF7cPG-B&*g%VDSQ
zMGsyO>bF-5_1>+U+1Z}lzmk?)stl|=5wf+Cxy!|{DyEl=4jJL}lYd{PGV$P-Y)x)u
zcYU*%(gP<#_-bxzTIFSiSKC^rn!&!hR$F%>s^=8(&<$!548$oZxh5KY>)&m?e}S2Q
zvptwoNan;X{ZODasym`;$VDX`?cr(@UN4Pe)Fktfc-tdXq<&ypqY&&id(__^?1Lxw
zE?ZWT6gE`YR@DfHNa0uRkltw<lfY?^Ymeo0AwNpKe>|tdyOsmEh3g65nr=$WJ(>8p
zZ<#%n<W|?)&}L6NeYtc!)}Bdh(xnu2%&ET52dbSh&o%`FX_g~j+cw8D@ms|=k*Z}|
z;@PIbeJrK?*ZOn&xd2dz6k9M5w4|DJ2W`90qXrv09ALuusINL~PD19e^t^1lb3u1a
zv%T?qWCFq*U$cEN@Fvve?Y2KT6;^Zj_kj~dW3FFWPgs}O!NJ@FL^b8}(@&oN?&S96
z{zAY%6f<l^qxR(8m+Xze`*1IgJD3mq@Fn~J#i!c_eO;(Qb|gkpIvOk-Y4|mJJ9kj}
zpvW;S7W?d-ScS-Er@iZIR?T30FTU|xwrF}lth@02n4#FLIWweB*$45qN3R#9B`&NF
z6L(M#YsQwe**?m#i5<+?XScz8>~G`KT82cR8b4H-h!Gra-3+hWrvW$lksP1T@;FH^
zM8Q59*me!V^>_W7Tw9crLv}2ALJx7Ar3>HjYE9(k6+!(uJCS=z0TaTXPYy$(L@(Vw
z&iYQ}i!@`0vQFoU>2;<R;}5nodB~Q+V$UW6!XPyiQ_+trRZBc*=X`PVnh1oYu0#Sm
zlY#)HJLQF#K!;GlPz3$iPd)**9weBb9UpK0AA~8W|Kl$?MANEbfgps;38oqLW?z-t
zrFB`4UGi1%N>c1s%a&}Heib(`O_4$DW&ZB(C*5Y3f8|n`yI57fTnS;ysJ-GV7qA@f
zC2q^E{K~Ix*C@1189?PPFK60D#P6~Ph;0_TOY!`<&+n=jQzrR;<4a{bT^-NBGL$YI
zcDr5kRS`^<sUYvQIcLcND*JHJnY-?*B9)cf1QFe?&$ZGrmm_<F4~04KfZh025iAw`
z*KIfD%2ChckJ9bvj)Re36_InnA8gcD5nCty$8Pq&7Lxt}eT@FfuM1YrSud_<ZG5>;
zce0^QRJVK;MmIS(NXW7=iPn<7=8pQY336+jx@(QAFtWYiR~gnd2}x-SKEc;K*mrk(
zsZKXB2gN(F#bL2YzF0~1_t|9soDh)bA>%2zO>CW9ZLkAA68b%~|Mo-;&&tgn5y)HZ
zj`&gwJy7UezEXK!eYv&qyDXYY4LQW=8Xi6{JpffDPbG)$RA_z$yr+njY)J<m2NDq+
zvjdzI6;*2*&daBBzDlDXHbLsQVqBUVFUzt)%Vy+U>hYFJy>)Ry8e46i|C55RwxF6_
zRVT{2Y+*dp$`FrAQQ1Ygk+gMb(M0#;9u(7Tw=MQDYie$_C3yg+u{$M1_}+_Vk#fGn
z8gfg!xn8U>-r&#_oXKO{?9cEMV;DlLlRnqt)BI-f7+db7jk?ofZT|Syalf;6rzW*$
zku9+^-<mVcx&k>mxzZR*4O&>>R_jS%meo_<Xk{)y<@s03xw9%U-9Uy)d&v=7<5Nx;
zcbh$sxM=TP+Sb%+Ykm80;Bf-iIp(B1lN`cW>tm^6e3>KhV4S8jIH|Umpj4fNp*rag
zC0R_HJJ%leWBSMW_J}`Lag#24bl4i@BOxQ3l`3j^L(Hoj5CxU&=MLK#kfP@yHsv{-
zM*dm1Js$6?R`N@dC%AK-@Rd>)7Th8tJn2Yi!ioG;^}2WCQ3v>^{paa37TPn3LGq6}
zoGSWq(`sx}VsLJg>MGghT<JnIj4c(yXwnB-@IG0;f7`KUwGG72rDZ}K#O(<TJXQ*(
z)Z-n2<KdD>$OTwt5LMxXyKPs_yqqH3yJHwms?<5<9@^dnm7>_~xUTmF!i6B2`*fog
zc4+$(pEHTOCyG8^^f}{JXBP(SVEm$kAyb)KprvRY3WlN<az*BWVGXm2*c%Ql5eztJ
z4m(a->(|&|wSMtX4l;Vy5hs{G%(Ay*KCiapQB2SX65KnvxH<xR;63m9>7hMMm%SHb
z%A%l~O3j?!_c3R-+6T!XDDs1KpvOLpNhInhSX)KwTzme=2VJU^qmT1=-dAs*__3=e
zO?WH+f3Cwm&4Y+uiI78Ih|j7WQKej&t?58mp-+7_y+Uf?%62S%(NQ7@4dt)c86bw^
zISz<@9+d7!qdSqX{9&4%bcWFDOypF&&0X=<ddP_gd^=sC!<~fezMZMw8z_}(<fJ$|
z?3wf-Ind8l5X+((X}A3HvYpRkLJR^?vIH{P1qTaf?gaZO1}g;`pmo(q`MH`vCDs*A
zzr}uWy694Lvi<67!QLvfpq(tj1=fmgyX2$`G@@PA9VXC*K~?O+O;MMgEW$-6!6o;r
zBiCgo!@c)Mw&L=WMdWIv5%3lMPogsH%99~CVO^AHiEwBmPKNwNfFiSAm7~nU@N#v$
zug-$f&y}cf&B-D)M0Yw^=CuwpdMd`I_&|Fqr>1%<UhgAPwF%yF!$mK6WYvwnkW#o<
z>4-Pw-|#)%W+P9A%qS-me)V=6b<!onY~XPSv$Tr`H&+lxTp@cfIuI%2p!%K2ls!EA
z<Q4~z?jeXyV|;Q1+vgB$U$?Q2y=i!i$K^u*GSX@S0_PXSv(L!VT_&6yvbV|DR@BdK
zo0uOjpfz&<=>Tbe_t@kBa)?LCayS&iIK|OE*e7AKv8mZ^zX$@C^mO`PusdR?%rGro
zlS^Yum!50mOL=hQEFq|ytx^&2={ZTUp|iI$9UGa#ZnK<Nx|#?oBgtWQ?t_d-98?()
z{J73ZGU_A%bZ)FSLj#_}WmI=EOdfO)>hXl7Zky*w>T2!2&lXf;E6xyt!p|?MeWxvq
z;lz<89d9KfcTv)e${<PtI_VNHnRrEOkj}dmqeu8xb&%kwi(?KIpH}=usr^%^e96g>
z-b(Z=-k)x}H(nMfOgrpaYDj<pMd}9Enl(BCdfSwn-0VcO<S(~aOT4QSBGDZFYs;&p
zD#F_{i98Iptr}UK<cq*y?YT)@agA+kQs>E{36R>Vxw_&LH^4~9JYYS;CN0&E2+&#S
zU&;H~LxuX}0_Xvpf;I7h->u?{&)WluV%&F9GS*h_?Bc9=bmO{&?JlIVWt{i*@sg|#
zN)=Sys|WKLIX`0W3W0pcF*JTsjXmr?V_~As8zw}Fc#kBoqec<>{b+7ZGgf#ijj-Pv
z^3h#ntKCn#!Y138<NSH*ym?dSF1T~Xe0wa{fKyuiNFEO|(HSfyAY`**X9bq4d}O59
z6S0@2YS5GZhaZ5aV&}FNKbi-L^dtU1Yfn2uRku;icAiNd1`}+hY3)+$^Kq1>*=3u2
z`FGFzC)-?+nF4vDyn%dSwk7da_!w3m?vnm|hTx@(VHaBKK*dU`QZp<^VIGQS-EZ6T
z8}>xjA=QJ}5zjmkLNe(=+vN~)+izQ8yZwbyH5Bj_Dqws4g}FHUx@=!=JzJ*gtG3_w
zjHz`a9|)jydLmS#+-Ch8ZAHW#JXsXmxeM)31)c)nWfb0JZ^Se93HP|{k-P11LPVf>
zDLfCF2a`^EP`T_#d?f|{WxkSQ<?RZ+Y7T_Q0g2Edcqib+N`kcTZjMC2#8P`N_+U%B
zw>x`3Pb@4Y?5IF0q40zJrP4AL_aYR3=zILvzy70r6o82dY%7N89rkhcp&VdX-yuSe
zuKgsh4l>q0tq`GnWVK=0Z=c1p0)RLrrr@LTOx01PKr6TGST#eqHIM=P*dEUrW=*vd
z@s@<b=QSD^xjQ;e`gXULfZ|jBN5}ornoh@T#D;-X{IekXOa*~@H|eT7XlG+=;^9iv
z){t_3&XGe)Ry*&&s8O(8$X&p@gn*T&mr|xOoBv!bTvKqJ;Wzpvb`he59)9(|S~-P)
zX<xheBi)`8{uTwi?Gk^c5cftg8!?khzb*`Q1?j|SwU>Pz;zt8TJc(=Vw#)tZ!M=rx
zqr0LS^<|)NasJA$i){AhMx?ckcrr(P?E*^)mk~`PPB~Rv^>v~9?k>Cf>kz~Uo5#6$
zO$_Tof^>{m?b>*TCW_{y;t<c-b@?8;@)juKb&kvRUmx-Ql_g)bwSmYDmUY9|M^r*f
ztsd=0UqaKWrR4bAO%5(KSqWHzEsp%UNRLzJ&9_kj`8KLYDZE%U@QZS6H-8=GqlTt#
z>H>7zXkW^or`jzE6Bf}OyiNyt8}oG$PkXx>*om=;)4SxX#{+KT5`<dzw!YVDa&!t$
z5R2Go6C6DX#UZ5ACi-G(I~rHpq?k@cVkrwz@*pPXGmOVFw<>{eN<J%=7YQnR?DoX^
zEVZq(I}(gm!d!7V*ZL=jUCJ%=*tF?6j9f`&=xsK$!gvd@%jn-FjdxZK-rb27XtS#i
zMJeUZu{pWUKQ_@!kHyUOoza<toa??Wa>%e01DofFtdrbi3u4F=8<n`V&|hd+tsv<|
z@q!l_@IbPAa?bmk+M2cBi*rs^wAYq=?TUvKWrR?+yrgsTZ!zxyyEjNezsriD3=C&}
z4L<W=UoD=@#u%oA(-$Jx%9{5gf&(m9ezWi3VBg$%){@JB-E8G#4B%2&?&S2_G1eBp
zs$@;-@9hCH*PsHTs#TWq>ctk&70)#1-<V&u(0Y=ZRjWcUVOsEWxH31eigZ%?$f{~w
z6?-YUezxXo7cGgkfd#9CM;MA2Pd1PeT4WEzm)e+qR;>JtwZ55yee@$EoK68{>wFmq
z!OE{&Uw!mC#!aR5ZL<esp^DMiV-NXXgDTf`z#jJZ2m7YAAT!fO5(VxNN86Hrwnt+$
z@o?>v%CT-p`X*jS&FMGBI}-7Dr+8W(%Uzs?J^t~;Gt<9fPxziSN=<t*J{49}=UjK(
zQ`NJbaEl^6+SAn(WSRqpMEGZXoX({Ut=8xNCWeTfhY8;DUJ^LFZF4SnSw~y5ZK+_P
zyr=aind0J+oIp?}GyV;SglxOpZ6IOs0IMR_Z}(?Kia`Ku6))Nm^RGaN*_As)_5=B)
zV)na}j8Vpv&WQ(H6O>y~joS7GP|^YWZC^}UZeh*~z69H!i<l2<vjchfD2}0EB|8|O
zh;Ee3Av_*R{@Busl|ivSZzN`5*(872;oPM8#7qwQAN7qL_mk)sG%f+rBRAxSRkrKf
zIW@I7id6beKA6{p91Zcjo3qR$>lk3)%US+J4*UDQj_FM_O=_|alDD<M{;?<}ng(U7
zE~{YqLv0W~s;25_pq(9c-UCcodOa#!f|#7op9Dtbso7$mItbAF^Ii5?;7Y`-^GNM4
zJNk7x&k2}UW|l0=$NUsfWBhn-xr%BNdn>{9L;@RC2OyV1KwCKh^t7~-2`q6`B_H}!
z;$bB*0;dBAmz6~eFW8xQrX{P3)iwz5Y;J_M)K7D;kaM|06vj<<J~2tTDXi@m9Q2hd
z?WY`|l#`LVD=&b;$$kzVB!O!kKqJ<}eYQ!UE0>Z}(Jx7O6beHY%D;&0TxxczT=5IA
z6<u<wh&vF;4p`}>r!v%FC^OS-k;4_!%T5(hYOoIpqd96yU+&KsXQ%7{kl!J@BA&_Q
z$00@Vdd03h<-(9hN63ZI57;UwFd}|Ypy5`q_$;bIx!LTh7>7tf@hN5)u&d*3ed|@+
z-A*uI*TmaWn29GSO7^w!w&Yqx0PEIrG+pPYn0VVbyFO5%_JRBDhH61x7n3?$cB8*s
zGhve5<o~>D>RcOnD#EWMKf*^wohqW%oBxYa0yq1ZFllw;9$k%js8BGzCcDMoZlFc2
zjX4z&NHAoXlS`6iS)k<Amu>8+A}`MWqmA>gD8=7nHK#%rt8MLBN|=pJ@E-^J=FPN;
zr@}DUKxodSn13VcxSD4E2Rh5-c!ofUvQSzl+LYW1X*Iwx%mWv1%jD$dwCC-PQ*j1|
z<(*q?wLi*=mIzkrf}re8cg&3=8*yf0XJ%{L3Y+Dtn_F+Q0}oPAWScI5pM;z%O6^so
zgE>CL^gHL<+yF?yud>7IfYc>I=yF~P1<XsliUb)?X#+(9<G*UowBS@o-*>K9(c5kd
z^ZUEeZPgE8QA|=ymhv{RZ}-GAm#vB+%@+GJrI%%k6I+tcR&+EbgSyw@U)9}a4Zc4e
zoeGX>ti~*sUJl1*Yd&@4s*=5wcDS%9wj_{K2?436c<S<e(9qRUJo|02>R}18?B!4j
zUVF@@<EcObRrKwQXSzjo)F?}7T@Ed+<g6!tmZwU|UeZBUR?q06jt;T|{clw@^!}1!
z!O68X@yrb+3S-)B59Bj1X2%w9x3x~aO^TdZmuu3|k{Yw#e<XIH%^q~>@$1VT@?R(<
z*<}yMu<po1dqDwyBnVPjwW6tgXbXPjJet3RAyXJ){aoG~VjLL<an^~U*v5E9Vgc%|
zjILYlu^a*Gk)%TT9V9Lvu*ZFSTbB#?Cz9OfK$rGpu8krB%X(V6?Wq_=BAX1BB|4jO
zqWkUX+(~Xe%tBZ|xU32l!_fHsnWW!lw5cBJ3lK%(5~#)5H~BMEU5}9#Je!?yp`;$-
z|Ja!P0Ny>*`eOu-xADUJ1My7uQh6>FX|uh0CT-q<*^%t<{?_hBZkSy@>b+xa_s|02
z!o?93TzK!PP_rp>)3XfYq3uh))q3}Q+n;|dx@(Rd2xPcqiz4^dzXJ_nZv0o?#Dk}b
z)|5J%1cF0}`02{)cq72jKJk%kOaLRsaX6N<72}vn!@Nwm$)G<&YOL=<Y%hJmj`(<d
z`x54sxBagUDkGCS^^TvGKh2nL@5XRCOkTxez}|~b<xbyk@8`nClal^{BcW7S*-Tu>
z%l4tubXPBT$47B2x={Q+_Bm?C-)f&E=I{bxU1V23t@ecM49+O0(mtyo*^bp5)kUI{
z57N+sz?46ivI?bsxgF18a1^2Yld<z}cA}y)IBScY^jDxM+8~q~|J`;fsk#e8hKiG)
z&ao9o#MFe|*qI?hC{hdhIoQT9WoHv1-6TJrOI#5>w%^XjK&lz?Do9c_w+r!xlG0wV
zpW+RfcuJjm2}?uNQs|^aw{uJ-lgSJ4OMv3uB$wbgH)y)Q#%Me%d41}JV&9}&PO66g
zf(nE}udJd2cF8y4WAprl!G1LIOTP*6Kndp<t}csrqTZ@tvCH!v^pIBM<14-?B1B28
zyYibNto_};jIj~^c(8Bg0=p_dA-qJH^jF8Q@HDuO-$&QvJKYHVXgk*qO|3*Dcn75o
zz6pV>2&%(<wAaTk*>6{^mgKWIWH&fSw~ilYH&!b~A1YGYKD()U2AhQfsi|)xzX>6M
ztFfn^EO~|Ajru0;-5H#%)poOgN8BO4zo?TnI>#!~x?fVaR3IY^xL+ao$e3?J+PQ0L
zosD&DwX0_DxNls{RoL>YR^xvWpwf<ybV6+Ea2Xm^4}DA);ypGoK=6D)d0{Df;1ji~
zJ%~+C7>oMjTxd!>6OAZTiE_nud%&GC>w)5SP*~9&@e8H`84;Mg2>bj}8^4sy4YpNI
zK~ZVHP0v};+of9XCV*;YEJaEI-aX{pZJIlif^C+6BfpPqe==3s><TJbv^=WLFUC2E
z&H1ZW*j(T9sl*=C#msV+Y%Y8Lip~2bqRw2z8aiR@e^s%?7UYQIfA>3E_)WTY)3&<k
zn~-FqLx9~A>v5GbLXh5C2b@J0x5Wtw>XGt_)#YG`16oT4)4lPtf36#|-x{1M#*Zfj
zyD@&xz!RmjH9Lr8Toa$zQjPyI<4bg)olcOF3SqrqZMmWuT<Pucx=L;rFCnJSPJf-6
z-!l1h<#vW-J(PmhlV`OK!vlBL%A`@Ys>N1S1R}GXWQ$CBU6W@>RU4^`5HHGpq_Tlx
zy7B|OU~7HhB-1Unb-8dPxNZEdbc*%yrQfU4%n4IN3wzccjJH9cL?;B3-S$vU`v;Yq
zw1)#JWY;3rJ>pxioEV8m6N8A7#hwVC-cV7gJg?eGdGJ-2T8(=LY@-v#j8(-0s0;kD
zY7}`dyoL4S2?4?s>FrPCE7<>NJj|6#lO11j{W9*)rxNp$sDjMhmi@H9#f|%%J(Ihz
z61ECg?W-sf?}S=zZ%Wu@QYy5KmMY~eIMC*5qm*kZD}loNRI7MPh2Mka4py~o*FpQ8
zhDfo6nGECreDo|6>|-1Jz;FL1+|r8rKxRZ?*b$8QcjIiA|8ouwtKESLDRVC{`(AFL
zy-BA_>zi#~qR!3f`{R9i{-73Cv)>NHvwqCA3Ripy4gFw#HV=WM+YSY!!wT-lg5L1S
zD7Q!4$zew?HLn_NFrgq%)s7_ax7pkNytjmyy_0;QRJl|z+q*e2wdfh~z5G+rub8G{
zOFShd?7Bta^enIs5`dZ0>_eY(?8uS!QMKwgQpgh@SI?9JRO&<7C$UWTu*vu&pPCY9
zK211ga=7U)9UqIJwe`yWJ(~1`kK{h8Q`oV1sVK6A>ig|@JoA`(I6QWt6Atv9QVUPU
z?-Kl@CYqf}LY<C+D!Wwv9qQ1TWFL6H+U;x{Pf12E*ts09m=-*;=e#4P=GKXJp;~a|
zS>q~EY2KgWLj};t-uf*1(9by$;(n9;;$TxU3Y(K}Dn1)hRqb@SfRPm`ZETmEj*I^;
zav+=R($mGR@jq^{%TBvAq7ZELFsZCscDWA#DQ?C?d&TLDAW`sUX}sQTSLUd*>hG^_
zwGsIz#cbG~t4>!AKt(U3aZ<)0rc2W?!Dd0i*Z6dIDru?Ru8sHnBJngD=`XwNy3-f=
zOX7a^n&{l8c6|at@kwl|H~6qB5<(cjjs6!7rutQ#c9UbF?vFO|bP@TQx>wt%({V}C
zHKV0?@^1EzATVSC8||-lG&I;P{?D3QYi!KvA_sLKpS0Q71fB{to|5{4ZV)t=aS7aN
z+-ew-P8X?D1;}hdV5>v}FxMgY@<g9t0^x3x;%6~9c|d5TVUw%9^IC;Ufc41CDe;T0
zmQ6(m(4{1K|8@s-HF9sa-I1GGJeMU~akbUvOo%zs+qV;!HQk?4@B-4J6pz%NnZ6bo
zdgUIT<pbAI;j+zU#~7Mly52zw#?3h$V!49K8tZLtg(anJ^<&?X@lf{oy4d0rjXZXs
z>Um-Fau>O<+H670rEj-!z-0_xm}3Kdx$qER1j~!IC^r*>%7EPy-%6O<!dlV1Y;inu
zlgsm{LX?TheN%D?!Abd9sUo18d$%>jGs&M)t}x<RV?1*z#uUW8&>R2|M@2?^OT3}@
zAz8(Vmb2yY3<s-j0`6(rImd!Z5*Zt+is#}ZyR5x>n`)yxqcE>^CJlEI@PdOEG3a!;
z%80|}By^Rc7(Iz)CY5DIN*6dPikt%5Ru22UAI?B4HkP$3F5%Z~RZddhNVNtP3ti)r
z4EFt*cpoHPETaz$L5Gs$?RnQGGTS&rt8HB@tprbr@TIjDI#}<=V;=GA_F#^!#=>2#
z%}B)$Rh(cb@}^n_o{fk7^ST++7T6<!^*#btciN*#if#4O&g!-eiCu_7&NZxJLI~Vu
zW56$&UI_p)&pqbPl;&BoZ#r;YKJF*A^}Y^!B54nH#Yt7NeXQrn_{DP~z%>f8uk<O$
zC7aWzI9qHcPdo6lm9z0o00H((=`KY)^~K_K!B7$zBEih?Ruyt*n{tVFq1qwPZO+}g
zOL-#-dDxPtsE!W3)*lGs(uGS5_+O+W$S<`$<}Y4dkf2kt!{KAEXV@-3`ZG(Vlidj{
zrvilzacYb0O^gj0tZiSshf#?8UFlaeY0gb1V~6;F;=uE<X_tyHf6)J!R<GRNLotBv
zCwcM|0Q^Q^MQ2?Ou2*HZ!$-5Z#F=fj!#==_k{LVbOPGofWJf9@q$->!fb->B^R|C7
zpHf=bnj9}MfgW)BZiTa*@QC-S-E{}o5I26`-=o}_YO;Kg3qSm9F`=nH{fE_BByhN)
z<Rjlf42Apo|Eu<Kwc{#MBcVkyPgR-9K8Z25fOwTdPz=VWrz13l4KKT1a!@{VVy$Is
zkNOq(mr-`ifgJqb^*=ja4Mx`)nDaL5Gbc`mvPe`Nl>)o%WYU;#tDW+3??nAS9l$2?
zRmz<EyYK+9_%js}b}J|VTf-hZTOFJsS1Pw5lKi>&1@m9I{FG+ryK_FtnoPmQUfBjN
z#K;Jzpcl;^o6}E2WKr0gY^QRh>r%i!Wj{M5a86q77vJyt7V7`oug<DqIb`v7n|<rT
zQn5XOd==;@{N674w%p{n1*pv)Zxsmjd)R4SvCF;<YZdEIG#KKtT^`TG6a>}ZkVUVk
z-g}O#(dC{itM^{fMag-2-A4E`(t@ZQsM`PVva1}D!M;D%+tvB+1vD3o?_HKs-6L8I
ze(SQ7u%J*{GVb`V*Laed`?~mD*W*Be1Hvkn2Kn08+w~PpYDpqf<BN7f&bhGFZVXKL
zaU|)W^tn_k`J3WP98M6tN8XK$w{;g%e+p(MTZ*hTqrMHw>n0sk7x2yB7U7ngL9zIw
zzYUSCySHm4%(KyM`PRdqO2YJ*m`iy+3aV1Q$+5YUCBzYXJMe3#`fc2|A*R%NS?d~~
zYOt>XDzFJTRY|U=?3GQdMn|BCy_$5<TdIaIxmsJqS0k2bUw7J+Z{vt628s?Dl;uO3
zx;<xIu^Msej&F;2I&IzztE~X{^7)=1qa{uEPo^syiCFS#Ym2!)?szxdZ$|K#m0%aO
zx+K-u?3@&3qTA+td!&@E9bQmU_l!H1RV+$oFxfP>*j(Rqm3Fea+${2!NN{SidAZlc
zE>%<+TaeRpeW{eAsCi+vv>oh|&SxeQF3Mk^L)@c9`*(5no^L}fsdCSezZWM2Xo-yk
z&)br3Ly-r-sHk{v;Afb|!+}O<V_PKsGz5TRBPF1cquTgwC{fcWV70=Ub6rJCbZ>i0
z3|r7rkh3jLj5GD+2^vv?Tp``omSd8wPndUmj47k1xB;F6>#T646d816wy!InmD5<9
zjMkG_f@gYc1>(D`gSOI-0}+S!5oHyxu@S}Mx<f`tQGu;V;Qn+UjfBxj9*9|_RJ_1l
zu2jUeIWt}^RTRRiV(Xl6;K|6Wwm#qBB2tG3#Tq<VaS{=13X?-Y-w%CT6w-gHvxog(
z|NPJPh*N!2dxt$5Ln7fYWGU`o8)6-x3fV_efb<9TGu{|)r<GAos>gh=TNY2T$DLS-
zaoKK9RIqws=>hDSPx|lFUtMKSCFOKWfh-jWo{q&zq1|H7B)hn)<StmBgTMB-w#kPf
zmi9K=9N)-kzSXwmp3fm9s>}NG&r%7@1_C6Pwq)DbYTNyL(k-aUYdd_7rZu+9KP9)j
z2zk4c@01R}vYzct#Bp$SXR6Y|zNCiw_Eok&Ke?}Ibwh_*9vuk0N(IfV7(Z+WV?$^)
zf5^wc5(u|_BM{cX6?wQqXz6n+9nuE<cM@rc5|7_qlhDPo$KI}<sc46cF0a@-iIf)V
z9WJ$Z9Rb8T))O1`ULw6fOlhbXz?~`5ptsluKB>eW`_O-v#7;^-)up&6KMJ7K1m|O4
zSZ{ZOeNqi2Bf6Jc`LsGQuNJngZVH77*=G*O49p!x_&AzSD=dG79rHN``zQ!|JYlB5
zQb$*VorsBK3+C=Ud;mG+WIiiyLAU#!%I{`%l&0RNeF`tXF~-hRpQwWM)<OJ+b~c|C
zTY*c0=i-^XuaX<3o}SP1PO5HkS6xUlB^+j({S=?bJCX*BpQ~Zi{>G)VUy_9C8x@^~
zVEk(iypS^B?e#0n&bXjaLLOb5<YOh%x)^uKnJ`x2!lcsE85fhp)+AO?bHy%;XDWdu
z4j@~`<!8z*-^P{SWmlXza<wY07af`@T#lDgaOIhh9@=}U#)Rc!gwNK9p^VdXReUG`
zOa^nAnH{RDt6!v?4|5{fHPtVbKY=e*DDB!a<*L4qV{u))<ceOAJ0Pe{FpcZ0#i%b`
zKiMKL*$we*n`Ggl(Cx7st7)jg2+0k#sko{7MF6-==BA&K4#4<H<89QLkc+TgdpFOU
zV;=wsD)8vF(o&7xP$=H$0K$76N>n6XssdJ+afjfTGjS6x<DTfYv1ih>(?eTrTHfR*
z<#sB8x>j?hNXWE6>9z@H;y!Ncplw#`8jv;dOt}DN5nrB5Lsi#p>}4gT22A#))ztjX
zro;^Kv;2T|yKgHppRj*N^+xF@QS2hMNeT5!J)>y4&-&j++svG`R7CUWKc|9OF_6=T
zS2UV^CejW&$=nUm1EqaV45B^}A{dHFvj9obb7K&$sZuxmby!?oLQ_|dMFTy5UZ6<f
z7z=!@Q|HXLg?Tup{BOVgtu1np+uQ7(7}E<J$We^(DmJ);SqXX2O_9|(GROk9B)-up
zQJ}PvA@B9^p-qmYHN+c)-W2gb_N2yKuHP&8cT2si=3GGiN(`Uiyrmk6^JmKw?uITZ
z5w%t0Vr>T|$=dA|(%s9X$U<kVo$(9NgoWjb2(+33>&g}2fs{0)eEyzTHO7|8;ELf=
zc7<X{S{dJnT3+C6sk8j5Y8?9Aad=gubd4juzL$2)4m{;#$gi@sxzalk(iQi;F1|*G
zgY8O_+h^;2XQnQ)2lLk=@xp+ifQJG%ipC<AJk}>2K+5{x5W*bRN2+h1ML$(E$*cBg
zJkv>%cMI!;ZSZGG+#?~&4R9OdJ*~538&Rj=F=vuXSHtA(@jSO`Bi3$D#QTyd2f*b4
z_|TKpGu6ZNeR|42tM6G_3SK;2ZJUy(O5_7eP(Ic(zDQ1m_-tPcrKm!+jo3`|!zQQk
zQbuL7BdlkIZAnsTZUx)@F}_?s5GibLcK|XFNK_G48qmaF5Gtg-qgq>u`q))bkm7;4
z7PU{<YyaJL=L%h$Km*-sdt>KUblAS+nv}t!tRrN-KcTx{C-Z<ayV99mF%<`6IG1R2
zSw3fne7H%ICfOVDv-V^=%TSE^;kdX+LvOOdY8AAywj)W(B3zL?gkSn~fMKGr?d_z*
z7OZwc@;mupeot>{x%RG4NXUUpC3@*SpK!2G5uESmAIqEWw+|}tWM!95@QQtyFWgD3
zn~wsIian$PF{e(HDlz+GM|cUJdU1?>;_Fb*H#2@3<H~=4+g7G*`^>R2uIBf4G=6q#
z1eucT!LeAnTOgRp&0`1&e%&9hKov(X4l6qm&+>AH?(JlfkH=|n2NfL*1@2U!8dAh^
ze%j$DPQ4WHIFr*4eE>d6tDM3_&c;*(Toz_@2*!9Wo;{B|qT(7aqadA+XG+B4P?cS~
zkRTQ5LYf2T0VEPaRsEvaI5F`wMEo4%l&CmVrhP2nms|kz9kgE)BH&|*kTU^O-^Eq@
z2l_%_c(~-dkgS&0L#vm5=Ry)}hifSHBjhl<>^qmaygQ{g^1M99LX)RXiCvLnwJ06^
z${5RC?jE!*6U$~B@m)v<>UP#+SH;lmsItlnGOzwF;)a$XCSQ{;sGUYx$4<L8pOsB4
z_HDZ^W|SyJv<LeTxes>zcSn?LhvlQ&Zip}W28z~wO&obV^LS!pJPzSaxf(P8lJMdU
z?~qdhf^1a2(b++4%gx_aGAvNaZlI>fRjWc9?J$sh)^4}>m#n29%kUUqtP1y#j8`GG
z5yx4hjZ4sqx&>o{tZEWF;|YA95HINB$o#2^`H@s|SEp@KJae9*OYU`>9M2qD=h0If
zHnP`MPw)2Jt8O|VG}#^12a@@vyKl1Egt)V-o=UPdJrE(mTsNm;`{lOr>HvFeW+IAh
zqO!1b&LC{>VLmHYcPlvnvmI-gD_d+%MNwX=rj#US;9WNNyCPNJJ!iVr`9IeXT|Y0d
zETrwT1u-pIi2{hqpq_`O<OsQ*7y5@a)H+*K5t6D?sBHKSAe2gRPplsj<2b7S*iKtq
zA=bl;#le@vGbvn(hE-?Cdxzmp7Z$AB8j^&@|9;{mYm7ILj>Of<h1mREXeVA^fIiC=
z`ebx15fz2jF~G?GW6LXWQZn9aZGj2#@0YCI5zJ+UkGC^k--3*xaL@yU(sxzsSN%Y*
z{v=lSob|-ps$C&vNby}OeT$a%bhX*43OKeUfGc*)kV@9XL6k@Pf!yad<f*mQlpg&%
zB+#wP*XB@{YW;Ub4fyi{doUp(`$SYmV?cXo*dXd_qXX`R(;oJDXDuLJ`H@_)?5q6#
zI<n}axev9yjg7V;-tYj!QYgbVR?l`IpDM=m-|aC!9OZ`FGuj@{fr)Qd-rI}3@`O)Z
z-_S~cd9I`utLIbkvT&%}@>ejhJ^fuMWGJkyh&qk-Or9<2rdTEVe29AKRkkTXf;Gs(
zg`><iCk)WE^!<IDjxF)bPrmYUsR~t1Pn8`=O?H@pH~|=>wkOreo7&O2+IIN+NDcCh
zDIu@Pb_G@GUVFiIe-}!dIuugEpBn9Za}roF+@di5eX&O+lrsbv59{1t?cPopw@hKO
zmn#SDfKQ}cUPOn3F{)p*oJ*9@LwUFhmz3Q>0r}gDX;$U6-iWb13J+%>$ng_>_`9$W
zm-zH@*;BdTyCb(0V?j|+!D1Q61x1A_yUL`k+Y6xkI2cEA_3GUq3sR%K9dK$TL}eBq
z);kqOc5z7*ACtWsuxoonkIMb@Y&LJm@LsZDmB=Kz>boPgf~|x+@n28a<8TU@Vm5MM
zh^-dI!>ig2HBRK6;^LAG`va%VCOT}`hZXXb(xsR!RgBhU^HG2|TzS&Ah@C6;<Kx^8
z<O?~8)DGg4YMC!rj?z#4S*d<Z>~m`aZj8_3z3hlCDm<FYZEjUBm1EUV+EWi8J066`
zUy97iQ`3wzQ6h^@`oaeLMo+O*@vf{WFWc#yFic;hb=jFbmg6VhW@nQeXuT}8sf4ne
z%QvR?QlB%A1_hByt=9{II%)DR!sqCv!_MwR)ceV~)$;r7=NzGyN`k-S9F47HFUTT>
zjq%qCGKqK|>1Jn-NNnk7Z=?X#Sr;!b_@Z|!f>CBJ1-*D3p-ayO@1T+%HEHohU3S(*
zHm57W^^#qFHpIVD=%?MT@OQK+-U%5i#g*|L<3jdsWgKxfZXpE$;U%%Fs=-zHu=on>
zY9AaUDiIkRfNRdWG^u1WRTrli0&hyOzxHfMo#>e@c3r&ZA=PP`(0kXPE$qITatZmx
zpNJ>28v_|)DY}|HQ%*tJ67;v(O#xjgs8N1@!A9oNaRrxH_gp4)i&R;HXm5^Zo?+*~
zucOaq>lC$7=;&0!3Hx=6zp6Bp2>zteb1X#SC^>c<LA3EA&yLNvX$?SR7ky~xOA~LG
zHvk;(NZ9pi0t~O7>UYhA3Q6VW<!xxKj6Lye5q^F%>bExOZ1CVrwjmb2S`zFERB+4N
zl#6C&OOVZqfNr;+4bi-gsy#Rs<6|*Yk(=rx)m8&4EnX~)(02!c@D!WrG)tgclg;wc
zAjoozX8RtJ`andR&B^_k)k^s`s-s`DXc-$Yfh9OJZ*J|fc}ay++k42-Uu_F=4E5LK
zMp}3_3r#`pg+F;k8@MzkK$V+Pr<g@aOEZ*ZevkhJgI)(}7W+!-SFN@sLC#Qox5^LP
zo5(0i8dm`7R70LROsIIo7_KpKTLNC6XP=Z$ty`=)CL&V?sj=m3+&fT2OGl$uvRQsM
z)SmgI-?!z7FNScJ6WoX(B7e==W2Vxwn9V6^v4tW!eGlsHp1;7l5^A<knle?9ttZh&
z?P{%TWx~=zOk;^_<n*peEG-~2-`1QR%6cp*#*y{-fKSyRv+i2|_nZF{9Itacw!-_?
z`{TjB+B$nM7dV{;+Yk949R!2jZx1IXDS0ew6=kMkGsO#EEYgGI0YyusPJuz*9<5k`
z`{-pa3!jOsthOO%tfw5;#(3F>5i%quc+iNFj~)wT4YADE?eTb4%J%p%wI`AgJC)xD
ztWU;X_*p7WIR=(p@2U8usPmAa@<o83{d+ozdj;1zMX9P6>0VoydTRQ8Lt7{TL4;+S
zatr^|PE-fA8S-zUer1!|lFJ@{>+h|<TEcc_cMIx3l08DqF56x$i2hkk4IQ>4$D3Bl
zw%ab>#GWShdAINF7_7~E{W~natwe27h+to=0(OQ(GOQkl?GGSyb;-7%+>Qf2s>0Xp
zVC)U@6oxRjbojrE5y%czSU^hPcU8fDBS^&eOBRkDoN?9QhkulmG6fSI4wy<QU>!z<
z6qJK9=wlKD%@w>FeHX@lB*}H&-S&2FQEf*DsdMinQV7Ibg@^y$07(~|44tZ7|DNMj
zxn)?j-;b%?ZB%UiG4%)Wt<#0Xvd#9PPd8%~k}eqes3PrQt>fXh!P)F%2ZvsvEnLct
z_DP=DH#ZRP@#$HQ&+z+H)cZPXm-OdNg-cLPC?22Xy+N!d*P8<Cj^<hk+v_3%?pO}K
zR4JRstIcpdkYF}DaW*1Vi2rJ)`h(k?Pv)Cp5TOBroyxUynx%ddJz%E;Q#wI9l(Klc
zLE=olO{guB5ib6}+1YAKU&SY&<Ltbl-_AJ>mD#Pp%k#-rw6d-quH6fMGL(_sZ9i4e
z$ri!MKzR5$kwIZ(&&jf1s@J5}deFqL6}=86)Vs$_4HP$<eIIu*B|`|Wli%}_?_D5r
za@CE$MF#<@xio%pq24*VljvpNm)m_llzDlMNG5g^(YhkO^Rc#+x|06gmHBc#fg{Wh
zeT?`%uAEZ0zC>zcefvHnLOK=JL*rNbe67nm?3(Y7Nc<!$WT9RAeUaGM618N%&VQj^
z!FIdeKd6~7(Qf#@NQqSG;L1=W!i@<YLaXF7iM9^jO+Lc?t@pLsNFQ4Tzff^UeeYr`
zDt*zeXjx=8e;>jvMO8|coQ=-qsdgYa3o@$3hq@ir9*gJ_8{<P0hN^P*V*?z|rE=BD
z!5LQ#CxN_2LBBQMm%CWLZz?z>me719G)xS1$hWr7CMBY#_1NTUCx<F_2%-Zv#Rnw0
z05R$I9B@8fX5yFcs6bORs!S%zZ&%b!Z7j<#cpZ-&Ha(ve_mI-?sE+O+p&?WB%4X&G
z6eHjO+w2Of3a6Euag)u-IpLdFkC1<4F_7tgZoUoRTY-k`607^(6_AwAv*h+m{4Vik
zUXJF5`C;z01>dJoi5n9kEe4kaRqk@`!bJ{z3-NnIIo%UO5fLaGKN|5aCc!^qi@y&Z
zBmRmPNymzoHnAJeqvQR}?ydep0=I^kO^I-X`4&S*qtmJSn)<<PjtRX=kBmv~+9Cko
z2b)!@d)o3G8msY6YfCHvpaLT?30k|8g6fl5XRe!2!KP;Giq{ofA;UlBJL~bobKAJz
z+sf)^rKri9ywg_YE-q+k5;t2@EvAHjNtmz);#mZ4va_$)+I*&BZ;T~-_PUFnNkw_x
z*88(T6^?0CWI5)Q2fshkP)_||-~Ye=vxgEtUOnw$KOZxv{%H;Y`2nqhMrFY!rQ06O
zXWV5{mDB-fLq3z!la=m-t~Y+~T7c-XSVf9Es6=mg0KW{=_E-{GsaW36GdyWUDtJ7x
zF1tQhe<I;rh!?xVo{TB|PFDu-EB2J{W5a49^J82$aQThk<31A~ij-wMRcuqA)62A3
zQ)z|c)HT?L4Ax_tt7)+1K)76$w#5mTHj-WS4GpmL{+LWMvS@0*q+e?NY_fs)+$~AY
zO18(_vgOKDQ{s6`1mKQ5SJLnAZ)s{Hlx&&pN`SZt@3Y<c7oE<C91C{f_NtI;OSGok
zMtf~vTno3}YWov^l<Y%xIFRV20E_%T2V=XW;piC2U!VvHnKbF8g2+u>e>;?4Ev1Mr
z_K-J{V=qvtK5h=GaOEp2sL4fpjKKiWO(wn{rIIg~!yydUM=A`L3XpKtP4>3$2X$hX
z+B?<nUejzV_S|=?F^7EH9v7-B_&IwoP^|N%tk(l2=|;Zq_mlnq%U}L#ALK^LdZ2uc
z4;}W7r7f5TnENA#1CMO2eH>$!YXsEnfG1;JCOq+zfVPc{-%oRUrmGdkXYoxj`|u6L
zErPm_CLC0XwPU$8%7$BQ$Mc?TY*IB}2<(K<peE3EGH1Yhf1jNS?6{=w2CvgY1O|nR
z!4HIC;f#;dpk$-7@s=MTtxDd>UEE>k5-sQw5J#h(&q+j6q6!$^1%JjNpgbm~oF?A;
zDYqUyM0w&r`*eR$pO;^fQp?4ptChH6S;IIf^a8VA9U+7o;m)>mA+;gmFLyK6CFfiq
z!NCTgu8drIE<`C9ClYgO(S0t97ljdJi>X1)<>y@F+cI=hGsLUPm^^nx($)A|$$P)@
zT-=adt;^_SGoqT2qL{i~apqo?gL5WS;F)|QyZT&+MNkdKQsk6te4x2h71aU0Ha|jZ
zP@I7H-gW1~ly|@Kalqa6=Zb{6qP5;`@Xt%>5`MWcfk!BKfpe&uQF2)CwVQGfq$6n`
zHqvK8xWP}@(Tf>@R7qdq%w8->H=hgdXsN}cokxx))=(kHI|xG=9gr5T$TPaf*)8!*
zWerrV5(AlyaR~1!MiU~-hK4PAbxDx*vV(<Tj>{oDRFGq^)oK!U6_->tz=RwGwx}$r
ze+TIj#1+nfQNboAZ18QFW@V+G6sRF^MD8bk+01&R@J~KhB;{x=I>rAQ?4v6{yWO5-
z14HfVZS5(=YwpOQq69%DH(G5@4E3uE?p~mu?gIiKx#2c52lQ?r>fzLFv*H!07Kd!M
z|0RbK!q=R0Dem_+V0N>)v3A))i1<T#@K~A`i@<fz>du8lYBoaUJg1buj{1!)@KOJ@
zw@H`5!hCB!a`>X8%0_i^De~O8V#Xcw&$c+$S}sIZq#)X&u-g(J5K%<h=e_Y>KPR4~
zd#*}V)7=Al8+@vcUM?zYOw?(gfY^;cctot(r~LDs){?N@g+9K#!YC~yf+4oWGkJ5B
zS&z(V?M}Lbea*er87T5lf7eXu*q|gUX{pwgke6tNp8RV%R!>`*w1$<BE2Z03#bn6Y
zD)i4S@uICsp2O0qvZD1rkT@ilx`)cg-L^Jgnz4#@k|hUlU9M~v0{;5^Lgwf$vF-;0
zIj(K`ZSR=?4+Rj2RdRyk5GNZlSx`fLBWL%K3eHzijGgoz^;bZAX;8JH8f71(<08(+
zc*cP!;uy4SkNLB~zV;?M6uOz{@qDi|FXhgEYrQ?;%c1PU^ONn#L}RVE$5R)PO&HlY
zhkNYl_{GgAAqL6)XV3U>zxn&oqkn6Cxep%H+i07Ta$sk&5la;*a+z)RnK3l1BNTUw
zpO=zl_aB%p`g5StZx9wckSDOQX(`53*C)1Ds43<?SkXS)Q7yUn$s~!|u6X8!1SG2M
zvE8{5Gn8O%d#in=^>Nn<_H|!w`)sOg?9ZP&Iy-Ty<Muh=KmX&%v19+<4*FpCwRyhU
zA)l|v`*<mfQR0n&L0Bzb8elkFk&c{Mm?=t@v;rUo;}?$&a*e)3&DfD#3$m4*QYxD8
zb~UynUc@`bekUPDhE>u%bxs}Q4Xs4HthV>6xji6mC=%fPYR|kHqk=+Ru>HVCXr!<J
zT;aoNRyB`vYt2XTtVHevJ%8-%>{gVoPpY52dk~Qv_GvYYoSm{U?Xu4*_PH7_&?-Bc
zdodMWdaS~Yu4t!HKNiQU;a#sM#@<fE+X}fLpkSD@=wvV)6jsh&(XPQ=X^|&?w4KV)
z6vM+sAbxi`o|RNdp3(fn&Nx;mW7>4*bURyt-c+inp37U0q#Vk7qDv}zTUUI!=<R&|
z&V49fucEt$*Hpd`GwN+A9qRb>Q=X2y=$uOXD*L&5NyeNKt!uv|sPG-N?Unuc*LYUK
z;drL(L;E4*t|GjsBE&ws<cE-W?xMy7c8ylMG~bfpf-hN&UG_to-g=<&;(NOMhmd+^
zco7@B;)l2hWj@C?cjXT*sKEp<k-FMe^duV*Gsq>SXnY0nT@}wVK2Z_3SN{-K<*3nr
zXV>Hlg9i)$lvncFoN|uaWUl)mEGp$Pk;k4uD8(XysQqw<>wh>R39%?&Hw1XaBwFZK
zyFrYZ-B<xWlzrk<cT@br-WG${%Ozk^5qKjV)IY26wT(&)D(92+5wUDH=bHJe=w#f9
zqjP<;(5-HX7l=ufHNO<K8x!!VTh-Vc05hC6nTnw{?uU?a+d9-<sV31uCFYfsu%1wT
za<B+fd&v2n=ySIAD0O#IyilrJ`1!KQKNR7$=2l|<bBigk%M2t*`UDiIcdmR#E?^$}
zT^oaO2jUkiN|jBI4JnM62^8EZ@-D3np~#tmP?ug*8%+3;5oVSzeJVk7Hrr{G0<*I0
z$?iKR>9MP+lcG>EJkPE6Q!2Z!$ak$SfXJQRg#~9AC$M?GDjWoo9=mP94@KTwQQt|b
zfh|lNK}Q;wx{0<Zp2>128^<fq;ysCca{XJ0id^h?Y$d>XN&a+qgL;S9z45v7Zp8>_
zH*O8_4Cab?=6Rf}+pICqCOuSCEOfOs$5Q0?kQq^KddVTUjiuRJON9hiuoc$lNLZdQ
zEbm=Ghb3!s(rKlUw&cxrUo(R-y>;fx^=%Ekgp71m!-{9=R+sKe^8@SgZNhloW-D_E
zC0+*W0E}Z*epxgD=WKkB-oNGt_ohhBATox`mvSXQ05U0}FNptfB8i)^2QI?52QRO6
z{N8=n3|kkI=tgnyvLyKWT*yjqD*m7^6QgJsi+;%e!rt9N;O@gQ38%I2J)+?3kqX~_
zX{w&xaF+dO%$$y}{YZa8&q3ReN8{Fs<7{IL<9;HKFf3T$eo!FeV==XkPD#7KlK8j-
zhh#B*?o@jsMsTStGN+12DD%!Xd(yX4MqH&5KNYXbNx#D(;tEsCYCm01SKFd0P<|nn
zq3xN(n7Ryc_WFDsGU8JjEI!);cLK!(V9DquTg+yMR{g56wQO;!09C}rC5mJTWGXM#
z@-kP{Kt(Si-~xng_vxpWmN(e!c2t0t*i`v}pbQLJwkvsG4^3LAWA6Z{Y9<7;Hvuhr
zcJa}(ekFG9OK=!O*#QT@k1D;=f%=EAK#=elZMD1YfKypVW4#?rZb+*+CB@sJ7^=kT
zFuVuqD}eWnd=;BVvkEsHPTDF+@H=%dpK+EYba0mKNVQy5doAH&95lt-y`8)OM_|)Z
z)iQb~pK$l#hl4A=n<PUD?09=G#+Rz0P<avB`#F9QK8D+DKFDuzpETQt@vUfCx7AAo
zIDS--po@tl<4OQ6h8bDjKlV+BU|a2zc$Xj`en%O2OL3z^!F`_w9O2#rbuX|^*%Cg>
za{~$3(Evj%ZVxD;?2H}rX)sT=+wlsd0l0@36)0{1rbKhK6ETgGJ}5@M)slAENuLI+
zL9VBUtyX4awPZS-*llg6D83AeXPgvPtnRk6@qvOzlrO1d`g2J>_|MXjV&~&UT@e^u
zSjG;!P%Wwi)fW@TPkHJjJ0WoY{6mD@#Gx6IpbG62vGNuB#gD*zzR~vo>RVU}B~Uu!
zybF7hY4)&cdPLbJ@yu&Kc_8nl{%o+Xqe135R!UEx^X{AMLtcK}E<f*rTm*p**=OvE
z7;lJ7Nd7$UGMID|l}1$ySzZ6Sb5$g8M0`^$K?SaTcGdYJ=+W(+^qi~DhaB0})T~wz
z*90)iVc%!h=I3(fh$N7$u8U{7oJB`aon3!EL?b$J%MN!wUFn5G74fCpZp@dKl6KZ^
zH^oenOLyDI7<CKhQ(iSmPovHk;i#u$mED|JfvqT^4BrE*kxDb#7tme&kX*I5oDWHP
zI<`nfx{b*-$Ws8k5+B9}3Wyq{TTr|pDG(7{c^6g_^9`@s<mQqI@r&OLC3==Y6{j&V
z_oBqxdX@ANUj}QDq&w++mNa#*<WRV*Y;t>Il*#AQC`##ZdJ^~8l=DT7p4ZD&bGxIY
z=JylqjtXo-*kIFK5>{J*Kp|-l3Jh`pqZn<c2hP>WaIeku!B%vvv{{aj)@Cf{HakX6
zj!36Vo0DH6FTQMZlWKa~DG}afb-A_Nhe*2fVgQdhLcm0~THv2xiP**e_u9hTr)7L1
zcg3PWtjiTehXQt7?siWFPiVJT4i`J@i~nV`Ejb_RQXv7A!n1qN7mMBec54_0isD*T
zp1(2H(vNVz#hR;mRRc>lz0&`I$l8|p){O<)%Kf(7pJ6DLj}+?`{@ym<pt*mVGsoKF
z8&z}N{3g{un6EQ-#m6r3ONB8?ZqoX?Di{Z3)kmP)Z9NVnRW31d+sgcFX&Z?ls{&%(
zC>ZKdtusb%O^{foIVqG{t__$%X>3!R@b*BpE*c5NU9~oz!Ca&cbCYbcbvaMmqw;CF
zNqN1m7=vH8J?Nx^REMSOp%{kir%*JdjO^k6G>ocqi|xN)kL26XzY9hrgZglTeGR?M
zwjnU*0j*9G!zIFv@w@I!VFI&zU{>oGu*a&E(*LUbvfUmJaAG%<7gNs29(y8xkxm4N
zO6_ww=1Hfgy1Dk0|EqD0JzW9+GIL8ve9@lCb3r>G#2V|1cSHxzl{|~xwkc;Rl;+e>
zLWga091&GP=Kr=OzE-5a{D?A&^gDX~`d5|R9Z1Tj@EzEflDXaCoBUhb5nsz>q{_0d
z;c(mK`!<t^6Wi^-5B4bp1>PbdYi~6pGzdNqqATr7p7K^zeod^mY@&F|{^}dm-F2pX
zAlMYCkOvb+q-Ba07f~+HA2P`E_J+eWX7QNCb~wh;6<34@wAR7IDOH{85eJS`^XZz?
zseInvjuF+toZ=dRVL^!is|*YNOR)uex1yKQXbaPjTHTap$lzzUy%%Fk$WXsb1qZ*M
zmzn0LO0KJ|AH*D`mqDSdjA#BZd0|(_YObG;V#rLn+J#{FarMlzg;e<S6Gy{f-^eNU
zY4!U~>{*_z_nB{eLn-QeG+t0L82f~zv)_*Sv&fikrQi;6Za!!$0rhrb2(G0BN}Y6+
zq3Eht#HrlmVyANnE+r7z>`o^G`eRd5r=6)5Uvf$~%gWqSvyHPp31LPwwQjd_LpuSd
z@xt#~&UQPWOC&f)y+#SpwhOt_&2lag-4?s545l6Svs3n{f3RN?RZahDzs3sNSGSq{
z7@=1P2CslNN;=vlKc=3c+K@y&>X2L-FX|4LMM2d@FZ<CYDfc9~DD4`w{`;(59^a@4
zvJ(Da4|c_mML?-vx!SJ$v0TzrC~mhAF_f~h)g3`*i(w8byDH&C{YTzEOaQ-K{bLcz
zxV#|oYvKj!R3NHLI-r=DPp}5v(5_4P`2`PT(Fw2ru}G1Febc0p-jIJ`q$Z9n-j{z&
zRu%Oax#`Cc+1}i6_}~f~nRrsE^w#hLjQTNdR#=4YX!hjh1PxPx3t|H{`o|Et6i(dS
zOL9TNSF(7-mUr41e+GQT<y=&#u|HNuOGO5wK5Iy2M2`D0q@Z5RLX`MiVKq6iTZS;k
z?6L`oXrlVE4byPeG|`bdgR0IH-iQG_!KhT|waM{J79P1lRU~9ePDGfFoQbbcq2TtI
z$c+guOb<UEcT|ALWHa3U=AvP3BD<OVi8h<=dpX#*pw(t3j*B&+U8=>b+`Q!-_tEIl
ziLB<f8k_TD5!S(=nq-o5@M95k{>ML9osT$f94!>)<z$>eqI?(lPj&Td%)&spf)gdX
zD>2NCK8t)#dUcbiXZHjST#BJ!5#sA&pPX>YR@}~Aw!|l|oon~{w`|?>h|UewrpkTc
zoSd29q>TXyZVqeqtr_gQ`%eq3CDBK|ZyHGavR78{E?XYpz?Wd?xH!4m+v1sgW~h-R
zT(aFKpvXd#b>{ETpe}fI@#{)lH`AaF4#{gh`C1Xd{q|bv!_2CuC&{YB7#vD9BxDXo
zw3pk{_vN-*C)fjtnz|0Tux+jHK-%ip<@uoWej8N`Y<(_%I^CfjtQN1zA+Ed$YgrNG
zp@6W2s5oue!|_avnj5U-<~{ObC|&p|DT*3O*Q3?f3RCfvxD7e|tlkFO=!;!2gJg@x
zs+V`lsH|`(d%W7t(y5#)J53Z%_?K{tCVMizlv1dZti;-<;+YmBL$pNir{kIY1X9MN
zH$4;Nql@@i>vObDxNV|sO1$;7zygR2*qng3!4S5@XAFfGyMJi1V((T+=YS72b=oxB
z{$pf85oOZbOne%9kL<c5&er(f+peSy7~L{bi%}(a=L|?b1GYCeS^etaC;NQ)Z8DAQ
z_kW}NwA+CM^6zh-Vh3{xxV%*f$PVQ+WUzI2w&*r`BS6>ns9Lw%P^b?(1tCUuO2ZiR
zvo`tf?1;le-j)I~-u8KaGiKB%dnce(9Ke9R8-u!ogQBTO2=C<zNV;f+v%Fu?wp#un
z=@v_@f<Qj-H7r#)oqg!A;Q3bfGW#e--b$?}X)ChxeVnu>bBt0jN@U0<Ip19bLjmol
z7m?C%r@&{vjyL}&si;T&-wOJ+W4YR8D&2PcqCJ+4px;jT9P@DsB3dHWII#EiHc*eC
z*G?s{+}iSx%T9GV@0N0_Va~BLxhbRu$|{2-d^SETfu_0{pr>>3Ol;_H5INyLcHZaX
zax8@{FT`Qi!Nl2CLV4sYQ0k{>{ihh>KTr<UCQCl~pM7&@-EF`4(y1iZXutZvx-WYv
zcxe~hNCbz%6?MslG&jjTC;9IcyYxb&I1N=H*1zmB|B_e`)n&Up#*y5*&#s7ZIHHnN
z;L<PImGOo;A`RGx3w}k*N}~icv>fK!iFM|x_*Og-Ih?5Ae!KcYSd-*Q!bf(^1sAH+
ziQpy1SG6O*9m_3j*XHP4{nA9GcZmo0+jTLzWN<MY)i1jKLU6doE-KNMbQeNBZn#h`
z!QZU48!v<hbAD5s-E_ePG990hF^xFUt^lC8$81yre3z;h6X#XE2?HsJm0d=;Rik}8
z<nzC?TLQQ$!4YmZ=0cGLr_+sKtiyr4+0ks{swJxTp$i`pH1=6dwM6BYI(<xtx4rj-
zl*7#RnHX<-tVeNy*ra&dyBm}+lo)4nj!lYjmWi5jp-6mmyeKZ^+v61_IZ1~YK-an>
zpB05%($4c%8_$a2mG@LwetKfMX2Lj|>6lqsPw$Lb75-#I_1<{Cve`)rh+2E#a)&Up
zpuQj?c+CEsoI1*C=(Nm@XOh5`&wT*7wa%ZB$-io;{HhX|=EZy5SBzJbZCj9dFH0Go
zEV0n>fTGkQWoA(|@$N!t{MYUYU`s`Gz}&%k_aS6`*lUZEY{c(q5V$1I*dN5F?u`*u
zg;>6Z-7xfqd}E$Up*1E5t5&oj&o?{mbRhb-<j6!JDj06L|5P(!g0(r)Y2ksBu-%_`
zbu?IKQf7OnbybT|S%%_}vYvQ`3W95*{MT0IG<3G;VwzPK!Zul=DMWYp9RG8<F<X;3
zEdhP*Cn&`T?15@t5A{2Mb!TlZl4}t>ugi~%ke70f&ieZ5W3`5J8Rfz1ZPiICVqlQ=
zLw?@>+i&gRe5XVndQm@?jj#Wa;9;thc*q{D5b%O__%MM4wjmKf3WWj*H(m%;2mMEh
zR#4Srj#u)E=v;jvt+LUQ-JYlheX+QOUxY_I832hHs)Uo6C3z{F^{0m3F4Cy#dp(`t
zzqPL1P425x-o!I`yrv_+_xT?^9jZUNshUBp%7#ZaZO#P|swdHAOF}#q+Ymt{tE@lY
z=;|f-zRL#kjRh0~?^F=Uc3<(Gb7$UdJMwMrbF~xS^?#A~=HYo(Rlo4+C?X;vA|fJI
zgo34(wv<v4={(TMCbSG?c#=FB+9XfNleA44go22O2#AP?n9k`ur)`?HSke{|5fKp)
z5fOP25fPEj?fL!Iu#>*uxxVi@-?`3Thf8zyx$k}NVePfoUh}fu91VwMRtj1huaQcZ
zVNdG=eudPF(oH~`ea)0&*L#HeCZ9izRLq5{inbjK0J|1%y#%o-^yJh6LH>+a40GTk
zYDBgLR*kHO3?z^HNj2w5Bb%DU5tlC<dt`?#33k#tX^RPg#8A7l3b?n+-am0W+)AGG
z$(dS!dI~zl(%SZvwZe6JlRO=h!uArIT5)GS6VD8{!Q9T3XYCtgM=g}+3?J*Oki)!0
zp7%K;_aH?Va1-RCEW;3aiWfd`c^Y2A8q@ir9rj9W(Myimbc)$7F9)3hf5Pw)YQ%W4
zxiBoR_`F!Li(ymrT6i_)3W`r3QWS8I*J3xx06~Cx-P@Qs3wHy!pEqJGb%8dH!R~Uq
zO5cocAdOHRv<z^2D~-hjV(e|Dyq%!T^rtGAp*6|<J24CRUt~)f<Xt=IRnz4?|7P({
z>L9tZ^gJMs|Gtl)YXdCTKyUg%;Ns|Nq)cZVr|eCGvciX{c8$^4pWaXrOb&s6+4VtR
zFg=@K%EzD#<6>qpvYuPzux~2Ofv^DXQI?FUDC(HVgGn(u$20oy%(A<x>mFs1igS6`
zcCuB{-P)i}hK%ZsFURaLri(~Tbway}`dDwx5-H08wyn?%9hZ8*O!17T<oG>d7i{4s
zL!C)(6cx_#Z>(ERuvv`PIbo(J?s1fZ{t610dO2y2vNLCW^(;AgkF#BxRgpFi`e~lB
zM;THCYf~sVb&v8E6SKADaO9EG_C$O^MJ;?@Y~`n?RjK7mQI?WolbqqRjTrUykPQI|
zjjLFKL)yc}8mtCWIdhNG;;LK`ger()XE_AS0mDpPJ$<w&#Hprr)e@ZRZxrdUqYTL>
z3O7qH>lrXW)*!e)A)VPC<sxxuBkgOz87fPr*<}1y-=q4X*@PBEh*Q%4=r<kx@%;_g
z-i(o%^@b#8nGB3wK<31HN+E;x_^ni%ol%ixbso9cV1cq%;1A@IXpqll8S?~w_|o*K
zzPK2{C0NA}TLrR^A-!3G)dJO}AjyY}h#3L9=}e}nv@|)|s6Eb7k1HN2V*;s&Vt$CI
zfu@aVE0n}zy=x}&;(^&fOJRJ&dl|^bxD`wIh<}6ZPxPT3$-#Iqv(YN}BTh=aX+{|h
zPb{=&ZH;4Ky)IDcSSEWrb2x+67q|Nr-YYg2a*fg_%$MP)SH*%R)YQmS!`imSC!hlz
zC)e7cfNifHF<F&zz5ULWG$6?{ER;pwt04{#+D4)ej#G)Rl3_q352_>qf>U|B1|nXm
z&(U%|!IA@|R(Y%(LL#BL8Z3j+!D-Q3d)`v;pqtd%!OR(z2{M+khqYy@4J{CGl`X21
znm{Dlq3EMr8?Q%?FPtEC1_WfYhFxHWy#mymVH2Je4dJp?F0=g|RmRKYCZh@c&c8rz
z383Q?(B%XHd2*}0&XiZnoLCz@2GBGAnm+yxLq&#{$enSJAt@14^KzH3ya_T1g#;~5
z00<+mB8FHmckfYIZ`_q~PwW+vc#_$;H&FcQGv&U3Htrzmyl4k_AXWlqYmq$YEof7t
z+{*ydhuUXD7Y`h@hXWSSf)vRkX-#y0p?eHVziTE8YlrHRrC_rj^S<pWQTvcICUCK4
z01i9Gn-_D?yZ}Z37FbX9Ow0-vE?NKrz$9=YtEDORL{%y0zfP708YA)%yz25G05Er;
zo-A3}l}5tVsUCjy_CHx2SW*sM2^YN9hR<96hV}NVt#K?EKxvMFVYDDn;aM-6V!JR6
z=4DtMP9kKBpfYapZy-ADgi8`}MOyut69K*Ww>fA{#cAetD%sb_<1vSEnd)lUVGBj1
zeSIZt5p6LgWf|@4>@w6wr4fX>K50l%k<Y<_D|$jz6QsUKGXp0M1egz2@flwYZe7gQ
z-1I`mp%(ht^vwz=UCCrN%X6_BAQ>q?GW^T)ftp3gVGCCI1z#OuLQVA=XohUEFIr2}
ztU(j_QnUnXMm>dUwsqTIPUDaxfRi(nH?MeWn=kEBB)TjT_-dLdMOrf)zviu}(4wb2
zq+};vuLt_T9~~S0h8^QRknG=#QI&X;fX<h<{27TH{Slcj?d>2~;M$q{ggM<hLhhY3
zQ=|#8Q5^#Bc6g(vGKAe&Hm>(#R<MhJ_P!se1{2m?G@;wESzwG=`@ld6rP?Ma)#l4y
z+iuw07eX7l-`CXd0y${^%vH~p!#=p)444o$*%u^xg8`$*q6u<#$948M41z@V3o-$C
zk2pu^x>p%MJX_6JW*muy(%qlg-DI)W$L#ehb;RgGId-oSw6YpGZm)4{7Lf<}2+Jg{
zAXb*92|0eRvjVgeF@Ir+oZ!71r6KdQO-}Ur`H0}$!9arMt(B8vE?@u=Fadv|)pBxl
z3ffhSq;{YIr|dOFBRg0jr+Rn9AxdQ|JHS(ln%&-=rDxajqjLIQ=OWn~G}m&*Q8U!=
zJn6C5b%X^o!JJgfnen>_#o(~2vm6lQ6`*w?y;hh~R$%NxPlF#URZxqbYw(*kO?vrI
zK2t{VNbkMMl@2a}Y6=IjPlABOTF}tgHy{8?nJLOpl<l9I;e=yQ66*mr&bW*Wj1N#N
zwnEjuGRQwL^As7^HF9w@v-CEE>?QF?YZk%)feMvJ@0M{4$`TI=sDK&8&d4y|7Qzso
zTedJ%HiuIJic{`&j@Wg|3uH`S6QmYR6{#!+bu(sB8yg*rs{m^w<Kri+EYO!uv;~zT
z?I-wayFw<VMY8^z;Xt83Gj}7U2r@aAgvDr(zDiZ(53cYzQIcY!WCpv^=Ie@IxyrvZ
z_6l&<ap;0E)u3NjBiF`CiyKe~fi}?V?cMm%Qsm#$jzsqz*r2$3U|dQJWz?=Cr9NpY
zQt3a!u-tP0)J_sb0(#6W2DzK&t$BHIqJ)0sUYBmLg}7&McR~WH-n)y@E;+io*BsRy
z?e|oEQ^c*c?kFXKF*C^4`PME{oAwO<mQZ4Wu=b*Ivuy1!UPUqc*#UnD_VoCi4_Mt4
z&vtA8!%azdiQMAP5OfQeGF#<Jxz(SUI~D7WY>_$f40i<CFZ`7|(pt=Un+y!)$;EP~
z&AY8}6g|K0vi0}w-Mg3EZU4UcM))Y*vsVSlLZIorwhh2p_@CStJ2@Gu8RRR-$OAz=
zCS{>Of^hSob<@^}2)BoPYHo)BtwSz<pBmLen=mPOiDARxC@+uLmk>wu@@TXIC|LT0
zVM?|ckHvt4i$P5w(>B_QO0wk*P%X_1LOl~`wjdTe0(Kr*<g)?eOI85a-co6@<^?eH
zRmxItpedH^#*^hfkwq|OY%#W!JQJc4SH?mI50=$2&`h{)$lA1QszG@snq>XnUD8S9
zXfx0MeZQnRVP<&JqHYm&C9)}CNS7aJaTuzm8W0B;Capn>q0^wY37fyoyCX$_FClJD
zC@s~s=5eEr>c2O?r|gLS)K|*fC))fOg(#}@z$449z#p2fpG*zNaOzDVPwkDM9Ogu|
z1RuP`@o#7lp~C|U>|x>Zv>l_WTwb1u)s^O8yL{H@-=It7IUhm=1&}X_iqFSAK+Mr|
zOYxKr!=&Kz<ORb#FxOdmu>(9eD2v4iZ{GDU#aV@slf{i|<>k0AaJRx5@k*=?VML7Z
zU<uV%;~A7?Je>%(k=JY$6G1Ouw~^~B!6v-nP3h=GpsxjWdDEYf4ze~l>wUhJo~=^V
zA%&l}t;^m$P%O$9OQKIF1ZHBP3T*>MUx$#@cVp5sGnJq~@BL4jcLe$SHbX>eqU@%_
zhAK>E`+&n8Pc^QNz5d4hRq#f&gnYBV9iFO$phZllBL`y#5o1~>hkaGtH*&vQz`n{p
zV<xF!1%bg0(j=YtxrY~WRMbn?eeS<CBunO^yMMqNjP7Zy^mfd?(85uvryhZMId)&-
z(QtXyBVX>mu&TBvh|SCK-k;jL&>w@yPVi?`B(4SlV=lZC_bGWFFkpb3v`=q9Xj7KT
z$@`SaQ+E75XqZlk^=O<K@8;EVYU+-)UP0KQLOLzA+W@T`%CbUXe7X%Wdi-T_Mtq`-
zs0D8Kh-cI{fkj1CB01Ba88pZ<&6TqPVD-3q8>HtxWy6vD12@yT`<%H>WK?>^BNXdp
zrY`Vcy<>8~EQ<u(Bz<hrJ<d7jE7CUrF!6HfZ~rc;MG=w#28_1G+DtL`YT!O&^-a(<
zBaXQWSem3kJ`S!1>?O@`bL3*%HX3u_=nJg5Bsx*foJVdAwxnGKymX(5FeV64L?PbN
zxnsya=N3`xr%aPK5A(jjV1vsy-$#pIP))gv2ozx+9i^6B#wZ`b_%$m|M@`UH851~`
z@dO)V(;ktjAsf;p<D(J4tJ;T&9X3sUh_I29qETrMKFZ*j3k?~vj1`RyEe4;8yd4H{
zNHKMCRlGv7I|R8*^`5Dj-3Ux$D(!32Xw?52Lva1R|966H@G&u_C^mcOa4Cu5@-wM`
zDNVyM8J_7K%l#Rp8fuT|jy5e8hnzDNv!SM`jGbW!DoTlgt4y<`Ll}+pTcpOHb%@-Q
z+Q8OQ+B;Cn5uR<nspfTSsSeB=h=i+EW~5n?QMX9NSw><&Zs$S0Gdt!CwKl*8Z|Svi
zQxGBST2^j}_H9&r+!|m|`;E*=^EB%ngNN?$4~SiB<W7SjEedj1d_0;rvfORQlitjg
za*y{!trKe#-iC?X>&qk6qy`f1{@BIUkTz6GmHVQ<YUCd&kq7pLftxA{Rao*PTJW$s
z-uEQ24+czd5d~5_WHTwiCe_Kq_5%rufQyerYw(|{$j24+XdDXcGl?2&z3N|i%v&?X
z0aKZhg>H-<aBZ*}7SpNBiykO2D8ITHJut1|g4B8y=r5KD5H1QhRU+|HlQ-Z7(@h6)
ze6rNL^WZ-X1K};UN|wi`6Ef3frGM&<CkXprFRKF`p`gmuHNcU0ZOi~LgStyqQ|p7e
zY3km*GzW$Zt*xIen+(TYXu(Fb_<W4rX@dD+uAjcv*fAzntCekmzs^#ziV2~{;1R|h
zJRalVl4+J5-mPAhfDNt=K#eNIwlp9ka(VJdT*yf8N^7=oXdE$l(tv`5Sdd7b3c9C4
zIE+GY*DH~yy~7THD27V%j7>@1<{ISLnBce!#E|D~u(rmOq&hFp2MrlESYAleq+D#*
zu2H2I1HdSXxdqCsmx63CqJk0oFZ(*66vU@ALT!;(Vr}?@G>GZRUQIA!yemliPAFv|
zjI_#Yw(4m)7%$~@Z$^`<TC+4YALEYSux14{@bHs2eFQgpX)d9+Y(Xl{z3uM-Yt=#B
zhP`;l-k)<#&o9Wk0iK~Oiuqow0sLBo!o8nBLIWt7YT_I;TJnLnHYF{c?Uu{lm^N5e
z>WWwqmNGuHf7Tk>k_v<#BHt=7`C!ZrghRP~FeHc5989-2AMQr7-xx7EZqaqV8F1{J
zo}pwJw;n8V*Zt0%L5GBpPWy$bqtF1$G5dE>HIQMNyJPp;wX9+FnB^9nWZUNBq6aOU
ze0+?7)ocbP<I<87{9E|M!DL7kVSIa{jZj((^gL<5Gk8N`#-w0#@_xVXq2yzFWW*2b
zSEg<B(97i1{q7ix$XlANp{~jSTThG03`Y80IX!x)MTnj1$j<QTF;fc<6@#C8#K0NY
z9hnmUOmAc%+ZIi8)_!MO;U@wayXSu8!k7&O8F2Tz3KVpiT36;<=;$KSYriT>$4-{s
z_OH6>(x-iDjK#zH`+6^0174M<^cbMXL$bF<`o}bfq9hT`d;_ervoDc>KH<f<>R8&6
zX><m8cc`WSlBYElz>+b~j0OY9#jzlifrQ2E5(8~T$t=0ldxH)D$5K@VGGxCiKg*zY
z0}nMUM#S|)ip@&rBMkPK?s~X*j7mV!bR8Jk(2!mXRA7$T?{1&@`n=|QcANOvm{(Pn
zWy8iBpy7>De3?v)ca=HHrjvXU^rU90&J8eM7y7U&Mgo3t-o}71)?-?<r%m=r(k{1A
zu1GzQ2Pd1gNUn+zOOW1cYBT_3Db20I{c~+>9jGg!{BeI>@2x0Bsh$~btwsBtHD_Wl
z^Pee6AW8Buum!Wj48ccBX$%I-8lsNWJH9-29UdXb)&j>mKTO8GQs#tBUlF7-eZGZB
zz%(;Zb({y}B`lGe*hQ+W>)~~`-<S)PmcdZW0o&J+mj;P|_*9qrWr{qWnRzn9pV4-P
zWGsNoJeg(B5ZVvbZ+&eo&+zORF`t8yyiRV~ADn7?-aD`vAnVjPL2V9%2m3*lm)zoQ
z)g=ex!Z3j?+NzQ^^4bv3VN6jkWx3Tst9mI_z60lRc~D<KeZrgoBLefGQg4tu9E|N8
z6<pp_33F%c|3v7q?(%*~EJ2)M=7(-UeN%PTa<{`l&z!^N9wTbkoOku5a<84yVTE#E
zU=4IT$Swyq>;c;w5H1>u9!xzkZ;IZd@=zcQIU=GbAYfOAvViiiod?K*>TdOjcL9t-
zLz=S|$)l;;Eo$OI<g7eqXMMu>LTQXo03~X@X%?OrM+6XGCkw1`zrGj9qSTm{7%Umu
zlBPf~q-xH~QfoXEjPUY!F@>17@+fe#XeHghnL=T8bPB4_0?C=~*lP`I;6~N}y-ilu
z$GJxV6R4@14XCh~Z`$8cQ&5Xk%CG{o_*lrE1`dEE6c>?R^Q}R1VCTtIL3`$A=51*X
zys9a3%$3LO88qG~s|(j6Mm+3D*wF6Zv=B4Kse!p?o3te?q`QdaPtZj37#~S?WS4<w
z+=OBBWE^8?<|EmdiMdZjC(1{2qbWeG0s8TnZI`F_yWV3cmXNE4<W`9~0nhjnEFlh2
zWvM(HV`xt}U|PWlnF{1N-(p0P<A6V(el=r~FZhNJ93U_Lx2c$?C*4{|G1kjV)-;cV
zcrW|-HedWjzptcCr&GEq#0+s?O|!KG0*tbHEf8iLqBrFA_yx|VM&7W2Dv&z&&A^3%
zx=d-QycIvGGHsBz<FgXb8+pf$^5uEz)87q(%&kYX6E6Dq?4-6eP981q+Yq_R>GFZk
zZXtvY@?vzV+iP8+78nHO{V`}63lYn~KsfMtEx2Au-Ki{JEQe!EUIONwD+iQ^nF<p_
z=L2DRwG6L}D(-r~c?ld27^A6N!ryd-O7{aU4?<Mpdo^4GFE(PHObxHRv>bClxjIeE
zhiX$k4qMP4qY0>bTzW=S=9^V=`~l~25Di(IkrR9>p!hiMhFBd1nG<agRWC1<lMXoh
zR#Gt|R|`J-<O9wzGJ+YS!iN%2*@B%u#otiuW$HDH;z14pIn`%nQa`!Y4RV@2YimRf
z0pwxPaFx>!C`UDMgq(4}ui3no$f_u-mL3O`H=D><UODrC@~}fPNS%C^FUU--=Jugx
zMc^jHIq4Zo7@w`s?7HVh=Wv$?%|_WC>E$!Dgnx9vSS`H|D07sLSe`yMCIVoPZOzBD
zaa4Z8u5D!C1SD;6b+I~7Ns<|sKp2p|p(Je{W@N)WwX+O7;8YI_MD3?~Rn85v-Dqn>
zNz04m;+Q#(1>A4Tz+YbCGv_&+i(Z+&@KWDGtRHgv5dVc}beQ~x1vWr;SDTX&zG>kH
zLZ<`D>rFB$z&5TvUs(aI`WPR{94bt$F*bS|lOfAIK6-=ZjvQfoq3wzOF;J4>IU0R8
zseNKQ)=8o5Z(AeqOeQB>vOp=1KLlHU#Q_%|$fni9F;}kguE`{X1Txi*W8Vu0$hGkS
zT<NLW0bl1IsLB{~VL6E+`=AgW?j^DCVkEJq3reXUPESt=ZUKP)e;m$oLl3aD6*A4I
zPhKWGW{u;n^s$-hjhrV-=2aV_Sqg42z%`C2mVcFk2ah)k8Jb$GHu_N4XS<{7qL0fh
z(riWoijo>A#OBB>hd0DGI9trN7FvT6ne@QdAyKaFK|wi0MY$!K!`p6~+?tj|Ifz(J
z^T9Sy!#Ob>+<8>$GH2u+{*0P`CNx|RuKdoJ68dWfMMF+oxyyhyY;fUVx!cB@X}K!y
z@z$haP`%@#ZIXKf+u#)ruJgV$2(wEohbs^Gq)B9!$%Ebk{`(9ng4$|@JQSV4jSiRp
z!#0D-<Q^Zfe<R&@gFI@WK*~PcD31knnf?TqPGc+u@&ijiAs3p=ODN0ZjpGQRm|$8c
z3#_XQ2whedB{*ux0Z#2EqXp7rD2NQjvef!#{%@#WGo^GFS>C={fL?7=vNCWRMr<U3
z1R+~($V;s&<Y#=g*2e>yD?3iH(dyi<k432AkeZCG((KQWdI<ar{=y8uDF&wg3i}Js
zP-zK5lxM(-hNrfsfoQO_;yK3pYzv5^nkUfvac@YJ)`$hAfFwCP3<CYm?<;L~71W`E
zR7{hpov??r%RK4NC?QdSPX*jl{><F4Kzpz!Pshm6LBkLH89!J&mK$)Gn4X*Uh_?V6
z!4Jr@vCVn7je}4;7hP)YQ#vKAlIQ&yOc<ax=C|~MZ7>pL0xMti<uN^4B$=a9YN5Pj
z%}OC$=Hz9=HR#6c<dxK?avI<ZmdLB#4bU1c4mM-c%)DmZ;O2z3;B`k?<R6ze4ENg_
zkrNIIG4GU$+09t&8XV%c;+gs0s}IQAeqt7ZDlCw9yag>dEe`E>11zPLIXWM_=WQU4
zfQ^Br1((eG-V10dj`aue1+I1IUSzL-u>^`Ex~(8YQ})M5YG>rL^>Q$lMnUTE0q2N8
z>8mw684YD}(6}qYA6hsZy%y<w(6wFTs1)ydaMy{@Z*j?S=rJUtZ@R}D5MMo_s=cKt
z_ZWXeMksQ$7_^VIzHyY{%OS@dbZ@z#1qE_^{6O4-oN!Qim+9qlqW5K{F69Ss99SFS
zB!3h6gXmUiqnsRFF$3Ev2bEz%>I#<PkW<r&;fD@l!(2S4`7;ylRnV8y?K2cSL6Nz<
zoN>@OH)Qq&)hdx52VL1yo|(<Kz%v7C(EY&AEt(roo0p!^9mKxubUF8+b0>6$K`B?w
zRNpJQ7z@pf^tNA_Dui>PAZG$L<2+0@G|r;$LBGPM)mOuN2z*2TgUUT45=;i9St2kL
z$Tl#h0Xyjp^)krDsK}$@ZB{NmXlylsY|eQ|76ox1Iw@R6Epmy!r`?*NNDM3Hg+kG5
zEe`Xg9k$)xDEEUehXgnZRHn$wFq>k1cD9T-xQhzZNjdoOj`EE#pEF)`i)2h}W@V;K
z#>S6=>?|1{D~HaPqGw(v+H2_Q@MBV74zQZSEtZApk}?*?enWO)O6k?OPbWvuIRukk
zanKp&8bpd9x*tHjD&9}ZOp~dxQxhq8T^pM@+zV||`FFjKv6zlu0OJ<8t`ym-`}V&`
zO47zaRsuSe#{Q~t!_3ASnkwbq5rQ$GkT$dUv{;(vzn~zCBF-Rf+odvQhM+eD4#X%$
z$W#w%S`(9Uv6;ZAHTbtRo_)5|#k!%}hGvTEp5a50*r0$R=sQ?2Kn&3Y(HTxT19;)y
zA+!A%k%LM;I%?h&3xLLsPH)IqF1Oe-WI8~BXPhW{LEake$N&)WaK`5ZN@Ag0{7Ti&
zv}*(Jh~-YnA=l=e-UB1qc?;1g@2&s=69V8$zdL@H!O{*1?4Gnf+KF^!N>bO!y)nmJ
zadCZ(x%1s;UqA6ZD1z<<);wT4ikuwsU?2!Hk31CL4hO^2U<FSej*-EJQ~^I93G}VZ
zpe6}%=Fw=8$s_YbjXY*sgBmWgq%pRs;7Xb2>xAwdd%~SrEeq^h=wPbz@Pd~`0U1bj
zFdhbafjeQ`3KIP5-=(tD`vNnJHNn*vrl944(1;z)(Se<LjaPa@3*<K=+iLGtL5+{s
zf<GsUtdX@bJU*_L_3<rkPcXG0x{IVap5dkpd*-Gz8LP?1^64!x8AvCPY>AY4X$@%f
z87SLqDBMsiTkv=c3=o&gj-YpS^+f=An{C(V;lpHCz#8}A3=oqigG4dJueLNRPx)N@
zYDC!5TzNWyky?M}jc^$@%QLYWT$6eh;Uo5}VFdD410$YG4Pd7Lb|b~DJa5yPfwX@w
z_&>97)$&114hA??jZ2I2OK}>M$E0P<GA_VNT~%u@2S$}32Pc#H$}6!d#+cEq{MDF-
zix&(#c`ZF7b3~0?le}(VL^c$fUEYXp;Q0-{yG-7Uk>JcQL-nd}1*Cz1@V|OHPQ&^A
zFO+xe^NLa+rdA_*H!YEQ6FS&h<-HgQ_D~e6zFOY*$x%e4`xvX5eBjTPfGS|Z;0D&n
zUWX0Bv*|1YYdfNV(FH9V%-FyG;I2>8P6yD_TSgsk@VAv7SlF|Taxl$_j08Ai&9+L9
z!^7#`TT739gxj)Hs|>XgI(<20_om&2%wV(}YoedBDa$(_^2=Tirv|p=rRyQ%yeuTZ
zXjS(^rnl-)v<m}iYKf%XS}DgI+QkdAh}xQC4<%j=$Lu)!#VG<?j*k&kX{-~L6Z{$P
zK^{8fD3cTYS@K<g1I<Z?jO&ArNyUWr^!z9L8*@1@>z`rsDc+B(!su=!r^fsW$V;B)
zEi{e~qc6dcI^BV1!Uwhj8$zQ5-8~_Q${GHR6`^3J%dOJGo`Gl2)z;Qq(akf{jF-_T
z5NZYT`MluHI;8AS1ro^gOjFb-7F=U%<lIBf$YqLZ$#?bgO>kwD+Fk4rn%xDvyz_SH
z?L848S>6?uE2YmN=Mi)DbttflGMzpr{RHb}#!m9i&C)-b(qUjGRKo*&R04ojj1)%%
zOoBfLkbDskf8Zfym@_3MVAlqD@8;t|vkQVs)Ti}ghbe7!MAaa=N_yd5bBVuK2xBm_
zw6%L*T$<)hdkD!K#X=s1Apv=0=4UK1leY~!q`dkyH_HgaOytb&C!-R87y-|SnawiB
z$F*z@)OJ#xvkW?1vU<y8Y+y@n2Ha4`dk>H-sgsAE0|`Q+g%*;Ybja=7(6?woCzxG9
ztqT3a#W?<M5t7LPEIm$i#Dx^6R<A4k(6H(eY7fMqZWM(RfJ}|enu(-gq<+`N1LGl}
z1M`J-aP(7#2=!@T09q+UHb!|?3$&K_7<3e7o(GTsDGe+n`NSH^z0E?9GcKO2%(O$U
zdWOonv`#8hH!H#Z84;`YXVl)&FPHHTHMTzmL!{QGF~LquU91A~`E)L8hJQ>^ehIe(
z^P1%=Guno~IG7#j^=yYLv(!*d&=6pq?M*i9EM!beYk@00&I(;|ZcWd$d=?5f+%9tx
z`t+AO?B6{ZWpHOq6!=Dal-%VXyY3XxPK)L4XjFpofXdjV5tacw2Elv%8D(u|Z31e`
zeF4?M&;VD;1L-qMh6e9qdC(>^u@VRMAz!fR$AQB;8BWN2v=2KXvC`!{c_il36E_*l
z(mfh@wqp^9^J7MoE-x4^jd4AI6M=st_#3dP^L!d$mlfjkpmAJaUDBc5xY0$ioSn>^
zpxmG|1tPJ`HQei#2I7P1a<49I?FNvm<u)M{7@v_XmX)z?DB(6cxK^jlrH_uG05>1T
z$3!Hlp6fuOfQQgn;=D84;(CJ{n22g=j(Nf5mian1#cz5DfnRNj@#*-0URZ#%)yN$<
z(BcN!79ZrxRd4dR{f3dUT5ZES5~hYB4SHn|MIM8dIK^!-fH~dZ&Yv&4{24cJ3rP|9
z%_q|{I6q)%^ksa?_YS%-?A_De!rWKPLFpNrYOaO{=~-L%%{SlNQ=aq2JRq&$NaiEB
z^Z7$AvLina&O7ulFT^gXRhkr;eoLr9Brn>~!w`m)E0dSJH(Iah14qBcm;KvNkx?Kw
z^Xw~W|JGrbz>p{<rq%ORL%)6l`^#(5CGIeUFVho7UO(jWFKXtMWe|gn$Wc2c)F*R^
zZ~9k&92f*%B%9@}L+vUi)&nyneLEJ5f~v4QVXNOUSWPL1<F&jSLqWv~6F=Qq-m~{@
zjT1o#-#1v6Qu_bEUX;M_RxEqlXFzWV(}Km9Gb$?OV1g;{8JeI8G5nRoc92FuxX)?c
z1v#uNLM7Y+J0CXA$GC6WfmmXWNn*Pm4qcf#QiFKUmF|ZfP^^k#`6`Y%tamgVmS9{x
zHhr_5ksn|h*UE9$09l}ypq%DmWBHivgRXImH16Pmgb6dnzzK(y&ExI>eNH@V{3Z?=
zE$ATf8bVL8!W3gMDT*>M)4)=Ea?Aq_j0&uiQ~X&IWCtYC#0|VvPCdM9DtODb|AXa~
z<+Q`fze22>EvNe&xO_-4s#iE8&6N(nru6O+OQQ1`oTq^}XZl8g_W}u~hII`{6wT0Q
z*>3jj-%onlSgcDU=f--?f0Tln^g0{{S$e?&-FkbkCeZ>3DCdzRNgwN~t#Rs?r0-$p
z8K-E46X|d7G`-FMABJ+WrX`0f!N4@M-mD;z^JGvAJv39JwG1wfPRlDShg<4s4VGpK
z<x73*p=ZY_UkEma_Tpw*!J*$mJ~vzXgh>T5Easbsj?GXGCn|#yoYtnIv|7d-c0x3v
zvH=mV$R#m04L}128AwK`P?oV9bi7ZBthJS0ZIOxA`%GkYtd>dMK=n8Hh!vR%4=aaQ
z1xsX1qz+f@6|t%GG@xr@gIpDBrio3ZI0T4Bv$45)Xeu++8M)Ta1KcE$5ajwaC%Syo
zk9@5Z*^pDeASKZyb>h6lz-g5FvtTW$^e8{<%w!1*M5!|i3Ei|fdU|109uBK=RfZ-a
zRgqPDZ_ZUxCnq)b3@+|qTB^&S@~M_uZ%5c^ED$JdU7TgA-m#U8i|y#{HpAa{qzPue
zpl49C%nIP-aWE{q<?N$S7=vlYEckL$Kz{_phg)Lrk>idWMWx(oORa|qKU*bpf)FB3
z=V+739fm+vs10$aLx*z5Jh>}A!7|9%C-*hEJ8&E2qUsQrbdSv>rv{sQV+~bEZHpqK
zIEMEnP}%iDKamG)OD7MO2M;?V%xq}Y@=(AZBy0=l;ap7RVV@Z8Z(ztHM|DS<x<DR{
zsqcW@SRS)Aopa9Dq%jT->`L?Ekg`ArBViW=)DSxjr87`=ku9Y{S?H!%X91LUu)Hsg
znO4-|;*{lSrW<H!Q*yU5daQ%@ZA}hh{OYuUEII}yDs>dp@yc48GtxBIK|Q$Mdvi4h
zZi7~7wsn*v0nR4-CrVjLON<K@8OVm_9B8$^K)~ouvCS3@@K?*@?a<QmgwroOQV*=t
zLyyYk(iV%u)s1U?mp4)46PmE<*OM{fA_DIzM`p_MjB!|@emhkEQ7N-oo{k>CFf&e)
zH|#Thkk654ZHj;p9PXpv46C$1A5e$a5_5~a5d8qaTkSZ>i)jI@WPsd{z^Io3@o3U1
zRoU@nzmGd8f{pT_n>Dziti<*!ffmIXvktr(y{nh4B~W?I;6D@Qtk+{g>XwiD8r;$w
zL1|U7Zs~AH&*pH#pdfD=UcveZJNH}On54t?Hdw=Qy_{%AeLK-5P_E!a-Z6m9&f@ZW
z*Jogn`AQVzy_glw4Y9EW*7f%t_l#_q-a|PI`M_Eu(9H@3vJ&s!IPiM=Qh~G(^mjhE
z=>2J{k@uN8MTW`%B`71V$CYp};dr|Mq54!gY_n$V0+A!G-a-M}taLsStQmza4Hh7O
z4k@+k5x>SO^3dwTD&E~boW)eoc{%2Yv1D1M^E~#5vL9`Y!wcoOBgWEM`9E4%7Rd4T
zX$HQ;`1FJ$yUqkdgwaUSOz1PGs18(GMWYpRV(N_=MirOjq$6QT!VLiI#N~BzOaL*D
zryNPDL$J}tbdbCQOt{sX<W!rC4vYXwA3!ylI>0J9-Jh|xQY&!+U!CF4n5q(_0}>r-
zFdcE4S)HS;?@Wg!rF`=5>R1AO$znO{h;#9oGI-1)7Z0-RRZ7n^%qn+SX_9keA(z4B
z%NIi0LiIKsji}4A>}$)Vx3wL@<J!mjQV3yb>w)ckqkk}?bmv0QwDeDNB!^4hE(Fs7
zv6$Ly8GMWe29PuGOn{;r*Y+R-9Fmiw1cqGP4rV9^Ic_V)XfNV-$q{8x@<_1eV}@dr
zIqAWI(uNshj)3x|T!#6C$%MCJw_y5^5eW$G&e|p<n$z&8BhGGRn6!j^`550{-k<Ab
zto;Ichc}vxKjKX0xGZhJGVzErDAk$j955NVzfj{$CfVFaUnPZc3P1+6cH!i-n1oTf
zc=&xqQ%IlLY9QrRu^jU<gD2G%nHm$I?HK!Zt+#1{x&eUzifY#%G36@EphdAdgq}iH
z#VubFojOv`qF3rGhI$`lpgeYt!6&$TaLT6zRt>&PD!l_Jdi2E4hlR5`Jwr+yJs4Z1
zCce3>L23ht@bSA)>S9uA)~SUEJ;R6r)E+aZ%PgM_WAa;Nb|4(AwPbi%T3Qdj+~k9z
zRtY$6V=->AZ%5{GWjIi`21N4+K!;c<a{^yxA`&GdcX$uVp5hATWqqfQj_?3-Pu`*l
z3@lEv<{rH}7B;G0?uofV6T4XM_0}Y+>emVl)P2@Mmx>g(0oy#(qZt(`9yDB;c%?iP
z(?y(23Cnss9IYu_o007iAA*+-?RwNfKI+d@9Sh*FcJ5>GX(1hdsh*O?^eOGY(DUQM
zn`dhojQnS^Ak7|yWpF*<xXB`iKDBzZ|1M9LNt1n9Tu}pkR!kfyNp~_?9uJV$L+gXB
z^uEb|E(X3M*^;b|O&<$Y)LO&YJglngeM}ekRM4TNWQ{bZv$KUFxtcbu%-^Qi6NpPK
zu_hQ7THEIjLPax>nUOZjHrtmrRAYkc?c-_fRP|c~%MRaL){9vnZ8pL4FFId##i!#k
z`1HvHRu)0T(%^yjlz&RyYpXmRXBO$=ve2bGlLmu$kNtvkvpj3v_q(W%Ja^Rk!SsN*
zn;ZV~2C-Zjk^2SjMbl(qTok;L7mxZ>Wn{eB3TEV`!2B9|q{z$uu_=eiwcIMNq-Run
zP`|QDUX9PF89{jfo8^$P!{v2v9X1a7M809mc5j?FqlMuP2>e=k%K<~byY2F}x3Kih
zO=y9L#k7R*i>0AuF&6i(Q*zjm=v|3i-%J+8OoyhKe9t)wWEjcH``)#wLCvS`gV?u`
z^>9qsYriT=^c3z7j2&+3UODKa(Xqw~OfvmzWGriv!*L}~s4dG>SIjoCQ)KstR9Ayj
z2gjyPvimr4D!kq3cf<I$&azvbw@7r<dBaV~Ez(7HAFChdXI4~cv|3mHo;Ub>@Kp>1
z>*mkszCr~v9Q~xb?Cypyr)Y|<52Zm-=E}GK7=Naa1k^S{<Mv_MeF`UMxr(um>S#zd
z$rO&qTBmJ|L*YgJku;M9s?%Yl$#F5?2sW0F+9yya<2F8Cc7NEGLxmSE&yUIOkJ^KX
zoC6x4;NT!*!m=?fda2Ml{PAec7)oV(z$u=XR<BvWvHFeji5O_8az?1BX}fSkJ!Dsm
z`$^f|1>^Qb%8Zj^8K5U%H^A6zlTY~!X!HTwMF9{PrT+c_0eFnar(?CXIjZ{6@2N2t
zjn@G`gsb#3@eH0nmbGi6oaWDn1++e)sH}X}pOMJH1c<}dET_xv?%WW_!XUk$i@_&N
zD3CMK95v!tg=$l*ecn5wpB&|T=A`Rk3myXf#usFFXY@I}NY0eqG*RbfH<Za2W%n`o
zH$0<j;p>x?vtrgzazjJUzjAhTj6_`IoSA2$r$3`#xtemJt~)2@IdsBgIoBI3hI1-e
zMtCsFmwZ`3Rov&j>=#_YnQrMkn-p+tr1X~EAJIu6io<;pz8v!`&eb7GSNeGOfF!oc
z46EnM?rZh4w*Qj~(!S}`ZpEp!MB*kj^@}54!W&)s8=4@mH$)XJTzi477)liSZVd=D
zf}LC|YRHA&bx;~2iCK0B`WxOrjC%$l$6qMa<e~)Ut<cA6oevolkY$n)`HG)z=HDWP
zMEOTdn|0IGlR+*H(2fPWp0AXz#@FMK3_vci{%Os(s`Nnjuf-RZ3QjKdIk0Bce02@k
zE`tLY@F4?tIwU?BS(`x~e;FEEjMrt=$npZO!=twp*Vq~v?$7AM5cxbtq-U$uBm(P+
zjI@IfUoAL;*8upV(k?MGDr8s1Tsj<#_BX~<(5eIl2Tt7v8Dl4;c!peNAb>rX#;>tH
zATLMA|Ed=n7r5RR)sn`i25T7rs$~Zz*t53AX{hT3n<0IlCwjXjI1}3deUi?v%kJZe
zwBK#JTmP@9E0gRrR3VM26xdKGJ&*HQn1-UY*4W?6{h4Ym@QiLFlYP^PD%@xd<5$}n
z`Nh>TCFVc8vO=y1vN<eQ52JZru5>!t|9rVB4aoZdv>6K8tJ8qgA(C@|IFj1IV-wHT
zA>GY2Hb^~8x7Bj3!=3st3k0P7j#clk^SRQbMzlu8LZT_=4HO6Yzdm(MA4^K4+a;3@
z1Ot&QZp3)tqCnru92|d(?OSSk_0&-(SdzwT&tGEfgWaoH?^{R59JWI{!obD(QW`VC
zFKjdf_%%`%&vf=&!+a^XIiOY++3gBD3h)(H@2qKYGKy>Kp=X~SWFS*nkE_1EN-Ayh
z@o9xrdGAcCfyiZbh_3c$hQStVlZ!P`ouIdfg~_oPH-qJB0xF{l3n$7AaiK!MtSG9$
z<;X<U(Jcm1IsutCAB%(Xg-MQD(N$fb4%7YAr_-di81=`2egGoSj5zTqamq|%Gh>}2
z=tKcX$Sgy<ahe(;2cL-sn<S{%CMbPpCxl5EMF4V!Zr>RA%J{T8?C4Et$7o$%xFaRm
zkej0^m@ue};O-EF$I627_Wp)1(h{&xd%9U}^{Fg`v+6Rr&3+-f0-}Xvj<u%+6sf&H
z%0VM;56)ts+>t;sKZS!ENWSR<k#dt((-rg1m>>`o;eKWEtpK^_C7>PQF5g}Ci2*J$
zB>+2tDOuI9-!@o-yRxh0?$k%K`a3pZE5DPTLG_FCMXUQg-Y*P%>WXBg#>wWYJCuCa
zFmcM@NyFt{n<&hv&f$J90i6{bZ8|(MkaT^h`6^}UU%B6(87VRq=L7yM7DR#R`w4nf
zJzAM3QkZS>U~mZI%H#*OA(J#WzdR(nT@9#}=2hoo&ORK6PJyW;nEWu%5_U@JA|Fn3
zXr^Zb@N1R)D6Ne9WGmwVTrZFK8zVUO#uUas_I|A1mI_?*Xsj7qrCEv~0{+Cn2*VRX
zz3Ghnm@TWKrlvxEYFGpvg*5s|G>5>;3-&_>88S)5+`wSmO|{7Pf{SHdYD`-oW8TO~
zG7a+lKrgT$>Ugujx10ozXaLp)D<qV-npRn8(?Ld!I$0D)mU-SYR7@@o=#2z%1#i(5
zw5+7I9uAL75)hUEhnV$Xt1R`Cgo;!^_hkVRFe5Z|Go{ROpM6KC9gC4!5v_p$)K~%$
zWTijjF9Yg4S!I*1$`qH#>X<ZK%8M(RDr-%Ap>%<AAUvQG#7HZm+&XnA>*9_2qS+SX
zFVK2@^bGcnnP9WBA@!^|;J{_9mF57^IOe;VDI4Q=K8q@ZC9)~?&(mfuO4O|{k<A7v
zrV?Qe0B^S#sumCnu*^#VQO<Bn0R4t~=I?2ZNw~)zqhh!EGwNwj;Q*fgb7fmPKjUP(
zcWx$4jc?@@2Z`bFc%zbqsu3|srXc#64U1shGJ2fv@F6I&>s^C5rzh-cc;c?MEfvxh
z_r#euO_iPAEsi=6*~DjC%`WS<C|fs^^&EHmahN?jBTuGlit)QtAA>&r+_<!<=gCtx
z1BeXHngStz5uY$W0`0^&W%6{IDp?O|_gdta2`yLRTt5?M3K4$PuwkEmmG-S30W8S&
z8M~i_M|vd_8U5N8sC6FXIqLxu3p~4w3jfB3HO+(mmFNALaTO$Qo8`CZ87<ESl@}6a
zUIFQmArmyiF<GhdPx3qe#t4H2u)KKGGmAL+eR_t%f#`v5qA&R~I&E*|Z32$v4*`Je
zET+dHpkKD-A%hm~w?D==YBr|YZjrp=&&V5DBBurNrx;IdPPCP+lUHT;>0ohzMijLH
zew;YtT{NT(tD{Z$&jIg(hLY-RgS^(>6Il<qjBl5}#4|Xgkj}BH!Rx+}6pdQtul@_>
zOZeivVZWf-gD;o-SNfXy^BK^)LEiLdESy3@xDCAJ-{PxL;9TV`--!;C03)~iT>L#+
zQ^_E2dw=9o$sw|Q??2MGR0cCi{X%)io*^+-10t_$sg#$08ucGJVv@X@4nJ=}^>6~1
zcxC@9I&%OsDaU*MjKX*;Knu;P{5!yzDT91HL*BQ}2lTl>{^OlvMO1U>MW7e0eBf`4
z6DP(lL3cED?TPN--O31py=fW@#iuf<N%jSSDI6~Q9coTmfCp4vT5ts*EeGrre{GN)
z3~ENBNmdR;Ct!XL`yh)^?-aUN5d2MYB;8-?=fz+ZY)8sQB+|*0YP?M77rjvpUSvo+
zbuuj%CX1^_^&|MpBV}mXxs$5S+8Pm)BV9U~z7B>v3d&kgr>l2uxZA-eOuBV)EebeH
z-pAcLxdMYR&2S)sA?QP$Tn7gFpxF+NNxgLpp5jD5+$pqU4KR<vFOS#8u|6S&K5Pao
zSkLPt{tVY5z-t<x<6;#P;AHpF_z20Fwg7<;IVi_>QXLpV4CP~;R1rkSXF0*Aqq#M3
z5X>fpjNS5aTU>@~VN~~t_WOzZz(3pvR{0aLc$C#8;XKKDtw-Tl`DDxx#^>$MVVvAa
zm0Kj+RYmftfXf7Uxzxg`<CHY#HJX=cJLIlU`!jl5;8g8Ed2l&3paeRff-67c4GpzH
zWO47E<||7@ewA&2-;I1WZ2?ai$qAxT=&&cJ+ZvGI;7a*itZ#Bfz8=T)j5L|KdX=@p
zh5Yl;hKc(TThPOCjeD{kh`$P02&ZOnftEq>3$_UruxXx|08bmXfuF93>*R}_cAE;h
zfTeNeEFYKKNOxFfay>iEg_-xvlG8Ju!GnV`8w(zuW9zA*AbzgDCqbj?1DVI<OE#YZ
zXl8oFe8@lI^iznjJLx>zxE$ySQe^b@C9<0hHZ*`|$;xQv{&H+=aYJ3M4#~j#1ZG|a
zAf9i1O{pj;L*!%MSZNuow|VIo(5%dsW#C-gKYka~)t3N8F0j=M9t;2M0r9RjQv}e<
zh0#?J6d83gu#;=j&9NEic~LZgr!qo^2RStC0w6J5C|~hs<bKGcF2|*Cu|cR3x}~qi
zC&-4%4JC7QNlarTB8Xvi_*#rRVR~f-fqEIa)K&~Mpn+s?w4Z{S*X7WUC8%4mbCx`y
zrlv`Tw!>LNxaq*X$ULW&8yps3odRXTaGT}e`Vtvo&ta}vH&aIXlzBtU*K@dBM#YC^
z4OKGQ8=!X}EaRh^WQ^~(`C42StA?)yec|AfAY<d1hPRmIX<R%5_fOUnky<i7R!jd6
z?7@U+L+>u2;9T5@6XO}aTLCTybv)(k{){+ARVWQnlYCal#SF>6AbQSQyndm70lo<O
zF}u4iw}lQHKIHO|GT8wQk6KkG&DFbrrF5qFr{;tMl*hTg!dsK+0w!P@p>(@41_DXq
zZGyb#a+PnD-NSYRS^ST>4$0Lqw&yuuxQ0o>Q`0i2<;%-8_WOzNLBziofWFp7Yirb`
z9A!urS|-<}-YBrAS%W3I-sc7oGF}S|N3&!M2$!QqLSBjthR|S;k1b9Z)~+j5<%DA3
zYAH!eWz-$-y%xy^e&bp|>@ucYng->?j0<Gv94Yf>%3}g+Z-W?L?&koeRj3jxQr}vs
zgTcx&EpE?g5JD?TD}XT5(^_flF?k+`0A7=ocAR^u+Ifpq`C3#niVbUb_0TIps=Xst
zVX_z_3p?K`xik}2Jm&W;Qsb>7mVnu5%w%>$9G>eC0#j>qh^&_R=n3NbJg1#?{tT6r
z>1vMivRUfw8Eh}rB@E=HR5c^!2a`5}!$5<n?%x7`zM0+w)}hK{=KY)%XfqtTILM3*
zF<<Djp)OZN2-~;F?DU4!ggR<FZ%ni0h*n?Zrg#SBL#*KDPHI28_FB0m_J<|zz!O01
z`i6HOXk$hhYAbH_H<rpy{l!6E=!AA#AYrB+J_mD*rnfcr=_|L#@MyONQ)0fiJ7NSV
zIGa@*K*|mc`~7BIG(CU=%+h;jK&1puE4Tx`mAcbtCS1eo<gSFj2m#M9xzxA46=@AC
zwa~rsZXfh0Ejk8;HqBd;CaF;#X7hUtLZwymU4M&I){x#PqivLX4M}i>D<NXa_u9KA
zLbThK<-UN+5K8y=$3VEJv2j3!CGtSP0-ON23q@rU?Y{4=5lKPb1VYm)dC(vWbxl@&
zVDD+o0lNim4ub7N=^GwzE`NpmFb*ktSS}CSt#ZNn{p3dmTf|VJ;2mK5h@UsIpbVf{
zhP%Ode$^?;k8N=1p6cXL`}Y%bfXzRNP2&|`31`K};y{gt(dnnYK-0@`O7v!D3WLVr
z`s#AX{*Wuv=;pZrBd|^LNVI_P^J0Tb;Jc*uw)x&FB!kU><br^Q#!Et107O{mg9i!O
zrmE3ZxQ-UZ8<o9O9>dO2q+aZAOuI==lGisgAUCCNEDZz<_X^|XvLxQ9S&1G+Ov}M5
ze`%U2*(vl1H=boNQ~EDipuH>)U}!Q(Wa4Laf))NTImvLCSQ#C{GEEHw62Hi*IK`Qw
z5`<`JLM7jI2D&VgHL)hF1_~%_ErmfUpw=2bjDQY*o$YCkdbag}xCmLSl?{G{-3Q6z
zen`R1hX0d?PbwHaVZ3bgDe27#lvxERZi*4%CqfdO%t5j_b&Yg!(qgzW0#QeK+2S9O
z&e7w{bmIhLs}EuLvsk_6wgy_10(~{F$~HT-myMF`333cB1dYV?^0++%jYeKLh7<rn
ze%3ysq8qPqYVa1z4)4KaLt<=`JdqZq0;{nQ*yd(wGcacI8QB>_1GDJ-NNf2nKU!2J
zcm`!q&dtC^Dnp<iOy7X)IZi#9p3%43cI)T<3}Aq`uhc@G@@JN<ombjA`9*+cW|k}|
z7W%YpZYeXQ{nEcsQJ6dhwPr!mVA83|%C7>WX29HukP-CothE?ckw-1_2KlwEcr4`2
z=MoUHRoFIOOWeYx@|z&vNO;YF;pgqQS3}FoZ`;=c@T-PWUP!&E@ldq_xQlRZerH2w
zZ>W|RV}3}YKAiy~W%7IPOCd$36B>q>yf129EEO^+sp9<~4DSFPOzUND2F(WeT+a!k
z#NdwsNGL5?kQzmLd<iJWjN7JC{uI;?EJJN&MOI#o&fxpP1G)`ZfHA>={yFWGCEZ|A
zwbu+~$5zN+{5wTfQVLjNY_6}z-pqveL7n_H=9EFsyb{uhHw-Y~JmkOPbPWc7E^m6D
z=2b^1Sh}F)^0zcg&0p^n3l2^4R&+n4zM@jysQzx<&&rp`+r9)^);1}Ju}=OGn|yf<
zF-6|-Hldawck|DfR$(STUEU31N1G*_4Hdus6@V;<_c$KQdxo;eoLMFRj!A($GC3E*
z6>S5HaU133KRzT4s0>l0s`3M0VFxx1Z<uF$kFBRXQz3i3krFL{yPYuZvl-*rR#3eC
z_DQaxL=M<L`@-k-pdaf+@a3g(Nfj+{$58zY5cZHkCWk5rdTkw!B@Qkv%~asNJ`&(G
z-ItnBq_e53aEt7mgVVGgr?XRM)$Sl{8F7>L-tK**e`kzum(HpP>fg7IbnWc=t86`-
zG_%sJvn#8rbMREe>F?gzbz=pHZ26E6M!HH5a=LIH)7g}3u%H8&W)omgEp5@5l6=^|
zSw!2g0(+|*>(4M<c)rnlajSf!v#IaYivjJdRw_N)j`KHmR3Y6_a%Adp`KXW0`H)*>
zTn=4^;lBj2Y?&PI<CxV!tL4bYI;#@O`Q{TkyS@%a11}E$@%BZ63fSdwVnAVRu2??d
z4d`;n6jIb`o)rB6<7(@*a?vM!ayk_>gZ)}*HBCQYxaDM<2)v+jutt8$7pas`r%!p4
z3WHOk>pc9hi5PkLbZ6HOO$HBz+-bJo^;OX9eJ1r<i}W?8B|x*9G5saC%V!gynZcxE
z)sEBC!tCVJY4me3dKE}}nerKD#OQfMRcGb%{uOz9W-nr*(H=G<U{FrJVAwVk8R^dq
zYy*_)fzV%!-*xp>aF&s?VgkS{I(D+e#M$=wI1taC={1ErLUEOxW9x;XrXq*Zq`o#}
z($u>J83?fIl`ubk$-0?%d7<?3ZX(OfGTh(ib#`qLEa@!CTPD4&V_@e>kcTe^&>-~l
z&Sy1xcx0mEKGtqxMs;K78=_3gF*2a9K>?CYKj{|(psZEBQn3E^7U5`lxxlBy!RTuP
z1E&m#qc|#4i@X~pa$!JgI?{Th`+@Nr?07l!Ocy!Abzq-JWgujjHe`^2pzlQ&$X8-l
z;V7%sH1l$?y&VSf`PDR6svnW^T`uwY(QstmAkFeMoB!ZSWNErIK+4+^yb7IE2Rq2@
z*pj@bXxKoYnUKHoGSpG_?zV5YHOesmjEWQHa%xXEJUrH?*N+`5Fab~t*u^)(y2W*X
zyS7?J`f~eSB%=~|x2w8*6?S~I_ZF8OO637sEu>MJ%xFwPC0a0SG;^}MtaDhk%xi)N
z8}GpjMk}()V-rY_2pWB&MxBgHAh8ez(hu94@!p#4$a*k!#)QthFNO8&-2cu0WMUAn
znTQPidi*GY<K3hLYiij%Bxc6~+(iYkMM&cbb9Z6Fh?y022rDd+%N=~Y1mGzSjAbHC
zn$(T^n%r5H2E8wqDTXIS6=eu5PRm45G<>xHI@c@XpyAh6xyq+xIYh|yt_MER=5e+4
zQ^<(Lsg5#@C=gXG*F>*+Y0#RzMXpWTtKf(3fLTx%fQP11rCF{^<IKaBcjT_G_uiP!
zp8=k9%gqGpVVz!$1Do9*cvuui5h-d(6N_V`DSBYUR&aHc@SsxjH`r+C?wX(y4!JcO
z=e-u0QV{wpr9KhdoM=Rvvq8#YU#he9b+u9+`+@*A^D(Q4C&<adV!b6Y&EZQ#Uo6uT
zte9(#OvoCk^pk>=U{&6jCA^nNb^DZH)4?s3OR!n1<tvtfG}hR@)MilN<A&5bgd$+V
zG6+(&-X)XOwo2Y1X$1yI>8O-q;!1M~sf+nnR)AT{WB&ElXS<Iih{O!<ljT~r%1i?a
zQlhCCKPx7w$U}~-MH*rfJK%jHv+bAW8}BbS#+0ztxbQW%0H;SgGcFpr$={R9pf1`}
zTu3C$f8$i&Y`wubf}->-KHbKNlmYa@mT$y(71O69Bq@I7>l!NMwhpMJUM&#SjGs|f
zK<3!yl*2vnc84@?j5`7mMk7+JQob2KhvsTw+qly~09z4=s8;2XZ^gub7T^NaLGiBm
z45e{!gM2%#1tc*mnORW+5B|FozG~5E_(HV+L?m$ENn@;ll7=|GN$&C1h_ZoEo!W2$
zl&!yO%a~CyL+*{kG8lCNA&-2|J2IWmLa-4_<UU8q^9RcP=`78qKE@plse5@KXxYdS
z6XpAXIN&58C*THuFjg~x224~+gBHvn2#-62r`t3f&+_uav=In51#P0o$YRww(oBIo
zLN=v77xa-I*}$;*))&hoG0aRv&4UjC%Kq4P7%0u~wMYHeVuTv3mY*1596VYc3p|*C
zYNGPf1kUBKelpr)ku)aA%u`PrhD6NuXD}zK)sPv^<~fM`&cq$!cahAGv7r{+EDLPj
z85Fdvm4!Bu`g&R9pR0l!Dr!VTU;?=IWO3)+AES?QV+KVip^$g=Q<H(BzN}1^1c@lE
zK%TZD1cJr<R8>Y|yk3??BP|A2R3Xa)1u7A6i5SHdF$3ld1Xa{5IvRhm(!1kfFhB6s
zvdU(D_Ia|}_7fEks$`9|L^^T!_vNA7#eT2#udy=vCPP42DeIzLo|QErP$JLI`VL2&
zE=}fw46j5xin5`z8|Jm|Qlm4?fiNzP88<*(4==X0vM~liEI+jTL_*nQkUR67Ua~p*
zM4SLLw7^+Vj~0VeA!RT43uI-BZ{Q+O@A=Z|a5C(q`wh)9JYQRVrS$k>5GbSkw{_mF
z9_GyE3QUFL8rQnrTUsB8@`p<BaqriX{}cJZt3k1~On&Cfj(K84WDIX_^4~PH(NkcD
z_vyQ&cjgn`C#7uOanOn4FXH5-&0E8bS~cLz!MoF2>uQX!B0~CBwBBVPtAs1d?(`uo
zk?`e#t@25KMte1RawKJupL=IaxJV;Bi58vi7s^xKnimfl5R~DRUqla<o|Yb0@^l($
z8R)U*J(gb@vb8lL^z9iVoV5*>v+Y-@Ig`s6(m&fi6jpA+%&&u1lxAi?;9;Vr=X|cl
zwS#ruImfs3H>uO5IM2pZKW`_P0s!=33BuH*b+jbB5aS?e3P?Gr;qT%Ms$5^xATL_?
zjDE~Bgp+*B@9kvGtdy5h-(<t>hUV_J$R84L+bcesZ^O%Ro6OA67v_&Xf2#OsDQN<w
zdL=#vU7$sc@xyTXh7$5mj_70)n&ef#>N1%5pA7+Sy6#4KEhzr5$;0F?-an5u-8Z&j
zFJJdFNq(HMR}_0yPQzHdGX84)!ZJ`IZx}@ENW*te{wwBRn48{^lQ$jpjjJcCLCKHS
zL(^s0!TX!_crD8-zU4DgEX0aglKbzSclRa<y}RvyAvQ_gb~qXAz;uEX*?9kmLrsr#
zIxM1$EwpF`qx!m1`Df5pkijx6^IZdM8S^vc<zGHU<|UwK6`V`uy#OTKhZ*p&9=rT+
zpMaf5dY)KSgeH04H;TIGmGYk$4SF_uJAGhp5wBMPui-thU!zg4Yp>0*@A((XzL*81
zLb6l{=a>D~cBp#p9!UGhl9G#HHQk72q_8Tb$iXyaT2;*x<B%cm<l%DI_Jk#O<cPPV
zX@;s7rk9c~sy`_P385Qir!J~5Lz&Q8>D<MA#lVJ`t34bSmB3rbN|!DXp@t%HW%Y1H
zkgnESTO-N|0|MQ;IMs!ntVA7kX>04=#ofx3*Ifc+-69`~XVj_dO|x8%>7wc>Y&Oy$
z$cMYQQcTUU^ay80ZzdE!*1x?ErjPsKhj2f9mp;-(^?JG5B00`K41!LU#w?p3wMK)B
zP;WdYJ_4nCO5|hqW(M-)pWwaY!~<xH;WxKMKHg>b89?Og{y+Yc6Qft=uguCPtk-`1
z2FOXawzkG$+2Tt1q>n*T?14&havGPRH}v9KDW3`$U`HrLaYneI!FZq2MRk+7-58tc
zpVA|irZWUX%Bfvky9q5B%}1Y!Hd?maJXB8e?hVnY0YD`;#N{rQ&qix<`Ff`#Ev~Jm
zxHu!9i`5jvDMQY%xgh^3#M95)xTQD()wGZFIBE=Xg-oUkg@}B?-^8-oZ>lqGbfjXh
zhh%37PQU0gqOG3N2}?enbyVLZ(cxNhc1#e72kNC~7uRsX;}kB}I2q?Ss+;E-X8=@Z
z%n;^AFSK*x3*1$f`s_<FKIX0$4iK@Ik54K|DHRDO@jWwypO@Y&(=)mf$~c9u^bQos
zD39~yIDfsSN*|jvr4yj!`EeTExMx#Lsd%NYPsltHw?1(nWRQOT%rX;`m)k7;ySQ2#
zGO^lWF7QTN^(H@h0}?za+gaf<z9;iS^VEegYY<-KFdyj687fQKNS=tBN*~Sb;E68k
z04%yg!BbTR`7<L%)b_2EulO@M6wse&8929#4b8@8%jB#6nkp8QVp0FBOX3r_$I^Xq
zGmhui3<293OF`u?wW*@)mqwQiPAgju_CyV5_=`|I9BtSiWJrKVU8w-~q0t$Xnlj8s
zQ%x1VV;Se+{>)O=+s!n>_d85_Er6ot+m@03hHzviFyu@IfnyNVC_5>2xdMcD<E|fV
z3!gCQG8to7ab*@H_OdRjO{znsiC8+W1l7*KIzKKD^)g^heev`L8Se|`fNopICboz3
zP4Ee+^D}^eC?mgKC;A&8GUz|$2AC9Rx{$A@H&jmZ8>;od1IwfsKU<B8Wn?W1IxLv-
zavYb9QW*1xn>raF+IuhWvioFE*026g|C7nK7jWlmu9hh=3G5#IOV%h5m~QlnI7ByI
zDpz_dqiM#i@D6R2s~j*L7tW<p=m8m)s{<2CVQkC5Np))4=x|J2C)d~}oilZ+Tx%P1
z^?7n#S{U3burNAVT;C1`5bZEhGKxH7#~fEEorsET?sesmfQnN;@EQZIzzIl7j#`hM
zNmDas+oy*^NN=;bQtHphpfEZC!laZL5T`N3LCS4#bxT_(72X^g3^1eWCn?j?Bq(T@
z95UUXx!Dgg>ME)9XMyt6ELQn5<}_u=0gy$hj&0LtCSe;RlZ)06Ul1G%i3lN^!DAr%
z4orcpmUDx5z=0^sfR_bAGplB;bpStwSrBEJh`CPk-l;=?-e6J}1JMHkG=n%(@6Q-r
zN`v}(sPkq7S;I5u8NfNTSGJuQq+t>gsLk>YO!%THe6=+AGnyWV2=tGf9mrWxAvXpr
zK$%G(kZ%Y>Q?+rEbv+zO%q!()L$VCwQ*Q~RM1oNeiM)IxSS+u4ZT8YzkHQk#31LHm
zh`dT}i*>dUr-<(@GAF2cjWW=;8zSF4Nbcxj>V0ZMU>#Q8k6?!njI!ZCKe#V`vwfAu
zkKmTMGd*J(7|Pcx<y*G$F$M5^xy$}N4NeI1Z3nXkg%A_5``lL0A;{e=a<{epJpPmK
zc)y&*B5XUdzso%x#-f&jo=@NPXAn}<I}GYfxi^@A5@dLk?*;5JhP$pf-McTqfto3(
z&33>296D<7#KPf|<N=#aEgcoVAG5)pu@5pK587t+z2JQLfiH|odx*ji;lZxq7B%Hf
zi#!zg1q@_T6(-K3BEzzlw#pCFB3hN7K-Lp^*q<%I(pj;Y3;R(_I5U%Zk{$`PECiqT
z<5)Dj8nW`Jy&3~fEH6K?mB1Bki9F_0v6E`2lEg(x73lI$(^Pn5^w3FTdWPzmn95R^
zndf$K?GWo8n94UV^BmMptyH%Fi#XKn^Suj`D=ito0&m?xh{u2od0yymxN0H?<8=wO
zCRybDC>`+WE%s*`w2EiQKq5^B4P*=B4ZFmNA0^p^veZGacxT6iI!y6Ik$$->OLNuo
z4Lj!Jj#!?aEq4>b23ZliQ3pkJk*tgj7>uH3vMNoq1#SUmhF|SxeHK-bYwQQExJ{_#
z7du&5Um@$FqiWW!U7ygGMP(O5IZO9vs=qeG_+yYZr$L&1L~1cf1Q^)8F{lXBuu(YO
zlzLN4Q-h32lAELHuq?y|#1^!q{<Pu-IJFkp;-8~x0+5&6Oo@dKyRCr_%=iVdShd_+
zy^AQ&LFbumvAV)hg9~N5{X$-E@cZ((w_k)?48#Nr*)BgzLux*M9Qxq(cLXXsUjVlS
z>>w{M!=fi*B)D|dNt^d=WE+SLi{H**VhEl}c!<ibv}mT%GbU)ak;zdL<;m0s;E%CZ
zgOB`qn%`oLTtx;!$JM92HNnKv6;v+&BH#}$CajNoCgkZvf;%Q&;y8Ni+kGUzj9sG8
z1wDv76CHxU!;A}>N>5=XiT+huN((&)K>ZfWvmqRoLjMKE<JW156am$pP@W60mE=&h
z?KklS?qx=oJ|DfRDxOPQCcm{WpctTQ{0qJ<cn3A2MsXiS41X6r;BbdP_hKvy7gH5g
z2fq*8M(`;P@JseSkF1qe!0Hb%EqLEmB7XAa*vYXOxPUPe)*pQs@ZB)9EdochQeKIr
z6;;62_$O<Dis8sQg)kqwNxkae<ibp{qkW1O&7bXC^aQ<>ycSE=vHlVi1(zkm%;8O~
zxI;%P=1hOx=Yp)Hs#)Dgty~&u3xueSE-7!smf|)CF5?2Et)1-ne;Jv8QVX}~o8F8!
zB-{Zy+|z$E;N>wo!sq0z)E{H6xUz7^`Fp^~bOV^(ZSuCQtO86bF1~*RTyr%w$XkQ0
zcqi6^dkHFD+;#KipFV=k)y%v`uirgtY#O7P&x@w*e|bM<pf<hsd(k)Sm#mfo%$I)$
zYOu&PC-8ot2Ao=fDgTL|C@Te{;0LiJ=;-OzKu!Oi;4P7?vbq7bGDLLRSI85f%;Uc3
zWI{3AFlB$NdPc^rKL^rm)g=`;wLuQ1XXY(Im%2kilX%>S3D_Ez5HhFFt#a6|;wmV!
z<w%DOq_aD*#wfdVb#_b>7g0aK`>(y0gofrwr>-jCSJchU)-u#cI(IdF8_zNA+{~EN
zrK{`iP$#tpuGCdE(zUCq{)Xmi8l+oSQ?^Cbd#1C&cpx9$yCy?QHm`-ED7Ad3tEuW>
zS%c>J7@sHY&AgRZbIrn24IOQsANFbSEI?j?U<hxxuBLLdb4G18O{WG&>a0J~)!7Rk
z_y$S@82mV&Y`Cv11~M7IKI%i@m?#%Ww}InhSaeQxMk}<BA4|`W*%AYC8&B|F)mnkg
zLS=C*g2ejcF$u;G)K(PLlf0eS)gwxC@`>n;{2-|327v7(e};uoipD*;Kt9>w8NKV^
za3d%CGc~7UR0f!R%78oE+!9YoqchIdD>{7IC(AR&d$3(obE<d9(?t;k!BFxUe-<Ib
z8X%gMo8o)W8u@JGv%V+hkkx@AIgn4c>9sW?i)e{_F17w|F*e@%jIOG_sIHOE8@|CM
z>;mc0Rn?d9LKx6rzF;pv!83AZnkv0ts4LngU+n6tOl0H2!IiV34ODe&F$@T!oE^(i
znT2VSY5ZV<@}AzB6`lwx2D_gVOPE3<%elS;yI@J?kG?^_<UN?^Y1XV>wgB88bvZfD
zPGC+`A?6t-+I6=DvF4w+1Lxz*hAUH`o$8b3PIYG!G-Ipe{FpoDi!eu6BB+ey-Q71<
zo}HBiXCnQgCCDxF%URJ=GQ|D6x{8xA9q??clM8$#MgXg0KSmnhkgh-w;f3B86Y@nF
zV`#e}G7j`F=^?{|3?-XfWJ^R!DR`!^EW)7l07qgv4#Zc|S8-*|m5XBxB?)?YIa+)*
zp$%hrd0v@KhjvmBOE{if;$M<jP?Nw25<BBxi}L`BdqtU21StJ%VFk=BGdMuJ18Hf{
z9qS1W@n?(N5`ue1hT1pCuTdhy0(>wTR-k@!i43<r>(~DR8DanId*MJC>HX+b&^L3<
zj*9Vi-VT$r*}q5oGm;jQ0T;>`>jpfBT;{7=pu7>+%?ju!swQLoLpyfdkNGmrun7zi
z+@mXHy!A6-%6OUJP((k_$Y~@rV2Ql-fF+7y$i$c$s2L+Up?sIG8|E@xB`cG>0d-c)
z4ndMske-n;7zr#)9bt3)+2wJipx~lMEKc@jOF;FQVtoPTFeN^<s}TI@X1T(jkp^;K
z=L0vd^k*gp!W9YZ)m8p1E<HL_TpgriEcEzO4GY_|o?H{7fzh+%IjiMbhqj%UFo{C%
zAlJqDq9;*Vtz2(k!x{$WAX^SR$V(>83$p}tIoG0ScSAkm3~(nw?^*2a2t<k=P`*h?
zdd3J1-VEy{YiL&kUri~EzE#O&_+4hZo%74CJTS9f)96EXtB5B|uZNN~aC5URZ@&4Q
zGiAD?gz30h{G8?XjN_vdUS$jk?vw^KhSpUXMj?t{s{J)Bp;U{3agm%qLyiqSt8_1~
zN!>7o2(%#(j&BHrD9e&TN6u2I4FKosi%Q`Cly66ToC}i&jiA=q97bF|Uh0E7mr&A}
zVK4gh>mxILP;>M!{&AMAeH2EN2KzVClNHPCARsQ=lH3}AZ%j+1RzYDSHyKhvfeu;f
z=7cXi6jWSL1OPtCEwO5<*6@tkvA)rDcOf*`q+UB2g#hyr-q15N?sV$RDIIGJdXj~x
zOJz>xWzY*y@3jqW$lBp7tKZ~iK>5~mzLDOTZ}8si(dt7ew>fw{kD|;m$c=z)9D%X7
z`#OwR@oP(+woUHvb(s6WLZ~)Y%Qp>pP~()#oe2Udzpw-ETS4|wTSP5scX?BGkK=)5
zQs!Zfz;FAGW{c$R1UKHxAU1gYo#=^qw%|c2_r$>-jnu*4^)F1#uXmW-8<amg%Zo@M
zgz|eGrfLb<H~>mp$$j31mWRVW!-6jN+aiIQd3hiW!2(n!^nTyokI&A`%Y*hFS6)3F
z1gR-4ksrjxj({c%{!m~l;^d0thp9tL*|-vucsS+^_c&5-?d3-?Gx$1_&t3<K;Sqmk
z2t@L|Reo%<X=}W$mpp3Ffu3kdnf%1@!t%YbOa^^OsrVrEStgJ9dr~XnF3tZ`bZay}
z?tq^<o@li(X|!Qb3pX!w9h^ZjwkR$#&G5YR#(ba1<eNA;->0rNYv2j+SQgj_?vA)H
zEtqBrGM4jXQR<J%YKW*b=h2cVi~Wt^t#J*cM@`b?Z!~62&jZp#sS<2S0^dBiy#R=S
z!qN~+c5G2|3X}GjPFoQa-}8(u3p^YLDG*NivfP2!vA&UND)9j?r4>FDv`cF5h1pY9
z+U^wA)<dvb6-2KvQvn`XRvXe(!b_0Z8`cEUU@zy&+UN}ngd9zG2w7(zK%c7766?F}
zHc!NSu7=f#=!ytj+z{B2pCQfO3t2W`6v)YX*%;?({J2T7DM6c&FMjTUG3#V=G#i>j
z^7tU(WFw&o0gi8p_e1c$)yFZWQk~^wYitTgpM_*9%HdwwmU^qwbikPOcJBaeg(|<H
zvyjIPAewi+S$-Br3~mdVnX<!Hi2|;D<q7-uP}+IgqHk&>=t7v6odHTp1d8ulWS4Jd
zh&~KF3yO|{w#d7rf^1vk<#m~Ic``cH%72(ZgZ$j&msv&fR5Z;L<tpVDam4<+Hy7?y
z=mzBJ0637H!5<7BQ35Z&j0GZ5AH0m8Nn6L_juij5$*=rl29r?=q7I&(bqnR$*z`&D
z^6Lak24lkc1E#L$V!cQ^K}4YOfj*@?TP*qco3vS5RaV`GQ*0;bdFy!+11EnQ7>qDV
z)(3bYMuPhWPaGw#-}$Z~dK5OH7t?ZKT*HXl^lGK2%I{-F!T%{Z+M}Fbid}QZlyx{1
ze@Nr41pOfwyiQ(@c2qS1l$3h@XrL@ZLP)y7zS6!Y3T=&12HA7nsr<<&xj?T%UdgJ}
z!<l~7J2wpkcWa6KIV}n9rx=hd9KT#Hui4F34Da2)7!*cU=8C|YzaDc0e-O6wza}b#
z?7-L&IK}~jw30EK5iTCcK8xiIM{NWy%70lej9bLHL6|zcdEX4o#DiG~%7!!R-+~X+
z@)(qS-ijHcOR6>KIj!ud{5`Zd)zFUB!W|brl&FXvJ&XfEsikc3kJ$Kv2K=s;ckJ*Z
z`zeC3{@Gy;V9ltXeb<qo5u4j2|4QRqR;Tvj((gH7c)PKv;#v@%e<zqTs&O%B>uPyF
zW~52bS+49q0a8#~vPkqs`5--8p^hhTCy+gHI#KZudcBzHVB~Bdb~MPo*t8PaAFC{C
z(4sU4tUq-Jbp?+cOtV{|N`@7<HV?&jHTi6PNlp&iruFU9_k1~G-G8;mmqog{znkB?
zbAY{_x|za^X(CogXK%rTeRLV3TS=F0s*Y=GJbS8i?dHC3OtWN*_Koc3Z#1rf*99m8
zfPxd)y_>33kR2FSjt_M+y&9QCU_bt<IhukARR?lRH`QL<d}B}fu+PQZ<!ClvC&$L;
zpg$BJ=zK`f^K+QX<s<Fysqh1dn+tS<OR^HbCONK~Ya0i{3r8;!`KS*Ag&IXw;-(xQ
zLxXft7rh#w`IvzLAyc(-g3Yp`wn9GcP1T--^V1BI%ZUj}C~t_FvD&zK2!6s_!w^pT
zhQOE2a+2);15co>_+<3gw!+Qgt#Y!pXlooiapG|Kl>K4>ASh{1iQT!XT0Y%P6|QZK
zLp*)wsoh-pGAYX_Yh0S2>84t*k;CM)ZmtRf{|8&`XJd>!i;10Xdj_}ai3RdGUk>d+
z^T@)eE|nI(GrH}zdIcq@_Ki#Q^DzM`#29qZBN`OfHc&VIg@8Y*_dz$qBA91J6SzDh
zy`FqA=BySEesq>Wy9S;wXGi104B}Mkq-V@}Ob*^(=d=&6weIEIv=NNHh`d46-+sx!
zj3A1VEc`)X5^RuO(Hh9h1aWj5CtJ2e&a+_<m%tmow<DRw#?U1}BRhbEbb)M^FZ&iQ
zWPAsEKpxsY2KVdElk<I)ql6#G1#_iuLSB|LF~O{Z04DunvH9r;BkrG235f}?D`Z12
zNF&oD+OE$5_7z0oWpE0((0@_sxK#$)tVSVD)xRX4M%k0`K7*n+tdIw21FTG6vE9kF
z^C1`etZ{zSzR5Uxvn$y6UroL60=7)Tmvq}*0@1YXf5H3u|8HCQS{#s}m&v7p);)Uk
zlEFSPlRv1PS|URXJka1Voz74L&(L}qmfD)rGsU>!=~!4m18W%B?LQ)p796&aFmGh+
z+Z0|_qXNHCyri}P9HNXi{LWS7WK04Ua=rkJl<hBb0Gs}aXhCCbdkF4WThv6>wUBW!
zQLt-tT!eq3j1Rop+0mbXDq3}~kO{GaXw965v<VqkVTz%N@g65%-Jx*)DV0J&Mgl8e
zk3)^*J~$#YUrma0igflow*>)_x*V<ig|SR!y%s^*Y?aGnZ%VQiGC97dUk=XslmHqe
zjB+glaYaA{91VF;1RCg(jlkn8qlfX#r=OLpVz=BgbTu^=DxqE7O-(sZuwKRjtC!JJ
zrUo2&tteZxRj#oa7BJnCTx)0r%~o%@&gX(d2%2p=;SR?^uJ_K!q*2qc21hUB&rG*Z
zjpQaNGDH{$+i!7#sYXyTy09bwR0N$w)_%ZD2d9A2K(Lw|iqVz@!jGLS<<Sii?gKqY
zO)G4Wa}n({Eq0|yEtS*l!|}rl3Zyas4p$K7P|i!0PndBxJmD0GtD}SRh8pCQ%y|>F
zY669}<MyvfpKSzkF&pkmxuIK_4^8+;?$Ll9;6}a!mvF63bb1cjIa@7@1mvYIAV<aw
z6JZETeV}Luw*?C^(liq?Bh6wB?_H3x`7$$Llh43>h&Z~e%nA^XtFM-Z7zf-icp;p-
z^)frYLBdvo1-HhH_D!bt3aAlpig}=QIK!Be5dY4Vo4rqBf|enI7X20jVgVUoaJt_x
z#D%*@9Y>H`6XF`8Z=f+<5%)F&&<rRQ=h#148Q^wrZ&5MG=m_S++TRgJ0<1dRU*wxU
z8MQ><@Gx`2orVJ$*0%jtELQWYgI&tYU9sUqkjgtN-;U*iP?-Xdu5idkjRUz7l;`e%
zCn%kKC-AZaDz$(uv@2B3tp^;wo8Hfod*dc&0Tf`ae6QQ?bE(Xv?<tf;umCeqCZ%7p
z75WWt%HjL0yC)i<aJfHVqB55$M9qQpfq=<ODErDA<okA%s!(hPsP>>u6aM~M=ueHj
zKM1I3(l!lle#n+N8Ce^0^27EmR*rz4t`FN>AU0*>N8Q|R4aYQS7~M|vh_$F;4v8P9
z9=51RM%(A3Hbjnv`{XCK+kFQPkjLWaBX?1a{M24tc)<nI=>J4MnYl6I6ZeDLMo<b|
z^0&!6YcO)s1eqU)9WDx}8V*bLg0w&Dus_VVvQ8EnsGj&9825XTgJ+SqVVyEML$SE*
zVq3=W@k578lT8g`I|Z^OX;HK(nFG|)*kx#-ApAARGMiHs{3MnKrsa`|Ix8z;2f^&)
zVkB4QoYhM2SToz$?Io*hXyC(k5R%o=)s#`gaX{8MV%!b}c5QsD{P;pRv9!v%^o%JO
z8E3pw*84M7DnxuGuy(s_h}naBYHV|}bqs%<V%cc_KBxCJJ!O-VyKpr!y$b~>HKyV(
zUDr0-+EL$LT6{ES_u_%t1Y7f#^lS;(EZB(ftTlk5(%3?@3y03Ff!1&@B+=g%Xo5nY
z7s&SbHJL>3apQ6ZT_->DF*qViR=~SlcBC;D@4Ow)X#F-1fhYV8j}shtL3zmrleuoQ
zrA#QACOZT7V94@-SD@D}|5TYSbhm8h9H&vcZ3=CT)ySelH)~MXCu6#-bs#^tlK^yw
zd&pDKoMFT2XPuW{baSnw9#BmaN0ffr7e}Uw7M}IE0)FYwmVig6GwC)={u!rfq+_%g
zP}-qkfClhQkhOnhhedPCkP|^I0`EmN8t9qEz+VU0U^-#wFT-Pu3O|>?M!gNP?Sp&Q
zFbMgLF(QyjYLPqHD1$s7i!XrIQ-141v5XM4HCu5BzmO)}Zj0fq$P5|_Ar$=1Nyki9
z#(B~3^5&awzD|DcotgO&^Tu`kQtS*$Mq%E#Vk5@!52-a{;c1s$EH59mmX5(ae~iNc
zx{5fd8hOPIA^k4NApXfvgcZ?V4eEo!)fUq9XM5Rqz(w*}zytx=5Gj>f%3l(|8GlYG
zdX>B$hz9p_mh`PwdGoK<Q$G;IHw+BCW1-0kk^qhs@xE{RI&_Dy8?9I&m3#B#Z|y??
z{hLrxaIw4<Ynxc!kf*`s@4f@%TU7W_#8F32RcgF#dpI;#g-YG>5ATwy2D$;zrS~0Q
z@kImVpWSwUTxBDTfdw0Ym=C>p-c7((?*}mJv`#P?#=oq8T03!%zSrSbll(hYnnw<^
zoV;%@3t&eE82%G4S$(uzK1f^4;1e~`${qvT)N>h2yEip(k1>$^eL?lk{<7@%p_z+q
znH=y5(*Q(VYOnfs(BCv6R*F_Jy26>A>rfhDE#M8}i&J{opV6p9n#ts!M{HE2GOrft
zt}3jGQdXvyPWB5jJqcgu?yjLIq@PMwx^!3l$JIThYj<}>gWv~#1qWuEbnEVSGqM99
zU6B$Xgo%1fYO6^6QQ%ZQ)ZO%0AeFe~`4@UksHno0dyLHjkzZ6>f4IA;%}8N9mn9A_
zW37(u?*2xw@W@Bv2l=?;V&>&JoS0hqsLz5|CmaZ1rr0dU`!mKy67i_k{+N#yl6<>;
z`2>Gsnb#C`ICGr(kGHRe{m{7#y8>G&{<6}?iQUa3Z|vp6<P&L3&3^&6iB>tOyOUfs
zCZ?1WpNuI(7$r4o0T*#{cUL>r!B~nEYu-e66(hHTTRJ5@TaJr`^mLniI-pUMfq@q_
zKI1p+`5AC(pNXNMNugqGJ@D_e?xFq4AsZ|7TJqU=GJF;ywGoLer~7wWrh_PtGygf;
z24oQ^$;ufq8C(a@__xu^;`6ajXdyk)$nz)>s|n%@vB*haw2@*h0FP^i=L-HbXow3&
zzL<umBQ+#PoaVE-yK7+{3Af?%t)-sMjv*(3&Fh)IvNTqpG;&T13F-%rgL8dHNx?O!
zMEP^|TE$(oLcSF6D}r+c`0HNLKdQmWdEHeB+xMb@(mS5_>C;EPoW|vB6`7@vAvzB+
z<goL*yM7hQynL>522_B3tp{R8o%D<Dh9T`$|L(4?M0skeC*^`Tesr6GTFz9K1AJ7w
zUdfxRk_+uqvpLe>J<uBggfMVGB!n~NMX}p(+Q#l<BL)R%3rJJdU|A_&aZp9}BPzRi
zM0gO8Z$>UoE8K!3_f;Q=&LNB{-3G1ECDvk8wvvGpUrUGGwA1vR;LWg1E=@<Xg=7Fo
z%8(A&S!8fv2n9PMg*=)#T0`vUjbPr9p*}Xk4piI0Wd6g_!Z(8ufrx>O4G+LF*(guT
z2=AUXtr1C#;|NPA{LdE3$h4dlm<+>R7sx1Ytr<3W4+8Oxwv3ML8xO6+7(=03{6D!Y
zFmN2GGU{{&1$^QTDDKA{wLpvGrHr^z#z#9yT$M5*e$~MRKqjV{Y@>;ae6j)zfu9DJ
zK|Rj7Q6>dg==Fy8l(-74A4q?jfC+`(g6cV3xzIW-mCL;ai?42y$w$qPpeKY_yK{<v
zOYVk-OQ<pA3L70kAq{e6^aGg$v7B+Ph69I@r0?qZ4Lfrwil|Rbb6pDs03G9+<(i|$
zRvj`nA9`51Ha4;d1>VSNUgv|sP6z67eVjI=BCLSZ3(j#S#^V75+f@{FKqX%@MOYA>
z`2!Zmd^9i+PRHoBq{A^R)l7u6z-6uHq3|y$qI-YoQL`p7Ap6gJE@f%g)V#|Z0_&k!
zwA|JMT;_p?>`~$CxqZh1)K;BlLuBfi@EEI|ZgeP9RVkG|Px7kdK1uSb{F&ZiIEqZc
zP#x{Sc+sf@&Kr^om{jIw@HEs|xBdDJlpE5;$Et}M9f*&Z=2U~Vq}G~_9#JUy=vnil
z*bM3n4}fJGq&`MLk~jJv&+zuViz4D?W>6#~N2p~g3Yq1wvRj?K@*byWXhUo+ye%qF
zv<UQZwlOc`^K#=+ODBW1Ty9EWr@I5RUl6K@>PzJ2pm3lYaO8(Q$}<%4a>ez2i><N-
zk?r4zS^HH5PT+sA_uk=kR>!;OKmhkf2?PiU9aG)FnC>DMaKp0cZA*Kr*pjwTu`OEY
zp|{YhTqXD3Tfj1=21p>(5K4ehLP>=rz<u8-OL9)mbDn$8ea>I!cTW7EZ-3t^vu4ej
zHS^AlT<M=9Is`I8LFq~!o+}8mm#d5l$Yh+Ct9RPE8?@K|9yNW9O^GS+;0(=jO4r)A
zIJAo7x}Y$$03s7maY(MWq4hlMJo$MLxRSZDvT}nlQ%boaH^vJv;?9E{yD8dZ<^zay
zT!-Z5m>*PSCPrP2K5mIMP}8ub4zRo0t^Nx&KP9Q^dgSEWVrLfT>!DY|K;;dSBtY}D
zY`{r=hjTZC&=X7L&Ugb7pPB@*3U|ej!S9TM{zWu9uuc_!x!dSdhzpI}W3(7mFiP$<
zK}_#>mE0GNR3jrX2y?%6+tPSu8H@=JSPQ2X3?45J#(eacVJ$<_8G5xmw3C@+DLvgV
zj$cq^7ysgdrHs@h5Bv89=ymvpYJiWl*+_IxfbFnLQ>2(YY6BkBzz~JUVwSklAuniF
z8sh;f1Zvf)X|W_cH}xbWW|mnarpFfIEf2ng8krHxjOziG89DDzNgz#>I59Ow@>GTB
zo@L}X>5!hs%WVIW;-2kqfTG#tYetj63~5el#)5r#4`OGIEyF;hLzo-mu18d1HF7`B
z^Y`aKg=PgP1_;iN<v;;qWL1m5G^${s_lCGQgc2>X6BLr0&RtoYXlfBrgr;zMI0G4{
zD@)SgX7E(UePW|54e~_RcUcxsYH47mJ!N^CHgETOYmpUc;PX_zGKb<yf9B!P&>hyv
zsyO6vnFa2Db^JLc&xFZqY?V*1r>$abtRXCDjBg;Q#kSSD7&xXuOb*b1vfiKB6}$-t
z(FWO&-kYgc5ujdWWAum}0Xy5KASV*ORLC!FY0ien=y7|`ygG-<6W$c;77&ZPQ!J9r
zK0tG(;XXujADC>ZwD__xCD3}=5))~YBOLc4*gu}M`O>ISi^79;$x|*-ZesKxGnGD-
z<^~?dGbBZmr$ZHm3kN(8%LX>aGtOFrhs(1uvEBpv^^)h(@E6iDhXZ1fJRjS>v>u0a
zR(@rJIW<FB@j@Ci@-<_=j(~&S^I|lm(hfTndd5rs3?>O^P*8rM)c)FAuy08*0*8py
z{!RP>9}pE-LcwfdU$#NlG~nV=A+OkMaPgMkdIuymQGUGpyc#Q64_AIxUP~k9mZ#K#
z9*^2`Gvsyq7;!xSQ)Pj@L)&k7g9^AngM?kYX#_yB!OSFiE4|PR-XWZPrTos$&v9qV
z+xF_@a^h)!@B54@Ju5=4;8hI5f*y1xCw#|$=V^iL3vgk1HbE>UqbJ|>u2>0=@rz7Q
z|6VK(z~xNh_P)PkbmXmUjr<|ng`8!c3{;}Tyjv6LeBcx@7h&XkfZbR=^wxm;)GR)b
z4DwOv3#aBFL^H64${$VEq2fxNd>s5{t_-H&Z;(%d$qH!&`!uy`#pTIlpZOS#Dzw)!
zHEe52gg?dHY4}w^y;wf?XFN-g;2bO2BwvK2EJmF-p59;D3XIfxYhU?H!*xpZg8sMs
z*`|Qh0CVLpG1}e(fVuyxvClXnd2!+WTWhC~GgPOh%iq)B8G)(>CM+4slwwdo3ClnH
zcglZ?R?lUs_cd<suf0tg2M8c#UfyqFBIE08CIPV`|BQ~YLpbWSI8Fa$lLL+trQiN-
zUyQ-s&<uIsre=BB(@+!nE~GTB4JBZhtb88}26zgLlxm9-X{XkcLIr*|$ad}2nsvx2
zvVA+V(9l=%gLWZ^XqXR|Dh=~Hv@@F*1&CiWWk-LTS6;>g69#C-J$Le74Clf#y$M91
z_S3GNTh(w*0mUZmQ@e;?LL1C9)ge8b3z22=p<{Zc_Hb%d(#hJ&SHQc=Jo=se9lL#U
zPp*|NHoKxc5Ds12xz%7GBHXHA((D$Eax-EBu|8n;cBamlEH=vlF=;gACmXHwNL>*p
z#>5PZWoLT^a6Q;cBgV6`i@&EJXe<~#xw|@Xc#R@pMom@pUdwLnl4%fJ<)9-71;iae
zK)?d#zeA}6LGdpQovlJec5i2hJX(gA${x{=3ScwZnb!UzAJrTM>vLh2$)5f#iVA_W
zd&RgB^P`b;&|da7a^rdk7&VlPAA9@U11y|?40ogvm}MVt3s^u1w4VfBfY4+}G6Ca1
z4K^rY6num1Yut%4x<HW6lKs4Co|TjumhKfiFZ;Js1JkGxqve2Tqc#Vbm6ZdxnJ-60
z$QfNK2ibTk=ore$!9F)l!r-o%L)vZK5tk2`doZz{sUfHrJWt>dp$ge7hx(^3pKwQ<
zDTi6(EsaGru<0Fc6M>GS(bq@Bys7~`&C8MgJ(-IjU^)YjvaazxG}2z}+zN(Be_Y#n
z=j<KitIk(5K_P*kebN+QR^%ixw!SuwdKRrNmVQA&{HAwEhUBR4|7hD0WsK84#`;FB
z5Wv6tr$%QmH!rl<rE+W{1(RuOM<4|ck^w>efmP)wCLzZ;Z+dVo_7Z?!FnDj&e^<)!
zsU@a7;_57duviIRJHfeyRUVeeiMGHafigJBMhXeONlx~`F-^O|$01}-3CaxzYI$J%
zF}5r(gZynMOk6?S=$FgjSPG=f0D*?Yc&S5TW|Z^BGemXGl7e`BGD6EMWT-Wd`y&p&
zVctA3&vZg8g#B`O^j`{S=Lm1y95wV9GJz1%!*f&PcsVte8O04yKmt}wIW0XZX5o0Q
z{77$w#|M*>X$B<PFYLc%RLC===d2@GdUT9Sui)qi)E$#ZL0pNsMaCM9`}FH0r#tZx
zpbLFzT)gZlFi^)_3S(l0xIK=KQJPbYqnr@G56u+G8NSM@-e7&#0V+oMvRKaarcEhq
zdSX6qEoa#RBe7zoobA8Qh64z)7!rBDoRbE}N+;*Wbb+GE=5cnM*QQAZ3sHQZZ){m!
zs|RMST;T7}ZfF4qi{!%S3YRR}^bn3K7uir+8b^<kO!|fuB%#-%-J*8x5%9|+f+J(9
zn)X<?i|?>(`W~eV&^~>?Ie(iwABy~z6v`kaMn8Z^P<f-&K4tnZjR)hgR@QE-#mW@c
zgZq*QEl<4=Fs-3?GSQz|c3Zj}R?DOqC+u_-h^?re3V)j+87y50!b_!(rj*`?svt^z
zndDnno<X}r)>Ow9L6TQ!D;EbN499({#y$nowpMDrA6kLndd9!Yq%P=*-QC3hNqu}?
zg!1c?qiLGEYGjK2hI<<rZmKs2Qf{H6Zw1bnhIn879&(8@P7_L4TqTzpyZ1Y$mt5vs
z3a2f$6gktgZTZq&9#RbeDx}VmD}qE@fOrRsEx}#jXHHPNZ|aFk$p~lIl|dsWqsSo2
zq*ktK)4&{RN6?I~Hb!P33Xjq{xyD~2MuJO8_ujQez)8%few|IHrLhoooaFlSHegp^
zFQ$I`dCZ?D9KcpaFX(@0{YHPEwhX3#TMb9_O+E)cqs@u={Y|dl9B<=1$F&WjbEDki
z<fMdNB)7(Bs`8Wba+~!GV-{c<h+T{yffQONcWgr{xBuZHhp5FM6>bwOgNXC{uK2Ws
zr}-~@9E4HOcn)B-+}*CNgF7;JKsw(OpF-U&lY7%5W;D9|zO(~)g`}sDU>Bkg?+-47
zjosj<4_LQ=uOlQ1S~v3oJQ%HSz7BShhpcr6Gi{QGW0Mq&mPZm<7pb8XmsRwHJZJK#
zPY1br)Z)U7Dv#Nh<x}BbY>a-joCEts;9|_9u8hVlnV!yr6t}};TH(HAf*CeWEiGIt
zGyQE^AR}Ia78@d|VWyT@vEK%Rr)JxyP)U&7RhoRT^f1wQ$yknNe@3SV*Mo`K=lFCf
z)OgEbl=v*QD9`n`;m+49`6gV#=C#v|!$<uu{!ixnx4aNBj+jc)g4nkh5tK1-EeX32
z|IMWE^JQTy5zO5+vM4PB-3XQd#>_5@V-{*1)QgEM@wc0BO`(H;&TF&iEsc>F+0gd}
zSr#k<w<HoGW0{u66T~%Pzd^TH(M~T{6V8&A-ZrU1gQ>ZhSW<aa>a{gjg%(_u)!wT)
zwhdjrCgxXLUtBC}O`!EW@;q7Rgu%5{W%vS+L=j3@4j3?HeQJz%G>9x%B{rmIbXe-L
z$wq$$T?1Aw%m1}0gbE-Y06_fGUSUWPG<j5#!b~3z76sN4|6iVH=N@SIXaGI1;*6WU
zZNqFSB1&4KZE93G*%D_vMIwp%WMatm(6P6A3swk>EHPW2G9vT^@a1W14AI-b3{8?}
z?3Zc?!e_l18X?xobH?&4L{lZt+gnE-yr29khzS3NS>SoEGE)Y{#0xQ47(tQh;zer|
zS{nliU-B*(>oZ$^?IT5aAG1ne*;mVNtXtd<05W;mx-CJfgjb?lnjCZ;SY!vqF8wh2
zt$oL|V3Xw47y~jE!IA%35FYvnAgA(r>KE5He7^=7?;AccmOff8ZzdY9XY`lq#BZ4d
z!hw#gVj&NPVF$hKFY`o-i^K0@jmoRa;QFtVcVY}#_@kf}znebdMlws)d;ZL#0+_^R
zro8XZcqz7Q%>ZE;S#UX&5B_6k=sgan(M<W!Kb(Wp8}*-%foO(&WRt|LyBw;<AA?HJ
zlYp50*gt4)nV~2>)`IxNsSb^h$~ytX5D}LAlCjXA`tOk4+)GfRiL&E=^WO#XS+oh=
z;u@?lrTd@Kj2L%s_pHyOALL=BshVnS6Y`i}<09<~n{l6``^uL#>QUwJNCCk5m5Gn4
zGWl~%PD{7`B~F^r!{x6*dqjq?y42s2pjo0p9!&6H<M_L+>*m{@SYIjsh&4nk1P<LA
z`8q}pw=m0bGt%dqXbb9*2kBy-JOut}BdExOd;b+Dcufwd@GL>y|Mte{y&%3G#E@@o
zGD8Q+cm6AHS%e`W7qfgH-9w9Gs5p(L(q8Z4lpQ(QuDxTeVD}^y5#lS`xA#jjU@p}R
z8>x~X*f)q21U7Al_G$(Jnzjn{9e0esN;4>>v{PyiE*)(0=@??W_GWgW8btIG^TXRQ
zpch_gZ!HYZs1LY<w*X5RD6Q(BbWG3Y!dzlT_D<;;y)X25&6LjV-2yd~<w8LIF6{%}
zm8y^7bAqwD+UK+;A!G_MMUb}L;#n~b|7ITOo?4)#v59fTWK@u;hYvvW&R{}xPwd=Y
zt$;0!#fXAKny>bHP?b!TUA-BV{8$_X>#DtAxAtmrIb*~S`JoX4_kHSXySI-kQ9W4@
zyKfI$1+rSH{K&iKawC)*3iJ}$GggiO+w!V<RKeQIU$a_-w24iVy?sI2n$<1;BN0%`
z<i{~HC_glu0@Wz{*f1}K(*Be9S#uWhG+ZM;ZLe0&md0NFWZ&o#iZ;~<M&9k`&uFQq
z`;DSy|MUzQhVdTm$Z|mN$`;<>NN5}wIB6}9EfRbAH{9<m0)Y&Oa!~uNR=5Y3W-J<x
zMqXnM_BZ{+#g^G1hZq6jhJhI98C1XosD;s@LQ^Yu7C{V$#_-HvuLi2ad>Jt?9OKB?
zqo#ElUJv)ys15?0hx3qHT<gG(@YZlYAV3}()=W7vwZ<)hwp0|dwWFeMh)=p}7vPrA
z%ip$pQD}q6(j>k87e%R|9bAa<FO0&{CwR!qz&A+W7zf>CI@86{&%bB7K{dsxF0~E<
z>1cl&R|}M7q+&2vj`3%aA)d?MzrCBCOUgC$=2&BgBGg=w0qwUQ!vK*cjOf2)tic$l
zbAo~Ouwq67Mf>}fKF;ZC79{f&gxix6?)Wr9f*v)Wh@4={4E3R`PEL%Wjk02cC)pMq
zIbv|Z=)wXy+4#C9H%U&hUj{4wFe?L7%WKulFb%kYK|xQ1WwC$@A`$&93O5Km?3gAQ
zV*QtChV}v@MO_{yz@biPn+A8TChysDjSNejHR)B6IFsQf4$frZqZ%1u1FfzomQ%eM
z!l5I_4=){S<g^%HL{7;_`>ejYPDX`vgaHADPUQWM2lYke7o!ZXjIkbZDkH>M#(Fn;
zHrZa|zqk&Zp1J`Y5O9jH!uGD1D##le6GWI4+SCW5YMG4p3EJ9$69D;`kmg~<OINF#
zcOz~AXV_9OBQOmmXNFYJtw=oqCkWzX=^7w>@2uz@S>0*e)6#-x$L!$4&1DcLcTOyo
z`b)rcxcHpw&$t%0Emz8U{*0I@3IpJD<@`49AuI~pVUb*5)T%7cW}#wVn7*<Udvudu
zWQ@SuZlz@W1s-3_?1dcZQWVcX53VGGjf?#m<SwndkVP9M8w7wW+$z4&T1>a3?petN
zlcVq?B0Z%f2to>KgcKxVX=;Q?xoGv&=Ap?ntE0ax1_SFP?RHG@RGyZ|aCh_=sozQ_
z8e^e8mP`uakJ3sNMik^jBNVMNnuIpOkP4b-s{9%4FbwOTgM%yY-L#cpnq5GugM_s(
ze_kBZtgnDyzE*1Dyvt+JD>YG4txuA2IPT0`T&Qc>dtDk4on}q2lP;6`7;{k$H;Bpc
z*VuB*U`irdis1q27$a!v-&A`WEzQN2X)vNptS^;I66KhMUAf3Vxip&81%_o&MLh6%
zrYFl~F@Vii<05-`{8g6$66*>dKG}z9Y2Y%IpLtJo4q;fTqUb$GuJpGxiwwrVW9_Q;
zTYITEM!$*+1QXF*M*|hF&&C&;Vg$^^o0ILat5aKQhQLjuNv=sVAzzpaj1D)#+kl3V
zYh%5jMCakpzs?vXUsTi}*IRpx;L9Oh7O>O}^7EJ-?h_2aV7ZVRLe`8$;Uc7|L~LVh
z?8%Js0i@z4lNLqb<CPfI&E5(P7z_`k0rM7PBl<FPa;p!U*13)N$b5{<TV}N@Z;N3-
zrqVrzM8k4>YLsP&VLJd)<c=5!!v3|8$(=Ue3gCG(t?pemU)1$RA%LL8;4!0zju<9)
zC;nj2h-xzY7t-|}TX9f~dd$5(SnATWB5Upgs{WuciWka#-VX1=jJ#y_K4Z=Mt#{lU
zvhsirzX`IFUjlU^4|>aBXH3&T)?{r`9vOw@A)`F#4z)@#tPdM|4}(eYk!Vn_Tz1=d
z)Sk68s$mp?yXg2a?*u!WET)RD`BbS2&Xq=MGKbsPG;cEek?SD%ru#FiFw4;`kQp{F
zU^|z|%=jH#PXq`Lk6G5_Byh%TBV{do;W&PqOdUV`P-%`$HH1l<=d|B?0Jnr4bPS;4
zmY^18<^ltFpmQA5bB&Q$9x^BMtQpXb!hC*=1tv3s@MpjZxFF5R;d0m&3;h{XI5Ur9
zPK&%H@<FpK4!K4%4@}*#{49xW0EA{mz5$e7YSSzPoMc%D$<c!b%5rZWK@S*){wpi2
z`9e$Cv(j0{G|9FwgjMz}HYKv^<4VMYKC6Rq%j&C2p*5`W_D$olz>Kwt5Gw(nS!aL2
z=F5A`dVilcB$_QYKu_7Q%?L0KFfH#UGi9SKE^^G*$R=xtXGsz4V~Noz9Bk}LLU}xG
zkyZeZ!(u?<rSgPPsvly_Hpk#`zc5VCd})aZ^00(YVN3j-DJcPNQ=UwNWs<P4We{Jx
z)dmZjaz>u=mbr(+M)7p}t$n~)|L>|9wjlS0JmUnNrPT(WjZxD?N`VDNd@hc7Vs4>l
zLo(LN^Tr5FRGyPx1y>9kGge-(p|&)R$;yk~9L?1Pn>67v_ENk&cF0iqbs{xmlqgvo
zmMp)CU!Z(2UJ4a6T-v;Bo2#N)UWsKxo|sDcZBP>iOQboFSMAsGy3N<%bbHN~rf<Jv
z<@Gel8EPS3ij(<`*a$FzAd5WXiQcr=Xno+`0E8+hZ^dh{qZ%pXcWI6@)Zm8<xAJ!4
z6T-<jEmec)G=tb|mfzc8@)y_3J4UVxF1_?Tc{j<)HhI)>C@>?(T*%S)V&G86d0G-_
z|Gv|SHCT|71J>g(^Zg+P26+yZi1tYNAU#u)JTBa8<U>2u02|WuKp(Xd9$qO5LQjS&
zfAo)e!w0SrAJ3PM?HTtmSemtRYvmK4hUI4B(3wjVcRCDWCaxQwT8E?E%KKR&R>BHL
zw>X0~$e(;bs<f)KBR-zch0lFJvjE!y$_yZdwem%<7u;C#Wymskrg4HpMtl{sgG$5{
zvna9QZ1v|v8pdX+gqFVqgPq1(#$Wwgp5rXnuu4sdtepF|_FJ#q4%_elKr);E+sFT$
z0`hm;R~Z~8|A?il%H#6<bsHYz(pd~<GD65V_BKxXnNY$0nK+J)>1J4TSj!aY{^Vb=
z|0hGiGX&JXjp4u`r490}GZAf7#*wS!J8#XB>S7D1t&05ny$!d|fB_;M%sfh13?UE%
zUT)XH&7*MJ<SXIGu9NLMn1z#<86xV4HnjuF!tQ_I&A@QWk)yGVvO@<q6*8GE>Y&;y
z@czID0;fDxzn%Ou^IPLn_*mL?a1$pkKnmAv-$Bi}z;(^6ln(v{eR7(N0jFQb4q*e;
zD@JRY4(a5>Zfluo3v=!4tpVDuN%N4mS-SW$+N4my@d+sV($#wio76JB@Q{?<(hR1<
zUBkUOSGxN%hy~TIY54A8&nVPcE?hGe?c6~PqkWF<FT12q=xYVW0y@&J_H!YO^SgC$
zgEXwT)$+stn6Ja5abe%xn=z<9-6Ku1hc$B=tRFuzO13oiM?~eG9o(Wdm^BpYWG`>k
zE;((MSoRKnh|pv}_hON1P@8QEk^=1$t*MHE5(pjCFnrvh@>6d&*o7g)+z4xBUmtD2
zJAtT>{nFdKJM-kGk)3bvALGMcyiVwKT?0T4h<3nsG@~*d*?}G0wp>a82tlm}#Vn97
z7Bz*DHdYRfL1#*9a!@V~>7X`V;5PEo)3_Ib*ziJD$f2<Us7Y8ShuL4_5z1a#CWpsw
zytK;^9o$gF6Fp<e_~l3+B5$?4a53-lQL#UY@_Fo!UNH--8vVyCpwv4(18Cb0C+U-(
zEk>v`9a9UXuPp~MQDCbPveC~OpZ#qEOY=kLX#btKae_lNr6K%L^f)r(;4#KD#~$5F
z`UletnlN6Djm~knvNZHG+|LHsJdk>xeo;BjU*TSW)fr~{nR0v(8aH8zlM~WA7W&D0
z&nLz+Sc+);ViA>-(wY-AY{qgxf^ii&+1_r^e4s3<1!;1M4P<;7>e69Z1_qBJp9kH|
zGRS+NW<Vqxug!yP$v58zfi=W`r-6Xs=X3FUL5zG9l7OQO*HHT#Y8bA(_<NWUDpPlf
z47Z<vTUm_%ZIThj7N976e{yR23m{%O&AT#x9<8O*Wn^?kjs+n{`5IZOTilMJ%#o-B
z5-)*acyy2eT0CMjHCB6!6O3=u<PY-?{zX+#8Jl{<`l5x62);Yr-wutRARbPBg^Ww>
zEW~Y^Zxc;V{A^*ccX<^GQPvTpS}WsYZ)F=$;261>D`bNA$0LR5<d|pX41Y#^J&pd1
z^f@!$(~C2&JBV@hXDmw293a#n&$h*75`DV&&xy@5Qf<fQrj`+j3H53^z=7xa(q4<N
z&kx#W^Ox`jbU}PRD$@WU;Dx~|Q!<Q&E0v49y=LG)$;5=RW@4nU9WcQsVk@O6CO#Cw
z@^H&w;>EFF#zW|2W1o49!aC^BwC)1I2OH#c$=MjoP_jTu?7t7`dBjOl>hw3ENna3a
z{js`MSu|f<nU(U8JR@r0!<?7~J&mR|WvWRrXjN6YVP{E2TK1*11As~B_gjv0uF~HI
z><t~7D5whBjHso_r&jW*NksF5)+j?F)$tN8@7Nc0a&gFCpgfT60OsSG7%R|OQ>51R
zO`oF&NL`XuNOwdbr@pWRypj5#G5EYZ3m+=beL;kwaC1%Z>8OJMTt#i3@S6=XHAV{L
zEYnD@#}pfE<zc(g4Rc8_MjfoV80DpD(^-7je~3@HEZ&E7FT#;=d5okEp+;ADe-^c%
z70u*l+pHwyE!N&VidO~?mO!;xDOVXqD7mh-Mr)C#D<{{aNs!!TK)W`vRjMzIDYu62
zT<5<N#=+!4mU)@QPMN)Fg<Nla)dFYv^Pp~FDj;`5%xN$Z>)jZC^g%kWn|v&R(MFGQ
zbM%D`L1Y!6<lv^(YRI?PH%KjnYZ`Jg(hq%Wq9QJ{a9rHx?HaoqHthBw&)^Iry6)Jv
zYuNK=%AHQKY+mm2pTW&4sy0BA{6(}-+)xMm<lP<Ii@NzXaM*1)WN(9^<{lf0M#vBX
zeQz*sRjwXsqo5_+=MxMpwv~R>s$Q7pM!CPk*6y@{;3$F{8`)(UUJNVl1OA;E-Qwh@
z5!Uw1gEl-bf*=q10YSUYVtF{Is&F3aokXgcP-mvFN7A5hRdoSvIuSR}TzND$g&hHU
zaaRtJ$DC4hFEPvv?@MDy1jVwqUA<}cHucs>S1Z$FCj&?{MP|f-z%&E6>#CV%rgt(2
z%VDuu^JSLFw+SN$j+5EZENY-kq&^H=W0TRNI+NAHq|NclC`S6=sDYboPV}q=JLg)1
zFnA;S)jPDizt%#K`S!ckFIf=FnyZ?a*ZfQi{R{K*sz+89C91Jng=sLf+AWpE{x+YH
z%NN2rzGRy<qWPLfCEaWmthhA#R38<{Bg<_3kchNpEl;yy(Ef@v1ELZ&v{6=i4|Y4F
zPcf7^Tjf>$4R{8s8#!l*tWI;Fo=goI$FJ(iYwT?n?<lI(OGuThjh$%OM`c}_r&VZy
zK}gn{E<d6|HpJV=C&6QKqy5bzn21x^)M4u`>Ux2Rkb#*BMdQf=<7S#3ig-aJEkvx^
z<33cn-I+{FKhlHfiFg|WX8<qVhMTRk&9`HN-v;U4;$N%Ag0E4z9%SRCY)hO*0Lhh?
z*5~BO7&kUs4oBG5)I3w95z8<`o{GU}fI4>^&Fu^)d)h?ch|w&4{7lG8cnykkS&EEj
zW1Vp(RKgrpE6=4hSVHvyn&ccz0trCm`P4J3VZ%O$yUMR(xPx-#6G_$={3W39IVWfz
zE9FHS1b`hdGQ1S6K!sWbal+{FUpuexJZH*eRtGbDh9P%;^B?nL{xajDm%TMRI_+}u
zO3Vp0?13tSD)Zae=SZ+kby_3lUX9nX5S7&&`)fW7$~$;-G?$zGx3Z$V9z&-D2Z5Pk
z@(tSv2$-IM1LaMV>XTp}e=BxKuCk~`ei!n5B<%<CwtvMu`!nVDiDnyA-$asKdB;AO
z2&_~Uj*xfbC>w=!d@q(redXNHOlh_jcIWqFIciagl2iIayb3I`l{;V@^jt`Q5BwK{
ze1*Y=iH|=t*)wX)kOKKASO$PkJ!g>h&4w}z*Ul>WIOavgzfL|$6oS5@mesZLsdwZD
zB()nX@jBcqK8w*JXAz)5C`;;}b5j8}^+YpHgU{n7cvv-^#~0pSvkH6@FLO|ndaZmJ
z%M8po5q=sm@s+<0u2LV@Qrcm01D`2>PGp~s!4mDiM*b2L&?FAVV}DH(KpY~Z65;K<
z&&`&<#n?t5iANPC@ONKVGU9UiM~oEW)nnxSCh~Q31KSx<pv3uplQzM8M#%v5iGc#L
zqL)W#fB*E>&^44Hs>4)Pur=jh-Wm)(s(&tmA?M%z3{!wrYYu}Q<=YS?V_`;sG3GlP
zZ4r4aC*Q|p3Z^201Nx+NRQnp@adaPPOy71L)y!}yP4C;=@1Q0pF$<>)svh_=v*MT@
zze7hirXf@d0&U0i3YZO_0nZ>ibqq7uN`ST)RA$hb1=6mgTF{6csgU*^)s~b&=6&q2
z4*roT&!(i<igirC&>mrd*9OV$)KTqL69DMz9Mb~?1%=5V4YSgvqnlpH&J1&xt~TM3
zdB6?2r3o+Q6+=Os-R<|ndL#nu5x=)A0z*+~Hp|Y@OHHnnak1Q#yBI+b;;xR{UE>u%
z--^k&&@y-PrVL8QGjO5&u(f$u8937ovb&L|0LYRC*~9*5X&eQW@kbra?5T0LU^~`#
zYI~yYnOX~*)B@QnErTrpbIww0pCx;zx2dBlTgs0+x)rW4mtmT?ef%9H7zWE3K;S1q
zP=Fo{An;Rv3+bk4D+1VXjqK}V#he)Irvi;kKv_ix#eP06{?3zu08Mk{>>pzo4YENI
zJ0Q(K(S3A=_?A1+`=BQR?P6o;zZ#=Q$NoWW#y}Q~z<geMsg)jVoKp%M?IDR6hJpo2
zq-PqU#t>lbqr}gl>1|Y@flvd)P!2N=VBkO&y2s(h*H|AwwdDx=vDkAsA88$zWQye|
z@5DqOz$Dx(kXtjPm(hVG+%!c&Z~Lg{5$8#tj&76&UKnV#TIuVpfHoQjMgd_4l76u^
z(65>BjJ4*E_GcV8Mb#RZ6OOUgM->c_{_(S8AE2nqvC(4$+z*IVf|O)%&v7=xfmJwq
z5jT8%q7Y2@=oT@3a)R%BW$Y7UTv#>u#h~-dmXoYas8q9|Qmv4atxZ(>7(M2UfpSW8
zIWC8i*Htpm_Nmr@801S$kHBI~ovL0yOy^;B2FG|X5<`D1mLV}uupm|3Iau?8j_z#0
z)d}Jv4|nv?_;d@3ZtEUcB*WqvVXk<#2_zccaqB+RHnG2<#nIPGS1#f{5uBh&b(#G|
zM#P*DWB}9cslIR)2?DlE3wK(~7GY`3z&+C6v`aMelQ9?3LSQmR#hXJ=1|C9wv=0vV
zdbDO3^D&7XnelC`QEo7+Fr03`v@{MvX+;?qjHF47<2)wjuh721)oZehPd#kZ`y)^O
z3HCec6qU&t@$Q)oa;7oXkqGWNtD{=s&m1CWN2hR7SIaqxnj0C=2JT-V=lU~pCk=$~
zP|A7M)FBsMDCZj)3+=up7o--D1(v%T>ehwEZV0ObnBk&WSJ;D?{~rfKrsGzNL&a4n
zixN(-n6WNQQBVqZ#br|L4fA5kELltho%Lt5LjvXowzx)eiEb=(Lx=~JUglvdv0>o4
ziWC{8@tP$sg1lsg4j_Z2%t$tX*(J;4tu0sS-T?dnoW!fd#GqRiTChxtS(}axj@u*^
zLHfaiq|&-Bg^pe)RUx(ui;-%I`8x9Px$2zMK4wU@&z>H@RdTU4fYS#F(Q4v#^=sHs
zSR3QXRb$iF#ke8op*^UgV^I_JK{9B8j2oK=U2U?z?WPDQI>d%CAYzJji$Edp=hXNV
zE5KMkY&dD~XWZ?CXR@H;CH{<}eJyCe0Y<D#<69cnA%#}8Oqa_#x<e$JtIPv20sl*}
zT%JCrl4~c_6*daQD*%M^v)E&`fO|qKyfXcblqAr!u8OJSAmfsn`;rIU)oBbsu;3S5
zh~yf78vy}o@<ifeGA;P%+SVgWvj`e!oa8idew}{_OFC@nWb^=9UhiLqrj-tQ^f?2X
z>(A2^HHQ|)xlC^GXPB@~8y5<>G57>|1q{A)Q|z6ZoEk?UdT$OX20g2)SZ;}Spg%3e
zdJsZ;t37LJL_7^`u%zN`K_x1dATv;JwpZl#;I#3>hRPlOnh8bP9wPYX&TS^4scPje
zpLiR^8rKw)%dCz2iy+LFtFfD~X+gNQVCK7HDF?#bj@$S>-Uoe-dN+XZyEocUo0nbj
zUGsv0ce&hW^q|7Rd=~fHY#>yS6Z-+5jV%`U4Xq{sb;^UORXQ{{GA7$@wN*Y8hb|NS
zLPW{K@c_xNaUIOcBOSMn{r?1b`k$i*_=`N+hR!hcBZuEr@|Zs}>kf6nh0++$)Y^pO
z2$=n8+su~;eK?BvuS_>88H{ME%&_BN_;8u&glLOx;oe}1;EgbM%}RVjj-g3`nf2CV
zH_tXo<8A}YUXzc8pTdBH*P5i+pV7-kd@?OabK*URw6BDyn484@d{`jowfTw|Iwb6t
z`RS`A)B~{fX37HlN?i)IvM{}|6zfUhv`7~DGvuMcwGY-iaMR*o_H08=mYBfKR3aDr
z()eNQgmJRW{s6!qIj@&{Gh{vPoFz~tR#@jS3NvDEW$aRe-}6$kDw>0qD65V5rSL}5
zZL+2f??L~DoiNhjnxI*m=7Msk*e_IZ*Tt>}-V|}E>(e&nX0v>9YGIKL@v-KZ!!;dj
zvC*G-_z8V@vMKnv2DPYhOOjv41HcT_g!i~nr|;1?7N3Y8Cm{}Q5>D{V)+8|;m3dfy
zdCO`^-OPn(B&)2FExt;XwLnzLlZn~v=3$1PHEN&2fi7FU1p}%m8bF>h&11~aq4IR_
zqF!lqBfD&V#v4}FK-b*PylUa(f#_K;&srn-qH=l8xs7yPD$g5T;Wx%XT`Rvz?WAxf
zuv>~DkQdShqt(E$=F~;z$%`Soi}Pi)eb>oLM!(H>GM3ldZF9-M*NjS_sB4zr7`v2W
zakKKWjj*M$2<#%S#L5pI!@#-U29ec}uYyZnbv{Oo19{D6T8T8juiN96#v}KWH_`yi
zI7&oZBy&SFn7kRoM@AH$s#$r<8k&p?)$iimcKaCut83+L?+E#0fA7=<Z@&|L9&ot4
z>smRbuV?77<--1C1QX-!=g51B<W}&GdMF9&(LVBiM|IX9w-eIgAj(Pc_40?<U!%(>
z$p?uhtp3I{d(-8^P%)ur;9vlQYNX6Z{wl{roreh|{urka3|&kC`*F~JP<=^>d}5R9
z+qb`b8vS6!9rYxir3QKGn`P)vv0I>#BRY19d~Ty<a*8VXBAOpPuyCM!8Gj9}K=OpI
z;>!_;lSeWi`Lq2@crp(0zc~Au@sC~rOSZC5{_4Z9lg!kzzooZZEp2M(lE3>e&B*RW
zGzl4z*)e9xKb&dr15bRNn4JdPSuHbEVWNfn8(&%Map04g^fMH1sOprp*r1E$pNU+H
zXq(`DO#YR|IS-mO)5R>6f2U^*Co<&Ow=s6y-&nDFk;VDp1o_S=lBM^*Qogs{F%b1O
z2)C3@VfhM!1GP7n``WIPn%Xk3jmq|&4C>ar<w}(ktD(IAz`u+W&WcaaD8EA|HIo&?
zyaW-pV<)wswKNWa9eJntV|00`wDT4;Un8c$#Ons)Z14SrMUh2{I>e{gZPL;HfcJNO
zm2`?<O3*`R`xGw4dQAjJt4x<pZh{zs+&LNGL%POGa64pSeB(>qI{9t78tzL}{3y>z
zcbf?IJrhRu@IfKJJnn6(y2{Q$KI{udcJg0NVX^GuZzE$cR0GA|%C0fcQJIOdTPHJ+
zG0;}yXlP6UfXM<OkZ{iEpUp#q-#xmfA_MtK#@oXNhr~1}zbijVtHByuG}sXa&t%1W
z`nOsV39QSC!+UjdlW2Wa1^keE$J8~ml$MhEv5{;nObePzT=wat_u&a;@{`mRF=J}i
zRK8p;KlNYipaK6Tu=~DFNumW;$$rs2?z_0h?;kTMsxPgT1A?R)3*M$$m>ifWmXZN(
zL$QNmrk?8??nXH{mL9&Py7H0+Im8>`?jjs$A@*q7gQaKC0vRuJCH2g=d8n@h^@zhd
zZDsA$@=IKkIXsp$S2bCVNb~1ut-djn-*RMHKw3M^U2s$yF1)1JvADKNuk;=xju8kC
z#_w&<P+X}V*eQSjAm#c5$;Rh!x9;0Xt;+e*Qt9UnP&GF&0RnY9nXyL|3^YmwK1we#
zFR>|6$0WkEI>9Wm##H70wrD7Ap;qH#6Je(?ZAexI1SNX+?IXwedeaTJOpcGy4$0-J
z<%Bjvq_vf?sV7F~Y7^8Qgn?>W1pK>PPV)Aci*ciz>@AR)nJxggBsnGeD9Kk$l7Y5l
zd4dTB*^*f_|D;+O+{umXBcTe{XHZULh_Blm*f2DLA5ztbQjlnml(W?txS)Vz^WIgd
z*OPa>4D)9UjWa*jd>L-5Ql3Gr&=?#u3(BeSS8-W+rJR<Ui#ok4Wu$$2_<4uRC?8u`
zC|c{4j&_>aZll>!OJ4B?!>hWC@rDgj!v%*i&ANs$mp<pyQ`gL$Zy7Sj#WS!8lcf>9
zRoKbR5QWI)dr7WF<JZOqSw}%hge7l6kkx7^K)pS~IASP%t(7y|?2X8;N%eY>oMmfO
zUJdDewoSZWuVdvL9|!>n6s?eqa&BT1RejWGgQHFlMt}$8yx6=MI4v74;qh~RjGZAN
zR=O`I7Z~%Zj-z$1F7(k^%8}Mc|7N);rU~0MQo2ee4a=~mye<?an$4sE8P3J^Qf%KL
zjWQBQ0PUFd9;m}{g}4c^a&Xb){A+SEt^FHu6D*0%RF|J5rQ48>rzPE8ywaA%;LcQ>
zgp|h`Xnh?9`%W~mU2wWgYSWEXE0`-4F>R<Ha7REDVSL|gSOP24lz4xonRuR5h1ecW
zGrHvM&w4L(Rc(FaM2(2Ri_?_bMv0(ggVb2-w6|bUh*?3VI-7EFInBa#ZKg~oU8~c!
z-lupg0;g`||C7na)0E<n15<q9Mke+0$ka|-HHMFMUl62aj$42pRKmh2XWF6qT%;jL
zs=j)J>UMJI6Oz&)2cG3NluHwNG_qe&5SRHglL4yE$mMO`quxjB=@r4G;{x;}KMP8M
z&Yn*YtFao2<CTfyxgGVI?h`Npb(KxNph`2gT<z^Nfr)U-MQe0Fv0}_MM%_`st7qie
z5J1(L1_a*vNAwIZwA)k%0#>gNcFLj(N)?k$|J>G`e-N;FLmUE(bf|}E@y2LNbtIEK
z#^pCzp8%wBeQ!>5;VRJHF<)*8_ClTr2#j0(6%vg=5R|i&+u|84iu8gsZ@F4-H<2;m
z*j{pn_m10=T0LmRfU*RG>z&b`nvcm9tK=?!n|qN!8D_@enKNI0;cpucQNAvfyVKh#
zjWz~Aqv<^{DCI8R#Ma8a@l2(%K`QS{OG&%GLJ+J4{7W?8{fTyJ4^(fBE%F0FW|oKH
zP5D7T(Xms((5A}R0pY+Q#T!2qLL^^QD-WkR(cQ!oVuL(lk{3R!dU@3T8z%i@<gs{O
zc$zfEs1T+`I2yy;%vmtaTSS}~l~Vc#WV$Uz&l8T48UC|raOzf)nKlu~h&sSvpft=%
zt!u~}ZC8Hs&bF4$zu<6bis4*#qBO^x#vm8loH#9d%iLfdSQ}xSniurP9SrhA4e|5S
zfC#K3+H9RHNYB<M<NQKvidMQ>oX?9oZMDjNM{N6lve-oH&{49)B@iI#*s@D~PLM}i
zXrE8kDp}TgVykD{!l_%ar^`+D&}FEYy%n9d{?II@H~o(`P+1xLgbl)d%Qy$aAsNzj
zl~Jb<NY&M`;fv~lwOtc5908qmty91hMuv?qlyyb{2N=rwAST2dVKMkuHl!9Q7B*T}
z2*Q_5MwRnVlwUekAjlx~9=D|*1^><zz&~>GMEZFZlz`14YFJWC!~R+l(-R$zlYC2z
z6r4z$ATL2r#xTaRsMA)byq)6B0DT)|e1SY=Q@;E{0!N+>l^~nPB@zA6Qt*tmGy-{F
z#*LR}Z9+YJ9w5(oqjqGsJu05}UkuvA<3kU2sO-P8?odP>C;tonnrjsb-ogq&;=LGz
zwWxCJqgnD&5CVY^8}Yv-@@s#KDGM}Z8)VLJtONMkv58&|-bdzo$lzC;u+5NY^X0cO
z*-<Pl#p+Vk<??E3lIeAzVZkN*S{zGIpkY~|ru%wu7;WX;F+?-y#X>FA@<uGtIF@}w
z%-)-}Q!^}>`BtJ^QsY%p&6nSK?=U8*O_0NU+n)i8$k<_Ey@5-!XVek^e8(*vOz@7s
z2SJ3*Wgv`qy(ha4Sj>l63E%Sxv<+h81Ux6FydUgSoT)9AKRDrlp2DT&1Djk)jh3`4
zZjcX+Cnn*x{gKTRu~0~RXa8M~v|#eFkEAW0-bDB(Y0g&iemxfU(_joW*)UlPb8Zk;
z^I3YEwS26FKiQHao*bL!bAJIjL*XefdHjWqug`$~@@1#3C;V@^TI8!(TwH<?i(4Uo
zcJfe7oGyO}ro+m^M)y|}U9jCw$Ny%@-+X{fgw5<Bxc185Z3f7-ggfg$;ssUW%?$W;
zEXO#adcTPQ3@oSKUMK%F(K2PK{41?E6lguVaSCc}&42sJ2wuR=7Rn4v<q(RrYvYh*
zLTqM%_|`gt>9)QoC*S!{Xf$K+2i>~g$MB$<%)v8yZKbo?weorom+d;6eT=5x?K?Z1
z0tx{xO>PECg5Ov02c6B<M$L|RJVKJ~`n!X*j9Mu)XYbhAtZhs}%20F~5O?x#!&|>b
z)h4>x+r_to^4SJyAKzjOk^Z}bKXdc-6Vq^I>DYN|J90>4reab<g>;Hn6n8{IESJvy
ztO;i*O;gJNB<|8Vj8<)<CsOJ@>FUFTRa3PL8WmM=O1JbSpW*sK-3*LR+zFYZdm6ZT
zLz(2Nhqu<|R6}bhjUzkz+jEc|f|l6n0E+C=+3)q>R(NDkU1Zm2Q?rN}XkfR_YQRCc
z8N^>Aoau)_zTN{6rLlYbHHc9`dqj)$0zxMRs?BFXQNx^P%=o8a#?<GqEA~v|nhuCA
z-r91F?A2K<XGk2mcOp7$D3$W#&Tes}a*b;!#M?fd-B?;%UL-&9#?0zUP%3ii#gwoK
zSF!_`L-tLbnqoum%6^G-kY?aA8X--N>>q<DuA#mBfS^ZlK2JZ@ft}q}nwc_5x$z+5
zbGZEK<lq?I86!r>A@;*SR*ml&tqf$s3B-KH&;esW%9$-!gP#uTywx+QU?4UaP;jGS
zsn<hyQQrcdFz4{${yh_=&{{xG<Pkxx5miXtt@$I4?5t^xkNjW!pBxqA9)hHly|&>|
zy-2ImYnAl&US~jPEmKtgO6e0b0i*+Ao6zv4OW*%!fqEXj#xtc~5Vsa7({gfjkYyP5
zHKgP*&OWZO!>mlX#JCm?W0~|1;*&X9CGFVGYS}yTh$Ccx4@6fHE00d)IDf_z-K)WZ
zD`BiT-um2(M3DI35;-B5crr7)mEvY~Vww`SnsLQR{)`;OMCAau$jPZk)`~Ku#3?aC
zSh4`6%gMlaiB+sI4W?urWSr5`xZi;?IDW@2r;Uj~r6K7VB5>I-6v~2FqHKMU4D}aC
z6^7X{Dd(`(XKImvSd!sJZs@T<wq#{QVtTqWXCo4iwgYAnKGj;!)Ye1QJ1yqGYc0GK
zGBQ>TNtNJ|8x=j_^xz&_kBh?S_y+b0)TFJIG2T~zTTre6OUYo1v9ZWR&7yyrQ`=%V
zPWLY<WmkivmN~{^p^P(0F@BH~MjNo9^VY?r<GKJ&jPc$E3B(Xl28LLm=!DLSve@>n
z^1uI=oDuU?^Ak-A1eh{Y-<e4!vKl8D2x>6U3VLaWCG4z_scIoOJGMkkSwme}rJNHp
z0hx)MSq$ACGzFKO8~Z(zRR=Llx##&Z!mL{%=eN=b(nFa9f{qSN9+^nN7ppq9456uV
zVGLsw7FjL|79H0h8DDRr=4rN{kL!MsKO@+aph<?&7st>G>KUIdS$_x4cRk4ovdBfJ
zbu^1hNi+@VK+6+@l}lsX2GVVrfiuu-xU3ztR?5;q3F<D7PS1b_46EBjYmqnttHue<
zbW-#JR2|j%`PlXq{*3k<S{mVzluD;H6E71W$o$gSu^3ZTdYjnhW{9M@lJ{q-N>fZ>
z8&n6`hL_95v13OA?p+hZEY0T`vRWHI9(lk4QrFq-$CCjN0d|qDWvTb(!7$MHP3xK*
zF9JnOkiDw>Q-Y)=8XGt@7zX+reHTo2plYx*1mVUqz2GIC-Eaq^39@12X(7$Zr8a?6
zX{@>|7_1Ub6xjBFf4V&AfL%+$vjOM#73mrE9PDS<(SPR8=mo^_s)7Gk`ZJ{PAYCDn
z5t*9an@40hGQ&-itNmFMd>zytmg07NO?tLgu^otLkZWUjg!gLDkX#p}zzrHO+v{UJ
zu{MNu0~CpVp?k^CZJI3HDL2?s5H(sVH>N=v+c8i5P0o;Qq54qTC_({moh>(8uLWAu
zJS(@v0zf>{;=nSk&5*V8<W_Hd4(vG05k4E%?c4lqlVN&Fs+^o9x5u2fTn$;m{X0|c
z2r&y2k!F0pGx!RY;)dc1xywJIp^l~^PWBhAWPliC*bZ=45GN>FkvcaV;Y=u{cHZ6N
zzZm$=u$lJDy{R=MBvA?wb3$|Iecm^>6yq<L<?#OCvT9(7^A-70jo^R41o>fy%Y(KM
zC5WAqhrDG);eupL&_U?+GI==oK2uf20IgbDBp!*TQQ4rL9vcl2dDJw5zDFM`kEQVt
z?N7nGRvMEqZp}y4R=7-yM(eF33jvvK<E=zUs_h|&fEkcxGh>NN(S^WRC$rKDY=Y{o
zxOJIrvqUYKY0_l>4Ut_d&E6W!qyQey@%gvG2a$aMD4CnuWVLQWB9_R!V1*(Sw<qh)
zkC(W!up~5VFYsqPOL^V^YAy?dWkzKxXy0BG4GqbaWRPb>7W+qPzQsqpUoDBTwK;H&
zqDw=6LdiV9vmx7-8MW$)WO*7A_|3UwMRZ@*fbfE;vNHar)e~}WRs0QSE@Sal$M67G
zK#H0QS(6yhZc6lhC=gB7#=tRKqELv-T<6c|S*L_)#^JR-J!8IU8ay`223wYvMg%8p
z^xx5fO0T7eU)bb~gD72vASjL%=R2*Z*yr*~oAkKB@^}ygSE5QSM)X7o5RLzA<G00l
zctzcun3A`qC_}6TJ8}9flNKA>BzlLpn6&EMr*FTb<;f6%1(CZ~wx&&VjbgBQ!RI`h
z<thKg?w+_AV($2F;PaohJyV{^%QHd!iP!)*PM-~lN{zcto=cQbs2T33E9H5c`7nUp
zQV&#;?J7=7ShKtk&DK}zrRl}k(CU}N-X+51r64ix=DgR#Zis&o#p#^$Yu_!rF2Tg-
zL(zkTVjZf~%h6439+tiuc_o$t(X>{`=ePb9qzf<D>R5O+III>)cX$ZCmS$<FLgI{G
zPh2-$wKyi)P~j+Vq=Bq(eRh_-8DfNoYj!fAfb^rk<!?qbER)^+&L?e~UK4RZ^rhUY
z^0uwF`qt(5{<6ist3}6OHoMq6MxpX5#j(Ge1_s)wJe?u$87mEfw^rV_|1Qjc_J4?X
zaj?T+2C2JIKJaH`A4aeu)4Y5bQ^27_yAMlCeH1)0czD6!)8vm%Dbt_K9tujqsPb_r
zJtg_PeA3z@?l3fh%#}~mGmGSAJ>$>(nQA~-vBfYW{3*Rh?75!B^0_~w;hVW!nabyj
z|FUT6qLD9+QX@32k$e@)Rh!C~fero~^TOGRSnR)e2UsIQh@j`MmA_gu)Hb0s|IPl&
zWGB)X@OOWowrpB?)<BQ>NAUgNGNk-LVtE_`^0jd-jM;1Pzs2%R8nxydc_M-&7s)?k
z-f-9;PmHQn{|fFc&Sx~8_`f@AEX;ottCw%na1@-U0ZiW|=HpE_a`h0TWheUg{tW?M
z%w5f52GT_hM2gyGri|^nsMQbUF*7)Xwzpqf8fk<8kNu#FUeJdW43!<as6mi9TVzLj
zTo0@APF>8J*Muto&r9BA+jVj48;h0IO8YKm7*tCcWLA`zS|A<#W0TZe+O^WLi<^xq
ziN#SPRwsWA)<-J#5L-*7bBwzdMo#I{#f@VqAVF7gR=Rd^Lk|M|YKV;L7E@E(MOM1o
z9D4RVPkMAQv!9=s8u_Jsrn_^Y>>NWXE=Q7a6fVxmE?r#bhbmuQ<TZ46P1D)P6lD<6
z(`7ds$Jh$arX2XuA38N~`~uj)h#Rj0EW5{)h}@Aqys@@^o*((MSu}-EMS~)-XBV}p
z<&grXSoZ3o%mce;5ikrWo@olu-ZA${0HlF9KTa#sYOOSBxKAuyW-2}4Ke0dS8c-b>
z`N4l`eF1$4l<&SV80a&M=U)QL{(eCzXahxd5!&CrsjsANae#f((wHxXPr@L$2iijP
zJ?2<B$QQymg(ow&$vmjSOpMbk2OAFpSX3p4#7Ijk7|+tvM+2v}-c}dFBzLI2KAynP
z!-52wQE`<V-eqfvT8RHg*W>^HPvnS1D`?fKa-al;F9M8&R6`U5f{g&NbX4$e4!M9?
zm%3M)+$uuEpFlMm>FxV5i)8S9VqFTx4wSyZMgzfc($D@v`v`)Rql0KDVgzOF7$ZG0
zm`VTC1@26En-+m%eK~MW!1%?h5aDQ>z<PFF9+1Al{Q<MWks-%<e@zIsT_wlI994Bv
z6E(gCzEMuF_W)a>@_XVo@2Sq}OMa67LPL{hEW_xicAXr&2KjDfV&s&xkz27IynwTg
zHcm1bXajC(ESRbWZ(h^~MKid$1L?(>p~11aAxTO!DI<>9Awgp%UM&bZvLYTZbG8-$
ztB#OiM$<n1`^s>q4`c~Us6+~^lM(5eR_i2SLryi`#OXno=qfqQ-?42#qv9eN89gFx
z7HZp!3bH7U!sf^Y*dU|*7d|6fA(1>}E)ESDV<YR;ub+(VqDK2MwQ{=mMZ+|6AuuRs
zT<V`G8LdEcVek=-&sj1)eYGB19+lB`G9d^w3><lez24GT!c+)n#$UOL`eHdNjh#7G
z%vfwX?%CeSHBZcif8!kM1P;>jY&q;DNTMv~#x@>>B}aD2^NjNKRX{$T@9iS5jYc+V
zwtC#An&kpp&~46CxzIjkQYN{`7wwq7l8LE-a<ptBAX$q18Mi0X(X537wb-BW4aQFJ
zLZ0>UPlwjLMsofJ*ML4DJ}XIM<d(r2VY6{{R!FH0p(i}}WvM&H5Ks|YF6F7QW{Lm^
z#U`0(pJd81GAVur51`l<HVfcy|Ftg)m`eK)xh<v2s8BFu#CXZa!eIAPFJm%z@E{;V
z)zLi6LzG_}@Wrv-$Xy~e=>=>}yuc(YwXrMqJ6!7QfF2CUDm=+*AoI>>mP9H)tvF68
zzeXm<FjR5nRUNrIA-<Yqim`3^#WL0Y>x{9|5VM3@K;W;ZibMaG_%Egy8CNKFLN2wD
zoyxolm&G(j4;(L-`!tX`9y&7}CG+J98#=DniVMRhKl6sUaOR(N$fsQC-_vFaGN`6<
zmF-+@ZP{^hbqH0=i;*pgl7aGswE4IgO_OVEY=tQ1BG=k~!?cx?>%7zO7E*lJadCan
z9xh_FxIq>Fxpg`eNsfp%yCIDWH!To*Aryxj6M3u--D=F(>=ie~!X0v<+#FjsTaMfQ
zEk1a|VZkv;|Gnl{;sRbJxBBnU)WAV{XoFW*<El#M!A7~w`?O7OkP($~a(l?PY5<&a
za)%F?LX)A!v>4u*Ccz}bChhL>2F==NcmdKBC^s886d<T*a<@$hfmj*2CmpZI9n=8p
z?!CTkiHuk;_XRl>;7{PuY`NdqEn6lJ_`5Km!mjdQ4D*Q@fRMn&2cykHL9s$WE9Bv5
z4FZCJkhHWt5=Rb#_gq7JG`+T3FScl29!tcEkYL(BDGVA@w?N}U_RtnPE%<m)rc|cK
z9}@}-WQP5LqD`<_&GZ%08z-7$ESs&9S_x&A5$e>;<c!SrZYTqoa-FxurqHyDfxlJ5
zRkOcERhB5*nc%HC@g*3WZZjsPo9jJO5JoM$Hl9M67cC=MEzG0f+WAfZ^#-ElO^`h_
zuK?KQ?VJ0_oT>}k^oc6GV42ynC{`BpM6hSf9X=4cE3Eu4i{Vh!Tq#RTd#<f!3hEN%
zAIxdd&!y2Y6c<{<QEye2rDv<uq(|fAa+9jZ9MfM`7!?7?Izm>)Sh;jO25>`K6+AL7
zk67Z>A;)1IOD|8?4TMR{+Vl)bsC1~)WSy<)XaGmbWPP*-D^HPZh_xSxpEmlpGznN7
z)<)T6-HjP@n*7qnN$0pc9`w(%kQ@yuPb3CZ*8nbV5lyk#n<d<vA!js&w4|-buxUJl
z<;QqtOYH5jAZ(pHY5l`mBU@vvI2Yt8Z`3pw-n&=J)AmB1QN{9%&65SpYpUc~`wJ3|
ziotU+BG~kyRDcVY%k%M!)(MPgBY5Cf{*3kz&Dw#3;)Qqy*HTo1e9>CRSH#P_WOPC)
zl}YmJ;36&OYgK7DWIf`sw!iV#B7^TjKn`CHwPj#M5i@zc;;qxg2oRf1vIZ8rT6s03
zQyrrl0sDK+$Tb9}hSyCDO)Qo-VrsbBQ>-vq+MD)bK|z7M6{E+2x?X<gBX+5cXRPD8
zm_QY<#VmO{{mv4Ayr5ItPX$Hf_r9`rkkYcxZ2%jOyki3zSugJ<8mh}y)uH!-qsD3S
zONjON?a2t(TW}d>naV#HWe7t9E;=h8_*fX56DeppA}WfMA|Lu#7-UEnEffZvK!AzK
zM_ml1#6yvmN6_C6NEMJc7_mk(<d0ppT29UkK;2sL9?A&%{Wy3OY1ond>JuM^Ei8nk
z?mSC_&AjiQreU<&6SW$qd=?D>)R(K1KgG&VJRPBi4f1&`ugZSdCBCp1Hs6NjAfVw(
zTRPP3$)U{bSN0bHVj$L^jZtuE!fyCW>>Wg{$X{cL6#Wl#J+`W01ODa<L$XjagPJdY
zk3qr4Lx|-+63gP+>}t!`#`2>E0ww)TYQwY+`cMDqJyI-dOb{q!NBO_}7tju)AO$1K
z)4vmwF!hgG(&by5Mzxx5zw;M}7Nq}c14O|0#@2m~?I+Szja8FjI>3!=yRHs611nM=
z;3lWoUoYEt-Rcn$Gt_&Bs9t3STD7Np6o5%`<W#dZ5t>_r80j6l8k*@e*|Dn`wlo_!
z^;38ecIxU@pRpOpBe-u_B&eOg$gWv@5>Rhx-_>DQ74=hz$><Ot!jud>3dxEg2RqtJ
z$T|%r1@q~Y8d{|;zD3y6ox8dztpug)E2K;OS;*7@(zR<VUJWb{eVJwL@c+0ma9DRs
zb3k1YU)S!@L}f0E7JJx)_dQQ`?z(kHwTMi?jcphI601kf$7Y--yLNRWTc!xP+Ka$9
zyZKvfBO!ih^wf$srLuc!hj75zP=8j-9x;Ln0F(hb^H;FHabg(IXwR-{Yy?PYmF(5k
z4T9L274^BQOER)|%v+NtE1xT$Q4I%<Uw&+3XldO10CeGfx?23;<h=YO4R{5>DGa2z
zT7GIj_c^w|?Az6ihFDw#-%gYLtRZL}v~Ta9zJ~6Huh-zNen5;4p-?PN1qps&Jc8+_
zoNQGr2gR4*5rPt!3gKX%gk36wF{zK)?}s>hYa)LvE`Krov8T62paX(Bu%)NVp=mQ%
zl2|MlSsuK2SPTaeqy}euZKWLU4VW*9N8EBb!k<~(AMt)LQ3V}O%~e&&QSk;8I4F8N
z`lVND3HM4oV-fD&@s+0a>*H@2&cdeF*9WWD4LiMjYW-3lGxVAZ_=6m6EPB9ka!gE8
z_j(wl5V0ct(=W`rhKm=(hh4vp^<S`qsg^)J2#5E8SZy3nNHCU@<Dzq1q6xX7mFf8S
zTrF6$AYLhC@UNT@Z{wE7x0hf|PxQCx0`g-Szwo3yDI`J;xs(9KI5};CdWMIcVsk7w
zTLyZw%@Bc<QLMW%C_Q5Y372(&436zu+fXS(Y`=^v7(8AITF1ppCG)JxP=BU|AdH`e
zqG8rR4a^6XGCaMoiKhg15%QyT-D;sBIl|MC3L#Tk#T0mQs?X96M`}_G{-O$Un)5X0
zOihZ}>1S<0p$FZb^JS!WqsRt4+knZ93Q9nGg?)T<h)QGx=NYKiw=v$MS?EoVvAr-h
z)(0A2n-aUws!oq@5cUeJ2Szj>dzLxr$HiJyTJng(v=c3h2X#3jscC!R#dExm+L9ny
zg1iaQ#$XiChGqMVSdPMi!E&a3peQB?1<mbwmMu|BW3T>lwl^5lA^!86_@*4W!r?pT
z32|<4Xb~Vkw3D7^Y%*bxoFA_vj0J9LHK|+>&1&gnGs0YGUyK|f7x}N$-w1+&9Z51q
z`zerjRH=(>5+z7!R3^pNIj(sYNGe$;Xd7q;mlfwsE{HKs10JQsnl*?(gOtX&)W(g1
z%CH`l#p<|!tTkY?++{9ISh!V8jQ8x2$96EI$#j{Njw6;=BVSgeesD{~(6~)1tu^M)
zg#uOOZ39RFhh{##k4Q5d@dR&H`!m+=q82$97rTr7D~7(VmYSGs4%r22N~Jbh!LDM6
zGSlPL`7_-Wc*bC!`WSCf4d60d&B;b#z%J@7Y4Vid{p#8VI63U>m}=)!UlapsaJnLX
zfbiK%Y+PmKS)y+*jd7_3B`cRX<zlzcjY7dVTP}Bghu*A^DxNI#tr6nM1WZ@NNRht?
zt|;{MGyj76HY4(w`>EAKHchUykqjMw=D=|Ua#f7DpImMKS&rEKYy6z;-(Rjx<Y6sw
z-t-pBbxCwEM20R0Ue8)D=hw$KfGjc>7(*6*9;8=Ti`4_ZA>P9&MRg2u<8q@vV;bo-
zOePH?+!V*6ZZ^g2-Q4C=`k>d~=5~vX8fnwy)<jUt_KLzKernxj3yI)GSLJTEug)4Z
zQtpUlK{5)axxLf&*w`Th#}~+5wmZ)|P<|0K=T=vc#obO2Q?A;=PE7LMV*=$g!b<M-
zk=qSz4tRH?+?Qr!MLjpc{Ct1xCIH$fpP{^CHCJGtJ<!$hgw<uh@n|wWtowtlgafC5
zaNzBSY^?`D0fpRq*rwiEBjOSNgqA9g*q28`#1|nxu2vq4{RH)GlQh~qdJpK|Tc)Kk
zYc(4z^<tSGlC^i=-ZH}oi-2&M>F=6~%Cl&r%nBybONpiBRh|bU(84o2sEmZUb=26T
zDJ?1u@@A220)MZR=GOL=A>pW%IsVK-@Q7wuCv$B!a4He+G|%6#6OIhLT;_Wpng+*A
z*`PYZLKYbHki3C;yl|;oXpK!|&h7?TWV7vg`Q@@W*a#P8mNZ=A{mg;imB8~Q7|Bwj
zRIyr6m!+Z3!u<@x0V$WKXTZ>cBy>=(u#ZqZ5-YPZ=!e@gBaj*X%`ykH6yTJXRly}d
zn!`$lqL!twP^|W56^>+I$(op1F}!L}8)U6FZAyV5Whic}=mj#Gc_WO(fw2b1!g5(3
zXXc18vLUgq<rSsmP|+qEW6-+j7AZ>AVugt)*%YLkV)@p7={;FWo+c#yStpOjGqs_s
zGfbZFXI40v%(PTC2Qv;U7%wfhJef&lvc*OUuVS@4nTS-adK0R(R3m!R$4^FO-e<cN
z!_*>AB}t@V?z}=g9eg?%p*dQ{;+Zy|(?&y&qS@S^Z9`aENi>HJ<vax|d^gYeZkvu1
zf0;b*lQLi9y3Kb&$Ng3G1_KNNbxP{w1)I+1>tI8D(R((~JFWTDb{W9*QtaGgddaW7
zT_hR7h0*H#{l-Ve#o%?HV1sa9aAYsXH@N3(5fey*SNs`M*YJ+D943+9`m<TE^YBoZ
z4YcK}K{P1u$goa>$ZNJXxLjv*^16);wS24Pjp&D8;(PANH=`eDr@+Q&hWoed8yxC+
z`CTv#ONhZ0^R}%uFpcv2G#ug-X?$8B@0cLJ@DO>|$rWnWJfNjbV87SZ!5Q^cmQVnm
zWbB^zyN2CLYf)p2Ga#8Y|2*UiUAUAKfACeDgV2N3@<9v%7js(XH$h$hFji5+xqw`h
zk8DumhR7ddH}XJd6{e59fhI5t<^UqhkWYNM7(LI+3r{&-PI<2*a{p5sLOt%efPZ`z
zTn?6RO*8eUG_XZr6U(RexqSyeF5p1&MZ5uk0G&r`F`X}i#n3aazx0<X3FIqV5-93*
z@@IdGS1p!GBB7M2<S&WQTI2G_CJ6Px{AS8uV+4~J%=tHKvz9S*e~;0qYD%fJSpMPt
zaPDD9v=#h1&apA$h7^pGZ+uM{&$~+gX&h3MLjY$;iLsCp{|-*NxL!+ud~548ArQ(9
z@?Gp;jjxOyqE#}u6!N{b+^cV2k#1&{qWmya+Dh5ZpOG%qjF1Rgw)bbebuU&`X1V;J
zo7<9@=%-G0h|e^cg@P7k$9UyTAPjS|Q#Utb6=x8GR4eVexru0SeVw%LrpCCD4Cb$r
z4&B@s=iprijTzB8+IzqxW%Je2DVjJNrnAmI1d0|`<YjTtF7`UcybcDKuHD>RTg+U1
z$U~``3D8{rC6#>oyNR>yZqHg8S#1$H^JyXL5uF2WpOc+!J^*9YVFy;qF5MjXh0}&*
zGnm5ID68BhyZWTUPGV>&ZpYmcW$J5S43i)Fc$#r!qP`7I))lh5y;G>yXIRVj=;ns3
zYzDQwa2<i|D-nQs&uKp0Bzwj*pbxNQ6mJT9#isyO;X_vQ*xTvExTuBlV{b?=xM+xx
zTKo9yI3rJcr+dpIs^ll#+}Z_W9x@_U$WP+|Jh03UA^Uo((4uj<Rz$Gu7YnU3H5CCp
z;F`35H?_0%J@!~Rz&`}x^uzVALmg-<pDE_(5Aw0K%~{hH3wLnz%!5Wn_8d9HpP2@s
zx`XtzXDy8t_3)B2xiJ!&#hQ<$L*XzVuQDU%Z|RN+ia*@j3)Pw_!jA}!&>IC0P|BmZ
zYK&%L&Lh+8SqF+rB{OO%Xv<q8wKW|yL~^gxHwC$JDKs1Q-Md?;+YGEIMU_A1Qt0Dt
zsL9ABU|%B}jv<h(UpF_w!W_c{^R;qxJlSli`ZvijvBOGX*XeJ6V2?sqKQ<<8$#&H;
zG{9cKu@5lFafz#;LV_WVH*rvlC1a|w6P!39p>T%lcA+noB{NSn2DsqR1LUL_7F3}6
z3M8wPlVeq|F{|Yi=U(7ufTbB2VrM*%eUoHRH?{s|fclZaHfEL?%Q0^!QG`RbAvNR3
zXGwuI$fF61J~SGvEFU*|pbU#Y2W2KfM<4FrH^GF-8t$wjIU=>GwIv{Tm&mE<Jtmc*
zsA-ndte@d58Biu8eSr-uul6jLdZS_$#*Y~yqhsz6ym=YpWQ2ym>o;N=h}xLX92B}F
zV`CK%16L1zK0Rnz0;DNeyjI4=lg)R)j7`W`p%27BJIr;&I77x7kM~J1BM&b)Oqw_$
zHhNJ7YN~-EXZTC55+lrTp`6(*Ag*}@1FM7PtZrKmwbTp$C1=Zj15V1>sc8b27)#bH
z=hy;kLd0`}HY17<<9c4~7`W(zxu9XeOn-j2t-a7QX0?r>Pw`W0R4qh+BHCwQ7j3;v
zs>bW%1x{^~>pTysiZ0;whW7=z(3rNSx>7C*vH%UP@-O3~VK5YHv#(R=BePr;`7dl0
z@(Ef-*qe*v8(g87Zj2S7?I6syA?W0C4SB1TTsL(%op-#H1jit|o0JCC^d7^5VXl<f
z3#fR9Bv<9$4V5)mZ)j7)v{_(kqD^Qx6qHH9_rQ_Dv|5o^!OBlB#34}`bGE96RZ^82
zZdKQ9{CWSJFc~$MVcS#(Cyr;n0UlKs+k|=^EH&2qnYjvRq==b}g$B~RPU>uw1ILe-
z`k46O5d&p%nz)4?F2X@N#fQXYB{miDR!-8X)@B|6Wyp$#ATW^h<<QjC?s|!95S~xX
zvZ^7k+NEhWyh*Agm&-!>YeE6KhnQWQUSG>Wa=Cp2R3jl0yepx!r?+GLMZ>^S`B~7t
z0A^fNDDYRzD`2j<s}h?~Q^Lnovtf@}BUh)lSqckEERLOPVwHwqS`|>Eu8jv3juN;o
z6eZSHLZVYB4{8&>KB!(;ehCtsdE0alGmx0U<2NMTq8xe=>h)E~jkZ5G-w)gFO}08T
zuuhVj<9i&BMB}c6c6y7?galxo66P*6*Y2(Uw%z>n7r8B+Ayx!hfp2Zh0JnQ<R0hm>
zOxpt^749&`y5dN=vvo`3t{KSb))diq1qBiN21W%IZ=L+YmV_yVV9}JjgDivcQ!3=1
zwC*Wb15-$8{p8-(X)DZwSpe=c2JQ=U=KX0c!7@Ax721iyobrIZG_kl+9*h+$tjoc^
zS0@kId<K@t!`?lutE^Q*HI(+2a8W<v?ZYC@D-&RX#1SF=tUMai$ijLgkEL-Cug&u+
zg7zEZy|DmZ$h6pSm8c~$JrN7xI(nS1ml-y+(Pzs{<Aj#R!Dq=VYXn+&8E%lXZEBZa
zaHuo|dDL8sw61`{HXFT<RM_KEnPV+hm&jaeaa5*ulFUm3Zq1Ko$L##rLD1^pBVG_Z
zT9bnpIFIYTEc8z8o=T#xltuoGsUiRs10fg6Vka)bQL&pTLI`SItfW~20k|Z6GZUw=
z-JX`FwxLJ+nYk>UVUd6{g$!IO%k3G0XBy~0DJKwVg}+DDcQM){d}yVQ!9X=;kTl$@
zmAhQ!znctkv<Wl<$|WE<WVN>$H)BW{Srb~qXp|%ZD0OX{eyQQZ4J_+I%Y?`+uhxij
zyt&@i5!f*Vh{%R$8yk+WAnFGj{TcY2xB=S4WK;BqIAEqG|D~^;u9xo?J0~7bt4b`e
zTIDew=FfW~^#zb1+C|MP+3as~zcHmxD_W!_b_@!=An!_6w)p3osvlQ-7A$_!Kd1kk
za)&VrHjJ&&nZ_V$!Ap7SzYv<o84hpqt~_m|!7v!3@Jw1ZW(dI>IMAD*J421Zvd)ob
zjY*kG<%!ux%5$m1O)$P`+>Si&&zR$dAzCw`$^0rk<BlV0V~xDv&j=^uy`DP8i~h`R
zA?ERTDZa&OF>VNzmF3s|OrvN)X{K8FjqT1}ef!DFX*rmI+%AYlp;tl<6d=p}Z+)^R
zPN{JaP2DQWUJVgaPm}L!_9@s3!3eKA)v)cLf4t$tvE)V8)0?rKvw4=Wek=7wz#Vj7
z2ZTle<liO2QwQNu0F*1l-YQ?(w`0cOOgL%zz5L!r#6a~1dB<Ovjaq?M;sqwMd^bJ<
zqY+y)4-tm%g<OG~84*hid3rx2-pC>HhtO`uL6@ima!)>pT~M1rP+?X+G`Uuok&mo@
zSRWCT{YPsgn}-$d<J2#qdd)Z=7sx06jO<P_Q{Vp7*GVlgcn0lEWq7lEmfqf=JdQjb
z@~0RG-B+-`d>)&nEKgukseBR25R>H=!4AsuvtN2kBq@Oe8=xS6l{O$yoM1<$b<xcz
zfA&SL%c1PaU(z&a*|XTQzovE4!yJ=g5?44t(e!$3puhPyl)9Vn3^ptIyEWR<I59I>
z{$X^g&7pwA*G9^LQ{@}~jN*p~1{TWur;n1!O_>b<NlTp>{*~s01i5&IL_G5ESXX#3
z^EmGj&-rbzAnHJ(x9?(^CsfGyX==D?aG1EdOZQeo8*R=^54K%*y_dE$j+rFeclVn(
z4M|!Q@CV)90>-#bXKIrwJ9OX5Y-B|U*ylBRM{nAksTOjxlYeZEB~E!77J;d*+HL!b
z8qv)p+&&Evj&5|w#Ks-Et5L3{aXg&N($U|bFOJ}Ts1(x4pZR$RaKtR>+}-VZwfP3=
z;xF)k!kt17gsyF#LC0eSPU&W|LbKSUs3s88hs}gVzgYE^kjuH!BS^&@)taVbXMY#g
zXRtr#2|0+1%rx1>J0h<H(q-460GO8d4hAZzEk$;V*&+c|gZ$9H#TL_(oTudO_6AyC
zg|lXlGz}Idu+ja<pIKHc{yS6lw5h=@mqA(z*{i!7?pVvT6!~AWvbTSw#sjh=Hqwvd
z8AeDVHA5=xv&}Q6e_Sm;iTS9xmp4+Jg#@X{PrIuHtr7;seS@5|{_|%wgE0rse#V8+
zHyULB7${I?h_pQ*eFTRfMnG4P3E%@`3J?Pxt$L7=hK40bhJ&po8hU7-JEXf?ezQ3S
z&`QrV=GLecEq^G7rapPv)0fXIo^<aVX0sVHcDx)OpJNFb%!Bh2<VU^|Il>48s7yH^
zuW-G9xJTLyafd0Eqq?h+5}`_^7EjvCo1)2uh(f!ia;Nl;F%!KG<7yw9QZ-cLDx?wW
z8$TCS<e5vbpD}%o7brQ}TUBc5_}vL3#{>sKFm90k>AkjPEg95ttPNDtU@%@*23YrF
zYiP5sl;dKl7=K(V$2()-=1wOAHo_`7!R9@741zXJ>>j3RqB9^Q;F8faM_~PR&^=DF
zcPpR|%gH_{WoRrohi1!uinqcA!CewZ=z6q510{SGxK|ELzZ-r%F^myb26b2K!~qw`
zU|Y<Baie93%@nHxgbG8s3hcKE<i{OqMBRMNmTPKdn9mUi1KiV0u^(<Njyp?6_^3G^
zMV%|-(3E;=@C23-@&cFAtej?jj77}fNM8|VK$|Y3q7&?Th9R$nUOqaOx^O&D7KJj#
zJ}w7SjJ3ftFfosOXr*#`clR)igC7e(K*XFY8h4z{q#CIxYNgOS(WJ9Dn;GXbJ{A;r
zlF}-f;IAl*2;)EoF*zeTt1ZJ30j2rO7+4XiY+)*AwV}V6yY%Exv%^dPT+UAHrsxF5
zSn2vrmvdUD4V?gaZXx{7jn)fMc?3So^K9BDoph3%Z~vLk=H-GIHwy+%k_){d?p%7z
z3CX+2-sTCNNh6@G)YL-bt)etxCP*N#bGa0Um;t_`$})Ooqq{<wl&d6XWG%~8Aunf%
zH$h+(Lq;e?OA|Be{#Jx9Jbgg*mIXE8!&Ed}dAw@KVP^Fsmw|&O26dPr1P!mkU3ro{
zYiYzGR7AiSIJd%B3F$Iv{;u@Z&>JjPc@Z3-)byES6_IXI<*l0hQd6Sj<C#h}MAku>
zuTJeSI0!nZLMk<tKH#TI<zi<oCgqZvL|~>_<0>$$VFnB<we}iweWIYJ_XCF}#4OxQ
zz^~PNKap{Wp@dW?qEy!8G>&x`0dGi1Zca}(-Xd&oig&Glp-mR|;H+24)Syf60mmF8
z4QY<kaBOnFZjwt<+st2YVpHRQ*aA4=(pVd{2@pd<ZI&3F%hH#I`TGyaBbWP^CffN8
z4^zgYUlIJxO^U7iv;UaCm93%y*shyb8XF_Wv0RnNhm>FVc>@4aSNk(u4`^`FbR5^D
z&NcY84fkE^Z?g>IYV2>u++7!w<ef{cz_hT_op5~&4d*)H63~G4_plQGJai((TTl+&
z5KnM|Ohp<CSmA6DFj+##Z;B-uM!|8jw{Ipz5DAlRlK@0+@zL@ZRNsMlc%|HG<g3VC
zTnk0CPHqdyhZ-KxVT*G>Au`C|_LvnUDoeWZ_C;vVT)D%4!6g#cDqNv(aaN<1+?i;}
zgg+1yI9%?Eb_k^Aaj*UdpbF#{{<eAEXv$h7cia3xD_qs(p4Rc>uxi4Qw^8o(_E@-;
z!ji}Y6Giuh*s<I#B<KCM_NC=Hc_1i<+G{y^FwFrqAx)3O*p^^BJ><V|#;jPxNMdXM
zVcUraUxWJoNE#Z!25OU*N9~1{#?hH-c`VI)365@VC}dXEP0?tFHZZ`)$TV*T>C!+&
z+?s%NpWfZQf&leGK-DQMRA!`kQz19a+Qv0A^+Y9+vV%rhLUU%>oHMBEgDdE4?~9yH
zL<4i*G=(%RoB}i|@V?D42l}(;c*_Rev20Uz6`1R7G3<bM3R)`WrDp(}gPIE=Am&G(
zFr{z<!}uTz>?|0UqoFw~3!RBgGTEJ!!)N|-<~&&BO{kdz+RK$NAuW!jrgn&=B;eyE
zsTm8$Ci08R5+0zXF*>y{fxlp$To(J27&y#g0luLx+XGW&xqrj`!-Pr%KCVc^A*`L0
zStl!<&P;@rl~w+Rg+Hq%Agi5{<bz0}VXDv?W6?@PfUfobMCk6iScbA(gRGB5LbeWC
z7-d82xKxQ_(v^+D1-f7LJ_&wf#==c8OJD>bqzQ25K4icvJXW*im$5Sta+{OKy=~qT
z;g=+SQ=UluF%<#lyiqp$GwxeL>X}xk#ab!^!Uqo8Ej}U|>R8JFQ9r(uo{Sy~DQULa
zv_~Qv^i%N(Fi&*3)4gs?{&XS;<lZy3gXrLpXZ;f%5wsE!55(EcmFME1D5@`_!}xg{
zF|%URDFh=FvYpvTSTr1L^+KBaD$I#T;S6~(=!yfJ>7}Vw8wTQ~pa8gskUN@}e(lff
zerk93-*^k#czWUF*&r|52qz;qk-D^BiIsxc2lZ+ZmN^&ewM=7GX5rHJTW{Uwszw@l
zHO8krY;kx-ve)7lkdek5reL;TUXPK3<7w8Qb@>gW#7JOGpc1{=rY9()P*`TlTmFnm
zeqfh?F$9^<q9vS!R5*w@c{}>`U4?39>;wDl_wlv@OJU!@UFscMSl$(?<lRI7)w<k9
zXSJX}2zk%$4TC4h`@V3{t1z@bc$b8OXewc7H#XG|yhS?}Z6jM|8QT69%7<x4$Z&#%
zC4}gs^vyb4E-ht5TVu>0LqIb|y&PtDbuoTyEwwb_=q?2ag|P^q__$2pQZWg(;?jTW
z{c<T;l85lY&*JbeE(4Yk9$xuVuyA3fBnJ?{=SHl7wGD`r{vtKOG?*IsC|`!2T{{V7
z{p2g(TDWxlIZYI$it#oF@fT|@1MpIn{55*td>dxXAk@G4=!4zp)1%N@EPqc+Oz@)}
zZmhl7EdQ`3YjMcd$=9||2g6q&-*k5%0RUOK8moFN|MWpA)*4%g#X<gMD?0=xO@Pn-
z9iM5<a%k&3*1xq*c=xQ9?`)2z=8NQeZ_cz8Wd@NRYQj^6A9Zs8Aj!&hJ<Lew6H^O`
zIjpvCeY+~_MKgcUL+vpG;V|66KP1{jZ)yZ7t6P4@9%^z!F&&D&oqD*9a9EkNi~k-y
zUfQ>5z8JYZSZu9B4>htNCJ#Bw@OMZ18^$|k%deGA_6?}RoK^<v>fFOkb!QGABVGIj
zT)iOEfjYx(?%E@C2<XUg3ufW>%}Tc(Zo=Y0RZ{{Nv2^dDR!JPjS~RprbcPFt>Z-V?
z?Cj6f1nWBQE-^@WWtjVD*B)l5Nd6mWWuWdBcg)Ie)(r}z()K4mbUKDw$$(;V(IVO1
zJLA$&VIZn!kD$7v3e7o$vhDIC|HX<}X`KnMHn2OgXZmKE-u;l+na3e^*<L-I5i3}N
z2s#LIqsZQ|NatN4KMvv!0?GD?$>XR599A`wpV(SsbTn}Nw1)%eZ0%vRv+s+`kdjRH
z3xYzH0dfMtzC!koXHYSD$}z(50DIQb2)k7AB-p?}r2~BqZKCioJ4P@Z6tkn!Y!U_S
zP~AQ_2t7H2aB(@r1_xY5Rjq;@dm1zM=|4aYjX9q=QVz5KE*M=fP7b$MvQV0ju*bOn
z9V<ursONw+pz<k5>8K!8pWeNsm;J#Y%3A5|Z*v`aGsN0RpB`>42auF$wrZtsVjJE9
zRbqp~X{JUTFchnP{<}F#3BW`Vf;Kq`j`rVa6Qhz%cbKXtXp5GuV}csB$R*t%{e4jM
z?}m}!SQ|A0eVZ~eAXXNuJzb8A@iHt!jfux6x+{Q?>J!ck7R5NhC^ZCzQaLdwg_s&r
zYAH^xll&P$l+2aI=(m%-HJk_>J>7!-%PIZ~s0XgIqg#D;GBBn9io&ssE9F`l6bm_Q
zybSg)x3OM;GZxDbTh<BVhe|;YHRw&Mk)f$MXsci@6jYI6F)WaiyO^hoX)?p(ZK8|u
zHq%Coux4S~l2eU(NS{VwbDGm~Ho#BIWu$!yV>+;a@USs){HPv=dFS<&b75k4Rf@5a
zO$v%*mXy)KW0fc<JjNO=E!V`^V{PJjR7Q}~<26Jn%eWA9wM@>b_t`>wcf{yng)-h}
zs7!(`*Fn_S-F8Bdk5EQ1A-!D8=rBXhuzA+xb8=>U4yBR%mq2yXHqLSasZku#q8ak$
z?2tFO$AL0tEIlWhuz+HEZK#by+Mesb;C4%9r`rn+5xjDqH;#h<zv#_Z&JQ`9fiVk2
zzaajGTaw7<3yoe{bf;D>vcHP*I1p+jv+WqvV?$reGAa6xUlywIi}eJH)9705{5JJ$
z{DK?9a_Fu|nB?Lv=bMxlID%O=V6iEz_%EX8q$I}35DqB~af-q<nhd<m7OsTTDNiI=
z3I(04xke_&S|g$|Gbx*&QYDj|)khXcg%68zhyjJIi8d?KfS0Q3+M1BG%6s8;ocnh=
zWLG|hHI@mXaR;mRmvkMlZJ9*i;uszwT-YD5oJftW$mTnc4D=2Ng4!4^ip{WqJM7AJ
zX(Z|w(cPf?K<ZP2EMq{)xI`ugxdy_LAyeACv0N##UZ#3`CQsY=N~IxsqIm%A(bsFs
zcS&rY;ka;KYGSj##xlKMX5yl7v|R4rw=ptt<KS9+g()0ol;@@(y&L?ka3~>S)YcB!
zHdo4(K|v<^XO5ezVj`nh3G3=4%C&wyj~6T#tuNH$Hpw;iHpB#DMXrramCdQ^RqeRf
z*(+n=_JKZoeNaZF^eh-t5h@y=F**h^T^08m(xjLItu5l~M&qNx@*26xTQeIbO<VKj
z<~W^rCS$pon8_3%3zfMgmfEsHP;<RC2Bjux4r;O7=FfO=(f-Z49=Er7j|w&%^>W8H
z%S!tf(W@{_%bh{V!3^oR%a#&e6NtrMgg_|J%oK9BtqA;4m2yv<j<{>&Cds{R`lLXk
z)Vt50S#LO0uoCyjGu-)TNmz(y4+LX*ybbd!5XG`g9*kBY5tyxZ6>V(6UzV8mVW%T+
zL7~l{dzuijb~EIW;G9g=R7F=k+GA@sO;Q4dxf<6k+`?c=lE=I&(<%+y(HL{g<|^tk
zGR?<HabTDdS|;E<l<EGlUDYfUY=$3De4Abot7VyTX+kaoA|x~YONx6`cTxR&R+4%D
zfm%T;bWO~TZ;(?PfpQ4+S|v>`DQ<^2;`OW9)^Q-LKBY3p)@>Ako&OJS?;Txdb+n5Q
z2^~UlZ$Rjn>TXPpdoNhF=|$yOI<}-EbR^p{z4s0Q0wHqmy~_=-KtKWsBqSsuKqvtM
zgc=Bha-V0)mb~uo_s4hdS~uVNSk`;qckkJ=XV0EKve-XX`!+sC3Y9F0eMSXo8ljfj
z$JydsNk*2%k8n59wYuC#2HwhLg^k=;fB50D(ta2+sy}YYRXz<JCk%X=2WGoEJ<|*q
zK!1&_vC*|VBHbx#{T*YurXH<}Nrv{9^)^Z2%p5eO4J|eV$YZYy=SUkJ`NqYlnJ4p(
zUVu<{z;ZVwTv)4P80@9Nj&1*O<7Bh_vq#?^vc=y-J8qi3pldFXt<AGz8zRS=H7vFT
z%0OO}XSXL(purg;1MZ*TuQUm1-#K|aD0MglZk7dlA|}N-Wi%N29LSSFy<|ct!Am~|
zy!bo(RhpuZz=gnpRnlnS_w3V8p0a<Qg08bq8!~~u5^0JV3(++f)s^KL!){~!0GPL)
zP1CP}_F?(K&&3PF@WQX{g~s~QY<0QCGs|z{nQ4h|3FPJZW<-I;;ds*Pmz4a1PeETN
z(aI>S-x`3ZX_ArO1rvlbFb`c3jQ_=8SEFjDX5=Lw+A5{tzchMLERx^Hdkof*KLp&R
zg)`;l_z4MQrSeDn3!)Y6e18h~v2`#G*D{v*m4Ff<6k)IabI|0Naq<@%frW_i_FrSF
z5vU9KYAgV@W<<SGTl`xf9jykmO8;v<JY+%WW2Sq&?$3-ZK;QvYy<t<bT8RAJZq<>M
z@}}`9ZN|3k7<nt!#^z)4b~7wx@{W%Q$p`KyGdcdlUdQg@|K#0-&6G9j4DLOnXxw^P
zHd>MQV-dBe$43JEXBu2pZ$=Br2eB|J)g*fue?E+(fi__%$}1oFJ2c?a+D@O?zigp9
zuEYBuC)!~n%Mk6U^nS9_FXR_c`I*!#pW63BYGLR4%ooTs;05w|Km--abhribh2aR7
z8g7Aq`=F|_U^vEWs<~Uf3_x+hG}9QvePt6(M;$u(j{|)z^p&rDQkbm~mG_MuK2+kC
z|N0<g0Zc<!gA@I29MeAXo&U_E4#yHJpvw0NWOV@25+Fa=p2LoZl*UStR%*~GWMdrJ
zrIned%&kYeh^FA;s~@#eV-^k|t`YF7c^Li2t=y<H5Dq+SEC$#$9wAgIQ-j+b{We>v
zoyVe7FO=PVAnq2K&0^Tg_h{vY49%{D^>R;#Bf@J``&PYwuT~ylHjGu(_ikmzAFY+4
zW^1@1TS(gb*y0eNh-9gKTQ%*eclU(p_#1&{HY+Z1Mkj)0?&qM<WXw2v_UPF^fs6Ee
zkk~r-mJje*l_;?U6W-YW)GqG>(>QbCeo%QrTD4NUToJ6c(z=yfBbaxGL^>F;Y*oPP
z7U*ql)0`+mfHOoFS37^EVYsfawRbq&1_h0r$$&{<kPiM0M`cSNT=>#4=EjALR1WM<
zr&elYoD2kXwyi?6HZJRG>0-YW;Fp5}?WjmvAqV^R*ts#!gfY%Tnt#EyrfdQegCWRI
z9T@ARV;QQ4+V=1csFA}G?U}<DRyLd-mYY31U>*qfP7wme<Ol<l{Z+GI<VOa=Xv_w}
z5SlGV`7=t1)QGl2wmmuqf<Ckz|G!O+vA1BO1uob}0SM`8%kB-s(y=j!dIh1iKt(++
z=FX$VXN?^1VAHOx=*QUvr@=u!A?OVkE#k`L#8ypS`mq-J6AJNJVIw&yjbOe&dMyl_
z-JC3FVP7iUV`7jEu7Vz|%!YwY&?)JDf}#tFfa<R1U*MALX><pbqeOc7YGT6}DA3z!
z$Y`E^FD3&vC!F?)jqlzM9fSJXXpQwH)8*tAxY3FbsL#>QpE2fyh7FdNI3@NKas{2g
zq*64|!`V7Dw!dJ)csVVGgS5vBhFPko`!iGeba8TqfiRR63(riL;~il(6~o-m5@VR7
z4eT1!#<Uz4yqORC`^cnV#=GN^lmUSX^}|w3(+bxx4^rK4kbwz8Mwn7fhG{F%Q&kJf
z8EkKWENg3W6^Q4COfV#UWu^6&<Bl1cP(s#WFqm^@SnyAkAmCp{+s#-cFh|L6D;=-O
z6GjC7#)D;!bm&`Xt{{ZvdRa5RVN?(T?#<D$je~1sO#FM;I2juhTZQr}fPI|(maRb}
zhqGF#<!;h=8SgWZ@C*-38qzRA;B18y#3(pV>;XI>QSmBZY+_mt4ag)B27SYsBa;IE
zT56D3Vg4d|1meH6)|~B_q=^gJ@?g&^<s5&;vdn~K8#Kam<CITEmsc%6ex9A2D)wCm
zM$h+;0`ZFh^aXJRjV-UMU^3r@4mI;fHpoTB6Kcxxa<P9+eVg2QITYke;u#|I7~Zo2
zXQ41gWLPQl!i$_%n7MkZ6bF_dXi#@938;(fYH|oK%Sgr{fVzN#I@za~2kC>kEA*n3
zIxe+>BKCk?k!ch}uBO2|1AH|b6H#P@Yivrek?FXLYGi7RtA6?UPhATZSMJXY(;CYl
z6+V}hD54cD9G#VEM`@W;-!<T4<aI7kFc8HMkkc<!23B$TRJqiM7o9%AnX1#)v$zfu
zT`48O9DyyR%Re9ZRadedm6`w&p+V|+2hM9_ZOEoTJn}Sug;CYZcC4U1S*H6g)FT&q
zvdl1ma+Nu$vzL+S4c&;m8I&?J7BHbKgKB6Qxy;u}(m|{X{>w@wR2eQ$zqmOM_%Z$d
zidHF+nWi5L@4Pa=C<2knRSqc8%7he1?s9b+ka0ewDXdZ@xNG9vS0mcz+E_9UEXtx>
z7g&MmY6Dpdf|;&Q2No=g&mi5H$;UvyAz;~Y1DxnL2H7FoZ>`+q<5}`Cb>UTVvqMI!
z3B+c#a@k0?1p6v1wTS7A-0Ck|yfk<_?K3QZF1Iyrx?We1zJXB-H|D0_p1#}+x{6DS
zxW2=7ZPeJIa;L**)-i+Pu7pk0*~5_AU_Q7zaMm3Gh4;jw;74G@90j9$19V2=(*B#5
z`y2pX=Q!0E2aO_hv+uVv1rdVj@0;X-IAiF~#8z!TH^zxH;3|30CtCmmH!rD;IA9MY
z03e=VS9#ezY~Nhb^$7H}w0~z%&s82tt653f(1Z1806PF>0J4bJM_8EQq)*?I<uU)7
z{SbfUyNMiDn7`_MESiF7)y0ZrmSG0W8a!uqS~7dcX>B2p3JVX+iDiKM!q*Fazh370
z+Z2Qu^vJ|$dwZV$Vo4kzUfA4ex`h2HmQGS~L3Tj`OK)Qp^cVUwW>Zi%W4Xvh{tWhQ
zd~3Oliw%6%9LviRpH_DXAJI2QP}j-Q*rEQ`EpS<|+e~3KJ^LtqmOx&kK|||cHZh~Z
z3dbjc!DVH7i|K-fAft&@L1Ci`Yh`tud9`FrgKZkzWKEp2(QNRx)<*5otw#@8mlj5i
z!CDNjk7tm{Xn#U>k!*+&V5ZwH8`GLlcON<VoBRtqs}z^6lLkLznxKmH@R|<!klkrF
z2dw~mTbx#Koxtg{)xV@82thZX70pdj-oXr|Z3#Zh_T@L*{TZsafvz>hM;^ClkQuG!
z02TfxY}P^)w3jD?ma~=A)7VUFha;KnOK)`ujvB@+jsA<KjsyKnQqv4vc`E(VoO_~7
zPM!`XKDOGb%r&Jw+z7Iy)yu*p(d5$MH@%a5c)dKEp4pYh0Mh5uv(*|RwG8gFUk4+e
z3^z@s{KjA58rTkksWZ>}GZsi>3dAf7|ALWCL2<4Ib<g|-2Ao=3w%PJK!&UcQ-MYz(
zF#xy_vrQSh^HQrOEjDg2TWBn%17~V!xilL6y?;;7KMoZv4Cs34&&+nsqqR_8PS4g6
zL*N#(<&Tav@YEEW1PcN5vp)rbVF|3Q03p2+sDQr@xh8)OszulCO8HCLD4Me&a3Cjy
zLd#0|Ypjl<ovv<?yc%QYbLFf!BY!hFbmT~REf%919)VmVuLsd8xtg*;j<gK&;w&)g
zH`4Fidi7!OkmFGBT3jZ74^RuBAIqEZ*XY9P8S<6`&g^dLc$UIQ-uSkUYGEKP{5YUZ
z-dFF$K?RAkuGK$mOLyE0x%}N&pf(DGnE0MC@<D?O<bA^<#58KA^3PZ>VlUt_r>r0!
z1ga+?W~o{}OdE&zOGV*2`N${?DhZ5w|MC|!8P6B}ae#+jzzkCQ#FlNoBl$FD0QTW6
z`z*~sKQ*FoLHeH?$VYeWDqlE~h)#=#v$|4C<lnY&XN{LH?e(rFoFQMe@@S#T92+vY
zM*N?cl5wFW^0iNC3ZAMA@=c&z?Vj914M0{kj{$%GjdjxL0Wew4@!NQY;6r*W0ik?n
zYh@-86w2>oGmwmnblW=l!FH;#enhQEYc=7F$BiMov^H}X)n-`NA+VzAzx_TWmqX*q
z)@r80Cas6Df7f^$WR64#2om#UH-DzVfS?naP<L+~0X=01#lgz=XzeDeVl~#wp8gSG
zr^9D9B%8fraFfGODrSZ3-CAury?b?&ef(#wOmA1B?Atn$yC`n6N*;TRr}$v`i9wG#
zWfVa63rH=dS<T7&r)P}mwdy1X1R&TF+9#=p9B6=4qvTR6|Cvrz$U%BsTieeQ#*UFT
z_A^>DqNPl(G{=Cnwt+=ptU=nP@sa+C@fo7g-UjG}g?6x)3oD>&)JVtHX33;;fa+3n
zRgg~hUSs_*l5^)YsD>zO>`s^TjQO|}Ugpa|HqkMM9WDpA_7FI@0OgRFM`MR*hlEZD
zhO|z8+FBLn^G}gO?OXOKJFK-aYYJdWmcX8xNq~QrT(szDjeL>A)8<2g1RFh~wHi_p
zMO>Cajm#tCXQh=oIcjI9sm{sKt<}=V5F`BmI<V$r{4EvJY&z+h*1%jEO1lirrPO?E
zdV3SCn_64?IDZD;A5I(o3t~P#aN55TnT;p-cp5Qg#DPUp`8v@FFXS`Z6|Hm*DQB6S
z<X>u51HPfzUfHR1v%`ge<r?W8D<5USI6Vx%J$v_)p7GNlt6kO0fJIF{#(vhI{8pv(
zj{O^hw8%d30qoNYq3`r<K7yzhI6W^XJILhsTE<<Glj^iczqBMI(epafykWD@Go$zv
z16<?HPj%Gs+GIUD8aGc%&o)5KTZDM$WpcWqvjmy*P!7-VH^M~6tMW{LMne#`4CeT)
z^0Qb9TAQ-mTL1WtCk13o>3cxh8W@H35VBb>dCx$@3iETTWl+3^U7@wf;^J7?VQ_1=
z`NM@s{~?&v5F;|2T(}UJ)+j>*I@E6f`^TDA%djAk$+S8Qj|sq-nfnC8Bc`D@e1w1N
zfy<gEIx^m-`itYWM)g5t|HvpiWEY(%qm2kqRa3^aZt6orKMgJa<NuSfHW|u=TGybg
zcI9MJ@;KYd<M2N@D;ABmI+n{f-T>&<^JFOqc+~|&W{tcn*v62wnf63`GJ;X8mx-<Y
z4n!L)gv%n|l)mgq!K3jV1ti*57{NrNIKMjERzBuzIVV7wfI@!Y4b^gP+N46%Rjgw?
z_<3pKIlA#o893k8lcAO9g4SV8X2}`Yl3|=PAzLe#fGa>DlnZSW5mkdINx8`R+Kh}`
zY}7N37B;!W=TZd?WR5acQW(suKx?TM`75A2nyqnT%#-3+2mCSUSpZ?B#HV9m?H0)x
zxH;DUnA}Y3T%nYvQ);Gfp1&4>h*Fl8uNdT2DcPXNk;BIfk}3WMt!Qj_j;tw}n!Z|v
z6U1jLq&)4dcFF^rr6@svbcMgo{bn8%Gma|#8Lfyclg7M`oa3rGmzykA#&2jBLZJ*G
z?^53gs~^j;kOjefL9B=77Dj$FNOg<3nD5T{^L~OW*>1eX{y>!@nEYyE9k}Tk3&0Ak
z(|jGlidijP>-NcX`xNIHOk_s-l-^7`qIK~MEJf`}n0#h@G!BmI%Y1C=U-Y{g=j7?W
zJZ5EFA%0OAQrov(VLM7DjRy2}a%JrB03?T@;@DL|K;z1oQg(H0-Vip9xF%i(`LP$l
zwP}-*ISI(8?drOKvsdqKa(#M@HABc>X)YjNxglN~i3qG4eZ5u~$)aIziWw0!P1*cr
zABqJfkwAmM6;}SD@PA9(C>gYSL`Z|&8tB8`vKG`v;N`aX927#7BHIzzp>lP5EVL>M
zBgGvtZGnaR-|1tSK*$xWlDpFQC~#nxsNC)6!y;-Zf}v+DXp%_;8*z8s6WA?8g~yEC
z8%GgNT$GxFWPG0wO2-|q!3`+ybiY3%L2;YmEtUt;LRhni1i)ZcY_$B`K!?3-BdUHc
zk_Xe9sn8k5WhcLf0%Kt2uxggb!|{NLqLXXO<re{8z7}R&a7=k5*d<s6!z`jZ8$&^n
zmo1Ol0t=u7(Lnc0pPU+Eq?J>mTrZFLFMclYJG~opcmK+ltIC9)xdl>hbW#cHaGA^s
z9BBj@qaRnv>|pK_GR0+;x$+z`Lh~}mp~beTdxK1=Wpw#mXM8jiY?paK0tI8n<z>DD
zsN?}KsAu6y0E1kRKt~ZRP#-GM==QIag+3k(`bg|v<Z!AB1-rw*nZ>P}EU?Yo5jcZ2
z1ayh556zE36-)g)3aPYGp?{?;3xXVznNEA-^44zbM)6P&bIi+%v|X7QNIb#mUTNdj
zpvz^ktg@?Y*l<}LD29f{i*6gJY)#CnMnp*DaG<RX9B2s>8jUGLFtB5tU48J)q5?s_
zrVcD@y)T}jD-E(CkXcP5+{QE^<tR0B$)+TuCs#UtVZ|0&Fsam1cDF#w2D_8WAgRgb
z0MsH8wG7>scm_Wf3qesQ+G=ZOv4<MjmR{h!tT7|9Jyr+C$%5J&VE$s$_096Q|H565
zdJPL2SHaHrMEV9+Po)RS@H9F-8OSdN^;L3Hci33~qttrEjs9tpFKvxD%uhL)(n1cm
zGL&pei<H5z4Y)&|cEP)QA8GP+*s&q6T_ew=wV|D?L-VYmzuF}C=X~loY{>lwdRYSE
z*GYuYmd(uc`HjC!aTI(LPHt+MQ2gb2pUAq5+70<aJcDexhERjc{Wbx#2DhD-Qj^~W
z<3Nm$GBJ%fZ>|XRVh}i+lp-<Zr5HmQ4L1TUHv?E!|Kb_1lRw0;;FoS7(3P<S?Hq8w
zm;Lva@ysoqBY*V8+D_B^LlKfH@1FtyTvWs=#er9h49jW5_;d4qLp-lQ9ge^FShV{x
z#*rmx?QZ(3oxBm_<keuY2$q|IFlqUlPXjDL<DscxrM%`R34t+SORw8sgR~N~yph1N
z&1Qzi-~AZ_CkQID+~16w25EL>S$QjcW+K*F2xf2FD~O@1W6`j8{FNYj^gx$?_%mA!
zSRszUyRn2)R@sE{=Dpx)yv1@%|4<{(`~Iz6(quB|J&?AmSzsU`LQ7LW^9RA!Qf54^
zr4JpiE}Y}EQhcC&_@iJl18_0OzXC9nhO5rO|NgOYfU^q`Atj$UI7|XpYXhj`(;#&S
zs%%m)OT9$QrTHvIE@6+X&;47*ZBSlk0QVPx;2k$NZqCZT15=3j-hi8I0b<y{j6rsw
z{*`=X&=5i8@}C5aZkh7buhS7g)HH$~zp)P*>&NHH<-ajL$Qt@Da&s5Tw+SpBXcF%R
z`7V~k&;>kONr?-*WV3u9S0vKdP&yxOD27D+kdV6qT!*Pn45gw#C2idPatwOa@6smX
z(iTD9uor&h&loF#@L~AfR?Cmus5NHdNZHjshnAI}EW5QaGm+6&3s;Bio_=A@3lI#h
zXnOV@ZTyNG4x2(1j7U&}WltL)mD%?451pq`)u34brhvU2E?ff;Iw&vAk$u{zp^!D*
zkY2v8zd@OW^-36G@)LviME+0qYoqFS26kiw?`aaj{%zDOgwA+M-3PQW+n&mGICG?D
z{u}ZhavETf^WU*z^{u1~BuqzxXsh_1Z{wL6XH@T2J-&4tH%X4DEkR*BX%l}}A=(2%
zLfbYT+=klWW>jrw05;Ym5CY0U``Etj{d!1;gpyT=`Z1JrY~xnY>I}t$8tD``g5Eh_
zIy-<=9a&q47c*51NKnA_Jn7QL)#>42e`ub9XekKY2PK-AM^`1{%2vz4X>yalHo@X`
zh(C*nY#>g4>O0E0px9xSD5G+yzl{PyH2#3Uu9Cz2ZNd~~BUM(D!-H^OSwem9V$cn&
zdI1=u3N8fU(sE=xgKJ6)2ty+|Dgn_fE*m`_9dBc=;5GpetK}F&GDLTnQM<+l<PmII
zBggvtOkHG#7mIcuXDCMRFP1_)-u{BRlAl^AC#3NqK~V+*m7Ew$Db7}-z&<t_eBq=Z
zg-QKKO@#ZRn}3M#S&$NKdW)ockOwxmwv?6MdbClqdSg8rtM>H0#kI~ur)97`CJl($
zAiaWiigG2=JBF&s)xy&b0Q$61vv*nHbm<#ZfD(E*=pZ9H7EiY0gX&w^awyyUre7Nu
zd?CXj7?UBH4AwctphEs>KRMO@jh6J~a+*O<Tf#)2(_=V#c@et`J@5=)GfzB^98BSI
zrh(Pt<bLuq-$kB2>R#l9#<cqThbDztt&;)v;T2s^mw~Z|$i~N|G050W`7{~qBO1Hm
zxXhSnK)m1(2UR@@Lw$I}NicW{BGf_|w$rDcexlZz;r1!C)xwO7NRTZzH3hf$$T*S^
zcTf?{5Hy6gztPPrgla>J2mMgXK*3||8z6-Ntz+$*VVUwO8Rs)3mWhLSR?O;A1x&ab
zZ}>RogbSn~Xl*PlUI;P^${2wfCKLTzt($@~#ca4q_HAjVLeBP|EjYpIx}4L@7Q<Kv
z<`<cHOXb|yY9xJSN^_)~^L9cXx#w)$6YQ6IeqvFxc>m$%#l@}{feE)4G-CqExEIFI
z16fKHrIC<K83=*b-vR`;$i@DQ6Y=(#h1kkV(tE4H)xdGEHHCqh2^!%mMX_3t2KW{(
zgymA~&lpBVpjY6YDRH1JP=~W1p8yQVtNTkPFg$AT1eu&(W^&8|O`d5PyHn~DkfW1l
zuq1a`kZ^GwOW9_P02=Gh&dL;j-CQB&E12rfOaP_owM5Dt<>u=)81-tFiXg~h^ofwl
z_zRWJGm=Zd(6fPB?yyHz`7`y^r~z`TT<XuPeTDf~tBq_XP8cS6U$cqqR-D6vzxCfz
z69_^!snn+N5$XgCv+~%qosI^B8d2H@H^*$59`KGVL$LLXfCE7<jBI0CLtPuQ&)N~O
z3ti1U+;yC24f~Xt!4hF%nS=kIE0+a(gI=Gj0P$K2lgnc-%E9vUa)qrDW(>J<C#uHT
zp^b`UAXmjVs00K~c(v`{WE8WMYhtx(cQK30wf@YyJX$NR>wIv#8~w;=jQ90vggG#H
znnCl1KxheK`EHCq3Lq!l6f|B~Qo(ALHwVVlT%(eK+>%CQ0VumFZ*B1&>vU6zy)CXX
z<}u~-l&SJ^d)n(25Wp8A6plI&Zc3E|H^?0U%@APXP9IejN@aOl<gP&0bkt9&mb-)4
zOKRkv00OETy&fyU4e#~sZE@Mzy@25qd{!5d`+`$|1@Y2FnT7ie6P7>)R{Veu4%~wM
zQi5D4Kex$u%mR0KFaRG^I8`1pva3c>Sz6gb?m{^(EeiCDok#&k7(PJNMCFm@S0`h6
zd3n@^rI!oFH5(QP+YYw>nEhq-vgB7mp(Ehots%A7`z&^*Vad|cZcG`N<-ZUqjCV)@
z!!kRuAJ%>(Q)RpeBeKXRmdc!jV#|+Zq4&A5OF%eHtSD9{^AfgLHdf_3nV-Ie2Nb6d
zsj;#kJ)_az{GbaH4OXHvSh*}p>rnFq*gelK4YG^<>xD>Sg^&xXTP;h{h#P2NgnEsV
zW$~U$j&?^c3p#>np*kze<4?HwAb_k$7@~WKW@BoQbUUsLjskr?BdgjpS?WP`rVOEC
zOm|Ty(`uiQ0mQ`JQn=aI1gTCKEo&1*Rw^DLw6ZQFE7)Yz$_8&0I(Npt7L4^Syc--g
z<5=`6GYo1wEWsNCa|O6YK`)z(cqV6OfC?L8O8QCw(k7ga&GEr>bbf%yxWx$!-clF?
zxB3^FEW$$#EoWPd0vkFtP=?BEk7sJfLNX?F-^T-=$lx!pkSF4F6reN3lZgkQdLf8;
z4ed=zo(!;CAUo_^loUY#cU~HIS}*Af(W3b83VF&fT8q0~p0<A@o3BEeY#9jcpyTHm
z{~jk76InSi{<}e*HSXNAPhWX102~Kh{nvpO?J@-enT>~nBeZ@z@5`mY&b-FeFvz^%
zh+BxeXPx{uz8_*O6@C|p#BGHrLA<kGUQE<bD=!86LG~`#E{#pUPY00=fRxFT=^)}w
z{b=%s^gHE*U`njzron0QvJKDrwDQLwp**M)w|HLuv=e!2c^xc-<yBsBP<e174*k!z
zxW@WPgNF{0zr@ZC%puuE{u+PQ)Kd6()!v0c2=apbEf7%xZ4p-C*AgVfkZ)0>N?y15
z7{r)IEvGule{UG<$dS&;-)&#|m$G`!n`zN3Q)#hy7TBR@LEefJR9k5=R;tIpZM)o~
zZ!dYrS7*`%FUv*p58t}^Fb6M`cl`~+C~>`B-b>hGKLd4l*&QP9$9NDLwcHCFBLB4U
zw1sX)J_v@+idjtB_%MC40frAEeU*IV&v*wg_<b&9nt%CP`36cD($<1)PCjJ$*nd~k
zH-4c>&BF5ZK*=YuX`|6JNj{BTq9j}`p9NG!2##d8xzBwxEE{(^V~AM1q`5Zv7mlO_
z@JONjzhA0$bD4hA(%x#j@nujolcI4FVa)l;7wT6ES-JctKpW4N<nncTWhq_qT7vHz
z-vHxyb_}fD-hcfU!p)+eny5hovV0qxrhyRlyLnl^3(UduPG?;M4j8_KUaVlr_YQ&j
zbg)AkA=m$4G?atUMB1uaU08vf<z3pEkxVZ~rHnQ5qqb^zQWmvTe%#h=F35AxOzB<Y
z0j@R{36$M@3<mtH(;nsvW%ssbJ!8d&=qarKmOX5ADD~MXm9l4B)ysP#Hhr(QW}YG?
zfMGy!U{+<bDSz*_Zrm$^aI=rUOcn+q8qAP3B1R0@*6r&sX^JHVL$yKqi9ciT0(+)0
z_+USqzuU>(WPd|TWBoBF$^kxn3&q9xE(#4oc0MprgrPP_D_^Xg<d)~Tb-+9X#@;qQ
zD5XVYmceGmTg+^pbWpW5So-wtE$w1?CM)og!u-wBeEYT@YlD1DI2C9W?O<O|MIqUa
zu|ue80t0tmI{C_th|DF}Ibnm@Y(`jJ+PVb~T}4>b;UHg_3FSPT%(rHN=7ZDz(uu)P
zI`e+5mO~tL`-U?T*JxO;q6|6-1hO<#RfTWAg%5RF=+{pUbKtN^w4SM%`0xN2#w<!-
z5Rv5we@6Qrkg596k!kPPrwT$eEWPTc1wliWk)xdaVdIgb15*exg<xAJ#{>q(A@dhz
z^sZ^r6<jAIhz2><-m%hn_~y8_Zr7~HA%Kaw_{ZB6-MV!<SxyKL8?8{-d`K-P1|>ka
zVn!nqA62ozzuC|Dq#&Cr_&~cQ5Gb*#FIc+AAsT@EQR(4t0NGlZi?QE54JgQXQTL`-
zfQdX=b{>J5%(ud<8So;)vW$w82$t>>^hMd+`pEW;DJPf1h<0)yPKUz9*spC9JBc*b
zBOn^Suu}|2J05}+>tWnVr?z!lS_vo%1!I}(1S$2jSR3R^SZQ_tPPfG)>K7DvhGU9!
zgiD0+O#o3_P^)|E%tTGJL({DZZ<5M%KQn3?IChluZ`(9QVf&xQeEpw&Ap_F77O1X;
zLm&fpIx@y3Eb?TKzfF;a{x=xrWpI!{5qlRnyCJt&Aw&K3`M748{=s^>!xBT<2=Q&W
zt-W??Rz}3GS7l(cV}!)WgfeFI+U-pC-yox6U@%qsh~ejv(eXW8!Wq=^$jF#DUmAW!
zIhF^58kQ07SRaKx7UuKP-8U}2fslMH=45$p+{Dn$xTP@Z?Tj0YMjt<tX=S_}r5vbG
z3gXycP8M}Q6HPasDPhP2pP5IA(2%K4bV9rbZ0K%ij*|?pFr`7vK0C;KU~M&8an#^2
zpR*Gr^VMOG#72l&=LX&oQ9+BoIj|UVd7i;IE(fRA`N2RS4nh}~3ldHj(_sH3>}GPI
zuR}9Lu}<htcCPVnCP!Z6-yk*yQX-fR^wx{x_!VW*G31hXL!%s6?dn=7OrtjEANk4F
z*qNeOY&9e(WXD5JE%r~0OOt(A)DI0H^rw~hG`ulL5UZfuWC9SR7*@&Tv^++!f#<`J
zP#P0?vqxaHEFLjz2hsdsK3PXvf7s8av{e_}G*n5mr${%PC*_XtMGyt7aRAL_i(xje
z@V9;eBSNKZs4_aq`GOo^wfVFR_Ny}Bp;R#2_tLbZjAh00aaJ{uQmXxzxWt-+uJXQM
zCcEMihitY$YJwS7Lyx1TS{sK3%upL_$TT0&1V9`*_&6;#e!4MEC^hOSnUQ9OS%=F_
zC!x+@&sr<#nT7&58EfRSod&nk)=-Jy5xYD#P)+5Gx?Ct%__MI;Z^fowX$)iB;GuGr
zO;e2?$B-(n4x(z2>O~h!bOgR8*!~IppIn>Bfnjr|SzhNS4jNUZT<<TC)~NIY6LN#k
zp@vE<nF$*VDxIavUn@5nHcJ^qY>v&Ff(8dvXQs=|hKgg4?RuizVhgE6aP+MKOLN?v
z+}73|v~bB(mQ>_4CieEW?iOqr7A$wz9D2}~$(=r!$+JK)kam}&orjj$)l332rgL{p
zje|#n(iUvRJ&82gi_9pXrMh#k5s?=D&<x}IY+1O0?vVR^Kw7#~&O!L!12H{9C>S4v
z5GFsiKliC@5^_UztE-iTn)`#XsSszB11pzd4k_t<t32c*)7owa&wBH$l!tA*S>6s-
zM)^e$J^~#nhsq-X6L{JtES$D8V<nHK6|9EHV>kUTZL<cC87PmXpVu(yOBJAB`6*w4
zh?08$9ZRC66<5c6ndQ%*_o}s&hO*gdoF(98JU+8!jse|OYZ+~qx$)UknqQ_e$Glhw
zeUI!fK0l4aL}BvBWwOBFfmGL578+4AXs=Qh`Km29fh>yokPN(CY@Z<3tXh`D&A^)m
zQ!H&U#R9y)4V+|I?D?p1vON9XK*t@@g%-&Qn-+;42xeayrvR?Vp0dhcC#)<6!P0`O
z<Mk2fp0*}I*{pt$XjjVGL=P#i9=DDK2B74jtc!*B=qBs^KD`s$bSFfO4FM(Y!)P(G
zF{nU8Id})Lb}o3VYziDqE=22gv<8hgRp;{_-+J|Ij$iN{Bi))?UTld)v5)lD;6H`M
z#b8(492zZn2Qs0mvfZC)12v}l;|6{__8U@-CUid$7ZhwbENX$v@kyJfu!floI}C7-
z!p}%!EL-ER;hTqH^{IG9!3hf{YM*x0lHW036=hGj7#uY_7EOj)%$Jd8VuN5qM2aj9
z4-O>a>#mk(<31TY4b{pRwD4R!$-&AF$qZG~3==GJN$7uLD?}75N-lE+&j)nWh0Poz
zFZeSC7;RIYyIp=87=^73{UU!Cn1cBS_TqAR(Z5+h)fcN+4p#J1d;_~LDEs$`o~#90
zb6A2c`VTQCtc7roSIf)3a@Xc@Y9mDYj{!AKJcBb>?dDG=Efkl^D}e>ZUcusAB!4!R
z(zAO{`HSseQ6Wm2%3p)<%ChR)e>IKEo)ZkDTrPjJ50DPEDkrZsgM^42O%cUn*$wmc
zSS<T&=H-nv66$7Bz{9TpJ#h4t_4GjxVR^IpGhT#m1rD@87#_Us6Y~68*e<g=HiFN+
zlR#-co`%SO_%ARrL9C{cpUw_O)4$u+oPs=dhBsF1To2A*1y=Rn@?M&g^^~D(tHMl=
z{QG|Uv8~v4`Dd(A#Rv)*8|8yoqx!zod?z2qeoTkwUF~!qCD=F6$^d-O$j2T>|B83V
zqvW~f^nDzwftW%`uQ~Vj6NiSUXrX)>(7^YID-Zu&BcBE198n;j``XFER;ulR(V$ws
z2&>VSbqI<6A`vMAitVawDRunYp;-u+89=A8iyCJ6rOlP8L1gb&4vV=mVXneG@t@e~
ze0Kbp{_?d?)WThdO-N@#zVV6F><vJ$jlPNhjfJY=g?7@F@~w}idK2GMwU`<N?U!ri
zy95woumLv1ioXw91830oWIx1dn2zeiFo8?E2$f?TF9TFI$S&>7f(Mf>XiIbKd8vVt
zf7H(H9gR$(LMoN~*uN(QvJeIv$?n?D?S?#?G(zp>FNNmI#I)VpsVQ>cSo9&-!~RXd
zO!oB2c^Fuc#43IvCs6ikrv}4XoS41anZXcdUZ4znNAX>w+Kd|aY3IV8G8rVAk`3-_
zvmj!wtNg?u((V{#Nao`2`?Yh+8t#Q@*=bCS-rs>CN4Bn0iyRXNv{QQ-irJw{FBso}
z?bLGGSg*oxEC0@v2n(LD)7jeIgU*7icxe-uRzojEyoe`kyYs%_Mp-BA{25JN2p{Hu
zrF~$+RB~RRzyxbiS#nkfL)vikyX+X7S6q$G;DzNd4#xBto)u;+k!&v7fp+$($QUe+
zV3Tx-C1zL(>7ev2`#`vxjq;`(Z0~^>mVt#G5<rtVVU7)8<3IIh)YVkJguHa9KcoGg
z?_p*+%#e)ISw+~B5;@%GR2B%rv?fkG8%H=A%-TockLzJ35+f=a<VgR0C%IS&hMw03
zIm%(?yz}MgSm^{N&K(nUfp%nwD5D##tCM0yg&f<i$*f=fOW6_aI0Lyl<9F}z_U1$+
zlbsNlz#)SblPpY53`}S$fR!jd$v&y6E|+fU=Z%_Iza9qY?lxg|ChU74ydHsVMD3um
zn}5Jt3w;hS$2{rf&nQ4so?-=%-f7@vDzM<c(kJKx0u^E&a3cDqw;E^!054iAC;Kxv
zVfEr+cxXSzu4Su4P!n}(yR1&}edCVNE(_la{!5|#)MhGDO{BR3_-XCb=1=7~FQ*4b
z0oN*LFYPm8zqNB9ENvK9qh^$a&5|>lp`_Umb1j#j`Lji^?4S+?@U>F<8&5bA|C0ee
zrfE&&KfD;3tVYk#Tp8%Fs$C0sw7dk$nS*S{dY#-`2B$%qT^Zy8G)gfmVMzL&1)K~a
zlx^7U0LgSR)Bs}Gi405VU!_)JNcS?lT~jw0H1XvUrj&s%G1Gpt8h=rJ2i%F)QWILR
zOqk2bLYK&hz#xJiA-m*dWYG8kbb66dJ{8$Geg3qTkG5yfI8e4ui*VM-7y}Q%0);Zx
zMnI-0w`g2o7Y2Vz^j{%owR2IVnKc_l0Awaq8hws8f<8NoGzs`!3mna+LSmCt!=q#~
zA@Biw@uFn>*u-EFQ1Rx%ST-qrolH8I3?>OV+rK7*p;sE$t(=ocKA*=bK=S9NIkZqK
zWd}LWpK-UaPZ~OYzCUC7gvrzmazUUJofT^2!dSf8On7+c9$k)|qAWwrlBISoa(V#s
z#Q`88v+Yx}&*b7@F9TGVy(9o0on;|aRJ|;;Ck#1<+nQ~Wiwq^GeS{>qii{LDqr_}o
zxL0Y=n=2(TID&d%anJbXA}2^D2Oe;*FxZbl%%!nrWY4g5=XM0CmDx9q^{8&0P2X7J
z0aL@L7%)mCR8FSau|{=I*ptg+KyPmW;a@78pF#P7#9e8_qaIRk$@$ynCA1WcDjSZ1
zeMT<zg;FM_IS*bwkQ>t8q`F<xu{<>@f8yB^gmaoSthGw?j8KB7*_wAfip#ZAyX4>j
zQe#uF2X(pB`V_o^xrGfdRhf-=THtc*1Zt<#gBB*SB<+lZ8AO->`7EbXXH4V5BTkf=
zHpLZ3TzJ_Pa+!Sz=?u9Sa(P-4#5usvS|L`g@`^wnBva;vg2%{}HYSSr%2ffW-jZf<
zH*S1&!sABe#vp3BCIQHhdu@&<*E(tHMmxVVYd=r~Md3PwS`9#Qy$_&<KMcSVcSD@b
z$+>bCt-LXCGK2-EZwgYbz}2oL7;iQnRgz`+7_FPP#2SEY_MKtH0b9eZaVN8+aAhe=
zFy7`6fVJRang@^cBDp;+i+D4Q`i{hbRohaGvM4q(s<EB(<xc-*A<Vq&QVO|P?n<kp
zd<A^~n)%&n9GDgGt#&X}zl>d!_q5nK=2z2<!Tp;r_Zs2`PZ&0KklYtY8rOgPgy93_
z{`8qeZ>4lYc_2-=QEh8bTx~%77m}O&JdWaF7t4cb;fpnLc?qtLhYY=$$tbCVmg+U~
za9S19au)&zv*i~pCUIF5!>57|@<<@Q3|&VG(H5>w9<^=t8iv12gA!F5u_TWLG>RmZ
z?B!PmZQ+!{8B*_SCd;SHPn$c1;UzLF5lOZZ3>u=>Y#$3IbiBI=J&h3AX9xvTfw(r;
zU*Nq*;~C1<%De>BR-ldmES33bATv@yFO~)VjQd3Y8X`Ppp`EFoef!9w7)HYlG-^Z^
z8>2YpP+4NIQstmQpLRa`bP)8cWdY(57hNXHgI7$2#bLdyXxF60x=jAJPM0`h%EL@|
zMVf;X{>nHK&26W2Ju79EZ3Dg6a1*U==ay}_ajm{21M=3SH8b*xeH53;S_5)Gbxzi$
z<ukU@+>7g-AQk`_^h&UtUkiOzHpE)S(aR_s9a}y7$tDLC7ay)ZBAv1yTQW8T5f$R3
z!#MzV@a8xMNY7zi^DRcKY|dgPf~^i4;t}C)qOi0=w%Oc~$gw@ZS26>uf-d`b@Eojf
zx;&94h6Z1(3h|Q;C8av74^T#u{f+<$r2r8EB#kjVG#OU;TZ-%Lso2({>RNf)hoa0v
zuK;5vsDvz(rt~RoG8Dwu$uj{9YJBDMsQXYM&+fFj3!b`GLk!_%mFMiVpt~a?{9;M4
zZ`H_eLa;-)hZ5cMzG0SrWP+C0O@ZPuOJ0a&c~hL<2Gpum;_!ic5*Z8RcL6oztp(UX
znzdgvKz2NYWA{=_jd1T8CeZ)hKj2=IdCY~-_J^3Rumnz=yu54#k}a*4KN|AN88;|@
zijzJFVFJZqCi05CISL}pp94lZb^%FqDA`|v-g?Pj{U!6I1O3=ZyDeS~K!>sm4bH*e
z?5kn5Me>?|WiV^1BB+$2<?Ah0!z3(>CvW&O^<wZM215THG|(O4J8#+_$SkU;leglR
z0t9osov_JrP?Y|*$~*B4cTLzm|KV~Yj}41W^4`~Dqu;~@_^zQfRYiHvM>0DWN8cdt
zJ8pXQm4EswJ!`au-u*$~23Ax}J^e6XQb&&6SRdI)y}R|6fB8se+5pz*Y=J<tN<Q`v
zSqhI{Cv=08Pwc~SqXx>S!I8j_s&ggmW{GqES)2l-A|UXau_>P?j<cB#Aw_`eTriKX
zkuQRxaBe8F|J&y^WlH&seCf|9%rT;jmBqhG3uA~q%gbz)|0I}^apBbRb&LnHXMhtO
ze&3{TwU!roko-5^!*Qm@gHRs%Hekn%tR+~qrX1rNdFU6)cWE4!bysdG-=}BGp%=sR
z2GQvUf2QIqwx6wwq<xs>XeQF}@HVqScWLiNJeW^;=h9aFBY&oOJa~_6=*Rv{>jq%g
z80E2R`zAI(;2nVw;d0rnz3SbK^~3XpvU_{AY5{<4K-L~DrXfRT@vc4X4Kx{tuR-?m
zHxMdEnMh49d$)IMAo}7VX0THB@s|{K=<AT}WM7*#2Ukv&{3O0W#qoSCc*B0aa#rJ7
zr!~3!zp{Vegvv8OLcLMn#RFok5cXLNi~Va3Y_Ar?{sa0;tM*~+Q&|YIg|u$(#?8VK
zG`nR8Qkxh9wgRgpxIx;sH`Al_Trgckd;b9MYk}i->6;}`=<Fc2cPNz`;ThSNVZVc6
zpLR&d4U?r~ptFjR=bhr$VW?gQ(WbL+#{%d^Yov?Mq#X@09tzdSLGe9$*X8Bl_G)9z
z<|}e?NQ_lnR}2Y3e%jtmq;#LrrOTsqsA0ajb;n^gLSYf^ox^=TyG&Hzl_Tu;=~;xn
z9hu<Sz#LaC*n5;iJRi-<ua%=6cUEUoON@a?%<4$*^f9sVMR2__w6|+})xu_0%CWK0
zdIRvNvjP&!!yOk0r#Ap-n2k+1KBmQc>*a(rtr`qzcvFh@mPvA=kDWqJ)hht;RZeQ}
z#=SwPO_7mq2?(ohOAED0G~FGD6rm64q=)0K4v6g;U$Tu20Jpjxw1f8w@NvC*^s@A}
zFX4X7OP@q<C}#i~11H3sEFQ)~xm5b5nOS<&xX{V|jG25i0I?%pzw`{bevsf$u2fD*
z&s2E<r(G_m8tNdzGDqaJpc(`?mq0=}-Pgf1VV2)rs)7!qt<Fd*wEiaZldg7VAQ62|
zil@rYVn{>+A|z1yI}p^q*<)`(;~W`apA@rdBj|phZztEYQ3eI&YSvo5Do4q7u)PB7
z7Of0J{Gg91kf8~GYk^-1F}Ry7s0kb!`K|%Cix$nidU(vrG6OQgIMgVuemT-%w%i6l
z3_)j<J%hQdx{$>(m|8sA;6SrT8Do3eSU+H_jP*J3IoPC09{lf88D}S;JR1hLv(gr^
zYymG_1jZR)$UDfv@K)dp<W-E+LCt+gl#B&V@DW%xh}CS^u4H2DK|x^!GEiiaEe_`G
z%w#z`_+SONqB@Dr2_S$m92!-V&kY8OEkd3QxyN~Ne8=ImFz@;NL@ksD$>}L!Gk59&
z9}x~Gm|W-_+$0wUos`w0c)45@lY+6)VAX(EF7{{Cj$kcg&j-0AhJsg`AuG7>r7(TF
z9AwI>X^W-E@kG^uL3a#tFw=$!iN%4FVl+!f;97}~Vbu=ow#(Q%v_a=;OUq<(nkTG<
z4N@BLKnt*9SL>uKHh*ePva$J~8jZe&v(6xx5?@kIckVXT-`+`y0wNJ3XSCI)PdBNE
zm(=~kj>we1Siz~%X_$d@=o)I^XVC}&rNPm8A%VXI5)lldh{wFtSnm19O114rt{C2b
zyqx!!7pRRAVg)43njn?3Tm|*Eywv(zTACb~--2nU1$ek1S@@Unf@YVS9($sJOs4tT
zahMNB%wn10U-B7y_|QE-u>saITRoBmW#&#NoMv{`Dx{XE;TCdPtWtvuIMfokJU}vx
zS#4>ceT!USs~(O>WwuVZ(sr(nWtpyOK51$lpgV=e3R;fl;<mWjC!!mK4h~}InmDWQ
zT2#ojfgSutC)W~)rPtX?N+5w;Z*$jHAR7ti=?42#tNg;=eq#&|YdVW8Gv((dUx!tj
zF(VThL4({JZ|kX3a+h0z#M#UiChA+`9UPivimiEan}IVNuC3dHYD-XKF3xsA4Z{N~
z<j(dHX28yv=4*pAq-qQ#U2YcSk-K6>RK!HEirgK4k7Ze+YPrYZH`(1hLqzEkxi{W~
zD8T%w`~1xap<%uzZNQu4egiXClA9q981&<YkCC4n_Q1ka(Rk2U95m>Xj67t2G}aFR
z3=g+=Q%Nm4C>5fFgXOsfZgAH!R+?wykw5}GC}@&cBaa3XL`)M6@VI(@>3A_Xt+n_6
zo?rBs0oPbxlqsyp%dg@}L!F7zGGw`Mq<SYemabyx1pJs-F#>V1_h^$Qv*LH9VY80X
zb-<j5MYxe?+Z+hU&{)SgJCQt(0l6y0uDQNc)&WAtS7vm`Jbz|Z44%2=GT*Q^a9Dr%
z>=*2`bGVo32EnOYBnxd1dUo$Fi-P*mv$94OJ2RpwX`3vGHAkmTu6e0XOSzhBW+|X$
z%{zfTryG5qEca(xB$V%A-W3L3Inu&1veM=nh=R1Zd{^0Dqcc2NvN}Fio2<p+u5n}n
zBe+?3ue0|p1zB}x%G&l#W&aNo31nSD1g}mj^tRqnZdE`nqQ(rY8yxzGSTSEgvK`9C
z*ax`LC(EWd*Avc>h9JaZ1S(FI%|5Nx!T~9;Q2CbhjFp+;5rt5(H8u~bJTe2&h!QHr
zHrx1axRJKU0qb90B9A+Cx^<?E1Gi)wO0M!myj+AZt|#LU#Hq`Um>SAD4N7Z44voeb
zil<iQrn3&@Q^6=IL1Xy)=>RPcab{+&stgVSX^JymnW=@{3$dFF+IYqQ9yNS`Ji8M)
zQ*EPr2NJ0~XV@B<MR3Zm1Gq8JV}IkHs3ie3h&Tkl$e$0SPA+F@y%&OQ<PjLfrh~tY
zFPXrhhtQLN<ahpyX^KXSc9Xwouojk9$V&;IRzr9>Y}y*eK&;}v{k>rg&FyQ|lTX#<
z4*>%lg?SoYUJe)sqC^vVmfE#nUMzodoa2V3xJK`#yb@m{?yZEi75*GZFJ)8HYWa&H
z3Fhq0@>d@~J^UbD$lda4FiPY`AX<;TF#cvU4$8}GzG#{%SdEOaXe!6b>oE=5B{L-P
zji7UA3ABx|*_Ot+=zn2~Gdf54Pl`8hnsfqYjW!W_3A`2P)fjUv%cefpCGxh-gxY7D
zc3cZ8d#4!-#wUS=fB5Vw3+jxF)m*JTap)0x*S4rP=>I*R$fCVy*jG^xoAbV5YBUnE
z<)1zY6<PHrLY@7<mVm9CgIoN=7E>~$2NniyGFXDjIr32&VLA9OZ_ZWnFC)^y<xoQ8
z<3uZomjT0D2(w;kpR{kX5}z$PS3~&yG&Ves#%Hi)%4ca)ms4^<spn<#xlIJ;F}Cvy
ze}UnJRL$x9{dYhIy^RG!DCK_{TZ(qARfyaDDp(T^WnTW1ra^Rwp;*3-FA8dSe&m~=
zTwH4~6aUvgp*EnZ9ZVzN#xT4nfiRkLzq8#LGjX(h-@d7nG4ubQYT)vN4?ukpo=4oe
zi$pq@m5mo9vwImDuuBKEus~F-kRSQ)EG+~poVI`XaR+lFSIMp&JU|T9%cjb19Xx=o
zMP!6pOqCmKlHKiX%(5PKlRa#fDYZ~3_w3NL8(^=k<T2c<gP8=4U04NQ0@fgV`!AN-
zXa6-VV;`TI5&+YZHmjkMQmksTvae0XCcN?!Uy`Z&RGYWRem*jdzO=Vc;INVR_qU;|
zfwx<r{{bDEtiS5`3`?8IfgRk)rjdCtUrVbFZetmT=xg*rjXzOB0cW^P2NN@?`ECV)
ztI37>4`!Z?n_80}q;2z>)EvD*^#qC#6pQ9bJ4YwD2F`YShZEW#MnIQBkYxvd14e5Y
zPdj#Sn<cCuDvNdUPb@@$*2Hbnxr1BKC=H^5uyU9#{vzRJVI@=omxGK5PCr5pwpkFB
z!Fn=>_`5Wf@fu(+`k%(WK{ud5nxeE}_fUVEjt%NUC=4lwb+8=jetqTeKx}3POeR?<
z&_~2ZAykO%EY&K90L{g6r2k^Ie(5M8KVh8QQMUQR5uYeWJ5p&wr_&pi+VsB=jBt|Y
z%Q3N34F@q1rE8*Vk_DAOi;{@^lw*y0p$*h#<hZm}#z)bpMKd~$Nypo3xNUI2PcWpx
zMp_1_PV`w>a~c^Ygi20IsIia_TH3oAY9@{!F5NqXFBFBuYpY>g<OSFxz-JX5%uDx9
z&!AS=aM)+Om%l^0qsbjced!(F;htj7`C>fl<J3dbZiDm<#H%bx)r*u%t@Y#}8P;G0
z$)MqVo%D;H!}35$cyfw?F`ULP9)eSSKplic1p|g2KRL~hhYG1A7wTd-9`kt@JjHxV
znlVpru?#BE@Y%>2hGLj%Kq_Z;@LK>kXNyRppT${(4-TxSzYnAC9_-h4WakWs4cFLJ
z%1x{+X<>ed`<8(=O=JB?NKAu#9SRlp4vj(a-&K$q2B)Dn=ncp4ks-0~xuQ&+3{7Z9
zbc!;9Wil+@K)v;>43DXXk1voBM#zozhn*lJ6Z%*ZsAZ_?C?7P+tkPsMI)So^ZXz(~
z6*9(0&=_8woS_G-|2@{<Mr&=Rri1rxlyR}4<Z%c|$jezVreew5Ipcln%$aV6V?#_!
zHMpRI`zaVtia45VPA2$Z_Oh{<iAL?y5son_NCFXMAOM7WIvHT)Td^W52zz#{2$vb7
z2EA4TsP~+}+^A7<u5qq0qvbrGH2B(f1ZSL|o~alMyU=R6z)u1SP$Qb;LjMbC>2UH~
zWPg=Wqpy^UZJq))+_@wOZQx8)M3+J*Dll)=b~iF_tOb`Uiq~1yrc{b;Bu2DLNf5!P
z;iDzvi^P&~E!gE@)IT{k#I7n_5(}i%ft`=a^h>2It^jB_2*g5Zx<u)S;Q?N0Nal@8
z`(gvGgDG}s855(*#Z;eyGC9-h=o~7KL0Q;@?OZGTB|^peKFbkzS?N$4NoXPnV5gri
zm)@pS!XXz(RoZTgM@5}%W1?KIu9|N9OKl^k*GaXlvax;wqX+Y8&8V9yHGy{}3fL|f
z8w{Ad(umX=qDRz%M@@@yN(!e}%5;YrSChwj&`ZJG<QcJ7g?V(Bm%4-%(_L0!!87Ay
z1oKh~F}oTUd|8|%CV-Zs*0x;k@7hW9G-_g4u85Ht+KByuTrOAo@*qaxud57{!eX$0
zF!!td1F(L0R;YZ;k!xauQ2ZcUg_o|igH?ru=j#GY5IQ4SnzQ7tH@wtV;aa&N?GaT-
z)_C0}H>PVPR}E{2+!R=ckg!&64ouF>0ztQ=WiVTk;>b3+)%TclGNK*Q2j*uT_?Dsm
zx22igBnj%cJv}2cuuJHU*pWQs=W4k#aEwx0b~D~(NKmuBRw24OUKw3jS_mjJa*yMJ
zDi|~p$OEkK!?Nf11_oJE47NA9&p{yP;Qfx0-g3Voj2W-;fGr<uhYeYNp7xh%GQr|#
zN_;SJLe^8qr+IlOK|#L+hh}~$^A;brZ5udhr2Hb#TYv;QdBjQFu1X+8PjPdRxJP|R
zbJDu}MFX7t(qKlAS*1Ld*2G>V%w=9Czw&3y17itCNC#3Mh=X>8$aPpJW~B|W9xkkm
zJ=>p=l`n$k3#KP?Vl>lgAhs=(xei<S0$`H^owN19yg)jthG?<g`7Nel><UeK3j&xa
zwNRtjhJ0a&Bd|K6dURnq+7`(oCuZLGivxGV&;(GH1mQr|#Zs2qpLF$5o>~^e=aBSM
zP2=wJ;Kc*_7sv{S$RcM@>KiCqnHIT4wO6LvuCk8`CiI^$SXKvs+R|fMRxKB6VzF##
zMWf-`=EVZ@R$g~qIx-Y{t)!3#kpLOAGO)zvdLxtmSlou#-XqVDjS1GZs$#5wG_uK1
z0V>}FTs9=i(|}x!OtpsPo9)}idPGsvsgAG;+0qOkxU-U$Yz<_hKrWo;+v3lH=~=i@
zw%Zyl2$<zcA2;xJJcv8?i2x6GEFx6FpqI;&{){5GMyNt6-4VMt6n%uGF}=a+ohD~L
z70ZC!MxVHq@?b*E)4p7$NvxA5yFSL}<r)7Xd@+=ZDfKLsX9FrQgE^pL*2{S=##c6h
zC~}CWdZy*qX&kaNvJMLK4G>I!<Ks{wr&hETrrqab%g4jZ#7;ji*v8eC*UE4GeG8RY
z2v-gw*ZdhQ9WYr^V4g28#u(!!jvqI+V6eOt07LW9w0s=J-=}8;ORsnNgM&(5?y4?+
zVM4;o{x*a6uT_q`TK?z^CT?^p7r3dhaNgH*<WEkR#(KDL<UyzfpJ*gUXHDx@Y-IQc
zkeyLhgX;=5x44~Kgnm-ihnlIP*I(?Gfo;tW<X;_bLPFaS<192@MP5y-T&6L^D2gS2
zOV5l5up)Ua&cvjQydM7^S|@K9ag4z{FfIPw;UfPfbyEc}bzS{Y@}`eW{E_gM;9hzw
zaEf?R;@+YKn2P(hzda9XIfYgHSKf*H6DEW_Y)1cxtt`prW{|7D>mRuS1W)irdCw`B
z;@kpx-zUU1uUkb8#RB*LX$!mH3IyVR5TL4VXV>M2aWWyaqAKo3>6K+XXy~qhgw|LM
zt~f-7s?tuM#7g<t=yWu6fKOr^D8F!qeCpr8N=l~X*>MzPKC>5RmSwVNcqX4Ggs`pw
z&&WpkBCUq9C5Z^)7TW{Jztgacb)iGfZ&CTu$Kk4Ipkuh&S7~f!hO_h5Z26CYeDSfz
z%GbfbAuGx^{w|isa5e~sE9JkoF+1*tJ@Ia6gWvi)<g}LYv_`%Q9-g_h7Iy{u#Uf1j
zdqaAz5Ltac#4N)G4ixEVMmCy_DPJ$ak=doAS-B`lv&c1#pj`7(`BBHF4lst{Ju6#e
zc~syT*aRxRKkn#8Hx$q=m0deFwdUKnpY<S9Syuh1M0V@wYSw77Y#5L8vb)d9Bg6iJ
zso}Nk5u;vOTbP$U9b)EOA(LK16!-FHbc><Nms)oAPOH#z8xS{1Dui;M^ft;+>&~o~
zef^oP6Kkdx`jeP<KrPyv?Pq@>riMM*_wU$rFxQ6nRqQ0&(Zh~8)K#i7@&O&2n5+nc
z75kpZfj-H6a1nAT^yZdU2^c1|^R`<ptsTwaO?tMO%77}=N;2eWZ92M9Z$PGUN+FH=
zZ9A%AYBW6P?K-LnbV`Y|cMR}=u%Z{WQ47uNkbVI^fC2WV)Eh=R+A8UFfSsa~17dq(
z?h&oK>h#dr-bNZ3<$^8=Cl;*4_>ky$+e{+|`GP_HYS!iLcW_5F>mpXTP7X<Qz)X1}
ziMJ9R&p++xRzhTA6;ee!v}2PQJIk~1GuFsqHgscswsH#MTUAUtJPoHYF&Yd(&c8{H
zi0y@RRw76Gdn`QHoQOgGa8%GLBKI^vszi>CiD7YII^Qv|iS{v|T`XPwnI_$lCXv^2
zY)lD^<>k2a6-vJoqfcEY$ERm(Pr~S$m2yICH&kPI2OtWb7-Pfn4A}}i3bAt?)liMd
zP$czr>lo37rKPBiWfxR;Ba*`6Lg^8xL^zjL-RNn*_3Yh8dIbo$#@Qg1zU|)W86(_y
z+cZd@M5WtdHLwg=vN;Oqb73XyYeV+x-Ahh(<kF(O1fN4@0M!w>{nFZ8%aBw2-Gz{0
zv|24C9^L;_0|-xwU`X9*4n#PnsS8-qGv!-ydisWmIh5U2${Dr`46=k5^~|)r<)s<=
zR*L0k@ez)UGIfMF^-s?>>OI7if&q!PHtD@iLoqS1QU>}j%wz>~#5STMG5ts~C}G~Z
zqsibju*EbXR8fZbGmG7!D6vq62Bx65;hq9hA7-bfTet4LWw`wt^=bObh@direP?8(
zW5}3}aY$|;C0Q^Q8D(D<5f`KF-&IV*9%KJ*tRGY$V*^GU9Edi!C05J0z*|Wz3b|<i
z#j|`WG92osn@j-(Zd7}h@o{We>7}w(3KCgR9EMc@@)#NL7MWnQGivY{nHW1v)m}*p
z$;2qez*-_JV9J~na7=^O5e7hjezt**z>XPmPD14xusUi(tK{5(0~!vZ&*Z!~I>osn
zIY0gw0;|gf{<U3Un&KfB8f-%cOpuFw0c0A8?odN8LhK+HTpU=}v~AKIgGqSbke6<d
zOJbIb&_AirW@)U4F9CYF@~xtn8B_!DUK2u#18!9PMy*LH@pmjjn3CWI$=K%O5WsIH
zlM@}bsN<(5|595;MNy%YrENpAL_oP(vwb(XVQGtG(+(hr8lW&ZyJfQfl%Tf(nQ~;D
z+MS2?E)W8yO1Xbb21r{tokT6A)e2t&)emciyjUvz8O_A>q%6js<pN>TQS-7Y;oXWC
zu&O6QY@wH3n$|*T!-Rh1A>t64Tff>!gUNJ~4|LMj3$?MvegoX-*6KWh5%hVPmNrB0
zOXcs=(@d6nU_s`VmB=Wc8TJi|DVR@}5gT<Kn@)i|(6ahj<Nv5Rvdj$j0oVT)2G?BX
zpz++01g}Mi+vWbuGz9uj*UJ^@y-j-Av!~LP{*1OQ#!@gv=Bl6=)sb*+!8TP8l&k%1
zvLjl5C~sfmRFs9~E=|e{)XaO=1%_enV2!)$<L`Vq5}@;PLo-k+8!pGSe4_(kC91eA
z-U77I+!SB4<V{hX+-&<*ltZM#EkPbb5Po!P5bRFc&KzYQ%OQr}7B7!R#n0Ol5-9E3
z>AfSc%p46i*t#<wproS8I(ON^(OygL4#r%907le(fweB4KuBM`;~rR9?o9-ns?h*N
zbb;L$iyB!;%~S3-*px%c1NH|pb<i^E=LV>jKWB**8i5{+uZwE4<t4Dt#Zjz6H7a@7
zK8Nobul~Xr18i$g&XY%c)hy0H8xCS;<<W%sRqD`Ef`Z(<RDS7i)BMONes<@5%%35s
z8>h><ll;nL&%=+G`kly#G9&d9_I90Rn5%~KV0J)Xh;%;mNSxEL$rCeEjsosZBnIPC
zi!!C|MvZ;8%(bb87RtPkbu!Fsb7g#f8X9@5xf)pzYvb89YY8_1t6u2qAdRtd@@m)-
z7a3(W){ht~iw)jE5c8HKG;CHDNZ0AoKm#Boq0;TREG?ead6*2<fy{<W%l&OCY?PzW
z3sP42GhQMP!*P9Vl9k3pvy|K*S+7bHn=}kABdg=yLn%zyX!zQipi<NZt1JOWTWeTA
zbA%jhkkzuzX@IT6m&y7d78DtrDjU)q@Es|_*T}}e8fCR~_p%Pbrg+;EEZ~ik2Ai#b
zb&T1*YjYg4{!lcx_)rvS8BxjoB(_${Rv*fgSk>lbTPzMYUA7bnp{26jwF@AKmTE()
zDpHnYXS&_vvF}4sGJQs!JYl<s^~;mKc!WAbcH7ah$pUe-oGD^#?b>aO+lxM^3VF&t
zFRYy`Pp8l4K#Xfjx=wT9Muh;FYc?xA<4dC)N1b*Zu>P!JB{!9h%I5;Xnk0jTfQQ>c
ze(gLAp)@!;C~F1@k_KxFxmhp2ac~$Lu~eQ<Z<AlEks4f^;oDkpL0+)aGIEssHi5*m
zM8Ra1{LUFELWbqVz)MZ)mCVq+mtrpm7mSnN$L1qg4gPQJ;s*JHjWGp@Q1Wt`gFbQI
zKWpWW31Q6&b68=_SHdXjKcz1hYvL}|^;i5EW9kv}iqYoFpVND0H{65+@fXJ>RYq;U
zrgAzQt@76(^r_jBN_jPXyA1h`Dq_puY}=t=)nuSfzZU0z{6Kl#Unje!?gGn)yb<J?
znNe1V7RFG?{_d|~b`9fO0+ZOAzDE$40P|bv1dy>ZxWUrpSIXOg+-!BVyc37GDu+hu
z82KLo3_9Ar8$<}pz&3d=h+Q*tYUF+2$3k3=|4i_p{tzgF&hroK_c|mP$cOfAQ7t$J
zt(_kkrT6I5UH+9;yc*ozEW95Fp(=I54+zq-oPtjR2=sC;1Gu01*A(7p7g_=)`I$ds
z`2fhlNT*vOpQn>n3?Ht1VH7%kocueca%yK3teuH3W2$k5)i^%#m7%Htj;tEk?hwuv
z7cb(FkzomFzD`#PIT!^`2FPrLWsm>nQDG$tJb~{0P3&rM4#BW`zy3D{M0w#-3|J%I
z8V(VJSt;KIJkVxXaN~PhCVdgG3uwDT`5^|)Rzhsi!e`P+jWxYd<8v4LX9@iCKkB5m
zqQ?4(==xnQKknqFv>n&tvfZ_l8c3kr50~9KsiL1LqJMUG`)VMIQ^CTtM<+G@Tz0wa
zY5%(74B4xbb2`QjTjL={$?o0BEPv+Gq`bCL_UUAnKjUObA_Sq@&Ax9Zv;4KNZs91J
z{SSJp{G^lHW(Y>MI4}DJ{9~&TfxEvC+%htsw&pEzfUOKE`E_z&nna8BL%pLXoLo$O
z0n*CG#3c=Lr?mD3Xvi7ZF1AzJ_&nTVIzL%H*=VP&A<F6-<fUDDk#(aOO~P<`inQ$w
zz2k=u9X3Ha7#7dTO2>pQGi9yBu5^lLdX~v|=1ON<5A%DaOCY5hiYW+G4)O({Q3#fP
zu>CbPg9f(Ma!7)Roz1DJ@|Ht|9QmpLqVZ75P8mVMtMkxKO=bYgLxTYU9u@$BJ**;C
z<7znE*3(#DFnEF-5idYmp;4F8%8~IxkDlG+D5uSC-Q{RoNinbTV-g5dfJ_CXF1JFu
z`iKncTjx_AYlB>T@iB6o|IF%ntj7cmSdNbg`y(gjgidOynpnxUc(st;GElouv;k@n
zl75oUJm2m1OBetGhr<Ht78C{!RwCUKYUjZ$)FZ7ESvz>22DzU8jE&T<0~(pN-7Aew
z;UD;*9YT%7z5Q)c62u{@_R^=5+FjL}3!60n?He#;P%c(ZHjJU)bfxt3kts4;o+ue6
zgiJZ5#mF=Yz@jFn`rFjuX&GkOT4w6bmec$fJCZE5#PA|k$v@r4fqty!brxPa!?8fg
za+RFvZ}9$MEo(T><!64z%;65RwZ<Z<Kc9(z(m%dI3IO9FxM2hQo95^T-owqZ$iP@D
zo>}Qp%!bI6H8Lo{h=35#%_3;>gVShQ>K;NkMWbd`GsM==zqS-Ns|*bcBa*#ThIMK>
zkP0-DJ}dJqj;*H6ybL$g@3;X*j}gI1NAU)gkpY7>Q3s!xEu-uij9ah()uI>~Zp6_x
z1ab&w$e1`-h)c;;6v@~i*@6AX$v9tt8uYMDrkX%dBb&m3I?D!v8M;Wu`-ULK42JuH
zKrYQtK+y*9n+g7mzEjj9V^+0HOwY({c#hFd!Olc>YMGpE6VP}i=fsdyuDC0pjB|}n
zX$J?M7{h&@!CzQi0WNWVd_v(5pD@b!g7^fv=vfx+kPDl)2~5;7bZE##N(#B9TohlZ
zF$(1+<YFJ!NR?18lS_<&o^+TL?o7y>6)R)vSy6n4`-D|w7lG#$+q1@cT<r)uz%Ca{
zNv9@N4|yaR-p-=p7PO_Uf<UZvZknY%6C|l=#)R8LYGtzj(sK7$CUc>b8r`78s7_ZF
z+<SVq5>+Vkk_|X{jIr;$Ev9rbKako|b*KKn+{;saP8>Q-<F?FMGT(9^6G4_Z#K1Bz
zTM@*^zM$BzO5X`Hp4pM*rDv7v)=DnHwHb$uVV-NGDt&_%i+WfY9j&4i19|0A|HfFJ
zT?ExJ@AO<XjS2ZUdgE&$P}cYtBr#QVA#&FS^f+2(ci$w_93w^rQ0Jf`SbnsHaItX0
zDw%F1+`V6~eljDSr%l+`<io?ib;VM*6Xnov#5O31UY41$%PuOgdj-|V1#($XZ?iTA
z%vdocgi6chJB^9xA+S}e&vb>Y8gA}N)OEVjaKc(C8fAEuox&a`_mrz+3N`C8CTp7#
z06hi}Fy$Km9gYljBy)0_96WMLt~C^#J$C#cxh}nn9xlp8ka5~#i?6qL8Luoiq>Z3u
zSOr+QF}AL%wwxu(Z;E$}{2@clmz#r!72<T?;xEv@jlu%p1ncD1*zZbur*89o<T*ip
z5C-DeGLg(WX|d}1?ZInH%Mms#cO;x_VsQw{WcY}dU)?sjGvGwB7>=3TW$P-$|K#pC
zZw0kQOz4+;{Krz5)b6!^!@tp6?u*6w6*L#x#Qg#LfE<K@2ND>~MrY$)41_{6R@+4=
z3G!eJ2DStHDWr@o@{m769ckF_c+)MChn-+pt5Y>`WnYZ-{DlqH4P}2Gv48eL+k!_8
z;A%3CUj~v`w<9l)naHEXkhxG?K1fORIjJo}-yy@2`c6&z(aDV9Ayr;hJURQ)OJ$ZV
zb;nZd>ROp?yNa$_7<rEUL6<)Yvju}5hXUbQHZGgzbA&8`O2RVVMnko(O0b#*wt%t7
zB3zgj95!|nNw&%&|HbU;mU3&{aIsHdN-$fku979Oc(iAMJ8EgbK=})j46=P?neF@_
zV8`k>EVq|{p<0CZnsIAIkOzx$K}?jDJ`&CeEJ}6`R{6%e%bRQ+`)9PsK*-g$D~<I7
z2FV&j77Xc&U@7x8@#wO90C+RG4`M9V1s-%`lo(t@T<^b7#AB~V@(@L>YzSEEvKcMK
zxY0=&rJ!Y#odGYkmY0SgR<J5HykVFYDQ~hl^fy##KuSg=Q*H?ysi_zmBs_v!n^8h$
zL+hAq3o;m}d4RCxY!A4RJ;9T0NgOn_JZ|qI)jG=}Z^KLE30ny&YB-%f=~Lh{oB*nT
zvt>stHwPuTGCNZm4Yf!oL3H_3ojlgV0u^b?d)f!F&ep&gMDZqv%4+Fq0{{{}?!#x&
zl#6Hz#{IKVo{f#1R4C8I3K29}t77@DJ2jQ7nD{^QuKeG;Ex)lfL)^k1KOcNf%f463
z3wEyhqM+Ar4f6hj291~B#eRXdnA<W3R)`ma-ayuj*Pu*eF_tg6{G+9e@b+p|ZRY5m
z*}xJ7&4Ft2`v72ES+*Q5&OaC{QPV^TYk^IpTK95bPFo1D8_OT<wC0Lg8Vb!{oi^?2
zw*OiY5B;ya;tOz{3LtHhKc{ER9KkQN&6yZDxmNyanCRKNhrH@9GUyt{aHKUMEH2<i
zqCTuD@>*aBv5d$*12#6x>j^_-#gsEWU_;)pY0e%ie^2l5_G)>RzUjYMX+`^*lEDUf
ztHn2ApfmH}+b+d+m3M;Ft8y0k`wv?W98&Ub9OM?RPDdOJyNl($!1naQYMQk3@_t|d
zav@bgs;U1p`sm)TcQ^UK1|K-8zkC=MTyY`dJ3g`>h7TGn|MD^PTwshj&>uebl`vb1
zdIrP}=S-i(+Xx=i6b)QppV~9zFT>Q1uD`saKJ&GhTpM{8)E5{dwn;v>Lp`Jx)*Tqi
z8B+U2@TtL-)3Vh$?H>h-`nO+KP(pECeHqMv#UIGVOXMqCaAgK=y#K^1anbjXuUjCL
z6(H4zt>TCYtlz}jn(FXhJAlZvC@ht4?GFHl=+*CnigC;~D8W#1pi7=&<RbY#P~NAP
z{9qr#mjo$EI;#;5Jd$Z|yL2{-87-pBcxV6~e$?44W{!Ob2Q1%@(oyo`&K8mdTi&kr
zPed)rZvMVq!AJ<C0Ube&b7Xh_r6vCZbz-DFI-BjS`BI?kq`4;Jo@q?V+Gq`!yxn^_
z{LIx|Dtj9kxv7P+Pkhp%R;a21vTtWKL&DUFptrpIq_c;bO{RB}!pwe-1r034Y;0$^
ze_#fOovA~!agq;+FB$s=<5#5|7+*5v7GJJ}oZiZ&Ypfq!Q3y=6?yUC9Q4^$%Lug8m
zdWEoP8ok<f_WK0j(BFsH!FG<JmW9q}j$>~IYPanT56F*T=g|%hhniym?J}6gJ0={r
zEUD6*a?&ZmPO}HXXYoyEdxl!vT1t50sQ%-nOJ_CImS)jS;2{5oG6mh1ENpmi%nkpQ
zRXkWBhxjv%oCQ2ghx%zO5~k$h48nd<EG|$rKpS@)7Wf^7tP=H*9d7$u3oG>z2_dY@
z!)}pmN7JC{A+RV%#>e@o@E$Vw;V658O)wL2ogAIOw=}-ZP}PsI<srDEQo5$!EgTj>
zo^otwwN2v;jv6n=8P*ZjG7a@OkN4M+Oam^0YXQRN2{GKT0y)v3&t_`nq|R#AL@F@y
zcCqqqMinK~rF+n=GB!IzMz=j|=0eRK>lrf_mRFTYufVUWXzD+d-u}!RSHnNIUHbSl
zgmE!CidD^}Z#qHSa6_CNUm=a{I&uXV==udJ5s`yN9YEzNiFDa>08$uvJ3Xj!YJd+`
z#vtUBb0eo2Ch{nUBd6QHk+obdX9VnEU2I9a45<-k8ahf5sFr4G_CgxZ=<nY#=bBO}
zja~yfyNy0yp4Hkp1B1xGzl-n)62YJ#vcag^Q6+<Y0u{t~MJLuW#0DD=;sZIN812yX
zD*L!`)iB<}xVc4!87x}IrvHH9qlQm7SB3{BA@tJZikMj$kzlscz!WV<re~CMcpEQ}
zQJtIGSQO3wR!%@h`}_2@(&V`YiujnIvZ5*(n=odX6fh(ct2lAvjQ)Cc>mg_PODsuF
zmmjrQ<2~c;r6Ui!LJHCfPz(pBn-nx5o<Zf<j!$UanwX9vxrtgobVzISvRo$lWYK$*
z#kft2I6Jln%^*r@^{zZ82pj5JUe1jTEy<$b1`f)3fh}f(z)r{Bedqf#Q;BG%m?Ib1
z0Sh;kTo?ll1DBs(SSJ_7IUWgmyg03jmLJMe^W~BNq%4;!k-~UG%^Mo3EJf+rD#)hR
zt*h9+f^!5ZOHyJ`AgNT1xRSBI87d8imzT*2a`H;9X0epoJ77T7nas?LlsWcL{flP}
zs(LsJvyOMt79&V$SONJm!VUNB3YlU{MDt^KtEMJkSyjdkYq<|ihrTwEz%QhNg;HUk
zR%!j-$~ZCTN0XT*IUfn!7XA>HRhuhS>6sRLrm!!U8Y<xTVs)?TSQh?fTkE`kK@wuB
zJ_~BpBtn6SROJ_v7zKn{|Hb&Bc@I?RlW8`KDiF(MdVo4X6a8oSXP89g21<l_#HB7?
zEWy>CmzmCxiqO;evcQz08coY^z%KV$z}gWBaD~6Y3z$Li%u2n|pCM2WiAQQHldJrz
zdaQA=Ty0ZofpSptHNJb=wg(8XOa$0Iu@tt_xYj5(2hW)i&vn6uS;nG{=HlyP5&7&4
zaJC!bg~6qD2%ds;cVk)#y9m<4y+&?|lYqKwmF38Tzd1q9;x}r1m0OGp6xX2jX(AGf
zkV*YX<+k_*oPjd9T2W{}liLHZNyx{XDR%^efX_yKmg@U~QUr2k$L2c&@bc``46@ZS
za+l4Iv;?_3cw5P2xhElhJ9vl&S<1a}kSCyE`hBrr&m3aJ#r?59Fl+?cLxWo`52S$@
zkD{zpe(uj0rNnkFYY{~DU|J$WMk%SWXafxi5BV<^mJKSUCdf!GdDsA|Km>f5{KEcV
zW0jmdV%OTRN%E-w3)S6x9D4t#!X#s|JQgP=4^0)!<X3@3T&kKIt7aUHv;ZBhkb1*e
z!Qg>1E5L@&n_xGnXA(Onv+bp6xnh}<)~jb0JYc!Z^$+L3Yc_ue{tE#UssP*^^E$ha
z9WG7QkRSttq&YvZKLGJZ3j!y&Pc#TY&FF}Rl7(?f!JL=KBKxQiqEB8H8$`Xj^^+w5
z5rj_$(=q94X&e(Z%<3IQo`7BBzqiV=SSpMrI~C!S<wiJl<={Rmf*=w43NQ6apNN_W
z?`FyltNa;s^t8#VV|6er$(oo6n4&rk^Mq5$TK|;v!4M3bAz2q|LDftKlr!skeVT#+
zRETOqH0lP&3~CpnSj)zk!+8&A(CKZHy@&XS446lQj|3x<a&zP<%jPsX4=REh5J9jd
z){O=(xaYRUXSivY4}+Ivn*%Tp?7Ts?2X}yIigvVlc|0D#Du)N9@`QsJtk?>KZ^NbW
zq#+!NZ6Dc@MqLip*BFDq&!x2<pGvTj3m^yp=CP*(*)?b|+vG2Bt0~xTg8cGKyfAL?
zczHIBp!XSGn=8*bXch!xUAgAt)GwD`$1m7iR<&w~vnIbu;}8-Pzn=GJaI`S&8YM?J
z%M1RDwXrD-u?){|ZNWRPM`+;h5}lIOdLb|^l|fFjP+p9krI@U@72Qc_rzJ1N&O#hT
z#{)wZ_@=Y{KLr09Q7A75U|^f<+C!VmA05*?tc&GOu^ecHER9RFyy9;lfB``NIW2_Y
zan{M|Fa8Yrw-%>b&XBS+l4Whs7xJnv!tPTVNO5&zj@9xvUp`D2uvlpt0S*HFSTsYd
zkk@=P`mUH4$h?f#JEs&L<eZ^=DCL4Tj58j7qWnDpWz`3^Loj|bk<|ty@V{lOqbItg
zy&cd33-sSISNI)!Sqq^f#WQAubzSYuC}J+}rq48_Q3LGXGcwbHJ@S59v02Tj=c-3Z
z{^_8AWN`g|;P26Pw6l}!LkF@?Px;7SGWulfdem_HS3(cVxH4LOwR{}p&OoqF3@uk4
zBcJ*p#J=9TWB?%O`SMwUfj6;-RVpSvx1q8XG$nlDud(U__!xL~2<rbfZUKa8u<e)W
zMOy#EOu@_Vs~~nn$vGKgAcXpYq-i(!Pn`E5!zYXxTu>li`$@S0c}CwD;u`A_nTKL)
z=`V=8@~z{6rEYOA^QM#U0uO`G6#V<NVT{3N(eY*SLp%ee#bqvC%xV=SPc%e@)stkG
zE^6^=tlx1rxZjVun8k|~beIMJ5lDXQ&$wa}1lG!~T_P`(^;s5xsn80vQa2FQm}Ivu
z;U%Y?o4z7#*1*%`Vlwh)WcM`NQmqEcT+}@RQa>5II<=6I?Ab+44MU)PPL;j7xS47k
zqKJ@pQX+f%hsKD>D7op_dfCT+F`ah7Q`Yrm5kk}VO#rM{RSj85@)LhH4~LSa?HI+q
zpH1CZKN>gm{_)57!Xi1q{y;IS>LNKX4ae*tr4VW5&os&bm|)ah>%ch*$5u+41Rf8a
zi98nY(>8tq_hPh~2F~b`+S#Z~Akyxt3}<QYZ>z><XC9-GK0Cyc7>)yMZUNbK^tThb
zo<fNh5dS>s)Wt&&#|@Fr{w38Ca!1%mq)Qhy7#3P2jvVBqV7VuZXQJA!G0b|{4vxvy
z`J~Z(a!49hqj~YoEct1A&pLO|%y?*e#_%?3X)EP0e>NW=Sv3-bymfI4A!^a0_ZE6P
z$PsA@WIteKn;;S#>6=wf-{?_Y)Lh%Edp9}SabdP^6JD*o$}zT61=)%!l=J8sOKs8N
z09H-arC3P!SRaR?m}OAYs{q`|aSqJ_HD=T0yH$>l(bRUpvMOxz+Q61X$~8`~eJDfw
zH94`1TXH5*d4}=jq`(Z0s0Fw-NVhb7OOqzTMx?vF&Hi0_k+St}k03(qV@7(~roxWb
zf5520(kqS^j6Lui0fOH4dY0vWGt$SWhb2X0q*?R6Zvttyig?SQFq~{i&cj*KFXlt9
z*%~>;U)Gd992H2ga%w=SwSi~~%*$zMq6Rw0KtmhkbQ|=r3y+sG{AUgTF$7o0N;$K|
zClEhjvs)rR>(X=y9qX!`DwDU$iag*EP|g_3&4?$#SUd}A*=ObpJypnm#JORaXtT-V
z9dLS{3i_H@O4DkN^|w7k#gP&j5LD$xG;K)%PKadD1odWUA`GtFDuetP!bQLZ=+#*%
zgAHYahEEtic8m=1H&`H;^;IZi4oz#_jwCNetfOdKJVW#kEQqv=$#8q-VRTH+DwYvl
znj*R?t(T%;#G*2G8?9YcHW?Y;v(64EkXth<PSe!7av5z4uVuK|7y~=UK6n`!8{nv<
zw+NRAQc1g*y%&M>ki#=_mVGfEovBdoV!Vx2fO-Z};7GJMNTQuiEJk5W@NpSF#A<x3
zr!g@dDf6r|0*9o*DtL7GO$sU>T8$ova<&g{x_XQKm2=WB^d_2s!rfjh=NhPt|3<aV
z^8!40-lHTSCtNA#$1kw%(;7xlv^0-gU?VVx7Oi(K?4k~k(L?2;SP)CQD%jiPVnan~
zd2O*=5{QHghj;2mxLyl`FBBmPhR3wXHU-Vd!LN#aQy8jV;;5mV1YZ_84Cs0}ZDb5^
zk&KUyHa@T)LybqP(8;z<Gax8wlx}H^p?V^3b&CGHn@lt*vx8Zlt&*%SUxU-|^HN-h
zb{S2vmug|2nCf#<&SrI*EeN|Q@6rS}90ET7U;dv|B;1>1gp(tczJ!(`x+V_fYzL9_
z#EYcLu#5|$MlN*>*-c_zR4teV8iIF9s%`OSqbyiH2&sl<_!<WQ&VSWbuLjqyO#p1u
zTD|LV-|=c$Dbwt}gUPA}Ekx7@G2Mm(77-*SGi>MJsX`6|=(Ns%g>e#`dZzuALpW(w
zom>`&0%rn5bGZ@Zupx4VFNR{l0zXG~%Uf;CD;+*M$Czo&R>=%p*d$j4I_Zo9t73rX
z)&9&Xyiy~aC)fB0TFn@w)Er@at-oyw61D)VzRsROQqJZoG$8~DriOgPMs1QC?B9*`
zM_ep725f3b<dL*k^(}H!jI6c6X))L&H~ZUk`?5U7QXBz`47<f~U~YYf6Vgr85xUhm
znTN8=Z9cl;Nki%6_P`_VYF_coeQ%K+Eq54E^gOw*+!_0Z{0=x^<t`r%a)Y*NqecAg
zW~AZvfL5uIAV`{!dtxUMLklJ8-UJH;N4sY3v#p1?e<uW@`{M&teunSq0Uw#@VrU#o
z&uL95`FRYbTZC#~V73n?*sR(JO_vY3-0xedT+!k#m>0zAI;`~ea6$;oJU$OM8a7$2
z?(z#88PQ<*x?&B0s*^|3Y>;S`q*lqJw&6&Pua#dK=8%B94*$PF9`m)}J_VMzBJgpr
zp}F!apS=ZFC+UGuDJ8CzdOs@+6<8*-61EYU1V+{zg*)5FF^Szf0>Pk1j=Bx2oa5uH
zfY7p4=K9-!1d@a4Vctj)ro1gO&(@NusmaQGUn}0l%}hHrip!<-Alg?Wy-27u2`)?!
zD%t7|sRSyEn#WPA5hNX1+{LX7RT?LR^8KKcB|#Mf(MVX922~72nr>E>*_F~*KM9F)
z%Y8^zA!iK`8WvXA*N7-&lk1hSxe#(0K&WLnV1|HL3+LQsS(U&gdC??8@th&Ts~usy
zC)UXtn-(HMRbJMnk5JDKyTINCcAc!VkF;uHUe^12l=BdU+#+;jgWZ(HNZ_u_kZo`D
zS5+7%oVd(P?b>AL@rtfjNP{nzmQfzvEf735#{!#+l3^y+Al7&+*<$F+mYAPns}I48
zk^#<a>bI?VV<1v8iC~UwxAoSdE=Q$29_!T;)N;K~IFjhMStCydEEt@|d^_mP-OYCx
zRdwx(VpEMF(+@^jB~5mF%6_HSU7ijwF$M#c7XYLthk`aM#+h3HCpF4vLhjHK@vu|<
zFV5aOyw2iiA0BGxp@a|!?hTvfZqsp*3%Fq!Oflt1I+ldh(2;D*bQ1ze=%IxY<SM!M
zCO2ZsIP@9<2_cjKp$7;x1it&8vJ&~de|@>G*Drd`^X$&f&dyGsM;$D>MH}R?v_2|A
zmhtj<;2Vd7b{?&QrU&kcm>QXsqy_vdPsTI6O&c-_;L1}zg5CyrkHPJnZ-qRaexd!B
z*Dia&ahzuiF7Ug1aHT&RV8iEDoQ2N(Tmn?R!#F!D<oTEyTg9jq)+~O(hSByCXtC;V
z@Vekqn2h?O?*tu2^xM;Y!K#StI@+{tuKYPp(XgR|<fVXP7_2;O`}T6|qiU8+cxSv(
zOAf<dd`3#DtckJ(2I9ZE{7$8aC1+sJ0JK(afECDSfzp0mICQTh1W~_KK8G)0_LaZ+
z%*Jc<OewdFp!QeOmm48|@Y?dgEctuDw1asWhNsv3ZK`$JP62`oH)^fCZl@xP-iTH3
zXuV-)tO~NpoBoY9L%}%gg>7g1AN~uiCRF~s1p{n*Z^b=XRf=%l4f1x-Hrtiwa*%i5
zi8rR<0)c1dpFyCwXEw>Z2~LZ{BUT9WJbBN)1DF>7|9-+xsuqCVB~h9x^MQ#=MN{QN
z9~1TM_4Z`wnVBp;a!jem5%bVk`*F}i{+My{uK+)9dWM#UoP6S+@Zy5OA3l^V@~Pph
z_rPBAS&(%>7I_>0_Lp67(Q0U)`+DbZo4eyW`6AvY-OHDD?hyqvMZRh^5iN*}!T37f
zu7Gh(zVSD#m_2uMll;e@F~y8*lM#a7wi;=TQrA}bZ{S&Pce^Rg+5TOtx7EZ8+<*Ur
zXL`Uz+W38?31BSTg3aBfja&b!vy<VN!40-+8#7#*u%R7tWw$nFq*UFIn+6t1-a*o)
zDz|$ZH>w~a3SrDOvPT;=385C6>=`q_@?&%U$VWq8FttUISDCtOX0yE<0`3@vCzSJ&
zz5N-yE_&%BPhIwjXAG6mqqao$ZR0_Odi>2)Xh&ebHg29BgBwovZ{w!9qmGsX(u#RF
z?D+lI7SJ1^ln2HFEJPc|{eyfGvtaO?(b~v3oP%T7GfU+Ve~T&`3s%g6U4gP0iqP5M
zMF_S$G_95CO)SrgzN|J#3Cl$u7O)wQ)^Nw4#WP$j)W_)?KD>?E0h;S;3y?5dD@UYp
zA@5^cdX;|e&mvrgHEWKHv2^?e5R<NzHvUZAm%uCS0c{O?WWji&U3`W6fqrWiHEACZ
zP?v&K(4o~%aQRCifZ0xT^!<-ObuD_+iXC>c^>@GW;>)G8kJ2K1(h$EGLSdKKmLdf7
zAS}2@y0%esDpWO^Ris-Rx1583=HWat$Ke01sdtnE<GE2#3h4sh1iKpu0dYy)0~xR;
znLQEK$yh2Ts>6{v+TLld2ggKVZEeVVOd6NXJY9cnb)6pT<C+}ISoKwMTr8jp1J%g!
zHcbxJ8?8jyGqA3LDR1S~(kngNfHOc^*d)E%xD7W`ROGce`uJ<O;ULZFztT502*(GG
z8qhC^AALroUmRx*(*Y0dpGX-p30~5CNlh>q-~+P=CeA8kh)psu_KYTT>^TD3@Gm%p
zoRFYpu}PZsSICLM4H!=hL_sTuP=2jaEuD-f`8VVStPzLHKu(TvU`A(nD9wJS81j&S
zH-+Y#Q-eI1^g+#DPD>LpEP?S%EH_P0BAHKvobH3to5`4#t+*%#wNabWnPX*eun%+r
zsF5M@*I+mvWT>6z3|eO9*&xlRW5DZ2OVA8+d|(AihzoFP$p{)`xKGKtP^i}nFt*Bw
zVAD`sD;UoOL0d+~UNOi7NV3+vE5Lfsu)P`r`-9fsl{0-z6F}^W80F81Ll&6Q<`MqP
zQX<UgIwrvbMM<e?iH!ZhZW>Rvq_=SiC=+y9)11FAm$TA1^Z{x+R~c{5G_w{c)Xc4X
zL+|u5nGhsbP+f^qP<3*4>>tYk6qTZv*g4J?F1-9onHVQ!2+Qruxovi`(-4{@qRR4l
zu`t*RSqFfXvCfa-k<FzUP#4%>s8~`glMITQ8FHb2YUv<6VH@S5m}*cNi-uhsXFpe4
zUJk3{B{3T|m!3OZjgpDkCL<k)T48|^NIAl^3;lhbcp6~YVzwyOT7ZJlTE;#bvrIrJ
zIR==V0Aoo91~+Yy;yC8hO10`TZu1h~5vs!s;8qb;ce^w|!Yxd>PA>J+U}WM(87hxh
z11%s^3@P1Dm9pSu7znK#)CO;nftr&m<$)mdokL<qMSwmOb=m=XrTvYL>Oci*epMOf
zQ930fm)SAOhuHy1i`DVvpt4L3k@mSbWT@?ul^O%6%PCT8yPZYb8`PJxk6@`-v;B75
zt<xOZR%RhEAL?84cg%#(ZMvffX9!$?Az9=C%drb}LFf={tZ@)q7trlFBM2Qz7QNC;
z{HGOWfm|NHXjm96!t`P+!F6~=&;_Ie8mVBAlPhEIA;hS049>c%0()gj$yX<2lRUiI
zG^7GIRLC{0hNZAaYu&Xm1a3W0<V+ue?8eJ4=&__I#j+vS86wAL<QMT_A&ezixjyCw
z>tlWgo&DB$ik^Y>@=K%QLHQHphFA&0D-|@jB5w50o$=_cLhAvsjO<tWmF*P5yvqvM
zSnStE2sFCNZw!*`bQHh>Bscl+<R&yWkyqYquatp3@X+33X9iVa7(~so5x4qYaNPvI
zC8PN^hs=+)V!su|%kBPU2u6$}HpY2}ExGr=-tyZxq=@oIiVHxv(-3*$@p4xnlGVd<
za(9BnqC;CErSGvdvuuLg>u(vNDdzGfqEX{_|AWaCdF!nEKL3*D4%V=>3M_1jb$=Xl
zWF2Zf{|6Giz)B2EA__zv3}^?VPf5A_KAoWDI7KVup_pCI0S&dxUS_XRYUvG9??dub
zk>PKGcsMI<*e1xj4Z!dQneF3Hla3ru@)l;`Y=lvMj{nY6Lq%c}D9P0TghDmR+yHZG
zrWl<_<^^|{jJSF5y7{&>Ou%Re_J3(+nlvV~TKXDVJK-F$>c@f@5Y>F(DqH9;syWxm
zo!%Ve-P$fe?5dH)_E9C1v|&zK61VeY^wHIdc_4(PF+Uq|Ela4NzNrQXSstkPLB2a{
zciFwWB4`Vhj!MysBqu9l$fLR+C#zx;)i#C5s%<M}by_B_DGY;%J1mnHr8PbotscyR
zr1aam%lcX$jg6;aX=C{iSr-7(4w`JD3`f*UMAipgA&a(#IoBHkVl=8nL?1SxDKG<W
zjU#Gyv5kT9V%%&w*%W_{tf+!iv)R``D+&v>A<xa%u*H9&<t3zND_*!YE`{y)B7FMc
zK+{P0NB<B%jc1p~Z2>Yv^;xERdptnx6NuM2X-<n?hFpbEsZ&y@?1(k8SPGi@)X5_;
z9lOQsOuNDca-MJN&*)2ty5NucG*p;ry4?sv9(`Eg&B%7URvxoKue|ao)U$m&4a#x>
z{G>^qFswnv(Z1zRCfqEA#lTJsPt;<0D)0$=D*Fwhma60PX`|@M$&inpiA|WGHC~>L
zRhmN@u~N^aVR-a$Z9=m!%j)w%S}NJo4jZOYbLYJf&?AhAPFh4|%b(KnP4;Bx1-z|(
zIJN_MF~k_kB@hgO=06*RmENx9CEM59Vw4lDk(UE}mW4t)SNl;fi6;4L5Fs`l1}xA4
zIK?Xtm02^XslZqlkYWw2xHBJ;_p5=4(Xd;fvKl%`_~2BFu9epuFLuH4ZX(;F7X7-9
zrb+`wV};l^(hN)^QU)q-#tf4)r8Om!krVKbcuNgoymXr6Eq{h_G!z4M!R76s?K3F!
zyc0MmhD{<X|BN5Uq27zU8;sI!Q+Y3j15Gma3FWKg{rCcnX>#&GS|(hk@?mU0P9^QH
z3_yaf3AWph0(|ti1%hx<e;g=L-w>5E+DRdH%D-Y(wWDfJ0{O(>M%)F?fXcp3pP#0;
z86t0d>@$C+IVT`MmSFmKf(<r%5E$$JFh#FPKKEZB*O4DmU;)xu?(~Zodk5q{tcPyj
zFZ~&$SzL?kCca+23Y?6|<Vxi0_-fl~++GZvTqNHFW^+}gY?kw%U?3m`o@THRtO_|k
zfR@$LzYTn$GBvB)7Ri6@tXH#gf_#^zU`124Y{~a&8I-<w{&~gna7$a&aYk$A%r0$J
zFP(Ih?Aq21KWK?mUQsH$wROYRIpbybwq^_?sk+k^D`6Pk9&OE_$G;dANQu|(3za5h
z&$ezm!8$B@k#{p=Y>ez}H-6Mst$dYL8QIG}Fk=`Iui3)T{4M-PnUU=6(|Up_NQ3o2
z=*3mXi0tD4TTLrvk32cp-hJD;O#_+FXgW{Y+s|gNnp!CPw{@GPawLMdMGo+1mM9E0
z5W@D4<2?osg3MUO<G>aq(N2L)gu$PxjvUl>CtFZdRv@ew2Cai_z7x<h`4At6rBo4a
zp<y3CX}fc8ygd}*FsMBV8^NJIa4Q60G0bB5X~KSsh|)Ud^RTvVsvJs&_p|gi<EymN
z4v^XirqY13yjBi(VB#h~P_rD-*3G)56_Zh>68@Q=#}mj-S}a743<Rl0r*3v>V>9DK
ztVINA+js*W5k^<7fwjGzKZDf4%>}iumiDnyx{!Fnl*KPb-gJlfg<J>lu2ubsno~y~
zhaLd<-WeAroeYJ|^(<YB+dw)S(2&=t>rv+JVgMrUstC-#YuY<Dyzu%mj5&~Q_Fg3|
zKRG$d4h2Gca?(8(K*80hQ+gy|Hh|oXxQ`Cd=^_HzQ*dQZ0u<p3RE}ZO18S<r#+YCe
zbl_|T7e6lEn#vFYss+c}2_B*uWIfX-?9piP^u3J2`t=+rz0=Qgz_iRv&?lb3c}kIP
z6Fhl+?UUnr93%bwXC-wW8_-67`*{GW>J0FoDS|_+!wtPs1_r`Q>Wa{79t~doCCXj6
z`cVsUmYnF%7QpI+D$o2^PHOd@%KPAsC;Kxs^<l5)0K8NDnWa8i@xfDL(j0BoP*_iM
zOrs%(-FF&50Ai8T{TEussDz;WKL#fG3`&!2HYEgPI5>b(BbjBv3`vJ)3&>3GS1gMO
z!PsS-{-L%fNE1NAq`ZK3+X_h2U}De(hNS_J`KTMg$VJBX*}XH|zoDi}KRJELBm5ce
zS|u@-?i=Y0(gL_h^|T?-K-!9B${GHpMY);T{!B-LN1UrBX5NNTM%EcLAC}Sn7MY(J
z{G;6Q7^8tc1NzF?#6=+=RhIK;jPn!U0Fi8yoMj_!zaFa?Z>T}2_Q(m6ADcF{G>giB
zIhhcUvOWUV4L7u$9e2@)%(P5tshkrZL1Bh<qev$DWRz@H%DM46Epi~f^L$9_$zbR|
z-vFvDM}+tV{(FmcZk<euXDX%A-?LOMv;`w6TQ0KyEUmswE=~xdM??{*^w21m7*xoH
zqaKs-4Yy-t`8M1xZa9mu7X;K`PzchL!h{HR(KM)w{24ccI^bN~_E|$GqK<KwOtvWz
zQ^suG;=uIyynHFK71tFaf~?eGCPy(fh!z>vBfd0#vDV^FXbg#+;=eGTh{Q$j8u7A1
z%4`~@L+7O2F1O}-REsHs)h_MKRs_OSI{dH=YZf<L@ls`9Z@*<*(_p#GUWdW86lT-v
zfWKGY{*v?g8*rKt>Hx-W5pSEM#x|oGZm(MVf<>b%WvU+yXE3l`dL>a@3pn#w1MSt*
z3?@vO1?xBUa(duiLw#8^YrWLPfyHHHRiYVM$zo>ojx3QGX>S<eV19|q6VBGD0hG$L
z2})P^XxK?m0Oc|Fc>=0c+h)s^KAQ0q#$v3Js{%Q6w93_S=1Ni9C0AO2ZN0``7zFn^
zyu#NefJ~u))lO#mw5_d}t&wqE?B01Wh9Gv&{w%JptgS-jsq1|?l*Cz~8P*o#HS)`}
zIpjQ4h&F&Ywt_rw@L%*M!Z#YZtyv5=I%$yb*2}N_FJ|>EmtXr=RD3Loo0c<L8GqAu
zCkwFgF4q=QcEoQAO352DRBnzN61o;l2vB5h35Ed!u=R>aE1~kJvOE8U_4_u*8s!#N
zbEE`+dwRwWP=slL+z~^TXNonv^S6OICY?`~J7crZ5Ug14@)2lqR6h$uiMwMj)G6hc
z6a-vq8K^WY_r#jU!O3*54Tr@ru={sD9Fw=mjv1bIpAQ!f08RXYYC=T@j%2w%zEO_G
z6qF6}fIo95CIZQ4%Y()zX3PLV{oa0u*9m2^xk_-;ha7d5IM#}gq(1PTDMGp~Md?{S
zKKT{UjKv|L*Prq3r7A@;-<+Uj<c?#1DCN!#{L;)q32mOiMaPBAxBo_Egq$=486mE9
zt~C1SywllW4gX&j_%ob7@+>Wty3lyW2?GYoqC{&f@vB9&NT-l87yE~FqtS-S;>t^G
zYLsow$kH?!!(v&{r$LtaXc`*<0GM>E|H|@U@Qm=l3Rk2zQ9TWtWX&lVIKR@gu!8Bb
z%Kocjx~vX*0-MXq8k-qzdCIFWQ^Q-hwuPmr@_@Rfu6r@VV7aWbCHL&xPu2%U;GBi{
zjCih%vLWbqOcr%Uq{&BXZQCXxxsgRS`tKSt$az$<X<>(({CCQW-aMO|4MEGq;LW~z
zBaO&<+kz`-ON<OIf-R&#3EVRQmKr<4@cc`Ya_mx{52BtW4+j~+b`MJ|m8(A(0(+b&
z+X8_|_w6Oy9RoBy(-_IEYqmG4QHuaF#g2G?2pZeTBMDrZ7+T-*J7ZFshH6p$s963O
zbOC&^U^B!+c{FzW2boD|)*z1sT@>Jap)bngag1OS%~hdQygXqvhuYhP@}$#WL1B?R
z6<?wtV=376(`j3oj7TosD9;!dK^`;s#<TtcRq|*t^PH`7dp%jRJRd_AK?gydi5C*G
zwZ<m&TdM7}CjBW01ot5p38rXiq!ec9Y<SVeMU_@2H~cw~1KONm1mqC|i!JW6mmD(Q
zgfwvSzI!<s-~<S1f3ZK*{VjhDWU492h{w3gE9sXO8zQ|-^W|@`pV>O(I=$-e@v35l
z2kKXxwMqV-$WJ@PXng}CLUNN;@|qo^n%V++J?%3SQ<*uoMBWGh$aL|{>VL7L;+wVt
zm}q9fKIfk@Y?a}D<cx2nPv@#mp(b-^C3EHNfW4>|m9O715`Z$BlYcscsFrvAYZv}i
zGNF{VPTp&~(<*e2&Kv)Oia+l=5Q=WXiqrQ&LJTctb}{hxmGWV%RwJ<~`Yn@>(xg<E
zjOm({?c>;R^@`z`fg*uM@KO2)=E=VdjibkpluvwMQ-45RQ2IXgXS^^&&iu?LQ^E$7
zg`IS*{M%_5GAY*kx&Ji}Cbs4Bg@2{4Qh;bx7wCTdGB{p7qekT`BYiaPHX-oqG@XWv
zX>bfLy$$k>-A(N6Zbc>j6N_ce1P(m=;u3P!gjfuk`POy}jmJvlzpc<{Y=5<U7ndE5
zDe@Gm<a>JwQVxQU=$gb4l6Gnpn~eY5rJdUBFdi=HUF|v4F4--fm!SgC?(NiY)LefS
zDsVD`Wsi6O*{-Gpsat!tb9+z;V!Px=?abI^Q6Sv;Yh|x^rUfo+XZQAJy!zpB06*Ix
z``E;Jqw~g0kbTn&bYLj!mi^j!kRS|4=&_C~9V*fO{$gmLysQssr*=koe~`y(E6W=-
zR1OS4P@tW`r;w-RpmrXUXKo2-QB+WAo;Wx`V1^?G9vl+5ITM}0f6~san`0qL<32q!
zL4v9T*fd0V$WP;gA!s>sSUa`Z71dRsxoe^P%>Hby2YbeT9BzPCV|$NCBO&AtBW;FE
z`Ey&-m6MK{BuBP03n?ujbaHG4&_?)_IOj;4KoqNlmR40k8p=uAv{FmfUO@F<O=Ru-
z7Yz#q7luZKUuX<$?-Ma-m%_7g6Lt_#e}#02<>loHrK7)W-NG$EuXVOvr$DI2zSxb{
z*{5z5#f+7(5>=M~mvXEgRIN^q+o)?$-0;eZ8PY9g1rv<x02>G&6I~sf<*0-U#@s`q
zq`|Sf;kw5~J*0=>6}dDea<oH8ix4juxL@R$_#TnMAUG>wb!;pJ;v5S<kT#ABT;bzj
zQ`RJw{;5Ve-WO!Nl^kj|<Va>#^h^tab5eTwd-eteMo=V}XQT8^vs$$c`i=X<vu4_c
zDKM^;zW&VXaqym$e*R2b!h!aY^eO!f3;j{ydVu|B_m0QNz&PPmh*Zqt!au=2)vQw3
zEi^dlM5BGQS}j7>V2Pa6PHp~!M##xNk$F(8_WCJ7CPVY`<y1p>A^s<)InE3xl!z!t
zkz0c^Ly|b%alt4pH}Zlft_R;8WH<w6RwL$auun=Gnzb5ODnsIU)>g65nhb5{R{D}G
zLI<%9*wpL<4jFa!(2-+?%P@x~6h>yDs-ouF`EZ|w=0k+%qYOloj0kKELR|+aiZZgD
zTeso-2aCeNJ|iZk#9qc2DmgR3+n`2i77H0=D`FiAoYv8{A$<o9lraI8M`ckdv$zFP
z_;MK=Ck257O91L7^@>ou8rN=T59qnnM{piA6c!6&(Qyh-%%tXlW{VfAm((C<Ia+CY
zfR)YvLiaq}>iM>D2$m|939X2fQAK8yfT$p6r(fv33Du6m+&R8zdJ~u~vKr(&(VtN}
zf=;Vh%;)+Dd`3-KYp-(ca-O{nRg!(6ti8wiMq99v0<RaO$$17<3eY>WK_)p8S&Y}4
zJoxWaP8Q0AMhYljue}*AiVcDoio?vhkaDp<GaIrB5OPVp=hw$7U@hZVW2)pjDQLG-
z^RJmoMO!3T1=hs>HZj6TU%9sw+L!r||8XZ5`8c%YZjfv&4LuncWru?fd@;Rpau7Nq
z+93W*u`i_-1)@RZB?&}?Wa@{FQtH$-pr2eCXY*2!>J)!hJ?eI)NSQx#U7ZV+as!Qh
z4AD-!!iInn$&kAA`f@nn?Y6IKx6_;QQ8h=txN+gjd`ZC(my#sa4pHsD@H~@LR1fFb
zmYfk`bN!i>Qsa9{BT{tNQOlT5TN_8Q2<A}r&`b^BAVl(tMhdn}Gw6Hw#Q$V^LI6*U
z3H^0(#9<Me%HZu8{*rB=7Rr~)?HNfErnCx#-xSLgX?;u8H?S0v49geIlPeAXMF<_b
zDn>`J3Su~Oa<%=|r`JHaCJjJ~7*!$~A8|j*wT6Z|uxObXBQSP^mD#UzywHcpTbns=
ztqpX)@L>@H5$<DzMO}*<;rbvYn8+C)`AcK#r*)7UY{wBsjzCGdF=4*Nc^d)<9p1l+
zanLdn?hE;KI(;p{G+Km77qv`&V{5)NGhJ>9<Z2+dIaF?L*%h#Nt3!E9Y%V*=APEmm
zC*n&8FeDmgs@|4RfiwVKDypKPVK>l%{yuWM;c*x|dU8jgg~|hdr+|bnz;}Kdi!aHb
zpxvE*qV<Xbzok{d7H{3XEE8O{6}ih{!k%dg1wE2?2XfF#4$<FoPuk_RfKH?B!66Xl
zDEB(_5XjVPX|bKEB|#}$CBHMckX}}SZND$zhq$pq*Gd5cQjy#rk|qRVi0lv8qTzqc
z$b<2dnY(|Va5$UvucrBjf<v$kV7<LSp%w1MoCXBX3iMQzf;MKyh`F*Xo4@*qCU_W!
zpBo>HU?S?gw8|9Cp{zOfXnwq1oh`{qgJBQRfUEI;%cU{sU^HrJm9l=z0s~-FDIB1v
z*}pKBgv*Y(9!#@c6jw5GB#UUTFO<dpnwdVe52-BiXO@>ug<+{I4QOggN~>jA8r-5w
zHe*+o$FpsylM5VTH)MtHO>0(VH^|DgYb)WzrLxVqaHi0%@^N@yavVAuR{Jy4sT)A)
zD`ZWK0KwBM=41ES<6><bI|dUNEd#TxSF09ES?6B{173(cMm1B&dY^|UhB3%YY}jBc
z#|TTHV>J2LJRw>}6~dC80^|7`19~XSu=l~Xwoo<&+7V7QRmCIO9Ly`1EuFzqXj}X{
zD>KT>q(<55bNyd;DD}OE4YEvi85)4gAA%zCYhem5#Rav^$5iVoFu4N${Oy4oHF8i|
zMT#yYA=YU_Mx`{zDJ?FmE37P&9q~SP1KFN_>RASfI9e=^#3>mE65Qz%5}8O3#5OII
zKl%VJm8esKXrb4bp^=aJH){F9Ga6DJ^JgmP0X!yXJ)WL1J(}o86rOJ+#R3X<j2ojl
z^*VVn@QeOusPvwbr|j@gEuC5^Pp2K7hajG3Vt2B*h>)Fw8|>L2_Q9o4rJu8znMzb#
z3%@Aq3q2n*LoPtjht~eH^}OJ4+8quOV!T~!4Kn)Jp|n_D^tbpsZ8R|P$e#n%mP_PD
zla^XjFZrAdHf2d37AFGMaIRmr#i2U~g70E3B&t-aB=px%<up(fN<mdgoxBnYaGjAM
zxhh1+-~1QrW&$x$4G~25tAW()bXGOa$=_qGuw$}NG578@e`a}pq=XIfy4`cV2lSFR
z47KCYUI~7-H|>Z11AEFp?1y4Yws<Rz!jN&(*WXUFZ9-_5iuLkNJcDvg+r}E0i~nhi
zu(qtM6mpBa8x)P|zeq@y_xuy&j>2DL=l=b8M$H0vU?qwV{26!-gCF$f`OpwugdUgj
zkpVm{TQCJ)?T;O6BuwflY}KheNB(6e1f66M;QWbwr2d?(@@aw|v5t&31Ec*c$PsE&
zWldcbl%jvfN6-eg%I8id&<s>aq-f!u2As%+@<r^dS`5v4Xr%dN97XUo=2TnMD;!t&
zZ-!!hWxG*2BO_l2Lco&AgmLENn>gyZ%MfShFTiZ5Iu?)ew{ZndLlvc*{I{LE8_Rh`
zQGcRe>pS}pt?=0JucC-qCf_^Aru(-lk}T5ROg;cs)37!0omK&0>lzqVcWG~SEL|WG
zP8&Oz=N#G9e=#0nAzFrp-P)V6P92n3`fP-h?A~6@c_<i>L0{rM{96woQK?x&sANxv
zN9{yp{IF7x5z=Kqk?BW1HqUifNcIXCvA<|g)MBuaz0*7!!4I`23;t!3pncl!WGhr$
z^-cJ{^|G(O$t)sD@Ma(<()aTjDQeNN+=Q?9PtO?tV8{A^G#7(nu;GvjY)wD*x5Jf7
z;y=)zQQ;;zLm!cY(tGp3^Pu8FBRaUf+V&8T0;f@(9O9Gkz|u5Nw#<{WR(|4$Ys3vg
zwnf!OW99HWyjpT-dpA=c<eBy==o>$6zjGgTSC^yDHMzJP*4}NmWwjOZvjigRQfh9A
z9G*bjh%>tYw0WHz5$}OjQUyk>D*3r>Hz0t!K#u&si42mnv9vb!c60r3wsvkC<256V
z5^H3yxpw}n0gRlHscZw*K2TE#6Y&)35R`+@7J}oWm?bC`7d`Gb)XtEO?ag?L10kJ!
zIK4dp1ao;i8$|%qwdk8wCtU(z<(YDn`eX(0uI<%ojcCH!LJex@=GzgcgXRj-&IUQk
zXMl+V0)Rz@cTby_%EjPq2v^d>-=?5LW&<;!9Bq@q%#H{LZq6}@WLcwTr5qbrgP2|<
z$F<+N8#eQljveLrSRL3bg^w-p()6^CM?;m<oU2{|xt<x+&Si-XbrSb(!6WV*LPVv7
zNjK_hmOeovV^9jZZ|p3uCIG$=$INKYT-p6v4HnVQ8be0%hH<mM4F*|xT7?X-9Xg{D
znG6F1`=|q#W6{}SIl-4jIg24d$UBo0(=)51NLBu%^lX`0MtIy#_Ge@z@P^UFD5s?N
z7UDfp)<!us(4tXlG)<rmht{w=EwEikANT1_-}IhukU{>HWz|r^1j=X-lEJO!V6rRo
zGKU1Ajj60=^Ms)>@{qBkM@ydV;|MqchWW=3)EP63Jf<}=JO+gHVpf+GGQyCkj#e4z
zFHrGjv^Ypl&hTesp@2g5jWgS;WwoSOMj6`C4Xsi}x9k`31hU}f!xIcLqC`B#*F7JB
z3bSRbj|&F6{dR~9w}Ul~i`5M0Wpq};r*<DhXcfHX%7DiQoJ{2hoJ?=ePk5ve+g+#B
znq`8GIzG#C3Uao;#{g<t8y70aV|2nfacICuN?>Y`i3vm=n#}}L&b61+SfpVi=lM%y
z#-!pUus5D>&rn4QF``JiKrWA5U>t`BC<l#nQhbK{3-S_d?n~rC+rGRpgXAK6TMf7{
zpIsd6ig|Rka!Gr4I^m8kgaKE>yEF05V1&%oNkPJgWeHiOEM&$>q3v0(o~S=s<a^S9
zJy|c=z(7@H5fUA0WO6(}I6eh%DK>IpC!abg@il-G>S2L>Q%Y?O-LE`GF14}Uur|f1
z!|oUwsOT9n8xpZ!mUbKM&XsdpRm{aw-rhx(AtkUwvz-E+eii;c?=fUlI^$-sh03^^
zF4&Glxoz_=l&V0s7FuRR>0-IepEc~5Nr9EkBBa`YAByvn3v3RB{i<AQ;!mcG=A_n8
z*jztqtW0%4$a%xQOR`!k(^_o-FCLn3C<8Rgbem`pEDd$()M%AX2KHbzsCNq-u)*>S
zn|T6~chJW`F1J4iqYpXx+7&*9(YxjsgZ5d}=}KRp1=<my1#(qdMJ<Cpua4t}7>P!?
z#)slTXF6z$ZgZ_|I;=6Mhc(kczq0#9a-Fe(Q!bES*uyb-<K_Aw3h-FA?uI%bzie;W
z5e$N$joI9%8tRG5!Jo(t?cE}a+Aq8`XP8CyMu*FY#i}D%eCDqb@oq$j^{?$O282K?
z|BZj8dM>8K^#a^8FXW~`833FEF$9fkfN(e+<mO<mHI*>j=j4_j%&fi@s3g~d#BL4Z
z0a{rEf+p|V{2A|XD(oO|xjjg+5T$HY47kJJGU=5FSuVegFI6PtHM3mqG>jG&mdafY
zM)<FZJrfS^_Fp_!h)x5XDZ1e9vCqfl=gYmZ>7yCYC%^l_qCDV-dZyeLM1AGua({rN
zw;1od`SO5&sWKx_N`IyapAXvGj9r{+iHztx8>|DtJ~gGVKRo1wP8r)`gL%x&m4WOR
zsSl<G0|}af(=Sygvm6ef6Y?DHCCH4k9S-Wi)QjiP!^Dk{IkD72hygj7>mNf<ge+tF
z#k^R)$}YAP?!tPR??f5_oeWub<0%9b)|zR8{@UQ*>p2ERnRdYdy2hA%0@CQpWPwjk
zRfvMrT9DpCe?}<;Jv4FFH_4)ShEr>Ho;9-Ab^^z8fh_Ug8!-;8k0jKkb|9PU2jP}l
z7Saxr0onYnMwSO5BP5};C?_jo=P-sUj1btUh^>{CX*o2SQ&eNLOA{oGRW>g>HKH&F
z93<fWs~r}z{xzUp>vgimV6A2I!?ivd<7&u47z_or2kySk=y3uJU+ev)IM1ofv1|x1
z5jC`qgba1xRDvesG#xLIjW$qbN{wtv&@iHhT)jy)r(32a57zQz87z$$yuc>2<8b)r
zfbO@(GuRMW>kH)t<zb^<B#*&b$6Eb=a2^N&E+^XpP~1AK-N@)b&1RGBF&3m1idf9)
zXpY?&4D&Sd_;&=Bu;XAJY6$?Az*ZhfTZzzm)Iz~jJ7Yk|*DHV%Hc0uS4*;X!j_W{!
z8-ORfR~}76YitDMJe9(6<2+_~>(jfBJZ}G4HKkad2tpbMXY`YC{&94kvj0X^*6H%J
zgTVWU$#q+h5BH4yj?SS2<yrsTs&qBrU~QG>;xLdSl*#iBmz!&F!!TNdu}UxahOtov
zyR+E!$9g$xPL@9f@*qY~B6%^m1OzLX%nIws?Ef4Ps>hS<Mi$CT@gYLyDNVi{h*Trz
zJWXKzORRX}c=>DG!@3`2&ohDUmO#E&{IX#~o(1x^Sl{3*d!)V^cq~I(uD_>EfHw#D
zVyYFT&86~Mz{L6;$e(%LVS;hV?bTc4jd%vH>@3J7?ELv=&_z*JYvdqW<R3AKmgRsU
z2{Bp5;ofpU!J9zbZ~Jh(`_}p!@A$W-43URvs5o}xpN5dq8JOYSK+e!J<h{7ZP!kge
z0)of;hJeakb~>W4<byOMgB>^|6=)<B$cGL9cpce3{#_v-*)zz;Os+%89t`+#tP!GR
zA^IDF`Tr|sMQjz^CSX9H7@lEKug=J)2?gt52B92BtHx3UOMez0jzq8nS|pXpzhg0H
zkCV>>!+?@8>D2MQ@Z-_9uY76$nV&yOzA{uc*Kdc0C|{>FF6P;$@52%LO}vMeik0#o
zAD_Kj8~lxL{aao|t^8@&p#S>Y+#ptqLG=OoE}$*Q$oD}r(D)z~iF7cdT5KMz{pbV$
z{d1QNW{#t<gKigk?~uFTn#Gx1DZBcY<jxwvM^lqZ3%mK35e31Hf?CG_zte2Idj~bZ
zk)z}6@6kc61*M3~+p~k(d|(a5Dg04~ooW`l9gK>q(q0|ZXflRjEPHp@xgSktl-9Ls
z7H*h*Y*e-Zg$-g~8!8_)f=XmRe?3H=)!3r_V+bw=pv;C*^niE|Oq&VCj5htTftR^7
zBL{YH1K0#IaXH8*tSClUu^jB*dhH*T*bi|;kbolrm{ts4X0)~ZB));+omPIt;>w{N
z+!B_nsjblFs6XwX>hV>_%VGXLx$`C*=2>8kKl5im1+AtK0_Mr#0Vt$bc<_!$uka#t
ztC#_gz8iFr9_f2&a=+R$01oTuw25y`AB8h#mbCR}%sOMf>3Uc&+W83lmZ)RtBQr*}
z_h0OSo(HLt))#oGrGxD#lFMqOqmN084Sz?txpcDM>ma^K=LCl}*I|fp7yF$xrKIZ*
zR$xp>S0mjLw5%wqW@kAnji#ni(DoeZo;HhqGKCQTa<eScut)mF>nHD+DMuTP4nYhe
z0x&A%7~85-yUVfhwr8y{G48lT_lyfV-YB%WzE{5i(lfRLrh*FT6*MrUvK;z8YHRe4
z;Xsp!1A!@Xmh_2l5S3afedC)!@P%Zc$4fu^-Ku9u|Co;ZJ)nc@z{r4uXBd4(m@YH0
z!%n79BXASWq{_)Pa>5Ugs!<?72D98w40zP-YA*Pb?4(T1WaZ=^?749WWbVMDaf<%}
z{6g4Z)jK*5X`ebZO~&#^zJqG%IV~1CsmG)qa=QJMDXt_Z9@N2Y7dd707H2%zp$nN1
zL1Qw+hqv$~$OR0v27BKiLle7dDP6$qR91pVIYP;cDGD(K9PS!kIqU~3BwysVFl~Lf
z4`(-sSF?qSfR1_Q^Nw)fFa<Q8Mp#M~$w*rh{+bIZ`V4==;t5PeJTs1TH9~%iWK^sW
zQG;9Y*(`WLMjP>s86{%^VFf5IFJt}PVDS(HWZVzHXD9~mdot&<j0j*E@s*Ab!m6z(
z%S>a@jC}vt+@7KIFqm<IecW6>9$AgNhR^o7xt&nVRd_!q;hS-4X7}d5EcXSUj!d*e
z1>x2LG0zQHu}aMZN0^Dpr{|^7t&<pAjhx@IS|~sZfbvapfj{F4(Vq{y-XxP8<#UjB
zvOzA4O#?nnPh&n9m>K@1igZyd*<I-rrYsZmVxxK#*pW+o8neAxcyY!D);s{97?L1>
zN^!hWVEc+vMw201NMS$<vS7<yh{#f8kme(B56M*71PN2%=E1;A!xn=NC#RKLL>aSQ
zi|t$Z=ooiZ5<H+HTP~$Oka>MAkk-(8X^;<V?8y|{4ur#?#R=}^GUK|L0?7E~J|%i>
zs8-4{rxiY~onTUjS!7m8Wh(%e<N9LZ)T;Dsl{)>_Lf*P8o<S9%WWjcj)#(r-E)u$o
zdggNh=m;1Eq{aa2bpnFiY8?n*CkaS~rKt(Lja2k`(O_HXU7HpQh36g;p-c~AMs+r1
zqs)~$L*FoVZkHJjWK?OT&VRW-LrEuGg8X-mToH?*FUBquxiX!AWEBJRfSUre2_J`C
zWkdmUYz+>9Tx|<ueLuOz{u>VOtXvyB0bU_U*fTr0TMggg^wvVaz0MFc8mbP6>lY4+
z?F_YqIdZ+dq0K<&$S(t-$i}RY8xjJhTU)Rj(*V5Q8$f#~D`)hdpDmVO8*Fg)*2r(t
z2GAx;1{2NbZi+d|ses)ae<2;TTyBYDsLDEJSk|*i9m3_-K&GmEZYYZpGQ|jqxZS|%
z+p~|{VcS()PW#Dk<4e#ZY@up)yEBf%7^FnsWq;JbzLb@_{cWm|Bo8{}?uijpZ>1+<
zrrhh#_=XnQh5Pn*!7q8Wvi^nK7xUy570LYpZcSxTr95B*uqx+N&6B}CKA2br3qH`-
zLVwqCm|8S5Q+{ttys+~z@{s-KbhH4fcTl*wvjH-r-)GsxxU(P;&kky#hbt#@{C&!+
z8a4=?I5%K}T8N;vmJw5$-2Ak-3hkv+Bn|!+4>}F!OwMSGXSmo|-HBX%L41kYWQa>C
zM}t@=5xYjHW;550k?)JrFIwpqt}14oZGp*aaiFpYGU1Y#rrB)W^x`dzY2b*gg4wP{
zmIdd)tQAF*5z3FFu-rd231I=az$#f`$gD1%AuH3!+5sHG`chdH+roqfFkUDJ_?Kn|
z+9+1Z>Huvjiu$eb0YJP^{OJLnEo%+CTCuH8*4b=5P+xAnog0K3mdgfTI!;Vjd@u{k
z`Zn1IJ^S>Rje#Gv@bYjm`I~HVQ}6)fx)&lM2%e42@iGNBpwVjA&W6msC9M({B7mfP
zwzUN@BtC4%e0ez76)a3F+(FwZt|s^+jdA|LXCT8{vK>j*xOn;Bi)EYP1Wl`9S=^qc
zTcol8Ya%oUQ6gXl-n1RTg5i%~u*f5Z8CWLkWM}N`)C|nn)DZt@2qhB-(LCB==YbkR
zK%tr4^HfmF$zw4s&Mg%rDs_(=XG6?%Wi7<(oIK%6Ft5uzB`90mlK~-vaI54g`!k38
zhtAih9R%`igqNkMAX5O<2sV^w<4lbhbWWZ;ml#~?2W^H80JZ@t&GRuHsv9hj7yK`T
zu0crrQ-YBO3hDsVTWR%rF;FuaUIC=Wp@f@5qGt~?u%YZr0Ug}2AX*rN=E%!NRC(aU
zf3cMnKpU$roi2Y3az<ZDc_s7()Qrg?$sHRCM)fxb$h1aFLA1U%ug0P;Wd)_b`@|$s
zMpre0P+zl&`wr+Yulvu9_<?ryxeyE92tvp^cU<0(3G${9LUpznwyGeCv-9LF|CT41
zWqepN=4}Vd&b9}(GIUAai8VpXl7ISZ$SuPW1>2<HrykyS(=QhB$|9&#8)nISaU&uO
zK;z%v|G`|A&H)|>)A0xX3(?Y`mlzdEI2PnX|3!r@POMVq3NWva3@J5m=zbiV6&^g?
zHvjTxwBo@vsP_I(oS--<*e?d7{l*GHufhrZG?qLVNq3*6LCKXEQm|P59rQg8j4vyn
zJ8-mr<Jy5{%vQ2chU5!Fd#?cyuf9xZ2Z3QeR2Y@7{29y`YC~c%g|9o9r;c$8jL4m%
zUIE6-aq&#%nv0RYaWW_KZ<7B6kXQxHwu~qL*5+oJB>8UuT%|>WzKhqvBuPC8AeZlL
zKy+WM1ieT{GZt|fVPe~*qne<~Sv|8(cJ1gkm%<V>JY?Pf-8!oAsTt*2O4ti%_l~YG
z!(__ZSIB+<*Vw~Gfl~w7k+NsUoeX=KT#g*kjyqWpyB0cBne65Bo1@Y?!0er1qUJ@T
z$sF0opSj^1y2={ax1*}$&Gkdd5by~j=6)T+US}zejCwZO+~0p`Db>FSoN*E8=75en
zcUR6hXqx=EqgnMx@6o*}I)Ad!z=7#^syOh5X;gkoHo-xG2nv~q^)qew;EpY#r$@aW
zVv}k!-k$^p)SyoNYpWb;xT~xIx^wc=^ahLdQs-)r!wiWzbVx%LHhZC&$tM(U8IpRq
zZ4%w`&0BGg9ub=c?vJ<R=RT+g3n?Wv%8?!2hF4ULxCUuse^yo-M{65|fMy)3eM&nY
z1NSD53_H#?N&Aj!kQ+Q)Iv9w|Nr4H=H-P329!2v-vXNycd#$;CEJBMq2b3(k1o@<k
ze+J7J?WAD*+);>n($#<Q5_*(E{l@L)zl74jC?B)79hKl^A19MiyC*zsRb_DlI9ZP%
zEHq~;p>lAvzhoCXTXQzbG5(Bt^BOFkAw$Q;dpMpjljE>2l;a#6G9l|c&FTkR!O@O)
za%w<&9ZshQkfLlJ!3E9t^xyxd7YR&&ji|OUM|v5R4AJ5y(z|1@6P1a<PN;<-k$@}+
z+rt3ZxzZ=LxEg1*TKdMH`5Lg(FW`sN#Qwg7vcEs$Ib^WUYM7-5#AlhJOKXc{pzV2w
zqvV8GlIkcH`m8x0n4TidiGd82kW}eZCcjut3U*OYJ7b2N?32>;5yc=*iAnQNtG-T7
zO>ogZY|QesbU=B4Ex?TuG~Fch8=vmK)AY|qUzR3}D*-qH2b4j!@m3^}X6(U1k2RI>
ztE0C<AP9ppq7(J;WM~lZNQm8ev1izAs^_$rgk=T$BEuX3t@qTdhQovCSV@+d5F>(b
zAQ&WPapoiA9fYT-h;@d;L&qfg<xx$LGvk~>mWRLtL2Oh&hyY|6olpxJ!nUjhrH%1t
zT7MRMybcF!tgl36FRTOwi+&)Djf<6FtC`Y>WE8wb`{!BlC2n*ohzw3MO2nza$m0V*
zjdY^oo}b2o9s{~VaR-dZtk?<8PHN@sKs~N;x_cpF%Q>;IUVTrHi3t}J8X93inJwq~
zGp4pP0*&=$&-2MBTr387@T_A>xP0aO*llD~v4rCVfo@oUOAwJGlZ;5Q8pOuPg+5T2
zv{v}ly<9G8MQ0{*kQ(Xx(<U!+v4O=Pjv9F2GAz7xiQNf9sFq|}E)v+Uv(;7L@;NDp
zt7_aiQs@g~@mkvb)&m<w2?6Za!<^|YlC?v_(@m*#a(cyZ1jC{f`!n_wWe@<2zfuyX
zvAh=bW=oy5w7wUkPA)Cgs-()LhP|P19<x`3O!09k0kMx3GttT%Wg1%u+%wwHm|1yj
z<rIW8R|EyQC7P`tY;q7fr2R@CdOkFH#0_it-Kw+<I+D#+beTyV{d@M3>O^9#d)dzh
z+HygTWTZ5#N=?8r4%Xz_1oCX0Ws(pb%?zWPY70g`<17O9s%4tZ*7pSDkvcFDfWvM*
zbv_VT5!1&}hlq>Xah=QvasgRc{cYY@nu$WsqFkPqt0pn6s&Pd?&EoGG9eJg{MHbEA
zI95ox%2raInF9E(wzG8F#d3{7N;k%}=_{)>!G=~d6ZDi|>2GS1>(YC(a5&K~6rAQ4
z2{p{oremDKJhP#$Prq0vNA|4-w}-y`%Q$0sNT5TF#~XsEN9M_mX%1`hKtI;6{8<F$
z!=@m=wpAgJ79zrL`~~tJHgQJ2f!q}2n_q)dd$UbhsbUP6`YmZA=sLGzU8XDE>c6P@
z)%Ti8@oc#*(2WQk4O5ZZ6R?`5P79^nVR#)dV1WEKK*&{P3gyoDys#1(x@38GrO#=0
zXEe|Xx!V9`o-ulA=j0w6u1~MNa&HWfLq`=DyHUT*c5`U0+!s@gM-M!?KZtkam?82&
zEXakA)f83Kp!=YG1eT7<@A7+_IUj8^i`cH}A%_mv7eboxFW7c{LNxaj7JP)d%<^Xt
zULb3mh%q}I-j$$mQ#4y!R^|i+z}5yomCW@qW<p5u|C4#adEwqfa&K1V8!E=3Ykfm3
z7x!C%G{#?uZ>W<6_7}QRF_36sg4psODFQ6=XSgY;Z-K$gmBleej`62Uj3S}MpgNoV
z0?nTNbe5&nS`RK3<yan4$*2s=ov*N6nK(*T#u0$-1vfmsX*iTk?0FhH1Z%vk4s1bW
zM4XxjuBzHg)1It}jfW8wf&omAYYhrSbd;pbsC5n*856~h6%ee~`?Cd*4d_7M2my72
zKeOtJb{{r{oKXbRzJ<DOj7NjYW|U;Lmi;DMCnD4kLr^W7eISZJJlZfu%N9cxwtOy_
z!&YCv#~I;<vI6f5<l$Hn%7|vrKm*+o5)zm)z|mL%sijr$HXjR%01uAHCFTchPYd6o
zA?qMSX>N60v6>Hq{U{&Oz_mH42|Hrr=mXBkBZ&tmYdio!Pfwf>R*c&juUD|fto$)B
zgeob>WaI_@sF5g=kma!;C<JR%%j0QvEHsTupioE<pFr1-`p=&5-}N%UvsA>-kLr`Y
zMn>sDp2C4$Cr<@9xoOO@e!AmMtukTSDiKH*9eZqY;f!TmrW}~PO`b_8V}TQkS9vyw
z)#O^uO0txf=UU#zU0}T9`8dJZaul1-$_xIEDhI$9a|Hh6BhvmzyA~7HR3MiZgZu}h
zB-@|U&N7w2YP#AaFBv-f_3SS%J7F;%Y_9wzmOcTFs5<#;40kEW@|84POYL}CQyS%O
zanD?GiM*QL(yHa)z4W+khE3-0u`-nXEU26=uld&4CBap!!kdW8!&>N8eBFSoMQ^D$
z(k{`BNzz#=Zzkkfu~Qaz`$vF}SOW-kZ}}^XBw~;*TII>x0Yz1&49+EaCq6-!mm<__
zGageoUEYl!kj0J&i}!*kM`tQfR0_;o-jA~eIz=I4s4yP{BD$R{AKK)&Qc3`go&Aqm
zU{;lzrC&bwXRKmw9(5F7^Jm83Em`D~^sU+LP|{95wZ+h!T*f-fpT%i#?JHDqq=_Oi
z0Bsa}4C<TYo0}1C^?ApgRzL*unUx??iRTN$etGG5`O^Ngu&PwP3SgncQV(EnhOd3c
zsZ^*d9rB8i**E@+NxyWYAZH1C(J223<RU;0m&vy_!whs#|8JtlxfuI9pASU|8Ho!6
zGJ2f09BH-cQfTyu1?=!Dkxn~VzMl#Jot3+EGQ$;m*Wt)ggu*%3fp+a=CMmlZ{@)gn
zyLB@Ao%xAKpi5=<PHxf~Hd^-Rw3EiRVW|IBBYVcns(mvKsZoB^NzHr3*_u49O;jU$
z8EjKat7}<_Xm1-KSAvu@*~h`5TLq25DTykdp-G2<<@@>!FzmnsL{2bU_VdNDnIF9(
zC?@8oXMdZ!xqfhUW#yEzx(YeKK!;T;D?bjTz}mV7Y%{6sa$qO7s3GHyrCwnTILKdC
z@e`LFEEkxH=j~vFzqx({)TToM13;A4N`Q3z6Mu#@SXa<i%AuXyv{;B**A&-&>f;-=
zv-mt+wkYW;hxzA32~AL7sPeN;YC%;(S|NuAxS-yO9E=M&Il^8*JJ9ON$?|jitFpWf
zc5^whlbcs3WVBmk8=C-$!iab*EYrxIwmvC2s^;s#I@vA&(a;PFz-aH!R9)7cS}Pq4
z$0%XiM>+;_hC{sR6dR?;Vc5=mtOvdr>71tHxn=`~rP9S#HL^nQPt`8FrUAk?z;s}C
zT<I3*!{vpjINYdal36211=R5AQG#Rz|L%dE3g(&iu)pvXqs5Q*_o;$2Vh!#dIVM)g
zKt)jEMyQC#c2YYedNUr^Nlnk0l1e!~mfNduAL*G!V&zTqZuE+gAkT6$SHUsTJ7{4_
zwhkRUi=>Z@q6J5!Z_w}%M4-U0Rwn&ok`AYwBK>1ChpMJ80}QOpWil{WKou18teoI8
zQ&Luw0oHJ0K&WY-j2LZ@lN^)8E7P1=(Z=@r<W6P)v{n<Ar()!4J}4`0dWub(DJy2U
zmYnKSaG<D}dRm&6mQD6$TPLU6Cge&os3s|cVuKI_las;zKGX!YdTxNCKg6biY7FxQ
zawq*2@&FnZ+0rO4@YH)iPZ<_3z*yk9Dl$CafMEhwA;@u?WrRO7(;>7yD4QeGK)rg)
z8FAQ9w4*G8CgeqOrfpI3PSnY$cnKZiQ3-<wd~_NZiQ|}z`66QyuGMsd`jyscW}b|-
zDaTGmF)JBoe+(Hva@+(t%g4bz07_@n+g2GL=*muqzLTGp#d^mo=E?+n138SPh3GFD
z?@wUw?sJ0v)T@B1{u6CIaOJ`-REjugIkyE+m}9tOBxra~<UE57YE<cDIX}(Jz&N^K
zSIY(QOszBxh+$nVlVZjqG>pL|AQw8@H&YH}8)ny$7x3RDa*^#e9Dm^4b#if#HbTjf
z@qI~P8%})JVKR0m`=Q=NLA=Z!Uo}$b@GyFvS?a8MQ{>B|i;yPt)%eA3@2tOVG;dKm
zll_^U0G^XZDNgULgVB>KUneCtF%BkTU~*DwY!PxwMlSXBa&=S^*{sa7^x%hGEmI8q
z!pX%_W}hL6s!qx~xtLN}ptZut6u}TG9Q?4ZZUy$tiBTDA9}c@yR;v62QwroAX~_bN
zBc|sAlBcGmTxM&5W33jwo(fA@qPp6ruuCYeKL&2)eC70eQmJBaY)!((65Jlt4c1F-
z@G;FnqGx)ltqgtLYGj&!L712`%{YPSt&pVQqToc@#sI%J;bN>yGqa0~Sy*SJXKO&H
z5cr@rUT!QBW)+Y!&~-&Xf~|lh1R}_lvHzNc1Dh0d8QcG>d_3cGj0IUCR|oZuhOEfb
zaZM|sg{GDTJ7xkzOYNhF<JvSlBe&?NAnRwm=gb(26{vv-xz2C`VPc5<!j2Yh@ac7O
zeG4+w<3deOolcmKdg?DbxsMb%$t@M|ZgB8}q?t~5qoJd`+?V((o2I#bY8Hjmd>ZcN
zI#39!N3`bdzcJ2%PnY8!y(umq6kDnVBfL4V4~A+t^AfqGlg9|d5M9E+5GFL=8V}L)
zF@v6u8UcNq5q+P2z2x?QX?W#Sxg+*#Q1ukZH;DE8t<8&bRiQa@ccx{cx(eV$jEvmn
zqxI<}cgHC}eGAaqJ^nUn4R%aeKsU*~zV6oAWS9ZSA$}KIsP$%u9bOpZ;rH3#5Gc_`
zHYfMn;Pgtv$zl%L2mE6rBAO`~ytEWz`hx*k6;g4KXBjI+6dxN`<>aB*4|ix<cA6R~
zsA))je1WtYXsg(RSw5LTuOcc)gl?1Bfz-Y|`^ubzgsp1Lo&_43YqG;dN69?@Ri!gX
znW$YM^W)MOO`)kFC>Pd8l_DGclX;MvtUStcSrGIw3|a`T^@V{&7*QEu%hjm0S{B7R
zawuL5J1Xj>`MYMUVU#)p>&S7I#MD}G3igksJ~eM}vNA4qnL&%xBM!FQ-|({s_JN2q
zBapEO3|bj*!uU#m_bS5_it#~kS{>h#>0osFl-I;D%~Z--|J~TB1#+zmT%pTsv8<2#
zsS5E<CE5&hLu~xSOdaNE@-afoS_50}#+YM5zHIW}X>?~)^fHL!n-dz~;nB)w1~szu
zph9l_0nO2-!<0}MN#x;xZ6f+G${&1W3u{4z6c})}+027+re%8&Aj0fm_yf_gC-Gc}
zFb#-)nk&sgRK0Q4?ub8n_3JN>#5{=4$jQ#M0{Z9e()^=8Q=H-0&PEjPqe;jl69!Jf
z#?4|y&pzVhwYtJ%{v}ON^napcp*)^uU<#>TNAiS^W*)TAaT)7N;$sdb3F=9s`-!9F
zDWi!g2qJtsog>YC2jV?XOrGiFv2K)JwA~#%xnOF~#)5~Gp_<2YLD5zv0$l6)MA4B&
zsLjo=l`q&<A|%)O<DY_Ja796>V<hE^K{22x7^42{`+~-FxXQIOiJICNKTq@6OKCDq
zDua<OjB79ZFTkBHiGdBI-6irDn*tHoOxu&c29kj)lpSH<n!J)eZotOiGYmnrE|k9+
z>iYHTC$9#mXQ%<~@Ak`}p+n_0LlJgyDJ(ay+h4eZ8PD^EPe)j2R%w77^rlb8$Uw?T
zjOsAj`G?Uc`s-xmEt_u;vUPwzdD~H=!9IX)QPghXr&z-~ad5_!RAws>ZTC+H5D}p7
z1_e(5j^A^Ls0A>Ln$FDk{h5|V1pTZ9ulXR(2;QT(qZ$UT2$o%eD?>gEq?uVn_1^iK
z%=J;ASc_S}t_NNC<9Hi<j2`i|swMO0{#RgQ+d8mVHUK3~mZKcYC+QoSLCx){aS8Hi
z{DRAhxWiTfLK>Vu^HoDP`L}}+3Km<qejd~e!(O3$kpOLgjYSJ+$(OOsu$ci{U->Hx
zTcUi!B>u01u`;|ClF~Ou*C1-Rq)T%pxUc>b7|%ya!?&?Drt_n;nS+m800gX^RF&Qq
z$#+gl2poW{{e7C8u6+x+l+J2-f<Fu$*w8<>T6XDd#w<lN-a<sJ?AlpPDUi6bWmU3U
zXEkl%h*ik$_TR|dUnhHXb^{s`R_KFdW}0Oj?P+gyxS&T5`H_t^Jgc2)_e!%-&enTO
z_O>^Go-JUs`}i9?L1gO;P~5k(8q^T3G;Emcm!?fs3-B7{z5V@DivUrPRSt+xN5B^M
zWB;^uUw7QK;5G;Pr(k}#`WrOl^Pu>29K!Vv?i@D3mMSQ=WaXoBNM|zy;G|L;1Z6RF
zip9ZN-!nfk^f%X^UBOzv(DDxTi7Ab;_vsvDTl_R8o(}u+Vetv%!BQlm=HzFF-JvYF
zcz9>Evvfb{C^^C*<VbXKv-%oT6B=C+m9SQR?lU3i5i`-!yGV}oZ(3KBvuH7XzgXJ%
zH}jiULq5a#hF!RAK&<&8FbUzbx9jYtyXtH~Em9Aqy}{UA4+*D(!Ppm0g^vC$%Pb*A
z3ioG|bP8evQ7x6uw)Pxzi=<0u7vNEQoN0%4r*yTk5vrDvZuT+j;lt^CRLc?>?hIzI
zLb?b3m}LN?Lr!|w6zB(mWagv&6T44Q86G_NnDmTofXJ2B$+7;7oR)4LI5*|EM2R#X
zfnXs38#_Hd{bG@KR2kGAD?MXxt8pLpGR#y#6Yd=tgb4*9;5q3NaKkOlHVAdn*9Rrb
zXO1G`+MV$GIdC*TFlb{VSXqC2#zYS5?~2=afWOy(fDVQsu@lTdf2JBKMx_=01cR%v
zAR{OG;iOSup`2vUtJ)$br&-yu2g?VeJ|z(p(ZuV4k@?KmrcAp|P7Mfn!f?)LxIN9D
zF-I>~g`$j%K|I~|xS$lKzd@bVHfxr0=&FMQRS0=kA#X?mgvl|iX|h&^+D8S5gj7*H
z&z5uoyuZV2NjX;JtdZgVF%B*+ka-a6H^_)s1;l}HjY7;rsF4=!kdgj7(;OS^8NRdx
zPSV^No!wSm1Ur{0Eoa(-N-IzcdQ?z+Nfyf)?b}KO(X4E3Ls{5sjICipWj+GnaV3rQ
zH(d$aHkYO>NS_O2TpEMIN9%OHvl4DdoeUP3E#ng*C1nFx+Xy2?eh^(Xg$U;CPKeDJ
z2{M{mTEt=qXZx6pW7~+?(4SGxaXM+(#?I&yV+QzI%`eBGFUpnYCUnC@glD+R<vf38
z5g;6ItDNuS)k8j5CKq&eg&CO^aHz2!SdmN$U}0orW=T#i^nFqN38-WQJ^l@GSuXO4
zOmo67Fl1hAgwen6K)J-BZb?0)(ngnKCO)L^7>5py3@PxpX$s&<Hscp7@mv^ht8Bw*
z7D-WH#Q6tJ33L;q52(M*0C&dW(cnBeMrfu;Nj;=Niv1ZRbr2m5ZEK_?J!8%kugpbK
z8sDlyNhxu*Tx#cb1TxzzWQxCF^&2P)E|oGP&q1|F+%69mH4JiKMQ6MBs5@|#Y=poY
z%^eC->FY8}9c3xrH!LAomA>S`u*&nQZ2_kU#*)ht7>t4kmcR<CW9nXgdrQuCrt2w_
zq$bvcwhU}fu?)9lZESz0i0(<m#80&k3SbqS76T1OW47sOUqOL@gjCBhmPwtRFhtd(
zUJ01N44(&FoK7~Zra>+@)F9<&x?B-(Kvt(ot_e58l{UuqTX56f0{QN$7?YJ5L6le9
z_oHA7k!w1ugQTQDt_|q%jwY_jOnYza+33V{oy~({2q*%R$srG-NPf}T-IuWKqlCou
z3Go`fxr4pzA^3oM|1vFtA&GjZ<OY96fj|ilfZiB@A`gw_q<$3>Y@fMp4U#8=i4|1V
zA^MzkG=CFs<c%7cKS*veh=+|IJ6dkGm0j3FZV4Pg&LFKIz)^1XXV&(FwKNyVZ86x8
zv12BTm)p}3uR+CpgxRo%!5xMrNCxW>VuOF>w@y`73QprKLHCOxoiswaUMP1uT6hU+
z7@cxGFlV_d<|~9#H79o)u}r#hlHB9G25te&aBo5%uP~meh4Q=jHXlNJR_;s7fe?iy
zHsFZeAJ{|`18wRLw2=RfnIvyY$MD2xP$~d@FfGKm6xpGhyP3}?zi+uDAPqcZsIEaw
zq0}b;c~wwlT_m#%{S{MXw*SiDVR)&ap*F~zI1Lwco+NW)U^vgpWL|)c;)9j&bIkY8
z)dNLw9)!|hgLmsFjfv>g*Nf|qQ9cZrT9BqiPzQj5<SPqfEcMcwfUqbqSqZaGRu=m!
z#&V5WE%9d-u5B*)rT&aj0@Qh^&M)iiHV0M&hZlqbA#D!J({Z0e<_R)noz9I=I#<ME
z5!}rbgO#yIxVvd2V@Gd@cGxd1sxGVIeY9181xZ%N-w;NNaU|CSr0~3h4%Aby)}#~2
zGM>q|&Q?4TJv^A`x!&jwEu3US?8F(hV9QPR18VchM%(SuA{2n%<S-iTXvie!#0*9h
z(KkB`^l7nJIewEZX`A?-x;;R9$jOkcHgaJln9;-jCH>UAJeU;whj^w&Bq-95X11jx
zWjY%zHO!e?CfkFFO=LSYHZN&T7%?@7jURRdBY`7#3N`1fJQ7<3px_<YaUJw|wcN|j
zKrya#ZqXl&_C{9AqX{(JFSNI9TPctEGe|@#PHx7*c|3rE<p_;XAU8eX=bH*<qdaMc
z0kRjudY<y1SXiS$o=#vPOG(v}D8%zjizWg&h(0C`{H(uk5n$#4vY^)IY}2a|Ne;r4
z=L2?kakClW3;qleAFLwQu=q~_0A(d?nbsF3amtHnDkec_;1vGdh<o(UJ`Up}U_*z5
zvQb`&XR4E^yeKb+*o;nGI4U{$i@k)HMr-SY$MV;BsTfxRm-vd$3kIfcKv*(-^Z({^
zSXeof5~$Iy29z3q35^G6v<RcWr=w=Ym*z5A7p2yy7;J*b{aOG6RS!|6=!9J)ucsMU
zVVSbSDtW`7u@3<)pD@qJn?5W1H$pCfzD5G2Uiyz9J><w`PzN4-|E)N5u(VXm+eR#T
zpj}+5T<|+-QVdOH0>0UQ8d2ghuBl^j+joPXPN*!9_tFRSKN*zo+cAT<RV^Rbf8&nL
z$%g??0px2`j>CK(8G+Xo$;b9$Wey_tzy1f2NUBwlWwm^grn8>D%**;T)-<vPTJvXa
zico=@H%~LfFR*InJ3`~bIM_t_cN#|pJ0>m3=l)CuIDE+%^e^H-PAGvcRU}^qWZ*#%
z{?$rjBYx#uYz3!mgDB3@8p^JKUk3@`#v%_#;2C~F^+esQRq{<hL^B{x-!(9H{wF<S
zim(<PkZ%q9?EHqXFb9Wvxs_?6MQ;O`;CKEh7$QJ0P2+oin+`5=9JQXSULsx84A!i@
zq_mlJvFy@CjVp+!kX^fYmZP#NDrtJBLx0$<i-jFAGj*Qq-o-6d6DliZ5Bq5tEDgAp
z_OuzmqIupK0=`Oq)W!80#7<X(-4(EY;$B_cn8GFu<sjL;4GxrulYJbHxB<*t&6xdt
zeR#zAsPn1`d%T|k&|JU$dQi#!0o(9#`Er2$kwda$k^I>HfWriC?E|~G@>)e3J4iGq
z2gNv(8R30!j5B7$C^^Id3jCAUH1J7&66@8Xn<Rk^a%c>+{a!ese;R+ysDyLjus|`i
zB2-s^KEV7rB<RY|d?fQ!(ImJ?4v)9V)0KeSH7iHhs7fr$<mZ7FPyws-tOv(DvP<YR
zmZHoeWEPxF{S76?{2SW9I4YBX+a|2Qp_oOYKu+4F1u{aFav19FJ6H@jrbs#$q(|4#
zR_V}1tr*SqXP{{zBF;M6-`j5mL%jv4>Eu9Bkz~|?)&vLYY+1{svyV(fq8}I37sOOj
z>X$BE)IJNJc$IVw+JORJNA}XqHVZ+lK=D!btJ#Bahe>y%tbBA7>5;}Kuc1)K&~fkv
zIoj5#M+1=8(iU}0pufdmsn^kbIo5xnl4rq9yhEX*$Z<jJh^f(>A3|`vBN0|h2(y&7
z7fDZl#<KG)ElU}+SK2eWnbqS3@MvE(n?~<IFK#!)j!K^((q8@gN?%*%DIJfMez9o~
zdRR9e+a~?fvvoK+W`7va#kH-;)mfa7f$0_Ngbry-PVi?mrSei_jlvTh%+~!_E$l%H
zG0RC^TxP|A$KgtC`A!bp&@+Y*DzuL+mQ!L&z;qZu!zNN{W|C85wG2%Jhd9kh0GS;q
zzn?&5trJuK2E}SJ<%leh!GRn^JIWA$pC_1tGE?V=2DM|WkdI4GYMvh}%HwlnSd7rS
ze@_{n799=zXbI7>dm!V5GNOw*9h&PAwr{z;BLhaYr*Ozsa)wWUG$-8LXZjn~e#$(^
zqq?Z}d}>D-9n%jPG)Bgx!FXd^e#O{$1`&!mn$)Pq1-m)36h+9+GL$vfpEFv<2Y@xW
z;t-yemHgO`aoKA1yQnS91{kjOo8|1-9f%0gN8N<KbNm;(fZ=0;;SgX!KAISC_d-C&
zxd}!Kb%HxX&P&f&N75<~pP!yt3tFb(Utj~F3RF{tOftM2f3#ef4xg3HY=mRR>~0su
z65(<}@D&5bj3%N5;w3(dMNGpPi9?r(XD}jBZeTz}L7Kv?YdC^xi<ZK`?+^s;NRj<j
zXli9P)&P!1Ql?TpIi9Id+^7XciUaS6A_A{L$~)Aq5~nY6ZzkC>?H1=&*(4)nOZ|6x
zP#F-R9q(NNWi^sZ1F+HXA;=V;){^iUkiACAd<3dBaOGj~2zJxaC<HB$ayxJ1ASi1z
zYQ_KTFfL#~!~n}m+Zg0?fw@)rRMwe=rV!NSZ?%{omf)PH&Yx=khItnas9G}z;(X4g
zFD*m|FxFhCiOsBH?O)bFtBobH@3c$}+CH;Zrule89xX)6Af-&V-z$-UAazDOB~xUE
zzZqd#MBU}-85j?~UI}Z$75<EtLukFYgHVm%pHXt;<}jc2Dn}Bvdb&TBz|6;(x=nI*
zjKDkr&=pI`UlY%$*x|i3;IP;F2=i1Vz`s-)R9u^BZ^OGmcLA$+SIc#V1NI#Q4u0X|
znKO|8Zj|f&84W#9xp1+g!kFP!-Bx6}H`oR;MW8}%jO%nHGJ@$X{Z(K-S5jRI|7nr@
zI?Y2VK;<&|jg8S<KW6X{xyfIE_Ng}<ttB@br~rY~cZ>hUtADNBYMV0|iD7bEd<#7X
z*~Q?gE9CZoiM`wyA9qK3X_FrJwKS92jn}95K)KVA7zTpW4OH%mk>G2qtgbGRyW?6Z
zKnws1mE4oowgieAMWn5AZ_LJEJqSlM<^HbA&eM_b5SfJkfB(PS=VMcVr|F$_A?}Z>
z3r*YLFMS|LLd`S`KBbt2a1?nkZ4EoERbXp=Z<uVZA2LQBN<&jDf&2jCu1{m(&cpIF
zF$rS)5}9R)*nT^#-M1s`a&~Menup39|4fAyK$+TS$=n1h>wU<)guXc%-AMn({Pc{m
z5M1bTX-Ke{#gU1hjR70(Muz0l=Aoo43v5+`s*%aHF!8MxM^)<*ri*+O8h>C<z-`Mg
zclyc~`<#f@U;+&?+Be7&8yU6x%4DfwU?@c0a^_htGlCk0NtdTp(w{0T(u%CtF(Pec
zrHw?TMpnh|dVJ@pfoiosgC2%R9r}S+${ITbYR5wanf=vVUxdnnvd-V9t-!AMIkG<1
zjuG9xh&@d0IAw!>V}`SNG;&E(fDZW+IrbX^J>W`+k%X|q<cme%xSQf=XeAXCoZ1`|
zkjq{sThfx&V`{x{WNT~~j&yiFDW$KLhtn+313-E$Nr-=lU*NK5k|7T{#jxeFEv94f
zp6!7P)OjqD=9pYnQOdwgvLjv}pP43)BxE(I_P{WvopG;X;f$YGpXeX$MHTD|k<mz+
zeKdj8QuNlOn#U5pt?E9LAlwQMIemdV9x#l9cJoBbEk|P=Ee<>>IC@;)5_!_6v5XQk
z078O>_u{GGL6xxC;I4Z*7N1`U@Scgka4hG`vo<9}Y1kB>3%t;)ri{2ro{zy0mW(n=
z@`8h9iD%U8=y?L0!TVdGkN?TfJ^Ds&p9y#MiwTuhEp{`^&VP=1podxmtl9EX%X>Q*
z{lRU2*_X)JOZpL7Yz==gu#0L7aj^bs2aAPxUJ0bA^-t}gAWHf!{$|Sti6GryUX77m
z`QlF4ss6jahaP!={xw?-x?9xLvZm4N|2w{t<3_;xhCzftOdQxZ9Z3eV>TZ~rQN*g1
z%0B`D`pn$FZ*{To7Y#Jk6G82fhf&pMXsXviB6!<a^w6x8Y{Rj7#~8)1T1brlwEu=_
zV70s(#0Ggqlk=cQt2G*kM4Z07XD}e77$&wFG)lmY|9*mtax$HFOXPz<2)1{bd>G__
zQ&5pBl#haBd-d%lAN%Jm<pSr*zv9@i(G@gkSe$b5N$`f@a5pf~=F=EvTy24T7Ju~W
z*GK;CV^a9CB#L?Rc><GNB8*nPu%%yk`IX1Xm;SpNl(sjL|I1gwx5lv+o_y`E;8Y>}
zlbYp5`NoDo9on4yCr(Kbb=PlWLRGflJ_h^$Z%l$cgtP^6nJeGf2<S{yA>XGNG&qi0
zlt@=Kkzoh;%yK+;@n`eEUa>>U$an1;p-_wnV>kx$cay0X)WM5zx2|ejLz7$Cy{j45
zOkZL|1I&LoFRNvbt{#XsdhGZ-*|Vz~*M<VmKT0!Oa1EnR_UfuidsznNRV;jOAJ-y?
zX+45=4-3z1*{7?U*chIHFcR4})&_Zo0fwtF`F>r^Oaunu&R|}|2&7gX)+qb?HyR?K
z)Y&Kp#5j<SEE{}-{MesyVP;-vk^|#C*gB~!H$Y}P$e&RUU@;X$p~=Da4DEj@>4QPT
z<atPZ3!*|kD0u5e`H4TXIgH018sq6jVTAV6u4*gFAkIb(Gvp$y9?fm!XMqw_^QM|1
zhx_uO4MBgUbfjg#pq3J}IdX&#?-v??nIk_>V64F<NQU1eNBT3~&2&&P6}Qd*?0Ab?
zv~5?nKH`8P=br*byELe|0Tux-?bGDeA&*JS9gL#-_8%x6?LS9Agzsc4rbj_K`&V#Q
zq(J8`4hp@E6tSSD`~G$fP$uKvMP&+r(#>(ghzzqP(iRo}y+n>m6lY}s7}DQ>L$O}E
z$J7ut=;)=4*CS94gCg{poE+_+o5*4WYnSga_BqPns+&`<>SF^w)T*k{ZoP7x&qAg<
z*J=Iuc<mf4qOz7=x}Na}qrs8M+q0KptQ_GIz2lhWYBSO&!Oh}_8Yd`yySgcp=|f16
z>1XdW*CYMii=<0`|Bh~eHGl;F%7FL|Wi`-&Oa?kOans=@nva{#YN11-XVkEqkf?+;
z7&Nrhna7C+Z4GRiCnYjfi;3%>QTbqtoZNyE8V@u-<&;*>$d((R2cBxHE@hyhoaQeu
zb<!%?E|=5oxKN4)M;;V=Hw5};EiQq<fwf-!ddrXis#O4)S+6a}X{f=UDJzh?AQ=^p
z=i_*=P7#H=VO`y9hFyb>3;GzI-sM{LCd9ia1|lPTLY#cn)Sy=K=7lh#*`G4f7t{zZ
zFJ)YE;4}Oga?}{KgP@EiInyWu9=Jl7xuNu;vVVFFryi5iLOVJDN!EAR;K2dr%9y~M
zR$+tf3+9HgvG3Iy^eW@h*J^*qwbZ~Phb5mC{JseFL1et4lx31@B;R<u_UDD8a)L82
zSilxk$k{$C^(jPw<L*F2ntg!S9ynEUGBNfiH>Fh0_4gTb#R@S?!0yirD2uc3y41+|
zX-xPGKsYREa6y8c1V+?r9k3nO$)tc>wWkYX>qeDM$xbT;>$xb+xD*!*OF1o+i;W5{
zJZ6$y;se`liJevJA9pEZaM!{&R3Qa1%T)Mgq%i)<L#~+=1x2#{2}%n{)~9T#g1lKK
z`!kvvn61T7$zlVePoLgWVk0ut7BH6v%7)k0$)$##LH<9P($!;g;lY@{{T8spTOpd3
z#lllXvT_>|NS=eJmkOV&e_yF|P&BboxoCq_rDq$Ux{!J{%4M-G#&RNrLaL3UW6mv-
z^AGe41Ff6ztR^<T3g-s3!~#pbPC$&`RR4ft5V;Lan$rxZ3og7|rY8U_CSn~-(RD!>
zldC~(^JIph<(xsIQ1AEh0Km8x?1zix3InjYepsGd88}2s7P-n_rZ~vhc3MHNju)m@
z*2y*T$GAb`<=Qka<6Kcr-@q&^uafKHr-Ir#CdvIGa5NDi7QCpgHxQ2N*!?8=WnxCH
z>Iy^Qr_#g2d_x+n5+3QS+!#2pq)o7}dGf3DY#9y=gec_b|2p9rO+C<&(zmv6f4Rx2
znYIzVQ*|!d6K}Q!0vQz<xy8O14=4JqK7Ir?P~N;PjY8p>24AK%@l4$AzfkvpcNq#F
zV~No$Mec~34X#U2G0XJ))?c=~J@Of7ZMbn7<W2*o0R5S!$X)&#eReE1$3t(k-5msk
zD;o8|{5|51P_fDgulPlBudjin8sJ=pOuayU7aN0O#+s&mUxEe;0sXNw3WR68dVf&O
zghBE^SM_upbDTUF$4pfgWa8nru9M%#uxhr%h4ql{@efK|LUn+#WE=B_>unABY8sPS
zHr1dBLq^DKhgu!dSP!d8&xsZ1Dp6s4ZtU}9B&_6QUi^t`sklVu$GfnX)<}c>g>fM4
zNn`9P_J%hM*t0C~XT&hIG*(ky7+fQtiYjAM78!)K@bSyySR1{Fpjl?K5OMH=>NYNM
zlyXEkupsSH%p?UZ4QkNJLRzeOSzr!DdN*!c*in|p`qe>f2|_D;6>vL2Yh0P$piIx|
z7n@~OY$JA=C0~(X0h(c`@M<IKe#j?X6ML7dg`@;m8Z`bISsOqNC!^MgwRLd<$7H7>
zZ%Ed6^{{}+C<lX$S8;Dcy5hAguL|`>)2iP{Q#$5Wh*=9jwhH*zm=I3`75#_{WK&F>
zDS{@HmCdmcdD^575tv)-4OS{bXJX_~#uw=115da-Y?L~qO8yXpkSl|wWn27(#J_6U
z?sGFUi&2?S*`zt1sTUVbRp4;ik)F-bJ9i`4%p?9xGgh$i=sO`h{TXF%x*O<a{G&f(
zsx!RC^e4-s0qxLi0oA<6jI0I^%FmO><F|<@2?cfNi8z1xgXPKi4UtIM@&b7(@ZxS;
zn5BXMz(`qi2VElY?-uAb&o~^$Y1FhO&-&ibRLIa5wk<GOg+#kVp7U=g_^GuQ&z?67
z<}-}ASYC+v!ZrknndCoD{*>lp{apn6!AJ078gCiclhw@mv%_Fz8%)vBTs*LICW5@=
z-{_?PoX&x<|K-?4oR+onmpD5hNw_{x2Fz%E1jO)H(jkmEoXD;Go4r(vsv55bTEWLE
zQG)63aov>`mC9=^>wsti`UH)<?xWK5gP3=$m#rpI^GDtYE`%LIde)l(=$UYk$v^xv
zR<F?xx@uizK-gRH*{BRM(dBJ}t+2RM-ia+PQWXn)=bsJ`6<k`gkisbM24bN{u-Fj;
z`QA$h5BG)L8PLjS17P!uzHhK;W_C_KNN+Q6h1>7-vi!FBFxUe+M}lN|<~~Y0!&0QA
z`IYi<0?z9F@>=|toyDU&%O?p;?gp9cdigXa#`&Nz&1&<}^=#ht&tl>tc%IDh`tLXr
z9gmaGV@Sq1ka&U3F9I;wOGvh?u>WP6#+#(j>4i>_(7kXU&6BVE8|_1@c-kso$JjU+
z3@Bz2+&4asRTKxFl(^|bRHi5YiJcsm0dtjagS5aeOVP&rzkYfuHLm}=AhfY#^5pw~
z8ut+iZ7Wn$>86IJsg-55i1ym0TUffN9+>ZN*KTH{Qhgj>wG<xS&EF;)ih3ctcMIE5
z(!uZzwKii2dvsIlSD#)3WY09&5(pF~a{S1jfz~wwiMr8V-8>>CGmTN;dv{Y~7zKbD
z*{7SUimbO<&bA5rc5@5Y&};<+k14WWKnYoeoq@K3a`#VDLk=`Uh8*DQ#BSo$Sgx7b
z`hM(dGMhd9zG2p8INcUGu-nc<;RQVQ|N5UC<R4Q@M5!MLPpIXk8HAd3E(ZrFxYsB$
zLOPd23`wwU$WLN!WuSD20n?->tc0e5Lk&4NF%`_I`f1FD7Z-uZ82m`9748an^*703
z>3913A*zupLl<Op0gBPd&jMMxWfp;Sc;Hye%uq-<B49><7nN$}GW|S#ITsoX?H;S-
zNLxqq{TE1^Zf-hZm@6!`ZEc$nfQ86S`q|obQv)4>atoz>8XqnpK!z+s>0pEktu|XR
zLpu6qk&S5CNyxTFV4Z?mknMou(Aho%%Gcw1=#mCr3-OaCi8azSPy*+Lrfgx#$dhFw
zpKi8@{>b7ys+-&BAkr2?kcai7dr*yjZ5`ZrW*15i|And+Lj%onu@#rg(f$h=8hIdp
z=R&s1G5!lphJ2=_N$Eq9V-1S4aYr9#sALx$RIAO%@wVc#E2dOb!kN<35IJld#E)LF
zDcD?A+eQ+(^!D|d=Ezd?^xiYX$^w@A1foaKQ`pz$&7oa=zu3ZD9c)E4(%%PXsKyd)
zTee6B*yN}(UM2(mclA<pF|Y=n5KGtFTx}mf4ckSXnD&k7;c}7<G-!BUzMO3TJsJy^
zQv$0L0gNM(|AKT+jkUtYX=NZ*%W1w=2wpp8VwH^WBc@rfidxy}wj|cNqMRp#Vhp&r
zc;Pb1e{cYX0tN^&9uionHV!+~0kSDHO%ebx-agMKZ-AskVl%@7loBu;Jdt7krN@!0
zT+0(VONKibluw|m<FZ{TBb<<VTG6JmWfM^{S%rQqsih9#8O9`$yH$XaWVVN+k@Yg_
z2Xk7DSF=@(_Ft5Eu>#hGWjxSY8RO7Uv7_;AE7;ChpTAYxE&YWdk8DCa&POAwf`t<y
z&q$Po47^&-^3lk?&<h?j%#!g5?JOKiS$UJ>`v@kLVY`$QGy!iDS|G(Uv(jrha5+2h
z1{sR=N#G8Xa{?1NoQH}2scsM!Nz<+o=DEHgD;wiTz?L9#2=@j`o2AV}svo{&pDtz&
zGy8?1Sm(#7LHu)>T;TIqaHl1v+XXktL9~QIc57TJ7si!^hHQvZE0T+h;L)+=8o4;u
zM7b1<i5}-mf+LSY<?f0cw2@5g2MAX)vLWd(-;c%?)=w<(QPRSYc~BUH4ZjzLDDqcm
z2jI!t1noI%lyfQE36uS2?hXU==wvL85vG-t=B7wV0HLG|Nr`6;9)7cgG)ie4ohr12
z0dKi94apcQN>=k_ip|*jgx*pXTfzh#3^WVsZG)X9?IBej!=aEi?X?v?Tm#J7)JfJM
zkg760gPIRqFjTe5H^;6~YlUQ+b6N2BiYx+*3OQ(XEO7|r>q^PRR<W8iu~rl0)vMn?
zsf~SMpl^0MZpx{)lXRELv;+%F=2K`~FVp=Qx?td>gCF8-)!C4_GV<RUjs&YOM@phP
zfbJ%`L-XbG00>$B=v!JOR~P~-Yp2VVi3;7#q<0GMrmKQlN3$d2)d7TBfYcHz*Tfs!
zZ^Hq-wguH7x17whNvffrUFRR0R%cq+F9K=sN1}b_^$v$|Uiznb&*E;AUk3IjlorSh
z2|tUNkdDJOOK$W}jYlF10I|z7gv^y+`7iSk(njsi+>Q+x^w&NPWf1C$^T5M?V-q1D
z02^^r%sOOnzTBJ^&tPU&Dp@DD*fTgEv$R~m`<Rnk0|gm&w<wg`VvUngMRAJU?!%CQ
z(@r=K7r-4c3_@pWVcnrh^IOAQpWXxH&TeY!I`2HWD`3#Z@|47y<Zhd<OLuMHH%IR2
zwzDlm&CEr{X}~>QCk7(}z96j1y*7O3V>`?50>JzXgXQk)=4NYjZd3b<+E(uOTZP%?
z8{~n6(^eHmDA>q@akSOD(Mo=GGW@--fN>g3C`5-<c_>h#H81Rh)%z;UCq!cw{a-9V
z2r&qx-2gb6WY!N(KA5ehKDIED*$$^23gUtx8q@^mBsjJI5x5~kDre(&E7UqS$hsVH
zjxsOy6p55<HafrCPHSA)s+becR6|gqdW7u|lI|L$(av{a2Da%1{+5wBuNCeJC4q&u
z6&aLIq#t{cBcatLi-l`_PF{xql}#WP2TGw`G3|X85VOReS@@C3b4v|`BJ8p(3(6ip
zXtXT%mq8cMIn1@TB37ISfd}5CmG(g$Y?xVD6%<np!*G$TPSn8jsQLkKTb#W$@!|+{
z?#jtpe^C_yOvEDn6jK(<x^8aIW1SN8Ovb^<$@+i`awF|=P!wfDEEZ)bSW}D=Pm{q`
z4b3Je8)KvjG$YbHL?CjLy}@=l(`0ijjdmkAcDMMD#1h+Mtp)4f8Vo**QiJB%emEY%
zp2EII*zP|B8)qTL9Fv2$C9<i7Y`Q)Es+@+mn+^Y{;xS%!#Ft+2HYbk+e;$k)?-^KL
zcKV5HKm;IccY!(x87w8zqKQ4~@6Ct52Kk<b@rl4UWCVHKj(GvP<vwB4;|kBplTQ7u
zf|iv3ararc@F{y2#h81`(}5mJTi69U#GdhI7EP%A59C=xYe^~cCm1L9Tzm&EOfwEO
z^ylLl{vucU|5*FZ@H(qw-J!RHkkCRxNvQ6Y+)bAowsFHUm=0UgwzO?YTWCwRWx6S$
zcR~kOx%b`^TaMBz1VZm2p(GGWAR*lMow6k7ocrfK_uMDG=-c17%B)$lX5J|aS5L~*
zwl>J(2gROAFKp4yc+31B{n>o=`PoK!)+b;IjZhKc$JMIKZrnYW+EEaPF;em;f2L8o
zc$-S!^XWapJcM0W$P2M*1IaBf+VD_f7;@mHBsEY*09%o;mohN^<wfx?8}(_rG7J6{
z`xbyb!t`DZ`XfO0wf|a8f(ww%u!LXt9u#o|g&=M17w9Rh3igJJ1WSLST+{*{Nxy}D
z0O|f_>^pb_!Fg|`S+0ZBurdH|+o5vineu1<m7mhrPw?;^8+jv&-`2}tVt<1?EK`rU
z(&T{#@>gRoT$&^AMjs5G#n1B)G4P&0)3RR9I`5}vyy1!>aIt*g&v0>)dW?9i{ms}3
z`I<oY0ZjirwZDbQA|U-D`7ovlM@xB?eB`f?*CGu5AF-`a{tmk!D<Atyb|etwYnD%9
zS=ILD(G;nVh{lJb{L>g8)C&3papjzR7UO{`hJy@l^>y;k5D}>;g?#Q!b4;4%MU4f}
zD0dpKL;1p{0M&r{x%|sX9XIyci?GS%-)XKpZUp0gnYw2j4atb&W-$!~raW2%gGIS6
z0zvXsd}C^%Ugj<x6xKt@ca#B=-Qq8Jf_usC9Sr274$oppyli+6dq#~1+B7Q^)XSb7
z9Cs@P4qqjEb#Q=Y%vrK`2Z!AtX&coYvQNBRPS1_(+rePI9zA3~?~Z0ZKwJfj_HP@)
z&g+eEazF<|4(Fl9flXbDr(pj!V>_&ouXS)>4}~)kxK$xv@6gKI23lqe%uSX9;}6*J
z@(pha^n)qG@LI}Nts1E55C?To&FefQc-m)mg(fiCw_;XEr9;^doFNDMI5krP4bew<
zh!GPW@i{mo4~==FVjQfb!(tdiky&_nhgQ8Oh1;3Y(oBP(@<4PE7foQtw%s!DZKEy%
zQ5)olXrxC`5BW}z4|0T+ESX^P-SliN^c$+uv*pMR0n_Ro7sOQ1C3T{sVu)k)a?GQn
z^>LuoG0~ERF=C@2m9v9uinxU+UycY<L=b1>`~IdD&Br)cSM^wbW|8;Mv;l@5*P(S^
zT@{E*0ETkBuZ3CgAzRaZg>r%|tGcZbebXQ(#`xfCk&|N4M+_S{codvh9lT3FD)kbN
zj`js_>&Qzde*xGAv>tN67U}G?QJ_faMXq30eV1T5?j)0bRK=LK185gXoyNzFH96_l
z!4z-`^td~TwhUbghB%Sf-P8E!<-q0dAW2d1RY}EUMr>C|u{Dc}Po+nUO{E9U1qG_$
zK#`vIQt99k(#v^{*Bj23-rfhMfSp3_S}A>k>;(uMWIFbZzsulCftl7X$ifgtjF`Yq
zf7=F4ljP(y2+oIM2ltY&T#%gN-|-pQk23t!m=E#wYy}R%)6$3*VEESh=}s>*QtaTH
zukrBQiY!^lU44d664!B@nT81L82gp?&zJZ!<2nvvy$lF`E`cF~TQY6u?6@Dm$RMAG
z&48{vF7!Mcu!EiAnF%r^NHaE%%q9ev1;dp$OqHQ&zE&!AE5ap)1rwG-b70!|@V1v|
znJ@xI&h}@9h&6AAJtHHm)3JjG$;jx0DbS;Qn1+m+MTa3n9~>R?MNJELsmYia@0dp9
z=jY{|=n04-F0p{kxLn2>5f}g^rO{O}0z$_*gOds@MP?vmn<;f}%xxem4T7BKwaHE%
z+R#gxh^<lF&{|f?_+5X;E~4X#sAe++_e?q8=SF~)_srsi#1gOw%>~g_1LP1dmAlZ$
zGi<~lxhN=t4aMLH4QW*Ab+Pw1U#$h`QWTt;KBsu6Tw=4WMVN4<T-w0_dM`5x8{snR
zqyl8W+zDq_7SJKHP%`mFDeWPpK$(xl+*;gT5pYqS<_6nOMLD-Pw|#{-tze?jOS1k<
zuk->92)<AFueDnM-+!e)qtk)g4g-@a+whFBgbd9^OW1z!b`YDDiLncY<|<LCV3NN=
zmp5esuVbh-rWjKNF;ipzJ){~`s!bg#hOWge^Zv|W980RGOT5!oF2HgdXc{b&E7CWc
z6j9#_2CffgEkWEdaw-ZgxoCvWkp>$@$=HEIrP05JxCICRse~3M`!h&)us>zSDwz^3
zmgnno*nCs1jWh6n(qygYCg38?f=o-}VLXf`(aV+53)<&35(L<;O7F3vk5zuXI_3&D
znJMkW*-W`3+;mM^`nKbh3$azMjV~1%(H`(6ZkACW??$=K<)9*7${GZ>C@0P9VuAVB
z+pOwp8!BYF$>oW)azo60Y#xVjrQDe4!&G05d6Jvbvt@u4Y3nSLn|(|?$^hBx4LI$M
z%O6;~D8vFI271md-UjV17K&pu;jQ)znWi;0jkS3oybW?&NPff@!W)7oxBL1+cZLwW
zBepooMk}a%XWC2148pW6cX>H<hr7H>K4Y*J;pV%8r$^;kKK~wTdqh6N`*H8JF%N=|
z3-?P^$bB(ps5*4@Gw$jB=pQv9$`CRx4;Wd84lWra5Bdl9K##fyQsiDh??3cs^fDMG
zyFnhZZ(3$_>LL&O_$dVzU|Yd*1BqF;=aJ}Y0xETr5|7&8K~bQLP4bxiIX1`Kq6&FD
zxTIYW0*!}=?0X`um|vo;7%2@NW(3ccC*ym_@zof}T7uyaDYQ|bP0CY79o&~=)R|$l
zLnULBL6w<44d|e>zj6MC!+n<jVq+oq@RXmAQ*U-G1{|sNd6|<2(02VQMJOvTQ7M>f
zR6~kWgEXgAhjav+2SU3{=J_w-aGVKs!zAc@UkMnPiUV$fHMAhuEjzUw1}`@M!Wb0t
zl7Kud@{gF^P1sQrJnb1x8I+?l&|*m%JZOmy7wMtJ@{IFw(9V>lv6v$eL_i2=SsD#h
zJM&yEx043?6=0ARMqgHKg$l9K+s8qv)-O`$F>R2=)yP%$cBTyFNGfHuHH2sd;L2-a
z3UHc`mhg(n+IXghvk9$rG4o;dnJaR#-U(?c1OkLqo)9Y9;2)Ev&FMBzHg<3vu8|jL
zA*Pnt-zIAjk&)^?*=z@8$-q+C;*Far1~CUcgQx9G+3FKzx0chgnnE?P@wOl&@KtC!
zmGYxRc@~(T@A@A@kSPxr%a23C;c_Py1>bI?C<B_Z!~V0wu_s82x6;1AJuNBd;aU-I
zr%fe~a7SKh_7mepTz$|Wt)UH-)Yr;Soi<dqKx{4D^)u_HPf=g_xxHJc%HibtMSNH~
zs8oI#9Ieqq^C1V^>ngvp*T&)I(t7!|wUwEem*03-;BF>0u`G`KHg&&I&v6u^m)|)f
zl|xNOE@x0)L0_&s?+m7e{s*IMc@BvH8hP+^(y#R_fNP72Q~P>;G|`?(4MI%;=iB~g
zA)!AyO##1ycjmcW0NH@Ikgp}r8WG{B%gJ;8-!+)ypKQfgMOyjhc_&)7p|MV0u-QOq
z;o`sO^QH1hK0&^wyku{*%(!6mxnt$!;1^tHhAUGl<Q4yfrxg<v82a&Qnkna?cMs_9
zqMn1;2=D+SeP1&k7ze)_3z@tgjAt?m(^IS&@h)@u-||Kpo5hRLVe@8M<+l33^H4%%
zlf30`Gw_YYB#5TI9b+HNTbh~E8K2`wV=9jw7SuT2v9UoWA=tDae=%~l%ovY*1wzI-
z{#=lEV@piP<KmZUc`x<N!UO6pk@t;?nEE=LVjrZrqW(Ylh3-M97z$v~qW@bO7{frg
z4n(Vn<p13mb_7s|4?`11UTGB}_JVvAYgNe$9sd#YMc4_VcIxG0`?y!nBKgF}$c@Fo
zbjWr2)StO$2P)$h`7Gv!6a_}n{xf~Ys-@KX=ga3towM>Nr~ZX?(Y5R8@-KhYwKDAf
zh4OEIM&}puL1D2jkuQVRc4gjwQV%$_!P<1zf8~2t@4=Fe4kO`8BH68@f?Q2CupD;p
z=*9$1Y70f)qoZL|VdTwJmvo!Ea?g$it&zl3Cr~L)_Uf3RI0)UnI~q)*Rw=}SUOz%8
zUG_=8EQ9%rlNegDq2l{S3#IwmCfU!1ScRK_W&bpG*nJ1YA3gi@manDOxa4-++ClsE
zj;#jJ6j1mVQ$Db5%s9Jv03mZTzE*K(v!O8BA_sLeU=LRseluRu)Im;&#x{k{w>mm<
zSdOYfK+$6SsIpuohxli44U5I(&>MWHj|3VGBkW-6&630HTM9i`Ifr+2+!0C@qsEZE
zY2)bwr0j^8|F8ig<vSf6Eu>vb|Lu2u>ByltmSJl}*;%dKEJsGu10xvdsE)0659~-r
zq{5L6hd&^*ql13s`2y}>lw*QT%Illz0GKt_%lB;1J8sx{13vhEd;sg57mOY2(=<eq
zr{^3wE)gXe4!+Zm@7Ss!;=H!V2QH^(`ZFfRSdseaRyo0Xy5b6mbEF;1i9wN(Bge=|
z-YsD|F4PJjNgdJ>LYK770lCE_{*FPv0f=Yev^#Y)_)D2cWeAifurJR+!rAm3G)rf1
zP{XS@3XV|i)h=!HS2nYIr%^i-Z^88`U8BR23S2LYe3NcIUOTfWO;=0zm;(ewWMUWj
zGiGA*_V}4Vwu+-M1W~ZUrSwRREmF*6CQj6zY4WV?O(tjbJfeoAmoZf@faSeY>)UAZ
z<6iSs(kB*_KzSxBeeG|g>!C_K_IN*U*lck-uRw&I)1|)?p&2f1=%JW6cU-cr<>W-Z
zZ5Rx<6!Uu+Ykf+ZId2R{n*e=IwQa@{@OjMfv{>!o$SXSCTZ%|Vie**Q<P0A}JJ<;f
zwGPf|RX5~JlP^f9!VLu0WC88h$bcY)hCMMaaJ38!x&rS)B`81%b7hcsZJa{O&Md1{
zGT6H|*i`iz<8p{@ro#;J*2q2dNzMUydX~4QqL7@gBq8Q9G_}_5w1yf)c%AtH!~FLJ
zhyviVxsd;4ZeFL&tI3DQj)yh44*x^#wOvOJm$8Kl&XN%!!bZ|xI5H%|E^z`fDkK4_
zJY=aK;pPIH5&SSvZwh6N^>Cz&F#=;@m&rLHPW#K)MBWu-Q7G8+rPMba1KBN%4r>oN
zjLaJhPTjBIgK;rz7ZKcaChO025)g*-Tg1V2wp0uUJ}-z1_}d(k43eT1H9m>(DpZJ<
z^L=n+8kQrah`1nVQk|LBBp3P{sO+Q&-ByT~i{gy}^U*H$!8Bu-bhoK@6cU0R^VGP+
z2cyZE=mUlsHwS*XG$!bi=HW^?j!e=RCY=5#EPZ*98}S%X{38<|t6z!d8w?j-FCb;{
zHZFLGbnx<!90Rh|mAS?msYo-&{ma~pAWqg+Ei<W6Ciu^&_o^nra)^&g8x*`56*=TR
zRoN!uO)a&OGm@ZWWmQu_CU#UuRLxYG<n0(vj%h<$RLA6yHHhGj8v7InBfP{?>(eq7
zm>?Q_qZBBT5ALkZ<|?boplDU)^A%DTGeZD1(9lY`BHo+G!knohW4*B_QUfzmXt%{6
z=)omCI+^lAUs$s=gvjY%R3wc)1#?Rn15OU=f$TWa;ZxZrQ>+6d@0ZI|n-e@!EMVSb
zzu>x~p)$>yMr;*)LRWe-rim~wVWV7?#%8eq41BauA?y0;m~S-#bvfT_e6RH>mTTi%
zh%(j`C96dSA)-a(x{huy=gW{*VIg30y^U@x6L_WvQxW~evB1<x=s!^7Zb+-N2@dxg
zQ%Bf#*wm=OfeqYtW#diuHb4_ve>eNv++{pbk)w-$k@p44(bM|_qe*>^0rR&c2Edic
z+$hXzW+;{18Z)oU=21AIAh+281<NQ~&j6&`y&u&*RMe=nCD;TMqFL@Rw!G?ixzjt*
z`UiG^xM;a6ez9a(2&mQm%iaD9j}wY0?pAa5+!F_9b*4fgy)*@)Yp_U=+ok`?eMal6
zufF<Hxj%Y>u?H^$Q!X%3h8I2H^JvDXd7LtCgPKs(QFt&o5-~3Ll3J7eFr@xiFH|oN
z#Tu!Qr5XV`VAS+z(0Y0}`qzWntQnn~Jd)aH1RG`5GI=z3_@WN-nDwO|Iou%wRPb@*
z!5;mK<%!tc**fm-C%qTSMIP03KbeQ)sa-U$_Kq}ow&^}s)nZ00y2dzSk!Skg;>cVC
zlWdkV=WHf!$n1`-2dk_0e<ZHRoY>f9xrs72$c$r|k@K@~zBKPl0f^EZGZ{8&+~Bb?
zFa8{YdZ9U)@8dwhdj@<%daRKJ_N-;b0F)@Gtk-n)g-!rd8(?-L9)x6GB#WX29P2pQ
zuoYIxVt>Y?2(8l1Buo5RGmZvIC`bfZYJ+LYH#RhuAwVvugukZ{mtCI7z8srP!+o5M
zSHu`AfC(~Q8RqIr?|~{U&r?RZtg<EvQCcOj)rntNmX#+)V4Q1g_2EX8wMO6!V&by0
z&cD#GQS3*^5&TQkXT5hwwU)6ZTQJQHwmXN6myQ0NU2X^xpg)7vHW?SPl8kJ2{=smm
zZh}<2=uK~1$4LU{T+i4oA@WAH3)R~i<T3!ZAlqUGj~G0TYyBf*s1|1BAaE|ogFbNU
zA0pQ_$oBMXD-KBR)VZ?5p5d04+J>seLM~INk(NZYR8G?8veV^uzy9(Q8xJIUrn(CE
zBuJ}&1*X8<njwuDXhH4pr!mq@S)L31v)JWKXQY(T+CGd!(2ybdxxLR^X}E=J<rmiL
zpj=jd>74`t+f<fc`Q&(u0>Xg!ulza$z}S)U8*8LrQ9t=@$JW!i*=}o@@&Eq+<af~_
z){p7eI4R`!u}BC3*UFrKh|N0z_su@-U#J;@c{9%M85>qD^05BsKXa$iF9pwlJR3`~
zi@;@K+H=trayl6sDSt9v>~OL??+w}YE0o2H<pt+Fc*tR15S*mmW+g6pD%eWBn1-Nf
z?2zAU<t1M&Xf(-z)h?=0UbcQKM#w8|ivum6W+L;V3B0_TMzjpr62Lf^QLly0gE}wy
z8ldrp>%49RVT#Bb!O>%w5cH;xqrI`DrYO#Q#;CsKeegW8h`nX<wuzj&>PF;<739x8
zz%X~(idnxCM-gtl$GMC+)4#;Z)go8tujxZ-4JvG$V(-Rl6S2MD^Vdi!E!m?&hg;x%
zADmU;#gT@i{e##RQxF&imFsVbNQ~O07O_zN9waD-D3A|h%~6>c0eSWEQT#lCcLx8%
z8&FY=9Y{?3WBUZ6trF2EP(VHjGC@RfDySSPpV~gbbtDb)Su8iqH;#g_Gg?vBMZiCO
zj!fQ03>5?<9TE({{ydQaVX<FWH*i@pp9`U~(E9!r<N-K{lrE@_tXRyb?0=_5lL1aD
zL&ARPzgQta>y-1GFaNPY4;(jkgnZSGH)s<X6m3<mn2Ro*S}i_pC>)WybuwfS4u6=o
ztc<@%cJJh9Rtd9?sUGan$+5x-c}!<d|3>j`Y<6BKwO1#_-%dI4IN7_CVwGc%BPaWG
za;y@9x;leWq7|}le2qO#rIRQAe*O&fQ(=r;1=+uogLSxV8_Ruw{aK1wXyz+?&A*^$
zkcDa3%GXoNwoc4xJg}1?qY4Ar3gcgT6E$Q*zR}5WPkTm%c_m23{JMku7uvWC)`#Wm
zZ2ryog^HE7Nev%Jm3+&8G4a4SI#rf%-XHA0K#c;YFfEL2(w}*VQLbnI-g2l>4oF|6
z9Ojdm#~W2JL52?>9%KTyFqwnw`fYz^@G}MhZaBi9S*>=BT9NO>(9|-}vjrsJJ$B!X
z7KkxnI^?G#joPEj8&Ku9UXI%3W3&1htGZZ@wjt+XQOYsFSi>Q&zUQq%*}+(?O2qd&
zIrv%1+r%bJkYoKNvhsA$V>xX9<Ge2&2n8opXyWLr<oMu-`fLt0AWld<tW|e2t6iNK
z%MG2Hxy4piE>$a&lhOz>HHAiyxkFkPa-;2FwOOR2H}8=rM4Y8ld|rWS%-k5A(@T~P
zPnPSlE3tqf(8I$pfUdC{#x>RXCEDF~c{!54kXa$!jj|b*nJV&b$i;*uSOKWmzjUk~
z2dL)GgH3w)+v*<1=}()zXRsD-v+Cu)a~&D2fm#64+kS7Ean=Co<G%yp1SVlsRQe`5
zF%`h*WXEJbf13vZcLOSAN&o195Gy8+*cCZBo?)bxuDM)J3AwW45$xYngFKUSWnN?D
zH1F0xc}|>l=O8BJ^tR@-3+|6II;nZ=sY#lkv{=q`masSzUtz_SV3GlrC`k>?Rqr@U
zl?@0^V@!Un4D?s9$JDN&lpbWyaG`G5L{?`YmOt3vW0W0Y*5N)`Awz6TsMo=odMy1l
zr-PgoeZrke*})v^q45lr08m8|s>y5_W;+Tg({NV|j~<x?n5)bokVMW-9}#ObGkZil
z19-|fFaUBg(i*C1g!fxU+3a|oR|UlRXn!BaAM_EU?wC%k=8VbKqA*CcoD)k^Xe^^$
zgbenvsT=x*?C2^@&(^4MfofASE;Vo0Sx`=%>(9V@2zr6d1|9A^Une_dVQPSc<R;2V
zn4U7;_XCs{Ip5iWJV!55s{AAu7)ODAjO)5E#KHDih_W&hT#r(G!#wMY5^pffhGzgB
z7bGR)!kXpcI0<<Pa4pZBOKjvQ2Y?!nfJf--xzva|8P|np<+4Nq^6ESs8*}CI5F3Ll
zXxtSf6Z2PxiK)wFJ|~qvm@^OT`C5s;+~1xbX?WO875=tarI3|4Bov^OY|M!u0a-$K
zo82{lgf*GqUz#h#vsggKg;JUNp~|p;6;q_j7{OpTXiT|ajG<X1drXmuc6#+fmg%JU
z387p0imX%{=PRk|n4HKhQj=Dh^P7vbDy>5eigPWONUe{7e=!P%E1&l@Wt!|<sq^t0
z;Gi*KATyQdD{L(Nagx@1GX$k+NzOy;V!@xOhk!x}o~VZ4YrPf&N_VX^+G-+is<A>Q
z$2Y7-jlF@uc$t!h$kL9i3cO6FruPhrVB~F+KjYcQ+JpeYWLl@zp|G|%B4~ZKc<K|w
z^yvq-q!^yd%s=Fv+@_M4U8vbMSmw4c*((!i)?y1?Wq;wy31&oHZR9Az4a?VbQatyH
za=F&2r&1kVun4y92)QoiJp|cc*C!@PnP0j<i)DH&Ae1BqLNH`T@gR(y==%+}O>t%E
zEV(gS!D6DS2Rt1SJA6~@h_YP12A6oq&1qbAkaFtOq5v*B83FZC$A%%bNN$NipnLU6
z?3}K{Q*#9H`>jqE5_rAb7A-*9a(~mAe|vhy3z01e@eY4Rsl(mG8n$<)_m=P(IB=HS
z<!345Nbk1)ra2+^#Hax4@q{xEvKo{a^)k6P+JdgV!aaib1p^|*0BuZ$f_%RX9;LLa
z@>zKxO$8{VJZKA9KUsblAEJB$v2?(+55)$=rOeD&c-U!bt}f=G5ul`X@<{yBR{R#@
zSpb*DqtQ94F6HGhpAiGDXuevJO+V9N4nH1)C~rdPSInSEo-nQ(JoW;4vTaSV@X+zp
zal#Czr-Bkd-Kf1Uhpe4p{D%`4>@d?`P@YoG+A6dB**pdY>;7FUvt!D7vX#r6^v-6D
zhe51~%#Fi*Oe6Pqvkh=$p;G3>WY2;RUFIjEEaPxl^oXY?e83B=ms~>?<_JZ1q4#2#
zDEBhJz_w>BybTuEB8!6|;AMyw70Cv(G8}e^_h{ZLN-0m^kfpxr&?M0RGCPJ!P|Y`2
z*2oal<u-|)J$lKCm>$FBYGq|?<C1~nWL1y{7nZ~GC#&P{I#$kHAZvWkbaZKSF(mL>
zZ@Oo1S(hlnBs&wyRLi%>`j{;ozH|uF8@9pZJ#S_KYCv(1^KzrV$UVgyM0l6NChMVp
zzrM0L283W#WQ53;M6ou5npbx*Vt*DWw$+;>hDWmiF}kuX+CWGJl6&Mw@pm0UGl9kZ
zI93@k#Rb{k$+L(GfKO2h4*^A>=MLLf!?V(2gBv$`tn4&W<|-I0@)M^8m0}H)=79^p
z$y{j-S^ybk)B&pz|1>>A@Hw7g4)QZ^!H+hOocF&|m;JddAxctCChYnPlPTkdjT|h$
zOx@C<LuHN^tg&SJulyG>A$1$pQc(0;e%;AJGssZk+@W%(Ay!lq;pn7471B|D;{!#-
z4{R77aP#H2Hk?UWcmQA`{mx%8XRTGm`F(t5NG(OP{2>jM!F-tI&KvB6lc)V{cF9bF
zO*na<u}%o;a!dWuUV+d9ry~4^9X}fj2ia04&)I5BKzw(t{K@&m3d%9$5@#oLYod9q
z3-G*siqh%eMR~zT#t6G5daSSr9X&NK#@sL_*8N0Ot-R!IQ(dI@j0Xgr6;X)j<q&$v
z%fRiq+>WnUcW|CE=kwJV_ekCz`I<LphrO*9k9H1$viNmpxRSHvjWilwImk`9R^Ifn
zx6d=trsOT_2D%oB^|t*r3G)R;_h(}vgjo-fcR~+??ylv8c!k$r>;)AWe@)B7T-OBv
zjMmG$@l4M$l460p=Vx7QMPA<b-}wtQbUMmEh(jF!MguY*<ZtmnE4t!D`nyktTY#Yt
zDW^d`44M_1uw_4T8PubX{KNjUq%J2Pd!LL!U={{1U-~47P=i2C`831_U|vmI2e%VC
z;R5n9u+MA;5H=0+Pj8JeFC_LN0MMVufKhm%I*$gwhz7AeSRIz0PwU=tT>ljY0=5cH
zOfd4ltqa_!Q6*o7B!WhqfywkAUuuArUxk=PIZlzz4!Bk2an(=0vPpL9tZ>@!T#fAB
z*$^roDIQgi(L(ge9{vmWJ|puic9;Pxdv<obt^#MW>=n%;#yZPumSykGhWgRUiR9{i
zIxB3q<00rz`*wEo4;v8`QDr}W-$KKP(=3twqYaFO$RVD}0i9b7TD)LzseH{}Kp|kt
zMZiV4qM@_Fagj{BQR%rcCDR^g1IOXC0#1N$bXHl7+Sw?7${eTxO~WeWAa9pUWD<0~
zeAAy9kx)Mzi~X&bHm|vcE>$N7$MD7?A@h)!V*$|!P`eJb`H#Sbk8+s(H{^bU9PUj=
zDHp=)-?pY}GPT(6nw)inPb`ji#x5(qM=pmz`i`{-tFcVJYyaJ?TenN(NNZi8Vchbd
zvFS&3wve`61rVe5qZ%`>nLgS`RXVO@+_|N4jJ1i}f(H4X{eogfDDNWQcP`=DanEVk
z69u6h>%W-3t3Ux*9N{E#TxZ8#wVbsaACzRuRyC_Rp3vC?5hehtK}rj9O$u^ikoPPU
z$37{o(q^@T87$TzF)V%I)EXk)qoeWr7{r5gvYC`-t0zcj8wc-srbW{wRv9=s!y}O6
zCtdv+FFt044i>KJmipHGKx~h#(%qk##e#4fY?C4za8|=Giv0!T#uBhYg5gT(;m@d{
zva%8RzGu9rz^sYlUY#B0E(czYN*K^HdI#Ax$w|XkL40DAeQd5JW6zSlL0Hte#M%Qs
z=odS8$gmNErGHR#WNxaQ?0bVVZI)BK{r0&KbWTr~Q`52%90_=E9k8g=Vwkv66Q?Lc
z9ZvVzFm;g?Dd;&qBW==6nw3MXpE8D7@&uF4jAllcP0r;T3vkzzbXEiN*sEkfkj&Uk
zHN%1FnccXDvs4E8D5=9ScZp*)Vh;AVn}MD17hX>_BnZ`5huVTX@6L)rOh(<H28CXR
z8nqAxju56{PA%Rgut|m+Q^BY&$k{>EfmIn9k%nd}I)W@@q|cwp5v;a{3XR;wtOG&D
zlTpT%12h=DN=8Q$cAVkg*)k^0VX>l<fcWGbTlaHIOJr>N1h6H}CIsF{X=eq;OUhX;
zk+<cH^Or2pl&ii~&W%}DWaCoy^K3N3Gqt$gtx3iQ1Jxp#RnE70gD1=7g6LFNW<KgM
z%#sT`w{{{P(LNB52Z<Jx)szc5zUUaEwoUlxB5QHSGU%-rhX@@E(+JFTNl<zGNVzmo
zfVrVG<(uWQp!|7IQ7-ov7-Gpl1mc~USZ1xIz-2BoqCp>&^7tBI8B!6=0}*G`A^nP3
zd$)gau}nzZ1-KE5XO{<1-KzBQF9M@i1+7p42O)5YKC_?;5^7&06Wg|ef)CUPp}c`6
z&5$5VWm2#Tggn2o-&wIlLouY<hSU#X;5DiFc^X8@h^X3V9ymEG8mojh%v^m#dH;rJ
z4RsVs?Yf{*396-_fLm6sNVJ%V4a2aT8>BuYcrDJ-xl-^~2%7PtO|C{m@RD9<f_N<4
zS{f~3HJH%50vmO5v{Zu}5mbw_8FjhnOquF$$Fvc&AWi;^9viz&b*)UZPfohz6uHvm
z#ObHVRsNzum(<#~$<^s&+M6bBukn@Sm|4u%5`_p0Uh8i+!<Gi^LA(WW-7d0^M=xT$
zpq#Fg>rI!yoj4VUnpfi#a%;LDwW`2^Pk2FJ05_Wy66VeIHw2#`=)6I0jI)n*ne@!Y
z(R5P`Mr$9i<F#_L4~8ZLeTSA0McnxZ{!23qL)3nO#?60u<>M_z?J=+%XiD4~6oIP-
zI2(?kRdSm@n+HH8$|2nDj6{g08T%FJ0#>=L<n<k~cP0a;y3>0nKM{9jNS8_eyE-d8
z`9I>W{vRL6-A-N;Da7son5vhYXb^qr+>-Ys+F7_7U6}W#jTIOEXweKVw1&dpmuQ>f
zv0Mt=@2#;P-sH6bi~m3frqSc%LF2|=eT(FW@hZ|*sX){7`H-zOLZ#{Vc-Y>6dIs=-
zNt-bC9*M5ilWN$Fxf34s_Mi$wT9B~o<gv8WIDxUH)X_;7gFNoP*bWHmnkCkrh}9|?
zJ5HYT*Icz#c}HA)t~_PD0;OK-+Rm^^Hlfgs%#2=Pw-WN?&2}I*MC`1XGqyMu04Zv-
zWp>E%^Q_$VoFLbL5}E7oYlM`?Nj0Z7SSUu1NtqYJWR)$<oqLi@wOQu-QqqQ|Gti>q
zH!JYZ-xqjmh>(HMfp7%QxiEwd{QWoz7ulMj%$O{WvDTpIHk>a@5*4>-zvk%B5Ig&5
zV96L+mdHU&mO45@tyIHX?wzY@#yyJdzQW%ApG+?+(<Ci94_6vsN$JI36~h{usb>+Z
z)wYC_^Qg2WYl4Q5QM`D05tN9v(J>G-tk6!@*Inn$>v5*5#zSE-Oz-vnHn|NEZ16_R
zk_|S$PMt23jj02*5-_txvdQ-z;WO&4_yk>ZhiOuS@0QK+C6<NJy3F|6vdb{Ih{(NG
ziMBNa;Siel+iU>&%Dnt2nrT5qFn1CZnjiZ!6MP2TZ?{h}lXJ2oD2BkkdZ@Q8(KdXN
zb+XexX74N*J_l#ZPy89rB`SuSaD=zUS37P5JowYLuK+%2_@VqP-Y5aIBtQ4L5r<*)
zNn1gWU&N;CQPf+08Ha35W~$Y1|CJ551fXV90Xc}jj^RV|;orXr2@LUMqDlb{^^x#S
z>EA~4BLH9hE^UdFaSYnG6(RV&u|LYS78uq22iuQO=$ho|Gy(Dw$z!nUe0e6wLpUA#
z2tFkFV|ts89%x~Ve^h1ZS^tIGfOSgeVBIv<M4pSiSI9Ram;F!HIn>R`@_Zr-Qm;_h
zJ}WQ8w8r5Od@*Q3m<fFUQnbYA#f-dci(k~Uzr13La_*QxCF2Imt3L9$hl43euf=R|
zNYSoaj5Mp)gT@h7Ktrt^i1bGEk1`L))si>84Hi~`{)#hny}adZkaCPAz`fJ>C8a0J
z+nrl2yDn3q;kCT&g*mh~_Wu3*$~!UTF$0ImUt;KR1F{Pq(0}!3w3euB(U1CWC>k#Q
zC>~>*n?=3?-lq4gm(nqV<$Z6&;(8Wf#_Qw*TZ%$CycK`5S4$8K_;({FCc`_`KMW#a
zg46~n5inU*>3N06N7e|gm_(3VOl9QIY4UMQ1@JO5To@H8pTy=VWv<<)iES4XSc9&?
zuxkq(gpiTXthoZ*She!cXbzO2i-20`=kW}>4%2aIlzw3ospd7Q{xkEi7&i}XTK?^P
zYQ!?-fm-Ye`O@De9yc4CeKk~x{}_XUv>VlXvq8S{w}{<^qffdR{6!3yFyRtN*xkA)
zWQGDjxKvhl_cuT|2ol>%d-yX;_QyGfb862n2F@}1pUdA?aBDAr8xk9W6@^hTm%aUM
z^Ipt@%-1RfV2^#eDEL;#E0y=P|HftYgJi!hiUYzGAp3iRX3QWj9w7Pw@r<yCAH-ja
zA;I%cvTTJ2vMvAhF6~n#+Xw8EgxpF577p|_Sw4TAmd{@*->@NN8HjpNm)3nWI==*W
z#?;C;(>z%x*DkmEmeI0h#>C9DX>zc?f$(?Y^DVr`=8zaTtWeTz6Ar9HyEy2GSeb>W
zw0@YiLLCM5#t)A+i~INPBj1iTajfC=L>2Z$a)i;UXW#zvooF76!W+I(I8DB5Kaav@
zKGLSpgxFRVem%<m<kho`;5ga|s0nAv<&>s$;2yKfFD^l-JIeREDDVl87h>o8G1Rko
z#r3h)dMS!qA_zi`i`BijgB+h`VdXiff1Y41F-I3Q=YrOy4Ov*gCv{PXK*2)jd<E&S
z%Zj=w1*T?_4&qlpI!1nX^lpgkL)5FC(Wy%SvG@v{qZiR5DvCw1h;eidR#hbzB3RuY
zOL5|M3EIM_hEty^jdYC}PX-($-O{v}P(-6@jdZt};DY+v9L)M6d!=Ot#4E~SXhdwW
zf0p(xfao6ni~?*f6c^sK(=&#yL*qeCe;le?QYAoicduv%f5B*0HLtfnYlg#{834Rf
zq)&Q}CIq(t?um$L<6F9{34QS*y9LtE*r!+DzS7?ZfUZ^lTL`t4HZUE6OyxP*8&*o9
zk!_gBDfaofIPFgL&%*>_ntn2h;9ET{jg<zY1|`Yq(WELY<h(_4hL6=O5R6q_9za3m
z%rrAroS%<SsAegN7BEj1>LJWKz@HKPTZ8vF_<>zot=2To2cT-mFCo5NkU?GCM^T$)
zrQ_PX3=U2nGDwER-W`qF1=vGpbx}NU?71>D&9*I?*{aFHF@uzwVgBY28J>0*--K1c
za0SeFt(=|avzeX*tmsA=VJw$Jt~-E>k=B1Xj0Lb)K}N-Bc^8q^-571}V>RjC8k1(u
zbHXgObL<W1Jamwa4VFgPq;e_s_X(BbwgU)C1h&QItj!eQBRSV!aj^l7b-kRIo&l!C
zWELZ?c)W2!r%or!`R!V@!aR%R0&5jXBpIre7si%n4NB&+T;x=u#Yq){YQ2f~i(?H&
zB6I^t;3dHySn;@$#NNC#4Qr)(P;eiET-HTR|4#p({!cEqX=WxhNhYztVs1#a{H;>B
zy7r~aUS+w|avw8<668Ah+`{tKNJW=+k>ri)^2{2k;MLN*peOjOv~n@_^8yH~${^oJ
za0y)qRcY#|GXN%A0Ieh!A3}06XN$nz#5554%`L!51-?v53&8SGs*3mmRHwErXKOYv
zpc;S1ATXBZo40+o)W)F7kPE;QKOZYsR^P}QYU=C^Xqi!kEVnD-ZBuHiq<)tH=-~#U
zGWE33#l12)*1l~l*W;`qZH0P-TG7|uXbrT?sD#_5KF<=T%=1LYlf7wEwk%?n2Kzjj
z63otm^eA>`d%wOBd#%ZL7N9KbifJJg#+J&J_6u$?B_?r|H%_rgO@Tr97O{G@w?MhP
zL9R*Sm4)MI>{$g=+?k*eX4l40E<_#lO1UmkopDL5K(<P*w^v3ZeHtW|>4{=Aw=|Ys
z7jCKCkS5IfIWCs~4>Ze-)<nJ@2}PjUP5$S!rb-@hH`|C=VhtF?4}8RWhG0(6Cll{7
z8~hd<VxUM9h$7wU!v-Wqd5i$YnR1)S!jfrndyGr#N&=mP?rMAY4*$guc<7K3Q+H<<
zk8D8-H=M|*oX^F+E2yc64An+3EfgXi^u+V;HeN##IDBvS*y45Uh+As!O+;&l88Dqq
z?u-45L)ifeOt%ftx!+nDfr8EQfVW~&QDf?`0KBB-!L&0hPna&-9|q$WBN_UkpaDP`
z#uzSEqo2sa!_lZ3QwVa=vmZK?JQCk%Qaz;z21rFk9*u1@fYC`vynD?12Mg|;4wB4-
z?euuO)QF-_^}vUp@Rzvt35Wx|lP7KMTV{;N6*BUazp+3glI&!dk)}qS7~f)KGgC_&
zm~qY!9GR8gW56v=dXQkYKMT8-iTiVWRu)7>twh5ZVLZ)EZ?A`aZ%B8u@1~ntw%~TG
zc`@2CyrFu2^hh4(`>?7P1c%TK1x_S>p>PlWT>)roVGv~qWCYFnMQQz)LWW}HK|hcN
zTBoHsERN*?GNu^;3JJm7mL;)uOSN*xQXdT>S}}hMYg!f~gm$1Mf8l3ZZU;oK9=&Bn
zOb2E_e_0s~0M}9>0Ya}zqf=}LtZvu(GMbvVxUcqa>~Ntnu|d{EI|`I)fT67QXLQ}7
z1QhLPS?A9v%~9V=!@t&t(A8@iGV}=4%Lbb$g6{OB+89Eyv|cv(^1!UY`LNkrr09r^
zyd~(TK1W`L#pPyL)7CgBDjRc%ZIf+5&oL+i)FeN$?K67ppuuD1$1zrnAAyXTlDR$2
z$ugv<EbfRl3K_&!NsG0SN6b}|><nrTV1nIGVt4kyLbTdXIozI6ivo6W$mD8a`?Fx$
z&%CQ<XiOv@N&j<ezz~&RB&OCAmX-=;iCQ`G%Q)&0y+YI-C*-fvqO!;c&jE(9wdJ(^
zIwpadO1S6hH{Ldl14~-xEtT{U{?=a1*3x5KA-@Y!>oo#2plG48u;uS#{LrxIw4gWe
z4@UlcV;N=$>F~6Ft>+THW2Tld#WU$^EF8871>PV18TE0N&|;C+XZ;zHFrn$NP}ov=
zE{Hy~5#Y|BjCFWVk31i2l*!2p@fL3xkrz`tjIguFv6s@^7zW7xf(S4BFO-keJ6UAM
z$n#3<b;xct)LspaLs<eOkH~BB_aG$oVqsp7g-1*kXpaNujnp^+HX3HkOJcq1H+_)&
zi_vckl7GwA1Njxa*h=1x<}7#xH2!nU0%Ij~jsK^Ou9tVBIlyf276NyjFMsi8w3;GD
z_^-jX#kgMNUF)yRt4F<O6vgD~<b7XV#PmTODwhx1O@tT&L(PeHEtkIql?Z_%V_*L6
z-I5WsTP)*;!Ls-ejurq!tOS?JXvs(Z4IMu;HJ5@y|LD?sD&6s9L>|!pWy#}-I!5tD
zRLyCpM}x|A9(|n1g3FH3BQg!;6Jr)^nOQKoKJ~wlx&V@W7Ara?Q=Nfy|EEzXS8G*)
zJ`X+{2sBu}h^<);vkC_aQsDmO^E5w0co^t&FeKsMv7~xK4%zt2Xp)yu$bbBM;`vne
z;c(rBPG8x0T4qdvpKFpxSI6#X8EO|BWjB9@T1h)tMfYAr+3em`5t%NCvg#sx_zS*1
zsx0r>RRO<oh`GxYWG{b%fk6a6X?E`2)lFVl!4O0cj{Dd*faWp-t7YG=YUGX|FZ*?M
zOl&}w=}G%XV^}C^l?)p?AR2?*wt8z{i^ed-Zu0eLLUl?xu&YBoh||gPI-hSun~rz^
zRg{Ce8i+%#49}YR*NAIxly63#h%Cmz4LkW;-iiS_CY291N&%dLd*P6-4#HI-y$R_5
zp*9X2V;<0bSd4RIhV^C-k1hr?x0xW-x1$TaIf^RAMmeIZ!K8NPYC(3jhG)um{B5|6
zaQ+!s^W7j!VG3@$KQjJ;%c)V0^7qjarc7w?dO5nQL-ue{z=5d)#c6m<R}W|ediXt?
z24W%R@P8<N=Og0j&cGupS<muo$J(lnIaiL0sgA`S&&u&#4L(&bJQ$UbIv{>1H<pn@
z6?n6pU<53wuf)a0VD=Mzz*d>o{KO~4Gk|n7S%}4SuxBkZ6oy2=4z^ZDn|cmKX<)^4
zis1}IO{uJOw&|f*#YE{6tB|WjeQSQ&HReaX0b+}Wj~-Lft&M^(z^D^#lkTzVL$iS6
zkZ)=o<2y`Kiv0_OykN4V)ne(9TG;{}kYah2^o*4n*jSd4UOoaEsOk`gmS_N7@92$E
zAIegn_<J(zXB4DwjJ;h}y-Ggm*Hz*0{{8z&|1_wL&;_74z$0*SbO{Za>K=@3U8jw5
zioK2S@<sqcATf5-sn+d|yP+W71?B2AUtcqRO$1YXQX|mojcJR?=`qQnEIfXO_ekNT
zY!qQUa;7a0?pCXnl9(Xs02BQj5Su$whk#IkO#^KU;#$<Y;W8*KY^4m2-k{*s$dI%w
zs6HWQB?{2zYRBnNACE=-(E&x*FBP9*X{7X!`}xQK5E&lrfTLMkXbmLG*)c*~L{(ph
zV|YZn&v_W?4Hh`1*2zd4ZetljV@A0^vV9KI7QBc5VR(G9Oh%`!pqatuQYg+_rk;!I
zG=V33jcPf^-_~#O3*L~i{<ev^xlmoq=`c%5Z5-oi@Q+JvGHZf52%UUTAmrS(Hh0oP
z3@|{>3+mJFHa-o}5-?GTM$Y$V%+_FcpsQREyAc=z(nBun>hM$sSc;`ITogl9yu&j4
zIF-e6v6G3uA)=4}ttegMzgtj(njYk41@z&^l1qI&#74Ln$WV~WY-^MtYV`78H{9z&
z5tOmlTV`;-!O6iq%(58%S=k1Ni*j#~QSH=2>0hfbg4Qr{J=?XFC7BwOmVx164ROd!
z_+Jho?s+;*SZb`&n6@EbiMvas%85cffsv0)s8$?Za=~)tDEpSLK@qwbB2XU03r?~>
zfh9FG0;;Nx*T=xq%URaMR;kL=RZDG<2!ery09bf2?|mC7snA7fu)25~oWd}xRoF0B
z1QmIxV2{?v$bm4SdIY54dMq7XHU+1ECi8Gks`49jN%*J1mZlHNb~dIVu#%d^)Y$%<
z91Mgjyff2Kb*&&%j4!JoA?jsnbPqiqN&D18eP=f%;a^#6m+DN5cQsgu)_`6yfyKHq
ze$iH#cutIcF+9CjISKq)Hpso@a&>yPjrxV6AJ@d7aNCt!8#|}uiblE4_@y3kIdZ-C
zL*7PWnHCzbc~18>Epvxv?<~0?ksqNn;68+A$c@%?22gKaZnCxK^(@u!-`yPVtA1rc
zWIym8wWb9YrY)%Lmgo^IM*)M>Qn@v_X9B7@QzYNkwe<*U-dqy8o>{WZ4gy`zjNBf5
z0@MgQ>JERl0Fmg7KxFZhJA-SH&W1zgu2|MOTm%CM8kHOGj=m>0)@Xv&Jw`Y%a)sO*
zT!$jtfS}|)BSRjJvr4%?QI*(fWCl~|e;_`aNI1734_cEbGEpx-j3G4E78=V{4SXmD
ziP|}n3i7begR4dBiUD|!7}J&lNq97-j-MG(KuGE_8yBKW0l7SG?_S-pi#%cfIWbcu
zPg=hv<+!MUG{!0NRP518oP#nW2muWj)T))4zL8uaLJ3F2CoDM%9Nu0x%LhW7%j&i-
zmf30DS&_*QTs^yGj&+8c!We@!x2t30aQPy|l$W-YOS8QKcG&_wa$X|C24z|X>CN|N
z3?b2KhQMqV*c_|juUP0j4~0S&ITfzDSQf{=7>8}V#G9a0wQ@0=WND11v<lk=$Pw>v
zT;@|@C<BX*r}zz7?oFUd9qlR{8(U<BzfH%V#_A?{v@#Zacy<CH9jfW8($p9)q<chG
z`!g-t%Tov|xhD8%c)qF@MZMRCYTZ~12NbRdgQNwQu8VanuSSN+`t%orz&99g;TL7#
zjsFYJ^5CB@oBS8xehjTw=#6^ZX4xDgBSwN>SVn(KI%e83(h2VnV&MHwTayq@S5Dx|
zwz!^cwo*`VE96Jc84xD58Tqk)NyL>#H`AFw4QTf?Mi<HUGy~=+7&2;A@OH!!C<ICq
zMyq}xE~F*Rz{&tFhF^N8KLhp)_R_cz`AIy3sn2jUvRrF=hGcU{L*A_Z)6^clw}26_
zRzI_~Lb`sXS{gr((HKaemI-&lJo!bM0V1m)?^B&y`DL2BMP)K92{TlGI5c1Rm2VJo
z%v$+%8U+fLfI2+<e&f$H8wbxeBXZ-n-k+Hrs`Tp)UMRn__NKs0Mm*8){XHlOsCS^1
z%lu6HL(l+c6C>_<-NMtsW&`s$|DUnlHn!vf`J=zg0WphtHdyFcd)9(Vj<u8O8|xa%
zo8&n^QlN(*tj%X`=1IFO86Uy(-YDe{4X7pHgcrJohlM&vJ(6{aO}?0B3&$beTnxoV
z3;D=Pv3rmVn45;|P#E(FE_&Hm5y20W<rNpy5XACoT8yP|Tab8!^kIBpPH^|gYd$2}
ztj?nHdK~_!6`a$P?2SYX9=%i)HprXN<G3O6R_di~$xSq=zAkU4=`3fAmO=}5bpP4U
zBF&|HCyo#lphTwDUt(<Bz#t*Ej{MbIG6`iMDT5Uj$h)ZzR$bBq9~%Nr5P8pkX`dj-
zfY}Z5ek?z#9n{JPadyBbQO?@-e@g>kQYl@Ro8|9`6fEpQ-drvp#-KFV1qVF_U>d|n
zsf7s0n<f8<PW5KbkNpipPfUt_;?Fb?M$ezQ@~QWsxG!3t1C>IvljXB$LE}p(1D4A_
zV{vLn%I9%#jHwzVU)T>rMvodN|MCy5oEt^vboqCp6*W#8d%PoQF)+O^?MoElSSJ6m
zDPUW#01JN=9gGJ|F5L{4p+e02g?ZA>hkmnFcI##+8-tq}L4wR`+1=k}AUd@{LzQ{s
z$R7SnJ5vvf%4-_;?4}S|*KV>`H-+1<%#5AbJKA@tzZHh$KHc1e3ZSqc`*u@%zfW;L
z*{_@8c>{+I9(cCwZ_S=tA_v5v)X3sFKHWF)*Sre=@>C!w+~w;wh<q7$uN-J^ASI-h
z*YxRT{)TmvLxx;d4(jHx9<CL`{`{to2G<vUD}EV}q2uLX?+am?YFy*+kV9gM(3f~M
zAo>2#^b8&(2;y~kc38Jar9^TGBLe7AKHR5efd|y!wN5ak{<p2?BKto%!p2lO7*|Vp
z<G65H&(qDYG{4);Fe{eGX7nSZ9YHqe!3woGsg6wR$BH(*mJKm^T5kQQ^b6y!n$_FP
zgy}_cbhp4B+p5$sMV65BHk??;7^z!k06If8xNZ)mR^_WvPXYOP-#0?zk^odLIo5}%
z(n=ffpcedh8)AN(&God4kCo%?Yt~3<k`sI=7R}A+(Y47WE&Xpf(K>Iet!=D9H9P1p
zC&l1)+ye(_hgdxji3;aZI959bG3mKLYwN+ZoxC3gw9}<UP{?oy!YNy&b0P`Ly;`z3
zji&2BQr42`;=h}5G9O1O%`!c5A?r6uSMO6*G0dE9sBUR)TK?JQ2D5G*zGU)5ci&bp
zqaaF(x;bi$)J9HKisN$(mVmt8im65}#lbr1k+u)5N}h!*h0N@No<0~7l7%O5`I7Xq
z1-R<cE2VcEV<_;|cD+HL^vgCV3Ffx9yWsl91_uO=`g!E@e*UtVTIw*sDCsNgpGKz9
zB6!ANn@zA(Pwv+GO_*x`BRxY-iT06f58RDm+o#$r0IU@Xc@&PM@_L$)3@QHfIGvvQ
zv8u>)w%TSrBbF1}6p$3qU8X@$_EB3{EoZu@0HC}|N}?GUEi^;c(pum~lL5hNxiL^Z
z2gZN^L~2niAPa*MsZo*wUBFzH!QB+R2K14WA^tln^f2TLK^$^c`i6m0R*Yb1dgiGC
zz#fK$pn?McVEXVFrpCQ<y?L!7D>|R;AKNBEq9HJ)5#3zhL6l}~VFJiSeP?7hbrB2(
zD$$h3<{cG#yCE~RF4F|A8*O}Y(j^^bjPoZCl67)U8X-qb^?jR+4W1o|OL-`DOMQ%P
zeuLty9zdIPT>5+|!GBottK{5nt$Gg)w*w1?0lma8sLKLEF)}pjwX6iu&I@9UgzG9R
z<GndtVFSgV#8@Ne`*>$U?^-Dr_)usL@C{sELp1`VxiD507jI<cqC`Ag;Gt*DBDvU~
zDKlY6OhUXQo{@ntv<)~`FOB)b8;05^1-Z;cdQ**DZlp$}7VdDtPnqZk<e&nMDN9*;
z#!I)2rsdJ*1cYcaqO8J(TgNNpvXYHq03qkzhT<R-5@#^Co_2I31y;6E7(`Ip8oPnE
zc{}?HUl3)k+J#{4GZ1spr^c5k%qtUP=!9k|P?3B!$)C~oVqEWH@Md+AVr?;`rXK6U
zNKJ4cPEXhsu=Z=y-c7aAI9h;>n*1br?}a)X^*_iMsq?W%l}8k+!fY8)d_|01%Lvkw
zzFg}4ZL$q-2cVn9Xi$h<!y+Q65W!7TkOo`Ti#v6aMq@)1`@z-h4KmqVQe=wE116(F
zzEq~f^4H{1h&L-!6CIf483&AIeDW}8^51PQs1lBr7|+8e%Vb)x1p;`1Jypq-X}O5e
zn_nJ&Tm&XuWx@?X5<p_FPN!{MlU(D(P|q?*$2{h1{TYQM?{&qNmFwae!r_=`$k^BG
z6VYH0la;XS+^jQYdU|^+PDK<lfaZFGbqilSIQT~U!#qYe1z{#3tP>*r=Gcw6)-)?W
z2#pH)Q;2H>S`6svmUsmSK!e=s16i<rHYUy;z#%dL_cm8W7C;R1#H7PrZ%M?zb6^nR
z)*iUTZcl5rNGtU%fs&xINA9qRu<BtBXsl+8-WmJ}FAmTc%;v6ctwn?zO^epZCy<}<
zSH{<_%78oV5&?Jj%iVS?;G)cGxhJNHJ4ewEh=gh08`B)ZGU4|*RV}gE*z<mW#_NZ8
zFBQ%dc_6;3&cXZlU}Cd$cLxSX99`ix`JuPKVjIltp`P-PwSbkI4WRB}e*^9%-Es_d
zc_dEffulza8a8g&=uz^h&8C_zu*ZVOsvFDXar*;<0Y39Ye28GxMmiA+@}&KayP$wx
zJ{4aMf**fI`lDo|%=A@ir=(&1SUhW1Oc>}%t;~)IZQrWM2Qle6{;kEsG0cJiW<c#_
zZjvp_6whO3Hj#8fqmTyDY&*R&UnBFZ;g%T_a@9;E<MBD)8O)D298z|#GySt#Bp#Lu
zl45;GQMGR)?7f9COdx(Vz!*Qa$OqG2C*?QDdO>8{dRZK6X?8h!peKw4?^y|Eg)DL5
zZUv#KfAd^!gV=c5sS_1rTI&7lNy+ViF)d3QX{JJTOt?2bTizyYhy-S3MGVzrGkIQ9
zMPmLBt0S*8qb{41Rk2?xa^}vL)y65RuCIw-#~{}Le4mxIJ}`6UbMLH`bv`gkaA*W-
zx!7s3KGqM08RMd0Fhdkk4OclJ8?3K9^fTEQyoY-v%ByNJ^^;^%>Wpzc1g2NX=6D97
zL1D{teTzRM)9|Qdskp7t9Q5=0>O6u>Wn0Xq0#{Q*X)VZ)e0;Vfs`;QB6xh;HE|(u0
ze~pBqwLQ!`4L!3+LF23TcGC_U35+GwD{zt2Vv~eVaDwcN%~0P6RQ@MH9)_rbB<mIa
zvhruVKC{(#I^!2n=q^;M0riY8{h3c%HFb=dsnI`=88$RRC1(-cUw9V`qoh@dJA}w|
z_}7Yx|I%Awcn3mDfzfP~U&Vm0=_J4Q@3D3oll9-FyMN<rZ)!MKod!5B>icaR9Lyfa
zy|BLvTElS1380Xh>O**c9gd*i2NR>#tyb>+L&%qGdAU3te`KmF<(V{fh!0EzwbbN~
zF%cxF(IQ+8MeErZyowa=TP>{!Z?x^3=VFNPYO+GlpS%qm_ev%zz0cb_JMMuUdM`-)
zLSl9c=z}2g)?TzX5oS<<>!J$sQml1t7U%!VHUYr)I7DBupAbOFI%coNiwFbD%4=yN
zOq(_p;q~;4SK?Yxus4iekaUQ#U=ZD#wrs#X%jK<@ImD!GIT%!G(Sqdgzr5`Y0z6e9
zmFb^<w(pCIP}Jp}V1Znf{Keh{|3bMc$3X!x@K@u-F&PGL*UG!z9IZ;$!DbBRdp-n}
z<X8~abF*!uUUqpu4UtueXuOmE5>P511iQo23dQJeCbGs4l)neLY8WZ|p?3}81FwvU
zm5*ZYPoUTWQ48{qn9#^t`8Y-kUn#F_f!->g#3#d0at|J1V8@@vYrueM;w*=${8{vk
z(00_dGx0E`AupeIYdscP#{Y7+%YWVWB44Ct4JSaXqx>s&k?v2xx7Z<edj30DbU=M}
zDyrmtX<zedb1lyZKau<=C^aaTsjQW+g0QKM61SUHV}KKPp-6WH3eOuYyLC5UNmFL=
z&TOWzwaaGQy}Ln^L_B7(tU8YKJ^UFYuZ6Y4XxOv6!-@6TN>*tBz4!9hn3BgJKb|^!
zcUMEZWyX1V*(d#u62a)2B75KNjsY^KrXc%uHyCpUWc)1IKYgH1dDVRm=<X48<q&tc
z<KSy)Y@4+R2jW}g>)r~keuUw1pud50m8LGrBrlL}*s$Q9&B#H~3ued5br}5f&F%)>
z^8CT(gV8|fmN4zNoPMmW4le@YP7U~@kB~Xo2gihl(^V{o`0M65wKT?gJEuc2XElJe
za%guCt^>@@D{EPP3tG%!-5t<l7!;&>xg4J6$y5o*vh7Rd+x`qz0ydKRGUSM$PXTu=
z!)^MV?yX0gDxB&HW272~0X>Fu!~T))Ca!>?!;ML|kOGyEC7|e$-4z~gnbD)DmmKB9
zv7^v*X$yWkIt|FejF6}$$M`d#`sV${&ih_a1xw=h7vl6aK#TRgzVAI~MR1Z|5my}k
za%_-ar)>0eMprBIxM0hMd}Dc41+N!4-nao#rM%npghWSzv82QjInkdHz#|;L3JAnW
z@of7nbWXE;o^*&|w?im>?<j)VF@3|#9hFYfDRC0Tq_%@zsnd6kah=WkcDfixBDcRF
zU1Jhs!t{`C-5q+QlEJ9H?%tQJ5^X);lc|5ieuE$An=0<^_;;2O;J_=GlHDVA<+y>;
zGgvKGgXA;DLG%jhLrPM4hCR;Rii5FtOh!Gw=GQ=3Es&;9cXcK7M%kafPNjBXDfpaw
zYc6)sCg~TW#5qM*9zn<c(XbkXq>=;Wa&mfm8Q6!=;3_%A-!_H9Zdfwq^Hl$Z2Kf@U
zttJE=si!4+&7`gX4qqXs$1~N!n7Fn<&hTg4p1gZ?7WVX+-m2Qon8h;0P_tyl94U!k
zG+ElpXA$7tA_G#}oVkui2By8qFv$rr$ln1MLJAx3KiIee2`6~~xI<#!75C{YXBmwf
zfQStZ-a()Rit#0Spx%?8`E(Xdm*Hs?yz&e}2*|<N(dKzLTSoX|aVN2SxkWl_rV^jX
z$QUxx&8RMovN81ORV1Tr4CPr~qbg(U2V8|M=Y$9>ZD5hDvC#xVm>JYr8r;of7G_3P
z%DCXcQkeA!N2`}}t#KSpFmdI)m~JWJXJx$q2KK=e&yUG#G?tcbg2Z5<kPCeBRCrj+
zh0z8VCXy3*AXhJ!i_#*lfSO3QWhw#l9WFLf@!HA?xx`;H(1P27weji3ztjgrlVXWn
zmYQR9q()(48uY$izUwbIpP8z*6k2O0Ekg=Jr2N4CAm=QWvN$FX`<SiQ>_g^BmiuQc
zoX=~jc^|3$%g?q-g{=fHtR}#f^~qB`qsqPv`0#`f1tY4Z(&y3KvKq*Gm5<g8B{Zly
zG&RYb|Kc}YGtm*Ggg;#-S_^~J10j>FZ5&^?uEK}PfoYfyAsq$JfPJ9N8gG-odp#sQ
zi<_l3Mh)qqx;$u{4<4?os*}3zt%^|-A0RqSliytKemfU3XIJ?8UlGiQc$*4b(+ej?
zy>ae{(F3IrQWlvS@K|Q0A^ORnKyqX7bs3;^0E<3i&YY$576Q;x52;M?5t}1hO=OvB
zwCK?bSLQYayN{kgqZB6FG(V-=-A2nZ9{dAo1M#bO)ykD2%Q5%@5JpuMu1XwDhndOg
ztJAY(7!(Ewh3%S{N_|yUu8k4EGmFbdz}?r`E`?8f9seiS`+Awvh_)4#{Mn8|tdi+z
zJ()|R`v9}oGZj8P1RX{XaP8t;!5nXl(T_lQ3T|lEx&b%ENO6Esk)lO=b3DU>M+KK3
z#L7&>oxx=NTb%my02=f13rcm<z;1OhiJ;co{3WwdOzknx-0jZlI4Gfo-eG^S!adTB
z?zDBTQq`+T?y{AGH8@l54kDhFJx}hjAF8q_Lvn8#Gkt($;>~hjJOgY&1|fofe|Pu!
zz_6TyCYe#QQ67k=%<`z$%mxtb!Ng#MjaZ|IM<}wf<cEp)ZB9TMD+Heyiu+J->WFL^
z0Jw+aPXI+Qs~)jQTz2J^S6(cSdMDhA)cKe$O^?kYc`Qg?o+;F^M*ri{%vdC;*UA%V
zdc-cNfh>_H6Tx|Fkp>olcNy~jl)tU1ewYTUpv;KAx37gX;FS>c+OL8>Ff-cL8#K&?
zHY+%ajy=F34Kmv%g1<qnIdROC;bKuR^jvS;1de+U;E`D_&EB|T%~+QWIIibKhp<Ks
zwp}dqqr-}N)@8;ywIEH81t+Z7Gd16>ve18FBomKvggE<(FEZX@7z6~$V&fSEjyA}W
z=u<6dDxb?`sXsH}uRax7=55n7iBu2crsd8Dh_F#-u&I41Vn9};-w7}+fO5r<qc-?*
zrHy82zAh`P(hSsWgv?-W6$=ut_AxQCj<w7ftF<PLldd~UHCUTkXv>!}2gW-8g~AM1
zvoPik?XWoIdLJQY4~kLE26NbduYjen!385n;^8+qHjoVtwz7M&nccI|=24TYXpl|a
zTUlWR`)UpTw*~)O&;ORq#w_D;kxH#>i78<aBq>WtZnbB?JdjuHH#2Taq^Bc`njhd)
z&1wEoY=(0haap<i*hZkYCv0z{4bU9&-vZfTzqia7FhE-3zb6)CXQCOy(TLCS?qEua
zpLi=g)wq?I4A2_SEHWFR@+ivZ&#bze^11vh(Hp^fIz8nZxQKu5Zv$z-F^jRmrus$j
z`M9bK?(9GnlwbO)7-Vkci~KA_T*DmsRZJfC0%Y;8W9vb9v1Y_?Lec}Upw*!sy5FXm
z8h;U_qE$RseiuZ>nCLC!ampoGEWh{B>ctuO1s6QaAJPm^Q3&E<HFn?AMp5L<733Lj
z&D0j)jC7H}Oh<;Wf9V-XJMIESpS7XmViUk=P^_QxcL;Sb{DcY9e@Y`?2^EM+4?$=e
z70>(IyuOl!{!y?}UNG9A!fCD!h~SI<3Ty)mWG<p77syMAyvyhy#<r$qxQXs`{JTtE
z4s8#RKpj}I%<|P<F%GUpM8pDlHE2Gl5oLsG<u&g|EgRiP8{~C+XC!nP;0hUeBMsLQ
z0dVn)yy=tU;mB+v)6O`#w}LHj0j!m6dpi+=8OH5v-T&Eo8U<IA2FAP-43Y!Litt72
zm~dhZgK1ew`>);@^>TWmwdw-n;pAOc$TCypJ^N2o_mcM$A!k!_2RJcbK8S8mC4UO=
z!k8bHjj#d*$lr}<S2QBwQa((CXMTuD5ZgCa^7W%=00tq`T9{V-k2GoCy*wYMEQ5nc
zppRn-fH<O1l6;a5%sli%97&)0r>vU7kgrv`kty~*Glqrohf01y=W?XhK)Wx<=RQpH
z7tz<itY_xa(2=r6zKAcX^SJ4o2h+cN;0B^lWZ<G(f5xj#P(h3T%9qh^RZjk6wC&Ze
zk9_3=Rlhh0$BG|NtYAY)k->p99pQSy5j|gaD@x!PG!8qdlyPMDqE;(jcXk#f%ViJy
zv((cf_DtW=_M*=v@@Dpm?}ngAe?j&xa<r)!MJV<uGQg5n9P|1u13yI*$-b#o-XX$q
zP=*lt*w4pKm4d1b{+0dx8DmFjzOgpb0p0_vx8UryT7!^6a6duRuavLF=msF7eTsbD
zU$>!9Gp378br<*{mdk-Q`ttfFTo&<-7#$>STQL__;5?|v5glZ|R%an8zUi-qRn&Ih
z__ut}foH6TP&(L0(LSRF+(FNz9@>W#S%ecEu!s8h<Upc(Aio?|)XFO;p@z=I|CY<)
zHtOtDh;aF~eOv{fw;bWaGSNVZI8XV1xqK(+tKbe3)Bz+$SAba&u>s?DrX1-*h1nT3
zM=(gx7G#6AB}e%;nl6YnA!vKFJws?tHjn6>x+*}R#}qjzO1~r>Ao9H;MV)%~>LK4R
zQslU0hTfZaY$7|T0d`oV!6qzjb6gCNOREdKNsjktJbIX-00A#2SbGd&gN=A%S{te!
zu+bIMVS6VTsm9>kln$|^BXWoigaGImBOV8p0?}7Z(#ad5mCF6J34+wl`_752QY8fZ
zzLFtnz&WmyF3wiSRh}wceZ(y0hfo!aXT5Z@kxYWFD&4&sZ5s3EeMm46aP}10Th&<v
zi5A;`_v#Isv`3NS#5)<^398|=im$0$VyIZEp58BG%IU6Hi(|D{Y!A$_zFvB#dEgTa
zjnzXr+sGv8<4q88qMLmoh}hSO$SmA`K6K+Z1^k@f`p1{9|1E=-bh1C=d2Nw}TjUgD
z?UotlnUPcd9lHQPC4`*j&n(hPYgWnW!Kd23EpmIS>J0ye*$0e9Vy4@fiO}?=vTf#`
z&=*<~YY!Bx7Dd!5WI&7qzzZvCYy%`TFxo_<%yppFLddN_?RZ4*$Wu*;I^kgdCBoF1
z@uVoM4Dogd?SO?DFC}Ni!XSqLklN5_7PV^Vo?$Ez2fRdv6}c}5b^yaD02K`{3cmyk
z5i_R?W>vME9UKnu1pA^OBaACEh<ihsz#u%svJjs!Du@F%XI_<;Nyqs&I>rK_PNNoK
zX#5w3*qGSILxFS2Inm-khR!s|*kCSq8mhNKMH0jb41v-Z-@r@_zOR*WHoh8MLewDV
zI+2LWXrXY-{k$UejCDUl#wQYL<PI2jrJNuAV}esMO>#l72o8zxo8j2NUZEdgwp<wF
zgCvQYGu6<+>}c^`6w9s2YIMD?l8b}-@U>RJiwtpdiN6MEtI*mCY`sfuHOJKF5$%GD
zaD%;SvJgEevLTndFf8sbnIPgwfVNVWrj4QUYy{ty2X9<}<EKh0{1r3ttjtz6dSJ-|
z#`ektf6K7~NYJHHX}w<Bse@ESulXFx4CJCoz=O!o0ME~siN5TH$Jh?Xcy`_(ll*N(
zX~C8o&~&vmotrdCYD_>hqM(7)#+vA*lr@sK*TxPWCv~>ksOk+!<_d4ppg>4SXwCI8
zL?qtVNFluetz7rQVrlSa43lOJ<uy<N8?A>5tjy%}Gkw=;3&<2}6id8i`y82SLq!5X
ze`)f!nHw_)5`%#&(~2CeY^$v8;+F6kVURX|_?58)W2#UUopE|s#m1o>W`W5V<l^ls
z<?2M+l>C5S5L+$Rq%rdw1`#5pFj_9J&g4*IszI)^Sq#Z#t1Aj}eelOvZ2Vf89yEtG
z57)fh;7wA9(`@Ahl{d!FaDdair@doG+fDv9gHUKiFW$aLZZ2}aR4rWZNa=^(hf05m
zlB==#Zb_fn@kK1+R(qp1GZ~l4$ZfWH2bIa~{w_F&-73y=hrQb}V=$tD<jzE&2=9+j
zs=MO7p&6uF-5pz8L2~dY*nrjv&Mz?h2D!)E*%cP2biUWeKvRl0hcFWEK7Yn^1bP)%
zo%w#F?J(U|)$%~l7U8O_Z~S0<qUbm`$#VIjKV!N8OA;VDK^}@>Ps!nMdDw?#1hB2e
zVK0?OjC=c_uJfb*k{#@tHlRDRSswFQyVVTg$Cx2v5<K`9$m7;XQQzM3g#G7)+yr^j
zw<Qolc`Bv>!AlR$jNl+#Hjk9`nIYu{Ldt2i8<}OhwPi+aezMFCvcjXnN@dh*jC<w;
zSs~{Mts-Jh=K3>clG8=WQj5(&H#|c%Sn7N-FVT(HBpF6IKL&=RV8lNzusI-vt{@AY
z3K1=(tB%$5SQS|mbA%T)i*V4zv5P9Q<xuc|Ixg`RSavFkNQL^v!Y)gFn#^Wpj`<>R
zg&B3rY#jAD5No*)n-T%LT(#&GX>8OSsZlWx&vpj6Z)LQskQR`1$S30ITVz%GB_V_4
zWgM5Q6M2?n%Q7Z+xvcSL44u--nXpyXdJDlfv;j;{Ul$Y@fD79Z=v0vPv9WQSRYEJ!
zy09DKVTDx%*I?R>-V@AYg@4hiY%-pJIt`!EW}h;#k*Hmz=O`vl18+;L<Pg*ptCFq$
zx|+X`bBv*%iyg<vQX~V(Hj@^Ybe11kFThO@;QC|Z=l=cs$abG6qXMZ#A#YAM+j7|v
z^DH#hR;x16;_s$Y3<xXM9Q%8x&kKEPzAxmt{A8EaGqi(WVkMb_E3LNrh^;EfPyKuQ
zU8RZq%r-k_G81RW&tvCQ0n7YFbO)A;Q|y<!kk*c=U$vW@g{((K#9xObDdegr$#05U
zzX!|EQ{euu|DXIe1~D8VS-<lR!{x<riQoI2Jw#$BYpGJBp_hVo;UBD9m`aovPy5^8
zI}PYu4ju#Y2^#_u2)Noa)*T$iuy6ht3<;6U{cj?jmp(rmvjlPJ;H8<b>V`b$77n)_
z#TtPGgE~ywklc&qPi?~haEz|$jf5NxGwu248NU!zrT0rA5P89LPs9SU2n2M67h@<m
z9vN~#OUOjpOF_E&>=p8|zZSr!A#f~q^-5v{bMvsC<g5M+K`)x)vk6=FHGf9N=bcPj
zz#^}Qs2YF@G$8LAq0lpTt&WIn6>pk+Gx)kX+c-(@W_ZgQQ>?BI#d(=s@wUAMloV>U
zLR5dY=6M^VdLiGj86gP_<!=6x#)D*6uq+Ft|26h%Sq}FP%eyhW29zQMfc0MdJT_B}
z$jSGEbb94O`_te9-w3=WVz&G(r~&4J>l^q-v;4itU2G^r0bbGZ739O<@gb<WE+5%n
z$V{ck`bVf_=7<3_WS82J0SSe*=Z}vQQ!iC5CxO+zw4cP>0TC)3h0XtIOm?i6Sdq_s
zcIqug>+5g`kkeMkKZ6XsiaaZyn`k->e!4HbU8@5~y=$BND`rZZ7<JTbDgGTCjB^ml
z6;QkxF9y|<;tA^Dm$Br75FJ&WN7mebg4!B=W8qc|&|M>6Ikf>n04S4UQ(X-_<A}Kf
zw(3UZJ|T3*+)yH%yA=mWs9>Lla8YV+k==_GwCPvWSN142Y|lWfs8P|T|L$4r7+H0`
zDN_y3+N;=69@MApCwmvW-8!+cLW?vMWS@8hCa<PvyI%J7PB1SBbv-#i^Q{WUA=6*>
zD>jIeyO=2L3iyKdw=oWZj(I@(nKA6mI2C~i`ZLNgn6La-zMkGQ1ad3LcVN5+M1jB(
zRATu?v4fjEihIjJ{sJP4z-PFpj<JU^1&BJBI5fPz6?B3;WKcXjGWeIpxte!Q2dHy!
zv9kQ7yxOB>b$9#!<d8HX+*Rc4d8oHUoM5G%CMqO#bi5Dyuo&H#{3JQNxYgj?O#S(&
z#`5i0jSQ?*MkyYVI-tz5D=)wEU(aA4$aiDaQ10eDIkH%hNvH?4a#YN9T(-OlN);@e
zqn&Zw<EN2kCN+<Vv7HUq{`caKlIk+~e$a=pXlYguholwO;aKYu-nUXYZWoeU(AV+a
zr+F3)`O&CCIic7w#{%>_%`H07zef-s5KL-82o@|h@C+VSqU1P~n9*uie!&j#mG2Nk
z9(+YXIu<MbiX!xl8t~M~IBsCBUOEQ_^l)oe_DZ^>zUgn82i<0?bS-Yx8zmTX-0pwU
z&E{Os3*NfNsylLqQ(A2i4YWbhL6QG%#~C9eAo|+H6BNf_2W2K9m0x;9+toSg*)}F}
z5x$)zy;6IO1hyb!Za+)KnJ>NlcUmCKP_Q6cT;vQ&@e<BH-X_-0QN2~tH<+;mWo0U*
zpYfD>rpl#%46r)W#N*`T;5bB3R?8{75LA;Ru{?~sIyJsS=Eh_Y>a<v*;(op5bpMPV
zB2NfcaXMShh@DVZ&9Jx%IWtYTO;c5C1Cz&;q$EuUX%Ehn1EOz){u0|77_5dXH&77)
zl53Fh#7N|F4UU)9#abal;_rdD(Y&Ti&aw%^uwnq|(3r-!e3J}I#O4j>Dq}DL`l2iq
zeYlU!Qh8NllCxv#?aT}=(^eS~w1k&|_#w3r>K$ukWQ@NOweAs4k846=keaB8#N`;)
zuu-Kl#=BhrX+~}&W_pf2gT>Bj-vAt8V~zFUYn(+K$oPgx39tSx^#Pf4o&$=c$8kmz
ztl0{P`E#T5;=X<4JpT#Nq|i;qr#Zopf}#d{5_4iW#`!)-Mi#No05QG`V(~$Gt*U%s
zFk1#yN#!Ej8z_`GNiGi3Vx&}0nDcUpKV!HjBP?LB%cW@!+Z2@baBjSPng2pL&*BL@
zCOF6Wa(VO$r6Mbt*d54L1hQXdgUH|t2Pt>q)DBo;6DsoC!VM~-xgB={GsxQaKxJxW
zLJR@g5rbZMJgZ(Ll|c#Q0B8Z<s+cds)6ogi1te!ooT+M*iHVx?wIytRnG{@uI5EYd
zq&nWBuhRueO}vHLA_$N|VS!p775#5(z&^Zq7s3aQl6<ivORUa_cm_SH>VoRbUc=(j
z7Qe!uQKe#PH`SK<IFL{~6p?0>C51G9ixuIh%vNYH(aZ1>{Ley7?!^h-7_3%<O5&5#
zKoKVdyL&p&1bhz6b4ofj2tQg5npxO9RhGgYp9KhGYBW|rU`A7{QY|7iWLk0Sp|I3i
zX8ez04PCp+m9aSEnUjWjT;&6?E*LRqwcFRq)xkU!fVq)^Y+vRrVjPaQYlA522$;ec
z-0R{K)PDjReZ9ABhHW@Vc(K#;AUxy+qg{D1!ws>T#1f(56y(O3WVdeJ<))bAAXYND
z*?OW)MV;dZ!5x5Qm^utMirkV!1f3rSjhOMQR-W8y4KkV@TjjQ5cWn$tK$e!-*7{zz
zd&6WXo<nq!-jO=9AWUM|cN#yJ3>zhP#oSaavyg<l(@6j?EjV%`eecl38SH}M7jjRr
zT?x*E&h2k>#<(JE7fp4y*o8L7-@Qfvf(~59`-0SdF$8bmR(GA;AL3TSZl=lurWN8|
zlxkLs^90!kW1f&a%*mWXc+t9;fO2!p^m@pq*uQs?JnU~%U=Rgc0~7m^#7J}!@)y+>
zu-K1A+wea@7eHpyMtLk2TLV8R(I5BktuPg`5V0e;fmg^A{ylD{2mkY~+V%2ev<ZRC
z48|3B_LM&(yJ@Ttp3Sg9_w3tKW~LKQ3m>Q)(dc?i7aSq8(gY~)8N36f8o<N?nH>#7
zOweNm6;9?@!~KeiWp4VJ_^3ir(j4<cdF*U0?tPSbzF>&cnUCbY`O%vu=EhpD2ZFmG
zjWd;mGsyxTMv#;&j9oDS?u<pTD<;4Om&Gv>uo0tYsJkzT*LxK8lcoLwL+a?PM9F4Z
z=FbRkvNko!Q^<0c^o(d;VKc2nki*L84Q$gStNcT5Dnd*w(Xl!;#&Re`%;w9QHs)8+
zZPrm_6D*ar_6@KF%B1q#G^kGN+O^FxTO8<m>m3WQ1<7k0Y+eZYL=fsmBQ6R6<QdDl
z$-mY+*C7s=*0I^%M<4+1!QB$e<55iIvejCEJVAB>F8MIw#)J$=BQ0<7qyJhlDD+qn
z)=~IztQasP7As_0y27(Q@a>5xglBnBFOePojAk42`C+z6OT4Fv(N@u6XFRinKn5Dg
zPh#{qUsmk67p=AW2&ly}(3JhqW@X+I6Qq9Xt?>GCB$;U_;m`aVll<I5dR9WIH8J;d
z`=({aS-kfIhaSlC3#S25srBHMrJ7wwuKZ<Chi#(;UdC<9_ABq(JZ|%luQyMA?R~c^
zR-z)qAoc6yH~tNSLl}~^28{gM^bDo_IXi6c-^Fo;QmtBu@Av)^4SFsG)B65kD}VLX
zS6_CeJnb)-Im7+ES)OsmB5JWz{@B(8*xKyZXQQ9e2JYvAJZEhVz<qJ$@+bSFWyVN|
z|L21hvvqkuS}%BO^I8}Rz~t)d<wfgn2yUvxHhRe^XHJ&(sMgD|&SMbU+*k=S_!V!>
z9Qt<G@~dgQR0Gv(F0U209tIuYf6)N&NM3j9tNRsSEW|2OTW7Z73dD@OVe`Qzhm-G3
zAD+h$v}3BboNU}*bLH(cv@K{zu^ahw5T-seP2NeASOQ8wK*5FZm&8v@Cy$^p<nPK~
z{cT$(rH#C6&sgRoUx$j#wG|n8FTH2QAebufzDcaE0LVV@4on3xCbuK^Z~mddR-8V<
z0^_V4igOjF)`zi{RS04!$VdJOvpS$I!Vq66|A;A8KvaI52E&4kSPD$>lW1cgZrjbu
zr=c}vYsy&43GDNkkI^@x<^|9bvq1jo?J0zb(-4Kb<#U@{A4u~rY`BmDOc+=z|4OfO
zhv*S1{|>#kj(3y!Y5!%+6V7WUVf@G19Ec(<Us)S)zhxVypwtJI3+Z847{O}TKpJhh
zTMtLsGL2J-_d-M5y@%s+h(_S;cCv?mw*Yu8gMOfp$eukM=$ew5Dtq}0x(&e+^yTi|
z!y!T(v&5pdQa=X4m&-oUJ`xY~^1*$3C=!MHauJl+P4@FIc^(mX;;rol^6%fn0Ya36
zZjb|_SJe~Dr2CqG+r9)Qe_1A9?_nWxh`E&m)0apJbTRsk9*zZ7;@Ub~OjD49?2DEe
zLx#vV{R;=?F)}0z8wmilqaN15JuFzQcTYK_hvA;>a$d0|h#L}tS}lhr%2>6{<p}+>
z`5orJQ1BC^-l#$xM}ZvPjwbXEGVPFd#A^ArjY2QAk|S)Gyj~4Ajluy6<l#Fu4J5KO
z$aiDnYUdezcBDUxVn@hJlcTIrq+p}=$kB;KDtO5RlfjRPb`++xbLe}i9lInI6&K|D
z@o9D{GFw%{J+_BKW8-m4687eC@i*>9WlYcU-VC65x^R$9fN;Ik9ZXZqn@Oc7CMwbs
zs2ZlP<VigYuLI4=u1wCuze!2wT0nE{;Hw1N8!8m8=TMuL97ldxDjm}*wxz7`pj7(~
z-*@u<?MABh^>;hR))-lGzI2J{O#$AKZIG@$6?VvI37&r4Vk#|kDUwx)7BhjqyT1)(
zSzX1vGoq-6#xxB(TZ-FFl<bnAb!8B$u#}=bg0V;CP|ln7SWnwVmZ2%VVvT@QK|x>z
z^xiQ@7<9-^*v^tveY`X3g4mB<ia=Ep&DAg_=o??c-bPrn+FAW#*P?8AUit??>R3*t
zNlp%FK>9E~_!MsfVv;Zr;u`16sqqpotYv`IX}&tU<d9RkfP8cmo*s=r-J<>qKupfC
zE!w|#Z#gq)0ufiCwLMDw0|SY%wP3;zh$Z&0yKNYc-E}x{muaf7kwGye4J|PbB%Q>Y
zK|ofp9PF)eUKEHZUyH#K>=|P6uVuzSTyX-l&lk8-hQ<%oxtcO$0$Z;Ln{9-z4)!gm
zH9SV*Fe~HG&bEmT$DLCnywiEOu8MdR3#QmgjkK3?wNqu3w}GP*RxQW0Sw=^@h}hWd
zqGn8bwwmBLNV`hT@n;lPhSV&Tv9aKzO2&<r(r8_yPBlzZ#@REZO;zZM^6Kz&;~B6b
zgvlwebDqhrp7=i*AG-t_n%vHR2^+4K^WzunCh9)GE6Jd+!7hkRu6nK=Pp0!)67hvT
zHa=qwDI^`qMMjujkmMJ~M9!+u;!fa1%NuZ(Q5Hmu(WN%b9MSX3Z1)42LU#1!{)uVw
z_!ovVo#+cB6FtK~hNceh{%R=;!7;w1RLcDY?l*JjFOiDGV#}}qs?u`MkcoP}O4h~;
z579cPG!tU<LwP?yWg-&G{HhGZS4b6>s@M#ulVW&UF7_xgFD5}oPfUZQmy8zT3YiqV
zS%6j5Al3egWdPz|*b+4%S*kNrSThK!SFLU2QbnJL_~c_}!#=BIY-(LR7|L>WK*g>w
z;v(UuPU@WiCVqHrC>>l3e4(P%ECm}>zMKkDxika?Fo0TV^dXYX>9t`Au*oqqh+F#G
z5Ii7LVi`u243epd<jkrj*&;cOavEg=Qv-ITG|Qq)^JjbmQW&<$I=RxxPXNlI<Yr=5
zF%A8yv@ljN&TKveN66JZS~d0AX2{g<Hm4rb{PcU>oAc$G9*&D5PP;lc&DQbSSjR$s
zLIbdw>tbfeERpNe-eAfW$I5Cu>p)Td_hOkI#KYC4wQ@t^<i(1RZc!+k((6W_ZDcl3
zUf-0cLKl(Qk2lBK@p!fMLQhBpbAj}OSa`&l^VS-<B?eI3zgJ(mH4WG9*<e~a(WxjU
zahrFUP;^9V-fr{Hmp3(__(ehP2!=#~-&|!QmApHRyIN*ke3{&pM!ix?lCYNC-9|FV
zWzg>)e}U#MI7m0`y)nuudBhjqmll&5CUmN>9udX={r)YJz3q~|`SO5uK8`lZgE4(t
z=#(8lOdp_X6}Bp)A0G1N&BQQEQ|ng9!?B6$VdBHKt&m4zC;;kt=?x{EWj{Qceql-}
zbBAaRFp%jn?~^fiREki)OdkKQK4&SOwhUJlJmGKi#Na;UZFNuj+lp~1CsAD`y7H7!
zvb3SGqDf}>r=|vh;eZ6qj46yS89TOQ)VK>|R%(ns6EaI=I?lG0-EkLSg}VR}&xt7k
z?qk%I)xF(>?KL+IADScx0F?+^hRJcwX|jvd>Cp_!ab6l1{Wn^9LFUH_mm&077DP|x
z-2^^CjS&|g%0mB^Av(-2W2Ns!v8uV6n#S6EK^7-@!ZTFQ0#XV8ZD|Tij71Up1p#Ke
zcxgylUN7H8mZf1^kTP+Y<u*jLu>c6@3Zpa-$`u(|`G1&u&+xpeV{Lfoy>~+I9ov#s
zTmZQn+^`I$n4&J(k{+Q)vMtjgB%y@fLYi#jB3HS08?Znj1PCP5K<E$%ge3F;;k)lC
zOX8fI>w5p3?|q#sLfU)nRc6haHGKlA@IC_;BLaR|>Ojq2fW>NA{vS@4>YV{gX46^`
z?|~*zo#KjxnF!X7$aJ*QFu56WrmV8b41#+RO!=$*OMXLbu9*}qH`sdc(QD#NjsW#<
zY(Q;MSsQ?aL5v!I<bRWOv5nmmAmR!}72p6dlzf_kWP{D60pY8%F(}}f@I->w)+Xbv
z`d%pyB`BDdnNeF0$KHZpB(Bgyl(@N89*KUC3mXyqvf013>V=FYTq7Me?f|;5J!&xE
zeTqBnv1qMEHd3f{(&=L{r-K?<lcxVSY5AxLkK>6%q4Kyx@%s+IU5dfq+SpZ$Jb%J|
z?`Rz~LVjncz+(1Uc`_DZIs|#j?u=5L2ROn*)wuwI{nI`#T{l*bG}G7uy4(KV)&w02
zjpEM)UU>$&tUTKVQ8fS(CV#MRke#0jh(8)86Sv_NF3$yGXQx0fBhSYbR#{SttRjZ|
zFIcCj)MyeU4d5?YryZ^E6Y4E5xnN?PJL#C0?ZfG!3BwTh<>2ai%AaDMpc2Govdpg-
zoB%VZ$6s|2FpQd%lSLBgW3^WPoDhUI9iTMMtx>nv0?kzlbI&sJBmgeO)z<^fy^Iua
z@r_tU14JKrGtEl10#L8ckhjt^llKu3ByT6CpvyUpyiCMunxWq`E$;*-)J&6iQ(u;i
zbvpWg&zEDMi@-NJC;lbik<+V(yq`W@1Y{uBwM0I!1{K&#`1{#7D9BYIA0`l4N-R1z
zs)ABJ@@|yc1E3pWZ~I&7y4{R5pfw){nwDf+YEkp+?|~Q1^>8Z!y7AurhYg8XB<S<J
z6WSliRV{~^;1k<#B(s-KV+?o;F+)H7sXq&%0lG}beEHmPk(>vs>wFPi0`qB1VbZ5B
zV+$)TYp!Wx4Q%<!zoNtf4)b5JcB-R6jH#I*kFr+2PW>@e6U<d$kG~1@A5&5fLiw$K
zY&UiHZTxpMM0x|v?K|&F>%Xb7zFj28v~A|i0t!RKS2NQVIj+~5Qd3(Fr@0E*GDp>I
z9jy~l9Jf}s%2CZ#3vh7joX&03<UDlH7}+Mrw1w;vCyE}mvaJt8xyzhpsGx&_MYhXP
z)f{wquz|?-Iqv$0hz2b-rd%@>dFETNLrzlB=*_40I6W=S&4UA-Cp+f2BC=v?Nfo__
zck(GQ0&YD9#uX|%=cqo35hm)EvP-mqyB$rmydrkZQQcoh>)0`}o4<!`RP{bIdh8zW
zhmMy$>~~zi<+7)@rNVR-C^0Sed)at!?J8lWAlW;O2l|QKCabeg+Kq0vrQ+Sb>1(*-
zqHgwn{xx_TjRk}HT=uv9p?g9T{qSsmdiUrh2gc6PC1a}mz=sKCCsm9lQyk>*2MX#%
zmM#h>cL%3oqrPan9AXOxP0f=-V^LsWX^Wt_{4m>L0Fh2-hkFa-`qf)hj!4+SbYsSd
z+i^S6)-Ymlu^eT;QeAy?fC-EQRlKy++I4$Oj8DeOkL<C5oJDmla$Nk;t5>cZ-!-$A
zdO5+@P5~V(9KsJMvC-_$4V)t<df!N2j*FK{+BI^LJwvW(<`;!rirJkUy*6dwsBlWx
zjPMnz8Njf9YQUdci&>)55)K_5O?t#OL&ePsr#<~^5E2H0Vm6DVm(P<dQ54+mUa=}C
zy;ql0oDRzu2KJV}ORfQOFubjL2Ox1&nU2nCH;~Td#p$O*CZe5DAozPzp#gc56vRQ(
z&}Jy*3UgG)c=Qp{C$_6z`udl~hntqapHI@1$Hw&I+Enog0in|0XGV1mHK!X?qm80I
za$0)Bm;rJ@kmMA2r`wRD29K9BoT@-)28&)%>;XN29Fqt5Gj}u8-5hAkVSY%+lDN<Z
zS)-0tMC&$zRx?ZD;50s~52*S8fQ6&E4DmJ1=@#Wf1ZQ&o71;Kn-C$Z-!&J7z{EO~6
znY-JlhR67H<;HbGuirBRg9nZtA!pe-FVTN8BCw~JDIKe1qyc0+Rx&D<3$BpL03_~`
z(f({MO|p2#0QWK01rk-1O@c%yXUAw4UUZU-^$%#_W*`_-{uD=pUhv!==ddvghDuAm
zH|N+yk^Gb*>bU_!NIW#}ZX_fz{tOu(bHbrul~klZk@M^ss&pgM0zwIDz{;JUdN6}G
zR|OJ))))9299qZ?W_n_#`1x|7<HzXnGQp-(%<?p~a#5N#PZYc1?Zo_47h5|Z01C;;
zC9#3c4FFIL)KZuFba943iSQ*ZmJ*|f$hBWArLiL=wWYM9l%<h$42|Wp15w@$wv?)w
zPQAjPdH5YR4JAVY-h*q)r7Gw;3bHj<NwuSs8ii~&z;-64F=|?qM(>^=jK_Uqx6xq>
zR?Ft37^$edPHN+%BN;7paOF~Gz|85@Q|e<D*#?HGWdbG;yP3@oHbZH!CNwCMOS&uo
zsXB{BXh<6UBUXK2DT;NFU9vgJ#8|G$rcux%r*k)DG=NBy21s!6b+b+8+@i5E+1E(1
zk^Vbf`BA2%H8RmKp}s_>rq(o0*v>0cEG_;9M_0qWDE?C;w_2uoKN>QNZ<tZ@a+^S5
zL4jNmA7XQ<I5FD-6<$}S58(uW3zr!!u5x4<G;Ww&ZNC>m6DQXM;(<3t+-;gBv7OU|
zDv*I)a$UfUZaWx_h3Kx2g(_o21V9*WJi}KsjsQ4vgFiDKg1U(1@*~^y;iJwi8X-Tn
z38_)@C*GL49ldjb={NebP$e$`5xgld9`0uFSC*R{CyhTd(cgt%o$M_!F#w8&2I{zf
z>d)qYQM(QVgb_}+M!y3pU^Oq7+j2VT;n4BZ|Bz_ARc?=7!B#O@0(MF6@MpX?>Ffy^
zOzw<!;663)0n|x%C17ieN;jwCyW>O*8+fkVle%N3BCxrD^1V48WTa|p%A&;1`@ALC
zT{5iz;6}OMmP5y)T2@#7nSr`jkDMOzK$-wcM(`|HI;b}JxgWZ`Uh<1{bs&xk$V5+9
zcyQ@82_=jCG7c<m8VJdC@++T3l*`+IK&f9_e*<xDTGMhM#-S%Nl+%3-OhKcXG^`Zd
zQD|moBvhv+m_{O2&}A;7nQ6(Iz)VT|g9be$2?WWVW%ymkOi<-A+vZULK+N$ksRlE?
zrCsI*KvW~y-MrlvB%)#}@+3j|i)h_gI181~bTB`T(uko0#>s*pJ;>_?s~K*|h0!}p
zpn>*Xt0EA&u|+{`U}At$3fC_PPF5B>Sd`pjKS^LoG{~nQSV-Faz3~Q0AOO&=WGhS3
z5}`uHmrQW6G(DrzRku@?#WP|I081LO+$XCR81(8Yc%bbHf1}<lHWU1$47oDUwr5_B
ztcp%yjG=~=qR?u8MmrPzQeo_qHU3N^t96kZWo_5Olg_`J3?b_b(wHh_eOYgNGcZ%n
zvcw9x4LO~WlNHRhCW^rw5AuO)!nV7sBFIKxk0o-p6cp-OC!1mks%K-S>BaI;ni@6x
zbB*#TtuwEghhy8R3qVHodU+&mJM|6B?`x?kSIXvY{WDaKyi!N(Dfp*wlOuB(sd^s`
z=tCrN-KL8Tj|D!0rPMHt&Uk=QRWMf~vN!EdelK}E?L|8cOQ_?tNq!p>a=wcSvM2l*
zg?T2KLoreLUGxkG{i-GuX?!wn=5h$^wenP)gfkg?|8(kg4b56WmD%!pUj|4Otjv0r
zajwZBJQJ-+wK1<{aIAaQj%{^Qo%|uaUPfrDzs?`MGq6%1QM@>4*=Jdz=Ny{YRlNyk
zfaDNY!zS~5+D+!~p>r%9)EI7SJ{B*;;$g#N2Fn%b^F@CK;s++5ciK96$q{l6*nqY2
zvbTU^6fh30z;FI3EqEpUjKFq9sNyU3Z5=XQKw<N0*ZwDs8M)l3PxR-2^pFfPwaRNT
z6tsU4-urscs)0a2XqDlK^hRLiP#EUR$r-(AqYi??23qL1(mJWo<_*0>-i~J`Ezx~c
z-tlLQnO7%1B#?R6ZWN^LT_*3@M&;$^$zK9TimEUI=J38lK(hm3mo`G>Le=Cf`5?Vv
zz8TG2;mGjUuI1r^wnB&>rg<@h+Oi#e6qojt5}<y${4JnV-OLKEAE#;4Myk|D{+<AB
zWYzetuBz<+2&BTDK3&Nii7y?U{~4W^HJ8GM`-!a?MK9%3`?X|RjeHg}s-!<_lYHJa
z9PVZl26^4l&dc->U!+w*8H*|MTKv+VK`KQ%<(}j#e-`mOtmgTz^z97Ds7#+TTfVku
zaM_50T*dND>W^7dRS_-U8cnEz+f7#fojO{B%^~NC7~JRx0y>3B_0!8$b=8C`k2*=V
z$Th_tFA8$DA=%83Ev;2Y>-o*HRj%slSVV+%mY|z#k*(tyqTj&6G*gwjjeUdcPW7^F
zu4?Q$TF0PLm~3Zn8Z%I?Y#(nb>L+Gohg@ghi{KXrUL9FAc8sCNm&i^&K|-6#a^?dv
zZ?>ItU46#@u_mC+F3~T--|A#n|IjW!noxL^uzucdxu!iNF=v+kWg6wLiBcJhA-nr~
zGSECGS~~zoCcI|%@SgErH^OY0WYhPIDKYT29%aq<G9Ur4s$`+JvFx3zTF{{gt#6Ti
zya81M<fkYr?`yqOK$M5S6hbQY^G0Ag)Nvd>@v?ueYgwBz<*@r55Pw4Pg`n&M0~FOI
z6?O6h`wQn7@iKCdPv1`}I8riYB;E(-s;V{@=I}#o3pW1<6FxN8ZU1F05X6wrn31-J
zS%Xt9y8KExyla&>9jdF5BkVl{u&f+u<5PJHHsUB7rKz%Bj<&rohe5eE)1Z+w$JpC^
z_^lori>S?Dg~#Q(q8Xm-GdJH370>a3Ig{!$Ws|gA#|b_Y=x@=84xbC;M4L$!%9Y7U
z)_Yk7o^tjWmifcwl&%Fp9k0<%a;j|;*Z>IRJz`_Ma-j6gbrv3ci<g^+A+A@h%bhrF
zk-h}5%JJ?kZ=o?$xz;^SIJoNGz9n%Ai7dpfd6F0VItnsCi{yL5S?c&jeU~}(t(Ss?
zlzgVDNZ-}Mv^bj4p-_d6sgJkrCN<dMm%cFqgjk9?>&Pb|f9EXemqt)8MzuZ|-t>=e
zGzEa=;6QjNc-+(c4LnL!M6<LLaAlFxyJn@-k$H~Khz+mLfPfTv3sa=>Mqefa(x%|N
z4~*t8l|Tt+TiZLgPzHC6s;*jHj&(zFJFTFN-^s92L2@!tWi^$`nq_OKk8E}T<<DhU
z%!cJxfa5bEz6`e^OR5{?On=SWiB~W++GjZ(q0rZr8WcfL<f5Rr84+~IOIc9{wL+8B
zEu(x!dLIHFT@7h1(T$f;HvI8dULvF8C?F^eG@c5;F=?r0EJ_*<T)GS<<kqf~vkjnd
zFo#Suwu|uL98+_^I4WizP|c$h2htZemu6w50=dRTFwO@;`T^ChF=<M-iv+(Dk<alx
z=@JCEShys~xoNl5Cjy)?1$@$@@xG-HcF&6~0UEHN44szG_YbXBj9#36HCzz10#;sB
zuYmM+Vc^OXc$~?Ev``B_qaBxuMVf0Fq;^qmC-a*Qf_z>@SYDiVSJ^)%WROeznHhRW
z;1|fHHVURc3*Lbx-C7{ABZv_y0>{jd=!&8!bbUbxkXuB`(?BMo6Xz@ZnNfeHah)TT
zv2>*arady4Ratk4B4{p`Y6IY?(Zy2ZUzkA!8dFMBNG-ffCMGoNDk<ukW}4SzQqaE0
zIaceWovwRzY4hORrtBQI6?d)FcMV8A8W`;r$^tHvj1L%QVa9?}d}19CczH^L^*<OT
z;pH;h3DEmZp!+kV(N>g)yu(>*hR}#f()3^8(G`@}>{~*83jFaLmg+!=ZDJ(B6dS^P
z2BBJ*YJVeq1gs9oOiNsU;IU8|owrn0rWsw%LS7<zT^{=e?1g*?v(_fLBIcwfd(XH5
z7Ic+dX>%G01r-8&uCl*iO`sa$YFpsw;$pcbaHS#J0xU+<#kD>Rg&?{aF#+*)J`D6!
zREcWh1*<P~y>$$&bTzVJ|L}X$>AE`F<%ZNNE3_FV|Hz(ow8D(YGzDDbkG-7<T83^v
z@n=@7Sv7HTV~m7rgGtuuETs<ga#Osa_6KwIOt_FqZ}vAz`G5^1$u^T)VkVW17Ke&0
z`KbYE0PfaXeFAQX1G~RTZu4i9yy+iDMaAtwyRdzY6{vy2cmTP>Tk>|H1PMc?+!+UQ
z1h|B|42FXl4xW{}ZQM+zLGJM{&5+KiE|z;!Z^)d6V?dX;`}~<IxRn;JmHQL<QO7LR
z428m!(!|fYFpw4+dBFajQo%4@EdS>YEPSKlH?n2`z|0u<i!=i&uSw!($}a<r=mB5F
zjr>*G29Qo|6;6`B4wOOQCJOeZxi3%~3H*g*x<4a@S2s&28D`j5*ie{6Wv2H|S9V@A
znxP#E);aQE*SfL&YCGrXsuyMI`dQw08)`)$rb0s!XQyXvR62qBwaXmuhk9(v`m8jH
z-XPCg%+_4{22MTonn)W;C28|pO3@aCCOP%uch<KsFUomA>mu#=d?Pa1mijVTV7uS{
zh)ZN)+_y!nj?@Iprz}c<A@`=P@Uqw^8k^sB+t}%8H!4zEn^FIRCAQB)kdvV_gA_VT
z(>ltKsG~~x?q#ueiZ!M)u`+ziyP(V4+pr2c3|h|U+_}OhXb{kQSXOqMG^KIWe3ezf
za>8XA?i{NF86k}@Y3~}_Y#6%WgDz{M@pw=|MAXL4JOWV*M1L(vyVWz%>s8jr*%~-}
ztZayH;mTLvA{*0Os4t<L%2L^s+u5XsoQ~H2zyBu>`Fv(WX->%#S|52h&C%q!HE>mW
z#GiRyWjYng=GY|=KSIFtuk-=zFeEBOK=`BHh)Am;FJM*;FCGiL!L1nH+!m8-F;$&D
zBv7B63vD9uo7_%Unz8^24<9@dpzotmU_O{{)qdcZP=EKh_fIyKRvH?g)BprBLJ9e|
zwlgDugrpZw_(Zu5bvRcO3^}yl*=mZ0kB}#WbRgYxR-W=1@*E%$B<8VPo(`ZOA4-G>
zC}`yOxt)9KHvjiXRC(rm;HAB_U7ih`Y^nx*tdKth+8I0K#w*6gAX7DXwLiw(N}9oO
zJ!h+9kqx8|l;>>?Cry|jFZc{e&I$J=$mxaq3yi~yK^Wi~Qv()2Udrv<3se5@6Fym9
zw$X|K2xwFPl=jl{OSmc?{bLj`#;^GLa9PoI2<6=>c{M=~OrOu-&)w!mhX#sL%wW1m
zUh~#)CkCR@8-A9&Zu3P(e)t2B{=E^mKo%+-I{(U>w(gw5TzM<tsW(iuyzOUv8dKoO
zJ8^X)nyJz<Y2f00H&=Z)DkjQ%)*7&0hV}d<27(p0wvid)-cRU5Trh4OauR}6KS=ww
z2(H)a$p{F*VG=*78vAR^OH**S$cOQFQwt29Rq|1S2{Y2rZ^g1IPM5#MepR9z5B=Xi
zjt3c3KC7eKU;1`El%@=@3pL0;{EhK6v*C!%7^@jLH~)<Bn-Th0QzxHzOJEA!tJD`r
zMIZUp?v0MtBTkUd(*9WfUCM@^+k1FJ<;xfTo@!+jI0DIDFJA`6o(sl9z6!Eg1h(*B
z@k&4bI{reonhN<QAr6gftV=RSzO|ts7NTrN8fu-U7+~q5@BA6j5n6I35YpTI{nV}B
zyl-gYSub1kcDFrKSn;Q98P6b=(VLOC%~ritgrdrj*;i!i-mV%!?S<y95Rz?tAPvOC
zrz()E5Wa11Q$`i!%XZ#B!OTzvn15w^|1k8?(757I&;-0gntO_31v=3uXvgRWybEjn
zZ33^jlRvW~Z!xbsr%5rJsA@9M1GjCL-ksYjp&B>#oWZhdZ&jp1cfM41^L{AkfZ{U^
zgY4crB6X=KG@00WQlRP#*`s&oj@lc`+##MeD|`CLv<6;F0>M=Q#;atnm^(E2j0j-N
z_}=N67gWHQ`}mmV*g!U)N&{Rg_=$u~vTtwK8&Uvmk^N$+Y7Wsv2!03c-@B8}nx&MW
z<$%~t>KDL)9+*a9S_*ns(TBl`6w%-L2QkVRX8n|dtS{!PtCoX()@=}2m{Fd0=poU6
zbqTohayc~GKmw(TmWh83^Jlz?X<OR__4eU04r-#|vm?BTW;_5Y&jWLMq>Vieo_CYv
zsNSk<#vkyKvOI)G_jc_jHF*e7Rb%flvCk^q9qZk~g%JBlvan2!bHG7i03OtK%A8n_
zH>{{`f#>oG-b#ZR(Q1uqNO+D;Oe<WgDX2C;5<MwKrMX*9wthNV&l)YK*dmI@6^$Jy
zrv@|!j~y#L;v0k~)JRV|G&%V_rB`pI&yyxfjt@ntk?{_!Gn$*0#JkwkUcJ+jy4<~0
z(I|P*8VYE}4$r`3^X(Zj0I0ot3P2|Xffj0kCs3#}XAV)N`cfE68a7(`cw5W&G#k#-
zRKJ0#K>xTSrEiRh`-~bhxb{gu`#v)nCb0hLJ<Baw_RG_J`MEvi^xkeAV9Hs#5}aY*
zLhj2-k*}0YCdec-GrW0W;FSRe&gLv6`>YHM;BLMZit9mvq?NeWae)pt2w!@<42gAB
zHp$TVedsV5mgcyKBLLx-%5Wc>$_GSdf)IcM(>G@Z^l)4`ABK3@0XDJbEL#%H9c2}&
zD<9$ABaIa<KFd=*(lETE6=s<lM5V#TFv?Z|#GMIVXta-__95df@F_iN>9^D-V|r_j
zssHRT^PiY-IXn8NQ<jX4H3J9eY)5y5V%wJirKK`1y<QI1(-L-^<IhZ0ppvkho4{mg
zT)Hw<jgK>5j)W3ba$ed?Mm7@X^d~rx^Zku#TY+0p_7JKdPZ!vXTWF>!mkaH$F}2N*
zj%7mZGGm4C_M(`asyAt&St%E%F<GjS93nJ_dU0NorcF$ClQv+mR4z5VKuU@dDe*or
zTNRPX2bMY%+wg3bl=&`EFNUiMQAw8Gf~rwg;wcYY(NF-rBCs|FEt&*T5kNxyHk3lC
zv<(;ul}D9P50v!<S5zIOWZdX6Qsc8C+d%CM?S~V4o5GwvLwdmmm&Enbg+WzllMH5K
zan-9zxVCqv%OZ4|=q_xVg~Q@kuzH-TI-6P{YE;+zqR7a>%EvfhENUr+q@+JrGS+lb
zwniEZn#kbUB$vgGZEgkUzeXB;9bgS`f-KfPn|5)d@}N)_NRvOKV*rb@!4XuN6I?X`
zywb5*&{MI75yJRna^MuOf$|kHDI5KeDLx}cm`x2DaON0kF%ToQB4UuHr6E~d9GndF
zGnZR~qULg>FSx>%ScFWjSH_edS<WmXSlW8IDlV&Tj*W74+9Op5f!<OwfSb0?4~30L
z-{cza&ZHWH{k3VemROw`9!Le~JP2KdT$fg>nQY)|MK*%k@9P~~=L0WR$`8G7S{5L>
zGdrN%U@bt1SRnLAfsduo6ST;W<K>tPc*CE>G`htZY0Mi2<8i&wz**A(Vd5s+Y$$|R
z<NvGVW?#HrZl(+BcI>eI41&HTI4#5|Qq_XMT$@2Q1KVSx+!`}bEs_O(8#_({JzH)|
z2i}MbFH*ZOZvWnax6CP=^*Xu3W_=Fw7R#Nf74y)62LyL#2lcMBEzA?Ec|zoFZ{6KO
zOH#2`?g>;cM;JEdeQ(TLMOA<wnjb=fIq_&GxX;k3{}GqV{XP&&0V84o2WpA@EOrI8
zeG$X|fR77j3Mvn94KWMMDtM0k-1^C0)-1p9erA)ZVZ!u)`(;c+-SDaYrs9vBh27=9
z@{Vyo@Vo#^$PP2X=db+@K}OmJzfR!K3cP8Jalrnv;uMQgO^*petwNP&R~@k#LCC5a
zn`LH#sVX*bNLR{(>5ypV5NJlh$>G{%ZWm}yabRGilUV@}sN@iEH9OFzEYsWw{=7xz
z_)4KF!$<a$hBDmOI?P-{E39M_^eAmH5u7QyCeej&UTk4eZCwVgCG&$(8IwWs5lHF_
zV$hK=<)d;>i7ZT=ZbZz(q5vlJ%gB$cKMj`_`{Z~XP<>7tI5pekPhl07cJGJhW0oxO
z$(condf0YZ8WV<;0aqK1EMJx&_g_>Ogj|ZmSRN<<tYJV5Yh-r2G-O4P#F0Z~rEM1C
z>~O!X@=4Ktw@_C51k|-iJ6IuW0w9$cxC&G5Uz?^m6ITQU!kMxzJzEI$C*RR7>upf5
zYZ=)PaEC?&URAQu#_ebwR}G`jCjXifXHeaG5Zi|w(Da>Sxs-+E)Raqn*!u0&6SeOi
z@h_Fo;Lt2Zc*JJ^QfFaj06!5>DIE?brN6+LZjou+G<h`U0V+!eD<mnB$L#P818#JB
z$I1|5;jCh8d;gnQ-nk`6#F>@Hy(3lugmzpR%-=ew=@-!^Pgu8f0Q_CxAQD@k-dnjm
zX**O=SBgKLimq_p$j{R??`eNVF_Q{TQi0#w6iglf$Nfxv2NN=*`RHW&Z0ra)_&@k`
zn0;uT{Lx!eT}*upeM6tK;|Nc<25L8+4;&ibEHBt#!<mMa9$GK<c83h+{i%%b-j}>n
zstxH2PfGA|obmE0C5@Hxr^GO4>TwmIJ9#Ch@+fW5mhAgggI0czT={bss?#h24v5(a
zsBAF8|C#|0RdcZqultVaJ)#Y8IjeR1jTi+iE979+7`$mB#qo#BTQSQ4Ma9F5<!x{5
zN#Zq>`kizx8Y?QM@m#!Xt&m|b1D`Al8htO;Hvrl7n1Zcd{t_eOZe+0|77lpdpPBxG
zyw?o*z@L#Npz}CnartWuIvk-!jSP$VFySOwUXyWYS)M5$#Sw%0u2KGG5QG&=K8}vT
zMQN1@@PU7ij%yf(3wxXVBR+@acZvMdv4uv84f09Q`?H{I{xt19Zgv*kR0rD6(&x8=
zPx#!2r2d6Ygk(Lxh`phAB=(sspY6?;u_l}wrpWB7DgISV14k3Zbkjm;@zWmUYkx*R
z7&;xItcrXS2P|99+|cFnZIEiXK(M%*sZLPLhe!(E^xv`lqq9(L$am3P?N*kmMbn<3
zIrSit=UN?T4v>~<i@Z=gQT=1hyK&*NWuB?`sLEijBXj{#3^KTEm1pWbW@_QNhG|K*
z_Gfe(rI#u-v)iP03xPtwd+f=!_N=2d113wIP|9VyJlC$2luv~D){N)d=eaJBTaBCy
z%%%hE-67iIAtH7r*TCQ_JLaj5t+E1fOgrVd_H9H4sd0CXDL_PGc7gRUg6xvlNym+j
zR%nq*Ao}ha3suE}<F@X|Zh4(lZ%wJFKozImqvL_2$DA*F_(#gSfOH}evEL_q`jF=8
zVkWA+Y=ZQPlf518YWl@Dbj8Ac>g@sDyzJxOTPPiM{i|i)JXcg85(YVBv$CIcs6h$)
z+bqkQ%YZWn_}Ao|%-wmF9B2)Y^d5m-Kd@!6<}DNv2c^Lnu%Nu59GquLRg^|P#C{z;
zcBC9?-(eTZKphT?2_g6$>EN5>a2sO;=vkv2;k^dRA|OhR^k*hosM=hPav;v?VA46&
z2`fMr%hBGNE&(cN;G#a}d;7#|lWL|_;9!rnX;!zC%W<(U;Jv6HF=>~PK*!q*Yb)wd
z%JPJmKa*0B!#FVx2f{O<13bybw}2qzvO+dyIXTTq)6d|%FOyU9I#cM-e+wUyQv-)5
zH`l^(qr}Us_lW(dsjHHn23M?PA((<*@v3*vUXtT$Aumk%p6>8Qh;#i-a0qn5GUapc
zSjPY$_aw=){%SV=5L=L+7GOX{K0hfiM5<>zbz#5`r<a-gXfiia*C%EI*9q#`DC60i
zzWzr4!sapEp}NFLKOc?RYU%LV2Gw|fZ>>Qo2(4g^Q+xwe=V@^&aPX)J<ar>)U^oQ}
zew-e2*L*DGu-fH}yv_slApa+g6Sq!^3^9sE7Rvx1fXo{Ob(a65;)H3@k<&*8`T)kL
zb0KSGP<qBv*{YV1!GSr76)R4XAqi?LArR60!2D`7d=B-`S<;<!5@~v5m_Jii8qbhJ
zONIvw!FqT+6yQaX^vs|+m%*uD&Pv~!H40FW5ovVh{xTl|+Wl6@NJlI<)IPF6M)|j<
zHc^YKj85N1C?I!(isv!i?grwVG4)IjKwvG9vja_GUXroijb2KC63RN^6Eb_n-3DMu
z1Cu%F6*(?X70*@E<ec~#8&4zfN-dE@JIc9voo4x{tC`x=%<{kEZ99um)9gH74^;rP
z7p{k;@O)cwO$lOh%H;y@*f7(QF2VYSi||70CKn+L6Ko7P@HW6kO!u9OQpZpyV?GQ-
zyx5<?3ls`2HI>UHc5sjgu|zKQQw9;DUP=PeaCBh~YHC7C?HLl~*H=QRrg;U+5^xAz
zWBpVplk%7X)e2Y}V{|Iw8Av7dE3oNHrCns_BKAx(fmB5s6<l%i0Jo~+8O}6iXGW^l
z1Xh7oH_1eAVRt{HkLSsx0KG-4O{<XF=m+Ny_d5i^byDZgK;~in<T+j__34;V>`?p$
z5EtV#QxCRYGJ!e8Wl+%BL~6l>g8W?OP%JA!GWy0u4@^7AL^I|_WdSAGm?12;$Z5(%
zd*9n{byLCkbWUvcH;UV!qfAgh{UtPYX;mtFGs0J<*fh$t9Q4$*N!00ZXbTisDJ?+<
z22~(_8+L$c@c=ulArzOVVa*T33WQRjafQF(T}Ztpt##NMZ0VKh8;-=#DZBxTuuQH>
zGtn?ecj_P+ygDsH>z(7`-z3+herO|9o2Q<?wQ{YsX5#A#I;E4Ty3Tu`EUsQGxKpps
z^N<Us!fbA&jQ7J}oJLs1zZ>iblpsT|{3xzEG%qQK?S;YlsJ9zHLAgTZ-+mI^j)r8Q
zMj^S;umBZ$YhZG}DPV(DlV%Whq{j2)X6uWb3;c?ZKH-)avq3Xv@KmYRE>0#wG;g)d
z91b6^3b`!-iZwIHc`cUP{Tb;kZ#(G!<&HQX6UhqPnLuf={MZ!5<u<v?zon{VA!sjp
zkh{|un$m%?yWA7~jUPT@gxs52tipY667PL+YY){Ro%@5y&9)#_G+Yyi-#<$;wfr5J
zhdf|ClaIyz{5*OY0x!O-{31beb;NUXI56<P^f#(*0CqrF*CV||e&xr`_%13<=$4yG
z6w0p?FgEaN1`p(7t+b|PZJ>h*=qCIC(=$r1w1Uo(8Ag211f`rA6Vzis4GQQbw_P6e
z4=Iu(=?#iZFOyk`vM{HL!VPagube{8PVX0Exuj4EUHmi^`W){Qx&w_ZhIwD+#tcB4
zsHvNS84%?bNL#?NcuEcU<ayo@5rL|t>q>+$-@8S9DKOkbg0-?BFc4zw!ZZZcIOI5Z
zUCn%CzAW-LL?0R_Ngs@HSe%x>ijkzyI&P45Z%tJKTC3K9Uv7{k-kPijBN!pO%2MY`
z|5M?flVw>Vu9RuHU^+pmf>Jfv%k9*RtbuoutS}rygQd{ouT0CG37(UKEt6Gwojt&Z
zu_JJLwB^G4UIk?}+~YZJ=NLdO2NZah%W7Xze5VQd@Me~*2`UlkV6(C|h!3J7;VdHS
z{1VSEko7j*j@IErhscK5IHc96YmklJ3@$XA?AKE+xkMhaMyNB?$Q&XMr#<4KrQVE8
zE_wb(qHBg6ZjNR^dj@L~H|;j+unmXdpiUkQl2_H3X{H$en1KvNj*<rC?80qVC!G#P
zpee?pN69+*jqeJnwb8uD9Bx?gcwVR0;cLx8;DUn|B{AAbej5XhWxjQoEuM(Zao}J*
zM(#Jn^V;Hz;UNBG+CrRi_Z)dDpoW_Z0^HNao(&l#zqjuyp=x=?zvFhYGCi10dDfrB
zrUN(S57Ac<oa3A1kM<|Xy%wQ+&N{rTIU~=fiNHr)@8y~Df<FUX)!CZG`d&;YyDL8d
zs5k9xO0QpvZ<JqxP3zI?<Yn8DhUyl0*ZID6g_Y3vb{#a7ujHBKlAHm>-{qjl#xHUc
zcr~+h@~eS*bx<3^lkm^Md%(D~7XRNUui0qLRL#Hc=Vc~1wFUBqA<&qygW;9+rcIRf
z$Ta|mK6h_9>XcMv<!xUG)1fl*8%vRQeDP%2X&zt*qA|1Jw_)yh*T13m7LFTm_M0j1
z#WRQ;6dad>t@%sz4ZMcJ0j-zH`{@AE>DXA%IiM^n<pbNL9yz__ul^w~3tEnOLw=YB
zVB$(U10SUU;7|rl(M+&2e@iG$*DFwXAdYH849fDcf5Su)q|=M2;|u5_j+8zwD4h{c
zCc?r3eM46M89S;#QC<#IhT8m3(uic(><y#?8gf)7`_#Y<T}YXH=A&b$AawBOZ28>3
z*Z36DERaUx0(II;fQw()Hyy3n3ShVCq`nM<N2y_)_pb~A=u`=#!M_rWEy9=f7NxlI
zwN2%C{F84CO^O;&?@_)De4A3Ejz<6Xd6HPr(TS<>zl%d(1_c<k@sgiZM5rqVL1v5m
zaL;2-Fw1OaFxWERG=J19@Z#t7PHqVHeCQa~$yWKU--Bs~6{>laZ=HHG_LBi_bh%n1
z+jwh|0D3quWAV26syIXT*9WU*yL^@Pt~^Gz&v)G%Ob89KgHOXW$EK<5?q|Paey8c_
z%AqO5f_Ad6i^q)}K5D4!oUb~+Olfna>|(<qEgYgTc8zXv&r?!@D2bV_aLeHu+0D99
z!q+Cdr**M-w)$t-H?l{*s!QR#dj$*K>}h}2IUenmuS&AXH5IbAcj1Hs7aub@PyxM<
zgT^R2t#C_ZU;D6V)cLYsdOe>}eX7&gKf!(-PL@IQ0NYx)`!})3T7w)Idp)eC7T(FI
zsS$gim$>N;4vJ?qkMneZF*(>k+E9(rW<VM{BtTr6ful@|9O|&MrI6&f=?G_l{#J)Y
zpJ2$T#wRm*xIfbqk9G*qlOt@AIfXstNQaIcLKDmE!=ut0WNE`v`BOzSIXd4J!th{d
z#3fzNY@5gUI3~X-qR6qfDIKjy<^>h)aW?EU=t_<^^nhsABq#WnG-+AD?@~E2t%Ph3
zZD<?hq;Ah3D+6Mi<m7-5JToSjRLCjT240FS4H<CeS+F9f2F88QvYoPWNss7*kzk-_
zJ>&1<!6OC_94EbeT>3Us&%YjMlH<Su@2dhF<zJrH<&v95>oQN9P25g*@3e=!Fc9hl
zL9$z3TFDET71vZj6PoXX^YSALjH=C2;9Y=DDYr^-iSB<?AfV`%6vhHTc_?Z0u`b6@
zqJ$@MUxR%G@^eYQKvhjvic{P_-?chW;G<|2DCBAJ1jtQgY*tRUBZ|Hu*T@+OPb`Ue
z^4OK41UP+zJFQIy*jyT$VZeoGJJ99=dYzR)zO2w^CC6Dm(}QCmoJ0yY^f4!?ULZsK
zjSA^F+(<|5Z-%DzHJ2iAzEOt7Al(>XnnS5KLU^MLH)?p{sdA>zKJ;ct=PPHWDRkwA
z;S}0sge?S#)w41(wNm1SD;ca1R=-+C*$JqrsFKls1`B%07$3=$j<%$;{aNfQ6_#V;
z+(8sK1_cl-HsU&lK9VvnkY`+DGhA2C@rjVYF|H6rPaT%$+UQx>SH=g3G&T)a)p<dI
zJ7z)#J3rqIJ=td7ezg9u0@MX*eqhwd8h{bI(4J9^28XqB7(B8v!P^1FA4S#;s{Q36
z-veSZqrO<)qI+7)i(}eqShM8w<^Xj`U^Wu!!2MS)jo#6dIzOWKl7JHP2TsE+UtQB6
zr3RRiIxS!&Wwt~X7lEm*+<T$COu+%RD5>ygBxFXO7f5CF+zjPRl~!g)V!rgTRa9OH
z2kq+k822?}P?#{RCOuovh)VFnvt*)8gKP>^E0Y{7VbKFN>z?tR!S7NVlOD*J<2vt8
zqkVLYWm2D>tpYKCYX`J|8Gi=R7dI{BcTlf}05E)YqFlOBr^{lut6Pv@yF98yG<qjA
z*f83Vhc4^WWvxytUP?1088YjtDc_iP{R@+!QN!Y^tOPZ!%=4u=s0)pS8mK+l#)7yA
zJ1A%l?bB1@8IZgYrpYWQ)zsiBdgd2Oi{bMC#Ntj1k}`g@T<#0yjH7e~XR=7H2s!|1
z609wlqotT5!*@u=*U6OuHedtSMOTKa{F$u<*n%tfYVXHH6nc+CR8k8sGg!@&Yhr!s
zuU?5WdTpH1V(87Q<+|9_&9{RyyFTC5dJxTOD&gLa(zreX(TK`^MtG=l%;9`Pz+rSL
ziskwg!+_F`Df~DFM3{9oYG%ZYK+O>gjRSvUd<6FuhWX-}lbalJE_GmCVOPJ|TZ5>e
zwS^1c64NdR8B*K5{M4K2al*l;|Lm={OSuKPa+~*W*kQG?h@5S5dqPes-YvB>_7CvC
z!&W-3XuycUa%Ze+ENU-S$X&MU%&jDM`_N>07*9&td5^sxKCq}t?hObadYeY(TDdPe
z8^}~~a=%>y4a4MT_E|$SqQD*q;sbL&y^4PxU;`&Y25A+<!CxeV>r(8Rwf&c=^&D_H
zM978ms{kAF%{4<>L8reCxC3C6F`>Z0Pnf8rH4p_!k1AxkcZXrq8fQcc1bLOo%mhbu
zOn^e2ReA}F2mRaG$Og+`Gd`P@@R)v)<N&EMWtCf+2Uf{!14-FLnd4t$383iaGCeoV
zTJz;&tVq<;7W0I7s=2m-<nz)q(pysg1v20IA6Fv_?6>pImxZwcge+Lj*U&XDN^@AR
z)R{3*i*2iV^yt-F+WkFq$ZYT&GD~8QQRM>alBM~b>ffwvztT@5{ZuV&pesJT!0<ag
zIhI*>9jz$s02W9c_m`(7TXiPp*fYj?MJx}O9Zi|YY$+>a@X4r`*eI(Kv?*8{!(;wn
zWNT3WHc)LO%u!&j*LauAphs%71{Q#|-iN8i?TU{`mbGA)*QGJKg5yj#1WpIS|9Zm~
zI9!tsZ%DnZ#Nwf7g>a_P2uf>K$VOkThAU!GU3KU;rExT;4DL`ky@8U%J@Al^0}U)P
zDm?5fAq`zZfZ?uvBsh?Ul9pOV8*WapT?Joc5(m2KLZ+3DfGvojDn-FkKANV^<Z@(1
zxQG0bBy*KKW)rBbX=;M>-s#gJ=RHq;<NXoUm!dtrJ0A~V<Nl02{jEPE55?3dw8B1N
zJt6%Ms?^EvqQxU?p+&*$u<K9y%+&UPIX#$+J-j03$x}uTi$<L#PkU3Acwvfgx~tEE
zwun|O)?EF)(LP#svhqxV&muHnsUhYHTZe1y*}yXrGPvjYV!=g1@AOCijMS8>d~)S>
z`k%9Q^qqrS<@3H2r=OV7N_oM*ovY{Ai~(jve$f_j<tY>7rI?pW^;Cm-V41w^U7ATv
z8D05P>Jq7maP3h0WcJusVij5vOkNHAC}jW`q6PjO#{<>yP-<0Pv-W42{z+apO3+Au
z!#50Wh`<_rzP#z+m<#AHkhg5&*~wM%w(}}_&ltuh6hgUX>;ZFryc1jOj5IBR?*<Ye
zf^dVp7XWw!*;!1B=Mer&zWX+o)l3Et>eaX2_gyzve?=>RfKme`@ul)X+6mTqVeYAG
z<*)XPyekw~h=hff^25}Q*#k+27Rg5e#EB)7QGoSt28fatid`T3Ji&u0N@`HA#v}gS
zd!o9C!U8C?{3F0UvAMhocho-}jMqQ10da+&#LBAatr(ebKeau8Ywl3_%m6(AC;Rib
zdMRFJN+Di;5gWv0?q9|!fHIF>jg%!#N>D#p`O4qNyv>u}Y>=dR|MJ-zmeK)kj(i<M
z)|FJ%)2&p#3CK|6ot1C>Bi_kS_CwRQRQ?^?S~Pr6v3!@nWqMHNFtk8HDd^lzeOv}s
z)10wI{8>?3id5`d7N|lEFyYm{Re@>zY@At;t~^>!wk|M@9*-o|$D5$(+@`=eK?F!u
zH>t3^ZGmd)rqs!H{sGlFq!3KjvweZ8yn6S@ksWLr7*_o<cJzUfAstFYD$N;_Lm!Tv
z3S5(uZ7wYbVrrQA&IPI+giPLCA-fc)9`EAIkDDO7#_E{tA}hPubjA#j-Q$b1z<=*i
zVCqZ1k(3uOxq#Gut?XG4b`rRGCO6-#?B%nh3Y)4<N*d}QCVTs8Nu-(mtPMu3eF9`E
zvzeFsz6OE*6E2ti{Iy07k<@{O+258@g6mrji1(nKJTP=8I?#3t;m%Cl@q_Q}mL|Rg
zpqUi%AnONt{jzd!^rJzpbdOmkhxjv^>**^6enJi{aAhgXUQGb&VcxsxeF+}QYsl;@
zho`S;aI{<R2!k(@|4nL<Bdr0h9d&kw(yB#{vZ?C<%*xSe2N);eP@_Ev3q>XuIVKIk
zSZ{i^t(RkCg5{a!QXGooY>^Gj%yhiBC)uXTiu#TdVvKHiUHzKVOY}rvAj95FZ$)=U
z@-`<K+LTU|lf4gWUf_TQYbh+WxNW?Ip`7CH?J_q-IHeBSQD#7-J~d&C#ur*OZMEI@
zC~z&7rzOlv&sZNS6;f-}%h%-Y`?%HCOHQ{frwNKwQMoJ0jsB43B%>6xx8d3qC_J8*
zrU6Ep#21$<FUa+h?}JbhXICrW5msP;?NxxdgTk1UW@4+AJ_UYBBo_^8R{I7R2tEZr
zEnhV4+vVM_zy$4XT10!}R~UuGjndx^Z&^h{6Jqz}G#eU)E}CGKESJ-5T74(<ogin#
zYKJkAF^y(L@kE_F5PvHJ5)M)IX$NSaPqh0)XoR4r3@Eh>GAspmcrBdN2B&3pm7cL1
z&{_|1u-Z)oW&)(qEb2pJHk!GI<-Udm0XpjBqhz>k4N98Inf7Z<*)%!JyJP33lX1U)
z84&}4`m!{NvawWT$jF5FOn}7eXqJphdrEFoiIt4@g(0;yyn)Af3n~)GgF*_Bvkly^
zMl$4Wt&H_CSsI#kVMy|jtQb6FvA<DjpezInW=sbQHZHKG3`K>oW#`za#~**{amSr3
z=h{-oAWz(QC;tu56Jo2*b3k`j)R=3Dc?+H&>sNk_>t{7z`Zr$Sm_>%dop}SZa6wZc
zKo;mio7a$v(nh#Jw#bCkA0wtoUg*t3Tf{{^6A}_5MizEY1tJ(<x%hv^WL#Phe3Z0Y
z64y|+78Vz|)UW~Rm1?Pa0JfIc26VKdeg)IR@VuACwt?W&lu4JBvS`=PNUtT7`9k(S
zI|m5yEt84>ILpDneo+|<Q!N((MJfTBDplS)NIh>@{*-EehW9YWP^&USYT{Ba1tW{l
zmGZ1ij6(xvpVsYm2(-!t19D8wJjuHOVN`jP%BkA)tR1=pqXKo&b$JDN9QBT?cW`zq
z!eEUtLyD`ad6kU8PvyWF__<mdynCayCPZE4)1=FwmQ+yGrUNkLy2aAy-^_w!NHybn
z@Qqn}#{40GF!jOZ(iBsN9t(0JL}WMs#_8?Vv!_fBpf@2gc2h>CM8{xHLzy!*)(Xc^
z+~(}JCB~|$udXSD<`-P<v_KSi{DBx;o_c1|psZYxdL{vygXDs!r)y`bqjgwQQ-fUP
zE%e^ULJ>nISI093``Av&HTKn1lyX9<-)rrHJK|WmE+&NpDxR4_t~Ufj@;a#Yyf3=e
z<F1E^V5!`Yo`IL(bqMtOkw1gxNG*|+QhyvoumL#@)tT=nj=HeH%Z-kz^qFFGJJ@H;
zl`P*)-j}Hp6*=YR=$|ePV7)d%0eDM+8mt+&1^+fc^QYE*M{D2X<<>OG)hfkPAAFmA
zK#$u-<&ADP@FV{@gCgV(|HcG8YPMF(ooV1r)E8i}>*X$=I*F6U*)snh)a5he?l^yt
z#^_eYWK2d_?uogbewy5ymX4#M<edBmkIj8CniYM7j$)16Z(El^lmsN-pLq+5TvnsI
zJP_E~Ge1{;9+;OY1t*8X67b{m@vM3QSUITF?MVJItt5FBU?y}7{FRLg4L2S8f1Su8
zuO4W4saL~_$xN{1Olxd4s_vNCY`V{LE_8Q+I5PtNsJfx@#!LfB15%XAgK3B^%4>=v
zz3*gJnyO6!cLe}U&*s?)B%n%o#sq?MY-f*9AN!RuH=3w&gdE>=X-m(R61Bn2FOqoy
z$;lPdWWF^(Zt><)_*`j**9AfUp)+G9Qgi>OLK65(>ImSjm4)6lQHZxD&Gw68IOyx3
zeF1_jk;QSAX|<GgU!mRqj5{V;a5I+p8=eRT%rYj$1pcMzjaeL#fCx*0+C-N5c$g1e
zb68JAmM0ji!yaxdT6S?6;5|?5idcS6IK{1u>Cl6n91-0+R1>=@_HhaXYPe;|>VnSm
zC}SdG;(z0Rvc~64JrX@LDR|*0rln3FvNo-i_HX4JWSvi+OtL4bLTmgk>tiR7<3g?F
zP%dvUOdNgAs6n#Ps4(q1^;J;NYzjbt1t%lPYTwuia9<C3Pc91LN?9Tgr{0-gg~kv#
z0VXbmT$-kTGEicFS7|T_%QkPeLj;Q_4sCS>v`8H`s}h>oAGLnf?YbOtGqWT<7Q62v
zEJF*8HSKhW5zkmIls_rw;bP=Z`AtA4RU)REn`Iv#H=JxPt7Wd8-=@{D(gqXSPM0Tw
zlwCMMeis)Q(!EtQ%acB=(HNX5XIF{Ha3z$d;=U)hjRCC=4TTc38kv;*bX<dUga>;4
z-X{S=A}kK@eZ`S@CN>3}4zETF>8Cr<vw_6Ji$}{JVnQ(PUPn^PD_sp#8jvr43^)+6
z(GC5yY!c5!H&BPk^F9NTB5Hlwv8gWv@PQVLLZ=nhF0dEleBerDDm!^8`s#_4NiTa}
z)SpqBgw!j4iZ8J52rRCYS5l)*(4111S|+askwwPn60m0QBXs<zpH?ET`4^m=8sQ$j
zUS2nj`{dIukvD9kz-VfoqBn!`Kq<IN-tup)wj2G_mdV@JH&V=J<(+sB{sOt<ux%UV
zU0*&0BI>>uP#vuI_j?X=0E^cIC=cCG8OuWWua&>p*wE!<z>B|cOCOwa2!0T-fPQ!q
zoooLZdkDIgDk*eh4?hfascWu<Nm4$F*<v&hE|9MW<!=c(vw#O&;R5;C=7Fu-2yNEi
z4Z39MG~yW5c>}kHG2T`F8T*Y}o$iU#_3j1-AfH%A!?TSgs8aZ;{ngRhl0jt8XMtdl
zG8tagmNeR*$2#la<u(uHxW6!n!=Tb4Uq&YbXzlwd@TXVLJo#7j2!7oF8?oMB`y#2W
z;Fd6i{+n1hY-9`}TL-fBtt}O6U^Oo!>5zZh07vqld}p7~Nupx1)(CD9DOA;6CYzBh
z3Oo0wv$l#>*iv3E-E%N(S*Qv}I5(>9a;rjD;FTeZCUQ|^D8p=B=ze?%%bg_K6uNQ|
zx<V$LMlhmm>(8vP9ZC8U*)BZ;_l_|cp0s`RG%izKE;|&eF0i~An$8{VcPyhyb}H;-
z94;P%oa|ianlAS?+NCg5F*tdM^DmcOt$AHb_Oe>p&6b{r{P4RMdPK8Q3sOTTJ+g<t
z;SEg^#R5`xS?(F@r$9xI?s|w+ds&}_s6Vl{gHvxUzg>(DG0wkT_K7i71*Jt2uyh8;
z?dxCizNaaMY#mi;9GOz@ehI8}Y%v?}{sFF%sn~`CY@N7MGR<<JHwQ-tA_zS`bZ9^D
zXAt`E3{~78St<v`+zaJkgXGyr1)*Z?A*l;yWuf9~t{m#1hm>0{hxuZls6Yb}1@gzt
z337OB0?shfLqJA8!p0d`G)|5T=+f2+w4Db5PN|C=WsSx{t9x`Tm+TL)h&~j@7!H7W
zE0JRhRb371W?6MTGaaJ<Z+wD<@jAl&F2|=S(tt?Xz&)k``h;#fL8as-HB3MwBPZG#
z8Doj+9|mH`sk3qZPPQRBkf4Z#QJUx&b4sCV$xpseP7PEnWBgrKdiaDW$kCx43Av<a
zJcICsOgVVAUV0U(=+l)aHOCt<tb^uw3cR^KUr0>gF{m5jL8TXFZ=3G`ChC?vg9{uI
z%{l^JJl~*t%7hE0AP`1pUT9~#q@JFOjnXGZ)$5!=DeclXjS3|!AV+Nhd8V1t&s)3I
z?a^z%yZ2ATg!eBrQGh837m&kddRk0w7>pNS!Ex!G9y1z*XazYVb!rSSLPw;?pD`tY
zMW-x~0fAqLd_bakXru=k3=zd&jWdbW3^HWt(F2jwgAD>T6KiBhI%ZaR3Th`^7EQQ}
z@E96<Sl+05*I_XuNOZ81Vnf<wcpS8e@T)sBH2_|5otW}j{)~DyMl4ep8Ijt-O$S4=
zuGC2D2;vh~Fv=TfB7R^CQwWcC@bf~Zp+*a&X@;FKws%(ZyI#)rhO<Gj=nw>of-z{e
zAM4!$c#vL@<r17ku_HCD0SjeZkk)es$~mzeJqvRR<=pfw<G67~0HE<{ns)gyg7&<4
z1`-hY`_4~*UeC&r03Yn<5Hcfk;RUt;hT76@dSRifqQj9Eu5{%xA*O@y^?JD|Ez%4f
zt01qNviV}405nV*T8)G}a!Hy1!`D6Yg;G8fk1loK!{sX@C4o?^oX5ytDK(tQ?_D5e
z-aA9?d5%}WkY64%+k7jS{R(S55<F^U+z@0Jk@8V>8CB6e@&+Lq8ML7~){At>%_;h$
zCbpusStk0Z5ImSH1{}v~ndECC?@Z@<=2$TkHx(~4r8eN8N6&(H>td#7Qx9LCTFiq{
zQNNW;I^<nWL}q$w@Ha$c)iVORrdhl!Rs<K{Yz@erG}?wj7s&b+lJ$XHuMAdurZm}9
zP@8E|bAvS7225sHfJ{z%tKrY?I?Xpz{0)UPioTiz0EE0<rW&-SwSe7f@m3aw#X+gr
zfbB-6rHL?SmwJB5<^Ier1iO5$uszH}XxNnwW<(o7IyC1KJ;msfDdw;8!3YXQR3SzU
zh)9j|)d>h%%j=q$yHO9Y+7#&Nr+01I2Q3W>!n#SWORc*yz?%d8^)X+NUJCiMfeSy3
z9+Vl!6Zo>-;5|@+qBYa>Rx}9yD7}HlHVH&h<o-ClQKAi<7?WWl>#Y1F=v+Av=SCln
zDk4Vjtpf9KQ%nI84I|T;D(Pl_HV0>&jwS6lEVuX&lr<1y24o_Kvlep4Pi+fP@j4^7
zdMm24>AJ;=>V#^$+~#k{o-&4h3B*mJ?o7Gez8^REym4|z%*J>pDmvuOAm<<}6^*5G
zm+vg>AS(AS1ZiM@^m&C}mE7%p@X7{*jazN0+!K9FY=$prR_-+{21NpAD))UKeAK*f
zZID6k5AcmdddHsyy{^PsE93#2_lOc?CiuCJNrNQIM9qNP(l5MMA}H&ou2lsj&_aHh
zU}1J6di|62S|q=+Auunf<FoSXv=q%%2UcOew5D#E582L&A@1}z&y<eJsDGUiH)^Ex
zpJ@jqr?5aC^nT1zsx%m2q_uTc+60n)GbzmW*0i-UWjlB|yrG3|j)ADMwiz-0bA4|h
zd;?2N3BW>_<aQQ-5VRSiHEMv&3z9~$MCQjtH8=$E2Mc1u&PT`q+#(jn-(@A)n)-|^
zvQbAenVc*RFoAJpQFpLp(r(X`&6@}#54;*=iH%;|B1>&_CCM9PnSW3JF1pcdg#X_1
z*!km*J5E*<cB<oFHx-wyo|IIbz%<1P(0R42jD?ldAy9EuAS<W`WAv_rf=l@V{kcI_
zr}wPQsj$Wbv`#0uhh?(HknGCiFTPyX1`I0c-hp^o+srdd%4EI0q9>}leQfY<!_kif
zS!I2vaT|Tb)P#~4A(ygj@^7s4mrYX1D)y1w_Cx*+&l>TS$oO#T2f05neb7;P#QVXw
zYS^O6$yC#nh33d+M<wkQ-_YT})aD%l^RsZ;9`&wi=pbju5Sqtq#l!HrGoddnu$&Y9
zX{mkwO&o*r61ea17(AZtbI6il4B#G~mEQ&<TiIM&iTe($$`i2~${koC;&kM9amjDK
z4fGKy#qi$;<w<YJH9`#I=E_q+9w2*xd41Z!p#CTh06P#=QqPs&NAoO_Q8majaRSC*
z4=HUuYiosBU;f|}7By$&k2W$YG@1V|jhD~)Y)HLXaST_zJRisc?FGs#$_ugFESMwc
zd9d6U15@FM)FdxOC-mt8Kp|hs%MRq(;M*BKlM;l>pRD`A7TuARS7Nwf&2TZRmsf4}
znMoP>bL<DAKP$>f)?SN!9yzE;UXKMaSR?}@HTdW^3Olu4pLx0RrlZKRMip-x{7P9U
zZ>HHX6xBE>L)Eu@Eoy$pyqU1&ZG$(&eMFGe%R8}L@YK}D62sq3TSnE41-5n9#=jTu
zy9_PBUpm**c)3LW5;&=LN9;zL>iY(ce3XCrz!s0<rR5p<Yal@ls!o^6hX%!#dQTAX
zk!?JpAW<FH2ajXpM;a^V!Ja1{2T++a>+ilgR-~h(!qxr5J79!A+-0EJp@tcEyRKn_
z{4)WK*;<wTl1~CFS(3K~IP+<GfMB_)hz9#C;0O*%jW+OR`aJbSr*2ReOz#UPq1|q3
zj7z2Se<z_|1~OGP*H?jW`6_Ui9)M7_!Z!ae1L&AxW8~|2OVhno%K9ccU?CJ}IllD?
z*=-3f8>H*s23w?jZ;|g}8_q$NzDAKgs*j>cOhZYw=;Io*Dasck00JuHE&HfutE2VG
zqhzZ-rh%h|Z-Xv~g1wc5B>%c~A63>hXQ5i_W2!Y?r1XYh%=)%{+#?V+22iH$`j}En
z)lI-&$SS%Wi)4F$Lk5HD5vC?pK~Z){-A-3VmNvj0Q|s>0He^6{@;4TENdp~OGT!|=
zr?EF8LI;EmC~Rfrc8N(s+=FF&SA$VU>j*@R0cN}PaRp)-ve(M)sT(!rQE%oq&mKNe
z9!<;nF-P_c_)?ODPO^rsDto0Bt+LAtOyu4+=;k|sy0^omwof0`B96_#A4&F&2@Xf$
z>PFem-gUJ0f_LNoF)Gg~piYVRfB=GGA?HI+9(rjX=x;2$0G){tM-2R)FF%Me#-cWD
zg&Y*i#AZ@HWpJ%tnR0NN{YKvVstGtGJ)@bPrn60Qs6U&fuAR0AhoxtXb!dZmXo(!&
zN7M6TM`-rvPiC^}o<t5zwoy8xgmQ%8C(J-uInsONbw$+zf|KPa$64d(I5K_YERhm(
zkQ^NwirZ|8923x_FZHqZ3JNuJD9812vA-$<XK*c)h%<Y<gT_qq)KXB`B};RH56sC#
zfE8ejn+yAb8}~#zEYn~@IjK+QPU<RKl118jrQ0X>QAHbUz>PJNOHj(>lmsn0$Lgt*
zQ~%4pEQQYrhg%>$Vgp$@1gZL>p$eu}9@F%)y==jw&lx*#u;lm{gjJiLhuoM5LVMt0
z-zdGK5nQf$XmEu}o>P(t)yWT{F?fg+*a&qc6LFjhyMUtmu#$umd;;@vlWFP`t8{N6
zifMffQ2mdXApQI`Y%AoufiiQYf1ge!Tces2Ts7h=<urRTu<Oa`*2T%kpCo5EDC~Yw
zK@e1jxnGOCbE{))jNbsqj+;7e?6^q=+EU<qgF4@X5=>3ZR{>Uu%V6(_TR|Nz;~s|u
zL?+eB&_qTPJ59+Q?1(yY$}q<}x-cNeo~@>%bwC9QcGgz3$eC%PtD)q8cMgn?XT@yn
zh$4)I9zkn>;3N8|(zmfnMn+E{K2S#DV2|>iSWCd9;!QHzd$QTHQWf(;BYm8V@p*SM
zx?uC*Cx8QmJg;(gpfR)5g82l8Gd6}CQVU*;)xL|<<SE13l|3%@<WXes01(=gM8SwB
z=Xh%?qcRIu`Vu+UTf5N^nl6^iWx=iSf%QeG;|=e=^s!wd3=ls*!JethOheG!5%U6r
z*4a&RVGxD}BrcZ;K^V%AZ2O{ERSVcGBnxYji@n<r1<lPNZLeGs+d~Aw3hC9Wr%W!5
z34y64&7#4)TdiR!F~mX5`K*-M??@unC}nAzpv=)z58e>Ujm1))Hg%Ck{<7F|MH-)B
zt=QT}X_z6qRO!ez6TI(gsS4QQH1e*r?CMl(RQo&)WU=jCi9Kr!_8G5THZe`FD=vWI
z`v#e04XM3sgmhOMV9r8%mAd$A2;vW={(EB@m)xB@&%}NsSq`ga1pW;~hFwJ9!<>1U
z^*5GWXq_}VWB_@EGHOOiHa2$D(1FsF1~dyYuOG4}EEIqk9BED+;Of8|)@hy`SO)75
zwd(Vs?wjKMd@p!K<E~5zJT*bbE-EUCsm5F=Ew;ry3-fx)v_2*tI2}>1Ivxb7H#OxI
zP-dlEZeKLhz;lK7YoMYIO3hr*6I|&t!fJpnCgAgGrp_PwZn?^Osi=ns{?);!RKcA&
zay4BOIIqNm(YM#e6Pi|0b*@6LOT(MP3e|5qdAZ)-P}nfXR!g6{Kz<nT@0p(?Hw3p;
z+k(gt#O%i3Q1c*fl>FFX%?li+LacAK{3K9j*jU^)H`-s-RG4JtCi@<>;b<nz%FX^g
z;YVhZWd9b2GLIwU*K|i9`A}-0bBDrOernT%!3J1$s{yN{_2LO~n}2SJfGl6aTqJN|
zAS+#iZIIi2TYD7rkUM-5d{2K+I!)c_rzhJ0CKDmM-efM+=9FRPLxXvD@E=IA4W~&I
zOr}hHkKeqFRm~cEd9U4c2%whxQYX}9abXMP{+Rgy1n^o@sKDicgdIHa?m-Xe>YV^i
z@#krgYry&uZ#*jz+9u1H{33P>_i#B_+h3*$YygR#qXgzxv0IJJ4Ndav_@WqAHYny>
zyPXSqMpNrRb@6JMZr_#FG{Un`X4rSQ4b822rhgY#0<$KWX7fRR!_g_fuHu>?;3F%a
zzhM$V=y1XJlJHX1H#^OOw9D9CJL_|NMH-lk-oWq4Tz_VmV?som%@(z+Sm0uwx1ggw
z3pOu<ws(H8RmiJa1Ix24@J4nh8Pl~=7RGQm=s5K11iQ#~Ew^_sSsZ<gMH<g35J%d*
zy+(v#n`VM;FA0FxBD_JCMvo8;Xmx;=UzWw@!{eB7rw|PYA{E<UM3v>SNr>&PDQ%P$
z-p3p;yD%x}tXA4H`ZI$$HY>|2!)XdTB`AQq+CMXc2T$8dSrc8cJPGg|8DeFvFA(Hl
zU5q?7Q&$hhXT7bZ7U5V>cWtn@CFQV5!GbIsy)!E9!FDNgxXITBi5T1lhxVjj*an8(
zKNRPop-LY1Z8s&JG4$#;Bac|C8pa!MHU{DRV(EzXdZ(*f7f=GWS9vs0%jFCtYz165
zU^gC%8;ef$S?TmC1N&ewV&3#s@|$>uji7%r5>U$H(IKtq=*s9iUFm=8ttl$eLy;Wu
z6S3v=?~&g{f10e-?y)DmKXb&Pa}NCqc+Ec*AkQKc6FJ|XP5{)BClv5#TBaKem^D<z
z2=ZMFwfHj$lW6a>s3aZ>%Z9pMp7p7yHxQQrGt?gpwr5qKBKIHdJ>oZ+gf*kP`dlAR
zvcihY@Zw<tRmfe>+ZuT<;|_koa0b5X@}k{Jj7CB{8A;?zu~-d6^Q#k_O}o78M4e9J
z@+WWB&3F%tgeR@Mk|4MVY@3pGd9_ccr8HFEuLi<O2m7<5N<C2JwE#!AionEs5@T{g
zug9(;KPB};Gz1z`_C}m+9w%^^@}@0!gavB9<sVYOhszyN1V0%s@OFSj<A~l#6Hle;
z<z0Uk89iX$m-k{KVDk|)0pS7pXw=o`F9xA8BZdziB<}}a^(;gQ!Vlu;Olgw81}iWY
zc{DVK^@j<Ms*wq9l6+*Xhm9LIM*fzjPYkEvr<Wxl{jpCDP~bfSztA@Mdx9dSs!TIE
zFP4+yky<SOh*=G1C6RvyKK9HjluvvhNO`J9VgX^px=(E&x|&r%N31a+pV_B%P_j&w
z&*M{I027&#(X|`pi}<v$;xhF=0x9}3%~7j4;=X1antNoy%JP-3i!Mi+i5$rHFM|kl
zc}?=Qzej;A<<4N=k#7RN$glBj%*J_06E8MEK>D|TNhJwGSK+}T-v#z$;67f3i&*-q
zUT)avb7Twqbp)8BE&Hlqhzbc?^;P{6k~lOp%GQ0|DGr2P-ALf4xs81_d|a_?8?S-F
zyh=HV?fe;a*HkW1Al<%iC@^p^=%uiOH_|&6uw;JW9s8OJ&<sM|)P1s3-%fS*BgB}T
zmf@|YX;#l3*}1RkE)i8iSIw;K(l_ZASq^Den{bcZ9NEo#qUy;LFK~1~OSxp(y{{@b
z2Q_9IWRF;!uAO?Z*|MkgoT&v~K%(6%nvjB^3$AtVc&0wV8dD_u#QICh7+9E<efz4O
zg0aJ;77epse5Uxq1Gi50@2mPn;QksYuMY6J6Ee)-pt3mb1)2e913oapPtEnJuh)vs
z@&oVNf|6K1WVsv^+g#pKF9#a{J6fkyqYz|P4)LydJ^*T01|p?{9O`dicEX`F&l0dx
zV3ZE)S^!8FC@ZFPxGkWg6$$op<OpwpLqx+Bg%3H>01g91R*v!oT5)O91X%!rsT7X3
z-ZJ$Xu6Inpv}nwj!9^pA<yeOn<Sh7X9G4*6RMDUnw2$}U*)c1#t$>!|1Yb^2VkWmV
zi`t3NJyK!SqKuZDlxDvUEF$-Mv7DUHNrgj-V0dFXaI8{J@$OamYTOZsI#bi1+U-lG
z6xAwu>Ya;g8$_Z<>_|EAr5bU$fzowIX;vn^d@{yo@)RtVoU}g6HceY)u0K=u4@}^6
z(AC~Pw0T3r%M*ebU}8Vvn~>MHbDAn3|2Lv2{!jXp{J_o9rg|xem5;7uKAOV#(}aXR
zzUlA^1qtcfZ97^1m!=lGUi$?gz~IyLMP+#ZzM*L?Ltb^gu}|v;SBh)opiVaw?a4~R
zXC(AzEhA&|xCaZQD0NoHusQ}L5Ae>wSu4h%B)SYVT2$Dhw+ymh^LqD`!S-uTVJ{hy
z5TnbXpVocSa~1|2YO@(rjU1uFd;s1qWYd|&Wq87SOS%pXp+OUu8^Cv#eSXP=OD4#O
z0Oddw=PQwsfvmVm>AkrYJfw1xGRme`g`AzE0|3hBk|5In_8_k2F)<k&KbBD-19`TM
zz{2$OMHm}Mec=iCcajvR5$Tq$h>xwI_usf^SqY~r_^zB|Ye0zv++gR%LGPeVdAv<-
zbXl347hfUM6>A-xZ?DFV9xoTfULf@V*$m82F6^4Ix*Z}SQzpc-4mbh<gTP|TMZP6i
zt()yB9GVG!24tp-ZP+nTq-GFaE0@Hws8a)y&=O_Nu$Ox8L>aQ~3^*@I5Vb1_<^c50
zI%TDa1eG+xrk|BETMq*BrQCjP0C!U19grmP5&|9o7eL>nG7YDm&>9y}m4>6q+p{{U
zpi*sM?LdO>@)D^DSb}w~sceyn4)oao`U060Z3i`$Op)67tGJ{VYC8uP%kjmoVs6kJ
zn^aLBr+XR_4rY8h<QAyJp?9mXy$uPGRkDY@;Ie>6B?`~Nn_3!ueJ1H+OW}{6?b~?(
zs26hG{^x(v6q5zxS}4u&*XG-yUZ0$H+V;n!U#)}=5(`ycOucC2Aem~PVpR))>Mg#I
z?(Q-wJjyit2o^(V#V(IC*U-Xh@mF{s)Va@*D--fTjzPa97*|EVicE+(1*5@)L)XgH
zebqusf6baIrbVNCbB&#ETv!OO_3pxllmRu@#rA`JV3Y&H`wZUK+f?8ONw>KlMmu+d
z#%%|rV#1XhqF0@Ynvdj1F>&00w4j1TlOG2`tVFhJ>=q90PhtZHz^+~i{^`a<?5GJR
z%|PuOux8Zzvs%ba0q=3cN6O7<O3Y{)f;o-8w*;;tT8;teKMmA@_eBZpB;|5znqa$8
zN~phXvm=yksFmCMcB-!vj*jX(SeCJyLKJwp2Fe<{4({-2bPEgBt(qZs#`2nwV2MhI
zyL@@z$Jwp=Mo|2q(vwkccgOOIf!+7`6pep?R16ArZ@>p|rBfmbA`wbTt!Os9FBUo$
zCVd17X662P0wEM0uHfZ<7W;q#C(P;ztMLPAC~6QGi?>F89?f7hgVKij_h0zv-IQJs
z^tE0!slWWvyRr1-6j@Pu!=V!t4#XUIL)G}N4fEm3z%&8U8cTy?0(=><C|=8S@-*5t
zJ$8&#OBcOXW*Bm384Nr#y*DWfHFf2|SUk=I$|k_$WmaHDQEesNe`cGQbIB1h$0w@D
z4?0LAuu9!B*Uu!EwOHEHf>PZm%m&JAnddXWQuV4>gCMN={*2DnxFm34w8?^a22BYg
zMwy#@VVb=bpH=VNMYg&jsOTt*4Kt7?3YllyqcP5hW)(#TOMF`D*o#gv5CF#vgiHIn
zX&PP~brn&pZJ8~q5?Y{@jj}v_-4y|2cK8(rpv$td(nmGkRk?9lWmAOKzfxAG_w(r;
zOk0(#iD%F_P+5+A!LruJ(y%<>u$331^WD0jh46v`cd$M|T7?cnG;$%bA<hx}TGdpb
zmJI6Vd4(eBo^`}DRgIusn{2IGULq?G*)eadnJ5o?Kel7G^ISIlG^jmz#HY{F)b#z|
zh)Hga>4OA-D}pRYn>~$X9f4agZdO1o^JsjnxMc1RkM%XH4dsjP+Y{*A$M85rQA=mw
z?dDrSoo>a={hM?;$OV#9r<q@Gb`AG<JOB-_(*i{2p5OXR%^YWDMNLO8Poy`jM%`8A
zalQP``?37orX+qc)&Y08pj1x<X@lsc;gC-|D^gP}zxVdW4J)ybXZ#se$GFzj*Yeq*
zFawaRu>v*5O5_hdG203-49c&s#!dXk=m0oRlEBc@=lmIfsUAmkKYKp52Q8FLU^wz=
z3*-fFZQ`=UrM#GWMNk;P26r%dDV{m)XH@>nvE1?sNarX}|0hE!cpzrri2I7YuF(Qn
zNN6Un`YgWhddi5hKL?3FtEN_7i$8J@#`n5Gu&_t2yx~)zW`l7k<bdCd5hqqaG=3{h
za<SWF+U0HQAQ$Nt-m%6&lo_yzzw5-K0ofYlJ^vDx4Lzn1Kjkk8yGUD2y{|e1$L#lG
z>`7q$<O5$E!OExsWZVa(TmeJ!UsK=eJq`fQmk$G%S<DsN+#(-&D`1A7fz|T2m;nd{
zOA?}DhTrJaCr_7;{k^3~(wI1ID>j#Sy;%NkeRi~#*EG~JS{66TKVs7wE2e;MR>(hn
z;NU7*^5c^jxVwFWMo;h^2n>=>6F$-Bk`CPTacYNY=d)NCs9#+JBcDFE4Jbx<<`-!w
z`nBT*gR11q*c8MbFOaX&z+{nNSONEKh6wV;>*QYn^RYGXnETqN@jY%5%<*jb#z(>-
zaF3gf@~uBpYZ4YR8<vxQ2XP%yL3hIMqI;z8XcFmXYCW?Jk?&e4Tlh2Qd(?+w%YLe+
zLJU3}h%4y}xm7<?#nEg=QoN3rH*bENCOSH8-Opuev%|2C&o<F9*hhxoGuQIA{Y<k(
zPt!TFT|d>jAzN^ZY#$vU84koo+IDv6=jyx>aPlva9s6~%pavbUARSJTo%*@!pxeP&
zpb8Wo?9<NvMr(iI85t+b6t{~nqg#Fpexo-EJvVoaK0&`i0jBR|w|=TV9Ws1`?4Dq;
z3G5vsUOZtj20;o8*~8y+U?zkh_eM^V5phev+V9!V;(;qGVa43b8-G6$T+%no-u|A~
zF*qenG;LAs7TC0pbv7b11#VgU_ET+c!(`bn&7sSQ0Utw-HFd)NzR>Quk5smm18nah
z4pJx1NT~z+*=6QjGNwSvO!YQMmMRMtVO2j!xJ90xtnW-Y$e&T9VIeenp&Z-=Vwm48
zaiBUh$RV**U>DUIG()o==6PWT<3rO3tm_85Bc$lV{27(G^gUz{=;8fb>x?WC$cI`k
zN5qhN_!)Y}^f>gNKQc{Fix6O#HaRMuq3{qb&-f`vr(vjfodYkAHF8W0GXzo%<EoGK
zMs#%LrBBP&asKRfT-SF1Pws&7@%Y&CF5wthvYKutRPD<N{oLBst#&j-yc2yDa~}Pk
zgA^{ClY9`;6*JqemXqylRb-)%I>p};lhwxrbm>$(iOlGcsgfT3oOvV%jr32=n&G2o
z9KePedMNgaPar|(NKUL43>zR%Te0NYpwK5zmEM72U5W{9wLtP>pMgVA4rmw=EzWj|
zkQUJ2Q&Y^8Cz@*;co0Q63*x<cec8nXHw|1R#HKbWjCqb_k%vC9h)JL+$QGEUWmZ(4
zeM%v7q+i04E<=%VeErjef$&%aET#~MnBPic<>fR7h56N(Ky<nt#j}d!jF<o<tjts?
z3b=s_Ay-c<8DLGC>tL$^;TxEaV6_+H92B@V5O+{5ls|(52B0W(YWKw`gUlX>#D<y!
zWV5+Q42_AZ`2g;Q&~gp)XL@~Laf@++hR2DAfkCZ`*oVk=P29HAeP#kB%Fn3$PDdj;
z6`vIV<AK8Lm}$jguk`AZ5&oVCX_Bp$FH5TEk>0bBX3!lq)X*(sRG>9-L5z;e1gCzv
zj4>FN!z+e<CualxTm0{31^PeHXaB#^M#iR1x2n|4#Z>GY(9P=++r%I{3Uuu<F22FB
zrY{gKeK{v?Xd1Jib^xND8_npcf*ocI`1rsl6@~4#IWJ)fi?Cr=fwR0F*UI^U>hMqo
zhXTK<3(_#S=q;w{Lhq6vGLH*TP$qbn-K}Comv*@*CI#|AZy#LGa&bI^&V%V0*28D+
zk~mYW<U#Y{rN&^uILGA6Q>3ID7D6rOf+-!5N~1e)h-8Xkw3o89Mw;QMY1W+lu#A>R
z>qjs!fJv=Lg|QPg6HBDhU$f3Oa;!trTq;%GnF{A*L{(N@2P;i=kl6FilbUYFh%)gy
zb(ytXlsGgp(R-#>Ass_eU{fZ=WN^Kbeu1J%Z4isiRuQnB|CKtQg9&MbBMI<K^jmL`
zV;Tna43tbjt~{e|6b%U$NYMzyS_1J#m9=tNEQ|s?F@dbTiq(y2Vl=T>e0bKF8`P;n
z^0_8kc^&>qvlA$jc2iCc>IZRAMCRAYf#qt5QJn$qZqyq<_cS3)u^}^!RVDS2a(t@M
z?XekYNsHk0^;{5XL0qQ=(q>V<4y<a6T<()MSShN=6>$yK(~b4YXn|b<(+C~;I=QN!
z$J$W$24^7Udan+&gdPV`N9A&juhlSJC3~J#Rj`4a8W82$_zd|3wF1#~2CnAH%B);(
zgM+7o10?el;q3k}_6!m^#kCs@)(CO|tbY`Lj4CdYAE&Mx8X0sXKQS29mdcGzf~iMa
zBR2(Huwus9SIEuQPbSNBUbpn?G;J%cPPNi$NHY^OF}Wn~LI%&0t!<N^Mpqg}HyM)5
ztqwp4tk^)HBgoWk*2=8wT15QzgkQ{mLHvrCu{->Go>qhnU=Ew)PDc>lFytsHsGw(+
z0Kt#RUGW}-hvL8jz0Hg(SqCQS?%3j?Fc!-_0cD6Q>Iwvf-o5FLT7?b2FRg)5NIE;Y
z-zdTGfpEt3$(zS8x%{a6_&|W87}*M|s!M9*=dpdGo0}StisTmwTPeDdd8Q?S`C)z;
zy{aVvso$|Y6GVQM&Ib*q?gOc2JUf!V_UYO&HTJ#Ld((@a{$Bty^=6A?y7z{2h7AH;
z!F8@T2Y!dRC^HNf&Vfi-D>J=oGZ2w++8_@m=;%eoU`3f_@WbJNN=9bKq6ajiRz;)C
z3Cx6OUc$hFxp8@bb*Ai!9cVN7(zr3tSOi-3X)~E0jp?1p+IHo#z^7n26nNU(Wnn@B
z^|!_~z=UyH?wHb;EJ_doZ2<bPmGOX!13lmom63Muo6Hl9DoX&9C9xnxu1?BKmZjbo
z_s^1jYgKP(otFg$V6vLiRhGx%OW}}A?~N5{qIS^}r!B0@Fl}W_0Qnf0Pkb?8zREx{
zXz-Atb4HAl)zK&FpjV-~l&tY@tZp;8w+*t^C<|<HC9=-nJHo+PX1c7mXY>Ot%b=zJ
zBiuH`6$4EzL)4++#*8;6&{>cSW0yeram{T?{W4gJs5f06^7Wa_Xd|jv5BfnH?!yUq
zRBxHdOrsR#k=O{>yGv!W_tPEW)i8r?Ma6pQNbkYq>rKQ!V_Y)WTqe_eG`+WLmGW+P
zXSB!Sy-GI90il=Vdn%p&-Yzdos^zviXgHw+_V+wuAOJuz(lEMv@*-a>kJ~{iF3U6^
zM`GL|-6S`8!mkX@nk*gF3Cg8l9$2H_xW9{i!l#vuxTLU8PdeOrE+f6bQ?}pC1R+n`
z?+CoFkl%Y-Y5}HW6h_}X)AdXfA+Ykov+3D-Ed)iUr9b#y*@YRlhQ-tXduzg4U9us5
zKWE1TmUcMmArR;JpqWMTf_DVdxzZa(UsAT8Cod*+U>YKz2u%~}s`gT#!vI9H$;;6W
zr(Ky?RWj44^(O-~O^J>2iv0ms1gIEajUI-UKrsGuAh#y5LUJ$lPJPW6LbZzpIkm~_
z(Xs|&935QMc>9Lkr_;*h&F_JQJO=|6QD;ZqvL)yD$dkA2%uYcqRYvf>6X@o#ovfH?
zyr5C>cWvSwt<#zt<vpL2$5Me<A(Z>S42sNsdZOCoeQzCZQq&iI5c>n%B1X<bVB22<
z9L@EhxF328QXM8fVLZ}D0c5C1p^U?MHjCC;`CEL80<m@SaiFti#lmr=V&(7YV?Ans
z2NNGN#o9j%Du9Fh(|#?0hs`H3MHm*SR9S+n{?k|rI43xvDVP0Ov;%|6r7*AE=LRjX
zO*jo-_`rAtY{q$2jAWYVFJnqd?q^&6AN3mfDs@T&6I1dmk$=S&F#Y`3-hg|=ELC*B
z|Hk)F`BJ+*c(-cQT^l&gZ(}&jlp$I4;ozpU$-fi$SkUOZ)VG43tc*y1)w04BuTZw=
zuR2szoW%h~PNOaRt3nuQ$TS69Lk8F4R{c%Y3)>J08|_Y5h9*?PT<6Nx_6=~1S+?j~
zwnDb)uiDW;gU^#~`-eNH^GV55XuI^G)sm*0>h|8cn<JVc?>gDRr;&^R8gg$gn@s$^
zWB*QORct_dnsV8xzv-2cxDIUA&S@ZXV?;2g>=G+SLJcN?+|@p>0qQi#ZvFp9a~%L4
z$#!<{@6MYQ_29x-y>kzrJwZVPS^$}bG<u=z=~GZ8Kfs5ZP4@C<bg^QTAAJ90Z<|JO
z6MVd7pZ==89oQoK+Mac^BCYLywl=tDkU;MrZw3vM0}}8eY&`0h9+-x2=TNYpE|nj|
z#^6}40gf?akV*Ro_3xYvQ2yW1SIEJRI|w>E#Guj9S`0<mp|N9dfkJ*?Ijn!@9y-gh
zNSCjrdO<x|-QsWq$>#eYHyqKmBWe*O|8E($k93eQIEV823TTs#it)w_I8ToDZte&B
zaZEIZ5`%s>C>tfm_75E?;*Xl^NSnZM9%r*e;8mjvAjc=jFH}Bp9;^x{q)Cw{ruKn}
z0m&1e=&cPUs5IClCwc2&-H5~V$6F*P#}49NprF*1YVs8CC%j`QrJmZ~%_q?CBDSc9
zp%LOL7?ayG{z3}gP14Jjj_QviBqt5Oj@%fid{jZp^+U#lSDU1FOdX_wTpNi|o^53w
zY@7Lhe9cfmCWgvEnhPu#uE0NM=5pTHOnFo2&t}8QINhP%$EQODzk6#t?E0qlAw5A>
z`q}K@XI%|ezkdQ1Qww&tBAk}yZB=m=VJ)ZoWJ%$}@<5vf{C!nWKO^3QiJ&V75?rEB
z8s}+aY?C5e_c%CO4)E@^P8Qa{YFPvASx4(&1ah#v3^riU_ohM77bRc(ndLr2c`jhc
z5MKx{Q%ViI0RczIB|}qd)nVhdm@C85Go^gWHV+RtWi>GV%!KpJa0|7G7xu`P5&kLH
zMROgi{EiHihuvZZltNStt&&mE!^CFPX*T?)6+v0EF+n8V87eDhrxjSl5`U_}T*ms4
zM&fDEX_Ml#f(@Wg#=(rU0Z~>fgR(CDJ9khrP}Z2ya@*&oUYIkL8G%;G_`p60k<1AL
zKWSdy=#l1zb~(>_oCsk$Bj<Yu#)|?KaW7q9fT^u0l?(kf6XD_Rg-U#eOz@*H2lDSM
zxhM_>jW}|#kFA0Xc8rC2OsT-^FLUG)KP1T1cxh_N1l8t}t<hun0QE?v#G0OP+zC<|
z=Rp0U)J;xTv{A}zejTlu(rg8)ywhY?9=k9a8k?GW9Eb|*YCL|e{Qi1*!%3BQZ!j=^
zhYY}CsZNXA2wi4P`gWxrN8ZE}1HoYVhu^@Y7#osmDcB5zcl$?-Y*~tBGW%OyJi`Ga
zEt>(=alJi*bssfa7_N&uF%w4*j25XTO{%kCe`632eSjLGA5H2@ZfcQS=Fh--Lg1jF
zyIdO60+9~^SVHP-*p-b<P_dYZZ{i+v`azc{O$mQDGD-lnmo83l$&%(k#X1znVV<qY
zKCn%b2-hZ4yn7yCoGW~eoF-T<i8}oyr`pEk7NAOVTI)KDq!GUuuZsbhmb#)~2pu4}
zz-|Uyxjb&nvc{GM)P}hta0ZHb_an!eAqUBo&hSt+p=$dY2!&+SwNw=tHiOJnX*uQ=
zyBeE)wOthrHDI}~@m|1nL2~*Ze#*5m3&mUtqm;7ff^(hC7jy&RC*bq1w^NCVEfBGO
zn7~e>rD`73Rb-~zU{gS*sB`5<hMgU)P~um~j|~%>E1EJ;PRmdFyHf(vL^YN)!x}6%
z#$;d!VBUPB$5g()O>VNqYBe5><OAZ~Y$IW<jdDwTiElg;sb54t?E;f(yAjAQw|Z+E
z!PG^Tp#C;*%{z(S12YjXa=XnN^&T_LRn>Aw9H?x4$)pOoGY*uM{K35L@;=m!jE4fG
z|L*k6^gi^Xy(hk`WkfCm9Pf?Mv1Qou|Bt=z46m~~)}7ExLJgt!-rWnuwyJG#!!nov
zfoy3@mavM^mTb#F0tQk@LP?{8Tx83=H@O8Yqd;f@0)YU5Py>V#AOr|--*?K|@;T@J
zK7UT`J$}%)zi+Ksvu4ejKB@-YW;jI5EW>GT4<bUbLIzvk5k#rx!!U2L4l%8E?Q*C8
zZgvcAujZNSrOxAXSKy6dEWq_f_n54ZyN&X5nDm|myxnimUgX~R&?7<fl@;$xzi8DJ
zF2X6cHLj*HuvN^6l^|4$og8NRBxqs6gS-{>31wDHj4PEvPRt0M9auXD{X8|HzAYwH
z%QbrZ;w0N{&hgdrOZv1B*^ghWq#D`WT>sMeD#Hl13=j@dd-~EUgwV%nd_xWWGB2R0
z)IwhnE#~_dQNBs9eH<hvg@KM_f$d`srHB!yxiG$m-<xiloGgk}=uR;FmlqRV-;0Cb
zXa#MQCH^8*85N#pfLZaxcF58+1bt1|YB=4E{FcQ*Y@(@uc?`(XLpfRDkh6?584(S{
zjChFH@|Cu2<4|HdE314<teUMF&SZ7kI$X~R%RE_QhZBTdE^Ff(RjnZ6Lz-L{2mnA<
zalK=k&49>C7J&PFwyY1j8IOqU4K^^ujdIx-gG{Nc#eD_^Y?Jfro_%Dq>1%}<`CY)L
zxfp7ol>41Kn#@K4*$xKX_4y={2W&DVxYWxQA0_e{;2f9EfUKsbMz*GnVf-#7)abtO
zpdlO$fYrIv<s$)mZtX{{NPorxI=sj5uRLVChSLL^f;=4fff-l3%-|+uRwlZx7!>HI
z3;aBi=rDOga9QQ;qmKqDAWo#0Jm%9<M`j~Cwg7!RE>096f*rU?o=9}HhK?2(J{dSg
zd(0pG7pw$}B^D#J=E>L^YJ<?fvhtKc3A-9BNlzzavQUL(*8C}+!Fa?P0Mx*r@n@_!
z1RIa~BA*S;F}6~L^*<-PThAE81^W4aE&$hDN#60%%I0|=*El9)2^Yy<9CKU-VPe!q
zHy6CTkTA=#guMD$6N@($wg;B?7ky0f300>6|4V@&<LD5RXs5=%@>hSG<OYuy!Z17J
z<yhH-S_IPN<duZP3>qIWdh%}x2JUW(ECI*k4$hZXjnd%z)#|IS*`H{zpceht<KQz4
zv_;<VGZ4e%yKBL^#dtPF_nSfEW#^BSw-VH_EGiBb$=?HNdIKTBz8x34WxpVRP~M3#
zn8EOF!ZTS{6mVv*9%?t}{d~4NUO{`1mzkxKllRgl8(C9VSGeW<SQN}Y*%tXAX0Mr0
z-_WR5fDhwQLt_@=HVp&+w1Yje0UcsLvi~kDC^}a@4nRv_j{U@rN<mp!$=Gr7FF%c}
zioZra?FNtbURV1zFQ|VfGGI_4(q>T^=CjyHh77?Ql#|cn7($w=moEaX2o+{P9l`?T
zKXD8*ENh5H;a{dHNtt1Qz&#^h`RDVXm#`-CI=IijPGD>V7__U*78%q<s~qrz9%kPd
zWZ|YNX7E%a)?hn-#xn5;Eg}`j_JcgQ6!rm6x54D-vIjUfK+fL5zM=aXW^)u7f<Jx7
zL2isIVwER$%HGLGw#Y07qs)?>{TVZ-7}mQ0PTO4usjae4zy7l8AgeS;6rwaRjVu6j
zrm{@64w&$b>x4~>Slx2BL7{y%V2Mb)+1+R0(dTu>1cz_>Gh-<Xe^@Es@-^Zxf_2ku
zpt}${z3eeaZGWR`WY1U!EyH*dwpZC}klHudJ_$#3Z$mP2I#$%mKK?qT4)Px6qVGFM
zO}>b!$;r0^QONB<dJ;VU`wdbfa3*^>JbYlA-|4ml=FL*f-#?9NDirHV9N^E)v;p}Q
zrq%<~vY6-><|mVaP##<k%8vu+C@L(IgZ&%oN3fSd`C!5+tttF>h8$w>b+&40g$6m)
z$6(iTpjZPp4>N=yGMFuj4)-^-`aE|Eyz+=3DoUw%X0>j#@*`)IBMpf#vBCs!ln-FO
zYYs3|j<zZA0*FqIF{&9bpsySovmu|I0t5|t$90>H>LFtY+B%upNjYny93OO31|b3g
z7jnWNGn2<5qwh)U|H+9y6e)@sDmVe$6r~g-JIG1#r3&QSa9AjDAlJ*u1|#||$|(+G
z`2KiaPc^E5c9oIS?A3<yjGXTCQBPi@Gf|*q9Z<E`NsquoZKkeVdd5P~4+XaKtn{)`
zv$!dxcVIz-g*Z2jRTkB;0nxgB9CTIPv0S7?aYC@7zOm=<MKWKxpTW-~uU*XOukWAM
zgnUF?9XPczAa-#?@dz21rdNFz6T@=>DKA=Qo)Iti>e*Y)9Hiz^c&9N%R?dn|MWQ$f
zaV9cr&USF^ie^Jini45$53-NdnA*@HgS)NEdcJs(2^o?$3)%n5LqfSONE7n9QSnd;
zV^~TwjZ!qo+!7W*#kkX`xW)bU8eCF5C@!<Q@`kK($DEYdt~4<0p){5~wq&Rb&EqK9
zQ@hEgGR(o^7Ll%ILc|#E-<b4e$phUo6y%&>F^C#KEDDI?TpI<YRghGuPR^W+Fp$R7
zA|iQYpha0gn+C?29x-Z=>g?qxdSlhiax&Tounq~ypn@^}DZ}O5yNHtpqk^1gG+mu-
zl(E|&m#1f@l=+8b<O{Jfs<3vEjq|rHuHH<(=f^XKq2LfRwDAI4LCGkTKpCG#TLJdf
z2KkxRX<iW*`q4s^%Xj1GjLcv=FY-?)y?H_cK7_(67yD$)8TXC@a*4lf8nw0E<l7zc
zJp*F`Y$kH)pe~K6G8sof$H}lb3hmUP;VL-DTc^uqKFeIh%TjxX#wy?US@I-6^BVK|
zGeL{>2wp*d8Jo1Tb=)|q@PQcjL$Poj1k0=uSp!NSRZ8Wcu07O7TLXVE(mp4kT56+I
zrB$wi+DKl%R3@aMW~+a1CDP)mZRTtxir~ViRuf1>R&Tvr9&j9cwM=xnB;_!PoJ`j^
z7vwg`Nwq<j!_g!{gG=iib+G$NeT>@x@kbgQ5T0`8r81FVl0V~^6vv?r%N72NtEN=O
zRL4f&1k129`m$}Kp^tntAJ~@LW*$=JLTQRK2|<J58WcCUCJTI7EX}@2W<($@LW#!4
z&XdV$`{v-1qli!Q0@IeJ7!nJSGEc|w)L3jO1a)cg&y@G-BB)`I5~ro9mSK;qo}qGU
zxiUC3;u|L5h+pME;}(amV#Z1=l#UzRd#mK?G!hGqm;zvW$q!7_nbs^nG^jB5S~!uf
zv7=859Akl*Lv*cU3!h^U?baaE?O6&hng-eZItPZ@>Sj_TE{N_T#q|!1SyLfVVKkCF
zTF@9U<4fg7G5r<IQ8nzx`OVg#iVg_cKZ$dJLS&fvr#>@Bqe{m}I%P+I0-3{dW89Yo
z<&7EU4E`+e2>-<t7(IXPb1(~%zNO{RPkv#Csky!aTX&PgZboBnhF&pX1UJVo6veOw
zD4vYnXO8ME#)F2JmC3CE0J33llymY+|BBMEW;1|t%<%QAAd+$v8JAx>VB9qJ_Ja&9
zzlp)1b5E{8t?b|0fhrg`ZmirE$E^S@NK|jWJy2TFi1IaXLm(~PM<uJ`g}4}daA&+$
zMEPF5RCmQh<tQMMX|9mF^E*hT*<)Rxj>|nUBK!*cWeMQCH$F#<l=eltFD;s`Q{%vp
zPQWBu{gPFOtTPdH)PO+L@WtaOGvW>{Ypju(KBL)b%v?7sSYA^zbm59-PzT~@X9u?$
zT`X<!Ms*GD#6~o@U=6!DhVwEjggZBIuBt!dn%7CYKO@H>M_2-mF)t1dYm&))UmSS`
zD}hjXQ-7%}2x2|Iu^gnd&_1hc(AtTM47jG6NjX{UL(oDQW#~-QSrUkMCqh0bb!m`G
z73#>yG8+xTWmcBQmS_%uvS(QlZ;u2S%1S>r+zD#=th_)Xqzc7s_Eq*a3JKGex!Pec
z0fhU;<YLC*tnryx0GsR>%B`%86IPC(wu)+5XJZwkh&qg$9R?@59aPHtgl>k7%>hfD
zEgO6;k{F|bnXYYzYGWW1HxmZh<R6puF<@#I5(75d5g4D*hTW*wEWdN1q^3#kHyWZ3
zI42JTI@q5?w#3jC4NP9?44jiED_&?hXIsUsjyDZ)19VYyhs`3vcH}|-o!~&Xf-{R<
zT2b^G05T7@%Yi0oQR>_XeEr_wVoQlSc_?1MUBd8DsA%$VzzSo>1o?vl$KTwaipgoK
zdn6=L6#hZ^PgI>i6@y2E>?;~_b+p4O**_LTg6=Ej@i<+*5Z?Ag`lDAb`D3@q7%au=
zeOBf0$w0p)#PVNx%0>mV+5p;kI@n4HY%<^@@T~mFU*LwWmuF&d)ud>3<c7oUEYBJV
z)j^)t^w&QptS~5oSLIT9F5pCPyF4E>*NkA5Dio6R4~(XP9z+?uV3S{f>i)>!dNH8W
z_6lfQ(<Cq1h%BSrD1WuLVPK*|_hko_ml|z(TJsUs^fmH|&qPrId8Rx`@;85bo71Jm
zeqQyr%``#7%xZbf-)2QlIz6o{atG|RuLm0)nt^`}ops)b+m2yvDju+?k```(QU6+b
zGci*uFu}ebk^0I+eM(pB<gGNJS=yLP4DfNI{XKsU*!B4BghlWr>^{Rf-|=VkPP1Iq
zdcgT^JcC8t^smkGkGPF+e?X(ek$cal)$Ic)Xmg1%*&y%z^b1-u^WrvuKnNeiE;Xog
zoifRXwoCL8*2+KqOB2nk<`b7PUp|U4!8d7tU?z<5k&hi(<u~{RmJ|8J5sMrIS`IJ*
zE9I?amj@W*XUf0gXjQZ{HL#>GtHFJmc7ri`ycu9o1^u$u3gd<TZL-Z(SIcLCjIw6L
zzE8$g@VNuQTZN?)7~1)Tk(`$8>m&d1v%;P>OXW-38DxNHobXp^b=K9AT-;XobsUQk
zAaWUO7C`I6gTz=I%WwEIx(}K5$IOxK2Aig9Xzb?m#_y|@*`vd^Y(H2ncf*TD%MOFp
zAXYTCq+nc$>}Wp>g`Tq0V6~%FSIEvbPiHHo3i>j^jdvN`#T;^6&lY*R>tM6OFx|+l
zjk^5{WjA{h$zU~Gt_QE!J^mhrmYBfbH~mYJks8gk%s#3N$+rfZHPd7ib$DPyXH$Mv
z*=!GAKeF{Tz=&*l&%y4guFXK`tEQE1uNVgB)+S(K*o6?;+7ZyYK=uv*n4AYHqJ3)~
zf?t-%K52s0yqMs#oGJSbc5^90E37`yw+&}K`u3Im`~$`YP$DFE`;I^3<-w=`MjY)Q
zgCIx08D{eeIUt@)fNe-~R}UPlW)<KKR_7);Xs}u$ncO4?`-)g5fpm?!0YD6`t%QEH
z9FjiOW~zre1SWsZ?Sw)ihb5@!4WgXgCWrg57Cys61;@7Pf|@rcN5nU%k3fIca+n~G
z^lz{_H6qN1@v>V@06EIY3)=4J;?Z(+pjxk6)4$Y*C&#2;*lVY|#&N9A%9F*ik<gUo
zIDgx^ZJC`~RWLa|sH?G|yaBB01p6B1FX+ywQP>1c<HW&kdLD&(uISYyCmFtYAnUUY
za&n*#EzXd7bxOkaLVASgKLT3pw6gx(sSb<=AOk>?5vV+zmO!V`5lDjwaJnG`<1j9)
z$2Q+GFp54WR=w`&-$II0#G=TP^h(cWsgZ;h>fS~d&`qnPk3-6tSOA><E>oGw9A-&h
zpDpYfbG48LV;=e$R8wl2aAc%^AOL?u>>J?IVy8Npk;4e1fj$*4OV)X0!k0y#k?Aj#
zGZI}O&yhfA+Lt-#+3@hp1gh5DgLun@Z_>9A&LL-Q!^grL!+2XY%!BAxK)5~Iz}A9T
zIT;kR20m3MgM$-f5!h_b+#!SAz{g5nU_KfbRbZ<lf0_WrO3z-QuUHiutPW02DRPJ{
z9R@s!x&h7z<1p8QCW~z&w%h{fOVUQsaLaprv6LE{IPIttWvK51RZtqcs7aX=J}jY?
zh7Q=hao1_FI{Y`i0K;P|(AOPb!v4Na&M`)P)n#&S+F#a947r<D^ASF7)TySLYt07i
z)*NZTk4M=T8Re+&=7~0jixC;=V5lE$Uv##faIB08sGy2+gDAqD7teqq+NNnn8oSLi
zRkNim7#~Z{fL-TgTpUlWYK4-2=MV0pE3C71G?Gj%Ff{h**;mFVH1c>EQ7edYp--ed
z0Sl$E@4Nm?HFN@mXBYXiXsZm1m0X-gjne((<jm#1B$ixKBHuGUgHZBJIhn+zab!^J
z1C1FkO8~6XJZ=0d-#5rn4@olVXR9g1d!jtSw?NqpIH^?FC$(^!WP`X73(!w09Y|jA
zJUDrQyUKr2Ng1R^fyQWSf=$rbT7?j<Y%^-|Hc54`%Tj2bkd<o;)lh43?k|t^PHT{f
zKCG$oW+c_TN~yKCP|d$m>SCwLD{G`a{utJRGL^nMlV6mn$)q4kwpP$n#1aQf)uhWV
z=~wvgk&%RqV@fe#T6X_8#+bO6nF_NODp=0Y1aoNv0=}9Ccl7|}V7KAOQXPz>xCw8o
zKISenC0NYN!fNDX^Q75VYjz2)nE{7LHI>OeC#6FcC1r}jlpt6Hx=f+OF*Puawmk5&
ztF^tw0Gj}>72NsLd>Eb!nkbpUaiu?_pum^}it$$&0bF|0C33a@OtqeHQmyzw+H5A!
zh2aJ9-SWf4NZ^FT+pLCfluxFyNv;X*sKsftNXfPSNw>RAkB&7PStrwj#0yZ=q)x80
z1reXP>aI5`KyFN5xglOeT3usCev|-*3k0C71b9FGAIOO}=L-2r(D{~|aR7d5>qoBw
zxzWK>Sp>Un$;I&K00P)^<Yz&x!$!)_gRC30JE;637(e3|(e$9993t0E@j+p=-0VCL
zLjgt;IC-^Ejb&Vz5!@{fHFZNqQnHeY6@R=npdQhT?sC6OI|)G$D{hC|`d8^0!<TSw
zai9M>FoV!3#>}yx#Bb6w$^tg;Z__rjLJ~+9I-LD-o4*Yw3}isvX1U$pCj6`{L<+OF
z???v<QX<9yrD3foyT}zo>AW++2-h6GBn93T7jZWuh4UNt=@aU^W6LKYm?bCo*p)Q=
z9Jx1^qaukFO1Uql!M(@~{-u!HTjLDXp#3gS$c*^Vuh>lx8fN-4E4oYGwnk<dg$%E+
zlG(;^vJE_gZN64l=2)HqdzF{k=EQ`sB4t@2d9Ks*T$sZaNqd|RbeDy7ng;55MlW0F
zt)CwR1qx?Ddic|FvLH>#6xBJ1$Y5A%$|9770ls>gsVU8oMK);Bi1W*2ae$PmEpKM=
z%_Tv`^w^>Lz*65?tCB})(fVm`lw~$F>(6JJat#bpmF2bxBSy-Kpj$OSQ@5dYf`Qj7
zW8dHsg3h-pL1<P7l!ublzAESl6e?i|(|U7IQ%!VO<2;tu3C5DMOqO3e5N}A<+GW{;
z2Q<(QckgCd>vNfyp?qJ~88(_xep)&L7FcE>Tf$tkO4g@u+Tcc@(Pg=8uy0O2_2g4!
zqb+B=H?-d5pc_!`kb>kyg@;tT*@q)b)tCxA`<>BKKjiq`Z}e2(0N1uW5NOEL5P*VA
z#&ZOM*%H4)m10`tH4uyE;NoB`>n8l(8FQ)imzU6H*=nG7ww`m2JeYoGMkiVuL1gS2
z+@)<%ZtUFV|H<#uT&$aBtk7I$8c{0`+0LJSv^*Sd>M6>*Ygl;*HhK9&;w)^MX5uV)
z;B<M!e^JhXH7=vX$W_Us0rd&rm&g3jpwufg>Br+#pu_-$X}$QK@PVw(frYFZYx!dk
z-?>`s_esZqbztFTHcOteEf~$hBJy;+fWwUHTU)PzjPNG|2FX{O<e7v*`dM(PLJxU1
zju83|HleimpB*7!3smXV0B+hQ0;*x@dCtxtFYfvr(<YveLkWAl)<FA993~JA^R3v!
zhsazfFBmea5J!cgH!nIgx>mfwG?bV8S;!Tu@$9cL9taEF$#D78g-Ct;W&g#>15;Uu
zAk|kwwCXu<p!_W%SX-mwaK*jxs>#;SAl8E+!Ll!}8M=D*K%Ua;fligsRXC70d@NWA
zu`Husb%V&8K}pI(%#=Xy4sRItVtOl(kCZdo$eE}4_p~%7{LrbYDMYx}-?jx*RZ_ov
zC+J|(mK$Iode=uXRG5wW9|kw-fmF(S35jUNfDK@Ei}&qWXY1ve$!Po9ScC3dA2`I=
z02O{{-2o4PPE6L@E*}PIWTqfFpkDqNJ54_?uaJ*oA^=Y97K!}waU%F72(bFZ->`#4
zCqbM1D{c;UCIKYnmGY_alybym|J%Qy&yb<7h~AUW98G3Gpw}9_jd5-AdF=DK@cMrd
zpF&)QT*4q0`A=XN1mg|Y{I35pRs#oP9sFeSmA^_?1&=xzl!*mj8~)*vK)1ZC$dE3o
zTL_BVP%YmW;<l11D%n8FcJTncbx|R@Lbe~G21E!S9l*~HL$<L$YM4KDWkx|;Z#yJ?
z$c{sjH4_dMiY*Iery*`wLkECblmkXf>mh1^Y?&sz7)YJ1y?XbRUHxNnZH8Gfd1<#H
zVSc1^#R0A1(M;Jrh8+n>ydnb+!#9Vh@slwS-x^{T#%_WE_L(W_ma?@+8a5)jn9xVJ
z+MW)CHQ&;^PxeZHF{@k!FWEcLRgaq6#NR$XB+n$Hg6WsSbqQg9U;i8-<Z7qf03X`7
z{cRWBwbh7i+J62Eshey;yB^R>CEppMmO|)>eP#cc9;IxzTnD?~0Wk?y!SFg($~|z1
zS{j?s=8;8-4sxWZVGg+DwN5FRCO>$M<luxfhT2hiVwi{Lq=1DUlF(qqahHRT4>g?k
z>DOBhGkWPn<yi#x9&W&PwjOn<91&ZEUKjPOPAW$ZQ7dz2>qzUCe3Va(MMKJ9%m`DD
zSaFi#%qBV7mJ3$hgkAyAZ;y%fSAZ*^@{Js8fSh{jQF5HaN%etVKdeKJ57==%Q0ixw
zlN0Qd3RvM!jN!5%(0TxPlA(?v6iskqoE(3aOhx5hIc10&qY=&p{g^`JsYALfpl3oQ
z@;uRdR-GTGZ37B#j7@TSe4K&0g;d-LEz-je7qQKIKPB%0dJS>qEiXNtGN{EcArOip
zP(^vCm#w4)_WRy;qF^M=wn!hJ-a>`jSnNUi`mtua?#0qCjhU*CJAwC47(=!v=>WE$
z0k&<(aqYnWZIXeu$)m;=l*t+SEd&=+_lk{#|7INGnfc$b#U_22`f6VEv+{JWA}AeS
zN(gdxkkTc`$e<vy%bU?IZEz5lO6$DUhotqW?Tvz@%_Rk~#W2mIP79jQBywlov`I0v
z9x+0MEd+`TsUwjmRczmmugoG@Gm93IC3#?R1(`)1A{r1{8XWCh#Ip_c*DMl`22Pl>
za7V3>VTRq#*80jRIjxw|D8m!RwRa4(=7n;O4Tq9tU@_;$3uYXGs<}W$_!Jf@N)?hd
zrC2+CWGpz_09OR~|0sVoD2F;Mxy&LN9pqisfPTO-CdO2@M7|h`@_B}nY7pAk00!y-
zy%ZhL<I4OQStM(YGj4TU>`~?9hFXoPINt_oo|usfd=F6n1^RwujzXdWdC!74F+LGZ
zz5xaYXN5xa`jQKWxZxKE5zGY<9AG`)4SG501i8ou=-!Nb?pBtERTr9E?5ANSOyDpC
z8Xkhld@n6gk0s=U&2p*Z#$2RY-4yh~`ZJdWzR`CYmP1rajL$VJm7IQwWYU6|v#Bl{
zoN*>QDL<C-U=CFcjpa321kg)`GY69`HKkB7WwrUK@-WFTYl}v-_cW>WiS@3d@DFXO
z%10rEL&G!IlE=$3@K!6T$pj-e`cqn@+F>Q8SQM8~68WrU01t!I1l7O}2$p%d-ReV!
z%S3}^SO#)Yt)C(omg?+3`}G(o^?7K(qDh2KVYV$GCze2J2n4Fo%22zRGAZznH=^*W
zU7S}Wa82oBxIIU9hlw=CoihnmCwP!@!Ai<tD3B&61a_2KBF(XvbR9J!x?pk~3Eb|C
zj^eN;KTq*r%zzr9i40nt8mu6TIx;P>7OhO9C9!2%yfC(*S*~;d*i3VoT;;D*XklYN
z+WM~!fJ;z}Q+|*JWQ%s*2@B+h`DYBtg|xI;u1RQ7bsg#oE&0sq1*~9m*9OJ0z#7sZ
z!Dpt&L}U5abqN&4PMP|4eZ0|gV2>VhgHJ*84X3s4$c)icr<(jISiqK>KzldQOz2}$
z6<`JjrInvJBF$76;X7RTPZI=`H!TO?Mt`Oek9dY4B>7nymGT470tMpdMsy=eE|6as
zx~r91Z}N9p)SVq9SZV%dd$%GtS#I&&V95V$xz)ae<UFHMe(68KTm)7BS3XV~{_60n
z=GXo#u|aIjZ)|L+4)R<7*{Wb0AG$4lLRTCmm$`C#Ft<@ihd}<3<}BS2G*d8gjNIv8
z!SVy!1g<TJN_WKp(QQzl4wgsH&E4@fmW+UG*h?14JwfPYsJBJ`|Ggo{l-D%mTIw+^
zqx9~J*P2n-XVRn=X-!+c97>H+AGR?c{y!rK1hG6hnQ5${YKqJX-hAHqGTWZkOp&%&
z)TGP=nd4xp0?3`gd7K;5sSI!KDvL;<;<rlL;~UhHSOZqd9t`sw5!_oA(VjuCw}!;c
zkJV)07i@t1wIIF9tU@YMkx#fV2F5K%a{!Yb78xTMSKZvG!ocE$aXJOfWl9FNOqTf9
zG_FC`rcNMB{cYnR)N&$Nby+Z|Vl=4&(wF<!mPnxfXIWvOBOxPKp{?{*#z#|-ScltU
zRqWrEo3Y6^0UI~r1g*9ch}iQ+^oU0(NLk~9TZuoivQ4r!7R#$j31q3PvyH@cg!&EA
zkwBs$*@Vya`2|6+vjihkVmAcqq2n7woeG%yIoW7q!$E~>OE#t17r?bfHn>hU+o>53
zv*GVzFaQ^cVU#i%1w@<L{e}q0rL4iSLLLa38V8%r79ZOpu^3&jP&$1;GrBqnYp95^
z+pR$*=vr6-)3ZG2gOCOC+FgP()#Zz)I5L|j5ENk7(eKkQvmp=DEJdA`KKX}y9NKFs
z$0Gqp9=3C_h3?2dq<QG4AeqmSN8&uSzyvCf`Ww9ZY3ZMZll+)JV`mP=F)RbqdEC%f
zh}-#zSdDqm(1qIH$6_I?nF#Zw!%S%@tcE5GKb2-=O?IS&gQ%Vkwq9L{zIZh4{wc;(
zW0&FK8Gn{%>+9s%G>-;@tJXuFhmwR&637Wf)v@Oc{yqcy%JYVxF%V_tF9D1lN=Uo2
z<%Lk8XnDcrFr`gR2w2)uY?q<_#ULVF;AowIFQMYPBZ9vEmwaS)T)Ymq3c_XOuO{Yf
zfpMmD_7Hj5Ujf0u9%o(NUWwg-N>?p^^GPhs*v>h)JY;_(tXE?zkeCO3qDfwJbl6K~
z_O-~I*Zmij&h!L9Jyag{hLJboFL5m2beNd`#}LBV$VGfB7(=!J!7*9+dms&qW|9{B
z;k@n72qUYGGxYGCcn^{Stx-&;eK!F(Z!2{{wH(m#BmYRhFdfc9xoqFx^KnQ^>#zyD
z(cZWH9076XgE*eUP;d0Zxap!UFugASv;jI>i@nwjZl{lo3Q(shD<6m8l9|+iS}v^2
z@rmt28A>BUN&A=0)~{Dz`PBZiD3g<a`|f}skUon&VoWI3&!T}0C-^L_h&4gox1cEh
z+(yTVWz54D2~TWW1Fa4x_dh;1ZxHJN!$L0V!>3P8zVuUm8~*#Npqa9U=34nWWa-)r
zBJo5D%<QF!A{c`1Em}gB$u|njvV|?vz-97b)rEk3W@~KcU+QI}&kz)>MQ67!FxwpH
zOLLNFUfUr((>B%kWwq?+Gms;a0GMLmt@n!TRG?<W&en4gm%OvTY#=hwo61Grd$LPR
zVDyiBaFy)p&*(qlJ&SZg+0CDU?4gx1B5#H4o|eK64tjR5pe3N~Z>CXMkjL^zzg3{-
z%1Ow&Z>g0%3e*-@fopWn0=0tCE`^G{vR7Ij1vVR%vJg%FpaR+3$EF=mQw6Zs``BoG
zVO-xgeuj)eBN=+#%eU?4-UE8eegU&Z?P23AMy6f9<MUdgHMUF@^?d1~{SAJc6t>nq
zApYWABL^0^xeRfwNcok6VoBv_YEvr*+m4LPTn=mUAq8%3MTe?NR1BRchuT=3t)~w<
z>vTEH{<@$hhYS=sJVu5NRGFD9M--@yk;SCtNMEXn2v(HtsK7kr;%Gj2w6EFto!x<S
zkG9D%{tNjFPZ_&F^Y&aQ$EI->tN0CjfgERZUw!G7C(H4HN01+C-GlYgkB6isIl+)U
zqOwj-OaQ8-5S)L5oD>6t;FbcJ4A(n3D7NQ-{&I@p4vF$GBxdE*AYi=DN?3?K1Cc?!
z#i^q8`ZRmJuCfkED>*s6AQ>F05N=~lLwXc+9cIxeRO*A7q?kB|qNf2~P*o+pViYhc
z7-ztqQSYvIS~9a()l-DAuyE<)#4xLK7JL(Z{XH6$pl!KAhgeDJ=WC~dXBCjP99z)e
zpM{r>?0i5FL?J@{WuTAOEy@oNGnVv>z@;&Ky;jcjXC_sXiFU|Y{)~<wiXJo-p6xKx
z>poKkrIGN3@|3kQIIUOB$xa*`Z;1b5SEmVij5M4f1u+wZA}YP>aQX_nz0Ft|w$3n?
zP~>~dad>ai|I)p?L2(=-#0EB_V@!}qamgqtEpQVrD)pmuUlro#hUQ6y;xVKahTiZ5
zs}wK83f!eKp|Jr@lolCo1l`$s%vo|yU_fzXcsn;RkHf^F62-^}|D}60lUwx+=5ftt
zh#cj^TCBhM)JFTqDky-M^Hmtf#IP`^(DtX*vl&4l=Ox5y&0@5GxJJeXK6(!5CuOmC
z;1|jxX~C8aT48+lI9n@9p4Z9w_TQO0T(=kalvuG!WFC5hduDuqiPck9OU^vG(11Zt
z!Ab;KnYa17X`!n;KytZU6#Iv&yhxn8*k75mb-GGsD<4=cmlSlFS-uV?)QXx(8O^x)
zUgD9cG=NCITDdg#E{80AT0Jg{#rNvjN51bPYfO;hZG&X|8AHX^0|_(%mdBDTTg$F(
z)%q&@Z3ZOLF9aqb*%$%09g`DakCjS0HlR3`r>pWeXoaAenGFCZ_(o820fVE_8Q)S2
zutH7M4ucoRgCsJ~8Bk5Y027zm*0INz#|*R|fdnDn$wVIk!MPNV)e+8nd85=C!W%O+
z+T5njL7A&TIQB0EVkRKrU#Sl^m#u{JmKzDgHUwgajVO>wJ{5OgZA&VK@`AV`c+_}A
z`ATDej*EaT)EQ)v^JhHq1QXqZrO7!ltVSNU49`1+5*}APb3`;7C+^d;mrS-5O;E2=
zPNw9?1j0=qp6btB6vWZ#kQRRiXA-V)Y6c5sTD+&ahXy;ym4QkR#<D)-Owzf^V5ljt
zldJ6xZo!4bHqG*bKr5ndSwEHO*4+c=e`ph-v_dv|3CcA-(Oh~Lu#aqx$~$<oTpNfy
z>a^2Nlj#m5PC4T~u($1UT>!2#@qV5!*ZX*s^<nqOUo~=rzfHr8>6vTfN45ermcmpe
zKTgYLW0;WhSQ`5$hHsSDhB0ln{4{YDm2lzgqH#HuB$68wXcQfYgw67^AfR6GgZ@0^
z7kI4E;zE8AKW0&R^(OxsR(f^%F<j^7G$J~~f=*er;1-915_DuJb1s+`sG7>HzMK$2
z*})hnhG6zfAI%(DYWb931;SBlG$X$@fXZM!LI0EZ>6jCaKU#hpJf|u%899Dgxh>#;
zLa*)<(7MK;`_aGM;UL9RtYo&UNjP@|9y?cHrGP`tA{!wAX)R^B)2J<5(;{~nK%+*F
zD=Cw^?fE#Q?B3%`iNSGY%DsWjo&yHReFi*nhLyM4Pt9njo|PH?En5LnqGy8O%-HFX
zl`XJ8%(AbWs*!BgB(oEF;Vb~l7Us+-?Y7t-u*g-)-WJLne_PjzXI?KB0M50WrX~xK
zzrEW6C}V)4WL_+wH^Nuv+hT_omC1rUz>F4BDRE&u!<D|AvIXLW7ug~*<w%FZg}&H^
zL%~?+qDveq6?I7upk2n`E|;Z_sW#0YSQdMuk*$_E-8~w5d5~}+D`i#7iXaPAX|9?A
zJ$9wdItjnF$SPaJQD<B%tK)7%JUv?8(JzjhY>iFczznWNSzDl?7AKu4>l`SSiJKuE
z{ufzXhpbOaXSlJ2+HWv0pnOA2M!`7QXfyZd-BUK%f1=o(Z1%ezCBo!)K8%|{!6{eE
z{k}s?u71EVL8)dA_7TNjvG?TEq=03z#W9NSHIR{J2HTs?01#IaXr2@xTW#cO{7)YA
zFOm3#UEtZIgO7<7UH&DEXt3zCV{aGB?*rpwkR^nI$q(6bt~}-vdDsWyDX?mlAVMXN
zKcwwOc8>F7%b0j1mdzsQI38GxJnGNL5~!iBf#&j<0X$-O**JOJ@ZX8*J;;4{BHqTu
z#@ierZ1TtSj0OM5mDk9V_6!A=>MI(W>6*&PQ#KMBKsDneKJ8yoV}d0JJbJzS$)8az
zSdP`q!sySW(P_PCN5sJ@dDfrNTfhqyt{M4r5HHSjnAOM{*?$aTn>-g+d^M{!;f|-b
z;CV+CcR2E&%~t#u|Ao8G>!$<SmZA5BkhdxG!A1@L>5Bmm;-VOE%Us@<f~(Z){qfiM
zih&q#Hv$7M$I1{zv{s31y}aVj;3k5-)Y$aj;u!>TDs2qQdNn;`D_P!f>*Tc{7KVNx
z`s{Tl_$Fk>zY)+BY`F=-=9~UGWnGvKaJQ_Gw;X|CXE!m66_WKW`Fl(W%BEuu35oJ{
zOj*OKtT}nd4=X8prM&B3&;SW<m}+YOh-X`;gWL?G@5QsuMXFu!zIs3SA<6!Om>P<Q
zmF=D_AKL1UJNvjx<evdLs29$8NDN4227{xj!pb-w*}LV<sNeT-fLxE*F-Uoz_}FIB
z#A$-9@L%x`IvgR#T|Ujf!%F~TE|Gt`07sJ+<B}F>FdZF=d_oI575dL?=CLL4B7L4e
zmvr+uQux*v!EQ8Eo5_#ULA7XWr>gxj`A=v8Xr5q-{FlLc(E_fDWqQ6!ID|O^gQJdt
zd~KgWM9+HDP$_f^(}+wt%rW06G~*qh2OHxd(0Up;E!!2URf-O@?F-c!hx$+TS!65L
zH=yRz4uxub8DCjBQFgRfd-v}pI~AJA4^}bUu2hF{Wj)Y=)vzhT8riu}Eng^}Up8ix
z>{95qqLCT)bCq3fA^^{Tb$&5Nb_)n?XK}J{7=gK^vb%u*9VjE;EObi}>XHCIh_3lo
zVV5Or_3qJA_HeMeJHC0j@Ziw+vu9wX2@-rx_A)4SpEt|i4uQ*;&|=^>fl}Ea`vkm@
ze#s)hDr8@OMmjQA-fH=_j{p-X7`uvNdZgNAKU+gHu2cC=%;PZ_FzMsnWr$wbKk$U;
zIQHK<Ai!lfaitt+gLJl*j+TRbdJDIu!)CD@?5htYd#O6q^X9xm3cJ)Mq?K-sSz=;Z
zIxG}LWl1@-+rlgob)FoSK*otRTsqkw9w3$$j3_ISBLY)e^MQ2MCPy0Cj6$1=qkNn>
z>S9z+ypMCV|H4bojB9iKmt%sIknht=j`cNm3qr$6P=Ju=1}8c$Z~z5XuS9GF#^iQT
zm5}4}^MJ+bwTAUTp^y`Tv@)plpk+}{^mp}u<1GtNU}VclL3F)PS@7hT5*jbl(V2)~
z@x)4oa*D0L585xETIl8&?KuSp<7tjB-i=fXm%y=nx-EEYC7a<%k3u*4z`qJ*m5+M*
zK6ZEJkPNX~P{-r;>Q(4w283g_)Il!7oORN>P=&9_<<cjpX+)v)bud)<QmnK|zj#K{
z#In{yH|ihYs(;ENDFy`IvJGIf<WK`+$535<mYlH-7Q4YIXBuMq_w6TV`M^Xq>n@-e
zj+`CHVo)(VN)3t!to_H3+QD{SN+5Y?yz3C>6lh=|1p&m^45IQ1ZK^65I*S}0^CH+G
zH5Oj%K)?t^dmU_Zv!ukIf!=^wZs-yzjV&ocfr9dKR_YoW1Thi??T4iq!Yyq5J~znl
zM7_Lo4Hs6^j?~*O=lE<8*XWu9D_kPy7Diqn>!tA<YEu!8oNRQeuaLbn%48M<=n;W;
z&;%X*@Cx8-I(tV3e9)TsjFdwbzEVa71EMJ!)ucuT#wRsqpfTlSjAM8j<oNUK<y1Fn
zY`k1kk923S*D{|6GvNF&uiC>hWSo5p4IzVzu1U^MZ?cF7cZM<C3{AKo_)BdAV3P5H
zKsAcf(9CGsWx7E0`IHO&X#6(=3quEB=Cz!a@5VBRBkv+xDHjE;jLg8vcd>5**$Tc;
zTjV8i#<5MbFwjray@|v3d;#=AW0ANOS;waXir1w<EE;N_nE+SGWi}m5vQyxi{eJAZ
z#jG<{UNT0mjV(DTj}_yhr<}M@Ym}z#mx=^gsw4}T*W*4`wkX-O*)(TFYzN7p(kC|a
zvu1uvRXl^x2Sx&y^EEOdumhDmC)EkuO{&(glvPcdW1WUa&(?rd$~%__6RNApA;=Y%
zE*aRw!md>R(9>XnQ!3mlYywJ>fnC|P!CpopN?YTs>*B19LZ6p<BRV7jY5GG}8Vp*<
zQ|sW@n`GqET&HPsazy}-tzpUu4Wx~ZK)O)1=@71Xv#J<>%LTC%pr{W-#3tLDiO_4C
z1CpUoi6`6hA|%&MaeUatp?+o_JxCH$15<i{X_de_aX>dni-R@??iEY9T`$voEj-q<
z<;pl;P%G5Lag~3<_&~NNL3t>-I##H{1?@3Q<p%-gh#FMyOD`}Oj%w|CxyDa*t~w*v
z2EMBrnsE^#d}n%L7rej!3ysF)x_||NL=&r+=yZLs12!tk$qn&mg;vk_QBdkg6y%a0
z`^FlB<0a3MpY$bIAj3}_6a1Y<E5z-{PrG4<DKpDq_Px=dK>4R8R1Rmz(a!=dtCR^3
z=g$KWR1bq?=@;7&fYxNdaa@V}<ffQ_9w|&f4d`Y=KWd0H$t^}jO*IqcR)?QblvP~j
znWj{#c6RxttrliFD!9KgGDSgfRL=vM{yNAOk=*R9^&1~eg&jQ|3|L-+^J<sEZ;ePs
zHnX$GZ4RZW=dheZYrH)MENg0nk|B4*-&H6v3~qm?e`B^VwLZ#Su^|_+dh*@zhTcdv
z?L8sOKy_zX<`ppDazhyZ$lMq+{M>68>eUO8o%aPpMYmv7rKJwg>geW&yf*U!{|x^{
zqvY@la<Mo(PU1`(66HI}WtIs=SHK9k&Bp1a2oRP+b!dw{&%wQ|?wC3LE_HG`3JhrK
z=W~6bRxD-?JTL9i?r&Sc8ug1{`&kWdUW|;=RMug4etZw*dI>2L*ZG223D81;o#g^8
zOJU(QFwcV&x)wVNT~!wOp7SQ~`v9Y{&MbssaL2_#wX6a(QI^=BJZ!w{m&S}>le~GB
z0T$b|Wd<>v46M<Ce`UE*4`Qxhj#!bvXW%FSYL}Hk30rOlf>s%ogK%Ya{4gm45pRuu
zYCFj`l^wFS+cUNbql~=HpOHXeTmWy~EFEdJD`CDfLl2#TIA#=AWxcP5o~tDG&}l0h
z{7bbVV2u|4sIfSV8<CCvJ3VDI!_J04tocd2=|LNt0vcF_sI?;9Nj4iskh}ni_;<m9
zN1@K*{XUvTu@R$y(7a`O6Uzg!D{R-4lP!TDWc5@_r%k1iBM^uX1KA{7<0O{XYQyaZ
zgP?M3SX)!oB3*{>;V1+A`%nVmPYv6HDN}4@NLw85?T6B&4O4RRu&<ArF+Iz)`~AV*
z*fM?V^v)aQk(d&qdXxIp9!=jY)*adeN#Zg4X1rDYdE6l(WzinVx&++{W8eueBv05k
z5FTix|D%se4oE8~h2|&y8GXmJn$gn#l;hZna2rRUb<m3QZjh&AeRa^P<WEk$<q$FC
znYd<AtwJjxKO5%@#$yWY+%Gfa{yE{23TT|HyqsmYd@eAd>x4XsEBX1L4pc^wzu4|v
zdh#*yLjGMXPt<{kxEF1Ei(x#_B<Gjx+wmYQ6dW`*hKk+w@^V5B6P2{Gmb~K6z?EVD
zLp5ZO```STr3D(0uiBsmV~3WElh=#_A;lwtGApl#5Jwf1wlvI~l{XwM?y~oin{?d2
znHIrXaV)PlN8a+cd8JZkiE7D6+%<fcr98T~{Z;Et1x^M!HNNvsH@VT%@UY)`d^66J
z>H{if$h&DmW<7>y5bw%A{Fg|S(I&Yt{w|aE0x^2R+mIu+S>E?&T2Tv&nTOr`AdQ`R
zM)|q<(CLn6gev5E%x&D4*4%mW&$zs>T(~eA!SX)x34l5HZNYpn>Bni9Zc!p+H5iO0
z?oZ;l<(hMoP;e?I|FTgSp8$h5{y#X!r$*$ZD9DZd`nS(xc}?>$+6wtBErdww&MZGS
zdO$5TD70TVIEHJuv2cShpxu(u7t4Q)fG0LhmM{HPV~-|ok)blwYBn;!zKWq#9W&FY
z#?fGl*v9?qfD#G_-Lz=BD^ldP%)*B9@)r3<QI}c#(3L^9D>B1oG+@jJO>SSL)=U6u
zZO?es?%>ZjE)BQrJGf(dZ=>U2p6pcQb{bH81$!jyY;)GOAc#hG33$fVFzaR4BDZc9
z)Ymh9Z8ryL9&#Tn{+hbrGDI=#UZhsh8fec5Ao!+@$*wDP@-2tL%8f8$eXZ<K<Q5zl
zr|VHW(irQWMP|iAj<X4Gs*18KU-#_Py=*E>%v$Gr$9kZCDNozSrqgSLNousp?rZBf
z2hpBzbyUi?10=*@FpqsRGCB4uQfqO)J_F@DMQ)0%M*R@P#I(r%@dPEGYEb5<MGmkb
z%h2Hx3Ai~qFb!!AE113Hpdz&}cecX)d~lJOmF<cqJ+i_dqa@54a7cWQ%1k_1+`mKp
z8@o<%`1Nh5rN9RZ%^c?2VagbVKVH#n%yf9b3;P)A^dJM_2*+A0uKgu)q>qem)SPQ6
z82nc~rgD@+qiV7pb~!pQU5C^qV)d8++3K`X|Bz!1i)F(K<TxXO@|s4N0*?20Er5=}
zS1?N&dYY$AYmpO!@QP~?=O-tnp&8Ch`(%flY(p1~D3Vi()YNz4S#oMZ0mIHM3|B+z
z<g`R4kOAcMBDXbWvZyu+kJkk0;dAh~Q>>&o)H8v}n)d7rI$wIlAwV-k-b=lmO3e~a
zJ(PwTTD>T2%y)L##{naYq~DcMz~)rz8=xS?r~<TvbJZ`l3%L}_x7-Rq`p1(d%&PgM
z1ByHra!Llpc`NHDG*#zhpwB{~l6Iw~Drms1ENA#iEE(Bu!ZQ<0n_&VWr=Y?}q3Eo@
zH?$O}?Vyo%Ios9^Umq*}W@V6n9X5C*8bf4d-^{_=U`@xBApx$IgG47U^i`9B01AMD
z{9zAhmqJG;KVvSIT?=`+$p9=0;zPk#1o7vjIOYT?F-8f!>!l=+R!3E@2FA<M;4^xc
zAY=tGKP@uUzoiwD<)}LzSRliK^-h{pC&SYpwHY}l0o{&U%)mJ}ogeEUy;;5LdEfd7
zqg9lGkda2&MGY`)K(`#_Se)x37ejS*(K6cS(fbSZI1fJhF$PEtOiPt=UfO_JdR+o+
zC8^Ca)=pA!4UE8LM)zzJgeb~!u??ED4H<_OA<p-Ki~?x?jo94_f;Qa_jcp%qW0z!6
z^!`HoZ^%l>`?dcbJ_`Qzi)^_CwKa0FJ>GH)k|8g#r9-x}IKTuyWBsWRGkAcP*QLI6
z*u`imgCMX{E;I5Qk4j8hhw1yVu3-&WWhVVqGPXp@ZBj%XW~Cy{Vi&&^IAbH^?6$vv
zaF{)`38$?x5yV{8f7Za9Q577i0BLJ2G9lrNb;wOg1=T@hs!sG_4ZOiQq$cqQ>knui
zjO%pU^5^DqLtDSTePp5!&x?+-2eo-d0nU@!fTL_mC4#-`{Kb5qCWb3)1W&Jz7t5je
z=cFNzQ%p;~i*(+=fI2zsTG6G*6^Wqp7ecqd5NQl12zNyTtU@^hy-)vv(&Q6aRF<l(
z(rk1K6%O2Oa-3b(Mufm7Q=F=ynpeuy1m{vcM63(e5?>(7pppvvwDi(UFbP;rklu2o
zjndhA$z>PIRSAELptEWlW@WuPm;#I{$j1W%|3MIWK^-B=q6x+iZF~lZOorzJHw3x~
z8;KRdb_Hqd+B8Ab2v>d!V5bKhS6+RlT=!qLXPp{)7)*VA`X#SX8LPY@aErK2_Otv^
z+6yvh4Q$}`lw{H$J2Jdz5>C*1`AMu2oB|GDh%7%1a$quRqul6&T7hFTk5K_Z{Q~g6
zIr1|HkGzQSn@n6b$>isO8&=Rjj6^dw?H6_d4{MN{?7v23<z@%VGyHIf5$hMA;s$4Q
z9f|N(f5z|-$j-_De(BGs&XW9B!DI6)e@2Ly88aW>{@M;Y_(@KF<G)*es%`dfyX_Lq
z8I0NB=~yba6?Lg$UUf$pH?Grzp6VVd4`zG^Z^PStWQr-A8UM;1X*QG}HvGt)J{#?y
zQ1FrW6VYvV`P=lu(rd|B$h%`yU<PI`jaRuRMul62ED@(i?hV#95w>u-FO6iZ2~wc6
zcH1kYEJE;#$gLShN?UG%Cdy3InL%0%UWH&jD_}s)jp<m~GPpQq$1^~reSF~BX$#^h
zs%D($92<s8GGg~DWp04Dr4<5Fdm4a3EuCR2WL~^6w4ks+=KC9(8innJLL>{)R;>Z}
zrbuzC7s|po{Rj#H>7r3^e9q__U~>VkjK$q>K*28`54gnNZimT=<_yTNvNYC{L&$uE
zEK36}!W}?1-XY74qaA<Qm9oNr)*M85df`2hmHx~`OaKaDcU78cDV!ZP)9Un|N(0#W
zc3ESyH>1Z2i`=d?BJOOxV2rHu*TIWbc*mAVM^c0G9yt@b7<R33W+2k_{v`t(kbXkT
zuxv;GnSjT7s2gMHjSXyiwJ9dwvPd>N4h?%`Mhr@)-Trq$H@QqL>_~FIv6o^Lb7^c+
zSJeYCEVGFlWlQ2&`LyHikpi6wEvsSuv8cVRzRqs8XhkBep6tYDpgb5745+UPbLs3|
zF`O1uGH2fJ)8d)HMpVs}hl16^$YxE{<zb`DF)b+6R4;$9T}P6(JmT0;9sq>GLv36^
z9*x-;v;(kM`~R^-zM8y?+312~EdAqgAM_m9U!F+PtyY;)wU2xo$_3aTL!Mb?_K!9J
zV$5?XF8#?k1>+IRoWXtcltTfdJMTbnky-L|%vFGZrauLk>ilkF%!WK;>tpxRXZ>fp
z))||*N&Xyb29G4#*TXwunJ>>dVvtnwe2j}j#(aGi?)pnyCm4pgoGjI4d1Lnb3qc;^
z;BIe`7ZbLuOzL{Q4G?iEFU1OqP&o6i!8L}Vv#7jmq*_`!e5||@1UUumWSRAA*Uq_P
z<W>6yw)$C6d|wMB!4ZRaM4YNs@_G;fJR)e7kOjHE;jfY5Flw94THZ{!KnOkdl69-`
z`_}8^Eq@!nQ$2pmFwx)rnYGE%-U;%y0leip9O8FkDOm_G@@{&2G43)%DPh!q7>Z~e
zWPtN~h6p$+5$F4UT<1Upa&n+Xd=LoN%{6_K{5#T|KXf!;55e#L8S}xC0*ZyDWSxBE
zZ_xO-Og{G2Mif4sF0jpjt9=ryg^L0Xf6TX1{*~Tl-~b~o%$K!LKJ~W^B9lb_?a%Z8
z1BYZopT#q1!thUHduGVzu`#u*0|uL6i+te|k$N>~6Qn*D((nrTPrS?;VQAR|4gSlR
z8-{us?|FHA6?p1g1lh(K#(_CPSEYQqLcTV-u56knQfvk`{Scq0lqvZ}u^DR^GP_!-
zh$a~>n#SfuvR$zoh|p80R*TkdA5TydpQRzPvO}>O+6qvDjXStwv03fNB^U&}5;U-r
zKV#uf=KapVZLxE)n<`3?rGV{6yyq^(ZYCUw`Es&r{D}fj2*uwm4MLX1(i?5EyRU$2
za}rUO$lC*JkZ-1M@HWphLM^`)tL|pN>>fV1hkpZK4t7Pw4%xF<O<W8tlf425WYLP1
zz2ix9{S*Yl04w_#=moV|EUa4gEp~%v5qjHyJHfOVRuIdt+Am;2JOQJUDgM)_YU@Ci
zzGAnE<SMFJO^aP%W&dI|<rRz@BL~=g+P)9dA6TqrN(_VuLOI9>?{-n)Y_ePx>)foC
zgNxNldG<E{lS6!rS#WnOlS6$BvKlt9qP#~nmeMDy+#F`GoO04ha=3j+%OspJM|7h;
z3w7d$FeAy4vAwmlR?1P177f|OCbARc(f+YYy*evvx-(h(m|`{AQdGk!$;q+)rpYa|
zLoq3sT<y4+rmBXz{qa7H8SCLo$DutTc3h)w(5eKVo#+G7u*C>hR;Vzs`J@EsQaVTJ
zftQnwKDJDU@P0~ALq((+o|@()8@75crv;V@>L=97=?MZV*XlTz9<k?8_u*%?hEzR+
zl17${lwLM#XDg~U_l|SeT#pK+kb(R7j#2Dm83x?hn3_TMzzg_aU!T$(m&E!C>1SB&
zY(@ORMAVn+Uu>4@?haRe4Y?z)TFZd6JaeqkLp3nIgsqSPw^|uP`4f4fcRH3c9LXk&
zpg|32%c4BbOb9_#8*eV?CTIDID4EbUfol5rmkxIF<4tmQ;AS!m2!mq(5Ot;LSKt|1
zmske-FLw1AL+4;KWk~ECnwDCZ<$`Shq)(o;7g)l+FmOLECq-$*)Vv;Vu`f%pgG@Du
z$s;q>ySxleH4{Hd4B=X_kzR^YU$)9ic!yS_p=oqWXUu!GhiwBLHU^ijI!k1@k7I#F
zyr4Sd9A_5=1#+$dUpjj1NEwla*T}tY9ehUmFDhc;E<+~((h>X`qkLJ=<8TCcBH1J^
zUsQ6m?><rGDWv$FS}09EENlT@Bx7Q+un|%VN4Y6EFSY>Uc(cmDV}mB4mp3D1v@Fgu
zN{iJr$hg2T_L+m2MK$k4IzM5;s~TgH3u0mL^RwU(l`3=^fU9A=k7fZs#3)GB-4}E?
zT<E{^8#)o_*)(o#yy3gWYO%lU47tcqjLv(Fa&at_cM?#5Yg#U`-_dfZUcQ$`&07rZ
zNSM{?Qd=?`)UiY8Wog^liWx!!WS@4NiDmNrSS*+l33V07Cu1KsL#08uqLin#Gjh%R
zn`92FrNTe9EwTz4S>H%vg4~<U<hoV&c8*x-U*at&uH49#Qswj5Gs?`sG*kX%LTqSZ
zV+||FRwtm<O$3Exxzxlavqoi;TprA_wh{#}C;A8U{vk;kQr}{!brc4mTD}<)p!K7j
zEp>r2STAVN;>Odbv03WvSd6Hwlm;KfYKj9tSPig26AH9}l`_fq3@sJqiU6qyL3Yw;
zf8knXS;U+HNHZJS=~)&?Q*d*{e4tjK3B%3t;sh43tyg>8<k%o2DKP04E^J5hlz1E0
z0sJM}V0Eg2eCq$_f6@}8L9k%wC6+p-D}9Mf^IvGJQ2T+B10~}tjgKN1WonCD6$diY
zIH4IHk`}o-_HR6cxM62ZQ}7&_daY8X{Lp_1_Ra!eMEU}`#_`2W!R2yoj6nJTc(e_Z
zirQp)011Mo@rRL5cG9jhAdyo7rQ&*j4ZE+*9z7y30B#AnA^nnP&1CsA<wuE%z%q5x
zP4eSd8WI#~wEanJ8mf#UGY$*~>;Gv$3usuhfGx;)GUmvQMoRSvt^S#h&Ny%|d;Tjw
z_YbW^AGJ_00}$CY@(Y`{v66HFq3I_3dknk%GKc48|I#G<?l<TyJ`b4`GbmWJ?AA0l
zlf3DJT`9lxXB2c<%Wfvf`B$+z^-i<mJ&aJl_A@XWtQ*058l`OwwXAOLZ*1q#w3I^I
zZ+%!G9*#NwJ5O#)AR(L;Bfz{Qw|84QRc{uhP-2uj(g-YAPo-_6+!?YAT)GWVndB}z
zc!eW|7s=hR5*XlUAD98T<(}BLDTtha{CjV(k7D%q%*cHK5m<$nnUGe8i1f;eO^ie5
z?XgH^_}jE!z*vcEZ=KAH0g;M<WzUNFD%g!0L5Ges24FEl-jeHCIbHpt(v~K(@J8N&
zc0}gHQjradg6cV$>#v$z0wTseL!ox5v?olF%hI$vQ|9^G<gd^bK#NRqpp9(4gJii-
z+!>xdQ^^p4DhopWf=$Tdffo9e1?55(#R^BkI0+Bq;xsvoh?rstuF)k%?l^(6)PJ@Z
z8^%^Gk!Ai2mFOTIXd!u7?(mXNGLnl)`G5_FV?~UhCTp6t&~GGwKr>jlG-Q=2DiBW_
zE45pY9+uUC!d%T%4R>A>thu-bd$ZODrYOYTsYuh2b^eTJ!xD0rN=JYTWN#po!L71B
zo~bIx1JEHG0<n6Tlb2fP=uAVeY&24-se(ETE9j=+pIHRY<7{VTvk!_B2K|mI@iO_H
zp`PuP<bLOkG=Qy>2mEcGK}MCY!`^L)6{8jkJUcn*jFXIGIUza#Z#6)mF;=o>%7cLw
zAg^A!Vpu9bATXv${@zz+4l|F#z(&woN@e;X|J^)b#2wY2byx&l+#T|8Kv4nL8Pa()
z$m|b}j(X%0qrTcBP5`c_sS3`(!P)YtLuArDL+<CvV}=&^zK)m217aY8MpHy_$P>j~
zdum{NeI-PR@`lFhh6V&={W0jI2By?na5uF)JsAk7Z-7evRQ%O*AZoEZ9lL}!3~YW4
zD`U_egisn`o{7JU!IMFY&)WARM;D(zqD21esG*9@I+C=4&bG@)&*^gxys6-}i_gd6
z#umz7{C61rRf<KC1$n{8R5u_NjB`!a0y~B|T3!q)fMl1+HcXY5Y(N-|%H^+d(CFy_
zV_=A#37#*9megxtZ+RugfXSXobMiM^7R-#Z;of-FNAabA)>(0rby8l7-B8~X!*}P(
z>ozh9)x$siMr^bj@|b$3!^oTd3&TVh0|=h$cluj401DW{dGUASE@LxO<ZYWw<tapr
zYo^#cjvZX1Fo2nsZj!u}OL#ZFsY6;Y#3c};{38Lq0#>)VU}B5qz1UZDcH&k0e*B%w
zWV3MMe&FBmo?vA+$cFM^+8HXRwx?Jdv@ddHm?ixilOVzD5qR-Y(A#7Ln#jiq2Q(Bz
z&IESsKK#TtfPtQ94`N8@jjj6B2c=?3iJ}8mmw(&OOiB98&RHQW1rQJAbNeMzQ6pcZ
zRnuq8YiE)C$LH@>g9U4w30K0Gfma+bR`68v!6vI@2M6O<@jLDkSjj=Jj5uBj?V7UO
zX8GEQumaufrNl$~+<==(i^(@i)O0lfe)a82x(;BSGvf$EAT1P(HnLdP?Muv_2yu|8
zQk8gz60>i$g9o+AjwNn_gdu1W-G4i!A+^pM)F9Q>G<1nA_;&W+sfO}~$DZM2QE1e?
zlwD%Lk<G{yz^T}^L=BTDQKY<JH(wX`93drJ^QN@j9auA}m|pcw2bOWeus-@5swDw`
zaB9jPu|%*##_3ZC-_yS8Y~6B4=PcQ)L=A{=LE(bf+g|T%rP45koUcjt@gdOp3f3O;
zy6&4tgKiJ?d$p9{UtYQ2F6r7InE(H&4*LZb;2NCSx#|S@PJTNfOqj@_MikjU76m>-
z>2ekdKO7KH0>Dage-P<`C2B@J?SxA%mV^A8xzI_-(vYYv2gfR)Ueh$n1|^52XV#_`
z<{UXRK|+Pp5@O8VdRR~fM7ddVxPMDntw_ZJIl@qh9PHtxa-_dt1gDu7a+IUZ^?q$3
z9f1OL8Xnzkayk>3<YHRQF==vKnsRJOm%8F9fsxPgx->4L+XMcyNlASJB%R}oz#8ZX
z#tt7Js4qtZJHx<ENOQs>ggF;eG6j-4<itP^k!6QSPVyCyJ}H)1@j80r6jgF^eB+lt
z^Da-x!wyz{s*gY^6rp~&&n(}987puC%V`de#vpPAB?xBKpPt6%O{I919=7dR%w*~5
zzoQ!;wnYDxUTJl#$>34=8bI4S!J%z-fNQp=%!?d-{7arR&Sru)JQ&iq#C;c8tu+Ir
z_VbzP%7RD(KDb8u+cUKDZARo6STn3O1N^-<=%K7gLTP%S&qE%~!2AW2%JtGaBSuw;
zR`DTc#s~<QCdL@+bCy4&0!B;nTHFa|8y6e~m*Ajn7J?I`jDV>vgJUpFeA6wJA^tYD
zN39v=uO-_kaD>qU&WKViM5{(ImFlE0*b2m+0a8@rKA5qpd9pxmahh(engv#YI!o*s
z<!<zPMOY~<l?Hpi9=&C#L!)8|Hh~T%6H$hxMY9DagL^t;xY0Jkb#v%bgIoEWG$UI+
zS<Sn1eMn+o!yZ(~X3pyff18M4E(iO*jZ7o3YXMFgia3o*&)6WBP0?1!=ol5Kq@meD
zuIY;yW9P54wH(DOYvsHWcem8k<3PfU&p@fMhB~;e%h4LHEKpZD6|xg_x2%kdLC%H!
zy<X0bLwo_`Pb7TG1wK73+{~()fdgyCjPdEfb`K6^&n1(rE({FA48<s09%b``eAj?t
zp={^@7X<`WHB%u3T<kai#WQ3FSHn!XB=Ca2nBTw#Ti=Uk<S5)g1WH|6(zT0<*6=1H
zoLnwTghf84sj2`TJ$QTx-**tXFUV9wD)cn~l(la&cKYEiUMuA>s`b&73ZEI?M%8ms
zGeojJ(j45*Flj*l4*9S$!HN2a7@gL{Dx<^R1A5AY01<bInPlibqX<Bt5W(o5lbz8g
z!I<6}e}-6K++6y$T%Le}O%-SZyO4=#?0mxujDSUIG_=$P&|pp!CJ?u-jB2IS`In~9
zn?OYq7d8af{Y<Hk-?6uV8rMWb%-h}CV9+4YbrQ0!px`yv$)q^Ns9#VgSEO}mX+H{+
z(irTosRaq8v}NUDRiHaq;2DA{O(kAW#4*CHCFh{+m=!ry-H>LxF~^S`KCVP2+aJSN
z-$bSuy`g|YBXy*yL4HLe3d+i)#SQ>cY&F)fNu~uEt5ivZFjyQ4v|Jg$fx~N{Csd59
z;u(&Mb~wfsUTx2)3Ba(9Kx1&C9|TQZkkKyHKa6+iRs_a!s5*0vA?)HyPWs*nN6WSL
zg~#~hWO{s2r50)Q>dMM>vEi=fwy8-mESw#3y;1eJN`zJ3VE>Isr!x6bw{t-0z_L8)
zvixztIn>Gq{KVH6Cc;_JMt<tg%*3a9f!r9Rr8j(d8)m~t`Ljd@Od|ZbzemZE?ika!
z$TNRod)5qDwq9-uFwjY7a%GF$91DT6bBhnrEgzk%6HfD6ZELH%FWoO=kVe+(Ym{Hv
zAQ-<=e(h^eNr4;&zJ=d(gN^2QmPIxX*>8i@;{HVzG_v5pYH#x)cy7tiH{%|>J<bWp
zhOP+K-R<t0xx=SmvMr+|tkwpiTjb6(8ifS1t`%}ukZKk-qMY3AgOe%KK1oY;cggOa
z7+eF0)K2Jp@7~xkxVOkp;eV3*VzoF#XgUFx7u^uhLZmg0C5i*Vn*-OzjF=gWgBEUN
z3WMd$hk;O2uViKbT9TQ79hw!0N3>KW>MYMr$Ywza%e8LvXLiph-;p`SChHOPKi8-R
z7tngR0orZnTG*y#p0fkq0G=TUC`{r^NSFEcJtR|F{}<R~EID;K&ZEXdEcEryff$3p
zQk8;KERaRMT2QE!lqAlSjjE+n7RQ}w6%ra{Ngyl>&Ife|NzL&J0wDlEPL?I8Ni0?x
z$q0v-*CH)AE6Wp1Dl)03ffC#tS<ww$Bo9C;E7RLFd5|9HilH}lm9GGjEm%A$+3c!h
z`mA}g+M#K~-1K6sP~OELt8PzZO^NymvlC>k&#rc6i~&ts*7+Q5z#k3xsK>?I8w9Vi
zv2?__ub}T~eNf2cn#r)RZSVn2rGeKKZq1FcM35ItxUyF4CfoQ)EIqI}CZzKXQ{}~d
zTh#B;n2ep|$jjyaphVPl$sk(ifmjGk0yyH3Q95Kx90VAIc{tL)(wPu!MhR}6DJNU~
z7i=NS++=1{b=d6vL0dM;)z{<5cE$X#)6^j2<o7Y+1FJC?W0fOd<DvYB>e5)EYru^1
zu;HM&enL+E5Z@DHFbc2$j1lD{A?i^KG%v}c#!(xa=?!?yAVI?tT4U?w@xV4D3~W0E
z7<nRLklAZoX}kO}ForNtrn_&1sQsirL+T;kQwOs=W#gA+kmtu16i?gau!)16|H<ji
zBr+QMn9xP54Nox>pV)TrnE;8I?TXYkc{UFcv??QtKL;%pXp`mV60sW}rcME_g8Vri
zt#)}nFoSNdRaLldbMlvNqf+<mz^(Q|Y*k|=O4-%Qi?)D<iWUf#FZl=XU_)@61KL_E
zfAzg^=A#y0s7%1m%W0R;Rs+K`FV$q|SK^ng(^W=B+Yu}ZfmHIh_yt!H98xg%fGOd>
zm&&WgOR_K;F-5IKUQ58xcSW;7o4g(j2eEt1tdck4%AzZ`)@?#>`j-^gb=^SETSi{!
zpWIXaZVY%#ajCpr(lu2@>(18yjsGX_IH;z)gUR5IeAl0m)tOUsw*15KW?{<U3RKUi
z=E{3c7kan&8SPflO5P7NqL>oTBHV8Bfjy%V2|mS8$eer_&qicuHJ-p+wtuE?m+<_<
zv(P3V`F!0?GTI~uV<R3-KK3i^x(C|pD&-UVtDwGxg>L_4B#p)u)s#^`jooDE7)sE{
zzkQW7A@j1%S7`ps$aXkOqQZV9pQktT6`k9lSk9F%f+C<G=PE0ZK!SVeKXH~Iq++dK
z`qw-X^qX{0b>e@^SH5DqDDB$++DNCN1$VxbMnolxV?(uqAp8w~1|f=7ZQ&ePDBG2q
zC6$_?b)Vx=N6#x@5hqu+FLgU9)<aXK?BIj21_mrtsyOTz&yerT`wc0|vQw#=?W&Q)
zDLa>{>9wF%S=}zBT{|l&XDiFWlXtZjaUWF5Zt)Ye1N^kReTNZOfEd168ZoJQi{=$#
z`Br|hYOiGi@E*P}Mq#YMq1UJ!EGa5v?HR~WJGBWxd!-qY&yz^UXnUu(wa5Zk$OhTR
zpo6vb22^s|H`WUq_P5hGWP!9jF+zcXh5I>F*iaZefITQo<~lQm^qu_gNt$9tIhhZ#
z17!bL73z^1|5aHfEOv50scPV;ibeB5tsEF|W1q-|nHGMKW3?5M<}x|h-)P4L&&v;1
zDmldAx5G}EmhMV+#X8i+hs~q0p?+GW9OiGjAzp71{`U$w+{W!}EyhI)36BxXN96IM
z$5cZNR6;s3puk1J6|o$`Q2_<^j)`s7>x^FQqa6#hywgSjO8|anV2@Takz-1`tk**2
zR4ZJe66Uc!2V<roDT6NB<v4%FiXl7_>;QCpdS;z}QEx;}D0QP(b3OV>pzcOiPPErp
zdm)oUl@25V%SrwzT|Tq|%z(k}<We^(<|=EZHM0QeDM2_RARNf4{tZtuV=i$4%W1LA
zDwpuwqfDd{FI8U{OcRJR!7J$ioF_eeTj`*fB|ZHw#yZZFUPhd7(l<$OM=P!8thTuT
zH=oK)(kBrgr6|3~Fe{Ib`D**7-x0MAglk?m?pa1J&>+??zrFOtU~@>Ktiako$OfTV
z#N7aYvkhD8?XI{X;5P#kbQDLrhghHCYmUsLwOgmlnSm!5evw|Kx47T$Op}qb{CC}K
zZ0~%$#b$RFmNs9`E>)LOzn(p0P-&MZ!D*j!&^)XCXq8tOM42d{U7ENFGC07;TOrAl
z!z`2`u`w_W!A}ppV6zk$@;h6HmPlbjA+79Su@Lt7B`v<te55F$ko<wa(At4C8Y%W)
ztQMHoG9$5ohD#C>^1+gMXwe8dphctyQsR`1?8~9)%Oxs*AXOH#C9Am%3(`~`Nzz;>
z!~K~F{Jf0N8#=#g#Tq&0xrW*kPdrIRH~_J842Y$BXtj(?i-NHOw8%Xt#ch^R{)<u`
zeo2l+8J#9uNrr{Hew~a7K!H#kZM}fbGnzxmfHwT!JQ*A3usT!DM)esfi!%Z?yiUd?
zG@%s*X-RG2boZX`qZ0vII1MNCf`A<B&~Oy2$+mEOjDu0hW}q(G%H+b*uD$i*>sBUT
zqYS=VY9=*3h5#Q&7<)_oFRZW(_`1jzR91<`TQqoF9Bc%<r(7;^Aek9WZiwn}^1a}L
zkXzu+W>g^p#V)m-LGe_%EMUb3lf<q20XrUlKYn4V6o9evwy4)2nK)CBhMCg1T5Wza
zsd->8HcPqxZvAjAh#B0FT!Q+MR0QV0ANV~g%wowXnIqXWCvDY404%O5U{r=&a6YQP
zRZ5kuU|d6kOfaM{eF>Hk6snBFHuxy1DOCq3jCnABULM;&hRJ>~1<FJRlwdOjhD;Va
zpeh2Z)z`+P<268}&Zz~?U|b0Gi6(e>)hR3u{>-Ymm>xFCKrJdOl`G<BD0d`M#s)Xq
z&;9%Mmt3j)51>~xOsKD!M#;0uQDR494p72YZwtZFn;rU~85SskI~4MbnSdwTY@Mx*
z@MS^i%K`^e90FAR1V^Lydag|MXXFd?nlOU0#h)=Ig<>Z|AEz0lxm>MW=|CBK)Sz^*
zx`i;?Uu8&U^GVdSn+8sQbzGdxFVF(l@&ntVhGx*>57X=xWw#XE^_uj|3<n)Zhq^XR
znQTf}E&3hFbpM61hxD%{bF^IN9Gz;tS6at6$s(de4Mk8ditE#gH-Hhfq21;(xgkVf
zRA{Is3H>NG1L3`Prs-E%j&1yL0>a3LrSn)eZj=1PXs`ylKu&&Ye{{Bvs+=S@`jikx
zAo8%wEIP{gIM@apVg@!cXMvHoKR4>rjJOK<ML=Rkac~+sGEt^ZZb}QW*j2O!m7CM3
zo551(2<FwsOXU{Z)B$}5$gSyELlVTXjM738Rr||yS|}Bn<-lU`OjY|;+BGKl&{nrn
zejQ8HGGt8ax>kPUz*reFvzn?JD8G%j!C2^Z#yoPHKO;iO4dMNk+haKt7+i*?Wo^3L
z5%?X6MwWNRALAi%-R0vTVjp9ZB4*0nMu%I#=;a<mL}|@bxi=62ZbhDG5=tt%BKHL%
z!1=+sa0P?&cMJD#4OvIqd*@_^zu!I1li1gwO9TMVOv|zM8mcmY-!7L~4m1xcq8+fm
zn`L%-MsE;Znd_x3o~b{I?&f7OCsqO7JntSS$A-EAUOP94yNDG_rZ!1?+$K51hR8g_
zM@2=2%=dfrc1Vl2Lt$7DjFdgATNr4t(C4DK!U&xCxR=$_2r@z!Ls=9U0LSMoF+(jv
zseqd-4h-N+hSM-`$O?Qc@s%+nm;F0X<w=%0A8y4_UoOl1+c`LgY|X&Jkh*8GJU9>B
zs1*#!$;ygAl)BQWJ<OAp>DgvgfLY&Ul@H3>fbkKQIs|c6R(D$}EjC)>j8O-E#jlCY
zFKDVp307I_qZ2jeWkwZCBTO{GI_)Uw@E2%NAvb|QDC+}X@H)Wu#Djo;8RSRVXpU^i
zgQr3yy?>@YZuDR5x-jr*b(<}lO1oqS`ClmK`@dnW`oB5<4+H;)f&U2uvN?p>DNqH_
zC@as;!09Xk(frR86XwbN`QNu*2lk{E8(4WLZZLiF0cXZ&t_oRli`^P$3^__VgRep0
zvZ6!nvNf0^A`;-al?US;Xm?DvT!10Ek}9-e>vW0{^sA^EZgQsax!*fj5Y+INsnOj_
z{XgV<kgiqwl~Ss@JRJ9)M$MtK<sT9d)Vfs8kVgVZ6E&y&(Zp5NG@veZ^%cou{xR$y
z_A!i8kH>%-drD(dhdhx+u+Z4WK(_fN{us9@{Q|H9H_DT75u@`-PkG8$N(&1O1{7}r
zsM=TPh<w^W9SSEpJ2w6)CfBMH%pF}K&-gP1nYS4P)@S`0qZcEf=+A)~%kHR^=VHGq
znh~AdAkX_pCLA&dd#?N?-f3(=m*=d!;QP1j4FuB$?&f?5-irYy^g>EDT**rbBx=ks
zRYF(%tB+<bQ<JhxF}50`Y6g*)9Y<VX2>KdWtoP+CdBw3$Q!uskP42o{Ab$(!QAVvE
zX<Dzw-(xE?6W~{PE#Op_EhVRA@_Ks4_VwJ6mGXvt2Y-R34ZIn9SXP~>mbcOu3so;b
z!%O*lJaZEe8@az7y6bl<<sDy>8HZ_kpb6A!D6;g~yZ${6tD4+l$CH2fGYT7Y0fScL
zJ%46L+dNqBr@5KQf-?lE>Vu#|hHgP}sgw`>Ei*My5ATqF=Fe3(7dCBwADK{t^-ijk
zkGlavA+-ZS%_lJ_j--WHEtY>7zbHpC(5E4-*VDby?On!kK_8IMd}@wHA#tt-B;h#8
z=k_)##K4CF+GRxk7mhwn-BI@)5yG-V6sMNUe_|nEPc%Q!;bE-hOaH|PSWT<=odqzz
z3J`k^7${#mh_^s#z6GcGR=D|PsJc8LCBxS;H2Gg}ZlRul!nYgRwG-?=HI<A(knR1A
zHa&vc!$o!&YA%jB*g?iJ^Iw|Nm&%U*i#2UG#Y+PPWT&BO^2hSgAiAYab~Y#)P={1@
zNl=(+57i81SNpxQb#$TZHq>nn2>%C1Mr0)p!S41#QKoK^eAE6LkrlP_EeC<b$YcWE
z1%|Ud(lpHIGc(wp0oEiGF@?=$uk<y;^g*JyPweZx{p-2#Fi<hw3~_azKm;SbT*m_o
zg#NhiP<KdGf?rMn9=|;_asjxBwA)$b0!HS|gkxquTS85JeT#g@{<|DWsItHPH!GLQ
z0Yl9IXR<c6Dvd^fs%V^tPuk=_2g>baAY>}SOn^9OsN47v;DY*V;DZPIqG(Q46D~;j
zkOV0udJ-}Wbx3DpHW`zt4;|`mi%dlY5;BlGT8r&CY^YkXI$O(7&*X3;A(Y{xEA)uK
z1x_9EgInZCUxyKIx9ob%qk)moA(zQf4hL*^nA`7FnnCQom7^2v3`8^jaEw2*I8ufT
z%#~x)dyHQ+L&|aKJ;rE&2jPGo@6ULc;9rNeVz!(RtLg(k_lZN@yP|!2S@Kv;8ro$}
z!ukx|s3#9qLwy7M7pKImAVW({St_R{;-dXUb3i~y=$<O4*)C`5vCF5YFB#vdoK<?*
z?^j=a$|ce>5giO;7)5(`DtGMVc%^O|Dg6Jf*^u<Mu`a~_q>lkS2GJIM6L2i5p>h=l
zXGP$Cfh-NNv4R!-ZHIwQviSi)SX9%szTChd62$y7N2#V9m3QTg7zr$cso$^x%bETR
z#D#h3be=C~`7=)^hS_zaoE_LN&mc-h>;De2*E(C-B{#>QshkW>)bDk@lh!%J>B+bf
zZy#8hwrWHLY!U5J;J;gCMLou#c)0ku2e3sH4(+mxJY*T7GkR1&DT=k9Q(0fn)JQ2F
zsu57r%B93m-Pt;JxRl1)H9_ApqK5j0{r3o$?xvbyMubIZNYyCA4e^;;1OVjZoLH@g
zveCap57h=a*Qe98V?D)7Wki6$FmgrpIvHu>!R^D;wo$&V>;^nvM*I7~2*x!wI3{ri
zR`_m%+(_zX4P7<O%6S11DH7i>`PjZ3n`UQcFIryOq%1u%k(~w4#`!a?>Wgm~&3nGl
zY5)EM<bnhg61@SyO!yqe2T(A|z+eS`7A~Z2^8balESRckt^Tf$V9m(Q#dDGER2Gqc
za<MPL*b@nbtP#<2a*6Hi$Qt>czr#htJ%AZe5K5WsrM6q-5m~t`Sl^WD2CISVJI+xV
z!oW%uc2kA8x)yhMCTJZw^30nD(B;8tGROsOl8S@{+nRNdma`5Mbsie~^UkPB$B)`O
zfl?|6+Rp<JF!Ha;M?>9ehzPS)OqdXKovVY)Tn)BeZ7W0FCz!-*f^N=b9LMFhKchz#
z%S0cG`O=K{1cQ@WdxiiRq+B!Y9Xddr!(_B?SDG4*fDiWYsE-9tLU1pN;RYtrKS0wt
zC#BW$j;311=vIb4T@lm()=Jaoe>GS(2Iv={C5hx5UOnV^o1ybfu?Ch-!>9<zm1cit
zei&MWm&)YWC$`gN(<j7@2b_~KH5NG@5(ayuLTYTWzfmLIVjrfZb!b-xUVXp-y|h>Q
zjBSXb;M&j~Rj!H)sl292t~RU>?9oqt5bxsHF>itK5kGW11O;1<DT)K8kX#d>R*`#R
zgRk{LsOd0XAM&|O_h%q(4y<1L>k`Ox{#mD9yMM0_FyIqdh+oiOLvBdVSU8eK^tJLM
zTLx<Rq9xCdgVNB|4?PM1$WQzXX{XDdCa7H3hOPESUs`wB0b7Th1y+4rB|l4ZGL4my
zR_o>GcEpR(jZJ<LOJ|%ETTkBP1J2ca5|<irBf)`!f9J`~MvV=K<(h=5?Y9JRvUX;@
z+?qgScqu7qnfx+=0Hm4eU49jmsS2{zGa}7GQ!p0!Yey{j0?<d^LIHr<!*9CnrN+{z
zZG(A2x@`HaElwMJ-R3jfzMJ@c`+s=`z9V<UIVjAOx5%9dr6%^7aJg1<^5C<QyV3~j
zlzqchDt8C{qXi?$8658(f6YQ|+Ch$+<=&X6l-U~h`R0N_)=I0dm=x>bbdaMNRyxB6
z0yg34u?~PU6O0Sg>A?-}?hc;iqeTl>Gtm=&v+bx<BP*aSF6GJ1GAE#ghL1Z5f$j@r
zu03NkpDv2rx7`OOYo+^+Z=3yoo((*FLcPq-LyUZO<6;Z!v?F-VytxbQZ)D{`eFfz#
zipdHlA=e$kG^l@ZS|vQFI7FKu-ZHI*|H3IIOY*b;Vc6{cG$BPY(9#$a5xq6&lfwlp
zbJ%eFC^SN-Unt8B5p~U|8L+~(2o+X&ZLRbfNQb5Z(C4~DJ+I6HSmmQJ+KS2ndr7X2
zZ}RspzR`l}YRg&U-^_tm*Wx?cWo;ZIG$)^zlXZs6EaGT8;w0!v0VRXS8(plAJ=S^w
z3=gC$g${`gwg$wlk`On>eCn2>hP^>HCAfL}s*EL@4GX9`R4>2t-yse{w$N}6xj%>(
zFwT?*g5pciX&e_>MT=}PsC)DoAe}bWu%^0N+3I82L7)S6ojjOeV)P34bhC8Xv(DBD
zD3o4rzUSZDkmX44c_<L!5?<aT_i!3aGp;};8{`k^?N#cNK@uvaVetMV=@)LP23`Pd
z8=>Po>f_MqxSqlY`0HaqrnL}sAyq$~zNY_*0u~JwOo4mC&{J7m!;ptR#&<)xm`Zul
zzauMBaR+S8MBk@k>x-({5B=%1m^FBXV)s1xlVP)wF?r7f0MIuqm)%=E%d`Fqr7tZ#
zqmdCT<NN1Ck*ReDCfCd?7zJm`b3P|e1M5$qW`jI$3$QR0WX%30cE2ezwMJgZFNgF9
zSilC9TfG<;%Baews#^GGU-H3Wat686@Z2tcwFPyyR^{a7n4ESKvpUKvw(VCRbBX-T
zer7%4TxA_v;6aamH9mp3O`)5u3tmgBpACuvFW8ET5p?s)>oGzF>KSP-jhws@+XE{K
zik!-ujv<ZJ00ziu>^`ze6K@$n1Nse+zsG)+UC|7w>Ft0;ZBlGG(=P7>4#w7?EDI{1
zzZ)03C6+?y`-ch8#plU;KEPbCn7O1nL@>Rdo-x({{y=Qv2eAPcmZJ{HhiOY$T$X8)
z3{0~YfJi8jf2Iv$j|T)wZk>+bSrOo)IJk9aziB-ZKep91Pt3_D2KtnmiD+w=#j*QW
zEVmw+4=wU(Ag;jlu7BGd#=b|@@>u|-w>I;fNWqq*1tn8HAKJB-22@>6StIL#rnT}#
z42rF2u0vFH!s!wX8DreYm)k7Jd>$0F*2!1heqqsiHXSl){%aqHsThpNMIMjHFf;T~
zLZ-=S32yIi3=5MK0}f5y;_p=Tum;&~nCh`-A1m7rQ!~d@2t#aUxWh2D5aNbpXQdtO
zFWi-EFtw9U#E8ln7>Jke&M`TKYwQ{$D0lH^6#1FWKs904cm_izv~E~n*2!)TBm-+v
zjR*c`kSMLZSOgP_c2A2~1YrrZfP<+~J_;+sv4^QiW(4Y^eruSUsG1roWe=Z=q)YEL
z1JU*zmWq0)FMBb$oSH!P3MjzGV7+^QF@2;y+C6c#$lhs%GqDX&<zQ{!$45hq4CP+H
zOr?}j@%tKLr@<;X%*})deS@j2N%r%J=HlM-7%2>=Wr2Xa1@fH)qPcS5j3R+=y-xPG
zp-{8W#fSrjshyWW)^cEi&jOE-Y$XTzGhTDZ4gh91$ie;$>aH?4MB2$A{>(0W+R8S`
zp@!A|J$lJuX#>>I$dpRm8cGaI+dn*p#70xI=>TUxBA$UlNe~?JNPk9o8;%ubm&sA_
z3>=N?+<@!q=wWV!L#P%KD=4lU6B}FGJVA~%TJ1z0Z5@6)&Jo5N2n9_rI3qMxfps3A
zP`!!5B(T8HwEXWU_~Zmn!)KueoR~&SrAi4FFn5x_-Hs#xN|LNudGau~d5=Lt6N<N-
zVpu!*%1h+b7#!ve+7Ea|pB6(ruu7A!=>a%BR#TXvfYQTW;o_R>bF5T1L3;XUX0D)O
zx=wnf;h8V4%t3kw4a2q*#%fC`<QZ(2KEvEP3{SyI{NGaP8>6TYM)$);=@-xN7cUD^
zP5)SD27%*nYYa$~kJAsv!cZ%<y2-%Uslw(KIl~wSz{7nnXF7OF&7cW22yIQlO>&lf
zIu5BW6Xoo*7Ah<MeeVVZM5vnsm}GF^2;_%yM9>G@Wk|e(z&T`mNI}5XbKn3e47`JJ
zpu$*@e{WGgs+dV}9L4`f+<QRRSsiVo1B8+Qp(P=e1VS^R*v8oOM3$RuHA}`ObV=t}
zI<}-EbR^p{y_bZRP(z8_<=%S>Sj3b-=p-Q!N+6-vPyz|}d8Ta1|9Ag;*Sc%n_2*k3
z%X-iI?lOD!?AgyWP>+6g1Xa2qrRK{}+j(WViCGzDYkpy$!)17IZgaLi-<D&7h!N3e
zBRoAz5^2eb3QfvLdjVDKnrmcKyg<+k6lJp2QtPW`DrQiYMoH+0r{Bd=KEiAUkUtRY
zCWVYs)Wigd$|TPfA<+jT>RBq&!mI$#k4~+!Mh5*wFOe$iwKm&4RjPxIBiic`k}hNX
zeFciKkW{r!@n>;iU}c%HG57v`kCSmROf_L{V?rgI3Bod)(5W#@*pM>vb6)_8F8Z5V
zQc)s6rv(Gv2gh9w87dgJ>2f)JM8sS`a-g=6Wj#8>1oDIyIn#Q8zQj~hXN8E+#ii_G
zN?=CkP$<lmvx5cn<i~A^#~W_4i{+g7g$82uof1_`jd7M2Ws};NQe8VbtcxiDHlvY=
zF8z$pSR@($MI%YkNe_&Z_0_@&3e0wV+8HSv#yA0kfJ!3u-VU`QuI60qVv9~~@HL`2
zo`FEL!sUWxvr~|!P)$Y?jG_CSeyB{eo=fxf$nPkPHnH0FoEi?Bj70!iXQVl(L3|Ed
zMN1ms_+BS@V*=!HfOKo|5yVpRd_Ol%$Z^?P)S$~stG6|$Cz4#J)+L4X%L2u?Scwdx
z4Sdt)zvv;S$1@G^$dV@QsY~YcLH=@_q?2OrBg0-3+>wL72e+e+RI^MO;p!X`W_I8f
zh>Dr&9^c`eP?(cJm;#Z=`6ihfl*Jyj$bl|7FFGmfs2*J^=i8rSbA=oubS_B4TBv6a
z11K&GN&rQH1tcRE8HJ1U&>JrH&T$h45)EF5;%nOAB_q^$njhcT&T8y+>Xd(JkjgxB
z@0*TsS#a59)@XO<A&1N5!Ap!>MRU~Ds5&yeC1Yf-@ZWjn5Gad?6|@EW=4!dpSfw`C
zBv)-SA!5TT<?7grP=07NGBRFcd$Bkx*ZMa+x*4d$JN_?RPMCY9-q0~{y7l*VR{f=Q
zTvT2z*KIRXZd+E0oCVIhe#BM{rP1?Sw>46-pex!rsk|%fi)NWV{f3}@6)e!qC3K^Y
zLxnIuP{{z#VSF|nH&7gR=;9_D$9Mo;zlwd1>j#ob_^34FpiYe{Jay%pM{NBOE(<^Y
zfBB!>(u+6%`)Oy9Tm2c=gJuOdbINVeZy}46$G0be#&p9vH}p;sr3r834r4=2cL0$8
zH>p7g9K4RhR_+XufGQrqXKD7lEBYvIMM}fFgKMbNfE>(JO=~K1qf?UI<BbAE0DOs&
z1Go6CFW8Ww(&<0b<jgulGhw)%7K_*`#kF2AWP1FfjBNZsIXqux_%KKYN`GbwGjZ2N
z;VLt2TTiMf)U~7Rgv@frzoGjwnVqHpghm5{nRpIP!piuwWlqo_0~ZvAR~MwqKbZ@6
zf7X?@x@<hB?S!3cGy*-407K?kXYGycIhmg(tl6s7_PrqShxzEQ$JtFLTPF)c8Ua#l
zX~@c=I7LUG{>I|;885!TC!qY9Zn?yNp*1AZAegOnX__lTaLsC_xdQQTIy^1&;c=m;
zsjde+v^-SBCWsWK>XQ}T3QtEUmtZN2W@U^IL+e(?UN2<p83JQ_eU(jSFp7O@g#Bt`
zgJ$GySmOhx;f>lklTGO0WNi=sU@A<nv{u&n+muTTTH!InUEC$>ePHxoi3lM|0cvd6
z1G2$dL&m{YxKdd-%SPYa=6PTaX=>c#WU#aBdb}wlWofl+j#VkNDERohV34{z)7~)-
z>Rum+DT*dB?BA9Q@IKpmFvR9sVei`FGjgrImu;gvI0-29TDd>=UU_j*k^DXu1{g=S
zo*L%X7+GCwJI-%;Ah-=sGK~wgu0EKUgDMooDg6p~`aTqkk5Chy01PsHIGzC{;9(9b
z$shcgXElH{nlF!}_n`IRTjt??)SppU6LMoAsz1iW+aOSC<T0C2wz-x`U>@Iw4K&jX
zl~{PcKM{;S3LZpHntVp>Mdz%iyd&;=vd<En9ZyF``4-qJ<e4-Ui^^o9&)O>pOGbs>
z4*63Mw-~^U{5g6oF0YoqB)aBNvnDIg8S4%}@#(+DBGfb@4O}h!>!0_vrzXk*A-wbP
zzXrE%loxy`ykbFQVh>G|zxm8feo^7$B@`APc`-&cIM<qym*OvIS}pQ&YH2ABYnm%o
z$}7gG9W1;4_iflpMG8iO2qt;8hd*Jxz~Ki&hdRsPzi@Ar*P@@+?1a4hBdrS!Y#dj_
zhW>M#Ul<Hw&W?-a^)xnK2yvsw#p4a%@u)KXW>9o8DzxC{{#JB2i7sAm$BsH>qUI`v
z+v7XdUUw%lmqHFhsx-*E*4`joDc<u5+2APbj36c%mdN{QIIDCx^RfOP*b-}QJozvN
zl&`C6Z|RVaqCbF}IasB+z5ZqW^*Qtm`Ploz{=vTBO^04f3uvMOCNNz-3HB?X{BuD*
z_3mwbsW5Gj&uq%M4Dt^l(XGGXQUcAcmd}GIB}0b}l`s4aDDaT?XbA_ze^R^je>0QT
ze}i-TBQxTcF|E2h%4Q&yZHdBSgJH_7m+eX%UC|6=sNa*7?d|tEts}WZiQ$N*qML~Q
zioMX?*$=sVcZ@%(k!pP>`vWFZIEC$8V(6ORK5>*YBAd62UA$2{uwcOjyM47Jz)4<)
zxh3eI&D#p$0^ow)+ALskEJUr&2C6~rYOQs5PJnF}!9O(b?v~zW%>t?kyx}l|%I+nK
z52AFLe7!^gGX(uEkUe}zE`C)|?OEcmXMxE#_KHF4jg{*DJlQ)vBMwgma=Gl2T4Lls
zFm~+VZ&*)AI#(s%EOFSSkZUrY`j&q|rHS^vC2+*u*U3Q=z?}~^2YPs*Knhx9oaMJ|
z8pi>t|Bk<Es;L_1Ksl><Oi1awu|oZhJzBn360Vi-U_gmQcvf}D_hX{SNZlqsu;IXJ
zk#DsTW@NuKQA-lV9E3lNexdOhPM|Cq%9s2ot!OX5LY&@Q50FOo4?;k*X;eq517g*Z
zK1)Sz9Z;wr2Z^Aa>v{q#JkZ;Rq5|TZ#%UObZH$A0wGmZob=&3OAO#B)9%6r>?g0!}
zhk6H$|D*bP2?*Dx#GU;LG*`-D@o~T72Fg!@WCxxBbnkHgnuDb`#(Kny9ud<AY)r_C
zXMpB;1oN`$(vi`62B_vlIVxH&#{Ic(iNeGgl<ufwC2{Fjvei;pmgj5pnGWf1A6K@*
z-V1Bl(cVAdOqRNZom>V)gQwBi&j+LIh!>a&ms*il2f$zk#>*%cC4&+zrgO}=H?Njs
zje|>(4PTD4&eb6fBII~~Jv282YRU;l9*Cpm(78@DW*s=7zntVh(Bij5PEO;7?F0M`
z+phdH*dkvG*B=@ge`b^KvjxCV_r^n|$i|Xs%}$oV_FD#-F7#4fT%tCNjwv$4+r~6i
zy<rK^p)oV~81O8H-;@k<rt|HCpbqXqdPxm8q6}>&5<4P>h3k3&>TA7}#D^LWN3D?>
z6Cv=C!FkYcGBAy`PLxr-SetdzSeeiu(o&mEcjv^Ml==9O3J7Kgcsii5a)0K>8Y3vA
z!oE2d8lF_f(haF7lhJ9?S`8XgM}~oD6k0M}E^y6~YF`HG*_&jHH3!pUjhqrKKmnnh
z2K!CMMi>2#9U$X^uL%V!EAcgRyP+NO)DlPBhE9RiQ8NSL%J_5RXsF$_xPJkmJ<SP{
z1TNIJE@-&Z<aB$rrk>J~Fwq%lNS3<}u!NlH&w%Qv#|pSq&Pu$s4mXgqqemD-pkK7f
zIk5!F_t0-?l2^-3O}vVOUOkv=V~+4lgSff>GRk1cUF1+!TWei31tM7yA}YL8pfhP)
zG!~)AC{V+!v444fl8g^ZHfE<ty}zrTBcSAZoD>adt|UJ$H;^Pb?@do%p8626;8yI)
z3Dz9M)3i33ShCgpRT-)8i4~w$HQM6BQMFZ?j6&U=lk>ozo9!0_CbUW3`{1<)P%?HW
zEZs26EtZyEqvTZ!3XYuX^l?1qemd3mieZ;llMq$y*eOul3f?#r8*Mh7@X{8)5Gg>n
zfP6qSp406%xF%MgpOmB*k5t2K)XWKD&;x#QA{y{`43UuVln}4*y8)oz5!)A8&yeF7
zv!9w8#R-jdW%hqNZ2@6pj_1Wvl@%4s`QF1ER9@k2A2}f8g6LK4f@UebFm^t&?=(+t
z;NUMRaX=orJ}rE7L&i!%CfUE$(OoW%@wGN&fiK{?a!Dc$Pkr8jX6qMA<kGYv7QH(g
zB)H6ZAD3RZ+g%>32`-^(Y88!mJ-)(6W5*>(384kuj?zY}a`4@Sa%Hemq)f1lcvb3{
zs;1HYYVRt}U!Z4T&Ly~pToV)M1*W5U0It_M>GdzHC-2z!w|b*6a{L!@+6}2`os8Sw
zFB7Gi$=B>0*Y#RHqE6H<)za)z<a+z2yAz2uAcxxMP;rB|!?mJOd^v8nH~J`p_oz#l
z#rmdTfMUXV(OeD6vV*IM3I5uL0Ox6SK0q`Mrm>MeB4Fe>a<j4U(BUQJa*Is~J9QqD
zj?1lS`0D2cg3;O1vOvs)a&=p<@0Kg>+mw~tO@vfKYrP`~PIOKe!Kzg)mfytg-f|<f
zmpgli2z;eM?g|crlUN<1aO&mm=wmSQ)iugJF(lZx2*0iaYlKtr8UTU6b?T8tSVw0u
zTnamFityO&kZJzAY4tRLERpHXmdp|Wh%SbJ8ZWq#V+b?TWO!{Sc+YHatDx`CGV0d?
z_%6uoSc<Yrl$G*+SP_;^TW}<lE~8H^D&NiZY528+e%y}b*{MD+4HFZ=G#QRIKWJ5e
zLmXu00-IKAdu@j-v?(A}27&rT@wG~RGj=WZwY?GT!IUaXg04BF8*h`P{t09gr5jXv
zfNaqrQq5BQF1L4>dX~xP)G2m_(}whBj3wg_byBj@TXcI74&Ozx%18v$CQjScL1<QC
z($rRK{1fn)8u}R?xYnO>>uCZ?Xh`c)(@Hl*Ue^0Fx`DBZ09~6l1haq(dCoBtE$_5s
z8P>bnXfyyQMF(2hWCYFD0`S>v{~l9y%24^8t=sV8V!1c<jBIDh3M=Hk*qmTC?ya5x
z!Yw5M=C9RjC6cOQ6pIvqhNnCJ&HV4UXh1f08)=}nVI$vfy~8>(MSky%Xh?wNe)`}2
zSGGDQcLEukClA>4TJpMH@}Q5?e8uU@!M+K#Jrt@%X-OHbe?#P9e>K>1EkrLO>h4=3
ze@MU3vfRslD~|;GkE)p}k2;-ZVKqFgRQ?#ZI<5VHYE|%)kjIR+XPkeAJRaYuE==(^
zU7kp*I#-eM>DZu88e5_;958YS%%>9l8DL}zz8%j`r`00Ofnq^%d&YF{#wqfwb=`uJ
zJ%0*LM=Thd{BsNiN);7H;sJm0XUvlk_w?uN8E&C9wKxuey&#I$-?JStr+X>RlILwb
z`}RG$ue{()NqY~$EE<FPE+IC0w)`zrQ#Hos;0M4{;YI(-oxq6_<Am>t3VtcpMU4?`
z4^|A?zU1XtTvVJ=Jo^=6l5tkZ=I<c~>NHsFRr|3CZYo%v*J4eJhE>Wxf+%R55e!Um
z`A>VDp2P6mdp(heMmIZp--umagG~=eEF*8GT}eZ-874L0%m5R4%YUc-;d$t>c3gjh
zb>H^wm>nzcq^42=Lsvo+o|1Qaeuv}+vJ3qUv{B)L^IYDurb>p6llOywYO^sn0T2+@
z*bfqa(Yr}+-Drdt{tsim@S|leMY!z9NB)e3fp|bMmHNND1zc3<?x6h57P{nPV~pA?
zyiGqbS`GuC0o%o=iQ4MViMQ9vXZH357o2^D{5zJlq^z>E1bg=L)CmobBp%fzMsa>&
ztB3?Wp#FcXjV*UzC;c~84=J5n<x3+*zN1EDr2CwKBcX$+vjcf*p=>wOUG^|NIxFM0
zCfoZnS|H3$xm0%W7U0W6$QIH+Y>8hPselDMV8+8LAv^ku9yEbY=*r84SUY*MRAlK%
zL8WYG`xpV{@S$mwT}CQ6hNMuX@>Op)+op&1*G2|ZM_+8-j;KQ-tP!MP<iM^Yx9$j&
z=9bIvTiPtUc~{7_MwEuZIkNjmzaly^W8nMO{Y_9q8Iq{q9`UtZFrla}l|6%k1j>?P
z%tT9jr3uqv))JiBwQO%25^~#|d8q6&(s8|f20QB;@t2zQzG;6kH#Yd@Ti!iY&dbms
zU1#<U@@v?rf|FPTnsUEwWWsr_o@n10sUQwgogtQ&%FcJ~ckF1Wfv6BQQjy5~#CG|<
z_XB?t?FX4v<p(h?cqC1f{etGb%v!y7_74-Gm@`q^ksqbe&wx%!%mH3^BOOSp(l|{y
zV5H-}gt*%HANjGpf-In@l3SmZ14k;Pa@J9DkT+<ij4muTQk6P6m~M-3C({)fISz>-
z0|KRw<a*W67wLtWxa6TWK~(02U%mA4x3TIx{S;(XF^7NiVtttLV^LdMtNg@9)Y>A4
z`-sc|f)>76a)h^JHxji$Q!pe8;E~1)r9+FVhssgGijbiiw5nQO-(dctk<u?skg1Zb
zAxfAf(m&B0Nz1?;$l)bN2hEW&Har*y#7gE^h3}Z)AM6`O^ALnn?*<tdYXRq*Cg{h2
zJ_gxXu?wO29-G=W_>NoYIDdw~D;V4%84(P1y!SvqQbO)}zlF=kd^y2M0~xHbl+31P
zn4~S)iD^a)iCIFLu8@;PI;=3hy*X2(wp?oaCkKhj%H^l=8T{<%Cp!(-=bw#qn7b0b
zK0248G<^DGbV0B$lEMCrumfGGP?1K8qcvD8=TopiI1NdADcc15b3uj%A4B75g)wQ^
z$gTT8=A86D|0lzPAJjjAvNqfUD9P8!2wNUlZR=rBDzRO8*inbc$esaXElJJgGRhX@
zETm^!B&CVa%-}(w)ZFyfNSXHuFkiX(e>KmSd(*sM@$g-M!?40Aio9TXsSHsum=X6f
z+Fz%9MFKvB@TwBG)WOFG%v&ebF<_h)JQ*lK#{>r*cGTf=ioZafQ1FXxc`*s)6Ke9b
zA|c>I#u;HSvDG;0PK_miMxUYQ_s?UDWpL+dkkg{0%F-e^J<X4W;RvfNk~0!b^p1^=
zXs9*!bH-br=>wrM$&-;+Vua-YX*tU%v*mWs{OmMGsti;D=EylV(7L>8i8Uh~a4}^{
zC)CT&*QHafO^3CgC+9L4f1NdgO#U<$XVPE|hog@_!Ch8?UWqQVCW4Aik@3bb!+>el
z+dmr+_1WN)!2}`vOpuZF7Dz6piM>j<F+yIp`zH7=`j@I1;q*Q+MgyoR64G+BaE~%A
zYNL%E#feZhR+`dER1KBp7zp$YI)^BNMleb8u||53si%Xq1cRvawuz;4W1WdBHcD&!
zSww5M6rvT}_nC!uBQ{B!_sI#<Y^6Td>ha&6MnhK>+lxv9nH1}a)1P^E76D40oZ6;m
zk;-$K65UrfW|6M4!(SoXKrMvUwyEApZ|l2J_&lQw59F+z@2_>K?`7{gbr(3fZ6W3)
zy&u7Opxi9E(7%j@Ldq$f%Lch9MmGjg(yelFtVI9*N6RI?o#5yM>R)R7SJbQ#dl2=P
z8S@m6DV599ax03@hy<nLN^E`F7p{nA3oR{jWg;n)5ki`~C(Z)7%4x%33wR;mc0OIM
zPHz)*(~arecTF@}iui$R)2G~J)TFxP7inLy2;}VU6;M@(=yFu>4#f6Yio4t|<6Ou#
z$aOa3?#@!0&aaPOhStgrL5ps>8r<mfr96iF6_tIoM(0{vwwprKO-A?)%w?0C<yWyp
zAPB=wcxmh9*eAb^PZ0i-8{a|o>t@@sBL>SYu0J%(t=q7Va)R7uV?)|2+9__gb?WX!
zs*3_H#&So}7=S0Vwhcvi6x?@y6USWvfE64X8syHPG~KeXE;vguPP*;g6=NILit}`G
z9_FFDjW)II<HyTAakQyrG9$nBhG_UD)Ys!$-07n?RN5Sorv(SLq7Gt*OpkF5uIXU;
zux6Q&rmRUx5L>iSX8JP&P;D2@3VtXn8Y8pQM`l;%Hk)H~vTD`R<!?|Aq#NI2Tr=h-
z5+Fk+cZhNf6Nt?7w=FX!Qz5L9`TlknV~Ej$6`X>5nJkDo0kr{=xiEdKkt6gylSQ#A
z$@IY`UmUa7Y{RWMw3Y-*B3pH%EKP&9EW=B2*evsBgbXN%umxqg(So(#8f8U#L36C)
z1X?C5;~7pm>MRR!-mOZrS58KnS!9mWi^A%d4}yYm)~rc?BGo}w)|wzF8eA;vg3Iq`
zRSUa#_A>L!`rxG2Y_19Q9^sd?As(St9m|kGvTO`0z@ChZ%aBW(tc!6#CpX*F#shLh
z-aTW($%K#YO`B2y6jjJ+=i@k7sq#Lbo(XyIHx(V_rY%kk*dj1x{jYTU+vd1ory}k;
zAl?18B8_>NN?Yak{vPl1WFy{Bjc8lF)j80^kiZ|(Y_&XK<7F(BJZSAe7;(iO@^%c}
zpd%q~K(o7N$iqf$I8(A}r~JWR1Zc=xy*g4`hmTm}$UKvkM@PEb1oBa_W++nJAV}6m
zgmg4DO|?1Rotovb*fst84U)$bSInY|6>N=b<O$o|9_HC0PkR5rcU^~M1TXj3Q}$-D
zhI4}No=$91(}>=mvEdaFUwAeay*!U9%36EoPl;UX6&ghVz5F@##T3QvW`udKm%sSi
zwmf!ye9k)P?mXfo`D=7wsgQ7Yl;_hYzz=vZg5_U`Ayzifsp4-zjO>&Kc`<h05G13L
zm+Vax^=`zTmY03;=ybYRUP%%#-;}F^6Pu5_H;`dq{;S5?^o@lpAesHO7-<dM_4D<}
z_AUP~aX+eD{%KPhUz2N;*S%5G&8RDQ+GKenx*rSQ3xwgk8Gl1nWX#@MX*el38^;wr
z2HuXH2BT9gZjv4Hj_tHkSh3zs+<;UfSYyCl@}4)P9H$$c87CPW_I~_=&Bi2fJZCK*
z{Rh##=2ZrwLhFF~9en6LP>Dx634rGF<s;w5R9n@2i}PIXJ@{rRw(-BxH#1xxUm_pJ
zGUArAvP?el4(Ytkyz6wzLK9%9Pt&$&C%S97?mu%yrfi7(I~Jo3Ib3p$L=!%@CCFt!
zkuOricIWSj-u+L|3fE(HMn{eR_6Q^>&aOK`2@W^OFJmxSI36^^?7`z(Mk%O<xN5|F
zG2{R^-FBlKm>HXE)fm?8?Ij2ZMt0%9xN+_<%7KxVyt)2m<SU~TUem125Rf~LQv9X_
z&WbyE1Gc41uI)TZQLuuh0oo;A&{KfG4WeJIxf0Y4XwRU?(5L_gk#r5y+d%7-nzDDb
z7rQ%0AdopPyLnqYA!+%fM!36mRx)I0`55_n%r6IF&*Hgzj0(UGObWgr;A<k$B;B6T
zwVGLYkL4MLcwX7dy3R9?4^FDRM=2<H*kOon+s8km<(y2*SaL%bzu~`F<_G8$I2jB(
z`(}KDLz6(Nx=a$Ym2agbScx-(i@QYj9pw-LKwB0j&d9g@ElaaW6XY8CPRw=~N}r8K
z%=mYMTs3q*knee)XcObKdJ0tI5$HNfaX<<Vq|yI@{e_t<2D9zwU)hS<$*Xo=`C%FZ
zF+rWZ{KyGx!CET5xxFBG^?TVrmQ1Zu7zr5cfSz$F3)2u}%kkrwKq(S{HOqlP8n|At
z!_{yJImmc#K)-%+u>WjGq#7~g5c~PKL4)MbG~AUqHYw~DNFVRood3DIkZ?r{N*!hk
z4B-WoEx1O*3b-Wn&FINEc6j<-{m5t$0ax+gb~qoAdgc|_V$P0CJ%{cTZcs<1X{=&O
zFC-0_BYjg3DMhBT1^`aAzw`^TYl;!bQ_Rp@;AmytB>jVs)P!o}=-ASAH7yVhR1gP@
zQaBU|uol8}yG)Mp@sU9c>5rK9fx(QZtDDCKvLJ)f{5FGSHSd}nYx@v*9SC@wHI8z}
zQ{V{JBF7tX$Fbbe3F+en%p|5Z)e~cU5MYf!%@B}tQjABz8SYH5l$M4#+1Rzab1)_&
zKaIcOe%>lSvr*JcLKw0X#lR7}0mTIY{DbZ1G4Rc5m0}+VF<n+6f^e51F^m3#j*+2h
zq)6Zo^27Kc!(yb!n*o@ZrAvpWN9|4Uc;mS{BFNv8n}oXXC8HEat{N^Q?O#pfWmIb1
zcm$akr8G8lE0ldMQdt@zt8p+6aIKWbhgfu?vP@!95g!7nWRQ+j2Dd>(QBpF4oe84R
z&N;n`nDNGeb!5hYl~R?uT!Ahv1!A>-V_0PvyXkOg+&0F0v$LG22JM@?IiF(m1PqFj
zeq(KKw%69kxEL^~s0Cq9P0P*X2aFS8XgOS=<>&EgZLY3^PLZeCis#x;E*<63jkVh8
zsdz?^M!_P*-Q~;}8*Yz8>3gNGi$y|s7mnn!t)c1;lmI)&KT>H79_^`8SL45cZz0>s
zy-+P+%33S6X$IJi=m;$9QmxL11?5sL^N1{xj4?ucUb6nHmlKC`-dY(y%7DDyv46dL
zR5eEG({y3S^W|*_W*j;x50_@@g*ktXJ|<QgcbZJ_73Bo!VL>R)&>+*mC&mUYf@5H&
zMjB%)!EG9Toj4Gi?2VG*qI&+nG#mLQ<QtkLpQIB5$#{`OW<-@y)HAxI#onzfsv0Bb
z`r!jV85|9*X{?ZfO8Dtgh_MVtN^4{=X^UyY|An?L)GLy9=Lu%ZY?euBk)~mGtcawA
zgwXzE|JGDrjq{W#{!D}ZczD5F(2?k~0lH(H06m7jsWDFM8M-eqr^b2lj2I&rU;*&5
z^L-?<5IaPCXTGig4Urc_uV5s`O;97fFr-g0`0*lt!44MwceY#{&j1t~guF^F@uAMe
zjZvdn=xB%`z$UD>%u8cf*iH2I!++&6AJ&|1bHZTxNjrfrPd&iB2+w9fA-}@8k*q;x
zp;s}JiCB`4aLu_gSfqpblqtQgN|UEY3V9YPCMLQ}t~Lg!$JJY|v40~c*hIP3nSrc|
z;yW=G4GCl6Je1wBx6eNBh%@B6G_%DJayI_!V-&e&I`QMabwd!ffB!*pqkm#<yYwPj
zDK{B^VF2dq_bVGjcV`_Wy!<+S!60i81YC4;yiuOTb?TN;5m=CARa9%_R$mpa1Sv$v
zmR_i=+?M9H1OSoxE6D9JQWmGWBl>87i<$f;O?8115sIbUX>XJuI|kBZ-erH0?_dwU
z+yA0l&Puswn>kqKhTod>1CWghs!h@veOIDV&$Jkv#uu2Y*K`xZ#SN?}tZ16Q#LYqd
z7pk?)^lqVx!6G;-)&j^BQk3BApPi;bUt2>Z=J+?LehGN2%hs1^f@H3}4cD*&l8Cpq
zP!FHyZxid7E%QUr03eL~aRw)V;y8X5cz*~AQN<t0>|~)oGZm2BJYN>YGbjO@F)oM=
zSsW_}j&;8i`~u%&Y*xD2F7e+9{NtPdakW^Q+GP1$K$I{rFAFO7JGP%J_ibgTCddfA
z%VaQ(?m*0VW0Dnd{0+<1<62jcmGPh~-->XsdRY}BA8;>xo%v|>sI6Am4$F&80OJMl
z0HzFI6AJ`wcC)OF;S^N^akHNY_rlh(K6*h;)QQw`HuT!ebT?5O74b^h81lKO72d@e
z*_4Ef<wRj{Grd&1WOMY_STjj}=k(E)(gv5xz4i>MU9J^*K<XQ6n&iG{9{$Um{}$tr
z1CBorMJ<BE3JvXTnLN<(`@Pj3$F|M#`{=eEEV$K_kbwh_lL!1=DhhDEg0jCt9vr1X
z>D~WNMV+th?v#gu=-_QAFiigaupQ~+TICNhjZC(oBh$(>u=0qr3ykF&fkGV0whrS(
z7s;bO80hz^Sg_jiADxnRt>W8ca^6lKi|&gi0nt&*{o~#Mt=dfV4KtHG;m_uFuheI_
zm&ubs&&J%j$de#X89lo@x7;F6M;o})(tDcOX`hMTN7Tx*M!Y(N2>vP91r^g<<j=8L
z)Ja3NLae!3Z<qo@{t`<t=$O9pT$&b*TQMyrANcF2t(vGH1=L_0&>~BpLih&bziae@
z4&6$gk1jzzq?LndiO>kMG*bG^kr!N&v}C8s-{M`U`czGM7*m<0p8~IiFWS3XF2mL7
zrCu~Zwo;nF<mH%bp*@qAS3+@wRiwRE{_aEpF!HK@OQ$9J)XxFz`I<eWgDccv+(U@V
z*^vPftyE^%(7hH_DS}2Y^ULco3P!@%Smll2_DUqh(5?Sww2kX>TLbJdNW=4%jTLoq
z>BsZ7)0$LfggSj&?WBAs*7cBs<lX2^0TAj9YvsL=02Aoy2|(}tcz|r7njhtZpvo}D
zFMsGXrU8W3I2OU;C4k}lwElkNeUK`=j37~<{3|GlGJ23PAG<^}FB`o}+hFl}k5}R+
zF<5lO$O(oDeHueyT`~DANg#xyfyjEp!rQCl-~KJOJjkJ5m-4yo+hL>Ri}W2OmzmZq
zSIB>2pg^yY<Kw@v63k%&IaQD^ZSjEVBHl(y6?)?}MiukTfLyjK4J0ZVuYGRH?6Q5S
zLvu*>HJ-+h9ZD5*gJB!TAIf}Nn}G9xHD(NN+%ekh<r+mLL+vm)Zgz@s!081M@Gu`N
zlAZk-d{uNSA?QnXiM}Tx+_$l@M!s6A0AJq$1LbR_dUrkTG}$$^M)!MW8JH)#l^S%%
zC05HN&{yiRYh?G-4>1C4VZd5nFICtHryK%1WRG|Qu+wVL0(FmNPj9cc*8)Q|X3Jh_
zuJC)%?ap|YR6kMn_AgO&h$F#xIAfM4+{gRjWsNZujKu#&sls`a5qv4%O!K1PWZc(u
z`BteynBASD02J)&P17<%^Y20o!1mC${ZsNZD_8Qs{*Fx&XBfPO<h!YJ7P(PZG5MZ7
zt3d(<6kf8FLf<d78((jr-xP!1jtf5s@~A&4@1>mTW?ZTE^L~iEnLqjl`C&ZMBwF+x
zT_ZnoGTIiyvZ-!DX|J+D_V?eBUQj)#p#OtpOsz)!eZCwJWX2*}R1RH=2vYyp80CP|
z<-q@&rJ@`q@sU(j4l?e#;KZZk;2_o*-~e)ne`vNC3=XIt_b65cQ(Y;C+J_UcCVgUV
zmUmZ$z+wJ5m<LBFS(0#BvXaYB>~kg{J3OYL>bW66wxma-hLPL~L_ws59BJGFRH-0G
zl{z#F-z(mQ5we=TV6Z>CT`2v$ee6aI#UdEYP}1Ljp)r$XSJt_AYnL2t?R0kzWh4Zy
z60m6wF!~IGUWG{?W8<i1a;Sl&ir?XuVi%4<X%q}jG#C71{h8G?A!{s`<E)J;`XwA6
z`yN$NprZhr^AE71*tB{fmJ^L?MdihEQoLUc*rqL8FDLtC=zl;Yd=|c<cDzP@ntowg
z7Y%#X$s!q@ZLM5$9?*Z0jRr;OD`l|FwXPBRRGSnBHIObELLIl1f(&umU)OzYH~+s=
zhT7mxMHb>=i2&*zWpS+57pro#$O!9;YLAq}{_b~de;Jv^N{-P5B&R}$80Ek42%+`Y
z%=oy=Noh1tJ)~61VlA>wwXGd_DNkhN)d)%nRIv(YPK|;AM=-onIRZjq35>y&-W6pd
zVNnbFpgZ^I=&CfEmnv_1&i$9-@SB5^dbw1`%wg-Ldp7sj82bnoM>u$$66*#xa2zig
z8S9;GQ*cHN^y5mmUJ7qQm_Gjx|4&XW-MXueq__bvL6(N@pC=m8h{AJ*7f-rzoE9Vi
z8U&>cWSK9gCsJJk>=@Z|unK3yN)1LhH+IyS={v1U2hz=!vy4x$+iT=(<4JgO(tLJK
zysZ!j?!{=a#>WIj(yag#i#+XWOSgVqCjhU(9O_D-oD=lsIv4bFaegugi!8;1g=m-F
zlpv<pNyZNUmiAWGpp&c(2{!H~8E<m7wn6HHrPRHXz`ZA)^3$my4UNSM{D!I*v7EJ6
zNBClbb=BQj253Jk6Mc?m?6io(Mx#QubrN(PY4SIiaD!$Z9t+LhmC6{*$MVU-D^&8<
z739<;X^E~TsT=sYK~4xJre&RnQ>4|NLCvX$w<W+7ki8H?vV>EBG&W0{HAQ403oAgo
zu{$zX<B~hcIHwJ+#4MpjW9VdWg4PpiM@%k1C6R}_)ZB>GFee@Ub{CXXq>ut^J5#2{
z9ztwup$$src`+(gO}XS<a=t&C0o<%>3;qjsx*$lS#v}$UFwh4DgyljXG*ZIV%0>1E
zKr3w6i-S#TAh|H3OKbt~-3;7DE;W)O(6%6#rJ*v!jt-<V<Z^#TB^QiOw<NhDo*_XC
zd=_v>&5<jkm*Pek46|~T{aw+TL-82&dUfo;_U6e@3%T~!81q#%$hAH#Q&M>8WB?rO
zyYh>k6SCPLxL-P>nv)fwOnN~xYk(ne*Tq2Kk6I7?`FbBLuTf--MZgYj@cyhkfcm>r
zH-IR<(ch-X=8jr|{c}@rnmP!}uhJ_TGJ@^9PJSH=S_~zukgbuMjjXu>7WbC;K=lJ0
zYv6!eZIr1V-)->@R3$=qaQTtj?K~<zrAqEdKeK2YVt#O!{!L7DaBBxWjqdb!42~p{
zv|R48ahb;wU%K00qLu<p2<|`A<ep$ET-52+z%XaC)%-SfYau>Jk0_n~w&f4kI_N5+
zx@1}qp*D}8Ff8!&Xjt!WT~M&8Vbbz7qt^n^NsXDBwPTr?<j@RNpf#x$v&3NQECGHz
z%Vs?kHkuX$hs*33B5^bj4%Z2pW9*Lz5hOP5@)zjzO*jrAS~54d3QCxjPMMdQA$+Ze
zv&;_y4@1H}sNW0x9bQD~^a&?oSs33b&zj@bqEg3jndQ&YGZtiV5Ql#D)W(*?7@&Yb
zumC(+8n28-Mb0*0mdmVh$S=eq(uW88XnE`-Iy_>7BS2|I8UvG9Qe~PcD{YhG;($x0
ztcpDY<vSIofSN@Bo{rC}gTt%G6pa}wYg|Ii!YRH~)~0FChXmjaV!5n~F7?>d!&BB9
z%cCqtPBx_9nInA;r0f#e7^`ylfd|Sa`vYNRbqHnkx4A{B`K<<Fe-|7A<z^a|ZB-0h
z?v1y3@xye15XgP;OkMWPvww@Xz?%(0#+0l@x@}=$s%E6|O1Zxmd3fcd(xiA5{>|{T
z-v=Wg-a7;EVr$&5m}mps3#<?Ar_dODwFF!I0YB*AYzyV!K_B7V`(_Z5gKS+a57`DE
zhQRiReXw?0p^vIV(DDZ#9KBN`I-*xRggg?2QaFb0mh?|&^1w&EOHT79++aD?KgKY?
z2?S22V@E$03xgU3@_6u9HN3k4$H7tSiA0{B)C)ZOB8k$ICdY@i$y3GxTdv1h{dD@O
zH%6get&o6o<(af(xU0K)=~<tP#wvlZ3!ur+e7j8klopfmQ5Ge|x8RKzybEBd6=iI_
zSxP}-cua)wTs#ADW}-pX`}}Kc5cp_x$n!x&xIoZ7!?OL;CFliPBUG-FzxB)rx}utp
z&;h+ZW`j-U$cx^M0f)T%FO-*3*95W@&XSjH#Mr!9T-xOoBO06?+U4&tDwgb&SM8Ug
z^5H|}wGbB*vmMy_nS%Voxlk2U<v(22uq<>Rp`(Ygg8b75z3m;`mV||>UXPhW8&G>3
zzo#sfH*743<vb6$ncj?1*44>d_6vGghlPAQrikURXxIhvj!`X(iXHE!HBCh}&;d#N
zp1su6GD+U|9eF*N@&;&bA0$%J+lUfjiF{~eg9hKq2<VUO>#B-!`Bx9sYR1dQ_7^ct
zK;@r!OS7OqF?%2#Vm`G`Yo>O{XTk2s0R`=kiHiRnU#Js~<=gu_=BRs}C)GmvA|{4o
z9hWR@S@4!J!hb@LpW>&>f33qo{rbw6zH0x&#)EIw$`6j6CsO7B9%hSa6lA+HLje_c
z10OAelfd@&th>`rF<`+N*}>m4<c^7(*U4A>+3fCR>OxP)-W|&fK?JW7Q|P(<U>Nk9
z+^Nj4I5X2yPaxC{e@NNcTeE9Db@X|%OY{n-62v(`Qf-j0mMQ9WBvJvam#>v6@Hb+d
z?CKp`Vo?M46&A(XKqHmi%C=f*P@>WevU{0c+=ez!%E2jvNTqz;J5%Khg5wf!@g8Lg
z!8PV;3$kbYj$er1Gx~0?GQU2-bm)mkK&-vfR2au?LUW%)m>y%jrGdls1<>m^$`p1)
zZc;)H-?VjVf+sI-7dRTeWliG9#x)a}MCmiJZ<(V!aNoq`t*Jx4ZNtSqCL`ZTl-B4R
z9)8q6zy#m*;hColt*`6kd;W|78}1pvT{p}3%eD^0-4I+stFqmidYg(}t5)rk>wPKS
zVR8o^sx#yVWeU&M;WoIREg9|Y1^J;(l!5awoc+lAh1UUi$oOji7#}1wjgk<!a)3Xx
z<)s%8B4OiO7+;v+3FDLleb#1R<sFI+0uWzvP<k7Hf-c`2IoK$O(>FIMr%`i<*ju18
z>U1=~ng?_4q24C1MZ9V8tT7y}kIxZqjfbUa)wI^hPhuFV`r0QJ0CITjz_NCh0cw#W
z65|px4*wPcn>=Jz$&uE_DOKfiRB#)7-Wk_o2@?9IiP43HSNt{7&!5qPfg?rD3jK{a
zkWw%&N5?3Nkm-u9n*;m<iXdi>!F^JWvCTDVung?=tLhKfn}gCbnzJ;ggB)9C0C8{I
zh#L9OVlq(6aW<nuORF58b_Ehekv+Myy61wG^u&=9yc1mG)z^wnDJS|fHcWRFW@kOA
zOo7IHJ<V6JOr4y*+oTXXkTLmbY?pSp?;)BYBR{jRP&lWdqkx4jN?S&&Qb7@ij;V;#
zVE=_9WC?&&qC|>&F$X<@ksAWU0M%tkj1xK;v^gO1C`I9GfJl8AmVU<-o?l|!hTBd-
zT8(x%=q1h|7jtjXPjd<O1b691DG4S+tS%kARx5bD6xXwn@r$w_ZlKCpi=ncN@=jF}
zpdB87rM4c&9CNIc1>MjXbzs&TE)ULzAz-{z1kGBrwYZW>rFW$_Bs6Hb-zn-x#~^Tp
zVbxAT)ke)KpBVIM0%5%A;!Ap@0cM0OtPaXR4`4}cO4>0!{VL3LN{G5M50J6(@3!$W
z&fCUC69R=t3n9z}=!eSPY&kVJzio0BS!z+XA|4DzGyzg5$Z6g?4=1b}PhmN|Y^&LP
z{lH>64LGEs2l^O6nbb2@b9BiW&S!vryF0gB3Pt!#rv#l)jmr^Ju1W9DvXPG~88br8
z4my`oxS!+SkmIO%X{aGqWtr5(_+Uhvj&>NlTWjB{{=Y%$jEfLL)0UTvt!@dR>1?ky
zH4h~Op{>YxZ<EJ6MLl`3-k%}KG#!MYG;flISOp}jrD)7qzsL}pukVlvF@xfYN>rY;
z@4GtzUF7PJQ3|(}#?<B_3eo$Kn?joW8O}-URqz6?nay!p*8=lFwOMNziPr%7NQ+Oy
z^2n%Op#+*I=lbIFG8wridIw~+MpuZ+V4=j7V61|5MSWsAjIV7mhXNAtN_$$lncxL@
z4}p)2A14Ldpk%4~cutO~VFHwq%*iz+o`H!Nc@I2YIzkNL!p`Dj1(|Bgg>+Q7zMmKU
zH|A>*1nU#(337!XP`yyiHF80(6{X`>&p~*hca0~&;F;5Oy<8O1r3l`V7aQvV?3gH*
zcr#>RT1|WM{9Wo2UD-sL4QMtMDhg{{gj62MWx*CBP!<W^HJ5wmDDQ_mq7@du!k$6c
zqSzg>^pnZ2^y!&a!Sjy3pC~PZZ|E9zRgfRR6l~<U-$en|tBp)}dlRJEHR(N&0OMy!
zgIzYRO=Q@l5ju;p9e-iEUw0?cT{Qz$0HXBEL;(6}Sime?k>P1^oj2ezA5&j1*ZVUI
z+F^*#BDukz&4qRkpF{e4$c_Gtrvlg!Xwz!BDb^R2K-=V3iD4-nt<Da-#>^oAqWRZp
zVl*bGECR_XU2l$GaIVmkVKrot8kFUh=wZu^82_zltSdlVUV3NAZLwTcc|fQcxjlHf
zzR{TQj-XHtN(IuT3H4}Wevmgj8J6NSyVFNM7nfD~Su$o>aTEL(`O4+47)d2UKjdy7
zF_zYmeqQt$N_UU{LfyskArd8q=ZpN-T7Y)O=+I7MS=?M|WLj*anmVXLxTDK-{}pJo
z%&-|!D~D=dkeT*dpQ8>vRA$Ah0E8tR4Ciy19TR4Pxn|8DJEv@`*O!N-uf4Tab#4@v
z@Uf~+g?Y|%50X_f*Qki=9^8BJw#<v^4=EZ`B=bY00ky$xVnMWnc}9A2!{-<JFZ9Nv
zVb0PS5nEW~zie}ksHI_XtY|f2uyOZa;wz&qVr9%5mJtFb@0SJ*Xn|B(&yZyy9r~km
z>hjb8?{T_JK3O)oL{`LXs*$z4Lsq8Qs9#qvWp7pLiJ@RrQWnW-e@2~wq4O+axF%+8
zwQ}IJM?r+bld?9}W{6hzpnA717`8DBpcST*^{M~f`v~})Z19N~4U{j@ZVyDc(J0rB
zvYe<k-z=N#Z-nR8Xo$z=Sj6hKT&w&pjhwu$p7nBXa3aJBGuam8KKllh%4mbx;@_At
z%`6?=i7J+Nn&<lcF;8UmZp$HF4;k?6HM(WdMu4doWNQooiZ6{cEZy^fPoK&H&mJy<
zE%bx&He@FOQ>>mm6a&`aPRikjgZQ{tG5sl9p$J2r+SD=~NGYcSQ+vb-+gsPSFHLoS
z)YwUVapjLjRX~O7@&8NYvEa2)xr{s>t2q{zE{5(r;hox{Lw`CZErJ)gJZY506j2gg
zp0ZutuW$bW^0ZI13kotrbQ!AhOfV@{259xO(N88%6EoZcV{L0&;a&Xa*h4s%+w0oo
zFEPdeeUFjnd`S`9k#CW|M#HT+*n;Hwm@4FuykLXs?i^PnfAc|6t)h8_CW;qL9~+2V
z%`XMnsn>z*Juzl4TO)av9Dc=G2X2AY|GR&}BdX`5V{q%MHoHmW6N+_t%{wP*Oy^b<
zm6Cs0yTv)&QUB?`n;Vlw&%N%?U?S5y3!%n0QaeZ~ioL^lmp6?ZLs7Ayb-cV~f0f`G
zjUDuM8scizx>u<VPsIP7)Y4i;0YLj(Chw+a=FPbdl|kN1&){@QH4Osi{qzjFDY*e$
z^g*mxOCu7DH_C_p4!|tP71NMDvP<kKP4X}QH8u&bSot_f2Sf3hamj=o<?cH9B=#h1
zxD(p7g3G7zXe=r!Peq!78u`rTuN!BX{M*|!#gx4f`Q%J&OMY%|!)8twnJ?_GvGCGF
z^wED}Ycege#wYwY_5#jEoby=As80W7+17pk_g)Yp<q9YbJ+-m|5iHx4J62T92zlhK
zm+kHM0+d&DyhFJHT@CrhCizNviWRWJ%M`gg+UrIK*{NJXxONz?GqQ8Jd->><svx_R
z8>VQ4Hu&wU{*3a98HnfLIQyDE(+C$V^D?MbyV|qvPFST{QOczi+jO^byL{m*0HPX8
z145vfgZ`J9O;`n6Ni4fZo6G}>1!3}uuiMa(WtTqUd)R2pfUBw5dQbbcyAzv!qU`10
zvq%7QG%=I(-u_JetW*bHD*G6{x;w{I$~Vd#RBgy2+Xm$SH{-`4L*!f0{Ly{;_LY5O
zdWsiXzP@jlD^NCQ&_MZ4{HzsEu#<uOeK(#V@k@9Le6QS4RE=uH@y!G>s{G6Mqq{mX
z*AHSKC8K4(_@x@onexLlCnku6eG%C0kIDnPCClOl1FsKCcNV{7UY-5@8x~=p$$lxE
z2@bGUalu#%_t_uYRNC97%7NvE%$CVPMsQR>GF|Lo8x*o3737d$9T<hi!}0x4`>iTJ
zxgdSg5SQq!3*`mmFrPRVmS6~>O+A3*C&u7#zDGeg-8qLFk8;yy<cRbk-AvS56Q9$X
zb!5y6v0pUjQBWOKu3&L%2TB|uu%)kmZ0C`hhnyt+?8WZRqN!9v`rD-I$IH>yDj)!b
z0SE=i0Po4v4=DS@*n$9XR%2@8f#vQ{gq#U*WT}J59~6(^Ytn|W`Ak-hwW&2Bi$jMT
zXMX?{L1~KP?T_wGBzKn+Vrps|w9v^DW5k2u@zWtE+4|wumW5&X<oIzi0v<E+Q(FzV
zlH}!Q-WE4CooJzFNl^@bI51C~RD)xJ3Om_tq&Ph@@3eI|s)pEijMUd!Z$q8Fww|yu
zX^a?q7#3fv&BXHa4EMJo0&wTSc+|#Ed)x>cK8@aVNGWlCW|SN6jZ8vB9}qabLusBZ
zBV%`rEtOHmx%t{gDGhof4Q)%Ku1Z<Zh4;m*l=}$Gs0zy~C_=F`S1QujXb?yEDsdv5
zb8Drt+@MjycsOIU4pZnivjN=M_8lFYG{fBVRf*=xwCXKb?UTYS3W_;TDT2n-A^{KL
zv>uZv#+WU`Do!z`XxI8dV`GTe(~O<n2z7~ZwhLvP|6;>3=}l%|C#PBiaEMw3z|6>&
z?Er_I793EMV<?QAZe$$LZ=jrE|D04)CujPY7%@ll0N?{T%Vr7!Yg}v4yoeIN6dWh#
zSU0t_6xYP0aABYq3le)st&i5mNq8z^!Rr#)SlR`^F}llyjGhb!Hy{v_wIyXF5{+}?
zV*uFkZ44f(lloYT;f?TNYOqFWfKf3_wK*4uD(u;8s2~$!`}H3<KqguPSXve=XtbY7
z%1dBbY_gwVaBXYokY?wNkS{$+38@nj(}zMXc_U~&Hg=1V9W|%hG_2;_MBtU6k8SkU
zSZ63~L~Gz<gOSjGy<iPBO=^+0m}42!=(HQjDT!oK56d^>CNkOC&Y%pa<cJyQ<*6f6
z(h4y<Afdcv($TX*idn0Qg<ZHrrdme`WkKate=7`^#GW|q`EtIEb5P%-<$}~aJRl%_
zXy&~zo<VWY^i?>3E;1q?cIJ5(94Qya3vlJI_}IB}NiZzJv}$lvUmCrG#7sfN|0=L6
zm)X}qiE2=a&<Yb@Zp}ABA<D=VHi;%iuv{5knVdpx$~AITu++HR#2oY_2*|6;lerK}
z2U)J*#5LAX148BGTAOn_5{dl6+cc*?8c^oRFZ~&Hb~=|5Ww|a`+lp(q%Jn`cg@(Zm
zi0r_M5NNp}G`l*4%dn<04j8%7x`%J&Sh>mDw<KShQAU0hbHkzre(-A}1j;GGPyJ>a
zb2U`8TSDaa#>pXKykwZ%>RSS^9j>jn1%>EP*bdAD`P{rSRlhMoIt<+5Vjaasm=OFo
z);IF4Al>?%i7!zG34CKmhZ<DC_iN>@_#IM!cUsLfp#Qc^19#il^C*aRkCV+zhm=*o
z%am-tjlF=Va?H0g5ro;q0FT2KGtEEdal|wxGk9C>rWt5@Y8D5Z-k?3yR%ZAwyf4#x
zs^=h^nR=ioP*E?l;u+9c(`=W@>}VTsT&i+P?_!zb^XAQyQS&??R8fLFgrt<N)Ecdh
z>H{Qmy*1U&F$bn^GaKZ*5Ofop!R^r1WWFr|>RMve7o^TC;~T^sCQr<hr{Z*3=$$b#
zbsEWnvwBf93B3>=v~W{fY)dg^*f3cVj8SH}?vbryskgyU3o8DU0L#+iqK*KbvCc>D
zSc>I7A@f`$tnFqrSrOC6u3Dp$glk(o(?}GGqvf(HEztr54S^|nlcqgOR!84ZE_nH1
zea2Y|pBU4wNxuWT$2W>NVVS5nv&7Td=nW$<%yXfv^JhG=c%E;76j`5^7$__F5e7Rj
z1oK&Mu)PGi0^^HpO#NA!CVCzaL*;nkog|w=O2FZ`G2c;;%`r9Ci#9VjNE`m04};UC
z*$l-S<lZz4L~i4~jo?dz4DL&B+eH_gCR_X&{Rk+g;QuMz)<SnDvVXAt5u~3375Dq`
z2SG>053np{Fwbt&*(|H(?@hNpaevvmO^*!gSS}C5^i=@s-UhLVgPA|&gZ@oa2!a+U
z4}~;r$hEZO>rmDbN9)6Jw8Hbg05{)1Shp7(B##)&*VH#deR$OSGdYFOBg~WN8ePMn
ztjQlk(nGIrE<nDidi0nzj=WuS`2X^FYyxD1%xOZhCqfwKvht+84fq+y4-TBCY=a!z
zcc45S$Ji<5TDIbuw3t@HNqLUEEzgGRqK{#X{3&)8eMMVUdH8es)-o!pRTB<3i{vjp
zNzY!bw=Fu`YX&InzvqmoFzMjhrKd@U{53{anuVKGMxKxU8?-R-3&ym_z|M@we+yAw
z3{dUG;83QEV};6>;#Vl)^0F_W;kfYcfmQaE7&%0mRoR)ZdwGWZ-C4@?CuIR+ZtMw5
zhOrzn=+(4{8p^K8KU8@F%Pxb?{+g|Q#T5BR45c9t;4FQnH)L!wTHk5pR&J2j6Mt$*
zgJK<TgrZf)6Z1`fL(yCCDf6wr6}$j*DNAt_<n36i3^MyNmMJ6eI21ySkSG`|PZ%pp
zz}tO*dI63X{CBRrYh%`0WAa{*Pp=;kQF!6~SSoM^6cPEr-+>fJ9xfloxDig9k&j}G
z#e}P1o0Wg1ZveQWW0t7?*p~&wchSazF^t0h#3^TrE>CL$yRh_ua&&$Bw1=`hJm{H(
z-;rV5chzUnn*n@Sq<Ql17*nAE2NN$LpQpDFtfBkAOTMsu)wgfIqvb!|iwRS!jX^0F
zxvu`}zpEao(hDX*tTMq*=Ct@S*bYv?3!Lys8G;QX6$<4vA;a``6#=G%Q4^;GaNf|M
zwy#il4h=7cy?ciWh2H?5X5}mPZ+Lk%%Z?R_v%zf}MwXqtNjqjtI+Jc14hH3L=Zf&<
zo2AA;m;_~)3PXyhUGUnD!pj&jkB_g~SO*@{U%uuq8Nfx}XIg0NDcRLW2@wvkVmJSb
zY&&0e_u&w0*8pZ;>=@41ZR{*mnv*@Ox1SHKmOX8LNIN`S_Nq{zg~eefLbco5XFwUs
zFd-J4+{d4Bg=lX^c5?Yfg`=5_T7vgFAUB_aMXOP3VG0a7S%1q1O;C8X?CY&@k}M?3
zIE8PgUe_>oK~=u**x;d3qweE(y>ad<9!*SrYlqtRqB$s=ym%pPntb1%`Q~PwdijCV
zN&~#G8IZ!zRY7bLn56ehL)ZXq!n{d;SmBoiSe2UdlN+-zH5BAWsdZY&)CXPm4@w^m
zN6-T*9F0Y;7^V~Zar%IzCm8iKO%99?iV73upjbzEiE(y}i8t2@yQv@Cvqm6?98%#3
zRdK!vP#Xouq1FY@fjkhaLLSIopR^zw>4Bpz1xB;O5;YMTt^F~vE3FDY2?obfGkTeh
zdgg3EjcBGE?!Oy*ko9RNo}oGpD@h#@Q^ha*9<-GsgB%SISin?{GIpR>bRCer_-kwq
zV!59&X-hsM{i7*JO4ZGAYdt!s*1!L8G9bM$AJ=O_OdI5wG-n39Fc%gD7cCV7t>r$4
zA1Z^4{UNPd<XHQ6(U^)VInMrl(g8pFi5%~Z&jEBqEoeTJ=o1oga5SM=gfnkLoD(Y?
zg>HeiKwq1aV&<dU;i^?9C&%AIkt`CO|1{S1Q0O8{<!8YLK4YYuASSP7Al@B|(u5a4
z*;ZXg1{*~X28?`U#laYLd3e|<95Tdy*RpLgG#bF=9{yg4-n~qQImNJ+pj9t_Fd6P|
zL*y!K$C|k#V%tNhvPhtkSiddPAfHAC*>I$+G!n_Ev?c}utOd^z_?;%D!H2lOk?)x-
zhIf-CQf95zG&UimQp)Wwb@RzcMf5nNq<V~0+Fw|r<xu)Y2ZbA{>lCEQR{fwpCrEXL
z!+&b(GeM%~K8DE~qe^obV>~d7mWWgA<1wYxGS+%&VWHGk85eBW0&#F^nx4g2DaxY-
z;(s2?TZZ7wf}9p#4KEubr#qnxx+EAmUC!t=!f;ZH>?o+==PQ0@EJ0Haa5P;j&We7<
zWLujGa<=uu0)kC)PVf}4M(Rk6A@Qpaw4#MlV-p+CtRq>ewHc$btkn6mDD-iU>G2bk
zOz;JK=@_feBO=?g?tB|?&wLpl+=%pA@Ux{^xjtxxMwz?ba`w`@xS>Lg6|Ivc=S}ex
z;I)o7mkH54#6J;a+)rd;%o-&)ad&OBh1)U@qiISbBtFfhUMbD-I?9Itb<g`NVZ}u`
zS7}L8UV_n4H=)xcExYGB6)6Yk=R8kyV$k$~%>fbD>a9U&)6}Y0<3h|8#GvOHVoC8|
z<eQK-|HbYs8Ze|1X0Eh*I}ll1Cj?<Gmq|8oT9xxqK_&->*Fq;pn!716uLAuOq$7Go
zf4q<rld<!3YDgzopQzo+c}DW4bLIRP5Y-YWeHbkMi)bdfz!;*as!T47j*9b)ZMn#}
zY|HicZ77qAW2v*vQ#;^Rg^R@{X$_zP==A5wrD=LhZmI%QE_0?g?^UV|c8>s5q$if0
zfS1SMv(1gP@W~b46zgR%DvG(Huk>eTpTwTRv454XoH_bCI)xi5u^4!?$<;wl?j__j
zyCw~Lo!*=AU%A$ZP(7wheqlcYgT@&PgG(Fe_{(Tw7*ajpe7nx6N#KVL0tSdNpy_%a
zjCq;ioaYrD!iP7A1+ZP*kS4$j4F&uLxiRD$rLf%Of6>@COMaCWo5e!RG2e`^zYbZ`
z@3?+)vo{MkL+xwi^;@=Ca&D1GyK!r*Gt78E*l&xL#^xr<?a`8!;j)<eJ7W3>iKKVT
zZ+w#EJ+JnI>Gc4RcX}Ht^Kb}lf{WZ;X*37~h4^5=jUGsHw=wC+3c1JMQEt{2Y4X{C
zHT<o&XtI(lv|c)679(J6K<U-^7~aj>MD)q@^zl3{xL%)T*n%I|?--eB1kafbbCx$_
zI<B1|a7l%Pm~CXjsSgP}C#DDiNBxSLQI~&8(=GEG&euQ&UIXTu<f=iaugtT5<BHlU
z^HVFdzLH%x%YxVknJnDlK<$OG^Ke7OiO<W|qO<^v(8Mm%tH^ZyFOwiGwxI&Gs&AGh
zq2ks7<e*oAwz@Q0MT&Ununa$17Cj?3Z~?V1WqCTSn0kr<7W9Q;P%o|FOfGW?_-utw
z+wPkdSitjko~(3MX3myrvdSl`${Jdu;<(x;LSK98Tbq=an^pEQOnQy~ZmzlXVK5s5
zb!@5^YZD=?_Awz!LyK71c%7_EOHaoall1Ft=?4txCmVc#T@V*5WMdkjd4p3M+7ztc
zi24Q@M7u_Vm=ZNoxIoMAY(+=VDfiy=JJY)H<Xk5ARk*81dvi-WvP8i}A|qSuMTSLW
zoAYfQ&^x3%Sh!o`d^8Zy_Q?I#MR(`0YWaQY0$C@p_)HzX)#h|;-vRPKEQjJX*p|dh
zq1i8x2mM1FVbBB3Q}&_6v&_jEk_>Ji^05D68$rGBKw56RKSU2;2;R}u)=IXJN78cf
z4rd;evvI)$9!<4G9yP@#>db+;{}}WwLOy9AS&sz|jx}#Qc|2yT7iLfc8qE_%4PsyN
z<Tk^^R)9pp7KhsVR16dPJs~SzWGIBE$<y8@qsk4Bm@dy41xL0w%d^p?$#tBSYvoVI
zDZG8PXXMZRh8AYixlNb9BvvFsj!Q3grdgoD1rQ(4rB+RnQ>plCYKMxDp#jgkWIAS$
zyx>1`?7WAdkh%OVvFt`19WRFHo|Ky;FU1&#Q%!i;KemiJ%!Id6Ua8o+r+PE7_$%O8
z++j2-@9);<Sw|cpuX_6o>suwS#d6>Sh>vFZM-WxPSc`|I$~0I0>HVqMA3KBT#9sGj
z)GeV5-hT;}<&9oE!Ys9PD>AcW7bNq1dDFkKlxY@E@>XyOvg{Gkd^?5(H{@E@v3e)y
zFr>U%-t{qBf{0MBDTm)P#;F)Lv`XH$Hk-0d^ia>r2SJ-L1d~}x14<4~#}Dmz>d47Q
z(XK1nL?{f5{Fnbik8PNU0IAV>LKD%)J{!hSakJ6~^^+KE)zC8e)X71Rfe0RxR8h{E
zoo%svW(_F(%M6bc|NoAYZw%7>HOuGTgn6zTaso97qEul&`NI0E2Iko){|QoNr{t-`
z{Wk^(d8t(~pn`lEedY_SO`?V)sZ`W$9KzkVt8_dGZvAL+`}h-{GDva0L#3i@)nhQZ
zuT&aXX0Vc7zjmxN%&vFZaf6Zw6m5{5DiwCp45l=l7G!5@vEOk((RQg+M6bKE8u6I&
z)k*{2G%+o9TaSCf*Q^!TTkD3H|NX9&4%xKjQM4)}yZP8ufP&;JU<caW+cQm-hrbn^
zMZ&uDw(eBr<UOJfhKsf$t*Y$lZy72@ZdfjR*@%$?hWTapP9tVJu0L`Q$UZSMcxx2o
z8<h?K;VkA7;9Bs_O2;0`;H|-%@3$($c^;MlY<cC^g6!*qV9E`qTx2?$Z^uxpvK>s#
zT9EJbOi1q!44nIJ3=RXQ1AtkfxN*K$sSqjhZX>hm_hY1`yt)4%=Iq67Ss`-2V2KQ}
z_aYiK`odAcyPf*M`7@3$5&-Y-tf7P>^Fn#{>_d9W9gx~t=5_#h!}*xOVs;V4(WW|z
z92m2)urAL*%Gn&`ztH;1Fp_zy#ga}32c6u-4M<#Se-q#?-UW&6vL5H5{)<UxqDVYt
z=(pb|k;H0Z5+Oaz*#K4I07d>JZ2}gOF&6BV;$05+7Kn)BnxI(n5&p~sEG`Z>4UP<o
zK)ldBgL6147#%l$Mm6_M-!XuMP~HaVXS?~}gU^uu_Rn%8aXZ>J+9V>H17Zmali|`N
z$3#aKh7Q@bKnB{B)K>>(T762rutbQn9>D)t|GjrEgWkU%XX7{!=?acdovA4oH|{lZ
zLa%2G{+tPCofG{T1wAdGL_1DO>$On>J6K4@=y$Ts?vOq}ihdgFjO$Der34D{Gn?=D
z8q`~W)GzX;&HYA=oH95zAZk;f&Pqm#eNCx#tdSw<uQgcjp)p?d0JAK7!)yfbXB{HL
zW0fZ1q=E0xh<G2!pgD3(mlA7{UW({#Wb6)T-Bc|ho-nhXoE)R#cj!3bwM+gXNK39S
zO(F#WJC?Apta9u5st?e0^8d{LNqG<g;fn~?mI~v8b`&eAwE1@<4-Qo8(N3Xo1*HC8
zm8izVVKnBilImEd`u3dqsE+ZMbaP^}0hp3goK1*8(B#C^Sv@0VtiR1DSxrU?-DX^R
zwpe{2aMhJly#<XqMho;kunqY0G^owcsd1~rhC3~0S;%G?6@7XdlSN}Q@$(spb(vdy
zHa0Wk6(~^7^tX92j`-fQ(y4f~oSk}P{0~z&ESGbF425jtc-%Rp#`(iMVdg+VwVDaM
zOEJ@FZSW6eEu}yB&{QbPl2#W}0FJ^bWTI2I$|6KpvOWbfZP<N{cF-;v@2zWJm=+Co
zimdq*>+OVL;aT_r`pbCDW7UWisdtA$nz?Q0fm;i7=?Sr=p){a!3<fwc1`HexNgkvz
z4S_%%wvQeaJ?;xlF(FtBx$?+PA<b!(8Le)<s5YT|Y<B<@tEDBxWm8Q7klnf76nz1C
z%rozAt2bp<Y->s_o>EBDSpX@HlLUxITTBOMZcT=PH0?$q>;@!>o0R6l>{{w%jU}_p
zFiZ+H*_!O`oS18bAAsr$Q_|uvL7`<n;_<pvI?`}g;hM~6a374#SJZ~%;Jg?jG+57?
zB<IH)%`De^K~MnU3<&hO(6%u^47td^r872huti{$i|yM&wvH~{m&9mCpe|lp15~@L
zTxzQa7kksgFZ0m{%)pp)#EBxV@N#b-$Ednt!T(jRu-(xg0ReKQ(~8P9U6fgRnfF;9
zyhunSR~cd9&J4@j)s<U^P$VXg^8fE*QLd@<i$AIcw;`ZX{jjg~(PC$-&5m*RzX;ZV
zHwnR9)IY%bOoR8uFVoyMfoFNdb;)%>_wlHmD%V%K1#`<ySXE0EwMlNUcWRnyrs6ca
zF%F)JBDpDARAfMb7x`7l3KmRdfb_5ZCG%?}$8MIJeURJURG7Trmf%$^p83w-YRpi?
z_;mnDx5W!Eis7`wo#l38k3JXlkvmdf%9h<MH1eB7kmb-AA@;Cjrq$gUlZQW1LGJSR
zOb1g-g3FV;y;Yu&1Sxn0zsDKD(iYf7sdq|*-^LE>f9$c+88j;$BhzfOb(v<F9;7Ou
zPA#?Y89m1;ZYw~CGcwa=KNK%8^n6yYy$V+bbwmULqm~s!8p-UmBIZ(w>cuiAO^ohg
zoE{_3R|8b*ie;c8hhNshM-{RPM4KBsuq0E164>+NPuyjJ=Fd+gweQG%rWh<p`@<^I
zF%vXB03+G*LjTf0Iz@7@7>tiz<RfRqGs>so+*vP+jaIN(%aT~DqILihOJkY<RCuxw
z&tDcz_cDtU5||~+;}?Jn^zwwzz<&X)$chl_^{4`bpsAIyYNbGCRz;T+bD;d{=n_W|
zOFq%Wx+cbqG_cI%OO<(Tdd8xWmd<sZKf^hJ%Z9m0tPegOgL@DZ#TMBR!_;I4yx;R(
zTB9}w`Rm}(zg#vsk$b5cAae2>mCeCd#qcFUY^wauKj7LceD%Mq%DsuJG!P%Mjh+fC
z)xHYHSh>$f#-o;LgmKhxw%D#)Vit7L$A-;Hcdz*qxklbBx!>0Df<ArZ_o*pbQ7vP!
zN&(rL2EkkqEYQFROt^iK2w2|-{L8?Omk<V0ZS=ty1Zr6$$0!0kAF_2sTmV?-;a;;R
z=M#FSe2<j6fAHV=jN4l=H>fNuo3|3m?jykz^@!Jb)Tak}-;Xj95c;U7;m^R%c&5sQ
zJQj=T$u4`sqaRPd%rmAT1`eK`Ay35i&mhMm4np{PKWUUZ^q@ZSlnIe7x8ctBbTGtu
za0OT;&!i@;q_uj-DQTYdq1ly|9J*Zo6zqZA0k|*}<j?-n95}iXk0)C9m((P|1u_mW
zdiZB9kmvjt!<XQ`OM9gzp82a2(h&AwHU_-#-y7t4?}z!v=sW<FN?wSWRKsNsC2lL#
zuv;`)UNmAL+YxD@@YzeY5cIull9%H^ggyel8AK4_=JiUYgOa$EAPc|7hF}lK-+j!c
zC7LghMLoZo=8n)=*REcRF~OGzNyGmU>tDnm+*qS55>Nps!KL{1c+`pj)5+?h`9`81
zt-eH0tnkSUdDDansJsBG-doYTo-F2#_qIQyT0z%9Bu$ccY_ffT{7lC%0UY26de=KK
zEryFjH5f>uycZ(|JSFdk7#p95#qxt_9s7%~6Y>8rp7DBtRYpXIeB{q)9c9W@WE7Ks
zRc_r6`V2c!Ovh67AVY_VL(UUkiXZ!zWH!13(Run4f5!d@=EDU@J`LFm6^(b&&(a!M
zrSdsctTl-I-^Tgy1sous$6KnP8fDEE^F^<I8Onu#L-|h-Vt5wyHIjho(Z97+I`pG8
zxS4%vgepT2H}nh{?RcY_7tOxCoj;=z1NZ|AG)K0#XUs8MM_(%5_;(m>xa9xv&(n=C
zUA{6}0njt{KSOpL?NDNCyZQ|8G};^pTUkSO=h5bK$cirWWtY*4h61}BAYUEr4t0h0
zTI40eCjMHC*YI(i8wk$zXCxEhyR~pz-OZnwy#ZP;&X?W&8SxY#%Y@eC>oG9oan1pF
z+r!2@3@f##zYa;KiZf~o%U=EruDjUV>~OK{?PRda1de&XfDx)k8V=bfy1;E6`j31g
z+T|^eY6GPHH{%)P41xi0^lXxEdAqz^;GnwyV!f#L1oiD3V*x4H7d$8A+x84~5Kzsy
z6@??AMSREKbCUs+J;`^C{&03I9wOiK7c4KKx^c?)V^XShfu_p-t05#choSty8YnI;
zs;-v(MmwmC`NC$XpAed!CO=Gl!&MFpxB#s5BY(!ToY*EjSY>}}4t4~b1_u~Fz?*<s
zD}J0_C$<b_0JpCLy~B_RG=*w$C^^W!1=<aPDF=^M02udq&+KzZG^Xj$yI@0|Cx-?t
zE&!0$$LCE^&iHudQQ@73{K6V~hsDc~i}I87M}D##9#e&0!;Q&;phx&KQ^qjb`xeQO
z{tVZ8jM`FP9Ob?8u$ZaGq-n4;VIdQ+^tE|(ch=_A_$>WKJJ^lGEt^rQ^^ZqY`3w?L
zwWD;?(WAFoa!Hgnt<TAT*iGYs#i6!YL5>-1qK^Kb6q6c+3JFV37jtqZ3^p*9*AIRg
z4AsY523d;+{c2v4V~r$sqdhK7l~?Yr`=$fVKRyP5i!6$VVxOJh8-C!ResW?gEH<_~
zj6;qx<<3biB)TB+rpw9k9*=*>WqJ|Mg(vMM`DqLmXAMz7Dm$h>{mduJr3HcmQlkcI
zPb>x&m@$AwK4#l>MBdlPU>hN-@D-%kf4B0F%Ih-3pV2od(jW}=ae$1_HhTS6hS}S(
zM>W`OxVJ#8hfdDWHe`giV8{<gP2Gf-h?2xEblCOFFAeLM?;03JMj9&=xjPPJ^(dP{
zcPC10ltyRwtumJqSh~xCpuKG1>T)9G{);B_HOUScz`*+oYZJG+4yoLxP4g5U?a!D*
zlZp^gy{a@Rb1$@fHNd>3+JDi*8cW8oXcj(QE@O;k=t~KFj~<n$SkLE@$;X;_-Eun=
zu-mbn$A!3U%TMNQ^VB2}c>DOdlg{k06i8NfhKU6hfwZUf8U=TLPx8XkeKxF~$aLr|
zd3;7<N(P3W>2H83;LS<n#5y@^^wz_uOmII@*wGlGGK}gzSeGUvMx{ZS#B=O^IXm_-
z<Ge8e?2vPO(3aSX8V95-gpxsgsH0(rRwd$5YGeF*xbz->osSNT1R10DKgswrT<?(v
zel|AC3dx47scg-Ur;0E>nlz^w52Bw5Qr8D*k390o1Ej&WK5cIW$%SMAcFe1d=KI9@
z@k)yOhVs!unUKcBuqso@Ct5S3+u>F!jWL*}9CNx9q$y}jyD-*_Au7!=3(O3`)^HBq
zBzfZy<a0w}KK~uG2Pi__a;BVXzvGx}mDbUY2U$=Q6>ZpW3icJe+A?%A1n7qHxi&VW
zAA@6Ydm0Mr7hovN^ExSp0^osnaGEnG`!l^}ac|>XoMMt0^i}h8NAPSr(z(DwI5mAj
zF96MBCg<5AA9vgUIo}2kr-iIsV2=@EA{WL&^`M#*4lsIM<P1a(BY4ex--e8vd&kB8
zrMq&1!^4DjiOm?9i1HKVQtP*j4t<x|4~(13%jNz(g%nGK*m+C{_lj7+3~b7j_E#Fu
z(9(dvuIlw6V`6CAqxWf-TpcnF$0T9PP4t1o@ZI7s*Non(sV~tbgbRfRfl`PX-FR!Y
z6}sn8%fHt8gcD5`R^eYban*SrZ@^zpei@`g4?L7na)R0&^!)3Lz74s0x!$JC;!6d&
z!N1_h7ErihWKbw!PrK3Ex6zO!)8!_g4MPMN_i8>1w4$1G-LGOy>Qlf)>5^Z^m_j&E
zGsKwY$j#A%(nA+jZV48mg9oxq-0GsKQ}5ijd0VD1>6p+K{k?AYws^<K`2j6+q1+Ms
z8mYqa@|$RRWEQm?I^<5H2(To%Ynx_yrDC2UWKWa3)8LuOtyg_zxyJ`aPan7`rm4_Y
zVc>pebYC$UxW=^jYfM8;PNw_&mO*%~o)0r(6u@eLd_b<kGZ40ji)H3$hm)atXOMyy
zaUmI*6)ahnhX+7&jm!>-o1IbzTN5|`9Ph?dVrF(&AYJ~9W)5&1tvxc=o<T@J(y)9n
zzMVIE>p2kl|9{tjlKHVN*huYAn->^+wzn~H#6qWuo0V`PG2blR#uo)W;RQQL7WXt#
zSA+1|C28U)$$|~ZfYhZyMLJEidXeU3se=_DIys6-kJsb4ULIO!7OA9~WrerMS5V3v
zTWg7|j9v@bmI4aHt+G#`l&UJb+II``bdo7FuV-)MZ%xPs{e?%U;!%pU(MfD=OmQ4I
zR@S988V`9(_vDPM_d!`$r7F9!AqF*0^Cii~I2N|t2w8q3_BYbR$tL3-CQ9Osv_m$B
zAS{J?@w>$KOVrk^)!Q+Pg>tX=7{@+6=<oAqvv65pgf2bxwj@cho?i2iYrN7@_oHoa
zo^+?VCr1Gs+{@JO1K6nC@4w^zMMqaiRH{*z0i)1LcJlktTdlk?&#(q0na@o%*lDW|
zT`^yD$=Y*T?nWMnVd8#3=Vkaw$%FokLt}t8fs=>gJ%}&Xh=qzG52xX3_yG1RgLH5k
z_j}|YyjMCFvRutdsOx67cqEpi5_%k*rXEdiT4*6X$W5{TV=Q-Nu6d$7mPSF>S6%=%
z$m8}5+Z2_t9KCoV+8+!bG2|<HG6<uWMaKF9sUvef70tJ1;p+Hwe3os=(VOxa?~X&K
zw<I$|JR9A?iUq(!rNW>5J@PRP=c}M>{@I@)=m_g?X7;}%4q6EXNADN-g<)vVjaJa`
z<OAiezSEIXuTh?l;bpSzQ{)AwcQ4B(7T)~3p<~J4V)%uLsEwYL7i0Kv58;|&9pt6h
z@6E{l2IoOIYjw!W(M@GlMX|hMX8=Nx<nO`%#fTPr)tfVTnde-WycW+eOG46%z&`&7
zW@9`8mK{NLpdI(bKcl%OglE3)uhFHQM<p{jsQN2!M5DMTFoPX#3-V?Rv8E2k{9FEp
zBZk`e+wlxKBC-EX5Ek!vAEsQ8U$|}P{rheRvI)q*S&;V<S+eag*rKF50zBW3KFjly
zYUG2s+~N?1W(PU^Vd@c?Qn`143Vh_xEasbc5P*(x1|eA;mNkn)W*_@&VSZhL^X3!N
zz^b7<$)`SW?rv%ZJ<$Z8r7u|_VNN&GmHs<Ag{H!*KJt#p=l+b61@$g)_Ro+njQ)l8
zmKL}^+8-DZrt;r-Th~!Hv3zMSFmq)=q{`t2lp+HlvR#$(>!An9_EieQXn+We6$RO$
zN)bO`5IiuyQspS$WC~tB*|AD7z)^@Vm7VP0Z8Ui7TxDpbt*KpxckySMOWGyESF5(}
zf+M)Q6NV95%fDtL!>XV#VWaG7O(NKGoa`23gK<hD1$M6r4@m=PnH$RF%-5?FwCwIY
z{mc_&5APPMgDV|z?xnJ4l>wrp%~B{mh=IV8YlJhjutl<$|4v91cQ<@9jqJ>GOLeim
zjVv(z(R*<p9|$)kuQfmgeIVaRQ&!y<>%UCCS*2iWJ%U8PRpnYwiCR7o4u}1_ugw`6
zT|vHWE7LeZzEc%WlPDM80){QfcWn?1>cdKXuV={a13;TP_Z-F5TKRsg`H*Zas*n93
zwb#*xEbmj`owT2~079tRL^idW>IRzs(E97{JS_{2@<;wQRW9CCsP<D>65oR<O7@SH
zt#9n8%NOK;SS{){SOyfTw5=dNb|MjXpbZRFR^>n&ZRDCf$hxk|)^n>K9Bfz&Qq<%S
z2XaX8Mo|X$PcY7*-X;)l^kycoK8X>i7E-HSDTk$5z<z3r=Av9(H03AJa4~E=P)OwP
zs;!pQ*$*Ez>D0*)!Eu9;<oC$bIg1(4rDnYx<sJD28^VjXC}vjDJB;+T85RMvN2SMp
z)?6jFQbGFLPk5bP*GKzbOrf<|26$KIIH{IUImT$0ubrSK`hjUQt5n7?%wbSG1Ic-R
zS_F*XSYIFj@#dVR4kd&Yj|*M_49FuG66E;wGN))4W^Rx<azaoH<&fYt-61DNUy8R=
zX;~{L`7?Y7JkXSqlY>Hv6jEKGwn`sI9wZCpr>PxQm_cQ4_{06opW!`#7rH~GC>RgM
zMBq*`*xn&LfMh@xXIPxxfy8WWmm$$MbOUAx-2`QNs1X|8={QAXnEiE_{!fPc`|5ED
z`GJ(}G9orJt}wtS>R<vaNQu2z4u5owP9pn`Or5Rb?!={WmW+zdz|yr+8Vp|Bn716C
zW%m1!p<_#mhf2ALh%Gmw+X`=u1S45kM!eFW@dzhcvmVFFXcs%p?NSv#W1cKc$TUIK
zRSvCW7<$4yp&aU%VC(+<kCju>9|H!-SZ~9kKuseU7gUAD3a4>7)!)HpS51nUrhgt{
zK<y_AKkE8+$!Q^=>hhqVoF0E7X8=O$&xrnjyK+wfzm+qCFU|H@hwH*wsd;tu<0aWe
z@!2sLNJFH70OPNcbDZTU`ZZ(^;J0CZ3O7rQ(HZ&Nf!F!Fps60>>h2AnCf&x6cXdXc
ztk!bKq?avHSRqoJBFqi}uV=~HU|5c}xviebY&v9oke8n4^|s!a=n4pr2A`FI61)mg
z^PvKp>zRRW7y7sKRxl={SzCEJ0yYzal*nAq9OSG#)|dv&+dch0SK=35T&Q88#)ch4
zL&i#T46}d#fs*%unLj&MMHA`4gS^E%fhP>4i5<)5hU~1%1A44wN*ZPcZVh6<M8e{k
ztvCb={yDjtH;_wUFlh76^fUs~sKbV8i|vU%z19(j?OK`S^Wt16Al9m7ioSF<q{%T%
zoJV$j+aObd7MOX23eXFPY)81K!@to??4Z?RXoXX&%szur>Djp$z%OmNL|GUNZM~h>
zvs(HVv1E<2Q+<A!obSI=NYbW7wWGJ|{008IrL@s>;5a@AQJG$SVaVQ!7N|(rL>Ku7
zhGT_qghAL0TDv&(tI=(8N&GuMRW40rv15@+`(^%2ktYnB85%E71EO_Dt*qEU%yh$r
zTPRl;8w1UQ{wP;Q3wk|OrCP2EnSlr%_;SkCiKDE76@{9e57)%xP~017-dp8be>I|@
z7*b)Tp<h@>-JKQSCY5A}b2j-`)Knwa**`043v#`;L5dTkPo{pk!N}3?I21FuF_8lz
z%xQg-Po3^}Oe=#VXL?_z@s#<-e-$!(5?ntJ7Fm#Ar!Q;)G`T43GrVokDsgk{tQtVw
zw^;YtW<d0}Cc2sT@+wgNwwOGwj9@RlS@4`(BDaU&Y|iHj9dbvUEbvBcl;3z$)E@L?
z0PKBdEc~!axhuwidCT4LG7ffL_UV;=PmBes9@(1tKz|$WLEWT}6pN~M24O)AY+(Jb
zZGvgh0)0I|au!vc9&6hQ4X+8bnh|UPD{4neR%Y4@T31$W<+FOV0w*c8{>`?RhQW0-
z%PO=o$6pGY&T8D5yZjj$6cDu<+2=ZM3?DA@{O36^BC}!)HTL=Tj8znHF##Zsi_ZdI
zOg2LuZZX|j)wz{Mv`b}SuWoompg52$a+;_o0Tz9O@-{WB#s5od^fIH~2!vbRrk1#P
zu`*F9!ki`tMdIzPr9C{MS`0lOsQNCIWwyk~M4FN1w#3K}kC?|5#w>{VfUjIeR@!e!
zEsq<;s^A5rd6U)NHhC<ZHhCf5Bx|f~5G5~beb#Esz}SF-%Q}BX><r#Kur1G)^|n&b
zbwEz0*4W_Z-E25+&6bV+31uHF*jiy_QyliNwt-7Hlg+_3wSdN2aeMq-8l(o5g6J#<
zgk<A#Zx3mdZ5di(IjHZ8Sq_Jmu_dh<r4L2xQt9?*OdicE718ATV=xGp$+Lh+L4I#!
zXselABU^noU<)+xfG;je8!E#yAB`aj>P{#RS_f1#Atp`Vd&oF)DB!e*edxNzI1bF*
zj|s>h(rA&T9xIKcNAiew=4t3yG6-X)R?E9=mptmfo2!IcaKMR*YVj@kqjh{VG88>#
z|D2eaAdef7;NlEZ-xD!;_$D;wrgAqt8Let^KTPy~;=Dc17*ADgRT~gXanS5U^CI1F
zX&7AMQh;X$#vxBzx3%ziea0tfZkd$P^W|A5M<5HV2+vKpK>p-!&&CCuZeyDPU;WuP
z4lnxHQZ91n7r`q7QEoV9%U}FUtI^a2IAFd!7p+0i(f@Y_tQ3Du?;$@Pg&&-To)2+5
zUL!|d2u6eQz(^(dkfSR);{MG)*2^gN0YYHp#YDHBWCvIi=DN|uG%tnl9$A6l)R$8W
zaI80(FR#QiTqBslU@c79e-B+Amm9d|Ti$$m)!ziMR7s>C-)pgh8uCa?i!7l3=vmuF
z)Gz@eA^)_suZODEEUzc((~;1yXWrXq$QwSDZSa8m6SOUFdiUn}NoHLyZ`o4!MY@8w
zV;)eiwFo5w1m5v&1lfs`=iP0(W<fhD>jaJ8i^C${T1(H5_v7fzHtT_rkq>&!9Ts54
z&px!TU@KUH(1IoMQP85A>8Oy4`(HjI+E(bOPh0F9uqe-YTtehy8y-DqCT8UmA0BxT
z22f(^Sj45cSbv%pnaBq{GH8UR+rQ?Ol+S{LCc&F}w*1@2idUc>a!)Ll&;1!o0MXf-
z0q9?Nv%QK*;ul`W!PM|n`%n628W`As(*OE1UJL>5(6>syj4k4M5>pmoP}g!!a92vT
zA!fFP=>39I9kP%XUnDS-?W)7GOed*|6b+0^W&89C4UB|{kU$Ht?t0lF+5{rXLWJ<I
z!|z&8LQ%G_RBzphE*YI5=?vL1-Ubbr+?F|yDNmxb&ra0_Il_fYr4K{lcD5F3nmSNZ
zOLnPNu(B8hOXREZHY6k=GCHe$&7aM|naZVR(a&A06|Jkr;jmi_KoP87Hv%jYw0_y$
zKexkAHA4Bie-nXy$ng(B>^K+hZ+rMR5ZCavutKwYMn5!(gCG{ulZ;5RSGA&>NQ(fV
zYj3CS|HapPz<F66eY}(2d+&)cu_YD+1WOXSC?K*+2hFm(z{;|V``-muV`529Y^kQB
zbfkmylIR-Jd+)vX-gCddGc#w`-1px1{Yc>XKhHUH=FFKheS}x-vas*`=1DP(6Ru7Q
zGTaa~=E=EQC45T_B&6jjbqSLDrGRX*r}}gQt4}!b8hcv4fZ+xd72)LRu}ySNg3Udn
zDQv{c`x>n!E9jZ|q`ib^0Q9q(iv4*iLHfEqo}DwCO{1jeIKmg4aIQVKu0Tmc3e8hE
zyjSqYN@dS$3Y#bg7H6jtf}dZ}ahEhO$W#G1s&?{%rZ8=j;zy2XkDbsI#@FRkl96==
zZfP%c5}8OGx}BK&KxR^wUU8jH!CPzWMZSuG)hZ^}uove9i@FhOSk_5R$KAHwMl&wb
z#!k+E6I&{O&?)|XOQ7-8K%-Q-yX>?ij0i_kD%DH;4JIVd(q0-%b+0RXRj6L^)2HW&
zL(h^QM~Vie5|Px;NK6kKIl|6N0>C~b4TgAaJ1b$t>LSDHusW2amgqNKLJ!Wan^Q9r
z7-O&GH3I*(*zougU8k@yrNYacx5h@qyehO)l6euRMpnAoj%ZLSH`+N4>_zPj1jJGC
z0=FmY)aAazMkjEnVA3QM5#oIA+}xTm7#fz^dAZdQB;O&c{rNc}J{28G2-+^lEr&_V
z6s=%d$xSk>DV^?RIqvfQ#+9A+@_Ze7kflHY=rZR*tX%JuL`=#jGO2^Ami;!yse4A>
zA{(3YBP8O}k>%0G)o0?o#iwZFog|4u9Agvw^O3{Ow285J$uTPbBR96TdXG)2=W#m%
z--Ssui%Ea8$#vIcsTZ<T?+ynD-%^UqF7nA%Eb8dCiyi;C)KIpk<aQS8tS?o^>rR^*
zFcVjWp@XhI+||?KnOL0ynADW_l0;UVV0l|)4fAM(JvP1RxEFBgrd>ihyI~fWIvA98
zBGYR|A}B&_)Dg2OS4f%2<<?x^P;WrU1%YU0Q;MIm5fl>~G^S!*X{?YBo$Cx1ILp7+
zA!}6?MxEr@x$Vszi?C75i79md$tS9eg1Nc6A?87saLLZ&CYt9!s^%%q-1#wrbU^IQ
zsQ(*mfunxdNYXkNCNF6%EdjA>`j1qzqC#R_q4z3O&gm;FHMrSaBA(03EwP?AllB>j
zpSCD5Mg_=TYfS>9LwH+^BZHA*4|M}Mmv46*z|s|AXN&!#vqnRTOJWL@Z>0%S$v9c+
zuhZB>b<nLN7t};?ux_~J70HR_lIOWBR^+Acy#8WmJkwDZciKcq#qvB)fZfwAr+!z!
zOYm>8>f)j0*phE#-OkBo5D7zbp{2H>QvO}=C>?5r@5|b$Q?T2vjQQ9uesK=elYAqK
z#mJ`C$SIM`#9ig;_Kd&YM5HpBeSJ7lG2pC>v5Qd8%Zj^N<u8+%+hJG5?;DT;G`Dip
z`V(V|`W8`lmM*_dVWFo4q&gqDh5nj~Ca;yKREbw5nkcu@-@SNo!>)C{aptHI_Ufi^
zV3A?FM5YuU=QaMuA}rn4Ip}@lEOpuSv086eDh3W;;@oey8!AEvdnN04SuQYK+Kst@
zQcaeV`PxJrHWw*@)#GNb%frK|&aOv(K0UJe^>vZn;rRfZ>J3g{%i8-f^|aX=>npqA
z=IddKhwV+dR1pC7jaDcZhVtg7YO4(^YY~}HL)ElR-mGn0E^o;*ze>GV?5#0==@$&k
zRQr*e9A~2_ntZeWwR=gA-BKaZWmp`jcI*9~iTF0yZ8;*UkSFgPu(t&wiVL+6ih`aG
z`%xq)<n6f|>;V0z%Z200?>iFPYN?_mx_8zoJqUR-2zl4wGx#84BE0e4xz4ejR9v##
zV}^}f<wQ$zG$d6j-s3&_VtXG}dV8<)5Y-2QL%uIYE+q{C>>%iQe?0S6cxn{2**;L;
zQ-Mke+Ux9tl`f<#BU>Skx(NHB`nEd$N}}FqtD8!&!i1T2NBml@9JYK72Io8LSn|(A
ziZ^K5VQVV6y5>5_CyOd+Yg6)>iVKHYhU=1eB7lthM_0l6#1FT&@)=~w;FjYW+EAyK
zo`bbZCj}JmvyGL!CCcmWlykssGDpZ@+~mYVkeY3FtRh7y1LZE~5m@6oY>T5G(<Vo@
z)3(OqwTX(z!4YNK0s@71%3iypXlLrbwLQ6Q6Lai{PuGWbz5ze@&I+RYwdDIRY9u=W
zrfgUKUd+vUrs()uNZNM$uAFzi?TI%xFuH7an>YlD|4`5^z}nqOsNE}CD1gv!djlNV
zaFN3lMQr<WI_OW<T{W*CYBJlOlN4#B5HiH1%Uu?&wRRvm-~{Cp5j$rG^UAxT9mDXU
zoD+N1b$p{;IGp@-Y5$7u-lb#@6HjGFVkRwFam}<I1KUby<q41;^^x!{6NbYVs%U)7
z=klx0KJ0{rjnFTmkHp+LqU!71;~Pzd@rK=7fA`X>I(q{?hUZ(ZIoa-ucXT9l;lUy7
zqrS(}m$p#k;CLQZEQp8<9}CJ*`A!va)1QS7?c@GpSD~wheWDIrvNwy5@ssh43@3<P
ztm%6DR9<x*&GzX)QAs`qMO3<D%XXBryq(2;CO5MP)uktM#8|j-<FiSe!_Ge2K9@V^
z;R|ry&sVCJ&A~Ziv8&k^94pfiFiOMPF9x7)DaFKu+xtuLOxV^{kEOxB9M5#|E89s$
zHNFxDkK6zb-dAG>xg5E9zm_oJpzJI={PiSak`YO<Y_)GBJ|<1JZ-%t#<`d47vuYIT
zEXq~<KOEWp_N~M%UPPo(VaUE+pKax6D^g~?eaE+{ucy%`o7{JEx^c8U$*m?tW$|MB
zz9WT{qW0yB6b?9u0Qjn${k!dl{&@?ku>HvYK5fFBX8W-(8|i`?O7@eAcsJQ701(Lp
zzxhvd`K8I~l1}^C5R7IP*UyJMlX^k!h5aJW7<M_jREKNsmnmkIv<{iT$q4*ad@lc}
z%tmA?*sqh4N|^;1hHPwws*0*+zX^!7+0uzr!dEivI{WS5F$N3iA(sP6d$Q4fm%zfK
z1guEs_c2;XWmpkz<ZUH^{-I)~6f+H7gjfsvV~iukM^$enN&Kn4Js2sqK@L0n^ALFD
z?^U#x^zgm*7vBj2-_%#{uYN+PugbRk&Hr%<&F>D_-*fXPO_(-ehW#U!Dd9>3%7I*5
z!(~Z#+dq?%i{C=`*}rltq3ym-ILE(zW)cXx33vKWh28z5Zq>xdWdDs{2uV~V7b}&o
zTa{9+xtPpm5~B8i=3;RhSWO(>1DhixP}Wy217+7ds5vY+EsNQ69J_7y;N}o>ajTTN
zsX*DVhs4Levyc)HJT#su&r8t?cW_TVtU1k)N*xbg173I7!xNPDS@wwLVpvoMxSr+q
z$mXzEsRL@s@_AHq*ou1lmUnaJ?9nl7*&K!)?1J)`X19y#&Xosn8@E0{ipY@b*4bm5
zkGm0Qc{_a{+w5`8MSP)S=;IT75XhYZn^<E{$a#w1ti`7K^~A)KDkYa@k@ln*fD>7S
zG#L<{++1vfFeQ%AQ*s$nWmTr+sWE)oWRwFWs@rZ)tILqINZ|#HqP`?&wNH2Kkwn~V
z&&W~A<qO_Z(z0hJUb*UsNX+*Ji)JvJWU*%_V9^@Up5wzW?P*_Z&&|)u@ZO2bb8nuP
zxTG@)YNb6tfKJ<&vcHG2gBK)*sh@Z~i#m)@dV(WpVs|S>r5E}?<k($mC&s*4lYyKY
zc16XnUlikbWSmQTFOFvtRD~*1gzcpIOx1owB^&MJgaJzjRvzx_U3QA232Xmh+-ave
zMO?AmPD^mcw~(crI6{3QZmq{&8t^HcOP!_F+{9J<(<{alXC$U0pA#(5&Zyt0e|(`x
zzLdxL%z#VRlk!;k)6NRGD6xjjF)S%=%H%O)Cr_E&JlD?lZMye%gxcZp_buoZBOGvO
zjL_eaF|r3!cqxN(VyBeQPl+vLW8uf`lC-b0QSnXjlS`ffn$bBwdcJBAaF?B%TO_%J
zn{m)?b>0v(OF>mk-T4VfaW6=GJZu-#_Xq>#J*5J^%%35`EvH#*gD}A!dwHNj7Ww`5
zilpnYYJ4}w$?!6Ag*OlExy;7q_shvQQ=2Uy9p{^R^656d&Zu&QYzso;gm@-<fb69#
z*e2%MIQp0s_$BS@lj0XP9C5{sN*;!Mp~E$`1$%OrP0sg<X6>kUCUlYS)xhdkoM0E%
zw+73SJ4s9ls9f%MC(=}(fzUcx$EPJtVoRj9W5X_q!Cc}Z37CK}6|^~O)0>ZbM*GTk
zHT}6X##S9AtU0CRz>NB_EOG8j^l<m4_{E>ep1ivd#bIl%m{Xpr6oWN3vp!RDj$#Im
z*{r%sl~__do3wl7zRb?Oys&S{Qk#>!A9<;-#JJDR-)Sk=Z}aj<4m*3e&96+-d-`~@
zhnoFk3z9!65rY%E&@aN_BS+h14&%V;@$Kq$SE_Pc?mv$kGtOR_s0go?!V`Q`T4GL_
zL!xF2zu<Tdi}9k0Du+WBe-F2Bb!eJaw71Z@Z;7=f2&#9j8X7>@UPqULtpg>4yn-5w
zrN<WMWsH|pnN@e=DZSH{#8*T7gM2ku*Li8Krf_g^B866|^jC*d#d2D3NUpgecFxr*
zFE_KSE~v<Gd>oc_>r5)X<|JDlUqYn-=BVmQ>KsR1VR)u>`&I(mAv~BX;uCRfDJV+t
zxH6uJ7b(~sN3ZqdFlvm6_Gk@fMA(99>|A=#Td#wnv~TOnF^c#m0^VRN>obvL>5*cq
zeC^}3a^$S5a+)ssNVTIv^vA+=MJi;Sx^(=h9&A@95zb$Dja`!zOZ0^@ARG3oSmp+-
zE(h(}_}4}T*kZ2^K-7UzmMyj7d`&!qN5W6|6K;N8GFxsN<v|gad42A3fk%8?HU&Jo
zVq*8*Xg5@N6#ypzP(c+c>2hOTJvAV}zI3kFIzv45T6<ki558EZY!&wU{6f1TT0)e5
zLp;-*^7|po+Zz)TGn>ZSn-b<RRLGE$zr^0``!cSDP`9__yW?g|u(t-V*~!(LKIkI6
zsZJx0h$h=*HzznUbe5xV$Zo0cZ7CPIY>>C+@xaQvtewV;#nydWeG^6)hw*Ls3?6<L
z`=DAOZ_lrF8t{$EveU44G>2OfqmQddtm~b*mN8Q<vUkN;5!Hrwa;v?2@YEE`|6pGC
z?fL2OQD@qF@*iUwSn%owYL$o*5|WB=@IHU7jSkmK?fuCR&l-L9DEmNsM1TWMQAM$+
z>KW<$_Q5*b<`Uo_8Ms6%*@uFXoD&pPj<crR2j^6)V?l+kpd?lDxFbiCXCA5Udc;y>
zL425ZCInEN9BrbzA|=|IM4gvyc7xqo=du&VQLHdG0hLjK&+0j_KAuUclf6zIjyA+I
zb$iyeflJRe#;n=OR0)=jtWEKYqB7D;t<jF!<~-t4TdpdZk$2_UB|yHn*lD)pRKCoC
zyOCSA)}`0*3mrM6mu<c;4f&;Pd#+Yi33Wb^b7Dt5v|{3FEgiAPEKMN}JM&s@>g~r5
zwJRZPEQM!6XuEy0(KA)uU{65D<)c_BPQy04+m9G&2u-%ve;-&)tu&R_l(A%A4mp-`
zBXkzF{V|A+mTnhGng^1mmbSNd+QCX6va88-B++c}7CKasq)_Me5d021oQJ>BPLc->
zkv2OLuaTS1H70p)s~vT=bKy)omROZBn?VjCBzxxThx3<WNhq=j(@V$wNM7h<@u^AV
zJ@ICQ+(3HDc2bAxdlRoyJFuPDK(22Q>Z9G~|B`OPCVn&q5smCAR45PWF*g2qyhnI9
zZ2Dt$ITBS%<d1#4ej9ZOil``1Kal_+#8Ou6lNIu=qE&w?KJ+8#0i;f(pRUN=$sQ}5
zbenx9_wb5#th6ZQpY_p^8Bo(3_PO{*IaIpZ#o#}m;DQyoGl+4qFXTAHTi_S$v@a&j
zQkbE&r-L4ETr^+u6WZKzm3`SUK=_0_^j}FVPM9=thJ7^wW&fqN68~ji3(C=SvdeB(
zJ=7s0!`BlHN<!nzd?OE_O12b-@;4JiSojj1fN#Y#8!(=Di3==pfxWE0;;}E1J-?Gc
zIUB+UK^v3$Orb@^Nn_uufT}Qtf^iD#DtCeMFTNk&D34n<tF;I$KS*Rz>>igN(g=#h
z51WsBR`}BH22zd!F#Y8J$g#MJ@)tj@Okfk|Nq4e`!ie80gz_i;>X^2b_S2lBO*^Z~
zhKTt;OPpNNvaDDFf9@X)KYN7zBKM>u&HgedRt3fP*st<-xumDVejPup$2fh|elvK)
zx%3O$XS)W?=)d(DPCn@j`(2=d4^^!NF1^iu@6QHS<7+~EqF4Yiees^3BN+n^*dG(R
zrLqtFDYmWf>9Te&I<eXQTsKtJU);vv-25+zkPgKTsu>&<`qx+sH$T$B-|A88!WCwJ
zkEOY7$VFKl3@uYpO8*ErBSZtKXSe+`xfl8-K1tP5`&WHU-cg;y-S+RgfW1mF=OR*Z
z0ojfJ(|p{6B|KTHQo(-{0n_D`vzcykL-s?2;_Rb)K47L>$Q&u|I&lZz1LN)DJO%M`
zp+0D4sBu^HEvF{&gJ*`l4pmp?TejgLGmDAuY=td6G)9nOASze0fQQX22C(tG+zkJ)
zhtCY#+iZ3Benfl`S}k|fL3`xPFsTg&t1(vb)FWK)vqx1pc?zafzwOa8k3We}!4_D$
za&U+8fXBpevMR_V$PhxspoOji?Z!>@vV7dkFbR%rU)r)#$tI6aura%?w<pXjCeBgk
zjI<~E4~m86cs^-nNN-Rnh5ZPkf_$FruZZA5=~En~-ROa>6ry;l<8NT~<T>`Vna3Ye
zvZ05aeU?4l=Q!cobL|<iXWBu}VnT+*ptQlB8NY-qNGPBDx)RTy70WIWFz{nlb;yID
z(qzx}c?k~gvgg#ry$1p7xr0!QvgakTVTW#5*=^5{Nu^Gdia)TiLP++4#3kCGhs+(b
z6Y2<ixUB$=;IJ3^+WKgHaH7wK_YD60qWV42YV4KTA-_0D5PghG;iOpC|73I(K~3`f
z$xd$rtH*TpQ$@y3@vV#+8p<^|HTTHNJbELZ(_&_Yf53pzN)Fje5-sB{nL6EG>IkFi
z=SVv}FyM6NF|E>ociI`bT5cq6ZA6hf?abJJ9MxQ|XU+8R1084NAz6uIW2ApZK|dQ7
zGq|aiccE9$EIGU4XdA?!xewUzT(l%Z$Z$k`M>hZ@aNi9!GM;%SI%?j+49<yJ<c;>r
zKU394<vLh_;FIZObRAWV*Q7`9wR2}4e;_3Lg3IkZ-^kKv-p-F#Dcne6m=s`w>*Co4
zJj2wgFBO4a=76_R=de}W<mI{9WBYnA8@(cD#4RIwP+^QQL$+@vZRCsGJ2qcL6;Z;2
zl3vEuXGCGM1lq*g_@o}VhP=~Cv~j^y29GT^Vdn9tv78J3@Bg!jxdcuF;d62p*`&Ht
z-rU<mim<`jg-+!wX+1ePPx<(1c2N+3+b`Xa>z1w(fhOYP<BM}D@nIbsYP3wr6@dud
z2h_x~sj(s%h2$hxXT~Dy>ssS9PYdwGFVrFjcMyL9q+K#Iyc`7jU?FO=={Y6hzLFu2
zAk18v&+g$uU`j<3%t$bh=*3ZqK(MBmMm|+Jg5(D&4mxX2luIOJRCQL*&;x4T*i0vZ
zF%;~^YSwPEW)=y@<-_>1C@Zt`@{#$!K|Bv70~7a1)}@d*imXtu-R8zG;x|wZ+A5T-
zycC;P;c@PzaFC(e#{5{FQp04Qkb1qqck;x`7TCgEyr<R_1I=YIq^=k_)>TzU$3x;h
zb>Zyt{DwV{rcFMpy|N<A%Yi7Z%A=iH^7292Zc%X?TjbMUOb}IT9uVZ!&SlmXBYWhn
ze4PiaJr{s=RE>ZS5X8N>LQvZI77AMuP~){jK$Mt>=d-9?t`{whZ)AUk&FeO{jv;7O
z?kZ}FWPOCcE0P6K0a)SH%PLM(Hlv&_ok8v=AF0z%#I-M>mg{m~-_n+D>l(5<-oH>C
zv?}FTcLGJ6sS+_1E4{)$A3bWcU76^)sIRkw)Z#>CX?Lp&x{+uXFJZZui=F#gZ+<$a
zuZ`%}K4)wvon|ZZoM2`zX@#o-2BA!466MpmD#qOeCF+?v%(z6^@6gR?Uu9QUj1O_o
zLPyFyKa4l`ni#m$yWyFx%2!o@tDysg;6kx$>n{qADQ~kQy4zn}$GHR7VyU2MugS}-
ziNXwaUH+><+grOn(L`3Z4q=<!5Mvj6Sjkyg!@enJ{>Hj^nRUHX)obHfIV^0-5!Uj$
zoS}(m8hgF(_9@q%yujX2;l@-7+($T?MS-(727ZchmS1GjR@9J6ZXvwcoBWDtp@9-%
ze05T)ZLvtSYkNx!Ek*4R3{ZcHXYH`JR-tC}nRZirs=|V(lEoWwvqRX88N_Z$PD4Nf
zVUX%2cxx`_tPu#hw<WPpr;Rmy=xxE$U9l11ipuc=wF8nbs*b(AF7I%$aEMJ{N+{be
zYxFyETGSgFhDsHCXHH8jX1l$s!mT<{uB*R0o|T*=CXzet_KKRF434CPi0~c<8EL1x
z6wZ5Z?n1Hq6(j}0zAx#W;N6uSNM-L2xU~u5?egBaw&Mv*ABgGLN(A!qr+v_={vxvY
zKI8|B`cx%?Zgu>#op^32?T&y$0;KYyh#Z6uO18Z-59D;Ydu>hr6VZS`#kGmdVxAr%
z@B@I(&ALILig)}N&(`NHJg&_)B#ugqvzzTV#+WW!{=YWUHsx+KEhPeSa}oi4kqC>n
zyAp4NnPS19&R<L+HYvxHq_C~dRwm7|Z4S&u?F8*?4<rPGlo2{MV@FN`DJsikXZ~A?
zd85^KRcTW8vSMB1%MuA!QqVllaJTOUxuv!zrvl&FMg^%#6kTU`=g`FEwL$WGbF~vl
z4z_(WkCzlqPdIipSFUEa^@xynu}H?jVX;4Ht60?J08o(^fU*NVCY^MYGkY))#JyMS
zly{RIawwifE~eBuJDlU8S86`Rp=pmc*^&6gd4&Wr1z})lw4=$>xG23&?6IU6-cq`V
zn76z1CAx324+nMJG;ouBB&nlV3xS<ebx%Gk;fNkLac^M7$B7<V3L4NcfbJB%B-v91
zN(XYaj+Js(-xu%`5)7MAV6dC9_S#32nv{&Ki2LISehIjd*->(hOeikbd@L7y@690d
z76iPHCqG%#-QA0ag#3?B1PqFnC3Qj4qJ1*=y$3TV*1b>VJB2os_Po_T?F2OH%+dCl
z`m>tH6qMU%<C(J&SIIx;9Jd44agTjIcMz5=kx^Z}{d)XD-NC`ca&fWE_QhCm#GxQ%
z$VI89kvi?0?Mp5w61}0EiW<t^urDW4Aq25WZ)7hi^(%=Zf_jl6A;GWa6ADT7tir!%
zU#mC8*~9GXu`p4aa(mRp;2T3AR*6`JZ~Oi9%|rv|LM}!`UHeu&MP!}Xx8qv{f|Lj~
z)UKkT+jktpv#=h2H#yyMLPYl3_k09hXQ)5l&pE;xw3So!gLp%6JBRIuzJP8C)b^u_
z#L~>2gC+}l6+i69IUMLMeN&Lgf0BzvNs<9mjlh4JIF;9=uvh@cCw<6%=K7d#CvU@6
zTI}cfdWq3+9mg-?Uj(chi(LB4iVrzdWWd#h&7bzG079CPOPV|E*YUQy2*l^*<B(qW
zo222sE_Suoe(PVW=rOd>($HP{yE>wBMO@4JeLO21K!`yH{6j@qE!18X?6t5D+aHra
zE*&@C{*)89lGeg?_UAmMc(H^b*M^kp0J!K*fAPH?H?yf}@+H&kug-J0Wqa*!d0g=-
zFYYTL0)Nj1%WJ}DCC9HY>wo0DG=Y(P^3RzOHb?#ArJej&?%IDPCDE4W?&bFH7)}xS
z>9W;X+h_l&c!0*CX~ok0H=d=hM0c{ya*LvzXITE(sZI8PSrKzrxUXM6ezqSNZ?A_V
zsxFFz^#{!gi{e}zS$lB2f`<<VSkkNQA@PiV6~$d<orn5oc-&jnSm$B0LU(PHq|$zS
z_^e`voFaSMBW7hxp5t8_=Q`{d8B6y_fA_32&m3-#s=yCr+De6d^sF%Dv9s)mTY?{x
z;G*5>xTrAmV`B*EI4ZNCc(}s-?QuDTf+=y&6%*m(1G*I0OVvS%Z5(hKli3p-I<lS*
z*%N0KJ7df8HhYpkhMCfsB%$EmQZppQ^gd-)*bbZeuC}Mv7rb+cdWSwOz$`=#=(11g
z0^ep&kGGYlCv{F2>oY2p2c)v;RP$$zd$DK6H==4)`d0SjvvO;D+o@-0&rXOYkDp-A
ziT6E@r5s>;ZoV*V_%M6kAb1GRDpe%<BB;ckf4}W4rYP}3ZV>JPY~e4s-|vzj<?bjM
z2gRFxf@ALFQ%*VEUN|cZyA&8MA@6o#d|fKRFm5T5hZ3hS<`Vo@;gUCBH-Maza=E=Y
zeo<zNc-LNxcizVBq&jAeP~#VBM*$a!J9ctz4+M^puvo%QiEliy-NP6qV@P^X`K43i
z8_5W1Mm=qo%Vd(h<SJ2Ny!9v)GAl~PS&x#NobDx#g;TCQ@pOBs|9#w;DO2q9ScTh$
zwfV9zE9c{k+%ndz6ccPmc4j=&r7F!yMaa*}vxq&Fi1J<=Hp?xX63Qa|8pP5=l|f31
zPw{V_J!E)=?-DR=!vh#CAL)i|P=yewv?D6Al{X{fAtGl%+{mOldfnqkIcHY!F9>TK
z;!#OrQ^|rD?K2LnCJU43CwbKDTz^4z*vJzOkcu(89LHUDemrvtK!&Y@c7e}<i(K;3
z%Mt+HJPmt!tbRy%BEZ1ISff%S#2W6lSHyZE3A(J*6}mCA!pGu9Q8Bc!i7F1DOj2r0
zIWDHbK#V1`M4E#tm5k7hC>@IIY8yYRm}g&hflWwEEWqG0F$XNitQ;AelnX5d%Sz1{
zyD*5-g?=Y08Md$$!{!dy<lOq6Zly}l$T@LAJOX0h#W7~q9Z;;#lstmdi!+&^g{cnY
z)yr&Je(v$Pm&E5H9tElH=JV-6^Fzv{K<$zzoy3M3(JzhfF+?IJbH&JbnRw`<Y;KeD
z*Cj3O);z2DsnAJVmo4hP%4Yf*>F(^MbBX^($Js(SHaq`Ka%V9^CfEwCYuMa+q}1AP
z{lFe<_pT|=^ZBmoXtDVfHdUijmHflDpgxmRz#A?tOmr=35n!m5m&K-tN=rnkBx<b<
z%8;q!@+2*S3CV$@fPe~R=n{S9AU+GfQ0vncr{#gwZQX#r$N_AnifL;MlSV#WYcj?h
zw6+9OzD!6DABMH(GmH&VtJb2*Ee;Zu-CkAAcZKY!Gf*ugw#1Q%sC$qEhgMr!f!x3u
zF9s;<NFc}2!`!aOEgajj%$DU(7gE;2I%ge!CieUP?~?wuJV`bxj9ZrvK)VsOmh8@T
zaP`RFg#pD@#4{mWahcQ+wk!Q=XOsAUaB%FN%`Mh5_#%R{&fzX=I2}!#IAMnM`roHh
z)zkXwE|Gl3GN~dMi(Q#J-bU~RF8Ov_<-0P4GP+kKd?jGW4aWWPOc+-BEIw_!+DWpw
zhe7H!x$3c)QdC6!RlaIi7%eL8+8`;dH)20Ue!5~MbK0wO)<R}p{^m8Y0Li#4Hv9uz
zSaw}bM%`aBC<rB-b=-^QW~mJiS-2tJQwA4_&~{^<qlqmYu;bUptn%q97E-k>UYAFY
zR$;WIwAbhMFKS=1#NH5J=mJx5r-NYPjq$9+qA`jZCA=vwTO=NZ5tl+?T&zkNRnzD<
z$2Vx!{9gXFw>WTP+bR5w;QH2-?Z%gc3}4YrzELBFjj)>&uqM@Nz9mU^*jc0P);c;A
z#ilDt;I?>%AqtT|IS3o=ZFv(YzFQH+{r2|y+BPHuOmXx9dPi_X77e43wWT=cG&Dq#
zvOtk2?47x3MLuwbUc5#2uKG(B4Uuwx5c%ufF(-Zkj<(`r_S)@%Km<Q7HX%*M_`yn)
z?}=|DCm_eM9c%5q0jIVgwi9UKO#JD|7T(p2R2X}oBfF!E#6)h0_XluNT$CVx!2d)b
z`9RXpN`>!mp?|P)M`?$ik6dgR%%XeqLv<)H!oj@*<$!I#)wMcz3}K*QchvWEb?fLW
z9Yh7%@65vzICQ(b2M<UtTvLDdVAhiE3op=`4-(h2Hs|T>CZu9reS4?^qWDdMNWlL3
z<OZ~*#Bj?BH&onG8<rJ7wze^zDdJDfC?F!+<nz$<dMW&7v;T|VHgf>(idi=nTU24|
zP?Pjs9kyMz#p$(sc?&+1tuZ({C_<kC<PO+2e>SjsMhEHAJ#A!Q+V*%)r%xvPqny_r
z`7NR@u2$kW_u9_fGb%L_`qHpn_rtYY(Te5FcF)ReFea%HwtI2_c1o>Q?e2tTdQZ^@
z6$-i6m$0JaD$F4J0t~+=H?YP7wm*+oDWQ)F-mn9K7DQ<o2k+T~tm&n^+>Uy%uEt{&
z$;Gilb#C(R`)KEf4e4;a2XPmrU^_<4Be~J#fTlbw0pFFyqq!&yb(D0cykTG^WO;HJ
zN`jdzl?i4PAki80G(S<po%WIBqnH8@PqBOQpOnpqt={XD(2u8ws{I2=2}u6wa6_+0
zaww&p?Y_j)xLSDrqmJ1r-JCc(9#G1tq@zjgpEyq+^V^P`nx*#f#6Q<14YG?T=@ZEl
zN1t_;eKIk2&qnUB!piJZJ^<E?7W;JkF0YDWBnVcv&%`rPqy|;r&(9{(imk*8<q9bW
z>~j_WHO*4@HfB9{r+wae_sDZb+ZS>xR3k&3)xKC4>3T4AZ#XWBrTdb@Nahb6NBr*?
zB(Jis#1}>I0=6ARNY_#QYQQ1Zh(&@MRid6cla*V`zE<agIkE}f?^fd5*K;e{XlO!s
zI7I%9x?Ek{MOA{&i3@$xcNwFv%2#~L7e~z$`?mi;gV+^xN%>CTP&wg#RqV0vRz~eC
zsU-UsLpb+vk>N`IR*BsIUT)%aF5~Yz0+erJKgbP*WvkTOCM;_|tV>5cWVv#X4b}92
z<WP>GE&Y#^W;@U-8}<_ix>GvdGW%&=wa4<-0wngcm{oU-mZSUY=l+Zy6Lirb>1PGy
z+<y^}s9F<vLyyJ3i7+7TN;RH;dB5$GU%fcCAh2J>3_7&rpb_}E&wd@xoSV|$6yp5N
z;0Uk)aaCeol-^^%jk&#eM2QAqZnta8Nd3D+=~&SC`&<x`*?z7w$##E;0R)yD{Gv(G
zj?5;cMcE%?YTr}I?+QndQ|wPM8scWLCvN9D`|}{6Vk3jen)a8Rhq<Gi5D*OZb^mK#
zO^Z9b-KqGuiX6#bGN~M~zvl`{1v;5DcH2J^1tq1eoKFz1I$Wiwwf!>%#uAoF{GE$e
z|8mUWpW+()x0A$#u2OI4KTdiCqPN+9b4KREB*UGv!)$hBBKLLJ17;srPVTg(Cd&B|
zCi%eGVLF>mKKz4bAAcf03a!O>A&-)4PY<5$woi6mcR6&j%N|mnshX(E5fAm@$tP@W
zCve3cHoNHH7dGtSvy1I>Hix&J;=7NSopws?AehNY9t=dC9eZT_uDzr3fi!M736&F_
z+1sN6)RL4Bz*KYO%L9<mqZ0&z;8szM=rQ?k6?{S+c<gL9k|^-=ustpop+FABda0)K
z<6~^6A?X=vlu6MRdqO}}3_{RD>2@+i^m}5ArgTN6Zc#SVo|KTvd{UlC2b9qD<oHD*
zh-3y8;(`*mFQ1a%lq*AY6LE6()P#Vy#XHrzOP8eNW*G#ZmSb|$Djr;(4d2422W)B2
zkb&<Rbqndx&D*lhJTsnQHOBghT^336S+m0~qncKvi6gf>J0^0&XtkZ;yLwJeB)-S8
z_7ku1+<e9nlFCT;VS8RI#aq=CxrFHk!vLS}6f>~;-q+oCeTTik_ibSHqIRY~p<+mU
zK$8^R4Om~8@Ut)4148@l#C%rtHJ8p_6wh?(#4c4%?Zvahv`+hYOr&jgQoN*4Y?YAQ
z3xS<nQ7q|rsJw7W9;V5Z4?H#h(KO)_J1v19i$m64l0Qx9SZps%*p&MNB_G4$clzvN
z`yPJIFgqhRoHGZQ{D~8bOC6^=O8Xi+GqFT&9TvE={L4`z&#__t3&w?w^7fsbAk31(
zWq8Hk+A2Va&UV;{x@4JDOH8qi^ijt&O}BGmf^$aMs2IcB9O`R0Wf5;iCmtuYdrjwa
z{j0HD+vmktGsfBZF#y}gR_RvJJ-!b+&INgBd#FMnJLCOu?$uV@0{ilus(C4KBiP+u
z;qY_EE5?FQD7f4hUpoctWSJWqD^yneeu!~9#5k_*0_g_;afn?Q??99mdJ;G$%ud^`
z6rxlxA#eV^Q~bNQx+l&~-w9^GCEYz8S3^gW0{-F$;&oAz%PJREBuXbMWLPeOT{byD
zlJ8Ug0E(R1MHO_)+`^RT*J~H&#**#YuqjSk&BY!&b#|Ei)w_mHacwp&2EgJ2hvGj@
z$0Z5eppS#;l&ds7-Y)rKJd@A!(s-uWubpg~3Te%#@2Rc0zt>d3>LDMsIo=R`sI#0n
z<IFnJCYdOuQrfKg-X1xbc(%u8*Jnz(DZ!pLCt*ei(H^R?iSxO+@wR$jXB^r*KSTqo
z|8L`%&G#D;pTJyO;Pg3>nDmAI*M+B?Vwd?}FTZBKUG8MjQYxyUBEGUNLq*;ZOwdAF
z>N89#AOr`TElR+g?BIkOtTmnqja$lva>Uw_*>=g5Njzb@wb$V`;??(v<i&vjzc)N4
zNv5=Yw!|;pm7TUU_n>+5)Cty+qm(Q&rGp-|E0PsWq_5Soe4TX1k=B{7BfqJsmnwcQ
zPi}**iC3X3UUB2QEA8C@m#l+|m<pE^V&RH9pcmgzdzUMNy_NRC5cCne9<sly5Dj{R
z^*BfznSBV+jTpPsmE=9ym_31`*Vi$9`Xw`(tuKiKyb}t?DakzPmC0Ae_jOUQ%2wry
zC(th0t_lEAOmr5DtEN8ykhO$fCw{mwPWF=WiV~<gS=ZF*sP(DCtveYGLN&tN%H?=L
zX(fO<b!{S{7%|-^_Ud>h4oD0DV$2?UO&(oEc=U^jU6+V0g}X#&dv*Na0M`d&a{HIx
z5h&TgHKbBX+_+)3clmZdp1_o^aIh{9u^6&fY1RBGU(byJm+}emv*_-&*CvkS(dNC)
zy!X1;M0sM6^gXH$JB)666$pKOMG#d+S)VFBx;X!aq$d($xC`Hy6C&3r<UosiQ*6K*
zq^&#b&AEXJO#~@#iJ?6xOTJa*vdfP7)&NQdFey0Izyr|gPkd9%B`J?;cDjP>=J-b3
z&wHK9Kvxi=+!?#2E<`O){05M`y2Wnwg^+Mq`t#iu0}Y+l#YIUtm^bs*BJxJ^l)XKm
z(aodUCj4pds5gnaYSA}HW}0^nf>4xa#lH%~E%q*FQBx`U`)-HE3)}bG?T$i359Bv{
zPmJKU6?xF5k1EIZy@`HyS`~{4u(J2XGtGhVL3fe8zrs)pmIDK}SNlLBu;`$k?krnM
zdSO18>n^1kxQjoO{HFz9-|D1>u|(x#hRhveXrxojb!+U-c(#EcF72v{F0Bc0+}<Ez
z0&5%Oy~ftw?-yO6V#a7jwl0rXaokds!qx|!fqUgaCtSlvjk1loDhw-f3n&*&9u8^D
znr4$<AL`m_n=5uRm6tMW3P-fNa>%y!t87cGL*+R{X@?kQYkpcNPJz&Qr)_iC5QX1O
zy!CcR)A*j2RXPGYeD({~ZgXc0oBf%Tval<LT`vTJGFP}{C)w>^b|^}DPX$OxFSu&~
z=<a-W-*phB>KPuiy)l|wmrDW!Ct6qczC4<$0J5^f_B+t<J&HWof%wkxR`?#U2)=^}
zbBUKK>WLl7nK^cF3|VUJa6W@)iRK9|wj=S38&*oa+%|SJxS@)%NClE5TJ)_l`%>77
z#BRru0LI}5QWrA&@a(X+lKv>&*vnmfB;UhisHELiyC<nZt7iO?GzGroiAIUEdvmj<
zb#oL40z%R~N(z`*+I=yC3<@~;flaP{oh1?PLHlU@&P9*3FMfIhzsr&6uGHi4yRS-r
zYVD<1fYd_!Sj;IalQeUcSpImPJQ_kTCNR-HQPJsz+>pWSlNH!<;)*`-sd~An*qN>s
z#a@<r^xBp<0-p|$N(KtR%3UI1wgiA8OMfOtE@r91IrrLU<C`Q&*#O+)3_-EEpUZDJ
za9%Wbvwc3E$!?;%Up}`l#5f{D)rOJ1QJ21Il#;n;U(5}eF=5Pj`%?a^34Ou79N!MT
z6m_XdnpY|JuOv%aqzF;_Y6VI8@3Ng8u&*U%C-q3)@3*hJ{B>o!eItlI?C6~9T|lPJ
z6W^=@WmoOP_N{oP4OOx$QjUE)MkwxWXTLxui|BW%HUi18$qFIO5+5=&yF>iD^|(rj
zkglUbhDw$BUV<$%scQazKlxKLw$UHlZ(blmAd=d$&wf}Tr?)5b%3AZIczb9a2(4m=
z{kXneiWUuQgvx$0`?$MATj=;nJ)lK2Jf|4yPZQO>u(gK$EdNOneB25@PXdF&(^l`d
zt8<56B*08tGS92<|H}%(okeRV%E^9J-`=NN6yYHH64|f)Z4B9iUB`cuprBQCw%Bj$
zD_yiR@EpqD)k#!VzL+%a_wfvSgMh)hkNq;9ZGfc~8Y2enkMT@QQMnUh0e`AAE$g@l
zO|^8&zLvA`=ZbN9bTi*JZbPSwzr<(~+9bWGXVYKfZE5Q6+Dxm^-;!xJh!y2L{5`(b
zB`5Jxu>7Mw)46cT_Mh=gwHb<?30nPE63~MA^X=d9g3K!_G^(tf{|q^7@(9W}B{Gre
zMmaA3O~Sx4FQSO_ew&lV$7peg@8JP+(p0$*Pl2dEn_WI|&T%CYs~}CAJt#h12Qrl2
zNlg%Y@SISJ@eTHrj?E9rw~$s<J!2ih*F)zN%O$<qALawnu(ExYJ$z2F9^wyLPF}_%
z{9kzTl&bc~fT<|<EU#Kt>``;v+Ns6*4x`_|JtNM`9_`}{tR``*%^s8BU=<es<PxyQ
z#@M<5l&Xe0#xJmTd)%C2-Ng^tZjaBmipnDhvLfjdVoY6%3eQkl%oB4S_@<CU8y0)g
zoMIowx7n^50QTe@5pF16KUn?oDe<)$V~8pg4*JwNak8l)jJbi^vV?s;&6l@|269V#
z?diGLnd$`cj0%`)oqE{9Gvk>K4s>$gjST%P-vuse{5a1Je7eGu!B29;Z{(zGx98+4
zs8LILtv%Oq)I~I_Jug-iVd&-B<WwM6J%3JE|5{YsfNFOya2jZCu@mAW-`e%}-N3!f
z<oDfZFAPvp0sr4v?i1@;6hNZ`b;Mre;FYRcQXpTP6gigq#fw0hos={uliSG&v21JF
z1LRv*5>JWAi%o=IHgiu7X$N4Z&T)&C>?5T+Dby!1P%N*+Qy0PQv^mAvIG)_Nmm~@r
zE0)6A8urqd!415{$HHMosVhr4Xr~8kQvDQGt>PXV*s#0oj5%R5CQBLquE^yxedpEG
z2$DXlE=zSkBsi2PE-v(8Ng<>XVYnQ`Ak(6Ua0v}h{E*RSBkC4-p;n1y@KSCpc^Eg@
z$T`LJKwArP9J=hB1X)pneyMS>s`ksMoVTeJq8Xi^&A<)XZ|4RWE??lfLSH;DmjIVl
zAk$!s@cDDX3dkL&`lChnzaU2hkY)Y$GM{ctGtux(_VUD62We^)Pau;26?Nqj2z32!
zvoSHH1St^%9Wa=9D=O&N_{G&AmnX}`VdHY>ahNUC)}!6VCn%&lA^lI76S6No(>l~$
zq8$KiV$3hW3U};%H^9RuCFZA-V|rn}P$V=7n#XK%o;~vCRv}hj6ktjC77dn46va(@
zaS|IQ>dn1id`jXU=HG5pb8y5Jw=|H^9nYj$%D9IPW|!oIEUg>2;dCeGb1zt6m)6&{
zGaKo@c$dw{1<XJ?Zwk^8GgcgdlE-wk5Ywb9ShGV!MOnGsX69kwpvilSJ_OzB@50a~
zn^l+UX*b)dQF3;o`@;S<3~h6qA5FoVVsjIO;s^Je-fa#gUfkx@-}eEQ%c-s#>b&o;
z`H4k>WO}<XQ!j``lf2Cl(5BpF3o9ZNT>}?G3%RVm$2;XXZ??<pGwDuTs}T7sbJZ{r
zwJ%dGY!_7!@R5I8k|Nnu1@cSK)J3<&7R6|a$Hlsd;Cr{V#xrq3IaXDbug%f1xSOo*
z_DaRDC?HiRq7ZS=77u<|`~*8#*OGW!>?bZE_5zvCmio>y!(nvOj{05=bSo=9Xjjx<
zs8`KyVwGhCr7JFrUy7CrB%2Y4+}PNe=c%iG2@x7954t=zn5slr;k%N^CSEefx|6JD
zkc(z30x-1@*kxDd?ok7vT>d?=O8G&=QWPN9$ot{mn~82#dRuSoi45s-8Vd8;Er~+1
z<n-l~_yl@vrQ>&C_3~a@6{`~d#c%x@P^X})@|%kZxU>FbVAN`9BYm|nu&alBtOKx}
zE3Nq0?V9){&63g=TpoQ@(l0DlK1MJ9<h(%2@U{6R>c6Y8sxGIR!ml2jjnO<b3QZ7g
z@--Dn7!WwuM=%du7teG;<RzAlbbW5r#a-R2NKry`xWT!_i@cw}TDvhZ33rv8O1KP5
z8q~F4TZdgkw45k;t-US)Q__l(?5M)SglkcEUhn&^sIXppLwuT?PqOqIoleL|#KO?9
zHzl1+>AT9_TrpLP9dX<Seco@0-+hHcZkt%`Ta%VrFpjD`(@g<DxNpj7y}2S)-ZrHf
zm8+c7dW*vXGv3><TN85>C+Tl3PPE%9ro0-yT+edM%M*#$*4~y>O3@&%hxc~Jb)nju
z*|vA&8%P#p1S!Yc-yn7zgivqvU3rXFw$N<J-W>~&mRXFHVDR?DYO&iw`dA9ylQaYy
z;E>*PurNMo(B7MuL!*;SriQ&QQG%yL2}?&&2;T3LmVO4j@`2ow-i}4~LH|t!pX7)9
zP)>w$?zSIM3l=q7?Y9oOYj#HtFau6-XRc@*=G$eq=708B7Re!6v#kw@4VJ8{Xh=-*
z(CCwOu?kKRq^Cdk*m{3PVRw*;+fjn$h8S(A8M?4X*o0Da3%j$8i4(9V5uCOQwke*8
zsYuh%MZ4LTIC|7byDMpddssaGfK0A3w)knVExFT8WFQwq{Z{{YDTVd7CG<!Q>a(+z
zJ=h-4+>->eg&=n%J$IE1G7`Z3CAtOE0ZXk6QI>WkerG`SUA8-M-_$h4_T+DvN*DLq
z-8o0`L6(?k+nY>}SC-hme1mIAw%5JL3;P}L;b)Dq1Nm|j)p6}$eOYyur5bIuLqQ+P
zF_x&U05=_1te5e&wzLHK@%E9#Qftc!&k8^4lbu4bm1DjV3UXlM{cx_NQHlqDB&I0`
zk3*=w$kKr~+C8x`LoNeALs8*RxpeN$CA9W~t${op9B@r@Us5L`kZcE(2#>FwpMKQA
zXlVmpM7!e-A{`@1KKhs|ON;yM;|c5}3ib8cCjwaAOp1ptG1fXi_DR2&X0AYL`BZLP
z^Nca}>3Ym$!$^Tv$Luq8eKH$xa#&gEu?LzJ{`uMZrLJqv%zfmDm%HF|4&S*a+UFAo
zbG40MNK!6TQ3N?g9Bi%p#as<h_Y@)cQr-34MgG2%qxWSWa4}AHDDo@05+wzFHUBZ4
zUT<H^iRUSt;OlWXWyDnQnQm|YyW6R?xf|^pj-z?Vw)V}G(u=U*WltOSt$?WXP6NNH
zZ}e@SX=!_>$~1q+f#@d1q+#FnpGw!zHv8WH;8V^47Y&@Az8{0r;FRmnZK6M@q^xp%
zo|^Q-y2-Lb`c~^=+-^Tggca>os>)&eaq=aeDFj_nmMMJ8ZBm}tPXaa<?$WlYlp){I
zpQeaXa(-8}7wf{$a%lH+aObIR{3eLSU4;JW&vOY-qhwlqq4taVOd&<GrxObq&s1zj
zIq))H{>l-klIh$Qzs@}+{Cb7`rlLVkD4$WzogMbuWM&jSTtVqW`&}M>g8V(1?)S0Y
zk_pUYG+G9XQmhrueANCBz@TyzTQ621`(sY87Tory9Fx{5y_oa5?axCNrM)7;*#6?=
zu!!!K4f|_A?0EueA*G8}tmJP_dCNN58us`4mSUa?%Lc2MgzO*ji(5uu=&<y%{r1mT
zPLY}U2EmQxNbvJL_*eZ>sZ_#}G!8ja|6Q@J{61Z8WNF7}E*HzwqQqv1PZ|m5<iGVB
z0#i7^>XKq}2hDB@A{3hxF{w%zr_CNPw-`Q}CQq^l&Mj6${G$X>K4`8R9Np-Fu7fc8
z;JKk8qnt-|OwRB_<`&}{*>Owlq4^a$#2RK$=*q+9h7qK<<*F4e{c_+wd~UHaVHZG3
zluzLi@tx|hNXtKjZRC*-&XNu^aeI`%pthW)GN%guK6-AkX*3r*+j4u%+%Sh0=Z7wL
zNu&~iA6ubP(P{<7Y_P}0Gl}jBo|A#`@i_vf3|%J9Ue(_E2@VSpaR_WrjMtovOSx36
zgW|@X6oZmFT<BJgs3*q;s2`iM0rw&-VmvGMLe)QeiVqDN#{clt7+P0XF`>)mP}cCY
zxy9(pL{v_&r^leuK8xlJLlC@5#a24rGh!U&=eQfh*R#c*S=X&Ky7TB+xv$-=t$l=b
zz?l)8p6xSorWX^?{+znB;`5bZ*Pc7~xHq>UazuCG^M<@5++$I&=jZYZ>qFRN2SSZ0
z>w1CH-#H^j+X+q;q#|}KDYbZCn3O}HiYgnj=O<R&E9KKSS!SRE_96#&VD<D%nwss!
zbBpPE;v754sicWY%*r)6+0i|)`s_1@*(m|AZ#B0k64$8-@Sr6Np5e9q_7h!tT6`lJ
zQqD>d(@W|&6o@U)cG*jPA_$dPPtW~8!Be26^0M3?aK>CWPwK2lns&oob&@fd_|Ta#
zfs2M^*>;vMm>h-W%rLBi?Y?a}5;tMvki$!UhqL{Zw)8AnX~Po-&Euxqh{PK`&f9Ec
z?!lO@4w8Qwc21?p{l$=;7U~T)$`L!TdP!$DynJ--3~|g}z*beO<>x!sr=t-a!I|eJ
zx0u*x=T}s!Tb9%L1@TOjB28YdwU_14-0LMwg<sYaDtg#n?%RCfBzuJ~V^IeljTJV=
zXHmCW8|xId1Y_Pf|7#1KKgP#Ap0%b^wZSIjJZwFdl#*4f=%6h&F@AAji`ZT9aooTg
zZPMKEq%5LHPfK5Gk6joemm7dpQ9aKlCz7YDl+Q&uH*CN?=7;R!cqXx5C;t$3qAB?v
z+&~Vd{p{n^dh!n!H>RrMh!KUTYSVm)i(3}iB^BsG5~#U~{J2{&jdd~={dZ&0rM^qe
z-F@n_951j2qLt9`Q81h8M>~l<aJp<xa#3NtZW1eI`a&u9Eidq_fFL~8a?Hmz&(06A
z#&(hjZgUbt6UI+&wz<BF;iE^}yquOr3Rw!Knjc8Wq^A;99=WM1U9uqNm8VA+>fu`Z
zV`0UK{EV=_qS@(^xU3$aA>m+*Nl{I;ezVIHj-tF!C<1l=mGKOBnb=w}m$uYrn5}|H
z7R9U{rzmX5Pc5yzHQsg(M2aG&uC|;hn1M=0v^#qEias#5*#CvT)CEsmk_T}TCXa?K
zjmhNVmdzM}hd<?->Zo8SJwq-dQLzlDSNP_m#Vl{wvWg?gu^zqXDV&}0i&rfuzHqQB
z8^iMWMf#$uIS`w~pUQXY@~v#|U2NUScgA)vu@$jO_uJ4{fqlC&R}PL94zi!<$eui4
zC4NvQ=ti!0DZyR6)*Dkxo{@WQO}Sd7==D{g)W1Lu>0`)XYlzQZ>5wcYMWfAD1#4V_
zneeK*!okuVB<dhB_Rl?D3ZOYnRk5oaLYaZwC@kS|y<MFbBe{HH|JUTgr?yV9S5>IQ
z&EWV&^HFK&YxA{PgxI_~UMu)zvs9^kJ)-4NdyTK3>e9s^d0o7yyFu1Fg-rRbULU`R
z667^fEnYvjH`H5)PKOMxGzT}hO9IQzQ#Zz2g){S|@?MG=1gN(cwRkt4yw^I^RE{K_
z@O8f1FFWBpd%ZtjN^Apmz<Rp2B9O>0FX3abXV4pyA1v?iSgAJ+l0u<U$i&|48-+ck
z-`<iRbai)KEyeJy$w0<+)3^F2-vL$4D_Z@{feOD8y<A&~Nw8ZgLoK3Iu{L>>fji`H
ztw>aGQ88qIWW*}F%@^3WOm(T>=9^MFVPM32dw?%dU5*r~vG1rG<wf_@#LO?fcUFE@
za#?gFb*|u0{0X_ft0Ke88{dTi_T703E~PtKm))NKoZM(>X|?wxDRomL;k}6(Vw%sk
z_tnYbsB*|lApzZ0FtzvlWLTJxNCSNz2+&0XQR0zr6uteyINnRT?L&_9dvAs3-Uc6A
zoqJx~B@$Ij&+4+B>JWZMV!o&_g`;vvEBCcK>!j)WOTt`}JYi8!_bMDy4O{D^aZxv6
zGVAi!CeOSeuCZcXab6Pn?*OxkFWr!sB_Oxp9A36D-gaS3RFIJ(n-a-mOXAq(iq|uT
z4Y#}UOcCOw+geenesbIo3F+V;+mb_&n1LyNtM5~zvmMQOTLPm@b=TEwdwr&Gvr<CF
zcI4(Z_xIRNU-`i5`Ip(Qcv(UXch7y)&#~R{OimyS_4lmd*4Y!Fy6z(CR7Abl5#D0O
z=|Bia?e08cOS`*D;>6xWtak?`?tfo0x;A{<<XP``V9q`1q?gzM|2r1RUOVW2zjq57
z=b->7SQ=i(wRX6oT&52XhAW)Lj?^U`#=C{%8IAI&OMO$P*|8Y+cEUD39Ao%G6sykV
z#pdm`k0j0^7ZRZFNwtDzY>K(Y=yz|-;&&#?$0Ik8FaaUbppXgCC`RYLBwWy(rgm2m
z+=}v1r;83cv2djvuPbm({!UoI$8rRuJ=M;X-K&)U0^)6~<>Qs};?)jTfJX-675`T7
zibR%ABpsZ*;C%aJE)q`#bp<|^v`OA}mwnnX*57KMxgVA&Qpass-1+v|3KMCe%pgB7
z1X2zR{N6t2xL(vviKow3S|4h+1FB$<!p0{1LLyq3I9_h+i}`43i3m_^!<T&A$<vVg
zzU+TrvIL{#S9}FZ&_P1^YFxj%<#ct(>LFYxw$l@o2QQPa#TuQ3hP9P@_ON|D_TpA7
zFt_n7E$!?ZbuFGltza@w=Neia^_%gf##Bm}>MnjOo_T1Amqhw@?#YA+lk7XW^+nJ}
zCSvQq>n|*C>0i`t-wQ}3DoN8<V?W>9@B8XWcqXvQeh>pG^t2QVzy|lj3Xv-3>a^^%
zAH{ePPFkYNe7qm0j^EeS=qoXYi>X=r6UX-|LLTg=^;>1~d!s8ch|Tu1`~@ggpHY>}
ziml6j9=~{}DJ)x1Q;A}+U&Jq*sA7bc_~M@9Uk2R778E<1EHA&R>swngHdMv<*D->w
zKxS1MFC$2<`%U6_*s!zgx0P%MYuBiaj@SqqgZ(bZh{7ePPrr}1!#Tk{t~xO&TOMC%
ze{etsR+ml+-QCNO5C7;$qT5cx{*+TsZMoY196v0gtHfVo1;qpppO()3*W`|+IAo9g
zE#8qrqFg_vjr~1W&NGGV@5RpdkJ$T;C75IEpNUCmUbQ`x+3tJ#uVm1r9ScDGJ6?_Y
z6(#S;YXto#SvMKg-07_9zlk>nEb)*f@>r><3hml&HqXszE};6Kp0M(Od13HtyV@Q&
z&&`C&9Z`7-4yHXQzFd!`^$q~mUij8OIDaW_GASiWA`mNkNc@87wFsH(i1_qh9y+gB
zCEGhYt{@I^UW62p61J?pmv-!UT^>HKn8t`4RP^Q}=DERBGN!wARisrVGPJJ4_Q;q?
zhgMaVWzKS;_)$JicPn)QmfNG}73(5p1Jr5J9^+s@7~RY4vH6d#_HKJzzDPqR*xBPN
zMs~@KPU6QM_Jn{5>|m~Qb|JFXQ5F$AF}`u(+b^vr%_}xiT7auTB=^XZ>q}I>1))+j
zo-*&aTj0E*l|9v$z_Q30u&2eex>=-7Df37Qq+(7KgZK2DVg=oObPYZuccHK&X`>jc
zh$h@>&x~ILud>rrcmYO$6#T4t#nLwWVtaPH1B*s-jZ`Z1p{@3u03B>2_iyCS1NPi`
z#qfMJ4Xd6PKWiJPg~k+z?D@G`^~3{BhY4?ZK}?`qT2d%cJ^ZPVf)nEH_4mLb3kTX|
zFZ4~qE`tN^M4xTch!OT8|7(AHue~@aXKXk5_$TGR&KiBz2s^m~;4Kq&U{*b)V!=(r
z;VD9MJv9N~kjmty<nGfdB}$icHkN+6O>m}{B;l+e*n}`cKUSFCy6p7WN4YtU+8Hqr
zjH=wpQVd+1b=CAtpIpXq^&l>h3TMp=yA+Mb2|a{43=0sIIH>+XiXA%pe->0bLD=wk
z8?a@OFLf(N<bpUlrAlN_lE_@Oibhr>$kZ+S@AY<0Jk!BeRYD30*r;4jik$4Q!t!MU
zWLItUJhzFa3?k`BhlaaRj(MRHG5y>+a*1C@@7-qS`KsG6SyLHE9>()4QOX>XSYzYt
zg52)u-2^VbEJpJpjU`yXUY_5Uur%!ItiYMlD}1l0(9vdNoStS7Ryo#5PEDXOHs<!W
ztGT_6kHLzkpdeIY1dbh>5L3(jBhiD?V-o{K63tZ8%^k*4(<LhGvq?Gl;*LdF6fc|?
zJ{Ei>1P3ccXL25HwWd;G>x=4;|Lbt{HC>$e7)P*+P4QJNc+rVAHM!kvu8nE=7QDCD
zE(y@p#;6plLG7FFxFI;9%Pvg}%IwI#Dd56)Z$?5>?4707D!5f74s06Pu_^Zh6x{&=
zRCS;^pP_{+yy*zC$;?VCQeTF8IX!Y(>A>4;R-&u1tbIkV%}#Wo;x_s)(X`o|JTL8C
z2<=OFGB;q?36R&WWQMZNbvDmeNOy!To1fp&P_V0*8WzNNG6gBqQ&i*E-@@e9<eI1x
z)n)PK1~d%SmOTP%6%X8KmphhMwi89vW3P;LD{NVHUhiPtEx|1m(6cC)Q_3Tg+0?Ms
z`~*)_5tLz)Z9^7Rf=SDPqppUv=MGOEH`5j;;f(1c!57_niBCo7+zMM7<GOBK#Bl3~
zXU-~IMZLn8s;jdYsg@0yd(aK7h?-8H8#M+0*YcnWT{G$b@<c1A(XPZ;;T*D9DIXKh
zwskv@v}fC7D{?tYJKF5ZfKI+-><?IUcUn(khpiRaiXAlSGgVP^H(GB@N7`Ds^#)^g
z`uupT;-Ia}*~TsHwpFnJ_YS#I==Z&Fwa%^@^1H%f#LnV(pjdDHF{AF#a=F3k<+MFu
zR}aCmf-sf5c#UJ3aJ99i?!&7b36xnR$vP17CTys$P9p8-QoudL{hAmJ{)60BS}NFe
z_1T~ra=jlMFP7YGH`LSLM^?y<LqNjffjRH6z1F9txb_HpU0qD=g1}<r5_)~oSYw3;
z0lvWpKIN2CPd)ji_Qn`kd=GgEL(`r1rre8Y|FD;9$q&n${cRQ8Y3(aLliTerdEjTa
zcT#`st+_P#2<}&EnC!Bfd<#!lc%I!HxWXM{w>TtlXvJUM>c7yqz1eQ_55{&cvbQB3
zCa;9cy}gcJB=ll-c}F~RtsPMp6Y@LbS-1hX>vq|@>U+{V-6#L<SgIlg6os*`7;aQS
z>-Kybu~M91CtAjPlB*P4_S$=$9`WJgk9l8w;8AoXFT&oRxLGE`u@CrOc(jD#ct7Y5
z#t>4{rM3$8AzyeacFooP*GBsiyQ4xW>$0S=bc5eniF~McdN9X&O$ADp2Cu5PHbyJw
zoM-ZbdJW@s4zBDNls&Wc@v#CrrL|p;jcY@a6G4@&eUSG?2Xnz?^KDb2XA&hwaX@bN
ztsCF5Qh^6|#k6jWlF^7$snQ3wCFnp=KSyk9;!I$)*|wZ;hUyLWwcB>5ldB{}?+6I}
z;wZ)MZD)Sp)lOE4?FtNK`yZJYWwS5oaFE^Z#4UbFm10q<y>q}lKKPh6nz9r>Jyy88
zlLHN-WX0Z?UD^&8=RKS8m+h-KQztFIVD~3pr~yGqSUV8!VCX^<bWQGHQvbNcb|^qD
zA}7Bqh+cstI!+S)4#)Rm-QLjZupPPI<vi5E;NMFYqE*Yh<Bug<O%>_Tx4FZ5R88`j
zZ{bDV?e^hZ5=T;(*-=FCk0gddece}lKg1Q<JqaihC5k7Mw23=HF*k*&+P(2*MkeiN
z%Z04NHc-LE$;LbixYalElBwp9HWce$j4$Rxhd0zr<%MH0j2-&u5Xe+TqO`IE@{M*p
z)<G*E$uCsmTyG!qXE3iN<jyZClH<qo*@`|&A~(qCXxJwbNqt>P8C~JQ=%0)cWo4lk
zH5bKE`&2yB0a1lIwK@NEeWpG$eq(-S@EHfG+_?7H#0{1xZLF%AdA!W$@|Sy5d0%T+
z)Dh*8?DKiK5N)Nd6%G4BeYX@88&Xp4i#{h?O_#V1uZ#4hczgZ*WkgQFm*W=&(J8Y*
zZjqt~*jM5gq%!Q2ZetRQTVr3Xzm%#FW%p4*;hkTLUy{SnwVJi5n(fz}%v*YUTkIQw
z6^*9i_2i}7H{(mmuf-t>r{?e2;Jy{KpvYI*#N2JU7MR2jH`ur9mxG-F{TzSCv9_4d
zmWa3cZXU=BTY46uk$f+uy6MCQQMQdJsP88cwD$LQ_jX^^(W<23A0#uPsuh8Bz4k+A
zHe~Q)5NX(ta!DXV5lHGa{^Pp&evhix<4*!uMOiD?d@BdX1=pV@D5X?tsl@WLx^m?V
zh#qvl#1y<f#?NCgnKX1$dE*(_wf!RLrIk!2T1~avFB8DxeB%y<Df}v*xxEXA@*(?m
z#il}DJ@HWnoJ01T9FyWBJ=6<q*l!bLoMYXsN>H}nxus&&BKv*54!RVn(6B!wJ}`)@
zF9vQd`(x$VEvmx*r{E6d%Hi}XnY-ksl)ZLgZHxUmt_?+$ZM45Q6B|8Zl>ODW_uiX1
z`ut~qi^U-PgKcp_U7{*d^Y?(oJ<6(%ahLrgA(p(PjB)#Cd@n1fgal%ff&u+2w<gTL
z2RJ-pU8=~qodf&tx-}kAQJZ+#f8xva11cIMa_5FE0dfj9+++WZ#i7`6JZ-+){)*Dh
zwyW&r1LlWq@bok6f%DxG=d!F+Y8jq(aIxzs@3jZT_hp5^M!!vaaD8?_hF}?b>>=}u
zrRbCkPPT`}&zJ|Xg_f!i_OLpQxB4t)89<~mRgj$Q;qkkh8)atz-CT0v)3Hb7?|2C0
zVUyP87n}Yh4$+rHg@A@V%0U@1a+p23f}$co(%tvlW9l>6qCGY2v2{fyM4roQpFJ*~
zg}lLaaL^t<-;Ew_x);t=tmA)!cAt>I4q*OchQqLP)qhktC+$A$iSvuqvb6`VJEF*w
z=7$yN%0BW3B+ETH1`t7ndznokzK1^*kn)uI#~-Ttmp9&bBj)a>&UYg)Z$l7+mj500
zv=~R5SF#Xgnt*Y*C1;;K-Nz~AT8`Q?5^Q;43bA{^LO}rMuxC2h1FH!_LuY%|{9;p_
zicK*wT4>Q(%X0##%mUsHXbMn0YR`?gJt;|g-7$M!qJSb&Ev3ou^L-O%kC|&P$a$x9
zcK5g03I5BqXDqN6&Oh#!Rn<vGqC7FzGi&nX$@Zcgw0rSlcr50f7su3Y$<QVqwUgqR
zNMDkGq9snQ@0D_s%38Bia<nn6*dmsrQ=S@2^t)<Eof$hVzIS0t#-KaxCHZ}06>enh
z`%8!Hn`{szcNLxCkewdiD`d&d<6G>EgcE&3y(m=Q=*+|nVVB7LXT^Y?f#cl9ouR{W
za&m4kfdKs3b+}y&=g!69j@fajBqJP?OG}E5jf{bl&B$w8Vp!~)Txp4yE@G*T3OaKC
zoM&zgy>&*<KVE`#N<1I*s0!)2l{wL+vW};cxfki&++jkBWuQAR7t(+RN!jVQ^XG?P
znl~^SM8hsf%pmXJO@3K`>02dHeTTi=Sw<gyitQD-3{JGn_6jr{6VD3QV7bc49h(F;
zUKttV;*E_Y(J0svVa3MJ505}ouUgbLY(hdzC1;6=Z8p(A8(5ur^=Jo^5?+J_^rs8w
zANMBn&4lGluBW(|>|rYDo^UDUBHy;jU9EQUAW6YEWU5npV?jP85!p&0C^Jm0Z^~6I
zvw#=Zy3?j5C^JjY#wGdBzRp%QV0uC}?Cf*wQlI(qYfiHn0VJ%V+#TuwG?YO%1+Ge7
zB*cCY$4GO<uor^fhS_<h<8P8|q)KJ8Splecz80gfydtmI?8M`xeKdQZ;0Ze4oEXxJ
zb#E9@FMzpq8kO@CPsKV|ajI%Ya8+(9jjd=98NBAlH!><w(iz+GPFs*$OZ^;FfF4^I
zgUIB6BgCr&YuyH!n|9g#CieKE%i|dqWDShae7C)_zNb`qcTu<0u{}{iyX5IM+{>sb
zi;@z@jK6Tawa)jFdG1DMi3OhGC*<vtjm=dmyINuW&}rM?K908b0H9Qphmh4nc=7z=
z#Uq%W1_P{?pgydO{%~M*D`h8WZe2>kEls?P?@%L%HtUGt!mNThcAH(1&t%pC;iMke
zvO2Bmnv+qEywU0FT1*Bq7+#)3Pie<gLdHXvlkvT`;Ag%WPTif<+1dh=w-x?%Q~&Zs
z)bhMCNXJjT<Qciv<d*2EMCCn*b+FtW+;EZ}Mebm)a~9fAs<LWkG4(k}SN37BFZHQb
z`kYhSl^JTQVq`HiDIkPKBZjIa<yDpB6da-ad1>%_Lm1y={q@Vi&<+tg#T<RnBy@<I
z5vf>58cb@`S&OuAdsX6^Bnq8EgeAMy*VWs)!d{(dN6eF%mpEyAO`Ti~k3C1G_zUd1
zB(&*0*wn6%Z`Dt29|TGHExTd<@y8d_z?Jk>ke<|~S|$y<G2Znk@r^QB*lV5aXqisg
zlGo+Xgn+NG*T?I~lPIqW(%EWn2(VDpiVRhXS>Z9i+uoRLpXkA^Mq6J|(Cke<MoGN9
zf-6xf!JFr&+XV!6FPGu{Eg{B<7s!Sn%Z~mo8-hf9=c{jZ@bDQ)cDN}IOv_arc5{WR
z_!wg^IcauFMWmO#A8aXdYpkyvuQ+744FbZ64P}Y_l1En-9)*(JX>W5t_y%R)-kxhz
ze^jEfr8T`HW_3p9*R*O(?zDHtw|5{)9JF^0fm?1e`PF@~?@oSG?u;UpK^pSD7lG1l
zuj|65T}&}5_@;Kj@9}$diu9fLI+7?f-?!Y}SIKl}ro^Cy@%?#rr6hOR2lArqg_~m>
z@39Z&oRBC2M~U-x<Lig&&@~c~v$PEz%I#*W1J`mtkq%d)7w(8>T9s_`81n4S<S)zn
zC`*Kjxh7vKl_1ovVu!7*lMR;U^o)cO+HULW@9K+<M-`*@(0UN-V@xq_ovqEt(i=iw
z>1@B+Hab%s$F(+SAlu|GEK?2o&5o!(ykMQk3U?*Yie->ha7zFnK7;#|lOq$1tfcF>
z0FT+$x+pT3!P8y1VzwpnS9Z`n!?ycBJ7{O6f+{=W3vnsQ1xGO1?TlHx#=W}%bzGF3
zzX1VxSM1g%f&o++LB=+hK6fX4<mn3YWMT0;XPUoL)ztQcKq%Qki457jAd7gg77ir`
z6b<a|x)zFGfwBECmA%Oe@Upfniw+XDFW=;AT=8IKqwRMtIHs}84g|O;@I|87A$DAR
z4hI9>GWx(j4%Knw$l3@;-e`woMb0$Jol@4Spf~l;JCZy{mj|Z-1JF@dm#*lvV@cNK
ziWnLT{^4YZv(=B~Bl#ZI6_r|5M8F0dR2%G`<V9x>Khy5bot&gNb{k0QXzA+j?Vz#O
zeKCwXG(5n>1*nha1O*V?$r-zQJk}>UP5B(+15zBt8g=GB7GLJgs*37`b=t>+--v(h
zuur7aNe+N4IG?QJ4yy4(RfA7C)vsvh^ncpdMUX&={`^dmWVteB{McZh&9%0*w-Qs)
zW1mYNLJbA^c^mfmN_z^GEOG~bV)kRRl_~ZGAC{0IYGy7+-Qf=ZqEpGLmOlGZE?^Ok
zV<z~r1E}x~`-=Y^A4Q9OwL-TA8#WzBkTJiO-%`?H6>|UAb1JgvX^&AVO?|`1m@13l
zHyw4&^2L%Z?Lc&Oue5LH;9OCJprUQ;u<yjPVm4($gk$?|Jc9u#iK1A8z8C9v3zTrJ
zV#@CFIgDKJeJ7=9ps!&+aA*iqeyRPi4u-{6gCU6gD7GEV8BWImE_rei+kc#ImjH4x
zvYqymTt5lOAl81G!XrV*<!;uo_*u}4ESa={sjH#@9+RqG$}b!d=ZqL`zpS&AJiLLs
z5#@dr&&1Z`!&j!vuYKkj?N|08iL}{olI#XH-6yNBg1Wp8snD|5ew*Z6vVsSdp}XvN
z0p3P>+1_BkufT3Z?)gJKg|w3UWBy~70uYlf$)P5FrPu!Kh((0zwZ9}V+-C%km-sE`
z+<u+>wVujSKMiiGvqgmi`&%rvNCj}+-SE!8$1|N#C6=m8(Ld^J^qfxaKshIj)BnoX
z(IRDP+yn;yoft&R(bcY6q5ttAdf6%aZ+;`H1{8obTi|9znmnPs<G=zR4_M%KDD-!Z
znzM1gdLI~XD?1&g50LBZLGjG1E2469#XNX{8zaloBsR-l#oJLw(;gDvAhQ;65Ov2M
z8nb%LWqNW{$Npgpij@py1zp7+9^Wc5gH>SA-DQu6Z-stom4m5okF2BC;)tvqk<K2q
zpcpzwpEJ@PofAx^?6o~+f!hXkCwL)WCH=uK`B1I%{4RW~e*+&q%#J=TM_8e<(vMFB
z788AOq4AYMO`Z^lkycTn)VOtp7)5g`F5-y`!s@UJ{Y!GplM>^_aP9v5Cs*vTQb3~=
zr#&T}aS?MeBskt-PmL9h8g5UkKP&ZJcN=NVi8PLyS3pP~|1%aGcXJ%x(=M9oJaa(=
zf|9{caXn(0kzSs);P?a47FI2`XD2Es9kkP)<Fk$&HO!v7px8$*StJp@Vb60QNE%*B
zbBIPax`RF6Um@_5QWg|I%=OdMnmnT(J0Tx*G#aGdEU_2X4b?S@g&Si+N%FT7E7lcu
z;N~ii+;~y02SY7gUG2q*4c#2t8RU_Z>dus~0uG1bXa}9dC)aO|Y0tPrciAcV3`t0J
zGj*k(x*%-w6UN$U@w-d3euQ39p;RHQA=x1IQhyugWYJSgK;7v>W>7ZgF_`xm0jaW`
zbO)#=Rvs~#(9cW^i7BwU!S1hTEeHek3Y<Zx<%B6N2%{Up`N#P0DF5xTvlkqXdbw@I
zwAJv+)WdyeDc)-%60UO8MB7_!WIih;zDgR7ofEL&dPi22Q8pd7qjF|4-YL@=MxxO%
zhV*U8tJ;}!<C$AMidJdo<y;&KmC3+vW9KgjTidkB7up3$H?7@hix@&*mUu6Buiu}(
zBQMXfV5ROG+H9{#xJ&4hIDz~Fvhj@Zw+B|MMo?>eiMJZ-M`6-58|S-USoHWhuEGjr
z7QV|S{Li=wyeh^Tn;5&QODzj|NX00%O~AsW+&iwKmF@jph8O1e(hs%I_pGzY^_c?b
zN=H7sD4ux~zI%Bt9-IwoF1&}`!O6-vlFp~Z@0ijG>)yj|h`Fk4jFL^J#+VY2_(k?8
zo3<eI^AZ@>(uq)TNuF+EUp95xbjJ>zDSGYFJd=GWDu}x?a>JHWt3)AHO@S(1lCrAr
zV52<?yg5HzT-^0G)A>(J8+lU=n-$+_({wxrU0}15Kt_x>)8;rakAVHpP1LrvuOKcC
z)Xqzk6?>w~sPp5Qh*HYUUQ9s?5)BAyzzDn8YYP{o36DOkJd=^gl_(X#6&h>jR#qjm
z&}BLQbfP><5!}lixT(mnuXLPYLTy+}EX{){6p-W-U`4DvL3(s%RC-vSw8Yl?(Vdrs
z070CZwpfDX7rEedrMK7JuMPqbSqeESzrYqJz)gyORm$0txHEOai_0l(qI}1>MHsoB
zv$NF6VdBKeGp0@$ZygC#x!c@&!!f}LdBuX`#j~IUMLdh1P=K(jOqoN)l;KR8Hhs2i
zL9trFxQ+w^+pYh#&bo(db*woghi$n7-oRsrmZ0c~uB6bh7^+E{j*kjw0O}hN6jsDD
zPEJwZ0K=|~XC9<bPLB1&vl7k79?1LKNK8?-0arO;sJ-!)#43fdpy^v*?jAs@#=LHw
zl6(i@kTz;7hmf@L=*2}vSUYT0j=zc?&TT#ItL&<H*E>oVEss%@_0eBP@n(wBrV-H4
za<5L9fQFG-=bd&<z*MND+;{4JSZ&busywfRWEF$ywMq8?CZ=_$SWnzUplD&~67oY1
zMkT+xA~7s>#ErbBlHOp{`vzF2BAJy9c%7e&%P&9Ou8+@=s73SqdBkq;pGTc}w%u5N
zK3wA7h#j)m#^@Ug#T8l$!gfF0>yoozfM~zE-Cmz#6k}RZqU{aNMVd;@g*R3#9Vr3t
zy3<`hS60+fdsB!;TX9Mrv^OWQP3@*ut-U3;intgBf9z7mQK@tB)|gi-m%vFfuiX@H
z4>6+fi!R8UlL#mxjEMrN>6YYl%GkMe$Y7F-)E4tL=PFI)3EJE8<*|KB?Ct)K?h-2U
zjs?e`g^2fm_pmBY?VT|=kjTJLs)xQSm$|sF0W*DfonMU4gZ*!hXHc<JmD1+!Jr$RV
z%*W#{yAT^Cwy*fx_xj#W9y86}m!Bj1>SytR_a_#X5blle+-4t$w<IDdheM$_A55|x
zj{~)n)b<a>jZL>_TU|G#L==SdCpXd^0b8-fGL&*rL~)ASP}1I=@uk#IZKXQ+cs1QM
zK7L07Os@?tSx$Y({Z54L@9}fGF2L3W#tFs#a=_N-Qc8ArAz#}Nv?as;TH6?J$XXx|
zJ&rKj<Zq~42qurs`37<-b~{MtnB5i6)}eGOjd~k0%9fP&TTvQGiS%lUK9syC(APF!
z)WGUVO}5=Bre$R}u>?DOJXEz#u5tNQ$e9i}*AX1I$#w;%oWpv>Klu{Z*=`>cD_e{0
z@oxrJU##2Z?t~KB^{V@uY_D%5!rkJIp5^TJzTAI&O6&M<gY8drr@PFBsUj8BYtjxl
z{1j8;hCP@wfuqum0A7D6Aod8I(z3fy<Uu=}>;j*aBF2s+{gg0CSv}+smmTqFlHIg!
z#S-@0u@shCT4mdiX7k~gTQQ&F|I}f%kL1}LGrP&|Ng~6BqqA0vX4t)X;s`?~`D7qp
zELNh@GN_#U?RH<zQDg_KO$70M)bUP&1?kg1pPHCA>|^ynA`*(%fIk;4AX<!vACKQ<
zLKg9;yMjAK<SANbpNQYr6Yr_2rSi>kK<-07EkUfG3}EDTUjw34HRV%@DAKtpRnxFf
zhbT~ZfG8B7l7s-2Z~EU0<ulG9nrL_QSzpvyqcN<0E*D**XS~7W=i?cN47ou^?U;Qb
z;VqIz;jJ7}br{oI2kncFWpAKGDs>Fu{k%xFqISfW1M@-MuqRj0kJ(ppi+Yy!6Ugw@
z#2sRL$>l*+lY>SkA|?8LEn#MnO6yc;*4JZjL0d{uy4e6D;y3bbMb-kdtd`%5XUbia
zsX;XXzm?z(R?TB<_44|5z*C|%V7u2Nv^ksdEc@>yQy8-Z?<kd?R9fk~z9!C~)I|GU
z%qaLg2yT;oKc2}a&mJS)ZL%LYku54gQ$Nfzge<9gGe1g6l0hW)yWK^l{A8x26d)Ty
zmzO-+ML%QM5~~2>Ah6n>#D0=ogtBU__R~B|!-fyHpVhrCg@b_0OBZR%ZCJ2=o*1P}
z8DX3f%<LBxd)hMvLFuyB4gAYsahPinw0@QE72ssjmGq-K{MWhIW_58{YQIT7I%B;3
zHpZYD73}<XIX`C$fg167hy6ZzhKhSBu7XI)JTBvu;SaO@W4=S9n9>HJ-~Qz5rtUu3
z6Mr7uBPdQRYK#3P;7wQeLCH&8VF&E5@wObAGPCSMuKrtKP8$JbA`+5)C@~f+sjB7Q
z>pEn+`$xjVWzyNFkQMu93=kzL6bQ#K#|Tv`U1jWF2~IiRF8uq=`0pfAkSW7&^&+zn
z#m4+6i4;AMbPDfd{a;_cbi-b=g_)??O{Z4EA?yJQ!`O+0rh8K!5(pU0=z$B}h^b>K
z#>a-U&|L_zVrcv9K?{okaoX5%_TYubJ<NxE2z$svH>v8#Df&{f%R?82MXj~Fgp56G
zVKJU{Ew+a*4CCeuRm*rpKqG{5cQ{Lx=4FqJU!<YQcW?)Hiy!Jo#e9mxlc#MBDR__e
z`BbzR8y*SO3yUQVj#|t<kBv`cAs0_p2i(WS1fu8-LP3cgl%@3XzB)xIF1IJ-QkVDj
z+Y=M`C7mq%Nehn`txc+=3qfixEUisKpNFGnHuRHop{4U>NdgHq<tYK1vZ@tzx*N{-
z)O@>~D0h{VQdagfUl`^XYT-8Q>G`n}XHl=8Q88C5o0jG6uxIASi#j<y_N>GTr(N;*
zY6tr4e1?Fe7<`<u_M9LL34n6X=zI^6%I&!e!+h4(zO)~6J@xu&vtrNlJwN$0d;Y>=
zBb+Bg{tFU@Uj9X!=@^@>E%in@L**<g<4frqFU$oZB&dc30<xW$TQZJrV=wZtsmeXY
zUhLb$eM$C?ofMx+^OQoYY}k`?IHucE=xM*5l5}7E8hLFGR`EGCepdp7!tiyyh{_d!
zL5X)e?S7jjLy5=M?zET0Y;l1V3FoCrLg+4Sc6z+A9$`Zf3(|4V$o<2$pl95-{HaUw
z%y?VRPy=9Qo9wK_=d_NiF!c^|++blsx;wj$R0^qh*o;g!HavcDTfAoqx|?7`g=lBt
zQ?i_ojA!mNArcq~@0^A1HIQ3nBN$gx3fBX4!KOcURN&4lle$df9jzR9qZfw#zgPKJ
z(B!#JG}pe|&Pzbt?5h}mJHH~J7^+LZ#0wmubbRS0bn0dK6%118&Z;f=@_43gP_&$a
z&0mq5D#Nx6A7XQyz-%8k#u*A3p9k($;@(&%v*|M~8E4}bhI?SK6tBW{8#X>yGNyBp
zN?X|kXO~NQ`fQ?qOc{k`Hp%}^@Ft1X6<D`yv8y)O2d6O2oda7>xt9uzaHNa!#o{Z(
zG{q^n*q@ceL=+dLh1-+_;==9~%TzC7YJBJ=xX{%3a9ZrW(;S0Itm2@ru}kuE5@>q+
zSG3#oWaDZzg~E7g-A+n4#VuIy3086H*o;bWvWako6nBF<l{Do=L}hst4^Dn_p5kGn
zN7&4y;+}R&z1gfvo>V%CeUZwx-Dbyv#B-4e6~JV37KR~KVFRiKw8G|A<jBIQJi~^~
z%V#v)Kw2o4?D;VqCyEQb#C+R=ct+F`5}B_3gSIf9x&7CDSC={A(WF>0KbPklCGgUj
z=_`}S<oT8qkDlY@GD?4CYsr})KSU<{2-n*pKLE3*TWkC(5?FIDS*nm>o6p(XepN5U
zTiWw`xS6a9oLE~NbLv_H`ZBV|UwRzplDdT|*x@xUhiDeIG|n}L1-=uO*P{Q+HYdBg
zvPe3T<s%CdZM%{~u#!E!!odTS-LQ>iv5Wdp)G(;Obygs2nG&Yj0}?Q~A!;m7a_|!1
zUH(rfr!-yY&R<&*%G+#3U_t3rMRkG3uN-_P85)UVv-Q-KgUe!RQJWn}W5p|FjY|E!
z9X%Sjw|=0PbTO#1#b6dfpC9EF1e@E+n6Nm-K<g1KG^?BgjGI2wu5yrPs}&@~(w`ew
zgfQ`?9d>mE8RLkHLUv6+u;D&<P_S3tmhQxPTW7Cw2=2Wd9&v4+RU*vWmv`B#V<U?J
zOv^1K=8n5vTx2d{zNVg7{USk6#ou>umaoebi$`gh+S}Xp0gd3rU{$OPie8YdiMlU#
zgRgfwu>v>dVk8ZUW;jEyt&3KBS)HqG_PSi4lIUP4%A((hy|?(nP#|6(Z1^@z9Jk>*
zctbAol48%cHzp}4_Kt2ZNpXmi(p3bfm%(_mzqq`;1%DEzNJpIFe(kM3W~)+wWtX@q
z>0knQ)SYy50$$A3+Tm?>O91Zdvlu<tAHwgUeVMRtbzF?bFne2$M@wd69((L<F&?y3
ziUAx&+<kl92Q_5K<Se<!-;vzy!V=G^?A3S1%MvyfVxR~Lv<vL7@2bP>)OI3q$O0(~
z{<|x{-UAcw83-CI34eP4B|%KV61w-_<Hwman;Rjv_xc+c!92zte}cVl;qfPw5ZdBj
zg|YU=b@={@vqKm#O38BjKvL9_K0>0Uoqcd{`(PvXq1@fEL=R#xUhO|qGrqGI-}xPR
zV=Q6jI|o<FMK8?(LjWvhkF80lXXr{?n`4sKpdt|Kd}CTxEwlAigxOLY3|@JxhJ^Vp
z*ajz+fz=bo+QvFGXUsOm8YT896+!WloBa(Ew|iL7T|U`JoNin4Q|NeuZS|FL87_jz
zY@7eHi8jfSk+;`@6%SR`W;+s9J$*|S_1jJ#xwB=3?Fvk|4Mxe3B{2YCb|>ZeX=&J=
z7}!NH=Tdi9L=KkLcHXczQAa&2z_fiiTo<;g6-18g&rOEHq@SvC=m7_8)W|dKVEl}z
zCmUF?M<0sWbT;MP8tfN+xWchb0&n40?gKgE`;O5Q|KQQsG*SKiJ1&-7_(Hk+?O2Gx
z?F7HshXam{WuHn`ynQ6svXs!&RoKPv$^9XFq|NU2O>b_YVcI~PoFTU)v(OwHvWAh`
zeHH(@A0_E(&kk}LKN=Iu&`=zqoT=k=v|Zfle%geLtX1~0A)}SfL*>qu!1r;N3|@7C
zeWIeuW3{DnN)F#`pLFc}|9CL?$v##0ZHq$BWMi~XJI3z4`JTg__L%_EU3=jQ<xZp8
zKWv-E(gA=w21w2<+mAieK3DhrAY@1mB{q@IJK$Hg*ca-rrOK2_xY_?h+<QRTSyf@f
zTa@05NLM<9mR{(SNrg-jGD+x3?#$dN$u#brNiu;T#R8}l6~R8KGbyC^Nzr8D*c$>O
zQUs(|ktU+w^Xzi&#kIcmuYaw7tzVZhdGGt4vd=#I?EUO=9u6*p8(r^zNkn5h0p5UO
z`bZE32MISjv+mVlB9Hpp9mt3!K)|4*$80Lt?GPG&wsjrF>uPiIxW8^5VYPeYFFsjb
zx52ZD$18CU7CNTY^+XIXhbvuRbN@9>eibC8#vHI?oN!Nma(d7@7RpnFXBePutfy^{
z^hcuSGa(ZZ!kO2QpueTX-3H02^u|DV0mQR5km;>eHF_>RXH43g^<kecbb}*QPbK-g
zPYhGyj#3d>Y`DIVR+)|+>Irm1ZK7tcw^+RB!yMdCUh+W|#Xn&4GCt(xIG!?9k#qk_
zngJ!MlEx?gs_i;JQ_STD`uUm<+w@GcKwr1z>u&Gczn{Dj`!w59iJO^f<jo)$!ljwb
z)F^MI?x@3?7s43f?U*1&%Af?Qp*9=?-~ZviJDv*AnUUSRs#<*r@=nNHmas!~qz~2<
zVXnzyc{dTJ1Ink}@nW67m)_ovjc)qd`_Ul0Jax_Tfzb{zzw|oI%)(FdVN4I^TNMD3
zy!<Py&3s!!a|?5tAO%12kuQM>Y$aS-ypeG+mJRgpkY>0@Yo>hcgR~REw~rdn|4Bz6
zRTmy%v?JE?IKw7mYlPPS>r)|o#+ZrK2z!$$jx5sX(sZg%O;KCE9MQ<AD}}l2F-1YZ
z@ffihV4t3%0Ag7Q0K_SQLh6Qw>{4AEddi;u3z?SLMuodf_L^e&r0q~bwA6i=|HlY|
zO|thC#UW8is;&~}DsElP$UaltR0j+QTcQcZTAS=^FJdCBGp=Tt?C0GYL=jpo6{r3E
z7xnQ_JU~>d`%(^wZ!De|LY^UCL=X=&QgpYE%Q3sJNe=Q?iMO)in&#RO+>?WSFdfX(
z#>QO<8}1Mz)W8A#<WQgV5@_2@OW+kCb~bX{^W`w3F>*4I9gEoe!=vX}8EhDsWJgSK
zsCEqMOnx@r&NpcB(IdTcAlVG6gBFk6>nRG>p>#x@9Oc6{MUh7?he)mT=rkcFub6D%
zVYX3@394X{und5abAec%>e%#)>Pv8SV=fS!9dew{(llrTy6u2hoKcRC-;3N~#rKb#
z5c2?NtsW)BJbyl(Va2$uIf5@taioBm)m9toM4K*v<2E_T{@#SrA#!pI7CV4aqj0P4
zDN_{7zW97O)!*SziP3E{jRdC9X;U2KzLGNQ^jKr)Nxaf=8)7NXuxD0R3eZ&pw2m{Q
zPc<Zr3}m=f(#u<*Cs4sxVC-kb1iRPZ#G(ndNX~Xb=~9BY^d3T7(R_t@-6iJ)drW})
zHwRGUT<?;M3AJCV<}QX^(jn(faa1=u8`sBS1H)T|P2JmCXa&gCh$ODO^zjxT|6sS0
zQ<!(pAc4M96ol_?pM*LXZPG6mk%AM)jpf@yCZSu5QTm5O!5u3Av>^HhL<5&2EH@(q
zje8Xe*>UFCGRPaGNDl<x)UOWk)+QMotPiwahS-WAfdQpm<@{g-28!q5q{zwzv1G`Y
zsF4d}X%)4gu^L{{i)>K1S_DT}RxY+dRaVv^=5KAIToUBKoB*IhgP?hgcrdxtcJ+{f
zL!=~@VL14Zp~6Fr^Jw=-X&N@J#~c=KB575H=y3bBLj7qYf@B#SrcE{QNsNqHlxL7u
zIx10!H?7dgLWE@@lu_as9WvS)%M-_x%Yq{j?O)}2iI+$Bqp~@yc#DjQ_L0E{c+n@z
z?LlQ$eJIeuHpti@0}frp*(*G3%Fj3-llu{1Xfn|>#R$sy=rU6^6O|=%az%7mkwKRe
zoa?cnb*pd3PAE@1f#GuuokWhdQM9{#>Pa#wk-=<RI$HGjHl;@<`}a$BLbT!+1sdbz
zU_qpO9?@*z^{YaD;?M&*<Z7QKHW#%n@F9ax8MJ(jQMN@dyt%e$%ybS>fL@nKM@88R
zgAx;8FW37o=0c`WrEKA`G9^gEbkN3ZR;C(Dje)H^&Hg$arg3IUkJTx_dXQ_Ont~|m
zGRC3G$OUXSJZF4aMH(7CFPdy4mFXGocmt<wM#x^34~he!Z$`cJ44{?ll%AUv0_bl4
z|L?z2oyciXumz`hjWNR9xwx*pHWnI7K-WqqL}Fd+p;Dy0=@y&r>|!it=TI02T%Z|=
zofL^vrKku8l6pHJE-RIW5Hn4=8lKOP@s0kmNo6a_&L9L97G?xxlMfNg15-x<IcfH1
z@E2<W5JUTJh(+@A3D>7cE)Bep!)-npJKFQ{JJ>++X4Quopor06AGJ(LnIJ_h!zfUS
zb*p8*Qd)yQVZ5UT9KO^9L|95QV|xG%(wuVJxzJ0Hohh?@#F|S0;?v>YW*nW<><Flh
zGROAK_>u`FWpZQee62}zllQB57QWXEPSv627SVz_D<l4Hj&H}Zy!993B+0{SsFGV!
zQ@DSNJ49_Jthe0yiRHmbsB%s3WrtgQq1-mb69||fNs|p{Ku&JA&d~J^D575~@FK~m
zNCp>jo_yKfYi-oy?JKdYaG22r!|Qjx>Prx+dU42)2?F}T?yxVr+p~?iruurd2JbY=
zb+?y{DVMu~sk04;XyZLZU-OAUNR!v#+hB5BL81G4OuQ&mrWl_2h7Xo9iiS23^}GEW
zi?r8jfS3j^=A{|AOzv^Aw8OI6DEG!13`e9>quggd;uvMR<2R!zMZ_)WjEK7z9FT9>
z+s&w?gUNh5792V^UAwqmMeesZ08fq|Jx;!pR(K<!7y6CCGH{c~1MwazEwspkX${s;
zg`-X?-;G@hr4Uz-(0BB`;Ot%P5Tu*r`zCG2w^qmxOup<|4iM#sX<SUF(Jd}NiXJc<
zMgwfaD*xDLOBIhH?KHlqV=xg8C!w_aH&`Hyn@7L`nP=n1+k9ibzrlrI2sOg@WPv|p
zumt1RkTNL?Q}4@hHZi^7c3Bh@7?WwJki{`uLbz55t|NWKN?n%O(&^pf3K2joOPsv4
zwwXsgPpweid9&1iF$N|=Ml_kBZp(~8>tL4TWVutI$SZ`7a1|ED2-cK`jKI9Bc!iIK
z3MLP)t$>+U`saq@Kpw(kO;P?3(W`?05&E7*&~KHj4hGYRQVmbSxOj&#O_0U1CVhdj
z#4VM`pjjLJ0;W>w3}bMeQK6|4afgW03_?~n!9B9Uk6nzC%m!9ss4nyqFoKnj#wG3;
zB`!&ojj>tFMwiMa?}(rw6cdyaStgr<j1at50gtvN8j~$)peS2`pzSK@@^=X{7S+K9
zAhk7G?&ie@i=fwUk!?2Ub9-HX<J_}ld#Ds6>(p;8JM8c}={)&~{cXfZ`DqMB=Rq8l
zNrzA<WM^~@Q(7-1!VI~27s@XGWihN1R>fcjuwARz9dw1g#Hm4^$nF>hyfDl&lAi@>
zM=)f&C&<;e{}B1PZ8#KoY?5DiFGUxM_{e_gzhE^0YjLzhs2KzDt1xU($rvU~-5hmU
z$gk5>c;KnC6cl6C$KSY2<}P!X0$lpPca8ow)<tu=Y4-gt<^ao}Lw+B79=koNt};}D
zWI-8r`Gb+AyB#<s?~bN><&V}TjkUa8diZeBw5Y@~@R}HDr~Ju((a<pbvJ{e0bw7E;
zsD~37k;advVHBJm%&q4QgGe%w$4t`EylBH6o$_bfh;!$jC6C9ZLryV~N$Qg3hv$)t
zo3ul~!!HZx#}jF&$;*U~c<J+By&;}I0X$mX3>#V|PX;$&<JQrM^Hdr%(?|_<=E5?e
z1E{4u9Usi7s|0XXBhT2Vhap<yZ_aLJ!V}P1D$m-=VDzY*DbM*pHK!V_TYw%*X+7`1
zkPi^s<)VW*On;BL=8&bCmlqQ8S(MZaQo|5m^zLb3GhzrAoytpT2e76P?&*cQ_2p13
z00!0}CcILu?N`DIforXbN<YKHU$rGdu^7g@G|FqSJ1US5^Lm=DquD?Y-tcE|g)%b_
zVq}%PX%w1_K+(6-Lj8~6lDDJfQi$w-IMI0j#6$9q{R@@M5V#}%v>s56ptXf}{Nm7*
zcWs{{6R1Mov);<eN6Py_Q*22pYwL_)V2cl8V>UO{w;^ecj&<A^`Js<K$#)!}|4Ose
z3#%Z(cgRQn3}!BlM9VV&w|C~*6*!HEU$VZ;$Enq&gom&Z=gWU=PKf2NL%;vE$(Lwc
zl1w$g(#|VL2Z(s0<V)mJQyuE5$e{TD9#eZPmDul!{j`6#2uBow4hSGS4&*aaBW#j)
z-%=CUAbU=AP!&$X1-tIuxkmPys%YFW+(;vP`xmN9VuD1@_L*v+D2^Mb;lN#X-v$B9
zaw&jDan|oUwP$a&&JdOahwFaUfhC0H>yYxZe>95H2rSltdY^K@)E+C0glbuUoUxr5
zzS|}TPVKR<&lXgeILJPv4qC<FE$Y<=$7rA=;I0KtpQ3=*dl_JD9um~h<Dz(*92&pi
zq#_QrMWX>}^d1)dAY%d2T1%T8?)}(ROmVBm8Z}y2Smg-+-Mkr=6Y0k?M9w1l>{N$^
zQF1n$tEOPAmm`hTxN5vYKIg-s(BLV~vY!TD95q!z?1Au29PPbO{h%7b-4jxV!41bm
zgF{Elu})@JIaCWgZmOe^K-lV=P(QCqjt^$KoR`9%;2qc*1``t<5a43;1hc<BKUI;)
zkw`1IzfBk_Ct3?&g83lCNil@+)pD{AL7j4#4J$63;?Ka=Xj?Pw<kVo{30Yo*B&Q`}
znifO&$b>H^(<P@*^%%Y@q1Yk5LeB6;@e-spoouK=8GJalwE?oESL!#ea$h26c|X_+
z?$3ngR7ho!v#0hbtYI-rRFrw!j}^#9MG0;p97{w<JLH_iPK?t;JSeskqf1btGPW23
zrkhZLt3}T9cj@M&bfC#)*NpY{QJVHZP_sk&#G+ZY5P{9UMsg^Hh-#32(E-B!rZewc
z`ln`4!B6+?4jJIjXyceNwN3{5(zSzb>t#@oL*Y`RrH7z0TR{fLK(dX~o7&U_91<%>
z>w89<obR0!E~MKm7sNu;!Ys_mg=uW8ooW|2UKGzXLE56FFOJbbxPt6ztHl0huE-_n
z8@=5cV@6WB&1YPrx`#@lV+bFZO+#b&%xJ>Rz}SGL#>A7$Wtg#TpT7P2%kb24CzS=Q
zBRe8!tRLg3l#$jKu2Mjhn2fT2jYggZ)b!Ct7Rb|8kgKcZGH(&*x#D$n!(8t3vpK23
zg6&Ya^>0kHkCF<&J*|cAcJz-rZd&+tY-+)8QUsmoo<nF3{@ftrZ2sjeV+u#jc$@5;
zyj&3zglfVV#07AoObBu-GNi5^s2|9X0s-L8kaGXsOT)uRU;sB2GEQK%ObqHG!lb$m
z^skXgu_g4}<pz#OugNCX5uOAIbY;Azj7j!Hz`O?EUgi5wU76S+i%s=ZCCuR4g6sI|
zn3D$$u=3qCLBh*%37T9R{ixT{R65fRuZv$aTmw1_b~QDZRdRhIu>u1K@ug8j8xF1}
zK#^18d@8AEm8m`i#4rH$X4vIgnHHjIf}U8&Fi&%01dC;8m~7LlGW9$(V_=<BSnB}V
zLCwnODwA)PD({M$l;-?O95LCL1&(~?M$!jkCrEW{1l;xjM7_p8wnEkHe1X&&|HD$Q
zm%2Esy!sNZY|P4Zqa|YPsUFX;*E0~;Qt!<XyJ5Hn3e`(POdGLp&GksiZcMb<7?DrX
zWKA}+1|otg?O%F7CcwrtMv6!XLLd-1e@3y$yEzbWCiz%m+%b*2vwgWpmVlg!V?|oi
zD0r8Erk%)4YXK#!W&q!w<!_i7O&kQ5+yv!M!ZpxsPGwjw;8XL!8Mn?c5usHU?Xs>L
zLpt(CG~6bAQwWi|#%i1cD9Jn5TLahQ1R_+!Bp@c}-y9PgmYLNkUkq}ncbdc6D7VBi
zX(XHm6K3VsL;<D~0M}sjN|7>gn<*2$Q6lPgn-*%J!4>l*`v;4Gq1t3jxZj`w@|75f
z-c$Wm9|+?Ph;rfRk~{469PH+t+!>>y{g5?BhTIjySDzoP$9eL##I8&^U8GW;xacbR
zx{a}JMvHvII)~K(876ldR~2dt+!F*I+FH-+&hPbp$kTe~EA-j>ydN|D_zT=)&|MeF
zH$!A80U=ZK@-1%wI|>ZKLuZwII|wi)gZ$h3qd5Sz48dbwgkjy^iAh4fFrf`Lfjr>P
zNG@LVM)$>o(OxNv`^$IjUzN2@%-s53a0pycINhNEf8Ty*I4m~F58_)mV`j?_jqtds
z9P;5uv1sL^%jCx~HQxJ)Os!g6b>l-}t!dt7$1`YNbcYy^)*Lm9`H6olJBjxkAU94G
zz!lK$7fdz4Ue9ezpK1=n;OH<~4oiTptc9{LO`8fFq5dveWFiB_<8bMXO#_<AatN*t
zp9ZemXp~OxKp_UBtW#OT`F!9@yaQTMU>+b}Yh-EKOAj%nhr7u4(lUD+Ws#V+Rs~ZL
zl_1mGyo5zp8gEl}JruIb>d)k5g^!QnU9!@cvz4VEV-PqHPzOcv_SLaKr<`%p$+9Ni
zP%DfahzqQ}V|t7kM28T3>{DcK$vUUKsmtmVbhVdu%L-X<U5|j%VnZ}cFC7+uiBucY
zGsdZ?Ss<IdGZO)%373O0gq!0R%{eew$M|@Qx5GqZ2oI%gmk*8Z31q-2Y+4kZC|hGv
zj@Q{qfZZY6Y|#Dt43h0>fYd<E0kp%Ob+?!D@(k3&MIgja(g4U^jK`QSKQ->@Zog{0
z?2Ms9G1=i1B6F;iT}J&9EL3Ywx}zN(G>!D7vfI|O33_v*{LEia-vI_hZM(-CXPgfc
z_sqni!uj*moQYOsvB)pt83YO<{DHJ0Dchd`7E8SOtEuk9VeMn2qp%`-qx{<6glCEH
z7S?5r{3eOC48k4q^4pj!HW>qPSP4&cB>7$RjT^)e*b0qot^D4mH!)jT+bDnVLpp~F
zvwsW~paIa}5_vd%i}HhNd#sT^#Vd#k1RnZ`cg(b)g_aDWDMHKTQR`#^Toq07nEj)>
z{mO9@<j-kBEc8TnS}BjmGX`-{;Q`tH5^|@a9-;oM#rZ@$fqu(WX9PX^cU1crwL>fW
z$uz{a9I{cK3R>Z+I!e!{V*?d=l&CsHPV+NqD7=EqF6#Q5@#hf0yw4VqnpP7pvojyr
zbN14}fqmq8V=<UAEvVb}cbokfM4+|E3#MStY?c>|@(5yIFE80E45e?Bmtz8zP1#(f
zyy64W#Y11~;L&;2M_~}YrBMOJfr|Z_Pmi%+wElH71MA=&2e{f+UJt=lio((e@{l*Y
zc?)(hEc(rO2{kJ;!r(0%>)g3>Z<M!#=0tL^^B9Wv58uFp`p7#mTm_?elrEQl+Ki?n
z%7uvn@A@kYmQzLrxcZ*Yh%Q~?PF75du>kMKU?7)arB}%Zac%*Hg|*DAtq;9Lc1Nub
ziA?gZG!WJ*qJ&r>AH^E=m48RGxPhw?ffVwwtpPIjFvS0)26@>Mg0!%=|C{6+bCeC*
z>e$T`m<~0iWSW6-9gvo=m{3$v;b<InpPCk+8Y=%{co12#$214kFs*C@1KB_AZ*lnQ
z%LL^<GtG`tTj(NbN<9JAl~pg@b6O7(u+inC$BmV}>>rSpSC5yyr#a9z63*<r>@&?#
z!O^g8Wnb?QcXJ`&8h8{awBIyE3Za;x(jQV%_Mc`*9z%v$SHkLe>3uVLiXY%@n5l$_
z8YnLZPIKsmmpIcAeNcL7HCTq=_!c?XskInv$#j2ST}9QdQx1s<l#MHwLv1{~=T{4q
z9p>YKfrYR(HL&nm%F>m?Q&+4)&Lew=9ATt}-CYO43gA>e>#ew-4>&p_QW%D#3WOXv
z&5<}>ub5Y_M_xW>#H&D55?1Rd`xiw3g2|8eZY>toK+a>PDJIu@z(6@Reg=Os!ecd5
z^W$u3P$(9IYLnyb?ULrp<OCmGm=6plC25FVvZCbYV-?4u^qqVm{<Z6FxU)}8WMu*+
zz$LIRb_KMulY*@MdiR%;jjYxFA34Pufk0?#Xp>XZqO%b<4OG;bSMoHUj#=C4kC)S9
zieMuaX4?qDpOLt?O95u$0m`wMNFm(oOdm4EH??J!VKUXTR}6VP?p{30{?UY+DQ3#q
z-kaquLXW|*eU6XX9L==Anb#QAub0TV{td87s`o74BIm^_LQUwB-qBwPvP%2-utKI#
z2ZKC?7qo9syeS9BpkI0qwLQp1^p-;Uu9N=h7j^bw)mbkD`$z^%bBq<P?z);*D%Ati
z=y<mkI)%g;6wF^<TbHe`lEG<RU;?2jT)2ZjiM;Zb%f&Lpe<xzfXmMJKdca~p)H=?$
zJ{XJKh|_2e0hkM{k1`Y!scQybU+9A-VxytGn6~MO7lkY-(~N+NeZahsNqGz!*d_kV
za%}Ay^-H5cs|-%r?8{r?Z+BpDdSd{%bZE#9VEbBnwKRGiT1VA-Sdat)4K)j8cw#ZL
zBDT>P50@d;b*uQnnk^&JSQ&3dymE((jD{<5^F+0bO0=UpmC{v3qB{0?as235G7t$$
z4WmOYvn_iW{9Bj%Bnt*4R2y{yjtQPB%gm6nm@(BzwJ^J6Y>W*64dQL^%VrtpgW*M-
z3>~C!3=ELzxa0kIus_;?xKRj>`IDG6e8X1+m6$pR0X)HnN2%FK=2GV~ic?Ct&7uF`
zJ~A<8SXq~=1Wy8bm}IMg3)o<G)gaDxavH3bltY}w3c1pn7&4@fT;=2X<Q<=@(Hoc*
zC&SggT6S7+uXt20e&?a7rh{A)M<b4AlurSZUhCb`24c_y0xRUY*o6oJ#pN+U=CPy7
zWlEZEal{$jHdA8(Kqd|fMutp_Px}t&D;XCEi+2L#mkNKKP$vO50Ps>7qXTNY68~E$
zRnY)$9FuHNNL5)@TwN~J(_FJ>C`%Kn#%N{Ini%EI`KsB|G&Z?c8`D)N*koy)zYP+B
zg@HU0N7MMr=}vUxS#$MJ`cWA1Je`py!JCB{La<coqX$Nns<>Js4Kbdif6NCDgDIua
zwhHVK*d%Sz6x@}sg<77MX6K>)edUH=OwBE45gkHHjEPe(ujGtW*hA3X^Zs&iNU1s8
zZDK7(X+;0n4aZf|nutqh41_y^o@NHyHUmZkzd%Er6~}jV4goRDVgV1$wn?Dgv6?Pz
z)6|8z>o%ASbD~)k6VgC^c(HD@-%ANQ-egSCoXgZSWMpo12=603k$B<e;0Y#6HMU@*
zd@-a8z%n=jZV9o5Xy`lxE^l?3_30<K*<T0s>n*p(67c*0;h6eEyMDQRDTYW-A=rf1
z27EboTR{m`mX7GOuh<wyqTCD#@YSFSu2Y=}s7Ol<+>zKym0Cy!tx|;^@ATHq0mWPl
zk_Pu(gI!pf@~+^5;jBH0TVR3)6)2&je%G)2n7Ik)!a_ZC`G!BE3}US(#*p2eda|^e
zH4NQ^vAh7^z<D6|_!td5HU_s+L-#rrVRPLV!vWlkoF)0Df8-{Tx`wnGwS22E1T|n7
z*!AroykaF*EoBi-))jKU|HAlCYA04xkmUc4iwX+qE%Jc3%7Ro(Jwa6@dC*zTVg}^9
z{z^E$m~#8QkR6jtCdl_~I%i&Yz5F0Gr)iW}_096bka#FFh3j0LZIB1|<7qudt2_Pw
zl3^+BvGhP+2?E0jn3Z|ng;_EtW#$K;vMM9eQDnh1b>^OVnk=-xU2?H3vcJ_-$zmIQ
z6U%2vM;h`n2umg_JtUp}j1hSOrZ4g3rWzb~I?*`wVnwplzrp!Qd{u@0vh<CG!FZh;
zgpV0SwA@Hj4yPL+-iQ8Y?biXI?10Z|g{-jm6jI2`$`~=sWXdV5t*lDDcd=d!j;O`5
z+Iy!~Mod5h;?xJGw39W#dqXoV4Zv*Hn(R1x%C(ouIwRKH3uS$3g`w}pm%5Q@)C2t7
z;G^cIKtLCuLToiXp*H%L)Mj~LZiQy1DkfB!?HJ#tppI@|tJMWPS`8#*AIj$RJFG1g
zD9oy3bhrBOWJ_$_Oa&klim$FTR(j7kZ;A!t(}wkJwOL(snrw@mqrnq~ky;wZ_7H(N
zz^QrJ5if$wR=S<`43iFi;v}a`ASekiMt)lK%!<yflAY1u<@H%0bGu@t+|2-OSGs-N
zX4shyTb#zRJ7$`PP*exx&ukm@I`?epv1O~`Z81NOzFHeu`sEki7k4-5!E^PO>Dg9=
zEEvM^tLPuPpK=OBnb$>9?)tTVLx&T9UByj*^9gP(F5UQBf173pD=n_ZTK+BuR$2=!
zLVh0;g2SVw4zZW=hj@apEG--_f3$hkqRNLnoSHQFm4p7%Cv@nDor74!L<?c~kAxC~
z%VC)5VWD7;8tX$;Sh&h#E)<DRZ;?Oy)X0*o6vB8<!=WCJY1C(M|GE4neYOS2;}cFr
zb0AnY0+<8Rn$%e*fA!z#5wn`ZL{*kRe=@`CNt^JPCP<~HV!{X;xZQT0JnhdA`H4M7
zwp%XGc$?U9H1sfh$}4(q$kQmH7~iNWNSVJ`o;C4@Ye;~FK4&v(0F`;#J|7dz14L_-
zzlUH2b7|!73pPev-rOiJrna>#CvF<vATJq3SSSw5{IV00_qq_I;GtkCJ+Fl9Da$nC
zLMwUI=V5p?P<R>|MfGG}OY~TxcH<Hpu&>)5$ino0!+)k&Wwh2jAQf+>&Y710Jsskd
zoUvNoiq1#SKF!P9ahi=9ccuIzW(8E55d;Xqkaz3@fH?>T!D#=PnpUbq!gTn&-VH4a
z6>M2V5hQ)jNHQ9(nVOuu@6G$cOvF$=@Mp%urX_u7&$`<Y5eY!OGTT%w|B7d|s3(P+
z$BAWs<lpil0E$<-OaAS{(+#hT0stR6&~o`Wv5S|e1#_9tne2q1J^x*gFa44gLDeD!
zA#km<k$eEaA;}mzh{y$vl)zmY@~Mmm>`lmWpUEDXo}+NWvmtP#A^eDM2lt!B@acFT
zAp8O-XP?Qac3Z}H@I5n*0pfx)G__Yo!NN0p^*ZAm+1uVzAZZ`_JCOB8**D`r;TXb(
z`}qh-2IxSLLknepACYN9L~W3x%}wtkI>5iN6OJ0yQaLbV2q|MVqIx2ebT`XE8O6#*
zPnakNTi5V7lR*y2I8p*+yow^OQV#W2BBZ!b<@vCTW2(3xg)*v54!2P-d|4q)-avMQ
zk1TE&WxasUW*qQ?W`s&bxLyUD?nr-;CKIO(;Ub@lXMlWJwQW1XK94er)~onEItGlJ
zfjv3(82`q|g9+pCI5yKWf-2ix|6gRM$#Jp#4Xu?;`HUQ&ao8B6gf&23rZ5A$xRXzF
zPEN@5@Pd&k;2E&wdG>!(<numU=x;Cv@aN8j@&)6K;TiDLiNPEBR&2_X>|d~{=s-T%
z-s^53-6W^jnl{a<ms7n-?j6Re<8CQAEwOYV3ZeL@KHa~u8X7zTm&h5u^d<lmcvQjC
zDQBi0n8i(-X^ZrVfuh)RUe2;r!lG2><!sv%5Zqeu;hcE0&k)?hdv3G^nL;C*SuE$p
z8>}TMy=`iZEpw!glR6-c0@P<Vy6w-tiSr4TX&j$6j9XsPl+^fv1Y{^VQ~dj<Zx{oM
zx)z!&!NBqXvE0zRo2v6N(BJHUKEd$%!l6CLiCk2?AM9WeTlDY__Nf9`Rz$H23p2zf
zH)OyNIX@@?6Q0K*CEo?M#C-?#?kyJ@RWu;JQ7-bg>9Jo6)fs=v#Tkb_Q2UpaF^t}q
zq^U7MLn(}5^E&xU{TECEgrY2sT3IDA78I1z3MBA@4NXF*P(ch|CbOW_Thj`=AS)}-
zs|b~08MSMAT_?khFD9UdMWc+ce{{DGEg2&t6LA<UMHpj|j7p<moVbF_GTM77Qr-0Y
zKullaaf)p_MQBgcs^1QP{PGaZ%}vd%^)kj^H|^B|h%M-aI>>e@OU?2AR@($-MJcbz
z*fcoa2Sb^Q>WQ{Q<0@DXPsS(0<GLzv0z(a9WJ8``k#VQZDBO?1({_TtYDp|SQCGNV
zFZW+)9MMsgvaDrdG!KEo*s6K-;bZ4+kx6N#3fJ*+QZ&qHOixZ<!l8m^Owh8ebEU1+
z<cwSu(^It!c|-W_YGc;HgNMj9wkzPfo(EKCi(KnXsxZJXwyTqeu8`|uuG6!#vh~0@
zuTQVCpn@uVTrD&*C0>Pws{x}^V<H7FJ19+M+KicLags2qW)?!_=(5bDk;7jKf^P)=
zT;UCM;OJb5;cZhBi#eYfGAxz;4fHMO7QEnkjZ_(XwxX;?i)8%;vK4ENtbit09gSjX
zyc9H68P+9O6(_XR#6WNeshC74Rc$)45ZQ-%%uvVcjFhENA5a1!C(~0`+>gNMc6P`N
zf2R68I$^=%dhgyyZt<$%4<-B-c&-}K@2Rwu4xIkkm7qzZ|8A_EOk}PJ)_kor`8pZR
z$l&BMCMY+@(LWLh0B#zslpA7iLF+@jcrMf9>F0$|;bN&su|Q_t49S~M-7MG%FM!g|
zn6bqs-Q8{%R=07Zw8pcU*&44jCo|JT3Z9zC6k+(<EbjrrfVc<1E*2+evc+sCtJ{1K
zS<9p?JtHh=0&Gs4w3AxoMsH0S2L#!OQQLTLij$%oP%WYh=LS1B*IBiun|+}C$XrHt
z;%OEa80w4OH5?lrezHPtu|0?AZn-sjR<EDVz<a*UUMeXsmD~L#vM9NK8_OXmbp0j&
zg}??wR85LA>g>xtE`o5_9SW)yFVR>0OZC^H&Ez_iuLfbpx5ynqIGAh9G-T1#JA))u
zRv_pud#Ae{eqNONLc#;E%s2)H^zI{HPhHUkt`R`s3EH9Gh~{<u==)hBcSrMOKt-Xz
z-V^UYjiBpd5e|`iO(w(Yz|DSgUn0*2XvJo8a!+^4H*FZ>C(^I~t<XRkQ5QqLZL{p%
z8!=P&JFQ3`s|Nz47DH0TLb%X(VkeYC9M#JM_UQyf)yRWsbgTu(SnYN4T_2C$h!*}@
zfHFJedm$-!vr7*34zu$8cv*wG^i~JR2cil;NKMlbYuSW9^k*GVW_Y_OMbeLA=X%sS
z3NG4An<^2*-X`radLSH#YMy6&Ro6IO=6m0|A<!C&-_gwMk_EATBg-esLVr!w7YvTK
zCs_=}B5N8&h4dnY#py-Ho-x)zHyaL&4r{Lo_dDT&=~&_M^fu&Vi4B)03|x&%)6OS0
z0fknTZmPYuBbUYb0Nt$x{&pxtd08G!A}XgE=fy+TUuCudxsfXhRA;;_W0+U^NV#B;
z$2i)kpP<r{tTOTpLZQRePHccL>O@3Rr>u!ZfzaHF5rIu)ZM>%mT8p6dFOhX=`@!jj
z?S%^?@F!0>o5gznhM8Hk*tuml#6p#~Ah}sK7FLI@)SdI7*czO=$r$j=xu={ho6}5I
zYMKH4lK{tXvTunsEE!fRT`@+e6m*D^!M0j6vzctTEtm*MBlH}QaC`cKVcgV-SILgF
zQK($%s6Vk$eiE}Lw{zA+-4F^(9G1D_KlQ<ym)FWw^5EJmJ7ci*K<0OS;x*fq226GP
zpfrdQG+}Lp-DyyaI;K&&92CHRF+i5T_*w8qwxNOtS&z3)tr=<XU@-bo5Q!^4_h0Pv
zwZqMVGky^)1eDHI@L&2+%zmdPI8T0Mrvow#S<ek*{dEuon}%WgJUSud2?A)q!Ee%7
zSiHtERTVmu-zM@~pec2!-x-6z2G8Xo8Gj!mP&zY?VY~dn8#J-5<~6$FCB%OW?i$vH
z;$`x%zi3&7JRg`YNzCR?#uokh!ajK<mL(6j<)c9aLQn8!L(V@I>jAdn_~%2f`LoZN
z3(E`*mgz56p&qv#b|xxT{v|~6c!tM6krpy3#1w_V>!_jtwT!oSJpfbo$rw!~P)e(l
z{!|>!mt`BZPWsb9A@!8tl4Jl&&-gP_s|;!XTcR4uIFN6_v6gS}tdA*@go)tWH4y3_
zkmsy5l;;5KfqX+Ae$U4+s#>ia`QL*g=5*$*L*xSJ0~QSYf)A-fL2pX``SM~6sVHI$
zaURg&UW%2|`vmFxrM^!A@^Y+J-+_bV6@QIKJ9Q)7i>e{L>hhLl!Pd)biAF>aai2WA
z+potAaDwDnZtjf`51FdzraZrCyJ~V>cGl3g(N*$R?6M3h^5KiOU0tMAic~!$=`rDw
z(n;CT1d2V8f24(H1R3|@L)7Pe5w^=a(KaMLGLz+>X*BgS;8DjZ`mVh(3`s7Cz{<&c
zg(C%$@co#gf&)+>fV_SX>;otjp*yq(K8$YiD0PS_{L30ah!L{_J__y|QHRi#e+PvS
z%EH3;^09IC;K9A+Kk+8+<H*T>({k~W=!gpHf`eEp47ao;)cf}2Qx%5r*<HJ^h(J8k
zW%fwF@Lpic*0N>)w6%kpa}2E`LYtS*R0Q4$Nsmov5nBa2X3q-8H0w~brbYJhp(uuc
zRl!x?vUi0-n3Jrk%|6jRY%h~xa2;Ze*2=!tR1@Nf^6-4^SE2A|cl*TQWwL*T!l-?E
z_vs@C_>hQRIVOQ=Ffzvk)PWTa*_JT}AR`A=xItZ=Mb<dv-N8OcN?huZJbMpG&rHol
zy>U4-HB?!efuHLz@0g;9*7`PuOghz7MzMOh4H6J}n;hZO<iM;X!FKe7;@|RF@5xF?
z*lo8*ddo$=K<uEx1>v7dQ#M_OML+mEgLdh-JIdMwt>BAlt*J$ty&UcR*s(`1=4Ls@
zpIKE}f-#HbSbqks3~H7kUdILP`wkc+$NN;6i^33<MM^Zvi4$V+hgYExj(on*Bj`gz
zbiI5bT8G@A=EAziC&n{)lu4CkhzB~ULSdlGhRMklia^%^PCF%DR6$C2>>@eUpXo8>
z^6E4vPsh%s08_$QRB$$*ZVQ8il4dz0+EfJ882`+)ux48UF*8CaM{ST^g>OJ2gLuyJ
zZ+ONt%#z`aiY4hcOXTdt0A?!k0%yvt9db_m;^+zMg4#IFt#H_UJaY~RkDeD#5dH#9
z0HAsAV9Ih_CYVEXq><~)bI*{zX-J+@uTU428Af7mmVUMoD{yQhT(!SViUGZ}(-D%}
zCIgI6bLXCNjtulklYcNksEr$CPz;?;9u=_EUC9cAjblns<+o9WSVxF_9yfNPoSzyZ
zSY}7?1^&!cVyzT+p>>XWkTr6VcZ8En51qnAhZp-K?F`@zA1r51Bu|f4xx~LL4nDOZ
zUK(FQZ{o(W3w*c3qX3Z!rK_aET@MVRK(xIc5krl<Wy2>*sgD_70tcc_xL$_E4%nrI
zv4`8l^|q~g8IeAK4GG?7UEGnje7Ng9ua{Mf^5HH96<I=r4g&6oB{JH2M!65UEWI?J
zA!j(LwemGi_sgyS#%wF1tAJbj6dldbR6^({yKlpES=y;AaY#}81okBRq>4M%`W!xP
zgp8|je@{yeW*PL6Dj6ROgqwk{2%(xq21q^JyAzB7n3-^$l>5xp6vE6AfiDyN8IwwE
z*DsMt(FyP5v@&p$D;%c=Mc3kgoAJNZa;0@$G8>s(SH(Kv8X8urK&8dJTpcWjl8QBQ
zO^h^$ki|q3yV)Rt($2Zp#fEBZnllG+x7Rzdd4$0?rWY*p<Xk0F(lVs$GO?vKnlY$>
zV#csU8)ceNqZO7yon(CG3b5h}!q*k?1w?gyQ_~EojA4RJ2qHo><4@#D*fx`@#7xMY
zBBMZ&&H8|>%z>dcG_cl7wGn{jt)wP|c{6fgYSZk@!wn~x)1fX#G$xBuAh0>68*k;a
z22jl~3TyeItknB(?R+y$*W7bf!K%TBn?OgPD2?7X2g+y+diHH=HpLM8_U|Lj#)Q3*
z<a0xesVvJIhH4}iFODcJm3-nT3SSC@E@`Rg>1*!1|CbIuX-y-xENupQ%?#d{Sc{<N
zHkoBZ%g@NjY=3<Lf&`W#*J=SK*k&XG!><I|GsmY}3~nmGC^v@G7?aIFq~GLCEy7uj
z<Aut*%#E4Dj;7mPZnmyRPq;$9=<id(XB_`#@Y5}EPyoD{nPm~YThjnxE@FWAQ*N_o
zT9^Ro7mOp7+kKes#n7Xc2E<DFQtYCkljO^(Px1}V@zwGbf5x}zsE1`BU-f6qkYOSi
zO0dWssRt@-){p6;cUljV%7+h=yV942ZcrY5&A&7cGg4P6xUoTycAU106|SS#Os^3E
z{b-$i!+(#;smyzzA<09SNgsEoG4WC_YJ!{aJt79>9vcT}#7KNp+5#=_tuQw(O$w?&
zW_YWEWM`KiYAm{UUohvGOr?C&r$bvD=?^R&E8mJu1J<WMh8GmFNxftNs^Pao9K!m@
zs&y{+`zP2(;Bj-*SUoOv>d1GT<rrV`kUZd>YD9{z$U1p2(RDi!8+uFRyU`JosZqud
z*A{#)coKQeO*QiU_zE$|ln@9)kso+p;3Y8o4=Z|#?sl=;CABEkkJ6$ofB@A{Pv7Q0
z4yMil@x>h|(ry$hA2oW!M44yIkikhN^R27y_5r<t9xO<Ur9skWgwqC=h1S4z*P~jU
zEJ_`BK-nU66W6&dj``wfZvfhd1Ea%wDAO2O=}arb>T5h<nfs<!t}>KDmc*%q0BELX
zAjS&fV5#+vn|EMLE(=+~YkW}eZ+T(>)^z7yT`Ldy2(97_cREx@tjJnf5#+B%l??<$
z`p1wfIPv-CN?8>hAj}uR%d$GQ{CEP~y!2*`&54#Qf<@LEJ#)Y`WL>akN$GG|A8QLf
zMi?0MyLqy~pAi_`2m&xkZ=-!QRLehaiWj;&!SK|2x5#Fn4uZF!WKsY#YLI~WmJl7o
z=SWvf4H>9aS=k!X@SD-LMI$gTh+smk!XBT8owz+F2(%S7O0f@j_#mj0Vu3Vj?<Y2C
zR=9yQ`)N#ccw1I>`s>I^g+`C@L2&HyXFRyn<qF?Qw>8lU&mub6o!$W5bwFM6vouIX
zH7tP$V)XSE2)v%ckg!n{rI4Td+Z-mt*2s?7GFawcSWjRFY-ssqOb}WF3y<gISE(O{
z@mN|{2d4Ru{5tl|NL(de0}s+~>_8X<py0Pg?KV~>|DF9Ek@|W0y^oE84^}qXStox8
z0-b*P+44v4&Jx@xibxX5@rMi3!YQqKoczf*NwHIqXE{K~-SC^qBZ+WpXzOVyo=5!~
z2X26FTBQbpIIL0in9ZUL2J4@LPY|^Z=;d*n2=Pz(OOk@-Z>Q(Lh->WjMDz?OmT5IC
zK=Ri#11pTn6Ya_L%+C1LFesm@(BPYHL?%HnFd++h4wdq>4Qwzj%zh@QL`w(sLA40y
zy}#LVcDGj}OG=(iqbvl{V3aTx3@3js(XSY|D8P5mr^V+TBU}w?HGq}n@3HtZv9(z{
z<AvxL5dwL6(LYoRSwE!1gA;x!IBX0GR49`Ca;$&9e*NSXYlg+@!8)(TPyKN%*=zBS
z;S(;G*W(`uc$YU~m})o>_+&vWCbYk4D~Q-64di>vsZBO!-a6|pzwIC6iLN1UDAP(R
z@(-J1CRc;xV&*Zt6M8;S*@jlTIZXcP-&?*R@>&qx%c~x?%eywlp|D0~$$N<+tj)Y+
z=Nd|N9A(?({TMyNfVsON%@i+}4{XK9O)8fUtu46efNA~9r-7&yRTYpHDIWzvN`TD$
z+h3p~fF}-7m5&oI6r2eb4^G>xOa2qg&Kn0%Ot7NIo#MRagEwyxikVMq55xl+fkR?7
zo-c$MB9(5Lq5dH-N`j=Hs#MIRHxhsLu)kh+(^UC%rNar#9#;?fXDUsPF>J~9I}<zz
zmZnN}vS+1Xel)&4UIBw4E=KXTMfQqzt5FO{kD9$J4S}>4O{d?d(hx}OMv$0p7uE@f
z{v`YQ+w=p{Wx@kvzrx$<Let8-nxDOYrJLo@Y^!kZfPxnXR5~~~61OKF=-)<C+Cpfa
zhQ}XN=m`W>TSX4`XCN|qw=?7rdj{ka@rv$pIn>|t;5H&2a#;G7r&I^-5#At&d+*dC
zgP}O5RdPfuR9yqIIG}0ahLw0!I&S>c6Go35HBpXqYFo{1Xt&S_EbfRp^lJH>w~rF@
zpdhPj5-^1flA|gW6m81S$jZ@`4szzQ)ve6LKgKAeH<`$>l?sWD031D*1#^$HFT2}M
znIgvrDMvL`$qDviHNYk$o#*BAK4BgT%nDq9ljsXR;9`)WxDeloK_oC141H!k%SmZ@
zH_?@+t=Yn#Z2UT+X||k_*3lyH3U~IMS{MdoB7HgY<g}Q>sG)Lt8WOh*_aV|wk$17s
zx6Bzft?u>-rd6M5{|2<Me%C#Sx#|^b2nQ~Lt1N6z-6v<og5km-mOI25eYST-uQ|QF
zJgF6Q$G}(v{u~=D<D9E;njn^;P0sZRFsz081M6d)7u%?e;Rh&sF1^z*wA7oMRh`&!
zed1kQI;oCB>FX`hy+qw>E%b?gK_*B#MvpMcqJJU{FP`&|Luf86cMK<#3@EG&O&gsJ
z0-ymv4NRk9kT5Ya6mNzDZMzKeUo6~;MoHX$GB`${Zf%OfP8nh=JZMmVIlt1x9F`vI
zWG5UoPH21p@(6gKhy}=qiXU=;H)Wawe#c(BFnBfBS|t}nUkY<6`XLur_UuLTzkS4n
z;lt&Uv{HI~6~?LA%~WWh5v`C*eI6aV3-_m*@G41dFM-&DxP&unsC7OmpFu_2(lk2c
zB)b$DW-nZZa17)_2lc3=f~8o35>tg2X+-MPY6=x#kdZOHe703aMFYU)S&D_X1M8j4
zGP+WIE){5x8t7&J(;mSU1AH%!SwQ%bQ#XN=#`xO|oMCYca+*OlWkx$L0-PUX{XG-u
z_!At19S;q9oVUti8_;fX1aFn`{)`6;Q{j<rDp%OE?)Db=BM_5ZjljwYv7IVWXAbHX
zL}IyrY8|n{N2g5mXAl(BfpB~;kV!!t7>hF5Ux+Km*CJT#$`FCD@UYDR6vhBxkReeI
z%Lbe<wM+z?UdnlLjZc+l1|8}&&P{?}8{4j=g6XRW__)sBRb&Gr;|XcN^Ytc=##RlV
zEmPtxL%OIPW3j0JOie7wI3Man$k4-S!5~Z8VbfwPpaTP?Ec0$T&Vh_=qp{=jQsI+`
zV*=3tQt2NO|1;&0Pz#+yRYkv03&PDl{!2D>whl6cM?AY+h47T>pydeE-fTq;s~Y1u
zLi&?{>DSs|Mxl^Soz1%y1;J%{P@EbZ;ufe?&WJAoh60VOPoqJFCoI4`5SvHw7HP1C
z#zP^J##rnT$mwqirpae#BeRB<ce4p_T)0?=8`iUOL(GKG1m>^}I+JNjIse6$mCDXS
z5S+JU<o%a0P3Whw9o1q(M)k8g79MR)B2-gAaq3&ieTrDb?aIuUHG+m%%7+0Pvy4EM
zaEr~hzqX(rU7JlS2W7lo=J+ZqV#{3zy5AV{fUH8?iHFLff{NVaZz~joxuK$;%nib5
z7#K{KtlXUVi*}g>f|#~vHvJd98E${#I!nP>x5Nr%XIG-6$E`jFY)F_c1a(0db@JS1
zD}+O>QEvB!)LcYIC<H2B3h7(QI~=|oT|qHo3-p(K#ah^PKj{C})B+t4+;B_fjx>B;
z-(&;lCZM8r3r?#$gFb$^qwE2w#MIEc(jwWFTHA?|!;!+OP%F^$*Zg-AGvRl#Dhyi?
z+WGb1c&KXdWw*&UZ0`^AKph}FcgOT_vrDzylgLE<mW+eQGX5#|20f9(g8Zoae4bWt
zg(uHC`KEvDdk%BA!t0&#t)L2KO)n3@;BSWt42`0`mGBbw)%`Kx%d*ulKilLx!40Tm
z;vtm}cw@AP>Ho){@}NKKfRF}y%y4q~u0I1%!hzuM?vn3C`@C<BvitkdK1|N$I$n<_
zKM47V%Qt~vz!d$V(WtJa7Kw@Sqcj=29F{j}{5a8#v0ufPdbEd@256d~)ja!J6UHEe
zbYIO+BSmlvM!Eu^$^zR}xK#kK!9s6-5u#cs%yFRLwXJ}w7o}h9GKy58L>7CaECG#(
z5#{xcH1wsk$l!ZkAf5h9FN{~xt(7Il(HRufl%@9f?)K~E$})eK{Kk9X>HJ&nJXoAo
z(1AK)NJL@ujXdPvP>*DO1T-#LQR)6Lt$PV1O#L7$qo*s&WtGn>Vz0JB?X?K1)v-L_
z5Q}G7lV-Js7Uu3-;ip)edR+#*i}8VrWL>OvJx@~9c%LEbjjKmM)^O=J_*c{x=(Jb|
z`fMz;4jshy<)*aUgwD;WreL;g4hFMwnt9n`J(T$+>|HT)oV<-FHpKPM%huEfE4Ofe
zER$_cSJis3ev}HuLT|e<CGrNC8nvS^BCIS`QO2R5-h%ujMnp6f#<l!3k!C4&H_?Xm
zveQ3TL>JE#-G%U8E4%#dkOi9|qPt@WV5A{H1}DRAe~XMk5rwKj^0PSZu7up|iES`E
zqp@*6_btVO<XQQJ{R<qm0%!X#<J}hIPa>c4SKc^tMMwjdh52=?<*+(jWr&!+Hu;S;
zg8D<OlKR_9jlGy&A;0s_)t83PVerZC{TZ+7q%hqC_~Q@po*t(tpp2O0KSrBlGPCmX
zu)Tr&9Y*-EI!Uek$!2*q@Z=hKB&~v7EU?9eXoox+4MEAz%46xhg2U3|1MOVRW$Wbe
zSh$I%YRX>{M_2)Q=)Nio;N*erxmBJ>i%3%wCouM)?fbvR)J$jQpYo)?t?^9UTQI>D
z_mZcg9g+oH1V^{>(n5JU$Ofkpy&$l-<Qad)-9iT_wUEE1_ZC7<7*P0Zn)7yr8JL&C
z^RZK&^Lat8sUfU39X%8}yS4Isup8X8C=~=3%ioPk^&t5R{;{*X)+Go`ipTehF+AWh
zj90*9<R#ma!2N(<<m6?4#q30$>7DXQn!yGD6qLAK@@h1Na+id}7?A&(b1qb8u*>T{
zn1G%&p%v@|q#U$bEN=wY68qPS)a1=H{-Pu<2<Ca{U;~!vt+Y(knu{0X?KqKh$g06r
zCH|s;h#lPvaMH3cSC_mK8?j%X0rJmaL_j8u@@`BM2l!?Pe~2=9&!@!DsLk?zno=R>
z$HFr~E@0gs1he3Pp@=3TW=!{?w+%J{BWgO0{401Gv7Z%nz(YTZNiaSIk%QivWj8_%
z!v=y(fhq8DBJ|qbx66O*sOU4K|3LX~EDLN(${rT7m8u@QlBCMujZal6w1$i+lv~)N
z%78u7@@PRWmrwgMD}W6LFA#yx_%kY*#GY}}rR-UyW@@=tz1b^Xf$GY%lhsi9_qM*e
z+xre0Ec;X`gr(J9HU4$qDmUhl4BB7zv*sXCpaSJ(|EdT~WV#T;99bnHoTMw{fGP(+
zVP>K>CE`X7tWq&|>5Xzw8p;}I2PT3K_Gc{ms8Me4E*=ukATwC7i8DRapW)ml?t&gS
z%3=NtK|DAM;p>IeJv_)-iXvXXc8;i0u&#78@)<vCQyI-Xc;tZ`>7QHBheecpE<Uej
zrS2*@Drmt(c{#euu+_+Sqg_H}cfK4`<?2FsGk`nh<=AMgxfQABc{$EVF)1g<S2?;0
zL$3w(h!m|l!D*{;nR<>>DKrd$&iBs;ZO6k^445!4Uq}myTIDC&REG^O9U~{js#UjU
z(ecSu3RDk380RTIhYkQF%#~HoI=xA9YBXPisk8o#oR)~boY@jM)|SZW)@5TWse6XM
zpqLlN%0$XDs~jiK%xr>{$e6cY@u&{BGs5qUVBNE9u&DmJK+ZNYWf0A%%H}zV4$C0d
zDJtQtz`92g-?_FZnHpAilk<E+rg<zzz~wx!If`9K@5CP)RSjnxV;_G;V8?4-N?&Ua
z<6i~|*e?hK5l3$Tutl`TaHIa|+a+qm@m}r$@eI<Hbq`m8R|gukhxG0%gMy~Gm=gEo
zfTIT6j}=Je8R9)rcC&Z@gN5-Yu&VQ&0-O_L=KjY3mJ8xb<QDP_vK$DWR2TY}Y8m27
zrq>a;rLPEALCQtxOUpc7pn5vZr;D8mP&ELeU*hv(+srIj0W0v*G%DVSNL0$wF3F`x
zkdYGqMsuB@<=zgO4fSuBK%!v*s+D6T_=YfhsgH-Zaq!Mh=0%c<t7Vum+C*Gr4_MDh
zI%tTD@GqzVv-k=BR82xg`V4q1JS3y44D?*oUAW`PsA4rK*gqXG2-=j<zDyRpgg@o7
zShxx}ahhgf;V$=nsgSLQR6@IcjEr$!Mgco1a~jxT&;GW^*w~DDR8X5G<D4X%ETLZ_
z26`T0Fm&&a50ZSMxFMrGuPA(pWykl}5(Z<~c{U*iU6Vl(2jUXtw)0e#$KC1~M7mEj
zA#};PGAX176G|q>AWPs=&B&Fp{u)S4KHM%>#WSoWgLil;U2SvEWUDfAjSpP)6Q7b{
zVb_KLXwKx2TmzI!uJf0%wZQxYxGC0Y4qWfQpi?lId1h>WQ)~d;?e2nueIiqRP&~aE
zk_r!}O!F4(sARe>>#_5fjK9rt{F)yO#;>pg5EvUP3c-A?G-ew-u%A>#OS^BQmq`s(
z1u-NWv_|}--ELQ%$g@U6i&!VW##hV^N~mancA)Y=tXaLWTK}Eu7*$gyfEIbAq%LNR
zGeV6?FyeG)bKKtwg>FWoBZUYR2F3tTsj83DXCf=z<)y*LM`s8WFU@mn{2z{j2tgYK
z-*}U;NT#t3CRB3_2F@-pmoDQCsewY$4@301GYRtM{CCycxy(>PC{ua;D9R#A-Uo%y
zc$KLH##*AEus|{R)|iamtCW|S#&@GDU`WmKHaL0K=tHdrnf-}pi!tR*(&o=>%ka;Y
zGRKGs^#(|)+*s9f0{j>lC$B(Uhz&-_k(K%w<YqCU%jKqM8x}7lAc(z4=K3>!$r5AS
z8#fyt^+s9EFUIZ~Q&-WJky~OmV?am572IlMo^su_sF-$}O^qkF+-`s8IVxWYJ}q~}
z{L8U_dE3EPeCTF|;~3NgI*Xhrtmjv)y-Ferclgqn2bEQAxHL3sP-_F?{7xG)?xRvb
z`L19z=viDnD-ea;R{5IGoq9cC`7RhRUytok!U~kgH<NEfcXTVbbbuR}?240OjofWC
z$-`dP9H4u=)x`*AjKGh3(-?4>JO)eL$l`(Sv$p|S!RgB|F+hpm3?-`)0w-5j-CBd)
z^Q|-t*1V%Nz}S+7^j~5yi{;xf@O+!cQ{0~pMy-12-(wd+^H=c(VtBC*Nbd*I?!+-D
z5Bh*jz6UV67Ru{)gCmI7SIYNdteG6Fj$9^JhY5T?#sYPffh1Y^K`;&qpUeQr{X_e#
zhRT@=D3q9%9|d1Dx8&r<i80qfow0<qc3;`1dO#ahGS3Ie8~I{oH6&i<r<G+5dB(YH
zlm$)zRebdskv=6L1kwgsm<U`ba^xb7dW<YeB+(H=H(`_((@Ph}<Z7~rSj$UCntlUl
zLU_M3o}e~6UV`{v;v=+UbfFqzORIX0!!G{66ob^qvMgwHIS|j~Ap|N~ftNhwO)_tY
zOg0~C<BF=D-V_WLKU0*WjwcKnF=vLYa<TkgY2AzmY+EU-f_tGk(RM0;TdSi>*m0~7
z2jpGW*mh^&07@xAi(XqqI0mWVn8kWBG-aI$8sLSpKJ`I&8ud21BM`VH8xrF(^3^aq
zyB3^LE{kNN4OQ>mkWH}^<1;AWk(15AtyQfJC@Zxk))-QNhc{)Op)__-clqybBrD`m
zT*zf>Oa)sG+9UQq62jpA-yqxK7pg*d)}<};!hr2*W|rs9un#UyhwQMyLXvXudHIR=
zUK{~lTn7B7Hq%DlsU<tpRJ2(Dn-D3erN^rs5VtJ5tZ|h}EI8O5Qg{^LG-fmIHsbTj
z=%1NtHT6vCF*ytCe~|n<SkyB@h-UsG7+FV*vLo_K<3SYRq1^wK4H3o1>YC-(sR0yP
zb!qt<>i}s)mGWC#9Nw7;(J8<4*R61kW-Q9@eU^~#YO`X=K|B~9m@5Ri{K0x3hRZyW
z$@@p^eGax<Rvu2hQ?+1-C@Ll<q7|1g)fyI%L+j;{|5>X7&i7~<NO4>Qu7HFj;d%U+
z_r}{87{ke6#Xoy%7HCACpp>G-c-&||DpMtYu`$CliMuYJNUgA{FEPAT^4CNbl;-CW
zLs&l<b5joFg0GaP(tE6~LdmvTp7v+zbw_(FxBrYk!w~}T2v8I3w!g(QEjhsJtvcmd
zAARwGv5=aDJm+r{Ke0I3CGvcTqAO`7|2^7+#>AA#1z4&Vj3IYDfYCjGytWrpD;Dww
zG(=u<wV}SLMPByb^?1WL5#lAU1VvE_YaZlH;R^Iu(`cDF7k*+IAFuf@=0)KQD8sn_
zUiUS{Gxevb#pwD0V`1KKX1{CKZJ-qXCvVzfb+^x*dZWA*vIaJCL$*QQHWJY95699!
zyz!!2eLivj=sQ8MIZX`}^3SRsyCWBFFdUJ&8eZ#&TZ7*%5R!FCp=_>|_o{l1|9?0p
z&XM;Mb*%^)!a?MN7=7O%gXKf(pI%@r*uP9V04l6%Rd?Y>F*!uMHzEH0-_{=Wl!jIS
zWUZ)4`ms}%@#dYH)3R9q?3Dk++QAB=ghY)b`L7d%_!$)#B5(?Hf>(*)O4hJLnkl>i
z2l3YOsjP#I(8aQ-!@oz?A-)+jp+4<j7$0yK;Nq2RBr<PTBNdoABoBSTd-~f{7-6!h
z_}I(erUL^=2Hvyej^(m<HsOT~DIxZ5<j2UMaLPXZC3%Nl0*#PDD?4Q0tm2Y%LsI`l
z6~?RsOSsV$>Rc7Vf%eb(a%Y$T3D*&7os-H3WEFFqos$E7w6Xihj|XLY?84-J2;x52
zNP+WoDUP2*e7Lq*_z|6iEYroQt&>B&IjTyuR~I5;>aaw<`3UNvu<4Y;eOi|B%V{yj
zp4;e%M3931SX4MoKAY{?4_pZ!Q7d!Wa55ZeBjhbm_}AyM3h#E~l2c~(AC=arAWwB=
z<Y=2E%7Lv#SYleBRnTd;&{Pfx*BxuEBdL{O+Hq-dsnEfR3swI3_!dwk3&JqT-;Asi
z{B1Z^^t{yDj|<Q6wq14i1%F1Xgh@B^(CUft44!a{5#1~&c~2DkynBOny{vG^$w8GY
zfYfYrzD-U^)0@{Ir)Cv?{hv*LT2@idv(7t5PS0A*W@TlQoDpjbDL`nC_YUCV9yM8E
zp30fVJm;Mvy%KThGq8{h6^U||_1=ttiz+!g8&)G6cW96a6}A)1w89?6nR!l{FflpJ
z9h7tZ8Q)_b{VMFQ^I~{3M?eI!Ac=11=RE1{zZb=RMsI3I^+|(P4N&DU5x^bNH(G|&
zW_TB6iss;g#n;0=?w3YmRWngGNBa9SZUR;*!=ExB_RpxHlVxD688$rueC!f5N+n=W
z+F1yTl)>@Q#B4@}1QTD@it}2|kJX*(|B(y4Ti2)+y1dZaLlpq<&qdKMkRv)1<`X>K
zeH($yL;%9RjI>~(LISv#gcyYNKyJU(-()zR6`dg-$2Uu?e`aYTIfZ%JSs5DSQ{WJ2
zi)xCz+kp|=tEAM&O!3FXq^KDd-$2FI>zH((6y(WpW6_d{BV~jShUgh@Z&}H}78MyI
zgFFay;uT;y8D$fJRX`9XFQc8kb1fOUEQWziPQoBmUoQ7&l(K|=0A$ISkgs{va-*Cn
z%eqg(suSC!$NHCm*MNh=+<+H%oGpaLDkD8P*0MC$lxdMG5__hb!7zwIqWFZgZpKg@
zP%}(smU~x}mIQ%$m`${<S{w73>a0w%A>V+DCCwh3Z1aYrs<&Jjq{d#_gk`6e$IXfm
zzumXO8*o+Xm<&UmiygCW(AC~C5n&?!s3|4a_%oKs1}uV0uI9<Lu?`4{uhhh->#WW0
z_F&xjf4Sa!v-mf&eduyoD^t=h7Pt;C0#*cjRi?%k83VXvS`0wJGNVw&KP@(F7&Ww6
zDvS^e5R%G7$1Xi}7ywq4THB&N&}HyDXPsamf}+X9{)&^Ps$)2%lZQ!-H{eEvM)K9h
z%8$i;8n_bKN4Eq>D{nN@qPvVFmFfPm1v7U*eO@6mVw9LVh0!*s+4a#0h|WC=p(qV;
z>X&<2V%(E#Y|vBz4H0?z8ENvKtvU-YVbgE`7}3q%qSZR)B_12)hBO*gm|56Qa;XJo
zt4HiJTuG8IPzk?~RvLv!LrfQ~#aknC#xbL=iL@HmcDIkjZ74D`+jESb=>IoBLS}g%
z<~ZSrgz>BRr$n14v;B9Ti!@16P6wEkp{8v<1~M48H(*Jb6U<5#6KS70xiQwWtf@k7
z@(C|0g!;O)nd`q8qQo7-m`<uLH+wfMSwiRkMs)qfSRRz*#0Bb<s<(ucErSvaC*iH}
zZ-AZDU~V(&&=J^NE4LS|1|uUk0;T^_h^8!Bt${iGW%~pRvl%=2D`^~^s?6iOm9Iu8
zqllHwk~@MoF)U<W-1&)Kcv|d`yL_Ui8j{<vraTnoYyP%@)~4%x-JjWo;D+9R!=DlO
zWPuWfWZ(TiJ>jArxyRQJeP~DtH7FW~5kYZp^oPAnqYCGi+!r*s5=kM9>-=UsU|2QE
z56ZXf>^P%WFZs4@e`avW{V{V~^3l{F-?5*D&u&KL!&r212v0^h*1;`2?;cFEL(H?f
z?v}}SV`)?Jt$feAM5!$KzR!u~cvM>D-#w>5eh_m)Ada4pFs*->?a_2A9Bz6Jg1#!E
zDR)pv928^@x*9X`BkO7=u5kWwuyQE?OF)4cY4?eABEyttoL&J(w~EY**`t0&L$l0}
zQGktgcR~?c5c8VFgAQ)88d(@T53evv{DBGV0P2uM-hfr+woH@7{>-#8Di|xJBRyk6
zmsU-a&KOHY6YnrsVy{rAK<y5Tt6Z8sW6>-o1gwx{=^5`}<2KwX%R@qdOV_x3vIG#C
zYY1UXyzDh-G)xI{LRO?Tr=E*ICj{NtX|giN3Tw4R6PZ^Tuawahvf95h#0cv{(1kFs
ztnu-2$E?Lz7zVsH^$+JU7NiSda9W&4*2UWpG&Jzw=8^SjKNM7D9h03uvcWm~q;q9s
zyb7VmJesv?60A_A0R>iMQ!M8A<XPMtWFOPkh>2`T1TaKYbEehy?22D-RMBuNXark*
z(40K6qBS_Tx7oAq_U!E1I^xmbo9)&W{tA_(R(3>#P_Sqk^G{PTequ8kIdPQyH1;uk
zp;~Bqr~Nh<S0U_5%d8%M7(mSaTICq)HtDtnI&&(Dz3jHX_wL=duly{CjFB?>mY%hq
zIE3<T2zL2-^cvAGzp$U1>QTlvC%?3_0oR(~Apez*0!sp&kN$32HFQl<vEMGg_TIR_
zH1T-F&TnD_V8``XMr>{QZ7f82EB#l$i&t<)g4glc3i*9J!(l>Whi<b!q~^Qmjbu`l
z{4ucsVHnkUlrWYYvrQgO?P%tTp3HypZ#rOg0x3s;0mI}p!x6CfR(T}8LEIf{tnqr_
zN6q#tX_d#~v>bt<|Jh$<^pRBpS|^WZdwS6&K!=+}IZ|s!<xc)-N-dNGN6@fww#Z+i
z6|5*tC3s2Y3F`)F<8T7~)z$=e-e<8UPp0vaamkAF<*A~5M#!BbXFlX&dD`D*rXpDY
zOC!(34#%Co)yOEx%ipZK4A7FiJZp?s-<Fr>{A*nm(2BbE^Uf0}=7vk0;{^x;mhnXp
zyS(7fDE7Ha2qeE~ZI+ack(a`rU~wMY&7vNhm;D=hk5~a06$0dySSXlLK(t^!8NU$%
zAVa)r11V{Sppw^usYYdTRc-Qm?Eg$7uA+yFtV-TUjGt+oA#Wy%Q8HR+f@w`}naZ&1
zUIcsI2Y=_=v3t=eohHnY2E#+h{38weA(|E#;6w6GOdjR}4dZQ4hW?rD;e`@Nz0%O>
z$o+oTUtQGANE8aXZStN!v+M*0QZA79qe-B$JPdJg$_GvxmS2hbM2j#$0vC)e`Ov?K
za6}pd|4RMw)&dJB{wR@wuoLO3fr#M1f7{zWXk(L)y;ll+#%MxWL4Oe5Trq6h|CsD)
z%~c?WI=!1RR`Av;-MXuR7D%-ri9D`&otfo9p>T|RsybnVXg@FxUdzQ9E|op19km=;
zSHT1H(}nkOKJ(y1+&u74MmSMr`b@RrJ^lOjl|AjRjoGa1<zE?;P=s|W6@XAyJD^rx
zRx)Oc>{G2sQ5~~;_pNrIu&i{1>{o5LDwId)9FTp$`}g-}sQV1XoBO^~4zLDhp`eHy
z=zSOYzVIb+S;z&@@u1Wu@EK5zs@TERrX*2W^s6xNHWhz`H!#;!#y}1U+A3$6jJ9$*
zhZ+F}ppf8UF^Hm@oe22Aw+U8UBZvDSxWJ5Y*$(UBh-!nhAf!BMlqNOnJsIo-of#71
zv(`yjrbdo5Dz-MxXl$C*D4(nDVL~Wi?1ivCkMd!{rXes4>y8=QG%1d*cC>St22Nv<
zj|q}e6UW52K|mZ^UC2d;lK}OT0JR_IUEr<+s5!@1J2+ep#m~xFpAf{wb|QGb0v6`y
zgBBoY+`ywwCSQ;*M5EBe=gEn|fKWVJ8ye-L_&4&?03gZ9sWrPOh1$JsM4R0yG43o&
z7bE(LH(#D=Ooif7nC)pX|6H~Pb_%9-dMqiJi7p?6dh4I7<qR8Jc6KwX?VcH{Mx6{a
zrSjX$n#MIjK!(rqhS&rHJD8$v!#dlC4JnWee-s*o$_4|Sl@k4T%WR<>qVGSY4$HYl
z2F;A2;eVd@WKzX6WfPpe(;#_l(>&cEef$}7=@`AS0z18L5E3e7)KOlHWkc>RqRci+
zKV!6u&bdzd8>3YqC=Ug?WI!x3Y*H4#WCi4bu}n2E7x2!Ycn4AOlx9#wWpJVa<zca5
zrX-gk{tY1uE&xl2o^PWV0hH?kqcmc&zy=pae@)GG2zix@g7RQF-tWPCiR~o2IQ44s
zSBsP=08L5TA(zBRK`9>M$YGF6{TWvS#uen<LMid>X);Yu3f)!=ms>>*nL!CM)W$os
zwOUH6d-lM7oLwrzOg!ycj@>uh{*Gc6H8R5Js&@h)vUsG=A<is9C5sjH8RdO*+b~y>
zFeqn7wZ8!KjrN%qOMNOAK)ykh9dcQS*3nSWFAounD{K*=IHuY)t52AgsPc(^7-?>X
zhx(74qswe-FtG-@?pWW%#8B7BxU`a-gF+-S-e+gNOyYhllmcNVSH!%F1Oz!^wM?jX
ze^djn)&g6V+Z2%5gIzr_eMeh@JxBKDq@w;AE>2(c<RoUWFXhS@eWn3`$W`7FA{Ibd
zo|n*u^gO&e_>5k><`%ie7WK4K<XT%uEMARV=lxkky{?j6?`^utucFmt!4%_>Y;}uF
z^$){4gd<$0`Lo5S2||^Xb>}nw49Rt>QBjUpcn7A|@o=Qmf}XI-APf&TEEki+aF9XY
zs<L)8WQzsBiLqy^U4udnCoU31Y@7y=SBG@Z&r)~?1h26v<L0&*QX5NCnnTbQp~&bI
zAPb^Mu$x#Yk2=Nl5XlH|ugumq)e}jX;jdH3*)>J9PZr@YuMhd$+6=ggwXhmu_E=(~
zCy2w5#t>gX`5CmrWDGJ@gcmp20*~cAiCJkjZZAije%5ZjA$3lq&0?@LJ4bS97L3)V
zw07uG@}Ky{Fj><pTl|;eeW#oxtx5Jow#GcP$?D$Ry<25wObkoSJgZf3;mnGSiO3q{
z<;_k_P<#b`lwfRWGlm>>h0L+l>25EdBR6`(ar$qC;&fBcl{u$ca(Aw`U7Sn^xCPpW
z>MC+`&{)g&DsSn)9g~$W1~;J$K~t`>jp+%u_%<V2w^DA6rAI6+P^6kx6ug$(yjiIG
zs%`OJZDuzzD3C_t?f$!KV32@_e8)<lXe=DUfGTYw@CeMbjAi+9>}Ggg7s^+Hs>&3s
zDzr?#>TU88RK~$`b?tmM4e*W_D1ycSPnqF#XY341NNsWe>8>E;2wX_gN-g<owxh2e
zH%Y!;V6EM^JEE)(Mfrvg%Mf!Of1Bj)SnNJ1{?}XXiDgrD%~HMY^=FLLQXdu2*!z4o
z8nth{gxN5RRHe0iGd8CdJjM?CmWeUM!o$Uc{qXHH7Q?mS>XiG_$irNL=+;B+JKmjz
z-B?}!Vl4H6APDaRt!C|r2feMhO_drIWf$_V<6VSK<y*16<$FfF>Dd|beOt%MTCC#|
z`GF6T`wC||PQ6v~!&s+0iaDY_m>Rf0^4<&!4V9cS^2hNDHD8cO>pSb^U+p2-n-Q`C
zc{I=GNQ1)gd{%<C`@2|_V7~uOk-)@`g@(<cRaG`*f%inS2!@g#AxwCgFAGBwRW!kP
zUKAsOfV5m79P&^zHp}8PT^?kn%yszNH12o^^G`cOIzv1&%LOtG+UJtQ6_%kw*o93K
zJ)z~ZG!`940rwIsEiMZN91D6j$#VN=DT8+*(;qV7a@81FVUuXhVIOB@Wq}xK(^99h
z@P5sUkySxloZB;yufIApkMi0kSrfxB)ezEryR7wROJMsjqhL7%;W~e2N(XN~>6G>U
z3_FSkwN7h;52h&GkbH~HqItZsG0{+?^}*4s!^R-0O~H%U^qMOqn-dAUKs<QO6j+<*
zM)U&N604{D#N*9T3+eJ-Xd+X7?!Y{3d|T5z^i~M~QC+gl$f+WKi)@cKfYL34jE84K
zcK9<SKZDbthynlji9h4gAvQu)`=7=$l-&U&rkSie6Gh1Nda$XG2)VuM6SH8@65XN8
zp(1w2LX81HwL1;VYLRl_KMTo%AWDVtd*Z>2I^>c8pZ$3-re5xW*w7aFg`G@gC6^DE
zUnbUJu$w9BCb|ehEtX%!YUx=X9)B9?zxLtKCT8O0R)~n-7{6Akt(ceJCSo9W3dCIi
z%j9?V3>O9<o^&R>*pS!1k7d>%a&}C$!c1{gFOoml@^rVCwL-W3W1<7G9kW#)j-DWO
z>8nGAzWgc34h<->*wnhyw!7pJ*Ww9taqD+s8f0W34sba}<iVrfGI3u*kW4s#EcIr=
zQE=tRpVO#_`%v^PlE?iS*$-$IHdu%J#pa@QEL!9Vdt-DxixN=*1U3E|E$OLfF4ia0
zGlIKl9W41&>}upfWWXO!`%)E&Rn7{i+B4}FhJ)}9Oj=f4#J_pFOaf4L1Zw`QwVTN|
zqrN$!d!GvqLTyDVsW`u$w@Fl^-Xdi6-+lKGQ(P}Eq@mIb=s*m}GI`MlU`7)E^mGz=
zDc;6jGeuylyc|LhLEivUUh$7%EyGSlP8}H4kk(fnC*mnfrL#j<kVWS~A-=1*e<lN8
z3pfXWF~7H4rkKTO!RjW)UytG7jN_d@wA0>*;ZUinl{eFHxSOdc(bK;Y8r@sL#_CLg
zt_M6H?1_O`<8MbpD7psw@*gJdypj-C{+()1p#{KE-vo?D`Qe`?1o~(};&;6d3j|h$
z9=#J<pux3B-m}wt2s|n8$8fNd7+cHiL*SSgE*Ib-`M_T7-?zVfXn(Cj5b?j90;cWh
zkt-jCD%*yVld-F;ekG1);+9(L5<+;pe4KivZ48kMU7^TYmH&j0ABQVanyOHI@4rEF
z9VZiS){E3A>{tn105v9S40fcXp*Mv7vps4II3@m1NH*%0!kY^1c%6K@#vx5aNPx1R
zsWAlBG_>OIh&{b==(-3wu=-(p)hGZ8EF&X(*SO7#a{1MCtMB7K*5e@E*PA6lsW-5s
zgMrQad9%2t62H*rvPAZ;F+>m?13_?r|Em*qJi6pS`xnGw4vN=7HI4}~@Jh=e9-O|#
zHjqPtBJh^-vNR^l9_nLJfrzm%wZeqQVKs(iDi`AIVy34XM(Y*I;i=7eJZrTE;Sq^A
zyzm_-KM=vs*7R6>aY`vNHa*hl(i^I?d@g;b*M$MzxmAuz&uCy6oOHB3Bk<1qxKL^*
zC&&0a?8YBn{zyM69c;(eD1tM2vK$wy%Ot$K9PgimBxi&drNK%$!Pz8aBDbt9mIYOQ
zJ`IZd0{Qod?fQa0GwEUK<cY?Q2xplvTuzFQAhQ_7K)~(fc%~T+y7%Q29~M|EoV9BA
zua;B&Z7mGz(&4lkh3?Ng;|w`Hm;pgw$S_Y$GY_8U)OtD7pIHqbI?cCBuk_wVkW2-S
zoaN7`4H5y}0p#rL#7b-7!8*sfEp3`C=hpOCDOkM5T9;PXfRP%33J@%%cCcK|i){h5
zUz56fNAHTwt4NYQ(Yv1L>LHN6{*01=?sS0b($A=yZ)LHG{x$CMxULufD+8h@#33Sm
zEh__kO<{a1^OK#irt}x70y@ZMGy@prU~3rn4YW4M5POd7GC4m;0DyI^To6oxf6e3k
zz&<<>J>FF=^!?)oa6vgBU7*-#7KvOG`ws@PrExQ|gJp0o_9;WO!NtL9svG5!^bG&P
zdn@6{x-@1AU!fLHC@IhiiY-l{9WpdHaRS0ZN)w0i&I2kPbm9ZhkYP62Gfp~BhI=bq
zMEcBtWy^?YrSITDGSXRK;zSwc4O%cT_7KE_xk5+#FBmGV3q4sm?Co+{EEpUF!1(2I
z69g<6En|Wxs-WXEF{}kBV_9rr09F9<#`+o<r<tU(jcSZM&Ktru)xAWZh}7?t@jkr{
zEf4PyW(-MVr}8o(QG33kHc0uGa_`al!gp#rpnN2m3CH6^?~(svOl%4mmPv`AW@s&k
z?J_wCS`PK-N`K*Am^Sx<FYkppd6f})FuYM$r@2s(w^M^$H(##tF;V&DVa*MXKvf(}
zoa-{V)*89~`itc{qw)w+`Fih<PzJLO;1ZW9sgtEJy{HG0V66Wz)tMitH6X}oK?cf8
zlbUG3^2_M>&D6N#32-Dbu}s!iq{g{dSOuHB4m(IHeXkJd^7ufdC(%HFNiOTEe84<;
z==5XRfNTi8W>j02YJUT}1)H2!`~s<oT??IZhtwuo6&xnTv72>869W}tWYhd6C}Wn*
zbpKuD8Gd1b2vZMqMP!Dps4C@BZ-gHTxU9hl4}C2ojWOojgs~;k6tnX91Kcqt&Gtr1
z2DdrZ$PNAmjS0H`*FyT_5}&0MF|Ib`LEhh{yT*>KP0|vxR_&683X+#p3m}NLRa*Ub
z3Pi$B1*3+Wb)C!%KJAA}>$4J(adS7824$&bcI?aXK%VQR?GtmNZnakCL>r~Jmi5LM
z5WB~eIzUD{ZB07dWJ0YGDlK#wY>v69#TDq(0J587zn0)~Mp&p{O#H<FZp?+)9oTxQ
z+>)rJmzUrfUEe&BZuNE%Sp!z*ep)8C*;-Y$VkWnzHs?cSwOwbuf61rdF|W`iS%E;Y
zZE(#(`LfM-=(usXO86_W<da|>$yfbTQ^$GtF<}Olb1;?V4*Rqms(fDV46P2A0XCz4
zLxbEE62G~=72Z_bIvuZ!N8uoK2YfvZn-|UCq8@BG3=X*LH+;et`=^G8+?}2=wuEks
z&2mp_!I0fT=>5I^3*<b#1IkinS>0#+jxyo$%~%`c6|&Caw_?nrD-ptnt@>^6Q!5gI
zi(pyF{i$WW7f-DY^2P@Fj&EYbtzr{DU<<@cALYRqJD?LB!>DS4i>QOU)UBXn!zQIt
zNfdIse9yQQ%%#PEzVF=@xL>QhV=w(6maVxKuKFLw9O^UZ8aD2aV!Q?<0f)fNrV8l%
z*j5m74**<y3?44g#WK$*gOmgu*jV5B_UiE2u;gXICt6&rcsUh80FXg1SUL>*hTS2H
z{231$t{WovWwFnl&v;DeVM73ze@CoGM~(aYkthyboSvCZZ>AGkGfVdaNRcJp44H<k
zvjW@Hv>P(*Qdw$qf}@GMb6FZ3ayRto?~vs_70dG`o@)x%7I`S>SPt8m#i>_#XJ+<#
z#Uiz=0KUTLSs6`XFbpkX1|^JG;3unMfhJ_(KVI$cnfS+h5CA*^SIL?hw~+&AwHs$b
zMFT!V-y=a|?hD53m@>N*xw7jLDd^-<;|RTPQB{IbR_kN;fttvn135PMIBX3}#efO|
z(X!DNySsgSNvUk|*Jyje8ZB7<vf1g`0o{_ZSCBZe#kmMK<3T&@vaL1*<*T;F+-h5q
z1P%LNt43_|CQ!S5H_N@Q1|qgSdY{--FFS$}N)b_EC1ii%1EO1li=_!`I7YV0Ph)|1
z@{SqvrSJ4-wjNYk*U7GU4_rz+9#MbN9czThpcdJk-dF<0GKk}6{)`-oC}QlOP0|z3
zKmyiLnh&Sq&wV9$bTJuk3*;vKewsA~De;S#QXQ((*TCV~CcpG%xD=+!bjh!LnqJ-s
zFhrLMUZ9gPIEFgl*?I_z-=w#B0;udqr#vTq>%XvVA_%;I+s=mHB|;K>G6AF`klz=L
zhaL)4)s{a51HmpuG%QThKicZ{9?(Y~4(ipwJtu#13Ydcr2e}?(TB{0{(E@qInpe}f
zrUp44P#+%k!59*zN`h)^^W`!Bg^muI%cwew72z$+e-1+6kks@r=zNbC4V~o*8F%p)
ze`Zc4%JPNsMC@h=jfmpl%5oK1*Y>Yz6&EOREJ}7I_MR#2PkNi3D4|coaR*L@r$S_u
z)xoh1vH!GD=-hKpl4nesXS@j7-_p>TjE8DHa3($*j6W7Ryw8QM0tr}^mFKOceANv3
zyPYeg<I2X%3#l2+%q0#cFIqEzmk<&9l6}|f++Ol>jBY}v4bGA(c_r3g0|uBIL6nkV
z60h2FqC_My(AS(4<PB~(3(}xQhbK6HUiX8pp}9@ou>Q_C{akr77-nd$6`@h6Q}>pS
zmes&`-ZPH*?br=C<g&cW8Q@`B@Rd|Hpf75m5RUtI5|gO6LG@k*;N+h+28_2#-VHiV
z9DbF&7wd!#v<;0dL`dBGiE03ExSrUxA9&N@Tk4?7Oumv2gG?I915Fa&P_iNt(>fZe
zN!K5xQ76404XlSz@^2g4NL;uhAKTw?C(3mBj}JrloSt55<iGyR5DMspK&7PCFhVF9
z5Jub@E96tPh7od8=nd?Q<I`9%nzU<Vk6K5vQpG~ztw;v1HLy@s5S==65Ky2&K2xi}
zAbiX<vS;+D#~UH}C9+qoB7EJbJ=;<%dwV;?4RDqkd`*l{_OW(w+wCc`Z><B2xa0%C
zaFy&=o01Sw3jkUsxBmXMjtI84)YaF`0S6pV+rt#@Oun+NP7bVfU7!x7dYSQGDF^xZ
z%r|E$#ldOJv_E*DcFG|(7TgumD2Mv*VP*68!(x2cWO~8$;Z9bXFVy(gz|mEF%ft~z
zgi403<m9t4@DXvd%aK8(Q5l$1pYtyDo<g)UA3SqZ8ag+So?&vd4YRHSYS=OM_e?V?
zog5qALz%&ChpIUpmk7@)eXJQkLxbD!_!vzdSFXYHo|hBSlxfO#VApMt&&LQSWFQ{C
z5Z?d+Mwl4tIG^b48*~6u0-8Dg1ju=k|6*c5!3f<UopN#-r`06b4BYFKAT45<pj_wW
z)LO;9a+#TR@U^taY4$=Mrc+&Y9eCsP7)uQ*k!tG18MO{{XIpA<%j%iY<QNn`<!yt#
zj8?lI1jC=@W6<<CW$Z3FJLWp9tbClDV*{9=$vFr{I=8k*tARn}aV6A*SGIwYlLCF@
zyr3b>(XChm0$jZlB^Wh{O%81XRDq_~$KO^j7TP3y+V_7b%Iaw53FsGXt8K-i?VZx!
zpJ};T2nB3;8BnY6>A->gWT3MLqD^H`>dXpd;U-EM?9W)fk9FubL!lZH?_u!Nj}iGU
z=f|KhErhaxHMz_TIlsVqKy1%QxiGc^v~84dM|uwZzN>)rnH%yV|J^+{t{WC`_lu)v
z1=~oxrE-b)9I+`{!yY|%%B9g8j%0#)o!zWaV9DwvWQi?jZCh0iu9cxCBf8s5CY4HQ
z8VETZf>P0B(}Y<`W>~yjS2bIPhd@DGNF8(h+hjyMf_a1NEnZkS(ukf%lHDjDvS}c?
z(&&R}n4^Q-ju31kX6VnjPmn9a95%T;-ovDLn87NQG5*XndO#fTS(e!p@dj)Cj!gt9
zq=Z_4w(-xn_+1@EsFkUKB89-T@!mrROgh2>w5+a3&z5m9aR#rI3AQu3+e_i}L}*5N
zNNF4%IgML}V4CP3h7yG!DVb!?y4#WL)S@}td713*VOl&W8RjTg8ZECs_e{CUW{B#$
z#I3LP8PIfN{<(TVi{ePG@jdD^SLY)Y5YO$#7f=M`T9XqKXVp<I$aOItlwRYRcYUp!
zRlDwi27C`bz8fa!ln^evZrj;)woDDSA3Is51&y%A9vCl~m_AmOm&>p;eT7Zyq)RTA
zO8e`?;bl@4TTAV2gTqh}6|0*~tj`d621BUd0{l_!t@Ch)wV<un_}jLX7Xx}XFZOb&
zP28>-?XW#)X%^PK&Rav|AURNd^V4m9Be23U!+#G?F++#yV}8(v?B0-7(hwxdqb5sM
z8q;UX;83Cgv`m`(83YwQ!Z3hVNplR3)6l~SDMes!dX#~S$qmsJ<Ov7>$Hr30g{VgQ
zYrW(XV`Hn}4d&7^RGWS}#;CT$3YW7UV4Jks(J>B37y{~OQO%6@0AA*0R&4*0(lIhS
z{sH7l+7gkM=pn!)7qz%#juEU0fq!yioIDw{&I^xj3iS$6WsU!bwD%6Mv#QpIm!cF!
z5m7|3&>;vRjTEZMq(T}=2rZN|Gbb}AQ_jhmnIr=Q>Aj;gDU&+sne?7OG7y^d-aASM
zMT&^O=UHXX;r{OTeSds<%}w6(zPqfw_S$Pbt863!6;TZcH<`)Q7+5*rWZo}rY>}Cq
zlRw6YP|mSR&UG4qv9YW)IcCXu(Mw|%P7)vv=i3m5A;U^8u(in50jPqBbzw}RthrGx
z@<~LzGwSsD&bT;rL#1dA?P9qk#&|*wPylkN{cG|bb1(t<Q-WeZm&FNb!M#u~|CbnT
zxjcpqqp+4YL|sv>zRZ*B<;vJp!;Y1!V#G}i<#Kgmn?+FZ&FHu$2y{#yQTNvx<@%2p
zCf8Ye1CN*M{WE^R8XQZ&IX5`3@aSgvNGh=h5+K4${$yX5og{y@4#rzl@QsPbJpWDu
zM2VD}P`Sy*Tb`eSDcx+N8CljKw}f!4Kr|o1kXVvPZncT_%T?h<7aydW-zs4j%Wc*e
zJYR6w-|ju@4$>DaUXb<gQKv@kNIc04cJ(G>aQ_<2<xX3=eUW&BYHWK+OOULh89Xt~
zSOl?B3dm1S%`i0EHXzXw)OZZ)v+xQxGEA$P>Fso@uM=XHzs(Q@Mwd{FoShbel{$%Z
zER$BF!AR8IlR2Tw_01JA9#C+uw{PAz77k)L07~z9HVK@RaLUU37#nj;F>Pi6FYs>}
zc1OF>GJRP^dZGVfi$ANqqu%u*|3zVaw6hMt3ZcPSve>yA6rp<(7!bqnc$1u|B&5*-
zY4dT|aYy@pmMn>}!J$TvETTYUX?l-U>TxkT=CI5cn8gHGL3RdAB<?Qa1k7Yx?qf60
zHP2v8bcGSN!uSOD5rL^zMtAsBy|!q2l|R#i8>ArYygG(p9urJ*j;u-C-Kv1S)jwJe
zzzp_1bi%d%I~-*+bm5Kj>dKfxEBaYyyyS+BtWR_{ytWOQ!CGToHbg^eU#piIG6ejI
zKXJ8*Y>WZpcWzng8=I{0GP|u#?y`U4zF)-R&XT*mMe0v<PLetnyl4Ef**0`-QMQCg
zsX)k4dk_^MJxfQj1|nlq+T@<<t#+q6Qb?g9R3nGLZriyxhy>l=v_`u%jgCp6Lb5do
zi?bC+5BA9nx$j$)u<#@;VlDUk+Z~AHTzSBsSwbF_p*iwk43u5~>>Xy@(sW-g52cwa
zujuhvf;P)v{Y#|-ISNX-T@&+gqVgJ5@lh919!Y)kS_<y}1@dS-1C|WU2KyEX+38!x
zh0!6mS|0bF=;>h*0N%~QqiU@@kw&jk^4K>FeqDlH`ed8~HCV=#_4V?UH_WOZ1gj}H
zp7zf<cn0cez#6!3xjYlc5}2PN_gNDsTP^@d^qhaiEk`FXF9+pjUMbJV;ITms9G@*O
z_y#vJ8Db6jiSyk6KD$})MgJ0>T`)797<1(%8^f6Fl*X|++>!Tkm<8}(0e>b~@k$Ic
zQwhr<E3c*jEW@A(zHE}$f)dC=Y0PGt<n`1}Lmmot0V>5C={uEXAjBGZ)2FHNi)uNr
zKpK50Z&|O|x&nL=uvgypuenDUq_Glq$2)Nt3}iXkzr{Mi+~Y2?%6u$Z_4gPI$|$SL
zyFuQKO$=$kk_3cmsT?x2jv(E8vC#l3%KI_9dKRm~R6d9kk(p+i&;UAp7<&tv1d$Y;
zK`sMTyZy*Un+FmJQ}*M;`P}n!RQ8*-1??B&>XQ%=#cVk)HUBh-r-&D~AFx4h;{U2o
zq^pkXZ9jtbH2TaN2<grc$A3km49LPqP-f-xZ;wk0n5aIDF;e<};lIP4!ycnkid8j<
zG*UTPEngZF1J*3cSJ6IX3u{+xf|&hxoRd~x2WIEjL0v#*>X-YEeFht=n45&^cvX1m
z8(&;YA|rHxNCRFLA^)`w+gpwwDv~qYkfC1~oH`x9lXI*MHwa8p&)#=)h9nZK0I&{C
zbTtM?q|vJ7Y~#P1D;7$C>d7=VaI!$ae=p}4AJX^XjLypUa|$HZWU-02^&!|0=OA`@
ziENkKYV{e<gaNp{zreg8R(@EDmE6I5WXunJ89?f=94XX8evngGsM8h%XCq$hhdF~Z
zxfiG%D4+luuaq65r%`2N<VP_SFeLN8c<s!OWA0ci<}6!0D=#j>8HnF^$|>-2z+U6z
zC*F&la3+H6JlZ*@z-Sr5`Vk@Y(;zwuPo0NumLubKm*^T~rK4;<IOk_JwtlGMgA1&4
zTNR3;Q<34}ih1BQ9-_3N=HTd=ExTGroHp3?7bVliF`$qY^A~2yZr+_85K2G!WekBB
z4|)K$BEPbQnmV=9IN9APK_zd!F}HF-#Q*-fq&G%YLW@8vOQ96~#>Y-w#0v=EmjzHP
z@uxXL!6m;nO7=iCli&GJO2ZWPFlw;yQJw4=9M#YWBopkuSI$vmB>Vs&wmJ}d$6I6I
zye~%;k9~r*fdOFWKuYdwt)g&UWrOTjLV<8T>>q7Z0f??gPY0xzcwd2)+31v0;8L$K
zXJEsD@eX(lUPc(Pv*q{cnR#<r>-ZphhVO7|YX!rLvU0GI9MS$KNM~P8=o$dG4simP
zE&vJKCS81Kh!c721$Zb`SC_8-r9v%Ebc=xRmKMz-D;ShwJloxWF`SaxF|nX!K(u<;
z=E7aiaL824p*}Gx%e3Mt^m@j}3fK54tA3+iX+Mz{DDKUajM%8&iOU!nNQjfDqI!?D
z7HJ<R35POM4vW@2u-j@Cr0ZVfa32Q|1n3gLBbdleXZI1-4jkCHp;(UeIgsIqiEad3
ze3Wyo32T0Zy@p-J03Sl$nlLX%$Lx{V$6c`s`|c0^40wmaP`GGF`o>7m1I<~M(l$c(
z^EuGSqRD_D2@skpE&p63W&XSGbApBFJJsBh{xNo_phzPGt)|I<oZ8oePm+P&uZ>@o
z5AS!7|DqNG`Xz<%7EXx4iK+=O`j)CO#D6a>rGfK66BFhjLh2aX4q23y)DZq-V}!cf
ziDXQdq24>3Xbd)i*rSh##@;a70A2gQX*@i>#JD1sWkj%FSvBGtWn>T=8V*%R;;YB`
zKEJ&E^7hO4Um2Ci(1FMEz*sG#{d>mMFhq_j;}{b~6}h||A0tp%3-b{{1Y?7DkYZYt
z6H)`bU5;ShVmUF^jOHY0YcQT&m&D|^lWdNt2a%VP{avmQy;4ZU#h;dz3-w)2Nkg~0
zWO>%<YRWkOk~@!%9wy?K;Oug0jI%K_B_pT#+eGJC;!Fdfbl;GxXUgfmvyj;ak#a_E
zt4Bkdn-gopRZ1$10fV}uB;!I9`P-FJ9-YE_mzN4B4oScyjg>f5DnnY1MzjV3hq6YD
zCSdI{-rwg08TG@}tV#s7LKkG8>NNF^8??=WQhP*lwt|^rlT3&K_0NqTFEt@v2lbcQ
z_yxt0GEx_ko*r#hai|YQ8wokrDETxD3zmmfEQR!po^dkIT$$+4IAxrTkU49l=(Gin
zyybjph_CP)?<GQxrZmPg=+`)XDXF9>_K3p7CWll4$)xl)Yr2`WGC4hCgc7UIu9GSL
zjB=JYAI^j6(d^HliegDf=mm0S&K;;Yf^h9{HS(#>Dq$B2YO*C4UE`Lp3C{Lr6hgx<
z%%_?f-!Qfo-<W4`VLL=TLq6r4Sb5x%30+A3=xspNpxE6!19r^0-UfOEA@Jut2!6z&
z5}@08X_+QAWwU1j0!^$yrP}0}`Men)x>_!9+5>WcY`f6Aq+68%KmeKKB5wf$fLteN
z!KliP%=3$5*nLrl5#EAJ{C!LvQfUs%a~4+N-zid-%BA)`^b6FkgkR(`>$AND|FsLo
zTIBNJJrs%u9_lNt-($J}u9SRp<jS=BH8uw_-XgRY$W_4|TGtF7dMJ~sFv-<vX&3`B
z2Z%c=P&yUhB8JerHobxCmVEK9^JjLBn{je|8W%OqfGt>J>V_D9*WSo@`IA$mv_`*j
zmXGMqp?D?N>W%gUC^iDCwZy|s-WQQuE;h1|@iqLBa&uZiTv{);e2e-#y%4+!<9HRu
zN7;L;&x476%*@1P1o9VurZ6PlhAuC+nbd;41V8cZi6B;P2PJ9cju`YvB=&0>(w+Vm
zgDz<Q@nV>k7!I~9y*_YNfaE+3Y3nVOX^Dg^yh{xY;S_q%O%M4pEX$G=GrS>&(J~Li
z%39Ig#gAHLX6(`Ilst+iuy(;Ln?ifbC|n;u+ay~q3w@@wWT2$ZJjjAMu}0vb$d)KI
zROcoV(7I93N#^-CY9Vklf(tB$#InTa#}rTj1-9G*8v^py>SbYc2Str(YmD5~d6TiR
z5f=q@8>?~E5#6baV>T6qCTWY7Yqwkg<Y-CkdR9Wky+5+l$x5jhg|RGVZJBcm!meQH
zEcanh!-ybXJNqoPX+?C}?a=PB(moxc5v{T+SbWP(kV~sW2+-37^khwpzYJP?RUWp)
z+GuJpt{Pnz+zei0z!<6<g9xahBI}*b7Tn4GI9E3K+tklkeONR7uudCgqcu93xqz}M
z@e=cVX@W<Uio4Q+(<<RFGv#i7hT|LyPRn|gY|eRk6-HQ&$FXJ_+Lkn0f-vmg!xE+4
z8{yi+CL#hL_e5u~C0n_i&=>A?I$NC>?nGFrP^m0bd68_5ZvZxt9q^~z7aUEWS+f=a
zxj!BukOxamSL6d}(#$X9ekbO{dKC`_kDSuEbLUg#q2M@+xTQap2PvG6nkXQDHSUB5
zrYH{wTL8(?h<z{-qr6aGNs-D&e7FPw2s$hQ2R!NnnWJjGb2JpSRq|LO-(vMX5Gv3u
zBaf%SGdG3d6L8lcm<lLq>VZBro6wU!Dvje&V}lB%MXxeJ=Bbj_z|0_gB)&c!<Rwt5
z%d||M@z%`iuXjA*aN|uclwYglSy%B&inZF<%yIiI+C7(g=1Ghw6c{!p#h$laGO&pv
zQeN=QM+J7ayqM^s4m9A)s!1Tk<E2=-ehj6MmlH7zTVS=8SJLpg!@+;_zCnScp2cDj
zuX^LHEJcYaF}zX#F0aMbL?9R{?jS|z^%#hfj0QBlB@3agy^(&g01Bq^YGj?f8QUMl
z<qws&62s8J#6_W_OG_1!#pG>MI7b1hdB^yyn5)Ui-$Ewz<3&8)2J0u78U=pV5t)17
z-JC}u;KF!BRRG3hTJ(Dzq+&k$`}WVy3T<ta54=s&86rY+k)EAvp>ll~U!xc_u8039
z5!(0w$wTsS5K>J7FM>eFyL@6Z0Rei;ry>8wTGfJocuQ6hRby`DpZ<)iPyg~Ftn_DQ
ziVn)jzx>yza76#f=RW)ZlzES>Wpyf+Az#=cjL>j3`7#zu;eAGGGoXN+_EpFYmXw(U
ziHvKzWBn?5o&DG8)1@lgz{18e@*i8dE{FD(Z~ku-3FR$=f+Jv{WJD$?mZyjflCvim
zvPi3gfH-d+uzm5}2?ni_*|b_D1lBh8ZF|d>o58@}o8Yi$KZLvGtD5Ee{t2(MMoa^c
zf>znqJ23r{t`}z2(@x&be<5N(eU%3+e_t=#dpjT@lxNK(mL2>VFa@0Mb8r&PmLJ43
zNNd)3p&b8V`j+*w8RxJ`cC=xm?lzMGel)?s*}h|j$&b_UsA{V9j6b2Y$WHzXecVis
zqY2!4&-p;Kf8t$$GB95BXq27(ZOUuVk>(G6p`!Uyd%L}507?Zops>pX#gE%tDhl$m
zpy{zqxQXTGF(n)mgsRD7yQXJ68x=s6U&I*nr~~$l_ja4$+piM0n^qu*0{?4}U;4->
zfLQAo?o^BnYbw98QJ@lJz3lGo;VqDy6#-_;ul-B&Qk1$Ie`8}-P-Uk4c7nm{X1ZGz
zm8mPF*Gl=Fw+&<%QrF_T_L!gmH)>H=W)W1f=LA32D{$#~Qynh)#ptuLR~kEuD3!*A
z?H!BQtyeGE$2*|SL$!h#`TP2yFkw|_X<Ly)Sm$OFmSw*nAO=P5gLWnRr*`I3gM?h3
zDF>u>xbIX$2l*I&V!^7NY#>9k2&$6<jiR|~RATsjnzaI6*jdx%pcwT)<b51GAq5J<
zQ=zi5K{`)x=S3yp-69kSImFtm$kbLymqZV<D(TLE$V9+VSN}!NW<2XCZPd-jp^@JB
zh46WI8wXTTgrW3E12t=K72y0s<0EKIxWJxvrSx=4lvV~RLC~aEYKLY2Y>)T0!4DWT
zV3hPRrYSV$sA?ZJLH!O>r=B8*`-e2#2sRO5IwE$(SX|yJNBSGoKh!$GE<7qYs3PBl
zt0|9m0+<D=xl=xdKTL3NdmxGnBK)`S1jQEjI#~Ke#}LQVZgI#<Sw~mk7y!<&Jo?9A
zpgGsefC*dSc#^P$VH5xV{;v$|=#`2w_;{ua3jSz7EO4#glQTFbQ;CbXfj14AU@+IW
zV)|8QG<PshAs*5Uh&aYb-lJQ0IX0TfHr3FLFx08XC<$UJRM`ww8)mF>K&MWf4wvEf
zj|2H184(nReNNHYQ4@FM1V>{k5av{&%YEF0t-F$FEnBXIO3ke5>tIdV|DP_SY)Qrp
zm(flL4kV6)F-h7ih8BCgzd>y`s{D*i!w0ser<exifoV^O9oKi{NI5Y$_*-FK^x*LL
zTOud<@234zU@CxvCK)oeQBF=>Z(w$`VqB;AGs{2+(V<I?v-#7~gRgiz12#{Mt`!O^
ziFg39v6tmE-{lBWm(zpYkY!pcXBeL&O|c%TrewT7p0SMM<K;-@X<RE{wJ7gPg_A`0
zAsB#pRAhtcQfY6uw=^`BH{>X1nkDNkU@(xNOz2u6<6{+p9>SB#$bc$ulw8F;$F+nm
z@X1W6Hag?>9_;O0aE4wi8JIn9@tI&O2%T1Hyi1Mqf-<-U23761*AZ4A0bwHAb^l{J
z)Z)2li=^J$AqGbS4`sn5@6^VDK#8In3{?TG7m3yiA!X~}?aDNwEYL*bz+q#h=<UHi
z!bxau#0KAG7Hz6VBzn^Hez-Vz`3KW24MS`UxiqAqp(!g((HyviuK<mfNij`E4&xa)
z(B$Rhcw0?CtQVBnDe<=MO;bmxG|rP||HTeyt{sndDjjDgcB4rM!AkErY|c<&&WbNF
za1xDvCmX@pF?>|_g!6i8>YQhdf-0aVYv-S196K5wcKM_IeKIQa%eltAjn%m#fWh;8
zLaGUvMJDIRp6q#O7rDUxF$CdC8M!b--soHv%=;XYP!RfQJX3*K(~HxzOj}FUdch@3
z%(=t{-~afsF+)ZSmrHF?PaGwe*#NtA>DpT^_bG)9M6{U{S}a$1^Q@f8s&;0tcVvNG
znO292jKvOIm4;-V_4$syLT2DRSu9rv0eQNhAKHREY1Ya$p{^z8DXpbzgJ6hIp-I0M
zHr{ncfssflx!xvH#Zt33*z{_U!)-KgDxhw5QU2^rSYZb{)@}@r$6W~ZHRGFbm5vYG
z+;X8BbvV&(_Exyp5nHKo4Y$~{_7()y)&qOV0eQUD-{Y>J@&(Btf0?jVo&BMWK<7{q
z%Kc|Cf`qMa^DntenGAu<L%BVKPNAt>Z$U?LcM|_A0czeE!m*JzLbW(`P0D$2J&wN_
zGA-U4JaTB6Oix`gchoS|8L@KMORy9W%E$Z=*0i*ol9@3vAer<_vQEe>Bmcl0Zu6Y&
z-*ewF!>j|JGq|QT%%cqBrV+3`CmQs#jnSQRW4n#bvy$jM+YNct{VtdJF@?rFgzAF$
zT2FVH-5X_LnilmUE6TLUj`nI`vb@q&7TY^|>kBfr)Ff~F7TE|`duWI(iSbmZ&uFQ?
zXpj+S%8*u;5?>aZxD3$k@|db4&9Wkfj)KnPWu><Q69SYbxEJ8dDxbhyfL$B{Q(jl6
zNg?D5E5ab9HU5kqWe9!<`DwD&=}!^Oplv#MO0OnZ=fB%-wz2b1dck?i;`+4z3I*sX
zjdl^?hIk6)WD}H>=~Pl=qs{h&!2?i0YLgF!LIW;-xGy%!T`_1?hRnXY+lY`aG|&#)
zoJdJ?SH-ll+Zx%D+FOYoWKOvD*as{Ci=Z<!aF4&H_Z5NPmIGG4H(rCu#~K3Ik6XQ;
zlJqDLhoRN?+1l=-$jdsp-{oq4l00A|)f082JQy9Ream#}HS$o}EG&;icZKDr5Qf6J
zDt`_6(vSt`G=ke7j#;U4jU|Ik{fIq-{K%r>Lj_=tN_jNs0=EG*Rit>~oIrM(JmxKc
zU4VpAoX(NQqaVZ_BIX1-$rJGe_cl~$RQQv&1E4XDhX`qur)<-nz%Znzy*EW&Kty0K
zm^JG)JQM6h_dX9x!tDtF-YCxo1zdyEgmOKm=|^dm=VBc24CagF@V`FqUr^>z{xOHn
zYbt;ybL54f#w4Uyl(TwTv%DDN(_VOFTj`kT@=|Q6>Krcic-d)V^3-6HS8PFXv4gzY
z(KSeC8W@PG2sqWX*X-3H$QgV+&D)ZrS)tNfRf@mizmwae00sVzRWi`;O*@4qRLNT%
z>vaz?gtg#{xBVGsLU$N}&v$Gy;o@eTcYm{KRc3+C)XU!!Gw=k)v|#p|M)R)!qR|(=
zJm>;S<9pr(0@3icRr7k^-)2M~bRue9_>(T)4-(-vQ0QTS=gNnvSBC!wNf}=EkxlNP
zQ|066pG$|S&s0f>(_TIaN!YDVANe#4D}*v)+sZ#eo<f^KoveSRp3sZ3A8(tZm-n+o
zJbF-=d$>yeWi2D?3^7@s`=@rM;z}tsvT@s+d|~|E-g0nf`O=#)Y|+mCuY3g!KjAtu
ziZ>m4@^52?-l!$^wSNw5S^ah_c=8{kL3XTsV}CodQvRFf!GZ-=meTAFsWI4)b~t^@
z^xb~PpIMzpRm$bN=@}%MZ8X`&pE2c^G0=?5{a#IgOS~*xYuZzsVOvn?Zj<l(H!Sx~
z{Cp*zZJVA^+d_p#NXqSM4CpjX0Z~Snp9Z4I_BD<NcI(koc8Gs;@7hg%P~%`sd8VSK
zfd9!4?PpZ>fcJ988pS^$XBcjeK#(8RxZQgk(t9l#{m0Q>9hB*vY83Vx#48?tVt<E+
zs9ARQvDq1AqDM_c`DvmO&piOo$Vb~HX2P6y(1PL7y4><J8&Gx%;CIE(ex5#GjK#3e
zclFOfKI~HRftP5KUpT4Zph_|xRuGI6lkvONC~}Odzr3ORm)^D=sobQi<yX<4WgF=B
z+&wLzsnbibbli~43uN(*U;8(+!77GSOqbsT;VV#;9%bRP^4l7Pi^@ifl;1h!m?O4F
z&DK2(i}ARG{~e2wJ-w5;!1;(2^SXe&(sb7`Seb?Oj1Sd)v`qH4F3Yit_pvTnu2S}`
zapZDL5#c|wpSSFB8_@4J$o}>vO5e)?@iubAGtJT|)(&Vlqt(JS)+z@E8Gtx(d(!Lw
z`*?;0VI%-b*2_Wuj8X~*F=e_OoW_7lu&q_;?6ZsFEL5?%JLomQK_rJHHkif(7vqiq
zl`h6byuZC7-vGyc*BS>ZQS!M0vHfs$W~H0W@$|FKlJ4GpM;X_((!-lCDK2N-%c1@Y
zg*QdBRq}=CmYzOjatAXm=y#_WA(q2@rCzC@HXdp)(q>EV8U^>N^Qg$oAcQ`)-&k}3
z7-JB>7Wa}g$l>0zwP_NNc)@Bp!ds>Y<CR5pPaJ974#r>%*HJZQ(eNw1^$oD7N>^nj
z_YcU?B{Q&a4!)~(OVILC`9mxeo>Az5=+f8vVR2d*H2qQwTIvq6njX>>*n?nN;6!CM
z-}V;0cf4Nu+cF@ZL<WSYDTDidppPC02}JlH`wAuDP@)2Rcd%1}LCor{z@8+UvP^~~
z8Y}g|UQ6W|>!TL}42}&#VR6WjYoQtrjY&ajLrufU*eb)Ud(^`nJYcvCw|_Bv8e(&V
zzfW!?>dO4Ck!e7PzQwaeP{C{&jtk~P0mF(sgaKAoM)~aM`JFAJW6ZEXIS(d>jR`tx
z(74V`y9(F)@zEi)J+72xBVo=mw#M%6AqYqZ5<<@KCBq6#f#TuAB;5!N*eEBY{t2Gi
z@pfW7gB}Uk2j8rdlYC5ANvI<n6Du|{=n>W&=5eyk3X+?3L{5q29F-X_<5F{6FdeR(
z8mEm`waAjRr}=BwB9!Y|7`piXa=P~lZ4D}w=>;XldF70t1Qr6+QSDbbPBPvaeP0Zo
zqOw(PoRou_kNP?dQenI?2AW)@H)JQQMFDZ+ZjfwnLthr%8}H3RAy&z=6cV#4CVSYy
zout}FtVbYdZwePHRZ}Lcmz*_+3$$u*Tkr(mncv1SI&@VFF083>-&n2)!<MxSiJ;V)
zTyJlwDo9<B80$`Rp4AEJV<c+ax6+#hdv6*vS-LY9Nj`RCLv=n61)yMU09w^*@LKa@
zVmzzP0&Qe1hO888w(>Ho1CNr1)T;(yB4lQXG{!UV9@8{<SwWM{nRY68V^WA{SW%NQ
z0Es8ZA_8TkGzZ(-pEAW8bBCOYwq}1j;&iO^BTz^=(<jU`f-RsX1&F?mrv}c<nQ~T0
z)DaaGO$9kS=7utSm|GLaHq~fSz@<2fHJsy}sg;DrVB&+iG+zJczm&q4*j+dY&J7A+
z0L(U;0g-Z^4~@KO5iweQkD+F&K%XBZz+U5yWxaG0As2)Qfnkb73X2%JFlJJOf1)TC
z1&hMPQ;>^O{{;GYUaXZ%;#oW5QPBPz?C48N`ebG-yexQ|c=cueHUK|0N*Gvsxj$oQ
zAY<qVc3k1lOpCGtomWOrdMLA)&uY0UwZ~0Edw#xLoqA`urNNcha8L-YvDxU^HB+v2
z%2}OvL)#*cp8R&5w>?{fsZI5`-k;H13#|;6V4K_!<AS6>krmctkU!ZDLJ5t6{5hCM
zQ`QXDrmK`D)q1S=jj;}RIh(1d*4qMaiq18IQZumR=Gay>_3E*^B_@iFOO90x=~n**
z9!ike%GB7R|0PsNY|SFVDH`QA8|r{csFSzH>i5Sb7PWGR{j&+>xd><9>0{DjeHKh4
zEx{mA<dBD+WG}wY;^0R4$TY|m>NcWx02(g1d4@H|YqOy}sqm4RHkd6JKoOngQs}z&
z3uJZ(!UBr*w|bL?8tc~Jp2U*Psd0Zf@_iL2H)~^V$U^K0THAV4gyMxJON})1V(<;9
zp9zmB*nfUdYgkitrmha|>jl1;s-&YSbDk*vw|O~&ZWfy0A2?u)EOKtcDueSE2WiWp
zZd6KJOsHF*Ub4jAgan2OAxnLFW^dRIFp4d+#)l3#NtVaRaA{@^Hxy-MMGU`=H)kLl
zvr1MP1;=Ed^sVw%*&YqXR4!ju`?7knAWwE#^Q|$MwSzAa_p>&=t*$fV!!MC_{tOZa
zY6|zza#^2tm-^&!E)cjNqDp&fxojwDhh_#Ohir_UQWBhi^Ggk3@QrNpZ#WX(#m+>&
zyV5gW*2B|yzT6${l!U}W--LX!(|WUg)82x^x{;AZpw*VxTmy<ITF@x%v0n8wTIHUg
zcNyq?CZomgwZrPD{bj4IWgXKV<i1!~d?)wEKL!E$dmxQ;gBn)TA-x}r6KhBvk|=TX
zJ`}v|d3KRU*Cv1U&!HJ&q72NMBM-;tb@@pdq?*bjF^iFTEr|1|f5fYeX+<u<CV9-e
zpnHpYI4dZc>hySgZKc}fiI7YU)G)JfJ~cMWlQE|P>LWds=0pc6Q>hor(?PkxtTrmo
zq{Zj}(ky&aH=jJ~%-;k@Zc(0#w+XUga0`${L|vFF7zzc+^D!YXFtucuLGnUOr~}Ev
zglHDL=&hOVRT7l}CZuERC7bY;YY^{x4UC(YW9Xw%C!kqgN#unxK_+=MxL_=92jErE
z@>(JQ6X_IFm)C7Z$N~bz<PCp=U?c%<RLzq&{Tb9Hg|?wI%UjVIQ5AsJ^0qZIU`ho9
zu)LE9!xP+4N1keC*e;O2{VzM#l1lz=j9Qo|@5c1X5y_pE_hKdMGIjEP5+6;qP*E6L
z_koF<9=*HChw<tG<BpP#Y+(5uFZh#>t@*R|l}~~?U`5j=KD7>;%I%;3un)QKkh1en
zCxV7bgLjtTO!zD~5E>j*C;694Zh~I0VN}N~sdwa>%;z@H{sYS73vU#$BH$VRxlF#a
zMiD7l)lfjZ=2t;VRV0*Czz-}RW<@&w9g`i6e3`GkLA<6f0B)22qy}j1v_iRL2>Q8z
z3b>m#$~QL8(Te%YfBkK1fo>k!xClA$(Mzq_&~}^|lt&|w4r<iqMVs@1qkpH?;kdGF
zg?zVGA*c41u6?@ok!@<X?gq6PJWs*GPl@=ME0|pVJ^x*8O70`<DGnxlR^P8RY%;<Z
za5TuawGQV@XsW~_Z&$nZ2zJ3HsPt;Iu>RegS7=U8_m1inBq%i`Rv)nQb^BU_8dU~^
zUg;ts*`Zbez@~DzbAC{}bx-tk@c-9;<cDe0C>xG0=zZJK=z`b<p58x-dDHig8K=$S
z1NyNKh|$F?*0NNi0T#o7xRd{a^IaDe_uTk5KMBeY8Zn~3?Cd>IKVuPq^Pq52l(s5;
z*UL|9)u0*HB)ixYkrG@ZKl71?R3h&5^Po5KkpYA5>aWm=M=ByeOMX%7uxp_SRg`I`
z?H0p?$;cq;^=cOprTJwV7|R+_FSfgUnS00Z;a>$ejlq2)yGLgTf-^mGvHUv5NP`b^
zNA8IHCcVcyRMaIUzm0JZMxbB4{4Vv&a`P5$Mp4V$ck1<f#FD5!pw+Ws^>kqFd#2yX
zXI3%D_yv0b?(4m*yMBI2^xm-=CuC|HWuG)=f>Gq~^|Eh_xqa?E2ufA#d_RAN=JAa2
z;?re+?^w-BY&^^CwA&MKZpi`u4b^+Enj@WJK`L>rq23{Tpf3nyJvjzLW)?l;eSaSW
zZ)~cPgJMvdag6f?9OVb6K>-utibIX>938-2jFKI&A`h`<$`JC@AYEei<r(B_cTJ>Y
z&`oL3DT5oQOSf7_;j{G<^3AF(bg$jI6ZRVH@;q{430>e0_6q5d+QlU}a%inW%$b@R
z>6yAn*qyRFksLBeuUbVN+glKzUz4qs-cAmJt*kGm_c4GR%cPHOj8TK-u;{D;cel#n
zK^Z;%nZ~zJj)-TlU|~%nYH5ZX=_BC@2d6pC1-e_f%^7~uDn|vmhi9AQ=vsA1w70+w
zicseU`GbuU#Uu-i*pR-lY=LL79Z0uz(l01PJ07pYW@0o-S<o_r>j=%p=x_7L1C1+W
zWq^+vXsjkcgCS&Knuj{^VAw5?L9u-XB44di1_u}FmQWBD17S@FEr2CM(m)AAB0~`a
zHQ{=U|6+A+5_|^b{McY3hCSfE%?247vZ!&24D&&8w=gzmC6LSEJ{8zy;D!-4#HwuL
zaNf!}vf~5Y2#cZ1ALq~LIO4DvsX8inp>IV6@3bBrLx)bzyis73m>TWpG4^$P3xb_^
zUm%_O$2%c7EaHH?WQEuQea}RP#|GDoga$7s1UG1y3IQ}meef3gR-gp9r6zj7DgtcM
z3bQBqm{GS&$xG>Uav~z#qg2lHvJnJDoKkD(r{+W87rYH(5!a+r85wZQfS{9toHNcw
zI2b-B*mS2F$r17bG%|RzZw-PPy-NdoU?5@8Oqy6DX9NLZ852^>*uPOJm}!+r$hLJE
zFeO_7fJrK99m#8g>cl(8lu;|K-&4n(dbnhrN~TK@_~OaZaVO6B*e01u^<_zwlZdyK
z5t79<M^c^0HV-mZ%k@aE)*T_$IfRU09~a<pm|$XK%<zFyW1IbqaZ(#oAC3FW>rzw9
z_n<|JsKuCc6A;1LO39~p7*j$-Yn2qzDq)W*R{%AF4mMvV`rB}40qM1kQH)uY=kpo{
z(GU~|0qAm@r}1PwG#l-6hG8^Rpgt@Yt|>-bfQ~#Vc56j7tEf+o!9xD#DjH-;w2XKj
zUb{9^*^n(a2fL!xf}H8Y<Ss=>1;D&za#jozWqu%0>g4PsV#@Ql-;~hz)OaUT*Nlr4
zE9D$-2#SXmzyK%9AAJzmCpg?_C+G<S9Q|D51*G|8<UG?k5v=|H;vkgsy?rKNQUoEC
zPc8^?0;@QWi`;AFLTh(OzvIWqMIpc-msz?1_&rXDbsz*0^d)j}@F0vidbxO027L+i
zdh!OiOMLXEtl=B%P0e$YOVe03s`U&RDwla{^hwYsMo(&hLc}UA_irFVHI$9HPglgE
z^+ghQZAPy2cg$d*)Nhlk{23U8urXs^<m&j2g=(NQT@$=e*~I%ZueDFeJG}Xv7mCVt
z(KLepIJ}L3MXvWDhuRFlQdytIIWH`@!N1XR)A)u4^`DHupmwwT*}5HxRQ^%fNphn%
zXD2M9Nm$KA)5KtptdW~+>rJYWo4sM}2Jf(d&n?j~82wiNeYWDn^ct*`zj*7AcdEs2
z1V>RKlgcY`ir*GA*>W>v`pt5CusQr0S-HdhMTrf0c&E2-D{B=XT7rhSC<xZ|v=}0R
zWiZbs3}Jd=2zG1ECshb#h8@ifO_MMhN)TFL3uUIsoi4q)$SnKop$KlB9X*Z7=WznI
zdQS%TYL!`;lL)|EL`Im-mbvK}!E+quz!c`i27_|K*m0K5Q=MPt`!Au_uZ4JGNYFZ2
zkiOA8Tm|3kXj&NC2^Ej(fZ$ZhqV%p^a*}L1S!~PI-ZFT&v<0m+f}P91T$aQn$3yxz
z%F_5}78T_iWtsgGkTkTQT3POEOACYNF%2EhmOzuUWrhE4=8Tnn;@+Ygaz0S|l{OgY
z6t%J{_yG4cLrs;{A*hN9>NLojG|T!-0myTsto7ziPsJgP1+)|?cDPp7B`TykfiBS3
zhhXemi}Y+1b*z&OZb2L2DNt#Kt=BLRrUOs{Es%}zAuiypfNBF}W|K{#OP3y9<Sy?~
zEA4PMLU7V>YL&a~twGT7WwX;Azql2>RJIsp5Af8@_B3?Hh|tcZu|#&dC$ZCVO&(u_
z6YpMsW<`b6c$cj~9QbtT-2*bh0a0&4YqRA(W9&``A0YSJ20u0<52OWSWOj)o!|C>5
zEDGwHLpi9BheBgO!3_<|QJdhewrl$j88rZ<(;oJ@Sh+cRO6Vz`3!(Ig)7-*xXkyrf
z<TQKeyM>&<W_dJ-swrsH=czX_zU8ry6<zyu?IVxJj1`8_4X(nEH(~PteZqeyo}jRx
zrUrE6bw3%~!8CYU5Au}D9|pfJlBc6Tl&D-S&!jb4g0m?yzxma(_8v5#ChU^u;yuzG
zGe<>v=ceb=%orL?smKERv`t=!xp<VqM(iwV#*F1g`6qcX_GJZbN2!*Vydj36QN6`Q
zLX<Qh@yo#)P<s%Cf<M>FEAeq2^H?LVntH-*S1qskglAJ6%ImQysdGTCyb-MdKPFk{
zU`xCiL^+lg<XbUW`cg0%9;bBjEt9wXTg&D#75tshjJlzq)8Bl~G^uglu$K3>LI~=Y
zl)uL$u@9&VYy!GP{dtzW>rHdpX-p?3`CbgfZcayXI7<@F$8LGw-}ZoJ1}Dh}i8mPV
zOvnag#5qoslnLh_dTT(ZA?apA#nwnp*xVoamxi{1w=|w4)tHx${Y#dDAyRIJ82Ry&
zSYwuc!0NLs0z8P|Zk3essV!3uIW)wL{}Gea%pW$lQT}P07)q&p<_pJ|;uZ2Q-&t5j
zw5rwwkk57CJO+S-Ui-Oi{3!$E3vWE^-$gjtzfANdwxTH+pawt<`6`XP<koabe6R#^
zcI4kddyPXfFp&=Nweq!pITw<Z)L?R;o#rNA|6>!%WHRzibnhr6ifm#KGvR#rZxYoS
zbER0M!9W%%rW;(O&e2DN1mZ&K@6;)-KtNru`T1_0A*Qw;4G#|4pGup_Hg$oQ)n;J!
z!G!u=odTUKs7Ze#4i6+}M^`7n0&kXW?VtI2hC<ZKc6EkX(r3k3Yr8O-y65(F2D(}T
zHzgL&8J?m%B{#?pbz8Ruwo;8tPrv{_s8jS7rzMQrA6g$6Ks8Q7+0pxe3;|rE#~-Lm
zS@a{DH}RQTdgFiWecD-|IvP(0_4S_z?Rlqoe?((rW^zV;;w>9()PO_FyuD&)Z;8Fp
zhGGmhSJ7-N0GCE7js|;A=P0|>1%PDt5@IGl^Jajd=@!)xaZH1;7L3&Wd5}+?c&!k`
z%)}xdwrl*N6=$vNEA8I7@(ZI5oT02!D!UmS+gpJ7A-dB3g(!F4a`UUYt#)a5rh?h7
zjhS-Rh-{YKW91S4pMg80L4IxYAJ}*3=mGMZIz?r1%gJwjn$)%#qQuK?e&^5BU;zPm
zjP2p0)x1|F3SkZcl6xi!EFc_0d=bA>UACa^y}UJs7x3s?fMwb{*jIzFXczEaHp?(1
zG^CPO_Ni0cZD`+qvTx8}P`-A&?Dwq!TKM+<{tSo6J&0b#IXhPli1iwoDZ=t)RC%Xh
zkjflXH6&)oycAk9&wwFKzmHBZBSx<+fLr6B#C~a$VaD@(O=ru&X-#pjEq-CDfWf|<
zqu-`___6?Frj|;4g4&ZQI9*ap9jR=jP@8o1UvynSDIUH+Suwh9vFIq|GoF?HrF)&a
z6;Mb2eCgrem1e!@u`h?(OkoJKq@d<3OV1#9L%uR2y-L2-8htQYxmngpZ~r9>B*H@6
zGIOMltzHkfXAZN*`r*z1?BT<0P{2s5<%sA8=3=89=_JSU>j6zf^(gPa7|$!{6X%qp
z(*kKEvdZ>9_*7b|k!t~O7h~W0CQ_?x!o46|PUMvWO}PM5=$9DS9AK7;#8`Mk=gPbv
zN;TdBOX<<y8JJ2iQwyjl*vW5@rRi(~Qp*Yd$Gc0=ipJ{8z}N;VHLX4gxEbmXjV1yk
zNV1nfP67BqD2X;>9QboBbhyEZ0*Vd6M%^Ss{O#G$y>RZ8=-$U9IwU<BXElpnqc?2v
zWBqqJ!I<YlzZEwvH#P6p8ye@tmWwv8&&n`c-%7}&;XV$Jrqr>G@MrMJXf!X*Z5e6L
z+FP&}8;ecMUy|cYF0oj6QASyty%A74I^M<JMU<{y{4pllbVYF$v%dWq^h?AZagCPA
z*hKe*DhvK+Nc9QPc3q|dkT!0P&B}@14qG5D3)53hiWygCvp76Yj{Q}E<Fp|+zFAHQ
z?%%u)Y{0yOY50_3!Q+D18YYju3Z*Zy+`w9LYKSHT$nkbhWbT|6GXZZZst3dC^ytzq
ze!@jyXN1HDk_(8ArYy+VEE)gWPCDviZBp*T!@tycU$6O^Gl9fcME4z~M=IkHs*>k%
zrv^@!tnJi6BZrKT@j;8x8RRrbRcu^1Vj877hzBK~o<AC!Ie$j;nh+OjuuQP&<Zy8e
zqm!`WHFfHlpvNCUM^M3QqbURvAzBX}lVwt8oq_8&VBgm#(z0R+IwJ^CB0BTl0#A0v
z)u8B{6r2)TumU@kg%IuBn3#r%!qP6Ri~h`F+)SJ`Bw8Y3O|2PT9fl&^(&*#iB9n70
zV56gKUy~7bNIk+JM>pcWM8K%&6Ua~uk!frt;fpdk2DN!UjV^5YSu!Qo2Z1OJFkf+a
zHQU~)f-kUM&Wsa=H#}tP<Sg$)jZF-TSc-udcDkNj=a{voQeZuJUDVXnF_UJLwsMX?
zGdw{9>g12X_^RGADuEubwZJ#eHJ(8dUrx@mztf{3=ldG6j^GNpAW@Ss1%#mK>tw9|
zg~3>bCcsn|8P}n<x~7&~Z1SOEQlnfF)2*T+TPv5w$U#hcN?6YJvLFE@ny!q3kanQ*
z;-$;|8;kog?f(jY#yK$|7}kSa>Ca}X^{Zv#<*M`ymnVR*ZLln^_GdV&)QO?V-09g8
z)~bV4UL)70Zy~`zHz<vAovCDpA0*e?s^Sv5OtswLgferEs~CA4f3ls^-U8zAc3}Cl
zx60Z;v`bi*^v38NX-yY^D(w9xqwt7+#}MDR*(eO;W?uXHPI61h!tfB_6&H4x-Ri%n
zJ#2eimmDBzh=xKw3r+Jj+vLMg`>x8uCU1{*(9#rh5d%j{g1D4%(;dbqwRwPaccuwz
zWIdEC1|Gmvl9uSEHk+wwkZFGYlvl`fqut;pnc+Rz8AfLwi=Oe%R`^F&%S@wfF3Y=`
zW(C2}2?<8+YIb^tqf<LxA+0X8z|-6%07d5bTghhyu{75QPG(|g@;op$BtBSrUV3{S
zyp`Ah*v<2Ophz9y)&U+$b;Bmx2Gw<e4V3w!jk2(0QRuPa#bCN;WRbnrh>NAmWpOah
zs9d?U#Xl<ZIay-=z{SOw_R^Rlczcs+Ewap>(JlmOAPiu4ERU5?Je;}^JQ3~laOSSC
zm55OPmBvs5QDGMneO1V*0Zqtrs?6fBSe@2C<?LYm9d{M1iM<R8=TWj}ZQWKc(Tt-M
zivV37q+}?9fqFOxs90nC8)co3GR{339^58K<2J5pZ^3=M8=?nX`vpI?Y>Yv&T8<t*
zvMG^;u5^Rw?g|oUZ81}C_~#P2+gNUBvuw6*w#>(>ZV96Ep0xJ(SaD0g7YNHeOYZTm
zELVh5c7@#Q&kV^|1W~ruZMEnGT&{rdbSgni1->sPQvt6VOzZpoQyzE(=Ws`{JP@7x
z>C0P9SV{MMc`)XQ4Ty@E>N0pJS{zwK4qv_eHOMoZ#VRK={PSUNkeC(2VGwKtA~Cq?
zk=U1wNK~qoM?+;nI1KNOZ;;1~D&}~2-2RbKm+cer>3}8{9Dgzm4ch{oi8wKNDvqfF
zQ@N_4d>|C?X=f1bSt#BRD2#-pz?f6wc~mE;JYyZ^fJ0^ESz9bXP8oU5dn301k<%07
z`8xL@muOe0{TSqd$%uMf@}kqix;LI=q}qIWDHxzUi&X5OVy@18*78cs5T>*Gp5d=~
z)myXcKFynMy2)$N00Ww{@YWUOb${8^RExegknfF>p`&6YZiOH~@TLuD6zJO^Z$<BV
z*#}3zP~MK^LP@boURNgX_**EwVCN-toWI4XJTg;TCx8DRgP`?ec+I=Eprh%6hj-yU
zZ%RW}z%Hn<A@BRg7R(SNB%t&`bkdJ`gdf_!GEImwZIq9KQ&2bw%FV|%SwzXN*>XN`
zcm4nJiLGv?p)f^0^{zF725d1OsOmr5-Xig$pth+;bi&Z(MP&1z@g=r~m#>n~f&);t
zilX*ku`@FT-0+JeeS=*-PlAr7WU0<&yYh=*j|N<@^kpIraYTAwm&jNC46z%qW~foE
zl7Gih0Yo<<!Wa)iA;qHHU#D+yCg2${<^LEZ;N-~5H}*TyHc_{pce5km_`lw!Nln_U
zj4fsCigAV1D^7?hP<dB4N4`^UNT9JEwPMDzvuy5n>lNw4>5)Oc!#4h2M5)p*K#%NN
z`JT5*G>vK8eD?i%4^7fcDRB6<{y9-0?rL}furSNPs@qvNkYjUk-fmy-2<b=^8gG&v
zV#J$KrVgLsqF(tyj6`oQxA9Onn<+nxUbR3kB+8EU2B;cAqC5?q_muZ^_rhNwKS~n-
zqu>`N6#h7Myc#1Vg11g~@@MKK#<z48|0KrCBGOseIeoT)wg{qua3!}tgC?vNI7hlm
zy`ik&Q0Pk<o&;W5s<{Nb9qVWQO)K0caJB+oTPHuauF8Q3Rsw6<wO+wv0EkUUacPua
z#ByaCn+vjAy@j6UkTw0wdIyZlfEi-r;Yx1%6^H1-vU>~}Y)A76m)gm%gOC-~SbH@7
zn{SO5+yvRG@RT1+zpZx+mDMG)4ZH~VcfL$&Xek4-QZ$n3WRD;Sj!1fjX5zr#Gd<(w
zaa2golfArCTRpRKV2F`Hc`C@>(Ww?+*7IziU^94K7(&Jqhb*^N_O&S&8SAy5H*8y(
zhAo+qz`g1tvcIz=9O&?S9Z<j3Zg=X@wTpC0<7Dl9Uhp_e4)kZ?h+&Aq?|mSe%)u=W
zo-;voPz+JwC2mX}yaYB6_FrQAn!A^GYIKfX!L{s_$%u#evsOS~3?EzqDt7S}OcZi@
z2qIMR*EP7si-mU!5}<!N3YlaFZ#&3EnYLKETkGvD3Q^QH!$ToGymcz6Gz=NU!trmE
zLxTe9=4AToYUvp}1G$m#2K0(o4A;kJi=}tGQfNY5Q0Y^z$na>WDu>x$8*y{);r^jf
z3cC)RafClJp=I1~WKd57z^q~izgaIw#kk?jC62*FUjqS;b}B@X4O;B{h6$mp5c-GM
z2jd}OYmn>G*E=RMLBA;y)TCdqDHJgVpwqM9d5?&aoF`?8)XY5&d8|G#wSfA^@W7`$
z31(u^FbKr?4Dc^GB>Klz`S1X^W5H&~Kp&0OX|t=w21Rcen^pB@_M%$PP^dP@VDHU@
z7&f0SIicMlG1-RdCRU)7W8wj-vt}nlY#f`~=13VRt+lz;?+BP*hNfkKlK><njxsF7
z(oni`hx;4u&4l%~R7RxvFV>g_jl*G@kc{+S42cEJv58DOJucX}qDYKMM)}B8BhVcW
z-9>%dM1`iyXm5$DqdxpJ<}v<DD?^e%<Av~N+m<;#{Q}bs>xL@m*eJ|{80){-HnX?@
z_1Vb@A!-2#LFI;+KG8=;jfFK7SwooXNkE-ze3G|izA9R{p5!1W#~h(%!hn)fqHU)2
z!<1`~aUlS!puc7i{g9PYZR6~}PiHyJ{yJ_iTm>E*4JF(#rZapMDESEDF>5hnQZ7@C
z#g=jtmZL_D7$X&F(Q)`<(IF`DC+|1Xze{DkdYuMlB%21z60KPo?^DA>VRZxT!uOg6
zB30IYO%`TwwF_HbO}0{UJ~Y_QPN;=4A+Z&2?825pC|gZpr1=~b^oeCs>(9c8j%#qG
z&Rd^>tvFZe(^oWEEm$ib;{B+jj*@~;mZ=lG9E{}>_59GizDy>@6o=(2vp}+n)_mvA
z<M4mdU=;7st(P=<Pg+zQ(s?GT>^8*$mSaD$WV1|)9X7Fvk>y31Y@gwpuZCs>W5U<b
zEK_2DwfQD#_IAxYnF|GDwVavQ5!H9FDt7G<>lvM8pH7{6)>-4^?C8D{dC^*PU}{=g
zaC}-?U0`Bja*oZmNh5W^H-EI|iU`q|CFcg?Y7jG}Dm{e6gU|C{v^En6htmaHifQfV
z`{)@L#b5@y4=xBDjp72tE&&t3W!&(92;@Q=3W{yvhEDL-ML{_zr*x*v#Xc9~X%2@e
zA9}d*K#)sf<r;W#Q@LDfTcQp~2CePOe9XG_(I*5r7ak7p%WY9vkNFB8s>Rj(KdF;f
z24SF+68<4VNF%9Lu1bW_Vo3_CvlHZYCw5t{O4uBj^w*>znP!5V0knf_R^;0Ht%{`B
z2+Ub%Q5xh`T?S~4oc(&a&bHd&dmVLzTpyg&KU*$0c=LP>!V<9?%q8MM@TU;j0Be!v
z9fT`KY%rMMMkCw+f*0V}o6?eUlhbmTBRBgqLxTWNVCip3Gi3C+di>;88(SX#k-r3Y
z4&yDM_1QD!HtV=QVk2-TYopw59kYxamQ3z2xih&&?u=>G=c}Z}`=N14JdGZoX+Bl@
z->e*vN?(j>x^oS8x?N^B4U(R!9b>ZB%ZwP6I&>zZq|Z!mpY&!hCkyw?G8Q>t|Gi~)
zz2ol)#i&EFwgQW-LBcUO7U0L9WAol}E%w;lv<&8R<h>iV4D<XKt|=Yq)Kaln&}HZQ
zFQxh`m$6M2bWj;H<!a29g=zMi5O-rdV;gl*%x-cv%hb#US?urf?AH7$2w{toX|t9A
zOk)e?kcWozXGxqtx%vq#HHm$>G<c_)qHCFN80wU&)Tq&DrS=$DX9w+aBL^Z%YGs8_
zfF>%VwA!%uSEgq(I03EiwaV5U37J5#R>!Pyda3|~SXpBO!JV@;VA{1wuG7SWx~<j$
z#1Oh5+O;mk3CvDpQe<U)oR|Y^nJly+QJnEt76_rHh-{1rBi^s4Y_h2g8H1WUcLjBk
zb_yW_ljv@rXd|$W%`qu@y>Zg8aN3rjKJ+5O%=Ao~>DeA_;f7)cAy>*h{w@V3W7rwG
zVC1~F-q9|s6r59<Fwr1e)B6^*v?#d%?u%1Eg)+shC0im50<haMxj*{Ovz!n%qdZ{z
zUN&@qJZMwO7vPeAD5iuVv49bR>*cSh+m46_ljsjew^X9>Hj}nXGMVL(V6Fx@#ZXop
zLi$mobrHpr;A4I)w*4rio;@D_sOg`PC(^DiMTAmBn>^_Q)jduMg9Dz5;iw=~B?G^7
zc=exl=>vnf3gmw#vD`X}88|fO%d;_t9=N37xis;Pd{Jt5Yvg(Vg?pJg26bh_*<J`P
z1k9GJhsE_`2o7vcURU#y5dd_W3A^)U9|N~Oy*dnGcqRCDOuh->`>zHs)Mk(yO`kHe
zkY4K;0qhD?2f{H+6L5pP9-kxf2zL>_VKbU~8uf=agKCgR<W6dhZ`E&gEW4PmVx0%*
zr7)A;P94y#$}|p!Me+!I#~U%!omSRD`I|q3IKvr4Q9M`v9s?f_Rno*L+I=_ZftxIu
zrT$)td|{XXJ@qu?X}__GuW1<X#}>t{4ddklJ2wtIOFr~b5~e}HP5^^+OVBCxQ5w-=
z9tBVa=g7y2K^R4<$R9{Yb>0BGPy8F){)!EW1qFFN4RQ|40i%$A#7?SXE_wkI`lk)%
zl+N;53=)+=fkfK2D(6oB-2T4hBIpEPqzQM_l!B$Ckecx~5AZKjx7kXVZD0Aw0Pce_
zG+Nimzx^3%0YN5fQFi&-=@BOA8XyBKZLvZA6O1_o_zuzJZ-PLGIz`1YG`Ug!o0bfj
zBy?(U*+lYNRaY^0I9ZiR%`BeH;TQx*OBxl@<U4r<txz_sDBrcex3`=yM7GI0ZmEz1
zvmzJB_wt5Y;<^vTS%GU~u&i(MeSaGkF1Dt6US!+!_8RC{tg!=U9yJ8n&fjM24J<#L
zdvj!adj{lFy%UH+F31jfkBfwSU<q}QCCvkVX!$`t9PeCPT<OEf|IoYTw22(2n19*P
zn=3uS881rZjX@qi@-OvZP|juuB@h1}$Mirl>TT;G33kdWYI=O363S7f{KRN?LIyH(
z=b+=TM)|3KN^^<65mx2iCGW^xwzdg(H_(RsS?YmA<=7X<&pSTF0Z6!YmF()zEG2O^
zu&J5yi@bxz$3jPt-J)?g4jQsW+N{5{q2j)1EyVvTdmn_Kjk)a}E$S`1R@?R0dBbqc
z;?X4r;dzW(WUW!<H)(<S3&#wQ1OUJ+`E7LA*i?X>@H>A|^=>r8C?Y#I_vly>4e%q6
zQcto=YFlN`d|=v$T&e?^_lgI{H9=>EF1&YMP2nR>mwk-th>fe4eeLhf@F>WBsdonW
zxmQpDy8Y8H9R`pIqXY5^i}gf>(oQy)vAJR+qxcW>Z`?@+{|7EgQ~~z;{8qy(#~n0I
z4sxoJQq0?)iJY2)eX^8o@D~Cg;#ORwfx6_f&LzZfDS%a9q7q>f&bmXq>!g-~-fVxK
zbcwF}R#iz?@7i2NggF;UH-E<Fh{Z51rh9N{)0xsEufX<}Yq7(wMcnwI{_ksWPF{_j
z-!pdF5GbNZP}iN-%NvE~6J`i7Q6Qwfjd1z$y!5f9;$7?Z{$bWAgVx~CY?Q-0XoQ&%
zi&!j2lsqGK82)^uHAHiO<v}ausI&&yO-ceI@mx7N<_YD17#uHC{DYH~mp!sct!)N)
zgt136;4khQze7`?TCtkCfd&)kfhqmGP0~|?ivWd``7?SEcz`kEw|{Ir4QkT_=K*QO
za71vRoF{l;t_(EdXU@cQ2Bi+5f#Y{Nf(9EM%Lb2-A&D)_gG3aw1elj&^6qNDcEF8r
zIN^@<FD#~#W+pBea_0*)9~o*L!mh|@Nl&J94U5sEUKLLgjB9wj)_~wXmcYWZ5wV7o
zkoS*ZzmYc19^HG$acQz<+%oCOcKIlupaQ@k1M8=a4xt4Ire4PQh%MWLnjn4RtJEzq
z6L}fOrxuvAWAq*y&k!O_-G~<N2{An4EihP5^moWeR4Q0y?j(Q47$nAr!U~m>z5j1V
z;HR46a!Sb>w8A>WBB3>^*Jf3Zi`fiEe3hIUvw@^TWKIX=ogSx!c&w;Jx|E#$Ey9_W
zqX`&tM*3wPlcmTx7%-PGV{~VpY`OPFy0QW>eaOaYCLAh^)o}mi{!*D}#<GL9I@uUa
zCDPbXJJMPn-G8W5#V-wTXg5oB5CH;OFO`HfS-MEMApY=-MuboB(HqWD64Wa-J`4_%
zGeX*d)TW_9R$^!yh$IG60u8HR9(q>lZFt${tmJKCxmu`s1wXND;*2rZ;1w`_$rG9A
zedz55IR7?587cZR2IxX#=IxQvQ1T3CT07(R8)NqBGN2iQ9u)3Y)k>P;%tT<CRnDeH
z%K8~}BGBl70Fg;S6K#deM@*MD@=f+WF<wYhFc{RTDaKYsX!u~OW`Bc{7J9Y<`Dey?
z2x|ofOog14c#ouIh!k%b2Cjl8nKga3NsERiUYCdq>>6cinyeKJz%G?@f}fG7I7$9!
z|7x)J&y8UyUP0A%g`Af-1vuq0Ip1F)R*EgA7n@v=y4k4Vs7Mfz3lo1+SL4x40n3cD
zi~Mc98bQ}^fn4lwW2>v9S20&brZ4fgxdrHbgL*2L8h7Ag%3gAre+eML^CNh2@({c{
zt!in)7t5IG+Q}8(GCglNJTTcdxiXQM>#CvUz7Lci<f<6E9yq$6<!XNhiwC03gw8eb
z42KB$gW0{;rf16(lw|nTb^eTgVrri(iE+Ix3BpWhq~8$pgJ!`{0_M(Wu{-&bj}@c9
zfd^IF6s!bQ{@L3`)`Q}{ZE)P&=&eN}CKXqk#Z9ryge8k|bEqA|`i_!Y{3{jjTrqO7
z+Q+wgAC|;Ndd-x-L?4PdQQ${#;%)wm7Qe(=qJD*%2(#t(paXUdONP#eVseLxgWf&+
z$ekS`KsTKsxK<B<n1WuZDlPsNEjS*=*c14ZXBBrzz+=-A*>F~=)xw%E$>x*k##|$+
zb9tHJEy4)ZIt8$1$UO8OVxTmRnF}{FNY|Jrd^RgcH+;kiGCTE_l6~-PM`T{B|IUc$
z^EqbDucYlc>HCgjf?|&sUoD0>o14BTlExjY;UGk#Q0;e}%yYsrd7Z^3766->pPo&F
zp$?gVIV`YQS21*MVJs7)WI$UjvyPdz$)e~DI+WMG!I;-j0ukbu#Ua#zcM!dR&DItx
z3z=p?GV^3f;t<Um0xZTsvxBAnHc7;K^3<)DrDxOB0A&#I@>rM-g9BXS$D{huX}GTl
z)fO%a#9`Wnx+@b!DMcwNm?@%GDD1ddvMOe;mkrb3so8dDZgtEal(zxXUz^}pSz|OD
zlSMVxwP}hBxHII7D>hfw`7gGapvS@}q#m(I)~AWmb))gtvLQyJ2TcehVyqkeZ3Im~
z?AUp~$)90!0y2b10Qv3;?gY$-I~eXxG%4n*a|oQ}B5h8iWw8Q^LUmpMVA|p;Lt{qT
zqj7`uaKg@#dz>VS*-$co_GCgTax5KjuQzNdH15Dq18r+e7AFY5!PVJvUm`FPpwXJA
z=0on2<=4yo{tf>HX^DPj$pbbYO{(A2X$t|EMe?A(s2~|$Tn5nPA%DiWAgjQKVwv8d
z1wPO^WiSjA<E_T>!~V@&s4+Zc+n`TA67(!!vuEVdv_Nz<5O@iP^JDR1844TL%i~7L
zLT-vY5hwd(RBWM;f71I=)c}nwfl2sOA_msh))xl}GL<nl@T@#--|%KCxbRS}1?@az
zYFa~neB<N{0M2JaZ5xGTegJ&Wd7C6Hh>kx2)M3c5bHOWf<@vBTkQNHgK#MQLjz&yh
zrl6_+FZzrP*W%}K#lTC!M%cHsa(RlAsB`3HqXo>%NzA-_#kCIeM7<ib$NnI6&BOaO
z6QX5fP8vBtUJrT!Ek@CuMtLJ1AaWWpv54Swp*?cIaCysq%vUzo;|?zr+<Du6MD46v
zc_;NnK$AyftNhKM=`rnx_21(?jmjVhxm@0jUPooivlSH?dC$qIMhljWu~eGxiNB)=
zmAr4$gk{Sk^#f-B_~JfH+k%lKWM^Jh$7KGG(yDjV+u-jWuOc7&xTw-X0^^ri@`<fl
z0bT>I=k#f;8oj%?vJMFc|FE%~w&ncIi%ye&rfwFfzuy8-pfh9q)ImNA-YrMcRi*r^
zV_`8~z_d{5%#ZbXjK?zRu;~}e7yh={sHO;;lJTXH;LvWp<tqbs@<Zj{HnR2>q(E27
z*WMu6fIuc<+~hyb)~I`}wVA$2i?f(p8QRq>`ERr`7-?rB1;x!!|Bi)5zGMGFYb4iq
z3kH9Jti*4y+1g~ALcoz=2`mUvT*?<Zr`E$I`MrX{qpY9;VT-Y~y2$;$|ALAP2v^d}
zu<#3}F57yW2sQ;>S@f1B+&XSx{e8RiyWOg%S7Ru4&IMm6+uNv7#|0&+WCwqn2Mz;R
z|EDSV10RRO2Y&Yx-N!%l`7%&e{r_-6?O5<=$C0SAg5<;>72L^%3w7#H(aK7H{J5a7
z7rbrQvpW?O6rF_p4qjmMlW3s~@G?AnJ9{q}8N3s4&A_husnh*}%`>s}b_q&Bz@DqA
zCTeNoi27N<FiEqy$Tkqbib-+}SQ>sMa;Se^usq$15Dm9$%$622)`%F4Wm^9t^+JUT
zkPwsAHRfSAn+?j<!ISe#n+<Rhb>aNV+RoL17Q5Tu+gnBrl3)9*Obwtixlw*oPyjJs
zT`Rx!pK(^JhY_A-`JE4-bcx7h^JEVr8HFf<IvZrqG>HwWyHS%MEW1qh@?WSUc;#NO
zInBnsV|rQ=!Qz|9E(>HI8`kJzh3xAu84Xz)f;JEBjs4OpS?o2gNRa)5&-xN0IKbaf
z9z>_S@Vb)?9CZz?mhXY_#g<#p@$c;m!29$bvjSza9ONS~4`d0}EC+l06dzjj3$RV+
zG+nyzRIpGe;utt2_!1ZSF?t+mBMzl59q-eU!l6h>u>^$cYHJK3x305vvxThZ6%XBG
zy$9x|K%n)A8H^e*dW;-u|3Zub6h-Ojt<xq$pc+)OuwQybBLl_`7(PaN+h;I~7b6p>
zPeIY@L(Y=Jyn%o!7Qg@^nz&95k8Y1ac$plL1VBQ=I1O-tnrCukA$$q-m3qfwgB%s>
z2XPyUz|r1}0bGhMf$v#P#(W=t@ZU*MYE@d!8#JTY0|h|(#=gQ4*Dq0);S=0_+&v7b
zDKqveK;`N0gYaVyh}jw$;2$zZm=H%trI3M6rCF$Ax=IG6<>%dj6czwP{ae>=Kq+e1
zbP^7+PJ4IjBFFfL$`|+$fWI6YGEvPD95iZ|+QEt9W->H2yn^_T27Z9F*ha(ROO@m-
z9p`p1`3^iiJQfO(g7q@O`vyG}%iu;FnVunCP$@ZGj`L@<#etgh1d~zzjJ*>9Mqq)A
zPTy{XwX4vnj0sLhQcVWFUrP!+J~$5PtEi(s*4VqfrSIteazgy0ThHEdqEi`~3LK>;
zc?XQFp!Kl;*~}-W6`BdV*Vah|gq#wcls8pYHp@8sm{>l{I0U#e4$>B45^ROjf;7W`
zOUdcpYJvt-F^5?vXT&5uIE}dm8Gpu`Xb`@RWm_iY1@{9Xuo-F+F2#uleJwCkDy{Ph
zRLak$X)=U=e3~Xzb_K(wD!PD~&WdY@PL^sPB-enBx+M^*id;%AXxP0=FPUJCA75X~
zJ0fcmz32jCE(wun-bqj!8xHPvhCQ;Lk0}cjLaU`NW)3A#O#%qsdOzaKtH33PneZ*h
z3ITtT_b-w6#O<jtV4<KsfHK^@G9Laf<YY|@xq!;yKx=5S7mY-F?X{OQBmx*df<L9v
z=fap~f@n4)IwP9?$5@e*E0g@0S(dbiE|AHIgbShM0#9+<d96%|!&@bn`J(MWCZQE5
zMf^~+2^kj5mosg!g$m?f2mL4KCdk<_N=s71revxJGCE|c|Dv1Ip;;^@IQ^VpmVQjy
z0ek+@UuBFaGu@ELZxfbtec2Q}gFd#5AwHOhT`oFD&WolhGD!FbQO}RnK>Y_|D-4vq
zz@+?`X1OqiJ~}f+E=nY$kC&d(nR2ncP?^CkA#zEg^K_&a@Lsq1a%uW#A$AQeS_C?9
zSqxv}`pKkRFZ!S^k9{|^86xQle=mXyI^0iJdaJxym&gXLIh3o?+YGSK3W!dWt8FdL
zgn@cZ8jj8Zq1!9vTK|x`Bg+VE0=~<j$r*B8bOQ=QhUpbG*V`hux4?CAgYQ)0#0%w5
z_6vMiS^2ZK$;Ct=EY;A6{>+jaOMao2OqfrD4CN;OMKw7*n+EWWC-V}yIrLCCT&URH
z;^T7n9c1tdxz)$gf)j9t{Kea%8N(<S*gNb6bGj`^JcRD&+r6dMc0?m#P0s@(-Vve_
zVU@@lyVIuCw-V(FS`z!<P7V6crrG*uVB}7b>CyDCJQSE2-Zaliv;W#;rmaNJ9$jUY
zZTa>V81YQ(!)@8|IW(?0(i(Ht!0-l{V?8yMqqvdG_307Qp!A_v%}6ybk(Y-%$xjdk
zx~t4ji@5=&3XB{OV?i8|Py*=Sq}pIEAw{8Np|?gQ!fN_4NR7t+qO=4wQn-M#WU);k
zgL~?ML$sy&ldZtR;O`~z6&8dwCf6dDVQK1vg*7>@K)RL{JU)b1dKJgp#dgb$4)usI
zsgo5>ex#x|%gX2w0-7!~_$g#nG>&I<GOt#7(TdZ5wG9^2odZy0O;8)!XCp#u*Lqt4
z8Znc2U91nH%!{%<O=S@;;b5-2YzWSwCYzOwvBW6TT*;_P*<??6{c)ih1*jY3u5YaX
z{MJ~t5@S;Cj_p>D1Z$fuHG?s-CHhwY7lIN%aeJEW24Hl2#)4fq2;`oSi(nNmV*_e1
zhKnq+Q110FOF&Zpa;yJh?oxs&nko-J!+mi?A&(CQxri~|Z~T2c%3M6)^yRK!E)Uu?
zVA^Noq0oW=H-Qd}v4a=&uO%ZS6D|N4N7DzosaBkN*vDb<V+NubD?SqAU`h63rWvaL
zqfQ)7j*cn1Tp#lx)Ys?aap%-CG3cB;5xcsOo0R7jnNOxgAt&*`rv_(S#cRT!N_6HC
zZ%6{WS|Cqb!-)7~2<$Wdo(%(ks%}o`>shA`&L+T(&qaGsqli2(C7Q8yGvxU+l2n`<
zBe7nYH}Hag$zmeZZr9;^^m7mHyYVkw5r7C_a0JT%Tkh9OJ_5$EO_!I0bAjHWX8kMP
zEZLvNS4S$|tHFcadUluBTrIGRN;C_UYX%nP%BjtcxxU^p2X(Dja6f~2G_YFUu#up&
zT@7geX6l-WByid&T!^yK@|L$lbpgLXaWc;m>Qf>pq)p!TMaP+f5DTr?A@5k*kgv2U
z|K{W6dO)4iEDTRYDw*QnO%`ln@XR$(a^FqNWC>{8Pw)9NidLdT%&>dkM_%IEA=iK_
zK1eOl9YJTxa{17oxnCUuc!7MBS};H7T1YG6B{Sq>W8SHqjvObSBs$E;YLn6o;ASv0
z1^7n!)HdNrMo*S>OqODmuYU%)k2xM<<g?)YY-JPE1ms`#^&tlzBA<s)9EMv)aO{5J
z!($Q<BIR*p5|~>H-Sf+6bIY~J#Qn<O<EcTdS*=}G%#?p8R`AFl>`tAawliD4HZ~le
z=RHT2@}C%7gI~JzO(Odukex9jZSvp3R<Du4Gfd+-zz~RPM9=i%QAw~kkGUcfRnPWn
zQIuHUnP{LOH?g`ku#Ffg#0ncowUzHqR3H(TwNxO~k0F}1vdu)bX8Q78Nkz-QH&M}|
zA!TFa`}WsiLq^NC6BWxElZ7g`-9*RYs(2PO$oAGOb^sC)v2O5>$PPXf(^;s8X{fy(
z$&3^HK@0^&oLN|$g?oSK?L<Hkb9{H4xU~x}=+S&kN>XZI(BP)2d|*vMo(qher0PKY
zld^5K{K$xd^mv5caG3IA9}#oDnHoVc5ZW2iLS&~HQKr70Y6|z*PbMlVTP$amq3rDA
z<BrpNSil57wKp)PHpFl25|m*<65NsXvxy!a4O-(OLFQ31Cg<lq6bi?vV3Y_2_;^>N
z`<|UT%P;J&Wp$OZ+e8I7aX~dk_DlN*%I%?e^RFfv*lP|9y;q4O!Lhcxtu7TaAYJ(G
z*NMg`F$=2F<@%e63Y50DlxqmkZ>?#dMANF|clH~QvSxn1$3#bX#v&sJ!Q{}w_w>n7
zUR&e<OU9;X1lh|6XR@2%I_w+hrn6*kn_IRvFZ(zdk+ut~d*2u&*oH?4oN=<B4Wz&M
zNMwKeHdlY99N+`M9#-9yJ~3MN468s0vvl&-ERK$c?Rq&d$k!e9)PEnVhDtX%EYCp`
z9o>beh!G5W^)Ed9jCZiGUrNn`(;}Ly9YzW|#JTCZMrZ$>GL4S+MKqAr<6_F~A<^qd
zuqV#fE;dYLT(h79N~l63?`o6otj@xP(#<wud&{`Pq`Uoud=;Rxb<)Ep$w&oD-<TtZ
z24NV*rzj-~r<G_SJ#7T4edeIE_cEFe8!%dWr$w`<F*7_E4AaN@g%?j#Zw^bdL}Dz~
z-O`uTPJ)nJE{DhKW8v^?lq1rR=xXCu#3n=`G!lqJJu>)9;|KLZ-J@cuP|I14j<uv`
z8swY@wfGPIj0SY5h<zuzwv9?agm{r6*ev~`bBKI^IhyTG7dNB%dDfK0SjQtxwpsd@
z>?$5gh)BSTo1kD1FhapIC<DDsEAvS1W==d!58GRV?A4yVddOgJ53zm>FoE((hGNm!
zAt9{bp~X#Z2uD4}n<cC8EmEuGY6UrVqMZ{4RD<m>Hr-U9DQjFpnba@n@*bLKwiGN1
zg$cgpVTlwh>`FRPmEt+Zk_N3Z+&Ha*<wi#Yqr!U(fJ;XDx74Ss002y1ZfKB{F|5ZW
zR@A5ikYuHdvS+wPg4ukyni2=%Xm5v`hE^9HBMi8*vVtpQO!SLya4(4*AF>Qbdo|K;
zWo$g?hx$g1`C2)_DQxs04H)o7zB)1eLg0-(y2>OcSuYjL%;On)a^fbOxEK$s*PY_e
zSfXSTX$zZiT+E1J^DyUF!06Oi$1>EP0`^rYry292U@4B;(-T8kz6mfFAKn?pc@;TE
ztkp}#`{O(q^4A9Jr93_cq|1nOW);~;Dol2QtlY$vF)oymLX3DL^L?`Zt&3oUuQ0-h
z!SUl`a}_Z4D({v`3<1QAuyv~Anc7zxY9P7zXgGG{gfvz{E!<A4q{dga#MKT;l3Ucr
zB(<^88uB?LQP){}h>x#EOl2%8Q+$d*HS)HDM)vJ5g^B91IQ4Xyn1;ppBh}udXmn_A
z$(Mt>8te@y;!Wi&smZIz8?B?!igq-mEx@cJ1<OFt^=eroIZh_|k;GbBYh`i>6abjY
zz>L425~DDv9J<*eY4&H9SwNL!iJTdnjM8bWR&bV0X;7vpXD7xavzF>3Q$yGfu99;S
zLA8V~&?K_vG^mIy!h*|j((i3*5pzZ*F>7*e=pjrAM8Y&H!9gIN=VPVf#-p1BZO-@o
zfJhsyIekImJG*B#L11#BcglUnunA`Ncx`;1#Bx!r@^MXBh|`M`vDT_OyBxOXB^}}(
z8R<|k8syU0Lk%p!M4jWZG|K7VFzyM2N~TdE3`eDClvlYTn&d^z>V%Rjjoihma=9v4
zw+iW)SKB{OfDG2fHNkh~x%~JX%3)pW>yILsP|2=~CCrU!f-h9AH&#64bh#nT3)T^a
z!Xn0h@@Ldksbw(%{m*HQIuZg&I<ha5?QisNxT!Ts8PmKeHm(Ejw4n?^xY^%kg>njT
z8YH)*Ub(@!oEzm<pG7NTJ=^3j-kvF&K#&;mOXii^Cb~bJI0ks65yAVoH`{oS1=8dW
z`+F_^BX_28vDhG&p2CQ@Qj5=$b}Pm}=WyEkgpj9MD^sVAJ4mJn$yu9_Wr=3^$CwV9
zWyTnT8afL8%#06E-w7`NW|`$PRx1>=XY{~qTd6Hq!%(^g$6sq|l-2=9N$;1{A7CWK
zoDNE2tagOWO_ZeZL`ZU)%nQau5$f#ZJOHry-ZGC|`g85_Ksp*Y0fZ=eE4Ios@TSW`
z|Hf9pAqoHTT<nNNA;bXP6=iWS^QbHW9V@vT+Jf{Jc&ruf?TlOEzd*(+?gnoI2GR=n
zWU1{$)Ph5rQl%`5O@LBTvfRJIBHjZ(D1&76beS(Jf-n}6fat;vveKUse56%L$Kt9W
zFc_C;r-I<D--6b#cCy;GUuHs9)|AYikt~dwEA_Rn_20=+JfL|69fm|L9B0xDS?A}Z
zrHaV<G#btCfCplkZ188gkji=+Wn)aZd)HpF$@?%2kY+va1WDP+a+j?D;@q-Oe(nw;
zcn&M;fNeG&E=L&;+2X$=p@;GtoUmBhW7UX1Ra5ztd;B$4nI(P<uZG-fqk{5+cD6=8
z*y)Jn2Pv1xeQBg6?sh0zlo-5V@c#G)Dh{Cm{3#FE*fQg5GV-7+i9Fkg7t_UU_A>q?
zbMS;mTZo5~hk`^yAq)Q+l6xppoEqd|?}ww};EC=$;?KbQU>mE5`lvl?ZvoyhzNuD2
zMjrF`Oh&M5_!4<Mwa3!!t%x?-AW!%+%TF)|(vu-`2NhZKOrCO9)r-?GZdS3?vN`g!
zz1c|<Sr;B4&lt<$2$N@HV*RsK@|;V<7Mz;%<oRe1?lPR0IOmx@MRoCo_yy9OB@=K>
z2&6ay0OZBc0}vyJ%8dmW<Bjr?zphSx@EkQ+++P8h<7Hn$ZuI}9Hle(d$U|rlVOG$+
zUky^ALmu#JaiF}G-lKwIBzryBLvN+wVlu{>RiDiCc*DQpB@o1tHekwc`bf1xFWO=9
zg6@2IE47Ae`>>#Euq`5r9Hy<j?XB@PUZPP;upQs=Z_Jmj9uTq_HZ7MMcXtOL74j@@
z^Y=t1bF;7{J6^=F-c60SY4lVp8h<Z#5j8HvP(pCM@BMuXlvHKa(2ft%vLTNO>%fS2
zR!&?g9~wiHjTs~#d0!+I!|7R|@#8pH1|hc$D=VLbuxEUtUK;*s9FPso^%d1cK<!MI
z`$xP7H<1-d`=_%WN`-vp!!eAT*g@QC`Y->*6msrN1Tj;elh6GZ#+2C(W$YYJ&M#te
z0Q8u#X=TCycF32(;@x_6m#<RO<Okb!rXKt|Ca$>y{AIm-9nT;Hm_C9HF8_&bR|_Mi
zsaC$3xYceWX+Vt+plMMS^S`k`b-3<{(H@9mEGk|GxG4wV{X0d+35zK2i6GeT#;^5F
zu->*QDxwG=p+>%EzhI{=MSR!yiw0vF@`3}5VP+`G;>m8ZZTg-13qgb#pvM~M#1`VY
zowb%jQ6)rnZC_M8^Rxp`mmP|Ve`3*jUa-2wQu%@XjFK|)L;q`Sz3f;FL=7h#c^ZpB
z#SFzFt#JfiG5hn!={<9HFb<yY&6AyqTP*{(d)Myr6Ki?u)PrSb9|PDAXoa=;(-;?y
z4|;eANA6<Z*Q16Ea73T6C5%#^Ab3KpgBDAh{M;Ha?5L_XUyg5f^+p4WTLc~R7cr=A
zz52**{(^;o0{RE?_RIJP!jMK0D<}S{=<zO9fYez6XLp-T&+gsj*VbU2UXuNr)OydJ
z^4p@K(+IKyLPeI5-#M+oc{nl*(6xQLhxbY;Njrr1IqeyZ4MmT86%COq-T0(}>&`B<
z_l_>J^|<DY#{51`w~|toY~Id6oMQWW#}u-bBC|yH^I_T<X~;6A)LPl!ThqKew1Z+E
zU=3stIiU{lQPPhR7j_XaBDh7RNJk-^f_qi#)AIQEezBfq@XE=7HgQ}W3BBR>u^xzn
zflHr%%0afRK-UQfZEcW)W1FF1{v^QYo&6ir<<to(ha^5tt`&631ePxTHst})GsJI5
z*OIplR_FPF3k#%MYMXZ!^Ba_k^k*a$@gr0*l^*`gZrx;Uy%}<-w`w+`aRKqK+0ruz
z%RmifY)hb^IkgnQ%)N}^Mjby~dV3=#A9*HodqNIKpO7wt@+dTTm`jvd@JRxN^bQ<-
zR_&+tp7W`$!t`A!M-&~_Zp1wftfSW;NBUxd4b*HPv++)4?pY?bar@1XqY|Z9>q}L9
z7uQFZ(3NhoS(wKk(le$Vl3uf=uRlZBFg7(&gT>O%pXnPqz1dP`6KcppmR3ps_zD!`
zSeTkVz^GJ?h~R-SEQRzS{sDTaHZ6nvLmrE?Y-lkJj>bSu9s&%r8d6lK<;c!*OyXAy
z0ZXZ1y8H{|SntcuGO`3UcVzg<P#aL+vc59RyI`6DLwgZ9DZ|qs*|3&|)%h=y5&kwB
zWB40EcNyu=Ec}LYbG;mA9puV0GAhKyu}$?dx~L8f#IaPdy7m}*tq)Ge<5MebdT6la
zu|5u(h6}KIdgT{UHaQ^#3WM^2S~kmxi5R$70rSv5@g1kkFHZ8`t?VoGIf#hTqv_-z
z77jJNG7Nv(8E{Ha1Nswho?!K8!a<v4TnROZjnV`;H9ezB4w{0Vr=_vcN2zXHyiI>0
zv^5C-)BPJ_wNRWff~9gs$1})e6dQpFkuiC)1+^FOza>(h+T}fgY>TaPX6Aga@ZWiE
z!Bwrw%xrM8d$y!9#G)GA7QSRg9jj?lgaVG2tgRVShl%=)_eMEgs;C;wQ_^Uat!XbL
z)K<s&lqt`myhFX@d{VTrS>}ZS>Jx$odPt4!q@rGjBenjrMZc>CFLnNmhCc(kHe#RF
zhj<w|5(%dHqC57qE*$K<$|m}o3pNSpH1N?k%0zF6M+K`q&%oX;n%V<bYhD_P4!VHO
z%kjT;Tdu>_YxJ*tD>2Vfnw%Vb#<FOvJr{qU57|1&TAh$DZ;;8sNd$JBbn4odDRELD
zr2ue(GzXC(B<Y@I45yjnD(U3R)Do-R+RCxl^>S8BbX1nk^s~Jo3#}ps#vrkj3LsOX
zV-(bB%vZ=cA#W$;il|}FVxfQZS+|0}c#6ROkaJVBnqIA%zVkI%LC%YnKNfE3ETsPV
z(JqkbYLt@$gmr<pt4cJf0cu|8^Wt2H9;P!<F7mgHb<7yN*xxQqno|CkOM(heZ{fka
z)D{7k(w`=m`A2kUFz%hf<NR~3TyBiiR9B!Nzrx?+R;OU>xQg@2APA-kr4VG<C|4OB
ztFjYutr85+t3yB`SuWdz#C*BN<aYm)<XY<o_tw*Nx-Rv@uwaXEp(}KWTwgL$EI$-K
zI%A+C$qgY>yB*p`{^YMw;i0r<jYP`qwen}*-=HRBh-n>qI#QZLRv0jNV+^%EKRHvG
zX>O34e07blD5>b4WN;%|Lc2Hn@BA0#9(4oSozw~-apjg+{Q?qtNXuJ|{;=S(@|W1I
z{qiX6f1CY+Vp1o_?e@!-797bl<c?^h|A2vgj~_Zl?liKuw+ulN5sSqjQ=-L}f))d-
z;?YMvElD+o;gP%ODrPM|TEG^bgloP{>~uPNoXl_nTD7E72ldQga|PcR*TjM&jB8MH
zUS`D#c_JwI2&5%WF+14Rs@_1Pua;Ju>)^h_`^%hIG-#w0$K)<MZsx|k>5gL%$Glh{
z_>3$#alR9ZaLamGVE@WaLG{f_S(qk+a|DCMKz&Gy>^)SHB7nR&Ha<XepuE!N1LkUT
zl8`vEB;Kh+bT|ydrAD)I*oDiiG2{UwYf_dQZ%@q0ir9E4tcU{lE90NAPk~Oa3aWrT
z*$EHP)jpXP#p2gk7j=bdSsP4V)_07ovxeY0(_o49-VmJk(AVfOTPGX*S-_Svv86Yr
zSy{DyR!!U#?e@=C%3Z#~B|0ZogDSMfZOPqfg|W-=tiz_*98E*r1Tf1#WlM4EfsBda
z5pGx@Yzy)OB+NYBBwXiGwWB=-y?F-g90YTtjLkj%%)lFqUc1+y@erq10dab=)o6A2
zxWnbX7!1yfY1juSKqmLcGr!4z7UKh{8QgS%>CnghV0uQ-Pkq_)kUxtoVMq)4tFbRi
zW%Q7T{ddEmG!9B0aZ*|+5Vdpq$5U*UJQ}=E-c&^3?qfk&sK`{q)4iPXxHrytJLXHF
z_Pab0$9fq&9#%1tX5^FJ2swaI!zOrTZ38~#ji87N;To+e@U*{e^3ZGo*jfNKo{271
zsbMbBB6-%InfINtZl*jJ??FH_R0&sS%k%!sgIY1;CGvtXHGG^<P~}Bir~#07F9i{B
z-7jpJm*ZOq7za$`m3Rhe#T&jC;&^#A<_Qp*f$mnPXeN4pt%ET2R_9R*+rC~*LCuVt
zg*ghZ;v3d7qRddjGT(?CwPtzKH?mdke9P7t?^ny)-=Y@I=M=v3j*rkaPH@q5`J2D3
zMv6KG!teKYn|+Ty-R0eorwG}gjk6j%gB;CQS^na^kbUJ@*1>L&_hWminGUbDCXsd2
z>G&WTRE-Kb`jiH+hs^?BARpR%ShQ0<O8lA9%=NgXSA|s&K90UX_ek6=)oDKQS!=El
zHuM76$Df95&Q2-RLPP?04B9|<0mg~bc)I*EtsW6vJ2i<it(VW@tOCZv068m2#5>=i
zdn@h?ctZXaWEqXR<G3zTK98Yj>3chLzepXghO$WJUMOFtDYsEwMhF1Q(1FtU6RC0X
zRfxjA{rk(mJG2E_6CGzI{@zjU`|Duf|HIu|N5^?x@4`Dt!^{mclNp+%i7d;u5>qTg
zI2lPJh$W4p(THrffu>0cfi?_{j56CYv)fp%%gm6JnVFK@=h<*((%-%7`{P^ppPR0F
z+3&pXfqnMbXYXgj8PkMV9C^cga8%T_q_xnP>@<7RN5yarsvxYh@KzcX6$?U}ELEm=
zGSWFElHYd-q-a%?9@sDMkRP@s!+p&6g!-Sdf8Jn2ZhRPRnAv(cz@Jf5qTYrKVR=Vh
zk;XX}-!2E{9sPo))6`hs&@Kn%BQ{gDB+Of*OEkU%1C)dF`!w>?d<YcYJg_adlD;<8
zqPVKS)ekAQMavU%m}#Gwl=Il3HaH{%tJ@HlMCIi$pB~eQSQ7>$a^f8x6IUY(hYEr}
zlq2Pcyh5~y#Hf`ceZs6zL5vR8j2sn2fFFUkAX1$?I<F8gDjA^Sqa0%-8I35rWBp~<
zaj~6u%W-K?o3Z6cSzgx!r-U4DU1I}mk`wX{rIzP9<V0_Vl%QJ7BL62@p98(H`0M24
zm=APCfaf}2jiZuxI*m<7wS54L1#5u6YiyM54~BA{k~h?wmnWC}PfpD{RHRtAlJ>N`
z0;Dy#)*E*=qI&D;iB@p=>iTSzGmH`~*?Z(nAE3dVNKI35gA3T?tm22*s0z%=*}m8m
zOO~m{kRZa%WReYXPToyX_(|&Em+-o)=Xxi7P!fn)5tiX**(v8mC+hH`60uIsPea<6
z%n%Jmkqi76GsJmjk=;QpxiEMfkxmF%1i*ihEmV6oN?FLo-i@s{_TSzI<&x-{zHqE9
zD_C6`t2-3eCR}F!K*|l2#mkLkBZm)>EBwPJ@!1m?A!<X)m8k<B>UJJqWuN3{SIeLv
zfNDmzl{6kh1{-goLSa^hcq1SU9BYc{%1~=$)X2dy%s2}(-U>N9+?Rq18heZip>(1G
zDUcD)U!XR`kWs$0?yc=DF8F*IX)VJ$Otf)SaBw|*EHXNgiQzmZWZ#unOba(dAonIO
zR~zwhAZXQPSU=aqM{qo<3m>}MwKkJsC|~jJw2qAEq?EC0BD>`}pH*og2HL+2Az_B!
z^-0<Q)Wf3brN}@aSc!$Vl9f4)aMvifVvG-itueTSHD=3ggRo!*)G8d3V}p<?$!$BD
z=T}K26+w=w0^-fbd6T|vS#nCo`!g~&YZ~*?u?aTYEDD`e+FvWj-6A(6x}YcyhL2E3
zxzU~tpqRZTQvj}!iHToW`Hcb+Z$sj5l}XlES97yWw!hB9ohnme4h|nM{E7IoJ04|f
z^wE;V0f0~oL>Av<i>Pkuh3CpNZ=%$pz_?H*G6C?Cs#x<n)P-r1={`d0GU}9K7^&KU
zGrS>l<v=M`-yt;~Zuzw`)6P1$4BB(mg*urP4ObxW2B@a(x}k&cpWG6?KpHcxjVh@6
z#^qLDXt>Sb+P=-{$`ymV8yjG;-0sidO5*XQX8Il0H<E#~DlPBy_PKIY-IMt&cSVbu
zPHV1+yZsq6;1uYF#K`zFrfyKvz<Nn_NRc5UM@UV6-win02R<;M*WUmC|5Ivh5=bY&
z<p7{qHZf>TL#Nc4K)K?IOD~oBc(V#2r4V8b@#aMMBc;(7%A8O<mFR0{9c)Wdy(fqd
z{eX$znkWSlEt1*si|U{}D!I6Ho;M}3F?-t#x#n14q#WSTY>9u~Q)okcN6xr-_~7Bv
znwn+VFJ{pZaL1o=Z%MPPsle!xHh;#X5(0*Ft>@pio_awN_yti=(jIL`91lZ+h;B&1
z22j`DDjoR<wr5J3$sogGsZ?pKF1B`Oq8oEW12MuT1&u&U>58uKFESMa-W8pf?wER6
zQ?@H3J=SwmAtQ5qGR}z5l-I~yTS;Uk&X9SrayPIZ`Fnf>6b$r+(FSz869RX$yf;om
zzfmSH_ZcTtbY+@Rcseil`xxnL)T*dBBbaKrMji;A0>ZpjPXu`|s1J`gqtdnk!+u|y
zBuRyHRiyxweR(L>1rw)Q$9k6Uw~3=`)X%Sxhog6J4l!*8sw-kIj~GeD!*qMprr%7o
z<pVJS1;MQj#0P^sv)koEsX<B`GmUAKuapn_moVGGB}6@!%42zlx@$6R)j3`&N4)&;
zG$TUQ#Jo5!M%z6Ry{l<&{PbjcMsxy)6J)_h{26(X_B}PCkH&jis*YP_gFKa%k3HEz
zOlv|J44TQ)aYUhRIc&lfdBz(7cj!7Tzydt$!zITsl#n`=)=ZJ-QuE{vDieMBJ0oH@
zA`J6+|Ggxl27+q~4hzQe^vMg}GR}JSFlx*EySx}JV`-@pYCb@Tj{Z`b>;g?_Vr1FN
z-kJwZkh7In`{iTyHV^|8v2Co=oSs*_1!(jL2gBZj;?V28l0z8L0PP10xl(;iNM)4y
z-h&=LoT=IbyxDg_35^)bz+@|WgVxEym<e_et<Zkx2f!Uv!(o8SqP&9%L?ks00(kaf
z9~_;=-0Xeu^e#!AuSK`KMU<%rxLi~Ef<Mo_UY5p+PDSn(lWCUO5rL=>s94KWyCkUD
z94l-iG~<HymGN<F8$$Az<z-cfkq($tEwVaMIMrO|s^h3#<G*mIu#+Gti5M{#Yppe3
zU+9+pB9SQOv7F#tRm9f?k)V1pHjbOs*v(G*^(Iuxa|o!R2VjHG!)nu7B=N>HH&{<-
zgGlkF5Ko<mNmS3nW`7Gepei!55W8TD&60tl1k|_2iXaRymFV3T6I8I9tj^fDen6w!
z;|#thGh23~iCAJbaRzXMCLgoZ+)i(%4>6>y?y^!%YbKcPiW9UvgDlRx?Dm0zqOfEb
zLAg{u?yd7{^0tNP^1QG*h@Xfy0B<AHQC<z}8f2wKrgn947nKSG*<(yJHk)mg0q=)6
zBYBnvD*uc-1NO!&vCGv)=N<$|YhK+Q@=2d%WVaOOzc0;dvo0~O`TUf(W-y1Hf+a?x
zeA>Pl=*4xE7K5CZ&je#m=Pg6>*;s?A731Y|X^A;&Gq*nP&-4aK-I=@O3;v8+C+iua
zeyV&icEEJR-jNx;WbE66S^!`6_wC*Rs-rvPEB*`;100XL?1A}yHJC03%eX0rOvSGy
z3*4$1ZX;D8M;Qi;t5os#<mr>ITZ;%MZBhNV5$O`&u(ip+`z7B@Qw8|>tvEvKG9AcE
zq)qkhm_b#c4K)n1@|}1vmY3}T9R6;a2M}mzPEbzdd;ZL1no0NX`!nWAFuiXRq}gkU
zGnl|_zEetPA~8F`p|3}0@NNRK%gYbazR{Ew>>I|6GspFZ-js=Jk`vMZ)_w#E|Hvji
ziDkil9PdJ~P+zlB7OWX(`}`;II~bk%a?}cmNU@*#R4l2OH`cC{pT%Y=wB&iq(9ccr
z$+b3g!<+Jp__eN)si*SGP}rdM*deI!0uD)2jm+Kmt5}?gaI5^fNK@=!fU?NI{Ea_z
znV_b^Z-Yc`7Mgfp3pnU^iQ4KGa~NN1536|ledua~hropSL;MQ{Pp$khT7`V3JBSxX
z67TMlKlv~4Ib--#RCo#QW&l%vj$gb;wFb3dN!H3=5`pbDXJV=D4TY3nCV%zcX_l~P
z6Fq{^36iVlZ@!pS<$3vge&7DEQznj+e|Ud}@=-n!8^x7;@=v1^_Ixhaj+;PHJyiZ>
z^Mebs4{G+mos#5BuXrN=3G&Z`y8nj1K;0V(3v~14@}@r{Ip}d>z3;c8Av8;{XbVx&
zcEdi+)>5@s_G=F)lTjIpsequxjNkp+6%Rw`A5x8);TS)lU4hZ-uDed&(XQ|y-iO^T
z2evzyma9QE91Yn!$lkr<j?3lXc8?HkZI?sZ75=atQ!R(K8(2$~k#HEff<ss?hqdps
zxXWxyogCgC#s%Cj{+fvxZdKa%JR6T_cLWXjzirJ?hE<OAX~MqHvx)gnc3N3n<WWYN
zf!?-k9Y7eZi+i+9Z=koTY+98Z<HV!JZ)q}X<=A$Gou_463)Q?4<~W-e&Q}fQJKm;W
zS#htN5Tr-E4%J=WJbPlyTg`GT5&2dpagx7nO6?M$w0&}NECe&wD8t_w#JMrkDyP^#
zut*3ZJk=W}k65Wx0$eD(CZ{<Wh@cbwCTm89KGQo+PaVT)kKcI<*%|2@vcD=um_RY!
z`%M2`^?H1RY#ICx1r)-8Ve@DCXm}J+{WK>9*Xr!_3)2m$Sg};$ZYaIy#N37s87k+-
zbj)~R9*Ujua$c;^M9^E#_rYt_BDMjF49W$;l(P}40II<*y3og~dLqV%%a_qN)oKkC
zk&A3maj6mPH)NJw9NnXYAj>dQ6|`l##Fl%Ymv~5CF167#u>L__F0&u8$#b26+%LC(
z)-m}3yZ(xpU|X(7u1s`DrFOYvp~GF}zaY#D%u$LI53&X_by<W(5BAsmRt7D(VnZAf
z{gjQLEJK5Z>Zk}WZjoWepygFnGQ3^I*Yv!Mi1|WI;<ZM_dxS;?6AxI8R63eQ*}jI0
zceIQy;T~=;<od|F{x?I@+Z*KScDEw;JO=~od5G)hA;z!q_ia1y&PZAc%<a0?Mqi8A
zgC<05<F=kOP)*i@CP%x5mj87z>k3?3bbX8fu9fwm>vmOBh+LHUc(G5Qt`M4CAY)=w
zZODm7zzFOMZ0vI9nUYNkZYf5ok1bLLJWM`XAQfpQj2h+~;#?nRZx8f(tQp<X<Kx-X
zVj<-T?e2I&>UOS;;ro^Gq!ymorY!dQ4Y8&u)&=D*FE>U*NUKA`nk+xj7O9mtEl%>G
zLbiBL!3vq2p0Ow~CJ7o&@n`N+QG;e`dd3>eYA(o4M(aI~4D4)_X)&6HOb>$6@=_Hm
z(OO$4(~I5Eh=o|&Av2sx3QvP)^fEbxqGzUcV?`bw=uuB+mbb$dv$Tj+a<ji}nHd@l
zg%zcf_Ldkz)z};5R)4K;fMwP6N)R;7Xz1I#QMj0u<#r$v?{@pNkY~-8JNyL(knDt-
zunVD^ccxw`X{qf&O#rDj%U$*<&O+c#ciZM?o&hCy#(rZmSzfA>c;&|wnN%3zr`P6|
zNsYZc9e@`S!b9S;BPORwvi2MGi7u%tp)RA1m{nFEFAPHw(}ob#dd`xdga;Yk$M}~G
z(&!D)&V_}626o6jiCzpzp*_H;T8gIG@iqhm!5Rh{H$@L>H8IvwtpF@G#5#_G=J*|3
zQLDoNYy!7YX;Cj!i}%Kb1DXQ+AwryTu4=G|UZ1pjJH{@iv>=Wv4iULGmI~?-BhZMy
zw}k?7{q>TMZG`k7m=`z{+KsS{eyL_54wmyT#$}lu@jB`fAaod5S7-db=LM*VUEZBT
z|8T{vk?!CO7*E9fkWLPgV$(^Fy^Z`1IOg-&E+(JPNrZvk4Pv3*pA8#)wj5T)+&IRl
z)gWOKpvSy;G{KG-z+CT1eHsF{0*tIygS<D!+k|`U<-XveJhJQC<bMCmN~i@6$_VHO
zVmnvy{&JXX4;nvZ2%O9N?DQBuYJ@!GP3WPcg25p(dH>sfvEVYTLoN^7f(#loe2_fi
z6!f4w)~=FAgLj9F7%d-g0vXt6>rVBCW)*e(55|hC-cN&Tg?#93-<U4tVn(abd^moG
zXb!ZxRB$}zZ9<%D$v0k=_jphwgA=qL(fLmpWf2;~(t=Nhj;aN+Av$$6?F^$0Xd?3?
zP7~^_;0+z}Qt0mM<fCawmVb;aVsHbzG1wsMVck9DzXNnri<1xrUAyzqGAgXe)1kYJ
z%_8uxS)PgE*5_(rw<1o{=VV&8MQcBo=D9<6C8L|3_h-x`;!)1D&KGQ|Wi8!gq8DSR
zYPQl|QP3BMt9qB^C2xUIDJ-yr(?(wQiQ7gY5KjD`m1sU@Z=>vQGtw~g=%7|!iNT{Z
zzgBtZjn855Dcq=r0)y1%2Y=V+I$LBx@HgN=z-kNAXO<Yuf^=rAFA8#%^S+}RS?puR
zXw_jzC3Q(6In(2fLG^?*#aa5ibt=Y)-p69~<4;ttk)=TaQ*06WwM3TLo?(HL<uN@t
z`a3z_74}<WeWR=l-XD|A%Bq-LYZKJb)u~mx2ZrAJH9m&Lkbz9Bn-3M02-|X5n|`s}
z0$y1FR)HzdpXSAqZcz1It43Td;S5<9>x*`xMhOi9L<$e=fc3IIem9f{=N4@@UZ%rS
zVySHKo_QH4M`LIk>uPVbd5+1VGN#FZal9ILbBJKODQ8QNpt8JNw)&)LeDP?47B1To
zA*`k(z**TI9}OKeLUx#hWcVN5CO}zbr}e?u4j#n2VhO5T?(a_IVyR;8DegR`R(;%m
zDLH2FJHzKc5rViKE<#|_a1Fm2kF-80Vr8?k$7VVvhm{|&za#Krf$X)v!`hf2pNtuq
z9)Yd5Qug^X>fz*hh8cb;o`JD6IsnY}=~#|pBr4xl-_klP=VxMO2wr4LuOU{v-{P||
zSj3&9$a534w$G)P)xkn(s-NHr@WeISVFjzu0>3T9ab?NdUx?o|FNI}Gm|4PVcjk2Z
zVo)6Mf;hMG@+F^(fwFozm3-M49^+pInfI00%?$+_w_o)hc(4%%B||I)_pSn4fE@BQ
zV*!M)WZLEHK`F2#z+9Zn>*O10C24K3$PM!wSnuMSCBIN{VbmBq=#y{xQHR+1Z^v*d
zTJFhp%Xj>nK5TC$Yx~8+P^Z58wt+(ALD+#~V8@wh_B|gPiW`8-^vsv<dutW|Mlh9u
z8T7`#7OT{W2yct&e%%K(dXW4eMnU)h3&PtCKv6h<e(1fBS0S2tOPu`3M+VPuQo~HY
zHf`g_i4Mg}%J4np>`!cE%C)M9{4~Y{gA5stOilP%u=Aez^7Hun^h)`Kcd1EgAbP(M
z14A-m{L&i61uKyK@~bey5E6q3btG-Wwvk`kR2a<<qpZC^T=+LW)Wz_3(Kx3@U{1&1
zhHPgElx&myE?5EDEoB^Ct-o(~FV$2e5M^QRvuMp9e5jnU;T1f-timQO-ap#3X7WOx
zKm8Zo4Rpk*E`Rp!EMm&Aw!heyWiZO*uZbWlH8_Bvh6VCa-mI6u+2W2EHCq0jM!r%{
zAYM=Z4_7oaLO_)~dO}b?9gP(4!~Ulk^<yT;zpU#$bm{)v2Td%D20nwSZ595L21VST
zK6iS@=~ScdZ>hYICbd9yc2pCWHv@3fTmu@NT8`)~f7K$gHE>U)u#Zui1HI#NxKzAN
z_A9u3ix7PTyWxs~{R;{%!kZ}v6awy}4Ei4c2u4RG@f`(&G<iQbiXC%Ca-cusqA_v+
z{t`K;pzz}e{3iz&_8qRnzzVXi1=#2i`+3i^0BH~P&uKwHrO^cD!wLpb8czU+0im@>
zgB)H6d=!~(Sww{#;h*!)JC+GYY=InE*mscjctZ}HcvSprOj-Gja<n&XkeumJS|eSK
zDHt^CysbF8#_{=T9b57{ZN{ZqYG{YC6Rg+c3JU4me!UzYgodXWaoU*h3H~jQ6{h6T
z@n>cSb!Q}@`z+w&0w)m4*(N91;@~zfIoa0<Hx<0oTe8C@GAUT))OhNYlGWIVTNfUL
zPIan(<G8w;NJTj1G+Pds#|?6N%o5_5Y2DN?O#j!sL^;Fh2b2`M!?F)dyE`)$5_9B{
zM-hLP(d1nh$k~N`2V+gjF>OY^oRg-(P;uK<3PPi^wE*VN4c;A7XoqSo=lT0@3kmcP
zUxtvJAMGi)3tqwzX;>pM>wdYwC^)IHRxS*FfDO?s7X?E>$Ra=~FBhk20GhyPc<uHj
zi61d3Oq%s5jog<O6f#FSn{1t27K0|L-B>S|r?nuwt2<k+FgoDw8H9c|%a#5L%zus;
zVCzP?Dn@|$Q3rvWTn450V3J`k?8cl1TT_UfM}?0z8DjrDUkhGtl%Y1`YA8U%V#d>Q
zg?1V4lc6Lky15vwI3mUm$3qs_)ySYtuA0Rw5Nc^%4D^m^>xLI`bntmihDu~1FYhYs
zvl_hrdn<(w6xepOdHdyRpHg@rc*)B(F?IDXljC;DwXyUVuhut&lEr=7&wT_{@!hE%
zmh+-_3W#Ld<t$dEQUN=5CtADS7L1n$T`pxlTn>R=o1H185u!~sm|{h$F}6v{Sr50|
z8tJIbK|>mw`sZFzF(wryH0GhkTx{L3GR{8DLmkV=c%Pc5729!&NipYwfol`I8Ra!L
z!=nL~>l&#HelI~;O9M}D2tq0j$^mSX8)HxqZS+t;M?yagJfG-qlc3b9*u$hCVTp!I
zLv4vnPJNrD%rcmzR`C@7y~KP$n--eM`_-lf!L;5g$`l|j<)+}QG8oRaGA%f7Tn6_=
z19GoQBcyq&9w?da&n$A2y5Me^5$#OZvcog&6~^qqi(S)}scV;6wgNX5azK`D_LJx#
zsIs@%4=A*msh3;5amy{D5LqU-#r~*iD6}Au`1W{*VYLtosBj>6*gIeoz5YP%jMHyY
zV^e3QFdGKuUH%y>uCLPFx=Zd3Y8&HYo7kNft0fcEWU9Aq9sJ26EZj_+r8-Cj(5Shw
zwq0ssd9YZlDruJ>Tig(Dcd4~0=L=1Bk~MPi+Q51oQ*}Nv-d{&01+qr!V`LiUA5k?d
z=hEQ6P|v5Qic!40%xZx&`Y+0N7y?5v2~%vBd!h#ni_irT<JtZUo-yW`XxY#ve@4X-
z{vu{0N^^R)0D^$)y+v9Qb(rV@dj^u(v_~C}<YLJ(P~xEnwie&V@n$NE9miCX7s|aR
zHRoM&v9x)2w)vnY!Z68)B(2ZFm})OJ7KKbev=mh%g=hiDEGEmNNjeJdhCyZ4e6EIB
zve3XgLy`j4;;n~y>564-)u6gI=??C!KrD4nS`>yUK@Ud;2F;80GAAZnhU*Zqljiyu
z83YX9z1F+3zf;FtBJ=!rb3*H3g9U@G$5sKKzQ?*{40gS|x3KSMRjd$V28^gB3z-iz
z_r5e5RFP9gpl-;Bw)>M-NpF$LeYDQMEAxU{#C#sGKH3ZLvB-nQo(Q+j%lk_Bh*j8V
zA*xCy57|16ub3w9_m_N~bT4wh_Q}Kki#kTpuVn;LU3er>#R91jEr5nMV%Dlr$)mAy
z(3My-Y$49355(H<c@oV3!B~5|L0f=-+7>Xz`;fg|SD-N^ACASV1?=@$;$t(P7{~Is
z%{teG-=FZ`xgUtW5Da|MnyUgAW1oJ+C&nY32}M*YZTUYM@(-n3TXXW1kCLxgeF@aZ
zD4zTBbjd7;N-=QX-lcK0Q=ajm5E|eK#t6x0gERA)TCG#}ob}n7>278+!1Fd06tc?6
z3ndE(2RBd(>b<-<@kRe-F(%2A+;VIfll_v>Di6iER$jJwtN6~#$9%YSJ8BsYhE}3c
zlv7j?>=pl=1>2YtMJSE#X+ou2AZdF2cT1R8!&N8EKq{ui&-X!LF^T0v8-v1uGjW|P
zh~a8bimBZu@E4}lVr6}*^UGvWY-bwk?TyV@S?q5uLq7;K(J-DRZzDdHer_8pW3?L2
zpgx;BJRtBw$Wr?kX0jSM!!r99;zSy{5s?|2z6rO77FuORj8<Vl+w`{8R>o+-1cYUg
z?JTRJt%(Tll-0Jg?m@WHnqa1C+>wTheEeI|%>#9%-=A3wG4+R?vMx3~L=IhE3`?^G
z_Sy+FbY;DNnPMn$cy5T1;7*^cZ1h<GdP0Au#9`&Jj+>%a%OHe5d7*4}YO*IH2znYR
z6lOEu5;Vf)4ctRpUH6M(m%MD5C&420!~`DBh35Tcn7!NL5XWY&X3>py+3sY5mvD#u
zwS0V;?6kiQ8#Y{a**0isgvzlytp{yfnnqgc00z>>6A!R>sCzK>7PZb#*r*|L76ZTB
zDX%6Dw)8jy40*)cEcc|%i8u&it^+|T$S^L(YM|76{h5W0n5_I{de7Xyeb^5B(gL8W
znzA-4ejqPhJ{1$fnP!kERTIkGPkTFz(B@6n%b{L=Cin<&EL>@y^^dtmcC|f45B&<C
z^IuG0Te>#gqkZyu=YHw~E948&y(u>(MM%Er@4+wyk9{fL!^uI{75;~wU{%SNLm<zY
zce#8eu?`Cdnq`f_K~|9omdaOsPE=?CQ7XdmwY0@y5FlItB{mZUHDX4$);EIbrp_2M
zv9esgX-tvDJ$~}7V0K&>UL)W3uMLV+`$oPK!&kH|7!qPlzMDqOXnDngLH}YV|M#Np
z3E7Dm`F>iHB@mdZ9?5IoHqU9Ea8T{h4v(VOePhsvM8`Ix^h`T2&oBK*hy}!nvQd8E
z<0vgI#~>1(6+AzF7@|PWUW)*;bGkH~<3~O|Yz%BagJyqR*k_d!QT8T}|F?A`J_HBC
zPohVdL5vRM{B3j^3oAcOG_$0fojB1*pEdF`qy5B+ant4J{uXpH93*}msOtI)8;jQ7
z!{sXUP|GjV+Z(WP-I|05-z2}X?KYM1XnCGjzYfArMbxhR#wVt+qJEP6Hr_zcUc3Cx
z-vF8SK7co%vdizIJ7qV5`dk2p`Trrfwz<1T{uq>?NF>xu<Vc;H{BxQ7$$zJpD#GCz
zBS6g#Q3&#9?|LaXooWUz<@<}#X!4Zl74p~EwYa|)fy;jj8CZimzVeJpmcQF@Q0cbN
zCI2WWxLb>%u7Ch$S-!RM&omX4D(Hi87Q4YY|FWJ2dZ*OOzkQ@08fVv88^WyU*ni@8
z@Em!SP>FLp)@-x95ql9qwpkXJl{dX#iWm*FRgUD<Xm5F2_=P|UNW2p*GWwI=*?y4@
zM*-Ck!ye$W5VFF$GeKs*4ud95_0sq!N(l(B{o_k$QH%tqp{mL%4|RNbfPY!yv_Q`|
zaO52w3d4>qyG0K4msstW#mQC!xMv9LZaFA^!I=iz48oQdG_Wkz!5xl?qW&<BrXD$@
zLy^dtSvj;rk<i*4g0|b_unt9K^Eevh@D9b}U>PI$vRjVmaA>{`cQUl2AQ!EYBYj5L
zUpQbaIE>mG<7>A8**+>d2F}@r(tmQabqCjcC%QYP!@y!%R>T`nR8@|RXOP(lNx}A8
zD#!UVl8H7P^C6Cp_drYvAr=fgp+oU(oP{fJ7H*UiI~-W9MRp|pA$d8;W>r9MgiufJ
zaKszmEW$Ut<efG`CaF!IAg6dQTtF1zp;jrU8j%oTH(XAOQQ(-P@w5ge&FR4aP+XBa
zgRMpfG^$q08McfQD<{jDF*IV-yuBqWXL(ypY9d&GnyhlR^@Up_b8?Q&ZlD*?0X+V4
zt`nb(M%B5rUJW7k^HO_s>su<+`PK$Px2sV?s9P>Da-hH#;+fjz!VU#14Pv>-z8}zQ
zMNqk(<wq{|xtY#JMpeHSB+n)O3j_x?U&)_xX@@xu!c)6L>vvLrHtQO83Apz%qfIjk
zNnail?ZhpB*sE8>ziS#>spwwm6Qe;xhaszuT@~!f((bqdMZL>|VlA`XfcXc<T34{p
z2kYPs@!1o*glHb>?b^<@bh}~xjHV@{35fCxkNzP6OrrM72!BS^kd+gl>dVOJ9;`A%
zMmZT=hH2ESopeA21LBs^v14j+cNYtgzRQ**(~Rt5x!Q@W*g53Q0({4#W1U=Mufc7N
zkc4Za5xTosf9l;fQCvEqN91)z&^?Uvd=Urw^(EVwC2<r-Le@8KUxy2Kft1<gaDNW`
z^cD9aKq@a|VmjCp2t<R-q0f}7N*o8)CUo+JQtkteBfVHxd2A3CB0q94D~bdgJVeGN
zUb7orSV5e?!x|ZH)6qmb=t&cjTxCpyYA-5_=#6Fp(#rT9x~(OjTiO>x9(gzQ4IK_z
z%%CTisb6{-z>P*yOE8%j^t~rnFOy=NH)MKxWU`H>Cflm@%%;T7&KFvF-4OU}YR5hW
zy)Ayhv|2r{F-$wEinZcz^0hI2e!EPwzVzM*6yJ=kfb;|uIg#nc)6-Z51p%iJ+cRt*
zRiOsl%tT^JWXrah6-33}F$ZT{u6T3$r5|T#NqK0w#oM7Thh=A2#rD=De3+?W;t_c|
zdhXJla+?u_x0e$3zdgncZ&wam?T+BkX2dsekKY;et;%%DU1{Jv7l`cv|MX{l*mMLc
zR>L;R1RqQR>6@k6+8~gNd&B*G<3>vEZBpyfbT+Kmtp83mo}7yd>7~v|paD5}wwu3D
zyOMgN>I4j~!Cwj~VWHPsa4<B+9EJ`aEcc}Tw=3d^WGtDTNDK@BgxmyjS^?H3|And%
zO(?>3X0J6{qx0rndH(s*lBk}pr@_`ESkGvVe#r${$L3lAm9<H0$h;d00G5bv-dmhE
zup@d-7fM@<3o0wK$gP?afw#QB%@dkV`mKnGYd5N6(Xr2)r4X9|`PKE(VZ-H(Pbiid
ztKL??eGawKWv!vi2EaQ6M>EH=J9<!cNR0{UNzdq<F*?lgXJ9o=q($vvnHw7$CkSKt
z7!yFV0#F@cOqrM3*`;w6yP-<GCsA5+YE+5CN&qo=Z){xn090FPko&v=hI~+0L(O}+
z-#dj2hUTZ58%Y6O(FXc}Q2^oU@ZUU`MzCD<v|VsVy)US(l^6)A(Yi&#t~_L1)`>u@
z_otP%#17uOR|;=FoEj+&S70>3BSG{`bB0y)AGPJ03XP4h#0L@`srUko2itrwjZ*PO
zCVa|=f|(~a_Q;3rA8@q5srQ(7U3!pNa>wJw3LwNP5b6m(AxzRK?nOZX{w_W6o{ZsE
z;$j>mjeNxRDBL>j@=;S$XHAx;;w{|th#Lar=@?xn()ON7!)Ijz!bL0P*?0!RGD41C
z|L2^hREk-Rdm#h~V&`~<;=Eoc&-=v4*pY#*G?y1b(BLLCxJzFQmY;-Dukw<0#Z*e%
z@(I?#E_*pGUx`6NR7=gUb@DN%6hfw87`);gQ{~{EWZ6AEOQkp3#xb7D*W{oR$b4_x
zssg~Mg`a}L$U-+)$^xG{)4zF#2Q*Mw=sl>8ja4U{O~ZbnEb_Ne1646%s3NjBs5=f9
zTrUaI6T5}2+vgvUak)7e{j$`bad|ZXVDE#nEZ77?Gt8d>`LvLhCpv3-vse0I4Eu^`
zTisVw>Nm?uV+bT=x5z605)M6Xbyj8rK>_c|>KJ7et_H;dtZ^dIM-lhtxD^MLW~r?0
zuriNafAS1-cx`_%k$^S27`TPpX9!}8XzsVoIc)SOS?9yFm$+(MWqqQseaLJ;g-&Ec
zP;>00GTE3u)eJ@&^|HyI5o=)4^c@fdo4q-LQSKR5fxAGq7(E6LLPX`(pchysG|_D#
z#-ea?UbfrN=iPR@?0DNSX+E;RK1<{8jA2$o!`YR{SE73<POAi!-7yX5ibRRw;gyeD
z1Bk<3(k7pX^=Qf!YGF9M8vkx@z%8Kdvd1?7jh(eJ5c7cL&jPwzu|j)&q}+{+cVYq?
z9ED)!PsVPn&g20??+fzbEF)Jhz*s(Iugt<tQq(R!orb$(fC?(pfj*PkkGz|RefzAx
z&7QrSuYAs^j@;&M`FzlDZz^)Hf^DgM!GE!^ddr;Gvq8St;a)wBenYgJ8Yo{fs^GAM
z#e!P6U-q5}X|jSH?`rr;8aoaT@a1~M7k@S0gTcvI2Oia5OA5J~gIrULiUG5wZvU@`
z@PNX@i$D4S9q9<Ym2afEv9cKj1#8dJocyNg79c0=RU%p6@(%5DR*7Os29#XEXT?PG
zZ-+_(L=!>X_3|C>fg%g3Sz7+kX7F8OAVmJ5xZU@hL8x?YmhbzpOJT&)2+7xqOXdu{
zP+m_{EuQq24M`1Wnf$;90tt%4n_wofW@hR9(3XW50{*{Sev}vvWl!7X$962um?1xj
zCY;*D(KxT2^3ymXaF<%E{LF`^_6F92H+lX%c0`$49rBClWIV3B%*!wR6TVEYX05DW
z`7@rV41K3D`s>i}Sd2)+)_)TM04fM#*BNI0TPFoMn#*Ic=bNy$x(f0;<L-Q+HYdN2
zvFclZmY}ly!Jk3J+WP>;4>eK#7;`5?gJ-PSgZ3fc|75M<g2^WNa}2Kv^@o9jcgtV=
zJ44Y~nTcmH6m<Ektv}2!pdEip^QX;75w%$Uo`%7aDU2joCI3iMD`vCWF{GPN{+YO-
zG`ZcbG-GPwzihCR5p62}4iVUi9I^j+XQsXoYgi|5B&oFl66#HVgNwr=P75)dx6<r3
zD{j0J>}VMk(ixGF%)(lN8849iIt@$olL-(GLfxQ?Ab1RtxnK70RNQa$upx4Q{dFYb
zPu~$O;gT+3*Eq&^$$_1U+)T{0p%}|Soetg2$}(W>V1J*h&+Cgf;{8MX84I}b870V}
zoo)xSehm;O*EbGJpPTP%6Q+5%KVvWzbH7=s^oULeK9O`_dIwP~XgG3Yr{Z!zl2FPJ
z?0A&F$*h^Ea$wwibZU2<@5ya)Oth;=gGPCRALxiU)_+0&41UL1M5va+>bUrZtctfS
znwPO8$9F2k2|qe+{z<(d@C&mbf_`GB0;0tFz~m?S8{D`I&0PRx=VWgim7bwA*ukV0
zqr9`z@kW$v1mTI0o?^}Mp44oOoEk6S#@Y^pP)|!eEQ93ac3UN<$1{j`P4a_kbVjFR
zxV+_@mAudFG!4d>ow^duh?TU!aeDZvfJq^5UzN^Xa+a~sFkCHkcC1J{3f6%0&#`|4
zh{W-YOLET*%3zNX3}CT@^ZXf9J{mi~HFwMTP79_K(9Oh30JZ`b*i?x?K@m~3_`=|J
zDxP?G6~_wg%_1My%SF)vH98Z>a<S2Gpm*@-!Gq)ye<L!=w?aR>G=>Vel>tLY2bIh0
zD<mn^V{(_<cDPt81Fn=S`~$8j;i9b&a98@!$OD9h%%x?4=}f-4DuxD|oyjw6a3&4%
zZ@8`K9--@x!OblAOn>v>7<w6OaT#L7?}7$BG`44}#!L=NeXGY7xIT;d6swL5?<`Rr
zP)~q*II_aiEY?5>Sm@};MBwe_S^$CLTw~76C=(JWpfF>MjE>!oI3qAY4lBju`Cax-
zrWWUkT-})h!t>c?)||g4b;~m7OP~Su%eC=L{fo4ScgeeLn+^0%oF>=#@2quC$+rfQ
z@A|Zf7AS)eNH!@{=5KRLQCH)p9b<#4L%>pA$~z5DR@@jAr*9jZK=I@qU@Wf34jCK!
zC4&oAbM@U)5oBadh8Ey_4UMh1%*VzFNd86{pO|@p?xK}2h`77f%7h?XYrb15eSlP}
zcugmK(Q-o&suKL()`&&EF;>4N)6p#xW54APtq90+QjiSCfR;AGNi?~0pM}QanIi(a
z%ji8=hqt_l5BW@{1Pe`v<FiGk`p7AWc?whO<KNfFP5z5{an-Ztbe`tSLS@L1P}&(Z
zBMF&pmnt6)lUE2+F)!I_xzN?Xw=7wO*XvG?)voE08EGs$a9Q-#61TR?OrNx6M6ll_
zGAl;0_kIPwm|#;9a3MFxG9tPdS+;q(rPKXTs654m)f+%>4f0ku;;Qhx+?JMz*1d({
z+-}bXP?i+=C(yvLNOwf<ZrZ5Ysx}`b=>u}7x1*eicAzoIU9qr}7@mE%{R?B>i1#!8
zKIzP3k!O5$8q00~S9bSGO^ml0kO~lt+RlB05H^P|NHa1pRZ-p>AC?3mdE3e%Zg^s6
zO{m;7qeALZ*NE3uzS%DIX{e3u?b2XO+=e@4+NCjGhrGl{U)=zKb&t<wA;LQEli9&y
z;}Oo%B26J%z*fd6%*aC68-~a&(i}UWBL|d-v7#-pX>$2y$;D7m0wyP|vD4J6!Wq$(
z3A;t^we_A--NA5isEcihzzd;(<)h7M0HWG`4j#Qn&{PWkI^1WdgVtgHf`;9e?Uc^+
zKJR;@mAP5ET$oUR*sVV%O!#*i1m@H2Haf%dk)G7S8bC&Kd?K6z>y@B56_*d^^<3{v
z{chMejM1AH&#)=zO=Km>_q?qIbL^O&@!l9NbQm*QH_LsYKv+EoLhJW?1ID97CF%QS
zq|F1tWz$iO8QlJ$H%RcDE>qrj_dc81XjJxpsAMG6YwQGKfvERKLs}xgO&)e$;F4(p
zLp?`}0eX+vBBsmX<<T_8r5GLdzj8SvXFd@7yAIdPa5^9KR~3{{rtlO5`A{5KOvC`z
zhT_iDus&>jit-!CTI!a^Y(lk-t@3y<OCt`hCxV?D0MfO|leUZs*wxBMoHBNdEQTh(
zOg`$*tR{s87(Zo=mo>G^(`j|q)7yp2^2PE@=RTI4I`?*YHvMWDpVU^LOV22zmqOFS
zH9Jm89y6@c(+6Q$3h2KO<N^G}Bw2YeMg@jqR3yXgU+Ua<ng-oo@qhYHUQT_pA0Dmi
z<zq&XNriU3P3DzoSr?U3f2;JyGsPqL*-Dx3&!}=BG6rQxx66XGXpGz?t#}zLO9wBE
zjx}Ii)hk&Pq6F=E2@z}^>H+E!k;U=5M(_|g=5AFzfhwCUvD2n83p26L+lNi7XEurl
z$<lZR`3vSDBq7U;kc};@1Vy%4?)#V_M~h`eOiJMhRa9hUYKD<FdWE2@3NFghMv~QO
z%kb_(uq-6DslRLdV`QB{A?(A6xk1*(yk12}DmR#F!Ip<jKV*OEa1j&$b4si;jvGE=
zq^!5UR=}d$V4Pjhx>FmySzC~jJEpMC(28V}H9OFo100ZRDd5Hd*=!`7MGwrDm{VP$
zr>9%C#wc)xG3JG_INQ7n@Eb;M!j69uePxQlcBg};t72warEu>9Sc@Ir8sCP$5iv+x
zWoLBQjJs5uWLGF=6YWwMR2anRvD?=Nq72gX<H2BhV3>#6O!80o&|F@qy$5E&irMe4
zrU7ilS!FkU?TG;(%^ihW212Ojk^R)r3CEP|H3D<XW#yAbc-VBZ&s!sK%9}@d2!G0?
zBlHn17cZYq`YGM|G|YJAU~<y=SD&%@RW|Ec`PnqTT`>On4dYAUhLF$2azM|e8b*K3
z=hK`mS`^z-#d*Ydej!?8aFm-^L`xRQ7t=5Fx?9|@`S`!&qt?74v__XPB5MGDQ#O&g
zVMEK8<9iDwgds;og^K1^(m0X)jdAiu+^_nxl4Mc&TD*sqWTBf?V2H09H`e43A@mI+
z_PLkbA>WKwH8zm73>M0_eCozys(N6divb7U_HSUFFnoc3%6CGDW|8Dd&E&iOlC!Wz
zDts?JGYQF(q2G@Wk($#)q48QwA14IU{CGrKy20yl)(jmoM1J6HMSSWacujxkpQ>#L
znxZhVbHa~e$lxZzKloFAY*d&wWupAVh%<QbprP{9=;f}v?vkI`cLTi@ET_Orvuot%
z##N|@-Hze@A{e_om%$}QoWU=hek?ht^@tTERs$A`xki4K=xawi^U|<LYGUpLv;R7#
zrn$|;sTay`($rY~2{In6!|k|Pej5ZVXDlQ9(Z7o^XS-_PAl1tWe;<=N_mZpS58l2>
z8iRTM=+791$y<H6E&dehCb09`TniM%KZjwU%01@F!ViD3wyWD%AN8-^4w)NU34gAY
zzoq^N0#k2VCV!9j)G8qRua<upVIeOODkT53Ex+e6(B(0}ME{BrLZw4|DsDwtApdr%
zSQQ^yy-{xLKS@BOs6mW|(Wq9z+P2yop>#E6GqZtWyctxcryuXHXPzVs!!7cbi8G{m
zY6`w|IjC67r~{|z&9VS#7s`JAyP8EHXS#4#|Gr7~?^4il(u5v4pi9B9f!^sz|B`of
zdAOds>g}96Fg>HEj3?lFIjGAFbp02)W)0&1DF6@dGPuz6R?nLUKP{F+x*SHst&?DC
z-ZOrvccTo4aqfT(d|2vQD{o?%*2&@V9uTXI)UctP9O2KX(op6B+mR!^a|e<X-Qz;9
zm7}^8ADlR)Y^)q@l&GqlB*&!AXf11yfgGDUTcf;#LW**nKVxHzfo5mNTayF5SKcls
z*zYJchlF5wiaEIxeL*Zhj<E)eaW2M4pVXyr?gVf;dHUooM*{N&M%p6d+kVFlJ(*TW
zw^L%MP=AQSE(F|rYVZV4EL$cl3~Re#d7$i`)}@;3y!XiIX;O5!@eExrXLKpJjF?Ec
z-p@?UL+qoeHGo9Va&pq9K=T?@LCs*kobA7upOwGIA$5*X1=oZ%%ekpd#vW0EY>@N1
zBH$E3R1GlNiB6sGFH)i6nZ%^z3#_3+wmv5p22*k6!6#<ap~9eC6kp;@<mJDFl`c-t
zRzu7oLI_bHmv{^It!1U!Ie%%_KFe)m(cwZH&TY+Ry3F=SV>^)3MRIwpby<a6k!HnN
zt8Fh=1}TS(7$R52!d0MtSBng?Z!@!Tsl;GkFpFHbxOxpgHR{R`AEuQW=3q%T7#1?r
zcOT>=<*=@WEwzPRFt-nPh<p18sH1WVp0hd~FhX$7u8`qfj{MDRLYitbt_d4qn_<|n
zQ8Lp0THBbBQ8s3PhYd11rZW-J7m<rGeP|wbOzxJeyY?NcrV64P;S6X@2+lR}?c_`k
zqO-1z;UTV;Dm-r%Lo6`=PC!ZC9fGN1W(5F7pE6mFkO{Ba<uPHfR%GH|Z=?ql0FEmQ
zM$rvndrnhVWlRiPl_%14m6XSpn~v)3Jjcc+&cQ(i^63@fUGC**MEoFh$Z^(Rrlkcn
z@5d*$Dr(Rc`Zd93UIx`OhuvN2EeG!sOSr*5W`F_%1d8|F-st4Vox?KGzXVQ*w=9`-
zl0Rd<F2okT-zAe{oUJgJ0SipA0hCXiQB^r*vP?~#>3yrv6I7i;9OX^^iy}2>jFs56
zrl3rVZm_NBZ{7*-SXG+HR?REz$L5-z+U7)ji4?1t5i(zoY9zeW%1m#7bz$jSfLbB5
zya8e-#6;-#0i^&W+#C(S&Dfr8M&za3VnXfOtK?Q=*z&STa$A}N<(ri}TY^opKyLT`
z=zyc-Llrl<Beo&9j<~<&wXqHjPm^tOr?*DanAyUtWvU@za#v6pUQote-5vifZ^$*G
zUV0{Gr|GDG51~j^$Jlrbg6m<WgSTj7)O5KIAdhfyG>1q&s64f4mk`j23g?@#1m;J|
zdV3m0j~SVVg1}~7h@l(c2CI*~UXK#w8jabIWQZxlYU#n`VGLZ-S){RRpJm{fTMu#9
zR59~%Pg)8fEsb?dGM;U-CHTrJiwej$IRzk=l-}#1dNtd)P_+gzJS{QEAtMG$&fliu
zkC<~9&#R=>pV9loRG$?%?eC4PN+#gVFm2A>4FHUALga17j2eOhob4fWP+}Us>Voyx
z3I9NkbOi0x7X*7IMPzpRFH5jW1PiwU{j$K6uHY7haa9~ElM_bV4&}Pr#+J#?&T6%B
z=}9CYOvH`dFLP4!ERPGThlvHO<0f-M=nWn{WVp-=g21$7H5Fj~@}4w+V&Mf7B$}Zi
z?~Ms};t~p6qSLkSEJ)-36Kdqo<^DuwR0q(FxK<wUp6Dix+#+V&AgPv~0(mfzj~)f?
zbXo=osFnA{xz19b;NwOHDL&*~^ub?@vjQ-|E_r{N%x3oz14uzME(?(opnlky^na+9
z*fYo^0go^S^pVsu&UXv}(V_BaFleS3$(l&6!-YJ4x}hA=2Yu+Km>MQWY-Oc<$X+Tm
zwaSORSrb8&i0kArA25-7R$wK|Q2nBw&Kf?C8-F#z<Ju-q1ao&I-0DgHvQz`Ka8^Ua
zJ`yVcF9~6H#EHpA?b$$Y3ltw-rVTPa6+)&Kk%CZ{p7wS*KAM{>lxtc!;egHZOmI{u
zA^vC6oX9<#*;aYZiL8inXm;3w%wJ*xt)W!Uzl~(}F2(XHc_CyA@)wYP0B!e0ACu)V
z8sJTIkN%QX5CJb4kB+sn((-b!3#4AQjkPGTr$1(!7;0EvUWpf`q1qq3?z)`4LEU_&
zAoF8gz?uJDFrDeh3(}4wj$qaCtR(irL=hIaRPis1{B0(GYpxlN5j%nw$0~u&glb?x
z%aZ6D$~a@^c^3k;V5l?H%%m?#+T^;#TU63j8sHFs>ra*j!`i*+l>N)RVHVkAUIb%H
zDT}P^-Y!{gBpT?w<aSx%1M>q{@tT!Ep&_G3$g229MfoII9sd}U&B&VIs4-cjfz`^|
znB(3DiMK)xUnl+c4Eg-9RK0-LI)BeLr(ONH-k<TfW%v#vPh>;1k01fWLCHp^he1<>
zkq8oz(%eY|+IcJDCgYKOO<N=5$~OlsAV6u@lNKd7mTU>~Wl+-$5(gZ#)jweXJRPir
z=NRhGLmsJIvdy-79t^fUy-FdC%E4MufpKnjHt&e(1O7vC={&W#oz5hnIGVH=5h6dh
z&fHt$mSpjjerTz5&XOR^Y1r+*LzpUz!pQbgWANj}mI1%oTzNlJzkI^U%i8%&LS=c4
zSB+4(kEK0}lYWm=*LaJEDDN9J&g6xJ%KI`9<f?;v#G(!MCK@oe(44_qKoUHMsC)UO
zkIGn=0tyfXr1&b?=fBXk%nq3OphuE?D(tB!9|+|Er~apH$d%9l8sX&oOzN4gQEwj%
zgi$E-eb#@sV~(R?YAIk5P;C>G*3a2K*z+Jn=I4DZCZLGn(&xh4Ex+)#Uo?aqeW8Z9
ze9?bF%X$z~HzW$tCtpf4E7Aly0TA~u`<M_Er&g%CbFtnOsyqU}VpBxbJ?I*>@>S<%
zMuma^U$YTUt1O=`UyqjHEN2xo@Ce#SF)H#6Uvuc9;CcO1z8U<e&gE7u^71WzkG26T
zM6E&q;<t^(ac!u4Cry$DGdiJ}p^E2s{X<)9wLs;2Hr|2Wu_|l7AM#?xxN>>T{!xo)
zzczWjYoA6m*s7W8$lxR`ct!5D?ak0$`GI#tjhs1X1mb=eE!T9n$dA%UEa?y#r}E=8
zZWDYg?qlQrNyrt1d9X_9PgC=}Sb|x-yX0s7H82d=z_t%G-&cN~m=1a2#^S1SGRB$#
zTEB=d)gUvmstA_+GH6rPo^5TAUxnVj=W)#UF>ui1;GAE_;ILwwfXe^I2B+80@=h0`
zJij&LrM^>sXMgMHlHc2LjiIi`vGoW47eWMJBmObS1K)vNK8q`O{24V>$_kWyl|TE?
zDN`sb-J-#y{u0|B7v3YCjBb#>dIJnSLBtk(>Z|2%iRNw_0C}L8TP}Z(<(ghTP5xp3
zU=~b^{L`qyy2?~b{uSiYgV6bcVWjeJn^RQ-f=6rRKXDj9k}%Jd0Vjkis46q=+?M){
zXibmz(%h0aV?Z77^1Kyu(K@wM=#b0Z9Wj}^p{r2~*&+LN8!kz6iN&T^*@D7)ne5+f
zz^bS0VC@O7F<6Hu+Cn*?dmnGdxd&nkU}?PU5{VIIc0N-^ES3YkYx;VKj__DJD4qf0
z<$S@^a&Whzv4e(>ltbcY7^uW)7=?SNKO-cnm2+Tb9M-L1-mJVF?sL)mZJdCns~q9K
z-~b}RuhI3^;*mb^r4XMmW^q6<)1r7$?NQz%^G=x&yBu2L(f-VCkKT@;=`C_h5CEpF
zWn~gaC<cZfYoi=HS&mC%F&Ec%%!FcjmmKfE7~ID-!`@ZA-~|6gDbHs-K#A+iiHS<|
z8*_a&%SkbolHfXPjYscNIoW^rLkfG`!pP|Vge;VIb}J~1omnfVB#Lj*(-Wo3<kX-E
zL^Bh37C^T?&7Uy`mHrUsvYZ~zpn}n8+zsc~89o!n8E|%HHPG31roYW!Lav?F?YLY!
zQwC5YP|i+ms?mc@vR=-yXQ<vzrJ0qNvvRJ#hq@e45_wQ=m-CEcku_X1R?fG7Oash}
z#PDthv!G*TJM(g3cZ6NB#inclc@=VzH_6HIY}*0=^x_~Tj06TJLyqf2BP#L||DA>n
z$jdVvUv7eAq%4{*mm0s9wU^0d{%T1zY48OiuYKcRZmTjKIi+$%%w#;2yDNhvv=enX
z7RpueOjU5bR7?hW2Nd9O8mVCL4h9FwRADve^$>qXVL|O-0c7t`Z-=u5%XxwIekl27
zZw|A8?s*m{({TSDGshm+WFr~T?TCH{ydAV@Ms_RW50`zRy)!GLynQXOs&w)wBOmAJ
zv{-QQBi;L5&Qf+T>4xz)uQuv-WE<p~=mR$~;efo>`><`%2N44u(n9}m*U|m*?jUe|
zwpy+W!XW0Bb|tEI$o1YS6xY2EKrUGDEN&l^GXKTY04KHv^pCMIWIGA*cT0KD79Lq{
zy>-w5#s<STWU6rqBSu<bvjxW5n!$ClGR_3ajkm~no83fQSvbM|id%Z5(x^XZ5U#Ge
zA!aca8pVx4R>Vj{@7D#<s1+yH#8@gUD|5t<sx6Zesi9Lt8cepv0fjfqloGPjA-^7+
zN@=@ErrIQCLZz*>sHL0yUG)lJs>Lca)1p`P2ogA2E>(%P4DM4o2iBu~Io;<#jo#+q
ziZObo+haz23AsQignCPu3DaF?`j;#hswW4ePl>ldW_6nv&-{064H^Jd2#mHUp>Gbk
zfM_h>5C|@WNV=t4-G->8TZ>qeThml677A?O1isDQ8|bZR%C*bwu?y}+;Z3RucLXIM
zvDi1W)b5OzP#hO&La3D)JOg#ru<_k;xA(XdWl0#5U5rZ4M3VzJQa~cqgp_J0gsHd;
z%b_jC#3u%#;=-Tq6xCmot8W1SA+@OkR$!o|Pt%6}<ZNmz-Le6yT8)?`Fb(Sb%h(GV
zx<Mf+_5KSs9xQ7=*Yred@Lxg)hX#tFu`olU_r`VWQ(hy=#yzPvv)p65V9>~sGCMeb
zTCPT#V((8xHOY)L$Dm1KWgLR|3sojFD4-<`V3E?3WfD#I${G7t=Mak3>TgkTn6DIM
z^b2#Y?~P3}b=s7fm1CtXvDSY99cZ;!04N`W9N&gwLeg#%sATM)6nq`wd9jcHC>$Lz
z!=ZzRNoR3N;CS0*;9-b)&=)(ZD^WXD@dC}LQFZ&5ioD^=<?!2Es)xpzN>8-xiXFY6
z>L$fx;VY3j-R>qt4fQOF`{BN8GV@&jjDui?zgozI1@U(VYtFM7p{x{&C!q*Bzj;qk
zU`)0h$k}^?j4d!N5YKv_ukMK9a=*<T`5TB%c)%6~b@|80gK6oQrB#}3^}ga(Fbcs8
ze#kZ?aC3xF%lrKe3n?Siuu~q6VIm+AF(-Os<|F<duOz1=p$Yt`wS<3dfvovJk=m-k
zQHq%h_Je7BEZs?Mji);+O?@aP-;hPbzI-?agOdqrImF@)c`Q0syq$F1Adkm+0l!NI
zNj~-RM2L(gs7%y^o{SZSZpR=lvYZ~&SZC&me<bM4`;EbQAB_g83vH+kn3tzY#=_g>
zcwO1kX?;tBw~IFcJ>%aHcr}JxBhT6bRuyXHxmdID<K+1mpN>oc3fI8t09-FGIL(Z!
z@u%vSls+#8%_d}9+8X7hkX%#Y*g^3eh~}4z<f)kr_Xx05`=&h8$}%|eiY=V`ZF;>c
z6IX~sRt$teKl5Wgup@EsXfXc*e?}mUrV;DqF7(N8X?ZBJfGF*LQ_Yvkq8L!7u?-=-
zVBf{w3Z*l@0Dvw_iqEt+UX$JzBXDG85#ZFN-kbSeN*vxUua+6b=iPp_ERVrr0@T^*
z6<^_8iz_qH>B^vHSv3XFD*pifL10WYxR%K3G%f-Vn!qh<Vq920VtX49K)KfGsFB^6
z6RmEOp#7;;XkHiwF{^b!GYul42}<pn-liq+aBr6N_6_RaP~U@(4Z>kV+DN8;@la<R
z>nho3`q89uvdNf?4g=X7(z3cKCtHH!5lY1!vNe{yBtefB!y4J<zfiBCddnkStw-6O
z2CznxkEm1et+lcveWRJ&m<Y{ysL8U^e_<9ls}$S8g3vhGWeb%@IsV-?I8=T@<^n<v
zJ|4?bpeZ7sNK3#N2(xZqjrpKf5eg^B9`A?of`kXHA8J7Z{<TR)vgan*>s)}~Mrc@{
z^fxHo7htpVcC&q{Pc2XaN7EMhRC;d}#)0fM9DJYlmgeJ3-XxzflFyrW=bVe>v-Z<K
z?~R1jKj$CQ9Hj+;LOSw!f5uDMsdUk;_Jx>xdpFbezG(A=_KbLVFu|8X8=G3pagr~4
zI}mX&<xRwWB@qOB1a0%GL*vJ<MiWhqtvS>QQ{4D#HrT0563ffi6MN!bDNx)3EvPyu
z94+6lwHseCS-u$y0TB+~l5ZIY;<8L!A@0P1QVCQ4J4GgeW=7<cC?Ly4eAnK?S~H2^
zduim!5v@a5E8q8Fl{(SPC$3e$<+ViPHL$$YuP3jksW7(0Y`;=x)DL2n${VsB@<VR{
z!FqTbHWHZoN3P4zNT;+SYtfwm`G@#BkL8sg`>0Kh5=JG=%4@lP67nxsSC>ZwNLGHD
z#>5EO#qI#+cG55)`I(Igq3j3(ke?^IL4cu6hMnm=<QLuh?q;mb2L^ioZ~jky8FO=C
z!4YsPuasXoZ5-5aXd{9zLjwNVj#re5%JDA8TKSFfPbChg-`ZcN=Vr_AjCE#BnRcW6
z-q;EShY)Df)-8YVA&`;jp3n*ikXM`KkM`A69xQ*d{WrC-i;3yha$`+F{!-i^P{%0e
z*2-TK<2Ge$ky8D)m?WeCv$z()#)az*k)eP0<??bkTAR}FHaa`j%0Hr|ssip0X_tS-
zp^R%^XwYK9|MDg*ADU>2xfJR7VHC>0Z4xspD`v@mf(NNJlHcH?c*CFBiAjZFx4h|X
zTG>xi75FyMf2z4}`8Rz4b(y};E99idafoTzjz*$B`^AH{hHQcL8~5)~AZ$|Ev|HqW
z9z`Z+A~x(DJ%*Wb01ODHN??-sz@CU1Mc}%!=Mp){pRsH+4mKQ8o8;j1-eSPl>L!vy
ztTW_!G|HjY*|dsDa+tr0eX1#mG|Lb70od(M%sLOi?1&zZh6OrB7c)dWhEX)PkMu8b
zG|=qQdTbo}COIm#NgtfmI<!e0-D9cO1PkaEW}GF?d3?{)Ue^w%FExF5WAW!2ImU)R
zV$>)(wnwp@dY8Ayc`qRu=)pAO?f5h@lXNNDTuw+M<9cdriL>pA{tL}@UWKB-1{P(x
zoa9u~`;#zE1}vN$&lDVB)utWt&K?IC8(TWykAnVuiod1oPL4y|)2Tgb;Sb=R7u5MW
zEzN;9h7kIp+#?u!dXK|~v}c(4%2Vo$9yh2p-vXvayPO##G1L$i`%*b8W|`oIX!YzK
zgC-a}KsmZY9r3g!-A8~gmdQDZ7Bunf`X{7tLV-TlmSCV4QXSc|=XrBnNEPAgJX`<*
znxEA0t@DF0%$P@*RF_=ftVIsi3P<>&GzIWNAHXt5rEuPHhwPAxjI&|v)yl;^j<lk3
zV|AucE^$)OPNg${1*URoECCJ<nBIu??32s<*)rhEQMu;wv=C;~u*QqU&|l$i(?}w)
z0r!bq>CcR4mM3+UEiC##T89j>u_06wA!k^*!N$hJMh}r8v54>!H9|DtXdW75&Ewv}
zIvExd^ECY7@o%UqP~~LA+t!ERMnCrq5;jKqAXu$sr;G|-n_M=jLPp1+5eh@1BGpXZ
z<t-^BhBi{Tnp~Zxy#shA0xse7xyFb;rV)3MUmLRkn%Guor7(WCuWE!9(XYZLY2~w+
z?sYxtUFjMp*T=vbvQ3C6HQF@e@{KVuMkEX~EO!&eSRSGj>Rz@B{frHAAg3Hf5v9UD
z8bFbZ+Q@Aj7w@>{H9j~F!i<MKL;ZMOSysdjncy8`yatOvFRgU$HUU%m4dgULmT~tD
z-ZD)yLiH?5N$$5elx1>bj|U*CM*-&;3X3A-)aE=D<x(aYRrjpGc{RB>44V5|t7D3<
zksgX%XWDjP#$_@!$k^F{<MAeIr#=g9ahiX_Kn3E#+n~3ZmRglooN@lt*ch?N2+`?1
z>V!j$14RDJh=owG&-rYWnel8di|Av2;S8S@9hD%#Htd`IZ9O0{Y}zxo_%nu?v0I2|
zxz&a}Y}BAZa$8!ZRZz(oylnm~fMO7|8|3zoB%uMw9Wm%x6Ds6R|C|eJwlYsF!;2WJ
za92<gGMv9q<uR(<ZIncia$9pwGHFR!!>bR>tG)^Rom+i}RNIgl1uQkeXf2FlZ<AV|
zVljMYp=9HosYo+`hq^9FVWJ$mIq>I7Kx|xzh>ocbzHjJ8S)Z<MY4G6_x}#57>k5$5
zq|rOI$x?}3FZYB@nTX5Udt|naxw56MNt%L;s2eyN(fGB}94~hw(z8WcV!(icnE#WD
zb-}hoSy9-obUj&}o_-h_t^Pasgxn8*07V9&3OLKLZuk25l+i#?CQh^&$uR;{Uy^*h
zfp6)LTn_uOJ$l8SWhKI`I3-L?A@41hLW~mV5(^x3B(f?F3@C-fwYmUB{Ib)jXh)#A
zYt$N-t`KPmt6?1<2<Gm1#2e+{gFrrhk4dq6>ts&!U&w17@LZn*0bqlH7t6fFO8sCt
z%hfTN$Uj%fdrS_@$^fJcB5DvD^Z}5XS`lhkVVCyFecrrX5Xtn?6>@*9B#J~?2=N2{
z7Sa;{ztXDNC=bRnKf(xrzb{t229Zjw@{qAdTcKvQygz<oAaGtDwk-x1E(%LN;!XKo
z$e{5Sc{JXig{##O5E@(Z=9_Pp58C?!y+}fVj{}u7^YWpfi8?>QCs2Jz4T3npjOD{d
zli{O=%VXA7BXWM)<?-0QfQKku^`aVly-1#jE{6;sB~N<ucDG3w6k2oz8)vb6B!*kp
z*oK=0KWg9R3Qf2J{i$eZ!q|!Ov^QiJH4j*73shWTC_`93<Fv4A&n+tN+0?lOFlmUh
z7Kb%K^IY1)mO=y_MHhLgJReKRV7R72ZB}0J_L(h7u$#umi+*lYOqZ8z9*9pvq)3~*
z9Mj6MaG-q5KX#FzV&aux*nAeks$Y83JFIGH95p{EFa-6o7WfO?w}f7?qh(<zrifZ^
z$zx%VS|y9pw5UsRF<IAV1-AL(L=e;Osa>l}A8!#BTN1;@<<<aE`_dY3M32TBHuI%v
zECBk@1mFT$mR6`JJUu)I0LX4x9%Id+h*Vn#cWGW$*bEu>o0pZTRhkOwZ<19(79NWj
z10Wt*9nTDaCC|#5ASWuPlRww`XgGa>LYt|~>y{%2ZIgbV8z36kz@a6v`tUmch1oRp
zhpopEPx-V%)~6Y6!ok49pMC-s>ZS7DFB@#%!kv(jjYi7aY#lwUvdQLIl`8-y+Z^M!
z%P<&HC|ms5GUz<>F&;qFvel&9W!K6!|5arp#svR@Y)`9W@h0Y}V{L;SHjaVb@h!45
zSZMr=O4${i@KD0g5rHSWLvKO2%L4hhzd;p&iJp*r@`=PXEP+ZzfdX|CxXKWkSA+D0
z79=G#$(}S=l1YWS4ESKF+iK6MvVlKkuWh2r$z}4%7z$DfYh|B*!o5!Ya2G&((_B8~
zznCs)CM!$0;Z`2+I&2_b!1S4*#?2Guv%%mu<Qm`s`CK%F<zw~PT^O7BK|b%l*laCD
zTJbRXLa30S*AD!D7yiFVz8Lxe3V^`3+8|%DUZG+($d|E>`|*O}|Bn=f|L1>}uO#+T
z<3_Iy|7slTouHMLt(UI_)8=a$<?D$LptXYih}CoNZIN#T(VJ2HgGSFc<GTq+B>Ps{
ztwni6_GA5q2XXX#+sEPwN02W|<vXb_mca4d1}U>xz8kD!3IWbWmab&!#_xGM77Bzd
zgB|#Ne;eKbD5?~$JpLIP_*&0C4JI{k$3lk4)mm>DHcP2Eem!kG0zvjJ`J5M`01}ZO
zq@HOfn~eIQ@f|LKtd$?7*=SZcPD|!Zu!zf#lXkWZcC<!P;7=YGW=+9Mfb^=L1cg`}
z48PD3_S2v_OFj@+`I&d3;u?e0D^xUL_~+@Dl(9uY3Elq}sbM0_G-#H~FHMS|Rz<#%
zCE9*va-lVk%Z8i|q+_g#W#U@)_HX=m?s{r3Ok&eG2K-4=<F{!b#p`cP1NfcsULEe(
zj%>8whcrV?Zl0!pFrH!!jlBFZ^#u3`lVvEU9kBQlSj3;~W#q<S27eC5?r6;8vI%4e
z$zRe;)bI4_S_MBDje?XRsE~gRSv93u{^msDxy!IuUi9_%Xp*U!yp^F={t@GC#=jwb
z{~4db!_0$=hs(cW(I&x{+>EO={+%eo?5w_5m%{=yVgH|Kt`h0NxSZ|{AF=Hm=UN3h
zNbSY)W^B1(!v@J)(YKXBgG#woWR4-1dT3!9WI`~!Vp#y3WWPC%*4E(;1sY`g`v)f1
z)zJe=<8v|!XnepNMINUWS~Mcz9ditt<pQ7tqvkUnXs^I}W}<`~G)LiHl$aPV2hZ8Z
z=+=Q=D%yw4QS5g#fS^O?81`%63y%V5JPhlT!~7SrspAM-a|FyC?p?#VtI~isMjYX9
z(~wMFCA79xj-2CQ;|)2)5+CKWBq<PJ!k$1JtwoNy3)|&r8(Jme;Ess_=mymqTu=~J
z!SQ%(8UXz@4DVPW$3-`YM$60b{sx5y#pYIgb;2Bn>WJYqx5$Y>k}~KmCnf3=zJ~;a
z%q`W9PWIo~AQN_|d)N-pLsH&p9S-zDaYlhFy|?9*c$TTX2L&|LK07s@RiYwPUaLXM
zX+Bs)<6^b-PdVMzuBN8FL(X^`A(%qR5h6`m&WuUJvT2vIyd45>1moxlIol?QD;r=%
zos$*`1~Ho8R7{sS*FUA_oh|?%({i3aW6C*I5d10UdkZZ53H1*P2bh)XBNy1h4H`aF
zE_7PaQBDMt?sE69Xc?M|<`@9X^N5jQq^0dMqo|!z7e^~vW0YqOOPOv3*O;nu$sEVD
zNN+1B2C;Bye1-QJQ$-MlF>s8{Wp#?nf=1Y?babu3>36wN5f?1fpoJ@f+8UExYMWna
zQydOAz*Tb;Ml8Qs2F2@sK=K?QTW^)YF~G(aEf8*@gF}1(^c7o}@){W$ZIsmfut1z)
z{x&xur?4I_so~LXEx}uO{55TN#2iQd!+9(t=QvPbNl!anLNY3Z4&wcA|7u=Fhs2mg
zsW0yeej3Y)dsoK>8CQ9;Tx0)0q8A){*QQCZf;rh>i@e*PX{|LM>UF*vJfpc$xq1k<
zQK)z;WX$!kB{31!W-p8W>oWio$N1Oe4W2rdz*TgllzWe`dg-c#ZMs#)&T*Foh*N`I
zh>|aeIZZ?Bhwh5A7x|~-oaXea!|2B`q^kk(Hr{3m>vTbjOo&ebe{d7Ps**~3VQgcj
zzC~^@rWi4Jgxr{TkI_)fu#Y@T)88isvB#i#NUcnY;X|pS+jBL18<TC@<I1TKGR24i
zJBku*Y9eJ(sWH{dfPkBPK2W!C(&(;6n>=Zzc|ROH{Y_gTII7}ksGKd+{YxFBliB<~
zGg6!C6;`v&G?|&{8>%LDc%Xe8Q?p`{@D>4@yg635tQNXii`-)W9yeoh`E=GRxHWie
zOr}e2E6x$(mj)2jNA(kOw=9y|eFF|3CU^J%G}8blBxCGt(0BTD>27DT8tV<+l^TXU
z3$i2!BEF%Q_iq1&0XoEa5kVxG*eiur=6U6%+Fvr44Jo`@YP?~>FSPF#DlD*)$D)e5
zS|6s~wT|C;oSQ0|js2Pj=j%Pob;0}qsG6nTdouxKL3Ir=Bm~dGtCN>Te~G6lqgGhh
z_#UGu4ut9^neF2vWJHONYI*o`B^Xm9Wu?jbL?&1R@Y&{&EYJ!Y3VBpx#kQ0d>jbeb
z5VT$xI~Sv`;k9Cb9;7v>L;tk)u~P1}^{T1KNt+YGuuwbkZCB*|G$7JWM=i7f%;dk%
z*Y+41kSRoFX6)oFM5|zQR<q!J2s|kkNQW`_tPDcGI{htDiQY4wj&^8u`Jk!MS^<c~
z(j8(mQ%j{hi)`JVm@NYO>k&m$FLS)fkm7cg>MCGOa}yoxj%uVo$UI*KoD1p$A?#L?
z4IuA{1Cizpj!IM<0||&0D##3?;PjREddt{>fNp3It&sZ?W1EXBjKm0(z2Ck;tXToY
zw6UBZ#{=FDQAtV)ddnZQUCSbzoZkD4hsHHFHRt6a;}MJq1?%PgiI|LpwF)~F?_1^J
zG(Igo0%dufJmSwZV*?Ac6MERAsehn-XqBeQ2mBc&nyL(t&>u`xVUVsV5MBjC!AvsA
zTKSNTu7F?+gpz#Nzu~rLP6!2Y{C!*UKRxEX5$a+nEA_|6ZErOoc}^4P<cW}64Y+h0
z;`&MJs0up;_Q6M-42G?#P<18zqagr$vej)FdCFfzeG9^BIwE7fr-KgEY4w+N^30rl
zR+J-^S1_&N*`N>3DV9m6V*OlvkGM|y&S6x^^ZpEbP<?LwpyLZ^sDu}}^_a27Ilt&{
zqyDKXNV_0DUP`>0E^`O3(wrwYyHQ^D-)$pv3x^_X^8aH&HrkQcJFi6V7QF&F&b&qD
zGWI%O(N<e8^V7)bE7K?hSrFsHYT9KN>tvxnV}>+?7ZL9-i-P30WUyGWxY#9?*%IV?
ziT|R>+!zM8m`*|ZqI)z8N1IweOYIqTdaVr!Jxi8_9OAXb;5jRRvD~E1m_{fYD{Svn
zLrB%iN|S(0q*&#BlO1W2Fd>=jzEoDnlyD;weN3odE^E@prp&S%p!?wS4AgmTj2BD8
zT(k{1di!k@7hiJ8C9=+YQ&vlok$3;C_b(|8nNz~jH5-Cr(1&?l2{jixO*W=qY%5WZ
zTMS9POg4Et3~)t`A^owk+25v31U0yXzFUkch{(nbTw8r|422-T4l;lsu+h*pV}rMu
zc3syX+tXl~u1C-r0jcThqG4K^)Uq=TV#D4CwOtN{DS5lBHDs)yh&awrC|0uD-{!K=
z!BLED{<y8(?RQ))pYSeO`@y_Eu1NWwf7ORcYKP9TCk_o@76>{)ip792Jk0ZMpy7Li
zX%V=M8#zDeO?Uzb4z!iBFEK%}yaa)HIs_0rAfJjZVDrvaC+Vlt+&ANl;d$c;Me>={
zR;sFoR#Hg~`E2Un?nr<RBcDstrk!W7k3}U}EaT^cp6!J;lw#3(kY5NLrKvlw-i<GM
z!^Zdge6@Tj)XzyX<jX!}t0-a)Gah)`<SQ;9mVoN(<f}12MAo1@0My(K^0l;h)PERg
zKs<-$=+|RtAU!J)@KVWd#B}r+rfo<8V&QrO2)-FB!aF+h@~xmH*qB~WrrO%&<=?h}
zO)r}^9zL+|_`p2h4pAyw<hwzIW`Ie!cul_NFCnE-9r-Nh`h8btk=gcIlA5_D-~#fx
z57n&`g(FHK(jUY*!RuU6q{*sj=jDfqNb7Nq0=@*C`6GWu`OFvs=2`qW##c_CV{^6q
zBvDf9&Xq9PPyKTXem4aAXHEbV6{*k5&!bcL)OmaEI`{~G5wt98sYi(VFKwzLMh=o+
z#V(l60N7vqx2B+4y_DYsG4j|MIC+1YSeLOfDl@U63H?BhV;lU=$BqLE#{}(9GAnOW
z{C)b;)FU;W)ULN9=nvjI4JD!hJOKX~o2sF^HP_w%9PLlu%2MSU^Ni6WXE6dx<<BLw
z(<@1!h4PNOW%8G_jI>&5#H&A1i&3Gjru;Qp#^H^n(Ub!o&>Ra6W%*mss(Jd9Y4Z2@
zR~0m1`A1TZ7+qi*wPA;J*n1eG{4-V$b`Aj>FtPlrI9`VF;H=|0KtI60)7uLOP{U|g
zE&oZY0pu0x;2ZJpsR->v?WH$^UR7G$3{T#Q-95EhWUj%KEbrx5zw9?RFh^EZrI2D^
z5aapx_qU;3028td>jUOGP>QR#G(Yhja}~X;9GjN|=eo5|feu4iy&(sABal%L#*98#
zDhJP1yyoJ$a!3qF>t|Blua-l7K#F^*fq?J8U|Zy{Xbb0(1pqCT!{>TTs+uZTR7hGg
zL;2@?IbyEkF1Yjpad)WdDo5HJP~5JrvA%%CJ1UK0H$)cQs0-xixen0|897*v@fR!!
z#&R@{o$C-!eOs=e6{3!dX0#@`X%ol$@C?1A(B?j2q5x&xQaQm$an<E=Vsu!J$d8kP
z%&hI(*4i#7dplImc7rH(@wa6wA*5E`>3t(t0$?HwtS^vL{2La=SfJ%G-u9=QI#)65
zSqK$t!8NI;%~gPUS_LZmogRZ3I%t%f5$hV&v1pW>X>ZS}tg4W+(%=XqM=f0HypYvc
z(X;K{Nm;lg&q)J?y@0@=2e7W^`uE)UZVZA*rXQoiiJTW4L1PjuiX3W85mk8Q{L~ur
zF@Vw`wCIBLjKvZNDJ+u<jZ4Q&s+5bou|DiH7WzZN1U^Izpj@1$Q9R&vCr~SeI)8~b
zOxu_U*J?P7E=`?jG8T8QTxN?mZOYgg2CH12K14%2EMenaG1p*MsAYQ}pkAaJ6>l>z
zUEU7PD~<FO9XPfo0>NRSTe&J&+G-@xiX4=Nfr^9>`ow+)#~O}D_2D751OvUd&5@x_
zfh7Y}$-=bdIX%oiDQ{>*7|ifk6(CJ4zGeZvWI4?UlM!<r!Dp%kRPB*Ih9$V!i|8mU
zJQ)?l#%dAe=u0ZW=-57cp2soqJhsBSY%AQ`CRfK-isG7;X4(DM1SPQz2u$pR9(k?5
zO{|R$LSC!K({-D?JAOe58Y)}@PNGYW7VdS4UdzA)roLSNU%eJ<KbH9~rgNF9ZtF9~
z+o8TpjhN9y<%tf2W3kPFVvhA^TAPwe5kz@~bAYL0sOzo~4Y<w_>qi|&qj+VUuYwy%
zU^=>Ib?*R~*ec@_(-uSAs1^j~Fu^##yllKw8V~Gw9DI925I`FwX?DnsL9UXMn}b^`
z6H_PKH44I#2}x7}rCViEnjZ7fERc9|^q^s9tgSm=rle=^s9~CmA-SgdGj0JoJ7Dq2
zO)(O!cS1dc=MRGFWtwqlrqG7c4pku~Z(x#dO-80i`zlBbO`Kt~#+_%QWv0)AXEGBt
zh&_2hF?uv)mbKo^tB2)gZwx^&P+^G7^vNxA_Z<Ys9R{-wp*FYrFl`0-XB1|<Ew~#u
zN@>|Ix!u;GvAR+2NDXV90j_}D>Ca$DgGo_@465Q?F&lv5xXJ0&g*cP%jwzR;ghLAg
z3^M+byL7;Z>!do7PfHA_az-F-v(z{NRC=lkMB$f06QtI^wCowqh$<vw4YSdvGK{I~
z_#Jzx&IpXM4fRrQqZ#N$N;B`6g4mJ<dsfkfNY_T{-%#Qj?HdBma!<&q8e9o9+XrZd
zO=u&<ME0gcaz^5qia=<3u`~zCad!km&}yY6(GU6#rpb8WoC&QxPhx|xvQDd=fdjqO
z8Kx%PJ9nQ3Mqp+t8d>ciuoZ@!GS4isVcMd34d|kJv|IB2%)EJ!6p(A}#xuNjZ|Gnt
zq&Ie}#NQ4@ufqo6^=PFtk=hJGYRhD})zTHKU(>+rdh4azxa!(-uaq8Z7;%-kT38_(
zS2@So6^1ABe&%}D6gAW*De>o}9l@9s3VDWA@%~ZjC%96{f44YQMMLmAEjwN&yGGs{
zy&`b3&2QelFD6`FXqhedryh9Hs#1o2SPzPT!xnkK)(zTNEe6e+w-3fj*1?~qw&MHZ
z$uwlCk%i=;M8=gGW4|6Py*~~lNCU)D$ip#?e4({X9!YXwlsxKRn~j2*!q@$Q7$?}>
zN?sGgHeVOvxeo@XG`7$T`H(lEo(B>R!s)}#JqApu7*r`<Of0ZiN7m+fER-q$5;aJP
z){9ym_m$@X3$HR#uk6!}@<fbSE16hb#V38lRF1ix3S`65A*v4jP(BjhV1Lktp#b=3
zP_nrZ1@GGn)$){|`Q;T=Rq}Mq3T7G3*H`BQv3w>*I;pXxCCe)SQB?fdpq3rP)Q_l`
zDQE*}CC~Xxl*=gTHFgNVraT|peaNU0@<J>bmWY*HUknDH(ZV3`7I`Vn0rj0QAO`Nd
z?0r)_Sris$T=YbK%)jCNm8j5y5<-3QN;J&O6zcrFww#kP$b&{c<NR3N=?KZl$buNx
zm_i-ma%5rr8)Y1uAzBud41sa2^ye64Vy4^T*uo0Bsa~L4c&#i+OUyzh+ztdi8R*m(
zL#XLS4HsEz|ExoL2zUB2|IUsp8m~OMSW0bquz5ok>4Pg$`^zyV3P={(+X)?brH|Pi
z@NiENf3UO{pir_Z)*dGvU`1$CyJU5=rlIpz&6@}Ha#<5Zwu@VEI6WY1{TXp!-nm2R
z*PmKo_%07#D;v91*7+|C+osjUpl|a@tT(APeB^N1V7s=p0QcEO`#b77Wn`04Cx^Nj
zutQ~Yj7AMC3lHAn&v;fb<7Ef@C0i5ojT|Q1`~?KYtMPziP`0OMEICEpc)RR)`(k6b
z%wOVhU@~;4D{@qC={lf_W_(Y1>ak$5D_E!+TfGjKGWqmLPVP>W&(T(K`MZx{8Hmnv
zA8vB_L@XY58so5tlMyQ*F&P#7s!Mt<_%hiOEHNPqFmoUg!V0T%S@wn;XoMHBg%b6X
zY34LYc-zJ<*%#z$%@*WSK48zq20X*^M+oMSPy66XAs-wSAcJw}XOgY66GXH0AN9N9
z@9=+Pimd<s*=S$k2J>QDaLMP2nlhx4D{LNsQ0(RNF{N7EC!tY+Ux;y0sZ>H~0HFy>
zcu&3<Jz^iB&O0o{a7BJ82v}%=v*61?hM_}-$yZ`^pgkgr0)aT&<*UZ96)hd|wY01j
zCYYcm`Fe~GsU)}p348h*-jp6ddzSH7mTyMSRk=EZk4Dd#&WwE9I0#or;4+tPR2oeq
z(ZmqQ`~~ve)V*PBR^|~m1<LnQXFTbN>@Jq?``fxq;69iyzHMLgw;fr-=3FGN$IwtS
zeU$tleQP(-dU+^V3qLH99quvi0p3j45BH{R?;m;FL{nIpKq1gz5g>Vt34pGKk#OUd
zs;57RJyZ^*8t%NG`WTFvaBAWd(&#|>SsDZH0{Wl93qOzfX&8l7qWDD+d+4ZP^2<a>
zdO}PewC(<@wA~?nF}%%?#6TYC%+^!0Piq<c+N{}$H_30_Hpo)n$Zt)c4fIYyWl|$9
zqv5a_fA7Q7%EOS~{SZTcNbRs9FHtJS#`A{mKl(3RMj{^2MesXOhCewObb=twZU{P4
z7fc=avyA{Xr|RV|_V+xC8~-&%Hlb4fmPoaNen$=6{d=MgZ5|%Ph{Fd<ZIpk+;_|LJ
zyuAy;=AZtI?{Q?jckN%+Lt8h(Q2w1p$<m<=AYv%$YP6{+yZk3cfTBo*O?gM!8*!jd
zCGH85y_q`Mg$0A64xWE2UV=z~l&ck)r}o=G@2m-B(`CPTjtv5^&%i?h5M%#&hLBp~
zHIYxmdipbkJ-F0R^xrX0Vb3eiy-g0Br$`|Rm7>P?LGu(-L*Xe6Fg)1)Nq;!Qi4U3Q
z=qXZLvdx(Jq49t|DO6mS!@Msp68RP$J~@1zK}EEo=?SFm48m&I!bY%QAI^dGazsoG
zKJhj=(%YdJX64ZpFbj{GXQ}p@LVy++=SlUIAO(VD{LJF6b6j7Jjy9)GuNo`I%rm6e
zn28Z!bfHt*JJx$+@Dpv7HPE<@o2PK%xF)#uj*oUG(G7cow?iO>DZQ*CqZ>?4G;+a)
z!IcUPa?(7*cFEBkBy|UaU>X0oQ%;_zIP#>M<(=N79lMyco)gHVkW-v^;3r_lR-np!
zoaVc;<MY%+c8UwTocOdfDTYaFZFo7|TMLY{c;6QfUhK!c55R$YhD}4$JK&{0(+39&
zRM*c`GFC=ED|O3&WZrd#7}4~M_zG>L-EvN9f!83KEvD{5IoC;K$g%}CtET{~z-@A#
z{|<#w3!w0ZOd6*wYQ7tq!udwRi8JH^??Zt*6&3WIQdhWeo+I73r5Ryln9)VX{nK#W
z!o~L28StTAVjF=9bw)1rX_y-jD}wEKS*$1awmC^qw*&!*)HVr%T%IQLUy(CjtLchh
zBUtwZxzgVT>nMcFL@9#qz|Ua1?v|^3txC`^YbP-}wcbJV+-U*K0Fixm57l7fUBIqw
zG9-8ssy~#ep+P1$#4N0kd27RB#*78RjOe}^o}SU#hx~;-HDaFQro8(KmdMDUil&g8
zkC8>3h|KUl)lu<23hpD0K}Ltns8-tqB}(IkJZnwo<!W08+{z8#<u!>)DZ~NoVn=~N
zpr^~VJ|DYZ(HO{N{JRq)(rJMm2R!$>^lYU@Md3mfx!#@)^rAXC+$%8I+odd8H3S#2
zN<92oCu58a4hYehu~HrjIvW-6WNgd>d4!1TkqY~D@F4yt<6@!eo0w?R0H`1<<L4P(
zgRMgR70QF91;YgbY?!xhGO0izZkga?r?$oDJ0`DG#tDEfJ8-!WLOX8=X{({A>X}3X
z6bo=kfNZ?}X^Y%wV^hZ6EE8=Lw6-?OBp(q`BZ@-U`uLDeLDm+r@Om$dA(MSXGyzb?
z3qrVGro@P<FdVyHcdF5)ENW%k6eO%I<Ybz$>Ok+<SyGh-#oabNewk7(Gt$nYSA|-)
z7vjd>S%hn5;*#PW3mRiiF#*eNRvIZ|PkD39YPs3JDY-|6Z!2^}u<927hGpK}t0A|}
z+jkW4Q)&X{cG(K+?Y7`(TsTtOi3hjGT7#FEehwrtFL%6+rwqa5F}X_aG=ktg9I1E3
zer0KloZKD68b57{WPC`}YN<kPfv~QQ7iMQ?N{zpOpm#k7c4H6MT7%Qb+~AL_5xE*^
z>QE@^oCu|mPiZh=eGmx{0edu4lyyT)nQZXi4f3Z2sTuM-SuwS((rD~}osBvmfVi<u
z?r{!_P=2T-x*{+i0psHg(x^0Zsn1U9X6C>;$e*VF#oBv->2(x&zeQSd&T+vQ9Kgnw
zEeEoaE!mPSB-thy%!F5(u|_juf@B$6YDwg1lQXiim2(a-8m-7VvxFsw1(vX+W#Rt*
zRn=$4d++mo_qq48&G2^jIeogTtE($_1;U7~)A2hvY(ZkqdQ^LBtv8?0px7yxgb!O^
zLf;Q+CgKtkT8Q<wuwoqt`8Io1J|}gx@<q#D6jSZbsmhWeoy1}I!wAh4WjJaB`BhYq
z3Ps|u+-HON4k_PYEQB?(Yvf;8v!U9~`N0_tS0FD)Qd0mvU2nZ8#x|*~x2?0y7T1}F
z#?}Roo3f6Pq{zYDQca!M8K1G;9J}f2Yo<@OS67so(%)~lG(#XY5P*T~J7$p;K^|mw
zYo%&E80v7rTN)j;b&!s4S@WT-E-C4r*V2GVhal4sH>=4267=A<JZR0Cr{7>J@^wj$
z(r}Qp+*aOom&9GA;(hEjl^%3XwAbcSuB~#)oXTH&UG^E?NaDPX*z2=SOtD-U4$|8j
zVp^@6(3dHO`i<EmbQPSbqMOo}zNu-^4RanMOS>40Hz(%Y38~#{ZwaU_>egL|Z>=Dn
z-8SFe){It^n=|TUweY>Y0bm43>m0r<UWEgPlH8sT1k!|m-8<?O*<m`^l!xS<3H4Co
z)#!#%<-)t7Q|=1pw%~U+@Br6b4TG;}Gm@>M<!+mMZ!Fb?<1VuIRjNe99QNAxH%KE=
z;J3RlkjCt69|*SWWzVw@=G?VMWbss%Dj%MGDEsi;!x}0)GYlrI5H$P4&D0~kB*A3%
zkvekDKygb-5Bbq%XdUE?vsKxxyE*egDJW$zfG$A@NLE_$xjj5_R|l=0#P}QIpsmU8
zgs9-EX<J)o;vUsv{wrlz7pH>!qT4$-+<RL}+`c}bckA`H*oJy2Dp`umN}aYb7Gy%t
z!mc*Evq7=r5>Ww_uCSAv_NIEYbRl8kB4eBDY%d?eCDyj&KsT^gb%mk)FSo`c(z5gs
zX|OGdm*M^-#qsuhBZIQtziLN=pt3ZjTQL*dU_0yghhX-^PVUM_NeKe|<FC6YYB%hz
zBO4&7kR+rPML0R+7=GAUHjmPD%gx>#hnR=BI&X*V%R0YZ1T^73Cfi^8Q5Gted>ab4
zF-Nu?XvVJ06WX{ni0okH`xTCaUt7x#)wRE}P5EW*aK0O27oLGWC8fDO5?fo692T9U
zbvIAOEBsgmAmVKG_dZ@fB^Z$dt7RXH4q8%%x9sD!JEZ*~8?GoSf(W>!!w$@kRNg1@
z3mb7Z4!7Q2Hd3MM<}o|jbTJx=5fZV-JlMZfl$?e_#_fr^+3r*w#)<c#s{2Hp2;8Cc
zSR9x3$?RA3M$|jD>lRmA$m#V{wTBK|1?|&yy(FtSB|2)K$vS$F_7@xZM*D1i3(ecb
z_PKn*Ey8?PDGEOyPmrgEP!#(@eF9f<?T58ke!ke?fo{cc#~i-sW={Eh*_YyY=;$N=
z)|Z?8#@#WxrhA@!C6=(GkL!r+t3jS#^dkFO#fXJN0}5gLdc`IV3FUTFXte~eE%uED
z@fDIH^%d$J_?@lx&E}oP0T$dv=&)~9`1q+R{mXXycHL2(Bvu@N68vk>iSfibey1+-
zRRr90*>~%s6G_Mn*|6^gl<0jwZb#p5pd=m01~mG|?B8-Mq3fw6P3D;&)R}g+Es!ed
zhqc_cs81I~{3zet0P85>R*4S73FgNQ<O#xX_C)&7CKr+Yq-l>^aCW+(dGua@pEeSS
zn0Jk->T1LzN_duK?`JU^^5AtX8Su-1m)Os9hEko#iEcZy{zV;Uie}O>1eX1>&S(<O
zdl2A%&nLBhF1$h++}vruYWkHz5q_6d{B`rRqPk?*lG0=!s=nXk(-Jth(8EzM`adc~
zm@sRG{Wf1+&oLtx$kAZX-^F7y$XGIHzpw9H&IJ*F$W8+s;yQ5P%zkSA*imBn$7Tq+
z>ODNL{*+H^Sh@viG^X><`LxnfiTQ#g*k9@voZUlms=wwFI(q}KTkOAT9qxVgR))XT
znT;?v4$$Rrrr}TV-`OHUp{#2MaY)gYKV*N8enwVxA_2~eh?9TRzByDxC=esembxC{
z#Opv8oNVoGOI;;76}o))XhO(FkA}YH9!o>3$%z2#6xzUhE)C_HuMv|}{<^RL1iOB(
zd?(Hd$7lPer5VG1#%Xr%Y}ffC6Q+DH?vs6nG^6PVV@B_zyl=dJS&RL1zkFW;h+LYE
zGWGi}ZJad+(U~g5;{i)U=_u|tVw~Alp#64xU>%@uIPo%f+JoxToD9<Ai6;_-kJy9r
zX}wE}D?}espLSxTfmW;KL-T2o0XS>EaL^vsJiQB6K{CBqayd~vJX_M`UDulVgpS!G
zvPGZ}Ar?^##oQy8rrrsodf%c98-CQ%)U-|Ng00r{9~}^DCkM4XCOeDMR;D*O)GF`L
zW8*18%KGeaqo0&{0y~~mJ%l{k<Cli>k`n&!vnMnF*)9bfNWRCOxYV`Rc*FGJY;8|k
z>YDAVA)X8Dp98%v<c5SN$E#o{0f?!jcuIE1j&Qk;{5*DAgJQZ)$=$JB)~ICUsY_EG
zj#K(#9Wtlq2YZ%~7vX7RCgOq+@YWiwwWnu`+#tjg%d#AV>5S1)P$yaAn4OtV%bfyS
zqiZOH6I<-8e44wwm5>Gh{N;9b)`=b{Y+oGmGqNr)#Pf&QGn;pHomV(yoYU0lI&Xbr
zwVhk%Nai%OR!i(zF~@1Mr(JcqotJM2Hd4uQBch$Z)O93^iy>4T>B{}6$K@<W7e8Wh
z*VqNItR$A;9(cR38G{n3Mr6wP_~gi{3y6!ps4j|xLJ(LiRH)WP2{zirOT*d_d)S>2
z1I_OuTwIc$5n@a6fb*hV+Vm_jJXDF&HDj$n#yV!tu2aydPn{xyV*@=WdlTW-NyA@z
zZnMZ_{-X{Hq(BOlcwV-o$m1CeXgRsko?k&R8;5LrL5|0>mIN7KRC1k^@Ug*O7*B^@
z%_hVix^l{_t8C)ZFf8=iqzX>AjmmR-lTFU=yW42=B_u?0n9@A0jAT+iuD8pwPBOf-
zI^qH^uX9T-i1QnGz!%xn+5+b&1omhGC->8GHgc{Yz)&8W^tMU1Ag!cbQRhCThg4rz
z2HMRYYLAOuZF)4HSOMF$!e+z-mkileIXq#j92(^Ke|3H;j0z&{nS9K_x6x+CJESie
z8sr9Yk7IpNGw=xE37J#3w9cwPflrr9h4Qh?4not}Jx@`4HYbW)Z|>P{uBlV$(j8UT
zX3v~@F88OD2_jrqdlve!f0h31#aV}S$DKxDZ|o%%|1UXjoV_$3(76VLJt*g8b(79k
zx=#+!>jN32akZD{_v7&Na|SqVf$Y8_yK>8Yo+&p}5_)-yEcEuuIz&Z-gT*`XmTimE
znM|79ZWU|~$~t+*1nY?J&RRG~-q8GR?0Ec7IiRH@*lw*{AFY||!<dLRuM!-{rZG1^
z8bLzAO3$m)md6GI;pTVis&F~u+-F&Lvs<?LsZB~q@ULN8?8c@&;=0K73g(S+sq>qe
za{0XD=RxZkGf+Otp&MpALSEg!WI@xX?y3o;k+iZt9T2g%BEsAruBYmYFQ6vCn#u*(
zy6k*mMY}v&)sG*u_Nsc8LPm9=R<HGsE~Es;<ehbG)j$OfL|>{yF8i>-99t+}M0w=C
zgkcHUN$x5`6=$&Zx?*@yRLoHlMAGjM+TvKuo4fjk?56lV>6&SCY)P#5-0pU}IjXyO
z$6kFGV8l5Exd7+G?d_I2p#f|xME?xftrgeW`xXO>rAx!U>$DJPlVQb2%$-XbvCo)%
zl`u2pHRLedWXobC#P3^45Qq^VU<-lauq|)u6ak>gudx+bS5v;~#9CRO!~83MVO?YS
znxr>G7;~t+Hs;B7Ol|hMyUf#bDKNBAUj6kkD@?#{vVTM639swx@0_VPQF~+bHdP^O
zZ)z4bIAiSyA=e%D=876&se|{H|E|o{Z>i&T)k=`$x7Me_bzE$f@3zDCwtPD1Ph4I1
z_MC&$%Yar2b`kM%TeiDk#kpR&d!<1avA%r_462Jv5_XQtUG|RVn-b8)B5$>K)=?vK
z$W@GcXzg9GgBP@<WqJ1~<IsaG<mM0^-jfTV^LVGdx1!8^5?u_~`?60RxjJNVi?{bz
z&a|DR6Uv1lg7|?PaA1{AAB9o~E&O(I`(R+w4d*=HK2)j7)syDfhntCd-Zuwp@C3rc
z1NM<RDQDtD0k+#mvradl-VS#nKkZdHi#6O9vC{5{uZ<snzO9a3=h=A4k1=3t^6m95
zD)AR^uC06ZioO9d$&iCIFla#^Oy~Ms`Jl7GmV|`3A!gD>EbX9eto{0#EUG}Z(Mq2t
zft$9ePJGssNwzsBKE@aXd?hW)9=6rCRDk+%?hFa*?1~78(0bb%U2|XC5V@kZ)sKU}
zl7a}8x9#-{a#Bg`6QeKJ>}`N-M*!)X%ct7Td|hq}GC>m0XS?b$bDdSco4q-RcP~Bl
zU{6ej{zP}gr1-g~`ILM$HZV*fVEk5g%e{5x)4S)}zWkWTFhL$LdwXnuGk~M;5<;SF
zb|5=VTaq)P5!ki?a<$zK#xbd2LIO|w?NFtE1YAPT=Pj|rF~iHI&9Wo$vlr#m(PUsl
z(vzd69EW9i1IjR-aYi1?uZK5mgw>4=%sXBoj)GTqWUbl9>WJOTG<uVAO+n~Kkug7B
z(P4UD-$EM>!i7R@XhHAd?oI@%6BP_77L*U`05aD|JeK6He~1v?lMMu=BnyYOf2nmz
z#}pTLs@CDaDy6M>w@+l9LIC9qcW8LCeKN-_enEC3(Ke9-<xYjGuutU<2>B)=k;JF%
zw@=rFojvVx`%HGNZY0hr<MC%>%ET=zP2uO7k4GkExuAVMx)|&q0svpAUjoOKCohuM
z%3ATos<P<r<pzY7eW}87h7@;Sj^*uMh=XP9W#rRV|6*Mr`PG~q)C&$tG%UkyfrzB@
zuf;c?_ka7xzFvvpgdQn%zL6b?VUlS=YUT9lrDl(PGlr;pD_u45tvckE{4SQ*w`0h&
zTCl_|`_~xqrQ^=C@6@^U4#5_Dw|<86R36XA*i2)+-M$w<98q2bVto8Mi0>z0m5xqq
z*A7VMzr_b;&Y5LDsLM1J5AYx6vvKys1s6YR76zjTOGnlM`*E%Xd&A@AbUk}`e*7e#
z79b*Z)*kz5eC-wIoNhm>V7sat<hGyZe8zfMxsEI0|BGxN<Lk=|l&W4L(Qf-?el5lU
zjl>B0?{&N=ux0$v?)z2t!@d?9iDFzf%O>=}ex2{g1|YvoXrcWkAkTA$;Mf17F4$GA
z#X8h}TUm>NmI5qW@QA-lpu}O|m&Y&-MPC2D&KgHS9QFQC*A#nYoBe0L29G33WK2ZW
zA1f@~+M)aI*4dw$*&UTfh<3F7IY+7<HrQX9wb50R?xRMp3Pbs89pLb04qwqAISk0o
z`d>|p;??3DlVjgT`&%6=xpaE%ziX+jhr8?ko@0@}f3UW@?H_f_oF)=N$$(_GEc9^t
zB3SFWdc-Skw!1BJrIBYn;5fbevd}eZ0P=FcsoL(5b-_z}dN;f0vd}*YSeXC|f)F`Z
zSU|g1w#Q1y+9sj}*<;J#t!V#L-{&1zJ{JLNI0GJxcJF1WOT*Q=#gzsFcAu!xfdXqJ
zf6V*FxRg_-xl!kS%Yuv6QL6NG8WD1igE@2=jm*P%l*jh|%R*xkUP3ZNp_hZZsaaWJ
z>_F)UEKA*H8-85M$Ua~XT;}?<shB30*n<+lF<)&B9GXrRX7b==u66D2Yj4>@@~P^P
zv?Jx6<Mz-Bcef?Qf%ULuq45(_y44<@PY7Mah|06@5w$;-U&gq~L~pl8Ht-!KQ#!E}
z9|oIZk6LyrL**K?HSa5ZU;0dqkqsZc%oTkE1xAdQJtn4ff~B{|1{6kCo%WQ|>~VEG
zArOQrm65Ku$7hR9pEQB<rDJe1PskP><*5`)((}Zog@b?@OE0w%SZuZ@)fQZ~<w3g~
zHe0C8RY}WE-3Jy3n<wHq85W+rELozVUb575c60BkQ0FPR=;>hwO3F7{;T~g^=%-aA
zoQqkGpu8pa)LaDllj`0(l62VV&0r*cYSG=}9IEiNY+-1qze2w(0PL`*2T1X|Y~e|Q
z^mj(CmkvNl?!r~ceVX?+R4!-6HsuQ40Y!|TRiQqcL7g3VLauhqefEs#dSunip0*`&
zd3t7!!tcOjmO2%JeLp8#M3V#EBOexe&!$CjM&d}73=ST}o>d{hoFuF5V{y;Rr-7Z8
zh;5G79`huMz4>-rpCI`T&Jw&Cw=6a2y+fFY(LyN)-v!yZFjcWJ{Iv_0r9u__^Q8ZN
zdXHb`I=(As*+l^g6ws{e#iJd7W*apxyQCSNJRvn#Q~^4F?9#yFK3qrb*>zqcTy2yH
z_?+A%;j-&zv6iDzylPz~nCm=#ZoZEr0$9Lb4xQ&k|C3lXdwwqPs2&VX8i=~DDGU69
z3UT+Y&D$EC1A5^Y80z{Q$R`M*(_}*SxsGq@JXQ)WA&?Zq81QUjC0d=Wc5>E9M?Wd>
zsGlPfcUhZUQD*@r1DldxavZ(OI$oBu6Bj^^2)?()F0Uxd3BwmrfJ%9zU}tUYLpC+Z
z*D6Bzj@z{80hfXnTaYM3n5--6{=y3c#a+h|yE6OKv4W`LTyPv7`tlx|9*f=Hujs-V
zb@1r@dhMzlVtpKT(ynfRs?-$j`w8eap=g{La2Q!Ny=^{-_o93`&fV~;oRu%jmY@?^
zDYSW+!(euN7#_9{kBB*S7KDbuNy1296YGBKSvS1eu5BQ9$Sn<IEU@e9in3x#aw9{V
z_RczcaXzi}lfvpC9PmqO{{(vtap?;8$X6(FSV>;=z%|~GWkz?wysWOf-_q{;<@!qS
zIN9XP8op86K=pLMfcWjVmsi>`c?p5<IAFaZzZ93VY&fZ;X=AbehJeJ91+3_Ry|RuA
zRiP-UJO#(93)-6H@Z>_^uwq1{dT)<;%L|#?@sJt#c1J+*((}$c-#YWnIO4+Z5CsOu
zM{AkeO{Z`d=G^#6Jc-R~z7%}sohQTRY?F00J?XFm_8c94Xj^xMJvjh5qM&TM?8c@a
zt-c_w`=cb6%&(wY&`lm+khv#@`=V<mOt%G<X7x(AXzv?B88**)^8>+KMPy)Kyfw0F
z!b@#owkuhI?qiZ?+p7ZMi6&!K{qgWXt2N&SD*c2Z;)V!?RS(!;_J9hF35S+;sG0Ci
zO%=Yb=IL9s?9^3}4&kWcx#2u?`imtG>rsd`7}18;!{gBO|K>4U9LUGzKny9ASU1JO
zEXE?br0zT7a@jjK*Xb{~)?Qs<?HWIhdp8*=8|EzyVDNTk71kg<-&%1Tr!s}qvF(=T
zW9aGKD#l-1RyWVwHnOUTZ7Wfur3d}kVzxZrlcb<4JmmhoqSir<C1`VjrLBy?wv=}7
zHFbsM)5cZNq0fG818e!+jm?($x;pRKSjGu%Pui#)oV)D}`7yUG181<+-q=86>{W~~
zc=o1XaYk0n;=)Kk`sREaIYW_G^ti{~k~=ZHz&$O$Gqp82)mtm}PS)iz_(r|0c_GqZ
z==}8?B7g@6y1l)EdkPBqK1joW-PTMt%v{oII&HAqE6}@SU+RNBen&G#zgR&!40~tg
zVP*{t+Pku|N}0kjL;T^p8$b|%0A`XAO`l<Jyr%&h3a8w#@2$X5YEkaWeP7e>A?KR)
zq~1nRT4ywMh`1Phv$>!tJCy7#C#U1cM<1*-4W+5#mq-<#rqs~be5e@&`ViU`(f8qe
zXYAD-x*|_z(__fBAIa56j>f|F(F%t-Tp+rt;z?JlMFi;``Q%1u=|Ni^`>6ZIK3h}g
z#j&dpT(XNeMsKmT`Ed?pM9eURqlALDjv(rF*)C~Ll)$q%JL}*axUCNePKF38eS>YN
zBbtrEeu-_&56i7cX?{I#jRX(7Ge&pmCFk3wX3g#tGv%B+IwjoKH&@WOgyeUF(@B#R
z%eDkK;a>au$TR3^GPY)K60Q~ck)at_Vz*T^Uerx~Sk%1R>tGl3=?<+Oxj3=cCHd^O
zodMF3Rj(R?X6(woLv9mhNH-<}#O|g=o|U7_`w`nyNiP>M;qTZsU&MWHFr+sQ+P>yR
zi7hDGMOK;OLEE1#A`H3(PN%<{=<Ps)#p7D23up8u1)m*kgw!p1M`q|ahX99bNL?3m
zDEo&6LO~hqj_Y#laNQPMt=EmF_(<KMgjOzYCm6zx1{R)sVg*QltbyAR&JlP0u;W=L
zg{G1w=n(u^4%;DnVk4rDSZ7juvF0+Ae7sq0F(>%wq5hUIff5VF>zy{7eaajqR}jVP
zpNMxx{-3Vu<^pdUX(p+lU9r2$E+k3*Wb}srORKAGQ8%Y9sX*&QNqiA^?w+bBGHFRW
z2=Iv<LP~US^at#dby)D)q{oEwTxXxEo$F4?RvYJWzkND~!j$}IJq!hmw2Pl<CN3jO
zpq18Ercqs%_}PktC*BK6elEW!G!3`YP4@X%4<e<dcx}2HJq#oris9A;mS4;cC_x9D
z)YzA@PRx?b%-9a>%M}{e!7)n&u&=~tZsgL3LHlaXCpH6d93Dz470}n>=sWK`99X}e
z9S3^=rw(+!k#EX9P>P;?5GBp#h<!7M4hQM_U74R{D3ZE@3y|zviP;c-P)*Fa{O^6|
zjA$}qmUh~=W6@@{$o_$?>|g7AU70AJ<&b@+83IO4hOpE1zX~P%Zno$a0XYS^5Xy$x
zZr^MA5o@EZwB5d6VeO8|u89e3W>|{g__u%<(R*Bc{DU05OlmqEj>!WulrdvmM1L5g
zCw=Ll<dz@RlLxiJ5a}@BgMM5WhMSM^EZ`EAp9I9^qh&wMVam?8-g*2x?PpmhExhVv
zr}=rF7F=8kMfydZ2&}zErKrs@7ESwQ5YNk%l@11$+vc;&0Kcc^w_jx^RcytB2MCTP
zDEoDNCuzKx=HJvM4@!Wg!39=vL@YUq+`0p~qOH5ze%lmVLx`lm3;2$#x^Uch`+c1`
z5{0hzJ%YC354oHoA-bSfw}c4257~d#o;tew`@8IqsSA{wzQRh$c;pYtFI)0YfJAB^
zZiE`%jZ)%IafF?9?rHYt{BkhFK?46$nZBW3d}qj;=S~cNtstNys4ff&-lX+lx^bw}
z<?g@5Lpa?iyOBgPvVZ18!mucE?P)RbUif>U1FY7KLm2T`n|}m4T-Mh|7<_TPASMQy
z3Z;jMi?0R)Ilcv^-mQd=Rys|oq~txSGhQ+u-UWqr*8xf<aNq_^AXtJGsw`-$sItVY
zyO*jLJ}T8e$bFztcV9O*&cM&D3$6M+awu#u=plFOf^mMCcA@0~F-QSEB3JI2PixzW
z+wwR7sXt3?am`A$cq&@ES2GL_3tG7Eyin&^-9$T%+CLT7XOFCE-Gtq-P|OcYbdU;X
zz1=(O+!fF<VneBQ;V&yr<FMVQr0%ejT^l-PmZDI7Pv$~kD_1@xh{cphmX+E!7%P`l
z3++-zj;!kDp5CFG?0z|CY2l=+!J8`n7FzDXljHI6yMF_ixU}AbjKne*T6{K$p^zK@
zoIRk{u{pFip>Bn``0RqA9yDwAzyJ}%1y-d{@?MnjH>DJNP*X>~B^C(MU8vJUM^;Un
zU=PlA?UrzP*ixvAue7zJmRMj9Y5F(}`S(i|O05$eg#-2d|Ip^SwSIQP<rnH6!?#$W
z(Hn=yh8I38+ciJgu<>=IH#Z+i@l<NNE5`ggG9u|ZW;wv^vxhg|+~f8NJP`_Ycjucg
znq-eCuF5B=Luac{@C~I6c_*y(dOj8G&LfNKEYBqyt1KD%d%RE?BhN!iS6hbtKwQx_
zdsKE8R8{=A#3c4&p~crhBUBy7tv$MqlS4#~sjN_;&Ji3lF-PHJ8kEy5I`pUK#0Bs)
zr&ys~yE0q=NS9@g%>ks(PjrlkytdGCe_!AMNlVD;9rn1pj5aJjVqcDyySDhT&ld^Y
zv&RSOwzYQ_e)^PBTjp6zo)9}yo*Q^<6p9KWR-^;6A-WStM~6MJSzoycV&ec?fRZyz
z2u_9egcY(I^Tpm_Ps*`|HDwGxRA}+1gR;uty3_u-Qcc3tIt!&IF^VKEamq)^H@6q&
zf<HOGpxhlwK7>r=lE8*Si?4bxEP{{+06+v-M?zblk|W5ZGKi8#3Uxs*CtW>b#)PXT
z+iBT}^h)B2nww9&lS74e$HbUmFwU-vx2I-{KI}#lze64kX`v-HC08oQsDFBXLxa~8
z!n3})!m`lr$h8KG>t=ge{UZIs$GP7{P%>Hs(z;+S3qHL8&k;TCiOwC(3oW8%u$6X3
zLc*<K90z0#E!6qMTv@?t>na_s(ndS8d3ps>gdgjLI)`<*SgB4iFG?z+)M(r6ton|!
zBGC^Si3)YPqwmCPN7lC4*%c;Tt@id-Z=vky5uRQ6N+9w)qnY_=fR#4u`a(-QjWi%3
zW*^)1na$Iqa1v~H7g`cAbRvk;Rb~TRDfkOLv2*IwA=lVpn2eJD!S9yZq6vtP5HA_h
z#JK^@frSHwpAI=u3rVV?fR26ktmeZ?L(I~Jy1KH{rM7ENcm^q45kx&iHxz7rUSKB~
z0^tVXPYM;eNQ42U&O?0e9d>>*`?ap6ma@h%vC!hP7l8|ZS9Kq^?_l1uaX~?0T@(}4
zk1Za8OJ%OFZohDPU0(tehwOrSq+ijuq@z_R>R4JxP9VFmZtB@HXU{H_;$p_>ZHsJt
zZJIvqswst1F5(@WoKh|xwTtq*9y2IdFk}Q6cCzp4Lc6~sSWE4wjF}h5Qe1fH`Guc8
zY!P42is#3#JeA_KxFlcKdWu&eZ&Dep3xX|AuuGfIpk(CJ<dG=UHCUu@G$s2PoA=|{
z6{81ww95<SW~~&XwBFG|Q4-IoFqtxYPNCp4k|YE|iZ8V1)(6IqyP!}AFhPLjurD>+
z^KvX=niU1}KO%Vw?b_=vKl6HfenkvtZ=~i>5W||`&4rdcY6hhXmrLFY>H_IvfI_JP
zTfC!`Q})8T@B`clJy0md=v|N&o*cIcStlG8UkeQ>qjidk@)n!etaX#Hh8FD}kH0Il
zI{<@!(28phtwU)uDZe1MHU)WbrWfjhl22^)%v0Km$<5G$$wp8t)OnXUB%_VdiRdOx
z)?`Y3og1-~wWKr^atv1B&}FeG6Lqy<q2Tj&dpXFZn`bU>>eRU(4TZWy0*TLBysZKx
z8t$oecu+`C5pRn_rPc{Pk@H}kO^c<x==|}8pShE@A)+(p;Mxq~)fL$#?i&D6f47uc
zCu`9mHk9JLuKXW^iv0mCD77V6%Fqw6@mSR96(wg~J)=;_mGQJtZftboXfrATDRQb%
z^5=v)QpCy^!ndYfl~d6cluww4+Kc8nZfq;HonshDRVFl<Fq?L}y5g=Bl0h4((y1v`
z6DhG~*3#6r1?`2>y}gY~4N0Z+q6Uy-JRjIZQvE@t3+;hR?cDt)P2#L(X=NQ2JF%xw
z7u%<;or`wF#oO!v5{Xf+nOP`Adw;84=?OuhIdxE^#_27T96D9CEw{(m-q*zE+7=NZ
z?~eI{OA2-O^C?V{_iQJc)N5-EuDbF;=qS`URt7H)!H1SyR~K%^l*xrcVvx=PvBYV8
z0EML`e{nM;iDud#_=<70#i2q=bPeUyt*Xj#^O9uP#%N80&aza}hH7K(EIbuk#dR<W
zT-s+Z&F^Sa@8BJU_KE;*FSO_+VMaanvWg_GQVN!QOt!K@yZ5YAh@ygc#@qE3>NC*+
zOI_Yws7k`UMjCfMCcS%@6M(tBJiF0gHnT<W;wjVx!f;hNbnz8|5RJoQp+W>!SL@zT
zPU%iFI^OPlAL<zSTNdioMfU>28PFEIvJTe$u3T?ksPmD<{t#M$ft&!M2fCxUt-eED
zL~QLSwNCc0O#(OER(s&pgq{UNP83QBaB0bPmC;)3s94Z5)K;3pbxf_ZSqjfMEYoko
zptfu{DHHRC;X9>91SQs5E5G1r-GsByF6t7sVo%-ad%01K+uVw+{gH8cpitxZFWToi
zXbzchQJGhdQul}$D3o0?XhIFernzi4#s$*4as=X_Bxy>uD{)@CimNsY?ZJjGYLk}6
zy6cE$vCaDX+Byq0&3Xh6CnbnNOeMFjvm5Ih149dtn@~KA){wIWU#w1>A77b`ld`jH
zl!Gk6UJY7LLN0Jk8)SQ_b<9p#58$FV+k$2dl^)%<Q>k^*uPXvoM*7}_p&W_iCaF|0
z!Mf6-?_P}q1WZ|9g^jj(^vBZ*#dRfbp*==PF!mx_7}!0rZ?N#wk%<AbJmJgms=7Jm
z_OvY?C=~q>uvpeDX|t@qoH~uKpYy-{r|_i1Tt~9>y+i>pK2drx>c?y#;CbS0XusYD
zZKkZ9cM-Qr4OSrWxTuIk?OG?!NSZF}IX0A^5F;U7zA}jM7YgV;M}>CZ%+s&5MKQHW
z@@VW`QYZ>#PP~n|znwUj#dYQ+gYRt>ig8Z(+H(Nhlq2Fx&gNhODb$GrK7X8Yt}SWS
z3-c4R4V$h+4O&9_Wq5*sZk2X(rthx}f-6Rfn=1w`80zOgFp!#DIa0jYtLp%9K^W>S
z6oV85h1LA+mS(y98a!?2QtQNC$%A??_Uc=M5>LGvCY9gWk3>8%X$Ng-9SR}w_znRM
zYWyh>i^&Pax7f0*)9T2dO{9W4!FwW!M@pqj<1QXoC^$igt?kbOB6;F2&t6a)g9dEp
zksYPhxhg?~3A4hAW}MF7q4(WfsEdPBalhTYLknzWgL1lbTjSKR5Q(M1@<O{-j6~Sv
z%zI5;<0&tmIg5<$g$kb~A-BFGj=&YaHt=McQqAI%SSTk_>q0%*{$%8kXPV^B-S)aV
zrhzt@z*+-^8hvxp)z@4#$6nvO9`0j1T&`@K1Pkq2bM!;=Q7611N=#w@@RmZUp<#zL
zInI<ZS`>R@wu|=QRpRzR4J|23p*?VV4nl$aE8mp!mH(Ig!?7b3>T;!|&XcZ05|NUL
zyt&S-A%N73+2CV71xkbX_Lc^o2qpBw&3uJAe@EHVB0Iw|4xrKewQtSuhnzQNdW9BW
zop4Ifs;;xQWgP^Ct>t8;)H*5f_^KbbZ;w@!{Z$^Rg|hJUvhKNdTQrTVn%YN}x+R6O
zmTO+C4;}mMIogo0lg))2kjaZ+o<h5vdPXpn{-68T-jOg65eaLHJVmz*uPd}T57mud
z+B++Gn9<hRSt#YY5pB!p*y;S0q4r&sa4h8Nt)4<L&dQ#+Y~|f`5l6Wb=TvmIN^eSg
z@JkcA7QpPhr#|jbV9hzi;=Uc&5gw;qBg8%S8hdY!N&8U2tn#JQ0kE#n;%gZ#vMn93
z_hnt=!9oFmey7$+Hzevn!X<luK8J?Kg)!2qlv*e0U#vdS)AoUU4iOHuOE7=b#pJul
zo`@p!gLQIZ$Oj8WM@|iEtByn;HUWhUTh~f{C?_P|WV?5{8dS%Hwb0VM=~_v27&&c-
z$#VN}4n|TX2by+esda2IxElA{u#eP1yM<3$)?<Y_fwY&)9qps}9q~X|i&!(|Z}#W`
z`rK7$cQk;L;Ft=oDmLS)=FL&RbCE1<e4LmH?Li&7mdw);Xm{ibvV+P9f!WvTf%tLm
z(=D`X4W(Ub3}>rz_@It}roC5cU8r^C+Ofgb<S;#{k)P$e0RU`osV!sGMB%!=Hv86i
z0KRMSs0#`-yc!0@V}Bh>WPntB=DKp~JjO_!C^U=c$~aF70Os!Eo1@0KF^(~X_XP6!
zgzMY-3I~)`TlsF}F>V|o3hj~x!_+uM*IL?!czqFm@g;wctm+!FjXCYmdxSRIx!FRC
zs1AuKX3ViW^NT_ZA(;M(MvDq?QFx(UqdWbES6^?N;`jOIoqti`x9_T8xFQyAbMwiA
zVx2K19;QcJLkcZHPaU4n8kmqRK^Mngd}-ll9IgW>4?86TZ>@{wnw*gA9k7L#5Quv*
zGW6pMLKJ2`TsM*2SjhY_vbE6S-wj*3BItJ8_U36BkKLfkUolU#_|teb!$PyA;9quB
zRJ}|&PY0G1Dq>sD>08n_*k(H`X;e0{LMbR16c2`ZR|6%*MnN4;E-%yt_$N;qu-z3f
zFq-Rf7}Uk?z_HO|0`}Bi;HyL)xIw*8=h&!|=O{1Y)yiK5S_B5~%~>YZb8$6hIocwP
zjY3c)8i;!Di@}rmh9wv%6k}D;3@mQz?;NoG`BCW?W%-vTMp^=AOJJsG57<m`2F4CF
z$mLnlK`Re)^zDPYd7(YHPFAvI2XmU?x#og;sdeFL=6XFl)GVMR1y`gq0j?7)w8VjM
z>zQ^qds~mxCgy@T=R%!REc$_T$M{jcf)qyr8M)20@H3Wh5=pfO?dX^llJcH8m!?jd
zW7Uo9wPVd90TGaY4`Kv${ym-$BhS+znQgV>@x8O2b$#JyoTkkknHugLP?V2_8T0B1
zg<p=DTFoH&2khg`1aWWxTTyP6T2~W5$iZ_sfYRMNzq{}|<~El*q5AAZGd|@nm+AqX
zfl#H;62#8Qm?Mr8yJNoI8*OA%mX@^X(yhC~0S&eftHR}Eg{P8a#ub3TUE3<AFuci5
z#uQ{kUjzdyuf&pdoF_E}<5v5ZW~zJ8)I!!dL#T_XOz6ZEZl`i8(R~g`5?!Gsb_%8e
ziH7n+5=U3B`us$;E5;dX2zy^5u;2JdzJgtac5TNgokPg`(!YH&M-xuHVoUZF>V(NU
zS%-Jor|Nj=1~SfJbEr@k+x`MBy!&*s=6-FAcoH6y%7~V+6bp~&E%ur095ppNpHdLO
zX^I7BAzSv@TBU&T1vtMA6sqpdfR;YRJ{RLBI(w)?S1c7OVwWPH%H8+#0ZE16^t1*`
z7Dq@@hWE35VHB{CbFoFb`vY;R(BebY(W@{D?V~ToXJ7El!Y^;lkVIV(VPC3~b5P@N
z0OY6(YS&Nr2$#Xymn+%fLiskB&(1=Pui#i?t<GGjfys|xUulLYIU-ijDO#y5FdsVC
z_$vX0oEG-g*nsVAHy3_3n@hVH6tS;mm%;T2sr)X4kt2l`A53@}N$18<2R&36n7m8B
z9?(7UK0ft+_5kI814?s+bby5Rjfx91W=|><>gNHlqACvCH}msVsXMAvrX{g?)YvCU
zIi&>qR<@7jHr!>CjVs$lZ5MCssna<(LgI+<&2N`eZvk)q&q)~n|9dKoNn5>RC{k|y
zYd{#~TfaohLOCJaF#YzOda}$S|7FP%AlQu~(Rb_Gj<FiV(MXGHPqh7=uGd4`S9lHY
zOVs>c-RrOz2MVS9x_q6(Dfj!$B5Z-3752zMk5FfDWE<>kZ5Qp3VX)o}r`QlBZuz%n
zgv!X}%qlH9d1w)xAu_gmVatAy-xjlz&g+G`*euLJr>$$KWf!6!7$f^(^w8JdKG13v
zN^$sTW<mrP!G*^^s$W7YHLp<0lT7ixemv11<F_9-Jt)Q^$gdP;drEDQe*owym4)Uy
zyu*HyV{{=}?>NKknMh<(E3|9DyD-6s8T@H8rLE$rIT+*)rM=FA0~FeWQCZwi;*hyR
zJ@&JxL0ZV$E=$&6*CE2pozH)s1DD{V;pw8UKm%G_2<nEk{vuuJf{kV=ILO^oBL$+!
zP2qqmJnN_+AxEOlhLbDpmzBI_p5j79oCD*y&FJ5AF+7Z@Q1Yb{SY)aedo?y8>93mh
zn#xnuE^dlL$ysO*!cL~*0q!9EbsZ0U<8GD<N3BAgo_U%Z_ST+hzp1cmm)9`y*oB%D
z4<=o1|B+otqnbe=g*x9QqJ4_k(0MKjsAE%c&t7l84ImEmp#II@((~HyYWK6J%qkQd
zVp5EA@hdBFz<!@~8z{L(0;)xwMty1iV}Hn188dqZQlYYqx$feu_dtolv(mquar!gu
zKVwX|X74Th_D^g4TX+<w4f|u(N$6DgzSJPKj=4Ffb9{J%{i#`qG0qU~sTG({yZTlj
zkK)&zYxr}b8@7a#E1(8<@>T8sjfEDSfI`+Ggob(cm(ihVV@TvK)Wyc1A*}psb|ywv
znk~7Rc~-GPOW+_Jq%=~?N4DC3HNF{L{lx7ID9*i58OQ>&-$Ft6w}8zv&pr3t!f&BK
zIF9{yv%y6og56f;qt%5LXL=;h9kstFP{NoUjtFw>aTGxxN_!<;obvyuXmAa-H43Nm
zJ*KJ{ezuLY_m0@%za{0AY$%|**>WHEiQB=)+lG%6%G#_m`v_sV+j3v(zV6<VKZgje
zyZdtga2R0=B|l6^JT6cFdn^ya7H6Q4LE0k@^+HRG_bQor?zuen=9oBG5Tv>2z0e+*
z%ti=KV~^b{>*PBhhKQquI)Qy?kdJGv|7rQDn%M`BaAZiopsG+Uq&@GV3+>+VbACs6
z;is6on|tg&&9F6KU2=<!jwQdN)RtI=(EUf+u>0n41Yb8lLk_6ZclyXVb-y|T_#-(|
zbH7NTE)ElBB^oIXIW~;L!R{Zxzh?TJ*@fSE@I=(uHhH!{^u=GmG};4V=4k8Chjn!q
z%KG%v&N;^(7;Ex^aK9`R>()HEGU%W^C?^~qbMENCrsY5@v;-EyVC}L8=W{VhT{7yb
zD|AD(MY4gkmI|cwonsGKUXOjbwfLq4m>XQgRT^@3AdDYRX}q(hOt**TCsz(185k&(
zVz-kCVrXDcsk!W76`1lqk0})DlDwnrXYd<b4$`e&VTF76a^H2GaOJ&)LOvY&ei>B7
zmWW+AHoV;)5u<4>94!2dojf??cb`0R6gL2097a06Ru@_lu#vSMN@S0UxeP2OHC|Vt
z<mY^x@Zm=tT+gEuNCF==C+2lDj!VEJS2^hcHWr@JZfsZ3-D6@ti;0ftEfmzy-$H<T
zY>fSe8_p{Htlt9~l~E<Bf;}!C>Viq(r^8|_%N;b&anT-M7oc_1!tV5Oqb}j67;LFj
zaV`amoylR^Psnyzd=zWomNxKP&OxO8EVL(Mr0aVn;;>}S{yi~Uq*D?@bKFgEIx1)p
zhAbj0?~`KgBde~OGN({*u!pu;NlUiaKWClvXBr=8RiREW8d>$?YwgL|E}hOUzKlat
zq3(g+-bQZXh`=4Nr^H8x+WHGW^**_Y2JE!E^dA(BI-e65?#G+<Y0T-V4YZWcBH#?$
zCXFD^9xb#dG!*OTVMvU57f0dg*%A`BRwdcwLW{3}GAT*w+HzN&pgk={J9)yKDTUu6
zY||&&)3ZOx>Dr4r*<8>rv;;lI&#ceRs7QYq4(e@#g>oPg>KyruL>N3|(avnfA@Kk&
z52)K{?^{Z37cR-CQFdKPVWJVT`<#_Mk3At=X;f&@dUxO0va_3E)3qaoxFLQQT4HzN
zz@t<Q&sd)FF85V)1Pm4`B+qmFM-mof?q>$}^uUn~77AhyIJH~4fnmVT32KR!*PX~H
zY7~9#OYGbR(WJn1$-7V&V0+cfNwe%(^|--Xw#CIo#3;#bNVQ{VaGQ=TcHZ*js_{7J
z?8AMnP!nC9xDD@{^BaN?(&<_RUdJYOP?Nc;@RYBMlQ<`O+Y{%;MQ;oJ!lpvWN3hY;
zj%!z~vkMv=4tp_KrORmDX8)etgfDDHkJ62=<F#3+3oQxQGOq{k(taDCL&=M7IplLo
z7h1#<;IIw{yudDMzAK9pyz}tdLY=R5Z@XRG?Ct5Z3I#212cgD<TQ9Lo>UXDIHEH@a
zlSw9Bs0!4?D7BHvANBBBMJiufae-Ufi9Ku;Dq_?Vd-}*S*5dllXIK20gpGH8t5DWJ
zF1+9(drpp1m@BP4TK7VoXaH8~4ts8eJ*TjODjILyLW@H?)T&_)dS0D1r=2t_ob{!t
zLpA7G-m#}VzrKsdaNH=gy-=q?T{8YsdqJYbD1jEi5bNm4_}GBlU|e`g8~%z}_QHUD
z$RjT(SsYn)<q)a&Y(h?9tPM&N-B)PQC(pY6Y@683Q@2fN*M`gf_Ckx(Acd?{$x*mf
zgGVOSIb*|wDj=wCF4V;uv^u)_I?#J}+vK|9IQJI{3A@M_qvYIE@;k1ejuV7Qj|ODb
zcI_-Tr|YnDZM{+t-(i<6PX;(8NBJED>n2()wX<9#NwYP4r0}eebz~J+c#_`c@;U`S
z$krhoA1O_p>$%;r*QVCZ%UvZG7D^77@M8;>C(ww$!<%ec-Tb=q&pnW+@i!P2WKwdD
z{7V{?vMU1dCd`~XweUNJ2!{&;e&zC05f}o8+tIxcFApLk3$+q5CiVeho1U|vS1`ph
zXq2oQc`GLGH^#p<qxL*e;%cGj*xuiFqg_?I0NKQj;G$Hhi?y6Yj(6Nk&~{!OuW*OM
zg2L}S)*L_Kk%X&qVVv14vSOa4a@bd>%csQ>j4sRudr>~EBm&xquxX{%X}?Q$fwiz%
zb-xcT=K6TXPeoG42%}cV?$l?q>$bm+oFF|*3gy5$x#}@UB~Y4^b;>>{E>(O>){U&{
zfE9(4yt&nB*W_2mWbvd248et#fDhTKa60Md?76nV3`O`$Ya?K8u8#19b{}%+$lUdb
z87pbgbrolbY#nl4J~io$B!6s0r74WM367^<;BGI@9@!i+Wrc?zn|NQLCGcowtCKqe
zU$XqvLml&3%vOKj0O_Up+dWWdaNwkN+7rCIH0#6)Ai6k%MO_?Va1ev`vdT`8bFh11
zC>INr@kkGe2kiPf06f_T2MR@@?5l=I(rYiT8v=e#DTwyU|A&*$9pi>^{>=7f&$|q$
z$ZNm<$X-#AfoomBg@HnadY|O?QQeTk7hk38%<ljkwNBiLVv?1k?3E4hwmF|DL=IIU
zExutT$;n==+nZgINwBvyU)bg>3pe)`>inI%2INR_>23tp-n1*3S@23@B99C0+K*?v
z{AFiY$MRE;<#*%l`agdZ9uNh8`72&+ojI0BMJ@^FP@ygspES7WjaqfMC>!J!plGAf
zx_y#=VWoH4+~zr=580($@KLDKqRJ%8c9>U}7{}guh0>U#Fml9nWu0q))CV+`x}caY
zM0mIiEwYBlH+GkG*LOJIM6fY&7wW>cF=^T?PceRDo|6}hFZ>j!{95t(wQsj}`$h(#
zkinRzj$zt^@rQs`f^0qcl7?43UYbJ}_|<k1uZ-De#ny<Q=1kdd3jz#NZt9Q_Ny)c$
zxN(F>RFJ~nIx(iCO^hj|P!~+?jJDn(e60KGGw|z@ANHe-q)tG;jPxC7tQJ<FpCE5S
zp;V1ND0#`|@T%YkF6%A)iVZlkk8C{swPV+2Gd-LMYMs=u&Yq1NvVnjm7oxQle*21g
zf=rN|OGIoiK1i+za>RC^{-Po={g~9`evpCB&|l7Fq=unbsk3f*Md4?qE+<#;D1pzm
zsHqc+=x1i3E{Hj}sno{Jtx~o)KbG2<q)75q4cww#J7EfW?ry4`xMnZ>)Wdh4mVo(n
zZSx51w<XQaCyk|}A{tK@!fEjViyK+%=y`|TTv5zpnl?ZVLDJC_ou#&mb|?%}kxfL!
zanx&wTkO@%*OG%_orV6}3oTm9g(TO->dy&sOMXh`Hrcx+T8G1nCk`vL2N@q2BFB>C
z9t@Mxh27c!RUSR=&tGa?)qRLtTACjSi7fQw8l}|G+_YM1dvJn*(N#4~h=?t#kn92X
zdIw5VZd)kauoB}Sq`z}P)hAGwN12BsaUe?VCT~wP2+-F!%9aNjPGy@Gerr*uPP^7t
zG!v7qIZTt(IReVMBi{J1+uoJO!o9F<W%I#i%kpLu*AyQtJf+=oVigOv*Itv0<OG!E
zU5z_PI49HMU(;RY5(P**#CPPJ<Q|;%+U2Jde&AOJX)Vaq#&zNI3iV>$PH@rEi50M}
z{{Qy6>`lpwB%Vu2CKHI9Q)IuP-4T#)yHhN#IQIG+TZ)M6XIa<|4{tBDNEX0}?+y80
z=ze7@+EZ$s=A>K7bno69^SN*a6pwPCP#3!oCSLAfvUtVZ-W2C*U%N~Z6q0?Bi+BVD
zhJNzRql*NM1glVwQs-dP`756;oJ?}PT_tUAX~3#IsM9sBFIJe5&M?}8aQBe19bvyq
zsrTMmJN1Y}5La5dQPotT-G^8`TfhyvR?^M4<q)M%mevqMcA-vO!^o<>1-<t6z&{e}
zU0V1ZG?x2LJQLV$wW7Dx>f~<NLPZj+W6o3Rj&_&b?e^$Yj8R*m5cHH4mRlay5qn4N
ztG$g=w7;`Z5yKdF>4o;r`bH=$F+!{_wN9Qm5@onO$lev$%e5TyQ4vy#@AyUSv7z3b
zuu3R_<+>_rFWZjO5x4l-u2A1hJj2q7J%aB;&G()FNZY`Ifx_?XUkYI7U+ulIN&ANu
z6@GfND2afJlxXkEI!~7jijx_uP#1KV%fI-~-k;AXszItB^j3vB$D1U3Fd?F*N^_?3
z^RRuOZtmd9T<5e7g|Ku^<Q(No*s=;wg<Z3=dr@D12dv8n17Jf$^`m&E6x=(}K^U1J
zzxJWJ3^Tgt7YcPoP-`V=-fkbxx-iNL^LG^L5)TxqCHYt<iCFWG)D~t%#sp?4-5L7l
zLQ9N|{AwKtQ=RtFK#9q?Y=p#^6(sCX4Nt7f#lR+3`0d}U#t5mz#ZMl$J1XBd4+BYO
zq39Hjybi<bZFL3ybO3+=MOm$ppwrew3C&mwl%$mFtqEyax7ynLtavsl0kqTsY+AIe
z_-t%B@xEcMqR1x&BGPYSjPy{pk>iCH@0YvV7_{){kgcy9W*z}_6oQ4Bt?NHF)Y4>}
zpbI6(jIozrZD8eP=i81-Xk&hMJ<6i=evK2n(Bf~Bb6wm&WKP~p)go}n?yPUJO4=lA
zAh5oQg_gjRiEYI9d4T_>x_`NxY=FB%3l)LYmv;|N9cs7DF(G2Pfii^{G$(rX;0ncg
zIE`${PmeC9k{*a_t}Q{|g8vC;!>#!NQX+K~3Jy{_mDN&*h;7SGVe=r6Iv}fko)(>~
z7hiauZLj#=wzyR&h@~D`HTfFbkv+0}V%q3;U!gAXT&x*#na*u2`YUZ`ojvYTeck<q
za!|7w9aq_|eC;lW0GnV(p+zLN-Uf#|XVrAzWzg2y0d3SFN^Ri7OtHFqYTsNTQkudA
z<1VqiF^bkr9k`5RiKiS;VJ(xlvwiW5qD2V}Yn3d^<yT8(`(s~TE+Mj4xkC#zf)<xZ
z+kx7lOv>;F+z(o)i>XfLI;4X+Re1nQk-<)9HFgpKTw7d`0X+SxINCSbq3jgiixbti
zS)tDLGTrlAb~yS(iKBm4js9C`@%i1!k>VLFhBw=hW{qT23i_$=yETOt^*N;67e^Bz
zBk@<r$U2b^vYtDKcbD3(_hlImM};OKx*dy2PrCXl!jdQmK-@%Y$KxAV;`>Ve9$9tf
zQ|)8<z3OAI4<g|@YWn7Pp*>_(iDwJ#<59%Yj)J|XP>hq9Xg|U}25dMd9^w4jHCs!q
z696Tm9fby+sOtsk-lG$3_(-XB;<rNCxx+^42PRLLG2!wlvkMi0aaZbQi<3<^5v_(F
zVX1ZD9_u``{YyS4UYXUGrKHrlKz^q;b}COVL3{_pI_L+wSW{0a%Bm?oi>I`@x@uZC
zbFgj$_K5)X#0j$sKLhV3O`2?<tmB<X_*tP8O&45nv3;tdB;2ZExCsVMvQ}HPx#biq
zLunjapRVWud2}>(dw0`bXm>&&Cqj`M-EJ6HMSdoD(WgAC@Jk@1$6avNpRLCoHkdw`
zxg~{~IKVg`6#jpNt>Stz`&@!i?&#}jQ7{;FN%lwcB?fZ4NN%6cb}9dKBI(R=J-35*
zPodqnisu9Hx^4D_X1Sb$^yQ*Xo209o%Za|&Y^6%Ik1#@n+d_-?MoKe3L<oPeFU5R0
z&KI=}6iR`zbGk|N%JrCEPEsce)F79j7AgY87Q;>2R~m@!mCV70i^L%t3oSl#Ip1q5
za>%hsbxDRhhuBw}QHlu__Dh+uk4;RAj}lDxQ%4q)uT_S6)PA+3P#L>#_S6Z}r`y+K
zWCN|6h>s|gVy?kw%aCE;XnK}sp0wZlO0BCmODLb%H}g3`O&1g)VixK|EpC1FE%vRt
zs7_AdUQ*HPOu_J$GTI&%GC36?toUT$@9f(t_*_@`72Fg#A(!BA$2B!+5zzTmJ<i;D
z4zy6FTJ2cQ1ipNz@N6#U7W(#Z9Q#hQAQH|L;p9O=JWY!~tyoHl=iIz(-%Vs!cKaQL
zQamxd*|kN`Q2So&$%)rYn?AWvI9q!|%&zvFT+n2qaJ`rBB@(;~uG9!>N9_9vDoPC7
z*O$MN1hk{K7usXX&%Sz+{adpc#`?H6E<lBLA77{zV$1)ajv3wo5u1;$E!5?CFS$&^
zk#0%?<PY;{omr@AAi%o{bx9PYI1EIUrfj|asICs_LHki>%`4Oh<L2Q@D8<E(^VLvA
zxF3$sYiFTF^Xl-6;C_-{1@Q$WrFARR<ySM(Pg+{{Q2uGQi0cl`#vOnfuDs9^*uupZ
z3%cR1?PqldAmbt!P;>xk%6|O=zHL8GqOJ8%yxDPeDq#h{=ps=^%)(QmAJZ<IHpzYw
zn10~}=a>9Bn_!z?)_uPafBix!zvNj5K*v4y?{zIFPnlFG<XS0Ts{0_?uWH?}lF&++
z-B_q|a;xZ;{r2mo+d~|D7}ANG$vV=D;q{KuAC<ZNrUAT=FN_fZBU%DRIEfwLe@E;;
zny*H7Nql_@b((H>XRFPAn^>o8PRb3-WddM9WZ*|wEwuYgLa8A_yu*H1`NoBP+)3J5
zC<ob`$rurO?DzQ+o7Vk+wxj%{E-2^pX;)6MKg1Iw<b3QWrvYr+K*>7ygp(hltSkSS
zJtcioz)q=kvhbmf=LFbde~eYTbo{u&PaiQEE#Rd%t>JY4RAm7IEW7=>%0iW+ps+>q
zeO$*(aFqSIe$(?g2y>1X>ip>-A7T}x0TLtlmwftmjF-2u%Wp$_L!ERxPkFBWH9-9$
z(PVySA2L@+?2}y_gOvSO_7P)ooll`IfGW3Lt@_{UAwb|qzw!YqY=O#vss~>Echk>t
z|AL}_s1xNL!d>I<*&GK|7?f)rC9sma;?V;Cqf+=WiZ`V_g7Ff~FSLi$)PtoM?X9iw
z?GDpL;4g(3cTYDKqs{@l+lo92F%d`&s2wk$ig==JD6|V$b2{wqE3)6Y?g3pW2bFN|
z7U5=W>qB;r>=P~>1|;<V(gKSY<7(SNdl1$O&%4;}xgrokZXPn{!%-FLT(MapF1?&@
zA=ja~(Vu{Xd)3dRyRlqlL_-uWv>VvcF6duIFgalOr<lmBp~5f0ysNdy?wxbcj6H@4
zZHXxF!9q*oX{d^i3~#dgG(%P7tzYz1sB>0AL(PMQSt??51KxUv?Y{X1KNX<{Qcn7@
zRBDSfbdqR11<*+pg?7IfYR}xlFVTXLRb01l|K{sbVGF89ORcNf7Kv-L2dqfW1L>-F
zexVq6J?0XyhzNLKog!!>F6`*Yo>J?4-!drd#P*<gj)*7{zL0?&nXXV5y4ki)IRtUH
zUfc=s;EE)3y0t9|W&e62Fd9YC1mX(qAqlt;gkpzFLz?+g3hl9X@C_Je56!W%Q0#4h
zjf#l6KtL#sl!UsJ%)RyOVfEb^eVyHfk}stctLS$b!y0>dfcT}Y!mmKa>(4&h9?^V8
zisP_x)Vg}17+I0+k@=h`uIHsk!8}?g8D5&TjrOP<qO3TQ_{EgOSwKs$pl;x=dCugz
z;4+*&I;L>V!XBk>q2yB#mPkFmA3E@u+$5pKklkSgm?tJT9FzeSOg>Nfn;V%tewuEE
zer)q?_aoG1VX398)VY3VsqG?lo9NAL$*`O5M8?uRuFeW2s=VRF>TABc3oQZ6YY7~F
zd;?5*c1rTtRjBhdOC=;!RxwJ-hWvzV@y(IPrbmhtTAUgL1qgjjWDAZ-%kgcyJ#j^~
zM-Ji55n*wKO3fY<0yk=LU5?@~o>XT)hT>s0i8(;=Xcyv<$4WOf{d1In&tBZ(DWy!^
z!i&M4Tmj+Urjf#fr5xT=nwAk|h08G`F|V<w)CCEhGlY9bp)U5Q`!om{<XB6ng`E}z
zpuKHqLE(4oh&jYn+f#EI@+Clp3sy@)53`RFp3xpCgYSaO0Q|Mn^Bs{p&rZ#fN|l0&
z0>s)_Xm7?5%3+9=ug=p}B<s?-SQ)=yH>nY6>XBFN)AK#i6%XLjVWsy9Ej|h)EeEiI
ztj`GgI|=`6ehOa{XE|tRR$_PdxrN_861QxEJEN|XXVrO;O$eVKF++tqL5BO-$r5E}
z=MyLfIPsmIEYu~L*36yl5aK$^OCBP7M!qA)Pi)El;WdRiofBu>`aFAP9S&fYFN3^P
zrFU3eXz>Z^GY1$bd*9B<Iyslg-yGhp)H+d&2+mt)=T>6LB`@<^orQAtC6^s7(iVGG
zGq8hDnvEimg}MZe7&w>|_PqA$Av-Ty+$!L}k#t86#X^hsEx|Q3Vn^)!s6!jl-Jz5)
z<b5jCX<08*2<N!?je%`f;di2j*dx)I!#d&6-d~VCJNLn;j}LDv)H%2XDV=TdUzK7*
zJgfqfFU;OVZzA)f41h3fg_eYYU$3sQ+xYtJF_(<6W67gI%PdBFR~_D{Ht+xYFS{sq
z`gEwTusw?%1ZXJS_Tn7nMmRD#p2NBn>IAtqoT%)QW)_Sy@TE}a5Db6TF^Uwsv{KCr
z#$Qq>H~=JrLdr4cuC-_9t07DPXt!P;#i555;poes`7(P>rBrOZ!-bM}vjGAFBYScM
z*zw$473n4wZ_Ku0YzhFE+I3B#{USP}Mc#@vW>vpyL|G){7fY@v3VB{M>ub-8`Qzj{
zSop1d%~eGJ<N0;sF4>JyHk8^TIxc%WGYWLI7vwvB0=OkIGshKL0zObQF|3<FkQZj1
z4p=#z;U8A0%L5j7T;S+jJv@Rea-ysgDkn^Mi>I5SLPX>QR#hxM6Kiw3E_Vb2s0h;E
z-V(n*DF+x?eM3d&=We0J2e@wd4wo)b#o6T11#|+5iJ4)N*1Oaej|<DQHrzv|#CRqv
z1i7bBij#H1jOlh+B^+(!#ls#&IU28<@DjT`CmwQQxa?@1#CS<QpgoR}8Et5nr`FNJ
ztLws}J?JQ?bI=J7!iz)fm25wvNn34Nz9auD&0{A5WUUj5Dx?%1)vn0rL;<z`;Y~`d
z10!8ktea}Cj1dj?-9)T)$<oNG*+fU#^a^&^&hEK|VgeWR<<qEnZAOI#>Ae@VdJ5$T
zUm_(5|LIy}?b)j;N^#W5(I7II$Xrc}cUq|$DpI>TpBBA}97Kskg_eX)Bz$#S4rs_O
zfqUO&Gh<7RN7&`3cTDbfjt$S?ZZ8_GLljWmx<Xw@WhYh+?}MMVS;_Lq8%B{J+ye-z
zD}nwPLvrCMNqTt9^WN@Wn;koFWYu$DQ24D?CL4*oN9I%nxoqO>LLp!%VFMWA&1Kib
z<GM<cya1G9iolm0sl4kvw~a6gBn2$k*G6}Ib<bd-5Ohu3iigF+I)iT4H7(*zwed4N
zr_ka<5EYcTru<qj&N{s-4<E7k)TJ+Ic<G|9e@XN0wK!Q;z|HOM@i4`gHk}-Ca|m~=
z6zbAV2K@wEY&7K!#19TglFt@<*@{!AYx@V29;VQcN(AzhEO=IY63@nZ%_3|OcLck>
zf)uTW7%Nn`g}T^h2rw7f%M%vzu#UYSN%(h^+9C#7>=<@^J_BcNuZZ=;Xx3Kv?GTd?
zBV2%Ovm0XlNr<-)vs-_m>?<K^BspWButs`Bd*zDsc2+<q{$`!f<<3H__o-+YUDbPJ
z_?Wd-Mz5d52XnE@7HR@raI%hjTQ^&K#ZTvi#BGRQX%gtO|Hb_Al#fSj=?cb!G6nD+
zV#jyHu0}(Ji)(+Os85Wn8fuqXVo(WmI_p5@qEGCV^Aa_&uP62O&h4Hz6j4>ys<=IK
z(u6|62OFknsQ&Oan_Gu5Mq%RWic(w1`oLMDWsBgV<2En*!K4V;+W-TzrBIi6v>gQX
zOw&48<%{h8!Levv71Yy+ogrI=-1`O!wPFq?atV!f=Xm1OBVf=~H*oHSmOz==U0lw7
zW3vmnc!ckT3Q!CMT=DSULVL^=BB5(@??A4dUpI^v8dVglOqC8Uv{LSA+7smIl1tAk
z{PgvS!%>rnbA6vJsMGhsNw!;8sEg^3QMYK9DyRonbbtW2YrXlV$VJ`l)Wx(2HyX0O
zx_P+5!Nc&Vh|_Bh(%HiL>;PfPg;L_4Yog&|xTG?{5+YzI>R39&3+Gr0?LJQGk`jr3
ziI_=Dm@Fo&fc0n3;@F{0qCkZ@CwrqB);-Kh>gsSQ7|3>EdYWGNX2Vhx+5_KN9%9?s
zd2s9y_zgDL%wsR(j5vU-(>X_^4lbY&x1su+PHOZIx)Ff7+(o)pHskex_?@;WpO)!D
zq?yDGrPd+hz@I4y_n<9q=H<z%LH4_NX%yPE0H;0sw9}qxH`V1qnXSu&3Kc<v7m^XM
zPidHJNi$5PXZL+mYMl;Nu?yV0W;e%kI4yF2RVP|e{MD&TMh1_LSLa$ul3=tcU5(oL
zeXR;lHT;O=1lhEz?<Ce3yQK~pb+ztxE>wsszVL!e?AAI|`EM{R4!Bb5f=>FmYD+8L
zu@>!xl7BJO6cMAKFG>S~&BvByuhPg$>O(v&wN73x0*c~`OC8w++!saZsBnb74axI%
zb~XQR%Y)J(agkS-(|Vv#<DZv4JG$5YkBC+w?}|zr7q`fhscgN}IFbYn69Owk@EyY&
zZDmZ%IgKNQl7ByJE&kVDQ>T6zrlgW3KPUItYx8+mX8NSIms;m~7e~8|_PQ}BrtKn)
zoYbHa6_0UZ((d1tGrfzS0P*!X=E&43W-~N<8w)LAC}}Ned2S4QL(ose)WwD0K@^BC
zl8X|2iM_G1iBj||C=_%3BH16ZA`k9oIMPY)wl`IlacIyJOH&T8uxC9E8?1VBeWs(Y
zf8ju@Q1)lUOUppNn||I>K{u5<YYV0337zDdU@P7l=-!P_VyjT_Vab(cKai2HtL7~^
zX6Nt$ds}=GtT+mG>@1Z1c@bD~U~+(VBM-k}1MKZ}`?z%Dycz4}cNzh8ne&dJZs)1s
zR%W!_h@pIm-BtmQ1y;0iD@!7*-CJl8rRF2thG(}ofL5aDaP6cnr!bm;v4%*)8reIV
zDePqmn+oMvN?pFOxUFZty|cm<4?@uw%E5SXJj+uxNXqCHH!Lv6ZBXrZRe&JkNsqn*
zDw*sVD|evKuBl!JIc-hnxU+ZHk8%9<zG|pY);T<~YVKTnPcuCc+t4*qm&8`5Z5#yg
z>Gx(!*t*<H*bj%}rMAexyB|O*`z(w*oX7iCxL5h}PCNJ9XN~=@(BN?9qZp#!p96`T
z-{pX_won&jax#bF2l4@V2r61>E$k(ANtPi1%nm{OgB2UE=xfIlv%Zw>8?v;0C?_%Y
z?9@56xzLiEDp-CzXY9i{N(C)!FO)PQZWbHpvX2CXz4Vd`O8&fL{Q36L+6e-SZ16-m
z6zUwJWAGNZiZZRr#dP>OVO1Wr$W1_~ZJ}L2>S>XOt<&zPD|v0JmvkeAvRW~Xh+AA;
zC+3?o)+m(Ow8V3ZaX_&(`JCt^SLHysBR|r*GPWhtZ6s}5J31@*R^nWBv{07>W;DTz
z8_$^=sA=ososm@p3L$<fpGl$4-w|ODBP?C*m;=VvXFI;cBg)F9Jbn^Mlb#mZV;^Ty
z4%<*w-b7);sUruHd5TK^io8QTH>$qnx>Ct;hiqfcMuLbg3)+iNUhA~&<<=;*-klYq
z5%VKPB1r%(G6_Cwscot_I2!gwm9V6nIPR8h&OSn#5~YSs6s;<>=!Cvz_Eg)_fF<N<
z?$r}ANcn}yYDJ-4J@%1@q|3HukH7`!Ddh`sbP_2FEkfYQibm>g_YC#6t!|<T?fr#9
zZ2ha2*!Bd1bhvvNk}i#D>YUN$pn}8NQGvjzFZb<mP=NlI+M<jDqFP7T`fy-q*>+Zh
zn=}=dGD_~k`(E1LJG{wuH6W3sSWVUZF0_Q*mMf8T_wnv#jH^YSppIKgtrJZY{gV-X
zPk{2`@#hzQMi(A?vB~zvN?kFb@XNP%NVM+vWc#wN?iO8XL7jk&WPisZ+h6euVFC){
z0WpQj_{Kn7N_rqp-bupqR%f9Y+l^@Nd3LaV0tzDi3Eupr)+uwI8-Iq^*`a)nt>H2v
z!h5N8AziqI(+<ZPUUJ?g=M{eEX}i^x6ga=yk*Je@ANHPJ+`yL<>S9DhJM`Moifs@8
z`JXFWPHYvRd2)HF?c(3uL7VomNA$6Pc0X6I@H6%^c<p{Gh}&T-b1i`#ZzjLjQ3OGu
zP#0Q6s2F7HV-=WFh6qY-70TKc*IhZ)K3<32w@B`q&9|Csn)6$(R<q$s8_)-~_2aKd
zg+qs|Q*bMu*3eF5C&3~Fs~C}zy4q}4p*;x9z>u>3blOOEBA#BrIEWZXU4)Me5DP!(
zMuP!68S66f>Z_+0e#aEa-m-u@kz-2p`g#gKeN3SV3I~l-^=Wc*&+9J~{SggrI~#5_
zYK~9jmJw{p^N7}JJ!6q^lNMp?v;i{P+9zvIGuwuG`U*uC1C=;ef$gAH&>l8xv-I|<
z`qFfQVR{S2Jght*1okPle)j2@#>DRS!p|I|Uo`_)w#h!z5Q-2q<yIGzZnQ=DOyh_b
zLhZB9R_={?DVip@5Yb4T5Q-BeB}(C`IOnJJ>MR++vGH?B5z##Llq`&_8h6PB_WAm7
z+X9%@E>6ZmRZ0e<ER{2R_J!z%WLMn3MnTkZHolhj#mckICZ>Z-rmaFvV1|tSu;lio
zM5SPcrS8?eP&%tzATP9Q)6Ks6npu;k*q7@HI-e?mOFnkEkd^=0qLVMqpRdG#xcBs;
z!f*A)X|)TX>Z=uv97=A2FQaw3FhgV5WnatZ!XcY<HDC;N2@Mwvltf9}n-lx%`E;Bt
zl3k7#>V#ipDaEY#jb_=UpLDm{LS3=}?k|8g6ODljZLS%yZ)Up&DK(OONrc83Tvuq1
z#nsKd?m}VT$`M9#lB%YbXB90$$Yyj*>alMJ>5yA~Zwr}`vRWa#Bzv-d%@2oLiPIw>
zu9PL(>e~wKv5TAZ2fXy1N{#TgC9!3p?6X_%!o+%Qf1S**4enoU-wlZ1{npP-jS7D0
zEYvwSFJ!0}|H%-maCad!Xs+2yjcVT;T?`~^^+}8fY|EiSOH!{G^0DEy_Wi6A&wPL$
z#70mTBgC&pES2uL3(BpthU>FO5s$vbK8FosOKDp0(h*RUxz+o$ALMtq)s8;(cVnSW
zh%nL9qy4b1kYi7^(<##wEk5j+3Ijn`{-e5ekPu17GFYhahY~Gew)WVMvrd{+Jcp$H
zE!63oBdboo{ucX5MHIlS6ywNERH+|W%&u0)e3UbUTKzQp(XzW=FNQGc5=Ksd(1S>0
z*Caf!pVfDqA=N4!F4X0Q+c&(yeqNt&?Y3Kj3UzvdNx^s8FX}f=u5g(7&Eya{Y!}M7
zUTBXqf#?I$y8W_FZN`*2r76tno@f7_pIYx7ZD7F;7V3P@hUr7dq@^Y+pZzM44Jzcv
zw0!nL>!AC@xy*~dipoQISa>!B(HX?|w&eK#>zKkEuG0dXSsGc@KExI2zljk_$4vIg
z*cYrGaURg$|H#>n#c?sW<Y+=ed7#klD~vwUUHQeC{5Ic#G)dnmKU>Fj?Fg{8)b=_R
z-N6`k+VBqkUEQ0P!zV0iD@}QB8*z98_WL-W#uLS#KQB4|68l31^whRRr72v0%k}o3
z6)mT9^x&A^DpbUv-76HW&G13{W3zo^N>~q*o56sEmYgCh1cZgee+qb_?dxrI6iWVt
zZ$xKRSKFVnu4<o-5t2_`qNgZwnA>Ed(MBeC(f-o3`(4M*oR3jBl-gba8>vS}!)pGT
zr>q>D1ky-5q?^**(JBcEPsITjYVv;juV#XqIg*@DFVyLb1P^-cZ`oe`-?4p_rVc%(
z-oeA({<{JNE719clB05RW$X~03b+8--<$pv7$|vSb*XjIQ%aP>erErOb-sGeoWf7<
zOx6IoRyJuZ%~mF^lcfUL8L~`Gu9wu+Y;rz%x0UsM?VI4Z;`FA)*F-{xU`7%b`ghKg
z2-f9x_v}sYNK1laN2zs=IsWu9yGQ*3oD3q&Nx6%K8$2x`jJ<f8-LrY>h<kuW{vMHc
zT7m>ln{k=lYo!Ywooz!na$)VF9F2nm3+<n>4=3R|(V3BBS)s*e9=;Q9@w0npop?#f
zk8*mib&*0ZM7jGU93&=2l0yjtJH_WipVk!Gy>AI8y0%~q7nt66r4z<$u9`My%H%>R
z&RP5<6_hG3NxNTNp2<)bf>~OH8VOb1y&XLslXU+&w}FLSg@S)2cr8JDjXj{kZ}Me@
z-vNjV#+`2uYzj~U3i)OjU==j+pb9CDnwlae%yqQ|t%&QxVkhaQvIo~rr(mC_5IoCN
zf{r@c#2P&$;iO|!3c1y2sn8Bf;qI0y+I<|t7g^YLK`kD-(vJiz9|R<J6pDW0pg!Tm
zc~~@Zwzc)boKp1ng^v*$vELqE=Y|-m6!~&;qAs_W<`xz+PeE&sXmCs%pXgN>Om-An
zToXaQ{zt}F`~6~|zCuw$8d)_17ZH0@_8`Viw8LE^xmRFcp(UOh>LttKqw_hje2Oa3
zu@J(0(}Mm^_T<B`LtrHiV~>fs;9*NBhg{&<vDAqob&^x62i-@%JvJvSi&9mTQy0QG
zkCE%b9#=6-X^k<Y4isvFygE%1UD5WBZH<?=J$_|~NhyBtl(@6-3za5DOLnC2lu)d{
z8&j-3p+Y&i`N*0e;h7daHgWpZldiNU*0C)b>Ou4_l!L?)pw1nb1NNkb$|$Q7^u^69
zTV1)1-L=pz8ga=u(E6X7^^*}SQjowz4(}<n_yl9;hg(&+>T-slTzzr{8(e@A6miNF
z>Yhl?Bdh)wm-yLJ>hL@~Oni}!CxsS(KEvw^?cps2b=DE+cg}Ih1gZSr@xG_!u!8J5
zvJ~oKBhQ{N+n(Aypnw!nv~^{)j{8qJkCiy&^jttO=2b-%uox5FR(LA6*RGMZ_OvLm
zxyadBDEaOZCo9Joo%~Oap$slj9t8^ifNv#+vI%bdjAnScbKEmAm0G9x({%W?Gh-3Q
z@lWA*GIH{7tZ|u0PIgw)B3+JPBKs-=YoR3u=kd7M@EF|L6&W0l#b97?#6Nj$p~atu
zDRD`Q%vbh|tgDs<a<7(JCuze4ZN~7-TyK&74T1Np_hx4x3CS$9i%`sH-DKy~LAkD9
zaNb?0b3BxThM%S<<rE#xb+Ev8Za{Q4In7#yf<r<I4$|<W4tFyu8v18d>N*!jzZ=yL
zi7*3%CstyVI<KP0<$e8yg3ng>A#Zg99~m&S=RkoVxNZe5v;>)2*w)`;<MQ2_r;Y4O
zrPi&3Jcc*ZLAxNQrQ=^j>kykXe#fCfd+h3gB}(|vWfwNEIfMdDC*$ryorA7>(8kwM
za0$X-p(Kn)wFp~%Q4U?3K}jj(bra$PEmc}o+6Lqw*!PR`aUF}Rfgvdmpj;9dXwjI&
zDTgZclIGC}qk(>Cfzb!hqLEFTKE*E0hr+^6Kj5%KvC!h1=S~PW&X|?1O(w+z(gEdM
z__(p$&9pGEnZ&`KlkH+^31#vKkTEi9q1|yohmyn>Gz*S>TkN?NRvi;&%q)}yD;xz-
z(L67qEJ05m*PPgeI)8xd1^xE??6uMtMc>41Xpt;NG?>!s*$ZMi((W#jutPbHcS36M
zeUzK~3oEV=BPe!hKdvgZPRpd2PmVd8P{(`zrRQHzD8<Pm-wGXc7&XcE4}J8VXcOb<
z1(9`-QsiUoUcmL9-8YBN-lPia8Ex~pm90>Aj92bVnSPH#4JOy`NYbQk8}J(|AkZGH
z=HewbC0A0q$2&`wboC)D<Yn>P`IlaJe&Kf<Dr1=>*D)aBVHdbOL5<2|uu$*?<h*gI
z<)HHi@-dwnkBzLF*wIrc#Qqab+O%fX-6sRNF@3cPEj~vW1nITKG}{$99$6dYW9R4h
z=viH8mq}*PT)VPzPS<(vY>IKDjJxD~n_l<oTtq&F22zpa@{q_7KsjJDn%RuH&xlt!
zTxij@>cwqzK)kG7RcEvq`#_<TieVWBI!OIO>e;I+>R?`$Up>cXp)QmGP&XnhX66S)
zETtv#xyg&L)D{s0*;vH7IS$*4DxX7+NW62&8$?a+B60?TXInWS{RFXD6`==*q$!iE
zPgSCkxCwVzwAoQdW|$r>>~Qoe)M=z`;L}b>_?(K3Yz@>Hg>tOu3=%QgH5JbmEhzl<
z(P(>09ofXe*|quUXkk~N*sFC?%f{+%vFqw^WP54dG^h~8rq1)CW6vb4?Zx?}&??4d
z1#xoH5@7D=>%{%_C3W$uOFI=JeRE(fUs@mS;tG}}g<^td3wu@p`Md08F}riFKj-?w
zZ=aoVWJqJThO^emsa;<o)CG<Ybri}$Ou8|?cRSgBc^$`swoav`FI0GMabm^#%XGZK
zUXd+hv{j5o*0<UcVP)cK$teexY_l73r|68Tkx+7$@T-y__+#6Z_4bhSxN+qTOqRVe
z=MIMlYQS{yHR`m^WU>c&tS#Vn$;B5Iema>DOSF&uvl>oKTxWz(9<lbEK!jD&Jv@Ok
z+J%-pG<3(N%i}(^)w~ZwWgP*fxnyf;_20Chr%;hwRU(u_$vP{1e6CV<Z2=0HjSDUQ
zwA39$4C`i6Yvt4WfV`-5-JJWi6<Q=k;IFan#BCTR=Ei)l;C6S!PfCf)?mp{NmF{x8
z&8x&^^3_)r3c(0;uzv)nt}!E)_YKY~g*xHf#TQ>{-3bBN(>0JiVxnPWBr_@yg!Z6Z
z6K7Ah8?!HILX<gR2fo3zt_tV6X<>dom+_DYD7#Co(~RXg%Pg#??p&N<dkQ5t6^+qA
z2c6+d8=NMgV6g>te^0))@Y`YHJ59LnYa(B}-fRgKmvnmyu97l?7M(-0ySOa9ufi18
zF}<bH9j+W+TWU+3+ltXpY6@ILZDI3Gxh{nHp8dyJOiS!n)ENRI-Wv9*TvG)--;T&b
zIjF&0c44RW*H<{R2xZdWLR}u%a&Qt@g^jWrIT&mp-$5UQ5G00sW1%hq;a5%jk;D-R
zGccG>W7_t!Wdk}Q>Ou#L`W2@L#xd0FAO$wM7y&^MQyywj+7tQY92B@BS!kavs!NEe
zcd$?rwrIC3vBed2F1T=fp%8AXkg*!Nr5@z+MA`-Z9vt3iH#KwPvTXsEZBJe9ZkcU_
zm6H2)0k~YDHrSGS;1T)HG1R9pl0s$rJE7~*aO~!~F>p*E%nJ|5N)+QRcJZ>Wj(sr!
zt1>^e8E1vEx!-QdQ3R5OrOjQjfYxewYudH%2D`8fkWRYOZf#y5GaL)3L<%+LNVK!I
zi$apnB{&aGo_e+}tutuv?prvZO#IYnNPYeDNx(>Ehh-J?GbYTuvSbOM>;g;67=*yT
z!Baa*u-sl~(cUKol`U_ejo8799A1SwUug|XVGar=(BRkEiaMZy!J+o{zW&}qjkeNU
z6dbm)4rX>+e_NrHgip8jI9;~a#M8a*-B_PS$J1l4jZ!$@7E1oEEC`3$MhEb;d|d)l
z%$O{N3etApaiFhI+1frP|DV$`VH!aj_WFE3{F=o0N{^_cMGAwBh4xCQz5ktv)^Diu
zC7z|%y+)~V*u)XJVMJHO--+AST6<#yZuudFi0h|Gp(Tidrw$+=3YPRY<!E?<!_kR{
z&qL`fh{P?m9bA)Q8c{?pDy$DPN55=uuJ5x(q5yH7@Ez<6x$3MewClJRE+R_!De#uy
zOwV{;;g{C`LbN$=&9TLIw95NRtwWq=Cc=Va_O_f|KwKJr($46ND73^{Wmq!-`S$j#
zLkM&N$=^a<U==oJPdA}?vwIKNZ4Kn&I*(w~<cL#SJjDl3yUx*%joWK)j84H)yP$(X
z+-cDkA%!=(LUd&Bh%)SH_khgnl!KPIkquTwoPZO5rz`f(N+guGP3k`?@&yu!LZlRX
zfdv-gqde<t>|NQeje?pHpm_cr_#8=)YYOdx_C=STXYalXv}Ha$Sg4ERU|yei*Z1Tj
zEHgs8dfQ)WowPY}K1E4x@6G4p?qXefy1!5t6c5)E{Q0p(y{{QtBf|5(Hl%p8V}<s3
zgA4`1Q@y`Ve?o7kGHVwq8oDp0W<5NWC;ze!)S-%@ppHPZLY+Sy7VfYBqE|(!^TB*t
zgecgiErmK42z8b6hnnw*xrq}&yJKUaC7Ao`2r9D==R2$B_ZNN(_eL<hDt~HQJ8bwz
z@;kb-O*vloATLoT7!$1mm(jA1#;N4KdWAydPk_Bw;ux1V2-I3tL4#eMoC;jJL`{t2
z#M{B~J4OK|32U%W;dnB}kr1*ee%Eghx78H@Y;d{3a2I@`P5?lmw8_>~L}2as8pWhX
zk&s3XW^3c|TW@{F^@ZQj?g14NhIiV!m8YIccmE4rYHcb!8d&Y9FD1HMpCd$G9s%oP
zH2rX8${s<{za|=_JiAa_+pzLfUTJ=elRD@s)IxzBZtd0T$-`VMz1vFs6ggaY&v8WJ
zjBFHJeZweddSkYPrLu;->SCAL5*K`hzxH~1-eq^@$n=E>j$Bu$3oK+Dk}CarC(%gT
z)O<U<ITd;sbb}Tx1rEaJ+UD$2swZVbf-fnxF0u;CZTFaMiLHIw^PgV$nahEbz3<Tt
zwzaY)pcY)#AlJ<e7AljRDD*d=BS@u!-?n9YoG>n4Fv3uQDLWhOKDS^?N4=Q#*!G}w
z3S+^?eXum;n?nrPQ8Ai}un_lQC#cC)6-sitfQCD(`Ug?Ei-awa5S?1#ey3kWX}g;D
z<jCLwv!&KSo<jD*I=Q<-n^R1u9O<$W+Oz~O2<?zPa%E%e@bnctOGJU~iLXwKD?uoU
zvwYe0XLGB|-bw~K+6%vZ8UT|muOsS!?ThyF&cpb|?*y6nFqu6?q0vs*{%nz0==n2*
zye=cq;(a1L;H7~%4Y}<=E{&*iB&!q46wxf((00-u?|^nnVk~cigAE9Fpg9R$=6Y8W
z)|O!G#W3g=KuOz&V(e`T+Byoqy-%0g94wpdaOJ8m=YC=eKDm1Bmh9=A<AC5uwnz+-
zVo*MNp&_TmJJs2a9|3~uIy+i<;hAC=h*)bYO;wn}$UbGqVrU35+e`i)S@nvu?ReDR
zdh08=vOIg|qyRg_sqM9oHCW%+G8v_ZilK=*Txd_w>H1dLZx7kWqYmnh<E8RafU(qt
zEV^K5o(<<qu@z(--(9FnD$Jq6)DA(e!Xo5%*+XjMLmHgg6;DT<gAk;H*{Ul_RyZ43
zc`5=Or}Zw9pp@Dus&%BF`HZvdWcDYS7WILo%u?&fHHS9D^-K0I6##y&qswrn+Kus^
zXo>a)cObA}!(ltsyff;L9s24>&Q#ljB!yOc%-yE;i696a(C{`nmQYSqK?(*c$0g_^
zLf&DYtfWD9aY=548XcLaE<J4dQw`W6&4AKaQ#X2XPahgRw;g;spWMJ1;ik9YHHEsm
zD9*=f-KBw)Jgcy=&jdWV6szzv$LO{sSZkq>*u>9P^0rW+R-L6OWBZ<uboaS9kURPo
zF5yzIuE9cuXco7?wBs1{`MSt#OScN4ba8@FXz|hcSQwEqoN098wb3To7jj0@CwW{1
zIbCWU+ZM%}9I+1D7xOvp8IipO>|S1Kog^zs^17JkOF5Fjf8UXuo=RLniz7ieE<z?o
zj@y^B4n>N~zGPV_)P;tplU$q9@O&kv3hh!3a!PT;z#8asAu-(IMdi6d1dn~S8K|@3
zunwftY~I=K4zsd5D;!(4isNv(U#r7F1ESCYC{>{@_6?qSa_LnVgR6|}>&??r#K{P}
zr%<Os;|n|3hkU%yzOgd$L;?qw5JgDgV<40D?Bo9{mEfG%R(M1E?&AN8vG)Me^D6GX
zC%qRE(koz^X|iO?76cMmTH6AvAjvj_y5*6yc(uFivuk?|B!fF7F(#FCSzU5bx2$F*
z>s3haof3L)={@;=|Cu@Wu6e)jy*~N6*7uonKlf>K=1f1!udu(V7Lw-6Bj<=WSVRHR
z>QN$R{<d*dx*%=PDW)293%Bi|6d)1mdm3nx?5)aAfCp!NylmL=RI>NB!G<EQ0wkqh
zacW}Kl3&4tb#AY{FBb$CP#o2ql6VWkbw=&|^&Kgk<#mvkGs);jjlL63oVx0j7!v!t
z*12o&JRoNhcRc|T30!}(uKD;twlgx=$3NI9X@1@?(YFtFNS10NX&6K&cRDpGa71D-
z`%u?=`4%N%0alw(QgV{ly{Ha$+lL#P#EA_Ou)M-k8KXGez11fZEW$)q-Y^J8y?vy9
zpQb`2AP_W7uL^!U;MAu;vToBRqDMgOk2XlJPtF%04n+H!b3qBzKh~|)d5-Y#T617}
z16EGl+{e2&6e61UYXFNa!l@`kXEchbS-4;06K$MGx?GPGqm`H1{mh1BWa^XlZN}^O
z#y*)D)z?q9ec8K3-q|A#K*H$!j;$o#B<Sf=T~~r}(vERpMShfzNLUL&sp&;yWS`FU
zn&A^;_zqGOE^Y52qkN`5r$9AH>dL^_V0?nJT>-Oa?6V8A4x2A+LM9#xRs8%}r(Spi
z7TM<-ng^Mbc3HFohc>51lNt_JXYKRdNTi0#8=rv*Gp|NjGHqX|HXBnGUf!YY6Ta+p
z>g&MU$)T$_%t!6-+jQacEE$Nhn%tmEAgApgaw4WH?G?rNKwvdUhPSf)W9xlrdqr;R
zVrZwPds_hk<bJS!>ZWBO&N=C1*oK|@_>EFPOQ=y){Xe%BL|s7oPW|H4#M@*eMHxY+
z6*#$nY1<c#BNi4mumeh#VyZW~9S_&c*JuBlOR#zU2Io;LJilYrsQp`O4uyksL#O$4
zuoX^Caza;*j@iF=ldT*mIU_;qSvxo#Z}mD@^qEM@$sB<+*!Paxe=KrcQG#ZrnT{Mf
zBFLerLHq5zRrEd`c;fy`#<|=6vuZe`AmW-sqHe`aIW<~B%>P3~QvFvBZRwKBmN?Hz
zWH^(MhZ9|675}@NcO?pr+4lw$@2wkeL2vt?w%x9p>LIC&JAfo@Ls*EoMZCTP1I)&k
zTyWZfC4sqVbn65;_WNdc#Jb_~H9i5}!Dfzl!Ptkl$>9?9(mDH*I~q|G_XJA~Y1Ae=
z>eOh#bbW$-X+BR(PY*iJx%)Q`4vRc|S$#q5IVWZ8@$m3+Y6?3wr)lH-@;e$`DB2Cf
zk_VDV&@QJwM>53rkav)MMSW!-oSozkZuD?q`4-lErHmZ2ue_r%59kKBGgtt%5H0rx
zx-<p+*;n21?u!^03o4hBPQvH8wn^|xpxYc?o890to-_Bm7`LykZl!rBOo$GwP8brY
zg;Y}dJNud%_5wShNKYJVKt(o|Qy)`x0R0Yw6x!I=wuxR%78vLT<wVq&Vws>r%L~eU
zUDaqrXqI7S6n3oyH`D|fW;Af8wdHA15ybFj_Vr!85=HN?6X6lq_~6W(`dsXO-n0D;
zcf|f2M58qB6pAAg^C@p~sjvx%QbTm>8}Eo+g8#!Hcm7gJx&v=6_Za0b_@;uT*Gl}>
zf#op4?C{m1<H0)FH`m)b*dlfgCq7`>9Gk))pl45Nh|xqF2O^3xsJn#iTN;+oeHe|t
z1JD_9&A<lx)^0Q;MPqu-a4dFl%$yX|=Qdn7F@$da$<<9!udpPr8B@rnCg*Ik=4{Ho
zts9UmJgNH#pJ!jFQLtvPdqL*T+PBwG;YIk3*7bk`%jpi~R$K%hZ!4gK&1>J0o0R(}
z#+~Oneiw+|V*S#0ws~6Do5tFYvjLFgzQvztUw>D1;Q)w&8H-gr>%cUbGeZ;8GxptW
zc~=qm=#*52s0F?Fo~)u{C+2oKrC3oNnu^F&5Y}!Xq<!xl#kfwANQ-`q^c_&16RT(K
z`>G3aFvHl+u6V$K#mLe}Qv^#oCS9c?UD<y7{yWM|L<zk{dc?%w0{~FCH@^eVit&+*
zW%+^nxY#+}Hi?&k153Js>(wMY(6s$vw|G+E(rcMPP`%yn)ChHY+h^=4t)Id9F-&$|
zgN1R5pXlzVr*=(}{(y;4+z2ooCG$IO>T%>ft*{kFe#9x%PLS~+!wG~w2koMKZROY)
z`6?(RVvbHtjL+KBTM^qD7pio{u^F7h?HR3hU2hVX4~j{5!hH&n6n*70yIv~gq=tOh
zfyJx`VXtjb`b-J>_N?~VEeIq%r<jWr;r7x3iQx`~Djuv4+q1i7k3;T4MD+&KIRXjf
z%-mxaCo(+qHP3#L^PFp_dtkVQN*+SbsqchH%ZY}Z=D=#!P{G7@leU7W7<+E6*zoL_
z^BCinf<ty;x!*jmVEdQ_9ar*60w!#{nl$D1{Dzs&dZF`Nzt2by>9Jpsi-I@Srj<@1
zdLb6DrG8>6=MLJEu12xpi!!e}H5b#EcnsKcd0I%zzL2?Q5$T{7Y1J-iW5uM1s*`OX
z%1T~Or>1pKo^0avY-z!yoQ7DjT@E?0BoxTLx&vOEUxu^_%a7QywnH~U^`@pt(BpuF
z_)C^wZkM)JXj$ze_;?44O)DMzLA$Kz!Cqz9bV?!>eYvXI@`fs*4bJHhp^%;e%yR0J
znyy7#rqq~Sha<!+wBU>M8s6&VC!4B))KG}>3cIgp?IGpLUt1ST;F(7Elc}UW=^$n{
zZcy*piu!?+9a6ieL!@qtk+o^RQ?Kb*arshvVf*b$7~pZX7l2p=d6bCe$zl9OZO+)q
zN~=W@8wVEHl71zq3_?Hj04w}M?QJ+-2^1VQz5r7mb-boAM7_Aapxq<$gIp}~IHD%W
z!^*LtneFzH`k>^~OsIa)8%(KQOI2(?TpyI0aW7pz)0|tJn$$LMjp^9*HZEhZA87->
zaTKSk*}ek^w9@_(gQqi6OhK^mtisB!7s`hVDx-J-mNOqW9JAC{c1xY2ct*cwt{6p4
zwt8_Ma<@?Sgs=d{-}`M<L!GU-v=Xa8xd`R>(8Mjay7~|qHRv54Al_hNPn2*09?%B5
zzG{0#6pjNcB7;Sp<|I1k?;*RYP<<D~=%`cb<{`6XBt=z}3f6DCy6vT<&STQd*n^^I
zx=!X4>lo*aUDJBMl6xS=onoFFHw=#SY)zXx_#RRdj5vVkVEW33b+y=pmsUGT%5$q#
zP^03Do%;CYVyP;v4=?Km6Dfsa&x@Ae!*>CB%Bjy^1_jqKD3zp?_Se=F?uItWB=Zog
zHL#k($T=SB6uyJrc5S_l>VkFEOdNJ#F+1w7SZ8I2&4q&!&h4^w-69>=E~TAa4lG)z
z-+<f^o{YS^zO{{%hF*f5-3(^b#E!)z2(n6R{+MlOGmnGK#H3S9CIMbz>!_@=FHeGo
zkOfOgp~j^^vqe5%wz0mQ{`C5L)Pd=MA$o6xZL0dzcJD_#*yq3sE4oa$%;|y);O4G1
zY3j=hI;3N2ayxAz#Ah7O`Rlq@b%iTS0b?B+E2kzm#=tl%gk9ebvSpWC;S^#*_HlfR
z36ZY%hHf!rdJ=8OTT123DFQ`({02L9u=*f^y&^Gv)vA@w<Nt@5Y%vS`%0zHd?319<
zt2BZhpDRXwv^tP07g!T(UYRYOdUYTIp31yougdQYPAlnvQz~2~8?PpVQ##q`S6Aop
z7lx9GQQzkcMtg8F3hO&%ujwX1hgI86!tdDHPJK??1!sCf#<kbBB>?Z_5G`>nsWde@
z<|c-RXC~!9_qv>}f&Q@b7#~uqU1%xfewV$zV1%kem(xoDSZ$HWL7uD^d?%tAobyo|
zyd#{rFM6@_6klZ6pr}+Tr){YFwZ!;z_>cp9WF7%gsjo##TURi`b33I+%#h=GZfdjP
z`bzqqi1!t1t?(0(lZU;lk8>f>H5J<-<Jd?|P#P$OP0a&_n!?dH;;U@E972Y)Nx#i$
z3oKumCcn%9+tP|R_eg%=6muHmFbegpZEe#Sv@G>~!18z_Gcvd)o?}~$Dza|n^0RnK
z+_-Ok7oLsO2QsDi9<vHELH$w2jn-+PbMTy33G?%|k-oj5^^brVq-lyl36?Cm!ftG%
zTSEjDVH0%1foVUZ78#$oS(lHFwa%y14l{DX8;pI8gh5X(+-}<&iMjMr=UJ>5ZY2zB
zyoU7zni?O&;)gUueGF?c3y^zcqJD}0XdLMZNsx0(BLLK^8I*k+H`rwLQwl5TC6L@4
zm|oa~5He+Ntgof|LgxnR^YHqPoUTrNuIpy3m3^DaiQw9&;Tx}2;#L4jkV#xY&KPcP
z>$iOt`rgbu8k6DFBvo9sZ4~Kz+@`ysJQ#;f=T%V~llt&EV~P3-VqxB9>L;Ol6G`Oo
z=d^%-mQ&Dir#{=j*9b40zTK3gCA|adOd;?vY)UjD(c$JC;lw22t{^hSe1*@+Q{(vD
zPP?TW(gLApNXTZ?#J-OMgEc2XesU=~mf4fDPxxgfP!2tmcQqlmlDraK!vSfB?AF3t
z#nK*gO3}h%YkXTQi8OC&{RPYAw}&IN1B>5anc#2*g(RSG&}wHW7vY<8$g8;0b34W<
zilTpy-wQRq7h&uzO^oME6}B>UV-B#6Rw<ms^*3AVk7a-7caWd4=alNbxb&u2c5Xnn
zAJ4rozvJ4goadNjsa<6jWG9;!Y%hun@oRXqII#K}DY8T@IXyWRh^u{TS0hzbsy;<I
z_j#+)c`3<5Zf_DD4%%(mgPW8)hvz~C2^COCyi0EHdXTOYXYeCvut*BScOIHQZg;dT
zHaT^xQ^*fN&}HV`jUwdEg2`kpZ5ijB1B;==E*7lh@`voMsu4;_6{ReRAw_tr(PT)o
zr)1p{6dty_8>L)D9%3aU?mM8|s@LG>(YJdV2ad0=l(Lj_r*4`V8R?CqJAYekje)Dz
ztn(`1>oe1{x3|7Z?L%MXfD?u-rzUx(Wy|a*>a7kP<NQUO_g2%CLwSEGfo%4Z^&7Yi
z7<w7VWT9dSof^q<Tc9#O)i8=+vrC*(ceW}RRLr^bA9dxF-COl-y+g&%R8|Twr#^8A
zxdKxK`%>4o;=5w+==O<lI-M)xtYpYgZ>PR+MB2lQHoq_XHXT+vr&RcfUYf5S4KCpb
z3XgY|raKi1se4YZQofBMOTl%@xS`lz-i;9^L!TM-7I_`pPq(Eb8PB%SAr!hmQUcjL
zuxYdXO!YIJ1B3E-gVhPotR1wUt@rkWkP>YewF9dWL>lJIngqHXx1XzCNvvxLj+!~<
z5Q|KGS}bu!1EKnHRNBwi_wp1J8tVcg-6Ts*_8GJ5cKd}kx3nl{a#(T)CJcsyxF3Ck
zy({r^vr;rUg>JuUE1aBr(0(zwDmEccQ5)Naxka;YztoLZ*%mbqN4&u#-YD3FzxK;*
z-{8^970DEncrJ1zOY$N|>OJ-=T}LpEvRk=0*9%5~8FPWNrH;y9ZToN4V6RV%Jmp+u
z_#Ih5BM=s`wXC7`YxP}R^P+3#A>_bP0W*SkJZ#6Ww;@H2pM9`+dmLExH`9L{2^agk
zu}#eXj;h}s)Ft#F(QEC#6WDVW>YHmK7Bcp|G*=*q-CxXS(yiQNna^Irhmu*Mk8$3L
z1z`J#(O{FEvz^s)sPouTQb8zjgB<9U1eAKwl5H3SY*)RT>Wb_u9%^woD8hz%p$7aX
zX7DN7-LRI)l&(O2-uwvBmYV2@2(0$kAyk=rTIUeXB%msDF0h(5IWO;Hz_MU32zPrM
z9-|7BDpRR_Qe2=WF%f%_@Jcrq?yIk*o<FqV!mdy}5zp_fUJ7sN#5fP>hXi0TGDmHH
zlJlw1nIkbmuK}r<`5pBicUjwkh7S;ZWdfueo3|->*k&3Y64JrO2aWXxx6a__20&b&
zF3srM1GyE&GtKRCO1XBB1!c;DeYFSM5%<Q8o1KDo?zIE!uHJl&J=D(eu_5PK?3P$9
zRQ|{Y4`=1*Bz`44=Z1$k%-Er7MTAcxiIQpot3}8gYcpkA2IENBJ>13z?#eF2O*SkA
zx;dxjA{y3I23DO%4ook~u&N6g5U~SdN4f>u9dE%3dxObfCnkXenRc{#vll*G`aUgE
zKI_mq<J2c_c{Q;(LfT^uumul-30&I&#Cp)WkVY&_A;|G=7qPb(Jzq?8qMgIigNOli
z-YVoxUClWdofF+R6$YZ6CzBU}guYNC8*e7O$!P6Ix_J%FC9Evg_smcybCpxCm7(nu
zurnv?Z8?sm7b-hg&i65bV@`cC7TadFkB-=*U5BI&QpcORlbRIY;5Ud8MlQ5d-Kuj<
z884F|&Vh*&LBiWM9+z9&>4w++Nla*cryTR3G#E8ccBWg!NSY!3!GVcB4%|YTefU&6
zn>}5<e%0nzZWwTiwLRo#lM?kl)MdNuTrwMLHV!zCi*XPJx#@CZSJ`7#zbMX%RanSp
zM|}mME&Os3qo1p6=X1taj@>+nik4FD?TNT^Dssa9M(YQ-D0W?J4nc8H%iTfPsLxf{
zzH!~kzWrv~pd%9_P9f(jbkEoV(EH7G^Ob3FNd4i!V%;S#9=FHa8)8Cek*wzqOkCJ7
zIb7yz_R@cmPK5alQ=C&1%|S1NGdMMW(k`^k4!b$+l-fps&27i02+MC(jW&nqv;5Q}
zfW)b(h0r#I;zo_kF*<b4*}EH(+>pvX$~kCyA;%d~7uavNS-JF*<xU}9$UPGQi1s_(
zm_uBG;2-e@%h{GP`k4K0bp|5Nex(EOsiG!wOM8^u=hP>vH<qk#zt@fIe(eZcwlfXf
zX{RPTa={^1;=UfV-%moha?R>Z&U4|Z;v0}yVmxxJ{DW>zI(}9Mt_(%mtWJG=S@KOL
z4mxLlSYMOwN4L7mF>E)7!a1iVmS5JEJ@!Xkt7))6;Z55N->K2Inv{!~oNWHMTH$nK
zy}8I3QUy$hH#xdT>`xZz!(O4wWr4+*zBs^pr~PSLqsuN?;uK;4V!P!PPckn1vu;q>
z;n+W*k1xRR{G<y?kMrlX^CVx1Tgj+!onecaYu?Izvvq8Ca-01{Te#JPJ1XDEtOL?e
zR!x&T<u4QeU-P_|IFI>5>?~Wi*<a;bGxEdXIeHDd8T!RtsGI(}eVFJ@atmU>azH6^
z>AIcmat->MT2I|N6AE$U1?_>-6FT)Z9Ve+FP#qh{-?pz@iQu+PN*5sNFac>(_MWVL
z9M>!M=WXBwkiv=Zc>rB^+k4xbU%la4r;r;BAs#0jbgc@B+|5}fjq6T(UmG0=q`QN-
zk<@|ZS$_2x*WA3nl{VdsR)vyi#;f1_TKl_veCD$+avn8(WV~iRw%Z2^A5zV%O{TL$
znZ2k<h1JI1&Gx}oxOB-<rw}-V;x}{`viGJE>_gdsmE--J@rI?O4qSfuGW&4#U8G57
zFD07-7URwHO<P4L=mM+vkv5vK(cX+x%AKt|sgu(L{@O>oe#lWPT>Y3ZT=AV6t;CWg
z%a_>4@-s~6Ll6i`i9H*XUE4n1<}{QNoGj@0faSoini{oF<oo1!B`RRpDb~;*2{ryH
z)<|UVlzlRrB$vrj=Q-cPprsr8<aHQ<ZuY6xJoc)IzEcz?atGLMwiNa;`*i&@Rbh)_
zSe=^e)>hWzGYy#((t&F0G0yJ>6FpQ?DVcUZ+lceEz45Kn+njRSAv(cj@3zmiby(Ef
zMeE?y=agMd!qctx`7+5vmx5|?2`v~SbK_p*NjV=YQ<j<+11DLQU&M5zsIeQ&7b$=s
z*caLoUoBaZQZ_U^G0H*r_c^1ex^8uzv)?f95?s&o<9}!v!Q$X6+QXa)NON@6nt^Lq
z+dp=zuvc2x@UaOI)Ts%~Q&%#3tyWP7pE!u)_D|W`{0^eSb)y}((mF7q6xjlY?4Pq4
z(tyrPIfdMUk8>S{{Y!OJiWCkd1g;YfEE!bpHH1|gw|{M#JX~=|Lkv%bCJ$gOjk0cH
zFky)Q)|L-%3%GTstWM1DSijc(y=$p18pKu$tQ#223X+V{!Ah9bN%N5XN7XN8DN{QU
zp^Hv^qfv->={@wH4F>Bee?TR>0G69^1A6)hB(eWWq5q|W&Qp9OKMYd9*!_2{MIm%(
z)D>K~)2XQ@Gw4@=82^(~I^M%uY04?JL2>4%!=BWKVn!gP_P;qP%a$!&;ymZqs<XT1
zcAMQ<<UX%8BNtc)*{+)=$H<ZOC3luYf72k&=F?6&7j<xw<YQlYXDuKO5RthSKy*oZ
z+lrZ|GYaSzgLv4!tm;!Z7l&J7J?lWCHE;FOoa&m!$XFO#S(Emdefgb1IoEDl<vhlS
z_RdMvr5aWAS7h~)70bOIU-GPH+gG+eqm;kVDaG$9G?!*~+gG)5yb@pb9t=GI;nU;y
z(dJj**><V+XZn5WBzVZF$@8+0oFC`q*VMbRUod8*bO}`CvU6$@GX}?oCrFNf=iArb
z8JiL(Iw&3moe7sMyWGC6CQ0gf`Ppz<$43tE7j1=8e~~(;+X$MUSvl;nt>gyZ3G(*!
z`Pm_;2#?84%^$jOn|(v|G>n#-J~R!fFQ6tzbu-b%3JA7uY^#PAQhLWD65ph=5?42W
z!drdP=G1QJ*lMTWH{DtEi=0P9)lp1>JvKgU-(25*qPVgfMU5O94m$Okmx(PC_M|&g
za*=*UnPe35E)-EnL&!23xAT_19toE0Te`(Q6bFtnE&>axs|};{wz6C1DgCVtQPxjS
zPdKI27xo}_B*WD8om=h6IXlC%WZ58Z)##uDQG){txw3D|!l(jXokDJw%@`)^+pFCc
z4MNTZT35^nHA&xJN;=<J`;K~1?!r<a%WTWtJTSV{)Gn5Ooh`#j_ML44<95)z!Y`G=
zXtfwF_3;}hxTwV3{fHgk)nV8|jGzQf+-nrL?5#c(uUb$AvI(m=o=6Gw-POa~#@f14
ziQ)z8)MzrUoS4A<{CnE?r?w6aIwh^-ic6N*_qJsTJ76fPy}`;5OxXHC`@U|ej?jYC
zvP$Fk2&z|VVpNLx6UGxcV&C5ocM@(^_Y*jfSZc%vi6|mLwbVb*cKjAnaVRPQfUthk
z%;cm({#fZBtS0y|n65$YbYQ9A#CH)>u|4I^u#>;(`OZ@<jdp%WQ})!ht~A01?21zm
z&k=iC>q01udp!q`XaISb*lHJb?MvQ?1HyrY@Gc%h=TI6dP)^pf6c%Ug>8&s9L2(>9
znSd46E8k3<YD|a{5Iv)<63zs-I;EiC$rhw6ozd4J727lGJCGtce$frF9s7{5sLAd#
z^^=qKtR$hNti8&4uEz05jj1f1g+049CtfiJNr$_9{od+h`uU~oy>YsMiQ+CVuAzd7
za>SHa>A<ohXe*fK=j1$dH7UI1NN>aeRBsN&A(QHMh_pSo`x+Slv^nP=@disBiC`z}
zd2Q*g8s%n9BC-G^|8c`Adw#<aDSz>nrku^OUFAUZU^r{Z)#(M@HWc+mv&yE1zloKV
zKVz%k!!O0h>-G?>0*Wb}EViVzL}VC9>y-11xqfg=`b}IWFR8{;B+=oecoKAu-fE=K
z&0wfvjBIKBMsp6UiqQf6Ck;O8)C7`ig%oFwzp5?E9Y4Qg+47}Mp>{!f0-k0QUs|6_
zS4CY@E<X`viKHeMY8AHZo?VucNR;)M^PKw|y{pn$Ghka@Fz6Be66&4w2FsxP(0kbB
zZFM&3LJHg=0OYP+$qlGu6Lv-SafC_c6-A0S*j@+<gO2%KQR|}=BhroNZz^!b4xe%A
zMbWkl-eNE8<}8xmg!(<)xt*Fk1V(z~uY6Ir9*Qpw9#HtqK}St&yi!1vv#K&$%2#)f
z{ZKZxWcf1ZQJ>U3a{V!T@txttz6zEid>oW&)By(4Ne3-8i82|6j?`X~Bx}R8*V8LD
z+YfiWlEA2SkG!9A2c4Q&2Vm$dlFRy$hV0LJuJauAI;*s-gge^G8p)!}m$5Uoqt3(j
zQ=be*@0LNhew-ezEV3q7AoZ3Lp7UK1;(Nl~AQ=<hDt=c+I%(u}Ixe^gtg60fp;WKL
zS-^50=%12s$P6oPVVAA08Zjzzf<RIfeZeF{P4Xw3NX0tP3Rl7wRzafAtX);zkqu9W
z6OR5g3g)nfWk9Dsr$z@O1KG_Hb#>S1egtGqw4B{BdpR{7!lT$=B2L>i-4`ASE6gtN
z6j;hk$mygm+h%Lp+%W>hoay*U4O4pANMb$fycKho%tt0Sqj_-zk~!?94NG+^hEqy%
zh92o<)d?7-;6`u=bYStdeG5DTEr(yXFwL&bN$cGr?GJ@oHW>+s_eD3dsB5-1U!Fjw
z={p6rJUoFeeqFbAdFME*28N-*a%z%rT!oM_Y3o~&=vb^w6l*HahrUSHb8KkKNV-g?
z5J=J3m(;d{n)dSgsa8hvi+pB)#oGzn13QlLnH4^18{7J51A_t5H3LiJS+;DsZOYLM
z-U3DAS$)RX8hqsDHm!UQuRLUg;|?sR74-vdg7H^GR#B>)5O!T#d#F7r#Ks&zj5HbV
z$X;XDH)2PoPA;K$%9>GFqCvZ%h5|1KT}dr*V1ZK;UqVO1QKLO^&|Z<hB%AXtujlz4
z>(<#T7Zz0b10$d_xr?-S>Z?~ul&5KX$bPh7V(R6|j{CC%i{XkdlX+3eQKVdyXBe(9
z_Nw~LeeCmh!h7Dw-UJql0auKT2-z=~cZu+?ZeN2JjIU`UM)vrK%yQ}zE9e1MVCfI-
zH3^dwNn`sQSUE2)sIX-v7SojCsC166t*_x?$x_MgbJT%prs?Z3dtIv!zJ`=%C^{QV
z+_?h4bgSR%>%Elq#A!d`4HoWI%n0<#2HVj-I6RE?Qfgp8F@1{egLMj{*<x}y)Ve^8
z)TYvM11pJSYGT%g8?LThzu76o=Tg5KRDlso(#pIcBW-@!;pL;sM`a5{WzVOn*QarY
z*aP$2OKe}QtQe%Vq?kL7MHZ1_ukrU$4vQ@fT}Sxk?kT67$P<Jtm@eDeMhe>zR5P^D
z3kXO3MKs+~JN@7OZ`-OTvNptN9?C6*4R19Pt0f2>3!R-@5=PsBwQ_tG<vRQxkm}{5
zArT*h6@5c{OS*L3jO0L)7Oa}Q+HS0d*~%AWnwQgMo@><9eo@p&aKAQIFi~>tmxG*T
z4JJoB*%UajY<o=&HaMM`z(U=~g0psb=u;o7vI7_2LpI))pV+oBr&NQ;TLbX$u!($Q
zbSjb)R|y^-(~3=YdvD?JO3w&tif}2CSxyWhX(^V-W#x@+YFOg~(~9r}q9x<qaAkro
zl?z>{<regf)WS5{RGY@B$?Z-dIt`k_r-``iw0`?McT-H<!a3qV+LY?p$8d1OrWaw6
zg4x+OOjvg_hPV0@p)-Gjx+Kpd-nN<grqswvZOVb@4TeDyQkSV{Z%T~bG>z+raD{Si
ztP67cOcP8J;OK+hoMdAI?3?pkdzyf2O-KZHamxxL#k<dLsd_PV#FEexNq)M!S;YK|
zQ!h?eq!Kc;$7UPZxM{R<)jJ?f&HRq5X2!!um6P_?Ze<^ct(B(CoqUj*WH*RT)!lCP
zriMbgs>&$|!-;jkgXqoO(qt|*5}G-%+yd8+4%%Dl1#vvm2=2ddn*+<ET(@Np+mAhw
z`hpjt<)Xy0k;#9FRQ}Fe?MTi|%5aJKhMfF(*9R2M{6=C%gXIA(bH{1(9LB{LKks=L
zU+k0;e}gh>xy3E(`d{s`hwQE0vdDs<@5l>EGEkxc^>Jp16%^-uhFuWqt9Vz;Av~tt
z*5;DTgk0@D>i~3MY*1L#?aAKXh`7RI66qn*{Ma4!?Y7J@w=pByJGy5;9umA22>%XH
z;a;f7y%3~3+qNLeU&3JDfpji#^gwsOVJ^L^W-}#^ysj*Q>zo>OaNYWK19o?vZa5w}
z&(WHYQf6HEal5C@A!l-nVwBsO91(RsP#<fySZf)5x3|?VmBB&*mJhSNNM6W>3eLb=
zz2;p(_{z_-NL^@ePaH#Q!Vs%KjO@gW{Y16d#b)C+$)=jy?bIZJz$l|{a&eFSWCv5!
zxmq~pz%&D!HV<sHpX%0SbYjFQWsdo6Wb7yPBHHp<yElca&9^vD*?`i=YmdUvyrTvh
zR7<E4XQ5P6!LU2^`VBXNw=|4>yRU{IQ?vZK^tB|q)FiFQ{m;h87`+hLLAQ5S4^l;j
z#RdBxuvi3%raHt!ywo@Cr`txobomme5S`D6FfpHp?Pm&>Jl7uFXbyV^Lx#xs*$3=r
zs|kfrDq=+*?=qO2a%$qM+LdCL(L&hIH3GG|T`uH+LKD99Qv3P(Md~2Jhu(qZ?sys6
z9Q%d#1{|7*y1aQ8U~;Wh0+{{wt`0ko!udk3BN#=7Y`MbkcIvZl^fW!7PunlH{kmn0
z5bNHE11RdOcv4LZDNcz|5GCF{chG*R4Q7y=N~gCeE*p^WXX{qB_%Bx*T7IS8<$56p
z7N68XQ_gkNJP759{cgX~&8DQ(<jUgAbZV-{iVfwPXh?pw5z{H8xLKzd)LA#B+OO67
z$rFcZ#T!gLl2a5q3H$X%fmt7&;-YAw$)HErB8uMdR`wf<wd_qflS(PcS2!$pG_(y4
zJnu0{N<~flQphKqIUh%XVx?HAxyS7OHi@JV+KvSiItfTiP67|<Wq0O42<O?3$44Zr
za$wE6f@7N)Fxyq*l>uMgWar3z0xa4-z|!ej7l~HFLuJ?3ts%P;^|s9nI*&2Eps_rq
zQG4&SJ+0+|vA8`=eFIkF!WWC*n&RED%TL?hZo0&^=A<qn=uUl}g4<9O58A#~92(t%
zC6%I(Ncs0{fAPz2pVxmooN}T8J=9=i_&d-|v1~LVq6)duE#-TlIgdK^r4J*A_4KqI
z?0To*#`I)L;{$9ZXEgQ61Z?U}hid$RHr>kVqaeAy14^aNmfqBMB47IUU}9e{!ZE7w
zmS?@`O|Q0xTA$&u6+(U#Mgv$$2+Gp9XAWebC)kke;kE@J(sKC-H3N*nLrzW3+9pDS
zZnQ(WdTTaravo!$#bL?Xp1K)0D3BQYyB%)C8?7+JR*7?G9bgV2Z-_v3e55|ZSznVU
zb$Z8+IrTY_Ou-b{-j8+@g#jEPzyTx{(W+z$)Jt}(#uGv=xX>xSz11i~cGC_8t+(Sj
zv?Z6Va2^xY(PhrUNx(FnXrNWwXqLM;#vD+jrWJNq7!A!YvYUa{@}Y|>qIp&vI8H0}
zNH#P%xYc>ot^@0OxX_$zZJ^!fs8cR>Kso)4eSE?mZCE6Q5{*SN0vLprYPqYYOH%L2
zIZfO!=NS^}UOUw$bCNMZZ7B=uH>_OsGCQ5!y7;*lJCAC06ZgOAa-lP=)u6^=y^{IR
zo;`^^hWa9thhf^%`}u6a#OW(JCQ(Pe08@St5wRn7uD-3WYV~COTpUqA{l#=8Pj4nu
z4qZ|%rbCyEwwblOi=42>a!(ACtf=o4iX%dl;3JV@C^V?&JCv716kHHp5}!v+QO^Z0
zsL1cY#*hN*H*(R~RA<rNQi{QgzU$P3g|Odz!WT-;mv&zhhL(c*MA?Yfi+f;fuKP{r
zoX67T_2TFooq8SF8#ZmU#~XHQS8F!PxbUF28jgMJAkDNWbK-n1=}OM2L831yYT;LH
z8=STa?Q`KbP1<dVP+@eYX3@fGi|@DJ>iXIFtT3-i9W0S!zf)ho(M64+bFp_9Y(Kum
z3e-R^<G?P`!b+f%VJ00<w0mS0%h2X_^Ym}GZ*s!u_b^kw029fQw(Fq%PTMUTx4}#}
z#aR43FwdyGS^Ur{&Oc(m+ZY@PkWQ&vLMhk};{g48_1k3F8xyW*0&28p=XY$KnwhlU
zujvp6D5fS&6S8*)@8Z<gw{>w|P*(ec_HBBkJ0nkqDgooJQxk7vpF*`Jw&{%hVH-tZ
zeL6ELH!3x;i0jv`wLhx20)<g0iOr9gSf?fzBm!+C+Q9y}-p>9?jUip=nG3gjtI_V%
z89~ZnCj1Nw|0i_{rwuJmK`lAIT;vVxycKh<?I2^<ZmI5A+ynNfRg+2!W|^84KB%cV
zQn-muq(k;+-MmU67#E0B(-B&4^}&=;qccBE4`HCRKd;}bAJXm0*(&3k<Lq_n#kciu
z9^Eo+e^DKXOCV?rt)T-;J_aHR9|OS^@??GjT@~?{?Hocdi4b-GI@*Tu2P5g5uKD<@
z+_EFMDe$P~AoK`>e_hSBB>6b5`T`8@f$=EJ=5G?6!*d>862+ABmn~Ugf7`Z?)VMIM
zicAe@pYA`XbFlKa975;KyZYgkQxhCA?@96L`cda8zlVo~y|=9n167n<BxeY6*lNUp
zN?)SWCia89ufr?}AeoX1Snz7;!=o7a(8qMRkmBh3YjWX$gdd>>Se$>2`s9Y<&J^4&
z|E`+aHNT4=@E6^7gY9FR2z_^>1^Ymq$$9YqU#A7MTyU@9cd^?0naK16mB)i_iC2)=
zS^Hp3Pj0VNEv2|ejb?FTavRahAL?c#lL>Lxkps(5b6P8L*G{RqL)jgBz&_l5gD@;H
z`N)O0I<Pu|<X=}@44w$e-6F^0kK9>G2DDJ;(L(Ly|BpJ*_?56-k+Eq^l53mP%ctz4
zPZ&g=8yR^_jc8g7cAtGLIX9-6R5s6!|Lt~wNtzO|5XN@d$6MEtvvjs$Ja=Hx37u-X
z<uW2OBGn9%;)UDn6V(ZI5(y2-LGmL3Q%+5CUF&AZ${L~FpKK_)WoD~WNJ0h*V(hzs
zux+2JuO-WhZWd-d&mn5`HX+EmT0xfjt@i0gx(7$$4Jal4UwX+U_L)U9u4t^2G1m&2
z6W;H4>KF0O(6FXjw$HXMgx(+oyxhM|O|9aY2m_|e3{^Q80?q7mNgpm<eyQ`Q8~tAM
zO#6I3-FS=hl)T?sEb6oNg&KZN9Q$f+zXR*2ZBlC~&7=<g`+9p%TqUC8@stD0^DuU@
z&ZoTmW1sy)PQ}pZ)~#bsAzov5(BZvvGA8OWj$QVTdEk(@?G4Uz5p8+eGa;x#|EX(K
zVf$+IkORx(a$suoM*HXbHn}@PB2*HS<hojoWOfdQIINRc{7X$^raTe%fItHer~7$P
zP&@TGEX*0Yo_rt5g4hD~uZ@fhj%>%Zf5ZXgloD?`HfjIXEzp8_IZ2lA>4;OW*SX4o
zyU|AM-`m7+cHz$kd5|Jr7D;L}$bn6p)~sJ=|Ir4?=_rqWGCS5FsY&r?)%tatH?CZ@
z+5R(UKD;x$3iCU*<D2wfU2D2~dXe(Nsfnq_7#ZqXxmW&o8!Cy&C*&;-Ads(Za~<a?
z3W&+xYyXp?z#EXF2+Ha^W$lK+$V&CU?Q7h@gmr1e0aTMxj_FuGZg$rK@=49V;88@4
zrh`GBqHAArSKDVxu5b#fn%_afhe`X=yV?#}V0!l;KbH@WQy<OB{-#C8>)XEUt|AjN
z<1>A7k04b73Nc0!>X+XYeDdhXtn(PZK7dk%GZnAsN9-%|)rnikI>&RIRfSmP1`$pp
ztZk)&v9D}BzLNacPAR5U!)<ml85>b7L68pESG69q3#HoDt!WFe6a|$s{HwE@%P+so
z>+y<9FSD=7soF-WlPRwP%8D&p>}#_Zqr`#YB21~4gIHED3`yO-u3+MSq@^U8fCKC3
zgIH<_G|Vro^4H%LV)f&<(QD~AT3Lo*T?+UO*}q9{zTmMiNlFgZK3Gf1Q|U<mjoI-0
zj+;lvlpIJI9t##Cs1{RKxP4P~5W*w6BF`poFzsqdVMpeU+Ba9HGr2!-CKm|<Cr)Z&
z*m41p(goLMdr~c;zL0kn%P&PfHflkK=VNn+oVRKWCqk_lIZa?o<&~nG&fn6tc`)q8
z`B%~)HT6qr2h*W_&qAUd`__g;(WWN1k_^yi`E=_0=bg8TrizbDhFO14*^_gkH(j^Z
zd5or7_vJI8k3@DTR`}a$_%ZFAByv;+mc$QN-IU5D87qtWx8D`SYe<(aacwO{Z9HxO
zAKilgYgXb8a^o~Pv?%80CpDXWXN^cSl#js>?qb!&OekTm2Wmawyp^Q%%F&7buzgp4
zA+Be23X)ZOw+!2N=i|V}_0D4qFTcycq!EEQCWH8ZeNS~_gbN3pLX0ctB3)|sLKzO&
z_qM52iH^yoAyuu^XcRidrtJH^Xfgx!8OvV!j~>^l7tP?R`Vss7?Bq1rJ9&(DQxPlY
zF<C(IlC~eH52iRDXcg`1@RQFo^{J>LKIq2Yto>lUuOW)B<MJ#}HsY_8`Ov9PN;q&6
zal=okc6BN%O<VfG!|nM(P2OB4*8tTYwx_lpX?p=v_OEO~9OEf(_2~>DiXc`?`jT`$
zd0O^h0!2DnR!XrvnMtYSvqf<M7v=Z}ZoCQe?}$^@W>iXRs1+(nQt>>!Aq7{vaF-}W
z%`rDk_w5-qz!WV)?O2CmtI+^5xvmH+JJg=p2Do4%X37_IgVZOIgrT{4x5(l4Sw-G5
zf6Y0k{G{kCu{y)#jYDE1F)n7=XXhM%+FYIElony#9`q|>Vb39tUfk7Stc>|roPsVy
zqb75$hp*uF-gb5zUZKxvv?>)~c$hrwz~b`~yJbz?2YuQTk-5+9-aZ>DW{LN}gdkh{
zSoHVU^KveSbgd!JK?g+~(uw~ed;VP|GT}BI_cQ<~!X?9j^nX&5Ea=N$(CWhiu7Fsv
zU2&bc%nMF^bXIB~nUca_4aafDmeem1pitL{&1)D7+XQA)FX?U71ozoQYKBW%Q{l52
zR{_zKqQ~Tcc-p&qj!XFtijsqdLF11>otL(*a0<vPLut;!-r>|}%g03)wrpWq7>l&=
zw>dENrLLHQjlQ&A%GBPGzRiKfP+RIrqMl<@pRvojp>}~$I#KAK*uO(gy{I6WNpZOt
zx8>ae=U(h0bex)4vix!UwaXifwUO%Z81BCfrX8L}p@;2?dJj`C2WJXPz?vdG?;Nxh
zMYjJBr%pN?(u`Vh-b$*oa>JUwy|6nx^Qs`FC=gOft0>cA@Cwdb`I~D9-LV%{ZxWZJ
zX~eO%TOjf<(ViT(f^yprRgIWV*-r9CO*t4+(;Wxe22#T&)lKckix*A3Sp9V7IR)Dv
z8V;wvX7m9x$I{w8U@xiNjNuDY6s7hCy19rGv>z_mJ|yA@wU~R%J6LGl_F`Or%zmWa
zgU=JAgA3YT2Ua~!kw-d`L{n|(L$<Q&WvR@QTTXimI;&Q{m$lXWqyxMXLZH%|-)UDS
zuHt#9Fl0z90MXCct&rAL-4tiIqwrC}<ik#l_5j`oT*kP%kpZG(<dEV35>W=XPW5_{
zf(IG{jjD<(*+~WBfKuPGWt6+e(YIVxt?3G0`TXatJJP#SqqF^rrI*>&^->7j!W+&X
zVb6o!YP8<yAz^{Ac`>_hwQIV06&VjRwKtgZnU(R=*0gnEUD0g8ZJ%;r+H4q;dZSy%
z?WNtwu*YE9?mLiJAo)L}fkCFWF6o%PEc<>Xu_!zy`?7szd}P9|Z79BpnRiNoIFOJg
z;@R5vgEidQi4$!#1=lanz{p#iZvLT*v~^XVUbxAziM_3eJ*QrKim5$@j(7cCrOzQl
zU2iHg4E7zU`pC^BqgyaRZOGnVe)$sTQNJb#XKQcJUf!*P9LS}xQbfK2E2z<84UTNH
zjfqS1J2qb9JO{nwLMb+#NKMPOxv7mDVF&{PMBotz7H_91rx|PEjcs#&e)Yfx=dq?t
z=_#a9-euR-57VPU2e+s)X;qw>+?k>I;{?I5e*K~ut9zRD9Zs0oDc<UnH)=Y49T>vD
z!*)ZpCdo`Eoo?9TU06M8)Ea7wS>jXBEuouzMfCuc5$l(l@&*%!*YNO_^`0zG`l`}H
z0~4pWp4087(caDWqYZg(BCn293aPUC8EX#BmZToMSJK5-b#wE@84jt*xnUi}HzHfw
ztGnLH?N5~Vr~^xN*}%*cVRFPN*lQX{7f)g0LsA5k>vBOi^@tjsv)5J+WoQviJ`55h
zpRIc3z<OkEm-CiXQX^N|>*^ahB3VR^IWUo?!4Y_<*LOV@Q3^TRfhG5V?L}HkSzxh{
z%$>KvdRx0)E*iRt7+CTu!4!ng9zDv7!E{mRP}QTF=R2|*NJl4~0awPL;if)Dss4+F
zd<+UcTrka+WFqvY4lH(^u;+}8R3CHq2CndJED26c5)%3&Zdo)OEJHIM7+}35x&4VP
zaAOo|H+2_v@UjksE%li+^lSQ*_(XTWvzO47QlH(Xi!y$)*tWHUMecV{=3xgW3eVoX
z`byi@7Fjnc+%!6cWdKl4Ho7y3L`uIC2a}C9)Et?GK2nPJY9WIe4ED|5P_3qgR4nfi
z2UcG_7A_0>nFNKT?YA52?YKdZ>ET{W2PRUpe8nX;RzH*tNsQEkDe}(kcWTOq1P}eK
zabxuMyUNF_KZ9%OD3+7ySqB*Fgu;PyalGrCY)S~BtYm|wO-D+<T{h7Md&w0mmN+Ht
z(*6vAxro3v*$qehaFc^*!Ko3QUCr&y2sZY{Zt**fR)>t-Qs@6~_2oSUEBuH}wZjz6
z^BB2{96(Kbmq(vj!Rg2Pxr_Sp^W?d3m@||HPY@S9ons)_UsoIToU-2G_O|rG9VDb{
zrkiakJs{g~V5!xoCQtRvxK0dC&e}~iG7wtKes~l~ld5}*4#O}}pPz^S7)8*NfOK=$
zt&IOA(ToF2%m{fpRo<}>c1yjT(@RVl)7xOuBuFbmW-6P_fk|gHGQF+ul!ZMnz;or+
zT6-lK+3l2KRw^(6Dt6S~R4~b?<xArYrlCnPgHGC;^P|^oUhh11oIxrVWmN1EuGm|$
zQLbR0q&Jju9<Cdk9z}HMAvgS3n>+T#6H-RTQG3#<*9b0KvBG{l@oCH8=oq||QxvHj
z#(C1bx~@GVlk~$V>43#-MwSi`lZ$|<!C&0ode^)2+Guzbm^)pT5@-NC$t6>y3CJ5<
zx&TMJt!qU|kwpN{Ik5Dl)D<jTGCHc<Gq<;CfgYmX37c<3R`DZ6v{J8qh_?2+jce?V
z_SuLVV2-g^FTj+%llv%f@w&4H1dhvepr{DZeuheorh<I0*VtW+7QCUHr27u2j)4U&
zTIvhnR4{wpUG?%&khzz`xJECV?ZZxeA`1i2mBDN`yQdDjJm6y6(Ip8)VwB|hto$Xs
z6)nrS&2a^$Dqvk|-Duz5mL0ziGl26P>y;u&-b@tw410ThMMs7FmiMAd4En0QHur#2
zuOoH)Hv5VC2C5}VxyZG0oWTRisfjV@(2=1KErI=H2b0JVsw4+iOS28D?oTZmImBb`
zL6|>aHJiKBX+{_qU&Y`G6~DJJoipQDI6~q;mku=5So1ri7`At`vu6$3Zwf(|rJ6l$
z_qA8I&Emi`<&?8a1O@l)o!O=N9T6mj8$5pczzT77PzB$nsp-Oze!A*YyUzAtTGZuo
zuX5_uxza&R+0V4jZCHP;Q;3<<kL3_`eD0+EY;Hxou){`AF%QxemtSE&SMSRRUH+)v
zU=i6!({g<H`L5HVluiA&U>c-e!yX%)l>qh&ZBjt%t&nWWdE}8ubV@fCY`1rH4fA=B
zk@L}k=?vO}7ESl3{bIeMlUgkOnYj}VEEP?hE5zQ*K#F&y{Ze&L8XP%XN=}wxOjq!c
zR?(>!(yiTUzub1;$cXctBwU^bycxU{&P%85SF$JbJI2vFk;k-6*_w?j?N{50((jSl
zl?4TmdjaKw=vlhh;I3=G)?s)h;!L#}5+pT&1R@|(dwfxli2ZsuJr5|~9X!oB2bQzA
za&l4$E~b07qZ(k_M5vBbL$UFk8l8Kgk5i@ryT4v-1P=cJcG0mpa+K!ey!-~+*|n-z
z?S1n{5F8pzs~|##n%8zUBp>G9Ri_l=U${eJci77xwA~F4#wSKeI_{L?&EmZxD??Bt
z+fy~=o)Q^OWxIILsmZ-~GdH~9-MqKW0@uGOM2*h^iXFKF%Dcz*<sfvsD<-6jn+KK(
zfVC6b29bY8Y=5pW0SknI&pO3ek33>iEs+d+IA+*^ni=t@QVfZbHJG%QQpiXeJlOSa
za$<TcZbt^B{Whjc>Ft5mIYeBN4vpU08*yM7A-8!>k^mlu)@n<$B-;(U%z-UJ)pWFy
zdg`=2)YYK#1(Q18OQ)u&sF=O*1F?tO%8}Zp@08+=bZ1Jo{!lw319gU{Mu+Gkz+40z
zV28PB?r`p+DI#^8=kiR>j1-|;kJC|l0mZ#*Vvcl!Iwcbl2>pNqOG+3_kz}wmn1?!d
zZm%6pc47e0n#UNw)JkL=#EU{vZTMQ)u{PsZ_6Au^<=E&zH%F`nNeb+E+t9102zGIb
zF>vV-L$3xOdZN98p?iALDb|cDv!3_|*=RVU?2$IF>~!UqP;zx($)%8zLa|}ewaPj9
zWcQVpFR9(?LQV1_+fjEwnP;~{@*b_%bAu!`ixa}BQIElDGCn1XW~b`ycC0H9j?8$V
z%NNxki0KTG8u@f$;}pSh<P&W}zJ^rQgLbA3jMJukSRGhQL)e`m!e>KE%Y~PAwhaxd
zft(j*=cnIjLc1CmJj`a{;2AY`u4@5tNo)tU17PBZmMvSd#2(8IL39t|@$D27KQSLk
zF|SmX=i7R(;>wENh(td?IyVTf1!t%2Hxg}bnZ4C{)VCkM50T_I8{Hn{1e)A7>j2{G
zGHpq{b`mAST<aj*l~P72m=##zUWQ<t<L$7^+T*P+lVn<%(xrtCD0j-5akluh&1Vlr
zdfO+Q=Op0$8+#-;m1J?Dy*{qg6oXDV59guYt@c}4kc``R3L<jcTuHo#y*tOara$gH
z)+8zaRh)PB+c_hV$#MIvQ;ZQLTIm(JT;fJ-ztfFZE^QJV&N{H1&JA(K{BFIXTc$#@
zhK&eVj#8mrP}ASbxh0f>gtH*N7hs?P?DyMV7tsPwuqW|XCLq~X&U;rM+7Hz$1<xMN
zWog%s+8^XRl6mb`=Q&4=j$NMiN_zQ+MTIJ@1z)K=Q#ub>FtY2gO?fLRPP7t+7rvjj
z2w@_(KWanYhEGM`DMmvf-O~_fyZv!Dmw5UD+ZWa+v>HxLd^{PCKzMBk_@!_Ww?C<0
z?wvad^$MRl1~1ncYEnq$y7uX<GvoHBjm<&JJC3fGD=g*m0Jt`zuKP|tK8}0)v#LRL
zgXkM;3r6Fi;?%?#A~cM_N=fkNIZXBsYUJ%sF=r*RzbBGphPtq`*k9BqL}#R0d6>fw
zn8?@4p`oeXO`}Ea{<3@V9GsBkk%JCQGT|G@ZU0yKI6XM&JjO_pbcwwcSBV3s{dIkV
z)Az#dG`45%5pOW9gVc2RP1)bHIYVC&4ln=;D-^7+!ILAjo0jCC<Dv+b?66alQ%)#D
z_{F`aZB9}GA_s`E1C%71V3~>WQ7)Coa`4`UBAaHgn>$5q%P{iYV()9)ch%ZWPC+9l
zzo5OpO$<Vj!qg;J0?Wmei<q!fVk#`pZu`4h55k{gM?VyYU6>!F*>LJ(q^QNhGOuWl
z4}8&thP@Nb^21I|A~7=}iwh1J_Q7sm!j%d0fD)W@PCGU6HaieITWFZ=L+uNsLYAma
zf(&o4Kqz_-p|lU*RhLD^-6FU(hF*HX_c+ZExwnIbaekrM<C{lW(vM^b8DEAUN;%3n
zdh_)9Xu;BnK7~MFxsun+q=)0ja^A04w#0c%qN)4r?c?3p!iykIp9T{ON>3*X+9&c?
zqqoj%9~^Q@(GLZ^%m0csu}>CE(f#NQF5HHX0<i4IFuCQ&af|;{y(9uA^9bIe4lD`h
zjnf=*$L!OsJiToSPShy}GnHnPFeEX#As;UoPV6%alt#`>sZ$Cs(lGEj^%}O)dyV$&
zv(*pn9$j6<-75~PD9K^T`lzO5rjyse=c*>{z#3YHr0jKSbeOK=pMAdVmJxbAwN=Ck
zP@J_7%<Z)=RFhh3`Hm<k0$4Ped|tRYRg-_8Z_DM642RsT15XdEh^YbtQy-I}i}alP
z?H^KTdEx7wC$&ow&k6g-j*=-JL^%g_6PeDE^mwN})-rBPq+beV{HL~VoYxcVa|aL|
z)$GgYF99_+hy8Q?9BC~iJ<(TS(NW3T3gJH&{$U}s&z<J;|I!U^aC~%&qMSw?So}tt
zhi)eO8-jkUxczH4c!CC4oir(s3PxmY{*Y6jSfsS1B*6Q(ZfYgS>g17!z-go>YH|at
zB~sJ=y-g3!fJC}OZ*|Cu1(rl}*t>ds6iE9P@)^dEaJD(c2K|p3zf`OG4Y#s7upCSH
zsEI$9;Y|iz?%_Zj`_B%a&ZP>2BR^#tYSgzGlyK-4|Es<&9z^CWl3+P7ZK;W=t;`*w
zlKpqC`qQ7|JZaStM*8+Y3Egs&^Azi=kH`w6WU~L8Zw?OkW}WBgte8D1$3j>VBgA0j
z)nRu>m%te@ew2~|A;-vT2&JTb$=&sZ1(T|hmIXug@Z3&s^?{{f_0?b$5hD@eVf)hV
z_lkkk{*X&*DpKIjz163WblmAgz3grhz0cT}Ri8xrlo<Rh+!(M7w;vlsI=~ZcViHs7
zm$%-FuG147UcJFG;96}Qvah&1(Ovz+ZX|I==s;hnso8uCR%fs5|Mr!42kPn2?~$3%
zDVJ%TYwyQL$!PsmZG@a_8e7H~Vx45Pr;tsk*RnI8$61}PZoi+IM5<A0CLl>C79nZ)
zklNRDBRU)_2C_S_SkX`%vf_#yk$PR8o~N;ZeQnhzSCdRdgyA#BedwI1*CD-p#WMT4
zyNiJLBJ1$10}wMlf}(iZzP@W5#~_A+HpGEdSL=?MprtZud_$W$4vpp~11xBJa(>C9
zi8LZzvIj+FG4|Ov-kprqh*%1=Cc*dtOtlvt4ahhnk!54nzUl7R1YAQr>^vuF)1qqH
zncQ!_J2?5#$>7hNV!Wcx6HzYJu{|lrh|oIJa|$t;dEzwwQf_!!fMfP8`GFy_-;Yj&
z5IXC?>Nml!h0Dxd<h6>)w{NXEN`^AnaC(Q1%&8F~tR2{V&HB~$<m?F6kp84osIwv*
zP8X8c*taGA5qUN?gLcQe@)q5z(YJ4}?xZ?7S6zS0yT)66c~i2p6-FVWOqg)&J8F@F
z^I_OhH)K-7MK)sRE$I`V_f-4Nn$u4HMP&v|1DoGLWZ^dZu0{BM!YL(9(PD!};+wv^
zMk}J%?k<(Nbf;63^a@;ULGn7rB~`=W+4m&Mu35M0+Ur)YS$DMqh|def#m+ni7JP5R
z>l?0Nss}bYfSBF%G-7r_rKP~e?`w+{JXfm0bg-RrYU0CisFL<|blCT|#-N>f&GC8f
zVC;Ue6kOL~KahKD^}w~xV+<%JoBrj=cAk?nh=cuLF7F1k>5`9~ax|OEr9JbQJ*ANh
zj5wo_786Je2zxKYD~vKUEpFziT^G-@Yaw9>fxwbc8=08FpxC#k)z52Y)4@*BUrtS~
zq;4Rxi*h@$F$o{-JH<j$WgFcmu|jkKX8h@GvLbP6>>US|!athpX+ozt&7M&oz{NzT
z6|94_w~DW%MqPX(Hx#8F>6s1fb-N^JPgyeh#(|e#r<=~6)uxGj(7Qvl+~>fu@7EAI
zW6!R(<h-r$QbnqVUyf6gOM(QV7z=qMh|S0(Y5VQs`epfVV12~<0!i_ZQ=>EZYH6aM
zliOD_GvZaiWVULfJvY0~&Fa`ksHCi}qp1lE+4HLF;(XF^;)nwiQRe1Zdwv`E)ucZm
zy&dI1i15mb-X6CX6ikW@>ACO%cVKEoXOeEfSyF=xfffvrOM|dUmO;#Z-s<s8$b*XM
z(Gi}LaNI7*Cgyj@Pk4|k=ZP)>mPg;v%+yvTkg%oMD0!@Kgq(JYg_eZRiN1h>$w)3s
zB42*PvWoP34p3c2gNU%ww@VwEqFSQCLGJ+laNzCRWo>T2PNnIioE_T2&WP*7mM0F}
zfMyEY6s1_{#Xc{R(zH*(J-fX9e#6G~*RNSUu-buWg=PD+E85`Jjgn==DW;}s*@`8$
zqQ0ytZ4n_lf)tf`J6>zl$0>XHCHBI6ymD&Lc}zM4&!K>TOy?yP>x=TGD=uH^Jm$VZ
zyhGsTN=N&lyWf2gy@Gs%*ur8e{saCd{z^T(kZp^1^XKRYLQ&DP7Z<wZf@lIB@ditC
zy1IVUUQ&Zfm$l@E58$#OQYWW$>T|lsw4=3TKiqv;^1(h%!G|4~w(`nIp&$pPAIY(;
zT6wMWSkzD~TE=t9>Ye^*582B4UfPt@gGU`$Q61%@PY+@paA)nxu5~#G#p&(9;%!Y8
zuyQq%-C$KWF(^Cmwj+-mprr8%%3zNlMBTc&!3blLmY2T{Oo$3rV8Pg816?l<%QlFd
z`hWxLsGV*=>~FEFx~|A$I=mTq8INkGCMngx>dh<d>iS%ss9MVd-e4LinG_H*cG@+q
z+2r*@k=DVWqMA^6Pg~Qz{`h?tZgWaWKj5b2@Fx?(ORHyKZ}_=Nlbr$)HNt=06=E-|
zUlwN$ULSQ}xe1mpS!&m|4NLl%kv`!j0CGM<T#UK1wfRLdz2iUG`jRIOf6v;w*4t}l
zwrp`q`aWj1(J5PB*uo@8Gsc#cQ*MMxGf6w0`WzZvlZ+)kbVGg)<P%+{lmsA5oy{c?
zT;@9@db5`|jNTY7nEV1r{6eM(sbUF_r{#T;<XDP5wlNu@XT89As)^kfu1?42_S&X~
ztHH`k*irb0Y>w3A-XXPpo&%ff_Xh|#kak+WVZekKk7L+Ffqz}MOeaLLVA@)iAZqm9
zM#bq(+4c3Sogmj4rxZINHH{)rZEkMpW<x3ljS7w_c1Rq8-s+{cYEe&#fVvTj>*HIG
z;bP=IiA~CZX=}VN44_BtmATqnYBB6QC$4Oo9J3#7vlUA(KIjn#mM^TmW|h6FI@dOM
zr^8_4mCK*{+SuY6a<8P%_^Z2NN%GQOQ8pWSMCsT&;nWMKb&T0-sv8U|kcmKK2Nu+I
zv2?6qV~PaLB+;d#hGgs4<_QEHz|e{|lyWZX;PfmqC<cz#wKIwoOY*X&tfiXYaVsa5
zy}p}5`G)RAQa$DkhN7B%xDQU{tPQsFk~BqVt*5r5w|0Pmap|WZmh_YjwS#7<^O$TN
z7jRFB1sZNmBR<{elwtxa$O%1{GOy~Mt&#i&$FTEQv)n?b&=hFrTd#ecy@}<f?*OVd
z;hxWNbi%e24Bj?OWH{+Nu)@a-qS3;}^4ZpUJB5xEmVqTsT#eOrTPwU2;!*`JW5S1N
zw9WFCsnHQqn^2C<z^8%Q5(D<z8@jQSrA{&tQp5qLCRdG{Ug2Zlh;M8y_0iQ+$eH4p
zjw_)%kD#%Jq)8E={G$K+;laan<E>uGZTX_>l<S<5IfJ%$-8>lX3&hDgu<Fd{;8y5v
z&&Kb5_mhP(Z=ZkL7yswf1$~lRZ7O`|0ef6}PC3m^bhBM;!-9ottamjawOC=I?~6H`
ztl!8>RmWg#pg3NrNxd5vN#6(q|HigrL5o8{<iKLKM4#nRBU+tI-&FNAWgpE6v43$c
zXQXwj7bh$_sas3MgG!&KoICxt4hZgY@YP?`T-aEsDcY+G<M%4zgiUv3K=~~9fwWSL
z$U9ej#xAEmv=Pa&Ei%=s&!C1#!a<v<HYJ*JD8|VWWX7q<xj@@Jg=fP}ZQMGGTE9WO
z!%j_|I!bbyMm=PrFwo&)VmEhKpa7f{Er8W7IR!Q$dS7@eJwjP#x7639dY9)~Bq*g#
zof)j8Q=k1Jrc>8n%(huW8=3xYEiizZB35HJ^(K_$>DDBT+<8ClJV&n=v$PFMgjsLu
z=3KaoL&MyD2PWz-pCKIV?alQ{WCdXmQkMdY&x-lP?_9KmA9q2z_LgokIAB@cQ}9Q?
zl4h<P#;5GZ8s$kBs7VJ9tFJXuwjgDt`0?DpIK6Ji`LZodXWvW5zT@T{5_Gv;WBD`n
zP%(_oTisMh9eq;#8!Uv>LPS{DTeE)~<hnfKl*%7Oadg3k<OYZZREj(up10LE;Q(P{
z<>xNau@A-qA`BUzUJ?TeZo55sjE&bjPchQ=QP~0SXp?ip==iu(O4jt6-r!BMc4xH`
z{$+e$_CJ0MG6k9PR^OsS(Z(>QF2kPP)n+ba4D}B!3`fNgr>5pa84$Fwbbe`j1zTlz
zcP)2P5D2R14>3-=occVSHX%IrkR9)7V}mY;Fe}ce!MOB1)Y;_1w71oJS^#Asz-z%9
ztS(Rqzo56*dtvYcD@eKF4JMPj%p(X+_7mMeLO;(biuljB%Y5S0$4=CpxT5PPMNLG>
z?I*JZoaA~WE5W5=bj*R}?&DG?ncRM=>n-YaCWs`Y4lK{K<(Dq8d%G8OQzyh0zn})o
z#N?W_{q~M-KE<pF2cPJ?N_{ST4-3+y-Pb;gbSI|>k|sDXwL3I3Jv(Oa?7l~)E2bhY
zN)BUwq=$6s3)SRRlf>|+Ybrz_HA#6NsZ^P*rZmgq&}x!?CL4t$oE>ut+Q?&EM={c~
zpKUS{tkHOnEL^h=D6yhHtLz+O_H)%uIo^nk%S)sl<7AL)_+w6eiNc3CmDMqA%FNKT
zPwECZc3t2DWwf8KZ|y^=+0u0FgEQA=Mos6`=Q$tLw1n%>qWwa*r3tI&`v@wur95(b
zVA2$Wr<>$>D?2|jO5V!tgZ8f6xT~*rp0t}s2Z!tz+d@sw^!u|;vF16YE__(*mpWMN
z#4t@Qz#fR5hV!%iayL1rLCsYANp*c&Alf0PUdZ_PJ6Wy!a3%khdS4<)vH@~ZIIKO3
zVsMXBF9caNF+n{1ueQ;^a*ks>@&aIqO6M;7wOq|BN2j;LB~g+}c=Idm*V_o;h<hVW
zF&P56v)AOsSztTbMi?OUfkF_JMD4WAwcK~CyWO8t_?qXx%6TjbtD{!Vd*x-vLPV~l
z5axE~4^~`yiSrnftg}JRcR1VHt`3=XU1XDc6zfKfw$jr=hS_Div-oW8XHb|CMJ?rU
zZ`$^>k%j-P)SRG_bm>k_G#E-)8l+dYw}UB&LPFIsZ?GT?pwuY{+`fi}u$%G(6f+_v
z-9e`&rYLqJHDW3Qb>sYa(Drvj%54xC@1!SpYSif~mS18AsvGb~95(v9%Yh~4Zkr%;
z2l<r`=1cQCrg~d+PpR%f0Hy_g%?s>-?h^`*kul?#11nt91rq1&!Cc#I6VmKaQ1h2B
zUuq8}($Fw0gHuX!IfTz3<nrOHl>%`R_sm7}izY3Ivhek=9jY2_QkcMq<J~7KrMDU}
z1I?BYLa!a}z6SYb_A+c<m$pz-=)rw5&|r*FcBEiUbirlJ8!U2b6n#6|rj9g<igR}W
zg~yoRKzA)d;jxZ-JJz_a7GL8)VuMKH5m}Y(Id*{^uTG$&L&%i@<{@veux|rXkjshW
z+m>IpWQp^fsK@E1O%7vYk5r%XE<fyVkcdSg;!b^0(FKuNJ!rlaaRjls2-~!i3lk%j
zg;w@?gTbrE7AJMAJ=)Df$K27|qY;&+@K5U1PYiE31@z^|&aLqEo7Bw2QpZf_+{N+I
zsgJSAYIqEjt^)W@w^L-4*+<ic6A7SL0)k>fqn{B3ie(Q%ZD%^{k(M7<0RBeK45vnj
zx^;5K&Nft>ls_M(XgN8r2o|(+?F~%9V@@gFh*^q#wcj3Ved}WhOYJ`(9f9byh41IP
zA1UBRSMKfd21B(4RaY3wF8huA6^xP+!cvN+<P{>?&1q}D*=S<yk`VD2lT>6GK@lvA
z^H!rEq4JTJRHB`x6!*_Yx(9TrjZ>7W<nh~4gg)M?6nfovirI%qJgp<~n9V-{ru!=&
zb70!xGh<jHWFojw@919YbC6WRZ@j@4i#fm5C<A#cup&C;Y7h4v`h1S1e|N#;{a=}y
z99XV>Y`kI!IHWnMIrfj(Z|4dVO6NS*C#6C{`-KC<@8l@^lf+x&0Y@=jfdPfqY8xE4
z-)(dx1W>6iB5(vMOHB$?WRx7=s`dT74$}x-q6hf<vdX$Rn;PK(%_^zd?^lUzpL0i?
zQl6Ny!GWKO%SaW;AG8*RX7JD4#swchqKA73_Yq%skS+R$`L+2SkH3RnQK~m2a)ukE
z{ZX4;E}PZOZIt6g4J_d+(r_VwivTQgyZYnmrQ(s#I;HNCgPVHequ94_WBHSOXBd}3
zh#{qHyRZ_=t^H}inmSZo48X*BpxU~*Z-3S;VV8oMZpcwDD1uWTAEgAn42N&~^KPdp
zf)owegaazF`dGwJ=QR@wsr!pI`VA9Y6W({qdHmsBFoOS;{bk}kinD3wIl1+b(Od1W
z5<S2_td!jMUOh?f0os5I?O%7z#;tz21*H>mYQ$II?n4Y8cfPkDhYdayAr7np^pxEj
zvWR`w{<h|0u_UhT+MIf+kNMESg<kps7Cd`Tn?sgQx@yhjA~zhTJ{>B^+%(-A-Zm~@
zxmkN}4z@Rb>uAp@XzLP2f>#Zig!kpF%<p*iMb5LvGTbBj7(9P}bs=d<NVpEHEKiIh
z3`F*K^_q@yWfp)Rb6~Y$rJ|LdUDSSojr%}btQ&M!a?dH}XGu%1ltWyi{=r<oVQz;A
zUAhAh-FnMg-tuhwP_?NqD^z?ZTIvQPxerPP5zABd;aWH8l@w~0>F9B^acW|#5)a5t
z2p?(4f)7Ysa|s}(VgkE3v-Ht!Dnez<`X6;*`oiG$N&M^VV-2HLPmDQ*SUgcGIRU_r
z*vG3YnVdBFyrkdM)P7P<4eiZG?GrVDQUOXkE#==LC8JZX{*4m+Mi1?i_3>n<(nV0%
zwBM;o<ipU6S~bp6`&6|kdVRL&HL#>}90aT^=H*Y<N8@4$wyA=)f@yGS)cX96>sh`*
z`%KqaCF7MGE`@^(m?8&{IrXt=D_tL^EqNeHef!z=p}{dkMX%yjJ%VE496pz1eFK67
zk6QfEY53UBS97Iw6EB<`I(;O^pHr{-RIHTzd5DafvM*E@<Ov4CC>{<XL!L3GCT1U1
zK!^-n`DlM%u>G7aq@H8sM;(}Y@CNOSe`x3nZ6R5-Q&!z2D=uAT|5(4>$N7W-2E;q!
z3o!L{FEIya?Vq|y%T0fhEt1TUrslL$AN^7~B78+JC@JbcH#EJuHx@o30OF<lIIiwP
ze$}-_|56_gEsBU*NY8b%As9iu*r7gX|F6mY-?Y_vO6e=49vO^7KM>?dx`Mz|`?s3K
z<j|#FWw|-)c6zH>9H|gZ+rQ^rO@xI5dp>2&)k@r&?LX>!fvVBf)H)q<YGPo_1v3^-
z3-+J&cC-6r9uOfpH9>A<ZI}I1S&M01W<9CR?Z4VD;k%R<Bf>r90_)Uh6AxfwgHrx?
zTi0!)aU(nBKwS18a}ZJ>-&L;4|H)Cn{ET@$vm4gi|JFPP<%cjS%}k!ZX+jX+cFK7x
zMz<fCUALg%ePnh|EXzuQae0hp^M^U5BELuMOYX^mvadoRuwe9HxN`@c`uK3@hUhA&
zWnWrflej6ZMyKm?YJ>@w;lS`^^+Vk-DYG9hIk4)+De<h#tyIEffU+;YrzAYGD$I~p
zhjL+$B7Rd8i8u9r<Ur>-HT#P8J;YPxu-96t+xTR*E+P;p+)pZV(ZOBzmD$B9jBGsW
z%pTr0L$0c?>Lx<DJDT%hb8R`})a&J!%h$-hI=d@}LHHS*F{?Q-k8HY7Y$ZMVnl>=*
z#uTeVS;+eMyKvw5+EzzA%P?0HQ?BU&G3WyFPi>R0tIkC%qt5ONrJBupZ}oY0r}UxC
zqSIX2@xDG=x(W3?PE|dptX_;t{r(O2<hi92?GdMxe9W??E9@KF;zvH|pg#^Q&oSNp
z(zkDFJ!9G{6)O|1q${+{LbQ67@TEr1#3@f=KrF{$`{rz&d?g$^PARZWecuI9C5*tH
zRI`%qzxkcal-X+XAd#pp&u`ra`7QZXE@CF)%qbNPOAQc~6Z_Wss^&*L#?By?=xd#t
zcsu;jf>7+TCl@S||FAMj*4<zljhv*9;z#gpi&~P6<~W=SFijY0yl~|D_L{Jy0GZjQ
zEtSHfs4uOR^h9~_R&HA;0BS{W8EWwz`B5|=iX?Z6$sNH*NJcq>2LT`L$klh=Q>^PC
zEbkCbSpf17=eV6=B^63y-<4Rgf!hV0=S37lH)`sNC2X<w-Gwjc6y^>&rI;;oox1T!
zcUpW;ThkjRU|qUWf7{H6eQ%pdjx1T{q*nr#ZGqRfv%xccn{?;jcTaI~-1!6#4#qj4
zn4b*A7piGVWwg@q)HclbCmIh<j>6(oi0?}aD}e<<z+ckI+Ye+<uUfxx?aIwgDS1XN
zXd_;0kdT2Nyr-!=@bqT0i3O2PF7KOCyoCG&r8s`rd3Vv2rN&Jz5PR(@HRxPt5N7_Q
z1B<uQCxP<_H;vGh+f&;V1p!F6NMN}wUk0cDw0a@0I1%3OabUqaPR4b%7u{2=7s3DX
zS;7l55oTR)RL7=0y}dBfn;awld&Gg%;9>u!X2S8ZVNG0pK=}7Fl6GFc;tG4_69y~O
zoNTfC=t-+#u*hOT1^7#<H+xq7M(P#uYw{P-lsx3rh#svT+&*s4&PFa>cDeJIqrMjG
zytrZ2wd*&oc1kgQd^tQ3b<s?C1lV)h#42+I*d)!pcKLp9H5s#m=1ST!BInN8bK7)~
zGhl31Xa+=0EALR>o|oObd2npRdDcX;=A7X6{MMNKkZDX@j#%bXPL19pTBo;d%3e@Q
z7TQPf^)L~=z!u}nf?n8`)OJq17U7eVa;J3jR<BS%wMPys4IW-U&H0zK-^X<aky?k}
zchM0}{o<Cf&_rHA6dOz1?&2)fR*^YM1Wszj&d1z&D@i0Lsuq%5leVnGpDkPPauHPm
z)MU1>K@;c0%Xtr-?l!wLr}Z-FBYBP+k|I?xZ^?oqp8}<WQEucs(^5JUo=NtSSaVsM
z*h(Fvbp|GiyAjmJB4Nw3*uvf^#^9wV6+8VfXHnpXU7n}p<M(osy#vZkS*q-pt+6X|
zF%ij?T8UDTAT3PbHh)tAuhhLk+HFPkEW+=oKa0Y@!S*XgHqM7L_QHk9#D_sl%wF$c
zb+(*=<$O`Sr#X;UuH4OmrDA-QGKTc+hZ-K=fHMyXvic50JL|dpvlr*iTD5MS^H^*M
znJ^osPWCVjQ;J1nKw7Q^SB7FmemH-L@qf&DE-rZ?{gt>hrInFm&yQ4>^N^L|<3VpQ
znO8V4_&yGt^&J^u6{)NU8+rB<KrUB0Z}lxA<)C!5kts4I;VW~6O9kN+#f74rvsLxw
zMGmPs4!#eo>(m!AEZkU<F7d1DmqIojrWt|M`E)t;A`pYy$suh6^)erY8tK#F)4~8^
zwVZmrymCyj?pJjud?(O?%#oHNw4`}KHgCn;O0(PMq;AMkf?&J4qeI7W9;p)X9)z0P
zH4UXAzl4tF#~fJFqbsqY+M0$qTwFc{TS~b*#$_!+GPjpj3!*aToKn1zF@cc)#9mgj
zmsg>L5rF}7ZJ<+gF#~2Ql-Frr2Jk@_#h<q%FKl7PCjA5%E$rI-e!FcmB8!2k&jagL
zUU}_+t<7cKggTSQR2OjL9Wh+%y55uRkQTVq+=1oR08<GNu=O>cY#f+pZ6B^3FH1JJ
zZ#wlkK71_pYP&+9HZ&Z#phSG>L<2rZCBZ08JN2pIfU}>9#G5Z~RFF)&P#;F&fV9=9
zpSF$l!N$;aH%meb;6R=F=tL^WWU4)GoAURAoYT&;1~ocWw$;tG0HTiK^%X`M))lAb
zA}v6=Y{YhCqQ$thw&m#@9#wW-TSiXt-7J>O@)2e3)YJx#pf@H>%XI+4AUd_nuFnP9
zf*KrhKq;qVNRdx+YrG-x9#{Bn(@sI1Cy_^Afg$#a)-3zI+>*3i=5{(YTHeTSgtGXR
z`4!Sb^(UM{>{t<8N#+rUAuiB<G|BbFPxX3QKVYw_w{y@i>2Xq?Lrc?YBD#aA3pE`2
zhm-c|*4%ZQuW|}SAR?z-#Cm4Nwb#^Iakm>sWg;E0dRzAXeelbo1}6v~d~M^U)=y7Q
zY{zj5fDV_HH`wdi&Kw~(mQxB8NbUN5q6r_c*B4Bj1qLnj?hiY#ID|wk<!vO!Upzp|
z&l!xD4b}pRPNtO@H;-`Mje4g(T5IX7HIGAG53mw2Fpggbmefd8vt~=&3BpD=xeT}O
zaAeEzMgCa8>Nf|0={q5muxjZTJkoaPR_^RYh^8ze`Lgkeo5!u!eeh8_A%)>F2Ntu-
zwynXCR9Sw|wj~jUkoYjJQq-65ykEK2wzla->(_HiF<>1$I{jgqY+D;J_?`LW_j?Dc
ziRE5m8?8N6M<X@5ClGH;`X~R96wpE$<h&d0NJ%3W3SSd@LoHb%n_w-CwbTtV%RB^P
zwzm>xpLqOGV*f#C?Z!5$#~++O@03Ji-uya}B93JZ@15Xll)CLA<9H{T+xGm;mL66~
zo=c}YGZY2|CFL3Kh9JF4+z{vuCLN-DNpOm{iS{1UTtYj7odXl)o*?9%oj%!CdaWXY
z2Ay&?$F&bl8*gvy_Kvje5_@zINu@?jQhNyM@HU&O_L5IvNU(ejCJ{hadnngQzdo2d
zN&K@y_|Y~wHQIXvGd7)*aLwS@7U#L9N*)u7&MBM84%{UFZCqR^)-I38D6G{A4U+cx
zrmjf|0^lk1QpfKBrzYsJ)Q7V4<K)QS#diq__oV6GZ#UOhGz=x2&^0rWxzVXf6BJh<
z6W%myx8#{PG%?~lYL@zNf+YO1*<69TagLG@gb3JayS3&oDW8r8E|$-?=XVT@z(2gH
zt=HAT$vfph=5#QC)Zr%e<~G!j6jDafoIU8&<l9hFrOuQf|CZKqiKhsh4j_-`D|^H%
zjoFX2vjaDhnFzB1qArsktY<%--DAesHBKpxhe@uZAmYSZPhzWTQp@wr7it768V|iZ
zVQ=klNO?gsr*nhtnO7zejQ9Aj-<JJe1)G;H;{c*p*gov`YFJ38eJPqRGy6%WCZ-^a
z8%%*BmhWg&e94NX%bZf-knRFiprO1Og+q7NZ>8v$x(5p5z(mlL3uCL@)%{ebfh@J6
zt5UX6lanfAI8y%I^(HIEh>{hE!^^3O4vM@;8KNN2d)i@-fiMMg&Jtb3Mm<>K)CUST
zggge?Yk;9-)C54#GC*X%y)E%_<?2<=V|EHV=@53Ux3^<~9j@$of!$%-*{A$|PJPY{
z#6>dwemcOS!jAn!!x1j?#@z@t7LZbw#bp#sf}#_2$#4V)^K1NS>?hlCFf=jb6yj8p
zdR1Z+6pX*hwj_!Eu>DjuzZdM)MwP>$#!r&Wd*1yF7U6mSzy8_1iKL3)AlHCXjAlVX
zup}E7LaGa-KDKwX1!29_?0Mu*V98Hy><!CoIBWMM4h@e^4bN<MO7YFbG2?*<j2lyt
zy|c|ccyoGe+9`|dJ@@JM(+w9lO$?7ZCE;|xH`}+L$%UI5yv2FUQ*_y-m)p;_wjdWG
zu%~5;3ie&db(>Qk2&FOUuu#^lpG(;M4xB{B`{?x%G=S-&gSSEl?C0B3quG>`<ni!s
zJ~?;4Q==}84-bvnFXS_tlo98-kR!Gn%3iQn`ZRl2bwF~VA{<GN;SC0o>8KTpZNJ#P
zcaWX0t-0TU#YFB8eTkyYWliX~Qv0Qxt}7=<uG@D?xp%@-F)lB+Uv4lkMZVL@(+Mn(
zt$|Rv{YqPnmE$95<(zVThTKhb<OCLhU(KIjS3s<}(J7`*4(l9OV{6KOExR(*!^pkG
zDb+V*x)Sqz(tf>75eK6Z@g(WP5yq*}q`ruMwxfQjY`Ib%p~H4yAvcp*>;9^4r{FoP
zH^5?+q<9ezC^;^<&oCa@&aN|&L{^G52NvrUGAO@kd2SBduC9jL2^oG~8=*KyYGP0%
z#SMjs{F-;S6=Dq;6nj&C@CQ?^gkptyb$#RdO}3}&26GB!MUC#jaw$f7TgOIiZ?z2@
z8Mg57)CZ;aIyKSuJ}^N}b|g)<ecjufWWc1Fij?WzYJ#Q)(nnP9{7&27zQgo}?<4aE
zEaoB^yR;bN=nc&|&}J5VmR(GQ5gs`<n$$ehCz%7gWa7~L=4qBpWS+4Ha}HKck2#Np
z7yCI~prP={4-^db0%kX596sW}k~U9Hk&<@A9&BCgxMPXB@^+=BzL=RhkZkOs>h>bl
zWSEteqvHU-p}r<l*^x7e!^72r5JE~vU1@Lt1q0W|*VmGQiGIr!>a4=(dH0+h%HeI;
zw90u5A|i{L9qzs;QZCA>G&M)&A9ZSSbZi#*Ogqx(X~+p^xE)BKbOk0vJ`Pw!IReQC
z-;U-N&<5^({NA|-opOvwUfoJicId+GcC5`Alv|mcm82kL@K&SU6$;!C(d34Ays<+g
z82;GS4nX+3eCbj<@r3d03#ZkLbD$<#@8*tKdE}mQuZpX+@E#b~tt3Fjej{aThE6sz
zvZ1*zIH2Uf^$Kc=@T+|Eo_A+<&XwGONg_AMb~s7`uT6Z2nTb)y3AM*gm70eGvJ*UI
z5|JqbTaZu{#vn;WXhof0sAuy#Ubb%i4R*RAeh9Y73nXF(TjSK{3J#1*^J-_>3c^d^
zu7J*=AaKLCaKR%N0@&GZjq-Fwzw5wa1bq8~l$y$3pa7HIcCPj31`PC0Dc;)yTOe!Y
zBf-$1wz9|CVYu{)rA{FhBAE<j33|xR7fddw${eBaU10x*qBwW|B4-x+jW$SSPG7m!
zDKDZEdk`XYHj@3(ezSFU^$eN<N`;K_2@=XbU~{c!!4>WW@g8$vNw;CLS~e%q4Y|0J
zQOO=}^|2XgDGYhIIkJrrYn^(vBQNj%D8~*Wkj>|Ga77AdI)&&+Q(`LK=t8%%WEe>d
zSQIkk)F*GVnrH{U(zoC0CM!)TI=9j;D{e@8>Ktl1-m1aKO;KtM`3Gf;k9fUzcPsd4
zky4$lvg=Z#({KIt0~_tPyWUBIpZ!h|Cf}t#7EUa)Tol5rBgY%wMVrGH6%hYBISJPh
zKsV(S^m%TM8M5E4PAEKqt)P>zI?-xmLKegSNN|_RIQx6mrFP<0?2H6{2=FPVK6!zm
zmQ%=nzm2twkleRGUnStS8kmu=l;n;)2fg@%Bs6b%@w1%A!eUNe`looaKWsfm1FT?f
zskZX6W@<!pQT1VG_@johSV|SbPdPDwu8N%|-M{^DwzT5%OPoiEty{Ql4obs0^C#W>
zOP!QFiZtcQhNiWmKB>go-uR6DX={yp5vFGPPB~hG-s&Wi#l-%sV3GX+ykR)rVDDgF
z?&YAr7yrFKulJM#hYOIQQs=#3^E(I?B&`S<@xJ{<n_H$phP{WB3Q7#5Q&WervTWuJ
zDoMFACxz(BLHo=4eQx)nnH=noI5nEci4oMB_E$Mj8aTwKDaF8I#%T^OG5hO!L&@|W
za!UC|u;J2d?6JQo80y_X*3(#*fz`pLZ6uCCUCm6BNQ(BijsCA$KXA*i((K~i3sBVd
z>GjXvlPEoekAU;66%k>>u=C!wdspf9Z8W_OB#BmX4vQ%4`?{Gt6k?_PErjH|oEi;g
zeg|H_+#oV5D)9d7!*&wLj7sB8S)AK;SZ}P<_IKH(;X&dxd5$L2f(~iN!9;%#+6Quq
zaqN8Uo18*TYJ|?o<G?<cFRWR&df*jKA?8Y9t70AGQ4l)TurZys59P}P*AJ}Q>=eog
zcqWk9KAb|;Q(o*m<>3*woUm%}HC<Etk(xZUaLy@3hs1~>>R}i>Vjt~(8@JxU-<G`M
z)Z`>p2n0qw`&fNJflAV8;nv{5M1Sxn@7u?#UPoz~ppH7Q?D$m~sAHd~CdJE~XTVBX
z4Gm;JEYyV04m;&U@Ks0bliidngp*a2Q$#3`s8RFVdSlz|Q>`i8J)rD&0Md<^&gm{6
zt|%j0vwgZ@(J<E*lfk#=K=c&?|HkYyHHdJyqb*r>I~e;`yB9@;eYW09tDC%1?ss6Z
z6pO9Nsa^S8H)_eL!2_^JM^RZW)MOIAwC5t{+UMJ5SwBSb%SZw<>j1^EU2*wk_J#Up
z=Dt}Z68%54y?2;hS9SNl=$$}9NFxPf;sO}kvSgdCYRSlwkYo(lhFt01DQITyFn31M
zn3ybVAaTW{0jcVvW_7FBvYd>@aeD8)_e!tf_xY~1_n9%t?|I%o-dBcs&R+Z6bM{$#
z?X_3mPEYc&wWH%>_Fv_`+y_hI237IT`fShK(@bOy`c&aIlKEsu)~=T^4NA|G{e+#!
zxhpq^6;VA3Yr1_pk-L6itUG~$ePY~!<PsH&<x!->^7kx4&}Zh*3(PLq)RSQ-`7@8%
zXPZ50)|-$PAjv1&o&Kb3WC$S_)yqDYFiJuxC}4~1QVuLuoIa|Q8PUpx(T3n`pRYWR
z)%=jt62m-Cg``XD3zfERASFSPa|39TTZ?7UzSs=%u`mRv#<GGZYw2G4=b%9@RS81u
zL-wV7?WXP)PFZqVDSHx>-cOq5m&?axlzgxW2PV7|vCTQ$P%VeKex*JheNr)pnD&Ly
z=@U**PSPR{IT+{IwE_ET5&N&b(z#5De3-Jhd>9qiDA}3sBuGbw*pZD&>j5Syo!#s(
z|9_PKAB~;uYdK!(;}WFLE=fDtQOMwF>`j^1?#iB!$Qj%tm}$cb<+VrCt#jHV{84@i
zd&*sfeQ^GZ@)8_C5yWC6Wy=rY14%+Z-fvGW{nckrXb@<YYNc&-w>SNS;Y9r_Oy#La
zhFQs;);wQ6cTtA~1>scU#R;cBH2&_mJ-vKStZ>L~wulQov1`eeddi+rhlJQ$<2KTU
zC4<6tr$5M78mM$rsx#U%%d^3Rj1A%x{H2ne*PcCC!6f8K?2p~vo^@A|Z8tI#5wr%R
z$dg)nPjbdZP+$0r?Adn(p(CKKU*R-j=sUP!zda|1hDVFslPm?~z=VP8mfvR2Eh8_M
zXv_$wyuxI3k%gJW7JFWUDVcCGWk^wdk2*c^?b4s6tFm6~`2}C0e}i_}n8e)_K8?ZD
z>CfND<-;3@?X(xvk(A`&#}I3PiL2j!;U$;Z3o9K_d@a1f0D_KmP`gTipggdp><&vH
z4mSISyUMDOO3L96grNb-t*j5j=1KO$7nPUFzbgZ<oQ1&jQi>uSu@~1L;CNLOrUNK5
zon-YOshr<fU>NVPIb_@hu;c;jNZZ7A{Y`ad7q&MK_LS}lq^vSgT`Es$PJc=1K?%TE
z#adH8($O)mJ()wI_bIfHWn<r*16@PGMe3--DFMV+d>A{_OOre<z4X?LU*R<33&eQk
z65cU=+P<Z{v9S*@Q!E?;Xy}Q%r5z|>jNDQC)`EE6i*ul8QnNaa(_emb0E)m=_V}Il
zZ3&y%h9yBqd&?UJEH$}hL=(5cuzh>Q+Vx#xK3h&Za3noWF`vSOc+V00jwIm)S6ty-
z$~e7jxxFkYU&Z8MceFBptks+3{10Y9r;&YU8Jt8AElAntftA-_y5WNm7<d_$Jo~On
z$YG7kr96kDIQ{vKV5T31V}BIE;=6MkGuxn{V@@m3A>^uF)c)y%_B}Zxtjd)OokpPq
zC9q2>k{RG3>|`%5KTYb0b>wqkkITQoK`(YePBzx!A?Lb0P3#r^g$x#|p}nG+x3FbI
zJY7cBW$*<31r4*A#UXLng*n`n-7dlAG?KJ+ND|xTD--{(f30(sua0E8ssqBf?0cKF
zI;$3^ax6UW@!5W5_LN|6<_1|Hg7q${q=H3!km}1V2NGOv|Mc<4@34#OI_2yGRd;j%
z>Za;Dd$L_p(S0TRC5@PmeV7{Ma8Q-rOA~>Eo4T8w>v-~B^x9MQs%$On_2lrNRZyfV
z2<YN$fVnlkx{hFZaHpOFh}Ymw2&&aQz2>eb)-x)ddG>$%Kj+ychNc@Pu710$BAtC2
zyJX~e*MqD$eZ}d|iIgo7Yu{_@6WFOa(W=vSAZh)6D`85Lc6oVVRHXrP+PSM>$^Po}
z$0sYsNv7OA(^Gau;x8(2_MBFski!v?t`beGC-llIb1+OF41!nFPVTPL-gOUaD_5Q<
z;;_w22C$^n_Hx;#RN_^2qBpcQIgRo%<Y5MS4tXtY$@6ni1i|xNPAl=SVTe7Fu+Q2a
z>-1A+)M+;hSk^z>J9c&1movmwL{t)lS&>b0t8-UDh$`5k*ug!oDKPR%`LNiwyuc*t
zc7{iXy0WNWTXE0%kfch}464)?al~GiU*BrK*SU<ZEINb&g!Qdhv|e8~z%`K^PK{Fs
zROnY*6~XxkAW{3@-cV7p81dLkiw99f!rs{YYPQIwM45w5PZHfms@=u$9dD|`?%??D
zI?Z^YWC=Ok;uM_7w0LvHAj@3rRt1t`mkXv`7QS<rM3u|0oo{a`k4nQxEyY9fk<(KQ
zEAnfnNhR+RFE44VwlKet9Ct-y(~7AP7mzs+Kb0+NCRzj|8{9Dm79YSih9QUmHq5&z
ztSwG5AzM8S8MNbxQ{1C6Tx?dhq`*YE#1>#wJLeT9MxRx}58Bd-;mAag)uq(8aeCsZ
zaLX`%pnTf03d2^6d6O8!<6dC|4Mhm)BZySjl{LiMWxNK_GD9+Y{;C!}{UHYqP1y4C
zNZlm6rgY&7)8NxH^02Ka1I_tLnJ=Ukr$-OFX8x6SePWt=+3Q!Ya$1Gv!lK0fR}XW0
z>s_I_Xi3SCAbz~n0gC6K>C3caE1O@3R)okbF<6dcdZgW9SVnTVp^RT6ipUpG0!&MW
z454nXWvi+r&4z;n%j81(q0pwB8)i)BuH<+G7*P>$b<Q!h&m~p0opx}gRDeJXb~9-$
zX)BqsHMx(hZH;Vlu8W>CYdud9hM7ksdRsG#$I@{R%7K;N$X6Ivb|xbam&dwpYx9%F
z!VB)S%Hk4tEF%2lcOdbtE6<V4kw;PK?@4UJg6wnp>o1{<6z8>~Puu!3D(PB^$)aM6
zSD0uA%7$y4vm473;@ktMUBZ9V{c(B<=gx2n)r?LznO4SWdwZRTf_$;T5>W&tE5w0P
zfzDkqLaYReK*?hP`Ta4wsq~Zpxs)}O73sA{1S=akwHM1;eMk8!xx=Ks?3~``z{*^z
zNP0L4z~K%y>&Q8@cUF=jE<i<@u!K4=5&3+)ZSQJ+9^qloQ`DK1KBFjZr$5;NFB3&7
zuHxppvrvy}7$JssSsL<O)Cq*d?Uu?|+d8&pOKZ}B#Jvhrkg9T~(y!fG@7>tN?MB*(
z1BrLb5w4VNB}%@#u7-6TLMCa&9T9X52`fAEJ-MKwxP;S?{;+J}ik0?#<>ozb4UA{-
zs9g>$hJ)!EzLdu72w}2TekA2*NZx17(AGMg*|v;=s2lRJU1dXXove*V$YU$N>BfX9
z@&hX_*1qwM7Z?-zpo~836x0x{-@H_XFghqaKr}VCqbhdl*utKgr;y<)KticKxml(G
z6FCN6*;l4hv?GC8>En`c%Hg3-f9x^;GhQ|~eA#_ZSbvoQ+Z`+~xY_`uai&dd>QI-H
zB?!eBsCTXB^hdUQ2Ot&Zb+_7JnS&0e+c}M7MjRj5nY8!j3^Mmf5>6}8dKLA-Hx*QB
zg(^@R+V={J{Q>0Rc($SX=HZEPLa9d_Ku#lzMKTgKhl3i9*l^C&c>*S!>sTs6i_CMX
zPZlM%`1>fVY-CQq=veWox)+m+;-0<sV=alDWS`zuDj-Nb=$27PPl*D6@q9wdoSqB8
z4A0D@Min7XhL~JCtuV16*l6Nsko_3-k5*X&`(Yr2;A1R*Uqf+O7l0}A3QOiDd(TlD
zEB9pu5J`L|y~1FY;f*K0z<SMhKNdm+$FBq{Q&<Y$<@85xl6F-E3n~RWWgyu>6knmQ
zie<yBVow2UD$p=qS)?rFnvakaWdX{P4YIEtaPE@a8pZ0B%rH^t0}}|+J*~8L^mCr!
zhRKd?PQVDY1j7#G08)6{OeifzlAdiT@0BMWO9qTy3D`}9F_0aI4cwJaZch-N#n}W~
z8+2)ilFxVfEG(<?i~UVjl3lxWz0-*K5uM1&bcU_rwt~-Gp2a4@rjTb!6KvX0tE7~u
zs3SUq0^eVLIoD{&yoKgP>1C(C@H?ukF+@4VzrQ>UXFMW_k~EZwh!e-@NsVB*+nylo
z=k|&z!X)C%2nSN07ism_$PU^E%Ja0vDkxA<B~ka=>503UX=qUl=P~<%`i?ae-6zwY
zc44TBeAp9l1fsuv<~grJxVt$skyhN7VkV+QtQABA`(VL=tTaTH>vUm#X<V~Hoc`n(
zH*{~aJDSGIb>1}Aa=y@=Ww44-kX%C$kOR|CTO0>GZXYTm2+Z>aMYG4*I!;e~XG&KQ
zd2AnUzFl>ISqJ!ufHnLC(|uSxBz{PESNXM%l;5O^Db`JhY6DXSobY<Uez1XUXA4hD
zG_WFm=$xzM4>C3AJ;WB-59M!gw|9po@K6%@;lPs7QgQ`F_^$HUbhl_=XB=3fozGQ}
zxyr7E%wTudiGz%^(Vvst6jcl8Ns;&}yC>(6BUDGIWk^daXcGww1QYdl?B0T3ofP0S
z0*f^Un7On28jMM04$KDL*xivl_Bj2CdG<)flpjR-xW8foGlf)~$S}juxg9QoDDLzZ
zwP`juY);iSK1ICwqZKzjsx)qLn(2#B&YaZcF?*oAr|>t#Rf&S)ctIOZe~gV2v5D-_
z-~INnDo8DGE(>PGJ+S%c=di~!_QM4$l=kX(8gZL=y{wMfUF6}LvL9)_GdX1D>ch#Q
z5*^k{fBZ5x$7+d}kL*Wtk5jUzbDcCmZ7tl%KUM~v(RNWO)q&-MGRP-HzKGznAJ36S
zIR>W@-@AJn=?YvE5NIE7K2qU|d)bXx3$aJgqq}FeQ9ox`9&v*6KT*C&7GB{};fiy$
z6#8Qz!8J9rAwt7j?30Pt%NHzgE^`=6MKMm;Pt+eGA&87n<M=BlX6>meQEC(_i2IWj
z6PZ5Bc5uckOwmeV5@9j^RJm`8_o*<#K2$JZ1@#Kbo5G;2BDE4nf4Z{2<=7iq15P{9
zzY>Ld#D1m@K0=SeSnI$Fy{3A8CvN*tXCNA#@18rs&z3zmWP<1yIb|H35=?<dZiVvF
zatD#3aK`za)+8qsZdJ_Rq;1PNU8z#5PD4w%&9Rf)bL8ET2^~Z^=Cma#ldm#jJDQ2p
zt1?zA2Pe^Lztf*8DJAdjw4LSdxdK)3WX0X|i|fi)F>brc*H@)bRUC7&*zZI4*o7gL
zy8_{oFXT{@)kQ0RukB6*Qx2I(oQ~5B>R_WmoKar+G27Gp_+*4)KNO_OwbPS#BeW@J
z3|3-Ovv-HpJQk$1F|R#(y%Om|bC)67_BPLXSUMML{Xqv-a6;^Y?Q14}yUf?n=0UG8
z#nfwy=DaZ5U+IUeBMyId+M0zmD_1PG19iKmA|Uo(bPymZhR<~c9ISE-)%ZBSr|kfe
z1_>G&YTJW_{c!N*h|?&miIVFu-8-ju+o1{zK?4Eead%)TI^cxaKzhpII>#Z#FfDA#
zbM*@P>vJW!M=_8i`7JB(=Q-E0^Wuu{+0h)}Xlqicgwss#Lr+END8+$1He|=j8*2gg
zW9(EZ3dNFe`h!F&@nmYJO0FNT-@mFgPGk<vx=Rx*l3YfXAxc0g_rgvz{VHpk{TYHR
zV&CcXYt5(<)kFDY+&*1)45z7{`B%@glXpEaz~%o7llaX4Vq|yTm-E9ZUj$Owsl@fW
z`R826T<l|$l%3-!^mw}9UtV-CjhSupu32Db%H0xQB=strtim!rPaF7@JzVZn?jlKD
zxa)Q~uoR)!kq(5mVvi((CVDE2Hs&;QO<aevZfA=UG{?)0gfW#$GqaoYB<1O(b8lIx
zr^30cLu4K;GtRUy0tvt2-ec|2qg8`W(6e*pE{RrJ<K^t{>AhZiGTc!UJ;ikENTkP_
zw;c#urff?UCMAL@ZESys?C0t#V8fQJ1#06B2VvlG?ewQqF`#;Hs+49wUmmUaBKg!0
zAvXeRxLZ~{C^kHs&JUsZO3%yn`Tel8^Yzn}Ar*s;91~>x+M^vr`E!a3ie+Lm&6|Z{
z77a2)g~@QL<Y681{CK%fr;lq}W1sg5Q;8}`#Rq3j+Aq|P4}{3vj(m?y7P=LNO!_l=
zdU%2)M-G!jCKXLQkye3)?{cn^9LUidw_hyptr|4iiv!a{Uc*28rR<NQbM>fgU|MlQ
z9Bu#0g^6Hbo#2!gnyCenCK8o~>{l9`QSpdxRibqKb$TjCz+N>a&vW3kT-&dfM{|0q
zPAsOGT$hrr{n&z#t9tfp6&oP~Nt4a=;*yc;wI59w%_WI7_h}XyT<6#8(!pgVbC)*&
z&Z{J5qGP{N@b)(6O3afmoPlxs&GK1Qbd-@IY&TAiZev+rxxjv_jBjq?2Q8dZqsr-z
z1%@}4Ei5JN-!3pM(Bgn_U@7`U5O3BFimlLor`)Z9NGq9|;S~Ai35qZYBa{7ZfoT)e
z+mAxEz%-BD(XRbo<zx{bHSDy@%2!1}k@YeA{fewns&I~MoOA#=+i^Cdj8T73r7FHr
zwJJ3qbA>>S@33=MPUQxCNkg4JDiXKt4-2Xk?EEC1h=s)ADNcWkC_gk*TKl6Xebm0`
zoy3V(n08eK&nw{Mk1P5xFcJR1*MyUZ6iIq=C`7a{{}39R-k&61+MC<XWqu;=@KSZO
zE+0zGPx_zctrQluKdXa5ozquis&HUhYdV02^8x<68HP$dOOjGjj+~wJXeMqZar!Uv
zl6X1iGG{&bjnoEtP85#vm$^bOzx=BC&ULQmWka2<_E!y>YDJ>c4osp*l!$-u_4e1<
zY;DBH28W|LXA4jC2K%?aDLvv^ArV7iIbM2V?P7yt)k>$1K)cWk5#8)>D=tu7wMp>T
zffasx7~y0e86}EOcohBZ?@EtsC#3C9V`LO|5dP>1n^)Si#J81=-AE03vo<c6zbf>E
zV@jKX{e8uW+=$+^C^9x`|4_!pa)l@CMcUut6{c-Q9GNe+e{4pj<XkO*$g{_APfsq6
zx(O3B@K3qy5wG`;I1Nq78nWk5Y5%!$gx;2}(+DgG&-Lt%SRwW=Ij3aKY;vwI)QqJS
zKSCj8rI5BwOta!(6&n^O_u9YabW(eV(}pLVW=;)Towl_MmzoWs`?q`z#}M~>xu!->
z&;GsnT&cWC1~OwEbb8`*Mbc<^3gol@D0hpI)GWL-chQslZ~m2++ke)dGIs51VA6r5
z(9|1@5~lxE=V}Q7;7%)FhI<Bs(KP9i_NiuQ5)2eDplqz|(+9lvBu(tb8lWHo25O%!
z&ymGZLFcCsrz?z|S>fpP9JbFCbY=3K)+8z@tV1qS<Wi@<e7x#6>)q$<vvtJFRJDLs
z+#$VEqRvi6{JDlzlF=p-1Ry?wAEQg;u}(+-e9psV7hH0ga~+S!Y2BwlHv2+>$xkBx
zG2wj<EV)r{VniPDj(xEVzYu?vtPg`svR$WNTRE{oJ^NB}7Ahz^mvi=BejNO$38%F$
z*B2?aYIFek9f)`chTZW_t8HJ&=eCpx!F7y2iB+{fkJ?u&6)ak@&}kGtrs|%0n^rM0
zlKpqRtu@+p8f8!lHp)RM1W=+4hEe-kmFo!TCKDcGmjlwPvA<0c>OjsJmTS8^7T@`U
z(7_I;Ro*R`O^yp~)SRhhPbv3Nrj$L2R@t`aU29Km8dqQKG|J>q03H!^2OG-M?k*xG
z`LGhSaLxcq{b`sA)x^%G@#X3Hl9_Gi@dbL*8g6lb)idtS?<M&o^n0CNEkmz{siU~)
z*QPMoo3LlzUDu2}Ega-t#sr^&SLTv@Rvq**9DFpyfk4c_*q}YTyh%u}RitteC=%d=
z(;uHGcbF1Pa8cNE%5PGn%`=FjCKRJX=6Cw_21a~n`h-3A?$ANTJ6p%M47MqO51?@S
zjI&ejQu4Xy-5r!fzD1;@(=7Z-7UHZm$J~QzY5Wh_^P5ji0Y_1lb767|;a#>Dloux5
zWUbt=mN-30B-gh#x1gQD(Q$iW#eK=i#=aDrNfZnv5S9Kg4;8&Of|qe5AUY{kSsV}c
zZQsxkS9;2#h=4Y0rzho#fk`65H)hR&7bVJwK3Dh*t-z2BzZ6j5BP0PT3G2mo$D+ks
zw%559U3TI;f^V#-TcqqDr<v1+OrQ)Towxl>bx0BF6!D`hpGSETJ%Q-VnirZ>PA@eT
z-fb@_6DpJ2qfRTi^g1M>!4doBf{>gb8KX*^F}u(XxMgIjI(Ow{X6OmFMi~pGpeZzO
zr@gc;lE}qZV3?RA(7*(0BGV4{g};jv@3n7f#w@)`Qi+a;XbP4&{l$u>!ez;TAZ%%0
z;e(VEalpPc@wo)Yr*oa(MP#3Hz`m_6ly;Y79$Lu-*5kYQ_KFY$x|2>TkWGj+OiVK5
z8T*c}8(vbVau(AJNPqkW2V01GtTlUC^RaB@+?V>Z(n{!wj}dKYdm>Q|VMykXxv8Sx
zdG`}9jg`&rIWf`=9XG7KWG_uRz2%K3m1T*)<&=F_GxA4udKNP99S$sQ20G9TS>C?8
z+_V?_pX3}QJ_nZAxMB4Y`<_aBBuT;aX=?emlp@f!mls5&KT_T#dDIk=lpclJAIm0@
z!(aK`V#l=$>T-dXQevVP0oDNZ`rB`RHDd8A%J|YI6tU|LEa!5P<$$I8!ukyg1LfoS
z3I`Un#WfTwO<6$)q_f*A^Vhu&TZyQUc5be#h}2WYv3+lYI9Yt9n4EB6d6!DFg7>yv
z)DVFRU{;IjY(ZA~g(<?431F~`>(0@nJbMEo2bz)^El$zAB&VmpGrkT7D}w`|c~ya8
zxtb!H>C(EGR6RG-q`4DlPTIOvYwcCd7swwO>7HI;MKWUdqw;8OHRGL4&Sla$k{wxH
zNPxTSHT9$MNo>&o9Y`$nIKVK-Mddhhkt+A?vN|=&MuAn)PUgcBSKuT^1jqe~#*~x#
zwWVK#7w6y-!<A8aQJ>rCk7>;0igZVY@xv~!J;9?zR5jd>o@7Q5L>WE=-lh_*($sg`
z6|-LoxysCV+AAz7o&np<D!;P4p(?VC1Kn$n$V!nwCwPu_Eh0+u`=faM0|Y(L4y@{o
zS5kS!L+q+LzKB#(gtKb+$`C-kWt1QK<2S+<wI1L?kJ$Y38+~V*pmgr!DR3I2bNb6~
zvVJ%=b6ZegQr1+j0@tVmOZlK5f0bQb=a+K;A$>Nd0wL3N@U|#`XV=vA-$vQ(dDCPo
zq7?bH&HPF2)FP0piSq#XA}Rwp{gOk;o;ql+tMBGu;1Q=)7btU{A+R~KufO|=%-6#n
z&4yVlSq@l|C<(|zItAnnbwRbGX1fLtNNcpafgD4uII!|JR&pFbLEAOvv?W???%ZZ?
zYTjSP(&}oYCvX{d9Y&_pSl(Rj)^@E8r-`g{dgAU3aY#|;E#>Z%>-RE)7#|7K*yHqQ
z0Jq$Fvn?z)Jvl0coM5LX9*Gr2@g3}=M{H3pZ{kuA%1dJz=?LStIPad>HV{GM7|7ZV
zEV&itxN*#xJzG+z5u|biB+a_#uxn+QL4+3bx~3GrEp4bS<1H115w=+FQpO$qWe!z2
zwW{MJ{MfPv13S=0r9Mfr*KS9TFtF~%MRr}C!od*=3T=?J50L2PmfPR+ep{Z`Iz`C2
z&V4Hn(CXNVBqm7i!RbRzOSA!-Bs*-<t}o-B6AmLbuFVOJVyD-BafN#Rn#7AzpTOQ)
zr$+KP-<9SEg%Z(|v=PFOA{GzX%6fNDMY?5ElBY)&=_Tv{L*3C~KG1Hc6v@E|L!C*d
z9VjdUXVmKxD>-MY%Bxb9Pvp(QD9p&WA4E`I-7Kr&<^4`8m>-iNlULDbC5a)rP<m`l
z)32Q>O3X0*$Y@-1RHoCP+sM0bev7@Weyv>nybdF-14~q|?{30bxwaX<EW9yF{=&GN
zp0bQm;6WI&b#<&9VLj}$a#F`+TV*3(-#koC`1G1fs5s#Ch-wyGGtX{p@In*-qdqXb
z-+=`dW%E?9U1Yz=!e(!;?^wKi;hObJ*E)bicqvZ6^FfV@Y}`~(Becr6rHD>(v?wFa
z80gm+=3h16-cgZ&gO24!Gr8(!GAr$!Wq`@6;Q^QvfaRwm;l$A#@2WqMQtPzJtjVlW
zBA57Z>C31Uc5@l@PT0?`IN;{7u<9N3+N0T3z&Ixd#cz(-EzL8bEnt$AwPA_bV$JZ`
zep%>cqmdq>YbB2#uv;6}{jhVL!|(R@;H10l-DN<Pa~wowfaK7i=qYl@nkDw0d<Nl(
zZRa{W7WOvU_tn8cD)0wBQr2dbkeGV>4y11SCE~G<9^?N<tX1BZLcWB$8u~EXlP8DC
z>zs=5hC0)Hv?|rf?<@X2<@5w;YR$7Sh{1T!HkP{~UU*pqseu$3Y3WI3ifW%HxwrFE
z))8Df?ljV7r<vj>M2ay0=_(S8R6}nzuAoz2oDo%g>P`n11SxV+RynpW*c6UF>zf^t
zh!onEgwBawPEXKF2vNL;0|5Ao4XOxk{Yjt1ykLKBd1-E7bZ1D49w?(Ik~$=Q%z>3B
z&WR0*>UgZ_=_wmbByi*&QKf9QJui?k${Qi<>DavYCR(;G<Db)r2j;LeM2y&l9m(;#
zsiGp}w|zuLA9G-Vin0JAjvBp-%$}k0wA_;;P8L)}!Qc@0=}!WcVP+6V<ZuO$Osxzv
zp9M^gVAQ9;Xc+(zlhPUxl1@0VG@GqjWZgP1Y~KjPGQ=^j_&1!M*gR$bCYBiTkskbY
z)s$^?cE~e|#gk_wa2a!W*6A11psd4+g)8lS<!v?o=@7q(v@)kBO$uX#_F!__sBL7d
zAr0-U$~;xQ6m&nQM_g)AtKWL%3$;xlVn!4#u+*JYZ3F>syxfyP1)|>F4ooawd}eyb
z#WvB*WIC%b6!5M*>{I?~`qS@<q>(z6ZLV*@Bu0jx)7BC?za3M`N&U4g71QfiIoHV<
zZg~4*+gck}T{X{X<l108slf1d_B2Hap0UX?sNLn;b_KI;Qj5IyYrV*9qaczwMr*sR
z&Q4ZY;^YvC(g)$b;hS*!6YC<Vu9BW1nfE7dFS`8}=Q8og`Gz{u;`<wF3PD&Aat<uX
z8DW8V2gV@b_$bWg_Hs82CM??oKsm5vGwn_v8D;2<AE<K^%O<G-SlKCbxLQWn%05!W
zFZ+S|U2qe5Z3vU%yH0CYWbIEm{jrH%d9`^n2<`6KU17-@q%iNX57w>fdYP|irEtn8
zX5DP+l-6lV*^%K&=0k3Fdg680TxEBbk(I)FEL&|EvuCZ+pAW`LtAa(CQtU$&0|Y9h
z?SHRVSXzCA!6WwJ`rA;cLSpg?ll&|-kHg{YBNfNYD$JW*2}oI8gw=OZ9Ap6a!SX$1
zkrXyA^dXZtJt@>6J!o^5{P#nNAC*F$;W+qm2fJ3)0`0EKj8;j^MJ)ypkCAdGZ-)}c
z?k+D)Nl-r23<x>l^rU#Pg#y!dPhB||yvn(5_6|vSvVe)lJZ$&o8pdvqf4=QBwZwWv
zTlVa}D!mZ0gf>C5d=uw&^K<G?+Wlp+YAO^vE^8j~19ta8SOooLvW`U9vr+{=T8ByS
z!Kl*`M;_`>x;7Ti0}W+zOfjfaNd{RaA`FBqar(8QNZo8xtFvt%E8iA7-rRlB>B%(^
zyKG3y@Jb3b@;sgX`@^&R8N!|;lL|w~LOGS(6;BCQF~6krJEPh!(v>7^KT@W3FBAWe
z(+W&Wfl$2)B{9WjWIvjyz^>ibr#v$12syA6k{3<1+XMDv<;N+}iah1AKaSmj9&I?J
z1&`Q|H=i{t6N$paxImWIen}`oos!h#08{WkW#;VT^@C)vVsP^3DcVgIbrn3KKQ^35
z4z*9zA(F4!-*K9Sk`703O&lI51B!jJ^oW;atb(G?b(}Umg$Se=g1ea=_7eqbaxk2B
zT5$)4%n<NYO!Ft}XGW}cM$FHx1wnuOCbxIVs(bCH%5Pvx;%7>Ii!rDKFpf^AU-%AQ
z4T1l3d4z^sjvR1cS^;ZTuUNI-ex`0&wW!><14{P03ZK+ahp4Wf&AI8c@vCs1AEHB2
z1K-<iTivOyy!HyG5jTihm87XM&)W;;!coCgq!LAzmeZrVXH@g6R86!U`3(@!#uh;q
zG|PmjI&E5#mELsHcIJBt3?!1DMhZSuHqx2LwyUy-zP59n8|$*c4Ys?P$Ea{OTV2rU
z(QSj|^9<OY@*ydMiE?&3u-xI513c1J^`xnCTb7a1@D4dJJ!j$2P<Ny4t&Ei;P-TG#
znGs;hS2wh6UwML3_4Yx|I=G|4Fj)o9rY@}Q&oAUW9Udba0f4}amb{W(&OoqYdWY;l
z?lyfy<8xW&BRplS>jQSMd86E6dsx@U9ay^W$P$Y}V50;05+BS5QJCGiOy<jHkJzDR
zR-~w9rk_;pJDi@f<jLh7x5Lf*tBGN^S6J%X8KQEeE?P+pg+?$K1&34?3H?cUq2Iv>
zh*dwDoKgn!z$)#KNB^gfMtn-)aUjW#HD9l!%t$_UV9C{DU4@B|9)(sPud5V#mflE}
z7Lc5NT$Nd;KgJb4rjUWp*ojIeS?EA%cYrkEOXNF#sImsO;_Xf=m%s>7MO*L)4NTg}
z`qB`1!VwQ!QS8nX%J502Uj(MYYOk?VWdvGr`+~)tbYNxCM1U<UP>Q_1=XB{wnO%Io
zR%WCp&5awoNL|_?iP9XmGqV=5;Io=<$pfl}8Bvo6|3{p=QVbj=nih}i!$~;A{WzCN
zS)9N-NQE&X(kWf}Nclvi4aRx0;pUl*JD@-Cn(-bnBX+j=P39{oZA1}`r52adpCr1L
zG%kCz+^wn}@JCixdfjCnC+^T|zi>d@Ld3F&cWd|L!|Yt8iP%Oe1xfg!Ct0L?7>E@;
zd#q04dTKL_P!^Bn2~3m3sg{_Bey)<lJG*0>2wfmn5K!#9Vl6dKPo8<jP*9A?e!eme
zUVsw>i45-sz?|4QDSVk>J6(H<M@{91a)YNwv?l|ooi8sfDlaOVM35#~$my4ydNp$D
zOmk~<Tnjr6b@c;(`<?!np**8D%@dC|-;^_yRa2CgC+Xve)1M4<<76DjOx2)Y$kAa<
zIp<tAWUGx-YGr(bJyCzy8lJ#oCHw-4_eZ=7H?twxFE;Nl#~W$44NHP75l5Z=qOIV2
zhrj(&zPx{Qu<Klx`ItR1Pfn3I$b$X)m-Ajuwr+d5-omy0N?FdL#3`p0pCC@5qJZ+Z
zh1g=hTE<(rT14aki@S3=GRFl2&CE{wwfe!FKd!SF@NEK0@@`Sk&wjl;PRm}3-NRVs
zfW_kki-0tlHpN5iH|7irUy3rS3U$zvXoU1?q{(lVyNel(N#Da8K#SL&WE?4ShG1#G
zmFSt-_WIX5*NNs0W35drkKZn+tzxI7Xjzl7FG<{V`pZ<RXpe;9N9=bBEW(4?VN_&g
z7Iu=~#@p?8n=ezO=Dw&yBbo28)05&c%UNeAD6(9CYaaXWC9N-Maj1N*RrwBj_WKpj
zeVvK6(<)EF6UeBQ0w^&~k>cbZ*dLT}ioMG0Mfj5gE7K~>vBjjP8%N_GRxyS6v!3FL
z0BFH2?C-Zfx;qk=*KBaElEDn%lpohYJbzr?5)y0@lmpXGuUT-V{Yl+)k^DAvoo4>N
zPo9xM)bc;g*=cjMqjN1fQuWTUmi<{iWx*BmoJ(a+zjwg?Ja0e$02_L_#^Eq-e^EXx
zU7ZN!AR}X1=}F?GPG6t>W%B&NQHGkX1Jy?anrq{69Z52W2<@*b!SseP)b*Tp-WkO?
zk!`U#X+<5jzb?~(q{S4(jt+5TJw_QRYkvy-VbJDH%DNQ+pR>QoZyej?Tm{kP0)k{9
zy1%V-u#(zEH1tU%#N%4Czxz6RJVf{%FtK<N?1?=6`#NT1z*IF9i$@5L(^EK$YJ5iY
z6<9;JqF+`0{-F}ZSf_7HMRE!NRKW9f3jIIUy?NHvEz=|5IiDaj9kG9Uk{LEqK=m5Q
z4q?X37$!a%Rde=Uc~jG{Dnw`t%0R%`65apYAYMKRt@%<5Ca(8-(~rn1mZv=>je*Jj
z{-x5SilKK0+dT)8s|3cWm8l}hdhQ|n*NWGX!G5PPhc82OVQ|PTOY+>m6(mI1V9|pr
zj0n-tPbqk4|6ZOmH;yzSNI?vKA*a7+5D{3?w*RO^xjd`f&=x&}sIb%iv%EG9nv%}$
zbYP+ff_vH>`>*DGrQol^yAahCp^u%)>CgAw)Y%XP;y;y(xV0Gp3Xzd!VLY;@&voI6
zVh`G<bK$OAz1X=7OlnzXeJ5$eh|Tty91$`X2rmuvIwHV6+n`mFf)<BrfGYVSHa(y1
z&)`UT{i_fpw12Ky4~mMF+hodTVTxhbhIrCGfA<rw1SkJT$F(>&7sjPSK9K9}3;6*Y
zS^~4Ap?&52$3W!+gn@rCF@|8-*}A&VY38=0L;rdM_N9E<Iy}8N&p83cf#e<iSbPv}
zzg!TPzGr<N?3$Ii-RX&?MM$9BCCuAd`%1&(MkmIcmY%h$JHAYVwXe>u|JWR;hs`9&
zSxQft$v8~`8I#QeQ-*|_M<DnAzWa&g`Yg~2KMk7#laQx#N45gf8o41s`BJ>m3qf#*
zM!ZAYTP$dPwQKs2^Sm%NE|_m$OGZwiUxG!SwB97@hAM-L)$YlUtqK=8fnH(C&X(7n
z>PYsKa-Ra#mCSOMmjO%KV__<~O6L*9!0f5z?w~{{1S5(AOO82j{?+!h@>3<pD|J>Y
zJA?_RXAbsbQ@5w*PZ1Qi-{v%O@lc?3gK8E%qZ!6=*`-*~U>_VF^V%bWo}%Q_M$Ej}
zGw%sibul(jj^A{edL~5`Ik}6nlJ>0Z=QysR!S1-zj6sQM%eBSf1orHE3PGtj-uSrF
z4s2+N%IvHO)#vT7=QKSLSHPB~1QDGHT6^G`U{wQDs3>XX_S}lCt4KknkvO4CH6Cz0
z*z*cZuTupK2}u>Ef_exnD2&JK`S&D6%$w(2#$$wN#ktE!`ho(>^rRpquP||PC2KG$
zdtv#846BsMNddv&!Rd*o$|Du29atg0p_!75713c1JO|Z_p0XgMZpm?^^FoC&{e940
zR1wcIroIkMF<y#A$Q$tDds53U)IZ1esm(5o5i`4;yHeADnNrjS9o)XLj8Xhe2WSxA
z<qeE6XmpCW{ib`0DV$8uG})kd0gQ>wtyH%3WU(jhCFNss15*Ztyzx4BL$-zOPX9|7
z{mgy;GcD7;xjeOg3oMgaL`Zdd0!6h)hd*a!yqvO^HV8izDjd_~z{+pr=w<VTkQf8V
zP3h)a?g{zsRj+lfVk^U(0GWMjMK?Tg<ABr5wKX`hnT?DX(MleKXt7SAsD~X`;#W~e
zak#mCdqoi<AJPP-zc3-A6w#BbeqDEieMfm(<*Y}^C;^rP!Gf3yZ;t#pX`RAaXfI3p
z9o#hGT*j?)j4)V5u*I|AQ0vORGgstw9f{702(4eU4x=<SM*FTpKgsj4tHR&s?yiAx
z=dMD~C&Jf8;ST%md&=%KF|xTcsQ7yz!ehIOcX`adr{a;?A_%cG^P|EW1ZBq#su;oW
zr`yZ#dE!ET%!kTkFTcDvhK!?HUrN)QjpgIsc-GvK61iO{@c6J@P}drEftpqmCS|8b
zL!8;RYW4c1_KJIoF@~dtuqY_jjcpNVLH`lvEG0lmg9_(V3E~Tra29TIuC%obGmoz<
z7}AoJvdMXo4ook@z|n{A`FqQtf~S`SuHsfloSwk0Se_(vY=d@DVsCJCfH7@#n(2^Q
zGt`<S2s=VPFRmm8alv|^7KJJu;8ZPo^hAl`Is{lrCw57a-OBC+0edv^mFZ_kB{*xB
zR=T`u-sMgsZqWKt=&+pguev8qCj+f5&UN7c$;ckX6So_ZJ8Q3Qe#!Y>{E`*ozyhU;
zdQj+;GPI)n3xCf*ORuTpT^eqPuxJ#R@DVNxY>jqV-C$OB`k5@JEo07d4jQu8R(fku
z1egM&l{|5xV1t6xFR$Os4kUK9WBRODm?As0O37ljE1LOl&JJV|s+_d+^L<W#B6qag
zgJvjkG-+2h&xVOG2}*1PmYk!_Su#4IWL}Q2yRo}b$5YT0AHZlvaN1Sn0ZH<arRl)Z
zc)W#(=1H4>&l4}xq~85MS?A2Q{Li^3MLIB|RObcdu~|z`k>e*FSehUyZlHu{*|vLj
zbzK<@t<Pzt9HZl;AT68~U`%4dHBC=)G>K$UI?CzE@7<(qUnuU{a%*z!%rpoN?DS~f
zsM);AURP14wZKbsx+<{zj3G`y;z_Ts?RBF}?wHdSW+vFnRDkgf+3ZpC9tp^4<{}^O
zw%hi`=J~4IBHG*Qz~b{V6IZ21WURp7WN)fthVJE#Q+?Yh%ovBA9z7LOpvs56xp{)h
z6N-GQ35lLW!{(N1%(d+;6+q01*suf8*KKUIi4nlXw=fA|No$Zo;xuC6X?s&bIVrrh
zDAC2bA91d;AM-sM8Y<mxTU>@lU16LTv<m;q^9XV9L)o*{2ba{9p%}`@M*~u3JAM9B
z5%z!{YD@1aX7JAN1}a`qr76}X;3OfLQBleV6klq~Dy9*pB~pMl2PPuF^_E+yo_Af2
zvo(QXrsbtIvu*wAwd-wp`Ei~zBc8w4D@-O0#ST9<z28<eFX8AVChfuLgAOcQ9W^FX
z3c@}PFBHc<2<5V<;M><H7B^6~*|{$4OExy`&bn(q`4MleoOVfLp6@xp!VM)gD}eX7
zt!y5l5Yfzah<JaeC-^=NLj<iG5(5+xge%b~n9k^UP9@-{5T>mvJw$sWR)D-vjTIY9
zj~=%$&YQ8-<vv1!B7XzZ4+MG+URHZ@xrJZ^^Hll8HFc<w&|0FZWoK}Df>u)Aj0)nh
zHA(VTB>vmVhwG%DqkK3<C$BIGbdp!85o>Gf0==QrMgei!fw0}6N_9-7qOYqXXR+uA
zVioO*Kt`u0w})^pq=r;%HIkle{p|Q*5x9VrZltZm=}AItZG7JZ#v)b-?vxvIU*<ed
z=Q1Zc#ISU=%Sf@$-d>mGWLlOSK#Vzk==!?dc2i|kD>*BEmI%UdMVhm{qrmb=yo_!K
z7O&K9At{<U<ZH2o+B@qj@zNrEO*()=SE(Suo$OutO7gM0&UHLW!i_TfWJO1-Wnylw
z2w$Qqo?}iso|L?dm?(-X-%?<E5$ENmVv~1Zfoe$$?9SS#Qpe}H_>Sz>`azmjmIV1f
z-oVf(WDFCZe0P0L_!A;Bk;&94*^!^@^v8Hqc1D&-A|LHNb&L@@El<=*2bO-e$N}Rp
zZMJoi?!T|{sP<T^FMK^f#7IU5$Hxb)Rq&FeE}pI!Dsgsta?d244D0oV^2(%d3<c=&
zz_j^L^KQ&G)?q?<3dB3d!R5djYE<n3afpoYkA2YE`J0(-iz4Yjxt(5MW!;i+#B6t(
z$(s$H3|ZIdkzr?H<mp6_;l3w*5NjGlry-7>KqQ~{INCE3?MdrTT*$dL<TRQktccyn
zdq+)`fA0{I<Uo0}BnxfgIxQrJK~FrIva1kuD5Ex5VM_aCUxPI|u-uJ%6IhCR_TF4x
zm)v@>b17WjgbU9$J?R@#DTKffaOn{TqeR_AU!M(?S4b4lv|++=U}c%=K&&9~Tsbye
zW{dsDxsJPH`1EVh#QfMueK*^-=IjXD_Fe>vT~1FL(Z*X_Q*7>5$Y|CnX*-bIvRg#;
zjSRMIv<xw=^1(G6SXqlYEi8u+?91MlG^Y?)6`!S<OEFCoeC2K%D?iS|pJJQH?{j)`
z64=S)7<>#C(5n;(_Lx$ZuasyRc~gs_KlZyq{&SRm{4Tl|EbOuI{Hn1Irh*Xy=^a2J
zrPK1U5^8k7ChF@U^H=~#I=~Vgoe9n_>a=Wg#n@t;ji?lp4y24a1CuFoVOt6;6c`YM
zl<mM0`>?xlRqD5`NxSFopFX?NX^66}DF1A-Jc7?=bCAFAi~|#iU2*07%k8$(FT<KD
zp&pyw<G>1aN|@@{`wOZSua!aH#l_|H%!!DAd=wukhahN|X)4fZ-=EK0%z&j8I_+3r
zlJVHK%Kef9q_3T_+w=Zq9B(e~x6{6m@oIWeaRUv-MJ5#%bypTurbyiO17-OvT)WE4
z^&$>ovLDD5wr2H8=Q1_F&Sp&2s!M|^<CWamV&wXh&KVnx(-WiRT{WgnV<>fj9<vYT
zgdn#~M3J7h0}*})+5}_Vk@!V##DItItcR)yEEipTxeTQ;ij$shcb3`6u-~YU>J^s8
zGl^7>*@x<kz`V2-&K3)&Cvh9RsRV^X!bHg<03NpwC*uHTL<l>rvI4d6tFR(d+(!~$
zFTeR>=Q3`Rd?AZVMC63#@`Kqkv+eu?aEFiahzGFMKR|#tJ!QqJ!aX_N$lO1abSjeS
zIgOyfYp=Y<?kYD6--^tam9fIISheDBiQFvTmV3A6NZff3uZ|tm>DNnVwkbXhi}WCR
z&OObuQyw^tI1j9R5JyZwqsn@Yxwd-?J<kzx!LEv$SI-I*=-d^*e^S+<K*)W~2#b$D
zW@k3ai2gteTo5N1I{|XX{dF|#EsFb7P@x3M$GrAbTfMRzl&*QyK3aYw94NpdUF$0T
zN>8$f4#d_b^41=xtNF%$aME(x!u%z-+<KXPEN6K+k$2_thVEwj;WEnNwLru@;K1UW
zWw6n1msPZ|kJyitZ%#d1D^f`yY}bqxm$uViNa0X8xH0?Kk5+!zH-tM1X9ob;eHP4H
zU_Vygl&lz`89qYlCpHgyg0|<BN#NV%{|!H_{dno0LMTzNJ{4ra<ZL!O{mIub`SdvJ
z?c?=3SySxoQdzvhG~e3q;hOe|DgiCT*}cJO7BZV0@9M0g;f{T>JW3Oy9ds8*a{&vK
z#nKLzqS9Q_0e_;Rgk4nxs6r^f(uO&2-Zl1<<x8~X%iJv&nlP&b)`u7+{knO$(<aB~
zr^?M@kFu^Q)O}77E~h^xK=v6K_F?q))0Iqy2{nYr_8dqcR}l<~bX6J!HX2A@zJk5>
zGtE=93FzdC0}d=FB#4%82W2Vir*!_amD3XKMV!VKbkq?Cn4-&7mtTIlZL6q&7osro
zh!YO1AWrAZCI!gueaf~sGtusD=r~P1ij$k*Otz!^EX~;YH}LKBh~&px8+x`gug`yo
zc%X8v3TU>g;(y`bu+vDHx1}RXZFj{Map8kH8`pt|UYD+VyX|R^phG{9<k^eTV5~@+
zy-vU8syBq4#ir_$uIrB6)*9;E=0NgeL!A=#OB9Li&5`2v={najhw_OW0>`pM@2fK+
z^#$@cHcj=-f)SO3I%50l9ayPk2!blSfwAGppC${`f#&Jwkjz6w^9qypEMM!6$M3L%
zb@T*}BbQ-u08&P;ItxhD@?6C;A1pl*>l7(Q0GwzOVThhI?NhC10%Iy%`cR^Cq`TF*
z)SO1DV$Tkjr(lE)N}PQPJt<O+O>EepXB;WKI}M9jMkGaMLTr;u!n?Ukv%#!xqC{uK
z4;a+pZ7?+lmh14Mi(YTX$_?`4D=`1~^g|9TZ@A^=OYL|W-#!?g1d-VTC~Hp$w6XeS
z58)sPJ5e=k%!EVTUYAoPNmT%<$I30;XAjknTRV~GUjisAO^jD{DOJ2alWoY6wv%O`
znbV)v(h6gP44#C&%1+f8By?B_-;)kR?<I%_Dd=>;VP#`Y6h921ITzcxD<)lu@v4rP
zan1*)kJ*`K|CHHFW(P%#W~j$(>?L;vF{`?C)|ftQ4_8(((&_Jxmn?aJ@zqkacfpO$
z$>|y!KJ+lX+a9S5eenbfv*)x6r;|Ne)~zFUwxL69ta7618L(BSC%4LWXEU$l^B-+~
zzc)<u3Mv97dc;wPly2wBh@|YxN2~a&YIoUBr62lrH}TVQ@IIEfo!Pc{^*fwKnL_QL
z$kOa<b4oDS&*jZ4u}qIR4RI{u06O;b4Z0P_0+s6L5*?kMB*m+)m~Yc%@F}mrHBr?n
z4AW19(-3-N=j%s;&Iqx`h5;<b273uDKy|O@_{(i3*XOm@T<Khv70&@#+$7;O*0nue
znWk!<^qgj(VmQt~{e|o#A9;|lFsB(WOhVINyh=F!WcDN2%ATk*J3css#n)-aTgx7Y
zU?K5|lb7rlE6E`Wt2#tvmBz~88cVx~@T7BBP=*v7c1Ql6v|q|arXX7pf5!m`o#?BS
zH2CEl{>-*bEI}_V)|>)0ex*TIBj#vcLX21CX!`Z(y!&B1^S|04JjX+RtOJt{!Z+|=
zQ0o5o*{?N&JQ;+(pDh+xd^n5`NdWt={JzLH581EhoGhE5s%D`U@&fHQDow0gc!M{M
z<?GhiZ|0~jxc$;gsfk8QJJ<!Ud%yiwT?kaw+N9`W2ci$?6E|wxZ|AKm-nQ1c%#CrJ
zIEh{MJ9U~UmL@L(?HFw?kMw$m3E6%(hd4+Pf9E<TpVH&nD_M;46*QFw?e}tGqEL<2
zr56Czw(#tznBMPK{>bTflu7M408JeQV_F=ye$NvxrlDdk<b4p^ydes@^qh`DMN%C~
z3UFAK{;+<?1XYg-{-s?w39A~eL6so;ql%{9#D)p#MUOa;_z(%1rKSNh+SIJDKdw(<
zsp0yNl;prdWv-OWajLO9Xu&nYp8ZLA@|*x)c9+NnRp5@(pNK}LQ?-S?Qc;w6YJXbD
zqTBUyIm`(Xe^$S53pxN3MOzb$#Z58ze_ozOLORNCTFRyDA9s543uhxzL-m$ujNom5
zk<UH<Nv7}<(^FpBsJnw4j@Pxn%xPcBv7XMgC~hl^_OI%2dYyg_?sM9K?#$o`Ze@R6
zhuGWL+U&IE7-*7hMD`Dqqci;SNT0t+GJrOxx=uq6o!Pc|kT_uxQqTUj8b_IB3jGYW
z1MYxhXoH*W@9NOTyZzK&a@v7`RBuAjWWLUXk3h2g-`8QT!DZvLk|SR?F=qczzDkCF
z3C6@5II#GRxg=A=q6nyq-Eyz}WAjof=|O(V!wTDtSOxZD`ZeBb7tFJNsw;BQ80D&*
zW<-N*%kouyLld&vlFN({|DSWH@;X7{v<iuHusow9JIMZ}PIqWcS+!!0SJ`L#rEv=L
z#mmRIkag`}b6oG`xGXNUgHe&@g#BA$1=k{#t!Tu1Q;XljMz+)by^baA8CbxNJFwK&
zI1jiVF7zMuL#Z=1(sG&#CVs<f>_78r#f{EYtcsMm;v9&Wt7G<G%}Y};S9B_{!ljQ)
zAF@x?dA#Aqb?cp03{uu%8SRms?b8)GbK)&z4N(|A;*Qgw2*tXHw1O1;kbNef+7|&|
zPD@;0<_T4gPRrA06H$xsnK;)P65@=+htK7QVaei3WYy{yB0gW<BT~>VsB^Y~#g|X9
zhbrJfyI|^IUno6VGDU6>P9DF*Yme-p^83v);4hXpq~}n4X}4Dx_E4T37`87pWGatn
zjZ~ers?wkHfzwyfohqLYaW(ek@=HafpxUyb$*&`$TWW0d2QK9pRF36Y@?O7^$XVEa
zFIIl1Rs2*sg>)-KBG2$wE05081TfJ8iS?^Bt1)nbEPqcazA(7@e^<eiWzr$7UhXaj
zsCiITz`?G4tuE?bXVhsFp5M5R7~d`wRWYwAvwKTF++Gr*jQ-M~rIVAG?oEHhiIo01
z@kLL$w@?wuX$k;&5UH}l^qgXxw5Q%1L+o-M0t2R%`*MFv#ZRA>%{XIx-BPD1c5(F;
zSK8C>ErcRN1VoD8m6zZiae9(`!z`}&@yF~L<!)J>R8UN@s^P|Sdg5-`CSv!V-fz#G
z15;@|>IXQm^hU9161qR0JFz{hzAW~YJxFqg9avCHyZ~w-mzI|)W&7FX?lgf0?Rka4
zkwXqWI(^Wd(+ohC5J`SVyuu1~N^Z92mN#hy$)Jo@s`ugjq^E=$bcYCP?b`G1ZIpyw
zC(hXgl1nWxD0`SR!|i0zKR>6WGZG$sT7e-IjgZ+zZl)L98%%L#+eH^T*ZG_hm@dOn
ze6+pr-m*u491Sn3KeKyC@sl$`CYIyQU2~R*in>rR^WMTUIjE)0=0#g{yn5NPbxYUV
zi^?w`IxLNP*KP+EqtHSWxhbwyb9&fbTnJfTC_<JUZH~m|^e2B_s?#@z>>JCIWUJNv
zItn-!Gd&`|`Bz+R-*j);ALq@#(rF~J6$Lqg740R>+L2K=bUcPMo5xf88EkBb$|RV=
zJbQ;oc={3h=K8?J<k=w9I{<w@)%6h)?WJXeDXL3JMs$MpoOF6(QO}9+40a~=D#y>a
z6wJvk99bcd3NUSOgB(U;-&(&NfnW7R4l`GprK3(yF3g4F7#;1~>ZD<uRP~#E*ld7Z
z$b)Dp!$@3diYW+JNA5nVvCiYjWU_BhRKDq=*E^Tm6UJH;y|VA9%Z!*xK7BANEIQyK
zY-YZkG8)BZWj9wX@G$Ay%L=y;vtxs%Swb%(G-w#-@po1{wy}VD(^}3UQ}$gAZdAxg
z&c{c+!s>!m6x(;#PZ#wD-cNgl$%6swXGV6|_cU~O07Rt0OGS?*CKGz$^oyESw71&J
zD{|r7L?+ObLXT+IhpkN#&jon{URG&&PBVx-EL03Na;>87%`55$u2B^<S}DaXBen9v
zoacU24d=QDgR^Fs<N)b?dS&@W))9J|N)H@#V7Xz;dXT|B1S?!kKV;uq29;ZPPJgZ&
zdJ64e1nSsDN!?4)izcz{Pdb30K!Szr;wL|a*%enh#QW*Vk6DRUdr28gkUayGi9GDj
zWvSAiuOv5ra?~!ZXz8hr7_D3-b4<h%2UPh)qVTKo$8j{p=8n@$kv(LRL{paC2_ZJ8
z)?S?`o7qM=p&(oIJz&XNhFWoy&TA?YUPnj@^?pYjNbYF!?XtvzBGsJhvMa>~APa}R
zwoZB|W-(7-UU?JrB!@<EifRW4Y?oK!han+SNc#jNCelZ_n7)o(QPH_@;X<boPpDZ#
zs`Ytg^FC#b<!aNaAbhbIoqQTvdq-?ueS%`^!!dovfd!eSWi9wLd<ZX)U6n&onB>H;
z(<(bi#+<N;G806r+Wck=N1-$Eb4>a&FniO3MM%tsa?fS<7Sy?3y=vt<otAj>n2JCu
z9>cD#n^HLGQX7hC=R;0}8jp16+?5P>3t@(KO>J$Up19LWWHBwu&lIw*Z3Zfos5G*O
z>A=iBr>9IM5hN@a9nAK+0@D^FhmFqA0an&@QyWc4C+pz#m0ZzkrCmo6DMDJrqzJEf
z`g8js`G@K$U3)_XtfMZc-1`91Q!Kl`4*QXa`Nrmj=?#|wM;hK)rzaU_n^16-{q5MB
zl3K85cE_AX8LIq_l@B9)*qigVbpx%jQKu1Ku6m~m%0EKkowwxBx`?{t9g593fY{k7
zdy4|UFqgwnXTrHGc+(D&I>4e_B0Y>F&UFkTHz{F4?mT2DTU=KiW*_RAIqimVWlh)2
zyVKz>xS?%H^Er7|x#CvC5$E(IB5?VqU)+|Kn~TpsEI%orH9g2SsvWNh5~ts5%j$qQ
z2d%?&&@PCHokl{VT~`JPHW3NrYGV*=o{H*%u{r(8N|@j+RQ_FFSxJ=W>0)>XB3811
z<78Ic8#?8djm}lUqQo*87Y^C=1txc_+>46C0ajR|ggmOp5X@f_1;K~CwfsinbjF%Y
zIj~$KS6p+=<+d_gWC)G3q-n+T(u=0gk%_i3z1t1>JQ9gpW3kbbIsz>1*qb>ONReH(
zsvvtR%!N_eHdxC!r>D%Rk{`qvbl^kdBDK}cFHbTVvXGS$(mvaxtwn)zilCAAXifP|
zI_knzn=HlYnParc;xsAF&fNgtR&f~XRNO=<8uEjspi95-b=}(4H?CS@YjdfrT)lYV
zN~fW@TeM=8t;@^di6Q4w`{K;DEj?RbCZM{Z6xInWkf3d-M(k-hsBo~3v$-)J)#__G
zmvO5WD+Q#8V3c<^k2l6IW^XV3sj~z@v*z?zjs1|*A9O^WV=x^S3A-t`$<{E@%%p|U
ztfU47E8E~fdq;lc%(m-Wo18|hSS6OK5RWui-S&vRv-vo=)}vyq1B($SZ`S&jQ&JQg
z1{Zr*nUmz;(t4zJMAU2q5$YQbICqJraEy*kj_Z&YyZPSGw-=7E{|>hCOO0d5cc6ub
z#$mf-c1z-JbhPDMmTwo2kq>8w-P+(&S!y}Plt03$@AT_8v(f->es}#!xVolDK?Mai
zvkg0ng0Zl#*n8?a=w&8O$AL)nS*eH<`@STj8(NBgcUom|$_fnsp0(=7EZsVQp?cf_
zh>$3Qa;a@7BbUZb0$}`2W|ieW=x;=AVN_YINhyV1LPXkl?-Q5uL}r9Ap4gKzM^3B!
zEHfcnj8<H8REYD7GM``PY-0?lgVxS54iHx6Tx+FZ5I~N%PW?pa5kXo`P@jeA@bEZo
z-TKOXiV#<@!C9gafK}y=L+{3|ztCU^5uob?+6z1Q^j@bYx!FQQY#S)g3R?)@q}p^9
zR-NJ`kJ(_kPex7gP3-FqEWb3wOHwS$#+vr=v-aNloACaqp7%)y7FZXZBgnun%dy;y
zF0z|gx4Ugq`K4BIuHT&wtZc^cFy`YCHjW7BwV~!cDitJ~%U-WA-J!BmSOsmk;^=)7
z7@YSx?HGTVNk*xVzy?z`Qr@UoR*7R|!SY^JlAtGuQ5Jr9pA<*y7MMgw`RR35C$MJB
zsx*Fz@o6vKYokf&>xQMD(a1&FBO*<H&G$9Wl3z5j4NSD%o<O!6%l;XR#&{grLvY^}
z7;ll7BM!3Ldih{9A%%=|+72KFuA<Mf0&#SQjh8{@!8)XGh7_!@dwK#vng+RN=f+hL
zb!QXFI~H7VwR5S_o&N~d<IUwI!j60~od4j!M4yCWFn%f;Z>e91P(YBB>}lS>7+mtr
zL$<XHq1r0<vkEFFB)Q7z4-98AFH&b%hOnI4WR+=%#vdIRYxNI1pfceyZ%Tht<dXt4
z^o3{bw%X5BXZe1eM<%01#1T9FscbFocE{TG{_+jlKeV^17F`45G+wz7PRtyz@2@Ma
zk8*ZSD<}K<2_m@(z`nhb8MA>MCd?<nVtuwp2dT)!ARQFKvJcdUv?oXKbWb{flqr^T
zjQQXQviuK}(dp9^1o|+lC9tGNt#fAPTtu1Z05z{0*rMy#IE|o^(0Bqj584OogrSEV
zNUS+9vFqhHC?@TWB!>4A9L8m`m_eN$cV}(Al}tRRm7h;M%7~nmA1V*7I>ITm5@0b-
z%vZr<2_<48h?BAp*EKv!A%3WycH!cW;L#)ASu-I)5=3!yjJP8c`;juStYCIA8HIGJ
z&#^FjQ3=E8Px+TKqQ|=Z#G~5})^QGCGtmYFAlH2C&0$)#AF2egq}4{hry0Y}qBDqG
zh-@KN*j<&fQb3eZlVQjvp?ZM(oql0*#SIIuTWWXL(F~4owj&`};|@raw?%~@h#<Zv
zv9Nkn+=2jj+VZVj@E*Ij453O8xo}w*UVD%s!9gzR#Lj)?%jECWQHaQLUSS!^7;|Iy
zSCq&XN94Ez2$W!zX2a3~c*s7QFkHVSSsFbGnCM|pk>%s|Kz{3*o~r54D5NGCOjR@7
zK9;<skAoDP>$1MFD<^N~E=jRJTt~B+uq^oi0K}sZj$pI;zz6L|n#XSsgTEppfeF{t
zVxxe~kLJ)8QJN0(BduH?FaQP0sGel_RP4v<`dYGL-TJjF7O!_8ddz6ldie439xb!1
zG`h=y<(5qG3F)C9Z?--aTF}FFM)s^4k~R+RDqlf72%oC+k3Lc9ij77V1HLek;I${i
z|MG7JcYLz^CiyvA7y>|OX-<!bJz^2dl>9{V-I6|qI=tVkqQgGhb5QcpoOrRHEF`UX
z{aVB(sY)`@!(RI%k(1Ar%|2=*!0i=^@KfcTMegM(;_M_5Iz3_$DC9bX6u;U}C#uPn
zT(Z=g#-baR*w55;g*c(%g-Sd9Fa)km^OVz{PABY82*9_c2k&R=%nT69ik?bao7dWP
ztJm4K#J@rp21nC^;J^x5WOY2u9l|AiczY#_;coAJ6N3c)0f{{ZfhsIDjiVptba;A?
z?MOB`|JwP^rK}bGe727IlA`sn?X29A(`cz;SzGHml%U)-OU+DE>=fR>ShjL(QnTE4
z*YBgQ3yG~xJI5c2a88m~>lxcq7ZJsuH~^~O0mQBFH}pwG&y-`WY$^x7Y;4fE3>r;E
zKmx}p+gnGvh7<mrRy-zD7)BarsN*-;Vf&gMC50uap~7H5qDN1i+14#p77&E?7i{L6
zGV%l>ticmbe|oFfb~vhBMe!6ha3Cl2ifiXx>E(K68#S~K)>n)uuOsrU99RrBbXUIk
zAvCYZud@fsXD02+tas9Z#oZMqn@v!#9cu0lQ!|r&%z*{t8`(s006ScMtIw6gMM8*{
z<PoPQcki(d%z|nocBJ?i5K@qkwGm6#Xp+2w)C<8o_omg0<iE}sqMoFRqh&bq9bu&j
zhaIpOyrQJ#q>W-Ep+O%@I>2fZE2QTDG;$78VDjvE^KB0X>M5P*6(;FNBSHGH6O{;-
zQ6tM~<yOCF3`23dV-J=0&pA_zPdhTb$LZ1QX13kbMS(n7S8bkYBXcIJ6lkujWH{ku
z(aYs{!hU3@>Snx{1DS}E%rVcXlq~kw7V?IjuCs^7Up5hC$;yRv%;^yeP=#ANlY=3$
zh(J#-Ei$(VOtOdTOZtW=!{#)#39QCm+OkLL)0H<FbA~peW#+XfqYtuzEF?X>*Ur|x
zo`m>`QGD5x4oGNgQseB=f_>$<D&<Ee5nzchR_(Z*Yu=o}{z}CTpSjZ`gUd)~lRZ{{
za-L{3r=<_R{q|dKv7f870{@h#NC^N3ma;BRRi^Ie^Y)o-)Un_?mVYJ=#c4}6oexL5
z!M)(L3Zco85GE2R<S=^V`T8wIA14&mffY3)^F*iEs_U~=bYZ4E4LfseeAqh}Br()Z
zPthjPG9dCyiAqN~7C?=L?ep>a&Y4|kW}(SDuw2Jbf(irGwYb-Qq1-Lgl8}KL*ntUI
z>z1y4yFJmcB)z9`w6mgNW{=<L^e1-0>Lk0^ar?#kwpel*aTZ*|>503w1*jqexjObs
z4L;&XLP_r6P*MJ{cx+Vk=iD#qPIBUS{W`uui73Jgux6<sVm^LH*bXHN%6>Fuzfyj)
zm#;^$(_aTxjp1k%gj#L1Urid&=?cTxI%x_Eoss^o{aSLYc9#Q+Fle*d3q(38ei+5U
z{(2qfVx5XgD|sB!BL;kmLHtIAY@jepYm#RGiE-<6P3iqAhk#We)HFQE_M1t~7r*&+
zUM|-Uj@oZk+C)(1dJb($6F6qS-Hc;mXxwSV9a%<OCw%U--zhMJBt%EDF`4sO*im?1
z#I@M(){S)|v0GHep&ieXuNNLDnzrAobFC*sAE^V&UP4b|7=M6_G)OkTpF?ktP4qk0
zIjtL7NXdUt9$iE#j1&7En7GYgTk)=cSZ)Zh5w3vp=)iQt%(i6{L-t4cJP>#eqM?<j
zh3&vG(S7qgauL6>n*KOP(i)j;IoCN=l$#vfILPw<lk#hkq0Y&X`j{h&jnL^Y+oHVG
z+M!imP(H>#Ej?*u%?LmwA;i4elS_zPXwaO$`LpI}5tR<n%lo3riM2=X#zh3Nm*d{-
z&nr5@S(Py_6&7Y9o**B^UnEvq6ROYQw8{iNTB0cq*<V&9VzPD`IVsn**qHvR`ItDh
z1#wEqz~=9@N63_&jy0|xxYu7-#Ip(&f_E6cG`9pT{mCmy@Ka8MB2W2Pc;(+zYD0AG
z5YtRM?kprBUcvo3e)n(dog2Fxh0u1|Pr5S?U*r@a`@5W4Ou=zl4Z0(cq?xO(zTE!4
z3|1nF(nFHfHia{s{^r4A!m2_ggV&V(L;XyClc&N&`UD#xpM-&u{`h4ToFwjxCLXna
zER@&Z8FsFd@<^&c^+jm<r)ET99^|8$@nD8waCG`LL1HRI6902~3v4^ASSbIDPkvN-
z^yKq*z*0Ez0tND4D#cJTM8%b077i?tap8p*+rO3@l-8-U?Wl$0z~VE4Py+#BLSo(j
zTfLiX=L`!y>%a;h#&yiB6K|&dTe0(?>4^P%k@C_?D;Y{iR~#B=4uhbQTqp!P_ok?e
zJQLv)smuIF9r2<YZgLt)tpi<LXaCvQMjGohD4%P@Lwa@Y(r<)r3bFiG{R;L{lvLSG
z9aym~>o~Fj`&4;uy1VO6t=+I%jFm>kU6MR{tu6NHd<CB~#dcp>6g}bOVf#!)XapU^
zVq){>;3=mkSvlyB`b(r4e73-{07g^{b6|nwFf@ub;3F&%`DTIxeXjIt*@U4V^v_*@
z^e578TE29ZeZI;)<YIEJCrx$o0}3|T7aAhY7`B5UkMZSRdqkQMksa}oih}%N8M0ol
z8P&-^z=WI2uesK~R36D%kXTbiJV=e6MEh_TiGIYsoFm)Z9nzWCPBYLWl`qOv5N`05
zh8VhSr%}Xj6<UXTeYN&3Td~S%6wW7`g2aM{?7y4OlRg@jtHhzxlk1ZFPaU-0Z(pkr
z9Gh*qa~w!n;SoH~6Kk13yDxjRD&;XwZ?FU_dh#QPCsM+YicdY|zMRiO3%I(bG@mh|
zM-WtMzsOH}fJ)i!Cfq=k@)S`2)O;oiABRFJJA!F!VC{*q7AXxXSoO5~iiEq2le)G#
zO`$8x*U1R~^!s9x=3jA@bJ={FYRxpkI`)kFLM9@5Z_v3;<}v^B1@_GH6>>Y*0tVWF
zG{;A0PCNZcb&ClhuxI7FHf-4FT;{t#miB|4Y<@||&n}NtZPkd0z|Jo%LbY<Vq0*mv
zV8kn{LK<n3&&ek%bCJvP6{_S`Qis^R9=7K;Jt{Y^?OkjW)05YpWVyx6gqpD1o>$Pb
zV4iasv#3MH6+weaYtO&0P%qWn<h7?Q<Z#dr%IOR0v?~sqHxbw6z+!aitxv0xzH6tw
zu-q-bclbu(hQQ)(gd5CcvSUi0LoBgxXr5DpCXwP}3!vvxt>uui!CamOlLxV<Nlh_+
zu_<wn^w}5XH(s+~zH^!LrP^xt;^rk;`KU=Dae9IqrAi^iU8ru(EV6GbA0wGWLFY2E
zRhamrHi$<F*7&CTo-8bO+72jr5liBry`+-w`qnTxIy95`2Cxr*bM0G)DPY8DCMRH%
zM6TatFRd7rGa(k&X|J#h8m6=Wme_A8_Z5mKw)ucpm?V{uevm)D^}fO$I$Jjm>g){%
z5+s346M_{{;r4BHi0f{AyO+i#Y{;kV+jA9-bT_x0YklR!7<s!L`;PJ)&59^f`w_65
zih&kl?#s#@(sC3+r?URQ5*eMb3H#1EBQQz~l^9f5JP}(K)-tC*`Rq7R=@^f`D<6N$
zrTWilNe||*>aZYucT&WHdGnpie91=a$lud^Z)&_~YhGc}I4LO^o6*b5eGv};m1$=O
zRz!1*cbbhlC;Wz!jO~Kv1NarcVNLJ>61`-T+UfKQA5_NRz#@A^o%_D-P`l?ebCl~)
z=q@ZzN*Yhs7Yr1gCp~cV>t-yEgs8r<(0By0N+D-(u_rh^$+bh&2@jclZ-W4xlcWWC
z)MsHrfC{%9wTtR+39wOq+@mm=28LxD*Cm41LA$s<kF}#f9R516-06i%R#j6w8#bwc
zs7uOkv|A7)CDqG;Y5ux|HFc;4cxiKsSTXpVKJ38CV-Lcr<Cr^f$FzjQ9beTf_9zIX
zI1xTjvJ*Xp%9Y={2cEIRUY)eewlm;d2bY46g+f%^c&H(-xi2jjB1dL(ZiWi}VX6Yi
zM(*5Im`Oz7u?9rCaxsL-*e=VD+sL5t4LZ$odlMi8h7V%Ul$x;9Ui-iFSD{ow4E<^I
zA?uTJ*_W5EE4B%3u09Jp6ZM%W0%KR)_r!&u7@UdOv2C+b#U1{)x~1H!S0MTk;X7tm
z76gR_kDqc9FUUY~c6thGl>n-Mt}`~TfgO^`kpZ1>U~+xEW1(G@Tn@6wG$-W_Kp|6=
zla?=bzs+xEWcG0SSeZ^wQUQWX$>?V&WGV0>B0X$DGt`5?L}Ds9N;Zts6ThsspmSvb
zDCh1V*$`J(2D1cF1b08}LKRBhMIkEs!d|<kzJncBmOg4es~;MrC#RMrr@dLxX(+j*
z$!lkR%&yIq%VAJhN=pMDMt9U+ml!WOWll4h=S`{<^!kdT$P@`Dx&upDk3vUy&L?~J
zhI$w4PAV*+oenJT%DgpB%zR_sMeKfuTqzDFAi^71DlPb5h6Z(lSHw5GDW_^}hhkb1
zO&vgTsa5Y>Y;VpXE{^jkoR%i|rbYIa20L2*b2%VAX<1n_v0-SiH-O%_up*2NLHtM+
zIbs#?Lox)s_Gb`3yD?gWB(qXr$`;i<lVTHtxEQ?wEE^o_=}_jy%^SYH$YcsM^yg|K
zzIwQ8OUmfOyvnO|dXsqXNv}N;jJ3x#xv93a{sxIt)<K=Ooz))bku7d9|7=+^kmMXX
z!zaDGu?n63bb^icCN~a5fH2c_UB#IaeK0BEOb3*5!VR4<Dm_fv@|>3C;a{Z@WSdJn
zWNikkl<pO!N86DYl<Wm88W<!!qBdHpJahe=)^Vp*R`_926++EvdutsO`n+=RI&B9O
zw^5&)H6cChQCpeg!f8v+42`@2m02EBLWS*y`X$8QPPADP019iPTJ1(#Rr@LLG_JZB
z0EEx+i47fFox>c&-p}J{#p83uD?aF)t*P@8F+=hoWW4fTrzh@KZAP?wc=b_xTP_t2
zn2$(%q$41S=s4a4YF*m^Mg~VmDUb@FC|rYM!`O5>wywUDU4~6E_KoCR^yHFWw&J>V
zw!SW<b*q;+je;9_2Z?EC7H(`l=nO;DMz+g=74|Y)a$drg+z6`U^7f`*2Y+e>A<!sj
zC;Im~{X*YDP7t=6%61?@MkzCx;8}Td@-JpXtekt}z1lj8PnCP6TuWv?VDG3Ovvg#@
zX=o1P_}yc#ws$sECxJ=fav2tO*y)L{&f0Z~6+CY5s!zJ~()T!x#%~Zs)Zo}6YuTK&
zm{V4}xf0`gB&(hbBn~Kk7>vK9LFTv3F@xAGb@HflLXHkAt3aq@gu8~Ssy(~4A(Z`y
zb1K~9USZluq?4=o{JV2<IYER{i%u)am>5sWj=U##7K&k{i;Q+G(Oii7B5@5_F21k6
z4l#~>nMrxfD=ft&Eu1~ps(TfO5phnL(+*VT-X=4_Y1>d>5*?KB$BOg{LkDA(D!AdW
zZLE{aeHm4tl*A^3&*_iRQYt(o+K8uQ+8?xbAtCm*&_u(zABw)bTAVI2hlIE{Nt`WA
zf{tpIcFHuxJOUH(<phPq%`6|9KI7b#tB^3HI9au?p>o7>5GQw=f^-!;t{^u0wavWo
zg3B&@ZT`>t>oar=*~D<q8vu2f+x;x<fl5We(sX{Y11dpFs%e@_Iqjlz1VJ0jHN^oy
zL(X-fXk~CmShKyiz%+eq8%*pO2bL&Bu_*B#&=I1FZK~e}RVi-}8$|L-{<6ZI{`jRP
zH783HsI-is=H2I_J|e5yf#rrf){((k2BG29jBZ@9-pl1kYm<%SvfDZ^K0M^5F|%z6
z{)=v%w?MGy;+O+VNs2fVnUASi94dlRULKp?X`|&g;;Ra+BD&if7%6A6#fojbuMTfO
z^@_*~r6gqqaav^uS%eXfl~1lz64JFIT%4ZdsGN|_ap4V8br8MeEhrEZ<vFJ(@56u9
zv+;(jD7G~;1Sui1&V{o&6e;Jfm`Wj6+m(VoCMt6ppB!@<7eQtJM+fA*<~6ptjE>@V
z=pdQcBooYDrze>!Qwo1WKbAcy4itR+4vGhCX|SpEM`3Qtf#tp<v8S^I5q)bjPn@yJ
zU7Stez%CUh4nc>%*h)k!kKf64WKjbNLr?@Pf?u1gd}V!ySj-WpoqPr9SGIL(WY}$W
zEipx1xdR(Om#}v70qy;j{S6Snd(vqN{k?H2WN^}WoeWB%_dBh^3iIH@1EH=UINI$M
z*=t)HorX59nQbdq*$2v)w0CPiN0avoQ|%@lD~`v<exRXJW!49q1}})k%(G(M=`Vi1
zVT@fehnSbqPzZ|8)nAzrrE}4f9DP~0+wR#1a}~8lCY;M8uJ!L+v()Y=gDdr$r8GFO
zyyGT#wcXjEG%FIyR8nCuyigrgs^>#>ilf$wilYH3%dUwCDO@QGe)i!q`aPkVrj{0I
z4PJZ1W2Cw3yyGA~`$&$B>Zt=9l|w6SENdyaPet$_tYaliR>q8n!33~G8#0>`=2)RW
zl(TdG<E%-p3u{yScZPZePq(}BRXm35NT-!teSDzBQJZbMyWzG9*3sx>|Bnirl5JSf
zO7rKgd^}bItVHz4?0SY}a!(FoDarZHb&##Ntri<Qgx%X<PlkAHy%9FR0tljT`s2w{
zEINgyHYcpwefhMRZHpPIa<Yk)1Qv4<yf<;l-wd%kIK9j6ub;<HLUjTVKaioh#P~Dt
z#wu(dP3-h@WF+PwTJk)+6;H<l<y(^d#n}Wt3zMlLEGYJ|`pqTy596@|OMPwLRrBnJ
zn;UjUi~=<X9axTP)>dBj;i!dXKhhvAbNQuBhc*OPCp6vZPejA7WF}Jf%#Y?otXR5g
zy_f5yTXB{DSdkWLKrK57Pf>~`N(!n)3Tv8}lg_=lWH^a-i~V>riWE<G24yL3IC89<
z{=!7W+f)i#W+k0s_VG$C^5Ze7Dz74rf6(bEBiAk<HR+6fqQGRimxqr+!(L%BMXEe6
zGiINhwcr!_OED*LGJsk;D-_%{f~oN*%EPqgWP&&~ecTI-6}@qAl+8_tW&C7W76&k8
zopV~6_(c<w_ER}|M0LQ3X@o42Zlt7MQNP%CbI_apbotygR;22k{HyjP6$u43==x_0
za`#2C4GipIvuVysar%=dU~E!Q^UvmY&1|C((Lk#g;oZPAU2DcVeXR|)t$YzAg7`*D
zzd!!~+ZzA3?G?%JI-)5V{Z0p#+;w3%lXo=py(4Tc8j|)!2}H-IcR2l;JC3ZU$R?ao
zJZNWKzF~AMbj9AV-=EwQ@-xijD2Wlf>aNlrb*?K7ftpmFMXCF3cSA#w(Gms2w7;Hl
zdV+>D9a6DixwfZ%9*U0A?D6J%g~_X^Bj2cqVN(T3oP>d(Ml05#_GqM|khL^PrS7Qh
ztp*x=Mnn!$5-H60fKv?UADSdsr#oc(%Ii}>l706Pudoa@Pj0xs8M-`nd@jrYSS~t_
zRvoYdm3K^x_MAq1scejrs_q1P2MY{$686NX)#AW%bJ$48?)W6U%pR<ByK3Fym5ZIG
zcJg&Ca=s5Gu@lqnT<YB{u(K!@hnv^sS<yRKh`<Wx&jKzIV1k<MNTpC}hw-1&j=?H8
zRvIU}xE;Oki8tu^)f^>8k-G~Edyb?X2b)DHK-oG~YyPy1lG2WyXT%6n2at_D8NyyW
zR!2X592Xg_@_y}G(wQH&;{`@3Fz_9!lmlzj+b4&`#aTpr9jnn!)SZM<df5B>96$_1
zD!+0$l{uB^CJ!||kzWDNE1~ILdqiJyz@k?%V>(#;WEq6$Q1xezMoG`4bj3ua1kGKt
z?!9%1ovOnbr=X(v9)S4T#4I8ij-u0bt|XiVBZb8(5XR|=yM<SUUoa1LCQ0IoYp=Z8
zxz4AduF5<O!?%YE4CO;17c9M<USWBpIW`WS^GLZ5wKHg%R6qw-=1hieNflCik-Y3|
z!(>!&NJ}`XtHcI}a_jWRNRX!33?wq{v`5Qts!DxEf-_58X=lu`(=X!Q(jDrvbLGwY
zQUw5=lz*8GXs6R7Za3N??7Q8u$Lg#xlaiR40};ITKr~_YjDqHoRq=Dp;P*q~acBht
ziO2;8y2I($#@{cuy8V3l61h^-RsEm?i}j}JX*+GYGQ#VZIM@2_nQfD*g5Mb;9oo*<
zN#8(-;6SJC00KG6ZB*S)6`!7|%xL2z#yZT_4kVB-6sQckPMS=rd%W}z$BjH5F|dqZ
zBd13j-~tZ8{DnHI^*1@!IUAYhg;(mRJ<$w82h3=0k2|n<Wdv{p@hT=o`ou5RO<O79
zrRsnKENe-5EjiBJ*z3>OFBJ?czd|;AX}WU^&rW~*GB?y&c6-KtxgxM{VyG|d>-`Qa
zHbKr3IDbE+`YZQ6@glzR17G*gf#*JexS%M}?ea&#pj17WT3Gh0W!i-CDKD*=Z4`hT
zvtO%>cKyUK5-QEyw|fI)<R99v7i?uQLM58-aA3Lb;vGPT{EY^oDts@e-)XNfJhn0r
zLhL8)H!G6CjVg;`6pjUS(CN|a-7vV-eyiLhq^cA`V_9^1%3Q`N>qzqF?6)iNbo=@u
z;mm;qg>KXOv)^fEuZy{vQ@h9?aDZBFi@Sto{BC8YnYvZ}#?p)h8uR4b6|c^j#)OI8
z_xI}8NKIg%vM<WcaolHn#1FInfUE=i?Dy;DxZ;L2P9w3n7WH7j{-D94WZSgiRhVR4
z9W^I{{KIAqM~YF%En*O&HdeLMpBj=>GpY^RA5~TzR@gL@A&|lv=Eg~JSCTZGSQ~>*
zyv!H=aYJ0E!ve3u9dk*#=@0)(J~>dt6cfQ|e^Q?O<Rzg4lxt6zhj^v}t}^}WPYW`n
z8YobWde085+SNkXVW)-2CG5zMD1DA=`?JJqi=gHf(d;xYWUk{zq(L$W$$f%CCA%M%
z5+n-^FGYT`Kd(|qo60gA`$g6nphV_P9S$_2Ce~jxGsqrBqljNkd=igsJW}?TrC;Bt
z3T$#3<aIcpo&Lai*svL#m>Ojn@OA%HrRx<`PNGiRqyve2Q)rZ)85VU!!oN;(zIy)U
zmpj*S7feykMb?P$f~aVJQ{Pnr$~fYev0&D4dV;vwy`<aD9De)```ZTBV!_%0D18!)
zm|2Qocv&d8EB${2hG<t(wj!>xzpG$r`6OAD6lTEj9-Vxl<*#Z$e_z)?53}7Sr<v#m
z|HHDZ{6nthnQeVoh@4ig7|hpWo9rKxRjH&&lz5|Erb9K#6t2`{!Af$me<}lzvXb80
z?G8-r4&ivSs%_an7c}MGlEPf<Q(k){2otx?0#Wv$R^-1FWJN1!3BG$WqnW~LEisw9
za%#|B<zzXe$S#n%%l@^DD>Kz2pTa9F4Ypwqv45-gvD%9l!-0w2;S3%^h5Gk$#}rFZ
zu@3y&fu*otRH;YpKPp-3?W_O=?1d=LsEIrA+Aq_w2olqC-D?rXcG`a?Q(ZwZ85uTo
zT4%=r=F|;z*eMljnwT9N$*e7>oy4BGg1DEoe79q~{#4U*6vEIMRf>=$n%3#jzOZ!d
zV*7M1ytN3k<4!}f#$lRLC42Un`tmihNO1#z5C=HC;j{JoH?+w3kr4(+5Ms)}A>7&L
zD&}A%at`1sc3?qI;_W)#HwvrMgZ)9Z;q3FZA01PN<-pRQh}lDl>5ua)xBdScd+iI&
zw@0iT!br);b9zMVH;z(2gO%{b{A9}Dk}F8!674x#5~U*;(3hHrsS0A2M}gQOK6<p}
zVRM+YFXxj;dD=LDdEaRlo}q1TAMP4r+CuyJN_mHvq>gGlJiXg1OnQvuSCx(WYPnBn
zBkw>0sRN6rir-bmIG4+R7qn?JRj|X?ot5hJm+@pk&Z)>bkS*?OO^>P>$}CfkFLQd5
zGmwlnK5lk@F<`A~t#TTHDG6ipim=PrQ|=FZOt$b_l<V{Rr%&5c?{D_xL3R(C4U%=_
z$6?s*)9$ZyJnO|sU&a5{lPSJWzrS)OrA}zK5jr15v==)+>GVtPZne9FbUdTHAbjjF
z3Kg{vf^&NEE0<9aYto*1e__81XKSwF3<n(e<8*0eNNH0lnY07n@U!kO%z{m&-)W_l
z@4}mJv1ivVTpa^WtNCsev9NMI=l)o8)Pk8D8*H_mcB-M_xsy*&N|xgIa-(B<kyxKw
zzC3c#=~vBJNh`Rt6rJd`U%w>7KdOw(DQjuZyFV84n&t0wE{l+xc?qT5E(i@zusuI-
zT}Wj?E=BHr7!&LT6=~}v!8=WLDN@p2STLo+0+|pl1K>d{p!CGMG^{G%D@h0!q!_n-
zL!yfTpPM=1wBl}=Xt0w9LzQSRwM8LRFRI8Lh5tA$&EoY`9_X>KUtI9`|1tO8QF>O_
z{qIii#U<_)V;iuIB_x__BrzZv31mrtyEB@3C5<(j8P8}WgiW#ycA8^A5Xb2e>b-z^
zBN0PL@4ffld(ZuR_ul837bkbE`}^ardtqtTbM|@O^PIBJKD!@n2jPPpqU$7q0~Ke9
z&C_|STM>OicBN+*Os7K##qiz0YV*k9JkA%bAN=Zbs*|#C2`x?pd#!7}5(odcdg(Id
zqol7rw|=i}B=y7zB)F*I_e!@$eIlpgS#VOV&P76$Tp<URF3gMMs!EKGqIsX!O?5=w
zpd(6iXDQ&F`WTTWS7NBFGNDSb=eO|(v!nSt<iKJnG|J}Z)sIgvXh?tkHFKOo$~wz9
zu?ENNh4rc3h)(@}z2}^oXjbNM_)R7}S^j2#e^D)BD!Mdn(qHW3Em6btquAlRmET*s
zWWmCv1NP#?*rmNq&ZE#0rggQwq+4sba4^T5;=oE}*Bc|x$Z*eIT5n~p95~xchrNTz
zDS{42h4ivEBf-R_cX^+4rBk06!zzcuBK9#^t?cF5-RZoId&(&%uYq~)qP?P<l0$J`
z6E)m7*i%k@U~Jx4BYx*mdu4|so4^vMn2QxcOdsbAz9{L=WmS)^Bveu|(3T}#&frdc
z&L=rNxXj32)ggr3JktLcSPW07ibMTV-Cb06d23x7rq~2xgSe58iBpOL>SF-q(ghK`
zV6QHiwyCZdIPlbc4y@ruw=vR%Zx5`~X?snZ;MqtwzJO@x)Z~`IZwr;@YpZdsI1Z5V
z0rCb*X<q7yGj>J8h2^1=7@2S&p=TShD;s_flBtD4wol;s^hUd?bwnSh3z-c<NnBe^
z|E@`Y``2}d9oc=sDE(4?<r2-X-9Y9pI&YQGy&ueYeOs1Ax-5zOp>5ko7SmE+y&-?H
zcx0Hv$SEcFI6gkKMdah^ZY~sUE%oPK2d3p#0<^d^`;ED0h=&|;o?{VMy1vEK{pOLy
z(&pq=d+TMFJCC(qodXRY=;F_8Zig{=Lh|$dkia0w-fAR$l@^j;D#6Y*ZQ92tS38A5
z9Oax+k4P-q<{lff4w3FcZl|V<ZS&^a{5B7=^$_as0J3)eyld^+>Up~2>aU2mY%&?n
zLA_ok2RYKwb#3u)LY7_c6dQle6{mw+?E1u+%U<X8boV`WLpFegX>8aj<dm}e#OdXt
z%NZ_N<{KXf$3v0`5bJOSep^^6u5_R=c<s~h$-K13-qa9w&B$6ZIly}W$=)b~rcQx3
z=PQ*JgwPfTQv8Jk>f|Ys>BRWM87Xe;l)a_VBtnu0ol<<9D;zjWL)Bq#ZJ$~=I692y
zumg~6aOaYh_O@CTO(hxo;x7?=FwE5C8bT0Sj*Pe0+wq1ZxZvfG2A!HDL~EvpWR(8e
z`n56)%9j<>gEv@W=q_wE=WRj$4%={e&N^jLd3*^;HX=tBTi9A+k~2%$0S8tHEZ<^e
zDE4F@MYl!y8(p{}2S~|iUNHLd1bbg2ecWzpqet+UM0y4f6Hd`f`<<F@i)fF>9dvOe
zDYOao#6V8rwbx(oJc{%zyUT8F2)9gugif(`fx>5$+wL*DCD$Tch2zFn03sY*cQuVM
zZgGcka)OtI1)~u^R05hqh=bHelj6k01d#pPmUJ+MJ?)v%P4>VF*U5`0<KsK(B^}(N
z2*<Im0n5`&!LoL1y&+ym$MprQYrt|<7fHL=v!%1|z=?!r)n{Q!-<7kotPO*m564p8
zK?h(5!5KIu>eJe#l|g&KGT)X>lO09wR&XQm@dcwZx<)wn-0ihwDcfcAS=?9R)Z~)I
z=8!D|(H%ZS!$sF;%iA8rV`^PQ^a05USU|EZduJmwQ}F*3a^w@^WbvA`6<IkwMsi1|
zAXa5&+k(Zmvc8%j(K@&!C}aNQ+u9hdzTf`bPit4TK8)cX<CNm84wI-fVW4+1F?Y0@
z$fDM+ufrq5Ta8-P-K%G%y=2CA*`3w5@O!6MN*))1FuZl-(spn4nqP5y`kVeYn(L!>
zSHq%B<3n&f`yGG?2T9rNU5WJ{nslC$O3t1jAp1@zGj3Ct_S(C<UdVQsjJ;S@#j2jg
z@rC-pX)e_ef=I7t?`a6a4HttXNka7nD9?c@qH5opJ8ewSb|mVe7)@wb%kp+VQAO{|
z4vwyKo)X(uD}5-jrFf^-Y-SW&@JWeZ-fG0P3V*~RXg1H-2kIkO7ejDRB9IQotTVSr
zQ3vdU?XVvUl7=rH@>mBXhG}Gc?W6(^?CTote0FRsvxM;C)N8u$Sg_Ov6HPw!&S~Av
zOi2}^L&LVZ4V}|T>Rk>72Ub|5I}|cDAk;}V)W#5|5!vPrIIu+Zp+OS%4cc&#VC2IX
zz?1{g8+YH$OrQa`Ubdo0h)Hhp;Yzfg7mS`OEvU@*`)y5qjn$5o=dLgZre(bM-g_^<
z+ScYcKKQ}cJCD(4k)L$)kwOVJ&DP~wb2G9inNv)33qC@Kq>Jy<_5h08NYy8zXEb`|
z=#n!2ytjHyh?KDMiWHlwO(TWuhw^h=sl|2HlTJ}nef^DdY<<l-7yI#;ltW^Zjds|i
zyb=8t8)^!-rH@Jj*@lM5E5_G3g&2)CpR@;x%7RAXG=@xx8hWgF0M$#0)Vzr_=938K
z$(Hd}%}J`>rB+Ha1X~RDW4t<;F!_Q=8Lylj)To0qTv3ecZhC}FBwI7H?a2p`ARlCA
z9=LSDfhH@xZhV}pDnFdX@lh^~#G*t|(}Y6QCL2C+5sWw;$}tDI=OZbwO?9y3-CzkE
zm_9)21_Bd?Y-3I_W5!PAluC%n*x+P8mGEM^#-8UE4Jn%pEU6S-*fhM=@5W*(AtwT5
zkSumV>ag?MrdjGRN))6}3{ZSb!bbS}=?sAdIA@!?exHdTH3UWn7L(8Tgt<^wxvHdB
zrcxl?o%?BUh@&pAj|ZajLk2n9(k)#eV#w9Ow$;*7pJZ>z)R^7VI)+{w;sjf&aC2pB
zrKT`IEQ97EvedHel0E-O&gROIwaTYRDOa5hJT;91nLz4~b}giz`dR%JOoWW9Q%=2B
zc>ayo*}bh#%Wu2IDRdYdiXxFA=}UycaCl!|Q>Mc}H~7+D2bSAsCDF$A4OtojuSR-K
zN$k|!3(WH~WK_obLm{vzx$s2mk99BttXZMsoN2SLh=xTOz1{9>CqK?L!45f)SYcf_
zQO4PvZ@WLwJUM7at~Cdsaj%=$WFODRjp#*rOrm)UC(7useWDud-^W4roSIy~J5hCg
zV@(zd#2F$IBwuu`eEiU<uTfqMztO{xfN#nMNJ0wXp`hN}wfJ`X=5{(SA6Fv3Ne7^r
zd)IRNmiEn!>&Bfzv?<dD*2SRU-`clm%E`*&l<GoU%UP#B7ibaAcJ|5EE69_~hOn~^
zEDu%P$3QN@p(*>ehMl4O!O<XuGf2O8HDO&=D2t*}WI6g&`w%Li5vP<iW);TAjrKr&
zSCm63MbxtfYigV$GY9R#dQT#NG`$#_y}_i15D%pj{Grz8AujAfUBXG)fmV)BZ(O%o
zPS{(!jeyfOzdS9%3&D^Z!*K`-^+k$w@o7>uXg(tD!#VAXCbvv%bP8Ij#oSu*NJEz4
zaQk)2wIXK`ICOaJvv2RdDxoS8VFk@;uK3i)Z!~!de-nT9>7*m87B5-pJSRdDlOu-p
zJL>&7LwWxK)=FSnRk(yB(6Y}oBw5IXhZG9&GEbML^*cLQQqHsJ*MJqBAuui}1Be4k
zFgY@Wd{>V81}<jhQHW2V9n|H!`w*^axp7YbLuB(i=B-{n5z=;?ruB3S4&9}FPj-Fb
z9XH>Mn=+-`Qws>8|K93mN(3TVXS~6*I`FiJnkVf0+Meg&Y*z_rCQdpvYHo0?+)uwh
zaS-D-dEthgVoc0d2<}#lh4RGwfo66XoE%z5##)j-%Qy=-mtc5$!vr$$wl;rk@Z>Ze
zcVHs@cPOjfqxn2NJUHw;|4%B)fg*kHeiVV*t3D;smWx~mj+XX>Q=i&VqEJz>%dfLR
z$PZ0o+F_DyR)4tMC>U)Sly>HN-8lC~ug4Wck?-nua%e<@y9yn<Q<H)v>KSNc&vs`6
zZ@uib&SMe{Y~gp?p0=0QjuPJB6k~wlo5unx6-DlA9bji{Z)=#K*`Q_sG@E24g#BN;
z#rC!KZW&p<ddex*E|g_nEy+s^_Q>|PWf)vd%+i!oF0Tp69N-75ANv)&kOeOZ(Se$r
zaDOgp3@iuq9t@{;utuN|E9rG4ljfDzxdg3VI#H)b3c&B*8i(==vcZHGl`^ve3#Iqi
zAWoeRq@#u$2c<Gm%D&va;-;l*?UKnm?+Rm;)t9`R-BcXx$at{(PEDE{xY}WId}?T%
ztj2b<+XoQf*ut<vGiOoxP?OsV1*08ng?aPlIt9(d5SWFm{#do31HO{s(^D8Pr>4kA
zI^GeoWRK?<=iWHyTIady<&r1AjAM4Z&GJe(Nv9O+%8Hg8APWLI5IfOU2=-MTj)(C2
z1{PQqJDfR_hif1IZ6_Np+%k<{-JnyR-6m)W$Lv)0ZiSrgc^1ij_wrSCx;{3`$er^B
z%aBh6b)2v>t@BSlgeXoaI-e1L$!&)*%+9u&G#ay@;eP|Eq9&$98d2ru!{lq{TDKRD
z;}5vjDaSiPRo$y`2v^bb+1ZcW`+Dawuwj&3+IFG6D~2^T#!&~Bn=kZOY5D5Yblb)H
zjhGNwd5>M%>A+$vS|pizGR1{5&^*z-eVnbAffc}#_DxQYO5gN@^~FxK7Ms=Q%0_*R
zh-HS)4Bp*-s9^b<G%Wy&&D-aw1=<{vAtNF$a`RrQUrNH0%}dE58ceDV9hgdh|74A6
zFIhZc>G4{S!r=+dpV^u+cf!TWfVszJ+DPEGrN0tWpxu=wtyA9W`)+sbAE`ant-LIW
z!JP0-rGTJD+|Ii3%@RB8him_f&ymzIn-E;f@(iQ0^KQ{J@sWK2+H`O>%=GER_DFJj
zioZJ@i0I8q{3q;3Qq0p51|KSM;+O-~LQeL2@+g|JAFbhwW0l4xFJ{TzO`>(Dz7{gP
z!9naN?8ove<C{h%o#*tBB0N-nN|nL*$7@#D9jFWP_JdeUr$(D_(YkT_i8gf-+!>iR
zjp)LDr=~*w<>D9W6bRu@=7hr7;T|>W6m?c1c!f@Jt^HJPRxV$b$Vw@h&6_95)NViB
zFqm}fWbbs!h5sUxnjFY6xOCcnreRwshhl$wgDD-k96&DMjQ_K(lT*6B3+Fku@{|LO
zcC{O1SynK3V)4&)n5KI^QkB?eba3ips&xQCJhi{==Nn5qMI`7rw=DJ?kj6!b$A~g5
z|3Wv3v-zUppmb`2Du}sNrdO0fj4+Zt2x7n3zQ;aM4lXRB4lE}P12$<=f2p0?3sx*y
zxddqtKvF=+bGV@g+pL?pe>vCnZCv%vqp&cNu>4AWB&L8)C~I1S<xW64L6`KaU1!wU
z=%xdUk7;wO>+*`o0O?=Ll@_Ta?wnFgoKin=@sBRKv|n$_fT6g#%Ef8x)ab0!ej(Gs
zZ{&bi3@ja3z|FN3W6pHE=fz%fWbfQU?Kj)_>5LMPit|gnrl$Q8ra#rR$<2jr&wi`T
z>=0=KC!AvRId8ibfxKY9o%1%c?bb!kbF&{QANB_Oo%%r}4dk1+ggCHV;qcrFD%(#i
z?04%ojb%B5!n(m^U6R%52*Szl)g;JN$#s+zlTX0=%znRp7dB4r%6JBQgNd+&o&r71
zA9R?i-9YYp(s;HZ<Ms|g`oqMKd+r%=9`l?Y8YhJWh-`n9@7^@BR+nx#MNx_Q^KP_1
zt`0~iphHGiNdwCsE?Bk7{v@Xj#(2zm&S}H58}@a|{<JL^2d^&4;!;!EfRA))^ocDS
z#_Z4P6XJm*4>Goq4#tj=YLENN?9bb2RnfAYjWcne8s;sb|C~A3++crEE$8_bo&w%r
z@~68G>_Vq!f7z%L=0ddbPFd(dFc~>;)}hmeE&HqN@VfDy^H|uUu->!={B=Wy!uUwR
z0})9HbGK8kZY^Iiu+sjf-W97Ok7V5!lfJZZ0D7x$L`oj9lE|g{!41m(wr^+3Lqxv1
zxecgEy}>O5w+*aVuxj8Y`@3A)cPfPrnjVVL_oN6?iEg*Q?^a1?V{RT<ea<*F@pkI!
z;{Hhchk}Lg5hzSTBX2P21K5%}`t2VROP>7r%#JCiplv?@8HIoOr>wgC{U5pa%}yb)
z@lL6Gm1zB+Yko4&KqpfK&O?7h7|uKOB9p6g(DpBd7da$}9lCA^Z8g)S>q?XG@Rk-}
z5x241zt(_d(UAX{4p%izZp~a}?w+%Mt6szibuw~7rw$|ti&LL-eOqty5J|7;_`i44
zb(A9&ks0y`EU1KV7`@D_m{Gm|kNQmu(<LCmfhC;}Eqq&GsXPC(>-nkAa<FeTnl8jH
zP*38Av_SurV_I_afb*zjnAx^kYPJ8au4Tk#T9xFX!fbQI9(3xpN0%*FVfJ9m-Zbfv
zoM&zS$sU*F+UGo&7^3+RE;3u*2ZQ|#EuejFy{EyTn?skh+ks_oFrBZr&wH?)H*&-h
zISofNY$(*^dy9}^?DHRNw3#4g6ip5wkQAXsQDkVRFWMJ07>ANPO4vdUOw<Wm+rBaV
zu`jGI!h}l;r6_<{=L1en5;B-7R9U0;McEz`hbK^PSQ%3sSau86W6Hj`8(>QH%1sZf
z&TDzZNN=kxtQ;d>(mpUaN-{Id*#MHJ&YM5izO)stoiooV6f%e{7JE_-eD-DCCvhCj
z#r6g(0|v8r-oCtAkfbM_o1)WMFI0!rB*PJRk=j?Z24M*aK<C8+4lF*OsIOz7ef}#S
z4D`5u?mXu)r(z8jH2bPVmYHqCL(X%fAX;}FS@zYf{<1ea&rzSTh0+|HvS+p5M7R-@
z4__u=-C<j|oiLK42y9%8#sT>3)(*~yaw*V32`nkg_-X~`z{5VLeR=U>=Q#$Jm@5uk
z%!Wj;=jOn!yJ4>L7@w7Y0q89!Ezt{(vCzQU*HnFE7-3SSY|bc<u#h?R@sW)Dm9%$)
z&K*WDe_nos3yz(~nxh~QFu5n~`Q3PD^8)l!X6kdsZ&gap7u07$+d~KCnaj8?opoxo
zhe^9@FRagKqvMy4hvgmzmPCvQg<)NnGi5Jovo?&&%P`Sm4x~1iGCXLdK4vd&0~e!A
zV<(w)p$M-0tnnVFK1Itka-bKI?M{XZiT0u}B4;nDx72S4F}fKv(P5W|!3rl{e6XFj
zG={Z=i@h|5jcSl>>6Bu2bx<mdq;sXRmp%A&O3Z_+R}*{Tlw$~K`q%qM?ByLy?zuvP
zgWh0L8*9B4SN4jWqvgFxWpN637eF~9sNv;FatgWZmDQOhVMaDHSem2W)zF;uEySYo
z8)As<vK$Uq3n{$<r4A2BsRD=mguN<B!6I@$ZEy;CmVn-s<uUp#rb<-h@(yX_W*9~f
zYC6Wjsn1!!h9GSSf9=)P?cCpN0CC<rH=LRj1LM8{=wMt8^O{DR=3jrUQ>Y&6R-IH*
zzqZyi<FV4(CjFIEGIChU#(ArLc|;mRIEM3fMb{M$P90%7J)$er<W44*di~_|#74Wa
zQ7%-n$deQk69N%+pdqZFiG9qj>gG)O_S=26+K1A_9ChkN0xNdiUe^e-vc`;Z84&;>
z-;#k9_WIU~-iKt&mFr>qZ1`kLewQT$CBMC)^-p@RDesEGPTJMo=!FqF>&`i_Slm>-
z=$w?)rAUMb*S9y;Z)9#q;f+!KF$WgE5%S6jN};UjW){2O=2ZXW(Im=G+cl%q=!Bci
zMK!mYCwDJfF;$kp@)#ty?J#*3?3xENbq}qxHevjv_aPa8+lu(wgJn%TgcMx}>dr4>
zVN%yB`+<-Zb(hU+wD__&InRL&$-IR(md!7iD6>Q#Nglv*j>L5q>Zb}yih`4NZT$w3
zOXND{2$&eZc{j|NW7k!^=vBsDAMysv)UB|Mv~zv+de#m7AXBmrB_uUErw2!+8@}Pe
z(g5rJ9W4Cvw{n1arXtMYOmkz0BD#-6&Fim3AZlXJsbI<lds8=PNhe}S9`#w6%q>A4
z?akSZUJo0yQ&2;QD|hQ=5Gd3ZdrLRc^WagafE<{52I<6kfRvkk$Rksmt~<!ZOm|P}
ztAEnbh*})6x7BE3Te8GTM-xhaZ#B8l=|?Xv^n|^=n@sWKVy6+BAdLH+nmC+#6ZW-j
z)>bUJW$`MfnC#H}>#ny2NhSHw$xC~jQs6*o|8-3Uotvep3#&fQky0N(d87yfN1gh>
z$i~;FszHf27X5EE$gCHgnmElw&06?CO@Wj-OdHr~H`VW(S}7cuVI1~3H9C$^rI9KV
z<J3Sm_1Zgmu4@|E@e&i9`rLT8a4^V!@a9C~r8qb_&#^8kGK$LWwp-eu!>K|>4%q=a
z%b`;rzmdjWdTS{>7I*6_r=U2T9hle(dD}9eCAmHm8%CVRn6<P8RQO_u7w_mky<P0L
z&NVfr^FS^?PJPZDGzxBw;Osz|aG}5nhkYeKwp-hX(PT=a=m69c?i5or!qWOkIAt=j
zW1Lcfk&rV@F!kEx6T`htwyZ6qy2s=y%8(t$K}=1}g2<(Woa1&|!J2OFJ`}mYVz8+i
zPU5P-ok#8VTF9PGLoD2sqs3gjAhWi-gApR7-9gM_gK5pA!F(K+^_?{_BG_{0n`MzS
zN|8x2*g0?I(YJgM&z*6U#w*%{vexOisjy#ob^acf&x$-kowu~NuDfQwt!#K1yFe%)
zS6^j@pr-bc{M+)Z7XMc2cUx7T%7_#-->hZ~i!)kyaKd>@lg||_Q+7u#+sw9&%#Tyb
zZM<^%61%hUWcUoPhfq?^%3DSz?XFgsckMi<5c3~I9R$g2<*b>#t07xzQt6Pg%IR1O
z#B=InJ!GX22ZXz(y}Jz`tkI2ah_<40-~%;!n|sN&plplK0DDhcJq{01TmCw*=uep8
zVnt8VD*^}Yz4f;IsKnqL#h0tW^6-W!Vkf;X$s5~O$pf8Y%sVG2x*f=9Z~>W7Rl|IL
zeNY5cUL_&Y2Pf2O;t1iFj|DMB<g^d8!SwF#4G~m>CdGlpcM9#o03~!T1&Mtykqd_n
zl=krpASDro+2reL7p}p;H|&&R9vbP4T?7k!&IY@h*(*u)p@U9M3M!-6Q*Cw4m+B<U
zvNyfKb_XeifxtUrs5;J$f^h4Im)jyvNIajsw8N?I+Z98&2@WTAJ^27!-UFzvDCc)E
zP7YhI+RIS9%*B<@$ZVGmJN5GJdHw5bP5T|Br00}s<>jiH9)v;oZEeGo*(G#fwXCy=
zZnfP|0xC0rt!urOWEIm5M7>09eg-4#Nd3N!mXJ1^y-sd7)N98~3~pREWgkk4Nq$N$
z#+5z-KrMDA;7Ybk&ysic+xkXg2dCFYKs$h3&LEV5sNk9uqN8nal$42X56jeG;)yET
zfNkieHHmCoPt`|osV_YW^5x+0l#M-OgpxUwkTgSm6|zKqbg6<%rSGQq$_O-Gzf`EH
zTnuE5m-D`MIT?qX`b1oWSOV=Q+A=Pjo?P#ga-%a(BUq@!ynQ&S)|FRYezo%~Vu|yE
zP3Geof@65p;*94Fyi=_M3eFZU0Wzk+GM3A!PtwC=AA-Q(Rb?9s2IA0N6wn+6z;c-&
zM+!oekKF8iABSzaeuJ(?yas#!qywv^m7J|(M!!7A<RSPGdAq58qhlyT`kruLLY9#Y
ztBLvB+%2{83UIE4>oO`frzSN3%WwrG{~OA#yK`n1<1*+x_u)vXF;IXUwk?ffldavm
z!jhFM?VdLG%P?s;rCgX5+@3RPAE^d~*Sda94knUOx&6E~IB(uH_R;La%(go=IL|S9
zu`MDRazfSomXEvL+rBWR`$@bjhL~nw-_1=(-{`e3I<OoS^c4h)U{!I7`|TUr+k7RA
z)aaXm#oIcN6?vfVr@Z0qNF(%&d!(JR@Mytp1NO1%bcP*r+`{VgUD)i@Cw<~n-(>gI
zCo(`lh5-psFjP)W3_eX&c~qnlP@KgXyT3#-Eq$?%%&O9_q^?eTOl~CU_zl1A{32Qm
zu~N$VaJJQc65oH&KAuxBgV!?IZgtZj9+@>QsZo!^$3D?5OIT8nra4xE=L}iGsgG&X
zSt}t|3x;xpN&d$AWoRw5f58EbPi78DMy;OXPuj*%1oWBg?VGA)Z9iR@onGqaS=g>d
zifA^E%=JmHzq#ve^dmTjP*-a~N#sU-k}3%5@F=#cZ+Y<P%UMe%=wnQch>Ql2VK`Fr
zHHAG9yUx3@_qCg)+d7Q1-nVv3dJcM_t2`CdD8&jjLG<OVb;LfII5R%6#(7Lc<73jl
zDu0G%=c0XE)o9Ixf}Mx9QzOyh$*pwhAvlsxb#p64e!5Uet*S&V@-(Kt?^1=l=ZT~$
z4>YEd$!uJ@xDPpeoI?{A=;yqZ9YjYb#!D#%a(W+Z(}zE?=&J)r@2KA0gSxFFXzoKb
zawXw7<CJRFq`#Jn&QaUi4OGdVK;Dq<h_rBOf*Ri~$9H?UL&$>>#dT=rhyyExoAv(<
zZJa$)HA?KI8zz+jPssO#Q<EnH*Lb1Fr<YUq?Ong{+v06`Z_UD_Itv|*eY*7lCNZyz
z-RZzW+R>3K4MV?Qhmt7r4|eH0+C0WWH(I7t4b+shBsod(-g5Pp!+G-H!oaiknbzy2
z<CtVNk<P2<1=D-_4vW3VzB8%lLc#{Uo}WC(x%6G#Oeu#rTs(i{s(Pm;srABe68`Q+
zlh+UToI<Qu$^mg1rF~CF0ehq!-sqI$9Y~!{NfCDY-a^UHuCpn1@1{hY3iEgBMK!Oz
zX0Cl-KHjwCPUo?(Lpc?tTRU#wpW{)8uJc^I5kJBhxGMJtTJJ$R(y3GYpA<-r47RN`
z4kBVhhgzKQ28+91Vvlh}d-oAc3VeixcVKy9l1~pk$~f6lx94kEJbDPP1opr>7A(Y6
zSr*RBoU$E_0bDaWJwAzG>p-H5````acCw$(Ko{^o-q~h$luIbk@=&h9=*9xgtZcQq
z19Ml^2<K#&I1Jwqc;eL5$(h!<yx|^(9l-`&yX~&wX)PsiqJ?*0!hkJfgS!9S_S7q)
z#kv6NK2#{ct`f~b#RC7O*;9j3gyhN7Z!Jmku(vuQVuI#XW)}XQj%*zVKzr&FZGsiW
zQGiooU%}WOPsxi-PSy>U(z@c0cG&)Qk}gtud`g8j=m)fQn92hUi|<^r+$q%Yr&zvZ
z&)X3~q_G`rU4dtlBb+EUu$X(1UnK_YuEDfJ4Y%Xg+B{I0gy77S>5=;E4z1<!yvq(3
z3=WSy!m)?A4lH_=E=zgLY9-?UKv{4k2Z?)z!u2TSbj5V;#)`k+juuP;lAP$cdBB0i
z&bncq9cv#5mR=`_vbJ|0aO%6gB8#aQdY#-S?XgCZV0_eYs0zXiS@B&lMZA@}Vz`GN
zY$U&ZJimV3yz8CE4iV)he44|wGS@LNXYF{K5+*nl)$B)L(W4+V^p%{o;2(#voyhg%
zZUg8Zg_^k?;IOER?9iP6W+&S|TFy0IAPD7pKa$2GLTFu%@HddxPPMt^bQY2#d;2U*
zynSGSov!z=LP9+w6Q=_Uo_h?lv7M<8C1;cR)utLG%-yNa@nQUUlrAX0lVqX8cD5a+
zw~om;Jmml+p1$o3b}lKP@W6RatPIwvpD#UMHH}f|P9dizFel7^?0BN+?Lx5-53=MB
zz<yCK>=Mz#-S(;b?P3ikjcmE_L1U;_<YB9CdLapRiN#3(p6F&vo<QQHk2o;R*8AT7
zzV}~kKUgn>OA1&9d*#4lsY*8owu+<a_(c1muA|(&4Yr=fSO`qYD6RmQv`cLsr)0Fp
zm;*pxBqi}E=|elZp{xp&`X(mh$!EInKLIn6VZm&#F!FFWmv{B0zzHJT$8$c_mLarN
zIbBImT1{#gB}~A5!$v%2Kb-w1>%_aAXLWaG+q@gEwWqtnb@QA;^cHCtxk5bR4*QYX
zr^-$sUjC@4vQ9gmMfjO?7#ZG5q~SMuOWEc^8E8M+W+0B1$X@2a;%!MB&HkMW`eR)W
zxK@=GxWn8VOyaDRygF%ryxwC=!qe)a<a6F&BECw$h`!Q(qUO3(`Oq4q!&%nbsZXSg
z6C#~TezJ|FaH7la=F|k{XnutmghKOI*@$F;IbuK6t!$T4ochiw*r^XJnnd^h)1SrC
zNt%(iF~q~>4dJ|1s*dS3TvgVypGiOj1uBHa0jSL<@5ikBv)!!7<c|sfcNkzv50KcT
z?ULJ9SX1ohy1Blfx1}%f4u+OU79}Nu`uW*WOO`_2EBA;aoVdc;gHC<Sp>~RnKK6_K
zLcwG+kT)UrUk4^uTOr6?<@k$<c#&k+Da6EvHSEc~{7AaQy6zAs@Lx(=LZUp>4#Z$k
zu7!}0m-HwXB}v^c*XPwi@wjXP2d0Udd*j?|>{nV>Zq+?3l%lm5E&~vnX&B0*!Z~ff
zT1(Z3I9bC4;YGq<!seW}v{d87&a9qR2;Q&dL3RDy8=S{JG|L0<*Sk%SZaj)ii$lq&
ziMNv*OR3;Ds*mcI+>w-juJ1x_r#_ZB$%(dR83gS&JMI;W`=nE>u4IzZ<Ck{XZ#4#z
zT!TaFiK=rT)tsyy%_j6X@BVhK+vZ7}iJgMx^c`^N_B&m}`iZ!e6Amo-{wZXIDf`_T
zMEKJ4$^j-%)2)0n(n?UDbfHfIXKxcr^?TJg=N*=g(2_f_ShjTXX?_~gQN&$L_WRv5
z!h18dr!MVqVDTGqiJ4YU$!KJw6xVXt{-DhS>OAt>a<BtQpO1yT!D;f>{b74!O%H{s
zLXUwYkFp6P9AVJ7?EO*KU11U7aSX1(WXD$A6B$D6k8}PA2^w>rYhjdPUI8P!?N16O
z&O`axk@g)}l7!o*H*!f{&;GQ=6b5kingk(j1}XDM8|c&ryEgAS`?LHW?(%m#&)IG6
zOH%$puJiu$b{H%j+2j-w3qzVrQt%h8KTMexjq4;FSWXcBYSWWz?Jt{%J~GC=-~{g~
zW)q+gMSgDfS2=@YgPTThewV`okY`|<)WeZN(EhsnVEDL88|1+9{2J|z*x$4liZ|PV
z+8tP-QRfpxUuDhs+qV4j?qUt3p3=gn57{24zQ&CC69h%2Z-3X}@D?z_t2i<^x<yGm
z|GrTYt}GmNO7Z*fQlNcB$^ApY6h<BXH+UQXiyo$LN1CD}XC1JAtRCuE*Imr1mX1_6
zFmxxJ`c#0dAuT3B5pgHYKQ){jT#MFz3f2xtF7C#`4XfprxWWE8mw>yk*E-K4#UoSl
zJF<VNZ^}leR0iG6(Mjc;dM)0xw7CD;MguDqto)_34lG8KAv#cF3}TX5cG<tx+uG_%
zk4lCS2PT>@JU+5o&L98YEr~RJFn7(Fhi7T+sISJwHcEZXzU2%$X8+L|=LA-$4SyY&
zaGm7C_MbJzp)iS1%9UYzD3TOU5Bsk~Bs6oqQLn<xwrl6ixBt!&&z(DWj`N%}j2m&F
z>h@4uVx`NAMRQ=$M-GBc`5wN)_Bjv5L{H%J11+HxogmA)0*2-KVW0a@trU@d<5&zK
z(SV{kc_uUh3EfAa`Mhd?Ykgy-uyi{xttE~{BWt$U=Vyt=&th_>q?H3%d-etOTd5nQ
zb0ph=1JiFKZm4Hp_)rqZ!~>l)uv?*k6_MWM)Q1)+j(vu%Bxm_7DM4$F0@1#x-iou6
zW$Drpuxy#?LSN>sB;K?vzfsI(`Wt#Q#CiMTnuDy-9xautQ=|33OLqixw=a39{K=Gj
zY#iX&G&2n}EqRgbT~3-`+NLf{Ea7V4z+$4MKVZA_SEdm*fPGnYPpY!8nZw!SfP&w4
z>TAWpQ3fu9sV0m|<{v6@xNLlE61RTJ^`_jHG-(+MzSq9up}^IZD;7DA@si?PbU!80
zE_CE9zI<O<HKP5ZQWBHH0PWNy8C*X;Wnb0CyJ37dUGjk>DdQj^sz=hGeRX?_tn-=D
z42Te58kIawn$N0_i3G@fj{BaA1hg6@tKnAZ_?yhpv)lN9?E&rfa@Gd_@6==mr%2=X
zoW#W5*!pqjIo4h{t=lNNt6!KNL>7B)*L4XnF>(i%r_A)&gc4AEP4~`$;M2ryG#G*h
zU6JL{?Rnk&5ely*f`yHad#eflikT(V7r#;Vn&)?IXLci{RR$))a{H>q_JSPBP2<yR
zooCJ88t!Xli@mTm9~=o5drEZ}><^wBe%xM^$UC!bvA74PsGsxo^N0^Gt{*^0854bl
zYXx?>-bk-F8E3R7#WD5?SvewmN&P+%HTYo%CJHdPmdgNN+SXur)On8132AltrZO?V
ztXuxLnJs2f?(YYEwnh=o+Kr0Pmxs;E>o=+G)moL8y|<b^V;t*y5L3`Aa@0!ohQ^gr
z@-q{ITSThum4zOCf!fK@F=X*G)K}Mu0@cvr+b(NVlpuXrFqsMfY2<4sdt2;P-A__u
zsTBp5=&4Bc%e$!y*9i`ZV-BoSx`zrVqCVu3if<j1LdbO1UY*S>8dQi=6>!Kx;s2WU
zfi=WbQYd7IwPV(I+H13iBWnilcAkR*<P4^XNnTBd?27iwPys|BvjY>}ko%p(%dX6J
zxKrj@=Q+QFMN((^`Af&_s%m?Wyv~@!?%p47HKhg%gHBTM*LAp{l)U0ov^$g?t*@q2
zUnD7beK&@@8&0;06f3DIYA+AVrUpC%wNJkC4Y|8kjN{XqkI1%mb>S;lj5*J>=wjBT
z*1FH$*!n0wRLh|>LQs=EPEBL;Ls*rEc>k3abE-8hq<m}9>p8IGtaKeUmx_=iX~^bw
zq$)fo#gWkE?xrrMKF=8x^czQ}xRmOeu2aF!z&y*~qQqmZCh#UGayqXcw0Yg&WH=EA
zf8K!=Lg>pHTyw9P-@Xu9&14~U0PlBdWQF5<B8pOnogJK*uxs1qz>F|PNGrL`03}l+
zo@1?D*L_UN44wJ%s79sX)a35u@`Ba6dFA^2<;pvjFTZ`oDyJCT4F0;zzEY!{v>VzQ
zV2oFIgD98G^RidM2RQYyqRAJB5!-I8)-qXT(lM#&_B%DAzgNBS_4cOx;mkH}i-QM{
zQrLm%c*rC%=;+@3P^dQ7-t9c)HXj@vvbWSXW!4TU8ed{yHG)V(ML%VU!>DO*ZEvU2
zOWch&m=074jqpnLw(dPS5QfFyfpz?izJz0{mIV@$imA7^^Xw)r{|3J)$82!uvTGtF
z)4sOaNX-U&8AOQ0u>3tRv(Ks5pnGF+RU`MyR~wCwNBVJKF}h5umZ2uOTNhT1q!g`>
zc(agLyw&ud%%VF_*rG)Dfk@&Pe04#z7!rTSC+(&-pP?iaQzexa>0z82>6-`cvVnZM
zb(Qm^b^X9qSJ}<EQtNtyo3=QGq+?}O1(ofVf~CWQdc4PhB_a>&`V(7RA7@vvlXBMx
zosT#*iEf~yY~6})UebO%Fg(I(+!a@jjBXmVcXVxO?@1Js3Hvb2D>Zp+OQ&YHwr<I8
z#2d)p-e6ifxISD#cG%K3Bv^0pNSq`NETbWC;vE|%&qxIDFRO)z9b(C&|NNbe{%iHQ
zLHg7{yhR|F-PUlPgm_M&W|F`Jy2l+Wc6)t56C@A+{ZM6{>N_=MNOY}U2d)e!9J@hT
zwUXE>Y<XhdKzN4thMjU?yWB#sf#cLFCRa|ui8t?T6TO@`|DIEf*F(aqRm$hkfZB>i
zutvCYk~>w{;2j7OCf{W^`ju_&2u>ck2i9NrxHwSlcQ)0QQCpQ`M>K#dSUnxehJvw&
zKoa_j$CcgD_ChF0H7hz5P<1;sY7TD|(#hOe{g#DN#!@Bc&W#f;{Z4&D+A#m@MFV|T
zo0<@=XGe@U!|{H^sn6fQ?n?&QOUTN*syn)AL_s~6Lg+4<EvF{Q@f2n%#Z|q#0p!yS
z6Yl_on={*Bd?xHYt#t?nCzlq+f#sI@@bqfr<oDJq;?3mt_UNTYy}<}J)fzI~$L)Q2
z8n2eMi|1VL($r>J{rwURG%5D}hJX_z!%iW4v|K#$2Wsyq7V3mks?kepB~9E8`(VM6
zF;`xeGu~jR6`H4teO>WqsU*=(+gAdT)})dIId4T*)Aujj5rlD1RiKFt*4K1_r^Mp(
zMArfnVV&%KczT2^QMS6h19PE(^9wWE9heB$wb#$Lp~gAfMJ_UhZ4%t?0E6)69#-s-
zTzI5T9)76o+_R!}#|?z5W7<Od#=tMMVgyTVk(Wv@KfQ3}P0nL{B_oqFt8$3Zwx;#(
z8t1XLTSUr*hhK=X)uFZ3h?odn$b}O#6QRImYEtT2OFR@?Ze7>X0Z>2Gu>6&765Tk`
z3iX;4*(pcz@!l)m;5<erlPHJEB-+1VAF7Gfj#CIDT2BWSy%is-2y|J!<342mo}0O7
z>$Bk-=FW8<qv5ptS9;-?<I$?w6US)_xH&z;HSvjm81<)7eQjv`8*(!Vo>?~_fsxAB
zC2NPs!&t$B`m5`ZNMN}Wxpt65{*yM|A;}q(0pWV<z-laeMPZqFxs-Qdur`rnxqj|^
z=dmV9?ue-h(VZhZ?8EtzQXKObos%nr)DDQ-Hd#O1%?3bBNm{K*aBAZ1ys0Z<A8o3e
zGI@W8iH^S<8cvPYdL_X=N_A};Tf4!t;ey9OmRLcJF7#V4w0>-S^Js4v7yUT4rc0h7
zh{1UlCQNUbwoPsJPzj1bSMGga$(7v%K5uRWVg-JIPAPO&y7lMo`l+T|hc8NCpM}Z!
zFbs;er5zYhdV=Pql?RlAhlD8zR94V?3I?r^&4B~&fCGzpmzXNeqEyDo9eku}<W|OU
z5Ydsr<~lXGEx3VI=?6dBy(#fG%|sE>;?yKIVTh7S0xO5zTQF%cb<Ak;fu$UQDw5n8
z<bwP968v=_OKtpEV8yKQCqwE(&FZ|X$?i=J+Ba0=+EG$m;iBiWu)3N9Udr17Ts!9)
zrx0iwN<kJ`u0Yv<=IEq-tj4FWDW`#a&!uEJ?bM`W8M-2!Ub;V0wv|lx5~&sV*nQQ0
zG4%1Sh=F(;#9^mCTKS*#!I07-p~rTAoAov5{syO<a?CCFzamKT7d^no3-%C^m=D3a
zJOr)*)9hS{JHRLM@y2;`oJXBXcj8sT=KaPjrGVs=3MbO~rTh29!yU43>Q-78J)-4_
znUGjaP4q_M^##!KiA&q<n;VUoAP=Z6aRd+zN2+!fRr+?(+Y|OJ^>f*MWc(m?kpnBb
ztxUw>k5jrA@oJCSx3+_P!Q|lTDW_Pk%b{4rQcm`V!_EZ<Y@h7=xLlAfBEaO`Y1C*p
zE?T-^$uj%4T8z-vL&x?*f3@?(vP$71wct7DElnS;YC_geRnxIUqftyya1Cdin%ag6
zWeF+79%#5XZ~lCzkeH9z5vwVu?SpM9(h*ccbzm`Z7)V1x<|cq7M)pv?&1HLB8%e5M
z?R{W{&*?jXM>B6rg(2nlPTQI!W*JH4o#*JF@Lj$L+$UfUXZ_>|gpB9-vg&m#rW}ql
zls(c7EwboH8utbh+pD1L{g?LJx93!i^hT#J^HHh^QIvt)Q}^1ZyM@H{XsuHe!e0H}
z57>8fZy#r8agO1#3an-fN{7*4(ms<v7+yE%Jc@eGY+JR|zB8NUo)%54Q;d<M5U7zT
zfjmw(_Fe5OW$Keh2QV!#Jv_?^_}%#hvSK-peUqEUQ_*L+vM6W&_jE{)X^$L14j>3R
zrYig1Y!x2>V$PgW%mqJ`IIbDA?`yl9Ok`_O4^qBd{SYQAJJoFB0*ye9iOdObMXG;)
zwYQ60aq?(5g{;<K><4H<#?t-3Gw6*(`O?II;>mO<Q3-EF@1W2j8iT;Lt!8B(*-j5R
zrI^u@!@w=s%3-1H(QdBOZrN!KeYPgO6~Q<rPJH3^`d$(TP91!K8VouJbFOQ=j@yn}
zfJn>DtVwX+!`l+x&ciEHAJjB)weK`ymO9Oq?W|TLhY}Vqg$I9Ub8xr%3=0tpAqT6Y
zP?k^_*seA^1ZRsturdy;_KH*$am7RJ(%tQV9=IDrbKK_+C=l$^=vzK*d%CF%FGr3L
z7F*1hQ=|80wk^bJVSDqlTph_lnThEfSiW%+b`0CsO?_Ni1H$A5A!nVMTu_i+K6nZm
z*q_*nnoReZaJ>@%@!HD61|MMu5_tzV_lBM4JoZ=j*iyrG@S&%l6+J`JnjD9p99caQ
z>5!->uf-8=E|A}@9cnVd2Hld(TgpBQPzsUvU2cb~Yb6DT#-hQ}yDvSAjx<Vfn{K;y
ziZLR{8rE1&dp6L~f{E6qv*jTN7JD`sWpQyjyvo@K|C(cM&JaBmg>)V!D0w|<YR-`D
zK@h0AJ(hFw<R{r)JjasHZcKTyJ_=ELJm12rdeC`JHlbhSk-a<m%wZK9;CMF)@`VXU
zF9#Mq?-%mud0GqEXeSDb<luGODMfE3W60jj`D!P-`8Xpp46NS)2Nt+TzBmZ<sd_`v
z7$QT&VR(b7XEF++8?n>*-i$(6VqI}v0H&qBb6~})B`XJ(k&@KT<iJVJKkhtN_q*E;
znApy7(#{rjLlMo8sR^Tc=TXM=%*O2UZVXPIE>hRv9%bjMeeLxYz9ha<nONYOoccT~
zCI&YJmu}}9DscWw-t7!Fxd=|Z@L#SiN>Oy7UhnG&or<dvarHazt^S!R5k>LFknCdH
z3aOeMxu@rV+5rOPQbLmQz-5}r1!_;UX=XnTPYn%Dfat(<B(9zyO=)ruBU3u*DMVWh
z{VEEFWla0~5B=ZP$HC>)C&Iwgh<+*z;!+)(Fl@caO-?a)fTS(G0@Zl3gQZ-z@6zKA
zth%DRe?-KEfv|`iDoV3GlSAlDt#=-4(}eX{VojjfNqedqmKj6B9Eo5YSTx*6opW@o
zF)M99T)%1l@v<HOt200cbMv<nH!BrgdHdPZ`AdlD^ae6;vN%&-FxuALcwtGfAL;tb
zg+^)t9xQngYLfKM;XnJ)dM{&)(*HEJwo{`?oY@Acj7v&TCj3~}$KbqzG-w@>pfFpt
zuqNKp%HMGP9Q*NlyzV;ZQQswim>?kgiTZ8|zkM8mQ&YHhfyGPen(Fk>zS&Q9^Sy`t
zCu7wSZ!n>}VnV60pK7|sSyduRNubfXzN^I2+u^mvcS|}#;cq{ks5*?UZi`cp)GSRo
z2MZ)bQE*Jd&m_8WLvC0yC{~YC4-gJya5wT(!!gl*wq{K;r9||)Xf}0Up+4a)nT>fX
zT7kELZzOSv>_OnTGG%b{mi=7LA66^r3@PQoMp!FXc-halGWnM`(sjzwTE8h*-xO-g
z&F2@|*O-Tj_iIE#2Z~c4*eGcuLz-%JHD}Cm@F@0+xxBA^?Q5OK*dUUFBn@d7g$m7n
zsUZ@g+SC?~XaaJ8#oK{uya#Cvh1x0m<*I4qLD}Er)_mNliCvxEFj}s|_ABjh9cC>C
zonnnTq`<KKCC{0-^Q+akczxa3{TNz(=`8m-HPyJ>Po)w)Yrpo4*0P;A;sWaqd#jhB
zUw$IGxlr{4#6fO%+OO9frXE6mwV5GV;i}Dz;;i#l5-aQ!a?arf8vBj<au`kHZQXep
z@fOlCI`yLXx}IFJ+;4WXnVcBsYIO0KQxlD=^FsG?2*khDWVs3U1BLDdDIYNQ*>4xQ
z-Xu4`D&Ts4!?pH1-Hdd%PqTQL+Ep=y)JJRkU{vID{@A5`_PYgBkc$+MC(*<Lll1!J
zCuVkz+wZmWhLnt4AIX3mP_&bZN9mu=+wa%6@zZ8ch*1z->+bV$>T^dBuRo5v$se>f
z*gZ<Uj&$$9M1#gBdShe?_``ZbOoVj%`=KPjM7zeuhwYD&DcC%+@gC<{6F#$z&>}MT
z|8Z8$Y`bZ*^BkzGzBCVWwfKbnN%c6jFA~nBY}dWAI`6kT^^&qa@V=`)V1HVD(P1S+
z?gcb84HjPEjQAj4PJh;oilqcIbgL$?9QhhT+L6Zpyv-E7&>fIXk6qN%sSo+AI9K|;
zM>rPhNm;-{s!QI&{-Q<~S~Pl3cqM-yM^#BptdTeXX(hJf3i+3_FoXe7zf%sZsH42<
z(dtR-6miMPLjF~^OGG%bPg-}*(jX#v8u6A^PE0>6&R>7loS%jhNJ|Md@wL#~@ttE>
z4T492?h#~Ve^WiwYDra1*o!w<o_;c&?6bd33`H|ZVo3@$)a`IdSef9z|E|H}APF^`
zcQECF<2_xA`}c`C8wT$l*)YAqDF(J^zUA0)1Ui&%Isedva1L7~ka7S*55fcv*+1qp
zp73*==je@&hzxc)X#dnE5o3#(2l<Pnbx^2?um=Nr)Okx-jM`;DS1tT=Ho9^GnVsj<
zjc;BzGPNH6n}4Z^6yel%OJ!zv6M<8&jj{;t`Cqf0>B-?O&U5YDjL62OYyZ~4qzOo7
zeHJDYmhkw1{d>)g3<#XZ(*LqTu&GW>wYnX=&^qn0|ELcB55=5tnN_7;)Ayl?-dg+5
zYB>*;#4)|K->Hd~n^HvnLI2fS2B$guBC;mW?!-mvbFRc$;6!5oof{AtacnK=b)9l!
z$a~Wp%(j-r817s*LyZlMz_kI*F<~pr)EdX_bGF7#eqH_Nlr_mKH;&uq*6Yo9Nvc9`
zu*T_=?9V>0y@zE84;mfT4lFjjOcZ&<C{C7)zib*(<FO>6eng--VxPaY$Qdfy@V;Yn
zr{b8J>agy7lxNZf`+}_{Hc#PBhbQOIlmiO;__xmV{cm5mHTJJmjOg4bC1XW8LvHKB
zb`f6X_C+~<xLa6Gv`v)L=CJ72o9v6*w-()g+o}~y7T&SSI}oni;)ch1_9fN-Bxtz@
zkgGu=YNG#1$^s4TOWT+@9<*a66#y%OC6dUo4ha#vC>I1t1z(n(8n|u6?YB9FK1L>$
zsX^WC%eR&sw_$`^K~}TkK;qp{lEChWA4vDbuc%K-I}JyIBq5`oPKw}7o%&cn<-HfK
z9LFdnYl5PKp3nk?Mk$DpePwp<mT9cZPNB%5!l+Vv-9%O_3-(proQPtiY9E+LEx5xr
zw6C5m!K6}(In4YYp_{O0b#3T?SBLjIu=uWsvixU`2bS%Gl}sV{*;{K1p=;nK+ez;L
zJ+f?bXNx_jCM#VEH6=<GCj$mGAzsY8!JgYKUn-7HqJIMx-Ip8War>IBK?R7SdzbSZ
zoli?$It)Fp4_z8x7HSr!MxPnnG;YtYA7<DJ%ZHavgQb#M*mmAtQ13}Vu83j>_P-zq
z*D=@&8?vt%<s)NGIo^~_S3b6JAjA3`vlr!%xU^j{zm#$x%5Kj-lp^xQHE6M1$DC5M
z1onfwS`o4Mm~HGO_1k?gy-J~^v`fH>eHfe`C9T*?+ooH-a)DFO#}on9-4*<@HcZZE
z%*Z;199SY6lw1*iX=RyBkn-|wH?ZNkCa32>+HMG`q%IhG>VA7gLyje*qhzF-bO7q%
zA{<7D*qpLg=4iRacEiX<r<4tacB+qQu*=%wFL{UaoTTK6E8lFdYK5B?yxS>M*E?T#
zW=4!<h-Y?r)%P)*)XSe%e;@T$9|uY?8`_FuW{CW@S8pxek)aTJN--BFk}zv=WOX16
z-(J%?GD~b|Lcz&#208ULW5=YZKwY6Np&*&p)-O}GgnL5T7Z0VR9Eb!?eWA>Ou&myP
z5&DWY%j?D`-~iS-fHsfdRobo?6AN->ZiVY^yv}*lG)|$L8n&w%MTTD)9Ve8i=YV1|
z`?XXl62=K@udAM@Uy30=2)Y7G1e8vLW?$d%3j?+s2{g$Krnqm!%dic<p&QlC*a(O?
z4opYy_-YOoaQW&i-UaSzh!jN|@a5+62YX{H-8Rj23zYIyCqq?l1DR##bo0@v;8|T5
z8mA^P22om4m0~``l*n6>skgaRpFR-U&x#dC!^&V_b)5P{f$+siPeZXld+nNTdN@j$
z9_WYzOW6bU@zfTZ*R8d@+hToLEFx!hj3W(DpR@Un!69yXwE0;^dl8$*DF-e{Q>7z~
zJWzIRH{;4Q6NzgbSaCU%OeNk0*EPWEY5WH`odHx!kY$F3!QZZLoxN?Es5+;Zh{v*t
z2FbtwhSouRZ#O~DD2J*c_m}7%d<kNLd>fC-Pyv$cwHw=F!4L0nO4+%MIJmO1_NIo4
zxcd*RxC_Ou0}B56Gne+*n`_1cpJ`EIj?TedPEEG7i2G<1E%cV^a?Y&g^o#>b;<50K
zmG;)QdeTbJmh|tjdfwH*b>UVBFD~9mV#nJW`CEbK30D)3Ignf??$D7=>Fpg(br*ur
zYAQ#ytki2Xq$nguYV{y!y7VSeJV@8+)YOv1<+E`%EXZGO*syM`^PITE1>0!h7Pf}^
zsW!tp<J6>UHTovqSgV`7ZBgs<0C@sek2>YR;}HJnts;UrQ!ZV!n;Oe2v>Q}D9d0>t
zpxS;TW2<{a9S+oplPT9~NYkSWcta8F)W-olvEFX34?QE_Ae8UCQ<KZ5+ts=F$Zlyv
zimP$7e2PNiaB*siG@a)g!Ar1rap(hemB(yxL!q_3x;okcW$RO;J+c5!*^(T=u(;h}
zrxfFlGdTPvWiKbtz~0g3S$7xqoMMvJa8?Y22D#BB+pTS)LdD2F!##S`f#x!;#8Hx@
zr-^>!%ERImqOHcpD%)3NFtlauSF6_b#)q9^V%YQ;`GC3mJ?C?Ti(h!GHp&e@SX|WK
zcKhtQZ@}$bX1^&1nB8Yi6^o;Q-Q+u#w+UJn-1n3Nh)EWc6b}Bu!$QINfgH5oSqLgL
zIi_s`IV{-BHpv2DmM+9v(RSR*Wh<RRPRBy>aZTCERyTj{HBKS-#jOi%Red}hmS|jY
z2Ym`mpREt4mrzB>RY)-nxbfK?Tc7@T?6H|`|6l*_^hM*s33jdB+3o5RU~_Ps-e9`n
zK&A-gth;OL)7Pl~pF;HdKN^Aa%b>xD$tmqiQ7rvJ7wm2CYBYFiaCq1$i6|1%yLO$u
zJFD1M$6<~s1sMtwOXp;=druog=BMKZ(!&lc@sRUeUP@v4U_krqz1be|o66-wDY2>#
z0mWl-=5e^5w)eHIJGEhaTuGtGteV(9HoC#yUtg)-fQRY#bi}FA5J=2Ukn;y}#__Nn
zbDpDPsX%I~2yxmUtghs#T}F5*=ouyv3QoP&ZEEx2gneBNS%zSlDz;zR?!Xe4Zz1VG
z&j$0AnQhCtonVCMAawV@(!(5IsI&aL$iEA=x;`xFPmUXCt%F?3Lq>kN)VH||U#`ga
zA+gDZx|!d}Z(=YGELoU7eVpbp+{uZF9By3?UyVr7;0*=>hUF9eiS_EeJ+vboIJxE=
zSQ4HU>$u!?%GTtRknwuT>v?9|l1;WY;WOK=_=xjd7!%f$ryRS<y6U5jL5$0NJq<K<
z>UEq<tP8<^q>U1^RVqic(he+_5f^$<qh_BuXdlXz-y(yhQ>fQu#M1tf1w%K0tncfB
z6p~6K7%^V_)jfXHYqjUi#gsPMR(28LZ0nt(?6|O!qvON~lk9s#-xwvoE0I00__W5@
zvL;4QwvBb`ExmHsFIe@1Fq!ghIpNgDj44D#HX;HaZM>~2dPV68lqso1b+CTFQ(v~_
zNnQVrCUv44Ae6YUik*ceGwwBU>NUrR2#PlSaHG772X1u=(SckTkcV9UlU=7m)fZaF
zXTl`#oU^GmxKOcm3b~+eQp-`Vjk!cYuC}q}NN0w0<B`EqyWgn^vL!ugdO0fMs!Pfh
z6y>Z<H+&2Y)wom6of0>HkxTfg`|32_v^CsA-ul`9ITghv4vwzHEMl9pgE$YYc(+sN
zHtYp9>rwtaVRz@$&1_q+Y+!{`%1yCU;ZNhXr7Z=W>Wz)z90^4H&M>Xt)25ak8Fni4
znFR~4xeeUHpt#GA)HI4$Z5q=QO{JCJ=hWv&gAGb%C?&>^b~U=)K5s)26MNpN$-yYj
zthaI97V??)?A|s3k-8rG3LwV=)1@rLUte9xgFkr$vMkCNCbtKtJ{@de21&4Z4=nFD
zG+qPe95PG5$;0{p46G6-i2Kr0IC_~EOdV8)7w)t528+0>fIVg(%Na**BDmWr#o$q-
zK|a)yj*A`WY3{4OHo<h(^byv7r#`pI*dU1*j^GP)f14EDnE<D$8*;Wd(8P+n){!3v
zO#OH+;p*|#&SOW=gp{P~v`_R+y0imQc}OO@&#6fPW7+Mu**Dft)#XQFD|e9Z4p<B(
z59lTsh(vu;)1`iRTKL@(c{*^R(;c@hvTttVSvk!%hIMrywFMMEApTuCezc$VEj5RV
z@!R2);+P0CJ2Wp{<Mu*KA^04;-`cI#@vwLh{p!HtH|iv+3rww|sj~@NKG`s51(~2x
z{ZWpYN>jD^a^AkJU~KHrQ>ib&9$-NKXPnLd+i%;a+I(?j?QlvluNr$g_b8|n!>~P2
zHEG6`tR$8gj>4(Qqj+W?gh-@JA!$0kquw6O`C7DM$?{c8P&rf9nX_yJ&$)*ZZ?O6i
zqeUSbLL01$tmRFN-BTtjY#pa<Yfb7tPD3pC%FE*oro<H-dZa}?Vh`7QGRrG<!Kq7o
z9awIO)uZEUaq?Vik7T!25d}9z9G_DzgqKHD2=Mmp9qcsXJdV-?ZvyM4X!gWoneEdJ
zDZ@<B{2Mhz)FiRNdqfdQ-_c0xLULz1r4HfLqON=iq66B=p7WV@FyB17Wo*nTww)PA
z9+ySgch)MR(7g1hw~`#lkK1>3ySzu9Pjp-Wl7+i#aCB;&eRsAu$qf~XR^x`P2?wS%
zS~{?5z`m!wF4G!`QS0-f(28D=hsgJ~?LU9cbxt9MBdd<`OMg~63;Vv-0Q5t~FiiH|
zVA{3HsVXhh_jm8fCYV}lV46s%+IjqzC+!F79mS}k!a?73+zU3djXdDE01&R*vu*9Y
zU_SR@%0B17l7_?$5PUcMrpRJtkJbouU@MK2d>7ktigr^Uy?6-u0`}Inw;5W1Zg|)!
z7IDgJCUgN(ZiOA~yl+O$u;F$du=$qrR+43I$p%xWY-b;WG$p^(qf&n<l!N*_n+Auu
z^^@@q+pfN!RN@3)CTf7cx-f(K#DReoD{fz5yW3o+;`=FG9s(?B`n>Dr+n#zMvY3db
z6SD@aTWBpLsSx=)>#w`FePaqe-jq}9lhK}$r3F4qN_R2)``Si*@&PvV1D7s1knHLj
zE~npU`?o%wd7mdndiRWnOD_DX1CCuGw`?q7(CcG%pgP#^DJaGoUmfY%o%;BVa$3uc
z`UKxU*tK&w+%8Z;JFpmNx=d;sVKqZ$60@??4%KgTWy6Eus(KBqCMR;O!)+^ZpU!{D
zSr_ea9&hvK-sn6AzUW9&VvHT3o<KgbBds%ZI-*=y7GTkErst4Lj11-x@E(!4N4pP)
zOKV6S{YnJtV;LkS9D=@fgc|G<CaROi?N}0k<x4Rz@SNzi8LjzRZlUPeW7TC|mg_=g
z-DO^!nwl$;WI)Cue<lP`4k|FJ_IQqa84M$jiC`;N-ekuci9}5W0z2ib99X{6P9zF&
zo6q){lUF#!ST}@<K)I82_+))ods{xpaRWK4BR6}ZU7Y%kY`6I1;Kb}y(avpf2fLk;
zC?7X*OxfwS5pQ3<<Tf2$01_1!kw<FE&U9^RD%G2IdnGlkhloX?iLkQ;OBbtfV{>59
zE8&qGPCEYtJ6CVlhSR{k!PJ7Ry)f5yzFH7QE3_+h_0TP+Mhw(~$w|A=J`Ib6TN^6A
z{SGWOq-)S`j*dVmNi}z|?Y;^4=JB-zj{wQVU)-BS3p(7p$DXK9rkjSm_2s2=7@3Zm
z7`E64bw-|gKiG!I5MT_zaiwiS*m7#(Z8mE#4)US+p$3EXmfyd&DKMeJf>C^+?NaL@
zk~vCTu59Qa8}Q)hO>!zcnae~vq!IXBr>KdzWx-OLspdINpzK2RQx2>!Lk>QfVqORD
zetW8$+dRo66DH3?O?Dj-%!O_Eh5vAC8xpAVid$G6SnLK(Na=BTJ9CvN(DLcl4or$p
znnMh?gE2`2@N&RDVL#Ggx1368Efk(gteM$Ps2A4WzI^F`{b<{PP$=<m$}O2fRh)57
zz50$IM*P9}W}Wx;W8EB-ETNMd_Wit5qr+&!II$1Yql1(7<2kuV5yR3`QH<H^*Mg^%
z@)J4j#K}94HC@uzhM$uCWc^S_nBv6c<rSWWPR$j3@(uqF|FfTJ1H!(%4$Zc-%mo9}
z<R`0vAF-d#NgDhRw1?;92cO&uR_YM{nd+lhTWv*Yc7YWU&&a_DJLddMH-PQ-v-O)m
z3gJ8Zs=)$xnRsM=`?;<i-qG@fu0;NjQh)An>SGdhG-PsZ#kc5+?4PeMs2%y;$(jR9
zi!FPP{X*-(r32vr4+VE%*&5ui99@+2GR^0|*tK>7d0XiJxC3ir)FI4KET!ozivChA
zBPW6Ln2gB~iay;H{L5_?2c{=+N`!|8lE{@|v6H-Lzf%3lS<0|peP^FjBkc_SGI&kc
zujaVWt=-`~#~6Y{GX}Z&Xcf=cuhkeNT4`nRN%jVlcRKS2jd{#|y%Das^>a|xJV5#!
zgVcWgjW(J{NP|fWK+T&rb;Nh^su;rGtU1bU%-^C-6-MIJ=fsdFf}A`X?YDB$-hS0p
zUXM2|SY^N6&a`<qUhfp*Yl!zT<I-P=F7B}3+4}V5>Sl)6NL!{jy!6i0m-|n1-W6?J
zjT}L;)7JhQ-}p+W5bFx&i0(4q&$z6>es60;0=?}c&XZ2HK_*VQ^nbs#vY2C-f>>Q?
zYwjHBZMHvXmGkD`;1uE$BJ|3!bc|1gy%0pj7s388U!F#&pL7bj<EC(RhIgN`KWg*M
zj#JvE?UADdbE8v}ht4n_5_y-!9~T-fT<JX4%*lbI|IqxC>X7!H7><k5yY_iuQXezh
zZ-<MW4*kh|?MmtmPjzX#{b|=$myJ1gt{8B#Fgx`@)8u!n`9&FGf7Wo5Hx%wgB(noc
zOvaT+z1201X*>J#ruiUx9ZfSipw`CySHKGoyJYUs)=vA2+#$=yIn{X9s%_4ku=ba2
zE6u%e&b3Y{Iww9zS+sP!!e8Z&xvTyP=Q&o6dYH{5NQvBG1+c%a`j!Pjo|=)eJ_y=&
zr(Q?uirbgl-&Ak)^R%MJxqaz?Q<J-UfQvosZ`=CcOaO1htO3bgbJrFv^IH46_6qKB
z6XU(HK?jme;21n?f8Tz<$*-$e9Dpbx{t$QCKh%&#B_c~E^cz_I{3Zkt`^RoYXCpc#
zo|Y8v)W@X7`iBzkg8fs$bb#uL1EPf-nD&H{$>Ghr(f+yGEw3TiNSz<a0UhB)pkBRL
z*xN!P8T*%-ukwe7SrA=^3wLU2(eztdfCc&2ZdBy+;=NOs4m+^iNdtEi)j?E}{ad}I
zTmtyii2gY+^^%;ecs?<?|DJO-XWsR5o#*&c$_?l`W)9nbWH0%~ZYVRQ{KnWg$Ekk*
zpZWVGWL$Ed<M*6i^gi@__FwH6cnPIL4v@rr?G14LC<2w{<*5C4qR-&wjn1P8*e$(r
zOzQS<&FoTKHn1d7j_*XWLRREFVe6k$FtsGs8tUl4;!DkBs%yg|?}~kH)rgKug6gjA
zb!uV@<f)YIFL+&tOR`*9KCix`8hKjd`RTxF+XSu?{HT+4Z*Z?%5^SIUaNr{$E`!c<
z(K{vt&(2}{f`^Oa9>ypbfu!h*0p?~8jhVQ$4Bt3lU-)n%R(*GS!QX2bx_>&3L!KF|
z^KPQ#0>VhRNBxT)j#<Slw+JuPZlswliqn*JP|6Tt^-=rcho4TtJjE#v0SQpb0g;ye
z&>J{!6;)CcUvt?-=C?1YnF8;^qf7Y}I#|%vkZJ8pyB6X~GEOALA1dh!HHlC1iD0fT
zqEq^^g6%B{mMov%U~*wsR1cD}eR+)zT#i#)w?zea5nRJwr#{!=CU#1QZC{Zs&um+X
zBOgJn!svoAE`^9{8BW<(R(~_3PIjtDWq9eRQ&am#vD;GDU$C#Le)K6AG>gIZGuuvm
z;b@ngpMHwF`08p+Ur@?zcrOQ**x9!xnsZ$>q`N?$)kY2fD4S>4l0%UdN(!gGcBqWt
znvbLQ?1$sTSb+==BkvTG{I0<{L_7RBiM@+&UpC<NEC)q<?!$E&hmVI63p*gu|EYD8
z_BE~doNwYbamjXIqIe@~;r}M=c{TFfY3OZHs4H-Ik5d!9%-iCWWbK$ezkRKDH*S`Z
zKMRm(1a9P@_6w@Dc7KM}OAN?KZ#6kjvH~89mE3PHY}-Qt*1d739J8A=I8}SO%3wVE
z_(iQ>_&Dlz8Fs%nnDo*l+tFoO_TrpTFbNK4y;CeSmhK}GN7zf+EN>!lwwzG_6b%WE
zFEVZQIoH`ss}bq4rLB8>X1fC`SsizZb=cdqTrbN>MsSD8_A2p%(^x@X-cXkf&YfpS
zf}pO5pSPM^j6%@3`t%iTVj~xD>K#;n(5WeOm#1KOjM*zYm;@{gx*Uwa^0=KC+&pHN
zwMk#b9iba~PB}J!^lahy0J_NUW2e0;e;4YVolYs<lef3t=t#C*-sYDrAodQUxOXr}
zLLQnI?A0~J{krGIK#O&wKFzM;Qe*R61aVQ+Gq${71YbydP4_jdLLxSpv4GVWx&$qS
zVz135;^>>?l6iSLbAAD=*SW_aVh)j5A~W9=)i@LU)cp)nX@-CWW;{cbhEAu%eA2FL
z9SS|2X!AaAFcu|Fp<|c!+g0_RY#8N==nW<&U4E;(F;cy*z6VCb1nMl1Vw8v?rWU&F
z)N45>#y5*we|<w_l9Erz1_waQ*Yas>2yZBQFgWBq7P7=q1^(`}t80OEo&8y-r0qPj
z?Y7?C_Qvk@$B-Um_So(X#w7~TF~#=J$tH&2k_p15l(cJfoNENIIM1z_PK1sWuPhl+
zd`?aDXSaAK7MR<q?V7f4(PE5kaf<OuBLQ+Loma3_iteT-?hO`VTHwkqn_rXIuVWSu
zt^F_2lzI&oM-BM2YukrHzK+XM9a!K9jdHXXB?!mI>vG65+unH9Rc~=hfgP2qo5mG)
zq*Gc<{sFr_mtbt<Zs)NODU-)$Fq&+if#r5XTYyc26HXys&hIB{p-g%5pg=caH+KDl
z#_EgUtIo$=PEF42%npc**pUnNrfMy>Q??_iTo()IwioJawEIyUr)(R%h`qVd{i$AW
zIP4w{NUJ=h97)6ST6oLDPtO6sx75EpHNaFj^OWDxk2v{voe7z-=cx0$v>6~w4wJSX
z3JQB^Z*6EWGPaH|If^ND-aI{)c|PoIZKmd3JJ%`1cx4jS3809<x3?yUS3w&_F@~3D
z2&0wK)l>J|*LDM!XRD}@g2QASq$U<1r>qVmj(rK0ig{R&3p>gkDLmGU%cv%);j}F*
zm~u1}*XzKF?nrMU?toNxfm4gxNrK50V|(VO1gO*e25wiosiBUt1ZjOh<n&kxUeT$~
zmV@`rxzCmdvL7?shVU81Lu3P5Hektn;gh~$WSVQ$Z*EIU;!!yvIDkAe)=aZOZmGd&
z{uIcm%oz=qnMC3~I9uGh5o$vYX|}c&<h)Z8Q=}GfQo3}EV|q#Z#wya*I;BMTV4X6I
zsj+>alD#A8+BE4Soabmwhzjc!jNYx?7tVpgp-OOIIf@a)cw5>^w=v01sisU8y9gl8
z&0x!_r9Csm(=gdPaCvDkFa&m10_R1$ty)M|pR}nY?od;7klP+4BHv!GHWDPYa`Nxg
z<f|Le$H0l$@_JQFkZkqmVyP8OPEAS^gP43VSHH7s6!jyUPq#4S_o>NoKlu<|;w#$B
zLJ;tlXJejXq+I~MQ=c<|X1Pm$geLx&t?W)@2`8yTN@0$53=)Ual=D{hD?-x7Y*qI?
zozT)%$h06Mr^Jf$PJOBerq<XU_3hl5DJ<x%LrzVozc!3W60keF<qZi4B-!V{^2C}q
z|3<s3UdU_L^R`i7NjlIaj-XM8J$qN{TyW)Or<-LsysO8(Liu_RilqrABzBd(yFSA-
zLPQjuaN2<d>S`~_Z;~i&YPy^1?LE1XNt5k72Ih86LP_4+%|Ut=D-nVy(LSdp-WJc5
zQIkjQeQnT7Z@<MUh(x^a{co`M=d7at9dVwcLlW+V!LsA_n}9!1^_uChTZn8)$r%Lu
zq_=w359L@Ujc6VwOuv2b;d17}ZwWzdC=$vz(CBR*07$^>+OO-z%W)D|u-6+b*?~Pz
z-N$h@*a-IU_-43qib<hHh}{np*y;}1&mfELLY|SS4Vl-e$wekq6So5L@z_w|t*}bQ
zRD$%ud*!*5m?QTHd0Fz7ChuCV10U{M*CpO5F~>3C)Fi118vw@uoLsNlvy-?y?45D|
zdI=&M^P!+1TT?%j*dy<D?O<SerjXxM$}kDV+D{U^c438G+ji1|;jrRV8PO2!H3#jw
zZe5iWLM|yfVwIyQxPNc;$yIAbn=4GJHb>fEU{g>Qz=-bh4i*ZsU}Nk<^<LUe<z;rl
z8%#I@@nZiUw)OQMGaLdL*yg};>DFx-p2Y28(ni|?v5`5AM|;B#NK^kog61~Vn38@5
zJJOCu-DT|5Cv_PndlpgAWBJoXT$aLP?*ATI_VM=KErdg3RiK=EhbUwmmQV-pw26jg
zLxbyaeV~}XUpX*fAMWNR89u_`9GGYj>k6MZVUziZHIsvDrJiugIn>^oHJQR?s$m9Z
zP{Qqp9e@UU$Ep>!v8yAP8>_AZ5aPl`Da0hQPfq71UPJf1o(IPWlG>C69kPHW#VIA<
zG(@Hjt_VhxIK?5exf{XuP>Fzd4lIwCWn{e{$7}uW?xiEa*yEw#z(Vj!9j$Cyk~Q$F
z3KUf)0c>x!r7fd^Zx|cP2uSlIra^wA_jF_>1B|3`g$aP1X(JwV>g&i=@_!+cbUF$J
z{7C&$yi;)7kUND^>o{VkJM}r!x5wp&$UPrz924XR=SgBbSWtk8EBH$5POf{a*-ZJY
z6pOH@&{ADogAvZD*Xk(z>g#J>)Klp%5zZZ0Y}bV8ugHil49M;D4fPu+MFim;#2KK$
z@;!wB?Kdyh1z#@bxo-|c!88-hj(w~l+bwv;PC3Q+xQJ1?*Yl#HnS$)ueO*H`x3F|j
z1R6q+qb5lfJ{PKi%GmvlVm~SW9eKk%2#*UirxnHL7EE!&_VKpJvg#8$0Ga~IO|Xcl
ziE;ZxD=wva%qi9xeB%xFjrHZ!CG3UAIOo7(z?vi}ej?+X2)BJx{&ayzf>S6udtR}E
z%-cTu=K84!Stk9}5HX!NHSsnAV8|=z80}kfF>hadoAa1^pR_P?bbeeur42V0tXxTs
zHm97!T1jrNo_(@@64CB_Uzf52D{|dM8nFX)DZLirq<vfKAGRzol@q_YFj13d(x9%>
z8}4x?eX8`va`eLSp7qjsIBQCR*A+EK&AZ!qx150z8AZqCSSeG73}V4O*aOu)Df)F4
zqI~U?tPAA=*nP;U7oSCTzP<KfH@%8@PoZxXCS_$tCOlL_$`eE_ODCjbmYU?+u70a+
zefa5zXnAJa|G)ls`jva8PyO1&!n2o1=h{Y>Cnw|5R>$Absmam3>N0yIF@sAULo-M*
zH_0lToaHEL-`-YtXk>D7-8huS0p;%(z1u$B=2{m!Fs~PJssWabU_;YkGG*UUZ|RIp
zW6cEz)(8%qx>iXpr!x9W4RFLhQ@@cq0vB$SLXSJJ%eBhMb@iX3`c1P+B*uJa{a%-I
zi7x_Uyup-uOOhD@F7{n*uiT=`C~!^9t#vCmLw<KP5!wxii=s_{<sq|yxO(zK*!Oh(
zRLX4;%UupEUkdUDasJ+VDLLuTiaD^_aY9oVL?$pdMliAVeQg?tl`si5%YoFAhu<r7
zTlRWgikimm@2^cjB$)GDq)1A|WWP`(L3@9oS_vN*ey>)7|4V}9)F+i&Pm<kj)$vql
zM?|{=i>6b(5y^b{00vEaw5vJB{6aX`pPIb=PEFzjxs!XV$H&*(_J&v+*D*dQ$CLv~
z9W%1KJbgk&+|j7{0^&eNHo|)-NFHdOemGB~*0!AwhbzJ5_c~8~0}sxe>5*UOu4F7m
z(`TMiogB(=C~@Yr?QTQAb9&S%WJCCH3$>KED6#|Co~n_^E^A3~0L(fynmaE1pig`A
zdvYf1ZE#9;UL*TLBlqR2gm=L@QOIv$2bK3ochiRL@6hW=D2MUoabV3_CZy-k+#S6T
z`gNHz?LfD|y70rOm;FYn@>ZX}DaE6-z)^p}4t97VrGZWXxsNLTE5vVa_1d6HiE}(G
ztmF@=z=lIPz(vgWm{W)WYUid~o>ahu%G=?#4zNS=62nix8%!HT0s6}IaipycS28(;
z+&iIh$`uepkQeKnwWIYp)yN|_m5`vQw;FEq6RRP_U=K*ib*z10U~I^%uoAn}ggw^q
za%g-|hO4j$0W9?BA{UvS3sA1dt43pyh^!l(!?^3z<T<Qeu+ffpo9h(Jnt0LU4y>A$
zPX_{H%qo_5JCSeWsRZ5Gl$`)3rf$))JMCoKPlzF_$K|o&fP`}7h`aFQ#|fY~l_P}Z
zLVry`dJ|Rw{I}D66DEfh-CG6!0*6IS?Gxp{k)G>_oyp(MY`X*hcBd4Jky!=xw^&^!
zRsqpxJ8mEo?TD#N9V3!pC`c8_Ax+x}=iU4olc5yRyV<wr+I-@hNX!pYBBoS+iqsVE
zvy1+shCl;!{u!vj7vTTG)Ti;lDNtX(iThjRg=rTWjT!E7V4)R^!w_hkHjrZ`kal7l
zBqJYkEGS<FYP*;#J4%<OP^B2F$WF;aLD?~3$&cC-Id!YU6T>ORd!gB6DU^{fkFOuB
znkdzQ7l)iv69Z7dfg}e4D(#25B~UV7`418r6VWRXi0stIV(SFhX_u-6oeFYPS0+7R
zv5328<O|l6k)3Q~Pu6enWfGeyDZqi%D2aIq<X8Nv3|BMVBrA>29x{_3^9Iu$5%Nqr
zkBQ@{B-^k1sPhz~(jrOgdy13lhue-^v|y!EsE>u>8Rdj@8{EijPuEYI^Ei9ZfoZ@q
z+it&eV1@li>pb{a38NfX4LGl_l@pHipGpM<OKv~<@YBzW7LX!N@b3u!F7Qtg0cV`r
zJkKUZHgYVX1NpJsilfs*&SQRjYI<zF{dgn%tH+Vmhn=$a?9Hnd*iU5D!s(IG(1B1E
zXT=2@WA>9dc+#V79&-xW;$lVb|5Pj8JTke#DaAngne9H-$bLE-T0XUf#Q#;e2YRr_
zex}g_t~GZG`H6)%zSz&U8I!vw^8)HQuvme<R9Vaw+7SD>dOLlZgWNi>=vIFw0dd#T
zjEYMBe6^Mco+KPiHew|;feI4QrGZC*Y`>7bV~d<XKSQbB;CgciCz;&r7js%y47}5M
ztm?yQ&!YO5+VGb3xZi+MydlX|fy$hE_R9rB5r~3Ee9<`vrUl^>kK3<Q7Zj4Y(<znI
z6G_x1_55lzkVheQ+)H~LSaGZhsmPSXxw)i{m0w)vw`jlC{r-e>cyP3*9ayf))mLBr
zdi(XpSFV{B7c}JnVoBt$f+7`OFw}l%8Ga*o*1HFsr#y59Nwha;zgb@iTLoV^#9$gM
z^=pa;-EY6uzWn6<GZ&pwd??*=W!_Tg{BfLNf4eawp*ZeiU1<BL)W+De_d0JCHek>T
zZJap6&JD$Wr`4DGdDgi}lEGo8UW4H18(Td+YQI~9NISJq<S_@9eCy?Jc%S`Vy&(M5
zMFG;THJCWkNDR!O|NFJ5n!3YIDU<49EFvHey%z!ASn2+tdM6e`&KNO<U@B5xq(0G!
zvC1xS0*lEXw%roUoGvxXBmSaO6LXO`pZCxOOGV2l)$Na}H4$(r*YUk_V7d3g#~@rp
zm_IbW_Qzd~Zg&?<8udqU(xxWoyylb}sFQF*q?I`3PwJN{sYVepBL91w8e!dn4TIb;
zXMdU{<*NbZD8*P)WC-!W{;Z9^!YI(Ld%<S5amiHhOwh<bFI;JkGh)kf7E=1o!;`ma
z`^F_kY|(x87uDBXJNZb+F-6yZQWI~hMu*uEv@L(xjr>AHsL4etGbHQ8IOfz}rESe<
z`Au>mc^HOF7&PY@UK#dRNnLK3H_v%YJR6)?XMddqCHUUrRUpytb=TS7RF9uY-_D}o
zYc)A?om$c{!(`jvwz-9qsenBPmfLqZ2zihFU3O$>^|13;E1|d><)Pebf8WhU*d9ac
z-C!ZWLL_v${X@MMu3XFjT)6`)d?A+)vm2M^m8Fd@*gtl^;l>4U?9`<r-e9t^$rb%5
zEY3eQD#gt&+z~eA0CET$Fd;l@|J-4$bUnM-EyX-IH955_R<5*v>4qXQp|H%o4oofK
z1*WSf>|bYh5-tJ3Z8DR(t$7cE)Z<Qlw`&lq*j-JZ^l$A@zG=ydfkmsFa*Un*Tw*v(
zNSMkpgWbP3YPgW=U7S+PeA?e|KxI__(P8MU$N_~zWk5}81IX5eBkM-{&-yuhkqn;t
z>%ep?#@hAvUsWr~UbmBQ52AOlAZMY>u>a1pN#W?ubD=_Jw8Um^_DBvs5imW`gMD{u
z#Jmg+NvBRM-IRUKBf*oPT6P|5Y&d3WkKu;+xdlt>Sg6h&Sk3LRFz_o{J3eKf_edO?
z;|gi!Ia<?cQ?3E_`H$pW_wTsobD+Rb>Z4@|mXgdbif5BNU|&$bjBCb#p>3jG7tq9Y
zpx)}i1n@!nMJd&hSmLdn_JxlWA}+>OJ>?XmbD6{-%*6IZ2~+4UD(n%_JO`Gd`w)3j
zV2-}{k<wEv!J5%L7y!22!|uq~(p2Y!p3}bMk+LI_re=yskF&*rW*4An%H4$}#=i8C
zVl5OD7rC4qP<WlRBlQ4eO+pQPN=W~ep(J?ov}vT)_RDfzujD>L9!1mn#u3~Lzx<Kf
zX{+$Rp^*D?(d6KojrJAY`pi11cj^ID#neZiB_GOx1_hFRWoz$F{I4j)0z#76Od@k)
z<8;r_S2dhu%g7)5VCcFbNKQ?y1$UaoZw}g5C!XTv&Ye~0+NK;xQIZiH&+S>&)70P+
zi787Vm{O_aeYXBI*teICh}qLU9r4$mozwIF4}QdXOhyZj&nbIOhumle+D!BF{oZQQ
zH2_vA*gE;PKDYZ#sIZZcL?orJp{6D**$4KteN8QgjI<Y>QWCy-*Uh!()u@=bj;bm3
zxnyK~6`gub!sg!UDfD6X{Dz&2l}2i#Q_hWvB!#gwJa07nd+i0ShE1Isqcnx@cWR=C
zTyg^OO#j{&RyWX*(2f`br5_Zr4v14<-#N-5Dzs8&8JGxrQ8$2D&6BpNR5_B4TYVYZ
zIk@CsDAP{d?qn~nmJ{iihSYPS7@{UuY2}jl4A@KBKI9@eJlw-!$^lllq(KmWElMsQ
z&zE-H`s`~lBy?VI>SGeXCL}%45Vx1L)=)Ib1s??}HF7VfCZ|kdJzqG*+b?ejyIHDa
zrQHIeg<U#u=fF~X#UpVLtdhPnjH&>luZ={9y<Q#+_R8+NDKT}_)v1Z^hV08mA>o%@
z)=&WJB-T-<oTph|UIB5Kx=uqU`KzjpcKs`O=D_L%RN|xN`I255T(@DDKhl(r8AKHP
z1rnMl?8)Hc&Ji}%-V*J`nv&di@SnL=u&)!?^y>C4xixomI;1^u%TB$PS>Cu%{nzAq
zGA$>0o}-uPp`jC6r;1eEnu)Xa+O{fBrIrZGlLO1mb?w}_c12CB1R3FTXEtzZ^cpup
ztnb;C^%^`ADw)*bCGt{J^E7)cB$5zhK=RNIyQ<;9v~t{xOs#PsdH!$YrY`Qov)4Uy
z+50F3HF?tDYbd3{>l?l!_-t@WwM%2#HAH(uLx&V3nUKTYVBs*#7|AVaS2u)`wg)}-
zuHY@S8(U4A#?<+s<*_$b+g#xqYLO$vKL87q&CFQIqiFl!9@};!!c&)`9bgV<qmsg{
zwYk+~%6<~RcX@+hxehr*7<x^8rq5Tch)1i5kFx6Ou4JU$HZK?O$}8W<aYrE)G((ek
z2kX=GYvEEMA#I6VQIz2~Y$#5B5e{AZDF5WhzKAS+Z9BR#B=6$iZd5?09B8Zt7eaAH
zV2UK-^SW*|BBo1%q66#O<kevq<s)`|y$!QagV^<d5qI9<bsfjqpY&dx-Yv_v<RX(|
zcV`zRl9CCk*>br6E&+%DE`$q^f@Ila$#!DLHBoaC+Z20mqS#9ULUDTUz4zXGzTZ2u
zdjKZi|1m8*XJ_}Gvt@R6`Z};cFQKzcVI<6Ekh59Wme#k0c2BM;k*ow|C^gX^O$KKx
z^!So3E3HLC6{B&vI^a4rd6KU0kc!q`U4C3V-ZeR1VUi9aRG0nrn&u-X<CLYwft4;d
z@;p0VNgYDRVXrN}$s8Bb{ygWvVp56?mt8Njf>ejEOO)q^1umm=O4^hp30u9%UOzjU
zA~J|Q)Ew8EFe!iz(&{HxzM-yNTtSz9yj=&Xxr$6+<qjfyV`9uk>`4QuT?Zmf-J%7{
zY<be4-l0y{d5(!rPc;rUcKwQGjQHp?dQixLrH?;bQ?XJJCL%eHdXDA9XIt6KggiF#
z+H*xqTybiW(#$!?ORh~NE4EcledJ;Y)6JZ#U7Y&przAtIpIo$8=a5#fUGF?5zUw&C
z9uNr&;c!iTaX3{a_K5+}$*%;BwLXkBg*b932D5$)^1ilwSkzU%0*I4dVZjmcHE~S0
zH0{C@4a#)bf#q(Ae1jtJvd<7PVCyPd98N^!V;y%uTA=l-@3C7OBovR8DYL3_4Pn5k
zk6Q=dWw+JiU==yWC?rF1J05kmzLsv~zCx#zB;?-7flikU^R}UMIJq$p4+$7pna0#I
zoTkG;N^N85lCI{GBvlf{DnwPT%T9eFAsi<L8(a&!c6<4Pgc12UoSWY7z~VG~2DIK~
zchp|pPA*%1Htaw`Ny%QcH|4hG3U~2_PBA+d#x-T#5I1`QadlH2>TRT`bxPS4E`j1+
z6EP>&Q@WA@x>o+61A8>~&HuGCgl#U9o0l5|Q31uZ_JQb5eIS2qCGY_`$?mK)uV)fB
z7^hfF`2;_=$k2CJPGq-(I-KYDePUontK8k-nB<zWp^;(ase`MJQy=tV>3qAVG!^G5
zy@<T<g^o%PMtqM`pWWl|mwEmi$L+o4H!{d-zh?e>dCExMJ>b-9o!1QE+V<xB{mhOw
zSBnzsiI&oV#pf3<wYSvOognR`Q;Lu2?2*C+rMcbLj8E#4@I~_qgJ96<jBR{tX(!cu
zO+5mv#xh7D%m3R-vhcQMdO~ht0~ghL3aL(gXx%iW8HOw8RuP{xSjBz<qEzU4q0ifM
zM2nX#a~{ijC{v44L?5zuG!rI4HTjpEYF(Fc!E3$FQbm7MXJb@;K|ovidK5s;V{DYv
zCZw)V=ycwC%ZCy4blC`crIQY<P-ao{L<DKgm5~sQqV<(}$@B6l=18VymP@9-5V;K8
z7ZWN0YoGPkH*;FR8B4Q7jV`?l0`1f%KQ+?fp5`fQS87mF=nQ)U;pW(OKNo%5mK^3w
zZ@kaT<I<a#+Cbe515<EFPBDg*H|GjTL(U88%M#gOP6M6|3dSB|)w>Q%glTh&L_Om+
zR0kF8xLAKN&N`i`DQhRU<UE(}FR)aUh|js`z=F`BR-4{$Tg&@&@n$%a6r?%g)RgI;
z8+n@2uGB=s<(Jw*GWRJ5NkkPnHOckdG2D8nWh0ekZQ*K(vL_7t0K;UHyJhZ8!hsO<
zzj14^(cF?NCbw*Ho^`@(p0rNgeUR3Umy%p~An;?g2Wyl{%qf;3AD`ZB50rt;qAD&5
zX?+G0&O7xnk&;KmD;!5I8*e7}1QV{4<$?ptVGfP<O^#c)G?CF<91{$S1B*E<qeL|-
zL~GJ~+C-T{=)P)@HI!HbwYyW3m;zfU<d+;3y0biWZL;ZPB;FEQJFs#{X08F!VZ~ND
zZ`)@3F~_u`@Qx7iyw*$UC|)9OX=v^_v+WI*uuZ6!gLoLQoEou?SKl{%xor<N%?_d4
z&Sqh5cWM%uP(a9jqN`zbi?AGmb-ngfGrt@H%(?P40jo~LXq@nhQZ#|YG3pigpOoXF
zM2`MR97}o5vu1T`e0T`e)x*uW#WbfKOPp3wqB`|jLc;XpFn(w41sg>M^$XJny~5Z!
z;Rv8>Q6DL<O^uiqS0NqJ=hHvV5b6Vi!df7j0tUUS>B?pJ#9YvjWi=a+QR)+$#@hXO
z1irg8&%Bs<G0{E*1|4(qB&B}NTB{40(QUtZsoBr@qI(Pvz$%Bz)2YdBVU7feq2BxK
zJ@q>T11I|=Nliws918Wt0!sjauAA=KqxC1VDLYw_*$iydXV-!_i8YXSs_Z6@m0yOt
z1YP5b#dKqe{ddZ1eLZB9M0zjVB~_J2?7j8PGsifdA{RbyMj7VR%)zfz*9uqU(}-o$
z-j}o6GchThuv08EaVXqeq)&T)67aQLt;}Pt1xGFy5`1UnWw)Zar%<TFF}4q`YS_)b
ztG*6hIDCeZf;+IXK*#9C9%zZwgxttvA4qBx2{+nKsdR<ImKMk~_Q87o^l=R-Nv<d-
zkF47qBJJaMH_OsTZd$^mx(+B0t)3RyKZoo?Ii6dF2AoF`&3iZ4hbz&%W1RR=O7U%p
z5ZvN|9_umtNPW%V<OrFLopK>ziqc5$4++E}%uFA!kJdM?=JIi;lrwkxu&yTmSX~xS
zT&_cV!8;w%LI~Isr(Vn!r^2XxJfDf@=sM43@=_rvZ}>dXz9-k>4vah#PC@<O$dwo4
zWCV*#7rr-pfP6U&<L(s8o8YYZsuJ8tE;?Y3*N$_ncAQdneEA5Me%KR9NjiO`PKzW8
z0Ai}N)07iksdAsJONVtpTGum>iwcwQsf5MY66~qkPR7F=hgX<_+jRjgN#yM5y39TJ
zpM-N1kg}B|yUmdV?GtsGZb?T=%9_b7!z6k8zVf~FED!UUS6JLr$PAEj&c44AShN^H
zPT7t+&7lBc?+7l>X{Ez+Y_OyIdw(FYYT3eh&Z8>0>7_l{XP>Meh1yA5NJec37PynF
zZg?`pene%@PB~_uD!;)B+Nex`#c$?XpTxNe-=N6$>GB&H^kmVLNB}HRf5XtQI=Q2s
z1LaC+2cWL@V9-QI^-P)7B;4t@m@WphHPQv8H>y~|M{Q@~U7rG$okH%}$Z)`G*f+bf
z4<$wTxKoZc;V;Jm6QFCm6Q)bbl+}>fvQY;XORR07Aem$lP?*`CTvB#-*LjY&Ro@;&
zFx#75;-*GSfD|$tIyt6D(WdQzd(*!1GhvSWR>_#;6&5%UC&N+O-*n<QI3Q74S?ZOs
zg_>ekZDX_Po3aB*OSataJXM0zPvQqVSj*k^cFhuHq2$btyYI0>r6ZaCFIYyeur#*?
zn%d#gPAxdL;P!WWg=wM0u<U|{9?5pNX?93_B(ua;2UhrnwDmXI9|qniAnj=V<?s+2
zJ(*Ep8p+j9Kq5YYlJ!`5ldSLBb~;Ft=MbeV9`H`9h9qdx&fscyyxFj_i)h-5r@d1X
zy-S^XN*igK(fME8eeFbfCrltrO0t5$M4E=%_`pG$PS$PL-x=+5N}6Quby#4}Hd{w-
z{#b(OLd{x+`Os^<*e^|PhJflWli2BYD&HKpn~giA*ofI-kQ}O=Y_j&vMSHHPZ**4R
zjAY+1Yj{)7B;cUSfHX)6SuDr&VLM%?Dp<mBj)<D`PE8_v=u;%fp0qQibyc07NR2F`
zq9zZ=&g4WMj%Um7(^HGD!7nv1w6pT8Jj*hltKW&_?O{iCVAl&3z<%Z>M#?fuz>K7M
z%z(_F(tYl;^Ud!SO|*aJ+38&ltnf>uh~}EYcE4R{w$LOx1QMhG$c@7dFaumkaIw5c
z-&De3(#txqa&E~ql{vM9eCO;^uK11ksyWZOM$EBvGLf$kX;8s8IQrV<+C#{SL>c9&
z1D1%$Z}NIosRrZ^s>~TuohX~@mGVo{P79GqxP8=tCBfg!{<a^i!_uX7L>4%JLdoP~
zo_;VoYA@IiH3JMIwG?CyELv-dH9$EL3O&x-bZIS$c>1cYIeylGWzYf#a+@TG;oN_<
zj2<fs{gy^b0(=BLanA4b+4^#|Fe|6g$#DQzHJhnxMGi|cT}s#n6w9AU7Sh*6(^?c>
z^ib{q(oCMT=j+cOY>(<9t}PP|C<av-SSh4}fwdp5D-o)g2#R%JWx|qIsG5@l_L<6Y
zF5;Y}km$0qHB5k<{YaKKq0i)Ax^br*+eI&x{axM-KbkZHUUyt6xhYC%HQFAuAFIoz
zLF?+(2GA#UCJMSAuj5)d$Z<KOgqnboJK7*eT>FWN2zR%8d!15zA8LoqUo6%E`^maA
zh%7ZWZBE#PQ(uOjX}H2122*m-ekxab`Kr5!spERQ4Ok@{VBop(#%lNJS}f=rh@-HD
z1NPIE2J~?1g_AE3jrh)0tL<mXw^9q18l+TTUR#s29YO?g`q?_CNT=F3XQ%f%upqU|
z7R<Mw%NbyCPZ9A%DM=6t&i+Vw`SazQ(sar1tOOsy@=~Ksgj-ODxTard=1yJ{dttlw
zIItM3HYNBUUZC0~GAJIvz2g^iXd9Prb{^Fw?mF0Rzm#i4N+Ml*;1pw3GICNj^>g;i
zb?r&`41X}}0MclCXU|IemGa$?$ieYUTYE`U&4F>*sSi>AvFX!W?N?_DBTk_#k9NNL
zTEjuo<L=j*5h?54tQV35(^JeW_1RY_gY>W(<*(PiZW|h(m;{Fb<ThAC2w7jpexuN_
z+-v`9qWKK!1*etp;HK-|ZogU94A){98%ps_JZpmay?|c-x5}zXS`|Ub$Wn-|Q<EFu
z4%B#E`|a{$odb$FK=^fFT8sN$@$h~2JEdMj(Mb&7@4#}dR}=XCyUhz?e&Lx<J1~6%
zw20W}->ch@(4|hFu8wj*h1{cDf|My_DrY+gxcz>!f})&3$myKYA&EP=;<a8LggU0h
zourXEY#{ariN~EW@+&J#>68P>N#BOQIzGC8STUU#4tQ8ZX&^C=OgF?%4k}mdj}pc$
zE@Q)?Spn66X(8^p=N|jxI+-{s?iqKAs#$dNGW(NUBE?#R!lg7cikWi6{xrK%GAiOC
zoAXI|aKiqqVql|R635o0ym#tzwdLWaIfG;W^V*FS2zf217)>;x$}}fp^cNLnrg0Z>
zN(E&ku7pj*{<1;sS?zb#w(2+3YjE@MIQpyXa8G+;a?B|tUp>ahxun$oy3X|ab{E~1
zQ=W5_zz>I^nw&!RH`yt8Uy{%`rRY=@V>r3&Z!4<xbS9mG2#z*14C>#N*D}h8<I>eP
z8u3M^rc9B@P9VSieSOhd4gm_$K-54#k~_jb)UPZ0oz>?A@Cr-)sa&tl+dtOPp`Vqr
zD&_1ohjv=(PTM~<*sV}EY0czkrud<t2u^+C;@Bj28}-{i*C~V_7(ff+0Fn%a`cBNV
zBGnp)Z2Om7iw&I#=TVzgDgyi0Yi2rwgyhh5#;>h`JCTbl*1}WvZ*^k&@e8F8->pj!
zflM&<iW^l}`oGr&yG05WN~LvPm~tfeVjY7)$?QK$chnJW-1MM8-{90F&Tt`7hywpv
zn(ig^N8Xw85e*%aQ<MCX?wu9Wsk<uwt69MPaVVdL&jgn5Tp979{r2CbQFNUFkbK*L
z1uiBmPZVV=I&1cL)>Lo()9fytqvxHPf|qhRBf}fne?IH+L`5#~!9Sdm)?jAGy)FCf
z(mK5hg<D4*SbjB>5fU0iO0ntM=ad?86i8#Ni>TcpWJPLXa=}rk(LR2e-f7&gi5~K@
zeeUDsq~^{TQI)O(Nvzt`8tJyD?DMjZTsk7p2ugXfvnM0jt$lv>dIcxFmuFlj@3Aka
zE$<lTQs*{r1`>Tq2cbO26=5v_h`c8Dh0Q4B{iIMyDgWgYLQQ^^H5uj#vaWqmc~=N2
zv0;xfr%m!Ur(TB(-zKxs7e5{}=aHMPcOH{vzh%|(mG&jgEGrsUyXvR|i&;(`i#QPF
z?0E(%@};#G2r_+bj8P6G)>N)aiXtTp&b}<auz`e#@S>D*%{X<cZ-1<heR&x{wM)jf
z<oJ}Ih7M}#i=tt&cXGV9YhO`23sa!WR0tq&V0C$A28bOFe_fSF_LT*}b`Su2z$q1>
zftZ}3J}g1<W^H<SNt{_YWCXEuYBX2lS08)+v6g*x^ZB$YD`P>0iD1g&D4*OHB>tez
zNnHZv=%Nmgnkxrt5`vX3t@@fe1bOlEec40iSVFwk&()F7iK?)#eY}cq$M9PXMI?V`
zos{Vtju^DcUZ78wR$1wr=ZZdGS7!lnNP7n{5AGB5lh>M563C3L`!nz+vahc_k3{H^
z0G3}$ACjqPsV_VJNVvMlp!p4tr#w~l9g1k0lV3dg6b^#ZyPUQXojLgtL<WNg`o_}D
z<nfZ55~sJ{sfmwC9xo}9?VAb={R8F!9a9&)!a}hgdd;2o&CPFR=FuCDJ20Kooo!S0
zEloA4o?}*?abSf4YUkvgKr)XXux~B1!p?*SXJ)9hEQV3jsn^b&eCQ$a*?e0L5d|O)
z?aG!v<$#j@teEWYA8gsTH-iw4Y37ePu;f0cp!*Z{9ZeI3SV<>y&Vl7d*NxBi!qTZ!
zx9Q$hO>&ce%xk@NRg+a%T5*<AhMMcL)7ZYp+D@U2{WuPam8He0(O#63wq)te&SS0z
zEWo3D>)DyZc75&kTI8aZQ%u#!LrCzgEv&J<I0sJ}0wV2T5-Dr4?&rYj+Dqz?He!T#
zN`cXV-+Vw9A%!lG9w9i}OG_{Ja&{;l{7NKSQ%)Xg;+NOhdpW&ywH&u<+YNOE#&kJi
zzf&$>BCdo9k>6`Z@Z>(YvCc=(IPEIAWXs8unxxxG#;zm^_OhIz{@wxSu^c7J+zpo7
z%gcwv>L?nQ8xh#5$_<bjH0opGr5sTdJz=;(`j6Nv$}goNDSj5^m!Y|L>Jwr5hC8jT
z_R8$VOW%5vm&ZFu0d-TJWLL?S<~%3$?H!6lVfLy*Np)ZybV_Bn$<aGBR5tHg7y{;b
z2bSNx71cI=sP(O+&+p>Q-~eK3L*K|)q+=5a24ggtc5@+aNK}yrBJRAJbCl#X(<;BR
zv9ql`Yzqn&F1gux%w0erL$58Y3k4}wA{$w?rDxpf)Z|)lg*uveCSy@)o1{HOhY;JH
znlg52;=+f=7B_D`5ZHc#!!AT0r=}pAe1Ek!HD7V<m~@}EBr*KDNA6=WD8#E_9L22&
zhiqw`f<jp{v#8W_qR2JV2d#21^>nzwid2LXwydJ@Mov2{UzBr}w(DwYdv#6?EhR~o
z5_^Y05PMDi=_=mU8F0!uJ*ZS+LD2ZTwzT-*%m?_l!zq`ZXGTtXX0NO0jFnS9I?@Gb
zS&M6-Q(x$a&OGsMB}s<8zPv5>cA&chi`Lr7^6Syta)#bezXQRN8e4h_eW&=SJN4Q7
zJ0{1wx@*+lSUJt{3FkQ$D9`8Ay{9Cs1>9xJbAbpi<o*N-(TO=zA6R?dRy0^4^+6gc
zfaNCQhQ%==1Z-vbIYbt&Dm)?RDH5ntqoxx$mUSn%xz<*drYT%SFjMUH!%j^hM+!O>
zt-zcdEU}ouM{RZe>Hsz$I2;F%CrX6EkK3AxD}9s2iOd1jaY_2%ZE>GA?9AFyqiHAJ
z1p8&t4tuSslXZ$kjAy$gC;ZM)+(jd6odeN6gz3geZ|j=5ItrVi9dy!xl~G+ACL(Wf
zh0WfYZ&<f+{kn~-ol>Bi7$vbYIKA0ziQlB*7NvEHWt(ctrM$7eng8$yRtyz5#E$Y>
zlS+N5w}}CkLeVxf?}>zv2>#DHu(GhY;SCX6u(9?BmPtnOU9eoh)M&W9g>!uzh1~6R
zNh9_&Rrsk2qUL%{4`zt)#TWlSC#w`w&}7-PJL+nZ)r_zPrz}Qg<(jqjrZR_`75QHA
za{h)Gp=YN)5n&L;F$e0VLRzX3P7Fq&ba2x0S(T-PRuaqZAgh`6JRXh%_dnn~1;)e{
z1TS{cKK+<Cm$jCjd!JJ(gh452LeXS**4YY9Ug-<YdW9))P;#;?@^>{;uRMKmhqzZ5
zUR9iQ;pA&~H+?y+U~9OWdL17%nw)j(@3ni%V6(#_>-d`Jyl0qlateE%(<+C>QKZ)s
zZOey6hwm+uD|A$B)Bn8eh5AAx9~|nBYogy=x-ULM-ItFTu*96z_)~XArtB@{gQ;W{
z`>fzTU9C(_;&`~}OF>O)*2HlVKcdI)bBbkh&&QE@oPNHw3OQk*m-Uy10P}e0BDY2U
zWqVs~$Ge8OmnlM{0BXM0w0iCBWf@UvB1cmy1VT=@V!5^L$WmeNC^gbE=-S@dtiROc
zSe3JMl9;(x?XJ8Vz!q7bDlEekBzPRO-lm<DL6CBkNWh{4njztb+*Lz?Z++S4<!d)P
zk0LY7tGv4V>j-Op8C`u1P_8f6zslN4*cZ*?e@<ZzI?HWQ61ajVx8yr#cC77?0gF;O
zEp#_+9_QiY2rj|C*g$!Yt_+s`KD^W52#-58@k=pF;1qgp{f(=Rk)Lq&%t6m6!z-~o
zgGx{ZY;LbYjiq3VQ#MqlDA~hw_OCEuwX`6xk9L1;7h;hx@sI;cridG(mfF_Z0tXf8
zMwAYz!gLVG`)>F285=I`BH<)fxX0iUfh9p9<638=Z6i6Q+^ddqfI>k|xiEy%!$!-1
zM9^d+KN@F-{0gYa?Smq#J-&r(6P-k#xXCwiAOH|@Ziay!OPaJ5nZbEZzGlP5P0M@i
zfo9GWe<3ZAXuP;_YEt38VDS<gufG~ac{t@1bF<Ct;#VOVShv0jbXB6}b4Y*rCTh|Z
zJ8m7sy8?j(o^hi508$cMV2UKirgu3tIl3VPfnFJbC$lqeW~K(5f>xv}*NuMLR#!xh
zpTd#peO_VVCdG<if3odmg79@p{WyR?u~=q?ekAm^I1|!sNpTO>4Y{o~G6oaq0Ll+e
zhZ$bJ)H>)Dhe;k3o63I4D}r89nnMGs=a)1A50#p9A&?MKwSLsZ7m}0vU$oaA&Mrk{
zcDU06&}lQ+*|NpnStddOlW9ge;=ppXm&cj<NF9>AXSj&a0SHMJFI;5rs`D5k6cXah
zB?p$ckB)zfy}M}<94bCY=e&U-B<u=?<$`%>mzf7NnJ2x%I6ct2AfO;~z9&DvZ1H^O
zF$Rt^OUUF((HwRmruG7B@n{{#?IhT1JH_1Tk*O`@N)`~!h&@(b6oN{_=*_C>YrU*L
z$~t?BvnE(7J_I7zd-Fp>BfYKO?QN&1O+aWjsTSW?$DjHh*oI>cEN5u_5cZw-mliT?
zDeI^1*#?$F;|vWZfuV^ZmcuEowDz54Eb#3xlftCpXYbU+E>mA6I*D!ju13C57MBPa
z$z8_yLt=ntU2t0E0fxN*k8#rQ*azyKQhtj)=<*y$ne&r!8V)H>;y9{F@hA4dT%MIg
zlkuoypwnaDU1fpoL(X%-@bF8Ze-h)!BxfHgHAUl{`ZdOX&}$7C4yDmwY1R<n?ZXu}
zMbRf=)c^=L$EQZ^Bh9KdsRQ|KMD}yuo^k36KEy?9h>Ei8X?8za=1@9HITVsx%L@!O
z5{G*p`S!7FH?yN$r6}h|U@<}5jsQtkY=y3eknIoGHjF_ZuRZ40y;lFkpaY0*6g6-#
zN~!etp7JU&QcCo2im4KhBK;F`u~RRKhr=D&OS$i1%D%TQ;m|1dMCIIfK;?)Cl`Ut=
zA$z=;>2zA3y-|glXr1$vyW@%ahHzQbNxa)DOkES*CZV-GSx+kNl2T%B%36v??t7s<
zRYp>i%F8q_(Qj`|_!*}@QLleUbE_+Hp01n)sk6Hs`Pac?0TktWsM8s-Pc&T>n<?k+
zXB}8>tPMllX7zogMY2r_Ks@fi)Pk>b&xn10Gqkg?8j>jYJ236uVEyd}iqs$!muf-B
zjpk-nX(a19_&dGdX`AO8I%;{KNmKpF+MyMkxK1f#5b?Ll{1F+*V9kpl2&QGO*uBWa
zpQ>-et5dozoM;_bpo*06bsxtLwoliAVDr&=#tGxVwErF$vmLb;>)?o;Qe9)tKH>Ng
zd#1cuh?iOq@#J_=IyJd8GTkWMQQLOr3~k0$WxG=dd{3TBLbR;pv)Sf$RrF?OJ;O|8
zN(A<W*P7hV8Q7R|!0tLuxGdd0aUqO~c<ebf$qkJUakE<2_SD^UTdO-bGBoZ0k^vc=
z?6<wmtSI%LJpV2@utHg-hE;zrGw+R6$M%)qusUH1JdOzjSh#QH%I9@B>UGefIGG3)
zvHeZ2rAU)pgPTHRRH!v@O7QggmQXj^f$}vuEprm;uLFx!MoR>i@R!quznoJCo4F=h
zmiH)!!5lk>Q=faCS;i+GhpS6=sKW57k-OO`>^{Jfw-OENPHf>)An8UA*Yk2SD(J3L
z%#~lx!QbK1@~$1Jg#5a9IL|rL8`s`$N6S}qoedmZ_>Te8T8VW+OEP80N}KflkO|<h
z0~4zeDK79`KVCaa*R`ifoZ!InJF7ZvJ8=ym375b;cvxLY0Mr*om2j>pJ6RXIY+PLd
zpR7}4!*}XMPw==JwP*8rsE>`xbJAq#MOZ}nuW6$kw^OAiX>x3G>LH^}O-?@FQU(}q
zwzTK!w1C1g-(Q6B2Uau$<k7%#czVto;n$JM<6?2_Bh(~Et7`#!b@k+#=6x5j8;D$-
zaA3JiE8&#$hKaNF6KE<m9Lg_am~-h19h}V&mbhG1-h}<zxw^0VrY1UF?lAxmquURo
z7fM0ciu2h>S45;84uBfX>=@;8f$^bjXg2LaPV4&HSKj74=g@m#Cw6u0VtHvQP-OSQ
z3A0Rx&KaSAWQ$k;yHwW}*DC}!im^soiS!4OQB4N0%k>o@>8oCh5ND24lj3rBE2d@S
zfGcxeBH!Q0vjgRR)~SgBV|!xH%Fpkz{a_v6&}b2vrW{yw1Esy}v!IG}8oQ=<+Yi-l
zg#U}gZe{P$G1b(By;3}fcsSjRHeD(8t@x)nrQ{`oKB<N1b@=VE3<vCLp0JDHg?SVv
z54VQ;ZKfd;QoShJLB#Kz2(wHuUhDUBJA{<mlB8<7G!!WId|j3_l!Ch#99W#21eo#R
zA1+@I;)JUmTCGel7b)?b`kd#L<Pz_2HPvUz52QK1<~}4od}?~XQy&!e|Dz_Glb-rV
za=bUZ<*m+ROnZt$3SJB2`1Bt8(WduGYa2Qrudwt^)|D$K?Z-+xeG<byBYDPw<>A0o
zqu}PRvY1J>4hK5>@$ywRPt2)yb&_~$0$oJh74|HB>Q7X3@1Mjc>p`bnjvl4=lCY7q
zMMQizX3d|h%tnkx#7%C1dC5$qh7E>62jvx~Z4$cS$pPZ~f2s~9xLEiUl%v36ieP`j
zM-{DuItCH@>C%xop0(6SLx&C*ew+IE<s5rK*4xk2&yqC&_kW7zU=&vxZaT~qiU68(
z#PgS}Yd@QvBcwYNa&Zqgu<ZW|u9w+rKUezCWJ3Ou+W{tOab08}*?oikeCbtg=)jvo
z0;$O(V&VLS_6rsFS|qZUObwurE;;weCuFDnVm$%nkqZNjccXlE0Sn9V#M6aozm%^f
zTV{K^Qz)~UHkONQ9jSkh*e{p&>tMMIJ|oL`10$?M23}c9e<feuH`(gvZZ=BMD;*Td
ztSRNeuV&XF_v8T_BJvhM9FKT9pc%7YtJHznA9G4&m7tqNYEAy>*DIpIL5Q*h?~3%S
zc%HLpX{gWbiFR9h2m6g2=?J;U`kjI%fs0oWb?i5ju*_dJf4=ja^So%mJo~M>f;iP;
zMy6OdKyoGMEPgxN!#PflIE82+QYuO<kN4lnm-jE9SnfRM;lPDMiYNZv(musOov~MZ
z7N+3(&_dhqH4`ZTPU^yo4ouW=X2*tm?Dy+%P)AHSrJQ!uZemn*A;TZkVJ)1$z$run
z#FG+`bKL&0?ibh+S<zC5gY5zgI0h$6ea_39Hm|ooN+Pvv(E{f&*o%pYN&DkE=OeA*
z?JX2T0OCXN78vDt8PRBeQeU)T{R*d`b?8(6jA8rJteV1Se!wZjZ<GHJXL7{;tj^!<
zxSu2VQO=Qu*rr3~wEcPZZHmMgJmxBs*RIe0qTmuc4=NCyiOD(a)FiHKYLj~dB>c-V
z8C5wY=S~!N5_#f`rCu+-N(d@6?XOBNJ_a-VF^=+&BRm3=VDj>t?5~?~9K>ZN4hOF=
zZE&UFV!PSjH0`A636p^XlXwY(Vt-q@lzxP1^Q+2!?3GS^PBq#Rx%cUk=fA7+<;u?J
z7TjuKVcHHbad2asXyz&V`&#Y}V)pKL%7xI}ywLukyj8x<^01cv6IhUVrhC@V6F#v*
z%q#YfWu8SUrNdA%g8U^l%?T(!Xr1ypUO9$9Ra_B}+)vs+m2Z;5DS~nM>%fvtgq_4v
zhQHN6XRDbV1ZcKooM3+fD@Pl7k(pLGG3foedrUg`e@R>=GuaX6If?W3q4sdU{cDbz
z>kh~*LLpHP4n)3O%A)yiwO3&g7KJ_Iz{*S{zoOjuDY1kUWdB~@!?-wiLSm>edHICg
zH+{<fBR|ePF5DtYp)4fn^MwaY>3<fOe1l7lc?YH`nWL)tuckXJJ{?iqiiTgas!$)q
zG$&pvdWA8IZ2xz@AH@$R6_j}dJOCE0Bk&^x5JlSJR&Mr04jZAB^Do~f0hZJ`+7C{B
zP9OA7yo2<VpH+U7$FW#M#nj~GZcb7T>Vtg`e+_CBDIUj6S;jdW3VRT?**^P;f~Nx*
z2`DrtmsYH?$lB*LHDCb4ZQLiEn%Kl5*5~YVv!ocx_Ex7<rs^E$j)o}Z6j}=Vyi$|?
z19d8iMsjM(`%Vy-|M|^;<)@mqlsK2en)8fPA76+|Nf@L!lT?X)!4t9Rm)*R`c@%La
ziT6(X!Y2yCvPRvuQ!Ho_{(jnj_C?toi`}U6TyR)HKzpRGu`ez#eMhlTXW)Z?RfbLM
zVE8X(kg*uRefA}#UcZq%uHBS6BB`NIJM}RRVZ68nMMtIq_e)C+DLmNW668|Ld&p~z
zb|?{Dy2&Rggnd~qD!$xk4Jjn08rFf(wl6O)q!Z6SMkkuYTwZHrVv+~FI+6V8S2W#I
zRtgPI37(YHoSGy6+uC~hS5|ad)w97V#N_3hQsVlm0wb*<e7G{k9q|ea6*Q9rhiG42
zCJbj*sR^WDqA!i&)oXnciuA@7AAAr5lTz#jwO`%#gRpB9qbppv4GmE!-81$zWl#!y
zhr&*idynZ*dJk$6E2a;{6zMYhuWhDH_g>5;A=D&Q1;NhYuXWPCE-`2wp)k&~4j&wr
z?dzXN!d>}Z=AI9-AsCtFH5^TyR{A&r1f0##5c!6GL%wABsvhStFfC-@z=sS9lmMWx
zxo^z<)8EkrgI-EAJ66c$>YExYy%Zxxw_Wf?oXSC`J{1JFU=Gha{=T_x=#{!4fl}g9
z`$3YbbnRQ}a&KC>(kV35L#__b$hQ?)_JVzDna>m>ad^ap3@lpbWE2X(0`9YKE4>uk
zAU;&WPeoR=s!n~bPTW?S7J+ZCzkzvC@{9{wH?&^bODE9aQ=jWkR&lL0kxcd-<u_H!
zrt6u}ss(~`XRcF!y~u{nQ|;FD1(wJ%$F2^Xlx_CHIuUD@zsV^qjvt2qW>w|QO)MgY
zG2J6$(Rj(h=O5*@TxD$6<yvo_8g(9XJ;u<2obM%bw!NryI^U|3A}_nt{m^17s87CQ
z`Ksl1eVOB+Bv4E-vf&-;wPqh<3q7NRGFI%Iy||evZiN9muwXf`pvi>uLH}OTtfriM
zgI#oBd20Z9+{U=zlJpXA=3RSfb|-?uoI)Xz2r-OVZp|3SQGv!)Hg-d98`9-(b)NM}
zr7ysH?#4P0h$Sw{#qAlE$Ek^h%_v|3{+@r#URL*h4>`Gponr7y>Br4fa~AzCzchP!
z?Od!gYYy80DPmcKn}?nH+?K8O_#`oWoo>rsQFs32=rFQryWfE%4~(!qz}>X=$|u6h
zy~S{wTP_@6%%^g@p0}II5UMpxSF8ccjW#ge=2|d&Rnt59t*0Jn&KR8f+!JCg@@%wu
zb<!ZRQs^iPTPCBbHH?gNNJ4=$`JB!+zYv<W%bmwUXcSnZZ6rT``7*_UV>g#+1rwpV
zm4az|XiMm@)W)Ng&fYCZRJNdWQlC<AAt_oMSg!MU2N7~%`6l=QPlVG1v_FpX{IXNu
z9Fs!ZU>L<3+oC#H4CXq!cJlk&5w#}sgUew;K65OoEv~;&A_GujH{(bh4K=xp>xU)=
zaZ<D;&9>@odnw?gGd^HTYf~=imZH!B#8PC=EsgSmEo=IHO(+0IO|8jM>e^aJ9)#<J
zy}EhNS*V51<Wmkzhe>z1W3Q>-g>h%Ibsd0E25*&a+;I8Y24!>~Ql9B~OLat&JM}ub
zd!Y!Z{p@w6i%AWW!IXC8pw}8LbZS0zM<l6A?Db6-BSHaA=#m4gXaf=mQ<c*_1K`Op
zZEt9hKv_d#D|>;#HFCQm{}IJ$Z>;0Hwbl24B1{1!^(eYpc|c}o)467O89*K=bCsyn
z2nSmdcvh5Fm7L2hB@81@O<6Z>R1N)-t!z4*26gVmbzpgL1?kM(Xsj~fcB?8buj0mE
zr<9a<?dCgdbuBDhw9qN!L5UJsN*DbVMjo&=<$Ln%6hDBIivvscqC>Jv)b=B`w$e8Y
z?=W~2OULD^tr<UKw^W$iIa%S$)4Ls5uFM?Wgghm5N~8C&b%|UrzUc<%G4X$7sy*C6
zQG9Ft<mzDrU@yg22LpH5Zfgc4iJ_)k)(t4NhF(P^VG~tj>nnvrYayR!%37qE9WQ;m
zZKzWm3c}pxI&=3sHEMOkjW@o+HdeaMrR_3=5S12_nfa)~&^!H%(<){M&M%};T-)sS
z29-2p8ohEw?Su%Vp@jOJQSeqLKl)I+qmGdomGX%D+L>3~WkQWkY19JcUc9cosm>pF
zsSV+5>;MuOHjs*CQ!Ol5y3{F@)sx03SE8q!54w;2fKw{ylJc15FqIvf%Q(vY5OJ~<
zq9_j-r#{|||65r3PEQ}OJIim9;6$zjAz$!I*e0ENjb>GQ3%(t337p-P{p}eVfd!;c
z=JP@zuY4=+PV88>@vi0T*R67jrNd&GQWlkV`iR|=%hw(m8+4wdRiZ|ATP)uxySGkB
zI5A3baN2=s0!jHq4um(C2D$kn!@5^k@wSl!Wp61MA}kkfNV};INtqrB`(T-9l^YEG
z4w*7wx9@wRr0ZFUuL!g?ZB{_bN`=4;H9T)`&2EwTg#>q{FiGZ`342@98o%`*zX!a+
z<gqCjacB1SI+hf4bnA@+%k78TA$z>v-cdITA%x^_cFGw$hXVxk9>opzvsP_1(jl|S
zs8iNnnK$p|`PSP^c6y8+t)2u<O`+F1ni3!8Sp2N7?iF~2R87d=e^wBy^@>o$iv|8U
zQp^N9yA$8Y{yGm_%%O;3G=^SboC{)v<=1QNX0595sojv?;*B8cw8}N$mr7P9QEy8b
zo1B%Uwco}1hScrU6#iLm=r|?y+d!F=C5sk0&$%gwC&z5CB5er2ahD3;2OXW76mf~<
zcwo|o>aP|oyxA!P5=)TXtz7*aO7U#>m!66NkbX{@E?~ilrh~RHQMWkG&U+kPwzYoX
z&D<YOA(n^6ESVG_*SJXAaOpq>y-5zPH%!zTeMU<E_vkOryKSWD8dq%r!@0zP<(>_T
z5)+Ov6bz<~mYPHgsZ8ZivXgisYOV*tlHJr~K}1l&C>km~f*G`K(l|PO**Zl)_~6Wk
zW_HZ}bKr$m9N~9j2XF+}wzRR*%UKa+zXQvSfiZ?ix(R!r8NO&^y7!4sAx~(7DaygC
zI~UH_c*O^<=_YPQCLciMjcv@(7!$hncj*;}zU7bui%!bpPFmPQ(z0r}HqmrnnN+IO
zAbEkBSU9b}?3lqAhkY$oDw7qfR`+9;bBYDA#f64@k8R5hjW+zk)sMp~QZ94VWI0l2
z@X_90w|*BK-PUr-Wi_>RQk^7@=eRxCZ1+qEaY`{>!U_hYX{+@qSrSYhocgIyA(OGx
zrjki`<P{G)kA-r`eONx!NZagKeUOE>hjOcJ-n4w9^PFw@lzh2QLI0A;7Ynz~9xlU|
zmkm~NJoZjFFfqI9Zr^0@%*WBzp!2BRI7Bijd7?a$A5mTv=ef*}Y(8=o5YZ=5fxRn#
zKS?TD9?SVfx;5|u|5|%@o&4YgWSrRNz~W^=omdcMnn}d;hVPi;TQe*2WGkmto+=aD
zBdTk_-ZP78y-E}50Lm(HB{@G+Le2cdquC)C;tm-Oi3`HS4@|@?^KCHrzOFr%gU92h
z?L5cJbgs+tRQ!h+-FutSr8@W^7!54ZeqwxT4BMx@uMR2@HJ65E0`um$Bs#71Gmvax
zG;*@OKPPtU(1`OG6I%~h5&!S3BaFOCibs+Qr>-}yHHA9GaU9#nzN@?^bn2|)(dp+L
zSd!qCgboha2lA`!Z5`)PG<=H8e{K6<GY7ilAvO`+M~Zi90{_zdpX6=t^N)tv=-Bj8
z`|f(6N_uMPmThPQK)suD1E%dml_!{!C%NlDwaSVo)n)M<Fdu#*Ojb9%!+9#`7*W{I
zzX$Rdr=)$Pu1MH>VzM1r>7h1b;+rN%qKL@hN6T;09uqva11rCo^-vSLByn7B1s|(D
zfd7#QufIBp4m&jxyGcztj)UaK>)8tlz+k`$iaP1}M^Vx!ie5;ms4Q9~&a7w+4EDGB
z?R%2yv=9mK>}fmYGGOgLDdR;76gm36rADMvmbxY<fK!v(5g{~a5OUPxiPk-vH{I#w
zxl0}vd!lxrGrAI8*j5J+2M{AeiKN-mc`D3Dl@_LvM(1Kr=1XUGtm$)}^Wfij=LUPK
zE`}6&IoS>@8>}Vv=IN$^^j#n+h8AESWd3Fkvj6KR`$X*}XN>Y`T*eir!nEbYZvwOL
z%e`}F&no9Jl}tVD@!p}9eSbY3`^YfV##L+B0cm-{!r#~4-co+CAIQzPfz008B@~mi
z%l^vrr_+Nh1YswzPc|E8baZHNs9YTmRK$Jh{6+SuI@!0ZxZNp;X+;ctqkXyzNz%1q
z;oxl?m{#WMqtnM<VLM6#u|<qeVQLOcFBl}fpFL9%W>lfbaOeP%AslTJJ1Zrg?W~gk
z*W3?Lrktrd-=yGFvH&QpvhAvGYW3sf!NdSi|5vYEWxMln8<u(=_3>8UP{;O^&dF%1
zEu@sI6_z&%1m|UY>#l)OP*Tcp2G{m=YO*UsL~M<A(Cm<P-uBh^BGl=cQ!f7XSy)D?
zv-50!d2e3rneSzb;O;|>Hh<6dNEde?OZ|ktAlXyW2=EV_u!H#}8S}V^W^6)Q4PZ%f
z*alT8u8k)aSL&TZ_1MFegXm=32Y|emH$qBgk4t{R$s$XS0@oVugErD<C)ZK*ek|%~
zJDgLzaM=>)u{mbs1i1@E-jOoY?1%`DJQJXrUTgNk$x9ESIKrf-M;n?K`Dr9CQ*%jK
zUV;VA-_t4&9+KtbfqAU?5!Lx3&`tuLJ{^`ZSO;3g92M=9C}509t;FmczouUErmIVz
zK{+3(U7Y$tN9C8MMBMV2A_(11l$yLI3V{}OG%R3FO-zgQd~p{C4A{wvDVPR{N`|-I
z;}xbAkZ~VhKYO;c6TdIL?Qz)GS=bSUZbLfoXf5qH4p_ZBf`G+XQyD!g7)nLB6Zg>P
za^=>o+2}lK>yC0%jN9qlP~)vJcA^qw1JF#9<Dm~9Vmnhvdxq}G?OI_G|NpDfXqyMq
zmc0MY&eo1XG8F7~35pFYSF#WNV%yGDqI2)$R;Q3JXb<BE#&$nn8zXObwm8L@T)6}4
zmS9X4cHtU^NvDpq3JU+u40@5`(`gk;iHloUQxs|$@`+un9oTgHI;Sw_<eYPXv|Xw-
zK^;<85+a3DQ&wDw)Z_!R&o0-O-i@Y)LcH`^qd&ZlT`7wuuf6n~jlRRVr~Zrj(w*i)
zBzi55;s?t|S$t01KsE=andR~z+==?_hw5ZaKZg<^wBQa*6lZ3~)lZ-{o~~55HPV|Z
zCI_UW=;}uy2p>g6y_y5Lb&`v@oI-RbsYWKVf<b!zQJcwT2!n)tP)O1Q7caY1*z;y@
z|Fh@oe6XPu$j<iRFLC5fjk-s6rUzk#f4FHUnJu}<oR?aYHl$Rk>sUH%pDDl5j8=Vj
z2<uLbM#c?WceREm+x8=M!-x%LdZc?~VDN_F58$<4Uo4rpsgYDAx^~=tv=DFH@mXU^
zfdr6Dwtbk%?Z@f@ptqE8cLLXMVCqGO%q9^W`QxP*BDjh`i8Qvfl=48l9`#z&<#C^6
z*M8!fDao(~(T)>NO*$h9J(_`#vj<uxSCMFCKbfy1Ds^ITYS<|ze$hk<Lj1@>*>K8!
zs`OnK%I!)`idR@3r1DEVZ$DkyWyUVvKqMTncx%aa7NgPpp0=MUHQ+?he>^C&t4>YK
zpRh$*0!5eoY-K@SeS`BX%vk;47{{6YTxrd%2JCZMPw#SIiTx9Uy}cv$^Q8&y7-B=Q
z0AKJ5Q_?7<lY$4fU$|yQGfF-jMx2`DQC5#)W+WJ>YrmL8bm{!X&SUN*u0R^LU#b(y
zF_=>*=M$1`I5n{u=16Z0S*{C7%YV7_leBa=q)0t-!hw}@E8FdZIHXD=kiveYQkQkB
zx=x{xpQAy3FlpPb*2AA|zKj13I}ou++gcO$YfWEMD=g3EM0skG4{ddo-OGMG7hq@`
zp@BRnC%$0eV*8Ccx#1HaygBC;mO51(rpN6!o55wadn959mJ<^0R#HlFB*xLKwZCk?
z)%@mE=t24G6-FKkd>Lfoykfsy$JjI3dp~DY*MVq^OSn+zck*#)V%&Mm^X6_+hT@~z
z%<Zh2G<jtSWV@XDLUON!5OXwm3MFy;y=J1%#WuMxeaL~84RodPQA_`t-!C;%V9LAX
z1WFEBK2JF{4XVrLB^%=()Y)A-IpGw_ib`V^%!~bDeI0Wxmjx;Rl0s0EtxHI)#Qi^N
z&d~=6DjXkj067WEhhZ-5k86E*7|k}^CV*I!)Gb1c;*exxmHz!nsn57PC96COuFfj2
zsgF*G@W~@-H`w#1^*3R{*I_HVq+m|*mAjmJ^{MC1l`B`1MB~rOE95_=G(>nkIIz^+
zpy!C}jP~dCplM-grBK!?F_{B^{+{RCf6>%rn5|OOCL>vE<~p^6sA_*%exs5UXbRm3
z>!jBj`8FtaLh=EJ3Wva7HOnGL9wDagcv7S%H8CT)HY;o|uquDu^h%cCR9}V@o!1(j
ztm0mAAj<yQ-_$R_6QWLttQW!Kd9C5A^psz+CPC}%Zxj1(xZy_UG09Y4i|D$)t1pAz
zE1`{Uy3A>*HL}`dsw|y}QdItZ#V-U6hKTz5qytNI+CntiR{Mv#Ot2oQu>m4RZpDx?
zeiMgg|5%&cIgIgCj=Mne9Rrah;h*Xb&@~Mdk`QeshdH@N|2YZKym^ZkInRNr#hFb?
zb9yRY649Lf_Ah0ODMRd|*Cl1BQzMZ|DJ$(?Ykv@pPz$#E9Z1<niuy_Y(7)AvqMOFH
zz}f(aMp`H7R4}*Ozc(v=hON1G<|L9Gu)xf~tXLai6w<T(M`_LG<umcufhF6!jwH7O
zSdjj+jvwtfOu19exmqvJa{I6PMYg=~Q634v@*QI>(eAGOcQ%^YkvZ0+Q+Hr<=1wRh
zBZe~g$bxgi?8!pp*KO``3c*pR8NDw)ROh5zX3{ktp~q)ES-Q4n-KL(+PBCflBKzzo
zW4cGVVaj>V)1!|&P(J6$xsxN#r@GqKnjoQUF$t=eFc6_Wx4vK_OGQCsA@aS1Nzgv;
z$$F^#PZPBE<TFmaw!v21_y@Z7`K5J<Z4PJf3{J_-K>hW!W>REB$$cuFsirAziDxB~
zu{~yA@MLfalX0UOJ?d*ap!_-bs4S*>Ns5(~Q_ZXA@cFSX%;{T;*0MD^<rIaYSQ*aQ
z7d=@vL9Ze{yG~h1&>?t3X*It1$wG3OKamVE4izSqt%9o{#P%hn-5wZ2i6>m_;J^Ys
zW4QECp;38keQ8;AY4RQ|L$6Gc?yi!O;Gox5$@QW0L}ZQ=pLfX#Df!<oOI%oi`y-E<
z|Eq^#oln`9m-*L;tE);5aWbTBB<G*{y31hV<t1~$zM>8nnOCVu&dlue3QL9*BQ%u%
zE9-Dsueh;;X#p0))%Vz;sGFsplo<6@&6gr=0D6!U4lFi_2BeGE&glG}^|ZCGZuW()
zdaB-oa`zIGE5{u2=sB(8n{zMdiP6~$axOM+$HYD5lyYmc{sS%hnu^Fp1EQm5t{#N?
zIV+d4m5!og_O<nwb5u7ez+#eQvTD6}Eooq+;9-@kJ>I^q{619&i5X%#OB{pO`eXrS
zS*CQEIc#5_Q#rFkrU5w%A{7D4!L>(vJHz%3xo_oBi`QzKYeIn}(T%Nwh=d{j8=IL6
z!3*r5yP-s)rjTg%Bs~2|*u`(k*__#-yS1HCQbAk@`X|XO`OURgp<(ARBI7Wyn8V~c
zwP<316^8mP<pXm2Q9{C<AP=w<1NuAL@h)t^Lw;*%CO!tqiob^)SS}0`iytvKri9(3
zEVFN`y@qOXf}<xq<-oEFaQ(u4#X=@_e0%*I>5+B!Gvp3X-3~~nl4hT^@2JCG(V3WV
zO3^|Igw*dVY?v3;4uv5}jF;j@#CTC78ohYQ0=up>A~+?CRi_L8vfrtRT_BOS`i|%X
z6GJLC{Y6g}&xfIrvDVOdxJNs{oGSsLQXFO1moMlP5f3JRLLIIroEpUxz4~E$aT0f~
zZpA}~LQG8_J2FJ)5qH2|Qfg`lRio^5YVy0cNnx%_0PLmBtnY@2PM!={vK3>U5Wa7C
z@-y>9m4pbgQ(i>u&J&yMK|hOdQutSekcFny`CY!JZ%Q{^*^Q-RXagbm`z4PvQeC)5
zO)QMOZshxi!P;Jy(?2r!kn^Zxkti*g@|QQutjJ8=X@A*)l~quZRi=bROx;sr%PZ=p
zPreZGTXHSEOl)%Mb)InV(K35w=_Q8L@aa_iR-lDb6JLP|phm7EI`z^}t~3cZ<y!_v
zTK&#*jOb(VKOckX`Z(TluPSd2n#w2@qM<`&pVyjH1!$`&=^!Rto7b$55^M!_IItua
zTiV1w+5G0U$2n8PnMS&!>B~-yFnDBS+-@#?pR1;+WDzdfYrPI8oh4EXX2`OV=3lY}
zbuCwM7alAU<y^7Q>MQl&QCnF1#l_fU#itk;nt{xbY>z>av)LA9za}3VavtNvQ(w;8
z;#_F5Uz63JLO}}A;Lr-%lG+VYI?MV&IXzTDJ5}}qh$uJ3*-RZv%bTR+kiekivcOVi
z>ggMV-I;jEmeqGJXG?O=D`gQ$<<-iZL9eb@R7iz3R$@`Cs8g@8D3b_2;IGMj`_9)o
zPq|(M2@`z#+Ga@=o{q|8dcRkgI0eNk@z-8g$IleYoP>$WD=cn+10BfcY_Bg<2<HON
zK`CZkG)~5AqD2x(?G2^-ax{=w3ilCY7LBU5h0oUSl&%!SOzULP6WbeeaaLhj8FUJ<
z7o;PS&+}RKqF&4q0e5i5me=W(;N02<Zv!OHoEH8gc!k-DLe6mi?K*`*>E#$x7E(N^
zt*nd3vdRfVF22B`bws8G{gI1286#~~?!h)Gv7P7WW2L6pGH9zSAjE85u+{a=p(Ymx
zyvKnFEp$aUQ30%%t*L0SZHRmDMxAm!JI6<5=3QF>plxq2NZ9~ZbBQJAj@^=-(jAh{
zb6|DyxZ#*&>vATrTH^jsAsfI^DH`q+8|K!gr*eOdq%K}zk}{Hm+i$nk-$Gk-KiUyo
zCn`*NyCtk12DR4LE)!;hPws>RD0`Lr@M4l;QDGD})P*EGN13zw9Y_*e<PLc)gr|vk
ztAb}6ODFcBg@V~bvOMj;a<{?I5X#B4-k#f|{{iQzEPbdaU<;Ks2qlx<(bUL5FRQS^
z2l3!^YP8iDB0d84raFHsI)hFjUYGPAE)(8ETE*kGDVr&3bkr$`<a1xdYU`=U%oilP
zKN@Hpxdjwd=G5oxP+wBY28tfD&E=OMEVMVd2PhSULB3u4o%-nK0g_8xwmY+gO&=QJ
zl#(p0<jT-dyQ{7p=aR$>{yMNAZ=5omymEGw!dnSmCGKI<f}HNOyBq8{7af<~3|PT^
z5s$i`GCTywW4kARGqVG`=ax44<FN(<i(wXdFV6EAhTYrDsoXK)KOqp}`;R*{1)b%&
z6}DS@bAf5W#7yBr<rS9akAg1`+gqA;7@zc3xE)~8?=*M7UMkuQ)wq1^4o~l~`^s;m
zsV-N}JFr9`nKqb{CgjlGT5CW;S$}s+$HbiNcWTlj6-9a6w<hh&+bVz5%RMI~^X95d
zK)KuTRNJNinYTarnanVRd7H=~w1Uz#IN<1arh(FYi)@OPAt<0~en+DF>)+u##f;^e
z1W#+NioiV&3_FENA*5R6D86jH*`R+C9{?|fnH>ukEVjPdI><%bXyOeFUO<wyB0Bp^
zyYv)|W8Q(q#Az3yFNQ4==fM3L)-J=z?L&PCpi-?xjJZ=E`}xB&ADoH*Y|ETs$^3ho
zMX#`Zp<NPlx!(p#yLvvI#a#+mqIvO**TE{nCuvy;C}UsQV6$Z79SJRQr!7Eju$dk6
zmoAxSL;3QV9g4wFstL$dfiVp-F9pA8_gB1NtLkd*xNMtsy5iKt<im%`Fb1w1vaQLk
zEFT+p9`!j)zG9w-OJnKPkpDQLFFLS9#!!BMq_Azt%8k_h4s(OMGsVO}RBPnZOa;y8
z?5|<KbiB#_Dl$$@%#l_}40dw$qB?0ZQD)cO*2%-FuQi5G2VyhjMBhH$D2TOltP<k9
zu#~)lKs2OH+(d>?%>#8lM_LcHol+T0<U~yOukZ;NuLu%U0!|uMM*0PvFmgh!_3CJ`
z??=7t^=D^xS-08ev;MczArzNxr#{vpb;Z&sAG3)%f$TXXAc}R{2fv~unQ{TVY?F16
z8*rQ*Y;W&6kZ6nIp^!uAf)ok4t?7ua;Zk=MRU6*1JRfxGCB$%ZcS}}V+g{&|7&bXR
z)g5sln!u}vICL-CgL$rOKv(ZPt40nf#1>99{Y}Rioi%B5r6vZOgcp{QoP!UQ)<w8N
zygWtkxD~{y(ObhOc&k0!AS{Y@Fl`@7l?tmQ<|Jyech2Ifgdb%JbznLJUw7U058ES2
zVZZ`#vlPf{1E2P;#G#N>wdovyrsV2JvH5%iZsOf_t;kxS^+m36V2R!9+XF)+bqTK|
z&z`SCmCbQeN(&a;Ja6e@drxk%B_w0vQ4;TMgZ5~%II`ACF2ddAd56~wxyz|9_?aHG
zjdjT$tJ4QpD4F#Nr;}LT{Z386EHTKEU>~#hHn2lzB^1@R+kt76V|eaKy<qRlHe_CD
zj}UB(-ekmq#rMnQ@sLj0LJ0b`_ct}o#RR-p`HiT_@5eP;G6kQs?<}pWG%3YrD0H2g
zXw5kYk_#Em+ILkHBw+*TOyRA8gwGN)1p1bMCNwNE0xBIwI5@|zr6`WM3XKYSc}e1I
znmzV`ob@|zah`Isgq))=1*Jj2&&xho&#7&#A*Yby{npX&ME~y6RoGKb9r3rPyuy$s
zz~3OW_Mv7-$2cA%#@d0!C+SLfe(F9Ic~u{_4_6eJbvWWT4HNIwm*qrZ#8;I`_ajLt
zu<4PgmP_ZGCWqnub>g0}k2W9FHcahf*p{7|RFX{g5<Y>y$EbZQadSmy%6SxSZn(uh
zo^vt9t*2ZuUrE#5Q1c1<o;pvfah0b~_+?$28|*-!a*crh-dZ1vmne{_0mRxLaq2Zf
zRCi=T2k&HIf4upPqRlzYgDsQeR;Znv`kYVsU@%|LaF*`pL^@<ol*6n$<~)i4<t8h(
zXP-S;(F%cVpp7%314(8K?v|u{l2rND2~bA4<4)OAb*ff#bL>=ar+>-;C8IywnjB^J
zNq5?GRiVJ*>txT7)+D87>hp-=RF7z;Pn6}-JySYWmG-E@lz&dDn@dpc@2iu%hI{ZR
z1Q8bFLH5&T#dU8$qGE+fASp?`?FXv*0Qo?YtBhoFy$e+uC>Z^8TFsu+n9$=gsZW*}
z7KakLU2r+{`A*ICOe-&v`A6$f6?*BQo+c70!7kgUvfFQc_>ueG_7*QCJS9LX`*axu
z*iPqy?e_|kzp(CXyEJ{!cGL+Z#AIAob^s_;NlPVdB!Nuwm9uBcc<FR_sLJ<l)(1{X
z2AM}_RkjqD*HW)5Y<E_+mqamSC>2fAQ2;oXyPr@`(%BB$uEej68#XzQaZn8mgSEE1
zS=LM$LTHKu%e5bAf!d_`-IFI$cML~No~6B7gd@e?rbm%F1U})QSD3a6m)SyFFWbJ_
z9Waj$;_VBh5S9?_WX)hn2=27~%{wE-2Fr2IffX)3FO5{1{=7O(XU_#Y(ELX4kwIs_
z1B)-{uodi3m<f-di#V9$m+2H<ic*<XX%B<!*`Z28mydS0vvQP6Zx!j3R4McG9iFp`
z<Ss9#@vv{RX7Zn*UPt-L%|xFZDPO{W224&3E3kO8Oci^X0VP;PFEYK~j@CDGI@Xl5
zaxPGC9A9zj<Cmcn2b*-eRO|~omf!3l<D~PPd!*A#c%Qg)B|R-2Nd-*E5)MpA(HT(g
zCOc6l1zt>QR=Jh{i$!2ULlD*nQ(Dj)&e_Sj$b<N2p=)hBkepiH!UiTs>`8mJyoI>$
zum+?jt&C!F&%WZ+r>S+_9oy_w*|MQ0&@O;Q$~o4$XNJ_0aKdSu#~ub8u3NR|n)#Fe
zm^9zYVtpdg>Hu--%epJEedw?3bbZGbQruvVp`5A!uE0jceWsEn1n5+*Nk=EGri5-O
zh4wK|w8|xgIhMR6N9?rEoUJu%eQjzvT!#4UO-<3==%%R1MP%plV0+0cou}NCJxI2L
z?T74q89|z~ggfUPSbSVqFP<m;xH*9ALN0JD!%?Rcqfbp;#K_pi=6%wD!-}G>04B8H
z96`{wOLN{{uRRp+bHJ%dJj(!<*yMJ(zK{eAk&6p1xx(~G3H@CBVOPp)6&HsJg@Yhc
zWnq@8HOU->Umd?utnd%kxh%p@!{RwLxoG70g+KhEickv{FLMexY^f__qg}A+@}~5+
zjWuLFnLekcv{qi^P<`0ddbW~cAkz#xpu$LGPP#L2gjgYh?Lif1Gx?Ksx37Ls?(d?{
zfC};*5^D~Jam1dlZ$|Hg_Cro4fW+0j<zf5bYi2$TpJ7~bY7*}^<)nY6mix(B-tKqG
z`6(=SLPvOQ`;pRPX@VriN76SYQeDB*%c+kOVsxrCIB7pxmnWz+o<%{au;n;4IdTp>
zbnan0)K=h}W|0oqkCpm7j5SOhjyoljA~>CN>UEUd-<q-?uifr#4-C+4%7MtKqY#oQ
zf4P#D{zR#fi<8h)mvd(-idrMDVx{dk-H^neobwwgU`dtC(X^W`P*d0(+0+R3_^Ar$
zZYO9Cp$k9~+VF%(LX9xRQT5YJO*pd99|=Qnem+~HgOP-)!70F8+0T^a0HYC_!lK~7
z5@R<_4)jjh&*rQwS+K}?tlJqrCW$<MG5q}8?0XQT^qwm|3sW8qIT>8GpD*o-A)n>l
z<-oMOGdtFCJ-q!weG~H`7VyOLk2x^4AbQI=Fn_UG#tgTRM|3W>;w<+%_1S*{fKbK#
zQiByz9IK(ECeiL7lTLkNtc<^ejs0@zfRr+FgOV@}EHF#dz8E^F$*au8p#4ho8=NUw
zP;`(EtegNHn4SUq)$)#XrAogyY1<Wdd}!vpH}#yVLas|<Ya^Qw`wQRNUn}HZL@M<o
zGGC>S1UFwYG9>_~RY4D>RZYLUJcob1{vJ-C2mtIq9UV!?ayDp{i_uSTk^M$p3{+Q=
z65$lR!o-)048djDZ<co2yGpO5yyX=pE<tvE+;{A^k{4LKWa%>JS-Wk?()sq=`SS=2
z#R%L{C=xK8yc^r{2>zWiDtQk{QFew|4tBw*$;DPmp?>?_{1xeIQT<TJI>kC}wco20
zZM?mW%WSaPIG{3HI^tMe{)*V{wBN7$ZWJvd4j2wV%ZdN*fvZnX@3TLs%iGu4-WlKO
zlyl<VjGo;7up$KKZSWf8#091SZ@hh@{ZV$4q*B}tNTGHDcCjumX@=~N^B0}|Ugt3x
zkp)Y+vcCC6`WiJ#ey1kLE7v<{!H|HD*q;^{dMCw<%K+xUk|S@8kUvj~i$ANau++#F
zNP;>CR*sTz(!``7->b~XcNQM&&+8nl>vomQhH}mUIbOHeUzA=-U{b(OE*ezFsVVPm
za+xvhl90l@W`EiI^0?>>E~0Y(&@ZXUeiJn|j@aF`zp9X{AEDJpU~+-PeD2}UQz+Rf
z=}6Jl9ksu%s{$<+6Mkgok^@VM+nqvN3%&T8=I10VEjW=Mr=}c}QgkbS^%48qG8C`|
z22wJlVnkD$avq-p>Z3EV!C-cWJu5Ci_ILHew_q^rZ%sLXm=$Ef6u2Z{ibArrf!N>Y
z=dlY7O+19-F%hZ`OgOcE)q4AfeC$D=%VQ$TotxI#KQ_IFD#c`8^a_*hAx>PhTmMui
zl6)v!H-$eDkVMSs(=4DSQl7dTANJ3=mAaF#@mPhx9|u$>=L83?R`ay|OEZ<I$bl(K
zq>QQuoSK~Vk{ikXHM@NEy)(N;ycA}3tXjRs{w-^WjBoWig-Ak|XKG1Ch00$0CbQ7W
zbqrx9X01tF&ajmh^xu>G%<Q=CM(4ToHw8`QfcuXElZje;fY5aZmbe<)`#i-(R;h=1
zIcEP^ezR{zw|b?c2C(=|I1r1VFv-$Jl%Cmt)pv%N1=*qh`<`zilb!mQ_#8*r-T4}x
zo%Y{#-bQf_CREh{Xc#j)?&xeKo!$W!&+Mt-({7t`o+R(VEVu2mo+{L^g{HOb6cb0*
zAW(nyQ#lzWl`n?=D_&txeKwF5?}UBMQ)ymE#X>4A(!ERiOeK&++H_jQ@Kf^Cl<Cqw
zu7zWINd04<TfQcd15G~Dc*udpkkbuBG+5e1Ce=Q#d{a75vFHaOLa`rtmeduw2Ax*<
z`yhNGIW?o(=a;WBfov$*Al~B@CVhZ1yB(a~ZC_B{Ab%=JwMzLC?4DB-s3M{$1SavE
zePL-G1dE<F$>qJ)2nGYA3_(HxcGwp+<K7h>QrtM{z>=WJW2T?rC;Q^3;-JB!4^LG#
z7UkS3Yla9lMtAZh<<nu{=hJi*{FWV=nwPQM3`qC%h;2|p!ekdXPw2XYOWB#EF=Juu
zOG}sXkWMmJVT5Ti;StnpUzY8lJ@_(ows*-2=fFgJ7A;$9UtW=EEly1oVkxxb8NXb;
z$-iRGqzR><5%dWqiBl6(tc9<!tY2AR(C**_`Rf&?{wm{!vhIG>Q*r7L(x7W7DaD&m
zjLU^zjJ<tz_OwGlCXZ2FEX=IvbM}JfL(u2sR2*0##<FEcC$)uTSzsqTXJ6CQ&zkfl
z7L==uocdffC2k(;PTAL%ALzr0+7YcKc5-?*C`o-a;)l|6Hl<!#2Q#$g>zdCtw3N;@
zZGE<W);SkjZ7BDdKl}Pz0g_QFM?!ma+Yn9#aV;k`!QXrwbR}82eZx~FO#2S`9wLe;
zA-DrgwRd-Nd~AHEi{|zlOP3I1kfd~AX-Mp;pc!YJdNBn(t2f=XdXs%qV#;`%+u@u-
zU`nbCG!EI;l+68`n`IK$5V{Em7OlmE%X1>K^WlYY*1qMbBJNL)b|=TgRIzUWmR*~g
z+C8BAw>BS?$y@H*icZ#rbJP^Hff{K^@T~r}>?>pzxr>5^!46orX1RTPMffe~Whj)6
zN#B<ehB)g3_8s{nQNQQiDaALc_J=T3dtsdqH1<v*8dL^?^UJQQ9YYTx?`7Pkyuz3(
zsb1L&2kk{k-chKDjd6<VF3}6`;S#><>#}X?NZT&;FQBCVH;{Dh#q}d#u=0MNpV{ZY
zVpLEtsBBO$Pz_z5y`;3>r^|~AjRcl+0HssT*Ig3I#Io5->xY)F-|Q5U+>H$*ZAmP;
zp>{3JRO&k*=^1WaVK?R|n197pQ4)OHq`j>4AT^XCapxUaF3`&5>(|@M>sQyTUE>sD
zUbR7Gb7BkIE1vqy`(r(4cKqM}b9$3pMrhnZre?C`e=OX;Vy{ecGSuxK3YP{4kSs=Q
zng}>yjV?leME0Mxo9aYwCM}(O4IM}j7$J>Pv5?!H9Ux~ZNui1YAOgo;m89~1cr6~w
zmPk|JL3CsD>Ld@g$y7oqTEsC)e{qeo`4tf$mO5^AkjFAPHAODRFJjDYuJ{!VQ+3Oc
zfTF9`h%u69&`#v=#H9uK!I>Rs%iAOPbb~FxbPTm8Mr`3V{n7by#w!eF1;h*$lPxOk
zg27<IbQ@xYiCs}HWlSHoxS_UHi4;l$rZ0WJsZYFufskwAA-4RIGOV2D99C*Es3}-j
z)m-OoX=$CO!kjx3otiTHCI2Od&a!0ZhT4-4I?u7;#4zcURZ<1|4oP`+`DLgsn0tln
z0854F@KCE)*OB3!Yp==0Y>jSJcmbs(2pd`*du=U@P4+v5_>`!ke99Vz!d};OQZYQ?
zOz*QWX}8gZFp;k>?W)0&#5N8r-U_V{i4ScYjLY+n+8gpHLLERFAAB7t$5bfId1QB>
z%^S<FwOq>ShR=QCh*Og;FnLU9vu6$8I4oVL?i;b?Po?<{)C_th!}DC5-W)E_X`5#`
zBBv6r*ovH;p6%`Sgi}b$7y8J#;_=G5SvE_|r<AjZ@SwX}<Yy|*^N_7-W<q_{395M$
zTSQILw2k;m+v=PLg3!1TavX}-?)L&?geb}w8y?xVrhX|jJ;{N_?sjT&Bp3m^B*U?_
zrKwzJ#ELh8mrhNh!<)%qd`l(Ly(FsKGNj`bP(hdF_u9Jhn&itP(HSEkH@H)ih$lCz
zaZ>t<=HFT=0V!Bu*(g@Z#45~<0TSWJEF>50wt@v(vDDIBbYNmy;6ysaHcr|4N=A_k
z6gw3D>5x&$z4*U6_34&}k-T>Zhi!g_&T&JI<?eN>Hs9hD%EpW9w=s$zvW*Rl8|R>u
zx_1s(vO+iywB<H6Znrm|KLAoe`&h_1S=F8TKmoaECjVPd)9xrA*iS+o8B+;;bzmZF
zo$YOVQ=RPl9=YC|!s_vsZAuJb)`*92O1V=|gz7p)@<m%uT>--VneL1&0Vu(6?a=tJ
zZLUZ$L4Gg=i~`6RV86@qs~c$UY)GMm*%Uf750b@BPw#i?%WqIcgWZSNqj%*1mMxm^
zJeEz8A&;4tM(n}e4eIE;5Jk?UeaF4lXdab+7d|*QvfWdDlbI3s;IFU2o<jFSjj(bg
zjPLi>S?g?>a0=1EI03ZraP59`gLt!ka{oJkj#g!83RMqY)!tHiASG}~{#H&oY7*fG
zxm9bT!}UB++WTrxR&2P#DI_7_WS1vp8iXYvzO_v9+@wN9#S5!?#;Mon@HIm5^|sP2
zX<KDJK*iy}isUVSCcRp2+VZcB6|=WDO#W_|{IgCuk7QlRKtk5a^CPg3P^R1IV|O^v
z-09FeYc&JU%Qr77d6-TFr>00uvy-VF$hm=yV7;}!OueKiREiEPU%i0vslH}~PNUA2
za^aX)n4~!*SByr-t-o$OWDwL~iZcOJrc}#OQq@EdTD#5&`#ALFh!`1pC1&ub*Ltzs
z@_JI@68RIuaZrE4wlqsZ7#MFxR-j&KlGUZ1PJPbcdZo&@fzn*Zp1gSer)G!xBtTn7
zJKKlb<berH9n7~(j*^E9KOu@$%Q-@n42wOKTWRUd3!F#InXWqO+WiS9pSAOt_68hv
zM#j3fwf1a-^g>QCr-(j<`x!@}__otFTo+-{;(1OXPY4{BCx~ybkviL$?D0>Oxe!pi
z3XLF?m}{MRRhoxKpDG?^k)S5+<g7j?lji)I>?tg;=y>5Nkin$FBi*L+j_d(CQK6-E
za*Rus%y%9|wtCpmHdb)?Bd9b!g1(A!ZoEmRjHJ$!T`sVCpg~vdG9jv{3mlc#n(#NK
zGm7XF$9UMr^CRTV8F8Lts6t^h(J02GkLE4bt+UqC-tH7?Gq~!+FP&W%ZK5DG>=|0H
z@N05l$-9RQG}U{}PC@vctjn+TkG;BQ5Qx^iKd!0WR$d8t2W4?3$1*(&doCoL6YP`i
zrBh)Qi?tSuM9`d5Q-p_|f$otH2{IvrD?ZBhU`3@q^2u`6Q!eepp@gGU2WrDu*;LM0
zAFdA0b8hKd*7ev!wXkU35~olo@H6ZU&Ok)Ths&GOJ2`F$XOa}97>txtUnX`oU|fjf
z_Rhk*ZWl-4l!O>FJKnUy9x30Em@VA_25<+K+jL}VVz6uPs(e*1c`uz(Vq7okHIy~>
z?(Fsu8rBh~5M4+_3562aE)mk}1cwA8dQyCteYPfOo%jVhf}e-Ir~Iadf#WxshniEP
zO)vL764Qw-c{Dp+E}nCWu?jl7l+#LvYaRO$u7eHsvGQthamv;t9iWtuXPlZO4P=1e
zCQf^A?x;;`S2&N7se8BD`?66BQ;zJFPB}i82WHT4K8O1_{lydB-rsbmF+VZJz2XQs
z#GiHQ%e*NAWrjG1n$o_r{8HO3%>rlGdO9gW)#cO&qQE*qIYOUe-&N;LI)!Mh?X1*@
z)FiR&CAlL`23`9=F5iPw1L$oiBnyo&vE4pc*d_erruR9e7-GmxV4M=e<`AX8I$_^k
zXB0+KY~K}#KO_q3LCrOjLz2M2Tjj7<=G70CUy30|SA`5#^JY+^!OiUGZ4*?}9+cDD
zhigB=IGwI!nh>0?n4$wteR3cmk~Cl1qaP{10U<*}mKT{QHMOYvvrc_8{Yt=iidM2C
z9kh>@nw)ZBQ@j(!JvG|9T^RhXeXKM~o}QIsQ|d2iR?8%cr#@gGPrT9n^sQl3gp{MR
zDSpaSa?HMGwlhJVq{-J>AM@FoV{t4=woF<7d&_&|6C}l#wB@rfDZRr8WRKT(LvO{4
zWa=>-pq!fMiHM$}-sOPs#Ozdt@>ve<z!HPdOZ0Mj4BM0Sm<DNjLtw_vMTXVfH|yKm
z?WsAF7#YICpG)^(Q<DU6^ZFipx_*w$FA~5iIk0T7YTaG-iP>dj<Z%Ecb>W%w!)rYU
z6mtf9jK(q-PLNCXePu27vb5)&Qk;*g-*l&ae?BYz$axNQO@mab7egZCTU<b4KagFz
z4a4}TmqPE5GCX}UYgpO#fKv#ZN_DtSLl%Y!zhIwg7D3(^NhG7oKBp#jLU^HRLG07z
zHzJ&HBrJ21fg?U!6Va#Wc#D&*?a28gbf<}iabQW^NGTxc@|h%?eTA-4)O5?5GGROG
z+%LR&nN!F^p{0BywyQD_L!<p%K8sX`b2Grep^RYFmGTE|_chays}UMcrzTn}noT~n
zawpqUPZz}lXoJiGhDf16y7#@6&TmJEZF?zhR5&rXxG!rnZQP5TV%c6|tHk`{q*`E_
z!{-8Va8hkLsUrw(TKiw3njNU~&MGMIOgw7ZEvYGkmBYVkF4@6?IvS>&?37+4!|p_z
z4m$NI2KONBY#r<jkJzDtGourOEvFFgO#6SvsU1!jQDGmP`9S{j0s{|X8$(*o*^#GG
zNfQEwdLM%Stol+IR+5T1ZAItlAy>%T(YmFVhgGQ6bs&YU<X-lE=-aW{akME!ttbu(
zP*EL8GSk)w3%tA|i(r5?-j3&y`TDmwPr2No`bp*0i8_aAmy-&@D@-he1Pc<;*~w;o
z(wd<Qx1sKYl+>EKn$lY)Nqjcx&k&bT@)%n&?i#$R#DW*&qK3<MDkp+iw`KDdJ0)e?
z#hHgZmIH%?%<C9@u68`4OF|9m6?Oo|Ox|<c<87x)J5f-X+vKSVEDni@e$oM-sk3_9
z<T&#<=m4a&co5m-Y~5aBV@I80lD)Mfy>_m=SYckpH^eI}eN_1*68ibt(NJ@TaoB++
zxyBb9MJ5u$h5Dr%ZhY$v_c_Hp)*&pjiAx)QLc7>}{5Z1)nqz+kmeK=CA!HnK!E+_i
zrShW60%cgMB7RdnM}1lUG=0IB+2yibb!*1yo19{-lc;&oWP#!sYwb#%k40SCK_Rj8
zuJ+(iA1vAr7Su0+6!H+`qm7|24s>OIkS2<@NyqvXX=*=IdL<*hGCk<v1Xi$rd~$Rv
z7NO}Rnk72xIj0<R#u0@`kjc_@+KYuG#L})d=z5lODk5_ON3J+ETJiNmTWls#uQeVL
zCnZ%7A#2aqu`Zmy#3|&bLXn)iS|Ydd=p3BcWj|c{pXPS;|AGUH--!2TH$*}a#eB%Z
zAam?z%FfNRBz&Ae*V)^u#WNjuTIKmg4j-8(iJRfCG%r6==Z6EIdj}>QfI9!`8((QZ
z+ALuTvSLvp*43#AdXe-=r#lQt_;lo`a@YP?>3c4pbblNT(MnE1JDvK#B4`Y~iu?TR
z#~aL2sF`T*VF#uSG||y5)IU)=)u3h!h?sAuCZ;O95FuRY1{#EiPoJ}&tmuKYC=ylS
z=mjh$CSA2+N3iavS@)l6-X_mG#EHf@9RqRd%P)6<-wF|yd?3;8rwc)5Bf!GelD=@%
zex~VtI4?*b7owuUtTh_4nCS-N?PnVlh-CmRB!o71v3RY|-iu(9Xm&OP%zmztjWI-=
z@csj$BZKTlL>k)9H_ITg8s-K|6tIGJvxgsZNN+g;r;ph$B$ljMv&MN0Tu64gxC%rk
z@;)a`-G2MU@<Nf<)DOwRmwp;#PIBtSz+L~ghwrmr%JuGcCds5hA&xi1V&?CcoB5M&
zkL?i5C9Hr`qZhpHy6awVzfu{a2MAIl=NaX)@d)w)sYK55SWm{_SIZ29uRSBgC(?nH
z^%dC*o|L~Z6ZUJRN4tpCde$ig60`P72!cuaoc(&zrOg<eB~VgKx;;4A?M3YA+Hd6a
zJ@U@CIgc^E!^-qyzgc@8UY%1;IeQVdlj=Ook!inG#)Rh*X1VkpQYe&yTqQbDpLAU(
z;BZH&f(Pukn~9iB1e6<TFTCa%uk}d=HAN_Og4>0@KrMc!qDt{mmS-Zc=whzFNW8Mq
zklNFJx3pH$Zv~FY9#>(=oJfvRf*rEoD{n~dS#p_#B$<JziG?eDM5dJ9=Ti8nNr(RX
zIe9&Oaq&GRiS(~}vHd}LqkbokADQo>e@;!4pP<`^RzQbd*Z#0s#mk}UCMmH4)5<72
z6I|0DRZ2eI+U69J_ZuVnfP)im^^fbyt?RaW(MnP-D;JrsAj1*BgL>*u^1Y<593-^}
zr5Iv*WXXYC3~0vQ%FXUiD}nAE?%<I>?hRl|s~00u*Z!=23A&cf={3_CH9|aGL$E;M
zv(j&WUa?|1cZ!Yc$meEvfN_wm+hBjuyep?nv!Mv7OHPeuV`j$`8RqOSORqr@nDKPG
zN*}toV8^L1q+R2d%gQnPs{-RUoU|I_J_nY^v*b--=wFu(gpVQHDQ{>Jp4A%GGy*&1
z=AD-JzbP+Hq8aC+H?S1nbqd<wHl4Zz?uF*ffyFYxQOD$Epbm-D?C(mAPF7)^SgJT*
zotoUwQc<FM8MD7{hITQ8d6wLP6{3^ISDm~mhVIuHX#Y^B8U?X@)8v*Qu5h<gBT}~h
z_Kj=pA4{v)4-g>rGR|11Cf=XM+laj2_tNvGG4!8Gz4pMYt9#zki8q|r`q`L&v>|ai
zbC6&@`{y#I=q=-iFLhvv)S>3i>rL~l|CdsubyI-XewiV8+b*X@QtH)@Ayde|>R&4&
z-PWG!I;FCcT4XJNtL?LYs|_S1Ii-RjT>i`TM)vQss0T66+#KqiLYH+*BK3)SVZ@7@
zQ|v#Q^@$XD%;G)=mh-1xYUab&T9Q`i!|XqES1ek3v-4Q6SqcBs81r8Rwi9x74he+6
z4y@R7BbQK+oqEV=VT3+=?7wq)u~tku&nX)$ShUdW=_1d!xw?c5Ta@+LNJ7&d7{_0|
z-#+W<_`zg<zg$_Lo=bW~!i7R|`2FnsDGQ<4523)b4y+)IguKYC!cZGl(LU$tK)X(t
zl<qv|ZoeCm(LT3%Tk=#SNpWC8((%bHTkP|$d2@=7CBj3_?@oPL5>S^fa|xS$ehy@2
z$319XC`1FX_3|eniMf42fysc7DS8e&u)IP=ce#+=pIc6_5q;s)u^-{RPs>gKjRYc!
zwq<hMzG${@!M$kg)no4<TB};GPb)<e+yV7z6H`K#FK)Uaoh=M6>@=`4lgeI@B64zE
zddsu+B{?>PYfN+$qH!iBN)19-VnT>HW?%Yr5n{+XIMjBEb(J_s(~~%TD?*`fUzSL6
z`xb(XWu&B>=R6d2qBYRLQ+C3>ym^V(NU=$DJ@$6e4(iK3&Sa8uFA;4yYF|-?Tirki
z`|%3PX^A)3S3X_1KHjZ+)+NPfXXz#kQ=Yc8DhnSDEc9THeN_&CJhd{r4>;v$!)3;D
zKxDPg*jJbSiMf#>mpuN(-Hw_>%vS&X9ecskg?e=RuxD`NPuBr8Lz8+}2>^mGvaiVx
z&5$SnX;IgJwjG!d<)x2YZ(n=OW|EZ$s}LIHlTJ-;(#`D&`?|VZ$=gxPv!FZ4di28x
ze0`aF`L7ApGRQpBwn;fled%CaZH;+WURM%~zoFEGqYf}hX%TVh)WjTVjg%Nb&WPVw
z7aj#=q!8(lCk|jvczB~K0{f(WQ{C5$w>(SHXWu;AHX{DXx$X^&<1X3s^Y$&JU2?}E
z9XYVTQmMtNDou=A-tT?czBRwObkV$d&U5L!bTw#t;eo!b{9Y_(v6E|8#)NVBIQ8)x
z<+hUN6PZ)&+ne{CEu!I02UZUxQD$ic67}pm>SVH)^O#62#Brx4-YqRUIYVX+VpMx!
z_IGB-n>)_4Pz=4}9)$AiYEKwx9BCi~u-v25Cpj+4(akX)7gIvvUzAw2VXDu0yk0L8
z36{xc)~Q--P^RJ31e}AVzrGoaCIYKV@HygxsS&Y#<Q4bbXD@DADGP{pUdBCNa%!4I
zkxl`74Fq1hDuUeYCHWTktgiDc#McEWFMT@L=7&a{r(#KmmhS4<4Rxkq-L;oUdFa3j
zeUkc4YKdygPDbFy{K))e^A|eLu{xkFssn8Z8Cs#B_Off%KNt?`&%n}!*OpVS0Y7lZ
z%1!q2Tn}Yrg`AcW*?*<Fensu^hSi&IxqX#WE_jnx8-)+Ovcl-UD7X}p6(-UmJXZ>^
zn@VrfJ}_4dp(e%*KF~u>&Ilu&Wfflabm39hig@gH9YAzL^aC2eO1RBS*vt;%@;eVr
z4Pd8tVCBUM?toQ@6+bMBDl4<iPwKJ?38ROcYa~(wQU)mbNn+_?ySdUt66Fs&B~2ZZ
zdDa$G>a<0{Ze;v&AW6%%l1X`C8F3yE5O#P4nbmL}EuYC)4(-w-wy2JB3)d>iIu0QE
z5*F=Zlv-TT19h6XPRVMyuzJZwul0c}d`zLH^-P3J+miZ`L3nFrz620BBvxPU8)7fD
z;||->#M>9&=R66Sdq_91tmzZ;M{^?K99WKd1^UHTmj)6F<@tCXJtDB8{yr2qodJ2x
zrN|5eZH>LA9yYqUJxzXqG}o<(?xejoztY#y_M#9|gC`x*xn}k`dtIIFnCE~c#d2!I
z1j8+mGOs(-OE+JiZ=Tt)@^0rji5!E^RF*n%Nz-R<xMmv#{|in?WfKYur#?s-q#gS5
z#xhK>G9+jraIV7~J)YA_T_-A;LjrULPi=XFr;7hNz?mf`=_oi0(~13*`aA_OvcttA
z$E~PuAyFuYEdn}8OlzzD&SdYz7J|H2HX}PuUt|PS+zwvXPEFA&DfFmeqS<(o?~unp
zfqk|rXQ_))#d$8XRPGLDvy1TTx7Bqq*o_KtkOLO5#EDIJZnQO}fyl6OCL`)Ou!46<
zjAath)E>6ANpsS&K=1<PY{^+A21bd-6k&1MZYl2-zRJZ<a&(2U{t?{6H*H<*ATj;Y
z%ga6tOuFqsqNCXPc5C^F+}(8Rr;MU;YB)71LAA!l#tCtQ1-Y%fkKG$fEbS<;GCh(A
z#Ab;TS30Qm*VJ&<YM%*3yw?0b-rh4n&+54Qp7fsBiPNhcY-8L22_)<|iI4=6DUu}t
zc8trtySH{1ZSxj^L~zF7G$(OMvuTM?mk^Q=pk9(?3F*D}-t(mQ_Wu4ebFN*I=ga%y
zeZj1+Gw0mbb;_JM(@%w3p$E``LsW#_(xz|Ob<3QBcEMHGaWna?iE)t{W5y|IV@~u&
zZA0}5i6qrKXc>W3pQKub7Sl5g?YM1hWQ5-6l+n#A3R}g8O=@qaUQ*%pAY1H2drN+4
z6;ct8@zz3ZBHiQ}cM9#{rbZ}9;WX})md;4zI@IbNw9Q>#6#$@Bf7}}^nPAxlPT7`*
z&~Kg`nTS;NKrW+SjOA?Vd(ssJ4KehFZHb2zvm1zRw>4o#d5;xM$~M{IK(+3&UQZ+2
z++Lle6QQ5zvNk(4iNQCmS!;K+k*$XOIHk0U;C4)4|69A=`y~5b!;zUwNgz_oMt$is
z@aNT4q4u^8fr5U+!#?4_vei2BSwzaBZ>YWrz2(=m8(~ODe%7f8@|T)79r=6g?d_=E
zsFb=poN}~5Clc01F3Pbx8!V7eeC}ZfmV<#ja%R1w6=l1eaf;ci-ryjY{Xu!&*}jd^
zNu0G@?t#@(D_l|b57~stgk$fjxk<&_0S<UMLna%^vr{h&UUBU;_U`uOSq$nsoMQGr
z*w#2atyeIaB`@Ut@!nu54@9CD8|c=KMBo&nbIp6aqc=9#1~)i41`*)M1)@b?f2-~~
z8fw+35=cHgg*#}s(#p80$Xi{38^L?h$Xmba$$BlL1)E&V&k11`__g1uk1>hDOG9=V
zBDOVQ1lsCW<e7oF>0t*JT^26stWn^okouerw{z=`>z(JCAz9)>=(drD1#!pnuv3mU
z^eE&Y<zgJoHfW2EKl>3dfO59s%+p1cm;~*vHg;^G;4IFWK#~rwAps%zJO_8!wr<%K
zOfE-i3Byk<oOWtNvaVNV;j!F1tCgsB%qbOrtMfmMQb8?SDCc-vy5(11=M)k*2PfeH
zY@$s+2qp&+1;>b9==n4%OzO3Ile2y9-JR@CI5~Eu`XLEzoT;pc?3c8P?j*Ya@k6%G
z^E+&4Hr3{s!v`}Jt9;gh<(}ZKvTbtgv-e~xg%1sCkIR`An0Nzb55du<vunhia8OeS
zDi*RDXu@Sp9l=1LWvO-3s}}!=&E!npci(-lah?M$UJ#%w7R+X=CPdfxrKq}6Ewq|a
z(wBgX&|`D0)3`h9Do}Ji4our|_3ACQy~DwD5K7l%IU-S`#T_O4qut$5pp<dx*+7B=
zr$*z)qh@-?nB7zJxNm-sjz5HiOAf5md7`}{`4DhcK)!NH>mIblcC`0wTy^7Grxds<
zdX2)5)<<n;!KB8K0Cb#pH<-R8m~zS7+olf@8PtP9%ifAVwNn#s)#0mXW<{iZW8xm_
zXVNIlIK^mk2y=}@39eAI``X~c{X&eBG+a`qcQr9B5N!kxNC57=?`_*?78jivt|A(8
zK(*WDOT%HDa_jsN`=+)TmaVwfDQMaWxgRhuv|XxITgqfopS{K2kugwD*!%L!SS2$>
z8i_jr1@?n*5;H@YnKTC`347H2O$U;fIE|4=ZpZ;9>-b!kSb!-LXP<x5WBY-&b%*9e
z%_-&#Bdg(EB)P=i-&U4Av(+i(d3!5I>NnR;#*b;MQ_#RM1x?zw{9iw=3(|AfY9#kN
zt&-&HwB&2@E*8cdwr@=Y->_w^^B9d9iyo;gL7_|bZOQ%MsXl{xmTdtb<w%$e-Cdyk
zpOOGt2+_CaTi2}J;ylJ%8+GcW9Q#1MMBNdmf}9Kza#+Tz&O7xnMGPT$7|Dd+(fUZn
z@=JX9F~t?@_ugtG$cm{Cep9I`>^s}EuD|t$4NfU_4jX$Dz44xXSM{Q7yBWF0!ZS_8
z_Uw1+Q{TC|Pk44@TI|~gb2&x_r)J1!nuA@}zsEk*x);u~&`@C!QovcKCU^g8oVV=5
zU5OxJr&PG0>@Pu>>?7T@MNTIrXs3p&z@3_;w&21**s#94+6q%qn~2HIHFj!ho%H5p
zYlRnVSbBBBzNZadya|)0%L7xrB);m9Q{O0(^aaR+(*0D1%{c3275(1USJ-AT`SOj^
z`SGN;8of=KsR)zeExLg1`~D|08kTFI!ffu2AGYtWzF`~AFuh>OGrua%2u^*@8oh=_
z!FKPlk5*6Q10dcof+azZC!Cs+T81L29fqBcb@&zJ7EXDO1B(HR<VE&CI-CR`oeRPu
zZVzTR7j|tLb)GfF;bOPze+M|*Lp2PY%VfD>KS@kX-M%zK)YmX1r$|~kY!BCOv=mrp
zSyly8%b`>KI^opUZ<Iql=&U`Gu!!Z<MO6wN1eUmt!FY_Yxku~0iknz80p`_-&mbRB
zFDXY(nMy=3Yd_FdiLF1~C;hHGmK|tJlH8VrX^{X+xJmcL2MW*5+Ycs=N0*N}rNVJ3
z-9p1|yK>`94v#pGK_xW4x<^1SLzsT7{g6J(_944h2ZQ`1Dxb35ja*Jl4mgD(I=ZMz
z6PUw&()P5P#7GE8+AE%PYJ__n2it9L_H5;fWzJ)=T&reg$XB;7YhL%RH#(1rQX2*!
zB!d`VY=7IEW0L~|J7jGDlJ@_%khe*E5V9!^59H*mTf1t_+D%SLT*j8w8|`2XSJXMU
zWcX2FH5!SB>6)Bk@u5}&iIm5D*h1Ogdz_k@Z@KSC;keHZH>$&Jm(xxuTF5XiH5^&(
z?eT)82Y%Qx9azwiP+I139w-;rKq9;zY45>XLdeM4JFxss2Hq$;7GH?8NTjAbVo!8*
zMYJM`buzA*GbNb@S{3zzs55Sb(1)HWdtpb5ZPWok%GofeB}dyt@kI|p+NN2Tg&OLQ
z@33PHz-5)&ly?k3ZWf#z(EZx+9M{6GZKFdI=sL8Tz-lE<h=GCfC@n0~n?0G>gC6^m
z^Bk?zV>w7j0TQ&=PSjlO6O#&8huzT|EHk5OF3#Fh4e16X_&UY<tWLF5Nby1%KHVmt
z8%ds)N|ap;ELyio5CSM}`$>DICQKbyS|vqB%MV1ZQMyE1BBKUGegda$PRGrYZ2P{Q
zY+qtAC8uy;c)=ESp$55!juPW*r(W=NkVd+QU~j1jgiCs=<vwYrJD3iYu!cG?ZA21)
zOxl^8h%wSU&tZI|s8f`%9WwUV*}@NL0+b94CZ11qo<>f6(R>`c2teM618YM}r6fyb
zq7YLD2;AEtr(UCm%JD(=>G`%0Q@t_n;G>vl@9GF^MssnYsN4hdhoOv=LVCYyg<WiG
zCiNdAsIG~aKkU@RJXG`~+jYNP$}uBZ_MGQfzb4h>_QZX}eyGv7<;#~l1z|MQB=6p^
zA8ucSMv5cn#K?`})WjEwL`A^e!+(2izFNnL!kR1nOKfp(H4+G={uOHwlWWiBU=e{)
zl{=-_o{eXdyZBO@%R)O#IrEg~DGZz;qb6piGKf~SF?WuRvrhJ0c6MPGJT}qykpT&q
zJgwgQPWzD>O71H8u<v$Y(Gek*WGN~4(nmBtWY2fMQT|;~x?>KkXiInfa<{Rt8`zK5
zZ{(1zI19Y-9hka$!z8z!&MFN3$J+4-a={jdoIB${6G;Y9*($2kemoJeUpKOgk^s_@
zEbN*kZAkvoexkPL(ilA*`%0tv4Ck2m8S&Z5wn5wIza+jF<P)H;#I)N_c4Mv#RY;%>
zmUKyp4DPp|YVGnn%-|&lmhX<sn;QZ&oy$r4=|Y{P$lBauT;LtO)vyvsHR3dt-Az2=
z&$LMi-H{9}Kyr7MRDu5FXB!9-JuxYRn*$2em2q@mm`D%X&m|h-GR*&+lH{b>Ny1p|
z=W`!%N$HUDoSZu!n46iBU0}$5p^cYQPsTa2TlpeP<~EVnXY3cN9p{bEItW;U?O#Zl
zjq}WYspd)>Trmua+DHN&8anUlG5kn5qg0+5$jdvK|6i`T%q_9RqrqE^ycy*OFPTk$
zpRiv^qQqHqKzAZhj46^PU9m7^8nj<cLPSzs=P@OZb-J$c*IH>>&aRZ|r*Ye_(ni~_
zcQd5)7-_u#CX!5uN8f%UzdpyQsk4b<P^)Ak8Zt<|@SEL>g0qF1>~mnb+s7j}O3!|)
z8p(%VKH`#Afz<&f{e{S+LL$zQaO1bT{_4aJlHv`PdPi91`Q7$AU8fdBOEj(+`l*Q#
z%U~imYp%Ap-z^w-ra?GGMRq%|JpR&|(SEPm=<;mwW&Cwun%QS}L2CEf@3(Kl2`ddN
z$xm1e-Ty*O4ti#C2e%?m*dOHTth{!)^O!5ZR~0Z%|LqSOPC>}Epo$uf<DT#CnSb0{
zeeTTIc{;vy%tTD&A9e2x)?UVD2UcWuiY!H3s{Y5--DH@Uc)1Je2z6>SbaFvqOXJ?|
zKWQj4(HrV{m6*DV3#aT)>n#xCP||cN0H-F1h$J3Sf1$6YLy9JDxU@gZPD)-Wt-_20
zi{C`lZ}uimPPya|3T=O01KSf`Qz8k69as@UvJ7aNg4L8r5|(}Yi;m}cQcgXo_64Ot
zZ4L@$Qj|~{IS8D#u^8!8oVTL=Wx-0VO+OAeu)=xWXCZ`<+FNAqui9J@6Lg(ZDp8$m
zE`*qoM77Nd3Lh!-QW%mdG5ru)eGrGPUMZtl8_3dMcdK?Ph%T3mLuTIxVpEf@YNK2s
zC=oZzx_^_Hww@etJSH6=&F&8S+Zq<9Nzf(ywj5Y8UN>#Fzgv{?<l@3Pds+hxnIw`E
zVYEEyw2FzyDTtjzTi8nb`!@9Jr+Y)h-wrttq1Bx)&42A5@@voDFAvl<@gRtpQ6kCh
z9~-l?ft)@RYTLpiG8f!|WB=5CM7Q>XAy0$mz)E3p4g!%_UD&jUhxX6i+jO9cVxIK|
z(;(&OAny8KYPC`|dH@cQ`;M6~`Q4DC)v1r+YYWOxL)TJcFhx=LuWh@np1gZtQhuko
z5ZmYa_HT7Ykr>E%E{u~ECJqk!_xkD*g_^Kb0s}PkkW*h0DKh=6{YMd#l~*~BiEl&)
z$}@4s{<Gl=V-!;)(iE(dQzLYk9wjx#jQv-&BTIbjhPa$viW+Y<K^tPX+|T?g+2enA
zeNQ8S;y8fCZfrcEJg}DTdV8?y;r<c)q=e#aarj7==hPQD4Cg!u@?QJ62Lo{s%;Ty`
zr(A8$>VT4Z+dlrmT#2Q}sCtPPYPD7`WXw6uG4croOW7w6YG9gGW$KrG;uEuo+t2vi
z@UT-9lR@|@0W6>NV56mYh3fK306FR<qf%xb-3Nd=-9EYMr4LsAHzgEEzB{&*Q=fR0
zM@JYV69Ye`ekpvDa`ha{ND9g_rtNX+wd=9oLPZ-3YM)x;)8^K;B`u{l7)G|R6)_w3
zY2Dl?>QG+wyB%1KCEWpKQ_5Kl46sjcJ%ZI~o_NXdL}Fm4KFQ3cTUU+RXXM%~?7~Tf
zTlJiB;_g;fbH+Y1XL58q0fx%Y;sA=ENQ_HHHNX2=)mNg<L-am~=nPCGMfaN^tj*YG
zcjK0uPZ~ZN45ui@EfeKg`<xnsJcA|m3Nw~qsZ$>qmWQs$+G%@HV!*<#VSJt_<hHHL
zyU<G7i>vFYjmZ0Y<x_dasYy<^KLI5iWMkUr)(awf0h&SxrZ*V#7>pM=q3rY8Pcuc3
z(qsVA6w2hN>;W&SHz_R>m!(Q`?Z9FdWwKjzXXR31`}~G!841Q3rpo4^Q&Z<?c&I_^
z<P_HF9PJCb`b=MPi1m@QryFf#;aOTw(hPQU<**J7!zKH|2SbE=>1&)P9Wyte2(vG0
z{Rt_#%=dA$P!m%u98eODvuF#xxL|1j7B$=Bi?CEaYu3J`-l~;TD77SE?ZhMAYC>?-
z%5#1P17%;@{U)Qel5(NCacW{_LNH*o++E9W@-9&&V*Il1m#KbNE-#JbkW-V`zIN04
z&8zIo6NONtPVaCE8sEaMJ4o~kN{|`Qz9MG^$Cl;JbD@}=1EdQ$WMA3APK1~Nuj9ay
zQ5_v0pCfnOSGE1O3Js)FinoB_(A)THFRk%Jo(9H~dml#uMn#4Zr~V3h#*DR)NX=3B
z;)KIsQUir)DIyMRv6r>65aHBEY9GPN09f|xmeFC7`Po-L7^;!W-~LLk#~bHH?XsK_
z<;|XP3i*?}M|bMRt(RvdWDOq^3ON$aj5J=IwO7>4O93R0m$3ZkuTzu8vum!r#$LH(
zfhyoIhI-bi$uCYRlJ+UPyuA>l3-4L6i4iVIjgD4aXXwnFeAHf*w5vCTd)}myRshJ=
zMx)5>wO6zv67TSYQ_LQ18XdA%FZqr%gqe&3#PU{?N3{?*-I2(nS`#3Z8YawO`hXQ`
zW|G6m%pMF%D3|QDUHxJ*IUORMTWXF<On9UXoUzxnY33-CUg7EagWh0-ABOFCKL_sX
z+md628k$4%fldQh9FjdU9@~{UJsT&d$Y?BM1b{?%0te(CIs_MbReepG0<pl4&+l_!
zddpCs#1M9MYryF(9+<?%4lJk}YW>)0a#4?S)RsM1693vDwEf7p3^12vW^8n7%9giv
zxPE%nyTY<-uC;5rMUi-~eIb_?(Y972@k+~)i*kDYuwC2w2ZJm%0WqTvERkxa7grRo
zXan!b^u%5EOa`z#5=KUKft&DYWo}$z<aGHnrR??f5r%VJ`wsJ<;FNO<2OU^8S%<IA
zjJ=@^Y}qx}IfdHyl2A0Um+XzLAJ`$t_Nmjuf$1Fs#BxaI{WUcOnofy>VMh-CNNVDR
zlF((t<9N0=wJ%;jH|P|q!3FfmC@tpq+1IuP%dfuLDI{}>9~F~;82ojMUjxf=jIX$x
zV@^%H3|$t|2_zTW#MkG`xSXfQefRBUq0js<y!xsRD|ClKr;b6mpeA0nPbLn!PKqJB
zzF<&987o=6v)*7~3Ssavh^(#+6xwQjtDH9x1jrY-XT_=S45ZEU+PoLLY)rPMt<<Le
z6nRM~N-;GwIXEW_4Yi@IZ6hLa#0a4(;|?UbLW~7u!)J>mDL801w6Sv9XA(>GO0qjO
ziF<3v>Bfn2WAz`|1SvdoVWs<X##K<Svv@W69c*1raF5*heW#GfwQBn$1Z(}`f-nOq
zhX7L(;(wr>d+p6jR#U7)>Q&|YN=+<)7;Vf8^M~gT+D+B5I!rjqm9?j0U^qId*Fc1p
zW21B9c5{}|yHAZG_Bdr}gl`<NTVC)ceM5K3A;_>?9Bod0;gU?cqMoPh);58mr3ufb
z+UnG3P`7ZED~_JFp|Sc<Fzr$Y5N}I6lmyYewy~RJ*_)c8K$b13BaS=u^&4CmILeh9
z>Md<+a7&F-h?Q2{kQ|Q1wa5^3+%`R!o>7o{HbQvJFt`f{$r+n14z+SjJ8iX=_r2@h
z%WZQvKyke~$@O8lHK(SI3?-{?d5StUnQbYIPBvDAK*xaG_j0Kc&uA4x3PUb$(e*dC
zHO^x#DJ`8+wUDAs1b)BWo`as&{j^S@(7Ov{2@JSkcXVCUB}WPz6p|ldpmD1C|E)g7
z8XXXa?X5M0hFiK+Tw7O@c0L5oz4LpVR;dv~G=VJ7*xR}}BaVp=K9S@Iewv!1=y*KC
zy?;Za-kT$$Tl5-GHn|1s02}Y^OV(T$my?Idfk`x?gbKm#tQO4}50mP^Vot(qlkIwr
zjUWE$_Kt>Q%tQEZg`Ok=ut9<SPJR4RUh*uJu2C)&dS{A=x`Z8nL)1`lRs#(t?*QD5
zyjI1uYMHfbGKB>PAJ0;e#SIVku4Lc{?ROr9X-I85?A_f!<))R{nHo%~SuTcK>PhQ$
z?eK2ai!ovVD-_U`Zo+#do|WhHK(!9v^uXV<4lFPiZ4BJ6ymPoj7X}5o7MBhUKKT5L
z=+(lmA@WmA_lI@F$?*{Q5++GsQB+tsse~qSsFAuMQmy0DnyZ9Q80u~7cU|l}p<`D+
z;jKnWrW7BQz39pL)3!CI=kiy+?)A=dYy<A<jT80p`NwUzduQa3)Yiz&<%>1NZbw%Y
zQ!vu4*W%!poX?uP;_76}m&TW0#IP~nG2s{mLMyFjHZ#K1XgBOp?qzn0ft*5!C+)7*
z?dR^VmD%mUVzp9Zs;H#kyRb`ctJb>hSAMS$cn7BK1+zCuHd8DwV~yf%ylJ&l5Hql5
z{pQuTZQi_|o4UqZcUBFcpmd5I&D62SC6_i)Fp0|wn;=TlfyF47D9AuUu2rD^7%Gm*
zs+V%EQ2{x{D@G0PET>);Te5%JRE@8NP^PO9*;t$!v83Fa&Q)#iX<x)}F14c$R*Bu>
zmk<Y?`r3)IM=z4mvo_r(6~7s*7o2$~9au7goHuX+Qa#ONmqsQBokx9`6AvrmZ2epW
zL#G6p8+;L_ls5b1R$z0j9ojR@;&Rk=V2KRD(j{S3*X1&*bSc}bUb{n1r}8IGvwHYF
zIrYiBDOGLgTJG)^0~hKbAO{>+ItB%?%`6S#iA6{{qGI>t_IS;`_ulC|$1(?z3nbld
zI|`;HWyz!&@CFkfrkg+T;<TOZx9GRHQy%BEC%nNJbr?-fFI=>HTRR-|2@%t|>cC<?
z_zhG(5h~6okcU)3_Kj6f?~yyCEyx*HG6YEk)K|ZWEvNoT_RM|t8znP|qzRDdbdB8q
z9Fg))eX>-fqC_?O-old>v!<Lzn*JkBO`au@-ia)!?O@%24Z$gi_!6x)Yu@1+Av}=8
z_whGEfxOkot|l2aI4fC??0xkczNIrFsE@vT-l?fcIT>ER;{Jud690+Fu={h+8|G%G
z=4Lm~;@Cazz|@;fYuDan53~~%v*WM?LI7gYCT_9!FMbsKIuJ5^&LlN4lZkB{M5xlf
zxy^1IOF^q0m_D?y>)y-kTiQ&Am@e(SHkyVQW;drkR|orw;-H>k*xy>P1CVQ-8;{`x
z04%mDp{oov{|2Vow{=sYV^Gs00p$>HOX5m>Zd=lhlfIio;r8v_5ar5}teReS)~V4h
zUf8vn#MK{Yi@6S0TBlTaq`L~#{GfeD!P2ujbY0$H(q&<TfS)^L-`S>nws%h`PXW}h
zb&*(R<@m1r_Us%<5xok;gJQ@1V6{$hCbt=|?CB<wjD4u>c`gi}BU^6Ify4*2|Hw@S
zKY7YN+{OcMG47NCg(UyUN^(pnrf{5mwvV(K*DhilpkKV_pi`4G%84=68%8>{@6J~b
zNCKX5N<~-mx=<PVLP)po>1yEq_#DRP+NRV9X|7ti@=E*OoG$8*J$L``(3As+zM>~?
zA}uSeh`d;D`@Z}<(L>wWc9d%DWGm+-xaHdScW+o^VVAh)IrTZFn0^H}+DF?2Md~Dx
zV7c8Wgiq6X)TxhA%Agv4lKhnvanL@NYeT>UhI<OxEELe`f~iC3h&@;}?3&QfN+byl
zs8gfGB`FvEc&IJgO%Nle6oWemosdv~C!al>Emn_APIAR3CC%je%^Nq_BT2|4i{ba>
z6muRqHFOAurI($lBN$f09&L1QMw(BjSktZxtd{btv9;_6O5|ZHLqPMwom^8bi!KtD
zf`RAf9&p-<r1pk!$YBfqVA7k{yw`csvHs3i-)XxVPA@0@ABEf^a_C?cmEiZWB{F=b
zfU6{^MI4h+pjA$ha<0V=ykNU)DCv5b+Up2nqFynO)YmG7KRLUWKsaH(?dfWy-BpG+
z`RH-IqEk~n#Jmp~*-yOY-gY3MY?J6jxwb;&4S`CwYk&8(meZ5rs89Ja=7DL<ySWs?
zPvVPkLeb@yfyaDB_ulT$IlHSj=scE(Z~B!UNl}RI>%5Sv73R+lR9A($y1$Oac3?GR
zF(j$HPjBCY-30DYjJkYi9GI9+tZY*#AP!}VxO9lXWbR#=abVFCE((x|m1RRhDUP{|
zC$_`ggq&d_B`eDOkKxOy$+L8JWLg*8KHkbVk8-v9tW%EuN;R%XJ!!XsyxEa92aqrM
z^kk5Bdf_CUk$IF>v0Ny^!~Fm+DSM*O;su$G>ekH1;aRDVF-cP(qAy0XOkGQi)JMCw
zoGVraAJ;`^O{ZSCuPj}%C><+~8nx*}q^WTrvG&?F4X?u<$pk4Ct@spu9xrq|-ll$Z
z0%A88#|bdei!~D0?a4$Al#|Y53^}(Wx-+zIC)y^vVQ!mKh&jz8T_ck&og>tqYS_Wr
zBv%rqc9faaB&kGOE~yQH;puk%5SVw7|DJXr*+@5Vp5zLZ3412#5^HwKdCs+7wR($0
zij#?5q#+O{QK}6rZ&9ReJbzAg{gL@S&Oj#Q!2C(?YL={i;DDW8GCI*_Ie!F&c4}(t
z%IiGnxt*z16uXYO6f72#f-e(`fS}>6KJN-o#=972Hswh>Tff&bwX-ermP-t~->Hv9
zjwr=qta8EAX(9KD7|_S;T<RVBq+G^j6yHPyGY1%*-jDCF!YHsy+xa$;GjpMgrW^~J
zgHUWC`bZ^Qw+n3oA<A0HR2wUdmYVcO$>1twSVrN0vAsF&$eDG@`E?Y%a9Hz_{V#RX
zNM;CacJ!<bhSU${!ms0F@k8xh1CtzSGfp`bI=ZfG8wm8nx&B*6BY71ip)-z!=riZL
zZ?M?(`wZUzrUoy&?;V$aHP=t)DJ7>m`R$Qwp1;U<5OBqlP#;qww>O#2<(N%9PD%^)
zYIZ+#PHXXm14}wLfOglO>!wp1GY+Brt+$#K(DFcu6|^5|Zw=NtY8+S!6l+HBLFM#(
zLzQCbD^+eFwp2va$6PB%g#3Dy|M5q&W2nHyjZ%sh5*m{0Jz_uBeuWJo^mxovgNd3b
zJ&+tBf4n1V!P%j1z-bY^pe9e`QkM54POIFaGgI))efx>lXLtuR2XVAVAS%DY%;eN(
zj}UDXUZSAlpG;D;4e5hNaV2AHar>!iDrKj91IZTQz=G1U3c-k`FKn0gg%3*D`|0*e
zsBb0p@di7<&e0tiWYe>s>5#p<I4n`#smYbtNKoF4{cNLLc%7pRgnI;%$4v6QA+ZEz
z+s}1~l<w=g15(P0^G?kYMyVY@atYKvJ8M7R@LfhpNrnK55nGJ*S~7C`g%@-pDF)0h
z_}OYi73J+LY8dV__KWQ!IQMh)){p~8O~4Jx=QuGmX}?sjEjv3CR*r~HjnICOqn9LM
zzuY##%^PlC>y&cmqko8P!iavQ-n5UC={$>bG7b)yU#CV3O-#~G`_&qztYGqP;VK!<
zGPF+28>zs&)h{8VfjLsoU4U-<TI<KN51^c<l!w*)Gx(KgyNbS?TOjXhwtL+bSG?oi
z`<zm>%MeQJTTBt`fc-|-ACdUny1_H@y{Ay>QXeBtSM8AJ>^JK;-8tzzYLeoL?`iw3
zZVi>2NBoFfg~ZSBGB!|Gk8=~8I$OQu1BmB=`91dA^)VUdVZd3*a}KP|);@MVGPX_F
z?^LrKuk;)H99WDnq)<MmOjsN{82Wy9v9;tiWb_l>U`?@piFUtN?R4B|BVinIV1bkp
zb-J{c;6JJ}w&*_l{q`IBAYT;{COEM8P350OzoU~0HnM*62lX5A+G5RwAq5M6(!1u+
zrf~rLVb1dG_5tTH=U@o8yMal$7yMD9MOzSEX9k^e91!wQw?A&njS@<fUwofFbit{K
z<qW?7YS6mapX3^n<)b$-;gn(`Q-LKzLV6V+u|KWfN<5ZQ412l*)A~=+iNRU>v-WnW
z0+Q491`D$rU&M}Oe_r26Ym@Z-8K}71srg@2AbGG*l-OU?7t*sml`z19PBzEe9Iccd
zL5dQo?Ju*ZED~abQ_^(Z%5~5~_E+6Y({_c1(t#xh0DU68Fd{vCo&BHO$%7+!d-0qT
z6<OC6qokabu!i>6-TTf-+*sh|N^h`?rIA+QxcyBHHg&d2v>`7RVXm0ZeNKIzH(SSa
zDTMuPD-*;q0M9_Vdhi$|FI-n^uz%Mkm~EBr5)yF~20~3Cz<zB@B(#0@_xWCV;VCJI
zQ?9*~Layu}Df#_F`!zNTNuXr7bYRgsDX5H>O~m@g#Nf?S)6S#tL|P;^@IU1=F6_E~
z$m@9$&58YUy<WqYYZnnQ4lIAU9w$8emu|{r4{e-=&O|BNE;{u|n4gpCU(p9=>|fht
z#PL;p3A%y{-fF}$WaNXKqhz}Lx9p)}wVdZ7fx4eZ`{=m+dz&K3l3HS%pv6Hw<<tnX
z5Ot>~>_58s(uR=aa@c{zeAVG5ML#u<&tJ6vZ0jqha`IwPj?EA$R{4m;#}vDuQRvYA
ztLhj1D_@Xo!Ksfg$&z&`Zw86KN9@1rmq|g>LW)g2v>x)(rCx(r*T>tz?4iboD2hh*
z7im6l4UGY5-n34%N-p9Cj@ljeaSz2CMsV(!bqcwLcZ|-?+Q&bX^QoD{sU05lm;=)X
z7k1s$8?aBP4@$nz*UN;OUXkn@XfytFPODsM_z8vab3KfGV!;sKQL{u269<;N2wwov
z)laH62qJtw{yMOjX?00~FbV|L*&r?8GcZ}7+y*zr#o;@gVo|9KG9y}?S;%x)r*Q86
zl&TMNGu_R*+?lz=Sf@Vc;}!++f9gZU`V-rr)C>+FcNAliVKYTg-Q)9VZOlv>Iv4TE
z4lJi``Bf|J)9a^8rS~*?x0;OdQFy>bCg>1S+Go^D6_tC6-kzC1=)m&Cnd;9@Pmc9*
zV))E@by_v?sgvOkiR($HCb_Cr)5GBuVxRR;a6F?km?U`)Aums6osn{5!*QN9?eg{u
ziQsNNyX#5>3u7-i@4#|=%dWf1KBpCyUwe&HsEwIdnGqmoFKRG$i*!?R^aqwI3pgY7
z`CLR85_|{!OT&d!lrOGdhQ~3pj&Gnhm@W@3wXA(^>m#^Qsc0OSq_@ZdBHtwYybeS5
z25zL;RxFk{p|eiCRD3s+cHt#$NE5wjrw~I*;-DzE2ki6P2SYb3-RfBfmb;Ar%HEWH
zLBnxeqvUT$S^XK2nh2k)zI|aWp2QR7(bPdFWGBOF)d+uOV~j)=q2~Lde90X7=+YJq
zAa`77z9r)vurIEG#-$t}l(K|k(Pp-Uw-=mx4RngDyuYLyN8XKbWZ|>}6Kgyw>i?xn
z5Da0B?U`ovBTkK4ZJy)^{<55|%_EQ*r%?Mg49MV?tmASg)DX|xm)9$GHfrUS6;X`M
z5vN8Ih-CnCP~X0y`bJ*=XX`%)rmx~%IA~wl<}Wy#+>&AxPdPP0PomP{?!T%HG16E>
zsGI}SW(V6(AhW(STac!@w;j*7&{YA`W}E2C*o;FRs)m<6^!$qf;SPZC!I?}H^GLL%
zB4|S%eRPERmKaYS)$ZGIsXz$&z=}|%Lr>@X7qfKMF00?@z}2Y@_u|0v3-As~FWt!|
z_VR{hYlraG9dgPV`V4pcOxP<Pdj8rRX52K#JVg>H=@(#**=jI0@?a8ort?;@VR8o)
zD|FiVby7sG{aI(PY+Z;GyYz<PGfGVkh_{6+VVwMnAxxLIbqaNUhC(HmNKGKG`XTI6
zTJ2X=>pTE6eQub1otkQ`;25FIlXgXY`+suA<>Ttq=Wl}L*CF+ky}Ev*^HMoIlxt>5
zgsfAaTS^KkCR--(*Az@5m&Ccp=8rhAK%!l6uV?JF*#gZo>3C2dh-m{BGZX41=3JNZ
zq<F)g=ID4`J7s$Bk;m(x1JKUAX8Ce^eVa89Nz4f1kp4m)s0q@ZH*oRSuIzezj7^^R
zDFf3HJH(w?x--SD>b|A46a66v5ICj40^}ghHATCfncr_$H$p~4z}@6yZ78O+aU9mm
z>W9VYfiYsEU<LezHHs$G%<@uO-f(G+d;uxdVfsJw{9IEb5!%b?T{;(F)zMQh5c0>9
z8JxU+c5T({k7(wwjRui9;?(5vH_#)fls#-K5`v8xcSN@B4hItJjNz?JbWhmIwu)Tc
zLe7}JJoJF+U2CSX%TCF?)~?G5f)0$2qVc4dyR*~9Xj3Up=M5dIDIzJ{p6H-1LW(Pp
zoO<onNfN3}+Z)>|20e>lK?f#!cl+8+TkLDPxq*nW%5Z)TEO~;ByaG>hB$78(GfB%#
z7kaLz9GLLx#+%pL*S5vI8J*^gQ!Hecr=a}x&)U}&Op`2Pu$#jTSZ;FhH>e3v|FR9j
z2NW85(7wJ+I6)YZ_|bvXvg!JW%uNs|W2+kVg_>P*N_lL)xi^6a!u9oW9awTuB+0D<
zi&t@gr05hv8O|5Bda)*~x17y%Sdrt@=(wFEOnZcdoUk=H2P^O#=25e{2|J{%t<LED
z*z1(4(+X26moS;>?S?i{6-*`m8dzh4uCp6k1LlT|GhA_X)E8m$wUx&5u&sNje99*W
z?n<>exCk(oK?;g9pvVPoP+Q-o9J(Q1R{RPQ5-$=*aLK7J{Dn(QwYI%^$$ZG|1~VMX
zbbjF(rzSClO-$0j-rTI+)Hdba_}w|BcwGr&OyV)SxnKvLQ;bCDT^(3rd%$$EmyXJA
zX>Wieq@wz`H(00&Ag0G4@wc`&#D3Ahl+F`A(5Y!h30Ou?(8p<ED#e&>XcK<p*ql?y
z^J(okfj=9&mD359c9A?~s7Wdup1k2D#j87=t-Ymv6ylektLtYNuw0<2Pqjnbtq$K6
z%EwKuE6c9C)+r<o&y2H|ZF8d9t6udcugBh9qqe13-VlqJ3dz%wLnKb5)gCRW)2c8q
z{Y!}q2XSQLy{%F3{yk(0Mb`iziQEuZ>CKLi1n~BUo|k{jO}98tIXSCIp=ft>sFJtK
zV|sL8F)bma3%yuJxy7}VYj15Gp$797`yL9$nxAlLI>eK3jtp#X%OP)Ez212&+M5^t
z%-A>7tMn;(W$Ws%ObSCybt2THtdIl*u{W{T?d?@B9lR9(`U_S=P~xw*di-1Y%^|uc
zUsZf%LeM;IcQ%yF_0~Lk2MbGFYQNsmrt20ZX`~bb6l%&D>%_tyduJQ8B3KwSeyrYL
z=(i#1;x5kFyV_|vI7VW&8K;~mjaa5TDs)$>xRnaNwRhKhab1BglPYSD0}J6u;c%#Q
zpn<AU7=qLa=}33LdWnIn=5*aT1!3UkSu~ZLb^~n<bPZSd=K@K*Lc)+*RLWD$+EE*<
zn$SElGI~J$NsbL_5{Yok?b%TEDsfgbAm`f1%^L5_-A1b-Bag`^k`*CqYnwaHX|eox
zoI0>1BXwvhjUxOFR>iFywg@N64jC)gYEo=Xt7r7ohTB)eedQpRV?OS{k~-fE`!kY5
zVe0W;cS<?{h>N@0Mq9I+CUIJ=)k<kXI2blkRy?+e-Bo={#TqLUT2dn3otj!_vYs-F
zQe>36ndif{t?Kno6#X0$a-8T6P^UhPjayl<ov;IAZGu8V7MjXVBk_=$+V_Z&agFA9
z_wh>AN4Lbpb9#%>cj|LdFwV$>6BDCN)NiB#OYcI3IS4I8cMm!Bnxk=D(>a;DyGNc$
z9%J27YLIGCgj!idrrJbq9-DLu-4>GnlfoL_lWU+${9ukL<$%#}<5@jz)A@UDAwiQ{
z1)^j7TWzL$^SZT8K}dPq#OQm7eV=V3!w@b(UUzRan9e3`Rr%!1<p}P)?>?``Rhw_M
z?cKbk<GcQf;g@BQMRV%)t<}>6Ys}i+`Hl^n*E)}yob7~K+C5ztBKZQG@L>lg?mM)%
zx}I%EwP_({Bp7gDg)w_W0)dTU2b^YqYtf#xo!tT{PZ>2BuY{L%7nW0>O0<n~i5{|h
zvrk+~dX4j}j`t@f#1ejEef<!63PRSQGzT46emyWX<Ytyh-YRkG*?sLuyLA%BOsA;z
zlq<~;awa_W4tsBYgf;4o4>+Y*@kDB^_(}Vwf{BnS@IZ-Ufdv9-LutTar9?|`#=N?u
z)-B3bq*7U+AX-j+d?8iCq20IlwOz7AZ<U0In#<@hUr?Szy+L_E7uBUD>S}-3{cT5a
z0|Pu+pB$`RU*r}TqoMy9R*9RCJy7-OohNZvJc@L>5rOTkzBy<pc}tRg@Db+j{fSFA
zp<m)L>C|{{#J;%|Hgl3v5SA_MS~Ftb(khlOyUHmP>L^2xc&B6#73HIw>b|vRO8fal
zAk86fFgfl^EE7xdZSA1R=xwR7c!TtOnG+z;&1qGnI+dy|@z62*_Qq5qR-tf^`$q@b
z*qg?IQ-0<jXun6GmOOztOrIZr?tZ5xvFc`ik5A1ADyi?tL9a!{<2*}l*$NJSXNNXn
zd6$DUi>lk^s1dFQ&-aXd*AkRVUx<_5U@7ItU;AKwW^8QxxbqwX)B*?R0s7@G_)zP5
zoKnf`CK0A4mSCwCVacVY_Tha0)b!kt^W5z<y<hh~R6x5tw(KL-XEDiQg)efR6fAt#
z5Jj2XA^Yx}j*V+qJCD_6xgaE#F|(*DzNgXExd{?%V#Nm#^P{_{F4^~XsFCR}Ps|^2
zVCB`YxIJ<crU#QwY%;9YFT=tgC)gPWrgyKJAVO=}zCS4vac*)Mn|8|8Njzj@;Y+t<
zAMIu&x!k;-M?vVpPCE5T+cuFiAA85gs(tn<NlcLrNy|=sX2rxtquivS9G$TT+gyxJ
z$W^rO0J@ddDV^yLWj%VRtwHD)QisxD=@JlgVGlQqfU^m=%`gWjs+pSH$8!^-_DH=h
zsZVIt9auKFeRRwo?bhJ{{FXHQ=e)tRiAC#=;WY3A-Ar}ipJ2#Y=2N!%Jx+a*-E-og
z;15sO54PDuZ+hU)w>xDqhwI0Cc=q*eSE44dW_{<m&{)R19Ox3}#~xe!d|1Ziw5V+$
zwY1_usE^N!w#%wXrU~0!FcwR(gwVz_4lEb+CT{qEyV=v;8*H%zZBggW|H7%S_v)(X
z$c|upJ8D%a=#VlS10udB={~LMmrBkTN+H|V4m0Qx8lA3q10h*EZGY>_lGuzT{;l5X
zGlE+_NFgIVhpK_&c%L0;U!2E~Lyr_j1S}=Ob#voWBX+Qwgr1C_5j*4!rq!1Fl$7I#
z+L1~Yn7&hrq2Ra#T>#%nUURtNU)+nO{h*s5B-c=*DSq}r$oqqc*^k$-lIlsaNTFQP
zF*-DF_31G!Gr4^Hvz7R{Bi*#5&OtYeW**l_)8W)78;U*wd6Yn-C#r8DQ_)sA)k2T!
ztw#OT<uYB~K|5Ms3Gc>07MRiJy2R&BeawL*A01nwm&XbbTKKV?Po?V++m>lDNw(p%
z3e=WdmRHg$`s6qqy;k9Z9dCyh5%FVvM4TZ9q$!jmIu?j&dop**s@t}#bDnh;ZJ3&|
z6DeljHs(Bu{w?gfnf#shRDE#?msZkj%!pH;t^4Dn!}fHR*4~3TVa6#XJB;l}zLfS%
zO=j@43~vv6iRr`1nVJ}ww6&7Ik0FR@H|o@eeYBG`Jz~+NEK(9~O}w6^9FYKb)M=F@
z6MGB33tTq9?Ts^bsv+*=PN`}s7Qqx(pkK<h7_+IJZZ##?hC6Dwpi&b{rfcAplMEG}
zohcaol<y75e#jSLCvb_2!_Urk%dK0<^}Qz@SQ=jN`@r`y^_-&^w9h@1F{cbJ^@6bb
zFk!N91QR1xOZxCkkE9qzev!ionR<!V-_EzSSts9KN<~C)R}$8dp6Wt3EU`hcYrVl(
zU#M@`Y3*XZiOf8@1C>&)tNf5j*=&~@CDhf+9M6<fy_6%{8hMNChw>$IOP=mIrRa`O
znXJ5Qf`j(MZJ|a`xz0GnFdZygY4iEGdc!8?QAf_!={e+=XS+dZEmKVjEO06GZ>)jr
zuqE7viDDu9F^J8wKjf5>ienobfFXG<H^Q4|oF}oWS6{j88vBv_{?`70^H{t0K)Br;
zL__j?gF$f=y?+tf1x)?qt|-zt{Af2LBn+IMo{KB49cayg=<hzL>`D8vwwxG7lsG&>
zm4g^kQww;QeWQc^ar^PsIu@T%N^L5mm{XJEB8#Nbi2Ov3Yx{`vlxOzl)vN6%YclYl
zAi$MU!zF2Er3M32&fo;{Q&pqsM5tca8#r{1P1kCWn59%hXl>U${&f8&H)`gt2pQX{
z$s;;^y@g{Df(DhB=zhe0riS_edWZVY8%&wXbqF5gZ28&xg5Iix7E14w+!i&WOQO5;
zr|jpto}h(}!(M*oN-a-K^dvQOvhT>Nlhk>T1nfNa^KIND7m#WGf|Lo`iB65C@kTCp
z8n$1^+1|jV-U?84%C%PVZIYSnN&CgbC#6|-o<j=g-)+CtmSg!f*E)rwgPpN4wxDw6
zA&=QFx9)O$R=@z=m4Ged$!@1UkZvdOr}iry(NtV-x->5_v;xr?ORHFO)eEhTiu6ji
zXusM95oenerDq*jZLTh9O3akdmlSvQYwb7UVR;+Ybzg+Z0-;myfc<(y&+X8Iy<<+f
zFi#g!;u533>Yx2ay)jQlIjmlEV0lh+;nZMchWX8EB}pb(sHYuRy)!JstS3>Lu@t}6
z2+gV?($zV|+!5=C*V=Dq&vs4`Y^<yWl(keVu3llklPie1e7o};6QSJ}p;n-<B=Wiz
z*?u>PB*M3{2@qlrATUyTiHr7oSz6e2OK)bIQ%Z_D#GTTJjrRN1bsf9NuQ8}_JoZ+z
z)b%Eq7yE+_ca*MD%6M5nFNORr6)g490kXm&H2q-?^X?H8Za9%S<-}MRaVbnt&DbCP
z&oOe|%2O%ge9_ub-+8>S@qz{=(;}us8E%eo$oz3b88B36BJT-BQo#OU$=m1D$1>^2
zlg1(MhJ6|(>z}kfg<3oPUV-IW&2Xj7pH_q9wA5g?1IryvNglzS^}q1x&$^c?XRP+a
z9tW0-xrBAsu`cfw!~^^D!~r6(oX6^IsCVhJ1UyvT_80BNB)-*+D_j-iz@qg*#;ZK?
zpdYPu<H}#QPsVtYMCRX^L8o5pv9PN@Iy^E<#xT-;|5d%MRMhZC817>eQ<JUn?h(Eb
z0cZcG`ok*3Kyfi}U_$2M!9n}$wg+aOeE<%OVyu4>2o&cFhwX0)Che(ioVZ9deFrNZ
zO>T^T+uGr>D-uFhAn#ysdBQ`nr+=59+}7_yCuf{uO{EYvPYxMn_S@gLMP9aYnJ*Sr
zTy6i*O{kpKS{wu39!^aRK-`k<tGF<K!v3*QnTh`0@Ptk|+9Yd<85qUVKh;9y<U{iz
zrui&w{L@aow#~w>XFmqh_s?w!I0uE`$)bu6qb8T2f6vs|B%CM5!oRfdUq8p?BC}4p
z$@DOOL>G|~od}PAt(PW?7t&UA;k;9mgHrYwW|oY4_HVg{t5)CWJmz>nEV<f)jP~zM
z1|SXRcu&_aIUo%IQ@?_m|6|F3D~^}7#n+Ktd3636r@kiX2@X4H_#U_atlu1jFw1;!
zObfOoT#x#;hf~jq?2Co~-1x83S+<bW%qg-M`LcvdbK1t2HT~)Qi4SQvH!|V`rQ^*?
zB0)`2;1kI1!B`Xx3Le|+;iY{_CkwH-$DEoxKG5+I<;6mMT!Y-zC$f5aXomyQPPy{B
z754ED=ZA#EB2bHJh1`79$KX@5l+F+g3zh08JX}mzW@vKC)iO8Kq9ybbA1<=CY{g2a
zP_|cYOqsDyYHuO_;qKlJryOqyw@8M38sDK@`%mtAiczJoSO!q>atalqK7T2qQ9CXc
zz&_>SI)PRVZI^e61IRvb2Oq|!PklI02y(-{$xcZ`n4^35tbN+UWriOjt`gdVAp}q|
zpKIo(M|$J-=?@1hy>5=WdRFo%`;68O$zr%a<dk!-$tBo6v&|MXTme6_MJOHRNpCe$
z215nlPeNIYqUN(|2H^YP0}jaZG(-R*cBGJT>T5&E6Dr~p?6d1PdCV?)<0SPz>C_8d
z2FGOP**c1l@Hr2cHB%;)&>L!o<qRB1&lfoyoK}ILV%=I66e)z@UJ-Mjwing-;$UO7
zv}Wi5Q7uuE^S61buX_n!ob&hWN9XrC&v^ijBDQ~SYseLl{ob5Y*6<c~L0$&A5?WIC
z=hcw)QQeeulGI^~Fd<J!&-Rk`E^b>u&U4C12k4Rn!Kd+1{``g&Guz-BoMPg0@WqjG
zfqB5S+rFT^Y#g_vkr}6~IozzsRa^}RXY31egP}}v9yMGpvxIDdY+saBTYEFkW1`{R
zx+ucFxIUO$Q3_|-_Kwc)c4~53R^vqcC9QD%+AU5YmmE{OGJMH(`b#@l3c1oAbTGEL
zv=6!?;LB<W(~79Sk5Qw9PoYh7>h)D6{U5R~ua}DLkq7^298-ZMXE)W`frs;$eMP-7
zRGsu4vsg4$h7}<V^?Kvx2>tuYeB;8daJVHL5JnAHe3B@oY5S`BB!?DbLx;iIHkc4u
z_D{K$y|jHb93!Q>mEpY!uunPl+IaNmX@s7aJ^cI>rPu!d{Ezcp(cDd|?y#?Jlwfvr
zYSJm?EKX1E=#8N!xvT+h7@g5gL=Hedy>sCKdwE+1My{wglz`q~A_yqss5xq{XvoWY
zOV8Aib#FCc)(^sw9wf)@m2E-TAXAvtxG>v+6;YMrK-%L%DzM92F9-VE8O@dgptYME
z$A)RIs(#9$K<I*|*n#DSVy`OLIxn9W?{r0b7Xm`iSO7s?SZ3HeDX^7HLbmG&1G87R
zAN0qGs2!ek0PzFe6Cicckzk-Y-*|gX_2)xSgb%{we+WYqFl|2~fnY-??6p}W>nQ$K
zqfRk)mQY>F+{dwxzOKFk?70VG7s(Q@e872FwtB|KsKuVk^-<>F;Ct-#Z3ZB%qD&W<
z2l;!y&#8&89fGq?S!=&tSzl8eq;!ENlh>5y9OmtiQ=fZ35`6J3%pzAMF}?3q_uc6{
zHy%Q|Mh=n+N9LJiV0Lx>o?GlteD!do#+3$G{6<Hv{Jb%92%{7T1bsMW%NpsM9KfY%
z#wka8S~4S=E8T=}QO**O$_utUQTN)bS9(3(%F%dDHxH@VlB1}8d)}$ZL-LLdn{Kje
z6Q$_khSg4?BXtthE-sw56&);sosoYp_#%u86KIETiLI=5lJk`WoDMp$L>cVsq4o?+
zi_5ZDvI}<I!_U7O^!R^4C!C+uDoW=S*QePV>NC=pmY*qO?F@HLy?(Q>YjvG!oDpx#
zRUDo~-^g>kQ(C^q_~jWC5nt2Ix+rXh>k8X=zmldn^*M`cZdrSay{T=MjUzp{W>U32
z9ByYf-)LW31Q{|j<W+zmuNttgYu%0wtWzG1hy2G2)v1qeMHo4o8O@k|eQ^$Hj7enB
zU-2Re$s<sXd8d^Q#Ah?9!!x$3-h_sWwMk1Juw0u+AS8i3oY}A_FagiqZ`XGdkWLA5
zwn@+MOv_5W)SHA}^=x&$KR2Zg6VkRevKA)p)<+zDZB6S9Y`EfU?}uY)Fio+Tab13}
zwtA4(An85uuHr7&`(xXmbLwM$_ATs@&XD8ioZZmgjh0B7Hbt=j6K!90%PPCE`-!M^
zMkWEP{UwabZ1Ld`5_|KYt!uwQAs}-P;p`494-`BfNzgNE>l>eO6RwaWPO+9wjB*Ht
z_U3L!MqYC^lJwTnx=~Z?F8*=YSZ?a}<n^nag68?E>#ka1H#b%pt7Yc!a6t8?!(sl1
zOS4<*G{%Sq4Wks})8dHHmv>1n2Y+FCZ>=|oflL>?6Amn~o4i!xwxR2}f)b?q*SUzZ
zrc+bw$&TsXI7uHi=3K9v8}WMF3Y+<s+zt0^cb<|TyE$|Wo9f%R48WZ0Bs=L1mS?Tl
z_T9F*jRDq3+9Tb1EajjA4=}_#^(m3D{f5G4VoUuZ#g~-sbVe!+f_r$J`lX^wsHzeD
z(X8CojaHO%sZyAlnClF|2>NZexA~9oU;>e0+JI@0Burw{&)6Ngf~#&>yV-e8pW`)g
zshn~VzBN~WaC(P~A7f5A=Y9a*#NO5>Y?X{7PBFT05NSkHaCrW-eM9RO2dLb7baN{(
z;r!*Vdbhp3{UigTA-ww>Sh~@?b=2<cNXnTIkb-*+VJY1*z!LO3w2Ju>=13zgf*xr)
z>>V}J%rX2@sKvoD-Tf!Ov9nJ7WppK0O_VeOZz9*OC@uW3I?TSq&(7XiU&-}KWlD;k
z)Z_}oM<kd3uKI3j6vYqgykFEqP_JEe1MbHYgaX*R>)l0YWrG8IE)r!$jo7M9Vh*jB
zQ?rGzC+AtS*qa@*fexSQ@<ZXC^pAXV56<M&CvL<gWJ-*5)&|?mhbmE{-S5C+H$tpL
z8Tg9;O2!qmCoJiH8)~!wS3dB;yYkKbaV5R&ci#~~W(DAEfF;J<g3q&UZKZV^ZeHV*
zG=$OqT{henYB2*+zJO25Z-V+9LJUi|b=*eUg0NsOjD?~mTEk9-xWi_#(Sm8CmG0Al
zC2^aX=xt*++*RnXgkc7cJ6&la6%dG|(>9PUy@^xoNUi?1`VvGf8C~ww1D3R9<4m91
zx5io%Iw<Mr06Odj)4WIO==r@iUhNcyty>@tE<Ejvuyay=h>y03Y9|awKdZkEEU|J^
ze;B!xT)mTRxLfY-k8STe<!qtbMfz-To2o&<)gp6MOdGW9ai=Ejo+CSkaA53B56DsS
zJ-MFjA40X9l5lll*OpD|ZMt67bTlbgf*9UvlF3X<To^HJ=Hcg+89hR*pU80Q%N<~0
zukNGJfqKYh+x&sIlKbT)++dRUb(sKHR@q#)NJ`@q3Run9{7I)K--=@f@*?YS(Y812
zylG<c9;cM@ipJ0Tq=w9dgLk)+W!>b~t)zT#07<gAJNTaV$&TU9BkhP&Qv(sVDnaeA
z?Woq$lSl*9#W{GZ(UBnasIr8mg=c4On=4mb<vi-cgmTW>z4`pox8p*@bM`&v2ZR+9
zroZ-$jmS^;$0xT(KqR2zw!%?JM3}~Mo{ZpLyD$5_uxtGUK1jWA2QC;SGjbx4oa2(c
zH#>0kl~*rwo)fj^CI&{itaZq~sVx~(U8#W_Se`gDc_RT*zqje519_XF^byjDQd4gB
zyCa|=ydR}fXH^zc!M(5hJr)><$NO?GVbP%`KX!AEfZUO(-jLnjrin}((r-bq*koLV
z>eM9ju%?G1o15DnXun2@h`~bEGe9D3*K_tk{NA6)zk}98PC-@ljHqY(<~DRzLoxt@
zi8q)?jf|3FpueTHgNu<!avT#Tu%yW`#ys)E@t!>=pWSb*-{g+X0~T1UezTLEM)HeO
z%f78!>ZLZ;O!p&dO?{#0UM3`6=)Qe%lEhnq){^Mu*-nkHkBeM}>;u&b9D}*dspRkc
zVW%cW5NCCqyBVKSacTA)`OSr0@FR_L1lH|6Y2!n)AxYc5v)(3VT%IWkhRsYr@-sf*
z)YmWNEQs1u_yO<sxP4dm%L|<8;yC4lkV{T5ty5p8IDQPnT!>;HtX13vU$NgQ<zX?=
zpPQZ>w-42q5DM6!qJR={9s^UU(Ybaz*WP@%k;Jiq8K=<gN+lbVT0`-A=j|iiybBw0
z-hj3Kpijoj*}l6QfwC*4GPfN($MMjiK2{IM@eqBnX4>~Onug=jtW(N&4uZ5K8kn)~
z%?{qpm2rKi5FJdda~dQfCyaexHd}w|nhke2g;>!%`b23<hYC?ci)i294!gAjlVd%n
zsAFL>3WlD2wAJa(vPq|?y8HO&GW%HT>s2esO+YC|Bd#uF1B?z2RtL+EkiDaM6c0#E
zVlGx3VjW>O57|SFVvxQ~MmXVuNT?(U)Z>B0=bcu0YLPk79<C1*D;X@kOhTu;)ig`U
zqNd=Hd;!U@u(wmFen^HQz90S8qa7C;8Y#vjJ5%UB7U`GMs`iQONYd>^1PM5z&YZU&
zc=-9uZ!a1gL9v;^4|ptE>O6J&RYF4h!S0Q6xJpmRh-BrBe8Z_%UkQTd${w`;3cTHw
z;?SD6I#0>6U|l4~#$#=9Hjoj)DMe>Ve<R6W_6H`FX=j2-9&WpnH0;31+<8vI4f`U}
zkQaz8xyK+}m-cj>T{Q7SOgJe+;`UZACr%|8!{!qVihMEUOR=|`_*9B;&7A|wUWUU_
zZuwGT>Xfs6ImAuj4nOP^g$Wy{6-2Xt@qP3`V>{@;N~FY3T)y%L+J<2lE8UhZDMm8D
zA)K0=y&EvY&Dg=h5nXKM6mpv}aD|7_C4L3{utW7xt)<SjBlE``STU)0bC2eX9nSI)
z)N)u}UX%mlgqX@Gq$>y>&&3+u8Q-H6(|`gM2{j#ogj3RMqUW<CNv~F{T;V*%%gR(P
z=UIEAtvn~)j8p15uNTOhGX!@#x}-+I$*@d(jJ0uUG;bJg$j~6&{;_=J=DC4!s1l_d
zq3$w=3q9V-xTcR2ZcABn$vsDL(mt8Eg`&3SJZs5NS&!oASsf7JLbJ*aQ%*SLkEbSv
zmCh8Jqf_=&!7_G2ikx!}EM`hnRhfXa2W7`#IK1b)J>BLSijw}<`y5#PCh}E=KY@Cn
zwIF4iJ=5wX$nc)962lkB{S4Pcr(UPUz}yZynUAxhTb)Opt+<FCwNo{&m`rG_T*vcD
z5o!`4M<>{&Y~Rz>H1jct7|4wR9@42vG-1V1?MlS6Gi}0Pwlq~;%%fA2=!?ae>?mj3
zxgqBUXwQrTNp5PeH#dXU9uDYSec>P+lt>>Y8V8nQL)@cGA{7pS89SfjeCMlP<2)wy
zDmARE0-;m43sobhhuGxkzjJR<6WkS_K<9n2O8Z!|{Z1)|iJ%CXc#ceTcBu^$oR>93
zS;8a+o_1=oHLEKZev;<d4<!YB#jBkst<3W4mfH{Ka$GMZ2+s=J-7seJ`82W9dD3!C
zN@?<J_fq)_ij6zyz|>`8gJk_(=t{kbo>PiWqzNu!M(w#m8ttNr!T_BXl}glW;dN^p
zxjV5M{YaZfDZFWV7|OyPM^L4v7G4g1aZ1?pxosA9y{$Lm6k-XqEp+6ZW8VhLZ9iHy
zibav<8LslYnuE)!(bR0%vcZ0=brsB4?wAb1?0o*6`g;8#2&oR!v^X8IAJ3OuPs|OE
z(J0-2ng&Ucm_O0Fx{VMO3bFFaLHmULWc{e|Ueb{w$Ib6`YHD5Kp@AVk+fOa_Q+!wS
zmVM4<NFqhOgimSW$I020ikzQrqzIv5x<BERwf3ur>}RrH6VJXMtr~^+UPks{(*ZAk
zwi|ccfPtKusRE8UHMJrs6iZV5xq_ik551Ev4h5FSSGb=xtATvTguP?{&CNwf@rb;v
z2V)yL(4?lsA?fJ(g|<_5`&@rOnL+{OZ1Wnay&|<MwBn%sV%N0`A!O}Afe9>W0Vy>g
zF~8J}Rq0X_9e~9Pu~l(y>1>g<?G#GyUvA?<ClQt_qzqs}o^`9o>{r_AZl0a=DkyvJ
zwEb%CC?c|EoacDGLSbbdlNc2rv0uyAFYHp@lu|o)upWNDqqCeTGxqCkDnT6aAiE<>
z<r$7Z$=4^G`uIX}chWC~kb|P^H?oJl0b+5SLN$}1o;pa;lVkRqRU??JS&EHGyrL%g
z%WJN_#(t~%q29No+`39rchMh(jcxVfa^zI^7-UH*kMP0ZsQ&G49E%r|pg{@$U7fV8
zz8w`^#4-m4N^$$0S`L*`S$)xgX*upd!uZ`rn697Oxzi~X{)yS;v<lw<ND>nW`(?k^
zT4QM#b4rEP=a3!g04;vMp$_UnqU$L}6OD^&_kwTxgZ7d&)3{4e%1zjt(#bQc%n9~~
zZIQQ*a#P^|zACd0EE^(4Op#2XM_h*eQN3T@L*dT3@Pq>kWE&s1Kdv>($q_jXTt!Bu
zCdo5G%CP-OD{W#=Ii;8dbvi5zoI`1B(((F~{b{kctbFisnee8&(_M(l!qQ4ufYJ<$
z%%4?9@|qeM?J6kHsmU4AtMyZ2%Kkie0^#r1I?qwB8iio-*Z!hl;0Z_>OsE6P9TV;l
z$)CqowZE*^kVLVVAnHB`mLvnUt|Ei}s)qc(<UqctSRLw<BJ@Z+jW=e)lVB}c?ii91
zvQ?FZllo)=HuWpiPm!a4owNMxW5{LuoRXI5n(MB$zv<qtB+*VWh7AA3HXsutHkZH6
z*7x1_dgn2ghq*wk55)wZjU*Uy1F*lVnOIciw==BOh-hKvhD(anXHdbvZ(DZVz$&K@
z9}urBt&DcQp5ZQZeW3kAH~Pc8MoXe`FQ#&#KF=(cdB6Q*jYjvJDNO_x4R0_>XsDrJ
zoKDz3<*(NCw~FpiD!PM_GQJT;RcyikTwfPgDAK4iSmaA7XR(<%^<qqxUw!43_AmAA
zq+8PD>~Uc6X2>AQF9~mA1W93G|5`PnJ>i{V8zbY8x0>X$BX=4#*x&5m>I+GtGvF6k
z(!XWbuC#ye8Yro+x=6myww1TwA0EdpGQfYdQwvq|P$cYiK&682gQb%@cMKyC*nj35
zb;WpGBMyJ*z>+P#Ljk6+rvKIWS58WtyKvA2l1<>B=Hw>%-wC({J;<OE95?{2+sqU`
zIDNB6Y6X~EG?I{KLVrWJX*DsQT6qr6KtubuM`~NWxi^TzgjabL7xB#yCu<*H?-Vf>
z<HemJI-2GWc&pKSbmx$CnfsI>%0A(dU_f4buk)lGGdw$OpZG{_kz9U$G=Io}74epe
zTniHFRJxst;7_XGC@?=1Ve?14!Q?Hd#Ehp<wSV#>DZOidNr|FM#gc%Ec3_l<)Je2T
z#nRTX-f$Q>Kc$;6lxHEXG>2|#gb;JHvtxbx)Ev(pw>wY5ROLIePs?Xks^>hH-goiL
z4OdA9q9Dg0N%rYAT$Pf))SwPi&rXBBwECrz-O^qB8Ev}2WBGY2kStFKS!a$p^@%S-
zVI1Y|`Zj-{f?Q?J6T1N{8_AiYKQ(Kg)ivT~1K|dWqy}T!*|G8hI&7cazFu(#i3!qx
zM?w{+UTZywqV;pyHzp@?TL1|97-Ak{M)tKA6$}Lmi>s6CtT&j(MnWFO`INo5zMD&#
z`T8U0BK_J8r$&9fTRuI^HyHT29qdBrDshx>VCnpgcsDu#+u5I1CX3H&STPG1hok?D
z1F3UKQC5la$p^foYLrP$rY>!XysV8H4StG{&Cl;vS;SuUoXT><tImPI)JK0~bgVpn
zR`vynZUp6z;j7siNa4=Ja5UQ&HW=uF;TQxu?ZCA8r?{&P?(vJ-kVDlg4c0*imS-la
zGchS&Tx~QasTP%21X!{TA%};u1tjPqcI-=DFy^#`B~hV9o8boJP5aW;Ke%5xRN`RZ
zz>;Ahl7Gs+to1KAd9Ffx?tTZR7Ld0|On+ZqEo5gBM+vX*z@m3@QIngOu2B=gk+UUC
zo_)n5CE^q46BiYaIiMseWS*0Lgd2-^S$CbHePw+{a<`^ZI|o>PCTy6h(awy!nEBW#
z`>L*M5-+8pdD<H+9fx8M*h{Njdgv?a<fJ!P`hAFbzhp0~@8wR{8N`XUgvD<4__u_J
zBcFmiwd2MC`|9rZOEZyl*-@SPz%ea^!d3Cxu*=$}oSEG*CMO#Jxjy(4Znu{|Qv8x4
zl4hJ@l8IGg!xG5s6>SQT_9ap&$PrkOJz<;dahXtAN@HP|M0{P`^U7{=Bgtr}lf#y(
zdIVGIqqFQOm{a{dZ<i;@eaooxRENKmfd?@4E8JNA+pnscR8wnuB+y|{aB6Cc$tPSV
z#TmPzT2~l8qtIQuav!25TBAD%gA!ZPUY$Ef=P>BK!ztGrr4#0^6%_risd1tGpija|
zG2Kdh90bm(ubwMAN60ohn^5+=Hs6Lsi0xr^bPVN)cQ9P>vFDMmiJR{0tNRhIb)zm}
zPTs*F$}$ohM@{(pZYHFCf#e{@OH=Z=Q&XEktWmPPptpS0*imQf%G^Jt^+ZE9<A5~O
zg<T|_AT9vq=MKB7A;-Eo-Ni;RkptBbyvw*<-3ZNQtoW$jDaX)LR1JLx6Qm1LmnESb
z-P&^=V@@+4S8N4a-u*BgWOU1;c!jf0O|fVzSJ*Y}dxQ8Nf~}Nem`nG|k~Cj-ZGB2E
z^2Hi>gE6ho>Ba~oCtHzQVioBJCWf3+j6zB;u@#V1IwNwtGDopt&D)&EQZmLBC%RkZ
zy4n`Sl?K0?q$^MJ@H%%|i8$Xv{)qLn{c(FkzH`O$mCmCm>vl|P_QvW&I17mx7XwCG
zi1SWO&dBnsm)Y0U42Y;pONcKzJ?^*>U#wp$0V<X-)o5?(W+)S0qMLJIOIA?`u7r75
zGwo}uW<P7H6bWhjlyKkGKoVMtshqQ~>()qOW2(=zX?12R6u8xk4GjNcM8EWFOC9ys
zw~1aIceP050F)g>VW(3^!dITus~W<t#-nY_DMrJnWZQGOcIaJC+4VK<;<=D54tRr=
z;YbP{TiwkMq>VQ`jaD33VBHH`aimIRL`0?9n#BcL;S>Vf`MCxrmnB=<x|P@6g^nbZ
zRn$m&g)nm?T;FbJ8*YkRk=mpH;?tTy*)~8*{+<lV3@>Xp*4K3KDeEXM1`aGS75QSQ
z7p`^d9_cg`Xy-?#g4O{PpN!H%dMSmltZ$zmNS8bZAk-r~1`K_3n``CJ-sTi*QFIU}
znGtFZY0mAYRufEuj*Cbh%`s0+O^%Xz#GvfvgozGeI+?^7Xvl%(j5A>phh?Eya+KID
z^@U^+nl?muNllWh;W>mg)JV5B60&7-n)f^9+)Z1M`!}?O0oxU{th<H-%bl9&skl)|
zp|T}!%vT}DO~H5%Ipq{Ih!x;kP-MHewD(~%6X`t*x&|(Kt69`JhrtS^@}?Rq^8~Y|
z=#;Rsve4ow_dE4Mkx7gyV1;eY6LaU?&QoeWhDcRAWLp;JR=<S@J?p@-iNe1R*=<So
zpeuXk_c)~(V%#Vdr^yMsy>*7J>vC5qkd}B{JN3y+Yy@lUj%uwl4@Cs*lmdFXSS_bM
zz94r)Wsif1*jrnd)+nOiDdvuaF6&s;J<xi^aeG@^$tk(WFx>#kc1sLBP02HvlgnQz
zs^#Q&zkS0a&tJxL1;f{z+{0?Cy{Pj!!c`*v-f5dVh1HzH+(%O1w^x6l-3SGdFN0T|
zn?LH*B;u@Dd$ZkH&B7xP<PTMa7SO3l#DUFIg#BSSh<DUHg^^9?To72fR)@QYw|b7K
zI0wV%4#O%J0bE1uo%OXPsy@2*3{oC7u~GC(U4yH@jC5VjR4eal(+d~Tf@WYA;mN6q
z*<5mWgvKFJ1~RQ+5d}28yGFAQM<LnM>5Vs-)>G}|+R<x^wRu$TmopAvDRE(M477z`
z-P0wuU9mql`JOo&%nxDzCwLY^l5hi<gsm%AF1MjJt4I~Z7%I%f8|)zcmG<xmhLwI>
zmk5j(6D=KsyxXZMT+B?eWG1q$x&E`G+qyIafC3-mEMr2#onNwr4d-FKu<OQKo#zs`
z=15gHWFy_Lq`AriDpQG46Q9)r;W5ahp0UxcwfuBMb)@>2W|o@VUYjS!Che{^sWaOU
z;88yUNZi8uB{Myugl%g?X#K=i1Vqv?1If;<>!IPDu(6y*jQ`GK(QIW?$ao<Z6|c?l
zu13^c7_al1t?bm)2r}^UiTOP?(LOObEwR)ow_{4Q2V_G#xfr)>lZ9+TeWA<&&gK7|
zB7;`iVXHQ6TJ<)YYM<CRGKt>HDOWe1fgnrWebnC5!IaBN*m=|&OrMhFAGzJ8+ddZ!
z9V1c9tOJT3r!Wu#LYzktGXx^d)T^~6WH>#9xB)DFqirnjPnkNju4NRp*(IA%yn*Ul
zo(1yDIQ6-@gT0xa&9yT*ypNziLV19Be65=sAe?N-wl@ID=;63096-KgxHmU!cQ>N6
zdTi7wh@D*6weBXnr!`wcqEV+5lb1q~#=OUN6fDRx6H1CO2Nolg8BSp#kw^rtPbVPv
z+1bu+)*@d9qDQ$e(-w^KRv%hhek0wHPWeh4X7?ry*gV@ulR4oOb3#LVr=yP#Et$}7
z?7Dp%Vyu^*bzmY28&_?y`*OTQ-;X)ZdD3yG8XA!IW-I(koyQz|i9^RRq<vExJ0F&5
zLmq^{3L`{#<3>0-qcknaYI@G-=8@je<lRmwTGL(e2RgArw#Y0ZPulw)DFrcCoRXK8
z`~#%Oa-fOCFlw0-fUE2qqOV4`F);0vW!8Poz4kyZu5Oq;<P`F!c?kfr-wAtvZCVgA
z$WE~ixcpEPt#wI?7${|B`{oX|CpP~c2u_2E`H&J>m$`pS8(Qgj#Cd8FVOTifPdW8T
zU$B1-K>WV7-m6p`+HuIH4lI}!B6fAEep}An>dDD%qkX5Ov!OTCY1qHLVe1q(lRKrP
zQXwHU{LwM<fo_$#Lv6+>=3dx1H#1`2(RTUV1f1^}Y0Vu_tWx^gh>?*8Z@7eriMH>|
z8D@e{jyZ)yAub1p(fY2o^KPA--a0w9%>l%m>J*mUhO^Q>STNngB+ba<gh>D^)HoI-
zpTHbyG8hQMxqT>ca}X0Fj~YM**V>1(1F|bETzaEZjA;;erBhu;5tG4JKhn)O7d9o#
zfyGx3uuEi(lZ=%7Ld2ojcemkSqMRM!00tmx@v1kx=}!BewxsM#P7$Y^Q$Y$};^M!z
z6-RbVP0l*S{1kH-vY+fvk{9_aSGMnaq`U)BH^@O5bwdaoz*94eH^9UUy2>eZ3Vr+j
z98E=EkCU+gKh2z@Ye)<6(GJPs@AaKvQShjBYHFf&nzd6$DG_CSIBg&6@JN@CMtlq>
z8w=&s#8S%tS+XucefD6zO)AJrlw4>p`f$OiPh=yMuWt{%pjWv)NKU?R$f-#ZKFAq9
zf#2=J^)X3=BC40>_d2j82XV6gyabiQBaNbO>g{w2xs6LcIf|#nqiwBbCJ7W7a>~h<
zuDs?-`+<6|kX%QZg56R8p(dAfGxwSd*$=igLj2o`38<|RaA5_W^^>;i1)Y>1LAod$
z_Er;JX0~OBVdKv~ZjWVeh@huy6p~UPFv+2Fcbj&oiEc0;=-3+!HVEEV0lsH!Pqo7#
z8M-FL#bXXEcLRdAjGo%WLAY&i>n8J?VZ}0f!fH7+Ix7g-wSCnnH3Oyo=BRnvsi{s*
z^hmpJ`x_o`e(UZ=DX3cBXyqfNUTd<jYZWdX#GXm%eW2mS4Xf7Q>{S}z<RwYM!G;g0
zU(o^f9YAd#nWg1Nf~JSVNjDtg^=60KN=-nQ&N{`ShDzoi3R(Jd5sW>YbTk|eab_Fr
zJCG$LV+n_4k2l`nj!EZP7^Cz#Lw2No5_)!C56xh934U~H0{L_Z>MjsmmhFjxiIghW
z1{mhRVm#2IW+Yn5+}Y9YJr`IrQth#t4TfHGLBXV4wQ9%eJ>uEYpCE04GI8!k3r&4W
zYeU2bp-jb4dc1xTj^VLeWdjgb66@yFhmezbp+pss%Gnh1$do&;_U(Cl@{#8=OJr{_
z8N`_p*%)S=3Qg4LtwU??u@l|5&LCv9vxM^>WRqt>tJgc5JIxX&A$zKR4+cuoMK=X@
zL2Hy==@hvtsgLPK$0p4lLd}!*bp3{RrmhT5s)#x?>zq@c8i*mzO*FDY_Dm-vizcQy
zCGrjNmGnfFO@me;Sz!e~S)Wz6<)?C-6$K_DaOIUNuC!C#_P`B|+=Y~*n{qG14-b>W
zc{|;$u$<tP$?%i|%WcJdF_QzMcBcI<E)vjze9(afx~gk%7mM+BcD7neArPGX{B8%<
z2=MTBJJ<b|zzznd*q`)oq$cpOoq_T@($dR`Sr-nS&-V~?)SKSslr)9UK8Bs?tp3`C
zV!&SEJOwc%#gkoZUuV0GIHl-$p0-l@O8P#BOnIs6xl)lOYuR8TBT5MJ421HB8akoE
zmDWUT<kGyi)W<x_lppSsoDDy`h(e5}6yFi)El<sZzF1Gl0rQzoh#G&6eEyN=?_{hC
zA6VG+|JA?FH+0Uh=wWt!wkAq4hZrPf!w9!hr>2%knra!6Wb9jHau(VI1pAXq3oKFK
z&7?T8=jsiiC(|3)(a8!&Y?4zSi<BO4VU^_zM;4CQk96NwcDan2(U~AJM@mGeJ|`cx
zO?tsS^H16HDX+nuf#k2x|8sybh@~2^vb)HxX+K&o5^ap|5cor2mkHx?b1Dc#evcJy
z4s)~QptT>%0ppl-j_poK^yvED#I|Ys@osc-K@`1HW-(MUPK{=1)#@$lH{5DJ(NI+$
zA{;b$aq~ISwmCJT0;2<y_LHrj5E`AHrx90y=`|>f(Z1PFwf#Icsf7DZSyQyI3%L~6
zG)|vs`{{Oak?xR#j$*FX&B~$lGxZ6P2YjgM$K&2$TK90@m_KMg+decf-R}?0IOW<1
za&8c-w9kI7eQ1=lz7tL{PloNIxXet<l85!@8*XsMiw-UgCzMH(`ecXhyyDI)uCQOI
z<{8umvO)`+@OO4$w^N^lmXuBM93Qh^Y~!DT@FL+61L43lp__@<`lU7s_$yc+r(C2i
zQcj1wm~!PWxAoBy$Kp3{O=@D~sP$v-pntJnX|ss3Be-r3YG8Q)ORfh89+O16XcMwu
zZC%ZH*%4$X2UY}35x(d*I+Mz;EkS8raeGcoD+I2bB%S)?Zxu}<AM;<YF6m%RrT!BR
zELaQ@zxC}mvVcYdq@b{5IYaSczgeA0u0q;ki4}4_q9%qWqq?-ILCaI|68y+-wV{W%
zfoP>f1n&M5qR;Q~#d;AI9ke<t&N8IP7Gl4h%d~1@$LxsoFO)Sc3rg(8vd+#m;l+NZ
z^*crV;7NhSB!=>%dY6>;cdJHXk-nIreL8xnDY7R8Gm17iT!q)m@CXNNzt^eCbRO$e
zR`7GDf}!6sbI>YLBcj66mj3%~VYu{SrjIcjfY9+xm)~iBkV`-w)VXa=p?<7r!H71o
zKkSCm-BdyiUS3X3wEhT;+K1sxK0JTg{-_(o;*~@}S@{i$Xs14YBMY2-jh<ZCZGYUh
z&;%*Qa7>u>f{@4&7oEt^_$S%pjVM@n6!Bp0uz}j2cHQIdK4zu6I?kzyPw7x0yRmpj
z`?K~bJh7om2)jUXi=e^CZ6l@KapoSe5v#$U*Q_TWo(F=$T1t|rS*Mlyq};I>>;9rq
z9h~&>$QJql(u_k=>@U03z+5&lyVU^%vWV=<6D%Fwm0kL;8ZDZGw9Goi<ecVE-=e?$
zKQ%%*5-9-$_lSCfk$VVS4i^^t>(&m^Af5?rkOM2a)x+HoI%R)TGm+x5I1-5QqRzmn
z7hxonoVyQ)>~HI(T&ft#Iizm~wgm0#7z;1#zsuHgfXv;~A97$Z^}-D0L)pXmlM5J%
zdHef%Tk@wECv+mCSvcs_B#mCJ*l_!Y>WX%G-fVr;8w?Z+HSJUL$L$}homdKa8&c!I
zqRZM(m4=>~R6eGETGG9i3QVko<W5OJ)F&##r%B0xEy4b|extW3%OH2jh;T!sgAaA;
z<2Ulqh#Ocr*8Zj8!!|71q&#r|(Ix5H6`z#}p}F44{<Ug^faT6t2-s?*WRr1AnMe-e
z3HOZsTm2?UQOY>-3cBdjBsy^{4Rc8Ud+Qu0Y4|zJ?{#3p+bNFe|7g|FT7_!k80f%a
zFN(?7YyVj-G-8e$mg=N8n9wBB(86T>S9OWQot>#+O8SGpaB31OZ)7K-YwO#8C#Ss~
zQ?&D}vtwb`B;Bz`g9TsMH8JfxtCyq0bM|o!{IYwU=Nuqj`VxUY{?SE*)0y!UL`y0P
zt=?f2G}NaZG2CNw2P`=*MRVb`KjG0_p7j2cKZf!;>2^43G#d-M-~mtDCssRVCPI1S
z%NLi9J2g4V*%88IWsCl#)|Q<t;qMtKnqY4aI5ol;HqP{P#g5n~w|+8z&CNGXML2&F
zwFL<|<g`j$lLDU6DuCxxI+z?bWMJCk!1CB0A?1Lk=~Jsw^37^=#(@=1N}y?C4M_;>
z(;lrwx$TAxP9eUDgcU3|cy6D*WY*h05?d$ry4)eD@6cTo=Uf<y=(BxB)iCc`sjxog
zDc5S`-JtiRpdfnvncZ*H1DPu#gc{;RP3^t=B^R>pC+)MUZy9PW6MT8LY1{D|>R-a_
z>u7uSBk&F%MhB9|-C6tWN1qRFUGi?43OJ7=`b_=RrEgLYg|WwJt2xHiYiEWbeomqa
zg56ofYDzH$sTDvi8ygHZ$6i#O&KcK>FFLS1vKGZ+KGR+0V=rz4MbMCwD&dt5ENQJ)
zlh$Fs32U1aJ5Sl?<__AhWtH=o#B~a%%u)Nih91Mylbp|Z$7a{xaqoS1zU%Gwl5D-X
zKP=6NQ|uBf6WPQO8&Q{^UjtMEZr#Qy55jO5_f`|@N(~19_)BoIeL;4Kv&nhPjeq_6
z8|@2gc##8xUz{Uu6j;z9m{r_NzNo$i{gvOr{Hb)I#EeBvzG(UKYwU~LJcc(=Y&Slg
zUd^0IPQ6}Af_iC`zNBlNIX}?of_R-Dw^JYJr6X4kIdD++rEMw)i13<miXu}hu3ctd
zR;`6367rv%-{Zh?T~YSQvZxfrY-^Z39T%tU%WGJ9k$`+##D7bj*=mx(4MPj4h!d@S
zMfaQ3Eh%_b(|^{fQD?3tt^QY5XFxA~Om~wyutdpH-^;H)#geaTT@UopQSyv8n0O|q
z(<kRI+DmJH=y)$rOFH#*a84IOk#$1*L$uau8($NLFAha&V<ibYV=wClfNh_exVMzs
zFW%?W)Hc|Sm0!Q<;zIf5Ui<1t!(8_A*E>&1m*rnzm%U)L+EOPRSmOMe$sI5hFRvyM
zguv=3>j<#uVzS)gLs%n7>v4NUk>lid8jBEX$O#O!#P+e<X;s}#a*+uFx;wn>m2IDm
z;88gWlMWye3V}@;`#ta^mnZ)2=#5NHI)&`?G+gtby{fgB(SmJ%3ELnr>8U1%akIFh
z2CxrCOZro>_%fOt^H!6Jj&M!{wc%jw)s1e&P2gOj0HElIV#gI*q%3A~29iSSxV@&U
z&qFUY)1pJt(>&$W$1EnQrT$TD+a-H#JJf_wyiF`b9i8DY9Qqrlm6mcNt{@n>UzfvM
zLvB~+S=Q0<+wJv<*>{g_a~=zerA0|yU@V|r+0{hSwU{%wd;DKojZR^8Tb|8qoB92A
zRU2HKI-z(fQkvWy^;)roUEJc%9pTDBWLH004tAA6c!#6`fEva0+;*`z%NDzWr7ZdX
zjGU*YXo`eXWnZ{x%L|rnPI>Vrux?g0Obu34;EY}KXkKR|3L*!dU<E?oBgAihqD!ox
z&M!nF*WbGFwk>vTEy_|4=jNy;Z)c7BF{f1*=7<P1j%|0wRxH^Si&isvd5GB}g6ugb
zq?J}NLg=7ii1x>9WjAUx-L%Fb#~X~?VhRHQc}ad+%aKP%>td3yy&ymJwSo44>iW%5
zdqcN`67i&14xx`{k){Jq{S}O!Wsm$#^hfc)WUY-2Ax(#HU8k+MAkxR%8*?p{%MLz?
zQhZZ#3@JzA6!@CN$KLi~=P|briW|<k)Fy%IEq&jcmMo;K?y`%7{NU8YFH<j&+edya
zGOZq8c-p?UaU(^qWelaQR5&0)>XysZ*J+ztu_;|Tgc*L?zAjniXFr52{~`2mlr@g^
zw{F?AVa;u+FZ}wp%ouyISVc&y1FRh;J$q{9<0!RNIqYkeFLNHFQLYp$Gpt!;{j=+9
z_QVh#cS?26ui9j*Ys5KLp{tZv3LOmHiBq3c9si~=Tho5JYHX`h2u41?x0HoXMUV1O
z9kR9E2bLyt$;5jX=neIac!ba=RgEapA-kbl3>o)>e$gq#N{R|T;nbI<x3^V38g^s$
z6$dy1{hU%(Z|tEo#9w?}PSsugzVjHQp*H}}VC$FQgES7ukt&m+qb4^F3?1|-oN(UU
z4Jb847^kyCbWRKn^)Vp1pGr<%y#3&&CXpzf8NZYX2NG=$A>AUWq8%YbD}1YNZeNBb
zDw!0{(GDyn8jxrIvmeAjbxR`D5Cl%wNjv4lxz#u?jbiqnv0EFj7?%Jk`(Iq$1UN<p
zV<QJD{k0AG>otAu07#b70qOKB#jb74DZKL9>zv0F`_QlAPGE1Tp~xtyy{l`q*;Ft&
z)F@bUtiP4uaBg|w?v}~BZBsjm7SA2FqIBeW7=$9vX%!}*os(m>xx)v!ScK?6+iqJs
z_2RWwUU!vk>DEYo7{X-5@aK9d8jJc|qu`{Z4435^vT3)~FQraPXJiq=(~<30t4qB;
zvaz?#Zm%ii#RyZPMQ0sYyg0~7ppsJlD#6Mf)jG{aA+~sfB?}ER0f~8Q*L9uEvfIR|
z9}GKYC8s`-VC_9}NQ2dSTlb5g$`Z0QSbh;h`hS>v4=BB_;_iP^Nw1KEbifeXfU(?7
z3$i2^kZd8zrkRjOyN_0Dt#-v0%WE(h3@MNV3@Jd0z3ROeTRmB?<MiHp@4ff@Ki`?T
zzkLkv`=0mw&-tJ8mk-NxXMW}W?wvb#`gQOr!Q47FF*|H;t}cqnDwZ?L3G*5Uqg8<|
zeKUpI-;(Pmlk6up_Z)!c;_9ofzuxYxuhDEv)@2Xz2Gd&=y-0+<y|wx#@l+`$k~(%1
znW&H^Sx3MJ8Ae_rFw&Xq8`?bsT2wl>wJGj%@JYJpR9av>+1}Rf36?L5=T63*`X4-#
zd>C(diDIq#>!O45uzh3QpPRvMj{i^$w>&%28jd>n9Cd__kr}f5te1o^#K7<~I+S+g
zq@(qXZ^FcfoK}gYI!-!Bg}>bd+t4D}ge6)<UqgL#Fmb1zDt~Wo*2qwNI6hn(u6O4%
z(?SBQh){&7DQc0-{lZ{J(*`RYBp027iTA)};~sVw-UeTw9|Tg`hQ>{GXvP#pX946c
za@U$$ZKN>+tZVX4(H5zU;?K8@-8&@~RTdJQZkvOtiB!<5gp*x1+8%rD&DS`MxS$+D
z$TElQzDCWk`Qg6PtY_#fHAQfouuW|;8R2H@VE^XrQXVci_!^%K7p^pYEZ1X2^CLt+
z4=G&?K(3vwy-~K1&21D`=(KT))&Yn|rB;G;#3V)5ijB7qxn=1(r;&S+PJEHTI?)IM
z8K*LPv(9&HB9A-xc&;>4Gyo<-B{(q&+LrEvcSJZn`9I!Z#M8);k36-<-d?iChgb<8
zQebX-BiWMMnTGu#SgM}2jH(B1vb~WsJEcBx9C48bm>gmTPmvcGzig_J_mdxe>OD>?
zw;dg#GaIRF)9rO(gBlbY^F%`MH3t$4%F4hCF4|0^rL~;chOvWot{wQ5l%YsTS@gQw
z{cQ@+xbj#!Q7*q7j5^Lq+#CBd*nAalvo)9Pom45|r`R25sA@lYpkCAVpss8L3l%4j
zgHL*2bL%pjts!Ytbzy$DH<*MM*&q@8?ZI3J*c2z6-$_cF`^sy7N4*cu7=ojs-n3JM
zAC~z7pCq#ZkFN}d@2u|>kyX$vH}KoRWY^Y>-enI}*Y+YQXa3ht2c}J{ho|{n^%S|C
zl*Q2n+c_8w2P1qQ_28S@l(ODAVh8i71BxjV7fxOtfvE&UdN&*wwNYZG-R{7ey3(gG
zm<&`?_U^iH7fV^&2BqE|SbpdV3P}%7OxeS2pqAcp*WK%!W{H`LQMQ?j+VTs?y0G`u
zRXWvj(JCmu68sAa(}A>Hl@u{XA!H<`-q5VQHwR~;zuEa2lwK?o``u~#<|IdaW>iLo
zBhao<=;EBT&m)X|OU-DrP)INw2<!5O&uFPG<bc%}S&R0q-IJA3tz)H?0mnIo4kr4g
z8X*!;ad_CbwPnIl*yt8cJFa0{M+en*an#;dPvm=$B{T%=eTorvFlj-I+$m}!jSV9t
z+xy$BQ$YjuQeI3zA`2=xD|!6eyC*91F&%Z#13pJE*~c<@=x~PbsH-K*$y1_ejyPM`
z!H6Q)u3K*3*)2THuXZNM)7rUWDLVLEH%ke3p0y9OF2=!c%A0dQ#UW%rYmO}KuJHu>
zU>mBnGh4RsNwo8NL<60h#bAlKVp9vYXpa;vWiO-;$U>oWUk6Q*TqRS;`mOe%HbiK7
zQ%)-hBz9~LSRwu=%n0`38ln{Ua<6W%%4CA=UbOG(o{;YhpBX;r6Ang`LAVD;FxYqJ
zy64#Gi7ieeUMrKkR;MyP_biZQYTr|zD_4s0T_l4w&70a54t_w7lZJ&Gprmq65RzJg
z9{0Unr*@(u%7j?X9S0MP<300xIivm~^^vk6D3^>1yWU`8jZZKbr^tBuXkDkUEve5J
z*hhfXL>`TON%nIDacO_*kA1A8v2<q1X96tlX?#0&O0KSI5cYjZY1c2g+4)&ZL}~_I
zfpvX<qm5z3&<s26q&CGP^z8@Q)UKPrqC_hmAYm)D(4+Q)HF~L!q!^aYil_<1ND-Fi
zTord}JmiKuIKR^#EtqyxRpH=k9axP(sw^L;T-sxGPnK#CX+Gh=QZ}7Ja_8hcDz~<$
z<Yvk0ld!j#r3^lLBOA3+=J{*e+WfL%v&}Mstz;*dS}u<(UV-O?JCSE1vN4m}+Zc`a
z3EHF;&rUO7{IwnRY*8f!Zk9xifrF{<?Oat_z*sJ_qSDUxw1~lxEAx^A({{IF%^gcu
z+OC}5t8cv4`I#E&5cOoRw(PFAJ|j963f%8t>aB-jEgWOTU{n#yVF%u2d)oLAS&H8P
z_3*d@6O~KNCcbHJzIM&hRnE`ceeYPd!uHjjqI7Mq3btK@DGpvn&b_ujZ@O#E(mR(s
zjTka<7a6jZM|q&z&obLMVh=tc4uz{?(GRdz2yPGC*1@)`>NKZOavcD~OC%#6VCHm)
z=Ap%T;oFnZX~r%&81d)JFTdOlrwnsD8Pd-0Sn5j1I%`Mjky*kcGa?&Ik>qL1IAKS-
ziC_zgLCJ#4dmT&)jxtNI-Wb_qX)RC;!>}q|;nRU-1C>H?#X}F>YsYi!Hco9C-QqN2
z1akT7tzt_DIM(+>!!Ui)mSl&K(s{(elqk3QHanROOx@UwXnX=WrmG?IIZtIb7UtgZ
zLa*NwTajx{*L|Vt2NCJmfyk#qi9O`tOO3#;ve6!IT?=_n@`8qiV#wZMs*b$^Nzo^e
zxSyz3$m5j7+VnX}yk#ezKjPpoV}#%%i&<g1Rx3@GYOD+Cpqqcf&b0T3y(QIFZM-KO
zOs-zSHe?7?K*!leIf#rnhZHH{4MxZqYn8KYc<Q;XhmtnnP+?b=3p}`hgU|K3UJ2}!
zeyv_$DL~#yo|<+MFx3kQ`=PVjg&e{w^PkfUO5734>0+C=U=wopYVn<d%j0b!aua;k
z3&V1DxU6RsgY7yX%`|2C@P9j)q>$xE>OYkHb?G?gvxKA7f#jD=_9-ngZ9iQ7<&{_*
z@}w1DDb~U-*jhH)d@iX?RLte)m71&k=J2_2*J50I?PUW2QcrdwiK>7}y^-J|VjmE9
zoU3Bo;L5CVS>MjuLc$j2h=klc5n{LlD>BzElX>!TNXab;FSMs}aCH(W4pmwu&>Uw+
z7Y{G*hy6(VZcGGW1;Y6YEUreBVo-y3@@+~NeR}b&vcZHisCEfNrQ;4h?o4wXB2y_S
z`|L-%JysJH@-KAQjPHc+GSx?&tKzG&vPD|p)LXgr*zoPgy64M?&e~z0lT~?#H;fE{
zI^RQQ-pE#d+<v^iFvT+ke-Yo?zA(%qxw^t*VL#EXW-hg&x3Qxy!gS)e0>+NoPu6u(
z!pgJfz(mkQpiwizeyUN}U6@QU6nE`N=o+)1PGY+1%2zl)bJ&n)6nFV<kw$#*&(wf~
zz*Q5ck5jmlT;br8j5o+8(YK#%-IBYT>(9@hcVOz)!rU8rW5f1yjgPJD&7yPAEXq(d
zj#MUP>Dy&LU#}jd>YQt6o)~HHL8)0zID*9N7ut#$8tdb09CiSC5q2s$;bDgMi}gzF
zl8V8RBo_qZU`oI4QP#(PsWq|X5Y<Gp$WpNpsXr=4Z@2w&4(07j*E&D*V{f54ko`()
zExGyX8=Y1T*4k-V!0cD^wg(@a@cMb{%De2>+GIq0p{OeaO{f!0-ao2CC5P<SYlK1!
z<r8$uhCW0-Mb&8mU(=drMJd;hYmw)FqwZALjj9C7EK*Mi?=SfLiq*X>J^Rh}-Yui7
z=3%E@bEYs|l~q&ztitMkt8H#i9&vsrCkd%hbb8T#yI`?0m=2wt(_p*cNZ5Q?Z9DCE
z7QYPXN-I#BNT<9z`0m-5{WM1QyA1~Qs45LEHwPB$SyEK+9Vzy#TI~qzX!d*EI4s7B
z420@B_~gI}Lsfmq->)0ubE;#>1|+2K^M+3wWl#r0(;cR0f6xtGmd3mUlMM>=hLH_Z
zLGn70AlKj@)`zjlDrO^&B?gus#-wMal>#%77hpY{8)$#jI!;iosvS9i7_m$!4DG`H
zxJmteeBR0*bRaRa8YAtw+Vw>j`|VHie!{9}D8iDt(7>V-S%n~t(HZ|F*T;7I)9z_m
zmFWU9a(K4AVM@l8$+bTA&sxW+ut!P4tqvfGO_jo#R#<ag`sYccOsHz2(@J8;cpw9&
zCX}(k?)(?kO-XKI{fjw<@qrFrJ2hvRV7~avZk#f1<S3GCgUKo?%0Z^FziMk@3Qrts
zf_8N?%;M6-s+)h^jgp=y-X;A-=MBkMam>Nz%%IcAIJtA-i2Y3`x-aT9$*81NmtN{z
z72{7DCLb0nJN8NK8+ae=ZwsHRO_BnqIwIK0xTSnI@LJ^~W3x*A_`B*yhQ`I->cHZw
z%H$58kn|?*{P*4WDrl%BDp3a5!PJ_=Pao@9J}3!N+QU!SKRn}^dYTf>V8Y%o;sH8E
zOQ$%-s<(e^eG8vpc=a7va>G?K(=+(6{;5$8Wm+laGwA@L^AZS;w$(wj^kMtwoblc?
z{&A;~yo8ZuBT(*z{Y#S;#EpVx73`*sBHqncA>%k#70Kyv*vv=FJ^x<|6FZe=CgdW8
zbAt&PTTvBqsmRnQ^HEFq)@qX^D=*f+wVsi^q*{i!ZyZ=|QR`<Y_h<i}7Nk38oL^FJ
zSnYlLkM`xk>U6rTw$=++n874t$HmgO|E%jY2~`3^oyFV&Mz6sdz%e^x_FrxMaRkY&
z#8UKvQSXJ?Qq;JWf%(6??>>iL4XY2%XpP8O2a^^C$_^>?WXk@ho+&G(RzKhD!17cY
zq8(Iko~(ig<?FCVV~O^3n7Y$w6BZ1Gv&`(19&M~h>W)H0Bsg}l8xFpv>Nsku4o1$3
z2nX$xA5GI$D~)yl(#r`MfH+s_or-8iea}vPO7(+@$7Le!a@2u|m$1K6f8bNwOT$4*
zY9RY>Jk-I&Cm?xa7)&<*X^#dy4{gr2)6TV5Hn%M2bOt-xr#~8#$hqbZP*R3gTFFQH
z<l^HW=bRIJ)}ys?4O5bWVRQgFXb^)G%RLJ_>@yy%IoYBjdW^FJ5qYp!umRd<K3W#u
zIG|6dM@;gR1CB?rp~JBFi_7`59u0a4W5nI7oTi9zVQ%4l_SyAJogw*z(~7UkM4<4{
z=RRYf^JuM%v>pZz2-`J7dtNe4tQGC#l*Vl%E4!k`6avp3{d2p{={&=*19^kVGfS$4
zR-S!cP84Zx3?YqhZp@6?=V#9;yEg0mj=pMn$hoiLkSV2qLA?c<jnqtWuLhGGQ_;Nk
zg+-PypmgZar{u#SFFE*J6-!AhM`oR}FG?y{nA^~s;rw=|t*+g4<Bj&ki6UW+T90fw
zw4-}sFH)xv<`^U-3VU|fJ%w=zf5?I5ln|NwoUU6c;3s^9uA*{>#NyzKSRPN2QBl4*
zU=KZX#bsAH4QZeM_s3b=mnMObb079|n$h>&M6Jl6d?^gLuoe5VZY7J|4p_^%O5!4T
ziDdWm#zu)*A+7hh?a4zk>!D7X-KdDC9YJ{uMbEyxO=rX=A$o+ksbI&3i2zVWQk!lr
z%CBf66@g^I_Jh$YzzE6T!Pi>Aww{}-eP!JfLJ&_=y5K<ix`U5<!i~X(a~w<4SJfh+
z0?Yg<rxjbM!dTk54VVNpMu@fh?W@~c=O2%rQhU>Z#XVGpll%kA$*57pxFGniY4|Gt
ztpkp8=950a=Zr)ozDzVy3`nT8uT3Jo0~T@rD9IjU4n%DF$w$~%AAzuS`Zi8tdtU3e
zgndR`N3jo__Tu2<%O!p)65+u7aeIC@8me@mz8rC2HHTd)5l<7Je?ehdaIa#GQ3{zu
zF!}xTJXnG@*3#`H>R;A9QyzP<5mi%D9W^l7E5<4H+SB&JdYTIBXBmEtrj-78rh^ZM
z2mR$eDaHnU($7rFAzjs8)Q#yuwBfxN#FWP^W&uWwcGb+(*yv{ax?Gz&(8~F(9Yoby
z$L;dgwXArfeZ+y)t0m8c8qQF1o-^ckF(eeMYj#B&DC{_4J(CCk5g&FisnZaLqN7^v
z%6#M4J;Vw+jTnsXlZg0zal2z;d@Dg=Qx2dm5?g~kLX$6Puq<}28TJNCW0L~9j@V1P
zk-ZQB^`RRvgH?jm8(y@hsB^g-<SJ5X^kLGZuIk=%G#ES*zIQMLDI%^A&UV_%l5wtF
z`X=Y6OvKAny@OItY+`nGJw?yNw}#>JpaV;b-OAB1ThbaUHf(SjNg+%_CNQ6|YYLWO
z5Nu~Sl7OX1IyJL?W^%n<+sPK)0Y)0FkhM_I`K`KkTvbY@j+o&<r|a4>V#ibf83q?f
zRkN)))gul*mdE0br~qI21g>u#5SL|Rqa9buWK+HtK31Ce4Q;s~qp2hK&LGe%$cU%a
z!Ph(Fg%BH7PefTaHh#N&W_@qeX_j5|5%!AAxjAJwB}GGDo1EXFUm+8X+s(!3@K`OJ
zbXuh=bU_3mn-HlYY2033y%<pE_)@CvIhb1Ri)0cye~cH=z8#t97%DCIqYf;mXL1HY
zp0!t&#L+Y!SwRNnk~H{7kd$+63eLCaD_&J^mqZ~pGH6jHS<S)cpsb!*&pd?5cgkMf
z#xx`ZEr$%kIO<@u%ipl%I{W%Z!zcW}a_3iiHpt&s^+5Q+Yr3_sR7FW2(yAzp4bc;Q
z&0mfybgl|1P8!2GTUsq8mK}9kF>YB|mnR_Zl9B<P=_`)fEd%4G__t%w8?YQ}olwHn
zs<fI$kMu$=t9wWadJ0Q^oLC4f=?Ci7J*wN1Tk7)qhNLzyoe@s@p|(mLj>uK@&QpA&
zwgn~0#-R~P__7rZADi`n&88SE_*k-1dt@ob(rdSN-=^b^Vx}FK#tI=><9l0MitxrE
z&PQ|r$?5nETwco6wcG2}gO&vGemUlJ7;lG1FOWei6S)>`KzHPNxZ<I=z2jb|(Y_N7
z9<BpkBW+nLeviF2`?qQWL4L|<sXGgETiG9Q)XGQsy2U<I02MnWfFyIBkxDSk%53*3
zoH#Mt?>p@jI(wYGiO+1>Rz3Q3me0YmgKUXtmDz6&BB`Ew!^8o1t5J5LG6Rct985Au
zwnH{wY7rc;yB>Y|3i|qgEh9QN#KzV_P0lH5nTeD9rpWjR2)5N-@9~IC$n*k4{N;89
zyW8ut$7_csrl()$v=m@LHN`d6_a+ih%~4-K!~mGosFhybv$c)!FQ3`s)d(kst!p2N
zaVx6G&r_bakvWU2Tpvsvi_`8dj4C3?>coLVd;!US-Z0po<qQu~QP{QYo+5@Y&xHD{
zP=_=Cqfk5e7@^#plRc%pAqiJIla|ICnnK7??b6E?Hst_y>%!dsX+G?Y)oB^dWQ)=!
z6(Wv<$r<Jol*({?;gr3p`^25vzFSHqp9n_dzG}0*x$Y}-Im_s@0~^>|QVf=Ca?#$>
z=3*69Cx)G-$d8nbhgk~uwnv5WEbXlgmN9)9X7bi{UC5te{f8Y`F0+v5@+eLzbR>!>
zyV$;=eI6~DvlJ^2thyP;Z%W>0gr%drt<6)|S2RGX5V2UC34HWh<rj|GH+Ij-R4Zj=
z!*~hG0wcCRF*{+su7k1<N|zPC>#Wiaz8KLF@l(cem#weY>_vsx<Fw)mbw_)(grK2%
zwY*Y?YNy!4)dw6*?oj==&y137hn)w{`Ed0jr48)}x=U{F+y%kkq6r~t8f7H+&Hs({
ziXJFDP&Od#kg^BG>D1iC4q^QwWt1%jkMQCc6e+T?%!`Y2RSeMmqqd<wPd;&#SE0_O
z1B-Fi0Gzgwx`MEwxPtIS2NrWE&K$$qHXGa6*ycTCYm++a0OHcz)1^3)0E-w)KyiuZ
zoBIzsziS?}d&f80eI1P55-FT~PX{J-dvp`h!ls7DcrDYkL{?x)kYRO<Ivu=C1*c-K
zw6Vs-Zy~aj^-DY6Tl`D;P`0@dR%gzVQA@rtF!6MqYzZB)j<XO+JIi^zo~FJ^fkMsl
z221-zuo#<Yy`WYb$0$1O=!HTg<vrF3r&|(+XJB-aikJ~_3oLmmk+IVw_V&6>gQ_58
zV$vK~h7aOY9vh{y9*l|t_LDi|>nU<lKgT&ScB(PuHEZv28aWhkeoi<PY&t1{!nu8>
zl2-135H`me+e~(+H>m{SDW_R8tvYgwcGI|#;%@in+wWMu-1%9Q!(wBI^3W`;{;lm@
zEAV+bEq!WyJc&RLbR!>I7Sf08dwEhfMh*v`BeRV1xg$2)25&=eit@p1-@^_pxjab@
zGJ_wNEt4sh1{%Se2z<Nk!R+ew974g*{9Jr9vX;K12B@SWVk-}LgGu63IU3cbduQv*
zJ>>NCoaQqYfFzdC-0Y$HsOA=GxiB)y$$kP}%XECCXYX2E*CE|9s=1b-ZdGs!rt4f)
zb}z)59#&@irUvPaPC{%La)4-YFU(Ozxo6(iOBgn^DXP?@(&l8|xai=MpG9TTEm3QE
z*xsFAI5IZF5%qnisYl+)Nsw5-AI@Hnj7~T|a|GdoSjO!=^){%2=&etbF2M&0N-mXH
zD312tHtO(3%y9?=r@g_XoyoX{C}`hY*GaYA&3MQ-2P}qO8q7s_E&nlUWH8#dG$mkV
zUj;a2$^{fJXK~BOE<M@4wM``=5!`}iJWoa=CB>>hWuEx<ZMl+$l<np;;<8{Gd`6Zs
z#L8js>lWP7<xZnsxDWY0;`;3U$=@Ejj8m8S9eBywktjaD-M+ov!x_fIPDAu@^-W9c
zJL;8+Y>-ppBKurwEntFYN!M%%R~j+tCz$H&JG)+}D6g!^Cmh&wf<~xWKQlhD5jvir
z*4gH;fq!6tYEx@h0I@3ZhJ@+h#iG|xiPAn;4;LFqd(U<UmiB<_uD{V9NeT)3f$DV9
z7LudL#)s@f^`@}CGJSa-Z0A#P@FMrsea`5KLap}UZdwWKkt{#yfHX)qVCns?q*G!T
zMrQg>Oa1Ik&2F~u?gl~Ss#C26COl`N_3V49C)#8rre(FPNQl&S?2Gu$#tC5@2}|f<
zrtNzh9qFLRNvD~n3EWB>P>Jj#9ievC)7W0NIhb@q$Q=|pryMWw(XP9v!yT*qHegzW
zvTu<h_OZG`ap}^3l)(=y72o~?xHCx49<uLi-^Fr~P?3k9pVK$apTsEE;So2`!&q9X
zUeSdQ+xOSt-gLe5GuQeC9oRByKhQoY9L(vz4(o$=n9`3!I?-?Y!EWKmn8-Z{-;%q|
zI+*0|{W#3@(QW`F458lfdBBpf5frk-9&5D2CY*~$(xEK!l63ZX!wc4gqYO_9bDzz%
zH^34_*_d#=!DJbg4kc&&wr=5_3QeIIAND#JEnk$%js5-@7O2rl+g?{^D0g$}Q?J0m
z=mYVakPg$g9qskVn=+bH{J?>w^bsa@aUZSWa=Xu8w4H5?dSfHKzSB(hg6N{k5!AY|
zU2W}Md*jVcBiK~>6SM3=+7foxt$cQv`S;ErbYO9HR^LczF7^LC4a0oFq1iV?aSVhv
z49A6qgw`2)ZEt-G&x!B56nuO!4jO!H){-L9d>xXOWZJU15!Yt>8u?Olgp7{zS|x)Z
zbIQTzI$OJulEb7Z?5`UJ{U5mv#Ww2TYs?52VAiQaWC!Z=G#K(4$k89e<P9V9g7oM-
z_CR4|2XmaaazaDjX(S(KB}n_Yhza{p+m{j3D1+9Yb|6XNL^yFu^5Nt@3v<tVzVkad
z<`gUYNOdO!G`g6(Ke2QUCf*~GCQTZt__Q5uw=NwWb{e^vbFi5_&~ROx%g5TA-tq$H
zcRk{4<Wwl{_IQn%wqpf2;$7e)4k-N~He>KX7}B-Vz}QSuC-UV_zIS2AwA0AB;4D3I
z5l8G~@{UcU3i2~4G1%*X1<Ow5q9#9Th6r|A(Y>-*sWzjXZqpN6kTwZ1=tFG7U~0WA
zMjJ>2)6&JDWRKU=;7V9wB^R7?VA+$PJ4x$udlScNPvrbeqS!HXPP4vR56`rMTs&v8
z%Ok3}HSV;MEv}`!{Al0Kw)-FhnT{2nsl{~I!Q?EjT(`o`Ro^%yn~hUzbEgBVQ96i}
z)>*T#E}rk!fh1zFr;11`LIp3@I5as+LW*5zFIwN<;51TXyt6liMFSuA#d>v$62Uhd
zSPW$r=a6kh9ECG)c-h7IgLbKP4FyVJ&)Ow(jesffqYH>u2PHnimyX*HwGVyj-C-&x
z$JK$UyBOO!t-yY`>(9vu3+1l^%U-XY;eaqzwV7`aDyK2fLxl3yecE)w56j!0tbR6O
zXuGsm_O{2pVUn@6Fkaa(7V0?>JWSsiVh5(Vdg!5tE_;bRRihil4xLueD25bfVHZ?&
z*nXsYts;tKh0vBGn@z*OvykW|Fv{n^Zf#HJE08<;!(I)7LN@j6N4o)0iSRUe0LuYc
zMGpTc1q$uQ>ME5L*V~R@r3O|j;$m0_;`DDOvq{B2-Y8@3?Mt181gGn-yU~6kf0EPg
z{FGG0VWU&_ll582iexd9EsnnqW`GrmAsw}!YFD#mRaFnJmg5CXTrI6fYMqLk{B-w`
zaiEH}JO`FL{CzzNto%%CjZ938IIVmJyjX$xD$Vw@9qgDEJ=?<x2d1xDwq~{cTsJEz
zfyiYpcmFX5Q;SwoZ{L2tjqchNYwjVLWUT{HS5`&Xw*5jrbZnyU{1hUVQr~`Y;1$}y
z<XB)dpm8v{+OK1Lf2leuKU6xS4|{`&JIUGn@WL_s<$7EO;z*y#z454nsR6&B95K||
zUG^&-S5Tsg{2Ah9q2WbT&>}wEZ!8l9>~>kr_N#3VE_<2s1yZF+{+kP&tAZ;PyHGVX
zo}$w(el1C*uL8jQuKr0+R^8|W_Umm_brf|}PXnUKx$60^u-~Z1mP{I<C}F~JFdAX1
zbo7aP{>|?7WGHKF*|&Aic5fJo5O~Frf8_BzkD&Bhbx+I$ol%7&NrjgjOe~P(U`@{N
z62a}avx8UPc#ZQ@3t`#PwJYp*+LtW7e>C$DfF!Bj`_8w$!+!VCr=QEU>lv5`OhOw{
zaGnBhBd*lBATLMV8n)kS-yV!&z@_A1l4{m2U2VVLs1&teW@Mw&PULIvUT%Mo-CJ_=
zHP<@7>#MXqq?J};lnikv5xf8J8514Row3i$4xxDSoM^7f9gjG6^gc#fA$+pj{89Fv
zX`Ez}rd8H-xy~YRugFHTQ(~O`aYqm$SV_W199WVPi%_=x^I_z{uWx_SJSoe%05;W@
zo^lXzi(J_Xno{<K!q5)cpO!!!0^@F{sS&2f$JosOtll2xIAAi}0@D>U>!~j=Zhzhe
z^L4$gP9vx{wk$3ude#1-1~os5uod2{te<e)&+m2c(X$x5#oc;-kNsu$=}M*5iaPHN
zrtV=Lq6WYz`>UM7g}FONoZr!Xl@UE?f89Mn)%s*$xa7c+goY>`YJZcu-jt4jLZO%(
zeZT<~6&#Ep<O7J@X+YKvc;5cD+_n;tw9xY)%KWWGD^qm~DdpPYwin>6Qq+#v-=%Zn
zwJV%oTE?qZtg*k(UfsQ((wZ|)D`mh<IPwA5KeSPPZEvH~hz`mM9u`*~t??lIN=Xs+
zkNGU6KhI1#t(2GX1QYcCPwnf722o51QJLJfXe(%V>GqPiq`N1x3$Kv<bA6HO{pbKV
zMUwzaB_!mToaijsHsF=8e`#tpPBWIk29^MzTq==kSix8l`0QWnk%JsODI~oTp<Xyw
z<!(;Eql(N=*}pY2KxU!jDI5TEI8o*&HW1<P?;Tx;va~;(a9}a216igF&&itnbM_xe
zWgCdBbbiOQ$~Y?f7GY5KpY<(i!jY=8EuMVD!PKfyMGDC~d+fhjKNTJ?6+=!kfb}^P
zp%2gKf9F)LM%6`@rd8|y2;GwLN}Nf?zyE1>ayHNXlpvUNAY!{~W^n-6W7UC}pAgm^
zSPi|L?-~MJs@m@ClOC%DavO8xw6gOEXsWmfH?~hMSkg!wH15E%Z!9teib!-LH3*^Z
zQ`++=ce2T874h=6d?Vk`rAOGOwti6rplUBC4|Om(L2#jv4iL=-oJ00$)hi9XirA~@
z4Y2%5SRLj8_BL7KwCR8PV`VRjlJ-c#idO+xEcpzJBYaLefB2aZO!lnmhzMOe)ah7-
zV%bvnPCNKiAD7)lS+qNDxqIDRYu4IlJQlLjFbBV?>b=uWKD26LV%R>j4KsGdtQ5Ax
z8;pQDI<GUmF4||+yEEoqx!8K^nF^*m<lwt$Md^^@nTymudticew8s;vfM5^;Qv(~v
zcH(bK#-O+b>=E`k*>jH9n&4owzSEBDLwaFyj*%83c}b<56~pnl^)1N=$o*W{?*lMa
zoYHMC*ypux=8f`YAvbx0#hOGn*AefZ-+l9T1m`ppOAD0;kcyDr@R51Rb7Wr7yJeWb
z9A#fn?}=IhNx}{+ZF~x5g;${8e_;)f^3OsoWI}VTfXM-Zwp%>pK=Pe^Q8(xkjpbsH
zT^ujHcNl0^{=d-Y6f$C8{8;fM>dRy5oOJ+~Y2)SlBubE;EZ@*##LFNrdUm5zNLgk*
zh5RLV%vj_U0J$nT|8fqj?AvqN_aQvXm!l{G3DSpr5z{4a<pA23w5cB0Ewt|`B#Wdd
z2QMNfK~3kdeQ6CeX&5465t0K7f=pu2exOsIR5$(09xHA~q4=>`j<I(jMTko3Xm+AZ
z%i>w~+}0b!7jdhPuk;FD`|0!FdX;^7{v4lJ@BAztAWK5l=}Fb`D+-fV`EZZ$R-HZw
zCKa2ZjpUs6?JFM(Zn|X2jW;^KYfqQEHBwLPs|qHEjV5FtnoEOedZL^ep^RVMeIf2~
zec@9N^MsVUv7vdxKeGmm3cJVj+Sk;3w25fn-j9x5Fru_LOLamq^27GE?K+}A6bZYB
z{s7C}MW+@Kls|0GYmho9z3)JBV`Fb=wMU+xbKIN4OWk)`xtCG|Z*;?kDSJVE5)+IC
zCIcZ92qJ3?c{C2bcE9uJRZ1V*W|y_sL^%La_BIC=(<IB3R{1tG{QT$_Had<&;GAYY
zFzgiwK}gmwYI98Fgyisy?WBO8gGn~K<fa?0wy$f;US0w@Q>GkHth-Fth~P~Pm{fGs
z4+YnD)>gZ`zO~j`2>sIWWb$XFI{4HqW;U<4D_WzsspmAb9gmK0oUkifmo^e4<g}8U
z=(?N+oNiz*E|?aLiiHzH=D=cnm`WLqvM4<SaK>KJ21H>a^qqs;9at?pGI@#{Mv$1j
zw2@%wvCXs|9%4Zpe7*+{nG)VqR3_9syDHzaFvoct)JqJJ1y~Z*(&>!~VjHrT<=4@V
z!}B|wRuNP<_b_{%u&djAu8gujG~<$tei!?D+?G@)VeoR$J@v3wOIkfxVW+&|Q*sqY
zl5`!ajf7&mrgdwYu(N5UmH>&*%FIdG=kVjmdu-RXE-oD5Q`v+V1Ynx3sF!?Q;YHvI
zct%!sV6`K}q#?gR+DwK5D&@$(zPl$JG;CJK99Z1bxQ!yRWK%C)!EUH~^aknS7_1#w
zuC^6JGZO3U#x`~oxWph#Grnqo2y3!#>b`d|1VfwVUI!yCxfbi`&GpJX3kvR!uy}7U
zsk9;6Vy1t2c0*3v>G0eF$nA5j4&ucb)3;Y-&p1D6Vgyl&rq&(bag-p?S2mJgF^iFd
zR`pstQ5tCZ0<L>iZEWfH2{%~DreTp_Wh0?FSH(!`d!z&^Bq_DAWEih*vs1C)SzJ22
z4l%pnW6A$-Lx*DFODer+U*Ci|qS><aD0J^YLr_dcki9K?0sDt^a)K@FHT9iZ#!}Q|
z)XZHGjJ%avtjGmHSzByr8y5JXDlR`B`G-=lz-TbnP=;)5%x>v+f8`y8TBbn=`77*V
z!2!4`21zlpk`a}v!S<^SM{ZeL)(sLtdvut=q$9aKD_Xh3hi)cIltkwz!jpdruCu(6
zz&fHhsSbw-1E^SsqT2WIy(vc>vK7e-Hz52@IgR9a5go+DVTHK0E!h48iiScM0Fu(n
z1~j1|yRGO6-wFm=TIucOsMnEkQfJ}hc6(ba>!LR8kOPPpF}?_$QEbBQXfF@;r?vLD
zcQEEoEDwfnuWb~&MukIZ6_3)+A~Ao5y{=%;FZ3rZXpaL6s#j3q0bAL2fa{lB?=+HV
z8EI4zSeje|i+jte_DK;z5;<0r4lw1AVJ$mCU>7NhNh1hP#eLZAE;yLndY0l-zN;=v
z$uR#$4HgF@#s(Qp*y=XI_wn#;PBXqy3e7pbhcmS7_3axO^(d#!>I0@X-80(XI%R9R
z*N)6Mjd()E46*W0+uG`i?20OqMW1W41e0n~Z)8OE2nfF-1aRFVS;n_1j7i+Od}R(k
z_wx-ss#)IM5qZaZ8Pq(@fxiwuxjuGrjg+#ABFxS1Nn&Z@|D*$mOT&vr7bT#DmBHT7
zVU)SA#f)=YMo9+~A53In<fY1``Nl@cm>SU9`@;?-nfOx75jw@?O>I`G5iyQJMLQo;
zn=PYcZ*Dz@CKcV8h26yrz|<%!=|-ad6ZV!iYH6bl9#nNES!8D%e0`isCy8Q|9?mD&
zy*ZxLNSr3!k5&<Bo+qZLa?Kg^AW72RnpabfU=+_et=#<A6YqpbIc48a_YY)U#em7*
zfkDI@zFa$@Pvw9o=YCuF)e%n|fe(_)_c)lO@4N3<XWv*1T=9eQNMdG)MXB_ugKXkr
ztJ)G&Dv4|3)06$?i{W#8danjji#3E~@=gX}`e@=4ZGBy*tzA>Wusg6As7x~3MX1~}
zjf6vWk1DCk)5wXB4lEaQ1kP`^;Wm*l5QG33CmdKyx`3&T4>6|w2AUu}hY+mEo&(A!
zjVUo>+&0vc@S<U?<`~k2vkpw#{7u*1U?bfSNP?-I5{>ShH;kkr`2+H7vm;n-Hs(xI
z<_I&Q(~3cCx{7=RvOAx((V7Zrc7kc`1!b{>^y$TW!%oPxLAeZv1l8aV_vPXkrYKp6
zlnx+A78L~VEgOCwyH^Hz+msz!js`O{?X+rkpHP<|)sT&u-LdXv=|wy>zu$r7XRwb+
z9bfd}Y|f`H%-uu8nA55SsY0k})flgS)!8YKrkuK1Bl$#RYzLq0Y$HXAh&#okG_~0#
z8jsx)>x*WN03)qRH~h6NU8mxBO1#CACC|?9axkqMp^m7e{M!r0hR>dcvIH9f7N3>I
zbw(dOH^$v2yG}2Lb<0Emkx2G+@VUuiEXEf)l@I4gPU>m)@7JMe4y?v(JJc`kCwGS;
z38%Z)#@NF#F__5F;b78^hFmG4ILl0?sMO*x2z|Gi|ApoycT3g=Qy#}+Wpu8MUgT2-
zmH7MHT39u4zXW~S$tbS7W{GXhX@wTgz&UB<vO&Taf)78?K9?!$6Xq;R2Nqoqu_Qh>
z^ClEGtRj2Kr)Fyw=tu-ik^<$xig3xa45Cm{g4Td|-(GvLaZA-h8df1zAT@8Kl*PML
zT$HQ~dq+JhHybH#N|M_jhJGA}4qgckSG~yI*$8)RVq)59CB?816xxE3&mPJFQ4-0p
z(})Ywei)}K+Peyt$_R#gOhUkt4`H7c%Yn&!Q}tE`I9Yfzc0othfGIj9b_zzu73R;l
zy772bl53a<)?ieVW5|2;?lwPnQg$2Cqg~@8Eh{wuG2vbIaQ7LDu3cafZ5HOPe3iYY
z@VZ**Uysipc3|-|&1`0+%f5$W$ljY{NI<j*h*o_~e&^AJ1NO~D6<wHBECu;|WPw(Z
zT$qe=RqdWxS~QOX(N#$`-;z&WL(Fj){Bh?xu-rp?oLB%|j@Y-hp}#d_XohATP<5Sz
zVF*E$?K0Zt+ls*D%RAt-a==2k)h?&2sRd*2E6kvcR6e7F4kkW#FQ=E(@!j_RCb2LL
z6qp%z0P$d%jkPe9Z4_B5_U&!7BA7;M$Uz5|TxBcKx6}3=-5aPT21n;OCvPwert~pP
zdG?(hvyvk@#cFYMomhT^uLCbRio-JT{<41T1J$cC7R%7g<Bt&~2u4yOn+ghoeX#X0
zRT+eXgAOd_s%h;h)Ty*-M)W^Yc$tOH&;ga!99YfX@aely2W7ET*`1&-`_SS$Wr2fj
z?sQ-=iYz{9_;3YL^+`hEKitMQHWc}E6dEX23MQsUkvMP?j6cWiyV~^JhJ1<Y;{bxO
zLC|r*cQ+n*+uGZ`8WWo^4chnA<KVL`a?~448o!bVF$~-Hc5F~pCtyPm@h8Tib5#tY
zEO@+IBAy0;p-^<=ar;Qy2a<I%{Qs+`aFsq$^)jdX6lc*<ogb|M$t_K+S8@E3xxpk8
zpPU&Pfr)>tu1te`p1~7|XbwhXI4YCH&T0F;ZbsyvjiV|YSkiS^<})-S%1PPxw;{ac
zj&)8WS%NZ<83$xF`+>%ISN6xpomP@OVT!mRf3OBdyihxbBEYg>aco@<J{~9$M9R6`
zwvV<CqaY(Om`*#c(d#b7?r_i^YcNcL&Bxqe9Iz~3Zb(&gHSprql*f&~ZR_q5UY)Cw
z$#0pMqLw&^_Sv?_>RcnbB>#!5uyhiwf6r6WSj#WLOiS8Q>?n661Jju`*nVbhdrF=+
z5>dg=P&X%gBetXaW`*JgCv{*FQaAPQv7LpM1TPzfj`;+Zqk7-yMx6D#n({i;bAHF@
zsFaoJEpWb{?e5mJ3S~*)*HImp=xXe$4!*b;vQHgWkgI)9VN!C<!~J3%JDBXkddg%@
z+1_pfNcm!Xl=%isoAIV0+m}Q$8Kn$p)cld7uf>7OYk#Av)%_u-k*e&yijF&wA4i?~
zap$*QI<rN`;tbotL@l}IrmLObML9YRLVke#cBt_x=uFz3yeGLFWDGvw;6tXVGFnKT
zk#djHlLnvg<=f#J18M4V0LEWDIlH}Kl)9$_?$g>zY>pji6f`k3G_xhnEbBmP`|EW=
z9scK|`Jq$k_@q_ias(2SL*Td_>&72BfOq58)8F}%4klQP_@68udsr_i1TsJ9AIICO
zfpI9LfOaa|%a*RbV;$v{Pn4+i0_Rtb_vV=m8|-AeaArJm5FJ4DOT`GZZL;y$se(zS
zR>0Z$`8^ISGsmcQv~glM?47Kf(>Z5Tvt(0jo^qN&YwD%K8|4E^Dd6$8U~mNYN4L;t
zAX+8UqnK`<XcV$#W8Z0Hk5|qh7@Vm`N*{^NWOOSAEHE`lPk_-v*J)%F;$sjgMP+9j
z5wBghbj>=aSzmexvqda5J6DfjjzbyFGj(Kr7fU_l;8VSs>=TMTJTo*ZHhsQP9=oob
zATl+I+f?y62VeX(4UbZnB0k10<h-qE1!-xg*gnNc78moo#W>;kw6ePjZshCE&+oQN
z?ef)k-03vp0zElv^j)wY8W?~12F``)M#+g_;%fSdsUW-#_QM4W2R`4y@p9f^yI2GA
zKFBIK-{y|o7y?ew7l0HGl^0n-sP<&{>L=nxZ8wpr>0t8J)F@CzNzU4|g`AsZdLm__
zX?9cD)enkD(4?NHl4IUN6be6Ma0mD02-vkB$?+Rw@sTS;OG{$u*a#8x(;GM2(`{Q8
zjcZM@4pW8_x!}8I9xpmq6}Q}p>;s>agu}kG-F`ItFf%md{H(c&QY;d%?8jQaN{s1h
z5jvQ-TK)#<={Ric$2$^A%Z<7ug-5|5GIxQ`{bgE-sEEw=6K%5afhmESQrQkHI$OI3
zlO+{QVfjB<&lfw9m4`rD{>pNmUUKjQ3ZHD_a^Mc0VUO$Hekwot`WvqE`guFN<)<6-
zne4Oek9*f%OYtli-_LZTqxp_t6$e(crc^K#6eJ3%DF^Lm^Buh{6V6X@ge5mGx!!)R
zk@|JlUh6dKJC8F?x;w&U*#v&ReP<{nd|L!=b!8~M;nQaEi2~~{bp1Li6JdzTIRjv7
zgyhW1T$kKd^LJWFe=+HJc#<kjXi+qib@eFUZoibsCl$)F^OauB1#)(k?)b~~>eMMy
zFO*z;*cUMpK?Y8(wqMD?+Csqze#V%jg|Ex>SvX?9+9-uBFY5t5?hPjDP;3lap#56c
z0p&@`qY?oQIL*O)n2A`J`~Ta2_Ukp@P%QkVgq`#N!!AxZ_+(6xl^BBAZ`5$?V;`4y
zpQAM#SkBYB+gH5FezT@JCZ31N*h89iI)WkL5Cxs9a!C>n)ElN4u>DqpU`LwnleywR
zavz4}P_t{we!ClE+2fdMgrcA#2U9#mj=EHcey4*;q=8jqy=$<n3nA{Y$9}g)A$47K
zPlDf~7YlqmTbZa5DR9)<@6|onmw0w0@Wo1!MUWsj2d~Aq;<j7Y*zYIdj#0yZlFDN=
zCF;#?us`UCPWru6@=W61<zTdvjVi9l{;)0!1u!_Ca?ubc9E=8@f{tt>_D9uE`W_*q
zWal`rpzAcPN(7OL$;NDdTvw;-R80Gf11nuj&S+Wa^rqY~B`Ez#*WrN#lc?wgCN(PM
zJ0fQPX+2$Huu=kb+77VXp0B*}E%s+^|3EENDEuWPY$fr_rdt~ujGL>f`!HoXegc-8
zEcwst1xi1Xbxq-yz%(3B&hgCc_7?*zM?wr0)_ArKvy&<sp(W5RSH&B&yUP(OU%KS^
zqxP5iN%u$iECyZL)t4l`jAcBFN8_)mFJk#peh(ur1526zo*wby_SbcD>Z005RQm`s
zl!GbGScdiLhDKgHP;S|O)6JG5yJXSc<-oK&MyXs%GB|lTf7{l}s)<R2=?MoAUDn4-
zedw|Z?C+YVknr9(K3K7%0P}LxCrW9-6m%@gehxSJ`yz~7)0v&qLWh+O2R?K(ZCL-%
zNP%SDEm+`a=d`b8-~Go%7ZHIXvGJ%kn8amSdXHkp`=`1t1M=lKI^zwNm0h%N9kqYX
zhc8`s_nH+>BZhMD9FP!R`Sjy#NRm&E*}t?dKsl19592ZcCdyfH$DQ}ECI4%iyElvu
zPj7UZvHtPrv3E%k-(mmOU^<zK2NSa8z;eo8-y^We{=JO@hGGOgryZj!>Qi=vqGf{@
z+JAJ+B<ru_5fzalgAs10gO{h8@f_MT0e9pqw5|4^ja@$ZLA(|ppe_TDBG8MHe<JK!
z?#d_Zzp`i4Y-d{~M!5BCh=gkbNB_IMpTjD}D5bDhwj2y{w}aO<(8C6@hyR^IQ~94-
zY8u#WPD|5I^!#mRbIC&n{L=8ph$@Ob1$^`}xy!&{*eA^u562Fv&_5mZ)WKj1N;QI~
z*(cX^`-pFpT~B6tV9B|-Yq1uKWDV#E<JhOn#oXv@Ak|x<SuFij*xosO$v$;1SlBX-
zIFc*bY1e$od?ntbSYj&Q*r&|}2fyke=a-hCiZO&a);@hMS&;k{r<_*qN;1QRJ?o4;
ztHIb9<;Lb<EC<#J>`nK|z?Cp2*k^QI(HDiGSWFDhBp4Cy!rYUOVE*~cZtUZ*uCU-r
z8f|fO;Kik(mSa%OPW!AJm4&(0lMIa0%!j>pX1HgconQIJdz@c2T4|@I1g(9}T<za0
zs5v#s!GaD%L@?7|KW3j>cW9AH6yHgrSA(T^Ec!oepI6r@+A|Bz9(G_w##!<~`H#u-
zKfm<@X(D})dmLB{-C)g7ZPBuqLNfLR`N^EzFgi9g@qp8gZy{|(8VStin0;Z+;=){<
zeuYA|2vFr3)UhvWBQr^QxzkFH#+h^W#q9-CD&;lhwBrSmYtaJa<FjWMY&X_XWO#Xm
z9au1{!F5vX6>cZZLU%r=?o>xr9$$Xr2Ez^?0m00#eMx=-{8s4-P&j~E5J%;`U}utw
z&~4lHr2`}<sY5}6-4QfsxZw2}3v<&GGeaAPr8n7^&4s1y#+z^U`q|six93(D(~A*V
zOx|Ewico^(G5hjb8Wi3_5TljzpduK&`5Y#OuPB&mACTEfIOIVG7Sy0(i<v*F-Cmhc
zUs+#}@%u#Pus#O@FC4sxeRP^iEF9osUscyP>qKUdEBKix2jXVIM~@_jVFAFVe$c);
zX_}mmzVlnFgXO2_!mr8JsH|&KPAi#O$Z(k$9m1A<Z9O#=Y&L#mjs{CBTR83Qc?0uN
zd0SaPk{wDWYw$^P*Ij#~J-_Q}RD(tfIqtx6@e)WeZ7--RQmm1)V6OwyBZ!U~nzqa8
zeFLU;&A%9?Y*!AxR$)YXv!WDvqj7&>yEEz@%SbO9VA{Kgj&|^Ir#uj{C96VQ?sG5d
zx~)nQ#L%#}0Sh{XQcy!%(8W3KQP+>y*R{7nCkipknCHdN4DkBwwb*vi0xz$357-xZ
zCR?C)7#Rd=_mu#4&aUWsAYWT5_Ys|?ZJi#1FA5})fF2wTd)8T0zAL*iN^5k~U5lhv
zF!5DcmXe*kQwdX?dvRkLTPPegwb82$6F>XphhWSvY20-@jtM7}4T!AVcv0=8?Pl%g
zoBB>W2sd8?+hf1Gs!_w#=9v+v^(<~yN;9NvFKhDv;b>L~-tz_%Q&C>frTGJPb$vy0
zhm;GTk>C`p9&dQKcSHlIAYmF7k-DCGk1fd^<d!!$zhVrN!G$4<_bN&2n!-pV?&^56
zigr1e6e?c%YP)u^Cvfe2ih3oAiTbnM7g3i>IzFP|bp=xug?3y7up9_#RE$1g*XJu<
z&OhgOE`JQixOU<ZHzaIf?$(JJbW%9$MhBJ~PUwlre8k`)i;mac*fx_b)Mg=f+5r?_
zxfJ5acET2J%GrFdPifh10P;*1CCjCG^X4`Hm%YaMT@TUHfrT=>FRwe<9K%Ma^EgDZ
z`y5Pti-P+Uw0PcLQTvT_fy`~eX@rCziTm+zf9Q-#u8KQF0Mgqac6()`H&y;o=xsLw
zsss}3iFauEVn=)GVS7~@fmQPRL%RTK1*x8gxL(t-UOo5p)et5Rk&u#f7JEHBS;~*R
zq33G8<%dwcASzHh_qA%DB;-c9o(%o_`qqEM3(1gdc@AuV_%v5iYp~DQYjTw^+uM<)
zXw?|+d+OoXrO(*Xxu;)B2menMCqDt_+!PO@EQn+bBI$P7EjfsYUg!AVC8wRXa4LRo
zp2-u{rfgZe6Dzb#j3`82Fhb_<?-9WrMP8RDX>8z}Y3KLA=Seq{4vO!^R&-22$zrtP
zx-<na*3#}I>)WllDzM3i*^Op(U-GrK!y|TEUPAP6pMiFo;<_YrspR7AMG#^=3hYt1
z|KQ0L;M0yXN<1UEQg+8Px*P^!Z4nDQ98A1BY-&78m5N?lS94u>CS)uT>ptjUVoR1a
z<e<H-?#%!TBrUeF22<B4sKV!-wv}yIh@>2ITJ^@mXbyVr<F;y%-9XiXu@>Zfi;d%$
zgU=6I-oIZZ&Fs#(GL)^QRv!iXFkS-6QN*Ph8JqJv?XGqindbycO|X~%rQzhxUL2{p
zy0xSJ)U4Bv$H+MK?&!)cdwt_#9K|=Z(P_pdvf{~ULh!h)DOgC6^qpK9?_gXZ;?icm
zwv8>DzH&5_MHf*M-Y^hK5OHMVNTaZINqn#7pYuE4En$uB!3XW`Had5$S-xV;iscSK
zd)HWR$nI&M$HCJZXVyFIAW8|Sx?g2T#m^`Ko6?F>>&Bf%OtiMRllI0$4H0iT<usD4
z;oFO)2UL>cP5Igtcdl@Litn$G=mWvNxxE$oQe@=;sS66ca4@xTD_lw#EhdRTh$Haa
zw`Bis>2K`aKgyz_Ei>kDA0rB0*P*`M+rD0viAR}X0J-N7Wx;umZ*4OZqM~edXB}8x
z(tlu75wPtNh;|<@CX3#IC7ryEqk-_6y{&yVCX_5V!EyGHEe9s1MAE!`{u<10tnt%1
zk>{LN4LZl#q*B3BBhkuwb5G~tHFs3aqOAMvz_k}EO5WDDxj^#U=d|)|Xj@OvS7qxC
z6)aLG8CW(P2bKeia(}*Y^pNRf#yx1mIj0nf#zab^C|a{1syb%<Hm)0c7@~2M04RQz
z3a)}FrSt8!4c+qnxWpw$oAbxK;hQuftDSU46*bVdFw&-Dtam?iOf!0;ol{m(_>gUE
zJY={(Gwrm}nB6BBTP7A8tuDyIrF~Q}`U;u8$PiKd)f>JN5LuDh+PVtzc3-_yO2l#E
zCdk+F1{0^0N{0c?Hr0Dn8C`{zFGhxjjH#h;ICxFm`cX91v0M+sllMD6^+^P2DsFxx
zA_O+KPg*&_ildbzxrX}pvo_v17m1er$%*v@3<0{5XYg?s6u2HKyxO}X7lp-~=4$O>
zxrsbs6FJAvfB9S9`Vyz1B~3XEL<HN?jhBu-8&KntRJ`FK-+F;Km{Kib67B6J-DECW
z*-oBXj5d*6wB5NX>>oPx<3mjJWW80pgLHLOc#!30cfA#Sk(f^6*8Ehpdz-5FNn}wN
zo75w1tg48y=&R+b9K;R%X-?jouGi@%CDf(Rs0ZlxDF<IZk*Ksps@6=Afx@xF)j<Rp
zFmcljz5DI{>R%DEgm&Jqa9%J$vwD@(ucH{mw`Ok;WhuHb;xtR6yzXXupc{!UUY76e
zPCRq)F%p_!<(zcoDQvd8GfKN?_hDuheWTzzW}|vc;-z{D!K{VBTf@gm-BZn+Z5YoR
zCT&;x1{AZ{d*0CvNQTK#XVZbj=CFtPzGUyLD^d{&oyZqqX;NaZvWME-tfqSFxYLXQ
zld7!Fh-<#9dz8))QrAy8u#i}_A3xKpARYG^jw6Oa^=D&E*f+HWlx04LDF+5f6F|j6
zc)qvU*xDO$8d|kWuD<prdv}em#s|8LQA(LRLJ1wb$YF-UwGU^17v}P~R1$P%9ay^X
z#zu+IB1U=2-c#eA3Z1N|r;FYjM*ChIBLid1_4VF16jU-$Q3m*x11n9407%w=>IBF$
z%<!Q^?zeAF_3!re&aa{YrB7##Xc?mp+qXOelfgyI_fzk2Fu54EA}OG*4coW2Yh|?K
zDLRoc7`B6n$?HU0=J(sWuZ%<kR)DOa(5-{1wUJ5oIw0tM-M5GbC`6T1ALvdA)||WG
z;0MS<!5w?;{q=^noF%(JrFbg9#MOA(q^`~%Bt`VwTYn;CvuRDouyKPC9bl#+9`!pK
z9jslk+`GoI*>U^MHXzGq$H{La?HEW5YpYrm`mjFHKG4`9JFI*SI}n`H0;R3!SWP!q
z#RTdD*t^leV5d;NTu%&*eXwo0WcGyS2OwTZjXuR{ve`=9m#N^9T7xJa_$6uObp!%Q
zzDk-niaGW}bCq;AHB0D=vLV??0G7~|%!f&(H~8*j<9otBoJ*Bc(5E)`hn;4Pu5{r&
z()8@Ry7|Eoi0*a<Sq@l=Oe};^jJV%j*T^7nghw+p4lJ4A-PD5mp2iY}N$j3<n$@jE
z&Zd+8l=vXq?DuvwqtJV;Ohr^*>|{ysdN$E`y&3yRy<E}`=^&~P;lQ+XCmx)zj}}@R
zfy|g>VmdJO9|duAVwkMxk2L~VJwc|M@+yF&9>M{hQ>wZ7efh+@m%YyG=lb4y`~Ef^
zY)$gA>R4vY`9TL$OE|L!J}y4aexSM)%2eEQz=6f&Le}i;dXB6xI-KMO+kn-ckL1d*
zNQ4We#2bEJu(FI63WbaDP*tQy+xWw!NRXu|zTO|_Afgd_tX`XXm|h#sa`+IKpa*dT
z&fk_yZ*#Q(283W%wL&&S7h*|B9^k6Dlc%8_a)K+IU|YfzNsNM##eWK>y_MO|p!Mzb
zvDp(@N-pr3i$w##*W#3;PpOb-R_KsBZAVVh$mp2!Q;dFk6p4Okf*+XO?EFls+|YZ#
zcC}F%=}YXT8P83HP~Ti><F>mSA?Z@WloQ0k#CNLyDlQ_=>7If~Jl7N0&b(mQ31^@>
zj%FP0+unL=i0NEM0o{CpgGmm(2}@T<z5D7~Nv*1^lC`D8&B0`{Eyqd`ATShGjQx##
z*Aw?9k)!~M&hJGL^8+==;ko63D!?iStb73Yg&e%bk5rf;JJ?<!2gu|!n>xG#V0^g}
z8Q}}zbUKuDh0p8*^E;eYV}ijdHCvn(v7i{9T?;$xa6Lg^L-w;gI1WskD#bD;D-Xkt
zw80ocF(e9e%7K)T0B`*C##uXBPf|d4RM15a045f&jQFNwxm`{@;QWdy4htz?x>&vB
zcs&B5V`GV9tG&TOfn|a?G}2DAC%~dLR>u_g8li9vlLsCvbK!Vm>|}QjLKF8WSRk(Q
zhN;tD7=HzwNrXGqCL94;UVtonPMnp43HMgHA0@o;OfhWvP@7nBy6%U#kcd-^4NI_t
zNs{FY6#Gh~vOV5?{e@Ue*aE!4G{Fka*|u=To~WntB#auh)yCI@$%VC*;`wKqXvD$q
z#LIIwwgXCXQ7#U5syOJ`w*HjjywhpMCFvkn6`~_{u3)=RuO;756V`#{B!;w-`dP|S
z=i5s#lG8opuagce_ZnGcn8Zv)4&4IQ__$q2@>)*dED8K_@BtC&BRx*q#qM1y#3hdO
z1f6F)cQ84OD$>a?A~oBkdRGQ*%7G)Zz490(P3>~<(Q|PS1$mQqs%K*T_@R0llOHZJ
zIaYv)ORrq9YQ>%F?1#IaoR7$2X|zdrVA9_iW{Ip7fU|{RhLb@YdfQQ(ufApRn&kBv
zOr6CW!6(mAR4Pejvqbx}aSj7O5L%po>|k=TXU4}+b=yuSxm_QiFfpCUCy8f|H+<9;
z0#BJg@EFtM+OzgliR9NTxyJcjRHFF#9Mm5vm}Gtl2-qRKgN0(2g3{C7y(;;o6>-sl
z6-$?L6<N|?_AuZ-+F`WbBaABc%54rN`asd*+>$VC{#d~@l-hELWOiUF{}Yv^BL%R2
z{dhf8<;_%u^&E4F`<#m#rbT`s25J#39I~INNAG|q?ji&eL$3o9@4Eh`8|^1^jjTuh
z^@T%DGdkB;^+Bw0Kh<XA&b!t)jUe|ld1%b!hUmm?`{_2RFl?#UiX;z?<Y3~_Drj?9
zL_WXYex|$nAbbI$VOEmpz+6Tb`N;0DEbxm7rQ(c=3L`7ies*!Jr7FZzWMZI>ZE|wT
zr670dXyTu%OBoxCP742SRNgS!RZ$tA(%HUIXUBfN?n&E5N(~Xo;SG~PaEhpr?!o%8
zU&t}*A;ir(jU3d_DrK@bX}{RP<hc!_xi?tot9&^;(0-||!`CbkiC0{3V4A+W?p$lX
zoNIVZzbCniRtklzB@MfDHdzp4!NOE#ztRo2wg_NcmBS4tr&V%JEG*V_CO5KO9JgO>
z#E7e*YU&h-sGEq*K?<;6>!uhzBe=s+UxX>DLRQnG_Um<>gvTc!_tOj89avgiMhILV
zv)`yoBvzlJLr1V!0!tRMrcbFW8J~W$x}7y5lG?N*X??2V0Qg)(Q4mte?DkuAi#9V3
z-;BRvDjE+tB*EuPrtk*b--A#6b~2D#FkU*p#WPps=?%ZLNO<u|MP*9^xa43uV%NFu
zirf6%CcX5>@1J#AwbCLTi7Eg`?f3GE&178;^_^CcO@_cxgu{NnU@Bgtm~sVA0?UzG
zH$lpc{XtzK^T1Jf(tegTuvDn<R7?!d+8?&AU`*TZv@-fmCEOthna;%<6&~Ap#_f-i
z>V`JXIzNl;$-a}@bTDT7<Mwh+EKtCwtUSPSheUP48WhYzS5skwUF*F4N#3(CH$*rR
zWg;bl0n3TIheBh@37oP&tw--bs@f?=h86^@I&vX=ENQYuR<S?JNkjeZJHKnFWh~Sd
zea!y+889vSGY%}LerRGdOqX~5MPq%-SKQ__;$=-&I1y%R7{C2xzHDlux5fFL!sFfJ
zqv109s~l?tNo9=Ciif0^t%>HyEcVw8qb!T)rRwKIplRzX3zo%+q2iMLO~Rz?+{Y;}
z%HMNf@oYIyB?_msEoaf+7B_`oL#Pol%na+CddY0tHH9Jei*&g*Mhh6SXB-@Fe^;=`
zjfty|I<Qg<w!i`=CL=rJ?`y6}Y(z5nguf0xhnQ#vjzMBQ{zE>Ra`dxf6GO@*n{r^f
zYM2spkc50Q|JVj6q$~|fGWu-}CKR!am}&p1>t@zMQ>a6O;ZIjki|k+a&)s#>EyU3;
zIxwy5g*gIJHrv0{`(#g(=Rky$oDxh)Lpo+!_i7(G%2HFW#B~+7`LB(Br}}94QAz=j
zmg$`n(|68L$LZhN6-%cH23e0S-2ug;d1yFF&j%CL3y>7pl%nRe{d-%>vMo@Pk9Luz
z;$t#WEC$H_qdrpvD;KJcHf=EN+~HP*?fhq5Cn+UXK8v=&l>bHaR%`<Q)!w|hkFSSl
z)OH{KeR=%#;s0GQ)*xIY6mADrltNwV{>F(h1^oO^QA!?Rtk`?`02O$&mj~@~@Ii)B
zmE>We%#qo)qMps8B>d^)f#@pSAv(ie2Tu0wleUHQxKSs0IgNNsZrf7!kJ%@0t2I~~
z<fW%1p*@a&2~2LF17f^pqq#OerJg2}(J6K%9kT^2+1sKNM0YcFmQl7(-B#>*f*Q>C
zJI&~jV)?a5sjOw6)*cJj%%ai<9hf%xJ65f>Pv6$K_b5T%eW#u4JZeuVFkOBc5%#m%
zy9f^MO*zf#noj$|Qh)^?<{*U)JLoa{jCzoyom>a94Qe%jQL4k0FSgIjek{Fnsq<4y
zZKVeJvvQD!I2v}?X+(!K$XW-A(-2G9YoFZ>e}<VUsS{XStuCuqa)ZX&vd?Mb!0vE~
zJ}ULF!6d<`h?9=-{@l8bG;Z`T$-HMASZ;BOts2^7pVtkNHnptiB-VMr!NjN_!iE-y
zcbGoQu@hEi`}{@(5Q{u>N}>)2Eq#SiJNP0g`I{us9JeoM1HXle-4jkTw=Y?tkRvb;
zf8oGaYZr_&iIF~zIhc+_Wssvxm3>inwPZMbwXy(I(h`g&A>2}5++8KzGtXj{{PDPh
zDMFT4uEtK|R3rWD0m@K5Lk3D~CM3TebMVn8)sK@&0b$#o(>=3eMo<$*3-CJ-&%o!-
zw@w-h;^3FGhtMU3VVs!X?!b~H;Sb^&as;L=BXZ;IOADXtH4VP$&I$p;!RJd>jSg=`
zkF+mKp0HLS_PiLm2}oXYzfLeiZF+9fDzd=nu+zw?iRkAIh_m+Ptxp3hIJj93mUsBb
z4HiF8JR?ksfm7<<SJeI5SF;e5Hm7uRAy_*2T=G<G(*&?t*jE-z2~XO8mFx(t2(vqN
zm$8Y8KvGBStJ-u!m>Tr%R7USG!O=1jGRVQcI>}{W?($c8{SNl~HEodA;Q5|}sQ}1{
z2|Zjk%7=M*cjkqCZQ;fGWJx{6N+0-O2cI8Vx(BHo_Php@$D4aNKFxs@Wu{5{AlgY-
z5TGTw@SmUSc<I#Cj0&PSZSlbyIRNAZ_5CqgxmdaN0&x)^kD>memKb{0Wo^3q4{+8j
zg%0`-DCuL(=rAWhj@b)aHx+oqT<m920t+UU_AODPSfg?XUbL<G5p_VVFq(@lu9d60
zvVLF}6$|dm*WQZmLfW}~U3)E-s6%i&BJdoTEFUC{3{TtT`RTWgPCGv(oiK(+2tvG~
z?i6p5=T#*Dfu+r4_1fijWnG|!EggUo;@)6+G?@l_yS=!vt)=7R{Rf<8aiZ`QvtUoy
zOWL3lYOQ_aEaa+XdECL&Oe#yS9vdn|a9ew6;aMc26<(<H?6fXXG|?Nr4e{Av5-R-`
zdxBk6y^(c4Ww3o-FmwfE&OBtwUbZd8rX-6Ta;eblu6T;7gq?F$(l<6MNg9N?!ZD-p
zJ^kBlS2x*$IwYJfN8KGDDYjs?L$RW^$Hu-b$%znKr{*Os4f-@iP&e5%)nCLcMS{-n
zz?%;&xy>zTjO!<Qlf!mx*O5$`&a=C<mgzbY+B5OV2b+BsTdQ4{?_7&mxzTCp`Mu5S
zM@K5c@cQf_BGG-l2b@;Wac*Gww5Zu{H#CCTbo+x&quY~Zh*=E%xUs%Ux|&jq$fI^(
zNo&{Nc(vX1zgj71D!Uv^>6MX`7>p&A#Zdfs802d=Z+rU9e8R%}7v}!I`OkSnecTZ-
zHI|U_QsCnD@)UDkH|hLJ@t>fXDSJh7&EELznDaXpg?f7yGM8BM%8s;C&6P_Fn`6l6
z#L+nTQYN4!dsX-TW3i2^L{7Zl!6Z>K;Hu|wNt$2yc~hExb-Ob-e&kiKgs80PU~Xm9
zJ|Tql{}sT#zI`L=m7MJq-*aHmw_H%-B#Jms$1S^+y{4P@({fkE>}o$h>R_b45qdpN
zk_8oWZE0PL%ETBd%TIyK7h!l{QXw23okFHZ)w?CfUdMAezoj$jcn=8_%NpH@CDMrg
zX<k&ZBqeTbC+fu!TV8mTG?G4jF$~t?$H5#Nyoi|C3d~I_x{o;&*7h#n(!tc`O>!vw
zkAgFHYX?)Ho+Rd74op*y>42bMyR93%ZsX9DOU9Jd%@^@1?8Nd$8Plpe!5eg(+U<3J
z)M%r_N?9#W#8VG1>~QcY?LLY55WCPFjaL!rJ&f<OH%|B)Ky973P`>T8ZS$EP9oeYJ
zZU9Bq5|6{wXRqr9A;Zuj|0slOFnZEMm%YMPwr{U=0_0N%mJ5uXNJ5=78-48|Th)EX
zb}c>BDn-<$r*7x`9tWS(PgFr~i;{irT)ZRfuo36ZeTu~+$iTrz$Hnbs-4si_Yg+}h
z$tcg8&nhM?s&Yy~(Zb52hh2dj=Z@sd<gm+<h?Kax4biemPI+PY0BA;`Uld7ueX)e0
zp;52jq^7O5HElc)2t^CL!-3`UUvks+SKHb~{4gH*ctnCJ@PQEtPHmOKvaWm3c1S`d
z<x37MyE!_(fg?&(4(RTW+wDYOd(fMek}CN<YteI+dVt?+FXx)ylaF4QL)^ho6v3?p
z<E|(Nz|0-CH)MZarGH+Hg}Li)yw={>28VjgA(jD%)i=0#D?Wh`Gd7BI_NF!jcduDF
z<+N^4cO<hajZPV1m4F|lJy7gXTk|ud=+pM*ZUC`RFqvnh5~0c+cQ8^@$uoIN{@S>~
z`6UJE$&X+qySMdX>DC_gZ5w?P|DfPhY->TL2kouh*F<bs7`}r;(_<Ntz^ClS@`hpb
z?Hjr-X=@V;iX@2n{oXK&85P@6XqndU5qn#`M6z|u1;xC^{Do)1!6&DI51fQTV3PKY
zjR3-mM9R^S0}@YMm|NdhO{!k^)JoW-KZAiH;FD8K_I0eXt#7w48|%#sJI#_06g@}f
zUbunI_C04qtvj^j5K-JqgY9GVGnnw44c8O+;Na?u!Bk+xHQw>i71l4R9TbjJUQ2F}
zBfwQ{1W>NDmM+<bg0-As9h3nqI<4=Flsl+DS=dNjO?GZXjL57DtgcqwSCzj|QZOH6
z8{1ec?@z9ub(-;dImu*UCZyg*Yk0G>xp*ra1qQdn!Rs5ZzUc<LuWm1SHxv_Kxxvfu
z#9N~P0S<3PsEc!LYW$B=6?FLSumibF<eM96ic9(O0b_iUrH$37>Wx)LCOM156ww$<
z>6~*_a%XAryjR<GJkd56rX^M?_A@XJCK-=T&BJcR<~v@W7sdcOaAba)55Q3Hw4!#}
zL|b#?oP*)Cg2inl57oBRXvnIdZCaTJQj|qXN>>LT6QHzCX%r;A+S~JS2;UQ!^Jo>h
zo<tmp6N0#xBv6~oiMaQnmpDJ`gJnEX3?1r&^l_VNF9`cy8jgTzfskB2id0XmZ+Ew%
zH?)n(Oov0&!9@3>mb~on^A~NVV7g8bT5NC*EMxH}`y-enCr2MlDd_%sGNi}nXG{hZ
zh0G78EWfI9A={}e4BMX>17Tac_wQ$`*#niHabRVR?);&AYlb3p2U^+#_2#f32Zz-L
z`8aQG7|BPPj1>G8W;@%xO*O$}Ryywu7Ey-uP=$aWOn!U2jz;5mjnDzUaGzYt_KwDY
zw#<yF2rccBD|G;K`0$y?#l-1%c0`irL+F%;5SsX6LOdOOGPHFhwaxVHp`@^dxpf;S
zokk2tu9$SPP{Jm7tfH#4M_#aZwKayItAH3KSLdfo(UJ{J)1a3gb*|0Vt)`42uluHF
zykQ_nj@5<`%Du9maaA$tnK6R!y)7p;U}|=vyE>qh5mwx@;!b;a+h8zWZpOUs0FrT}
zQ#p*E8Igmrg!qDoyLrDP`va`9(JI0~I7d*VovXy1ZYJ~iJ^4QIou}U8v|^%?yk+vZ
zVDBxM{KoR?!GU}c7OPZ*W#3%HB#mD2CrT3*Q#j;c;;Ayd<Y_ji?OW<YlE)A(5H{D7
zlF%a03w$irv=691Xy@91?OW^46z##VC<cUm2OpeLOu2lCkW&}5$igQYEX4);wr)68
zUo!nUQVf)hp;SHa(R10T%57oqYrGjJ>5eBKK^y}jW~Ku8)As&0Ww)b!6X)mvG}t%X
ze7${pJ&w7Et?3Zk5U`xl0X;Y)d-ZAO>^lmdF~OgZnBd@R%H;f#9(v5avkf>Vk~9+^
z^$x})D6AFpwSA!V2eXI_$)vM5u;>re64t~pF+;p*E<Rx&tb1_BgjBZ=ZLPsFNJ5i!
z&>m^eo+mJuXDf*TSaIMk`HUXxtfmir-0~4`DWYIgP#C(B3Asv(j#8+zTXjI*huiv@
z*@BWN<FW&im^Up0-go8isp-wr&hHu~ndjnh_T2^3x|aSyP_+Zg^|yXzV%EN=F_D?c
zQKu2TuiZY3ph152z4ai4%oBhUbq;`~3Nu0xXB@FZ_L27W@IU;u;e8XIKk5yG`Bb$X
zfMFJmIFFCEdxCfG$^*?q#s!n43Efn<hJ7qMzhxtJEC?p9D_AJ8uUueOzOUVdXIrv1
zQ8eCQs`#mb`jl0+?@ubeDoz-@bj4K;V330j7&EocN`FzN;14vexSqO&l&hs(6Mh8F
z$u$V-DdGHJH<;Q;D}p=B-C!C))146L?a^FA%O;jOKLefPC(qqkd#qq#17X+L#-aok
zbc`z^f}ZkRIOTnWsoLh+IwjR_g5B5wB;gn53jbnMpNSBA+g5mqIFdpnNo82st_1>+
z9DH&t3^8o^v6*Y=_uBTx^@+zBo^_h>BBTO_Lwb$aO&%n!koq0Pz*NLAt$&$;l^%NO
zCUFe3*s*hM8Y6ERotPRzQc?2i&QyBtn{|F^H7r>|$-rH0>Wj!o^6~~t&P*o=0o~mO
zGd$yBD0>}PDLTE$A%doCPhF(G=v<gRguXV|?r<B0ifwzlhiwmOfs=Y1m^u|T0QYrc
zp^6G2IY1!!o6W+(=W<-f`N+1vJ&gq`Ne7B{V6~-Aj!tc|1NA`p4kC6O*>xaM!W+J6
z9O0)&PA=bt9jsR<+=T>8!eq`nu>5+2nK1?a9!hdtzkbU38RVFI2RjsO^>BkpxQyj|
z#(@=eM07S}D^W|X@gq4L3v;*J>ipJDvM{&u_GNao9wSvqk#Q6Tbzt$BmLU@EQz;R4
ztRv6PNXx6GY-WZ%__z~}^_2YDkpclHIbg>J?rhdZW)JcXo!b}jS-D3}!5ls<+p~6}
z?oZ)LqLe;bZ&tL`B?q5#xEkBEovg1?plziN`65ilG?~42+NqCYID;u0i#q~WX-9bS
zW9(laWfV@=SE_#s<HG1pFBHH9CRSa_y6O1UCf`hJk9T<GOvFS+|HDblN&}xOjl5v8
zt`Nwu4?eLd(oal|aO7Ev^bWiYb_5A5niS47+Q=;xQF*%qOZvKl@(R82X*-)E7QFbp
z(`paX%&~9CYf6m|J6D(xjMQ5+?Q~{0OqdPD@nR2F1l0L_S)Tyu8Ht<@pxBYZL4rGr
zyCFDA_S$0?>LoJHORGK~8ng_$U{V_-`fnVoE>_x$O@x*#!D9}fV^u9cb~W+d3*0Ua
ze9^g9OGi!{Sx@l6P1&WK&%4%7^(V1NIc-^T?i#iqs%|8eDJ=(|DV;g%U~<un&y3p-
z*JV<<6y1M>m94e8&%xv<FXs#voq9dr7T%hb&hHpweSWjG$D8cQ!mxQEB4QwjEsV$;
z2a`Kde3L?-_t-+CZlovC-^uwS4lFm80f|cQL4Ra@Y_q58p5USk`Z2}^So9|Y5mk&s
z4lC7E`jPzFsTuh!X=Kp}<j`Ztqr3S3llF9e*t_n1*S*f~;@l)HP|3i4w5>+A_KQyI
z*)azQO#LuYj}s6Zs)+qq*Ke7j^)0CasZKsV4F@mIJ2kEH<kR-!`PM5Qy6iINcgVcg
zUSmH|Uy=r$iXvn}_BfarKXztT?`}5#6ZVsBmEp)EgCxQv9AJu4(-ek|sEMCyG#j_3
zQ=Y*T%aV@Z;8XpD!C}}}>0*Yj%VFgMh>*0O&Y`$fS!k0^ORQv+H2-mhviwZFW59|R
zq1tTLA%KC8?t@2n(eUiDpDh@Sg{7(u>Z}8c=?~8zy~xn#)HhT<`seb!3v*j0Mx_YT
zj;YTBAvnDWT~vzjCHwgt@Kxl7ku&c!i!TsrQoJtKm;FMU^etmO4riWn0MUos2}?G~
z%6ju>sqpfPT^}U^68BS4qzb|~gykB%w(Q#_RsT}{CW3Ox`JGE|BhHOqZXaK2GQxxd
zi$01UE9)bIb;3UIEA47&2J~LX3j$LWb*_uykqWkTdcuCSJp|2rHylW|?O*^7Cd<YU
zJ39(}XIu?d`(JBMCs+uZNlaLSarz`$QM}Q9eQ^xMA?X>`PUNn$4koEHeE3AWk+ks}
z**^^C?PMw<P^TQQMrR6Fh5csZi7Vw%QSAT+6oaphM}<fwW`uVt#>0Lq7t2i~BJfjB
zTD^4H3j6K4Eg9Z`DI`YVkb}={b-1^A%6_LCqcrlQuZ6AJ!ITKq8-LJ#xBFn3SyD%q
z<7>Nvi4Sgrk3ed_*I=PPYsii`FmV)&k`VX~4A{Tly;?jUiI7P`vzXuMVCs<vFhpkL
z$se?FWfoMGQiV@}B^&F_5PrkLA5v)4AJzb9$5Z%Qcp4#B#l;hT>EPq(Nf5Z(<`3H+
zEq)#gPze!5bq*#;b?u!??T=f7`Z!J_2WUBprTs}mt+`{h(<nB)7%djoTKm)X8CoCO
z;vjZl#prdKV>%ddIwI1D-yE|)YqX&-bpjPBb^@qgunn3VJevK_8(Y}iBWrKmX~&g$
zf~F*wGRoRtH1>!NJU#<OI*?rM(DO4+ZD71PyTSglQQUQ2KgsyBzp6K6jWgLc^%2|Q
z)L>$yMYSz|=C8YPTCvh;)CH-)!rtv~x)G9>QgXu%Z!n!qBwyfpbjH8U>lWrR^^Rn_
zSqGL=LcGtk{ax2FIg&KpG6-NNcQEl@giba{F~2kR_jR?z4lQ%#7eq~R2a|hYZ}`3m
z;yLUelA$ilEmfqn)6BP0E_2rYu^Yqz$5V*SVMz6cmkJ~bkc&tQ8kav>-$nbUHbBFZ
zeW#I+7Duput}a$t3|WWuS@UNdOw3%0VY2MWE|a>UA`JeeK3D^Yu*MoX=?$ierrPF^
zQ|w<mQd~T1Cgu`648j{;C-F9A6)}mx%PcqaZ*{+H0t&Ch6v&!Z_>V$8b~*T@+o<yi
zEVeuO|Np(w?MRO@Uc*j1uGB(Mpf#tH$Hbqv|0s;y6sJQbl&mhT7K|hd$%WWO?LW6Y
zeG5%)3pYJomMuY1)uCU9e77eb%J0bZmte(j<X9ew_@7wkNurm9Ih8o&xHaW*{8#ea
zCqF#D!}+ag;!a!y_TPzFm|OmO=XX$A?%(@hKqu{gYLJovX#yq2EZRN6r`)oP*m1;)
zDYNahP7ouXV5k+|2Tbq0<BrXQHrOX^Pst^H`wAeF=RC9O7K>bSRdO2WURki`nfOQ#
z3rXRuee(7kboETMsT|+%iZ~bv)GvRfeM-J}VUA$O5vLX3m#65Z<o~Jl7EQmjRRyTX
z#KIwR@UtThKIQ+-vmAO7)#N{IdwufS*>Q@%Z|*ygnE%|y2-<?i#JzUer?<xs7?Th^
z!YhNQ=HN9jlsD19JgYj$5{rqKC85Ds_+isj?X%C=9%39NaC<nZ#+<hJ{?aX55DPxD
z^>~`hs4WvRxC0Wc-8DA5c?(9m&#LjyrTL66UtEUHRWVwcb5-wFV9959uru=bGjlr}
zST5(%LsVs@#SF{O_c^Vzy&;7mI_)Y(A|;gST~axaEHkS?aSaCp&QZg$0`mNV^Y*#<
zkqdKMC$>6`=v=uK<-xJf%drfNX`j<7V%`zz5xTW~eh1qtMkYq*z=|6Uoa4#DlbT2X
zuzkVy5LXuFp7%WGw`Pm#!8&&N3%92NaLvumPsudvC~aq7RK$QGQ-;#;dy1i)bTGN{
z*hzQV7q@X?lEw-6p&cE`m9`s`XS(Ra%R9f{o?UoxUS({>Vd;yNqrKq=PMYK@smtL-
zu;;Y*5sNp)woE%$?(&R`n6)p-E2z{EhhjNx4dKEZJ7<^!P+q^Z8!q{CQ<w}FVXbuV
z8VL^loUkuzBawz(R+z#?WvFPFFjE2xbsur+v^}@(5s#5^MaGMqC@_s4S?_5S{qln4
z&ZLO!6AmnyF&jtoJBOFpzM^56*xKTuFl|X-iq{;CuOg#Z7)+C~r0pvQ<|e;d#kVd*
z&YL1o!B^j8Indl~x38*u^mU77f@2UPA_f}(_}r<{#E49pvaeqJbUL175b+2OCc38u
zr;<%9EB<Onjos6}CMWL6M-UwznLp*ULp79ah`XKm1u`}yPjdZH?WuBv+t(&JuED85
z;t?(Jh=sW|y)C1|_Pn;*IOoY}<XVzWt0|m6Z_jV!w2Xa*cu?ALjS8p89HDdIU(lX;
z2QjvjPP1l4VZxE)kft2FtT0qCh;5QMFkXXY(ZR%HCHGXYvAwVjM|DRuUJ^ZKwX+Uh
zH2AX1FMp}MsOx5rlF+kGThCjV<B$rXfg-o~>)K4bp*KdY*%1ejCtWF3Yr_Uab&Srl
zc6mN4f~0gA?PMJ=ckN&>d^p*u*hh$>vnyIZHDF9A=QlC;CnHu$fsEj#jjuw#g8NU|
zm2E|?#Xv?7;DiH-5fhc_C??eqQsl>JdvQlBDu*knL0MWVy4)~nP!SnrMA|`nNjHrM
zKs@a{S2q|(?J?$A8Q`VcTLQ9*)Jc6wLJLntC_Y`7w{vZd=N+6($#9N7*te_NcTli^
zT95r<2a+=%oIo5*$pDJGdRd#Fjo6*JlD2pxA-v?m+SRRf8^VRts;4MCA?qD!862}E
z?Ts?2kQy=NK!R?x%VTtBZcf=X?Onai<Pr@#?YIVON(h?UsQ+|ro3l<ERCWounp(AF
zdVr5U<l7||OZieEnd?&Oc-=DRS6q|gK>AM<M6koIZ#`Jqo27=F)2@Y`$sie@tzg@*
z@z@R3Lmkf{+B%6q0IaC)gnYQDKl1fug%i|LyRqGaK~1ssS`ol<aTAt`BmbssPGKHZ
zMIM@xmo9&$-CR9Ss!5-rY)ad`VOoHN;szw2p0<})f4Bz$xp+AoP-3>|vEen}A!dil
zrT$X!V19?aqTX3E1-(iPrzV7h$<aZFVlPm_d$Gkh3CCX9MsryR(M~(YUHvMlA+A4Y
zuiD<CB&AYjA5+p!pGHa4AxleRJMY{P@0Jv(5E>bEVjtzV4mEytzFXDvhtTGo_JB1`
z6j7>c^gwPcf;sH#+uV@)60!bJ6tH*`@0BE^=q%oW8I<vhVB=f^#ATva`D@xYuct5?
zWwDnw-bMcKIBdfK$Sir&H8<ET$p@DcJ?Z>T)4+8%+-%G0mAR87nTA#(LSo>-$HI^~
z&x6Fa+wwL|^HeJV7Kb&!_#SsKA<*rC9k+TaVpvgEPFr<EP9ya){W&#w%Gp6yWKSPt
zgW+d79-?j<S(4HqZ>?THfDDXCa-T24<ZRW#ykxi4b=pt}-ca0u151J@Jz;prZm$p4
zYLSsyKC3vzBIm`y#|JAx2?M_V+8ym4teH|!<8SOZIG>O+8a}lNxtZCN*>ELqU0@x)
zHlImdBDz4BD@WKnFzrnE!QNo6Ycui&PV%FXY;VcUS6^o<yRnY)+Kilh<)n2um>ei}
zCY>8E$-!2&%R?Fr`s8w=K^#ncmkiIi<Du+Rc4xjfj{k5PDFpVmlAxhczbgrh!!>kh
z-mKHs8>v==?{y2|aMCAN=hRJN802R%QAl2cdYs_G*B2~pDJ)#PoDHUV6ivvtv!>k(
zwLC$W4-%ZvU~Hz{5?EUwn5Fi_C=^2`#S%;kQRKgAIBZ>AT5(MA;FKT1L`PLfLaFeI
zu)8~9vdd6AYI<T6GX^m2%BapLS7=N3)DzWlnKmQjpOM_<U~<xs2)5cA+I(Jj{Sv1U
zi&`U)>tH{_-<=j<Z>-LUf~9ckoNi!o4<fJZ5s0ty|EjcKuqt~~Q|w50VXpXI2PArW
z!|hAg*_)H(*W%(BAD;4RMiBwbKyS(3p=XlbLnBuQG6ovUy=@S6<h2S=0f?c}Cu_|T
zPGWB@SQyK53Hc(7nF9usv~O6vmrY(zxahz%qziLVoomY8)?T<C`D=pR8bFfu|HIyU
zN9TPUcfXU~tJ8aI#j<2uRv{(Ii5ycTCD9@&nxrHbxjX<Kk`O@v#sf%%WI1EWP8=t3
zNpX5Ykp#hBR9HljN=PtH@4ffld(ZuRXJ&tZ{J#I(bI*IuJ?CEW2<O?^U)kU6?CkXA
z-0iX6&;$q5^zE$;y_ep7`BA5pq#TwM1-YKGUWesl(2q>F!dq02(qE_l1@T_W536e5
z_%;;K?QzcXQ1NR9$gXt&TLnkC;DO*VOo@)tK!S6wikF5hfVn?<@q!I^1Cbpt%UJ<S
zG8|BTO{VIR4hs~Fgile($QvwjkQnEaL}m8t3&a<j%CW@A;q=D<<`=L+bB{`aZ|zpY
z<K+5f=1i^zdSVmeX>dZ}USemLhgTcQw%QEoN0D4=bca*JRp;>}Xf<}JuFf54yB29_
zpNY(t?F|v|cNr^H*$S9Mi7Q`Wqe+@@sB!0aQ1%i<FDQsIGm3Ap`#UUXGIBpkvsYw{
zI{k@WPO1+{c}KlkE(_^`CrMy$FoH+5$Fo75va$B!DDRFuLXWmMoc9XQboz6CL71W=
zbR_T2ZgGt5zyn1y#$H+%Vw?ihrAx^jcEHBlY(&X)2`L$K^o-M!4#JylVv&c0nZL_i
z@Ya)ZRr#)vX`i&o_KkWCqdswQ-V3&%96-`pW+?w<Z>#IHOl3Ky2$TcM1=m-`>%gXR
zD66)-&iR=X#a5^o)N>0@*`T2b5lzV^8TBS7iz0RRI{nG$R*#O1+jM<QdfZjROiBW7
zbf+hcWt`F~FDB_vGi}G)I_dn*$E@$M(L7LmcIRl;(W6~+p#vA9)}cL^cgze;Oi#}^
ztr~jN@L{8W!gdvm6)eqL)yW)~5Wi{VZ8qBwv2vIr7Y0tdX0uN1V=K@+KF*g?yVD-Z
z$3-DUxg_`s2Nw5eB}ow?T!z1$3Dh25oLn#}L0|D5tU9Mh1G(yvH{09W*tKnih2We6
zD|gMB&GtxLko+r36*#bBEB)S91p~gLVesD3t$n8z_z>fZ;&((geO>YSVd_mdjko~W
zE<_j|ir^jWB-Vo99u6#(8aBlioh?ym-rm_!#?ukx@ezrtPJg_CZ4nV&!L&#0(KhNk
zW+=eQ@&}OS2sBqQIkCD(mrE|oZRiR1SPeqtr|^VG$B<9S>52QPV;YteZHY~2Rp@u#
z|E})C(Dh(zoCD#&;=_{V$_l3){1DsJyW5Aws*&43Sr?siz*~QEiwr)AQXovQq`R+g
zg3}1nHM`Y_7X+URc6W{t8L@9@XhncS+ljwn6?A$MZE&Fs9yx9AsjKtLLKA1Kn3A0y
z(Frx2c<47KMUS%dbw-rat~r*wDFj@5Z(DQ^J?Q+-H7Nm32k{N;n;Hn2M|FBI%>c@E
zf#Qa`9AZ=OUOpEal6`aA0ai)DCX{7LIlw>zz3DYc4k@|iynRa}ZDMAeMh+343@76i
z`|N!U7O6Ao=?0eTVX$+;-d`6;LXsgEkB<Wr1DWQ$PDH&AbY!K3=gtxg9WFIUa+O4+
zM{areKKo!&W^VvzlI6z%Byy<mnFpBLx7K_4`dX;_9a!3fD6}v>@gR1JZ4#Bgt$jmi
z1sdpK2No#Fy(S{bp=DfH_U+x(3Ru^gea3+;L10>p<pe%$-_iA?KTzQ*fncX6rb;%A
zDpC)6G8JE;HXmwOTz>5er&0V)yLrmb8B0&R0A&1~)vpH#hZ#4g?2D9H=#K$r{E9fG
z&TWvR_q&p}4UJ7XKZ^;dG>jf~!M?kVlGVa$X3fd{N5ThxS^wVprG<&*XlIu7Pth1M
z)qhXpbSsetomSyVwIXHE%%`2V@2z)Zi3z>=<i$e{EX7;N@7wMBYAESK%H_3G#-cwT
zj@PA<shFNU7Z2O_w~rvuRjvx<iRklG20#Tc>8}rubKH;$p0OV&n8NWTQYfYySnVLX
zSE`~2Al&WfJV)#Y7e4}RK`~Mk2=49?(zLE+7U@ozwcUyT9jklJPm%b{40hYc+lI$6
zH$773Xcv3P3@8}9?Md8%_2=M`v}$!`@PSH4>@AqI9c3s{Guwd$n%FWR53OP`3Nk)y
zUwbE|eqnu1yZF5jCWTx(bC!vSSk&P6b2;jb)B<&W#~6rDM_lH5R>1yx9|iCN#(_la
z7piMckJ!KfF?6ui=JPpQ;knN57$cK}AgOfn@G9XfKG0|a$%!G-j0+{}N`TCAp&-Bx
zcIXhsrN2Gt4W_x*-cGc&9cs&L*$vk_jT*mledWHJv%~Fz!Nic$i1&rem@A04VtBM8
z)gu8f7DvQG%LGMFJXHQ`(F^`(J6f=$LTwV~9au61t?1kggFZzs7^dx5`yv*h4CRD&
zdxL37#*$;e$J>&-Q;}D+IyNKggM1FTb?e<n?L<SU>W+^%P0`>^3<P#EN%QV4tDK)n
zTo`_3{5!*Vo~SR@NK{E`Z>Uk|4^B_YJE~c^4eQpEb-OZkWnrfXw*%Ac&x~-Q%!obJ
zU4a{4^oXeen8dD|Fo8Xt`pA}honJZQA%MjcRp9-pHbxm8!Lz)s9gMXI39*(>+37tW
zeh-gb*!};Gf6jPnhQeu$?4=?VT`6}vQ{P1a8s<3bX9A0RQqd6GRkD(fpAD0xoz11Z
z4sH3K(f)%@JFq5vDHLH3)G7Xd%$`XkUwhLE=co2<qSH`E&NZ^1-ZkwsG*^A30M3(s
zwjP$m%U`5c{XOKZCnUKd;b7EuzPm?xo*IVY4^)eao+2$B?iXiv+l59zk`O2j=m3Hh
zrRi2LmJ?|)Uoc<tT)S=i>=1{_4je#AB101s_Cxt18Hpa4$iPtt7I?%e2ZL0lA~_kt
z5ryQx-+nlk!*$nQf1UF?#-L9{QwqMXG3axoZ)*)wM7X2Rz06|@4lz1!J+NVU-(r=A
zx@eag1+YKKtSMS$jq+rz)FV!RplF^0Fpe-vMY1h4X1$Smn1fNLt;KTfiWPP_7t5Uk
z=U3dwkb{qMQjGmb!^vvO<#dfr6o&b5L*LfXDV?q600KLrFO^iMwj!hJYA)K3*0&yH
z-1`1>q8z?B@ATB99%9FpPIkn8tUZ2`1i^`^5eJ})@DZV3Oigdu&e@?qp6qC6Z%2QK
z6LV>b@Fpjw(c$bTQp_WSoA9xnk;D&3{A%@G_inJC%!{{RP({|Dr5=S2+fUU%n{X*9
zMxMrZIz76IxS8ASr`rHgx?o%b2ax2m{QB$dXF3^EdcuQfl@2IYmFQ8ffk>B-?qxsQ
zA)aEfF{7uv!C2d}#<NeHv!CnM@$p!fhv3e_vm|AlKu&*h(#e^J9ulAY`JCqEH(clS
z^QL9X?H6*!H{I*}k|B6zTtyNvq|VweHsV^cT#_D=Ea`8oLH;TUOv25@1NKYp&Lu*N
z$QC~e+wBuhf9Z27hnyw5P&u$r$NtVID7o&J>nVt<;VY38AgitXU^+Xt^=IspwmRg4
zSP010Y?Z%KsLUAS`0Kz*p`O`Exz&OFYBv<6I;CAFr7Wpy+V}QAm0Xp3N1Psk6)W~p
z1;qSX4U!?V_9XK?j+EVH^g8`P&k}Q$HAfjR`}L+ePHrC^!)3GGffOp`HqtuQU+lm3
z8+CuhE3yrqbzq_{f@v_4{${&?n}}J0xg78YlW|9)BgQKGt@a7A{RzX0D?#9KdfJi;
z$Brbr-!7Q0OS9x22UgpSDve|zfGo`{kJ(xKoesAI9+5)i0P4ZY?GcxI(th`O!-!La
z!7&28)1$At^zMt#_U-qQE?=|W`6c#5{FeQGdx1()p?^65Epn!9AGW?fSRAvqYzaFK
zED_7gW$ftuxb}x_i4K!GO`~RMZ=eM;gZ)u1@S&-ZDd%^5x;T~gKZQ7H)wa-@Kh7Cl
z^FVJbD^WQRJ&}O-agG_sbJy$JpVV9OBoSzh0WQJ_r(~+0V`u%-dYjZC?N*|RoUlCP
zkp9Fgi%B#`Su*>x_5}*B0i#DS8vsiq;DAamgYEX`g^UACYQ~pDa5Mx%PatwP^n1eo
zB2jb83t#lYS2>OR5aj0J84d?T@BPc1h?Se~Si9l&Eq85j09R-tv_nanhT4OQC9=t}
zEvZ-JMv#L2RlX|=!8xt?mUN<pnrMIBrsnpUF{cq1q<2``_*wg#e8j?TPL1B~v}!9>
zHcap)5U^;T``g?Zut;LJn0A`AmKUQ&Q}+40hW8k)RG&DT0k)#?KjriX+e$4uv-gwj
zWGIVR*dhD-eBJ8)V8`^tq|+3~S;HB7(<`SqWbYpuNoFVnvJ_PhFasTTdg4=)i{uXf
zj|EfEhhkU>Eb;~;xgVAoWmN2+8cN^_<6bQ!yJIMG|J<GdbCjyhVQ1c8Vc%sjKXLJ_
z{Y!nH+#I|sB>s{OrAH)covTDE1khmJvwy8;%iX51KSD;m!4~!9Bld6Y+i+|v>sd!g
z0ZaN^-RG#kar^hhRmR(r^@ta$^wg7Z2>O#0sfRHg@o2O5A4Pd$b<(y~nsKRo7p4~|
zlA;FAwK4Mtz`W=)4-z;FyZg_^StdsnuQG4|K~o1%Sfm8%Ot~<K@`;qxu>WfQDi{&9
z9l{^t0F$y(!$1<(=0@!Z`|qR)%=S}#PEV#86CFM#o_HEb<A3VrvSWm$EXdMZPi_lK
z1X;7ajbLPV(YEu9H&_aoQQyHns=F>ztmN4O7K2rq6JHocwj|(>-dodxjSW%2X{R-1
z^UV18sC~@d;L}8&a*8VrNkPaxWA?H2mR!m*172`o@s<!;YGT5yXdjo4S=hb&x@AtI
z(2YhcF2Y{iOh)*q?BnavB<_eK!lj;aU^$zhN-IOOoonV3_Leg-4g}iSo7&Nzc7V0G
z_QFo*FF#5I<|lSAlKLSynHYiPWAPH_&L<Xpf^GYg>T0n!*^0Gy0@Grc8QY3jXP;c3
zf$N&B0b?3VJD46%z?$v!#~2P^Zb-41h}ya**?h7S?aNtT`-sz@7D~)uQxgx0{eEh`
z6vBCm++bRT;ze_yCjW?iTK75%><eqe!~@H#OV?F=AS(?Ve|q&0kc)Yp&kK%?rtkD<
zonHC?vib+u89$>NGKM}zaW=G$mY=oP=@(IK<j@!U%r?ziCZ@e>JkT4l&uYlWZy0`J
zh;k98I{c}deRg-9VA>O(abWQkj2!hOy^1pUpR*U#)si;k3?qQv8%!dBI67sN>~rcm
zaXuX$Oy-j}*gWdxeoY+il+WE;0wZD8V_FRWv>K@AqEaUIdHJmqW79q7cXGyCNnW$h
zZ+C8@EHbH74j^V<OkCUqTey8ep;A;5vBJ<r5`HjGr#~D>ue`;+aBm3c8>c2lX5u6&
z2a*p&LWtyyi*xoxd&^RtCpHqVuCqTJu%_hJwOcmZ7w^q&ECV(3^kgwxoCgsXlhQp`
z>0^|hNXqM%bZC{_7;=;Yt4RpUD<~Ax#&Bg{+9v(>Th}{{*fk@5h56G8R~Cm(U)ArN
z<uA+E%?$5QZV#>6(lPfjtOx>zlPdg6B^bWEh9jX~bp>P=0+x|1JNvLBGCb`o>M0to
zf+uC(tXvROil>}@`Ks64VPBb(x~sN0n!-8@&R*C)<+jzEoJI^QH|W?%d0<z!-Ck6~
z((;oreJ@%{gK_c__|dE$uqzrzWv`)f5Y1!@5riN|=05wXf+1P3x{^MD1;SehO2*<U
zU-reV2M4w^hrJTeQI(!^dL(|ZV-EPA$ek;fAVrsIq!Ay(w@WafUt;(WK9YgGBxhxy
zI>~*fRj0qBN6W)SGGgDZs&|W^CGW`=#TiBH*YxCiSf`x$tGhXkvfS_=Y?`v77t-l3
zYF3hw%sHp+rS)t%OXP3YNouLI(32WA=Ooz6x?4HyG(LRJffb|6V%6#1z;0wOudDfX
z)*&TG2)${rZi87kY}d3WRAt4qTr22^9x+7f&Q45@Ve)!K`_KmnAaYtkCEAZM$HI2;
zSRfe5(X8Z+Yuon3ad9)mi;PYZ00S`VX5n8$rkIYjbvD)O>R}oDD49&F=Rz34P=%d-
zZ4WCsXJK??Vp5ZKeM8M^&g!EPSd|>Ek7XEkL*q<%1knt}96%y-IhCfD)n_M5NNbfG
zr%yRO@vKNe<5>i;+w#1%PuS4FX~YGT($lhq7woYc3#R3-fIZ5*I<VBjw``x-Nfe{q
zlnd^ro0dC2W8PAZPpd%anzo|79>JU)9AY@*4HiV9C5|d|b9*B~HAQ0-bmhQePS`3S
zuGsLH4TWY!xS+kV`V%JBj~KGVVnKg?I`1Y*ngFfS_SMDM61DLhJz481l%C|HTovDw
zPmcOKDXH1lBo}z?3?_h?olaAeu<EXLcWts;>aGmems*uIQlEBuw5SFg*oP?asv4K1
zEf!w{Jpt2$uvCsr*w;2z(;Gnxb(%TT%Wha^ukJoths`z(Ozy_eaGm~ILW_wMv0d28
zu1Bu;C}3Bsm>v!9zBgU<riUM~TkAU6BwO}d)PD2UBZ`*UNVaRC{e-Q`<uS?3sEmSC
zuhR}N$Gz&-RknJ`Qx&Ha=@pXyou0gGFf(Cm+NW>op?o>bB7sHgFFQ+U*yzl5TlH%T
zlNBv2#gGUmuAyh0bNaO@Jo3of?e-c#V_!moN*Ww<dh#XAm=F_zG$Cr`$KBCrGn9I^
zuZ7dzV0jLozM9z6CWEs+b*KXEc&@0Z^89#u$3pEjUBBwRNnTZ5G7gSn@JjTiUxR<e
ztFEy-^U;cnRc%A3T?G1gNYF}+k%j+BTh~NS!mCwGVzlpoBu8wW>c7p_*Nv^1UvOB^
zMrGTEg|JD9&xYlm*$g&j8`@UQxqwtj#^wqnS3Lq-C0+$N9I(6ED&xt`KA5~A>xUAz
zhvSH=GJcRUN0%PE{21BQJ=?gqqW>_#AT{YA(msKO7&x6S5y}XwXvAKdG|S+abEg$U
zfdVe$LS$NH@+Dx{Hr3=M`w)d3abR(^gf@L%JBa+4ZEm!<{+^9aBfsym+>K>slulq<
z5~B;d*SyyGoy;16T$z*9Qn$MsEX1EM@j9@0nvMcgO|Rf%GPT=1^(hA>S~;zH9!0JB
z6b5JawvmW0CU*=<Iqme+)wt<+I|XR$bqOOJiqg*GoZ^BMUohSqIeX$w_WE`&JX5m-
zA{<!J>~VGj)-l4jgarAMd4s*7?vXA=peuj(Ixx}c`n6l^jnyrZKhMi3%;F6uJDkF_
z6&`<Iy+#%QZMQm*eTf?7^rsYZ-L)&OwKvu0YxbJ2Us}SP(-Xtx30+ONC4!#4xh+W&
zr#al)X~$6e_t;z7eulGQz^S+nAeFW?tlum&^427WdsIKqX%vk~bQU>eQ2Jr(B{?Vq
z&-on_qOmsleyF>1kxK<V`eM(bP84nn8*a;OVu*~7J|$IvEFlb0_dy$JFvv*>4@BrY
zu-vHEYx%R^^xKE^h@Y9DCgi{YrFnqE%g_|I)&NtVO(UAZgDyBd;vEaS*Y<2%_c9%O
z$bNb8hyyEekt7w{-mQCWNY9_#V~Rh(GIQR!D%VwqLQxmt6*k)5kOx~Zsq+pjrg-V(
zFOdlB{<<g6J&rR*9av434%dv7(Be!x>S;o<^821+JvDZrBm3wVfvnzikB!x(@_5T6
zZ~@C=2Kz`E+~@SyXK;2a4@nK2&$OM@lMW~qfypx3N1UFR!}2Gx)jVnAITzw}lnQj3
zwE`6%2%qNRVFut8c(QDwdw6$FVn{^?eX%Fq{?hvL*viTFS_nkeR`MksSm7v-1tBX)
z8z?@Qw^fe>NcMom@D*=8^6p3lK=+^@u4Pk62UJC$<`4&3sdTK|MljFJ*fbIBV>U?D
z?KO8gzv|sOe`yK5*>rn1tV8aH^mriIIX$(o6#PK0GGfO}(c9APUltrK1uB>^LP`Ey
zl?#G(O~S#nJ<y>z<NTunv=}=#97Z5${rYspX6Q4Q#2(p$jWQUX+*^|GwP7k{Ph5%q
z7^)<9rTMcv+pdD4c*>BV;W@C3P(z}oNcS$pdin5XmtaSa%=jz*dh3yci;yF96<NS!
z2zn^LQOG975N|GkT;W6<b9&js_4+Wj@kz+Z4lF3G$)P%Ixf$?eMtOVPBWX^o?#RU_
z99Z77B<MIKYv=&$y*<*shrC{%#@cdVHQQ%p(qePY<?xKXqeDuR`9$_V>cHX+955pH
zH#>|LpXfka{RR8Fx<?UbGLRA1>A-4KidEE9DCAc>h{<6;_9kg4##0BeNrnPgpD!zT
z%EQuGDbK({d&=IKtl+NoE1jS51#;2IvIu7j8TYI`nt122Bq)%k-~d$k2d}}5YW}R*
zHQb+!!28GA>xhG+sy9Ya9Sba?i_^=Ueo^V#Eo(Nd+_Ltr4fd{Ny##g4I=_qfl_Q`J
z+-L7zM4qMx33#6aOHv>7W|W?4Ute#P;#awU8h^Z%#0#gt9;{<axr+7w4GB}|1cfp%
zTY)TK(w-?09b2Qlr+cm_QJg+w2<0%6`H}uGAhj@++%a@GI6{H8PY`|5zOil3TgN6I
z?2kBY?Iujo1N||3Z|miR^)cj4+Bpl-<QS{!+vGO*O>Kv}Yf7E~?sq`(Vd>+LP)|H;
zZs!w``$r?5R!TN94)JanFep;n;Y8bY5Q78D<wUthw$%4^PuH<%TJ@*A!Nm1t)jke`
zdVibraC8ebA|I8`2&0>~{zl-U#NZ);Mrzfz4_nQN3uKR>4Fb|g*2&HfQ~h94>Cn`S
z^HY2E5M*fI+V<s5l<sm`sba0U{Wklywx#cwaeimVYx>*l+Y2Mo(u1#{k0b}BCy~k8
zmBkY$n|()pS`J+qHBlR#^#=sAF#Xy|hx+&1hng(&hS8lvy`cyD4kYQ6Jc5ZRL7RKS
z%dYb8%==c}z1qGjN4J?A<AKwNG33@6?AX4$Mw|x9EI<yQ!XrRWN7phSQpCu<r(hZ2
z!Wv)~1Qt}Fy-s<#()~iIw(qT;rTa;2=uVHm?An`dw(qNpB<V`p%ettF;-N>=G+Mzd
z-(MFBTe9782$)z7nj}3j(p0K5>i7rRw?>c`Ws2EU8cf_rJWl78{a_<6V2e@3$|_Ia
z1^w}C>8#2|QR3z9f?<HYjL9xk@CFmlRAl>JIQirGm9yjH&QCpNVRtHHkuJWcJ_vM&
zq#p;C$UEw-M`k%0K$P|oo0#qWh~Y*lob-KbNqg&W2AvdO%F}>vWM5<B96mq{hLlh$
z$<{+tzl=aYVU|rcbX=RO{tTdwWmn-?5x3^_YYA+|Tsm(1bJ=j~Ddvz?5C?`TP?eU<
z2zs~q_VxHEQ+Tbd)9Fb@F%^d!A1G`ql!T%Vhzf@sn7AA-c^b>+!EDK{h*ghP?tYpV
zcKAaf1%;nMxfOP(%^vcyk`tdH4i8wqlFICxUT=r<gEq3wO*;+ogb|9B*pYnG%ucqd
zS*Ml!FHFE$dtAKkfE``(4iPm}<G|wTvhP44lJ9l$&#}5X(G$D71Jeo_?@!qA?lo{?
za41tqE~D0yB1hH~7BfC!Cu;M^DP39=Kb8VdIgT*IRq2V|)E_~)vy+87>6sW(LUv1q
zCd@caf1sU}x^T#zXsQZgjY{X2N&`}t8Und0rXwA=kwn6iyJLCm$tJ<!y`V}DX5j6k
zlLH5=SK)3aeaW6`^RtdKn#Y}Hu_lF5LZoC>Ki&B7^6Qs5jhs&<K=$oaT`%E7MroBj
z1Xco1q^t+SP6Cg@4|ci@;1acGomR2|&aqMMAS=_(v@7oDJ-EYZ#q&}XlE5dyUKs^?
z>LELuS8wX+5K6CB>`}y&E`E=MG)21kQW#DZ=IfN5OUa1nnkj0WZ5<<~FIO-f+6<7)
z<@TZ&(zD%6N<H{rqZ#P$NH?QwWWJ?7xSh{&LYi{pw3bdrnX?N?lsDgaz4J5ChaxBQ
zgJRv6kd{4{Gl1J}TB%SplmB6RpOg!o(7YdNj|hXE1P2MV=M<BRBzDs24{C$zaUc9}
z8*#o6p;pW1;Kfr;Pt1*|PC*OGXS>+0W@(CTKXI94SGJ7v-g*>wo(UKgscDxQ1ua{)
z!fC`qQxs7i`59ZNhsNTIsgtZmP&@|r!bS)wQ97BI9$_-!S8%Ay)z9LEAlNY&j`n!D
z=MY$&{?zv1VHH>bxA~FoeaeD~W4W0mMKREm3wI?JKAcP+Zts^#L(3}`q}+$WVVr)U
zlE^tuTDKpq=a!Qrc1Usk!%k0qwKPC>M2POk8Y+3V7`66=WIyx-4S<6nGU=O9Op~j0
zOvsOS<DY~1W1QmK!?WS7CyLGTO?q;$D*K7<OSDI)e*@+d?od@wqF;N>0HJzdKUo*2
z`jnSt&<aY=#b=%V_#9F1vpU^}b@)^56PYTN1|ZGCfhGDU@a|2J<b@q-U_af4f?y+M
z<#E&vHX%5pxZzQ6{UyGEKnWF!`w~L-+Rrq}lhwM5^CdAr0ZOj+0C8aUvqj+SQRor0
zLf%hZGHEQgaed6`!q3ljvvD45w)hofE~iI4k6k@dUw^(yj9Ys<Q=bJA*bt%>3W7*v
zzmS{TDk6-X-^CF-J)D<Hb`a5>vtO+H8xQ9UYA;wT@EBEyf2n(p2#~oEC(KnFg;83+
za-c<^p)ZznPXLJhauWCY2@ZoF?T<L^v?@Lb(zdHAtNlt+!TSCvyp%@C3lwV@wqH%G
zZjJxETCX7h8JYgq5-B|j#=$660|tobY10HO$$q^pD~buoJWo3?R-rwT&FnV{rdgNo
z1_{f774OeNky+AD!7P%34%lzDsYUoNX8`o1G<qqO;zBaAa8<l58AF@~V!u_ec6x-0
zT{oBnF=EAKTDRXWQaZ5kG#+SL$=zdNX%WdUF4NcVR1Y~}SdtG&Sz~tD0iB-w28c~g
zKt<LsWnsVDE#^2cjWv-DzY|W6n9;g^Zz?Ka{$A5edc#w*I=-v}O!bFmM#rY@_q*}Q
zVH3(b?^22rJra9vym^KF!6G`8uq{M7up$L2(QTJ|-u|%ZVz=M!{0<C?$!HhB?qq+o
z_!!oE(5?eZooyMZGJo6#10p3@MV*U42|YFa=isX8G5wQv>m)_BXw>@@^G47dcKJUo
zJT5K!BPx;BinQ+Z$F6n|_Wz{)S&sgytFC^z^E>I73iQ35_UC&uxO8a>Vx*fVxoAJo
z{t|v5uGK?_ds8F<*k9zZ7RZ(-ct?~>+1N(ftB|tr-T!5~=GNYTI58EZa9}|K;yM-L
z3&M`a{!T=-{Z%95@VX|45?RT*gu+&2%vI@=QZapmow3oGo%Yv-<TQlEWP%UrI`jk%
zMd30^Dhc#&>eEy9OeUyRaVXZCjFQUi5mt{l*9NU?W|f7mhzY4==j?BDu}^R3JhN@v
zCn-AxB(9BkblycoWDeT?E^*J1W1RRjaGJ@dsiWVwzt1_J8l~R0)2Ov2_g+wKJ>wq=
zeHpVJ5;HUrIYw}|lTLq8i_U<F1C{L`yU`PY42)_5jO#3+8T7=}lEr0&+k5e#{ZkI4
zw{6?#gwx2tpRGO0u&8?XKj-4P@+J2<KSkjyIYV&T{w2Q#i4kIOTJhlA{guLV*#5N}
z1~1}eDr+LPank9jL8mi(57YB+-4j%(UySU80}Hfd1<cYBp`Ek#@7;oc$+0NW+=VtN
z7_h^m-uh8D(7KmsQCy|CmH((GtIU$>gDI{>mP-O#+mjIk<!MtuWXL-r0Bn~3*=BJi
zF#%31@pbzY!85qB?7tF8_dRmW6<0V7sd-aKtM=dZfl1u56&`nBNfN6kCZ_Cv64h5<
z{SN16VhE*3Ni78HnC(lL2tC7d=e@x)k2C}Y`>1{KP|R&gkEzze?n|6fWgoq-5vyu5
ziVUz4BFG$fdSabs0pFZQtw06$Diiu+_7#^5m8!RmL-YV;oXuw9{x3at@sw&F!972A
zU+~+Jsh!Txn88#T_oMtB=Y7hd__%$=E?3^R&S?lwXvRJJ`0OAPk}xS6c|SqUvaE9!
zHsg`T{RxGQEUOh;o&!rR@c@VAg+UZc+$SbF;vB}kPAeWLa!lvH*fXM}eNxvW^SxG0
z+e7KeNiNnmuupC;>W%a!$wZQd=>T&dA>fE|Prb2$eagP#zC;L&PdH6|CN8<YeQFMC
zbgK%OJFOU2rY1M51ZtmsTFuK+o|w{HnkwlpO39GE3m50@)AQL2yD5NmgLkcsc%1Ds
z8WoW$s;X#*7tT4b5;f#~3pqPhQ-+b8x6j-cn&7AkPu{U*{W=Gb7#UE<UxsM=tPXqR
zU<VhK8Y2f0J;fSG(&w>qWqo!c;$B6<@jLo+Mza3}-GHKSX$F8SaUIbwME2h{K4G7e
zgH!$9DHTn007=QLll?Ii{1q@u#BC_~q<wCO>x?Vujw*9{ayhP}j*@*|qtF{}Ugk6k
zMLO)Q^Z`j1xI&yA`t$3dd^1iC{cVTy(i7j8kzO(rp0F>-M^g!txv$2;?wRSW_J!Tx
zbXw|C^@g50x_~1-5v^5EB|I2k)V*32`c-#F^D5!5^>749m;e+tnPJ^uT%Qv>P2yhg
zTt#1|U7dc3OjB5jz9f;-(+YpwX$7I~W1o}-Ax|w)IleDl(i8QPVrNz57J5=Ly7%5S
z_GRq@LyWBS4=DsrPZ2^1gIGnsJSTc#_dQ#@`~Aw7+gBuf^Om(+oZo4*h7_Z!$5+<J
z$s)wYgRUU^2X=p0_yvxkp}&1Ycmz%$a|gogg>|Q-1Ig5~`!AKI>5n@l7wBk1u%Ly4
zqwhs+@o(W=6l#M^_8m~_)2tgUkt^!qCF6i|<@{%Kie`)c_<kN53q{uauWHYzd`jAJ
zO^PovDm`v5Zr`u0B2_T~kfVf$rQTRBSG%&!#IhS!IE{Fg#-Mp;zq6Ngv%>}j8LOT|
z<P<tR`8AaN8t?b)s_x2pRF0-o%gsuUXlHn$Z&!DpoK<!+T9veN9x6{O{W*0o%qb<8
zzO-#A#Jlq~qtq=GjXQ9yObp$m#M;a1-MOoXVs#9>0**eiC++V#hV<q2wd|hZ=X(6&
z5pOVwga}HAc~96i9abb$u%mL_B8+o@nQ$pRuW0xnkxTVYX$OOd-6{BLnx4f5bZX1B
z^<0sY>bBrNbzmu+Z{l#2X}hj_wa%<i(WZS4tR9P88{&x0MC!7w?D|CZefQnx_4CS?
z-Dfu>x*nn$6uK-et-l^t%PCvd#(`ubC%C2oSiNMiaw;p5L1g9B<*gqDPAoi*c!f43
zA8{J0(_bqWSs_JWyRjoed42amV>w69IX#-oH6yB$XgBRkjYQ@RZ5d}#veHLP*bLH1
zM85JDzq4~~?y!UcjZTUK+KPQ2zM6+28mQ1!=&jtoSZX)@9NOk&mM2=oxuZ1u-a`mo
zH@9c@cwU&Y9FPcjHNCG)KEvsRedl*$8zGZ*`iy<`zA*KS>hyt*FFVl0O87;?{Agd(
z2!5>Ji|WV#k~iM7af98GZ@X^Ajn_KAQ<lMNIzHGsqZA2yRZT?NgY#rE{Z%lHLa*t+
zLfBXmvz45UZ^tGTdzD0S@^I~I^UVnGxMi!ca{b12YwXqa7A8&N8WW>nd)^sZ#8Xax
zAR?DvmgJT0f5BGf)GqAaKxs>-6$ogS4u$LATF;hgP(^B5IEjSZd4ksHFC=%F_k4lu
z9joeTu}mRc(C@(FY8fXLXR4yFs;GUyR<|DZTXCDTpveJBPs%Ri)GzEQA!AKa#P-ph
zgZ|i7rx_0o^Ee+)5myFB(6-%{gt2Dlq}R_KlRdk=mKQv%?=))sByTVeUBMo^qs{cj
zm0O%f^4<ldrG`Pw*0v8sG?fh(ZN?i+vR;I>;az-9K1k=sPdSb5L5k{EeW^Pek)f!O
zakAY31X(PpsLFSVB{y4X=clxF-2)G^fgiwNzt@4~x*Ketz%4vZ=H>bZB1m*}xQ~h1
z0VQIoD?m0j=?oj{Su#t=@5kY84lH=hUiSYJ2<0E~*Y4U^!8ehZNa)ohu8n~M)|!09
z3twd$+th{`KBZMzdrvq$x#A+KLY($-du{t37Qf6l1bsO$iI0!m_sA8tsUb_21qC|*
z$cG@;>4+G7$F{k92p%pTB0m?fT)ErECTvT6l?XBH+1TQ-?;}3I5g}Ci#dNSR&k&|-
zcXyX}4rkH71OipDaQYLKL)gCp-}kgp@c|kEb-^5%K7-t_op$fO$iJC+(D{}8?T#5-
zBKEq%R%4Ay#MSRJPEV4K5;_?zes_C)x2yHWCnjf|cFb1p1}VAhTTB$QH?+B6(Zy~V
z`kkbi)}yF<=@F77BsQJ1H||TDG|$BM3Ui3Ku*EY{web<2nja_YligP%m7Xrw93=xB
zSS_?O@njYJNg?h{`3;xehj8}3i%&ak?KhjZ5Wlm^-rS8i62ccv#7>XUxv=}nhwUx-
zn-iC2oZp)0mBfyZ+gs~)S?(j`mrt0BMM94*8$w{&X}!h>uXxFo*ElUr$*PFB>)TM$
z@{Z9F=cjJMP==9txO+CiQ=){0D$VEg#8jn1F8+PMMhcA;;RxNx=hSUVpj@G8T%~pM
zmbbjh`t8f>n`J%>j5$3T1se?nv$ZXV+sLPaS{*=OI@x`z!lruj+Zw4|di3JcV@@;K
z1p3Kje;U8JZLjTFAs8Co(}?0(A67;?d9jrPB4(~KkcaTaa7TD7=Ai;nln#DwGFz7k
zYNPpC8%K9JKZV@U0c@Elz55gMJnb2b=+)v-_o2b0?PzqAWn-e6JFt{%BUEhG#unQ%
zQzNrZyLd{*FKQFpnfPLx+qlVT2*X&VZ9Gwjf06=PDt`hXP$v^g26`x#k>z0%jRPP|
zg*XTu!IDGnlk<-L61S;=JLubFn*?bvAW}6-!fodC==~c=`hHu@qoVemMlBF6y~+|)
z5ROe%kEkW{6IwnxJ@K(=XU;&4!Si6@;dEdkn9;BVPPdPQt`tGv)rD|+f_tPq7l|;C
zWklL&L$;ZAC#0?%zdd|fNPtdHVO=LO1b4Rw3Z^Th8lPO4cVO`bgcR(2(q@&y8ixUD
zcJ0Buhq~dj<5Xvs)5;6Rb!X%(p`g=I+tseyN(~&ej)&M<f#nq3!vT*h{n>gK*(uML
zXrbt*2=b#}{A2yy>$ccKMU*4Rm^g%J#^<DYCki0j!wrUnqB?cYFzdi_sin*n8Tdlj
z3441ji&Xr8v6(H2eXWJeRrwyILDAD0So$M1?6$~KCtf_}lz!c#Bd&TQY`P?$**n@j
zOSUudBb^o|j|%;XnxPqv!W`JwwfC|!Mr;voUtqB=hPRq`EJg8z?B)Q3s?i59NWHV}
z)wIfdOZJij6U!#N9j0rKc4R^f1GKM`%pw-T7kiR`(#@i?q8_WKR`@y0;DiIK&1`A-
zc~@KC-c|ie4^CN9l*Q=}?4lF}CZsUGyV%VE$-hqP3dRi)%ZnDu6~v+B_c^vXZBX5j
zIQI1&&UEr{oaE=ga_=4<-PW73Z)jq`1`3NtjZGjz>W$Z5XYa{3_s3(k(@IW*SwTjx
zPMNZA%v*?eb$-SN<TfIv^rXGFeN>!#7oxfY3q(oOZVrTq)nkRRt=Km;@+jA%>e?(>
z<KFu7G~Inctoi9!;7G{*=I))(u>!HK<8oPqWmPf^W%ey?L!q#$ltrX7KruAsR0zX3
zxnTueI+MMxt)TER=d0OvF1yoHj9!ty!}k8VQd#T4d&N=3jlK1h<`b)s9{l_t=r$*K
zI9D%vUSsoOeo^(Es{+l4&RoUci~H?^ZCuQZ^d1u2YR$GDja44iBT!4&5A9pqh^JXK
zV*{t1){>zKl3iJL-&POJLUp=Hd2-r5=n<W)Shm8xJ!z^p#zw+^R!x>W_Pja!j>V~v
zvJGoH><v}{=#2G{eW<Q$20Mvhz_bgHZnbHpeP>O~($mL+S@20ZEtGyOX&uo=g|Y9d
z$8k;^4!U`!V9tT1_^<(2<@juG`;2{ehcyD~_%zNz;^IA+PFfE!0I?b99B>8RAF}VM
zdvw5y6t~Bbs(Aq%iJq7jsasNgFkIR9CQe6HOrJ=nfdi;5Tglg9V-K=;#=ftuM$Q-`
zBLp9G2jP=+2=JJFf8Cw_6KNYQ0FfTCylbz${$~4uVj3#3@~qQpY$*s4IXp;9=R!hO
z_Ja-e5YlrDo3r~ISlx4m{ZSi%QX!rS2etfTyXzhWdvc-;HQnFoz`A?FL=Q^$v2-vF
zLOF-*@tjU&W9;y1EbLx(<8`(t36ZTat#PyqM+!=aiXOb3vi!EUp_ny|X_W6f;J`$S
zkKFQd+gC$KQL;tn9>u$=fgExA)1#zV@o6G8vO=KE)nnNs6rciII^e)!5Meif$l~Nn
zDCDs1ug#Z~Ow=c@=2aA-Q~tpG5;WvtPBJ@%woVSnb56g8d&@0X+JWx1+A!E+2wnr0
zi!cI7(d7=-MVf47A|bFiuq0(!gF}b4Lmkd!ybBGi7yv!Bt(=1_^_IhSI1zD)1IR9Y
z5S{0P7_vWz=M;D<hIrU<f}>*}X-s1MC^b!{oVKo@D14v8A9_QxcC=ZNu&CZYlZgTj
zF!`u7Nv$kccxaVOCkm?_E5Tj<3WYE)#sp*2zylIC8Sth$5c|=8_BeNlzwVkPC~0bR
zmmN=nAJ)+t{4RotAnCBHryP7Dm%_sCb-kg1(+X517nE{v%1$=)AX~t5mG0=k62I$*
z$?+eTpZbZ02M%SVV1?7JM;&F4L0rI_@MOU<+eRdJ)En#oK4h8EBr-jfB(P@l8s}#`
zrV>L@V2|0;1;eI>ye^f0j{^&}CA6-fW>J--G(A$j0!jU++SiU?*Kk^e73lyPP9|$S
zYp2^l1HU1sJFvEVCFJIeg?bbUvz=*#iONiX<iG*M<Mm)^UCIPV)8E;8R0@;jk6>{?
zml~$iU$|3@opN0XCoRz?gAaIS=xwsR=Lc4-e^@)XHttukRB8SEwR7DwqwE_bbHafo
z0w4&bppt1@h<dg|0L3f#SpGUNanQAE)~vDf$*Zv0B2Cgz$92~(zt%2vSW;q57&X9b
z9tQOkS`q#6%Kbzn$t$9onzHenv*)TumQp1!>2M>V?ugT)=P&Gj%S-Hs+K_Q#!}=nc
zIfW*A^wA4CPGW?rG6cxj54X3$x+5tnlo_2^OHT&o$?d{8PSBsUvqY-5i%lLR0c~`f
z(+ssM>WN(HP?Ra0M;DGeupH@{SzG83j_eC2Fd#ax+_7(1ah+Xm4^U_@OPrd}4otiB
z#1NUc1N)H<B}bqgAwwn)<zv!Q#JDI*FEKjXhr2ronu!m?{2sJLPkgvMvQchEOfwW;
ziCZv}aHsxg8$4IBgQ~EQ14{u;p2&mS?8owBsgyhI^_#kM6ZYfX6v<^Gc}Q$|-^F>S
zCz+mtFBp{+<IMZskJ!A&exlxb5CK8vWHeO=CX6r!(ueFP+p|O0oyUeD!!^N%PES00
zse&d&B14LvwV!HJxpi!Y#4VaN6y-4J^-AFB!k+A>yP?cOr%DgM;0>l^6m8l_mi9C0
zdI+kkS|_CBcThHv=lK7EHW2&S4yk+5pv1MGa$ur}tsE}C-F~hvkeyD&olYZY0m}s=
zDGn~u6`8T0Pm|*-uX26~M+>{BH*r?_FVs`CI8<tzkVFR-Pf3G+@f`|)K5oC*28Ea@
z9)PFXfoTh-mW%yTn{?EliP?eEF4mG^LNd$$a(g)924TePz^XDJ#}+-gS+5)$p5Vk`
z`;~S(Dx3ntp1b_0H<%Pq)s=Y;N5-$#^hl9P4M_7StOR3l`t!k(D)uZqZNJtCgU?95
zpg<QfQ+jfy5#|&|ty9wHBx#86KvKV+$XnGLf1o#TTKRopHYM;GVc<6!4%SYoP>It{
zE%CO|F^)(2&DO*UckUXenM_rF5I#o{Z19%_%9ifSW3iKA$Duup<N)*+1q%2p1W0L<
z*IP&D>2Ls@b}DU0&Sr6{eK2mGqzT6XC&%`0cT*^r0$&AEfh9?d_3*I#PJRKQm3VN6
z`%XLg&}}o*_Pcd^`og31m^WBzpkj<i?e~iNalq%e(+E6OVW5aR{{6zV#95K*lk5OG
zYI#2stvclN$31dHP%8@#f5iTvur-fGO}070C%9TF6g~Qa!5C3gM1*z(Ath@ctl?R%
zdpa~Wb{eNY=LBCcmbgEv%VnkzCy6+!60&rLB>k`E;jj3A_@Di8n{3n^6)jfTk=RC@
zp7_}0a`N`<u|H{(a;w4wXvGzXeP}ugzd2%m+E4_?l`<-`NrQ=pDj}EbbNjP)FNnp8
zdznMH9as>$g!FeZ3W?N+_+oD`wkPb*8>$gnQ05RUW&cGacY4a^D?-b}qVzH^&Hng{
zdZz+9iYOge@)fE0JW~Z%!qj4aSv@J;WQ1TYNP0A}!WP^9UsX%Rc%E@u9Zuvn#3F5f
z-Ea&y*AZaa1?TkJ>8Y!kQ;3J-o$POtN-3>PXb+8Ai;HOvdcYC;+a|$|WQ}NI8~_Ct
zGW1NzR0`!*9&g|ef+b?u{;p{N6rku&an!K`Nl^lV4i6XpO3j?~!bR-=eVh9YYwq3b
zUF-JIi3jW-+B5H`90w{_&w<2Tk_-cFM3H)C{)CEoz`*A0AM-0I9=v1VG~z4uWl{$f
zp&#<8{ZsoScwcO&kf$Us8T*y3CG~&axvKWUMSDsXcZw)(`{$Y>rDdOYTFE<>FTdXY
zB^!jGk`F|qrbDr?VHvQ0ZFr`fdY|k>m=7QcNkhHS`|aOq=)$=i<CJ-0P-5GyCq={<
zNcPyjw|iL6%E~+);qE8Uuv-tsWMQN-Fi6By4Wj?3PfV|CNDM(;GL}jlq(4w7f2`&(
z&nA|{@t<|4?op`uV)Y5=jd#*7#=|}PF3u0^zZwqoSa`W|o;nyeE9#fnCj0NkuWqMW
z?6lLYDO5(P)VipdiLLK{62m>tRUbHwDh)(@4(g{8!_4N2$f1lhxM1j%ozUrt3DNVU
zgPmZb{HP@`R+t262Uc5TS0YMdh`EsK|D)$(#wNCo4mrPTq*;F<WKtOVLHn4wz{KbP
z1r!C4W|D9O{IHOi;KiCk7wlteAmr0U5UeI0m;{fN8`s#!wXb5OYul!fpvJ`3Mvv$x
zj&J+;xn}T9IatQ^G?8Uala@qDP0qC{k+}1I`-FTA@|yEAMky4D3F{zu2^}IFMr5DZ
zp-VN-a{vWX#$JD<zZEOMh<(!Bhu_S@7v8(D`~TcO=PiMJS<qBdRDVTyPuV9|4`yJ=
z69i9Ja*@*`bPaDO=>1dXV(sEpr07pI7IyELnX*sq?vKI+l(9hUk8OpX6tyP%lQ<R!
z_GxY6QfpxAf6{@yjF-Ip|Hgmz>2sw|5Ui&Y)oE*Az+gitfPF^mT}gRl$cc86ewD;z
z@~Fs~7s;rJ!ZhNvbF6)eAM;rItOnRJi(P`?Q~-fN1(vb>!BUkiAnEM0+h;-eN_m$h
z8dzQ(8Z73(uoFCkQSb%TBLWL6R;X}z8J(V3bS)JB@Wl)EIf?g$-RoDabsBL&(kZG3
zp198~m=2w*U>|R=a{M5J?YGZs11gWL=siq9PEYMZCm>X90%A2HXGsiESp+xM=eLy}
z>`5hDB(ErLHMuVRnuF1t=`UzkhYpt(8l3GS4J8?Q>rbxwzZOFM!ftUXqEfrLqywES
zAEfK_2YSn>9Wqa>Tl=CmZ)_QoaUFn=wy>LA)2hE=UtCB_C$%1Y&KoQ@;P4RGmn36a
z*xjF;*fr}k1OE!*V)CUu+LzWyatD&Du1fYgJ^5k^Ftgq{q+?)T)@A`MbbsM5++t2o
zJUeAN3C9QQ%je2w*c%~~2ASP~lq+PC4ROl8qV7_(k3tU4<A(>Pnc#%i5#IQfg-Pu{
ziV#<MjZCtOGCWRyU=oIa&<!e(lw`FzWiPB}OL~#@dN1ZtV8QgzzZ9Aq^pU@l>Ak4C
z=RD5|Dsf;M&!wa6w-@Y+HgkyTAxnlZp~?~TG`6qIE9NxKIBD=dZeP_f1f|O)qeCE;
zz+l2D;l8-8TS6P&V9DgN<iM55-bh`Ta()-~<q^woc}e#zlxgSL+V2`nD~p^cISB2l
zj#d{Hj5$_@%(5pJ_Bj1HXPg;J?GU@V%_=l8t^7p?5*T{y;+d(wy|kVuvgDhhoP)?w
zn+OhNr@z>$3ea+J@yl{kddaI_==@A8^CYnc@`~8Y8w0sx)fVp>>t=S?HLYjaie*lt
z$Uxhk6p>bw_!WsJcq>I+J2V{&+cScTUEA14I`3nV%0I9#gwGbAz*YLh)zn>^+G*F#
zl@)lfogk?Qrv#Kqx0(-91<8CyLb<*{G0h2_`5HijQdCGxvOZ<LnKMNPD}?a8-Oy$Z
zTY>Cj$C1;3N%&lkmweWi6~Yx9c;0Cxua*yy%kvyXl{s2oJ!#*OWQ=f#s&>v>k4#^B
zh0dD@Gn(Dlp1pA5@}o{md+5q>LW^!{*rTqb%n?}b9aw#eh&tVYX;TaroUO<S&14f?
ziG2qWq$*;FjMK}H+0ETk7R_U_!7K2PEJUY2NpOZRfw2*LWjB&T8a?9O0bhivoKX4_
zzPg^f)R;nCuEb{UdD7`m%!O?PstIQOB+lutX`B=LCy_`rlV-Kyl2&cE)Cl$?aElF4
zEXIKaF7%u#xI#}BbkZ5augYiQm&K3xMyDBwh(lt-<3?y~qI>g}O)H(>`OeMTdy}J_
zobu|Nnc1j{;Iv}fli7&eSH__zZ7BvUJ6vzz7tF3~<3uf!-&fvKU_~$5mK47$wUYU>
zTN|G5m>E(!F>THMW`ZYdRl^9Bjm}6jMlXR24OD+^b^9Xth73S5eKeR}pfo{}6>UwM
zZg@780MH&IMw98z-I0DRwuRldPLa5NTU%_x{TM||1`ep$-3jE?)VFT0uga}GDLNRT
z(^K;zKZgdQuU;%?a7T&K%@$6zdHNS4xtCn)Tq{bXQv5?io{rnvS}k)(NEe(|N830*
zF*h8x*R-)Jpr5x>2F)8Rv?R%5Y*%+S<iUYd(d*d4yaP)T7>rU(41?FYh6|#cXSS(`
zo&$>6Y^Rqg{w65d*0*^MG>aWqNg!YB$%4Ziyd6CAgl%XqAPX+*lsTYOdspAJ%I?a>
zsLb?EBVJ7=oDBUeXc_VJF&mele6c`hr`vh!*JQLR3{pho5>XAA*A_BV+$3G4M;%y>
z<8`Aw+tj|Tc-|4dAO(W12N9)3zlhg7YnvO{Ve=NjCByAFj?-V<K;dL*`hbtXN4M0S
z&!Y-*ArLr1&!s=<k*if9MMkH)+kA!{G_0Y%2&=*wD)!p$?F?xL99U3ny0+4;T#3{0
zj<a@en<dV&Lotz<2uOta(wFBydtD=)Eu%w<YH$GQxnwIM^H;&ZukT)+87Ihe4y=$M
z+nhE+Pyxc`><x9b0`56(5LQE_*Tyj7OyX&KV|UGSVdy_~@t^~XkIATZwg@B(38``o
zk@aWyb@!<BYe>`%Ebd8r1Ph!AlfMMX9y#!7dsE(nP>hPCgY3Zou-qhWRp8>Zy}5k}
zer-j7p6X<MdeUUNfm*nIdrRIJ`TZ(uK|4?o92Az7B{H!-+glfVAj>5b;`HQ7#pnKz
zR%);Du+=L!u60^LAo@D_qaZySDwrf&=@rCBJFsF^@(_t^l?0weDG_OGxCS6oib(WK
zY~IKZPEUT5su>P!q>YE!CWWK#07|l&h@)=$*01-=t*&}3m;oGEUNtm9uBg6aYa2OM
zK^f8%+6yd38FoijY|4$@h;3WLJTj-u2}*BIR!mQ(5<W09rZUd3ku2h}y-_CtSN&n8
zZ{UEghN4+w%J*TTjc^(jRMU2gfj#1Qb!><ziQUCUQ;@uh1F!fglgHccxBI)n=qwQk
zDx(gFHlim9lSjxLwnqvEieVi0!ffy8CJ0%F*YVeZC4zQRo|SS0BR19`(}=^IY&h;f
zVzTAcP$2Qs?87@78qv6AUw9`md%(KMma9^xDoy!?e3y;a-hwJc_y?_^%6T}B3UMzM
znVHC4dexeZTb$qV%oM)U6s{mAoz{fk#U`6Tx_uUtMpQb+YX`K5gL<gwnSFa(TTGZ2
z#t3Dho%D@;0k6bPo9aI3AU1IvOvIrAyH+#SwBm5&Aj3)+3Zo#@pBXI*raT^Yeqx(#
zkPp9&dgL>BX=WXO<c@{iS6yY(i3Q?vq-s=)S0JI=Okqf=r-!_?IAEUT^+4fEEG%8B
zawLm<>B)JAB`DEE#Z1#4BA4xx_FzXbO3>B9gf4InA9Z>XPirVtG%<<0Wmj%WDqS_=
zG;)=*A7(Y&Ih$>SP4?n6DHR-V4Cuu{*wCG7s$fFoEvlT-g*ZS1@|B&%((_O~l$d<j
zM0l7Sn5g@n346E<dN}a-nVS%1G1(0ze7wCrr3GQ5Yw%!fBTi55rZ?QQ%pR#*QKCbg
zl|{0_WcJYJPA;pxqkaD-W(u?Rhy$tZ0$qw}R_(tO0lqHh2DKlzl+#MgO$-muOsZ^?
zHwxQGji2EiPWuIXQ?K}mVn1ZkCw1H2*@(6`QhowSpg`OA%HIR8dBPs;V6tO}?c5tI
zYK!nh#AACb(YA8aM(1ZTPR=hQ@9te~SZh^lZQ!(Jn7xT$w0GyvHS1pQ{1kz%?sLTb
z*XKJivW(4ehB7<IhyznUC!7uJ8`=~5Gt*?_JM9>Ea`%v(>^&V!F)E1f?8LwdwXy&!
zIzg6kyu$X49phMX6fKOc%`N?*8l;Gm>WFq3u?>kk^6>1n_by=-I@9A6<dt=_i7r6(
zYyFH96JImxm6;fM;F}ua*6Da6ry0Wy+i9|U(e}lI_RS3e+cIi~c2+q~r5HO`6fXUi
zB$$QW!!ujAI;~`Pj7z#iS6;~8S3POC){>ERL+V`Xk%}u_LZ)0Pxclw>?P=^)awseL
zFH-ZIp0pk!w8{UV!)Zz>mR}>*(JA{tlJCmNNNm&LxIkij!mF$~&vCb$*=Oy8)uS=B
zIL=&iT@NWUN(tA|^5NmKZ*91wf&;dBnp!QKeJoK%+wiy5WC;<c0v%z!)vIBx&v@$>
zx0BhQ!X{i5wG|<pv-a%`xs(B(c3L7&(nbgN9UbW@9z+I+Jq}FFfh0ay+=trtppwYH
zr>w->&FQHPbg{y=LewSAuzKw~yFsgbpuQ9eg*S5c6i)ABPQP9=J%Ku9-&GesFJF|M
zOG!9qFqP4N4;TNR@iW+WH~g&Y%}hD1n%q1NmoRYxCskqWdkS0f)XQQ*OobN=1~9Xu
zH#HmOYrnTmC@6%QY32j*^(yG`*_a+^+P<%QAW^J*s-jSUN&gKj9klPSE8yr6v4n5W
z8w@L>f+YwXJ8wTw?@;WkYDmi<z&44Slb(*_Kpg1hhwTR&|KChfD2-&LNIxRx2;^<k
zj=-*IPZ@~>7vJ4>=d0ghk2mK8sYlziL+e})fSnK`kw$u5D6~9=8s4_2AsfraxYH7c
zU9n=B?d=A7HY`jmKyNT90ZLAN0x@`B+f9&FI1GVTc92_V#z<g5L!8UUKK%CkoS%XH
zPA-YPfmBM{-+g$hft1kj1{0+##*kPfo9~bjJRucG{1=Cvo<N;~?GHuB{Xsj>7~aN-
z{xlV7Y3mEGy?L1(tc{_^5mWpY_<kx|A8Htx<d`|9^}P8awayMNfhn4j>@{D6DefWS
zv+YP5X@#G3I+N2=Pkn|>UZ%Q(cC?9g$gP~lIqm=gRhe6@Xjojo%D=H=)x-KkLQb2A
z!o&|aJta?_1Yy#2RYFro!06!0V|KiuL$c)rF}d}m)TWiDSs{9MqI>%@Z0QQ<my+`w
z;tD-6c)bxLgA^Y-*=UanQWOYJEfAo1O4JOoClcEcp3HAy`>JcMuqW$bDG|hO=)mH`
zGXhU)g5(E+2kfc(Jnh%oI~7nMexEf6PI&7Ve^cm{T$74@ky%x7W=|)v;>?Is!ibax
zlB)P7{4!*H!QW5S>$ObIGL>^M%>78C7rc5X9XmQu{T|L{rzWEAIo(#lW;Rfnha5->
zb(?3W>`bGnRcM4xE4$XM-C$=M0>Wpi!Ol6bK!6k~Nxj%Gv>OnH@=PJ1DWbBlFDY#0
z^yj1BFfn20+PAT}Ao^)kfa=Lo0>lm^*byRW&o=UtZF~m?dH^xa5_1)kr=aKaZR(NS
zw11q$?f^_17+d3ZyHJnU)+iJ7G5BZ-`t(S&!%dCJbYRbQ6PXN2vZ|&?QE~EL(k}`j
zvfF;BeFdqVME~^>1OdssK<xWAjB$PY;d*Us?W~bx?3{V(^u$NZA(F}&1>3fZIpBre
zw{38KSAV2GBXgfY%T1VwOs+ywk%%g4*O6NgBz4Lby0u0HT|QRElzdNa$@Isw(;cST
z3485w!CHt1>a+t(sdo^y)_$bD3cezjhjt-gv8*$eG~>xoJCSM9QSjjoo#_aUFakDg
z1wRvx=k&*dd+AH;N83EF-LPfN?Q1qUfOvNLxbm+AnG8lhVn0^Ml{_Xtqw+5G_k_2e
z_*_P+G_3e*KVGvvtsvcj(~MW75Zsc37On9=(P8Bj^gIV9_c*ZRWbhZw0+F1&HcsMU
z!4rgA_h>iFBb&}Ru%hlX$wVj=Tf-CfQyqoKXaQ0}p}{-j^u#Pl3_|qAcz(owx~>kH
zS#O1QLN|4KQukgnNkLZ%y8TQ7DDO8r2;V*cp?dj^H`>o8p%I(0&G{X#nb%I0@|XQw
zW5&ZAgt~3kX(usaB}1+Q4JYj98*yMu8W}iEEjD!XzWqXBNNSW^SE_D-yue?%a0$2V
zFxVBIq+dB6`l2KDi`}f~43#n&BdtczFMOVUt+G|P%SOj%`u0o7TR823fzZ%rPChhj
zzue(ABbTFOj99bNQ;e>+)qbU$Fde)t3YvFdHDU6*X;BdQC+DC2YQtBof}lzVrpHlk
zmGVEoR-;$=F3S_eNL8355bx7zKu&+Hs-5Kj*stfka(j(CjX-o(Z<aZ0&VHkYauC^4
z^hQNzh*{b<Jwb^X0s#%f+R8=&$se@eOjK;28FGHs_bBRoNgYl5t?rdFE@XZlFnv#2
z34go$9#uOl6^|%6$$P5mLjM(9oao^J5o{=hDpVvLONVU=I7iso@3awv6`67M$`%*_
zA9VU-n7MG1Cg4w`ZvAfEsla^E)S=iqae6W$s){)MqKPp)9)$8t+3&TD?Li8sIISdL
z)GgF0rud}&e!c6!Mal13N)cleyBs|^Lw9o6<*fZdcXLFLiHZ9n>}*6G>Nxm6Z2LhS
z#ft(6D4(P(t4aH#wn!--#YM9YAj5W)-mk=NZP|a^<|=ecwn9ZFK(+^*9?i?dj4Jp3
zNexSisET1J&bNb!IkVr$k@lyZq7Ti0xy2|(rN${34pV}2RZ^a4uGZn76->r_)pdf+
zI<S}_$z}6gm&ZWKi}>etwQNw*IMD?hSSZfLAB%SLbmzmdc`J(SFPe-p(x+sp3@Sjh
zh_}y-+F#ZSlmDlVpB_MZYO$ysTI%qB)dm5%Nw&gx=D>=mlui?lKn8yH;(q(<#^)3!
zkJ3UruO#+gZm_gwBg)v{R8Q_d`$Nr$oMLZ1<dKHbt-2B76mEap?qQ2p@f+DPW0X!$
z8oCrIKv}0*`@8O2_J+#He(JztZ+z3s?C<MQZJ&vvfCxFKeX&QT^|TFY>pf}z&|V)o
zJbBPU@{QJ$EMI#G23P(*VgFdGOnewrAtp(F$jmmSk<g#(D2x+PxX=En5p~2aAh`la
z65+-;9Sl+ozmENLyL1exPq|S5Y6_Kwrnl~~e`)VPCQ=OU1ttcVl--5hSQH<!f6ZmE
zu=~d4%bi9ME?K0>{_Nk{Bf<y32VzHPFfDXtmSW_!f6q<phGo}Y>-^SIeM|ml|51-<
zT$j^y9av5qCvTCyNEqIK=K6()$rBt7mq`aG$vw`|4By49{a2x69$urhp*V;XKU+^z
zJ5{qWLUZlE8?Qt?S6Q>jfsEghL?3th6DODCmH(eb3JC>6BAVotvrf;|45w%#a|jgQ
zrv%t6vKAac(k*XMwLse+d%1#ZoZq=q-gfr}`>6fFWEG;qQA@K<JL6e!6AaIwa`dTS
zVIRFe=U-$fwvh$8h~Ia5Qgp&ltd~X6F3cnBNEfM{Y9CW~Vwa#lqI3&Q^u#Pna)+8&
ze|zm?>*}-}q+>n_oSq`9V-exUj&2{<#(L=kaBO9+egHlWG{vU%$n2zU*vHop=b*(p
zv<N)POPqX?ev!$b-&aJBeM0*pu{)s;k`*y1+(VJmtv_w5p+wQ6I47Ck=Ij$|r1}Ph
z<zSU`V6lGUPz+%yLRC2b%s#2Eo&#sveq~q!mRCd3u~>tB>umN<uCe9mE1Czz#*xtg
zPEl^0eogJA$1XoMVV}}a#%?Xu^(57sfd%*O3MH_RS)-URByuaIK-~LN+W<lmp3_E^
z%h2hmX_(V=gT=?~(+Z}H7Q_uuamIm#1krZ=C0+t#8u?Ha$SLk{)INQGFvXj$yV3cX
zzTBOh^o5h{GrGCRxX#Qs(8H1{=nqWk6I5LH9Fm=c(-ZcYIjIY~ABY-TG{tylMh5m-
zt!H{x!ZOX0{Ueu1-w&?DB{RYNv%9Z5BAI<b;e^0a2I!9?1=tJrr`>Khqj<_`DW$y8
zqiB-<35m^byYv|H_M;aMIBgAS`AsYAbMxnpi5<?*_@c5m#MZ8Wk~#alZc@@>E$wSR
z2%L9%l8N2gLyACVw$IN2PRvXXRFauy%$i76CPJP4{RJINM=)rs+v^RcdBGbg6T5w3
zejKHJSKPe9Y2}xrf)HN)MICYo*b>@cPQZf1mgMI$Cz4T1kPT<s7Z=VB6F_ssX{FGF
zb5$GYsF==|)CVUG$-JrkOA-q`@jV)+Xz2oG&@ZhAN-DkJw35ed9NlJLR+md`6_p^`
zIIzT7e;h7>@5a8ou1e)vMvn^)EX8qJ6+#!JTZtt075mFqjeu0{W1$xVgNBw|O-lkT
zgP|)Ai#u|Du@`CWN_dUXCHu<u0hpNgI<2B?o%xi$bSeD%?S*w6$^juowBkCcBT^OQ
zGeXEFR^mnVJt=aFr1v>6sdNjwAAY%A(NMH(#d5F41Ysq8`>J*))F%axj4}sS%S4J-
zsw{Fd=OTXAUfgCdJtfrWLFNOPbjN2Za*)4vWwK{Z!5E3?Vh0cxDcL4HU3={%-PE6f
zv!_EnxTe#S^LFX|;P-t9MOP(0Hr&1bRxAv(H1W5sUAJblU7ephq!46&Cq*cE;idI_
z4M|4LZs|KcdD*g?m)pzQ_z5o-``3v^!199S%dWGR*9933$7)2C1tuJ?<orgvrhQu|
zdc;~_-f1u`!<1T{wO7=2I(J5j3p<GeD^~rHdCRGgW!(6+-3ZQMIG|P<{&--yn%0iv
zK9A!k2X@{556gFa+pzOXU$Bn7eb%ne(O&uxZ{+uqWyKt$8{g1oB5Vm_+c>Q+EBlU~
zxLV6kajeQkTb6UNu=_PL*uQDz^z$TG3=;SSTi(&pg)j#q_VB&(S!l&P>-5(>N^4d0
z&b-~&Rw2_MXLH!1u)0bGMGklR<DPl=q&$DnrQOt|0Je2Pg|<0>T*?uvB|8RNMeuq3
zK5Z+SFfpvkz*Jm8oE>qXX|g4_W}1v()X<yjZK7AzZy<xqfyFA+=BG^}He88>_R3`K
zo9|lf{7lN(uyU1sbv;EggTf9m7CNxvHl1*S6s42jWcNR1UsG>kuZN$styST;^u(Qf
z6rUwb%I>2rI<RH8w711pkG6{_0IU!aSiPvE{-w&stGa#c%%u;q)|Ax!0sJJuwQqgx
zRoB?pCh=g48aTfb1;NrHFejrY?A7)15b=ur0t;6c><Y1FWer6G7e8Oz?9&)Se=)<w
zoTW6<-&)V^YK_5SfyIX<3vLHe5?9HS<%U|--J=qay6S=hOQDes^g&61tGkOhViiK+
zoMQ(TgXRb(R>qn}8pESg!+1E5tO0doATOajX!?7|ZfiI|zCg^wn|0WMwQ-`@mK7$x
zU3Q<7c6$?saL`aA&1sAE5ewT2vEPxq@enkRamHyjG(jKQhZJ6faA#|)r%9fR%1Y~z
z6s~X-ardAsdrb{5o*i#c(VPmECSQa8T6LUsuO76oI~z{$m@7<*Q3I;WNJ}hoTvxDU
zciP#HI<UY7?1aZ@(+OJy$cG+C<6B?%$m*yVeLl_`Y#xSG;i9&odyewLq7+mtNTneV
zMB((Of}`U7)cYY^9}C}IZEfC*gRX1bJ~3n)ThFo^Zgv`hVL8Lt-SPbqcFj6JX|HXZ
zj}(9*r<G#T@b-z(Dce+IS2RsU%R8|6UM<rU7bw4kQBomu3KQJiX8(2$q3`c>+E+9C
zU}hc}4*MVi$CAn&#SYRc$$J(~+m?E3R%6fte%67dhPjF(GwtqNzXV+0?)=VA3r3Go
z201t)Q0<<MtaJ(=JTv4)9pFw+uA0pl3aI~M_cj2=FarDr4j|vhNu5*nx`qd4FP?<B
z2rM^rI4mLN7*1Z_Mzw3y`5mK@VMYnA2PJ$+ld(5+&w4ss1<zbO<_$K-Iu~m`3<G{+
z8`7QQ6I$60AX)lOiLg_ONxH9%l9;!hPOFxY>@V6QG}Q_;d{fO?YL8O6B574o0@53t
z{<u@yfYQIDf9GO)bEC>dM@cNH45&Tm^yha*m_*nt>(002*k*9CGk99DnzE!R&%Suf
z-dZpP%2q)kZ!qm`a>A>^Qm<i}im}t-Hw2QiGciq`!B7($5Q%%#O;BNR2N-0Ew1F;~
zl9IfndHT&^8*UHfR1rcrcMcp#O_^K)DGi6{WFu|xa3;x+$Jx1n#q)#cgTE;DrmOUB
z{TyQN{!8yd)1j5?YJmLMx2;Jjv;61$PFjS9bPkPe>r~Jzb))e9B@!#|?_$uMYqh7{
zeBBMUJ%9G5C!L>xY$WwAylu3`yMz-mk;F{}%R1on=LCgU68QvDy}x@-9@U{^fq^9z
zA_PB>%I=OvBeK{O^`uu_D#-z-zvwbje_`zS4zL*Rv$5n+>j{A9JB^SVr3cHwE{99H
zv=;Gt%yuSTF*Kn!IIYyXv4BYf7;l83>YPq1zDeq%wr3ctO|+R9=}%AWa+*md(3=c|
z$|p`<6lr!gS+Ao8`^B?PD;}6)h0He3+S}R#*Y*4Od}zicIvZCu5b^|Vs=Xw{wpduQ
z{R#~*ttXuRxQFEvv<mUtAW4KM%n|2zAW*&%Wr>Q)M1?18y3NK=9O<zYLl_{TBo01D
zhPIjKy*wqylyc&)(=QUb@{yO>1NEL9p|nl~81I4o($n-@MF0@GK$+JE+n00D(bx{B
zDgEKP8*jSac6FHN=tAi1AhIT~K(%&kX^?4&)^8#Fyv^1<ngqp9%R&IG6JybKCEz?%
zkCAf>-7E~_;VRPXq43e4L&Xp{I>EWYh`;u5M<?5GNN#i5wbtcGk_}6aEM?HYy_>%$
zL$1`;3oJ&yKQn3-pvKHDZZwZHys^9L7yzBX6gYHx^3Dic;5}jQRm`Nsx1;usHV)Mt
zjE_+YnXD1?#A?VW4Xwwr9Wh#@V6J+4;kbQW<6_$=gR0|a9Z0+}g-BsptUfA<u@yaK
zUfe|}6E8a>Ttkasfzn9<iE2}a(Z{$}s@H2@WADs`I)Uu5-Dwo^B=0U}&mFKw8zF~c
znl>k3B8!=^-oPH~`embzXblG@dL=lQ0Nvqfdsn*#X06&wk6%3Hz~ZA)eV4R>Ow5jy
zfh(u&-E~i@rXr`~4lLP;<k)VCWhG0aXauO**VpI>`*nV6B%)@V0^95x8h!W1w#^U+
z+g^g^BW+FP?%&gdtaVhWqmj($e&wso+Ben>SZf2ZoVMt|VpJ%Iz%uts{xW&Z6f_XZ
zkG;41UhT)}l~kE;N!x^PI{n2_RdgwhF%o2t+Bdb`Wb0Uem!ggVh__%*9Nosrsxt$G
z#&7O!7ei(Vp-xDGe#Yq$Ggvv&vv0|tcTG+@KjU?>2jNDBRqFI4bYFX4N8hsINUW3O
zzF0(#{@nLoOEh`k-d~p&gTa&(W|(Hhb^3F#w~__YCk$$Q)IN|{ean6M&uJGeDEvYj
zdmUxRz@)K!FlTgPgaX$54#t|Y=RDH(Ui;R9Ni~&5gL3b{;+ab~a{`RT8o271eOvns
zHVqPPXr}lvG&4J}Z*Molx2h@|w{?4%)1Ta3A?Yz$r|df#sZ305>kW1~?Gz?=j$)XV
zRqsQM^W6m>%TO^usR2;kV@hcb!G*uGJ!*J@bM?lZwx)31=!|_=Qs;<5X1p4dv)#GV
zzPqD<2o_=evn+6=KkoFzXp`e8Z0?AC&*Ce{3hq1YvNc_Ivwd$pDh&eiRpcqqPdGjG
z;Yx*()Op;#uV7>qMywG8<G_-<Sn;T=u@lQ0de*+bP#b$Ow4a53bYgv0o}xeQQ~<g{
zlh3odgSPAk>Q0fhB=6t>FgrFIW_*v+uTOZ>!;jqeX8XbCJw~stq9RUDF6l5Z?y}uA
ze3hBouTQ#o(1FF0PNGgVqgYhslCW3U<4Kl1>JdA?lYS<9BO|t_-Y$+EM&VEiwePf^
zuxN#sMp599z3puTIo9V;C8wE_h5(`DA{}73ufYPV+8gJ*!SIVCXed@=-sakMSk1-F
z@xVH;+;*v@NC<;0kNdlQ<5<}1U{^dK&m17!Sql-VL`-$Rb8W8rTjRXh`Nor0-L-z>
z$_=k~0P3IMWadc@12~Y|%tO1JUy1GqN2%8`Wd}?8l18-OX~kE{G$4-;{L~KBY9}?w
z`JGsZjPek4?Qo-m#mH>vNEtM--8=m`RL&-u*l9;PO3<N(S`o@Yzz5*;h+rrgIx=cU
z+c*_VfvteHI<HY(IXjjM@WvaKIY0C1QS`))w|S?)(zMe`;#C+>ZBQYJ*oo?iAwp%c
z$V+o4thBAXsqF(hS(l0bOH^i`$^FCWNjy?|XzSL2JyDpMNAXX~(GdrhEGHd1M9e;!
zxLga{b$*MBZ0>Ehr}F2}@QCwMd-*m3H0|m9y|DYiN$0l+I`-s~cB+jFZ4z-+BmsHI
zc|lKd1kz+tmn8h!=^X2>U3|d_#6;S04J-WeV|>pkc=nkbtCDyg=&?6XJCNK3V7l78
zbXd!xnsC9+Hq7rFB~;UC)<d7xyv7=OCNCMn+e%id)6_TIg+55>oO2C3o2l#&1rva1
zE>}(sZ^z?o&$bubF*;0TFPg=}<f~m8XLP<z*@lg)oJOvE2xr(XbkmY4RYFmNN&ieG
z{enH$K9(s~rL(m9V-id$#OY78f(&^Du<Y6o<@EHnj`p12sbcOVSU@L-{&4#`v;^cX
z1Y#hCPnmF~Zeo<@aTPgb7aK=dvsK5~Drz2(zJ)?9=;X|Ok=%|MjDF*o&4YlJFT36r
z+DEVAWOs<6P;dv97rWfBGsOYHG(=vb_|oi0x-}q?E!RK_`1B;<;;|VeRd~!k+y*x}
zK{bmg@c}3&9L_Ic`~sTVkG5}&*q_Ftvyvl`aZZ2SGq<oeP%p#ak3DY~U1T5=MUSpV
zu<8vDE2kGG`QuIZ=<ga%#G;UOz}|qe2(f7o>?dkbQK*xjK|1<I33JVcaLj%(X^bje
z+le9cYEJdHV&~s(Kh-cEJZZvd3wu}H^6-83({09Aaq6nXKmg(h)BqUWs)`gp({K@v
zw@lWOF~3!sqd##G!dwi>ezw8bN0s`)7U012Qp{!p`?=~D>58EdlI{iLSjrydAyZ=c
z`FgjeGlS0a2x~>Ha9)qoFT&cal-_~;LcU=LYVrF0fR5+;#l#@#D_%bzdD9j4OO2kk
z;>V#8L@ms+vm)iF!2RR$wM4kUFL#(%IA#e6@~k;M_4yI%qVOO4m3ms5X47Mm0%JVQ
z>5qFd#vyhM`_=Xw#1I`_qM+SH*g>o>=+}rvzt)tMki?ceHQ2fi@i=vHuX9y=Pl`K;
zLomptzUg1D??Lej<C%8nIgD#3ot_$-0<1LMXYDuYYNb$R9J9joaGg6nx&05evlb(E
z|2OO66mphs_VmP~lYC|GoU`BR5HI~CO*ztB!ii1r4gEPB5F(o}{z?1oM$WJ#Wopun
zk4<D~HxCzD$bJOK+3$4Ip%9m(;5LP*$WDJPD~&<YX+-`aRsL=vd&w~X!C(~SulV4c
zPOgeEO3{}7sPvoPOUhK)@COG@OKVDHK9oshzuy*}e8gg^0CK)$1!7ZRNyL0sSquAv
zdc^@=5iz}Nf538w7*fo~AJ!9ci<gS0V6BK2L4SYB>DM3^D5cH0@?)H&G;4p<eEVEc
zY2XaMs|8lRvi%_tNrI!sYFjj%&24|&&5hC#W9c|BNdu$(QTvl6XotF3SrYfV!?c|K
zkcL&WOEK_afsBYG*&^iWg`eA>*5oX$u^{90F{dYKZ;P^TLc{s9L<z}C*vFk#yg(z9
zIwf(m1)bzUWz4?)c|$D^RX`@?UL06VM=sqA>`JOcI$1bqe^LF~{nAGvzmLxH(Yu~?
z`ZeFAn~x6>i*A3JSRo_9`6&VLA!%jySM_f03Par!2yVc1fsX$hx4&+4yg^k-Xr++2
zVPXe)V1Lt%T#1^Q+R$K%ypuM=ny|lJ{2Ju;me0{8>N)*EzZl>|Tp?@yU14%RT>q5Q
z(wJ6q9LKEveJ9^XKxg=aQ_|2zDMW<h8B6LJ=h~Df24jR0v+w;wzH4FkN{I$hA;E#w
z`r{yb%~KE~lDGY1*Ar*|fNcee$PYnJ%B{EcC__DN|CAVBzw&nHXQYsCv48G9Tm}Hr
z9K{90`4$R|)2}TFR1fT5x>sp$Rep}-d179io}{x)Y|Bh13+G=8JF2z1-)RN@@bV#a
z>0rh=`?of6J22Ket$c(mZ__gz=4k(3H%Wv@0qTHvFl=~=%R5BC_kXm9NA+0=S|=S?
zGQ7?gt`N6#VV0I@=RdoZl_>;R;}dOM)fI4Zu8OA+kq6J=$zfbMA1v?*7RG;dgYE=5
zK=opkR81z6CUd=h&i=bYc#?oZAD}>xA69;+zcBSAFN<_e`=2(GoBO*s-p*+!$!x+>
zkA2N-zTKpmiJTn=b~Q7{45>y_+hJ4FR2-J-zo`lwCHffvaQmow3YD8;eou0I5irgF
z-Q@1rN9XT_-MyakyUAIx5v5<{9PY7?sh-%gd2iN!AWg#QNrLQ+;c^?;$Ib_LRyiRE
zfL42=qN>XXq5R8_YhSmZ1B#<YkOPaWHQRD+$wDM#AFz+F9%Z^J+xr+gO7u8AH8tn8
z@L83z9CtHspU{oqS)6%dIKs~fYzOoNo6K1Cd6s`hgJMDibIMf{AO6Jro|S_E(Kb#i
zX$RXLrnSos?a9C{XCDzN_DR*BwP}){x8K@w=?_BC5nsxqMNG6$ZW!E3I6d5sb}kx{
z4v^ocx9_x1$=eoolO4N_bZ>2)z_bmHjiU06*r(3N{&PDxub-ROZnjUWk5BukK3<0u
zkksh(#6ZGH!r)I}#QOAt3H=%Te!_v(Vo^p<1hMj0L9VfU>@(WLAfZI1(J=f|v&`wK
zH)MtZ1zzkk^HW>UiPMe=m%TwNMnOI~;h)uz@<4B!&i|#|?P;>#EG}>@j8jQK_SuP(
zp&3#Sr=3=eASSZL`+|C7!nB4ZU7fsm+UXHKqXz$J`<%KLoWVR5kafm^Y1vFo>>RSs
z?Vhd0mqH=1_?nclGWR8k=<`|+d{erX>QqIK)1zl{Zt^z!{2GBc>2ar(3yYL$q&@qB
zhEE8U?~UC6SS>FFZ^(YgvI)E37tY7dHHm{~%4x(T$}cMYNhPgw=L!n7FRK3eFn;K-
zHgqM6MB<3kuc53N9|`r*zIeXOtVHU~@&Eoy??WyFnsf`1qIjb2*_Y%OZJZfekGx7N
zFPIoYrB?BN`_i^vSkY35QU+6^Y953_ztFR8<%ZSvWeoxH{?Jc7Dt|sqNiRq2%j>$P
ztl=f~4i+wv4B7vR`C_<tQ+t})KMtU1JxbBW0gx%AX_Xwbuk0SIQ?kWm2<GfiM_mkN
zC0w$hqJ$sOUf5o;u|I{8WZ(dDT8Ael$d9!bEj|ElC-Ftr7oCq0GvxHANqZQbf2b+P
zSLAZvc*jcTXYH{g_oLjy8tkib@*l(oF>o5S4mJ5=Q_tCpJ6PsJ67Ar?VuhtCb!i+#
zCZJ_h)|L6_h20NLOgN1~m}V>$j3|r7lH$SklIpK&yD@LMccfHJzr?GB-H$wSpItRy
zLLO<1Tc@37_N_whxVjs!0t6)zA9rAR!R?#xw3jBKE$p6xwwzYcyVCDsBEs;drN;!>
z%j$75or&ACXF9N2N_pad^kG@}582BLG08H-L>EK4A`KLjKJ8o;_vcmVFOA*Eu4ywK
zrleV?ouk7xF8v@^W0uCmnPsnN{UJ}Pz@x$$<K!8Vw~u)1*X|+vmL!{u2$aeHfL&Yn
zi+&{jcRB?Wg*(z)e=B6E2+etp4=MfMuFExtUAyo6PL>u?-g;wI|FY}bbWvphhey|(
zLGJ6@4Ta%|CxsfMTr`+4s9ME{QnsvJ7hyy47Ch;|QoqCCJY~!43QC5sC&O%!1A&QD
z7ruqkA^Ly!|L&jN*k)O>BkwzKnc}A%ZcZcnv60cEH5UbDNv64JkqX$5;DPZ@MRm|4
z>H5|?ZAIcIj^5#SY9<5KR<xVz_S~NoI7{gtM3%iyPm%Kk#ZK$83=q?HAx{Cz)9H~K
zIf?l9)eQ}IuUfa#Y32RPZo1CCrkj<Rwcr-XFX@TJu2m>$@DLK<Ee*%SvM9y%8BBA)
zk_(OWhX`-9S2Zp(m>EL6pk|Fk44^UlgmTI*6i2$qsc_J~w(Hr0NFleAD3LM}-g-1s
z+CP$$k^5q=o-eB;Jc>*JETBMyy$AZ+M_Kf?GQV*%w)1JH5ns~c8pTg;Etu#~%oYLL
zffa>P2sPG;%IHXPTUA$UEegL1@dg%aFSlULuc$zo#GxPA>gpGbMs;Jv^(9pbo&I=R
z%Aw2|`8f*I*<)+E@tzJ>9#X3VOFJx~r^7pjCU)6v3CS2rRYP_lN$tZ^GsAX!M^7jL
zvrf~XhyNXU|NpT69|Qki!GPV-7ScF|#U97uIgr$rr~3EfZ5i0w<eXSWou9EtYb%xy
zH%*K7nzp<~a4XL`&5~ahGZwVxD^5{s`p)h_64<lgM5rg;dc=t&CuBG{PDWeT81<I^
zJx-&0fs{)z!?}xlB1A;x3D(bl_zD=HY)|1VkMl0XykMyxQTN%oHsw4HWGolShK?I)
z4_DC%#c|^N_pXOPRhgEkOs~He587S1d(GgHaen76FiKb`39=jOd8y?i;7am9VmjgU
zlyDeU1|*)xwPQt1MByehW_xYN1EV?th(G7Ra^qUJ=62gu7eoRnjFs|X4lEh?WqIZ^
zlT$s%Y;$AQv4OXAiKyaCr1E1<e+(+;In{VYHl=?*1ZYZ@ZK<b}GeCAsZCDDhBid!3
z)4!Bjz`G)YsgP)QH}tH+$iH>qw8g5a+0)xOYWFl+(&5LCJI&k~D(G7~OYAasZ(^W_
zJT>Js^o$-pRqUXzD|{gMTz*s<1*isAG_z<Pj#?BUA{?~W*FCCUR<2BMu&iRD&HpKT
zLk&n3pOmPkif8Glr>8UmL?U2VU+s<gHB%F#Xm2~6W`0fB<)i3R?W*_X2i$e{mW_9B
zaau72OYJft`-8&frMqBnYKR<YYacl6+}jzbtT{TR_RVcj>8jVB?F~kB2*w3TGl%Rg
zZQ<ilAOLFM0P<_Fb4M)%#JIQCx+-fhu^c)jO4+F?C?lNc#gOp{=i0!xtN@EvVik)D
zAFuV=JY#Mj?mNxGOxBBv$b|5g50zXvI=;*KUD6Brec*!Au57q5v9)7kGpcdyKx$Pf
z$2fBdgTLgfV}QqOqzzydzUf(~nF7ODkI<@qjXp(5l;RNaRK+A#h0`AcQPc{-v=ZZO
zYe!+?H!!c}#Q?=}ZqXl~D2&J$tiPEtyRCYJnMw}?uZPGArYHGvZ!n;sf8VyZX(`VJ
zyvKpH2mfERy?1z3*Om9bruWJ8-jX<z*lETUGAV-v7#zSDaGc_li*%8$k#rSZ2`neh
ztk}t9YT``KR7{fuq8l)nG6P~fz4zXG?|t6Scdfn8y`Fb|?|(m@2R&!6eeOADuf6u#
zt8Z*|)eM64ZJqK-@`lK9s-yD`!LOd=@%Zwc{s^|%vcVq82QTlXq72@gna=El1CyCx
zqT3m_UCpEN)Yc;ruRo6|I7Z_1$4BVE%{ZblLi;GHXXB+urUkho^*6CdPncaL7N$Q|
zadrGAF)GnF)F%fk0k?_DtyPANnPTY!$`=;~&%x6%!xwC_^yu*v6aWX(3$cYeJ;^;c
z46U(l#qoG|5~J9t14#K2=Nj37goaKvQ`;Q103MYI8Q6CEgB%nH#^CW=oUnH`zd0dB
z!P^x&DiV_BywB;6?}oWBHwRfv3XqXicDnpB{iE7tL=;M`(7w3j{r*gvARLL`KO>(+
zNa(#fplw^$4sBSy(E;Qhr>v8iHbqEnrrt9pg{wW{0CLb>qDo-B2W>VN$huAzC!<*`
zEMpvOJz~4dnClYYJRGmEjMvQ3>}d$6IW3IItc9K~c)!0nT$zbiegHVZJ^OIQh753l
zq4Y$m;FL7T+CNAv(vo!41GajbE8hv$u>{HM{fq+(+{-{Ji+DrE_DG#I@+eiqM3s-y
zzLz9Rf1S1Psc2iwalR|xv%Gh-4MQQL0wWQyJ3<g)-jtIlpJW&kgAOGR<#BI{tEpnl
z)36QjhfMMAit!9c@TR?<y3%NukJeFyi#e6!2vl}@MBh`Ki`BFDlp)ECA#p}2-4P?f
zD`mK-{jyEfQWY{Is95zC@~Xb3L5T{IW~NP*8c4Eq`U52(ok^1|TmEcuM6WD%m~%O1
zoMtTdw4kJ50q$6vxt0#GD@u?%j^u;y@9I8#Z+W#eVHuQF!>sC+Y`;!_VZ%DbHzE@3
zu}bU&`SsdV<Y+q}q2tj<Uv|R{_IT}YPg27pQg#4k<bG#jLgy-X9)Z5!SMS4}H#|$E
zQQHB<>+~T?{nJ6@?{5e~)vY3rESBj$r>EX6wVJ^lwXZD$$~dQRZsdN+u&HY9^h<R|
zu~nv(eW2`Qx?NQiRUJ0yP&5~8k}_@1UHSQVE_<`%lpsNH|GGMB$iRvVQ=#7h>!Z`3
z>wOGY2nEi&_Vwj^Q}|Zd%gbJ2ndU00c*edVcZC(?{y5j#KsK)3Zr|9<mGUA}^se>*
ztyB7AS59VMlQ%{AecpD|zNwC7nsf<D+!Nq5>HvenNE42v4dpw=SK2oh*>Sj~+UvB6
z%t0l2Fs}L&Ihs%<?1OnXh&&DxprzeydGGrB>_c_;UqhmS(~9?n4;+LO!^K|}7k#4O
zs6>4bKC#~`jKLxg#lCMJu5-^O0pa3XxZuEqD^8)XZz&&|_=#>_a6$`*<$2KQ&ly#g
zAinO%Kl#==V<SYbj&%1-I*??^543yZq^5sceX9hg82M#yV0?)pm2nH&w>NnD-|Jw}
zALGxeU|DxmOzyCKNApWXM9VXxl=FTNM(x+y5UW#6-$0RR`_4KBLdzg<_8lNOXCc_m
zgDJ%s=yze#w(m-;aWu4ZSw#2~3{N~#gKR`5`$#VHPJ{u{h?}*>@bV+ot1=X;Oh05M
zzq_KC`u<&3K#xW50EN)Ib=aUBE#FfRx}T)$BZPq*ARZT(?on01C}8}(bx5qf-R(J)
zXdqg|6#9*0z`n1}O8PJKcCRpDP^nIo=(6vxLdmTwZ}HNYA|+(TejstMZTo8HQbfDr
z)|*$@-eyIq;)QbP6_ApL99<5V7239QS1xp7gwN`~Jy{_m9cp$m6AmPW-w9A*`|1aW
zKRP9*3d^{fK&tI;?o&m?6mo$DkJp}By<xBo7OZ}Qb{j}Y3$b?TBX*z=o0zKd)|5ON
zJF3%@R^XZ0=^b4=m{{0F#vYf#!lY6f59Rf>$63-`2jPlz#3?gH0}HUtmv2p8GmjK_
z!7EJPD%T%ndu*ZHmwS{5@t9ZGyn=c7!sE+VY_YsarLn|hByYrS?(~#BNlq2{5%$^P
zqBvFF-S>51KLt1G9LnvOdxKAg#epX)S|ERyatNC;!r75>mu4qzsizBh!&8uHsQ!vw
zp1YDkjLnWxtIC$@jIqqv>iK)bD-1b~Z7MhJqjjY*R?SHiPlZU8X>*Znf&PML$w&F6
zDWD%KztmUoMaXREX_37XJ|d@IGxyS$-uNy%-i#p=rH*L>jkpxaNYFBB(^Jk}xs56o
zGMLE)J5gbQJsoOs$}6lEF(jwNPB!;tITDg7y~2WC!@IWMPL=z_)-y`+vI9$@nKBmB
zk^O6@D?zTt%0Q@k#cpwKKUJQhIafjh^WeaO4Mo%lm{3+?EO+{KBJP>8B^|^IaK&jA
zk>Ql)2Z8N`olTf{lw$sIAq+dP8^9vi>O1VEhQ-0OVz~`3ShRC>Z4Y;QEvJ>ELZl&L
zszqqY)8(ZLEQ7FUpsNDYUUm1@wRXP#K71Fn;tMmT6^VrH<U*katz9T6Hkvqm#p#K=
zlMa${pQ%`D4U?KR=d?u;-Ra4$U94IO;cOHdm~jAw<)==41)&FWL(Y_vb=oe~;Y@dS
zbbIi!aR;P#F7L%XPN5>O*0aln(lj%QcPtSwjpQlw*na0OaY{UNoNqDK#iMkkjI7Ak
zm^ZSP-*KlW`CMovnFb)uy7)qak0ECF4UqPO6_4xM(<4qR1~wp)fw?&4O;K<5Lv=H_
zfBTTr$h|E0udCEf1`D!dg%E953kn}5<euwTnv#2vM%W~ugdDI4+O@je;-gaZ=rbfa
z$<e$y94vpYAw@VqW8YXVlc}^l8DB`?bYMkNP_ep@8H30#*!9u_$unDGkYXMR)}co>
zjW=}Kdp7B8Ap7C^V0f84O7qJ{9az%oEK*=b0zccJOp2hGcCkGqdvyBC;t`t;Gl%_1
zffdq}-0l@NuS815!69AtqxH8iJjKdfVlx32V?+K4QCr)SdQRGp<y5xvKucQXy(dEc
zquPW0c;0c10%yc;(kgd6txb_PF4#}hO?ZszIV!8+K+@JDiDVRR{gc&us2auXX@X8<
z<pWxGMQWVbPt}{2_x4@34zw_6ZG^bu8Px)`aLInU64th%^-d$F4}PII5yjLp2dwFz
zX$GH#C$eNL78IwaSl96UF%63K4sDHrH438rZ2EQI)OD`X*0P(bl|NT85lljJblQRC
zxG99LA==NEoARkzvsWBg84wF!EI{HM{N{xHLg|U((E1*H>0)J<ogPUwd>8A`ezA~G
z29sbLU|$^&2%Y}K>$Wit;j~|>sNqYMkR|z1`p+S!M{I(#qifj6FIV=W95z1TaM-@I
z38b<wapr_`R~bS^TbyQ2V*S~#G(8kJWCAWqVwEY!>B(iVt3%k@bPIL<R}(w%H>?1Z
zYgCd2FyVNuvZU?Tl6SxT6)$iu^OYe`^BF9Ru$tJf*I8LJ-EtbqU;2_fqi7OI?KheU
z(P4ENB2;0rip#8q#BRS?XBP}W6ju*lU2tH9(a8gl$WL)6j6C*Rl`XEgg94<q%9kj`
zAj*0%*->L$T+4pDO6>^S=tdc`2e3#?ar6dUL_UIa_cQ@;_B+iWq#VXW9azr!9=6t=
z{cbboiguRyD1^wX7oDD@(H=2ytVa92a_jtZS{|a7Crb9B(<AcW7=l?2y;W}C?^g&#
zvf3)2<3JkOVw9zgx#16zhi|@j>n7(qrmYMjUJ<&M{b9w{Ms{wel}l^m#x3?o4Ho(`
zfwX~$XRVTfbC)DSw8}m4+CLtkkZSx1mD%Y@*^xy#0i*eoW^#2rr;JI@IIx&5Eev8c
ziTKeJFTp5PZs&~sX`N2^OU5%uY0S_+$ie9k#;C1Fq6f#U+MiXJN<Q%(it;+Jc#rI-
zk}z@?CwQ!EfPY@+u}A5$VW*kub7XR>9xL4RFPd+?027{PLlVb3p!^(E`vXqDxCNQ&
zO8)uFY_(>ty^rstRn}2@F;BCpB4=JbXMa`SjJ0XUxYI0ae7@P8V)ff!SM;$`WSUeg
zC$OBq+izZFf79S9l+@Vd)BYIpwbL(=Y{Rxq_O}&9;^?q9>J>P!^5&D_^UTyu0$J?u
z>O_JZygN=ikS_MBU7wRn?C<L^c69fQ_nf8<o?@=DdKDi#Ix=PdP<}~11md}}B91sP
zp=~{(gd5k|KbEiPFTYUckSpjZQ=NQKRm}dWz*H|)RjZKe99S;N9h~PhZvWiOJjv66
zHi;1zu%%0;e8%Yy9<Pn|y!}i4@!hi+(41!AL?Mgv=}~BapPfPb|JTx!4mJ^<wlZv%
zPER3e!kPIE7EZ`S*+TPDVC~<UXL7I%ze#~!Iz2r}q|0#*T>Pzp|Ky|EzgIRep%`wA
zbPgnr0hB@@Yl)1(^OQ1FvX107Eer8h)Oa9G!A$#)f;S1-DriE6kONa7)jN0Fe^#oR
zqB<&#T*d2=qOd6bSK@1V@0#vt&uJyET1OF4jE>Xx-(`4W|FT#Yi5G2|o-(|?12qGn
zrQbbfwooRx=uq#azDxtdaw#(h<|g~7a$nSTOm5&6rr=&AXhm7sM=z9#K>U&eTSc@6
zOxRAp^oAR5c<@2{m~yX38bV^Lh~j19DXj3Po&H>CZ+Xj2_OaQZ1|M;wG|H&@Tr<%g
z+PHn(LXuF?fe{Ae^U>B!PYQMLB&8Yid;9q2X;EfGTNVT-UuK-3<MhXGr2EOrPk^+2
zLWNy_EXt0Yxq8@v6~a{{orqK(AXRJs#PWtT$fjPBT1^=1o&MZe+7A;yIHBrbpR`c-
zuCb`q;{cK-My6XkF%^Gu{emEKF@%c_Ov`X<)^)T`X$E$pG1%6fUyPjoeB{j79)uA4
z)bhyuOl>Gfy}~4MYbk3Te%eATU&s+j#%UFlW{DWmA?(xZn;{{sx2O<?;X)o%0wtXO
zK$WCh1szJ>)mk}W&s!*hDkPwkm@JW_)0T?vBPWw{R}es^8>C%6|3;M<`;2BP`aC2x
zC3<r45O=@-8pX8jGm|9m-gxg_&UIpBNaxaZ?X&9F^z*v-PUU!M`%%^x#V4J+^4M1@
z(qPP|Dp$&w_SyMpi9BAeuYCkB<L6YAfeZ~?(T>0h^-u;7nAD*tmSdk=?$#@FKi=;Z
zM)iO0j=%PKm5M0s*~QjDN-Ll;p6&!>WuM>7@KKgoWH30ez-8i3;yK%7NH1WY8To?x
zu`DCGbI1~KV2MR#-9=@=FRV1NaSqXzmL#v<+(bClhV6^$hs<?3*L%ikmzOFHBQ{%B
zeOUhBCMWHS^Kom#&Ed4fHMYzW3i>5^J)%>e%JuTz-iUo^{ebntHZ2XDIz2OC6Zo<^
z@Q9uepImZadg=ybQv32|kW?~*>{$H{EM_D(I*D>fAZ#rQ_7xS=1hG@8NHJJ|64T(F
zq67B){2U}y;yJyvBJ{CmFQ}`dXzS#shKZ~k>ge<*!@&fLYA;pFUf74>PGb*&;Q$LH
z%2FWvJM{vgNEH%`YrCNipsX($d?g$#dF|1lqM%mxVAy?8d1jayneAMesAh$M(qE7w
z<q`%jH@N=VSC(H6m`~sr&C!TXfBaIFi720Sop>Y0ahg!p7Y~d>Cb%>xDF651_4jh(
z6#apZ<P%Zj$zD?5hW4g!V-fK-nbCvWIQ{Qna${+n{r~T;-B|hb7HUAvLX!Y;X>6&7
z4d0YpS-M4Mrt38Gh7CIEY{%@ly|iw0+qyfQMv79S?Mc*-mz8nq$Vud&uq9~@i|fGY
zPsJX2lKB`gCk`6YUtaE#+yA+`uDE=f?k_oaHH%Ukx;Q}<CUsKUEArFAwAvnVnuTM@
zA1Fbn?6UUCTv?Cabkj}FbrMWk_amQP1ar}D&bw~D`PMs}>q1s3M~l;CQq_vQYsIRY
zS31`>Xi`&iQR<*%j4VNsMJiQ@LM-f<-BK3}l8EA?`Rl-nhps(5MxqPR-aWfDUog81
z*FV=mX%P&<kB;Cuy=`Cy<Tryp;}t6U2!@>gf`;^@pNrrEm7~4A(rD~u<l#ou5$$UX
zlFy-E%E29X+-7&w7fZ#Un>uV>V9R?|O^A$bX|Q+pU&fM@#W-x6*B^I!0ykO&vb9KH
zXL8ea5L0|UA}Msx@ydoA^6ZC*b%|1|F*6KYC!}##3{0D_1o0)F5(h=ws!GBf|2Yiz
z>p74>oTB4c!sG%{6S{KbjqTlYPD^(z@5KTzLbmPnuAY5WB4FjJRnBF>V*T()cO-u8
z)ww#NSiaMc=CHi?E=B6Urg>}D$ituGz=DWtinGkFd=+d;KA`>f+GaY_i_N+7!R92s
zE4!V3P0F^l_u1>pH%VrZ(=rav4Kb3Q{EV%{LfTg+1&xhA)N`&SwB9#MOuwzJo5Bpo
z6_7DWJD2`lvm+y8#1`6J%?KhrU5bfUSlZ-O<n@9L)lUg)Hc=;#5@nv3=bfJXl-IYW
zCv8phJdB~puP7uAtgwi_NF@SP)COi$W*f=a);7e7B{x=>${0|IUjxSwM!$ATVy%eW
zulwIfcbJ$NRc!{O?k#J#ZnwM3M=XR89WtH+OJRQX&^B9N8=Ho<IE^43J}Cqrj54;N
zeoP4CDfxv7EI-!i*V_<3C+(gFHT_5!p>DzH5o%=5g&bdBKYB<-9{K2L2UHA1DxS`+
zy|Fw?@irmrK=>N|g3}XUseCK-T*0EesTpV1az24%1gyOC6ub83h13-#buU4H(il^^
z)-EV5P}*at6ud_)OqD0*o!=F9DAgq$T(U}DuzMS9rBOrU1eUnnbnjN%QfY48Y*-m-
z$9kB@!lq0{amadHAY!@QR~{HX!lLm2D=R-vzzII9jbh2RmZ#}s&+xIz4GyIs&~()4
zk6)%$DU0GIeQnF{LjFC>|3gk&Iu8YtRh*do#}V6}xa-dDZntKgmKNg9k(S-R5Nh%x
ztDUPruS)Bjv^O;0uG59J@n3OZiHzHBS!oYcI%30-VjQP__BRZtKR#Eqg^eF&?ji$`
zDgBL=G}4Se;{bB)^2Oub4jR~-@=dEBnZ-UsqcFe=QWBxtMSF9h(ov$2xK8;{A;qmZ
zdrL)e<iFMl!3^&7#FMmt%h*Mdx;<E6pjv7*;_C?qmJ5H!IQ6^VTAl|jNy|c%mRlCO
z_u7*wOR=ezCI7bi1Zon^I;|j-v>|I!<#Z!fhPzRG?d{Ffr<E#5BP{GtF^TlGUs4*T
z#)w!GH5WU)y`%iTY$P!YDJu%^Pdfb?_eNG`YkFgAj#FK%RYn!oGQLJxo0t>P0O`rm
zuYLpr$c}=awA@D?w^vx2Q)TqHY{N-YlkG>G%UsY{OB5EyZeSw?rhQe~Kb{e<umupS
z<%z;-?JA>fCk%SdX(k()L4PM}-F7yU8&$VL=t}jIo<MS#%|d`s!Oq<B6n{Eeewn+t
z_U}AzY2N9{m&nodFu|Z>g{?>Raz6F~vac4YOrtW^ron@|M0eCuYV9B<)H-!}bog)@
zi9||7&DcX_^p!y=qYPNtO>qoBR@xiXkEDn7_^|D&kGt)TTb)KhDP<so+a0p;ii&|U
z6^Gz@XT0`Ez0#xOuY^RKs7MIv37>JWmj$mq)eo9^#@fCn%ez%HS0SMDSC<@Eh(y9l
zN!1JtDfpqJgI1@5b;~=I;08;>_9)Y~Y>V_HL)pCT0h?-GGmqY?n)v4&Sea9mCaEWr
zE8tpjH%INA<u^}2yfQlQ-+`sYXERaTV{KGco30bQQQitFfdP;vppx?v{d&nwum*LV
z>-?szYuDIJrN)g!053Ytm<2heHBA>eRbaNjG5~;W|LP&HF!4S_1vu7z+g*l|vVZma
z99CKI;3f3uvl%#8#-elH_7u!16t<@S=$Qwdo`M)5Bj^$IsAxqh01wxv@f?gDAx#4-
ze3Dvi!(%pA*)9~?LDU#^AlestV{`UMy%`}y1;1#=3)8<zmMlyV^%?bAO7uB<SA8K`
zh=_!3+xsxiI0uG1*xsF}=G5)3b6vIwWf+T;I3AkcpRq@)HqkFM*z*R+tm>pp9bjwk
zX;73hXXHuAR_nE=D%#BnO7=D7Hwz&|iS<dXM{ja^QbQ!tl5NN<Ev(tF#%TySv!k>2
z-tvHyUL`|_Kj1xZdbI4v=BNuhVvkiOaCdj4=d@zD+DGI~i)5;ZM0mVGuIe=ESWo#O
z8dIm!AFEAC+EN6sC<ID#%C#y1WVT7BWxub`@@{rcu9FGf-9Z<Ae_a^u2@?DxkQ9)J
zPBMSDr^5{Jr+#gjNnXIm$XIdIfraGW8nF*N2YDGY5H*yYo?JOYL)-1^^3^cUv&(0k
zR?04Xw9Wv@Y&HZi`}*>>G%#p52OU`QV{{-QRwnEl8vJm&BqKi(7PiI<>h$XeTjIzF
zQN=f=RP*u|Iam5}&JwY2Dz6HmJ3iuRuP|{n%!Sv<4YO}9AFCOanOQEM<oNU?Ye6+r
zp)tIGA8bCY?`C3raP;`0)1OFr^X#;JsC<E>QmOUG<qj-SgPo;>$1y@IvGzpiNscP9
zFarHyi}WO!4zu@6+lTWV6enztIE`H8#6J)LVc*h>T8dh1zg}VD7?OjKXzW|deOiDT
zugX~fi$z~1EG!^N^XvpyPuaIEJgcG+!&DUSDP$w^1s!N8Mah-&6wDN7^ExCG`}R79
zm)uD0U#A(bk|tSff!a#YFb^>ePuh1>)Kg}+I~Cj*kf!#<uXurdXQ9ujap$sdMlpm1
z#uiSGW;gn-f!S4H*fS7$;MVC$tgch7UHeG$Af;2{<c~Pas3{PND*Z9c+|y+%Df%AB
zQ?8=2RS2uC-!11uCXN6qd-(<E7|KID`ks0lzA!v>Ix_)KE?!I&Blf-JYtj#u!kq|I
zAwO*FPJcW!55y6RS+ehIM!?xaU^8-c21IoF^FyvbhU{@jikW?XU3+(v0!xi;2N3jG
zW|O^45zL3L9<U!MKTpOjE_lU(73|A>6rx|OG}*yqx7l0woYm~aN(OZx@r#r%#V%xR
zR9)cE*X+sWEs;bYf+Mh8R1aVY*jJZ{+`f#0gP+-Q-wW5FCvgHzX}3bGw*7U?)Bu+;
z9?8>z724|uZG~#e(opg11LZd=Qz(~<R0&}D+2KtfQV6lIsv6S$4h|qf#zUD>bJik?
z$LTN2o;UGJ9lEMTa>WiMEw5g;#ktftb+*cs&DVv!ZE_Abh|`YI)x435DB=qRmWoDb
zk`64VB;FDEK1@u?P}DTD#Y*;51fNYi%|r$(L-BI^!)SIaI3eZa?Qs1JO6&BvpLTr4
zLF}ih3Z#VQBlVt{j;hi*?Ih)q)+3MDQrQ|5ws6*K3mpn=>}d0}sJ9QXKIy>p>-J<%
zA-l)w9r%Tnw}Y1Nz@$KuxX7Vv_$7LFyb2JwAlfSAuBae@Vi+qqqvb@y1;RKKwvb>q
zPLH?%o+cD$N;}ur73Oye@vaY(@d{$aM1QK>2hNd+BvTx5U^1mF@4a`Uoi6>6a!kLf
zL;%Z~iGzZZMdpSs^XCP7s-p4rItgpqoA#cusA_bkJX1;}ksl$L11qzgv6op9T=qqV
zNASbhiWr>RWR&3V<iP-A4z)&e30<~xl|UgIZDzt#9Fy$y#NB#e?pWvT>0}$*W~X#2
z2~!Rr=sV-jq=K-$@{q%Jz7oUG>TPSCR-AR3()$72hNt*{*)HT=Tf~FroR%<&-JXKb
z_4mUnk4;N+I!)1wi{-o0CmJzT4lF(+Ppb{x1w^3w*`<nz-YkqR4y_q+fLd@j-gx6p
zH`(Qif%K|GvZ(gT!%k1qJVrcuYVhgVm2$VlVHJpz5+gD=;q+AON*3qs=j{iR%7#?1
z18mVuiAM(t=;YELs#CwG&1Ted+9`STI1daz7z^QQq2^m|aV`t8)2Ae@A}a-7%K=fl
zjY4ImL4wcfZMNKuF6y!Jy)t)m>W<Ts+;wAXmtC*e-rU*Mc3Lr%bc+e8m+Xh@r{1F#
zI_b326M#8pyqhwCJzL2#>W!z-IA=q$(9=$T839$vLQ0Xp?Hskxf24lE&Mv2c(aMKY
z>X|}SENp7`{Ae!!wc}K1&`A};4kW&F0c*L!J)gOH#D1*uIW`T7GSG}SmfdSvTckAH
z^Y-Hd^QFu^isw17#K{nX^iPzZhR0G8vci9nlTUf=3HL9{<v0(GjnjUzVhJ1lFiVbh
zN)PE%z)Ta7U@S5rBz@6W?58SStZz%Yn|1(&o-eXtXq%Uj1KXSZbbj^nUQYL$8;z(f
z2Nrh^hNE#e>e|m#dcui}u7haefC@KKJWG1XWyO}>Ew9hd<{DhRar=7bIu`&epoqx!
za|I^Xw#-`?WF1&C{yi<^)Sn+fc?w-cVsd)o-7H<Wrw&<^>$8blv|p&`-!Qt#Y2;ey
zOsXb_6!>3E?6xO%v`0o*PqgECDVu4EF`8iV$TGS|?U%~8DpLY$6qCQ-fh7sXW)Q+h
znY&-EZy(ya-D$+ovl^*b+5!8O0xKTkAO&EFdk}$r6cH;GpA+`0b+Fi#lxCy)ooTrX
z7VY%MFBd|eKg5|eqzS==Puj2LlJ8OzjmxBPVhr{3qX?0|UT7-Ar+8R|rZo9rIU46K
z@zTw0qRD=vDwcenLakKU18gD7GzHc%U`Wn<#<{o5`9*LBWgravW-0*7dv|ca6J_AY
z<%3%T)f?8W9$IU^m7lz;JMLUYMAnG?c13mUAm}|3V?q0VZ+ZeZ(lsh4`<>?Xvcd*M
zdmpyIwxJn0Y`@z)I}?c%$(Gt1JvjhuG>YWs9B%u)x-3U<A`#p`J8&<<A&U^loC+pj
zeX-v!JvwlNLm-%AxpEX5L{AV~&ZNX~h2%+H&{8;Ve^B2si}|%T=d{arQ0)=qk|=I}
zSe_weN0sMPh$8Glygdp-_i3I}Ktd4uqbf=d>1YWGT{{r%B12@a;)G#~_~S%=XRhU3
zCUwj3#zS#BGlvRN{*y}Gx8HV~(@5=cYFgz+>`xn7mAfj1lR;XgKQ}HZH#^~te^v%Z
zJqmPSX>(T`SiCz~ZBS03pFgh@Tim$H_ee!U)*{94oFtQpyYeNNEp(=?{Y8CA9OIRt
zWWaJ~xQjGISY^bf{AK=<D9_<exNHDK0{WoKhR+HjC;5(-_mBFkqD#NP$X_Kp6ZKu)
z=d==~QWBVP<N!Vxk<k9SZZ*`XK$D|+1M|X!HYqBxH{hgV-6_zsiH@_ssXrYWo90LY
znn5eFoyzh<c|QBwiVh|nd;Ww2NG{9SPLuX``IF_nIxrI>g-(cQU|7{Bd&AQmipc$a
zooXERUEGYc<5eR4GeNp1>>ui@*0f0hYK=JniCZ_Xywm<MFSmB*oXgalXYeev?Vk!_
zipm&z2T?y)kNb0b<Tjz+3V!YL_RkG!aRocA@+~SVlRN0YRCGfVP!_rp<}XYr3r<hm
zt^HOwbu7dGnmlNE@9W>>T<2~Od#yA(`K1+jBlY3D{ag9HCQqwEl^cLnpH$?`_w3*E
z3yA<j#C2LJ5Iod=$o`{znv7T(Vi3MAyp8mTbUBHGZECDFVgK2TB86wgqRDXgl+zQV
zsQgzO%YT)-6BZjicjJ+BdeZZZ;3iffUeC-$i0%cjkU0DAIx^I<`Q+>lEcaTrE;;V9
z218RD&}=cLcxdar&Sl|-I<#BLaE$BfA^WJse25fD)hNw6H$s1Sr=0#o562AIM>mgA
zDU&pOYSCSAdh)ido3`1<l-p8;N}G3t8aO?bG$SnK3gX<yHn7Nq6OVRaF*F{IAe9rX
zYsHf=t@d&Cb*l)AppmPFx~vb|$JZ9wE)z~GUibt~<tJd$PoRx`LV5T+b`n*x)Zg?q
z<ye}2S%;u685ENsxPl@-u?#O<ANn|EKKQWL9z7>jEs32RePW-qSlIC_JRf5W?ZU++
z_(d99ndwhn40<I<ayayv83z($5LG|L`$(d(Pw68FmbVc0Z&JZo4+_k_?6p6QF+!kB
zj*SFipV}b(8H5_0uW{ZR7&I0l4t9u7YwlB4L5$Ubm2cMunX++i!k=F6#^tGzW36#u
zF_nifL_$;;zJ<*==@Eha8GBxxul2K|U8fay%e<L6-zvTpg>?|3j@xJC8<zKy%b<WC
zBo!~PFt@9U2EGjY%*DdUrera2ngxXmA)c!k&1W@a@Ju*2b>>C-Vzt>Fa{9G{rpW<Q
z^k<i6WIC3Rx#Sh5sUc=e0+)TxV%g-zI0gwDwgV{JOFs{or2!d#Zutbs<w}eSD=DIi
z%r7!2(;vT-kR(dX#3i}%KCkp>QOE$T{WioKrzeH#QJuM{VlDRhiy^PAzQ?%=Jm_cm
zK>2&czMwoRSr6nc282=VlGCF_t<&yz*%#(KfnQu3URs<?FpZM?Ma^SnXpe{n2bKZ|
ziErp~_QlN?UtnelMhw}M2y>?=Cnt22ka_J(3TC7j)k;^n5>p?cKQJTetGU6kdhJU~
z&tU#~h<{E`Y<YLyVPCdbwNLzpT?u07O#tINWOL*=JZN~{zC1^<dIYhE(^8#wp?&$9
zWFn9-rHWKvQMm2eZO&zxaLE%@FOatO{O9y&7UToIf^C7Ge7SsxEHngldqIIIqCw@%
zu+)2n;dbY@3TnP=FRTb;aa{uaEFA}yOp&mKDZ8OQ31(L6emk&4A?9tVPGS-B%Ouy^
zi%O5OOXT*i(Rvr09)0EV-fgoxdbsp3(b-ov>*!Qe;n|PLoH0j88J1qBziN47_}6u)
z(~HaJAWMi0ahP0mVA3_FW+!*qOUehNjcu^57d`m_Gfi}&-B>>tz6I%q)+crDiqjL^
zMC(V|4~0IiEo8sS+u2QZsg87KcgWZUAhAaH%>jF9PX4tI!eT#&w}!TQ5db(b&0bbn
z%ob$y8K)T&s7X~3U!)&<dHFIa8Bqg^T~qxRot`q?nb{|$@D=55t#gGcE9@y{1bX6b
zEzXd|uRdk3Y+%&G3xYZ5zzX+NuroH3=;?F)n+r?hJZ}{2X{TKrB~$^Lu@#F&6bXfW
zKn`Y#`X@wkSG+{aQ8n+fTn}POTwkzTnttj8f(UAF^LIWN!y-JBeo0c>m3ueSnwYX%
zbH}@<wY#+~Bw7cOdctmWmqV1oxGg!@YhUSHiA**TA8EIjH^stdF&<`%nvY}?sT=#9
z{uD~Kx5n&_2D@1rMn&wjLtS)w%3P=l<smrpF}t(5TXJ3OdmsVc=k(-UfF$jbIC5=e
zg~bBY%g=d*Nu$qzwx?`WMNQ=VR28L}AE#eDGmwC^Z?7u%EAtQ!E?=e~5M|iNZ-3h9
zPuX^Y`uLP}AzE}XH5BccL_ZQbE+1U~*Ml+}_XY#N+o3=xEGVfWAFx+f{I8uLSq)7W
zKnSTK-si+GfB;2C2o2}#HRW@qSBmeQM^r5#ra&%v?PsHlA{feRA@+ieKyv3vdu<MX
zh%+o=<D{**X6@~_*VT^=0a=Ve-+^Jt>B%*_zSWtwudWk^2&Q-z*<`fvPk8OoW?SNt
znNnHYR+r!CJ_QhHGnG7i(CLZEmOYjF46u#^c2`4n8BVS6VLeB-VwuY57uB^U37xi~
zx};dS3Tc)KIv9-|(!Fz6vPd#;3HxRAwx+y!5kHxpf8^?t7Z?g5ep+r)4%FI8QjoIT
zQ)v`vQrDe+ZF;L#th9A`IZVzAmocssYr}41clSRvoRO;HxDb1T@N%@UBJRpvBRtOe
z`&3cZmuHGyi=0Sk^$MGhl-Kw=+fcqZWi7?EDdVY8;=J|?+hLx{915~LWJkGV_f%wY
z5@*JzJD^}mxz=*?8Njl=KIs{2OF3u}Kv~33v4{`G!OI&L!-2uUHl*ZGU@mBo5F_A8
zUsQVQ*{0@~3KY(`U0}KcTZK%Rn=2lonuq+-#Hv<UsL&bTac|{Lx8JhLY2*hL4x$pP
zTk4Bj6LX_ZEBCT6Y{)pJl+SsjBZ;Et?Y@SHQ)SXFgt3v6>gXxI)TW%K9|fSu(1xeg
zwpK1Ojcq_V!VV;5x1qTy$#>iG>z4P9%uY$Bh0{1Nv4q$BKmXbG9K!=04pN(RS|Y}+
zh=cbxA0;IV#Y(;wh3+uRs@6e&GN)Ww473DX+Z&QAtlmB}<XjiDos4V`Sppv@PtMpb
zMIj@@r0F7PlK%1=)m#hnsJ$^aLdvXja_NZEEaY)Hu!KHjZ_1IZ9*J}4N{g`Rg9r+5
zuK3yA9wj78fmRMI*EHUf*61{6SK3>eDWW6}Z$2It!czoT(IcL*vqkM1d$1XL+I?lt
zk}K><rzgHkY*_My3Vgh^eyg-C7AYq>c!e#Z7qW=ig3sIA@)fhpUe{@qJzV0o9;cYm
zCJ)ixULVSgln87GRvaCa#^Iq)u{v{O6nD(oId1PLPfi66o2Z=laR1`Aboz5^*DAwj
zVXaDi6tsyGp&W?J%lA&$j{N4)u`Z%Fja-L&I*5qF1E>g7H%bYx92Mr!n)08f7IClz
z8_8M5QA#v^&uJFRi$cz0p~i+G?duX-Q)ms=t}>FU)wP{wp-W`%a2BZy6tTv(v%FnO
z0p3Ayk|DE_e)huy=+Di1?F7NmbM3Z`R>(cwF%Gd~Q2~hwCOS$71yGOXq>Ys~Y8tgV
zv_^8<peM$w`BS1o6zxV~q9TfB$2z%;N9I(2dbsDbg~4$JjkWEeB+!Sa6(keqFFK%b
zna_0Cgxa<%SK9L42i7>(TDmvg^q`G5A5>$&O5mv19;EN^94QnzQ9n^v11GBYFrLE<
z0F|%Llugyz8?ffuWJNsZ8BMjFW?i;n1%ba=w;@Sc2!h*zAVmu2hAiP^$+=6Tnw*u)
z+q0=kC81i%6mf}<)LtHu+)jVYFuQrwM!pm?!BFg-rALpIPoX)%%4<*GH6@XpGy<Eh
zZws{}w<b6o>tD9#r=0#~Nr%M|hTJQ^39kcEhpGhY*cZtrXp~NWK}DE3(cF&MO!J#4
zSTSI~40*=sk5MH{NJT>H_Jqw=D&*k7@z!jw?LgxBSOOGng{OSUc2}%~bTl}2dg5+Y
zxjZGI4DV@RC&8ZL%1=44Vmwz2Y@H^>7^gH>j98Iq^KkQf+5F()Fh5|$kjq9)jhdd#
zH49kzCdiP%Rm0hc45AasFd;j4CCB0D5{%j-_DDfhs<&FMgebAN!sw69ntm;lYfH%c
zm+f6i_}huX<T7v=tB<b|b}|9)-Q{~j0*ux7v;#}+V?9T8JzD9XOpe{;O=HRjl;hbx
z21AqN@SgH~B{4^RPYLAgd}p1WG9#D5TzK~CA^VzU^z!(#cLuep;G6<J<$fjr&$+A2
zPDT{$Cs4@C02jg7VW%bhs0a^vxS;jDxn7iKPk5x$40)Er4X5p~I#8^TDS$uazy{cm
zRK=|4teSQQ5U(F^KH)GQCkN<R2UZrDY+45q(<Qu;Z)NXW%=%+88)qDS4O-D6Rn~V}
zO!rBaL6kq~{6eg0H7BjUKS#C|5t>kETGAib@d$>^>ucYA^|*7Lj6Z035rs-wqbheJ
zP8u!~l)4X;&rW@To~swU!bEHAc&PPH+t<~3zL&H3XatcD2K92bBAaY#Uh<10$VxA>
zuP=Cm_M~^h6vQnv1Cg^Xz4q5l6u%4oQn&S)7wsD=O{Mmyz$id5+iB!T+Wf}mH9Gw?
z-Qhx_yioYaoPMG2<~wh*Zz{LSoGE6MB$ZJN1UY-tABYnh)Uu0+IoU4lo0D>Oc5)-9
znVvj~Z&Hy@p<Be+mUl=sJHxLfAZYbb>k$S1ywhLA4B3ctyliL8HW6>ySmQjQ57nn!
z-G|_)h$(_iu$jt|k5$>}mkhCEw>?o&LO85e_!-nXVDS#&kzgnSDKctZTlHBftzw_{
z;Ut;XD3$lQ&c#ZO=>+FFjN7+Vs*ZZvAxS&1T=JXSUHjJZiK?=uFpxZF_C&~>)KT%;
zFZz-pfruUbt%0ESZ54aU&nHg}m9OUjQwSREjE>s3SJu{^o}H5Y3P2ik_RLO?wCy{p
z96#OOIX?Ta(+={0%0d%BIkxYt<A`l0g>PUvah&cH2QNz+_^y1m_E6<%(k#qSVG+p^
zQxW_~v$jQKidN0j0T-N}!i?ap%mwj0_T2?d5zq|HMbb&X_L|eL->8rVGl3w8a{N8z
zIZB6Aj;GG}0+t-5AIink4q$Zs-kirCbz+=r%}{4_ywkPstJAf1+g7JhXd*>ct+qI2
zl)tL-nq&L^@@9#5$;%eJ!r1U(*tzI_ATdusbqCAwsMC(G!@4BH1q_}*38?RY?M<{k
zfYAdxw9^#N+SS?BwI}QJjkO}UjCPs6u-;3*w0-5lxeOOTpI9V2pVL#2E!%|BaGtjP
z^)q0u5y{KPq-TzvxVx5ZJ;A3P$Zr|#c1NA-R0+4Vi3FXtgSEX?(IPFUtp&Gd)DBgw
zuHL-bX%s#mHTB^AnPYs;=F1E9$=WYqs17W461gITC6x*qFb&v3vZ~j;;APHb-g;f}
zI0QL`TdAqnY>Ul{`)<_`MZNY6N>^YH$Kc_*oJM8|gPHAl1Gt+7u!oGqBNa<&A%z(q
za$qq;F%$7uh*F=ve94xYyD!S@BhQKhD`YBpuCjh|!R%;dF`HTsb)kki2NGjUr&&sk
zlIC>a1&93Ev9eXB#p?fKALAWQ!hf_>&N;t`S+2`fO#68Kq7uWSP}s<Xae9)St%jGJ
zC?Au&gRqzoH~I&M&|irYgP>HGc{^DjuzACf(?|*$mOMXgr^*jB*<${e@wpWkv2SoJ
zyheDQ>~zITXL2{kd9#>+6k3&TAE!;)Q_aZY&<J?y0SBfxa@0jeo1Cflgf&Wn-#)J}
z9W^X2jZ4DL*5{+#D>e^vyaUSxOo)f7@!%xlukd%SejjDA<7g=d64kFVb|}sH1DcH}
zL0lgFKw5E{hq3b6)AdQiV+yK=pEr;PI@1`hdObUz8vtfPwgsnEXipkRhNVdETeJ&x
zR^>=!d1<4=ScLXKPhyX~Cgq;9_Dq3giD6a~vQC9b5YveX)JL+5<*C}uWR;YGBA-f6
zS-J`hsPv^u8qsf;8r(k_V<$e-f#q5VomJMojJ~;KmkTpaMKWx$<(Z5oM8z}c``i`p
zOu;aVbS586tIn0W8!GO;=QNX?H(-DI!TN4klDrsl5XtF!+UdzXbR%0I0lekIb&JBm
zbQp4lWf3)P+Gp)*`IubN1Et~f!@c%rr5r6V#dgSx7{`;^wPr9)JOLkc1!+XjdWI5V
zl>jfx3(yzfY`KyRtEz{SvgbeoG4kuf`uTDdu!-98Vjp(Bj7ht_c&`dwRagW;i(sxE
zv>&cyq3A8%sUow$gr3#gH{82<^+tQP+@fKrSeP(aVKPRM03X)XAE_cmq=rYh@0<hF
zX14>Qqy1=JFYkTPjm~w>)Lrdi`?2x{35=R4e5MX8C`ZVa{UtUEA<KTe=@HkfN7)nG
z;`HRaPUEkev!BSvQWuE|Gn}8U_%2|nD(;wWV`Un(pUe*-K5Mk?Gztq?!mARSET1F8
z(odD1`DGQz3rB64b$TSPjXXr^W!HYXe!|WU!3$2axax7!@5W)!wx3yiR`m%nLKly7
z*pwY`S?4nRg^>^9Ao0b1Hn+q(SKjViY75>zWj~j1)%JqniB@5kvd5@i#nbll6$U+&
zlasuG!SJZ=CU>*_LV2+?f?6>p^OWdGNp!_ZYP0-ef5yg~MheTy_1?j;#r8{$a3xPd
zESS2U3n62gcJ2!0NuQ0yLlO%VC1jUOdtCAvEOXj^xxt9UHPME`O8A}AlW#9tN85g-
zE(fU&nmCDWEI_4=lf&%v3vq8)XTO@*9PSP~mw~lpWqR~^`?Y-Z>?GlOBds0idqCo@
zF!u0hNwmLSaVBF8_*U(mh#Ycy5@(yp1F_%87G6+QPNkLGV{fcQnH~GhdhgWqBShjB
z1m@QyJ5Mx<0(EXG!?NG1^hK2=gk73iUiWU^_$K@924DT^*_pdtd*Tsd7QtTdBiZjX
zvnF*xZp|8MAz@u%YDZxR+?C5BGQQY9xhr-)``v;ZC8SBI!qVoz$}q`Co@(3gm3tP^
zrF9q~dw>JWWvD7jDk1y(4VEtivsSSYU`06UC!^O?TWmS@2l<<*r4uKO5bFmlS-_Uo
z++^GSuzYzUH8MhCd0cdQ^2Tkrg2wER>Ode?)$-sVFb5{&F7Ms7$Nsqa;IS^%XL}AH
zK29DDsidl@5h5;Fl&$zr@&`KgJj^Z*Am~WXmf=_RV*j+DUplGM@#US)>)_61FXFCz
z6~D=iUddifI=cN?=}*&Sn%&YkA6l-{U$|O&8&V@CCe!}B^yof~TB?Q0vmA1Ia)R!j
zWduYH{KaDAk18$%BVx!VkTlTvG(A8bSN#>~hT#aw=`WKOw$03vqs6vGyI^hLggM;O
zQt;*3`Kw%=t?8Z4W#B7?UY&3;Z+~52B2)xk7Wh$bV64K3LPpH`n>x9Zt;y~lr<uww
zg>I)>qhUI-zb#`|B8klS`&bk~CQeV1hIsg#{arG?EfQ&+hD<3BbtdfZ>+;>XY3l~3
zr46k&GrM!A{ljy<UF?WrvlXM;mCn^`*+13~kWU$^%o;e@@SGmKWW(ll_D{{+9SHWP
zC^FH;1*a!asUQwTAL#F3v9o`!)cd;EIoG))DfEkWZ~sz1{ejN9j?)sR*^!2aXM179
zv41U37qgJCS1dW9i&J#h=?^@#C+y$KC-PbN`KyE~`H~*tq%#2@wSTYs#{-=arxBBz
zcSj<u{YL|nfk2YyJ_nZUloZ!V`_Bd$5mLkMp{!vNcbuLSCvLlQmHk(FP#A~7%waYj
zU`eR#c_L0>OD;nxay-obJNZ5$HxvRA#vMqkn#5@sy|t~ev)bW8nQ&yq!V!t%!0H+k
z9kGo`SSPH?KI(Abe)ULf|07O2SJw2{+{_qB{2zVz+4W$F4^bFu$hGA<nO#gbua?lL
zPtIK3sytWDc1!^t;Tq(5+8v3jIiMn(tlf`E$nxIR(+p6qOCSaR5)OiR>|+Z|>sR4T
zidh3DyedzrXCGI_7#xC6QBfR?{-o2B%R1E_R;L{PiXD|5WS@Qf;h>hC<6Y-6Npagu
z8-2$<;c)I?8}HrfT*p_59;AU?z(Vnf38QRaXN2`QuDn-Zfo!cA`Awb<^Mb^*Pim$?
zk;bXQ219gup7V)Ewcia$8=rhQB!pML#<@yb+DJU~r&I)V;a)W283V4Th!~M1!@9Lk
zZ4jio`x460s$f!woSqn&%vGUV;@17N3PX{`gM#<ypaT=Vv)~rlLq5HEA4?9H>PJY(
zSH%h9B%J<S9aK2jWzTD{a)gP|+cO|;!E2BFH`4RSgyCsajj_+j+1QEidIuILr>$ju
z$BI?<nTHD(6d_GI&A@qx!;E(k{E9LU+h^6UxVuZ(Gp!sc8>#43X7JCh7`^R|TbxGR
zzz)SDREd1QeNHnus>c=weKD;~GKbTj6Am>=yi4bYO6-1aT}^N3au_7uTL5uu7Agu4
z3AlX>e1BdJXlJR_M^i(+^OjZi`T2c!QF49GX~Y9^VPynY#A4VNG}Dolikhe{kV%Mx
z)1O}%rXv}B(WrR!0sF$z6A0t(s5yW*bl&MHWGTtL|27rxR}}CU)mb3>lw-tb=IW6#
z5k~&S<-rj|2+A^8RUcD(<ctwl$&+xgJ=>R*2}rB8Ja<<dSbWBFY#99surD2$RRxpl
zSPNh=+@uI`jZBC<VPDoDN_pv`H;FZkZ^G%<27B%C>yN80$Cn>YyKEZ4c#yUcR38yL
z8wB&M&@%24u@jhr8Qi|2;$^bM30qDxzF%)aS75Z_h6<obG(NvUbDZcKkxlu<aweR1
z`a>@fY$?V!AJkpze-2h{dBNdQ`J5H2(BKTYQH+c7;tugU@t#53qlAlj=XW`G>$Bp_
z3v&p=T_SGUBTh5RClk*;)}0!gBMFMiIdgVH9V$h3$6HP_Hr1UV!d_J7Aa+X-6@FJd
zqgR}sSZ4HuWPez-Y3q|(#1igsnq{eyPrxRC8`xf4VDiD>{*p<^8yNDTlK)hA<s}V5
zkA>h*_(M!_3A=OpHCWy=+LbTm#<~UV>^#hb5eNY!rYz@M#R%jIJ<pm6WZO**d8vR?
z`hT>jk2yWLgSI%v4x7KdG?(&7XWY3=*=@Kr3TeG8hqJtQPj`A(D>H3@6}nXbWxDY~
zAh4GwiN!(O(@txE5lH=0I2M(B?G=@u?PyOpjliO8EEH`)5xQcpteBN1uB{t6wZbC3
zg{eA$mFDI``SMds*^+!NF(41{fS@4ci_Tr~Aqu-wf~4vMWTNAW(i35lh5TZEy!K#*
zL}1A9^3X5rmO68xxTMgb<j9hJ;Pl59B-imdyS3h3c6J2>%O(cm<QjGQ3k{`zcL|~Y
zw(^?+!zUwY>eD*?g*2C74^hO%Zm*lj*0mh^ODje!ZCk+s=fd=lf70$~zUQcR0w_{O
z&lCrhRz$zJ`YO2kohe!1i=T0>bElk~ogB86<sn+_ic=87$(0|LRqyvV6^3Iy>xHPN
zwyF-AK_RJvsHYuRell{Crczcb(j*8pepT&3ERvI0OC=%;oSr1sjk6CUNPksrVyj}I
z(@fMV#4orJRbcZLDlLWko8|zC=JJ<jc-~&q;P?Q>SSjWwy~5`4sOmgtMFze$*Xgy7
zu(Cgb&P_X2r3j=%U4jqU>k63)M><zRl?~JID4)8LBu-<#<gUaq1)N&;)#U^98%2&G
zk2tXS+<cDy$~`Pc_v$)y)@g6dX$EhQX{dyifiiK3l&SCPKRk447`<1R#--AatSB3*
z)DkY0TqQgq48hd5aQd4M(mB-fQDZQ|*1P6#G2AJ)B}<3`k~yH&=@FA#SrA1pR07BW
zi0;ORciFnc0|8h#GM!c-c43Vjf^dnuE2eg|kt%70TT+WqPU@(f$)|+lJ=g;FFI!)q
z151&O24>^HDw-PkFgO&Hr&H55)XB^fhekTXybhq$gY`I|9!dAhJ>@YH85L3^E4Zdb
z-qB}x4gDG$Q)91x4#6Q`N9+&z*G^CD>B?8R@+#X{5rijwcNdKhK%wSjb-0XeQ+Zn2
zn$p6Tq7*%%C5VAy@wU1CR!lROyH^<5Ua>u#zxuS@Tb?JWQz=vD(t*VRIkl%K%aKeM
z?AEq4zj+E}FmrdqZR_+TcEjqWQb3ZypAIwm($&*;U!6AzY7nL(zQQ`r-}D{9T%Ege
zRZP$Bn6s_rnVe${7x=&aI{m@ER@`FSQX$##h;tRF)10WViz+;9FAvhTl!T(#bGe(e
zlb&<>6WuT!`4XsfWB2F7-~O(5JD1U|G6G2H*QxkwZy0!u7LX1f0Tv775EW2Gbxmzv
zd!Vk^;pz6&jMIz<N`_NP0H<Ht8>_(6dRQ`m(@u7`ef?SvXM9uLMPKsZdz^;Q)SDqs
zXU5)KS0AgZE$hLE14`-XA*^ql7yOp;WkD~@i)vtom&j{R=ww02LW&mkgNcC7$gp#n
zN*>M>du#nL0(rYToMue?fDKJjPsETNyZ*Snt&CM}4|X5L4mdDP+_sq!dwUKABwV6{
zR<7=#l630C6lo8L-RBwV4!olojSMXc`c_VZ{2=rMV$spjDV4%;%375&Wh2q3Spj);
zm8fvP(_bckNz{ru@|5kUC?%zwV=?4p1*DR2in~zc7nTC7)Hd8umWuXa6O3>cxuio2
za{7fiW)#c9NPQD4O3Zl?$+N;j?BPdLrLuOpFP&my;`_XTMHIbsil=R7on>*X)~M4C
z`jdR75WdFzWup}@#qv1MxPfUBUiRQyY^-!EEkVU%<i-b9h*<|CrM8g?3!Qu|+pdb>
zw4HWg$^B~4Y+*=JDdIzJR7ZN~@UyCiq!SG|k2J&P74`=_X=_hs)OOX8hj14Q!+|x{
z?+J=-*?4o4%*e>$R}VO_96C%-QePz0AYC4{iApSR=R=G`RiBe>(i4<B$Q3gFDwwNH
z*7t<%M#iKw4lI5%U>xUZIxQd<O}F$5Cd-*h8AF<gO!-dw%Xg}pmdqFua&QvZRC(JX
zQkq<^9L(Xs;%!tJMT+Gwa;L8K_W)yiXT{y-ZR?zd=5yoL`)oQdZ(q5}xlHOVYBt_y
z>m^2KCl#t9>m7iy1Z0JphkaePncTq1X&iU1%d(1+N?_uo&E~u1V<keiu3~vMY<K12
zqp)$O6+@PDSCW|q9>}mg&1+9%V^LYbiw-PzfT7m(4tu!#Hko<Gbx817a(a?I-ZiIe
z+PQM0swarGg<=|Z26}SFB<X3EatDYcEuBjJ$iOQk-YPmvvj1hLCq`XNNsRiey{q};
zI5i8Nw%`pcb`%Puu!P@To|<ecV;z$MnjXwSO7uv>lXjd(6XRq|%sAIt6-2>z+V-CM
zu*Sxq$V@5|I6Y->QZ?jDk$0sj#$jJmImD*cG-ainb_wcIyr%r7wJZs>N7?_yg^rjv
zC}~nH)N}QM*PgU!X(nPj#jf?<M9lKu;m$NC4#O8l9GK>4rrXtux5vuI4iw_#V{;83
zx&D~bpHr_0F?+mXx(f89tN{~NrYJr#XYXrnNNWu~IR_T_&>k(D#&JUNk{H55YwvIH
zl?MZBht?G2L@avkmn)G&LnA~omv2Ne+t(J%gxCu)Dd!1g8?}?3W+tBEmny#`Ymzeh
zc)oq0vcHbvB{}uA%PB4-SpzQ&L31a2wjxAi%=)@&1lZ7<Zjn0Qb0B3U$x0y?tX7Q-
zdtYDoD^|@ETN7qqK*A=4*s#;tH&lAs#k|l6!d486+&HS#{*C4PpOb|)poY?4Sj0Z|
zVeQQ4?VBoRo+AI3Xgvpz=r6iCmMG94nm$9iY~NhTabtT_MbK&M8&?nObPxMr^H~bU
z3$K6!ORwVEQTtGXs_NO3>s-Fi+QU@APql^X^cErUiDIe>4TsN`)8IMc8T(YETVZqH
zUcGC@t+(H7A1>b#8!f+3zyz>5I#iE1Cj+siIq%ZGrSyy7Y7GJrQbvC-IsMI43x%o&
z8^mSb+93BVQbbsEynzu#sd_Go1^>3Zk41(6vYWy`t+oRzA``+6qt=<6vQ8<$4*LSv
z{MNobm*P(4vq1r+xpnQf`!{aiX5UesA$yGWNO_Pmb#lOp;@p)a9(DwUBFS1KDO=0!
zoPB5F@7iOS2)P!yEbkp6GhuQJ^U|b!R~<jU*P&94<J0L$;tIh?yYDmhkph#ORHh`w
zTLViBUfl<?Qjh3Mu8i+)o^V0o5hxyq9azp+Wc@~UQt;Rmm$vUIznQ<fAF5zuyn505
zF!JFOoA$lst$cA%SJsx$aSF93_I{{@xZ6T^-`A%FsSG?@;TXIrEJ0_T{<55)kfP{*
ze|Zm5n50PPLrG{l@t&T<Jd2&@AOVQ2HE~h)19e?(AsLZ+S+rAR8k+9np={aS2ARhZ
zE)>;v%7H1$<GwBSWJPOc%Yq_tb;_W5(-WL5r(Aqz!S?kB2muH+#4PrMAiegdMU{R;
zpuxqmzj>!ruKazJK5}|eAiRyj6bI_aV;4wo{eTGJ^cRW<kvin52rRIJwI_7C^!Vxn
zA9H&2f*!GxJv-FAAkRk0A_H=GhKT3%mx<yObS9E$4x3N$YSo>$d$}Civc(oE(r3sT
z*^SM%0jz8F#)$Q|#rmwmpe~^^0*e92St>)qWfnW1$znfT?glpzZ>VDf`Y>jvklK+l
zg8A$HQg`z6)`~^+m&L{TQL!6cu%*HlQ%68tj{W4R5LtD~v+}EDqSm70-uS(a?4%$Q
zD#9^4dbosT>0F%Y4tYe%WpKb{Fk0JbM-5rrj+LI2gy5$PHnUaeX`Bh-Br3zKAy=YW
zz^kEc?l{}r@uYLMH$(*Dk!i=LEnuS)$K|j=JCXOy4tH?Hot|{s+G7aC=-@s+S-vAs
zoQURaSCF-9Pr54y<+4*rFNEn5$wEV#wbZ{b)t_$YN1=&qCZR0#qk-s`RKD_-JMF0|
zvhIwaM4Ia9OzgBX$vKIknslxU`QNs}&Ng$Wyi<|?K?kNA9(~)pUh<%wE8~H)Ac|>!
zlg0TMNj>x=57^V*K^RBdp04&S;&OGsc*NlY%)8L#l|*Le%Xq{t73+&6!vlFRe-&d&
zf1;DXo}OJuY`o>cn;yLBtuJ|@(~b$r2^10!+B5l!-i%zMPAld1^(0M9+r>ISgeW;`
zAw=)Plp7>#%;R>cd{$c1Wmp!6RFZU9c%1&Ays&sOuNd@#sa&wjr3Z!<+`Bn(&gs#|
zs-D~!Q4FNywCzd`i0W#yds>|tryZQP(yeUyQMi!5RG+dROeT8ItaFu!SwpO{{ZM@l
z_8w@(Y3Hy>-XzlMYTk@Z>oA)Rt-zZ6iIGE(j&`lU<Q7n%4pHb1EaM}0cQAQ#nt0Ea
z%lD<6!8A!Y8wd%m{gMUb;#Zl7EJTqJa@Whw&ki2GVx<lIU*!RvyW#`0dY0_W7wm_t
zWH!CSxo)OiOpK#K?AbcvUYigw?>9NQWT*W|^F&369%KhT;lR>p6(<xE0b2=nc5)|X
z(dh^<1Sm9<Xin7%`>`rj-#3earC+HZc%mCG)<nyGyrBW<g%uZ<sgO=jqF&M>m^q6a
z^AiK3QE(}tRt~Jtfii1zyHWs&76&wIKUsbQ<7J~uO|Qb>HCR;^*#h5dKUIZ#V(_JW
zw;f2?+XMHg6p{cbAP%eL$Q%3V@?;8VDwqXMcibyXf$7p&QReJt3Lj+*Fmp)<N>z>I
z*Ep@c%A4fdIQJ%|Rp*Ctnm=1HO~@CL8tJ5PI)LI&cTf@W=W?z&b@5i`Ixv>Bq`#-_
z=j-$CX>suVw9_v4L{<b`>1q3g{{@5LePi~?`*HfqOvE8)?AmAT7wa?<_(+yM&A3ID
z%V)wBcHVxedDVs3w~n#F1Jmq};Zk@6d(khKTV(84wUJX8zktcEzq~gpZ%W_%mC6b6
zd9jxbJAhy*F`2L#o#ZFJwq(EByi>JM7q3#`$tz5;CL(Bz%l2!9R;7oYby`v{upD;n
z*UJ-RiN#XKrgzBuFqw`bYQTP@A&$YQx8r<COtaHp*!SdaLL7*aEgvmWP4#@1WSIj0
zXPus)vm8#Urqw<N#n^Av`3Y*tQ`iPdAUOS%X4*ZBnZI4z+ao~FX$QK}|EM>UtAurZ
zpZ!kaT4__`?bZ~EERfJy#7UV*wi@)kSb;*Y>J7*M{ksK8R87D31pC^D(b0hAI^=GJ
zE6biAG^lk*jga4a&ZLFc4w-`70;ea4NB4m~Cg_6wet`vhWo_Z(b6~muuJ88j4=R4~
zTVc}ZIRKH?$gGaAoU=b{9?F3EY!puimWU#D29K5fQTe0{1XI|kSgax!YmbB*4O;$D
z&iF&V`r|T8c6=7IXhBlpd>Us7grQ%ww6QfZXMa+jFQyc=%&KJQ^pqhhlO#g>>`$9%
zQeKlL2un1+v?rY&aaI*pWjXv=`E8y#n6=g_43-#)rI<bK&nt!y%2Y<d0p##kkMC$<
zrumCHV`yv27rDr6X6$V&Ca1s1j3uaARywxD6ZV(o6ErY6lQ;{&ftC4B5nU&Brkt6-
z`Ky5;E9`m^0k*<)0CE<4J8pkn-kRR<4DN}<B&SEH?L4%rZGTgr7Q9zQSuQ!S^1T;h
zC%A|SO}p=s{cY)i&u9-|7gq2O;dxGvxJ`>vL*ta57-`$zHSgEHlN21$5MF!IA(Mr-
zL;A44uar47w9#qg4!4JBw4VJ#`O?I!R{TM)u&P?9D$YMPI8~Yen7#hE1Iro1nFRN?
ze`-dpvpeL3J?j-F*MbVusJ`+)*YCOgmX%H;XieIoQW`nu#Qvqg(4^2D((etdu!&RQ
zc)l2^Bbx4iZF)!=;M2l3C#K-^46rT<d?j#(57EI3|5gVu!4VCuX{W7kS>C%rmatg4
z_U{#6Yu4U%|J_bAJ@?b?QKa##RP-M?^=lDKooi{w>o?nfCNv5GaUIf!G}V*#U*(HM
z$W>iYu=NqAC$LciVP&HD?>fFnF_F8Bf<sbwoE|;3)4KM4BtJWn7}9!|uqGc@#1A+<
zfd`R`)GM^n;3N{c*w*c%j+EVC^|%r;<{UtwRb|;GMWzquqnk%*J!U9eW}b?P%{%>C
z1$Vt^`&#>$BdMV&%u;c>vXHA@rEE)4vP{A$PiHH0?v>E6b;CCM*t~4PHM!IPNn@1o
z`*BC=PNEc_X{TM-ltR#g?(O3XOq48dh%&N(#qdRawVaJ4pvZxTfIgu<2Y#U#^%IN(
zSS}N^R7m?1^VM71!)>Jf5vQH6R_+k5CPCXisUS_>1_}Ez1c?Pb<MgD^P@`+@lPktC
zc~22!=m6q@!HxI^6}kMB#3f6s?OZ1&)}R&Gr<M_=B`bA|bnj;>3H{}1D!rnRhl}=U
z1tua;At#E9Ik4DI|F;vHnj-!7>2<KGsn5Vv7q`MVW*R(4ZF76xk)mcyb;suvKsfDy
zGTe>|MHG9ULbQU9F_)il<k|G<$*!pAj;NF~;=r|wPW9TeBlejEqY+XCQt}uE+$2s<
znfiT_-ryZ)?XwbHx2#xkvvXbgRY6bg)kXX4N<O{u*%8vvXa|v3<4_nR*<Y9JbDCeD
zXa9`gY6qqXSi>R2L}+|&ol!E{W~gG{b0A4%DLP3q3vMRM#}XRH=jFf<bcti3k<`_h
z?84MPzfSIWYe&y%C8vvLV3cBC4wb>apiGyDo9Cvl6S+WH$=cRZec-M#E>+@HaAErF
zPuLgM*RC7KyyrCIll7qL;2xkN`=S9VQ`(IfYdw5{idkO!(`OlzAc7AE5r>G`7w7Ay
zW>Mf5drq@p=qMI3RKnPT8=h=mQpX+EMETg+5BPh~=}FOQ`|PlNX}wMS1{5o&M$~-T
z=}D>&_e%QHzN~?%=83*;$tz4FiexwTq%Uu95#=Rdp$e%zi$I-Lb^4`3<73+}$;O}*
z!>?#QL?W{`%Tjm3=_!Z}wLgpwQhrPRojt$O7j!G5GiNF|u=ouQ^R#O|SK+i5REopc
zGZkTJKvIUkcXno6$$>8{A2b-IECzJJ=_wzi<)D)URMRe!=@AhR*$wrBAS{KhbEJa<
z8(1D9K=E)bj~C^<@*I-ehMi_G5rtQ({;jG|As>dJU4A~+7on_QSss^B7ik&Fk<$~?
zmQDl((Ot3^S5^}2VGl+99Z>o!B0?Zz){0Id_N!MCN%@lUFnrOm0OzmncVL>k5pfv;
znPpMEvCbX4u2N7>Dr$^*xa9OCqaAwV&_=r{_o9*Blye#Q$af}f90CKU?WOf6Fo#`E
zD~a{q`?uQ5n$J-UPnn;V99XUc1<&!;a4lhil+{et2wq-Z1bIftfif~<6+1n8G#Tuz
zhwT;3V-yO>x@9?OBa$75k_b+}ZXUXK^X9eo%5pOiFWeUPH7)4+e5b!m_9;ZrObSQ`
z$<3ul^M-#CB~S5N5|3+7Qg*uQ!nt58a@5OvC)=~wE5=oW0a!sF$JFwMV_~+#VQ04t
z5GRIQNK7$MiUXE1c*Sdfru&2!fx8s4m@cYYEB(ML<bRN34_Le{ly<)NKzLPgz}Ri&
zZe+z+;zZ3mu((^+Z-oP&sJzr}FLxheH?Ofmz+(QSgKHyIf=;@lb9vlR`e9A1fm9Jv
z)rft}YkwMCa<QsT9Xec`a(QR&V_odV&UNl%;o4C_?K8GAVKP5b|4xT^^c+}>AlYw*
z69u8#s(jj-`-iqW*JY&<4FyeMbGBDCqme=p^9SBl{ZyDd>hz0@?q0ij?N<A$Y_Ntn
ztd>S@V%z7~)9lr`3V19vH@&ntS*tVMwb#^P#1<~C=bQu6Lc_217&6#v^D{Z?Ytp$c
ztT3&PY=KNlll^P2>px9t2r!&|4lI}LwGUo@%)Yt~l)fqj0!xxWX)b0E5Uy~j)#ZJ%
z0ZS*7ODbh5q4T2Cp95VBCAaOa!V@NUI+tl!-PIYHr0g3;qoL+umldfP>v{oZB>r-g
zy^Q{t18q-XpT%Xl=1Axa+oqkX_{hp+j<7;Sl^1Mn{RmYQ>^aRc8dM-2jaLy;BeT`k
zB_><bEIO{sU7UVda2Wfz-CaJT6slmyJmkP6yRTTe>SkMCh5-hGAp}zofkZyrC!K!j
zemXZ`cc;C_HkA7_ek+$J%OQBuVVGX+7j{!gOT$IU=(c+jsW-j+Cg(B(=8&x>TBM?l
zak`_u{zxcZ13`4e4FF91w<xcaWwAEqz?S!J?N9-jFc@4Tzyb}a<Z1vV#=|xhG(ci-
zrL1!YET?;Td=}5o=Kd|Qqu_RQV9Dd(KwX7<%h09p7Bxgf#CjH&yXf?n9i=<D+qN`_
zClrf`BY(?*<xY;fuRV@deqYYSx~*&1db!@w8tvKE=gj9o8j;fzWEHjuC;`8*ZEL<f
z{m*#M9aw((ZFj7+?NuVyd64Z9r=8pe9xSiyB_48reeiW1%oB-L4lMAg^dV&<al9wX
z5dL9rD7erLQ46Q%(hZr_=}%ltb*FmvK*}^#mF5e8l)X{nT6)EV%Tnji-dNwv7Yf%(
z;)pjpJ!y=fLg#dMlf5ZXNEwWGIoD;)MP*N%2FL8pb*wnR*pHaIQ(j^6rYcbbH<G=j
z&dn$X3&|q`Kp4FC*wr%=_F#SMj@I-Xtc$kZ6vZ$`akO->h`lve_e&pjuJqH+<XC5i
zy{(ejln%Icnu(c-F28wu?OZoKi*`se(bAW67`6^(intAXM}wBE5hjEQZAYCRJ+{M_
zW94m?=TxJ#kSwktiw<r(r@!!vEApL+d_jW(g?1ElNdS-_qbNgQdUltJY11tmuHS$O
zM<KNOt7Ub{Yfmz@-1b7Aw2|_g?2%?EE_MW;PVI?Nzyf7Nk4dt2MburCcW_#P8F)=B
zxPGznB*>i&=Fox}i*jZWf8a~8M4kS+9&WSIW{^iP6v%^i#49XX{PZ%i>{z)^tTeJe
zvEEjgOu|B%EK{8_yj0U=(o_0_4$Gw{sdZbMGe<}5p@Pa`7Glp!YkBW0Z+iZ#ZC4pW
zx`V=(%eRTOCgsx~)GZl9K}<MwZM?t~si)*?9Uuct)5UK|)wYRBVdGsX!!&bA^ZD|P
zX6y`xBAYBd!k2vSM9qRbY=TZtJ`i{Re(P3D-hS&XPNS^jzKlB<+&xtustHkv0ajlb
z4LxZq**?~qAuVCr-kB>f&eY!AqGBF^WYAkE{xWCN6_@uA7>9G8c5Vuj#4f7(c+Ywj
zq=@XS1Oq4w#Ef9mAr=8CRX(4Yg1CeD76^<)$9&3x72Ko;K)$g>YAhbJ+2(HTzw$mD
z@dg&rN>DRCw%z5vj1op!hp>R%py|l<Ai+!Ps*kP^znH9Z1jg>E?*UP4UGndFg(<gJ
zq9ZvZ_HaSFgmp16opU6Qh4RFykmU4>+LbcXvbl1zmOimP%L)!k+hCp_M}N>2TS}Pv
z6vp~U?!qH&mOOTxP^KJU-k2IWV)!w8S2JRY_p+5-0FM=xJ|yUg__B8w8j=`%!fC~m
zivI=9ble_oupg$LNczzDjyOFjD5x%3FOJ#8v;Ce*TJqNg2Py%pPEQgBjtT|B;lTWw
zW^KuSBC;0S>sQ92KiR|T&6~|Dp$&B@d_C<=Th%+*wcnfl5A}9-Mn<S<Lp$iEpK^q6
zR)$-9tZpj9Z1_8=jSoQYndv@ERmR5?3u%TfTQe}xJaO|>SQ@qWHFFnMt2hxe7#c4i
zq|WIt%CeMvoB^SuC?QQ2cGYtrslwi0&>g4W5E{-&I1VgFylMM(``VnH<-PZ?DLJjO
z3n>&^^H(gWP?r+(KhPkvU(-%^X=NNoVWdBo_8672sRG!uuPYzGY|2!{8aVC13f7*K
zC$j88D)#dA<@vIXs_gqo*lDhtFgiH>`AhZ!Rq|1Ua2WUO8;)cxT)apTX_v~@K-PhT
z7e>+85u9HnZE;W{i5xoa03GKWQ%Zf;qhIA*merBx1fwh2H&xO@91lHGFJoaIa(d!!
z?HCH)WLvavE_dsMOWBZR3=gC_JxQpPyKl{?#=3p5PRkpqZAn9nH!4l^>_e59I)pn+
z5*-U5_ry>t%cu~QN_eq7QGc+e-P=X*jspn9X`e_9TG8NC6SNPP?>@xORgjCgZU>g^
zccQC6xNoVeVhc4z$qArc%s~ngj+tSu;Bs86Krn;9ddR-DJa~awjQAsTMqnD2r~|hD
zZHeofSFLm|rI@S{$$Wc*Nhu~eTx!vQCCv_PT4Uc)kw=ADL@OA+11f}pdctbp6y+jJ
zd}p;!kMxss!nj6<9;wrmJN8{=NPYLCDo$#P)cz$=i(p;nE`9#$Nle+}_K|FD+fJ(O
zCZ`o&AhAahv(mX`s$a72uESj6ToyjcW9g>Yw|!4tbMSm+rOhLR7w*pfa@6V1se`Y`
z*U2MuBem}>Puou%??o0j=iPXPMZs$JFy$V9Uw*&~Uhx7imn(0*#lAmL&e14s=UM|B
z?qI&3wjZd|Kh>Qb9`mNTMccLQE#H&=T8SUWcn>LfPESx}vMKFu$#6w*N=|(;zjt{r
z8!|CFT5k@l5atpJNf_wj?xVJ^?4D^TlGZ7-{?CJy@-#v2Ey5PnFk2(GKide__;O&r
zrM<kD<60@XYzLY#Wc5_V(}pY3=}ARxl$u-8_79d@3$uZ@${B*i-RTj{4kM)6p}fY;
zz00}I)k8Ir-jvNZBUFU0sw!)on1h_2LJeuBRgK$)`Z~-(Iu0T-*)b2CcY5M()wPn#
zg41kmv0ywb*pNLZTEsRlLnCPnPQPe;h;*!7cDOu%a3eh$hJDF_#Zs08pj<`yTt${j
z^-=~yir7bTMBAxRcUSjerx`zz*+LfM!D`$Ymxt15OG$N)v`3xGSZcl2u8tk8_$bm-
zzKku|>4_0$6k3!yvttD&3R1iZXK#Cj>HT80EKNJ!JhdOJ8={W)dort2$D>5Fnw`i=
zBV2ymxz?mbVRyvIlXVU`%7Q#<n(<}QTylrsZ>I`OCQ8tW;qH`Im{QqQwB;mX_~{g=
zNbf`gUz-)b9AJ{!<Sdb1go`{?c^Wj+8JTU3_Z(1NPDp2LWqi1XtWyGHZ$FctziIW>
z*E`p8u5#k4oh_J1jVs(M`Wd;b^d!BpiWZ<2d?R+Q5N>;<?OYcYHc(SoZYO&>(Xpqk
z12B3{Gf6R=yz-u1vh#K8uQg;Hh}Op~t8Tv8E|gyo&=Xq*o*xI6OFxRIhqw*Bog}35
znYx@g15ezYcHEhar79={x!c7$Pn>8q({Y-)ZBDn_JMB^(Z*jAs-FP3S0*TtYPTA#h
zUwXmAP~^axy55=fJv?JqD&gZ=W)WksEJ6Js^DKIzX6LwbR|-Mo+fM2T9Q(n1ff(P;
zB4{|UnB75}_n`jtLy5iB+t+S$uH#<00OTpm8hMl*<-HhC?P}?l=t&v@xk)NBt|~`t
zBrF*E^NZHBck)TF_-oBCBDj;CPWG~<9<9?adU~icW6Sv=szK;n=Qcdj+HKb>vLWoA
z)2eeMp^-%?I;Gx{{c!!!P-~~t5K&CGaQ;49JMOxB$Z6z?SdWAwcJ(6-$z(_ab?v>v
zBxjsq@Cd#3qvgI*4+Iu=(t)Lf0vD~7J`}Db2dZ-O3Cv>w{aE>>axpb|xIY}2zOMzI
z?8hsC26L1@NXO^M`9M!hNQww?6l93ycoY%NFWOHe`tBqsl1n{sTNgU`$-I65F5_G$
zD(?c7Kh=yuB8NEIIj^t?9b>HUcKhi%ZPKN<FJkEl!Bzmcv@7~^g(Fid!id6I_A>>h
zJvz<NCmdLAQ92iUhy84MY_6~LWGI$K^rMs5=`RA`S?sC`kXKTx*l7*<kY(`aDhpdD
zt8VP>fJAsxD)l&pEIDI8UtT4vmfZG)#FTMF%(mBlN&QNMRsn#lKc~^0v|mURUOU1p
zbDgVwh=5u9#X4KDaShZxa(YC}q{`VZH3(8t2s+Z$1+TE&_d-LoUoH=;u)c~zPhq0c
z0Ak~{Un#JN%;a+~LLwC=PaE5@$i{xPuEQyvgGv0611UUE=J)Wr@m0i0*{?NIEfk~)
zAd`jCCPaUHhYkytI6x^c`}I1l6!F^CbDBjPo|_yY+8OiUZ#2IWCkdCL1C#KynZ&yY
zbpOpn@pxzFjB}me7yhC&NQyl7+i#U8m6R}$zUaV`tll8?+kU&;B9(~vJ_PyG4y>?=
zu!e-%D*a{2ey1TqSu3$ih1eZbAP1b&9|K+_xJjZ4WwGsd8zeQZ_pGPGI>(#jwI9NU
zDaG2=pk?M}LeARn<<RjO&v4#N&uN$8D&jgBDR*8yWWQhDPE-*t-AMBE3PVZ@pE`f-
z51LV2kYxr@D(V)AmSYsPKWwf{D`k)vYh$0aKdkS;cQ0#^^6i0%&aWNEZ1JT1QBKbu
zjMdKd!1N4mAAel8&{95ycDXYJh_G|XuV#NzCyn@QPQId9CMZqJI?XZAXMdU;tFudl
zOCwl9N*QUUk1{=wVx)10qw~+oXymIWX;xK;9GHk?1Um=kd)uEU-(S-mo|W;v=KzY>
zqwG+fWFmu&vijoM{vt<4$_;jO3d7Kj67t%7B@H&EHNfltD_r3((`bMHjB}OSBY9s_
zQ`)fqs-df>Bpcg3NI)Hx1^bfIAGFAMvI@Lq^FL*OT~`~PD!hnNZUEh&i4%cKUXTT<
zUdyGg>l)VZH~BfQ`by_2rdxdAr2TC}5Fw945c3W!_m10F+-iSUCX6V%u$H7-1%~YO
z1STVjo|lC~C9EyN(sBFyI)~z{L7!+rxzMrmN#w*>=G+xyMNnb<sSx}_gP*Jzr3?}g
zNwovfpKF!XqGuvm**{i_+|-)vP?m{yN>~&N8OF=4VrKSF<&$_A+J_z~=dMf(q41PX
z;;tlS?oxEkDHf4@_2=!Mb8&3jvUbS1j=K~gDfe>ZPcsJx?O$?3Sc&1GT&^ce?_U$z
z9vO8mlaY-Qm(sI;tMAj~^9@R`OfF21W`a0TY)AjzY?HFEi%n87>VJxg&Ru0CMq$TL
zwlvTS_8)aWdy_IXX~lg?7?%s|RBR9gecOLF)TdC52tIRQMIwpw;^A&8SP~kKh|K=0
z(gJg#?MUYpY4hUuSZ<Gd?XSKY1zN~5t?)91to(PKy*O*4=d?A$%X@WTM0cXoLshn=
z{Dy%d*1SmpzDj*K?({2gcGG?KQA@ES-#fJ3xh%tk<?F)`Tn<v2?T~%+QWBA(HB`Ys
z2R&7o=tJH^X$v1yo-lZjfQWs;=?OBFO-6AXr+MnfE`>ZcGSZ!Kt_zu7mN!U2{`PT8
zv1xLgkAlFW-Z-GZt&-xk+I3iS-{O1BK0bef4Rv;eqh|3FI-vMTO6*Z!8D`&)g!_c@
zW=f&NqB({>1Wb}5C$+Zh6PJoKOaXO-n{fvapDN9@l!CNRDlkzGmLZnOX$O{+JBHpm
zW1n1bBW+(eq&@~XCQ8QXNt4MG0Vc4#PiY3(-%3Le<yWv^@AvDZ^Zbk?BUO$%W1m`n
z4-5h;0cr!bv)Z3JN)R0wQSgR7{J4Eu{qSwuH*DPKGy{S2*ZZlfNt=-qVxL}b-+~RW
z?KE?(7c;Lctq~<{&s&NWir!6_01aV>Ffq<B=-Fp16#;U^s+(_dTJcJ4qf!l{I$Lbu
zn(|}zna!NVf%cfFPdc#t4I_;tMT7&<CX%r-C+xGzZ$vs-!{ML<D-<smP7SxjY+;{W
zDOswyo)a4menzi75*SqJ|3bKRIn&TSXQ`O6an^8v(jzH&1C8$z&Fj6Y_a%#tT%(YQ
zeQrY+=fW%=^&<H015Qs0jT{mOnr0|1{CSla?`ci6@GCih+<(?|XYBKvkB;pPGlc>R
zh#hx&f@af#r=;n5`+@?~<|cbQxt(5N@GJHa#TygIT9FjmO`>!j^_EJn^=sObBuz}s
z*cX+tiZAD$syZ1;b&Dh=ul+HLz=$N~#YzK}o9eWEarq=7$%0Gl#{v#4<|u>6!<WLK
zR15A)nzu<jNdc}FvY_ATm$Ct#>`Rlp@Gek()M*WnC?BY>M7c%n%PK9yN;w~jW_h<_
zw9+OQ>&?Erp{BaQ9P|nk=M2Vx!S^fb{H*KFI*nZDY_jB@+w+snEbo2GJDlr6ipr9P
zG(xLHg0dGhNKq9FWjZc8u(&&n4h&rkJ7j8mVS^NH*x55^8Ls^)k)g>2VJoyF^fbGn
z4h%v~@BQNrEDZu%+Yw)AFRC|@8Ycf1;Rg<^41oQeSyjE-)XBcG^vJ6pJ~1hh5_p{+
z@!2+(H+yl#6anK-LmeD?GGZ@jejeop<y&PcB#qM}9zwZ{Ve!oy>*SG&%DD+N%j8Kt
z6y;az_f1I+2)#Nli&oypqEy9?WGPakU%FJK(Xn=(pbR9|k$h;XUP2uYds&?yhQzc(
z)bkFkd3T8w<H_>dy}S@*yX9O4XGq=~!h<GJ%0`$|UeOGk#mSIWNIGR5dg7OZr)Nn-
zkZ1Y8)l>G$$~hvUm(o7F9FTVW=RayU=jiVa^OMsO{LQOvMNU9#a)JP08iip<_mIDA
z(QYYkMOLnU80LOn(Vj~3cKTz`5w`&e6TOk&YqwSuvEQ+~Uw1?hVC8NqZE|-F6FFhG
zmG>Y<BN~aUReh9<1l~p*pVJ>w`8?v%VY_`PC3`U@6?{1psz<-dsf;#YPliY5-gu{s
z`RT@&13=SqN5v?dMM1s#jjVf;E~UQGANY{!h#U;A7wpcuEO6W?rpp1O_A*D3)))>B
zTbVyw-iu0A&maVrHhcJ~d<2Rc5IrkkJ<riNg7;xn`5@H^P-PV{)O3jjBXs)3_*HUf
z#$Ht^e<wEbT~0GTR!WUTxWgp)e^r5r+ewgDMMYqRN-6%xeF`m$FmZczgQyG2BLW>&
z%%(@rCQArmMXD9<x7Sp{$Yh*Ao(NfXci6hqU*3bP7R+1Blqg7hZNWu)R}gZT7Fm6x
z;1O0SJaFtY+!f?rWAH*74bedS@3g%x2|`D5bh=JUY=6})E9|Q)Pua41o6`ugAs&P}
z#jBeUN>|EFDi;?$x^H>!I)oFutGQ45RT6cmk4{fwIm+Wk;;0Q3m@FG2FwUiNVDa@L
zP?1*F)oH@PZB2PYq$o2B3LgmF&*@1Ca=JsUep}moN?M&o;~^J1Jz|$sC7rf)OV29V
z=h_G1!XJWfd=Me^gM_dE*TXioW@g5SA-y}Xbo;8?yj<S7>ULXS5w?1Er_;!3V-h6;
zWsI$=-*2cxhrpDx$~kaeVf{SV<91Ko@F)PT3?<s-i5DX(w6SvA>+{ZQqz_+vf-ElD
z14fMjc|8rr#kjk%u0S%{JH-DvfI?}fr18Nn#PiXeZBveHxI-$v(}<~y31><atG>A+
z`vLZnp3{t3Q>TQKlVRz~xsw6$-by7yV;i<Ot$1Ks#KJ*jTk5w&Y)tqWp@<_elYUR=
zeu{454Xks!uRitO>CULr5+&cf@>bj0e2#>=TrLCY<4%7eMivYsqHLXQTgpHTnQOJ=
z`|da`G)3T&I0SkmBkS)`=NDy(T*N59y&>EzVx)r#()2Df!a=9M=q(+wKf6Dv@Y)lw
z7Ovxa^66SIr|k^|A;1ti<ZThtQG^~nDa8(L-fj;xul*mn#(bnK{`8kQR<TjS{IoGZ
zf@kfG6?=CphlN&|WF=>69#7kw>Wl8#{CcNRsIjR{$M{ge?adXN;ck$ua>;=;%UIbE
zhm*j<x1vz+x763c_*C3oN2nAS5vuLUJcXub4_47;vg=%@;U%5n<1)M1TPs?J+EnBl
zcG_Cp_b4m#Z3V?o!WFr0rd%R}=;!Utkm3|CJfX4!NtviU;!Kn;LEV%5g7e(o(eU7=
zBq2}BwAQCXj901-cjbDJ0tTA2Dh8fqt<s;Qnf8joBtrW2waPB#EC9st|3ln+fazTw
z_rH_gyVHwfLTuzhF`~I7Rs&=tkR=&{!FIO1k{0dmyWV#fSP01&(&M-#j-9SvKwUx-
zl0eC771Dd}z4zX8Ki`=-zg_arll$D~x%as^+Pr7xSI+O8Idi5Tr1IgqC%Lb@*D^b(
zHon{eI~{y}A%m$DiDyXX9qC4A2+!iQ(@rI6Y*Ob~syxhS8`ZV={2nY9F2E$#Y5c3y
zvFo+vKBY`arx~9o+m-4sNp>lnn^p3ZZOMVZTLxEt=LRvQaL27R<ZVrAm4$xdVCt2j
z3EH;$9IXjLft9J~i!jwPRPy^N8>{QmjHng&j5nB8o~#FG)%HL)<eE4!9oaNBaS|k+
zaq#NrAOkVB=NGJ)oE&m~$G~Q}66sV&Y`mT(=}wjj8LV>khW_l}^B%H+m>`K3Hqpq9
z5QxZr45dYjw8Kd5;yFrf4t8vlFBmV9-}s&o795OrC_-O`re`OtUl(c`bl8UM^Et3!
zk}i-TWRWJ5#HQM$Y@p<=(~2wNdWID)VehEt(T#|eNV~hNvkeo&s=<Zo<E=K`K2Y*f
z4N6K0V8pV<$40hK+n^geSvxWYUzP!d@Zte>@Uff8A_Mo1Qg><It>4VuVVfyQHZ3xu
zCuKEAmCt=j;-40X%Di)JeBk@I=6$Gz?}fR|=5}$@tCu@JlSDU<5&HjNBgAzC$Uwlf
zgYjuv)ZOm9PJ5{F$K}WcG*a!mgGxqHupVxoDUQN8$%>ujwxwa7a`4e9$wJzF;W0MX
zdNQ`5KkBsN3I(|+k4|3WN7_3%LCk4r%X`HuUTKdmd7FqUE9P`1Q5}3f3lftT>FE4U
z`>KRt3#H075y|Ln4osghGskj;Hs<WHMhFPJiW*4!gi^G@<cgi>kIs(w?W?<iO3_WM
z3DXBwiG~U70ck|ue(kzv?Q2@MV!u_`*Kr3X7EZSwHgBK1eQ4Ba#7s!l;4Jt>-fZt|
zgRdh#RQ?x$o`HRAxjkO*$S@;0tuA_lRsC}&<cvMh`oe0J*8|zX7h%Mym;3!|>$=Q1
zjHm{0Fd0S=#Nk$l?Ca_}tydMFL@sn-!8w;uG8ipZEwo+s^>xod_^1?_qeyqaiU6hU
zX4oQcMJ)e@x*FLeHf<$J0MnO`^ruGc8&h3<;~Sk{IsK(>D6H+9>bWKBMA?C`7<t3M
z(IQ?%j6*BzN&DuyCzUJBq@3MsZw@BjP%C*sRRzB#$(b)aGr!+yi3vRR*ke?Adskhq
zK_tu`HoDszOk$2c^!WU<_O1WJew)6D)bf(Qa8(RZ6y#v|r5>kn^6q@k!VVoRO^mYK
zTfnqz`dh~BJxMrWHCIZE1BfSy`Lw9U2r#ktR*zNmS@{|=ZvxXN-9Cq#+TPbTSM*99
z^v-Gml>G|xuDE1on4PirH^LqnJm@rH?9xIZb#uRcTTx27Vzra$+&3NYE9cz_@9@|Z
z$U&kYkQ^UM3lC$JeS4#{2*u;ou|uV@NZzi4kFJqv3AJlJWr32V@B<wy>1+X^LF@!i
zsc{D{b~!vfIz6(*zN6@EcDU#K*1!@gWFKtbw`tSuP9we#p@fN1ZJ?9(p~mV~_9+=g
zE3T+)RMoDq?<|-cv&vS5nK`iNbqKfJj;pY}@2W{9HI<+FX|OEOR+hU^Z0~p1BW0MV
zBdi@*E=dGerH3OO^D=R34eCDoo_bpDcWJ`OMGvMry_@}RKX%1^_PyP^m98w_-ijZ9
zk?m-0-@dPVVn$YG)m%1x2NO@sl3U>uw(l=k7{KY@ZlYxyOp8sm0~F%DBLxBlo>9-j
zX~v@@Gh|%vS=-sz<eN4T76r8fXe^vGlb-FWvB-rVOFsAZ80td~KIO-ZsM9vLyFKD|
zPK-fXqOH%o^={kKghpKcqXVberjD_duq1ti4cqp1Z|$s(keW91jD!TA;~ZOhh=R7S
z8)rpSu>>k~Y_Ef<`CL3ZhPN>rQFF!i*VCk>N;%YV*1*zsuyq_k*$%Xa!^fp8P#Pws
z3I|jCJI`TMd0g#kau!RaI9Pbu>02NMlzaK{;N?es5r353R4yU;q@p60WNCY{Iww{k
zyD%pPIk0%AJc$HM!E+^RE}BH1s(WP3Q#y^bO<*}u<GtDK{gF8o&O<p->j^xkqR^bv
zE&@4&d=eX>9d3(~nod2Z6<0Lh`l0zlcBH*5{J(Nd9Q6iEqieV;>}b0#Ji5uHfyIY}
zw5X2@r=9|uMCoVjSoc01l%$}YgWh0lsklF)wv`=ky?phx*I(<jQt%qZyf^udSv!$v
zvS|7U4YYDU9Bi|b&EzPXH|e{h4kU}caP%n-IqZYmLajX2u3S601^*NY%RqvB<*!hZ
zwvJsFr`}~xH{v17i&z|*Y1bSY%>s{hx^9)jQ8q1=P)`#pn4~)@>GU7AGi@zR%*naw
zv}3^1=B*{8B@xPK2<d0L0gc5JjtvcwGS(=P<KUAKDujnA&MZH?Gxkh-vC7&%>@+oo
zBu$Vf!D%aYuF($x@6)rSQI0yGc$l=!$e)C$E9^CS{VLSQaX+7r!nIxnEiTl3;whvh
zpqDBMQE}v8g1*`jY|p_zpUvrd{S}Ws>ipJr^4MdqwdXq0&|ZKnMNlO$?L&0p(83|R
zm|rwTw3v)%G;`5p1Wr~IVd+FgOkxp!pzx9+<w8FjaR(W#LsW=2ycV4TI#AR2+ijxu
zgKesa2HMW(&@S@OVWgU4`CVjKny;>hWephBx*b@VaFDPs)dkvvm2`e^{<H(rx8a)a
zjgHwu_x0K;h-0LK4W_bHPe@Cef6^|u>lkEx6xE&_Sagin!=*~alQ?2O)IMQic0jQw
zr>#yYuzS>=uiH}AmPt|ZV2Y~&lk4rFF%FmI(jRU^S%I1=d#g*8gQ6-K;I+wIl4Sfi
zF1jD-zV{e9U*@s`i|JCi9J%dOUNo1BqBeKgk2Y`d5GM3FB?+RNI8YJL@K}GMPkgeL
zpZ!?wKQDQM^GhqA1ml7IcsB}KDNRO|I@>U*o5b71t6@h!(MSj`T$wbHFjD2@9ej{)
z@^`VbBleR86IsbINrIyT%SHZ<*)cl!Q}tMMbc7*k>L^1nZON460DHq{Cnd$Inpse$
zsMvnGULl1-i4Jn-1B)R}<@UJ!OzT?k8gbXB9hjz<%>YCF&n|u&d`#&9dOh|WUWL@>
z;ES9U!xDN(EP`Y9b8Y(Z$&QVXO9(*k-ZVQ3IY7=o-!34kgtO8n96&xxz9XKZK(F+?
z{X(O_O)MhHKGDt%C$7ou`WL&N>U@jXT<EWIj1`lA!aMwuX#6F|A=odq?og&+D;0(v
zKz&*UgdD-#8}fnua$dS(dTWLB0nm`GrmD@H{YrLjB{uOfRh#ByJO`!=dIQdsv0ts;
zr?3&4vIC2kHLG%-o&i<ZujQ>+JEw-6MqHNiTn4E~%Q<Dg-Y&cGx@(=r3-$vgDy8%x
zf$<spjp{+NXslB6=NwpalIfu>GxnQpgo#vA4G_r}4HmV?nCt!XyY08CLt=a~w|Cih
z4yNWvM>8s*{h0lBH>4D3f`U{~F)Jv6X`Az~O7Pn6)a_D{*#}UUPdKpn`rILSr7R^F
zy!YDgHUd!=%EX-0&bh+^fLvXXgYq!^USXs_kPaA|q=x34gGm{8Xp(Xr)AswV9pyWw
zownw1VaK}JN&AEDF&WS<+hf=b9E`Zh!j3!E+aKnN#+()UB+Z<?%S$UxD>nF#3d5Jh
z)-88yFd7F_%wtbDY;{ujA9sc?4zuIy<it&5Rx`Vh#N}L--?L_v6u43QlX^#L&3U{e
z)#Fi?7BPx*6zdlo(`fx^caN$L9AwME84SH$bTHa;Uv=gE_GkHX#r7GmpLdg7@aK)2
ztx$%r(~R$+hiSwhB<e4UOjF&{TyQ?U%%=HvZ+Pu3E#5FkDCN7qtc_jYa?WWL!O8L_
zCiSHKRWn*+t&LOwDKh}8E2u~@%;)T{+b|<~Nq^;VI&Uz2w!G6M(b?ZLp$#wKzDw`M
zx&<Vpi=AxxC+%-*z;Y047llywJDBuJ42@E8W`=-5sQmA01Tvf~7AMPMF(d(eisvh*
zXD0^_VHUN&ubyfk;Ke93&p5E0>XisnBwPJMTgw~AMte@H1||=P(!~B_&e^i7mpMQ4
z9m>p7eX*_fPaQAq;&++$sAeb8)WOFSbJd6K&i=Vz%E8`ql{WEZ4y^c47a1rP3W@U<
zIpklu_lWR<52e5hIb%O`@b#YZsj;p7Yj=;-b?v~+;iACi;I&LgIL&on|JLXixkTB!
zsfHb&Kj9q)I+2E--Ymit?cdu(QN6yW5CqW-z~~7{Ch|ngKRJKE{v*fmZOg9q`Z>;y
z`JeU5T#)gd8apVUVNz6*(ySN|MbZ6N-9zXj^0#u?FGeo=9tWeZ85`+gakj(BKiZ?k
z|3ZLpFe1VMQaF;shaJw&vP)=KL(u~!wzT|ulGQ%E!>c4%Mm_*bIXL2@<CqjiMbZ5k
zQ}7Xo)7V}y6GXk%VNHCT689w59Coe>4ktNKKC+8aZbOLt$a=n}UsvlyGG!zw$y_Cd
zz$b4E6(kpqp0d+Es$qD8L=2+mia&H1WFtjooe7Cr*$f|jxPk_UX17XDOM4#BTqc{z
z0-osG#~dyVjCLO-tmi;-8(P^L-vR~M#~uzgw}}L3e%786VQUdlpopa-_HkVgMI1}C
zRtKYb!Ib&-NAI_fuW`?M7P$2BAOfqW#vxci$Pu?}pU^PueJa738ULpc>A^@))Lbi0
z<rBN<jDmp#cN}zJIh=GbeA3ANIT4?9I4ElC5Eb~GMve&CA)H=opIoyX{FRwKhLD})
z8or2xI9JJZGyM+xlzL6>#i9Y7Zp3L54n|*2tQm~XKDD}-4lwcNy$&q*(_3dBuutnA
z6osB354ok35(h?I><{xC7^i)Dn>5@ISqBA3tWCMArGly0^JlcHNTUet!+`{!K0uVw
zUVBlt#wMp`Ny(s<lYliPmK4h-N9!|l+1#_OKji$*B?1rB7M0$M#Vq2_N?K7uBR{iC
z%vu~NgZls31<NF(3_1o@E-clClgKMkDZRk>eyS6nQ+I|6r1c0h0Tx}(jb^_hdcd$5
z*yq;O>fpub=0yjV^Q^2IWP#7CYqUZY3Jl9}V5xp=>dnkf+2^-0!E_^!n?f36OdL$|
zLS8Dv7!h*z1vNG@n93{~X>G7u%=2*QQx0CcUsR=zB9>dx0KTxTw$&6_<N#j>Ai}z%
zZ(p=X4~jpN)LL%`lb<5$f23zGZiB<pk%5R=@eUR?Yvk1V1NOx^lN<>%aDL~f!Z0~S
z_9flRx-dar#>#du@u||d6iG$~rCm{1?~C$_>0^)t^SC#Ru5RHZsv~#EzNDU`00+Eo
z$fM`I!Qh_~Lx>E0&aP}7TiNf=Y;&6NdQ!S&dXolkU)nti*Bqm(KxANPO1gCn=RAQn
z_R<DH+6Ym@f#_M`;E?=5@~*wCUDcml73td!BtMw$ghzb-DZ8qBV;q6Wj&;C+X*+ql
z4&qtXt*mEAf1=gt9w!(wiE|Os8DzHGUf$+!3jT_Dk#@3NA_hpwo3mHsGP-lkz0S|*
zjba-msp(*hSGGQ{nBvHeA*cpOF>MZ*;IvtLRRhe@gxnoKjt{dvPBH9(UEO_W97U--
z3kQ~qaY-;n^F(x-y<JmxD$FD5?V-5?tIboXVd?FP#Rxqrll|Hk;G;$+qY<SK<gf9D
zPhv`Omu(RXRxk^@t|QP)e#)|GiuDCkJSp_=NO<AcsoY{w6<%F;N(hj&7)B?pXa`e$
zmW&sAcCH?NKZQ^3`okp_t|Iyo4IXDHpj<Z44oqZu&ih>t6Bj({w6u0NPuLB~*tb(N
zd5>3PVaL$+8N0DPU<BMwrzw_<WUdIEo7$9=`hS*zZ7?lsb)V!JySeQfx3AdfG=f4F
z{fy`BmV%`Rbnn7$?A#5OMLi;b(q5AfVy9=R(TE4ZgHcLjKeN|%L)1=Pqz)8TO|c@G
zO~Gpio0^*@p!#*$co4t9BTgfpCaX)N#Le%v*Ehmn#}R3iKcKC<C;C*Mv@c7dzxleG
zoS#}+mp(9m*1o*m17*w7PCK~1bR|?0SRQJ$bYpu%K44_q!1<|RjdY!!t;p9b>{!11
zM(-Lgd)+m*vZE@w|79_W(sl4pFsTxtHsPSe0B)<=)V*nz)Ly3<9~ejbFfp9pVXL#V
zs(IO)#_C2p)>9e<^RI+S1(z)DnrqsslNAyS;v0RDh^154;6Z{oS7}g2ky`B5)|D08
zCNSL7OgZApM<2D@TF<f5j}Kww9d$rO=sIzQgV{s!?6S!0_QGdnv(!%`e;xq@w~K=p
z^VqBkT@jmWYir<Sg$b>Ky*szYhAH6yIY&X(3VwTIV?pc2Cb0vropc~2bY$?MQn<7W
z9Jf2#kVKSn%2q+LJgY0m?Oc`jhuE`33b%DNIjK;Qz>F9LEJi`%QAO;>lq=JVtO#4*
zjZ$Y}45pBE6`ytRG2!Y5F=}!l*qv?Um*0G&(?~8%&}g^@?rOcz0_0f~Aod1}7+%I#
zCfg159STm6#b4&TeVB$WIGA{+v=P}p<-UDW^;1k%!6B-Km6n6la~fW2qip!xDTCrZ
zDi`dujcpz%!!SCm7=9o{ACkZc1PkALCkfc5_F*fBhUdl!3jh!|z{jNvqTXnyjI=|o
zA={kvIW;~+uH2;4%rzU#S;k0NSI8K*H@6W&qSMJ&q#`@8npMS1sn2+fLxZ}z^?cK|
z%;I$*SBg`}Q7%`LLT+-yO*)B05uU5c3iotlc_s|G1Xnt+{Bo9_f<_}06>&|Ajvm=p
zv^U<?Be`CiHIN#sXR*~l<w!G40wTxT?rmRG0SHgcpL1ZzOooR>&@#pdLVHUa(rF4|
za8Qr~$ZNI@Q4pG-qMqH?u30sW*`r6*(%fL~nz8$9jCemnsn!~`(&)=s2NM*do+&Iw
zu~KhM)<G@0N1We*kEjgpq~N{1t--<uF2BoZ2bRRVngUTOUiS9x!8)rl-0lu6w?mA)
z!vp)uM#oT_?CsiwGa&~|yp>%e;+<7BaHvL7GG)YBY3_nTFbNPD3O?vSCiigg!nNXQ
zHe64`9~#^T;nso0yrhevm1&UcQ3mDbY@~H{<&gY^H1&By<6Ef>HnVNQM%$J_J<7gU
z3pa$U3BBrxMl<aV73t~?lNs?DpcgufZE2k01~xPrHRwupkJv=pS{;rl;KT5+#G=sA
zhG-BDUQBmYpR*c~K1e>^)_wM2hAQg2dxPzV#iY4muZ=ZQ+BC9lvWNX^+5zQkQj|pp
zyx9W@xHnvYqeF8JAoq~9TUXfj!_U8ncfbCMtFON1zx|(sOGy$+#&e0vJl-ure2pTh
z2yD;HwoHBE_|Ako*TzWbXrqga1V$&DsNse2MJmMp1?5)4>Ku$ZHO!%k1DnkKy*D=G
z{LJyJy;M~#`*|(q0AjRhB*%#vOM67D2%BoRqN!^xIXKQ2VRFbtB!#`BqfYod-O*85
zc}7lz9y$2pQfJXvn_ZY<vo@Wpt-l3VIKQi-Bs_~V&F{5A-6@Aa#dWZp(2=}h5ZH<X
zN&kI@(n~Yd8~KDKA}HKQ%D(blltu);hV>cvLGruV_HL}%iaW*V>A=ziyA1|24S(ho
zX?w8y6dk97TOcSA)0%^c$7*9z*bdy@9%=&F#{M`Ph||u!faP9E5eQ^mfVe%J&rl&u
zw4HDlIj|(h^+U7pww}$E#J7{m3;eE;)t;5wJt7b$+ebQ#h^wV!a<;g47_L)Rh0Guk
z;%bjJx?Z#LZl{sVJGRD9Ss8kEE^gtw?W@|5!91n2Nxx{YG(D(0N9?h#BRYgM%y%)s
z$Gk8L-@LG6VaKJ%(S_=teRcQjo!ZvW%lA02k}^3PFUuo*O#>-2o$(sXI3RIZvYRO_
zK4sp${CLRe;{BnQI~aK+Zo2tqduK`0%WraiE*a0N@o@Q_DAJEtm$ZMx=@VEX8;qdN
z*mpUB+MejvLfUeW=J?!9T~@6O@JZ#dfEjo8oUg5WQpppoM<@<!CDx&XPfDcD;uJ;o
za8&o~>#D0-ZpsNc6eb}NZo|mfs|Q_xv1=|)+SfP6zG_ZBR+=)7a1!=#e{yczzM=Y|
zTk<F;JS-@#%FUu$ls9}zNRk(^B1M2_${gR=9zTBR2{dGy`Ke8~La_Lz?iCTU#}ZW$
zsq&uBA9e7_T34@FW#7DHTnJGi;!f|2dIQ=59DK}{K2&@sPqUI6^jqrA%;!R`Mq<iY
zNXfh5b1TUV-Q-)LOxU~HmvFZ2gFUBNWFe2Gq&Jyh<cj;&!YGWcN_~RS;R0Y{{m9F$
zeM$^0Pkf2BWbaO@d+Y12aemh0wHBq&S8-^2PYq#Ywt-Pd>>@0q@u@gz@2!{ZMbaj8
zmIy<=QruQ?ZMz(NOoilwaIelkXYXtGupKJ4MA?r^DHVM5N-}}MLn+(9GY=3%`u?t0
zN_U{J0J0P?P28Hv2l{jNZQY~UYgup#^Jt73Yp8?oc%_66VhQcr3#Rg8qzWMx9dlsG
zgP;&a1WTQXV`h0Ko+bN0<F6Cs`ctikwmLD&`OKsC9j*T%E@)M$MlOe~c*97Xh!DH5
zL!X;JWFPE0p>(^@uN;_^$%P#+zrsFL9ZDS{rDu@?HAvZK&N=ub-#bWe`ObP+B^5FX
zyBt{V0E*=f%8w{YlnAMh#GAG+UQ%^IgRX<*PBS9lW8_$(3b<!uK4jmWFO7U2rx9v&
z2K}-!B>r$&hLtt#;Cqt3RjP!@7pED$3gxrbW3$C6rssQ;R+rs(rSmg7ph(IvvQc#5
z`)Z^q4~7ky@c@(S3`}lbB<+knf?W3f-J2A1EG0(K7urixN(QgNCctCJb~G__c#J?#
zTJd5gB{eAIn4R5dh$*AQQKY@W+Qud$_pZ85L;!7Z-DwAw<XzHamtSYQ4?jPw36FhA
zd(;0{|8w4?Z(7(fGBkzy7oiM$y1wpVN7RBn<iK(rglZZ}2V<x5rm}MGw!KZxL`4ms
ztOLn%&Sa&W;omX-U9^4eeH0l|{GE6kuowqyVi=#qt(rT3e|zdBWL0Krj{}jdh=rj;
z<`2}Jxr`*1A&bTWYnaGd%9(|M>_O!_SYIbkrfSd}!4248$sJMlajQPru2W$%C{j5H
zz|vC1x$Fabs?A@wlEjnc?C3Dz7fRyK4eC&P0*;L-r<KzAs7flx4u1IX^I0Q5>d|2i
z8Y7Nz;DA%~P;G4L9AFDSa`^eRjKnoB|9|N}=Yf((whkujX!q@5{bGiXIj~xxS%5(-
zN87YEM4SzFtVSd9OBib<v`d^g>R?jue4wYqo8$HHw1^yJt2*Jp;^9lVTFAhXl_0MZ
z-91VNi1qCa*6Pp6k9x9=Q`D!Q7(-PDRFr!N<v)~F{w|7~r;1&*jd;=XpA|cCIgvNb
zRk=YbHiGX~CYz*&L-ustpVmKpfH<Y}Lj^|2((*K8z*Q+gP0N@X?%jc%&Vk`5RO-ZT
zb(#`}Zs)khah0z+lbi_Af86<<OponV^&L)Pc01b*T33U%-W*)i!55=Erz{$H&Ix;_
zU~=ba8)hgSSgt=zDbfi@D9Gp7xklk^(8^Vm+dB6}FzPGh*qf4udcJ$Eg#D29yunhq
zNfXkAM)B*1wi2yCVk?k%yJ8OVWRJ?6v}fB^cEfcyI*nNP2e4!xv*%twjJe2Da)@A4
z2cHX78nj(p@_{-cQneJKe3^rJF`u!p<5mCLfA#}y_9&yM@<!O0XB=RX^(ve?J^R6K
z1R^zp(LU?IlJ;aK4SHd(Rrr8}8H_6P?F(<4o^={UV!87#x>GJCA>e-+bAHEFl8H1K
zo%PceI&xUlfthq!X*Kl@KJTpUriHaUXP5K&v*TF#`JE%Tb^w$4p(5_4y+v66;yY`P
z<*H&05;1cN$DYK#A8Xv6UmX7!hKPexFC0wWsrsq=FFy_eW)z3m_&NLG98_iU4bP5^
z%U9^Ya$}jD940cp_u!CRgFli(DQoq>Y2?(bB<5h;ezc~h>Z_>=v;#{yY0ZN}_G9gl
z_z$*lOhuL%0~(!^U{%HUlA|%~ih}v^HdL#&A;Hk9>65yzfzQLDkJwK%3|>*~I1xfE
zkAZ{9!G2@EH)lWDO;DaA%&9W0q;qJsVDfuZ<fdmo)m<nylR?F@HJoxV`M|01-oy6O
zjU<I*Tu&>awte@;u?OvE+U0i=(eJcUzF~rOoDkj&``LO`IvXN#wI;n`Bpb>of1a=A
zFN*5Vb%ZDdRBZeOW#_%&A>G(RA`&yki%BPh{pVY+P^t$`E60@`MWX&LWD+g%{q_rO
zRdUQEWpl-c1`aTH%8k&*)-kds?H9XwJb|5{lZ8WgBzM5UmmhVp07~r>A@z&nowHx+
zNL3{$HJH*7x(r7LFMs00EA5xtYz=MGu+oeV$aRqpy|8iFuhf92MLIGFyuqZxDV7OK
zmi=n?Y8CfM(n_T+#S-v2ad5Uh!F-R{uN5q+yvDG-(;H0n0JY8}SpRyHkI638kyQ>L
z2av=<+KO%cji%&m;m}hAVcIn%M~PsP*vwn8mYufWEaPI8o1;HW8PNmcw~-W$e^J&g
ztliGFk~bb1>)UU24-CggWk$<bv51j`t{qneRaBqux9gttcnc%egPFs@#BgM-V6ppy
z_B)NH27M0qqE%N!k`*%f5&PZlc?xRL{7SZ!X}e*Pr8PMLCTqW!izy;~oJNpQUZ8Jm
zmA90ds|7_3Nc;UZ-V?C_4je$-CRs`9%#&d#P}aDZu2SAr&HWD=2@nRXxaH7S*i}wx
z`Ooii@M32Ok%S!mVH-}UC!;{90Eggkz`<y8ISOvJ2_b*fvG}MQ5GFa<doh#BVsq)h
zWk-JDs-%ixQb3^;#FIa6RB;a_Ifou{+Tq>NZA$akE|hyMp8$8)fu70~mv!V%a>pUp
zd(!zWo~0r}WOMy#O^xU~1<0p-5vGiCMNl8KKg&h7Wo)bJlse5+ksp58{=7b4brh7C
zql~sRsDcq6y}dt7bXL#)qAtv#*Iq@62BDIBykXMLB#a6J9>-q%m+eJ&zQy?+WYk0^
zeC+mDjlX``YhL3ta(m+e&G}cx6@_@)U)QHk4d#dop_P$B_bAhTV1HAO-%n|na1PBs
z<-ls`X{DrwEZV<df7@NH;<0jHKIOpT>U6Ays{`6my8Yj^ZZ5z1dZ$tRL%}8K<^21C
z!L`B|LZqBG*rGY$fc-<G=N{6w5|jYMH_JUIQ<6&lMqZO>o5-twtOsZGn@l%H9hhcs
z+5NA)-~Oq6A##$8=kZrsr-KoVucB_qr2TVI8^am?CZ#Ge$=Z4snRabeTwBZiaM)BR
zYGMD<y2(0~xT|PIDYWhE6!4miO71yf|C(JyZQJ7fR!27Iz)$<PCOb?IQFkixCy;7^
zin-m2zVq*GsJHY-oJMpdS-yz;y#2=#GL*ISe8h#xq7SAd5CTi-q)re}!DDq*g=hb>
zQOw=g_8EKHT1s4beEvE6ucEHOKpP{id{zd@GVT>N!hjK1ZgwQ@RRZnEjMJ(i<LLQ#
z=d<=<1;g+{7FO~dU};{tvp0*$d(uAqNPPIJ)vKJJMF|<_h-H{}DR@Ae+XefGBgIp0
zB~)Simf7hU??B^Y6iS;>TG2-yN!obZeQ$7n>-(mMF+un2qmCqXH*G6R>{GMF!HAwP
z!T0Q=>nd%pIzuP}eM!r7F!_xU11LUw&_1TSnhZ$3n`cX^K~!xR38sX^DUBn^%RaX5
z$=xaConR>rMp|FSMn|0Vb645N<wxLfc!E$WTIwo_#h8738xTsL3{4>G%{ZX0kE*ZB
z8nRC~QqrI@o(4|yO1(t7zyZidAy@iXO%Yk&u!7F>WS^LCTRlCr)%l&W;<`1f2lh$z
zB8(kuf8mXBU@?JOSIVRn5ia%7e{y~JKK%J?%aO3BQ~VA%7?J<ocdfHeITCWw)I-j%
zeBsQt=~?^KZloogGG3?)T7ah<jE00Xj8XfvBPHw;JC{}ARp<a{t|E3r6=cB$Zl8Xn
z1p4JSEq5AmL7rTrc+uVV8SM*ok|x~z85Fcdm;%jo{OTEdQC$aDm#YZ|^prPPmadk+
znv$*Eb#iZL^v^pmb$x{#b)VJV$LJ_K1`DVIixEw?ouVZap|vZF1&SU2?7Al<G!fB7
zZ!pcQVzYRXeNGL_H!pls{4WfI4m5R|m2;2S=O*nbeln6@6;|#5Yv3eoz}qiBZlBkd
zX=n*zKQaQS+D^zV-taP?U1XM%7CMW(eSUk9EK|L)P{$l#ya>{Xc_r!Hz90wZ-XWAY
ze#g#~a$D4w7W*d1<O|ynU=PUZi)S5J>OY&eP4^$NFRGixJ{5&5(Jy{G7}4EBy{-1*
zdJaw^GzOgneb9lW6hl`j#-%KyFK(b!+xirSMFep`ddM81TQ4~hy8s!OnN~qNx}719
z$#}x9Z&w^yD*dI&Ptv;_u@csjw7E(%bS?i~XJ67y1}RZ|fQ~x>76ekUWDu36f$C`d
zJ%*wC%0@s@UPPzdXvav)G2)ejk8xM(i1u9y;Mtcp9t~$FIAq3`03I9~C;#T9`I)4O
zjS)mK>a<11>xs^_mmPWj75tuI#hhmS%MT<!c4ibSq_n<OVYYNqnQ7b-qm<)Qv<~Kw
zUDYOXFt&AK$Z5uf89xVsGgEPsEo;1D6?=ozs;`PVAmZ2d^2YWsZb-e5JQRzS-(rM{
zb0Sw5qTTIWThyjKDETznC-(bSwDE>*A!9{AOX_x7>L<P7$uNg$KxCJn;F+O@L3U5t
zE30QL7{*%*=9o8_ELxJ$&{pkLZ4TZsJ2Z;n+W|;j9KGAFZeub<VewI?8Q)4Tv~uQm
zv(I1C^<M|V>MNxKP?OykF%jppcu1LD+x1Kbbwx2k2bR>uy_ki9uC!IiWg;E-x_S?L
zyIgIEMp8jJ8NtECow9s%R#YtGa%d5tXs>P*L|8V*x;pKmW1X~^CG*=_SMBcWYq{hO
zTk(_VT$1?CH4Cb9mBwOW$GDD-8{3BVY0Hx`R!$E+LRLPl7_FTD=(pX_Xy>}?uD#Z2
zCGS{I*{gxw*d_uNfT%O<0Fnk-tjJX`W6=?rO?FdZ<N=WkRE{j@2~0}r9Kbd`G_ad<
zy=>mrpKyLhk6@R~63$AjD7Vzt{-46Cd=^sCh~jXriZN68AhHF<FoHG9=E~x-*R;`K
zFXWiY!2rDnST`@4WpM(=IsWRz%hzVlF1;T;mf!KE&@aMr=Duq7Puc6bHy@Etgl({$
zHy2NI@Ofth+hxK>RIs<#=SV0dKxR0nUGq*}9p*ntNlQYI?aNw!*|T-RRVWvzgl&WD
z@Zm6o^K}d!nUR(7<<*nipyY~|*st_`Flk{&8<*K9&hUIgM`mIV3=GRx0yO*RqL$!X
zRRbe-D&IAQ#3A(KwxUh+rlFbH=}D(85oQ=JiUVk6BP|^!HE^0ybM?JY3)M?6y^s0F
z_hqZP*;l<3Nh=Z;#M31QOZh;qcdn}0-;LN2j5Mxo)r(eFC)=izM-bz;b};G`I}}-(
zwx;MWjh$ira9~L#3O`~PB5enHuc!^XwVsVl8sSmaHEFA=Fb5{rSLo6apkTK(3TMaK
zmxJnyFmXtUPEXnGZSV#-hT-D0>kE}*QK7-1TBqlEZP&}Nphno#(&CW~pY*v{Q%Xnq
z$R7K~>X%C2$YUoB3RqDt6>AxF6;v0`Ic#^-RHC$?Inj#mY$QT}s;%q3Q+7;g0p}f9
za+8s5(1We7ODctdDBx$j!PqPq#$^9@HYua`@DwK<$gT<K8fNW(^E><xIJ@gexeRec
z^oDi3B@k`axFm;dLpOw>1;Es`T~Kz5AyN%y@Hw5hQ<V&I4wvkk>Yk<Mt+Z7_p7T#S
z_!z?6TqR3uyP(XEZ7fV?GUi`R#R&%^5$Y`jKyAvO6aB5u&w5h06=Uk7`jc(0eq}sp
z2ru4XVrPmtMYp#%x9frti3vRGz~ZSeQNA|RIEYKdSnqCM8>uh*<~iEIfhB_?;#YxW
zs`I|YacuV#K1C}|5<=QZ%_c_68(yqO8)4Q#jT5r?^5^U;>Yd8RQYJr9Uk)r4A5wq@
zc5mwyj6m*gYHT^M`i4c;A08@uOZ7+JrExzlGiMq|RW8!O*O)4hQ$ZO-E@B=$Y4>&S
zNhd9&?+vE0lCxT&nD^&(6l|CsW;Y||dDMZ$QyF_WV#b&UP1sv=0T3<#L+m-t{It!e
z@b<R4P25X$H$;BHuN;hs6bbWQd;5}4Y6<k>Q&>2VY@TuOBGQE&#0U%stD$t2eP!L#
z)@S(Gm>h}Ml*saiUut|@;(xWFx<BN>;BcCJ3FeWp<lu|X?LqobM$riyF4zHN!|>_y
z9$-N}Y4Xd7NPe1jwUNTh)}=aGSPq!tBOwRkc69JHT?%2({ZCVS(nebk7`51{P$z*&
zU6@tzN9)xEX+V@xaL|F}u)h30+j68r2g>&yVaU{}Z43jc;lVzs9JaOT)Qm65;!*2i
z;6Sry?Ah|^(|7tuAee2d{VNR#B8jw#sEC6sQpst#Dn>A?$zmeJO*>XF9XA$5u28vw
zrNwB9Q@KXwChUPmH1nq-m;e_KuUE<!7>QW~M)xSh*|Y845JBJN)dW_Xt^!P+<=<I0
z-s3i&Y-i<))m}e)BjYxa(Cuh>{LKC;qz!FB8tP>03NTrBbU>w;;6VqIks5Vi?7;f1
zGhts=;?x-jR<o;0A>npS`nRe46uS4c{yFV73VFiWJL*;qFYdGuoYFlHrW^#_$-K3+
zQu7KS$)>YA3p*afJ53EdRYL1vYaSjU^n&wQdN#-fy8=Vz=)kMRncHM$+f3_zo;e1G
z)MhD-4oohO(DJm%j@xWU>Rqy7kWHN-;^6Zh_8L8hrEL#3ScrVm*^hgJh4V&B=JFHv
zQ1v8*;3yrHi^;)=;S9*{Lz;g$7svq3V8&^rlmpc^PhtA;Qm342blk;E>l_ko(%J@)
z#w3$DSJkJ6*+#B&d!)f|a;1;vv;$khBAE*GqBI3O`T|VAVvBj&!DzXzx@*I|_Ep_g
z(n%$gpJFAEpyFVb40h_NkL71)DzeiOUl`(Syq<k^cexy~S~+_iSdy2Hf<WiDuc?c)
z)hW_VaXCw1u@UWL1$Z07)Cxle3rP?<R)q`j+Ss>Fj@dh#co%Wc1E-y$(eUi}@W39g
zCuJg7m_K3^nzjR8Pr7w{)}E+KTlkey>crn7ypUxDKA$|^NAk9>J(3mem*{H2gLL^J
z^wyM+;Yx@&p?f%Y=v|vRrQ++_K(Njf$is=O4ov8^R1*36LQ8_p@@f0M!L+MYnzVgG
zT_?MMIJ}O*ZZH{2B-N|#;Wy@tPstI??<EexbmQ_K=>%|F)df2}zuUg4=t2dw(wR}N
zS_hN#&EnCftdbN8JeCPj$LD@?^{G@F!W`UzrAcu44Y%00)C1DAkg6pjJsU5Rgy9u@
z5?)zIPt8AR@2Y#2<oYl-Dm5+z@}M`owhBoeIOese!lbJ_!f&nnMcSCM!s01Y7lXzO
z?{)AoC`+<&Y65dBC)<ed$#*9ONZuvA(`kwu6Rok0B511i{+_1&t(}<2pg9sc0T$C9
z<!DNURK2&+tfB*YTby=M);dbjPulz1vKj8P<!;^P0Ak45q9t2~C7mV2mqhhddw+XF
z#c&NAh-hggY2DwJbH8crZFf4q1)iq-#kc3rM<09C`I+1IihJw>O_HKa#HiCM9uRJG
zM7?M1I~q*h6DEOR8wVD{r)`666ZuK0TnY<GEj`wqeXxB75|4^#L!n7CqCCvo;NxwY
zB$=2`;Bo#?J-KWaC=s&tC^10XL&jsg)6P}8lY;&#Z$RY=T)5xa%|+ypWGyLx^+C#v
z7TdjO-_=HMe3mF+WE=<bavnz)6ffM=e=*3Vf?Wz+GSeNf@2=+(EQ5T!TSgZLmfLVh
z?~$9SZJcO_?>X}PoqE0~^#8%xq+-nMVJX&zVK8U`k%q*l^UhN>xf_S3?0f6g``I@#
zNkiolDFEpWlUB|M?4}!vBKW?#M=Vc5c^n54$(3M|fTwi!o0$Fgx93;vOw78x%ML~>
z9jDU4Dcg~R_0~rpyU+QZG=A4K@l&?5o{;tdeHh|~hSI@^8rSt$D7(5o9S(Pr4B6tS
z2fSgLG^j|(++TNh_sCzZt6LB@nB0!YLQ<To?a2{az4kWeXWG(6AHXrtx4re0HbR^T
z%`o#c2UGlSi4_AOMLbz4iu)R#y;#wia$z7&z(qr-aPTFaEf%g1cb4s6LYzqd$m>TP
zSiEiNYQhGHB_I`{1Km9wHY*q#NrNeI5*b=~1a>gzWdrW3fz!}jP{jlP!?-=!%mOOl
z)_<7w<A8FgHobZM`V|}RwWo5ZZo2U%ub+=#dOB3!lJh6KCs9!_@#Eew(kfc85QK~E
z>~Px)R>NRu#Cq%^bs)qe1uHE)_5@(Lf!r}WY)9Leux)k+)Gk_Z9DF>Koh-J2Ggx|$
zbuUuhv25b|9axQR6l`Mm(_)7gvG7q%?RepdbzQW|%O0Q0&cTajFyGEbfYOOJj>FSq
zTelG~J>!5x%o}Hi=j>#D!|EYQi#Uy34x89IdUk5bK&G^-VxkJE0TUmSu>qmX+SAps
z7O|BH|7TQ~Rzu?8gT%t@MrYFmyk~xoozDJTcMGws{4QQdKt%AOy>_O-!d<13$7dW^
zD%Zo~{SiCc4T^$;M1%@T)Jfi2YR4RW38b<V$fTwmZ6#ejQ+LYP+mfA>_#n%MggWqX
zCw30`c~RDsr9sUXJ6F9@(L+2vEW=X{EcZmFlC&-?4vIrn`a{}s&bP@Re2n9IQ92xu
z800ppO<rireKWa)I09)W$%N}ABnNx88;kU1pChOXST5cOHxxIojn1BHy+<sP5RbZ^
z49OcNWeW*eVq&V7ezDyXTV-2B6oPXwqIb@xnI5wrsE%mgQ?UagFC19X5806fKL23O
z(WOH?>9o^|vEhlBL36ugJ4I<o7CFE87044=BMUPGdlr~{?C|UaMxsk?>Sy|66HY4z
zDIc3&j)iWJ=ohbg67mC<%wuSjieaewm-DGqqQ~7a>NKN!^4T>1Pxv0_ih|RAsPI9|
ze6RdJkXX2VwL^eU!e#ii{!bw%JYSzgU2r1)6}&65is0V%G`OLZ!l%RJ*q!rdKiqaV
zG-4_4&pD8o<n*wm@|R|?A1QC21okxYMRC*M{VFFZc|8IPNzCP1^-@8@<FoiBr1t)3
zJz3W&#OFCJJiH61y<svgG;_Pq{NY?bmQ?$aWv_I8iYi7_3EF<Vjq<HkClf#oOV-y=
z1*M;;kC$S&xEWoLd0iO|4nFRQt&71qK_JjiHW;m_*+SGduyku^&mitRzUHUuLRqFn
zS_CRMu(pVU^pcxmfwP}(7^Ds5BPhZ!Eba~_>4lsF!e=M#XL5C3dN*rY$Hmc3?!OTU
z^=GrOY}qweUF9^yUvKXd;q-GkQ0s;U&d<DI?aI~m^X+?NjpfB>=Jz?U=w@YyMQmuU
zjbA8?bO;=0Y+q#@bTB!-kXUTySRa0|Jv%}`#4n$8VEH+Cq7)BhztkonOyH7^CAW44
zY;XARAo0p0&~C1=h;+7Jt}ZDYCW>FH;D?N+4ko%J)+@ilS^PY|(mc{jye#0^!GeZ)
z{-P|ywKe^Rl>*H(&)ToH&t#tD@888I0sBA9TiB26*K(W*f7~|XG;*Kd%+7)RdV5S5
z5`&{Uuo(3mTe)5o<8xtQul+{j))L@Hu%nDSpy*rFRbrJKyZpHQW;c>YB0NMf71B^u
z_WfBzVDRw*))!vu9M7mIWxv&ZDcU(#OKT{YR_JIC%2PP$Z`XCiu7w#xRTY2*zuJp%
z--?xM6KWYEzmr6`dGm%%&hIjlr9DqkMd?>=c4+!$zngbVZy6hPe#g_36f;aCRSRth
z7wq?vtZ%yhM(3x<`VmH`XTRUk_=9s!Lu1c^I)lsW5861(q{}eG0Xd<FgQ-suKS*s_
z_^9QHaQ267n2@^U7$g_gfoa=(r1yyZQK418h~qclsD~X`O{(Z6t&QsWA^YQcS=)e)
zGFy4ha>fCG*Mo0)^^Nu?ZQmIo`e5KRYll$UxNH=aK2jUlpSEs>4k+R}>cFBOIWu7g
z=W4}N|5^KJK1xMex+A8b9W)>B@JpEF(vskRUT@>85KJOK*BdOY`=Q*~Uv%#Y4T_Ck
zaTiA*@?ftHKI49HaZu}ND?8&~R;Q&5%Yf6JbmCx=i%iavq7G&LRl6H`MFuzm&;V)H
z*A0=6vCaOv@n{*;ODJ`KA{q3tXD94$s`2>z(_5WJYCH%)DoOvh)q923Xm8Wsi1f?9
z4qmKwXyk!0`@3$=qmC>~T;d;d?qC{|1g227XYB9W%%F`{F>q#q@HYoj<QNVt)Nyek
z1@%P2uKhz&$gS&EJ3rOk&3yVQ4okLwY!3~wU232_XcFteG<CH<wXHWYXa7{!hDVQ2
z!<qx}Vk2@ex>o5_<5;ZOMgBQIGwSQ`JGT)Ex!J#TFRU1!u<>D>b}+?7wdT`Q^sj9k
z%1WlgSC_zo8*}=){acNIx-OkAZ5n5A>ZwxRK?k2)koh3!oE3M@{=I{phL=T*mM_96
z7m{;f|IvN+#n27UN+gv<z`@kGM-W18XkqM^)#a%DXZMDqq3vj>3bwFg3t^iI<AY}Y
ztKI-3QI5cZGIdi90HdK>bL(9j*O(ox86kd6H5&#FL~`b=VzqhBhaJuSXoa<Oh1Be$
zfWL!}QNTGIMS_qt?86Htv06)q@$n8uW}=E>sff=<<aG-idpwS_FitiA7QAAq^+Epn
zllGC>v4tILMjxW&56vWJn1&1t#An+_wM&R5Al`zeE?L-tQ+*r<h7MEz=%c~eR&HG5
z{ESzn0?%QNun=ub&)CNlCMP{B7HAP2b}(9MoFY50k8N*TNleqI(~J(4^A*|gynWo!
zBGY@uwtLrDe#<TP@ketvTADkSM<U2BA$lV=#<{Avt?Gq_+R1o!f{%SdhtIQq<Dik?
zZV|-d;Kharhn9@{i7)tINhXK21!1!RlSWqdR&6diz&1jB=|_}$pic+wlj?0{5(u}A
z%)=7Ho<(#6A0r^OR$;}QQedClMqt(YyPZZ0uN-i2MSts@eM<KQvYm@7g;$g<PT3hx
zI(Y3wU-FV`>{E{hpBePWXM@Mm7V04+?oX>Xi%K)2<G2G$g=sy(H+}o`9Eyb<8z=$j
zw32!lEG=FPp7t5ttfwDMVQ)&WJM9g_(8F4?im=4wWA>u%le3K0Q^ZgJ%l%>SP>*oY
zY5U9<jC+orG?*tHjQS|gE4Cf`tgg#c4eL40e9&D;K=`xlv)ebkrH3t)mTq3Lg|JPm
z$)A(M@#gg#oS&l7g&iyJw9n0}IP50;@ifJ(<;UMHL)+)o_{*26^?iiBLDia+M}Yq0
z;A5cnQzyKllI-*I84EimhDM0GcG_Jg|2gSt{JUUZP*0XoLDgL_nLtwtqsBlwf5gFG
zNlzF~9jzVpuXraXrt=rs4$6F1+6XfFe*40srQ{$CZllW4BeDzt%;Dj~Ci+mp7v%`N
z^@>+IKQ#hWj~LjC^EV%N-ua!9@YL*<E%wE2yjQMUv1zk+&D&ONSbe*_B=5WWiYu;g
ze#bC~3+RM(#7(=RU@0vtuH`ugRtt7<L23y});m8>-Q2oIVUoDVkt%jOuw?n^Qj1gD
zWQ$OM1g8|(m5oQUWlqW`?m(h9O2d)@PGpCDX+1?f%uJUsu|{MJu4@OMUw-?V6|3!~
z`Gg3eNX3z$-W*R#?!qm`Ah(wtjS$oobIvbO&5B!ZUAuApn$>nyw^TAGO|=50d7t%$
zNmG&*iR1>n4z{ejN4{_Q9a<_4n4~d0Vu~z)bJ)xCjdDpd<Ks>{u91md-cV(?yrSKQ
zFf5Uom<$J&w5UP|cp)E&<lk4eA;u+@kzdj@x;$J5pLZdk4NXmPCaBiNtB!_l#s0w=
zJ2TV$$vS-pcnv?NXTm&?g6-<|b(CU~EUlvwv+N9*BvYQkzBI6Fjz0ex3?|Em5*IzG
zRczT=&>fUm+CBYoyS5%E<wU}2l<2wqI2I@{@%mDtproC4UAsC2=`cXYVm|0#;%ZHW
zwiONmwO7~GV*WC$c7^*LjHqM`cO>Sm>)ZN>2n(kv<Y4>QG)ECEZ=_zqMv=81dlS+%
zTP65p_T#e;W6D#6$_;f(<VVE9S6u#c(lv229d__3AKt;~&*MD~-MF!q^kesX{hXW{
zv6~t#Z-G*sR_x>X)r!c<(hIxn=JxGj4P;A13<efgYn|u>K?SRX8fdpPB4Jb{EM|fy
z(GB$UJZcR1?48P6VJV)l*W~6#3G$&4rxD9I;)O)>Y!y77K@Z!Ly|!^pinaAerk!@N
zimYqUA}eLuKm@keHJ-OVi-|>k2Ege30azrt7N}pGI)Ddsr@g)~NiV!O)Ibi_axh5&
zN$_m3&)Ju?9pb8&I=>4qV|vu`wl8lm1OjpsSlTCj5f)537WEt2b&71{LnTZ!m^Kqt
z$vaBm_=>J~-I<S!I+X2Ta_Oy^9HHcw>LQUFvoa}f(3`+iL*#6(p-pIKN+P6sZ==kW
zQIvpY(%LM%YmCgOfvs-KZd;!_F!%$|+PZ1^a$A!>Z{`T(`l(u8x8`i#L-FT<S3?YP
z*lue=%;tOUavHV#<fxQo=78PaF4(YcozsX5B&f->Ln6Pet?%D=S@lmvb#f!2@H&{h
z8VkB?>nfr6#*UN};?JmJ1y~ZO;^1TJ)GndzQmR6Q%fjHaY4RN4Bz)K%_2#V5z-+L!
z>b}zsCh2)>QU|v(s_WXAAwJJ?BA^2Z8WpS2GS~}m%S?Z?bDa0<tMiKEhLOOYPCBsE
zUdd9GjqT2oEtfd;>42h}Ghr;Fo4G26VLx)Sj-2PO-IdEc_KIUpOM1(Q4nVLC^;Kz_
z(YI&#*dYg#V&Cfi=+>USsZqw-%2uCqK#~opXEb9Q+r%xu`4*=UebDg8ML_id+tk5S
zuO@Yw29r@1Gc0n-3EN!PDUDiQHL7WNgHda+rtzG;xyk*VW1%sj4C&GY9dNFSo@lH!
zo#@?mcfk~RuHXpeD*?+XjB59?lHHS|{{VJ}5Rz%9qP$_~LHmmOa>mg4T_2lUBHE4J
z+tJE1iuqtvE8W7uBy&U7D0eB7xL73TE$#E+u?k92P8_h5IM&b2Y_t11cBk$CKtz=!
zO~(px@L$ZV{NGesd6nH?eQf%g*k*7$2NQkF#NrS-idb1^tg5$mcdD#}CRxsO%9A-5
zjfEmyM@ERQ(B}WPMHmI3M;P0I1Jf6&&Oc^{x7RlZQSdy1MZLik;&USVWnbA=l`={w
zvFZS7=r6`%Rp7x;n@MtALx*=D!RkdzGLCDT)QUN7!-d(W>@8qB^*O3S<zSK?Mz)Pp
z372?)kw*V3>iocg1IiLoeU2YNHI7V?Q5$XDTuB8Yl{*KJ^T)YCTdkM0LrI=V=XdS6
zGOMJTFIn}ZZK+R-c%e`vc6x)!X{79NC6sJ!*F}0uC>-8k`^8$=b6~But**m#!}5#8
z<-n2_6+i_gslciHnEHJRQEaSzNPo+gfr>E$NEgZ0sBs&jmOYSsVl`XCbf1E`^`h9q
zSo^XCZ*Oa7<?I&k8ml*Ku<^QDEM9?5IuaY0_8=Lour-9qdLkb{N+Ksc)2KDAKqm?9
zdu_5{nctWy1h5iN*i1!23(DrkN@e|y?dJ(}&4etCS0Vh8JY;Xzrq8vKI)`RwP#Vx4
zZK{#R7D6E>`vkZGNly8uSrXOl9gS~Otp*FI(-xcDd>1*o)74L@GRn@D+5#-;lO0fz
zz7Y$H7<bwRxxG%$&Nx4d)D^E2a)ixv)FZD??nuaPc&NeT3JxhY8DcuzVp?pr?riSI
zF6rID<a?M9Ibu*b?ZF0PdgRu5>hcr52vd^0cFtY)P`x#WT5r;BDy1~tl;Gppnc~M*
zIKR&x?w+j^du1Yg(t$<qb2k!yQBTrj%JJbg*M0Y*e_qS_n1d0MUiR44S6*q4w8yX$
z$Zx6n4;7EUSZo}81;k~h{E|GW#dvjlw3{7$C>><m&(B%JP<K$nBCo4<Bm1g)vqUyo
zf#gD8s;7a^o&8RkCkFP|3woOIWRdk1E8@I^7lCcS3rZ@0eRYkQK1p&vxuFg$x2Mfy
z?tM+t4H>4+PZ3{#D?u>cC|4@ow8|I&5q5Sf8p1o<OklK9HVv8P4lHIuZ<2J6hNQs@
z)2?Kq#|y746}d+xWn~ONxOW_Uve6kVdw6uF?TLDeD!6JONi1++<xrB_G&XMDrSfJ=
zVj=Boa}Y>r$4EGEn)26ie8@Dra8E|CFztL@BaZnKGBB`VNletD0COdG@aV-=FI%wA
znpK__vQ$Vb%Ch|Wb}g99m}4GgH|DPd@65N$LE_r{$Vprf=kR*iH!MyNwm1#gMF*A>
zw0^}p`^H98P^~0eITw<s8YWLKZ}Fc5Alf&z2Q0tgW~ZTdj`o<j9w*WD?3?S68QLow
z0&-@UQaTK@tL1wbX0Hi7X5UiNr&t1gUskH<j)8FSv0&4SuZ?++y{nt7(@?#vQHpQ)
zZ$(F)s~Wq4FoI<4TbB?E7BI>C^hT_U-prqJ@VOqt9?HHE#FnczldO<+?cMDutSa?6
z>hrMnREj|*zrn|wIlK?^jW)jv_MXN05c6c?Jnz6#(~(|8>3d2-zPEvHogV9rDu>$v
zN!MA)oV_pC%0^kt>i66TVSev#FD&PRG=c_WNFkXZKl`?}OHi|b@H3jleNKcWMbUjr
z?bc!Ov2X9LS}gVpwW2*p3y|1_h1t2P*u5AzoQRpxj;gdMn5}&v`OT%T1N#Kbo3;d!
zOFEfyuYE@&Mm|7UIu}`(5f|cM$}p%HT?qSwb!o&O@$xt(lUM9^FybVuaBJI#y6(ss
znTBsoafT`$aPTojBEqQCi$C`}+mq$Iq4TIkkS{g4988iDvarNAX@TF>V2rPVKMBfn
zV71-LF`hPzuo&8R=jY0DI~Z|Vaov)mr??vV%f6>Dl0!9XUBP|_Q<FuCHW-Dhr`bzp
zG@@)(B*4D6eK-50_Oi4bC}=3;a|a(^qE($$;?CRm)sr<o%`KFDUpiu3RYaXfvrDM@
z{w7tf7#SONTJcqs0mq)9<07BUGAOpAv8QcA!(%hEm;iwUBh>hFI4xqYwlAecM05<J
zue4Ed`vbN!=k?laufN&(oh)p{^pNdpdl;2>v?OS!`2mGV3E#w@q6qJ<j)z|(*!=--
zFlF+@5iRpi+MY&fL=ufnzGKz_)Xt;bCU_&nd?G}51bN%r&9+!~csl8^>>yxbuJWFA
zUds?J)*WRJkZm8geO<?tFr_TLV-75r5sZExvm8q@HvCA;u>JK;Z3Hk<)CoCv2yt~V
zNu$uC;ynmeu>);@qNMIQr=3zF#r9ZA2ist7x@(=&sQYzPM3`pn$#y*%I207=nrz;f
zl%J};#16VN0m>UD9P;!<A-q|{=ujJ(3Ta+Em{}G*$!cU*I#*?1V8F78pCbn7aJ?-l
zBlX!3rM+RKwbU6A+ym3LBlR4`mTLy_1vxNL(D<0L<Blc)Z&6w5fmdr`2O`cEJC?Zh
zs-rRPw37Toi{B4tJ8j3?_>FOMHD<@j)VC82a}NdRXymBR5~nn2C+kTvsqKUyR4N8o
zN@EjqI!?w;b)-kOJEW2MlKa^?^btoKd@+zbMkR%3$$d|^CAn%FDG#*bI?k0;&Iz$&
zlJ4bRJ>A|&Iu+#>oVNI}%yuXkXX?Hbfi%Gq4vt|6XqdKZO0S6^H#^$~8zrt%p&eNE
zlLanY=#g;C!iMdcw$b4)qs*L2Lj#J7)4G*0ttrf&Yma7=Rot%Z%fRAl`AwuW>69V{
zhwb@%VL}gqqEwVdob-@`X<R3Ef<=BOfh=qt@S%L?g`<^t6zPmhimyS*U8%fSCkQ#L
z!+p*jInP_iXR+Qq+h{=L#AyWi<d&El>vILuDyQ5N+vqU|me1Iz*nYcMbdX9?u=rv$
z(Mbm%kJU$NKgN1sKT!8bh;7+!l76%uC~h5m-J?Yz@%F6!U=10D76u^7I%x^9PlHLx
z5{{~ZKRugo&qwa53@P5+6!()3J|;$Tk+zdlAzeWycB#=Er}oVvf6xwQ7xAcEscOoK
z991P$DOn52_-Fd#qm(6#WlPu^i5HjKE5Zscx4Se0oQn=7?Hu=Ab^m?#Lv>wc>`S$e
z*Bo>(5~^0MT4m2SO2l@evV{F<2b4yJq0v5~87BbZasT0FYFVXvLemch(GDDNe1tw!
z!q8FeTW~y)`x)|Hek42oK!29HY&2?Yq*_R2RrC(0WZI8*#E^%EJcs`cSjs*tRSp-1
z{bTucJo(uCL8ny{d6c&y4`E~Z@kVr;x6L_?xPak-IzsujpJ>mN?G+n}1Bi>1T*km7
z(w?xNEH0gv7%l%aUTFV{bI2m(m#gAViP17zsYK3C)o94boMNt|lsq<XH%3nI;=br&
zaLJ!;#J+iUnnJ!bV<a<hKyS*(cR87fNVlKqW@UJiWBe#R!!bSpYeeN*Taw6cKijAa
zrXlly0>b4UtMF_GU)@jLn-Br}x$GXtBr;hvBu_r}*sJchpYQsn6g-itEL2(YR%x8z
z<6&{q7Z~<(`-QFp%Ajl3?XH}VgAWlH(;u73FShOwuC0Ww-Sej%ST2wV+~}mS;9c@S
z|5EqX6EfK&6P|To#p5Kk$RvRMt$5%CNg!}S6__Mo&VIR`F9~HAu*eP!=?#X%UJ1Sy
zKS$swq{n`xz47wnD6K(K4op1jnh^F5+pjhfoFcA)MvO}vE5gg|*V-DLBTw0B<pi?V
zOA6vxW&NGsZNJ`)h}2z~#r8O`7!fD}nKyQMrZ$2|krGbZZxmh_xsfIL?z@}?IdukG
z!%Nl_b!p|~^3VxVPup*Hy;0ev)EijSRQdpuUqTR;g5GuZ({JUcqWz8!ISrA^s&R5>
ze!Ct6Ik2tB<*29#ofQct9wV(tF%)n~`<=R46RhGF5zQ)#jf2VUfU}7acz?GjW(PC9
z&uPU>=iU{v5U*OSbP>BuzSe%PiN;hLROuM}-w4b=Yw<+Z0?yyF%&GFdf4_}0U6(YO
zl&+&V=_K0%c=68_Z(e18ke!Odv`3v*E`%I(iM#fPjUsMXeuLA<Udyi#%Thr?DZpgO
z%M3G6h5tXw&zIp-RV``8<f9hgkJXA6+mTG6P9L#9&KX$t*emaMekbol<RbfO&i<qg
zrRup8`&#|fm@5j9Gj{DyTi+Lyrz>INkOPY$(he-|FL4q4Wog-;)zvB9XTgvL)AAC_
z<ZLec^VSp2;Ug$--~c4-UHTB-mk;51`-}Ej44IA>O{0YPguKBg9DK_P!0$D!1Lp?z
zmpSVF0mpO=oK}pw5+lz<V&v2ISH;A&%SioG&BffObHCMs+T~o83e^;2^w*6Zf&`^w
zs^XhcPr<|sWJn}jU7NK1O?P$b*b3YT2J2vAqS~r&R#<dA1sK@hwk?-T-8gn^;DF)<
z>2upRkHOgfu1(o+A2*HDOd**-3u6B5@9P$=-}J1Wb6}x)A?l?5%p%iwAgF}5HUg*Z
zAF6-i64E|*VW`><M^ldHLHKUN!#DB#q(-{LrfC1T_~v-a1tJQ7sdsCr`4;|ye`;d@
zzfkU$Lb8@T!@*Zi70#(X%IL3T-lPC&Gbiky+c)3QA5hW5X~!_-l}hkGWdG9jM#8w3
z#c2nY)4GaevVZL+N}NSQgAqgP(;FtG`C$6ssp4z@R`+O#Byv;GflhL0k%Q6Fe>qi1
z{=MET`Bx>RbutsM+_M;O>Ga~TEK(M%ve5S1f3&$LHb<dyupvP4QIJbGzoXi1y`L>c
zj>|A&rOf-Ed3BtbqN3&Ul>S#OsrO)>Q7Q}VqNWS<UMu-ovtwl);EW9vyiz)VYT8~&
zHDdkGKI~X<ab%00^E(IeimR`%4?kASC;S=VCSo~>F9!2*@WCjw^reZ&4jozd_7TTY
zwd%8aANFeDpTq!eA9*Z0%1#rXi5B7wmcB?$zy<rL>?Wz#I<a8VX(ub>Z3?fGfG6J1
z{Ihm1+DF%O<>!#aN>&MAb&sMUwGvgTzf1VCk2zN6$eE#s`;(m2G~$3_9A#$HG|Jug
zvByFvSUon#LyV^#NY3A)mT4bX!$I71Sb;fJCb-YD4knlJhBbF>ShvPLzFk>S*($G;
zaRgwJhg0#NFwsf-gkurJ_7&@$Un%Er<`i4|#C+rwkzL{2r5zp8*vZ_i!Zx2&uhB=y
z&lzE{(g+-kHf{pC#_f}9jAZ{Py)I_CWha5J?V!mYdOpTZ`;_dg3{OKiYeyYG4mU4J
z)ifQ|I)RM5IQ!INv5wYmT<!c6p>3Ll>g>~w#hy<HO<(>Inl)=GoEKY<=8l;TQ-FPX
zP6vT))6<jvHPh1$L{A&<%|6nz&u9%oj+{pA_Y#&<!j520d(q;E@!dha4y>+DW#t66
znDB?$XXc14>{$JP^E;OsTaOIjD$n~_iNEgJS6}P=P7&sfNY$hE*=<eTc>T3bqixb8
zy`x)xPEM;TE%qjy){+sdN*jv>R_$}U$$Bb9M-;I1qibRA_J$WrR18Q4;ACkn4xzKp
z%h${L6G4(*00M<@+-sNp{Or$dxT;75p(R3z!wbLQ1%o4>97{^E?G7e|=GBD32Y33y
z?mHIK1Jg$$3;J#cA490bHhFr^%%8O{YWJ`}i5f}cAw(PXwCDFbcyZy0!KA&ox{};T
zG%qKi>}X)}!EgaF(PNO^7avQz2G?DFg8ACR!p2UAKBGV#DWzE`>nUCp0qxEmu~X^!
zt%MhJSbL3>F2aHBCC5ri+1MLU_71TNDGOlCU><}K_6HscQ{u%E@UAH8#MrUNX+&?~
zt$McH;Pxfm7chzWCI^;l5WQ4R)y`RDR~Fsfe8aWQ@3`t9GL$%^WPlLg>`UvJi~zE&
zxc5;9R?mzbIP40<J)lv%H0gJg<Wd=_XhtWcqUcNz7LvWJ1}Ful2#R1Z^&l-*>@E_Q
zoU24VYj0a~r(M-Pi`f-lJwx>tV4BfQtM9O7g;xD#g$H47^aj(epp?#C5XsB)mfp50
z=VvaZ;b}PcXwP0z&)CJ>K8b3k0*}DNddZiZ>a%~^E3-?mnMvn&eVBq%WVt<Ruc{9V
zI*5^O1xLbc8FO^TxhnYtl&7*h;+!e^UR}@B7_0mwMU5R;vi~96l$aW?IrjY3^yz;n
zfB3(EoV$uEs-l02FY0soFzwog4~MFR9VM^qSN|yM`%VX6w02NZD;}RCc3t*tdk=4}
z>c!BG?`kVW!c*{Hdv!g%<TV_8Djmkk;jdC2!N=3J7BM%VH0-nM8y}qLJ%EXcW_AD~
zlT{ZJi!HCmb_+Q)6V<(o*E;x|B9^G)*RYq_4edkY6D8!z`?hFg0H60nx`Vp%wB6Wf
zB|Pv%2LZ_4^tzjGxxsE~x4_Y{b-+80I<Odv5DxW4@Q&aCc5~|hR4R4s0^CQoBzfKB
z@&=#ma4Vs;VhVOkcSm>{BJeKgC1msq-tZZIkUo%{!C>L`n(7N@C+CdFPQ>T$aWFMN
z7a8}+R<qaE)j3zHZ6HG^CExso7=}(fkmo#SuWNIK5go;YrdIC4j*So5>+2QjmRO%E
zPB)k~M_Esm+x2D1+2$fTheoXsWD&+F{)QLv%e%3VzbE}*OZ0Z{@ZpeF-YIq`<|UX0
zTi6@A`-5vsZGzn_rtyHUc{_%`uSIg5K)G5`cZx=jGMBMygpL&R>fm$UwqR?cBzn(Q
zHVE+##Nm?E(m__fdDGfE*KFEktJ*c>1oa758g(GGiT}6LxEVa+3vG3yJ<h4=Ijy$l
z!d9gu(kz^_<^>p0SVq_$a4<3JRpTnkpWK?mPYm+RHm4C6h-qmPkpJPfP7b;8TBlJL
zkXfE%VYfGFcxp%vYt>%{6rZ9eD7`E!2QrjFP>1IC*xE)mn<y{8-D&2kT0b<`w>NgZ
zJXr?uV_q<H0g0|MJrC@T_Fn2sLww3c19T<y*GVvuNX$ycb1A1U)0$`X$yvItUUPt*
zN5(#mhHzjiKmMi_wm!Q=-#I1FX{Bg3I!m&v-Pt%Z6`DAZ!)eE;#16^Jl@54U%|fR1
zg=DRV5#|aet`;kkH-plUwxR1xmKcYFIxq!2MBU0Qd^IxRo3dB^>8Wk#9JFGTVKQuV
z;$?E`9gW@e0%p~X)rHJ2%4M6j3@}A%&#^Y?fHVnlJf3n9lHb($-ktdL`uOYsNK)p!
zx#6B|PSRytnjtNT!o3be>mDigT6=R|D(!2=Y2<p=_Rk=n!~U%mbja>5Oc`3(byY7#
zaj9UUa~WDCekbQl7qnvcba)+wP?k&7ZiYad$-$>{d!j!)M%DEcEx#h?Z0NzE$syG+
ztM<Z<5iIOeb9Qh0<N+eq$??7e&^K)w8lSPZ<Ts6}9;DL{0@wG9-B<IbUZOpQ{c{nf
zP<e?u=j{Hrh&Hf^&<Fz1K*+GS3oiK9f@QTU1sy!)z-pFcmkOf*S7)x6(qG=z4Q^Bl
zDpiLUdc(-<LEI0Ki}Ic*oAK>+kNo+nqkdGz-^e-#BO32l*4<aug*glIT&TF3(lfz`
z7PgFylV&#5h7AJk6EW-n+Q+f;A@MGpvf-|y(vSAh9pr9B;5eALI^#1GnYPnL^7&e6
z(@sOZioz8vfYGj3|IJn5VAQLH9XBk$*?Jv`NEpWK%R;Lu3K9dKuWgx87!J17JyI51
zeAN<Xrh_k1R^C?{YC`_7t=*lHWLw<?`JgrayBvI)CD$Mu2Q{KUZ)+^&64~$ToMvr5
zSz}eKAxuIxmV|pJDIZZwo_0`$s7J1d&~y;X+R@|>q^P@U#cHpg>sD;G?Tuw|>IwPv
zPP<sH$auWo4(+flE>-!#=sIBg=XcvgH*Oq!1;?P$g#(M8E51&rawumvjhB<%h-o(w
z3#?Lz4kpGwW7WhK&se|Sbb#49=d|JqF-aU_n9UB`RC_yLn}&&f-obdftOl?Ndq>h_
zobbZQQxgs#KlaigjO^1jfFbp8OLsyCzgZLx-u20kUw-`m&wn=P23F~*`ZQ&-p?^A<
z7#t0$!p3y&&P*dR>?#V~pH<;NfOTODSwCU3i?hPMr(hgq$}80gOb*iqw0wK85$NsO
zUY%x=Ka&;JvSg)=sH2BEOy@7FoeoSR)sPnCXUhXf6qB~w9_|)Sra&rUStba1=&;*(
z!zWove~oe{%Ege0!shD!edu^{IYe}T!aiyz051*d(tGgNzWeeM_DDAz=cQOnQgmRU
zvlj;n>vE{vjCJRYv_~81vA3(}wQL5!a=32qP2FZ+mE*r+oOmI8<7mKedVnF1G{d+Y
zIWKVG;3S|XDjmOIkLBZ5QF))Ly|kp`Zl|c-S9jy9k<VVKV9B!%CZ@N14WbSEn&g8h
zwgivS$W;`<SYehhug7V)<ZC1IhQIQ-fvKy{DSBEq7JFwK9;RBE&{`~ckAzZwYh^lD
z#hr4)$cV~jVUHI~)+nwE10paD+2-3fDjoNUnm$Nd2D8YhToSb84Icp?47y_N2``V}
zr?af{ukEJ)DM&*6P)eB2Ne>shgD=hmv1u#ft3m{`uWO5CFh^lj<+U(a0MjtAYP=`?
z%f7zu4=W^ZJ|Dahr;~#T!fpZ|1S-N-z#Aj*jD15ll21v^kc-8E#Xb3)#=T|73Z>k>
zu^X?GNMq4k2Uf0+5O1U=<`-~}_^dRjllDz1kgTG#h4Wi<vv!mq;{;O;dAsvFcgM|}
zR@k@Hn{ruck2>THrn!(g3pb;^tFBWNb!Mg?bzl;H7Ixg*w{LBG3h6Xs(><qM%Uxl9
zd?Cv{%$$6UqEz19<^!ICNXL9>SJ2pisYSS0!!p%SM2-kE7rS^*^+!BQ`6mi)0#^Gb
z0)>PM1qel4|9ji=<P5$Ur<D@?s=MyIbB(<(k(YMNvw=FT=(vK))n&Mgx+gI}zwP~n
zm(MP=p3BndGamMt`9luAI<Y_e@5hL;$*hlW%W)@pA8tS+E=xwH6``na#h872o1T>%
zl|w^AjL0))A83f>%U|s@vYXAi3B5P<f>E^Z$nja&5pm{9$C>kjAt#(d`^t#64|Y_Q
zm7B8+StnrU_&Hap8>_}St>Qy%y=<ib@Z{)2D)8U{<8$^dU?SFZu*`Glzq2N4Kjjld
zuqwm@EQO#4sboP=Wz<>uuJ%Q2WYS<{Xw{CdQ`M=w=io)QGn`R0Y2V!@bX-R)&?;h+
z$Z~);iM0~hlZ0pAQ+-IbBF44Tfklt^BS2`U$t-6iN&DXJY8@UM;pn~uCVIe?{!aV8
zx-OX?3wL3!1JjawsJG3&zk5tXGr)CEdV}HgL)K=?#xt~|T?hA+a-(#Z28)mwU?=AH
z+s?X9I)cu$C1BWrX;Z@npj3iU+tpU#7S!+g{Z2c1G43}-9I{$$cQZDmzkux=NYRoU
zjw$5J`zM+87;oNF_z17j$4d#r*vl-**z4f+7Tjz)fe_;lrD68AHMXfo@PyM8v!75g
zwmI9^K3UP*IY-`LNyK@E-~RTH!M2Teu5g;c8IoU$rzn_a@hIISxC7k+l#H2@ev%!S
z>?doYKQjk2+-e8At|*sG9O@Yd7US_kca~~pMcod2vhdQi7gIsIvs@i~@fV7cZnvlE
z6=GG2lZ2!lSSB$jkur}rIFxgD>3xVJR1b05X-XR4;6mD;v%?){>>R_npy+oOOyFD<
za)!QH0e%Y6l;uLqOkw|ucU15QFgubxg%57tNHjA|nLK75u%o%)`?FY=_#H$Pg9u5e
zQ0}qDi>7CLP9rW57f!3g89Uzbx3jTex;QcilUKvJ<Z?mBu@g0-l1dbQ6;_xC`qnI;
zbMSGGVuIvsIgBECvS3;XnUM6f1Jm*w+BRmVmQ0d{LFqzi@s7Ag2Op0_NQm{XngCC?
z*Wg8wQ%q+|0t;eb-eUXCmnqFKB5v(;-6O$7si^$*f-UTr>~Szj99?W@%AB;x`IY<c
z@YsZ%?HFe|7mrH34+#uRij<qj2x5Gu-L!hu3a3$fjN-8*48<0U#CNV6_7KSUYQ<~8
z)E$h52HT#UZ(WXjlL)Wy1}o`Qc5%DVC~&m@5H;jSdk!QABke<QEJ>_Kn8(PjvS%AU
zLo&i4B!N`s0=nwp#jl1&M+s&f4D7k~Ay^%7pM)VbR<VOg>c5*)EaYdji*;>QAyo1l
zA&z>FgNdmWCB&%~#4Y?l!EoWAy2W1ymV6F6Z7U-~;^V9B2iy8eXI_|vGykEaQf(5>
zRa)Ic!xW;S0>*rf>#CKjoS#}g>u$G837i-sT(>%s!fb?E+Cst7W~~Le(;G}C2bE<+
zX0XeRopaFq5W`Qq`%ndoQ&!A=sFPGlyQM&>14z1|7TTmeU!%4p>qtCt;jDv+(a;KO
zcG)Glj`qXd%XF4;*i0Q*@D{fD7(s;whx0~!i2YhiMhl0zwjXJoCwXOJXv%5ljAjLD
z_}P!v)6};Jm4!!D+33(MG&c&PqX?&SZIQghF4-sLDiJICu{Ho0^T;I}IDldn$HEjU
zBlC~v@)_2d22MlL%=R&u^H0>pIXz<0`yE)kORcDOT6}7s{bY_U@4qmAz-h&cBF0}*
z$~>XG`Rrwnz5agZci@q&c}}<??5Eo(tnZJGZ6RCDfwXy$@xF|s@^9rh{!H~oLX||T
z<GK4n_lp?SEYnUW$y_nxKU?=8g0e^I7zrhab4Zngsac@>Q^;LEm$Z7zb<3TfxpDP|
zwsM%z&o@e1k8t6%;%g-QD9(?SXTQ*3taQ0(I8EMxrK@t3jSZRP7wZ~~v!O1lav-o$
zuM{6o>h&*Wx5B<1l{g$oK4wA!=A`{{H%t|^BQlWPkaeJf5#XbHvNC3L+FARRHgsZV
z44ou;30)T*OtOMtPRXKV*Ua?BllH4QC~M!c#`!6B-CH-drDwmE(|hS%xTg8NWa7de
z$~zX1Df#t!ds`G;L1_n*pFb@Z{>+U1Mt5mc0>jwAZq2L_dYdo2?v`t>x8H25FH(at
zA|R}aS_!cs#V)U+C(gAYE47f{#grjYh)f5VE$p}23=gO%=d^OjSM`~>Y5VOQ(uEx>
z$g{*&$W%G76q^yRDLN>tA3arS-|uvHZ(O@+!$$Ag>-*Cxocy~DKc%$I38$?MY1w^m
zu-|K&+lrN&oJOtm&_!fMhq9i-jq>}A+8AUR({-jLFmXFp-+6YdKWMZTep_Up%a415
zL9sFhK`hVOAJ#i!wTO4c4#vuIFnZqD778oP*&nq&V-ryASK3Kr<S2_X?ImpIkMoY9
z;W^4pIIR@ysH?3&7Q%D>Bp>pU`p;>%59GC3pjz>wd+bjayD$45cCTFyEUr%FS*%2r
z^ihQE&)VdT_6Yl-Rm?_~uP_<dpBJpL4&v1vSay&?a{Y&b1N=o1GJGufdDU$S{Y(36
z!;8wO=_%8a#tbuCIvM|R$wH77`V2IYnoEiv;G=7j?!=7{lI*YAXF)r%u%E%sAzRUY
z2NTOm5-MZNb}0MPqOb0+^Ru_k5Ys#1w35z4;>ej)-%=owTq6J8-{j7`YRvgn!xXOD
z@NoTYK9NB?2Sw3}0oSv%yU8^V*TuH-cX{0y2Xt^&C#{^AU@o$sJkHoi_x}4_vtt7S
z7x^9U({pg4OJcHrs5?1qQC|%vsvKN?!oi68&`zYA{A1m#)DE5EhkeZ(Y#+Q9wUseC
zX#Z5#=}2^?V<|2MSYuA??lQ4T-(0lE|8rwO!~MAt{L&6UY;I^ULg}b~Y2PM}&Ri;a
zR<6v04kjIH<o(P}lhPQmTL0R5s`EVP0kNnKOq&Ugph*Rr{#$J`*I)1aEPB9_qa6!P
z<lk!^m3ooY9<oxbqW!b38rr9GReTj%gGM{7gy7LygLdsdI()_yBp*GAtx&Xg!oe4{
zNSG%)BKT|u>UA~``_KA%c`+0*!a+PGX@b%GEbLf6JJ@dj)m<lzs-4cEgsO&-=1!F-
zszza!kZfUg{P`F01T;Jy!ncS2uhZ`0BwPolr8(G!tTz}Rnu7&@*ztPd5W2O~lHj6a
zA?(AC#~_knI;LY$`2Pb3mdp^t=Xv{x<F%WUK8)~4Grm0~XUP`IZB#azGyr8dYDfRb
z<1xRg0yr@|wl%}D9ca+Hj5uKxP*{YdGyAB*h{edla4slCTmwwZvOHH>+sKOc(XC6H
z?^?a$UZ+{SO~-(1|JBB%Q_w%=ct^f48&zbBW3<@{^1M*+wX7)B3?axo{ju$75wss<
zm6PINVq8e&#Ps-a_3Vr}&g)=FbaEex!5IgiBu)lJp93uLuOsJweBIUbK-r6=h)N{m
zuY=D$J<i4G&)6p%j|DZNgPXh>3p-X3y!DA)=fj4Az3YSni=kj%VEvK8z^7t^u}{k9
z$eDSg)2Kn^Fh%++LHm>I))4DiPUJZ80HUiLOf8qiMWuxhg{~1+Kcz?_raqR4#ySHH
zwTr+4VM<(;L_acw1oWwG3OODT-yzNFcb<Zz16yNPwNFcGCQO4$*BZy31JaXd>TKvw
zZ{v)9H#9XMS_TuA0~L+0WIwgfI35bsJNnMAn8Cf6)OXv9+5k%MVEh&T-C$zdIz4MA
z#Xvr@^?Wr2aGh3+9nKDzhwLy1?6Z!SIE)!`Y(j1tAUVP;dxel<0=Lg@-~5i*{>+fm
zR2Q#Xe$#dKIUS+qkun!TF6b=R-th4{CQQ>t2M8?94t#Dmd9v1M=TgkA)(@C2fn`AT
zt=R!8)8}RHNVcA)7(JQ#q_Gr?hwqNX$zS{Y_Fecqw`enQAW5J1Y#W=I?9JI19Dn{s
zzJj6JA3+rd<kMM9Ibb*J>XT-VlBgJxL%5r`Te|Mr?e>K^8XG7X>ikYV5xa5b_=wA%
zurF#;mNo&n#aRcYEvNqwpJ6ZVF3{kGtnR?lQo=b;_Ql=A$SkYq&{Ga9-Y4FQO;S$0
zywzS(7(GbU=%fH>7d+_=BkP?MZIvbn)?-(+-Y-d$%eep(L?PuuJHnC6PuQ0<q7a)H
zahkQYC}bc@1WIg;(Bqg|ZF5()k3hIrROKF2vVtw_K**#HxB{B&OLMj&q-cC-a?}CT
z!dH6H!G%-vN9?7ozhRmRzY5=j9m&Dul3j!M*Isr!<oZpMtTL9KS9@W{BYk*$&#r33
zvYxCj^7bbkh-CAz{)8>-1~TF`7F|u=Fjey-Y`J3dl&-DLM(P7}sa&WAOLCLH{hYm`
z%`c1HX~bNHJ*wDH#{9ps9=9LKQ*|exR|i(CO{H6PIDtamWTVyJo%X86ed-`Ff=2-b
z581)=z0<C4+~=-oYSOaZNs#dlO16}oCX|pP<@TC-nH(nyk&2j=@H1x!75LPf*N@!6
z)_iR?XDPX^f=4t{t=c-Is?By?eW(_D7+&H7bW8-8RL)mYU43|n;&rd?MpA|HC?$ir
z9hlh9twR)izP^nhNB267ppleYq@6r#%L}I6Ix7^P^9B>El&1xojonb6*z`q7$KcTc
zV(mEi>>4I*hKkR)F}tyB*|Mvh-&&~4mc9JtSJ_R8*tGes4bJb{9%LBGy*3z&-Q1?T
zQ|79MfP;z8Xw$8L-CJ7s5QVk6F5p#@E)*wj!7RbY>eHA?$Cq7HpC|XP%#SJDye28&
z>bKp0-+gazS|airmS1PDO>TDEy<WfGyxm^c*3c@6$FxeG&Rj-qvf3PE238%r*XKvy
zbi)nHo!{}%(g~Der*2>#K=k;s!h|%MJ3w)JFe1)X81b;$eR&Q6;zGa4*bXeoXdV1u
zXx84)h!WnX{0Ehy3bjvs4q^hI*WbBf)2bC4ZAIGvMmd&?C~^lNI^q?Rwz4A=ncihy
zJMX~iGt#uTSi^f&UEQ3Q7y}(xE#EZvhM6uz%GLR4<c4g;SwO26uI!vae3~y=bK>eX
z-R!HHnKsF%9aua&l}ZhsN>?c_<JN9kmeevXg^VgN4nAp#m8%GAvTf|PHZVvXc>A4p
zvDS+rg(?^4_PplSvGLyKIeZ|rt7zE><qaqLqrol)wzhGR-V`$S#8{N{ccAf&D6cX|
zlYFB}=ZnXBy|LbyoKu_6aR*kTt>{6~vAoOLGwx{Ig^J5aA2j1a1-d9pS~)bam)W|;
z3~pPs%4x(`g;7pVqHNgux>ezWVwyWy<n#+O6PUajZ-`7QO2NIeo2iI>LCRvgRGwVy
zTl*b+`LKG#O~NYY?5=u@mbmOfJlq=$>pnbMe~;US_DOe8%y`bb=1NM8KWuO6x~UWa
z0<&2~)ZN8@0H!`l4*AG;vyBawmQfk`yun&&Z3INyRM*L9m_-&2IItMYAiav#6HDJ?
zo7<PL6eQ?lrg31&<~IznP1~EhCTEU~JI&Z$4zfk=vb&R`+;y+>D}|U%w{En1+6xhi
zsDFjJ93eFG7aUC7E;aw1_7&AqE$>54OD|SLHA0`=Tg;9%uaSwBpJf*D4Z|2E!XI<4
zDk40tK+D>=Q>Z0w)ZS7O+?4aH7J!(v^uyhDUv)HV*|5+hGAruE!Q?1hMYQ66yFYun
zu;W%09HAADl3O-%88C;uwR;pRHYbgjXcGsMya4C8v@gA0LZjS*%07Hs!-s4kqQW1q
z?LD93;A<&RqJ=y2C`?Y<vA1{c)W?ej%ZQ|eA~4az12AN1j;sRv%I-Z19HDQqSH<oj
zwbj9km7=ikma=X`$vH<bfH=RS4>D6|@Fa+Y3*CnEs>habj4{7O1S`flyU<3CmluT>
zDVP9zEnQSV8)m@RQKSWE5uSIhjqWmBl5k-Q`)o8}3p>bK!c-7N34moc77j8~VHe_Q
zxc{uzuBI9_hN}sJv>lj;ZT-+Rc{{cxJ9_DZ^ZyrX?*U!cb=-STdM}RS6g$nf99xPl
zOR^f>5;jSc#bT0bt^t8d03r&F3y`90Cu7M@aS~Zhbvk<&8(735GJ(J-z4zXG?>+DL
zpP79CCF_0PdT%{x5$l}U=bp36?CCR`Qnl@SA;CkqsKPf~=ocl*zTCM@#0t|aLtxI*
zK$jGJq`K1FCl?irDMOXh5`&RBO-4VE)kf=W43{+$PnVw-`zC)OZBBb|_C+Q@AtLLI
zagi_QARnr?Wc9_QtBceur#;^iuI1-`E5af9)<xUWHdbW2HpVTEm>3jP*?Fq0_q}b{
zcZX~&XLE84OTmP57t@nJImoMIXOSlfLO*xX#%p?#=4kgS5R!N_rzNH*e;NDQCc3v}
zaXd{77y|{1_#LM`xje@Vj!ka2$wqDNA02gWlDsW4|Ao2HwzicB1798i-(WI$VOyZH
zr|n(sx!BJN(cSOBx&ek!5h0PlJ_eXZxT)?pY53u<1Cy+^eEml2H~PVm3NvHQtp<(u
zgXqICSJc3_={6{xxJH&;-LHI@<{>70<~cnOPZBdgX$B`)EpP2<luFJk#-j{C457BI
ztuv-688(H%p*GZYM0z<-sn48=Yug)2D>^<<PT~rMH@vlI&6KOHb_fd`<l#053F^He
z=RWzsb=SVmc2o}}%e3rJIZ`WcCr9HvRr^JO{tD7OX^&J_Kp6$=AnPPqc*<$XeYp@g
z!F-h54Z(kRy-h}Fh0Lj1_X6yFf`vKxJxLA*9>(Fpb)jUGCB?#U+}_(whKzk8Q-oq_
z_MMiZ4C!LpjU<Di8X~71yU)I=?VM#C3#3yG0HmpHRJ8}GOE9Wf%dc*_#xiQ8ZXU(3
z2}GjmweNk$yUp7$GD;)}O9x{*I6f_-RjV^u%qLSrM5M(FWpR67F7FMu-(-)rd1eK|
z+>Mq6Osl{0&O7a~HY|iA?LHNALeb6GQ`iIT$q_lxX?oJ$-{hhY=CRcdJ0Q_pBBUk<
zC+!36`4u}Rh)Cx)5rdQf`<ndh*0E{lGT}Ebxz)aQVbnqTU{nq)raNthY2(o5-5-l`
z&c3dG6X&xq>OI6rHkg73#A7ba?X|D(p3AZ|r(l!7l5`+|$UC8E(@1Ht4|YAd64n;-
zQ5=}o2TSLGeM7ZJVGHI7TIImXIHFC8fD=kudXrSSKuC1fzOlZkugNl#!T35hkCq}i
z7Yms>HC!%W`=&*|RJf*SZkh#YN&2_}F9%U1k5^ArUsmitvYmo-v{`~^I8W79k~>#>
zThWzSaFP<6_Msf_at>+nx*i)C+F?(0v!-!s!;&fu=RV=BMekB+sXb8{<Y}1Bhubpu
zd!ydntnE+RH@B9XZ@<yIn_FIUyM0T;AT$wtf7}Yaa5(Ao!EdcjD1#;?_K2}V8=@t4
zbBe%n<XyCH>y}F;<if(?4Tg+_N1QHW?)vtV&t}d1%^d!MStL`b9B|B3s8-DUW$c@a
zi)DRB`=nwIArn||99Wsk`#VS)9<-0tL;AW5IOnjZ1Jjr`hMm*CvylXxs@uu78g?M@
zVU%Ae*Z=ry-*xiY*YkTi^K|GfC!v;!###bNB$;;|`M0PJN=M-onOJOvZi}Ar4xzOo
zknFn~B5fq1i!yQCr$MCmFhyp_im>mg9xbNn66=~PzqL0<5o9XJ2&gQIY~S0yuz7%u
z)N}3=36Xf^Cd;^oqtMjUW4|wZH@|aci*p?-i;fl};jevvd*22O22%sheG;K{>(}04
zJ3CZO`Ax@>L<vKurB<5AjD@7RT^(AbNUc(J5xwWM)NVN#vZo5rN8)2nqwP*?-9W1A
zs<msJ+afL6B$BqH7$!!jJvpMuDFO|+PU=T=u<h-BqWl3VuA;!u2d5<;yPpD}``QS{
zDNEwq#he5?=0|W^+h1Kx=W+_&>;_1=;;t?`?YTpgiOKNJ&F!%R1rzotxasKJZf`I}
zc54jqVmsKrhd)TqA<BD$$*8FioU@p14>kV4aaf$}CC440n!!aH9q1_Ka4%tfhjWxm
z$3}-IoEu>!>9Skx$fEHmcuJL)Gap*xIp(yd-9jf_Y^NkEm^F+a;b={0IZ{IiRmkrt
zZ!MAq4#aLyon~bfJ=wM`LfOO&+`9w3g8g>{GJLFNKywY&hg}0$thR`uWEf&9nQP=l
zJKoJ}B%jDt>cHx0#W_g2DGRBc$gflXNJSWWLk=Jw-A62jwntZX%TCsdaT~Jf<19S~
zRx`Yiu_r!Jq|r`wZ_`<QIzUC1QN=9M62F9ONPSK9lVz936baB~+nYfTxi2O3ELKu!
zkABJw6wwoD86k)5bR!|*0mdWjK(s{%M+n{P*_rw!GCw3NOkqqa(E?6O^hrW%M&K#j
z{wVHI>G>z@Y&W%1e6_F#9hlx7_S~oKTs_gathio-g>uf92&=NE>N$-|NnY}91Is=4
z-fOS7^C_^r;@X!umlaz!HPc5G8nz2<RwD`-5gg)_B04Qeu%ga?CO%*6z7;2ahky!9
zy&mI4ma)EF>fZOXRyJll0{xv9@st&45L2R__H<G|{9nqCrj~3zuoZ7g*zW9dNdk(X
zlDe}CA}euON@yY^1y!7<3L&X<&r_$uZ3FADD@oN<jf+Z#&TYEuaGm3GpcD>@$`9mL
zy>}Jq&s=NPaH|tq_k-O~g@Q>&ffYi__bN|HS<G#&o|b-vSxs&(2PP^oxs5=CXYv|b
z+JJMNR=KS@Mu(ve+I;u1BM}#)=+6d|iX@-e@%c-3wLORaA_9hI#evoMMf`X=)bl;C
zs~^gzbwat0?^BkP1FKzzk1&Ww*vCV5ge%yyO>S9+uea~qCWTJ=BVK7gT(ER!VaO)v
z1Xv>6-OTBav>&nKb-3DPBKCoWd?}q-{GoGk2kByh*(Yc0N9&hzBFo5$an*t4bVYS;
zNf!2F3mabk12#Dp2~bxFWeN|aUA^W2i=l1J8S>-pHKE2tU3EY)bD3Hyt6IgZ*-unU
zDrbvgFub*h*2pIG6eA=g0g3Y`>o=KEtQO%ZB?6t6m_&9m+&7eW`>8fz>qjP?n_Lul
zkSxm*O8GQ8Z&=jqr`v==K;(IqKo2apGGiRoAq)01O`7O)_ApVl+-qYF$d%BypKS{e
zTqAi$TA6b*sjCN_r;3#7h=>S?r9TUmh<>gHDbArJz4KR<0whU{miVQVIVE)|El5KR
z(>GG=&$sV`&SIK6s!U(jQzDVMay%7FlJ)2#c?D9fs`S6mEs5M@hr&e69DyLTXrB#J
z!oYs9&5C5;sF?<&FgUG596DmZRDI3bqG{60>%eKJy%sF(p+qU$FQ0t2PQ>IO=uwf7
z(XAlkz|;<>DK!|Zb<(eXCHIhmDmIM{IJeaUNf^pW*k`}m<`Ehq10xbOe~B>k*20De
zH?q<)VutP48h1yypP+)X1CRoI@2chY>p2H~${#q_i5b*0;S5MH<2R}YvZ{$0(GPF1
z^nip9!hW;;k_Z`yMQ_M~q~c7;j+r6*t=4UHGY~PXqamU>Eip)W@NxM;t@hb(w`rU^
zUA+3JH(11CCxY4Uw1L42CRCshz%>kvxeE6taz_%o45w4QEZMGeR(>}pe|%tgf^)wH
zo%>W>WQo$?f3NzSmY+B}lK#mD-&!;+a#>36i_zHccRkP8!o%~s9ax~IkTNP8u|2a3
zf6%pbF_7`L6HZH##ibKl><`-xfl_r9B*V8VH|0}Kd$4~M-+9XZsGCrZ)q)yG&;eEi
zYj2SH#2wWv{o{ICh6bHncLx3gEN5a^q9s+|{-o<?9&nlQEZX+81a&^dRu^N_!N2KO
zkXKqJMAgpx(;9#TSIt-00I*W%Sq_14+M|2Q@=%w>IEjOxeD<F;*1G|J-o!9b^>`ux
zs(*KpafZ}9GHrj}kZt44mfjBMHt|^gydL36_7~kKwU;x33Rrx!F_shXf(&K*%eIKf
zsF*ZlSaM)FgJ}jNoK?mq>O<RK)zc|j?T4n#dV}F@0D*-dNA0h3NvQ>gR=|x&&%h|h
zJ^oE2aF8bD`eIO{i~lWp&v{DwM8?#K3B_g-e*CxfX?_9G%mk%^aR6@*c}QhlcG%yY
z%o2A}TyzRe+L6Ml8Q+oBOd4O;>ii-nM}odqDcA>p?eE*<(9bX+%pY}N@w6haQ|eam
zGp7XGKQvU*QjAjT7eG<SQ|L1}GyhmHQDrSZc1;JC^9@eR)2~3$;57D6?UnZrjEz&g
z*#TUqO|O|t<-f9rc$I<$vgE2{Q)T(_68q<7n?zq4oFx2mx`%ZiXwL9d4_cC0m;X|j
z8-@#J4)X;P$5Ca?FxGQUyC^9_z5Q#O1c~YDar(OE&N(f?;X+U4hf;A<G}V6FZ~s>9
z&2AyLK}H#2*hAB#J?2cmQPGXlbBFEUyLT#C3X@?hkbE!d7~!<t{LO}ify34469ta#
zKiZb+b2uVFEF(RRj0c!ZIOIcK0ejY|RM^JEFzr7ZryAsN3RngAx#GpNm3<gY)&8qm
zGD0Hdco}M_L7QX>qsGM0q}iz)RI(PuL7(*oOYMzb;Hv(pdM*x>jeRle4JNmrhDGU8
z`{+|Sf{3hSoi8E^X)4GgMPxYdv=`zb4~M3--#+G48%2soI!jE3s`E}uK8>pctYHqW
z*vFm<GCRL>jcP7%7c-#B2r9f4sa?pO`|aaSB@>DaTY4<hQy?`o?P8ecJHwRoY`i4}
znuzXTJnT99`1Y2FBv17}e1x>`4a=p3J&8*cvOQJIc5H?k^Y#hd)HW6IqytNa!ZCC9
ziS+<VM~Di9PB^ec(yYHAH9kfqYmAOM`=surST;dVthx@YM7M==$66%%E~xh>cWV+c
z>exF%#R0v;1Lm~X=%tw}dqAaD_u|?7lvBlmBAF0<0)XU+84Ltlr>E>wPle>P0+V4>
z%i&)1bi;PEh)>I*q966gotrknQSnm_*$6|kPd}BcODcFfji51fg#e{=+C7ErBSNS|
zt!PPV1iBo+j`SHdIr4}I8StBUgUJajs;0BaKeJ5;@6mI6*=CERWT!nfO%yL8LJ-18
zseWF)JY$kVPxA&7o-2!<Jaqf4dU?tt=@m)56ejL-+G8q|s3?6u%CHcuaNeGuvq{NZ
zm3i(v_pz?w^ozA(T_9-XoVU-eb{Wj1R%qWqS7Tk3p+tM^o&(x>v{UQYUeHDXg*jQ4
zb70XUkvU}p1b)m@1kpa{ROxP;sW`EHQm0`6s+9~IFM^^tt5V`Vw_eRgk68tE9auaq
zMk+QXPqqFYx6kW3H_J3+%sDVg3|Q?3IVEcX#rpHRAs!}LC#+ihKH~;(5;^T6{I_AT
ze_^6#{20lX+hFZO6dHd~H$agGZ5r9Uq@2={+fBAhuqb&#ocG0+K5t);#3qOrA-Qex
z!n+<v2-+8RusA0Q!TORHEQEwUxmlxxBz#fZ`AaGA#!c;V$rm}S@Lkg_q7>|lt36e8
zs1~xSrcEJtI-isF5|ZH^gd5nG)DMD_Gmb0pKIxt$MXHLtw|4B>vWdgT`7AS_*xAb%
z%3j>9YLuh`RftgwqDD)8YKlyI$>T4nM|1nNYG;V_PD=`Wo2dOC5sWX*HHq3#7%#Yu
zC=a4L=l8zqRd!7y<<#^Zaqf!0N;(uepSPE_*+t$fJEa3l8hZ1R+wDuco7--8Zfa{~
zUB9d;LV}IGJl_)q6m-~uB2j_qJp@)!-TD>Xhg2?AO7L+9rlITHwP8hv+qE^z5@?$C
zn^}osP5Yeo+8g+hd5LPLr;2f1;^}o?^-AY5J$1OttFXT~5<9Q`RrQuH>-uvc=)5xV
zfQk5`m*O)RuveaXwh~<~hk5hE$TLr{`JcebPFo-t_z?acm6Q=nl^K_y7kgFp6qjb~
zK9wMfBq*n)SgvAyGQ{=z+)L1aN1W^SW=Nj!EmF@#%D6skug;M!UBA@1O!@<V3(FSt
z=Z22l%MO@kjFu3gqQXh;d8)Rhf(oV2h<D3ADog9E-B|6<(PSXzRm0qAk1uNrs8XA%
zzT8y5**8b+Aiga5UgDv=nHR1;=Cp^o1;icQWUpzsyiq$$B~XB*Dt!mXXieG84R==&
zN$lLk%8PhLl0g`c;6HXtL(p4pzrndlatOJR4ytTGEyy0bwY?zh;*G?y4V@OD=@@p`
zzTK8P=sx+4R7HvV&^dQPb5GdqHAE%bH(@6kLMr*R=U!evFiLS{O5Wx?&F>tet|ZnF
z9RmU^Ht9#0&5t1BzP53taU3ZWSLHs|Mw=kzZz*ili-4f8TK2kbZDdMKgKDmmG@Ct6
zyGZtWtRAngWonT}1n@er>?3~5I4AK9^-#nd1pPbV4JP-6>=-BK_SqZT`YsC2K%T>5
zmpDax+99`-)v*apW7xjDUY+S<A<%h)?ZaG+FwNrZvacwdUa{1<jM2*%pfKL2MM9A}
zWh=adnQUn-2%ZR}Iqo#p+h23DEjtzI96L5SSBXgQxUD0$yjDKx=>ayRtf|OtPK#I=
z2Ta~<D;gSMJ92K)U^O{4Fo>;tWgACiamh^iJ_nZbIMmxLerk7A&n4rwdQl<|m1F9(
z*T(0RcjoK}Yk6n(f7Qn2&SjwPLioB=4JftU)xB5&b{XXZ5k;m-9f)>K!d)xxu~oUN
zD9J+EayC`pfe3-u^tLO5`R<&WwKLOE41?SFCQJrrL~}}BoI7Z5${AQr^*s)1nR4#4
zcM)ii+vgyq&}QwPZdh`frdA8AwqYjHV&G0BvDIywZ@KYy=cXQ$%%yk~Wu}mcXlv@F
z5+|k2o}MFEx8`cYI$PUNjT*(h%}6rS4oGXca^uSNwyxeAz8>DFfJV_ZiqLjbxe(VL
zTQ@syZ_fF-f9d))tJd7<+{TjaL0I3*@j4uaVe9MVIR`=~S>5E)YAvanrM+EUc*-`^
zXOkSGDzRx8n!LtHI>(&$WbCK`SZO(QZexS2q#Dr_mOBTczL3hkiP(i<?5JaQZ<~8~
zxlXJ;3|=-^cx19?c3(XwbW{~=Ojh1tkpvmufpd2M!s;w>Zn9J8&lEs{I6lw~LuUBo
zD*Holkb(A$)1K=|*oG*R>Mx}-{$Tx57M!l$5VS?+Uh-g__L%6%ZRgDw=TF;P>NkNA
ztfI=(FR~YTYv-g9sVve-!J+O<ZF_5C{-!Bv>5Myfxux$4d%wM{_@uUzRFv>qFZ3wr
z;6djpZ7r29dV8CjVG?n;Da=Kthq0?Db!h+gj)r?^)KaAIHZtK)&K-4HI>s*jTfba1
z9(!laK|2Wv(}*{iD#l#V-oTIYm2D(_#Knmn&V9UB(oKlnLaXv0;a#zT>Jhy}%FTHq
zO>}Oz(<0h1zjI>`*TANFPa>&yg>)LlhzZp%dDot0<x;N-Q<h}r6E@gJ&TrHM6h{3L
zC4iP1`Ng2-5Exo?ox6RgeyMfWN{SkUsmEK3)>WEwS830BjRa8UVQ_>Z6o(v8Ad#q|
z?8m5(wz(UWmLmqjaxKX1w2S!@mcXVP&Y2|?VE5eq>#uk2W0hr(kiqEyE0$v&X~Z8Z
z<sd;z4j_h@2pIM*Ca*Jq*l6o6#RUhbM8t^_fD$7&j1DQA@u9@EB{$vdT#A@c`tBjy
z((rE8UCW)D>XB^6Eh2C)n&enFc+I<lL{EEz<uQO+<6=B1To#Jl;oL^Ql|3xv(V{a;
z@RrO@nW*+mi`JqE9Te~;W%jtY_AD2vcJTnzaDGowU7M_JmQqr>1|-KS1XY@CL{K|V
z)ppWmECYZxv~BJBCj$UQYeJ+~#EOvRkkhW28XQ2_dRIfE;87K90DV5?w4`XVPJtJY
zvT2*Dhr0}XSZm5LttDj}hSY+;tzUC2tk&6cvL`i|IxQMN^#Gf0I~jacBEWI>T7w;k
z7`&Kbn`u}?q8w4|eFxBO6#2W93L&)9oTxy_wl+^uh(@Ljo!%nvA}ty5R0e4DFUg<V
z>xJ6K(z=KmrH^UrzJj%o_M`yANCft9y*^z9i{eq7mQ)ERwvOA5_86M2^xKGChMb*r
zTB83lk+xc8l2jDVPe7wRQtjHI=R)UiX&tc)-r5sSG|h@&It%HO#C_4;-MYJc?TVGo
zU2X>EA<uVhS>f-g1<-C((E1*HL%?coC9)=tCBDXR?Y;H3?7OhHoLEjvN_yKk3bVI^
z61MI2AeSnoat5ITi_S>tR|S#F?85yhE2r(NyKd?r1Q|6X%PN9_mgIVHiK5)*-P6kY
z1&KwowfIg;3_xn8Tm_1Lg;pZC$rkyvy|3OTX-R^$d~(3zH+kN93)@To!ErG4(fo}p
zVid+ql?RykKJGzGE1MOO|5&|6`o$HI2ki))j<SYPHq>cPfqVPN=+v0Kzpc_;gdI9}
zMY0rJw2w`rBoS#-A86AE)yPnROWt7F)nM)RHTgbvJOMq<T}+zjhw2A#D!qMeL(f}o
zy3x5QVzN+)FA8ef*CoXr>@iY8Se*NAMoV$$g-^n<lb3&e`xHAyT~~!xVD<I|nKgOB
z2fMdLb;8g+BSgKnDD7dFSVTT~r-Bw2+Z?~4y#;%^j2k*fh^JWVuwFRrB`L|Vq?JEz
z-&lRpR#AD0OLC9M<K?t;Bb0O`Jzpxc>;Q`V`ljSLZ+yusoy%NA+(P(5AFuDEBq3pP
zw*%{DHjivCBe8v`$%?&)2ghcp{oz28V%)QKqdn10WKs-*TfD)vg36-Q39%op4?K5w
zQs2;NsjUL4GY|~=_RTdoN!ra{ESgbShfaHkDWX$*$-@1X`b{$i%DNJUe|m5c?4mvQ
zx~f_#v*=sf41%~4vB3<me!(X}d#tL&d3=jtiez5)EwAmeqU_t+TSzjI{Hv2ZbifW4
zVx;8#PJ3>WKx(ziKu12TeS7N=n`w0Oz{3t8ZRe}Tb&g`zcK(hW>io{pi4o4EjYM!@
z`A%q2oFpWIhEL_PeWaw?O+Du_=BN!!*_z+k&6}uaY<CBy<;1ngoP1Z4TzX@hCb##U
z`@*vHr$w2tkfjyvFl*nP8<=8)5EbVx<sJ^A9><UIJ+1G=HYnN)mD7QVFA?vcWSQ^n
zMxubu)HqVnY%SV>uo7si^hnGN_I>Rut9xXKjXC!r=us~D9{c|4uxPz1ji`R6CS9g1
z#hlWfn8RzuHdTLbXTD%S`5j(2n-L{OY*+i#kkT6mocokED1wyZvE2(JU~i(<L)Akh
za$0hqC^2MDTMY^ziR{Zr3alpg7(%G9I6TBq%&0SKd#gR7fEY!Db@ErbrOGYQUZ};}
zVl`zd)KNwI>X)L)lEjo9t4`^N3FR5mo~bP;fI~f5b#+e7{x;fvZ#xl!&b{yv3o{3h
z4PaEl$YlqTm_#(g^k(Na=2N&Nqt215kJ-V3NyRUj(HkuHe{zXKZGs-6Vih+juRJuc
zWzr6JLyuaD3hH2PWyOpD7^l7VrVd%nBf4XOlOi)r?XV;D7SYl)>MJgR6inWA8oMR!
z)p?FSL1-Zt!j9H&ppNigCS3w^EKLYHPWyFCil&<u=`FIN5n9DJGjUKlQu^(McvtMn
z`bt`<6p66kfeAMT#)-wWV~swq;~dZ(&TX_!?ZD1<yc>did{IrGmy40XMCPwJ?bRKf
zMl15Aj%1JvIp$oeGR7j>X(#H9FrB~}g_BDCl(87v>$Df$*c;K%IL7T{VOo^gVv3K3
zYgN)IEs5lBnOJM5(onZ?!nq363^P{-g)&QuOr5RQXRKfdx(=+KR#oIe^x$HrtFt;_
zD=U=A{w6U`0^U)lJvytIRuZkMSe;qm#ta~~IO`;Dk=&*gJmu}F%uGsYELW)P269W-
z*>3v7J04Mt;<d6|s?rSYN%07}BRI~^)kw4u3I$9nwKQc~qNlO}NIO1r^-+7OTMA+a
zcq0M$jH?SdblL;UBkWOMga)RkBO_gqLE!Y~bLXrhyol5U?vl*~zktMEs4fXTGc`;a
z+!11YXpsrzZs`8SRJJxzbH(eLtPx_G+w%5iwmLU4gj6%+^Ehlz*V}_)G1*c>gl!kR
zwd8V1*hr`OLA%^U3gYD>XVQUmlOsxa0e*g^J+*0Ia0>-_M;u6?3F=E2|I?Q#dGiMv
zn$V|q0Fgu{ZFE`0Xiqd@+);#=+}<Co-#o`mgy^gKOvmO<`9gc12&`B%Evw4@Dei2p
z_3th`_@gSi2PCH+rB8X!_Dma=^a#3w-^PIj(GMQMtJ0mXvn?bq9JKjd2?T&)@@|4A
zQbKJrR-CKtZ8$6xQLpU>ELlL<q(p=z^hlxnp>BAxkxDXE52!Q8Us9`__WEUKlc`@m
zTffZRkW>{xEZ8N|i_<P9NC}$$sLrjz!T-bE%XEz4K9v)x4$_hwQ&E+$Ok%!%B*!to
z6Z0pP8D!{nV5->ks+Zf3Cezsp7G$ymu6GBR+boX6nzkRSsi;tyt5|9sm?+WcR^Den
z-pnlTyxzHvk*b)bQfU+y5?Y%5M71Q9((I9sfC78SX-R~^W}R;L^Y)WTrk9h;+=DxE
z8?Qy5!QE5FhM4-FstG&<+fZBvsvI9&*w(b`<??ZlGx~7%{&XARhA||+3Fp4s5k!5;
zH*P<Z&)iQDBr*=BocnN*+_c1gHZSj7y2k7B?tyXpxtxWe(cX}A9o^N)3GNCRPU7`0
z`}y`M^b)1(;@5LvN!D+^<u&#T-2-IWfIb{H?!W?#Ls(<Q<v<jhmC~%UV}G%07vCz2
zka7d~0Bpc%PZKo9;f<?`uKiMzNk(=IO-&9d`;?egpz$3kN}?YT^0BDw>Mu9mxsou?
zY3DY#;2nf$X$$^J#|#!1nau$82w;bC9Cn`4OkVejSJ<!SoUNe#vU8oYuwh`>ey!as
zUAEM@spUBWS(A)9Yro#Qau+tsv(9Y_G(`3e*>BVjP$SqO2LfX|SQLy5!Nz{GJ*TA+
zB|=PTu&l@to+10KdXAapeWF;099Xam(b_ca!V+0%MJdXI6L9<OwlyA@ajt8252j<Q
zB+$&Qx@5mo?J|QyI9R#J%??v8>X6ee8{KLG@a%W%**uR^3|SfP;s+s%JMBdssi~%=
z`Mvsq_MHS*k?u^$mVN4o)2^4VA0>(8_jB^6C#lBf+~kI4RI(z)ab3*7Q}ze>lR*VT
zave{J?;QmzMZduxNbDcBm7(7JuydDNJ%UHW<YRx-Rw2`GVB_Z;SW+^A8kAW9YyIP{
zBbUR3NnxVEg@aCu`o4qkRr{0fWoa{1a40aT<Z{(*8kn*_?fQTU#5!F-Mh8~(MRh;Y
zYNDw+YJb)~8$yjpyHq0?l3Ghv-H*}6o;hoOUbqn_Lh%Jv5lI4{!OB=9VG2Bz8t8y3
z31QAwfw8~nCQ=8Qwhhm<(~>B@jz}8&%N)e~&V?i+2d0g+Y-ZU0DyM{W)xL9G#7H5$
zvE1|%@y7nTjT$nXfw)Rb!?xhG#5(18rCaQ>zp1Ae!o0M`3UH?-FgxXr5<|nP83VAt
zZHq!WzRLa*_U*tl94ORjjBbC|p5W`ULGagsEm{-OW+Ze%HV@k0*V8DdG9XJiaA1is
zlY@gZOeS{Nf2bFvh>q0>?vk9G7GcB+q8o`Xuz$?=&F@^Nh!q{e(Q{zQ1xW|mg3I@x
z8o^mJI*cC7Z7$wA$fwS>)^X?moQ(C32c0V|#o*|${Yy<wXf#Y%9uF>iWcG#jAS6Mi
zfGwOenA=EhmV@@MHJ0QkizIBPC6-E~gy8f0%l2<=g5W=r%84s+VELUe>uc*Z@0p_R
zQO_lH*uS@Nt$bLy$7uoqnCn}6AnnOoj|`>wk2W{!CULwvx3zGxzp#x$8vw2BKU)jT
zI(0;4Gd@k=7cKD#F?f|8<?J*1eS%+#(LHVdmFPkpzCOYqciQo9LWMJ%&9CJK)7K<4
zDkHeH1d@hUBI13SB%hC(ZRo8iL=p7M4lH^d{)k#9@tTA7(X;V2Duhis*D>3XN5<Gt
z`|M+8v+t=l38qvU?1I|*b9<cj_|1orcnEsp-zodp+5g>Yh<(UMsjzIG>b7o$G0=j`
z$Ud&?t=2Kv87<hdeW7JHa~Qr({t~5ZA7A}Y23kZrvu%JSnvd-uxvOWNFk9Pb#U%A=
z5hME!D5VF^ZqP|Nefz}O*t;^D;celrmRpK{3a_aweNufe4~CGB{;~s$r==c-_7ZwC
zwj=xGHnr#tir6|2l~{nOuG4<#=4tzs+49kZ6RMwurwRG3RcIs&w;DfKh~od$Zk1#v
z*KVWsm#h%#^4uk-J$Lj7b$W?g)0VbRt7S^fKFogfEDJP=Ku1cR^Hd;jaf0H;Dh!|d
z<I`)**&g(dPb<f|wMfB+yCOfx>8Kp?&zKDgvgF2Foy*+enNr7Yy=b3VFkvY6RM}6w
z!DPA9*ip&sd5OEQ@@>v_;qGaOtV%juvd@|=K|v0$8R#AWQLd%y?y~34K6?$9ynVqQ
z$#q1Ms5J1RrFNdpWkqOIXhga`p>_7zi9STDQFViZcY6+`P#`0C6>Ffl_kH$)n#lc3
zj&`vUJR~EF@KZ0>X^$RNOUTtA3Kj~no>sCL8h~N!b7srzhNeD19f8QuaG)s|-A!h$
z+{K?;osl^s(|+ZexkLmAEt&}(>Hc}MWz?s<1N8*Bjfs?mpsqa$_I`ewT#-@M9b=R?
zSP<nvvvc;s_8i1b;WXs%I<O*s^3jDxt}TnnOS|BNy(l}~*+CsZ%%D0TAC|1?83OqQ
zZG87rl$p)y0K{j<24*B4ePMlFnZrt-6ZZzDCnl#hvEA*9W<w0e3b}dOxyh{$cJzi&
zyX}h?eOcuIGHDFwk<+5-T5{8E_9cmQ5st`pbT!FydSSJL&fAN-S2ZFWk+X9bytPQ$
zl%~^_c(<3-Z=?gsh|`s7cUp3j07HTRHTK#|YhHyM(i%zq)!X=`nCfY#y*304k{Lfc
zx5ut2Sd`9VeWdsUdl5r>!~gz2ds%xGF6Hp*;nM~d%PQkdvp7M*WLE{*uGp96#IHvR
zp_l`A!JjLZI%*xJbxwIJp~}tex0lbR2{LO4Xs4?jx+*_P0M+>zWYB>&V$mV$IlqWO
zVm;NsRU^S((WX(!YW#-E5;as>UcgX;aLIsuh>26q%9H(>ZFX%Vk>dkno582>j{vNl
ze@f;EPU2K?txNFD>k<u!3tZw{=Ufo_r?RZ7SSg~Xe0zjX`?Bg~D6}lavAJX3V3M4e
zKPcT<d*$r2ucd$UALRd?|2O|TZ|D$3g-$>udsT;u;d2RNNQU~;QYeyV_sz~9vFp1x
z9A@4@6t+6BqG>AKrNSFhmM8>huWlT5d~#v{&kc#Zfa0ZEUWsQ<5?*ve2YWh}5oePF
ztGz?)I}0Q%jXhyEHiATsZhz9bO)kC-9fykZH+4PSr!abXM;%y(HmvXIutlvMNyYY>
zTqH~*TfDAE>89P>eVZM?Aa(r10f_W@rzIya$Zr@o?3RKh8P1RnV9|RKDG8?O1tTrM
zZmn;UIOkmF+t93}n99O<TboQ!S`Jdl74a6QB?;1IP2~h?;q8s!Z5^8#9_|e}fKn{5
z8QYRJ`-t&G`!i?<B=gt1-)gVzdad}%)arl*qVhfHZyAGON0LYs0R6i9O>V%0SSOD-
zuryr|bY!pZM27C9J9VBRsE&q8<B6p3RJw`#C>(uzL%mfN3nlua9yl-!WPay6-}%m0
z+8bw|y^Xt9{onoP{9G@C6GF!A%X0zN5$@w$XHPaG{|(t!w6Cu1bKE_5u{z?2as$Y(
zxwKnXS%$@!AwGyxr<|7dbOiAnM_OCf;gN7S);Dq3TZ@7o<?L0?rmWouZF#*%TV7cp
zI$9Fg3mL%w2MJ!WuyVEeBFrrCR#Q%U;iLGC1Scf};>5R=&k@FxoR*~#A*!NgoTqY&
z&qwTq(g^uItk>?Cef9?W#^_GS7RxLpajT5goHq!aoZ1r0ipM*@6U!VY<NeNDk!RjN
zcgeXHIt@^1pl5gG4Tq<o_i$4~kX=z3ys7a$ZL8X{Jv4wPx$gMnIl^Y|&UbTi)NRgn
zqUtTLxy{~GuUl-AO&dNfv650Bq>oF&xu+YyoNii4?bqX1A9Y%Sb!C<NX0%XlPV<K=
zS!i|p8gWFJ$FDkXmQsGnyB3M8o#Z3cYfVGW2?{NbIk(yS6)W$vwXJPrfXo=@Ha4z!
zmL@oiaJf|S9@)AYQg)AMGPW47eBbJ&ciNj5@CQ1pEh{J1LgEGOL0D9~>6EQ+l!!Vq
zBoR?p21s;3G_d7pMeZQ)LY%b?4Xa_=N;9RTssoFe!%QBID%3RQ498kG3zGjf=9ll)
z@i{%`F4pD%l8%JGV~7&>wz&lZrIjCI9)rm`ElDx&BKa9AKV<hcMu62eg7_3J&~w0=
zi#yk^T=^!uKT(DPA3f(fkgg;+dM7zq`H2ryPa;o>aL>bpw*rf&6Q>s0yX?VyCui0S
zI@k47))HSR+4PpSB+Dm9oSWopEAUzmPug4SgPCI?A#b+>OES4}WM<0V)+Ptm8UZX+
z9cVDwK}3aB*8A<Xm<J*A?%c)FEzVHH$PO8iPx^iKj@GGVV>8Z8Dk03gb^roEQta60
z3QLibCj*1t*}4M{lQmavOkg$k^fD_nDGd!@+182>74ic~asiU+%2H6AQyp!fVR@M4
z7ukV87>v_Sdj=cyh+x4*87jPOs&4358@WfMS{IDW|ADCpgC{t9u-+~!l=5UW10wp0
zc09>E(H^K*9%9Y_+HXTO93i<TLJ{Dh$T}@Ct1@WI5|Nd(@&F%_)~(mxM#u!(-c`p}
z-(U#iY=Vdww$0t@sD!B|JgauMq(R!n!H7^NQ+&dPyTOL~2%S3Xz!HZM);eAj{-cq4
zT9ZOcL~H;{TOy9Zhp;b@Mm5^?fm3Qh|EyTKK+0P?0)y&NNsLKMLl(v#X%96_UN$*N
zJd|^<nS|#K-8b6XzHG;;A=}ax`yQbEfgKJ&dslW2LU?2ZAFFPV$RrLMfA`OKr@c7s
z+Au_2^(Ka=M{K-XoQx>ekwWqb9ro6OtO5RJ5LGDpHqn+PP)u?c@yYcbG%l#R#Czop
zB9_}^8`w5V-jiR!eay6eSImho?u89)ZJX%6mFu0Gj###Hev>kXkZJEq*0_0U+_|j%
zazZ@~>YDpys{5AchrXyn5~|EZOVLQB^r>3$MeDckZ=)bGH!1ZDLcRMo-P+bIU*_Dz
zkLC9l-ZH^9Q=bw;QxPGUJh+Y((_4#5<1x9INYhPn!M5c$a4u8McF?(vaWvzF%6Ej%
z$+ox7j6vZ0&TTxHxefclAC_9;hCSR~!Gy~9oWZq3<v1-lo^ANU_Sue_Ugo5eViz)^
zc029$31t$C7m4p;a<oV4mys;QXXPwUqck%Q37}nUVnXRm_U?wUB;4%Kk+T4j+uSE?
zPzKI>+9ShMBqcf30mya>ici{m8}jtAHn1zWZ*(|R9i3&lD9ymWs=a;yx)z!IKnjl*
z@=xkmmWV0&>KbUq?#u8H7HKS>w8SdQ8Y*`iM=Y9mXmb`VRl@xh?_d!Rd=P*BA$wmV
z5Xve-8RoPoKuM`La~d!KUG`|>OB+^f*x=knKcxmKB;+Y+jxcK_rBmS09?O|{*9=^m
zBGJx$_3pUnAf)QFy+5a(1jsEt=O)?S7-#v|2kP-qXIV)yXdPJLTX=KPUdq&MU(;Gb
zvDRJ^tAs*0Ey-{nnv5jsudVl%4F<LSpf?!d6v6psk!`-No>QEY_#6@U-e7?@Pz$`1
z_Vo+zW9ZTv&N;A}n#J0lWOOzla_0xTu_^IJr1ydYOA5VVq_;WB{Ma{iuaP}5)y>Gb
zWI8}{Z|xZ<3oE|*C@)hr<FFFjH?|?5yUReNI_$u5QAyw<t##VIDVN}F?|J9D-{#z8
z>l$g6I9VT0qBRWZ=CTNo)Mx5(Kq@~}(_VxsEU7_6y|tv7Ifxs*#&=*(G{UrD!+P&-
zmfySHKHM#0(gH=J2d{Bjw8Q3izU>|M&Gl4MqGe{5cVw5-QdCYh4=w*$`<8AuRoy&x
zVP?jLquFU!5m8R<UP;`<^gR)Y3LW^?`pV+msMK?5sjnQNe_}lc?b{Z8rH*eY;+;`{
zw5L=)+?%v-?<Vp|$Y2=k9ayeDp+A$@q41l2M`xN+*)(Kw&P;TGiCG{(L`;bLNgm*@
zqKZD!upLhqLa^8UP;WEGdcYOFv;7{|nu0YHHXFZWQjUO9v@fEfc|iC9xbboOuC_Z@
zuivoLxr=^)R7%a{_)L|wkRc3>5%Kc7bHE5}Lw)B)t4H~xLHnM(M(Uq%t`~8Il!8RH
z4%+wD6oq)nD-R-M0!#8RG|ACj-<R(}16<-<=j`6{np^Dq>!+$-n=>@)z><~p-ZeA2
z&34u!R5QYsguG0G3b0&&rBjM7P)h8sZk?o+$j*Dvfz|x$=-w>%#Qeiivd;1D{0(W{
zy_qeOgF94A1Xxm_A&wQ?+_ybBkY{#sI?W#EE?cR2H^Eo;w!y5L-00kB%p96INkQCw
zIc5%MKJHxSZ1g$Z&-T|$=00pr8ySRY(cFxVV@9(B`GOm6xb;@&I?))FBfJE2N9|z2
zcnz{x$Wh*4DF!lGbBF9ueQt4BWh7EKE%8=uTa2a=5_>qifE)lj<t84G*riWkVYMSU
zmU~nVV#@1I5$jayIa(9mypJOMlEoF~KxCTJu6e-w<#YS&$?8~OQQ%F4EY6*ET5^4{
znkmbJK(b@`)1etWDShWII;Igy=8`Mvcv8RHZok#JOdQ+v$dsLE))lBd=b{ipqZ$u~
zxsHOZeE1}`PM*r2tXaO%PPQq7LrR+JT+L2Pd^m&Aq7u8EYSW0MCX22Z1hDevao{BK
z7pBG?He18r2@;=k?rQcgMjY=k1gz5?#vY73N7WwYg~1<@q@4Dgm_DK0#Mog#J=0{8
zXFdeAe4PAUAUQcy&Ov^Z6k%s;L@ldCDt_39ofd7kX_gmOVdvU|x|8~e1x)5}f?B)R
zD!dk(43y)kHfyU_t#EGQyT_4oo{sFx-FCiU(&FToqr$B>7_UM2izG|C&^CRaa{_lb
zx3zN@LV*`4n|85YoB57nvnRd5vSf0Y!0l4ouNlba+y@cce>HPWQjNkHU{BW@rGhKN
zl*%LyEGZEOgiMav<@z!FJU1bPCb<kP8QY+|41O#JZJ8@VUC<ubXII)vZPX#8Q<K{r
zNOVWC`<3LL{95s4KhPFhp*fReYxEsZ%?TA>zzXF7=m3Kz#s6U9|DE@)axP_JUN>WN
zIfU`aN2n4p;M^wt+C-)pDPhmF`*FnfLqp!(Pbuhd+U9Fk$~wvpI^_+PN@%c5yV~$z
z(~h2VQ$%uXKsiW1)V)OhR-_#)UkwJ|5Bc&4gzDM0I1ySH$@tow^GBVQeB*YCh`<<s
zIET7n<x1zW)<_Xd_9OXZe`b@8iJx}vW733wstR>7_Om38kiG0j>zmS>q){E|tTwcy
zC8Ll24Qn8?4iW3en#?sa$p?rG1rSKT4~&wHUMHjccy{ltZ+oe8S(I#YW_Uz}L+mHo
zV8UW`*triRAe9?oP+gEvqIv^AS??BwSB)8+Tm&qqV!4EmY5S?9DPaZN;oN0!plR~t
zpF>&r>6{^IGBbd_b6Yq^sdQ~)1YudG_Sw(WTuI_m&f;k_1sRi}{77e~o%VRST$6HF
zAi>(t*4vVorARIb;DFPTJa75xwHxf`5<yf1X2iKs7hoFl`4a-Qd32C`9M0SM`FdSy
z*|KN1AUj%WWE>^MT*w*s3+-uiZm|w|Uc>a^T?-*8qGA#D%*+z2+UAeIHEa~led5pX
z)aX#(eyP0@>G)g_KDmBG3}}fCK=TpC4v3l9FSjw?bORZS+~xd0ccnb&oW)<M_lYJ5
z2Ac3b{Ekjbe)vrjliMfkSCiY13~VP9wd*52oax!G)r<2qs)MSk<iHYZ=1-zwYx*m=
z%znLyk+iT@6ljr6FbR|Ml$3{|$%wMTwDTK@sUXr|4{C>L)&bqbGymr<AN$QVxQT&n
zqr=Xv)N;`>Y%~}d?6=wwmt*2Z^yOZz2<rXXZ?{Rkj{lsyw00_10AWI>11TL%`A2y^
zv|>+V{PsI-o~|keBKbP7v<M824U8*%$bNVBS=ECI7w&&^;yP`ygpsGr8zAELdj)Gn
z4OJtggGJ&{82jz_v)?dWvTU3?4QSKY7W;#|T)J+9b6M{{7dBe#?)HbZ?ZWy6iikML
zn5yU<M(DIlGP&-$S6+9W{ZTD@CiKX&uUrDk1UoH>eQ;F~3v!k0j}x|mF!|~J*4`wu
z@4%9PaOmV#*8NW!McWjr%#Z_!7iH#0)Ofc)ZDed_f|xw=n8qAXAYiMaDg&RaM7!<J
zszt8ZRI(#17)K>7iSyW`!0SJ6LK5W{AWkzf7(2jtujoJNO-zJj80Gr>MYSjkT8SGR
zwJyqq*?^YRgr|BM#gP4F0#;EjxhMOk19+Ym9XVB1b!3*#2&=zp_%YJcX%bjcI>-P#
zniF}a?XPnIufO(n&SmvZDn&Ry@~{0(H*L!04wP_UA@3@f`jY)^_XJKx7FS1`Hdsc7
zQbib{>hJ0~Ne%J<DE$vuQd17wn6|%fZ-Q=<bRaVC4JNCh9FNkc|Dk#tPJ2-wOw2;9
z(@skwVvrA*IOW3su{nb<$!2IjtH&zH>3<e<Ob0bjwKbeXKl-P3t04TobDR7Q`<lX1
zk=*~eftK~K^G8q+pmf+_!l@h<)Uz{S3+ylERoP?z(yg~d(GuAp{HL52ZHX~LzkBwt
z-5?e2nXZGVUE#E74^B-!V*l3WkF!ep&Rwix8K05S#Aa}~*}u1$O$(>os}3wz5!K;>
z{YUqNLK5VFeaeAJXU4*~#s0J2Eu2MZ7b7o{OG|D3$U{d5I%5CT-UjoM5e%cXFTk?G
zitKZCy8ZSP^Ap_17ho!<5m}D*QQdP9Qbdn;Ik4;KU2(H?@1@Eo!CnC%QH2DlR#=Io
zWFLJx=SNS0F}O-4YNE$EExFS+PLUtZ;lCeqx=E#ai3aeTf+kLTBE%3>L!`fNA6tFY
z!z>tr(`g5m`zpx1gvzW?aM?btjfh^#^TA&SmNblWYoz>teCySwnQ4v<pK<^p_AQQ>
zBu$gS$oO=~KA~Q!1U)4o9+=<l4JNA3!R8UD^oiZa({+ap(}C59sCqgKIqcadb@L_0
zB->xE<&#c})|?|6bYf0=!#=rQs+vl&6oT^(tmIV9i84xM`RWdEw@+#FLjjFhrWqZ9
zE@H@WTCzJh65>FpzI|$Yxa#dIYb0_}oR*xp)ug--efVjo198`5q;amLG;CUCpMJXW
zOdW+cMO_>q`Md4prDN{;jBW_UeVG?vNuuslpx$S;Zf)0LpE~Nt0VVEG>jzHsusyHd
zBHx$F8XQK{Z!o3>xhAc)pH<H(F`VEg`0D`&79@aZ_z8Rd>7YsFq;zg$GE=xs(Q8rK
zu6=esKELzM^=t24=iH{fa%jX}&<2KwMVY-Sj?-YW)a%GAszll6bk8aMEv=})f~slP
zwAl940<~U>lVec!xwSZH^+=H}wMbeLX)AVH){{J@C~AV+=e1?cs5t}$O`1|cXo+8D
z0EFCKEx9~&QTzP*rD%UD^_e<x!f6SbBwbva<G|b*d*SI6xfm(@hlq7rGAJeV0Jw3)
zdxkTaXB*z~Y+3+AqYtTM)?U<>4pCn6?-kO)fN8!bhqesZ7c`_HUbN@j6@}CZ1A(&k
zg>9z8L8_S~WVB#uh1ti(K4@Rm&1=*VLot?llfX--CHg3%ATt($KH`!Ja!@6%FK%B|
z?|`XF9kAqbI%K40U($Aj!ieu5<*+><5p%Pc!ia@b*dw3L;5^0RNWF(ZLu}93iwj8-
zF(T5677bS)Zo(W-)wC+#r?Dv&!jI@4dr5Wq33SdU2)?DJsRPSc4(vsF8fM1-DC+4D
zYDY-T_R_YKmnr<B@7%|)g7xT2_&5~8F8``B`lFwA&B7E(%)=0Pz#FVXH-cPVmb-F(
z=gqg@<lN-GdeaK~(hdVjx8(&SdIqyN>a^7ANaRI7VhyUuHMIViy}V8R(A4N=sjv<t
zDIN*Mh;upt`xR|eN@VU0Irs6>6kIeCUJaWU1JNeD_H;-F%U3UVE(7Ple8PevddIG7
zxCD-i$DfFurejKJDGW`OT(!@?tUeZ&Z(%L_rFf<|fzuwm4JyGhNTSMsixc+BZkd%8
zmv$p9go1&Y4W~T<v<9}=t8!*l?RmhtS;Q@c9vJC#M50~a`WHN1Jv#2d(zLl{V28cB
z9?(V-zs4-$4F<(!o??sH4b`<&G^7G+!4~pbX^)AJSzbZ%sta{v{YIFqxVdu(!qS@L
zwNPenYZrTy0526AD-swd!gauI%6F1f*4qX*Q^bJ-i=pC%g&%~V%3pLadreNx7V7wM
zS)GAm^WxoeN9^W;WlT+c%z-5b!q18Idctn$CUO@2Ij27Msne3?A8LbaB?e>j3|>~d
zwHx6=6^}&@Vy8X%Bn~@@p4e?&ClshoT<X=w99VW@=>#X&;3>Gh4V%O4@M_1ffu;SJ
z5D)@Kvpmd_dYgoM^_Jq7-e3`J6&}q~_S){Nitvd8#k|2%cb1X*guSjkN7Sbhp(u>u
zz_e*86tcx$U+ox#B36-jBL@~cE9qB?j`&*<phL;9H&i>~dg$O3Eug1SAVJ|gPJ2oT
z&peJw^u|VJmriXW7?<0ezNH)oqBzkaPLfk0gHY~B&V#bc*(Fc9>iY9kb_pJaWhQdI
zn>~-34E7am!q-o3!l+49X}wk!BQd3{V*q0*3=kbZY)k94Vu~tfDF1a{Yu6lc+H(`a
zZlf#G+nI?2wk+q3x_@KjElfe6rW}}twrXNCb$xJIE>G-Tx^jbasVOI~a@tnZOX&jW
zt`8|mL*clz1deBm#(&UO7A*a?>6`>sv_Zu{q<^Rs;Dx!Pc1NSf>xP``c$=gd#hB=X
zy*t~|F(HyMg=Ar9<F8J{pgn#gdMNpUT8?&C!I;F9*jZv=(F2i-{I)EsaxSduz91GQ
zfm|VuarO|h@HwYF_i<$J%P)T1?oR4O-Gv=}=O(Q=%O~q-!Z$UP>5q<$OwKs>$sBN~
zqHfCIb5H9dJ}4YOSVs0cu-t?MHnG2lY;{|0D3R1Q_6Hq&l7Y$=xyNb0l|>SU1g|Fe
zW<d}NQBf%c!YBz#$KaAYDP>?zWgXYFUM(M(A>v1|w18@zWZ055;3O|RHMh^!wiXDa
z+?JHgWB#6+Kku|?^NenuA^^qK)wuK_C6*}H7+CCu`Z#MO9ME9E1T?!hHyn8LM(0{@
zyQeoGW6b()mM%(Gr9V$Pu;^n7-V#$}$+^O+Y^d*KMY;$>PUO#(cG}~Y+Kq}okb5@@
z%GkzkbUF@K;*4-jMMP-1rluI23@b7%vfHclpH@yWHu0?8+X%;-$Smr6-7X*4s&dJ8
zUkPWy?CEhPRfj&#`_fz}cBcFfh*vAm==@G9(1Sg@f6+W*L=Bm-<X$bPxAvHIz939i
zR*jBUC0@}UsNXB@LAg<?v@3@R<Oc3^!drX9RnZ5<>uJp@@Yf#9@lB1Q(Zb-kO}r;{
zVh}&apuMF*IFyM>!o<k}$?b`qTLC1y;gN66C0>qEY<h=tCvjpU>e<^853qrcIoARY
z^e61?ZCdXjD09-eEs;`zd&-eh%4AnIR5*p3;vF^CRu-nRZeuJu7m)Von6^{Rg}t-M
zB#&T9<R*pfO;a;y(_dLTeq1hSZh{{T4H0}=-AUdf-<PPje0Yg0EED-5Je3?~h~u?v
zQ}^-N$ZSw5GB7=Hhjvll2HW#@acH`8SBRS@`zbR+D1XI<YM9~p0GVZX&xH+z$7xS7
zYx#!zte5m<67sSYgGAo}sCDz?BwmKi^%1nY&>l|(j?1u2i`L>L*S_0^yNARI2y7J(
z7NR&UHK8)-!Ci60j1)}rSSDXyabQ6V)Js{-qb%>jtl4P&Ce4^^Z|EE_^>gV)3TN9x
z)dd1akX#$fXMLTPL^hb84AsiC+)|%QIW~E=a9VCLRUU5bQW&%c75lUp=f_%i!xtx!
zQ7$aiQIH}0gwvkV0l`0GJ0@(rdY6`92r|@C?>p+PMd^)?@eVTBv-dX9p&CWqCY;;E
zmDQu1TsK*RN<@xYxZYqHQGvX}uG-r6USt4MAi{wNOQ$E7_3T~kozFaqw8venLGx(o
zM4lvSAr;wF4FY3NkjN|HT1Hdx)*?-^%Xbh<uztg|wNpc0H}Adn9q)dXO}B5aM3i*y
z5+gTkTx~Or1r3sYH9mnK)d8v3@4fb#YiwIiF3ulPv&bgkJ{jb?fgLhi+4kzF1kD%%
zs}~1aJ1tt5ab%r|X^wDtIHz%b=LU)|s+`lrkOND3R+W~~(ZUe8qh86PX)hsXrD^Mg
zx0Vb+jl=cWYa06_tzXnfRNX-bpe3X)=McT#okN=6dG~tfI=2bZ4VA#q{%7qyZG!Ia
z?HF_JYO||$m-wV?KJZa{Z~a#M3HvvnKJLJ3J9U*ec^X|6&UAoquCMCwU%H|2Tg#>+
z93?T$I_>eb#rp#*5SI6^POK!yfcVQH=Qcj1Qg<@Ji)9Ia6r)4bmU;J~?o4++q&m!D
z9dOz^<za-fO;L2{eGT)m+*0m_+vp6NF}|T(g4ihbXilMmz@6*FpoxKp?6I1Yhz#W6
zr|~)fi>@xF{pu>}0CwGT_Wm~CZ;^1uUCqEk-Tsj3?E@XB>B<Tfy@?Pth6j;zo=P@I
z-+B*Aw)>j;o%X$4f=8LEw7*vLGIHf!=cz<yf>oz+0NB^oP^7siwG&gm0~6_eEeGy=
zU0$vx1*9%fM(yj{3@WN&)Va;M89{pbV7pm1FzDRm+)6~IQ(aAN`-bkj%3cao^9GXx
zglfw4=!kt|dxP9~3<!rhy(Eg=X^9Ch!6-_KE%NT#H&wgr+-dCwr56rmbK2GGfz2=_
zdpx@{zY`phKFr2+V8Qc~;Vkm9(cP{to_(l|cj=0C&P|Q?bi~ed=aJbHZQ{oVROWuz
z0ccO^tY+e~IVXjHhz}=@ka9vuiq{>(rL>rRb4^$tlq(jt&w*(OIPB5Lzoj0Ka3Ks<
z*eo!Sgq06Vakp>nW<&;6qz2g=LQa<f{+!dEtc;D1j?Rn|B3dKLGS5P2MP{JFI4vow
zki1AF%D3kLRjZ^&8q1&qNRBc-Nu8wc=)Mga2ALp~UXB#VgAPofy?q<f6YduKNH?1z
zM@1XirN^9>ME_x2lcPG%>O1TAJjK|=rrjG%J6k4B&L+0+YQ0*9XWzL?!Wh10QI1RY
z-St&*CE0W2#FXvlfVUQ&lOG|B!l{~lPyHs7Zc;o5;^wq;q>@l5$fINuZSAz(d~fz(
zekU;)z5bvg^&OaIZq2~9WmHM}zBbR(Gt;Q-+-h^)vg9WF{-jUOd>f7+uJs%0WpK)m
z?QF!KtPQRMC({jA&Nz+Ac6D>EJh;FF2bL690cy+{=Ug)SGgxiYM7+C=M+QCCB+Ek_
zm7_P@Vy%6V8V4q+xCe7McgXhC?^A`TB&C>1{o;(%u5sKONd>mIt#f~7aIn|ItqLRw
z*xF4>2<qFuHi3jg_7u#8i2+y*LR+Ef)fa;Q$iKV4>x2jmf?^OSMV^Uz9(CHUqZ7#a
zPr~74(vxqwJ7l5gaiji+C{A^tXoY+Psqo5TfGIF&a8yA=SK_H64?<{7L7s<n#SS)!
zZiDP3edj(9Gy22#sGeena%xmQqBrE+bsHw7#}jrqpP1iy2PHMU?k0)%z@BoXeW6d8
zVvgnGUK1^8fg*cH+uP=Lvy>9Sj(UU1NU3@~#I4$s^_+N~Bm~t|1eWCg<u8A^9c%a@
zuOZW^U<eIDS@WVAw5!_$Y>g7@l#vI=+bWYoARS@If#kj<`AikNj!~laL<if&JM(P~
zwr_q<2^S~rWP4jbPP${NItcPt<N(>JdRInvXm=lWU^$HSOYgARdf))OTU-77+yMtB
z=?)je9{wG()7?<>5cjA;S>6`x3{HEjb`C$y#|qt?wKMstRjZe-f6%!}Hp5R<tuFI?
z<mhj~!%9CmTOU*Yzd~!H-Ugg4#58y|?KNZPl+!B?YUkQs8RR6stymQSG`m8WZqi0f
zqy?Wnjkqm&0V@ea{=7X^FO<DlNwg>C_Bt@pp5+fbVCUOxO%Dt?H;wIul|>#y<P;LJ
zFD%Rfp|awni4;%!LW?9I&9D{~yt9jq;N3qkP1PfAb7WyO5Qjf$m)g`LDYxTJlj;UN
zcG|_8DDzB|^pHK>Uco5kO(RX#I~YELF#p5sF1HC&VTNt^ga9Ou8=g`y&6OGrK^0hR
z<loSFd{|_imJ&uN`_4F&%9kc#(T&sg1J$l@3>n9i`mor0NPB*WN@$pEey|?TNZ&lG
z1g$`V9lYa+(;jnzQ-h8m7h_qQ%MFdRx&@rCyZN2VA6UNLo@qE1vCNtAfPD4e_&-m@
zGM1GamoKIoWry2*z8mX_eBj(kTcVuEO?I`LlBx}X-$a8v?5#y)P~0QRap;@&Lv1;*
z&nWX>Ivz0XzR7KrCbDM}&sK6`6_?3B;klv&^2)`OaxFvZ57+la#Fg=nxh`9rmi(m_
zNbz#9yOQ8jAKiYWEjGF<MTGkfK-hijlH2S@^Ku#ymdnKL5F%>L<I_J@!^pzWp~O0{
zpsG4r>%9GVo6z`p)-&Td#2s4Fp2T+7<cMoTjxE<{P&K%psHf7nB08d4Y_gs0gPBSs
zlGBQ(G<9$V<e;CdQD?|RXo}tIrFzj`bm9!=2sSDflb@<?>H{+ICEHT1_ST}j#LMvU
z$WBBqKVU!o9E?hxOvZ`G{=7<wa3yOxPsMcXWDO44&oocchM6gjg;%B)!4N=eg%67;
z!Ez;WXrn@WiEaJarmk~lm(DBUG*m!Zsh3~-TKl>D0`k-7#H4elhADKkpU=yY=>g|5
z#+-%`v9lBQ3yprQT)V-!Df*=_+Z6c}`+LHEv1?JJZ~8LR$VE$ZI|ZGr43wCFl7f5t
zmm0stX*D`9?%c;W7Ue{WF<yPle!04b-!ha2onr$mw<3C+jCwjMR?Kvd{YrgTetW6J
znF=Mf=n&L(x6`gEpWnG`a<p&1+P(^_kWz42=C5GHv}FIoXL<mdD1YH)rCC7N_S>&D
ziUjkR9B^*ao6u(&wnPN@uQ%nI4MeeKm^pN(15RW-ZJ0NHBiCXjai3g@c8m>Rk^IdX
zxbRNFLL{*}u;^B3gqk3kBiJNWuEl<<TaYkOL=>!^I_#}Q(zQ$&T}3MU?S}K3Al}B|
zR}L&DSeUErbTTqI<YxJudRo>djI?~=gg2N>8DfywA?$bC1PxD;mf+mxyx#|s+V8dP
z@jsaQVx@<tVz8mEWXIC5L5E2Dwcqc$B84M0e_*Nl3?j#|GGq1!waKW}Hh0>&D|Y#$
zxDzDyqWxh``smgkg<rFMe&<AQ2A}J|^vDkTqtg`=s<`Btj0cTeS{g!NNIn#PP1ZzZ
z5#>x)ctRR|<TU%^1$v5aadf}|L{9?C`57KO`;%_kC3}Er;S-?Xkkb-#5zgFLkmnXH
z`O`MP%P61i+~x8?jO5pm7!cpFKWpO;?ks-^1!CJ~zj?`P?9baz5o~mf$U<3tTB5^j
z^{^bHrP*J!Ri<dv<W}dlX7_|pQmn=PvR)`XT|1J4S89*ZlE^HHLqZwkF_pixcKubZ
z%R{~1R_8iK1Un3Qi?qGs0)L&mXW6}XcwOGWF#e{EY4c>yxzVnoZt8>fxAmeFsbnjG
za!UyDu7&-i%pOsDS@d_=-?f+BeEV(AO^z}AZ5ey?_XX2AY!X5w$^(nh<?JWb)Jvt{
zY9swabz~vkU5RGooj5J=1yL_CbusdP?8d(kJ{j!GX$fA*yTh7zk#2(}p?)gADD9O0
zRPU7Vf<Xz^G3&szC$aM)xY$3powKpWu^!U0fE1M_51f8TDpM5qar>7-{Qq(C4Y(G{
zE*K{b4?mpqteDQq2iDoYwyDDgBRL2gr31@_je-=4!%>JTu^#qswQ&!L@H=->F9=17
z!%pqrTX%7X_c=;^$UDeZm3)(X??1YC%dr_YPmvJ}eY7Mw8ks@)`_IG=vaN=k>sYpw
zrS?bd{{w`)YB(I^YX4PZ5M`8;>H_Sj!9*uTTa^uNXW|PwMHH6?cikvu!Ip(77xtsh
zv`rA<XiS{!cwO$P)1HeEL7q)wk&E`D&oneqjzAoD;=oc+UI&pGv5%=A$Ap6BN_tQ1
zq$P&gv}_%49xa8PAA6=@l}LJ<I)ye!FjjPuigWN(8y34VXqX0vukYh>`13ngkB)6|
z?s6|kn3aE!cdC%3%sjg=wS0W*_L|<qcxmhQ+R6?5gfoFbcdT6RTqb!ZN;b9QPdrn4
zDT!##P4p*qi^Z`WpVYNP{Gyng#sI6MC3ahi{b~LG<TK?#T7m5m$0PSSPN=5jN||+H
zzuBjBEnV4y+;L7zitPO@JM2^IBW%pTM525gSgzH*I;7A(ty>X=@1@T<j6R`cTz1-p
zvdrFD7XQ<0eKZEu-&P1-7u6ZDAfiD$6=Re5-%7oIMz<DOEKU)pAt|$R2n|kqPVvYN
z;%^Z_?K97WQHk0a{r(zUp$?#yT?dsVKah%i-kz5qoZq=pXC&j*Ve)|0;%m~C+L?|%
z5yC_BhwZa+U_-1YhF|WY%Sq1mV>R1j&#&(#^Gn$$(U_K8Jfv17Mv}K~pWVU2rGlh$
z!GUS=Mjt`o>e&mr_qCD&Wt4JS;(dHWN;-DmC+%|zb^wHJ)?HvR1I=D2oj?{(!gjl5
z)Pa3k7*RTC9d8pOr(LT~F{3R#>~o*jbtQ}S#>pQMjo(@nC(t5hg?W^4&GneydDq%|
zR;*mV!2#%<_w`0bb$s;;+uqx%lajl;5qx&`q8d!b6U4fE2NU69uY+~=1#Qr1QW@&^
zqyvjV^Q+he!62~!owF}&gGSy+BaPxEWCE3U=aSQ2`rOIj-b%JpPLje86!0c&KWksq
z^^H48+c@cg6`hi)Rm-A)KxMPUF0?PMcHR)2SLb*vI<DPmuc-Ga$}SP;Gi+Z{KMBnm
z?3IkBkP`(^Y17cI*}NOhpi^RB+&v!gM>69wwgU<hfInYy+S7?fq9DRQPBdZ#UecCe
z*@`uHcz4GEV&nGGGa-$DiNjvk1APS5YjRm<rbuRwilqP+AvYGrxxMzX1sKv4n*>W_
zgGsKyBq6G1U)r8SXOT_k>8p=9u-di~=UdgDqu7*QUQY{MWi5riIk0#-4rK&+<y?&T
zF?&VLh>UHre5x2+3$j3YblO8K7XOcWs3^RP^+4IaWbnQ;ciFBz6OlBpz23QM8$SfE
zi#QvuYty)T5+@^f@kE*>_tT@p_GL+BN4Jgk$DJGPa)MZfA*ioxKfXr_CU-d;HkS_e
z(_wsRgnLzct%6{Ii3l#hf}=+q>Gk!74BwIDST;zfB|FBuW%xRU%=qd?-N$E03U_W3
z4^>5X90&N2-H`noC(#i<I(LESOcXYp1E5(<DilXNW0&37Ua(FDjzwA>Pz*P-Wp3(L
zSMhvF(Idv#X{kN5Q1CA;)Zh;q_G?=ArOX9RUG@$prw>}!EEe6H+jDS1g_SC}5m-%Y
zDvzme6Dei4)YFH+5?KvS;FN4IMF^n_s4$P+TF=RhT_^E4u;{UNwZiMf-Gu83pHzv)
z+q$m`15<C$g`-8A7dOYS>fqd-jC`C>sLk9v_X^zFvelN<E1J-wWJu<I?vT^62oYsv
z-e<3EQpN;99NcK$IY%3v-d@+{oe_$V>>;FC&e==eTGHpe$j@xAuTLPSAWO-JxW^kz
zv3(&+p)|gsExNSXpezoka1Em~(|U?%`ZpE~J%+6&|JykSCh=qWJ@>9yX<vTk*&+4#
zzooIx@BDxNzw@o=AY_%bQ9R;7`-)tKXC5bl?s2LRbFXLSch=GN6jZULZLFIxpKwAp
zE@nVUu+ruhe<MuWmURfA2^4C`a<LaZftDmbAZ{TuEeNaZ`D0;53=|a^x8=o_U??bB
ziD0sfNY*FLJZQlQ=c()k143p2?WLiJ*RSY?8nI=l>n9ypJe{bUw&`GHwz7LqC`Y^v
z7NYcUrzNMXJo}vHXm@ll#dg5fu>Jr`#2XmgF=ltx`_cg~qj`{S%%`+8S|U*}k1K>_
z5pf!Yc<yTN4rV6L;~6j5{LVok8i$;Z_ia_;&_;^74mvkY&jiU1lFulCOWa#KPil{p
zlDo5u;PhkU-*6|zduD91Z*OW_=pOR!xGAy`BF928M!Tom`isH`7tz|*jyA?NX}ecS
zaYUEf>NYEiB9F1B7yqc!60;nu!2pgU=B(-ZaRsU?jZ`KGdE9A<Ib7`7mq;VgTte8|
zHiuAq9fPqO1`8~H(^LrsWh>-U<PQwAb!RG9U>aGZ&ooooddz|5W+3E=EIRVkrtQs5
z0a+%%lyfh#hB`I4d~%1a?*_@qWDGpiMJy|qotES?TRFXx0@)ZjH{>Eba|FGYYhjQS
zD}tOh));dC$O>}U7r{hJq+Zy)^_)y>WNN5Zf;X6ELxl$AxVx{qrJ5?LnakO#j9+1Z
zttHh>IYwJyjr+S5I2xL-@+%1CA)bJi+7+{bURUVJ1Fa|YL8)sBg#=bZ#)1d!fR;R1
zFGWt1wF54qsA36sw1_NHqZIqVTN+BPTeZ%)iO;227}}}5wQaaHc+Q=>B1h5!i*`p`
z*4rA^6aTx<xr^sC*d#$hDtMT^z3Enjdk&7O${>&$WT^itP-E}tV5;RVjdZs+SVbyg
zF*{-J?4DDuFXeujv<6d24EgINI(%g}vXep7<hlSRL{KeK=4+rKZe%1jM*xQ;(r@bY
zdu!JQX?90Bl@)yr*|AM$p3R!=S*SzR1_!q8aKJgG^JfB=IG5c9+vJo8AgYu_+h~cW
zpMY&-zS>0_YEvH6s~yvd<<V)8S!wOEyRBEf4+(%CEgbhK#6Ie@7qLlKo6bH=*<*9Q
z<Rx#v*0~JHV3`f))w?!1SGnLyo)j%n_&m&dzm0SgmMM;L$SXV}(ik9owAaYv)N&nE
zY@=;Ya~8Q$;JJ@y4n#VNP<6LGwCHzI=;7|(!9XV+hBgbw-_qX2DU%ovoO{uS0cyvv
z18l76B-@djdsC5<<N$^G%kEjV;Vv6*q=>0L$-m3|yXZhOe1zjG3B&5!ME+*O^uXrL
z&P{GK)R-R8nkYw;-3&=zmYqzZr_P{VXvy@F?s!z&+Qyj985qTR2exQyQItXvGf_C^
zUEM5+Hb+po1Jg(-(l9b^Q{7T1n>KMblW?^EE<5eXG;h3p$r9_gF;K=~YI;*|VA=tN
z#1Q#U%<nE_g)m+7G98Z8+>5G>Jzf12_fV!Fk#@1k_Iqo|-$WV)l=3ugika*Hj_V<U
z47iI`PnsL_(6+TMqWO(EcSSxT?+x6M>*);f#M`T-e#_QVEf<{WPD{L1N_8`gKinjt
z;PSGaw9ss+_-t*7oosL<8Ryx=SHfL5VKAhl9R(w#8Wf3w3l1!|lE_|dwIHnaNG(J2
zf~3#tDcN}$!EUEr_S<_YQN<ySJ$rXUK+@l+!8%Dy9J~-{^kWZ##Voc{$cw$Fn~%##
zLW%&P*ZRKG68u(@pSHA0mgd|jq2s8%H#wg?InK3q$mSXBq9})7l_)eaxt-n9ZdBOA
zzB+$IvA+rDx~4@R5Q98#-e@>(+Uzutd<r6^62;IVdtbrAwLqs&Ky!hWNP&U~Oo6&W
zAu%${5B6xar)fC?${1+y03o8&9+RK!bun5#R!dz*>xkpfX(%!!UlyW<gpl3^p3VKh
zxe3Ge{+yPL@FSIb=C)>3_0%N8N*3<20_+1dEa9UtpU`{<R`?o(l>yZuE%r6}sby45
zq5>9o`EIOqFb-IylDbr*EG4S1ZNtZPE5@^sCzbJ+w8u0*PV3`f<cCPF|GFgL%h_05
zO4cM(g$UfXB9Su&N+8%Y*nx~T;P0$`usw(DDq)dx3LKd1pQLk2$;N4G-_WQc$(n=m
zGyuq55Y7}~Tx0Ox*dD!;ctz>x0BUm1@{odU^WcRm_Dx;OE|eDu_{}y?OJN-Wxtb_?
z`gp-)T)>E`zYa`WYJTU|fvprzun)E0AzsM`geS;>h14F3Ms6G4s1%XdLSP}8SN25p
zrCC-wE1lDlI1A@jw(&w{d~W*>w|;?b!n1>L5n%Nj#Q|%2M3<v{H)Qj)eRD$uq!X!9
z3MpUoW~aUOlf+Hxm0<V!mNv2IhLfBn;{f6zEvGW+F5*)A)_QTCu1j2H$n;&HxYM5O
zjjDB2Hy^TZtCu{1oTl`6L=Oj+3M7n4-u`2#h8b9W5Cf5je&%IM3HbJudzKDvp*Gi8
zZ)jMV4SffkjB=_+AP3g{A^VQ%8mKM|h}lSAXXN&HYY*!k!#SKTNW`X`=Z|zNq~k}B
z^kpOpiI$d_2&N!{oruP??`#9zI;u3`zApeI4M2Xst37ewsN${r4&W}3iQ&nX&7$#X
zS+|vrro7j%jleT13Qx%nkV6@8Gae?iW(c524hcE--AP~v;Il(sH=7A-n6mFFedGf;
zkhreSE74QG+-KigzZ{$)ZlK+ejyqxBS5M0Vmr)7l99Zo|AzNr0fpWX;`#UsTNJ?uF
z^2+~e?UL)LVgN-$i49(vQ@(S7P|gfHciGe6kQ_;S^-<f^eLH4o{veAI#&M@5wty@P
zXsK1V{4|`LZHeBwyWS=vnzj{&IR}>8S*59{0=lQ^eCsI**mG{{mAiONBDV;qvPx8)
zVsDbd6`WqnW$q0j2(wDqqU|eKZgB!)m;qp+4a$Ia-uAcgj`!dMbq5YWZ0taNd@+Lo
z3E9{pX}i3&YqOt;`B$<q;um8_3$%mXOv#5LcBhDH__endX_Yd2MFzO~ZHKyV<`I?h
znd+UD+Uc~XP7GP@_w4YQXI1l=P^TIER@!xlzW6_Y$7Eo4f@K^CwsEA%#{-<o&RGoF
zwhl1gcONHtPutP9LpD-NVR&TP0Z8iHPN}d7d$K*edEgOdi`&4$Ot6M%K)B>?Jz&RL
zOUfwpUpq&K1k#e!sNcidLaOHR8t@{{&UOr|A}G4wyLJ_0<|W;^?L{L_wBNHyWy};`
zIdGLAu|z<pJvtep9Mc~M<EUG9vgSyh*))V*abQUYp?4ALuB4MJ1`X26f0NZ~h{Ui{
z?ai^3WHeB`fCN)oV)kVbMQ3J{?XlU;?50C|+Bv3dI!3El;fN}5o~jME2f7y9%1#%;
zDq!xIa~Drww&w};=Y*Xpm`10dC`Be0j1xpgIWBP1C^6)AHlIVv!(gx}g&$yA;5%Ub
zBX%yS6eq6}7+jwPW3N7DPqh`T7%Fu+gWhP#4rxELK!ktW`Sx^(A}!cnF`cxeAVviG
zcDvAZT#GBVv&(@chESJpP~~MV*33zJRn)OCK)xusGV<6m<=^v^B%%A4u3ur7+Ao5v
z2VtXQ3xr#<cFfyT+R}1Vs*rFKs->Xw52E#S&JV_`!D;6vkck}+vMq{wvdarCTGE9`
zGp8j-yjrf$D{TvHLwM;qx7m2t47ol(&`pO_>r^R$X<X=Wx7!akTo2o0a#E>V+FYD=
z;<V@6HjoBwbM<t}JTiES=q`{F+H?EHETyk2nZPsEqCF$!Lte;e@z$bm%lb(~TJ6L6
zuJ4J15I_eebH)94t+uP(#*$LeC}zt9K;8}>BpcX4=h-CsY)lq>e(zO6FG5(2LD&z~
zr{pJBFvaEh(+*6ZqI8Kp+x0|@S-#j^-e8p}MeO7$`{6d5yat^lB%fsaamu69UgRwz
z?Fv5#n6)2i9Y9Hx3H}0V8L;SdQix_vlG{Iww)Ug-8x>Vk_`qfCmcWwuZk^mZv(<jA
zdj(EH=18*>oJ6R=AZV|hrivrMmF&m6u0%vL2Dd8?EIJ+znOJf-7Ch#h{Y0%oY4VUs
z#p{KwHs(Kb;UWvkc{aYLSe`RzrG&58Pj*uhagY)Al)Rr7JxwXMq0y~bE#7`A*W{L4
zmpGTvrM4Hl8XES~iQki)SL|HZjLWc+*7m@1`<dK@#5W-YbCZw3CNkhzUTmVk?PqfX
zj1G?Iv>WbX1eiimJ+-w@+s}2sO|xK^mf*AqgE*+le!l$>{MMe)2^9?{;YHDfN5F$$
zsOR{&L^CMUJ_nYOBz4Bql>K6Za46z9sc@9lZIDf5GmxS*WWSUjUvkq;&ZTT}53I6Z
z&e7nOz-I1sH;U_&!sfrybup`S<9PB0YZ+8n6YN*pUSi|xSZQI&qKaEiyGVvgfgq*+
zS|cFSJGPEEcZEk?u9tQrR=r=ZuPcr~io6Vtz!I5>qZ^;J-#C-`QZl-q4b%Jo&tK;k
zIdjCi5rsg|!f$r-%Za5Tsu<OQMW@2cn{<x>;gPZ5s-Y^_hBLl(@+U2%269b;h|``O
z+&1}eZ_IwXNuE*Nd)m2=7i%;M2$4rx*Sqa^+9$&$yU0%Gw1|1i5kGCeTfLRtAsOKz
zZ!m2Vj`Bg5I%L1s`YD$=BS<A6qg5~BFQ&ajBrN;rSN8k$Vo?E=gr|mq0}FB^bYu@p
z#3&jXHZS{w`i*vX=8Ye8U^R7$_0)clJDS68><_!AkD%SeS*8vwzL3IHaWpJ#yM;gO
zwLfaZh+=V#$;dS(?dYJ>Uam9MFU)pvc=@a{)>KaIKQ1(hkW%`jbH0*D<T|kZ(aFGW
zn0^Ri&My*8*f&vQ(*7iO%!{vguEZ?aWPSV7Zq{TNNJl|~<)nvC$NsFFZdG;f4LN`q
zxFm+OB?<QyDh2&{_hJn!2#*7csZIWp-jw8}e^F113&`k$Z{LCC5R``3ClJE^vTZy{
zOHDd=899WSp&pzHTY^r9(FVrxq-v!0SIw9h*=MQEUvi+)lMK(6M5qbZ8H<124RpH4
z@f|}BpkrAQrg(_`t>EcYuKuQ8CFGUe^I!-_I!%BU&GR!yG5Y;&ZY$iKGksDwxDBGK
zFDvC9L#q8<Qn!JvQ_f{HCvlHRdi(o=C9XyQqc6aYmM(e0{-K+eNVQ(%3tcR4@su>m
zO;nzUQ<DDi%(KteJ%iGzb&GN>oTiLo;^@U*zt_{DRrXIgw)vfJe}!|M%bGf6i1?9=
zWB=S1jOw6UoVy%!XmVL%?OzI(RaBL|vfF{h`!ZWN2%1o0ZlC>YPSN__yPV6E;#Om*
zw|{H6g2IOo$^pedV%bt8!ekj)dJ7WkzvqWl&3TC8@ziAOIj}U&5Hw6yv54z`<Ty7E
z4E4OO$CS|apWUj94u*}1Nz*B>w8SVya|nq=zdvgKRTOZTV=A4S8eUQ}xq|aBA3K{r
zm4Q<>M+y4CvQJBOzK{IE_EFtqXOafPP#a8%C}bx|f3}Z4TU!*nc%x+6uJW&v1m(2X
zdRz)y4vg78=4=@SnQz%ll&esi1d1wKzg(p}m75=V2#SjSqW2$*>!*0j$DVyIMkus5
z%85HI@lE+sWSe`EsyH8aHhB5=8Ru%zwiFqVjTtMFef-(jOYjgX^PY{RM}d9c+-3WO
zdSg3V8b;v_CO)ikQ@AGW6YII=DI}fHfyMWrqL7fqVK|?3wuq0=1L<SjffPb1yHAS1
zIj~ZTCmMXoKKVIwlKw8u3^W1L<|U_H2&)qt2x$D2v%wSB-n;Ccm0mZb9Il}9ZW|8<
zRi&<Ak8xnytzohg27FpQLjV;vllbev5>cZ3zG9Q@({m7$xU)tSy3uza65oa?jQ1I5
zi_vp9#muyGtL|-8efz$BW__S(Fr)-|gQY?j^1MASe}!db5@ndXm`>?gGXF?j4YQ4X
z*4g6Lo5mExIOG6w6Qo2f^@DjrQnlx|eqm)NBc0nq7*4knLQ=fsvvWF#5gHuilq=^x
z5g8#|MMtU2YA<LXKsr@is7e}Xp^iE&`9=jp(SjGmKBp~SIG03~pA+-$wC5KPF4(P-
zVeNAZhU^BC#a{;|sbqPcQh-$Qd1q6=6NX8$JPYR*N4*HIO{b*naGE6SVPO{G7dbO?
zC!(;qj!ja>_t@uGi<DWtOM%;9AkihXq|i3PsZV`-VQv;et!JETt*eT2Abq{4UJ>EY
zbU{)T+UeCty|w65*oF{cTDamD)Z83oUb9%16eWo$Neo*1B4Loug*N~h@3ehk%@gmG
z`2r#3Ne6Zfy-K~4x5-(mIMF(YhBb^jXLt_w+ZW~FMkjF2;mqMyT(yq@-M+Z?l%hRO
zICr(1r4!23Bo#!py1%5GQ<*o^g`Cq9PD>%fVY;9dIB72~m{v&jPk7!NOx#E&Rst98
zCEdhG8qmaLwysDw$v#fIrhb0sjkn)=lfASdExukjsB<4*mK1@0j>+z*UDJ9)VfQw@
z!1R_)qr;rGU@z+sH8dJU4`}*8N*N%YcG@*N%a?B0XkS{-Y7P{XqT+Rnyw~2^!9J9L
zX5UtxaO9HM%k#^Vs5$Id?vmz*;vipZm;=GYjJtb9n+>qD{LIYLo`?zpQ=NA8jPQ>k
zySCPY-e~ObtX6i5mY5ao7wPq|K)bG+sVnRf$*8*>SoEgtBIJ2ipsOuwU)Fwu5kr;O
zpMsSFi{C^<5?{qg5nDjYD@k?s%C?I*c@|d<x498P7h$|gbjDwMRbvjzIDu(Gl^B8K
z<OGuxf@xT)Kp%@^*LU4PFa(Bz;J~zLIS6t6$~7D9)eS#34UBJ^BrG_aH!%syM7*I*
zSqNEW8HIM8^VX8KJJDsK;c>*J8yf;r4P}dWcWX%EvYWcmB?>9YJf^{ENn%313o5w2
zrV*37mpa$6I#H|*oF%x=Zthls$}tQrGC9JQ5IK%cyYQDnUOi0bx6~Wtw^an(1&orw
z0>jBsK(e~E>zG_>+Ep_Co@C8iOImEwSu3&sgx%J<tg~T8oZD_w%ldn9Zm->5uhoz1
zs5@;5gn-tfRSV~`&R$nok1eU8Y14~Hr|%Sxu{w@=NxP{1QVP-IE5I`F+T`kY4-8Ij
za&C$orK(Kru-s(!y63cnD#{z#bWTe{e-yf?b6~IUCPd*f88#YR)fZYK{8W9)UfLUS
zO68Qp?8jZs2P{QV`Phc-jRnh!^kLC+U|NXzoey$oiG6wA;p<j38h7!II0uvswI5df
z71cAbqNu<L=TJz;E~h1QGZB?$HB5yy4QU~m+0uGS99xO8Pl6IH{a#p{{Dr5~D~?Z8
zN!?}H4`e0IKclX)qW^)3wCvc5uV;DNMiltlhIxzo>UUM5hrq8$Dd_EQf1PtF+r<3N
z5h5N|F02cjPEvqKoGJy-l8@cbiD!4To11UD$+<~81gE@=zKb$mcNWr!mx$L#q)ur6
zX(FS&3>O?uDV}^++h8wynRA^h1<96&<}uRvSG5;HSal|Vj8mZ@t8ehq*4}(W$`%s&
zjL=e?5_fl4tRx1y1%Xd%H37?<ghG4ZDaQq3KRQ6kBGpX9b6{`Ed3rTE9+#ScfvJZ%
zPxGFJUEtGUue<r3D`s#_u5LIaUx{-U<4H0taevm<RDXnX!kVZN%nnl+Hk8q6Px>S0
zN8i@ADGADwa|o<>kwOnbH-tuX=lt2aZp$d{=fJ#5zou=2mOv&_3g-6No4a@GNQ?tm
zfzCRxdbf=CveLxCE^NmX`_?Z)$0N>Nq8Hp(9&f6b*@m2}dpE9HyT-Yx*Kv+OsEAMj
zSP!a{Ja@LSUZCt-1uf4~7YJCbNS-RAIxkxHLw4_?eI<&dz^!yF=DNl)0M1jA@^7Hh
z)qQ!n<mOwP%jjx((j;No{cW{Y$au<KVc%J?bu4e^?SbkGbdz34mC9I^gWg)g>`n{k
zTrdG7*4cyIq=+RXtpb*-x+PBBc}w=|o_ij2E@OriyOg_53Tk&yy1liVp*_&Qh$VMm
z@f(yruvHof>RV(-AeoXi`nL9)xy$qbMg*M)*3sk46jPF}B>A_uSr{JOH0|8Qa~UwO
zm|nf3S~8ln`PdMyytN=#Yv1a;rC8sS_RiK5$bl*>Ly^6~6lNEZF&K!xvYsQ_2LqBk
zxVwD;rs6AMWUztu9QY#jkZ(D#A}LfLP<HdCnjSUF(x^@`rA(hvFLpWYxzI>)Oag@N
zf(>?v+8rp((~4fu6dd-}4%)&*(#jgdk>8V=XhTVOS5S+MlEXdcKAFizBC4&IU!33B
zij7fQ6j&@Gwyf9=7{TbPlJz2S%r-YMYaO&T($0b8W?Be!qIWu$IsV#kV%6<SZu7ca
z&!L+mjU{Z@fn+u2+>2e@aKr64*l0t`<^PMh_m1<kI_~`^&FPMvIL&E}jcveSgXykO
z17swSB>^tDe6{=4YSC(Uz29BL;^2&poy1OJo5XQkQiOU#5(xE{td@}8d+)vXp8NjH
z%z1Y4y}#dm-TwkFmhU-pp65Ab&YbB7T_29OjkMo7I%)T|5!_1Rey3F1N8D=YMs0JO
z$M9N*OF`?`)x}{;eEQru8*3!~wzbahc#jlq@@L0Y+O~A>#&;ySW~VODaL<#SnI87R
zi8>w6PL4|N?G$RyrecRUKbvSBzx2&qe@98wVBN~4HrWtma^`^tNSx?YNU<;x3FXlk
zyyJN3`ew#;WdWJS`om6T^hDUC9xFN}^b$7J^*f_vbg>F`+4^acedL^Nkf!<%<0$%6
zA7nR3sUi1_5u0u+4^xH82V)R5se}v{4lD_1=xnBao)=<a<nJ!0E<WG5H93uhG->x0
z^61nQ9g(xBS}9tSctHmqGpVyvS?h??vaR(tIX8)J>pBWpT*4(9MBemqKuCfOX7v7g
zPkN1=y!hd{vre7nxi@^De4~eLw!Jr^yhMexHKAP&M98^v)r#fzK(5?0*B|mz)$`k?
zr|q2$pT*YWVL~}Elh5PRgP^k8zOl6-I1E*$Q_kJUh1gT};G&g~x;$bNb&<RS(K(OT
z3AXWf<wyJa2)}D~6b`Gi;f#G#)rEK!r}s&3b=)L^wJ&*ZubJ1jmMfQHJQs<Z4m{qi
zb&&EB8ffos>MY_0`Z)2RG^Z+-R<ftOlNGHQETX-qAmX_&B}3^RWmKsDy+Pt?GoI1G
z7Ac_Yp*BOylpAT-DXSB>Oin2QkMc>{dmA3EnHnW#hWzK$bQGxEHDf?JN>GW|`)VjM
zG&~GFdl*z{b;$tl6Drv^w;_i(C(&(>Qy1fFt~8SBb9~=Y5RF-syt{3}fe5ikl)MS?
zYGB{mz4w6Z6w)C&bwTg0zWOSAf43B{6Y(bTs0xJ%DcpfaAFu|;5vJ7op$D-KblWGr
zI;7oB87$d!;MtS)a-p2EZ)*$4MJz+QD+TA+#OGVqueWbcJe$I#Fzgi6K_Yav*moqN
z2JicrQ%cn4&_VYkHKBd5tCIj9gxIM|en31`oV*zUUGj*1sD`CQlFK{+UhI95D<GHi
z?Zx<*lP=m2twcdW)TKtAogZ#fl*vfhbSe8BaUjuKd1{Im5(Rm<Aq*USvr@&W%{x&g
zd4q?Vo1RHEsNbSMve$lR`<lY@VkxABfSfvzobdh>IU$GayXs54J$9(f`P?i9WsExv
zyuo9}c!#ojiiZ)J$V>1ewK}(~cYdbnZ5Zl{s~qmIIW#;MP#2T=TZuHO)4_7%wb$E6
za|M?wdnUi1LG|4$Z(n}L-S#~R85-rl=XXb&8AaWfd;}%J$GSC5#}`GJQ<ta+x@(5T
z1&9BUeY_j_{g`cropSP)w{Vfez&_DEn7QrKLc?zdB1G$Lp6b~ryO${FV9Y583W?Ef
zJ^-k{6^#ZZ*<s(?78lk`PCZxwoH`8^goA+F_dRWRnp(L9!NNO`6o=~sO}DIbe1~Dl
zVJ^h!@2}^ZBI0x~_lh|Ok^@`<EgZHV$YXq*7{O_1)sRz>R0*Q16W|<5CLyb6o!siS
zwda>Y);2lCIE@G~oE7FgQjcp#NZGOzSu8O=xozM{8aMXFhDOO`K!*11t!@&@0k2RU
z($r~s*G{kqI}$l`buT|v{Qz1u+u0bbiLp_q5c@@^u<!w&ZrfFLngk`nIfF#l>cV8s
zz|i8`?&|a6fWj2HLmlZ2-rg-AxJ51<wx_<MV?z6qX>#fc1tOyavmw9xBkbk83wLiV
zP^chz=6EQW(aJcGL<AJK7i?d>CvA^;(y1Tva^jHi1$Z)43T%|LA(vRz!1lN0+S1c?
zyA-2aZAEp>xQ<fY>V~tU9jI<eT_AOS*`!cg013RylSgDn)Lh+`ASady?6QLm?__@o
zLRcyr2a@DBh@8A2cH5(^b9j-Vd!d+Ahz!DH(Xc~pURgh(pR!9(7du%JucBZMmiqmm
z9d0v)?g4KU=2}}z1B|Fh2d=j-<yusEW7?7K?K?xQyPwTLo#uYk1W`hLdo1zpw%)As
zGZqE54AJ@VHp+<DYa;mB`I8PLsU7|_y8ClrN9!}8!DRo-u1j5#BlueRgDG|({59-}
zg0RH0mMS?3Ly?^nNF4|g;8=Y2;<2g|1A36v*v`ADi**u`pp8ZzBYKN<lzXM5B9H>K
zINoNSjB04MC~A1vsuizrqB@beQ-o*IKwrQN0Yqea33rv9tZzeg5SUYmNnPSdc%q|h
z4lzGfiCw}lJC%F*!3W=Ysq?!W0(yf@?EY{FINd~xJK{>c5vM}Cehr!K?M(hA(d3Zx
zyW8|Snhv65INRnn941rb7Ms1_fu!Q?cG9DqD{M)!%<c-FyGvK-4X$w$lf};FELEr=
zGeLTOJKrtG@i<?V=)Kh?Z>ouT%r0~rPTnNic4_JrKPwGYgEx&^sHnkrNeJMv{a~()
zXe}}prQAd$36pgaVcLGE>h?*)C%i|GM_nLT_Biz|!K2t*PQqPFR;^r)VW?39;Y!-x
z6Z518J7*U=xKyY*3*=yO7R(ezop9h<m(@y}KVRL*gu24M!`|wW*y=EN!k(<IiE)x;
zP5wt>UyeAC8eAAVU_Ex(4_6OLl^aw?ItKLy_izWEy8@CYwas>>?5X;~K1hVHp3oO2
z9#y6TN#(}v{jIu$+<v5?-&(v;$mAHB0V-?qQi-p#GkND9Er{MGc2R1oa9Z*P38Kt+
z_OM6unIG#$aE4<eE}h~Txsa&0dmOmtaQTXL%k0N<d#xIs^!hnG+_Rs^7W3Qgd7<;W
z!}z*iz<x4OL{8=yxhdr=f(}xF5gBYhRdp?|>nYZbx;p2>w<l1$#R%*AoS*KV*&k}J
zb90Y7b?WO9sI&b{^5l_Y0B@cx{b8r77BHg;6<+bPtqygjvblgfPMsVsFMGTFTuo>@
zRQ76r;ihoTf#hc0h>yb0=eps#)!%}>bJ(e<WzP`*^K%F67pg81SXQmWPF*ae>Li@9
zWGn%3Y4(dr013fZ-@kv_Db{Q(WZYN$s*)PX0r{8OFhsvO*9d)Z>f#wG3lvk5Tes6|
z*e@@@!Ph)-lamuAe$+cSR6gRFIfWPP2m6(JUeCz)UCQ|6$HZbBcHo+#&Ed%Nt99_a
zf4=^8YKl=*$V&`^&9D%Jl<4_u?I0j#fh%t*rsv0Se{yVMRtl6~Z<}Tv(PuF5PDP$x
zamR&vCzcuk$bKX7_r_~(aDEmpRG=58wcqUOlub?#o^$F#NkF{rg#A|6Kq)C<b}%VW
zSF;=>IyFHuw($b2!}i++m)1_kN~s0XBoVTv1J4r%Bfp~3<l1Y$lNfm0nDa~DUw!-9
zCF|^W6Tk6W892XnB+hT!s8i?ns%NY++Iu-29&+mR=5V#Wa2XEUx{3Ms+YlgWA`Uy|
zj(b<fh(+?GC;o%3U0kLDZ}T2+b!m~&{G72r?Apcc+##(|Cl+3L@03>Ik7@>VvX}c8
zn<!iu!Qdm_;EQP}M+fXtUgZ489iAb&QAc0{&npfJJV(Xc;LY$7fe0`K3NBVy&Yu)M
zB(pEZSX6f(L5@I@-eO9fCJDU#X(Gt$dLzzH5rKD%+n;sMhc*<AOr&K)0S6?INN0s;
zxHj&eS8MiJ5R^2einRxlt=al%=iOs}(a6XfR<3jkMGdr-()tw~wf$wz=#uG`*k12<
ziiNy~U=2}=lWOv>>R~vY;A#{anY15BN>r=+Q$w@%*Y)sj&==-67(VKXytK#{8LT@O
zTl<^F`9e-N^zL=ac?5;0K<XI6Dr0|}D2n!P;QWqm>dcn{21d*Nu6-Ts5mU~l@m8m$
zQ4*O8bNlV@+ZW<66^9_o3*mPlB_Iyz`n!K<`=@_@+?1uu&Ya3>vufonx7$C~@JbmG
zdsL^J{NOl$9C!>z_#%U#YzRTm?4OQTN{`<7IPr)>akK2esg4I2Gn{|KZq#x5&o#c-
zZ5MHd<5;IId$WaC4BNlt&Lt1w_|U9Vib*QV9>d*d|JwHUsyoPTPpKw5?;!3%XYJqW
zSxqDEM9Mzn)YVu%#q-Mky}c*At<xAD1~mr~%b&guAp<~=bx4Z$KVtvU!57T3DHw(k
z%sY5#__jyxXK5tTx!^zBKI3x62r{HXvd&b0WE>^&to>JRgS8XN@<TzCZ3Im{<o&-B
zv;u);i&Ln1mn)ezw1VuVL$wn*jdVV|xsK26b?WpY$f`1wecFj&4oELF<ovEl(uo`Q
zvhg-4?mzuRt<_S#zS${8!?e68QM80MJWA~|>VraB+{dt1>~QLubk&=ZF5f=$#8VlL
zIW#&oG_wUaKTPLjc%VL8qI)LgCsgvYPSiOd8<KjAam%SpiF0VGw-J@+Gja@Xedpy5
zI=>T!a5^>pfPHq3b_;xCe~VK}(st)G*Zp9u{+xXD;wQm(e&?g>A~4%Nx5g5rpFQyi
zRGGRMxlVjpd-<2CW?Jp@s<le;0@n^|b=qONsY_e^^G_sJg&@)K2W*MJ)F&Kx4Gu4R
z3^HHPEn(`sNzv@oWgGBh0tbRGY{ObP%p>e{DiU?s)QY0ijWdgmS9@m83dRIExKk2?
zI5ESK`9(EAF-Gzw$-8RkxG#cmn5LIe4*gl(L@60a&}OHuP~<$iq|k(lhRyG@XIJa=
zKvXWJJ-$$<6eZ{fFrt6)iR6Tu-cRq2DdcbGH{i5V-&ee0#WMSn6X{bZ`$C0Zun-HE
zEw%f<Frqnai@@k8La&U}_nZ^Oy{_S;qYw?$X?bxYO4^s^$8f9M<or%dkP?^A9M$=$
zoo-)NldmZhf&bJjbPXP?0#ht^qA)4(7MbL~yuK`ZtuD6}d2e-Mvz0kk84bRoYbTok
zt{p5Abw#t{Dqu~xzQ7RUJNA_e%aY!EKth4$4Wbk!;wSh*jpw=DJJa-kI4IZ-NIZun
z0yOZLDT2Kas&(1XPW!5cV3S-`>XaHXgq`86G$-N)?W-GMSvSdkbc(eq%Izphw#U9E
zS7nG)<fBd@8ibmR24_K(uWj#`nwZ##duZDX{rhp3hK77y^<R>*<mEkK8|u{F4m?=y
zTo0KTN(!Wp((#0ZZO`k*AnSN}cgz3Hf#gAkjcBmNzWzk#@T<>(t6G8_PZ#=4y>wjC
zaUOM^h(Reeg<^KGF52@KAyF+J-HF&+F#3oCk50!?#4}20m_l<NdtxgO+6x*%o!BtY
z4e-NGP4YAdLs9@8g)&IcM>nd(33f>hv&nXHABbQA2cmN$$lYmsVYLedBJUCVDfQz?
zZ;(2=`3_SE0X$<bYTsrYF(&k(LQXDs3uPJbT76wCuuBY~y|}BBVLZudtIJp+#jfqJ
zOWWSP{>H1ELg1Smmcya1v!h*BT}XWg2nIj6C`Y0L&uPsZM&ez;=ub^-a(Q)1dO9h}
zF#vh1OLk5Bd6&H;J2k@1a+{n&p{|4-9UVvQiq;(%^Ja|GyvfKewLkD8G9gCENWi(g
zWB!0$*^q7(QG~dk*2|?%QdS2AGQG6T0xu782<IZ5%?WRibR0@+X+HC^Zn{yi(^{t(
zb-D2JG`FkThu4e}9_SP$i9N_am)g~BUL;{KuP0E9Q<n%wpQP<O&x!=vv#+EodwJ5-
zC2Mb4=lsr{gusff_nI0&*pR^#WtL8B9}qRXb9|Iow1HjQ&O=-kdy@mFT<$Okh4OX1
zu92g4cP_itDJ>#Lsp5@AwCigCX}Op2_A#eU-?}11#e;T34tV2~93v>`^tq^bgB#m7
z2FQ2bm6nsu(QfL#m`e9Uu@sNb9dICe`$p0_%Wvlu)ll14@*LVU{>s1rB<3-xc?M3m
zzdChZ+2$S&UJ0<}v`AgfC%!FXJ$u!OQWL<$4J*m`z^T&O&TqT;G04TMPlTfQo+~bK
ze%2fv46mhwFevg#+GVdP2pAgJC@W*`Um!W{BQPQ4-LTg-$_lYlQt2n~8wuG0VzZx9
z22w>YX_ZLMNvOp8xHR)uMDQEBF)OkzSlUDY;219*cs!XG$ZKZO0qvkIIg!e$7dpSf
zMtRfqwltS+lsrx2PNBY<Icj7c0=b@iw=HXPGTg_DiBe63Bxrp@Q37DLEziY&@WEHT
z%=z6`N1BG<_;sGhd?fC0MXioBjiGm8+|;#a=r@F%C*=YwBawE-%_pAvM6AvHw*Q&`
zIz0@=apOk3UfZmZg5escB(g9;DAKsyS_2gq$oU5AjNjTq8b1e~Y(JTdrz4_yWvf{R
z4;-ogsi|d9zGp}>OqXcrL3>?(SeaK494SRREtG_kyp>7;WVdzG-319$UxF4p@Sx9<
z+okO%#@JTnyIDj@<dkB;l=Mz0nr>hjWJ#trTV3#U&yW@$WdoLVUG~XA;=psWF-6zg
z?QOS=D&OeHxKk0Y*I@}8&8fl;!D1x5#NWg2XrGUL1S<@taDd0baSW|G@Wg0vNE6Hn
z1e{LVnu3&U!Wn^2A2@cvfoRb4+qf~c-J0?G)|x30GcJ8<Zg3!Kt(-b>(TbFEYa3>x
z-IQ(d7#bkzvL_kVlLi><jXP~!*IKUZk@CT*i}OWJ3a4#-Vh-7CdwuT;x88B5-C56S
zxfFS>)ETWVHDfIL+<v>OeIu;kqBpy})n(=eDO7jc-POxHSQ4c)_s1PbpaOFo`~ewx
z><w+X6nZ=i?Wdv|$&4R3VQ;J@Y8gQ?5Jvvi>6hE!IfUSE8OVO5?LDxaPuQEfsn?`~
zLBV}N-W^Dad|X(r8+qQGlZG=KIqjU1pc|T%^qvz>Ju{XS%)(wjidQH-rZ$)x7~tSA
z(%f+X{!ENgWtg+KRI8^?i<WD#11YVv_3^G8v$xhvB-xYnb<U|10hr%*$xAM=x7Dzc
zSrvz*ix>{g9djVr?B*p)?Covg;?B0vV>oqrrYLFC)bv*Cb&U?k*{B;ls1uFfz!|rp
zTh#Nq)KFRw=25&95TV(OY!L$+s&>r>CDsT&__za!ZHOU(b(7RA@ovFDFkEo$Z1|B3
z)B>R@{i_2H<WI|9F+6O>{nqcMLHfiLH!@HRNN#=fH0c+vt*x}vu?n`a^_F8V)U05f
z3~EQbLDHO`a)zFH)JD3m9iVSW8^S2`ntL5cd@XRYuuOzTDhfnS+f?vHQhOYH;WN!Q
zffwfQhe=Vq$9WrV%M$VX5!&rt9c+wLyV4}x+hNy&ZAUT^ZU;fIa;<0;Uyv*$R~#H}
zY;#w)KiqGLFm>v3t|1AE*No6L?HU_vyv|@NZgG(@o~qnJ%lq7++qW%kQYFP{x*3d)
zje`zE$bI`Aq{kmm9%^;({?RRXFj368W542E#iOvvA>*K#hCQBW_<D1nOGTzeoeJUR
zRpghQOxVr3f(p`!^KD}PP=9oC+TPK1CtbEN8u5N<UV!8#x$fF)Y^s|EU2GVZOJAtd
zjTCz@UmUhUwNuWEaM+iZZGA$cY^j>#w@v4WSFR@*ncsDe?3JVsVz7AHLN-%_e|jLO
zlr146J8<!vC=*GKxxwzs3CB{WbZf&-x#&;mDP@kU8B9Stohi0<pO-y7oH+CI4kWNb
zDhfmi>FDkL_Rf%JMeJLuo8VUIm;%RWm3un;OT>uEu@dux4lMMfvvmY*W0ySclw{->
z+-ncy%uBkt$0@|;Ii2G~RPMTWwpj)pn$HJy!k+nUFM5G}V?#qMtx95xZjP#A%H_Fb
z57uyXHXmcS+_WBy%^?ROf<-8?bf)jh$sC`!_`dmbL!(YPNk)>U*gQJ_q*Ne*ihWb{
zOxhiVR*|aGTb-_0JrKEy_M5j6K#!%jL|x(3X#~qh2b=BPH8tWl&N!v){CZq%2KJun
zymqL3wH7i@0?EyGixPMZ?4g`VdD@OUg`7!voLv|9g-`o?vjGm^y<w*i4KihM+~HyG
zt8Zm|gbWpJ!hxjuv%8nkzPag*m1Jjl76(6^<fzQf<4rU=X5Z3b^G*@+|IHB_xH>Vv
z4f7KIHUq*u?OStZ1~Y?6=XVZq!^MZ@j!)V9a|AfKp`)e{BTzqea_T0X4`hS+ZP#A+
za;Fdt<o}`!3p>d#^KA`H7a{X5)Egp>WX^Tq@r;h6c6r~om(nxsGa<&ZiXtIW-Xcq&
zqig9`Xd63QxTg5?wCEsr*mtz4xfxr;z$ps#xao2W;S(QhzyaZ)1E*NMfF|%QtSt7S
zRu>0Ya#7SJ*;+k2un)J%8f+rjz`!X-SJKK*lpDtMv^`u1#~Fg_a+M_J<Ky(53#SJ|
z6X`;C^t8#6s7$NuAf!t-XJb>d@2V#k^X83T>73%6R<$Sx5XN+lK4u?jixNUzXq)28
zB6?U_9C%7E_o90zP;%D3JMr-jR0L#U5UF>nbPBK;?4vm?!??4Jcolk68%C$?dumSN
zWak@Q@(Tx2BT`1_)Kz?JVJ_%va9z+u#SV$SiKmlhomTOr%wTyR(*gVV!rKtbF)*XP
zqArnHNf$Qs>=T_XUAhXz?a2>B)_^=b6s8ku26U0MjkV;v!7(Y7X`GDq$@-GirFb}o
z$U$9_$IG5?-&=4UQu6!fBz5Y7uw|MKjaAW5Xjpij#NPIOjpE08EH;lXibHc+#arah
zr3B0iVS=Ihv+r-?MZg>aA1NotERzZju6V`bOZWqc`Eqoe=J2OjFFjvQ)2YT(yb`gV
zwyj&h!y&<i(k>iJfoRaT@?v`=fA@w*$DQA~(G<39+w1w9Xq~*u?|?ecf#fnq8jC;>
z+fl9A$N2h$5Ub<vNp~R8TJKTD7Hxo?$$TuqXLuF~V&GKdF%RJn$E?62_K)p4@sv^l
z5fV$juDIv6%fSvPw=EPz=Akp;<03DT-HixC{dxm*Sp%o)65d0zaDoYeRstzbA`mDE
zi(MKl+tX;wT5TAos1eL>8|uwaR>~yXn|tV9NH0ID8}LVk`@+aQwy#?v$^U#Atue9`
zc$p$Ba7{ShK0deI_E&4EMPoQ}ZmWJu762X-nKULWR(4?F3+x_f1V?VElcjd`25$nT
zm?-#a+KS6LX9wFm!DeOa#xm<&9eA3_BaYgmUEfmnNTopDV9uU!;JJQc%|Jj!ii(MS
z#10jt4uJ3#(5`*dfoSwn%V0J<Twma}weU-fj2j$B3J4^#rUPHZucb0QM0Q3kqgF($
z+mWuH3fB)a6r&0k4Af+29C-9ovlld*_s3e;V{H%KOn4n8ZpwL%4^7+$>&%|B$J_o`
zjef@|<&ZfIMFzznrXA^MJ8P!<o1H=-kMf3L72>+0Ct8C$*RAocu&O_6#~R0u)i=@(
zQW1-;8z_|}MA&`2zN{>dr{J7+Ab{!j83z)r(UqgUWQlSAIngcg0hSmM5}iABK|5q$
zYJvk=3oZAP#P`|B9I0ZDCa0ZJvMR*4b7q{%-cR5~H{ldwSu!hAinz&UEQq^D>~!uH
zqS~FGc@UH2%FeVmk6|pEb&73Dp|FX~L2~SDTaDXgbr05+su)A)PqeMVzS-r)v~%?`
zB^}V=Aj@XF(t#A^tk@wA5>Qqm4u`@{JKts}tOU88LZFJT?D7WJ+jhpvOIYPQ2;lFY
zKV}!&A-t-OJMt8>L`@7#syDIu<P70ml<WuFz$USslRlt7>{RJ^W~BB*wRZF#f+!LA
zfK!*NF#@Nz%jVjAt|u2&S6H%alU+>GJ~2Ex=KM}H3f~JkNgTKNR>xjQU*SDYT^<2%
zQe?-I9qOD0oswoz7n7$HF|rS8ziU4$Z3(DlKb-xU8oXm>!YR~Di}YpGn?@lLe;%#Z
zQ{7A-he-eut>n}sYtKFhUYlNFehndV)ifjN2=^n^O&m7E3zBojsf!s`XrPF@Ze7($
zb=-cmO$#m<{I1<4ya`jj{aDp$P<c2Ubm|gmuf6F8`|(z|?&UW)h1zjU4{M6Qul+<L
zC6hB#EUQzlDT~V+*p;$4D@lcvYWwUbtFM_VJ%T?prGzQ~Np^0-%;*@=rvv+`ZpLJ`
z6xIjR#JF>Lfaka&-qPL6%A5r5r`yuLeg>h2l31*ZAIGxxG3?eq(^xEsWsX`>$byw@
z(V^kA75{MW(3UOsvt57E2c@}c%Erf>J8+4w<M-On)vvA2uj*qYe9Gib97?<F=exPr
zK^6*gr%qp+8K1b%exXhMt?QRLh1yg)T_XmRp)$Gli(RKW-wa-+oDOl~JMef?4h%Z6
zv0($zy;r|fPo@kYghMumCy`Z2H=6(tgiOvMn3+IjIZZ4CNd9teQS_bnjrO-X#aMqS
z^DFf-vm^iQu<zTiG<?DLY7zoKxps~0p6U6n52~xAGqGRo<}cO#igwmUJmNsKsvDK~
zTKe!`%QZ*u3iI!j1QnA>ZX4WkJKUAGV_cu=)M>Wox81qPeq#~Kl)=e)lUe8ZDA}Gb
z@8{&9ZSDgwPS!DTpugFs89p)YW#C}XU_l_cpKiSNYWuB5(2&Nq4mrj2ww117Fba`7
zI&QyR^Iqz9CQq&q;lYr`JJx8Y(<-S~f40BXey5REeDa4jkHJ1VRT>@fD%1A64NOGn
zm?A=P*li23s)t%^ll@-0hF#2Cr;hS}UHEmxsZTxd(*7`x>%;c@ZJ)xd!?KSKJ2koI
zVTD?ZLVE3`zug}skspCP@Kcj9I>mu%f7s2V9OvYMe$1&$ap)E_2dvv4b?d)?-eP}_
zI*|D4Vi7}W)ml!eKdxy^WDR{?5%1KjWy&R56;+JbLPl`P{-n(o1|r<oBrZ}F&GHfx
ztPJ>5{b}ofoHpatVvQE=AO@}#nj)zb63w5rrG~HOPWug`N&-M-ycRDtX|d)GqCWof
zZZ?(YQ2GaKlnVOJ4et%khLwUN@{n<si?@jr`7dh5v;#?9p;KNP9_tX0oZF2vea-)0
zwsGR%2dA-t?MPJ~xErS?9_ZU&byzK5XG90~s6^eJ4rI}qD~weXI*Es_P5SFx=p_VM
z44guvULtj|ZRGW`*Z!s;>69mFH%(^;yg`^zkP4$yY+Gr(|F)h<hA!R{TkWhjNF<JH
zoH=aY3H!SfPdz)fdS<lUh)PgIW8eUyYw6ycS<%uV=kL4oIFzRBYsrL>Es&Gwv9zQ+
ztu&`8xRSZzAKJV?LUmf{mJCr1Ir}Ix9q>dX=!pdTQ@E7;V}}K2Img6T?{eyD%1&`w
zi{q9u3$G~qr~Kl@hY_zIM$u5^4dj|cupO~~ZsY^eUy3@krA(iCoV^L*&uNvouykTT
z9yR-yh6|)eADh_1U8YV|ogA`3E%svk*XnS}mtr1{ICa%p65S$M&i<`yEscPzZl|2O
z#4|#Oh6ncV)n{?>%K4m5Hwrx0x+p)ROmVcTIgvw3GGbCQ@3j9Yh&T;7D~dnOh$kS0
zr-7SrV!9VoDt^uve0BaaJ2$^=cm_u}u!2oMT|BdwxD!zm;XYrZAK`ar|CMt$HTpp6
zm?&$b-!;nBW8?PUiTSvA!lZVE=Rf~4vy+Y1Xp~a39P(DD36uZO;V0i?pLR0GnE^Pt
zgM|SH9Z2+S5z_OFLL#&oPTHr}7sOFYpguhJh*KBOXaspni)4fDW&`_-f;ZKGVpJD7
zra5q7EZG^vwWCKN1o<=DGqII*&gryV$fOB8R!W?dEZK)kWwqBns~~bl5SJ>^S1WnY
zf#gat6T)@zSQ2+-ih~QbXVlmjEjvfOJ>b-(;L44xn8wL(WS@Pq*vKI{$d8T<oSNuN
zx|8uTm}%)1_Sxs04C&*>Yp->FY7My>U^BV$KDQnTykj<W-$76X2cmYPiY6Je&pTOX
z`)V#DqEP4=JQ`Mx(ofjupRCO`Gd_tX*C`hf(A5RWjeSAB7_@<*aZ1r5(L&7W6ZVBC
zYtN8K1YJJGnq$pW+_Rhe#GY9*`M;}NL#prvxIAlej49Kn|H~JhtRsI+SM0DMsEDPJ
zv0n*@GP8^PK{7h*SvikOriPrKi7n_2;6lRH{n_<+Qla!l(!hj=N;d#pv?#p4mF547
z^K~~|bCdHk2AQ4A6$~cKz9gr9f}41!okC4l8Z=32YR}0Ay(#4-q#$OfuSNUPdgnqi
zSUG*j&tw6jbeD^{+%KzVB+#LM4j)n-z<V7?&CNn!U~+Ei?Oyxx9Ng%5WV)Vp%E48O
z32DN|Vh&u)?|%D=YNZ3>|DZ)x-_Y37DsOmQK8h!ha%XZ0ec!&aZGqdlR4h^mQWMk2
zeuHw#q9hj`0>tgP)%~<d<X@?CH6=200~fx}Z{t$h`|PV~p0#7;QwSU6)Fmk&pT+mg
zzPg$u@si#9s8bhnw8#%!n~FK!0~c`EzNWgD2n(9P4yN-RkUXZQw}MRr``Yg9ClGa&
zTllC`cQswS^8fW8`?{LsxMGkg)v1t(#v+6O1&TYZ#Ij=j+c|5`tC1||0(6KkWHMTS
zD`)^bdGbBHSb}1_*uK6-Dg|!jL4ig<BsmBg1RoI?Hb@+LdxgL5vFGP@Sh;S6^HUP$
z8)xhVH4&MyM!rHvyz0>1Zc?rzTmh%-l4>U@MOIE_=Ao{zp>wXF6Ue9b!dz`=9cQRh
zie05+Ufn#1tJRCTI%#);BXR0dL>QWgj6wF|Zci*6=NSp3xj5>;<HaptUx8|uR-JY?
zA47L@(z`l#8Uj9q|53ZF?H@|AxX7JcsH&h4sZ_KYXPs7cm}Cao^xlwDep1`ZPqy$D
zUHVN9F4>BxObyJFk<9_lNr|PTFxN}V>}4;>j?Ql*!b|==6QsPSE@xnDMhT1tc15n)
zP1juK{46@18US&GA;H*{ZO9=`M}&d5It@(WaCo8FORHT?VUQ~7+5kHRdgWM>6lO2$
zrb&FoVq?uB<-gOaR#>vcp83b12v@Zoe=CxtQ>xx~Sr?cMTHDpt1I<nIX$aTxSdJqO
zJm~@nt+=}D<<&eHQrMLk9?qsThY4`t^&!&3*NX0u>}%Jw4qbiYjW;=^#2pOlc;sN8
zI&IfBHh`cklHg$>cdDWpb0UI`{It)mYjtzmgDWA}#i`Toz9sU3Uf&uZE|H;!J%|@_
zqp=WJRK&dz1ZunoZpZ^=#Tc43e&<MsQ118b#(GQAh`hhF)oHR|URc)Cc2nDi>6giY
zSckC#QRkKNCqlTz*1w{C<d(HdZeQ;dYY%7iW~UyoSJp)5)=2JE*Lez?lnozvJ-M*^
zl}_PtdsWWG{I=_^yWT0p1d`E6>j4XP(q3J4<ST_d>I3_ox?Hzr4VF`ijt+ZG+Z02+
z4I5-dph6nHyU=tX{wfOWwe6(BM?<c1B)jeK_Ar{BJQ%?$nTJqNNgk%ozM(y*37?V<
zf|}S5Qp_qZKX*USK@cdH(U!EX!*vU79*45GI`KXVl_D6!mbP~CUepkwtJKwu$t6v~
ze|S*CV=Sw$rVViKlP+03EOdEAJ8)qa*DH^1wB_wJh%7h;L1dI^N;ka$$<E_cN|<nn
z0ulScZv;PE(bk^h9BM-``XgkLe4?x;;jm;kpRD-4$xRc}6J(zoaQDD{eWzZhxL6Rj
z;w|lSn5(7hB1tnHky)8j0JZ{n9lER&4AZUkz5G)8HN<ak5b5et=UiB3E9;p&Pjel^
zInx`YF$mkSp`sFC8^131)WlZOdh<KRn6jGKE(is5=eXV0u=H-4QK+Tq1PK-hVbNAK
z6q$r3Z;%d$ntUZ9rNl2KWtZ>Uep_7-vG|JPl;c&r#Tz7i?L_#3HYu^w?cJA^FMwyD
zM&e&iI*?e+D8+ZA&Og~qGImEpvG9Y6bU033JkzEt)Iq8Fi+tE?5(7pjk(Y_rq!?Ze
zYuDK8bJCy+tDWCs55_jz+OBt8z(Ma)UyFO11Brc|W`gjwv~^Xdji7`61ZoB9Qhhix
zZR_i88j&)wq;sxB@&-{rfgD?OhhHlF?(BN7;Kz^((FT!dFJqKK=YyBFyV}esl#REY
zMcG<)Bo>gyY*u%a+1=Gnq2@AVAxb)RF@p#~VVufTL5USu!R-yzAEA<bK#xePB)>%m
zB6KHGt3P!gYOy!wEZ#li{93ds%nlAcx#?m(?M=B1S6=x_=VyL?!RsfGf&g+R?XbPM
zbs5sA<ZmQ=_Ey*E8?Ib-PrXOPRaP3A+2P7#kDP=h0$vEOB#)T}C-=o$>PdA_ei7my
zT3ROzR}MTDKJP@60DRQmT5r?z%PgR;VX0#}8zk)_`cBJ;__lT&uUo%lty7BrFqH~d
z;XTrR6GLNfFNmB|6kj2iRc)Zt4kTrYk=ZTd)~j}s_!Y31Ta5I14n*y6T!Ry_4egKz
z5w>j6rNvHF`#5W!7#$?ch7x<Py6W2No!{9i*t+!BNuPz`niB1g^xiq<)Me+egb%D=
zy!Wz!^E2j8S`{u`l@FI~Y;zis0bvDp>dI{LDQnklLy@PE>Qvs-AbF$dL4YJL3B#f2
zbla5p89pJpy_~W(|4Uzdg^gxsBe4IdQ_^XJ1)mtMd%H7(gd78>n0OPUpds;TG+>(x
zlKPt@WJ&yhlzn80jddf9*c!0>lv9^eHJGKFTMGX&QA1jPl^$ANm56!{Jmych7I#^*
zE5_@Ya0X@V)fxs7YhkDs9eDKm>7-QmQB8DT)9oSQa-}B`w4`&|<-jjtE`qs~tScgu
zr%+;;w5-yrZl61AlR3`ouf6Ij=Xa6Ev+xPQaoan(UMSar&P$1QN?Qj+oZo8BReY3f
zs$*W1nKF#tIS`r$vxOkF-D#Bzyk@Guh2)GjXs7lRzN+X<MmNayX56XQdd_cKHalt4
z-Sp{Vdl?K4J9RZZx(P0odMV$`GzfB(WX~>>o&zZ~hI%q_aH)1*tCN>Hq~qM&ZeOU=
z4KWfNci7gp1?5S&d2|Z*Evoc7f&``}hG)_ecYj+OcEv(4Md+b*G&}I>f@~9#d(PQx
zR~NxMP?PiC>hfYVDP1112kPA#Y)cTfSX&7^-VOdRaNL}rsqd^h;h}gx+-1Dgi5e<n
z<0<UI-<TU~_4+l=&zRY`77YXz;|EHkoPDq@)~5a>!k@(KnK2ScQ6C=`X%~}-nZ({z
zZ(+MZ0;Td2krj&tq^2OFkqLgWZ(3-b?ZhQ=B0uUtqBTg%1`+GE*Ssb|KJcJ+8esTB
zov02xMxc{Q7=V8gZ_{;L@9w5UH}i(<?9_>-Y}qno?<sKdh5Uu3p{`!09N~4sNQ-U{
zRcqZyp6Q&oJ9TjurpaLOeiMUr!QNZXC=rWze2~+ri!Zd>yClzK%F*|=?n1uFV#HEh
zaO%=&Ff|S#g%9kT>m6d?G)+(NSED%S4Wh~5RE^M*WD~!oqt>62B}6D6e7=&*(@K<X
zYj2EW$G)|Z*46zHc+jr6j+Ceadw=y;hYF6Lx|i0eiy48&LP2y^$|{s*LnXHTK=+J#
zjusZdle(f{T^c=}q1T|0@;OOMCpiRlnSEOuIdddm0y(OL6~uug>SzNe^4YgH)|hyK
zjmiPz)KvfEiy*z8n8r>xx9@0A^oEA|lc+qLns`Dw-Pnhvda00k`(P(SC`HehQ`QXL
zoiq5MoI#uh5O9Z`q7EAJJWtq%+fZT<i_CHkB<W_K+_MsiA5Qv6A}&5aAqJ`?mj3L_
z+<E)Xs?%HGCuMPT>S||;sR~vlC)B>H4pZsgPB^8~({`6zX!<}{p}$lL7_IFi?F-Ts
zB*dvJ^p|gbGF8E%fl|5~=)3F9rOgg$Bh4KmWV%@Zb)M5IDFh}f`DL6zHS^J~ZfEGH
z70A`<B$mph?IFhfJ=IP@twIrbEmD_k34(3(w8<k-Hz__t(6W!UWt#yr#)iEsZy}ms
zbi_VhV+&&@PfCVjK1AI(ymvY9JhGOre4~A$y=-Yu!Qf6=L&Db9CuS5uhEzMBY>P@t
z8SIcG1FrSaETgcLpw7EbW0wEklTTgG<9tsJR{FLM`z{KUdeJyC=mZlRw#GDm#-YR0
zvF<t3GN^|$AWF~gD@fax3Q9@_2uQJVGL(oGr|+1YKEkle7mwNZ7xoU}YQ*o{ZeXvJ
zVt9hv4<v1U@RC<JKNI;D*=9A+gSM?EHaYspco&DfL{u!#4m@>C$nb<iJd*RgN%w*}
zh1?jVtd+x{ZSP(td`de<Vy(#G4hODPx?}yVE7sbM>J!3QOsH}-B9Wlk2XeV4JUKEQ
zAslRMrE(CWXG|GwC5Z&TW;@&PVLRlTpzN*GW$&OJQ!~g{yBaYeJ_(}&DPO1&ad;i>
zC5&x%%3tre-}$BbSdF8$CrdqDJ~Zu=YCU6Zv>WLPYiE1gfFk@gRR_|uJmw7&C0gt9
zi?C4GzV?hJo55rPXQkO}kaYUgVqp4@qi=s3Y)Hj%x_hgWkf5|Q3PC?mt07TJw;n3P
z@h;XOjoXJDxTqhON(}5^TRF&jyFMf|1U#V>qQDEkP`L=#2~U2s>a-k{X49$DJ4s14
zutVM0xSoJ#q{z*rH9*q+NHL!1igf3M9ZpOe=(b_@BgKxIDankq9edbgN_VWdyCZFr
zC*SInV_m>EC`@rDp~ro!wH(7w9kit!Ek&Ir_$l{D9M|@EH==Z<JOT?sT`Hx9v31Rm
zD7c<h3Ih3S>TfumIS{qH=kom5o_N~$W#++h6igtHn$43TKN4+j$GX+xXl1riX|E^@
zctJQf;PFi08s8;<tQ~K4frYWtoVq$vllKcPQAh#W7Dw$wTj_;kmM?^|8mu|wz-xA-
z2*}w~Z}!{CHlDyS;kTUB5ceHOd{g35x&iC_k>gZIy>_aeRC+L7+NT8Qq04n3F)Ag2
zRZ{fxI8&c)tzk!$sa^~`J(xS~Kxz(-v0IXnI#b_H9hyRom2sfkO%6QRnWWenk2FaN
zC{IOTyoL88G~3xmA=dSkXMJkm)QA!g7e!hlJC`WBM$samD8*VvnWRGrW#_XFTNnw}
z;4&mnQI{_n^fyh}g?tItVZ7MJoMOC0;wgE78S@$Y!LE)72gb>6cIxtkyZ`?C?T4~(
z$xGk-mUlXZcmv{0NN9S)T-EL4eP?rCeWC6V=?u|zowbYYS*Sy^JWwZ|eCcf8<{R?k
zB{+nzK{*H`sw_m3C+lGyYL(H-sjH)F!Ea84M0_WnBuq431NOrmD(PYn_`W#$h+1(V
z+C;0l<Q0dYS$nEOtz?1DpaWsMar{ewIO)IxT|@b&w`=pT$S3Vb+R3$sHRwCV!s|U8
zCt9rL>i?r{kP&e{=9Kf`Kz~6u2KHn1Xo|B<w}6c2K$4K@#yo|YPi*Pik9W;Ueaw!R
zfWjVz7iy3+6X@uYt(6p@_7gehtCp;DerlnK;m14jCu@qqTX-Gq@FPxLjwJb~qGB-6
z_EX)EF?$T0Vxd-Cq=3l%nEiBDCviRuCQe<vP__dS6~QA&!>3TVpJ^-1_EX*?d6Km{
z`BR7;<*2uxExb_B`AMe~-^GD0e9<7<_H%6ytl{PWuTrGl+iXAI_33oe#5r}j2t-g3
zRyS?G(C~lB8r+hr$C{8dgp2=&#jsy&5*l$8I1a(E%iE6H!ji*aTH=p3++O>oCRT&q
z!=n>hW+#SbWe=i0$^LR~q{F@7m)q2CnO=VT5~mnLOplD9*7hsiF!J`c^YEqA#fSAV
zQ877tC&&A%h1e3L<W{6BmKJg54!pXo1aOO~)UTa<>LWpu=C}P%``77P%?mDIj7l1V
zUoVo71{|mvXI>jmC;U3CV!Wx6OCdPq68nvssg$NN<`}LgXN?mYczjdynhBhb*>869
zo+*Nk$5|%786YKJ#qF+=QQ#K1WD5l9gi!mfhJF|fd$V9H6}6ChAu=wKa2^`JoowS3
zFK~Wo=c3NL>MHx4+|f6^{2J${m?H97j@a)uG+uVcI;W6ptWCu0(z8iw+TZISc^8ee
zH6lOwMHeI$iLKA13AW$w=AK(McpDLvPF<YLA%n76<uHewCms=z{iOXty)BjHI+^9B
zs5nv}b*9UgBCeePKjW5+2zmO$hQIJ0vYX@7$T#J7C!^e62Og^;g_u+~r{Ea>sEvwQ
z*KSA^m{tWyZns<D%u#0iaocW#-UhB2rx=TtWU#5yFGSXxoI0-Bu|LU`cwnRRODpl<
zWiP$t68qDJKS3{><d&|Oa_9$Mt3bXA?M}2{yI9~0_Gbm5_uy<EV<B+(0t7V`t(CPx
zwz}>1=k<){SN0x7=H?0BAevvf#>yo$`J=z6xk~!58<)yQyw#~S3KX<^2kbAq72Xa$
zOJ)9uQzzuQ=PmcV+5W0VDzYdfO|y!$2AOpPc#M<-JNQm<aB-&CU$;461@&f4s*ZpY
z-XM@y-nkHPh7Waz0peW3#+Fz`D~XQ4b6%lsI%}oAPM+;=b8o!!$_JgF>i+Gd^|!yP
zQ$+0BQKuxUy_^x^dck4)`}$xm;o{*jh`Id^yb+&>kHC+{{$XKi=|p7ock0r66GA`D
zO+XkC{;^&yQ?ara>K@oMC`g-r+8bQ^O%}r7eNnN6;RsIrpW0Z%M=@t9_)7;8`)H9`
zk2j}N3z?$qpX<pu6<FuIfkx_TAn{}zyM;-xIqWdW8{6$)+Hk=n@x1(X>T1qK1B8RJ
z4vV+_*Mi6brO+>VE`>tRf#kH!MX1T#ehhT~mK$hQzeoH(g*=N_aF^Xkf6V^9wqt4|
zIJH>W#(*tcIZ3Nj9o)UC&qM!_t2;8)U-$Y|PALgfFzmZgl_-KtC?@W0pZ#Z|$e?$h
z^HZBS^19f6CB9tz7>mO17;GITsa;jhe*5o6W!K$#yHm*XfmxH5OBcS|sd`lWcCB~3
z#(_k~gi>W6j+6>W`ON0y_Gzc`ZF|_++&DrKiHDrJnswZ$d0Wfpve!PnTI;IXG%}v{
zR+k3RP*&Jy)K^0Xmrm8Jj<Rd8QaC9CFIS&U6Px??nXNgBAf=Yfi&`aoWkeY85_%S{
zSr+k@)?^2oV38#$(6D9LXVtqz_(YVI2R3f3OQP)^9IiW~j-=zwuAwlrXXFs)(XOEt
zhv(`KJ9UXM@`{>-o`3eKpznC2k|T!(lqF+q7~f!@(=CN!1w@V{nkj-7h-kz7wx##l
z=T@soG(;;Udy4b`4kV|BXcGLeKQA8|?&GS=?^sJIi*e6q0u&*h7eFY;>+`c$;qOGi
zjZ+Rx(MgQDTol~CVBsCChMvX&^~vxeLyFw%z(XY3#|d}TzOcF}?7=Sq&9BtrV3q^X
zh#$P{N_%E@as91J);qszlgo4$h7$gE3P1azQ$?$Y_3ZZ_kUvRRrGT@D)3s;iK#1cy
zi(-4+DJShD?pv*T_Usy(R7^T{@K<)~w6U=BA(nn|*WeVK8|Gmo4eDwyNJmoJ3|&Ck
zhwMvgwCyA(sDxK5BL|R}Ntyd)Sl~Rc=d_`6x~NWk1Lo{@h&yo2<P3S2dK>IZtMxvl
zB>7;bp1aHB=fEp=e2UA*n3sWlS<dU7Ygepuers&76T^M`^6tw@jGFFQ$#*NTVuu6Q
zmr07ZV#RX%iiQOkV%RX+^<H&yzQp=QsJh&pyV=R)5YqKG^d;$|1J{bbm^`V^tyfB!
zA^$KeJibtuQPEP_eO22xOQwcN&OPB&XbViujLg_qH$+!*&w*1cWZfRp_gUgmzNV{-
z4DC$&9;Yt%$NF2>u2^AT+ch~98Z|}iQy1*jV#{$8bLA5qT#S8Pd(qge_Bh2r{B0oq
z4trj96url^^IOc^#DJ@M@LaU7@6bdl(?uK5fy5ZJ7*e-0h4%cejynL@v`UviUG0(u
zVM#W0Z7#F`_JX#67e5TkLMf4tZ7)mQ*}2_zN%cXDkd&1Y#1wU~-GQjL7aztZOyHrt
zuziay+(fcP+GlU@bSiI*FLq=4B5~(MtuM<_U^pe=;mx>7t*{q&6Cq^==Q3L{437>(
zi-f>BjE>SS?G`B=nMwBS)a5$fGBGnYY?pP@6fOp$Ro?1kNK<^riMbtid20tUL^u)p
zx~>k>wq#3^GTBR7yLspyb@062)$y!O&LbGLuc+w>M?-p#eTa!v2y_H-IPjozS6pIO
z<|`quWP2DPy~9x7tB!Sj0<2X`-vxVVo3Zg8IwneScI5#p>j0UBSxFHA32SjSFgS!I
zds+Jes+vCeu6184b$M#6zs;`dI=i5a$}Jj9kpmA*QJqK^E@O6eI}~s>;5vGzoK*6<
zmtSu$Z?hCx@4{)tTV2|w;*LPOrnS@DLgjRkP^(Kehfm6bxH_l~NHd^o;1&gwI`CX{
z`G`%}bqzN;A(Bna`?HIo^9A^VV|pq>(_#7g1q9?nu%R#|h&6IIZf}Mm%2T)nqJ-FP
zH`J5bZ_=Qg2j5|aIYoe^O`P3xlK;-=|8`@MlpL7MopR=t@j=`QOcDd$@3a!K+;sH~
zc2k??;i(y?5bM&69~>3-iq=DVP<Jr!mUI$9EcFF==BrmY;G)E>_R6*-Y?Vlyf_jF!
zVB9)cA}Fc+3tpAo+0s)a7lmk68e<&G>NK=h7ZK#j5maMUB7N8l*;ZbYui#KVH0>0U
zCPC>U7b}$RwOKd6O@^f@+@<hqp-$L~v-iNhq4iitD%dG1uuzu*_x|}OY)LnYNKVM{
zDo#R1Bp`XPU47&ASKHEV4G)LVnl2y+d=8}Mn{{DzF`wA7QxW77I=Xe!gpaMqaN2jO
zlh=itR(!Y?pwc53M0|qK>LkQBdFthDAk?8B;O{P5(T=)h=;*j8)TxL*$mvBQmeQr)
z-1wLbo5cx@8qLpbI)HCU>{$=d;AicV-~?{1cS+C^w<A?ZtJ6G65QV_m$~Nvem&KAi
z;nWo~e=u~On&<GJu-8><!mD5fC4`-E>XOBja)vL-pH5$uYp$4{+v*u`N{kMxvcst>
zWe1Tm#K5j<y@eXWfrCm1oVw@*`v(Fmf)0U=D9zQ`dVbqrGcFOMLtH`NRTpO~2C`sW
zZ*OmnJa<yf?DwuN(uOIQ#}2!rc4`Ws%mHHyg(iQwxUiOxR*6aL`q=8$v=I!(X5<-8
zIoA%|u~cxfEWf_$xG5K!!0pm7!Bq#Bz_qftFUfCZZOx$0WW~BDgB9No;Y-&B0ngbY
zIG8{PTbJ8v$*NU%EOQF_FvgbQQ1!|^V}18R2|wb#Iq*5^AsSHay$(DM2z-j|&P6zj
z{Ygm2!YQRzgQQ+V+K-cTPFnW68l_q_vDqoqyb#H5cXzmxp8Ya21r9opm?3fA(WyBy
z47XVe6^~45=bXA^YuAxcVGHp=L-xiRkERt3OvF1x(t<$2T@JiCC1J?MOdn1vZ(1P0
zVVsH-%JN8HM#3Kqo#%O{RZi|S*(xx%yt#WyJgqc9if=jwv0uyr(W<tH?z%{;Us#Ln
zo>NafkM)@U0GjFlpZ}b8WhOcmy_3EvH6&t@6*#1nrf=z1GHr=+FG>ext<)h0uE}S0
z5pC_QZC&mdY;_8`FWJ*|ucVeQ)Vv6Q_O><z++cJ-cljF}NKR*`TxTHm_HM1XU%mF|
zLfwg&22%W3uj{gIZi};%)(nl`D4892aGQBf9<vSA54n#pm8kyYXT@dW4kSt1`hJgm
zRzq2w8IQ|Xy^8E4%p7AjT(1E~n8`4*@fUY32U3)AXcKmfep~R#p0=M;ZgeIoOqisk
zg5H?ZdGUR7J4T&Cps!4QN@H@ABXXpMD!xagO;{z%mO_^I%^mj!PrMCDizSeMOh}iu
z>C{tK#p;EnSjq^Qax!zARvK16YC__c*^p%dPn_q(kge(ynrWle|5U1Kn<~sn&46T=
zE4CkR5xciTySP}TPRfjUlgT&~+)gVIh^w!<`sKE{2AL~K_r<%g1yK~s4Z(rezE-$F
zprws<byCyrgMz=?sjDrc?75<HN`I(%-cn!C(sh?#G7gzz6)n`@g-cQnYXYSjLJhIg
z#ygB#FutdTJRQ1#=N8}G*VT$P(Y4nVE?Q7=LsCd;36FY%D;6{25IF<Vm>)}IlU;wc
zR5Hro(3D<&$hG@@0WNb&oOnEcMn-&*E_+8qND>ori62F&ik6T(d3F_E$Tn5{zr>BO
z8jD}5HO%~qx-*eAfE%K1hrKc~=rVKtJz|4yzJ*F+Q*gNnVIPQAmozT}o351-iYnZh
zTyEa##3v{}uCh(dG%Q1Uj3`RhT4t1E4kYM|=0S<8H5+gSoY|4Z!S1X5{?OdMxbls%
zzD^*)z_xbJD&|YGOWu@NiPUTX&y`zoe_Vrmf3;85g@TafEh)SWzbyxzTYVDGf+?GA
z9M-x%*+y{~r6OogT*~+SH`@dCzSQPP$k(Vk3EP3^`_@d1jt|*8Yx9&I2t-u)SFoJ(
zM}&@XJ0`92p|vv`aI5^r#=$9x{kYUuUCmu&e~D}c4|cN)6^vEVY_QWDND_FGuAi`X
zwKbK*5-THDCmF?n<V>w4#t#<YoAMOJ8n2LKeZ;8>RN2cJj;JTE!P<b{wrhw!Oel`N
z3w82>Ou_!$^#*DGrSF!dM98++8zdMs>J+vtjnxHvPrX~>sw_m0K?<lVWroDr=Fz9i
z8$uyr4;5TfpjlnWR4JD82?rh{*#}z--Cc+pv-fr*375KXgQG4Pim}MIG}#-r_cg=_
zAC=+0Q*jAM8OSGhj8lOKx{Dy0!c<%78q)3d&22$9PQeM1Y0;^P@i4|<9d&6K?~sfH
zBYM)lCE4`rZo1m}DdbSDEW!irTf5mg0Jo7!--S9+d<mLI?EURzU$uU@Qz*Kw6IPTI
zKjx|nw<t|LoD_A{t<K1-1&EVc*1~TqbVeM3;Rt;~?53a=OhAe*w2HSR(x)CmZcHp@
zm{q>L79%N$XomEApqK+moLo9FHqFWV9gS0j=_lcGw^Z_Wj=gUm>_(Pyxx&k+%eSnX
z#^b+dAFA)=#hN6AfSc6c;|?Txs-8&68TZda@oF#Ghua*{JEe^}jg&#1<nNWM*V)5u
zT{obr;WDNvr>2g!eGr-;!S<c?kzMl&{|)TtuRf^X=Nxz)bNe8H851Xxa@@YF4If3@
zNvG6xmZ%a1fD32Z8E+rS6<@b?lndS{#7m_MgyJa<T}&yOqwns{^V6cf%!4c#3fh;F
zx6F-BtL#*qy3rlpvUl!r`)GS}xZ6qolOGegMjeRg^!&C<Uuoad(C_w@Yn_5r>uepR
zsK&d*KGwa2SI3K`dt)pPBxVp&9Rj>4!pA$bTk!7&iN#iM{FD{hX_W>PLVUR1bzq-p
zxWOsG3B;*=)TxU}e0s1*ry{pcwsA`@8N4N2XgGN{NSDB!(^<r4r+sgZd+h{H#rzg_
zY(&)lzWRKwHL?J?l&H&PiOCaHluq+ejCbGP*1Es6H|vzzT8kkFlw9NmVn5J`JYM6c
zoKnnL@Q4f)??2ntVO?ZZ(2*R=RDT_K&erJoeG{WxVEjlsgogWUJ*Sv7ken<y?N1}Y
zZm-T~M_`pj(%Bii9Y{HdD9c{COf2ej9tJzAOGy_h8wMw+oM+|o+u#}SNl1U}@-9=6
z?QBDm&O29y!D9KWUI5R1Pvk9bg0`!}bg7PYxTNV><hr|{*lAT$Eh$<~ZWybA!^@<N
zyBk7cx)#AD+|{XzWvrRdp^+Qg_SD;U!{E#N>tt{#oebr`HSJ^EH#1B`S<m)18q(V`
zJ2H%b;nd{T3l&x}H_{eAX8YQvS~Jmi3ehZ&JTVp8hZoSs?ynxpg+dx&rORn`+UYt0
zlmhuc_lDG%i;Iyb+X)8}Jj6qDkBraS!J0=MrHGmE*~;j3h9gVAlZV{vv}y=N?m?ul
zN841OoDowf^{TYLGNR%u?7(ZBd549FZ=i-9Iu$O0JabHw$k<P9bShJ){KD6~&<^J|
z>ydtmpZVRPN*{To?WB!Nu~Uk#q`ESEROt2r4ztG=4L?liP$DtiN4!Bwy(8qL7xDyt
zyxY^6S4o+DsLKv4m5~8y;b?Y%)Xd7HK&f=$*IZ*yG!%xdm(oj|U}l5PfP3I-d|!Wj
zbjpskb%cIN|Dg<F)D^Z$xl$)=Of84W+~an<eIO`)5)LLE$A-8~&Vh?Qty#9(PSoqg
zIO_mbBq4P<mJK~=H%9Daw@7NFMO09Y7&st}L~)cQbC{$qOq+J9gP#o}-)XpfJt;;<
zLF>SET7WA9JDt7yMlKNHXP}<8|3N#`w)9=2<aM5Q%84@^I6BsZIf*l8+caN){S8h*
z!@!?$s6VVL1J9kx@Q6hem#rR&3J~BhkYLpjrYT||-se1#Jo!d=kDi@B_0&gsW2fKP
z_CM*D)4MnTkZI(mroVQfrixu1)G0zyB|STkJYqR`q&AX7{e#_Vi7`oMrsVH(I)K+U
zN;8~vjgo%$+7C4X#>FDI^x`>6&7zH+R46FC&7FEmN%NEYTKxB=4yc$mg_-p2VxB+C
zcwy7?QxzD;wh<)`&PBTd<qc=eE}O4$2@A!1LGkgO&%R`x#w9(!qEnwtgy{{7wWN>}
zC^ay{7q$K2Mn_gD4_uYlM(9KBsp?8vH^d_AK<dK$y}?B$+cw3Y1?h(^iAT(jv?-!P
zazNpoJTEVR27wiWRhPvu4L6V5k0w=|-^P_Q8NksWb}H3z8QGbp&}G_>)ysw8I?>L{
zy`M3kb|5iKH0!}?9!9J8<DKnL&UMTadU(zhX((_skZQMuznpj4iac#1PwlAvM7>PQ
zq2u)!%5v(|pf^IkS^LS>fYr&6zSB<KrRwhoLe=?y-9`3Ojk8#gv03?}afsxpDt$Mt
z0^`cjb&;U@(;Yz@izKy9MIon-=JdkIE$}mowm{iu0>zcE9`@LQ=aZqz5=*SrgpS+K
zwoY=o>5StFSe{YpZbC|#{M<scipdwUN~4UxWk0tFrIWC>{3Pj=1U#ORzgYAZBh$|}
z%0wDxf*~lzu8@*f+LVQKc%rsP>=&~6vQ2nEjP{+PNaNT<WmUCbY#Wc8nH8)>Io`)1
z39ZIOP8yac-vcJ-ggI@$)P{j%kp}LaPF+mR(*~`rWWSu0eck$X&d+#p7*BawXe~K)
zq?R~_K>91~#o&B;g&cJ1@)-^u&PxWTRCWrf%{BYgdWIbY^2YXd>dNsoe3Nx<XKTKb
zX3%q-EWehRdf(`M6ZboX_)OZ2Bp*vw&FfHPa{l!k#qFb;ou4|N=C_UX;c<SW1}<S#
zv1kh^6oBMbWBX|zb9~xw<~P~FGs8|HUxdfUz<#Tmr-uY0Jd6c((JSra3L9dG-C1tG
z-A$NoM*{cf&id2q^yCivovxD$5zWGreGVjjp0-X*ZC0WH``vCOGxPn0;F+<fIS?I)
z+zN-ZFl@h9A5V|GpkPj2zHa%-b<0*QS-IMNzm53Pmt5u)YD(nnC)w#y`-5%_;z7B}
z3eJK$aR#G!z1bf&?8KbPhn;e6!mvRkQ>PQ=G5e#|Ax<SZ?_xV}>I&tgK~AHK80rcd
zwLdOYLG%>!2kq0{x(Y#&u#Q}zTog{LhTUL%DW62F?N4%e>(?$>w!$go==-F8B4~En
z{xrKpp3qIB<SGqS9(6kBNQ^=@EavggvOL{Wa!d$1Wlb|?V<p+LKd&AxXi?O+l;(9a
z6+AiKJG4ssOK*sTqZPqne^LET^A9#1>8_=LaUjW)4tn?7U)IEBJeeGYA9w0v$oXh^
zLTmf8V1LyJY9-QeDr!X(tfwm+<igFVY=2#lI(xxS>4A~Gyg{TrU@OOr>*P6Yf79N^
z*&}I9`rUvy5D~7M?^wHf$@-Od+-`rHn{xBysPi+&bLHh%+TYbHNCp_%MwtloqBGtg
z&<eC}B4??mJYs)eh@xpu5l1$<xU>Ui0ypXQfg7%Rx&1@KNiyvAxU7zH;I$@5W>Z}^
z@sCx9?m%h^UUJA=9nvKSrm*&;{ZqBWwvIsv=7OC@U9v)oAVXR5&)u9HjZlQg!BUA1
zr@cW+3J!rW3FE)CMdbkEVoO&2s8g3K1=}1MR+I%uRL=gjeQ~%qLRMF&ob$PAW`q4(
zdo0xENgQL#Igs4Y!B(Mnc=93p_nJu-Em(}qxD{*{TlRnhPgMcur&RaBJK&m1Qa279
zbxZ8%f7F}Bf{1#ap4;Wr6_I{4*uNdLj>OEYq!gO}*)$fDxP@%wwwzg~D))cjULrG}
zWbrdh=D*tP$MH($L3j!3>NC2xUV?#6zRpHw|DAMuHCKZpkWlKva*<F7j7ODXr(<5`
zx7|8Jn%D6W=~Jka;Pn7uS)Z0*-kOFxGB|jvOOs&iRr~a=U1Y*QdA`?M9s4L0-2{c$
zXPhp4A>kN~!%kTz3!c5o(EFLEbHe2-(1kQ(c2KwrI$b9G2qyupY8R(l8tFTO2A;Lg
zI$d)p9F7HpMS+EZI*OHb;CYU5Ve<|4jOw%UU`aj4o_FeE$zy4gWzu^PeRpFogK>_~
z{Lii@#nCG}73uYyx}x9GH&;MTuumW$tkCCl&pgU(q$QF%?H;^|@NFh<;pd(%k#=%s
z%j9OR6QiP>uRaJl{~#8b4|05bUVSR=S7X@^vqeR?q(H&-$ALeW*UoQy+yBQu_W7r4
z(<~=Ui9&3RPyvH&1P0p|baN~#op4I-5z2W2L?<2x-UIf9)he~I>VgDa<@W^=vjI7l
zav6=fOezwA5fkQi*)tm+T-*j-J?NAd&3i~|^hC-GdA5l3p^~vLs_(^&QYRCJ*6oM`
ziSISfaooRoUa1_g&GWry<;ruBJ=c0TrD%;H%^UJ-d4YMZ)wA0=45nw}T0bg;IMX=i
zj}PpNPlxN&{5H&H$b5)6aaB3hi2-m11hX?+GT)p7`y$hjeaY#klt%BRuXcXvb&QP>
z*mJsBlzl8`k+p!jHc05Kg9>Xc#3_Gitr@g9%yi^J^H!I(f#8Yl%No{&BbP+9olafq
zzF+)i`|`H-Xc1-e#^&hMY5UI%_3SIU&qSUd7W0Brmz(Im(a}-+$~ILS`kM&t8g?oQ
zoA6j>{3JcJ=hn=W4H7(&Hnz(H;|(6PpA|*H#N0^xw~Ms?XYH#_mx5!}!~_8=6zjx4
zEa4gEOzarQVqcxT$uuCMs#FwS$ppo&MiZ_iE`(e$b352PN~Q5N?fuK~#2z?B4T043
zc(&pTE?d^u)(|8Sk^D2wFD$OCBuL;6T$8#1V~~AaL++v8_$>64a!iSCn~+G4a?GCB
zP<hcQuBAWXKvJ7Ciqq4;zP=i2_t9UlIrbxa21xA02Z+9TetR7!wGvn>mz;>f315Ij
zNDw@j_9(Q@UQpfArK}h#rTTR03WsI>Ox=pUtm`x6k$XvgbBsIKoZm@@OGuPTA&<lt
z7Nq$Vq>kgn+;#_&{K6!6YLTPtMTtyn)~>jD<r|z*o%+dtDVJJsllJ23YxX60F{iH1
z0l60?$DCMyX+dH_7q7Gf$?+{vW=1I+doT-j+37TuwirYO=nJ)&Q)rlGPB%d&M?>4{
z?PD?L%9onszr1>@R3M7R!K=WjtJWuiqAE-!#6x>YL7;TyXcM`x*wGE5aY*I26Mi+n
za7Eiu_>geMg>z0aiJ=H~;Nprlj7<<Yab;rEa;}cwf~lP1bAaK;dxZavU6g11!|ce1
z*+4JN_33Sx<wz#MJ!R2?t)SS;+P>e&g%(aJ7ZMscM&698YRHXCVtYAtHDvj)htmU-
ziP6HYZW9EeDq>YJP_0f&EX~9&a`U`AKR-3|fb%mp5!u$T1u`RW;McTHY$8AQ*r-#k
z4@$Qfw1>aCru^E1KrMq?h)h&*;sZ%F@;Y`a?saxuHO(ABntJsR<aQw0bhGY2Mi+d2
zet(=yuo94*a*9<%MUh(oyWR~A-#EXdVv!b89La7*AP)%Ox!G3^lfA@l%oi-jlZP9&
z$GI(q8g24-Y=qX{bh_+Z+(j-&Kr(DmrS*B;#H_s{TW;)4ZE=2U`IXjm%3j$91P3Du
za)@;owm;@TVop=dAv5MtdsSl%ZYBjah1f2ldLot_XGv)b4PTw#1XVUVzoTtru>rLe
zow`U@ZLg`1X+ul6BSgcgQ}?H)$LzHQhVu=KM#OXK;+5)iXbr?iLII`1BW&Kjp}le*
z<rWYE7@WF%2FecA!i}XQ5TO}7#FB<HGVxB#I%Q3jTvZ0PG=V6b#%3o6eW#q`M*?Wt
z9A+VWOLlS$FDrNp7Ey$yGD0X7>k~K$0MC=84~7iu9-V<LuOmqZxtNbV%&f3XHug-~
z@c7&wr&XY`uBO}_S5wAc7|9h43rPqcmq>LL%ljLY)x>UYL+o);1DBRj)`>lXCKNNt
zE!|d#oZ&?XrEG7I^hx265Ydr;irrchofbYF)5n}Tp)Q|5!#iXv>$Rdq($t@CXS@Rm
zEX&J`MWr6G*VV*n$f9<lv*MLX!REjdGogtjI6=d4%x+7z>yGhd5TMO+CR4UAr#kr&
zTxxN)ZX0S-ZxqjEirPLe;vZXmy5t{4GIE1Rq>iAnWV{F{Q0P{YWxQLs#*^a*Uq`#W
ztz!>w%!v^k+J~LGq<~`+!y{-%?x+a}0?j)fgLxG)AMpmsJt1l*%_*vCThk3w2c8(#
zT~1x#T&_Y+Q2VVErLWH+O>96J$nTh@P$9+6+-Ga6jwfRyDXf6Hc&8@6#W&0ER~|Tw
z*w)oE5GZ=BCr>$b@yw$9w~Q)t80cn&_3hhucOX86fJI%*W>SjK8z-fR`H*U#D=F>H
zoVh#ju)t!q)hQ?098m}Y;;`LS-<Ek@2cNvT^0$;o7aX`m*+~?|xQ%1X8AUI7ckbjR
zLrNrAg(;ch><u|(AZpL~oo9s31(sJVnMl%kdt;)GSo%rsSfZFHh!rKcHtwo^Q|>*b
zha8=r?)ROFq~CXrkD`)%bM>2pJq#45==~xCy#ucU3bq!AToARwXYHPPGL@!c#HH{+
z*9=^j*#NH%!Id1$2&!&-OM9jcln~MDw+@#E*O|;^lA*JRfIFT?%VBS=ckX7(>M|_(
zXSBMoWP?%X_|Dt9UT1*Tp>QX`a^MS)V%YeOFyVx~J*S)${j<*R=zAr|1P3ozud6#5
z>dX^!JDoah*7<EOzT7r+`|=2?K;f-82<gWc@exj|niVB}36EkM$}{EC=fBGNSxa~}
zc(cdnp0MF=i8Lun+a*tlE~TOa&uvXg$==wg^&7dpm0*qs;O41_HO!r<o=L+m5r1QS
zPWwkDN3t?Hb<ui(6%IN-x8Fuu=TK_}!-wQfUG4q!82Z%o_nSI+=f<h|2r+msYCGWZ
zH6gUvjdXiaF8a~#Ynfb$`*EDQ^qV1JW;rC>+eSY%0e+7=<(NT{CN$MB0DEk6o56}Q
zIW3zVa-N(b_i!)(Pd(Gj_#mztmP5{1lbJDe4<`F$^u4c-d`7kQPo5zI+tMEG-_N$5
z!m<ycrBZ#E-=<`d{qb?h4ddC*%fc{w>{6$k(+pW94+L9EIs<&EY$B!ZD;{)yiP^Xf
zmT0oglQekjGkK<k*y%uGz`42@H#m6S(FP3V#@Ysc<Evh=-*R*So|8mt#aluTZK|zQ
z?8kJ~X@7O8YlDjhB*hh->uhp1XwS>`2IO?GjG<tOJDPyUfYVGYu~6v~VJdC9n~Hif
z3n!lXoC7KBI>dI!c*b^G!L8X$qr*dE2nsS`P@~NtlBheM?7n(KC{>s@UBDZmNBrfq
zap2;RxpkC$`jcq1wzgM>ss#SSsfo!FHzswdqAKohSSLk2KE#w`aLsO>>2aiNoo(Fd
zcz=UasI`jQtYqY~2dXYTRYJgW>T*{g97l=+n8g&M-r02~6xF<8k5eaxkSj69@o%mM
z83du@@32>$)IX*B>Dz;C(jl1A`shCMIPM)to)GIc^|p@NyV`_oA@S9$21`v!6$FA1
z6f|Yu)Gd=ZsZbclR+j!#7Yj7Frd%NzLA+s8l0Y6ePcpi6Y+{4+yAFP-wRNA9Wb~6r
zP<aW|ySwSq?RUy5e88!z**_vJ0)iMZgY5r3_MV)Y-t;tn2o$23vcby5VP`naoSNHh
z59N@1d;^4Zim^;7p^I}rWbdtheM(4rAhFWn0Z%Htag3Dw1AAYGczH3C;>oo|i<mm_
z_=ebSse~Yj_RVb`S=J`D?1GaFffKe2cs#QQ1rDK%Y_S9OEp_bVwxBiN(dLsjrOVpj
zw2Gl8W0HDEogw?|Tif!Yx*3}4qjjgIHl2>hMI@Qo5k|kidV$7)X=F?stJEb+d)JD!
zZ?q3|Efj1VCziK5^;#B0xETAku3g^h$)%@Gov?493qF{}Z?7hy<c-t)u(vwBnAfu{
z_t|%JuR4!(BD!_lsmnn_1?$?;8o~$LpyA&pH?j4dnp_ShP~j!0f$c-Nq_{bc%Hrsh
zwa>=;BNNm1;T-X5GS$y)aZ0%*Rwu=`hZBsEX}Y)BDJ8$SzBm4W()-$Xwh1G5;})k>
zQ{F9E&?m*S%3J2U3f>S1&t_2OFjQUiewPD}-s@0P*gv7__K~Vn*m`<%9rRWg*KQ;S
zY~P*j=C|RZz&(+q&K)>)g)UN5>3uROqyDsywvpV7#|JJC1E(gUrWU!f(5sY66wpJT
zLn)>}l=Q`sd1$YFPy0sP4HtQ;V}tB-Ahk8}Zabmr$8s$o^O5q}Dd&u$5Yqk3=k4PQ
zb0mGIwm<KS=q|qZgagkfR@@JtFw6=6iLM7yb);$?83S|%B6QvgdSGe(WShx2d&SQm
zcIwoES%F)&@2wWe>4k07sjE5cly$Ks58C&&BjJ`c1E-)D^(Ik#4D9>sMbg@(W$cKz
zx;nxFA%37HJ$4_jMPy>I3iFUTSu{ooV_~W|IH0;|ZHHasIiWBi;QOa-TbmJGE26wC
zPEGU~!8cZ1hI_5wX?vs{3afjhLve~ZnS+@P$S6b8wmsjxlo=a#3JFG9A_)RlE$wKt
zAr%(5EB6+uDGxc2JZ{#Ej<K(H=0a~lxrA@wuu~3wQE%3EwRzjXj-ybUTe0YwCM(&A
z-E9!z%96Icpj-~5wux*Up}n*{ZRajwo0C}4smQ%pnm(n$35kY}C?(3?M(9^fggKXT
zw2y@R;Dhon3!6n6F<INbHW!3PlR=XmLrrqA5Zs=shHZbF1Z4ysc8bwasobO{BB_!c
z=x{9k;kakWTb(GOj3KDY>|pDEP(2y4@R545(+oK9Tw|F$*(bV^R;abt9?fI;f&S>U
z^E-OLl*%L}L#-X^ru=afT50SG>~J8L2(8M>48Mt+0tZ*-_Yh;z`D@wX?)&n-3*#Ci
zgc)}r!UQk~OX7$f=?21iN0SW3h_GRVh20KZm~jhQu*VWJl(k^iDbxTn@aNI6&G6eE
z?;r{WqU*V9aLsLy%;toI6;9QU5iE4HdJ&wAmK5PBqw1s<dBPif(JrHvb|2P%#KtEY
z&P5uNxTV~wQ_nBE?5%dJj&IGQq~yba%;NDgq%u91X{8~}Z-dfZK<&Mq|4S8)OLCJs
z!z2KmMp>M!K=8M`5$fZ#%H!#dDK2V2_jMv=j7wkZ_3K@)u#@db!ovB4Qz{gc(od*-
z&Q7(luzk^cF-ymsy4r_w#mZSyE)~)yx7N&)QZXK%JMPp)Yl(XV)Pk01NwMRcv@_LO
z8fQhCK8B+Vb+xr3`8~&wQlrUtMF{(doy}#RK&vwD6eNq^v1Y|uJC_r@e(f8bpE>lq
zN5?kx$F|t{d}!sm^()pog(OrXQ=^lU!nzC9LycJDPVGCP#J&KL+CnyX-OPbkll@>f
z2U5vN_0=MM8bky{o+4e)yZummCh%Ig$dghRc!TH}kxHptuo5Az&vl2+4%$)Db|Jsn
z7}BX-Y!ir5k}SS+#N|76Mav>5W=LzP?bcy&*5<os4u#TLmrYSu+b2>-KtC`5gh}Gb
z?rrJlOF|D8LhNmv_YOP{mnR?EXg}P&<^bIR4Rf|IvNVVoH$`@2ig}~~d#aw1^;sSg
zQYIwj2BLM+B{RM8{siWmAIT+Nwr-{KGf$A^_-u~bk9LDPf(R--z(J=@UF%INnbD6W
zrt}7c6!AM|C}=yKiDf{$q$5Sl?vEEdMVGJvfXcelAUg&;W=kYNoR{oI$tnFrbt8>&
z1YOTP;?!y55BK}_leLzp%V9Vu1p!UVev;Owe=M|qNOL>vry9;e?l{OF8Fy+*_Q3IN
z)2#h;4v^rYf%CgiK%#o8Mp1Y%GqN;1Vn5Tm$q}O5J@~yjb$Z{$53%$gB;4s|s|`rQ
zb8*%>b(+y78z!(d|6Df<+MFp&rkev0ZNKQ}#fNZV`uWpOy-F3Lq!E-F=o4vN*>O_1
z#(B;$x{xKIbNmE)WPXqH4sG>&hc^t_FVv^QLL57+ET0ZUJ;YIH-Cg#JH3a3y4=p!t
zIn<@8ebL8K<*#K>_&K*7*)KJ+xDE$=r&MH7S?4&{hz9-T|APrO_7^kpSn5lkB=%4`
zVW(}9BY9DA)Z4Fg>vM`Nh$C`%kmS+G%N=+jSR^%I*5spzt=oPzJGKIsj^VyjO2->w
zS>yI$*>3FD5?8p|2HjMo%b_NjkF^AE<TX{lp3}Kh*WB_u&_yFlXAoJD@*oB4?KkqH
zDBF~*fl|I~%f!TZT#|0TnO|I^=#WT*=hSF>VuzTOM&`F_Ky|*cZ=?r011AV1=jI(V
zy4zyRe!C`Hxmx6Z0S$5LVm`$4O8Aqvown^h`<<eP$TdoAQALt{5vhdlZ9SJ8J!Ie7
z!a!n4Qz{nSKVrXEbs6**{Iye;gGEeETa5jF&6ey-%3#kuG|X1#LG2B$msV(w7Ee5#
zV$<ypI&3->yb}DPQx_|sF6t&pMK6Rw9*$r;{Nx|DNkPws<1z}1$DO)7#-=A`?uFwX
z*dH~tVuBPE-|=5Op=`>)Lw<SR{BBYl{Ba{7w{nRr94ckeshh69-u@&(FykW;woW-`
z3ziL^7w~bv{b{}oA7?Jyb4t-7gS6#OyT|^lTeW>5mk5{h8VMj}&dL<b;_@_f(S|P2
z-2OaKbfANt-?eQs9fuN`YlY;mg{qUmD*o;-x-Ukm$VicuK_Cqx8(t=tKExNfdc*!Q
zG3erZ=MHS>4V+@2ngaULkwP(SPp}4m)pkim--U%^UT%G{fgN~~o;B;%+h2FD5i>6r
zU?s01pN0cTgdZB?W@wP~Z#sMs0|{DlG_*RBU~wD@;Sbs0wnYfpne73?MqTw*IkUAT
zO4Ve4*DZ*`7homk4m)*uEG&wZTg<%ueJ<(zHWGDR<CH{jSFT@ntNlZLCvAiI9kdi!
zZ-D677`=~sr{$^nk6n+X0pL^=6=e9^y+M*w)p?=B9!Q4vPxXxS9$0{+DWJ!wj2uXw
z^0&=$rLX;ScIkB!gH2w)FMgP%`Ipv=ybE|<Q52rz88ySzng80>Wj(Qt6l%icE!5q&
z23z!RZE}LoZ9~*$xIxw(crNei^>^97w=b+^4V{ueT*$!%KL1hgS!`UCVms;9K?h!k
ziP9R$*Nr^GZD}JmM7q))PF)Ut-OvO!yZ`E58*z2zR4BLg<Vk?%283Z2Pa@XpziY5U
zbkwe)63Es-3JE6$9McEQ&LnAg1VVPuDJ9iI6Edocl>7E+XR14E`XeK=PEo{SQ-7;{
z`k9>4T36OuECuc_-XINY7F>Qm<4pE8cue8<+zF>HSFL|Px7>bay)|tLi%B<-pkkzk
zncHQb)!qcP6U{shoue)mSsFn(WE_u(FkQ6xjMi7SB2+v1Tz1z%2NHWj1fu4Yer1xE
zhw$<F?Cv{zII>d=YIQh|!T||o4%p|MNv2LD@Dz2YgAtY!VYspL_XIc|JfqV#M||;p
z^zC!c6p0IsQ=T4rrxd|yfyF1=X_aU(uEbpSd0FJ*D_!*B6t!osy6T!6?DN}LrNLlk
zx^yt!AUW2>vcVVBEF@>EzmM?s&Wh&^UTJwkz!6+_DWx*_!fuvyfw}hmIj1fdOF3}i
z$YsxL$Th_6d(%BGEODxmFo|7a!F0V?$dkN6zNcSQQ<DxpxoZ>M#G-T5g;yZ0>JVK}
zM2gIacS0F#&nh^|J7}_G1Su&p2di?_0>7Ne(@+$x6lPp4piU*RxFSi4f<&gJ2Hn0r
zyM{~Vq3AJxz18X9R=f*Gv3>EGq8*oh{iRN!Fio#+ACP`Wt9``2B<E{>+gh9q2zRC&
z*qleBxB}s{<do-ROYwAJf~5Q`2C#D3l99##*_Y;Kz5R~0&d)-hix183vM*~L)9n%z
z@<WVV9xhU#%S`^beR&7bNu9C&>iAA?5N)zB5OxR9SJX3NBystUBZrln-|j$)vSGa2
zY+u=4&Q$G>InWt~8>s_{nUZ^UcoW2ql0Q&T6nkz>Kl?m<fG*G>>S7S7QV0ypIkvBA
z^E&8_k<G#>$J^UNXH-l6C5N<qb=#ejv&aCHlH0nKn*rgn2lh4XefaonK!-><=#qBL
zF8kUxDo${@K8giOMG%nO^=lQ;^mR3c(8fVmxX;_Et8p)63=gS;aUVuBdtSk%=h21J
z(Te4gAjJi;4qU`}Vk3UJ3mY;Bi7r&7E_taS4A{HE+!*SoaKVV+TYG*R0Mb{;<tg&d
z?R6k=)W|nkR|{=NP56RV$JwQmPSJMM#V}GmD&!Wc)TL9iOS%PENN14fE#>Di2i|#r
z7V&_9oIw?K$X?jK$I&6P9$$3o^3+<h31(riX=2h|)M3K~B%8f3iSm_#_B)V7!bk~a
zFV2lHp+pN#K`S)XC+o$)E^Sj70<<{kL*D8#T3HIC19n-PFeFQ<3zf(!a{vH|2}=i9
zs5ujMdDmL_p|?udQDhX5Xe~ZSOsXs)FKMkq&BRo~Rpm(zBwA~F#ZM_>=!))p+@(gH
z!eOY3#h_hM9c}KoBM`0&Ug65NSP);uYjMfaViN4YbJ3XvDdfbrz4U2yLCKstoj~|y
z4D4mq>(o%@Ngmo_2a<b__sAf;o&CM&%(1K5o3Yv|1+Z*6)Wy!$NEN1co)N_S+SP3>
zSLix3r<lx5L=7Sry}aHAkqW(*5;IYE3H@Y8Ey%oe8w&eZdH{6o91`*az9!$lI$Q_`
zUd7?j-WD=`+O@gcFkLu5lSf>-ewkg@jUf!+eEV^yE~$tj&Lgtv`WjW9!p&{?%-k*q
z(&i|X*-C>dN|(M?N9=|+Dz==?+5@QBsLS_H;F^o)CBC|iA7Kt#Ts)F12cXzmsUH`D
z(P<kqoP9|j>BPUA3L=|>{5<41kn7zbZDXIpNG%P>L3>4qQD7?l!>kX^b}B6gF1}!X
z+ol<8D#P~5EVFBNApah9%1I|T^w@>3YMjlrH#)y_luK^DdF9&GE0)`<i=^rP5wQa&
zBzZGF4j$Nf?XP`c5Um1-m4;f*K;$B~*R+v^@lDFu4A$A}4I-5p&bgto<VkkpYwJB4
zlOiF(N2jjsFJ?8FfBBpw+keWwp`KyZSS@J=cQ|!Pu*N3FM@FY-hWoapdkv>+<55E3
z6lL7)z;g(Y7ey1C!0~Nq4Stc~Gl@eEkximxa<q!YO&71w0~Q;xtSuB1CP%;Ic;vf~
zbQO4fQ*i>~rAYZ^%e!fNG{lWZBPIZ$!+~UL`YRnU2b!(OEwY@%Y4sBqDZ9CC+f|qk
z2TrjbJPFleXt++_k~pwzVhrbXrxY!850pX-2+y`#yOq~E7mpMk;EAP4C219d55ApN
zN`IR?)XHwcMHkC=k7H1>KjB%v0N0fj_*Y}1trJ%_dD!dPAkv|eVp4!xXL@#9^;c^y
zN=_Cjr!JQv{ozokR0cC!RXthkY$DD_x>c^hbKtpOa5f<h(N?!HrBY1T9Jsj8fh>q`
zlY?2V@VmV=p4*O2zBIy|{6`!}OsagGBz$ufhTYLkG@f)Uh5U)A3k1h(#~Wm%l(lOi
z@^ei+Bgb-OK~tt>>gpNs^~xW@k;vpL=9`Q)_WFW1=0~EVxP?8uA5(<`&qI2d5<J`5
zYAzJj{#8=`++%DG40^z0Zo{{YwT$F*?9QFGu2GEMRvdN4`@>F+&Z$*g+hOY)6g4Wo
zy71i8ghamu&vd)9TgGrjmwv!oT^vm;tq$qCy6#oVL><UMkc0qx9e8vvZ9IvpVsv-e
z-Q7HNh82)Z!PcM}oD}fd8eHj7i2H_mModLZ;3%1<(w{l-nhZsgNoA;nDvGgx;{yK|
zx+$?9i@uiccUt9Ui+GF(;<Pt)Bh>Y_Iw0f@cp|jZL>D>mTrWHVMn>$-i<-BjvZ)IQ
zx!-|D7sCaT7s)v@cjY~~@7Ilv;eG9t#LrGm4%u7sCpJCjXHBO%lM5l_dBWb>@PFCF
z2oBHqa5_~nnhf>Gtf+6xk1#9S;fg8crjfTvsCw<~-PtL%r$j#KrBZjDH(+;O9!}fZ
zMr8D66uNG**N6dLhE5^55?o9F7jy3&XLohne@}Xk<2b$B#syb_sjiWPC<Y-}5@5&J
z$<^+acF`7j@2;f9u`|YT5<78A;@Igb2o)q$5JGiDR!d0lz4zYRFFn8aXJ*d#F3FSU
zujh{kUM%-}=37pgGkxal(D6x|&p-9C;D1ZI|Nr=})4$lfEg%lfBIpZSlyt6#Y^0k~
z1*?Ta*N1JQlbH}P+98rE=0)h6)0UGxh0%X{r@R*Zwh0O29O+}dcUqzYP2givc_qgX
zCnd2hxrJBv1_S4J!*+O-o)AiqiDp|nBvTMdBd*e28zh-^1YGcbPO-AFT)1&gj+t^U
zB!`TQDN@A7yVphia#W~{z1r_U3IiljDq45H-B)e#TH27|oVFU)IS5_&oA{gVawJ9D
zR`3H)cJUkG1&}iSE{H*cGf410*|mfNFjU0GCbIJca!yP*7x6KPJQDUod>iWGHaQ$l
zID=)|NR|midwDIV*GoE^dfEhww}!TXeWyVfF)V>k1UZs@V!EyJWS=^GNW255Nn82)
zS6pq|&p)L^<B9v7UnMrLT4itRrbm$)%UyHn!~kB?tjt@~02Gm#dM16Gi22}1TU8a5
z10K&Df(puGaEWOew2$6RdR1Kz?TSHuw%(S3U7FORd<W@^cT%U_fy>Oord4tI?q8d0
zJACC}AP(&`R9BT`ETlYUJKAS&=}$Wsu{tS`a06<cWIDE(lug@NaM_X+04{Ee2rcml
z<T&IFo|PSCpOU9m=DrBFvRy^#!&T21RMcK-mbk2<UBrD3J9kO2kPmT2A#(HeRn4t}
zbyN!{XCW0-;Mx!y=ds@Hvim#fGvCM&o6M7HpvHjp7paiDVx*~5<f;<5B$v93JNEWM
zuOUxqb<lw%2E$`3=s>JJ)&#SteSJ5{QF}qg5T`B43-PU+`$Q~mv3Jx5l<%NYL{W)E
zlk7ke`|f7#-<e}0Jd46>URO93lqLBMT~B0RjB$ZWaa<wW_q+p7?sv~v5AD~hWBUni
zVFKe@w8dbMR-{EE<W#{Ke(Tt&ceTN=Fba}ETk>jK+D)zF!PrhLeKYnzyN~p9CHhJ}
zq^+E4!?e=$OJ=x5Acnm=cipRA<opVfnM)$VsFFx3fANjoT(Lw#fY75r83$70?V3C6
zJ$0MbS<a{PQg+xadmTt1zL>H??++6b^-XPeOmChT=NLJsq2{6!bol<Ij7*~!-&`*d
z!b*@Ll)-7!_Xot!P1(1!uOI-c+#;FHUTrvl=Vp}yr>K$lwl6CZ9-p|=P|QN5e6z+>
z${p|P=2bhEnPGd-mW+EaK1rbLF8kIF4<6G_LjICU(8VrzgQvqK&OD;~Szb{=y5E+A
z7?a1|xrkW^s}Ns=YcLc^t{H##{*rz0vDVMo@xFba`l+cG21$USO%g4k5BJyy^Aikq
ze1~&U)K6lwlBEvXhnBrS3=9W`(<b~Cwu>QuxUJ7JKBPEUVFaYtf$Oz8IH=V=vv03w
z#FY+ciy#@O0IE{XrTqX8tBYhk>3DHiB)x~XMdYQPlbCwsNw@D<)<xMAC|KgHO=nq&
z$3oogJ9CCsuD!|m*<`rMxhds8+DE$i<y3ombWWOMT1=(CIq=-HTgVrfw(qKsa;iZH
zm6z}f(3a+g*|C{OskV=H8zkMc$1pmC)82t(57um*=Ug0FE<V=qXX7*k7E>|}LgLA7
z{h2BI?s`=f6_otSdg3Z~AYzqQa-{C}<P)pLXGnJELhC#|F;0|n>u#9rgYo&WO_rUs
z)O}yRt50PoekL~Lw1eWu?faAJE$yC~rtSz=u|32uq`}J@kOJ)ww8>m%cwWwD8$2gC
zQ`3|81hupu?7n?M#0KNqg40$TKo!OGq|V<7w3gAvc6Xaq7Ha7sIF!Zd97v2rd|EC6
zs*u~B^Oe<w<fi1FI9`Ot7?3JWf^{fx!W}G+wzm!Z@?jz>xDqa4f1-}*%WYphCIYDU
zXDq5rkqxp8we@}Ca~?eZ)H8U+(rzh|gh;ai??;rSIZ;T5C@99+_BU;Q1SxZF;9S=<
zq_<7jDp^^PI7HOjLv5uIL6i`P&DLp4WLwF}jLNd^+ktvbiiPF!rD}>o093yXxQLl%
zhnn8(kAwAqD6bAk%4imqLJD==f#)vg5gEr-0O-*0BQpE2GsO<ig$am~&<<P}dU?;{
z89SWg<dhA*NLdH2W1MM*%JPyydwCD)q&-}dUiut9p`-^7IFMNGp~`*Oj<hz2u);c2
z`K7gK3RDUWPmmq0`!X0a1xeZhE0X?$RuO(8aq^MeZcDqjZ69$i63b+iWS3)<w`1*w
z>#knmTx7pMX9bTU#M$v0rMQ+>P%f4zro?X@?*UwxHba$rlGf}*8#LzqhYI^a0D`@N
z)O4loBrABIfl0(r>}2+2d}Iud78gNXb3T_lOC>-KX)KNydo+7ZnbWBq)VE`OMxC~p
zkqpN}YKYhqow9Q(_ld{=k_K0GpWLG1bLX#2Adi&`by=Wnb`-c|iR{@jdBk}lW{-7Z
zdIdRQr`z7xz{$Nx3S8$Tgv~71FytxTr;OIe>m6E23AlK=owj<0uw-UE<Fg({pw+HA
zXlJ_d?_#TBekI3^HfbaqcG=lR7iN(ooGalop_OBHt{b}&12J!gWKU#+11bDeTh@#}
z-;IP6V<POij(aM^7%q6pE;QOOq6iu!Pp2iNtZT2k#xCZ^cVaSde#f9hO`Aa_2z+e6
z<F=6ZPWHxn&hNT6W12-#gGa#<<PE=+lwi<D7K_*c8Zsev{Y+nV_*8koo@kw$9EY4K
zO^6mr|2P;%i7Ee~nvOKN%N8Wtp0w*BBx*LXk!2O+09{*blXAyxcdl8v!MPTryY3ZN
z+2!09_(mX4Tb;|8uBNw%WG_`)(;+xs)LWa}y&{e(TJU7`tVC`3u@n6qNX!iLAWyz{
z34bH9O|@<8ha2id=3)^iAaV>SLLGPtgd@{Y+QOcyF3B#Pne=C!HVx{nFS^oxq}@Ll
zha=7Ood!+e((cP2TikCynolq7-YoTtlCp?;)y9II3S;Vn-1lP*D^}lpmva$YP_cMf
zDPOv>QcHh4KYaZwu5o^f%D{v+&tX3OiH5r2zL6p%=LM;d4kYI}lJA3e*iUx&6R~tH
zC<T!s#SGx_Osf3~rCG3_sy4}!5@0C^?6lQvD@+6}0mDJ!VWx!rbbAKei4^t4gWlRC
z4Tx}xwERpPaip+_3A8wIAk`Dqc9QIH)PA<j`Ic$QtAa{2BnKwoe>*1$&Dzi9DojdE
zn0K!7CE;w*OOD&mH$?4?z@q!kbxpcZ8iN1>3RTyBp`OmdZuq}25z`iT>yGm6f@T<x
z?H9W#kx?eYUBo{*lMLnmk^@&?mUfShqm;{jIj~>qo{8WTOc-KDUCAd0p5w!Qt@Px+
z{c_?9aUN%1=v)XDsQ(})Zokrv_OT^-si~FVw3VL2F$&n_>{qj4BX)tQ(H==QG{lI~
zPa0@~bdM-!zt;6m;hU*Zrhz-q!y8-?5Gqx5k*_Wn+kQO>*G2`e^IKH-ZDaNu*^29D
z+WDP(W<9wN_M6rJ^pPu6Xur3%kd2sWTn_eI)vrW5si;ps*%o+=NPI`eq&<?Y6!2ib
zU0=v;61F&}E$+rCLg-#>otV4h_B(aA2%u72-i0xownThndX<fl&$M|~mB;?O4ZEmu
zNl>_V*PA$wb7Ek>*Wf!j+5!yby84=FBhsLPIN9&lm#PLz9JL`1qpzw$DDd<gZJ_oY
z;>I7eaYWd?v<)(4kr0SO?!Iwe%KS2`<R2zBP0x_A=v>rhQF;HQ7k`u(#)+WRP1xpK
z7W)~NNDljFMBOdeA19ewNg3<<xo*Rq_9yjy)w9tdy~;!tmOtV^;(a>xMQbDiBi5%?
zQ)%2k?RqZOsL?3mQI&UrX!P6q^VDP>9oV1MozxRSl~`6^z#Bw{WBJ-C@ya)_tlFQ~
z*CoR#_-h};hgck>WZiqhfv4n!w5f6dGNa0cr9PlaMhoN$CUnOBA~BhSE4YPoRWh55
zmYPc0gjG$UF`h0v*#5HZ@$KXNnGxXxEx99?j`9^^k@*|5Vt>`R#cXffxzH3$;|xKA
z?b~11`lZcA({##dyOVq9f|efi;Z>r%s!*1xKZ@1Xwo+K5oTcf~e<<AJt+V`J(9qEy
zL{k2n+`>z{cg#D#i%!eP5q1;%+lHDGnFPtTwmDx$1+%|vlfzUgVL(n71!nJILeqBG
znU0)$3nxXKL(I0nuVqU_&QJ%DcF717Zgb$dF);jYo7-;xQ14K-i%xGM+skPS#8bdY
zER9k<hrT)c$65l6^DKS3hZF!oz;oLX8^%XQ#_gZlid^yPE1Zj(i-(au6j67|{!g_<
zzz*|6+KRWf^qnHoV;KAA_M-dH8@Wg-Bl}`4TdC|Y&xh<^8eF*tZQ;1L6`KWKR7D9l
z>?L?rZs>olC*yFtz$#UX>5Q-rTr5Rrf`f|oZ*{wTkhp^&fih%C=#O|ZZ}7wv)`zW|
zYaxu=XaC;K$m0-cwH<fbBzV8z%9p;@{-eyBNSaaSKwi(Nwani<E&*xIxhqxT5lYo^
zEb;w)`_Hy(HcU?{4S&jMiOCmLm;H~D*!ExTu63I=BCLe82o=}C`-vg43xz7TOsn>|
zb1fNttqPOd$6d%pNYx}qox4K{;8_P=lN{N75Eg|O5bkIne<8afTZ;_BibJKXHubPd
zCQE|M{t2xUTR8QOE^-|Mk?JgyhVorMv9*DZ@>`Ii=Cq}7Fox9#nRUcIskK}tjh_E5
zIxQ*ws~TT_dUl?|$M(q=LMA{VnC6&72o4Qek=tf^Z?jLi5F###h|bUC&YO@N>{Dw1
zsi`MT)6|@BAkpD;!-gJcpLQYeXleH?M53Q|E_1&v6PvU|q&G&;>!)8RGk3DpK>KAS
zs)DW@_wKInj*Hg7EIUab{ET;fMt$oc&Y(g~53fQ-1j{^h%o|+HTboNBr^ML^kh9Oc
z(5yZ3Sx%7@ACp9L5;2zGlzmo1j;d^x$7?BLT+t?{9e6Ca_*iB?q&4NBeRf+U(i&Jc
z7Mzy2myDD!JuChwhw_Xz=Lq5o*Ce7oAr2%tXNd6HEFym)lVD3aW1n-OZ5e1V7IIdZ
zCav<DIK#l6+3+bI5mP+uv<dmA2NTmf>~q_KgtU-EHrH3^+)n@Dt|FshXNoOQq&O^b
z&#Je`Q!0Kb%W#qk{0_V&PPFiLSJI(;9-=5xlO;wKoMNBXVVOcrq?#Ub+T#1N#bkt<
zf>9w;Y!PMreSVwk-drBy?X=`JEME<=d4U24?F({OP(5P|d+_*nr=h6%DezQ*LkH~(
z+g`mD|1?C0Ymv*P-RpKr2K%C%!cB9y6r2mqHz+i)FK);a;sICDiEhCp+(b~~m*fk2
zqpHWjMck$tVWTi381-QmXYEV#)mX!KIlluhCG(uLFKe@e{3l-rHb1AW-jhLKasCd3
z>WF=LhXy*1Kyu|dr%m1N&268xuV`47B%dBgN>ja}2pw_YIeg{<!}V9zhnkS3wBqQz
z11S=*mxV%k!ua`B?d9<0IP+Mqg}o_fT&EKQ&pgO2C`?<%w_*%~WIwyUp)z&Hm^Gc<
z6Mh;85)+{vea@cKeNm}3VfA*}Vj_}HWrBzDu^+JK=Cez??;&{Ixk{yJa$eMHU{@4T
zoS1fg=GsMV_{iTZ1bSY5Q4)=OW=fB0ZSqd*tYPZn*z@Z?C_4hB0#MF7ZGp&$c@jTE
z1eRY`W>VQlUXX}9il1=1&d8W?T4J;hEgr&H_QEz=R8i%a%gRWb7KejWDfB<ED-*SD
zR)U&y5u?ZoPV#282&5OaCB0QOpq)!uJl=Bez4qdK0Y`F744eyno??X(6JC-yLk!lW
z^SdTlC#&a~xZ%dLmv#deL?YiIAGYbJ!9W&!S=-g<JQ51?>;q{=nyiBkJk9Ya{3_)?
zLkLHYBEj|g{fpYvqV8wBwJ8=&^=J1Hq`jg&jumi>!g`!3BKd*lWT!V#(^0-iyDA5~
z0~|Qyb-{r)BO`WoLY8)~zGIDZk=Sx`;;6l%5chpc?;QpbIIXb-q+ybOqs6srI%Eo)
zK_Gt!#NHs`CxF`NC~b)yr|sIB9a$hmA<u`+KvCtPA9;hPUKYm|Ft5^tHM>Xcx(?O$
zhPbbC^=MvDW3k6`4qR;D@_QEd^zHgw{;}S6=ciU2j?O{xc0;?`CJx4xXvR2hdAqT#
zvM6wFXs4|<O-Z8Cx0r`h_R6|DD}sj9=Csvj7zV+io1=qAWi@+MoAC9-Ov=PfL%lvS
zPtjcz1wp)?y}HRiBfUw^yC853Mo43dlfA8!$x^P8y{6t!vTb+;wDS%+kXSa<>7Y!f
z7SGsgFSKLwl+qg(FHJ5>l$wGVU87~WoIH^_iJY*w`d`-&Xw%J`oQv@B>#-;m6k?GA
zRAkK5x=Bp@>YU@NZd`G-^E)~vB$ut}A_Q+`4MD3cK~z%_PdbpAPn{(qTtE#1dlbFO
z|7}&lvFe}<)$u-|_ym?Z@Fd=lH0?oobXM0h^)*CQRk)H<F$5|H9$56i;_;W+P2C(`
zlmI}u?Fpx?I;B{P%>GJ)e#SGdX)CvNVs1n}R@DTgu{{Lt${HfQJQMBlz}U?>M>mr=
z=loU=MrV3kX6=^5<5|iL@l$v_K6RgJZQR-xW^@{|I_q5PsioZ;*Ke@f@>VA6(BcK>
zDkmM*DQbBZG4S;b71vJR4_)jz4K-KcL{@B<-QIc~euXfqGV_v{gfxd8c)j?%>`V~j
zLv}}w<c1rrz0vs{$diY`z0cNm$lCesH5ZtCfh1+@?}99js{*sg^PSZJDT4BOMnuw+
z#2T@f%)SHfNVbd{XVMyg%WGXH1}e@@RsoT&5HmwYPpfgeOJZ1X9}qCLxY$EN8T`@f
z+q*zOiLu(yBIqewi3q9CId@&b0&t~Njd(wQ<t2`+TE4Eb?Lxbw_J&5Xd!x9!bU-03
zTE6k#{r&B>q0Re}EIR~*f-~jVmMzAC$EOtepz}*)cS@!#Z<0c8PtrN9)IQso2z325
z*Ev69wz5~+Xwq2`ZEaKc!JtFoTMjxze--%*JPD7^5yRX=X7ycd>cIr*4l;nJBOFLz
zMIJmJanE*l-5vWnCIQ_7qsm!t5KX9T$jT9l?PT{P_DDsHn$=tvwvn&E=yVLz8ynMK
zO|0C!a~Zer0_98+Uw6#j)cW3|G}H9d^yIwLB-voqT^sCcx+fx=QjFVai+M_#O4umY
z7j)Xi!}jJx?xo$g-{Jg@Z>h+Xq#-5r-rKqu)t7{Chn%+D`k}Q2Rpo9%N|-b<&Fn4p
zHOZc0E^;x)3OSIR^~pYg1DMOE?5#Ood}2Gs`#YV>n(EUsp(udMzP9UqZB%)>2sOCm
z4U(qtREi=V!oIFso;<Tggg8Q@y+IUjAPY)oaIs$P3SqQTBqQuuR)kvKBoW+|qmNxD
zoRUcpjfD40KHOYiuG;6!Vj8E<c!Pi#5k)R-H|E7gilJP8?F#32yk87hWJx<zn?Mc`
z_Nk3FYK+=5J(A~=IE|HZm}!P8iIj^iaa&c>;5626r~#g0Ho}B-!f8vrVQr5bRohZu
zPw$^rP>BGdGW<q?0WJbanAZe{9c;C&LmCm3YFRoq%;}Pz3L~#hrQ=>bD}ip5V-aku
zz7?l+&|O9S<!^~vXT8B?$<et(GN37zJjJg)Hl9d-Bc3>Z%BL~1)9&l0KMB?O;FHQg
z2U1%}QmppvBepFko(f2_&hMD`#t&2vQmjRl=Z+xlO|<7(c%g@4v}miXpkf2EFeq@K
zOA@fj>P?^qotHy2G;jy9Y{k;sp2#G7phAg}DQ&90iMjU4cY|QmU3c1wMJW5}0P0M{
zI3mGK*WIE+kuQ&lkG9(S@)HOf6y(F)*!I?gN*RrkAKKu2=PZA<4URZ>Y1Wo@uiS!X
zYcxc(x8*kOZP)n}tI^R%ou)*i*P{RoY$h+sD?-W|IG4HaS2l`egT^pc_i#Uv>*usc
za((3$SKMo}`TQWe>RiUssY?>FPE$`}t_GID6w0oVR;S<(ASpPkUb)J4)T4@zlVeoj
zMYQD&_rCe9wzF=~Ie=n1oYd>BtqvGuIuF>cZn}nNSry*_GnIYArgH9z=}v6T*xn@m
z^EtfRHmzCf{0_89yBJz=C{)3{;OiIc{x&w4kQBEp>&&7#kV4YVyA0J;2u4~(-d=Fc
zR0yr?%5>tTN&-A)Qjt}45|Vv=jZAJAMW@T{nj=fJO^(T3xrW624v^RSqxOzAx5%8D
zItm7PYZEh;1TA^_oo$zth;f3IXi4sxZs0L#3ZIwxLF7>;*R%Ew?F$NJi_L$`X^Sb4
zi4f;Wkl_p;mmA`gdA%*=YE4A?iY_>i7`#Z0zC%E(y{mlz<E0AfJ-oQjY0F(5Oig2u
zp;wF|qn#dTqUJ5r)1%nDorY2ybf$esAPhjxH6BLP$qo1J_F7)0z$w{R*z)mmowx<O
zHnvJbiP}FFF0E7cjrC+^8Q=ng=j2{E1Ytkv!1KvSRn?O+3+c4R_q2VvMfC`r%b0=G
zZ$#J++c&jadJ~%w?73`{1Wya+pnY>&d$v4BVPT7Pn({V85^B)u2sQhbYLgBtw-!e$
zIBkjQY!B&AaTbCyrc?g2_jb=<Mgo;7=is!75)c*8v-h><5tuby8Mvr0Em7MAZ*YkK
z$|6;&d+Fmm@wk0!z8|+K;q;UP!eimIiC7_*&(6(kwr{I0h^}O9yNA8C?T3G9Z#{;h
z_5N<6sO>uHT&9(xKZAiz!8{*GhID&x*7+%h^v>tl2isGsTR(6vQ+MCg8=tTb)xu#R
z1oIVMqmta@z&a3_GG<hKB3}OCnuwvhNa@)_)dhGgh>BR|IZhASx0ejAuqWptZb$<d
z!HSCy+jkV=E)TS9L~WS2%w4H6ubd&4t8d?#W0+R;v4Pjs9X(7A1N%r5Pv)km=#}0O
zr!kN~%MX=i%T~GugFUl&(!Q&4^6Q+R(aT(E_3JVFXyOA!SjL^-MJwd*hvqPEl5URK
z$Lh)MgbQr{tu3roV8@X~Y8rQMP{@%GT*?<BhBt_IhbWYaxT_+>Ui+RlcbGcEHikGu
zTh7X0JC>`yeQ($E3s9G2B(w!#c$!l+RH$Xzk9}W5k@2Y=l*r;ThKTBbtfMuS2qKpK
z{RPRrp8}prQ6F$1F>S4m1LaKZ2O5HI(J4_QSZ8QSE=`y$^!o>^cVV<;4zkYTi4a)X
z49u%kCGLt>OC(Gq=t0}vA)|!UOnFJN7Xq9E&u6e*h;E}!+Ma5IN{7fSe-&*>9cp)r
zSms#>X_MNU2z>d0#fM+wT*S>OAZ4lUJg+eVGi+bMMGAypDu={OAvi(kSYF`TllSek
z2NR(P2_tm4bs>zKnf)Vlkz?f&3}Synmkj2RJ)8l_QVb*(SKmuFr}PPGo)uGK3Nbl6
z)D2rz^&<Pohi!*LxkZPt1MNQa4vA<OfV{Pd50h^v6O<jS`^1*SU+~s@Ym>AsYPA;`
z;!vkjXQY7^S92oKHHom|twWx6?#<<*tds(dDVS@Ahw~DPq7o9)^3ZnJ1p6I$JVPB}
zo)Kd_ZVxA6Am^|LVc@E;jlj#E9jS?wsHXE<x>L>_NbV$Z0Q(eGIoeI^uuxEZG(XKG
zQ`*wGOH$nZiY~QB8ri`=fqRvj>sS|QR2P?a<DJ1i9rekNbx4#}JsDFX(>+<81JAy!
zo}Sq}KV`@BE8BYtP~$>sHW6=jqDCV&+X!(uExm&<*-};57x~vt)|6#IWPFT-KhTyY
z>upoASr6>dx=A>uGV$ouPFu+r-BCb|E-3suvD{9Tjcgyx<eYOA&tyo{avYU{Qm3=q
zV;7#vTstOcrbp1JT&(2t(jKcT%~QJwqqUvR^<1+Z1J5W?!ySS|&1<-^Vs<j(j+AJ7
zJh$8Gd%S*a+G=MS_nX!sOkBlBQo$4re$>vk?S}j(XA$xt{G0>xyg`=fSIjHU9XnS&
ziOAzvM=a>+a#omxH+YJ(vih;Xpl!;qL?g5F^-T$D^3<gbN+vKM(Zl7{p)a&vQ;|zH
zseNqJ)+WhARe}`ZcCqeb{n@-q$UpD21!<S4Ax@uvMcNm-KIqVM^-Ej{|FHuRMqd5^
zTU+P;UaI$E2t}d-L6D$nD+XJ!Ymz3y1}w?=QF|hvQ7#c?1FFZfJ7^Q;rS(<f%MaCy
zll3ap@~F4Arl2bLezCeL=BtPt$+4lsIFOjG2ufsXPA~4W%hjgkCFG2|A9C7~lE<@(
z3y>vbsP--ue3@%HP(i*tY!ME;uv6jxWL_*Dv?uEsQCQK%tPL$^ca|;%UV;OK-$4@c
zka;I^xBYNV1aTX)Dz>@RX-LU#`}9u2K;V>5b*rQ5A{vTtM@|VK(G@~GD%9gK`;j(0
z@K(pE$nDuC6%OaXMHo4NZUQ5p{b;_NRDqE(&K%@W$LTmiW8Q&C5a#&!TkXf%)={};
z=c*7%WNev+XFpyY(gzm;a?o3w=2&Qjo@YPN1{syEwdZMRt0%1y^m@#GvTNH1w+W8s
zv?X1?Ye4F<{Zu#F#~G(mc(uFoK$25WKH$J};*o2W?sNgxD&}?2e!6wx^2cxqemCZC
zS~5t6TmU8_W8{CP>q7=DNIa%3$u$0j!Stm4Y`dPLaxc{YxDE^pqQXa2GB{(ruxlbx
z{#<oRiz78P-1?HS9LWE`&9p2Frxxtzt2cO|z#z#8dz`lVS~&E@ANi)VJPEMp?HB56
zDU+mTm}R4Y#GI?dsj}}+BW(U+vYge7W5T(pSF5H99Yvf$^9gH^3iSR`8#rgWj-!`x
zom+zlPO%S}`+vFJvsx7a2hMecBGo|?@=W|nw@fly%bKA81L|2gkW^XVtVAu>eYyLk
zrpvVmiFP1qR=MfsP4;WuqCL!uWCGdmwB=YrXAfkm-NA1t+pl-e$e^N={*F3r$scb5
zPv;0X`;B^u$h3?O=TMoJm2L7NyekgGDs^R%>^B?sPEtc_+PMtud<er!NbSMyprZC$
z^`S(L&^GzpFi8!#W|5-IQY7bZnVBZS@wambOS?ndm{Ex+r!9ufE97lKLX_u*+&ku%
z;_&u6ZCbGSmNPbbtjz*bb>Ok2DhjD13`FgIHz@<T*IS(50mp+RTLWQG=k52}uu7HD
zX(d!qT-@hC@@*(S_|9OEm+beeEqfS7LT_zp!Hs$|_6Kdlk|_-X=NvCu>TL^r0UOie
zgZ78HWQ-0bJ3<;h4Y^t8sb)Mz9p$#BoEoDfKBpnxqN$dBPsKg{I7c`pL4n^zyE9iZ
zD&`S>bJG5#;pU*<+v!}zdss~AOp*Eirzx`ywr_EMCKyVT!tgo!vwF2yflT<CsN}mq
zgwDM!TgZj}^JT~&(}l#GtYe))X2A2+hy!w0QQ*k_q9m|XsAa{IsCis+v?5aYt0Tj?
zE3hL!o!QJXgJ``jES<K$tT!KAdKA=@s&>I?D|v7jY!-WvcSg2NZ2P}z(|%unW*gcc
z*Ba19#TwdQw+=HG>D@l-w8aesF>klO8O}MZEtyyZoH-D2rEB=d{<c|$wnMA&adMa+
zO-1i@A|X=|VKlfw?C%Over+}a%sYrK5WVuMt5#fPf8V;jd3q8bgL574W8gAt@mEXB
zVO{nQ&E7KM{8ZOBh-274=G##3Mx5U@y|VgCp2R{Qk%FmG|5LqvXe}&{y#q<IcMm0c
z?f=w#wTv{`##>w3;InYoKX>07Mi8W#CQNA#yfdR>Oc&+-m-cE_SQCc$>9lD)OS`YT
z`da(fZU(eHLMsPHVQt<ZS>;5QqlgNCK>S-Hb88iJ=v*erp9-D|uOQF#-&<c|JCBdR
z8fZ$<XJnq^|EMTp|4|<=KVR^+jB08S9qgr&#0q?gj@W<JGqPZ)%t91n(0+9w8A-Nk
zX1X^zf(KJG@Lz4o`!n+x>745%`gh<QfVi4n%<TnFVD%+zKIpU+o^_{wg8t-V*vAzl
zH=c-!?8y=wpAHcpbudtB`p0)~lSZg*q+}@x1V9Rdaz(SgW2Z>lCtR$hTD^XQb5WE^
z=gNi=8>EwC_Y-RfNu?DZf5>TzyW8$c&0p%kC)M3y2S{<rX{(JOB9#WDMB-1byJh27
zs>wr$QnU$u`T9YOFP~C_C{>HCC*5Cmi+GO%5Bdeoh8%!{Ov$I#GuR8*!ijdob2Y2A
z58*<~q=}N0;L4G$_GuTBN=e>PfkJp4e<`5J`=dMv3v<}HE4gEc$>{}@8YyM+C`9Aa
zF9!3t{6WMB)F0<s(kD4nI(E@MqYVu7lI;i!A0LtS1gT|;f#)vXu;vYS*=N=;OLzfs
zA9LDbTGK<6I$}=iXVu-J4azXPfHX^6bXUwx;S#bsr*9y-ZlB$KL1lT9Inb7yYOo#E
zOC>X&k%Ndb8YC=onVlSL8=th#sd-EE&C0&&mLjwjpBXZjb62bw;h$1kA&3X<nbkS*
z9}zqBuS8<4C>{pjts|&_6kV~;ZTO0uAjYS{jWGodB>OZoHh#Z7>tgN5(YeXUmZBl%
zfm(3EvULJTN_1HJ(8)h9AK%1zUIgHBCHj6JVuO8teKNe6d~A3^HjpHGYd3DRFUWOQ
z9O(V10?u`DSd|xua-8;s`4L12MVxRIcQiTc+~OJgqUz@Vw0xy25P0Fe*i>kZ_Qmy#
zmQ@zJ2zp$$SAaJY5rvoL?MqtEq!4od8D^TcB=c1XfS|}Pt=Fb9B1Wzl80m(|R)N=o
zo?(lH>T6$?^G3{7&-tB;$B`e~$EL?e?aR9-4x;8tA$h=CTVz5qAo!GhMZ-4OBf-pE
z#)rbH1GQLy-h5?sPSN>_H_+KJ((F}_62GJak7u-KEzkI?sx6Hp3-UTRZHYB&$#Ax3
zcW+Hoy6goqNMJ^e3dF#xCt)gseLZT=X|ubQfY#oma~<~}^(r(-f9<*5mW5U_<8tzc
zRm=Z7L>gn7pRcHALiMCg=Sk)E1fo9Ec8ZyJUQRh>UhZ;!CpBNLupq_nv*))d9fO7S
zoy&NI(uEJ$3%XAYbMqpE!-3R*6)&K!oVFKMo9LBHc^3$?pe>1F1Vv~SRn8-E1%=<P
zY@`+OU4A2$iYGJFrAy^-?uxHbGE$yWk!E{Qw<MV)L!~(|CktPR+X2^3?ad6v?Zs^=
zMhVWr{K<7H{t5!0vX?X*S9zL=F_N{NrlgMHxYkxs?~b#LBr-f|FKvs7fQ~TC-$%UK
zl-4-DMaQ8`%-hTI%jBtNS`61Q5~(W)8M18I%iA2txX)J<=OT`dxj&Xm8r>e}uCShc
z4AlD=YqR2FI00^Yz4NQ+zL02n%JfwYiOF4_aIWIpNeIG-U{~iX%_)kJ--#sQAjp11
za+1B`=@4X>^k80aAn{Jnkd=}8dcdy9ua3{mZAP);s=k^i%t_sLZF^t5DU<fYnD0R1
z?qt=>!s33rE~zPIZ-@sMfuNuepzCXyB$C$2P9hbWI%wX02Oh7JCaCE?i7WnweBI{x
zh%#<-*Ts(Lq#ojz?Z$=_w<#iAo>8YsOTeM5${5>juWTrYLrJII?7`5RJ(2u!$$=*g
zA>d?SuWDE8r(q;q1%k=hGz`#X*dPG5MSFGiykq7VM`?=%5*wHAhA6G%iuRg<L<Sw>
zROQtO)^;E@{W5NcfN!sD(~nJFyh2h>$WRU>?v}i(U>sQ$Uzekv?e7AK`#tBnwsah7
z2GXij#;_dMS2uX{GK0zWjfh0U5CS~Df%Puf1MU-BSuElGRLawT?{!)VsX=J^t@^Sl
z%&fXt0jb1qNa&v91W}j)&B;~nAY^<-dC{vIt0%E$Y%SsDG=#R9@(4tW`Ac_<C~pOF
z+D&aPBOpo2S`?OH{u?B{M-^DflLu|h#qg(Wo8IaC%tfEVzG*kt%QEAdldRa=@J+<O
z29XdXmsl5LX}8n_iVn&!1S;`YMP)RNM;v&5;Z5r{+-|pa-4z|oiH^dB4kT$<g}r77
zZg<v{-PZaBEz-N?(vC@SAVF(BOwV)@$Ls4(@nS?4{#KHm18G{|4r1vj&uq6hat29I
z;yJ~awQ}q~q~Vasl`+R%@?pN}&2~q7Il`_i3My<BB9#M4`GiWSqvKm`ZMzeqET+>j
zB?l5XC9)b3Oc2OBvrP`OJfj(-Vztp9Ekv*~TUXPxERZLrF{?{maNzlFOnXWtm!)fc
z_YB9xcQ_>GlhPRQ)cYv>uIK`LLv=lwcCJz+3weX&rq!0q4<EM;^<fDF+Khx1I&H<5
z(qN*$R0In+hF9BI-)TO7;Y7G{9Z2pE&gu&LhHdKJpTW~77xy}CiCY^ddOPf{cC~)P
znvKqtBEhx(+PKOe9L(*oyDx?&OgAV=0T9J?WZ<->p3bqF<0I2~fbl)`MYuTInHo}%
z1KP4*o3ZK2+i7o10u70!zzG*(;}n12Zf|PqAM&Xp(<<)Ofn>K=Q?=Q?rag1T^SpkK
zOiT~#&E4l!!v%JZfI?fY6v@0q0nXaJ-3-g|l0jt2je>7F@FI~H!y86omc6B+13R1S
zB4P#fq-qESsd3;%xcA{XP*}1|qC_a#Tbm@ZaZ+`A>T*L=*c{l`)^llUmCZ-d9Aa!h
zvfEoH<~2|DbxDD4XPx+&(#@?Zmu|f_Yq4vZ=UZjLfW%NBG{}(g*EZMPa`sV(plrO#
zQXfKeW-fK2l@TRdh>g^HglMw0E5DjH?OkPw&fy%g(Ogu5rJ>ba1Qs_6M>HttgQmdf
zz*y_o3nlO>pkMwe+LjUT%w8hsCL&aH;z6qFTM`jA&QUUP#<@x)r1BuqM;uaxNVYX!
zPC5q5T@H5I;^m4M&<CYME-M($Xcaj*R^3ujeuN$V9&_5{Nx$+fud?w(J3@&D&hL_?
zGTY_?dbf7{1-q}UT#r-idxY!K5UU`uR(2=C4Q*T76gbmo!<9!%zEI4AzMW@dN1%&M
zG=z=k)37gg0<98ncHp_eX8K#PbmIh^%(YnBef@PeI2W-NjVWf)gX7U(9hEoL4SInn
z05QD<AGYm*ZN@6vbl1UQC9{m21Cl#taUWty*feZ=`zo%e&TR#&RZT-e#J#~Kf9iaZ
z%qYtw8CkjBR_~08mMn!V8nhMdA~e-zI_ANkF$i*90)<pKaovZsA_Te*sxK&XY=a0d
zWrBytMGfIPN<pHIQgu{?BabqkZC~0rM?mQ==Q<~RBbvbM%-p2Sb)%Ixk$1-%js|0-
zCZ0EV+H<AlC@N4jW+HCecC<GRNFv5bGw-y>!mzc+!B4icCL@<n<oTe}rW<+_6k@Sm
zZNWF*wZXYa@_<G^J!13qa0q~`4uLV=+L{y&=dt_SJFdOvM&}}yRB|=hKL`W|?d`eQ
zvMAyU<9|n-#;}K9cf$(%`uwoO1ew>ljCaU7rj4(|Qr}UFBoZmvRv{@#LK{RjIPrc3
z!$#bgy|e3!93NVqLrz<ckqXQGS^I`Y2XIAB67{Ppc1~m7vqr0C-XxhGG2ELmE{`}(
zu{9L7QKwFWILBEwzOs1O-gWV**E7dUyI=JG(Lc@`3rVHQX{>b=i#<@I6x&cCCJ6(r
zEzu@yruOc(OdB??bS`3wHC?hCtBTw=wtX6gAQ6435?Xa2MU|DHpwuYRTj64g;fFTc
zdm5fkOz%{2AFfmOyN^tG5;&+F_D#9mgSipnZ;Pg;x+x;jBDUgnyp&oA_RZBD2}YuR
zN<S0zdf0*FHX5KxN4WmCG}5_RXRN|UXiyJyr0tw4!ExB|d%JmxqNcL)J8dzm$!?^^
zi^fx-&ED7DovJr}3#TpaM)1ZMA^j{ZiAU^Plh!}y9dC7h#$2VLT69i!4VXltt9@IW
zo9Q@H{eGAcO_@k1i|^b-9vv{S_qPQjjMlkG61G_;;SVfZKINsKC04Js#aYa;56;*J
z+f#R}+qluWYKtC0Kom5z4^^8g0_e1u(^$e<TbP4+6UWQihwHv(!Z}CSWNVYTPZ>b+
zX?}aRGnDcfOgo8byyu_;k3Pr)s)!is`Pp}LeK?J&22B;|i?*0XnGfWLWVU%jSGUK$
zGf_?k5B@voT<84Ej$^O<NWM|QKq|#DJq}XR7OU6J=2h^1cv0=U+L92NLOD%UZlOs^
z$Qw7>N2_zS-0YW2PFuW6RtO3HRD)#WDT?-EPlHJ8QGyU|BL@-#kH{&8qHweC?)KFg
z<id~wn8smLGVq)%GCNggYA?35?`g<S*2(lF#b}(C>J5@5&nr8Y_dof7eQzUVZ|G4z
z-?_{=g;L2(L+Gopy)iBOzW)U-3o*Fvz;ljondxn|?{A-_ThhbD_Cz#}IFJ;s`ny0@
zXgbBn><2DBbp<oZ2MOJ*UZ!qKM0+FVi7?K+)zQe+fo&;W<G~1_La?+Se450!Ox@%8
zA$R7km=~G+wfm4S_zU^8-33vyp_tTTVFHq?8c1>IS-j$81KFN7XtEId&Q;q&tVm3u
z?d?`p#~p?q=fgIgugaRE8t&`J-q5g<ytUC<NweG)^Q3o|WXNo#T7m<-n`l{KAM8eW
zLN@lG+p^v*IFJxqg;O{^U>6E_Nn547I^n6c{nc|ZEoFWnWjSpz_X=T`N9fFd^Zq<k
z-&9bxgnhYFk`V)`EvtSC!-dmAGVOy$>_C2vGj%Dw&SX+X!D*8)qB6_Vb}-RZE>m$4
zE@Q4$lUK*j5OHOPE<W{4?qAwHP7qYM$0Vmi@Bv7ixcpH(frs;VZ)Bcg!R@1~DI|c(
zG1i;3hpVGZ4s^U?%nN^+RtFx46GfaaqhiDKAmfNSN0L$yu#pvB7y(VWuUR9l0jHMQ
z(PRgcbGw|MF)2xvL+s7zN>sN;+NAORtW&8}#t0I}CUVSyr#iQzH$sH_yd7%~L`e{(
zU$OaM{dMfU15a%Tw#wj!-WOimC4JiQ?k&e5xuQ_=2upwn%d!Jc5omgJ6wT;F+b|4L
zDJTkTj=(Diq6OSZS-^pv%yGZ--nTeE^995QDde~DG8WCbh$&0EfL7t0ovIFr(lo1Q
z3YmdI4m_R_oDHR}+G7nBoAC@T1jiw^07xy$IYq@Ww+HQX+uE`Y&QR>yX-O**rhbAT
z?eTVf4+($3xeh##6ssaYkII6C--yz`1qn+5y@?~d!<7ys)|<Y^mXofmlau5Z)|nw^
zYk6XtX-g%LP(j`j=00Sf3(j5A9dDsi=L(sprgu2MV_3PX0;TMH!g1!j$@!fG`c|2h
zFErGF2_jlk6gbx;1n{G8wu{}A3Wp?g9dg=g50g*IWHymnzq!j6+Qdc;k2q`4X-jEO
ziROdtJrd0?b>t_`u?l{Zr4!^jm6AkCr*l_)OddIr3JNy|VN!e)^Jh=waNz!rE*)^L
zawp(DKnNPMA8OxKP5kju`A2Eeyxnm1RkoNI_v+Wa@XA*?7ljYve34gSm%HJI!3^?p
z&}oYvNmq~+Q2Dx%sVn9SLtSbZ(i<JcY~WnSdv#ox7_?#vp6tG*5|G#(pM1BsHc34Y
zO*wDuhwED{A43UavN?xLx%99D5A4%v3reZgahhTjr|qe3ky0iW+Y#PZte^u=yv01C
zg%b;idV}^OZC6sEq(9+Y)&g{A5Iv1b{AewsrbCej|IZb2xJ#41YTX;{$GSN^Dz79A
z<(ShJy-_lvY$rMkCMcZ!c-}30%k1n#Uj-zbHqnz1SnMbA&6DF>ou7sJ>BEs*@h68f
z4U>T~EI#hE#n(jaG#_0sqy1FZw+OqJ)Z$&6qB+1i?6mgN!!Ly`kR9r@saH$8uYJXh
z_A_k~2O!>*bD8UejH(>&^D$LQ!}{5J6YL8OCyML%u<blzP@diXbKQCl5Bz3xE4URZ
za>{|%n3RfyN*1QX2XGMne0>4oGt6BV7cV$%N!8Y=7WBY=p`OkNQPD32!zt<&NbQQV
zV*lXgWA=-6_dc!;LMis~Uj2T^fds3DRNimDRJS$K7#Sigz*z^P&#!JLp#5^ANvmgP
zCY-AnN@Ik|7T0z>X1`LMNuM8*EYyg$oc6nB#_d-dvLbloEJ#<2vvY!U0gqQNGr%yv
zG8+C`y;{6Oz8tM+Yf}V3SrY8myUr=PxuOlVR~<;s6suBkN%kAnA7PquQ6Gm^=y!8~
z0I%7lPAEMkpx%D7zM#F3K|Iod<dg=o%!rx3Cs9OI#E|@v9=G4>Cg~g;Ne6x`IBmI!
zp&fDNcFfMh;?rRDN&D?aH#Z`4jp2u)AwHE^Xkx$Z46wAc26wUFDLGW87J?FF&Semf
zi%|svrc5cB^SE<w3jCW@%#QSUYN*-oR=;vR^iYIK%0jHD9S0ukC8N0Pma-kowY*@z
zmpzVf<2xZ#PD2eb&trU)4WNE2$b$NC_WRw^9}U~xc_e2!ijF%F>2d3K*&igI9L!Ne
zzkXxM*=B#(_~)Bn?fkB7e(hEEM|nZO=86^0Md9sG64zoMw?FQd85;@%Ai%?Ei~eRH
zTn5DIZkai#xnO_Nh7RT{4n|r|+OmPflj+>m-4FqxAWtfV?oV@f-PD`yZKAxUa~=1K
zF%(&4Kb;qWP^BXKv+A9yJ?3Gsd!4p;rs^GnL_)%_S}H=^{=5;8wGz^2ooh)IR2zmx
z*<a)vsVv?nc0eB1QKxN%?qH0tL5)WS8%#B@Zk^myx$BTYo$}WmC1QV>B3XY6K@o)3
zj?6nPxdNM3-DZE4OS59d_18MTV*$j9lyW9Mo>soUZhIm^UXh~U0<=jRSlWH_?e;hI
zWD9@L!Ns}ffYh$0PHwQbN9}Lh<UnOIb+${i66N#;7cHqYS%%t08QR~~wn<t8Rgh?f
z{ass3yIGq{fh<zBxHrBp4zCcYN3nDMeYc>)snj#23OewlaypBJ$y8>N>`@2oA99&c
zaOB|Nst`%h*iqhteuShDRN)`%+e#wSj@*k~jke?*BNTZigzum79_meQ=g1j~;c}i?
zwXv}KAq$c%&fEX#-lcCT+VY&YHXYz7iH6bHKR2`wuDndfav(7ixw<HoXmS72h{|Yx
zY@Ps*_&kl0HMaN1?O*GeG?>s6wmWSx>U45Sq?Lt34s>m$<MwZj8QqGZ(z!|s1Z~nP
z=?tKZsQ>r&ZBiuh5aqXN)VHtaM2-Jwt03CV=p<!e&1Bp_r1$hlyO+oEKf4aK5IZ4_
zEUW3r23{x_u{}z?wf}0Jym@Y9%(-fJu(w1-_L(ho(*@&}U1-14wrs&gO(e0&a-zQ%
z?c)}T>L)wVGr0|Tept_syHbN7mH$5b`0kx@V5c8CqJx+$0_YrgO<?+&qtrKFrp$0X
zxxC0u+kohr4!qD<?3ZG9VdB-%vY)t678nlGQs%_KX-WNJz<D8}NCW$%oB&E5jLc1J
zXRfgT(3Y<Rd+-`SW=M9CTxXx$L9|*ajme(;xC0TnfsRd$+ox2YWvx|6P}oFb0}FW$
zI`E`T8;D^2)P*u<A)zVGIN~B9)UYHX4HOI}Cykcth;y%&aB256rbRj{Pur)}5~{SR
z(w#0Y9&p-{Fb9nbRuS1N@@3<Fv`;TMT=Gc@+{V|Ix%eHps2-{C<MtW3kW`c0Gd1dU
zwX}Oxzc*)}S?|#onvFIL{R~>(U*Or_5rV^d+vTaX&sr$$Yjb~8+N9GEcq!yZsDd0^
zvd>;90&?FRiCP2ay6_`{dc)KhGK4*&x<F)trjG&V29!1RB&J5-WkMttccwz-K4&4>
zOiD*7j0|<O=9A7S{^iKVb2*;1Dql|#PR8@hg%AwxUgiAKw%$hA*5`I(>M|*uwpbL%
zH<P{}f`>?k*Rk?C>{<Ev*8XPaXPXjkRf6C-`@F7AmH?ariw`<&^~#7dQt5!eny5>Q
zdGqsY+LkB#1{HH4>g&qY>+K8b#Tfu4a(uyQ3(2mPHxt_t(N`dfOcX3QUHZcA^AX;)
z$O(AdC7WGX6dZW28k<M1rnE;Z!2gTd4tUi|uXHYA6d?jKCAfcijf9ExK7Vn&M#fLo
z){A6xY#aw1h=|O}nXPl|gfFT4$m~MQp(AuOZ1fH!zL30DJ65L(NW#|NL-wT&Nj8sf
znVT6&DovxNXk=#2zO3FH+@3dgsrC+JnY5h-=(JC0Pw4xIE%xQz2xO(v6v@9x2&w~#
zl}0Gz?u-HWirjtFNttqfCnnHOc}6&F-oCPE&GaZieq1C?(=kcJ1&j~utLhoervfG)
z!)`!ZEQLfW<y;Y0ko3i#-3axL{?@I|Rc*8*Focw~(?R^4wq$GPN1TgVGNvSOLTrn(
z;rM+HLiyZo!jB8TN#GOr=+rF-E&?+#zInv1sCzSABA6c^P762&J#TPD^hoVjKAKJ=
zRR$uWzCEw|*05?$Aqh|Hb>OL`NQDD^E@Fqq*NCA=lhM3s&u`pf-HdXxu*cF8Qz4ND
zwTpX8yRjE^6AYgT6x!>w<pf7$Yt~zmIcl%HutCyz1pyYF!eih-VsTk0DOoZO+Lbk$
zG$v=TrO=`rBMv-y$Mvtc-d@x?jTMQMBm`($>XnKwq0R*O^5TV%B&O%6nSkKMW!r9Y
zToJhkTNqk*?IraJC6;JaA4Pv|Z3muwAmRoHb+wn)eNyTVpxQskh-iy<Yg-Ayba16?
z_EJ72xV@}nb`ob(kcHmB;QmM2v2(A4$W{W^US9LayJHinOm`3n2a<c>9@zJ6<cY0l
zO_L<>Z6)=3;52DJzu+~mwX2e5zu`9LS8Q@3(XMVAVH=d)yQ{l7=+a(MW6hwxLy)LT
zPFv0+<J%LXjH=J@upqH^O(QCtAD*=$n0y+ACQG}O0dV<4&@+M``R|lnyAaiRm;T#H
z&RqX5|2j{^S~seMf77mO$QGJ}x*Ijx5jub*;@m-)<Ms8Zuv(#zhfyM0TLu|3%P^1|
za-R~zK5%{)YH>iTb|^~5jrF;<cRM4Q1L-h9qH~y<v>(7VdHP>Dd=l&kGc9+?nA9M`
z8_5G`-(q(N@~^6QYIn)>q?!^Tt_L4oaNy}Gn3&U<QLpY6NTmj4D3Y+8H7E}ei~w9~
zPcXi{rj2tj?fg#V<)*%B_StJ|+;M_@kmyI8wh)j;wvF5Ch9gA)6LXQdP^Qu_r8@AS
z4D#!W@+k5qL@xX4#JG*KeU3tMuGD+dez#M}+g5h3m9IJ-z{{*T4qUwNwy7ON_1UWI
zG<+|@37u;}DOjP?)>b!}iz=9u3m!NPi42_2WjD2H8|%xx&1LjA<rG;9<c`JrW@{P-
z1of29^PtlfcN5AMIrNcM9lp_<YXDGm6vGG=hZ(6zrbfTl8$3>yhN7QG9+E9cwht62
zyQOG_7-hvW@|K}fCsp^<ez;dAeRv?I7@d!OYc3GSS4}#<6D_F5KR08ywH`yANG^{g
zyXb>=kWk~oxDH&w*ROB1VR{Dz%eWHa+<5g3c6<K3jlf@iYRXY1dRu#TM>mVI)&%wA
zT{0K~(S6LtMJVFhZkC24lOCoSJGywtf#+LaPmGz}S$8i}dBJ@h7@|DSM;v&qT`2v0
zK@uD3mA0;}9r!73hPBIS>o|eJ)aCDQ3a|F)`i3RFU47@G`QqhdCH(}E>kS=Z%~Lmx
z^Mq-Lsk!c2+mK6e(`_3!IlqN(bCBnrZLCfv-HM_xbf4w)1`&ympg-+f#Q5X3scp(z
z$43e7a;_7BH>_J_chw7JGn4D1D<<JUf@X;yX-3f=?C$pHbt|rQE*hDFQZ}ZmTnjPf
zBX&=>Ji}bC5PsQbvbF+`55&=Pe0mRL>W$SM=~{|3I?T3@Vtd{oWeNq&(g=FV-n7s~
z?r3tQoa&T2n$uOaxraD7n6$5HLk-a-h%{e-<T#K-l#LwkX>VROF0tU|e^SsD%C!S;
zhyuB)5fS%!-tKMFv!S=u>w+E;EA*BQg;EZuIEm91^S4a6V-_wdMi-KC!rt0_r;DTq
z@m0<YNIWUB2m?@6j1bPyBM~wV+1KXK*X$Z0wqDjl8sdRw1}1Zoq3;XQeh!`D>zbSw
zwosuEIt!FWBvCfWTrlp+5UK7Mg^20uwRuH0SE{NqZzHt9foO%7c2lx%I?iSt*ybj8
zshS1*a^SSYe2PgZXog@>W?I&?k;L4Y>5Rs5t~0B9iUS)E*~jmf7CG8x0~A&DWgQVK
z4FaRBu*4f&dP1a7uxf&+^sH2|emyTlm%vT#A8;kCiA_H0z-7LE+3Rde?&H-wDC?0P
zIO;S;_k#nobjk$DG=!~<c<MZfi_T@;C^b)HNWBqr7;AWQ>&-Vg7eYpY!<8j7-avCx
zlippdp-|6#PwRHB*)s2-15cR)%afcdiiF=*T@Ww8U4lO$D-<9129fknlr3@w&oDg)
zY@*%+oiB^>es67YngCOd)Mk@)p9Usw#Cx1JjYY{xv$HnU^;k;`*DCJV%&HBpO)10j
zaU}Onr@`eq-4=Q|aVWDN#zP05(gzt7)P%I{ZCm$tAwF^y^C9P_VrEXlo8FdpT;9V7
zE;(0;t|YV&zZfM0X4*T~O^oj#&Wr0<#3aC~Fd<om8ecZ(mNyL1Xl%+d5TR?3oNsYm
zc~0a+o~;o{-E5`(Wb>-RQ4U<P($enNkK0@~duhTJnb$TeN4+<A^7-sv+H^9`Xh$2#
zc1m+$oaVabL-{Lm0G+m-P4wS9PU$tOLybC3aT~iQJX`xvQFm2W$hF~7?Ik6gbmYQ;
zCp}%=+tRoB#{Cops-;2Eb{doTp+l*z>pnK6d^9proB~tt&sU7EA5$<eSK5M;EA8#=
z)sVKHbCs-Y^{s2{>)QZ{(ZRasT*p#L?TxDJ@d+}E>>cf~>#n=bxrhNuT288X$==zm
zGBH#1NLlAGm);;voK(ha7NhuvZvE4WqaCDubkTuiJR7M-RPZE}$LroYwK8^A;tCMq
z@Z}HCmk&Uh-_<BH5~JdMyBJn_6=!#7?>TqXT0RW%g{Uk(Xb+?~@q!mz;rz^vv7K^P
z5S4eg>#bzLvNgCaRyj>}Ewafs=CfnHZO+e{9?XPn%10JY+Iwo-D1ZBibCp{n$ceVU
zd@FLXeN#cS0!oCs1aA|eI^{re;jX#yYWwD#ADw6$=A(hrBD~)=pd`+>ba*fFtSCHL
zO@?QQHMr1FmNuQa+3aO#&-UJKu61aioSSJ9KJGvyaJ}QLFShs9M`~NkkHNN<N~Vpv
z&w;DIclPIIrtMq1HNm^blvGj*k1VqVIClkxO9%spVsdhmL?7+j8dj_%25)|VmQBkQ
zT6<VnM0ksnq~-)Qjf9adD%IYf%R=H+kEp#k<<Du;wi9J#@;{JsNcE|{*Y6c8IGg)}
zwV=%vM);l67E2iHh`v66W~)%C6ZWCH8xfPS#)-k++7cUOwK{DdZVS4G)3mtA{Xa5&
z|Ga&Bdj%#XRrmoLytPSOmxzyD*1n?=AKAJm`m(psR9pCXIPYYt4c=?t*}e2cP|Jf$
zr<}IjXG$lNJHkFvD^=7mNR+fP1Pu9O0vk>_ca?Nh<V>y$L|B6R<<7M4s$OISsDeTi
z`hbzuf#et8HrK;DMt<5yyP1q*1PWU@`N@H3lc^%*)PNFa_OWgV3n33&TztrBtA&t-
zReMQszlxvu?rtiSjwK&~JQ90gdq8scuEJYG_KAH@Zf~;X<*pkz*Rh40jE`^HzPH*W
z6shVsYw4}6=}=mf?`zK@sHmveIjlFWO}eOT(+WBN{zei=C4JDjD%_V(Np=PWU#Qq^
z><0URf=jEA&iaT9v~c}AtLT^m&mJ$HYPJCT!6r@JM2^~&a~Y;d#j=S6>xjDCEUNV^
zvk2tTc4xQXBD2#2=Smy?HCMdQ_T=uE;uLD<cM>phYsRR;`--eRW_$Ai&V6-$#!V^a
zbuv2}ZC`b^g&1%QsnZsBYm3MYFWa?f;z4_`y$95gg9}ndUnM30DI&fAAxp2>{<f1=
zD`<qPnwFE$Volf)d#Kv<c4C=uiX3p-a{9p^89#O2t`Z(~wg(Gt2O2y|K_tl(t|>?v
zv#t-}3N?|KtMhjl7vEIP--G{cTP^ps37z!B5$CRD)0)Pnl(-Mo_oXwcXBQDhRnSoy
ztIjY6p8RTu4jG)c!%18?3zj$(E|$GEU3U>zZV$Jq4c%RzQYwzN6_A<`S*@|61t*YC
zCc0%u61PUfYBg{!;}d#C1aw3cm~^tE-6yj4a}p$SGaN_|Z`L~I>>Ok%*Ih;&ncPmq
zU*g{Ekq%?d%ET{xb=qpnh#F`$bLN#=EVs|Gu1_*5rO_}WIEN5L+iC7)C_KD)#Ev(7
zQ34UTfw1bdCFi{b+uez}A>+N`A*U@(KE16vWBO$4R~g?Fqn9zmjEfu*;7I~kt3a~!
zu}9k+1$jUt4I3QNg98ctZD~u?WwKN4{Rp7wW%ORB_;fW|9eC}E$ZL(cc&r<jRw0Ef
z+HzdG<|piQH!ht6B?bAo(^gyZVYrC&icD!aXpdJH#2I9-%YqB4?WAQm?!eQ0qqwCW
z;Z?%PGwpk;Z(ZkHh_Gy(-DYR&Woa!9g2~z+fWU?2!h!47>#w`c&UH&FS)OgkB8Bwt
zKw{9bkw5_0fSvDN+(Kn__CaoNkq_W8J_ri_ndZfjrTIelosfX=l_*>Vrs6<KFqHLg
z#x8afdoiRaj#Tnt+cEVIqO)KN9kstmv>{i~+on3JKF6hwJja=U2!JAXoz$k$QnG=&
z0_i2OD1|SJo=4P#L`Hj}do!g$7}F6*D{Z+N)12t6a)7Wn`=Pp#OeWbNqUsrKg_+X`
zxk`UoY^aTGb93LhOy9@Iw4$aJr+GP7=ap~2*ZHY^P|<X@)F^Qep1~$)6haQ^<kP9k
zKUqDI(nZY^=|)al%0sWeb<KMF;d)k1KY42v%^#`K4kXpRyI0v$-Fq%@S5{A>P4KI3
zxatP`k>vW6vBSea7OB&qkI1o(<nyE5%u%+VEuvsQ+H^;6Y8#>U_G2~ol4``Fq_vcX
zPZFyG*T*>0;ePw^t}~~_so52eJ8ik2z`>JD!GgSk@<P~8bOV(2y8IK8A00^1;o}ib
znH*mPOPGGtelk~NlSHHn_@SvD(MdF|f(mK5LZQV&_ET+cSCZ!FT-7EZ*@t28i>*M?
zlKphOr&ZIGPq;#Jpi}!CxKzxqd!_wM4t#Q)qC5P~m0LAUnTVh51|}Pm)=yC<puGc0
z>T}=pto>ZOjTfuVcZPm*=&S>Y!6a|WYKqd=TMFaP*E|!P8Qnv`bJ`M(p|7KqYZ$d(
zXm+^497eP*DvahDyHq5J_ZW0zL4L8lV*3omkEV#3rzP(V<wRso7DUL1{ZhMk^)#j0
zr#VWbYvBNj-n=OMFSj*GWh?1?*xeFOJ2<GNWH@CeO3;j0J?ve-(#Q_yk&lcGoa>kt
z)(fe&{zi>KQO?8mtJzyTe;-6f_#i?Ws<BjL2O?^E&$<ovYi$Xh^Xiv77aakWF$LF&
z{dzaKs*oDurqfnPtn(C@0q{zhA`jYcv>6I>Oo$PQPY$HQ22e4ONjCTGH|yrqGm-_%
z?{?OKL`RgpE^Dg1zrR&&65{1gILmJ8+Auka4MNxLx9hFx|A8qYprxn?B<c3#_{hxk
zR{Nc{Xq);y4mY5-I4v<S0xIBXS@R-|s=r&16it)c!TCvZ0+O8nratGC+waw?Xg;(-
z#a2*uF)*aM06aUf3g$sw_up@XnWQk;&LWnSMq%$gD>vSH+nu*;T6d@YL2jwod}p1j
zLfSmWs0-t^Kdf2G{So@7x3)AT$ZvMo{-}+83>6GIKjO5+oD=yQ92H^N{<yVaz(^y+
zNvBPlwl@Pkv_GlQV#9`m<PvHb<krE?<qe*-m-e&wRSQlQjZhfuPxECf?_78HO6MX*
zm9v}Hy>yEFSsN9CjKWQi6Uahak+?KCr2k)aCZ%N_v_H=guEEneJEyv3PD{K$HMNZI
z1)<@)e^Kw(mz8<;Aa>2xwhwF61)>dzEwH~#A_uO_I=^!bLQM%BpD*M8vcGB&`q{)$
z-n@9&fh1MhiHkwyr|qxX$RGz2YB?;`TboF=!avc8?Qe4ZSC0`bH0E68NDwcTqp2X^
zv?%;-!^VxAr`~riBfMGkR%ISkb25Y&{)Rv_)T9`87Sfb_`FHu6m*qdsWjlys6oc+4
z^T|<7qxSdJ`wW1RsCL|GYs9}dYyZ&3$6is`KXyu|ElC0Oj}Q%2Qtltyi=e!sJp6Uq
zVu6tph!++H)iKHRRKceIl(@5g!@5oDye=-k`^k4v4!EXAQD`!gE5KYcaN2=L>mO`K
z{DJ+C+CP_IN&N?Y3bAjRn;jXmf60~J#P)N33$Up-*fy|#?G{&CPcn=QBI0Hpgeyr?
zs>Pbz;os_+17Xez$vfu%VFwbMF8wc|YuLY6o0NW~{gR;Kw8etUcct?aQrKqN?LX>n
zMj4s_0r}qARP9A4o>8I7{<ERpRt|$!Ng}5u3D?H4KJ_I2D@VMv`;BjMepk01k4)O*
zusFN)KlfW&(!+`Az@v-Fxuq${ttH9k<1RI$!-<+0StL`>ObZq69Oth1CUyi_k0G3o
z+Q(NL^-|bL@;je%+KP(LQ-x{DKB3;eEIOa>N}ZIWS85TbL^^kA4Jk9b@(%mNOChz~
zai8-`%}cviuU=`NbSVYJ6*t`A{I1c2-Afq>yicL5pIneUv0aPQ5@CP@!fI<th=}?w
z1afPKVym%F>EMb=NnJ{sG{h2j>0t*R-;5m(Jk!mTsqv`=(YU0A?UUw%ezwnn#5W}y
zWtN2U6yzNU1BQLtr80AEo*`<3U<(>@)dxF!Gn4k|IRVJ4)RN<(2G9kQ@G>NHhK_J+
zpK+<k*Cfsmf*uA=OD-DL2<-#3Zu`uxv%_@xWWYe;ZalAW2*@<-`m^fp<RFqSu+bp<
z9C09pgv2Flx(L1d?6&wgtH-x&ajt9l3O~py3tFb~g4;9N@JD8PJ16?1PQ!~>a6YcM
z5ftZmtjqMKMcgC|_OafV>~k)q>UraJUOy+Oe)G)goVFrssS<5YTdZGhb5VSWW%!!x
zbK8`ZpEHUgNdR;p(ie1)i9M^`MYyM8<A>=UKw>JyxpHU^+vl~ZK;)bvevE6;-7#`&
zh#&j>dO~I*rGx4m9NMCPj6L+m2kBpg3D_6ZG)f<lnN+^CWrue;@Z?*F!g%;7&+~=(
z{-xcc?F>9h{3b53F4$?ms6Li;T$aBKNU6_6P4TKP&MBCggGKW@)>A$b)zdg9AwQUY
z*0uYR8WtfYZ16Y*#jA}l7&*MNMCq5-*HiwIMX65BbRf&(XSIF9c*%xcmP}w@*6;>S
zfjwu`X~<6ZCdWq!P5AOQBb)n}rss*?q$M{GrkjXAW8TAf0j?>c=PPoUxb<h9-y%C4
zXoJ%EmDN}AU+G7e!bT09_Xd%DQ4Z5C*v-DGTi!@<gH&Ad))tOS_JZuD&u*&_+&-gC
zm=lyLAZ_P#aVUyHkc*jvVeC1HP6|y$bLT1$FkLE{h5X#E4?6xc&0w_U7Lf6X9(TH)
zUD1ZWYvBBj;ggR!WzTC<Jvu`Q9#c(2tdKS#WDIRc(qz;HCC>;~r(EFn{Pq}f^9U*Y
zPJ{5DZ1VZhzP%tnv9mX^ZQxua&ASItcdNayK0xJx@HQU_FS1OB5n;xv$uh)UC8hN7
z52(DLU0I`p-$OREwIWf4l>(BBABuyp3XKSooKZbz?M2xG6c3JQoNzAN<b!zW|5B!z
zZpTAnFYZ2+@xB*As?~W@Kyo39eo>0XUXo*_m!KgRaYJ*eU4X{Ev^~U(O7})cb=tHB
zvtyjcW-se*P~~Gy{DRY_`9w&fV1&KATUf;(NO%vU>Eb!>AP5@@5kdCm4YIwhsJ9}{
z@k|`y>$Jtl*%_$Y>Ld}t#D+yAD+K<kdM3+ON1&?HR&?t`bbkSvL`8p2+tmeWBJCsS
zdg1Bv1}WQ2s9|A%wO2I6*-C+k-sF_ikT@HT4Az!q5FvcnuDMi3Gip7~&*%g=(38eQ
zrY(I{Ubwb<T3kv>OW9)`NDNhP6jR2O0bIW$snD)VlCmMrRwuN8YiZ?U(>v_?*2kGX
zr{m$$qajumrUK?g?Jm2azAQ$gh^xoMfnh5997y6oNQ<Eg+KtsF--%8g#;oXF8}vgH
z$L_aRcBr{L2=aW`M&zFWp26+<5~JE9jV`SYd`%V<nWSD-UxS(<f@S%;)}|ha$&lb>
zudXrdC-hKt@=wKHM_zRxF?(r^D!L=LJ)4ws_Utu{0Iwr6DNd)Sr7-^yWMtJ{I&ZH{
zs`H|^yzsfsZ_V#b*qUC~O-4q7({XQYS#DKU*9-R5xk;9G-@banxd=Rwq#fDv2(v*^
zZDo5M^1Z~fBZ#TAE!*OWpy^E#H+tAs<@SP#XtQxu+hVwhq(MkfCiK`=*B5w0m>n;%
z60~UtbFM3BcvHP@Kfc$~&Q+06MQ5b>FQ}S`77K=Zfq=I)IfXORs_q>J3(^u(s5lKF
z1U5>$xf^zOQU}XU;GP4iA$ElW7<_mwrTo|}jc2Z2ch{Yp-sD`Tp?1wZc5C)G&N-NI
zuDanXo>JKdVuUT?=eB~g{9%MrlxgZ72jrS_;JGJP$C+)f&zT_x20tYi@os2RHy3HW
z4V~R?@4l@(4Y5rrSaQfoAUNf~i=~~XSMqi|z+bkZ-O<RyRDUPpu-wQ4r!hX)LS5wR
zk|hoKLn@E0&8>~lFsHbU5vL(#t1UA)O18{R*qya{u#t%0mJd1DIe&F}E#d(B5_d(v
zvdEAAo{oY)3%0KHm{nFN#wF=fD)Zq$>RTv**l)Jl`dpJy)l#qD_i>1Xy`jA$h~hGN
z*nuRU-9kX4ZD>2>*11{dA{Y3jFSL#If{+#Ha@LeVg7FR_y)<@<@=#TxWmAWK3!ws%
zP2{wt<3d%3r{;*YCeZ$_n&f5YNQ(60ItX{N0~bEr)89<)Z{O~2?}a*-Y1wJ3-7H;N
zds=Q!?3s2?8+vEc5<g(uIFLjoCcDyO*j#&{gQ39L8|ym??UVA0kKJjL7@>;I$_dq+
zzp2AGu!-9!jV=_44q&eLA$*9DCBlM2582mL2cx@mT|!&%T?dkr6}|v{4h!R(TVsFM
z*!bqig&sJKwd0U=nEunqKv^9z3HZIaloX{p<@`?BS6&@#CU2={BhH(r7Q)DI38J2o
z6Y!!U=`-W36?^Nl>5k}0Tmkyh1qTv$Yd_=qk)~o_+r8mI$wvtLN1V3Yhk=oLRz^gE
zSMBS%Y0p9oXm(CpEibVbd`a6dDy~|u>%EFwG)R<%;+^{(csv=VZsrv)d5+6Ax1OTy
zDS`tA=CtM9tnbaQ$5}Q~_b4esHD1ZEa@w8+3Q+lrl8*Sn=C(_X8mO}4Xk+wLxWJ~?
za~g7ouB5!;=HB>y)=#$e<}2=XekN@RAuEZY^z#LVp@`Tm4Ia))S^MPdRFY(n4+oxG
zcfHQ6+}gU1jzfQ(>s&U7tOEChYGWNK%2dp(KbfwtGt!(wROYTEznt3FpRn=jM#Z8^
zH|BL+{*^a)sPxgb6kG_TfPtR0`<}MaRR|QOzt4e$IaB#K3{4s|E2Mp&*<2E=6nY@f
za1^Ma9o}x+S_khQhg3V4ImyUc{m7}EXg6%^&5>Ebr9_1-R9u^ED<luLBIIchksHj;
zZMLZfL<!kS0Y>i{?i?3ip0eo<oA!kWm3XMJ0BJlrG;{^bH)&N;cucKrPx?wI^q~>w
zLZRgBLB^Gtg!{n%x7EV6{E-vbB4}Gi-XxEsK!(UQ(}s!xh_g7k)Nn|0n72VgUY0?*
zp=6zUYg70~vd4oqoAWoP(qkyyQ=?8xk`ta(*HsKy&Kp^qp@TNpCTi6!tJmGJZiCaX
ztbfX?lTb>bDBID^m^6vppDAwuN%nHvX4~0Nd96&DTouKMj1uiS+m+*-rm7c_2hOD)
z-YWJnU$={h=oqMz5M9~?kMjiYFL*f$)Gcpq5p@UZQ7GQt-u7GQ&53prEgB&N9^KM+
zl-P*qU|-*GPshoRs`Iqu+@T?OYzHNdNZ$>jZ|^90#zus}26%UVrUOr+uadFTQ`;w0
z#_XLr;j1TfP$?D+ry(X>K?ACOaVfMFl|}i6W)ce}DHKrPIs;5TES%KA?L>qhXNk@)
z?sc9B)J>5x1N<5_8k!#cQso`QX_Qi)6kkkh>JL;#Q#Efz`Q(^FD`Lzt8NVwR@22sA
zisU+%t@BXxgO~?p4ThQ918q82g?@&rL`&-JY{A+42Us@k-Sv19oW#zs_Bas9(9?q+
zd0OAtEtjOfRFi2-`VIr1A!qYFP2Rco71uZyN%ttHHND-wsV(HNAR}*436CgY<_*qp
z!R(f_E%77Oh;MESiJ&t@UGoshX%x+7Pn`pMzNI4vd$u_j`T6xoa|3&Cjgpl@|3Tvj
z3k;TAZ;-TxMdDWA*+F|>bvaET($>QKhbR&$)DAo_m|aYl)0p|KZOTb>9V19&tJ4w~
zEH0xOkjttfD<Rsxt$WKkv{zY2E;?<RSfVZAds97of4xJ-M<q(@ul7x*SDka<(G!hM
z!4wyk9<&d%2_yTGO1YzfD>MeKq{^jNkh2*9P#?^x#F9gedgrR{Q@jORjjqR<*oSg}
zSU#!}nA4JscgV$}b5iTiBQk!tI@byY5c%${O_L(`IH_9p?d?73B0?P1nVBlFrb#>M
zz!P^RAaEEs(%W}*bC;%nNzkqBKv3JzknKD3zNOvx)=?`s5o*+F%h#>#vr;M^X&<Q`
zppruX$|KAiRV1kc*ED0LY)3#2?7NorTG1{l7$~|dg<YHsc+Mo2PbHp_7ZT&yXCG~3
z|JJ#|X7LFciXtjTSJ~o^+sBgG-rV2m{1mdUo!%ii`Mc9ivumsKOPEf{Yzm?F?Ry&W
z=ZNl6=Sq8+3U@cIu<z}rOGg8VG+l7ol6sWnC>|UE)%JZgGL1$vps1_z46ru)9C$pb
zsP1Od)h0M$-`~xN%0{Hto9cbkm382fAYT097uyeX6mMANEGef1cv(lVWjlA(9$Hp+
za+zW52iusp4?te7qGO;<IK<^^jGzVE-R@KBjdPW}Va<ff_RG;fusz)(<o0?T)rPhp
zGqT7Xu)S?X*%k6Y%RiDD29W5cMx1GUZQfE|+1EbL2X{DEi5cR^%=xLn4;g+QtUic7
zisYd1dDj+<DDceoH~Oojyxxp+t$E$Lb=)3mO8^g%8l`AqUKgm0`_F*~u4!w_)r0NR
z4zvy_kdJ{$i3m~3fe6J&Jf!OTgdNPw%@k$FZRT7SB52Ddzmg5@pdD%*T#XkMa|73s
zBQ~wsaHk#ap4uDSTn@pQCI^xP1wNnArwS)0=$7741=b!ec#6<+C4$T$?qWE7@?8!*
z`XdfaN-^V*qf?PcM=n)FZB>xcX;=|oR7fi7D>WJ0wR3OqV(|qPOhB0AZ|7y8N5uTt
z(KZV3RQ*$7@Q_R#NQ?ptRoKGhvk|h$=dNQx9?2D;Y%(#(RHSBNoi@?^-UMZ5W_klV
z)^_jI_-5y#&{}&mmq($nkK6Ia>yS4Q2)Gow_U2`Eu@epH)=JXhDp~fbK2H8AJDFe5
zLEr=DqQ0Q?P?;r!*@UXtqYYgcfXW$jOqkOa6QoINZlaJyQvj?fKh-9PWoo%JqWT9N
zNTMKb6H7kIcOL8RmKRQ1B-)FR*MY>{t!(NkPF+9UNcgt?{N`zn+8j7dDM7JP+Bzz8
z9asTgs^CfD!Q*vTTxlvhFOO(O)#7yS(ssIS?K(SCUkBwPYevzEv}r0Ye~=|rUe(zS
zEi>$0Wp5OSt@?_@s(6DFM+@5Jr;`s$&^v7Bn&^P<nXn_GNomoxTf1iC#+A3MvGW~9
z$t-~sz-~^jRYXWQ@Y;t`4q&$kg0i`vv<nT7R-?*tA@snl&rykXu|33US7B1kI59pz
zV#d>q9X+*$hCa8;pVIU0`fh4R5&eCsdt361VaXdHu`oIAWg0IlKNRPm$c_@J^6nAm
zA|6R&W{$KTFUmkXqWw^PviVXq?OM@u-XJMV$}p&W4v8d-Z6?8q3T0(Jhh9_QN$`=h
z_S)r|2;K^F7XngIxV{=+VlkBo5zNh&8d7eYN0y4rM4EEO32Pg*C+l_1giYqE)0TH2
z)NivNZdW(sq4VzQ4kS5ys-A|&MC=}!GTz#P@FW4YA4yVs)r~8zbAH!OQ<iK9zxJcm
z7U9IK_@hJ?(pH35F1jRk;jPjd;qv*hdPe(KQX#{3+Ugk{K$J>E^ORF`%znI)+$f)>
zwU+TOf`Ird#@xk&IQN$6VU9X=3W@)TQoq84$g<+s!ag4r?0HlA(j-)8oWZ-%?Rg@_
zx*iq&GU4_oyR}jRl`6W54yaN!km$U&wX!1gH%*56)DiosCLxeiPU*xvte3{@B_-cD
z?-D2VPp2fOa&tICF<aA6J5i)j3cU1H9TxC21xcd=h8d<H6H0)H+HB->+@H<0TiSis
zX6JWxF_R!8kDj>xVqErf9X>n`IS8fPTbsN!It;@pQ2z7vp5>81i9-=C?!W^_DEJlI
zM2P<j4M$|)WdNeXw8hpS<_}i<i?v0y0%SOaHLE$LR9QvL^Ks{{nvN5>IYXvjK$iNY
z1}W;J%}byR@fbi-a9w%lYWw95@gi<)@d(C6+VZ7g>`9YP=;!t;b+>3mgcw2#XiIev
zjk&kYezp3&e<>|bL#0f(EyFuF>fDtABH97IX_gkj``5aO(y>y3O-`FSdmnV`0sHmp
zSR+*={70_l8Gkw?_m_Cw1^bPBfllh1aW1m28#&ux-hQ(_Sb`2B1~I@-`$plvh>-nO
zn-+Fg>DGnpK$7(<4jxPOZ`XT=Wsfd6ZI!kgT^ZQ#WXIO`W(lh1A_%hN;3$O<3T>*=
zIx`&XWBk8cn3gypjI1~_`f+6Ov;)t6^mh@ALr`#PGS7LD^Gmn|Z{a9S`~B))8y)^@
zr!5w}_-3r0T&5T84+@fBmGU50Gs3R}NldzJ<9hqUHa&$k$F(!jn$I{8$)=ogK5Ku}
z9$7_RITuMigKA}{%^!CZdX9(%uHudsym)DGzx_$=6Dqi1DC8>o9rYBU&{^+THIDzZ
zW=k}&%FwZEB;f`AIPiM+`G{kcp#EoV9f;{y%_Y7E5z+6)Dmj41xvRcOTq%Rp-qO6B
zv_G%si!ow`kkG44RmfiaJxjzLcU{eda5(Gk|K;ER1!8~EXz-xFRkc(4loD=DUCmef
z%VjI3DD<c`E}`0i1S#aOvcIaEYBPn2L45X{KOG{QpHx~37TI67sb75bNq4UFvTgk>
z_BS<`nTe}ee#Bc_S|{{^llHfbXTbzi^GX%dQesiwm#9N`2&i*!Vh7r^Pzhbl{;t{-
znF*i7)A*p%7IeG~L>6iciD2nNL|H##f1eX{m(D-gGCniOoYB;genmJvj!2wVDk0%P
z`-i8^jTou)H6_IW(MG~G-<!SP{;}Q1;woMCNLo^vLm=9Ud=mQ|29STsj^m6O9mh62
zI|eVSHl&-2yk9<cDg)OW(-N9}l9OSt14-I7LlvzV`{%aTXZo04=bh`o$uMNbnpX{S
z`<HennMa$)N1bc!+7T?HGxo1Fp)K!J8J)CgzjAWhg#BAvgvcHd8#?5}wh%V32A#Km
zug?q3(NvqBOYr#+vOG(2fhe*6s8t9H0?5D&BrDW5Wt$D*d)bB=tDJ5Z&C7!Q=cT8<
zl76x>$09o6BymqNjEQ$dAsj`NjX1Xj9vwt<)$z(_u%rA}o6LKNK%I3iwW^nof;Okj
zo+vS7<NV}yN?Oc2ElI(s$MA7a1lGUf`S&_M)l3NBjD39m?oD!p4!?6PAb;|kM-4`n
z_n3Xc6NS^zB-vsyd^&B3QX4o-%RcdmaJDS%<^<R2sqWAmnv>MtGQQP5sa;?Ib}nLg
zvZYExMtHMNt~SX(GA}AIlD5EkVG|o96u@Qrn|(^xHSDBxnrtAaEt%FzG7fbr&8OB&
zWX{h6-DIbcjL2UW8+eRd!9NL6b~Y-$eOlMevte8n|DY}9J<7}u#wpiepI+~mkSOo8
zDyF2=Be8D5f#)<J#M)<c(;SJyFe|D)5zBMbf#j5;nUC6Mw)=t+gb<|h!R&z)i9w)d
zo;du}KC9Y<Z>p4aig4N1CWa{1FS7F4?LLHNF$oIbIBhw{uyL`OjH=1AxiNRyGwK;l
zpDI-0nekzpl8zKK^RRu+6UE54al#^sCD*z4C+FladuG07X*V?lsLB*I=V=ozZk(Fl
zX`h?JXXdCa;anB|E0sz|Ae^*kbyFfKEp3m%?Yu#j8)ObB+QmMv4UY9zVngYc@`)e^
zpY#S_hE#?7;0AFkpWhaxH$sXCSL*Nh#Qc<fL4&NEn4R|S>Z%)VxYoXK*)&TvXD>Nz
zHO=8Nj<83i8#FoTiyGSA)0^4Jkp)geprstRa*xZsFOS;e_QlzaJNmsT=XX}D!Y4jO
z&8mo5urFyJ0|7;a5GMlV<sqOu4qUiK8KUjIt@fq$CB>zZML}4U@*q)29f$_JdVXZW
zzAVxBmbGi{+~iycuyqb+%9kf`=}pc$KZWX4X2e|e6}6(2a9%v^Tou~J;WzA+J@%FL
zVeLF6=x6O2C2<TP$HUqjCy0RD2kfhAXju}w(lvR$ttjyP4gvPEwJ7o9+4YQey&@_?
zYXWIymB-NNz-v!arx)}8ls%{GnKn-v8`7!^B-t=xM@?sL!oa8ndv4q9+o$h$F2ZKC
z{%*VCiIkR;<7cQ3<PZ5FEh-{j7o5A|?O5_cpI6k2!u_QOonYyo*RW*e+BF+)TfNe0
z&`R#;E1cV&-#)$>R?S5qkVdT>3R$7!>-K`SRcFWMCN~rG+l{sf4Uzz5B7NZ#r9XpN
zlPJ*WBQ$C0dlXx=D{D|f<PaBw=M9S`=9F<)EQ!3w8V-I>dr^Io(2*w<k|T?Ba!?>S
z1u`;3TZTM5lpNAtoNrq{GrdKTA)G@p;j|^q-#)GQj9Gh0PQ*%8r*kf9A`~1P(h`5E
zY-cZRi;18qdZ+__bLv4HxH<4=GuEZuFa3Z2$6nUvX_Wj|;{1uvrz!1Kyqi%+Fkut-
z9zo^`vC&@M<^irJX^dq~>$L+B?(7&Jw-q@&_BI)M$|a#eu(8=OyDG^n_S^Bze3i=>
zE0SNBQ_6PFCb+t7j6k|jADy<iTU!rz4v3;M%8HM9ML}3uHejj)(R&Ay3LEZn&dIfF
z>b>F|5>+lP9&y@Y4Hep>C~77CWQ0Hz&a`V=FPRn9ZBqJ15L^c$Tv^(^Qel>nd~es)
z-D0p&6c0WXwp<4in01IE)=KZcuyn+(Z*5U?UqbZ}r|k;f$~ev-ZzVlcq@mKHMPQJY
z=|KcB+6_gv-o1E=hy$(?v&aM+(|Kz*Ci+L9K>fb_5wx&Zwyn1b!Iq0$g$THb`WlJd
z_Nu&Xj_89~=OVhFBwqXUg1x$%co`JZj2TSc8${x&Hp|d?@S1w21+vR|mYWGkZaq8$
z;+coB{k*o@V8g%#1>#nol>?8?mkJS!CjsP`y{;`Q94#d!h4r4o3@Vf7AqTF>l6^$c
zLi6_3Z646nRGtA(G`B)n97wJlY#0@a+8hhEvd#5+6^lTGprLv~oog0Je`VS^Wvl9O
z35H7BlQn=gkwWI|(WNuCI?;6VoKkH1qt11pWT2U7wQ@*@$E(;)Pn30MOK(8VUtpeB
zV?-oo6Y9m9Co(T8-xRSAyr~N<<M%M)DuY0HLfdr}V$RL^a|3l(>*vNhSK2L!R4Z28
zaINz@ks?&!P|EGrZn!Fvbr2EhxYHKHRn+q(yRGi1NFrF=DQ|79BEd;?p4TV-z2hBM
zUhxv=BDX)p5cfzLCZBxQlkc+Ii+G9aDC+{D0eE+aY%L;p#hm6GNIWGYBJPcJB)cPr
zb=CD(UGMykZ%NM<L6Nc}96fGpYan}J0@T!PV&ffsvIy+Xwk!&~#wkWF?Z9a)yx7Mo
z%9JnBh%IdEs^4i@3AbZlOol;lug2iqRnIGTUAQfe>oMZP)_3C$r%4zXLUH5$owLum
zOC)d&>AmEj_w5b!K8hE~wi32z{XOnL#4N8~aU(T%Yn&oN7oDrxnX+F8X|s)O1EBA7
zQr^I6&>G{K8}Ciormov!(~`%P%b+uThLDI8W+5KSC9=C}Kp-pr^C6)@bO{kOcO!R2
zXJq`<n@L%*yK8YX2BzgFDbM!*Hh11}dR52WpE$kuUQKX;nBHw{XAuYp63DCo+koR|
zcOPjNR=e^(yAs0K8H3ZDnAqt(iv*~nh7dwRvQ|q-@4ffldr!W<nK^eK-sH{a{qz0k
z!*b7=``ml(Idi7Z9603CNe5oY1I1^0I2f5{Dt+s&j!ftfD0!xYTq|d|0heZQJ@JHF
zuspuG-7dnwOvwEPdy+!O$~^DD)9#Adorn!;?INJPrNj3umIfJc`cjOeLIEd~VV!+z
zTNn&ED(o-5n_<sDa{5#BWQ@D*?iz^HwstP0&{c8JI&Tzs{s!aPp5SMv?VkD#!)E3}
zh<gdD0Iyl<sul*?@3(skqF>6tz7jY9L`ZY*JKyy>ds{aqd5oo8$_1*RsN}T3<M}ce
zs+gau2IN-w${g+TVqftLvk10*uRfk|IJPzNzP-KPkUC~=8Q~U@Z_wr^9e8aPl@Y;|
z$d<`B^y-(=I^m(gV#ElJa~FcEufT_ku;p<j%8?yw>ovNWP{U1VGxX%U(y1d|MRU)_
z9PdV%Mu-z3d@I*)C&jgoJMfC$P#7C$$xKf5Y@(Z-IC&_<n+QjAAe!PWy}23d*Ig-`
zkysw4hNp?v0TT2r`RcIvk%QXMc>ayO%}yh*1JcIsQ{dWm+gN=%s^uw03h8tD;u}l;
zpikk}3d5Xi!vz=QJd+_e**K7Rb|olay*gr3?Te>2Q8oo>iF8h<GY7!}5;6!|J!to}
z$3n|ve+g09{0AIJA@nX3zYLDLzxoJal?P8EyVF;^^AsB+KX6$7m~PUq6iSZhAcmB#
z`pOevx4h4ak`;;FR1YJ?jDc_h3e+PijK>#0=D_op%#3UqN?{|gd}*)Ebi)uW6Lv;|
zZ}_#9@CQ5%ErgESZ1>g53(N)3Ry*lHM7<R@HYfVCx&7>l%P(~r@u|XQ<x(oao}VG&
z1e3FOH0c))pmYn`V(q;L7i_M*5lwDSEVe35py4?XVcYHOL>qMTOMPHIOu`U-DN=9Z
zNH!#l`6T-zs5cvDoR*+xKZuU}0nlV2$>$u05#V8Tn)&1Fsi)iT+m<#AD4YC@a?Xn8
z9d;lZ#)DJXP`BnQIIU_EC)v<SH|Zv#r%z(OnXv~NZiFV9+G*=+rvi{KVB7Mc%Zu-P
zwR0W#RNNQ7eApgreRz8$y2w~hUrcf)vJ1sE$&w3Vh3uVea0rq2JFUXLSO%H26SJ9J
ztUC5keXy$Ns{AL#b#g;F5HW<?2@suwp;ANST`SKLT85PqQU!8Kig?l+Je;r~pVCrg
zq?9jJKG1hJ4y{<>1t=;VY2{kqmzC?Q>iIIOhe{L3FBwWZ=`jZ`bXku>%e!HV?dkD4
zxPuC&-XIcI#92Gh+}_hr^%jEpXe7I%jvndUVDiH?*GlM<K^w{3=?gVfZmr1SKz9&>
z!^8uMA4?3hNAj11A5ouoAv}UkU$IkcyfJ(4t>^0;0&16nTBp4AsbC=N0b%qx`|4(t
z$}=CP@Estv(Oa~M@LbmE$T9_dt51%K43#?Q^aTMRG6tPRs%K`fJ!N0pCVPZfGm7{o
z>MiIL!>?<gRZ>2jrUWN4oR#7K_1(C|sG@ME)0c`}e}-T=a_isM4MpN4Yyul9eOg`2
zaZ!Qh8`{X&6VLFiadMKFp93iw1&MH8=P#fteq%jR<#V_lkIB<cU!r~pz;I4zSW@*6
zW9|LjP>)BX9!4LhFKH5@krbeGiyw;J4c)UxYgp2-QS!J;*$0Yw0TRDd)~QmLq~(S(
z8PxNXeN(};?L?(lg#9E;z+)*=v_}Ni)`L4ywl~MQyoeo!2vmvYPG6#0N|tab>Dpat
zVr=gH_Q9kzoudHVjIQ{h>|yX1c$W%1Yu8EaRJH?~!@jxhdXimxN|+{VJNJsnsl;7M
z+}!mtmTdAq`<D8+l*iM$U)owwUEuL?X&IBR@9EkN-`XvL;!_mReZ*T|Qj47g0PeMK
z>+X}-Emn5Q=@XH~*>;P4dn2n93x-_JObnc^WVuMnna=*SeMi?PJBSFy5vMQLe8nlq
zwv{nFJR|m<^&8o&6jrKW4EhSC-?4zNVa&d(TM!9{8B0GD8kt|~+!aGY&yfhJC=vVa
zWGW+(wnC-=?HG>4%&a;t_o|YcG_Qy4dveEz(S}H!^`y!}q3F#4@f__~``+S~m=U=Y
zv!37tr)~Cq9iD~r3H*7I1y<th3M6X?osveOueI;5VQU!j!5tuwyz7HNQD;FmF#CZ<
zGq1e(5~mTEHf$+Lxi7asB!byuu3=R5=Q?Ibeu+9S%O0!Yf(t8DUddw)B$rp4B$o2H
zZEthOCQ}e1j!}l21&b)0>%f!wjx5ZR+J+!!JL)lHJVUP2xI$lYn$3NRLD}QY=eD*t
zv#<gAFODL0y2W4~I`=|*qMKA5kShKv3X3S^Kx)NFKxdRYmiOC}ZLMyib`t@_PDd<x
znwy{<Y=yL)YO40x&U%QF|3r1<9HTGk4IAtb<5m#Ckm78+>NkY+gs%zR%IQlQ!>UTt
zN?A*pExQ{eLPE9GvuMH&2a<?&7Y+}R+C6P@I7J9bL{lFKJBy0dM8a>hy}77U<5XSb
zy51=@UxI@|&zKzB*RTbnk<*CVf|xRQUG|*q&)bks@Wj%{Mb}o)6ff<vr`mIFU%$#}
z)Kn^oN_vF+kg~po$<+?Dn{Sz&a2oN_49$upPkusTvK{OmvyxS>p=BE19tR%dfE9!N
zp*h&0*2nA}i9L11Tc0?N^mEDs+TnUD<{OZdcgs?a`hCiQBzkThY{LX9^gPl{pW<>A
z=MyR+45vXb`-b;LbUAbBp3dJalXS6d^Ze{M6cO8rtb9O(va1n}k0#maVY7=9^Jt4m
z^ryEFSNBXAadS0NNf#=DV&Tq|%=Ivzqr*+Pw-z{98q=rTekqd28hxJI;#|jsg}TbZ
z$=m&Gn=p2YBt&99oW8;X?f20A*a{e4?07*yTiGUq4}}Wt9i*6cloNqeownziz>cg>
zR-n_)jeX<7ePrC7Xkr%vzfPe$>GUM=)nItNuq=6fh^cn6!;4H(P>jDOMn*A;%scQP
zdRfys8JH~oBHq$5g8<s|A9^th*Y%KgW>o<yqII19+-s}R(Y4o3)i-C{R~Xnq81zAJ
zkaRv|Qf*qcPS>o$eHgq}@VK|W^q9*w{k(ms-KSM#U{qxC)+bp&=*s8WnYvGI92E^`
z<(<A1KOh%}8BtXHjXI~o71d}UK*N@5vZTePm0U#SP)mz&;K?Dv2sDgeDT|;N;}<(y
zbC6s?Q>vuJ9oR2{1WrlbPI|l7mfHdb_sjs9)JP+a$brXi6zw4%b9(8xy->dq`IPNj
zqFYj+G++VOa)BVH`0sH1pGq+HgDc3;lvaH?vl%0D;3=u1tE2iMtx{9)#fE|$p+?>=
z%_Ko#%0M~iNhFIb9kw4z>1xC6&Xu+%{`{FO_QUl9v9+8|iA~DC0+P2~KeFC_q<v=~
zpUBU%PG7t$T~H#$$ATaqvmfndGR~b7n|J#1aX0lJupjHDs5{(L6QxgOutywtZCeqE
z^uk1lr~P;%0&iO7T&JpV_ED@VKhcH=n}#z$2ah>@sbR0`O{<R0PqraPwZ)cSisjRR
z<nY#w++;u1em3Y$O;dG@c9Cb5mW+a2_S1!JiLJ^JlXfO&Bm)J1!r<rJ6(5q4taeY4
zeEXUF5S@PITqous{K#xZB(?o)^<}7(F!;37mm7K={w(Eu{#=`Z)pR(m{DvEF?)`j%
zU<G@Yw>zz3?@}~!R}I<z?H9Vc6>&(bc7o!}W6#?!))Qj4GsZjyrSn61gG<nml%zBu
zjas{lx(oJ8Ij0*r-g$l!13B#+k<QUzxd@(OT!h0FQTyeZ-;5H^wAqyR<n2(jj*@h_
zE1sgItRj#aZ>mIyI=_<J;{o9nNJz7`hcx$H3=GK#5^~CZ^(7Dx9F-zAtpiCc8^gC{
zzm{{{8*G}La2oOUe53yI<zTyJSW!s7UbBj|1S&@>aUp`C^6ubY_8V<3A_6I67Bm-+
zF&CUN<iKM+#5DBf%&`4thaXByiZCmuFUF(wm7KDY{uP|feyiXi=<s>MwnkwHYVrn;
z;zK~Bt>w(e5e)X*Z5%j<MS+n)oxa+tE43LzQnug8xyOwsi@wv08-?CFSW(N4dWJsm
zyB$RN-tvPfXhORZNFm}W$!C1lar?bo0hznzokp%dlCU1iX^nxKsy0dLwf%mBN4$;{
zV2QO^k_B?(4XzO@QCo1uiONV%u)qC5H_J~%%*n~6z25r75#bXI>IwV9T<QB3IF@4G
zY2`jeP8jRmH*0^?aOkf7^t{umuZqkbnZ@jn+lwIhLZXAq&pCaCF!F(Dy0w`k3jIlY
z-W{XYk2o#ud#bpuxxxOlexwzbO)dR@2+Izn_DV=RNM70k-H8YGXEg-Q;gxK-w1c2q
z`i3xE-zv?1KY#PP|GbmzB;q&Is}v3mS8B&jC?*aHh&uO*7!!~2kp8p3sAuP_Yd>cJ
zTWv{<?hZWp9WI9XNsiIEA5roz+tl94$;`8GJUZ&DvWDq))(!^t!o&Q1eEAvst5O~|
zQs9s4nD`u9IBx8(yGc$3Rt_5NjHew)E>ncpWzCkqY3%2^Tb%2Z{%)Gxh>!1Yy9a4i
zMgDiQeH9V~q#lIno`+Mq!v1b$D10`_JA@XuK9P{qki@6Zf&V@yP3C#dqnU8p@vJao
zaF2>f{6mxfCZZ~(%vg~mPJiATA{pnYES<7{Y<;k66~QB6&FM>XA!jj8Ww>u*6h=S{
z<^Pn8<;B(0v-ZzJ2tfc`M965Xke>;odD?*&&Qi25O%g@W|D{{*2$_aJvfN8utcwGW
zvM0p3v3N;FL-6Ctk(gXQKv8boBN8%6!rpHG+LmDpB;hfqSp!JSm*%I(9JGI{PY?GS
zQ%<yj(^uy6*$Ej>XXotS+c;N5OGW_e;ja=}fXA#x1vDPb**EqdE3*QD7V}fSipVFZ
z2AKh#sJx02R4PNJgDL*AO*@v(@deDZPS5$03}swYz@{SDg7ZVoDBr<#$8T#=?MxfY
zQ|vSHy)mU!-}2%dRXb;OQeWRbtnQP9BJy&Ky-i=tl;pQK{IVp?6ZWz*q1V0b{MR^_
z@wM7A&Fb~xXM)~fL0q@WX~=3u!JmPBM0+9<G~rnM1Zu&78z&t|O`5VycE0d1-0>s3
zJ_&(xvF-8JC%L?xvGq~iec`BvZtZaTO10H_j6I@O?W61EJGnuYrp{aGK$0Af#Tj;U
z_AzH(?DUNYOJZ@Eb7olFP)0e5`Z{;z{N&dttXYft!o&8lXOiX&%R0I~UK<{GZm0<g
z$g3cTeO&zlv<@l<B607^Q8vKiT?#fk3?%B;$v$l#-~Cc)kJ|9UiN?VWIPens?i|=B
z)MI2d&Zw~FhXj(6@MKgqwol9@KIei9oJ&1njqEY@NoRtnO=9v?><vw=F63;VeR8)B
zQA;972YoSaG4;J|+^p_y`;>xYhQP4S1SDq%M;vw5G4On9Jy|3~)EiC0>B~jCZf<s~
zivN9Dd*D!LM~F%eB#=vn;4JL<oPBx@oYcFu&b1^s+|-;gX`fMdi)v{rpj$hA`7%7Q
z9NI8vpV>_$k}mjwcEhc32nV9s8W||A<FoQv%ba3MPMM5V1E;UlP@NlrkN}&751<4l
zH4Bz#pWT)fDx)yfv=3(iZfLs$uSp_$2~mH-KIhDfm-FKPJ)H4>hC06uCb=T=iH$cT
zdd@z#=7eJ@VEysetB-U8%1Sc?E!5|A)55{=JVD31gi!84YPn{ymlOW;`S}4vWTjhW
z-04Xa+&~>BOh#W&4{LI&bPcpAY_UMn5|1z73+t}5*rtml{yLE4{PNE+n83=I&d{^%
zi|Ust%qP#!!2;<JMw&imQ2XL`H$+}^U?qPFNZh>>=}?KsnELHY>fN%SNz#yE2bZ%n
zebDNNH@J%3%b&N0X-(@v$f)-6dM1)OzlmZ!PG3$Tg~!>7+w7dYX?byta5CfC=+jNJ
zvh?c2PW#f_u*-|19Pcr3n!@#)$eQJxj&mDuouZwbX6)8-M6r=FI{ApbB4?Z|<4c_D
zq$sz{-e<3@fl_A__7I=zt&ae0iI5mQ_SmcP0q=VK1@}1DNt_3q@rqTKT<KTWV_=Qp
zE#xG6r%#_Ufkja{1p_;;n^dWgIAx$=?2K?R%l!}hTz(MZVaBGa5J9TUVvLPABHr5h
z`Q;;|ZY!mRroL(Qn%gjDd|A@ITY8(kF4uAR$ZNVK5%UVQ+v$^tN3hsN5<2aI4i~$y
z0@fRYxi<*I!4C$#kljVR#4gN<+0>hwaIUpD%Zry@aq-1=QF|eDUZFu!Q0bFJb>V@&
zUEIz5ksykiGWv20jYV3gUD7Qmr;ad`aIo{*V-BRmVny{x8a=c^g&5YQHJ3xiY=$74
zdi4a|V$<M==H7S~VJ3VQL|WnKFRUB{dRhCdRSRPSr&Y8j6CuUOcM<V{^q{|XdCuwR
z0#iDecbfX*>Hd^mu`<278)ZXALpk>wBvVJC>Mi3S-O;YB-$?b444P-8LvI~OY+G56
z6vMoOcVE@+hW53ncS)L0IFP_CwIU^+ndqoCB7xX@2K$|`tIvc1Z+Y=%4wdi6q4JXG
zoc>e@M#dHfeY>Uxkgg!=dBHvCt7S<IC5NF*Wif~S_S$aIRYWRG#ZI5FW|Wh65Kmv%
zmOO3dv9O`*IuKFKwY`lqc5U0k6ALp=BVHP-$(In@&Fh51kzr@AuSZt0L}8JV#a190
z11BT8_~x_rh8n1}nb2u42-26|f87)%JnW4*9*#bla;{TOsAZD;6IMj~@|PfD7Gq;@
z$YPOhz#L>zuH#F)oO|O#(!Qa)6!fmcC$;=1ZDi#`Ls*pt0M(Ur8HyNi%_Lg+IES2E
z*U+{{G^o=Oh1fQOm|&|C1mE3CY#Xh@=;1RjRf9lE>u$TgIDWFA;_|2$5*ZZGolJ+j
zlK2h<bA^xHZfJymFe`6--|5geL1#|NvDLXlZoY1<*X3)^d5zuJW;85u5iiF5*etw3
zv{16&M1gi~BfBYAgp7lpbDdQFiYqR&n`<>;vXWLLL3`^<7mdQM_t;m|eKJ8Qp5QU8
z6!gV+39qEli3v)j*lD-a-RYT9)YvJfuVz7q9BHn~1Z8XLZbg(o!{VPJQkuS!f)*%$
zEmzXoS_{S~fm0+l-X^;dwfnrmqi7d>;vV4`=Uq@DyS0Ab#03SbhJlj5PZMC=;Hkr@
zoQAlbP*8{5AGX`t*CK96Z$AZ%iWoo#q6zIi*uxdHu3kg6$1GeUOB$7N#4ZVobKqCz
z3a?b#kQ^dgiDY0AY>_~$h>1Q7Pr{)-Vz0faO+tl*z&n`uOif|>9C)tnSpR;zJs17b
zi!Z(0x$ZVli<6}eN($Y~&S-}RGUg@Twmw-yWuHGl+{*g4vYK;-iIc|Z3#65|_ki6|
zcVzsaj?CTZ^mPuJItQE&Jx-jS^cD1Z1$Wq;ZCl<kGu3xmC48wcHHK77I(K!SzfxXb
zV^A?AVsRaK{bsiuIpKDCb9*WgL*q`XC}JKuz{wKAMQ{nCf-v}&?puyyh@zEr4Tn9a
z1iZC=oF>d9q{<3Y{c4FR!0Yv}BR&e7zun!?20_1h2ts3YAY$p^4Oe*dJ@spCF{HE5
z)aeT`WV|;w#x3{ePsZ=t+M6f9+-Zx2UVYg$_O^OYOeNEG0#YB+sW*r|Q-;?S+3zdc
zmXv<O0`uEidLW5M=UjA=y}j+lg_+I0DJh55bC%+>e5eXo^=j~O#uJY^y7Yv%KGp$3
z%`Sc65gTi-3_A{c5rbhEEFDNpo<ee<0`TdlY`k0VuB<yB6@~<+^C<@&G$sA+Ej0v1
z*(TZtEI{FD#BEi`U)d525l&?NZrJ;z{X@uh`EZ|9Y59LQbb~{!qmM*zr%zmW&?87@
z;|iio&-a{We2XLyRlwV^^sG(R9*}ueua9DULo!FQ@^)7p%MD3TRU$=R<QX#|nb=hO
zy3wiIoJLL@TS+95oEy8ZE!nC>ub$JR(UR{yZuhsJPf!4gM&da2RXHbXx*NL6wPmhl
z>#GV25)f@u-6t^s&zK^Ry!8q1WTYb?&t_@~Gm@#*t5``LY;gJEvtIQ+7GDX?;_o|b
zwjm6TnUoyj5C*s%kofLo8%X0AO*c0MWNd0y@zorhLRU)8b3NqF342Gw8POcp=CRn8
zxJn&}rU(UU*5;B9^yU<#?bTXdoP>kgpuITEbQwj{`Xri&7mDf5cT-8&0a(5rF`quE
zK8mD^Vnnvk=23Cq+6oB!tuF&fGGJg!-4_IvO|XyT2Koy9^LRqV<jSA%xNS|saNa%F
zI+rm`I45Bv8reyEp!#@>s*x^Jzt8DQC3F>zI+Q2dRx2mEa3H9e*zO5f25cQGBJjjJ
z(DN8@0&?uZZvM&1otvG--A7k)o?vehO+@dNj7F>x3;fP}3^qlb$Bxl*^Q2cFFUn$d
z_;u}}HXJse0@<?Gm*UN(jh=AeqJ{U7N&2q598AqQmnlb3kC?DhRJnII3XfEB*lCqi
zqNAN#iZ7?8+gCM6=tHSxh|C}K25F(ta{h%q+&nR;KeMstT%XH$nB8H-j`NI=Cln4X
z`%pMD?LBP|QrL&kZkjb^I{P>emz3XCH69em`Ebp=$jf%7@h}2AeI<g=axjEFk|P<P
zotty6<8dPW3V3;5E5Jg^LWnYA?=846RCY0rZ{{z-lH9v<>8Lljvdk5F#$j&k8|n}s
ze7|Su3H$0Z5xzP4Cg)0PFgrJHUsKbpXpenPD>va}4}157eQiTP6hqNiUh1t+Q48AX
zR0+4QYhN`sJv%>enu(+R41=(*ujlCa8Kvj$Cqsh1_*_vTAz8Q$5E3e>-riRW`_e-@
z5%hcoB0qlOeW=Fv4Q&d8M>gRwQ(#Z>PT<-L%Zpn%MR_nub@U1Q#+(7nXua_{r<Fs;
zrWU5p_h-i}WEWkMuyi^!c{fpN`O)@Ms(a5nt^Da`CFyfe(>Jx*E6W$P=bgS>|2do?
zPuK@?!pVRQHPvb7W+ebxaUL9WZXfI>LMKqR806={a3G8EIm?UxKmW6DPDZ99<{xkx
z@rhC`YTfNya#p5wxDlDwG`m2W%0wU7x8`Ky@WSxzv=V#5i6g5ORvk8uNDdsxzAZP_
z&7<o_ooh)+*AdJ<Y2V%`4Q6(b9Bid0mt@TpNh0<gZ5b)7GsVFjPDfhHHcSo3f5Mz#
z-&t=L??dQSN>(gPCL3&b;KGp$E;!G=D<4NyGbPM#>^mLt{gPPZvcjZe-<=~|Uc7#E
zjnj~vzPz|`mV+g~r`-<W)(Ifktew6bF|qU0^Y*>nXfiNuC8|K<d&YqW@oegEMnPcB
zweMSjfNrUYhPBs$BqAaJo#G26=6!$LT|p9MEtLi&Mgi4B!2qs-&-EUh#d`b$jh3t?
z<(H$VX(tyfhE8Qu$QwGBEuMKX3vEzcXRvt^JL&{gbE>yD0S%$pOqy5iv2HU6d*q(f
z=r|<Yf#~f$0>YtO+gH8}HB?`&TIz?0hX9iEABIVVLLadmZLHV#SfV+nUH&K4Hc5#*
zOXLj2w%FtK4n23J)*_{O9Z0=n4^HMMnWLljL^q>y{RGwX)+eS<f`24&J=tbil$9^c
zYC+=N5(;)Z@MM*oRJVDvjQ%_8$&%d^BDNnlFMYK?hT<+F21APXe%saIOw%?*chgFa
zu<yX@Eq%&K=`Fjvo;jEwJx)(CENQmFWJOqs;q1i2^o;GvaV;+*Ms4YD6OnWJ>Nn|L
zj$b0&iHTv?x4msnSy?ImQ1^VY1IcYMJ-ua<?aLjun#JN$+k1mfNVNUgfFSoLoW>B!
zi9F~$ykjT|PZdOxf~;d*$y^5#OBV(yAfSvLXy`$i2ntU+?HtOD9HVCk+it{}%ke{9
zbFf+GDjsV0-l>c-TD7fCV0DcxWQV&?mN_gLp=>3m9Y~=ON6)bQJ8;S#=_W9b=ZIo%
z$SzV^6%+wHX~PJr=$t*>J%SQjArw62^p(hSNWL?v-@@^b7*{H<#6aw5K8{?1nck+p
z(~KFFRX`y)8dq3f?3ujrh8dmgJK;2o=IljENFTX4^+`L{K7gs!-coXI>AW)LIB-qw
zhTa&druOWa7j@VOh0>^;MwW7(tm$+s6=nP#c038aD!Dn=!i4GZ{?xQR*J$cZoSH@>
zS8VmveN%R#h7fWUFaJMlUo^N@UqmKCkH8VxDpVrjWW$I4419jdX{Q>F>SE95fKWl^
zo$Htdsn*KFJC5!kVeJ_^)gHkC+2D=Sj(ss))?+C6!cKQ+sq)xiH*oqSaNO8WaeSM7
zsIjg=zaPi!(jh){&Ku9MGdW|gc)4>GSe53-$d$FF_62v0+~C#VFc~(~+4=-#31rb+
zDRf9x5`dJoNJae9{+`(TBoeB`MZ8b8+>z3p0_7~FA4C=*@LepxxhqkUJt}!OBQt_D
zg@sRV+#jEPz+R}S3Qe5pRGdr%lQRaLVh3D8?0pLpSw#H@>w7=I?)o54QQS9uxl3;E
zjg3v&i;cDs#xqTd*|^gq1lzLJekiZr{d(t0il+j7?1vMfR80dP39Z_w+EI}KV?R=T
zSuip9q|;YBPXRM|($%nm4RtAMiBKHfezbm{XHXtw*vFi{r0Q5IMX$4rkNA1UaQ;~P
ztaXI3Y;&4{exh<2@TEC?ZfU3ectcEb&WYuL^cLmu1`&!KjTsetS4`VabRV=Y2ESwJ
zai_0)gF^b#K(%k_u>E9r_a5%v6#_!_X){x8CxdK$s!jboHvfo2q(_s9!!%vfDm3@g
z?XK&4IA6yX22PJq`_|sA_A?C;;i8dJ%nKtEkq;wviDAj(%UwBlm^zue{9BT_992JC
zYcuRr8bYhsw^G7!U^(~3XD7Ogc1D#}B)Ok!gl5A6#e`_({+E;&YowBNL<-r@x0NFJ
zXY;(%Ox>Xe4<U@qej$+<C1E!;-R7NkJSpr(JcAH`$jvxmzgXWTx*`i_$_uLFmZmD;
zwM5!Zxq-u!XTMax<SWD&BfFOzQ#oqo*Kpvm!G$uiEVM&@#YYaYnbP9%%Wb(P`nX2s
z(5f~&eSzm79(Fi68=7<hT1kPILdx%n7qVZeQ9vCb4GLLM{7jd@;|(q(tQ-Zcs|but
z751x5JYU-zA0ZBLVb1A_?I;_hvX3~f(0=XAi|fG`kU`mQ{|^J>cZybt%0;}YiFo3(
zGNoxxDZUQ7xbtjLlU2#Q?bnMokVKC0mzJo;n1q4fcnRJNS0@7!Lr&}yV?AKM*?oov
zl5#nH$!gZF8nNGMOeBs4nsM5xsZb_pVVWq8zWsLhMkQjUiC43C(1GYl<h74a+V6CW
zkiG^jxC8@v$aKtOgJ)cy^irAsNU=)_VZU3iQO$Op&-5Jjs;uAMAo`>ynYEK4qACaN
z_v$zD0bw3tlRlOc=s=3h?G?ie$BCr3;tclt?Yr0Xw#vRv`+TjfXqHj~73!Z`lOWi%
z>4hz~<L34U^@a*pfw$}3`W|iO{SI6+d2>}Gm)-vl8wpa`R02B&PEUnV5HZL}@c1SF
zC>Q%Yl>p*e6zZyLF10^S9<t^x=SrO$Cm*ywNi^xtJh%<Xik8rXgC)l7Pix^LF+M2v
zA*U}F{?^s2?9b{3Ee<hfP_Z3OU+i<3uZ(T<!4<;F_~#uCD4tEGe`WDg8QFovM9at(
zp->2=DRpG*?O$}u+WB}{aC}}4B!9WQgJlt`mu->f#BeDMTqD$iKh8GU0mji6zhUbH
zCCUP@a?=JUw7)92#D^ALscj}y!#d=^H3c_3z$q;D*KGh1r>cq-isjTHlo3+xz>@|*
zYUJUiAh!KY_kt|peo{`qkYQ0^UwFiUr&K_l(EGR9-`3YRO+5q+<)Jx{G*G~cFjt}v
zL9+j@ZrA(eA61^8`~$=dHwZR6tU9PRav(|!{QLIC`5p&@IL+i2VMW)s%Weo|w0|gw
zg2EIS0TTzJLs)<$m1gTnKuZDZ@vzM)GcIZUKh~#9Ov88>2;W2}#54z<Z>~%~ZMM({
z?Vk#Q!<Usz{vsUP=|F_C*Y&Z8{Bz><*ru({Wz75&%C?r(cloIOOTE4^OtA&AQl$=V
z=k>q~d5^LaE1xK2?SF0S3_>gLJL~f(u1z1Y%YkbUtFg`4zvVUM6la|4+%M~id;RzJ
zkRbjUq|)(uFcWxveIA6<MI_jNyaXaQgdEMxzXM5%9ggli)$>2wTxB1IbJ-_Jif|w~
zoGY%vue{V=mjz4ZP>xOEK#C2hs*ALZo%Uf%@rCzZc<#B*bsA_#rl)kO%**P28JzKF
z$YtjA35zjFQmm(EAHGy?KL7l4oks2oa8RlrCMf%erKak0To;kL6HZTROA8#UXdk(h
z+csU^GH^?Yc|KHr<;FMoOFaJUD)C6773`zxYgd9oL=dRl=s=Q)Fq%y1ZefC=!n^FF
za~!MI-+bGxqfSd>UtUDh-?T{wB2j=i=jQBVd^gvzh(-BJy~Gn>ADc5GLxhg5q1l`b
zgFe}^efzlf!4p#tKIpXKLE3*Z4n&n3>>7vd<GbmO{RJ)9>8-C(Dp|@tp?iX|E+lh_
zbwSA-NaDeLvkUf#_5D%~(yNh4JKEqXpA0-!O9CwZMgBf(pHzKuOeOR9Jl+xdv<EoG
zdYgT6!4)~A!UQ|;X3^IOZ!m>O{Ze_>B&_x+Z5_}9A|f8)Q6xt6O9zkv*Yn5GjqFn!
z2I}CD{(`TxsltGy{%LJ;QCO5ddD^?4n{k_L((!MfzEq5Jgz6yZ0T@Q;5?{RHn#=4n
z+G<TK5C!Bk6B#G^56s(Vw#Taq;shBNPET<n5r0jboUk&8q}uXXZ9?EwSOP(;!%koP
zMv`f2402_6S*!NhOC^qjYFN4kPDlJ8wIXpFx&AYjO|G8L$r)N+yyzb1I?!h*bte}8
zpnY!pGNPEtTBKQa&&UD*r#UmjKCeyAK<P8QzVCG9aED|l`2;145_M>wU!NGo%pv%v
z5#H%bN->%!{srCckQg5aOKBIQF9W%ZssEoNlmFLV`@;6AOodq5j>y3mo;QdDXGPsV
z9||ZaWnWYu!%D#f!g$0$M0Ybfs=@!8&4dY8&Hpb>OnLSB=Q)?cOO8C7urFz7vYAsT
zoz~D6Ru+voJ$4Fvc|n4O@yhOi7YCAvv?6Kpn+)&~3qGfQDWOYZ)IPKTA=+sN68ki(
z11Rz8fPHD3|5cM=_@o^pN{?s;_y(o2bL+iQdQ<^Wl8jV)`S|2;1D?VL<}WHWrrP|9
z`cW33j4(GoCZh9sz+nfTyJngiX7<W#QCSHOhto>D8!U{E=Mm5(D!;1XJU+Z}h^y04
zI1epAstOxp%uSl@U)_ycidm~jktSrm#8=>P>{ukoWV}~!Zac4jKjgOI6S~?v4t!`0
zLl0p+i2b$m+djH;3ZIVC%;^vDKg~{GRu2~$OVfu^P81#plZ4{~7ag9Mo!hX%UehLI
zqR*iSsuM!`kJBscx@+Vuqjo`F--uJtxz3%nK`y*GyRc35CbF@dRtzSHPmrU>?V?8X
zNAI6<8VwPGsGpJzR&jxLaeL5PCwsF_s~&VL)Xn4g?Jud{CW~toY;pHGkQ~(79)Wl>
zc4@wHR;d9_L&(4yDnIkG_Tb>r5ftzA<<H3?m}fU#o^2`+L4@&<(N2_^zy-?5Ej*RG
zuBhkkgBBfiTB+|`Wmo3q$o!;psaf1el89YZZw^+-M<c2_eTk@Bh@rQuTk9qwFPv6j
zpduU<%csL|*d@|Ek&I5-H64buGG`T|=0Ivk63r1&w5RR0b$7-aOS@GSI@!#D#N9G<
zC>P`ibmDc@mvy6Me|*;I)3=P`jIwLnNJplLgrJpsROVddPbO(#uWtyBsE_WAda2yk
zI6j35+vUJ>(qUy}i`yG&fKly;VGu9CM^j(XfuyE6-do_{O&!?v#`ZN4$C}J5P~RIw
zs4url5@1E{efiRh=L!$P=@>kz<?bg!rNu*~R!=zhrY<zViNC={8sFvIw5UMr^yK28
z?8{h?m2f!EhLP)K*VP1yQAuzo?9l1UFDBStKBZMTq05V}f2DJsFU0UqsMhuMeS8yc
zFztJ(nM&{hqE#47F3j5viD>6uc#m@#NX~g#EZ^$-dHS>nZ4F5rNzQ@9xOX#RWw1fQ
zd3=||U^f;VO(&xZ!uBGv7P5r{7qaL$NtO7Ijc`+2Im|)I*nlZB3w!a_0m+rSjw(&}
zQviA1ZqBW#gq<m;5vZ5xa9JqlE4r~u)RxB8(iDNj-6@1qtqXp5OZAD#X^N=m;PjQY
zaoLr&rZw*BPdN=u-lk1AZM3y{ec4qPU+i25Rnk_|7m(at>v(HEe~tvI=7sg@6C$zR
zQVgE5+wy5Qk6w4Hb6x(^6{G)<4>o^i>l!vpQkx0SIPDmNgscjBVGQWF_NIa;+(ofd
zlD-4W97wEY_Nm~`&OmG(WO|HI-QEb7=o*x<Pvpl&r&qYbP9o*ZM%xUc+NiLL$P#^t
zjiVSfH`)63nBF+aFiJ$#W9W?mM|>^A5=DvNJG-N8m+_uXu7iTolN403KC`&y0&(uF
ziE2rtc^Ww%%||HGTQdJa=dSq3EdHn+{ZybATa7(_#O_)OXT<}P&Q&rO1S$lw$l(w3
ztG&5HoveMTtXKMy_MUU$g%{adT5BEAqfRT{DfE+-Pjb*uaG<@li3@|xvRDtCj>JtE
zJrW;wpJ(mv>I1uF_&_Uu%3B|{DU5Pqhq8Ox<O~)tgCMa)DmA@{lW0zrr^@|gtJ=K<
z848VNLUt@Y<3O}-bNx{D>}@pwHlr%X;u_!U^wn%CXQG+A@*q|f!1>CgDXZSH&biDH
zfD>taJj#2BdbGFKgTvRu4j@R{=?e;k9K5v4dTlya&vLdTt=y5f;+h<*5hy%Rt{b9!
z_)8HE{wl^DxHjqX;;UY5<L!r2lQT{u#v|oU5oalH!<uZOEeGK-p|Q}B%Yir;HH(m7
z`yEnGlCtSEgB=Xpgqj&vrqCh#kZov_7VIXX7nnL#sf(&T4m@6`AXuUR_<PzmcAvRo
zxkou^K}-%j_3*(~PF)f|n5^$p)B`DA(1AF99aIHU4w9|7Ly$eD+K+qFV>pG8*XWV_
zan0pd+kLqqu6>1brOvhF+TEXDzkmctVSJ~lNk02EOOM%fQlhDGzUBbCpZ3r;!lu^Q
zC^vPGv<j@aFgr-3=j75#wB2blZEm2~iuRVgMqlE?l^0)bv)v686Jh6*Fyr(Emg+%@
z%2ti)w7Q6)K4F{lYkHeGE4n}7G-Li1EvL|ZNsI4jV+C8ne1N_cc3lUOvx&1Fulh!t
zYovG`-g~GO?J{k1GQ&K?nQ+>C1F>@);JtBz*rbh4Ukp-$N5-c_L=f)@8gBD70`LW1
z+FU7MhXYA+c=l0*x<}cZk1joL3vE$Jw<l_d>JjuL9lm91Zfu)vY0Z&|9&R(5THVpx
z*4Wne1zWLV;rpbWpE%Chj{3w0+H3@Al#_sbebFEXl4}7wux)i)Occ}AnF4(U64mms
zxJg6f)M<OL`%EQ`rFR?Yi)99q+alzbo!ouE-dSXVA})mI(uz9}ka(W_NvG|hHm#x1
zk!S#+95x~~cp_#hf0FhTDnq=wy(>|h(vaJn>slgtLO=`znIWOb-reSDEde@cFiuBd
z(s8gRn3{c64f!#4<PN7LDQ?}=X39|z#OnFFJ8!sgwR0^RNZHcqg$a94O-z{NL1I-W
z;WL?YAi9ryD!7L|+z8$c_dC}y;7l+`$*j4WR9EMbZc25CY(%Pi>(l2cQG~zt-nvhi
zDBlDH6P><93W^e_aISrI!x|2^Bf8UR$IBzTiH9d;?ut6t*R<6l+X~s5H9qF_#czZ$
zsuRz_hWt{3NF*2A*LH}fB6{g0$?7>kVyd5nTuPv!D6f58>*LwVTv8!m`f@ko*VFmo
zWEFpXZiGolEoZMgO-<+W;wnw&`#Ky}K44Qka3Xqxq~U89tW4WhT<;s|HxlOLDBK?Q
zU6n;@5Zxz#AehScZC~1B-`KtklGjqA!d>A&L|iYq7<=CPYko4`qbvXj-)LHp`I8Pj
zW<f=Nb%qdL0eiIigxjiI$?=U&Up({bYp%3!>V6}QE%X_uFSqg4S6^)(s2k)Jud`mf
z^<^$lN|qn2`xqR~2$^nnIeit_BCVjsE-B724D7rP3TEG&D=^E^(*(mi%~WmHF3hQt
z*|&5fJQ>+?a<b5uBV6Bmhka}N!P$NAqdiVrz_|!s`nH<c)XStV6Vf2pxdX|SRvd(V
zds~4ZYVraUO>!VPp4k~4b8g?!w%9puaIOnu!(YZnr=?eCC19$y@9f|!ewv|V4{vbP
z5O6_}^D2Cy0%4l&6ZTzA{HLYZoDA}l6I)**DF>C&xvP{xROWXq*mu{AX?b*vs)9CJ
zU!IMi(^-z#_tZCsYz+P@od!H%3V{?R?L|CNjoKady>)k^mGb$e>HtI&F30V_zOO!+
z>Y6gbA;D=f!UT@avDbkIPDlq->HH9fL-U0^$F@K<weL@Ubno5wI+r!a;Y2`Gk_GY^
z#v0m!{XoHU_I5=XCd9gk!;w`Mc)T!GJq<<i;IjD+eYY((f_7^Umo({kbSQ3V&8_xW
zUX719SK2rm6qCL^M>uucsMn=(NOrsgjdCzxIST+JJ|iW*GE1`TB2D|C#*ZgD2ID&E
zw2GifUKZhb!k(xn3~&E*B_*i<UW=xqI%PCJXiv8BpfLzFrOb$-?s6bQsN1&33hUe1
zjbJ+?$luVHRG&h=GlXc`uAGF0`xl(c8nGN`LeFPxcUv0BT07}C3_yty4m^Lub`#lC
z_4+;CxFXFY5>nG18(8GPQ*w{!=iLx(8IL(+!}d1fu1IIfk&#|ZSKO@Ibl@dX3f!!M
z{QDAT*HRdq8oGsT?>g7^w@JjFrE(UkO-f%(Rc5fj6vcf)agfiOI=c0nc2u%NA^LKd
z;H$KTuE>TRsJk`jNa`lBX{AqUzSM=?cCg;6pag|nD}_u;U-_gCyzn~d0m)<B)kS9H
z-n2t)99gbmNa1e~#sMYCuVUF=c*G9ZZ!}ZNL{%<g?sOp9>1W@&w0F#oEVW<}4(<$d
zyP~nThwvnsUaO|6_*plEe&-j7KO7T2(c3n#r*q=B&CffRr77sQhyYA-oDN_|n=CsT
zMp!gtdJ5N+@e-a7d!~`|Rn!c1TJ<zdh%6DxE|4|tSnV^VY(DR_l960La;H7ptxSeg
zbJD%jmk%nQoyRzx27`{*{40y76`B;<D+1>Tq)nfH+_`IrdZjHpkygUh9(%40U}};o
zyWPzh(V>TlPKu5qGT4cvKX<IZVbr<S94s%cpF?NMBQ;Ovx53W{hI3jPI`%a}XvuY)
znVp{9sIXsqzWb<S5hMUp5K|;F1V}zLnK6bcYUw6_US6c6-}t1{%tc&BoT03<r#p(C
z;=D?8h+YV36+IV02+mzO(;KJe>_e?Fa^r~8h$&7ZONw<NJlUCc+x4TPP9ypCOrKM>
zmm27{!MM|?)h6DP4X)rxk^AFzwjjbw#q#KAqEHPSNZ}9_m61oN`D@FC8&57hiQ-5r
z?vRU}&`sv&h`rEQt=s{)c4+54a4^+g(0TShSo4uIRU|@^wANC3epSqL?vl=~%$PBJ
zqZp!JtkrKw(&_iL`GCi}$ZiO>4tm-T)%WGQ5E?MjEBj%G11a(l<`5{bLTR#m-O(U2
z&vFXJ4|g+n2)0OR6@)}>`#laMr+eL+k+;~7<hz#_$2cfMp}G`{==#P9<eet;3S9Wn
z#L!zf+se5XqEKdP4$%M|;l~;xP4o%;a+<aGZd-3ZULW`53pxaFH!(juyz65?lIl+&
z&;CSBQksio7O1KdiE<7+mo)+_qh5jiWc8sNNaVoqa>VH?lvz1&C>hXCb(>IHNE2Xc
zE24=4*ThVaUQ67_PuFvXwcKP>KqM^OFinA!pQ&e(wil#Q$EksmSOU}206dj*cJAB;
z``I=#>q(BJ5nt4_;b%d^pX>TmbZHln-`kzO<Wiv#G*;h{5c~P|8+e4gki6aL%O$wt
z$}8*_x<Rg-6F3z8=D^F8LGc^qg$(Q$>zR3&izwPt-ulwID)Yb&`=z>1Gb3$}$Wf;+
z#~Yd9GV&{Qwo3rDUvA^&>oiXXvFFjJ@oq+8p0i)+mQdjXQLPFLmS+M;J^B!|Gj(Q>
zY2`2NhqeA{+r=X|fw8{O4rFM%NE9KYEc>;B$lxh)`v}UGXxlCaBD>4v?Ckw>%C=J$
z^;Y}!9Q5+y{pj;?PQt+HlXxC=_=$?Q-$;4#oR>RSxyROH8@Jy~W;p0?aV|UTluPKb
zz(iQ6{Z=<Cnbf41Qu+#eIMQOP{dW7h!6YWBzSFKP@igc!2iGq9oi;<^r3ej<FYc<r
zI`Enid@sx`l37wA``vCvy8PUrZU-Gm%=&Om-JQ@?#Z?joXTMkD*s1gzR*FnRufFgB
zNh6oi{rmMCZFtoZ(V0vUp2Pf8UEP6e)>cDPW+v<payIYkIaflCiG@v@w%H$c9{|&5
z29qVoU<f4lQwYP-#&+8uwZ4##q`N)i^lATc%Ill$kK1}Ej(}n4(=#X>NIfl_lAy+6
z`;*pJeheKM+4@4F6dF?8<WIY&N`e!DV();~9d;mz7`JVKUH@4<FGCbEDwDpL>?RSm
zeXi_qBB|}q>)C==gdcBJ$rK>*OWmh5%RFHg2us2tllB*lzt0mUq=PExNc=&5QgH^#
z{n%gT0BMma>a=vn@*+V=aoRB@-To?9V+L<m-)U$IoP87~BlF~6H<9joD(Fp<K<e})
z@=(lJj;VuUry)l{vvFMgP5Z_8EHy+G4o;8e=sHf%VDtQKTdnh6>s;4><!lrkI$(d-
z;J4r5T-TZ*Y=HIz@Y>&3U-*GQ*rQk(T3;H`Bg)eLq1^|oSMiAh_(579A_BIaNVB$o
z%=>O1cJ|R1EIGoA`9^uCLhOpa+GYd?+dsVob2GF(GSMAK{a!eYav5Sk+HGX<3F+Vd
zxnUI=jKpuPZ#>z7XvP*Ov4^<vFF8W+dqO!RMDGopzE}lOa#`I}idCYENFR!;{cFx5
zr{@iv>%`Hsi^SjlTWfL7-&XHh6mi*R|K4D8{mlcX6{JO&DVv8vZBjJ<kD6PFyea{E
z5S4+xygS?_$?r9ygZ7_oCD>&uB*AQW>(dkD!o^+!*5ztz4nr!(0!Ckar|?X|YMg7y
z-|!EB4Ig$k-<qT2FI-E6^N<5cj`{YtzujI|k0$CmtQ=_b-uiUhv+NsqD(u7aM^n9d
z4DmGT`;UpLLx2@vM~QCxh_gY*H_C#_jWo5~%kN))-}2)B?Vo*Q8)|%tP(W!$Aie`h
zuh7hFpGfb`^Y&3~DkMsQyhJrSefdFGUUG?jbUh5$ic&<Fb0xZn-2sUi6O$m?F|=CY
zZl1t-%04Cs%R2BcFfq>el4u>_8nMF8U9~e$l~BCTKDPRVsp&;L>Gb9Lun$}9tS}a8
zO%bs_u6_d}fo|;}ZFz^&7r#;Tm#{Y0l-aFteEaybLC<gH*iGkJ?vAPX1^a}vA*J?o
zE};tCIUTvnU|q@+50{lntaDw#4Xo8Co(=5hdrwgmkyfl}r`a*85aaEWYNX;RlKJ*7
z?Qr^Hq>@LQ;tY+6(3Ydns!y&FXy#<wRXGyDr9kAFMeKj)u9!>g)*;xOTH0%$(vSfo
zeM^grC>;k9cSBQyAxjnh)b4JXc6AhrN^l)|;SmQCb1=NxfulhU!j*{g_|w`rBbKd&
zS7fxSs)?mFnY(H&RWMRJQ-Kc8DHw#cJZhhwYw@nv-Q!%=t-I0wD>q5PA`^!OZ=aEC
zg%+Y?-RCBpj*J%S&rn$tl}!01pIJ}Iqhz#zNE_1cm_D77*5ER+DpoR%hQge~fZ0PQ
z?Xz+~8>c3m%fi=4++q!P+GlsO7Df(MgmsbK#et;6FuKJ)r<)rMU$2rISQI%|RdaLD
zxhtNpJ_V>iY3*|pvrxB3Ze8!Ra(muB`+$93Yg}>F<xWHN7mmAt7x(jX&~JO^-Ogn$
zb;OElX#4F8YEFiopxPptes$7;7Xqt_8v}Xa5tau%)4s5Nsf7!>UM!!oSb{h@@cIPe
zM1tP!v@hz`QJ<ahS7CPJH#&=87_M~eG`6ckK)~&b+Z3~QQv9$b^JG*p2ND>fRh5@P
z3B{?Oi%9RbFFE^S9uPDd@ehQma5@~Dn*R7cO|kS`rCUdp27CG0u=B3F%eks~6MK!k
zM?0tbGVV!z&pCa$B)3gW*q7#-_wGj_<~kk_fiDaW9%|=SUppg<5G1EB6)A;Yt8CC#
zdqqC;lB+K!gosAk^G8*pzHhH=Xs<HT6R>GIG=izYCdB1ebxTVG9&c6nSbDj&dk;8p
zjY>t*XE-`1w#cj72=1bE4Yf(JWY8s&c<+Vh*m*VPz(?4`33Lqll4~KQN<I^>OnKw{
zf)t&Kjh56NNTG&K15m-FqxNNO@@|<Oa~iS`%#QWOF&OYIuW9?4P~VBIl=6~xMrV=B
zI4vnuM6F=vXdbW&iWKW)qT^0W&s>;^_+F&v3zMrZFJk?g9XQQeB&s{`abTxi)bK4;
z0+iIW*z9v4ad(Ej2*;kWi`xey5a<wPSysaN;6Tph-F#|S?n$gN(gPF+l2#~T?RLAQ
zap&G<Rj}DLaC)=?Z|QHgOLLaU#qUoz4UrY<at-XVdbf6?;seQ7V=>q{4kRZ#c2nr9
zc6s#$3*hZ?T+*kf^|n$M^onj~Gb2gzy+o{x&^hG|F7ox#xXajoc4dxu)pXCfEc}qA
zeaO&dSLKI-y9B^4aCETKQ=2%lH4s*GaDnpSlgwY8U%$W!e|@JBHwzi1zekK@<yLY2
zmR(aX=Ih`YLP9YyF-RbZHwXdxE^2-Zf1SNHUqb~hs;1G%J-2RQY%TQkb@gL8Z<L40
zv6(z}Z)BHX(Qx1?Tq!D@8sBqvZBE8{=e^Up)EpCfyU|{s*VL$<bgs2;=lZxO><vQ;
zrbDbbUE5n<lU3m`_Qrg`y53yRxt6BCu0N;4=j_Xy5CT&_?X;3Et-5`!jWqrPf~z=A
z@T%lLAf%EsG)3Gc9&-B)qj#)XZ`ZYNg2qrvVcWpz(IdAakq*ept>>?sdcbKEy2%a^
zQlDMlA!izbpTZC<E#+|sQd@Y%w3MVc#$h+KJvvwzn^<s~n!ih~zUB&B-QIfDC6_sk
zv~-~KVQsvz2BAWniU-0p=Jcib#*489mwiJNeMoJ$n{v51NsPcns;SV{hmM31=jLwh
zsd&a0sKCxp0=qZ(e>c?1OuyH@qB!&B&C-)-#kVUpQwS&%uw*3_pNO#2wn<qj?Enn@
zeoMnfil1)<8`sp#%32{xs(@zs{Es-0z=$69SG<yI>&6hh;EXzY2*Uzue}m+-r)ZxM
zc()c7q3#h+L@PJKO;gkJNK?19SC=?^KpP=SO#l(Ez4jqeZRh4^H&CTzy#LS}IJ|pV
zma%Ieg7?0pR~zQqht^z=U~^;MC*-?(?a1gY_v%;rkh?d&v^nUSTA#aDtr?A5g&x;l
IG>Dh|7q`q4GXMYp

diff --git a/models/ggml-vocab-llama.bin b/models/ggml-vocab-llama.bin
deleted file mode 100644
index c1a3194c4012e36899b0bc0cebc47aaed2a261ac..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 466955
zcma&P37A|()i-|FcW5@*;SvH#AcW4oK;X`r1hO=lfv^O)J>5Msoy<%()6*Fe#P@;=
zD5$L>qM}~hQPE~c)(ev;F0Xdncdxkb<G#G8-|w8$r^5Yz@ALe>c`#PpsybU$ojP^u
zRQ1G`p`pD9e>d?z`X_nI;Q#l-|1PMHZM~4c?ZcUxB=_ayd79pz)7cxcSt+sH{8Vm1
zDz`9|Ta?NzPUV)Qa!XUWWvSfqRIVeH>rCakQn~I_t|yi2eR*zviof|O{^qCno1fxu
zeu}^ODgNfC_?w^NZ+?ou`6>S9r}&$n;%|P6zxgTt=BN0ZpW<&pioXRZ{uZS8Tae;!
zL5jZxDgG9u_*;<TZ$XN`1u6a(r1)Eq;%`BUzXd7&7Nq!Fkm7G)iob;^{uZYATbSZ+
zVT!+nDgG9w_*<CbZ()kRg(?0PrubW!;%{M!zlACO7N+=HnBs3yioZoE{uZV9Ta@B&
zQHsAsDgG9v_*<0XZ&8ZBMJfIkrTANv;%`xkzeOqj7Nz)Gl;Ur3ioeAv{uZbBTb$x=
zaf-jiDgG9x_*<OfZ*hvh#VP(4r}$f(;%{+^zr`v37N_`IoZ@duioYc({+6WpTaw~$
zNs7NEDgKtE_*;_VZ%K;3B`N-vr1)Es;%`Zcza=UDmZbPwlHzY^ioc~P{+6crTbkl;
zX^Ow4DgKtG_*<IdZ)u9Zr78ZFrubW$;%{k+zojYumZtbyn&NL+ioazk{+6ZqTbAN)
zS&F}9DgKtF_*<6ZZ&`}JWhwrarTANx;%`}szhx=@mZkVxmf~-DiofM4{+6fsTb|-?
zd5XW~DgKtH_*<UhZ+VKp<thG_r}$f*;%|A1zvU_ZmZ$hzp5m_~#a~B?zm61t9Vz}g
zQv7wK`0GgV*OB6{BgJ1wiocE&e;q0QI#T>~r1<Md@z;^!uQSD8XNteh6n~v5{yJ0q
zb*A|1O!3#5;;%EsUuTNH&J=&0DgHWB{B@@I>rC<2nc}Z2#a~y7zpfO2T`B&$Qv7wL
z`0GmX*OlV0E5%<|iodQDe_bj5x>EdgrTFVg@z<5&uRFzGcZ$F66o1_*{<>5Ab*K32
zPVv{B;;%c!Uw4YX?i7FBDgL@s{B@`J>rV04o#L-2#a~Z~zn&C-Jt_WrQvCI#`0GjW
z*OTI}C&gb+ioc!|e?2MwdQ$xLr1<Me@z;~$uQ$bCZ;HR(6o0)b{(4jV^``jiP4U;8
z;;%QwUvG-P-V}emDgJs>{Pm{z>rL_3yNACQ@46ZP&E7ipCEgh#IYAUYyEoFMar_f=
zsys&I$L;bw%H?YbUn5CXkcwm8adP`%qC_jy@a4Hf^|Rthqe8j&TpMhWDEk~E%Yy)4
zQWX%8({Df!Sj_@aRlb(dK9tmWiRd={hSrVecq=QU;QQSQ0k9%Tjo{ly4b~)4)6!L<
z$E8N44tBKF2+<2xK1kHE^eEA{q(%vJC-${M^b1Q55KSvI)cyN&yaO+@oalK;jtvp*
z`@9Gj{2VB$Joq_OQYGNcNX60^7&_+pwrsg0j(@%_G06F3sXYc7JZ*={Gp!sjW?LHV
z=1PqkYGe)E1PERvp9=uC%qp*eS3tgwquhV~conAWz?Ua@+z1muw%Wi}AgQ$ipa@y#
z6anz|8-m$HSIE~qWV6}6cA<|A1UD>NK_AhurO`DbmhQknu=Hl42}`5x4bQiSK%Ry3
zN-OEdfDyzM(0hx0-A42p`C0}!uhVjndz+Go_HU3>9pCP-Ht4=L$=4FvG_)K;?5&b2
zqs^UG^K7EK^efujEva$Lp?4XpJa+G~3=iq|N}cP$X(Xu%2I>0^+B%{S>cHTh{IGrA
zjxjBjYk>AKNo_;w6Ot+*^+`#Uk!nh+juG*owayVeZ0Sy-N2Nv{<UX!-A*?5?d?nFS
z@^u)no@tL!@Ec3A!X5X#BqvyZ?38AsGKxPV%_`vV1q0Si^m+MO#*q1<2q8Zq)8tFm
zfrG3CTGjyRuUL95(bufjcAm4}Fewgzh;Iqp5eVixLe4fO=X>%okGc5+Nll<)Vy(&$
z=!=%&srnPCGhQb8nWX9<{TGIP2GOsq@+4-@Z;X@zH~pPeTMac})wsw0D3Hpa<<C~7
zhv=`C?uD+98ilPy|FnjAKzvC)vvU2nq)O->n&P>a2akJAiB{p_y{Goo3A+(~KmE*z
z_m!kfhyA7w&>9oNtZet6+E252^z#AMdM31zrOzNbNKjSK(FaSaj$wa@m2ZOHlCQ%6
zcbI%F$VfQcYOEqULJ44zXCzswL7qoijrBxF>1Uo0M+>+rRPr&7I@CVaGCa&*G1Z>q
zRn}9-8B>#GmcsE00M$+~43OxFZ43Z)l2!o0CrfJ1`9!BkY6Q?uojO8g^8ZzKZIp!y
z`1h2fQ6M?pkOV|CjIJ)ObcVnzW5&<a>a5VuG(JY~{VZ#HBhf4?;Kg;eWC}VV=D3m@
zh|X3*(HrMTs*Vvo*T8HcI#<g9e4dtr#Ph8DVypl{N}09(e1lsA+Y7AD2BL*lz?x{0
zfGI;$E|#X1ad5Fj8RFzptIT?3nPtwvz*%nnGBM1`ZG`~Z3VP1BTEigo0z-T*(MrX@
z>h?ktWj|}1i!3)*L78vu$57Ov42@s}ciMMe=(_}60aLhJYK(K2^k`*n*DK}cokz4v
zQWcE2)j~^Yh!@*6QwQrfKwqjByLc-5e~*lpkCOj2&FkcTXg8o>*9wZ=H}AfMXq`}#
z=gF{MXejR>+90Vp^RRp>U6}2c8e^?zi8fk>)nlItW|B>aek<b1eVIV7)KTkllQE>R
zNs%&(SJ>xEh^~}$>*-dL=qmeuInmWZ#1I6w*?`wDDz1^wdC2Tqg~v6nlXjK*2u5ch
z6<E-Dt%?COAgLQbTR~~#`CgP{nY*kc<>z7`3>vQ1gG57?8GyPQmO3L~q^yW}%5M?G
zyF(^^t9%|{oj;;Ac&v|Vl1Isyf$PWjihQq+fTpT(jiZ(8Eqxi$xYg%Hpk@W=sR?b%
zDz<JzeK^mH?=~qJDHn()r2*vnI7}1ktOBd`?ZU|f=-eUIMX}r{ALq=4<zi^ppzcjl
zqs$h~t0Y-sPX5cV4$9yT71cA4Sr{sCOzfMD;;Sjtbbf$^eT%@Ufco$jL&cJMwZbpp
z`>UmCd5CrUYuXgx^J@&^ETY#MOL_EV_*w%b=9Fb>!xX;G+FU?%s}*#NQRp(eZ-(6^
zNb(bT3N=~g5%PMg4_Sq;mnyQ-zCn_e@gjwqEKgv7+^#VA>+MrV{jU?&rX=PoluLMr
z@i4%O=Zzw{N`XR6t;BNwCZnJV3f?3L>P*?2C0WH-3pJVN#k<j_Yb%BAB%AziQM$%z
z6uw1-(R!lw49rFafgTCpYIGD!7-?@)IsoKtZFtaMcS>@S5#A}~dEk4yq?oj~D-K>F
z?=q6Hp$PBN8tC13Xe)5{4#QUGQFXUf&jXChux<8EX;FtbHB|*|?-D#^Ea;&oWkkJO
z*?Jmmh3}RcgFFlGQRE=w9&N@m<UL9$1QTkq1i{dI1p?+VAiUQos$menPhcPwzE2zS
zq>JR^>^T%h(yqpGp?l@)c%8y~6&(xn{Zaw_y@JB`ONBg?*asw4*+$_94Dll{vL)Sm
zh>7^138~1SKV&40Zl};?Cc(~!T?s_;VFAmF{YUI`9^(6mkW(MWbpEJKTo|vA_eZ5<
z1+PW-DSJGb?latZp6eeIpo4=HYN`yWd|YajYGNjYA6ItSCi#Tn<$>}E;b%manft8(
z>L|Q_>Oh?ylfz;Ud{O{Z03d89nTJmq8C4z{pE3ZWSh|}g)hj7%3Vk!6xE`=7EWihp
zY`}R?$%CaCJ}4gx(2frYgbEMA@F8o*%Kc%(I*#rOAC|TyR`riai!CMAPmfroJe!M;
zTKhf}J*wDv|MQrA$CfjEOupyQnjW{-MV=y$3r$1l<xdNS`~>#*;inZ0cm5MoGyhj*
z@_#S-FH<jhSZ7Zf{u1bVQh|@~?&K*!Id2|?Ps!J+m{w2A*D~h)(?%dG7<$IW40Im=
zJtIvpRj{r+D{1JHCj2)-agAHV`uR!j>sX-jKzgpt(j*)xnjFH2d|t}0yAhjY`B-ID
zI3>vn7V&UOu-C`hCh1P&saU3Pr&P<2gQZ>ay{rRrm%!wK@)@P%k=D}`enx5!K^XK|
zg95ud{H$@ijW;7NNDIjJQVL&?7UR5Te$HC79;WbfQl~V=o&0%|&SYgAqH|d_XuhB_
zty1^}X#>p#W%@;dyS)nbzG&#qBKmJ5gss-l<?5_eza&jq8H8V&I!MPu$CxUfFFO*-
zkaPHDsaoEFRxN?P`xfq>mV6%Jo&9cOrV9(?Zo?{)`HD8imO9jAg^iD|>en$a_*H3J
z)eXYe1b*vDhOg-fQ8Qnc`gK;aUl$3KFy+1>sXD%WLz-ci4Ox6ss>)>3<nRPK=v(r&
z!j-?(#<BYjqHi1FW9YYU3%mmQ>^t@uo3ijb#t#o4`mR(0$nd)YvcOyG@5#p^H1+qa
zWgYYQ`z9@@kMB#B*;s;qU{xvu6#hUyv*P`sQi!n^{?IT&A!2VY1tl)fWSKo{Ka#IC
zHY9!|FaV8Sw1(H$A@3KZN|{I7kFB7B*%1END5wI^PpoLHgo*SM70wep8GkBijNEf5
z{HfsJiSaX~6B;c1nN*k<<|+Df1G9~l>(8A+RvY>kR?5>k{DrnF@dEP8b{e*H_)BZW
zAR&@pDFbYEX>y!53cr>X7=w>c`0G7zA*tU;2?+Kn{Edy_*0a#w^jl+@SNh)yGbQXn
ze`iB~yd(oH{GFi8mm!$n8|?fTME`pOum*^xrw-8>GiZ7woEDfj!iDz-NzIx|;U5Ip
z)O{H6e^j_^9Q{$KY&D@m=ubxF<S;hN;h&^w89n)DLpOvG_-AR67xnNLqiUE};=fq6
zOJM~3)quhNhJyI3;&~EsqQ6<a)?*a@O$tWyya@e$4>+iZzq@Kr5&gr8s=S8(LmHLF
zxUc_dh_H~N+x{uMl}2D<{g1UN!b=zakJTF~^Je)asW%Q+Pxz8RE06Q|{@0#1Ai?EE
zdEEY6o4~6T{@Z|IaG>NrHdYE^4}|}bn&L&Fo%?C;R4e4u&bDb#ua#MS(9XS4SQ?<1
zb{?SB)XlTk&V7)rOpcMimzC7SgwT=#mhX5k13eQRz4y+2(W3Rp?vQ*<kH})aw^f2=
z7HdJd#v77-cJ6N&N62@%7sIzvxR1~Utr72|gs>I4uQkXQi{$Ss1?3U+dpn8!cC1NR
zYW5R?V1rDMzn@edVVgrgZ)5*uyx&f2b@*7}|K&ek{+UJ8y^8$(g+lD%!u<vII5dB(
z$r9!~9U#Pt>$67w0a81Use6FbW{ngdAd;yM$bRxb`JTrb;2&rbt6|)_%%)+A4-}Tp
z#UI>v0<wnsnqqx*kOCQkAP$nM&<jvq2e)CPGY>Wyz8Ky~p%#=dY~ywvUULtT>fA^E
zA=dX}B{rK5G0LVQinUsqhu@)XQ4gVsrm;RlAsuSiilsq{4;6~baAqAQ7+B-^hZ%-R
z_FRUC8HSE=iVu?pldy;nm((c6iKZ?%kHW*P>PE=YWq4O`guyJ2VM_T&NaF#h4o&5E
zP<(_@b_RR2G6t&k0Qs3Vaj+`g_l^;Y_4^1P4;?A|wjRd9>>nxBCs<)=vIY|<Zr6B{
zwe3+x4m8my`9}#ktj3R0Fb@JutZ9rbY!;5T21VX7`$rpf*hqv&OY3W9Q+%|sz{?&T
zBV<%y%=pJx{mVd;e#da9_!wylE1Hg#%JX2N`o~)3*1Z(!_w5i`?7lz$arkAWei@r!
z{}lqHd;^90k#`UAD-6i)SHo><)KBJl>K!KySlJw>Bx22pkFy4Bm>+K$m<;~$A`{4A
z9)-tC1Kvc$#~aYf7@OE9Sd$v$bAnVW;d^+3w8p^zX21z1A33TxQOKAxhkQ-Vo<-q_
zR%5gR6ekLp{5IINCrNSwZh!xzoo#clHY|&Gc#@zliR&;v$;eV`>txI3N3f^$Pj*xm
zC_Gt6gVqIUCmZ4``H<rj!8Y{}u$^Kpwxa9w`)s(e<5Ns3uY@DpkYHPk-ab_rgYFAY
z-PvA;FRi!tj`67iy38i-Y1RxY#&+^gGfdrBnog5y6&CbqMp}&z)I7@#BIkPsVzk1G
zpetmPC-#iQ>)<;#;8XXJf4UUpC!kM4O%L)xQhd6>E&v$KkXj=Z9%D14`NX&^Q8Nr_
zKilo`3}Kh8?=ytxJe=IVChKgTYjTJYoFP^D<Y=ZXdlfOy{Fx4B9^Dbn6nu3ty5pHD
zGTup^X%r4ax%+1tgBbf~8iRcp%&uC2UD{_!5e(rf2LD-xXiFWFDm=>oR&iz)pQVBs
z8$zpDk}eOzwdl_h3dTzi@hoBCrkg09WsoMy><O5y%wvD-&z7dGr+42@;cTll$d1>z
zT?;!WnxhB`*o66WOdNw^bA)q@tTHc(bBwpvTX8I5nbjC-XA3p##naDqodjnKEfvf;
zI!7Rl=<M*%QAY9@{pT1QoI%Ft7?s5etLwP}b~}duTp@D2#KzoQX<P@r@my=X7M!1}
zwXvw4YoD>{3(qwk#&C`mpDT^G@z!9TisxZ0gubRLd~g`fvj#<Y!{T`Y3!VsQ0y@tC
zuNfr&JgZmZ4dr=?9VpH-CNQM1(aKsDGQ^0??mYGX*Ro)PqHO}PWfEr%Q+WbXnJ+Aq
zSfS6?zALe<GvD-4RSk!Dz7gAcA5MD=L=g(#Um!(L=PaNFLOBld;{`^l2!a+$!MPWb
zztCD^^ia4^u#V+1++t0a$9N0B$Pl8EubKR4m2LS&)_xLJZM;YrC>Nkl7i%9ph}F>7
zH0(!Ayv14t`^Q+5bv6u_*zm+QfHm9_ZT%GcJC;a`$~Z>R5}^b8S@v8mHKMjb_)Db$
zEE7z*rP=_#$#|*3g?R}3i<U`I2}_f|%$l#o;9e#Ix%NhiwZ=G%7h0~QV)gfzOL@;m
z3YVKCYZEYDVy!b+hF5rn)Pd;y6(*EzSWdzfMgxWk>!=k%>Uf3Y3(l7`98}}vpKtXr
zx^T1{o^MbF*t&4J5+C<mAh>SYOunXOTu9*sR=<XW#P|YZj}OCET2YbLx0TWu4))M}
zpTNd8cDWoM#a?J_CJ^Z0U+CE8!4Y04AWAUS_f%t}?INotW5~Zqdx8hCCP$c+iwts&
z&lY?E1-sYxjpRJsxV|(Uz+i}d5fc2;sNJE6v6^+5xC**RcNo+H_Q<iz<=HpcDZujE
zp+Pjs>rbf3`Z%h0DiQExbnO8vmSmSS#p+67msGlGCdFMs(dA_r;M8q+*%Rq^8xL%I
zcAGThd?0o;v8%%Qr6I}>uusD85tt)<wjcHg(or5QJqBZ>jOB!Sr9@PEuc?LdsJI)$
zb{UrKxYx)S!&F=)bR*?!Y6U#!tBmT&;R?m8jF!^4co<d-EwIS0AYaq@2}m_uE!azN
zRm7`}u7MJ72i8c*)O{GoYYg-_j*&wxki~C}5yc*+i-oAO&nEw3``mgLg%=Cv`Usnt
zn$C;2?h*$SArt;3hJ5Q7mg&$H*06xaE;r1>YpnnrW4pp%Yt34Zv42gAF!`W}H4T#v
zN@ATfK>!Pe%R0k5%%0_S!eyP!SywR5!*acY%x4Y$dTX@~e1z)--tG{Va;)iXW!`jc
z5XNyX#sg=AfM)MPxWRe@W-t!{EgEO9NKT|S3@PUn=&iix%9#W^z)8+<moSp(QYk8p
z@*%3P8FVXif2q~Wvo5|=&_mnfP+F2T))X6Uh`|<QKWDhn8udYI#~X!)D$`GWRx%Dj
z`F%!FAErT{31_5I*-CMrw8{_gO5MM+eKHTfg-me2-{5S6`y<pGb~_dHU)(R;7ocY@
z6B?^b<7N60N?bqYi+PIm<BSzJFBfcZv=_<0Tw3Kv;dZ=STHQ34;>(3p?9O?5Z8Ge6
zo|2o4eK=G3Tsd^bqrBQ}5*QoNlUE25UYE$fLcY$y9J|7RbpUyMg&-g2z2ucvu$qm`
zE2RNkHKBgreY2{IE0tPy-d-ii3Y;UpCJVeZzslq~!H(~EPj;9+M^{@Zc*j7#TCmgy
zDZE;U-TgYOt?|{ie-Qp9W^1z)kLRn{*l!lRI5nnlv-N1}9*Q>`1L9h~Mr$z^O|m5t
zUSkcoKd!L`62Nt>&{6?{Ypv2sEMV6PoN>gb#G0&g9NKl(v=<u4zs?$9Q3|gUQU}Lj
zZN{2zJ%yo7flZ*iY&-lw@ZE$Hk-&O%RPDZ4i)PG0Y=dBjSwnt1$@}%ZVaF_`IB(5i
zK+1vWfI!}`o_tN^u~8Th60uCOWvv+;4f2F92#PKqwVGtBFDwWb>&58O&*J$j8s-6D
z_KVV#cN<~RM9_l`eo+X>!@*KAria8w>6fHr&3Xz;)&LeqToP>gA-2K?4F&dK<PS=N
zL5xF94M6p{8a3YM3<(iz9r#1mv<{UK4jG2l6BKK$67~!<EDc5~(7nTgx5DXg*c#Vx
zWEyLwJnTCv8ytAM{Ibxq71K^rvo}**7J#E{0&TJJJyOQ8gzvI&zvCcli=abn&YLOT
zB8|!gaJ*GMW8?H@@->YC2}5YBRADD$yw&=+G9X8jG@^nml0RZ}ck&^^h!If8!(xwJ
zb~Mkv-chS7YxStrt@7DzIBL}0L9teZZ7j2IOb8i+ndFZN-gz@995b=B9->%lOnneu
zl!|imFxCLSBDMJxry{lChGdmlQN}n%xGJ?(TAFS>$o6&Bp!eKBan;C{Gv@1!Y;mvn
z*DJm!A-(W=YXwadUoTkdqv)1#`QF}ij|()8km40bi$EA#=33OJJASGe+7axM{hIIs
zM|4=T&dBS@T)d|uL_`xt1oWRjA=F{kLBOtH5U8}EET;!`6;qk~x>PUA0bpI|uS{b3
ztQ*TP3xHvpWp!V&O&X13HKfoLp@(5|wTtS!7)}~HRraAwY9Fu<f6~AWaCk{PDb&CM
zqZ=fJ9`J8a6tY>;uYLU#x4)Ko`PeQs;3x#snuZJh77Di;)*?@IEs;?`J4{YJAZLfG
z#(Tsaf_TnMiuHR7F-NRJZ<Javj#-WQHwwy;JhWJNqW~M1Sfcnw;SH4#<s^0Fw8?*^
z@mb2V59pPGq<}T^mC|@XoIZ4uF#^+6)38m>r0^yqmleCqtVGcAs|4NDQ&<UJWo*Fs
zrBDm<@Quf>5~%rc-dFyYrbnUe{Qt5Btbju;*?lW`ja^A)5Zk`cXdmPtEiV)hSVSR)
z&_J-+8EA=~Qfnje{bR!%g~BkalTdgml-bvpXod~22BNu?Ha09gO|NAW-@n<Au$>*;
zEF9P2X$dvS+9hg#WrxqrruVBA_T?lu3+j%3goWw2gRb&!u}av?@j88rNf^ddaEmdl
zUKp2!;a#NY79mHB5*Q3^m{76FErMum57DcIlM?S@y;loT9Kn#^{>)2$py?W)k%gLu
z)&}%;x`?B=_|*ciegh?%+5(0B8l$boegf|`M)w*p@fsZr_mwEn<X|1s(zPhEG$Yqy
z0B2EJ#vucIn4i2x=*Doslzpur-%=v)wel4V!+Z2vLl66O8wFZv0QRF+n#?0MEqbjG
z1-A|0#O=&D+s}#C!U+#Jd7U6Ykh%9dp_{{B{nr^ko4|zD*oyU3Ym8&-9=%RS2@Yo|
z*3WP>Q}Q|+Bw|M2s*-pV^1oHk3}S-ZYPNOh26fg1w;DIlg^<*(3VIYyoyZkQ<Yj!T
zfqyZ)hmu=O3&RlSRfTSI!)dsT<16nrLD~<4?lx=44yfQZp^M`v!`lpZFGqhww;4$I
z?s#t!-)54iVQAeZl)>u>xxHRsFgWk^f(8*=zJ46xk@b41J5tFP5m6`=peUl(8#>s+
zMT)gxN1l?`n^a&{>DKZMMlv6Nd2bL9>{5Dz0G5E~K)<f(MtGu%`U4e;-e6LID~Qk2
z93Tt}O5WfoD2>1cbGyLB+T`7C1%+YM^KUoCVQ>ex8!c?&gtsf}+&{MqG2<Mk7vC;;
z`ACreDA8&q3=6trPkYSzJET#mh`4oM)2%142@UQrlzq_6cW58sh(EeRXuJl853WY*
zEtF`}G9O^l8>I?NdQVgEllgBH2t|Yw1aCBvV7_oP$Qun)>pc{`QOFq@NAOnc3N}Dg
zZ#1fhc>Db(X>)zO0$#m03G5Nx^lP$$6U5+6)_^s1_$Cu5ybPG}uCU1dsW&N4b1Bi(
zHu!knYz?5{FlF8>bU`b_AF62_&u*o_RZ|=B&4LvB2A%@Zn~l@_I3MW8Z#GVc>Xc|z
zKrC{|c0))A=8@MBOuKJE<c;4D0Kg=R1_ccPFSZhlGHp}ZN>M{3P==z8wF-lU>oHMa
zt!HrhPHz#w!xK0z^)wA@66V2Mb{?ueY2C1}h6>(dlw&smHp90V<xm=t%k*GQzQyD;
z!Anu1fMGn}g;-ZXh*(1mv$qNWOaea0d8;tnvz~&t3gP4M{)cZhlqC^mq?I~=;;p6!
z21??5Nfa=297e+1tedeCljm|pK4N;CfrPgdb%Q;nFbu=D3GT~zDSn$UQ7CDERs1%a
z|IiEKgH7~{m?S#r=uVReF9&xDr*gRB-zi}8n8m@JCL0_HBf9BM9UQe9c2*SKX~gA+
z@*q9FQ-qSAq(oC)FiqYra3|S&=Dpnzw(cbV?Scf-pg`*s%HSi^bae!SEz&9+mJ(~S
zj2W4{-NcoHUZJ}LC#JP`mtev!84J!`Ms}@^gQP%<uzx8cvRH~Zf`fM=k*ir`iy^*C
zS%+gf(Ikvu#2_2W3HYSEcL<_xIl<O6hBvg@JFJsi&uT#MJB*4E_UlBhyjFspRiyYG
zA`ct{;ea<$NH7<m-|jZS4oaY@cehl8KOgpkf46YF9c<k##LAh{-9{nrJR_}AFH@{l
zIOHo)AZ(RIh>;c`2s-lKsT?C(&(~z7xSfJ`8a!U$!*>eD@H#v~kya{UcN}YS8>E_O
zl3nujE+bxTXYXCocpH0>{dd_!RZknMnBZMP$}9v$ze`A&fn(w5T_(C(o~?X^(RvgF
zB$|dvi};Us3v6hbSCjW{X<dOM$!~wY8Px;*oM+4O-GXW6Jc=|mgq`ENZA`)skhDQ?
zm)|4Mw!`%FG|B#1{~kvcADjmFNaK+b?~FsOhEYt>JvI`G>h_GatO)-@a*xWg$@=U)
zLeMBy%lDYPhu|0TU9NNkx>;*okIC^KZTm3TjoRr4!DIX$W1<u0-g|^{+yO##y41&N
zic$Ms!L_l6d`-av6ueie!i9t2bpx>xOTN}v*Fo`njjk>TCDBSSI02B}XR?^&s~z6^
z3;-fne9hz$EDlZF&cL>Vlr)3AF*+oApU|`e>xu&=;Wf$o3@sd3M3InLho<*5xt-@h
zB+T`*Q!t23-IdxWv|(h_!4rVxMcYX<gsY1KC(%oZLSo&CV0>ZY`a0j^@a{FRnD>Y&
z1=4$Ma3ca8V<XV&m{c1e!+V8JXljZyIdwmrg7?~(=O{B>Rue7W#slsBM(8Ln>)!iC
zBsDmy-mhH4krKS$#8H6C3*T>&u8mUUnv2SbwNV}plH~m&pOMlaxcY!VMF5ER0Yf)2
z4nxP+g4r`E_<*1*VK9cS&M5CjA}yJ^7wYi?M#y^BC?Bw44l_t1zCWli;I#BUXhW|^
z{s*lUPk^9Z!f|$?Ht1)G>1RmtB8s(v7zxi%@<Abf5VHD^5swq=0(l=2LEr!Y!_ohc
z;O6t$U{5vKh`B75WoY~l31z$Q!lk3w6=5F))$$=Bl&?O)_d_2xIp;8aTn776|HExr
z4N&l58xRw6Nh$oW@mj)F#7J9>$boJAVc}z<NXdtV1c|Doj~IX|HXRrMA29$t-u;gV
zrt%=T`iPN(1MN`1!v#u_YcL8`9lKl+5=tC!+_=CbrH`5b;s1DmypI~lehAQIcE5&#
zkGAn*#<r6@*gh&8;8I(xAIEvEZ<k{sr~8cdQLG8xeZpmBd;~+_J{u|=v&s|aJ^?j}
z!xq<wcNfupf*1!O6yImmwcbXF%dA6?)W-}DZxX$a2_8LP+EaovfZ$`sZjs|X!;c9-
z#KU5|L?07Gtkz-$&@a}Xrp0eX9~T;D&L!{T2538nK>8mSj93q!p+GBD;f(#b(Yp%Z
zwF+Df*advt$Q_5No@j#sxL`gZBuqjG-Y2ZeBur`F<v1K5_=LdKWA)G#btCdN`h*Rh
z0i2g(HT#56JYyau+Gs@2pYAvE3v3#D_Y3)3I3UcwUuYNrSAl*;Xd60RKj-1~j_wzj
z*Ih^P{dSCk2r7*!O71tFTkplaeam33dY^PX4k54)?~{V5jQg#DCb5j*-j1|^LPw8B
z`kmui<4+2WdGRlNQpm?igx;L^ln~U{L*A!MYGY{ae@YPKF((37fH+B90a943CL*nL
z9Rj#xO>Tr9N<L)*8f8P4n%ck5!o={JJI~g=1ys}j%YPjEa-g<=Uc)L9G!1xF?oKs@
z{EZtZ(o}%cIYo%G?_o-sB9XRN{sGApB&fuDzyMWXT9N;NNm$9Upc{@s<`w+`VWPrU
z%%TTOP{UYH;|GL-I`o5sh;21?A`glLpMMg2xCadvc3!uTuO+P0f(M19TA6Rqg%1i6
zaUw_Uf(da&XoIQy;2vtX#dd-o63na6bq@&?_@&X=zGlD%8*C35TVh6Ilc<g6&86re
zW21<>!|_8xEyvX*T4nAGT*I;|ytsM~3l;#zd5{0Fu>{4Nr@)o9-jDlQ)~pLNA$nLK
zk7GX)yIOsiy84}OKF}i$7;ja)M^xBiB)TFw4+tI+(AWU8nIAr4CI!1%d9X!~2#q-8
zMGwS}81wBTgX9q*W$Ipx%0~_50L~rAdsIrcqO1K!MXVhgDR@+najZH-89r+4ufYyc
ziy(_KOp4e6!^(|5dQ|8~A&v+iGd!C(aM^oIfDd8@KPJF&a>?^Kcua)^pGEkXF~#p0
zu+0}eW|()tAE`CBA;>X#Oz;fKZKB5wkDMrbk4uyM02|Z3D}&#;NWtSK7C4hQsxW-q
z6d(JgVM%IZ(2v6SabxCmN*=e4mH3TMOG$YGL)ZJXfnc)$gHX#9pQbU8anST>;lHnw
zqEELuhDvvh2XRZMNQpMsCT_zgq*@>5o%e)c=b7)j9L!4!o-p>sb_!jg_;VG81ed#d
zqj?+@Ct8nf6!b7XDa2K8An!>-#pb-P-*HbWcv7gzL-^s7!o92kk)|cuCf3URuo<7U
zZbOEi64+Z{%snLyVY5R_PdPliW;`VgH0&&NMK6Z$rAV94fSb*IhbNK}ZGx`q#58(Z
zsH<>)JZ<tA<+B3+X@SgEH?|N@??DQ;hpQ=@%1B{wWMAw`V9&vFd|K$n#aNv5JY(Ru
zaY&lyvO|?hk^E=0<^#N<KO=m1_2bH-wZTy}{1MLxu>+8A>`HKq1VnqXe5H||wc%RR
z5Jm4<X<3F@?mugD1xC79BEhqU1@RZ)?pYy4e0I^ZB82h?-xH6Y6}s|p96xJAcT}&s
z$5IV}xY+c0u@z$GU<~@P$?kIGf>?m`^--v)o90mDYLo^ML>a3n*x!@HLU)OSGoCXl
zYh_%&^q$jNPhl(SKPNEP^yAqp!PWMkhgt`_Ls%*=YtDwsa{{=4lZ@m!6BeTLWd1yF
z1mgh;ycK15tL$^~pBKElU&|fkDt2)Yjmz=Oi=G#n&N&Bf{TP5zj#N&bH*_*+u_jF!
zx~+&F_of8CEN=dkz_?}}o`%r=<16}8)_>Uah?6%`0DM{s#PO7g9||6(NaDcPuujlU
zE9Er^owZZY;q(x!?UY_bXfF;#f}KVt4wE?YD0D3c^Kz82)7ZkjD6HwOem^uMG-9&T
z*s5}9<St?5CM@ilDqT<hE~ABmmEmsIlB;J?xXWN;qYtxxmw_6GOETVNRv>nB8Yi9X
z5-xBiin0G0O+Sf&;b|JH1+10N2nF-zP@rG2FUQL986Dsij=zmG!}};KP@gfR6L>Ko
z(W3eU-&~;2S`kN6c%K!75**}z)(A()1(qn6g$WAxiEB5BdtuRMts@KV<KOtRLJTxh
z4#V=ZIzsu>`UUyUkuKP6ykN*L=E$BGq*jGpr7xH`OE<6z4_{D0JPkoaFBn~ee5f0{
zER^_Tl)NCK;WuqQCorJgpOM@A-scP%I+^^>8I&=;5D<LMhDdn~Zq!j+FE;vH@22Q;
z#vZ#`*wq;;aIhzfC?%g00x_$ghd*zi;Wxst{JeGIUF3gWSggY|4L)xOR$&QtO@=vg
zD*C*!UmoMD*s+4B3}Nyo3Syj})B1uy8>nE#(j?+|$p3=z$$Ns}3kC}3=iF~!5c#aZ
zNc@5^g~=y2ajY#dPI&|+Ul6)+>WTP)FDefYu@V18!8&yx)|D?Bsx@%I1};~Drz8BL
zAj0YsebHoFz*YwVAMqEBCY%kj!}E(q3159;i<kb}L^!}|-1~2l73L<y>;JdX!tt1a
ztA`CgI_<v=Fq~6}6^yhUdI{$bu`7gG1+ys8LflqoFY%WI=+sBqQ}rc7E{314MZ522
zd%B&z4RQ!I%~5a^eaSkr1NO|9Oe(xlPQE1K!VD$)vQZ8xepy=JxPd)Kno)mNptV5s
zTwEhJxcyN3UpBgG6L^p#c4fHAkbK!>jN>T`Qe3eVTyRu?p_UQEgUQztj;W!bC7m<`
zTUzT#D97UzwS?&LJeIJR6<`r-nS`;oh6P2tjiyb!Px5vfrXssucMA>J1Y<_K67Iur
zw~@<!8{B4zb{nJEAjK}z4KvDR)?-e8MIeGl42rJ^jb*;s^A(|jZ|CwD5440Wyznaq
zcg-4#v^8AJP#a$nN_o}6A!?!}c!dNkebsp2%bMO-1q)vn@xN-y8i$6uWedJ);=n~&
z4s`jdq3f(rq=4jX1UlFiVeTbg6}fNNh?neyqusB=2=%^Zn6Z^3|7(JI6t64=U)wW5
z5M9G7m=<p1v8T!EFitMNW@5#@f%`i7n$eTP@zK`>X~$afzOEg^k)^(V-MpECuNyPv
zQ8o`ktySVP?nu)(G(bSd*R9X7Di2cfb)l0(YUmq+4pRaFa^5!t3Dyp1+iz%JL5hKX
zE{zpoL~DHv^$x5dTBombJ;hoFZpDWw(K>MU@~Hl%NJ!4*uzY+|p-$Av*W?HSo`Y`+
zo9wI5)YdXZ-!uX87Dv20-`vB*C`Osf!q0%2`YnO64x;*&;9zsc|CVs4cc+7I33nB_
z(V_+A3Em1t-!ds;I6`u<R)jAVcEq=Yd>*F=2%&EqQRDChh>!f+f-KJ_`?qZ@@K!JQ
zw)NuxRQR{8mw6kEE{wh{w5;KPhgd<>u_;e9xedNc`i=m`g)8#DWAs5o;uSjoI|3Tv
z-n__by*f;R@H--cK5+D%m($7mIMz1e9HHbpCP{t4<hxdzuho_iV&Q#PAg#pBlkXZE
ztq)M}U7=|h<0kyBFaVnX*z_}uA&NCwf~A;f4eUf<6@AY{GdYZ5_C0Ak&SrxDJ(CPP
z=<MwZTvfis8h+16!)hV>?#Pu-Ahag_o={Z9yL5>r%k1y}zI-hsIMvgnguZ^?h^S!8
z8hl@ffcC@_C$40Sm!R(p^Rup@SW^?&a<n1xam){-VS$4*ydQ`lF`^_6*#Cib6mEMU
z947dIK!Q0M{y@aYp;y>QX)Crw_~L!6S)7W>k!<n<BQwY6BR{mRnnYL#c|SA^dOgbj
zp>@y9nH0Fp^H1V+4XZg&=ezXL58E=q!Q~H4D?%I8BsMb9QbbbYAr7h7frDc&5&Gw1
z7$iaz0x>vzGDuVeY%7JD1~WLe)LKJ0EP_Q4Cq@B|v!Q7dh1Q8}z>oIyYDL0ty&nnu
z4eIvue`HL<V~kDPkAz9~enOI(5lSffkr2281K~%8jL+~fZ?&X=SKnTg*5z%yiS%AH
z$mKDJ(|1`IRM10O6}%wAFnm#%#1lWzrEM6}%duuExTXiDv=$D-(N{lK5s%|+=Ev4A
zQy(S&$JV1VVSg-eG58Szrq$GBj9i=6-4tsZF@--(i9)VH8PiXsd6l2F@P1<4;C2+u
zte+U3i3$S#eqtna6e)C>RXCRUiIbH?V8pH+-ua-!0p$ZIJdYwwv6sNp<kUy8I{eg7
zp;su-f&q-z@Tbxv<CsQ275vzvu+LR9x+PD3YGjX=_zuI*gd=R99wATDWwDBXW~xCv
zp)losCJ-mf97h=bOc0%a7)9+ie0M4SnX$$f_s~5qkDw!X^XTWoO=%2<k0u9T<@rAs
zSUnhO`f(k&4fSJX2Sq<OW+!<c@pGeokYg<pmz4tr`h^Yc%f|GS*ZYOFF2FAEe<3`e
z_d#v&3&DHSY`hgCsjcXob_H2H<6nqq@-m{e0E3gSmj6-;u%*D7`b#4STSXWlzZ4e7
zag#sL<Oqg#=qll)1S(tG!`BEi=a(jD+@j-v<U|Y06WH$3uLKe3g*)$8f~dShW10M4
znXK`e7vdoTtvbMQ(&4X+X~fA>^eg4_T8b3}nCRmBD#@=z8VGR#TKcu1Vr!VZUkekn
z@v5f(YiSI-Hu$vx<%^r4exI|MqF);W&}RsRrdW&c>}8@bxhL>$j>!qtFq(g3tq^6v
z2AcmH>pBkE<Gu}kBNQQW5$4-(bmZ{vNI%0WhVXtPG}ZK4c=8+J2kiA=NAz21k;k#R
zr^#6e?eKrwW|>brgWnoSH4aq`UzTMH?ze&-!widrHiHcYTPJb#I{5Y={Z7Xee6e8b
zcY<~5A%vUgcTm$;r$E2+HMmfdd`KDn&P2KqPKx+<+5_y5OnztL;V@EcUg-A%7<!Yu
z-%C9dAj1Cl1_4^0g5MiMa_}DhUiz*$NRhVAu^sPPbYP?;E?40@oYN-24oJ;q3Q%s-
z0tycR1=BX}5G(>|gwsZ8pSl~PX(f*z?TlSzM3?avd0NPo2hT9%{$S*`9>6Vo?+-Qt
zV2p@Q!T*C`?LeU0AB?hf=%_!4Eb18bkpltG6BOMa1b*uwY($%sxWd-^;IX4WI&^#|
z!~3Jq35O3X)jvAam<;TY545z%R}I2H3chjPcKy-h-NTMVE#Qm2i7Vi1mGmbe9JZ(T
zC+)Xlk^DbdTfSHlXh8*bXQ;K;uBE8`k<Uo}WYn%<3qSdji5<}(xPeK377SYuL-l9D
zFgA5R`T7wnVDM+dH3)J-ms=03E&8+d7(BadfyW8~vD}ns5-Nu1FM`oMKIQ$z2$EO>
z|1Tmw=tfv1nijWr=&H#g8STmPz9w$B5`V&9goWH1JSt_(;dBVwd+)D8Jz^bsTjl>%
zFv$ZA!CzIJ_#A37kC>6@uQocm;D*=gcwGzZwCjxErG~!=wle0=-vr+7+pyyKe-n6o
z3KeLDOV{G<Dia=Jt0>a%8`n~-l?$*^lfN0p8Z3{$TYbJ8;kn!x9OAyqO|WD2?}8de
zAN0WA1q4hf1ocLL7docy!-nMVCg)*(r!o1v&<&Rr{X<d(XqkUl4<d?4Ts;4<HWH&9
z{6o;-4G*CChd|x|4x@h<(X7&3W)Rl9YY)F7Y=VDEn>C2R^8RVObz&g-|CEm62M**E
zE%>M5MT91f{QfC;5k&~GDU1R%WbA6K<|FAuOYj&Pp7Jt^YCFi&46cqM(9-`OA)!)5
z?1?7puxi5p5fR{Hq$w;7@&D{ROi#S!DlaS#EH8<+$?t**^^!p1-LUtPG(Y1E@?WwM
zS3oFLk%E_uk<ty={)I2?JX~R-3taec^pc4k>oF^a_$4DB5j)9C!hEkh&HOI|2mYZ-
z|E1l|J1qZSX8!XfI+;qrzl5yr9z0AYDUMG2m*HaXF;rBnFmn9pWunyvaQpAyu6vlj
ze;Znk*C7Aj27}EpXs^J1z#7c&DTJ<hA4frHb8J}QvBg|<(5fSRbmr;~^D~|QY4;Ci
z>wk;}oN<u<A17IPBP{rju(S@${l}Cn$Hg)q(SMA<YDpXeu>&<wf|JQL$dBmW4Mz!k
zqhAm(l%ZV*Xb<7J1<*jd_CdCY4Skk&?Te%==|1h+59#~}Z&7o5nqVVNfx^0G4pnJa
z`@N8Le3avIeA3aNUF~=Z9axde^FFIeO_#5YvgsA?X^o)lmgd-$M>u*yyV^IgakQG}
zbH-`fb&ztw@5VtD(_TWngbil)65@IJ2ZMVVjqHreYPB-2HNK|L#vkQetwA<G_Y$Fu
zz!g(<1-KkZq2}Zm_tVtS98|BwB=5EBP$daZ9<+DX%{Dr5&+N6U{Tvb>63AO!t-Xwk
z{D``39JY(`HrurF!4YP-*0Q%SHT97x-0sL|2Br(|=Jz%z(2o4zWp;1t;$ilR_<M_J
zWapUEhU|=lVY;_<k^0f9?TQ>49PVvmW8{s!MIcpdMWelCBw<)IU6oN>KZ+GHthU{6
zq?STr9ZQL}En_N9yB0(Iy#78)7dw_R?Q{XAiMNk%3P_XaFfG9Q1aHQhd`9mWhw4*K
z%kV4$R?>ZpUPK~LbswR(%xYXKVJp9t8k$D@2Bwf=*;eKkPn-Lg#Mm~&BY#R+5z=ez
zBmDNjd{q?uuu<#NG_CD5evtOH_O188e#-1?#9%2WPjd(>;oA_|eFZk+A>d2gSI1dp
zJLO!H8Z6piU!8t(d!wou$fFmEd|#7+u73?zSev9sGgy9B^-$B5Y{lBBSX&Rkerst#
zm%PlFC<!Hmj7=*E952p~U+yP#a^N3jG@YM-sJ#7zW;nAcyPt8bcLVnmqI=|(j@*7G
z0vz#C;A+Yp`09QlQ`~tDH8pDmH8iyaF4X;Ow&Mi?>FB0w4k03NdOwo^?w7%H-g3Cu
z`;#cKyI+r*(+cz^JOQ@9kUaobL`IX_a9hCJUznV_4<jhMzfB|fy*Yfv*Fp*R$!QyY
z<t5P1@Y0P?b${a$a~>}*3XT%u^Beoy7_Jm4+TY{^$R%oOwNW^7Vudxx&uq2!-^I%@
zv}IwdvlEgwGKhhu9n97{@SdY#9>e6z93Xt)*#!6^4lq8}@RL4cls!PCjVp%!0Y*rz
zJ~Tu*m&Mx`=p+R)&PM{(1C0Ke*ySH!^2YjudDl2VhX4XHA|-?8#O~Kp(-D<Jwb<o~
zu;*Gz#mZTfDD3(KyZfgV)(EUDI?&Lt&#*+91C7P<7`E023X3>y7^bW%fn9|6DXkrc
zdDbX*peb4oxRdQuaA3P<@>F$g`+y^KZEGbCqHbu#sr!Z~I?#kQ#v!In#b4s!jQBv4
z&iWFyv;bk5Sn3ZH+4Hk$>TH<S4?Q>76(2gtnCb?g%s~bS{p%eh9E!n_JxB=Vt(mWN
z2FCf}>zp=kzlR%W3%uVjOjWH`mbEB6NV<HqObx9fw{IhdyNIBMrsm4(zKs>()V(9r
za@F{4+~gn|P8=gPgdWpkI2w4>ql1Oesrx4=bFeUjXQ0uU-ofqeLWg7zHV|!pgMaX@
zww0)7;3})!!Pbu)N`z1=g$KQdBcrOqlXKZn(|CyF5o#PPoOX0kq+bhs$l5&ER0#HF
zaw{or)5EV<v^3q%57#A1e;Jx>T9b45XL=41t|nk~XAZFr8^O7+cZl$aWf#j?R%=b&
z2VHT9unT`viE@Y7z>!0QK#S!1UsYSwu@MS2Ssumesz7<^yp<xwho&5p)YMufSlV$L
z4$r_=8xC9-$sxun#9rdy$LT{<B-mdaDpd7ym_g=HqpHp?(|CtUr5zQ@9x4pVlOQEL
zmul7E8yltEp*BFLo}%DTk;(|RT~$|2A7*x0xoy-?kZ=cxVWc@c8_w%W(?R4TUnQ_y
zfCbmmYI1$xK}s~oiyn;0Y1gRrLA=%~!-*%Hy00>a30-W1dYa@&(d=QuJg&QftHZS0
zMmPv6*UpID7-$A~`U+IlCiO9><M1#O1fFmiqsCz(FT7wHX=+EAnu?4syTqF8fVZM`
zm<<@bh)0R$_|Y8PT$pxcJV)qoR}-s5#^u;w?j3G(2EX*eCRp}xqX##ZanIy%p$D7j
zoNLD$lfdORVjffv7a8EvTM6&eO1jQY*@mX&cdVkrW!lMrdmM&MQz64Vi8a;JPb~*^
z00vrexX2q36*R5MJU@ANgm42_ACsFoLa-q)8?Kxqq&~l|ls&>mE3V9o*VfmHa8_bL
zQV{5jB0_xvEh&xQ#G$GnX3q)mXl@(N1m8St93ef6KEh;<wALlC+nbs!!($#FVWPy7
zZhVT`QouNvgFoU36Duz1fFDP}CQv{bA$$hhxfxAfh-*?_#?}|ybk)GuY{q(WJ?s%*
z>#o6eI+ro<dhI|<_+n~RlLNeQ(j-TgH8Ro}IHgsc7->-rC(BJumf=H=HOUI8rO8zo
zgNY-8_11JoIsiS7fp(<uF~ElQkv5?(#)@;K^#)96=v=7kBc)aA$=z?lOTs!ZU~(aU
z2jz~GAy-7t9BJ$hz=)}8n*qLo7%KP<F<cwkV6;w=f)a;0tAyr})+I9C@ydl#SV6pI
zOOxe69Qg0SgZ<j{ktWLyc|PMPAr%9sMwz38v^h7Dca)K~4tTOh*^*Mk3W#vFqYMPJ
z+aTqRGXD7lnRlVVQP#K6*ZT7k)uTjsrBV1B9q^m5NHmmy93N>#S}Bj86lor15<crJ
zJfkd)M@rkMb(HZ82QR;`kT_`TupCb-G<hMKjuv*#-AtLIjbI#=LA@Pq++qJf*`p02
zoGo(0<!HgiZw%&+7U}TQ9u#OAc3qLG+J<9^)L9W8E#%^zHLNQQ&0qrNaD_yXwIdmu
z4jJ!9;X#kJZjXdvw;Y{(FiVM}a}*Zyw98$L(N4$cz#PXf%Vdr*4p<f82*f+a7(~1h
zWwk0Itl%CwM(8XTi<rT=V?->7J&+Y9I7Y`ew0>1fScQnvJ4PgV^)=Ku#)blxCCraV
z+aO4nn%agvM69l2#fmt|8nv{QD92}@922eD)85}tA7cXuuR+M|YC6_BY>Xe5$sB78
z;<7!DK<`*%5L<A{Y7RFV`PW4JV{PQ%usP=d!PdaQJXS{`6nXVn;ptVI@jkWz8-NSC
z;hI-5<6IGk-ZYOj<~8ywcED@wPHHLe8=y~;W2JWxBLag+)0g3B3r}tf{xUCAuQ04V
z95MF_X#l@;k+SVHzd`7`T5AC+_X-n>9ujCFUJc}`Rc(eOHiTLt2aVgQ@d}eKAFuJd
zgwZR6`B~>v^A#d3jpUCtgKo#0Vl9P%7%SE_iBirl|4qNb#4sf4fQ}QUIIc8voDhiP
zJ$X*RJI-pdiv@0o;{<ITophXVkLx3RV<M*z5C}_wCa3O&Mm%m;23t9O0%Om=cyZjI
z;?)zlK^w;j3w&W=lp+NT7HU{nn~D&zWU*3)?+6%zdek<2q+g=sIPEPw*-z819*&IQ
z5IbHdzO0`z$D0s~*hX@gvUj}eLCm@A@dBh)K@^35yhwzf8p>(14=YP>ybVk|RbSN%
zKd6K5(hNMotUVMEmRPu9k2ghH!zrb_Bi(dRz-`=yBF=(VJ2N%`t2#N}h6+L=pgfd}
zI@~vOf)EP#4K)4<!a`o(F7i$gviaS(tX4w&SQ9U?+w7BJk70O%2n8>T1SgnKbX$W?
zJHa{!FRfuPhYFM5>1b%>wie(=G@}ztB5R>Pv?iaM#G2fW=z`V>#s_<TC{em#G+|=7
z)&-o2pD0`vI2tT-qVTZ>Z}xd7S|4>YSDM2bE$@qKwKAq|?nI-SZ{WZmf1(kFShMPh
zMglY_j^ngh6>6xVNO2$rgZ4xd1>QuJw@8~>nn#E%o}t$ffuq!x!-a!v^mn4w#Dg~N
z$h{c11L!0Zd=2Xo>%o%*N*?<dPm|b4Q}!eyep24h^iPuR>41&yN^q?ZIypE=$0+*+
zs+!Jk2MOUx#$kn{A{%Wqcr`aV$+(hdF2H_M+pWYO#d`?U=MP)f5(=Z8#O2WGlgy{f
z5j}dqNhe#!<S|P!CmR7Ze*4%v**JnV0aGx0vUCSFU|7z4&G2D3EW?wHK0f{D_bLJf
zEHQh7R6SXTh5W|J$s#>Oyi-Fz14f0SlSN#UaK)Z%@~&59-HY7^Ocq{|TkZo61mt?A
z0s{Ckns!XI9>jSvog#3%vH3m4xRan{PYd8_%${O=*2F{SYsm(vjZ<vmi%rG75}aZZ
z<75BoDMo8)a8M#lPZ1d*WE*x%Lo@XuzV#8EB7B$O@Na68ErGcGbz%p#ls>%3D|SGl
zC1boXaRhFXtITw&i5r^$asOvd6|~z&IsV5xRirJi=VVVcrbqL<Dr>zy3CzhkFx+c$
z-hHY`f}?`4<x)u4jzEQ+YK-?_FVHwu#JawlA_d7N9XNC7*6^Fov6f+%4`+KzffYFF
zCeZ?{1Goh{?I5q^)4<bg(x1MPGN&1*h+vWz_Po<XjN&=dn)<Ajf0}q40TIg{+yAUv
za;FK+@Xc{lU2vMn1utV`-koN0*(5(a=8*Pa-=g*8ydhEu0Lw2UH8m#@VNc=b97F~i
z;KER%@^DQg+8Az$B2AxWqoZqBA}w+CrG3(q*Z91QX99=U0NVo3GqP~LKS){6WZs9<
ze%}+;XJH4I^Gq%Uj;Uf@=|IB_;VUXtPvnS`5~#@VWqA0-rG|nkW0w|rA{88PqWknS
zd@T6M38M@zE2607C}Za$B`!Ns9>fS$LPanS;wLJMO+JBy*EVyy;D8$&5f`Tm7qG-B
ztDkEXj$t|7IG3=joYrDrN^rU`qPn-L)nH}FlR}|Zy9w{fHyqLu%x}!i=yVg_AYO`S
zo^BH3vm@SZ$65|=Ex&2gQb;(ArKC+Eo;#V=<RpGskY*S-zQd83p~Db(ycs4b?7`ti
z&T3Hy27o_9n9}GSSo-Zo@O=gjS`FU1>I{(&3{}iC2d+cj)NQz$MFd4eE{E&()YP14
zedrQbwjOTuR+|i%W{Hxaa|7p%4g+E#`L|HdPyyF?9A(Zhl?S6v-v#o{5bET9cJ>V6
zo1ee(H8~Cu<g^CPU~zbLhIKyl8@s5g4kGrcstQ7FS{;T@SCK<<d>*IY`65hHLBe(h
zV_FCFKmQoDw4h$ZmMb|!W?)ArO>4_MoQO12*nznWYBDo*<h_PRo~Gro$?QyRiM`lN
zqqm^97jrYMqp`K6Kr8n2QB_kL`YF^JBO0X8m}&iwwTTboTpmv+V2{|GX(PLfi{m|6
zewD7J)Pgsh?<KAbMbICLho4n}zm3imzEJbL3n_D^(7zQYjAt5pj-i8f@A9=0u8H8M
zRix&2IM>b;Szxq)tuu{D*wIuy(@4zob1R_(q0fglwBZ0dy&~6;uUIs-1S%B9)tOF^
zIA_CtqUFHUsuFyxP1QX^(@NJI{BZPHhP_n4UAN3xLK@a2_5pfl3HcZ)*|UsM9C#0v
z@cS>8*J#w-Stg%V2!uRK#wqL%sygueG#h>=#<1Z=0~RuE#lLt5L;oz1(02T$0yQ0Y
zUCd*Jh;t}vX_9}$CviyG!aB<=takJwx`Sp3Rcx9-Ys|9A1tI8ldvBJYEDymv&1zkK
z>k{KqGkEk4rdLjD!NK%u3N$Sju&T3622)SLd4%7z5o~MZCQ4(L_B>9u9A1v_mHS!E
zStf=txFqlsIF^M~3QycD=>zQgdHa>j5@F(g3r#Bm3_`rzL9>k=_-w{2;v<+X4N=aI
zK6tYYD&ER^gtD`RpbdzBnXOV0yE3P3_!<EAIl*iZNP(46Rnxc(1HCv~6gXCQHZb{~
zYeP%%v<)~^fWV7jxTYhb$a;RZs%pH!)N-|^K1qpY;48z{V_GX!SbxlU`8dW(nwcYH
zujwLhj(}oEVRnwOrmqFe5vma-{c6g&A}9pxrCpYv;-cytlh8O$<ij~42n{1?xSE@=
zgi)k5*;@c^n~FgU;Mmew+FPy>eh3jN#AV0WpqO^qZ56EQbhhiU@gcoAn>kx3faLH-
zzjwBAB|ndmJ==yC|H=r)<k=>giazp})5f3{#{<FHA_KVZ&~s;tr1JbLs167Rw7|9F
zaNreL#HhP0e3W%+o^9(i6aZ!zhUVEOu+ju}q6Qd`*Sdl(ur{q|%80s<-9MdUtQ6Rt
zo;gR@)a#tyImQ?NtOdLV=a^XJw2uAizBcF31`a>WxfZzcf<YB1IDTavaW&@%vr`|$
zwp71Enyn8};~b-Wkezfc$8PNAIXdr)C0qt^CAhaqEiIA9u@VJ$say@4Ry2I$gXS9K
z4h-JRTw`@>2|Ijmu9?E^_;+4fvvY-iu@=eKjK*K(=31v#^$nb0uCT+uMpM;_*x5kB
z;anjZ;VXAhL+ey<3ovrn(PIT_x-4D}gdM365TJmW)N*8RV>@fE%&UuL(6m<43v_g@
zpy!vrGv^AM5~YG*fOB6qG{m!&)pAI?2BqbnEAZxBM>+SMe}E`ZfY<<0RjYJ#<2S8L
zg802PYPeh%IyXAk<cTQH5o(^RUB%npSYdVa_fX5V!ND&jS}^q}cA)2qsO#7fV#duA
zqH5PuW}bk-m8yB<%`@yfAfv37$fWnRghQ8cz?hpSvM3-NGni*Y6?rCCH4Bl;&MkDM
zqx`l?!)5uHh$00ubuZM2Cd=sD*b%`;^DQN!pPehoJSS;v-S~-^X=P)5AAZ2gNSDRh
z<?=<uHJm4$>rYE%&odzWYb0`Fah^>(4&#LfOB-_p+V0m<aGvm5#|!#Zg@b;bdOdy?
zOl1I5s&Sr(5EqeQX1O}E&!(mpAnp|gmcxRhPaICQv|_KmlA;v{_^#fx7O<YdwlgcB
zar+fUW7b$)$3a3_<AYzzU_P>0;{%pGX12elK17vsrB!+PD9D;}>43`s!4xowvoa_~
za9oxZiRsRvp)?@q0H%m*wO;%I7<)~JrVlXUtc`~&+27vM>_(inNe?P0wnJ!IaX^|^
z!Ie5+cx!JGGnzwDegwAGe35h?Ja*anLKhd{!i;{$fTrAh!I~e?2SI}QA~hVUJw?^|
z#wT7XQ9*^;u!L2$p~>0!k;tgs=mCxiPy+C5HoLcDM@|)IN-dWg<V(42%<y*%KtE4w
z76kIH0zb|s5}k1kWfquR;JQWcEfAhY5QmvvV2Z9oT{6BlS_99cD}wIl*MEWq*2T36
zeu1;<pmy-DPlnoZ0B-LE#x>{|0BP98EU@9T0mf6)wdDKH_<=Yf^s+V7Qp)NuCK6X?
z6L;RU*5MfJg;M(*9G)!{p!huu-rRc&MebYJ`ITKL9O4J1@FR2bS<a7h`kB4)fqpLI
z4Opt`=lsYZeuU2CQPAfB8w+iuvsx*S;p9bY-+=92Q<MDaVXW}rVZr`IDdaGF*BTF#
ziX&}WX+|G);4G9D32{>&gw>r{B;2rj5Iz%6bEwDR-PuLffwJN77uiT0!WKqhAg&P`
z%|#-cp)yrndzgm!r95kJSsyhtqc8SH+FG8QYPvS;iHuziQ76<|WDBU?vzH&qN*vhL
z{WQHuWM1LO&&9$P77A=#77J=z_QG`a7Mpn6;YHcSHZFJ*;V(8}JNLk18%pgbM8Tsr
zVZ#_Pqg6#E+k;S1L1vu@9B`0_5pEG_wLy+~X=(}d0KZukFSZSeUFd+hu~=APkA1RO
zB;VIQo|<+H;Aisq4J2A(J+qd>>oZHFWf`$z-V*H#IQN%`IB--8D}RYFj9?kwzvi?=
zwi|(_=e-!dhpL*E=i@_%0hfVZO$`MACi$nNq9xX`YhfoYu?fl#qvGeGUDZ)m9WK|w
z{ts6dN0ac{Pb;<8;fJ-C3bkUwXO<fHF6^q88mr>+&$=vr(ij&9mI~)XFm!VY0q4jt
z3KasbHe=zhY6jOBIY4Tub+)*eF*+Iw0uzrSt%*Yuc;=c*MUvOx2ND%BFHh83YD1?V
z<1^6`sXx8c<iLsxeY8wClGi3Pnl25D*U4KZBzLn{ahWiRv5jr6zf2&@rF1xQGzWp>
zr*<e<CbV>|rs^`IALmKfyW+5HnT>8SRUV^;BC8K#AVrD+ceAQcfLeex;?wxiN23-$
z#fKr(YGc?yiNoB16?R$yfR1(8kS!N_5rTqXt;}-a4`Bfa`OtKQeHPi}(qR_F;x89O
zawy9yOHT9dy`W&Zu(lSOeYptcrt9#dRSFNgEtne1ZK_~{h%K>rb`%HCIlc(q)Rq-I
zsTD65DOPKuV_S+0^$;t8-cOdB^sq+AWwL2SzXs;i3X>z>AI8d)Sz*I)9X8M_gv<*M
zqwEUPN&GNq`+B#(!iELkqseJY*cfc|t`IrNi@wzrA`EyB?vNim3ms0rMuuHzV}%G$
zVni@DBW(*$0HP0-0z5|)J7j#LsHNYBFdA2g067eES`mTqRr1qVbiVO2F-e*8g_jPT
znw>8#W?XmyWnB$C0PLS{vKUo=Nlr^A<h8uud|_4{-c&tbNJd|2kf>%6Hiz9B?0KyV
zLI$DTBL!2fvS6Buyu1bG#`z*6gtFgEEr%O%80<by6k-*9&z_ZOrG5i?>;h8_*s!9%
zT$X?3g1ifaw>++BW?jiJAAI>2D7L4teac;6^8mkmjU8NYfykzc9hYl_`3(QS1tLMb
z$_ZW5PzVKPBf3DOf!J<g%E7}315vS-;*b}Ciqz6}YkTmZT;kvt%SG7r+Rz0E(Ms)_
zxicuE$=P$@7Fa1%=ug^YSK6Q(fu{9Wnlzx~AEw+&?c_cRT(ufI;Z|DL%2Evdp~XY$
zI%p^YwY8#^whrp$5-bEw2do>xA+hG*K;v22S}Agch@t6~<{jK&n!F4bD(ONSh7fS(
zLZvWA-i6W-u1PFg7aD!oGqdZ-zfk)TKChf6OV~vR3JFj4@)pdsDUC@qaj2l#>~KZx
z<Aq3Z!Td$Ahr+v-AAr^5ARmuuvJ-dw62*dB%go1xrabui0{^TlT_hdA+a~-9T;?Jn
zbQC{`=V?Lv=hU(n38S-bB>y5EM59BP47qkjejF^gNJLN{g9fc?Ys{HGICDZR;qzT;
zXc`x4@s8L<(%(>tLlvB3G_?pfn{LO?%?hwGKmFZuz;K95i8cc04n7K<R+4Xm0PqAH
zU)kbokr_?$Gh?16tLW>jue8f`LSHk0bK_x@bD7rLDe#3^JbPSqpF8>}^xf#>SLz#D
z#`7CHEZ+o+i&$`TBC^OA!RB}H*#X7A%~3><@SO&i$4^$mv`Uo7si!cD92$g_;+JYW
z3@iKYDAQqM3%_8>Z<2T&!ZchWl<hDiop|sM$6y`8M^889TnT<S7CVrj!v^v0+o<X=
zT+B}>haEO0cfkH@D2zIy-ZWK(9jGYT*&l0V{6Gy~k8XA7>=}W}IdNb%@wQQ$z+T7C
z`bs7FNxe*`P*}%RO-)@rgR-55A5W9vmw$B}a0o|E(|9Zqp50E90|M!I@KifZPI93!
z)VlH(X+vRO2H!=eNl2a*q^4`Mj(_qZ?i5*WN3zu^5*n!R*-4_n#EhKQv>ZuOmx066
zj@Yedx`YPYtdLW8ugf^$_b0Ml!s~T&$#>toVc+C5*LH#iU1lnXD?u)WRuwEp>cu?b
z!!Dt69X8VqP4Z9nMqM@n`8a^X2bx+q48tr|9DK(Jj-Zz2;5Omk7)l%yd@~j6uI43z
zoUI+|mL8Coq%++n<kq{;30}8Q!Vi#UwN@AV|NU;^^rrJEr(f6LK|cz*t;hH#KC5wU
zy|x0ku7fM55{+&X9O4%!QXrTo@HlmwcqGzB9?*^*U_8dePjI%{WZ}duQBu}cXu7*S
z_gOzuk9AyLe#<1&WAkC>8_DZ2CMWpzb=H+(y!)ELq(|&^u1DA(XX{pzTUe93=D1Xb
z-<avKe$ntybXt#y2M@nTjtqXp1Kc@OJy@6&WCfd*R*!T6VxKV<_4Bnio=el3tivfs
zy^dxMLCEw<gVw|Rq*<@=%C0jg!CvDPCvd)2#Qmry<$8rK?9_u^VT>;rR2^Wr2Qc<}
zt!LJ;4WX^`LwuO*Di+wI)^rv75Ih<8nh^MTL0M=W75IVBMB$V;_J6w9dRX=p^7C}G
z%7zaA3>gAvGOJYb?4Vv{?DVgu>?$K~JJvRTl?f%!zoU?I*+Q8gG;no0@L)hyA(r^I
z+A0;^gY25tN^3Q!F<RwBiT;8;>mcU%+$YwCfB{3ok=2RUrd&PzKp3`+t3=RSF%DM?
zD(DVe3|MVUBBYA#IB&Hv!?!20nuDHY)5u?Klt3VSq@aNCYb_Xyt3@~?<tkMjh)bY^
z!`0RcUFZeg3N}`Ygka_BR!dvv2M4LCNjzeO9}PD#$}d&54Ezu!j^s)3g~e)Gt4dS^
ztBy57NdZf5W{q_d+;5(KFTrP-U1Ln<@e_i+%Z(vICbveUvzq+_fol(=r@F={f{Bg|
zv1>Jm6{Fz_VX@*JMFoQ69>f8x5y6k*&X9hUH+WlXOe}c2PA-!r4%v#8G_AyL+^`Nm
ze=hu7i3fW!7Yjdn80}qby@6n6;K^QW^oe)bzgYOd^%Q7(S0vZ^gLV<W5m~)hM1i2l
z$Kl}!H3y|%Wks(T<Wc~?&>U$yiNk6-kkHZ48%h~ox`ku6rS<ychZ8l8z4H?^?b@{N
z!PKWqtQX37IX82Okze4~61+=<d_;7hyDky7@pc>3mGA0J@|4b9V!KLyIjMaGDR7W$
zScj^Yn4t7tN~qw#B!1wnafzAgh$z(S$C09zU!`kaV%@|)J5zy)5np06W(?Oy;YVm)
zBK^C*gA#36*a|f;?MT2HOSD${aBZbP8BOv}A9-s{08^g?mur>dVN4-^ts%kBKJ)Ld
z<lKi`+r6u`@uMi{LxqD^v~e<y->SDkf58RRSZh*6m>u@TE{mNI%tZwvu??{!W;JGQ
zs|}vTpJ;lrjJ^G|0@o{{v`)tkejXfSKC@2f$Ke{@z|r(tK6}ru6W+1mAb*{)#g9Zm
zE;)tJ_Ra<CtfTa~=Bh%G=L$ou$s=<L=aQQ1(|XZ5V-lOi$Emr_1hDx6iq}c+%$`jx
z_Zh#<n6!z)O{r;x#Xn`s7p>O|webE?X1z4+$39CxKL6xg%B~k&=zFoY{PiLNtQs;H
zH3uTs!CN1!7draI+h1KT(%Ha6I9zXo3P0W~cMTfa5W3@8o@eWwnDCeZ0+5;tsXWA<
zfw)Z}+|Dhf5ZAC0t$`ozf?YqYxbUb7Tu&QJHvH@(Wn5M+2zf3mFCb)H7TN|=ZG(-^
zL9Dg84MsJ_i@azPY_R^qO+LOvwZR0BHypwZBE5AvYB(@(5K**21iYPBGOe-#yGYlv
z^$}`0M7(ITHJE5It`I=aO*?Q!sB6j@RWMy4T%0B6jOy+@4)k(@E?-2zgBEO*%T?rS
zu6>O*=Nb*b1rp?h_8VZ}=S+ME8pL6jLhFDe8`=W53h}TCMLA&|8Vbg2(*fJaE1Ao!
zhN5gK1RQEo;$RO>J&N5*&U$@V?pqLDx(7ENkI!5xoWQw5-lc+#2Nu@oOO0P1pT6es
zAV0suk<&JOdoIxA)O|3oF13DBzf|ZD;YLuM8kg!sz8g9^az%Iz5WZb4;xI9aFBKu-
z#haF+KrX&rYCWRcc+9X%jWagHc!}O<Rn)DIm3pJ`x>lZ5+h{Ft*;{VSZ?vJc1sCTx
z3g$I9f8A(y8<c<hTrJora^ihl)j=Jdx)&ZBEveLSn5Q5*OBA^}xR;68?Tx}b?lQq&
zxX~mF_ZPJs(NJCzE1W10Jd-%>fFOVk!3}-FjfP0!n4wST-Hwja0w@Svyg}FXnH=;A
zIpS0L1U=rF#LlzNx(RV{IDicLgt4jnaR%0Bl#Rjo4;B8JL25Yoy?k~O^_dtEBE&B!
zHWj=CddCU_EA!L4Z=sfhD93qyChm*DT%QTqp8CP{(td*s2LVUzWcsZG`Cx{;e&cMo
zJSr<#)>V)n()Rm>QM?gdrd+>q!vR=9zu-oIEB0trS5(7qeue!a$UGc64%#3eVMYBW
zD#UQ}o<?Edxkf&)>=(K5n_(?Y^4gau0fSTbVd^<(_(dk%m$*!5gM%IywJ$T;YEWL@
zWde$q^z3COB9Q63T6K1}X?wg$f%nl|wGCUTss&R|;g}@UbnSYmtcGh2c~Eqj$rcf4
zJky$rb?QS<{+i?qkFCp$XY2!@S(3}Fn@a9|gJ}n<$a@-$%*&0P%MtsRak)O&NuJBG
z)Uub05OBPMQK@ODVszH!!gocywt;5KMIKyL*B+Zr3NN=2fM5#oM>j4vzIc_xej;*J
z^p~BnA}Auvk5^HwuyOvtF4D`5S-6|=cx<A$@CFLHn5JDR9z}qaLYv%R%=5<0WozY~
z?<8-N&>;amvMtPRl3v{g)79T(Lew)&-XrBUSx;YZE(My=@Q<oC<X?CXH<@6oSUDS7
z3m2_07$aB28@;B>;X!GN6#^^8mMiGSek#!de(85w>4i{mAB3OArYnq~-LIp}6*`dy
zpa{Gxgpmz>{gl1JM1VWZa0a_3Q%_Lt3hNrV_=XiLP#~Zd-Pi`WjUU!>P!Pa~mBB&j
z>!Iih;h2AKi#?B5$kgn_J#}rf4udF<O=3&INgPt5?Kx@=Y&l@HZA4dE|Kh1nRwq{q
zCy3l9?@F`K+s|b~Z(b>=5uL=b+y0ftDwHqQgey%tcyJ@QQpoDXUfb2m;ms-vT`dGM
zG!z~r+=;`kNO5BTP}AY&cmlcV7HjDS4i8osdHJ1;L^0vA=d{+SAPk7EGSY_f9Kv^%
zP_qN!4_Dc6*@U;OvsVepcpwEqlm1o0`T)eA(^@$4;->%ttp$q^I;MJ+2@Fq&<Gdtv
z0JpG@L}B288;V>(AKnCPUL~UiP9i+q<A__wj)ayHw;j&8t8Aje-m0JrruTf{JNs}U
zTy5lU!?4R-ZS{-#p0TIJa<2LyWnE?b9vAsn3wveg*4)*`Lw<yx0t#FnlNxBNu3-^Z
z1wxnGgv$(#tF4#u^NEqd#622nYO=z&|6G-`&!d*h>?l!kwRH<lxnM6}E!~2kblNP$
zu{zCc7D<5;JdwOvD&U6*vYQ1h|G+ua;${J3KYfzhY+}GT<g2;CX5+6zPS>k0&o@NS
zXNqms+*#DvYy#yQo{{2Pc_TFyAJ#?mYV0blgLTq!xg2syyCRl9=tCte50%gNMCckL
zj5iwql(|OO9KjFRcv=Q>+OsL^D)YOISbh9!j5Pkik(@Ri!`;{58e^|8!Vh6quQ93O
zGBs>v*N#I38rN8V;i0GK8k^GkYA)*!hZZ}LClM+VYXb!KV6ki|))(Fhx0b_-O&13K
zv=&vcO`>a!0DWV{W#O$KAkSrCOkun2vUsc>Q3TgY7hiKV<*qepv9$^(cW|wYgvvNO
z{51_1FkV;CCJ5?jT&tsEj2{Jxv`V1}d!wmUuEh@LS`$`pc^t0FmNtQIh6YQnH4%4X
z#!f3`tnE0KimsEEZ7&R-GrP`~D1E+>f27B|&ZPUwS5o#m69YVM@FiR)z!Ad3Cl)!)
zLDj9}cn?>#j}LgOT8L@NA_)~Ss^;;WDNbdslTlO0n21~vMiiVPO-EB7w)J}696Q)>
z+do4sm&G|d);FaXDv!tO%UEpkMUg-{1HYD<2?QtJTp=$Iwz$`MP-g?{ruOFB*G82s
zlnaDoIbM960{wpa45|jwk##;b2?JXLH;ES!Zb_}zQ=~{!D{i8uwtxt(r&ysP%|Bk&
z(h?jMLSF{fxB63|(}DDcgk9j(Mnx)Dlk(P0v*(bPHwLinXZKn*Z~cZN5;&5zvOZLh
zb1f0T8n}vhmk_@Hys?6#b8H--Sn?vj)tGj98(CLE&XLlR=N~9&DnvXrhmMb3wQjaV
z@*?yy&LI*_^6LoG+PE}Q-a<5Bq>i)sof!~(P_y2E;p@e)%(~n<EEE2KN<vQ621F9-
zM-Cjc367Gg4wzu@Yd&}&hr$6H@%mHXJdPU*fMamL*MM{YBBk_!{-ze9tMJ-MtZ4ad
zD{47Rh-WWRqD8pn1U)h!Oyda|qJpq92e&s1LNfFPVoM9gvxHk@UAC9Kw0^<H0L)cP
z>0H5>7$`~jexRVrMeJUxn#P$Teon|$lsl~$hXoN0$Ay$Aa#(l|NX>!}xg8d9tVCY6
zo?4n>ehMOeKJrG5X$KiL6&tctwBa{nCS{5O3a=}XSCnQ@H+;;I)f|L3f?rtli#nA0
zDOWW9VPeBaqV0D;w^ubehI748D|NMlx*J84N|)@IqoOH(OpdW4A9FUfanJgGii;*Q
zd7uerhD8xSj_wc`q3JyTYU#A5TaV#5pGrb}`_=AD$yn@zf3;*3OE71)B&@96Kz_-b
z#)!hg3NLqEbFOI)=Z2cWt{J+bS`spF-h2y%nuZ;Y7e`A*FLsI?V;+@6rUhK=Yq~Nh
zUfe>YSm|Un4ys#@G`yIK&FCK77>OmD&*~k)tLcNr2sW`WBL{^ZJVOM{KWH?_Pujtj
z)Iz+wACW((^dQ<Pr^yMtofl|wD@JKmD;G*wfx|(Yy121L4VS~{rKk;*!%~{Aq#Mg<
ztiXn$QCfo{`!Y^L(OrW&ZgMpJf9QI*;Kr-t@w02IcFQlnTJKhEm8<qeKYr{Ye!l>E
zbg}>Br@0F<!)V6?@qbBvkv(INJTtZxUuHBzCJ_)o0Rm1S#6SolfgngA;x29zC1w)k
zt_cZot|;W@M?Ly+Ki#d<lZRcGAHGNDoIZWJ``r5U=^obA2TyV=wJ`xnMvq-$17ZGL
zqnbHpP*%+-DtFAFoM0E?Ux%)fV!$|silcPQBurgin!#4Zt0TP|uSzY79%Y-i=0(Z!
zPyvm|#ZjxYB6XPQrQ<1$$ieZxcPtTkWf&OHvDtl42FDV!R41<Law!|pnW|*eKbLsD
zn<*z2&gvVgTqJK$okzZ$#CGEeDS9o7iBV1syYob;l+ys>g{-f$wW!P8_nIe$qb(Zg
zeFa@!tDFWS(>c#xI_1O+UR+2oAn!4i8~o{)FZ}QBLl>I;fBtK@<e@KMHzhzbG?$>>
zh_CU{NNJp^%v|b@KC3>)yAjW3y3m)B`ME?R3nN~fb9q8E(4#~iHUb<{=WGakNX>6+
z(Z&kATAMSGx%_fz%o!sV7P(lAA{MK{D{s8T;@O)^M48~gHzR$ov|D#2{<%7hLg9Gg
zatg3BdOUFq^X<%W!-jc^v+42F{~c&i@?MO|fj$(E8$v596)7E0&Eg>%HFd?y=@l$`
zN(XmTBg3!x_hEFx((poh8_l(%x5MnAQ+p@iv1Mcl>m5&&UYxe)FMWr@>kTyM4u_pJ
zl#~;OJ8HleW%NX10Jr{}yicS)p5<xhiPSHBk&%xSfhDAP!a6pk`Y!)GuKFOauI>cs
zQ&hl!O4cH!VW1i(5;4rYNLrD5ni;O+sa<a2-;1)?yP#?5dwY6mHt<ii{gRWZ`w&YX
z5R9HQso1?&GAC0Xr`Y96?xgdC`<lF$DZ;2IbDcevi{#QG!Br|J4PE4l7hbCR@p}J$
zPS&DXi%JW}yeMM_?{I2xdomFbJ6q{oN+ANhD2wYK=8=8?<8Db$34>uDc8bg=#vI^6
zaXxkEteS*o=B)`l?x=sqoTorwMSD1Jyv7~?-Gb*J1Zp^6@*KU{kxD?{fd*(k@dk}G
z8<MqoBV+~pO$Ttm?nR~L0MPhwPMtTJZal%v>7N&FW0yRirY`fTV9!M66Pq|;s1F5M
zu+|7Wjv|(w=%}YbU+o*X6<)9oD9(~wNVLS3Sn^R)gP5}3MeYPLR9Y~C@oE7&Ru<B-
zgYrBbduvf&eNH2x3)Y7+!=SN{2)sOn?OQ97vud2u@ub>JDaJi-cfx&P_M<EgXxKtW
zk~p?Zm5Yguv)t`3T7y#+*7U_hDcyM5>&@JvcU&(3^NR_ej~XMrqU!%ji^iN)=99%>
z8(l!%wU{Wq$8FlzqJUZfvBF3Cw5koH6%~%D`(qaRA$|+o_c-2D94I-+MMadDsILZ&
z!x<*cO4_!gG-6^?-S%k3dx^K0OvQT7a>yY!S4sHtJOCS!d?hjW0J{2Oq%f6`lEaug
zffg*PO)$k`DO52su~>4&<Hb&}+BUqt*}!vH4QM-&{EinpsUt9;O&w*7NIwb|C^+zf
z`b<JTl^8d{)bc6o5-x@%^QqK}Ln>F2i*h{zctvK#;sO;vmAZ(}LMcUR9|%{XEPl-x
zthGqH2(NjYizosc-UiF5a_v(l8gs+PZJkdgN#~5rIS<CN(8rIuD;~ac>pMWa70KXJ
zMne@7ktORW)|+hgMwgOU`fU{>J+qWL?EcGhOU_=jBl#r*W3Nq$OR1}s3lB?aDaj%e
zs68}^>=TTjTEN7@izRez$%OhCYB7g?9A>l}G5lMw@$7h0bLeQLx0D74+M8+Vdlt(c
z9gAkldW2n0SYQmRtI+7O;m$p;WR?wgjBJ@^{2N^h(LAD#UQCl&TuvnMlc}oGy*wWM
zSpgFqR>Rt|k&<Z-KCLY$EU}KnTqu(BOoY<%YFuS;dR<Nw)?nsv&#}a<f%KP+FvUGG
zSWY92P7jUBWyRWK*<?x|U9rCH;2Jrzk^n9*PHX>GtgkGSjJlP?rvmEN;z~l2*P2ME
z!(ePlCCV*hqg(R~y`Wq3MwL2fo>vUNUAv^Ul4y++IqiSPtEfi07qIrBGVObv9dIjJ
z?`eE=Sxru=QLdUTr<vrfS|>4W;OaKBntDR7hTrPqD(~OeCEg>&)zl|DrIc1vhi^av
zDyv2%UY1HNGES<pdDUp9u_^RV8mqP)9IE8ZkFu9wp;D#cjC33^(nr~2Is%x!vS0BG
zd-WiL)k}P1A7=}QoU&f1oXF@YqhR)C$@tgay^=d+xM(B}+9wRdDMNY=2XjULTw?!M
zI%OT>n5w>FE2mPgSqx^SHf%yq9&27lA0;%R6ei$Ep@X0cWyV$KRFZTy*?ynbivp&X
zb)ahAD+32Qr8?Qu#v#@qri;<j36E21jg&cUs9$%(KFN78{E0HS^HI#Uz(vnP>A?$u
z(&@AlS0AZr^jL8ydfKR+HrnwPgpJDSB<H$CMjAcsZw~Y{6o3eUrI%RJxzy~&Gg!+9
zfCY}nEOe)n5VLL2m-GW~copS@R8z-*e_lqb*0FQvJ}#N6p=9ypbJZkmds$ub)r8<)
zd*4>{N?0*SshT=zh6bERE7jDvdDvt>9kBG>#M-bGbrIk2??v4^w*s%DGgvxyqDc+s
zi4r=p52*U0A4$&q?5Bqao#Gzrj0+i-`lG{7rufVm<E=k7$epo{-e9)ao`L9Gd!-c5
zq<&7ZL%=L5G7qBbi%g#D=`q2faP`2%@834<HN28){8{BZ<HDor_x4PpJhxyBwScF3
zP399XYSEAeXA%u@gQ)M0<*fBeL%@!nwUPGWk4WaMb;^&(xwDC53oQHjv(^?~F7YtR
z{OCz~d4s;uDV<GBGas82sCaFgCWtEKtsT)N&D)^n$FtJ#rcW{Ayu_3)X#txv8c%CH
zXA%1MqZw6%t9!A**+k$8i^m|h7}BZ@rIVw#SSL50W`gvLWwi<CG5IalQCya&_53Xf
zd0j!&r+o32M6HW=VI6v2N?zUr!B#xMttMNLn!}|#dsrvdZZSdNE*V+8#mIWa)zWe_
zaO*7{FFwU#gKd0oh=#=}=zHS21@*o(a6s37Q2V^dr&Aw~ns1}gPg^5QE4;PwZ#4{%
z+^4O<PZMnR(@8K+&v7g-e%i>O(Z5QmF1yvrr;P|Gnjw146TSmo^9qLm+VIaQ6alSI
zCxOtA(;bJjpk82l-rmyGF|#S?2g0V-mBFQooa*JAHH%?6YsIK1ARjLVGUp7>V_0nE
z{PW!9S4;j}VusE|$aV2tqK95g(<g_4Aqa!05~Z<+HM6yIMwK&cJp2m>ERBGmgkwi_
zF7<4h9e3wkq8{Uti6WBjT_eiwSe=zYKtgMP0LW*o3v;}z9Q{lJ${AKNkvf3VAorQH
z__4KGSk^UN@}Ef?R`qnn_DdJ}VpPLiO-i4!zTm#3@|lDqYb6%+kt9SaVi=(CesyxF
zTx@*C=;%>4^s=rMun(gp=zPYc3L-L-1UP%~)b}iGlNoUSZ8u-?*)(u?r)g#`qwxz3
zug-Ir$$U0Z0fSh{eKxU<;8w6Q|Jk&%S`0TORPnRc#{*pT_?LQ*{j5=SmHVbzw4h6y
znu@J?<_c3q<Fh_8bpO!upO<-f(D|$jy2U2%In>kaA^m4$@8TUY_-rC5cd%`dT0&d(
zS|c@|E-d`PXk_i=>^)XX(m8d~p6h062~bhZbFwujE1riX=U@<2rl9cvbImo^NWNx-
z!HD@b+u0N%7n27mdPTQ1os6tQP_R@>JkUTos#qx3j980ob}9~)9^&<oShQ8gZH_mc
zIz84LQtveu?v~WO-VG>h*Swxh-f9@UUNfHT-ysdJqH!tEIyIy7s4(>|w&exqSETLz
z=C*Q1I)MfUMfP1?M<Yd0>jlN|FFqri0hBjOx~}ydlnutMD0|4Xc~Mp!5KwvB%i@s*
zcgD9G)|X!;Bexn8FuZy}MsH30G5;H5w;H2&xof3NWMUmOFImsTI9hj-IfrubKCVK>
z{Tm$$LhRpj`mnm-HOzW|nd;Ug?jD9<a)`$p7jKo4V|VUmDf_Q#;xCmzfS7U9QuQnz
z7Qn5@vi))0v-sw?f2+wSeBL)F>y846>v_k}h*~%3eA^s$*G8wb90oz~mZk0AbIXjk
zPE^{ZD{%MLBx9IuE$ddL7X`RlJSm&rmojQ|hi{K&-jc0bV{jZbRqY_^&8mO<ZLe1)
zFy?K>p^aAnK60D!Z36Y4|7wPbW4EQLLcM^Q{AN5G->x%~z0IV4iilUa+tNhJoiHaR
z^R+kbW!3eK<h`thNh<i~xywH!#oK(4BXdV((tqE0i4k^NnhJKKD=OcXSjtm13RWUl
zRnJvVu@_ll@~+*Mx{tAz>Wk`LS!3SeJ(GR<ZAlhUS+-?8VCfdS;VC_QbAWF~9==nh
z<>)Qp|Fa!1aB##}yDjl*dFXSu8#QOazZa=x*5XZvGsn)L@1KP|hh)oX#bk7=3<7*b
zWqmuUs}Y#oo`}H)l$cc`w;QQ=iJ-H_=<TU@YRf5Op2^lo=bp^%#;=notFoSfz;Yib
zx!au$IDeXv@c=P9!zLwvdm^X>G;c{E@)uQ^QoP+H=f*uU=^2Yh+)Z97%IR)L%H9So
z-hdu$h&0hpLAvUAse`MUw`LD#?4?3Ipy(T7SsPZ=D12|px>sDCW*NFYiK?E`EHYv<
zssfqa?y!><!>fIJVi}PEP(ZQ+x!p#?;tAHnZsfYkox$hr4*i*Ldb?8?{Pm-e(^Kpg
zwmfH!yQV>qSe5qM&NfUhFs@_mlg}9&Ev=qvMn0DqhjAPVl+UFBpa<76PcfM+K!eQZ
zjN&{t$$l=)K?JMLIfN9rWIO=&zJ#3TKWDt3Tb`A|=ThHy&v+QNqGwJqnN0d89yO)p
z&=E7nLOBpx)|1BPj94rk3sUv(ooBA&PB%&&!-GoQ5yjkc?Q=<p^hGq(UXOxP8!t%1
z8=YXCYC2t1Fc(j%_S=y-J)^II+TIk3V65!|!aUD1y3v5zUg>tqv(Y5$6ucQ-X8Wo)
zztSJlgIRbhnsgV_18)TsJ|Wv)Oy5Du9f>7+)jD!V>XAMm&~win#ue;m$NX#W4$0h+
zgi=HQvDO@w>>Wlqw#g`=1EM|;9uJ7nL9HP3UKEx5q7?2(qyBIiNAyy>Bg9271tueD
zm(ry>e4w+jE(cV9vo95gYVUlifq<UtYMvTyE9wFAra9f7c&0iLoSv0+@8aczv3GEo
zM%*J!FXP75(hA@#L(;y(M3px|FsI{LsP<OM^V0RK*{PFstmmjR?qt*ZJc}y<o}Iz6
zz^HKez<$~CYIdO{15d!S1B<eKM-p&!^!BVy?lgj6(;_2xCdM(#tA1eg&O}OlaxTl*
zoyNy&FqpbCF<iGu*{H|qw8iVfT-3w7yd~q2L>jQBx-&6Mot(@_!GYc2zSoNZ%se<Q
z%Vhk(3qsDrQi{fx*~OOcG^s&1$aW=i_VYrZ>OEsyE;aw;wpQxVIzFwyX3ewk{vqpk
zCgP~yL){fO95`F#DQWsQ>kbp<ok@Ca=6?lC@U|0YH*rZEi=lqC8#o<c66|@Y8xP55
z{JyBMIU}`<r|_)+Z#$PaP`VDhbv=j4_V8OD(aT-NzK!Qt?(Rz5y58;xhB@fZ#_md!
zAq30Z<r5TxU9xu>AE#FA$d|h-F=hqJ?*$o;Y6oW}e^+!_9f;kE)pEgWBU$P%u6UP8
z$HAgZdJFdMrQ}HAD$2qWm+wj>Hm5#1UwUD@qR&Y+5MWD!Elw@cdNZ~lb??$4X0<hk
zt8asK*sMFMOkw)g#uNH5dP$mhC85TB!#%w>r4?B3XRd7Dm3C*009b3gfsW%uJolD#
zz0EVA+KZZ(XRnaW0K80$Cg$gU{6HK!bylzy3^}Z?mIjgJy~B!ZJ2ZPkid|7X(f){D
z^2ta&F;k7AWVCK<p2A1>SlyVZ2Xf4z{de5RO15r%-KB2}b6!R%a8$-U0h|8g!;<$3
zME#Y5e=c)*SM<6T&vUYFF?uL&7hLYkg5H<vY0MMlK+2KQK~!;2JfD`TgThe-uc4w-
z(-vm9UQaT_J0xfXa%)~3^%G2t;y7?8-S9lKYixRZoP1Q%-U{mMpPOazt0x{4T0<XB
zbm9lb$P$68yAFz#k_}7U1og^K$Yyk+!gj9j&Fi~R3-!7+Z2k-?j6nd@kpgbpfdTKh
zdCc?q#6E5>P-cyM-iS5yg*Y0y3-&s5%+nYhYPf5h$$vhL3Uz6t@7=OqA0`~q)$sGi
z{UsclkAFUO)xB>^-b>EmD49vB5RiuAEX8OWH=68wCnG&SrS2+9pHG6um}61&rf|9_
zm4HPw;W=(7rRr^QoUmv2feME{jeuPDy87sBE%4_Zf~-dy*Q-C!#;_b09nH_D;a6F#
zq-Z5A&!6FbyB#=aXblZ9(fPb_kIh;)+CMmbTsJPgXcVtjvibQm>Fs7O>RG%G&=cfV
z;B!#T>IacD4_DjXyhcKoyM2E2<Kl>?V9E)#W4YUyat{9$W0x`*8kxIoAhOk8lB^fH
zj0X+5yAzXViL#;UiSekmgH+y&9>GON;qEj}d<Y@}#k*5)=E^b|prAZ@cus|PC;7us
zH&>tKNaLA?3**Y&Mm1byRK3xmd*+%qI>i)Ij~4c#oV+_pg+3^db<e}HPp@wR_B^J6
zN2D3uIEr$p6*VmobUO<31Q_*LCr~H2nE5>G-kn6i^r!BQy%#VKK*ddmsredlxbMm2
z|8C3sut*;UcVGD5{{PJ6)Ijck^33BrC6IfJOWdz5$jCj3xI50u=sm{Q%dWjl#_mZ|
z-V`*<+>>a+^HkPO2cS;AC3jEat@><Y3X*ZJ%KELFgS<Czx~y8F!ac@|Lp!ACKclOK
z|C0g!&?;xLQY4ob*+Z8D74&#gaU55s*yTjI(=#)?WsD>bi_oG)wM7QVT4b(hx7QuY
z5>eJ0QI5zQ()6y`bLy7kh;cVtul7A@_Td4wE^(cxH@i3|-N-gAK2xm&LZx?JHvOAO
zmh;kgphfc<xfL)lV3L7n;R;4)S;z4BJDdvdP1INSMly1*5pmZZ8ND~*ghMKggg17t
zag%kB!)4~)G%$#qiUawqXY$k(g-6a4NU>@_?=@ztuAujl<UNbA{_j^x;a(F7!h}dM
z&^VyJu_uR>G>R>8fW4AldNPxG77)(rWO}cO8@Iq5>I36LiyX+f;*O#R;ZogO(|B}i
z{&}3KXgwOn@`<%E`c+1WghD3pz4lsZd1H9nIwoz;;Oaxy#b^k3($bBTs-=7PCK;}n
zqLUdkVD_Gqek7*nWXl^s!-H*RWPI|IvK?5ejhe_867{D}sh6mcFBlV5o2fhb(Jwe%
zv4t4>f-!ABH#^Lr{yT$RvR^Q|;k}np=HKWTSqB3$N&yB{-XTM6E)>s5;S0u0ZWnZ-
zEjsXh9AYan`Gq7ev+kU+^o7)Mj}Zpzyga~eLfh}4NL+W_byD@LMNZae|D#d8RF}G=
zt)2$fzF^{c=9sKI(hPxDq!FNYV+7sw96NKjBIhupsvQ_{S8UPXJJDU{5GJ4Q7t#t5
z`?H?6USV)-2DB;8<NYYB8U}qOBwJpVm&h`Rvii(%J5o1tGL!p^__)<NDI@nKRx$-+
zfHQiZ_3&eiH!sEeJ?=H{OQRG&a~$bCxr>Qg1G(kyOY`imgEAgPu_}2;^7ol6V0j^h
zDE3i!=fBV34Pi2BFmara(tXBby%A;Xl_OJ)h@^5~;_?v=$PAdO<69&m%|F$9QR?2<
zoGFCYqSlm#eqiAXsN7O>9*ktAjMZQ>Xu`psHd;;zLQ}P*9YuHO9d75o#4wgEoJF9X
z>_+W99_-QaGK-oO+4K}*x8ukW=|^gr-RxEn=;*WzqM==SKe+AmTBHHF-<Uyzwmlm0
zBoLWExM#oLnXBGj$L>$m(m-YOBojZdAJrFV+54S6SYPDsPphJ?0OJ9b5J4R7?+-Cy
zvoJ4($UZuAgp2w}9#aF($!JDn(bI0pkvg-?-mn~**bCJ-16LdeF1n@a<)Abk*6vSI
zmAy>rkx*L<N2cg{_cOBY09KcoKkrY2ewnen$XU1<m}4WUN^Nffa-gE|Z+C|pUBGox
zTe|n#IAB-SbLju^A6Lp|z|u!x#FmP#%v4!8QLH%)(RiYqmcji=IQ+_=0ls|+zu}=&
z)-Dy&pv)Y*M%IjlSljVPb}fxOvm74tEL=}uL$_wUQ!D?h1KEM6KY|C?wMA|%u~Qwv
zj0YIiM@fDy*vYe+lTvU{i%a-H@)p!sL?+h~HO(WdGIA|R)Dkl#Q%>1yvMfl&TO6Jm
zs?ok_xuZdKlEa+pxh&A?s(vlX?&P*)ea%RJ<$p=Tv1S8cVO*O5gu%p)ptTlTOf2TP
zzfgmSb`<iV-ib7Uc9=oBkydR8dl~Pq?i^%uNcdyOP2X|SphT?gTaMF>MH%?l)z`|l
zmpJVHk>!hqHO^B;zUX5LfA*taOax?3hR`V6zL?me4)Zf#G)|MXB3aMDz(h?ma$hvQ
zs~%t9_>VjEE3S}yfK&(5YE@PEq74)^_+nlS_`6Y%O*$0j2IhUZ=TMGt!Q{p0omx{>
zJmt6PQgwJUpTcxoYTlWRmzJdN$eJ!@EdUqQE#UgF3K%R%!&})gElvNKtu)udtw3Du
z3-w^vcGy|Mn0Fk5Q*)-M>3Rx|>TPAu+vTjvk-@7hVP4wzW|mp@yg0o)!IPvxq%<z(
z_7{_+bM3)OC|^pm3OkRHFD3TOV=1Gqv_`*_h8yp#xJ~+!vGPg`L^5A8KpIU||0Me*
zpJ(-vSI)mJnCQ5FV=RsVT0mooUXj9=d}XU~A9BAG1C!lLG8vdKt}K>7=}T!|VAj&<
zhOy?@9Om+@66LuhXWn0yYP5y%AlFI(ofWnoiEI8{)*M0IqRP5M#I>z7ynhf8gI_G<
znqNvIo#&N0-LwKosU+=a-(4y3p36H&{Ry@N-AElm`_K#63{h6F&4A6wUI5#^<8<;2
zMsZt1niB<N;Pp&fit{kp4meZO>b^i8NZhh;P9qN(Z#{(6XyldEqGarW#Omc$S1V*5
zFy3B<@9OLWiPq*hT#pSohedbJ{fmr8Vglcq0eDtVB85oy8lYHwAdyY4eOQkty_vHt
z$E637EULfD6H<P_rVN}VRUWwTKl>T>DC<agq_27bf+p7<FljSSBXx&3clNBTJ&<OT
z<BQX>?uG0nRKs%s#<K}&dX<eQa4_XC4<D4a|2Tz#Q75RwE|9k{>|z4)K6ZsY?<C0S
zf-ajuFTGTmQ_shJZ<`etJ<PCezn>mBMdl;w*0xg<%Nit59!xaB%LaDTBM&C}o#vK_
z0;66)9rK!x(6I-N$-0VQ_w%5UWuu9FW*<x(*438yGLn1HIJ~;?xQst&q`vx`<R47D
zw%CVy<S%&UjR&OY@VK7layQ`VyF-<U(t{>$*z?&6TMpFs@w}0}saIy(d%sj8?|N?I
z7><rL;<S5@)E`V-n?N$x94e#SoE@$^(i7-e8&PhV!*w%2OtFiJ#OX8A4w}(x-Sc!J
z7w5=bPjDrJhk>5sNN8Ekfe)tT(=N)~IJzHIOP9-5ATZ10je#TTLF|`hJ2DS3yUUl2
zHfnS*%Yk9Ulc<BuR&Vsn#tF>EWXwM~j7sLqi4HrFvFw+9+0RPu%gJs{{eZdm`tdIt
zr}6gwisYkSg&Dr!0GZvT=pfErCX-)IVr7enlpG)E&ndGa<u98MfrRqSL1Pp3kW{1G
z^=jf=bJ%;hw5Uf;W$7$j3mkNLUUztN?3_`$L{_m-l&06Clu@GPB|)TgwY0ySgq=GI
zH7)DJ4_u#O<}TfUwX6Zj0|#|oA)8-LqO3MYw&m=5dA5N1#A++b7f?{kz<aat#97()
z&qa-0{*Y0Y`v?}qhZ08)>(G11sDzG)#F!`401us+mm8kJvYthscrKWG$XKl&p2r_b
zeB@S=HD5hN<sV8@HXe80mcm1cPk1ng;6+D<O)N?n|5h6SDS3k^<pCt+htg2j&88Z9
zRJ?3anNf{SXt3p4G&rLh;rc^~<g6+!ePSM^86dLm*b~lJ8vcn3Txkyb#6yLae;(uP
z*Y-B`3AuC}in{HilieZaO9biaIdR-Sz-F|*&)o<11G@>Vgk>uz$(_K!X`=^WbNRXb
zkPVOtY|g%t`fqOLA@#^td?I|7iE#8Q#x~f21^L)loE;qLGG9rwhH6Z;*{>v0Ozg3^
z54o>c13UK0c!2oGM<gHRbYf*4D|{vKu%e2HVqn80c~TAJC%vUZ_U<39mcEk2YA;)`
zvg2y^5|yu{Td5Rw#rau{CWwWGCM}x4=8d5-Cw1@F##5XgqF-FPA)M=x)+MGQjcC2X
z6-?71Cl8w~M+QwTM~CR>&Y2~45z_IRHnuntI#HHpRNLG0dKbTl5N}33HF@L;qaQ!8
zn^W=JiXU)g%=#*W_+j6SImg(J4r9D>USvJ-@UqKfWIZvC$a5GLjjkL0_cFEVd^5J5
z_-K#Pc{Q+ZJYVFMVs<@I{)5`O<kl|@NMwK|lxxJOeHu#hLq2{edaTF7dYWcz7(;(4
z{?kvG5bEZ1GBCyyY$;$&>3~Jgy>9ZU+GVM%o0#!*r5eT9F6$&ybIdCk_tXRPqiFZm
zqCt&XL<~O+Y8>Qo^_|~XPXoe@VbI|?I%e=It$<Zw_tK8k5t{4-4(dmpj&y@a=DOm@
z3imEOZ_89rn}JGLuJ!|>Ug~Z+dl-MP3ui-KcSe2MYzJ)fJcMYV&Ux6lrN($84=4Ui
zO>@a@&bvn+PJ?Zy-kpy<oCM?MPe|rrqbi4Hwm%OiPOERNT-4&I%uF1$Fj>k!oQS2~
zCUNH{g@=;_?VmBP@5QLDH}NtV$-_MTm6BsTM<BLx0N#r-+|ilacm~#bhB|D3tx<ng
z&p~ysj|9qEwCpYwWqk<y$XRJb9-eOWR=oLedZ9M6aLKn|>)|vyxJZH!ZO7u`?RaYL
zI8j(0Ut&;ogS;qb^<d55>3NBv_s-2A4cakN9rhiXMd#V_KAp7*U=Yo+6vMmi=woGW
z<8_fo65ALr^D^>C;;fq8y91BWD8d;IW!)o*EjWtjUL6I`BdKSo@8DuKDlm45VDN}f
z*9T<$kt7s{_DKGb#LOb+LMESrSJNl%>Qt-vND`DeHW-ta+Q6|U3#|hs>`yt$eUwSQ
z5~&?%C#v3{N5-uM2weWKEvb9a<ue$_;}gVFX07?6tOuqOFgRfD(}{Ul5QPMLyw(uN
z3?^~R%ubsny=M1l*PWn6?+K1w-A8N?efYAx^d3pml|7`<jr*np^GDLsk7jY`B3nT&
zG*i6amBAx6s94$19{4x@*zw5sXyPdTSGn_eG&RG^U+zyaR3CjbQFi&vf{Z<CD5<q+
z=F!ARPK=C)Y=FYu3#P=E>T$;~9w}@>CI6_gwz$aLD}_fBYt^usxvJ>7YAm)WliuJF
zrpr<^u&ig)a?~4!FsVH1+l43b&ryxC=D?{IWsB~3xgN-?s+S=f<$GU;^6Qc0M1E8n
zf&C$v-;B&+88u_eLFv4yyDw=+_WTT(bRzjN4OiNYB*qtPej`ceQKUyh$1v9Gd#1j-
zS|R+vmeXYsJ8T#lB|pk&*pB{jKu5hV@>QcSjsQ6Jj(jyyn5T3+50=rdCe{_%BA~zf
zs#ET5=A%sH@j?)#c=oGFBz6&<#iEtwzG^~~A`py!)nq~~K=WTs!tybMzu;hZ>7Kv%
zRU6lz_=HS4YOEw@PfICKQ%4TeDSNs7c-{;|ZuCnLZ$>>w)*Qrcd%s)vgt}6~rJg62
z>{<W1*HHJfjX;dGTbhwtP#vD9h%2HRthRT~4b3^cd^OE+deUKS)5SD;@ZUZxy+HE_
zrycIIqQcO@BP(s+3Bk%VBU@g1?yQdCK_G(BnQVJ8ROB3B9&?twrK9q(w4Y=#FuP9Q
zPmVrjOy-Qk26XH(<AUD%W&HD`UWIs(xs4`G=N?Pk*~2C6xM#r!-iznG${rn$k0s%_
z(e9dyk0t3b_gZhsq?g5-Q%X^mlm6nYlp{U2fHw}QJeH<^zk@#`)yGT@h=5lMY<AqR
zTj~yU>J&cO9!n%Qqi)+ztb4V~Z;?hc!N#3gv>B!SExqncTaTq#?fO+T2kmI-W9*bV
z-XElL_~gCoP0#2CIDi)|n9HU&yG;0y0EzZ#O18X?s-XstUHD&I&81FG?Lh48*>(bR
zLCtuTucg(DO~A<4jJXw!BRBfBL|(gAVl(8K1PQ!GGG9v_nN_86Hj;=4SCn5%)MaQc
zsp@F_YbFGTb&ma7V)pVXhR5DKcAk_fdIEVSG5IxPH}e4GV5d=X5CxrKqqzz7vQi0%
z8eE_nDV?)x(TjZ^%%kpLR*z!rA6?(gSv0EYja$Q0Iw`S&Gy@1U)5BbA2!vaVcHn|R
zCwm|#5|bAPQ~%~g8)lEu*7c|oqAd$X>8YWS-B#os;G{Y5hB(&9b~N<=%rGO%<3`b&
zP*0EeXPHy{=;KCt?uwKDv9ZS!x#yVlGmj?`rYEjjlC0O!9W+5l)odpBcw#-ihV&)%
z_~RxQ(;V?69{?v<KMUUW5{LET<0eV&Jay84S9f!z$CIp_TtrsN4vXUthZFzCPK8~L
z*O;2dB)}mZH`Da`<2Eocm{@x}O#qA2dcm^pn6T^BDXS69*ga>{tDXC>w1y38V5GMH
zx^tIw{8zoWl&-f21@QOW3pnP&XY=u-w;MiA)a_?K%5cqvsc}${`=n*{SUE^!fp6Z>
zo+{LmhLMQHPwhAwX#pcf18QUlKiWuy*H<PiC}WMpy~AjGGF}}#v;QhtPoOP@6uGE4
zjFd3m@MY~<$v2F5dNz=P|5`<P*GNq1J<i@>vSHGMy@8axJYMwgm9l@^<Hky$vWy<9
z+Au-W3oSdC;rP*EfAVUnJ1}l6Wz7K`))-*xjl{lP7AUt7IIE!($JS=FplYC&_m2Z7
z{c8{LQ{|6W^BaDhRJ-1W*~avuTNgjaz4B%>fkGBH-O_iop`dz6(9_nILskR5fwwrL
z&d_Dsf1iQu@<ieaM+c6QG7>+q@~Ex;=o3ccGb)E;PZ%i)79kn`T!q=$C(@MYQJxrN
zPbBv1Ere;?#{(`F*y_wA?{F_Z#QI<GZ=OFfQ99r$0+LTYVIrlj&Y2^hNaBJoN+~~K
zGBClG#p_y_wyJ+~G-ROpH&Zxv;ZG!qSkbNJT7cTeQ=4^fkl+N;h@PlPi}twbMR9?x
zIJcq`1c_I}p|)dKS=wmwtY(PG46358m&By;m@iB{&n&9L_su8LOr_^~rYa`$exSIp
z$fCIA#qB&d@J3Y2!mcw=z`v57W##KeiCI+TUr)Vc$-!mg*Au~%Ut^Jpy#^=3uP45r
zUbVml*{>%F-tVLC>&9v<8f5(I#s+R|n49ySV97;}3*H1WuD)N3j>)baGWqqy|3$87
z)%lc^0{&^Uk|+m0*PA0VheYI|V{CD%f%*<+&zh%DUuw*xx~JF?sM*O{l>EOD-gO5$
zGYpN~@IEXskeb6jtSH*8z=hZ@xN2>CIpV`e=Mq}0Y$CfsCUarO6>{`ULy%Y~?@h10
z@e)JOv0>(DV{x?QZ89Klx=se3*mxGaW&7(sZ|TNBzLD5;8q2nkZy4FH;4X1AGUhob
zkA1`VbLlbZH<ApgB7KG_gRROp3@kTVJl_6>F^$H1cgna!)s1)l8%anQ-7iYPi?N+R
z-(CDh5(@UbYD75c^){Zvd1t_zyh+NQ(92{F^xsIsUqjO90k;~ztm1S?0|?Z<kpzP6
zB?fp1d_4djVEJ>*3mS!Z-80k(ft#x!0xPC813?bhI(N3bEE{1$9<;w<BY_iMCu;3N
zM!Np{hpv#G7gwTuQ#OMVl~s%%`d-IP{AA0?P09JGGdPnN3=vd^klPL~f0@XW#ycEC
zsbr2kX{2TOHZ#-FCyfIf>e(-ic^>d^ipzE5lg8$aXBQ;<q+xv}flZ$@_7SZVGY$Wh
zTlT!SVVIy`dor;|A2dPP;**K8Y_i;k{G>Nbjhk<l(xnO*s6n!_H$BbTUWvRNm`Gyf
zQT4nz9C+2D*_mT!r0%F`@DN!`g_t`1XFVZFBPwV#hNgoihZWU~HX@r{m$bd2oz6Rf
z1u=j1FyDPL2^7|@W?j&WBF9cHD(0I3jfW!Aj|%vJHwDU8kefJhxc`ad?nQ>&b`a>}
zH;X)F1Iu2Ms-57-Q;EWQNuh>QPdSrcVDELmj6G#c!b0PI$vl-19q!~M`&8OdY0q_d
zAn7@80e<kp9Vs4v%9yXN5AzNU%NLFdQH=O+%(TU)e2L?|l$T;Na*ve!TerAUc0fni
zgIByVswP+ESD!NBWBp+ft9cJVfYZCw18WqD7O`c`3*+ya?cKTqC04dHhQI56M4Fyp
z7pUXTv4GPU-*|>cpYM26h@#FP-KWxIgC>&6Gw?u1E1N^QzHEn!zC)Q<Im1e_6+F;~
zQSAK(p3BuNN6T$b(4F3^BHuLbaG24Z&B!-XgS^h*ed0HbS%;Tc8^*qA<h=UdCG$<A
zjr&{4dJ$S>+w#rCdXAe5%O_<#z@ZyFFL_U5xNPU+!Z&>oKZ)6*L&4%3TeqmhX&hS(
zM^F_l4A-*PS>Q}j86xxvseUu9y%Q@M{jBEDxc|RL>i%sj`kGhfs3Gg#2WW8FHPUcM
z6L`4r#`t*$s@9g*-9^jNj+_OyeI2jvo7Ju-(EXt*?s>Tx4^q7u6hYv5svvzw;lgc;
zvgNHFJu8Fgv0gCB_K+BA&?HYAjTTR{J$c%CiBk4Ols<RKsAny)D2+XByg9gEGEXNK
zD684%W#b1PEa?qI&a0_6{qd)bk&JkzguMTLh|O2w=`=2F)h4s}bXsKb3L}$V4=SCz
zQc9k+@w7Q}C_5%t#8jR(+4)$RqiWSrr$WJk7S&Q)+%64qu(@?Dki{fWN8Y*@TzKbE
zX*``obI%+PmzsgmQI@DyR78X{UP;?a@5J|)uCX0Q#)1uWqbBwP(sR_pa(mP9vDcH*
z_vV*4^>0NNRLjhW@~o?_x=OYk;2Z~9NF~igstNQDBhAE*dGnJiqfKKdn=ct_rY#7Q
zoq8??(M;2)$CsHNvrS_URJ*Pyxn}A<+cDO+W?Ht~O#stmJ{nfj|3WkIaGw2w8R-Xv
z6(%8>bVT;3m-J}pFng1zh`^lG6Mm(chF1lbt>~$}AyXbdB(<i=9$xh74$(r^1Vq%R
z>(MakWArCMk_lFN;5EI^=3=(xu!$qUsZ81q`|u&@1PtzgxdHZ!{TM_=uk}5|W)#DW
z1Izr#oM4XM3Yslzyb&F(14o$`4zleaue(X)8Dr5|{F9G7lc>030Y@Ah)<&O6#F$uB
zVvIeL*uAufqt|DQeV7Tdn|#Ix<-FvcNxb_QdrKMjLK^)^^1}*CI4Lg#sOte-jMO0(
z(aC2_j?9OzVP1MB1i)XrVb2{k&}T^{Qu}$%TMZP<KSk}Cw9ZcP^iEa8NNzo9U{5D&
z-XhY6-K}2Qtv{2*b?W4?;rsc9_i^VAX*$tPFy*xTM})TRKYoG`Zk?c{I?G{CCtWXg
z=Ct$zIaO`QW{?;mSmte^<K<ClWh<~$3y8rpHbzvx$;Q&NFg?(@L!M2XIfA&4Jevm7
z1aB*zO-!U7I=iuF(<oI39z55|JexSCmJ5=7)(0)NSGi{s-R%XHTCI(H?zHYlo;Cj4
z3{mj*S(5baX3-0+vH_eN7Qz#wlmZ<3FMk$W4oqj!Uj`y*A(%+L>Y|zl)FKNt_Po@i
zPkQ#o@!qpZbZsj|<#k6$*CA{r8j-K|%4)9Dbl|FTkXAs$1qZ5xw&(F=lt6AlEkK>`
zy(ZlNwJSvti0WpuDw|&2oNe}lFE_B8+VYg{6`8dLQBN;x=VUv|>Ooq*m1s|tPu4eH
za*o7L8XS?WE$8-cC1xs6<Z<j<iJ2!=A)Wb_$;HqaO7>fcOw0*$IFgdww~U9F$iZ4K
z%OGGuki6$05ld2toSR@{G1|n(y=p=yy(k9;c<kS&FlvzUx02+k_fYIjzh#reo=K^C
zDWWd0feDPcHeQtarAlYieZiVnnqeCg1(#HV&=_{@W!)r5^IImhnC?j{AbiYBD%-<?
z2m+gujyJuJSJ~2yejP_A)(iNYs<`U=R+<8m@S}M7mJKW>CAzca33~w`0|%>SDY6}a
zma$b6`F8LKxki{rzMa@`0HYpt3px7jpp@PJjeXmCbLHh9lFYXgLz$B+lDz~B)ffv_
z?%Rga%A#IMjYr{QdUnZwJIxB}cd+nnBmaTjQjF$MnV|5PjB=Or>ZRn3ALb^t?7$I;
zBd{-;=3zV*zEbsq>O{2`4HHk6+0}FAdDgTRZSi1^31r=~F2m5e;mx|&3u#6{W@O}|
z_3bnOhKFP}I?)PG2-?#5w(lR_z=X2vn4)@P<J|KM)*Nj2qi+-Rv>y<-RX}q#ET~)Y
zfkR-=283<zYkJp$=H<CGrPA?{=Mqos&Cuv`iF>;^$f16DE-@Ti2lbnqdCn-m_n>5-
zGsZ6J{rhuiSu%};d-WXme5}1BA5f07fTJlXJm>T60&1>eH2E>ExS3`HJwmRRo=cR|
zwalCXl%F$6*vp;-(pCb0l%MQzJQFTYAg;9lg@1^;w|T>k1AAmG+C+tkz1_OwiUdo;
zi=Ec--1P57^|an{$QXfe%^__k1A!4c(KoxnI3?Zqfw@p!toIzs#oN`=?YU%uqqDo&
zmZK!>JG5yGP6K})RLCGwJ5_rYfYp{;?~9&K%;FpkRY#sr%);Dx8rdCv-pGvViF^0w
zja?g$Gm}1V0x)G)uG#040Gw6>)|^)%Xc#wq0gtn&9)I(m$vX6!6fV`%Kmo<)O)jP}
zh>47f+M|~O2&RN;gj04PJ5BxRNX=l&Uv;ET&Q%cLTEKwEWvM@(W*N2nKf=Vl_I#2H
z!qK_=rgg_)c3Bz@bqRCr=JSc(!#7ZFL+nNOAC|Va%2fqZS0`GvlXEwcga+e;_`Jy-
z2pmI7H=j=;4<mgm#w!f3aj>F0GnnZ$5JtQ=%i#Gm3A=Jw?>V-IOnK&n$aj25;80CQ
zzGDnhV;%G`-$@L@`wk;=SPnUq%y$yQ^;J0fs%(_i;2M(qP7;vQ+=-!q`HrtFC(vhm
z6X3S1AO%k_vvQeR3>fq3acc5A#@M-4DS6ANYcZX0NE@%Bg85FOgoj};3Rl0ArfbZ=
zrRJsQ*wEGkWlVh3TlSjg;b>IWqY^jb((u%9#%Kms_~#`g*LM;z)nWzPAvOT*sD6}*
ztrHC)x!kBpH-50#zAs8INH8%Wn_iinQJd+w?_EV_4Ht(5*EekguX56Eg|<CGIA*q*
z(lYk(CJ46)En}bFd1C|KGMZybCS$F{4!xqpeBbNrS7Xeqe=|#CJJCuUC!jOKFsk9b
z3E9s_T5k=wuZ{FU^X^(~CE6cB88YeZZ#;8KN{-i&vOZoYM=3SGuC$UQVXCD!yj9Qk
z$b(Yznrhj>dC^f_IKwT>ntz)ndEE;!@vw?E0wC{Q^xWF?Tm<gslC-><ni5Dm5ZT39
zpwqIUZa{nz>qhqup%Wu~LX<y%_-wXJJg?14Kd81#od<4td%TC^Ofd*-SS8dFVY`*)
z4|{lO=Z+VQ>dUJY)3=PgkV?>J4nA~t^aUfVt)wW{UNDZU38-W|ht5s&urBM6Na>kL
zaxbLW%fpqz(H9Kqk6$MF$YUD8nwGmI2WyjY@rA?{g0U!$lP{PsRoL*AB2QmBU?fov
z916%*<pq<$T`Z|brRoU~I>KNaHvS>0zmVh+Uv5x#&C^3q#H@-Ak&UQcWH%f!6k@ka
z({pUQb6#2w9;f5Hz-R{=2WNQ~6<DIpm}586^|BjHy>#sbp^mEur_Dff?yU49QDA=>
zKzGx(LEvpS#<K00@q~zR_@ecD87o0f884;|F3##oJo=(BkOPD2r^mdgKJ4UXB8u|7
zjcGRfqOs^BJf?XuF@YdHYPdN5qA_6=XRmqx<Qhf_{;57ViVk=c<59+60EWgKV@h6b
z2K&6SqXSzsgi|HjnO<C3;bd5K6s|icwHK2Fxi=u5k~)+*{QRxGm}ZN^stR6rpz2x|
zId4Q0`xlQ$Gm7p~x3#T6O>b0D+q`HJJvVjXHtF~`*VJE;uJ`Al`F7|9E_Npe9XB19
z8?k2YM{|4Bj@x12o9Z4J4B_303BvY}KBm%;mlFENG0~KfmwXviqt(%ujL@fG`Ivv6
z#Aqh-k}=WZaB;+cDY1Bka~fxeNNcpWc^Qv1rkAAip0*2D4%||3Xf`z1#k`aTJ*ImE
zYUj<*WZ=h2!yNjO$rt5!PD$CZ+IS1@y<{w6&QI^hs{vWB;%lB7s?O>Glx;Rn+Z@gf
zdk)LGf9_IuP%oJXq8M*_ar9Q^B)1i{XDX}*?UzhI)l-~1yXyE4;5f6uT#hkJpkn(`
zN0nEudxo#eHY1-HIns9)TzGinIoXOFUd+qDYpQMG@Tz$`$P0ClK6%;5!$w8-qc0n$
zD(=y2)Uy_GKhNa(@+G#dGPPzr6O?bhLb6_u1bc1Cy=-LI1211TQL*3#h)4cq6OUb#
zDR>Swcm;(4=*x+X7P@XuhvCa<2!R??DaZRj!QxFxIa+0%k;=;^hpeDSq#8LD93<5O
zxV}xXeR18fpXN4ZRo1+wnuW@`<9={j8vb<{joZs<>teeAsD|Ifz~6RH?Lbx^jOguU
zCs4p_K)Rm7qq6F;dY(d0txLpabc@XcvsK@5Q+s&!GSN=&i`+O49I2&g+4e6Ec=C!7
zN<Dt*eLni^kyp|NkL|@g7mqeJUr7wyX)f%>UP&Y$Iy7cVeI<=-jYcBbNG`3iInKRe
z!U4-}ICxmbUoj?sbdTh{eSSyWEyzaUl{Dmf-;rWeo9BQt87V!llwR?<Lro3J(b{f{
zH{dW1+Z<6HmRmsETZ`t<L1A0Y1-%zNq)q~5EgHZ{j;z1p6A~JehQr6fu*ENLdL?w(
z$Xd(u)K*p%Q*Ezx<^T0B(s8O0rj-j{ompQ=YYS#IYVO-}ILqe5e>3_p%_Wueokkz#
zS?McDgzUAkJ^3GaO*J~<s2DYS4SfvtY6w4rac-8O`Kt9~2J?Z@SB<>bb+A((do>}X
zM@kUNyqfsz7y8)%1;uTV&An<=M;Xb+CQ9pr3+^^vHMZjAz3^&c_u;}GDZZLS<Cwm6
znT$*|=qh=onX^)kV%Ux|CcP!3ko~C_I!aK5SB)54Yusx_!>aSHd!G=Z35+bfuNzOp
zK!>nLgI71aQF!r?D&3-~-lH4<T9KU4LDi0a;T(ihOvgd);ju<HN*+VCAw6&TV{=D2
zw{ALWps}mWqauQW!$n6y_o1>iEUX59GKe-ftWj{=3!YwNb0M!8yHsU4@|qD+1&CAl
zsHbS*ay;vfy=JU1C)=6V5{dB<h2zB6l3lmGK@0Clxz~KS4~_m<oyYwbK&tP4<Kk;+
zY^PfXDZG}%;(+<OD@JQLs$#Zvd{;JJl2YX0N}aoV&(W2&61Z1*dtZGmNgnn^s<5cN
zX0o#z%}r#LIT8b8&4I86$+~|oap}{DVne?d%|P<7f|Qn%=i(zAjoYtXQVggqnjP7t
z6CjAOronl-UIjO@xKWHulz}tEZP<*iF>9|%-%;TejBLG@rmPF+W#GS~g^_Ll)E7X~
zPW@H$5BAg}ZR4>TJ&pRO`nMbN&ne!)W&G2;wY3xZHS!0h7Hy-|uKh9|IRwhXf*|iX
z69=WxHo-xQ_9i2w80GB!q)bMdD^3>VlGnr8UpaDSEM7y!;UFIcdRmR#Ln!HaF40b6
z*IQGowYEtlu`{lbwa8+L;B*l+51QMGhG*|MN@x@SH64MguaH*Th^ZGXTxbN4{raY*
z<23o#tFDr+e;%(8($i67X;q8l&8Sjb;@v>MohCzdS%STWY<W}Xu9ZRHgi?W-d^^(a
zOF>?@?y#-GWX->K?A*UcMqf|8#_JjnN?%Xx!6`bERmO|jZB6!dV>4<i$$2pbGM@GD
zz&rlBamZBiaL!Y<5Wg;k*AvzRrZG<|#b~p)hGucn;jW?<D!m@k_o$R3F*zv}M`Uq<
zlUUUuriGp{QF}d&a}9EDuMq1Fbz$bXtVOOrm4I;T(NFfu9HJVpCz;mkH{`PE?XeHR
z<AQ%Pono@)-w=musi-uJ?$PzkX*G%OMYnyw&uPQSGSo=+y=<8qN7)L3@cJ0<rGXr0
zGH}`UipcFwUS`M}X+7G3m%WiU68~<{Tl&#Aj1ZjOlh$0u-Y~Xpyv33jRcFC4`$pn5
ztY>483sCHOI4nl?(2g{+?cSl_ErAp#R5$1-zL7YsA;I*XVe$>1)>rlT_=b)2TW*o^
z8;Q$1DpCostY^1N)k|HsUus^L2L4Ox-UkabehICa72dn8doi^Jlg6byrho1i10nR_
z;KO#+0jVCe?VnuPNXHRkM~s578wlx}YU#a^Btx%?8Qz=T;0a}N-y0aN;B9;gY&rh;
z=@<kYoaW<)&oi?Br^uT=ovF3<+cNTIVjyocIXS**+}g=pI`*a!4B^kbX=L2J<M1BI
zdI5C-oAb}Bue@5u-%Pw@;$R%-9je|D=?!J!&D4ouES{p{0C6WcGE92uT|7ZkEA)~#
zhr%7*%$sRs=`Dr5ZKy<DHG-=;me=i*THpY+Ik48Fs@ioRBDgD!BGYG-9qW-+1v~yT
z-ZaT@8^UH};)e#bTG8%JOmOX>+etNx?|9+eDIiDJX)~j4Fufks)yvpT$9Gpz`okg%
zr)2BRv@J!4Xc0ZIw}}R{dD(W<7%9v^@>UvyTnfv`TgIpr;v)E0_9_{B%lLR~=2Mb+
zEAeJ8K{VLVyp=f4QRcklB8im@CQ<=<fy@8=TZ#PKPa%D@U5Ek)n7E3UQmko{(bO_3
z0><21Mo;cQ(6*KxIaHjLs5F6|wfa^Pj$LKK%HU4%EfcqM|8>j9rSAPe5h`n5|C+01
z-M=OlrSVo8^0>)mYs85^x`Tg6b}pBiqI2!Y#P)_YFfs`&C0%do!dn~+9RpsM;j1Rl
zBgT|I|Ll9Y2@NFkR&t)C8))p!-Cf2YsKyoh@R^xI#4R$hfZk5r+cAq@_DJd8dh~6h
z7(*R#9DCb11tslpn|a&la#}4vvQcX%N}k-?i46MWOdXt!N50x+<|B2AlUu>tVDMdi
zwG=&zUGaIQkIBgTZ~rMJZ=Q0TcHrOJCT1s4*$k0HqVfI}RSq7KS|kcM%82^RWb8d1
zeRY~I>;8@2mg7VqhchE-4q+2Bob#X;JD?hYc0fZdcAs>H5Yz&-`?d-DQ7k5U{_~V>
zlQ;e6GQs@%Z>LTC&}|92Cbh2I3ZTmsL~-Crt{?F#i08|<lk{5@PkV}eJ`J8d5L!mg
z8!LH+qI$>C^GT8!Eog~k?0k%C4LYsL9m$+GGVa(R*??s(+j8d<6L%>SIYW=1_u+#=
z4Rb}WSHgL)|IDF)HKynoz$Bd|CeJ600~pMreAc_N(s^G6F=3Fhf1mp#W2<sLb<ei_
zs)MdZ57^r4>&4pnG<g7!>8tL|;g?^jvF4ayMXr;ngFd7t`3?U*qXwDH=%z+Im)3b>
zma3!Mk#k~!m(-o07Hw%D96d%v_i!^Sy})Kh7k$|b09;1%{4=oUq^tM3TbCf}i}K|%
z5C@JL8z?*{ME%1mq4M3tfupNXlMTekcMaK%Cs`T4o2Z42Hx^_NV=T&}PG`f)4(GdR
zXdhxa%zoEceG8WQ-!&Gn%AAn#?;2O(*o|maqrwbpd%=q#d)y{|H%Xd?Es;qt$cFbk
z>W|V8-W;uzzni4)I8R#Wb;UD@bjHIN|Hh(ePHNsTPG69edVpui;ab=`y2v0{_bTNX
z*3gE3vpwT2g;&=*aW*oZfz2Xn)FEmF9$-V!@f7Rr#!I|M3*^|?s=t(8w0l}TNNzfu
zjVExL)*s@4naNgA7LhThgadC~j~(ntzni3T@dyvNMLI_Ip);Y8j?25c4<7CK5VDx>
zW1U3OoqA)R=_JBk{DRKilI<8+W<m~@X!&wc9-AF>G~Nn~(-#H#sI8hJB8e5^2QFHb
zo5iTEkriaJW72Tlq?96qI6N>g;NrboDvmL&5cs6(UEz`%^Eby1=S9R;<2tH-sXeu`
zKpF0EZme*C<%kiD9G#<fV+cu|&o-k$oEb{X;nLvTf?DLZqawC~r=;WAYWzw!@0V`J
zhQ!%-?vS2;ZajktV*qiV*!^hY2qL-_xyvi7vqTl`q<N8%lvc>-H}A*gd&a^k4V5(V
zy)-H-t4rz$T}Ho`__$YHWsH3<4XA_Y!@rl9%8OiG<g@X^DwdASO}X!xuwY%N*EQpw
zBIx1g(O%`fUDOO%r$)OPL{$k~{9c+}-FlDx`WcxVqN=Z#OW!jVPvbB%T4gRf0{tt2
z<OaUpyaV88UaV{1OX7?%92OmQ$BDzADiYT`1%GsnS^r*|hK4@RIXuQs9QE9!qZvP$
zy6XjmyLAa8gqbz59Y5e9R`(2@AZulj$+jC9AFN0(h@qa>R+(isJ$u!jRrR9>db5rf
z3?~ZeYAzie|AX#bcKdsY^=2N<B*5Lr_f4c&9%SVE#vHtKs1u#h?<Zz*zETUnvG_r+
z6p1I6@v5kyB>VkDMdo>iRqp$VQflD2gy(h{|9%oBz4~2ZE1Lg)5~XF_FG#_gXF6Zu
z>{a}}Ng*AE&7K1n4kL1<fQ_P0uZr+3=FNksUI4r-mGAq^W4qF7G{fm0&8q{`3saWC
zy7yrUGn6&2q_$E_#QytnF6$fLH^Dl1K$`xw2bU&ZmeqmL-gZ=3gE&q`W9GY{8+1`0
zx6HuN6IF<921U7_))-9v_<;?-&O2LC-@F6KAku}I1q5N+@qjXfostb>CGQ+%WFwK5
z&Jbm9Bk_{sJ_k7&+erLjyT$U0{YGZPXvS+M2%6nU{5);W5_1mEkf7}N#wA|b3qi?y
z1@3Eo&s~TH$afRz;)Ze5w2qU3kX>MK3*wm<??&><8zzVQQCKk*R05J(x7kajYG7C4
zS`MXa6w$c>b3Cq^H_~*$)XB?evn*P3z)-7*xUYMksA4DIMv!btXR2mkxVqr0?uLm$
z=|80%&8eRP)$mCtDl=5xmadmQ_X+8FzovDIvl(c!C&$2{A8GF7@GQ9H98hL*1a%O*
zcFVSdV1a*4<ekLV({r;-&+izeEMDTAj7BD|axmO^Cr!<}cS^=f0(%#Vlk7Xg!Nr1<
z^Y1AY_3?L%?aU5t>X3gY$;~oPap`BlVPYajqWF%<+r|@U5xgQV#dJtAMja@pKwb*G
z<7@9;sRV+(2bmSCUVjmb?^@(Ax15%GBoAw-x;2M}M(l{JdrNc#X<+a*9Qwa~NSclt
z0>x_33JAPtW{AF%1Xia|jmgvrg4~2Gbcg+7@<Hw6cpYLE+Kkp%JPCQ&k6M!l_RChJ
zc$+1IC_6pFNk_IF%sgf<av?Q8b@UWw3vlj&(e5V5IC{aD!L-YR#S3ZNK}V*@%mrgG
zy;TRi*$auI7oLz@Bo0o<cqDGY*R8|m_DKpClDrgjw^Q^S-cieBKwhSwrDzvL7}8gc
z+~e3&SG*b)C{hi87Q~|#?fxGhk$U9e%&;YE-meAo6dv#vxDsnbN&_}Ez1%_lZe2(M
zcf|D(Z7*tfWbAJ*(UDQ7SMuEpX(H2c3~zD0c_E!Q&6A&cZ84v2n+}Snk7^6s4;=JC
z|5gCz0h2xG7<gx3ke%bU9m)N?0}%Ov&$Ecx$PW@@Esh4u+~^OCn-?B^=PnuhLE;w=
zeKMC4Je2#vKcS7|cf=s~gVgmY%zkvij30VW=p-ijAJ{<JF)0Oagkw6E$bo{64?PS{
z{va_On(~-Gk~n55OWEsUI)gGO>gqGM8C+;pqwt9t3}kCRNb-n<-B~WRevoz$b|J@M
z9mSplB$Ox5!2{qCwlNwHv*E<rcop@67c{=L{O6e!qJsJ7oIdvWfr&d7)A~faJFNN^
z>!x@8zy4h|gDN;O&`aTd`~Vf%i_2DYfk&g9IUUMbHC!Enf13@?eIh?hV{0D8%*YQD
zudcX4MkDKhUN8SJb(&41y0Zal=7+|M8?Y$<VG@Di;v=~Lb=k*dJV3EBO5Rhmb4N7*
zQ{jguNUM6{^bDed=m~E!DovnfEk$Z4t4R5Wi6yA&b$#^K_7O;9N~)2m>*^7y{m`WG
zI`uSI|6vjflj1dp%4^Iy92>6>3#x~OhJUlin~|nNM*)e!U@IW(Q_mdjNa6#ItxqI(
zv*(kpQ|LoiNzc2fI?c_0!cX;2eQ#<;m04Rswu3BxUPo`>wxj#AJfdL4595Aab=5~?
zq-*`t=QuLjO@!HtP6SK2ZrTD2g<Ee3Gf|!A)ROh;xOPO3*G=S^&|B{DC^*5@c|KCp
z^t9mBPOGJOv1@{`XTMB30G{%gZ{m_y8HQjfcasdMf{VB2OhJ`y8u1!-j9qlKd*Od|
zM_HFJK5YSkYtc7bh^6ir{O1R?$7Id(_7bVj8eWgGbc91#qnqZMDAMdEVgJvMNGm$N
zQ$3Niqq@!~oJ^uRqsgAEOV?XpS|vnGFKVAw@06R~Yxc8swMg!oj&@niosV6Lb-m@y
zySvMQ1KW58-^bftdk^+#Z2RR$X=JGZn%WqT{K(j?bB~PvD0PeV9Q_`<*0CR@Ze>f7
z`H=|<?sxQ=bM{Ax)^5fkxgXisn_@(b|0oI0?2LK_&wEW4-4ALFDMar4hWc49UTWkD
znf#Fn4SQaWpl75Mt@tgNlpW!XCjndeQ4)dS!!>M4s}4+Elb~eq+Fb1o9b(l3oKsNN
zym+Oo_kHWb4!TpthGPQKC_LKIbgWr7m`PbXT7h-x<I;}AK@{;$h=V}vI=1LvvFY)j
zr<pc4qxkn*(*Kdk`+<Y971-faNy$G5x-H@gbvv@j+jt5^(T@{5x&E=+q>&$|AqSUQ
zG8)O9_~ezb9~+0%U1{dWiPxOyk1>FjCF{9dJxK1y#&+}(OgrNN1S&D%dLCrKHnH$y
zV>Az|5Isbw_+t}_;Uf!{%1JLhwLrLgDMi(fd|1j3)<OwGl1j94Ld`|0KQ^LNm~ek=
z5<0DOP2IoG>R4ZMbXQ^XdenuBXb`;A#dYOcWa(QiX+^U42)S)9X`xp-KX&Q9(`{Wn
zMK5n+<KOeP76>4?IqZ;~lKYpyo#)1M%k$<~8Br8CDS$9_%p6>7#}68$TNMRAv7S-l
zUu5JbiIwi|6%DQDnyzl_C&mwTPs*|KCuzRsqFk~+NyE_eFI+tT#Fs$AgN*+qjSr$+
zq7GA$%6s<AJa=CXNDY!uz!d|KK2C$Gla9b{4TV_xiOH4P(&~$ra+KZY<`@-6kgn|f
z?^5-`B#x_MvgR$S7nJ%>lGG`(+#+LH=dG}<LgyY})k+-cX+&2wgj3UtBCA*t{KVt~
zuj*|Fe*^obP88p%8lS*v7aWY1ncEhY)&6xeO7EJOkp53%%7Rj?ozV#mLn{N%Mnmjj
z;kScd^u(_4<)>-r;mjS?&rcJV^wr_$PZK9>OF3(+X;hfjcPW{l8UlLt&cXB*$^O&_
zE6RqP!$MW6yU+2TCQ&(}FDv;a|I;)}-+Y-Aeri&M@fU7wy&}6my&RwP78vk=VKpdu
zNwe)LJ2FYNRQaiq_sVOe8dw(f#9eb3@b+WscS-8eO^xNV7P-sl7@1IinuN3WIFsgi
zX#_M}f5Fj6!WDG4QJn4MUCdwY$UkU00Y{jrgqby{u!GQYT!2Z>aXy0jHsGLM)(7By
z&pLpO&#+@x$-sdf&_^QM0Zbi9=rEC=CCrcE<dGjner6Qe#cMkMI(LnX{me*xK&`$q
z{_DTuz3*ox2uanS3rHuY*ie*Z{AVT^SWlz*mV9JizsLy{8|I%S_HuMK8?55bOcIuu
z7AL($b$}ox|9NhfAn)Gz4s0X+M_@}eAWx&EMoelyOJc(HN!_u4k@GycT8pyS@o9XM
z^`E8TpvO0^HHUC9wQEMr6P)u~UeKO^s7I}~W4H^ok#w9A2o*Omu*|dO)VQ<f@Q&*^
zOr_kdOy2<>N?Nq7Kn=$!`dobw-IzEm+dqqewRqCP$?&lC=ZSX&%T#NTk@#T;a|doc
zMt^S1JTfa|KTo6fz{ZnYXGZdvzNXIp+!<^!Y=55SE_KdLf0$n#%n4!{{XCIt#aLEw
zI5X-d4*4nu0?Wf-wm(lhX*-9%W#h9HK=znlqH?75bvS3spQkBhnvqxyRA~bzN0G_>
zveYBBfP0m-D7t)d<4IZfUX?dqWrmG{x?OJmJW<oU;At3UX$2N;jokLM&ig1fXQUIr
zc|ye@#c6Woe@gG?CbrW|%$re$O($m*&p9|dD_cQTCby+U_5M5vl=g8t*>+@4XtWvm
zg|UOXJWREIk%YPTyo~<B$geg<whbBkg;7eU9Ok`@rx5j3UgKvSh;D!-_lwluOIzu2
zhoy6*dG5<c8Gkgd7wv^#q%m3^UTYM8k!0?u+COsO{Dt)%S0YmSMUpXlWL)+fy{xJP
zX2(#VRlQ3aPfu}nt3{<*eSTSw7IE0mB@<o*9L$lk#BCgQm!$)vuf{J-xUarOn*LQ@
zm6mrJMHw%%qC<LZFpr9dEo-39Kz;~)f7hv?(*t`d-k<*>Nx$~aHUXPagbNBZT>a=w
zfz@}*J98a8;q2+nYhVc3_Kpn?{(^Gpf8>OwI!=`AJ+sEWX>R89F27Au&GV=i!TZ&5
z_OQJ)()6;lBAH$yJZC9v)g{|AAyLZ+$@Pp08?O?+xR+SAY&U6Kmc%cM)3`}PD_ZCo
z<xcT(WC%n98%{=x$51M>7wV;9$dgYg_l&t3K~$|1Dh_mLP5Cwo$Z9VQbr&h4UCnDj
z-aQt`!SP(wJCkUwNbnrIemEv5w9x!SUp1%_E@zsK@MXNDjLzuHt3kWky~Hs!4q^Lz
zLOOxDhMDYo>WE)$_X15;K=q5RI+@M*QTJ=ZGhsgfS2R|no+7s#lyL~P?XWla%h-18
z2?{wSa?zNfk3L2&I*-&cVAOv;g^-P1H2R-qHNBW<%-yzRFQx&s_aq$7c?~WHW!yi{
z;NUucF^y$?q=_Xx6HOs1va;eC_hOO(>kb7bz5ahM?vYY7Nt{?IUrf}qSHf>Ht}0%3
z@f2oy7n2aOb<@fBp8+4yIIs1KX?i-2w*6u<GCwrKLOe9=UH2mD-(f}?QJx8_8C@B^
zLWLDAZ+$0j*W1ynf<AZYc#h6z-RQ%CspHZMED80%o3+h<s)xNtKd@rrohC{);)an)
zfG9HK4x&m?{Wfkpmj1*}erZG~dQ?SBzkZpzd<d@o(io`bGmPM|UmCyBDWNa=rHR2w
zHT(Ky8oRpd;;4}Ga#O^nk#Psl5YfXe!_^^jbAKx*x56)zNZ3KS=$UGg12HE(Mb!zm
zG9_>D;B_}h`Im{SOaiOyx-l>DT&$U->YrCLbJn796bYE-)V*)&kCq|&OB2)5wXz=Y
zbQxEZ0%<sq6ZD|@%cM5nIfuPY+yS*57=W2D+FpJNC0NIQ=Nd|teF1ubrPkc<_M%%A
zB5Z92xqA;u-|@vNlk1{cUX(Z?@!i7Y>)#jGk>PCzQJ7Jm0wVAF^!NtHm3PyC(=~te
z-6Y@W)ff$9k*lxAGVdBUSu%-z^==wcXyNAd%$9rCI5UUJPy06>RhQNMs#h%>=B}p@
zjT~A$!-7_HP($-ywF{bbQ1DR=T&47`(Y-h!<;Ykj%BNHuD!UQX?E!TkyJt?HQMC66
z1gO88BtV}`ah78teb?mn@SOV0Uw4umM)n(C6Ztj`a?^{=t!zAr9>Eh}R=P&o?<O%u
zs#Jg9aU9IDN)5NV@207#!qJ6;FRk{1E<3po+H`nq)}U|S6H_=3;vD*}EgzV7a6_0b
z0|!njoz&HGfCYTp!KrBux;gokag&jvW(Fg_vJN5_x|sed4IDN&i!%1BL=TPpsm~N7
z<CW=~s?f5(N}Xd8Q4b%vUzyOXu)iOVl30RD-kWh-*+OLPhY7`B4d;70J{dVg@{`i9
zoToQR`BzEicAP|ACl#-ASRJrb1L@vl4E~z8G)EvjW?IkNaTN0srl_?5!!111&HAsB
zfO4EdE6kI<=#g6{N;9Cd^?6%b4*i6QZ#%GE=D}shv7IPL*Qqddd{KJ-iK0_BJ^vJ8
zJpISgC$h2?Aa*mf2GJPdv0stxKzCw(4s_+$#w(UDwoN0yHsb91sEq#Fh|Y0F#(td`
zQaQP*)+w1^C&Hne;Lx2Nf|%iEMRHLZrNW!gZrmYZwIKPZcpWx*T%LH=F*J2Whr!{Q
z(Led?L_&-+?3Sw(HL+s^>+-MD?6I3IdnIy@ptM6J`|BifvqVpkn!{!rsT#dsn`Cfd
z0_*(SuGrQC#`U<moRLQ0pj*~ede-zdS3GKT%Yh=ay7lpn5N1a@{<*k%TDk!Oe~K1E
zu@`6;*;Z@@sU~<i$}sMGwjFKTvK84x(o-k@gZSYPZzs6r-VSgd#k5J@GnO4w=SuG-
z0<c81W%NB`>&-mWdoNKIr#S4xnOomW%p`~q?<KQQVg)CnYU7i8&qp(}9nVSvGKvn4
zq3<Qxa9`9?crR@}A)Ch@LDTo1(VPu88=&`WYjWuk_|7xe@n#`fg@4RDSb%vQtROhG
z__w{&uLc4v6e!%^OM<`-OzK{6;~99e7MSSinHOZL!(9)2O+WOCG~P?oksSy1imd5C
zZ2!@U^7hOe>pmx!KE%Rsq%55vAN3LOZd73G=}xWpUYd$E=9c^Y-i+Qd?;O)-*8S)e
z7v5}do#Y7DL7XW^6RZrpo86AGd-b`I{3i7P<JOjGu||Gl{65SL%Wo2S4lu~Zeq(%4
zZL4HF2l+tBA=xOWwhp=9Bq_lwEqnPhGX9&P45$_&|C_{P;;C?7Z6yi;fgrn53<w<g
zX4Kes(%YV5(v;F~lFTjWzCg;JcVdMxQ1Nf`ZBvbMOT=_rl-h5Sv>j!83EApiK<61<
z#bwP4AlSO)1=V%O8$w>UM;ZZc8HX6y2YIGn)wcp;;^)}mrTrTlPaI676PXo#i3(f0
z-dzvCAU%hygAE3}BAY>sd1m1Lka+4TTDBsKy|fG*uxbs^$h@gODk5KFUdU!*E)mb>
zP9A5!F|uiBZM=lWely8ddRB`y=VluE>S0VWURrgt*-hiVx$ewGO-xLNcbVgx#$#yp
z@efIUGm+Y^_;54gwUIcEHIb=Nq;-k)Ru-nFrL<|XSH|(3|9C(TsR8;FyIwCbu}f;v
zI}QZQ2rKB3qU6R`Wo<Lbi7usZGs?4TSvI?|hBw08B~7oo%A$m3*MHxKXNz`Fgcq`$
zUN%j*Z7KF6bT<kR+m&mxX!U@(qTh_PPHC_+dapjOk-{w}jb3WWz`xD!?RFHj$)+vx
zTca@BCrXa|HVsX@{K)8U6TK(wrd7s%YiwR+0nS9;JiaxNn(S{AkC*j4miw)deBVwP
z_XcjlAb>aUzfH2^E0}t0iZ(e(OYyfR5WH#O+AKi!o<M+0QBbcI%D*)}xK$}y@`~ej
zf<>q51-U((!6i)13z~{ir<J<nGo??0W$jWE4Vqa0ZJK2Cow&LIYB)ajNVge%o|%RD
zY=FI!SKln{sCag0gxvAG;tuIX8#k%mujd&PyJgccVM)VixgWjboFiL~G;*(V{FDp=
zhCVKpZEwzuO<i8(cZvLa6|<4w8FkcQoSjfce`jP>_s%l*J7*1!w*BkM|B$SI%@Q*r
z0I7318UJ092=!W@_hKLC!Aktj{!5C#OOnD8uD)j{J$o2lsTA4EC+*Btc66?v;P$lg
zyF_!`bK31(^>=B|b2U(lY?XA~n%95lbKr6A%>vnJJp9OdRNQey8qp}rhnmGSJy-8S
z(AY=rvcBuY>LfrBE&{E6fZDiKy51W-V^Ff^L|`+u>AxeZt4zOruYz?1TBa>$01QMi
ztz$6Y?Vgft?*Xbjk>C6B$9YIbexLfyv#n9jxKY(>W518Zt;&JLowR3upC+QI)B1)j
z`}^VGWbpkyjWX5Jo8n+RN^9Io*2@5i7Kp1mh7_8j-y5@cF`n587l)|Lqhp)&9Q;=6
z$*T1G#IhY+L#Y#j^01~qnyUOhsd{uS=5EHqE?0kVa(9F~n5fVCflh#@E8Z46&J(g0
zs2*h%pOW?8C(+)AC6+Y2=%TubYC6SG(=jMoPJn41O}2vo>IcS@Bc1rkOw805Np}cI
zj|e^Q0Q|zgu53D}?3LF_KU&B7@D<tePAsC}3vTT>CEE@PM|P<A{*V}h6#?79kv|w?
z4(Sc(AB@9n%h<n+{UHq+t}xYca3+2*f5mf>{e$ttVlu$hKO{OFTLk-YhXuwM^X9#c
zGK*aypqTG-Ox}l$+DnkhKP1-e9Qr3O{lO#<10)RI|6qnSy!Y2I%<(Lx5<Maa6`Ibf
z<3-a4P+-*@FTJfXR(f)R)lidbQFlLb%X%4g^)+C_bIp;6DhZm7JO+hqXw-#mE2`_&
z6jpuSQCxpX$NPl@u|?|!an54eB)w>bR%O$>c*81oNPREJmdQ@aTR|;7M}yHI8f2<M
zFt!75;qc+2$RE>uxV$<mBY#W-jbVwV0=A6)(YUP<T*sp9ZnjOCNKGBZ0g$I`ReD`=
zk*BWg$0K!&r9b~i8#Yr@Quw1u&mo&ai=N;CD?6&mKPIjoJ%jp$RkakDX@EP`T1(k6
z(@s|YXd?P=!#}A;i+fpvYkv&EEE~V;@dG<r)$gtS(PV~YxgqNg3t`whqiW8E=W*S+
zAk9A}k)6fggR~rj6U<g^Z;^{<ZuO+|PYtN5S-Me?V<@-U4im#T+~Y@kWp>Q?kcpxP
zR6DyBUD?AlK5#05ToFgykyWIg{K@)oNLAz`e@Zj&M^O%WhU)HQ%)e*!1}5_-BihCL
zdG4xFV6v$k;+OM0+F|-<{T=^P2!@`I^tDj_Pl>9oOqar+jIr5IN-=7)8wA?qpOTQN
ztx@SuY4+boq%huzl~Dthz4B*1Bb9*5CDlrqy=>H8L40a|N-SVI-<JBHOguN<K-1v(
z>wTfDN1ffc8fthB(YcqU897I=({4GTO#2{h2d&p1s-){gFE~I!;O?K&D&X#tnWdTa
zdX9K;zieK@>;YKbkA`+KjJ6zwiw`sB4*YxdBnWIrZ}HoqH?Q)3nr(@L%fdGD{-yH=
zJMxN#z8-x)(Et)*!5PV&8tXjseqs#X)7TbD_I+QISI)xN_l-&Cu9ER+=;#bbP|06P
zFRo6rc)p)lP6$@kT<$1}@0(b$nI*EsWVDK%?2MG6I$BLu+p=ftfkPd#RUC&^jW}0z
zFz!CP7GR2MyI+4liOPPIRBP`WcPqR}UVlGL3Ff^XPBy$KsObi$`F?PRU5)BpR2R{T
zPC$y2XViYS9S!c+o6C+PW<fQ&K{+)B<h^9ie@GWR%)FO+xwypq-VgY8siV5pEysD|
zO{~QNo-RH_sN43uRm{1utp6XlB0X?GM*68|99b|yAf83vx_#j`7S(>DhX%^a^g}qX
zfj}$hxw<Q%S}w{mXl2|}MVrVaAC2f@t(HoKeiEXcy1^?(5)*c~+c%jh>fkR0+}#{`
z%Kao4u55=Bm42FT7O+t9nwV&CRtccWC8>M4otRv+Eefa=JSVJsic2Cr;Whe6geQ1U
zf>xsG)Vlg|X+=lOj)so3omx27luooqyxuwKIx5WEJa_PKbg)F)il+CmkoBWO`sj=2
zoJ_j$liFz8ySzcP%X>&(YQs#2$T5&n`{%@%y{5Sz`Lj<;Pr%H`QXLtq=${k)Y~!YD
zY{pAp@re&fHcC(7Lm4+he>O3>k#naPItKpvKPP%)6;k-KNeFw9RmN5^pu=aLltG`#
zD1G@SrS#_{Z1WnGo13)qpOYN0{I#X>XJb4o!o5=cb7F&<dh-6I=GdVNVsucT9-xn4
zkm88o?MUs?)*a528b3Dz4u%|hU(pOK)wDwu57P45tT;zmAlrc^cH*pDJVXi@SkbFy
zsq6Gu!kui-+rm}=ZGFHwplYE=>E+Lsr#NHZ%6d0&<ay%;Nw*z@MQo4{hVv!hMm|X6
z?-1i}^n)}R6Ho^aR2nT}?1RLN!v`hf#oZb~vLBcf%n@}f7o~73KmI`?IbI4ezWpGP
zSs(xE(DZsob;^yR#Ox!A{RbuvH(x6yFT+a+DF=?z95*VS($$gI8&z-DpS;$*_DWUi
zQDrYwTKgbL?*wLHoKPJ06cM8uA0&C<T*bn_BF!j#9UAi~X?Y6=iMWaC)=7so3u@b8
zYbfdc>)d}xFS?2)p;p)oME0O|?R&u;H_2A~UMiJj@Ie|gyg~Hni`zjMy^a?7i<M9{
z#wv<BU71Jz60AdXNB@%gxWn8ajQu6?V6VQ4$o$0!`|npu)~j(epF$m$`-{nf8qmo2
zUlN~G6^Ony@0m+;T+SB$lEh%+Nh$ut7{B9iK_(qB<QfMtL`^(7E2TgMWeB_1av-V>
zfGS>h<H@pAqab$|?A~gDv&N5-x+mQVo=4b`i#J>rOh?W<sy7_wA>5xd9Zx-Fpuvj<
zSbLRyUS~?39d-gMOs1skz|{ph)RkUTWL{S{Y?0o@Ne4Hxfy?FDEB;NkBDpXHY6C~-
zEVJQuv~oG_KIE@y5oh*7tL+9WBY#bF=emoqlrs8PUmxFMef(=;D}lq<>Sdy=Nhf-|
zza}X-%<eH4*%NSst<?Bm(`K~yvU=)c_Tq`X1C9B);4s(~s>h(>UrloCqJZ%*84X}E
zDy6^&SL0Bz9O*qM)3Ho&)X#nB<5CS=R2#W6E49C-*+nf`)Q}p7rBQzYn^ty2-WoG7
z2F0@Os32T~h@6pzCp~m6Tiw6<JjU)oTK|M)Roc-K^M^e^!|i4KrfMM%mDddv3E`sN
zRC<oS25;K*#&)gZ;)R`7-*aYo{kb*l<GEX8;3$~S$L#=SHlMGH{LS!F4~;VNH)H9{
zrx>Y!Gy1yLhTW|v>EnrwS7Gqqk}#;#I7Bh~H)Aj`-(*tHMQId8C>8#e!~=CLWXMNt
z4lNr0z3?|1_Y*fs(J|0rF!{GM-Rz<oa|(Qoil(c_mgsM3Vl<7WvNoEAeZ(D&<n^mW
zh^Pesp17;2Up;=9;E;qMt^Lhr5VcRl`p@zDa8?>lC7ckjZbluIP2Jx*(k%Vb_B<6G
zPRBZchZXg()(tvLFRBxmp66kv#uV)B;-CYp`a#WCF5V_v!_;d~7mvc8f)hu$|J)7&
z9K?=O9fQi>ouBFnedO<nJZyZKF*t_)Jq<)m@ntOX2_(WrE60|<rwO_Db&gNjzb76n
zPFsf<yni>=Us`&{9puIv7~&j}d;q|n47<Nb>N!%1j@+UiKqmj5q~Zv6;P}M&d-9P5
z^&kn$oT_@t0f7*4Ow3X1Mg|DBy572u7B<PXC~RSe>rr?=J4iIUe>eF)vP^uazo#h&
zg%;XReU2*)@0+?^KUvnNz|DZSN1bl99DRt1so1b9r|^r~@o&zR#F+Iq=9jq<jpp}p
z?%-{b<GLFbqZaG`-6ZWgJ0fj)0<K;^Ap-~TU$<msI~bt$#ddEf|1buBl0O;wN16rI
zN>WDuk%lE9*G@p=v413Xt7v2*`?S94XIlP8Vu5bsIhX#!=%Zdo$NynN^6Y8J|07ZA
zW0y2&1xEtc%uL1pjW5&_GWm}rFrj@bMXS2Ahp7KAU+)&w_jTp@uEf~JHjZ(k*u)_*
zma&a&#o<8@RHAf1AQ{<6q6N0;37HXBfrQ{cTqLB(r|52V4>PTEMtzPd#dPb`kRDA9
z=~AlbL`gGg;#LiciwD&yT%`vd^gwl~YDhv7k|+sDNOC@FfB$<aXC991hyRbg_u6Y;
z*4}Hc>wxMzH*1AJ(4T>m|5L>0ma{e$pbn*oSPWn~Kw*I3*XyNd{su;lGs1cIt2V27
zVLD0jgE6Pz@_FWr&8vGz8Ii7jBcB?!8!^t66sPBXH-=kQ2Z*TMF>gg*u&yLAvDN4V
zJEoJiCRjW#c)DmJPa!(k)Evuvi2bqkKQm488cRw>9J(KtC*^VXR4vt~@`9lHTu%$}
z!V6Zho?F5eB}=EO^}ODQC*AC<W<9SrPIy+ko|f|+kGr^@Z#6VG!?J^}S8+dlS@U|{
z2p*FkI|jiaoS4&C#~&c*LWcFUO3dcXi#8BtTtIl=3)`j!n_SO3L1C;p|E8js16@PN
zNT$`xzKP^RaVbiBQ&hGYq4(;iK4r7fNgEM%Y%V_GPN2#d^c!sX1!2(3(JMSh#zxRJ
zEC=+n#KPk2RiP{uS5-JZf?Uvb-fQcL#JXIbzb-T|8Me*nH@%&=^yzU`{GhV$Y4gG~
z@M6(Ni=QM`{=GaEB1=@m6)MN%?W<U(d@pq;zX<)tOXtt5R+X6OfAb})D;E4IJaD~!
zA2`^u-<rRdcMr_SqP2c6uOt==SP^!pcx(cXYi_h-YKa!QU<2Pv`{@M3RD5PDd{0CV
zC8}BSw_^lT0hM<*`+NF(d7mAHpSvVHxq?~L8C6XV0<IaIjjAKPSRw?r$wp>gC@?}W
za*3ubM1?cXq~3@rXut+n$UCcTD{ARiB9AP$249U?hHX*^x3xeU2^>%0^}wag8DM>l
zBSGP8--@0ZktBRSZ}9`97yf=)9_+$li@u+y7s5cNQ}O*g`zRMMnUBwr0@A#i_=N8y
zdWoPia`#pGex~N4(8}t+pC=V_7i(5eoI=>Y!fv;IKWH6Z5ojIXPt-le!FNo62a@b^
zAW(dB*!N{2z<&DWmu)h>PQ$2AeLvs1GT18yXPFM%+OSvebHm)|-A(*Gyp$&8yI)!Q
zn$3PcZJQA+ilUm|49*(G47B%g$1kW=;#F@5PnhU$;#`3~H1I~YfQV})n(}&r)BS!j
z`F;6+`Dbe}pdR#8>w)<p8BW-yx+LZ!M%Gbxi70^fgFN5kXW*57kfyWiI3@~FFx+*{
zDt?fUP7*MVa+>hYtolK|6BFYBeymu{57Iv2#lcelAg`921u+qI_hGuQAz&hCOK%`p
zB_YrTpVIn+G@-9{TE`FaMkK!YF~MNQJe~Z&5AxO{a$27iq625cCV!CkX?NdnAFmq0
z8A{VI9mu&^BjkNapbo;$x|zp}29ktuvjGULDo>xe=%#^A=YNnlF=OmY8t*7?KZD)1
z#^er*%Rk76U{Mi>G%!Z6-*4$vH6<ihI)jA-4+!i|0%#LkZ&R~V=(jEXdb`bRBSVLC
z>`HfG=|-L(1hG)=MqX`#yjaCXn%AYT_24mAwV~slv(0KY^1?NyAzAH4T8jhiRu=#d
z4zo+@H`0a~#BQZIV4s1(Xx&J2e`2f{Iio`rc^ulYKr>}Taa)jq=+KQ<7;GWBIX+0j
zpGjdlo>bFQQO20yhc@!o9Gk$p4t|2eSiO0=wwdVd9O2WjH^lNpxWSEJM+P@Cr_mie
zWb+$&DN)q-x@u}6{=zF8we2apA=s0M(v~%6OJ8yAi-7EV(^gesH+S}0%*knpd3__x
z6kG-!AvT2#$A4SFBUEmwji+wouCgDdWx*m91<Vig>Um_c@*n2iJvi)!0u?{ZOI>j8
zt160@{@yc{)I>26Qs8n0t62^db^0O#wLc*0qiXV+XbwpIU8fNewF)I<0EirZn0E>r
z#hn+9{V;6^UaL0n!@Mx<?#s*L6$1TX+DJ`JHmP5q{j5#>FmGElWQxc){liT0M@jxx
zM9yC)Xt@I_rtp}>I+@I>Qxu~%7sz53h2hG4d=Cve`@awfAMG1+zsnmjSOg0^Pc=BS
z@i=wIe3SCZRs+(Zl)!sUou1(tw632gIn*{|Fv&>KRk~Y0%$us;?*nXX?MHcd{aE<@
zD6i_$S7DQ)=*7=jg+6#Dvumq<l-9Su7{9&dM|pOO_FhbWZGb{lZFN6NTc^Pzmevar
z_B23h79@5YiI1>iP^%G)gxntGN4f&w9ghhBZgucU`aR((LkiLUwDZZ6YT@b3rUVM>
zDXh(H`bTN8JI~uC{oO4lT_FFXv}qbcJ}_Gx=a|pnoLYT;m(2@s8+O(UKgydu*@D^)
z6*ZuiUXDSND8^O-H?DCc3h_yxp*~xSy3J7XdH{v{1K-X2rLYWJ!jf6)!nSgbRr_%!
zutPp=r9Vz6x%L6>gkTsy&fA!?ou?#hV8xH~I<q)cKTZ=nc9O7QYU-d<MAlkC_6Lo}
z@T~iB+7ZR)OufFSJVwa5<{#(HghleW$8&BKT<2>#0xq=4ll+*9-g&3m27avLeH2|-
zK@cTcv&kQ4_J_le(5x(yJ6cZ(CqmXHW0Prt>SFg^ih;W|F`~afu%+va&Hgx_2#DnU
zi9h$_yit3x#qqRdP-QgnaHtFMb(#oBH`EifV#_fe*I5$hEnp8|c*Y-A;|~YNaLyt^
zU#t<mLzo&h371aTrWS*wD&)xCQo*?MW44(VW_0PQm2Re48g$R|^3A;E@P@<^q(WuL
zT?r<&nYID}=Scp>*&0=oxtG;$=B?0#R42_}-DX<Tqzm~=t5<v0w9SqGn$_M5kt)wy
zt1#O2Id~tnON;UyI~MI&`?EHnLckDD$qIo5y9yRzGAeWY;_<tg7lwZ3-B&5f4=pWX
zPPCaeD>_72gBd}l7k|jj24wa%V%53LMA%=%FLX0+M<1u_A{PQ`GY+YCLx6CMNdc;(
zu+h)n6~XmA=T9YD72L)VTMM{jr!ZB6lGlZv`;23>xtaGAVk?im#s$1pTt<TU+fVZR
z`x}op3j&3o<kdt0ju5o`Cuu#}P)x#gRp^U}#0JEwev&rCA>u36{3K7>Y5f0L>e`>A
zmG3L&%&1dUwx`uc4+OB|7H$5CZth-mLajeZ6V}>p9Y4tvhItHc7y@YuB7GBqjckq4
zBQiPP(gnf6G=@;P(FchGA&@DR!EJNe>NhHLWPCB6GnCB4*FHif&;BGIN(UNk?kAZg
z;V4aRx&B(3*Qh4216_!=IM>*XK#sG*omQ8l*gZ#AG|W$T+NyrN#*wS;&ZMN<>-sY?
zz0IHG#{<@0ex-4_+m?n$yc|M?yENga*~2ltFxZU>3|RWpw0KOJhj=NEuTg)g{4@{m
zERx)xrj;b0oYm;(Vf4(kKh0|ZPnTqdbphg_yMq2S?-k<jp@eMyX<AA2+9THb(|ie>
z@t6}Gf{KTZ9s6lo<B1bCAQaep<N+%DH19?7*(j3X<WKXaN-`|0HlurD`Pj6&NXDHm
zuMKRl`Q<%56P3F$2cC^mv;j=vTy%#|1o8IZj^?8~oS8<jmsyB09Cex94GoW|LOf9P
zo0#XUraC$^*z3{xSAUv!N>WB~!K4R@#^2=W9RibH)FC)KM7JZpz*s4$Zr3S1Lw}YQ
z30FrbyYy#yy2;GWfXjcDr^jOfICaJ!dhzn0MOCyoge?FPzn|qf_Mo4MO|1P{-Xg7}
zAg%jZUNz#Fe$(oImbS&<3D#Dau$3CmTI<j9rp($Fw8|aPSr^{ei0FTocEi$DOdbaU
zGj{~xJzDr#+63)y*<^Hi%Een#fuf|g97)sBHQMn@Lg~AI&5K1Hpm)TXFrbir<eN69
zBK!gfQKaU>oJz2@fe6Q)pR_mP58TlyHsjCoHIHI3rMTxB_@=Ey2WMPTwi;+2?JIZ|
z09*~@Pk18`te2%-|5+j+8G5)Sylo0GS1j9BfIO0PG50dH`D4-}xt8A3^p7EBxR;hG
z$*-<k#l19pXJMA^<vl?(7Y_HDdzq@S?yrqP?7b!kf*ap$J>;~tdVM=_C5=Y2`a22c
zt@rXyaqX1H7qE^1edK^06FzJQ*RKx5R|5=R;a=XCet{CiVp4F2y2!$EFK@8JNYZTj
zUfLtPc(_-ih~UE~Z6><o;V>%*TxvEaq?263HNW#QsN>0NcR?Uq`jEWb2pG8VEeBdI
z@`HX>qQj(_T@64z1VI|GH32I4n{i$JbDP<wU^cWLJaX8!V$CqL9`~B|^R#CBITuRx
z(~UUGf1cMii%lwio)(T7LOjC-QKj#6yJ0A1HEKLafT3Fb+~tz_;$MkoQvdTj%ZK`q
z!()Q-^R)CXUSj=J5uu)S{5)-))7<_%N{{_KFF?v<IPmkd0O-;G%nDIN&c$!qWZ>*U
z%$?t!iV7H4vUopFyYQXI?b6TxaJQs$WhT%|P7??dRQolao<fMBp1_Q3UKmU`#03qZ
zA=y#gP_?3_eL3dGBSqo29k`tII1j5DZ$sODTMJ~kSP&;55Lo|tW`i2Kky%ajGeesJ
z;58U0+X|!_u=lWEq@lGnSm`hF`FO~!=YNrB5A7PFgecsHdbLW04FuR9L5fi$7&V`<
z+F#`3a^=4$NFz@5bt*pGXZ62GTL#6IHS1>=I^NcRO@L$T&{s0f?XzR5HbPkTfxu{Z
z>A%3+{z8ZD9Cy^@FY>6-D<Hb^2gQ`iU&m71ru938!@VT<e$?UlG4qRjsvrh(D}`U=
zQ^gf3fmjRi{^x~CAL_Y<K&6FH33j81B%!L7)e|=)R2Q&`V+3cgsxm*mY^{ie2bj8U
z>w&+g67UGPP~R`|u1|KgNNEKw9S5%QwqNEIA~*wP)xS)bd*Bmf2EWwgd;@2vieF}m
z<o+PjHDX#-Wm~qNv6^3|4M4u4+F$0mkiAshFY|^N;JII~FAs0VOQKmI<Zj&>J-7sD
z)LoJSHx5)*@t0}c+j)Nt1W*rmRrqBlr(sVi>~>3&s@sShZ%PPpsc^+Gs%fFxi+yH5
zLG)}h(OY}-{=+sK!>fMP<}?$)IBD~0e|Wzw=vVU*yAhC1LifvpG(t8bx9eLGE>M82
z{xU5`KlAGwg_sEf408e13EaV>wyC~H60^D$cpdd@IkuG-qL>cJO1IK7B@syZR-Wx6
z=;121bZEMg!}H2kZKc&6>hsq~jXqdEl&`gdvGg@wjdfdjOVEg5(<<um&Dgw^mS6~F
zKx@=0_*U!)XfJ)*j;TpXc{i|?cE;{%D{N&>&rP2^j<8eM<W^oh&q>9;oVO{#>R(zM
zKn$xffaalOjCg4)?`T9uZUizDh^+iy9K*AM$Am^&)H$KE&n3e1F+nZtjRlS8r7wQg
zZUi364++69N7rr16naIq@jkOv{p=yx!di^{jS*Yl%4djPa?aIks{26~jc-Nw-N?r6
zS7|*E|1!6w`tX60)%aI=9m!%~6~9U=d?v+P{8d_iaq)mB%#a#F?!3@yf0fCvm6s91
zx?iQ~@13-IH3U9xl8J$Dj=#JPf915%s*X60Fsl7kp6JBV5Pe#p(-rJT$H1@BX804f
zZXqDKXQ^A8PO5O%uH80O#1H<FLsWn7I*BKz;I(JP@QjK`jcKzne5~tjE-*pX>r^wZ
zmVQL18CdvL-m&o4ye5B@H=6vZQY~8+8tBga4WhmpNNB#Qzq^kH*0jZ$c5(i?rtYmq
z+l)!|s15_RrCJ1BLYw*PynkI@%DIL-{nr^C?5YQ>{MTuox)Ets{5q}10bY>5PD^^K
zdjw&?uk)Ir&tk*WssYXsRu@G$JtFp~SJB8Bg1P-VZ64=+d13rI?HA4)mz8t`PRZ`d
zH{r2gr)~58WkezulX37sV2K?fDL4h$Ep8^GQyikXWv5g$gc0d<6ya`*YtJS1%cgph
z%YQ~+5V4D6XQQmZSw9zF!z}Q8bdD<pFUR``O8~+{L&NKGGfrF9Yz?{7=E|=#cknvm
z)wV?Cwi@ur3dO}3XksYFg}NS{w<6ux)GB<b+P2h*qkq`!H<{*+BE&8IO<qLL1#acP
z$<**>H&4R}tN2Y?z(1<l^;xTmx+F(JZ1<bIp00Jks!M>omj0Ww9Cg1*D<PsvBB*-(
z1uQTlK$H(}8mJt=G@|u4X{D3zzIEL1i``<!0x!NiiS58|(rzQdfE85c_WdG($=~F+
z8k$j$>CIKiBWUV3X;*%s+NRZ$2cg6MCwk(5!l+9FM%2&qu{ff0Y^_0?3vBmssLlT-
zAAt?{mcmEy)?HBB<jZp-YPWOiF2~n?)_FxUhYQfj<d?r%Y&V8I)mqeS!2f<dki#nW
ztZi!44Mz#B88gCJglE0kb{;kNP7)NSL!)#%Eg&pE?+yLNnE<=J?Yz=s6G<3cwVn2b
zE6z9}YV^fod8WhKV#A&j9JSlU+ZLg!8#L9cY|w=(&D%v{^~0xC6%z<d>#ReCd^dB-
zs@1qV(+q5<g~7YLu$^}!z5~Z?ayu<@7lO2@?aUA0#KdIt{*M93R>Lj{Bo08B76c&T
zGxpkSjOH+Yp>wK;d<ky_m0*B_?$Z_mrfW7~d2a-qu?Y<Hmo<dJWNfq&W$xf*tMS?6
zR-tRfX~N7I;Vc1aG{`2{p6G=5^KKipB^=y)&so*qrUgR-4WIhk;x0#?{@b)XFyi0A
zU!&r;`8eiPj3|kRc-3#y+%vsz@J`gn(zk|@E7t16^RN(;=C^qw-ABRKss6W_j@yxp
zu!!mjK-jZdt57(_vEK39ygDF<Bqtyh#YQ&p+q|dnzIK^aAwZ#Du*qnT7zzF-`aM43
zrmWLZxe4L?C6y-Q9FL4NqtYWvHZWTp(xAUs=7i!PT*`dFz;DPFqC=#wT$nLwMECS~
zw4AuA595fK$5sTjw=X5D3Q%o?09(_l;6CtIwytU(-U-{h7=^s)wuFqkz}fHed|}p$
zpze2hp3$`OGB1x}vIl(GDpZVP0Vfs<RrR~H4@eLAWvlsJz6@}?k(&S>4ZqWQ=}~8_
zPPNgGVpZ_Fe8TTzcA8ZJ4~b1)l)TNp>-b&q#^(JmScBv4RW<Otyo}_n_Z?CQOmXca
zh?}4!a{xqhg73GKO~(*DFdz43^tX@wGowz}SPc_48$}0Mnz<Mtb1~(zK&-Lmf0sA;
z`+u4C{h!%_TFJb?Zm7GY0QvKkf41d-d4RiYC9ocG$@Z#7JA#}4T9lAo0+TM`g<A}h
zvMCq@<#q#^t>2}i=`)`;`+ZstM2D>8@AJa8d-(P8-)E9aIzb>-{5~y>EMlvEpVu7&
z1&E?1TEEeSaYpU$Qx}Z~t?u`kp=-E*f?K^{DVBKi@AJ~YBjUDXtpSX!`RCRlVEZ^=
zqh=EcetcEq4_I#a`=g){?HfTQHTnCr72erZZBxI`vtE3u;F}vGfJZ=FdFl5$i@uTI
z%m~4Dw^Fd#XkAoG&D&hG#zdRjZ$1!lt0vkk2<csq+l`n>4_fPysFwwtJ@X1qNAc14
znym_nLBE{)Y%S)2#4yB!iarsF!<=raGGGess}dLxS!|pzA*48RAI_#WE3x+;lOe61
zR^7Z<;!i2umJ>voz4xeU!3uM8z&HuQz9Y87-g`u4`&{h;H*D{f@Q3aR@TiQ2$gz29
zLgx0~hg3Z=-9;nxYJ0Dk^`(EX-)aQ1FEl;71u_LfskQgq^w*Q*dJv~Q_Fk#rkrWLM
zqAt)Og_%ED_S$<7s={kHAja4Zcl9uloFq+wBCppTzdf7P1C!V9mQ*bPk%$!OqPO>+
z5JbNu(RfSf+#dD%(F6H;6MsltVV`xzAKdQK*>`*I0U#ZLzv<NweIM9Bj7copkjlRM
zXDG}I#i=-P^Dk^XW~~87#K}K^?&p}Dk8$9<bjqf*WS@TCE~r~qbWTTCGO@fEKwPy9
zyK_l(v5mLO>e_`+HWOX9V`gwgb3B+HVY2~}W1khdDhRDj2W&1{xjy%r7GrP>D^V0T
z9w~G_x+fHeo7-PkpXuUW5TGWWHH#X{mp)}T{$Tl#B)}NlR448b2%ESZlh86q2;f`s
ziH2msR)n|^y{!)yQ+3&DY@3sVb|>(D9cF7yIHBtswYz~ZUOk?Qaa~9*edD-o2n2zz
zU<Nj0=Mhj6gK;98-4h(Vp3rvbce5K?ZO4!t6ORC6A9&A6>abIO$p_MY1}4h5(hp?v
zJ>V)fH@+@YIlK2@o0Y3!*D%Sqc0_^au-Att`Wv^(4<s5&R<nLmR(&8-<LgjZ^#>Bc
zIXMim)-?fUY07qgARi25yde4_*&(8Me~0Z6V47l4$+MtN=sf3XjrK+-N8E@kYGMV#
z&D|6OJ3<t^bmcb(WD=FN6jA;T!lgwPvNfQMk>RRc{WOh2Fdh1R)&mZ77HeIe`j^W@
z&k2A>#;rF-2`i3sar93Fb=hPLec-(^k^TLHsSvG@Qq9|Ve5SZrlOIUrJ&dO8{0GvB
z+tOxJAIRsWr$y)Tyr7|wqzZ17)tC+-Z0A9{C@j%Du-z_cFg+uMvvyg~&+_M)z!qlW
z-S<fvV^>sjf)M{U8xV1Zh9Kh;slsWS)1;6=leBDhO;9ldv-udXn@#Bcy&i23bwN%G
zv24g>dDAYcSHwFQvPM;$b9?P(eBzqU(?b31$5s568jbo6S<&y5$?kRl?;jxxysZX4
z?u?Bs4JYWGfZF9&6>HHXEt}nqi78(0@7j8-I#xa#Q6J*<do#WZpwMUcgbo5!gnZEz
znKFiKJ9b<vZ!G&@zJ?Thu;hbzm3z7fI)_Ec2NO*dqokI7Fw<0fAG@vmgNeS0z%pt(
zqNuuN=kr$a!L)}4LAFx04z}$(Y*qTR=ToSTuKVTCq9QjT-K|cczZQ7<T2++S9w_Zm
z#n065w7RI+3a7sJgPE$4`YnBxGob#1>CoZIMcUc)!AuCfC>PvF7SnrGpr^wP))GTV
zuH3C3eD84)j-Std>mu4}zhCEXp`4CDpdC?8r!ZTZ<BE*V3Lb$TgRmj%Rc9y(I2#9w
zsQH$#q5IQ5IO@u}LNxcIRHioLnn%uStPww$=n~tO^B>GdX)y~f&NqQn1Jidw&Ctri
z_-jawBS_MYV`7x{;>B2qF_I(V7kPh(|2LF%IX+Xulo^e&_2?1168&`3&umQPm%e0I
z^)WQqxAdQFF6t3s$F2z<n0N$lf;}-tmO!TAx?uOAGG7R2EoW^}0MeV|UAv)LxO>^n
z7<@8vPLY42Gq@bPrFojbayd2!+^vg(>~{PGwg&S8E>EE|C`qo}i3+>~Q+mj?Sf`@9
z`d#62l84KB;NqM{I{G&>vo0@4aWFSyrrq6Y%<kzM_S`KJFWOe%P8=EA)<kjXylVD9
z-r6I0Z@N!q$peYNTyS6dK)&}pD{Ywy-fU^N@(1z`#t8ydf3g$U@j#~GL(GI#JdkJ@
zCau>hAIN*jIX{e*kPJVNHrme5TXppEM<k4@Q7?atWzudz_#d<XW7cXP$ma=$I`DFP
z9uNWg9dC8|z#0e2+pCYmN$p=BuwQ2<H3bkq9MSUxTCe3?FfHm6$3RR&Ro%@~SN;*2
zY=0o1Vb1Dc>C_SMSZH2$o$9UK<#AXvX(WD9_r@$A#4Xjs91LjquYT5s9>@;>jM9=<
zzH9ah(HD{l=RwBRfXJ+q0c>OxOC!Q&oL5g+bJ)}a`7Ck;J29I$*DeH_c)@VgOlyqz
zC!ey5QNK6E+HrpVQcT(r>`^YqvLfyvAd{w-gYCsmyHbQSh<J5Yy<pZQr`xM4OPTKG
z0!A+>7VMhvZEt9_`B(y;N{W7SJt`+N@&$Fs)`V?XjA==>Uww8X{@~9iNd0E~!KG3}
zoLLTV*cFW0t(c6%IGy7_A~gHm9PPHca`#1)m}*yi(%78<^ZEobkTop_p0;<R!$t)e
zT#p`b)#DD`hze{QUyn`w!4nEQLtC(W!eQxKXo0qb!^pXFs3^8$B^og+v64iYL>IG?
zl1vU`L!RWlR7HJAf69ur5ZjiQBobWu3hc&?l0-19TN}X^qPYjz8na4O?Bde5s*+5`
z9xL9eOY#}usZeT4;<&&?$v4YxmG=w{7F;h^k~cGtNB8&OygF`sq7@sGJ<h8%{zAQe
zyn<}6&>qGbsJ<lAS$AJzQ=6hTK~9oAO|wShal*%}B}U_-a6H6?5;RL@HES=9&7%f%
z6idECDEa?xoiVhduxiKDNfHIRKj6F8t8x%d!99fF85VFT@ZqXQV27Dv%r3(n7e2gH
zCX!TnQZ3)X9{IdL7GvQ}1p+4!LR<)_`zNs!V@{@HDQf;@YSu1B1<t^;iBe<a+;rUi
zm!pQ{5+q-Y8i!N9s4MzLb}zB?Y#_dW+^%Y>+@Z_nO7fANTxppEyQa3x4H@tdj?L>E
z&*$Um0=OQ-;TE!m5^)kR!6oTTa_zj|Seyt8b|dNzB67Z|8A20o%L2+-gSjea>Meb<
zZ#mGA>8`{#?)}K3ZU>%8WWVTk(6I>#DXPnfMY|Kbiu^lg5YMg!3XL$@cQtm9BbKI6
zf349rRA)lUuuV<Yc#1!FuZZ*C!wKu>tN&Z4Z3|4ldtuvyc^`W)c2FpJFipwlKX0WE
zW~#!x(aIjoBzFdlfR(F|d*`1pZQP+RH$7zyDxzvLni{KoFj2Kf6h2{9QTDp0Q>l*6
z6BFE+H4p0az^HcjgPDpw)-WN2t+v=DZyuzGdsOwMUAwIA!F<q=K7?dBdsWoXdeG`)
zc*q(2Tx)vpJ%5!aAxCo*j`gJsXf0|ucmkQ_gZYk15~f~PCbkE%`#jfR$NlEI6gnTw
z+fl*FV7re+9dsb*ysS6=(3EZ=8&ID;XM1cYh71RiJeLZ>mNaX<Hm>2nEnD45O)By*
zuI}@yMj$Bsa{@^e7z=hGK0UIzO~<erJjKyzLZI12VbFC72T1{fANy0gtc~zC2{2UQ
zWOuJ!32;cSxKCp=ZsN_-U)54@!YQ@JTnxz_K!;!&uEk=tB^&zrfan5ryDkif+KQ%2
zBZjQTvD05HE>7>D-3X{TeB#aG-uP!HUR&1W6eip(_f|j|_KZd<l4JhuK;La8Amr6p
z!Q_pBv5wtQw@X*iuLPuC&W^kB)kgr9^_bX{?g!2HS=$IS8?j;T;r+QOVCT^4+=~Y6
z3_mN`oLed$_Lr)rDKkiB4`pI(!I-b)p+r)H3F*>@GNbF9v}{7G?4eB8;+d@cp|tS{
zFn!8)sJR>WBKmzO(HxILtBg*XxtXifqI;0%LUr_!6q|ZG)u?4xcem}1?vkI!H?8)e
zbc~!lWqb7N9BMrQT>4gjzwOmu-`;Qa0XkcqJDCbYmi|lRBQ5sB3D^?-V1b8qkhMOP
zX}{MciPj$9?mPvNK9uFY%Amx3-_Om_`A|MxkkIf@L%Z})I^T*cf%QIgpZ>{8v@{3(
z4+yFLQ4jGlB&6E0zbdFtS0?aE`*R!DII*Q9K%u}gkbaj1osWL^4<ihOF_?-9jqbXC
zAqEI9NfvmUjx`y8)43RE<;w>JRzP4m3wBuzc;5K!JEM7i;j?x{zrOU2&Fa_r^L91B
zB~~@Pi05Jk_f7aoe=U{*N9XxKZ{<HHoAc`r<ugxt$k9A41jcO~>5KYeY8FtJ-O#L!
z5;f+g;Eqgq-iPIYJM0T`)^4fd2+ZP&5W@V_eG9*Cw*x)_ENxX44<6cYcVco+bPXqA
z-da>8%!u7pj~dpxJ_ZSXJ!Tt4AXpM^s!6ZHF5c5p5*`$;H*j>R25Att1;XFQh<hK-
z6EV`wCFo{bB@gFKmh44v4ljK;5g{(p@W2lzx*H{~LAn0C@{c@&cRZZ$*+20OQ}J-#
zhdCoQFV*;)ztjdv&%smm;Y<fE2{~)k52y2@F+H%X=3$)|#5fza-O<Bqt}3aGPmdUG
zd+vW?m|}G?ERS55EPD0|7!iVTd8`-eylRK7NxxAK;dG%&?E5#-#>9sQ85CxeG>jwl
z)fYMwS6X&N**;F#PR-Cc8XeP5!ogXuet!2WMA*@wx%U*`WJm~L2-9PQ0Of&r$5nz+
z7OLTg^U=6JCGfKIg2<+^DgAt_p~Wu5M6l}?ZCVhzooTvwzsRlcF5UmU!i{k`;QPsr
zCGSkE4SI1D09XD1HAF}1a@W5`a8$%{*u9+f=QeZ~*J2G^@Pc|0%N~{XW3qN#(1wxq
zazHE;SDHj-<F=@dJthy&3>aF$clnFmd^jD@N8YgIKp&~#DZ6za+cPdOSc%>}5r^j&
zv)k%@@aF$+s}H~Tq~wI`JMMV8)#Y#}Hore<U1_%#yWlgQvb)h~H-ceX7dD+Q*hXyb
z9vsyXqp+Db<8S*XJvX8Yk?#qIK3;}f8rGP{d)`(jykW_NX{CwWQZSv8(o9!Ky%8Ig
ziu%}FWu=*N83HmuD7U;cQ4djGUH-76G)-0e0jns@)aap<dXR~aSfwBh^G2>J&6iDM
z(T^rYOQ|l+w26X@w_i<ZJ{NLKjNL*MW)6CcQs)9aC*vNKCPi6Y3}a=Mt7flIJGjs4
zOB0DhR&VuLlkn%`UoL#vnxhCGt`-5?xeH^t)<CQ^*^#%aV%IshSMP{vN%yjAIXVTF
zy@Yp+9lMWc(N?ZE;2lW8hX&Lav0uk*C{SwSg;a<ykKZ<4gqxx@*rZw!D*QWkUXx3B
zTAK>sCkSyjVi!s?8N=atIE3la%m)>tMEEwjs0sSQ7wl4?E7q5nsDR)6Pe_acpH@Vn
zSBjgXZ@^|_7Lv=Gi%zc=@w<S=W==JGPH|7h%ADyM!@4=pEeyB{@Va`yKibn8EELCe
zw#yb(k;@HBHKz537QtiOjM>cq3ipm9P@s95y?!eSJyf@?L?K}g>~_os<nMeT@n>vR
zsKaieg$V>t5WX0HT=k1lF0^`A75iNRxvq(J_4sHqSKo+!(@ShK76P0-)7QQDL!awW
zQIu*CS{rRU76BhVTtK1GM>1o(ABrt`B#|1yXG<SR`x#>)F4wY0GV#J>{Rb<5BvY(g
z^>e3<*bY_qTji)#MD<n|<W%ajA5;GnhALV!E4*Ks^Xf;^Y48>oZcVg3;xB^TDmluq
zYK3Vr75bh>GEwjMkT6yk2z+<e_6i90hMb2ZR<E+Q#3(d9lJ_mvtf-Z(SzVmwZfnuc
zgKh2Bs+l-4VC@0<B%0BfE)UDp`A9yG#&~I;N!BgL)VE8v`e8YG^@;G84&Y(cUd)sY
z#UOCcLEKP?!8ebO7i?VUoWP}c@{x3uj&^bRoEJh!<+!q^0tMH6D$VExRVFG4@*(|3
zzxn5OF^0~&#=}QkQU?SZvdcm&h4jD->ydm!GD%5Waz$;EM#%M=Hv32>_w+(1h5_vE
zRRJ7b`mZ(@)7CIx*J5CYU2{J#q(Ais|Ll6q1x6|CxrLZ%GIZObzFPgEM!2pR$Y`IN
z)Y;AAw0<Lr^OghqA+~^DD7RwIwel+9VnZhr{UvtMZVO+?4<XrVbb!{#x7VE*0D%h-
z?W(fp3hlGI`fztYsI@rXe~lf;Ms&ejzpvjm<9iIVCf=Z$EmFMLR-nvbkmh$g{?gxv
zV4KI6J(`Gf=Q~#NXxa_DF(A}O6YU&q-*07)CYp7Y9+#y@(-Oio6m5uYhgv}rIQ*&<
zdL@ux&q*%%%16^d(9(6P8~(9MP$xM9s~^pjMgS;mQ)*O%?ACS*B1t<i)qgb86=^Zu
z3T98V^Jvyks5(e*ZB#IV^2#fpwfaZ%aYAlTVvTc9Ha(gToUtyH0nGu5krBY6<$jBI
zizczHs?vn83JCzhBEcW#4(Sk<Z+YOIP9ed#G0d=Ifip5aH&pJ8X<-eBbx7PK!EGS+
zO7dkL3d}GE<{}MDFiF7^g}pX@|4*Kf*%h~wfxp{*_u2XAw?T@Ksesz;Ss*V6%Ql70
zntn7>TGEUSPq@tKVhrRsi+w5jEDVtNz=iUcqZRx^&f_to9Ws%EuwK!;B%XNo(R`8>
zxkv_bHIRa6uz85jeVzsSnkJY?0yZzG1ek+Tu4|XR_`EILpG<B~KOPqIqF`vN&fS*H
zZpQTVb8-yZvdWNH!VX8T&8{(A5%hj8*l3h8U)X9)f%8j%MVzZaB|S;LebxdgB%bi{
z<90WCc-xa9tjE0Fw^Xx@=*-mz$8A#>!ev3lDs!WFPyAzB8sx~y3ES4h^pjbZ#ITPg
z>SXJ^0|h^pC^)Tl>0@c`m0X_q!_MNDJ(h`%h<OlC`D2+NV4u;{JeH0EcfJUyJ9k^<
zW9clwThgi?%lG<mbU)Q9=s473HIKbl&Wqu5|I0ty{aB{ow+2T&)mp8(BsMuZGb|tU
zsqiuglLQeds_oTS&iNTtAKkve9&CzHv?7wkQ74e)f@sl))S|5d*9*(o{#d?yh~1v_
zM;!rlAG}*<Ohp@Ex9wPbwhSKcvfda*D`F)M<N-B4nqq<2kiOhsidij0+oLWiACI0<
zAVL4w2R@dMCQeOmCC<t78dKt3{9!sJ%m>&>7Xn-_F;Q$<^OD@vFKV!-lHblHH95oM
z=W-0K0h*eLFCv!P6@4+|ZY?q!z;VVHv8w`(1slQgRPp`4KxOrHyY^T<E;#y8q^Hc|
z^RW^nSYn;72Q<zG;sXhzet3dK0d91m%8h7sH02Au8Q(`-t!&Hj-AVu;6sOvjjR$Nc
z;3wJ5?Wj0{&n#w1u}UNT{jOZN6BP)0jDT=0mV&z_rLMReP2AoS>R#8By8rC3ZD<Mx
z`;*RVQ)3vNU_AE(S>}@Lm#x5z&<TCEEtKCnWLB05ASDeLvy!q*LYM+rX<4GcvqYtI
z4!x`_QM3E5@?I^=TNWu1kM=R!5p6wHNrl?_tuqMgR2e<FA0zWt73Ikqw7M)&_p6_>
z8ufxG6_~rr5_J=x*=hs)DNg@AW$CbBi|XxP7vS1@2F}>t_yqSpt5<E1DU7(4WmCX9
zGjhV3)gz@CceIqHQz|jht@{1`U%meq)*b^u<N$Kk5fk7hoybP5GvG9K*)g@n2L=L-
z?vcuKBM+$i3Hj>{#ZZ`JsI3sVbN!Cncy#=h0;^A!Wda2g7w?SQ`LewKxynwvf(AHK
z>V?o){wlo?VA9i!yqba3p|fp&++r7na6g(KyQJUTsa<wC#)(&ax<qGGikWktT?u3e
z4+ZV9GiJ4Bo=fLy3?;=ro-4~YuG~2suWNxuujiq+d0~yp-4EpJf;BPPZ40qx2Qda*
z49JJjTiOlP8zisO&A>NZW+1ezy8DVZ@-5ZvS^Ao-1k9tbcDL`h1YI{PUe${B42@tW
zb0?O!_nfV%W%9%My4?*pg~3T%SH%=#$~Izxcr$odAbev25TdYqu`V8Rz-7L%mP9x7
z$a+p&+ZqB)iCYhtJ)W;d9-HG<5}!ygzHOzCXCgjH-LfboQD(Q5Kd${fgk|-P#}joe
zeG6xC{dw5s29=NJi}*yc3$0RFO3PRsuv?kg8hw$+%^A(zk0%;#Y_nRmfYL{7Pq9Oi
z?OL56yK$CV5!>E?_7+>A{_%Vo?e9%FOqw3g+d3t|R4A}!^+G15mdDe1ggr5~U9IW{
z_GH}JW5})kvguF}R+G?Z6ybG(3g+=FOGFL-5B@MuDwPH66>RtX;Xc&|2Sm*mY$#?2
zO*@lc5I%=rv+?4n`w*K<>aTnE+WF#N{REwQJd-Y)!L!Yb*@c*b{iI=-*4OA@(*9i3
zmwWGX$z6($k0zPO<rvCQm+#EP>><T*I=rHC%xL}nVYAUUhX%RFxqGh)VtTh@yJ~Zf
zXZqy9=lVmSTzfp9^KBkZ3;D!+07`~-|DAQbuD;=TVC4LCUkKdVJ$C(~20~CF54L$D
z27)da1=P)`e8fes%c}hT|KZNcc1vH(J+?Hq5?>fF*ln$g{Ki&ekq#WRI|8}%_ee}p
z>#pfD=aRpx#ra~jt*gLg1E6Lb0Rt)7rust0%UNLeL@Vwbu%({_>9cJ$@W8sq%|4W;
z3Ym0b{!2cT>9p$v`sC6N<@p`<V4sOgDf>{ON2Z2=^dHIuK$J$G#vRcfZzc4mA4>EJ
zKVg*t$~hi^s`^l(dwdKXShXsj`|e*`&4&`9Fo*EPyVW5oFzk||U}V7dd?=CcXFg|j
z(QO-P&TOw*U>LdE>eUCJP9qFziV=G-VfV9XzF&@^R!d;nyY$r_Vc8mkB$1Je>e{18
zAKvvHfzq*0S!WcrK!(SnrJs6ee|j~<rR!X&A4+GI6r#MQhcuY>HY<E6(;V^nWsx%;
z<4ZwSCgamh3haDz+KVP-Dhhc__t^!_#rt2zTwz+M%3buL3I;ue^(7VXCgtODV2!c7
zZ?YNHgxMUmE2`N?)CA%lM8_?C-F8()2sjBwHCODa$6s650wQsgTnjd@!6H7pUYwDy
z5n!g+q?<RIE$Rcl1!}mVO2=J@RQ&tW99+_}{@z2CThRp!g&zYefm;*j*zK58C<hG@
z6;=hwbHJUpJH^Rxc4ke#IY&{4#(rr;(Lly-TaS*4tjvZwLWtlJu}wi`yX|kXds+(S
zkh2sWOf=;Uj$xN=$5M^ExWJys#CIGsRx5cT6Bnj3{?I9XB5w?i*Kd1rBrAI&k-*90
zR{li31Ca)vK5IMTyX&;aYpV$0I1mk5<rA6ou@_D`SgM{#v^qiFqv|ITiNnJaFz|^)
zqsfpB_V$TPm}C{LRW<x2u;UHzy{?k2yAPh=_s%x$jSBq)<hS}K^2ySG^9}C<_EwVy
zGkC_D1vx1dThwI_BF<CR8gSt?hTN8?J>Yu2DC>w%_ph~1VeiK!I2?=C$?Jr*?o~O<
zg$&CE)a~bX*1uvyG5d(h&<$8Y6Uv)%0X@VwZCu|ZNJ6v2bv3ER!yfeE{1aKpm7}vH
zCD&B+*2Y*aL=Puj$bsRJR=}^qi^au2YJVw41Q+sUyBsiFf-s{mZa=59E3xK;%<LPo
z+4zP&S){80YILIetj*=dDKzV~C`@q$=Jg53&pRB6QE{zZ7_%Vc-l?v(#c0&xZZkKc
z!!e$92#w;K7b0)@<tN^AkBR30kF@7jRGK_B#HFz!@J&P(x-AS-uw%>^gae|ZF}oAf
zz+2AN)TH5XligJ%x7fFJtv6?^N9ytXTpO`}k4)O8_FH?S-4itA|IYMm701fE+_nQb
zWmU2AOgG5+`g!(SNqHg(_h&<8TAoRG9|=aStUQg^@3Bdb<uK?GTFc9I@;kfWX%Bal
zXYRKNucZooxi|-ssEqDhbqoh)RnbEani{JvPY1wj6IP>NhxXa-s6v7+WJKlp*wDL=
z?X^cZgV#x1&eh}lf03BBy@5^lS^P@N)1g3c9&0Mk#}V<mJZ?a<%E%dFEgBN=PH~W7
zajzn>-&(tV4&MAv)=``(XTGgd=#ozOSTu%6@RjTh^q0O3Z#+<*4kM4jWkdIoBMEy!
zpxr5aJPHW!WRo%N$x>TP+<A55M#@QgHx=VyVzJY(3qp_Z<IL-H5$%)b>|(%f<B+}-
z*m`(@al0(^+B^>6j3BsbV9c(>Fq$!(v{{vbP%j3$`rsBtCp#zju7X_oC%dLE7*iyq
zHm@)De8bTBdh`wVf@8%(bitiBXp4a;z1R&ElCzX>r^9wr!*y3}7#|Lc<s#xPZn!0U
zUc-oez*b^n{CUT2xgFiUF@Z6St*RyHF4~>gpqSw&(dt@p=K3b=t}6b^)Rk9lUC6mF
zHMi(SG{b5fju*2(p@2YvAawVlM}mlP$Zth;g3{q2ja`vIcMmZQ_GBhHc*HYS5}$t2
zTj`Vera9ojY7jpRSlN^Lvh3pBMy%B*(>C8fS+E^XW&+0>GQEB)p3L;)?;4&E?^>n0
z7(yKE88@q*%oO|<r)>3;#SP0iYoenA$>C@BlbMLvbS`A8RRy<mOAP%UH75ERk<Fqp
z+&72I-Y3&Z<K`EmRv&o2mY9jN))cUNG1GyOYko4HK1m~}V8Sgi8X`1d`_KBe#te1$
zVPrJu{`&0!ei+W)Wt$yQixdJyC-u&lY9f)t?`j&mhK^frAk&pX_t-%E0rGISnjuYE
zy4R5*Y1~*1BI4CJx#5vbOvWTgsBvC>INnpL*OL;NTu?oiOtNVS#8#|#7T*~Ce0A4n
z>{6`M>oAj-1NL|dZ9k)LH;nf{wss|IH*tB*#$Z{y9=jTaE@hn)E}VexU$$#e)W+dB
zukKGTea)_`gEqV;AVI8m10-s72srR@BN+JJC)446bi{5(MXsF_wj8U21tbpl@V_y=
z{KapMl_%fxcm!_J0I%p#aK1y(#oIy?MxS|J6=LrY^z2Ub?5W#$f5r%o!rI)8FQ~`X
zh1DBSab7eo%1!-swBPRO*RC(vR-j2LgybB!9ph-n;K}U6nR4Gc*l@r~KAh;tL+Ot7
zTj__>=$p}*emK)4BDdtyRsP{jldQ$~NDr29J3gFQQ*0<IKCC<TFJU3y#YU!5-*2IZ
z-govOnwJmfLjlLDV$!bafazB|*BIg=A%xJ34Pm<j4cTm9?(ku~Z?XE^^WjAGOJCuD
ztka+I1G2p_LKyyDtB=p4euXv(J}OD~_G{J`d*}@erCOpl=n3s|m$gRYm9N12w`)wU
zG)rQRj%bimfOZNXtM6_w#{zr{yiu>dk_3b>T+y}xV+=QS8H(u~KWzmK&INupuHS=h
z%08(|`%_T2^MM{Z(bF~+Ac;AYw4E1(FynJIEoKiN&WGWc0-s;hym2Ggr6}{`0`dH1
zb>$IX34R%nFw+B%E5fPWO>t&pm|RwNRW+TY69YIGa8uTuF1r@h{}p*N=3@*9*RZsY
zLgMS%f_ho`7L&jD;rE_{yco;>|Nk{7hyR}ksd%W%)r&U;0ZSzGGBE}O{OFT!#e9tw
z6cKzSmZ1Ht-44u>joWH*iy^GP6AgXey6$4_KH5v)vAg<m*GDK}T@8<NpKJ&lI3f}$
zMU$gU_Pyx2nT!I(_P9@?R@jaS<UQtF+CGw3(X~J>+>JjR>f*Y^Mf@ZAww*|cR?9w;
zD3ve)AW{C2L^tRqQOJBGQ#JOwGP0`pNPbDN=kN3Jl^;ojH+Gf`KOael1D+8JMK9Xw
zK!D7L?xb0xDy$SU1dqTw_9OY=!MmT>J*tphGB_ai1Pnsud;IRY_`}#)WN%P*fH>qq
zKk79|_<WcKj*g}%PNsd$F*HsaEKOj*TA~JNE4YRAd7NbS1rTW0yc}|en2zWhQvNzm
zhR$e8F1cPirU_N1`(8D4vrHS%G&RDT4(ZRtA4CHDk#x@bMTvN4Jb+<FHfWQvTK)v-
zwewNj?wa%|;o@QxMDH}e5F;a^n=-*q$9(U1ewUTK7$YFpj>p-#6uqFVhdsU=y|luO
z+l&w&;OCWSib?s>*T~GOGGay8RSkrNb<@&06_EYgt_gh6YYgj$0MCSt+jUI_&t@FS
zHF?f%_u8U9aHrzqMy`cC>Shdy>;dd467HxA$jXC`&D*Wmsq9lv9^<jUziulr(7vSl
zxGf~Hkh4|6ALxTd?x<r%a+Iv=fo-clWADc2sX-o<>oFk&1N6z=h(C0VCQp>j=##nR
zsJW+U@+1MU&Rgm#>Bek3+P<M6bG9QBtBcg3<{fFTkd)2$Y-yA>cTMnSD%<g%yNXS?
zaZ$NyG6Xk~-mxPQxyNBa4!k3gYj@X#RjL+5;!>e16?ioF>K&Q*Bc5VSKzswE-`(nK
ze83HWcVzl4ba4jk(I;k(iHWX@(v`pFtlk@+Fq?!`6VwzInxGHD(|yiKRuVNiCAzIe
zH4_i@O>5ne54vJvr}jW^j7jbgdP%a<8E|0F@F3SHPhz0c8$giG4tZ?os7eqA$W&EP
zz*|88Q$D<L{XWrma@;0&WVU|apld>i+JUuFfEn&1x5ZSfQ=x0=Ybb3r$_X583B(e3
z`{U)Js=9hB?c+;3GP9NP=MZ(~a@1e?_dIgdlbcPs{QOE34h~>&s4qVW5C>h2!KSAM
zDL*y3SK&CXMLQ4AJg+a#2H$1Zg{=#J3I3;#<Qr#;fgI_u(hj<zmhdr7*f-U|sgZNG
zte;%Gc1v}}#?N}%!j<A&aJpfkq>1H;y4zNT?jRunIF#6ucht@q2cEiX(RX^wA<4O`
z#uyb6ephwb{jJ2E*w~Tz%LW$?ddg|rjP|_NdC-m8y{Lr-uBi6j5*!bngvfZixE(l+
z?4y}5{o#fg@JAErz5nmn;iVtVTk|+-4Oqppk0$Cnm7b^NA5Dxjc^!5{J@;$l;aedf
zhdjblr8;m|(qx@iB^V>ek8|u-e>Crkln(<5ZcP#D+qV0o`T4sK0<YEIKmV%j`Dos#
zyr#0AvM&DUo&=aTdvf)?AI;><a&eams{z8kSd)G;UxYx6j{FK~36v&=PY!zkzgE?4
z@Khk}F<tv`xrIDGnvS4`mKN*$XkHvP6~~}&xnlvko2OoH;8PgkSQrqj#6)^frXkJD
zeyk{lte{R0xWn6cv>Wghp9~zvaniypDT?>?VW2h@13-`j*A1Bpi*$YblT8OsV}z}}
z7!&7jQ_iHMV%4Y`S<1`NnVYJUASXII=OT|Q!Wr2+_vUQugMUj91h-VuqP5~vqk4@g
z0mij}(dYrZ=7li#+n>7)FXHy=sx4ox1<i!VW3a{I6agPQLJb6*-M{T-@#`N*Y&>Pl
zAI--;!8~QYc}sB6k``<wFzxDfs~x)?C=J5Q*=m&2)`cW@G?IO7wk9xaoH4r_lY=V_
zv#~DphDgA{GgcpC1AVOZ=6(2F^mb1`QYxP<)q@*^wc5UqH)hqxT^?aii6SrJ?OpOz
zBEZ(<*HHRYCS4ad^;+3e$-IxVf`r@Y7{fvz)4Os3Kn6xsv!^mqHDMD`@l+=8p1}!E
zrB(S<rpF!+R>Ov`dMZt1(;q~Vua15-wwk9hK}M`&T-~Ch|M5?+TCMN~wViuy&r_LD
z&ZMB4PAGLn9Nqn8?^8*}K;>|3VjAiN;v}aOhD%SSvuC*5ngcS9DlO3|Y=^Zzm1y&G
zgoSFqk8+=re1`@uM{qwXI%6i>sEKhO6DUju-2STfseF<QP2fg05F?Q6+ea`IRo-%I
z)q)0uY1~O0FAfg5tWD~-i)!q=e*0?KRG^Mws9g~1g9v=5Rp}IQ|LmeFVUy+()k_*w
zQ4+IYmjlGnzJ8lgI|4h8+m$GJjazXx#tM<TGWTly;Y45W2>W<WW6?QA=p@bdbDcIH
zHQmRTRk<Fc##9w$Ys|ABonyABjtQ3LqJkS56$XWNGXRko&3U9{t>Y;_IPF$IMIHHN
zTTz*Z6&|zOF%}OfxEgR=r_7CVN5jUjaNO2}4SE4t=G_&XZa7wK2nA>G_(>1)x*^ms
zPq5ACz<rDEsZAHjZwbY$4|ZmEmu&}<`}z<~+S8e`-T;k~r!#H3Sq`C^pUxYo-;cbq
zr!xukVx*9ga#;D(x=PqfJDyHt?)>j@tI*$%bQ97{eK48CdA6$PgNqhsty-mh_<X?}
z*XS!bg}-aNpU&qY5(pFmSd`Qjt2=MoXM6O;&|c^Kt9v?~Ic@uow%gvP^O@i`C$f91
z7e-x#dTfeup-V>A^z{7$WweKEa4k{Zl&(?@ESZ^DyE=hkKn#4pp$CA&KuI+rR*uqR
zf&vL7xBK3hJwo#%s&{Lgfxy4tpQV}_r0R0NEC}OwIy-GVFk;-un&Pv~ZT`;(9<XQb
zpEMPt_YAha9v6feflN6_rti<|)no?r9lIFI(lX-K_?MyzR%t|Q;;y<Z5FBOD%*@kS
z7NK|n3@35G(J_uz$>wY}{&H?&bZL=u=&Jt0jWt~4Tjw+l?puIW;x!Gsn=rNWnr!kt
zkJ<H@tAoS+w(xX1OV993jk(}P{ZqT4CZldubu;EA$vTiT3G~gsg%1`u#`evXf3%gr
z(&yJla$8loL2Na8;6;hLRXt!ghZCI8c>Ac`)$F)#8=L%Sf$0p7(hRF$n`)c_E8dGa
zmcKVA!<K3`V34;RFuFW+hE*g&>tXdvDiWa~>qIzMst+l_%Jcy}bbiUoD-t<O;HhOh
zDiS5*_`>)`tfC^3{rl5Q*<V<tzG$^`=T55<3}(#ZIH-tC<?etr6=~8BwAt>8OfuMd
z`bk);&hUmHG*S@)T5lY}h4(>jm%XZkHt(d>2LLy>f|i<up`TKHyz`rj7<x3gZ(B>$
zP9`gaLyA9)c)DCvs}=d^N(lk2qax9N5<g(0z`LMRh`@E&u^8~+kv8j9ReEmZUGJG#
z24Wi9kJ?aFego1e#Mhf#`s2~m)BF0JI~j<<T%O^sIUfy{zKS4NU${uP(fEaE?vi^y
zq^6_i)_%JfV<NyO+xe14@X9N@?XrF@eTC!>(W4VL7vmL8GtmM5qs^+`Npu2NEAnAF
zhP~T>Yv1Qov#XcP&ew_%{|<|!fH>-JF?>;D44pNdwh&)UtnXrsfx8TCD%pBf-U0=(
zt~UjD`ELhpIpE60)NTbztRlhHIm1_C8?<;@+1rA}rgffhH710>>^w2<1SYL+C#bpO
z(nDl_S6dY_SXU*w^aOnyMM5BafEz4q3L?$N{K~U?F&ZuggnGnSi9U$TYuf_b@S54j
z5<L$09k-H?<!cfy{!gv+V~Os%hj4=|Q!SVUEB{!Y|3;oSi0-ivjD|SL@BmfiL2jtZ
zkLBx&J((=gsy>#8p5vTdSpBhlZ8!2RsQFkT%P+oSy9HtCTVym9Y&IRo*{JPNi*!G{
zYjr}AD2Es$1*}uVyQx<RIti4h`t3SaYc6)rN)Vh(1<AXv^<#<fkfmTS!Wy*)=A($%
zJQZ?>Y9D^npPMlWym#FsnDp>532sDYy~Rm5H)sR*F~e(TC;%fZN_(XsoZP$xQv6sx
zk;cz}BsNIV`VjHg&POpm^q%QwO2r(3&Xrsc68?VhT{tZyx_Yp9_*i}?L~1rDF<>8s
z1TF{c7UxazMvWnz;r(J)ibFa+XtSCv&KmX?$L3Yd4OqNubE<Y0PWqaD9-_v)7PoKE
zt_LPAL2{!ANOK{&>L0;=Vlh6kdE&7tIQTr+&0<e~P5Q%S{U(b$E-;!zFnSYjwdex$
zJe!11#_bqvKSH_HsN$((32{Aw;m4f-!V0eGqlGsMj^XO6KmaXUkHu&Ai=sCIU)1yy
zwi(ld%{aUEUVJZo%Y{T+nv8dj*>;q`O;KXc#QC4j;gV-EwY7PQV1H?rK9eX29UAji
zrVnT1d22qCD3%u#H)S*=8i$KRa9N8#$=8tPm1@Beg2lr#nN|tkoPx(!3pkNhxD9Kb
zNe9K!SBZM`OlF`t>oK2o<G<Qx^5Jo+3kG11YGamaby0HI#h825Wfw_#tzN&0$adVC
zV&u(<r)*Z|Ea;!~S<5r|py=lf)f!cmpshXQc?yj|M|9`(Wt~FA9Vw5gx5R4p3ZUGE
zY(M}gIn`xD&%Ebab~+}*bhs}S0-aXEst8hRLwmDLMt_LC;9*sC+Hb)r;h8RPm+V|n
zRhPkXrr7lT3HPX^G`JXlVb{1>k6ntPzKu~2I+d7+!Ba#zQfD3#-mU~5o*KYr17Czx
zS5*(LfK+Ayv1?@MpNMS#Og<RU#bht5`PeG%)X9r?$gbaS2akZA!!zl$e&;1y44eke
z7VL&P9vV5@hp}LB23&XI&z|J~0;e!ZxTC=GZ9KyR6U;MZ;jm!0pULMTEUf1rTh&-b
z+=B5=%oi{8ako}n({k^6=Sy}s&}xO=*JIDpGu|xz%-K-ooqzPWZK^V{rrb*Bo(j$p
ziGE8zyG|h<QXMzbg^9F}=MDe1tNKblF7jBwW9{P^WmJ2fpt$Vgi7q_}xs|IP0qAXq
zAb?MbvRU!*yvG~Z3Y8yERPN^LBUTlC!9!zJFL+oV@zPYiK^`H<eFDTOB2Cq*tAmHz
zZBL-&VP`mJ>I9KjSJJNR6%_V7X2$gb-|W$<nm(S-qu00x$u}XGZY<NHpJQ+itx*eO
z4i4k?kLNw*0-tf@g<OywA5R<Rk9J#Upjkj$a!h@@vJPwdK$Kf_;yAh#_43ZY?zACQ
zbVVKvY)pxp<J!1DCWE`L+vLadsp%?^WZrc?`rL<lZ7L8(Qvt#kqT4oi0E0F)2NN)Z
zb}^cs;C#9C@qA1Yco}g~uU!sc4MW{LyJrG5@(d7FB_@M3Fj$(Qc=$(~)mYqejoVY5
z!K@(t(Rf)VyB0`sUSs!W^U)C@HF$wukB(A;g$3c6+)=OFViXRH5Mks-0BhvFgVDGd
zz#4j8)wdjI%A7tW{kRo>B#)Joy}!X%gf4=RZ`f_2OSlJH)%rGd+nt!O<9+?MR-8&7
z>0SNiF^Sc^rhpmdxY`J8lLPGL$MX$CzaH@p71zC3RJdG8XSM{w4XZHg)dyT6H;Yx~
zRU-FIaw#sU%=8M2-eskgiC#}|43t&IZXv5cBJT3aM19c1v7NS~GHu+JCab7SguHZ>
z0Oyqg(A3v$RRJ0k>C;vnpZolYSffT**kPj(MJ;_`SzDQ@i-!?!<vo>I8sUcmhZKAZ
zM{iwJQ+)2dmHB{?I?w7=Azd_n;4}sLZ@yv8LjUk#YpKj93`%!v6<*k({zq%qU{1p6
zbVQ|pnfg2H41{4LkOip?Z|V{2jf$u6vKbJB!G1y>sZzH`(JRC{_^$GZZR62w>8rgq
zSzHE8iO)yDp{1|dRAt@)Ske@O^<Id1Xiwfn(`w8`X&3LW;XLLP1QoJNDqZ@{DZ4C;
zc-Wn@ndrv7*ssL+PB}WWF|#C!OG*^AglezboLV0Ag9_>_!t8@3n-7@nepBkJ=wZDO
zNDp+m!TzG!A-l%5KFZ+kAn2O`2O}F>4rIo#Htr)PLktoFF};UYD&H$IE6guKGWprw
zj#0t=4%uq7BWajyQtt?hHY{6gO}!@flfSaN+B+PAu2WIb5jl%*sL@d(;@PHtuUulg
z-P3RMAei+A(&UYFyWj2jLz`cKq@4XkBF&_EOA2Z$`9$8A|MHjrr<HypQDzhLQTB;U
zvqXO&^@K{?D}2W%5+(9%MGEwZL`BZ(PdHCp8Q@yT22k~hOe|>@d31gvUrxOHK{;K3
zHBsx$f?qJUTR=GhNE*aFTN|k0+>FFF`e@}ksf$9SC72NjIZrQd^`FRx3um~?Rxl&_
zgFiWw$KBZciFDp9eZ}<;EyZDCv)QU{(xd-7)*hXr%eD?x8yO&CVD!({#C9kKgG!Ig
z+}10EIDs(>6p*F=*5w2}q>gc9;9xK4qbuEF<DW>UUf&6u)Svw@{DPeqC^&6y(U>FD
zlYqVuXzYFNQ=hWw_{7Nin|ARNd8;D$^r(I5Rdz}622Zlwmt%ZxBrp@BKrQ0g1Fi%-
zs0CV$-n&vT`Ky`_AlY1W)x+Bm!Et&^UDJprF&)wtWJD|qVNg+}#2B$-Sco0+n(Oq%
zA}vOnhN6A*4R!b?{MF6CHHrKeY*~Hn8?sw5C?2}DqT#+jjU?}Obd!|BP7B|%)z}<7
z2r<56E8Y>Tqm8x}SQ2sLZuHsJi;pxcZcGTua$Jr#)b6v#Y%{7Nu5kA6Ui9*Mm-`j5
z6Sg#(og|{(4nTdA19)87C-Y^>WFne_+x=uBOx*raAdqhOlZk+xyFqYN_Q{|)m+c}l
z`eY(j{EV;%{A9jaN%mIp$wawrDX$`ISmh`4p`Y|9i8H8D<D;IQueu1x1CG?FZ7-UQ
z6k2Y#zFAYea(Twpeli^+gZSU<QJ<tZ=iIK--^bV{do>E0G>r1BtXDO>{+RS8b=7^A
z-_82X;Qq{7ieNF?X;qILWb?j&)*eG^=(mnfCi1DNw@ytCy_{m#9#i{~-fruSdFu0c
zS;RbwAz)t*szU;}p9_bWDS_;v!EplF3#Yf;PGvII2Qj@LX?9+{A3A7L(Mdm&6wc!d
zpUel6r}jdpgOgey-Z)_w1^A@LX}P3=Mi^6!5u%rIeE65X3ZomKJn%OnBmBLKbAmd;
zxe=Qc1ZI@C_|=$+7G#!l`WhecnC*5gaCoiD=3^p|KJvJ_9$j<p;Cy)87NUm71>sf=
zR4|Uiju~>e5fjE_j@ixV+|4DIqkcFK-MT+h^4(j}Qk`D9YPX{dJ?I$YT#Y}R8iCP)
zNLBeHtjk*9b~Gh9x*K(m_YHVti}g?Dlh9K<AgxIeEjD719@Gx0yKTl_+8|PwvfK+$
zCsE@OpKUAtK=w=5iET&c1eS%Y?b);sKig>~&*pm%Gu)J>jEkrA*+iPnT>vk8Hs7{5
zwUMFX*-XaJp+}|J5#?`>kXNDN*?emcFMZo8pUtf7pCH^gZdC%Dd?2c$%;TcqLGx@r
zPk7O%vF{GhuX7q%t-5wjYt-{d*q&$8sX@|ln3B3@^Yk?-rxP3^iJ$_VH;{(ZKbv=X
zQaHPNQBzcS4WGj1XVdX;m}zbiswevfMy*xTRr9j7>(}Q#WgXG*ZTREP_;M56WBN*|
zjBs-k(1zFc+d#}7lkNBJkZO`G+*z{1v-zZx%Z~5q@qoOK7wu$BQnI9H63z>Hm)inH
z^X@)!j-&vp7ovfuR&d?lbj<r7Pq^pvMGbTWrGs4(Fib7&vi|)3-*AS_=r>|~3>SnC
zl#4RZW@Cwv4ij-O#+M>Pz#Yeo^IG$Jg0Kzh;~cE(d{l2wwzhUXz+p#)?SiU~z|JiO
zAdlAPX}s)43=MT&l8fCGZvI%~HN7lg9%#~IZQP1+6P%lv2rJP!UJOYo8Z&(y)^1gO
zx9xAYI~xAd9Mip~-%ux^2KBqSy}i}eqcP?%wxPyEg|<!oe3O8=_q1Qh?y~eX+Y<Ka
zb;jcaW3z4H>!}IcP{XPc?QmI&Cn~8*M2WNuX>V1gm1G;k#a338$rN+H(^ei8-saSU
zTd^HgnJ|xY8dXH~wv^>mg>M%FAXQ}=?;&(Jb5UKD>BbX3yG+Jv)M88<cXtuD)8JOC
z-v`l=+a8U9TW){y;jXJn^!3G>=WTCb(Cq0r>jU9N&M+t4rf7?nh(}{{Rpt>2z1E^m
zCj2?lDxhuu7H77qbgsSps&&M0+mR-9M%#82q|tLS>+6k?I5i9i7~`@b^>NOhCxvKp
zU}<sG#;fvSb3eK+>QPnVT1@QJc_FjwyiKVK4t39&dO=;B<Jg#vQJ*1uxRd+En0v&8
zu9m(eC<j{Xaugv^@~Ad5F~;V>6Luwv;XC}jKda)t6xPbF#u{=Va4)l$=VFEx%Hf(C
zKs=sDab8~vRAScyQ(st*oW7uPY!vYqu|>`B0d76q*`hLr<=FCJtf0B%Es8&CU~&`|
zGi5Ej6_})yTDGF<*y$0`LDSW__g~v;z-^Vd*6zfhU4+SvyH*?nX=RYD-;Lsiexzz$
zwjS%)fNaz@;uD{4@;qs--+96AMFGzY0y=1P?J1AUcHqx>$jQOO&z{Q!!lPJjt|iZ9
zg2Z02+e)8H)Qeo@udM93d@CLt_K@i1Dr3{&l^SJ3U7ljKBEYQtSEji#J_|13ta>id
z#Gz#UTdj|zx3U_+;TU%RgazCET&C$ZMUAO_E*~LiY?JI_&vWU7@Lh)yblr1_R7YR|
z_QvQGSeF~FUUebtM!3)>!P<wk!o6A`-sU`{(Ftpbu8Cjn)`YDw#>0KCDQ*u`Foq_D
zB#A0wnEU*A>eR@-dx0R7c1#~0ABDSBAK-Bg+SLcI89)q0Cnzd#r!NG?T_hA8e=Z#q
ztw(J#I%g>o37(JPCc&}ivY!fI1%wF$b|Ho|b_%)2fK3ZJH_rm@+ZXjkC?g;4rD*JN
z$&d_da1BL;&x}5slCM2x_ECF0si0@0t0NQRzFV$pYB@Eb;O8>;mRy2#QCt%QBoyQB
z25Iv#9^~-K1baQ&^`<0p90zA@L5N@>mEPcs>Ic36RjYpUSb&rRTv2lUbz6?*g5SqQ
zHKy8=ZLY+C`iNH#J!@lip?h4t-)^YyxZR1R;I>Etm^G~^j0z@acLOYu;B5W5bOgJJ
z6Bda%R>wvl=l)p`=df+YpOln?6}T7u_ITj2t@tEM(6?+mK9dU4c4lf#Y=)KW%opha
zr|X^hR(fmD%64Y{4I5^zNoPCCqx!zSLEEu29Rmn*32YW69S1wCa%Vm-4mfXH6@{lU
z52)UmiS_6N@?TEhny8+940j7|pK}(qQG_-XE_r9Bd`?JLCe+0r-1QMgY46Uwi@12f
zIZb{2&O{*%J=PTPI8*=Gn$?xFT-KuBt{}74o$0`sz;LWxKanF^M~uT$8F#8aJPrBo
z^&2uKAF3J*QDs0A+Lmmzhhp4rZ0q{^LLkr5xk`IHmIn)LUZ<1y*8txdS5#uA6eBo_
zo0LrnCqIkOpIi_gN1f-I7O)6zcQID?;1RnN$Rwu@r(~f$ifPG=x=rjH!Zo(#yj_X0
zx8a;V8(-tNfn8NoPb54S1xJbHcP)B>C^4kONFx5?3F6(E<9cx#7coFyh}NDcc`;_A
zfU_OX-Dr#TR7#I^^Zo>)AYsCm1&mt?%8i|XV4pmgY06guCO(CX`nDQ4b=hjbC!`6{
zpbK^<{xVwRJl0}G;5>(&pT%-QFr4eL^?=`;vR!UyYL>n>YMVRry-M8N32X{_>|U%T
zN9{LkOMr-)@ej6L{QVyhG1&8&oK|Mo;w8^#$|MX_I^U2_J)a31lX3{9EShxnFqP#|
z%*zJOVn=*qSIY%c@qDJMmA_%ZaXo)NQOsLSR`tB7+?llM=owC;7-p_UHG43ccs`R}
zWWGbHtR}^$fxHX%L~n14MXD1z7{dKk+H8BH0*4_vPOLtLgIu^^O+}nM1A24R^|0Ao
zqb<*86;qyPd>HXmje2fpj9{QW=3pqL3hIb&JWAP&ozLfUice3uYR8^Wr{z)ELH4SC
zBdXtSS8WVDpUw=liZ=9oUTTinl*q>lF@pVwk!)PE_OD2&8r?K_Fdi1;eEbE5BHgl6
z0{yw2U$P5o_{C0}jsZ22<BZe!VswF<BQam%6YQbyfO?c|<Km+kK^;Nye<ffQO?ziG
z!w@I|m||iG_{HmME+*I|af8GwO^KbZsXrc|r#;@%yr6e=_t<qoPh1YKA8etXPZtd@
zBM!Yq&1$p9-X#FPIz9^PWH)2Gk*Ir237wW>RA)TF$Sw6U-1l#7B}!KQjxBRrB_z7A
zRlyu{(bXOG29wn_gkmV#VCRX$jVAZCHd~Kshmf^w#LfoAe0pz2TeLDmb}we>47Vk0
zc6=Y7;9jup_$DSH6l&FZd%Zn4l!BI(ROekoa9Jd|+)1U?X;1BZXSbDAr#bTgEcl;S
z=YX1?Ux|a2_dsHFb_9%MZtIfl3N>WJ2M<`K01ji_P*t5s{9uDsS7$B&b*|M^$64Uc
z^Ha7viam~CZQzYvM7mw~#9tJFA8nC(fj_go)%jo|BQ8`@AK0`auu{DSLd!PmC;Ys%
z1UxBGxQ+t|3MQtd{XU9_bU?AAIvrA2GeNiNp_e#>J$6iktogk4R_8;s%~cKq@!8_0
z>_dUpn`2gpGJ;{Avhf&y1F<nD^_Ao@|J}|9COAF&icJNogg{0zbD=mDecd+w2XL^2
zicVmlpjIIO&dl-U0Mdw~NuSLK$Vo<bMI$}8^erM83Pj4vK&z{2_37e2n^UVUf~H+l
zEAA^y_4OOk2bXW4U+^&7+`xfVEAdQhF+SgRquU$RSyaFcJ3Qv5q_8nJW4(Gfchaj^
zV4dfePwFk9vh;W3wi2_@*!rg3j*-H=)8VQrwqeeFC;H}5*Kcd_Nf;j(=vc6ZleQi(
zyTNB82IUE%nA^?jd?R>NA95++H^BvPuQ(Bd{wP+zZ|wgQ+YXfc)JJypLZS}l&L7<+
zFXU~s^0)58`a-^oIq6gOdG6j9@|K7Hgl{R2Iv!6ARtBQn@j{--o)JPed4d&cR=kk5
zF);xzTICCgJYRaps-jDdg)XZWKG1X9S*Z1eeCfKD2ETr1)^@*;N%)9zYSl)YQEZ*}
zM6p{LGm1L(^_ef&-WQ4|v+Hc@Ur0yPA3Up?0uIqytXbt?0<E$6u{VyPbAZ7bwF+?C
z8EbzbA8ww97gx-GwvK4-2@*Y?LT5C0ktspe)ckc!F=7so6%Zogg>)`{{);s>6m8%r
z*+B(;C${xn-ml~L%elt~Y%&H9-C>bz=LOmgF5E(PDu#HvkH@iEv$!r+fT@mQG!;1@
zyQpCh&Bw)pP7s#@#7TJa%UaY?KmBI(8wCQ&UX5V_CVn=cqBABA0YZtZ!fJmC-ii;p
zI^mW2=P7{lf?bPo?AOzLK7brrS|olNH~01UlfVAmR?ij!w|+2PR4*>K@>B3e>{PBz
zm;`@G-xQkfywYjQ!X7H-;te<_;k#D?nU%lcWSqC#s!E>7<F=|jMJ!8}=T5Y1=7F>p
zUmg_Yt{QR0VmlSf_I3&kyb(RXTypT(rcj^2bjs1Zr%vFiF~Zsku%3ByTVOGp;?{jJ
zQ^qNzdCWk`i;22dzQTXAR{CP5?t>_%t?b48AjT%KJU$M!*$#au_FqK+<u+!MS!IC1
zCw;)G)CbYVxYe!t#YE(&s79<NS|Uv#T$1g6F`pg8jDsDjeKBqHKCF@U1khe|_Er})
z3m$pO_No(zagbPB^(sYygZ+rBhN85Gujc4BIRzntR&nx>e%D&{1?oOS?ZS8ID(~Ws
z0K+T8Is=9AzAk3ySj^B#Vn(PL!7Z@J&~H5ShixbtwKwc**>44b^mB`gd|V)ox|h98
z#)xr&Bs<lMnPcMll-^HMFWx_WJxqveJ}yKfLiOMv5T67{L?j)bZ}^k%QXqp{B8@Mr
z`!fU1BFqTu=5g{TUr|A8dKt_HkbNG>@2Y0=2)v<~u(>F2@GXBWCe4%7&qrZ$zP}zA
zIN`$@$I_q}?!iE2@x^oxRPVAI(en5Z4E@chDM^Pd3rE*X4co2a2ZC69bu<!=C|uE1
z;MYg1tp*G(YR(<IqnZ=`fL?nsAKBy|8|FzoYIjw<GPiQc)`hu?2@%r2Ya1`-o5<hW
zZdSrQw;3H^;pvg_?rAcP<8#CPuqAjNU)i>`)z7gE_EI7dNEuhxmomlf>w_PJCw?iB
z$jTz}9+w7}y_C7vCity#l`=e>n2}BFcqvmf$1~ops_YMEJQXXWu-W~Rt5irY?hI00
z?8i4I`Dv}@r9_GoW9Mx5OL=A_hGL+#FQv^%Fh<`ZPB(i36Ji2z_Sb1pON+qXD_8~I
zN%g`(jfm`096F9?*8EaFelTxsQNg~Ek(1UMpkMFm;X-d$=L{NMf(N{}j+nJU7t6yH
zYi!Kg0%Ue98sG^r>I_M55xZ{?bz&eIr+m0@$MFXjs3Vam#Kav$!EfWS1bt`wY*J`<
zC6?xVOjR2y<dmdlDr&cIbzBJ8SStD9IUTT*k;6qH1`migPmG(m`J925qpi#{Y$nEa
zpvaM2324MW>9yJT?Cm=T=?b`KKef34h8g41*9fwv!UE2)HXl86Z-Di6Enqii$byE;
z`;{eIjCu@(?RO)9V(Nq!m)(3Rk@E|?Y+1j)^pf2Qq!1br1Uz6M_ksbv9rLxHt+Ohu
z20c&Y9sN!soV93mAo<PQjj4cgQ=sMbSPER9ltO(&IC$PE4e0FwuF^mUwg&?YbzeLn
zw)K_b?p`;m$uytH@mncvWJyiF!}^E(`YWwT)QrBPU}ZI#nz3ndW7+bWOuep7*ij?)
z^Q>p4s;J4NJJyFhqcXrA>N-JjReUGz)2aiKKm2NH(maz`akqZ;oVHrkILs-{=Ghad
z>`Q@6tWE&^Q1Q5IwpZWo>kk_<X7$kkb%#rcnriZe<j0Bo)HMr#=PY4{^>F~>EuOG3
zn6@5Pu3cqNY(tZE#7H=IAg4}Y{9>maQ<vDj6|a_Ff#J$eitvG&bS9m0ue+LjCMAP%
z&)`^4Crmtt+qi(f^hY+SKVxDDL7cD2SLk?O&%kgBEix4|xt~#9h+!X@<Y>g0Du%*o
zFmD&NyyJsJ9loTN{M}VYm(@`VF0>ds27au(M|Xe&4nTMY6Zl($Aq}Ekjgn-BIj1fl
zV5m~RxxER%bRSU+UaqS<&y~kRRuI_1DFFVWT6Isr{oGJJ<{tLrW}xK`o7hqH+H#Z=
zkoZ=7eD`l`MNnWBQ2ppPhW_VlHSjoyT<wmU;FHHLTT`|7|C%VtcVi}x67ZEPVm*M@
ze9ksvH@P8sVsTZx^xxcJ@Ln{=p~IsrZRroNoUoGPOJB9^m^2=)I0)>@wCVIww1zF&
zm3Il|*<{CnJulsrDCKOAmFd?z=dFBKCcXWWeLc2gSK4MZ)&IsS^b=XDRqjfp+pTN4
zYF8!(4}s%`9n~uLE61<TD2EZ?a2&ARyVAKp93;Hb^_zDvjIO%qfbFUi2$rfH{nXw7
z(eH{`x~Pxxez=Jy;opx?IELBGR`Z~Y*0L*~O8b+zvQ@n_5v;F$S3Z<nsfetuBc{%~
zAq$3^=ygV4CwRRb+m-L;!$e*R*qD29_zVa(>4I%2ieRCL4p0cpI3&>K!X1ytWS{aZ
zE|cnq*V?yvubmIvQjA3eOoENZ$Sy<~qE(MhJT0V-?r&({Zx;jXSl{U`yA+l8@!Vqr
zUygDZSAz1#=gJjC0#^ho-J&+D)oK5e*8O%h2F}ej7mY9x9I|Usw)5ZEyjJ6-V|HE5
zySoOkwAhu;8-G(v7_z8ljmT<vgKB)ZM%hjM^n~BbF^z;R;)QoBMgzygHntV@vGN_l
z7~NL8u2CqnIGO+6nMJ!3OZfT-q#4@@wH>0vUb`E^X@IJ1Jvu#=((P<$M2)Z6rh0q-
zKS0&@qERww-HN@_4Tr(W>pF>TVbq52l)ap2sTa9;$;)|5wh{;EoRz+u2ns0zhK*MC
za^96K7%Rd$l&j?*zq#LbsHR6)$G4?IaNJZT#s8>$Iql8iG*eajfP|c<4Vb4elSaf-
zBbaX<Y_i>gNktI1+9H}RE>b}X$h=^6g6G!{v6KQH25|%xuzEG?;x24@IUfkT*jXg3
zj9<>D3hF#Up#^5g>aF$VbQpBGGP(WbEUzQq8x}fDj-Qqt@#mwSvMXmD>QqZCrpc9Y
zA0N&{G;F;^u&$ULcsZYaE(zg|de??tPG{fl$E_e_JbB}wjmK0FecdLb-_fP7_PY!<
z$-2%*Uv9;T^lmDK$MH*uQFTr9r9qpHx<js7B-D}$c_ShEjdnZ$kr$~brKg<{e&Z+Y
zN>oCkddg;Fa*umF3s&-KR5;@ihvs5$^^CZp{90fzK9a(D&8wSc_msOH(0E?(B7jVb
z8!T-pTMXEz{3Uba<$SYvh#8JdHz2$j9X2A*9<$|W)9!Aow=|NbR$GaR9$NwS_T_vc
z9L8rtzK^TLEr;zEe8ipNjJOrvT7bpH7<W5??dC3x0=%M$xD=n=gPd-gFTYpxBI-R$
zB84mgj&*KAsj?MK_D#SMZ%1J-S?lbTv_FY)Omxv#GS51U`%UR9X%Fphw6a$c-R;Ao
z9&wRAAg8XAwnHu4krM`^;+0I<sD?buROKsq3P$^m^91l~x$2cnEC(iV#;cAt7_@oH
zg&HA*#2u%*K<Gr@39Ef2oiMx%ZO<#21}Bo!52_heO(NKD+g??zdFj(uAN?UFz+&;0
zd^WuX8$%RB)g<nUwFJ<9Y}y>xR^c*u9Q{bUD)tk~p+hiS_v`8s>r}OFcji8(?%e;w
zZ4!H5dGBetr2nC&*je#pkv9;?Vd{nhdwjYHxE12F*Da67W2&6{<H(&<zY{J{J1?A^
zLrT*;rDbwe#%a5tis+%gY|{dYSOnJ`a@)L;H_&|*Afy|!bjT&dc3E8~+tU`emzoho
z*I(L|Sf%}_<Yxov6x!!%ab4U3alz(<)x<DDSi2TGspW{xYq#}v*>$xTOG%DwK^?*R
zp0h>$b}@$CxZlkTH{ead;_&A{T@LKq20hczEw$_S=&vj4=hR8NU7Vp5kZSdnd^9JS
zF>i)s-F`=P@IqkcuLZtKe_?l{54b`O<n`D*NQsE7fZbwDOCtt$n^E4#jABV{_g=}z
zb_(6&H^Ei_#UGzV-7XG({Qo8EeS)jLvOLeTV4(#aLBuc-ELc#KvMqPXSWe5fY|FNU
zo<M*E&(Z^ATiE*Z1QPmNNCH9P4waLPQcV$+R5jB@PBC3nwJKUUO?If9WR@~Z=~1*~
zhZb7M5iD4+ASzgB!GZ+~7A)p-?)Tk;W13*0-@E_L{d3Me_x#1?#A?%#Cnj7H4{<&2
zsI6Qr(9NpajCxoTx@G9j+Jv*nDwy;xeGp3rxOUfOEGH|q?WxV^xb<J~0NPud(W68p
zWFDiHG29TdT4Qv0I^3xVBC^(~TS!5%!)jIYav8IsI(5qB;c{!Mk1<;P>@=umOIvfN
zHP$8~^WH1gq+c=Ahv8MnFfS}hTWww^(q;4fL+I09n<&96XI{0UW`n5EV_TMDq!gaA
zTlL&)xJTWSwgP#sy1$tGLCUcllSrV3RK&BC`(RXy0k)!JaOse~4^;lybj<&i9-Gxz
z|MTkub~QG`WtJfFoCfE?O~|F%e93^t2osY`Z9bNrkzv%Yt2578cq692bt}=3Z|r7(
z+{Y8+R*aVNI$-_(vKVIbP854^5RUPJfa!2O=%QvoaA`=vT@5<IIxf8&^luVNv26FE
z*Kw|m<@ieZ;bPedAC9eq3gLO&%yd2GMsgz37;dN(!MLljHUqpCjtgSYErGZ8TP}zD
zf`}KVJ&0u+$7|x8IH!lP0YxH=#s{wZk83kNZ{10|hw&lwXS>oN2jYpwxq?ZjJ&oBE
z3%CFTz?ei}MDkpi=fV$eV1I<a(0s=R_1?GF64H2_40zQg+woe$DTE4%BdK~VBispa
zap!9ZYY9!qU8;&85G(^>B~(#C%x#Y<VxWkONF(^A>rAxmi!RT?B2~YZkf^28YF<lN
z--lpg*lMG95Q(SgYYAzPnOS|MYa}Dq5M8z6<=Ch%NUZDYo74k*#i(OqK|NgIE^gDG
zXSqS!RSC1H<Zw}ZEmNXojf0_~1|tFv;@be}n2Wr7RD?hqq}1PtDtLDa3OsI-xxY$M
zB%{&jeRwMI9Fqcskg7HvWf&-U+{syWgIBp{_PVNW&SFz!a{&izO4+VOOZ3=%t`3bK
zP%t_}eLa3i#^Y#MV%gmL<mPLc9~l@XfFpeBEx|=(P`j;x$v#7HmzPJy$%-vhX6k;j
zFt@SIM|hvl+Ff<h+;Ys8w6k!Q(35+Wwek><%K;ghKG$EY#N5!PAQg)8B<X~qt@H$;
z9kh)Yad4KvHMXe{U;VPjwp3Jr33eT_-LIUjX{@sB!E2eb<DKg94mgw#qpvY+I=L_7
zyEPdVKaM49!*^*rN;@#F7_leno+JMK-`Z1wvHlC}faCZ52kd!FW{ZHuI1jvdEmKo}
zQe8;4du9g`!m)L(dfRaz-yd#Dn;5pL0~vZ(1WYI4;LiBXNiaWh;CCHJ=mbn;JKudE
z-;~XM2knWn7NVscNVs*9cmBQu`TP?H4z|~-59BNL(%>M9ubQafncYE(TJ^+`oXom_
zmb<~tpK#F8bb#VG$<-PTB>KS}^r|&#*x@S^))b=*Tz1qg#J77YqH|Fv6^^w%Mqx9{
z;&nAWr)X~PQ;PPJuuX`LV)ldM=O?U3P;{O+S+w4$%FD&EHzw#8HuPyomC?hYCXJDa
zfWc|zK{**K*>lFGqp?5xzyF!ds&Ubo>#M4XhTCyi5KW9<_B3~PEdV?`OvtGD=oH@R
zdVCys-EQckwX^d?r`^<W&QNoDP%z!n57Q`K5I%85-d1%IlVT}-M<2u?@ryJ@Oo9Ct
zW1BEjw!8XjQ%amAZ4?%Cc2Dy`J4Ln&{oFA)W-GBSE!cOH3Sit;59G7$?p77s>UA}l
z!h>`p#^<?;U2u$l5QWWFY|1chdl&5fflQ%Ns83GFgQ!2Y{&)5;mWXKfj|3wO{GZw5
z0E`X!#$5(&JNj-u=jJU>V$X<5<aGUMK#qR|Mt_ONc_zrQzq04*Kcy*q5ny!U@NKVW
zBqRG7fkt0Xcr}Ukrt0;KM&pQ$u||45->R1|w%n!C#1^<ZYWM3I>D_)MMGD%Z1_TX)
zhT7iO^LhYb$`P|K2EaRS5_|sm?z=bvS5p&R62*q-P8{so_=%PI=kF25S>5ZIv?%?D
z`>)nVPyLAZ&ssx({u;p+8e_^GeF$KiR2kACb!*}EgzxUKWNpzAUZUqP34T4(nnT0q
z+~~X*eeuZP>eVEZ)q!IBcSnUXG9T-S4q;X}?7gunu%79CSbjZ`Kq7EJ1cstIK4W0O
zs3zjEOSs!6qwm(jk&;ada7>?nW3#U(4yB`HS5+yc27$KCMH}7}<VTJHCTDOvvUz<M
z*dtgz_+p)h8RAIa(4XUj<Ve<^1aP)nu_6#@&rNb$WhXmZ?T)4bHv{!p5a?LMdmQ`4
z00Yb$v%4ysnqFIgfnAD1SU?b1J-Y5<S@qCsc$X_$ydvJ0Nv?{Iot9wcA~a|{mbMR>
zvMbd#qTjK#Un2T_{kvTf6`b@a8-(s-#vHOOt=(CS`qf2&;0@ePezG1^02TZ2FjfzG
zh9&GHee>Z0Y3z`+030m1lI>V2th38l7(7w!ikfiAo(90#MdO*KgN^mDJ+CZ3YK9lD
zClP`Z2X5D;@2^2Oer{Hmsh9NMQ2Hs`QJ0aJy8yLmRb4t=L_S3{v$HPaEN5T2i+0te
z!|#FYY<FEI4Q0PI_tYi)_sa>}?Y+?!1p?}2Vp8vmu{=KKgjH)SuF?13x0<?){$z7N
ztHUj<{`{7O8>v&}(nRu*RZmc0x!p9>C92fi(rS%$83VblhtFbLD2RtiS7`-xLSh*g
zG*t?)$+z~pOeFLMbQygy!0qtkVWohlnHY=R@zu!@?1`ly_zKctjl(Sf<tqn72qN6t
zXh<EiPNT_8VKge>8tLSFGJbHgXQ;?@bU%)pzRd<?$CKp!s%A1Uv_`Ne{f*jo*seu0
zQdsxdd~6A90E+{Mcs+h<Av3&FIJ=<^&?Bzc%@}BagexF(Ou88pp4(B1q6=DeCpvRJ
zirgibJp=?P+~;DnFK}Jn)s|-g3RGz+%J*0Q#O`TV$(Ccwu_7>;Zi}-LKQv?8kH6lk
zI+JQ533JzD?EcH&eq|f_Bq@}?@oh8C8{QBphx1lt<;Zww_cgU70d+gq2Lk2b$;0+g
zdkbR)DR>mU;9!6c`Z;?Xqo$j9JK#i<=!^6uTEM3xl_IxXEZfwK+wIZqGmXK~U%O_{
z1sCUvSpCuaF=&X{8yRU1y1N5tx8sd`{QV%Z166M%)cE!rF6cKBZkKpUqT`8*Q-dhQ
zxuD~Rk?FNxL(2B3IzZs_us2Gfgk#(j``$=|!`;b-U6Ee>MkWT$9%rp4AnU9+G}T5A
zg$bOJu<uukBx?iR>fcB=((#klAdou<f7kd%A~(42V-=yl*+2{;-pG^*0~6N^kZ)P7
zzI?UE+EtSr5Try8KyX}e6Gbb%k?2(CAOb~nJpp%{do}i`kV$u<AnuK?4)~37R6Xnl
zbVHhuE)N$%j0%Ejw1Ja?fD8a+z0o>i6$%c{st;6kkk}Es8jFITlgkn2DxiA+^J`I<
zl2l+~s6tNVZ|!=3)`nl_jaU!pmq$0i!cfrkO%Ys1e-f?GZVQ;c^i;p|M#yAfEZuT9
zMHXGq*0KhpGbjmc@r`_Ol(DSyI7%GRyMo1|pxKfpwf1|rl|)_~>om%BZOfJUPm$=-
zRsvAZM-S>R*=n3K@=L78b`=LN;DLpDCsyGgwt{37+SJHLCs6WkMRhm<IC37vn(Mb;
z;V7$)U1ULs{+QI{lt(^!6zg|-h}huzLRs;1qSI~11PJ(f!kz>K#lZ=u=ucx3c+_}S
zHH;>r2{|mz+4E?^W_e!87aBnEJj{Nak;T(&+KwOR(|l_9f=h~ssTegn-Ggdpe71OG
z++9CTNc|<=I=iFfuqTt;^W%*F80W%9+uk3i(+<&eF!n`lM8=n_IzEd-m_^t8IKQq7
zC#*I;_P=IzKhF1KC3vMHtSX@MgL`lIaY9rN%Vv#&4z+|eX>^WKVW!^-YJqzL*wUsV
zT<OpeSo@C?_Ls5b|8eF>AWIysl6qTP;m*|{-rjmtjdPqL8d~p<(;;nfF4^;mVFUrj
z2G|7nfnlG7P7DP|Cs4VL#tJwEwaF-K!3<<NzKFeJvjLd%+88v&0<Ha(%|$;h^R;Vg
ziN4aM7W10IX`Z*&)rgw{9J&$x<DhIeH9J=j*e(5i<oKZzc3ZVj*Z56zr?ToW_~fP1
z2iy#-`C`E5CW%+<Zh(ZRAQEJC2@{XFN`H@g82IIw#_<!j5)cu_$W~Rv!w-<w<75#)
zsZ8mHW>vHQN48l3@oTIEzOpR=Lgf3e5b%6s_oF3G20tth0#biPkd#WzIa;Ie<&Odz
zlIU)X)A3kS!IO5<iR8B6x<Igav^p`(uC*2PukqOjNw8-EvUw5)IlxHB@H>0)<9vH6
z&IOiCZzcr5shN-oJKoI2PHH>`N!6PP?S=<OZ0DO9gE5cfHhnW;8Rz^Mcj4|g^AYX`
z#9`YLm4`f__g;-c<%<Z8)#n*v<yEVhlm^_nHR_b-3vS3(`)0y0<PgJFSNVGihOfTz
zH*PT2pubNz578JPt^ET*oB|{)^;z;lfYgHM#M%VK)Zmb{znM`Tu@m|N7a$ei%v1%F
zD(-(Psk>xx<?L8@bl2tlYLDuB=mNO~^#(8{7ND#zj)#i_hcxyyhIylJe)qrW#kF+<
zDen(cb5B-A)URLJv=%YhSKI8HiF%xvw5u`DD0xoj-pqt`>vt7V$+g&3vazOE1M}($
z9|R!39_w`Y!UPJB8|n|#`vXs?o2u$YuHa~;=nu(;^Y+Vuu~@qkqkt>eUSMb#<rlHr
zT8v@F`k=&iR}=W)Gh5OYqU_^g8{qqk5fvp5+`LhLYAagvlsE*OUX99ANq_>LtiPF^
zH@E(43bnoT{of|;dE?FWvQ6QQD}ML<{*7&Fa3lo3vn_S+hNABCdp~ygh%;snRC8tw
zOS6Z9=`s*Mih1>OZQ(o@vpzU6YuizZX$cI~lPK+i+CEj6n23>AE=D6TH{vNZ9KM*O
z6j<=bc}!xS|HG_4b29Eu#76qViPvp?M&U7%*ExG$Ri9DQ&7=v7TA%T_xs04)S2Tdh
zaXFdouFppwQIQ*j?WxbGewat$-uiU%S`XX4AN-E}Q+>kL3u}v9t1$+Ttg_YC=UWzy
zZC!nwP?w*(W}{wRzV;{9P@m{cD<-Fn>Mz+sH3dkBAw9TBA%^zEVr?-Tv~|?lqi|sA
zq7~H-?9n+ZML*asxIs|23eU01))NcGQ;wzVt^aN}sOdp!|CeJ^jv=lPQfVlrl+$C_
zXe_IH=9D3&lL9gmnCbfODpoT;*gu<%eoHL%Reh0v)D_Ef0sJY1*1W2(1;88t*zWmQ
zhLVTMx*i~S1`@aExuJb<cUtnNM|V6oT?T$jYk?Of_7u?;;{_Mn*&US<!lU07Vj#Fd
z9O$Aa)VnU&yc_EwYcO0cm!b}u)Y0K9b}vT2(>)#C<){L|9e~|fi66MvF7m@_eGqD7
z8*WyDDk)$=#yVyj(FRo%?7eO3L`{&rcT0bg8prO}X9hoo0~g7D5M6mbUwinnD{gt{
z!y_#L;G^6P@I9ih$4cDRA%11QJ&A^YI+J9SPh+E8L;&loeL!9cem;*aZ+6QleN`Te
z_EyGb0*xi6e#cuGIR^;*#IRNILqD2sxPzT<<wYV7Fs23*<Qv@eR>okGm)q`WLc9t$
zj@a{7!nzsm(7mdu5ID9^H8GtXoW`dJB!8>IF7)A7UZZ9`UshXfbm@vKSi8FTp$j3p
z8(?TEhybz%yWy=&h@9|PV|=pPm>+8-SO&PQf}lc(zxHdySSliTIw;+)A|mfQr76CZ
zh*QlUS?R4zX1M;9=crpC@CZwBS9l!sXa=JgW%jCNvCqnHB|<jo{Lj!^nRv3tQ1{V*
z>JXu#VpUyOHvMwOU=VF*V?rrF=GFK_%{gXsZ^gy$f&Gc4R<X_^(}zHL^*`8r3^zz@
z)fn#IZhmDq;_DnY$xU^hVncx_w=@OvB3-uI0T&Om#80cxT~};De;;%Ioy9n8cs`Bz
z4DPCwchtYyQgq|*?Li_>sPH7225mWjDPh2}QrUz9wi@tPU$ymEfRf7{HezYpQ^?&-
z_2=<6Z0oIjxg|DWm~^=Qs<^KnJO-u<v>pV&#6R$iXZA4QKwfN*)FXk4kUIw?XbZQa
z8nH<2iNO7Ezda3*&Ze|L&)&*ZqGC%oj9=69Xxxqv?uAa!6&PfDJDnQ@gLu}x9Z2Y6
zL4YTY%-b26UFGWY+4*+DCgpzK^>#YCmx%4Ais_U}+qLKId<-Yyh9_)qd?V%OYS<TD
z%VdpEjBjT;0Uh*raE+Q=BqU}s4Xb@S&rRnJtt^Qp-_A(G4KaxQ0-4kWBcX!f6&v18
zbR?<o;J5V!ijE`OY>FwT2XhKvQP7eM&vGW)0_LGXPw(FTb|N}wS*>DpbP;JsDJo9D
z0TSd*lk4#OqGKt7T~E|#8yx3J+gn+AcUvf{FO-CCl|B?T@g4w4N8io|%Y`wpAHPl(
zf|G&;VYN-`&m*NzZ1(MZMf;IQCC{P|LMB%B7<Lxee9UQulHY_~)9~c7axb0vn3&t=
zU5_t<z1R%_4P7UIOr@1mrhm3uZ)d{T;o|Arg6Jsb;&)z__$|E<pj0xMFRD)*HgFga
zSnSTp5nK8}O)ipYyQfkjnQ&h(s~J%iaa)g?Xf53WbhQGfn=Y?c;QXtkXWUR9EDjcR
zo3YZ#oyxXUh_$vOZp{49AZ~VX$_O1bjgd5V8!<7MlUaMDF)ytB1DLD7F@{EA6fG0O
z@2apTTAe8(XFZKl=$sp&FWa;D$(?4w>Pz-K)}{J0d+~DI6I~^<ck(^if+%N)zT9R7
zIaAd;>AcqLf8BPzld)TIDV@#U^-e||LjS=)LXCF6lM$YHRq!gvs@};vJLx%`_Qr7T
zbi7Yr#mih*NSEKqDDO-aEOj(ZQMYSV?Di!tZ?n1>9!(&kABL}w7MM8}tsy=!9Uy%9
zJBi#NxovtUbLZttR?sZ_5Nfu)lSxPmS<s-^?HZGK2BRn-#;vHjYyS_LB6TiDMo{*h
zjAk;4Ny<VTWn-k*@V|@hk4#<Y<5-k6|M7`${|>YMp?5NUN)H}Sf%z{s8b88V`X!zm
zW@GBZM1)OdXVWp)Lw;;#W2nROjlLR-J56T65u1w&XJ=9dm}~KaCwRkecRpHRzDbMg
z0+tZNE|IyRm3ZZqSL|jC-Q|{tx8kVyP3C4WNTk&k5fgD?>`s912BZY6CO}236B7AF
zef%kYkl06Rpc5xcwxnLb?{*k~d(q_6OSY^JiJDfT-YFJ&HRh3Gk=uHlQQSCOPh=Z0
zfm3dBuo)E+r_5?@35Is4$bI#VP|aQ8A4E%3HH2x5f$`k7NAYzOuI=$VnY7{-MvN;-
z4Y#9%LmsdEi3UP5ZBJuAP=2_{;IpWj{6n6{v^w1Y$X=-F7hlwvHDoNju=Yz_NqHkT
z<bz$nkr*{?RYO8cR2Lo$({@G+{5aD^y{jRih%>U6ZFfV)bhn>Jr1OpKQ3vcrvL@QA
z>U<vKZf;0)W7r)HqS}RFywPeJGKwEenR;tu+>1PJoc7j517ai&SbYGFS*t%J8r1jY
zF)~<2)x#GC&skGL#=we_<EI4RUABT6oWr-eO@E%~?67tLlBlK|ZWkNA+ri5hYez6U
zfaKEVcyy~-5htUb7?SrAA>5)SZXxLX%JJ=*LU?KQI_7uUXhYh6_iTpwO^yiM){_m1
zCY@@v>4r>*i?02RuRh`+UX6awrm(PcDup6A#lIFn`7_t%qcRo)j<?zx^u8S58yfHO
zh}~2liH{_&b2RGWu}XvrHEMA#!0uGe!T@g(TZpqzbaM$8?JmYH@@7MNOA>Z<4mohG
z-BNV!_=aRU0G_(S`Yy*FBTnkWB6mfti81O=>(z#Ap&>et_9>mvbwQ=mzY#-BT^fef
z+N`XQdmR12wqn2$pWJ=5<ggOu_JKamdq7Nk*pRQeiGJ6t_;dS_HjdhiPoQ{LG2+}8
z1Fqe6PmpaL68A)XqS4@gk~hOs4dA>KGB3g9J+aHrW8ruJCac63>c+#z*}LgT65#<K
zp?5PzLla35hwH(+>9{0`A5kmx1&lm?ZP&XQ%i!tJuI+v|&#Kd1-IAr@o@g;Xh$ZE_
z31g37Te0umd|^Wx$E-T)ku}9?qGz78YhPP!REBSzu{u>wmS^0?QMozUNHpjhIrqhM
zg*3jKi4rO1oM&p%jIm`y+$D%6Fx_l>H{aGHJbpD|Qc|)FMfHgR@w8;t>H=F#@K$B+
z7mHkAz<N}MD<7B}(<-}PUrYnXo3ahXY9FciSdB*YlPKsX-^~}bpGXXN{g6)wl*JjF
zeK(<Xfp}2SHF*EmHWzbFu5L)QG?r^C(kuO*%|{jX+6C>`HF$E{CqsLHQbq=SQ*%PD
zf_G31g=6}--Hut1Sqp6%_ij``^8U3g#IZ;*ER*1KF&cHb+XvE?yOoU~GwG7r`vzj$
zZueq;v0=yIdpW*4dC?BqiXi2>@Q1;wzKZa6^tv8Pm}dQZq)lJjhL$iLaoY@FiTi*D
z(YyKj?0o4qY4`POB7`2BUD0J9R5scJ)I5yUObNn~UTZ#IepzFWqYT6Elo2R?AVr5W
z=}%%)_rGsX)zY1>o@s=&dyMcr>RiDpOLfZJNA_OELQsQv<vZRJj$!Fi^<Jj+ZY;O+
zy^L|FfpJezy#b;y?T#+oi~;NC_Yz4!>&b)oy@Wky@d(vpd!IUW(FK=6^?MoTi9do{
zpugj4)ZA}yt5t2RtBBrdbpgT<DIBf-y@YWmxeB;(1K2Lt`@R?I28>d9NdMHCaWp(l
z0V^(|Tp9)SF@=7goA<qh5A7}1{$8eJ#K?qm#obR8QE$R1mI9EJk)u00Et8~%`}e)f
zh;Y%!PP<oKcbBZJ$`yUpk_~CPFr7FRsb3bDH#I(fYBVxn)0K6QL~~Ya-%4*+)w;R8
ztJUTLs*A2?yQbDfJaJXv<lK|mT-Q{HzlkmIjrTH@fDa*VJY31L-PEo@sFf_hZdHKx
zY*wxqxE<@myFcA?chu>*luvXawiVabK0+}d#b1oyltRuI^{$}5Fi1M8rGWk@!2#`F
zfZ%Cs*6!G{mf#3M5tzeDY<D$D-&X@1UKQko$D&d_dlnzti0wc)N?0d|)uz_wl^WZM
zjwZ25xvyC;@nghgjBQ1W%?us}NEMH=N6~fwdz#15_7t&HHNcMuaiIl10*m7D0-bZk
zY$LkDG}vRuJ&V~FNOoe+<FgH09eWX<m{I=J?EQ?2V{5;MZGJ!DTgk29tKQEC7cT3f
z54H3Cd>E%L6Cux^=DXB>xWe7<em~!E{-SfP`TdNQByn;#1D8KXHz^Mdn_~OkPxJxj
zao18*zn=+-OM@_p4oF7Uev2T2Jhp-pc_7Lr{SCW;^~U>|Xc8+D2GA|SL1^{PR`g*7
z*BDbq-h=Z|d}82DY~=VP&cn2|#V2WYaY72vNvGXE;Z8lp_Y;*P0H~D$Zq6>w5ZzIm
zjUl=^>)jLU0RvQVXz7hH+7JU+S<t%bl3)x&?<dOS;eKGLV|nq81nF#2gC9SE9kl)&
zgQm>BpNP+yFYT(j_RzsY(6Tu#74kvu;rFF2b@vB1r~L}c20uRYssq14J|GQ_;R!N?
z8!<a4u9!o|sur31t(Yrv=EO4E?SS^QiyFC(1r7Uxu%5W*>Hry?--U}Bi%sgSvAZ!A
z4|7^B#S*!-sojg7E9~HMWzGDp@s+KpwZA-2aDQd1D#bw9Eqd286iKCO8!?TG6PIl>
zpg^3m73Z#%$ogyl+-LU#j;^);I$;mAR_E6i5N}nsf>ZKHwFp5-tkui*SR=4w{(|0q
zKT}Co6;}|f6iGB4K#a{?wx<GO?H@eG;j=i>a5U2?`YNdST%bW{IVQ<l;zeu#CyS6C
zSQA-eIy>Z0W{4e)>716~+^QNgmb&PK<=xqs(Ej`DJPda=CW_#JYy6eHyD?J%NKV4e
zJ&hSV$&7~BeQ)&EJW6C5gk0(^#Q>>RZ;pCM0j;Jn^8xOETN{NQ?cM6s4x@%YSiL}}
z-TptehQ>^H#`}nH<Cd6>FMk?@kMk7GO@h=#-Q3X#&8!fkU4q86X#tA77usVAJfsH+
z2p3b}jJUB^Np<@{$L@fntHpX^*zi~D)gUBn^eA`bsC~N4h8pu$r}P*~?KB#b-ug!x
zOvcbtQYlSGIg7%(SmjdIc__-O0*zOZ&1u2#g&~+t>{Aih9k%8+AHPh^xH%~?lcQ@q
zak;M>6PX~A%FW6g{CQ1^Fa2I!W48l1q7=FQ=#B>WJU9z#1FPypcu@uK@Bh&5#;p5X
zn&7^vB-B)S5uw(-SfcZ|CdC%DalNi+<jYrVH9$S+!Cu$n6V){#rUEX{(!rtKRQHp}
zJhs&Rr9RIGbw9>x{Zo6OPFzUm61az2y%Z<(kq!hjbzr0)n~ujZ0G7C5zHO^AtQEFd
zQHTlvHpZT6bmujm>Cf5OS$iH+JB;h*i})JFZOlGMm^nP{D7NE+gnd_B=2fM?SvRiR
z5AxlNxEVHCMI&xWx4Y6FLOf`Dq8{i>cH0j^EaaKZf9<Lv4%$978G%DWb)*l_>nHD7
z%?Al7VO$-1Yd^^7eb}FoR`)@^Qj-#(!h6@N4ILCtbXY?`K*(rVW<iR#CQsZCGVWrD
zT=DPK2;hqZ_-!9#5(6omcITvaf!FE&8CKL#1sAWwFUP=xM78eLNXZeXCxGDQzi7QN
zMf@zv(Q0b_r#AFKB8BHMF&+)@(4i@a?c@iUI=XfQXr|SZ$oFh4Pm<l(RTU1qx0ucS
zpef<3iOv|~UR?X7&Bs>@1o?XObllAg?1tu&bR{=cle6uD|6BU<N1un5(sCspn~lYQ
z>yGAAhNLb;r<V~Q;(4!5-Ezq8#yU%vipx;|bPVjfr@BsIZ8`c$e(<)U?}{A@W*S;-
z7`Evl?tpbQ#TEhK@&}pat7Knf=*<`!F$%|mR6tQn@wA}oc0Ybg=J@F`@IXBivu)TO
zzC78kPbb?d5ACn*k!Dee<NR2QapHKZZEJ6khme2#iGC+mwmsFK7ucX@acGj`*Yi9r
zN!(tjgE3sHF{Au2pJZM?WG@3(c6^vI6wj0kRu!K^9%gjshw*ZRN&qbAn2gYNeV7lt
zOBP%MyZgh0f?@}3kIJ30fx2#@xA((Ds;4j7KGg*A*bA%vP$!uiu_ii2d6D?++7A<5
z`SQ!3^jMubIC#P81IRWvlx<Ld@Jc{ujUQ&@ClDK$+@${ZpQ*J%4E&=He`Ia37*|LU
zi%baKIDSh0w(g}={4k?Z^C%>ZV94=9daQh7-2qLqH|!CZ2oHK=xDtJr^?PTR4QZNq
ze>rhm0Xs$mX5tF+o|}x8Z3u^_HGqeR@Mq#|tdhHn5q;3EYKl(k<~~eRnYI0_$F2pa
zUCuV*tgn8ZNyb*<dH|002<=dOPP$0?W_-4~A=It-BsUkzp#WlvjOT;+=-tuU;lvVY
z0TyB!7YEPT;>#xFadn%?yYY*wJzX2T6r<p81v+asYb!3TU5?(mxPDe*ZqPi~e_M^F
z(ELeTkG|1}yDZH%w8$yeg42i1s4+NBRw{nk(g?5vaEW$bXOHYA*l@&XI1{>o{=*n;
z@&YdUj{-LCdV3tTv3mE{;C5w$`Z42(vS!cdja_s!9iJStXHlHAnFP9tqj8v5Hvu&S
z73h4y>?1un@#}W(_@iLe{+^TNjL>ha>Z62R<Lf`Oogd|+aeT&heU#B2A4#H9eUwi*
z{Jlpg*b|kT_S;@{;`D#tM+qU1aFJT|N12)=<&Bg18bQSIJ4}UIwR`>516HU0{2A6~
z^&jQyU29NT@CG$@7gkt;Mm7HO%P*}dfIUA$4$?w=Zv7YR5ZXS<#M;j-z62mpd(3C;
zFAy=1WL|I+Wx8yoDBSv+#4UDzlt{)Af8lw4xSp8BRNsvC>dTF;F@P;c*(5U@(zH-7
z!1#;?T!cOuv&r~8>~hNKj}rasXt7y!;rXzE?CM9El8y`xro7K{0*6K9nn9^>?d3W<
z|Lro5`Pl1(%Wgyc2fH52>Uk<X$3?n+Zo~k{lwr!S<IvLLv0}FZ$f+Te8@DTS<%~np
z@5CW!J#GuJYVgY*Pi`@Oz^=KU8Qu*zg9%-V?nzP)J_jhSK=!A=a#X+-6<bPmQATEh
zw^Vv2t!n3Ri8ycT`Ws&ZeEMH*1k`F%CB&IS^BaAl3BZUl;4Jl}sKO88z;wZ>pjvk=
zu-Xrgp{98hJ)ffI$59)j$gBZY?L+>ac%t2ME}G3*wx=<A{{*f(&th~UZ28fA9=qFu
z41>UsG1_T-H>@e)*V;eQ^NyyBu&3}lYRVXz0#hv6&Zh5jJUiGpY^Kv@?^1it76w{P
z`K5Y{_uwA&hIbo0)!vsy9D;rN8eaQvE)l41N(c3WUaN^QNvv@Ju~)77aKjzs&ry0|
zFnKi9H|3iby9<n58w3lXYI*EN8Ae55cT)gR-SZPGM5*hP5Ir|#oInsT?eD(!7%Zh-
zgr$xjItFKl33OL#$`lPB7$VO_muLGBA6t)#C&{egraQfw2@`>imZL9JGGvpmAwdUg
zh_u^ARXpU+EcZ0GNi|;km7Ca3H)Sf=Nl0I;FA-j8;@Hki*wvWmL6=^|S3j4?HBFP}
z{&kMe^7lWpd9@sL5%2YwQ2Pvw`Hg6b376e$%0#aPD@7d4&;Z<u8WU*CZZ{>$k4LfH
zQLk(YZ{7e`zUrX)Vl36cBo4eA14CSvZ7Dv>ykG2|M)3z3G-o*qTQD=TmH5OuXvS6r
zLxE=NS~H%acvo(y4=kbmcHGn#^zKx1yKM>BE|0@}KOj2-t{{Wf(L~7gP~Fj{G-HqA
zxUjbuU9tCA#fK0pOxbn-)i<^FD|_;?Q%_89Ph%<F%0hYSpXD9Iu<v<*cN{L}MVu9~
zNqZ1oYtH-MHaLu#ACz%Nb6WaAXi-&j+NUEpl>60}aCB#L#uF49q`lqMoKHJiG)z=*
z;8zEIMDp9yoR2lxAc(l!oaloSciX30m;5T$&+s<Hn$yo`UVYVSn=?y*Ct4<6b<LTQ
zkgUlaC+Y(pcRoed9PnUpd(M4z8&w0lDVK-vBGz#5OnCe@Co1EC(cH%&+PW!-Z$-P>
zA|?Qx_$~^(X<{0b)Rax!zrWVH1B~V??)TgiU8VH)EM{+WqD)<5R<3k{&(e@ONme0$
zu+f+tvB5m??__Kbm!j5xI_h|+W=tbz1CDcq4Z7N#$SyAn1Wa@KaNQ`WuC*5~h;k!?
z`RH0^er`J%yF~y5BGL)Wj^=K=sg7J$*>AU0P4)_yPBsTA_h<-7+0X4;fdi6)pjj&p
zpMKUB^|!lC+TB>Raag0vwiG-5{k8D(yQldPh2ABl%K;-oZW4T{x9@)eTd^7oQGnN5
z*K)UAv5go7pEoG`rbfvrAhoPp>K8Hq!;R6ETDNxCgP5M1*}1^T+3$xM%hhlY#z(Oy
z{xreH_pyddY%E-lZ3k$_Ipa^%r9Zfw^{KRZDrDn4q3P%=V?a1<&(sf^H~frHTRpFA
zsH=bMh4$wnr_Bx~#2Rz^(H#df&b5#d#i|Zw4EKwc3k12=L7fq30Qei#3Dm%~doW@D
z^x%l?(VuGzW42fQ`;}t*4rUBIg}n(EaJ72iV#M_c3|3tZU7yve3r}QI7tLV)3A<$V
z2Q%St)5#JR?hR^*J_@^M{XKJz>~Bqi7x*~Fg@eHta2(pA6V$aH+t}I<=Hoa%=vfZX
zwQ%<r4<_U*v{*^)J-R9Q*IfbDBPR45%#^L2RU$-stc)9gTUm2CPt73#gmw>rM)i9K
z52?vm>yF`Rg2{69r(@bYOdJ<x)t0cgw7nWb62gNh!*kI=CAugv>keirNFc3b4#ZV5
zAH6CEv0YcM?gz_TSP*$))SEE}oQg*6R<vVEP<y+rA6$Ck4gf^H5=1U6Nm`8sbxdR>
z80^JZ`*!f?uKI?oI`y=r*a{v3Q~uhx7lT*VymG*nW3q0YFl{SQ-0E&0tNJ{V%%G5H
z#ZvI7A#WTtHez@Jz1U_{fOue^z(aQ{*6JuBC}kRE_ca6X>TtU~2pC)&&P}ZOC&IS&
zTeOLfR2$|U_UExa2>6Rp(6&Cr<2}*2!DsEK_B6mc&oVsI;tyjo|2zts$yM}1NAPg7
zSxY`>NLx8sJ6iIQ?uUX`wImEIaGrMRPb7S{t0f)a@{sM0)`zeFz(k`Z9r$JfmP`zn
zJK@`$QtWNXSel+QQ?^eXjDWAv0a7*i)f$y_wOegVA|nLe?6W%kJ<@lDz*f<s1P#YU
zp(PQJ<}zAQYi!AQjJOtaSlDCDr(y?OP)CVuw!N%7gKmv2Ygbn&B$0L%TQcb)@7R=;
zG%O-mms2lUcg)FuLMqpzfx4I?fn$0#4`Q^IHKdaS8`1!5e@nBCMh`q6kt*3_OX6@S
zb(blRo=!dkwU`*Yu=Y<j8&e;IOXb0<EqV3*UYpY>JfnE!2IR@YWL}{B`+slORsTzj
zv2Un8A|`aVQGd!`Ew|#cuS{>ZRg+AC=xkdu{hr}bNrqsq)2QugEnED-q&$bC-Hr3q
ziYez(%mBd^+LL=x9c_(=OjwRh!KjtXjc^(Q8nm?4sLcz&qxi2!=Vu8$wGo}SCu^=v
zfi}rB{9(GKi6A^iA{A8$jf4*HftDHbNWV{GmTasm5U<;#n5Bnuv&S!MxG2~CinpUi
zo7+M^3CJ#z&-Q5r5`v9*Osr>d3|?up=TUP6Y2XWeY4T=;jGUdG=Gt}?Dwids)v5{!
zH^2O%$96_N{AJ(_cNOwJbzn~DB0<|-NH}}&i0#o|%@=KNA)y$gmX~WGANMJ)VyiA>
zMCwpj39E@dU1vs|62eyr91iRef=Cwf-QcvE_s%zj8L3yVZb#_%O@kUhLs_E6LZ%8X
zxkWSIq*4+t(sv=IL`)YRZEb@363M#)kY*I-+$7wng+z7$q!RlG8jW;H5Vg9qYsAaw
zE@UK=aeg0ZaC!oCFr8za-cr4>ARWWphUFOb(%J%qhd0$w)cE%AE)i!VYH(<hgP%=m
z5^gr`RKliXAcXRs>p5M*F;kaV6IH#CDR48l>zwuoTNo^_Vg$5vWpt@<&asGYRl(W1
zu2bQ6+zmD5P!HQp{oQwt^L;C}?J(YgG5Qd3x43g-^eGT*AppkX(P@%DCl)mXe7W7#
z5@5Fk`+;RfK!du_19set-p7(p*m6`rVCb^umCBy~YGCbGwi*jMHGz(PJ<i~z!BN<a
zLEF$TS=;5h-a;0D=AEA=t$=Ms)Bc$hHp=d6UQ-esJ<za7I{@;b=JV|@uy$50$f!T3
zAFBY{WZRCV?<-&Q=PKeV)%9b<)SqgBoiN+8sQi*yj;Y}q4!bP){3fX=b*wexAJ~e-
zsx=(~PK^_ns@8<4_+^vIBq|QMal$Trdb~@wU)$Z9h{71BWlw7+0v?<cz6e9-){NEe
z7U;T$eXSV@VOpFHRBPZ<yv!i@H8I%fzR?k@ZOzCt(SOmeOmMGGu;DUAnm&z=G|w6|
zl+5Z*4dXf4sEVIBtJ@UAwn9yK7Z(E9i`+Dc>uzh!#IFTW1kc~rMEpSV?{8UAAB5fH
zsxDQ=@=f!MRj|^U#o8iLZ16ieck75mF?yr()^nH1Yg$$VIv<#xwV?p-6g(`m8I7^q
za06CdWMPgC)9Q(R^3;ulozxnPuKm)kzFa+5gUrPkMJ%YUX*oQ>h0RCFIH?}4S9W^}
ziOY@Hu_6)0Tyk?WI>m$!ZBnH&)WmM9n+nH!M<w6?ISCjRVnWFdc+3_BB_?!sR~??J
zoS>zE#UEJrRIeAs!?M<ZcPkh7O2CEhuq%RWwKZKZ$B)>0tSz}UVVyPv2ysXM#5Usu
zRXl7tL0hr7FogFtU`xw!d!XO3Zb$n2Fs2QKMo=XvG5Pd|_+u5q4*N66wxh$ItM;U_
zKo`NCr-F>9@b7uj|CK$9xpqtp+4F#{#be)+6%N0msBsK7-wvfY;+wtWP{L2&aYQ3k
zhXOB>pwpwl?mUz*(UT7DQe_TT=U~6>K2$+w9x;2=;rOWSJ(Tgk=m&V;p?uiSyl>Uf
z{t%aYO?>UgEW1|2j9nbCx<eV4&%%NNLVf&Dk!v=@Ak%BV!Zbr|N6=O`36Kw8wL;Xv
z*AYXh_{4Vw7C8n#*>%v0hcbz2_PEjh0w^8IS24a;$sfi=6x}f$__Uq?3(2PYP4>nQ
zTYt~3@RgMXm@{KTTH513)WeL%u*s~`v2Zee@~usBk7<o5Hgz_3$Tb;omazL*H44Gm
zZ7$b!IrOzdSyhjx8Vo_`f9ACo6Ql=o2JCvQMOWdN-8huVRU2uj(fS_B{18`Hj--4m
zR+G#Zb9Ou0xa|&tD(ym{WD7bZYl~2aMFA%b^<7nYeZMWm@HPMQzuUd|8Um-6V@0_u
zmx!pm62<39p|e_<E}~3)e6<WS=g!%NIzQcPo3V|c-@u@oXfoq1LFFOg?7r&O*4l&U
z`xr0rhw*iZ&L71Lc^E;W9;=6qzr#3dJ8Je36YEKQVUur9V?UA8xh8?2?5cox(wW8Q
zv63!Gvlr32A33lcPPlinwaa!K&R3qlZ(U7PbvR?1+XmXs!x?K%L)oq%gi>SoDcOBE
zvtmc!6ZWVfVyhuew(X665U$MFzNp)eF1k9tE^yLnqAF&>@XacmMTC!GX5Ha@aK<K_
zd9PQgE1kD~Zw-eNAyC4X#>0tpaJ&#i>F=qLwLe(laOQ$0E|9uw_=@C+ZHF^@!jil7
z$ZxFu<!te80KbZ{J{(glZMaR<@vxh?cOOm!@4eToCmPZ})Wv!QCJz&DRMsRCr%R4P
z8`2`4XN1vMiV=>`Wb{K_H$|JStm?!SP8HE3HD9dES(>Z*06DmW=hRP%do*F!G{6{=
ziTPOF6l?Z+d?M$+WH;0YQWfa!%~)$J&^c4LVui{)KgZlS;&wFmlq+^eqe2(lh;|_g
zFFHfM7@u1UOy#bIa-!5{OVPEP5Mb5(&vq|X;y8%19F4I%#Si~*BAnb!%RgWXS2fL3
z<dj>lY`EL0Z|Lt!Zoap9I1`$Riq8q|mWIOi!x`E80ZYXh3Ofu{sJvbu1}q2t)_)ZB
z{Z+yH0S8`<`2Ek(t!)daBc(QbayZ|7Q-dg^(u3w{bk^?I_%pQ}U;EXVJ=Y4!rTRr|
zhT9h*IBwIO`tzjy-O-kNcB>g)?I>HT+A^w~W+Y6N)#BVBxr1Sa+cK)dQ~IONcDE&b
zIZIet+tZfuydUp=7JRP?l~ukiuc5y!((CPCthz1V!92ZdqI37Oa5_>O9aiiQ>e>?3
zIfZ(@UVl0-*$~y7l}G;GsBTGD22E*FMUtplLGa+lfGd7XbSMS#Zf{F?*jcio8W4Nk
zN^O~tc{F+JR(s}7u;xl%a52`aMS=vlqf$8<^^^J6hE(V`bwAN=1RBx@Wa5)illXZ^
zqGBE#GpxvMRxrI%ZC8I-gdc;=saGTiu!eR`U$8%2e>!UO8iO*9>2<-0_zhN9Gn+&g
zb5nmh!+1;8M^Lzo*=>P@Y;R_ax9**Qkmn3p`(lg0V`9gnkByPbgon8sU^;)t=u5H6
zojCLmj80<#=-S`fa<rWq9JZAhYU(n3zZ!)X*A{F&1}+RrN4^o|)5NLRtjr9GvDKDI
zGRpj6Og&M}#!tys-|yK6(IG<3L~S1mRM1rpZjS`2d({##L;W9Pv$rc7e+6fOC#vHi
z#K-NaCXg5=27UH&K12w#=kYlM#@h>ZIKm6ijs#}7V@L|2Ydel)OdB4D6Jp$|BY9q2
zBkm1GX!Ro*S<epklZ|ZGk$jm>yGg~7gj^mX7_qU6z&#XDwpR_^i9Gpa?Td!T`z~5_
zjO2=OWcOUJ>MOlqtu|l-6WmL}>Y~L&iu>hv?2(K*cx9lcJ(6jSwjhb8DlmE6i4|))
zB5uAsYXyzQvmZCNw&<Jd%zfCFtv&ia=n9ZxK;Md1w<PEig-ISM*8Q>vmn|bIvYsQ~
zxjTMJ%Gj&Xcmr`R%F*>X;#$~{KDo6A-Hd27X3>hXnpl9zBZ+VnPuTR4O#cvp;1puB
zFBdA=*1-@9)RY_-B#yy|vrP(4tpGM<HZS;aRK|@Urswt%H{xpy75hy!!9&Py#bNOC
z!BbUT5H|pO#{eW{qQ_8dK@BI++AjtaY(6oW)y@&zmh^W@^m|W#Pq<~>vRcDW*oyuv
zqR?B7gFH!4a$H9|ve<eoK7stOljPWLXbR4L!Denog%ktHV@_>pD8G^HzNW>#VX*Z5
zpW6d9_7o0m*TYziwST%`k7Amh8QZm!kE7gqt?ig8c2p>4D>L$|>S^pOo@`i!I_Z2C
z)iIyLnO^6Fv~zsCh_V#u%Z}>so+sk#(R}dn$l*dgns5?7iZRDn6~b|k`h~LVXgWJ*
zKCs<K6SDTFWSo2SF_gs0dynQ*g$CILdi&HPUgsX0uKH*?$;}=Ru14)V&^ec<3W)cE
zc1M5XEr8G7(Ts@|@ho(`;b`V=%WJ=cQ=r5z_eMdR-e_z~)6qn3u#?6N93(0n%@oNM
zV+hi|v9_4}m^(+eYkr-dSW$nq7p)}lxdJ@?Mz>~1csh)+ebIX~2*rmnFu{)EXr_+5
zLf{b-VC5Luwe-%s+E7f(Z71A9n$T1M4Z;7;*<^H0cx~i7(}L;1f3#UG9E0}kx2w@=
zdWf?=7p+EdFypS(T9zkRk@?sg)(wTKU5~a1ONoXN<w?u&Q@a^I;HI8@>Z5n;X;`t_
zN588CujL|0`3bp>zUH<4bGxIeT^?!Qb><7PGT0Eq-I12@vMt5{1WfSQ40km(Y;e#+
zEyX-L4%$6UWN@4aY|Db%tuCB0uN+OBGV6h2x}qz_2|+t|iP1s5b&Uikj*Yb|mNtX~
z{y5&$I3AZ0Y>v(1YIH!kkTo>M*xbbW8xb%uoR8*P&x5{?AjW)D>6HE-Yj`d&KYZJ<
zaH#U&RGz3C#P8?qsX%ki`U1wAh>K%e4tWHj=dnnh1DlKUh2VvO28--iK5<-cDA*8w
zAT2(YAv-<TpR!U^9m}U5p&^O^rn2)`!bR+DP!JtU6k&Y*7q(kJ!)Wx`o?{uCJwzyI
zx;K8<S|RZDpV_`+`4TVtU3@HG@mQWXimMb|o+~nH1FF_p=1Tu{#}Y-EK5O;IG97}F
zqZkTsn?2!)=Q(PO8f+oX2l0t~e8dVBsNG+pO`}fZQSVqCU8OX=eqf5x_Lw7HNs}MM
z@T>b+B5a46Pqvq=CuW0b4Y<_%!)CZ0M8{jV-NcY>9#RWOA6I$VXtbDg#lmDPR@We*
z*7fCdl2^d2mZJ3&yBcdu=zmw$&BZ{(u;Zcvo38y}*WLCt5u^Elw>an-P_JuVXW8Ey
zS{|hSsCV`El|hKit>`)(KhKYHJAT5$imL?S?j5bl+AoP;V+&D!+B4>~Air3Qc@5w|
zbyxkv>0@^x5W3t#xw1B#nnWO$1KNV8F0z%EonR@vs)hv9w)I#<lJxP^jpczZuKlXd
zHe)Sv^amKatvG8a^d_AG+*h@XC1QsjMBiA=P1wVL;UM?SBW(#>#$)|;vfZ|$UsRrK
z4ydfMKdP~(0Z3mzzG<jWIJVE!B~$jSoX_=zpy?}U<O{7Ugo{*F)}ApGit&w$XKY7%
z!qAjUuS$P=jA`51o(Ueak)tp;vnx9A>_<uDxm%6MaW`Rm^fxBGpt1fY(*@r3?fHgH
zylzsERJSK`aFMX6HMt9zwAvWq5Om5dPpqyzQ;5sR1rV>)#}8eKu-p)hl38qGr2~Fq
zs-UOxTtf)3qUETY`;J;6esBjQ5`eS`7Fa19irRI(JD2S$v0_YP3|C2HQX~N>wP!L2
z=Z$*_{<?V)$gS2>>47*CtoX9^#&Be!^4DTHATL9|k_*^Sd!i2>L}$!K+cW9K)59^)
zADxq0y5XtIHr<|SBY0%9?TOaC&huEmd*pw%Z7yncoNBRa?U@cvul;h&=3@r#f#KTW
z>rvsb+ali(KpdeHV|G&??j}U6c1@3x0yiEx1mhdoB~<T@;Q9EZEePbHY1BXgOc%o0
zyDE6)%mG`9Vaqe@)xB6NXB(j!%h7P0M1CukWx7Q68Cz8^lgK^SE6Wc-w++D~S8{sT
z6g&s9Q;xAZ{I0yOitgK=E|mu{R0n<y57i}11<5`IRmmNt+)2<LN7VvkYCHOMrLR2^
zToW_1_Eeo9)y7XK+7L>^p2tL6dB$=Hyijq99s2*8b>!pUEQ8k_9SLQ-j+{7dRUP^G
zk-l=u6;V4oQdosQ{{Q{6T^$(*E0HPK=-nOZ^!N)H!-+i|>EN-OkkY*!`I;R{=1}`O
z@<H^l8c9(68>@~=o-C7K&HDCrNutZzXcRsncpk)jB=@rVjzkH%1~IIP4#{=NTcJ?^
z0@692XiWkT@#4T`D|BR91`|(qQ?B|pElj~BW$mi#k)kK8sE;Hwvr>nU<Rk0u$cS*z
zjRSkMmiRkze<Tx}UJXy)N*;e!Rxg*57;q?NF*cJDEsh2}&VMJSaWbGkb>xUm>&w$W
z)7h*h4WDXP)h}EXbDRs9+6R$JU5f>Ab7R7<$0!)0prQs(uB+b!BN8pm?&1Zz8BL+!
z<r%xBVOpS}SXgPPQc#aODqxXcskMcem&dx{@Gk~HiT$|ya!$w(msIE;_)c{1sn^mB
za=qnfhb1D$j)IE>>Z_Xh`0$XeS9ZSZxNQWqErbsss*QlUJPJQyTQTTC$p8IVgNj~_
zsF12n7;={OVT_Cy-l#na6iGCfi~F%^Bd@|fVOzC%L?j01Nk=?4Dj`$+mP;4wQ_b$s
z1nK0TsTC{B1<gx1km~|6efEXLc%dpS6>on#o(~%^lAP#xI`p`%TGjE4nVu6s8RT}V
zcmf-rUHY>bdb;~~MoJ=d;aj%nc*g!u2bFtj@9|75lG#0Gk5MH>3xa4@A5Y|Cc>FAS
zUHx4|)MK>)AYxB)+sEg*VFEAKAJ6E95gk#ES;mHdCguFFMs>z|{k1j4uxIh-@n>uy
zAafreh)<ga;MyA>h8t^F>m!aFMg2X3y@-{LXGX*?CoaXbP2I6(CtNRj$#YcoXrcNz
z>%IE33-^Gs`T_;;4AS38-Z*+Z(Y{J3`AL0&1AgmH2do62>Ib;l_`&seaNSprXIe?>
z>A7fpg6r~HtQ3xwF4teSd4c0m0j>)Kq|Z3q>*uv!;nb&}J&cjvdfCC($maePqk-GU
zGxg<KLHET2AC@_Gs4#=I*g|Xvo*{@h7NeUE&kuwO(eB1H$~anL0<#otj$@Yu105CE
zBoDi`9P3>6NchRPb0y|}c*s`OgPg566ziJP=;d+Sh-Kk?*```eku;nb8JaFk6!NTq
z0V;x0_JJzG+Cv5&#vCzt!qMP=SqI*-kK;#fXj})dhQ(Sr1}`)ClUR!s2<4mauh~<9
zP~^F8&yJ@np|QuFs~PX|v-p|45GY;gs<jgd`Pfjb4B6BjClW%ZTpKWDRVUJk7@qb3
zn>$ZrG;|Ry=VjN4j6PG)hLjj__lZOS&LI}vb0VEZKP#~0d({(X@ldnvi-v5VD=0x=
zl}`}T7USg;37f~s7HG99AQBWH)Tw*dZ`KDiWT|onqv1rR8<H^gS)(9ukrmK3=?e(l
zP@$ma*!sIQo?~tlx1d95kFRf65It*^Xj(vU3Hk~Q653h!55Pbq1;VOq|Ep)LSDj4E
zKu&q~YEpe}(=ZerBH3bIqiQsKe#9nK4IULoKK%`Zqc*E1Oa{AYENQQ*tmf6%KD4>$
z&Y#N!$WYm^+cwV!fTViz*rC^>GFneq`x}+9Ni}pc_QfBe$r|aF=CHQtk|(<z%fcJY
z?nF=cR`3#CP^n)Y0I{eqfc-hEe^+1Wpy9w5{Y;h#B^NtK{4W$4Wm{Hx8$1X2vZ5o<
zoWkg?YHofd+4_lm$+#j?f&4bqnc}(+dBpb36WI(U$rs#EP!E@_IFp>S`x+AEwNLax
z1zeBg^_4voR<QpGxgQ1e*!0?C0o3>HSADjvwQN3KvM2dKul*WPRdm4F<D!i{ag_s%
z#Pe8qe^EJl*$Z`xt3NKd#eB_94H5)zM=>Fp>kg}m2~#lY8L*vE3j&X5s;GxvvIz#Y
zcNa5t-~Wp3DJJ6ZCwS5q6IM67dMw>S`-%yVV2lz;VEWY>ukXr?)u>ZeLe^xp#f0R;
zVms<$NM!gJNz?~K{RH>01`S1HTs!=Y->9MZ!R=z4zi5gU7`=d{F&tubrr1^tibF)H
z5pU5h2(T5ODEC`28sLOF38re&ZX<Wux}!Cz8zo=t(fZZA+Izry^*1*JPuQs12IEaQ
zZ|>P*K5X(T@WDp~ZcD{mW-?mg)-yxyRNk@zk16&;R%T<uB#cV0$*U@Nw=u57IemGC
z5#G_)^yTU7d6UiSivxotmR*nC9LA|nkh?Xt-BbZ01#s|Ithfu-P`}-d^=VGdKX;;=
zzVo)AeTG@H#mXVbVZX4+h?a@tw59m&!cQKZZ+!D)1diqS_Lmb#{1Oxo>P0Tk*y@9>
zbX!;b@d+G%{9&?D%q$Y9?(uQqb;`CGz4EZYsbWjtLwS?#N2lm-&e?;Q-bWwV!`Ph)
z9r#y!r1Gf&R@EMB`tWwPt+i&?1~6H85}o3$cFCT`Oa|B{*xqNctNnxIw*|qUt9`PV
zCqRZeK8W(nK2E4RL^nG=*8TWr$n}=2>f?N}$1o>IlK-6_Cxl)5A1*`P^>IRUUWvoD
z`{T?pe~fcD*>^wA7jsI-i=U828D9T$+xKy1vXZ{grC67(`r~|*J^ka_fAc6yH6JIm
zmtc$yMYW`1QUpU+C)Ww!fBQdg@ZaiHpCyt(I~TA(M5u~RNwCA;z!EmaaL3&zE9H|d
zs6q;lWo<E`l-JSPH4@4JH-;}(#`Wah16I-~&==y|=<h@LEL%^&<H1(kl&bgRL^TPM
zKRn5$U5*vTD&1Y5FWZo6VE+wR1cKBjQ`))%!JHI`M9}4~n+_1sU;xIffWrTHm}7bM
z<3vt>RBLnk>qNn>1=vTi6r9(V)DSi3x`x4dBRT%uh#ex%A9rq2@Z1bgre?6Xxb<;H
z25d*Ef!8wR^LC8T2}E~ZE<iFR!?8Yo!6++PKP(Cg0w~(uXw2KsmNZwUevG8N@lj~8
z<>*8v{}*gUt6uDbnON0yKl;+v^=p@}{KgL!3oVLmYS@7?uU=If!Q|WS2MB$aM(lyU
z+({aK;h{P%9<oPjJVS%Wu_~R&KW#fc;i*vcs>$&_d#XWrrNRd5?^pR}&ts{lk##1c
z8n9Cpl4#31GY^R#fXwn}ggP^>x#tT(FFG^o93;cALPG8AOo+<9z^Cns1_OhmGazec
zM(pHmKX?v&jVd?;z@qK#OhmzR4DRd9H~l%6?N@hZ%qH5_Z%Kqu6F&@cVqrIH)wYCD
zbzNs75xk;sAkg117n9ZyZJ2=jHZ)eogMDq%pQIPJLT92!&K-S!%i03$G0(x`Lb!JI
zgVAmmOgq%Is1_ATW+_0&XN4egFPD<m{s8UKSm)4%^r|j?({8t7<rvI~6#i`Z1O?D;
z^2-mS8jKU#F^H14Gn2Il0{M+i$KDK0V3w3Z5Y5I~A50FPSL55G=OCKW-Z{ZQ=^Jnm
z<3rrgd<^T6=dNo6?ytmz`&!wJs6kdbyBS~2_$l7fSK>H-Z?~iSLo;iCpVGqJ32?fw
zxkuX@iwI9KX^Vo6N#i$gH_Ey&%Ce<O-!9_+gWZeuLjFCvhVe%HFff50kw7-FWKQEb
z%$!$Sjq!=3F+O7JF$Uw0*+y(4%ppuYMv^fHm<^~V9dE@*LzgEf5OCht;rsSK&0Mqx
z(cu^;)*eQOsEGa5XpdsSik&CyvA%c{xrNAflwI*ybLz^`XzlN@Y>uAUS$OLBMEyKr
z&vn>1UiKp3<J@`_80*TMaPx87(Unn*ljI?DtD>~4!*+Hhq+kP@dBW|AqHq6TyQAnc
zTnqN78&rQ4J^9|~7oR<kq`EKax3WgnU74;Vy|UGG<pZ63{9L(=^h)FL=wPerx)PQ6
zVE=xr@5)qy%(c_j(3Nf<Ohg;iYZpPI(yO|u>pM4q>S*ObQs~Nrh!@7Md5A4pn?OoY
z<1lHCUXS`4tQhb@EUl!k6q^&L-6$izzx8xws_F7KZZPZBxX7nqrPVES;4MdRjF`Bm
z_w2LL0L$Nv3F^t{{R%iR9bmzw;cBlnA;SutUDd}Gq-}jRr)F#aIPEg$YXQ{AI1)!r
z+I&|g-Q<h(OBCt)bphJUxv(2*pFGKaV>h*^X9w(7UijDTc0h(Nquo*CzG;-V3;G)a
z_E}quf!tOa7J(JH+m&x{j;}(pb6G5@Kd$HGTCagFVl%oNr)P@Pc@t-CB}T@!!B(T#
z(TIgukJ%o=zZkDg+Xx_%=N=0H@<y``GA6;*Yzb<jDUh`{3O%Ggc>N%D29)D{#esYn
zn}^sB(b}WRl3@|gbMLWwOR<l(W7iI)^t~z?aK-7<C_)hq;zplG39FChNL6`<b2qCh
zOL)N#x|PxqJ~B?$lL{{HC?%}8gqvzrDIY_w8)y?Qe({SEF-+L5QpOylniICWq~|NA
zf$DoonG_&Fa>tgvD)zL=@H+cS3C)gTjiuiq>&xHITTL`O?6H~<S(Osj^}9bsovK1W
z1^}^MC4-nJSVJjMA?zlRWHjo7YDeLh)uwV=7c|4xmXfup7I7``jBnRe{9(+4%Zf1&
zf_f`yrRa}~kVp#9kM0;^;G*>e%%Fy2Qg0~>NjRU<0_p-R$BMyykJykt(}_UVIvS-Y
zR=Ac+nW}lOaLp=aib=FETgpg~t{|=t$IGh$`5}(sT=a-O?i;%nfHvc|Lx$w}m_~Dl
zU5}>t3fYYqojh)6?r&=4j`Qxi6-5Vei@apFOW73erIR%T@s9d)(>U(B1+{<ewf(mE
z5-L2a?5;}I{s-4^fG|x4tb3IX{NaM2`DGdDuU6EkrM=l!H3|1LMY&tbbSd%wT+A3x
z#>s6S9h~Jf+h+WtKy|!*V~-IKdsYQhz8cVzjze=E#3aMY>>I2n9>&jZ#yggsTpsCc
z5>EiXd>!=<du=-w?)2~oy1OTWd>p0Dus!|3ej;tb+Q*e5&tk2R-w`e{KCgH<+864&
zJYaS*A)kwH5oqi<nJ?u}XRPXEK1VRH+&(+?oie!l19q31`nBATBjCxrfMp~nqb{DY
zJ<;9%&uwpX*Ofw2?2B)tSiBLcqdSz3R--TXmEb(qsv~mY5I?o<WWsE?;W4X^PVlh-
zzZy;^O0@PXGV-3xSFLLXQa4TNlxz*|yi+)t$WW1N1x|X~D&3BwcWGA<Q}Z0iRy>)B
z8rBF&LsE)g{PFH?x!uuZ+UaXg0Co!LHVJwInlJ?@pUlU9qMz%7JQQr)lnuqAeY;|#
z>h;KY3e{qhCo^#*_2mqK<)<|xcwqv?*{o`#Rfg6OV=h*->oTSq(bmII!kt|U5Fv8#
z$@2l?lvI}Z-$xHq7;)Z+uj7YX>}F-D|2x5NV(uwS-tAbNG73wV7~P3qP7_txE#MXc
znsQoFTZ}%=_xZ~>#`?GUzp<s4pt+3F(fXb$U<~^Q9LtuqXrp~DWLSy5Pw_+_ylktn
zq&4r^y7u`30)5+vni!RGYHc&tWRj3a!?txYh-D?lfg{)bs5OE=^`JcvEU->4gnJmk
z>WKoWdQ<_{bGF{I$1ykebKZ{5a3q2$eR4A0aPNO<Pore=!WCBWnWjNrE2=$LCq<p*
z7cs-GQ)Z_U7ULApV|K@>d}KNicOvqtI+br<xWRL_Gs=0w{2oE`u2UK955aJt8dh%;
z(<m-%&#Al~>4qZagRjTl=n|*D3EOunv(X$Z_bwVHsDAaS%)}qZi>BsOKKon$w`A>H
zG)q={DxvT3lO0wUV4&s2Jf=S2tT^d4oJ#BTlh3SCUALyJ>eh5Bkp~XRuoX^a#DzuU
z*kfL)5g<9%t`06-vSN${GlU06Ddv;n|8_?&9?gks)Ow-<>@Y$ubvn-bs2p8&O+yE5
zC~6X^1)n5=%E9pW;-tPXg+*^?)9R&k33IDcnRanAdRT?8?P_#<3X~!OT~tY0xwD+4
zYnplA43gY=?HV!}wA0rs<18RX!|F=opf0kTDuB*i;a0d6^OPwr#vCyyf-*$1cPbNw
zIbMiO#yJB^sq|J{RDUj!;k@3B{)llpj8f5-qQ7$l4h4zT-y{TbIf|6NZ$;lpS49XG
zTdiy>Ndj!WQbtZluz$%mDhECh;>}p`>EZJ{jN=n=6%2NO-8bq$rx@=KUaqCzBF+yz
z%rot?N2l`5M(T4ihj{VhIE@6W?1$6ejxH4KhIhr27@17w$ivj53;LdEE|Wuq0(h=d
z==sK?e!FK1u+w>1?8(cD)3xJtzMqdG1v{OP<qAGoJ5MK^Zk<Kn!d15GbVgweDxhk+
zPiGpCG?%}#J*N{gjt>rTQ`la8LCaifweNI3doV1@GE=Pz&Q3xQYECD_J5Xb_r!%5g
z3=+Qn{x7ZWbizN_A+BdN<a`nJ%NkB+Y(x%<!XJ)XfckmqCu_SksYVh&;mwWurJ$LS
z<s@AjZ84KR*NkF~rSg>F4oobb&h*KpZZ2YR_MxOofN|b!`nolN(eu_LSTDH?c(4A(
zSpYS$))}FJN5)ZWe}K5f?x<Ri4cMe=!9dKSV~Q~*26&Foo(=-X3C(&uyBZ5}0iDL2
zzT7Sh48ImV9AZ6eUhwb;;vSBDZ^bRC0vu|OLA$A11mfiyRsl4O6hyaUrCNFEBKo)!
zhXOM_`do-_&wqk@VeC#zB^bzE_1E6o+|_DJQ5}st)-(6C42>Vya%Fugf$vr-;H9ug
ztAZlMkxMMty5L1V1AiTh(Z$PuGp3J8EqEKp5K*X`q%YZh9e5m}?Lkx?b*z4<o*+b?
zPUMjwU<`=R<EZIjwrpGD*Sz}vzq2PX{^Z0Jd#YA$DTTmV)!etqo~sA<zVqw-h1SX6
zovtHBpJ(0a#2-bUVLQ4LW?dTWAAyUl(idh=_y*Zd)%%b9vt8X8yIFh>$<5ugyF1^^
z+*U-1>`usjA-SFI?M}#kk$1tq?u^|o9-Opl4RUrm#ZanI#S_Jn)mBD7&pnUXm3l|A
zMZ90aX{dK_Vr||Ux)Vxwby{P0Mrl{EaC18k(WH*3OCUTe=qrU!8n?FYOfGUx3~N`{
zxMIQ^6r*hiqVAG_z(Eyj3@|ZfTt=K|J!*)3ray+YTE0A1j@3#JAb)M!P_)G_0KHLn
zCNAQpaad3XSo?sTQHtr|=p~HOW>rdHI)X(;SA(2Xo6`~wA$+(NYmILxiK^#SRGb>N
z>*^q>rEf$BTmQsebu(I_mPN2|E7oZ12DZ09gH^vBKlyti8ExH(9r^apY(XD{d*n&5
zsO>p|QFF|rZ_F7~3?&7IbG~*j*81Q<lt${}V**E%Y()^dHU<{PR@Dl*1d*xq_u2x&
zosEEeZ2|qaZB}*!WiV<ZRc!pD*Iu*x-T6vEYUVfM1C7nrxd{EC+PQfckso7qnlE6#
z`8YliWG>sbK2cD&x(fP<I!1SJflRO1!9P{GTpGd5%IR~BAn#gz`OS0=A+vVp6x7Z>
zNw{=!+y#$t!JmjF9vJK!w5m@MzG9+^QL_q=7^5EF^-1PY520t+ts(@!&iH<k@zo6h
zcwlj;_I{EOaSU>qY~1$g8#L)kiHobF4MF?3k32d|jfNOfw2azl(McQ@OshW0tIC7i
z=U4wpKGiO);puG+YJvxwKk~RjqXq_NWA5eE6rFS;(uB?yv=+ccrfx9ylSE2Jm}0yB
zoPwV#3NT+uwz?F(I@@{{Qr4|Sb1|0ne3Eg#-^~Ji`+H;QJibCXK$x7Ebl!GI15FS$
zaWrZU5dV8Jx<ak*RFI4aFB@N&aDmF~Cy6?p`44vWlT4j(sz~7s!L~W=6VrR8+O7#q
zG6-^mYS5bfsC#1x5X%t9HDJIO#BNrmHcIHMTk7mQmj~=`w1R`fwKqn0Q5sSTe^4w0
zv;~Y?iKMb9Xj}0IRP(Nx0cJs>u;{IocahzzoWf~1_vOmpF3(vB*bag_tD3`knD6x%
zDZMLKY$G~sn_&w!)rsc=v8^aVuYS?)>yuy#Yrh2M2eIe~K1p}UPY=}?9SPc?SkQ}@
zG{>k@$m}@v+nV1YY##;b;fV=%ReLH(J-+p`Pm-qqJ2F7n=b164i}O0XZgl+0kL*Rv
z^CXw7eVUJyVjn}`cYK<$v2=OTs`N$nNf*uS{50c5MN?zDqBd~a?oaa_Nx)3lq&@Ln
zhW>-PpXRHU_uG)|Q*m<H_XNk)0Syw>@hfY;1#zPVX<+asiW*(nNwn_ML}S1>kgPri
zoVw)pZw;R&n$vaU(^hK?P%)XeCVdec2@+w2N+;Ob@K}v8;eh_Lwa26f&q6L?Q;MG^
zR9^qt6)Q#Ei6KsS_e=QP-@$RZ=hM6^$(nW6ZOxpF^$LLG%J&B=tIL{K{@8{-&D3ZL
zPYq7ks3tMqk3xI$(@YSvB!K8~F|oD38n;<}jmgL1YQQ(zcL{&|=o=qrcNe?%X+~|H
z9m|R1`;E<inlOKw2<rOlLxQ~uh(UBub~8F1#17eR#R5Ss0ShiThH;-4Tt4rpHHP;;
zw}mLg7sqWJ7gb31Od_a9*Wkw$The3>@tlHJz85VKb9z{kWwj)>CT{w%3YQ?Ut8oy}
z*Yhe|k99-UR<T^!i0*KHbAQUs7(9hO+j?34b8@ZzrQMJ6aXi=TfeuaIh$|5u#v)+b
zmYfqG2_W1N{T=W)z$5I8ZAVYM0*JB^(01cla>Smh_ySP{lT=yL<Z>FY=Q=1jb=nKH
zO1?Y(9=Fdj=8mAB+VNS!3#h@?HLLn8ADc4yC=pJ;fb5LwEwCoLKFg@<5%9)r_h<PO
zZvCfZm9povjJrfv!OrTlOgUUTPLFWNQ3E2~XVvO|#_b~!J$;riao`FhwN`(lB1K6Q
zjrx%>TD?AT^hnWJ8Z-&`;k92|V^kfug41!+XYtls#&YIs-Aq=9iL{Vx598)&1*48z
zLVR-A|I&&v&C1y?#k?w-1p+}vo%WgXCGOdtsB_%um}7abru6Ms>p!!yAUNKIp3d=i
zC>FX6+!&3<TekojlU$pOt!h1I(^2~fuZ&ssOES>^Y*z#HW|uY1sbUgp{obxcvo;W8
zJ{AG2f_1q5Sw?gAADNspF?Qp#Ow9)`V9G^WmYV?x);qVN$1Xe{Z&yY`IDv`Moyrb+
zP^5+E2H6`9JL9&fQgqQoca2`IfZj_ng@Y~yy%#G>ypT)W2+L9XJZ)E^_AyeK+NwVL
zA)uZ#W<5%qr?|H^0*ni5zx1%rw)t5iV;{V3TTxcT7n)dnc0UlJFF7N+X5~T5{*OPi
zhgv~@0mJkx+oRYv<ck+B`;mSe&<h1@J394{<y^mP;FH)1faY>}8iOGFe^y!4dC1T2
z?D-EiDd}?j@Vxje-8B>Jj`gHJXLxe#=*e3*1_$k0hpL{8o?Xs3@9atFR)S92uAYqA
zm}Nn4clV@qIoxS`dJ>K3@DPf&w<oJX-GJFOkd9MG`S<nY!{h!}6~C%#HO1n9d%h+{
zIPIYmthOgpq9Wp`I*s?)YgVrrV4rLaJ((hTc0uW28e<MeJ*-@lhH#kSrZoiihyVaR
zZPO4Kym*Akc2)Hfu4m01Hv(i(YL-mP|AlSru7K^qRjj8cXqekFUSycwo=o<}SVpet
zvO4&p^3R4enLm2fMti=?fh;`Q7`FbP##fM$pW9^gg6F;O`m~^Rs~<LCR$s{y<u`Wq
z<zithIrkDK_l@==a7`5|Aq?lE0w>e6R{qAW3r2~%82(07A@jy1lw9OZ2#8WdO*rnT
zgSsE<J<iPS_-S0mX-J;6lvtY+rCm!mihU03b{s&|C&`%UZfxF^D@&KO)oZ^;F&zE6
z(kb0&%bL>KFBuqL!dL;1iMXm2K6ud9E1UIeu5jCk85~Q3GTRIgV70J{Kr5_)^2?AL
z#O_C>ezc=*2=YL1p5@qh(9egO;mp7Y&igTi^E1S|e%zBU7`GTAi*(9uV%vf?T{+*_
zlS+5IP5cG^RNY~yGGNcNE!@B}Za`+wqx;f{!U=m3v!1{o&(5TS2X9yr>hCy{QJlA)
zX9u;aGa1FPogB5DXEIu%@yE3IvY(Y*XA(x^K?_^4`%FfBgry+Eo-?|G*M5D;_Udnc
zBiOz(8C?&fsj7~Fla3tbsU}A9AUwktkueg-SpYM;t`wrx>ds`0hEc(4<}&D1uTeOe
zzp;ksmuLRqS&||*#)JrcKpZ>4-k!o#6$BIXmsqUU7K1r=0z$Q)$*Aj!-4xxos20RG
z_Jt@#3kfo<JBF`B)UlozesJ{Lzqa0(B_byn7@p)<aK#d~4v(WE8j7VtHDjYMOFKJH
zblPNm_qk!UO~?1Si9QrBvoRu)Nldt+vnkkyxtE2m%D#3cy&L@UNyawrj^KuC^JfzE
zK7>VP^nNOZy}S_}5aojRwbmYE-ph8Yax9WD>TMNZWQotAK=5}xH_(Erv8SZ|T8t%z
ztsh4T!_zuOY)xYDCH3U@zTK<z^sh;(z8p0RAR2j%0yYBiV>=x)KZ_7?J-$ZqsN7Je
z=Ln8vo6!xF44=p?eTy{%<L^heZVO}&V$=OpaD_ZnIk*0ZoAEwU`4RLWgDIZ);~4ok
zCOq33dFDL3@kD<k)h8)vi~?g#tQ{3$hzU-gNq4gQ(7cGga9=n3JhR_xzp@>lCj?6Z
zgx^`!=Ls#rNrW#*D0Y6H(6ZlEAiF-#_YWa#@LIIp`k@PzuH3WssIQa7LUXCr_I{r6
z0HGrCS-{)(c_ti}+#LWQt2H3ois+i7RTDKP#>t6_2B=oQq>En(g6lp{6Kg+l*y;s>
z%j%S#u_1swa=fdxvt*5x=6}H{a>KBnTa%i52DE~vhfmI=wP{s|vd4wet`E{)!7~;$
zEnwscWF>t;W9iPi1AfA%TqG8Ie7of^L3U!TXTjLY5J)wrZAfF2jsi!+XqB9pY%*4%
zj1<$83r$B2zqYT~tiIP4rYa6gpJxiEpw1}JY%bQ~DEBN@57(lT^w6~VXauL!C(ir&
z=lNz~ejX6<H{bsP3VEY4Ywf_z=!mE(2<UF9)Xno?-fw@NslOXIu~44y_0H$%hHw!z
zDj8dd86t5=ilIeqOB;`{yU~4$qHarS%LWiYd)e-3<ydayWKp#vMblV`Zv>sJ|Eii6
zCz2S`)?-OLh6U@op>o$Rdeq^~mm~S>XG<H#eo^&)bd6<{Yn)<9JVyG%IPJJIyB!56
zEjZaQ(i^EVQ#gpyU|Vx75H0+PDqR`;_V4UzEL>W9<rsMu9YgERf)e;K#ZTizsY`xY
zFVv%(*xDBfJ=T7E(RO^1Z_hTAi&bCbOL}sGJeWJbNVrW*5}w#!B)muZK&(XcF1x=-
zB*No^b2sawO#UQ{FTcoCAiYYLZJ&B2;Hir*tXiWFVl+_`qqhuASZ#a_tlhIZ^~rkF
z9I$$Q3|{hpT@3<-5~4E<p0&n+nq4?+O<%;}`e}ADu!074&ZsShdF7Q?tX+-A(S;1T
z*;e%JLPE@^wFErMkdJ+lF0V`2a_CoF9bvi!n)t|aG#^5VgVFRCnZl7n&-pOY3Tk*>
zZq5o!2J9G=fy9*+cP-@X7vKHw*nsOydlKd!hDE*_1Njqv%;xm%?^}dpOLk2iO(naI
zs5_iI4X=NZX{@_aa2MK*fC6(O_-!>tc}+5;*oM=?c3Ug(dX3%DEO^@gU*P_N8bH*W
zZ85gLsF<C1V>k~Yas?5@(ifQ&RV1-2#Jx&~9_Qj~TMm%N2QMKxT+zTE)!J&5bdmiJ
zO;mthM!#Vj(X|lP-!Gf7;a7OOZK(rzZgKPX<G8@4dX&HPnvP!CB|^Gue#xllk^XiW
zxIK==`SxFJTODJRPfUibU)z&dh~&t@>sc*RPT6OIiKmylxIfo$$?}3z^CCdP{>tNO
zxj?{rg^V~z+m7DM8=p^^VXJyG${m55=IrcL<%toNWLIxO@d20Ta3%L<6hGwL!JcT@
z(%G8s<mmbmd^#aJqCO`U3+mp416?PrrZ*9hOBl%3_NM*BjKrVwTGl#2;2~+PUJyWk
zUG&kQFV6K8TxO%b95pz$O;LuCnibT?fq$F3ZEd}oC24N4_5kXTYk!Kp`9Q<Gj}R%|
zN-z6CZ`-XoISb&T4eN>05|7qieI@k#XDi42yIfXcL%kV;Nu45b*{DECN?4nW1?+Ga
z{&AbGtP3kYtHt`szq6|vV?v(%bE>lT+Y!4Kpp&5jwtd{@)u{RlyB=VkhXc5w-%IUw
zQ_C{TKf5JZ@XLlk-456ZuMS&sCptJrn7xI7ln0Cx;>9TD(P4L^G4Z>h%}ZKiteO#F
zT(EokBpIFFz?B0=5NlgeKb~@oimUo+@%s|=jBRUawT;*|Ou~O{n>sce1n7|aPa?_P
zQWX{&-e}D32T)}Hf_6XX{jQR!y1?#WF`Q(94+RV-=$<`_NnL<Be;i*^_}_1<9%(&h
z>`8CFB`TpL{N435e)Md0!<g8CY|mngT4!Nu;d!5HC7az5Gy(8JP*=Q0%=&WQ$_fvA
zLxhgLe2}IHUBa=QvZ}s}wukyKb|b}LU&5*Wegf_2&olUs?e0sM&ujOR?TM}(l*HK~
z{EB|qt*I%jDb;v*eAcS_65dP=PGezI)0Yv&IR%%L*Y@T6|11vCb!ztkVOHy-puF}^
z*3g%z2BK_ahf~(rmkIbw`qZR>c-=rI3IS0H<Az#F9kAC$w;yPa4jhn~6wkM)4*b6R
zg_Zg;g)4LZu=<E^%oB&tT90N@K+jBw0H=?=S|y&QWCzvXC_=D}j*+MQfiW7};GUvo
zSM*P6V4fGC!L$Zy2BSPEtIY;fV<UDo7UdxNow@ir<U!t};_+6S7kF~l^5}$iU6sjm
z>vrTfqVnfIv70LLz@NNtZp98`@;;2|y*sDg*2dH{?YBFzv>dCi5icyLFBTq?vPD%q
zi6QG<{m$_woWW8Ixc@V|7at#^hKiMSqCvXBSRiMgXLz~UYF{GyfAqSo$KV(-dz6xm
zXiXMep30kX!akXpNujs4;ul0#?%-heV@F!u8s$OsNia8$AO0}PM@C^vA8C0A5rqA<
zJ&xMEkzJp;9lPLy^e5VRZq<2v8YL)j?O8NT84pR{tKwv?gvXMKfsd!+1&+M+XUuSA
z@Q(hB(w8vzMMC9oz^eXym%6)v?NkelJ3T?uuKtYgFclbxLP>V_XY`vI#OKswRH{kW
z<uMdUwzoeYdMsJk^?m&bclGYAj>0m;f_HHA?*ja>NtU8E;6a^ClHdMBP0m9H>tiw=
zH@qRb!rvCne`CK046fAHq?)e1gO4agkBA7t;{HUihLNaTXe)KJTYD_aG}aH}e(P`@
z$6!;)B}<wFj?U~}cK{8&8wG8-jQi8!`utC<S2YoD&`DV^z4Gd-@7qvx=ugSf_#(TO
zO-7H{lVVwiamRE3eU?;;Hmg-nj}Z?vx?)!~|MZ%eCWY8s>`TYmFOwJCwHU)s;Cy_o
z{RIIj1sR%{5xdc!2%(D}V9@cWKn`^?R(F`iC6<S##Vd*{`L-H!(PJB;ZjsbGW3~{@
z2*O0OMeQDz*>*RI&=|0POEKpZKf?vg_q2-5=WIEEokZ=jq8)R(3;Rb5iq*=X@Inl-
zu0g*2YcRR;=RY9062vq**=k#A(CVx)Ph5dQRKrSaN>wFVD3g1r!eo8!`ie*TP9_rH
zuriNh9*L{4Z7s&P*Vxr30X70zdm1&Jp`Em6s=2m^;^Vmh{j%p(dl5T*$&*spK*q0#
z3q6odc5|tv&8h|xHW3LNP<QHsCmw>9T?6?#<GtkJ;5hla2NLp}LoB{$Afq{pPb}Gi
zOc=Uwm){p(`#GuA1L-gi5(#=Bqkqac<0jR$0uPxmFTwBsmbYM?z(Y}jv^juB;FCn|
z4bk->mpALnuV9P}o756_P=ZhD?-`etwM8#2m1uzNnkg#|yI&m06s8N4lyNJ?)JVtm
z)&IxT`@lwhUHiUE6^c;Btx$#XP#%hS9GZ|EHwRUyLMhF~#s)(S@v$93Q#Zx-*dE(s
z&tzu&H$WUh2qC^H&E@5AT3#>7<>ZozRz4}s@yVsRD5eNisB$V)p$b(fLKUh|g(_5`
z3RUm3_L|?<C5q(V{Pu6{wb$Nz?X}lm+Y^xXzwpvJ>x@?<`NtWB<27F8J)=%;acm=7
zk_(pQgIwR(^iNiZ4!iN)B|B^or2Sh~jTXaPvM1DQql4QQNMU=UoGH)KF;WzczGx2h
zWSqCB$c=^IUrwYW0c_oxf3O(=tIVG#D^_&)OoJWQSmCC8b3&vUcdahOI;l_b{bE0=
z0YVD}7ul21psgwVna>0bKme~wq2+0A`YAcZqM7Z<1iFq?n#0Zo4)7kozW=Aq#nA9?
zaeK@8o=Y`p)-S6>J0wBoLiB<bALX)^z&PM9*QRIxcVDtAdb91>?Y0n4u)trSzr3pX
z@yy}L#D7iA^IV9o#}wg|O{SR}0kzOI+GjW81v|fq)XySMO4M>&<Z0%JBdpHuXt+DQ
zSRm!LgZPcl+P&y6yoT(4{NCJdmdj={sb)TL4<cKV%k~Po?XIb0F3M$_Y$pnXWokf9
zn4XkAYI!d0Y`-<wid?p;lUBu6>f`GC2ewL2d4g?L$CIB1cu=g(C7!@|ch=jw013`q
zY2!tuIyV0kTQB%na1k{(2x5QlkZlx<_JVDSSGa&-MY7Gg#6(&K$kLI^n8?C$w+ZF4
zl|odg3q>s&Vmhe2Qc<hAZ5;LF@*Oq&2o|(GdN<VDi%q>Vm$AOql$nofZL}uUbkTA`
zcnoUhYa}>`_F>iv`f{r$`9bU`<uZ1Nw}qzwDN&3emo3$#CUs{<2r(fnh%H3*w@-uI
zjt`toYfQuNS*8`|=%C2Db&uJRpc~Z^_<crAVUCf~@L29r%|wlIz+E>zA)yx%{)HV6
zD5Is4ozT#{UfM}LwX@7m37v|^I-C}^4dmO#(<3OvJ+EkI17fkCx@BsHHUUzf)BNVf
zZ0_&z^qR(Vc0Rxq-4Jjg{zk|eu#55C{JSG487>QzoQEo22|QD+R!qNo$z~<qNRQo<
zNI2I5AD=8gS?zkjCCr2y+YRC4qL2GM*iB7nm1HRH$$l#aH-J0CZ4syo26!iW8mUBP
z+q)V$s&N+8y+ufg%jfLAmMt7>6sg5}(>C0R9mJOOrp+`~^qE-No0fj5zt5KSW_!7d
z3+eLSY~{C-TNh*diujB@$YxsEo3VnO2u#VivZ^=RIV?{sMEc(S=s1tp)==;%UZ<{h
zv-9f$GCTt2xxP1J65ZIh^R_`?sW*zVWqirW%Tl#XY8(k>Hpg4HQ*cZ+_GX)nVsXj)
zpgGz%p-)Yo?JXM9-c8mT58>C}Aq?49w&(Bq!RgsLqiJ7(qt5omOAn+@Za=wrNtIFd
zR$gCngT}|Z(3>QWzHTe^X4{vo8Uf~3715gFDl(xTddq#q0>S)2MPeH;3k8n&UU|Lc
zLjf;Ix8#sbYyODF_bgRa1S=&ZnX_Z@eQ$2mHAlxarC^#euh|LFo*CvG7ClExup4B&
zQ~HL$YZnS|npUHvEFqleOe`^qB-q)&zQr^2&k8u)jQn<bO!nhBVg4MNtAH?4qW*%K
zwRaW9?4p|WaPI2e!rL4gSE94=t~`l2q8+L$7|PXn!I*~an&!Ecb$MOLBbJRI=h%(t
z`r?1X6-(gro}BuYW_G7X3E1syW^hZqqj&A;26{JoiQ)@(Pj5X0oz(3DOneLCabMa<
zJvAlHlD@QciFPAjY@ap<G(&&fm)ZE3Gq>7=XtKO7+q+!%Qut9{wzb>1C#~#D41jgR
zjbv2~JUhfz>$`<xNlUjjzDp}4XY15^hjT2e>r1?&k+W)jKw;w(8K};1(y@*4rqLC^
zoBGn?|JJu`bF`@~)5ZYQ#R!}GvOVijgKoHK2>_#>S-7<?V>>n1nw)j$1NaOcUDWUu
znlL6#zs?vga}U4UE9AB}Hd!t@o_}}H@`4j}LkT;Id4S<n2sE)$v?bZBRRbc~NH7Ra
zs2K`~aoeXy3W(!C8BGDwK4ymkL}y*69TAf4Xt0RmRreL-RM$Bi(`2;tauvedqaVnA
z=|+JQ8p{?q<H_iW&Ey+X;5Zz~yp4>k4oQ8FrtVB%#(t)B!9J@to-~OgHb_6>C-1qw
zOtRR*p@ZHd^vi+5Sj4=X?~~N%H~9ZV%6-!=2yOQR!o&KadO=5nRr2y;g8_G|vMWNZ
zlro#R349?)x8<1@yIR{H$u>(GoxaSaaN$If$*|NU_kioQHTLiRY&T*oT;Z_c3WJtj
zvVhwyEh}!zT)-gq?VzSJbhfkFSIE{zN*y7XSY>&HbL(Efr(p1a-B<hOq!qRPw1u&v
z;WFKyZN(NsbC&jJI~5%v7b#oTpOM*DK~kAA`sMv;$GW#6SKj`N31AB4nzm99pX2SS
ze$9brrkw`rtgY@(JJrh<ukFu{X*Z16x_I<6#0w1837-<4I~X6<FaE&ojLJM41fNT-
zC+gr!+t{D&OI&!$J??$8sXyDuNO5Er>a)#(ClZP4#Txsw^%sfGT)1u)ZpnqaCCJu@
z0@zw(1lUikL*r$&*d9G~aEf%sC&Wq>ZLd&w?@z9YIT4QADMQIe<J>6vv_gRKJh9yW
zN<zMpRDM`|Yk*Pn+JxRx_>npe<fjzkWLT=&v>w}?gB^-*Q6+q6N8;}`XNNQK_mqcb
zAB+B7v5MyFc+5A~TPFWRG;4jPt;0@6v-yt-uAV*>{bbP|r=#IOKl6Gf#_Bqq&)%dI
z?pckU^^NL%Ha?yIsGpl!-p+}N%}Aefddjm#&j(m^m8h8GZ=8<V$7<CB`?(ww)Y{f$
zSE3Q2HC0<sztnn1hj}$7r9HXW`rYK3piE&mxgJm|ID_BN^r2vP509IoY6Uj{yA_We
z1no7Tce$-eX{se8y%T7VycBn;_`5hxMEyOX#8QNc_X7itdp8<cKJ8VM7(=!spSE4E
zXB)Dm`Lu_=N<Q#q`D_Q4e)+HSKe6Tcw5M@A!DwtN^4Sh+^)qW_d^gG7UzN{xyH5h5
z<7zcUn>K7~0}hA96nBwz(L3oGF<aEd16}kMY<+;NjIkZ{Nx7E{(1F+%7GWCNrg&Q+
z#-2&u9B<*w&Jw9Sn9rhyP+g<gK+T&>=gk2gr(0_YYB*Q4)_k_3(Pp6-!SOkt?O`v^
z0G(k4d-53rBf+P$jFcL>Qv+@2%A=(UyjY!{yFOPVb4tIE&u6T}^`I$;q!9QpBr63z
zt<IXN@prfD2`6GGiT}GpIo?fv0NK81!+C)3V|1FTuad)EbBOB#g2xe!j_PLEVmOjT
z+HJd@wPR}E#tnwkSa0h|y?sLC`)7A=IH`$l@5xhG!A|LYJ26~#dhz2HoFD}nhgQBm
z8}wiqeEL6aHbyW$fJe<afrZl8hHXwiw4tg(GC8jwaKg(mtP26_+2lTbF}n5ZCyo!7
zg~WCi#}y&r=5f;F<g;^RONhe0S+52%efYoHHFdQ;Y?RkE#wIW0cO%*oR!#xYn85~;
zN!^Oi+%wWG@#NA73?uc)o%o0uz@RG(xHJ27dajUrLV8;-c{Ma?u9++MQg(14Z7~n_
zMiOGxE*VJsnMhR(xdYkR%PO>GdioapYxzKuO&i*5Mf^@RPTwRe10a);iZ&2mIFM~`
zPhD=S^|<igQ5dZqNGyZq3uLa-JH&<otJC8a&zQ14`e_)b4#Uni=))~jwoxsI=l^ZQ
zHZ4B-)q3+lwh@Oh$&xI?8e@=X{;BjEfN=7HXSQgJjk_DIRZsm~A3NfsJ~Fx59*vS3
z+Eo~4=Rmd%F^fiY%3if0W35Nl<kY4CJWv~9#Z?bacTHJAPaYjy()TXn!d$BQo*Y$N
zwm3*9)TWu@G#D!P4J5AX%xyZL;e(4qrX3o{crdnMzC051S1s6#dUd7Xv1pf!8_ore
z$6Kz+UR&*iNJ(ll7lQ0$(0V6+N1k8gl)Axb9!7IIp4+?<98BEK1Q9<wPImdT8V|0Z
zsNQ%zTkGwA`@wGK)F2fQ_Je6I##bETmK$w3j;pN43!3wk3G$)_)r{NWWuc9I4%G&l
zDm9NNg;)eg9~j}UY7C78dlH=>G}?Q}{p1(P>+yLRd#5kW8-mZ~_x$*FvxZz`ow=p=
z+t6Hjj`Z6xjx}}v#qJ1I_Yl7;#J2C+`i$MvOH?p+Kb{uOC4X)k%+@t!(&3v_Vi?R;
zAs5*Q++I4EHW=3rIOVdzY!Pr7AQ)l!V0Is_sjZ2*Br<e`z@R<%plBSb2IaL^^NwYP
z(z$_Re$`;cCp?A3>ROM)FCjh;X1gDh|HIZTwxnQdot|3pj9o9}iiJr;%ngHSDc85{
zvW@ZQ{LhAM(_khOAd<T9xLG~)VXC!8y(cirt2{L82mH0WFg~;lUh*WgNJ(ARs+X>3
zL^<6NDB&VOMJ?MSlu)v8Tyzd*hqd!hd?Uj}BLH^w4C10OsN0nnrQwRgmEui|0ePXx
z*80Q>@wknY0!cR=-MSW-pu(!?zoY0A1!J11H9%1emiyEIVE8JdyqnffIOn+g@|+zC
zs(HS;BZHabfvMGtWW$bTVxC~Vo1U^`0sphZX=NRc5i$GJ9}FnHaJ`f9^ZaQvM5i<!
zcy+&>uF(yifGBWAbn>{4a;?bn*+mqcHYY%FHa@H6<vFL%r~ryP0Q`kTI~Tor4oy2h
zn4PjNL8VAsVh}Y;42>h~VgO}e<Vd&6dh2>tA`k-{r(ekyw1Y?tL*eHEyBb7cqZ5d6
zEgmLETz_?aFdca1F}o2Ekg})jrpVCY=fW*L`X060iz9aBzq<n75poq6)7>C0^-}C!
z{5i-Ge_yi&3-x0e{%u3qLHXztwqz*V$6m0YWJ`y#ee5gF>+!&^xf~B=?7%f3xKvhb
z`B1i1fB6YslY)cS6i%cohcb@RU#k+dYA9`%j_o^}Z1qsK_1Zn}5h`i{+cy7!trIY^
zDzWwJ)O*-B;d<)@A3o`c9&LkQ`i{aZuyH8udd#@CNv)}uoAl9})w;cF*b@df4rT16
z(enf~>pkj$cMz3Z1Om~?J^8FP=!kWbhmPpXbLi}e-#L`oZJqJ^XrW|#V~i=$^ql45
zMK^9z`Jrq}k&lJb*yX)~`uF<~YN}E|MRmaFtML%(&xwFk?4~xM$msPQrv;<Z)z^!A
z&7q;R@b<l6N8%fbD%ng8Fv%K8G%Lz)$$MIp;{wqb36D<b2Tu;*0(Zb1%G;@!;Q2Eg
zuBU|ve)4uEo*K}iaFaM2PjMjFY+y2iOnFY7Z)<zz*;bni%&2OO@B4W{+=j96LVQ)p
zk=Q*DXlSy_nnGuWc17S|*Jwc&hO!-B86`JpDo9~hqmSgbXV)~xBpGL`U<VcquRlIw
zH)4!rP{q+I`<pqxrLGBI?zP*ZQu6S|vpL#UM$v)Yjgg=tMf-2}q6bty`2QdU-4BH3
z&r@y0tdRCtecP7i`Zilq$W8?ICf2^BZFAMt(n7X7he<<~i~-ByOE}jQwE&hE(w3*5
zI&5r3A#Hi*Askh6wlZE;yNe`gU8V259N6kY+7!t(f=#_Pz65BAP=KPXE2K?6$-(Xs
ziLm&(Xn`{&32b;?uP?kbJ8o9O6kq6ss%<P}ykM9dAW&peA?@4!oprXkka&iB)>xz3
z^>vNmOJ&VM6&sGV#8-`oey!1|Un0EvsdWUJj<$QJ!cg`^<1#n9&VWLEu%Bdm<5Qo+
zlvpAc2#gJ&d&~=r7u|LdwGiKOM7mkA6m%S|xRg`XK$F+izjJO+L~DeZD(rJ#^fXE~
zy;Lk^I(i!N=*>d`353b+btFDR2KMXrjNoF0kD$<FqH#Y}B`55-2=xrIFle4AWc(#X
z1f&|zC!;$qkQ~7A0cO1#Zl`0~=|9&CYdjNe3+{??R{f(>PAUeQ75LVa-OTl4=Y#`s
zZ(bgFu8>&IV^7(6K}cyJy4gJ!3Yjn@jX3-FVj(+<LZqyv_P;FR<R<J&JoI=%Dq9Fj
zIojA-u4V+Dj|`pH0!`1$=U3o{$?XW=8Yn7Na%#muQ<1To0eOTR@4a?Q!$GN!vczsj
zSDuEIRdXl4>!RSI-K}AiLh<ft$qXW3y4Cl7d`Rw2&naocnI34=Z}QNSg@^87$#7ct
zRJNVArNi0&NXZ$yY}s(O+qbzN9A5Xfyw;vbbz3o<oxH>IAA2DeTPX-IJuhm$YFKBT
zYeQELC+?6GgSIw4hh_E*btJ>hy5Yox2tsq$z`E!PXWAZHU;EIrd)kJ0=*qb?yiq+Y
z{E++Ers0hBa?eTEH@3*;;j|>0TdZ+7aU>i)qJ(&}hT!qQPT`i}Yzc26Lo<95BDKa>
zOgH=?SGJCL26H4nj5K=!Q$i}J&=P<5Qd7Y8#&~cxM2xdsJOO1lVtRg&RJmkQE@&VP
z^XJP}5@uUm3i6|PIHZg`+UXe3H`ds`_|d0kS`Z0UOV{V<upGJTk3>rjESp*E$}3GA
z3zU;l>9~*_qDI1rKw^X|B_+sRS~wX{THU2XtHVVd&H_UFg+h{`hwMyD6}jlhrE@l5
zbNc)7Gb<!I2%5BW(GuzqN-?K*)HSDCQp`z`5i=#x1x<7c!;Y}Ti^JLWCKj$p^23Dl
zWg(g>MY;~+N^L!1)+yUUOecEyNYi>%G$7u`uIb6+2(PQ#RQ|}biW2Xy^+B8!TkmE}
z&)WZLw=}A$F}od*oQWf6--$*|t#((R6V{29|6b5|l#s9cF*O8Yazu}0d+E7e<i;f<
znQ;A;{kC)@ZDY1HIq~9ed>jY^9?3-KHj2}&7|HgtD_7C=lNe7xw;=G@s*#KZAlh~l
zG_ramqv+pdIk&aZ%zcu=d3Bw>!$}FRO{*J8i?!r;WW7W1$IIU~2&w*G9<+^t?$f<w
z)7=z*<H*RSjlVg6$yq5ZOK#+8j-lgzHvfUO=-qR@y+hVIl3g?Kvc(wDF_QNFQ%~5Q
zXt>>j6gcHOMGUwN&O=DvdxhAekJedEcoRj27TfZAO+t4o1b%3_@O%!0dQ!UD>PW^G
zI|$w<Xw4?{VK>6(KD~ESex(obIdGgH2lPnBzLHGg*@&nTtRa;KCO9Q$qUQqcOm<A8
zci&Ju9<L}z$L5WGkiN2D>XTYB^&NJq*7OX~N~Z;7Cuf75(c|!B<wZLiEh!y3VzVMj
z${GBMovVTQV7SZXL{bdiQ9b8;;7$q7uvT0UDe9kVw~O)dXGc*6TwZ+Rs_G!gSDvs1
zk&9qyjEs|Zb>!0j715X}ef%6mav9b0jv=kuwHW0@a$vn4>u{8`b~n^PW&Ypnrk+|M
z>8*I;dgn{yc6{_~8yU1jG86`P7iWi9>2MCTdqFHTPVhXEX~nnSdE(n<#k2<qO|>P(
zY)!Oa@^v4YrNwMFV`(9FKN>U=wu?JS878v4m<c6mDMZ$gpW6z3hhrQ2x0r2dHe#1Y
z{o1NxVhogkWUH($CI-+&=1z3<Ym13ZkYTQD>!Mjt(umd-vt8q%b-lK}nDGHNhpPZK
z=!aD3F=rbCF#NjAS+J?7b(>P4Im)(KxOpT}N+;W>uiA;|ZB_%g4S5GRYFmow_-#Se
z+*KxnhsP|os&AJ>2~+9N4_qr|QReN5aV`7{y><!}6bK06X;1Er!6J?MK^(2UR5}{9
zY~XG4=Sg5th&h`-J!&PrLY;x)u^Ol`V&qIb43V;IBH;2{+obJN&%WnakFL*{*31+8
zjn;|64WdDgQ~iiO!mkgWoW%VZ4G{Z@-`$Q0iVrHeatWK{1&VHC+)fB%I4HdEq@KG9
zCB%kPfvvA{_IywQ9!029p!C^WiO-!Cg3VM-n$-}uwAs0MpiThWWKJl=9oKV2o{w)m
znGZbtLQwOu{cG)_hP3d%e7o9ZA@@vMBjshT#1PwD<H4ctQg*;{;~HD&YQWmR|GRcg
zut<&7%^7fA3kk1fk`6I5H#CTyJzMOi=!b)UihQ~iFi_Bu^2XhqZ)-FSTn@X7F4)}(
z8X=l<nbyWDV2`nT(LBo<Q=Md%?ne`kZ1r_(rECY|=ivc@<utiVO4-g&_pGI*v;y&-
za8)0nOQpm$(Cyjsl6K*{>_a5(Qaa4Oz0OvavV;A}uRdn00%U<b?x{pq$4ieYWD?ev
zGTsy39Bh64f)}8Baa&yg?)a9ik8k0mq&>F{@%xtDwlTUWBa8a6x~Y_yOsVLmsm&VE
z!Vg*Ujitn8I*5d_<^bn+a!_dz*8k5pziF)+o?9;2m>n_nK2#oi^lppm*id7aGC^ei
z9oG`T{WvD}20*_K5?PniM;;jlz<eO~TrGb|A>fm~+DfIwXV|#RpcjY+yQ;A39wo!H
zW-qIB%o+1N+owj2u-|Dt^#kBgOl9o+&raBpc-u6JcMG;=QOR9_?N}g_N8mdi-*LB}
zO$N{t@eFbexV5m4OBr`f#&UO6&e^GG<RvxzmT9Ne$SwH<q4{m*jAptziTfs-_H0Z}
zsY0gg*_cEkxuDfKeZTOL>ebH0{I_Ctj<#4_Ih!u1Esn=1PLo~lVr@XA{JR_j_cFFX
zzM|ofvU#j%3!0{y+T^N!*j?X3zH7S{ok4)@Q+7Qbw!*GS4lg2hv_0EoHv=?UGSBQd
zVz)G2PXy=`xE){n;Y^ZSeNB3Ftj|J>{4cP8dwQL+XWrKuNzO&g6npx`wC>%UK(_c7
zv%}Z(dgOeCFMaWn8)^n~-Bet9F`G8_KUbi18Cm{fcHj_Ogy^{9#k7%GGt_Ufl`m#H
z9>**07Pcz-gA!~0(N+g=3?6UU+7}a>z}er{y_hi+?1a=DelanL)?Kz<2ta6x!yx2-
zXd7Nk?0`j59_UV<qZ{@8Ru4?tBn+um4$(G8ub2N7g@0rGJ^$~l+UEGXly1!d6f{ZJ
zm>h{)1B?BSSVv&$$v*eQU}?6CYN?$XEQoQC+unfK;4xS^Vd2FldMy89wqel~jHG~5
zD@6Mw6qhs?u9Rn7?B{sRiNK9H{^r_KwS9qFbG=Q+ug`h%xI-F(AMc+3?}%vsjmK;z
znj(XuyB1wrsjy!2*gq~R=E^vpozUZ0%84iR*qViu)9loX8AoHhUTG;s^_|vtDJ{*J
z=+-l2*jeGT_Nz7<?NZ_#e3;J(nD58X@KQW2h)Au*&;gtmgzXJ>K@={{pJARa>KRvh
z?v9sZvdA6py!lGZ;!ec2g}{l5GF7`8e{-M2&pQ5g?*hB7nSN>iH|&OJi(WAK``lD#
zuHD1aI)+14mtK-u+ifk4lm`U0wthgC97JJ#S93rDgAZznmF@)<>Lb7RWCuA`$ZK(5
zBt>GLGApMonmnVIl+!LMV;fmoP8-uDMx-$8&gD!8&gS_Ll_2?t{ZzE&(a<$wREfkO
z74SSeimfbX`xm<qf{3jur_G4V^6GMSVn2s!d~E;$V7k26x_D-1>~=MT)v5Ki-acC&
zZ$mFhW`zxUOCfvPSWY~N`#Uy|O=X?-z2UyFIbOTCOM>cv<QWxGlCS1~gsI5o=$63t
zt98~|PCO+kP!PU40)|&~WETevCN9)podKgAM!-S6SGai`buNfA;O2n5#@JW(Anii5
z#LDL`!ligh3G;CLDrc-U<#u$zXd-Cq&VVV8v+WbH@n7+ROw;-ZU1|>%l-boW4(PhQ
zy?vJ*5p~8n5j;o8OgR(MQa~?~W;!^J#lmPJ;mvV@Xexl`35~+bijrl%oH3@j&Zo0E
zMvdiyV2PmVvyFBph`1HcRL<e(VE$)=HXE(y&tR=TC+f9fSDTABh=wC}UTBa!0pAg!
z(L#c|f?bSm+K242$dbZae{NUG*?k%DEZvd)dPdNlC&!t!1p(#|I&W8lMD>g3!?i$_
zbc%L8&|@cHt+gAXH>S%s{Rp$0wH5T3-Kx>RZIicaGo6aC*&QK>&S?bi=DUGGax8$=
z-;4h6cC-5$W&LQfC|A-ZLV@;u=6gvcZP}XKG;2#MX=jglz{RpkTENJKD7vvBS7M`M
zDOwS&DDgaCE92*4sn1pgXjZ(7w5uy=PsJI%Hej)cp2>3)SXarmSc*AfZ`Z}QH0bu-
zTFKV#!v70l2%@tMm5eXoneRoMHwOHH3K{)vQ(!UwGdx2Axo$6u)2NRYzMrm4&6P}`
zaF-pj1Nud<CB8tQ!F@9b0v_(N!?qRT<iHxTJp#xc@-5Y=_Q)bcr0)$xb^<o1K_V&Q
zE-`8O=xC>V#}|Z2E$rPz%ThFOS7oaPq;9;?Cq$gBq_*4_Pn=S&X_}5^o^$a~;JTAL
zzDpEGf<O$Ve$r+%4q{s-?N}u{Ide)nf)uSUURrA>G%{2OoaHA2Hx9Onor;DXMbF4}
zTG(yFo9&D~^Yw-Wxsq`QHjcCZ6nJ2>YKDm(I-e6^pXMxwdUG+Z{rkUS=S3>cWxta7
z=q^;!v9PtxE~@`_SoviQw88V=UkR)^5s6Gx!yVj@u0{u?AxgIdtZ{P3>udD$$jb3z
z7uR1+raSEVUFTMO-@&}zUR<v3R{I~Q(C(<O24v8?8r`^OWkx|QaLH=IFZV@|WY9FL
zW?O($!jov@q*=|*KTn&rw3;nt7uVQyuB=*CO`8s_DH6nTwco~VY(+I~XTl7~BY|s|
zKG%(?Y^$o7Xbl8Rz^mgqnMFMZ!rFkrlB%Q(JkcAE)jq4MW*ZbAAJ15_UX7basam>0
zu<`o?_Qq<6-(%>$^s1xih;I(K>_BF%v6}IU()`DyO+x9VU$9Ximf_r0y_6+@_JrBz
zL3pmFc54k-lNeXI9t+1EcwA%g*&`gc`enGYnw><PZ7E=Ylz7#|5~q5plb8#NdUj7{
zH6Kr2#0OSdh>ox*b2TlgBNSQgP+8Rwi8G?7iTD<^IgUHFFP>0&jFM_~T5a2rNZ8la
zw1u9D455Bqb^jncqCR$Z^-r>*X95b-F$VuV7I+WL|4(WVYohl4gPn*mqjoQN?A}Rz
z)rR-Msp#A9PLnQioQ`MuEKm9E?TlzoKpT<<$FQ))JD(P_(H9CbB;s?6FY%o(+FUj5
z+GkO7Extk0z^*LPySl<HD^<IwJ+Ldw;D0#)l-SuCzGr3PO1+>T21dp_MeSAn0LOI&
z)wMva-qU&8^?2}!M%NX?;-m>CHD@=2eBHkBZUwG!-V>~KTSV}*nr_s)6GWVUH<d@Y
zE0PwGkM6~ji>U~n_k|Q#y6>5dW_Qoz1-N81ZStC@`qFs9z0TvxmyM>Sh%;-qEg#MJ
zgQuNH*2)#5iMVUl*viptA19A1TQ!>YP=6{^(^X<#SC6LMH~+ICTPsKl|1;8g*t*ed
zFE>Zwle)m8vFN$jZG8=oUee8M7)@J-Wu5}I(Z0;t#sKbVFgFDxR}x~)hiz^i4N;pp
zASa&2PMJmuL<kd74uo%2OMr6EgI2vw{K(xnIs%c=ew535V$2*4A+&dnX8bX&kAY<O
z+8cymi;y-BNt|19pioQCZuy|$<8@Y0cg;l)4)o(W#>&^V^q>4z6{(wuh(pE>gvS~C
zK8<dht3sy(^rPRkLqP*{8eD#k!~;OT{gTbZ*A?#e$Mp3Uk2*a*s>{?l7Lun9AU=f~
z#|a@&O6e0%YTS|4s<%2-wNnCCo&Sh*wCZ!$fSnNlbbg#>XEpf+cjhJ$h$!LdAdwaE
z(H6WR><%^;^H3fb9J2GZB^$+PFT{YJVWD4)KikkwT~-&Zp1G%DSA;`@Crr1600L3A
zA>Ce8b0SsQuy!r_n?DVUxnBGGUHmd{2+T;)kHedy1+?xV&dB%bE&b@S1XpJlCvIy%
z?)YYR1jxk`4)ExNljYy-Ui?ihnUdXCJ9kkaE@~|8-pC$hOUAN2+UIP>!_=0Jr7c|d
z`Se@1EE+S5UYKY3SjGoPtCqJF@re7|pmXV!W7#Rl>B-V1OlmA`q;QQ}t&dyK6WiLc
zv?;4-O5BkY9n^@ei{5<cb2`+;)6U4<vVJUagUdhcvJHWt-{5WISjGli<3O<BCjHRe
z+t1CzHVbGy8vVwxY|B-WQ7n07GzY+;u3sLqmT2xSSX_Qu;|&FcFfxy29E)T#enZYi
z=LOw5<1<(SW_;T#Qq+(7m7I{pvBdZud(`rxxQE^+!(1Uq<2D}G!jv=win_P}U5$_1
zpSB6T!Y`%DJ!hYgLfFQ2VLF=DH`$>;uRR_1N7SsfYm{a)LCq&V{oi&h{@hO{q2n>y
zz5>C5CxoV3xLjVblL66L!ib&92JGI>c3K0byxm}hqwY+A^(@C{Yfb&OLNeOG@VQ-f
zZY-0#llzi8jQBdA(?a0}_^O?cg~Fm>gLvvfm#o#@b{Lu~+Al^QzH|@~E^C%+zFuCL
z<Vw(>J{b%VtQG<Y$+Tx{X?3s0_(>^2E@^d<?2#O8*W-KaXKx@){ioefi=CVuZc@G(
z1e$-t6?wM;BixZm(`L7W9C;M-cSKn((>N4jCc7J;m2T-?e4FalV{y{t;cy7Ev+;B@
z^6FZDw|SO~XFCB4bvKS6OUKhbuAks6U8a}JB<G+lA5UAEb0=N<R*YwByB+fq38u%>
zUS?|!6Sn9qX_Z=lCG)>}Jnqg(u2~xoP#-~Mo%-T>Get6ny6B6y*hVosk7sv;T}Xr*
z^t1yvl8w<XP5Zb(=UmmLUodke<HqLr0*g)_%Vnd0L(TtV%`xbK-d~=#mhntfCMSB;
z4fli}tGmyq-?t78%CiCygB)L{p#08whfL(Da(`}n)fV9l0_37C!UyGtYbcHM6Mt!i
z@x-dUcvVT*<CRo$E?EsydHyD%9RmhZ>wxVWPaI;DR0HvS&-@2=NS*$J|Fa`C1P9pq
zGtsO#%w1W{m_DSz<I$|!eGpFQ9VaE9oeTt?!MaW0vF2(U1+~EFbO7~$>@(xpjb-6o
z7ia!%qny>SfSC-Zv!Zp4`D1xG7qGh68FT6<iS6Xc3k*j*^ZW%tbNh{7JuYgNp275S
zxke?wYF!!Ed4){b?NxFY)U|8Hk;z~XS7VU8XF*<z&Rv#?xR~n!5G6A)6T<8{1Z7Ba
zQ$6^Bj9!W3SZfToUOY#}GrrYRq{<&kRyY^h>wx;4pZ<4a=mW?q_iCMyYsBtrTsv4N
zHjy@HUy%*EWFp(W^opX}mQG}+smrsD#j=UCtVZxZTt1ODpL-vZjVpfdMYvcQ5D~KQ
z5UKs`=&YW&lqEhZlXMC`>$wHjPNWt0-EY{s0MQ8Oy689YP0u~^b6Y==?N@vjZ9{b3
zJvxAXa-*=W@4@!VZk@;^1X2_X*k(PF*X!S`F}_7L3KwhE0Fwt518JGam=LQ9M$sB0
zz|W~-9f5?`z<7~;D!8PUD_N&7Y$J4Kubw`i{#VNhXctZvxLxbh6roTE8tmO`C5?p>
zh4R)`T_l<pF7Of|h))wjy9qsjE0_1hs748gB9%{c{#=-*4+)i?BC$^KglK_iY%>#?
zv+hn7-lDhkv6u^@xb3)bcQJsi9ItQ;wv&3**xQXI>Xfi&(|Zk-)6sV4_FZ-++V18m
zmwXe>)}}C(XP6D7(0fr>!p@0095o!p>IFR{yySd*$(52b;=)8GN|imowZ~WG>|(UQ
zftDlia!lZ7uYdM^yHe|v3Wnr_pU8G`==OZa>ZgX$94RYYgw~i?hwdYHT|ZQbu_H|8
zhS1HsX7r}sBXXjdv0F7FRFIwnm#)eXX+Yv{^u8l@SMv!waJ8A;n@~jI_mht1XGxPz
zQemz#_XFZfFLjztrcLNkl~{=<vrST8=(eSk*=}y1%#GlCvrIpDHXJfTFV_z}>_{$8
zD<(6hl2tCi0{p72j80gR%>Sy%!~r_jJZY;ZGYV3G!&9~W+}2KJo7{t{Jac0|On99T
zYk`N<1%xRi=k<cnh7)zNhHcPKj5~^iY}Dt2^B>!$cnWo}ZB}#VD5-Q=<78rK2&V*8
zL}P?UKR#RH&7<G>_M_H1nQ@6-MJlv(1X9StwkL+m`M>v3>zvH^0X45ces6%lFig5y
zy=(LMDa#8a7H_Jc3N<Iq!bdEgl77OD={{KO@Wqc@hccnDcEN!6#nbMAiie9#$5Z=g
zx92=Lq~1MPA0x766JtRRaeqdj8nEyjQ`gLlS2g2=j<Im^ciD;BxcpXkQcc+zY-G07
zDRt(af8BOEnv+p)!p^9l778_=ji<uoc##m6*~!G-JD#(1@kMz6>2oe9JV31<q88)D
zR!`-2A&Ak|6vfCds#gL^kd`k`X6F&XrRj{lq87sx9K_>8s3WA7yu6ED4TMI+Z{wO4
zXic%lsza>SBrLqgg1-@;Alkyn?527hng1ZA^|+;8l1m&HF1sy6ipY2z#DWO1%XU|j
z!`mJ*?cWoKJ}R{Tq=4e){$wU9rhEfQ&a|ntmbqKAy{8f_zw=MFbShhvuCb$L%T%_O
zF<qh1j)vX+wqh!6)=H&dE2k23NSp#|u^^Ca2_Ns(@iw_9qraZYHa7gobGnk+ClL7x
z|Ew<lrtCU9IC}U_-M4LnV0IV1_SVL!i~)7k4(8~Djn&TeZ}U`QT{u8;uWg*lW|r(C
zDfquxzc_1hd8#GgyJCgiJZr6jPv$)5XB{F#1J&_tkH+NrQ9FfQgnsXhLHy%CKK7F3
z)R#L*xOZV*u&J%%T)r@sF}CpfNcL(gMHi%I@!XPCf#LWoV6(M}7$esgEcE-Pva<@q
z4O$K~9VoG*BX<lOAINcL9tyn3S4bKH4aogOY$l*VzbC$F$3)D^{7>w-Mz<xo*_{yn
z?Jl6AzK%8V)T4GPKFo7z0OskKKOFk(O#C@o=(Dq00R6pQK?4zTwl?Wy3Z9*dzum#a
z=7dMf{0DYEP=q~lUArKp>Ol&>k3fnwMUu)%ymT%H9Ey-M+LeIg^%vmV3mQ4Pw0G>P
z@TrwsKuJB7?UpvrkL~7kyB@v69GJoz0<qI`C*IT}et2Z*657ur{ofY+n#}nhb|<LO
zN|54R(f!eXuzUKj(Vej43Ax4H0_>%1Um}~ZXqLQ`mN>WDp{gx?Dchu2oiIVgBS4b(
z*z%Xs=KkhaY{g5p;|smn%9pZjfsmb|`BuG@ZK`V5&@azpKYl4Y^c#1z+gg2qhZZ_5
zvSYrKSU@8OTb;f|lEJIo*6Rnz=c{2uAk00K>D#EcqpZg2sBL;FBe^RDhR7&@4N8sL
zS+O{1H%5Q+?~+_V$oQ6X5u_!?#sr|Jj=u2vVRdzA6cr-R_vp#-=!}nWsIJ=HpqfXQ
z&%aC3kbvRD>9xF2*-2VuRNaM_GB%cs3NG-L0&)_jur@6=nEUK4G(F$52>}_#t%e!i
zr#3GCq|;(r<08r=rOZFH_{+#-3OYR!i0*D`Yv2CNE}Mzz-P&ZwVl;%5_1W?G`-vx>
zsIwC<UCJ`w)>0~^=rqsPdoqx2DN;g&>+(y96M2D&lwgbmv7gP@>7c}OZQDDV?2PE=
zs(`Z^(pC&-9B8xAEzS!Vl;b;iO8qvQ(=a%tk{RN>7VR!K{9Sk{V_`MTzE?Q7C<;B}
zsVN3}Dt1|)<ve506}9l{DhMB>;0qkKpiZ-^g6fYU#oM(Sk^}?Vbs@R4)oy5T-Dufv
z#&RQPgWZZRn&H=WJH9~oikx#t@M})gewy6XSV<1TBvKi1kxu0<4_CV{v}^nvwb#CE
z3!o5(|G8|mM8Cm@Kp$D!mhOvP#|32BKJ8f4f~?c!YM<`Z$ysxS8to$80lac${NOIo
ztULUhSM5s-;_-F1dSAv!7S6Cb*6vIE0-Xl6)uLN`2dyst9N<R2KK`tB7i>fPS&vAv
zabLz;u&^PG;vKRt+b@s<pG}Va&H4%d&(UOkY77uN>i1Z4;I!$h)}k?ag5P4LNc!m3
zeHl0OB#&I518={zyXV3l4H+dlnknnl4<w8w^_}gFkx>eYE3{s^#)Qeq?@N5o4=2<s
z1@(_EklAuypi+E3?hCgXt$mp{Y_W;J3`lM?K!GiaqE#opabLE|(S!84-Jc0Lq#j1)
z({?0C(OVd{8R17PvMaHU=?6q8PsMUv<mf3@=Fi%RKo5i7l%3SO?G^X?Jf*p*&v_6J
z24;av9-Qd5_ocIi5jBn4*?2%x#MYXPv66VfMe1__nyS;VXU@pEn6tiqJ1?4a<JEmZ
zPbB2Bi-O+pj9u0%FUxjCOQM9QZHj1E2$=btT@~W(;ooT20#ad~0@ijt9uRz}#&jbH
z!1_kvwFnfip8*f?z9moy1mu@N=N06J!`L06MDj(utNFutifpZyl$o{rdO~q;wm(~x
zMA%jaY{~w#J5zzpv=f)A0V;i#>9WOVH}>1|{n~v$;>cgIKRbSRvgB6o&sGld0C|)B
zoUv8=vyDc&np^?yusS+r?x=H7?(MO)`xCqH%h$U7npdJF)VXVpu_y}^KDz}kF~4Dd
z+Wmvh*Eb4m`#^y$ZJX3Qg~@>Z=0LZ1gnat=jKs&Bl{i_)-v};5PwdamB3H;@Z$C(6
zLYPro13estn9h!XMuQpD`Fr+f>?1kKx<J+$<DwF-?Trtk7HCe0)l)#obHn8KU#e*b
zw<1=c7EU<<3xU+Y2r)$^fyY)dO@Zb)tBTxCXwFNwxlZht1sLnu-=V%wGl$<79$|so
z$owztQ2goTbdCr!H)W$#wHfuBuJLZTKUN#JCwcl0J06pi?oNI$IT6pP+=MH_Psa0h
zjv}h%keeW$|9O;^@AA!Qjl$C?L_Iw_qq!vgv}?uA3Ncc5!#^q3R>R^PcwuS|sH`dM
zr!cWO0YDn^g?wH=)Vs<S$4)_Xw<w>ki)yk9w!}A=7Z+ozU5WmjhxQ|(i<-$ie^rl!
z?Un3WV3UPw*>wSGZLu4RZ9RDGX22cnBB>PHUO4QAH#3mi>UM-u4tGQ>SBq15eEuh0
zb~m82$RYQ={pq0G)n@kt6mXFT{WP-!+2(Jcs65H*4`e&Dx0dE}X|&+XpbQmK!m<Nt
zd$Ub(rMKk=(q=<e#u@uSwolhQW-H?bb$7X1AIKO4)v<bPb-dd0Ok<O+)hiE8vvmiu
z4MXmCrhxwH4kYCNX|JspZkQu5Gf*t%Kw=WPUQ%=FOK$J5ODrXu4rG!^O|iijn-65{
z$ZMX}LZcc3A@VR2-+UmGO&U1WnD7?$iiy(G0a&Y=5FUY!&N|e0+7DjEcaP}m<<>aK
z4`g~omzh0jrx!x9y&~Du1C|plwjyj;KCo)>D5QeA+)f%O7=I~VdRVSiV@6noB`hZs
z@rK}MB=tpz$Zs~Sp;S<cIB^dNZYhtB2A>k)3HSIvX*hldm^mO0g6-k6@8OO4^CV;+
z6M%YRYv(_(<I%^WbEusNB;0?|yF3{~#vV`WQ-SPfZ=hm2tzkTa&&wG-kwd}GY9Jk>
z=w@d1xP6Q>;9NYyBOx`B7kZm~iq7k?(uHr=1<gzY14cC-DCBw}^X2%vSms!~qIO<j
zm4uxF)6G@NuIh2&e}Me8K({d@jO&Xp|38j_8yZVHmB4NaFo(h!?%nYg3$SNYy&Wj9
zV(|@&1x#RQEuxk8cVo%30qmXtMa<@Xbwj>cvzOC8?aBLH+m^hX_USMGJ6HRqFQ;A8
zhthJ{%W3;_DBzvLooRVIZ$n+00^3%+oNbo|bacdZxejLK%V|eeus*+>af=F>4ZVEx
z>gbiV$ZoQ=(G%Oo*2N1BpLYm^s#CM>sc~EXa<aK}<#J6PswF?g4S{rG8XM#9uBj=U
zNdn6*viZBWmu(JcZYbouY833o@~Ab(cdj?LmUx2_-ory$U(R+qh3ZlW<jdKCn~EtW
zeH95t;-lm@^&i&xa^i7COkwQamlIz@YcyIxlE}p)V}}-4p94NzwU_uJR(?V6pZBAs
zr2g?)b6?bI^o0LoO1Z1o=mthvsQWZ7PvkKDa>m?9R)ha!HvkUBSV^B-wIeTQoRpUw
zvnVEK^a<Gesmd}w7UN+nFkQ!Ex<-h|IT3$SUerzo)ScWh?39r5I~cjhP79`=)S(?d
z6NvWUoyiGvR-^87>1<Yy<ToI8N&r12CCc)-c%LJQ-Y_Vg4?G6vPuYdq=)Ac0MU4(I
zEb7f&{yXvzG9|e6!(CL5w1rqRzA!xzwOv&M6vl*g+chDUY*g0+930Rcj&8i1vG2vG
zGbnm9s22*lTY+Dzd;8yx7YIyP9qtHngJ+z)8}GoRjeMO#!_{R7E%#${9aw8NowjL5
z?;iWYblS@-aJ2rmWI8)eNxM|F2jZn)%3$D2r_%x@6$(r2i}6s8u*0o;TAP(^gQQ?z
zn$Fh9FllhH6qB?hI&FDoi+x$`(<KVG<rUM33hotIZC^|s^x$;jE$bemNbAbzk=q?B
z_Mz!)jiMbJ?)AW6TNNENZF$-r);PLw&$89iX+JJDT_Xh0EF=@bT7A*rBYH&m!&Y3}
z{e`WYPQ2;Kuh`>(6}f=O3_P9K9PutmUu91SGTzw9s<1v<kN3hDzNWWib7k)!X9UvS
z@RP5rPg1Z!2}r&%`h4L9doo7vK}P5fVqPfO;W05!1&BxYlkY5Ev{NaRL|zTiGf{u&
z{NsC12t8t3)sx$fQ?;+=_!0@nJq&gP$ga|mwFK>0Q`8CbM2EWq5<h(UzpPbn-8U9a
z)E4yMt%v-LT1U{ezPcSJvI*NA1i+>0;Ab<Kph@x|2$b=2N&9)tgx8R;&P9M+gR1sI
zyms-v$mZLtp>*|-oXSPFZXvh}&Z*BaC6LUdFK9wez#+@4r}ip}(7_-<s=Y}K5kYCM
zw~=_osK(hArGS9~wqj*1M6Q;E&Uvg{HK<1tDH~H4exBHb<{xsn+o(+m|GwVdoZE_R
zpD2=Mr(y@<ha`QbOU!iO#`5A`c~CGzn+3BLOx)~Tfs?btfimeLsDB-Mjlqzj3X-#(
z-IwiXpz0jo?`AW?e9!jn+wCpUk-#)A3db~Vzf-fAnTxmeNrkc4@gSNTS0KteHR{3&
zfPNxKL?oyEK=?NnN4aO?uoN#x3px9dx~Zh%Dju<PO5Jq3<@Y@`NeZNt9O1M`h!{IC
zT(S3K{B7<ua7Hh8cny)f{a9n8C;*6`)iYTNS@A#B2X2mH$IYrs#!eiP+m=3z9=@~g
zA0D%F`oJ@!@WDs=!809_Xkji|b7SqYk7KxB`Nn^-^V6A}&@@8Mz+x1?{v@W`Q_4=+
z1%awXW`3$4Dpc%dhg=jv@oq?)@{)eZxvh+CdO3b!Nkyusiv21EofTBJD}f3DQ;`%U
z*xRoI6>>Urqgx2>R6k|E(UjogN-Ce2YmX41uipk_;zo0LHj{pSI+OgkYtJtHr%U|y
zJ8ie;yX<;Ur)y-e*M6_bAQp}tbwk6gxqW{g<Qqc!bW?5ftopM3L1^LN-C?)%NH|BY
z{V~uV?6%vY&o-)}_jCOINxd{7jgWNnPE7eTqgA5N+wodMaroyz^E1LP?XF;_q9lI_
z3b4oOIi~OFsWM))zsA!r2`gBQ_haBm<7<D50T8|O^OQ2eUde<5);4p=eZe={7hbuv
zgmeUiPM{EfC7tD6L!<V<E1B5o2Nrb#mPWe)zkq!4mFzm+TA5_>mg%|dzLNH(SF)oY
z{|3z3uVj<th@fuqydhuKkC*`?@QG9h`mp5j<qrnj0Tx2Sb*1`oMOBYI^h$PeyP)gR
z0&Xs^q@%UvTlR4LwQjGiekHr|fAfH?c_j%Li7p<y^gUbqN~Q{IqaeT|LJT?~g{4)$
zE=K>$Pv(DOkE?m#FTZbf!U<PjHuw_(p4svW5#y}j_0gWOyO!%~!WkDbl%s^(*#<%E
zB8KSe(HG)2raZOw^2GQF!vLO)9^Dj}CU#SN-R3@IWqT^-aQ<|^ZH_r?cSA4*9s8}}
zl}wWApZ@@#2=3mE(I(w!NCHr?tuX~?;}gl71Drbp?GToYIM}yn+#Cx`@-AT-wiNHL
zRbLDum$iwoln3gmW!CQyn8vgZcGs35wx~T?3|pMlJ|ASkq(VV?0d7E8e?h}yI^RO6
z`90ek4bTI+Hmp0Ik~-jH%V{-?bz7ex4dYst*Q+Pj*`Q_;7-){lRtV%C{pM3PqQQLj
z-#q9u%|uBU<C@xQWj&%bnzgFnjk}vP$JLky?<(tln@|gAfgO0#rqp6^)Ln!231owZ
zcpuQCPq0mEFxxnK4#r@}Cd-L@C^p0JgdKh*oeL=^x(oARhCH%JLjtwYsk~}OMU3rx
zY(|I>qR8#~Ej{nVDS$Ivn6R;C?d@06U720LuAtVv#~M1fIUW=w3%b1{WcyegC*sM&
z1#oHqfuKF-u|X&Gm=2sDE#gWXq%(dyr54<iD)wHi6+&GG>~#Fygl~?$uVx+GMeU3p
zJ!y^oI9g8mE%<EUgWkp+6Mq_L_h5XPtsx5G;9?&Jjr{hEf+t$7`OXveQT*B6J7ja}
zgxf~hK8}ePEZKPxpba<FPyTNI)Ew~af{0wgGrMY^3KV&BxL!u*7-(1-m!ij~)_lV*
zYi%x^CvEF1Set)tzY?q-^r{8BqTaW(@3voyq%Kl<+6G(D34p>1Uw->dbcZ>L+s9R*
zvW;Wkek<tsE0DqLnm%TO=FuzvF3{`1SbbeR_u`-Zdp$Kl!W&w1^}Fr!n2#;K5pJr3
zZD=ie?GN#eRZpaHe1tiXas6>|@IPW{c-Y>Db~}2fK#BcHbY`Aw>3HvGFsNjQ?9bYU
zFdigpN+k!99f-WRfc-_yk$9;jrRJPd*a~v?SM^avDrNjgpZ8+`Ma(1kU-j7EH11~7
zsM^7F3ZtN|*cT3FTso;XNiAYa4rZsZD-!UldEj6=<yyDd(u3LAMskra9?TBJlzTK0
zWLY%eASS@+OYty-RgX-)(J4`y75j3$D3TKcchZA8u(5^_-$Gs>doWrfI}~l@!6YaQ
zpjvt8V0PBGdo77oLY{38J-8-4tQMp?IBKgk7R=XN=xt2^dx|Q<YY%3E#c=N+T;&lV
z!c=-tm3vjJJD3<4yoG!mo=r6+i+x<Z_(jF)L<w%{>|u->_Jp7}Cs};GFnPAMu^#t}
zuL*fJ&Z`UmK`88oKyH+i@#`WNxeX6yB10cJ0iF!vu|;8Vn}i4&{)#;n@4)tHta$ot
zJ1+DMYRNsBqgd$dE*1#;wypZ1e;|c`HtT~fD$>~wL2D&<Zi}9F)wkO&J$0}>TVoU#
ze^{`#7)3i-bF3r&9`fAzy9K++EsT3ad1uKU!3E`?4=_byxI1fqGD$D!OM>btEf8;d
zsKL`MAUFqcJxKYabAcHroFC172eX61l}4%1VoIo-7yd2eU>po0q8UNLE2zH;nsgh{
zOy^RJk8i+|nh@JswQ>L><i~9%)fjd1-Wt=A=-@~$*o0O9X*%F7aJs4JoBM*FznT(+
zLwCbDa4<=eJK(A9>9%Xr0*T6#5gybh+feDUKn?{qtY{3)c390Z<`G4DM9tga!pI~?
zqX!mHTQ4p+%*{*;fMfc+y%nD)1)wL$I;Pg1N6CY#-wswmx%1<qN*_5A?41D9(X^+-
zPGl?C1<W6a_8sKsIjKqHN?4v8$Hd!fKZ<Wz;pB|e2#V!yPV4vdT#pU~4+w$Vj@jdV
zAyM@(t24qE_lV@o@?-r{^QN`4v7*qB+E4X0p=Dg2)s{@eoUW7hp~zQ1l!8jmMH3E!
zl;akPf20LdNtH%jr8OtGjh+wcV-3NBY6&>fYn+6O_Q~RF<OX8<gBt4|+goR!2D0Vi
zlwFKJFTPr}OX31MamxsTkg3*J(60hxxE9tf*GgScXM?>&S^xTAc5m|qsdVuR;(~si
zOBEzo+rQEJtyN#Zv!gb(ByOnKZ#5Scyj-u<UU8%s?RV<-Ij@y`J>GF@Q1$!wF_QYW
zM!O-B7fZMsax{M)qkoPhopw_|uztT|e-Jv@2SWn9RijVWfc-I$@oj)k^tMogF!rYa
zL}Jg1-vsXj#kX?H-G!FH{;V1B<cy@=3>>^>pZ!HRjFC6O?gd;NSsmHG#!~}Zk-d3e
zU^?8p+x}L=h`>trmc5z|Wup74_Jvo|iB}{5WXY@9iHB9SY!B#h4><vCsUCBF#(pti
zw8LF&S^VAg%Yz)m@poki)A;gNv(r6gZ{{$Lmt&6bidWM~Gl<s#%lkn!B_(Bn*x8k@
zW+x`0a|m7!y_!h?IW`}6I6spE1RP`>M*r~Q7u;{VZ1t;|!kuJjXwCAr=G9C9soibP
z+u8tuIvokg9*Hp)hbvf!tF}%<CiiE~9v8r*FoPY}#n)K3{BHI{kdjp2wqBT(;qqS#
zIJkkO6Mcga?IWX0pM71=Q4w`xz(LJ#PrjOjAY8Ce!EF+axD=Hk;Zxb$UNp0t13M<z
z<q3BRX?QisQtdmfG3ZVrdk){|jl7Y5aly?ddK()UvK>MM-vn4%i|Do8lgxWW-md5p
zk|S2M#sm5mE*5Q~4t2njXK+VA-bRMo-J-{y?YnJ{nr`<HVUTz}C`71Y@-6EW;hQN}
zVlN1<;n6`>_g<}qkhDC0wmXLS=yx9bhUL^-7xvb^z>;WJ7+yYTvh5igR3nn&S|M<)
z;uka$fA*s{Dg{tZ=dzXK=f~ErwW{!U;~z}M#-br!KyTTEsMdn2d(=%(QyM@z3~Zm8
z;Q2y=Zpe2)KM~^_xdEmFL+sC9W$d5^-7w$+Xem#vp%@LJ4u=D~QN;EmS|;p@Ay(;8
zq1)$L1GdVHI@^gaA_Wdc>@A__Y@uSu1jXHiaIbkgdT1dF#Bohez1!m7(Ib}TlAVZ#
zn278L!a3>Xu}7k~u9518oLE1KPQ5@1OA}Ax7|%bhv-e`s8j{8Pw5F<^<Nf`3RgV-v
zo$q*n1tQV>I6zOr9_(zqV4;qb?WX~Ef`e7KHn6AdL$&&Dot=x%DST75k3=))!vh{x
z5X0)iwesUvvy;oOVe&*fzldggi+v&jK8@j(!ClY~@GlQD{<MZ}KeE%sSF@W55r+6{
zru}zG)5KBOhh+)>C;jAFu@w9BE6v;%WVs=`5{R}ox7x4uYHNK<dy_2)!xF;4Z-Odo
z_uExbg&Xv!{r2w%pJD^92@r9&kmq+AOCF06vwB^hG}Xe1e;=fUb#e6HP<wWZFZa*Y
zK3vRlb~C6k>dqg3h#}-zyte|i)+7dv*dG^3A1c>~ytys1xVJCzQNjKcKUD{Yrtr^9
zZsvD1%$_2<;Lm!*5rrbkTC7<%LvF-sg1EqqSoHG(5r5U35is?E-_l_u_cc;Na0;x@
zzX>(uflt`Xvu20l{O#-Ovo9RV&d3fI#0k+jln%a3@oNtp%5G%O<K(q8KBW$OmwoY&
z&TCIov+PiI-bD&%)YbZQ2YG3h2Vi!@Xt|qh`{hGns?NJ?vEop6G&Z<5v`^84`WmI5
zt<=}44wXHm&YQZ*BsDpd#0Xqe?BV$HS=QTXfxtdyYcy2ilfC-s+87M>+FpA^FBiV=
z#$;QkAIO^I=H|z1t<lO_-J!5EueB$lw?Z!|6k<T=1zC_^6E$`q==$|&gODk5rhHxC
z>#I<JsBOK*YtNnxSnX(DFePiOg#p~gpAsp$us|1V^Pz0-`VGD2W!#`fsF>)z@leJy
zNE%bC0?8S@HPH6}`sNrP`v1J`&~&6Eur4pP2w^U*Kc{NME_ItyNrn{K8hD3aXPfXQ
zcWvG}L=rCam`S)BM*}n(9QebwCvc8j9w-5y4`9WT7y9lzlnG$?$BtIG!=a$opgf5p
zaZ({t?2VC7c)nuY8VNfnOuxB!jh(e%eL}I(<4^KJF-Mj&*WV0^ju=H*sD=2r;I3OE
z@i!UJIjZ$#TW^J!4y?}l1?ev5QJ&Ss&Inw$F@bJ^L6dhPzzqzebePh+O3Ev=PpjkN
ze;cp^@#KN0ERX3L8ty4~@X)0!r+_EEW=AJieL)-SC2hDJt`RnQ6&w+TJR;YvA9l0`
zA4e!Q;!JIpJb092<E`ipau@BGMl@K#%j<1DA|Q0z@o3pKgxL3vmJeD>hk8OI+DSF`
zALyxqz<M&^P!@_f2rk8%=ETBWW91pqPU)N7==k3gxiN-PKArS8df$PQ!~5|>4MMlS
zo)JY5H`uN}j(yO~*>X0p-obF~r?nB3Vd%43@5r~VyYWO~9|oq)#c{vH+c|-a$b*l9
z(8T3X!(>h*@Y*h<K>avSO2K7>@oNBCot#fZP7gk^3jvIsgS`2vX06>bUHBL-s`IDI
zBp|j+F{UTJ`(3*%NLYJcv|s75esunS*p<Lzpqj^^Wxvk4@kHDUH59{(>NoKO3-3oE
zbyWzbU}O%<-$svJxiPyIO_SZ>U+j0HUB-LxHoPu+pqJ;8ZoiLjT&<3+;)aHU{)emS
z=P?(fUV`xE;)+Oy7K|;PX7UdK(M!1`KO(yoFp^1+LPPe)KyW*5RJVnF3pP3TxBQbj
zWQoA!?}$n*qvSLG^Wvi`$M5QK3&FPb7qO?5hBKw|xEGXk`z5;T7+jFb?u#It$p<}q
z9y-y#Mf*Nn012#mEo0cV;;ON5zwlaiFYQZ?G)rE~m_JS-z4(CP4q*?xmd<Q6VXgpJ
z`dX&Rrh3H}U%RBJI?lPDB`dV&6_~YU0eYLu_8znLC4o-*a>sMdmcN!A`Xx8bd^uo8
z#^)8njc5oK?1O5HEg$8!;9(UlPloCAq1Tcug02V?^K0o4-@a{^J*>9m{_8DwlYEXS
zk=HU=Wh)z;+tix)c*^xEYt?x@^=$g>k=N1@{m)#Zg+`TtL3{kQOiuAMiY^`2sR6<p
zmgay2uW=jLoUN~o(QOv?wbwF1#p9w=IQ)j7wp%J$wqF;p=X{xL3}9Tez3_Y4p44P~
zv(q-kLj?tY&YlV|eq-nzw9PRz9L{^Kq4u{&!!(KpNbyX!ZB>8xa8lD&WN<C%i0ufP
zK8tHbi(Vn1LA_miuE)?-u-53Nm@7_Ln}*qnb5cqt)gk<LqA=Y3T6P?#vSW#<?hz_D
zlya#Sq1`>0U@Pm4@sh!Z*?vLYvjK<5fGo;vX|Zl~Hi#l87l65BrGEPK<@~#-yDcx8
zq@>Yh8`KXAr`&nk3PFXg@v4o~Xg&ZeD+%L4obs)#RrlFz*eN{6>PJ=$vJ4EtcgBR4
zpZOE<Cb=g})p{5sXOza>O_>Qle-FfGq<8R>Z#p{l(kE`ZJs6aPF@~GBXMH^s(AKSa
z!VX6#^##mBM}+P;**@%OylQ@CXOqo{CZ3U!`?Q+oJRszl9!sP1AKKgTNDN2SbAlch
zui4R+FXx#n1m23h6I-b9X*(f;G9Hfz{6Tb0pze3<q{ds1sOzrAKN2!{pQFb;C9?J7
zT>M^*8sRp5T9_5OrY7zE=yRgW&Im02>|8br_G5u{H-Hg48&t~Sj`mZn*9z6ExZaX!
zL#?s9a$WjRV_0|(uJ^H>Q-d93nCY{RgbC^Sh!?8OuB)9s)+>a-VLQLL_@mGjv-*iJ
zNGfpz1iPRfi43Q9u6?Q>N|jzO6MQj7`(K~3OZptfhBFh&fy?@d8tro)O=Z6l%6OIH
zE`3D+;Rc>@@7J-q@fl}jEd+h$f7Wfki9a9v>R0WmCZmZotiN4c^6u!B(ooyAXc7rx
z?RW76FC-4zuIqD-8vA_^W^Dc=R6RI;#dzvV_IXeb`IFN+$OT>IlFida%s+^P^}}wz
zch7}ef#hgq$o?3;z=K`Nx-BG=Wg)H_e-ghSfc1{BdagKVe~wpF6T!}NH!#5qdcyu9
zlDem2(Ot~%)q3*l{CWFpG#(g2jecLm=C+8)^*43+=vU1SXV>EA5I(<fIGxHkcWzA@
zqa}ya8H3+kDn&-%#si15bDEAh{*umeX#n9S(v9Ht#dvaS6+B~EJh|wI2>hkPaZDB%
z(DK6>>vdO0&M;Qrmk(!W{PRUy5ub9Bpx3eo<H=3DD2Z1Fro2N8#yI*=V9GO5-}n_1
zt`)!J5NL6ta&O;<52rJK{wIEsMX`N2ovizxvNiF7k9QE6VQUX(=QHxKs{|g2uG6G*
zqgK}um-cv!DOZG_)CuKQ1cxUMCqdxx#~-)#f?07R1n$|<vbV=GbK8c1fsU-#zAhNU
z<i6jifi(_PFlSHNlX}lJ+BO|lX$C*$(T#Wr41!P?!=E~wg*-W3P)QO);lsDhqI9L(
z8e&K;ijh==TeiArAcA;nyxqbb-ph4Zv)+!v%y$IpTqJS&(wjCo)h-c=Ya+X<^>8M0
z;Pw!;mD)sk>S9k>hrS@6Hy7jGdgda=QM^Y~^l%_RKCcd+M$+w!F?HmLU$PfOINU6|
z+;L)WH0C?!!rj8g7Y8YuDwd0(47nYkPph+cBC}KE1AY-@+?WljL6hGniOJ?x9X&Yb
zM1&d<_5&m1@UfBx>!w>PM`PT`N163%tQD?tHWrxIwD^9uPQ+539JDFzzeR^?+ZT9s
zp(yLJ1H#Mm7`Q^XV$&ki@Gt+}4(h2?9<xKjYS@p6!y<8i@3^P9Jt8{QvLDad(de(6
zOacQoQ(OLCY~wApt@pFnnfEclLrnMU?%Vo-yAm3DJFbbM2wRcjd-je{f`_09R|6aa
zQz^F}9L~G~T&qs-t>-mC4RTVjxv<T`aeov;-Hrtm)z_(j8#=*)y%&$|P3<X_$LZRr
z&^C10`|*&UKaCsRnHVA!ueo8zq@p~UOZv65>T+^`Ouii9(dBSgiP8o(8=sQjkCQuK
z!e;)l)6PXR&IM|=*+-hRaB=s-+H)c%GL@?eeqkTSN3|-r=VQs#!_z+zhUD53MxGUe
z{raisSDpU}`YO)rXr6LOUkaFD)j~klWsx3jR@Hu`CA<Z*PnTVZArRGsWb$iu)X(DQ
zhlO~H(w%kq8*!elTqGsO!NV8x)#z}0q#yomJbBzXy0vRU6$cg^9`g3P7~~ko;dRlk
zX^Z_{6dOk5#k7Eq_=YIPe)s!p&OSe!DU!n8i7tlya8skiA-T`~5agJ;_!^4x_?w%0
z)&3YAIwgvBI}jmk(*6|N@yYMlooG$YC2Y>|H{Ehr3svNXzlcr*7PvFyy?E=+U%mF%
zc;KXA<=+n&q{OM%-)bw(8MzD7_If%SQ!YgN!t2@nl?;JpTN3XE`rTvTf!8xWi>G(0
zZedGbzvN!rmw$NiRr})W*^%#YSIb_{F5>ygoO^)1o(@0iFu*09cPxKB9h2YQY+qKN
zJ3Sm`h2XR!K|T0-cAEE6<0!m9Rz{2Y^C?dKp?G>0X&tM{s_55OKkV|HJuKi`JwyBI
z07uLOw?Et`y%N)!z~pJNA-$g6eKFs0U#!?8@s6N{b+2a<3ddneUI_HZV^l2i`lw@D
zCmJO+Fojv{3ANeQ+f8O=TdyDRcyeRd*P@epih0!NJ>~xTUst=taS5f;4=_+q@b_ep
z-z8$(B-{~~2(7EyQvuA8Yp~7Hl&fvk8Z<X}vT?pOisHL4C2SSN*7^mjIUaWx%PA-L
z4$TMOQk~3Yj+TJuk~!>o#C8R|;GxzUqkCenwFyOn9g|CKhsd?lEsP$+v-|aIXZ5C3
z_jnS(wD!cO=vauX=Ku5dyoiBj!FgS$@E+;P58DfZqZ17{d*iEAWhh-QyY<y>cdE?A
zV{fW<o3p-Xp4s=v8TEP+y-0p#gE2}!B3RA^(X_$q4BALMr(1z1c_?Y@q&t{&nY8?R
zrorMajfM;P^Izb_RsEdQO0IkI@5h3QQ<OLf6dQXhL!K&sN&~341%01BC&hjE=TsT@
z0kvxB&6BZ4OJn|Lb}*Jk^36OHf1~6m+u``TM7|9>65W%Z;AjjCWdnSdRx_H1<h3zp
zZ)v3DNiLS=Kd@u^rK8A252p8a&~E-`xdEI<j*E64?NhXOg2+t2oe<ur@5b#1>a4vd
zj|^=m1Fv>2)d8|XnidYOQ))=)uRHF(CqN4)G1s4t*6a;G4Bl63bhEyq&jbcNg9G;C
z_|OCNoe7>594w)n%0Z(PiZC0vG;g;L)yTDAcCH4E%fUyBZ{|NJ*<8FSF+(2*B#*YX
z^D#YB1NiLNC-GwbJT5O6v|o_J&#^E*u}|amj;1Xgb}=?5uGbVn^9vj9^p~Ow*VtZ;
zro)~y;a5R`mab7hAFsrliu*YIS||<{yHRnAHsl-ht@s<wW-iqpAzE0(Y(-fAtvJL8
zH30K|L$hnO-4JQ$36&DilxT(mQLktBiL?=F7Y>H^d#w#z;y~v{45tr7K96p3{HI8#
z-4tmmqbVdXXMYf0BP^9$f`pjO!u@05(BuWj5-YeJKcsZqC<`X-Px^vu0`quBN5B@s
zL3$AlS*(B7aJS_B4m6A~cQ-m79>{sv;9p`kix^&p>|VgcgVc3(75l5GiByYhi-}sj
zl8}~_`!{vKg~F4}u|Dlcb`-;!APmjEa3pc2cCOn?j$}ubD{}bXGh#mw@At2%v!zEw
zY}G{PcK1E|;*oTmrwGDjN3tvP_7?k6d`vKiEkBYS^Ut=o+n0}IciLU49)_)m9|j^S
z6QVtMB%P`4iMx~IvTQ5$O?AANyhHX-K(C@RTourh82mkZ_((bnr2C6-m(`1d&#sPZ
z;xi`(2jAKw8D~u<IX`1vKzu|aYOS|*@eqY%9}gVi21GQEdf~geBboSu5fJJ72|eQ)
zh|AD=J(p8@x~~N|u88cQ4M#E|ij+WHyrV4+T7EiW%8s`EJonp^N0Jy+p0G_p1&Wf{
zQvqNcQ-fb{H><NPsg74ev>fhkw#LP&j`UG}71|opSmAuOW(|6Or|r-~%XVvtR-{!P
zwq5aJ03CeJT7#JFtZ}e!i&vxE0V(tq*goEAyVWtK29|wN<nD>K@NhWH^MR$iICZMu
z<|cbVU&8IlyO#FeUXiTLsnV?_LJ}FT!IcYSspM6$z9X4P*xt*jz%C&jRCES`ZBU?|
z>mmT65Ga3josDSaZf}3aN`it6$i8#Yvs@zsCpy_#MRmr5seDWyu*`hGpfV93;6{cz
zXDV>$=DM>_UAe@4Knr$jrU{uA@GS*2vUX5NH+%i(LjejCi5(Uxa0+k_ts{a~i|hZ;
zj%xja8h6`nn+e479;9P$Em9owS9qt{F>MkQK>kg{WxgGR#z&aCh3cA9!mrft#7p#Q
z-F6~StZ!_zAH+6g`;<n#l3d)!PA=k!Ng5gFBm0p6!j{}8^OVTYHR-0N_cRbUU}43w
z)B2>F+;OO(-xsCE=0Bj0xW*=#j?1*b=1^pkASa%Q1M#!sKkcWnCZ6lCS+z_qaQ2~~
z<7C@o=QIXg%s)~K&pcvtK@v}GXCE&DL-4_5t<j8QIP-P=i9Swu<O@N=F4SG0#-Dzl
zzo^e5{~M-iedhIK?XvnAE^=hpue35fH%cF87YTZVIi6|xq)z>HpgKC-jqY1hhL)2m
zAIMCQ|2G2sbgDUbRnPd5vGjhc$rvk6y65DzSm&OLn{EC(q1iCNez>l$n_1#j`+dCL
zf(n=oXE%gldyhxV;X?a)fSZ3eDN=8$x30Kc{UHYM!b@N6v|HL??S7VvWK`<`v7Fsj
z=Ok}|EBs0AQ9;`sVLFa>#r~|%u@m^_w!5)^u)(ke{~|=Z(mKHJ=?7GJ@G$$U0Qd%U
z+w1+n6${WB`&;zQ5=!pfZ)B$f2}%frej}aR{oL-Cypf&Tjs#vZE?+$GMt0tIbLm+6
zMt1PmDp(PG^+r0VyNeANCYQaD-Cz-r-P7N`^hS0<<2L4zy33=(i?5dL%Wq_-TzURy
zSjAQZC^qs}zG4r)amnxY-9FDN1s2`>{6{#(JtRa3R^fn!Dy!lH+$~1z;rJW5n2f_3
zK?gi`P4tP$o6`mB?ONdw`jAHg2OMAFK<ffbdq;hTJsx!1k|z|L-s<$7AMz;0^;{Wd
zHQM^c=cMiXT7bedR<#ZB_oI)Iui)$Pcb|)?8}&`K=w`Dg<9UlS)gqf}lLqbiwUg8O
zQ$fu#p0%6b$i$8XPqFtAf|nP7XowC5s^~}>10Y6Yj!1%Bw#K`KA8^BK7H~AQev!`E
zj-Z9BV!Es)9;o-gUfQ+T*SlPbUDx3Ohpj<sjucO5+NP-|g2^SI4vmjAC@j+5npRIE
z>T=GWjQURE=e1HBT@lr(<w>Zu*SK_$FU0r@SljkS=kxCtNKInh8V(u2aVP{{P79$z
z=xv{#I!N@A7b-|R1KdAtPyn8zYSa)4HY)`F$$1F3L>me6y@<$WB@K2fS31^tS@7XT
zR*k2qwqs*|Hxw_=l4OvH+C&kdZd0OE4&xI>>v-o9xE+XhuHb>}KeuUhRNqbbb5M{y
zv}LOuiq9}ApaBnR(i;x=<dTba<GmhVee{h?B3E*)I7DYOs@?M+xC!MgVO~=Y_%V7c
zy5aIswYT+!`|Ti&9)BZAm^;5^@90%AH{{mj6B=dvKyRK1EHxk}wJS7FYG@Ts!ym;{
zEkNv)K1P)`)n)J1#@dY{__XFO{H2-v_oI1nWD7nC?E5onLYY-cZlR_8v3hIkBB#8N
zX~3Gz_4TLG0VnwXw%PdmvB!uf_%QzNAuF?;i(W8(q%aiwNJy4Q3P(0uE$O~4rt)LG
zF3+EXX`I)#;ttKW@)SR8;ZLI3r>EjSyAY_sf|J`f?q)HW^JkKyS>Q9}C(9*`lKtsd
z@?Y5Hn2)9=`&I0nVYIbZG+36Hv$J1oo?Xv2{{b|v+Je?rOTGOj;QZG|?W#U@iI6-0
zZ`Et)#!;(X3#7=j+-JWNorj?Mb?rP)4OxNf{yy4n^>VH^1bg9FN`m)!;6Q*1NBm8_
zO+j--`$N3-OiDd=OK9M1Q!S68>inZVATgl(5#Npp;k+C0vRZ$NLA_YDJAoFvFH<Z2
zInW*{{_=+xU$?vQA-vg>W!PT?xX4|dsl4}wZV7%JAsBkt{u&S}TtgwleGQz$cm6b<
zet(MzsYjJ!N3&zoSq2>U3rEvAoTOw*UmeX(PU@Fqm3TnSF*SL*yroC8Q@*bVqyOU3
zP_j=_-qMyGO-KIx`(^u*zC)vqW4tXtnsEYG9mXnjw*&j~(Mx_DQ-GK1Sw}M|fcUYI
z3gO`Ppa5~lbO*VW(IXAGi#;T4ynLgr3IwKld+cF>YG&>+V~~OLXhu0ga>t8#7aG>Y
z=zR$jL$g+$;Rxij_=uovVQ-UVl^Q;3!~Mc@`9CfkD&x3E)&+hIp3B>w(D#&Lbi;ea
z){DZVp+=AJH9c47e+>WFpl8er?Ch@xecWPd8}+UmL6mb`xFA^Ju4S777sTGA!+feX
zIzP5&ZF67-2XAV(hNGEG*2Xb`kl;JKG5TuSMxm;$(R%(oQi(Ndj9bd1Q??`E;)d+P
ze~aL**=)On5*8uO1Z$19^a@F=O|Q2_UGa{^*Hq=&t$vcMvqwmF5W6zs=FUlb{%GR+
zYyQzX16{AMWG@H_w3mAYbSo>STR@vVXet+9jbbISzW95fI)B>o@i$q8hip)UCfy`*
zcp;#3m4oGsM8l#Va3#ThuDNNKl{Me4ek;4YP}NUf+G*I?-I#{uxkN_eM4S-L9XO#(
ziB4lgdLp9jJDLe$Y2PM5;Bp`k!y}P_P79fxh{SeKFxY4u>UJm)cay0du6^h(nRY}U
z=8G_pqe6KrV?chQM(NSag52O%?6_+qxx6K;M{wXib~Nd=(6*LwrNM_F1suICV!9Us
z*Kd9EY{wRRN4pY@t*>&H@rh{RzE4O2Kd8~$t@S7MEm`>OM}pkvxt;Bl-sbSleJ`FU
z+l;o!PRA2m7q26GUtdxelSOex&m9Cg=IqC@$XZ+NY!GUM8!1P%I%vf-HLH&bDU;iW
z(X^66;m>JYS<1kieH0_%gc-28Xw1IY)@C0^ci&lK=Z|I=ofLwB&g7F|Y6HkI7vg)b
z>S&)v!<xIQr>ndeZwp>k`BMCilhBA=7Sf&H?zCTNu2IZ#pVlLL059+u61omwsnl~6
zItAFM@AlusszL}TBvr3nRg;lPS9^}wZ#6V{)HN+8JVrf3bJc#QFPf9)cf_vimoRH^
zRtF;7ost}DHw1>VlA*NwJi6&C;!<fhgMvLH_J{cEt8496EZTI~xQE*>(9OJwn&gi`
z9GD)Z=tP|HNqoUIX)sj$O#J$H=jf&Xn*)l8TKKyw3BVS;+Ws6g;Tu1*`R1;u+Kmdp
z{vzZ#{5*_i+U}{}RzlnV8W^zyxM9bz9Dy$T+tEukziqy(`RVzc=(7`MZ!R9yW0+sR
z@a82yNGa1qG}oKySms6rG7r3&ovJ>i$!~1wo9WP}vS<~DkuScP$pjtYFS1OaTT>j}
zm!i4XDuO*Me=`#Sb`$evU)Ce4w4ANbw{U+}(1QZb$Vn+@E7cy$<fJ_m*mVtKA*!*$
zRYDj;Hp`4v_wbw9aRt;=Sd^>AYJo<kbtmjKL4qw+Lc`X+nQbIkcnXJk<jo|WbP*M@
zE(Qc&`~<%ExZw0*OXvPr7vy8gD=*p;0@UOcemM=+$Je~}h=Q*PTgL*Bv$o;QbVNP5
z*S;Q*FneRt+^BwbcH#I)>M{M$&qbSDWt(CE6>?43Q-bRnK=^RQHovKAhzPB$=#SH?
zz#o9YH;AqnA4jZFpV#7her{U>Zye=4p?UM0nK(oo29Yj)dD$TtTU~A0vN)G$!U)5T
zarIHU_svTWve=kXkA%dNE_Aj~fKVc9)3~sbdDXBE(GW(>U~pAa6Qm$STic^1uCa94
z?0NN$?Gql|8R+BbLcT(KAy7@xUo4QlnzAWQ{Mt;T`pF3h0S*kQFFxzWZ<d3>^6?pA
zOB^4AF}%+GR*0`^u8)L#3ELD#_j4<Wx)_4}UQc>5t%?c~Q&nHok4$>nx-pH<%_UwJ
zZbECimRrF2@syyn)$g=@@l?x9c|c<!uN}*DT1a3r_Lwa)j;a@ug(Qv2p%~Qe?RL1f
z7E^A+BbsT7BoM3}pkcR^P=@w1BIk|)dn?+~C0E>GJEm^AUl4o3x_LWBn*!5P3i0Fd
zJyH-;`%Y~Xq~m1woCs(HcPF3OA85ckVT333<l6opX}~GXlbs5v?hr##ve+xQ&kWgV
zeYTTwJnsv^cBXuw3sd|VHE1TF_QwIIC!OhM7b(#-LP``(T*~`T?&_E(KT4g%#o_Y&
zzuAY;IVtSFZ|4M{;$8tC2>@k+%4Ct6)BAR$E&EvSxxtKZ6+IvC8@Nxo;`Ea@Gf5An
z_-xu?p6Kv`K5NH|1DbytLqhfM8nTNqB$S$Lu1liZu2DoxyBzz$3+etU5Fa8A!mfxY
zJGsaFIvR}+4BCP)-sZ-*-^8LqR^=`iK51HJ*qBKHk99m?c}NGl@tWQvNZ@KVV!so&
zA?*rV-d$p@2QCAisr&cQB^q^>ALj7Iy^$)zFkqb1G_QRaL=z%61L?M=ZT1KCIhOK0
zkz-%u+ma_e%9}|zg==#fgsvW#w*x?dV8lPw2<jdy-?KYHn3EXE2gRcOIfx911K+m0
z@jxCChW{76>HLP>Q@@Rfp??)#<SB$1Qh8d9%X#~o=5-gl8ZKlr8DD0a-3W^e_=TBt
zbXCYCx+MOj?5{mAlO62+>?UH3sJ9hwxq8_bXOav6vQLlNvUngOoiTrDCOhw`2H-d2
z<ujQszP_Cl;4_)5-{kUwOXTU=vSKC^I6ONjk=pUh-PxVe&e_VDOX((09F{#4{h_er
zm|HcI9o|%{p=J%OKemSjjq=3k^H*yy9Qq#jQL#0`(}SAORji%K&UDxcW-|QPW-?hM
zRm@nYFHpPqz2)&389Rx?(&_@M;rSnf)e~w<;+-M)UlRs9eR=pr<ZA-Il?WNz5Z`;4
z8*Xg1;n(cbwo%C8Fv|$4_N2xgbyE@*H^mIKvi_e6<XoO3BkbmY7e^_|u7)6It7iac
zjK4|TQnsyHA1Dym?j9J~EMj^+KvJ1SdoFrp3D>()vBafmSG*%?Y|>h5f4d(^Tm0Qq
zBoA&!^ntv5m;>D>_!HY5*y0j9|1Ju3jg~O&RJQDS(KAOvng5@$_m9o8%KCm+j3LDM
z8bb`n7-Jk>h5;UiA;cKN!w~v|^21Vk+KLRrFx>6k(jPZ%-}Y{yRSAb-czAgb5wRj7
z0wROFjE6yn;o)I;2t`5+A!v-@7-9?|#t>r&F@zXG&S&j)U*CG(zYa06*M051_S$Rj
zYp=D}Pp$#WO4D;e<(MrG+OZwSO|%t31QNVkqkbRWWqCa`H(E>dr^l~;^{>{dX>e#V
zt@bV1G3n~0L_-(73c`Yv5$;gkS;7KmI=<>|L77J&2+H)BG>c6I;`kefs6p%3!msQj
zmG~MB<~CvuFh6`?)&V;?xRA_D_?BUxUn@kYoaXK_>q>}_uNQKx4+`|bdpBsCz<#|D
zo1d2(wHZk`*VsmWS=$u&!9yBtv*y7U%588<3HQF0eYRB_gqtrembj+cHW7B_FKl}>
zru-V!*o1T^2Be`aNpOAXm$p*`=xgMJ*cHvlKL{VR-SL#9y_2>_Gxd~f;J#PLA>?v;
z?H4R=1Srwf-tUyi#Rr2H7<lcFnqo`pw!^}15#Q4hy+nR)wvS>O4W2*xs2bM5*p6v)
z;=Pkr>bSaMEdP_82o$g+;A7+r>!iq?ZWkU?e@Y;$;mW54z@u30OpINdBb~w7AX;&i
zor^AI=o>hT&j%T(i4785S1!aeUgj=s=IvsD#M2|Cq5D)wQw1q`G+zq35Q1ozwSF%z
zvnwLt0%SLk`&<OCC--2%z7Xn7tlgJt-rKy!-D|G~Fyf@xIbTIvzqeq2xTc;dPg%^8
z6NX$W`UJ9ECxi><?79%^qk@(2SV$}`TJw0hW>tZY<vUT32(oUwDT?Qt8~c)O`FjmM
zm}|Em1gsq8&Vb!gfAqj)_%7kb9<m>W8_dO(%f#<;)4WrfCovK@kqg`^jDDT=lc+M6
z=@FoEFQ`p0qur08_}1Nl!0gqGJ(5Sk9(XmeIegRbBY!oqA8h>TY#;Ti&LmV<kkq|g
zPul2LGqzRL*6H?0dnh1szWxbM>W2g3pfj{bUd?!AH3^c2|1$h{<c>#Q%@{UU>;iVE
zF#!=SlNGmk?A44pyPN!K8>?qNBrffb2c{gRWga{GL<~uOFSO>bW`aW1B6}(x+Ns0R
zX-@|b_p<}l^)^mG`Z^)g_?U$Ut}~|z0dnd%Oji>%V;D7gAUbVw%qx$w#z~d@3v5aW
zi%!zOP1PK!YviBHG|hl4kJxjjznWwqcO!a6c+`=|*PeYflY=~~0~xnE?75f*fex5c
zXT(EqD!x2ZebHEw%uK_!HaB_&p4p<t{0b}A%9o?wtW;aCa>JKj75ESDLv1rJ{-z>*
z(H3Y7lvm_X5fpL-Ta8vYF)o5D*&?;CW<np=Ti2ZP-PWmFJx>i9d=*AO&ZotC{X#>P
zEzyvbSki523A4YV<X#qkH~Lmy9)IUC;@Jv8B+;mU5{;s`OYO~;kLOUCa)oWt_I~l_
zaR#lWT~dmrv-Vi1JZ_JL7!E6Wk9BI5NSlSFiX)+0KP@B+5%`f?R7hczaMeLErs?5$
zh|c{1yP}%nr2(6SE#&o%p9d)N`f~id+I{5KibO3;Z=Ig%+t%26J-xtvW5drErIa7R
zHj1VVn3rr*4AO+d*yiYre!CB0VT*QqDH_SPY8IXifIHr{=#V((-)(z|VxBVZ7+%{u
zO1stVR5tYL1<&Q;Y~p7+v7N#tReao~A>nYzQ*(ET>ac@7A}d!H_asC!9wWLq_t{Rw
zr2PU1zbn`Q;ahTfIv5yz@(x)>?2t&+*w<o*W1v2~zK;YHzOocf_K}eC<TX@Cv7;L1
zvklDi*vKyLYxFM#E_giF3*TjGaj1DMA?kKgSm9Grz<PA5w4L&t|EEV@w&vW4?~Hb5
zXZUCLV9dzdS*<K~6t11g`|ex-s&=j1c?~+(3&dYg`()Stn8WX=UDS^(XGFVml@6M6
zq_(7bvzG*@!Y?bAgKEzFk%_KoO|I+#q=!Cyt{HIJdX=m%1Q3%JU-*~tJm0-*Jr~ne
zy<gazhmU_1&kJ!C<z7y0j*&6`lf?bUk$1liB)H|dwd$Lg_FSjSbxrBBe@lwXzmbaM
zh9Efa!^x?~zSW>!@`ivfEwa0~drq30`Xwdx<#YW0)g(q%mD>-mW@cZu`rrJbnC>@Q
z?3T7d3fG26O?EprT`p=_YB}BSoh=3t0DVUorn;n5d$zm!i4{BQPSy64cI<-TkL_M8
z47n3n`~kbK6MkN$*;XCA=%0NMT+X*94$mF5U=ME1E|_yY`fSwJj8`_a+32n5OJxJV
z<&wGDYaBfkJ%&G|<`IYJ!&@`v<Z=_aAc*SLPe~KuxjG-+nu!Rg&x-E4JZ5Xg^KsJe
zZtb<l)T_iINM9O=lDZMQc>GN*E6n2ROIC<ZGSfb}HSx|Kt};*Qokt}?(Wkd2=_Xxb
zxN`^g*miOrS~`AfTFE89vkCE{BEr~hVnBT54-;*20C4jr8+}SVp`&+4o2i;t5iPdw
zmTB=GqM_Ba>06W7!uJqWRh}7{BtAZ=g8Z{F65qHz7jq;+2ivz=UU}D#QEsR+1v*c@
ziCNJYGi0C5j%DlWz%Fj(qCk_EqQUGfBt2=iTU)E5?`PZDOY;PE9+%Pu+7D%XlvW+c
z&0jpnYV-yZ1$OTMQ9Y+&j@3mgs@qvZKuZ}8j@shbeQ4;Dh?UHpOM)bcK(Jw3s)>@W
z&12OF+z!}F@#nVXnoV0R5lyi5k=b~eb?!}#0g*3{@EMNbydbt9^CV?POCXOf8|K_)
zt=i^G=39G=My}qX6$E@$%7TMxJeqnU70Pdk29-Uz)z-5$6J{foP+Fy8bjYPdn*9MA
z`H^JdB6JPs%%DC@--(1TaN~P9=va?K&DwzC0v@SV)U&U1wXN6m5U{*N+)3M@2Ia_O
z8}+o9^s}2Z5b8Iy8=mW6bF^>;I!E9ZO}ZS9U3@ESYoLuwV9~Y(Qbc##c8%;l30xki
z&=DWd;#wOL6#weo4RWVOPBmq)>G#^Mfb9Mqwp)0<z&LyKL?v0<8<Zw>V*?+S?HBMQ
zJ;NLh1TLkTVZ{M75C;R7msZ#zom^P7WKTOBV^`0wnPW#{Taj^zz+flBM;g8n&-0`4
zg!JN`CC6ehD(95HFxQU9t8&-w_u2{V<_353x06Aw0>|>H__NY)*V;E8Y;I>XRymvM
ztjOdpV_hly|C}hDPDc-EJ+BXe>z_wbUWn=7Irh3;RC^Fc7090WGzKB3Acvh@iYLE6
z;?RFt2vp)Lb48u&yIP9aT1X6D>dd*156>4d$;hPUH}0SZQ{f=E8fbXNTl*?-EFGdy
z^_q}gN$<W!`#LCrkAfQ;?3;i@rUHyr*8>vQjaIv%j*WdOaoD%|!Rr@t5Puh~7b2?L
zP0^R~Z@u=tDC*1aw<`*HXrVm5wP3fRJHZJ#yDiw~Uo8ypqnf4Y<FmRW=p9}o?rz`=
zU&k8qlfZJxCQa<USo9{C(tYjOq%h<lw$~E3>BoCK{`$ofd+@c4e|n}ZgqnZ1QSovf
z99S>$ilLkwcZ!EzOB}_0{!-%KhhNJs?}L35zkDsRcJ9O^R0%M%ersc1%a}YtuGH&L
zmwG&S#|A(+g&z$CdpzE9l49<)Ctl0gG~;{d%#*KWI)J+24_i1;-S!OI)exj?XtSsF
z=tb&m+-sRwQUAP+FAWf-K__SccQz#5#l)Bb*X;jhlLdw|k?5}1GAY7ssBP%Krv~yc
zld7%)65upF6O&Gl=^A+Tpgj{W`FyAhO!};6W2BS?=ehVBDS<4d8r4Nan;Ea#R^iJ!
ztAx8}eE$!dT>|JzWG{G8{#p`4+R+_U#=yi8Or2^~nj^uZ_&eD={otlw7}5gGtlkaf
zh}YHe3+#|Bh*N#D)x_BKbLtk8`@2>>QsFq3pf2!i_R2*KngTA>+@<x{iO^w70uods
z_R*FGGJXS131MuRUVEyV<-(Q2(w)jyl<0v9#TrH5bN^-ectHvqJkPWcEkrZ4T5G&Z
z?y&7b9bFOSAy!%8wTzK+0Q&U(1Md_FuASUmx;2TpZ8$*m=rslw>c$jf1o9@3I9yE^
zy5@V0CRE+G#s)PZG&eoN|FoC2rRQRYAqodlV6j;%H0R+@>8U5z#cXl>Vq2`&_!xxP
zH5=k{oAW<xquy}*4FA<O2@wAM4B2U$h2INvs>*CjppNbmk27J%<+h*XPTO<_WZ8Ud
zyXG?wcE2OALgT<)b|@gbJGu+MJJoq9i~!9`r|pVOT2C_V-7&6v;MpD_O=(Y__eL`U
zhHSr3y0R;Y8wUinWBA12@Cj5x2Q?$Sw|ec6TDSz-HvDm?9gd%VUS0A?jDeSweWd=q
z&BK4QqaqnGPN_!dv1mKMraK;Jdr+R85Ze9Cp6B_bNb@4pI3=irOSao-O$7H8kB>Q{
z_enxK8)K}bpu{=B<qIXP&H2}K(Q{wg<d5^&zkl{;yAaf8+rZ~93LccEu%#Z|h-8-p
zGv}2_?Q#rPHGi&M(Rb9_VxOxWTc8DX_iJ*C;a;7%f6D)o#OW`8_UWMSC#O0`!BvrO
zu3JE1cVER=<b1Yk8Vftf*RN&w!)~$_;}bpnv3(Q2p%NWR_Ax)&t_vdKm{(k|8=71S
zrTSJ-`}^#>SfPb($hxVXy*Lcu9b?~XjmlX0A2eGo`5dvgN(AL-z#^@tbD6}CBTv}2
z?!*&E7#Cl=8$*)DkKpWBXSg7&$$c{JX*<F~!=wGa_5yj#JCIgwTjEikgn}#bw!}wU
z+$#Fuwv0q5M?6ZB#-pM!C+IJy+Gst%A0OJ5@hI0_yW7pf>Yuh!OB?+k*_Lr+ch9#+
z1420q&|zZ)!Oi$hoL`S^%VY$<(4o!D+t_W1)v#dp__ibgpe*+Y=qJ?H&wv+^y6nlA
z3|6*}?Wt`U`zHJ<37hGxeOfq%T<_=axNVur9pQuL@xqy}zQ-m623?K#jF$%RLg16*
z=arp(HYGYxO}6M}nW-AXBfn61p;ex?Ej#%AhRESU^p#EDmc*oA{fj*lQ%jnazuU8#
zng_m8F>1h`6HJc)MDjrHni1HLPP@luY7Q?Z9x^KkO{g0u;p{+lJQcpn<4;c=U%4%l
zcq+JDk^-kn^R9K}_dLBv%j}6B7U+EwrZB4(mR^#R4G8BBESO&Mcd^>&;TIG46RC??
zvY9HeUpEAb#V$YI7DrE4{G=$3CGoOi(F-*Vwp5cV#|F}2%e1y0O_J1mt~OX6%~Fi7
zD-u^|e-U91>%vAYK)(7#%j=mVfrF|g5G-$Qaha-B^G#LZ+k@OB6CAKYpf`NNjYpkY
z8(;4p+-uN?>o>9)H+PaSzr%W>(?VRc@GBE_lg??C^@}(~q7T;SwKK!d{>=tU3&;L^
zIsR<Jy?m`YwQy}+r>E5e1bPX5GJ>wO4Vo**JUZHqB?_nT?oAr5yN`lM_>rk~vCpII
zw?u0)$FQHa#>*tfJK?vfw};Zf6}OA>xxSUjqP9cyK;}rUps?^DDB8!O?bQ2xbKjb_
z;X|BFyJDC)$2i<}Ys0M~M-1{*kL`&WmsIC_1IhYsYJ}SU=<RCCbc;L?O>wjGWIP9h
zhLe^pvqPd<B~gEe^`ZVOgW?f2!F}95Qj-K6hgE2BR84AfHFiuCA;rJTbjP(A&1)#|
zb0UU+p{CMK1{Jz<1Oc8ZQQ502+G%xJ<+1-~G{ArV#c%AaUXq%(*UkkHf<SY2zBKWF
zgKZKbW*79e%<1&3izJ1lYK~njZD&HN?9&*R5A#>!>{7fKbm{i8NZ#9rjEDNxu4v@X
zUd0vX^H^#~9Ov3m`y%?cG_#GqjCod<{m!n&VOPcFIU4kN&B$v3ja#vQq4xC%BD3uq
z5wR?vjQiJv1Ws@Cc{hXzxeiv^w*eJKe#X9wztg_I87MR|&F`Z>#ic$ZbJq_7?@4vI
ziQE#Qu=aYyDrsAw-|Ya<M~2KF1HkY;9A$ELM=R0*&%PVY&^+2tn#G(!yQiJ)Ss1tk
zp?$tDvKBoVldHb$^{|R^xIFNB;&U7&Zo7T(^~{S4OOo;+;4~H;s))23{d&eW=eCui
zBp!M_@f(i8o`HIjxupDS55Jz>PJgq^9(g_EdiAht@}|6=n9<5X8xze)rh~2UvDY)*
zI5vXh>FL;xRc8{ua0`4qP-X+oBYORbK<(VIDfZ;+8I$pd0fKO!60WcuvN0h0JgqjU
zFx}N>TtI^x`=LBuuu=2cgm~)YNaG}$7$Z%Z^pZ_}J(Dz6;>&7N1T;sEe2$8#dVl2|
zbf?pz=O=Hs+4P_l$a{8`WajC$XEZ}tcyqTsD=ZeUub+E86Mi_o`~B)Q<Mj*SwU6Rt
zX0&bW<DetW+3T5D<#ieH3W9#KgTy&FL64QI0YV(C*jZ&|tj5IOSXBTX;GfNlZlCNc
zo`1&{#0$uZWm?#w8Mhiu6~PP*ww5qLj$DzEx4Iwzk*(Gs1R`ay8&?(!Mw7b&EfG0j
z<}AumZKb+0Tc$Pj8KS2ryP3K-w4yy)5#8rRs0n2~V(_N6Hp{D}C+)NrQP%T7JELmV
z@*scn2_#KjyLxrHZ(KpEU52I{-rp&FNy*38*{$}i?sC)<^s1ox5R)sacO|rk3G}Nq
zf&lfW_1^7dNMnQgp%uxWtLe+J79B+-&$S{=vBTEAet~4e9(3~m|G$YHotH;PC#p=L
zmfWDp5NC{*V`Gdq<@&m}ZIc!^?gw4u_3yFGLErY~e%qqCq|bb-9?KTsV!cf>VPb8~
zEw)`_dF~E^yn<4_Q5tth`)n~@0Xx+>vM1Rt{Xiu&zZmY;59B!Yk9kk*>t%niy+Xu2
z)op)J4f$>rr7>fIU7oAzpg@#E{9K<M3J^_e>~QQ`_ZbXpB%z;2gumN>i1dp61l#wx
zp_A~9qcI1>4n8(JCh-4fpVwco<N6^iBX&Y_UWGO4q|kOhZ#$)zHGc0_--V=w<tsX)
z*3aV%YiHHEFud2!>1iHeBImX7Ip?Y3iT8Jq#C?rh?V@&73;a63qAp5yp-cMyU|7zG
zkJ@En0^vQ!+ZFxbYC(A9=P^27-0t@Mg>Z3g*Ox+jVINY!)p+TtIuRVcioLa{?ss-A
z)`^l{1bS(JTCUBWl~faQlVKytKd#5h<yYAa4Uhu?`&NA`-2nmPwBF-3+iy3usN~CG
z<9#pC3n|g()B7PXAyS{Bm^v`Z@d4$!=$4?{njorw`%zyFVh6GZQPZ3>cVcGa$mf1H
zMnuqX=db^;p8_Ed=y6NjJt5TX)%NVZCg5oa_-t%@#`kdB@C(`l+Y^5n{?N0Q>SfY#
z{KH0V&ve@z+&e~Z&p2~`D(w5v_QctUEJ|0{hqos#PJlZ%nn&UthYM<;f<3B7qGSea
zj2^3-@sM~d9*Y!3^c%t0?MdXA`1@bk<HCr_1fKo$i5LWzFlhH=yp80teq(+rCWFem
zU{6O&k03_>9Uy-@ZoG|;Hia&(A`@bMz7Kj)e8&?Xt>*%mygd_MN(!H&HYGrMI$c7q
zw`Z)@Eg(MnH1$a4X0LrReS3Djb}7SC;BrboQyRy0*vIVIzz5I#PJ1q3#+Q(?5YC9V
zp8C~hYBegkN6Zpw+|QPnv)S<juHZ02-wfr#r;gki9)WF@T0+mrgB@B6!4XC#@pxT{
z!%te$1%XdGpfK!$ZFS&tWjB+n5qvjIq?%5(nlEPnx|#%~PIy-n0@<La4sI5UHH&J@
z`@IEQQX&o#E)Ic9^}3$ixiFAr`T>^L4~JX6J<|)i^}-7raBB1E*os&z_#}jFjE8op
zK!CPjq(W!S)*_VJuqd_$5yG0yx!kU4dhlY`0CI9+d#1#x=i1np^lBj28N^}xY_@Ju
zA_YOZPhHM>1d1i*h6-OOY6XbuZMObEl<#KF3$`ZUwKb6eXHY9w*XllrFKZHb`r|ci
zYfJMj5_x9pv^iOa<YT&Cb8Nu#X@f?sK^U}+dP#JRZ3;+SB~Tb_);KlItG!<O7PT!l
z7Hq2k)pv0|Zqp+-7npWD`Zd_4%N>GMAm<1l{*Z`tWnVhdya3$J7=Y8y{q=TfCW{KB
zpyr~tTb;-dQKS}%JOAww=sF5u6k2Sr=*3A&X!HJ{<=}uvb8XKg-%@-$V&}m)A<q5h
zBs&yMlNa&f7#<>aA%r{YJfbNMaz(O_G%$?YCwVle*PM*n$As&PDB((WjpH#vzXWqp
zJ+b{lskB;F`qK0Aozy5eT@>w90L2LiPu9~K(!I~&4`%}Rii$aQR?~K~&w!oNY7r`C
z=eMW36j4DJ;&(Tsa8g_h0I$3>(LNObzUkq=*d;wLai;>iTsjg`RU+2xxLt|PxUAae
zA}mgS!++&!m9sAdU=d|>dhAO*LtwbdRecj&<<XF^7GFh^<~6vd_(kkmEXYE)c7Gjz
zFP_h2ztP_CngH&=ab44Oeb5b2|H{Yc<NlTcqrcU7)ba`?(04)!cK{CBn}S=OlHhVf
z{~26N*=pa0zGs!)(lhcVR*l>0?^aS*PW%{H!+&0~JJH%>UXs7c-B{bcqUUk>DISOy
zfi(AGR5xv6Exj+OSh14w#@@(SR14Bksy6Vz8;N7rxLXCzWp8908+SgmWuxB6cvffW
zIyw4{*bxM2zmeEC5mj8eAATb{RY7vZt!~I3c_U+Jl}M7*=W@TgN8iYfLwAlVO^nPd
zfeW&qyF5JhMkaZnpYsUyvBJLBg9=<5^|+wddO}2;$N8Y12*fxN8?b(B1_e}jt`U6d
zjU+Aj^q!6ZX<0@>r*V3kKc5ukZzMUVYEj)nn-E>dpJWqDIC@Q{f7)c>m{PjrY|0zi
zNfAXW1{rW_;B)>}&%QBD$RS5g4`42Gu=}17@e6&VcClxpDJkHGKep%I$g~3K;x<DA
z7s-xH=b6zM^$!oT^cDv!7usg)2WsQ`zAx8U&+|>?tWvbadDAVr_>u=i%-bB7^MV@g
zSZ@oWFP;Q`PE#}_8se3M?!88Us5UW_RN%E*0_=vw{uHfFzu=R>F7CAkAxMFe98SH9
z^@FQj*_um62;f?>rF!gjTjer6)^l;R<*{Is$icfJ#;U|DPXq}kvNcBAjM3$_WYt{k
zkXuo(wZtx;S7EKCMIgZ*ML1*TY}lw&R}dAeyF9y2XHY3+HSLa3s_@@%BK3^S2hA3W
zg<_!C-`rvSdRIe~AuMx^X4BK$-erR$AD3sg@~j0!*yAzuvN}>dJMnaV56BB{ay*V;
zx-PI9K7jAudhKz1i)}+`Gb35xp0qKhy11<Rg*mn<I{D7D&C!f~KwP<PDQ#LeBBDZ%
zPQ&}K7T7iok8z<3>#kqnw+AGuptjnMXzltxvX94$^MC&2wC&W4pZDS|yF^(0@JXh#
zTNCBpf$GurXdaaicW*Rb*^G;35oWkwZ=YwD2jVTxY}E9GI~<I0Q@O-L8n@mr?}tZr
zZ&3QA9f{uQUhbZI9|;wBh%=a@+CNc{yal=KSPZd*e|B8R)L<?-5l>uqJ@fm?*aT~O
z^L8q_EX1YKPRE|bwZcd6GwqBJ^UDLDbeo;kN5sM~@1mX4WO9gbc3wn8^R+r}7h(jG
zSQYJ}kg7z_`Dvih!})Y6R){09qQWkZ43Qg1=8r3yEa`I#1FPL}`*XEgTrwPfAsSIP
z1^tPADa=qauCl9w;)*MG)Euy{w8?*7ZUkZFnx^jD&zXgN9W%kbm+Yp2Fg|8HT-OG?
z@;*L8ZoWHdH=+g7K709Fp@wq|_bdBOU6A`&7WcTen=t^6yd+Ljk0g_mlSMz2ybW&&
zM9Q^>s~_EN2M7SJv>&w+i&!fZ?(we2<E!p!C0gJ?KM52?8Iq5A(}3NJ<}f`j=+S&(
z@=Gz>k+`>Ot=QNP#1m`R%U$)s9f@PWS$NxIqjn_Lo>}&6bUf3zxy22;5ADcggH-Iu
ztzQrCNZdaicc&35PTC^@6ImnehDUcKRywDL45#tu@QGG?Oppty{=!&6u4k9Bn^Cjx
zNK(Wverrzz&UkmbVDRLQ#B_<hLBTFCRB`0p(>pTzfvY<BT1$7%ae)&P@$zLhK0yEa
zmo{NX#*0&ak>rdwQH{HsiLjr%Ba<;anJ*i4N+3Y3Px$tZB;cS&W%Ox53q}FcRM&61
z`jUpmztLxc1e0EwV$TZq7PP!rz117NGO8Uyv16dCfBAnl*vuW7j5Jac+hz$wJsh#s
zW()Jho-e9gk42XiDg)P^J{NzhG+LTCQ<<mM@Z{6BAc*c#75t$(`Xa$#Os|Po?nePR
zY6A&3*d%Rrogi1EtnyoPgCJMWscNvr0Shv%?zSbd6u)`NmWo!m@pyKHW$|j(1Y53o
z!uvjX-B!diI$n2mM#HZ6i2rEF^FcUtVI9^Ybjn=EjwIfyf!r7bYiL(@whOc=M0ZXO
zbnE(FxEEQsfK}jF=ixp*F*hWFWGyagKc#}LNmbDw;JC>VlO2B-+FF`zutZZ5RNy5Z
z8-r>wwl@Ckqt58MXzn$bZM{HaW+e{EFKHXJ-#H-N8jT9jh5U`tuX;|+T-y}Ga4^6_
zH^)Q78*Yh#dQqTk6$$3zaYaQhP5b%b)3#j{<FG_zupN3`jr;FVP=!wpC0WW{y))jo
z4GwzN<Xu{SREmwZTk8+6+apN7n)(~t8yImZwEdb|I*wHS>wtjGb-&$%dPKG%>Pz5i
znj7t~Mw(l0N8(inuF#w=A8AUt6x(S>W9S#>SK2YXOV-!pc0Ay8W5q4n35|%_F>fbh
zm)6(Vsen+nu+~niO?emYGiL;FE;dTgJR1~qrvmpnLkD(F3*a22%gzTdVy5guET-RN
z?P5#<mri%S`7}C_bFrIvy-vHN?b$^+8V{krEDC(`wq4O78=;`^|IdYJuWM<(5TdB1
zS|RV3I(Cu+M7mzuRS~%xb?{d^vWr?<Cl(|>q&>~RH4XM6suRw$mG-sJL%ih>!c#n^
z=Ei%wE>sqDp$ferRJbd8o~&>6o`pwiuE*y+AI;5}DAsNc&hO&^>mB6#L6m*I+0#wl
ziZS~-QH|dgkQ`;|>_<IT<a%Mj(Q=}D``#7q1U9wh(YXF31d+kqCG6gg3t749?Fj$4
z>%$M-LUGdW2c%EV@M+nb+5MR<%JTzy=32#)Js8gw1YYI+d>!>>VxSaow$X21aJwsI
zdT>dLpL6ZxNqIBlP<0-JORUzL*+p=khl4%(W~OmpiJ&p&&5V=q!6IY(v0;y?#~?;2
z8>>e@!R>JkKr|vJjXj|sBKM6w8JIDHoIMp2sKZ#mSwt!9H+AnP?^bdnVZC80<1`96
zV!Ca-fWj%AM@<NzR2C@M#F%a)iGC)lTao*vO;NYx<+jqMzM1iKekcWOPSbBCff>A}
zznOS974e|@GjC>E3}ik+F$vHX)u`NL&&3n|Ld~eSXXsrO%Q7=?Yb6!btkMd(yT@$d
z*@56-<?)ym@~+gfQ`OiV(nh9%4U6vc-pqupm0gr7qbP<g2#nh}q6=0X0FowzbHr+5
zXisltwE|vDv#S$ugnm!%-=GOJRMgtyfD1=PIlDxCJGrl}v86FP_w-q&Uf9lUxez2t
zD(UTEX)6K;#%Q+2((3({D`mcfI5D?eoJDGC93~xZt9~IeYWPF^zS~R7-P~k___J<K
zrFClj=aG=ROM`i0wI0#BrGPhhQOoUK07!!H%6_5C7W~s^{}VR0M)O|rM;nZBx|#^b
zE|Hcyyl+=qtFE~wL<Sg=t<!*%ufuO#Gw3FM4KF|2Ad=%-<<i_n0a?Wrw6n<lep87o
zxNu{}7Tqhl(5Kjz7$GHIThwA(N8oie_X2O5en=iX6v`G3ZUl9~Ylr4Hyx)dmeiilg
zwo}L_d*ZGjWP|5P+bt63*l>HIkGq)d)o>u|MIH7FNQECV2LcG<!GGGp=+=mv`JvJx
zxnbzAdT{&~?1&(eRSmZl`$!W=-lgpEqk&C@3zh{trs?#y+wph^W$TGI(_LZeOgkAp
z&>4E###4bv+wlLg)A47nqt(uo$WHyEUOO8IB2#fW*5n)TFg!o9i3+(!()}SjUQi37
zWbC4rl&>&O`cI>QOI{80?2<sZ%*JMiZ(P>2d&wclF?U5!P-XX|>(J+-6gj>ySjVHE
zZyvkyr8+dg^-*Yc+trwbI||uXB4e(A3_zjNYcb(mr@Og*9T;NCb@k>qnot>mPS-Uq
z_>!Mgk7Oq+*tgL(c@>_t?_!w6BMH`Ss(BNBjrP6J^c+t3yI{GFbsG2I?3Sp|(#-b9
zDLmGv42kteO#qEtw~Op|Vr1^p|ITN7HzwQWDGPs!ZXPCeuk_WrJ3Y&9_ccVTM+3rn
zhY}y0UqiG_gFP^m_%Qh~!~XW*P{sk``^4)^zYL#BmR1`*lyO>8n7cddLqmy&xJPY=
zJv@|7><W+jPgY7Yqr{M@2m&clVo1(e0ss;E$7q0D3so4M4-X|~KXv?g8>_}Rr@>Yq
z*VqgC5YhAYL<|QssFWN28^_?2F&sZFdZEFd8p<RygtFGSJ|i_gEx`B#aW5P<lt~L$
z4nW=UF(uN~BC<^w%7g{nFUU}rE;JL>p^36tHd%O(gTIs`C8ja%x3g?&fEzx*{m7;X
zzwUvRHa(s=iamGTGx6j(kx|b*tJ$*aDQRHO1sp8VIEc&`!C?5qg3VN8z9;t<pQZN{
z&T6*Vnz1_-c38R4C~{I)j=bdqY_cjr!?DGM=y~yj*KkBXjQ(K3P?E!jKjubN9k7N^
z;q_6lns{=xA1>FmT3-(qbb8hWVaQ}?4FQCkE)HlVBH;-P1C9sEeBk}Dv_uv3lAQU=
zG%p0Gf4Akr6e_raW`%x0a`c%u3WGW<fsR35E1K?Ao_q$DJe2e%zxm((S*zetzRQ!{
zw-05q6*VYG@P!0s1tFE|MQ-Vgd3p8*>(;CG=G7>;1+Eu!GvgNZ19~?PMC^~bvw;`>
z&emvSRp(YSA$$!6131oFWEYRN?t{{4Yc(zG2`P=lXMeVJG5Zw99ra(<j|}Ku{1~nE
z1Lg00i*Ae&FfhQMwNJV{eObK0;8?INnpm~_IBeD9YJc1|vZGRpknLK9SeCr)2w3M|
z1@)nr2mW82bUR~6u1hWf?-B*dRw3)Uf68uAh)DiD!k+9RFjd>Dk+@o5R%o^T0nqcm
z@mU=ZK=-6*bF;`nb?Im;xN_VM1#+awglB7nCcH(FQ;&>5@Y(>rC?Ca7_%yreH);Kj
zigJ`JvSXr08CvL0J04#_8+MBmqFl8*R-M%TiV67z=~T2P|Dm1M;0zT>V$P@sl081e
zu9`io(HGztbS^07)yI15{7{k<Dy!{+8s&%oY8T_FKH`Wz)o!XC38%RfWZ@)*i3dzL
zIu5~CVh&j1(Na^mOMnsN><c{=i2M68nhsDqkLt^IRVQ@PL;FS7z1Y7h0i@=X`v6@F
z0+1NbzSeTAYQsYr)wbw|ZmSpH%j_&KcURFHL3r1jas?T%Z-q>CZqU9HFx*5u-Pp|-
zsB+E{`(7>Q6O{5pO#KzAec3H7=IW#>x*eTXaVP&Vs8Z&e*X~3gipi|ByBeuHH{h|7
zKb5A$r*tnm4Il7I9ruNQV_)xp*<0CNxwq)yNe{f0O)phEcu<dsD{E}jTiM~nWQl()
zBBhRgD=}ye6wc#^;sIvMwt6@^W8?SmO?%|6jMKT8RZ0{8=v#5n?_+C@c`K6-%A!<&
zJ@!_TCs2?MA#09(E0YZ{a~0Slj|=&5^n4<kgUYHd5}3UeoF4PzQ!&cF^3R?Y6wjx}
z0W~h#b>gWze&iiiNIujFnlO7Gkpe-=CI&E6og@-{E4xCzIDEjBwN&Z~B#4LibR$#M
z05yf%O-|S}VaY|trpHs)Dmvg~_LgE6|2GW>I*w=65o@bGCtT6=bl?X#Bk*z?tIZT;
z2wf+G<E-e4#gas;HhToAWaURDDUSxVT!Yb`RmL*ne}-68rPjoHxQvb_Tj0@0ai%Va
z-h5(Q=d4-_-9=C~8dLwoY63fC*__pCBJ~5ram6;lP+8*)wIN_*W^$i|leRdPrv<5E
zi8hGuAP?4GD%?EZManU~EP&UuVfikW2Q&`dNQSf`&>(qypEbUfPO9HbvApKv7lbxz
ziTR^&N8)JJ7&wEiOew9}<Gm-U%vm8odhRmo6d>OV6aoq4S2i_SPc$RNc!w3^?}DFg
z{iW`t^tCn7-5viZnb2c{q9l~(#+i&~FKY}p_UCME(9Z=gx9Y45+*Y@p|MQ@&kMX;3
zklPU5!hDAd+r}98{M(3Fo3zb4n^z<1GMa4;I=Uj9r!4`fth~au3aN!Kr&c~8-fs(9
z{{AJ~9^;p}xM4fA(l2t94(T*`-Wlgk5!R#9Qp&Sk`ib6XyW!Bgwa;AvkH_|&7{G5F
zdD~lB350@9+aK^bjJbPZa#qjeMs(5+>IW2O@KQSz9bnxlBIvM2Z{r@w#?IRjp*+v^
z%pb*klGKXRSiFqmY>gcgc*^eiavl$OKhxUUiMKA4{Kk4B@-lrKCrf*vlOheL)S|63
zYTLWmb~@<ak2?5_7OT9ufA~a_8qaEI3<Ukm{hSD2Xy&3FPdFm@IeJ0wdeQ4%6gYP(
zi&8Y7#%k7bzYhA6;sD9*ay)RpQeXCpK1OyxirD>JkSGuXLES_D3n4cAA)<ZJz7&o9
z!rt75z;jjj;L>Yfy_MamlK7m=y&N&uq7h$(8<4*)5eyz?-)IODllk1QmtOXxUcVvO
zjX2wWtEU3O=XdeMMslC;oAG^cv5j&d-)rS@Zoqp?qj-RU2V6({eCBm~TkSD1;S;W*
znkjL9Cx}%lkmK_0-9R!|z#-$OpsUw?oMZPypyWa3iFED<Xiv1}XPWKI_+)=GNiL8C
z9?&m-9{1sq`ryvQIq_f4v3o}C%s6K~zBin}qjx5@OUZ6{b~OED{rNwmnR|F=#?eyH
zfkzKL5)D{(r|6@C6>)7gMuT&CMUh36*Un6cprm?}jTOiqg1|g0pgq1b6FYkPaB8zB
zqDLL9%**~g8R!*VHS?6lTjY_ZJ@&NF!CS<}Mbkd~IBmSfo$J@k2|F`^BUzdLktDx~
zBfZ`9yfd+Y+5wvqzmrS570-;R8k7?jXH}aP4JW^3(*<CVZSjmI(u=Z->euS%YG`nJ
zPB?Y>VLd~S?!;;{)!c){xa!RkF;{!0Bu=2&`eiYR3j9zl-<f?O9OiBY@tDz*Rw-Cq
z<lO1AYG;zn7E;`^VDmIETZDvVT9Z8a$5wf!q-sqC6#$sm1Z7iI_F5qUn@I}#I>CD$
z8-O+JOv2W0$4|7yB6}R`$!&0nn&-RFdo9&tzYFxsc4ntymi7-(-<E4a>>G+Z*a~5>
z7{6G&o2)T#M2FC2c`b}*Olr}SC!lH7+=_i9p|tj$nIK!$gPQ^OKBB$?)+g&Z6gvYy
zGy<G9#9x<4;lUX_LDgWAMWM)byN!dXKj^Sx(iB@0TPELbgTkejkM-pMz~}VKDYjMs
zaAI9WsP;N7OeE0dPO*Ljd0Y!9m#sc52A(~(G0?+E*=n0aS!A4oZPsIxI~Huwqvxx%
zt%9_`T@89|TcF?F-DcZ`^lI`0?uaLp8ScC{6ku2+Wcyy*xl{J12-_vIJEtf$x-?uZ
zcW>LRxg}xP9d!1@fE9}t+Fl`p(UWKZ+ppfF7Ae{RA%uC-4r&FFdpu7xpUR=21FD(<
zI~=WtPZ!+3=ZHZ0Az~kiq8uocqdqFouCYC)r@5psIj*PbBDK0=QM+<>QrjH)2hPU1
z+GVF=ZOOmQS#Vl|ax?C*Gci@t?%3JVe(D`WB`!E_-skG(NBntx;vTH*GD5mtUJ!{=
z4j%O5nh2$WV7H%YNOJNN?UD#i2$&xy*sEh`{4Oc5S)$>Nb=R~1>~o=m_k3UJe*8tC
zPT~gpQXoAUl-pnJsy<7u9>|Wkf(wmWPh)v49+4_Jdsf-kYEKXbhho9L(KHrwzrcem
zXzog8?h`lk7BPfP`mLas4<H$Q7c;^ci>xUkP(xZ2`@RIpD-mtAAJlU($~C(sM5;Uw
z8w20osrwgRXUxy<AW<UhHTzN9`~}a<!0vBu%X{e4ofxF(Rc~7DuF&m9-GeuozW4Sv
zyC?GT36OIh9ol^zerzEigi6BRPKP&=DLd(bw=;%X34gZ--%i|n_%E>IQEw+k&qDZR
zY4qC}_eNNVIG%^z&bVOm*oty(55Jx9UmOuRJ0E#Fi4<_iWRC;VqcO&S*F7Ecb|yYR
z5zkEg*xT6$pJX)OOo8>-w-X!S?qQF|pYApFgwV$ki^|!7a&i`@B&Sd=dD!3)F6;n|
z{&?Km*=g-lK!8B%8z1;3uYw6`zdEV-CdL4qu#_dV$$>q&3YvS;4RMOV`2`N&K$dW7
zbV04cPOxb~eB2P*du{sL8FNCt<q}L1*`E>G2=dKIFnM-_On+i^kJ@uVRq!AsO_Ez@
z1lEK>64xDnx8TBSvor&8M`P%iEl_1txCg&-feKR|mAs;hyBK&G8mkgQFOo`XFt?g~
zX|V(zGQHsKB<Qpe+#QPn#z~J@6H^7p%YU<4jpDg$dij#;v~+{H)`B(Yu{YO59GK81
zXEK`UC3<d#Z;~S|fU~JN&X)y=E27@h@<8Y}f3y{)J+p?4D$yTt#=pCxmJfUoG&7S^
z%i9;S%6C62gGBTHuvXCw|09=^+Vw-3AAPu;1qPqI(bvyM)u~-FA6Jy_(ok+@B&j`v
z_o#cy=zuOgCXcX?c0hm3U~a9g(I<u?e8`=?;eCTz4?d4{GqINi5&6_P+}hXzb#A#^
z7w8b&HhhSRUFz*pf4W|6h{l`<eYR1A#;XX!;U+CyYjYv_L2iy;N;eJ{T(-nwG$pgs
z)|iXePURc0ZNjW7RbsR4`hkdAU&<X?dF+t|8`67^2*%%GXHcWC(sl{;Dg;YHgQyIm
z)~o$|@3lR`zZ*Y%uJN%4a$Z<F0R3^29f%*x2n{|c3ga%|yTg5c4hf}&F3cR(dY0qe
zb|k2Ujh$P&8sgUkx7XlBkA}8mCBi1fc@Qb*wL0uXKqMiNTa8Y}QzA#Nof43WMT>E1
zJFPLU{DTwnjAradCQS0I-ha07vn_T`kJZh{Hs@oifB3_{+J(S|^xIqPVhlo^A{X<I
z+NVE<>OR<)1jcusUDo^NHGZ_&6|F^C5euq&_kJ!?l3g5bgF>Hcy5OMn#{MlhrSw==
zb!4LK@N>hNd?4qu^ZS)u3(_HrqZIpEi&2f>{*A!D;3=`Li+0^e0OUlADYIEHfd`Z>
zO4Rh&cd-{<c;0RXJYRbxg74qH@EC4de5fi!<dR6*6+Z}7k}seTbKlQf8W2Y^Sn_S*
zll<_16tb&usJNpiw;<n*r-i;F2=t0|n)K^gcCWN+$irgy10L2o*F2f+3bpn6v}h0P
zN}Sv^pH$M`m9f&&t!>n<jI|Tk)M=x4Wej{}b1MpY*6^WS!BbZ?+QYjN<Dc}(OJrQx
zmGSi#+dA61hd=aC)kkAwPIHgv86((98TnWMa)r>?z=?KqE9~+3omc{UBKo7(bEf!Y
zw0fb_uk23=a}M+2zvk@eU5TYMG}t)3a#av0j}M5g!~1N4CX{Z2|H&?$7+_sDfo)rT
z=SA|1DKW>s0vluMu1wQGf;yBm)X9#%E_;zCT(kZCAEwwdyDpSeYuyC8r^F}R1D+N9
z;oWRaF66s1(Z@p^@Bt4ZRm8rwnWeRJf$=q)B?=-rkoruIe8fXmF4*jiRjAo3W8O_G
z+pJ2%G!fB3{<CbDPIS9y%wv{tk|wKGQ!<e=$C`kWTis!`@hS_MwYr#o7wS#QXU0QC
z!|!V6*y3oJ4s_pxOGZeC%tWC8Te>TgqLR=C1X6};buhz4N!{g|<3e(u*a{(sHU>eW
zQO{*5pgu1Mt`5S};H0%^XsmFFd9;d9_;PvpU%LjWcY(GbAfEp_*`YgwKoq-xHw$3t
zqp%(Ue4gl9zN=!;V!(Zkt$$ahAYoQ59=wyZ_G<#<@P0RQYZ@)+|GMnupm`C_u~x`<
z`Gj8gMqDR!TigeFy*jU^J_V}v_!8Fs#x`mXu0oMT$XSxHHff)(<{R4_PtM0##4Y-a
zxe8J6D7dqxhg**qh1w>ZFi^PWn}{wEWlkyti!<7;Oe*IfL?TL#plFC}sI)DSiFU@|
z_^+Y3w_O5TOJp4-yqj&e=H`wIq*xJttpj{l^g?~JSB;WI+xCk-ox=xu>_9wHawnCp
zJQ(Du{hb}k7PUKPht;P8qqZH<lwE}9O7fA$=xl~6b~G@D6}vwCm`GY*?6cz~xSlZQ
zgciRdHh#F&oYYj~%ECgN(vUB7CJF1bfOa#<GeQuT&L%q>L>%{rNp>!V!-k2uTR>cO
zg#*9^txK*zHZ_z_7o+*WYO;KN8c<%DG|4V$f*$tLYnKHnibU8IAwUKctloN$>5GE7
z_C>(OaE2Rs{Eeh=#;$6r#IX^t08(G+r%o!>^|<rcwZNncr<Si{pt@!<Lhx;TqZ#zF
ziQU!wy1F;D4(}(+SfAYxHl_N{-^Le6&?>loC)9CZ$lFaldOV1IFK`^lXuX*H4?zrs
zrBpfCZV89(+(1(P-HsVvd5;<X7z^cP3yA-{6K_1)-|lJ&YbmPqQ{YtZZh&?#9>PX&
zU+8ljdv;ZOM~wM2P*2zc?<BT8{2>-Qd+?o%OCvsa+bBKOxanZ@I~hwMPc&@eA;IEq
z=hwT3qkU2m@!371_FQ?<mp>|Se)+I5?_@H-{B&}|SRQ*P`?S;UjW}?RL>sHN>54@9
zN_$+RBda;}pU@8-TyHTBJgH8(d^;=7x$x9G8FRoF&Tq4P`cDfj&v#_w-U&^^KS1)d
zjgO~vx5Z8#PsIOu)WpP?AClow#I^M{SqN0%HOu9F$~#GPSoTMos&V<~lNN2jrUfvx
zKPPc@ogTnScA3B1Gl3Oe$mtk-HXbY6^3|>)elCWpEEN>Pkz9D@yBclgJ4u!*XFJS_
zj**524>q-HY#X$4O})`e=v2mDV6ycuSXIElMr-p#NQ_x%Z&=F(?_BT*f_~43Umbt9
zd0~>8cSb}Rzn|u<R=Z$f8?m8Q7Z?!F;{iquB7#TVFAhkUBJ#E*#_z+PwNyy=f3nY(
zMI$d8ZOb)W!)nit&|@ovtGm*2!#!h-0cjG^7x|I52YnN=YApi623hSI@>Wr0A>7N_
z)rT@pe`5cgvO*wKh4T$abO!taY_2;fh;kisb5H!<!l_=24#;aBIodCX<QZ{MX28~H
zOb<n~!RYGlr7x>-c>MDrt_^7OJxeuQTdOv4wylqiQPx8yqiVnMZiqd&sKPc1(3Ks&
zx1r3Y0F>)Pano&^MF}qLDT&gS(#m?ix2^H_XS?9E+q9;&J%a<?wmsmmah9S7-7zu{
z_gBBm*pT*rJ(hMVB_Xozj825^`?YG9TKHu;qR(~<kgsD(IkZRL*n+OsPFzX$>c}AN
zPt=6x5=l_V{t~|LaG^X9f0JH-dYgjaRa;n%LqWf;l<EJlh6J3KRy}1$^iz_g{Z9Ci
zenL^!MrbgR*`?Y1<FnUL(i|&2MO~lcA{RUb+5AL_tnM^>GFF<x9(F4B0LM})opD-Q
zf*sOjXS6!uMBHU(^$VA-B!QiKC%YTGFu%cheNqITk1a6i7j{8grL)~G#sk?F*>Io!
zY({>U<gC0T@O)S95NDS~I=+JwEbA5haAhA$^mqF_=8X-;8O9fyH{u+Fe5q&7UOt?w
z?_>gOM(h;A_La5*cG8mn^R++*;rnEg`u=HO2YDL8EMnhi0_Dw}QNZzf442F2NkjpA
zx)IO5sEqTmefv&!wOrKbnyVxneHZIN54#zY@jw{7lfM@^Ts!Lo`k}PQUIyl*-4ZQ`
z2=jD9w}sJ4kB#zzyFZqCpyuPqzoP|49;cF5ysRU23AFktRtM`Ur|Lby_6+5)p8N5(
zrELwn(ROEi83PP{iVy70=H;aF?~k~^2X`lyux4P;M(xhHe5G6BNAFHYxrY_lL-9Aa
z%7Q(-J3Go55_5aEJ+eEy;&UE6s!=<Tf7@+LbWM>tkL}K+4CEp|qsHn7A~y+pv&VN|
z7!%nV@j(eExukF=iCPT8S%MAa$#@w4u(P?tp4y!_GjH6{=;;xVO~Z%So#O)E8r+>%
zhVpv6K%Lub6V%?-704eGH34Fqkp3pe^9yW+<nla4Wby*qHdW+6+E04cX+hF11i|Ss
z`PF_cct&vZB+B4wzB`jS*n}L(?tlGUbbV?3GMf>z@(d<$z?qt34?3Dz!nr!<uAVkq
zI4y9COnFSOnvavBB0|v9CAdUXRRZnFu-Dl<0lV@>+OZ3SD4z`bsCuLU(kIE2YD#+s
z!L?R+&E-l_r=Ddv<&jZELpQglJY=?5Eqa?h275{LM<47Om!<L6@B7YZY}xKi6y!2P
zF?Te>%VTOiBn*J52mJKql6j32y)hun=V<bxhL#tRF+t$K?Dy3c;f&;At-`Ts_~SgG
zkJdi&ow!-)jK>dHL3?J|c<WT36sC={eRsUBNohWNv>M%>=(s3&^ZfuNw-TuL3q&T$
ztr_V>*?JokmaZarSuZR6acvNWyCSX&@D7YN>qb^-A6BZ@Y`waad%Cm@u@y<YL(Wst
z6}=u;dUoEXSkoE5{<Uo$f%HLgnA{>%sxi@T73S{w<=AYC4k>*?GSqGt1pKj_3+@mJ
zQaB+z0tstKBrQjQUfKxQ-q{Gd)V>kBW^+oIX1k;PAW^IxwnwwW@=NrX_7#^t^lJO#
zwP%%Z?cITxJgO*6=bFv)9N&k8YHunqZHJ>NdeT-qqTQ2SMYxy?wRwEBNWJdPbW{y{
zhCg17;S4eFSfCK88c2m21RlWfdO{Q=ux9uW9Py<3*EhAUZpN9C{N=mTEdlT7)A3h*
zm7NLX@FK#B9^}BLjN`?*m{cwCGF*qx$I#UDNjkp^8qcrVPTz~WGl2jvJN!Q)v%;sM
z=^QSmc1cJzp{l`i9}(rUns|t~T~QNKdGW<Ltv?TJm;K7V2yAP8LB5P%7xviI_!atc
zF&Fjwx;<pcuEjbMlI;TGQTtl!y2xb*`$prxnB#oCuGhG#LbDq|6OI!Q`BpC}a*Qtf
zyFg*mAMK|0e^L^8>G|(Pqs~4|rSQBTv`;9tZMOn49D!kO3+;LCFiWlN062dnK;7=B
zd-w3ic6YZ#nq%-8LP*@YZ$Ame&c3exF1x2r72MzNYZ&)x#xd94%{Vli>Xbe3Zrss7
z&f`e&;JX=HPPT6>x1-+8_<tFdYkF<;yIJy*GOqkEGf(k=!2MA4bW(UO4gf#=ZW0P6
zt(anuyqj3I$0j8YTYFUF5Kxi=U&rVd(y&pE5f8X`vvZrH`R@p4XKd`d8IvwU5A?Vu
zf*Tm+M`FBMHo%iH9?X0WRY;)qQ!z>B`(!tdhvax`<1`9Y6}bJvs>Z7^$-XD(Jy-;<
zoA_?VbGhO7#n~`fWNPT@80fMo?<Tnibujr-^)98VvuOh9Zo!GeOpoU9?|ysc-Auyr
zWC|RO{;fUxZuYGWpK#q?PqMK;C&)GR)%f1eh-Z$J@ZOynGx1Cu<Wh?<7A~rsW3%J$
zS6+E#nU#xJ7_8j|urjC&&vL1|N>rYBecc|L7gVn2yjq}n^d>oruf1B0SGJK)xkjL!
zuOb%JY5|+vYdv65k<l97&9?T>l)k;TIN(*V8~tjuB+y2)(qT&jQ~{$o(&Mt&KC4G^
zO)eL9SGFT?xTSAJki#8xNuCz45Vm0OB6;akcEfp#kncsk1Iup}wJ}xn_E~#?!#c~%
z3mVStDtXe^SZ7SnMe6S83Ks4NbH;kqnko};b5^*h6^59e@E~2f`lDlSRvmGTK9iE%
zZ-XM~Jm~mxAU$VI%Am~owN@*#r~~nRUCbJtJ}J~P3i^h!4e?>4wiwDvFLA;Q*+z}n
z-5f<vH^saNw6x8FPE_Vm+fv$Zh=^UbRdY;NB42=Q`f^f?!0;jNY1>6+60Lbi&JImz
zA`<M7y5v^!H6eTxel{}d>tMJmXymz!5km+*dpGgCMez5SEy|lN+pA7|9jk1=;Np9N
zRF@dMwn3FUGzKR+7_*)9$CvC-U<L{+kqQrMex6l`kjp+h67#PpbYYMGNZW;|BPMz@
zUUCNEJ$Fn)Rkt;hIuJ*1N|Fd-Iw}396T%vH(qbpo;`x*ogNx*;nC;>+JFT|x*+x4P
zlcy%gKkaOc%~``fJ|~K=>_TF0w)4X5zt6G@(Jd9)zNocF@W!l$?me3N{^_+#dY#fB
z^MPI#ey+3bTE#JbC3>`bY>#~&z`1&_vM;m(DVJDc`(MUnCa<uoK|Q{NViz^a?W+JY
zanhto|7zD_O$dgxuQlmzuZ#PQ=EiYNwGg|m-Lx9r<qfrNbuIn3ngC&l%nzk|G+;&3
zh5e?$xowl;u-|J@J=1%^euxR-0gHuPlc(}8MF^0gZpTma7tKkrw!X=J6bhbN(C!Ez
zS7fXVcZDyCGyAEu`KBzhd+LX=pv&%;D4B=d*`CB~sP4Vb9@vvuBm$-1Tprw$cs&N?
zWH}iXk8DB8&9~U-Xp5(%OJxu3$&TP0$(AS~?HM=GyQ_~Egt14W4<K5a`|MG@obNR$
z#_ZA6oJy%2YL7+hI-KqSHFi&u2$~8V_V}Jm&=^EgW&hg~d(v0C;#c<Mo=oIGhepDi
zr}iWXg(0gJ+S3}boGoVKM$q9n8owvw%;-=&Xo`#)6M{IN^T{Tvv)}D(vL5SQ2{UC+
z#(48&zVZtVAIsD|nH0jo9U8Z30TFM<oK4q=Wt|uXh#Me%T7dO*nR$-fXEjw%`|IOA
zr-AU??CkZV19_VvWa`}8cBUSQ?L^ZaW8(NqsS?o#nHaIXJSbP*W|cwU;nR&)6&(tm
zT6do2ke1pPW<k&t_YT&jTFZ82SGvvD1XcY?hzwgBs1VMKtMq`?#qh4vO5bcl4EhJ$
z2Nnl4%1Bu5#qZ{$eX%8bGJ!FSd)Z!G8er-@TffKWE)$vf<_A1Ze7Q)#_ta@Cv<)Wy
zi#2N9pU*W~UQdG@EG?Q&g=YdH`b>aulqy)e=vRZgd5;zJ>`S`JI%A+BU%+bX7BIho
z;PBfMz#2(><cnQYXEYJ)m;RVWTG)u%TNA&Oc)yn}7!*vr)!o<k<ronQJ8GD<v2wMf
zTCjET0;M$sSg&piQ%+vnpm}k7?Xr!r5TtsyO(IAu`moJvmr@^X5z3xiqt~`ZziRqz
zi?LHx)9ulWT%<jDpZE^#BF|<`s*?d55?};O4C8@q6^0}1;u<Ag6l}Lp^2}j-)YCIv
z+uoQORmMu@{{4I8MC{(Qr5v!X!9~nCP(qkwI|rj9W@0xTwXU*5`nKpqb``E8Lv~nW
zj~h41j>On;^B%B|;`8b5u%kliMdFn0m<~keB}p=LtKD&J)aTpyKsdTi1U@}TyeH$&
zvf5fZ71ZF^n)ypR9T>R#7!3D}2C5`Kh@B0pbG-K3x%eB_hDV#w#u#K5^i)IC=EZ2o
z+H~2cF%qju;;Bnorjk=F_w7KciS2wP{v6=i{kayXi|?z~zEH<*DkhLu4V}7n9PH7x
zcvsbiGVFZg{Pa~o%&lp&Ynm2OjU+4&6x&EFWZ&pjk<3GOU1K-6Kf2wB*IZP(Fq_!7
z8iqrsxs#k8_Fc5_`~vQpb5ku~+jv2N$oD#mU8RJ_0$=A3YKCsRJ0<$MrJ-wkJPXI|
zSpLc5?8hK5oSxoxN8pevotwZ*?#5h)4^dG}kZU>ADOXQ+Q&Fo4of?wteeHnBQ_bGX
zcw;wt?YU+>@LqNg@9*OCdhor(92b#VW|ZEPc`U`~_c9j5k>nR4tmE$`mPow$>7qR>
z$fVNg<BtBwdl>^oRmd_v`rZY%r?z-Cr!fJWy#|5$?6G*P?`v%LBX_KZVj$EX%<AzH
zUPPup@m^xuT!Z-<T&TAvM=(W6WKXF(zA$n6JuO@sutN;kIQ8JjXtVL}WjA_CJlceK
z3YT)5_}&Fq<EIbLBboo{S0~T#XOjb``WMP=O3V${o19G*>{`OCZJLH&(B``G>7vt>
z9VuVvGw}n*I6E03#GVxZBD5$<5>&>Ik_*F(cp!fPdeWI{NNHn2!{5s!heR3F1Xy5Z
zzn27>3BR@S5wf^NvyGeqmG5Q3R?7Q<LOk`Uiar(a>>vfkG2BOO!ASE|9wb>_PFl6l
znAeBSxkitb&Ft*jAWcKv3v;YabI+snW*;=fT5>w!PV6^UTddx5U6;EgXk12$o~0!^
zP#z9<w-Hp%B;4I)^zXJj=J@mQixux>oZl}<e2#tLxYnq3!6kSVL1)?nsMxF}-lkkt
zn6j)j-li0J@a{kowvG1{x9*rC1$~Gc4}^a`>m}>fR93Uktw#fp48n@<WkO;~1C**1
z^amWO@R0p9W(#E&R@z`Z;KPO`;$?wFP*`Pa;~h}~FpG71m$FE%*CRpXZ`+1=bdiGG
z{kBoWuk)n$?tHE-vJ$_<Bs*``lDVD^Rri*dhF2wkr)>=wo<PERAAEdUyytuQ6Ro#v
z(p7Bb9V5U_yM5J$v<QT&^Xcx?bD2y3woA`QtoQ@&j-5H{H@2s=Kwg`{wX1v8vZ2Sl
z@%C%rMVuK2^pq!x<)CQK!f@+oK^!|2{kbFOinzkr;Sp5blyuaN1o2!}@-zJ-!C+fa
zfY1{$9o3qZ&#_}M<eYFwJ1%fl9vVdWS-j>1^YFTp`gHwBq+F-F3Z4p-++PNH_$@oF
zotzqR)pyAk_qG_e{gZvef3>rkX*EUO&S^nYEnXMt&#M8kqIMy+>mWH9E^4!&$6RHf
zMmvh3dp@~KLHJ+%$}X#!Ysu}3kSb+1ELDX2Ji1rl<wuy@7qLxzm!}fjDA8lDV`G4y
zi#TQ2kg+R&rLVh@BZP0v>o4`#HMMK;4UD^+7M(y>xaBv(t<qIX*JE<U+#0(P_~)qy
zs`<~O$#?Pl+#0(nSV^F=?={$!*OTe!hv+yTL4ZijTcRulskuY5yKigselIQfqjtz@
z<Re#H^x7Q_GQXx~PMzHiC`nE)MS1<Car46;wj<2l3rbb--tLPq?1`fsdAPV>_q4s)
zVZv36^l?3?%pcgB@wDN6@b$eJo6k}E$42eVcr)S$lFw*8lP4W9;Gw;ln9%EW1F7M?
zH*x>uW#H~vti~higSRV1upix<U8YHoyArjFjS&h9S)9iL1Sc9y+kFGkq-&U-W^xT<
zGv6C`bB?kn)I3Gqa1FL6<1LZz_EhwAn<aDso2U0C=>k6zs*S0ud&Cizh#$M=PY!RY
zh3c`(^^p@bl^hjtnrw0a<AT#{Q(_X>=LgXDPF2h2N!UQmv1rNu#6rF|bL}h1=_#5U
zzlmq|W{ep@($k{(0rITKh2J8s(p<=eRE;O&ov}AbSJf%9Wu{0uZrMzmrKeU@^|J#9
z4#+Mm*VM^X0eeFGQW;c3b7xfnk|iLM!8~D=G@~AY@poGg^W(ILY?0M_llaivXf=Ty
zd72>UfYl0xerk@^33cr3E3HB6dgZU^+|Srzp~clHi5a#e0I-8Sk<HS0s&A;9kLbBf
zz=uz@+w#CE)$ziS!d3)y%oN2?Pd5sBn@j0#7|#nK>Tz{iOFZIXocw-TgJPKBKeTqW
zghHKG5G=B?c~;kz)~S^u#dNZfc8{#|f599O2c1Q5E9(*Bi!p>0N62u3M#rpQKh$##
zV3V(jweeunL4CvSKS9|RGQGU4kx_NxD2!j6R*;lL>-JvwKf=SrZ6pYjp0x*Z`~TT?
z>oqe}>#J--jMWEo+bAfDR@$ccy}a0In@i|mE7&4bT`9@uy;b<38NuM>l-U+^WJ~wk
z_PrNMK^xf8s7?~RX@|a-daUyuHY5U76TP!Do={lvVc4#KmTRHL*6#Sb#qT?NO1}fq
z_G*-$NprS8T2SAE;$kst&xGvRW)F&lB%}6|#16$IXYm~!j`3ZRK_PNPOV6eZ=jD$y
zX}6CQ`v^g|qcPYkuS{HK$3)^f_gy(IqLIto&&U%R-6aH1^LH|O&s$XAV5eeuHiEN{
z(}7nn=g}E~z48|p@@zc1tm(?0b9*n89E(te<@|WSaGs9O^8uo(vC%Gw@C^)1S#K@*
z?`PPj(T;D`E@_6z1%mq$PH^#h(!VUTD+1$1Pq2u`4Bay|Opagdm5Kz-WL(Ig=azIa
z`%=B581hvOmR;`Q^+AwUENFHukR()?b^2OFAg~r+cI})xGQM0FWCBECH8<24u35Bi
z18CRk^MAJQ;_p@6NZmKp?8-(A65q!on~xmu-S&gfYwPrjs4L2EMem&3gl-FkIe9X;
z1(p6t(Y8B!F^>=JZge4tpx1tizxia^?Vb)?OsYtoUXR4?M}u0AFSmUe*Tw7EQ%F9r
zFEO>@1I0mma9`r*UQmgv->7{Vv+e5)8~^Bi8A~V13-{%R)Ph|?x?*(b0f7@32Sa;g
zU&fS@-PB!qDE1u#5;|J2G5a!+A?4G2Ovrhf5R4FG102kR)4(3rYornnlYBz2NgZQP
z2H;Q5aOgcH%od=_^t?e&$6Fl4UbArmltiX38^148JS^m{&SVB`g67(Y(X40_)r6UH
zE8-)Wyf1N2{A{7)lzo|e!~sCuqE}*_D*Wb<5O<oMdI=nwuBqdql(%Q}SPPT2X9G0n
z8Mc+@0+nQ}^fVYV0{!I3oT`}5)Y$Ic(v?2-StT;#xnQ$nWXge_f4ket_a%PYU$9CI
zdSxG+rD|W2j5@gS&5QAK?WFm!1u=dS2l?!*I>4cz@Zu9S@#LYzkfAnU<grxO#h=Z6
ztF0magyq}fXznUJj)5hCC8>~ewp0{n{}J7|EL!6D=$eP+B_g<ogon$o5bkvY@YKei
zlK>|o*OTSdgrjn$wZvGYg26dmui&ZwY3)H#Vk)7Lr?)BSeFZrxtWz7N3I#pYt-1p@
zHf9Se+Y=O>G0BQC1}RVfVf{hFs%EqfwniH=U5rsT4GIMHq3q=tu#XGAyHc&ybQWWg
zA0#<61wXX|QM$R!YL%@Ix{jN)!ZrlB6tL548}~_0^o&2ju+PANHbsx&k8xBFT*+%l
zg|L{xygJ*eRY><L+ZJ$>JpFgu9%!;JDc;0oTXQH!Av+XQLJ+`8>L#U~fiZebzQE38
z=HC_IsUST3m*Mv*6D9~KJZ+EYj1bnz3A0ypXc*{jo|AHH?iYsacG?^W(uRbI40PHK
z3TQoZJEZnSGEb93RZGF`l8i#s`_w3|I)w)#oP4CSi@l7%o+fluKVr%08Hl3D#{@ss
z#J1xxYa*@O4bJbjb|MfOK7hLbCB%a8NEaP;N*m2pCU#nw*N|}Pj9zwkeX`!piaOkv
zh=sOu@g9lPvwqw8Sliz60lN@GP$!(iVd~b{=f+z5R0!ANN8fChqF*cd)Go)Kt>_{F
zsa?^A3r#Vf_vg`caG=A!5E(r<(WC#q6kKY@ldk@%$cpp0-%V-rRSbZx51ZmOVMekl
z*GPRG(?|Yyc;Cbbex2zqrbN=$MUXijbaz7o_LE<faBv~K01FeV-D%&IKFB|@Ep7&M
z)&vRNzE|UlthUDw(I7>K|A*a*r}?<Gfa~oTnxm=F1*9K^i60quM>9b{`JbP?Zg+*+
zi_JZr7+uujE}8D>_e33d4hA*Po_Rm8B`G#BX8W_twL5S=us`EarA&A^dvJfoJULg3
zHfn#Sgk<$v+E!+%=0*!nBYWhb=u}FBWDoC89GxDBhkP54?9Vth`XArCMNX<m_b0ya
zi(mZ0#zcE($gVfH$AlJ{Ncf2h#n@=#icohDvd7f~W11)WBXE3wVw??S<<)cSNrA-d
zia7IByhOt9!aaTe1y|_vp`uXVpUES5ed4}FS{G3ujWi(|gE>V%^{S(bOgfVT=<p#w
zmZpdzb6`kQV|FAfwrSCd+11zCbPc!K#d`MRGs2YMTQrBysyPb3=VEYFl!TF$KvFVl
zX0$;gz^OhfkV1iwE`76wBhqJ+mG93)GqR-nCA3oGF(sH~m7Z}Mgdxm}XN2kqb6ueK
z9*0Ir63xIf+_^JvjapDW7KpWKvAUToGia3R)Pgh_&0W@@7R{bphApyKzfiB8nm#?Y
zB&Oc!3HX+3679`-ez9fovM_uYZ}cF%=NG^YV}+K1bI}?_%*ExF*BDL6>@KOd1Q@)T
z+pIPI#_eF0wTp&%e9^3+&4!PO`wZl*GZv#jhUIP{*Mupa@2w}^4Xi3yQGl=>^;`e`
zBqS}Yvo$dqYJj2*26a>?_2uZ_L`{{ofj`Ux)zLa(TeS58?0nERXx1H^%^PEw@iS~w
z^v4krif-2Y%e-a;5Vu6bjz-%Wu<;9}dT0FYsUK{+2>AP7+YSMU9EHs`6sWebRd))7
z24w7A`sBmd<(ax{ceL&!OZ=XgvoAH53BtMK6$zPae+*ymYV-r5JOvk!LO3!H21aZP
zR{N0N<1B&J2<#8X`>vFV|H%H#PQ*3Xr9&=VaL$kRXJWP2FX~P*#L>XOJz0*$RPwmY
zLAK)}^_)EJA}97=$SOeXP4~wC!XV(4j!)Xj{YjIMsp37)I{i<nf0?@(oE8z?JH!KU
z&gkn{)a06+vzk=tf-_*})MORZ8np8PlLI5Aj=CV)SLd4SVl+<oOQvR@?$5+F;*;SL
zzDbjB&!t%Y-%hp5!m-xFZm(##D9!PC&=ik%4v;U@wv*HY_NBIe7<%ohCP)riS7O>%
zLcqnTlHBO!Ag%??r_TJ1eXWgFQE%UrHcH2UT^D!`1`p%BA>7=HzR$i5bQs(fs`g#9
z59Pb_)SE)Lsu>Oa_aZLhO<(d-`yr?YYxC>_w?xYY(C)UHBOQBggl79uo5#;cSAmcR
zU4y(x=%d{g(!Kv9-AI1Y`v&A#yQeu+EPA1)!S2VH&K=o)_I_fY#8eVmj;ZQ__Y-65
z^QyJ>V7!EzV8@3Xqux&poxHlQp%)k(P2C$0uB@i8HYBfyqp7o5u8NQ7EgBI_5|8R_
zN3P4pynn%2B=;2b-|>d}I8<h|vFgY71Y3Ii{Y-|4bP~A8KcS!et0knM%bwIv&UJb`
zA;41_u#zM9>GzX3P!6{k_kJdVa7`)N`1jLS1=_gx-~xtWoR4NgwEx|ldYf39G`Hit
zO^(h)U3BJ97*2UVu|F?TIEYq+d@j+i18<#aF-MOcvFRGK*A4m41W5Qgyy;oZ_<0Xr
zx98NB8lLdw7y|rAUCtHCX1<?J`IbJL6@z({#cXvVnuY|bg2|nd=~R~Hz;2>$r(W{G
z!XV}aqNF<`YfCgFD;;M^_50a5hTjjK(5|_w5n1P=H>wqKSKde#mpc7W=gUrfm|&o%
z<-Bqo#Nrqbdo>Bwv~rlLyowI>N(6&UVW%kxwoE^eci2;iF4t^(If!kAaGLqc-&&&}
zvb^Z}S(SVYm=cM##NQ-z#%DbKp0lQ#vLr#3bR)wEYlZhSsgkS_UiBS|pLJ^BdiPH0
z)+4dpR1njn+cK>vL;(7qevDgx42aPrZ)=1#1)XhBePJ{Od-?rLZY*;V6$hZT(L8CX
z{%Py<-u1o6IJRB@su0a>gBE#Vo8RR(#=^3Ve)*DZ(o1I&t8BASS%}GZOFV%G_ea~R
zS&)~|*&C*8b;ltHfndAVC*O>`-eo%i0v<eEX>2IAX<KuX?Nm?9Z4A6iz|jPH@N3?7
zYi5Ke54r+)PY?~J(rJ73mJdB$zxKylZkMEG5e<evWX=amyB-p?+M&QA`GyON-r@im
zp^pftJNLUi@}tsckm{El4XBu++!w`;X|`w-{7g8mQ3*by9Es*sn<McHi(@AP910?5
z*m|FQJN$z<t&hYL^6&*i<TFB^24~f$-X$6jlRBrL;KVEytmAx4v4CCof~Mqo3w?}>
z!m<(LzI`ehkkSIF@=^@P{$QCdt1U@a;nx}}k8gN#-TJ(A^dJv>p~mewLLtA5<EfqU
z+ZuxKC{zJo#dEdWVx8Aqi)V+K0tdd<Fhml76o>dXqIEqR6<S}9dAZk7GBEvz-3Y{y
z{2|J>YLBtYuLj=*HdANYO`-YX@M$u%Xv?<%_6POzyda#3xAc1M@Cmyev+aXh{U{{4
z?NH&9FYk_4FJ*~Of%kXSjQk1_Lh_Twc%I{djdbt*OmHeC4*L@>4W1C>zSaV-T{{pQ
z*tOF~?SXi5ZuQ9-doZ3Fx>CM}Q3tZSSXrJ#0!16GA3VU458|N%iFI^Sx#U2`1bf}Y
z`pAKwSu2iy_UB}a^a`8*7h=?-2NLgW!xAv&K#~<0o-6iag5h^t8+$<4@+2xf9>}FT
zF-%*HYY`vpNzG$5*X^f-Xr*s3@}sJC3Pu<g!=mr!-S`*>94LK+Hfn-!T3BHd^%8e(
zEKie5NOYjz#%6sW6DIP_Ejb($QFljgk|A=&PAiRb%rnSOKag=hPn_%C)>JPkk%Eoz
ztWcbTLCv0vfeL6rAkB<uSI6NvQ!{lB&smx&``a}pvkzz!{LO8ZXo|@N3f0Snt9uEq
za@ny`ldDZRC9493#e`M4_HUjr`s^*-Vip{@;4)U-qD2i>9Y|n?Itl0+A;Ax4U`w_7
z0ks@gH>=a!dr7VmmV<^Ey|oz5rMCD$#vqfDJ(+rKi6C)VM`3154`hstLnvLSm&GRh
z#qVu-kQ+IGFLgy=U%4o<y)|n39%q*h<@^CjYE5pGLmkJb+amgw4}WN_0flqi1HY_2
z)}fP~P!MKZ{oM7XGw4HfEUIw<DdJ+@Bb-~@S(-tLfhpEHRD}InhVrD8TN8t{qFNe^
zcI?fGFWJj6(S`wAtFisi;;OptKz4EDlTYDqDeLh1fYim|xFP;*@9nmYfoORF&)Q8|
zWAYI7+U62o@akUM5)btr$6#9n$-1Ps;0xa-q`9EMqPJ_t6&(0G^t9rSf3zX3{z76r
zn%cVATsxyhp0It6$6adQh&R=4&APJ8_QXI7Nv${E&Fg#B6a|s(*Ar3Wc0jX90rLl=
zE1NP!;@P2~!K5h@?Qjf1ya20uBpzI!1gj3}R4=M{-j0gSh^lTPKNc^klSRcaVL7+K
zPH5zX1OwSgtvE`#E<_VMRT}%1SL}2E;k!I-XX58sGiKP?n966HK6{-r`<yl>x=xZy
zLNhQuFN83Jxj)ARjmtG7Zx{85s}%8DqD}?&z)OKJ_fT41j)zsbc9ii~)XnRO+vi$a
zybmz+d{LrLQ^)W>?MuDHcZIZ#SL3DUH0!jl4kZ1nyGgily5!7Drc-^}h*>Gx2DQf5
z`Xp&L<n8SnA=1ia+^z?8+DO%OBc{TU<~bX_jVEHV-1YanAZTl2*BZMSe^Y-O?Y5>r
z=)u|XrUPAq-HN~4nup)B+doIcrGUTNkJ_+vobi=;b@n>~HXnuQT|F%>FSDOSlX@69
zG1;01-ej$IKS;z^TvcWEL3V)E<pz*<^Y*|8!I3@5)JA*ogLEqSiIco;BJ5_=2N|cK
zFtZ0a6Px`(#-5vok9*O>hoUFLq)V$iH`46kfX=TK9QluYkXf&rlKSA$5g=({Y|IB4
zN2adYV;^K(y=f4EYpgnBxF)IF;{n=Z&$w|sp|_;I;)ed@2+E|tOKG8=5-!#3$EVc*
z`P4%)@hB5<>ibg2fsNNM*bI}XIN^gNK6JMA*+k*D*uCo}>#@U~ex}4@UpGQ9JASGl
z;aZq%MbkdWI1UjEq)Bhhkyv=TaLl1Dct#-cooTaYg)+BbZj<<VJ{Rbarq5;s1u9S(
zaJwa{D>}R32;;8J5(0j_`b}nbG$37dO5R%@LvvpJ6ZKrB5b4LMp-R+^xc+%DY>xUz
z3xt%uw`%?5*ExuRyZ8qeoPFm&#cG8~&NZ5dEp-}fki)ql+D=+d*4)Jc)jL200_8w$
zi8c*}WB1x#8tvOW9ATN-!(~zbu?3fpwD*|$S8atRvnoXeH)`~`C}Z=n1*vJzHKRof
zPzOz|RXs88fNA>&nNY=rFXGUw5TyAP2I0;C#cdl<-2z2z2e@+S9dVQ-ToFZ@&=R?(
zs$aj<VdU}tYc#VYU!zF1!I(JT8wpVbz5_|n*6KsTWVB8@&2<B@+iksYPsYRK2D70w
zcS1ipZKFUmb=xM5-rq*z3fBv74zh8JE)M_IwnS@p#pISPidGT`yDiWpb76;VkH6V}
zq*sepC6nlomdVT6!1H$ovC%dVTg#5z^+6K%3zXy@S%%I|OjLVf?iE-YT}8ZCkXV&g
zDhli`O^Y;Tl;w`jc#T0fjAaMa+UW{YJ0#@2j`U$Yb${}I+mU#3nWLx0J`xG}dQ%*J
z-i~S=stWie9E<tpng%G)Ewnkn*sCYv*((v_k6y5ofo5$gP=Bg~v)k+JbnK}PltpLa
zchdW@L}vpXmeln(=b}|JR^0PJE0hQqq6z%XRh}0$n|WARKGhR?N4pfLR@G74<#M2g
z<br&MjD1C@qx520pX>QWN>a7j7uv2aPGc&zFH7Tc4Pe}>q6+Tlt@f3=R}7+9vuj#|
za_+0J-mmpT$&cxqKyLUr2f%d!LmSX-H-aMRD6wyY2E(U2?7R5gO`4r{Q+>#qQ?&23
zod>ZU=RR4VvmdmW?4uqUv5#+kAlFN-7U_KRs<}=}&j)l{v##|@tJv{>6h%F`=^eet
zp)`G9cQvkiv2<`S*iY(GhiGswo>E$qkL`ZUx~Yw<YIZQ=!F<FhijjgJIG9*>80hRl
zHNYtYe|#=(qYh?*Ij&qG`nt2l=z|&Wt}k+PB*g1rV!Iv-1<O8|j%rH$)Y>Bl6Su2o
zuRVG&<K6Ye9LDZ3F*uyP1{vhB=;SeAHdf%fP_m&(eEeYga(^+^o``|yRJ+KYj6TE1
z(P=&ve`9YUXUf6<&-L9?#-EYr)3|8Z--X~fUNcU5PC~_O!pKh^8AX;on;3KGBW8Q@
z!6f~_B4JCO48x`z%$P8I&z+XNGR4$`*|AQnXEOnd=whbDbh+U8b#{7SQOUB}Gb63B
z9O9P4(e~`YOmrbrK)*ewCV1^}CGF*FpAk*UakwN=o6Qu_>H)xqJ}c<m?2ZY&&0MX6
zh=2H_m5cKAT`g9rrwXoYRYBn7zG3sUWKHbyURw~9_yc^kI(ki=`WybUnxG3C9FNZA
zC0koso71R5>f&$o^#j%re|s>NEk2lJMHbWTg*|W<7j%}yAh~vK%Yh6j6>yVY7Wn_J
z-j)Xp?$cPHV`N}p|HHV@YmNG)q&KsCP|6j1&UacymiRrtZ6dh0ioRYy0=bqWuw8o}
zVaUCD{R70UOe-_!31&Nk4z8Ma9z6n@>kYF0N$U{}IY`*{#Xz28uFd+@JlDkC*&}V%
zXc<v6coK_2jW^fhs$Y&rSP!#pw6y_^B<Oitr-htLH1>MIb_FUW?MR3geMyqVb+E}c
z#!z#UzSI*iYzhj`U1*y%C^GASZP7xxJjSIeokd$?JbVZ}W9>F|Nr6?{^;k8(s?2ui
zGvTB|eoR)Sp%`!ZuWhGhO$LxtXasiY2M!C|Ic>M#m9?X-*b`$%{VCfk8bcRU1pC#v
z!b6@9=rt)3`Em}%4x<2@9a76i@?6{DY#&Q#I1*dXtIcD{79jk%x%wQ9mnlZpj_D;K
zRcLUIYdKb88afeUHsJAhQgd6_MPz=noe~la9*=Z7-dD<b`b;b`X+GfC@i(&VukD<6
zdOZdzKQ_*5(IImqS0voXN~`4e+1Ko%#zqn8Nk2Z-*zRy?m-N`!hmYmu7>$bpAzXSz
zZEHAqyy6~1H1^M5!MZi`0_yUs@zmFw9B#i7mGV%DO58CAzQd$3`)q@K9qn=1b$@_w
z)E=cD*Wl|hLj)jB3>Ym8@P>wSZ<23?3(2Q??K@2vXG<!++fB8`Ptd;CQvvVfA7YX6
zja_ytmNw@Rj&@r}kfVV`_%U8~;akyZcScr$t?Ff}@0NCvGag)1IRt(Z0IE50wZEtH
z#>HvRP@cE@YD1(CGqt~ni@l4oWTZ6NNv`<+FI#T|n`OD>{jL~82yu)d9%78ga2Q6B
zm(4cD7{hir^n+5MlulcL&1SQC+UMy9efqKc1+7RJhV5Y-hRx=1co~M*L0(1y0V@Jl
zEEO#_#2ACd7>+T-7-Nhv#u#IaA?EznwVwOmJ#UUt)BCxvdtK{V*T-7d`n>JMj4`vB
zoL&3LMzxU_6GP`TtU$Xu?ZxbTCE5nHNthd_znI<NNsW+h7Bk{2F2a5S-SJ{#J&363
zw7gRwJ@);sc$zqlpiByt7Zay1>%!gm?iZ5?(34{EXT`&?PomxPVv+$?;}3A}RNtPG
z1O9oReu?a|IFG%U2^W!!)@H{KFx*H#Vsl>1B$*UZ>8o+BKH`-1RNN1!4Kh61;?A7S
zdoj~3yMK%;Vh{>E-e!J`7^<Oj7_<d1W`asHSqZn;!WiYmNsf+(V=SG7)jSe^)-PRZ
zkH()%>+9{Y_!wKkKcR~mwa4S*z5>n@i-aUNwQ%}8A&g2<OYWoU;RK6=%pJ)s%$5j0
zwDefBU(7@f)?yyf&aK#$nx-ewBvxq%H90&jtXgmTkPKF6@wjlz%8DqK8cjucp??dt
z?W{JCZ-F;iod~KGTmNDv=0rf1Peh||U4@0p)~E?GQ>s0LubJ>qLRqTyW$T4;<<jye
ziR}`cwgK*P8$~+GPf-HPHfcmfY8y1Ym;}TM+|B~!Ev~p~nHnd>f_yrd?(^0*g_B>M
zkhco9S=|v4J#58Tzm#p`OzhIgox3>mv>uU)c$ef9+ZzB<JP(_>KhPlDddLO?7!hVT
zt8CSmo;J%<91aV7O1aa_3~zgJGV^+hz_C#T(G(G9+qGnHS4tIi3J5El^E*T#?9@IJ
zJ4I2lexd=~^<w5+n$@)9L?$uTGooYz-ov{^h(@Y;*t4;+bHst~iFPTDk~}Q&<bIIQ
zV_yuh&DX2#k5`;Iu01&ruh?Am?RGGrBjl#D<&X|2_s)zT3GfQqp|>d75e=fp8J0a4
zqofEjQiK;kIvSlJR(pzvV;QMEfzol2dNoVWb$l=UuL~mL5KaUPPr}PsP6du`;cCa5
z;&fnxk$Kq83L@;)-BoNPK4o+@%WLgie2Sk~(oUY&LUXMK8h-ZaMGdGtH$-ovCHyHb
zM7_#O@zzfz-;_3{Ne#aT8;>8b5S0Rtu4ukE%scIEjj)q+X&y(N&XcP`JgqSo3!-0q
zdA)kTHTBbi;g|#HT`kI%0*T#y{k#`@tZt>fFKG4M;o5jTpdktp-DMxfQ^`jCky>{a
z&<oMq$HKX`y?|jn));C=`=oS`l3V6eeO89FNbr)~46x9;jv%nl^nOiQeMNOyy?w6t
zd3=$+(2#O1Fvz@psUfkATL@;reJ#G|-r}}1`&tb=+XolCf_)R~iSrMaiEo8AXV8#+
zH+761P5hbrqJ6Jkm-V_IsQsY%u4c}V7PKA6-2Ib!bcGeyM2d$71)2(5?3Vg#qTHta
zqT#JYaIq5^lPW{A+ika<$T<HzA{K8yk$AWV_WFH&nqCs~Wz$bcu<+A`Js`!zsLfDQ
z4EJ_Nz=7mj+)Ln2J(5_PFfg@k;O02<L?%}>cKf;6Yj+FBC@C;Y&t=>o2JId_ukf-n
zcCVhdv==%C>^?#CT#6p)dcRtb<btDlc06Z6KQ<?xk+~3<5O?Pa;f6wylr2J-s*5&H
zFDna$9-Dt62{UAA=&=PSk}wl!E)*JWCBjs051&W^h_f^{cJgp|Bzks3EFwL*&jlQ-
zcbh$SA~C3~&GvZU;F<;U&Mwk`5c$Z8^2F3<nBZ-3G^XyXd;Tmrk?~s(&T*-)&&mU9
z@?BH4<U}SpwHDL)TBRSzT<gijszrcK&iWPVd__vBT%&=u;2ZD9Ky9?g2#8kM>hzu{
znfAgK^zMRylQV*Gqu?(ispcBBPi{J#e(N+mN=4#nuE%<J7~3Fp>jsj?^v0M5mfwbl
zY*Rc8b-HUpQ+!zpOR~JaZ09ltKW`DK)*`O-Co#7zNQJU)$TkbL<%m!@NL>u@p8c=e
zx?;N4G+57xOwe|RKa7NEsE~tuW1vZheBJuheIBJS!qQ*>^UPcY+bSZF?Tk>q;rMpq
z3^Rh>T{I{sfZDbTv8Bkzt=uDbr~y2H;^1nqwoqibVH;QPl}roBJ#!)n+^wwC-6t|x
z(Q}n~a%_Z?XZ2%>KMcLb9${IQ?`E&;4J^F?3H01|B2#20nXQyj0@2Z3j1Sp<eax=J
zW#B+;5GIbZ)@xk%*+I?arfqgeD0|*(&dtLo{{JkmBX&fi#X{BZ{+`bTUdjDAZ%6f>
z<Fg5CupKMC?_#X2c3kVg^WwS?fWki3!Nf5*zzHGlvNj3;Ax>rr77q8J)1p@SRA%<G
z8bAdqWcK?=>Gbf}@^gWdryAmXy@-YBe9Xw-qd2>$9$_-*fJXIPhs!lZ%1&gr6j!V`
zKLv&dIF1F>l3c=N&Bmk851vI*xDq|UknL?P7!=p&1Wrtr;=r*GMj8xP;-D#;RKr{+
zp%vHkxXRb?yCN+@!T_;I_FnY21vYN)$KS|tXa-741)Cr_kv<e@RwYxfeH5E%6}80U
zBQJMmH#F{*xHgj*J>H@3#T8CnH!%kVPvN3@^hqH&l*8n+=oWLr08;kn@x+oVW3_z|
z<X8n?`cn9$rF+i4`l~ta>Y6wSsIT=CORNW}sc7E_hXxcCZeIB|Rxru#nD6h@T^$ZH
z-$!>HNHaeOR7EL=y#1&%ByAX&6D;?j{iGhSF5@~$7<kZx<c;Zcw_DK(p8u};_$A)r
zXBRc&?PSJq8;JjOr_S3>W?JK6cYD75WF`}k9mJ+Zb2J5b8>S$P>3UzpMSaG}#Pi6t
z$_(5QkQ&)$UfAi*lZiLNq9{2U?MN4cwq>S3R94p8-SOmp^mts&I+?Mn28yo}Z;$rn
z9^uUX;$)%9iCVFZl7l}f(bdYgxw|aiuU~M)FM3kF*(Vc69qhL`f!q)yv$>)GMX~et
zfF6m=Z?}0TGogu#GuPtz@!Zgr>p?C?nXl1wPW-1W)W<^wLT(<8#=fh4A=x9*E#cS$
z_Go<Q=L1{#v6GXf92aCqP0rGe@Z-@Yat7F<pd!ol!Qa>u0Sc`OSAX?7>>|q&!CD$c
zxr!cUQJ&3#d*)TfgGUn&S(QHTBf`w812Q)coY_xi;(^~|S;I9Nt}AeAMI3hw7`D2T
z8SA)yDDgR0WYudP@b={pX}3moSmP4InwXY8uK6XJ6uATU+4_@{*_C#`i)WBAS!K~2
zZ-^DZ6)qA(xPEJ6fcf15+axq93zVjgM14&$tdxj4e=>2qZuVYF{J9)Su}y>JD&?1w
z%^KeC7A{=)dn=yI7<)PvToU%G2wj?&Wvi`6W1IL-uCl!XF_3gj{pthX4)g#vDC)TB
zVd7P;`+~5_&*EW0!2KWAy-g6%=6k&0b~Q)|2y91u4`X*<A={}A=rNeTJ}6J+Jf9J{
z5`BHCj@s^+ii#YbBY4ThCzxPJy~aY`6T~F3c(3iv7FHiutbIYwhu~WK1A@mtaS|Pf
z+1d0PJE-of@IE6hE#55pTRR+3J3Y(VkpR@pPI*ovNLouY&`0BCHE@re%oKYCVuQKB
zcH40gu)Z5>$P0SRdu^N((Mc+oc`9CU;X$R7v(wR083uk2z+!2hjm99_Y9leuR?eWJ
zozpBNFHKbUF<N9&51qNFK3hCubyO5~*~U{f!vZg9Sk0s-^-`;28cVv%d)mQq5sPqP
z>@8QK8_s|Ru4Qj)<u%mWRrTrlYi&|rwlLN%yLK|u<+#;)>Yo2@46&~pE#G@W$psWR
zroDeM6D{PF=X`)DarH(>I)z6A6B1CdB0nlUcp(D&xU_WKA<^ZU8!>`*qL+F5iH4k{
zKDXk3D#-ZBcGyil;${yA7vUKse)#rz;Li*W*%x|A5h_<`eW{OEdrFP3N_~5(CNA%y
z5K&Z6^BXm$Pz2s{|7EwkGW~;p!vW?yVa)vyjS9l?_v#jVg>(8J^h0TwKWd_eo9w4}
z>z9MAb*K~V=NJMu%c9-Vr(1f53-A29{UWT;Pm;;yRK_8Rj?Uu{gdB9+sf>qH5&oa-
z_EU+6b6s@9^0ZTl4<}Er>1ybjahq`}+dIXa^GIjnrxHgWO2Xcqr_wRGymG1Cbt+>7
zJwvTNaWhY4l0+S|<iQAc3$v7WvesM83JB$hLiYq7tSchHPbH?xQQ~Uk5xXy%twz6b
zzi=YW82h511Z)C~@rJvyn^fnh`$%%xY;$AO?2LNy<UgQq64Oa!nx__xOxk>n2xkpk
z=50Z|b{jpf7X~5LRxhnzY7dKU*aD08NT5$wD+WACz@vfwFs9naPGuaB%Fg)jF~<}f
zjrR?x{TB&uPe|fR^ofi_d0QNRR;{kLC4mf^!gWgJ!nMwwVksgXuMw!)@m8%8%`&5n
zRqH1oKhZ5%;x&jBeh;sS0lMaa%A*<)H!rwi-KmVzr3H-wH&MMl5X?0XSY!N2_>Zko
zU!?qVn-Z7y0N%&#p~u#1%<g93PQ4q13V!x~vW>Ac@MjvfO@RrqZw2>SuqFZT;YeoR
zl7TI#;a+~&s9J`sMQvP{k~{+10=*_oYnx9csc-IYtQc4^Z{-At2MO^xcZZ%Ht(QtP
zxP6Hc_Xb*I<3l;Gd3D_Yrf}c)gZhYm5hXpkZ>t7dP5vy`<qwOtSNC|d#WoGTbBN8q
zJ@7!3@d7eC0>$1OE8TXgWuc4f&#q_*`aSqIJfk&1=L2XXcB^MUs{2VH_^jxHf!vE0
zy=i;&m3zElB;PCSD3Jzn_o)F1lv>-GY=3}FX^4_bj2#FZb8a8BgX$IiI;IiQzX&4B
z0hV<*nwR^=vLiv$D*mzOVvZhM^q?J$Z#;7^cZWVZCUPV66x<DyMR{CZdHz^mM=xjs
zRuzg(t|=+liBp**jRuyG;BY)RB~sQ5!8A?_N0POmFgY8jVC%OL^+MEQs&#!%boV4}
z=cDObu5cFxWhs6dWMG@>+DnR-O3GhS*Ao{hN--t^bAY=u;CO&Xx^?z=C8$NujOFdX
zg)=faf^h-7s;^g5yAy&=#_C?X+^z*2zY(ww-xbmoFvj-+>hi|rt849jq1`fZk#u9%
zHDa_&&_8D%3gvn{#_S_K4#FkuV}Tw<EV?0RtH_bwX`hsa!&-qCeH!!aWlij6P|lC%
z&!SOBN-%Gq3jz0NnRuSNxVm7^+n4I1p??S_^;JBFDhUDn>u8ON!c`03M3>#g9`x_R
ze7!59ztf`^A+zrTp-P+&e~3Q^NXYSHeBYHL+x|~F2dV#)5<$DnZ$AsYDz1RHv{*y3
z8uDZP7flp?KsY0ry_7NO@@_6Sx4ks!&dG#yw%P43C8o<A9!=adwX3EQtWAF@W4z?I
z^)%KqUP`<Z2@wqsen@x3$DVo`ZF_v|X?`7vyW$<f94vEYJaBxVi@E!yBqRJP{bRFU
z%9uG-k*UbrYxlgAF@R)4&lteHFJ%$~W=JF~`1*Z;4`A`df>+(ICS;Q8Y4XeM?3a=p
z@z6tyY)<q`YiF`^O9MxI=(Pu?23|mEH!nV#M>NfRb&*oi*#e<UTrWf+1}&%wtLwIh
z)tcmF5dM)ssIJ$mVY((3ANQ9sE{%}nH8H*J!DE4p%Pp|%zwUx_ohb6?;)6f23!jL;
z$#B8tTNC65EPBZ$F~sB!blJ)^V9zg7si!WqZdJk*$6!cTEz)^3VzLpe2t3kkb1lO@
zQKOGmQPz}j@*rDNcgiqkbwZqzvzxkFR$l_ajh2JNJqa5F3o3HEyTKaK9S+MK{H4S$
zDMhnhFUYK%w++IM;%I}g&0gE6;pCc{Y?COlEcjPbye6`{y?My;fq0jPs=8-XOFX}`
z3)?*TbF59DR~KAeXPec3J~vE~7d>`*fMAy%Nq$KnxxUWhb;lg{mexK(#E|vBlraUY
zusvDs{y~wlF6UcxYYhF@58AK>wkkJZ+u|veiEg*;0k{EnMq2cE)snmdY^R9tX-kSY
z+wF?RZebX*XN0l~t+rbP<#^wjoPM6wPvt1plIDF+>;o?ew^!f0(#!S<$!avM`_(ts
zVwCiGJD@>xsv<PAlezUA42%Ynb^DMgfCArnwfgU+7;O@AjtK6mi7^kbvFFrp7s1O%
z1)@CNAdU$H^URsVW;z~p?dr7`qVpC4H*<C(b`kOwNn!LA`x%!a4wlm~@GkU_XEktM
zG~0+!s$5!C_GC??ozsTyK_kwx?;+ImyqeXZ#CHF>i!WuDC$!G3k(zEakXyROE=7mb
zgRwDnm>-4#jqA}B{B|Wq!6CN1+un|^D!3+G4Z<YHi&2{tVODu^U%NI%wu$5RuAYV|
z@bjKJXef3M+WYZx03#Ia;kuT#Cy3&r`k_XfJh~%q=SQMoa;J6H@NuBFytdMA1dPJg
ztZv>XYQEB|b23l%X&{luZ`Tt>-W2{Q_n42*V!F6uP~Su-^|_{-0w)@}r>gu?2$r0h
zT%+`rzNz4xW*vPk6fi-c*F`S-CO#yI-qK=HID1yRZ#8T9K=pUKMb*9&a+`Cc-L&uH
zIi)>HiP|*xNpXoM)sOmuD%u1#{UnkX-JRiQAwygRk*|cw-BJtRjP{H0XHB~+r=8Ap
z<<19gJDoA=P8V0HO?EnCs-TTX-fh$319qX;1(>c6>TrAFESaGnidYmm!tOXd=`sQa
zm=3%1^wgD^65MuIz$$blf4P~bCo@$kFGDTLk>c4>?>?P`3l{6rDw}mW6GD)cI_;hS
z>fCssFqpG@PiO2M=D`@)yZ6QC?q}U)_iMz<Q6J3K09RpWYs39u&gsPHJKEc9uDX5a
zFZRIcOpD~kKO$@7Cod4F_IxGT{P?b$IU!|Rz+1FogDq4`tO6;i^TShP=_c>!BWjS+
z3V`~f@%2Mb&$GwWmK5(j_IQlLU0~^QkzV3&(PK~O5erJ6E!LRbS8s`iHShnla!ptd
z6>eZNmHPBaF9Kh6I<eBSl~%16_}6sU3PD-E%4#$TtP`AHYvVbDe)dhBdg*i%cfsmU
zXQvq20IKS_R<lu9*1}oV#1l2SN4Z|FQ|BmodTqVnk>k`2;~RobzyIBXwo$lHf02`J
zQvgcyb=jKqWdn{*Y>B+afg?`KX$e)HZEFNu<hZrP5_V;uZ5G}n7U(MwC7idSI)pnB
z*w7Wvh-qHB5$||)>8cdG`g%3cGGc=IOSDRLv$#GF3S1eKLYJ;Z;qHbVEQ(<P%nvbm
z9Ov{2Ml94o+XKtL+C$^ELm#a|tHYgsr?6<n_ik4Z7^f$k9`g2#UN$sT+wMRiv3T|)
zAw18DP$<JZD$;d$drBj5qk@Zhdxbd`3_LU1ZJ#!0BTjQYeq8R4nIlh@9f<D{g*=@&
zrOPzw2xColC}xU`oZO!d$7d7&jGA4zHX@Kdr*X0sT^=~9ACkCJuw!bUXCECGN`BOU
z`2~9+VE+NJ{e<S5g-vQ2SE!%Tiuu*w+v%y#T>x@@_gQtABbgk>_DFoNy1tU47U%Rq
z1qv4rFS7HRK93(p^BxNhhYs%JK?9Q9q_uQO__(GOB@g(F={J^WDontIJaC8vq~7R?
zkbdwfdpjuL0a@&=t0FUjfSAXl_moWUTKrAPUJT;_60I~rw0_HbPiXgf4G()i;BZIb
z+=Jx{Y{wRM&WCZ34C6Lm8WO{IHR#7`N3>1OZisqrOmg*l;6P#79{W^nF_JXfO)anr
zl#IUjJ`+v|Ih+fht3^Y<7ttWKn_A$@!$SHpo|(I}0qrZT4)g~0b!p9cG?9H1nBaO0
zL)SF+WoFdxqHQ-i(=GPBFdfca{a5=z+ylCzCQGhoZt?q3aD0Ig3zxT_0;4h<KS}s8
zWIvzI?y-0>gznfQEN_WW$wEl7bae^WBg5O7j1l@V%;Fxm+s<VCy&o+<)sk*MlbBr@
zs_$uMGD)BnXYlD~5|c#+ckc<C5ie0I_EU4^j(BoKCDy+?1(Z4_&Wr9kGwCe9Vs&k!
z&5WjGdBWkq?mm<FIDTs!QnR897`MlA**)>q!q0acuzSyB$8aN4<ho2&@O|p%`T<|L
zc7H$vz+k7%K9kwFS&HbC<_N^~BhEoM%U`#-XEJdHN4Rd+^gIx~KKRhD7TY{gV7R;6
z=Es|M3L4o0p^Db7+ZGDUKMYKqY$_1t7O1fOSSF9emlMa~l8?sUWOO6AE&lc(Nqap0
z#t^`*T+p!%blDSn<Sx=~i#2S|>cD<k5+kK2Pd15PeDv+vH}QuLxGOxxV^siRPm@xW
z1aG2HW%80+p?;`1++{U-jWZWY)!M1$bCg^6Ybe6%L{qdn7{Sw_Umx9j7XE@YPJKkN
zG!9dWn+Sh$fw&X-IzhpB>}qNcPG27j5xLkl#1QC$81z62Z>b1a+Z4^Z^Q2AJY^`U3
z<^>cV6IT;E<8BE^Tw`<Ara7y{+~8N&%_Z2WQXZo#3Sg%PxRt2NLDy$J@iB^rJs{$s
zdrKRq2>*`nNCMtH@v;rZQ_A;EoiYtR#ra&Y;VC>Xx-P@T8`~CmGQX(O^~Fl0NoH|}
z=B13?Mus;%*CgY@E<t3gxLwn2x6cIjY>mq0w!1XYQqY4vtBLRHM#9^paaY#V*j~LP
zPkN8-(}d#I9O;wx>!)RyP5V4L?|?p9=}`*@1L7K&^tRfe7+s+Wzxu;xlGv9@qY~GZ
z*y)jM0r{czoRIOPi${aX6;IaNu`}5r$Z6)cgU}ovpZXbj+)vfC=w6s2(f}%EJE7Tk
z4Ww_zQ)edKS%o+x&lSQ6>WQH2baa^Oq%^oNS;KX8B#=xszwMkzo^{A1o>v#HGxP)N
zVgRE`vyBEmAM7eg7?-BdejdTpgHgtDiyb?YgzefY8y7{%e%xtSgiPDSNqakx=`A!h
zk?t*e=|(#>8NZ{i<UGA52u(vAIPdE5`u_ym_w=}79)1q*$Cs>^O^fV$Jh%?EU>^o}
zGNjl?`evCcVLsL)S&RPBZiwWl%G{@23z?f$Z~IpJR0L|nmCNVgX29-kZ@15)&weDB
zX8T+#j1z_sChn<UhyZKwDf%*&6cYMn`zn?cs5?JGVFlo?)p{uh<~L_DuK~|%g&&<O
z@0g!&HR!Uf_MOmohDYq1uw_GWm+!M5G~QG?{>OOa65yN8e$sGj3$8M<p93IBap)l}
z-3pwSFRLaMYy3^obmG&`CdNfI>*gVM_rL9I#)oSP-FExg#7nt)58AY7$w*rf(Mcr}
z&*UL-3<@^mY~tBv_%gWP-W_K%X3s+LZ2ETwtYxnLx$Ep?$vYCU3USTPjj)f-Jex#<
z<^qLB&t?)tzMItVZYrA<FtD(alr{PTpOw9&%f46W<?<x8xBJd!%(f3FR`)x*|7>>Y
z!|lO+yFq$(0Gt>DsW}43JQi(kPy)Y3FMjxdkm3G)4o|zCHt%ev7Fdf@TRaS~TqqMR
zTq?OCb=tysiT5=J-oqjc*DDXQedKI93mC}KzX}+9RQNYz;&D^mV?u!-uOX~k=-Abf
zyKIt7X6iZRZ+t?14R_@>ljeJIKt;b5l`TeWiJG9q^0_IGakIL)j##DE1A)+1bvBcX
zJQ-_mM+&&97G9_&JfifUY=w|-7+`+kur(sqroXe=m}J*~*QsT0z#X;oR<9nnK%B<-
zZte2Q#%f!m;dwHaoUPMyOK|}G`g)B86Sg~nZxF0T&ZLdeygp0pxG5$d^zzoE#j(~)
zMy4did4a1(eb2^jIh%0-zZ@iOu<zuyv+3OY_21d%z!veAffNOjiw3fX8{WGDA-@_U
z5cUYY%}AuZ8qJfv?$+NQz$hGyUTILcu!h>%9a}{>__F6hvSF=XY(bs2ExMQcysN{u
z>m$O?J<Yc5P^T1&fdlSTKa^i^E9fpkZ6lMaJ)?Q-&)aT|aEO&tNVV*rRr~Av2>aWk
zM?zG)Y;S;XMs$o*a9_Od9wdBp;*9N&DItSl&JM)iYYXV@4~kODnTJC{dTDYcJFH=K
zQmDYyUPsjJ>WZ2Qdrl2<%tb4!{HVs+hq8;UbS&0n3wD6x(HWLd_t$<wU9VfW&Q53v
zxJZ@kRd0UqFLtVgH|*U`3p5_aNbGi2WP$_0d)&l25+8i9pWSj!8>qa1)Wf~d&I<<Y
zz?~v4YCy>zII72Pu1Iz%R$zaVjRk3{IB{{oL+>8*#d>9-UWuV_d=RT>Z;Ogaw7085
z16?EvQ+CB<5akbqy<bz0ja~};-GJ+^G1$iAfgOje^1i_KCHv0x=#-2}?%nvIP7pR(
zgr;&O5A;`c+s9g74PAC4AaQ($Lgo|Agk$ij`oY5zhI2DMCBp+F{VWhqg~at$d!AaO
z@m0z^$AY&n1*JX}%KS>t4TS=cK3|I{aJ8;JtVZ^YhEUf#al9|-i|pIz1;d5Ay?!Ss
z<(|afzK_|)+|_13=w1KBMQnaQ#+PmbU{Ze4O7&!+UWDxDAP)HlZ)pl}>Tb4Q^tg68
zhoilmDW)kQgLVAnjC;EAFs<a<Urwhtc2WeWl6AnQy`0@#VIZ7P(_hZGWTTRp%+PP;
zDbbIc2kv+|<L?as<m9;X<&5DrK&pxVAZ-2RBs~z?i?f@W`;tMeaQDke(12gML&U5=
zo|{pUjqV9po(Y}o&0dn@-aylfyBFLy{=S#vF7L9g-5&#kgN8)Rw!CDsql2~Xg9CF9
zTxy(a&W*pJ%aA<~t@yMOA!44eME${$ZS!AF;>etZb8Ug>OYQb{TPV7bT!7Q}VbRE?
zp<Z9w_J~N5XTp$_?9qVtAn+cG5x6;>@}e36VHGSi67ExEG#dYY?m~MadUFdWL;SNX
zR{O}@lo;`nfQXk6e&^*e$aeCASY`aZ#kXgbhRFHso`cmIh)2TU<Q7m!9*J6}CQv9}
z-e|RI*EQghVx0hCE%xlQ_0hJXtjrn%#>5HNTdffl(0PNyu&s;N5S_N`V+P#G8A5DO
zFR(zQ%#G>=X~_@5P0>pqDLtBEMu<5`6(aKLr`!wZV7G74<d*d|Tbn?yB?_X;Hou&4
z^U~H%`2?*PAHkyC<knT92FxA{`OAsXm;cs!HG+oa)-N2d{|A@D9X6<bpMGeeZJpYQ
zB~jOgg-U&)i!0SOEuOYi!rIyScCEKM&#$>7<`CsUpY7D=m4!Ym_q$3c@A%*^C{&+O
ze@PzmIo>VOppx}s?)I!u>?cp$p3;1s@(7>3@roEU_t4v?PxD+7{F1ysW}vmuW(T6F
zJNoeAV4w&QF4&=H)J)QV!y->32M>NJM}%NKvX_U)*>h@xVcYlXQQ`Mn1n6Tyi!DjY
zAhk@O>sgxf_JX#8Uj)+){DPf$d9oA?f|jb#gW28FE}jaY9MT@<eOgrf2j?($R?lSn
zWwu9RI1|rr&e^%rmqZY;`_79dcu*5uowJM4mI9FdHX46(&gJZqhSBikDjU<2J4=s8
zyLT?-hwVy`YvLr<@Br?n09K$on_SgOsNh7jNj(mBbFsaqwNo-p;~4U;XyR8`*GPIO
zc~A|X7w+xvYYrl`23z;_sW~Vn2dobTg|XnJ>msau6jZ{~j$(pP@nd0=EL1n9;HI*=
z+><_0i-A^7+)t-kyuv|!9s%fPPy&AfvaQG88@B!7Df?U)QFFm{yI+(xKNs3B1(kCg
zhqfRsN*@n2_*yi<*o~ut2SI(K(NLhAG%Illj~%_4Wb0G2kIMk{L<1mR{vJm3g8;fW
zFK&m3RzC(kIT<M7_mfb~^$k&A*M8Oy%%j(GxTQsdsuUl=Ci_K0sB)EwjU<+~xyjkd
zZ6k^C)GuFVw~u5Tx{PaJpG}KbE{#qf$#_28s?%nSB%WF-$a06i!yyFX-8qujF{<Y#
zyDM690xzq3(q_h+;r4fS+T8(t-u#Db)=0+UBiE$eQ$nijw{~wJ+XTZVolHQ3P<{!$
zKU#J74%lo#z)947jOM7{Rn5z6uAXYKt~?M}xNY0!X<)82vH61IS#J;r7U;`9Mzb)U
zCeAS1{q}HtOC682U>^}E>d#@)j>m>lc*<iUQu)$lB$If2B-7TooPsDGT1{INgKDQX
z_Y=_@V|5MHVgYFGMvk^6LbW03UaUMQ1*uW%RqAcgtH_f1Ek4Itt1X#2s)GnvEONG@
zG-1B&Y61$k2jc%~H6&D7eO4zF8vJr&^&^=GN4Ao7qG{N5jrxt<;DMF=wnoUf?w6B)
zof?t7p_J-k{Yc_@SoJsPEuy<y1VMk}h$56(<D6Cplb7?GNk4kfHjQLLC~8J(aC4Y7
ziF)M7vAjmIq8p=Ki$+pKMA5Ld1=TCdD{OOo$9eG=PMu<jG;Xr%(lAIAit*NZMDd<x
z>lH|JVeHy|^|;<IT7w!?%FOGsn{AEnUBSVvI{Njp8Q{B3&?-}Ivh5?8(A!nOpBfY3
z4#Ds39wv57k5%rmw=19}3nocm^wRyCY`1_FwxV5tyw9q;rXkOExhEq6$M2}^)xa@%
zI^z4HIl3K07ewIwrJaa}@&S#LTMP;<S7#gyB2saz%MPh0D(*tV!y}n+M7=VvI}_@;
zBY}X6%cz5%3;4x}=k2Iy{@@?%m<B@KB`>FJ$4iLgxx!KSf)FpmZp5ZM5txuXjZ84Y
z)XSk#jE9AJS^!<k>aMtFgLKFbtmYvb8Oan=q_XokDHL3GF3?+r!}j@72k2+*q824(
z_&g1DvyDbep6ycndG$|u8;hmi)oJ6JsAZm4;)?Khv5+JAZGrStgKJxlT@^vfw%DY|
z*2f61=~df+y(`Q)0~jw3YdEv1zg2ib+53V0yoY|X$*u<|1YEx^emIikoEr9hlA91m
z;3OaEb5s<6wvY9=ybqb=MhxqB8|{-AYZry0K2@8(AtvNzJYrZ+5jLM`z_r0QKUecQ
zKb^kN>gm9?)n{L7P>8d>dc1PPSNevv(`jF8PD$0-=Rt4Zh)Njv?OXL(9ap&T0^2&=
zMZQ;muwx|39|UWlYk;BzMf<TdG!z9tX>yhonkJ5OZrRU5rO6W~+|naKGyV39pb>t7
z`W!9!D;b|{^ITxJy^?WcZVpItx4)A3uBW3U8Thn#i_Z!1*F>XEe<gF>;@q67FWP4_
z1ho}kF}vfHYz?!dJ^i-3J>2<9k^?-h=PotJp303nx{0(>94&XhlEjClTT!mrtXDGD
zy*A|OdtOP$__8&2Z+w>=l03NhzW6R=;QKYQR<aMEAAcn~z)QK(s3sa1l{GZj+*i`-
zbE(_CS6p1hIQc+O!=*#^`aD70jP9Vz<_lslA~x@W_`nYp=5nD>N0IM0*@pu`_yM0k
zB5=gsBk4UFMA=ek>T`wfWBS;G{aO2ut2-*4d&)nY)n3WY?9C`2CY~pYYYb`d>c89K
z__G6dxFtfXfg`9~9dGI0jH6zqdP3%LyL*+UnR|!}@1*AUbJ(f_|95r{yfbDiO2kR=
zC{`0dE6Gk+>?ZuBHhOO-3vr!h8)*|4bdAF;jBb&(MqyOj-hp*<jqvr-OzWmbiCp5x
z{d$4OyXm$MCD8_TQnP&JQroD<D(AkN)Ir`O6|6}hBSCzV<pW>Ju3AfpaNTfib(AB5
zgpztVqQ5ffP8sC0>9u0CT#Z?^EB+in+uKtjj5}Zc*?QGAe313rud!|V)vs(&Z&3;W
zacjJFhRxHkW~RD;&obFt1MjZF7B`2v259>$nS_LC3pX&gC+*N@D{+|S3c6FnD4D3p
z=B74D)@`?E^n;%S@TJ``*(vzfp4HnLtXG`;d*UrCebrLit1*?61#@4JqNdPl`!x+D
z!bA_Q(Jt$Cm%)S4vK!<7q4+Zok2o9~eBs~Nktw`=mR$w&T(n!o!FV*Fv4X<e>{y`G
z0yvVlP4&!~Z7;-U&HV#*A|{Z+-OYAN2Nls?nK|=xG^qD_dS^?0yqd(oJvO3|R^hdK
zPE=uwjC^pw&PVg4aq-E&7@)TJ)Q-ldt}OQJv_8dF<nq{<5MJSYcwD4HAYzkVQO8AM
zv)|VGMS~ZaTVsH&&}C9Dxh%qp2khD_nFz!-9>E3WUpVAO?A?Hw!e0Jl@5NKX{erz8
z@E-cr2D>gI@h%*$#_hv+|Mc8n*+;@0{z%0#JrZcpZ#RMjDE|ou6CT~ou6iuFJANu!
zpcLpuFK{!yp&%HJXc`R*9MO;zuJP8Di}r=yt|B1ZW5U0TVXzAPCi<1gNOmL+=&$3P
zXW34fuwmCF;_<<^+D~QhOZYuI;k#H^1g!dPT5Hmcr>G9b?FWsxw?F}h9|J$nJ#Ie<
zKU|Vhm5ZM>kxh8N-6|~@KM6Ty?U&NXv3#4IOKfoBB>vxaTRgDsq3P}CvQ3ZWYcoc^
zE%=k`x8&09E+aPmT;k^4100gLl%LBuDKaJ5MeUArnJiGbw4uW8JeP5k{%$f?BDCLi
zE|U&cz{iJd=DBqCaXE%b$GfD*{bw+rbuMvGulk7`RT8u99zk$7%Yxk-!*It1F5Bla
zmK+|6h+p>y#sri#(f@2Uxc(Oi<Y5eR&Sj#7rvT?iw8%C$@S+}X!5+{k$yw$}app}C
zhGd%&3pc+s5QJu1a4yLc4Qp*-{MtQ8ywSq~)7m?6jCp@#s^fjgqJQFAa?T!|`tf;q
zWb%8p$6}16ibkM%TtmQCMF=$J_(h^~HK`s}v7MiYj&qz1wm2v-F_sD}EIF5irr-Vk
zAuHEsc;~0gUX{_5x{hwfs0vb}U2;8St5pXYZp}m$9)oPe!~C4i0$0Hk?gkl0tTs^i
z;{eO6F8;1_m;ZXL5w2Pt)~NY(ZrE#UqAB5Q*s$a8K~H|ZUZ2&tf@wo(^(M!RjWHo~
zJ(_0FkD^Uegp6r*Wm;1#DK=|Qch5i=REYSFQS<9>IRIMX30zyyAhwkjVKXNm?6BW8
zUO1@5;Q!l$Rt!AmJ+$ff^Q=oaUjKJiR!<BK*CL7r^4vRx5?9n@4jZ-pn2(nJ9vjrm
zy0#WF{>iqAI;|;GXjspKsEl&9ExPaaO=8<MjvfsDI|A@KFB9pluX^AF-L^{|!8lMu
zGuO`qsoRH<Wp-<&uHy9M-uG+(c8@qL^Lw;x>UzD*_OR_0d}^68Ir}tF*As<5qU{gj
zG<kU#3OgTA8$a=qm^^9+g_yfrphC*qAx)VJDDIT)(|<V7B}g16^T3YlNp~x5Y9d0B
zwSQCqohK!Y@iF~W*4vs~5A1kQ5l0F1EHA`hoc}s+vlD@LR$S#&O#PD;c3QJt@6<o5
zj{B)%KBAZ9uJ=Bt$2#l;cK%#;?O~h*81;E(6ZfmX7?`}Xt9f|HMzyg@Sp{8Z$1OZ2
zwxdw6F%cd^8~bcrQ`6)2Wyp6$XVt_%=ToX8ZsY-=xo60(1|1Rf`)x9S!E)O=+!}VR
zL}OeR?A^etJlRLyQ!n^Ip<%|qJjRak)Z+%O$9odWcucI9_4qJA;md?UTR2l-2qn_T
zT5Z+s{T={#BY?pNh@G`h^x97nM)c{_!trRIH(aG*H#MowhxqxNeHNo^;v)2U{JC(>
zL-s`s#HGv{`%<$)u04{k>fP`DU|$QO+nU)YBw!0L1*-FXD-s~sk}{T%F%h~i#a8;J
z`aVX|S7fh}wng8pbn7vh6Mu{j-LhM-pG22B_sX-MOJs0^%&6UpWw*N8ei1?ZsAYlL
ztN(?8c{vyAkiME2b&?|gWVgSX_;PPyizg$pX|HB{c`4}UxXw&}HSy(0ZpKh&yqY8h
z527JZ?A46Trx<el%HwS=#me2Kw}S&k7fWZpn#mI6KcuXX$4cBC=#;Vw%u@fI9@I~N
z_q;mkcc3!<v3p<5m^kSTqfW$qfgZC~;nk$<egTMeP8OG`PLbML)1_|CtBDWSmfPGw
z7>zvg-ves8k`fuD;sob;uO_be5Qq5uK&8dqRThY}6>L(E|L1sI7!8Q`!LqHv6LH<m
zls+QlaY*&km1-(N-m^7v4_6bc+v#|;$76mNAK9r2wkWXq^+OBn33b3>OW2$()(<5=
z`6U_-A{9mAQOaAn@J7q+N+v&BDq{}lWDCjEs^Y2k$yMcqzElSSL<w+fLAxF(KDoTn
zwNoinye8UvB|7L;8&9570*!N>AlDFuUSDd7DU}Omqwu}{qF?-NP0UK;fUVQ0TML-1
z*NaZWg9C7g4f?WS0OiDxZ7eO4+KQ#NNu$O|pmzvC7(cam4_$~I6YRWaeQy!$8crqL
z(xR#BW^A+O1FL`1k6KaCVLCknN#N;vHHo8;kNi!X2%+rj(XYt?*;fe5xmTzY5aE@+
z`eOm|@usJ2aOy1{@od(u@phoeRe{4o2jy_2)ZHfRQ)Ob?t|soClu}Xccs1iF6|QBR
zO2D-<2(h-%l%{c)K8=KCtl?)$yV=7OY<El^sz9`H_N=Ig%kiM?5#p5vSLWK@K+IJx
zX(0Q8tc5%ll>Gs=8~%ohT`Q!ABmf5mCqI#rHXhP?h3#`7B>Usx0N^nJ1v?^4Jg0;`
zH+2AThx0&yqtSKs(y~fBrs=~jU&<eTTmYz?Qm_|<P#Ns)L}?tTq_ELkuv5{syPR&d
z(*fB<Z96OSwNORLqugyokQ$Nl&qX)IQc>^o@ok4&*DeZ!N{=2GjYs#n>&`dXr4kC~
zKG+Qsf3h)wE#odft`&eel*|4VjlaRejZs&<9WC6%^uZy9a5c`50!PMV^n{radvDRM
ziTYTjJ+R?jbydBp!QK<`DD*^n8z=7j0jyZqY}aE%dCnR8P*{2V1DV-0LPA~P>>?J5
z*!HC7z7Z|luJ2{7KM?>|x)CEDsNtfCv!9<cHwBSU5w7lm>BLFUB1}9`Ne(t&#NY7d
zUi(u0Rz$`b%8bW{h?>N9*w^aT%R%Su8&NZxjc)~l2&i1%6D@rgYl1DjyxzVqb=Nmc
z#ql6c8yn<D0q&eQ1FQKdx=kVUXx0Uxqug#q(<JzEQ@39<D-Hip<4oGlXIvLSxT`=y
z_4A3HZCGfxpHJLsHIB5?&S(3+EYCpz)v&+2m)~@Kj^v9wg3Ztm)uHvc<9rek7|<Gy
z{5#Jl5up+TjWguC;w2U!zkJO+pPhYp<gg&{UgS`{I}qBk)n*A)KTG7E00m%fj)Z-m
zPqM}8O!B`ko;fhmWzp^zVohYVK;D=g4fx&#BLc<*j{u)nZSMIby{sWXsU$T&a6U;7
zMNTc7r;m}?+xl$2pmPT!z|R&$V|Z`Q7M@RHPDS?sLhQo<*U|3HZFXbFBhmEfr#C)j
zj|%RJloH^vK#J;IDW>4@pvCf)w&;8&{&=*9A0e2<&u1bUVuefk?q0lD5G%>~IAlvS
zOv1cyQVZO|h*K#5Tny7Vn9-^jnx~YBu+M5u1Fn^}BAzg&`(CXH1Sw=v8>{E<=RRe1
z8nW}b+{7zZA4A@J^-s;#s7^WO`f+1&1?QST3&w=8eVvBj<pOfHK0tTlleIx~$7Jf_
z??&|%{F6#)fgs$&Qw}r<`7V&p3o-%5P@_e}VTF^;B_4@-^dyD0S>ICQ&x&eW<vPMH
zVS?_>D<}2@Y?yzTTzg|kn^8Gzwf+F*_iygHgF?CquF=9QZ4ClZhZt5JZAly90{pfB
zi^?TQind+fqC-W<q#{amgb>OQcE-o0SG(dBi?4EpJrl3o=>f&_?pRPLh^Q6gdj?g)
zvw=m08yNQl(KpOlV0%S1#BP$d?TcoFiso#8{7v$Qf*puY{IY*A`l91nPQpXNxs4Ou
zm+s;CtlbaRBPFW4?1NS9xq#A+Tg6e)*p+DLsgLOg7EK2?;N#~rsou4T$%pC%{X{M`
zcOvMu6M+pHX%DwOrHL$hea6#5HIHWItX8KVXUXk|`ey=3en+nvnYVMoq^Zz`&BM;?
zhhomNy4l6(2GJu$ZH|`akzIo;6qUwBlMb9fohD=YrGHB}UyYYYfx7UD@G6<m-AC%}
z=psufb2VnWJ(cO4)ZxlWf^-VOSkl*oWlBweS9zet?vGNP@0I43TZz3dgvbl&Vc>RM
zr%*cVy@IKI7{76d`qkQtB9Gfg;vcN8H>s5#Ni#?7V=bkA&U%WbX)dV6M7`7Vlm9yx
zHkJ)H08hT3DmImT3_a(SD>FZhGpDBAZfZ2wk0nj<XTrFFUFlQ*xjJls^P`AAFSX)1
z;FF0Ns2A)@4UF(84jrG5uY_A4Whd<GfRwy#d=q~Y)JD6A_T<}GMBVuOL*?&c-ro63
zhkdX9sX@fq_Jca=D^S%w`Ahw%VKks2fi?XUKX{P68^TDKuCF57prdyy8oBn+T@$+O
zm*{mfdLp}!Y0MF7sDFAP<H=Mlu-h*rP6v^rqBZRPLM9Y=sWh0#bbUY?2b*yrv08%S
z){wP<e5>&cfN~zWxL-*8E|+4a?$S%ntNwPIc_G`Ou5RaA%(Y!j>fKyvvjST-8{y_d
zc8~Dm3y)T_d!s=||HP}V%)KxABkGXtd;f(bwYWR<><g2A-LJ%+V{@X9tOS(Jof?rl
zzQq%Fx-Hm%=f!vMPZ+z}^<a=*5KmnN*i^4AypZu*w}khQijs5kVGV|2V)#BhQX&TV
zn7iyz4W8zxH5vS4qBZ1l<M`t-xEcgPpVmd1I$jRj69LXw0iJLA76pr!^;;tN#1e5M
z=lgOEpuEswl`*uZAHt)(O6aH9WH;pCi!P9*;D;3&52R-))u@5n%fYu+9dMDfx+x4u
zkb*N<uhmDN?i6p0dhNZXt+qxTtir_DX6s^v4b-h5EL`oI++EKHv_b8u9L{Ro811Py
z%bMSGArp&Ml0qxFCRvj(#x~<clJmlN6(@X4ESMHlb8Tu#LhLPtR@<zW%PT2?8a;C-
zb#_{`E-g&Ikh{xKk8tx!G30NH?^E`H%hs>Ka4bd|mBFA*MOnjA+p6AE2B>};)(@o;
zh1&!Gj?v-uZEgF7$x<9=)xgAmc)8OQz69BKXyIHxh;CqK0ONv#n}&<-yY#-HouNKc
z!h}#q+pU2S{*tPcKO02ua&JuA6BKKxF0;KNkh|0F(^CiW8T*46#lj!^2$k%&1HwKf
z?IX9Q9n=UZFXXcDAwfWaZ-@1T%Q7iC;fNaa;41svg~^hWc}Z=bMBby(vAcdG-TJW@
zTFMi2JVwp1x)_<gAO!kyEuN9@gs6f9^k+Mz^~CCGZgc+zJFQ739V~X!v+)Uwt<9ry
zN8&X?RjB>X1+@mbJ)aLm(BW;hi_t99$jr~EdctFm_+kFDOBd2ztdkTk!V?M6Jxs?-
zou+7Z#DUW6EyXA3?U*YUt@`Y0Jh<TXTbqoBvVndq#@8-PmeOsp8c}6~Ny%fz-W3s+
zc`||b)VZtHk<i}P$w6Fj8YKtWbuAir6lEs3xC=~=dq01aO#_?m;~)xkOzlSe>=D8C
zNr@<EN3f#ov`>Qof0$=CHSbT>FSpO)iB_(K{#+B{fu-be)3V`2iAnfU)TMSk#Vx72
zY+nTq<?gK>f5MOK8zBuNC4ssgksb!^yMPn-X+%(WaQa@T^>I1>L1ai;aBP@A3O#7Y
zl>8LWxhCqJqV%($FJvN!>yg?$f308KZV6=$oMO2C|DyH2d}&Iik<=Y_F=N;8A%NO#
z7c={JMfK8(m3I5ZV8|?;oK3r!WP$dSbgj*%UrcMPa+S@97c3S08uNCCpf(NU;q`Xs
z#f)>iMmhP~+;uS%9mvAl3xl_r7ZcCTbrG_0_r;7qdddvsH#eBf5?DX|agn`8-Bx0~
zw|g%p{@pszZue;zgsl(>c>ff<F&8U=I%>0HY*?w8`8gMpV6!}l>NYp9Yoz*KE??(m
z9UlnnQl;j3(b2@24x1k|_`5l_Abu=#wb;V=li*Z)Sco)H&lDH9ctsijd-P&<0@a7v
zqaG>F9t))U2fVh$<AD^pkX^~KNaN=)@re2-N|S?994mStNbqgZmR!tu8IlkeP**;e
zM`LaX6DO^5iXLHnhNoDS;9oxqiC(knK=#1}e`6~&j=C1yN~}iDUR}Ltwf`01$Oq3*
zo3Sps#ALv|yk0G+qTGmUCiOV1F?v{9-(YJrnM+BJY3t&R$J(r)!s|c%(6bF{+T&3V
z8>b%eU$IR=w6th0SyPZ@z!hP6H7s|Bu9kqC^wF-Pur`scPWxtaKywk7!>6cKi!%)Z
zHhI(qv#yLl&YtMbYw%dFUe)_$yk8`1<CyNYK|K@SPV=qmX`sKwhD&qX?D3Y{w12Lr
zup1_8yNH5kgbUg`O2a1Th?8Y!0Czta_%6<GLcBQ%yOccBFP=`^#iQNoh4M-68dk7p
zV-^-uac55np?o*l^Y?0K#Bp(r*CS;)hHby5+LiE0W}vF6+Hmp{6t2;}Gd5(0)HPxV
z`XIVKte@bYxH@qB9nqL6KI!t|bAi*=9xo$tG)UiB0Q#|bF>&05j^lzvOm~R3FO-Po
zZVqqQiFm5&?p=u$=TtnoY6CWZI-ru>eZbCY$09V7TRqI(Bf^UIZS2!?8foXm$yPhB
z_a0%$Id)O+Sr=_K8n{L(>Pvc8R=1pzD0;*=f|fm|qk>?REAbVc#<0t`)vRKmDQ8!;
zj>CN6`qD}L=FT$tbV0ZlF#7tvY|Oh_56jw%eigO%M7LJ1PRTv`eKkr~G8WBsHNqzB
zm2La%!#ETF+uz$qdfC^5Ue+y79}7Yrox9(`Zj|OCUD556z^Sgcn~WEqYKt_a8{o}o
z%4Qsf&Y#7oJ6tA3fd4!oR}zKqUI0M;BIb^yYeoAq;PPoroqZK>y4^i3XJ4xqG%TD8
z-)Q+J7Z}eNG-}_*CkQ;`$*}JN8mBUQ>-%`+nPJKB^@DI+TUTX229Bwu6}}rkX`=c)
z&+gBGAcSbKTl&sjnSK!teQwDLmASo^7+f8v{cW#hC;V!c_gIbG#$QXky3V62Y}#vy
zpP@hUNZjeKWxOWK#pFiE8Lx>gBW3e17xLG#>pn@7+wIQRGA>=0{E^7e_FCf7E{EI9
z=mUP>$H(3Al(ZU`ZI%F03+o^4o<Mz&+rz!DWkO4`NVfMAg3QnNy_Sh0#fdXG+}i!o
z(}t(HaL$gu5!JbmtD%c<*Du=K821)Dxv65Vx4jIJvvZz?fcdc9#q9Y3thGq=*#e<S
z&=O@&^w@}Fh&`-t>xo3g*l&--AX7DmVS7{rO!wfC4|dsOG0@884fc3^oix||snpaW
zf$!-iv)>cZ*6ScIR!=Ma&|+;Y(Jy5wkUnSS(Yyz@n9BIGq3}nm3XE9TWoS#QU(4jT
zMz059D`Koo9$8YO-d7@WS*<?wGE{Z(VK+2P!bZKir`Tr-j%w7r`PGPxaN&c0Lxf%v
zV95IWM^etMizgO=OKj`aV~Y=IgXRX8vp3jA8)F#D%gSw&-mOKI&q$hHOVZ`Wd6pNH
zCNJ61qDPi&hqdWL;@&V_*k-+NYp1$pQI9K~t#s*ejhEu-5xrIs*V1miuVsR#`;J5o
zOzRIKesCBMy}`f!o(!u~D<E>DjoWaL0H+&IC$vosno-(#mOtB`%?V=X4!!1t3uWQX
zSk4IPO!O{I!-_%|N(`*wLZAbUobA?QE9H~zSy7`QUBGRR4vM5_r43ebeS>O*yLaG2
z$Vk!l>lGEFh!GE5@RPw#c2MuiaqP82@s98Z#I?g32NKG^Bon|9q2gtNIIo|ZYV#lI
z%J`Zc6}fd5kYqJJB#_(nr^r~xW0^FQ)cl2*uR`n4FvnbnozTHn7tRPNKjEpE0t6qg
z^L{$|O&&5{5QUX^R#b71BI?<4evJs<Rg{M(!Q;8;Yc<ox^4((RHGbxTP&BniAD15(
zw$bR7C^@g?cu6yd-Rg|<iLrRkPAv><wQ)fsdp_6j5;Z9g4?iz0I8TLfRV#E0rNi58
zQg9oP9=KN9HPMw#Nv=w6zwbs@UX>I(_j~H9%KeMp55&njm2R}?;-ha6tE#sT)i^ol
zex&XyS%5A{d_2{_jq^9ud(Xrrk2dVKPc+a*&-J4ADAI{e6X|<$nVJ1Fk&=v?uHE@u
zKS2CG*3K8P1bVqMd>IGBvPSz#px78lEIx~0i(K@EM)sQ$DRFbLZ}n~!`Kr6^JM~hH
zf|Q~{-v{E$TirF#e$dFUce?8!VdMJ3YXO4jPa;oM(w)Q8HQZImvUW>-c5&$D?U#7Q
zBLamqp*J>~vAvpst%zoJ+i2pzJvmbEf&cBJ83$gEM&G@jr;TQ;KNYM`ma6IT5i?lf
zg3k;+)(qg1f5&J#N^uprb2MWQIDjyoyVN^*!98KpO#R@=j@Wy5k0zl4$UUt#Ycw%`
zjL)QS8%@H9hpyinZQOsS&+b#-T(So2{%E$Utj=b~oAw^$9d564Ml<QdFWv4>YjXn-
zJ#+RvAXNGatrM5JY@SH53~Ps<h4V#<GAbRnH@Qw^L41~J^@eQWXm+H-m6CD@(dA)%
ziOsB}?R`Y>kyX7ms686wA__YMu*Wnu;t8I$$29^^CDHC@`y#a<TMajKwSZ?{U}?W%
zi-RB?*!7l(wq;dT9uLcF8*z85j7ASVG}o#$W$b=TqtyY@t2?2SNutvVHEBT)PHnSj
zf}wz;t%A%j1gDK=cAHgY^;V~+Rvbs_qchKM*BBGb!P#bO;%Aa_**cACEw@cuul|0^
zJ(@@~%~Zosr|0V27+^fhsBMbAFfw+LXR=A3KFN;7wOurS63=NbqTC|f8dyGU>KVg0
ze(id^{y!*0Q=B3%6}}Kk{=0Pr)@AFKTaOyodDi*f*ac<ee9?<Gzc>yGCp1$yTW!_r
zo?K7PhV>e2HqPnW^qRVUgrkjS;_UVRA+hL!?HJ7@Zaf_2$-h&uHha+n+ofgL-o?gE
z8ii*>P1IcDm<M<3qZY#H@OBRb+>qmf^BxhQJ2PwT4I(rl(e8_>n0VgR>cSkA3^($4
zB+3q!NXm3|lboCEkXln*hx@qDD!B_EiT3DOE+EW}+H*nQ4ZpFY8Wr5teY7}7j>WW=
z;&6@!R!h+i+6(Fj0lmdeXhoEBp>$_}QyO4M*<_|W9h(JRJTlZ-VTSF+!&FD~13bVp
zG@T37=KR{u2W2YUsy+qw`4kdz!AA9AkvIydcPW1G%RYg0P+?3Tp~9h%4!dDo_#=W8
z?Me)-!T0Lhr7bjuHa&*LDs`tKyb*0u-*vzY?3%_+05uK?efI9?<WyA)&k&!oBD+G=
z-qT!Fc=bDbUmue$KV8DF2j<<x@;r}bIA<RQ4Hhr5kF+##;7IvqIY;bcfrN<{n(T(q
zL#H!fpXjk~0Bzo<rIQ-bsn2f4Fsq=?XKIGHMLJfIE13Y14(to{gYOFq<4b+fhN{ZG
zntGhL2s@8{JUM;NzR^22V!^(hf&_aeXic+seZR-4+4uUPzTnOl_CqYpRqXRJJO`7F
z_{Uh7L@TiT?57w?0h83v!X2kO<R;$U3RdE>D*s}2^8NRVdcl|0y$bC0jIpI`ZN6-7
zdp+YpoG>X;jD>Ui>)Bm<pqFhuO^-x3;2-yTV*JiC{AfcRJLC1lec{s`cE{@(BX_qt
z^25BIi4mSggXMJB>)C<diD)|W^(-?AjEyY`4hi?Wg(@;^Qx11En-yrdCyATh?h%|G
zT&M3<w;YXJ7w&sKgn%Yg>;mO3jLCX7JKmP9s<Aoxus3B`m@BXqt846m*Tc$?_p~JQ
zL<jPekUl2pfB^DjhllCH;WgIH+{(dmVE{)@<LrHpJ^Xrh_F~p=;~s{I6?1xYS=O9<
zO&%2xw_zbNKc*@2t8Fr}?y$!-C?bvB*^k6-`p~2BQMEo1)3U`Y>@AKy>d|epT$aT6
zE6N(JJpSx$ZL&%Yy#}LCRgAWCpg$R8s`YX$D|&^-zT7h*akSM$XO(OSs|~7g2I37;
z7ax+*$Iqnt=+4uFHEIZcK`OQ)BCQdH$yo_^5At!axLJMN)~nZsRCy8xuV2__9*A!n
z0}sSDIJ|m@6l(B-rm3GitjXC!UY)}D`mIHeeFHqUX?-kf_sy_bR9=d2DeJkY_DL1o
zrS{$25^<S(U0m38x}MO$6%4&X++!2H-+p1{j>?|E&jzPv9!(#!?*Tho^<@;5C7DcM
zkg@`Ej%^blioiAxcj2PCT{x1K1&zE~kk^6q{P4P+F-!1F4^-QwbvosV#RXqWpnBrC
z?T(qmK5WnG#TKj{dt&x+*dd&9uU=6^(e~-7svE8^QsuY!341`jKIw9I@<_qWF8X=0
zzSa(j9wph>57)y}1bz<mJ*b-KCOZ;c@plpd)Nuobn4|HujIusG)HID1gAXtl`oO)r
z&#QgC5HE<t>ar7o1*S4P6<-r`>5e|9<Eg4WRd6}0$#Dn60fd*4*V7f}*T1rJuV>ez
z719!%53tp2po>8bXCPF_DA=e_UdG+>QsDB3r$}r*rcUrePyQq0(aC?CYggi{w6sU;
z?HCS6)Zad3SEB=xE`S=A`^k9e*7i=j77t#s5B{!+Sy?0<zP+b4@2VQe^u87l0a^V6
z&BUBs4+QZT!AP$5mGs3P`AF??)CjBJ$LgbJ;?;L9*$q9rZmZuuDZwk^g7Rr;P~22F
z4{yd(I+oD2YiZZ|t?+YEZB_G1`$9l_Fb;7;7h=Q}xN80@HCRd@FNs-8J&}iw{JY=8
zTi50$#nHDxBuu&}<G<5%vI-%a-*3K;>1@Xh_Xk0)N1Z|H7(EZ-qm0E|YuQT%^tGhh
z^3U=8a+2lW5~}QbPb~P0Xh`TZ0yc-KU7B=fFpL%G-7jUlJ%vHIHS+dLi3j!&SUoNN
zTuw2B=~E3o+|R>nXI#p7b`G1xpxtq4@_#G2D3+K!0j$nBy#Ugk@%`HN4(u78OSaGM
z()ZO97xD3!8NGq0-F+z*{r|J?^Xy-nbt#x?(QO0wT*_pPs^v9H?cPh7bl_X_zDt>O
zQnJ3eSaW}Tv5J7I9Grf3sR!(!%^h9cTYOW_i5|$0fj(>QrAz|Bodwd_1A(5$+s+FL
z;wM868I59Vy6-Je=S;o}I-HHRFgnjh{;*)TcWw6V{>T({UF9{#Dth!%rYM0=!|*Ys
zM^CF@+g#3z_IN-+2~{c^x9HMj$p@%yU~qs$9Nac2q7tL>@ZKR?EU4&1(&AYnTskH$
z4&vNUeo1N6Fq`Aw;S1??{_oMBuUlnl^?3M5m1ZID-s#oxNF+Bdd3jqA1fyD`)o6ao
zI0|d^*%l9%sng(H$&pxCy^v}MT}fkn;tBg*{#+Ad!uZ3vu}<LDbaMq?uOapgG&g%S
zwGEo<X5y<6b3txnAV;-C+oTz6?j2xhGzp(lw&HHfU&^9B8mM8>-qN33ml26tgeY;_
zUN)mG;1;-IY>q*D-5_>HQE0VJoWz5s+qy)Qnu#&%(R(yM=*ruyH$bh%nV|ntVh1$?
zHmI>6-lOou56!k}0M#jL?r>?RFbfn!i-)|2t=aY<Reh9Ww;jS@HD}LGJ*~v<w@V{y
zbT4TNsbmw+6?J#~jgKg0$<>g_@hk#+M88M^%2wYS?UOEdUm!H`Y7cp&_6O>v1W@C4
zAaKAE;CH-(mnO3tqciB&l7Y;9*~sZCI&cF(sp?se4vP*1iLgqJs7)#R0Q3Hwe(S`3
z5f<;G+BXb<Q`P$}+A&SYdX}j>A0LlVaFgz`7xWDQpQ!cS+Sq3&gh5BqFFmIOgLRDM
zGH0iQIEXo^Dy6f<voR@E1f}@e9*Jr1Vr8C-Kl2F3=LNZUz>C{n6l8A0>FzT+MJyNk
zye1>3bgEW#NmB%qB*%)4>FZ?_O-#4`SYZqP&aUVHgMaP7JnZ4nZ%;wsiVbtWsvlYi
zgl8vA3c>uu8KlK)F_Q$Fc*N?v>M$9}Jx_Zq1B8s$HhW*MIh|7y-s_qtZshEUX87cX
zYE{@uge|lAk(QckM%q2Y*~c+Spww+Q^ugLv^0-f?J{Z3GZ}0p&+T>58Q6ko5^6I>9
zH#L-qy7^2SV<~LvbAj+M8g3XP_C<W)S&CsbWUz^`HR5R1XJ5sG3rvIdwa%(^#s4Od
z<u*=;yLye%ncs;5UAbYvdQj2#YLoBwv+f5y*D`(r+%@2K_cX;g{8KEU<&>fMS&b|3
zHstm%Tn1g4@Jry*=j+zq$ixEJ7v*>`6yEkmV!s?Z94fcJkuk@9s$CJdI_-^&9}_X#
zZ`1XNUE$9*<BjY#!)EmhSE7^M@kYkt@K5R|;>GTKBZ(9qSeH_M-xdAfT#wU(8@gr+
zG0t%}K$5&&ebtsVl-Vqy*5o!0Zv1adCUK|8qPsT+;W_$;?7laW(2%tE-1h~!frLw`
ze8lY1kdv?LoHvqK)LhtXa|7afUt$layGHEO?1Om$6>W-d=K1lo)N^Q309B~gyV7G}
z3I3p0e0(^7lKG%$kGzpFayPa+J9P=>QGHDns2Y1LdZw;aKM}jo=JCMrp$!XcQ2_98
zxW7up_=M<P#pR^0&}NHEv|-oTl0cOWS>%>gu5n#|jx~P4D%E?l2lZ6NOwl^u?YZ!(
z-^gSxqRoovHT+Onp+@aE{SR7=XiViXavg9?)aobf^5Lakr>@<V5e`!?FfD|};5x5w
zeAy7;*Kcd|4NJ)rudNfh<sJ*XULaUANN^j}9z_H~xNTGeOibXlDZWoz5lN@%jSxr=
z{$ol`<|3EnMX-h(>0$$b$1$cepiMn^o+`w*&7wZV9q}Tg1iAoV*R`&gK7t>uM@<LY
z3C0G!UJZxv?mp{}d2_vt4aRqy^BuNTgh|;~JRQ}r(4d-#L)w;Y0rw)>9^Y1D8``0U
z4QOCkPCNBOPdMG|5?1&mk^6<Ky|ALc3${C;J^1wR>{($_OY~;3!1CP_@KQ}eL|BA^
zXg+bY%k~Knb|j<%Ju+KF&uMPR&c$RL6lT}|Z@*yJA+>i6j~&+bfp@so2-`&7jtI)y
zlvU-q7)}$OEbSgs+F?hvhF#W$_hA#YV?i$#dPScd*Td7B>;?7O>;5MvG-S?9SJ#Z#
zsW+0?wx--p>m{NEe$8;uvw;uYxI^%Wji||Hyo_Bkv2&%lqNV|1bQ-m9Ycz=$wS1}p
zHyYGjfxhHY{Q30L57}60p)9X!wDEw3cM6$nu80~rZqkIaPd&qoy6kE^lg1zz*Q5r^
z*#mdiBZa&A?Ojohl(&9Q@KCq+)CON8T;+q_$w9%-vp&171r<3In(f0NSmSd0NFO)4
z9r9z1)f4>Q&`Z~P*e7pf(mACqa1cd>7U5%`#^{$fG&a~xk&k;dwa@Icpwa@4%g?9K
zcbzLqzcI#r5fq5^3ZMB>D3qf){z~9Fx!AKMzmC2d3dI)tMlZ4dd$rAPg)piO^1JIa
z>GE4W+KnOmKEOvxI{QH%;C0h&KkD)N^O#zG(qyHa|Ms)`qu`#q0NzrA+8oXt_KTkD
z+Hfv38_TZlY|f!hyKO8n_x4V=jNLw#czJu@QgX)+*|f2YOIPC`=xbwoG+;~kMZ#u`
zW!&9a73;Rd-N(|AneP82NkAYSu{)z5zN{qgdo<tysh}n4LTsiQJpGj2J(fusO@tXj
zw^?J^HUFun7uY>xiLsZ}E<*vJ0o5ka|Gs#K%*D-LU|h`%{<GsHktI2s6Mu7`ZnwDt
z;(6ytT%+esk|A{4JORpMT|t7FKbG-n2A+ge=hF+uGT|xRXT9RxLQ$-|n+5xDjH8*f
zw&eDFWGtN$fBd68s%B-b5Pxhe<EaoaMWlT2j6E*sWiG|IG`~ncUHw<KM_0#yJt0U~
zLAdq{b6mAuT3@0$OQ%nS5R6-Sj93<rN@1|24S_qN-mAuvBsB5jfK><eYS>}6A_hgx
z3=hDqQIDz06pUVTheMpz3FR788;I^Xs}ER+O8C)>@+5mAKxK`f`OHAUby_<_B@hzq
z(#Co<>dB!thRJK?rjR*!z($QHIg+|d2&`T|GQ{wV=(8pPt0#}Yb9(_>LmIsTl`LCC
z<1HT4(-up&k(GeHKSqnDz%eUoIK5m0VfBLLmv}@&>k&N~JQT*4QEx1Zo+j(pOHaY*
zq1ra6pYYV6N5-~QBw-sSqZ|SALUFxs`)$GltQgbwz;c+fZrsIp=q+iaNEEg+281*R
zE83-(wJ6<l_KfILy5a3sR}HAs>{+#63DX|7JppXu3?;E_uRi9qaJPwlQ$K7)zA4!L
zSYucJwcQQ`lCGab#yA+OgbdL|JERe(yH2w!gAR{PW-0@}0mE361Fr{k9toIKX|v}7
z3fKKsI~q`E?C!U&U3lG&MN`7Z?Rfn8yWc-$FQ`A9^}DFn%VBvUAQ7w7Z>O}x*L4={
zw5AB-D(T|zW<0B5)^Rgrg^p+;WQBvy>8rs)TYHKdJ|7q$q}WCEkjL^d8hsNFh^Akl
zYq|++#ImoZc>CD6ri1WrrtgY|!o|7Ull&F!Z6U*yry7G-qvxgd1k4HC3TSps;09As
zr+1@ap>x3A(>i$iq2K(Cy{|{u>X=;@nvJ1;{7{|NP-viob^lSU6jBA?UiPu}rYEt8
zgnD*EG{ULd)t`C$BoJ6X)M}pwRg(VYriO_ixdmklr}}4sJ91aVM1CHhZb94rg@)DB
zJ_xzL{9jMWFcN*G{;@KX(-xp#Yg{=>t8?%ECa@q6YrlOfu%0QZ)4tPl19#d^`#wfo
zgN=pkHu0g~O6|uW`j!Go)^WBEL<iWaKZ`8A9zLD8uH6z?KM`{FiyF0(u7T8RK6<;H
z@f5Fc2?z1h=eEm<lcE}e`(MtuS1EN062Rq<4&cFgn|?V=<i!q~5x>L0>sHzwmlJO;
z!-tZ~!<~BdhllL0%OM6=+06L31_$Z8qdgyYP5h_L3JkjQNK*G)&IAtZ_sHhB*atM$
z8e#MIUCvYuE8Oq&{($qrUcd6#?0D+$?j-2O=3LGM2rjrDa5z^#_^lA3`T=24gbwor
z8qH;|%@-mqtTPWgTA*LLxiz#xj)el?eD>jZ@+7sqdn86n{wsS_kWnC`$`>3Y(LQ&A
zdOU`>9Cx<g+M>(ZZPQg4Y)^Q|6Ph@rQ|=7uQocC8avMm4EeRU3(mSlYL{CqwUODxp
zi{@-Yt~FJGY06Ch->q8HMR|@Dmosr1HR5>cDMGYsE@$G-^3|T&p9I`YPHjNy>zKHV
z`)r-)QBkwp>NV^S4s&=n#zLrAWozQE#!6dvIdkRk+M$tQ$hq=*O%*Ode&^i~Ln13!
zi)|GBwiKF@rfAdUOrq`E;^b-)9>d(`^TNX`3tLNo;)uvwTQp=FxW;+2dR^{{D(v+7
zp<zp^8PGM=HGI*#>Jji8FaE!~tXGeP!oU{m*IX)Llnn|f$G-c`Z593TJQ)`1Jvh{E
z3tXuVK!|%RD8yLX5kI@K#dfO4m5Eggy4OGE7<oqToBPo7@74&o^^*FMusF?u>zfeY
zOL?C51nw*u*K>}#LyGMULL(nHAn22OHW0z9XRqxK{GNVln;p=IDsjcrnX;Zbx@-Vy
zyhPJ7OeTk;X+=$0{Zcy;4@fi^@}3joHGT;^sz*<IYsWOsN-qn8e*SWH5n+Rjq$GJe
zv825a7`n`EC$vo51)eHwf>9ST<LPKj-di%>#NWMz7Gl{;zoQGZbMZG0)ctlo#(^%<
zE(*OKUlyYR-H;>3wB0VnXJox4qfN{)N?<JAHXgg-slT-=`iLQTveCDL9y!thp>$Ud
zl#WH^9(@lWNL-6%9^}+)?`oe$0q)QQzo$vS56RuXNHZsD5atWO>wy(xbibqz)mzg5
z%ASu(8)bMn2XpyYFTIMN-3U}_*tnkrDp*NSaek^E(5Z5fy%{6@!+iTJz$H&b0@q{Q
z%ZS!N0_W`fA{s9KoqZ_~e#fG8OOVlBBkgNp!Kv43-vmrQ9`P^xHlE-xJ$T-J7hp(U
z=+&Yyx@VI})utaZB0EujoH{PiE&<DB$Dg!koVDU&YCnsdC^xX>z&>titHXdX1~O^(
zOAyAL8f`o~qP(;YW!1(LCnZ?JZXZv~)1%@U`?T?lN5k6jBS|LF>En~G#_q+)VKV{}
z24+vbd%^Ay<eGNZ4BaUJNe}3!_YS)&y6fTQGIKoRi#=ZXtKaS(&v+`qdHBo)I#902
zd2x@R;|_s@aId~bEMcT}UqGJw^l$9`@gyoF4Y17?)CyO?&502ZQJb^5@i*5WQqIIT
z%X}}`yzz|r=L?8nd^SIRNkJ)L=~@uzc|jn!^uqWYCd%gzt4kIH;pyX<Fw;5ld~dTy
z?L4ZVD$qKlu-nJRGj{79-n~5@AOE<TIlF|6qqa!nMjB{`XFs9u+(zZDxg?p>cX;`S
zd2xww;&g7aa*=b3%V}1r*LdKP1YW%$g42;zmnPoJDy$H7Nb=(DURI;2!pYxi;|aaF
zQ>IR!oP{#Y^;7G@Gsk;^CrY<O|L!UQhqE;SVvuVnA>`}g<x+0`<O{R)<4KZR&)r(1
zAZt70-l$P<9m(6K@yU`a0u0Vw#gaa=DL()8cFV_-SxElh7EzIaWxRX6GIN`#m`txe
z(9NZ3P9=qk8W}S~e3EsEFa(h>^VCNzVNNu8JKq1j<C&Pz2tIx6Do(i${O9x`)t6G-
z=b(P!_LVfpTh$u^5#AX#tie*frq8y;V6OgO?Y3QrfAFtPUG9=Q^i$GSyQ}oh7z;N*
z5ZxtU@Hlkp0c_dQTH9Tk`HACL+0}|b$fE7hw3i_<?2U$-N$zU<)E(Z)?()iITKH4N
z%s1=-wL!&~%9`0hO_*y02^K*k)@O&*kl13bClpUUJf2C1s4D-(*AUyob$dqw30NJO
z>H@vwxA?jpjcMQ7V#lHl3Zh{<t`*coK0hq#!qPL~u!c^k0k?fbu~YG}=SH&A0dla(
z&c?4uZyh!gzpkycbMY&NL=@+m&=g<<gS;3^4ljYJyuG6$Ft(j8yQIflv9;UAMDGUb
zA&qOvcCzsawPG7zDXr~A_O^(Y%dTcu^-~D%ICD=1#5J+ct_8k*9asNk?*<__@cP47
z<vk4m8zcTx*vn&)<Vae7U5H|)8MF`86KCth87@@zQ3jjS<zs<D^TOqI$ZiCvVYdD!
zL2zHUu$58!RD&*3RrF>&HFc-^?Poel(lO$mZvSMTM^EK`fPSIrWI?%t`^)HP3276*
z(mPKaLr~z?@gd1hDdnPG*CQx>t3^OS4Y$AVV(U;Ahx6?F7&`)MVv#?@I|@W`Ui_#&
z8v4m5M(Vhjpz56c96gj(G%lwc-z^akZx&SL>X3p$2z&Ns#*7*0K}2A$f_vMWiOq!}
z>Gn4>Zc1)oU#QdG%+%Dw$SBj_%%sw=j3l?PahvgGV!4>LlN@=+n@JcLnt0Xjj5l!K
za<7kY*P9s+<Mg?Joc(8;sUG2h{e+R;tscv{)Xvgt*H6GU?s+p~s%yAWyC{8c41hkc
zY}}^~DdJ1sIE^?V{ATjso2?HhBsoVPVC06G>#@2Y#n%Js9%BcWD4VDEmAN6J+ya#Q
zE^(t)W4}$|9JWxna=$<ngr~om$xSWTmC={KIq8xRMliU!Ku=s`6FwGA>pA({&&wVU
zx<EE}SX>m1-I2ik=blj496%msxH!IEikhhrGnFk7@MZ0sx#gOOl2kQrl>*T-P(aCS
zRc}s~_+F`+QF~8MyH&^JZbq0|p~3hymDsj;<-sMjL6?a$o`9w<F!zP+y7BmmJxj<5
zCWFSQ`EFcmYf1!lb&jo@!t!w1$2h=GO{i-`UBx+O8#Fjl*bmx9jf99Bw{C9=#K;tH
zO;aGf*uf}=eLfJtGtomhTcYhC4lZqiQ{#1ndTsO6$L<wMNfRrI2KA|=P?x5>jPi>(
z09lVdf<<yw;c)8JM|ifdE&5{>VBp0B9t3b@Ab06ElZ>*Oi*&$s?!-Ccjyl@}q>k`}
zZrlE5CKV*jIwDoru66`K&%=TqJU;R&PD$ahOG6waLKM`Vc{7tA-Ms~lz;C<tTgg#{
z-=5VtSE1`f@oIZCk5$|^_X;~4MbYo>iz#Ywe_3XBzh><E^Efmgh;LngMD*9eH^r9!
zC;mEq|3$sY?e|bL^LQa<<FGn>=ij&35p_6>FAMi=k-Y&Y+ll{R864H=)!o~gBj0rp
z=vcs+7^8PC-l7Nz^pN-Lg{dEs0?SV52LwxJ;HRQ1Vrsic?VEg4PfsmVSKX7B<b!{<
zvqH&vY4Tql(H<*s$;5dk9scJ;9W;O3%VP$k62nCeBG>F14wnd-kI`tuAvn}+mqZ!Q
z?lh*U3im$l=t3mHK$Ng6B3~WHFxApENK(dnish@qiP+~Zn~e6XL<-`v53UI$Dv{<6
zR~5f25N_xqsPR1!B{G(z75IK29u6JXqZ2av;7ZJ`T@V^LQJK__H26r-JMp|b%Gt-U
zXX{os*bSlC>cZP6dhAG7c#$H-S+NjrPK~?ZcDB#7*8PG`+6j(?&trQw<1J`k#NWeL
z|Ab1tv?o0c+BrX;>??g;?>Xyw{XqFzzrn$p3*6F(e2;zI(rMpD{~p(pmiTwV5P3aC
z*L**PA=Zz){SeQ+EEuj_-0?NU`W%kSKLvvCJWmpnpEW5iw<JgPTUr{8xVrtKZ?W5>
zP`{FK|E1_IT8Hd5{m{dFU?sl&N_M%f_JXa`u4L?tjm@~GUrC(53A@+~eab=P@ili`
z$ryK&CnJKx-FYQrFr+_3Z0<_d*1OaW^LGlont3Iif`bEgcf1Q~vGPW3)|Dimpdj-I
zfqR5rt)Hv+M)S;dAML_@Y5+Bgc7HtP@U<W&#AaWSc=IZ!>_0h!K}4COkGwjf&DEge
zQvbk}$<StyUSXa_?XjS&i1}A^mW9IF&2nB&Y(W6!7*3&%3)P2XYY$(^1SKvIIG}O!
zdE`ox(*_WOa`tHS`G3rL=r{IQ;8R&>$)Tu0{eERKGaI_8oo+)4<^Ww3U^)J&NEWBx
z6B-S*f|=pP8Ufy9Uf^I!%qf;nD;JVAZu+lOQ#1zsP^PLh85D`;#HbD&KiCTyR>Yfu
zP9}87YW|AOP9^paZbnwCnXf1A49VB(gdEpD3Ut-$nN-(!EH>(k29zBbF>H-~foh4>
ztP8p!eW0sZFO(@!f<4AI1Y(?<o@8R9o_fhPwn@VUA4<eT=Zq~}Yx9Cd###>&Xo&%H
zWhcqJAQe!`Z5G~O&CRBGCF8P@F2RMAF8#)_jz5R9$)14N*FhXvZ$Kp9DXEgIUl2Jd
z@tYWoJ<?ciTjLGO%!B(zY*?5Ka`V|1?bfWe?HV0v=elf1bhX*tzjnqaNkeeicEuA8
z@?I+anHaTabO<Nq-NJjIx6gBtappZcMH<gz(rJ5yE|M_Qv{wx}+qe1|yH7u)Yt(DD
zU+A@XSki$&uZ;q5b}$g?>rON}6zg*srgT`l2Glu`(O4XbwiRAeg}c0ZL0a=0($T<6
zPRYmApn>=|7YBrR71AmZ@d1zZ(lBw-PDI~r6UXh8$k6TSK9kDfw8k)T7CGT;pqEl=
zxW$Rv%t(MsrIN@_r_S;@PsepW7PIev;JL=)Vs!ZXzn^2H0$%RLqb}*&CsQ$=F>Qnj
z7fiW8jO&;6#KSTgyAr=tbB7p4kYT~S9q=26dhDtO1oN0UgHoc~CUs($@?xTQ%-c0}
z-9~-hyuGWAa$Q?f4UqR@4|;79oaag_Er<5@x@Ih#VYu;qC}^oRlYJB<D(2d{(I9>t
z2$nTe+YQat^}nP${wJE73dEjIgFac7{hNV88>ypu?K9D?(S_5`17WW39%bO-%ol+~
zT0~wG`b#0z#9BdduM-S0nMx(xOg*RW*P<D&yCoOkZ%TWZyy5n(P;M%8xL)YHc-ib}
z7r&1_kPY#aMZ5b$V2n|PGR{BhOWYg0>ex@El|U`dpEd6K+_2pes`&4*h<*`?(+!#%
zlG$6?rM_w6vfcJp#?z5_$!haf#;40j-s`y?r@fVNc^m{>3!rC$MDbIZ@z$g>z;>95
z-4X9Rw<#wsJpL{5zXR@y#YBZa>0NIncIF$Y+GeU%lJ;Lq|986I9pGG*ojeO?y_F75
zws+<cv%~IrD+wJNhzq^<t-n%mz=JpdDa8Jkq#6oNI@i~LiTTdiY(18B;n6ThqjN%I
zXAfezYY>d=ff#FYyq*`xFofQTlkR=M_~*yhSX|vLy5AP4bJx-0H?vT^p^>2ooTj3c
zyW%5nNdVbTcOI?gm;Fae1jbxtkLiO7@`F7tWLJ^wlWMMlj=s0Up3q|rDt%iV-!@Qq
zg5AF)X5jxL>uq4Htg^h{6=S@Hc!@E_c#Sc<ybKKt%@|{hVQ7XbilV@(qNb`qhR`&p
z>eTmV*QpPx8Vy6!jF)B{UdNZA8OK2`<Dl0zG$KtiG-Gp$5)rXLL}ZL1#1LbQ*BC>L
zG3Nf(Ug!Cj_q{}udd{<-z1Lp*V|}kx(b7$#neupmj3a<m{3+u;+)Rp+=hs#_1(|a&
zUIxOQi>fre?v}P3Q5!3T6AV-6%ei~3`cE03M<ydgrncmwQ>%SO*>_^+3Rb6floKIe
zhX;MMSiROXH`qf(Fa+@yYYcEnPtLW!Mzc&o%WJ1#%8iV2RMrKOLiB|FST7hc`arGS
z^t(aOU^ViSYa9QR$<y$vur0`0q4*cdNBTTvVg65vJFTm*Rt=np?=fo^U`X@io!7V{
zxGv%<!MNSg8~Qjd9@xU(=xxDItX~kd;(F)yq=DGz4XbNxP`jWKHfUH+)ypbuQ{ct}
z+WKrPdP#~+S9iQ_o3*$VsIAaaVvCQC2T^ryiLFNB*IwHie`DBfwQc`PKgVo)yv0L?
zs;}ykJ&T(Kxs-Nj+59oMJlm;uZpFb37EU0#xxVGOxwcCPB&IvnffEi0^GUH(&qi~O
zTAS?-Ts>Z8d(@6r>E78JC`sxALgCdhG#n}0o}b#z&L_EE5Oav_3q)Ynw*6Y>RnT)(
z{4p>ID_{a*UD-IO47=@MAUtIQwU^@|Stsn3fXlOQReACcvfBmBts{0wFgYGLkB5aE
z6c%JlWXBwdK~qi7qgv$Uh`h(NwaQsVJFahsdc7_gyZl7VrM|4jPDVE=5dDpS4>pXO
z{nVc%Dx?@Y(2)5?jK?=~fBfS)Hle{_Ux}2DSO0Cnf(3R~d%6bo2cET1@N+-dD}GrV
zvhzY_Jt0|rHmT<_m{+RHM31wcc;e6OlHTFm4(oV121h4AeVP`#60a%w6Y|T|`1JHv
zlo%iBQ*2pqS0DdL;lIR>Kz<Os93IFE5{F%jPk%CV(O=nhO|hpBKaNkdI0Y}<Ll~T}
zzm#yZPlfIZ9EpdDFX1z<2Hn`lKNlo~=|Y6RP=9QE&v^3X4{#I~im&v^3K$B=zn&ZM
z{O+=w@yr=UHU+2O=sgaZxPN>b0Hgp;a@%P>h(Yc4z4p7?X*~oHcc+5csf-b*#;xF%
zQ;G89o$8V9F8SViDkGf@ePb?=+;%GC6*yVD?e<fN@^ic8Y}%=e_ToW6jumRk4B8#h
zpo|o7Ha!Nz6wCN_r@F?c%wrhtil#aD0>3+&I?aX1+;b}3u(j28Z~TsHB=#h`FM#Q9
z$8#K%-Y)<=M}sSnIRy`#N(=*x+@VCt{+3zIh+e8lSi!A2^HfIc$PpAVkh28a5G8I~
zajlClp!D^9%{J#$M$bre=h3nqeksfg_=obn9X4NRs_)~#ESMT1vTGi(2i2k{M^gQV
z;*t22L}MN<_2qf49yyh`ueyOgkH<p1Kb6=WTnucX;ADQkXJ(6{EpV2TL@;lQPbDS`
zafD;JL{QY0a*tbij8oAE{wvfOiMxhIkmlp1J5*a${0$W&^G*DXCuf&c>!TcaV1v|Z
zUX%Zmd`oq)r2wdcjpyjDSA%LdH8uvsu3}6|gEjhqP%{s1v9+;+bdhgqCf%Rki|h1T
zog3s5Ke0aMiedwU?1n&6S6{zvjI~P!Qj*+hc=C*qcK%exaAi@<Zj`Xrm|`25VXQqM
zj0A<8yH4S_3l9>!>Pq9D@EonZ+9(ZhXZ^xXN})>Hd3}r60Wt;y8qQ)8>W8B%?rE-!
zO)-VX$v`|-I+%&;b@jh(4ghgQ7_cYfZ|vOUteqN>oPlujTZK=YHG4Ypc)tsVLJxI_
zZM&Wud4!hhc{t||I6H)E<Oj;%xS!<C7;-d+iR|fk&yA5vGA4#~S4;$NPqxG}G0Eaf
z{e2!y@oc<fJ!_ZRZsDu8vfB1Uw@^c})@q^Lo5P;d=0(NuQZpJXdp=sYRmG|BK7GK6
zBB`CGL6!r~-~o-a4lU`60fdVceRl9v#_9PpaW?K82_|0Fg_78(L^kXdVVKFVi1w<+
z1+acQ6w6k>>Tx?P5IWkuzayFhh6db2`S@sDZJTG=F?9n3BX(R^DzCQ_TEtb!PZbTm
zT657B><#sdIEo>GgLO)Sr{H`W58V0++Jq+N4ks}=J0k=k5ZKvxB0q$u>o})(-FUQ|
z*XMsd1Yeq>vu!e(l#}=pLHMGuRG;%J%`ORGoZ(4StM}u7kDJV(UC}PXWLYX=ay3A9
z8=ifnJ2!$3A^zM0AFG>HM5C}VuT9Mw1^l307v^fYTU`_NNgx($7|m}4RCQ1wzrsI_
zN4G1x!_a4%XCE2mi1hXq`#io!yaMlE2=4<J>~K^<qSr+6?!~Mh0HxEh9XaZ20SM*5
zo^Vr`Ez1wvH_>?td$4cgRYy-ULVOobo=F;!`+H5?FF@ojD`tdwcmOruEyZ;4w{kK_
zW>?H4`lOoTuD`9AU3*A7T%Sa!yP@y)VkYR5UYk{#R?OIgnT!6y?kHvio~wp{^ps$K
zx;}yBf?D0F=Um>m+Fda=w*un0GnaO^I{nE{7TZ0=M13jl&M^*-?%rY|aLx2~Uor6)
zOWhOUe$B9Fh(Ln}rUt^gG-fk2NjT?}5^83&NQGg2v&>QpH)DE<%h~FqfWKOo&50qL
zooNFd^8$%3hvsa)+E%xCZiWTL#8_~RHM{hmY*`P+)KF%->hYoY94#dtBM)oVBP`D&
z>I@Qc`k(F5Xu8SsjN3v@v<4NVEs8JUDm~0;aSR*s9Nes@gIZD=2zwENe01$TPN_;^
zMZE0Ec`l;LV&dmqYZvz4D&eH?bNoHn8Y^Q#n3wwl<(wi^$0`twH)yq*M?I@sCnRvA
zaTB@MV12YkO3zuNTK6#PHL<X((R-~G0@7Zxb;ZQujg{H@;zY^oU<K#iv*fxOX@lTQ
z>ams)+Ze-89gqMw^%kKG`M@9!Q;8>O4HPs&&2h(5TN0_T$2))U%=K^GdPav;=(k=y
zV?AY4^y?J3J<xHCwRAx9e7vQ@1_R5W5FL;c;bB2v#udbUXPboOIuxH?0@B9RN1eOM
z$MjaOsIIUlHDuDd{AXg#w?qS~(=^%EVj^J3WB;FjwoN;!9``BRuBo6-z<cwlKzb9Y
z=>tPR-0-;V3_wVXi{ptst;vv0V~ok{Ql~g0wc0bWsKw)CuhR^M5tw(E7O1b68Z!E(
z(=!V0Ri9lj>I6YQr+qXsFvfxKjD61sbc+_-K7EY@*k=2urZMChtPW_{DtN9J1J_AE
zco>cD5j&{QNBY{k?PWct%kdTUwiGdnx^H?$)+Gbbq3ACiILy4LXbuZ{<a)nGjsyVs
z2YEEWQ7tW6B==b&xJ}bTmEyX1(0Dw?;Y#EP!&aV9uYi>wYPOStv)ogEyrFJOJW8*f
zid91m#r7GGb(k}ErcK1}u;Z|&XEYx#MownCI9~`1ue;GL5vG5zb1_))M0>JXpV!#r
zM(DLk^+EO$Xy8TdXTKZ$ZoQ-y7`|Jm1g8#Gpog+6fguXFV?(m5LI(~OE%4hc;zt3#
zTd+S?@2#G!firn6KKJYkTsqfdEY~2x;!;XR>+kIT4UG&zZnjU0nf^IpdMd^5XEEU1
zxf|_sbwCu2eG&T>XQ5vEQlKF1kjoNRdv%`_UtjAn6^gm3_Dge=$AT;UCWaaRJ1W9y
zg>hg3kM^B@NDj7#?EB~~yW(s-Bfk~3Wp<06$hBv;j!z`VF%ojwZF*DBWqA8|#tERv
zY_e(5u-pyqcZ_EgobBxSG=1YuAJ1g`>d0et=Xl~Ba=pcWr}Eo)#uBs+^>#Je-D>TQ
zifs3L#?#IHm$U8O0E7J1Ev({w0%{r6TJG0p&=?#{?1Aw_<ZFpFnW2_mA8lszSM#{d
z3dpG|ywhe!8<f}{kvC^*9<IqDr~%vmyqGtPS=y2FH6rBPUGElXCJi0k9P9_9`@E;I
zen=gnwN9!fPUyqq8QYY~eiA71NC3Ucz42l+j7ctk*g}ow8YVO&i{j&y2>>3NZMrzd
zgI*PENlbFd5-X314Rt5Esv^E3OLWTf%~zG<8Joeb@YKKnRy8%4n>ml!%9t7oE)RbH
zgH>y)%Sa~HWVPB7%|y#s-47<_nZuyjvE&|E*k+9}Wb*FCyN>&5i>=Wo?tJAZer-%A
zSHLK@E+#>-D);+YA6TkH4Yncv9EF+Q7(;O1GlQnmtTF1q;pF30N1Hn`wT@@%7&Nl(
z6?4Se1AM46+Wx$C#>Wt=>Q&3FJK#WPU$EZzyZ91Ew?8_Cf53-pK;V}%zDxLn8m45u
z>$YK`m0Qgv-%T251uE7tec_oUJYSP-*5Y95NsgH4(&M9=$cDOQJj)}Q6yquOkiRf7
zvP@gYGggNf4Bw+3c46D1gPdzSaiJGb!zf~4HJ*w&mV0g49U5QBVQgoteGGz1=xKos
z&kHY#6ZJ0jnbP9_(w@-|)n4|CqOSU(ikur{(%-FKsE0+BaTausHdUjiSKg~VH<lYg
z_4Hi8>Do?vUU1e!5<75V#_}G&rHq2^8}7Q@{qfEC-(&DR5Z!kqXTO5I7yuAR<2g~1
zlI&o#fcW*gpViB;*H%#E@s(Kn?o>SZRRM<18;jeaz;8O{-F8?z19q<ml3_>G9@Q<{
zfuq93;~qCMV8=A^wif(XkL&$F-zX&4PQ)yTo$j@hG1!pjv&G3yO<i$_czQ}*O&;VN
zkH@vp-fk0W3CBS7A3LK3XvH<as}b1Q(y?K|`s|$M!7_8F_1Ss#-RQonlj9R5uOPSS
zakAWJ&Ao#T*+sRibfw`XA;3AlKD(^TamAD)!WA`gQHmV&8sAR_i5=`C^}5W<SA47)
z)ga^)e{0vqGZ~b7%Q-9etN91J9@ukn+dk1Cc<Q72yb)dIivM1GowM<&x`as}!oWV$
z56A)3Dzwi96;QhR)V>H5cXrvAf)6_->FA<qx^moD@JIW4s=*MK%gsOoWp*J!9Rv=#
zBJ7)3U+$vExbD7<E%?(}^X<C;y$uar(qDXEIu3cjw9|>e=H0W)ZaJOZcf>p56iTPJ
z>Zi%QuIj(=w|3j<j7SqB??U43YC{?u{IWR!)AZ9MiFMM=d51pp`4fUL{d7jt*nwQQ
zB%?o_5twpk5Xl!dXm_2?Sctls6%}^(>5Oeargt^3$M)QFdcqkHcmSf@)dyFWlRU1m
z`{I3fC#3iO(}|H`dhXu)fV!jf9(GlHT>ET>AYI-uSVcwhnNu*}M`^P(xi*4ZH{pzL
zvrlK&C$59I3WiBz&gsNhw806@(_~#U=bkF_PbX$&Wx*Dl&UAjhJS8y`d%=UzqPET+
z5&#{j)a+q(=J_u?9pNMKi}Rsg<j_^~r!!_LSdNA2!u9jEDA2%W@gz%&^%75D9Q1{N
z_Kw8BlnYHGX#PivujMI|S`j1E;7*%Xq4Km2ksI(h&F;@{Re^+`{NyLLQivnE3ekcy
zP%SXK3CAK?y{1+PaYyd6x>&q?4uPm%Z-X}?XI=EZqNmN)2uODu9)fCOv0tmHqzKhf
zTNeXH<x=>!^@52#=brV%+-VE3Q_<}?Q?*gQ;Su9%0M@4|AY&fz4e_3Ow(hj!g@aF8
ztFVfa2WemUfgrl`NeIuKYK0kP)Vc+5%CLqSUq6I7)%rD0;zoft)x#K2FM!i*g8>@h
zc(~&UX-&im`{~>ihrSi|j~ejNe=;8-P~~m2x?1iW(UU?DYPis%Z_(ztzV#K`s;Bzm
zQQIbPJ)<xEZx2vBf4n^vO`X>#RAPtL9_mK;*Ule|MA=+VO7V1jJPNC3yJD%S9c0gF
z5RYyJ{m%;K9v0DdtLK${Za3K@6nKpems9p?tO_LUR6^@HeMIDS>!3X!;|-lYY5UZD
z^$=!?{TjY7gh76j9f*^<{x9ss0JzF?IUdv)ei*Qo?6jA4Pc--Sz)ply@hj?nu!HdY
zSGB;C|LQ7n6qf&Bho*k?LLVnD?BQ%3KAlOFFmZpRTB4h8J+Asl095nWb~I*Hst)o;
zI~Hrt!5`dY$EPOJ4a6j{708|va>OOWlWN0>bfwxGF-7jURy!rcqJB?iq;VaT!S>=y
z#r+;yJ`t_l90QcUzVItMb2_tz1F$<6g~|UU<k!jFS#?{1E0O!-oQuWqIKK0O(G&LJ
z*+gad0BJ>SL!({PGE$xxA?{Kjw-PI?r{=sI^Y`$MZo3lSEq7n`tC}aJ1l&dDBdtoV
zt*yg84w%B;>)eN3i>^*@ExtBl*L4aSplzQ7xVSOf4T0+^`tX;<WFNy`_&*WOK8u#C
z>MQK?5@?<w!@dwWq@v-*u`jjOu03Ft|1(`8hwQ8PjQb5|z^`NB8#?S}X_<!6;C&NK
zh%n@Qe=7u)r9vv-#q+Y>e1D(c{@+Jq<S#6AHW7Ask4N?37hLovGV+fYPf;|xbs|#@
zRN?Y>n_heIC|JYWClZ|-!VP~Sk;9G-O3h7VRQ>dC;J>C%l*D{bd3t9I8CxL<CwIk%
zgfQ9NF<4oB?TXsvcF#mc<OpR(e7{$Ny9Iz4)%zwgj)JS4Uz(Z!{c6;X4>)?12h<4a
z3q#}V|BL{FV^V{t>(|fBn8(kF%Ts$#t>QZ=Z?h*d;^)R$P<1>>!*c@iHX^v@O=Qd!
z+ucJn{>kPGl2v5U0aptqCY+HWi*36-7=QO;L9>UVA@F0{adPyqI;(Pq!v(-cv<ySd
zWA<o#jawC_c@0C%EnWn+D4?obvDOw(b;RCG{?JR}Cq|;*ltI$Uqi-%47r-ktQm(1l
zDg!SS16DPW@n%loTv|u4Tp6!_vCyiuSg>BmCbconn9IN}HL>({0d+2q6LGzuuFDZ|
z3#pGilVYO~1J`I3psuByIcsAdkxLf0^mWl01ROWJiHtw<#vHJO#SNNBExh1H?Md9T
z2dqipwxMLmSw8y44nbsGYYCPYddaLrMfiXvc__L>eTWOYQ#**LGi(>A^92OyWfGg_
zCh1-cOIYWS^=m;&N{`>$fG|-{#uKisLH*Fy<~c8HSRXaw?BxZBY*WmhWK^W9(^OC%
z0LbRjJ5N3EWK4?^YBby;Sjs$`?^fYwDTiyD2FVSzV*7)K-LCeyJ^$LCQVXi7v3)@N
z4h`LqWQCpaMM~!jiF`VC$7qM`QbSzK$0$dmXQqQ}@~jS6Epj)}xzu|>d$n5-q^-WE
zv{AXNSc1KRzYM+KbMaJ&$hqg$ADh+7(oj(Xh5A0NT7O@@*>fE2*FN(?l@L8bRuAa6
z$MKx(LXFU7FA5Ua-jjAvxbV{nwWmOxrc~$QnD)vP2!6Hx0k-jvfbgn*c25m5%SS$_
zL)k+4`9G}Q6XkTt`AEF?#ouH{H5BO=xJ-^s9ZlC#`VRe-9WPx@0DnU5aja;9=*JF2
z1y`zG|HecnqeQ%7$eeiX3B_bew{dMuPiP9SFrgo+Ft)Ju&IofUWg}TpM(u12j;~kn
z&F3^Z3>2G^Z^HS~j_pUXnT)@SFP#33U5vkRA|A6#fdH=TAI-JP@uIG*fjB&^43)t;
z?W!JIIhOWOeAb@a%RerSgKMu{3$WO>g_afI{CYI3EURu<W}i&$lYzcg--$N@6a6iH
zIFs6^nmD>rY})o&j7k7?FD||I`P6$dqhK5PLLIK|CwLu_@@0%SkQ?RUtEu*wf>0-X
ztzA;0%!iZ8;->J6D2p2cThqSLw~2E&Y2OB@Y<%Bu-^Df{zpH((PA8x7tNy~TSQ0AL
zy_res?BjllNZ2iJCPKB;-KK7RGb6$y@F@d!+nX89UDcGL2Uz0UqX#x?a&Co<nD%By
zFi{l40Roy=NxMTs6<S)7>3n(&HD~VJ`F5w?7GJ`(1m}ah^b4So=MR59fkPY?#O|JW
zCUn&K=GX1sH-pz;H<AHQjoWd2#Wo<=Nwxq<hWD}u;)CK#I4IkUH#0)QRNYe}`3uez
z6!5D99_&Bs&5RZ%PbH5wCxZXY#3^~rSe9kZn~5=6R&Mj&%($WEA%K`ag&Gf+Lyyi%
zQS9c;jPc?pC#??#X6Byyj|KLSUM}qzLg!`=znPe=6=h_!k9K38!~4<rdG5@)wlJXK
zyp(k}+oG5y(c8GyEf!kX{ggXjqTb88n_K}^t_J1M+F`5Eb91+QpjO5+j)IsES(K{i
zo!OxcT&do0Q*%zcx&)wC2(OKn?Wh#s2co5?(j4vaIFx$9To}eWAha*V9_BS~*62OX
zytX#_MQ-h|b$ZiGc8~Qj3Ugn&#x@B3NMz*L5qM=o#A?F{Of_C%dCk<7^>6}OtW`e@
z<Zh3lxujmOPIc&6^0*bd-^^G+v{&Q=MfKII{eqPAbh3v2!TL*`{Lw4UYpxCbgo^^!
zU@Rp`U2Iq*pwRR*2b=Um$p?2#Lr%Vs_|w;Hvu5HEG6e0b2_iE|98n!?(Q)7cq%NRq
zI=05JWi`uen_eTac$^R?eY@bq{VFMso)UWNeX)1wk-FPdwTbSCMz*K*DPjnzHFLHr
z7Of9~9KWJxrf@_w4YH(1<vcqD1Hv?}>NvRUjy_j&`uAv()L;T>dt>w;r+<HL&k0Y=
zAji3VUNCS*vFKwQ(1{;Hd_ZCcxKZ05d&2c01m<ZFbdD$xUz`He<yp5J{oW2v{TSII
z+&cEMc29ZBaPcTXuF2;8iaLBOsUlttJoqgM|9wb6*Yx2VHDZTjpKzAQ$`gMld+Skk
zLv;g)(lL$3h2k-ZuO~g#@wkZoYN4HoSv544)sxqL3Sd_*xO2c3dt<80QQ!Wjgs*DP
zn}v#gT<hBcmuC}dz(oir3=0seQ$e}Qv$1$!$c=jt2T*cOvnzNRDzfQChsg3nEvs5>
zQVWjy%EK;UR4?iqm)5vAF2xwjYS-9hy#swt0ImoNaC}=3Gd=mn>vmP!AKudGCMD^k
zmv+pSA8X!^=SD&EwKp?<r~xZdZ(I2g<+GXG^}t|<E9UGIp`wQD(sm=3rVbx*`!u><
z|HOLxOkLL@e8L(C;3|rZp@aV-Rv#>R&HFFo%W96&SLz>9jXz&>lxV5bhnvw72SWQM
zz;dG=rRO!^Qq=WiwVzsO_fnxI8Jeo^bzf5L%i!ClA((Ps){zCmDd#Qv1<9d_+uCkD
zlhOU<eZ8gJ*SDR?Zn|aE(IIZ4-)=vX=r8q=Sf^=c67R6gH6xy7<&HCnYSoh<dit3}
z9vA<m-5FnyED3h)E{)=s1gsfK<h##glyP-$k_zt85BaFqd+(Wy!cHDV>2lwh#J(hV
zuKQ!i?v_@2Ky$+SoE)ELOm)=J+6n=h832O_;t#o&XQ{#C?y5Urv(IGOkQ{*!a{_da
zf6CcBbsyovl&FodJGv-Z6E8zY#zI5hHB(n+L}`bIVtC&Q9O0Apu)sl3M-lc&j0bps
zle0%PX&6S-00Om~EC>h%i}VAvQvIr19N>^H&Xz<+#Uq1Ilk%7o?2tQYSHzgOzp#SL
zxH6E5_Z#JEO0#hNMq=q~rGTKST)S1rq#!538n}(AR(R=c8Np+cSbKexN_5o62tWGi
z+_~1M#xQ5YIWiTm3Fwxts;ykH+|~+fI1yrR$D`ENMGK6BZu40$oGnKgN1(mFNQ!;;
zGPaF@8XWjB2{}Yf!qw^(9ztb#!IA9L?A=xY{a7E^ZP(Vv&hN7#XV01J()eMDH248Q
z)G$;4i@kboW(wro(^!uYdE{hlK)<wX^28u5HmKpa)xB=)A8lCOtgiKpT$}Xlv4(gN
z=sBfzu+4h*l;CalWNe4#(JtE}9KnV~eK*@0uUD*Uux)3=%sZ2voOu9XyD&mofI)le
zOvV%Ym75s;Ew&?ihNV4*@nox$CEFS61;Ifmu%~17VMt&IN+6*TekK~R)vGa^JR2`^
zJy8GHy3RmC|4R=kG0k30u&--i{5SSoX&et?I%&_xldG}GKeA5?Udb;I+5JK=8y!~w
zR8R+E&{;?(FP4DxE8w6Y8}1vkm#2`>($;3L2!<xy8SwE_V@h?{A&rXu-Ay%zONXl)
zR_RC#mv8B^qj5Qwt+rz^K~#5al;i5G3}l_q$f@|lN$r`)fQcWUM&mB^L`A1GN4I=o
zpCdPJ!0Xy(Un84PlO77MrwElZ+8hmljBgr{oDE30GKxog?3^H}a@uiTI4E>%!dx?{
z?SmCI8DY5=3U)E(OExIGq={9beY>myv30rX4@6yw#$@5KtAVq^;7Ff+q)8($wz$>q
z<3Jb%Tj12L2?I^Jf@f^KerBTN^4mfRkKV*|K*-gshiDJDnj4zIs&0I7?NiM_>R%U}
zKMP2R$0iIfrT}6KxTJld9?81pxt_ihUO61jZ5^|(H0g$xpWD|l!I>ogzZu;U=8tf%
zMzc1rw{L}LLhZWjyLg#zB~}SX{pev0@Y`F79yY<EyT8gUZzY<{S!%*XKi)YBbmRJM
zZzWopQq0}{RwnA#!0i)l|5iqLM_`_zAa}f#(fdX+Bn)~i*YvkCf{4A*?u?OKg9j~t
zE4vW9q$o-hrv`UNdna%3oA(4TKt=>6)_o0F@7~x0cAvU%XFK?Tc62TGzm;7_?pc#^
zQNsAx1NyivM{yyP{4q5z)nGFPW*z%&Rt(Dxf<x<|&DNJVT6;#JIRXnGyWiV9y?3Z6
zV>Ca0SXzoKCZGLV6ImjhT}iLv#m8Oo@elUkTZz#jPUxY4q4*k9r_UZ1E}Pqjuvvo3
zN8ZZVFs`5!{`+Wr+_2mhmX?hniFmX{Z)MyVIk%B(d)WGmW4<eDS2Ws^_?ycPX+B1(
z8XWAek5o=p5w9o7b!V0FcLf}fRS8W6sQJoxgL8JDRmaNq<VKs&?Z1`yxYdoztWIBG
zX|?)jl`JpTD757WUPk&_6aD20U$M1%Ie=A$K-YEZ$Z6?%ZN&;|e|OjhJyV7Y8oDu_
zSCNSa?LF0JOQ4R8|I+g6sTLumH9&HgdaBK+o4<i%wOeO=hDZRJ*0zE_BNrS?{~C(d
zUiC()c<UG5y-s2md!8zD(aDfktRK|-;tBlQdI?Fg;pl->${k$7n>6uNk*#V>t6tVh
zszwBh&4FHSEPGO|OHqYeqBCkvIa{z*^C?3-%h@*dl?sM}nC;rieNEg9Pes?1|Fazd
zHjWh3nTfwSpp=<;E8US!5+Ix&N#iGVm$obUkep}A+cN<QFn9CxY;?h$L#ep{i%-KI
zyhpQGy1d?5`n~Zf$IZ$9a{;6CXC7?(d~Ev1%59&%qijSw<Z8d5sKY@uxh)>hPYsDB
z<c@h!b0$x0mmSn29uj!(MfdZ6YOlm!%RB7V*d)c5J#6q$K;MJ<I3=_^tTlDh#AuV}
zz&RqsSEV#BM>R%;m)AI^*KO38BxO!?fwI`6SKwNK@<fb@j)4jaCpFX7fm9*qjX)=C
z4pO*!BRt!Qwi=U1(n}dnL|0s+%R22$X)Z6g2m9GTG;h|R$2<2{;!qRs?k;&AdhPc5
z^MamgK&aqj>^fHyT#OBaByN{9EW(ky3S17j$Y#psLxXcg4P5EPal0A~2#R!n#zFfi
zpn>xG<^HioEw~l*S~Ld5Aq-vD!ZvbyB+r&l;(K=}!(@IVKqCskKGn#BTo33%K2zTn
zIXo>t*SWy5Mc6d?afFx^fbyl5xUU(%w8Wo$r50Tb_O&PxyfuF1UI@W%stK36eG_BW
z5b)yJI=<DI1V|^hx$iVeBwBTG<@=aK7yK%gk`i;$QRu_-x4*rkK#{q#8Bt!@SNy-Q
z$Db9E;86HEye(R|`hfor+wEr)xo594vuS5D(HJ{&&h9vy$Thtaf@0I-DRPt8oo5rJ
zs;aQNG$wAUc%<E3YKJ-;B7aZx-`0kW$L>9w@dyZ=uA8~<Y@!YBL2<u8jk>ORd*JMZ
zTfote57><8l&!gn1)F&`V=Y{`;pTYVW=%msZF#)#)Lo<)$2(swI0f9!^Uh`x*bkF7
zcG-LZiI$2(XbWPhzgS=o21Ix|wAn-I?yplBhKKbGBMJbI#IxIA`$k(@>`~!1)r(#j
z@RgjzT-mj#^vgE<e73vp#TK8<cp~h-zeL69X@i!ipC-5dmH(g(3+nZ|$(gGuO#W-S
zkSk*q$Wm7PtyKjsiJNoQVx{059QA4&)j}Me-l?`yZOocu+h=w0_fYRB8FMuXz`0tp
z@odIt)DWx6xnPCY=r>mqkRuS`Wvx(ElLYE@>d3=LA$#iuLj(Tnu29;bPjH1}2W-@S
zsK(K(DW>OqXx{SiDk=0%T5IgE0#{PIKxub(@|0hvGhXLe(C#TrPN^(<^^L2TxJuCX
z_QwF-+$Gq(2L##UOd8=^uouYq%2zfVeZWAo?}(yyhq6tYGW8H#W*G~_x__{3R+kN4
zdC#6K0g1oqmT1=7!bG=DodVBc<qE-VQ(saA1~+{@lL9X{Vo&{G^R)Nb4)xHA1ca8o
z)plyaeoZ6!aR8ncvdN5YyQU`S95t{%6X?%%k>}B#RiEy`nv&@4F3k;T(Dn$+be{QY
z?2Q5Nt@2yqIW@sPkrcA*t>=ZXTw;7|Ukp0_&%JEv{n0@go}Tbn2h;(Q1HR;NCa>Fz
zniv;KVXP~6dYc^#bgW*n0-A$SUSl+oRSEs%E1C-?0e3ORAx{`Z=Y*B%P=JAsgoQe+
zmz{821v?^i*OE5Xjz)LIqphtR-6Ss`i@Rm<FYLGmFLaU6FmETKv$mF24}T+1o;o9E
z&zQX-*f?w*)CeE2Qzbl;D!|4Au1LB!5ig)Cc<jgDLxZ^Oo)x^kLw@0(i{}nfL?^D}
zyoN0I+j{bBCQFt0J*pFMO<fcu)F9zQcEqI^WR;6rmo+h5&T)XZD_RMH**q8#CEL~b
zpfEU$oF1Ta*LOW`AM2Jz)^?X>j-WF{*VHS&w!mav*P^4VNY+?deWD2>z`|PHhza>Y
zB`c(TDoB#cYx3m%O#KvwQa)Ip&*wq}y46wpA`WCT?lxcQ&2p^5_Eo%@Gi#209eZ-9
zcf>8nHw7<|DlQ&>qeJx=JLg*g>gqxC@PQ8w^>O<?{`T@}W^X5g`Qybuvs>O4v8HsY
z-THP$`d1VRFs8S?orrg`ryR4})i^0(roEl0ej|#Gg541h;JblrIBJLy6&d~QM1$dS
zh`EcupLk-{0=xU|ME)24!tRN2(^Zkga&^!^JdWKbu;Aq}Zr=ZPVgcM(6L~rxP=i|X
z^mp5gw=>43j;u|d+ufZ<X2#g~iuKy8fR}XMpmFxw87n|hGSbKK)11K3I;8)3(bSDi
z-8TPiZC{tR|Bkx_bDS*@z~xxY?7`AV16(8z1+ZWSZy(&J?BSU9Qey5juKOh6cnXDA
zN0kI~*+P9>Gell6#BjZT%qws&emmpMxb!-lTU!!C7e?4G<!apSR;`M+GiIe!U?Uw+
zjO!Z2#d92~tJG^5sj2<IuMBhywdJfj2B=@Qw9#q<HF$8qUe(3J^7>V)>#SZ=D-fNZ
zbl{D#nLxzj=t|aTjD}SAp<ruEtLjPgpm0>N3Gf?+Ja!2id%!@x3fmAcq|755<L5<x
z_7iK05!_-2;`8wo*(<E|?TNd>;sn3{gWGe^+M_cr`Twv^VXVf}R(4Ne=AW?-AGTg?
zG!&697py<lvaQJm-p=@}W~}Lc`LT-z10_s(S(6RxTQ4l1vrYP9d5hCcxbJ2g(~sE2
z5NeWN;pV^<dg9&^>3>qqvH%Y*$t^MA|K~sd$bYs~D?m@NxwgejQ#Qrzfwp1Bd@6o!
z@~qc8re^G!9uC`1b>qwn&H%mkv=(xC@r3P)C-O$PSow_J)p2z`I|XJ3Cga}Zd$n6l
zDqEUtkM;qX0+P0EuZ|drx0{o;&9hxS7pSfpwdVu$j?N+5Cj>V)*?uja7eKZHf$we)
ztMk~(7vIh}OTwp6p}JSgLH*`#Rahywr(YHds@xa1*ItQ{H&GZ?BajQB*A8h#>-&0&
z`|Ys4Y{L=+r+Y+GasOR-@`wXF8r_hJrTFh~s>g)lGVC*iO6Bah2I?xF=(ZDjEdC0T
zb5i3ZmERtFBTza-wqx#)QvucrUoRJe#{)?y9$AtJbw!Xwubqi6@n$-KTyR$CoH@_V
zX?9pF2ue5~+ayOm6`K^ua7S>ue0wogr3PK#rD)mX;m4P?zF-|8lXN)IZ}J7B?yp@9
z^j6l{M;gALQ2FR@!Mwk;Yl69oI7Yi3A8dTWK8bt2IyZ`PZAC9?X}^eW#FsxIk?^P5
zYs<KKKT|*O>ePM>&|rI$DS|nCp$0jU2HTgKB^7h&8`kqHA;op7z4mp0+|<&7{pO}V
zN1*(reIqL5#b5h2g@*OpfWErMzEgAZ98%Fnu(tHJo1IHUbNn~?F}vklMwU@pJDEP0
z$Xh-8+Wp01`gYs7M2~Sx0nXdcCBlxnD-{=*b}plE?nJ{@CtyCBcQl>;AKf-RnsZSR
z0S3iBmmQT%-$}Nc0D?8-9zxr2_qoIyEW`B}p5h*LP&SO#@7{Bnke(79BBY<w^@t{b
z@tyy=|D3{2Qmxdqldv_~1L_TrlrEbg=;0aN+2y3o)Bps5V=1>;=Q2vzi;8soH~ltS
zO)5MKAHsV;%p_Nl5NDpA+<-A(DE4GDeYPO5gY6b{JScDq<ZZKu^ko+&wujG!`hiS$
zkHkj|P&HtWo)f1KK$DRIwdO)0!fo&+2ctzApvJR*E!F_d{X@1SW=0q^DuHCuD35W<
zxSa@_)5Hn7gz%BvP7^Qd9d(EOD*dn$(T0m_Wo)jF0{nZmzV~F>E^*YVIVmw*KTs#|
zsJ=<oT0OQBq+^YG>~-JuHSvgxeNyzURc8&&?QR@crypvFcEojMy?*e$1r6E|7|mk@
z*cgBg`Munvaq3;?kk{ifcBHk&M<}tZU2h)iK=I$H{;5xff4z0<eQgU&Zm$r)-Ae=P
z<Jt`Psk3ng)C1~NTq<o)8?ue)#RKh#ti$?gsUr=OgKbjBHTc7MHkGk+iQ|2|&Nj!y
z`k;9Bq=u_WB`de+xyg;kwpH6|WRq>vOGFC>*{(s#Q~Y_8Jr%bDyU^>dpq3x#z(BYo
z@UyI}-gc@BT)PoRpN?m4@)S93yVPYf+J_OR>Y7|d7B2CukhskKmv^gA*RYcnM_U-n
zlI>MvYyyySJ=QWodtQ$nD3@)Yu$?@SPul*#T4jYDh)GZ!#$MC_>&T3AFrL6XjwklA
zdLSWDCwjD30)t#T+!U|sZ6jsc0OyeA3uEdbn|3%x{CS5RQDbs-xc9(O9qV=sgq~Om
zk^>Eo^<?2GARB+@1{nE7X`TD|`=qew9{rANdqdq5fI5h7=9H#Y%^fwa!MvVN0e{JK
zZ%zan=FI)y?Tk8@+@J23voYXGKZ@rxF-R4L2J+LZ`7-neP=U#~>4SveFXpKxE=Knq
zm}xGlhZJna<#1W|SDWjulF`Yo2<l}Gb#}EhXDX}OM=>De&+V}grjNDQkGZDjng%Ev
zgiGU#>uNw27Bb+(##)6jEfL@wnnChPIb@$sVKsS!<7@VrHY3!m_^<Z4zG2P0zW5iK
z1mXJ_+><Nkm+=+4>Gr;%ws!j}HVxV|ZmqA?WlfG8-*z*eU4N6*SKmZq$TK_mTP>R_
zPTi;nKd{@r3%o2|zi5MfAM;L1+T?s{_D-VM9?Ih&-ttbO`HxdPwaN3h-TIDbC59B&
zMsZ#}slaW~kmHVP6_DQkPDThl6I>q-EYqTAZ{b>mJKo7?v*%x6n@oQvV-DDYZloWu
zJN1(*HHZ@<a6ZN<QQ$*%_dAI}@mdphk48wyxHld<f+4f}-btJSE+vC#CGL;!a?NNB
z9*8$Uwz6;3X2hGy>ayiFGv1&bbJl&9=I9A_I&8MUgJno&Lz^QI2q@^adGT0F^sdv6
z`BNR>NjPQ;)Vlb$ynaw{l&x;GhvH>fDUtBQ?<Brz8I)ku9(gCD>M1yD3*-ADJgT2v
z+2|Jq{Pam%sI?%j9yB7q>z8!i!RumyvK-vo5<%LIi!HYF0L=}?s8TBQmbIlqdZhqo
z)>aje;%mx{v@%9wry$1hKdEy{m?zba)kae!EN;ZQcQSUb#NF{*y?(1z1(C)8wWgFq
zaE<0tu`H^BuN8PG(b*U4;)90d(h8;5&U2*zgKUWReuLsbVH>rqcwVRE!`2iqdl4ne
z3sbdA%gU@(kCQJ<{;Rd?5j8BCbTrmdlAcnhrdyq-lt<2g_D1IzXt1+uzE1K-QWlbd
zz-|i}A>jfC)nnR(c*$$FeaZQB(>sY1+=Sh8Ou*!koous~sf@IHPwKlSE*HGw<GU_2
z=Ub;<;f+YboOd#|nO)_2XwL(%J=(XBz3i!&L1#ZI?;ZMfm3va!&VY9*4nmGg=)5q1
zsKhU9=UoDfC^&mY4N_sWV;E|meP<%`Gx7S$$*T@~nHQ4X9b-3l!1?aceEa)adTg&i
z>BaA!47V{1A|GMv_PhoygSp?QCz8L|etlWf;`I#<2<eY=V3^g5YTnmHdfR}3E18tG
zYT8gdfo0!b383I$llS$jnq7TM@(ZI7?K?@3a>h4AlH`hcIKB&?{vX>BZF7%dZ|x%*
zP9x>In};~n$JBL-nRh<n_|%O+&<&^1J<>^q7xR0v^v_PJtI2&_9ly6XG}_YS?6b*E
zX~z9%2FA5dOTKN`!X{!vjG!<)6OB=2c%bIl=nNC{5K_Bu`E!C4g*VAj0sg#D4#&-&
zn~dq#dEg#vc~M_9p<+z&o|m*MJztCyOY}3BV=1WEL{7Z8UGS$xdbnD8fzir7(pavR
zfg8~DAq}_;+O?QU*_txD9)FU&(>{qVpiYB!BWCE@c{IJBszbzD>}sD0eOQ2zs&RkU
z5a`&bM`mBdqC<2z_p$|i8AG(=MQL9J_(W{LyM8V7U*GaCc2nam#ZQ(TczTAvV$XdW
z&!J#tg5Sj?R&p7Bua3G$saJbG(RUWPsjr+pb<6ogWmqF_=UdNbYI%sjsNHrxyPq3c
z@GB%Id#~MoK3z7OfA$NT7N0VqUo5dZ&g;g20-@W$L~qm2XB0lHE3ERJ`YGx4JmT;!
z^@D2n?f{{ql#l410GBb`k8jZKJ)dzIBZdCrYh*sW@4Oh4=aRw9E4+9h>HX)$Z;+&v
ztNDTRnI5`7cltPq6H!xjkQQr(X5BP^7&ucyP^&tz<+Idaji&~kEtneEOw^<glynns
zI#0861qeu`f~{ItrsO~i&QG{!04cxF*Pl-u3bSi+tNawe++NV&hXo1YG0Bbgk<x16
z>W9u=U@e20w1paoz^Z`}TNJN*h!gfRxi}tAZ=?D`OQ$V4pG3m{{tGJ?%pR_V%gjGm
zg?^%-4%zjPZ7MYwg_bv2ReZi;ovkc^G}r+rP#vvE{+v|jwd$t<PnyKx)y4EvNwa!Q
zKi4~KjcUvZb$tUey*3~w%^?p=L0vz!(UOn0v$?iTJ$2%>xIV@}4(PTG>Xl50_^2|K
zjryr3SzQnXoAlENcP;!6HE7gj@o#a8kBPA=sFMrU9+Pf%psZ7f=hNIK>yF{8UAfj9
zf1ln)E`$E_>DDhfB{_El2W|sebz+-X8yk#Ika7OqhGT1uxm2%F@`HF{*;s%;))%(+
z<^Yd$lD(dA^+|0U^cpF}_7?q8=!iNhTQy~`n2zzY+qMZ5*HnY4?fQW$&>f+k(t^OB
z;}nqVL3J-g)p9ZIoLZjHt+}f8Y4!8CSKuTE&eS(V^*j?(gIvOiqIh^#2i3Vvr=vkc
zc58GbK(YXH#%xcZp?JUzbbA90tsP#2s9?{<qh}Q9XL`>^(=}z)B;eY2K4YkT*Sn7j
z-18yZAIPVA<!g3881Hgk#$JqfZH3~$*}+mf&x!T2Fxf-wrdwz2l^^zw6b7^YYYbek
zMvoVsWlYR?b|?UJX9dq_aySmvq9552Aq<EA7I>(mfv{#`K@et-ouBwW!l=Y1^4reE
zJtC5m0O0v_j|=n)c#c6kF$E%ek>q!GGPc(EuiEU5m`pP|Q#%zu&z<#S8;>zdfg%%A
zT`<WM6@NyfENjQ(!p_FG<N~#GF*9y9$Yryg*CRX+d2V8&GuPWiZI`60#_|LnE@`v6
z*#U;ZF2@It*Vq+dhMBnr_-YJFqBr{}07i0ag0g)aPmyQdt_1{WE|Zb}dh7-4lPvTn
znuiObka|Ij=LtS{nmNh<0@=8jJ`)s>f3K1Gd7!X=h|A=Q7?|jH7XiOiXI{wIzS4xz
z`E}6V&)L`U1=y-7TWUA;l;@)UCZ14Q;0kNs1|llg+IKNxHH5^z*91Jmw9D+>jIQ()
ze`~kAn^EB;M0=3nt?y=p9_ttO_Gp2Jb-H>7xyf(WNA7eDm<5~mZpIEQ!ygK7W&Dco
zcsH>Rp1lJX`FAJ6>5Vgd=ewEhfK$qry(>Ushf;sd?vC-%f~7p&_o(M`)F8|gnC^|v
z*jRYSznfA0rRmgTihnn;A;VbdABcuz7BAS0_`A^)BFua@qyNpt-}apToy~eTV^NSz
zAl!*DogH7~23f>8@i!6sB&rt*D?s6V^+nXOyL2uneMXLZdr<u*X8U!B=vI3u5EJe%
z55KF+#&z;pX{<*y#!~E4k7|rw=YR{gFvh4~*3e*!^pX<-e`t$^lD3q%sn?dgn<)D7
z<yNk>TTa$nwpq?9gyAxBQE)=QaAovQf^7oZ16=$NaPV3g$i<n{s$-!VhDd=xv{<dW
z@mk3d+EW*wy2XfWH`-X3FepXyu9!17@&7ShQJ#Hijpmalk03YHS|Jf`1!f_Y@?NLj
z@+~kO>)%ZTlw*>TaA1?}V5K(1_z)!IjEw?tDLyA0IM%9Z3RTE*E{)`4=ENfutTjGI
z0|Cn0HDC&{>(t|@J3DtrQ=~TNaj!Nwxdkxk_6zCwW4Rr+#Rg)S#`>~a8&q>D9P@5i
zK%s$k8fTj{*b2y=jp>7xLwPpmW+CVYV%3uYQA_f=-6EV~aK#8>TZN2ey_wN>TdeKC
zP^)bh2nZyDZdiUwb8sUOcE^J4i1xWnqqeg&SdxdIj)xTHWV>SPQDfy9^;zg=dvGZ{
zt6$K(3^^0ATW}(pdYZI7nt3&0DYjRSH7+MUHw6?viR68zOpRKU=bA=spHRidp`z>l
zI9I4Hb|Cg2m;xZ($Rxhi=iF`Nw~J+>x^>Q8j#uI6-D0l@RHPN6DD74Ku&SF|h1xWM
zAMRZ{{B9;+u$kD?iD`lFI1;Gp@YokS8V?H=Uf;6Aj%h<fQZbQ1e3;#FO@~lZ;5(s}
zhY2S48x$0!zCLBMU>!uw$EJBBrh+acZ>QAFSiTP|jh7%L40=L6Vo2Z~MB`nsGx2^Y
zK%NZ*HYBFo&S}oDT^$^m^QAqJ@3hHiiG~Gk)-DFBaoqKgM>jiOil)nI%IvbbsgD?k
zE4u!mCw4VjuE9|BQK{3XlC9!ntv}|A-{H)##jZsI?kbP^y8c5{z(h6?H<Y(eqQ9}0
zR=c6~URp)>pK9EOjwbs|FVQa!+UG(IdauLiqkix9d4Z02k!#&#U+QGx2I^KfzWhoZ
zu~&xdYk>_a+$nd9-Hf&8{wn^XeG^N?IVF-k{!ZyDzSFYRx46~!dyU=d71eu<#6hJP
z_EiH>GUt}bjNJG4b=s|y8EuDo8ME6a6Qv};!)~9PaQfUgG?KSzlZjgV)f~G+UstcF
zvFY(O{?1W3K0fifqjuM1Mo`<~aoI6<$4esgI_w_x0`q0}#v8;LD#k~TY*X}DJ$Ao(
z8SdyVzR+e5=!fPGKQuODGVv6ksiNLyPG)>WRgU01n-x7LK5@3*r4pt#N0=!?Ti=C^
zc-~~j`qVfB<=)rx10RfK3xp|ZHvQTjRCl=UVSRikTK;s-U)aO(3WYK!v<-Fq)JWaA
zrW|(QN275=#bdTG{#@O-)E32`Kl|ANTO89YQ!1(@lNoJ<DTO0J%Q%?yfaNj8;;TSg
z5x6KH#UyT(!tu&AHC7ccPz#>w_DMmzQczV>X@R_lR;^z!N0Rev)M}%TRC?`%)v3o`
z>KfF?U`v-e$I&Qkcm?W$t%(gYcebr9Ep7{XvvpcL4vH5RTpyotc>aSpfFrg+IIj2l
zq8s(t;8uqwJyHsx&GON6lSi_&s^{iDY=$JKi=I8ql+@p&)~N=Q&oI;<tvkNH@b`ex
ztA+Qw7~up09dkuE^eC=@QsXVaI2eET<zbYD1re7WL2Q8%s+YoDjs*l2c%vtwdb3aj
z&w_62$!OfrGDuP4Edr0Qy_VA_Z0ip|B5^eSF}6)@sO*<i?COzAo)v#e>jz`r<hIov
z>SdYhj@|!wr+%uUIxrzMxXJ}2Eh!t}t{BDHl;rMb&rD{Fd~d00A!K?oV~)A?u<|)+
zw%r1!hctLNz-uJJBl~c*$l2b~exlSXn0!tRnvnY()Aqck;JP&Tq24!{CGem8H-?I6
z9(MtFjjMLQP=a59b8iRY2sMpTBX%-lqn)~V#2%j?vV+lLpbrbY04}t+CgGJ>S=bv&
z#S1nMQcAkHL-FO@C+6DWz!>~GpB>RAz|)ZOGx3fKY4RmI7F{+KU+cEx8m=15Qy*5(
z?l48Y8TB(p>Uw)a5Kw(9Otz;qU8=ZJz}UtEU7ksVhz-{@O~imb9c|>Ci9@(>iJjFN
zw)A@WPZuN43Dqg+8;Sh<)GwQm(O@Pfqc=#&kX_X2Z{ym%B<NBIsVDlntj|{$hESgo
z`KC`iU+q<Wf{n&Q6%N@)fzoyavyb()yZpIA=9<24SXOV>Co^_Fse(DBzTTgx%?f-=
zZ>WzP85`W1^r>))Q&7%66UvdcVc<TG`9V2b><e|}d*RD?LxgsReWkBZhr#fC9q(4U
z@O)DPjCiH(Z`2f5JEzrC2PO0_hp5c_-Bcqo2ejGu0-GYR+&(zJU&x3jIu6P;*ew?l
zm2*OimHk4Z%%pT4wA(Hu`py>jC^x?qZoiPx_cr2(r(MV>qNgc91dq3bQG3|O^b3jH
zuWl%_JEJ4EEjE?TmM(1R7c$l$PaOxlJ9^>{X-|yQJ$l`UY-RVx;}A@T@2>mwwlG$F
z&F&A7yGKbe_`rp9p*XLUv}uI3*^CR>>f-3&Hu$ueXU32vqV)%xrA~24qk3F)S~pZ^
zw>cLw+U@s$Qh(2jsp4nte)aQXs+D9&TW}$}y)di{y7K(z-S(gmMG!y!2oDJ&u0ryu
z;6lKGcv!;_;)Nd%(*7g*(C<?!!VCY)2sBw3PhhAI_H2>X6^(-H<`(ORillJ1C0Yi=
zEKl`p<<Y~^`ekK}%dBDwfFZ0)@BvmC4ai^BxVpirVxHZ_eYWyK;@Y}~ty;ad7Y_`z
zlIEmV{gkz_r0(TXC+xO+m_@z5p<r~kOQDULX*H34B%{5M@gx_v`Ef>K-_Ftgy{#33
zArHS!dZhobbsCY-=)rc+D7s#=C-b5kT{j2>C|I{`)PS8sq;<5W3!(-4xYB>mU{hrc
z^KnGxQ-N*1wF&~RCoWgYz>Qja>=)ea3f5V|s@q1pty=>j*RxrAqov2&SbwzSVDwXO
zcOXWAIDw);p~$QK;jV4NLdoRz(6ACQzZ)aTrs#3lLrunn3F3;`>|8gS<0rD%HYbsd
ztK-QSm)p)uNNy2??t)aXt+C8=fBKiUt#miQGZP0V43_&v@{}<Acn5j`^zZt?#Xwr@
zjE>x_{dB+x=Y`Ha{w9{T#hy{0wVnX)S#^<11r2Ps-nV+n;yuC&#RW!)C$qh3(Bx<J
zx#*knoT%B=8F{kE{AK%u?uOOnwqLLD4#2t84qV8nIz~y7?NSyDS15ZizTz76<Prx3
zLCdE0oCnIid?88OKl@+)*(=&7)TY6Cja4M_P3wd6GU0BpNQZ>}nxS@biyw|DK<0>j
zi@&LeO|6=kX@v*Lu^o;DSe0&tCEQ(8E2mb;iC8u+5yZ5U0dY93xHZt;&_vlJk>k(h
z?o-hRK^BmF8yBR>@8WlGVH1KBhYB25&jfT{=g-b+9Z29jXy=67$^D$)*X?{j5gwM9
z2>#9{)mwhl_0t#Q*%OeZ8~RdAf(ToDPc91@)bm`L0C+{;kM%jOu4-k_DZ-mmGF~4A
z?+xetadhg&$DCc$dx-BqzpE#&>nGNmv`d&UK8b$4J{NF#AkGcVl_G3S_NnGN?D?%I
zRTFsO;yG!b2SwtDyCRVlVa2}?<e1}G$}gk8`j%Dz$L}s)C!G{o8SU$UnJ(6~+f5Bt
z)|yIKe-m%1i;9@}ZH%1bpM9rQ4l0A`$=R0gOHH6-M!MhsEzCDHZVkERVn*mtU%|?O
z_ggP!+(O*KcH6~_z)kLQ2>F@7IQ_^320ix(yKLIU#8(tw^8A^1!~=6^sJ7`BGd7|g
z@6$UkCYA%UIu`hgnLvQ;6QRPdox3k)Y(^bf)d+^y^GYJwNMRGb0gQ`%_r*X_q5l4h
ziAw$W345T_D5a&g85gsYKKUF95zklcTBDg46U)*@Md?|Zk=Hq359ZNe%nnGA5Q&cq
z)Fp<IK$m$_n55db&DSHo#}uTwm}yF2fJn(dXb;9D{2upP+#V9@frQuz_?CyG52wSP
zboh~}7H-eP$Yzgfap2E>hvHzNejshB=U88)X}GR7C5^xUAJZ7<fDv1wUUA$xmUK+z
z>IBY$w2M}uA141Xo#o2FaXs{G(5kc?J-DzD3Lm|@X5XrX29o*VSFfgzksP|+>eMkS
z={_CxY7b8YQxJ>hNe=K@U!%Q|bQDCf4%pfl499N@(Uev`^|ZdU-bixn&kdTTv$08;
zw^2i(_s8xoP>7KA1<IE?*f;sY<UVT^9*IZqA4Kj4FYPg%bP8BU>nwqbZ6Dyfr}kJ1
ze83Y0&q(WyJ@y3og8PMzvWBHAR@i_*TJ8~=g8>qmxV`2UTW&b|Z|rMxDPq&bjID);
zyC)NOQ@4$&k2(x`wmC){Mr-k8beC(wy>-ZaKDNZ+n6ufoMysXOD{NaJl<?H~wq1Rd
z_l=+e&?9jR$p3mQ=g`|uArKR`XIFlDYPp;my9U5^#X!!rKcgc7Gn!<AXVn8fbA33u
z?v4eerZ_5%Jz4>HzI5wwx$adb&dzf6*mI>>l4^)mMDl%JNOm+0**<;JGt@~^eKOpr
zMOm^&9ni$ub9wyj1XcrP8YX@)P=ukJdb9C&DggJ2`e|e&YRBujrw^IJ4r%06HG<pj
za4aa7Vm0jF5iKbDyVt2;Gv=tHf(CZ7&yGz&)6xx#WXA)g9XKGJP<OfRCOa7|U9W=7
ziz~nzfljAHM7gW&a1%at@rrDsP|XPqQRjrV!_LGchg|kLdokn7y;fR6aT2KPT)^R(
zqt3^lPdu@}CN*a~7ZO`>aq1{~d?i*$yChiJ;ruVh(<<r!A$DI0Bzp7+(Z3M+sa>VG
z#*YF%C~wX_j=zh)g?3$=8VXT?Z1}I+b-`0lAcnh{f1-WGHIT6$Hw4j-el*uURg>g#
zNW==<%4b@arW|Q|KMxEpT0Cd&U)mS(E|q+v>6ZaBS7*4|xI@}k0>n)z-r?7p8E)fj
zOpp4%sVz|tK_eGMDkAxfh8`#$Y_o5*&l(0>kb%Do^rJhq@8i#9^~=oOOEecs=SJaM
z-b;jW)oQ!-y^Q92>9b_MyY0Qi6|ir*TH1-EdoNMNPBiG#-b<u9N18&rBU-VYU_iTU
zdc2~pG-c=BOOzkBmotBt+BJ>BsNVfv#!yh&*qv1GiN<g*@B;TnFWl!jyDuPSfV!48
zyI<eMUa$vBu()<|M!YO|uC<hZVCH)nfyWx#L{(boH&W)Tn94HhZqHWlHSpYX-phCl
zL>IRCJYj{~iMV&PA@ie2(uKI^!h-iQnjT5C^7f#9aSv)jyX>L(r4Dr?i|5S#!vTSN
zN<I=3XMW3S>`^@ojoHFjof?wJ$0SqG<6^<+c{Io&FN{>R;0i!iS^WUDb?2rE0pto$
zYU^3$dl}I{$yMTLvAJuB+WK(+TNyC618%jtT#5x0>q4y%>zsg_lMs`iwmS7t?^k4f
z^x&mJ8{f+~I;T=`SFDNGecV}VOIS@V`WODm)(K>g<_;gW{s&9pDrM%iL8B3CMJYJ}
ziS_~B)0#A<3v!T7zRABgk@<0-bXEtu!!Zc5^VhBF4u*5s+J$Gg+T&Fx0B{lMN?EA7
z-^;3PO`<G-+$pmV=j)jD>Wj7^qOAQS_v@#`)>Ci524Wl%HU#M&)Q)_N#T*WxMp~K(
z<?gagv2YB;EfQQK4C|P%g;kwsW3t|qPK75|dNTGD>?1hY5-&*i&}&<xZ?=R7-P^VR
zq<A1@9q6^~f)$@+>|5wdpNg+Z(9ml;gdbvvi~nXj;~9My;hZe~(?VwhUWv9#i;+e7
zlXzYuBx8+x<~>{5bX*_1H2{_wmTr$)lghwdty*kvw6617w1GCFQMlUdc?}K&oU?uU
zv<#!RM+4dZXo0u^ABR?7AF$iH>_zRRr4cK6Fp!AcjLqZa_cGyUX{yBs&q{1voYuBx
zdsT<{zt6El@sNrb*kN^x&qB_#*~J9<>GWt!7yi)Y&7F2EKJ-ju1v?(IgY^0$o+ts+
z=P6uI#)In_bM{6&;ClpiP6_3(=mk=4umQ(~)ViE&QYN(9oMN5PM#6}U)`J={vEo3q
z)z0ZH_i_Q^Wampr<hC7x^`2A%&lBcH^kNK0vL?6G_t>R)X7PwXgKfDS^WOxAb|oNj
zGaVPeRdtobW$^QnR;bY}bsy`o&2zq8i-{vS!?a(Iu3TUM4WGml=y5`SLrXv%HP6)W
zX@E4^L%g(u_*wMiQ4fD^p9>7fymOXc2&tF{5XgqP8on$YDp%ir6{C?JFi~@jh9u^h
z%x<bfC<<8sNHiN0@lwM6#kcW22={bA-$l<)%!LpCKKg>I?8WKV^VeL;s1=ZLOW$%S
zlc$$@`8Yj6cI>uG+5L;YsiZ)^JzBK4xR7tt)LS8yEWG1VMxn`e<sm53FC~5;RW4&o
z-WiX?y)!x@`=yM5DC<he8&Kz8N^HWiZdUJ}_?t-6Ub{EGgyNB)9X-AP^-M4KYgmuZ
z$4vHsTGzWqX@(x_P+HkcecV}SwOKJsB)iCDfYs1ut3@N%&>TIjgV&xXpnLH`?(k}&
z^DiZqsBy#=TuSVV+j$1;!T64|=(OOWc-2pVA-v*aYA}}nX^&h=OcIst?9upwo6TK{
z7Dl77QHbZF_?tXb2;l(%tiF5Ophlo@yq3y+sUsIidGuSxN>*ITXt!%)hbR`|3`M2p
zMd@y<3i#08ql8!)50nk%+N+Mvk=V-yORgzadnsdKTtDVU-8%i!1l93vS+D-<C`Q<*
zMesUWwnl4KLtY{T__g|>wr^kr4rrY|Ux_Jty*}@8AEphLGC!qUo~?<B6zd^U{YEuw
z#<|m);*m*lnIM^Kj6r6GUskJ{=P6NZ?b^>I#ih4Sy)MfGbGKe2SaO(pV^NB~Lm=zd
zwjdGEu?zo%ygm>U=q&yg=S>^b*kdE8Fz}vJ=lMPg8*hq7Upx=&9*d7tA+A5zX5j**
ztvwklTl_bY!K>H75nMj;ms87ntUlW`Mr-jH0q^m$KQcc(Ra$W*#=Px_J_m+-Y^SDK
z<_c4LI@)peDmrbKUgFrrB0m$)6|6wwR-V;}sY>u}HLoF+PP^^VM&)jFn;koEZ+z14
z!7*&N=Y&a*V#vAS<N4TIWU%TZ;~h#3+ox$hhQV;Zp6WQS2VyEi#h2{GSRrhG1v?ni
zi1>z=1txXP^7e`j2r<O2GlTI$;~w26-;~$wP)sG5Fx2^nFJ<y<s1@Ml+zo*0Z<@dg
zm>xXSM(t?8!R17_d%)2()Qg+y@pyq1kHT*;KsBgbfu6`t#<em`f%pJ~dNfWTPi5b*
zqsL2Y=z3b4h>16Gc+SM19o&v*bue9%GKhWtoHh*+60UGMA3b(Dc#~Q&-1OktE=EtN
z+3Zp@Yvm)mtcApH1Pbbm@)dpBIKU-iS7YQ9C=gcQj|594Hzb;K`?xd#Y`G+riy8lH
zrd?M<QtY1ayaJyHXi_)0lK4i97KwYfKt2uViU-<IqhHE|IN09Z$Qm9W@kjeS2CGB3
zB#>PlrW7Xy`%*u+oXC)0#ncPK<f@4;IRi)@H)Eu};qiaAZ#3s>jA$g1i-yiZcH4Kc
zZhgbWKVpmieriq#RenR-`xyZY#)74|<^4oKdZ=aIW4FGa(bGJho$S-w^uy%8xJtlo
ze?QT0cjp1%Y3i>gi%`5nKjr(nQUO7m{{BRAp&LO~xHJA9?il|&yDR?Y`a#xyKU42>
z21!v9j}0j<5WTtHdU+YU@BKu<Py*Qf?`L#@Iu=~I#HqiZh$A$7Ml|ClvYBd@l2kIY
zSuwL_Pr+Y{Y-Y1HhO1j_jvn(vT(|Sy&v+BJarSU0P{loP!5VeJCNGE&5U26`qkwz7
zbm11?w};-(_D6Ce!`6>*bjThSvXmzIkwD7WDBe$x#^2Sr<=H~PGx?9mZHwX+ELEQ^
zem~=}h}gj;ZHXpJCX1v2Dvw3>0t;PMp`NPT{l7ALdSd<j^;T7Chk<6mR;n8h!tg}Z
z)j|Tydg9ell*cDtwy?wM1W6-y{({wO`@oX;;chH}4L4_N1P3lnzqGY_Z1)l>>)y}g
zm-@c^Kq*1Z`amjWV+S$bZ<rd5%YD=~YBaX$@945AU|a1q>MbArdp3RUcx#Q#(M>kl
zb}h~dtjwuuowF;Q?`I_6FFUl>rLf%Yn8W%1e0Qt$#@~IeW$KR}bN!<>AVj$zAU1w>
z_+Y?JD4bti!<y~iI91!Ejo>VIhtr@jA%N?|waA{PYO_F<Xzi!+$ryUWFaF%N2nju~
zq6lJJwTw$w*H^EqYqV`~LaM54yBZe;VG1ym3vi<NxikFU+aW;EPYgv-QrjuqV}!uL
z=4m}s6cB-amp;mD4s7>l)O!a7nL`ABR)|=N?%j3^S`rs@kK&lVr-V1)!SKB|o}dWs
zp7>nMyJi^Wd91{svv&L9=VpS=_iKiKJ;aqyraAq9mx1Rnc~L+OqE^KO$S4Q(8{s`3
zk`IIQasXO9=$Wh$rPYs@2u8+zRTHTr^4t#TeYdOny6texYz2<zP(j~dM`C+&e-!`T
zj=rC<%q&`O!SzX8F~>?1BpBj&%$yWjp8M~F@at^Ag<qxX<mA-uD3#dZApF7J5M<SE
z$~qN~BOuWIN5_TnI@jHCK}_hE1|)3fa?c3T>S{Y1$UzB^h#b19*X>*Ym-6mlK-Yjv
zi!VXpwKtkv!DSbPU2>Syy_?f6>9>9^B@b}IkRF}l{zz!0T~ROivQmEjYP<z|188<W
ziYJ!KdEbu%`DMeNlf)H0*W%+<Nf2S1UDwBrBS_zZw`-vIn%&U+y$G-KD4(jiSFZa^
z@12o$n)7)HC8S(HwR4CqkN>+mMZ7v}U&f;+;&IO7E3Kbr?1rIpBk9-bxU+b?+pV!T
zwTnFO4J+|YtOO-kIGNwZ0|96}e5a<U>HHLYuOHkRM3Gzy<2oJ2VudH&<&56ui%0Wz
zs~!n`#+~kRcKIfC;UDo++-kR9&d3ZKp5>Vq-yo|)^G5BC%b8@~SlwXL<B4tQ^6j0M
z6O%z28J6m<%Ndiw!SedHck4^aM%q1kszWPw@8yhR0kskNa-ZNR?`tAP<NnJT3&7o(
zcb}~XVmLH#lxvxx9#Slgvq3ZCJzM8*hHTd5jIu8$sHnYo04{m<RF~9mZZE!MbA%xL
zGGPY<11@nW*5=1+<d+J3cx{2^P{m#NpeF40bN0eR>I{c&gmb+w5M~OgdL-UMGu;?z
zj|RBq<fv-5g_jfWv#PSv775By<dH2_$G8%=+Y*7{zToB24R;S(hzhlY^BA^DA%Wsn
zIB`3Ps?sL#DY~Kx-P}rz*}z6|UcOrR<-$VO#ObPy-f&b3@!c=GI^n#c#_BaXK?MI`
zje@g&b*-(5ZJO(1o34$&n~4#&bs8EDF@=e)*LQi^ZV(!s{Br^~#>Xjizez7$WrH_C
zUhoX|SZj2ICEpjfU5kr)5r6$o4LsP<iG)Qpm{^3L{^Cd0D;Ua=CHwU$;uhTbfIf98
z+m*P3YEOm>8xBO2EF88;xJ1iAMNKZFF#&%1pGPoC*k=7u5zZ@5PK|*E+#~X~2)#VD
z+g2^ECzOQvaF1+LlVz@0+pbPA1GrB0sQ?rC^7RYbZO7$|ef6!7Ojj<i?ku6L8)Z8w
zYc%)jJ{JylX)<m)^}}n=XjCi(#H_~`(2+yQXLq-HAfyee=$^}omB&K7SABUr9K0s9
zKNb)iunC`!1<8?&b6)_3zRzPP_6u^zN~##jX2gFkw#J|b{=X=o*sZO0FdhlhC-Q|r
z_2}8bc}3uPvRj<v_2J};Ozuz&%${+wdN`iAW2(t{cSK9Taih>rlF^0p$6Y@|gn9<K
z*H5?O8jVzkIL)3=bE>R4i*-`#UNK5tan#dq1WHzf)E*BcU!d#pm_-X26(*t|Sdvm$
z@R`dQ=Rw9+7C-9Ro{i7L&GMXJ@v891?=kaAoa52`+hn}v_;gbt1_Ce0V5zVtLf&<k
zAl@ZsDZsrPLx7^%db=WIa=}s@Cf=czpybR)0rCp2rA>8yv3(rN=aFW1O?ZJ#E4~12
zFMZmVM=$Y7v>fpSO*gb8u%7m*zH_4(*TrZ0flWS)h4*tIhE(sRtj{>etJ&ngT-b?0
z@k_03mC8(frJ>xd@$0}oXU}fN3(s;+R@-mlsepnGw0$eI;q#p2_3!jcA>j>|xaex`
z`nhHwWb~LkP7s}2KFIV#$>o;(454ARTR+Gsb6FE5+Vxp`@hF1U?RtjZAT>|Zb8Dv;
z8Mxzvj6Rk)VYbus4-zK<>yJdQ_Aa*&@%BODERu;H>U?($?wk}w=I{9+Bj^2ygMP=^
zz0sas@VL|F`_!EW9Tn{U4>E3{cs$iFc_7-?m#h~v0u-#LM6btK9k@DjC(QaFF&zbD
z&7a$BjpBwmn-gETCEw;nyKb~Y5OM+7HA2lm_p@B^L3Tcp2OSkl{Q7xNNL`7(;-L=`
zRsPBPU)aOZb@8{DI3E%4jkyl$z}ll9h_P~e^zRefzVL(WcIED1382u!+&qgkLL&m!
z;tw(|2ie<md)pFqKDie!ye(EP)YYQG=(dWf4dHI-RLpkAU@{3*3K_Xg3{<5RUaG*%
zl`)~N;tN)-aguk=pw;U03T`H=i(#QyJ<fhxeW1`O1z9p(b6}0?g1jkyEZ0Q$p8RBO
z^oA~?ol42JE*|@Hqsjkwy}oSiNVP>a2rFe36}B;+JWCvr_Dx#6u<+7*UN~Vxq6BXh
z8q)@4A=@>ZY7+2Rr#_&#2-~w;0FY$f+1K7!UtHKw!s{imA?|ZDpw3p{L2H8_Bm!1J
zh+KTtuxfR^ZHf&C6NANIOvo>XpCZ*cywYaDKvg+N{*&qzq>?*6MkW1BK4m(wt<fQw
zbx_J~%eJ#azg+;9r-Wekl)5HxmGs{`gpZcqPU;rrZKqnSgavD|r!_e2=%DS2N4L4S
z1pbUR^m14aZU|SUJsXqd3d2GkD})T^r+be&YbqYaBR*iow(r8cR*#y;oJDr#^8$Zt
zq+|S7W42E}JmzMgUfVD9rF3Vg5Sr|OV1g|n8imW09JHEfjT@dh>IVZ#sBxdY9Q*O7
z8yDCs0;3hMU~_WzYP_WapB*ZF)Z5Z!hXoAd;232`qTR*?8||pDNGAkZ9E)cTp{sIU
zv*SVp-rW#)b=5!&3Kzx67zfAYf3Y`oKO{<(sy`DybxPA@SD-X-9~m2;8W-JZk4<Pc
ztGc>#?bKScGx}*wHy`zIxt!JTE@pz8bMZq`MI<}!dEwd5H83I*O->!f#yYzglW)Px
z{E~1T(G_Ht*Dq+9*a>*cMf2(vOYN#~f$tyXU}9|4K?9W?o~r%h7{|p6Y~R-eetAoG
zTMD?o9>}UAajkt)x)B(M4Rk||6OIUTw@>3t{<kOWv)Fy7Umz)geJ&uX-BkRA9v^dy
z?w6tg_-5saRY;lGzlsjHeK=}g$H44d4;j9x?;E((?Hj>CQZ5`uISStjCCEw1qWN7+
zV+aGDeXkYq%(|%Y>`F!l`}<P3>Md6i$thqWzV%9?4Q@#COOyR{+m(#u*RoKz>*+C)
zpijG!=whyI6kFRJS2E&;?Z3Ta7zT=Pt?5@14Mz{|0hV@W3|0K?nB65X>b-9GfZcsX
zX_fwefvFevp!B=v%7kkj##&grdjk+>_%Hm*?o(fF*j%^)?!S^zM<jVJg&r489*D`o
zANbC5L2!nCK_nd*#ieoPmBdy!ede>IT$>eBSjtxQOm4F^)>3jxK=tQDC#z8KQ0d9}
zg?SpCo3pQ{K#@9J>8@nlPebCs{CZxX-#V~FwUOrs+k4CmPg6V%?4cMOzp&(>{cvEs
zHRthVj|h@svLM1tKN<)c|5ZQv=;A{r%k>8fAdn{aL%ENmb9>zuM+c3pH3}4%nI%^;
z{sH}+bKJQ!<#P4a(}yvkLgQjf_WFL6`T@j}gwm?6B!&|Aq@>5PmFl3{!w9MaPO6DF
zSZ(}0>Wa@gHEY1wVD)<BVr{a<*gH*_uGVNk7RvQBYxRTMF>y^;7Z5OgHUREk>(v(l
zVkDc_K3-8?V;l9d5l+#X^avC=%ST(^q@3T@7>R`~rKWCISGB<0si((CD%35oTo=g}
z?A0NOqZ$>d`eRjHL)>Nq8okvi;Giago*Eu<IG&xJB(`RghG?L=CA|OGmBcA*{IPA;
z?&h|pj%>l6jNJ`ZaJNO*jCj69OAZ%FNG-+z${X2MK|v{m9^0l-Dv16>uf4sr7`R6~
z70?qG#`)QyhD1!a)BjGv@2UwL<pM!esP~hyOU*|LW22z=ndl4%A-~VYLs@xQ1G(m7
z#7<UZPxRp`1$1D01!}UmKc}6Z?0AkOr25!plYc;kxG&nf;O%E{zp&ImP&K+KJD^kP
z{v2>SP|+8ogJzGKJQ!fDsI6GG)?SVWchDpE?9|w9iHE78lGLlw0(uBZIusA&r^fr{
zur|v&nCv4l!Rl3y+fl*fQTr_jKgZ%bbhR9><61^6*rbPNWlsc}Q)DmdO*<K#{n^|F
z_J(l$cvy^1X@0B-=h00w<3i$ckARxc2>4>6V_<|c0(KZ(=vlSL1_6~1^bEU_{=6D4
z$Kzu%wqt(8E(RRP-X+b^B{h7E<9%5@cDnE9mFTg`IlrrVRffXCKAPIHM@sdR-E#1;
zx~pAXW7qUeA3pVVJzlcKM$ue+B9x$Ba>n3BARN_F@-sz-|1`$0URH^kUVFHa+>a>f
z<6XfWn!m^vhO%V#r5X}4hnwbCLgR(MM`A*1vah2_1J28-3h~Y8YsksbH)@^)EZ1{=
zD@YShUa;?CVM*p<-^ZU`5zIbJblbC&b1FE-cFTu}kdgT{d2)EH;H@8K6dLiw>shD$
zf7^$Nh!aPEdE)jD6Nzmaf<c`YFDS2r=I@RVGZjHoE{_dz`iGg;9}7#b-T7fgIQ$lg
zMZw5;SF}RSmUBJR-O;<J-nV<y!E(QI@6{ugqCvZ_1Qe<za>r}zr8td|vgQYExkF~e
zOt~^#t>gKCZ03g<t>yNoNG2+0n-yK+u*<5={xIVYk}ocatx2CT=fjLK;&${yF;6{n
zce!Z<ss4ZMneDJ3rj;W@kUjWe;-yHMY!Ahn0GDfny=dUWAEv9hywV<tX2t&-L4FF9
z=U6>#VIXc{>HVv{wkT#F>;)ub@rQ{a!>osm!Is3Vuxc}_@<2@HC08K;Rz!X)tBk&o
z8&bMTtD0IiSFEN&Ang5>(J7E`oNd)H9uJ<tqE;v=3{g(i?U!{^D8VJPmm3RneGHAO
z4A)EAW{v7*7-x$$TEa#*7O#z$6&}r%w{>b-zpT;L2X;|xVW-y)W?fO=hBw&8KvoYq
z>YB7wh6dS~me)9~#iQhr)d~Y=k-B;9`k?_61K911(He8;)Ve>+)L@gUCh9cDdbMPP
zjFLwQOB@oshU(3Yjo84{B+$+e3NFgfLk5QRj0_18+9b4k1e=>|ZR`hY!-BaX!fOt0
z7Q}V%XT)bdsjiwZLpR%&crI^gC+{^V*c#87`@e0A8MYvF+jf21lk%KC6_}i}V4m#|
z{H>@ec9uTkblKCfJG^c=s(JN51_Q)}X95c(tbyexuT5YIc|J+XI80y@+Z~;?5x{JF
zO3-<-G52%Yt63wVdZ>j9xhL(p7}IGg#5_QQ_xJi5`{GsIKwn?0?H6|NhVvXS2XqEX
z>SB8_Mn@h*aF3oK0e$2Ue0l0)53_=tYgLlp1~K0DYJ6RM6~Xh6=3ehA5j!lD6<^w9
zN3<Wxhq~MBsP+uS**vDgdA(ygI)xU7NcNQD(d%m1#pF$NLO>w3Ldqb%CpAzF7uy>e
zs3XsHcq-P=6)@u(GqIl@R{&{`OX$z%+L@RO91`?A{%#?6qMZ{SP{O%O)%gI)k<V3{
zv&k4AYU9<XFUC{xB?N|oU5byL`E!NIh+P&C?g!u3mLfK<2<EC>Hx_tdvbDtxJ|F3s
z!-6pLu^{poZTS6brTsvPh!(r9w;rd$=bvN~Bevm&hM#;M68yT8Nc&VjBI#0L1KthK
zI}N%9ibPV`Kx7RY;folq4aZ#S!>QehLXPaZU&UNp*Xj)Q*IKWXA-xw(!cDE$05x5{
ziO!nI5JvU(*q@mL1SZVyG@A=gCtKF{A7<V|RO1=Q;m`o-#!PNkGs(P*%<{KfO@wF8
z%=va}{5f~-Pwlp=8JR}7AO&4~lKA}puxWaU%BjolxEj(tc9*<Oznam)F0`@C`p&DF
z)|lIg3Ic6*SF|r{@*Di_s~La5f`Ytz)VwCwiQGh#n|f?W<yP%J{g9*52Iueos~HnP
z9dwRsN&bEy#u*xD0<IZXMYw}g^K_;&qeXoRGMP2iA{;3Fl9(MRDM$S=NAFVw_3-YT
zCsg#4%hBd*HjVD4ZVRqv#)jex>_B%PdQi~!<9|)2iigz2AUuXWtYKWV=RR?U;t?%b
z9>L(z_@dC#=0b`sj7JKclU-%h772rHO2+ZW7GKS{8dqwfV}Wm4Qd&`G`wm&T`mbQy
zRm3d*tn|++)gy8{c6s#m(H=(<LR@U6W`OWguxbsrs(ZAjpWBB>IL&H#2*55u)kQB!
zz=8B@6skmYctOQReMAZO{>VjSYc$xpTsyjZJ@)3v2(?ZVDIaRH_3^R}wWMocHe5|i
z>V@C9Wj+?RetEey#V%OoVFZ?sHfSY#tyL3Q>WpZrK;IsXdy{)>ry7vqyPM3qbkKcu
zVtQnWSb_C{N~(#A>5b3OPIDysHB<Ig0XDz}^h+!10zV{!@j(SUpv{I$UpGSlY?JmF
zK_v`07SDJrqE&`+Z59ghxbHlv9q1--XyPH;BB;Y8=EH4`PB7bf9fo7JEzm<XU)!!#
zz%$<keX>H>Qvs3lsGDqu9{y~u?G$+BL-~BWJ*}xZhwWqSQj29Mjfrp55|(?ep=b4o
zl5D_sYdQ^V%{}T3*IUR+-u7zno?&|~8W(?yll`DQucKB$4oOte`>tk;nJd=EibuJ^
zaacf>*C13?vID{yu5z4F7|0ieGaML7Cg6jb2gY$C@L!J4L>X`|ctvwryJ9U4dg^#t
z$p&qQ0?|#B#yuQcuZ<<QBkJ4J2T;ddUqaYM(LfN~uD&tbwA=BRg4@(NpeKZt<#l#4
z9<Zu|{Wk)GDaj8a8pq<4I_f|Dn+_WnE*!lR!r`iXSB_NCXY{(J&}L@^XE~;qHan;0
z<atN0WaqVDb*1VOlllZ-C0N>vnwqn(5P?y<6xdFYo4<6+I2Oap`iwG?T*p@gbv^2A
zyQ*>UO-!MSA8E-7EtHRfs~EM9)k$BW_*#=)6S#hWxx23GBj7LX*H1KK4y)aWRpKTk
zC{Eqs7=;S$Gj+$r+<^MI=7u0xaM9hqxSC|S`qlQOT0f2m&3*Tk=8y%lejQC%RMJ1W
z^V7|Mg>#6DpM9ebA^`wt?ld4V&fyjJT>zN`#Qy=&`(91J>j^i}oBdbD5saWsxaGex
z+RqfIE)y?@sQ$@)a@&7p^qK%wg#Fw9E7*s*PuMhlSLqVp9cqc6Cr4<yUYB)n%<t6e
zWfZ5!Xs>57WMEOa`~RiteSmYm+O^T^bUK|*$7edmn2uwN&#}E;pKUvi<2a7j_OGhm
zP#aa-XWK%eK}2>Eiq~uMPeepS8W9l$5kW*m#3Le1f~~fbrao0w)hAE#JV`XK`h1S*
zbUK}Lt>3kNOTU@S<a+M)?_Pg?Yu)Q!_qvz+y&$EfJO%^A#_D8qIUMh)*p{=282Ftp
zV9<3Io0#&zcm8_kDNnr#c<^^R2c3%{jt}#Q5Qcu|Eeb7XsUIBnoi8M;^KNK)p2UW$
zsMX0EFhZG3`3pw67NajtjMA?U&EaV*Mt|om`Z7=Fn5H@17#E#S!eXrBc<t&W9(QcU
zI9Kp|zRJqYuu;v8SHa1Yza}V4PPbFNGZKbQbd@v;SsB?+QeC1~uHmBpohOx1x@IM(
zC}8!fReU1OH7vzfN>4Lfd#vKsBg|Af`buRV&e9L#P=$tfww4KNjOkI7%VsKvCXeJD
zKU99Mv&Q%?Yigvam;cU})4W=wFdMThFH}9FX=%MkeBM4l)$3xHm^$AiEOF)0I*0XN
zDsIe0SPtrC-}z$abgqJ&RhGLJ-`G-t6)Ivfh1?ib$ryg{8CaJocj8n!#<`y1MO69}
zq+VXpdB(!oE#3w5xGX;Q#%?6IS{q|6y_=rsXmokUy^EO_lUxT<22Alx1yaP*!ddTQ
zqYBx=3p8|9qaQpDVJ)|Ocgso>-cFJ(@{EkJ^}FA>->c!0ZIm#^6xuA!+gwrd?a7^1
zs%sr{uFcL$cdfIQx~ELFI^~xXWT}dt($Odn8j-CwnjXh{pXTYQ9G4lNQjMlkG*=a&
zISukGq%Bas>%QbQyx~yb`rgw?Tw(z=UuC1vwJ7z|oNd;l=yf`Cj9Bb?h{6HA-<Rk&
zIin9M?s`gJM>ltq4!h{a_b*NH+=oR6h;(f1UCKD-YT$WU;PpAj9Cymp#u;rC8^6Mq
zW&A0UT~xZvd923O<h0sty74z!B^=MwU{t#vaMs>6ulX=4A(=DRI^Lr`ocm9xbJ@={
z$Ns2SmV7afPsX`1^#QFT<D+@42@MK2ZwH~#rTv(vGxRgBj#y7$Bi9`_3#i3zsbtD#
z7n?`JtmqBL9k0l62{wxjZmCEv5dp8ujL)yz8k?KqQr5-ejs_9U(|I8YcU_~dGs@Ql
zw7N0IC%e$*iZsU}DY&PcrmZ*ruJ60R>2%ai2exihc#l+b-FqmCePoaS@JQJwa^~Y4
z@>p%L&fE!n;+l2FYKn|O1+7k|$!*958A%+Su4(w11&_2j$wic{r=n;;x6<;Jw`4pO
zUcd(y(CfNG2fk+C^-ST>Mg{)dm9x<pu)R=ar*1LR-Ah+lo}KbtvR7{KdD69iG3)2=
zZuu@ajC!}h2<GI<JdyT`W7}6k<5$O!XP3{tCk^#92LAfjKLL;ZHPSqFV<x0G9{Xz~
z-wHPJ1!vqhANy-N6-s>B_pQhN?v>)qD&DuEdz;7p3Q2t=m(N(E(|^as<E0wPNEBuI
zh{SZ>A-XBC?_*8Tg{7gwRQ}=J$G+4N#Z7HL70%Bzn{vrT^ndJ2A#2TX$pBXaUZv(X
zha!N@kG7khQ&<diDW)4^o<S~!u>v>8L4zH|JjQHz8X^{Wb^lYEr3`)S$q0YhieZmE
zp}-h`VYo|0F~!@}k5EB;(t?5wE%9Ab`o43h?sqhv<C_x`iY23!>9mhNiozII_dnAX
zZLC75BBRe6g>Y>)>X7lSaC7cUgAz=5?5*n*1-{yeiQ>dq{4byl;3WOz`H*+#ZnC1%
zB7l3JDaw}?mz>=&&9#+x$#RBk%S3wVB(F0cThopN%yOe5VM78vLCjXre;zRuQI5Mr
zx~DNm{Bu>x0#D+`JeTbZn*DGuIsdUQ0dV@J@u4|4YpcgXHT>d5SR@oryL%P0dMz*F
zIc68EPuPs5inH$DuuK)?%PhPRz1$TRMW1h6C095Cuf7{mFGkUg@l^~~+3|Mit6cNl
zA<o67cMPMvT;q7<GH%3q4hGJw@v5QG{c*@7xbK0pvV95!iH=`BqD2dQ*MOeK9i*tY
zHyEwwMnyHsNgiQflY-XsHZHa($V@QCq2JcmCGnwTdTuiI65AAi=1ltKPIbk65N8aA
z(p>^h9cH<tMjV-H)db3YS!xY_t;{If)&6~*SLL{G`ZKz{@jPASs<+)7{9ll#+*Z+M
zJzvejVL4*tNE9gGPk;Ff3T-<Zhm*#CP?0Ms-W;jX#NG{}56o7VYek7t^9w+!>jWmw
zGg67C+C1z^`!iobFLO2L{V@*iV+u7|e#$T9YNG_c<&6p#ldr~bA*^&|uj8(oM(a4O
zR$)D+&qI}daPP*?yxM)zPAQHWaT*`XGkM!xMorCH)hQ`o>80z``rjqhJx;FZu6yj6
zhdJ_Z(9r&`|B6PXH<CB?_F~(c6U?-AUBxblU592tYR7TIRc1QPYmAJ3%T<@FDrb}1
zE{3t(G@fwpxNe@|-<7_riqN5z=LoOW^_MYY;SKFJg_|ePC`{s>%7|G%(|f)T_g#)L
z-pt2&_krpnW2%QP+<bOl8T+h9jzrsBI+vv*k;kgm7An7<xQ6mXxXTot0thNgcGJxD
z2)WTPmHxszT~PB;L|oRZsH@!2ax@9uk8fF#I)-;I%mI#R(x<8wGc+y((5peS)OdD$
z=7KneXo%+;b-MBQgcl0_aQf7RG)%%v7dwV$L3rhwkISby*7#XDC-cMszqmkSFvQ*U
zufOXcnnY20`OQU%GEQq_fbacvk=`gi!yDgwizo%3wOjz+)GyXz1~q|ixnB$T)XUr7
z`zyXt4e&68Q{+3}`}6S}PR~<m*y0^U^!eTw7v^~0f8x>Cl}abHRO&0jf|bUy)qDM<
zpUa7Sn<<3z3MaJw-+QZ-c@;An1HSj=iKJLspBS_3_rCX*+&}z120CVEv6h4M%RJ=A
z!|K7``@1`8y3FPLAucC#xx~8-JJiKl!0U4L2^i)ou{NH%DYKF*Q>igA!d0p50O$Xa
z-}}PPEY5eEF-pJrhF;2Q({H0)cF`L?HLtQ@jLU#ZLwdXU-XB*?PsM49SuZf|dtc=^
z$0))nAsGMJWM;ftW+9g*F6u7xKbu*I>b!~HdkgT~_dmfT*R*_}-kjv=V0^MGo_p0z
znBsm?_FIW*N<W8fJVVv?JiuSZOqENylxxH+`(w4y(a(0}nCm{qblwwNq7-KpEqr;`
z-nG(NZ%aSVmAb%;r}^LeM^1DTl)PnaG+$6{x$qely3YOc`Zz2SZ`+0~#p3U88Pm+>
zE3`LSB24il7E4tqtHN=-%>Gzu?gA{=k4>A+9%Y4oaJ;YM^KgzgqtxXJuVUyf7I7M8
zAAX88ZsbfiRWw_Xct={#j-x>&pL%!wF=w;q$q9+dEID=+lEkOy{xiizT+PQ}XfuP2
zin!>*1=!>ohx^6#*y8y6n4d>Hwz}VR$iltqHbq~bl*miT>yYXO^gJ@fZz<iyH}}0a
zBh&R47kBRby*oDnWVyDU+ft5fF=(#Odi#fq$Z=f7@s<j5UG(X6$&;rBHJUV@(Yu&Y
zW&>2<+No`?r)y|Jp{qX4Vz;bkj}(3HZxB9sZNggY%7|jUOI)_|%u@`d?kBfvG`Lq+
zEnJK;wPZ9`(__lOsCK!7Di@Pe<LN&K6$)Zknk{>!(nPUhr^Oqk8@|Cqsh2Kql#ZTE
zxms4cV(Dif1~sm6Js&-&bu7P6OQJg07nG%plBM1e==a{-0tmR4pv^lkV$jcxhDkJi
zx1iBQr?sw8hBmpj<oOU)Le#=scX5n8Fq^;Gu|y7O*>=Mf!FOL`jlZZ{E*G|oyxHA$
z?N6Ci@46@Bj+=y9P9!D&u0`&;@AY+Warci_)&8p;U+tyoS({^I{W^2g{hljBwJ`!U
z?mMD6Dfd+Oxp?3jnfLl)@X+yVY^04>3OsV{JcD1%V|Cw*ssD^8t~t!H6-}FhF0rwl
z@gBnssYAVY&Cuz>=hI@G6Bw=YTpqlfMFkqV9l<!0p*;N57WmP0^s2w8rjN!mVOj~S
z#B&wGIp)KU@xoPTBNe-N>AEB?dIjHAO28|{ZaKN0%Cw(d!_heor`um%7vI2P6u_^p
z_6vAU`<shS<0dY9c;YX^pO`%_#gaFkc#A(50nZBd%_sg|@H4LVPyE%D$K-J_bdQ6#
zpLoK7(Gsw(?>zAZ0CO0@#fkD?AD6?-Kcnvxe+lG<nA(zeUHRU%WKM4Vo_MR$hU69K
z?;>+@YiYm$_nTfVsQ-KKiN9ulM(s2|&4G?!lo%Au2R*rE=Kg5&O!i>$(Q<Snh5jL`
zrIE#qarscyVg<9|HgA}#1<mug<ucla;ZHm%V$Hu{gw2x{BN*wjnHkTAJ4U@OpFV}z
z9;0oU6=v}pqsSl6o$g6-V_ns!MaLz?(L!k4liw}xG&|riIy+>%tMvyh_5A23xH4C6
z;MJ;$u1xRNY78bh!a|ci8I#qb)aqf1@}iQ`?A8L9=JKWx#A$Da<0*+-$5_6=J=3*C
zGB3e;q5)>v3_1K}D+BM8-n_FN<(T6)&NF$JACF1AwQHVjJvJ-mtJSIcqoYkMeB$4+
z;5%)6U4|+eGlGrVh>K)6fJKVVJ~O7niydS0lISYK`&gGe@nubOY10@jvFoVEcP+xZ
z&nwq%#xggKjn;{$<H=a=GM~xbUg3Vt=6OYo>f3URZ_lYiS5V4|N1V%-FAbRse`{QI
z>Dt135HsGj2gf(xnsF)KU;N>Mg$t1AIPp$)=ne)+PkaG~YmsNCAYm-U^%ZX>ai_J>
z@%TQUM&0BxT4cOr83Xn$;+1zbSih|<_Dp`0+Z4o0+FW!~9aAy%SxYX{9f9XQya$9#
zSBnL7wvQ}VP2Pp%$fhfLM`2&AH~Q5a*FDA)yiP50)%Pon@mrqDa+UGTp6`m}v~5hq
z3S9V>mZNJ?=)#R(Dt;A3sxLd#S#uTJp_RgKtwf*e-;YMAsGRI*o#&Y1`bWOaldxs3
zVP>tPGukJ&`pL9nK8+#0Da_l0@-n>!!HQ@;K%*=vMaG~~d~$eu@2{VBd{Sv_thcMw
z(i_Zk0H}Wbfg68|6tm4Gu^Q#P<;DTjDmLvljYQz`(j5uZtK62OJe>?ECO?J+GtuDc
zC6%dZphlP4Xfr7|H#x!r@5U;wtK5J5$A3h#8(VB9-cP;Zew){|XqabfxgxHmoO$1N
z!8|DA!hFXKKHqO7?uysdn^^5u;ZMyarZ&e#A}<V37`x{>dR^?6)wusUoaQF9Q-5GH
zN~WKqhr$V~*;J2QM9#EnT)ZE<jx<~FCu(Ynq&%{s{*Ci7PpN`#f|z5jJn)2#w&e*3
zsU_A>eb}iYwwQ}a5mzmv7NAL15xQMVa}T<T1DP9^r<TJtTz-2USKKj~?ea5y8J@6W
z4!h_h$MHFI)KI)|3AEG2OYy?PK)g~jn}Z?Gs^(|+mGL`GF*=N2Tu0EQ4|SuziWPG)
z7{9r}#%&Bu3Cw8c-u?%FZ_g(o*WryHyake0UD0^+2Y>l7R&f3!7jOOG2>{Z6zwJ`5
zGbefP=qDW)k)0Uy`N7w=b4P~0Kln>M?;G$W<XxA&E>UxKqe$BC2XDFm>mQe){|}xl
z;`z*?z{>NE0pgCg%(>W$4-WX5(=hM{f90Q<6w5gXgOrVVt#Ggs4W0Oxp%@|_jDnNr
z4{I<~VO+*&bT1s|Dd2zb1)G*rbQx>D)e(+ll}Hyc@&|w8VGe}NAp-CBjS^o*5%noX
z|KQ6%(Mjg4(`d-XxKv!IJOd4WF=Jf?+K!k*LyU8+MHj0ahvU_>3*+byi8{Cmu4zBu
zIy2GsiT5JIB-bZAK25=7wM6`l?U<rM__i*2GcQ?9b9G9lX4!j=V}|26YCVl<csvk|
znU0nm&B?<nWtOAK)3&*|HrvtO-xSNYF^sM*>IY9w`q%%0xoY1vX8NAzYUfS7nC}?*
zh*mD13-^Vt8c`co^6tEAAfEMc6~tmk;KtF|E-X<+S8j>MQkR`4_ZdGI%YL}^ziJ}q
z!*S-OpQoJmPPmMvtvPF5E(SlOJcbpHK_iE*M2zDmF^OXHDp&ZYu{_pKL7c`aN28o0
z(|?X~V;8SSyz8y0lyMW>uMPCvlIVCPGsc2EfFxJ=RBEUB)SPPe2}O*J$~Bs|zpzQY
z-crs;*z%eqW8=vM-Vs<?9@Jx-iuG=etw*XW)-;is7Lo2qG)Sk4o_dN*m7K(NF-s*U
zuHzh>t&)u<j_=I4s?(JKhd3|OtLUi=<>ei&XnC(&&wR%Dt^wj(%2D7LwVqcCmAkoV
zAB!T5pZEU>#Uitb=gdKge(-(Cqo1Nw?In{0PRfU0=g6C}?42^lqW3@ur<r3vyp{}&
znS;?R@sPFLMWk7;cPdq(^5YUw>A0S^ijMJIl@^$DzA9H_EFV(g*;ln<CrRg6<AS}<
zyB4*si=#g?d(t}9XZ>f<bP-ywR{xj3pNxPjW~LDh8q^7GJkUp@V$rb`nnX5+H@NP4
zv8{~jixHHY-Ap^9rI809@qB%96>cbf)JM~BOX=4qQ@Fh?PH3A>|C@Ilo94a--;Mpj
zS%Gjb>1ltAJkA?QWT@3mY}_Vs^GmHM58+*l`X;gWRPKmBEX92_6344CAbj9Br!Z}v
z3*zqGRi8e?*5Z+PGg^H<9_PA{LXL4RL=#?415aF%=or3m71U2-FkxoHkP4qg^Sw^j
zpDCN;5&0eSMn|`jre^4N)nBomP9X8rwi7+gqStZj-NZ|!VaPMr&SrrzU2kT<=dPSN
ze3I{l8wu~n7-<<VRfgGhcy}mXsSJwE^n2$bx0y%!0{F!-M3?JSef{cwd!~-~%_Z2V
z`}PR>+x5ls@J7&I>z5ei;G02zd5&UpzNKH$>ynMx(c3{^4?vqGZXW3PF6b|1R5{ZA
z!ALrNg5F|1b-_aPRYCJ8!EzU`AI4h-t?cxJ&CmT{|DeCH(sq~&5rwS*j;pEWBFlU3
z=K^kcF;GQ}7&-}q#3AP^bF4Vn<wWbe6khO2#t@g(_?E)Zpf40~SN9o@ABP2f2_$B9
z4A~s6_${Y+M}b#WMkxOc39K%U*j;^%*AJSe@dm^wF|%Sl{o|-s^t{I#DPu%mzI-{x
z25%V+gYoc)acV@P2;#S3goyDf!ZX9?OYn|Co(EB*g^5Z`H;6n~R{=(#|7T2Ay4Cbk
z$IZ+X@qP324I8<<V4BN@L(WJYGhF9+vI8f;nXf4uWb8f5wM7cOu&=;u*PQgMF7Yx-
z`NnZa#mzfAZ?4|~Ttcs5o?|dM2J?e%VxV!q5sf+bnCtE4{B@zrdnJ`C-ioluwJ!B}
z=3txO|Kgx8Z_3>n8;l#rC5};EXQkR^scoZ1+Bhx~Lu{k<#{NzWt=N!co{Mz_O!dxh
zjQK{4i#})JBCJxI@L|+B>~G^NH_nC6+MI|ru30ylmnGLBUcLOsKQ2XrV=9I|xOt&Y
zzs(^yos!YGUQDJ$(~otEa^)Q~(>EJku8HeC&vu)X>&?9}*rI;*Ob0mHwz{Z%Yjr*C
z6Slc_G$+&E)Sl`(mTkya4`Pt6R*2nXw11hdkLLavS+07<PB>eIF&0&qT*ZT)oc4Er
zK(3;hZ7#QCc|j-RQbsbOFSkwku2i0$e!70fG!(eJKV~Hh9p`OjbkdZDA{EAtGUtFO
zcClyj9KS@3$K9{)NS3<P3w442u<Og&vF3qiW4K(V7N#`G!*?~|Dqh<y7nv^csHPWr
z3%|t5p#L6_k^VS1|0n4Wrxi0gZUd^6!D?RrK()vVV%DNY6`;k7rxdMq_4%{8*;40z
zeqwY0_3q~ib4N7bYWrs%cQm-6!;K9^jz(A8HQc$dW17^<lsnCe#g8Bv*KMtgm)B-B
z26v!l`)G7wmvO`8X_UUPxTO-!CrEMVwo=cT`r%yMv8k!>^R}SoL%#c(UApZ?t7~nU
zJ(+`{HdnwxbIy-@ZX_BD%B0oY$lq6)p5l#bxf>x|+$pSbukKobdpSIE8PNG2H=E4!
zu`7eeU$k_8;(ou+mY~~uv8i+97=%<SY6Kh6>Duu9<izNf^N6@nYn?|J>zZz5FqQsQ
zIM(n~y-T$>di9xnYTTmZnPW7G57$3;zqW8M`NI7&ud}`k`kJ|Ed{}9<u{+}<$kBN9
zx^l*msyXocS?xi)8lK5}`<q`}tMh%!B>Wom1=6XW*f5`tjCg+Zo0$CY{RKd~znsyU
z&Md5NwEJtRZDU@WZ}*n_l`#oxXiL}bp99l*x~CH{Vj7pux7$5AfS(S({4IE=-CMGF
zgiH&1VQRTopl`b`1Mmqq<1YyBw)--`)c8*+Zl%PnL%()U28f}VcmH-@pqR$Hh65aX
z-VP&^JB;^SgyDZ$ih*K-lY-fRV9@LMpPK_OszwI8_%t!Z5XGP4>A~suPIVo@H^2Xg
zVUAZLc^Z?<;f`1D@Z7seFv8Z97u+z?HK(_Yp<{bTu=?fEioIYZV~=sn@w2<wc7dPO
zSXZQcltyD*yDwjGgUv5_yqe~(qyB;kuC5D>HT6VQj7EbroWmsjqE}dBS3cQQ+ML8s
zdCe7HE9Rw?X)dQZ#_{J2$HfO+NsYhynT}On|KlB4_UbG#^5KGyFk6+PMD!`5Tvjvb
zG8%K$pzIGmPCU=mWtF*?fccKAPo~erLRYHs&cI`PEOL1*F=E?d7r}VPGghukTz`1h
z8dMW<EqAeusiBbpmWjQ!#xj1nW8q^SyYQKrcJC9JM?D>P#ha(|!!mn>7#A&`S23uP
zcbw2hn0{?t1DZ`fCyO<z33v6zwk2LovxV0`5*#=5UqJz$&%Zm0Va|A=OLCMrKFC3H
zOr$8gmU7QA)J9htZ<?Bmikloa)VgiJmUeHMoi+noRq)1Gz5^%7$8nJADl^aQACT_K
zGg};^#m{W_7i6}hJPmVmt{gs|GZ)z|#k7`2UIa-(j$?sREdAHI4ER9$>J`#%<*D7h
zdz$$!zPZ<c0ymyGKr?u-cM*jy#&pl!jhLn|znUhyDJWKxY^IKlRxBl|^2`L3x&bzS
zIu5%$lGpPTtISnAna{qa;FxPj-oZ19({dMKI*+kpQQ@k=i*@T_%{6tN#j6^#%wyT7
zg<DQ@5LCH3e3i+M%m}5`Vls-xSnS0bMWz-j7PVqCg;$t)E6(vjrHGNX>s|Y@{rSoh
z&FWm{<`mfTgWI6Yy%k|K8eIgV4l(*OG^xW{%J|s3s~8Vgs0wHnD@mKFDY@ZF=fx-9
zr^hW<x;bGmp0#e<=w>f@#}#DWm*ToZZ#QlcFzO{!bN)AK=T_H#v74ywm4~J_Ro(Nl
zg?p;R+V#eF9qv0;IHrsw`@r%2KEK0wJaoTJayqjWW6iZZoO|qgelD%+^n=zw94kTn
zp!p0B7egv>!TU5@<JEXq*Y{(z|BAT!Ei^B7n(aZiN->&uJXKyFQ*V@nUdIka0y;c?
zro6pxf*0f6Y&3O4%aK@KF<yrkuDrFYQ@Ap6fmepBqWK~iUbzOKGQpga{rvl;l`@2G
z->C@s#WgJ_Im4!zOMZ3yL`5%*U5wxKd-bPlxmRUb2>Hu9J)9Y%{Wn7XZhh)HcI2BO
ze{nG$-s#ac<gfnxNLPD%i?>7m^2{zZ=hW0byc6=)WTW7ta=MRxaf&zYIQqI0{*7-b
z;@yzH-hV(r=u>{W{T#)JqK2vdu84*F(mkyi1~|^l!|Sw?cQxUjgrgJ#9l^|3{Dua(
ze7RJwqGvx0b`%~@QflUjeCU?hNb_pR77P{k-kBa2^5;L##(=glJnI?mO5j$>C@x31
zszvd31V)BDxo83PlW`aoa&plj?@-yhUraH2bjWQ$>5?&#?-?8O+%X|<%8pOMSXX(o
zd>T)ET=QM!X>G#sJl@5Lq1?fjw>VHHggg;x0j(G(swVVQOK&-1s^wV1jSNgyfO(A+
zQylZgr52_s9xb%l0W*|Watz;!bw&RZwGgzdbHC~4id(nYuCteLHH`}S0;WBCGFOat
zVXkXA8a{F$VxCeNnT^uMe8&nE6f3dtcbQP6<a<6`5$dqWmCQAdo)j0mKKg1qC!0%j
zS$!ExTo-?wL~}+STP*!u*<6{ppLb*0)8!f;GRs}6b-YSso?KYrSfT**DPr_fY6PzS
zF8=fn<VQBnF~hs(<|e06#H<m=+y|PQQ8kEH8}q{$IuQx359v~2C5Iy|p%PtV&ZBB6
z$^Ckt4=|^=da!BDb@z>~9v_$oT#fEwli2(?(OgsHR^KH_rqU~UEw(zoIbC^LoNZ!>
zpA47BR2QFS8?*^Vx{GgaZW&Y8Oz||;^Z1S|*Ch-2FxF9IYkcvWU4b0OUKB0Vy(`+e
zF8^uX2|DsL&TbxDiG0`Kt7C5NMS=MG>tFx%FH2GA(#0gv{D}H`#~1Idc<*KxD>?Nn
z-qyK9jBGZ0hf+JT6L{+dhi!?C$tZIr(oawvj_D^I9eEnka#d{+g+T7(;!)vZM^j8g
zr8uCxltSfy^1o46@0grLLs-gCRVr!=y(d?@u;jI#8AXk2I3s3K8c?(HwP^Dx*SYGM
zGgb2rRlUoAt~1!T2q^O#+gGB&4U;*1oV!sqZ#hO)Q<LNQX31x(aa|2zoFXQp*_C0e
z@{H~A4P_a{kNTGDyQwkU#+Vx!x0RR*8}4fFxFSu^T{n_>!{0mN;PrS_&{H3nq2H#=
zS5thtC&)P&_id&WBWO*pOy{g$VLV4XR6U~k4daok2Q_4gc<flG(sQ*r&wb)J;Kqyx
znMQI5Dud6~)0sxdHMcpKqA9&{Sh-%lW7wV%H(gWd>^XR`Zo0Zdx2$ajXB|Br<XOg0
zPcP;v#B*EID;DYcg>pd*p1J<w+<%E{$8%MVy9v``FWhXwxq<o>(@c2jn4D+qpzzB5
z_V~olYPU_4*nUyboJeTK|Lg0@^SG7ovY0D!iaWzB`5oT6Fo#UOAvod>bcxg9FZ4Vg
z=F<39hc5@vtr$JmclfKcXMoQ!`HrJ}N{>Y5SfEdbx9EF&k_7a185$0F@ZRAqX!Kpt
z&t?5pF<YX)B5pD6Z_O>U0Ufu@5=>LHRd`Q$ZtQ0?tLyNW-~~J>S)F_{orZsdI{d|x
z9!A*G2}Tc#!LIC?vGk#UAu9V%!!cAmw$ZsOH~+&NCv%N=!Qt-bXRGNLX+(#Ub@-Y^
zY)VTRM!JkSU-B~@)!~azD?K-77_I!ds+s-h7#E#8O|CR!-EU*qN*6e;z-V)NH@?HE
zv&%hgfH|0@TAM>AfC<W(T}+S2nCM!6RZBz43R)LWa+FUfml>VxWHlZ?B|5P0@I;ZB
zd>J(b(_FDDjq&6Rm-Bq?M`yY!QPQbp|ID&gG8SAkwR6GdHaw5Wqf{qvO{1M0=DG-N
zWn?}L^IWCZt>qC8=8F~H-KYJ`LKmC&l*~qFkwO>nBx5rcyE@VOfR-#gZg+jMHg+?2
zUljb7Y7p~*A}n(?Nm(C@<*u5G%wo2}1t#&G%owGb|Is|GQae$z#C^L)%0EoO8b#;6
z#k79BE7!;xoRedb;D%n3b!A{?nMARin8b~Saf88CTiLVa>Fh?O{^5@kvC%dIeNJ<o
zR<Tn*!4?(!5nXq)wYR#q<6~*;<ZZS~&AC%*hpz{W<qL1j-psUIn9^;Wm2?7~>6#$Q
zXbrO5Pws3IknI|OnkkNZcGpCI_s3DlbroE~%>wsIG{Q(ho@0qxcy845f7k!@oY)Io
zrgNx<q;*iCD!PU*0db+$4;nA9xMC&sUYRlCQHlG)X;w~|owq>gs$NYqau^Of?)bba
z%G}s6OET}Z;+QJVOW5>m<GA1`q-Q`>s9^5c*mRYON%xGr+iv^8b68a68i>;oU!8Gf
ztTryb*y%McUwXkZu5N1`!8ozwNl~2}DZCj%r@~x4>tBD@hwS(K`j{%<+LM+G-2F7T
zGAU=qw^Y)_jp|J`sxg*ra;>m%{bpR(FUrIjZ2M-%KK+fy;D(xR#NXkTDrqJmzA)!v
zeY&2m<?gr|e7x4X^?TQa^JFOzt@`<)F+!$Bt<8l-t>!N4p8L5nnS<}X`#FbCgW!SK
zGfODnu6gJJdD?dok6caTtS_j?9o|;bJMS^3a#T&?i5nEO73YH;W|SkS4vhL}>I{Tj
zgMYArwKi9aJ6)T`&{tLxA}$K$g;;vribc1pHSY}5a12jXHs{x5y3OPclDdFi$M7OL
zT){J!h7v(+%Sk+U`C9EFPukVeE#)gY$-ESsJT%{gS89q)=C(6;@d}|oW}43_geAA^
z<jpwuiw}+CH}}h&kE7FHZa7(S`o$Za{wh7)T-|!J(^rMR&+dQA#pYE*v-Nws(_8uD
zDM7u{>8%GWbJ{fY>GW3I`R~u=iF&8M&?m0GnZA|-`dya|r(ok{t)C<C_6hq0{Z;Hn
zqX^~t?Sf-BQE5QWwJK|Y5rr_&ar}un_86p}bcE<R+8^BMZ<0OLft>Z4OY%b;TOS#X
z6wSkiy1LHb8aB*j`)9NB9<CbkjGdcu6;BC@+lY~_-hUd3Q7((M8~~%0g?ZY+Oh#i|
zj*ECEhOv$Tb24r2q>NKmvpFKEbk`4h-lv#7!L<O*mS|Lhi3;TvICj#cPG8cZ_aM&E
zlhrFDhEBm07ZjVo6Gu$zyk%V@a2evEn<onBbbE%_o6A#l)95o@k$l(BIDnef>1$<Z
z0%{z*Q1dd|Wlsr;hk#Kokx`(o-iWy_0vm+O%{<ozd>rFE=6CuMn0FlJiM?3p8gaqG
z`B>yQrZdS*8tWM0MI0|ypHMX6GVT(6n4Ey6uGpjv=Ed%1F4P=7E?483w;XxZ-LV|c
zt6MyNkLmOUKqHWOzIB=ZDwSol0Ep}K#kiJJw1zaruW^ZZvdiJbbX*DIUEcGkYox(h
zf@{P1v7Fky>$8c<nU9zAew=F$K77R2FS%#t+p}(<(9@2mz1pY>aR0zv@Fth1)6!v!
zqxlArJcf5|FqH#noBOqrmzGi;qaV?A*9TEZcU9yKA$oD+x1K3JH$~HbNtUZFZ~XHM
z)~l-78h3yE>uBV-ipYTgEp=&6r!x5=@(Kj<T$$#s5KmU~T|MV>o#T-b3RGlrJRK4i
zy6ChZ<2`W{b$a4|)O-}X-k!nxYMW8wqEffBlizx&YYtv4ru1sQ#=~O9OslM1nH!hW
zjnH;XIeOOB+&-b)Rma+j(RC6kT(O?BNmRNvOQxL+FAz|7?>M3P0B?k$N~M}xYMgzm
z?LhRdHPpB=KIJUKvs6>bS~Y1RW%)Xn*Yvnp^M%&C@{E@SZ?+4#IOc)_r>X`=(2Kli
zbOhsGo9f^u@nAgnB;mRX<^4c!4==*Z)k(=Ip3>rm8#jDx0JmJ1aXZ4HxdOKxg}$M^
z!FoppaAEl4UvO8+spd0U*jCqq3%xU_HdlP|dM>F+xaYbh(fAI&@4{PXBDjLyZZtqV
zH~$ZXc@7kdN3Ln85vF>DSv__n;|9TOit)r{N6nHEtAc9n4XfzjR$RwXtmNu1u017S
zBO;Ee<n_E$(5;x}ekFJ8E{rm09D0>7c{AncO?c+2>*<g2+?CH$10K}jh014R;HArG
zj@fO!(ofdLGa&z2b=|O@*C~EcqCX8Cjb8;<#;wI~E^ltR(l}ZX_E-8un!)l2<_$;S
zodNSY?VDkLf={OS!pVf6>|0@PA)jZiHN7oN$(Ya3g#E1oHF12ZnM2952Iv#^mm=D$
zQe5ln^7IYRlkslY7cr8JW0QViPjI011#f-z*AHIF;wL&lKQ?&gckhM$N!OfBm<v=G
z=n7*eaUVD+>`NNP@HdHv4TD`Bqc?0a_8LQ!B?sfb{0oMLec@*Mv;`O@ws_8KoGD_s
zE05|k&fPrh8lm{nRGIVB8tF<mG6{F?VNVwNgxmhnu7;c*>9^K#!6OdZtzoRoo2#(7
zf{~7KF1HnIIC`%M`zxc_Y<!i&-FpKjD3X~#*{6Fj(PeCI_;HX=QWx=&X-WaY#=jk=
zxR&{VzT()om=^Zb=td`O-cn15@eYG9BkWD8Yp9ifT?bzH!K~N(7+sPMh-NDjwk8h}
zQljZegR?+1qFnXnY@ohq9mmUDN8-X}{9jN(?+Tc;o-Uf_i(T*Fb)hTFv=1A=Tu4}?
z)?Z-EscAOvGBjS{yyazya%ijPXkRL9+#q9_@LFSsvRs)=k2ALRSm7F-?>epG5xm-y
z1E2Sr%pvtE^=VvUBI4Z0`D6puh<@|vYQ(!dC>EJbysK8Sxd@)90;$h36G@U-p;2ry
zQWV0*n?ci^jjk&x{qY<On^eTszy1YVT+9!6uDaFr<pNJ96pL-HEqE-z8#J7FQdJOb
zKDd1s=kr!`r+`d#@cT27<#OfaH)Oj!j6I+63xOPQuz`xJT*uv9^M2`S<hk!Z!D#c0
z(9R}|LB5#ND#_!Lt0-_8^WG(`i1<Aein-17{D~rQ;mM-uC|1a<*!V=02-AW;1*NY1
zkGN2faM(4%)OfmA;N65WH)eQ$*)yW%0((qVSRGB1c(nqjA!>6gTz=Cj#dAwiDK>cb
z<s#*U(~7$w&ZsA<#7DBZ7*wqQV+g{tcJ;gw68J?nq1KIrd1lQ)o%_vWNKUx*j^}6&
z2lM1*Kv~lzMl>2+JGCA4)~AbTbQ$m#E_XUjs?B-|1J~UsGZGD&^^@*bxJ>f8yV&HH
z!g=79tLr?n!SQh0rKMyagFCL_coUX>n7I(~xLw0!_B6D*;CV*HlI*EN+rqc3BN$^M
z?|ICjxO?w)#^(K!`;L7s-W>1`+;84OOQa>8xSYG-<M~l|<O1nQH+C%^yWbSZ*i=u{
zR0&228gw<HaLL9)$hE9q42wpm`2S=kB95sAJnlxf%hz+bhNo@{V_Q(;8p9QqRq3?@
zXT>@^b1CQuH<1s{xYD_3n{n{M_1tIX<sdGFc<DY5N;<Jg{A46_zH({TZJ^_VSp4ih
z!s&D8;ulrYb1CwxTIXhFa>{4;&GBFs@6UjUw{r1d)`*dBMEs@QyQRT1_K3e$>T_YI
zzZLO&%u_7z=`V8+;%%35-D(=ia2KzP>6_fR!RzDl;hJk46YN3Xh`+{~2U~a)hId^g
zs+;EU=@;(j9f_-YiucN2e@F6_As7(xWCfmAtlwZ7?LC#v32_w<`Ua{Xu6$Q0uo-i|
zL5^8oi>HSq40gQCi=zV#I(r=w@x+xl`m(~%h<{K_PK|qHS>+CMj2K_Ko_*W!h_`Z&
zZa#w%%7B*^xWL;+rbgvMn$<b(M*hQp!f5eMUZ)$&VT_6R+K-{MomVF2!@#(RFH~`f
z<dU+Ewp8O?Rx7A@pYS>c?>h2&b?h2UbWFPY6qA(c7Gq>IS*5W`v}T{;n&sno8idAS
zTEtt!r*Fm#amjhW^QwcHj%)MuXd-4gMpmy#z-*N;GnQjI$}x~kKlF4Rv>9_<6+c{o
zd2Tpyc(aeGJ>npnul&sYbPmgfD$N*ca`^C`yDQBY7%Wz4+!y6xi8_EUTw<w<l*~=v
zGDYG&_-HJ5Ax4S2!u^`c$0}kXzDP;arD&{jtj?K_IK||Lw3a5@Yh27lJeotiYY{V#
z#v{Q6x0Q2BPjo-&5GDyp?x$z0nc{x(>K+}^xxAuR@aZ>fa>H{DZx@^G%obH=(`r7v
zyY+X&({orz!xT2*HkS!C@0^HJRfbguAYD1oy%<&Tu6TaspCZeR19N1|tg>Br92FbL
ziTIi@?-apvdEo7Yb6vc)!!&}(Q{(A{g?!h{qeuS(3f!+16m1G!fo%u5VqHa%t1wUR
z%rPN~UDofL5mBP_biQwvvr-p*-uw^eOr8JFIPCJD&5c2sYl@9GkMbPpm_p67PHdQR
z7dnjs?PgTCKL2D6DqSG`v~xi`t!ASGEWR+~c$!L88mjfvm}(mb8a1yg#mj~i%xlFZ
zCqK{GAxCSS%X|*yW}ciY^VKW2(3g$4VB=ECXn37hDP@FIG`g>fFHRXDx)x2!eg-9C
zD(@(~;rjnnpm|uX*%e5Q56Auu{q~gF#udt~h`SYFGUZ{gXCG&-SDX8gxAk=`rTy9+
z*W|1^EdgS1R}9nRHhZM?_YCur1ln9%4dsZ(Jr~XsaN`#2z8l0;%`t&W#sgOjSH8{0
z!QMlc5udwB#v?Vv)RYgog6dWH$C^>hLPLWdW8!M88=trmDQK7tX;A#GjN8C_bRmV(
z<$DYr+wcSHjQG1IyK&&N*@!r9*71UU3?1=zs{v`79?N7gc<LIMCr~_*=yj#e{wJRH
z(m}1dn1@<)V&?|wycr+ig-gmcCUzxWIu582q0zEgVem@HDg184&q{8rbvU|yammeX
zH0tUdudJXqi~i<{i>60nboom?my`9hRez()Tf6uL;mt085~PEbM7-7IElHGbxgx)<
zA6&0yeS~+q{CzJCn9xT#Y@vkKx69wX^Q?kfw0GTa9$tN#f_`28;y*8LGu?~PR9^Y=
zr42?O(B-eVWX^Nn?cIEO&!tMC(<(Xy#y~M3zn~bTJZM4AkACp)D!4I$%Gx1ax6I+a
zXFxNw>-A#FdU+zuFjs3HcCEy4mp-1?>p3JwIKosO!tjE}$k&8;YTbDo&b7_ZzhJau
zR_|ETSDhGG&HW?BigByW9j9O1aGJ*f#&`Kb7v&(H8Bx5N;M&jF)bM`0ntB?w%#)NW
z{d*^1va7;0UWmq&*FVkC(=;Xe!yo@EW~e~V!*VTVx_I-9UxZm*{@9<hI^|P3J2$nT
ztrYC{1dhEZ{hC2z2TpZ!9e3QC^4NQxYh4~jnd@yFH}hR%@ks=u%UCFG%)N?5g3Now
z+Jx5sg`Gs}_^w+9q+qE)G?odh#&Q8Vp%I{_U4X`J0z51ih{qZM+O-SBAwggz5(PFO
zNnjmP1QM}P;8Sc8ScNSDqp?+B1-1$B6;^>!NEi4FnF1S;C9oFR0{=U51o%dazzF0C
z{5|po#-Ko8A_@gQM3KM(6nA+d87HF>k(Q!VU@Q&`Oh=i(P#hDOigKH*1{ESLM5VxA
za9ZH6s1o=GR15q8H3FMZYtsi%C(^&7-jXgNAkr2zSkhHAiu5Nm3A~T%0&~zTFb_8b
z*5H<saap@9(hS@Y_`h*iU?f@vqR?hzr{kVT({Nv4J|5VxG&~gP|G^`H{|%1?hT)08
zzac0v7a@WF1D&=4wTOr`6Wx|{1y4o#KhY~N4$lN8;JG4krG6pOM|deP3$JVj4ft84
z<@iNlJbo4U0KW<RUqH7v>_5R9-RkH%yeZOTyk$v^cw2}TFz+bvzoL)8zo4(cBD`y(
zUqe5UK1P3mNf;n78}AAH83S$X3m9ZG--E#x6=8@Cy@a7k@Q)ZK@Lw?8hMmU<A?~?G
zTH0ld65<LxS_%FcV{Ehw7;6J}V4Ov}Fy4lq#RMhzCrlJrhDkQ89Fs*Fjwv?bc1#m#
zD`p5R#!P|#j9CIxFxzI(h^TIFxeKL!j=2i`JIu3rWnsRJN4MBQZS`2BXiKoz23BH;
zMP*nj#Mcy-`56~txl;TmtgxYZh!Nr~*i}N!R}d%Eb`)zY?F{1mTn->X#IG`uXd@ON
z$)Y1jv8V(a{UoK>WRnzQiw!QtR*UknO{nE0QY|`xbRoJk&lJ<!kR=i=p|bs)4<W}A
zw<Fi0-N+MaD?`4}SNl<5X@^iKMDKJ(mbMqg78Rnzh8{<$MMrR0h&JG579GSf8+sb$
zLT!gpA$0R7DlP4EoVMr~s%&T@s)cxHR%2;Yu~~EwbwVxYQEzDn5U^-38vLpsMx%&r
z2hk+-)oxrDqW6Pl8+sf!EGot=8!rd9Eh@wvp_XH~Yf&{?EviMEP}_dov#1>RZRk!s
z5TZkjheFMl@JQ(P0X!DEnTaQsb`U|Kwi5^mwbY^0()J-@X_wJ0bn_gZ3N_cD*T$>F
zGoh~z;JKxp#0#O80A5;Dfmb$O34ZpQZ6|&avAGPt3bmcZZ#KyppvRl$S}x)Zi_YQA
z9{Xt)<1HVS;cbOC*Ww+ao9EF-s4WY9ZA9~=fhAU;pU_ta(O;;g2m>s-g7<8^a~No8
zRT$)#SAoGI-rS8LHewNmT2zf;Hc1+W3*Ak_2umx)NTKEij1s!}IYtZJNW&P5axvD1
zHej4k>v@d#Ym$x$B0fBci9#)@m}G<Z<zuo<U4|(_Ulm}Q4bCjW48JZpnCZh~m}Oyh
z0cKm6Re&g)|9;FBYNa~^8=Rhw`8GHc3oXjSBB7flSS)m7FP7NQ?O1AQ=dsK$I}gik
zk~3H#)Or;$LTv}IN~rlX;%w+;tP%P$6Y)adW+K7Quoj7eKV7?uB!#zRBgLk^f{ix#
z$~9~f@#a-*5&AM6TP-ac+bk^&sW$WUG^C5zmWfQE8)uLu)Or!wLd_?UBXlzjxk7E(
z$g`oxkuUV{JPK^QD=4&~*H9$%?RFIVwYq>35nD=7YJ;<I*rFVi34NJ?V?xb!DEAZZ
zLWPK}ji|K27jRnW<`Gm0eVK}Cp>MNMBlPebYK2-GQ0F&82I@s@*^PkEe`KS<CaFZD
z4gDNV7G>bN5Iw&)``J|ChKM&$;FeJ9CEOPJ;sEX_Vn;?J?uyuU3avIGr$w7jVJYs3
z*iwW0LihIKfl%{VJQVtJJ04luWjywo;KV56!_x=~eRBXIq1FrNw4oOevFxN(qg%uq
zNAT1}tV6HRce!}x^L+u&Mf`RLUI=}&4=;tj+=*90t>^HwWv3_`zgQxD(2Ceql8N6$
zY(51%)!dklH=b%n%-V}L1%GPH#ajyhk9@o>)LMmiY|PYs=p*98%jjz}+JSd%aC$2G
ziP(G|{cUhD1_<5QgZG4P7Gt1A1sEjMd=-OjHid;4BI1|180r^Yf?*=w_#DIi0=Hv?
zh%H$dDRjFSqlCUVh|xmd6k?1;l^84ZWfsQy*<@k7h&T3Qg3!aWm?-pZ8YWrVDNMHU
zYB0rTEFIG<%*w|M3-|5COhx^o5VI_CXE|nz_)RIIY)<)@EA;hI%oDnqi}^NQJ{Br*
z@ZxSP66{F7jKvCnScfGdhRaW4sSVFKgJlZ;>IjzmiOR7;Fm&V;Vk}HOhgFK&d=YV$
zn4OI^B7U2Wct24k5(GO^(~+p~d-+HbF<9P&6cHa*VWZF;PLT?3&&bCX5nIn<tI(G_
zuuZ|8fwM>z@teI!_tO+3Q!tdCjV!^AjBI4voT`!I6Hg&m#I`-i6Z$p-`F`>pD6rWd
zMxjNgQKaC`Kq`twY%N2Hh`~TDO8wlg;IIupjxvk(;+RE+C>Q#MN4G)`Yf)+QIk6k3
z{Vb28%4V6CiE2f?a|Ja*Z8@k_a43Bz>TGZ$>TQ~g(+Jq`)O<8pYWg`eT59%TG+D^}
zuPcKuccWS8=3(6M6CJ}X5nK1+w$K-+aYw<O4Gp+!gG<or=bnl-!C-j}?%A9&GH~BU
zC~Ck1!SLC9JX8kFRd^)yuo{nT%>A4-Mf^4oL7{JsA*A5Q{uAgF>`2>>h+uo*1iEe1
zGCUPAeEvLoZPe5>JX3hv4m`KS^j&x%7%AF;mkMv$i&r*0V<&#LnSPF6gubrFuQsD}
z{3g`;Iq;)q&L+Ge)Lei!Evm*_LXS@2ZA<$c@BFA4w&X1O2u4b)(AQEk_u*Z^VBi4y
zDb-hp&|m0oJqGx}XYrm5K8%4PcBGxbAi>DK8VnX}&*0=K7(U2nu$8Ly6o%PUsi_z)
z;=ku%gwR9!*c56m#3-Aj6r+_Sl$wDtHgy`OPfJZJ!#JOM1mkT^moP!-K`JKN;Ph*l
zBx1`oO!m{9#uS_8IHvhAFJXpYq^Jfn1%JA{53>~YP6lQR-DpCTO_XsKa|J_bb(kj@
zJhvP3Ep<l$7An=v<5(o(PmPo?MZ8^tB{osrJ}k9~cAmg8pLz|;1v@ehV}*t3pCd-F
z{qk|F@~HvD2}X*~VU1w#p*qA1hIVa7f?!u+4H5-A(z*E->`D6^DT2Yk4r~;U&6U_B
z^ld7(*x<{l*eVz~v=`eH-jah<p>GZ$UBRK$24o6$6;>ik;a}z<Tf}hrDdY(L)VLG5
zg2BdJ$P?_@w*&ct9cjBzAlMn$jY8$sT8<(S!)Gf{EaI&il!zE9-HB2WzYgHAh(9%E
zqfD?PqYB3q{^%UaMGU3xL4_Z_50!$!@=BbxFp!2SrTStos%_M?1E>-4oBgO2u``f^
zI>AtCF6u4JK8b)}`{lD}u&GX>QK+>EO+wA(xUPu7@@r@o?Bo!-VX1YMxFy(;eg?Od
z>gEO9vBb0{+_h0JrlD2g|L;+>Sz`G<+_SKv3HJrVHGA+tQNKBYhaz?#&BY_Zj<j=l
ztnfQ&cw!69M^LEcFhVwBdJZ~$coq?bw;n{di0v7iPX#*zJJD;Srco00Q&AGNsp|1U
z#Bd-TF9o~ucHxy^DDx10wya#nFE)Kz7Jl_%1%49@=QaSn-V7P6<D}Z_O|Q-AcvHmi
z<vP42;y2}ZTj=XEct_~wHS|%$P%5QSn^Ac=-nHR-(9aLwi~b_s&A<Sm8(DZy=-Ztb
zsE8dL9D{tg8-oSgGmc`2O_YkEA_mL3T^H;sJc8kZp`B$Ip{Q*;G18A3z$hD@Rfy3F
zZ#{xBHau_=V+GqYN-$0^lAVw7g5l~COz`8Lz(i%xT#rdY-yFnb1$U&K#1s)9G-8^c
z<|t+eb_PykreL@r7qe{oqnK@prIm;hv9$nm{q&qxE%5;6ix^6~h=qRmc`Q=+7e}$!
zh8Lz`iC_nXnWa8Fjb)13+=%5uU!TAV8%!xNM#KjJtP;Ap3vm{m#Tq{!`VtTfrPU)r
z;cZn&^ocb{5{wjGMv7oCKz*j7Ht)wK5qr*1pDEaWWe2t@y!8;a34L`OsS55$KaO<4
zaA^%P1tX;tO%=7J5ZNMzGLIq0PgQ|j3)8696bx^#LcU;o!%-Aim_f~^^7^6)MIr_-
zXQSAr+R=a#!N`SuD7Cy6<FJj&?k%%X^LOKzPd$cm3+u{JVN+G0Qp8XeHJpMS>A9#<
z_%{Wp7HUgFjnJ*5s8z(yz%JB@_}>noUg*ng1Z<XtjcBl0W*kSOrDm6)Nl}|m;JS!i
z+&DA~wqM?X8#YziKHO5&Z}V~6C+@)=!QhD<xU2BCGibHx8}rd-qh@5`o?vI70rxGu
zd=w7^!<P=>p&vH~kNgO`@YqH;yB|;d2xSNg22Z3SBpA7P37vj~B18m3JG0R(_)}vV
zp4vJzq1RUY9G)q-y)hNfEinTxY))MMUyAsk2Cqbn96FDmEiBH%FE;&Eim8^mzaGB{
z1_S$nzj-t17nOM9Z{GCUk+B1B3jP$>gSQlZCkJoa@bpT&BN*PjAAJP7Ph3M^!C-kU
z-nDT#8}{?ze)Ja%U%r3=7B+IDFW7#q2m@8vy@MDeVzB%o28;Nx0YijZ3NX|rudc^1
z%U}V9i`bEV5hDacJ9!XfsTrvlB^V6s!)O~JoARoK+$oQ>5z0?voM2Z~6~<d?T@5B!
zUg|N?rYg$BB*E~JJWLks$T)&2iuy%9rU^AyV1|Obn({DHu=7eiW(l@ulw!7`w&Wv9
z=)on-_4C|?d4fF|m6)&an^{;W^zZptB=k)&7As<~{2Z3p<Z0<xs_^C}EEBQ4;W(E2
z;YYE;!i*D$QPh@8SS4ck#0A6&cAYqkHG-i76^OUF9z%kNomWmEQ80LUCz33zs6mQg
znC-jKQcv#2Cc%!3G;C2`Uth#l8+GSyY!mF+u@|Wde~^xJpO}YC8#Q-7vIK*HGsqSU
zrPd%vup{j_a&6q4GUQp9cLe#$_D(GdMC`g+jY1n<y%R-(9qCml7K~)?MTua0pd6)w
z!MbxeEZEslg)+fVQ7VqvxETjg?!$eku;n(P(nigu#A@qUhbj@nd(NTShefEdFz*^_
z1v?vZQ70HGDnPwpSAG@(g55h#qrt|#b`*^kUOA2?!C>GbuB+T{D$p!qJGTNiY<PJE
zZV86c0=TWHchBLDh#hG;xGUIQL7`Q!{aOXu1cTL-S}pa;cHFnr6KQy$ylxfXq0k*Z
z1*+h1ZV?^}hDtxj6Pu`x%2UDKL&XRQ{<y0goq`bxCK2WJ?H+VnUir$G(0}Bj*XC4y
z8P5beGWO!RO<q%n7lQ2#*YMKD4HV*)jhoJsbiq&><yOJ)1x~S+nz|dm`Ek>MXWqmb
zIeQ#$2!@Kv@a8jb;(d4wZ;2QT9LC#%;jHa=N8v5G=p*#+h3IRO7gph2!Op9t=w}n<
zXQRJh`_%vj2zI1h!+V0EqDBm~5w6u^kfr7vz+l0jt{%n^8{x_>3{|<`7GRi&;e&@U
zT(Gkt3nK(0J5w>zr|!Zi!S17#7%dnKT)-H?aCR2P3Wm}yVVp|1cMju)no}{s#@tTf
z)rXasr0|C)Fj>T&9l4kyV(Upv6ERqxjv0a-8GA6(QUjToB^XM(g4uqAV~7&$$;iW8
z!N~4v%(HR#@4<Y*@TJ38XsOw!ut=~wbvqUd_M}oVU}45pEVXen%dyPD1H7wi;l5m~
z5bQ`RM~q-QH^{35Bj=AGPF4G+2y28|4k2FXznwyYA_gz#B2h4$O(E7&tFIwNQSYQ<
zqYXcI8Jh%qbIP$ru=DCQY_+K>u3?*C2NkHPg6-UQr&|X0AXCIpb}_O9dpRR!3kEAH
zkt5g{C_=7MeX$RDLSLUlzJl8+<rUaObw^O>Q}?3CQcvtcu^-_gN(3W^OHnEq-kpZS
zHbNkPGQsZs<v3>J9y)<?!Jb3MP$AfnwilJ+=Egppw#1CjQ6<=Y?EtC;drzjKMle`M
z%UK)Y=zi1*b{)BZdP`-G2Q2ke0~&07uc1-sVFj8L9HGMFx`=l!qgm+7eYm0EaDE<c
z2?hf-xNWK1GjPWyD#Tq8f2^oNtKd^cZ4->tUcf!Uu0l$(g55di@j$TsDs`<k!uczB
zBp52%i^rC_lm4MpSnGZSg_;8hDL8m84V{AR8Pv8a{NWjN+whB7cq$m$SB72-bF%PE
zup@OBo?Doai5G%h6og+2hI1<M%EF3t{A?qX9>Fgv?2B6bYQy(j!f%#?8sH~ymhCOu
zhc^Txhxg-63l9eHmW4HY@wQ-i&vCr-lQ#{2yBB>#>?&_SU%_BqHQp8M-Is}eHr2r<
z^cM_epT+>eNOlF@v+z(J1_}o2YB5N#BQ*zu1;g8SVu)bZ0V-ZC^;!mo2?ndnG2FuP
zQj8GnuIBs1g6&r?VU(p_I)~AI-2E6M7|K3@v4VfwSBi0hk<wI*SB2YlVS**@+=YoE
zeti{_MEt4o2qp`L5AVVh!S)7DwSpb#S1`jyxYB@`KJ_GK33diJ)e83R;t`hMkGZLs
zE7*Q53-bg+sYfwiu;(BpS{s31<|4tagVk6p9$SxMiHIHPN3m2e+)#mKf<5~#VYy(W
z_y|_mxSZl+1lzChq+jr9O)BC9I~xkIMkPGTLcEAS9xX+JU`NJDBntNI*o`DVRUT3V
z!+Z8(qYrDbNqIdCV2g-B>J_&NcBJvZ-$pIng;YhoaTe(|YV~Pk3icl2Q@(<oSIdws
z*mH6_as)%E1<1AVzzO6DhRaKkZ?ip#0-<{)C{%F!mCsRRV;(<)Vhax*M2TRiunwhy
z!Rl%pwh{JluC>&YJ8(>}{b~)$1w+~CsIYKf6)FWIRY!5!GIJGGHvC9Ess+2w>_Ux&
z+=SH%_Hcr&vvJE(Q7;%iP>F!x-}V)wL9nx79~y0hv&CpqVGl0jx+Pxy9L*xO?7|HZ
zL&e*1%Z8t8z-_@u>Pg%Y?7VygcLh6WjnQf$4>Q_qgsZ!8PcT?_0QXhGca&*G45v2Y
zp@n<O@kp>eqX3U>D$e3h1bYq~L(oz$?m$Q|R8)^nORcU!M6inzM7Ln~>BD$xBh&@Z
zE7;2w`kA=-_X0c@F_L)@F9buSg?MSH2h#CMu>Hzz{45x(JC9$K!Pj;8Rm6_;^Y~4$
zGf)aV_a@wMelFe+>^<IyHw8Pd9l~4By{WZjAKn(R{mNdvV_{<r`Uv(MDnMVsj`TXb
zE7-{{>n9iqY)5~=F3#fv1cQNWyr&W#@`{~}nz0>&o_kjI&ku3l8*JlN?86Yjc5V%a
zT6pCUh6#pivM^jQLa}dz;E#I_V5Ev@J%Ldoh6-s&D%imt`xwD6zml<*%Cm@Zg29Sv
zjQ6Q|n4r9FG-9F+FQ;DC!T=Ss3cr6IQ~dBMOcU%qOoIWzNZ~civ=O)@%o6O`Ux(R(
zKUJk6%EmoJr*nesmoqR=Fnn?s=GzE77+7ejTp1Tx>b?V5EZCiQ2}^8*?EP43Ar(@~
z1lu!?VYy&O#yPC;sRs}v*mZI*RtZLmPaw|1+ACNi7|!2`c){S+3M5z<s6?V#<O?2o
z*{FM{MG)-B*o}>X?WqB5vW<HRTWr)zC$QCqUp<d)f?Y>WAk{|Leg^3lmenECMku_3
zEDH}bAzLt1oP`|0?#g213Wl5ZBTuj+{VMVWBRq>K5NuDaLZM*q{(KY(1}kb&Y$IIe
zoGTdKegUPn5=U{^hF`saGQkj44#xyLG7g|zFjBSy6@owR%0{JN`_(#}R$dP-qDsVY
zc@e4wLkG%GBiNC?2epDdMW3TiFc>I9y<nv5C<21v?0PiV2$$2)XsJ0gNEPg%s^hw)
zo}fvpSh)8&ZV0s;z%2!bGC#*{%V9n4i1=d;1za0mn2%NqIqcd5Be^?pPq2GuE$&-*
ztq>0cdpM&$wA6F7O%)97JcGwJ!m&y`5p2K8*G2^+d9+Uz?9D7ir(h_HhFF3fX*>ZF
z3}4%ir{er>EqX=lIa`QlwzkLdT*SyhY7zvyDtF<f;8UK(zOt#PAN^S{c(oC~D1)1)
z@T(=(mEbqQF!!DK@7|2tRmvGxFub=CZwhv#amKYUJri#WcGq0OJAyqG2hc|_bl@EN
z+PG=Oc=x|c6_Ja6B1R}~_qS2Aw_|{Xg~fPJFr0k|0|mP(_F#~OJ9l8PU`KjAhS<2f
z_F<@nd+1hFus8EOhO69LoNz@9?W@2@8(vh0QG)Fm)fg?<)pQDD1cMZD#tL?yKZ<dJ
z;T+l*2!^<_O%VKXPbMY`KD|(eNrK(ic3`qt___&GMC>ZrfoT@*yo?!w?G0x!Q?O^>
zHOvz1-gO?c{Rk%zB^azbj=9RV?KI|DVnqP+1;g8E)9+Kuu}Cn&v);vmJqOCMM6mZz
zBbM5@C)2P@u>D8@mJ3E|0$3r~#nBcc*h^LFD#7sX48#exH_*yJu!kq(@#67jITA#S
zlpaE&V6gfKk`(^MaioYC-n|bSMZA-WO*T<kBevM^rc2l=*jtu~ZGxfV2BZph(BLFp
zFmg5*nS$Nd4kAmi=RiKP1;gcMkR$lFoB(nKLpjHgXCssrAYZU&2Pa&gN`-=8q>#sA
zf?Z{4D7NtIPLx=9;R;FxL)ncuEZ9wLYni2HUBofL-mF}d3x=t2sIV}f2V;WW6}xa+
zFw&5ZDz(!$X{feQs}G~b!rdjP75tI?TPN6iFdy}ndZ7*h!BEaVGzfN|<l&fLM`|gW
z1iK0^;ksb3f~s2^fro=PEKJ{xTY^8P7vr{#do>+*1iLbi;;z_jIfz!FyC={lbniUw
zDPn|6=Y1P|91lbc?c0xsK0Jj-HtM0Hcq|yKIEN>K?N`$f6bv`zA|%*(r3jsZU03%a
zA{eZ$MYmvRX9k`M{_Q{^dd0$b`|wP}p5kkGF4&%W9xntV*_ZHAFqC};uLPf7tisQN
z!K=sci(q@|0sLy?76tH|O8B}8c;U^sq5Zq?hG2N_RlI3o&3U{f*weHdZ(C~qLA+yO
z`5yFn;mxhhY3M6rcO@?f*zoPy=qDJWa;v{!J5P58`0xzg6AV_;NtA`v7coe%BkdFh
zTgYw25F6ogCWiWPY0l-xt-){$)2lH;jD39^BSma)%)}@go>7j`g5m7}jIpqx6l0a@
zZYIWw*t@F?;|05}WMYC~r0_B(3btRZ#U#P-Zk~_{b_Vh=MX<Xj1JeXU`wn4-U`IwB
zW(v0RJZhF;S0QIy!Om;dh!PB+4PdTd=)iHz6O7bnW4>TlRRtD`^E<g%B;t?tsaP!7
ze&ryR2!>Ldu++k$S}YR`U)_i0g2B2ASRvTEzXmZD9?HZj!H%?A#93-_8rBGg3d<2M
z7`{}C1i>Dve-Z`TQ!|ky`1Imgq}aICm$6YWQhFMjeCmE|5p1uc9fDv-+68PA>^)2=
zSFoq)0MZ46mvWFP7}=ALEVamuL&z2}y!|Y4Eab<LE7)6Biaa}5QjsrWXJZZuMEsJE
z4~ZD0l~0j~H)~KVVn=!gN(8%ThFB`te)$*<3wBZ`Stc0DIe}x!;4bf3*{D}&DQm-v
zc+*O-iz<xMf}NKSp-QmlAWda0ucuHWV*BBG)CzXfXs1pvc(n-ig1sd*2ndE7%g|sW
zlpjH(4=<s~M#$fb>w@jqPM}$^Cv_)o2tK`V5w`?8(ra;B+}z2>9T7XPq~We$xGWv5
z3jd}AZ6dbQ%=(^%jfJ=`7_8fl2bRhV9$G5r%SVE}nPqrvVR1R02zFlHj-X0tE=9;E
z?nI|xuzWirf}wqd=oXCZr;V4TrdFa?u=h|Zp4r-5!gCROXy@`mu<KF@UJ8bDj^dSt
zM|fCf;rYY(MX+c4Zu~0v<2g>ag56X=;D2~CZtybi90*40PT)<!a5e872zHef;%yt@
z0xc43gmPZVvXBbBzJl#EWPMjKw66sHY=k|#&|k3YOc@6J4{xGwt;BmGM$V^WpkT24
zI0gwmt>xtd!4NlhL;MJ5F;uYgN)?6)Mk<eDxL^-8uOn<+T4;^5)Pq+r%2F#%W3*WK
zx*TIf?BH4JSi#8Ia*PuUQzbm!QnQON!NOBi<62mD8IzRPjnkMcVprZyOc4wQE@PTt
zI|s-N!On(q%oGfNo`zY1-8;`=wqWpTCZYs8(obQojhlS{^8|Z}s8+DeScHWl{#ZqK
zIfC8C&S0@%@6j485sZ|b#!?%BI<94wTAG37f?ZekV1;1tL^@&wJJNSxm0<VPD#Qtf
z>+-Nh@W+$ch!+gy79&A0a;^}Gf<38aND{k`^N}KAFmMSQZPePM*d*AMSA#8roq-H&
zwUx}mHW9=7vXCkmJjWw5!AL1>X$5<VFCfcy$xdX8*nP49If7kh4j@<IUsJ>t@o)R;
zkS`eC!}(S)$lDu*g55`{$F-102F1#tnWhOgRrW!Y3U*zp#9>=NEy_e}5AfM_!Oq4*
zD7UFt?+U?SIVW7f4w^2U7K{YWp-Qmx$`w=#b{*j51i>KJ{8|fXr&T8yX3N#vxToq7
z@TrH;AlSRV3XK+~*PuzTJ+%<m1w;EO;R=S&W#NW}dw9{+!iGJ#ZQ~wD#T^T?kK?Xj
zclA-U3Puj{T9@q-Uga0D{puCm7xB(6JP@&yi^4;}j#SRLg5lilcq|yn+=C~Uns)_3
zOReSIF2PWs8l8e2yp9pEakCDfTd=!+JDv&#tIwfVu=7d|o(YBuv+-Ooa=aQZY}`u)
zcxfZ#?8hs?A1VL+Z2Ou|pNZH`b=t2Y{`(pHCSngoHoTOK%kTIN!5=>_#hWjsMLm+c
z8*hmmq1NDS3ojSpotNH(`5+U0L~Ktzi@t&(S{=Nr@Vn{gCt?Q;R{INfQl1(h7_Q!d
z_bhC@h=GFL=gwk~g}kdVSg^Bc7lsJ7U#`MX!JeI&7$z9rz6-+zL&ZBWLM${>jVts8
z-2vG2jps4i#-w4_7{RX0dW;qP@kArWDb>H9#CQ?IW#=(LFjBf36BT|V50fl0;{qlN
zhEj_#MX)2I0@DP;+*r>L?7n&xGX;A~c=1axlv#q=f{_cfP!Q}nRfV~NT{-obXCci~
z<_q?gmSLe_q;3}$33l>Uz+xNWWG0pf1}jRj)W*GZ1<M4x@;TvJ>M5R^Sy*@uF%}j#
zVU>lub|Ox&og&B@!5(&0yoDt>NDvGLijgQ7E~kiVA=U6Hf}tE*DOf6vHZ}<cE2vix
z?732at!n89Y1k&zd<dxuZohI8={Efy9-Ikw)8aBqFv9&(wvEe=FUQ8+or+w+9xk(a
zg1xn+$QO*9IEw<ouIl|L6bxVHoms*5!zCyd>^_-=5*wGg^->%6U>Obz{<u98W#at1
z0vxlcnm$Llh<Eu2htSuVs8n!>#xkdEP7Rk(CD@x8K($~b^$=<-JbMMTg57CFs1yA0
z94~|k{?y3pvx4FBIy49dFE*l4FqFL$O@iH=cCK4$VFsEl<ejJ+f|1;vxMiuk>6J?`
zTz(99Y=rYWaaXYON*-DTgEXpW6YM!wjC-o^e`Mjli0zFuxfKkb--m~SU8QMwBp9Sx
z`>|m6-ko?N__yK!f;K{B4MKulb!F%j>?t~ih~U#p0dx!g)OZk21%vgx8D`@iy@+Rm
z;qB+~+(O#aybz3Vl)n@V@{H=0rRJpKXTi?P`S`_B^G@Pd!LE~6@S9+mrxE!7yh*p6
zrjl<6cBE(FO$*Zs@s?ol0u>8_U0k}}vDAxA=pz`~cN~2MBiRAGYpFX5(N8eE<0|?K
zwl{o^0hU_Rg!e3@lrm7TlU84Y{{I^8KeVZHjRW|XaT#NbF)m|_F~%4oB11%m$QWY`
zks%^7L_{t!WLz#XhKR@z8EMzl^p|aIo7OamZA`PYw)xSdX?`^ksp5~)Ij_21k==9B
zoRgC#-DG%G#^rLK_51kWhvz)+Nz$~Z=e+OpyyS1?BHke#d5n3NG%^LeN7|_lY|!Z3
z0yb)HJdI77&=bOD(ki?0K3Pk<u!S@hMzHlc^Vmk3QkD3DM&}CHPFk@5c93>r7$1`L
zVW|uws|DGTMRVsP_(XGyA$&?YT@%=;3CS+(dQKNU)7*{$>?UjH^Vmbyhhx~Q3DVBA
zkF?x#_*^NK7W+wKs)PgoS^uwiuLlRIe^29(9*)TJnGVWja9E?VD)R+tF3jNw>BvR#
zOP=7Ste#On$>N!2O7Z?tveG7fx1?1Z!q;T2Z~@0iTRzNjo?vMbCm1v)JUFSTgLC+X
zTpMY_DYD{s<232ah4C$Ed!_f5TxpSd1+w<NbU2VkB!hO+$x9I~S&w+pNmdK8dPY{<
z12{_>(JcBj*Co|qr1_l}gQOK3ftR$qWinTz@+o}E(m4F2Bac!+a&<0&F-?$P{t#J{
z>rhyu@=zO5#$@@7w9^Glk;a1b;gT!T#u6hPStL%7X7B=L$=anH(wflIfs97y`j91!
z`CjBnd!!phWi*G9Cd5*>Og5&|xT4WW4vXYUYZ2G<%k<znb#ERwdD!X-V2N}xDJ+ww
zYZSN1roRn$IBGGDyL!Bs#eE$m6-_^r_46q_Agj_0^N_USvXh}Ej7lLcX@uKR(S-IM
zRLSO$2e#(Q*Y1#w#1K|B!8eH}S@D<gD{0Hh;Und22)~oHk{gfpoqO?wx_K7Qc-ShB
z;SaJtD6_d_qbrL4DJRpwll4E&#d*B&Wc~VUkIUQwS(nFw7fB~RftN@#7stz_H9Lt{
z$cFSlys8O>99|=<St(i2T+bQ2uFMbP4bt`}@h0EGPZ#kPgDWjQysfFZEZ)&n=}CQ;
zsW+rxK~tsRbOTwD+3k&FRhrK?k=B_4Hj`#i_BJ4$=nS@y6{-E)s%#y`HnNr(zz5{2
zd^y`m(=~)0q%jf3hkQ3b_TeKvUdrHOvJ#YgUDEoqM2u`o|L9I~MH;Vm@dQ_82N(vc
z`F`xyQT?6RqYQUoFX_aGu#c?!WJ@e^HQ~p8p6gy72XvGtjDyO~4jkg~TXLhT8pB~8
zt`+5aOB(Wp9U;wv+~<-`N$ztwYSD*(P!9t*s+m&M^OaKO3ce<--V~0Jje#(ZbJW!U
zPEh~s#YrAElG5>@l*x)y%2*PoNh>TZ4`l7n>M+uoOrlj20<CCMCI->23FD*i@Wt*$
z(8*vk5k)sym1V25q&+r|KC(Vt#5vLk7coe#N%@DDw7h<dkhV7tA6b{%9Y5Llz8gWl
z?s5=g44Sh2K!~iB!U&U2xgQbI8XrWIqkibb6oW=AifM91_6LcPwec7d%5(^`WJUhM
z(quglM258eqsVGPAdWm~W#w8+nuQcfWNmO3mq{nni7RA1HGxHCzJP1WQ8%uWc9*p3
zl4j6_C5`5%u&kWx!fn!t2XIHD$uZp3Xs;jl$@)cU)+N{4r0;<=qGdcJ?d&*y(OlOB
z82phvD50XMvk6r7_<6b7l2(gsf1nBSII^nD_|PQHg*<-M-1bpCQs$*1m#pPOc+3}D
zlx?*bbS`=EjKQl}{6Q^y@LwJ_3sN7ZXPOBBPuDN8^|2Yepu9MWzdT*P1mE)EMFuNU
zbn+5||LVfa8kDJoSIDZgG`_0i<-Yherv7*de`U~`PvdpcP6zP@>CC3_rj8oV;Vsfg
zNANaJa6=l)s3p15Qg_Dj9usTQmb5|7RPMt@vMK*vH|d#5No*!fX>WL+w4{1z3t5p%
z=2o(ri(ngR2W8=mtR+*}PCDfvc4)$Y%qnQ~q8lHPtEnPBCRbYe@Cj-4_TW?f4Bgnt
zV6_m#E<N5qg3m~MWCpuQBb~t>O>n2NR~ZXopVBji&owub#eU_y3kOJ3b~ZT3*In$w
zAw8a+$KUk0D~Q9&xek0mT1hvKkZaP5_vKU0F9iky_&dW!SnhAhT5kzQ`4YFz;4223
zA=wXsv^pnoOeuvz$90rU$e$q1-~di)LMeuC$lr?nIHm7=45t}fk)DumNk=wCaghxv
zMQkOT;eNE~sB>s%(2_@M4{6J7PN$Aa$lfxfDb15-HDNx9KBatyb4u@d43f>!L3lM9
z@5P8xipPEAny(vv(&^41sFanFG18D<{SavuzC)N?k)`AaS<R0kN?N_$m{Q6l*)-Xh
ziXx_z|0)U6kQw4xvhJEinrsGSV}$1kGOs{ZWsikCX%D$kB%P=SC9*E1vzN(LxhcLv
zHiNP^7iq~l{58_{lyRMGOu2EBtY*VlB28JNUe;V`61+{W4Nc;XMrT5}OV);_aG&4c
zjW~X0(2};`2b$`d!b8%LYs@cXb9e@ZCZuFufo#aCf0Z;BT(HTNmUD2PGlx~uDdo^q
kmZc<@tjiblh-{u8!tbO#n89Pxl6$u&WKD|IpFO?ye@WP0Pyhe`

diff --git a/models/ggml-vocab-llama.gguf b/models/ggml-vocab-llama.gguf
new file mode 100644
index 0000000000000000000000000000000000000000..8548445f8eac41d07a65e74ef3fc5d4a098c2615
GIT binary patch
literal 467382
zcma&P`Ez8~btb6US4~e(Oi#=bGtn+pn^hE9H1|a$?Zi@5pnxg}K(S<t6!nn#0K7tG
zzGN;yq3G(Mq?T;AWXpDY<VEhb*_OwmC)7icqC`m~9bt$O^JC&)&|fm&_nq(DEVX(%
zT457eneW|u?z!il{hWLG#+57o(Z}L{|1JLfkNE%ndRULE{-tW%J!qzNcRCx_|0TZv
z&+z~Kel@80kUm`Mw!>+Cd-_Vh9_~*M{!@JaF+BM{mEY^ZZr$rO!~J`o`0qdVv42>e
z*zLF7!&kcPY&hlX{}})Ozmy;M>bm#JUOPUn#=Uz-_~gev_P>^g#_g!S)N2OyaMHBH
z#Za%~iN#aZbXpImc;M1OUG)}E{bM}!zvKU7J^!w!UO7BU$bR|X@nF9?smHGj+wtI)
z@nG^wJ(@KAcKDy;$q%0V>f=P!v^}hc&7FFDX@7svzcgrjb^qt^`Hz0_A-yyCukZt(
z{Mg;V{<E{;;q(6q{{8IadBXJULzCw@)3d*F`&WLYe0sQidZc`Mw0wH3e0scmdZK)K
zvV3}~e0sWk>Xc7Q<<oNcv{F8;mQR=S)58V6hYNfU7x*47@I74Md$_>&aDngP0^h?0
zzK08Z4;T0zF7Q2E;Cr~h_i%yl;R4?y1-?fLe2*0P9x3oWQs8@}!1qXj?~wxEBL%)k
z3Ve?g_#P?nJyPI%q`>z`f$xz5-=hV-M+<z97Wf`5@I6}Kd$hp!Xo2t10^g$rzDEmu
zj~4hIE$}^B;Cr;d_h^Cd(E{IN1-{1$e2*3Q9xL!YR^WTA!1q{z@38{kV+Fp)3Ve?h
z_#P|pJyzg*tibnJf$y;b-{S?o#|wOq7x*48@I7AOd%VE+c!BTn0^j2WzQ+rEj~DnJ
zFYrBH;CsBl_jrNt@dDoy1->T=d`}eko+$7=QQ&)`!1qLf?}-B669v8}3Vcr#_?{^6
zJyGC$qQLh=f$xa|-;)KtCkuQ}7Wkel@I6`Jd$PdyWP$I=0^gGbz9$QOPZs!|Ebu*9
z;Cr&b_hf<Z$pYU~1-_>Wd`}hlo+|J?Rp5K7!1q*v@2LXcQw6@K3Vcr$_?{~8Jyqa)
zs=)VDf$ymT-_r%Yrwe>f7x<no@I77Nd%D2)bb;^b0^idGzNZU(PZ#)}F7Q2F;Cs5j
z_jG~p=>lJ;z}G48bqai)0$-=V*D3II3VfXcU#Gy=De!d)e4PScr@+@K@O27&odVxd
zfp4k6w^ZO;D)228_?8NMO9j5A0^d@BZ>hkyRNz}G@GTYimI{1J1-_*M-*SO(xxlwv
z;9D;6Ef@Hf3w+B3zU2bna)EETz_(oBTQ2Y|7x<P7e9HyC<pSSIfp4Y2w^HC+De$cn
z_*M#hD+Rul0^dr3Z>7MuQs7%D@U0a1RtkJ81-_L6-)ezxwZOMp;9D*5trqxJ3w)~u
zzSRQXYJqRHz_(i9TP^Ud7Wh^Re5(b%)dJt;0^j8V-{k_|<pSU30^j8V-{k_|<pSU3
z0^j8V-{k_|<pSU30^j8V-{k_|<$&*J-~1x}j~XHVO#fs2t;hUtGxT?rz8mtt?T~sr
zpF7Z(6RQ0&K>XLQO#HXUrz#EFYS1bN?edpq!vBtIeV$So)zJEPrRV##M>_W35ua^m
z{BPCwcYFRj;J0<p|7!hrS8oic`9g!&pW!~gYXn0j80tQMfRgX^mBnkFL#?BhkpT#E
zUZG@?uS{DG+La-Lht~M7ar-~xzk!-x{I7uMd&YK;|Lke)-JIzHzpDmb$+92qB~rHi
z!_0p>{`->u)<$VHOr-2^z&00fBnX7IR-065p|hM`XsY%5ENCLfxnFPU_i<&g#&sf3
zD1B9%)aJ%*2*Hya`h){awA!Osv;N!Yzq6#0O)q4LPgk05mcT#9?+)tNw;un%HsYUY
zJ?3k$>4NZ*7*_=I9)YZA%o#Z_5j{)(UH301?aV9M6<GZf4u2HKqY_<B%bJr6T2aNg
z;bgO%`kco(KJd`6eLd+_|1=S>O~SD1pD%Nq?x7#5G`J6+&`}AdChTXB*cL%-CR6^w
zw*MFSgKwt?Rl-+J&3Vu83C_^Q#wVrS<aY<!t1htWC$DJGX`p<PYrDp8>o~-tQ35*2
zWhD)c#;t)l2DBW1l;a-82%?&i&~Wine4yX{1N`^jIyVJ@I?>f22l;}3SZ+nHO%)i(
zW%nyD;xz+Si<Nar2>H!zO&XZk_2aknomk*3r+R@yH{Aoj*Go=w-CqRdH%EM|9miIT
z6OdU>_lZxG%?Unun6z||us)~}X2}_M4icHeHisK(H=2I3k$xPehHsAB;Ce%S_s~OJ
zN;~4h>iuT;U*j)u3w-gU{7U@m-$M4k|DRm$E{EcfXJ0(~5+|)D+Q;S&XS;YwbM6~4
z^S|?lcI@3cu98qH;i;YF0niGn!iPW2Ip!bC-}1h^$bt70W0MdFBLT3leQIZY?PM&z
z605r%bBv8Mq2<8`6S%kiVOPAV9{|L`I(vQo7RTPL$C{#hpg#rcJNoympWN-Pp}gTC
z3nRG0_hyr+xBj3y5-k0u>-D{t^zEar`MN>V7a#C&YZhbTdUe0=(e|+68s@|K+eFVa
z3G;LOw%YA$uC_?5SOt7^{JGgb5O$jIrD5mqGbq)9u@4NuB&FQ5{-TpqefC9RW~$@Y
zlsBhdhX8NK0=$`ucfmdPKVbe}BsC9Rh<<k#{9I!Dxn!v$jSMd%=N;w#h^4gS_%qxW
zOha+3`H$xx>d}54rd}<IZT*hU%(0pLA@LYfRFkB(y034=#Hy~bG#mB|a5eUaMk0MC
zG8|1fDLkV5gPrhp3zWT>Y$P3|35uu!X!s@hB<Sx$4?2;WF=@4=5E>UQ@|lTY9?vvW
zo&5P30$YVJmNOEAy{qZpnSW^Ddv(G*IPpBkx!Qq_Nzsfd?cHoVG|@pwY6qTq=plhL
zPPA5ip=nP@Xj;xxPsnZs-XMn$VciGY@Wk`?=(Pkn93Ozk^5J$6ggZ<|I4UR}B&+;&
zuJT+SkokxF=UxzWB{;e|VQ>}aHuW1Xeq3vB>Nx7Ra31_aRm9nLjMPn_s}MfrNjPOW
z;l7wxHLZib>+yMf`HO`8i{E(Z)viB;stF8=6<xZ-?HWOpaFXWG;DOk^2lS-#5$x=H
z+SwlmyP5=Y=RcXh$2a>yVD+(fU|+h#E3fe%qgimqVa*X|!+XVIFBsuigwb#VIaVNk
z{_gw(zQ13mJnCzsvBIOGXnclGjH~1j^Y0VHdf3%m+-ICqFN6$UYm}72X%U`Jil*?(
zUiv{Z@u%5Dv;hL9zZ;`<hCB!!?-+tni@l=*clM>gOTRrfqL))4!wU3ZsF3X}j_Xj0
z*NBT2UVjqu2SE){@hl99S{!M%&qtt53T6}4!88yXKhA%0_H|D@3~M39_BhejhP;Z`
zx}iq)QVpF6Em(!f`$Dq7R;uW69Fp;gF@wX#71iU=5W~WmOr^)SyME}^ncixu)P~zx
z?&PTNXNI+OcrZ>87jXPZ?oo`}#;KVe)G5-Bs&O7*%3u#-QiN_UOfvA}r}%0BG0|xs
zhdz!~zFXc+x%z8KCx2J?;Z*FYzU}QF*SR@WlMASZ$<)@S|4mTIo@ty5dF>_s_r3>+
zt<yKP@s%0$Tk{W+k`8LoeUnFYU&6Hl85_!jDmlw%W_}5BTRVmW_^$VT{zFp9pqeJ(
z9TW{{{=P>VjYX}Dq<J7Pyx7eCo}cM8d3^fKq0l$3PB?Tko+SV3l~XdPM6cr}6%QbN
zX#97{d%1RK=mv7s`nTBx;WuKZ-YrY=2{}#yFfn`X*2%%o%l@3dd-gg3>Vu(r+)9RO
zGwJ5yyKU~#&tCr*oM$E#t{vv_u6n^r`|&3LHEUtmk8t-x)Ez;Ky$}Bceg231A3?)&
zzrk5UpBlGfjCCbSmWe3Epb}Z)q;M6pUFd3@wF=&VAtq_`+>}=IApXZeC(<2)*;e}@
z#=kUwFW$}L_OsXjh+ovU61v(+;K$e=B8F+)o&F^sY-$~!cDAp%z%TQV^y(0_w_?29
zo#Tb&*U`~G7ULO6UOoGe2j5D;R!=;0GYs8gSC3bBOm?66#3%g9{ChD`s)QXe+J3lS
z=HCxSi?Au5nuI*VG5D~M(LiQeD3bHHi(dbbALv!n^ecqn^6~|P02anE+oF!$&>Yx)
zVm&1FrAwC*%s2S(Nzr-#{{QB`o31vexy^qpw28AXd-mfhd<UU~wZ+QcPc^BRE|*Ij
z0ZN5mY=`~iH4~}8bzi3vPJ<+AyCp7OB)At+$wpw8BUMLwYuXOY&n5?CqFhW(GtJXb
z%J^f|pX9NfNX5a4&8dF9l`eStPRiaxq3XiLPnr6EGuLbTeKIkB{LJhgU2T%A23#Xo
zw<1byhlRhFlf2iaJu+yEG6){;DPQr%oW98~s%bdP_8=I%zun+|4DIvL=@<(#m^+aB
z#M&KAUiG8|&qBoShZ4Q3Yy+H12&Te5cX0kV)~u+XMRPkS4yxM*Y&#ix_durtLU@#T
zzXj9HV2)ZHY3S)<*3Dk9g>fx;boZ`7o+7tE8#@X~@WI*F{L5Yw_E<GOBqo|}rNBNw
z-@Av0!yGeQ^FQ>5tC}H>$Bab5@!O!UzNmy>jH|AnuM&N|HY}YJ@sKf68TjC=N+o1n
z1D_-tuLoZ0a3+=JsFv#uXIpwe3!3%=y+&R&XwYr4tF;tVlTnPC^oniCk62s#e*gk3
z`QP(@o&#LopPFgt>ytoy-S7FVjKrK1U4N0lg{KEBAUv7vo;QFCAT>RxGy~yX3mn?1
z)C>z|8lk8fjEd2CY{y$y1_&6~hulQSz)&1Rf17m=QWc_-!zbqNhX`oLa*9Ad93{Sw
zq)~vw$A%syB}3eWmWDhTFK%59sn1JyAJ==_#-^5?xw_@mO{U>d*=r=g2T9dOZ3=#%
zKThSiZRFR5;gYHwcB#qT-K~@6>qb2Gt`w$oiMGisb)t5ZI1eLXn!4z;Xr-G))18j1
z1QiJ9M%4NG^f-Nm+zXL+C;<He%A^VTk+Qdvw{16&4EtKY)+eds+@pzhi}R`Kh9p1H
zo`LV^IE+OUlcu~1NvI#<@Z6Q(^u?w=_FAi=-s~pS^fCBXXh$BX+J$aS66nJ(%K-<4
zKlSaf1q(1KHv^h`d<rU(H+`s0kN-aNjK@DdNL{TSTXO~_f0px9(^iras<simTZcS8
zeKYrVJd;EKGxH=}DF*l7pQq^0IMS?Ng=GtC%kz%w<SK{5Hc>ID^XG#`o~Yr1ch7$q
zG9E-G#t9J?=Xas(9fXZCo8~Cv!YbwqU&Thia`yV!mkHZZ6IMBVO<f@jTC*_@u^j%O
zaU%=qK#Ms0t)h%~lBLw)6=23$O8EI(xxaCi!40?_lOmZR=U_+rq3sZ3qlNQeT<5-R
zCt38{R;<3-t^4-ALgyy2dX%Fu!=T}`p>}sA;YM+qx&ol;i44v?!^Ok!54B;S1>Svt
z+XhXwNU3TzEr!tq{z*?XsN`7K(hiUZ@%UX#n(D-$<2r@=)%KVVjAW#qeKB2VNO&Vk
z+ECiNbt*wq80@Av1Rhco1*{U@GllEUzB+%;x`yDjg*~oD(zfA^ObYyS!)#L;iYX`U
z1y={H_o~ERu=6(Gd`{VJ<M6}JdP!85+uY)@c=sM8C2<MP{8k=Ssv}(S-Per0fn*!M
zI(L!NLlo%GBrO?GpC5%Vv_tRuHNH?uu>c<nNe*{eX!*<bgiUoEu9?xnQuA_lYjG!Q
zAz0|Dm~VB<Kq}Lc6Rce`s?5dri=>1O#RvAJ!|&#qpG44X!HH+XRJG?n=Bn!O-u7F0
zG+O*MNRx2&RUx|oQdl~N@=INKiSG`Y-0yx9-XuH*mq0@lBS!EIKGbxz=J^L_-_Q}Z
zj{?S?aoFzC4X4E=4KB@+i*@0NHF*uH!`xQzPJcOqJ?q)NPQW3P?R6a18uL;*PUi0r
zn$vHKFF>z(mopBfPQ(1O+c>=FT#cyY?fH8~_v&q<1@n1os`+~ACuiUAQ{fAo4Wxog
zDdk&?mdxMcBbew&YuJ{i>mmEC&dPC`0hPR3`D5|6g<}kqmF#;jcq+UST8gmVm<zVY
z2_{%@A~W@pF}Rr*0Zd8{8qbQBJZfd#RmFd>TZIL6AktdB%|*lIC7`!=^e-qR{0l=a
z5clrcmkfU|2$dL4RHhrC7lQlZc|JNv93$B9Y&^yyIb6RAc0bN|1%BZ7b9jUde_eIu
zn(xY1QI4<yNoVmt@w7!ne(pID!uxqdAbj~nH3@!=TxO1dBbovOhnL?#mZX?y_foIi
zYa-kZhL`*nB9f1%r&GR#a7*{3Fn61UjW+*%&E;UyI#dIr4RV}7&BkJYpnUzSpK9W$
zg+j+&@9YTnS#hwz_nP$4O<6?k-}r)1*C7*Q5C25?9oJV0Ney1e0pV`ZKSAUCJE{9(
z<Tw}F=cX4?-I=1Ox02K*(1DHZHNk!Qrr_S2CAq;%l_t)v9A|X<BIB8p0L@y`E&U4s
zzGs*rHx<#_Rh*V7-khYKn91T8#s<Ml69-9BWf3;Lsq=9tiHaO>&Q-y2D8%)1lOd-0
z#pw?u3--pA+4XH10*0g+c6C~+L6a+Dw9tF@v)BLJEDb^(^lsigJ!ppVK<uRau&W#K
zK#zMFyXIG6e=YQqNt|hrB4M*lGmwwBA3M`;BaP#^=R6c-uxx`nl~(}KQY}<DxSL2e
zVSWHIA#80$OaH6XE@4iLlMsRK9J%gGb(rRMcMnQvW|T{VND!z^a20GXu|PK&|Hq%|
zg7QJ5Q(pI)u&aK2_GMGgwssJ%T*CVDr~K{w54d#=R0w_@s!`ZOz52G14O%H4^!dN!
z^ZVl}-B#E0$nQt2^3@r<J%p)WD{9z|Qy$|~d)3gE6BWH$>wozHVX*N7z^%>*M4wXa
z;WEq!2TqTSBIpwfa;U~|-g-s*v*ePIa`S~tg!t@BXJ6-XFp`)DpOFkvV5e5-vgu|J
z0X&y4%|8qe?7?&()I$o^PS^@11*gS0f|e2P7J$5$`q64`Grjd1@Po5&@-O0YI<g>&
z5Cf~%h_Z2&^j&kAL`f&ZGP`vu{BT&M%m$~-#h$)tNML!E-+K|1c`-zLw^ks=Gz^-i
z-O7K7#cSh?!(bQ1!BVIbB3^bG1XZ=4e6!j|mH}a{q~tD()gQxBN)r1ct_{$0<=rO2
zS|e0$@N2m0`KKvL=w{@t4dwNNW|?w!G7AR~bQh5kz2evp6HE^rn&VWBSP+D<@iGmY
zWd+_)PLJ(BFjJWS9fQJtA4Mbk$!BWF9dScAcEb4x7CkX`_c8#gnj5bAM2ak=foFua
z$FFCK`XTg_SolkRyDxhLyo12Dy)^I5Y@tEF%OC2o9y<!v2krq+E}6|`9Km`vA%+!&
zqrl<slWu+P;-|c<NfnyH{09-ShPajM38&%9f?z7Lgj~?P)mW_S{3U^!y2Wltuh3vl
z!smtqnp405$R+k`><q`m4jYrbVp8_nun^uld!0iMQb6oCN5)!xTYTVEowND(!!(8V
zl3|}Ejg4Fz9Uog4TdNvP8vygbnBI~~wbAj&%VGaRNEUa&F;YJKaCq=51mSfcDZ}Kn
zgiaaAxO3u#-V-W-F~gP4B6sK{B6vG6+ET!U7cprIqKc2hLtVtjl;=a|0cVn+ygCE2
z{4;2l@d<A_qWJxcE?f^m5cbFEcPrBXBS>P0oaX2r_i=<ECbENOlf-(rh?q%w5_uH(
zay66>tzYj+Z8%B~({vmH;=*}8RS!~J4WOOyo6|RP@N4z4w;y7|=_Dv<^S0$7F+srf
zpp&q3ApHcq^H;fRV$6|&kSYPZpmN8;(Gc1c{D(<_P^A13Y%|o#Z_eNLSEOx@iS2u@
z*c6+|p+g6R+aj4KlYEjSUsc$+PlqDuY$`R4kwiZ{f<lsV%(T$#r5Jip#xp{Y9&J$a
zC3|((10b8D7!A<Xzh|1<N$Af08FAa}1&f@>CATZKIj$lHp}67(p+vUh;>>02G?Lkl
z)5+o>D4ZjnY5p(#VIw&agbTNdz90i$%@Vu4_E;|>_8NPDqs51JEZx$kIsbrhe}YgT
z9{^v|ZjKsHcST=Ag-*xKIMje$d18+Xz{j)HF6{1=%hiCbG%100;a3&^J9mi_^ywcm
z2wj^!w;UMSwIdF$+n{$jgI;;Xu9QjJSBxX^^RN72Fa3&ea>5?dwsghe!B+STm_WEA
zTuNDir4s~rGMK4B`q^GJ;KO*bDzpKY4AE4Xv|o8mo`9_qOB^_j4>sC+_v1mFlmEc~
zGoj&pjgw)I;6NcbTi5yDFFe4R;Xjp;A+G_K)1@QcC~P5ie57MmE`u>>hf<L(;)NZA
zdtlb}n@~%Is=xoqC4U2rN=c4&oUKXYOfDMKoEHjF_|8t>6sxPJl1Nup`6KM`MKI@G
zgCsF7CGOG#2BTMncn#I+fph*EjFF>Y4SQ+C;V>7^&A;|z%T=flC3dkb`2rV=*Fe>U
zNJEZG;*`D!zRnm_|HN#Z<|{UibHV{RsSR+rDS~OH`mzhZ2ERe6DmSIG#laZ=FY<3K
z6`E$TFQeE1Ea;*^yNO^m2iOzWf#=S&3TLx{V(#!FMdgmdn@5evZx7yBk3)|p;jVTU
zt|Hjp>0N@6f~cKoi#B;$#}Uc{K)+;UTFN!kVr7#>OX=WvqWWP^0(*Oj7=Xvx<L)&G
zO!0vu9B%%!UFZxH?C~Ks4KihozjQDF0e$~Az{wIuAp^okHA>eL)CM9fa*nD%MDh$q
z75L*$#)o$V=q(8|=m*Im2O&=tW-<srd;MSW&2yjfzoGG&_Lf4M>L=+8;HThEl=zDQ
zQ|qOo$F)Nh_gWy&6e)v4|1E!Y1U23t_uE!Dn}07j?($AS!BB!@w@#X4$jv{(AN`i8
zbD}UGr)-9;X$DFGIldo;mrDCb8;yWBedM8Xu~kNV26Qd)Cj@&-^ai_KHt5B04sUPs
zwYwLMqk2#N5ZV}=zn&%%Ce6O2`uzQ~FY}GH9ljC%01^naxx^;ZZTNC`mJAz_GkoVM
zkZkQ$j9|a;-LtQhNYirh9D~w0&nG{b$iqU6A8m5#I)`;+41tO9`Ij#VZ{t++pa9WO
z=G^oSmaqyKLt*`sDno4KMbd8JeUnH*xVRnVw$!wD@?_lODnag1g0_QGd6=M$TIthw
zrOMZ@`b`F~Mem=x&r5?+Tx9U+o0fvm&J-6Xf*^C=(%+wiP7aPDQG4kkzlB#<W&$j+
z;Fr{&5{7A|1cccX*%;mcc)t1CJpm55#B=i?T9eBi*SpC@!*%J?vnhvC_dq9-5rLfR
zg;or-ou;8y-^S?`;2)H%l)cmUlZwFze#kxR$v*?-@WXo|WI(FXGemNPx1brqgiWu<
z{96W(W&##({E9JhQczdU1e+z2%71e9P146KL@k))m0b7%4va@=hDasl2<?7YTH{v2
zY`0aa4Pq9?B=ioyf*?h>Le~Zc3WP`b%_p(H*clE8k6^HdcOrm(`ew45wUUI=h9|3^
zB!$>19Pge7{<k_w8_FdDEy;CuExkdD;kRL|G7s9gNsg}G3*N}HZ|0~6PDDaxYy>t>
zzf&lJ0+6tu{fYnncG$0sczK{sk}s4~q8Aj&Hvoac3~+r!-kIzp{sP1+%7?!@xdD^u
zEb}gAgL34?*j!-_XV+Nw<}l_!>e-jtEH}ISzT-++8~WlyFAzFxFXy=loyjBh8s(BB
zPbx_mxbxJsF`CRiIs1}Vx{_k(nrt+%Us<6|PudJKB0&d40i>N@Sz9CgFwG?z$&E(x
z&LAs3SLCEUxJZ5Y;hd9NX7IsLOV;+1{A2RFB>LB6lw9B9JZE1$`=(b1RgI7gs!<-}
z77>VuFCT@?H#$h>27Ls0HOQOtrFPaiAPOhscYWz41Jex5wri-Ld;*S*^k{>t%6dTP
zfkS~r#^TQ4FFHZZOvexgWWwIrC5+ok*981G=ildhFr-6H$yXL{PNgRGBM2Tr*u%yP
zqxu6TXllo~NZk@Tw^;4QI%k2X;ONQ~<#a7<Gk(!b7pf7qFjfWIRz$%)-AzU~hRzHh
zfi}3rp`p5Q6wvZ+@+qpIanjf96Q7Ocray{!Xgx^Afu#Gs)L+=YnY$t0zMY;3&gru)
zg+I?UQY*ZCra8j*#l!I2O9o~BR|N7P9Dd-@f*AsZ-`1h@L5oQb|M5Y%`oOM}{1tsL
zB_R{Ba5d?({(jnu`bP&^TjNrN47_GhNgFC84&(=ZZy1#Z_q|Mzr!wK1$gb?dvCaiz
zXDC@ra>zP~`pQsp8TLuX12$VPOu`lE2{w!W<rwM09_8LNrSs>Fr@#M+f8Cum&d;5;
zK82v>NyZ(kSX^f$k^aC-oCwZzgd<5wWX1do8j!ZR59iUa=aFHngz>qQRtaQ7R%l$8
zx#vBF1+H^LYgII*8Hl7ixd{4$WsY_4h;)yj`1$Q-SR$L4Ak&#5`Uhf>amgk@R&{Ol
z!6_?-)r^zboqoqybRrUnLKG^onB?^*d`2SsM<l699PAMei=jb7qqGI4ZF0>{a`-DV
zBWLekPx>tNVmhBVEHr8UIsH!ZXJCe3?6%7A8OxZ2J`=HIskfefP)C(ER3}NV22$Ik
zEo)Ui&Et&sed^=fy#~2P1`A6rvK>K{$U>`!l5V@<X<wBQu@Ob5ZRQpq7kBvH!f#H;
zhY!$=p}d;LaJQ0)2MkAL`P~cL>_hpyU?hg<Lmq!Jm{u9PP)AZXf_kSTJ&r8kew!)o
z@Wx2DhV7^(N}x<+ia{ZXg+}5k=9Pxj+)+HnYe&V<Ski45T+DzXM}d&t{k+k;jH5wP
zK<FTbgd|k*+J&f=3>l0`k=Vx<FTP~t<4hrB&aWm~3$alsW|+n;JkdDMMj69#z$F1k
z1W_EPjv_@!YZ1mo>8&QsRgp|J(u4P1RFF+xLQgY+5>`G2ug6wN5YVxiB-lY_n@$*P
z;2^?xGK{b-Q3Q7hNfAyz<7lgvUJdtEx)~T&o#|zqO(A~R&4Mj(1w)aJ-kwC~X6Ok(
zU8i8b?dI9+q=@dyI{Hi=BRKI3eQ(Qr)O6h{6-Pv=La|rzQ<Bg8duLxu&R&nHTXSbR
z*Nn=IkQ$}v8P<*T)%iP63RAh9|HZ<FMP3Q#g#of)b&6;6+a6Leh_P{#l3_Ir#YJKz
zQuGMGZg8YtViw6Qe6bzNQMfK)zBmJG&>l>>@Y+Y}@+}2#hNs`<aO0eBui*Atn}QCO
zGo(L<M+~2Ta1=ON7T0)_Xj5?!IEy+$bOHFMdzc=NgPwvu&6zjX3>Z8{GiiH>HClbj
zq-t;`O*_a#zARG#l|KPaKeGHqKWD!kK_YmaIQ(bUO?rs?@ZVd)Ot&QMP4f662gFYf
z+tk&kkud-Pm!hVYd2{+^1|$x^2{>%4Y=~|iPNdS};3e@QPOmNRg^hHlgn)|1x|f6w
z`%nyYt;B_vqRfg(%=dV{n8d^}5%CiWx@ESjaG?@eD3CIVP9z4a!ER4FNIQ~;Wj2<4
z7|ACxo_-v!;6N#rvUoSEk0fHiWVyyqe8M}0j(`2Msp{Kx_rHsGfr<=Uo0EZef}jmX
z<bYj<a5i25vp7}*AGFm8CXkVgINi!B|2t0BOshM>oSQC?+|7i*rLuVX-OJpd(#+MP
z(%qzYW&1!7Kw*SMoX33BpQQjjsHAYiQ<t25me#sj5E877Rr3VKdjXFp9o1Br{n~DQ
z;D@Td^fA7;lT^_|kX}csmRdxXOGH(DsES4qPMh-jUQXB3MsfWh3B{YC&GKjJRHOWU
zaIa|6+Nm;}ecdp@?O4I5m)L@B;M0pRAORs;&XqVm(l|Z%e*DjC+_O-vGcbdsJpRNm
z*8E5OgkudfI|7Jw=g?Y7yA8|;?>n6uAnEk(RsA$AK$0(1Qp+fx<ik1ro72CNvIOx^
zb{i)Y0?v42^F~boZI4T#LAhtN6nhKOAbBSS=1z>lKX{O2HE55Lbbpt_@l5l%rk{)5
zZ-ans$TZzjS$8ap9J$Ni@VrB&N^lgULnooZYpwC3!UQUCWTAnGIqy*e&%IYf*`*cr
z!iHIud5*lwug!nNXQw<ges=yZi_HSI%Pn9bX=I4NA09Shn@gC~Lh8MlDVLM9pjy{i
zEi;j12V0652lnMqbPErz92m=4U04r=KjsL`KOoFd{4?)L%Y|`oNSX6S=vSH1Zi0f#
zE!TlOU7nb(`1XA&Z@-o1EHageP(R8VQsWsRwUVq2ON0QN{Z2U3@4iB8@4Cn(I<O#1
zZNXqrF(A5OIE0UaRE2b?|Ar3q%OC3f)YjYNgYX)NIGotDhcY3uRGiN4Ve)Pb2Cz+4
z^FQ{e!#3*mzdFxo(HcuOB)=k64(OiD_S1Stez_JlhG6no=P%^3pE;O~A2~c}<m@Y+
zXtTJsd(mm)x%>G3Yzj9;EI4%`+Jt%fo1wsi`y(^GAY(ioL#J6w#sC~usK8L3qU~ZN
zplV<s3OIc;S2A=+wuaN)E4dar`B9~PVY!tQZ}O?t+ucYF@^V(nOL02L@Hx3lA$P}B
z<X}auFtLxE?6C33lSo_x=?u)M^q9R$C!>+~h=ZNJk(-a2VRK8yruKylDBT!I&wW?4
zj8bvO8<D@6%vHL#rDXVLwI<_`p_Pm{zoe2{_$tIJ5b#i{a%Y*qcj0(U=G{z6cn=Ov
z{<`A=v?h{vE?)c;p_-P;ynZ96Fj!`^H#&VYcqHb^Qv_!RM`GF=-<?HTWHS(l012M`
z7Lf$C&TR)A_!^nwP(5Psl(HVq-i|c2%Zob3nX4S`$VsPk`2+wqRXVOu-zNn9`nF+X
zvG;;6C{qGfI#5rDSav=C!1Pl^wrV%Tuu?SM_TEDG@I&)&b8BF7hFQe9AY1tdb`)`d
z)JKbB1a<flIA=+3*;ba&Y}jM~8KniHrW^Y29eHb<eA6i!I;z$LENX`43DLq&K5A1M
z!B1eiLDv|FCNCQqqf!Qn!n#mswnC#u@|e!f3<|i`0P~n6rVWEMCcW>gKgjBiF|r#O
zT=1A{?ITGhn0T0$BOsu!(~No+?<~@M(?W)!nHp8ER(dW>PHC7IE~-frHz=Ylz)!f<
zK3z-qY=-G*KzSE50?q}eS;h{L_>;WnW-VRp+{H@-@cCySvciDOM+JSoj%!iUjR?Vl
zNEDqz0wZ7CbzOWVA3wZ>I)D6Sa%kV0EK3{`etD&opW*xvIAB)PfkxG)pcfXtN7ma2
z;b{BtdqOzG5nDhI;XQ3S4$p36yzU~Kjze^q3MvjH`&3Q>kh6DQWcwCDX(JwrT9e$e
z65#JRT^$=C5qa@7KeAB*Jw0%?Bp{rwJ?|pDq0kPEOPqIEyqvBcB;URp#lxF$SMi6c
zpDgN^f0_U8*GGmG1~dY%IklS;xT^fYm%JZToc#z+lHY{E0P(@mAbE*qrpJpGdbKPm
zd3AvmSzm0sokC($P$D%QE)Wk!{je(-R~fqZymU@_-a#q9dLcL_Y>n10Gu@!%s3I0a
zAkYsjtiE0^bSwC$li5SGcmfcjtd-BbY;KOZoh#axK1yo^$A}{$UMWLD2lB)Dk2u6?
z=)>L^WP3AX)ngA4Vl{xh8P#8}1nMk5)hM4FOIHVZEv1VS{5>Ra!1Yh%h86v#@jS}R
zV}vaXw4zRlG62jyx!r{JV|)Du+*ka%DU5100++DkB}t$kB&yd-N)QwE2`eh|@H3Hs
zKjJcS=%ovUt}!Ip@jTooZ<MkseAL#iX|?LyP6$wOXR!B4-uw!2SV9k%+dU0OXdcLL
zoyoFCIt)PM(+C3d&*>XebeS&+>XD}Y;Dfo3w;|hu^FvuO<|om&aBa=FLhwYVnJ(kt
zVrU!T-{4Ntf^O`1b9+$z5Gwajs}TU4wBa>JSqWj<E>?RbtDQv|L4|#&NW>r5Whz`u
zHl%`NJcaO7c__aDNc|)VxWvWSX2rzN9#pDF;k>9N<KT!>r41&M5MS9&Gd9$obhD$1
zE4+TljpqtY)v)9Y&w6my@X?$MQ&`d(;am_pNs(RoIhm<NSk|s<!$GqvK`P*{?g7x0
zq()}^Brh(pESy+v@F>I=99`M{4)w3BQRTc9Lc*M;oh9|d0UNlb#pUgJp=4%Ez@tJv
zt|IyYWz{rtob{WX)SHo37iS$TH6#Zb|C=CMge*8QY%NoSQqtoTJ%|7KtWQemU={6$
zs!?P6nb8k93k_veijgMpW*aJS_Q-XH(iQTTPTNID91JS*&K0+ly2KC>LF-0aK0L1!
zqT@`Fo{h-o3?HNixT^mF$A+`QF+@?(-ah*p&+<xgG6u%^#pKv9PH4UHo9la)LXt_q
zkXHuRvS&cP4Y)jM3Ss^Tn$Gr-=|Xs$zW!j}^{?q2q8fT=^66p8y}ZekQ==V92VEH)
z@??NG38{ir8r~=5Nr|g&Gr-|j=#$5RGW3j`=)6F!NwsY?YHzNs1f(DY54Q$c$nT&f
zPZdJ!B9g%`K;W+>cU~{&{+FKfl!(}5Z6Yi$zM%3TJL<sdsCVF85Mi+25Pjo1P&uDS
zzq#lZxhhspzlhO`suzAzbn1QWgP40Y)`L*d)04S;Fb1dgR#f-K?4V&6RRlie)K9i+
zQPO=EGaeeLs5u_Q0Mcx@8u<H5om7{VGupz)ErKIa8w<O^C`V;t3Ek0<G;;gMIxY)&
zyAUY2ri>z({z`a&ra<IO+HP`ZhLc3`q1Ee*b4+B5u-t?{-Ti!41=k8R8FhR0v3H2k
zL}<ADKJGT*TX10V&*W7};p5W94S6j59^rnk2>Nc!(+RCLA|XQFth7;>L`O*k)RS3t
z?rEXY3x_$Sd##{}I7hOuYpudfol=`Z@#x7Bw~VYOaLl^rBM5<4YDqFD79~Ymm7Q1c
zHyg<~b1;okg;WlG^m!Z|NDIU2;@?QtA?`s%$?pGNy|HG!Mogx&#0jbh@k7}QNp$wl
zB2L00OVVd2=j|$6&wByYlRvI*i-cv>;O-eXy1h*Bsj_6GW|tcEWqfT4T3%GLwdrod
zS2`#xH<-Wev-;5Uoa8VI<rsg<t(??Ih2gI!p*g}>HGq)$o~K$g2`QzjyD+T$AW~3-
zC8RnS!StJz+Rn)c@i#sS_K#!Fxh`hwnHqU!{0ThDO+XGZ#SvU?kpRm`uGex!u^;e%
z1IIEke?*pu<SIL!iN~FwH;2(F2F%@pe~Nz)$Mb+iLOBu{nmR>7c47nIN-x$b_Xi4>
zvz}8#0g)J$voL(V%qN&+<w;N9hk?Us)|gczEGNM-a?RW;gFb}H?vIeZ(%GOvLk%OW
z$jLFRM^(~0qqum&Td-&+$kq{@=(-qH>Hom=jBt^3acBr8l6E9ic(NjQlHZ~fBUO`I
z+8G?sUiiguH;<t>{WtsoJ1n&A_UHCJoKtJd1RRE=7e$?i<0Cqs4Xi-m#S#IN@Jhy&
z=2tkt@7@Fp<FC}T<1AB%b2R3n`XMx%%OfycF!!Uj9Vf=d`^vcV8t!HV1zVkXyV%Jk
z&uXjWBtDGUXYvui<cr|*ENcjn33r&YqFOke08pT23*pO%u@_fQ<>aeonH#+vLW16R
zYSUgz7jST9r5zeJwlGPSC9fNlX+BQ3IB3gHA!R4jCU!)SzvA@UlHl0J`Cl7~y@G7$
z1w<pTnR|^qLw>_zBtq~Rrg(8DkpbhkJL3fNu&zfI>CSrq+vHnd&{4ASbMd88zk#kg
zd<vS8+bF&sX`Ah6kzG<$Zd3#61RXU{SPjBaqe$_usz-tn)&yb%)D>BHcY%IdReo<0
zmHU@o@xM`E$QPkB@x*ji{55)l`P1^kuLILmsDfI`cRD+(eg;laesTH(9k^LG)|kA?
z$B|lf$<N&h{WOW0A@*oY+C50!vj=MBo`M@vV>2}j7WD{pIcc&t!k!Id8hZ=3F{j&s
zNX(T(WkW203%j;$5X@}xeAQEY=osPAR1?|xWuetDd@W(jXxQ0TIO%Mfk%QCU;lJTL
z;hab=O$xhkKfge2cZzDb86!>oa#kAb&Z02~mKeqZGAUE&Z_Lt1%xw4tEYL0r#cmW+
zVT3SBp6w1)ZaCYu+P?Nd8VaF-Q#)<WYtpStKxDg!)L(aZ*R9CK<7<=!BvRn&H*|VK
z`7<J`z6fXg`ZYgOX<M-QI+D9e*yk{CroEs1XNUuUMG<v%-_FL1dgyx*^+L(My~3Dv
zRQytV(NzSFn0q2+;Qk8UpL^L<0@GTCM88TGRY#@<31BUiVDJp^Ve^W~6qTlayMqpW
zd~!EM9>(9rbQx;9Uizdkca=}xSQAYY27eQz$L|lc37?d4fPN@Yn7PDTFtoJru^a*E
zQ-ybezw$yk{#y0-Qh`uOV3IP9Lo-zH?{=tka31hISP(k7{iGWd%F;OVE_I^9`Sjb7
zd`*+FWa1-*z)Pnh<6Yd3BaZA@u4F%Rc_1GBz>RHwP!|i%k1%_(Bv_xG6_uD~ijiZ?
z##Da&>JC@gM&uO?MZ+SgtF+XLkc<a{qGAMCy*44ib4cAJPv={aoRI1WNM|XdJ_nJu
z$DZpP6EuwoPZl-1yR3)?-V0)5J6u`b1GHB|k%_Ui4k=0Cn@B~8UZHS#YW-}QIe(iN
zp&G-#mwxuTx)`vngGWFz(wDt5sJX_-dS*DWAUA6$2Zi*Z|C?ChE{tCJuPDkyb;3g|
zXErch9i9s<KtDlf_e0fze^tsJQdL78Lt0;HZzXd+m^z4#gTk2;b!{)d#<h@D8StHm
zu<&=o#RDbQ%gRkYUGGM7^Oait20;T+o}%R5q>*l5arvq@82%{en5lY5G=!b87x^|b
z+9(_y{NzWT8LisFyjZ?&=sJ;P01DWUfJI25R@WQ2sP%RWJ)RhH0D=R_xe2br&smc~
z;$N5IdG{4z07ojbAfWU?JzWT=B?MbV)I(_yaXDnc4V;EvA@~Us#5nZYb{=ww_Ogu|
zI?2y}+-HY{u90Tz#W^PkeBSXLh#k%Yt^4b2_T@2|A{Jd^6c=z^2K=_1c{2|NvYB+k
z#f1?zHNaQc&e&u%Ri}lHi2mI3d56G-^PXoQ`+(0*S$Rhw!&yN50$&W$@*_fX(9Uw-
z_d=H$jVuS6TcR1HIDFCL!cPyG%r{5CGpq-~d3~E-U|;E(^G9|^A>ylHx&;mZl4#`p
zQu%?-tkIBtl$ti^Rr)omQR8fM0Sci<{s2B15!kJ8h{G|803}CVUoC}h=#3<Lxi|Q7
z?tycJ=0e>601dC4nSPr*1We-lTux`(dpfC#U}6|sr)*#9HW4fNWS$pEHidz#MAv8-
zWuQbthoy@QaS18GxUs(OZ5&Bc>}>KOn4x*Zps%_EKCXM^8+B-|D7S8l^KQsUIRU<r
z6uY<v_JV`dpN~S~z3>+-waPg`c50n|QkueU38};)buSxKG5{7W#4hN{7lf;WKtQ{Z
z762xlCf;@!A@ql8m){9P3deE3jWP^mNPUezp;LOQ^KgEeUOIL=-VK0}9>`@Q^M!v#
zQR4t}vs`iT+Q*H-`Tt22T*-6MNm|U#5v13Yzevo&81g?<cjyl>Jt0ESWJNyO{l%Rx
zr|+lUi%@Qs#nQ&hqfjG@DeT{IdM=hw&8%ySR#!YUnoE}8_StWHu`eJzZW_BYQ_|~9
zbTCpR0ges+IR7yjt*EEpOAdcCf_l&y_ygOtarHCZ6!W<(b4Hn#@a*Uj3LG=?fIuBK
z!%#4Mrs3gj5gBEEo5~3N(1tINIEn0z-4=qJFHv0bFI})+s_wDjYsgf|`~foi$Wedg
z0k3!^l#(!DE$yXlGodjNWNO$5#T6k2^J%&|V3?684nlCT=U|j}z$8G=BLl;XN%2i3
zh+wY?EUV}}!~D;py|(+dE<@rH2Y1!hc@2er-eqPbNB;eKo3=bXDP9cx;T#w=GcIui
z7yNqw6yu;2l>npWlT)lqt%6xLO}<{S_l(K;r@w0er^uT1ceRWm5OU-IjN?$;3S$fy
zW~pzHV+_?w_1?(gnT(&uV^!y=9mrvTSiR)O=U-C`so5a&!`PHUoBaL4s%d_ulQCLw
zqWz?j$U^f)0*C>JDxC)HM)2@ze^$i8qVLmolf{bQK@{0u?FDI2HS#x%QRyg*))}L0
z>lNW1#BBW1&Q)*7Nu4xQE&79WQC{NTJDV~LYTqjGcGRn_D}ZUzVV}bEnry;{L`>lM
zgl{PY8k%y+e|o6U1M;)dvH@5)S~+24so*R_ym$sIAZm}%DMuU9Z!&iI=_)z%GOz;G
zj#!0+o<Y?j5g{zaKPWgdy4jhE-}Fn1hGW1L3`<`0uwA;POe_Mmbw5pmg|}xIB?MQY
z6dSFjjS$p?R8ICvGAN|467JzPk{lX{GqA4O)QEaHCW?8mXVf8Z>-q-4MPGG(E$6GT
zf<S!HkZH9O??}zb6U+2wedStA1Uug=**>F6%uC4~ZY0fElW~M|FWN<=b2tUI-5d}4
z>W}@*{Lf2nXEVt;yU6I~obcq!Y)3zZ5_3ET$yys^g-xV8dDADwudDnVBjBj9x`0Td
zQcyWG6;Gp!EqvNu>HtHeJqCNf*5QNQ!;(4!-boy;w~>=qER)lJBQ`fVXtJ;pjlR(v
z#|LLq<~7GcLmTOY6q0AJ8@y+qO>`}K(m+%4rk3#G)##rB1x1q`H=!e!v_`h24Z1*%
zt_1!%4iw>q@?uC>3^dXS2`kPNprjMR$FfADa&-FF{Isp#a<6b3kAZ&aS@P5i7yQrY
zxz|Rv6;@{-7A!$NwzEkv{{AQOtpdL(rNCjZnVxW?5-;ANH-oTT$Ci78(EOBnA$$<D
zk9L{493)~#E?cvYv&u=DC}z3%1p?N>&md0%iWgGTW$C4t!^EXr88Wr~5hrA!m*<H>
ztb!FBg;TlB0))W*x=(Ur)2l#M5WzSJ%i@_*nE5w_)|BH%=uiq=w76br69SQpg(p&0
zqB(fg(-&KclyIg+x+Xx_ok<(Hor^&*Q9P>?p3g;Zd7Pr+y4k9eHD8Si7&sA)fjzX1
zG=|Yq5hB*#<1P%*Ka3H=&yC99EB2OPp$UeI=bhzt`er7!$R6@Q$W~H%1Y+okbyN!~
zRh;21lFTGa>yWr+)P-7{$CSdWP4y&P7%4o~7qMPAd10iQNN*`qa4m0s1@j2s88-~N
zn}Hjgz7fKDBHeH)MJRF+l}Cx3sbtA8tqdCu7VK`>v28jYpxz0i5hjwzI!PgkB%%!6
zM=Gz}>&;P<F$cur!Wnq_-4qzZ@M^7PSnD$UH{K?MTo7C5sbOLPcLE&tgVMEj7DWRf
zl^0qI_bf|EbV9*iL`ql%<%OXeIysqf!=|yE1-0|(1w*ij=ta<Y(QA3C#a4;AJ-D`f
za)6`IiBMh{j!D#H5&!0Qd+6`(xzG(NLXT)n<HLnC5<Uwy>>Vvr&f-$928i%a*F{dJ
z??-yj<kfPd0K2gUM=_|4BcA@Xm<JLB4JY)ECh4Tk9Bsdb){cwT9WR!ChufhS1Ibt!
zkp#C<>}oqy__(DdEnM}mSzZ!+P{uTz?zEXa3Ax<-`~E4qsA@0_|4%sv{Fx;T40HZi
zhKhRVWsngG-K9$sOC62YN%PY`<s1r}Q2pcj(Y@Vt5k5r&ipidRkh96{&j%L`^G7yG
zk=B;%63Uuag<^X?z39BQ68aV@xO08b0Hf?5lbn53M710F3)i=I3@dV6QzJV4v2lSG
zwq0a_@deCmWu-|I0l`b@HyCt#mB6sSR<tr$Qit9RYz`PXUq{89+<%PFrLHtNG6AqX
zci6yfCY8DkfLI=Ejd9UPaL`jlAIJ*5yygXLq8`cgcw^1DS#l2guxqUZ9G@W>&34_*
z5ZRlBY;Y6=KTV#pj}RcgzmrVi=7B&*p+Y3J?nX@*Twh3!=G2gogEY?2#-JoL-in|l
ziaZS0_Gi|K6&N)ceBOW(&dE+Peh+!ow%P4~L8l^3j*5Gv*WxM@DLL|2zMZ4T)kjTf
zGFA0awqj(17nEbvi?k1H;LRotv1qx6Tp<!w+fQWQ|B!K8N_*k;l3D+dS^(k*b43_k
zYVx%(0GWZy?SX5a+K1^8dx$hp!_%Q>*q<r?q($BLz<s%vXtws8oHO8mMG9^d$*}33
zJYaX44DJ|Y(Io1q%mn)T^{WH`Qv^jB8rEQz7=jzn3>Q~|pv8I~XC%%hDcDU7%rzMR
zi;C*VJjVB^M{{@glAZMb{%IywKpBf-uy#My@Ox?;;K9fua|ot<01YUF@0UMkz$ulZ
z**3b4=#ZjzUrFKnItXx4#Mj+kT-*KmY%p?}HwSS)apnt0SqXE?%`mtlwlv-{yNYs5
z92Qf;Z|1d7u;Rpwz%1Z-@y2ZwV|g2%L4LW7%;sV<z^fCLRI_r!9CoUga$p;(vwW{U
z+;C-NF))0^B1#;?Nb(SvZ^d1<6M2_V_rUjv<pm*x0?aek-Xb`11*)aqFP$Ul&K8#)
z-;{MJMQ>H&^_je>_f@YB4aWEm+<bZ7)s9h2fbMw>2=}H@b>X7-3@Y`aAsJjb+65(N
z`^jtDUNPzxQzeHh7|M!x(N$p*<n`+z@6cUQdyTTWa&^y2Si{XkDRCfL0#o;!TON4~
z6@h5D{~<5t@*=bw*#Hn`f?PJd-<l#~<VB#v5sE}i7190{`_2V09wM2Ds1X#baOeWs
zfmBe|E_pQMKL|(2s?vc+IR-(x_#9yc&8Li^Ti&l+hE;7o0aj9kH=C4)(;eNFH70z4
z?-^3%V2WB856E*&eUIGjB$UV-*En_DNki9e;Zu-H3*mh3JYPZ84v_}Fw#YT@&m@Sh
zrX&USBwy}CS<JAJIkX%J1zy22HCW3e#B+u{xEwkD>a`t%&VWi(K*Bv2q8A+>iJtin
z!!;LO2uwHI>AlU&*!Al++!1-QR_sSN#hanUx`)({&%VTAp(Fgrb3+S+kR&|aQsaQb
zjqE-yp69J(5F!01iF`8+xUdk?>kXw-+k#S=%=+D9YqwA<rg33mH)`9bad*LCq!@K^
zt)WR28D9Q`-6phfZsa#BQJsTpfw;g1o=B1RRRtX2P|4@EtA0TVTGPog<{P-ZG7_c^
zWeshm@^BPg{a9vtPf+Ug2L_~9I)178#^0il*q?TS6|(zFn_e^Ca}zfWtXxl36xsyF
zC$e#r_-9<_UgCzgq}glM{c)I1-{4aqK<`!*PMVmIwJE;oy}qk*2vM7l(fp6g{WV|>
z_jKRS6pdAxaaI3+!z7sP|L^|{ez>^Pb$?@P-K$^py+Tx}-q#-B2q44IaIl@O_6u8)
zW(fS)qVxw4y7oBJ=rev^r*GJ1nlO$Sso?begG{2_hBhr$vJ$6nwwvVW#zn@+zTyTr
zQ%PE#h_`^?!>fU&pgyFiz^~+%rmIu)_0W-+^X|Y_=XmteDma8w09KLg2`4&vWxGp#
zlkXrqM%Ej0m1C7K8{alE;H6I8^-em4ko{sDz@p%eR}mMv6}jFG5~vgH2rT(g1RD(m
z&TyM;$B!F{jb&wY_;<uF;vx$;(l|O|TY37yj-tMI<+9;}??^_*v{L7+&D^jx5Vs#0
zw}(n3z#b{%Q4;2ElBgn7o*7kgO`2qra{KP2N^<~7c%{<{v_@zs9BXCC8@Q2P;X4T3
za@nXlfdQWMQbp+@c7xxr6Ml$ldX+pvCpQY7f#@PK(!(3_O>(bbwTK;`zM)*9kG_;S
zpCo2*)6`_4#ULk~>q6UIwN&;R=}Qj=hxrW}V(I>sjUjW#wj<#InI`qv><P{XtEhgS
z_4^Vj2L;_gSF(mZNEU7aV@u#<oNZeXq1-$;8vPM{I8&kDleJTgYBLTBU1l+I`DmDu
zJN#)NlH>t<k-IR<O_~6dRIyi;8<8u=IZ9jOn5wu*U>}Uh^(e#WC<0|M?PnOTaV^YA
zsueJJaOmJa5amzxhAzDHG)My};q8r5mjG8hpKlIRE8E856H5V;)K7<@>nzfDz~7dF
z{q_q5Rz1W$ylYqEN2jmg>20Z)-I8Fp#l13R&A(&tM(n5L5l{o^Pgln2)kG~>VHetZ
zg55@rG1z6GzlkI^Qw8)_C^|ss#9vU-FY?(?VZ*KCRuC9OWUI_o4X;-z0Vur|s;)?s
znK}k=^pX4(SS1ee!ZbK66`_c7g};7Z8*?i{2FRn23JKgTlI9Am69wSF_Hy|L%`FoH
z0?^=8Tr>Qq1e-~a{0kc+4PfYD5Pzo5&@w}3KsOF8^`H(;$QpzsS42v)Q{n)JQy3Xe
zGs6j*S=@?*v~E)K=Bkxg02_%6L><^<RFib3=}Mcl?$id>m<RW2>41kUMqvXqW;{-2
zBDTp?x)E2R;_KyQuO9_<fSK!uCm@L<V#u9zTe1(Zi~eCSD@F?mM*!hg5BDgb5!OH#
z%IU+65_xSbwIC0CEijGjNq>Q>z)7a;!@~bjlH1e+A2k8fxJ*N@r@;|=7MUwcrQi*6
zk4%S%<~|Us7e2@TUcStc*f9oUe?zBopt6B+rJ81_o2YR%Ujj2xAsO%F;-g3|@h?bA
zNQ9&Hyr}_7L1|zKT7_NB`57AL%au#+*&x!R93k4npVqEjAp#rJ9wwbqr1Lu%1J!LS
zHLMi|1LqsiXgD_VZ1U(*)GLu9a!KZI`6aa4q65q`_D<i-YzyAbovvT7c4sfoN)9*1
zQL*{myNU|4*ReP1a>=y4oH=dqqI*6$!N=+2Q^jjy3b`!=x{sC<{M$nQCs_3??>k~U
z#-z0kut`4w{aX4!bEMruT|Lpz#4Tw`DB+OTkV`<U0sGHfG|5!p;u+$<o>xtDktHY~
z;AYWZ9zniI4~XdMqi>a}DtOFNeW5+8s2n>)`P874(mAZiu_ui^_dG|sSB)Kp*8!EO
zY}M=~T|+LCk*1p0g&7mQzPcsj<|TCpJd&)f7Pl*8DN^Y%NXG*7FE4yMdz;Ye4deUt
zZ=}XeSXLf>X!PDqg_PU|y(zq^v)^J#D;J^$nOYHxOicr;A-~{%l|SOXOWY2$IN;9}
zl$z+Gjah;fuVrWDXMXnj?-;%wYjC)05ZSQ@aAFgGznS&*$4!`@=oS{wMX3)A{V=K|
zZ8`B)bV|V06NQ(E%d>+tv5mW(=;uzv&=ryo#xX-;L*XmyqOkqw(Qx<tB_o9Ar3foJ
zY{hmEBq*t4h(jan#uc1MD;cE1LGOGD?%i^89cVBKo!R*Ys54tMbut5Q_7m&SJN?*3
z)itw&Xc|t)hb@2ea;y?HM%axo?W~&azz;Lq1Z^lYkd7)S-|EEmMyD8en!#z>i6AN#
zn<9-FSlkrE>GhXhGE*~h8L3zohQKz~$pa9O_Aa2wR;kfb?=>uraBgh#zrmDox2H!T
zdxN)$n>`H`7TA+vHHrSENEM1yh~sSOaP}217(UPUj5a_?O2$y7@4(WygvR>=<UaVN
zP3Y5HAH3sH`FBsNf`=Y@h>Jh~|4Wo29vg0Uj*V=Pa+knXDNtskBWyC^Z~M8!sGdB1
zlk+vmceIX--q|R};&ZsnhqFiqBA8CBtamsUygf)XGFifOb)aA%^#h*8WB^bGBdfz*
z(Y!MBQb>;nkY;vw4>4}RIPu_2(vDC(swNaVJ^vt)0w2)#JR0-&3DmMmd;BFY8RVBp
z0)RB-8)v^4?i|<<;-6@oEs~IzyExt1K1AdH`0R^Z@7>R73t(q>VDxX(!A3GkN(Rbz
zK83$o3wI?Io)=12u4_2o0#i-Z`8=A+X@i?dskC1PM@_0_=dY5B1Q((D3Lb=~i3vsT
z8%0Ra1}z!9J%p19_iYKf(|02Uz@2pv%uJ&OP2*ENLCvzrxY}=1qmHsuhYl-f+oT4h
zx$w{3^ACuCv2S4tzLO4vm15r@eXV$?VDfX>kOjE7*C4Bzf^3{<N?yb;Ox8RoPzU%P
z72l#LqlTCujH^uUK`~AW4<-gI&eYx_UK@d%CY%IUSGTmi?-utdoK(301is7~AaLWB
zGNc%T@{L@^Y6%lWj>TEONyZ3&pU{G$q9m2Qx=u?_jej%E0Ky}3F2xUh*@CB_WZ^jA
zc-J&EJ5ok1mh1nNRK(zo-(b!vfl_Um+B?m5Pf{O*w#IfHIXG+-0m;5#CNp&XzymrF
zE$*N3x8bVL`M9S~!ywZWj&YMMH=+su6@W!+gkTj6`kj5Hd`1O`Cr2PJi*liK5@H;P
z;Z&tPFJ%sUPeFb+&hd-$Ka67YW|Dn&R_fAPtvV2GqT)4((wPT<XcGDLU^P<0mrmAl
zoA<|JB$G_$WhQN*f$H>oT-kOvuVg?}Fj~tU*T{Av@!(?k6efc@J(!G5g+~?EJXV&c
zn3sJ-&T2DCt6t^)90_2)%-N80NQB^@u;+Y6m_QQl*Pdcmmpk56w!{NJ(w0Hor`TRE
z&Q!pT2&o_)?x$;IhxbC}R@gCu`RKqU1pO*I2}GVW-)3Hu!E)QLVD%p5MO;q9pIOAX
zt5(bBrKiIU|22Le2?J$aIv$922a;eCP7l<TOl`Oc0%(q8!*p!kx#n4}?rf}cYvr)=
z$=JxNY2d?P3LW1@?81lx$AG@=*HIrU_3)vK24o?6pk2PT$N<DWCw>K3Pd30p1~Khx
z>Ji}yRMnY=3$p#7vwHAJR|XgC$oytKvPGae5J-4M&0tXX6t*#{1WYQdr33@hzQHNO
zzjUB`*Q0fK6e+C86Cbv+XWj@E>&BM<-HtL10v!`PgwjxPBP>UTTOn*yCpe#eHx0`d
z;2$BZu+xn0SQJ{2Yf;sds-Bm@Cg7RbR9cQueLy@<-Q-AOV{yCd3Og`gh<HW`65FJD
z=_+oC^NP+684PAvIR!KGm<1MEUGshBse+UUAvEDul*Gi^JUwhLJ^zW3vECNf&~3#e
zz`cnC^fZ>9b5cU)=P%rsRD|ADMGHZXxJ3rq4l=GxD#%~a@l$SPg4(4+&kr!bq>VpR
zYzasYJ07Lb1Yp_OAA~%kMW0lrEV`ERy~&}TV?RgsR01~T6pI}D0yR34<hBP<A@!ES
zBauGM#$C2SH60+A3hN050y8NeL1T$9JQ)Y&c+i}nTY4qy=m?VZ59e@c%?tWL2YAi~
z?x>@7@nr)uiiT3xfeyCR<TK-wQM!mpMFmgD?@f||S0+m8ksh$Hn1<A8B~@@^6V+)`
z4ogQCxo}%iX)``Zg*rMJ5S|7F@=faUU>4Y!WXC9d9p~l^szqu&=ua9B3O79)x}Y}j
zE^<7mGccr+SYXdZ$rF;+yn5uI7<Dh87MGhjYmy9cJ<~nibn4+9kI&XJO=L4nGI2Mq
zY4D(^i1MLak%o!<5onB_=Q`3g{E?MWoD%h&d^1!;k8v0!E_W|vlbplIaYF|JiGp(v
zJ#TRKP%97B-tToH1*LnKEGbB_bl7RKpB2<#_%ITqXm}9C7$9C$i9{iEr0kDM*?l$1
zWd6%}bh=0|;aU-8EcH19gd_`oOT-KbcbO=IRMoiC7?YuS8Eew4)5>1fHC$fKMi3vr
z$LS*8Tn9+JwUg05!XJ$Ci~r~MZ$*j;{BXG?ojZK>+%s8<Q<W4Iq-~1)*&HEBi4MGw
zh2YEYMj9F}dkqtEtHY%u`c>kbTe!g0n>%Ssw}H*rlWCQ-hKprWF+>v8TZXkQ#Whoa
zODM;sQ!zTy@MNNPN&Fs(3dJ}>QZCq@$JlEdQ@gw$dI_eYdXAx+qB!toR+Iv32pVw>
zaK8ju-TLWQ3_r_wXrnBzdcYns;(ZMQdu&;@p!?-a_AJsdBNl|+52Kkn`TQN9g!?aY
zL1=s9;Wr>h39}Lf-seMfQl|2E=U6)dFY5=MeXE7ymRi*H(ACgSj)UGRotO|+N<N7`
z<|UQ&nP-SQRAdRzZ4$oGg%$w*1owdV9DSp~Vv}6v-}M4^r8K~j5xoO-ue_2wjX0dK
zQROv%v}~bqXw6fhcpY9RX~7_5ZDm$*WB2?vl<!hr0cT<M;&sl_74|nbHuhBHX@H|Y
zSxTlCvvO~3ZH=HKeQ7@fXz-v3BxH1RYfq%cfILzOSZU;Fqb_W;?6hfu^g^9Cs5k@D
zjBccy<6@DF<~MQ6w<Q@hcV<WNcVxu+cc&i$kvBqNM;TT!2psje%Hd|H@=ij*v#8aR
zS9}=^0X=8DH_O+(6^J_M{x*2^{nMwl;Ch%x55ynU0q|PR-}h?_=oquSS27I0O@+qk
z-Ha|s4GnldRk7X5;-nWvc64x(qyWV_1U1@lrPA7g-!zee_6R7$&!<<j3%DXBiK9+8
z`RJs{UgW(@u3Ls@izsd5qG#|$-WlW-;ULJ(16N~;ayCFKniu3Tvf220Q$j)u<q0jF
zQn`k^q)9~zmm;TAdLKv4xL_~lSn0{6NrfA+oY&*}ayWxY?!ZE}M!cY|E7;6R&iu{r
zW5m4-lO#7@m%S+eL_9UFoWT;L6tcWH_x#W2%7beZd->!i)1l9#v(zwj(O~i$I19W9
z;mW?pIiYs_<wq1w!nI215gE+gW+a|pK}8PI+83R77WY4xz){#?2x(9G+%PpsT%a9C
z??D6@lJBsBxoi+_gLP6B!u%W&ut0n9*zhuVKo+U!$bmQ)S2!+3dw`3&jt7jJtvCGw
z2&`MgEwVJ=w}Gd`KZ>l&$Y3)5ObbG?gj`+-*$+E|&7#8Rcw>_+<<c{Ra7Q)t{P!)H
zea$faWjOypKrRE#I5-%-62&7Bfu>(X8rH|$YmRW=eZ~(P1^5bN2E>xjtTfqV417d3
zOjkZdK!(T^p@GX9o@Mfb`c<SRkmyg}k?KvG#Cx;}m#!Yc$nr0D)L{au!0DU0L5O4G
zgz^0g(QX(wyV)HJEW+C7VxW6UK^r%>2>v8R+F?n(TJ&XvTPK{sI1Fv-90No}U`H7)
z<b{cGUOaz^d%;eb{2ymRxtUfmN5e_}8%YkSS%T#XqflBF7pkx1*#d2`r^td6@u+B)
zoz<-rDsc~$p+LYg#ojQp<{wh89*SjR1ZCT%6$H~uJ1?X+;+U8ff8h%rJ4%Hz)poli
zbZ=Fuj)Eiy$x1-IsO%SNPa?Q?@uJSSEM*jsdEROlP9Vd}R#=?%-u7Ath(H*C$xOGZ
z5$<yrAIS4g&ooYvp}qTYf7nh{?cN(`{<0Ko(=8QU%<k05G+F<uTX`ncKfoN=UEF2@
z3k*SQ&<n{Tn!d7Xfj~5OvWDqaE59|fx#G|>ErySrzM%nkuJa$b?%gYaq(qKUo~J9(
znHiIjDSKUX5V27%F$j(-v`3d!GVnom4&!POJt+zlDge5NX7RYR%s;T0oVC4tnw%$?
z!gucRqzy(@bnTv<{smWh>2AWoK7mlN`Q?obLIXGNs@K8VMBtv=ht_Arl6Un=xkd8*
za-GNv@Z2RiuB@6RZjW?iASYOqNi1zoQ!Ac?;!BYu3m+ejGxQ8-ffeud4H-CaCq>^I
zwEP@qxQvYnlq}#!zi{3Y?^R+z(d0RIVk-hRx_l)vXkpH04GqF+gwGe00GOrFb}n9U
ztd2Ee>ZU;~p>sv*XQRl;1=k=F@19G_C-X_lM_&d)h{2=vBw|nB^lvXf!{)@RrKQej
zd?)!R*e&1V9d~&$H?MB;P5eU0bKM!}gmI1#n<)hDPNZBIBa>NXE-)S6+YdGBd-T)u
z8Jshugc+)<a2gP*(>G+9wKK^CSpKO6?SuyjHcxzD&`=^VbTz{j>40#P8;266_f#{X
z$|bWy(D5NuPIUYd0n^7#xZvOb=Se03fom7kCQ#tDp-HUat|8#PdT^q*;=&(nCWwd2
zLCaSDL8&+Yz}d(@`);R$fEmXF+r?SpCL^C&6s??Xk9tKVlAdjQar^Co>;)>OEZ-$1
zY-fG4T4O{i)#=|B#@m%{g}x-F7aZYow6|B;mxY6LE*q22KW_l29+hJK5fpl31_zbN
z860iPjV+Cx%Z*G5`ef`0%a%aGrPqQ!Ybopy*JI?AxH<jnJo9kWjq@{EQaVXTkrUwx
z+dx~{gQa>5ZcjKwXR1CP*Ra(2+}R%#Ly{rX;<kY##1?zGD4Pc_P2|^1?l-^)Zlv;h
zE)N~L&PBt`%aHvzGk#CW%F*y}mI<ssd;L!eQ3Cg1n>oudFj4-oDMyQsi^vzaDgV@!
zRB7a0xZvq3609_0dI7kN`?(yb12{0m?5LNC$GzFe1G3u``s^t_aSQG_Z*2rfM2ZTw
zdtXO!F$}01_m+8T_M51CUYOGKv)Yd4l&8t4w8sIp7^Og;BvOr96pv9ZJRr;nEc)we
z7Y<`KSMZxl5Spgc&%Mmeg|eHc6LnmE3Tke0thKD4WhE<09S#3_eNWd5daC{A4$@l*
zK@+!!V$GLzM4Qt$r9?8}J>6e>w!yHg9yLlp^?vt`ze9xzJj%|7tB0kY(#H|btzbuI
zUp{+X1gv|`y?hwsREC2oNA<4v3i=97QlfyJn~0M;BzO<Yi#}2x8a1as^724a0}ss<
z5EUHwCnO9c9iTx5Z>9B6Fda+27wA`-NN2;Fu$2y?P1q5UK*_Xc6mxmgo08yyc5>la
z?CGEawV!TMMt7o~QC&(sEy`?O)pZRM-B>BqFw*<BXWmY&=}I`+sqlxag!3E%hIqoY
zOb5Uk{CHmE{EY}FJa_KbOruI%0wOtnB<>n;6nc0EVPzqARU8Ldo&1a573tAM4RrS8
zg@yw^XlWHPIT7=RL(D6J->lFMN<h+8gB<^yGCG@X(iZgGOcsc^13m2ut#@a6j3yY-
z&xIR?u$_MUAPJO3Oa6Tm)t1FaOH)3*Li2xJ92C%?(_YysZN@io<00YaS*D~jA_sMu
z<>n&Ljle)2BmkZwd_Rgpm`^s1_L?p774p=6y0Bm<({P(`dy&vx(hiZDhk7HUaFa0s
zt@49H>c|*u@@n*DbRA+BOhW1ICOnOZ7#qbmBZRdKeTDq;<(ExOcM_3IijVu7*4I4g
z4O-?L2Gy&4e1$oz{2+UtAI418<`XF4_C(;#{xm#&lb_gkBSJNiP;ugONbL#`m*CQ~
z&w1J|G^|W1cM)nN5rdP3wsyyKi=%{K9nk_;g%l_>$4VIt2BVm;WU!!yQ%8C6@DUPb
zv<>Ii&FOba9AUMT$H&F$t7QXj(7$V%3hyn4Ks4AuLj!L!a|ySnwb{Tg{M3q)KmV)|
z)48_7m#d>hg%TZr3&p6;zm|9ad&}C&$ffoMW7#jv!V`gyH!DeD>};e%b7=Gj_aK|h
zKIe{4GVIKt6rX{K^$+m}Bd}0e%5FgwoWZ3#SoaPQJVHM@FJG=NiBL57d&uWv@18Uy
zgfh&mL6&2fOPp?}G~J!dWR4(49QF+8ARIfOarF!5<5vso+6G}7af_h;LYcUt<~dqW
zFU_14@lUJ_UOc|agH$9q)&@mTFO;4Z2nxT>RpO9^)Pt{uv>{A;a=4TDDT0drZX;T7
zL7g%hegpRgaOf9TxA^Cv5<7<9*0pPXuQ|xJT!1EQC}Rv(EZW_tv_?&jE0J)$cJiq0
zTJhI$lz%f8CCQD_Km*Zt_94_P-B^t{hk(NMSzOdDgmj~8ZAP=(U59%h<w52ihQ3Tv
z&RzLkU<}BHUs@)j=%_S;qHh2&MGj)S5x_DbaUqJ)vpi`6{x0?*CwicmaIjUR3v)0q
zzb0i0dbYqFP!Pe~?h@|ViPj33^1?6iJy_9Ycca6KSe)V?)LZzk+Zlqv#mow<!>7)<
zk(%^%{=FxG5#9E=8I-G1<m6|4<q`pgcajh9Ac@T9Yne?=eJ~h&D%SwO1R`uuGJ}W0
z_~B#W>Bd3kI2kv=N_hE!hk+rEmJ6l_W~c)%oZLajn$0Ho?|PO>DO4D!k$xTA3ArO2
z{seo;2hkXYe`bsDBlZU&zEp~#MAN(B_u(CM;l$ZDqem{%z#_Z)AXGtkeo}~`2A0^}
zUgLah7stO)50x4KmXIVlU5MB#(dZRTPrO8iK+Mq5{s40u5HxUh5NI^UUH6&+t<(7{
zLaF;=R{U}k4KY%;)19!mvdtts^i34h@kC+tb9#8x9YH@O^yMmMu%wc$u$!gVh#4eA
z$1^rW@y@T4>WEi(<vBLIPcaEtX_#c72h?11hUH9<f;SG{pkkBaHlgCte*s*H?13%l
zqxKJE(TVm?FniGti@!z)!))X;(Z8;a_OXOz*lLSFbVd789}514K1W6iZaPZKO(lhC
zZ?T0BTLkmz(|3dloPzh}ic|uc7ZqDO{ac`fh$jO#B*ziuyYJ!y{$ddMifqMCDewKE
z6Qjlv(ChG|WxQ(sRRJIFBTSU)F4+8jA|LV`@g0HoT?uYH_t_VTc{*~_N@wdND5)l`
z8H#$)yu^eKF^d3WQ8+f_)4!UXT|pYDm4ic?NT2{?mpKUpd{4P~r3iI7vLPP0k3SRW
z|I6qd%_fSXMTpYmGN?{=wd|`XRwYgRnqu+1H23Uvz6chap=K!5B_R$4<T`w(e=GKT
z@f<OaZkeL1hw@Tp6iy|?Nbo8vhOkm!4MX_3Y(v458n|3E(dONPkqXxArwUi^M?oZZ
zfrv*oHD{-0<!wd*FXzQKyrU!(KA-|piw{ymHdym_s6=~P5Xq(o-9PQCmA40k4daK<
zCxb?Xaf-flY1XZ@#Si+@=ZIY-bg}S0laA@Yi7ZHDTH!ioLV{?)H$1Mjj79DRr4bTs
zy-&!BjwCabd-6?0NHS850{BQ}+g839go7dw6yd~9Hw*@tN^*@5=eo_)D}(lI2w;rT
zyUz#E{~$ym`uFPJln?v>DH(ANU|7H4_Qg;@u%$Q$Mhh&~WW}(-ITI3)pv-WHZ8oYm
zj+q!%$-x0PYH|F|RTmHC2ol8u^7<Gqdj~gJkV-Gz&!JSi!LKo55*pOV;UxD2_N97#
zASr-)tWN2-C18>9kYhtD7GcrQh#P5#jjZZRd6VKcZnskC5g~2imK>RbL8T!aAgyx`
z5v=X>j&<Tv=Vgp&JP$)VkYMIJupgt^Bbw|cJ-!WkR36etUy?G^(?w!D#=bZk=SbVy
z9;+~w0dW{IzeQ-qr*9I7>+2qjD6Nv5k-!Bn3?D-`5COpT^50u=YaZqh*DX16s1otw
z?q1{~aNW6+<L^lKzKLrpOr2NRv?|6@pLEF0a*hzLw7?(!^jqG})n=IPf?l!fu_f;D
zvxIsw5I=X8jZKU}_{TY2W-t&oWkG7`iw+8~`nA!xMUgMYP0$`<ySvE9A=B5Zb!9OC
zcEJjJ2lBpu_6;w8ZPQ>wjHOJ9bnlWQ14||O19&;~1HUoFf#D>p(V+l1-g(gvvULb%
zGX~L-)2U4~?7zu<Jk~N6?UV-T?1ovNzL4#5i4IG*xk~gf!CvtFoh<_aAFPzc^~}7-
zaqv$F8>DpEE~O3=UYr5yB;d$`;aF=g@*j6j-!Sl<xSY(uZ}X~eqj=nkIM8NBKwM9t
zZ$hqeL)oE=f7@|s81X`Rd8^eiR?Rvgu!MA0T}pM>R-SA<Qd@ymrsSJ<GIfu_y$~{D
zg?YT*W$SToH?IANoN6v`slx>xhTgcg$$z|j@l#$c3tk}m_}odluBiHZ`9SzYwqSH5
zHEi(KlYVH|iYvg4ZgmpicEiDjPup2vj^~QX&LV_1U0U?I6P|H(WPB(?3=8JBzT$^e
z&xygRGOscU@NxTDYShfKz2~O(y|g``!Hc^==}Qe2ZiCD&r5>DBVic%Q%Q+GG)cgY@
zh&ceT#JH~vt{N{>*!walkgl07S-ke){D*#RIXas7amA-O1oQli#A_j3C_78q2A!s9
zdP7k)PSCj${H|nLvUwqth)g%-03|c64-0_fLp-DK!mxnmdGRynM@TTG7Z?n`!44$J
zy;#y}(9;83!=J(IB$SWfg!f|9T$=>5`dL1KZY`M&1h;IF+$Tsi=R@UgwlOqwf%bhe
zcJ432j~?<SL$vx4h>te@SFoMN)U8<xJ!oMWcS}N%j~)lAk)^FH6D_n~f`b#p?|evP
zx~H3-4b4%$nyaGAf>2#N|A09N3XC!l4oHSSfJ8rS;rQ|c`_PLFBkD7`+Km+t2Q#AN
zWuCvFD+hM!RHC~{GU%~~3#>RRil03w`i@F=4N8g^67u3eoVzG#71{ru2pphFLa{=d
zBOx{AJ>mlW<G)GJVd{lj4}2>3e7m$FfM-Q9EG8Q8X~?~>S>Vl;MUkQKapt%&*UC^V
z<MgDr3VwKqHe%zlvIyGoJ*8y4TLTP?OGebp+j0v`IhwGcjxLYbO}K;8<1xva_PAal
z%EB%~RxC&l&UbIccCC#913Hwm<6389)nLM`eN&<bJ5S6U`#P%Doh%SgfD4D9AEdb^
zPfaNf68q+xXJ4Lw;7AKO2X>JvDBs*ol0bruAzCUHHuD;qTp{yma~h-$_>xqZOrg#F
zg;rnD4QkRjMzETbqoG>rgwQdqwt{E<FrPyM6M>QvYP86)K$sH@m7^@#DObmJdFVV|
z^hV${^S7B}f|lWCMQDEqDhmJRp1>T>2F62=`f@v>mzk5V?}+Bc$kQ=c)NuWA1LhQw
zgTc|BcVQWXcoP9%*S&U$Omq#fYzu?lG<5o=lH`!e6v{k;FiHKWpwTkn%gN)$?Bo5E
zJ#2_$vN(osO(a<zY8a5t@>N2QQ(Z10AE3I5>uO|?!Ti=KJ^ewp%Zk=5T)lT4X%Hmy
zrc{L?OoQ$T{0_y&ehA)Teu@)d<e+Xn<DHCTz$7d3zBiuv%wX5yg-hYj-mE2DwapPB
zjof@2z5}svbbJs+_35SJ>QDh1g&5dS9T~tLiX%d7Mh)GLLgTuFxZno4iOBBBgg#U@
zHTjv|lw|DJsfv23=)KRipS~%>qKkW!GCF5Z9ng+&7rd%`A-W3^OhyZ|fFZ7)AmP&5
z=MOkAC3X`RkMmoM^0_bYF}lJ%^>U@#v-N@wnpd|48!ds@sBUw6pSj>&;6)G|54m?G
z4gohCJm)?OO7`^OoXm8Cpnn0pP^zNmmAI$W2Zn_W&u(cmNd3!VI$CV%;J$Ru-+{Lv
z5-vf_!|&}zgCfjJWR98)@iQ}~ROnp|1#KBDG-KG^82ACNoEHmAL3fqC&KU%<gI=)S
zzf4#qYLt_JW)ahjv|iuKdK6d$=usED1SW3O-KwI(30u*qClNkeO$~jeq_4>{%bh|O
zSD|`>sAW<SZej__f${=h0jIbD`IbK;wJxr4dp=VX;u0nz1=o=C58OxmgZcu#j2oPa
z6A){U`&#N!g^F;qk-MuN?IeY@>^>J2I-=fYKLZatFyV<JmZBFmV7<-0Tgxa7H?HaR
z6a_TZM_=N4;pY8g!gQTE^WJIs?IJt!#yBnO)yas@qxMR+`Po;`zOJ?0Q^tD>AzVI#
z6^Y#(Mb%TkQP5ZY%~EljvZmKCu_IkkgJG=cvmBvNicO{H_+f%Zh!M@3_p@yaxKcU?
zmK51KeP5my@DJ+>9sy1lehO2roU0(+YpXllp5^P;i0fVy-N4a}LUTV^<F+4e3abEB
zu;m@9hu{&yJ?IKlLJT5vbTvY1o*Mm|h?bB3^z^&_2AR}+k}=JnyZsBRiO>oJMx;OR
zVwoVh7OgrK=)<jKL_ax9cHV`)K=4;3Y1L3Awqd!H`ydM}W5#Ng*OXEROowBf6s`VX
z1{HjXz~6IQ$!L};5Oawr^9ji=bXJKYhJ^{ob!nkH9m`5_Z>%mcO6g>JW8x2#drt^g
zI6tBK9}JG4e(qjNF#5=aVXj7wHUERS^<}IndukfaN;CSe&%R-i4))J(?VwH5-O|sM
z9xC{i1G4LncOKXLq~O1HU0GwF&J0eN<7|@SeVG40UcjJa_d++)d4SmSXe%HoKEFfy
z*KbITHX2|@Ay-1|Ws8NE9u7q?Bu?Dl$}xDU4?Id71OA1$eSifEIQo3*;-@~&)x|YU
zXMb!eMJUCg_6eJk7?xc)$GB0?o%a?gvz9JAj?O?uJ~{u(a+eQSShv|zwwJq!BW@7>
z1kfVY$l^?uBRkhN6hq~TVZ`PRFv&*kdixo(-MPAI==PC&WDFo6MmXzq(jgy0G>IM6
z0(v7Z03KOgYDc}?Te2Nlc7GFrM*FyEf|Ndv^KX^*W7{|YrVJPvwPf1rt7%}MkDAE*
zCU4a?zLpP=e1(R^+DOV{PJ5F9Nsj|%E`B_2Ljb_6<(i-ny*2-#S9d<D(-}U@je*(a
zTtXfV)N_WTg}BVX<k{DaS@t9W7Lz;LD`Nm~J8Y9u0L54m8`INwJkNpi_Q6dJ%c#ws
zl86OC`6t>}(yeoVyrLKaJd4COK7GRk7PSDLbSEz<Is3EuyG0?7t7t%6eg%Ex+%O6y
z-l<K%&|KW701hF06q!ig&yUo03swndNA{G5Vt64cZxDFmPpCuWA6$RKKd9C-WbDi%
zdRt!iH)z5abtBu?i2|suxR%quU12_?KO^xh!iCRdGGt9*YqVW0ru!D09=VP4-;cZk
z^u)<T3G@O<+P~gMYCkuk)6>;=MD-;_ljuOo(CNC1$bRX;2f376(f1N;6gItgadX^e
zYCOv*VEJ&UyU!SMDD6erP)<9~AjFn@FrB6KJa4GP%e%SIv?7=jy-l1JrWQv7(W1IB
zs@LV_5HX9<{>apb=1ccfa?F1}I)u!6@`!*zeDp1E<!9xvX@`A9>ylZ|-!&bxc{Q^S
zIs34_Rg_t18^Exa20dV$#wXo(Q8^Czp!1I7EO~^*q|p_fc!+>qq-q%C&`adtSy-?K
z@HQI1OV|xh2<BIO(Mu&iv@=bo(=~gpp-HEwf(tl(Kix-r#oz|IgP{W{kTw~BQXJV6
z(~|eXbr&GmzCmV-HTYq;kaiaP{hIpiW3WyJ&A{iB!Lhc^=kUwt3Cp4hGH7uom5vd#
zTp(CUiX-?TFozU5Ut0rr6L$|1KYE8@Oy>Tkkbyfjocf1Gt~oO9S|CMMNw8=jqNcPL
zepyG384KCS1_5?Wh)a<9N)X8k?hPmCDADGhOBXL0S{#DNNTly-;f0JoQ9hd6tpxcp
zQr}>G6BDO@QK$Zbx1i*eepgN>Uc2J)fq{}^$2bBalIbtmJ#pF+a6l)9><+~>{VwWK
zO3u!U<&vO2(l+?Oiks_jJ#}0I$KY9P<+A+ZNSd{nfgOqrK@<d;E;uyz0TfiOn1EB3
z2%GA^7sM;}>sJlSWp?lO%TSfFFF&FtsdEjB(;Ja)>|z;rW1Py*M&WT2sO7R_;O~hx
znWY+MyL5?JKH$Y^#P?la_lXXw1o5Fu{O@+=f8ZKRh@U(_sBcZRXM2`>6fu`PKWG$2
zlzsHRXdl^aNqp!R=GH&kt6MHWVq7>&27Hd-ewHjHTqx~N&qrr35dj^6s0ArU_X4bx
zsOR&?jxb=ym&&zE?hR*RLmbzi5SKBdi@WUQkk^$Xc>UR5^=3eWQrzb-^|@$LjYqC;
z`Sq(AKEO$f(l;~^Avmzbb21bR325{-)+6^chlIKIdwSZDW?U?xwFBow<UzA*MZ;wj
zu$d5@O$Ib-v<C<UpdvFpyXtJYNCjR+>I4^!G$Gt4W?ld)f)$kX%jUc^Xt?p6jt6Jy
zp4r(l+aE+N4xD|hl)W!p<dUGEl=Mtcy)}7^z6a+3bP28nGwCnxU4F^iEOi7vf>bq7
z<5>!=@p8l^7S{kGOCr2FTSy1tkSEYf4JNxR6H{;%_?yFFee9KCEJR;mpLy3&^6ev6
zGtJyDz|ot@qtIC4XU(^not#$+LrCQj-N8;9Gq`;aPGHI4({HB*jQj>f-j0$_!*v0H
zogi%0Lf>o36ai<-7TPzmQmJ;t7yduP$F@OSfK<yr=K4`|l`e=|NHYq5IJ&{<)FIlW
z%pE94#js>853$rnKpyw!NkNB*3uhiwNm)XOnTF1QMk&4@wcFrAbZ-=dIH<^A;SN9X
zkJC4zxDLrg#tOD1FNmkQld`kz`a=0A@DXjGjGlC){?HzGY2my&`(5$}ZW$Cm<*4`>
z#OjjoEplXN1(iFe;TQ!b{W!b1khs@A>o3qdDm@qI;^3nkjfUX(6~HJM8lkI3>Mw}u
zSP<P9K`BkktJj25gT`oS4Wa1MMAjg&G(kXIQ=Pt*wD@84@4fTWuW(aN-^>uI`YD)J
zalw;^nM`MUd!|WHR&Yqed|r2%S8&l8_qf@LT`Rm!+)I;u?s~QpK!z$9ws&-f2x~O^
zM)VUdCr*xrhrbiiAgJ)^+h-$CAfUM48Ht3Rp$sJWlSI_JIV#RwHw<O4%2U<)VNP~O
z@RD?G1LuM=2!qs127gL95WOn$L_q~qD8awVJj3<1$;?0uP1;4M=EjNl@<(mt#*<;o
ztBVh;(3{2ODl?{!86tY3#irm+5Z}J8%5S3`E{=N4ODnlwU^zk)?(8^w-7r8$gIC@E
zpnGwo^EPgBYr#kQ*-tAfZ@UWxm!*}dJxq}D&f&o}rqL+|0H-%~r&M3TsHAI4_75hf
zGv)g_N%u%=Xxxgq(*YW;zG`2J5Mx)eHE)F@`}zh?$Eu_pxQ)w7#=X=CnO|-^vA#Uz
z8VonD7(QV1?eagMp!f^{-KyI0yXgF0?sXlh-W2(a-YNi;*=SI`q4asLA=<!hqq-Lv
zUd07~PWaK7L=$wA{GCO}3E^LrF5!XT$W`hA*$ypN11qPFc;DGSk0j1@7+<AWj8Q*k
zba@55rZ$NZ9ASj8W8N_UPKC@~Mv6?$o56kY7&TijbBxpGnP)?CT_XMqziJqnUry`N
z4}+xCJN2$=%65SVg8^NmpIj9_zB(Imj5OTx3fBk*USfP`!9}20Q#wt31*I@@m|%}q
zmJIT87#!F>9fb<y!5KmQr{96g-hfY#z8cOJgbKsqaQcRv!pL$XS%%?b_rzLWc`^T^
z=*rz5r16Y%l6fRMXtzhG>AvzJf%(V||HK0#Z0H~<fmyR{e#hgy=;QTb$g>8AN4Ch;
zteA!<R?SoN*^TvF@Il(DyEQZ{p6)?O#IAa{SupZ{UJRj(oTP6Q=oU5Z2q>8yfAoew
zT;KMG*Y0vqc*MM?S3z+624x+FVR8?Q@HYF~TQfx@5Yh09yGZ5mnE&MYB#=Lh8!d-m
z-hK~vMfej=Au)O#f4y;C>HPees&`3S0sE%w+B8Ud1x_c{J5@!=$`97{@l~WcLgn!s
zNaprF3=kv269f#%O5JZAxhPyM`V9Fw_0j1{$^GbCKm9GQXzj)PkS!p1x|n5|#}px+
zZt*9*!NVqPRbC3H)yYfc{w#iHtUb>Xian&QAiR=kJWRjCjn>BHVO^3a5LvvFSB=WS
z2L)RYqdjJ>w+qn%lyGsUKM?5>hx}(fRnd48_s*~l>)Gpi?0wnHpTFNrLIOG*wf+ZI
z3jM;MAfZ#6e<z}zsDk+N+3y+SgJQVcmvM!SHVcs;WLe4*KD28Yk}{(*hR<d9HgHFd
zxrg5U*%eB0H+1*`U-ER@(IXa2VySQ+xI?0w`4x6m_jw>6RrTx|viZFZ?mowXW5icy
zuR(1fQ9?dSFM%;z%6zW4j}qoW&0mV16&NPm=?7#p!wxR<!M@^9HBzy>wCj?&@a7r#
zy?WQ59(?e5K6}qyu9sm`{C|ACZFA&jb?*BK^5Q&Ho!1N|o&kIO-e5@UVMdyfMvA1F
zfiMnkwOUf^sfDDLX5>r?Y{<?|I1sWa5<<=fC*GBKS9zDQjWNbRRT@{Fm(o|@lbqjm
zt+nnJd2vX}(zO14|L=!&ug7bxwUOFP5(2KVd>ja{9w&9Rma>ZXIRCiX_dQ^m>w9)&
zoxuhzKNqxaU{mr2>)|MGRMrHN3@mfYp;Sq?!QubTV~jMJ!I_H1<{HgtHIXmTzDb|m
zqDr6!&%ZONmgZT1%B}cPv4|<$`8AwkRZXl$(was;8XeX$&&Z_J%#r7Reet2}^eDLB
zY6ok0r{U_0<F~dQIh;kRwDQvm;1T!N-*tELeeh^(%)k<`-s`XEEu_@|=gk`sArW1a
zZlknKHHpt{|NS?9n(NrVQb*wWv)Xez9b48!c;7t1H+wA>5Fope$xdk$7Q?s(Bf%>t
zF=;PwNOKGe%|A#m#gO<=vw;6_t1sTk-F~GyWiR>OE**IJ*yR4_dIW4(_J5dZEiUET
zNK{Iu%HM|LU*^YrMoZu1m$<R6W_^9D<+@Ih>#p4bLX~@r*4-Kt<F?V<+?O(MBP%Ka
zvUE?20A$8do>;7J1Ry8GI>@T_=oeqo*Ii<^!M}D1QHbCVec{)+H`?rTXj>B+HTWbu
z7@E~Twjhr1X*nM{P#SWkGDoI{zf!C_Py0atKo0u0kGs-z>ubNc_()3|z#U5<P=Orh
zMsA;q7)PQYyZT4DjghYH*~oVE0L-_XSva2S+p<$41#`dHD%B@3<(Pt&*Wb4qp}<lz
z4h5zQJ8*^?2AXJgamzt(J7kc=GWC^Q$qq-W5y{MGvX`55Ff#QZy2|zxa-xM_Df^|}
zhD940|3x>8z1M2kO*M;Wx<o1Kq0K@YV0v{mFB0Wr8U&gik2bXN0Q{#i)N*s@j48NL
zRxZR-CaEKiaMOHfodBEfh!Q_O3>g0)z!R9%EGv(s7m1&_D_^xkgdm%ZLFGwAZ4`)_
zuSDMj#nJQ<`KoC1AQHUsLcVo*q8&fPR;z=hE>$k}SWR2<JT87Be_wu+vJYj1uv4|?
zh+Nu*=OasKNZRx(RpUF9(!&|2IrIwoiOYZ}hdx^Le{VdmFD<gDT(Us>kWtQpCS?3W
zk5Aite)@tR$2QStpZb(up01Sq#$FczV1@`X0q;b9d}oN(UCugG-)j1EG=5Jb3dn9w
zv1z1b$2Io&etvr$??T*r?n}!Q-yc*v6=Ias9?D&mnrEVgB=f571~@1x)nB};H%_Cx
zY4#ix9;c861c^0?ozTcsvKlO;MqNn@FQ=n+t{1;Zs%W``Xap|7RL6rtNt7bL(s9!`
zABD@nvyR2_o2B_}fe!Y@B5<4ZGg8Rb(rzV-(hfK*N}LUfR^{NwE~c8Ag7xN)i#B=B
zta62NZ-iPdy=lF9#~G{i1^m9bXB2u*(mLK7zp2@6k3T3MH)=s%EoU*Rly1?m&;j~g
z`~+jTj0)goaT~s*EnaV~Pz13L3lLVWoC|m~q%uLy<92^tFW;p?lP7_~lRJO$zw{o#
z`z0duxUNj=3OG8j>LX#Fb8RYu=f|BDKq3DY0BjuBc@MV3?UOr9HZr_UGbddlx23?h
zIMZ^fTLG#V|66xbPz9J}j;Mm#WD{zT-nkjHqGl0byXmB)?Fcqf10*LE^`Vf_MR>mj
zckq7hG@Y0Fm6@p76|r*UIC7emHW%D>o_$K$GIVLSRK&HHvWJfqWmnnz>F|PO0dlgv
zc8dDs9IYFvY@dh}$m_2)jX&7Rg;JLTTdYi&;X0{zQK<gZ9$ghh%q48$7$|NHd{Y~%
zYMiV@LuEww*A0FnHfw)LF~eEHM?RFTiaRe;K*~BdhFRH^D&TeKMBQsuOHa3-5@sdS
zunbTUtX8snSvs<3w{qSpxT5M?V7D~)qEnE#IXQ^C|Djci7QAb=w7}?yYAFC20z1(>
zd{^^6?YpqUd4T}%D<^T?qVANQz^SMoPkQG`5&~j^4r%^FxT>*A%$r@OGJ!YbE6cM<
zlhfUD2B8=NO$d;~@U$`;5QPVIa179+hG|}}9g_s(uBh2z1nc!D+?)-`e(~#^Rw72j
zD4c5LtLG5wnm>0i=d;+|$-}3-Osl^(hk>6k4=(?@+OkaWnajh9M=x^>nnHvAEk~E^
z@)E7sI#dWIgdzS2$EfzYMYob}38W^2J!c{X)^`}s_cxm!Sc<R0H&;T6K98mOdwJ{-
zyX%x|l|-oY<L1p5a$uDBsyxkd#Qk4+F~27^t;m7Cusb|USH{$Ucux7Ud^L5g^$ZP#
z^>5r{$0Z^TV6jSf&z(K&SB=SwAL{)?gK>35+eFG>`SM@>V*p|2`YQ0FamOFE3)px#
zHek1S8iuwA7}mL)*Beg*7QRIn)yd0{)V&$MrAO7(BG<n%r2T-tR=}#ue@OaP@LT(_
zyqX^zQ(aTvMq-rhNaxI!5OMKOm#^o}31X)D^xBTK0d+>SjJ2Q<=s}(V@6^fo%_8Mj
zUMNY~$8lj{=m^-AKpIgv6x#+Cof;WLr*0=6txGbHE8G);z@wN>L@iB>L*DcVFi(o|
zLrDo@=M0*U2M#tY%Awk(U^01pY3{L)TnsWEw5pTlfy*NDj+^Hf7NTMu-RLV$M8!ex
z>HXaP&K^prHg#>1)O7LxT)4>FfFgf3b2=M@H(I{{JlQRtf|`a0paEYTVa?M*4E<W0
zynFTdyADLJ&Y4_yEEF(~HKS95s9t&bdQN?&sU8>q+|m|S=@4u{D(r1Pyu^1k6lzqS
z-r?Q|D<>C+5fFYRAczO038BpYDmv=mq0a~BiA|bDw!&gW0gkFFY&C%ddE4?yVz1Gj
zfxpc33-8=R=PE5`lL|&!H{v!dN}NW%A9Ca6D0fKPEo~aV7HQ}{;uBZ*)t4DP!jY7d
z(gKI0v%Fg5qpJ2|?$YwGT#KuqKf`>o3R{Ifs0c{GliZvYusE%F$ld6JKsTT|u-789
z^sA5ZRV2=!SJcYYf~%?_Qbr#mCa9j<ST+-eLq$KLMnV&WSvSc0fFf}sMm_;Bt2t4=
zlYQjwSkuL`@8^a0&MCrlD$0h06WNa6I(U+A$)rE-XsJT7j|%E#l@T|cz9bn@^dcy<
zk40cL0vA=(i050nHMEEXOfjZ9YOWcPGv@x#S+3M{9L+6MSsp7m&PHwWE%;KgclC+E
zIurG;)qySsazd84ju`fjA(4kY@#$moqL0{J&o1?=uLiY^fX(VJ`3(xb<}6VY)ip`0
z?{b-w4iCiR_1W%8xQ3M<lYgVq<e+ZV24XmNNq0w_TH6P4NaP3igJS1r^Q&q?ZQf{i
zcg}zsQsRT6W|XA-@Nm+agSc#pTNX_ce2YNsN3uP<s8#1dXV3uB<gsGtaF^?)c1~_C
z73Kzev?12(#_yF2q#&+!SCt*7C$QJG3raK2Qzq0Ae$Kp&kDe=n5hC1AbCPAcsuFOh
z?FU_h!?&#hEH;<fshQR4bj}V12I_C6^vD048z**Jx)@SLmAcL3xqF(2yc+)Z?b)?<
zBO;&I;1;XJ`3Gv8R%~riWkK={@}d{ZP#uO)g5+>BX~3ecULK19#8xv1$@0m+uX2~d
zl{9M1cb^>L2G%%6uM=tCc7yf@IL`|`92~E8GC?cWyTLNmCAxT{9eyU&<Gv(H9aWWR
z;G*Nk>)4O>UN&dBCfcH|ptiXM7eBrDdD0nSC#C=BJp#w(8i3Eh&Na!KU(E;413;k9
zmJ37lAiXrJGkOE8F8~HmJi1lxt5PcJy#kDCI0vk+p2;oVrstJxC~Zm8QAq=D(}Mnc
z<8%6NOXgN0Vv;tti2n5GM*bT)m#ctHlV7fnqm);EeC$4$dW9PB<nl--zT&^$o|py7
zsI~+g?9}C0C(9?Pzf-MtVL^mhywSS+^(x^xO9)W1rb<nWzo}VC?}=E>$UQ+gXPTUA
z3_Sn6vK(<RG)6LG%LlVkvzN}5)42GhKDdt=SS<fB1nc4(c?{iZ%{SVsd!$mAQr`0k
z)uAkYi7|r@4idBHb{g7^pfBzMyQiqNfPXx<u7lewQ~Rb3LQ7+$s(@!2ZLb`I@E`Pa
zlN^${mbn3UE;_kQv`h=48<sUyOvk9Tyna$Lg5FFAGBT2HIE)7(+}}c_DO!k{I@fQ!
zSd{<wffO3n0nR!SL}FtFy6~(=cc+!UbF>k6E%Ja%6Ie{1t@Hj(gsvk~#^=^B=FBCS
zo)HVBElhsHwJ-uYD&$aZ-&EAH(98)qZ7C=NW}KY$enFJ-fGAkj)SJa<A1dz@TdBpZ
z!WH>7pIzYW%L>V4w;sDlDF{a-36`>Vq^0NMa;G_}NqU<|iW6#1CK}_^Rby05^IrBY
zMnp;b&DBXKZumQ{So2=Zt*tHRc^r?*Q3ZKo6{JU}RWF@sX__F$^c{1sU*@+dFOt7)
zMeFyj!DTBl$24m|sI7B?TR0DTNggWXr=ll1kDTPE^M5(wt)t@K#247-TvWihv}jmw
z1*mSRO{PA>{aFgwCRlxqz8mNnoWghu+aPiuBn6_5MtwH=W##T~qZQORi)diA+{P}J
z>j0J`=D&gqBX9m)Peit)CPZq6Rkq9TO{y<p5tIiLwsKSkFp?JZYpVnu4h}KW>tD-h
zxdLs@h1gPZmfNb1PtHr2pnv(7xkemnC)+jilVR~Z)Q}IgE#r?YM&)nJ5oh|YphD&2
z`c`igxJGIH+AQb^g04+MA?qI`XNj@`(DR=C0?Q7}a^WuiTT=<Af)`x=C`w3-_C}?W
zkn;dX%{5Y^wMEx$uTtr|*gQ+BWlRxJ%!QK8hbZ-CpPXAnK7GDz*rX)|UEbDft0_qe
zpB<R9Bh1f=RS-X~d0-hw>2E1NBm!#u5$*|{vdcOo^(=!$$qihHw!4QX#Ym#_g?E$J
zA#7^5kTSAY;1zitWcI3aP7b<2s8KIep`IdHREoe_mIJHJba-@BzCSyzu)Tf4zbflO
zjH_>vIXXEPMJ+BSt3zE<0<Rwb<h5D{MFHzD@cEl_OL8>SAVjMbFPmZHKw`?1h}0Cq
zGM56h!+W<3p%<Yw^(venO_(0oCr+55un5{UZTJ*Whw^704((s;ZeYF?r&D44funOb
zpUDY1CKD?oX*efComg8{a0yR}n^@k4+f5%A7!G)631#6S6#U9@!Bm1$HrGW>tTVVJ
z&B2@gMyay0@v6|eg<<m6XJ3A<c9)t1b|~DR(pR@yR#!wVkQWqmDgtpSpa$`L=4&|}
zS3Mq8Yy+0)2rt=RHhZ{%3M~iT33~_%iX>(J5FK*JULx)%TI&KzNLnKRfsZHOtZ;;s
z{XD#lqXHfGx`*oLm){{?ge9eQ91YAz@Tpgcjm4H$fn!VKpd-_-QYQ;$uWubev7ETW
z?Q1KoDm=xguzl$YnLKvQxoP2<*DD9d)=F5_c0u*YB5lh-e~77VG*>S52Td<ueNo>B
zfs@Ny?*|-1d9vTt+YD3HojIx8h+ECGxQZfW{jYmoeIXSqpxQ{D@H`$3&v{8@__aKu
zCidN*oN3IbpUd4NQpy84huNsC*m;oue&v-{n$)D6<0KtHu3`N}rojk<5$>B`nZWQm
za(^WG-@=2e*M8QTIcjt_f+)CQewY)H*%`L?(IDFAu&LuWy-90xISmq5sL0q_(_>IH
zbC-l5JxlfR+yZ<(&Fx7n(;O}kAT=wBkb7%XSwr7liBu11JSA%2v9Wz!u=lw{VkcUb
z%44wdd>j0t)hC=rqbmKCZn2-vO*rTV68F66IM2NLC>gUnFLFx|7|H#>1xHCrP3x17
zD=MRgDOJ|rCtN<~wPAg2l^tpwwxSOjH_b?v;ssg;(v*A}P0Ws$)^TsKabQYUG*=v<
z&m-p<KYi{|PEVHYNkP>WF%g)dIM4xca0?%9KEQxlKf08DjdHAJgGlmP!eak%x)rz+
z2Hcf1Nv09fwCxNfm&ke1&X9Sh0jEc#rbMi|zY+6%YOuWWW@|eYYqn{m>{orc|8=Vt
zVbK&x&?l=lSjmVH4#mCntJdTeL#cdLnCq?I1xgd;9!2A;t{ey1B+X9q826;aDO?TL
zdNNU4#I2M++gOcss0>?53NUQ0UM^!5wJsDs#aylOymZ~y8@ES3qjQekfQ{=j9S`z6
z`2dPu((L*W!Z!y-(1_)b_z+HTR{ZlzX9$qNN0Y0h{N7PHH<ySba{=^aIkIv&2zw~F
z?g@~SS|(C#QX=r;UoL)=dox`3#7w-6)#_#Aj|kvY)($be(odvY>ZL;zqFhA}Y{g#Y
znXyuq62+vF>L|lkOz}R1<jIMmT1yF)w#}gykhs=XddH`KU_A>atDQs}=<d<Kr^Na!
z;ag3Ig}Ykk@O1o!{*A8b;@11em9=~uJ;<a!!xj&%|IJGcOSy7qRV4Z6uU^lM5muo5
zjbGz`S`0T;QQbUc%HZv`pa(~hAD;;?URsn?5pzYdqD7kLXSu3}&Fb@J;vUdX@~^A^
z^HmE`vK-B%>Lwi+2*`)1ikdyh(3Zuy=<KpK8TWZ8a=05o+UIWO!Kl+*uKw&yuLlij
zPXP8;rI+EIQtfYoi6RPI4S)#sn~{3OSBO<vXGQI?<8Oax?j0edR_hjxPR_`D`BoxO
zqn?m`aG&{1QL9$lZX6~rHf3G_K!ZA`r#EgUufjuhV4Q+JrWHpLlIo(E0jgtYbJ=a7
z)8>kQ_nn*wjLT{8GgX-yk0AAK+8w>s-HZM*Q^xM(_#55K8oc@Z^EuEeSvzga#?S9Y
zed3L{-p9kkl8#TvE3^CAQCAInGQ=Lm;LuTYvxWsQfE-hLuzpp}2*Se$57T$O_yygH
zB-|j&(+VBFWj0^99Q>27<pzAs44Sk3q270Ez9qjSFh0&z;;HZYrL^Thn|A^dwyM*6
zc9>$+F)(4r+=LTseV);mqr?NGx-#@yVYh*}#Z5|3u~~@2jpF}`yA&CORh;WTLwM)Z
zJN6SvmC%!7kC?WgD{==ARZb<$MPMNw7Wu4(PeaY58~SFtNaB_)NiY^gL=a93?Bk@V
z;QjEhg?a=@(xjMmuVrtk=2UA*L?}h`;07rTpwe~-(7}5|vd4V~NVc!o!S@e1-=y~!
zK3j#YPy6S|V<SbBfG)?MTcy?z{g&e?*vimOyHDO2zpdkGZJ{Zc!q;+3*ZR>^T(Vn}
zL|OPbK%Q^b=>tKjV?jf6bNkQ!r~l<Nd)@V%$zlWqL{sF~mOEkmR^hTv&dlGtTQ(%2
zjDv0=LMZ%5TU~<5$s;W!cyCBKw4svV3KO-oh=Z0}i&1FZ%RFc`d?GgsEu1wisA$pJ
z{*v`K@l94H4E?D34rA<T`pMH3qK%Z)c|?1LBV;J%PT@suo|8^X4`hzwBe_LuXAUHz
z?kAb1W3dG5EZkU+lDd{W07a)*_sq}%@xx6W=EK#JQF04U62>vki)=_z5m}Dk?nOJj
zE_X54GVN(;oLg&22lEG8wq*u;C0WkS3@O<;o@6XggFIjBib%vMazF)2CLA`~W}V1g
zoWnD(0;e61h(aO<tnE`iFGoBl-H({6?;TQu*j)r1=gV1v(;JU|IaiRwxMLl>dVFjX
za{NYkpQi5XPCXZOm?16-;-`Kue?q1h=b^1CvsXJl#N!;ijR<#9ESY!t+~5F*dM@Q~
zQclms|MO$ALd;=QLO;%oDhv|^M@XF<jBb&EHbXxzMS=qXRHKQbsis(46RsAe)!34m
z86DT>KcBmWx6T=w$qiRbHZFrw5k#ONgwP=)bC_lp5#Ab=rbTb0rX{TORkJ?VqW&tn
zy?D5b`Y5TE(_d7PWes{3HGy{f<;HTr6&O&mfoP^@BIFo0i!R4+m9_rIFKPOyzu|3F
zia1v+SD&%BmWqkHK~w5vmqcPS%|}b2FAqK>Xmn_d6p>>+On*B2l+<%9mv)LypvhCB
zPUlYK3#5j&_R(MH--_r)pTOzT*FDE$R;j1HbVFYzr8hUj@n(J0X>MXPhkO_8xrl~j
zqvA7j-P%c{XUpxJYlVQiZtzi5#nSXDd2Wy%)wLSFfm#{GVA<QCoa5axQ7IM}pcE-l
z@F=b94^^oC<2)pg1eBTgc_L?WRnX=54Xx_w(`U4ggkQ}S!URka#08C(Fmz+nY)NHr
z_^!ZWFlnIztTr;sg>JRi!YzXOod-=r!YIV3S@Ia^+4<Y)(2x@7DCZ?VX0e=Y0(|Ws
z^T+B7lY7SKVF#S`cXW6ZQTn<$1bsk^SUm~|@F0=Yyc1Mq;JvWNJq#Wc^#REDI`mdI
z4@*bu)Ewnazx=Cw?J9v!J+BY(8rbx>@shK@*oH|>7xugSG~`F+8#u8j%)(^2_9b)}
z>z&Kjix%kTrS;%<jE)YB#PQ|`(8GpJ(U%Jss(E63=eOqooNCU1py0jv;WT$%-`7b=
z9TSO9S1FQX8ZahuR+#9jSC2<Rxq;EMfS6_Oi*!O%o-%83I`273vf9dRvNYlpA}i{7
zrr@9p0<~N;o2=oZKXc5vC=wwf>x_y?*3g&Nl5PpPpg)!1Bs;M1JOuS11yL;_7h?x`
zFGUgqP=f|xTTwkllFH7m<Z@JLf?LxmGPu@;%5vbE*=VP}1$Zi#z&!wnnuCEB)gHd^
zNdLWUt@2F8lJ(D5ucuN&)J?1op08ytyL`p)AW@UncsN;nhYiS%S0~3m>Q>RtC%&`P
z@YB5Rja)yGm8N}~kI%PgGWnR%hX}uN`Pa{I{_ESItu()pZwu!*te~K>yOn*KE5&7a
z(gXrIKsq+Hm7X}@i0ikEmGxn<BfKpw)Bz5MJj|9#0FaEb>q(6(X+vT`*c7IohE>M1
z0p0;4u|z`FsGOvwxzEh5B&}@VDk|pS`cPh5cRxarZ=#3ggH*4MBrHY~*1T7h8YnJ-
z#LlM{@tc=mqauXd1f>_yMlokhj<vmmgPrnCPZ!9mJWhiOWOqR)hobUkD-bX6(8b5v
z{Iv%;<UVDv^Gb!~-n{;o-_?RY`Mw7c>Z>O4{fMc7X)&z!fE!hw2(IK6Z>Ahd-$?e2
zL-IGI{UGfwex3uJVd>?o>CWL^k@-*1GVZ9T%NKzOAwzV@<P6TQFQHB3?UaH^mbm6S
zIJTB=`|N$K^huEg87hGc;~Z9Ja@&c)jhqCp4|~*m)#oBSnK~M8cUX`EdDQ4$P~MxR
zTS@fr@cgZFPy``Hdhx!#zt^o&srRq|wFES*(hevHPz7Q_xq<O%0$4?A$Z8KR6WbD!
zR$pD{H!t~KRXC98+3J<Z^O`H`k?iCxJA}b&CU88l^Um6QL%jfY5OhO-oej+m7N2{X
zW{0PvE~g(~zFs4fzH9bG<1>Br#*Lf$fuPEZyV%DUV&hw;lpn>w)Fz$6$y+rpRb|ww
zh#$RHwDseLOj4l?bcy@}!Cy0tr+-u=!fEJ^<SP+p2?V-b2F1?MplF(W_{chne`yZk
z#m==Td#j9wUGV!_NN066UjlPdf2-1QJO8OKFil`9px4>6owV9XQdsO)lnjqvDYa7p
zDeXjC5fSJ07n4U{{A(o0UjBnVzGq&7V@=b+kjg<RE}-8WnD-OIR>I?8{IAZ~8UOo1
zYv0-3!xV#AYLD-Lw$M+d!D|d%!;4?$CSV3d1rBI!XebiDEDYB5_WnV!?*$<9`ULr*
zk>HX|Yzo||&veB+x*YHCm%^nkm{LgdsE!q54{x2w99K%7F(CY4o+Nw(=X&px*OD6H
zKWb1b0%mS&j%R)0OF4hXyK;}VBj%@6y`cd-h9KuCvVe|S%MmWK7qk}jml&5+vOM_%
zmA@r6r3Ft192!Q1SdKQW8x<eMeUP)Kbn3mF-ZM9~89eX2@q4p6(z&Ia;BL_L@YwN8
zjSq!rYc@%A(Lwm8BLM86oL|wH!$SEm$CXJp$1DY=mx;0F3znjGh@C%Ki*-*;nsahd
z6m$G`Gf`kP?`Vh~2de4pbxYhFa-8|CXPvxQr30UP=1K@lIsQ;tZX8<B-zxU2?`@7h
z-fr?u0o9}c%%e%=BtRR6BC}wEG)}J9Y^JdGb3qFcLVNb*vPRJp?2ydkUlk-lGY4&_
zAHBoE+1;%&525XGd>VRyt0IS9A9;b8_EF)T{avhtBC5aW=`-?OMLwZ<FhsaHG24s$
z)veg6mmAH>K{-b0NUFPo%z|z-7+0u*b`~er$GMBX2Aca<T9`<PQKHrQ@Wzejw4Dl6
z1V`AlkY7vZtsbFtW-bW?Np3T?d|ouM0nI!~6<uMkd=%=>6zE^Q0$*LofzDb-{}#^2
zrtMkQ`h75+n=Yij#mh8yVdKfJg;TEq$6)oFcA~lg(bJsX7A0@-sf8Q?gp;ElQ_wk>
z%vn}4$~xS{{;@6cJ&uR59_{bt4=^RZgF*4V;v!TG(vA?7$w2`L2Pd7BIRf^&TN`#i
zm^MT^xkHBmrG*?0S_vZ{I;AP4@em?Tp<|ux9^lR7=65i)S53`1DK+WI*$|X#xwXqj
zrkhuv(O&bi`}gt?SGpxWRHHf1^M0%G{l)7yK3~fv#+D<?Lu1!pmnLTaYO^YcOlDQm
zA})2@vE?AVH8lAf6JFzSD)~NsBl51j1dXoGAi&5WOq~wzr)o~XCc_o>Rj)%yjisY;
zvFtd<ePXq&d(N>Hnr&)S&X*3jQF#8q-f-cravRoKzw+5=jD+ae74}>Ey?@+#x_9=9
z=7^S?6?9hCauwVo(A+s=1BWmmbQv@TYDwB>&9*-t$<WsX<fLn|yDB&KcI+KMTH|>N
zUOGUy=>Yq3f8|+hu!wT<&sj1n4Mo>rzql*d7CXzjI~7U=C?+&`PCQ}?rYwr`=kaV#
zD&XGh+S#NgR+=qA8a5`sD{tn-FY`Ocqn<-H10{mzX9B+EuRS8onss!;j8vDIl^lrr
zBI#ZfqA&yn9$@Q9C|r6miUfUp^J=2ef(mF#@?CGt_=AIiqo1m`R0%CAOgGuDHFeR-
zQzzbSmexuD^TMhxl%`S64&?C!6N^tKSA?1LPM*US4M~fdX1X>79K7MO>uu3lMu$h*
zaVoWk<qrbCl>=gZm*1F-W#y_{<u}{@xxT&8r&wA|e-!S3-(LQ${-m_4a@rgHr?mkw
z$)O3?mxu&TIss0;8_ooPFWj)mtDr>U50k+FnF?5og%Okz;)S?k#W<!y)*)w$&pix9
zpVW0tOTC%Pr9fS7P9i6?j6Bdv>>l?e5jV^2R8nS*H#I6>Yo|o!{9XajsW79~X&<);
z8J*}wC&m|qYh6Ry5%l`SZ*%KzwYr~lTQDGb*ei;qT8`B}T`ViaSd&;Kx>7G%Il&yQ
zXV)mxkqpm1#Q3Ho<GBzV@eSyCmp<tlW_rH7Fq8VxK0&zv>QaJDYwj$NOt0&UQx)i#
zp-5KSRiX;$LAt+F!$hduc;*kaq_WlE@sT`-E|$8Jz47^Uy32~D)rYQ>Y1c<HU%64W
zP?W%qBCeSuQa<;&{AK#KPgM-Kj4BE|mF&Pt(Ms_g4l3i9EV`4CJxntd017p2y+pCo
z(jcBDUJl?wt~5KfR4ha;CKJ*+T{GITNZqVRiIs@}Xe1?U4%}D9bzJ4S(l#pTLqtj=
zCS+LMDGJ@~72Bg+9V$~U_;y{{dv3<js&`}5_28m)wTXD;MCq8vN>WmYY2snF)1b~C
zpH_R=lC@4jG3MYNODk)ds?H!O&Eim>eezn@kcaNh@4lU@>>ORG*vrdvPJqT7dwb#2
z;uRAor$fFu8M%C;@EDj74bRjHHMtfQ0|!#(au#YOXHmX9Q+|uAm!Tf9neCiIljZ=N
zL}Q9UYYH=ccTIlv`8+*vYv3&%(GVyxpvw^30)1O^4HUX|l9XSp6OrjvUaQ?w0qzVO
zGsag~hX4YLQjvP;_&v)E636zsc@ap;b@Lg`4MdsMt~p#zIW4^*GgVXyUT@{Qy>c?$
zVhU>1k@PXYs$sfxZOjj@Kd)a@^w3@)8&KTFJgYmczQE>QtH?O6Bp5365MVE7&g_?S
z+llNRS^9s95??vcQsg?ZT|_00N~K&PTT*O1=W?Ew0mjtFEuukndQExeC<y6EDM~;1
z#_xaNuEUJER6#Y=i>e$&1rEm$(jae_s*dWvH4|L9X4?CMu1(j4)g%dw)hzo%(KIA=
z%MO7IjM~Rjleo&^p%t(v4dE<DKkbT)`4~}drHGu!SpJG;?-B=8*QA`D-cQ&qCoA{=
zRi3*@35?ua?DC;E1e2$f89=givw>1Oo-=#kI#x${5~>uK18h<TzI;0ZArpgX2_4-@
zRKVN!hm5;Aob;<814P?0WDj|TNfRTxmPj?9auqFii#0fTv@RVT4?-`GOzqs7wSLz(
z1iW$yH+o2#FAVxl1Dju+x)(=QalTCj==Bl4tE9Pwkpm@xB#cn_i_rrel38w553xAM
zZv{Xe7TYo-HgL=+SYBY5d!3>jV9TkIh##<flfvY*cB!kg7T`;KQ+24SG@@GPqIR4v
zReYQysO=aWK~nj+u5}B|z#qb#tenxT5L-*_g!+YNwX3O353&BX2q$TFdG5y)cdTnd
zPJj8MtB}(2O8IKmqmlu86#X4YIvD>Xx0FmBvq%&y?Z)WXm^f{F|F}5Hf+KI%g<Pxx
z_*>DD00(3R<zPq~DH5Zqpo4^$Nz~XavBBs*I0#F=F+6vr$~_B+wPvu}J<s<@Du}IU
z<<j4k6P&vsm|Angp<;X|=R$sq76!w*nXegw<OF-Ek#U&hV^nn22lp4&k^%tS;0l-D
zJgH-mRcIsxU%AyJH4*1lSgw63NBUwEIs4NnjP@Tt{f|nt!Zi_=6^D1~JtbljRCrG=
z@%;XsAZ{^KMQn5Bx3kWZW&q~fOhX}Uz^(5TS5rPz=@H2ANUvLyz3BS`cgP)>2V#m(
zM`v{N^X2>6V{-$+p5tLzATRXL{A_KVs<aQHV(7sr=~>QdmDjG|B}k|2c}j`wd?s>s
zkiWUer55+VoGAa+A3xISP=IXM5MJ{g)MwOQEVn8eu#ax7Q#S^6%o#R$PJ&(NB67st
z3Try{R_cwbGStnK1vV&li;ummmwN!GeIquRhlUdz(%L|(Qu($!qe$4jQ?ezJMw2Lk
z61Mo5;?3bMQ*m4gcN@Ip4^F}mIHyy0?hdWvM2uw8)fL~PIJQ3OZy0Ap0-LsF{KkQu
zHqgpy#f6lV=81P<;yFpx3dx~S`syr!gXMwr+#)pz0fkc7AnsYI!5l8MQN+XFH;Kt=
zHV54b{;l^VzH?-fJQRn~uBnDak)&J~R<fX3D(1ym2Y4P+>W3G<%qgx6Z61dgQQky-
z&ICx5&`W7*mha&m7?j4Eu>+wTqA`u+Ina|&13&PDRJfKGvooqy%|_Ms+@kAL?95@g
zr|LqPTMsAPDvt)9Uj9*Zey?hs;X6@%O)H2@WY%XRz+2v3LtR;bMr1uy84P?VDB&$J
zm|S=Es<>r0603I$0Vux^AcJkYOYuT$Anpxe?p|pRT*3)-_lU|Z`EI*TKEC)(9xiFo
z!PJ`LmfBZm4N@^3>K!k22ZGzvn%c;YI-TZ;@*B0Y<NxU&H~-Q%b`#c{JVQl1A_@FU
z#SaO|<{qEnebIpMEgFgFH-)d8CNPJGN^eod|4v;m$}8osJEL%WZnp+iZnu@iP|&ra
zp4GEAZsf^PnrO$<>@3gc^w9Iu`yd0N_-0>K@3gGJ&ZIr9dIC%U|5KA%i@|Jidm{<2
zb*#f*gikKVoiPW~xV<O6q%9<dkYc{PAA~E19&j*dExf1dX!;Y$StWIF{uD@Dl{1o^
zc2xT7lr%n#-rcucLgsx_%=)3d)Vu4YIAWDm-DW{2f(*HdF8h<y67Z80wA{E=;YM=T
z<>=6q$O<n3m92gU(HT*4DpFg0w7->D)?yAvgyQv)u4dQZ2m`@$a}*Egm2fk{2XwDV
z>&;SqHd^PyN%ta`(Zc&&ngzAxF{ABZJ3lD}i&M}Y4#n_bU~Qp<RL*E2(&6bSGRZUo
zj}@_toMOl<EuO*)4!$XPZ;C>J%z@^D6R!_Ws<ghHp4QKNIO~-!Z;X(y)vnFqbL%DJ
z4>kjmI-7b&Av79r;Z^ro$2h2+BR<sfcW^p143ofU$7%0J^)UGO=Acpzv1m~EEE1B%
zCo#E-mwFNF*fwciY;A$Qdm;xpYtZ;!*EXnae3Hc`qb2Rk9ylE+|1b?yJSU3$rIL2;
z({_U<7UL>gf-SX+67`B}C<^|}=k?!LpX$GV@fS5qk?a+%cWN|g0en__bMX^#;$pkL
ze=WO?j*OZM%b|;>j^^mt7P)$!mAQVPO|ro^+_00-j`0sK-py5z@T+gmFP1{I(>;)p
znSUk~Lwne!HlWah>1F=jtR7JsbwpQ4cdl(Z#CqsCJcP%z<*Y_`_9p5a(})xoV8qq|
zj{43fxcJ3oC1QMq**AU5EzA4GFr@L(jHTuJ$;z-cKW*hrvle``TsF6HBjtgrl12SW
zfF$=;*^r6?>5H>!ZzcLbo3)<>*H@Vg_shr{l;+ntu_}Te2?uBH+te^_@W5OcB~+XI
z$s6`tR8>8_fM84Inj(nw34yfI<YrwAPT%BE^y5KiAt#Q={a8<{O?G`W3Di?~teUfk
zbBz32xE4O?N#b-clN*2ToDc4(Iqzy~=q6(YH>Q`dIee^jsq1T)gowmU51_M;P(zxF
z>=hCC(%qa97HOHhClqNf-l$?g%H`LJ2EY5i|2KXxH_i{9&9G)Pc@w3+PJAveR$9$E
z&H%mX(Pxr4)a(mWWUp$A&OYve6~Um?J80^l>Cr*QG6|h^%ZGa<IR{UL7mAN3$${x0
zGQ!}Ww3%MJl`m8}R1g1!H|?N%m`rLqaZ$<)NzsF|Zqglr-xEK8U@Uec2};>4%~3{r
z9p_K8J)x@*u|W?C#q|4c{ZK^8N3$#X(zP2mKA$gJE(`{7yo1_r^ItqdIzrA0<Wq_6
zh&ASSZz=gm<8>dm5W?!HVm{;(`uj%W>nTcD{93!)JuW?chxEu|TSyvU@hJcj;m>jC
z(7sv~i$dg^41V{cis>vzEzzTXse6X;VO#i}-@5#v-r3tPqCe{%m(!qCjE|m5p4$_%
zJQyUAz!2ifsVP2<_ECY6(^eD!j!+>QVDGsCc><87<icngh~j$b4kZK(vUAsA{c#fu
zSK>mh=p*^!>O9frQj9PMD#hhL`tqOJ-d>*0b0NG*szHH*XSWRFed5;;Kx(Z_5M+Oo
z)T+#Tc%2`CE;=juQWa&;uY#^jaD=P`eKiot51L7}gyNu0TS=mCnw1Dm5tbynPkyQq
ziK%i|{lZl-`CJXra9qF|<}`TT{OX!{LZMrB3%-PJE$@;#Ujtu=b;(lO9M~+y;AfB^
zJ4JIV7z4bjMd#*DPHaW)3|&b1s2kA*EXG#+SRPQ964mLDp<&M-{?0lL^e7<znlgz-
zqH(80lx-M2B&=B5%OziD5{2UwVBAB~J{tdz+@MY!`8_<~TI2Iq{+E|c-GAynQ;k`#
zU6z7ZVw(;a6e97BdrziiQpO(J4pYigU(VX>Ii{G1H6HA=tvV3AU>seN0|0l6bJYp;
zlY$<`etiG(8#xSmcz?5oBz$F}pp;Wi0fN^a<iBX~A-}DT_}Wu#FglNU;{4jIwnM6j
zZC{vvVgvLPq*H+=5=48uA-NZ5Y8&2w#}gBq$gUL*Y<PVW!?(mEQjQ2q&Qui~EPSme
zTc`Ql=L(O650P3I6F*{ivkU|26wT%Ln+Tz6#2JdQp_zN_2Yusc9w1|MAY2Cm3`y_N
zLZdc=dw}0Y-|vu5A}nfdJfZ(WsS*|esnpJpG%+r3Yb`8nMUCCD5t1XUQ=w9CUv=79
zQlPzc2P15(oESwzk~R*vDIkNBB%eeEWIZ9B%1DXd+D+*Lc?(newzwO_ReA4IfKYWm
zK&I>4RF#Pc;vavy77i!PYmoIj6f4Y=+2m$eiNXpll*AllBAVGOKI*DH#mod8#SGBI
zhVmgL_sW-HHKUi*&OQXWpKF{%1oO>CmY66B((5w;MpRhKmwne2h27#LFYZ%Quy)$7
zW|hk4m#UHjl4aRK*r*KFc224=@wMkPEU3$N7UoPeyAqP0MhJiKzoZ}fY!Fq_8B-$z
zbfM*1q`x>H4vSk0IOmS#uy<Q7?K1X*wQMLdn2%kgng!^V+%4*TluwMurn)6KSIJ0C
z(png774UBv%|H0kit{P-PGQs9s#n;;92tig^j;w>ZZsS%{G66AD5#prb~~vJ@<Fr~
z!K}66`@&<zmQ6Hi=dCeODrV^2>;xPt=Lzt~0}v%Ny<dJC7u&ma_|o^x=JTv@kTvKh
zwC}T^y|iX9&Ajue71avO-<RYH&fk}Z<a>)kr>lLM?3?`KP|3V`fMga>mofi9!GUI+
z!lwrNqY57@-W>L?aIfa~G@pVLHJ_tVBo;=Px19^{dFura7V}*$-cnZ`SGxA)30b{l
zPiaGiO)W>_2J!<<Gihd0@(AB=O-Z=PwzrBzYq>Wg#EQ%?)1}}+4zV8wI>&7NJdb3x
zw_W+|%%e8-a7a`)zjrpYql6=ktWzlsTJz=KUVb~5h1!pupH!vrGWK-d$Y}%4z(N(i
z1@B03;gRu%+ILZ&{3kgj*lyXO4_u@pxe9NQ`BRC+usfIT56KgVvz(7ry=rp5_+<4@
z7c}8b8bs(PJhl7}ptp;W0!=7&o7636F|4bZmMkr2dve-<3{9e4q;<rd!@UhN(3Kb+
zRp+evgDSYOh+k8G8?4t}nc)S)o|<sr-uUg@`;}pNhOvDF8PKlw_J%EJwqaj}@mA#z
zI2=6IQ_l4&Q4%FC>P(jI=03p}*weg!WTq+u6PnXGAUx!8&m%zzEq0C7@KMfgxBZ>M
zo7UbC_1442AJ`5#03whVYe#Wu?S9p*hn|*0Po*9og4tJfRZl?G$t8}%-$6i4NP&Fo
zgO(FCyP;Z0B_CB>)9G;<>A>+DCFF-mPPLL;$}*aj-a@putO$@wM`^?2;{KTwYrB~2
zJQs?~nidnj{?%L!#ZaQD+P&MJF>YDC^$TB!`L`5k6iii*s#`o86|j%Uy9jy#j07$@
z2VJ>hYpej>%R|E2ffhA;!tK(H@=&=9?z5?HTjwSgp7e>5C?90wC96tJYu3Ail)nuB
z0ul(}IN>5|)GE<FM?3bM7(U(B-)spJT@TRW7Eo{=39*{|4M)C-PdK~e0Jr8f_woP4
z+KyN~tq{5KjxfLcjQ%G5olo-lCuTfB@4hX&*5OFr)B{SSZ+8M|3)jnoRj<ulKdlEQ
z1_MheR!_$4WT_y@%ps8l?OJo|t7QSWN4}^DoPYBDoLb+}F<f&+5JT7gdeI%eppR8b
z-H8GS>^_$paymFozNo~1Fje*W_>Fw)#`vwC-<s!5;V;a|e5`*|X4fmv=SK@+o$@-}
z%7Ckn-?sn@$7sY!)(TEWEfq&}a+{P?lE2FJo`Y_*reY|`VNd+<@|#+=A*5QAN*?P2
zbkePI5&$CM;Ph7nvdv3XyRS7a^qqZ)u{<(M0wmVb1j&oSGp9=v=HK{j58kEqj#rF4
ziRz16{WHIH{O{p!>a_+_2mnYt7ckTAoV@A;qfLIU%kIe|Ui6c@Tzb_sP8%vS=dOon
z(O@X^=s|hnvoqSO5Ku)$Xq5!rat}|C3i&Jj4QezLq={n175|6?EaMn=D};wtdRlIm
z?kRbMIS62{D*pi{K=qg`O-?p7dvp*3&`8~*G(}GF`q%U)FgrCRD%F=`N<~c&Q#tk~
z1yqvgIsmNDV?sk37&Iq`q=xK~s8PltGC|<j{J!d*<*&lVYO{28o2F`nQm_Z8^SJ9d
zzjW<SswkeEno)-Y{qjTn0Y4Xmp<tL>N@x`aD3s}ls|*x5;_{4F`=$Bp%7CzXono49
zN;Sz}UH&kiPfZ?f2;C(w15w=q70_qfw4>3w`5H`NvSE;s_YO@39ootR_zO$8r_Y#m
zvCf4FT<RD#umm33<Iv%kxvLgX&f@X@UNE1T`4vqziri}lF!SpAp#SOYJNIVv2P^}h
z3U^%DKV%{1`7b@zEn0I}6n_hvl<5IoI$wrGyIl$g_U0xkxxPwNy5;3fQBXAHs7ff~
zZy_68UpJd4^CNe3CQxKDb6UQG+`f}013kP{n$K+u#KoyR6$-$W`5gLT3nlH^^{(c5
zEuyjh#I9VDe)4kCk{kqvHiP;^xvG6Gp6>>gWA&H8?|$S}Ntw_DwkyOB0c{oqk3Why
zr*hYFNff-+*8xN42c%?`MIv;BB$I~{z(AH{&WS}6W6&!pe-Wkp2fH=K>Zsl>*t!|y
zDud~1xn7E6(C5Y)9lUPCe}sF6=T`2@i(hF*2xzQn{U(H8ODIH|jEaxL2TT4kl6IUt
zOwzFJW)XZD!4raM6%}8TxDC*wU=mUR8-C!fnzxI7c>TvIxPYCM$xZM`fBgM7{`sb6
zG=-C(6~+O9p<*9C1YAhfy@+JGR%?}nM}Wm<#H#R~w1GS=6GV(VIh5PWqV5-xkn-xm
z3uhl!BYIGlFsFidN|!>laTVx2N+XaAmnZrlLO?`A>Pp-aU~6)aXpW*P#SC@aS|S;P
zWl5@$k(Z@#LUC3VP)D6F0wuh&HB7Rxm(gLt)YQnh0@u>wUT%~<0QQ{wOD7CLQ*zyu
za0&2r?haKkOy=FaD!Naz9hRMMhMb7hYoX@b-dpAE>KNeOgUST6(I7%+9{O8Pg^;vp
zS-cQHU{asB_Q1^>FXqnO>DMuy9F70o&kHPar9;L{b1ZNnDiDShu6Mv3FMp6Dt#(iK
z_S5m(dg&_Y9Z0h70WRz+2%=z%Rl}8H=&-DvW?H(@Q7)ZWE?h_;61nw}vC7E#^~L*o
zHA+g99DI7?xnvkB!&j_zrJ+lpFma2!)_I8K*<L?#`1?kMIowR4p=#F%Xyw^FnC}?s
zt<?Ere`l_Ou$k#UF~ClSy+{RPSC*aP2aMrEuGz0OzX$k=)ruyG06<$5gv;-XZNu)&
zvoQ8JfZUOYJy;SB&ule~--xDjV7PLY(5cM_Uc3Su)q&u?1eusebGKU;3L3*3+;V*0
zh<x<Qn$1^cSx|kM9Zx8f<Y4qfv_(<A_YS^-46pFSi6dk(I(5V76=scb7RP3p{`f{q
zUEUwr`&oylOyWHEV;9;GHNkPNT!RP>Dx19gL4-3Sw{ne;POcUPr!hsw5~BHiSHCKL
z@n@~xkGe4A_sFX^U;N<m^-uAAmh=Dr|DwW0^>5O*h)5MT!Lf2sy~>UcX#&>5GHbO9
ziXm!w<DWG@g&?*q`QUWyL}bA%BZ9XQC6|#<m7lN@o52+4tk$s9cbdBf{SK1ImCMW3
z;{J=2SsE0UY@`CIS_aVccmY1l;<m)?9WYC6?c$AzUZSbfEm-7Y5;r2Y&JLx$TP`lB
z8s1+_0HnD&V!?psYlf6UnN+npeuAx7l)ae6ZqQlah|Te>@Mm5eZV}OtW+PIb$5%le
z(GKVBqoB`2fukVjcUzJxH2}>hT?t%gNaYI0MW&+>0J*1TyJDUVk4tE;7@1V0x^`=L
zueAX(O>F$;?8ruRmF>J)v(f{aUy>SF<oP81;|OoLE6Zpe+T-nEsl&iE#&4TsXm6VI
zN=RD|QT16vB*L!tn#Eblp&A<%Ar-@`1-=%UaC}nnooEdfx8@g8(=Oi0Q@6b-o;z`3
z^ktOl#Wcm$u5I@R9vDhosl0?`RN3N9gRKgfktjM@eD_G`rCdDgjf6QBa5=KW+<v>H
zsk=^o%xC5L@VQWUk8{sY=Tr<`(-}eTxj{f1MBK_nu>!f?wZj@h!T`_;s032%J9A?5
zn|DVx1x~gbVEB-Zcv=x5TqzK_xQp+%N>2!9a~GFbu;nJ4bv?)}ur*~lR@Ee<kXgQx
zjz*fktdty_VqxCr-Qj8O1C~22GV;%W`Tb`c-TI>H5bEbnf4BU7Qv-R8$eJiU?0~)r
zxnGCf@!Ou|NIge#4k%Rmvs`hafh1bjxjM8|;f%-s)oW5nY~t$}Y@02K+INeITP3pG
zx_P5~C6q<btDt*M&QhEUVNJqVz<eR4+H^vC5?RA;>#X*N27tqG_4)D1leCW&#q!Rf
zvOcGMqFy`jXL=s$ISGKK=2|BVF9AW-qLSrquS{n3Jya><oCjJrieag3cmTr%t#Oq&
z6){iL(A-iff!5A0|4R#8Z{ZS(j4c<#mm;!sD^F9nzVO5>lC7md1T9rWQ#&)<qUmQ&
z#C3g+?TK1?sMm82Q(fC%U%Ht(shYj{wtko&2pt1NfnwK5Fl?P~$TrrZD227v9>9k3
zRigv7p!0@y$d|~aO#zwpjp-~b<YsYTTv3iO=d(Uh6Nj1m6-Cu-#m*8ju+@4C)ivj{
z%}(d|G&^Z<fvNscx&JV%)b~Mez?XoO$#)$e3?{4oId=+j6^%@{n2{hKrNf<LVg#*G
z*Ch9zhSkJ)bR5g}N=>VvPtI#Pli}FS*A>^lbXu4K_ZBC@=xexH#_*-!{b2hvHDc2=
z5=)jslC#6h+G;R%_2%w|qCWYm0Rl2-#IqP7n$eVf<Qz`jRiIR{f<1}_`Q69cv}3F>
zh)L=lDy5})AxTfF#V3w;Cn@Of6nOZq{=U3D@!8Ad+Q=ei6k+zY1EmtcIs!JOFU@pI
z5cLKs>z2f#uC8b|TJX%I@<^Vk=a|RTY7-Q}{l~9vYA(?6wJ#(=ojEBPg3@G>h1HL@
zfzCpH?Q+Z9QvRR{za19}2QeWkc<I`wYYKgKQTrR9=xp4?OBKut{3Pyn%;Vh3y-^Fb
zhYy3<<ewhkkJ7{lIXY8RE+_9gW{!Sp_v$7D5O5Hk8urNz)gOoj!iqJqv}6ikztXoD
zrHDeY;oGg-6-0`&HYsGMzcCD`QrtZoQJ<>50>a@ob%YRjX~??Ho&bq~h;eNY0o5Fm
zLV3t2odk_mPpZZ@u1urXR5GAUh)Hs({J{YyC8+8Pj6cSaOj92c+(yh`YZt3mufuJh
zI|C+6&vez&GLL_~Y>}z{Ua^w9&5Boer|5i>Yni07f#cp@JPqFoNg}!LQ1#OKj_;z1
zGuiUuykW2dn+J4(ugJ8EC+1yMfplec>5Fy{lr3>5Dgdm`uB;`^i+kcv@xpd%$M6N0
z@3=?OUF+^L1KZR>9sUrR!0<~O>M2>#A{ZhKR|%vdXn3yDJy8|a$=%!RMhHMP0f2LA
z22US7O8VqbDh9vGmCy4<{1NKPnX5u9d!+f0l-)|tU%csXQEk=Ln&d<R-NC_!Mx%{k
zYK4JOXKH8wIXK`p#r^VM=cA+X$2}WCaJYksJQrRj3PP=~hmCmDNNoPV2KHI$)Ns{r
zYR}Y6F4tpw!h+~`KMwg$dAy5X<Xg1|%t#vpH_DAc%XEMcR%m|za7C+;&$(_r;hFr7
zgFi}+O|l65a}-4CwB&>?sSSEZFM2-!+Y5KA9d;TVf_r4dE*&pS-T<PQo6#Ycf6cPK
zohN$j+BH4Ts-ip@hlQU*GT_!~u9$c;H&ojILdR!^=Qgsk=7S^QJ*<Q?4i$b;3-5=y
z?$iCKZ6KkO@+=Kia<+tw3n^+NLx~{v_qu8PtS;|hX*KtG8oe7Ba4vPt{%zKLx2Y5N
z4@z@U$y@kqzMYUCc~rUh@mtPU2Cg#!WI$qaR_4gd4dRBVwI=zw&l$ioexo#E5psw>
z5b593gm;GpuZkX(q3@~9p8h)im%19nk7(VNb9r>#HD6i8Out+u@HxDyX6g%2VEE7e
z-yfI9I^d>BDC{IfJ(|X`Kv`Ff%%LMoQy5VIq$OR^BUo#M5k?QJN3ZH}hNR%3jt_5>
z@cH_t+6jvd=HA0-FfeNm$<P<IbZ%>Hlv~TAM+@=Ip%<d-&%xw#qSL5TxKD~k07%*O
zNVBHN*(&;{*}*Q_iB1n)yx`@}c@sadamN92-3;h6pZSfb6c234`2DVPtMTiE&l2CN
zhh<lMBZV5&3b8*3YMfMZ4gEV6nAKqS?k_Cl#543?x~zhEn#$n3{NnqAQem-@bQXG-
zLMkObsbGK50K1DVhmhLyw4v7ab2X6|+Cu3yjGz~Ed*`Xc5TbA%a3kQAe(7`zlGDXb
z&d{4YwXJd#aAt9Zd4mP;lh6<rqlRn0lv!C!3I;UwaFa$1d8EX&JSsRG?abx(FJIG7
zs%4iXy<b5MUZ?6hEFh_Svo#)6=~fzG&|_V@1Gyi$@W29klC_&d64Y|Tp%PPXz909u
zt7`HcviX`}F#Z4oNhJ96l_YbdR{N`b4dtg<Re4;rO}!YGcGOCg`eEW=Yg^tT5*kS|
zs)N_=u_azDPT!<cN+qb!hzW{LT9*9dx7Jg1!b1qL6AaDS@6&|Br@2uQ`yQsShU`+f
zRzJxmY_0^oF8Ajv#UVB$F^T27AYZ(9KRSr=p=eGo0FbHCB!ad18-{r4VLsXKW;!I*
ziwj1UuUYn7mOE>6IbmTMHISMfYGcimv2yfh+2h=i+6ZM@H<Oi<Q8*~i;vvDHB7(ub
z6>F6e$#<xPOTNGBLx8K-H@m~}ugZ;p?<o;%MH={$&}N<C#@gW_QBC0Kvq^B<PYlkX
zad6-9m@wX2(mY)Lae*QqR_$Fv$Kl~Ghb<zUI(FKGA(7c!E!y-6@>yGb<Efb@6IdRW
zV#vs#n^3Ftz_l;_k>*1VSYjy7&z-{5-M1uG$iBgiAL2IDcg7!<dWNfzq%OanpqKte
zaT4tu?@*vHm)Z$wC3r`LGtduwoJ63^<!R^CPJzBg8fyODT8c!a$pkxwWaL@O6FK#X
z%#kgYXL%dG=qa*94RPGBzYvIsI5b-!gXaXhG-}CxvHUKFUEdTr0ZabDL22rsk)Hm!
zn{DW_mdUKpycXyn<-2+8atkT^mMfo%kQy(=t23-^!*>I4UE4N7faaq+E&f2k1F*a0
zG3o5BH&vP1FRnGgc$9?VH8^g5q0LZ)&;ue!q7-6NU~n}PPs|M>#i6yZRKXg(+wfCM
zREFIiZdEjI8a~*dkev-ohh*{rEm0w&{2;6L!)ml9EU5RiWkCGJ^kxbHxD_@WrI|ui
zdn>dfcA)Wc0a}}4&w=4wD(52--qwp&tKe9QNy58H_0+Kvj5qHss)p%tr}0+I0i}Yw
ziW&=j<(ek9U{JyM&GAn(<C%Vy%mc#EK;x*6S{4z>k~0haR^bDs>*ypkei1zv3z^cc
zy=zb;6^DzkF)AjLWPJawrgHCI5&}+ay^lC|=xVK(=kLGqFS$CB;o{P#!pwU1qp`tY
z)7NU949wgYomQ5S<FWBuz8^ck`%%7aar{>9I1!mh&`(-fLoG0mfLRuG(-Hwb71_YQ
zcD)!BzzX+Ikk}-aH*3UBB+~_zdiBzIT`-dXUh-1X*1}En2PkA3T4%@_bF&d}QY~vC
zNK-nqdY2c1n?Rj)t`sD!AEWLn*cZuP0-S<$E!MqC^hEPBOvx4XxTI05Y{}qgEn<c2
zN-yT`g74uQDocSHm&E&XMrv9Vg}^ezhFr=W|NUS6y@{=9!W3U{Bm?h<hDJrTUqUL1
z((LkU3!{qiQJ@ZHnRgFeoGFMY=9}*yib>TOf^RCS1t?r&Ef4EMtPQV6<+<~a;TN(;
zDmh2+GMpKfMdLSMOqzT*V#rp;s~j-?Wc_KlU9yv7E-K*iQK?z0<MvyTbYC-HVt_WG
zqjbgTbQPk#g)Zq=^%XRTGQe<Ip48O8@Tu0G4A8^R^SCjPmE9@LPTw6+Ry{Xxl8=gu
zi0UqRm(aP2)pk{IL~Gz%je0OEL_Sd^`Ei-@Y?2xS=MBE`T%N{l`NA!t^7eIlSVdt9
z-*58F`Q{&LrBil}vQiXUP51JzFJI50tPd#2@IlikRH!prKdnsXn_I?1cLQv3X(l+O
zw;aO%{3}Tnm<EW@9O+J1_^-eXU3<B5F8hzS2i56{MPoAB$v_kqhfFB{NStWa%#3L)
zf@Pygg9Wn;=flVrjp|P=an1guQcqVPCx3}#4H4-E-XB%yOo9bDyKgcY^iQ=!_b1Si
z9PgIu&^hYbs|i9eLg?#H;wO`PEl$t-z{Pt<o0vP*vE$_WpCmQxhp{}|s3HJ;<)l@3
z8f-~u8M82}qLjtKi3Qw4;v#vxxJ7jRbd-x#$wD<ehiAFfDkF!#st9E|z~5-Dn$<MI
zx|{&?(Ow_FRXj|^D=?S5E8Lu)OSRQESaB<}xe#_Y-KePjk&C=t`2|?Dt=Ja#2LNPg
zLdvr!eojCU{6Pb?v*>bV64uQX{3pN{|4hu`>G;hgE_|S%GV&k$Lxhi&&)xA`d8#YI
z-sRu7`rBk6<n>ZX9Azg#VY#D9qQnD||5EQ6ER_7}E-u1R>1I^n#OBEtkg$Fn1pg?n
z4`F$ixeM3E9JSa=Gr)(O<3W?C4Eg!P;a<Uj55^zmxL?7D@R?u1<|tRBW{dW%x#tjf
zxK$cj67ER#b#yAjvnRdnT<%QNH$&2v+`HRs$#vP-z`~L~Z{M2{E#Dg&mAc#Cwgq|p
zvm&*Zp3jG9KbpT0j@NH44Q9V<=k{!_l8s27OQmRJREqn6V?h@=!c<EKF|s6bXuD^X
z?6Kab=Fyd8Z<VXTkC^-Pry~%>y4FJ+IGQ)1VDqzo%4$N0QoRT2o%9;SBT!&2X8^pP
z6sM3P{Ktw4-N14gYlLuG))%obGwR`1MRAUiQYs~vgoC?KVoUFo(s%I-9gz}z$$Q8-
zZ%}El-Vfqt5~$q@DCO$xnwCIw2s13G)IaCe)|PY22(xIdR9VoE`)jv!-c)`XzpYu=
z97SP@DlsY%Ly$r5hH*Ceg!NX6n&U0|<?1KwSs0*)+)$)SzI3Y<?O8>Y)`t(B0F~R+
zqwGbu{0S;bEPF;zQS_d}*NTXtIY>8;KRWk<s@DOwdHMRMlbGjl<m6^fbMGbw37*Yp
z?g_s?^gtujB<;ZJ*BzVt?9H4i{SJ~Zt<|y=;>_7VG5Qko(xlE&m-5ZG#vd1dm}J~r
z>ccLz6}kgA+|~@{k@Zl|BR!Wnol!L!vd;a$L@jsmLVkorHA#V;8|>7Rt`()p^_^0d
z)CiM-7q@CohVknmhzg5BzeQ3<?v*x`40e4^v>CjScHkBqu`~yO8N@cdZ|J2}reOtf
ze5JCZ%yDlEF-*I(9)01@^99RczUQQ68i$=t`A&4qS|Yv0dbQ^rZ0p=wu+C2QeEOOE
z09DAcdrkGD+Wk|Apyh#~{2J1l-yjC-Bq!{XzfJ0yKfuD!A#ylXE@DqjWb-Q*|D)B2
zJH%+#jw}k8s$cAD(>SmIVQ9*;vFc2(Mu0j4-)5a=WMmGHtF%HS-UkPS0Hz~QKZT_r
zgi|!lk}02?LwyYt=$!jKb*RtR09c^>W$BkwTrbV=lW&pmt6!8eQm%*daGfH8iUs2A
zX;m1yUpMYxc_jyCdqk4Czjhc956>AKqU4u^JKo2>@TG=peHeAuNN<`<SNVBX%R1rU
z(hkoU7hdzA+IFj-wKO`(li@b^Q!*OLWd@`PP{Q{_u*;kqeiB!3&1YK>%RXf~@s?f<
zpEoZ-*(pOIJrZ`maSG@!rA$!ZGuPUuo{8zXyHzkKKw%XTgA0{+2Z7RhgORZ)`Ds1q
z8{B8-Oyqopd4pI>p@FBv$XbZEhf-BE_=AfN^4BYFOp6KKRVI8Z)JlGPFi05*4gvOw
z$jC>g`gSa1=$HKd^8A-`)LY2=eC)Kj*;hvR16r1h@Q=DCeI!JFOD(~YEV;?(gZU3Z
zE-!wOrvUS14%!-VzD`B$xbMH!a1wR;DJPyd>gd?A5qmc0lVyC7J4Qi4#y*3+u5*f3
zg_EcoKy($$DJQ}9p`7K>frw}tJqo1?w6uL*0h*?F>0U*G$V$q(N~YBPVQ1%;@<j#>
zg@`}u?r6I$F4fKlP*u)*^+^FaPX^WoC{&YwS{aqtgg2}6QjV)4-6&+*j`7En)|Fo`
zzx*MfiF%f(n991<VsR=(`%*YurQ+jgp*E+sBzFP4WAJWKUA}V$UJ2Ab|8*47BPj@J
zO{H7>z?zXHQ;svIN}M{Z@!MsUZDnh+BG?Zv-qV-JcKDfpes9ve>K`wDk>iuv<vUpy
z45&t9_T64vHkww$3lN8I!N!Y!V#9wes(N5g7?-Ks#7Krnf7D-l>q&~ON}=__RB?=a
z%x!#V(WO-qZ2VgtX8wVTOO2&m9KHg_kpW~@$3&Hu=$9+4_Cg+_{Jt;+k&Oc#?WufM
zZY5~b;}Wj?s#2>Wk`)<H_Ipm_>Rm?|XuI)RCg<SlyYN=dXqKn7(m$Kca|AL0YpQ`6
zN}N|iymakVJ@|5gN~Woo*9wyZu`Kz)Oo~QVI3{MMZ_~>)7ZhQB9W4Y~^hF}w(LwnW
zS;;`1wJ3b1S)!z#X5LkI-OeCOB(yOb8fa+I`(#cz`qmy+5NAm!6nj2)(X(9d8Csy`
zbI|{Kesb|IQ4*WNKbq7r=38Di`g`Rw#G#F-xxJcyS{znHjaLfB!&S*=@m`gBc?K7)
z<Jua*fAZ+aPi;}~hl4#O_7z_jfE~W3DPDW}TxbdeSu#OcHOjd3%5UI>`5lMqGv!gd
z=9bqtbY{bRCnr4Bs**`LyZlCeOs;93J23-s+Wc2kVX+$&xtP<YGHtIaurK=-cOs=7
zoh82)PN=%=IMf<$pDV{}wZlZXtb1*co9ank|59$<DsG<qbpf(V=6Sy;i$vpMjGf3i
zwvF@4BrJf%6MdBu22s0h&@V@Pr)7a7$UFny-`K;nO3G^IcB78l`S=6B*A;N$sTcBp
zdaAo_Lk}Kjs=?eSd3LDM6E)>!qV>6=hZuKd{`2^;Lfg%YFThVK@6uz8QPpe4|Kdx{
zu}&}FuZvH4g0MT*R&s1Y_nY}E=|i<~oob@C-b5Qg{k>t8WK<@RyKZHAEV{E#{O;^m
z+KE%1MWn(yCyMpOk1t=(>jgJ)_3xr=<n=*uQsoi~t5>V;4O?M+kpcOn-w{K3_G!)>
z(l0#8>XY*G%9};1!-StkN0}twqhYzRgffUE+ZVqoy5Z(UNzREXm){|SrHIw>^tzy2
zyy!i*T1uhn-2(IcZ9YDY(}VTo*2ok$Pq<SKI&^$1AYXd<dOm|xJ_%T##QSer`W~#N
zId`B%kO<lW1s;gL=QL0I{w5nN6QFFAnUqeN5*bA@Um>FW%5LZ&N8D0vFiso?w&|9g
zoC5zaxc9D6bF)q~H>Xyc6&F4zBV^X9cYbM+HpLjA#cia;EB=}9+zh@a8;%>BRKfaS
z{0BYrh^L=(NnR+q^s?O+5xn>(Zm{bvQw!5ZGeFfa1R(jjZWS9Np~Ren-Y~hf4^<E&
zFFwMV%cL!&*N<K+0stbjU3<yF>FWSe>XPnbjmExRyutsuB${|9j^^Y#`{mnRT-7Je
zBhB~0g*m{aS}@r*T%v!qUV{s+WF1;RDOEXT@*e{z)NkDsHt?J0Ux8k2H^bq;Z4ah#
zy=(zY&8_YoRfy8SFRS84Av@~IG+2?|KYnws7LHbw!&6<ot!URc1d<iJsi4a1p|#G`
z!mQq=`xbFJfGjO&0t~aV7D1iEU<({?%mIpF?FlJEB@IcAE#%Q%p2*n%oeUX3Gv#zI
zo$Nc?XjxiKZtWCVqT2E$WUK{Ve%_go$YE4K=ja3;ueCztb<jA3d%M1;&)mqbZ|ybA
zU|1MBUwPgFj)^hvZRC0tBVAq2lZP>Lf3(RoJg7UV)_thL);uo${o?O53z@Ns7?wS$
zQj<K=yv0PP$<EYPZ-K)KHCZ(g#PiQX071UBh^anu)qAHNV)7OwNz4q&y$yi4)vAFl
z*t;nuxixN^10Sq!F%o*A=IBGFET7&vI<AHWVUC4&j$~UM2XvJ?3syaN%lPAf9MOJT
zyBZIfoY*!a#5j^D(>X;e!2J51Zf{tj7W_FXj#5`xt`w8M+Ob?#*B3QCL4cU5-#aV&
z%=Kw&Z{!$h7Q#G^_1v2Nlh6)@u1rq0iknzPtMZA3#)gP$O)j(o19~AcwW2fBQ#q~o
zMTjCrg3E|DvL}b6@TnX9bL$1Fl$uK<tUG?&v~vgmHg^O*6gA}62c9lE>|UX>nbM$G
z93e2!a}t-IxuO5Q`fL(2235@x8?OaPX2az<pf0A?BGV_|cuV-IoIX^m;t07tz|>qK
zTjL)Y{qb<SFm;F_OPRO;+Cb;*zn}DNvUI`oe9Pt@{77%chg3*7;xChcJVu4CUH8eP
zBHMAl<;^d4cZS6_(GnYcJ{87Yx2NqHBTCJC1W)eql#@a<2VD~7AL@5u(OG6$ytt4i
zmXods_(IJ^lGsfD7J)gFH8iZA)%qj9c4l&M+@HP!wts9}NR31qdxDM2X{|d2>M(Z}
zy>xg)MN19DhV-1R^XwP(Pr|URhWd$m^eK%a33T?jro8l+Yg@Dh#am>aWW@pgjmu1f
zEdZA-In8bh^<P0e(7bkjP(m^(N8x3i5JYRLggcj}JbQ0O!>P`pc`>K2Vkvr>?sncR
z_G4kTEiaTTeTUCa?BL{=PyR{YfYsXz0Ft^86Y~L#0*83&K`+05@vA(+6%x_|pUD8%
zUrIt1m;({77mWW6;A1`vxQI+E{liVb1}7~>fDu|6r06H2?m1o1GVaGGI)7pV<mq4J
zp3FY(8$fmanXmZ-qjB-N*>6ro={f{2bVOoR7a!*5Fmx`gX`Vc(&^IWbuVKEtU*W2?
zZn3{jIcOVPbTc&1&Br;#IgU;Vwn=caRRescC4I_axlyi74MpN5q0gk8o#;Wx7T&9C
zM98k|wNY|6hed4dS3N0+A5U(pO+gQ!07<FTY~>_6Z`L9T4_O`org-)WSefmt&E`HH
z_fL!MO`#gLN+SsJM#cYp`MS>2I=JX$?JLW(NkNUf5mL2Qz(THlP3QIOD<;D3c6B<E
zerzK{U!BIb@%(_VY2zsrol7OvsYPOuJitRW4oyD3tLM5$rp(L#D&d-2cxPB#GTMRH
zygQ^m=b$c~rCCki)@YZ1bMdQOzryIr8rP8o+<1IEdeiF2ke7{h<F{GXGC4+mIhs7d
z+}za#g0&G${?q$b$W;B~@_TwOS_N5cm*3NR>H4`{+-a{WTI>UPh>G#|=Vx*;OhRwp
z82X00Sb|gU#HxyS!lD4RdpjE58qJ!`Dw`Y^VgZU@`4g>}762RRv~dl6lES1`=SL(j
z^^dahm!1asP!e3}(*q+vfQb|+Nmd2C>^Iuc=l)3RQ{l?Oa8nT^$EC(h6(y+sQslYi
z4s!K<{m8m1pcMVjU=*4JI#{WfD2dkDC{<+1da|3RhxJ_{K>8hZ+0FceDv0IT_l6Uv
z^~q4FavEI?Ly{k$_Mp8)K+H#$W9*|rC28PJ#%-_O-Z?1EPMY$&29#AhF|(Y16gI!2
z()Ymn7?Wu>L_F@|mwG*1uY!rF<o24c7rO_ATrzEa&^!?v(TYZ(&?J?ah5?=$9hIuI
z5yg9{S5TJ4K{V8wH6UjSBube}UGoirL>meU*B9Q&v<;~^!e$|kb3bm4#y?4_C$LDS
zl@@|Gq->#`{*Gy~S0V0^fnNTz&GV1+=j+eDP=`m^hupE-h!;y@a#HeCXzwf7mEJMf
z+H;z~X4i(ce7bTc=})vn@;(Z8M11!D;v=r=Rj;qq3pvvArGy$Q>L^G?eoDKLx|EWj
zqi8gT5vch|LRHF@TtT+U@E!k9U&hP;s#T7YiHor~_oBQ4+BL_(7jUv@+VXYgDbBWn
zC8$B))`yPD`qGaMP%)CzE-Z=0`u4lfU=6>tx2Au5uAed$axEACc=2m}2j<eO0h$dY
zK%Q)Ljk9m>AduC5@c{b;&})Pi(bqxRG)ftNJI@J;I;eIuL>{gbz?(iQ$Ko_ACXk=<
z(Wyfdvjj|eTxA+cgh%58adMXaJuL$;K#41&t=tyP1iK*N*$RH<woxg$*8AV1RvG6G
z!Zxb-lGo`?UsP$@<Kbqp-sMHT<q6G5^ko>4d=EL!Y@W^ZTxUgF^y0I6766Oo8gA|8
z#-%K?S{B#?_w36zG$YJY)YBYEX_USR$5z)U>EkO$=gFo<91eW|sG~ibTx<cpn#<R;
z&o}dbdGOCZ(pdlT?#;Y=vVV_)LlTcm1_jk(v|h3Q7&+<@;fl?pxcEoki$Z>Pcgv)6
z-JD_<M8)$(O3nxamz;A4nzgNtFzmJ}eiwglQBY>?!LS$+^#OA`D2d_pB4iNiPf`*E
z!n(Gq!`+szoaRU}Hugavp3yvWi3gy2wk)-BW0vji;n>f*DQ_-*>RluNTeh+f;GFPt
zk_|mya{`0{_RXkdOu%tdf!^<17uQ2dB^5kR%Rs4@<j7H$!tNwhvKR;~KhhCgcteCk
zP_}pgOry0c*$hno^okbI#R#ZW76{CG2X>ynj)jqLn+>CiD~=CblOYjt4{8)yoCnX{
z49q{_FoA$_Gab=!JQdVd-Ilw!Cp11zh@c_XuJ?evN`RCypK^o&jl?5efi&=<ih=Te
z^%aoTRGiabo5Yv$un`|+3FA}lo|N3j8HvJ>V;HbeHyIRDvV14qMz%3o(i)g*RN+if
zX<4p_XDV!L3HN#Q>BriL&h%aLigq_9DY|$K@%7~V@F<mU&K9zH9c6P`bfN#4<Qt86
z<LM(&!3-NJf3MH$gh*OSM|4wR>6`JpPrSJfTuo8RkeG{B>yYUpfrARlT>gVz@w*=t
z?DAUOZWSMBrM9K29XZD7{U))n)g@5)T|zKtBS*`L4AT6!u0J0%`|QDDuD=7{%KI~I
zp?d$gPL@1IM1Ee6vwnZK&Q4XhkIX_DvN<8(HwtgnI`4HYh~Ql}w6sCA?Pb|CFl(DD
zG-V{vT7oBIauf8_Gj#hVv7tl6Rq3?-gB=s|%F_K(w9uSdBDR=vKq>M2m1#{zgQ9fB
z{vW-Z1Sr+~^4p@YWfTbQ2VUl&;o-p(l`PEdx;+>g<;yx+E$5pr=nJzeD@!XHP6`93
z6b`ZzoK6^8!7gdI5V+SK)j(=QC5P%X84(DAew%aoD{UH4NZ(CdQTpd%%NUpjUjZDc
z3}wOHB9f`cjz^_AqNbKwG1Z(?17L|Qjw064Wa$3yew?Hb6;F`D-^?SCM^uo<q=wEG
zmRs%}8BXw!WXfQ9NRE@-(M*V;Y5ppOS4-?ymI~GMY_iZUAqPFcJy6e`1&YJ$9rsx&
z52$6zO44Bb;BF_U;H?7D<Uk-156gJ&s)c8MUFyyL`d%5%tmtPl9P^4q>7q$Ph;`g5
zg|DTJND{zeH`;n~HSAv1`UE!1SA#I!DlvLX&2@`fQGzU8?3KS-zXd>cB}5}sTJa{t
zwDd@w2iX(N&H#%S$adw2H1lqDW`0c*0lwfo^TjpI7Ay=Fb}1p~J!PyMj)0J9!ZYj}
zBGk&sXx%h+(6{4*f(m)OFbL`;=``=J%xXH2OgLg^oCcZu0yWxydweo}v-wgHrKST8
zR!4>}h_C9%C;-IlxU7~rqtk|s>EbuQsh5M2m2;AHOzta7f|O8Z8xnKV>f#Jwe~2wn
z$WizrkAizRw5`rBax|FDacIqxFKA(S;jM7~RV%5GP$BIo9%XYyFz7~fUFoEG`Msq1
z@IkZJ#qQxZgQ{cRz-G{kYldl5C0f?Lu`2D27~4^?DV#L8grlDcd_non<(dEd=ab~A
zmV-B|w6uF6N!EsmkcAr<ex-BkOwZ<7!lLPYV3MT-rw)_CQxP)}(jqTH=#G=?M6=>3
znZ{H)t(?b7gH4dI;`j^Prj=#J7$EwjlVDwRY;NdQVAT`WAp&hT<N&C79^E~8M>Nge
z_V_LR#5G%b5v7$YDC%}7baKt>JxvvF-Aa7{zhpV;dlYplF4)<xi;82e^)+=@i3xx~
z>C(xrW=1wYDY9atq{U9JJ@cw|mdaSUvj}O7mbx-(73GA0aQVo{&*NTy^2uxZm7;0V
z&+{xn?zB3^RFKpz*rFzur`Jr~zpb-ExV2uD0B68wmy&gb8{?gvFcEYTU?PQ3p^FEy
zFY=-|%JH|fQJh`bzjwJOIiV>*bdHL{>wtqveDWJp%^Ip_X$kJBSO3^O8Nb)O%>A$D
zesFW?UZ+(Fo;R2{vO9!E@{KD2Ld$Q=xVfP=DoRZl0(|ejHe+$dU|wM8K4><wqIm>C
z<dd<p{OJvcuBok4{vd!i3d!nVSNa&WB_roRT;^%746KW=H;aS1pX!yZNd;rVD9{{E
zj;z~I=_RxvRRhyxaTQY3q3U*kM?=roCJXzOD|11gaACfG@j<?LZd@r-+&!V{{g&>r
zJ&t6xn(a#UF1sdMq{|}7E3-GkIg<>Dw$}20KrMQ?G6prd@kfgL)?w3M#FrMUXH*-g
zI^mFd<sWSnH7rxjo{&SXV2b!@RQ<X$+%E47oE{Ctg5;Ce+?Pk@Lovj|V=_u|tq-?D
z;rj+<Xx`Vn*lrT1mvZ#w{tzybHyVVMD)W`9M7^J~MY$eeuQg=kUgSV32b$lF{yu|R
z?+TneADYHps#ARW^tt@xHpMgZ{ey!YR0$2th2|Cc1a5<>b-YY}Cmjg$o>|IYu05Z;
zGZTw5M20iS=dbicy%Bd$?!wH-8DT(kf;Ryr#6Kf9!$v{Si2DTKZTzvlUV2uC4IMb^
zRfjz-EkU*I2nZDWCgBK$<gh8?OMy{gy1Lc*{^=;$$A{zzB?(ElDEnjl-CVeQ0s0QU
z&Q$6iF4<Ov`!vrZpkYy(95ebkfA7|kckm+xs!hBoE0$nLn}(%R2=+W*1Ip2%CKMA7
zK|$YgSl{zl%#@TUjfLWPP_zg<nO?%$+UwUo<5#|mx7}V7s-?V!QB)=-Y7U1{mj$iI
zES}@q8x<TpYCHv*;O0K$^6Gt5%nvi0DXfk^u4qCRlGlqnhvlj^!ThpVy7L(VmvBbp
zFhACn$^4H&zr5CjdQ0!U+YlY=ytZWCHia&d(oU_4y&Tn%!%@@!x>k-#1w7ctfcoK|
zBs}^j@F@MP79gx+{6<k3NR_4{PGgp6Nyq(aH-PDdk7i<25h{b1YZo0sL*(W8ts=*D
zg9Mj%`%g`Y7p-;8aXAW1RgHaewD*6+pDz9m55Sf3&mJ{Caps^g5V{ddRT_0~u$h<p
zf;K2JM{-xG#wyIawg@auD;IeCXgK~!PzCX#@9ED9nwy^afEaVm5-li4f}V&o)Ga;j
zR<LKk3kT3-UUGsjj>V>{g?rOPcpT7F7e8vrDbaLj*gDnxt4mb8bNLU_7^K}sH!T*#
zW#Ka?9iz!xWl+F+g4LZ*;Wyfk1l=bMg}V-WD;k;_w&LOCY7^;35)5IyGD#?8VTFrc
zk5a@+hTAOqn~K<Q3T&wUUdqZo;Cps4uae*mB}zLRE&77+ksX*!6;ld@E;sx3z<h4@
zcKf$)zUq`GF`>v6A8Mss?FY#a)AO!ZM5tR-eI*Wt84MOBho#6}4#!<DeXe(GXFVdS
z1PQ+V3?^-{3%Mxe9v}ao41V!<1S;I@Y;rs~{`p`hiLpuu6$z;nMkvR5c@jG+XkG4Q
zj<J@dh%szhS^EbV%PsqaFSLq~Ym@XgRZ_y&A%YfkwhT_xQO-%50(!F@7lpi24I$m4
zl1q@RCGl|ULi4$nh~!;<Cl3y9D{Z%PVdM>q4DwGf0+Pb=AMiq8At?i+y#K^H#b$3#
z{5J$b6dxEKxrP;`vQlYrxeT+8JT}$Q&EfGZ(m}zz41FM{qv`;eL;LKteDLcd*~fW`
zlu3<<1$KCY4$=8Wn;!~hjzmJAl$9K#M<Q7Bfvw}Ao*>f6BP4ooSgH&KuJb$ODiAo(
zx<(-sBdte&B^=*x?X3wn2JZUN(`WLla?G`qj#Zo@J=bqWb-&_;=KktJ<<79%bw5Jt
z;cywhXIVo*kdYTi#TIUnIM($~<=MjOU4KlgM$t=tqe<`e&-LW8#l#`<DR+?ljp|;!
z|CXyP^9#mcPJI<BYgXWugk@Z&`^Dr{^o6{`jKICC*WCDQ%Cde9`THNdmN#OnYYKIK
zSh#$tQC1FGls;IbFFk~=pLc=4<ROAk+hsQT6gFC*6zS7Z0RBGJ9_!i-cfBz;a~{tW
zh-3|0AeX+Nw0;eyT#00c45t1$F2bdwP<rabhy!D<YJ2$(d<gnBNl>LM-mfHF^0xK*
zFJ8-6ZK~W?-hzx<4$h02FZ9oJLg)RcPfE45yk+nxJ9O)w%O#usT>}>NiZiG9_@u9T
z?+(?lIA4wzrrBT-Myz{)n;dbZ$4*I8R%kEh7-y+AVo6wVpgu7GFO*UP(IU-N7ArJO
zf1s-hvc!pYXYjcB>`3X_w2r*l<k;z;l$0R>M6XkY<KcZ-(Hdf$>mM9>qv2uJZP5xa
z)pKz@q8gQ;BjW}u{XIohwWByF^kaVQE6JWkD!9p6Ozps3ihgcet1#wACrhojOmgJq
z#6I)1NrKUO;3J-dZ`aTAwl3n5(3|cIi}@~cZ}JL7#cO>RAW;Ek?w1A0d=R#xQZhwC
zxl`G+{I2kt;XvyZT;aR&XX<$t(-KAvanpyOD;nbbRs(O4E_MR=p$uLG%45yvh=vsU
z3DBVwWLOZ@F~g-tJ^ws??xesUALc&=mW867sK*{X^VQsj<)ziN+^y*_gA{LxEV=8~
zUP$tYoMP0A9F0^{2)!jWscm|ulWPA_O>;=XIQnh*2`sL-mjqks0IFyqqCouMZ@+ja
zhsPs{xzv2|gnE~Ay13}A?;V{dli2V*?Dh&BHBB|X;1=NL_i*^wrliZY1_-caWoty(
z#*WHj2G3V6*8H(LXKU^U!%u$PD`{vY41j&hs6;S}VFQ@WsSn^!SHgw2CJ5!*#wF(!
zpaoE`$Z3<Z<t?8?MFY=js|ZO)X|3o>B65B9*#fH4ioQg<IG{UxYr>de<jdrA6!2Ef
z`P{96t-Y^<(R6%x>3*I4Ld#kGGn0rM(50O{XdS*>pCcJNFeCZO@?0L^&d)zsV~F_k
zLO4P(8nJ})N!(c#dj<`ibaH33mDfEQz1XVFFWjPThgOc)xo+oSpRjBGSvaNK)D;Dq
z^yt^Vc0DHpv<WzBUF#VDz+0XOE+Qo?=Tl%ti%Cn+w@8E%yNIylo32Et?%JC*82=>y
zbTl|FNH*2mTe==iPSEj+*vX@<%1eaUQGztpC;PQAw~6YNBgyTmFqQvH1jx4rUe?me
zqA!bXHlPc!0iwXQdKh(}&6ZnnVsy3oD-fl~oZ*7XOKFKrU9MyM&jG>191Xz9EY8_H
zndPRrcaO{J+El8${torSVR8?o^5#L{o$on>ir+WNZay)H$+Eaa+$u_}S?jboycuB+
zeJKfdSz!feFi!|w5X;|E9J0rom~Rpd*8U#T6gnvcvy0Xrnj$&cOg5tx!B;ZbPj58E
z)TGa7Z01!7T~;0@wZ_ZKvup>;juMuWEe_Y^m0}7_R;ieYT$&a08vk>ps6^cuWK$t`
zB-5-;Fe~h%TqsJUR0fd-DUv#3G6U+5+*=9`I-p}~wG3sH-`bBMkn5bC;ev|N=26lZ
zOT*4j&*q77eGA-A)u$W*^>+MjK7drQF6hW%>F*uI6Z9~CjZ^6#x#yH9M2>03#2sS+
zyTw;`?Gm>GG`0rN!HIkpdYrhkJwVw%!zf1Jj3q;8-;>h;q*ZF=eBj}7W%myz02|qj
zayQqZxq!k<xon{G@8^6C6}0*y50Q0)_w!HeGa_Dt8;V|xaHE?zx7KS7@OHUcsW^ux
zu=pWfly~zOoNp{cyO@n&N#W6<^Szd)jRaSc2LO)bN!_#+v|P^vHk1m^GV=4d>3p&m
zKEeunSIaed<X(|4aFSHHGJ$vM6JJ^}5L+5L02JmrL>Hz`H|+J?4Zs_0L;i!3Mr?3y
z?zPX0pa994PTz0oJ?DYRZ72oEXv;i&{=g{)@SsVv7hJ&RguT{cP+@>@HzFAI-s+YR
zE+Wi`iw|pa2>=M5>c_>cb!bG{$db;D*rqh5rh6ZDQ&zxg_oNyWtIJ($J&`8>6g3#q
zG5&R9Mi-{@@W@+6x;wEBmuc(5aixOMh8l5|9%EE)K+(UvrqR<sXKM*TZC4{3fd1^1
zs$FZYlB+==vtM=qxwEj^?kqk+*$~U79I*~c!f1WuJnvP@JmdzoO;D;T^cBTqJONdL
z!(o%_L1!<?x22k=xvfB9yTJVS%c_Zzge=_T(jSz{d<UmO5(bshPsTsh_Rj;#Z{USz
zu0P6Gp$Sm~+;|~X_R1~7@Fv|W`no@RJ@4~j#8x1ayM~hT3ts7#f)*&ge6|-IUwGuf
zsy5~`FUcO7N2I8>vm|@bHohuUw<L=h3OMDjbHs`AR}xfT&tLb)ADO7u2s`EqkZ`9A
zmoq=ku8?|u5cw*(73?q`?%Ya>stJD6;0>;K#b#;xfUzgqi!7+pq14}#(u~?|Y*!m)
zh>L=+k_U`n6`G?6x;v<PrLs+O1%NdU^H*{PD$GLZ-13o94tjkE5KNbtB<rMp=-{xT
zHEbBwLt1h&!YRO1IqV#&yw)N2X!JW#N>XWCpgJK+4SsKFO+WUBuDJBF^OIF_U^{>)
z*s;g<xC-vW5vd%$eroy1j0}ho-OCy3!dtj==*z7_9zF+)-aLedGJk!p>=h1{v;d1&
z^3_+a=XfaGjd3eNUdO1K#TK#LB{Y0rAEyfjn*m^eq}EoX3o$A3bXW;Jo#mvQSt_=S
ze^x=S)$zX?BDFJq(_sS&%WITu=c!THIrQ2z>^V|kOJG*&wD(7T!T1gGVoUQ7$_oJm
zdFqug0FFei_#{?Njuee3Bu9PX&^<SrcT;9c7_tEPTw6nlJLsr6f3DIMN~Lt)ejRdW
zk1QQuYUObxU65u_iUnM$hy^QACrXgOz-Lb2+NX1R2lTR5wcGlqdhCKE@=^pNCuh7B
zb2$s>;d0Y?P@MZ^OK~O{)LZYCGj4S{hb@etc9;U0h}4v059e~LU3#dMT4RP6|I&Eq
z@gKU?qFie@6iF!-eWRUef#9wxx=qjBJ@6p8i^n5{!s{Y}(zPkOZk}(c{EisdGbMM)
zl7gl+FtxQbXGA*8Kp%Xba@+i1g0$YE-z%5cGIaU=S}v)zftedvUOcIg-V{iLJQ3W{
z<te_Ps!>$!pgEz2IIA{DykpjtJ!E8JNt!1m<evN8+g%@4<rO0RP`MsJv=!IIHG{~}
zq>&Z+@%XU;#f*D<w#?7HfbW4ekL2ZZ>^qIa+KQXuaxg!pF}Jl!oMjDj2>JHQbW+vq
z_v9jr)7_<oU}>SRzNlS9s6(-N^>3?BP~9OpXTlwz=a}HkQsFTv^Wc?(BrWa<Lc#c6
zjRY;(*^&p%RXqWarhQxyzDUE($VFnSiGlkEZbKx1Se^wOy%3Cpsxoo>yVtd0I}L_#
zq11Yk>r3R~X^OX)UF4E%`{>nl%x;!1heyZDT-IqT$THo$^7rJn*OtrYr01Xq95VV%
zq<Q(e@1pbyHZ}@MZ$#EJ%Xxx1>+@1Xs)*fwS94vSxEs&}-SV8KapL+tnWqD|2*i`^
zEa;9GEk$TnyKF;aqHqjl=0rXm+#{bhtVK$V330t&Sx!+_%!>58T2|C#&;~Evo7H|{
ze?;x;>FV`Z7e!z^ofFvZ(w6ptJQpRPg;9a86|n*pRbSjDRl77zZUHZcJQYN&O+xc$
zF`7=Wczv#$$SxBa6+1X7jhTHiH8h(T=aBRYE%WKqS0Cx&se18T^$MYw-{m)Q%eyKu
zt1+&<{G49NxqRk}pGoF;PI*~ur&^QoxS-OK)F|~b0UlYADzAv*wR-hdg<sbCZj-XI
zYKr-m-jDozr;c41?tJ-N6g5z~0jgzgiC}x#BZ#Z1b77iaEx+8VM6UNs3PMW{R8P%z
zky@#QS%PHwDYfcKw*rkw^v*sW08>>qCZ}}TYRKU$MDr^*zTBs5c8&&C&i3W_V;Et6
zbG0>&<c6Zg<fHJHV*hxtWg}`MjQNK|bItt&D!qjd7)nFZ>P)HFR6-UY<pk(*NZ-VK
zC%in*2gC`Qq|O%IyUt8isOthA99B@w9@4snx6!K2Rv~AvM7ZjHRV};t38&|}mm2xt
zci+zAxK}mzz<HMBMPC7BpAY=WHGKg-qlL`Fe3Hl%W-PYCvEB}+^$Rx~@&PY%R4F1I
z&pkFnixng`e0xuyR(YG9(SGT-7ytd@XL<b(RkUa@fjrVelk^77gK7{>-2FW^kV$f0
zS|T+Fb4QP}YbgMW29Bl$kd}pRHE}u)`lWE+#x@rHFHUO+)qZXv`Q3j1O1)%Q)>eab
zNyos}NFi=o7Ixy-DaBOV)7spGM({5n46P1;luGJ4-VXO177TvI-2B;lNdV{zlblNM
z4)jROb11=`!}KOLj-17lE|$Pu9c5@e;G(bwFd%mu0*h0a3>K<ACdaanICy9Zg0tqT
z1bINKNEqDTnKQT_?ktGm!Jq}BI1)2jjd%+kNLNN$1e;dg#`WQox*cTArALZUrm-c;
zbM&z5bEgeese%tGOyJbScT(^JZUQ>d-pCo2U;1v>j!Dpit%eSnIXdXU_}|dtb?`4$
zX>^JcQD9AnprBFB?x-&xVRAp>5ncPdNV6{&CPAteTYKiwtNEmSX{CLmyes1`Pt~O_
z>y9r{6TMezH<+t3mp8k(6m+_WR3{>Iq**bk<{)*eRP$j(E*=eyP1SaIJa<kHkKZPw
zpVXof0z{mx9uV6p(`%{~N?ngS+c;9b!!Z%b)4}ym##IGqVH8Mn&WDg)(!n%QT>yTq
zCMFG`#WX+<>)UE`ldlilcPrC3C6hG?my1&bK@$f{O6e!fdpk|2FdtO`b&WqWu%3mg
zdSpEd6a<_zum6M2f?mFgZzs=T9(r<hiDuNTRR$&ddYX>LTd4fyaOxh9->xsKD0wbd
zkq6f&*9%6#3iFls`h|I+;xhS2x#&51GXO2L*Q_5Hj(GsvVv$loib_VHI8un<{Ugj`
zgR*;Ea*t{~_SuM;X!Z2{)ccV?-t3oe+(;G#0x&OKk{Jcm@LA@21ad7=NN|4({I33j
zjFkQ1A(3BRl9CUb*i^_Po%Pk>T(dYC)CIeC?YVqKhc<>q@GfEAH%wyHu0+z+v|WyA
z{yHZ}5i^`zxm_?!?@Rs-O$Qd2JG|9(un>*tz~L|XRBdp}7(*$~AtH?0oXr6B?Ji&J
zd|CHF_OZQQ;q!Oc1aijPceOI9u8ECd^bu?Lf_&xz4-i+?f9djDg@e=_)wbY@10qbH
z%O737o(q^6za>U9SN;z$h24FfTb0)I(naTxJ4!|XxfYYVDa<Lu3aZg~av8RsK2o6q
zt86ikRLWQ0>)U(mhVzJ}crGa!4^nr7K&**FN%QhtnZd{<rt>|G<i&mB-+B9{Df3kO
z%yeK%{svUpaPA}tIP`zgcY)3En`V?^PkE>hh@FJwgJA-1vq8pTny`^S`51}}>0Nm#
z6gpm;T@lUWbF;2I+FCc0=6wxOc(-}pbfDDc%e})M#L<R-Dw11fEaR0O9);mZf$}|o
z5EfWbNo9dIv<*9bw?tQ|c)QdWyA@ef2M9E_Qb$mn65~sneghAZs?k~rW)}BGS^X4k
zR;Z~Jy^H$5It1Zq9WA6nQEol}vyvA~4X&b{l9eLRj6*pxOyn){DRTRYA1-vQ=5Wn*
zt4muYCl&1|2Y=@Jvp18%<>fE$#w<5MPtYKiTUM+bMT?q*P27<{Q}<UnvE=jR`tFc+
z93el=i0!?-knfNu+eV2m<oJW=afkJH;=4_ufe1QTe(Ko43D!N@32!9j`jg5#tIN7r
zw>vpFR(nf9UhlfX$4u9*9jsD4xl!`fh1f+T9g<3hgN+sBBDWw((CSq<^b6aJwRqrA
z!m1?<Nz#Tf?RL*A&+h-@>TSOxzp6XmzaTH~y)W)78Z+GucJm%%laf?==$6z%rS8Vq
z2A4`Dsq|1Ks#H=-PO|{x3mJ$Vk_my33Gu{C?3rN}&|r))&RQw=T6rn`3;ZSb^WEpW
zPl<6>=u*}1;hcT;*=Ij|_ulGD5St=eWSQ20NeIP<BrEs<O-Wlhe;-HbVNnSu=hVqd
z5MFaqm{^*w$E80@^XwKwJdDoH*CQiCY;RJ}HheW|Cf1fWy<l&u9ui9C;BYwknQ7ML
z1P|ZS&1a*O4C-3)ogyWo+JE{_+6|AMi8o<#rM@ykgEAM<GBk8GDn4fLWK7Bhm3;H4
zm-$`?rW%=E!)6UjM<Wkkf)|UOYa$U?hhi~E*wR;^y}00gXeAkOTUMk_gF0eJR#MeI
zDrZY3L*$SV7M@x)(N^X1`<{zZAkEg8Pv1g%rmeg<#Xg*j!<eMP1XN1R{rcxUsr_-B
zcV*yy^}5PBYmMYqk7NGPS0nYA@q@jp&eIjbuQFFf`I#mCZzIsPZ1ogSH2X^iS}AD>
z9<O|S<dGy?WNF9yQL+JD3)@X|RP5FAiv3)3$YkZ0?lKg(tSyAdlj+4)I8EW#z?qFi
zI<t%=Rfy4&ncoiS))ZLw|Hz#v;Fk7QMtR|GSVMZ7NxssnlJa!&HXEttIzIWyX%<#M
zJq^%fYz&|s+6|r`juh2Lp2qFvrHCbfxQ$YTidao5XLt~|@VVx8SEQk^7qU<r6y_-C
zx1C}-N`U1y-zA?eHIq>V?}6Ub{z3)PRzdgB#E5fobR?z$SLHNG<4_nqbFykXl-iP$
zpLZrdH%Uf=UPhcxy9XQ>85jr;-o<ZhX(si_56#Or$B||xRvd+CRhSNuudzM(ag3gA
zZifSS1#?eSxhKS@JGEqn#vvelK`IL#B5>@eQX&ae9A5qbeu#&#LI&Q0l}@R$i-TX+
zDW91*qsDYtS4N58oK>0EA&4Zt9utZK1}bZG_c@cQ5lyA>{giT>YE3gLPBma~acpCO
zN|2PWrdk|miC;hk@MD|onn509s^Bh7!<%*%6Ct4pmp>9A(HVw8=_L_R72#q;lXvi1
zq`{>Kf_8`>(%d;a$`iSBoZj~Pob$c#Ou$}?fz!a|gM)&<<)y_bE$UD))O3A-lgJWg
zt>nt7(2&3Bx5I@?=2^<kwK2Iawcin8(nzmUC2`$Sw17BbzjOIR(<1N#p|TUg)LZYD
zRq{Jf<}{*Lio*vzc9beVIAJ-(B(7GMJa6iXmb6jwyg%a=G7`6@OYVmxoSj`2Z9$cs
zC7U22f+mEntz4rLNEOqCQ0%3@h@E;;K>et{AE1)I;c>1EGSy~QIs8;0Oe)>!mCkbU
zY*F~)!{nVXq*-3ga9TKWy&H*<(!e#je)<sCSa=WR<!Alc`Ra$dR-hG!Cy5x_=D4mX
zjbb-@QI97-4_j{kI7I&QH~l@B%@mJdq|)5H*P8KWY^W{-h3PGZ!Nvu~ij&gA8sxjx
zdTd(h|EG$D244-rVY7P0SLH`Xcp6UhpCYa73_UlFbW=<oB2MOUoXiv_DZk@)u*)3C
zKY~8}V4Bu|(s&$xCRc0jjyYwKlFS({Mb!Oi{rHQ2ntH&I)WKtO8(RIzJH-qMkF!?i
zv)BCS8~^@4{fQ=H=52zizqtISzus#Hz>N)6bbH#n>H!ZLCMT@eq5qh8Pj6Dnq<()o
z9q)HaUOtRSr)<vxrEDBYH#9y^exsG3-w39~^u)Hr;b&kq(ED1gRfeIj-aH5s<X*F7
zB}lq0=Bb%OF;!3N0lTpD>cuJKyOB-05HoETuJAn=EW!1Rd9(VJhqVR=P=XLr4{@#3
zE=P32DLjpgC#g&;9L1?b6P7FtG8zP=-*Ygeq+Rbk?Ac_VK#tdvi4YP;YHGiOL!vzn
zNtQV0=brJjcn&*8^FeXqqL7L@^{Fu_?%k`dOIkSk9-LRm&6xS%oE3|@o1i{Q+#HLu
zAfzMjDzXd@+gTGqx%8$JK9|7IVxiR3d9~I{r4WcK^)9M1rZQ3*CVY?o$IqalK?;hC
zw@V~bn<@_ox#dAJ{L;&{TFf*ShFExC9Ed!7e|oHHzW0|RFH86k5ueI?X>8fOX#W7Q
z4bCMK-=MY<aJj0<&LHf3jkF*1op3W<zFAQGOB;*c<0G{dm15<IEhDP+emd8LaU2k5
z^S|O#!GYu;<LUJi6vaUUujthWJS`-^PL~?LskCqgOL5erw@J14f)f4Xl+~B2DsE*@
z=%~;He``TE&KjaZ3$aIhbMfT|?FIp3nV!8;OA6*Lx=`qv!%}_`xK7qG*i``yZTj^M
zT|nIt(?EEAjfpYG*c3C+N-Az$14o-w?A4c&%A0X$VGVR2`Vr)Wrh}Q~1JR-PTz|?V
z=PPhJx*P4DW7TXzTQ$v??}`@?(=kNvQ7;@X9J2a(A7@mt`UR!0rL$0_&b;m5U231?
ziLy0FH^0ux>Y7{!<drAM<nKKKYNmJs6rfL)hW?L!@LV_i*1q`oEEEa40GQKonyZ~y
z?jvTY<{lZ>Q+;07GvHhfkf5d`lx@uNPOk`x^8$ibDURX=WM5HzB*#U@sy~k5Y%Lts
z^^<c1PEz}sgwOT*oJO_7B`1&2c<DOcrc!+h%;UqXJW?9o20$kQPy%CL!ON|-Pdr;W
ztg%JIHVn<R3O&Tcu<p>v&8xY>RN=_Dcp(qwnaIhgKPuhMdRq|;Ks`T5y*)xJQy}*Y
z%j+d*7Ef;MmOwsvY~fcxm=-QBDUax0Ri>#1yQqkCrejI!QcVDWA=<^PYgt+V6N81!
z$weG9SSyO-rZz{di&zWdfO;nY<K!OV1s?3-Ae2kcVU{OakGLG{d-85n9&bk^U{Jl}
z5Is&JF}5DBV!irzKm=19z*(Sw)|%^Pgd+-g09hit&jcaal$fxO<S3x-A0pohp`gb}
zip4SX8;Q5e?`YgA%TpDa$<LCRsdu}@dE=P@mo!fjHw!Mt7|UoGXh5RaY-wLYxU!(>
zNv_G`UI}@cL2objA84%z*n<JWd7xX-DkPY+HZMP?ouhWR!F{`OCs?%L5k`1wT<TL-
zs#`Y5nTV}M@sAiCIZEN(0yUF}Qk(F@GuxeRcpFEl-X6MML8E|wcpm~~GLRg2@LT#&
z4go8NP<=uFyS!!+Ik%$Jz=s%Sfdzsh=`=+<;*6Eoz-uGVxDIt6G>Rbz0g*%_88y?X
zDREJvK)1fGiOPJqW^x}>3&$^`vwPv6twB|nnI}$GUyHGwhYRQTR0RXljqc^A5XU*_
zI!{yER#zg<euV}d^@>R$#p!pPp#<;+oSJIR;uQk4@3QOiu7_8Y@vh8>1H1k6&@dEn
zKr&cY#kp)>aOj%Rv@+Hk$p~%&N1AvCMObrJ?s2+vQAzajh<E_1DEJ|2+<7yrm<cUZ
z;^uN3ac;7(o+vKG0`qR~48q4RO(NRenq2b$L-IsphQlaeg5uE0lbi#zG|S1ffCt@8
zr5AgBF<x=Zlwsjxiap%__{XusiTC=^O3$um{>6tBsz@Dwt9|035-};2Qt&Hczk7a7
znky496(1gFw;&`)o=*)Gd!nwHX9Fpd57<s2U+{T{s5=H#UaPA_IbIgWx7}gpqF}cZ
z+UR`a-tvY=MOdf8Ubrk}(ZxS11Y5`)%!BLB4?CKo8d{||z4r6PuQY|`3nmC<WEh}n
z*ik5L%55Q13Ha0vw${b}t?|*oQG~_&64=)B?)a*Lh^moK1QCab8ih(Td6!%szlnBE
z2V1i>rri1XH9ZgQCDyGqh<LeHCpQ9IP5Jg(yM-=u8b_y_pb!c<riYIGORESA@60Bs
zKq?%IN35iNPL7IW%0wetSwr5#Gm+{f{`cOp`2d$G1Pi<J6F2;l_9nPL+)X9#oFTJ^
zC92r<-zqgP82-|R=gzM9h;{gbW~@4l`d76hyg?ZiI~0Ui<*mzaOTWMTzAmJ8Vvrmt
zKuBLI4rPK;=m2soi83eM@ZSg=<JYRyq#`C>_%rnQ!KweCusZDN3nK(S6&U2N;CqXb
zaK4%#1yxA>2PPKX<~1~qwvfc3VW2Zx(jEfu#6BA(Ocsp;h*O9WVcW&t7%u_$@FS@R
zhsNpAka`&P@g$C-$i8VltvaXa_C^Vlg5R`ipM6RTpvp4koR(C)Kbwn8&vY-R5JXdh
zWdU4*W&Yi;SOgi%T&Ov<T=i}Yf1cj+#VClidz=vH);Ku$`pdWcuB~x(N$MD`MQ8h^
z&?D{6c4A8%mBbne5xt{8vI~ZOX4%hLL5Au_;nq_~X=1&gCWk|q2^#{s2^i_q5+t<m
zmxEW-8jK^o7>`9>Lkfk7-$bpoVHWMX{B<LO>{{m5(%sc09)<}{Ho^Hn`G5T9#dr3C
zbtQ$Ce{K4fEzaoGAv|}hQYpfD@Eg-v1xCX&s`<!d_Uc!RQVzS=d750Sn};hkLo1SJ
zryU0h=58lL6DTZHXzhpcGRNtmfvFNioT6b51|wyAy&0Uenm1TY<PsbE`=T%PKv5)h
zD<JBsI312v7}Fdqy+a4Em*f7@l0K=da*6D4rt5xOXzhgBEq5!DKoEq~VF;$+Vm=O5
zkN1WcI?1pjl4C1oIm_M-B~%F-ruX%yy^_lHsm+dbbmX`S8bYXZ2CJ3c7;jb=1#u!#
zV{Jtw?)L(!(rT@GnP!2$hm1cv;7srXXKJ>FeScvgQo5sB4xkfLeTY3@68W&%N(E+=
zA00la9yvWll28IxBb=x4v3Q%ABa~ARQjV+(oyw^?TafJD0lYhQ8}L_iDB_KE;KYiJ
zaT+j?M$IxNvu-Cn<X@)49V=-v25cpNimavrK<o9l<7`~RMDj~9u`~3!)7#9&+#S&9
z%cUaO22nprb7Fo!@c!Dyj{yPvMq$B<_KryG_r?L4SYq(HYoFMM4-g47^mUQWbG$1*
z>ZxnP(QrDp$!;7xuuS|b`&a8$N}+fKL7w>CB%*DO^A9PrB?iT~w^H!lJ<ge7Gt1`^
zo;Q&08eS53^79Z?N}=-Z;51b8!8g3I5mET>P}Gz2=EnQEpFt^Ar<ls0<0VfGdy8|f
zz?2x$Dk!-;o*)`PD$`+0m02>Sj|&$<oEK1Bsabe%mfXwwY3GdSz7PcB5DR;b+t?wc
z@G$C1ElWlKESr|(9ICWJG6SO_l3U#<2dM}dsQzscX$=)zo9y#1iT>!@G&$24S`xY1
zX%JP2+HhKyu5kfw=^am#p>_Dyg4Ak?siKxz@0~#)=Cm#~C1<|jMSzfUe1@zTa6`<v
zPnAf$K3a0>5%@^W4uB{*Br2DZ8_^42v`n|!9Tg|PE;rb$9>cR04fo{3T-iB{<`Uei
zz-nw>fRK9awVN-C9_k`C3>R9PlfZ%ptF8#q-d-`{DZejL5?_sDG<+9Uoo?KG=6T;H
z)i}0I4?KhCL}@m>N4@}$X-SpqwL7Mflv72mG?vLbvckMe(EEHBxS*n;N*EPGAJ%@e
zP$U*FlemwzwoN_q@<5!1Ew+esDPpvtM4#9<h%z&;oSRYHTQ`=PCF;TL@QDq_a8=e+
zr~OzI9MB#tVsqN<O$_ZBfD6#e+E2a}x?^K?L+d^`m^@Cygtg4ACO1#T7PEU*NFeH^
zspqvN9@XoD>F`zR!32#GnTa4N3lsQ(I?`*`#rU6z8d43p_7e6|8l@?|DOPBo*OuIh
z{%QEVI4-mRa`sCN@EA5p)bP@x+fK{2Do*3GfxS#XCV5FUZE~^6-D=ZHAkJ)X9D}W_
zNd%C!)DKr|GBpWY*kS8xt~YMJa!darby9_mepKuiFJwX2<ndI{R^~BkqthC(lOlDq
zRNRrsiO$BcUV`bB*d^{!lA%Qel{hc@9*aZNP0Pn!y#OW%A)zl&_p5>eC*UG9+cpFO
z^DXV@7sG)lE(<kB>3Goyjt9mm?Ymm)itf*-m^H8^O)X8aNW9A|+!-XGuI~>(N1`<O
zxKi;@d16r1{WKJzt<ZCK(aNYbl1BkPu1LmaAYb@`DQ3R^y@pSh$O5V0+%#u*#+khg
z0x+$e4rCi<L!8YnwQ$Q`4m3)*KW3WiVz=FH6ocm-->e?0Lp9f9>Gi_N|8{@{tuos@
ze~|(&>79vEyd^~Z@ppW{6{8yK?ab_;5}eVNO%yN2AB2swCNg7DnqxA5_t$^?yT7*3
zC;*a>(&|y6v2gxW{MM@Qa?(|TDoTMfWuw-no`$o=9U~wn=cPE7y%YO*9-Oh?ql6X%
z&AL=Wr(dYaj*C(7<j`=yXz;q+`WOFPHZ>XkRW3%$Nxc>fF+fOd^-n#m|Ghisb-LXS
z9O#W#U()UcXe2P0DMrpIa#v2TlrG~z&&(rlK2lNm2|sDQHp9VR#n@9tQ_D9Z23l5{
zJo{qwN0A5`+=-)(4|AFQU*Iqyh1cj$t>J#*k?7QNfRT$)9!r91XDcM=K9CGQx_ugm
za{-qxzoDDgqO<t8CS<NpyH#<Xg?7DK9H}?}^q!miXA;!jhDEE#jPnBFk5NB8FRzeP
znHe`jh2TgT(*&VWt)c?#e4HpLmXMuW`@8{;Dek{%J;@z>E|#KX%+UbGe{@;#ay)PJ
zLJ>t{3leZRWKK1MF=nSfI0?eLgBqvQ<&VrsbmOJ6*-l6d0{$%&ks@}Bwd^Ga50wWd
zP9(J3aR2O-(!$`ERr1*n)CQyEM(XK;2UOi-rCUL6DvBpZYq}|R2ZXbf4rN->fowlN
zv%LplSc#!>3y`M~r<B*5j7x239kH@d6g2cNR6g{l#&$F_h=Ae;9<Dlg(ZV)Wh!(h&
z<nvi7QkXq&raPfUFiuYkw@aeq>D&nJ$1BsJA5`5tk|*I9P2H8uiGJ3+<@xCb;lM@X
z$ODT5gB@K-NQ0<BN%?&8x8p2;ND0z(jp7!Trs+wpt>4ur^fHt>e|J84OA9G>oBTAz
zITTLEgtn>}gk|(B-~+N}Jp0C~zeQVi`4cu@YsF7*4D-NylKlr?eEC`3g8Q-fw0%EV
zMz^lu{$`Cbhws>1E%1(VAcurUNh%)M=dNW8fC{*=S!Y8fZ;?<Y%~>0ErMbG{RYfbJ
z{dtkxkwKYS$!+U;yq`_22}I6x8sV*RCJ@SwMC$3e*k~bL84LPhTKg(-npTALO^qVS
zU28#bBE}&=<*4GNILC4ftxz4s+nL?6-yf)yy-y_Sb4sWd&q#xH&|}#we<+<TwgdM6
z+_lg8ZQwCA%Jy+_;sP9b=m!ZCfygf?1<hF2r8K*O)j%X0XL!<0pxcPDi-T<WK}^64
zCl;&xMu%B)iBix7f|0?e9=}2rGzSiGO89=2a;g4Q*btqrBJMCzOV+j~SSoE@$+Ma#
z(M!|o;Apil!g-zB={DtDV(rnfV~v<D{T;%%T~b%(!pFhewNe=3i~W@cQz^2O4g##`
z@IMQeQB4!#iU6lXZk07IXVL8l+cM<$FTd+aE}(i0;|STDVuL#E;%ddsuiZq<xBy&O
zj@xloO{H{mHuQe-91f#;MHG3gkU<<ODYB)&F+T6kPBvhGh2Sz9B&>DuLWYnu<<K-j
z_{0hf0(ihUa7bnBO;3pMC&Lt@T=9u3`ahL2^VRT{^VkGvfbDuG9bSzT58%Zt)2TDK
z6>?^^xt?Mk2<K^*6R5!FvT-BDD)hl;p2^)$mTcfGC;vN*r{jn53EdO{>Me1+8;Jz7
z%MH;;$klW~gf7F$nL(_^LH8lt;N!d`WTqI#X`m`P@lX}bu6+S(r0}E5H$C<N(11gO
zse)a12cc&@|B5I2phquhjYdW8Q+?7|Nq|R`_6lKx=ErmPsT7y35K}Vwt6}lwG{ZRT
zyMd;UO5t`*vZRg8%_;+gl0`c;w_N**2h>sugFgeHtL1HQaEgk;0TpAIx?4MeRz*w5
zl?10%<nrX_389duCb1P%;Q{ESDXoEyK@5a|>Tl)Asv5;6IbMBvoBWrQaUv)t2AOC%
z!u!M)v&nn9-;?y4h$d<ULB!;}Cm-qw6u?pnjG2@2q2pofICKOOCv^qhWA;%7oLXmj
zEuHUBVoV%1=yn!er;u*E$9y&#Wg`CS!mNbM<bNi3K%l`?5^;uNUHPvER9(Q{^Qg9B
z`kwC|HSYc7KYsUbyxR$F0rV}bO>q7X1=x2$Gl~?h_fcQEaa^+UH$Ep?)84B3iYD*E
z0lCnmTUF*W4j;Pt*L^U~A=Ws~uz@^)qEeN<FHt}C*B&F-WZwv|>^KrPpM8`s8lIO%
z3e=G+L4dX333FlVzCHjLFFq>9y^e|9&HfUcw3&#8o%}&52gF7592_C`26U-R;S`dN
ze1hpF0tlJaxmY}*%zCA$3y)YG*XkwzwfIg1-P78w-dpkD@%q+Q9|h2;efa8&ns<{N
zL>FeUmCi#s9@Q*SPhl(R;$c>J;;`CJ+e~!_%mvB_L&SRFQ4=EW+@HMVA+(P1F8Eh6
zVq#UoX%HF~VR`(Ddfmop1r|NlL@s{i2Sg2ZuV8o(w+r00JudhWY@U*IKY1HdG2V{*
z%qx4|&qUV}@=rya?NnM#^t%t!o|XZFK2XG@ES}2g<cjxjW_mz7By1N7bu+gz$~)@k
z@XaCp#kypd1V3U9<0*CTjH^PSx0m=aOt=h4wGT81)icRCOF|*EMI=%PEe$ElO{FEi
z_Vyzod9!M9B?G9!yQsPyB*sY9xM<QA-r@5+HkYY7<Hv&i*os^`v-j<|@%$_<Z9a_F
zv?c>Vhi*E@!#nyr(Y)W?MH1-~dk{-5x1O3x|G|a!kdnnry_IoqJNkch`Mnq*{b#&>
zWaydCu(@%#hyOuqvq^<;{SrnRXzJr&O{zRy5lL-%S%WjthGlV8M^Ru2uCxN#W97x{
z2T`El6cDe*0l+5gfoP&E7M9WkB0yIjc0?f#g@<_*3_9bZJpWnk^f!7U!*MhwQq1q-
zSKbZ)_v1Y%Whw8Qak4U<;yI@{K!7jVd$q)W6V10arX(y{`wOV<Q|f}VC8<MWTy8d6
zO|Kd4gL>hB`22G*pOV<As{lzBUXn&O)bS?@VDp4S8OcC3``pDO`gmCrXC&SP!0Zdu
z{$0~%Ls5Q7Pv<!mE2$JZJ^RMZXaCH^IT}Q+wIXPt0}IEdX9=;0_|`g0Kb3#@;zQkg
z8=sNxJ?(Dpms6h5m}^{ro&U_u5Cqo+>aPcebSnjB1#i`uPX_7nkO_OFM{XZ0M@v86
zUE9zXTYJdxm>L$Hs%UCL5M=UQ@j+_}?jsvZ%>&a$C&5#<M|xIngRDuUM*tbEVDgng
zSi^u{iA*FJw=i*aA+cx27?_S{7D^ay=3Eo-Gr9DG&>~{_h$<z5kx=cyX*=+WBxI&$
z)@8Ld)!lIlMzE>vXy&js3ZL&4uj$3lYRgDmww-iel(rhRWVm!71?fzsC*RZuwH64O
zvy}d(Hiemq6vZ|}0}!R4>C8+tGXEk82pd{6Vy<ZTVTs^WDJa3Ww^Nmxyp7*FD1;L@
z4BXFu;+aJ9V-dp9dM@O{6{71D>PwN^7}l7y>cH!5IGmffmL*Z~;S7)P8i(8~FwGbv
zKjSCVQ)OW6L~P=I$qBpHm+>3pfjI400K(?ME_e?PV@oi{`CiZlhd?b|BBwhFTrngM
z*}955dds#{VA!ipaw%RhvT8Jd-K!ZB>*Tnk#{&hMF<NZOZ-Yv$hk+23l9<XILBd1W
z&F9BDCUh;3K{e{gVcYM;1V`Pdy`~x0!nHcew+nJeCY<HLrd;qQJ5$B`>c<}qj(pA!
zMqvf;Yp0hQH!gMU(o`ieS1?wzyI%gZ4`;kokuF=3&AIvdQ(_&JFBSjZxEAhSzQ??&
zgl-3WjSr4O)7nD_r65or3O`xqA(jHizK1$Tki8R0QO&r#HKtsD4M;W7!}Mk{vqE-D
zB}?Wlb$0gJ`J~V{LxE>ti==w!M%7NTIKtmuwtc@pdAGJ0!m#cu!mEq%F=e3QB+uFr
zc4jFoFdzPug~0$Aps2ulX31(wI*bn)DnOS!TCRV}<Gj~T@m*>o6gLh-+&BvfNF~py
zS!pkpZ{^))OA?)HFklPWQcpn_3`gma#QvQb1IxUBK6x+cPr0f@lO-@!`x6eVD*T8h
z#j)XAOJ8pK^}tc|i3BXRLEJ#RC!a}h<%ec^Y6f!JhmQ8~>si!<W?d<=TEg?d%cfRD
z)gLNWXy9GI8~+fDb^ViOixKFevV#51(UCV~YV65!lQ%*(90sPwt9;NtO@;uYj22|Z
z!}g~0pLj%^e%#UjJ|D^MU3_f5xeGR7x|4Zb8g{Z@F25-r0mSmA#|VDQ;A(Xz_9Nh>
zb5U&`b!{t0f*DQvNHFq#?Ph)2OBEO2dHcX~#3fjwEq)0gLa?8H_?7fRVwb<Y{GPw>
zg9;3ZyN}H91Qduj@UlZ+u?k=#HOF_M!6biu33Stvz8;drib15IBG|%X2)I;S82GLX
zYW?t=`=!@2e0=e34}GgEO#+#h{8JhcbBVG#6NPf|w~DNI2xPS+m&lXN-FaXjBx1jf
z6(<hq-Z?Kw9o(vU4sT-P>PY~Slm#Oapc70U`{|wYkAIvx;`>B&)B^$*@Sd8bFk`+A
zB_!2zuJ~TkKKR9Z%3ht6;LGE!)0$qj3jd44MYbp3`uKIdVYP)1aVne_+bISve#AiV
z%4)RGrkM`-rNbzyMC|p49)AOyOSB9F?gLYGv6lz3I$UK!CA7zu)Pa+ma!APTAbow~
zl$x!h;>y54xcr8Oi<r(^upI$;#U$`Cr%iBPS$k$`U6c25S}}Ti?WElu%CBndn*mJ8
zJgCNK+IVX6uF5=U^F8-jk8};}GWnx<=myx*fm+D($rnXTC=f}3UJGp8;Tde<bt<fr
zLV{cgMDbyk^kEUIP%d#C6cD<~S}ttyEc{$7lmrmX@bu64DAl_|GzmOCUJ$xA1$j2u
z&*mb+awXScVO<NlqBPvZcBm3(fyk0p>GugcSDZn*9B+P=>kqI)Q?jiMB7^zGXEYkH
z3!a;*NroTE<GGEZr^jdvp;~9dJn;&_8P0kA91SjFvDsr9cfdb>TLQk3++w<-Z$<K!
z>3Ya?i<b0N|EGof#LYzZDFYCk46O7mc1?;!km$DR6<sQYBqOTw8*}hlJ7a!WWD&s7
z4<%jJ17WqJ`&F#{>q1XmzUiN+Y;=m_E=6l(L?<W>cYE}!_NDGA8)rMHkcg;6LzI;9
zm`)7Q#H7<?q}WY9$j8aaCUR?dw9{@E+eh-OXe=``v4)Qk-=^4|Fa-U0jy)m?;aEs5
z6Dj2%$&(|P5xZrZWx`_90*yQSd9?4153(w1ggm44pd(WfWsz&+N`ShRpi9<A87nC0
zS#8{%i?|-69&Con$GKY^FG+<`8P(SlOFl5G;&lg(a0Z1mO;ES>xWEuMx^D6~DlJI&
z@#-#=_RV^qeB14Oni7zDC)mA-WN$?(dp$c8TR1CeDj4&>@!(M3>6@971fOOo_{&OT
z)BBJM<gJ_dI%TR}CueWP8sV5L<~E}$MzkOA>>|&-d+_83o;zk42M{WFT58Q+|1=%G
zxU;<bo9pX$GbGU#pu*qp4BEAaB9?M`XGb5u?pF~$kM`^WKS_1dxuj**2Afa4;J(tE
zeFNDMp<e?WD8`k7=~{QSKM*OTvtRs9E5#%m%iaPQB>3E>vq)EmWPRNAdwEDa>NyxO
zTlP>8CWX;<MgGx8Pp_oYT!;}~#BMA&&vHB#zYyOm2IXbZN)!_qNxyH%z!LtYuIhA{
zYgP;5Y`pjJ>v{tom^_+MG-7KIfDZEroR;qa2JY!jey*>i++F-mbAdlGa!Oqa^}J06
z!8>0H!nAPjh7Vu&-CLV$Y&1<U3}3abNwZD$`O)4?Alw%9v>f!KgRmG$=rtCRV!#sD
z%Xb&_KG~pALeno_TBVELMqEy#u~_rCoGO2sF&M9KMv+BY1eGcX<(@)4(?C&SU=@#r
zbe5CMVix)!9VHJ#4Z=xW+Q3srC`TUvHZ%Sf)-=V({49y=aIQ|?rp6Bko_A#_?tlZ(
zbGVJ%?PkT3P*n5Kl}V|SGSw})<fLQ#xcL7Ib^JE6S&BbN;n9AgZxtM4g4DAX$ID%T
zJIkZRQjo(;)TLLV9}`tZygLZ*y)P<4lC<PklxdOk8xU2|rE6*oRITtM@*s*qBfln<
z4)ST%&Y)o>5mB^**#AsrTG9}hvSJQJKgCgK&YUSuow-45+4Js`pIEGLv%A<#_vbd<
zi1Ne%a1+LiW=fb}|4iP?;?j#16n8Y}lx!l>Tr@>226A$X|4Q{Zf)^;{GqsJS$%fa3
z#B!M!@q&`(%x;}6jm7!Ynf%&&>c%a7=$?{{XDX5{(9IM&D6{Q3sqlZCP`}bcIkij#
zcR}mpZ`vKC&ME^ph=L**uF+F-e_{GKkNvt`{5qM7K`lnXej<p`9?Kr}QV}F9aeO?U
z{}-=k3NDCzrL=+{C>1@_HtD_?hmndfLdW*+Js~d5ae@vZEBzM2RPh!Z(g{3!(37bC
z><vv8mHi(I00BYL@LmCk&LQQ;DKX@{mq3lgs)S>Sp<v*72>tHq=tvL#!qfhyMNOJC
z*tYN~<<1O-wVuVK&)YFG^o`J1yfk=cl&Oj~?TN_orG276mHw6Q!|VOz_j;px1v8?@
zo<x9>;!4%64H<;yQOvi9NR*5`20$E6xpcN2(ZEwo?5P@U%0M)Va>Nhz3zmeYhz!!B
zn@8&V%GvcMGTAhsfm%IHS^eOfRwQtp?gku_(!vrxF?CH#srVW!GJv|e=Esrypa>9?
zxE+Fvy8hN5zwwVILnD*T=9xnkmZN<tqWC#$?U<UxyG#v#yU`BZV6f>gbM0iOq%LX`
zQLhQ7ttYx7NPD>dn59<!)o{}f0@Ysje?(p&F%^YSNy2z~>Z^Y3UORke8_SvoT^l0t
zfRMsY9?+B>smI8jnm!uUX>#Z&M;`F~!9fW+f^@*72=*t-D_52Lo{(@VzgP6HS+QXn
znYCvbaE{03Q}&TCpu<pO*npf<X*p3T?;vB!bHxhNB5ES6(62o~x|jOXXg2^oluR0%
zy`7@K$>1+kAyK8TY9)bY>B1vgHv~@A3&d<jIQce;vnd2hhK3hwM7jyfpEkz4x<~{v
zWOFkTjb6TaOT$F#J@q5dE6T~4HYpUO4}})6*N;V^$zBDCAzK8u_QbYJZqsuwWnrVO
z$<H)MTp8H7i9o6QOB{r@Ra+)SDrCyTHKbM*5&iA4Y7UsQFH(+^npqu-gmZg=(4?$K
z9H_<jgUVXM08=$~gi>Vk2)(*x=Q}D<VV?$Er(|}&?J%|jTsu?zVSJh|-g!Q)Hnp<~
z)J|ElA_R4E!G8Geue_?Sl$>3&bAX-nfVKdL6;W^edG@Vl*?x8Tnu+o-R_Wfbr@vv3
zPQn|mz33-@F;60Gj12lpZJ?c82$oNL);kkW15DtZLDk&jXfDm@Ul@#)VrSui=_Zhm
zMOYIW_<oWir}aAa4@yLmbSE$tz5>12jS_e&PLg2#k1oG!zQEC&Ls3P`P{$+W6@1A%
zQ8B6UnO6@prhN)@0Ba3B(t{GjMpqHErr??vF&fcvGI$!JE_m0{j<I!ih-1wYCo)sg
z`$G5imo%%d1y$sHIi&5Q--Sy#q}g8Gw;)n|8(7`LP?dzy)_~+~tR6o?1D+V=02Vh^
zmc6y8y-wz_cCKeS+K*KQz+9y(`||)!k0{1hqCyVlMUf@fZ{Ez(ZB{!WHdX6P1{I1C
zDA&W^rYxo?B700mc1gAfEr{M$rq#;iT_FuU9Iq@-o&06VlH2Y@twNavXtiL5idDiD
zbe=6dPDXI?v?0&~0*|M~@pcVQ#XQ93pe{^1yg9X2TjW>ZbkK3pNcZ8<K?ZaLvg(8u
z289sRdeD;}gv$5ituzzvH8uRE)Pvx0GHD|>S=&U_Y~x4>v522YV~7yLM|CD~LWp6X
ze7hD2!l3tPmpJIdn>l%>RHS_LQkJ|srKphU=(-SkN@HVoQLm;FM+q{bChn#rLq}5z
zWeND~RGc*gNtS4}6MK5?=CdM8ch~D1)v@q|>XrBk&>}DYzU1%&drWK<Mv`16UeNhL
z3BVzK3rG}bYJ_sfSb`q<Y{^{JkfFpd?fw>_mW=jN((YTD+M}<$dPCerVHtuf;XmBP
ztC8dD&qg>-i}dX0KdT3Bf(MA=1+5qK?O}ym&vBawjGr)ijcL5FDU0H*LXU_caabgo
zZoF~rlPLfu9|o%M?u^%mU0oxSr?UF1zla!LiUbZ83S)+Hw%!a}gQmm7)1{*npdCuV
zRmAy5ltT(`QSObqVG^GqB3vP%s=riK3cM_NzMmG&@X@2MhVW+_EZi$*@4v~kHoa7#
zL#}qO)owdv0YHn7zoXF|w8vQ^xtZzrR}y3B2PIgd#p;TRMuyTgcTdkNktwmyVk=bx
znKmCOLgn}4w`iYFqn;}@20)%oN2zBT1)?X7?yy!oSn&Qi?8#gD6keb2Y0a7KN+Za7
zRQx-m<CEH=0IzPvF6?vzQ^EDEh|Uq}!*izPrZt_H`{KV;pBN58#^kYwP{aN-6U>dV
zkjs)*AA#DTr0;tzvP%ObPNHavyOUDYImF{Hyc_0~xKwQuA`wdUs6D&F<Sw{izZpzr
zI}eN{VwIJ)xCmVd?8c3oq6<}Y>M5<2kdkKib3_DW$+;!6XaMlY6b4rBPpz8JO?S9F
z)@G=M5nB;`F|@S7PC8SFd$m&!*~*x?1c%-&at$0L;b03rWAawNow*rXhGIs|*qkB^
zn<mcY*+Ge>DFlSlsW}1y8&)))%5qFwe;$*p0;)t)>s6FiyKL6}D6;lVQt$+I;jr@9
zrD;mx+n`Uo2Cin0df~|c>gbtAJ1I8X{Y3JYc6jE=ZhGohII!f9GKaC;spu-USzK5v
zkk>T7KY3SAU?0>Zr%n9R>0G3)YEei?`>p18B<bjW=uEjxvlc}R1xnH^n`-z2H+yWa
zd|n>{8{@aabEZZBAoTP00U}Xf3h=BjlfI;QOU*<^75`eaAKp&nGwfOb+?Ene@PM@z
z2(^y}2eHS&^~7w2_#cFl9w$9H^x_kb@m`+EkxUYLE*(~}tjzJeJ+c~2Rw$eY%=m~c
zw3Vj>z-WaCa%YA?Qu>WkKu1fvOxlwl!cNR(l8@LUc@SJqhbdpxY9C}TyH4tf7k^U_
zfv4@=@T_9Cl}*C9(pXvRC8mfYP_6jHM<4%H^7k&%$ZN17WK9t-$GNwF0`vpoRtoR<
zO2NAG0pO{^LhBUcNB4sqOlN}78e8@;LOXe~;wbS_MVp>HPu5+DU70iS7&yt-(fQdH
zW~@e@-W%F&n$KLRlnImPv<VLB8FN^LIv-wmAs985S~t+|!2BwtByntAJSLae>+BFu
z>Kp7CB6n$Zv$2Yi2ED7)TAFL>!IQW4{9p_<f+;CdMu4C6JIVE0U@}N*kP@Uw;-M*k
zo$j4hWb(cAbXKEpVXdVJ13DeAcRFd7yQpAk`jXj<O5y!A+YM762HVP85tdZLNCltL
zEC&OdpXaGGdy$o_73<JP;#wJ)jqU}pf_lYCV65`wmG(^=(i)9djH%=obNFH7R2)0j
z1gz%Q(s>Pt5iV;{XZCLB0L4eU<*@9<kK!D}i$&Pw%~(e@9S+fsN|=y{<t~*synJZJ
zD0v1LAViX#M+)Oi&29XZq`zwEDGr>f@S?13eTUaC#p80#o5FTQfvdU^u?IO@ip51y
zqbfadp$<~neNfW`F<dy#c|riuDMRedEeDf6N;aV%nCp*-6xFz?r0qEUCMdW{(%@=O
zn2-?-+H2u&PD`?*;;i5_!S?f<OTr-{+3!t=#vJCV=#L-DtiuZV(?)5)3KlLIE_x&;
z3!%M^L%wd&ibuC7Z+_Z~F-+5tc!=o)n^T;Vx;;60GP}}Y+VW0B<^PqIV4dx&f6sxM
z#KL2sUwo(ygsNNGrmfvw-_TE!_YRzjVx*Gp>+QG|3vjE;?_Iub)`mBVgKP309b*b+
z10G$hC-!bP3O8XJ$2kl*A%#?@^mtlZZ|eC#wo-ahOT47xw}8P$@`&ko-j0nwI!Z4B
zjuA&Oc;^UryB1>ukmJlCJM1(|61ZMLl6Z_^aC(-GIF_{7KBbe6uELNAHM5HgB$Uf%
zrZqU8{J4bM2$<j$_R6mz>$Ku$wTq32Qp@*a8KxvXcTsNje;4I>OsO6>I!R_3cB5>=
zEDM=K8-X=1ngUWGdW0KIhFb!I0@GhEZH<7O63sw6wo1-343I42P+GIW>fUrDVrwg?
zEQIGW)7}m4z~Xhy6+FFFiI#1x5SiY^@bs=Vww3<fL~v?Yb{<4bP(n<<A!}aO3PzRK
zG!Ah0*MozKe)y|WX<b;oXXaW*t4eNZ9Hxr@MeKHO>lWgz7Wi&~yECISmY4kq0_#Ph
zz##wrFB;8h@8ci)cBTUK!hft|s-2`}rw$q=o%UYl2}?D-c<dJ}Rn<){?4ISe)dH48
zFWR2AKUxZ(r)k`D$wy$@LCKlTDEgs@usq_SencvY9s*?&KZ?XttrT)TJcxxWh&@Gj
z15YfVHK$a;D1d}b`mwUU?S*cf<1_2}o9d;g_a;FU=t)*k2@sF?;%|$W5gJ~6*WV%j
zvsUXY87q3$Oz-%xrhm^S8uD=Rv((8LSTg1z?pdJmnj$BAM3&TjBS$-TKg%ra3<CYS
zO{Tj?Ig6fE{js=E10Chx$jm1uX-PiAJ20v8nRnWOPGk}wgMGb@|5|65oHfKvH69iz
zo^>2m!`PIXA8~zx&1Nh$We>-~aY}a^t>G*($D<ZgZN__XOX*t02fsqBio}O1L+x`~
z$1;;kMO3-`FPCq49)z(u*o&Nk3lZG@A&Lbf)U+Q5`z1c1R?E3=nI+7W7u#7xT$PW=
z67(_(r6W<qY0=yEc;*10=JDU^kT-_3)8!=ls5FzO1GB4!iwJ}N%x)&$w1CwHL``G5
z_Vl0WXTAi%;TJBg&foQym^AtXzzH0NpGp~T_H^0^D768m)(BCRYJi)-bAOQW_Tr`h
zD|-1&(*;k6fXy{3>WDm)GgiR^AUz?T3~nt?P=@l`{B`o4Kdp`Hsvn#k#_`@*U)2mJ
zj}zO;(LFtUDe<1b3qWtvsy<*Z-0YQ;D}hsT!IpYUgb?T}=2gIeIm!BH98*!Z!;+=l
z*~=ajaM-kMb7t(rg5lI%06LFvB8W`r@A8MH&+UN-ST-9EK#%ed+z58z@>~9s;E<pU
zQ5B86&8qo;E|P0Sv}0{Q|GZvs`TuDZ#ko+w#Mq1HkbXy*;EBwLbB<cPh<hJG!Xc@4
zD09o#YEC~Yp^=h<<V&fHiu;{CRP+_)MqBdgWDE3eQjZ~4L?JasOjFtJA)AqwS+#(q
z8;UT8Hv?ubm6Fc-$QT(r!vOuDWu!FQKn|Ic#kV<PIcKJ!)ltDfP(w;QLR&Xej%_Z{
zTvJxA-}1QB^F33#!mFur98kfJmnA|dIR)91k5Y0FNvriRcBa+FHkMYJ7I3tEw!=YN
zvzCG;o2%9qD1OsgJ)eE8<Y3<~&MRq?xOcrrJCvivI?gw=4=G+?n#_^+<nK}Me|YhS
zD*FgQ?uH44=X5vDJ+MY0c?3ieok*VuM*1!@50=~Po||yw#S}%bj%u*q3l$1=zhvBm
z$*so>HUm7(bg9GX>qqTv+AL{7|KcmUhGJ^5@u_~Dnjn3^YT3fMNW}8J0>+o{^nQ2@
zC-2k-;b2iT;u)-!qVE>Lqc`+xtaoAk?!7g=jmpp!YqHRb)Thzbz6LeZH2Jh)%@T-;
z?Sz?(#$;MWK_NXSVxuW8qvvG02#_B|^&k^H4iPGiC5TQr2cA)&GeT<{jZ4@IGaS#1
z-i(<<<fLQ_Pd=nllP1_$r%&a2LVHz_?Hgf1V!oL}X{9-mlSWrV`SVhHhC1J6>uGNw
z=E5~}Ommj3d_)O7%{;;g1;xS5M?hwx)D_)2wJ1a)G%|w7VESwNm`a4_Aq55^UhnxG
zlJijx#YZnaEs3N^kAK^4pAb{?F!6P1t>y~IO^T4Jxtg?y2skq?7=HX?-2pxrsi(1s
zm`YsG!)cA4K^-<mrt>2<(HFH`;}IuO$_Yi>epzr~96HCxF|+I>sHg3ND%D-kp;SL4
z9))V1g5c;VK&j@D9jbe#@A57>!a*wfft*pDWjzw*GiHywvAomF1|B6Qa4?O?;AaSP
zz}s$(Vzb927pIC0o{-YO;@F$ddqk0bu-R){LRZg6zov}G@m%8Fo>=Md&`}bb7h-E1
zR?oO!GOP7^<(Nt(o;h2}vl&x8JPazxr$_avbCNubGX_H?R!JX<o07trx=sp3F3}!`
zwEg4cTEjTDrN+|GnL7(xx~ow!H~Tq8a&HDr7ZjhQ0h%bQJ}~B%!^sCKUX?OKSuu&6
zm*Y&s2xC^4v$Z?>8|m-@_~%?xkPYFZW&8DtVzPyN_iA1Iq!t`i(zUis(LWc%)FIdr
zrLsL7;enxo*Kw9A4m`BF64?)1q0ulBPSjG_8n{qcbTp2OZA_4s<I)Rszj{(og7E@*
zP&1Pcr|KXlR8jkmcOyLe?6nYtwLTJ%Kt`np4x<)}ensHrqg**i4mQ7Ze;o1XKI-UB
zILdSf5G1pnjv;0Hn&zQJgyl}IW-s1@bZ6Rr!G_zHva=p0sG6$Yk&m9`uVh?VqX>9(
zKrhYOU<>J-7vt~%qLL@}LM$%suOKPDl&dXd!G(D?U?~0QD2=8J@08o%<vIbtc<Kv!
zmIP?2e5zXVKh%C}u6fG?H7!kX#yer_0~XB*KA>v9Z~s(OeS?+5z#>kNhVgoP&x4oY
zj!!1ZKg&UCLUQD9wyTR^S}8<7E4~$hbze)ci|Dl0<!GEz0%Zn-OwRgA1cL8$&*H!#
z;2T>Q8JyPR5#SmPn-Jo(5n#8w?+FvpKMdMDe@hFwgtG#{>q)S42sjjMBviXzD@KFl
z!-L^5P}D@b6-zG^j*imty-zj(t`joyg`3r&mTP^q9ir}GhJ53EQ2If_kdJ6`StKZ7
zFmsvW+<F=15rcU-`3!jW(lj>W=XZ9}>%W5B@y&;4qp#^#^(;4mqh7%`<Cb?1QwHI{
zjtX8P5hYBkrw9@>9<+4v<$o>}5cEwa%8Y#KmZuD`8HeWC|HD=M8s$;c&O~Rr@&q~E
zHPVOLT3XlfVZbT4bX4&+YbB8nHIb#zMfcX1JiTH1Re6|OgJ^PCW9-i;@$mAk7j+9&
zOg*4w9PpgArU7_^2=!2(vrWi^?jE-Dg8U^DB<&nEm^3HAl(b1Wjs>PvXHn7GWL}rt
zLozZ;F0?Fga~49pszfjnDvD_yMLE;)KWwM}rKBTj7=EXaXjf9}Tea>kI|wZOLEtl&
zaY3|)WNi?<&rQQAK7~rjkNu|B?Bee)eqjlL!jJMHK8>Bx1tA}NL1}Lfetq&`$9>3a
zVIy208G+!OCSu7+=Ve*XNr)d=8f*(wAIvpYSw^_GhIt`zROExy8IKy@6gr~s(g3mY
zeZEAL8#^NlxUkFX{%s!|?Z&EzTViXl=R*IxeB;U<T^uBKVK$P<s7F(S>P7d3BHE_-
zIIWf4eDqXql7AKZnlg9S=bGyp=DI8n1bmT2auk+Uh@dQdAE|PilXsJY+LS4SVwMFf
zovf+}SVkx}-GjkNS`aJhiO-;D(pRt?pr|8x<R9SM(j}RRb9`2?Btp%?2h%8?fRj-J
z!`v@jPxBz+AX7E$B<$0i=Ueb;G(&kYz}IZb&pi|1q67Gxv~P`F=*n78rbIXLj3(kL
zq=pWZjfmnYd~*4^{+_+B5ezxQ8KRixXn@+|QC<9`)S1};?WR{q_pSvhKk?ji9-cg<
z#bv%-o6pY5eoRw;@<XXD?Aj71kb2ok6HUmy65M}K@=hcqC{tldLrE2Du5aBOXQ!sD
zWQ_cZh!#Ydt4&v^peJ$`j=%ho2Hy-op=TzDfnG#qdtHal{tM{<W~CXzGyACz$nh;`
zLp)GrOQoIWe#PSs5RmzMf6z}+fjDj2qtQ45(ID_DV*KL0tdlF+uLZ$zj@IhyfEF;&
z<%^8Ll@wN$&8_J;lzTYIAW|_~ya@Lj%ha^+wqJ^BC-dqE;0bT6-)jbgo#VoxKbXg>
zq*g|++a#!S;i-a~ItqC<6|Am|X|GEJu%KROxN`N>FdFrex8M?yU~g-)V@eTiRS9UE
zJ*F}Evx6|$bzoe4;8_rqLqGB?u6;%i!NK%K@I)uIPmGgr(yXoe;ShqtS11Ibl@kV`
zH6fV!ocsI~tVUJtl6CkWKQ4j#$u9??peP$Q*HEqDN}uva(IK9MZ3~ChLDuQ!We9MM
zZw7<Kb5a4JOzK6??+4gb*GOToxRGOpu?2mQJP{s7C5s6yHgxd=k}f?IiF2G#rpL24
z^)u)Aq369uMeu@%@hWc(#~GJA!BgW`!lDPtGpvrcEmc_)3evY_I*&3b{H~{-(QgF8
zaoeO$DBdhyg4i_F-VA--cPu5)?nou)b0?uGgbqsm6w8Q}k_<z2tguqiN5lM(ci&Wg
z_0hAQKlZG)U$P9t3zdlqR`f8uVstSTGZ1GpO-M|i=${*rD^0yo>TsNGkn_?`+Y(J8
zx(^94Aub!1id#B`6_d{RxytW8Ry_nsP7O!aAeBAiU{eN5z8;@>3A6N5^?p!XA>0_s
zmgjO`=7Q(D&<ziKv6X8U0L9ToS@iQk7HC+ZIVz$+Uq`nOJxl0+w6Wk&vb@nLBwC(%
zTFVzkOTm_^DrYPTrhb@ydUp$^uTCxS&%vs8-5^f4hXzsOg(TjpJ0d|e-7k?J%<^dN
z3cg1D2NMFN3DwFBh`rvSDsrm3OLSxL0-qvvFspqq2H=;|S^`yz(Hv!o2^u~oOR4N?
zCI}RD9oejr<2?#6^u0L|Xr8Egi1^9lD7iIx=OlOW`XHdK9JQ&v1N2WOTPD#n@R6<R
z&=iFBH`bdu5<^#(I#`;El;l0gg&_M-(`r=tqdAZfCBnfucy>(>T%_NV*o&2lO^t3-
zw45k3Xc}Uh(O5zY84mqkl%^nBX%=dXZB}C}{@~&d`qt-I$fXklj5up?ZiJ_mrpO``
zC78oOJva(@16;z9FPeo+woJH7I~Olf;L$2SZ+5f7<Hg@G5deyHxTt@@oV^j%W;FT9
z<JjXg2S{!5;s;XMygwC6BBK?BQp4iV$u(9eDc6RaxLgT%Y$^>nKRPjul?ou^#3j@a
zi@SpB$8Rxo+{dz@4I0{JZD2ircU>!lbi**vPq;cngcY!{J{wuxM!NsOSS7G`t0s`7
zh<^1cFAi#d36LWI<HwDGD_iiag?5(p(;6OTK&upM36PGTqK;x|5vo^a(Xt*r`cw)y
zVe_c7(vLp7^Dz&e->NNrXK<^Cwn)qL(cUgiR*v@Lp3@8n;$Ml~vzqF!KmyjdH-_nZ
zkqpCem1PZPN`pjR8O<3Js~29Vnnw~&Q*%DjN5wF5Cz3Mfndn%A5pDTko!8S~VH}hP
zjzgY8#(hGl2jJ9{;*J88isT0H`(|VM-$afvLp<J2+4Lh!)Qo2@(#+P+Al|wa`($vE
z+Rw&-frJ<)Ej_1qH2Huuu8wyH{vnRRQ5J|$Ui60y;{9S;>RxruODD-~P!|;<GLJ%R
zeF`}kpqp4NdgO#3MOoQ{Q0`Sgtdvi|t6cFW7<B}kJ$f){MZLx~0-jbZjmbgMj(=})
zl0<cdyCsWTs;uHtCdx{C=d6}gJt-0T-sE30_w$v&)#FMLGDpvbpzDJ$L8Y^u`PAT<
z!*jGk(*{r>Cq@F0LBtedVh~mhp#zpBDkI(trGm$HI4!aWrOXI<t5UmMb%hy8&ZG1I
zDjTem0*1Dkr03U3Db-SI&`{Y8eGs8z>5yDy!vzX%ledUaMW$bV&l6C(H2kUwPu`N^
z?5Ph58b7wful+LKJjF&R>Z)rCx<sD{eS+)OW+V{5>HVPe26;y{pKI5=QW1&M0CuAA
zR{N+ITse81$mENEDQO*k@nRH?(mUuoiP{(M=nf-@28#r^8tW!vYotBvE*Z3V+KUep
zX*)>WS_Sz%HemJ<oegPO+hNB3BE+GWjCV?T3<OXpCMlI78Atv-Q<2n5w2x4^T*^fA
z{q)m*(W2Xnc`G<wyRuCa6k^Eb)7`YQ7uczK#PzRgNwF=qVv%8<7hlzIN2Sd(f6sz|
zO(pKvY)zux!cvu%wfughR9%m|WWRNf&r?x3p1c?58P}~Qwh?=^Hcg47p(K~mh&g{#
z@a*FY5{-tmdy3@`kckF&cNoXkdb8#ATT{kTa{r8qj_3wnN{>fh0$LWv)$=BA7XxPU
z)+i5=K3$(P5e{0Yix@oXN$Gq*ZRUuuL{!kyj1~<m^-v_%FZC}qvl33OFhCp`B@-=q
zLfgl%!KeTDjW;w<sEK0ir<2kch<8GldvLZH0LcmqMDQlUnZ9@#e}WcWMzjLt*k(EP
zVd)M})oJo*M796;NtjB~%1skl1{F}bN$1U~qBOpi>9Qk14rIPZ&hviz$+tCPh3GCm
z(1a;imhzEAo>D|WHKrnr0Q=)+_Y@jqsodd_zIol$5L`W>J@SQ3D<o3b3S~kp+PdT6
z5_37OQB0JLgOGZSrZ~7Iq3%UlYmW~CxPI&3f3Jx@`ut}#Qkq~t^lRnC(;1^X#ri=r
zdFGCi-jrS38<l(qDkx~mp;5(&|6tYnMufo>D4FsGn&fNVg)ov*X{%Md#3tk+=1d@c
z=C9HfPCHBEZ-fJL5JaO8#xJI20!U*>eG<Kj(hb4z_yYfulkImVGTps&=4oP+zl2_T
z8!!e^Aw`MZ@u4Yz@=U<Cxc)NzOV>tC^juH^x&5$tRqbxdU|tTq(wptD3>pu6X}&>K
z<fXWiGexST*F>#MLYTp(>e2!i`WN0_3@k`hyA@O?FX|UP+Rg3wTT|~@ttaj6TJVG3
zLEWut>P3gI0^z{}wmZlo6wLhJ#16gglMlU8)D}6GlReXb{`igm?!hS&DvUY-+Vvgh
zl}Z8p#=mnYIwKQ*K+w9s1GZGOrtqBCglLysQd$$`j>YN12_Ll}#sQP-RCviJ^`C~<
z=@l%{;P46w3lf-K<SFaG#gwL>(e=>_vp|8n;;;okoq11h>DZYvl!wOsa=s?V79mqo
zhuSsrAL>bn&i=HzihNyGmyok}J0Tv36QyZOi=6Hzm3aN=;+11FhzfjP7Fg#_VAnNc
z91n?|d4SUJC+RNOUun$%B{-;VV5&%Byt4yCEqH)z73q7$g28=0RoUYx*&b9z<<+>5
zO$MF$yg~PpFrgs|!4Mec?kKz)9C9H&Rjt}ypbil^?R{J)5>sD&R=3af!qrd;k)cOc
zJ%192arH0B6RFX`7o3?j!0<DyM-Jc5f>&tFlmm=tTXkWwYCgjLwIocdZqe6rmynv7
zjJ)0s(ZvG&@f$xj%hD3`AdhFEcr^A`UNP$+q3Hgc6O`{tWteKo>L$88&IXNc3=A%0
z*@JLDTdH*irQXNyr6w34u)<<i1chh>^4epOh!4%Iy_4aYIkoNdXkc`*Om)l3(yDH`
zk9w%UN?{_iAC3Wv23bTPvhqpT#ifQHD(G>Z?oq=+7&^n4I5Wk2m*2VkK|GY|mMJ2O
zZF&V0OEYm+-(zEKiD&wcV=#=nNmP0cruNKHjnxwq7!_c}Xg}BpKS4T>VSAa0S%>Ba
zVfREoAr(0&)uCQ}^(Bo?sL?bX8VBh}rISRzo528O5{<JbmpHVSF_Wx@gEu$|AF$Mg
z_Ag!gTpo+`JCUcx>(7B0M+4n+yu8b0k5IAScL;YYFJC@D1v5#+^W}S!s6ScU2fjo*
zMy38lbq@4U6F7_!sru&n%P(H{O0Eu0f^z5wnmauF%5}31HR&vMIa2-II@QrrLBo5U
zNM=pv;=PLd#aF7>Ty7P4<~>&0B<s`t-Uzp@KDEZ1<gEDV0=VZ9hcjnuCvTe>q+t)W
zRQWIbnMn6oX4J;a(S(~KztRs>#C20MCZjYG$C=-9FR%*-^aw5eTs9u)iXq*9MS-50
zzv&CQ6vYJAn^A<@Poz*u(HeLXjq0S6)WILrVQ`>F%XiIi9C3`g*~2hpHPfek@!u|A
z(`6j}Y(z^+svs-C>cx!kB}$%OG~f))yVd_BcL!$Be%?ZRqsh<d?BH`DF!veo!5RpS
z@8!?>f%u?mWDpxKe%uVl=4cd)6gwhcPa@bbECpEb{kr7rFXhI^x_pZNL_x}imx%h)
zQ79COZ4Rh+=kc~fFl*}k&^ibMe)2Ycbkl&W5`p9Wm$sN`E#n1F&S*GUr0hkr^9)rQ
z^+?7>Ub^QO5N!?UGoQKns-K6G3Yb-ng*snImbylYe1uP0k0N~r?(uDi$^-nm!6=+v
z>AJ(=g|&gj5h6(e*wfxt>F-8vZTj#PBk?A?^_(X@s+r`o3VoV3EgkBy4$oiLg=*+t
z+G!I=m7p6Rq;qqpOCAourXaXmLo1Nt=P#FRUt&B}6>RXL7$<8+67$(nEE+|f=5&v0
z^i=dcwqinvv+<ZYYF#xi<O0`(*a$U<wL<=`7lW8+Y6f2(CuH7d_HQ+dwZ^*Xas87X
z@7!=2YztSh_Stc_Ky%S__f!7kd5b6@Hz+cw^YyvlIs$<SokqD3A~H8fd!P2m`JUC5
zmy{@-K<F9|J(GPcrw-^4|G_j}#veVkf^f&I>~TcV8KM;WeNgpzZnaqTcG!cHw=%Io
zY{(I!g@Q1FIUw3JOsPLARUdg6@6bhnDq^xicTe65l};1NA+AuWp`UUu96TrknjONJ
zpm{wjF5kdfdec)mB<wTHV?}3^B#{XLD(XtPb8s#K!fGk+H`moJTO^)uYGlCvw)O9=
zTeqH0&DW`4ffhu2*FculAuf8-#u=+*K@#gPzppiau)eg>^je_GdwBCW0*lT?P$0_{
zH5NQT<epqXd2->Dv~6ck3~6GX^<AYw&>$!okxVB4q6$<TAYpCHVmkT>`Y1p{C-wqf
zv-Bsfta*?$^c9PcSD=|MA}=%zZ27A67)$}fdA#5{A!y2HSOf^6hhaGqUIkX16O27e
zwAH(dy~S|QD}pO)OLp7{Bo+Ib(51D0rroMqLI-mQBEi}oMzy+6-!Ks_-YIcl?2J$b
ziO=fkz0tY;Io^yN&={s40WTq^cqX|s4Pw<bt4WQw<bl(jw;Nv)XMFw{k4r@g2BX1I
zdaIR4919U$1uHplgkZ6G>?GQ|>EA6>Xn!e7rT7da;A+4c`xWXDhl)L7s0K7_%*;yO
zbC(=Cx%7BThF)WAk!|_oYy0nU#w0b3?m;pkxMZ&KAUM_lu=ybA3`UdxrPng`bZRY)
zh;w_v8>K6SZ;j}6YE{i>c1nb+s6C~Th8`P**>QP`hTGvoXD*wy_$o<y7LhZVmP8ps
zbCF84>#)(iyJ^I<N=#UH9jJh`K5#I##~L_%k%Cx7$$NPGJ$8gX`8>e{>>M*a3BTqZ
zoUkd&DShjoalDu^Aa+Ml5i1IrUV%@@pOeW+Vc{k>{!>Lou5?G~qxlH)S-No+FCh(x
zGXCj5r*08m=-}4l+@H<FVal_Rr!+W$1InFi!h6UQuScs#!^ULqfN2$tyW6KWOkAwq
zF}jHZXDj0*Zr22gC)}~3CPAuFDFGeh`!&TNg7_SW8XuK+bz#&{P2QkL8M!&uG3Gfc
zz-}n_HTa2MGhS0<_ctiA{Ju#dVjxpddD1%^&C_x?rY7;fcyE6ay&5R_<+m!Tf#L3F
zhnkLqUCYNKiI8deLXShA(&s2Q1V`!f5~7XjUg7WiqAYS`jYp-_fKu1dw52=+gz`%%
z!+}B+>_QF_;R`bf{@4n(=Jc8dn;LQhq)$-cHOS}(ud>xju$FDnOkv#E$#9_jiY5ZE
z$Z!fL98W92O>4{fJD@+!e6LLsTQbG{2x}y}Pu@ydfr&vSZdzRFDKJafWUK2M7WZ@Z
zA`LyS848_e0Xp~e)-R{%a|Z^!!MZ?OVh8rCrFAVr>eTg3orX$w;ozCQJJ-y80(8);
zN4*qCQ&gSLzG7PA9>N6_f_ST&(VDWg$C=jUXjm;xp!$FXBBudS^C5{bgG8)51Q)52
zy1WiRhOj}e@(~88=dbdEBRh+vQz&6uMbx|ichcnoz9|Ce%b?c$*d<cY3PxG5UoXSa
zAn?!tpEJ2YXQwAT!a$kobVs|`LW#g-rzn|2&7$AT-GXoE@=Y);(LvfTZGh*UysPI;
z1LGpMV#7RI!^WRJc~{iA{L5mV!AN&&u?_MAygVD@Vb*q8Iw^`T`x<iO`J;GN@?X1@
z<K|Ix5-&z3{d%K_(*VeWsi~r}NwG~-ykvx*mVwLxB(AR7i~m}aty}x)!HkZrvBg!e
z=y#e09MISC44~EM+Kb=Cp?93&qKZRhbhEpmgLihNZ|in?REL*tyB%pTs_QO2P6J-R
zOuqs&oWTYKtB3ZhGij6pSrrVYC79tjy&U$^D2Zf}nCYe8JU^8`i}p@U!SFKnt3l-E
zXrG^lW2;Ry8b$teryQ9z{cvs=N1c2m`VHp_P$w@vMHa1+GS4Wg8TMQ_B_ihbTdh7V
zqzaZ0t7|-}9etQX97XK*F1WNBsrvF%Tx~1;AS55LW3aM*E&BerLPObpR+UH!BG$Z_
z(biE_Qei}4*L@kIk+f}xLJaE^^3gn1RGcey-|3r{yI8v4X5K4F9i9b|LJ{lQ<d=?m
z$)w==lR!#u?nN)AEs{(R%d7S#|Ev|@FiHOc=TFp5^s-?P<lZ`LQwh|p97QYq15O4Y
z*=A<>>vSu-mE+8!V~P?iiI-uB>b<H>uVvx=H7j8TTgvif{jOgT4zBmbcwc`?q}k|x
zNiTsp@Y?vzq==Ph4JuxdP^7jLE4IObl!uU5ixa@|P9*~HB8TUC`ECMFLYW)vX!MHg
zAGd1Slm>m{=FM9gAlhS9>k(szT>*(Ty5{U1kEu=gPE)8qThw6{$fe_}W*A+{-{?nh
z@dr)yb7AzVUA%}64l8UrKLJ!mdVSPH%HNVb4}Zm(u{q3|qohYNi&5y3U-?Rqo3lU6
zWlAk!j%m!-A3ZIKwy3z;j<PABCPU1t-c)KC@+Yg>p04C1vWLHYlro>Xd^l{qc>n&d
zUOdI{t1nHk$<RC-$4eaWQ7eTI+_&iPG;L-=`AD=3Q%ALh%ymRNOH!58mqE?iZg&Wn
za#Z@ot0^u)^c;p8k^I+`<t?{h09nr~d`GLH86h`Li#HnsKk<D6M@Z=mB9XA>K}kND
z=1fDO@jS{2p~;8rxd+9<7K)QwidHND?(0<KiglOw6B3ni6n;wLFNsT3jpcyl6D`pY
zBweQEU%vz@+9U!zILPD(>C`Yp?SwdDY?`PXS7rSJ#1u9oh*ZyKZZJm2jxWatX||SP
zRfHowf2v|<ieDIC0{(l-9QVD$P`1~nj{MuI>y&om(UTw75=&&<)()#eU>&Cs%-F%N
zVcuxVy?o=7H~n~JzSQIyW$BjUwrI9t^nU8uKeHm<?2{imh~m>Cxzh5_{fvS-^&g|I
z&LdA=?Kxpq|FtEbVEK#XE9qx(m}e1Xy$S$bNqC9vsaH`uqGUcbGIt-w4QG%LIAtrS
zi594c^mCwN|Kj3z(q_z(UoQUd{$8r0U=wR<LxFi{+wRxeio0-4awou!4KEb%rP%m;
z>8o0x8u(6aozvt**^2Px-RPUsA}ZOAP~RmmQ$Hh%La{*_)p0*6jad#!lqm6}@EnK>
zqtg?q3^m9|kbxW#@xIm)NKpu=_I^z+QAGH$e(Rw$nUU8<XSLGJQ1N9ChB!j{W|c3Y
zXvyX4*6nd!7GPXzmq$o3J+H?CTrpi`&?QP<3V}|9h(RvDAi;a%6)(bbUl$|9e1vn$
zB#-rWE*reBYJ9AzM*rlaQlkkvZ9n;e=b=s}S-ekGg=kQ!J;>Ox{8C$puHV2NEr!}h
zkzt9-5xSie8tPJkOQ?;cYB8R5#Sue=o4g(F(3YIW>bzrtv@>~^m_vc8hvo%nK4^)z
zMI;dyO#@VO<H~Q;fS*<zL-=qIrW0dtkhylHMe`MlX-@p3XSS=4ZWb{h^5C}0!k06(
z$_r$DLk@k-9D5dvP@h7r|G=wok}+7?g88e@-Kaa18c1dS*PnYnE$hGH{-BU~hOm2@
zk{AD4T@a^&R8$n8+!?v=nFGG#JVy4$gsc3d8DQG}C?b2k7RuQ}mVPPMaQIMyA?lJu
zcAr#gFNc8F*fJ48WL=lOWHtfLS(7+caNg%uhWRoMTngn1qv>U%B8dmET`-!0<<u^J
z>TKqrN{HqA!G%iL_!o4y5Y<4kdRn6<@g)DWf&S``N)Xbo=eBvVhX?>`l{>|;rJ&8_
z>slOX(7{vJUcRLZHoJN9qAdl;a4)%hy`3#4k@Y1|YGNFLgR5gl{nvwWc<sa?{QR|N
zz0MaOk}CgAl?nEJ9qjk?V9mDfd&)?6q+CDGCyY2SG<u}c%=0x1h<87QSiw#vIKim9
zrK;XWW5LAkcG8K~oO-hFi$1ecrdhzE2(E`qLNQU9{><m4NcA@%85wmV5yhe`2cT~$
zLtwh&#mhDnY$CLbnA`?xLDe1EnCYVo?Wk&`$DHVl{UA^`ZuOhvrcbMeOYkGrjxeuD
zG4YuwBcP;zuPu1Xo#r4Thg_2h5~5Cr!)_msWVir8pvHR9U$$KZHQqmPSma0PwZtV#
zi-ye<(=GgyUM#d9=5d(6uwDMM$AsG&WnBb&{>Du&)nRw9I)NdnKy^eYtN@V9q3rf6
ziTF~kOlmZ8co@f$pR{;I5BZbUR500VYWk@ck9T6TFvZZLB<I_a=LWl0Eeq#s>Xmmg
z0<=^UozR&RJxJdVVB4PEPESx)uX}rRmOFc`)Dz%UtBsZ!aHB}rkAD)eju!P<MAm(3
z#rXavUeDYkM(UMfm4nULLR9DVm4Mcxs8r`Td6Qrjs*Y+==xs<sM6!V>GPVR{L<yfF
z8rtDi!8)mNLbXH8gN;K=xgQzX0cRUF8gMdvQyR4*d|rYL#Bl;4Nbz9bl**^*Jv&Jz
z4qO~i68^W@>!q2UZ^z`WA&3RUNx`tf6*!)K2@c3%c6V4uB2=u~I&DJ0_-a6i=MQyx
z7_cPR`JhoA7^n-W2dyOYHu(U$Wuy6}4Q+;~0iSC$NP?nX4$n?;3712VZtot=U`QQy
ze{NvEq!)gvx|oiE*yTYN4Wv>>chp}ziB;~k1QwXTQ$+!3o0yiyK8sjjRjvte@f$Bt
zsp5*M=V#E2E>j5w;!WuYFd;Vz<yJePo8Gwj%F`Oaa1h>#k6(*`F^swr;S4hUSt)Th
zyAIEuEjNDO>+WU+H7#<MTJf%Bh11Pzk>CnHCEP*sLDFgDXgy*)H_ijsP(~`1wvvtY
zITLU|J9osh*6RU@+MX}o?F@tCyR0WG;Yh8cSf>Uvn2>HFnVY5|p;gYK`KqCb*HDSV
zp|`1~q@RC6PJ^CAV4j!j(+sRx<0czIcEp<ilU@Wfskx-cO0OnA1Mh5ZS9J{bw_BFC
zEq0A2)qe7=c=>35@|I;vREa1O%P-%<+h5s&v}W3dd+^|+cM`_91v8=>)@(I;B;{4;
zyIKF0Bt|U7?D%k5bgmCI8QdkDTpH)#aD>M=xriL^3}uCpSgcXJdG6sy`3~I<i%`m5
zw-XMO^!CfuB_s<4M&;PxzFrvAgz837gp!+4%!bF&pkQ=BVSTMX#MRk=$2K<biU>ww
zmsmL}R!N}>gZHGD<_9r=iqct9F&2=IJ_$FP`hY!C?L_Ko3yq-hEHu|Whc7MNH|s44
zQe0p`z&P3HY0E59F3OWT9PeEzYM_%2EUAWh{pH$8r9BR{B|7<O9t#;aq)33?eLM*F
z)6O8g2T0_w3vNF3RLT&}J2s1+N+jwHZ?9$C8)MvW9tg3`iX0_+LlKUuJAw}m?7Sw3
zznT0JOOctL4(+4es9N%L6xkTp*3)p$eys|yU3wPy)hdloft_?JiuE&YtPs^sjDNY=
zGRrrnK65<Gac<!{^BBUvf)BI^@t+i#uL4vobLgF)xuIp7$y(?zN~>T>8%XxhVc4Rj
zrv4f1@5S!Ny$@49&cMU|<*QG}wmYeyG421}QN&r>2Q{S*H7`SYW{g&A*&9@yo9mgy
zp5u_vGtYKvI1V%>%ElF53~$R$#oWmH6Q0bAH_KTWJg~O7=C9@(%dLcC(9m9d^VA~~
z6m2<!FAQFx_vA(}&FS}z{z2)iY8JS$NWhk>F(kO^%5I1dVzAkm;D<PqI%Im~S{zI1
z)3W&<h%YmQ5C)u8p#@O>>)aLiJ=_lcUHy?DOX?Vzft#BO(Vxl_$Dq|9Ky&oA?0_;c
z>4ARHUBj;pXg38YMf`34^EDe%JVN@UlfWzZ<x)<BdgE0_vr67&rJO__Qnv>YSb^Kd
z!_DI7=ZvZz)fw>2BAp+Mq3RWg`g5LalLGIeBH<VPbFZ!^em@Tt%rX^YHGY(kRF*D&
zfD%)|bVMUcRW`h0p~rkKT<M5Nt)@Cq=OW+c;72|4m<GM9VAnWi<e|SY9U`Z?uZ1Cy
zo8y3D&>nspk8>(O(z;<$WUPP0{F>I*QMsh3{8_TY0H<pf_-FEDLFR-t?wOJxLFemS
zR;4G0_|{k=rfr&$<DY?Blylrq!8uqL2=*}aO5R5?S_enyI+i2E^2Z}|!wh>eKZPse
zh|tI=ke)^poEENOeT}a+d2<d7RZ3j46|sjrZ{iFEc6%^+C&Uz0NRn&TyCd*RxdNP^
zTcB1w7s|Lr_#3J>*`{H;^$vRpz8EQ=%8T{!Uxnf#aI7ni6VaS%R)}_3xs%&`uS>ZP
zQI49HRmuzPE0CO?$bR7m4I@CWMLa={)VKC%4CFy_00bVly<fevSk6<{UMZ~_nZu|Y
z?v#lGCsZT6%C3+FM|TI$<$m?Y;9vz#k({?NjOZeDF0beY7<wUT7E67++q9h3vT%qd
z8*={37D5luh!+Q^Qp|I=3B4mifCa0Kic}M;^i>UTV6KhpfO*;iuIWZNtag`bl`6Bj
z>~|<T6Jxve)vx+4yue@c$HsW_Q%$&cyo=+mJPKT3P(LDWrp8V?v5x^%F7B81Rv@0k
z@ALQG6S<!*zwaG`Zv_QhkCCgSRx2WcP+aF@7~Z~G`Ej6-a47-JS1GKBU5QR&3L#1)
zgO+wO0KAMJkr=cI$&w6G5HF8wy#+2x6L1z#ueVyHWh?zdvzDJUtm-noG+$L>nCb_Y
z-_teQgkG}|iHpwu2@@y?<QoZ)bcq8JUWkbY)<=(k|6aUNk&Vd5$IKSzE;A0$!-Dh}
z@P+yQQsUWde7X4u_m-X$)g(M=P(!u<82<K5=EU1NJPTKBa3V3Hpxi>%i~;c29d44~
znVzY6nB1)rI3xV<2KBrG1N+5PqmKb!TD~aVcWPL1c*b2EXY%ny%~RH<#>s?>78G3+
z(cwFF6|y{f9SXJzoF|m|xk#EA;haf#N532SYJf*VHwDj-jxw|t$Zeh`j-7h6tnyG!
z@3l5OBM!)lyKCUo3xPk_8f0?e8UZk+vYGK$+F6S?O8#AO=MXT@O<5W#2Bb&LeRAu;
zcl%L@la-rY_5=4OZ)uz;dshSqy*In*n2913ss3F2vw#&^x%O#n&ry&2JW<;0W(9uA
z*}ul__~_z2t#)sJoEiW!J`M=N0P&QQEM~k(`5{k})C9{}m73zD)DB@I7f1&g_PX=N
zwOe{ngQE3*{VACt`fBm+O0)7J5Fr1&1<1)CkH(W<=_6RBV{*T}-EX&dhr1YMC<dVs
z%-ylbU|gUp^=N>=FPqsKs0FZ_Uj%}rljhn^={EW&kcxx$c`0I;_9EC5l+|Qwx#WW{
z=w@Lp3<VU3LX8aKv#s#hi~q(hjr2|y@;2E%nT$x<Y7CR(DVdK$_Rs=`loBZ()1yu9
z4-Ns(5>z)ee*MSMs7py=smEW)3bTS+Y+)~?miRY_3FKcA090<~+rj<B4<r}BgOr%c
z<Z;wlzWK~EX1nI_JSDV(eNo0jKa~W0ePYjowf5zRE-lzeX^En7HKMm8=Xa<o5?bL?
zMJju`jYIj^i3*5Y#-rj`_&1$+DHPSSr>aYka%R?VdobA3=&ANJi0B`PiQJBAQ#1fB
zqANpX4>v_`zg+=F2Luf*Td(JAACxpQg4^@5xWq!lH>3RDmBxueBFm1}v(cbv)~HYN
z(v|J6<JjF>sAcbIDol*r>xbi!Nqjk#A8frCBrf2ZA%co9Ob%2vUQja{%NB?Kfe@|1
zHItu9;f>O<kDm9AxeXf0tbwjzk8cn>!Eu2d)E?QwpQ;x#mbGGufBMs(_Px<yBxcp^
zs0xR!$4*S)b~sK2aE?|6l*jgSP^Gink3_kOH>)ta3idKPL*O(YOlI&z8opOx6!u1D
zkNR;lr9k2=y1Gg)`P=a*JVq#^Ih^%&I$KaVs~dn&Leyw8WvE@95NYal)aFnqfTOhD
z1uv@=eD4h6fZx~!-)2!l2%64XAtAXAso*W91tGmGgqoTndc~=+FHk{=vJEgiPy9nu
zKvq?R*NoT^Q|+eLKdl~lxq5X0+QGP_q~mGONbHS4-fDlH?RlF_!4|P6FccGWB`m)j
z3MtzUpX7Eu8V0lq7C4;d$nJDck$EXTN_B>4KDY_uKH_coWzz_4NDec5GY+Z=s&ERY
z!4S`jiAV858i3{z#9m3okYKDz9@<+k4+<b(<+apIWQ5-@ie|s0Qv0D%6x%k8msYc-
zbP&xNV{s>r)7{D22a)1Qh3Ys~F~HuxVppEgf1{LXpaTnKcf{c2yr%!;GcU!`sCURO
zyl4fMLV&@9;hc;17`m`j@Rj^(6!#ZzP4g2M1^eDaIG+3*n=am_x)iz>+Aj~X2^*GP
zvhu952{kA6GfrJ=2c{|fgma~rJuTW>rH}{#5Xh#7E`{@2pM(vsHYEWXf(emYds0Lv
zZ*R_ehO~#k-%FaVXBb`Nh=SM<EB;mNpBgd&;1Fpy=O4cw$^yI|1k^0T<_9h+7X94I
zA}={(#pMe-de_%{3xGNVkM*qrfW^3j3FCUwj>-y3h@e#Hs>qD-#wI&9?o`9A<XTW2
z0TR-C|Klegzh**%wevB!dqy8ku_<+a*vSS`y!hvS<_a(^-_zPn2XmKNY7r69FVy-+
z<<zHpXeTB75H<+S0{xJAlw^?>3=!HSA!aJMi5z!Y<_27k=$xU$9A7zQBN?}DqDIu@
z&D|(B0=4wK!4|rkRC(|pM9-jN9J!GSgon8H=?e*v4i+{yzKP&?GzE`^V+=TB@_%DS
zKcn&{y@*QICf%#uw4R?rxzl9t4~DAoVLswossY-{l{SKiC|ntPTO+)caB)DNc&RhY
zzap<Erj<F4H2>n!FO9FTsX$Hh{~vnBoa^jQQ@h{ZO_xEHga;8zit1-q0A=?HKI^7#
zRO^kcmHx?yGm>1VbuCRL%`Y>mZK=<p?O<=3Y(V_a7{4`Bbm`hQI&D*#VzSe?M>9W<
zDlhm(_DKu$&_`@_Uq0<zQBl6j-Q6^8CXcl&8MDUtC|V?W9C~y=0EZX|$bsqyjR;3}
zRHBZ<;HV(Y+4vg+6v^U)nLOU=?PRP>XtLP8kl!Q3EIl>RTb*6>3cartddmn$S50XF
zI?TTFO&9-IN;&Z=gpO~=_J=bhDm#0DDi_F|jq7{xyv8vq7?0kuSOZ7sR&iG+jRUF}
zgfpEE52A{Y)DjJJ@F*qJCm$XsJ{A2#?Pj9(((UH#h8UG?6fI*EDc6IYYI8^~fL4Dw
zS|Y_JLH>GldvqN2Jjkb=cE2t)q4IcLtK~2H)S}##HW-{!O1AvQ4~T@Hp>-miOsO!{
zYPy6nH^cbeE^%u>H55;^Bp5|%#MH#CU71lc%HlUw^iOQB7e=`al>I~pq4w0`xwAl=
zx*WI1Dd=eJf^K$L%4i(+w(#VOvI87iVIUJJmknC2SmnG7)0aYWe|2)B7%9QCRv}}X
zF{#q!z12-LmW(9+pUeoY!=r0YCsAlo6@?O&r1_?nLI_Na1ddu-*2RFFv{?LouhlYP
zSEC+MOuC>?JE4NKLb8x6ohCAs=gGg)Q;tg!fKU<Ml<cWCY{0QhacrfkhH4`1+_QLU
zIie(~3QU&KtTi;uD4J_Q0m6AqaYu_29G6uCB{W;ZSVN_P>Q>aw%5^yT;y!|TlB6M{
zclaFsmEl9pfZ~8+(No+&(A+PXRSiD<dbqh*9Yaz!rJ@CABd1NyN8vPK<jYI?g;Ect
z_yB&Dpl}NwDO@l{uZ->g>6iRa+&{33oC10nvZEAQCw%6IFE)K8l7&3UE24OvS$Nq<
z=s_>n%c;JXZtYeV$pc44<}#D_K77zVh>*}TS=D=~b=QXvXDqzP@vV2WAf-Gw6JuOQ
z`B;`u11^Xc!id9yJ+Dm|SWYkcJkB{%-*eq&X`GW$?L}%mi@FbLbOwh#YMJi@9IjNk
zU)6&GldiX^QK6gmllP-{51a95*9QOwm9DF}Q}75Wa*#-<)1nZzWSHrVxj7$|M`Kl1
z4E-<PuU;6hUA$-Vvc;wa2-;^%-UE~K4iBC$Bl)RU-4rL2cXOQe3{4*ui9&;jNU%dC
zmwqQ{{b0ka7hz2=7C?ZG$COIAe^C~SrlYjcypaPc<;p3JLHNTLOJ5FFFEA{0bx)w=
zWTqcQ^>h6!O5)czt8!nqk$<^5+p4d?!}WUFAM{N|tZu3{ah}!i{T+!yKaiMWt<Z2n
zpUi?PR$Yz?6`JXlf>AKsBCHao$}Fw~n9?+8(IAjs8Bs82@|FyElYZKK;VZ8`6ZQuW
zY1KF_?NxbU!dytYscm0yRhUMn)2@Q}cEsf3Hyq`aMOAscS`(P3?8hq%3W-G!yl2-W
zR#`6K9ey^6D&hH}_pf)eUO@DRbHg%tvziXBViy+l=3v@f7Ue+*8_$>!qEWyfzm`I4
zNI$38yG*`R?bu}69wUC2$$Q&bX#$`XLboIxkf9lo6ZD4k3eb5@>xOF^$yGoTJTq*Q
zkPUQI3ss}F(b&)&XLo^dR{~3@=aDv|1!OX4V01(zT!BHX>o=eMGyT2oW1e>$(~BYY
z$!w|9n@p#A8v>t*jfdmxUl%E<?@3g+uk`W9uW8^olo6$+xb($8)Fv8O|5B(!w#xJ=
z-eX~vQsaCJMOo1S>?JLXdVN*4CGdz)G(VH}QY(%sZJaZyPzEx1)oE>crG}Eb3v2ub
zrE1OOtpn0j%-U4@AfQ`yl2L*8LXeGib{FP!>Y=y#h6hxb+sB~Pco1GnTL_TW*ElN8
zBlD<^)hEO5(u_(~v3mNb43^FmtwImfxM#P8V=lZE(n9nd?5x%@S(~e|ndj5&yl%z$
z6GDWi0gQ>W?6eDp+S*7>&I=*y(cPcC3voS?K%;^SD7p#?RPwoEO(maU<ptk3*xJX_
z?2)#5^vv@Q)VRRF_J*Q?l0FsCpD)IgJTF-mo3#;Z!CMsG9rS032^=RfmGZQd(yR53
z*4qgan8A}>vdfX)zZU^+-n{v&{_RjDLNCMd6p$*v_=wEjkY_9?{VEc7Je>UcAj6?D
zO;Kb?iz@!+fB)z5f0`R20ijE{Q7krj2N$>Ab&kZ!!=XevBZ;ip6ptNtlPWAB+p(^A
zrnc)*jApr^!{-avO+uZ#m#-yH#&Q*Nmnskm<=^k8Oi|T=QIQ1ml)CjZF8*4c%h(OH
z22A%$02l*6pzN?b)d&_vL(aVQr#kqxe~^zoOfC)XYjIk^kyp&&{@}?^KK@RKo|2H^
z2hhU|m#g)emeeT#(E$!o3HL!p=@G<}w<kXpp&GLrzE@jg7=ESDAo8@P7I;mb!3{^s
zsrUExU=aG|@Qf0sder2<r+$nF1yD=s;yC%s>M8`Fm7`d29%<kK`y(LT9s@cm0Z-A6
zA36!7SIJMk;Jll3Q^RE3Ntx0D{P&cO)Mc|P-f{VcA6R-WdqHR6Z8G5#C5pa84I9s9
zblQ%h)%2jRO&locDQ9|8VoQ3eIr@0=KkB)FJE(L%l4sVxZ*rKY^2K3Ccu(&^J&=8^
zjTxdlT%&Mdvt|BVM;vh&n8TY3OW~SS|Dn`hXNTixtqj}&9mHIG?aw`3RTrrufT{TF
zg{b3g$uKhrKLW4S+2j{*fW_)ZpER*t=>no^Zhk>6el%aj`=Wt$W4u#Z)ZX#rZT(Bd
zgbam})?7SusEpo5)5MYoi_KN9i_Fd8^ej+`I2nl^nsQ9Wu@yamjdntsPe8C9WGFxQ
zH*PdiUtLkWx+|y)rUguhFxN=smeu?Oi%2%u#V_em%`+y>wSm3MGxA^;)L$?4gnoNR
z+8<W_knj-xeF56c0g4EuFuu2uj#8Q>i%aakC$?53KDd|+EmDofPuNyn3D0?J5UvGE
zI%B2jJEXz1#R@e~Yc9<OV9>B|*48w+KALO}W357+50r7KpJ{ksTNh^9Y&c|*Tp34R
zSQk&ce9f9vz>|vo(%UUI1zN9ubQm6CEb<;ug<1a+xJrlUOE12Z+YJHdkMK>L8K|>b
z^%SlLY;ne>RYTAhw}D5{Ka=^Ls$OkPO0--Rzf!qh3=cx0Hi~;GwlaELocms9Dy@cL
zDspTO2uEwSop6gi^OP1pgqNb1f8c=rUa3OBCQS(2=~?EivC)aQpjQ6B+{OO7xS~#R
zbB$hVPF$-*y2_!OsP-NZ-W>UAzu~1{8@QfmfoSR_?P*09$5o}XUH-_7BX@Zyn3RGm
zYCK0^*31@(Gqa80=y5M=KW;=U#gAAM)=>jcO|@DH_dZxfuNvlL8Wc#JihyF|q;>il
zyrmKHp!FBWcWhx#5r;NQT1{D=pnA6ealJ{=1psS^duES^nM{Z{5tK_OLAxNL*vlVX
zq*rVkpLwxbApPiT=9Q+Z5Z7oWD4ry;66T03YOy~T>5sU)2T6++8r)4YRZzG~ej=R>
zDj7p&wMSWnlS3a4`GB>P{!FPD(*WmazS20eP$B`jVA=Vg2CoEai^IwH%tL#x13;Tn
zEOz_o|7N6+SLo3wq92aG=)h#6;Vr~6^}G=ttaBbgoLr~|{z(=#XQkhO54HpgiSUr4
zt06?lydEy+C3*EUReu0OC0A_|3;+U2`zHFTiIgur(qkQ7&bLR0$SnPW6IptE3bs*g
z9MI^Wkt*sj%3Vc$MIQmi=OZ(`C)OJ)`c$%^JrQz9Yt;$d%);@dJAh;pSveZu{pg;k
znzc!73i$*lBNiyhT(Lm6!!5=mQHis_D|!<ctHSI7I8-W!=FbUMiv1~tyCg3F<`xdX
z16~=#sSRi1q49vJm~<5376pMqr6o)hVxG%Z8ecM}MiU8oANom)y>PE}B-hO^?A46i
zN1dA>o5Ze2FgP4eex{v7`IjH~h7pTkrlIuN@H8LP;w1F<XkYXU*7Y;!c^L3E5pj}4
zduiA`YuUDlhLi^IxO3Q4A_#_eMS-r;KP^(#A-V=aJMmCSHOp@J!A^U79@zp8N3j5B
zzFNCtnwcT+M^knsxJD4Cnu(xZ7F^|>M0yRZP8dk<=~2BykHwp!Ky&g^RNi91xs$hX
z$m&IsE9=b#zwab+)so+4cYAj8Ac2Zcrr;3|3dl3|;+u6ltG+u1vIxN>zy6{=kcV;{
zdxdXade)0MH^_8Ha&y_ddCM?k9;<M5Rp=qJoV?}i9qmUEKS~vwPj=j{Me{%n^vmx=
zu{K(lSJw@dZ#XACfJST9`b|^kf;5vW7%^pvN3t3^wLp^L`UUS7M=!?pVXBW*`J&br
zS0rR<OTDM2!d^J2lm_rEFEW|gHn<9t|E^S1!u&z^(kS6L1goZVIQbt{D3jc@-_19y
zXh|5lnP?uAm#!B~F(OkaCgHpGO923=*x{5RsAr{Fi*jovpH}DT#ba$$o&GWk#ZhY@
zNcAs?IhlolQ2?vsC-bIOeapB8<udi)qruTQpiH0yBj5y=hP%2mIE}ywdW9-VipYUN
zCe%N9D}@(O?JMO4w@3skA<;N+Vza?f*KTO=6j|}IPyWFlx|vIU{ra;WA{)E2pB^7!
zki*|oar2Zm2b3YoD@a;#MT>x<V$cvewim&-OvKw6pQRQq|1#6lhKkJ-0iG20je|6Y
zsb1)c90YIBVKn;jFtMzNVii8*8D=38KsS5j^^5q(sFa{^YynNsgD=aMq9GHqCNysX
zeEt)<iE@Vtndpz>Ab9mt=9ckh*$w32A~S>IbgHIwQwn8ud!ZE6;3`^ALg}QdmZWJM
z#eOV4xXk;$TXVb7D?(%5pk9S{;A(AzO!P@q^Zhu-Uz04x14P^lfoD>LrUb^_w=dVK
zT0-9jlhVvWC*5&e(!gFQwgL67dBI6oupmo>mk(u*D2Y-*y*H@LCe(!114yT7_!lqv
zRJhGqS+kc~<lkoUd8hVY7<C^45zvib>kPd`&nn%K;8f5%8YMd>y^fg!0}MIYcbr>i
z1uSqFp$`uaDrrCpX4yyz#f{~YFg4EGduH+(jHfi}@F~QW#L7xGMOeux0cQB(cQID*
zMX3=TqCTg!xWF*mC!z*mP^}F7EhoR<M?QfjLtHYvBKYPdM{;kes5G8gVarY3`CebX
zQr*p-f9*0FcAp~vEsK^(4x9Yz<TusnikeNQAl;3`a^{C;D7Z^@0fyPClHNdJ(V)l=
zObc%QuI9K$4XjY&6ffoV-^vX@M+l8bVDpkJck9~CYzK*eCch#4XY`qg=9A(3k2m!9
zrx6IrC6A0u<bJ5>gt`2VbqXAqfq|lR$>=L(u*eIDH6N*vv#-UfP8ZE?ca!Lgxo~OL
zRG(?5V?$NI#vzSp6%{5A71-0$h>@4*Lh2-#!diAj`!8!L64YLKP3no*s+<gVij)+;
z^WMh1hc`#{ea(LI^WykWj}iU7g+tQYgd}rV$kdRA5|k+>dpn!y(xkFtdvs}H9>r{1
z1wxHSJRN1E>fyob=V8A>rWWr_H!)5lv<D4U5Xn-de|C9R!y+LkbgZ+|mV)HO)Jf^Z
z=*hKuSyBF8)stj^aR6Y=c-|Yh;WW0+IJh5Q$N2c%d~JlAr%zbZUFoso-Y<!XI@%Cv
zKh#=MK5rQMl*G`~HuQT`G1QMmB1+*2inNn|Fn|@=4U+2Ey!=rG@*o=%J6A$$dfStq
zke%X*ZjQJ0uyH|x09g}M&@WR`5xD|6if`9sPu$PqsuC>%BaxS^0WJ58iM3EnCH4dW
z$NX})&Gxl0FrH(GgDFord{?NdilFMoWvoxXdx7A4va2$99_x(s7m&7~(i2y{t>A}m
zoJFzHObg`M`|aYNxE%<QM}-vNfH!Zd1wnY@mxK<O7V}u?(h^+Pthm>EKI``sulaQ;
zUV|2iH%~)fn9hl|7&5GXdHFrxA$dt)EoH09gC%ZEFczHpCVvz?I0yv&{ES`4u~AVx
zLkIi5)$4$4bCQMcuC2PJCb4iWrMF8(T1L%SYW3#u0gDKrVQ#oRc`U+}ZWEC|&H1!A
z=l4+jfv!|&#hN1eYZ7_EC_}E@NUH!T8E$Oi9(v;BUFx$PWgJE4n_4k=5Me=(@9lYG
zLaN!4S1026xydb>LS-ZsIs9_ezMGa>8}!dZ&JNE~xfbHHKz>5;iX%()7CGLe6MX3f
z4R2PBp-bP*M=u8sqIBgDS@6RUMDp0wEP)j=xv;iFPAR8=WExKg0~2w+6*2~X7qLK*
z<rGxaQ(hWBVyhdzEL^3bS`7B_hv#llIj?-N3UA=q)x$3TrmojwH|z5>U{-D9rTa~@
zJ}#uVe7U(10U#<q>Ye{TWAD-&>2;lJ{sK=rI-*BCAWa=s$z*Y^3r{ERVBuyYs!9}#
zA`{32kU%06$_r2c4?84MbX(T#$W~C6Ww%r^-4f3Ug(Z=qBwpkQCu`aP%(|!kCH=f>
zziVe5joKjxQ9$PR?Y;Ke>-w&>5UR%E5`77^r;w|;A6a>lh6H7p!Xd^pz6|!mtLkoP
z44lb<ltrvZXa<f10#;io?4zgTY?Q+vBCIoV{-prL(AywAo|C9Z3P(@HG^woA`#<k|
zlwgT!5?`A06Z!_-oYvMxnc)Tp>STcj85dQ`<+%g3zIM_-Pu{1n42d2MllkIY)s1(%
zS)PWgiRVEaHW-0`No}CZggn$-m=&aL!%+g!?I$ZwfKe0?)4N-ymAIT#v+q*$=pr^B
z<W+9O{BWn2Rymz^MLf!nHP+Y<p_8j7-|0Q>|2Mwq1tY)qYNA(Q8pjUBEo-YAi#j?g
z%P6*E1Afu&AQXAQgo&&Ri*t9PztNncG${{v1j7+y^``IXKMU2m7HHhhn(JoZPH@$M
zAoiw2++JY0W*#gJ5U0V!U%URkKf@WQBqz>+x?3)%9xHqF>I1)EVX^5CHwXJ;Vp;KK
zG@<iPrE;<eK$cY)U!4W?USIAS$YC1j;Uw-2`2q!mw0jW#qmXNuO&My~ov7H*OIKgb
zk``abOz|C~onpeE2&&>F@@`)xVei8Z0|b;d;iuo}O&_L-!Rq{!W1u62vFR1+uVZAF
zGXA~Gylmm<4kH+U7OeHeGbW0J4@&4;5@;)3|B~zv6%+kfkMgplK{I%BISI{-jU1z^
zp8uj30x&nZui1K_l}IYW{&CH?-DonCgJ%QzvYy$VzPnElQ?clz4CM*AJw*VuJ?>cZ
zwholV60!Azl<+OX*fFvH%nNDGw4h?M-FCuhmj+B-H;KTNSIFIYl%v32bT7XCx)@9q
zN%}8RKWpWrcJ-_7Nr7K$eBad{^t1>z{99>|bX2Cnf!n!t%WDK2OJ2?YP<VKpQv{m|
zgqPS2qLL#82ot>7T-;`|hHq^&Tr5|eVRKp2O=c{A?4vNu>flB7B3Z3b*yCT%%89th
znP9|yF=78b=J8QplnplbK?X$@5{tSD4?@VaQ?XFxqH5_dwdsk;TD$!$cZX?C5x6QX
zRH-4zT*o>Q4=-$O^1y__D;iEeF-t5H(*hyJHum%NE99gpgVu^k3|1H*!Lw3yxA@gu
z2+#RtN=+<3(vIefJBv#Vogn^n>LZNbvs$Q@9)s7m(g_;|FQFSIs$eIxqLVKpZ18j<
z%t=FzU=~>2i=f3-v9kfEd4pH~Aa&vEbv*sxa{57at!SQ;8eUqvJAE&#^T#@nN_vuk
z^G-&c>94&-FmjrJr@vpE1jt4}NiXB9g97k&?v-6|kARdOc~PWg(9(XYG@_n!(9JG6
z(K#WFN8tb>kHqBnw{#91QI+RVA}_B3Y<_=BonrOKrPXEg+9;n5>Gyk??!lf8=^k}F
zGASyX5wD>qN;RRD6>Tkru=JzwGhh3R_2Mbr8NgRe-_yTw$=}YanfyRrfheo@E6AOn
za2AHvI)S`OR^C!BBNS^?gPS0FsYZiXRU$5zzIuF+zF{Ia*N_mFc!Dz6P+v$(m68t9
z|3tJtj#Qyfi9xz5o(MU!Pt?`h!bw4l;xT!(pSk59D)3ve3<6O8d>q?8+Q%Ceght<(
zgE+ZCh20sgX?~Qkh?iYbQ+uTaL>~ZGYGU#JJv5)TMC9;2-&3G@p|B+f{O)4A%;W-!
zq}I{F_0f@Muv_W`h?51s^#e6Yx7{&Uv+kM8CX`!RacbB9HLT7^F+TklA~PZwhVd2z
z2c!$#g}*P<C0gV%Xh}<s^qB5RPKg-66wB>>>gFx2V%D~09-%i(vA>0@nwKRC8cb6<
zmZuw}Hme+N@t_s8C#$GW$&lh|`u;7HDemKeuO~(eQM`g|n%}&$vEw^4N9-xX$qJP7
zxiI<l=fD=@1SgXUS=vsFj3}RZPE7|0qj0Xn^yFlCbxYR?(&{Bd+w=oO`eXDSVjQML
zLg$7~pN2CLSE7Y{4aS6TmSLQ6D^t7@p4b0J5VGWcx}^mgmtw!0LG$yAWJ#*VtB?S|
z822zB-KO>Fghp+d5g$*Cc&GBG#DT(Q!J_x41$V5@4x=CyHB5eVYCQM=4>^C&(6R^J
z=`Tw2wQ?8h<vgLz2VECexr9u_Fil4Orn92@r)X71iVlZJ+rD`}k-`hLyKS3wl`3H~
z0q_y?vwZM~38gQ$(b~wpBjKtp?L0V)WZ#{Q1s&aKs~eGc=B3mM&XZY*+@C-3ocZ%$
zQrZzRgfkcfKmGT-@mIdNi+xoSQ@~7Tv0w?HcNQv7KdE0_Vh!ZYnqZz13h<M!zjO7W
z7!VSaJ6YNu(HmT3SZll}Lvkrp(Zqd5v5v*5`GSO3F%-$JRoLsl^$Vy62n!S89d$zD
z6+L>l6#UYl?%gs19^O(B(+qD*4-OfJLQUuklgr2wGa-$zs83+0C=WbhuVyj3L#X-y
z<E)c1Z>&Ny)~6>i3s2w2m>}ls3?I?9S|1Pr{-w2sDYne)U?Gci)?C-&MUz6EtgB?z
z&)xTBZz8G#U33s_Kq+Ku<s2YlIg#;Z1^zGvJhdsrN|`zK@4tD*Oe3UD_ss7{d{Ko^
zqu#jEPC+4!Ef`qE-XNO4^dWDNgg0-#mfS;L-VbikU#1!!S3kafTl9GJqikEDYMPR0
z@hTD9bzI#GwQn0zL!)uTS$H(zO6*RM?|B{7^OFo!K}j&uu*xNrCYKp8Dz3WmoE~t&
z|LO1^Jsf}fWBr1yXJncPNadp4Q6`xM#U%FRp2MJUEgVYg1R#w5&fY9Fh48KnyPfB%
zRT67S!+YnjcwCFF!V+cny6L&7(A$E#jLyjHE_Mg3Bc0*bf?UcVRD*~}ZJ|wxYn_X6
zNUBRLIL}OiTI0O{T?Sc}fuPSgzT$wT;*w)N#HuN?1&_zm_ssR<<7asofBpaeKc6oh
z>?&*%-10C|LC-Lo&BQLY4@2Hc_rRl0*0qb(z2OFv;i~9XnM-nbqU%P{4%k=rZkesZ
zN=xOKze%@3j86Z9nsyjlJaYeKzeMO0?G}An#t96lqw~j^5yIFreSGhMctJVa5*5Av
zuGgZNS;Wr}@xrPRc-MMRAuyyL^OqC#2=YHlB+6?(>0~ljcn|+_quSl;l<ilpyG`AU
zKhc~1)IY7%#tz6C4Q$s|glPzf3$)k|dmFb)F*M(M1fPU*#-<?(X&^68?rjMig6c5z
zgU8?nnqKP+9ni~ZR?#ohSvR*|)b*m3U+!@W^SPVutHPO#Z50*Uc_msr+%H^EG9XW!
zRR5U?OTAL<TDDG0N|va%fR0aC`5d`$@^g-yu}H!;<}TC-x&%aZsTX|mKR>UhhJ{4x
zQ9BW+42t+=r|LNA1qseSlZhRsEJT8L`tCSUI^l!-&L?m@bEC33F~~KUx^WcLbf>7n
zwVFk<TPaM|Iwb=!rb(wY*o~K-_FCw?L#4YOM4GTf2cUXs%`R#X_K2jfP|2ZOzWC%*
z`aRH!u<JzoQJ_v_L{yBuou~jpt1UsoeH3+K;mIZnIio9Wib!*)l~xIKM_l7KdKXBU
z_<rFWRp;urD$zHE7P@H~3lA`iaSV(zlWSC2a+?|sA+UWOs(MOCkt-Nh`2wLhku{78
zT{Hs0SFwVtf2_qe3K<E2<u$|i_(b}a*RpnzKOJ}TIM+Zvmode%2x_f<lb!@Mo`<MB
z>s;8!RWwNrdypKknNA0pS`wxj(VEe`Xik_!=$2?_iC|Z6MBvZCMJLl$O+p!y5#TT~
zkRl>S_BY;1fJT)YO?LuvW7%GR*Vht6aeAqXcl6!vUa3mrUTiaMDz6(z8xR+<W{R;Y
zIFXpD1D<?-XOQvn4U}*4dABgMG6d$%jm|m!$o+Tbe|jlRy21NLtgh0{n1lwW;_vmS
zR~ejoV^<I3TGDJ$58~F6Wwbu@5LSPX%7-z%tRRgavrQwVUg`vHe1Sz@UW_ubZVY>|
zIRE{ln<9%)40?R@M^H*M@8Kn8T&{1RW=^7!W?$v>&wDtYCDx>c%1FNY(BdZBqs|7`
zkiZD{kpxxE)(dG;<k9^6q8VKwv@}Y91WHH_L%b?@Rj^U7zjnhPhPS6b_s?Hapg*oc
zbXbFi?-#owJvEZD<jMh+`U{)18%ysNBGfK|Bch9%h@p>l9z@k}<O>+88UMYzI=|@i
zP=6exrVkt+E06F!{Lf#@@wVikmtQznel?-{3NW9Gh65QSrpvZJ)5Vv{3MKS+tsqon
z7N<f-hirRIpDs!5lBb!Bsn^j-fsZomg%IoY*K@^`U2<E!EM}T_)n!K;*ifn><8Y*1
z>ljm`sNy7|HU-38tDVs;VHh$s`nI>ffR@%2g6^KZCM7$)$S?*FE2J6BT==9jLX8CG
zVS#?qmWqM_kQC<AP0Z9>4|f$$S4Yejd;*P=jM}ZPHbtSO&Fyu4gc>q^1RYNl-0zLj
zoop~v8Y5knP|>H~p$bGalnM<<g49VxQuHR2@9598Xep^%KKGd6IHm(c5q`K>)#f3A
zhbZ*B=@$&&_@r;cdnv?7bSpv}I*c%Pk$SSh^;wORQ6g0DCaW~eB1Fv;S4hNXOuSPq
zXZ<ho)Nn*fD4I~UCRsoycfTg>@!`@U9uY^aqU$t@J#RkwjGng;<Mj|$5m7~%NJ!c#
zSi_J+_je;#vN9h8qL?-KnZvK09wF8mL(~F9r|NdSX^9*N8#8rI@B$~%Ozo^@nKwXg
z)_~ZM@e|2_3iAl<58Az^H|U;?Og9gq7Ze_Szo_BqJ12>U1K$u#f!4VCSj)ZJSi7?r
zs)EzjpXB}w!fuE<<B<2$heUh)XaYhEHdJ}1zD0-H(zXVj9;wn#-+Gj_$G2&!<>Ll3
zvzLfHY$#l3b1x%&^pPAt@88}og<m8s>EBw?8EWO3=Yz^k>!eU3+T?{zPT1T8MNaW`
zpHpcXzF*=OU*iA#qKyJi!7QdLi!RNBQKb4cS5{|F>OxQ9CisOij$r9*s|c|`Fr}hw
z!5Q8#-r;#Xi2}R2*W!nhauDHRiTRuXc-jgOtQSj*g5JeBzz%cs-4g7~D|{+n#-zYi
z&Q5i5_LS64`{eQVQgz~qtOW@r9i3YJYa!$t>s*tof5xCvSTg!~V&T+b7$vJ~9zf4R
zG9fhdAb<wxMTBjVRz}JA`YTteuOtvw8ej9}j8tqw!w4%4!Q%N?_D@n6(o86DC3_JC
zMo<Qiu1Bs$AC{L_W9y?5GE}}8Ls=&gO+MgVp9<s9_fii;(>6x}<Enm4STbU<=nzQ>
zvBTmpxMKQ2X1jXo8)RN+a`OBu?*0>B)M`L`pa4V@Fk<P$#HdZjf#j0Fy~T$08LAbb
z6wR}lnTYxnB~<-^QwAaW@YM{V^kk-=YsAR#>AN`{khiO+r9Vsg5UbNJZodAE39wHj
zA?rM_FL9&d1pl9(|G(cz$x(Y``zuYPXI3rVC6Aq)XMHf7pt{VcvGAHnKt4y<Q<|>K
zphElXM=7uJ6-B%&FsA5+pRk65hc)eYtSv6>_^O=86#D00^WI2BQDGpeJ^!EbY<r_J
z2aVW);1g+6VCIxaHa=n|G#VODS_`KXSgZaUM^gk2w<A?7Hk;QsPqGx~nbwEYSA^Mj
zHk#wg#n9nnIfm8L>+Xl%0%2KDBGkU<JSN8x@l?-ef1M8I;M_0_%_m%BMX7OZVMaZ2
zZE+*AVqAbnA-4m^sfRJp))ZJKgO`T*?P+bdMdqNwc-AH{nh!R_X;i-qac%dKsT7&R
zz$K1$Cy1vC-Fy1hb9xiTrk5AjD%<_}u$EC5{t-P?N!Q7zEzjv)$3Q9$QZcsz)2c_`
zTV3$UMs)Xmo%$&e;gqZtK1IN$xuH)t*0=m72Arg{StfSsT1Xkgg2MgAgppK=GHI}N
zBBrceErzr=NhE!UI^t8!v-U9$X!A9Lu|N8eFMNT#)>bkjYGJ%;T)4wlB8em65SUB7
z1VWxxC}Ouy<gIw_nZ&(^`@%g;Od4hY7b@6CL3r=;g+VJ{C7-*NxIy&1khiHB64{7&
zmu5pmC2LmXc84-(ZGF<OeX?7$dGzDwTcZ%cH$Wno1T?qlwrgAvJxGi=TUo0&t>*S>
zVhsj}p8M#J)l}S2*A`PA)2^gL=K<oiVYuL<PQ(FeT^7-b)=|!8+!h!(FaHWDJ=5=?
zo31>Ji=CmX;gw$!u=9_S94Q`Xqu<X;<x+#BMG8@=g>4h);a*jHUfmy*Vu9Y*Z~AG}
z_++{YDaw)P0d&@htoLp$ngU0Y_HAS{XM;*l3UIYv@GvBb%%*c>g?FJ>DL=|<%QPbG
znS9o45cN_g%)-M`i%JDn?)?~Lpw+rDw@FUN<<cV&#@i#NcbtrN!1PQtSrU^lb*^+r
z4zGm!qTQ*)&JP)Dg(fCOjT_(a#GUylZRV$53fMk&ZM3r@E9C<tR_-f9Av~>P_Dj@`
z!wo9_@+5pJVSrP0(2G;C0W%*cy6L+orEskiQ@*zs7yRto#adwWY=#>4dy@#vvM3dz
zqhYJP!?C`3{mnY)^wn3ye}lci@JQ+F*u7E>_C@KY#G(CTDy37fhh$<+jy3^>8nI$Z
zBtmpjx^4mSNb`^8L8rmUq~kjeHq-fN4?3ed$3#)J+Gt9_SBH_<#`I@e=X|qiX3V(f
z4)u^GL#rPWjWnaz#mi|l<S)!~c|>aV_>ms9I|%y;vniPeF_U<0swGL=nDmiZFBpY+
zDdzCXSkfqZHittmwR`IeS_eKZCC7R}clzJ;pVqOPSaL)4e)L*ezdPMs-vts^%ne4*
zdMhBhd<Jl1NEH1+8gFN%ULT^4O!-KF48yO;&8Juft0=R~d#5fiNU!?{4$B|fLAv8q
zM*)nISl(`{hnz3XXIp*Jn0hl&;K$tSVY(gcC5cCcV#ub*=mui`ZY`ShD)z`3`_-yJ
zQ;D(Y_T-=Wxb{1+OUS>*MixHeX`VXTj_v%d(kh`#nXX6+lunlupHDdIG`J6iv<&I7
zx&aa_fC1E9M?BG6EIgVQZTewEaJ>r6Y%JwkLIE6W!Slk?N3=Ao@o40eR~?XLi8sLz
zt$n(HmTIM2Pd#P1K~i#@@U<5vm?K%~$<A8SU)Su$=A9jfNl=CqGs<kPRysMWgEU93
z-YJ)tHUTxBOB#a;Q5joFYuf6W))Wjxjf^Pv<QJcptkaasWVO?ah>0K)=-REbR-cF3
zl?0BVO^?1qWO|%N@80y?yeaS;C=tJv=6WeEmE6th;qh!%3bD356PhFrjHZ9ELIG-b
zizs~WQP{}S56O-MOUZ3LO><K~O!lqKP`RrMh1FZ&7@!d%B|xjE(gzX86uIQJp~$}F
zH?w+<)BivNvuZPj#`(^u8aAB@ghn#*p8J#k^Y>}tM*|sOO8Oj87>EY+OIxGkxb&({
z@UM<K=NaG_5q_>275qx~@q?4rU-aRvwa!^!N%4S~Drqe8Fg3*J5{T=!w1XTYT0MS=
zt5NQhHjx~=NJj7{g(g!F^&8Dx@B-InodFqZl;T@mzmrOZWMMac`kfNuA~#rs7iJrl
zfmh$~*eCH1yBeqcK~)OE)-=a7BJ2}lMzxxyNduo&hN4Rd${%CYHoO?V8Ku{0*6fiz
zuK_R!7+4~`9lV4>#nMHVskC47m24292(znul$UCaB%RN&ERI!)$Efl3UU*y-`c&~t
zQ+{;>_0wNy1rw^9VDw2^So0s40->LpfmgFf37k!>IqVy8+vB@P^HSAEG@YnLJ{iU7
zM==yYXLLbdNxD5}P*t%B8NdYI^;eunJy?qeL}4anS`lq0e8Bj5DYfTZ7F>kQV^c-5
zf}ivWkpn$!VKE~_s~bCFnb+y=XM&!~&C-WXTH%2=!~}6SxlCTdT<Zc;E0$FXG8vd%
zff2itQT7lUje)s&rRv_}A5k@F(O~0gR%9_8VU*L7QNW}M=|?PHu+<uuQP(`Nfz~s9
zSaD8`(k4fw#TkG5>JMHgp^;uE7Zb^EQ=i82TK%mjpV5;(&;R)ws!=U19s7gaEvkdA
zeluP{O`zQLMcnKR*xsAIBd0$ii(u0!M=`HOfcm3~z{l9q8U`nP=pr$nf8ue}QVOEB
zUKd`|@i;<~ir4z6fSFH%vqEcTVj^tb4~kE`g-Kk3%j6!%8IcC-B>N;WW%z96zT=!?
zGMd<oP{+|vzDEBiY^Zq=(uE-DhE)V%H{}9in$dnYqYyGs0LGgPs=ZK83vIp$0~9C^
zLV(=yIJ`5Owge2}xs2wZUxtCpCWtafcqYj~M*0=5Q*m^BRrQOeiOwJlii5p~RMAQ#
zfPOvhqmHag)t^49WGt<IlC_`+7ODb*evX3ik<^y*b8W+U_tUpxK<l6+I!n0m`n!7W
zr$2i1v00?wImaB}iQ20IALY_D=H^Y>bwoK?A$~8L>7~~sQX5hk=FWTddf!HvCLd{V
z1!a!@C!Z$DQ8kVNDy^t|lEqX)X9)bbL<j_IzLc_&#>yx%jv}&E*Aj94SSfg!Gh6L#
zCV_<!7Fs?GQ5`NK%-l)aNyw^|Jr)6Q-q7gNWF%;r0ETYA^mj*9>N1{9GVL(e@u(I*
zo?s3W^-P>z={rF1pL_1#jRxnYy%LJisY{gIS`lS|{~u=B_}LtG5A`2I3Db~^&2!ED
zLPq9c*_+ZZ>q&<}e|{Hm154{LDIz3Xt`FP<{{)jE1N|rp6j^)x@~k8$kaD}F$q+v=
zwp!2S(D0DNdKkaa3d*9A1VTPZAf7>(s%@hL{24XL_u@KboRyo2VF&2NtzJoDsHf=S
zjGLmZh9059Nsh(3`K&|nMa>jSHZa&uj|Usks=B(VLfg1|wBnApKYHWSxAs!#1*m5R
z_+{rNM|g#w_*EFWICz<yrj2<v;Zx5dRnX?0W^zTCel!|T-%xF8j2q-<ZVEM<<yg&9
z-^@IV)QiiQslIp-efyoXtLfEj9S5UCdYaxFhr=|mxcwyJ^~cY?<_}-iL@wi!Yi3K~
ziTW;vJ0}uGiTUJ#XtV82pQ|dYf&uSpI@heOUp4*Yd?bkdT?|H&;Ns(T8n~ui$FXFM
zT$_4CMhc}5gK1wHXsYA6Ux^4)z{Tuxn@o!C+ZUg@VX{QE2Qs|I-P*GeP-hq~zR6i0
zi%H|faZWTC@>}kb+b4Ub!@(EG>q3Q?$w#q}&vUF{XTGpY1gezI2qDef<!B6qrO!=@
zTcbi}?LGu*(gf7jK?wO*Z$6_VY9IL>TcsU17wa>L1|>1iG!?UbX+%&P&SS^Td~R&d
z<6td<rRf}@lal0n1ourr)k%9P8v9lJT*i)h5$6{nt2BbQ2yzI99W%=GW`PRRMV_1f
zJwI{_n>(+jGTJ1PbVl==>wXbIR**RUG-WAy;|Y<PlISV*t0*yG+@j!VCW4T6^4D}x
zd<0yWQ6AAO8n9euG(x_$aRj6XEwc!^3W>;_kN6tBIx3PLD>=sGQr;B8W^|d!ErsDR
z6A>*Im;yOV4`7vLq$6sbvR3avQ?0#L>YkDyK>l=P!ulaUT`aL5eC<E-e_kS)>&eVs
zra-~;R3V#Uxn@($Cv<bZX3OKT(bn9l80^mO0Vr{i!%pB|cd?a1yeKen{q+<|@+H04
zxr<WEP2w(7Svx^6aYQ6;zw1xl#i{f{mQ+{S8Kw<LmXURqSt4o&_&u}B9d%EFgF4+d
zfgL>%Ty}q)8M{+lsj!(@=uws|Vsp~t<@zmOT#k)zge#$|gr={j$AHyhO#j@Ob5%8G
zX%>nh92*ljvU|}w(Ig=GI}8M51eYVkEEdotD|0P7!esGILSO0tt5btLB`%+3;7H(w
z1sx!f)CSHLrDetgy6K~9Nhpe9+9~4im7S*x3sq>Zsnuv=*iaUzuZxUF^0bAdxHz1K
zh(p@<D8bU%Jr9;OFJa7dq-=6~8dEr%x`J9R`XBgc2(Nwc&0Y~bE&9nD4Z6NpeOtnl
zq3RU=S9>qN>>j0Xm#y=N69cc7pzDURWCe19v|RT0e`o!h>WdiWY)-pm?xlYmwf55W
zrx(O=E~&$6BSp@$N|0?ndd%rX&qXM?`=As|D76hDNQ{dDcZe=KMLyQHjgO`4rFqXa
z7TnPF4Wanx?<Qw8d$u|XY+i&#X&O_beo@H~cMHWJfuR*GZ&htYF`MQHe9v@(rNYYc
zp`h=Plw!nfF0L*NAVmhV)g#2?Alsu4a1)Ya$aWP>1Qb2&^psoHc5O*n36XFpaai@U
zg@63UrRa$0e}MX(`NG6a(+Cg(d330q^fJS~gyII4mZ`6E$(cNyn4zf4(YlOrRj=IC
zKN{OFnhHo2oeKFWAM!df%L-_&Osyw|o(h9JT2}s!g`AGf!$V>rP3H*fp5iT{ZhA&+
zAy+cLd3SZeFS+{N_1nq&!e(Zj%JpAef7>5f>+A=%;N_R2Z#Qi_zO5@juQi!A8Q9eU
zmg&LJn^0FiwevvN(nr-8<9cHld&_heL5dNmNfsv%k~ScSSW5JKwQ){#KZ=8^Rm$$u
zR<TRQ>Bg?r)Z!h=@_b8A)pk<E;b0;)FrejEcxiW{;#xGn{`&vS2awRUv07#d{dyH7
z9S8^Ehmu^ZVAIF&GADxsMY^70J?5n7m~TG&e4G$<IHD?1NO^EN8Ey$gjAK*rh|i)}
zjL8TzZZH84I)?;c#$hUTPa-&Y^YzzqM9RJHahG<Dc+}<_Zoi+APBhMzrOfiY76x`l
zpT&QLMf8_D%%w{HMTpQcMI64f<Aur1Dz87BKJ=o(+We^W>yb~9Utg)k853rjSxHhf
zienU^Qt807@?U!S8S6GiZgAAN8r7tj{v*en%qy5sxDNA$CxV7(s%m(l5%QA2#L0%3
z11FPU<Gq8N&AA7%$&cz2I|w#GI%P11jw8hoY@J7ov4rx9N(0R_&=2!FFK#W0Ekb~B
zrnyb@7)^4_K%pr4Mv5##jL}v&516_bvld=Z?R3x($USLH@{KG&b~_@<D52y-A@>K2
z#t&;^0QgJl2a@q=nkOZ0+Dzmx-v;elx-47>-(bs*748d2CfiP&JN;ROxzk0cG;wDD
zg<qtI55q%~+34G!SLK|*bO4V2_vd`RCpdh-2z8&$rRRkQCA%xA6myMcZIaGjCMhYJ
z(D{7?k`xumd54S9U^kn{BSO(yGSZL~5yILluOeqjkj9VL0`*9n?p&u530b~)cmWUc
z$=doTqT42KfBH@YKxs>+R;RigTTw{@2su5)V}jdWw#=4cO6q=x9mRFi?lb<;97ln6
zivlJSf4(qJq>Q}WVrmUi_uPr0a;D#D&GV7^j_mW3F9&R#fKIZAT)kl;saicMD2r2h
zao!Je<n^03&4Dq$#;Y&QiBg1c@$xX>j#BVO&@c=mv_W?H0gDI@sAdY7k4q<~No3py
zd-nRNc&HdEMjL|~&2!m1oc@gJ#DrLsp3=H$0B&ms91;o1?!rQuMhFQ!!;<Gm9YyzJ
zUq=9oa%{!PS3fD0LQ#qFiGCwgy-B&%I@7zDp5q{^Vm;n8B7&-U|AUH=x350XXQ%HH
zku-@gC$cfuhQmK-LwZ<LlK}gX|LNo5eANI1CUK3Mz`mb&ORFJ8CgaBq?t=9OSmn=z
z`9>+VSB9>f#l4O(gZufxR{IF5*Xm(%pR#<b;7~1?$bdoU0$|q87$N-*%%zb>+^Juo
zGIKWMJSc$D2L1jk7IrXvc$T3I@vxZtA=YvnEDQ_~5W2o1!$P;c#>r^$3vL>i7o$Ch
z@l$ifw<#q=nHb&eW)0ddvPb8@Vw-oQ(uf0_G6eeYl*R!iI3qkCDwu{-DWkz~%w`>}
zcfr7FuVBlJ%4f+~8u}KYqkyqbYMk%v(L?<~?WX;>opV(TBs0neMgsjYwEwMJc~<|I
z4)^*Ee}%(1CsJbWVe&u<xa)5a)&1n^L#=0tgwilHey`{pf+~+b)^qoZG(lg^nlDhJ
z60;fnOhF6vQBv(*LP6h@A<L`kkP*5V>fMOa<;HN>x>&<zlVsbR&vpMmJW(!AYeIj8
zC`aK-80IP<3(ArQ*MG17p^gFH8eU!`lKac8w321x%I-41G>NFw(Jd>qWZL;JGNuDj
z=v9cGD%CQlzrY-ie!_<-rz$<wRbmyzT-QG?*j;s^@i){(tAL0H>VlJ-`$x<zDhU10
zurRPS{ZB$-U7*LolvvB(D=}(b1_~*aHi7v<0)PObHVMSI;((4(qCN^djI1s!poaVx
zsTk=4>&H8*JMw1a*y;#wK4IqW2cU(j;%ZG}o?w&KCN}V%lAhiw{6osf|G0Vx>Edz|
z^O^}J)M%Ydf-D%O@`si|5alO)6vS!6af`cygwU{eI9XfUDPby;Yp&C;%|l!5HboT0
zV-|sV-g`FvH)+sMO9nP7dhsc7x&U3>Oerf8z~=OOLJqU|DTqKoAb#6{uyqvb#;0&&
zuRb!dH5^O7+@{Q>^wK7(Tg_{AC@C4I{qBRBHY4!Wx=5e2#`Ltl2Nx-A7KpePW-o1-
zu~aiv&Lat}2U3g^tsclRcEXXqBJ`NgLv<W8y~3Be?f;2SaG%K2l!R!6RPt&L1F!{9
zvXiI{l3{+<Xm9$9(sphGXvIMd3h}Ui|7o7T`f+@c+$v;|e!jyI2wq{5sd={L5|)kp
z`(WiVujU-H$KP*LxRISmI9x;pKAk@L+EdXbLX|FcBE_)w2wH0?HSTx4fcYxEUV~v%
z9IK4L(IR;i*I&!U|D8_XFNeJ<U|gEK6p%gWrOsF$WRR~xf;AuAIf|M^KV#GZl_KWQ
z;F6NIlx;6TX12lrJ5Yo{OB@2i4?{CJm`cm5m~p?#byGu5-w$dm08Px2T4<BhGG*#*
z@|D7Ie-xg_*ZTfb{#-z{yAc}SP^q{nbxNZLe|J@vow~AW_dcdZIc;Jhfz9DZTmXN4
z-rfLqC@BTxR1NiCJ>*(tw0aKfT0Ba@Q|5Q^T`}X!$$swHXY(-r1wSOmjjXj&4Jdoa
zoAe=LqpJ5tt`Ek~?MG3WG^4ja7tj*k#X<Np48o*cyNV|DnM)A4sUV3aXYDgaX6erq
zvzv?JQGZVzD+IV*k<O!NQmwt=vmg4j(ySWzl{q7-cIb)ufL);LA|pGJ;~*BlsKH(U
zVj3^z2R0d3+#zWlk%f&N3GQYY_ePf-ZG%{JDOcJ8`&Aqz`kl$I_3_K>h6YvwWKzkY
zM5TcbauMU!FJtb^7_?3G?F9?>pC}cCD;;3F|FpKf1);Q_^$52^gN)QqP6dzCodBT{
zQ|o;{Hl-u{+Iy|i7Y3S=vM<p^u1%G0Q7RC6>J{IeaJ*(f(f))mZ7bb;T5XFONTBMO
ztOyhk*8l1ZZ0=(`tu7H}CoS)<zpIG7m)V~T&eCq5Z=IPzg{By0**{V*rU7sTFB78{
zG4P`dCp7317639f*w7dhS=ThS30Fy5Q7y>ThdTS4uSNz+c(_0i_}~Cn3Bc=rQFcjt
z&=_UI6A>9IM$a}ljedHFq!L5%F=ol@UZew6ZA4$%0tTqSRHaK0R&qtv+<2~1yU0qP
zKh}JKE}2F<gX4=Rb0gf4A+abY_16VukQ5W6Ajh-wwF(Zn&NF_Tejq<KN=W?Vjpo`Q
zAX~`)KqKf}RTUC8;QHo5sL9t~e?9xb6U-)stAj=UVU3u<Zv4mB{!dZ2S>Of`Z4^4N
zHVHcDPE(CZ_FjEr-q^b^w0&r=`LV!42+uO0Fnm6|oKT!Sg5||@J+F1qro`3XqD&CY
z+J*oJK@yZ;;z`Wb1kHV5BV4KKz7bv;1z3aVZ%uP4zSXAONZZ5Q5aD=(`fa4EOvfW6
z`s~&VUsF;6`Y>|@qob!;%eIcUzfu6+iw~k>M2!n?9=F2XA$}8vx4J@ghC)c<cGQ+}
z+@!bt)YDEhs?{0+srpbrv4(!4!BU<680H%UN+G0cJN-qC*HUCj&c%zn@zi~PtaSkA
zVU(niarOP{x3xF*ctE{GaB(wI#W+ui_k*LdN|5{vniYX&NzG6>aV)=xEBQ*K@_9V7
z`h1Y8(9&pQ9prp(FiB7p=(pMwvzv^*#^glYCkz;WJ8n1sG$9aOIpghinX(!yzT5EM
zk;4&$rEk<HdLehI+I0ua*sSdh{ygHpYPxN@V~}9Khi=eWr5cD<5HsHcduVXd3jpGg
z@MyZHnBqyDs6|Y1|JheXSU4!=xoR~xG;33I0$l0^pVB+n42*E1o0L!Ne31Q;8RD42
z4Cp6o3|c@{8SDP4Vy}~wC+gA?#siuEG@$5kteGoCtZyy`Mwm16B2<#AwQ|`($q#GP
zd=kN{o6qFrH-bN<oiQUKwR1XnKh%2FFt8-z2}Ub|<>%6}J<9dSW^*4c7FraX(T-Y`
z-Dy*)VVC#4-$|Zd<s;r6`iM)ZPizCEnS|!w>m>~C5={@!Na?-$w6}p@`PJY!wzRbF
z<!IDizNVKN8lOfWyt9ZltESB#mObeon>ayc_cj%IGQw)u65$b^r^rFJIiUiLK7k%^
zJptuXnRdI5$5Pt_p6v&c1@D6v#=Z5+wr6MpaIdi{68{Ot#dbJ|I)J%=S57|;ZtsB(
z^L+_YL#$waZd-adj1tCz*Y0I<mJ&3KS|AyL-jwf`m-ywEpOLU=(6~fRBv<i(T((z&
zmOf)qx~YygU=UJv@ig30?*yKU^cP2P<1bA2+pPqvTz0Z=`aHe#CwU(zVo!%*0|-j`
zI3jKMTujcW(^_Rob>eMIr&8|ThmSW=oR}bP03x(N%X(V(n5{sqfKw2qh^m*d&RcGI
zN1&soP1d8&fj5Q48xaLsw0KL06|A1+VQg3hAi<_<g9s+8Gp$xfo%*oNqEouz@lzs`
z04<btBr%pW%Fvq$Bgn6!3en3@hYjr_B))<b335<Hl56)BEhr$DwZ#Q}m@u=#Z-?1e
zmk&~XgGu$apa0~OH}(8teMdY0(=_$(frIu1mYfRhO9aXXf16nz)Uryn4W~cVe;Q?k
zC8YSmXY_QX8EX^41!_{_!cYdT{~``E_dsyXi`f{I!7muMIwC0AJu=x05-4(DaIW`V
zj5A}yAc)$+{*-P&)K#XKM110{`XRJFhrdfVjMjs+P(hBC$Pq<aruCTWEg!}XVl)H<
zhSh-3l^?}L>h!58(HE%allfIws#)o^m%7|CvIw;ToVdrS3JLxAtsyRp<k)AQ6cLc>
zeA<c|{VYjUTE9$xk+Hiigb8ik(}&o@udE2I&)zblpVCLBBIwQrorn%>ZY+C)4VcyI
z)nFQgQQTXw<sWB)fikDbNwOqvZ^_~ABe?Vo#5bD?!QGtfYW!2(K56lO8Dg?os!}kI
z;T{<;`F5Nxf1Dgp+}5NvF{8(>MxwuHGb-v7c!|O^!HhykgX&R8s|0zQcjm-fq?J+~
zZ2`;nqhII}Gck3pejuMh|3HMm5hxQcjgVwef*f_Xl<28<kN-$Z=EPyej)WfmNQd{I
zzL?ulw6#0Rp)pZ!LVzScHy=(IK|)gNzFI=1$re|D#jjbJ)LN#6PU$eavwdgLYi|<_
zsm0~M8)V`7i_4A`H16z#h|{36e9a$YEdn=0N888XOP2I9+F;X3Br;7ipbIDI-{EHJ
zw*^>Iz@AaA8dm_ZO=d435HvmusqpOc+3oQhUhxg55F(9NMJ2OC4;{=h&bGdwrIX97
zHLECsI0;ZYHT$IMXKxw+`H%FY)C9e5ES|R2T&9-*_XBdpvM2KR@kvYDD0Ss2cDM>A
zBr6M81O8&%lNY^IQ%8z?UwvTW07<XPsv!%<zem%T8Su|P7X(vZA!<<k*PY=EQ^Ni5
zE84A$i$E+ar~RT5v6ow8QlGhQh2f%zMS6_=n3fzJ%{D1Y6X}q@bBlWHVk9X2)g{5z
z0tUuS1PEtaMAujzMv+I#4`DO>&>Rte#Eh4f5XW=$=wT_=;T>z8>;@YBiD!*Cz|BfJ
zUv^WE;AaOOo2KO`K3$h5p7z-(QsS4n>UY+eah|Yf8G<S<iMHF}*Af>>hrL1Exz{#_
z?{Zb(LG_QzjO%{PSu)`P@z&Vq0_MIJIc}Bqf?;T~sg*(Wj9A|AcS6Y}@xPy8w^aoH
zaf(kGTVqRCyu%qSHGL}Q=zTdDDMME`HWs%fC&ET4C_q2EJ$?6R`cuiSC4Qe*R>Ac5
zmGZg%njZ<*Qx-Zr%24a{VOm}-dZT&KsFVs6pcl$)>XdqB#<C+Lpp;+bUo!o>nVQzP
zPFx;y=K3ThI|5w!+*wm(`Z7dDj`=ATe}NjqR2FSPvuk&Xsj{cy%#(kqNoHgfaL|;y
z^a066o=jiK@}d?U#-ic`gDf0d;_nxooHloR`av&Pf-}$btnHLQdtd!NP51~%QH_y!
zeDcV^GSb-Jpy`2b4&&U)R4zE-$@^b00T|D$gW*!A?x!9DaaWrE;GO!?9c1t{ymTR%
zeyaoFcGL>!Q8;+itFiU;xXwQPu`dI0^(vr#iAu-_1twjn5UnS)McDMso$b|)Wv_Ai
zVGPnM1EPzINGOTHiV)dH)0MSCaNIAWhhG>wARiRc)(#1`zw5uUt3<s}nW<|1q#W%q
zRCb%N>oWW1J&C32$KI_)UZlS;NX_F-%${zTApmX7&&pfSYO9^DH*PqfhAb@UT>dXC
zi^C}TQ5sTY(SNJ1B(xI1`$>sv^5M;>eVrSJ5&}Bo*UORK&727yIzc*irF|2tJafz!
zl|*-{kaxaJO(O7_@L4N~_!!_57GrDwIPlr~_9f*{`ALsyQ@w+pP$3T04Pw?MfQA{m
zF}f|!7k-mAy845UjKdB9g^8s&)p!E*Tn~Kx^&9D;Z<EvUCh2X$owf>;Mxd-#pI+H4
z11vcJM3vq*s%F;J&)vcZ{f#x`Bp7aDOV@#*d0z9uZVvL?;lauFFMUprgLl2kNJC)L
zg$%g)7uktg&8@<>ufw6!eZiY9#vyxWNK?(|Mnh*mG>!V$3^UoHYdnzU%Fcj-Gw`)t
zMx?<BfMsi^4<u|b`QZH1w?q+!{G|bL^;ac{ib+oom(xmZ&otGiTeG;@VmoBsE3e(~
z-COB|(*n3Bcmmcsve+SnizZhuDm`OsHh&J31trn?<@B9#^bl7=xV^Mpn4F+^y*Usx
z;5bDyr$Rf262Vnj%lr8z#cP$83IYM0t-du*n0Z4bDw7O`!2SL;8U{Psd!bxNt9tau
zZpO%7z~81$OTQ(l_&t9q)LM!|FcXURrOak9Q4PN;(4Soi!4cFlaF`x^DBW`X-o;63
z;L#wBS<+_t;}*hKQkJ_o4s9ARQ4D7-C%_cM#j-qEX&#0H7PRsb*#y0V5bUQv3f+E|
zq`7){6$p-FTbD<lz7-Z6DlP^m=xaC@ZiFQGaw$1%Pr`KQOu`98yz7H<RZF=7Eg{0~
zhyQAq781>P4mwXbT_$+gaC~C>qLueXl@d{AnA(7Ci0YL8q#>nsX~$qu*$=rye{`PO
z3m?o90hqnC*!L#~vbH`KKr|)+{!TPp>fkY0LWIFwi<;?Erw2|j;Ad&NyF}P8X>O5)
zA2^$e<Auz9^5&~STY<}JAr2SVLYnbK5)IS$$^_Z-PE^0FY@4scz6&DDu83R&=bN46
zA<kSk#>gPxiO4$qhR+Q*)}hzkOU+;X=iKnDu?`vh{y3%m&c$hGQT5$^Go&szs%UC-
zREA)i)h{2@B_p*GRw@-+`82uj$l==D5PrZ&(E@!LiH;2~LW%f&(L-A@9i*Ppp0b0r
z@##B+_Dhn3=!g$fZjAm20XACs%hSQ5M0;Vwfv47o(dd92LK=<cBI7_W@VxE269s?0
zfGw~wA&(C`55xoWngwPhQz8z9wJ$F+z$zdpXiR^fp6vasEdx*@o3>a?__1|<SWG^S
z3ypolcg15K12=1cv50f*hn`WRp8nYjMB|`%!7ja7jsrSjndB<n3o7_UJ*57ipH@lf
zbi>u>sxb>FCAzMMF9gcvo7dm>EAMrN#~HZ9Ce=PH?U7*w5UJ+Ym7_67DiadlUV$rC
zGqAwDPtOHv?lXL{q(pplyCp*C@-e-hPa%`W$t3tzVgv@t=#D(`#1md3#BF7`is}~r
zS<0g2oGY2-)euBfacyLBcN%kRfq#Z&2=5NURl|}QgyjtLruHC1uJ-a~G(AnbWhF_s
z<H1qXhEIP1H+v_G!vLq!%)xeUGY5yA%Y3Ok0$8{oGd@{GD+c-CPKRkTIF7Gk;g|j>
zH~}&!y$m#Wsv^3haBocrmQ{gKG3`;m$G8aCu0JE(VM2dYkqY?0f1yin?HqjGp>n9S
zu3~MJbaWE+)ltn*z2@{xr~g^p$NsDk_Nde5Um{d*`rm>m!ts~F8y?3UHeK${S*q?~
zpJ18jD<CJLZ6E1?Y&k{(dE!GP@1od#+~8g-J4!IHmx>H=#l=bLDxz*@66NdLdMajp
z-0bXUtIzJaKNe-6p{#jxx`vXOKFs4mU<#l5oYqW&x!Q^i^)wm}^p4}mFOVlMpmRzc
zyup{A^#)5#ZwQobE5IWVVrP;GlM?Jxg`gHTIgb++X*H@;zoHOL+h`Ip$?FYXlF7nh
zQ$qSWt&Btjo>pOVW(3uuM@(U9TnxMl|4@InQ!$6$K=wvEZj<_~Ufv?(Z?hZSPs8T?
zH-b6p%KUrT>=|lN50(0z#%!3LGAQVA^VjCy%tMM~d3QhUpA&W&P&F<@mcl`wwN_Xc
zVUw~<J5b=BkwQ9f<)cLYFW&OS*qQ~2Q8@)br+H85G53M>Tga6_u}_D-0dlO<L!0Si
z+82p7x4!V44|Kj62XvfKQ7!=RR_{<qm<aFka5?aLRSB@n<m^(Xq5x4v6J;kwc0{3n
zV&5-R8II+kdV<93=q}40$^F~|O6vWLzA9_Vbl5@vdrdLNN^g+#^l)^6z)7nIL#y+z
zKT15llx}JMaY+m-lbRY0QCJ6rwg>rJk_<HfaNLS}gl!cuI{J^%S=kSV_XZiqC?yla
ztP-Ua7JbUvX&Xz54FTf&{w<Vt&2u#ujCqxyQ_@kS2?>UmOY-7kQ+ra}J<MQ`N_{5W
zh(c<<c}YwzZs4)AO_(4l*a^n#tH|(TEprOD1w!}V>EW6wLRy(%&WY{zC*lQ=N&XNB
zu;c`C83mv$+-s5m#x1>uD1!`toeI)ge@<x$s0yT{w@sH(pX}A&$$$4hwngsL!1(>#
zJG<n>t&_fl?bMkpLw-Wo?I+*Uqk%wZZW{sN_kH0r8IOm&#A^DX=v0g-xkHnR5YCGa
zPQxM$E0Dmh&9`mhgmhCUTImNjXXXpI?i9(2r&@aPfR7|f^l$=H5oLwuG?4Za5MrsO
z8dB-Cp*HTlZ>{15=puh%33J4}w%)~F_l5mTw6W*W5=OHCU6haU4s1}bi27ZL#yl(%
zSJg0Vcd$*u!X)#A*`;Ieb~1KN7;%{BLp)glce51%u;)vARS2ucAj0Y=q^)C_W<-r%
zAF*bFMWg78PGsHt8I3RYat=LSt86gsZm6DB+C5=81WYn09)bCvT>~!6d$rY~zz2fq
ztF*$b>7>Yqz(h9dR1^g#n|*~eHG!JaM!4qwCsG;md6sbNfzL#By?n)IvP)pzd!Q7f
z7Ch9#6AJCqaHFgP#b27RF+bK1IBH#9>Z%>(xB{SckKWKD*A&;a?9z7y*73?f?(nkO
zZm!9}mlIJ2j7S(xmS})g(@isqARp$A-gN7K^Q-SoKhC?jU$2ttPU`s1<>?2Z3{Ep%
zh$Tjc*IKC(om~Wo_qwP*2-uXb@@=leLG=M6FNOgSg(v3Sm!Hw)B-Zfz+9;&;C{TGj
z1JyeCK@A43)sTfisC!Y~dzzP#u?)4kUp3QjMQl`ESifG>TXzufZ`h^g*k$*M%=00P
z_d=eBeViHKEmVU=2Q#s}k_8=NDj@-Wv=QPp)hea6+b9B2ylZ_v*_&CA<psyYIPw(9
z*|kUt2d7gQTx$KMzrnOIqpE3vP#WP4bWc+bED|i#!g_cl#Z%ZWF-XaP3W|U~_<Jh+
zn5Th<L``|E=9U&YK_`2WTR<*CsklwNx#aqX*FV%n-rn5Y(OU^tFe=sL;2+Ge@IP=p
zF&Dk|5k)r<<<Zj*;ddq3EleYr65(NgGsb76@ZgHbs1~a}@sh}k+%KweVKb^a&1}L#
zN>F5!))UMLclp%jO5{4CU`2iQYfKa~lQfU^1)?`y>v1dFQtS(OQM_3qj{Qvg&&m1J
zjbJdEIrzqFTIfz%OS*j5Uw5tsRuAZ+Iy2~7j62YOjz$stm}3-%X-o}Q@z0Rg@k3MC
zAW?2R{EC{9!6+<ZnxmfY#3&|ydi9ko@mdG(#OT^G|FaErzFy_sP887b>4k<KyiRHB
zDp3^E3Mb1yfeQDk1ZdYsIpjb`YV9|lxuK1N(<$N;?b!-BzqUDl+oy?&PZ-s1$(!@m
z<sCnDeKrmjXOjtP(I@Zb>#yrD;LixZJ^#7S1+V?}_4ltn@&V&(Yb-AUIrRVmKa!|^
zE+$e&21&go-h93rRu6V-#wLzx%P;i>*Zx=?QUATt&v@^6n)MjuXb!?jRjp$TU{uOM
z0cisb{%C&2o(nWsogvTPW2T6gc8KvU;&mSdN)9)}eUZwb&QO!gG}C|Sy`9n6%Tma4
zj1Lfe+L`{uoNp<gby|xOjt#n@`vg}@^pS1dor|;PrzS!>4G*%3X*(<}S*vX`I099s
zCn%SH{Z}HoZcb#20uttz(-c~O=sbhH>H9S9^bv(&tHg2jgy*Yx*?SD8SP14D^xCr~
z)81uKd@Pq>bvGDNc*1U8KT7frQxok{imGxq?|H3e8`0~6rDm9}ISI+*Z4%o|70gIQ
zzxOVgQU7?1sjVL;rx*#<$RV|>n3_);WDu=TG$q{cZdFJPX|pu>v5v7y-woH6Qdses
zGz}JT7&Hp75ft{abnUUa^k!OSs1br5&AnqVnO|}?h-4nLiL|BIsdbRVNPhGY$!ugJ
z;i@o7Bn4|r-gd2SzAUc%&9A(6KLzRqNFUn4`Wh2hMwbTb0r~M~khr+A_4EVd{vBVU
z6QPd1+~rAFVf22lrqVWh_a_OA)rt-b%7QP{mY%-(s<;A6yMB$HM4@DN!I-9r$PihM
zIkOSlrK_4(s3n1FiAqEGESdfoO>2gG;t8LmZ2iJ5R$39@RH9FxjL{PQ%3ij$I_pww
z;Ycmk^mzJ9eL}^%x$}iV#4)Jojrb7f8d;1IdaV*tD1H$GFF})Ie=jA{OA`<~e-QQ|
zY4XSF(V)~}H@Zad+fq*D8Mk>EOUZ$cfF9NU>GaXR>L=-r;eP68A%u|PDKnQCpsurS
z?Oi0JaS035OP!U$oJ34(OAXzFJc>2y%#K@OTTw1snwvmUd?Wp(QworW!Oh4`TyC#u
zr4_w7;b#0|y>@APb3N#xA}owHGYU7M&UL>>`@76h$ecJWtW2=R^^QCZm`S#4n!m)6
zG`;7rX{+iFYx8}*bozd+vBRp5UBng9+NU3vbRwbCYGsnEGm~8sOKv@SBMmjqNz**W
zolH$*7R8;U+$Q#o#P?O-!hhn_Oa}_w{d*4<^z)UmIpz}hW82*zSCqWGr<F6zGIK95
zF~Nl*F-nlp&rt3N7c)_3n&}hjlca^o(i~%f2CoZ47iyd~Xj|I;TCGZHj$5sibQ^OG
zy<NCJKVpDW*(>m!TfQq>n}xQ;zb1|~nk99A6*)S_h?Q)bK6o~$6$0UN0}FioCR5R>
z^-Ovb5egBCXj5^9uP`!FKnOzr=4-I69e|(Ru#E&=y8fQtvOs&I`~(qBQc6|8fV@BR
zpUA3W!j4MwKW3tzQ;3<g;tK^#qZ#h_QW2MX`4yi~a}ZcWSa80WhLP?t7i&;uD1w(?
zyRXmRy7e6Y(|%wQvfIk;2hBb{j93Kmm&n`*oM+9d*ya#vg;EkZT1Rngd#Wgx=y<WI
z?~`~*wMV=Xz)$|m{-X2?2dzv~l310R7&J*5`WRscRD2-HO6Y=sK3GFuQv9zyDCANd
zXPIHxG+7IZXl_Mbookx<u(gNn%=I)maScd+`kBjMto0+}npKDRq~S@L@ym4vgq(Jj
zAL_Fd>0xHZ<Ceou$50R@mJ{TUp3kY)1$Gf-(pozCTY!~*w}5y4acOm_VYcFyqLUSM
z(=)%4MV6XEe*N|22nbmld^C<)G3&1MysPI9tk(~p_YiqFgmU0klt|Gq(SO~gw?^TI
zR0z6E>e`;(8F-F<_#A=Z^$UR8l?e+x^`H|Fmz_bv>DL+4QRBipY)I=a!>7@3;ulIs
z;=kAc8O8S~DzgV<*u`v%KxNpy>G$C%&$`MeL7n@U^g`=l_~h2;g&m3odn=Pn{3>l#
zCP3;)CNZYayK1wmuvy^E@Di`Tq0ayznAHK`xR&-JC@F3R0=NpxSyqMOxRg^Qr|GJ$
z$3QA_wf?DBDqGsSVf?p*2=yYUMJS?iK&~P!C}bh^9AUMEAk##oWO};GooCu!W`(-Z
zW@rdxS*r#p{)yW%Yd=Vm+ukA!7wLY5=)Ao7!MX20C@@@@2!&c<sP{ai)h98Mxmh{_
zRD9-<*(Vrv{dxKhMnoKRgK58FEX^2-`0S;KDuJ@u%ZOf-|JU9HrSd<iSegTfj}XG8
zT*aIcNK=;MyMY9tt%v`H-9nV%q~%zIazJu`Py*n<%sBWA*)VgDYTfe?hwqka!dPo9
zaeHx5|5@DL))5igc~h7#eQ?w{$hI`RYrmVYKO^6ihlk@uisMKX`kh2yQjKudZB*(k
z!Fgwx&8XIx7M<SYsCAk!OhQ(i;^VDs<~rkZ`bBC%`^Y+PJ~_b>wgB!Rl-IyK!%ThH
zJ*X8yp(LXRpjmQZaqiBtzjrvP{c~w|M@v0(3=?c5vF7KrjB^~lTrcVIfXD$F^it+!
z)}aUU?t-3<krlL5tbs-(GT6785CpHKB)`@Wn6iQ_0)ibFW6*<#c~d-`!bFHtCkAL|
zrmX-3HPkXtjAjf{X3?drg956B3`(Haf*z{J53(qj_8)Pqn#Gd$rCD12!}>dPfeJ-=
zQ0mx)HH}Xx&s+@HL-B&h#@*tNwqCX)`hkI<d*sWfjCT1&*hu&rS(<bzK`5T5kuw^@
z4c;(2NU-UFytTHW*~I(9m;E0PByc}kgc==`45K1`OzY4A5{yol2T`oljgNV4bwh_h
zE>T{!4^&<V(VHs5Vs?POpmu9z#Jp~tnAX>{<Q?=6B5e4E+!FH_!sWy(J)vAuSZn<?
zWskyBMJ<3gpm4>{+b7^$J9zE`de(V~AGdD%o7)<PHjaQBY8aHC);Eh^b8;sgN>6I_
z9dk7URfM}7m%o4J2Q|>n0OCxKtwCZ`?l!jPR~p;?eY>LMnKA>*R%zAW2QU=WjLGwP
zTPeJf?;Ns`J3Pu75AIlPK=RX&-*}nc^fCk5Yds0W#%joGFB1uJ_Ap5!2DICWzRPTM
zm-eh)Cc~h8#~VZtG8}p1{ugu_0u^Jk`=j9g(wx>Uf`6)`k)Z+t@;=3LBK-c5CLU=}
zHESd3oq_IpzB^j)<T_x66v!{A4VZs_-t8QRCC&y^KMYFOf1x;|1i=%Crr)C<QFgVP
zMYGdOGnr1IUfBZET|<iR6&I|bSNoI|c_U1&W0QD<!If>$lSvfQkhkXKiGW8+;6WKr
zuRb^*dSR&ES~){FHE2$cUmoUrBOsYHZ(MHrZRFgc1Ih~blC|k3K+w~NaMF3;hrgiq
z5nk**SQqOV0-HoA{CkjIEIR1~sG?S%F89*?i_}vEuF|eFk8qoCni*}2!Gz;5R3QHS
zo02aVC!`HTM?>KJ$u34>KgGg5RM1#AQAEaR&7jjNDzDWKbcL#EBjJ0u=}WIfEkRE_
zy<$Uz!v2FFuW>D<Q@W}F(?Zp-`cSe{4GPb=%Nzh}(_xWFI@98O?8u&6$vsBwfUxk>
z7a-SO#19TS`?kl`mQYPOYAe9NcZOf$gJ@`ayN*YVsru=NhYoeM1nOF_hrjn1;MP_q
zHBocwpPf43)&Jt95L@{D)gM)h!u}tR{AMR&B6}Jot@j4pB8@1n`)=!~{Jzh+Gxez6
z1X2RQYSW+9t|_uhGPXdK1qW7pvCh@5D_#@>j4>7C#()?WZB@R7Ecl=ld4BP=gjXZX
z_R@^m0>)9{A2>jgy?h4?R!^|b!I0E*um9%ild!mI?Kwe^#DEj|ETtO=${E^Jb5wn>
zbk9%DM>$Nu!0DOih;{uJy&7Vdi-P^?Ev-jLS&CvkqCskJ=wrGKtRkFY;tcuNUFg#}
zOn+X)0wi0EX2M02R?5moKfLbs+RG!@;Wt(S6eOyXZALaZ##6LhrsSJAAr5&_lv+Ux
zTZd8gdtAHm%)94{?@cfnLQg|8%=Sd7Nms}9r%*X0jrawm4lay#DOk5KLH>LbOe_g<
zQ}YtBYN<mdiavsuPgYo<CjY)S2*;2b`{M3t01S=kL~TT0TyRQ)>g=k8h0u<|$6Wu=
zBCwegxV#2<kXw`}4A?C|Imz8(b>I{JETEbm|F6?K3$Ooz6?*kF*3JMUX!+ztjO}4W
z#GC!rDYjv5<Yf9U{@z8W6>&Uqwhm2u0jeFP*18Y9o=r2V<O7(&0z{2&U-*2Ra!*HF
zWW1kT%_Zu`-&-sV{(7Ja^y%qCvab3;DCR`Fj6fLb#aL7+n`94HUci|$lPb`k&7iLS
zK4c}1q3I)hI`_~!TZ8dMs!_P0tsb`XD9z%(mAYih5fLusU6_)k5CT6@Uwi2pOT>i^
z?XROqR!dZ)4Bgm4v6fcb>9Jw%anZfZmzc`>a@wB?!O4OkL}_|0Z+eZ>f2m!4C-UQb
zOUa#J|9e*oDa|I`uOlD)5+FNOloQRFVK)QZuQwpznP!n4s(7k|zrRfj`Ci9c3_9re
z!?2lek%fs{r0BkAtc}xzB%;yNhoeCZN4Qx+Y`Raa2nfCUc|UrIOl8;zpZjxt^Ms2l
zlI@+QJrmw{3Dm93dY6Jt#sE%rwmqm(t2TYe^dO>I5$nK6!<~v=z6}ES0`HYD3oi~P
zJK#>GE{}|Nm=oxCpkSu_KK_+=FP&6IE1Tkbl(Wy;<c1y3?eOM+Dwk$^m<G*~^VCn_
zCGZ2mg0%9O7uk(urhaX}OI_+{P&<c7$skqjo7Lp2<WAZq8r~q@=_=3^Kd3>ERul<i
z^ikYNP|+u8V27q9a_{|u3rI9O@R?*<77MOMw1uJ>tk&$6>Namu@)6I*K&b`@sJt3b
zqp4B0ACmk~z~%Z!*Kb<%V}~LpD&~b^6kcR5@5XX{B`#y2fsCOCTe!=?aWjMLe0bS;
zl!bcu+NqJ$zKaA`g__Wr{Y|YFf0sbD9-{1Fk^qh9W%V?Ig-D2Aej(*5{#;#H^vaky
z2_el5TTl))kglKVrZpr?-<T_zJ<c!ORIZA`6u;7+HmgrUxgz{D&Ng`j(6FaxY15c}
zS%T>B$#yGeIfO9N*0dk~vQvpjq6BD+ey#X?$+q8HS=?UqK8bmz*ZcxQI0ItFxNvhj
zou|poLgxr?Hy0PZO%=Lu>}6Z=x24TcUoaAWnklm-P(ZvJMc3U+JXi+JOkS-WL|5|O
z3)tkrv`Sa|8es)?((etX0OKmE=>4s|Ktlpm^YbXF#cUkKLo{|cC0dgOvmWmo{LEXy
zH;0{=77V^I23yoc5oHE$uAL}rOPp^%UmzbRBmTRc=)>{U=gpaO;mv_V!5TCZED$vq
zNhs@eRKaG#TJ}Goa`-*D1#1%>T*Epm)byc;OoLCq4T8ywKe;F)CQ(Dg>-1_Sh_n*5
z-syieo1dK4@=g}b5Gfy@#KyRn{7vv&P22v>r=|f)weW)$h;`%@#bpPF;Wf<CMS6Ej
zG0>ET<lB&jUNc8g+Q-t`r!+cS;{SANPk%020jWCzHc<*!lvHO#y1SErAN2v?S0>PU
zr?)F_CIdi30)Qz$9v@k|s4np;CPBoSi0gZCpheY3+joTGOs(YJVd5AT$&a*ch_uyW
zwzP9M&8Hvi=a^>%NxvO}4yyohDkAXYLImz~BpGWf{Z0D4cEh;1x2jjm@~OtkeCY$Q
zy0f~Ob?p1o587$c;juxVOvr4Leavf%MDji^Yxkjb`lJMMh<MT`CDpE$R@x<^LLWbl
zMldD^krak7W{?hk_o+3<7X=<tT(hg+=$nzCDpi?GQ}ZgmJPTx~8KGXZu(_`POh!@A
z(r+E+y`@K5G_;Uy<CQNBs+<Xpry<W~Lh@TA(fpdZ!9}Qaps#_D!CtS>kQF-g`a65b
z`V!km3rAU?4n{39BBr6%3A{kRLy1<e@-QsEXXF1&7`i;u!8jFF#MCWtEFGvqM(fpZ
zBl*AD9C;pr4f;Gd@8->bOCgUfc(+gu8DCkU&%G#h<CRxTcj6S)mjZXpW@6Wf$K+57
z8%uw=t?06<C_r8vlvuPTx%yZel+H;W;!&!Bk@UHK>-uf|m#U4l9(D1v!MhZakA@l+
z|I{C3CGn^OeB2Ze3RE#4Q7ov9IauKV>$I+oZ88S@u{Iyv?g2ge;wxCcTokgvdJEmV
zXAM<OnSm>ly0!(Qn?wc(X?l2uXpIwts;%9y1Ndn<<e#1e>btuaGLz{67*His_vjnM
z;;C^mr>u13nwq=Xt;>WQoy>YYQH-5xtWBo1!0MuJhb07&U#1N`-VbN11<Pvb&pmD0
zt+zrDN4x5PtThpXp%57g0b82iR1VWC5E4a*((075H@OihZz!qdf6n1lD(WKsp;e<d
zB4^MqUYX>xF_{U<RIx^T=SbC;5*20vR1_D1nws$+(g0YB$?IEZh#K11&+eDv7^qHZ
zMD0ujt=Uj18|{bf&#1XlagWZ++Cg}llvv3hF~ZIZ=NH-icVTg<ac6DE=Eo_~5m@H*
zFhf#n1Uc|htPhA(97K7kh`xAN)S!P?9_wX|w<>U>Tfvf7<Ic@*FKRQ9=*UE2G)u@W
zPT%oQm#ysc*X#}AN=yjaRRoKQ;JPWq0v!^y;dyF;5Rk^#!d5q;$5$u&Px>Rb)fN)E
zY0+))2zb+$1~{+9V8mQXz8xw05&DazQ&Md*IqAj~G=I+{Gb+rU*gi3XK~#KE)Vf%(
zgH)lKisa_T14VTG;-i+w0Y&K;p18Rnoa}=P0F4ByGpT!|8oVD6c(nAK{=|O~qk2v7
z28N``I+H&qQuoQMECtgf;&b>o^wsfqw@AlTpgE(QxQ-HI1iS(%8m2TYyY=u+U%AtC
z-T7s4)%ggO^Jap@B`*(!pRod_j-4fT2SLS}imfbcN8)`}D`sz<_~KzFipr0>iQ0Vh
z{e*QB(Byc8cq8I%^5Mqn^g{hWudrP|cmKZsO2-3wd!%tXnhgvHXn(osH($Tub<?gT
zKxBdiJf={3iE1E%m1>>5tap9mKbxATbQr6y?HwAj>-scq``P4_5-7hK*e`Ak=TiED
z7UQP!*@;5m>Fi+b)14CU+h@E{DrYo>=lU%_97if0I_lsFlIv=`{8HG0yoa8lE(8<p
zG9A_=@wUJ3U7Vbn>7yF(SL!AvR*-jSeg@P(>VRO*EI;~tFF<{$OF{<K#E@GvOo7=b
zcsAZD(;=b84R)8wHKmA;t6Qa^qqXRY0ACr(_!zHc8fqO6szyE!N>d{ea3mt~E;O4f
zVuBFy8rAhi3O|&#qOoOIX9R^%vB^!CAEY&W>vl+*p7LQ)liEIA<d}3oe#(SLK^BN&
zyIm@0rr*r1?+j9l!k*{=b4jSGw3XX~L|W<^p{;+bSG`gJ?l1`DN4R7@rDwCID|SUZ
z#1Cbv5+GB3Xmfj^G+5%;=W%=lgR=k<zByvcBugSpNzszd&F2fc3S1oT^1nIyXxdYJ
zHB7T8=l4RMj)u1a`&AwUyjhGP01sjvS{sL^S8IR9R@Zh+BlUz5`LMOPuW9ohpSMnQ
zVJ_pc;#Ntcn#v%`GE1dET$zzhL*KlY1guig&EqJWFQEVms*YJ{jyrU5$$~y|R5r#n
z%me21?ZOw(76Cnbnh3r4QU*JGtG^HM-!rp$+4}T5Gq6XD@(jrtJ?6?S7X~u!O@jnF
zMJS0<B)d@J5M=dzlo9F7kQ36fdXs@4scG0TQJmwr=8e+OofjQ+qJ0#0iLw^j5H9j~
z`cwTR=eo{)pq~};O@}nUxv`@^Q}IS}Aa+hAS(<Rd1nN__{^M&u@a@@wZ1THnliI5v
zhzraXXDRkM;@UixF&;=NuKyo=$w-C{@Tm0Ur<q7+%3ls58`Gt!n~xnYH4}z^If?m3
zdJSw@j%NB!!5mZ7S(1I+GCh;PPBIR0o{WZLQ)T`|)emL#1W!^CbyzW~1P=y&fwE?;
z0Dh*lJ{e{~*-<Zy?#h`VCOKHT#zX%rN~#&nl0}p;<Ah@6(l=RMZc>}$-l~IB$}(zG
z0G*h?1f3Pu0GKXOfc>B?@u(6D{x!^M%V*F|{rTlqewjRsLJk(*n7<jB*dR=V*(*ta
z{!|~+2+^_VlR*eDUC;=2<2-zum$;i`qZg1Xnu-jj|EhzE9wHgJ47}QJQ*kiog)gI)
zv|?zSaOB%)=vLCJ0T(7o*91THw_z*AT8Rytl@?ZGu?8L`M+~~@o-eGe`Iq!dNhu1n
z;_;HRWkR+ZUgr<q21TSj+?95X`jgJ?AdKE<bqhW)MJ-cCs&zsUvQA?D&h}%ugoqGv
z5cw<gs(=YPR7=~n3E>GdygX=_R2~jkIn}Rl%T){)isd9V*4+}xj^5uUOFp!kJMx3M
zeEK5Yr1dV%OOfWtr93PL%o%<tdAgBC8J!`r{E4Bjag#c`yTN6Hy`Vb&brov63@S6(
zuKu}p8oPhvwfjDwSEGM|{P}nU6*vC)GA3qz^EY*>G4A6@ia4yiVK)m*nN*-Kpf)t@
zmii+!Phv^9JwWNIeS3Ejc|KgWPpZ}uFoFkIiY)dMF{S!Ea;0)stNv^xkU;`6b2~7d
z#>aWhYSpeEF?Pg|mdYC)xnUx=G*gG}#bWZ+<&~Ydo_m8x@X;lEPvd~SmbFfiN8nn`
z6bG0maS5q=si}nqc?EK>e$x=197I59ndQ3^8LAFdsrymmvwNLvq0wfX9VCnu5z)<!
z+#a^F-ix`nRY8tO;**aw`A9to8&cXfO+T(MR%jX*jhr$C5LSQy3MlQXGgIzG3UR{m
z<mFz_$awKb450OfF!ewg)kNf<0z3Bzoe_%kD*y;y5b;YJQ>pKuEP^U(nElMle(R_m
zi7YiS4PhF6u=p5`A79c5C@WKE!~nYXl~Q5;@K`{y+?IyHym$SkFDrApguh5O|BN`3
z*huQGiGUw@W0`oV&)r*H*je!rtAHn$%wU{uE!w9Vx>n8X3<A}&jWNU7hf&5WGhlsB
ztxVP2!MoiiR8!l2$N&<*h!J$ZpZknffH^SCuVXl&PYAM$2r`{ZqabK=RJ$SF83N1E
z7ipq=mbR7VoOLoih@4h!nQsO0weSNfOa@oc6akK|b|2n8O};w0Zdwfyk@XV}skApu
z$GOrAPO76G@&3%!f4zD?Ki(R}W9XJzfRFn9K-2T4ztokZ8aS{M_d0bJ)oM-3E-Llt
z<x!@kCP#kk=KR`xo+NZSs5RFk9KwDCe)cC}I+0=~CFPs^r7wLc#VZi)i0%E`^|xXj
zh;|<n$Oq2M7YpfwT;n17(DmQeT-;m-ZH4XW?;wWomb^H{H~r%P6Hcc)jv(O@5<z33
zL_f~IfP1+rU`0*#Flo%s_|Q1;#kYN>mK*Mn|I+uB77A$(H`bXtql}+Xbw*~{Hw7jP
zbkQ(Cjq7=Q&3GwdM7<a@Gse@s6f)PPc8Y3<qNr#jZaQ|`A%LU86h`qkw$*smKQ}v_
zT73tX_eVtEyqvA!^qn$kE^fE5!kG`AQ^OolnlNBiM2)8(>_+VdMg}Hga%&`k1HSQ0
zy8TpiW`BZvt2-+`0*xj$MHT!MH)tWmYEf{G_IS|<soKM1CktlG2&EW{aF0*C$pfnM
z+hmK=pM>NFj+}u}9Z^X+B3nsFXhB$Sy7AhDI;7CG!lPW&c%3;2#deo%U}_c10alIS
zjrWve3SOkoLNrKZPkD9=tvWzny+1>rKY8mZ(}CXViDUuvU%Q1NU@)WL8xxl1lGMCJ
zp7ZHjPl?kMMo;q&m=cFcqno&5FF$ollvCuDWo94dQspMW9DLrwXNn`GW<~!!hh0rW
z<OCUB6JJQLlG!}PGmx~NTix;BV(OE>M6?<IR)3#wH1r0|t4kq9EQIF=ARIZ_Xo$lE
zs;FpD#AQX9pwYOrdeUeR$w6>Gk0F}o)TFEyNHvD(f8WzYRErU_cGT|bU5IJIj9Qh1
zd$8x+i~1+4X|CqBe?r(T8sGBU*KgiR?zCy(Jp=ljzH8<?j2785F}_6&xM+zpkk@sg
zGiryyf?WIl^;?hYL}={8rH7rGtos&^_o8@k`rnK~1A{mVYI{_OsKbEgpd-nuQYgTr
zya;gY9j7a{!(|#~0(%=BDA=R7OQ3t!i}=NohCL6B5t>V#DKh7aT~e&?-`dW=tLvN+
zUI(e9B`984qfdV+$z@{ERiQw~rVoz-Dk~&gjRt|t0Gu6qhUTPkeYl$mV{YDAH$ebm
zh^mi1Y)9$F^xbe<Wt4NOBx8J&=DE54@odn2;4^aQ`YdrkGI?VlU@$c~pWSMx1U&5e
z+ex3E5_`VxtP7NZ)V_8`_LK&jrKKypq9|S-Y_uc3e#7gS9MmCUbW(eh-A5%c@&7PN
z?*ErpUh~`NGsJ}trmuc5G9*|BhYHY-_e0Q|1OhVEE29B9WjblD997B_?vF2nwkLe)
zFAIf08n~!BZbE5)$rc!&LsR0s2#nyU_8M+Gx!g=AEaAfF-W=^Bs<hhe)n%V`12Z#z
zg*Ny0k#^w|m1bN19LbcTObAni0^#+`TX%HAm)%Ei=%F`0d(%q+EJLO|Uet&SyNjR>
z^OH@X7xb6GFza<lo)bVc*R1a?`=^Zp30G{}3zPHWH>irKbhuz0qp~6f##4W*&(ar4
zhwiV@|1q4GW^oj}p}EE7)eX0pmu|uQDBiXf*F6Z=iM8U%B?TA`n5B!&@OLbVPSTp~
zse5l&=_LFP<xPn^M3LgM5Q2`73Kr0$q9<^;{P?B}FCREnFlFkl{|)B|h*}F`6i28)
z1*#fK*@jG`UIIZdQxwVUv|WxI0}B3R%2;wga|rlzWJh&6le~{F3`S`;FeP9d*LY!W
zK4i!HbmfQv-I&B_DpQaS7RyvQO$m>0Sbg6wLuKzNIwMG5>hp3wLMEwY(<EVP3#xsz
zu+?~Ww&^1s4lXmyxx{C*qKKxZMXUUpzu0+z@~BmKyBAb!mkl5eGKEFRm4)aa@S9LM
zWv1YuohCUJ7Y-kCMdwL3!Za}HPTWc5MejoO1`Aiz9K28vPQB?v)8y$_yeRs~sReW_
z<ngJ=4dQLZ5~1iaaan3}OQW-@mKC)GQXo>G8{mvAOQAGXO<HjN#KY^G8(J<_T1HE0
zW2sgeF@=Gj^`TGSir%~ZWXNR@HNIj}xwYAvZU{lWP;uP2MgTl?^;(f+;-|xjMH{@&
zdP!WS^jRnCrj^{d`T1xIj3I#ODV?+i&xbxkrPLq`DmQh*>S!1v-Cp{N*2<7DA3rlS
z;zkoR7c$HX-U#CRBU_xPZ`QGZab6s%U#EmQ<<D_OiT}~8ATU8bJYs)-E(v^8u2!%$
z$!8*xhZ%;Q^d}m?9!soA^ySX(^d0V>Uqh>sBr)Kx>{<b2iLgl`ZVstM^pdfJx*;!B
zV!a=~B?q(D1=!`FI`@0lTQ=GLfLx+C3R|fC2k8c_3*4C(qLCDONhzH>i`%CWaOr#9
zi}R2uS-z{!MPd<?R2=-cjc@>}KUJ`6ETctLquJES%JNY5MylS34>a%fo8BM2Q)>*r
z2)9`$#9=0&3D98e>cz;OS!|hzWaRSCJG<&8??WmkiX|G|%2j%s;fW(02GHFi1Va_{
z9|leOn48r}KOY=*a{L}rkSKU&ST*5V6V|nxAjG3LCXb>sv^H0wc?(qjY6_^H-aXAI
zAzC{}6e<cz8ZjfG8(n5kdkthkEf<c^F)*~Z(N-vURO9w*tG5?@ic3($qy#)Rb2nrs
zcXzc0TIR+^2bX?W3nUe-D&r<nmlV|4ytAzjO^%C5R02D@ZXh!&=<C&IpY_Wy@{81x
zz^*pw3UgJ9_r~NKg#vL4s+3pXZg$~gx(yJc_V01{F8>y`LE6B&0R=V{AY^}ijfh^A
zIO~h1-(0ESUaO7!DdNs()AEpXt~NR7ML4xRibl7${>)@ieuD*uFkJT>P&Y~w=P}Ow
zAn!2k3xc#v^p!Xx^S9^A2I|itRsBH|hT`hHhh8Nd?o}>Tgn`f$*FVrl!oz{+`kKb>
z>Z1xU3{#~NS3@8A)L%dz`yfzQos_{0crp9nm1b+BgIdLc(TPZ&WT*!Jg>=lfGP{Qp
zK(9Q#H`WrK$jxW(dt>}gwR;2;MZ(vgde$pJhNeK#40kObYbVingFU&&C7tYp!^o%2
zx;{tqz!ZOh3@uir9AR%_9S}JG(~_V*maDf+71=PmIn2V?A>IxKiA`=GPm%A1bDG#m
zDWepmRffqM7<MKBkyY;!zn8RFc3nc9?G`L-+*)patej}Kcw+i4m=OH{s}B<IO|<Wb
z0OfghLl4M`QYzX9BW(CFCDn(jTK75@7MmG*8xN?uu7=7B$nWqrFqlJNM_k~WSK-U4
zBP{yO=7#B`RCO#%$%8B;E1BT}10!3#TPNOKeIIcIbNZko<NrrR9<)<>7^hs-1m$XK
z;_((-qdugfC+UgiJ^gFI%t&vLaWOyJOULc%zoO6I!le35zUd)yH1hf)p=NpHT8Z5O
zq3&gja)nZd^;Ts1THFcIar&_u42czAdikc04J3^!3iW6DkP=c7Iq=HlGY9@s(bS05
z?X2e=P>7{WR*M3XuTBaJZvn&cxnS;Po<LGFI2V{=81iR?H2twPJ7tmJkjUe)n${i3
z^W1Q$bjOLb8-6-CWmePUhT@B*8kG0%cOU3eB<mHz;x<btipqcSw8(Yi=BqDffbQzY
z*WcD&Np=1H@sq@?<x-=mM5q06W&srqAoTy1x#ge@WX(UH;eM6MCoz^N$pttq0|XxZ
zYi6%ckv<ueKtK_$DE4sovc}{Y#>Ara0A)ODYI}>p$=!Y`tabU!rj7hAv>*9k5j5I+
z9j7%L`C!SO1R^ShfjgG)LMa*nIXD6_Ai(gjm@uyf2bl!!)k<{9IYk{T_Gy(D=>QRx
zghcUkB$pxFue28Mr}V-L(%+f)X>ZS{GoGY%J4Xz}f8QdeYMX(zYL1DjVh9K8Eswsj
zcbN!>$A(pe;yLc*Q4mZVRPm_zbIpHDzu6A_XMADrG-LxtP1uHr<HKVjfk6jS2YTO}
zVpF~&<lKl#M1F=qE&5Ksr&x%Afr{nmYvj?YZo!RBoxIXzgk<qkRK}gNCUj$FaV8<q
zK9cKU;zm~vu7DVDw-svl){{}I3}3BhhMIcU1WLY?^XV0bhPT3&?zGdu#t@(dKt@AG
zfk%b;#Ah+$KJC2QiZZu@+KLP(O{v-$MsJJ=ctxAF%@xZkiGd1j%v!jDdkHQ5mX3DV
zIagtqiMhyfTE+Y-MUW-1FA!K#EimH|sbU0j9l>%o>7n9n7_}5T73N60D5!lu|3tw0
zg2rs{f2JuVz|r@}SC-+~wYEG4mCqiG_B@agDXFnBCsrW(GD<ylo0d%)3mW#{&6jWJ
z?Dt96_(IOZv^2QEiqIdxpyif8W)IpG^fTF&1W0d16>F(yiEhW0_LZQ*E;RIWH=i-Z
z5e7gW{f~lqF-cH)6Z&p-awKhOpU(`r49TC5VkFYWo%OlJZM^`*4whFSDetWA$g2}f
zTohH8Nb@Z<6BTK`R*>duOIs6iS3j3=FKX?if5~LtlzNwEliCKqIl-xji`UqgU-YU8
z=xeA;X@kOk_Pukamn8$PanpjHa9dU$6Ri(l-A_T7F@z9BWgS878aE-7gnyI6&ykfT
z9v~yS2I_#hR<CAAQfQY_B{JEarE@WJqrHLIC%Q75vy>%MS)C<Ui~y|7KD3#8on4XD
zAMZplWPzxwDM|<jTwT0T?;QBUnU}cGQQ$x?mFuI0iLE;Ag8msGBQt4yI?f1pd#XZZ
z^CKdk%OG8;eG0FW*0^dYd}Cxs0@Tr?R~8!!CMhxZ(u*Vygp`n217bDW;T}jyQe6~C
zUG-K9^ovXoZ*`w8azd^9TRR2C-z&|(^|oV27J1}?@lnd34%(5RMC1N7F9_pWFRgJc
z!m(w3CBM~}J)l`4kqVuz|Gp#xI^EESghq)Q=YbQkCh+IgtgJV^QJU9LmIMJLZ+o31
zt*p&G@>vU8ltiO7Nv7`z0CY1Bq8Pm!{jwxEoniiw-X<{%SDJ!Ar$WU}-{o}$mqzla
zoM?sf&zCBqFvw8h;Get>wW>~#zp0e6ZHVLS212}lbvqt8x{R31;UIA$C(~aR3`4-)
z!_?)|$8$lua8j;5Hs3Y>a%~nGz-Z`q!c@I~OC;JHw$1B}HjFQyIXEn@)$f!h@IO|h
zf`i3*)!i~uXn{bOgU>|MFa~`?$uMyUWQ|Lz>O7jk@wIX1$OObH4aVEFr4uVJ!wVC%
zx=VFN?WZComIxxiXYYyM?*fD?R)wh4<K~&ZSS*Jrt+5AbvaSyxkHZ|8A53a{FZQdF
z3;p@^*T0gZ16~jmUkjzzwQxO>h^x2FV?sg2v`61dRE8{@8N^`$a6uv#fd9qmM3Bs1
z!1#<c70Rk)C<3--W>VGqi2yo$s~6u{)=7Qwsn>kPB_CVU7^I+aY<?(rlL_WvNMgPa
z0<KQ@Q@fp#X6R=cmryMUN<Xj2*OCVNH4vUex-!wn<OIZAQ#x=+M07xos;JkCrcb8N
zbqU|s41@1Uql`FMVQ9(7NN^qDIs`k&2CsH86Er*f1$|6E<`7PlN7!(|{0vR(1>?`7
zw8OSRnU{7WvtPkw3omF{N;;H^;Kv%^e_JO#kHaXM-hc8L?@-Mz!l3~g8~>%)A(S-+
zy|=hgZ|osbBn^ImKiar5oJ6Sf&~BZ}ARDT-hUZ{t)bCp^k65`m|C0g)1^eS}5y6-Z
zbLh=_jkdcFeA2pxkpj!=ENYk4+cZ-zyB(&nrq9Yq7T`!)BI-j(R&7|ErIDhhr)GYz
z82@Wbr+V&rY^9aXiNoc1V0x5`g+dLf?W-?6HwEp^Mypy5sDB9nsogCYdy$`c6B*#3
z&ROtp->eWYdvvnZzq#bmElF|3j6SN9@n6dglDhKIjgc|ws?u}r@PFmI@Bo&M8r{Sv
zo%>at9?rM!)LO>Yu6`y9-CJ+Mu^fXR+=L35(ZEYr8*oY1Y%>nzj*A7|%=iqnFZJc9
zE-|*KgST8zRs%3O#gu>a{k&E5X`5v21hWQ8#w~?59!#;y>VRa^s8sIqT76Ppu~G;T
z2RN#J7)mMSG+lkXxzb5HnMCapYhOi58nkCXj(4eINM)&p{Kaq={?0_<9*(p6L17on
z=fvV|E37JFz?>Qd==bs|p%e9o<MH%8NFvIB_H4x+wANf8cBWTk8#`aRekX#{*plb1
zK&2pOLsD_VrzLRR?;!RPMnb0;3^5|dX^+!{(6a>SL6ONHK7t!wFAUp5&uf7i_dGjM
zei?ud<$m%$;qcHdGp})!p;c(<Tqm#-uit#?mal5ypH-5SI<$a`OiRcz$r2F7%d2w5
zK+QK$R6`;$+)|>zX|U*tVnsh_oiJ$>LpABAl-|d#Qm|~spP{RGD-1{a0oPVvw<+_B
zw3*qCm>7+rP5brk0||lslMH6jq=V7){_E;Oumm=(L~jPF<SvmKs~aS1q0a}p{}b;C
zxgX^=A&DE!MZdcYO!GfV3(pjf1lZMM*jU%4$@C{`dmAvhYLH#MT~e4BW~DgP*Ro0Z
zX}k#-sX0<}f843OuS~ag*<5LOV?YJ`({aHy^*Uc20Xa?C9YwRVA}QLjVL$jApyTr3
z&`UxG`46X$-ms8#s3y$?2{Qh__yTTT#`5~C(4-i(_Gi_xAjrR0%@7P=<sM9IO8qBc
zvv8n?i6VArM>{+#UAFLD17hdM-0%G~var}W{J*5*Ndh!uu5|?XP6DAG(WUvtW>Z&K
zYB1yjcX8uI{ieZOy>piUa>|4~c#E?|J`N@jW(O06SqAmuGGRnGg{ovph-eCwzjkWG
zMW<hwlK!@#NkAuS`|@0fXuYAuzjDJgzq1`>|NMpO9TQ1_+Pom#ns}utK!}S*t8|qX
zvF1vs2KnX%AZSBEuvauhG{`Q=08~79Air`n$d(5YC8>Z0_F1NkF;#2O!?|_zReo7l
zy#_EIy$mi?4~<wSo~<rM@rT?KGXye@hj>HtD~t2DMSJYOGBKeWg_r>jT}Y#h9CHf(
z-3cT9l9v^sP;SW=gANFcB`$VQ|0wa$D)mqRvJNSi<hlCLgjpa`FNz&Z5(Efj>m+l`
z2r88&y;pp5qT0ppTn!UOZ{4B0WvjhKds3aJx=-t2(_bFzC_nS0{=ooB{RaT9_q<Dg
z+t7GG@qMedOIT_fEh=J+88;g?B)uVUN=!H|3pukO7QvE;LI4s@5C&QcgBAwV2e61L
zKFRq2WR{<E^;5R=JFV&IXa3A<+S(^2<Db<vy_XswdP>o8^t*a2A)uNlqhs_ADR25c
z=}C**O%rsH@@@SF%viqfUvydXCV()?AIa~C@#1hk6#XVRivHsqnGbzGVi@u<%wJo{
zUuM}>Ewmt!gwc_v)K+6>MUPZ(Ge2@0!`xduL|kNZ1hDE=&P~QqFNzcXC$O<fJTeqE
zJQRX1IacL_ki*G_dtn@8j1kB*@rX2)X%naPV#w_0pN{S?G$uIDBm|hhgb>j7DSMD9
z@KOTC-U$^EIj{cF#~H~(rQmbbUcYEbQV_EyEX>m>rpp65B}D6^lg?gL>}o`D3Kh*@
z$!Zg!YOX#KotFl~+nosA$t5Y6Z#v{A_i_X2r=m-rtjdpvHM2tHUE(?bNYbEeb{?Vi
zqlJcl*VtN3S`erTD}Mf@u(T2e{?z<rHw!#UhzTFw{Gnt~TSk!hbu=a*W8tvejq#_}
zZwHukf@AoPVHVayBF3z)!X;Ug6{(|BVgf|${0&C9k%9~Th>$=B>U~?oK|6i;rYub}
z>1Ck>vwsK^hwrceN?uX@Vj9t_Iw#DoZRtA$r7tt@Eg)CwnMenzHa(iCp`@to(3_DG
z#T-`Ee?Q^q_qL)qDq8B*Lew!{j&24U$nLGA*L>ecit<GWMR$0sPA=b+ZdG(8wkf2N
zmWgjhV#P8nYtsmpn#gjX`Cp!z1_qdVTah=5Wyl`UvyeZWntqBdL7(`j>A&}NE#+*G
zzDz_<J;uNirZ@d1U;=IG{_9WO(tj~i5_=-`dG&$TeGU>NO#>#eVU~C~gg(2N9Qxux
zuXf1<D3{V?hvbSsyAAS|0KOI)-H8&Ek-hD;*oD=mJ{^&cv0C7HGpB=3zoWm?@7Yhg
zUFL)PUya$g#?#Zl;50j5)tgY;jamz_qh3Z@E3=!jM)x+V&mSb$Dk;ZwvQQp=G|gcJ
zy34dFCY$0H*FP%Z61picj@86RQ+mDov$Aedlgx6_;#V7UCi^`?Hl;(+S2B~iOL3>a
z%+Y7O23#W{Cjb7hdz6@(+lVXw_O9ETvD0;MTwU{{axP`SxHf$K57%$bKA-Lo2TZcN
zK$A!=;R7fCsqYGBUT9SfCvIjqzG4ZClhfn^rr6?e<noX4Lj5kLK4xiOGmLM`n(+7e
zTYe6grA)h+wZ)~gUC20cL?<cP=Kv?fojvb&bMiRg<Zoddf7_H*OSMV$mVOdu=<o81
zPIj^87~>HJmB!{_ot3%+W((@*I|mnxhVC5uVv@^LfJODaI$;zQ{n$mK$PlR!)*w?^
zrRy#C(Cc92B+5+B+=P^{L-HH6%A15Q+MYx<hY%S(NX>A>_<m`b#F=&`4@^ut*n|zE
zStoHaD&eqNKBt{{iijo_J0`Y|lJlj$`uYuRKpkm&c}z=$(6m@8>B0hzXY<0PZNE=|
z121WT14dKiAMii?^wa5wLVTv!+$l|%c?uG&!k=0Wt(!Ddeb2loh%=f*(0(`xP}1^F
z<q-f{3bS~NJh4Oqu1Wd$ck3F^>&LF(v=2fIt{y0@>lG4*Rk}BAXl%MS>sZ;Ph@qsF
zvyL6$=k^f_5#1mwK@g*rfz7qa1JtC*aN42_w)PMIGEo~)7hZWZidU4+^qNOm`Ny@%
zURMTgKg&%?q|~+1V}w#7Z%0L9#UzEAr=^VjsH8W5rt20cAC-rR(HBkm)*D-TQ}+}d
zRRaPtx`i9imw4(7+}b&qewatROmdk!>@76t=G@)&O-au+K?&(}mCmdzk2qq9CYW8&
z_v#-jV$#zyl%n?!F86PIFe;(wB4d&=SnU1_p2KfpkzY$@9#0Zm+gQ^2S2uPRxAljc
zH=puh!TDv%gu2t=3&~dSpZ1B5gfLK&uHT-gA}Dih>$NxjK`RsnXmg1>E_t{$6bE``
z_pn_3yce;p^MINcC)9}RE2>u4^&&{xFP*TU$sM>i?OMhusZqE-!(W=MOtBIq&jY;r
zr|Y+K`*a6H$^1~eY<y|etJorpEY4}<FAgzAy(~PrcFM?82HpABM7u|+FzAKZP8Ie~
zRfT0yOiY{z!ezMcNDYAMO@x8vA(%CB0~xHFBnBcGn^3R}%6k(7hy1h&Mk<R;(}$zf
zWVl1|9TmI}hFReki8gPSMrL_s3%o(E!{4F#co`CR|MOlQl2^IE0wRFFi0bNX8ffol
zDer5}oWnpR-j<Y(sXABhi?N8kbn}hGyMX3;x5yp5Ws&~L=d_#oNk##$KygXI_=7!g
zN~Ne*Mm+$@Dubd6{!AXAR)TMu@U5Omb+wkqqlCyJL+o`QBYeDWfY>AvYJ+6;Z;wxM
z5IE_1-s(8%FVGZOhPakC0E_mJXhWdLpZ{EjqugMG`{J)7$U6z>!G}sF9ZgDR34vbV
z@=G&sg8B-RGD4bY0zZpfP<1HPL(e-poPH4Qg&N2t!@{KLTxwc+Uazaa6;{!t-R?*a
z>%iCH^@au`7!d(#h~pI%JHuVCg~Telt%QbU^YT9oIC#@f12aH2z#opPO^(Vo0yilT
zg;jV@hE-#NE(-EWHtC99{Z5RA%+5%hQmL82R~bmj)N<udCYP?VK`_Fu|KpH5>;R@q
zm1MUvGFbi=6{oGt`*N*Ib4@mr+#78*N<Z7i`kE<EJg?3>oZnmvnct=a#wS2MO#0V%
zjs-7Ho<G7mFTrzsB*oOeZAKwnPY*eSL#cUg#Kb{UQpRI#pTs{5YV-m@Ju7*eVnZl;
zqkKJZn^|5QQ(UUXj{Y+|$Pnz>AUf8ZOk%XQ$#Jhn3RYD~R`4VXFq3x4{VlECsj`Ck
z^*G8Ay%ZU8y@P5qQ)^uS8Sh0n{g1vh&FF-hQX5_rfL>>e`m8b|N-ZlMFCcY@B$F4R
znQCLi2ymYXliPa9*Nf!gYcmCf8T^?6)@V9v5kFAj0FHx~CX6W0_I3oGP!X+}R$eXC
z2TS8mmxjL~Vx#jy9pxVL<@+Ut4_{KoliWo8Lk!szsgbh)7k&P-3}EJ8R;KTWy2P4&
zF6WQl&<T)`YGp|Q4MY9Mr9s9A+vi9EKlP4mlE~mhK=01>#{Wmw+jU2NR%f4|ASZJ+
z>p6;Jf*XuE2Mm<n<<=W2b+>Ws$JCpol2j^DrIK1Qon<hF<S{QHBu*ScLK6saW}F2!
z7#nO@E7iT$9L*!0&7ATD<}1wa+V{2Z>da&%v0AG7|L=R>``)kD-rIYzEdlXUUHucw
z!Ryk{idfulqyqf#W(kBX(=P2Kyil+V5%d_2=^m(~1Kk&IAatnjfMX?brJt7i>da+=
zDKR`5MQ7npe!`07ne+y<*_BdzAz`Ia;eP${Bme|2zVuSIl>?ok+>6x3#~;_HSUA~C
zkq3JJN#G19R^W9*r`Y!UVa=)iW%9z!BihsWipf<v&V(=EzSBwPFyL2Q^SrwYJa3%5
zK@udQndYmL={pb8OOQVp3ntJO2PI3ZxCU{}(imabP#*1i=Vm^LU|R+z!Tb=8@gFqK
zq<{#hExmqCzg6TwPMNk%n7CGPajI++n@l9_SD3*zE=Zc*XW%9>h^)3d<6T+O`uys~
z(oF0SH}YMGSzLTXhfQE94Um8x5k1v6Pfnu=<?7m+zk~D+;J0_s(roFRCS>6bcsEL2
z3R?C~|B$#7qP3pU%Jkj+(B8a180K&P>JRz9+BP*twWw)nb#2ic1Dl%B9d$%4Np`FQ
z)@#+eK(svruwhg^iF%L8UGJO}50&=X6RCRbm>f}_!(G{!DG}6gC`Ksg;a+*=r~_C%
zDIxdpgX`CP50KXg`q0Pgh;9t@z+IvHOPj}(36rx4MI=~Kky#WZIb#9%Ejlp2>u~z+
zS?@H<FA1NMJJGbb_xb-q%)TF|2Gq*(vr9BoKY{`$^%{jlbRaeD8hdGzP?4YbG&v^{
zg??c(d>TDMO>*4*fB@S`zTx(;n&o3gV$(g**S!O78#AH2lrEv>jn0TZ3nrGkJq(RX
zXwqk?zk3?+idqK*_S@~7dZ$I=5CE)moc5+~l0`~)d2RZE>5jKWJS7~D&F!Y7$J)f}
z!o0yjCtv9Wtz=gN6(xjE$B>eKFcGkCwNy8`HVd8zv2*h{o#Y)MH9~y!aW6%Djr}t1
zAQ)=Kdo9K11yHE?fjDvc&PnOr?t-|G$3%Oun6m$;1huBw2}2GVqSBVCWdh(S-~d<E
zaeuA^s2J}7v8RZLo7Y$rQb6y>2=a4k_)wHT7Izd;vPrr~5zbN6G~aS~WP^>r!WTqb
zV{@pf=$MLF6`w2t+JaAXrx;13|8#4A@6m){i;oBkS~g^ER0>x;GbtZYXE&1&0ZZME
zzp+fSQ0syxKjMptTzomg-dT3r0ZLGXeHnh|gYHh>Dh)k>jY7`6p$eWCG(FqeOLRNZ
zL&YAZH<mga#f)=*sdpCyaV>k1N7HwEeC~K|@J=Zt_4#)$AH4DK2c<A5Hu)j-Qq}1b
z%FJxGxL-{p)y^@GMC%V+6z;9qMl%SjvJ110mCMu*0fzAKhNE4pcW~UggxEx0lNd}#
z&U4*h{V8tx%FuU1BCZY+elSm#I3q4+I1YNx7+&z!uqSANRzPW7u3l=jQ#$cg059m^
zwB(2xeb>ea=DUo-aFtD4@@l93ZtOuoO0^mhGXSX3Ak7L1MYbrYn()AlMzdj_X)8Vz
zYMIK_0&ZjDku-`ZPFiS9b?&Dnl1Pd{K8j>vr$Akw;~4$89rf!Ek%`v>W9yg}RsWsZ
zecSzJDH=M8B0j1O(kxTpU82`oiF2a6-I|_EUp(g1V0Y?<+<5*;-A=(FXQj)g#}Wn?
z?O;Q&!<FWeqhSSJ{S!R$E6e*1fVFzU)VKPUfJ{8XH{tU5kD3I3=HsJdA$11Ta3wzf
zD3V2~=31M+3q&l*r(*)KnTO4>QM~D8%pn{6CeEnV)8b@ctoRRv?n-ns*mRs})hLdX
zdh|w{Wnh!Ar4Vsu#mTNQg){~mBh6nRrYw>ILiL}|`T)FVZUoR}(NeyWK6lWqk#AM=
z08aC{)18WMi8;6s1#-4hXCOE6j%uUXlSn@KZRFFa6-wSfc1QK&NALYvQX53<QLp&U
z(Yq6~Ag<f=N1=0I5n^;%2nqG_JgF|Mk<F{`%a}5;<z-p6Bc~LZBi5AUzmpX^ZL&Xo
zVQ{nOl<Wm$C-GbDQJk3=lzU(DlShPYXLnrg)6b<!Z_oLcs1Pt0laM1}0)lb&pAJAr
zNb3as=m(Vcgjl?c)Z&dRB~txNE4l~VyLTfFW_4>v-y*v+OkQbWb*HrAZkKo@B?Q9K
zqyXezCDmyXm?i6)tc`R6JK;TPD4LEc5)8VRDbq`hLPs0i$(Y~{!Y41>^uC`UkHr=X
zt(aou)JbSqt$ww+(=-ENgVPsLJoe=CIbADa0W=@EeqFbqoPVuSpKE1_!vG_{u&E|n
z`gz3p$vk)wxF}hDb1VxVFC`>gDM|H*$h2g<^6LKe%f5ID6z6fwbJuHnccVoD#UElG
z-Endr40{niRe#{jcYnM{kC@aME987R6TehOlcl>vA3PA!U3y41OypE)4q~QTXwOWV
z-SB1nhkIFj(gqG%=KsJkFWq}mRBmi6FM6RHGJW(rHU2cE6v8NTo<L$nHWs@(z2-3U
z{N(|~2Px;fM81x)7MO*H!iPVS*P}a2MQ_gy2N7E7(6V4Z5-nf;vI)@O?P;AT@-DpV
zH~f@_G!Yh!|M}qC^zk@8Of#IQRbsk;g<wjg%P!bZf&WBoQUwtFJG<wfYglF=-^Fax
zpM?@ygA>KN^mZbSZo)*TybL>#H1#spa8WQ}f>Nlqymb`FjWXRRstxk2Ko9|Ko_QrF
zEzL8r01UWft>cTtyizaerm@tL!AMcggjI}BfB7~gvJ7-IB^i4i*@_)bBUX)fFw<d_
zAy3kTl~IpM_I{dzhH-Sd%O@GdR4KC{;r@&+0iH%psv$BJs7$x!xf%QO#-fcbY2jGx
zbvThyyTcHenwygp+aMs5RrbZo^iG<%pXRxn2=YL~GS0N0%*QQpbeTt^tx00(EFXNk
zq{1orDdpmH%Se1l*l8Ucv9zFNJtKhQf&pvQ_vy~4tAWWMximhIp2Yk5Gw^f$=g(ev
zDeYWr4LTy13k9Ao<EGCJ(q$v~OF3!HLT_P-){}R0HWT=)x359UAplW47Ngyx2-Ryy
zvUzfC2+%a8EZHrS2?+U7LTfOvhbqdgrwzMtDq8#E_80x+XR4QR)(8HkKlTQBv+mva
z^dI>sxQ|)UvlFhhS#OP|K8K|w45<Go8=k4CqZG+{Uw}vCzK1XU+(m>*<QJKODC~0_
zq0;^nofw!@t>wH%WKQ~L!wP>WH#>rHem>D_YR{?>%@A~EA?p;W<xf82Hr5%DO>E+j
z&TGWM!t9lmE25k;Py94|;7KnLTq8lYLn@ojx&;YP+@l<VQchNH`4yvg5uG-t5eQzS
zOLf9_#(yr<m>lKYg{_-<e}lYsnVtnxzjz}|hhVt1#+(oBc=}UqZOG)16d#J%mt)`@
zPcy^y40(Hf=lVN}5bLMY;eh*UCc-cV80vq>`m7<vqZ%B>15fA-vkEaq^OZv^qrDoT
zLk8>dYD!SH)g9#`2vDSyL$foTD+ItQsn0L^_~m-Qs<l}hnx4>@40NeA#jTp!FI*A`
zcSggcTnsY>*7QU_HOKoqtHN%Umf<4*E}j;!Z`*vUn`*TaBLlSV@!|BtTF*DXmX^%*
zKa@G2^Xp;YEM${1s%%T;7ZTd!o5yD<#s^i4ndqZ2jyy3l&G?)x>TMh>gJgZP)UZIi
z<2ppR6Sm>f`l+jO^zyixjH0bFpMZ&^x*zg?{p|n#x_|8TCbb1e7S;J_h#IVO)2E#p
z?W7g1DV=%Z`eO!YIdXpd#GAazPf#!D66IkJopCGYykj9#+TE+^B}&8qPDRDN&KT(t
z)h{ySL=3#$CI(JeAY`$=qpN=JrLTBp3mF!Cj{e0kPejfer3?Q2T4Q@rpF<0sBnj7`
zvDd1W243re+JI#;1BUu}6uy3~^u?9k;$smkCQp%JCg^1CpMC~Wgi2+8$PxNT2i^@3
zM|eQ^(SGF#$KY2sTi|d*dC<uqj+y`iQE_q4OMl`Iu@nm3wFk|#%00BO5=DkdDcBR(
zu47&;&Amf&S&)!FQ))4ddmzeXI97oedtL3xUQz#-40g;sr#fZoHEX4|hR9XT8{*d=
zM{VYd9}|;CRyhhsmO-LneUu?jpZznFO}1W8N@;4pk{enefQoGY+N^%k3QQ%1!Rmq+
z#6%JY5>Z7b^(y0?K^i#4GEE|u)V#Nh146v47SAJq32;I{ZH9!i7o~P*!NovG<6Z~0
z%5oVIuJKufX?4IOf=~Ju83K~^wU~vm&XbH^M_Nr%<7K_&WIwa2C+aSkaTx%YS%!F2
za|tKI^P0z348ADz$L|GVIHC%s*gO`LRs&i%t{o;>baq>QqziAihw3Yp2?5p$pI*CD
z)3svjAyjL5Du<9n8%*zyIx{5NJe|q}BevL$`35J1#Y|#}QMQ~hL8N{EDC5$K<q3k2
zHu$(#hOc-VNQi`ioJ%+Q*NnT#tAXjk1B39iAMz1X45SpI83n+Oej;`<YiK30S{18O
zZl&x&&sXXKiKwrfVY&{22^X>Gsb$oKdoJZJom^^zZA_@_T7ocCzZ&`5`uHdRT!;MB
z3s0CjfTl~gX;UY{y%wKDGHaVQC<boM8v=9%^~*{?nlpsZB5{Xzjp06S=wZSs=tL*X
ztiOuYrwBv}iO7hUjC3Chrhf3vtbM!wwvZQB@8`)WvO!!T1fz<u@>mFVYgfyQ8$o$y
zQ2H-`P?KvL?YNT%6#M&epeB_5PydImhn%VZh6@`l5G|YJ<pQ$8Q_K&5ea57^%=Rm|
z*01TzU@iPttpX7lhAUPzb&4YW-Cg}2hbtS?w5h#Kw^S8QH7<rFKMHG2HD~(jSs*si
zm;p<=v!DkN08N7<Kydwzj+AKsf9a)w0z;c7Jz7V5nGb26WalV)<(~rQD}Bm7`6*L?
z@i1WuChfymx%#E}@OJQ_)-XPN@T!iI3rn{~-(FrxT$CJMn3t`~48sx%%djH%fBF`N
zU3tbZnFAihzzu}hWr<Sh{1IE3f<EdWXaVh^>|V>exGKtkM$g<iF1neELnH7Z#`yeW
zPbR}rNOTod92C!|WAR)WkHxwHkDbQ!3tE^J(T9SUC>6`a)KovGc}g{7_>e0Rb#2UR
z#;Lguqa{NnG-|7;cozCPrU3;RdRheE>eF=K+8JxO>}H6hM5ovpdPa{X#GQKiV)~9>
zHb*ZZpAjbDudp+|>!%_&Pxr%K<*9Qu$|fJ8il*CG)R+}o5Fe@$a~dP3nIDVLu$r7M
zR2!iZ@J@ZZaym$=6MVv$^Wqj5G_PkXr<TTtEV{j%Z-9M-%1yntOB_1*o<Lw)4AXbZ
zq?&DRt8#{-65Avdq<1Lj@4PAN=R$#=le$~PE!!cC=XEUiCsq6_lKXa@`g1br5tQ&X
zcPBUvKN8y<R7EF~Hbe{{EbV9d7NI?@M9k3-w-nw5lC__wIJa{rHcsVree*vAFaIv%
zKgGR!@9G0Dl2$Ys;{#E~t<@Un43jrd%YLu;H)M{1!0PX3{rt|FdbHJ$tTOAg#InJ#
zEr-PtxIx=d5?FpJViXgrhJ?xRj7vHJJb*SJhc@_pGdv<mH;VC`c!K0x)XWX5K@2mY
z>LRA6FRO+pthXs0>>dGyoq?vU(G{h6kU;g{PTtfS#dUmdoP6JorGBD$NwgQNTT@tp
z3YfmSUqBY}!6u`uY5x~;P_m72xvyW-ql9s*1yv7!kO7!Wayu#Fn}7-!3y`+;qtl7>
z1dxPSBYA@fht(d6&2aSm7nC+aT+{|Pd`#N~<?oXP6qcbj=K3#v8j`sx#WFu?bSjA>
z2_W+bUTqu;^}nat+>4I=m=Dsz6~layQl3gUAA3pJ=HB8vs15VS7NR1<O|%dox9InQ
zyN=7riPNOGBC&8+I){WDyhjB9YscGCHKWrO3q+zs52F$bC`2k-1{DFyaR{@tR+!#b
zqEtM+wb}IR_fVq#@et+btpuj8&L`etm~bYDWUNz`Q2;8+_DaEAK6o{3ktng`=2s|n
z)TxeUSaKv|huH|l1K15nHMW@Wc?sFTb`hd1sYXO9C^+Jv_;re`?balG&4>3Nd^-;o
zT_iw@xp=3hK8ejI6&LY8_A<tM0<CI)la*PcUYNZ^9@;fN$c4keGi{hah&7ZsA4-cc
zF^(1G_1_7JtI9QuiZZ$<cF5_5j>UQ8ikRdZ$O3&W;W@KRB3-rhnHD(?@KqU!E<;&}
zM%RAI>LzMpg>kRrAXlEOrW3h-Iq#@)G5AFUuW4PV$0ITXJ@sm|3R4qfiG*dT(3qgk
z!r=bn$8LOD@A6Quir-b3UFoc_B(7i6BM?V)I}9=9)zOE!HajO1>A5XLWBu|qLi%QQ
zJcQGdr9!442(_o{@09W|+~?U?!LmCx1(Mp--v*~hIjUcZ`MH;>kCq-49MHtKA?W1q
zuv3XQT_&{6WyVZse?=_J{N!Q`sqkLXi+2Tcvc`ZDc=>cF0PLehW5Vb6d*ckPvWiI*
zUh}L{rhp={uYi~#XTn<a?eSs3A2R23HfEU{-j6WAPBckU9zuyhukx&aAilf3X`+IR
ziwdy=mxW=!xN{mn89JV5#j1=50EO2u8I4sFSz{4s4}u6LDAq&)1hDu)|1NUCn^yh!
z)vx{9U3?VnV1If)3WCN%?MZ^?vXw3+OF>eUkxldJzm<XuO={~dK@)z3CMZ`!W(i9m
zt;^>0O+9;m`V(&<ClaQ(nEpr64^cXzXkCeNM0>f6eRNxt7K$On_pBDrLl1r7>BqdO
zIhy5}3b)iG&CD||#b4WhYoa|X)8AIGo67haJ!j`4+xTNNGuD7BY8B*~Z+z<175nr@
z&bC^u<GbodCEis3s|Xz(wD!{qDT$$qD~Y~!;)SCYBV8jv(`aU6#H;T;V&~8IV-%3P
z<mmDHl-Q+J$@I&J_H(8}1fiC+^Ej|szw0*VladK0p%WHXFL*g00+56PWe8rTe~FK=
z-J{`aT#pXn4?P#13`(GEk|5V%ev*`fyBfam!VPoyb}M~5a<er(7#26yoj=r`rhuhw
zMpCj7BjED53G#PSN&v1siBL`-p``_>h8{_9nsW1Rl=6dkk%Jel;Qr={{~A`$ujBul
zHh|X1hn)r?VDrojMeyhOmue$sb>Y$F%{a*-4ONTOmz3FSr|GZ37-l>XBs84TTct8Y
z01s1qq><nKaoGbA->42W)fF$8<bVY{rqq*rt!(jp_18R7Xhu$4N7g6YXL`;Lqx&7p
zqC~CQfy5bdbo?429euV*)1^)nIxh5%3bqGPUQ0|Aw#C_cE!CMck=i4Hp!712dV8%-
zau7j2Vk_J)-xMm!R4w>ODvI{*U?;QknATPr&;o!e18=R<0UZf3+Qa{%B}$#%C_r_-
zvvqz^e41kB9_9N9xFNyuo_y7b(21jMi(><zim(Y4`0@FiYwn>W@YaGK9s)Y?^jP{?
zRd-H`nG}XcRD4yJ?D`db^Qou(wq4S##C#<@W)papW@yXB1#zMPj{$iLCtdDKs_+F|
zLOk$rs!mI37fk{}9S}^8%tHN6j4~x#^oVdK)p7NUw4(V@cEr{1Y#u0@rnyAolRv!t
zk2xd;tT=rjQ!o|`U)+8w|FyR}EB>42I72nDOt=j>Ar_gAHJ<Dt-4;V!OFETMxXHXp
z9r2|eOd&^DuN)Gta%k|VTo+{`+NAfXvT#KOEK&Bfn*#x~{#yBoBSMm)Shd+(7It5r
z1p%O@H$OR6-@~Jbdu%t>wGIK+^-k2|tgI>Ul(_X6_F{Z}?#3rRkrmRI0b#=a@l#Jd
z<@>vi`vRJv_>plox*)TjFN_H`0U8(q;l@9H@7_oJI{+`eKpW!+(2p#H8|~P_2GhGA
zeyutq!~ut+gfRn<K)hiP>RmeHt6%ElPe1mV&sp;ghDT{ZkJV%?WygH&AhiGhg2~1C
zhoc%t2Zme=m0%jGZge8Cw}?7<?W-t{l*~IINkT2a%Tdwh&ye^&j~4_mHgigtZu{mJ
z#ft?K#V^W*thK@d2R$#u23E6gWpj9miud;5z}K~7ZfB-q$q-P>LqD+~a)a_wu_~cq
zp6^GHmm~|R@6i}P-Y1z{;*c9_`Z(!nVs?7T$v&u)GFA_!QVoQH7DdjWRDT{)kpXe~
z)(V9jHCaJzS71>6&IxUbchl<G9%f#LLRTS$`&<UCXVN{+l<D?xm|=*kQa%35Jk!=e
z=oc_tl3(#X>itCPtr250K-)_4kUmI3_BO!a@A@p}DD*AeIOrLfS?LY;Khd)mhBX>R
zA>!mb7A62k5R6cd{!50ub_QVmvjh3ttG6vE7C2PM;YF_zLe&c7K1+G}Lv(x**vq$w
z5^2hGxyxyXl%a-2I^dM_V(8J~FiJ)un49)IwUCJnNK~}~PFm_7)kJ!8DHg7G@{sII
zT>R+=ZdlLxHiy-bQHg#+?=h*$M&M5w^dX}tX$=W~r*wiS)`G%6dMPP|u=j|I=P58I
zlDxS0O+tLo@22h<59sae^V#9Y^n1=2osqP`PAWsq_4td{AkZw_jOd*&#UCn#imJF9
z(>G6s__g{S+uDdJQTkzNO<zZPrutU~#llfDpgMSeIx$7QkjgWTO9PimgJA-$pPK}H
zLMBs2x%4x5G9{@ihPH@9i6`?T3xlA3HZP5SSKFI+Oo9GB-}RaZB9Ue<F8Nm;_g8pB
z3{Yc3bK559X5{hHYd`>GFbDINKv4}#1mIpY5H(Xrs_8+Pbh^bn6$3~=QBwiF%fCCS
zGe^HLQ9f73ivBc5p?7U4wjFZ>y-cEg8TWng?dkoRw<P>p8f0l1ie=hH!e#~(iqNl~
z*(OM>T0)L})z7ri26rFNPRC0Wu^<SykZRVcid!S`yuC7iM}ZA}9@DSVrpy_xC7GZW
z${h`~VJV=38c3)?+&F%2bkFpVfF5c0{ec2X`bix&Y8}{?B070D&lr)XFYUrnnal_`
zuU6J!o#9o1SSQeFCZJ_VoyhapXFlnzaGe0Dt5J?WhB_>gkF~{wBC!Lv0SP1J?X!FO
z>CcLvXzUJKhgOw<?38Nod0w7;J{eVWov7WVc?;_z+ZaYeV0P@kC>%@MRxf5ut~2hH
z)DgItAUAB1e_!iS>tNB=k;N~4rzJ)m!29|*-uA4)djG%=MpQlgf&P3~eNVjDbzEwF
z8I29hPLB;9ej955OFbS&JTClR<>OStpkLBPuORg0wMN|8d(qTwM7A%U@?K1=4Vx3B
zWk&9|`<V?EK2@>TPzPA)>2M7C2O{pR>JrD9$%MRYn9*0^d38bH34V9Dk-hV{5X?C7
zwgYOU-IC0cj?waWkmg}xaNrc|%&=jqY4G>(0mq5#AhKH_m5O<a4yPH}xzo#90Q{|)
zaBgt5xUZ(f7LK}FpqiJqT4DR$T3uU6@gXnh_cBqmemar0<z`?T{+Ivpscn<m^D=P}
zPNPX)JJS7ZsCh>(K?ab_v8<$8glwO`Okyn>(M^Ao?($hNrrA~AwgaMQ{_Y{h%JjY&
z?tF9!6QLh;6rcI5pNO@Yl?}9`sl&flhHafp_`fl##~(i*?x#-BW%fqqG4E1kVH=-C
zzt>_YlKRbxBy4%EiZ)0N3`9>dnqqsLqnC+$qnJ!haKg&9bt$i8lBJiz2!_m6TjxYb
z9XD<i2*W3CJSKjDZR$jEH8u~26T|IS3Ot7T{H`82{a|0QZa-TyOT0I#XKHdtBA>hO
zzYrkrWk;M<d^zom#_h!BycJFJzy>Lu;*iJY^%yo<vMZ`+=m)pUE5kg<0z*t3F9qkn
zlhF#maA;AtX3!<OaU(|f!8stu#C#;g5)h1qrkN*xj2uM&sHS0F%6y9!e<iy2;G4;D
zB!AemStvjTk?-&o^(|P~X!s}UX!T}I1NU#XE3OTvJ96`LnD@Rcu!U;GXNS5LwR}3N
z*75XbDJplF0-WZ$lFTKY{OLc{D>TJoR`fMWL=LixM2*O@F+B59=&dMrUyq?k^lko@
zeow#ya^!mCHAP@u<Uu*=guQvh_$>X}>@$U4JMDs4@|THngqv2)Iz_{fn_9$X8IMU|
zPPSB%6Hl!GMFO#peNoIr&2`d3KmU0xaB`&b3Y`(1Gu-swJjE%fWy7Nw5dsI|NE|@{
z-Lo)mks`<Jq0s>;<`YNbBPb&#DdcAXHw(Q$rd8Zkd32PCqSkm2tW6_67yc}j)|U)*
zjSHp(X{?&SS(#wSP`O&-hgqx~O872FXw6yi*{p3Yi=4z|LWMAqI4&@Hp^jsi<JsIG
zOA1FRlQp@Gvx1n_>f)=?Lb|{NL#jm*JlwqXsI+%|H7gUUTTv~Q0MGgifO*on3+fc?
zB9A=t^Jh3hr-96vqi?X^QHzIk4&`BWv=R0JI=x$yIzPlPNr|`?6_jR;6>3P1vb>%v
zN*D72XX1A65Sq5sxon*e%Z0-u_xraIp?x410m(;RsH!2lzKFgLzI_<pVn33)q13v3
zmbj^$M=?DD^VkXV)#GTMd87u-eqj@nBZwEQo<MR`EQ1AeQG^P{xx+icUS%+v@!HAy
zqu~XjSStdT9ahgo{R<N8Bu98w=lE}>wa$J{KSe9d23+nE?pGHeaUb{RG|NUIHFq=X
zc6JCIW+Pzr5ut-ci2Ygmb_qw=o0OK#sr0KgiQqrV#7jtqIE!Kki_%40Iv?LUhS~J3
z41alob=m|f)bztZ=af`vYx=Y4uPR>=;Q}R8<&OG*`YQ5vx%83_2kEpCTaCbGTrCBG
zGG<OY_i$=xgN%sl1Nuae_`cmN*Y<_SpVU$WdXw*=0|9Fko^WIiec8aCl5DS0`m29H
zMi6863c<p&Uw@PmCNLqj_2v}?Q@Zi~sMbG8dzn+##7Xn*oD?#(jA_wst*@p{g)OhJ
z0@+$E5c!{GCs$FTv*0(64hnFPr2=QLP^MKTv0bnL$&t(b)0Ntoev7e2UN=J=81B@|
z5HzV^jeB6v5kA>#CCXhMDy*dwcxLPfvPfqW_9&WlY;2h)`na1uI@Nm+(bm7D-_<y@
zc3BD4;;+D$36!OH&A$Zcko>UUms0ZfVle9To$h70YU@A?M5Sar1uo%mp;o1Yo_;!e
zpMvnie5ymDj(`E?p@4E^W~_DFdD;&}w*@Iy2l;YAAtp%OT2C2>-taL>SKL}<%-ITY
zlg9=Z;McN$A3b_f=6@Ntq1hh#Jw1w*u!>GnrUDMuq4$PhmH88fw*N|qD<J_<y_uJ>
zgn5%6)EpT2NA3N@>pJ2ppqP2YZE(+!SQ2FyRUCwm&>Y5J8sE!kH*=24j}>Y0&HBUF
z{&)YhJ$*OXo{kJ9&a2)s`m=1F+Ak1^d6ID#XcGk+TGIt4yAF<?ZhnwZz!l-bgv-R(
zJ~PA<GNY-_6k^}2%i*^vzL<I&X3^wfmU6C54%6Dx{I>pG@%Hnd)B;w(Tl+Z?s(ZP%
zU+N48sE`0inEa>Ak1&x5Q4&JI@&0c9LXD|@cm4NQKhI2@&_!R5R0QUhUby}Rt<W8;
zqWLbW?lc>W5MaSj@lsSieLk~F;aEGK+}AW_3Ck!<X9hdnpT0T$F^CtDNZ~Qw{@V4c
zj+_@^7VeGoLR1BsmJ6iG$V=%E4j;Vg_jB9?UnZe=S2vnFCM~U?ON`8==Heegs{#YC
zjJM)5q}30`imBGx+lvO7CB`gmqpv;u+Vq{6l_ocE66iHx7k<4;ZV5Y<X1jlR{d(A!
z+mS>f5;g1ifjQqz&@Jqy3d-NdrHN7l?rJ}<EzR&1SbGT|f!$9%*ieHmZEo?n8#64l
zH0;D~*pG7QM7gUEy)Vpq8ZMgmS6Y|;wVLXk#+xRA&F}X!sQvJb>05eai*#u)Zy~*m
zFYO8FogUWM-dNpOHjRN>*L0tvJ;xakjcbD<NFwipFR5)Kk<KV63^$+ZI%YF86x)3{
zi{q%(5i+Zn?FbbhW_!Ou9LKcIy?!Ec;cP*p{?7Z`q++VX>$e?aJ|V)WKSE+GYy_JR
zZbeE3lffjKbRzXddM~dZm}4>n+WXS#v-u^?TB=X~EcSX-SceSxx|Zmtwc1g0y^xAe
z0#a6dNo2hADHAXTyXaY$vZZC|m8edu+fCuQ)dF!GiYIV;M_Jp6^ey#KIy3qU^uzK+
zkk6`pac-z@WyfCBMxBogEF|s2J)DUs_P{qPW7`NRQMp#6zB0?Qz?r=4(*nIn>}lk1
zk*YeSI%a7fV6>^kqfU$wt3H}-s&6VRe{WB<Ag1aLNkuh*laQnPa)#S+9EuBu%U>+Y
zM__dgPYmWx&eyIFt;sI2&eF1lKLE`L*c3m~oe}ryIH4v@v7jz}v|u=k3vrRf)Ie1(
z^XG{9TFY%<n}bZIH&|&*$z#PXarj)mI_IG=a%T;d&CN8VT9VOi>G~B@l97HL>deyM
zqB$!9bCSW9J7lpBaUD`LVK!u%SX_(5Pcm%@7Mrh;2zVc{$pm@-N}UbqI1^z;@d^N~
zdIv(va=QAx4j>__ls?o!m(mr%0k7&RP)%vxCF#))rH-pwy3^bKDtOak-)6>;ZY<^k
z71dYMdWC>kbqAyLnQEOlBXWyjYnVd_Jl$2UAvomZw0NcWAHJb~1CEH&V~*paH}s3G
zMzd*wPWQqbDYpyTYM=Y`1&Z$r!l1pt3gaFaCjD`#vDRG7^I=3KC#~Z>VJz$;ICOPS
z)C&N1Yd)2CMmT|57=QWJ38429W*)UJsvuORXxPcaDiZr`xgzEQOk>gt9J`+xnj$(V
zaSb_B9Ra-o`~a{eh@6;TmYF^Hf=>F8c${}y54xPAd(!LPYzHc+0199L^(aDZ<Rg`>
zR=$NUjCtdGJ%!js*(Yw?c+$U}zDerlBxN|SrA{}4!{7kRj9ehOvNfVkl%0*zaotyo
z`(%9P_{#o`K<o5Pz!jD)ELm9-<%9iB^#5oE@Q<vUQ0!srsh4n3p^I}q{LbW&W&>=A
z7z}$&ToC_69BWrM{`0p_ncnmL34W@myxNF#+$DWT^v@fQB~+?$6i9l|<9oSZIB1a3
z>c&wUN6&X;4!NAoIXlb@iQjLjqc@3Q@f$0w@dr|!SRz``C7sKIRnatYZ&WINF#yYD
zNj^SH2VowU2&|5O4}d}6#|DlPd&#}aC}-xU`P(K7mSXDV2E&p{g_=IF>QyV4lNsC1
z(O!$ekKudi(bM}6@0(UJ$~&z<{2WR%_cN4;Rrj5U;s$}IAJ)+X%@ZKF?!~&BT#kcI
z)FZEv4B51<uI`84f5$5mf-FsQOl%6Gs~5usrhg~iicYJsp79?nFk#>-e1?>u95$0;
zduIC4H#}m4AgRr5iZt;;k#<N~W~YMRL6>2(0^KnOqd7}0tzLbrPKTzaT?w|t1ddO9
z-1}oiOmg=-;8`O3S@Jf)WQvhHrNfi_n7lFg)~6W-3Byd0yX;wys4T5C-uEq@>zz&C
z1q`gXxk-0Zu7>tfEJZJ7eMl}Zay#%|NLR&${S}7J9wicyiz8g=uY?VjrS`c`i9wq?
zjqRQED&#~&7(Bmz0ULRrYpNr>%=k|at?NH<iLofcuUcqCS9i3e0-gw_3SVQ=EOQEU
zX4v$T-sY|ek2@#`Y^{WoC0U?diyODSL$-?mO6jJNzZ!E)Y|)wpeJ*($cHVw3w|=K}
zah&DSO(J;5dvh8$TJB0;POCX+o$edD7h7Oabo|fXd*0i}aH|Vjp%J3fiSlVa)F;;}
z7b3!=A!f*3WRm_$l2`;*pffGCc|tjfq=}9=^hYdFb1pAk3Qf^1DlL$-{udS#gT>P>
zUmhNpejncaB0bW;HHGM<4_yf?F3O`JBdKL<FD~h{KO2Ry6tNxVlXN*yo3&Hctz%h5
zrn#!OYn<`w-GT#BOd)nwKys2zTw%zPUt-!->AJ+9z>!vodPk6jG}Bw>r7jD6UXCs-
zFzC4V{lh4=C_W{Ge&38D=Oo%9>1-5kgq>?cY)BpHU|##h1c#!eQ#dh(grtNQNiSZm
ztF}S;sSarROXg<kTA{)W6d`Q{C5z?C9K+&E$wvvbYQCY{<h~Y%*edFP*p`p%xOYxg
zJ24n)GbIN5=nFS)_yZ`*vsURGF~7O7w7TqFk|ZWRPqscp7RGa(qr8lRn!ue1?g%}h
zEdksi9LL+dzgw$<_4z#!1Ft}no5LE#Aoi~nj%@HfTDqB<ugQS&lwBQ`l|uC<C^|UD
zZ|L1s(Mt_#DDj6B@ddiIL1m(U7G^Xb7bO(_z1o<BL*BVKG!;;jr8lHbX?3%iS6_xm
zB)4irmNq_BUm+60=y_RBiQMBi_&28}TB3t6h;#}loX`@Tiq>1}1!A!^2t6x>9SlkZ
zO7+}_hoc8?#B&6^izg}^QtM}y_}4I}hQlzV1CD~n9>4KqymZF?XTxlw#+b7d?QPN%
z8A+zyRAp6(Eqx!s6^rin%5Z>NVqDzVQs)cXH)=hk8YTH`N=10aIl1q%0y9MniC4y1
zD~cg{S)vSR-Y#^P52MKv$aR)CaYd%@29}a{uFR~#asTdVq77r15>{}Ada1Y*<@AMC
z1}KRU%JnUEgj8Q?kK@zx%r#(__K03;lQjIQq2_jLOjm&PTjd3XGD(@m7%d=fW~)uc
zvYeQ4YUqI+h28`fJctbu2<e;HR38GmFb!l*!!U_(VaSkJ@v%?;k-o0UNZ!%oUwPir
z0JvpL9P93au?tCdyBEpyHs%UCTzi28?XX)UHXVFdnW~q(j+K+Q_#6YHX8CrgSU3eY
z=$8^h)zjq3W~x*zZV}#)T%VC>tkVTRzb1jUg$l`dhLW+!|NVYaNf{^!6)7y!WUocT
zw~$?Gb>!D?^+T|6O8^5Ta{>OKD$KM{lLJ3NZ!GQO=8cT*W?!hk3crUA{qP>a2`Q<A
z$X-t0L87IN*x8>Dd#(NhS4+@leYm--*U4J~k!oh8rwF$3pai~GH&%CaUR>0$aaT@H
zPqs_BoPK>pO|1UAfY$QX#;w3K0uTuh;-GcWj-$M_v*=F{{T1ya8s;sw09cm>M@(E!
z*d}ulc#6M-ob?ee(3d>p60{L4@^zDnD<U`%)@)rqc*Wn^Xzb`!g4pEOzlgYt6=L$#
zq=b5$z6{CVsl`sTjW<2ibMz2?DKJ*L)BE?+4`?EL44h8yqJ;(XZ@in$M=SaoyOc$b
z!YbnC`4a^hsxQ)a57F<fH82lTpU334s&<%J6pa*V%PD%3;D|UK%o>$G5<du4A<E!d
zUqEWoq|<^QSOa+ieU-Zuuj8W;Z}y7jm$xxvv?ACL@h-lQjs#>ZM!A{exM-)eU*^Pc
z6B)Fk2xfXe%}~9yTOfS~G95_vA~5<-zMzLKmm2+l2W6>^4+u%9*xPhfc<G`&tA|>7
zq;zP3R#B!EGt^2o5P$<Ay+T;(u%n%rceOjpj%TQ(#Igbbul=N`!=_!dzQ%SI0|ZF^
zypB)4-9M9d51U1Z9KpPa(ff{e((1+h4`Zm4?k|QLvP3JT%AP~*WE@k@VPQo;s>(wL
z$&tdy5kHk7BJoYg5XPEMb4L>>koLheBq_^06c9e%>M2*DLRJ@Ab#OV6Kp@-Uzp=&D
z8gLkP$ZJZ!=z;v@uwq8AF~F&);gJ7?Fwxp5&(X|WD@H)tN~!wLN^w7?mgsQ!;0-Gn
z##(@OIAJbAq86VSPX8rU6gmV7L^NOS0RQ*4P6A%3#2_{uOXn#-xU`;9HzPJ$SuiV*
z=Kpf(A5NI02==`_;%KG}!r1;0nGCPK^!tqqLA)e|%w~m~%K+nsL?0-KgLyW|jh7ZE
zDNZyTIibam7DA^G!?S}ywcMAU()A-7kp{~b9%*ZYGA~xYV!*?TUfR9tm#rl`@rW}R
z|N1()tZ!Vue)WEOof_5k3;B#@BG3`ls8hyOeZ~{q%zR9QgeXk~<@7BPm0&mOLgZA?
z<AUx{NheV22u!UUHjd!><oF_M=mg}v`n6eL9ZWC$RP$SM0I2P)*wXb!3~WJhYg@P`
z_!q9H50f-$ZSb}>jggAd=^@1k8)G?&#m1e!U%M}eit4EJ>+Ue~z}jzdHT*fbYFwLw
zT?MJkmGc|#u<$g@$8-?jlll(E1!oJAn}~rqiC#poyoe!Cy1(93$IN*7D7teVQ{kN^
zkZSwj?EEC1!qQNo47QbX?%KWmRMop!YD8PowU_9?%B;2$U0L#AWN8$WsC7t_DZf!!
ze-+Ng8nHOdA3P+C#bVo#-{W>9>)bDafNE|qM-hc3_IU7v<H8;dGU|`u<2ZD+W5|y6
zT%hsgVe8@7#e26_SCMNel-g9V4Rf#zAs;~|k^=Z|{gAx)F10V|0opKw3LBK`Yg^NI
zcWJH@?KyOhKAYWR$A=Ud$06?Z%3yReKlNBSFcYTi)2lr4lE~grJW*gP0wKS`Fq@nf
zr~()Lc!I#uarJ4lA@kdhoI`@73srP29fN9TBl=_U;X?Jzndt|71%R;m58xx^edq<f
zd3<^n){*4y9s7^i-GX4iqP*4&$8UPwjJV-zL?pQ4?J<B$%NH>)mDQEg3B?Jy;1eEC
zqNf1s^JYlag}2rjCaz_fh_hJuiN|9w_~iJUNVL}g(hD=%s^DflDY?F3i@@N~kw<MB
z^^8w^_C>$vm=?Td{nhc<MfZwprIeB*Bt~NvVrjSvJ(^Vy_WBW1LQBPa?lE;$=l`h~
z3=`J^v)r$0f<^PIuwx3OOF^-yJ@+3r=)aTIo?BGHsZR_^7@Y?ru>;yidy;y2X@scR
zx#5C7*N@<kf^?4-5CRlcjcs;lpCfu0Ph&uM>K?#-lSt~$=d=jr(n!>&gl9xpS`7@s
zq^8n4x=E*S-*ql)AFUaTb6%lc47JPHoEu05!=@1`)BGxDW;AHm!CrMdlf9tk_Tqe)
z#$-uKIB$7-^VXK7$QT5^+!!DFo!T0+8GV!D6;UjUbd2>jnX^cZiJng_hZd>|8U6D!
zA9-3&QADzg_#y#`iT}UwI}RHg=aJhTY6ju*Ta#d&5^mG+>IcM{-}XV`K>vGyQTN&@
z--xE`!j@DM4_?d=rW(j+xNx~A*@`vN@z;nEA{3Mx7Nv<f92ar#KPP~ype+BN-+##e
znnHXt4pCcG)v~sFR_=D23}&u6Cc;<I45dj4nW)Rt5A+lD7PyQ4-PSO>`UBV`rL0_^
z|FqN#ygOS1>HmnwNkfrNc$`%Hg;5lT&_9T{DWC2n#8yqDMZY6KD@htySZJKLrZ42b
zF;)O9AS=9u6<p#1ZN8hBLTVB+2(>*NOn>Z8?h#!(9;T24qkxSfA0yD~l>1D7R$KEl
zh=l>)%P`~N^n=9gPhYJ=Bg3Wrdu@M0r-@Qezp8jEMze$lj&l{yfR&2Z{#|0`@9D2>
z{jrakUx+M~Hkr(OOhxHPe!wW!i;p{CQND3d1;aIjSL7Axy)5D`ZTZnB10DENAN#a9
zbp@=cKJcY`GJMxs7uZZ7Rh3K*i}>lMAM-gbR`db3N+|QlV-Py3Qn5v22lEVFj(Nr~
zjZOLVA*b3*l1$VWB!0!_wi*Br?^f@AZQaY8103l0_m0E)eegy%TOAPG=tn?^q0>l^
zvtC)gc1r4I7~&Y0Pz`@7Cl_;HAL+LrelHIXG-2wX^v8=2L|PP%3r}Mas6;IgYB1@+
z+!F=H$^9HBY$!dL(5%wF4x3J~(JCIsb-sgaTrOL(Vvh>&EM(Mx!v}AqHd|0-ZMFA@
zIpLl_lbJ=uN}-#Pe>12fHJ~QoVLvp&@j<{vV6&B)hc1htYUBE4OAy7|OHwS?Onl7`
z>YM}$lbep|FHcP5Y;fDs_Ku!{HJa5pzQDNl2A{zkE6rE><IDWKnuTCJ`6HBtaAtlj
z9b;3k>;gjbZ;!#Z=trP63MlxCO=kX7Xu>)O)Qfy1cY+~<;{5KYxB@eqVjF7tv5TCx
zK7`}v-`woipbote!PWTp8@<}t2zkIC(g^^|GQ*5fXgA5=_+IaP`jh=??#^qIgp7P`
z*U5`g!K0gaXG@}ah@|wnt^?Lfu?5<#lp0E$W<UWmsXk~f@apXhlM+iM9HY5Y+Sg)Z
z+t1XXm*8-uNuamAjt*YLc_$VAf+0&u9<UYuz9KEz;g6`CKR~P}OQqCw-oIt6OP+pd
zDHsB2K?|aRn0e~CVN5Q7DkNxpo)n55xc$i^Wf^Mm*voDTD0o3(^~5iyF+wd0!6b|G
zeFB{-zC)aac8BN%p+`6NpfhusEF6T;5i$IBK)OtL_43w+36(A5>SWQD66Ggu{E@$K
z^<&*1J@gsaGYP5^*~$@%_ye8r(Tb=d$4H5WDABJKkpW0NPVgsqo!)T-?7Wi2X5kOI
z^Xk2bO~rnTu!AXX;!*k4jZ?H`?MzIiPYB~p0*VsMxoBl<az08ylU#qxhm2I5{<A*^
z9YN+<H~`Hjz^?STK1p0-f5^ZINdZ-KrJJ<gq9=(yAb%-cK1B9$=A`A!D{zS(h)x5a
zef~2s9w!0>k)#&~38KcU^`JX6<ZV+2nzB_DCU|r3Fv{ACCY@K~#i}434w5BVM#@sv
zJk+5J9J2!Q**i$sWUz*zLLrExbF|u0|0Q#@HlrU2n@CaWWJt%H7O#S_ei@m1W*Af0
zmev}}`V3wplEUj$K~=V!#pMsz%#Q~E6mQx9k<6R&3uJyXc{KfM?R4I_wWF^!N`nWT
z!6FhUTl7V1nU@u1uhXsMH0y$3Z{*xju=H#>SvL5-R9L}K%h1LvU?5XTLnr`ULVbKd
z9(t-6R<0G0k1z0V1w9ncC^#KW#MPg%{A$xhNoI-9wBk0hyOlPzUTS^K&kKvuojl3W
z>z4CqMtVOby&Ql7E7y%Df#J&y!XuhN=z6L-ok_hh-)Q*9tNT~~>>mjOWEO6og8OmG
z!ze-H-WS2)k&v%{RU-dfRWmwd29jV*_}0WBr)r3iBD(8$>v*6p_MPu_qn6dSxkMoo
z4qLn4EFQvHW_vh2fl<M6=vlqJSk1vn#r16}d#(4+6C%}~Q@&x=@vkbjK|`80(RlH_
z>sK>3qdwJ9rjBTEQ{hg(^u^%3PlmPa*)jN<5a5`a^hl{YN!x^CB5z>5M=v20g#o{;
z&>tQihCW7AWaOYZ{ebAbP68`H$PTY^yBE2|L#Z|y48_b+x?7(u7@bd;RfK8Z9VXme
zwf{9q9XR=!en18Z#V5o{bd>pX|Bu&t^&_s!ul>~dNMhaf%1`<g?5s}sI0j)mocH2H
zaYc-fCR{zidWUVi5dtF=<k!>*7Rgx{Gg|^=_OdYv`C%!}jDxA>NJ6$Gx9E0P8oVnk
zC2rLGRg+q7Qj!{xMA}z;(zxBwe<(!P$B}puzE@PH_Orw^uG5o<GGlqCBF6WmAEIpv
z#f2T~(WCasUh7hWggQ=M{=>^Z^kU}02jomk^kA`Liql{Gi)U(s)uW5hCJ?zjoEzTN
zGoe}O#|GOWe6j&mlNnpc=vKJtUwQE*eUwkU^m(f%Nj7{F!t$ngiE`EA3k6h3y+j_B
z@6Cq*?U$M{uUk3V)T8HP27NvJS}L7?j3A;qf&siWo1j4G+~{!FN$cTnC|1VZ>zDFC
z-q9XS@^SR%mdRpKbDO$Nzq84~#g>lR37?bm0ufT9E-8g2qEo_{7pPrZz!Pj?SLRh<
zVS*0?6-nQw4ZVos_Ui!hjpFJIXv_$&YC#XbPA8*YG)w%GmrOkL9~!n6UD70tirPAo
zh{n}X;j;Q%yy~<+eNzvjVTQsgengY3k~FNAt`>^EL9OGIKya-5h)m=XdYuTYmG(w}
zP-%NqbDRgQ82gM3l$4?xnH>IUTHDO2Ai#5!e?3PYpu18pPrn+rDbylV5p2-W3r!e4
z+KSl}Yg_>>`Qg_)mtuY=)1=X`PJHxW?4X%GhMo2^-LcDByV*S@11P#s!(WsVrPmBy
zN~DknpXZVFNnwoY4kCJ9CiB8Tigr}PnCpwtHw&5)@NW&`Li5Ru{!!iow@$BYs#!?Y
zh1@7}BM`A@_@fNSNt;XVhQO);U}CZ=M9@~IKOsn|Bi?0@B}b$mRsF)aE&cg+V{LUI
zLn&$lKz;{nBP=Mow!UK;Gt;P~rT4~V*iMU@z$zb0RCx9jEJO^L*Gh%g_20mefCltp
zB^lf{_cI^$fdJg;mBCyc@KnRm|H|he3x}o~E0fbh4p5}8csUE`mW;1pj>NPn9p!^p
zBOlSh$Jak^JpKu>ZUI^4YFVrh@qW_cKA8`FmYR00kT)Y@v#|ntD{37d%Wj7e_aKs2
zeSU?JP!Qn=u6swU4r5UdrEpXWz+h?*Ll2eyGkOu;w*vV(1LCWd&@yh**dcS1+=UYW
za<BSjh<*SyV&W49+vydc-+7vMpyttO`a>N)W2=|ltv9j(Qnwd9@^L_b2856Ite&`r
zurhIk!cE&?FnucpjErnddB2P(VXP7~M~H>f>6`wbz#Y6g8Gn&H9f*q4a6Do%iC%}k
z$=AWLrK2efNdHDXAJ0+!M$FiAxGph1P+#QW(i@)-ZV!r|^`j*TT~ESY+#%!dQ|V0K
z8a&KafY@3HZ^>#ss&!h3ImXVltXr7A<?n%@MzqDp=xLKzHF^XfXv+v@I%Xt*J6sT>
z{Qa!?TFrbciy;=az1VE#!uE+!rij^#;<sp|zPxNYE^AH(nRQd4L2U*cRi@3*JR}4n
zIfPQ#NU~gnYfdOT=;h=i^__-g3cD#Hmp&Gur`jG=Vf+t6W^#YMHadxN)4XZ44vIl-
z$9?qOUjMGGj3_K2rcE?VER9n(%<{u|@-;AIJuReP5+1ye#rdF)BoU0#&%XeKp{HRz
zRv8WUzm-`}@_&j2$3@uLZX|e-4$>9>NK`=b(GOfjT$;C^VYX<R=rfZ~Z|ezd;QbT$
zC17R3FCII%tgAhmWU(a%)+vE%AC>}fxa;8%wC)BfF6?JLIm}MowEcv9j!!t2IuROA
zh<Qipj9?bcs@d0oY<Uyw*rNF|a>(?}$N=peW5>Kb4(c-e8>Hs{sJ$$tCT0^8xSv5f
zu#*uWF94FfFIZimRZWLYn*9!<HyJX)b`!23yb+40=#sv$U)twWpZdeg|HF)dxFv+h
ziz_@6DDEgJkKsTA9HeYko^wD%Kx6Qtt{lEgs`<<Auw9agoy@8bDB=ziDG6%Nmrlyf
zFh;y?f-%hd(RsDHA?@&dum+^T0VfD%N<>0`UR-bZE8BrP^3_FV>94mN+xi7XQpB4~
zd%{n@An_1h6nBqcu6N89P||I1ZV|t|?CtvC=2yyNkiyM=Jx#XY`-PwQHoo0#<Pc_4
zYxl0XkF>fXkd>D?pDg@i8v-E&(t?Lizc13kw;{Du!eOS^W^z)Nee9|udrF<nQM5wy
zrcZ`h!-WEhHiLks;xs1bq3J#sxm(_}wi<zd@E7VFuaBtLtqHl$MJ~b)v}L*)C_;S<
z6Kr2pfmZl_iRwJ-Gs!fs9)p@-F0m6YTF5}?3Zo25&CLw&8?A9|lr@Ob(ACsuv@Yca
z%0p+!^L!2rC<^>P&m}e62{-VCCq5}65uAWi_cC1BU`$1;#F(24H}#ZTJM;R_i!VN<
zT{f~5&&~!)eV)@t;K<L_MyLN#;=gkjgvm|cXKt@+we*6gQ@sKMUi%syi>NR5zX+vL
zpBO>gXxA7?JeGXiYKtiI^JoyCt2%8*n)1e<261O|$6+Sw6ZJ8e>$l)|erXzP)63RB
zGGZZrq}5OuP;F3MKE)30>3X4|1-Ye7NdnBS{+h>qsP{1*IVUekCSXv77RBH+5)d2i
zm-^t4M)rOw7c(S3-Ea$1g*kiY>dWUXp=(M5oo4T}4hjW!*?ahb&lc8Ef3NIparL*N
zq`!?osdGLtO4`JC{jEjCA;@m|pDT1`209j&a!=L3+O<?nq;Gxmj@=Qox;JmLp(ro*
z>jsp%W;hZU>X0-?BwC?41*m>i8@o~z@$t&11he4=JXb&Ekp-Ei@cuisYL5H>no}&-
zzQ{PNq?Qnvd?h+tjE0qJJv;+Z8kLmxvA}ckN%f2y_il)H=<rpuG-QWk_i)zLoBC>d
z`c6gw)}T%*6eaJ{EWV^|jdw-Izz7kWTeekK9X_q9T0lYZVm)-w-;1FCr#^3*_Kwwt
zmb+$h^aK#E(;PKz(aH%il|lWzno|v!Erp2w@~trwZI3d)0p!7pBQ6%xT(tH5el{Ib
z3Kw(Qm$A9HotjEv2Dq31@4v0q+BC)y!fS~E(O5(uJ^FM%U=B#u%%J-WKzXO{&dk$-
z%w|e?B2}5?nsg?`S_4P+d6>@VH^7#)RV0moxm}s`7tE*^Jt!rBV-_f_FCuOMiwNb&
zjqq0X`Vp<+lvN@S?a6><BRE8xTg`?kNlfdYcbMi<m)=KO1@~oN3eI<T2RlzcPEa76
zpnp5QoW7%@KDs!GY6R&q#UijNzh(-PgizdFOrghG0KV@BkiHB`Zp{W&N%qKW)h57o
z;I?VGOQXIuCV{O}1yR17lmvV~$bx*XvE~y9;ctEbCP=<&`j#oHyB}FBsIpklpX_Wl
z7Yq*pf7G145CIF536oE4Ftn`s?Zw8<V!l9Qcwi&_fm*$tTTQL^X!;`=eTQfM_WtxI
z6Q*+bJB>Zai=HwS_JK0_b2Q4<tVFMN4YX#R;reymeJJ}4SiDvx3MeMu?6-b0=(^K=
zD%`z;=#2leUxQRknjK6iflfK}BMuBCD7u0m0RI;dhbQUYn8W9}dI$nqL`JHFbZcBB
z(xdFbE5(*CdS93_^WRCURM>Kyxjl8_?6i(DPWO!hq68iPdg&f{1}w`v0sjo+6wEB0
z3p!ykmTQU`H0!>o6j~8<FGK|$$6rBybF+(cr;qKQw@Bt@sgxu9Ee1w2hT)`nzNXd&
z1+J!IE1C!S_q();@L|#9=N#Oj|Imhq^(@-RCi$uRqbkwlXx>VgI}v{p^NyVmOr>c8
z?FvcJ>638?Sr^{ByWhelkoEoI|L1>v9h$mah=NYi4WI>($LjN}PiSVZWg#v@EYsOL
zGUN*aMi%>INfF=_r3^uqzTnanQ?Vp?6y@}SIhaz5iw&*?18hR<`_v-u@rw#=Cb@AD
zZ&fc&(}~6xc@9@UmR1!tONN`U&o|9b-3w%ZepUgbq(>;R`=AQ^Cc-jY_L_CS9I`*3
z3w|<oOKOB-pOjFtf~T1&iYfKD*tv2TAN?ADX!RK6okT~yW(sz*GxnM4`3jiquV>}S
zEEbAYsuPMY)fL;lT^18P5W=@}qiUqjb6M}$*j0;eZI3>?rFTw?|1q_TL?d*ED$+5k
zKe!jQj>+T2<NBdmYfh_33UkQSU+K;K>OKF01)ZuUrnF<2z6oSYZ<n4}Gp}y<;;Iv3
z&qxljS&S%4aXz|5m1YLio&n()tiX#fJ@v5YKradSiF<k^$BR{z(ZYYkaZ$$;yfVqk
zz2U$sQy(F<Mk)0tSf=6SMlRquVIAvGAWsE9&qt0Dqoq#4R7IPf>(@;h-U5c{`fK`*
z<TUB3{7W=`tWDjJqQAI)%~xJ;sXg@-D~LjiVrTUa)6ODMMemhrxS>fXTTapYeet$v
zv5jm@sOo?$P|2nd&G9xUzD-Krq;W>_1ym*M>(8LmLc^%opqs@*;df|sjTi$^Nr^n@
zTS0Ry%33ohvp{#uk0ZvDG74lcgo`iy=<5E}k9-th)lX<r0Q_B-FgOUh>SG;Z__by|
zG7qlz)>+<r?1nFJqdC92>enp;q!C%@B2vbj=s7#XJ?@On1O&P=Zrr&H$!-zA)DzEo
z5uIqYNlr}2sfekb_=^6*kaBGt=<bt#&)o1lzYex2CHWi-pNH~$81FsXqXen2&G7*A
zFf19f5{h2-kiS9xl&PsYM*5l061)Yb{a*FsWEya-^d%6QMCg#nCWN+@E%64E1EsZ+
zM2hHn>HDp2&8`w6a~l?t&)OnBQieo!cv_kVgCX{lR~?;{TB&Q+T9Je0p28!?lAxVQ
zS{tbCbf6eSglg0F`w!!6`9kX`B%S;pZ01QM4*B&s4dk<Sq%}2oC=>8o62#%5Pp}QG
z+|`3I$PmV#-a%!Z{&Vl%Q(iL}hDTHftn@qCGU5$hxPK9Lga)Q`V#7jTX(~*fUSdr}
zAS80wk<+)Teju*$nx<#?UE2VkX6P8$-)hBg`{+k+nDHZn4YI46zU|gYgd{%m*(YuU
zBr{q=mf|F?cu6&3dHY2QZ}DZX68H%RPQJ!`zKvj{PD3S}K6r}A61>r5pnJ^4#m^U|
z7Pl9*JR~JiEAjp2EsHL-R#eDR&)}R6d;8%p9`s9jB;~<=I*L7<1bS-6CNLMOh10&M
zj=iS;l-v;-sK|cgNktCVYNv7rqByDakRW5#YpNc=!ks8^n@+^Y@KR*ExVB&tueTCJ
zkNH)zfZVn*DB}X<i7KsnWTikbV<SRv02dUMw{W8&BGN)sB4cg!wYcsdCYPetdQ?LQ
zfa?3E8j(bxt@H@-rwl{_;~x3|<*?Hm(+7rgXlo!n!96B=>%Gh!u#A`|Iy*>(Evc{2
z9Om?Qd<ffIXMKa#wPp)RyLvB%55J}b1}D*F&C6huN~X*dJVdQF6xhBLP<R#S^1Kv=
z>SXIz6fG)>SP}p-b!_m0P~_P%M^XB;B1pR&sxmL<>);q5E^GA{e3&KZdkkC=52q%L
zuywKeD1-<x?o+>60HNo-HW2X;O3|_`%BLVz4niTy4nn$UMvB;Uj;QKUTx!ZDLJe{(
z0@#aI)1Lrroi$MhL&3iadXVCWjQ4|qp8ixH$7hT&8s2`BOtJ}<sQpZXNAL_ZSsZX>
z9HM0vdq#<YQhjltTf>l1Ah@`qemI6hnv@NK;6k_QFPKv3(^UDKY!)BP2WC|p#>TGW
zzviQVCYrpIA_OK2q4+0H=!4NP$h=2Bz158v`N=rMB&Z76vER}MRRLLS^dM%ugux#X
zu$LbllAOF*EzibO8B<bH1TOE}YT@npB$0k*O@h*RD`6q>rmy+8$P0~kkIf<xCZmj~
zs-_Xv)j?47rs#YK0*DA=v8W%W_=cn-Tv;zbnA8PB%u04<5z9U5M70$x>teXbC@Ky(
zTSCa2H1uJ_+0<(<2H>=Lt-d7`(#Lbj>QP?@lW5Fl^pTc)>k~rIk|^P2A$}rOA#m!X
z0!6l-lBAZ-d`h2(Ae3(D3HNS1{-nN##y$qbFiI)J8*3|x`Px1bk!F^}e=9*2J&sOl
z81~u&G#G<0%gs^Y2V6_Uj_!yhl?MKur>BS1IE+GwFj^On#_?Z}xoJ|;$R&2jgJ)Gj
z!rq^zrLYZ)n5(yN`m`QCd@zmzLafPXO2PlnF!jzP!`;ud5<JfgB}(|pZ2Fxj%~xD4
zj0t$tH8o65Dc%hV`}ewfbgR_cxQ)C+#Js#1ge)NG8bSxNMj0ktgR}>P--y~3eZMgp
zSbi%SFrn-endJ4;T0HIsD4DS9Vx*5~kConFURI=We8%z}XF%aL63oX89kyBt^;GY_
z<Phq^TB#Q{hNJUV`qj_W8HTjM$hI!6E=P^~PE@qQgF7V5MRmeTT$U8`rwksd5TPzD
zh|Lo;I;h`lFxkt$)~M}pQm4PDpbd()-n;(0tG7jmgXvo&$GzxF0C6>@N1OY|Q;C?M
zbM)b86!^Z$N3cK?p8hli*y%Yb^yH-Nn&jG@z89x_5>;|RaV*hDkB!dK5@^VY*WaM$
zLXLYOSy){E1t{}~rPOWJF&pcYNM-*M?yzE&y<pW7fx_zKFd>SVXd5?{$~}5qk|9k)
z_GX!*rIVZlbc8W*Vm0EMAzlJwWcH^wVm1&XPAwl2<rji6@;On-9sWHS-VAPqi-7)A
zh(FQWt0*V3mmxa2Xz`xI$?<L+6;aEu1lQR1D!!L&n6pqY^caJ=wK&1WgO-X5=p5Rm
z+Opq^rn+oK9u1S#eY7BwaMtwA174SMMDs`VWn4`CX_pUK_>hA^s$#tPqdOxmk1j(1
zdLp{)tFgo_2A&p+QNMhx_Ql8iM_MC%fI{{nD1BhrdSWTIDus;GpX_Bw36tP;Y^I1D
z>w#MsskwtFsmXuH{w%4H@mX4Z4`brk@a(}4yyi86h(RQ&;$Yz6+4RFQ+7<#SN5w6+
z=~9$LN!ShOhn{)KZgdy&2CjZ6<4hMoDXVG<tQ#fjM|FpX(WFx(rRjE%_N@QloBB6b
zDPu|SWUy*8@SAB>s%@B#1j9aQsFouo$2W@8;zs#TQHWjKFDAx~#~av#=2W>k{;@l$
z(Hq5G2b!egOMKXk)nHPSyA25?19&fu_((ny!<T{P&1I_zZRYag;RpKf8fjO*>+1Vg
zF^Ln^mYHYJX00qMq=G{pmCjZ`+{}lp7_xKk`6r)`zYzdSGhBcsH{syWFigTOi5sJ~
zpDE~eh}|c5piW?Ht!z9S9zmMz6)$q1_FDGxwufC8CW#`n20rbl7q){ysR>WTgkI6o
zZmG|_o}+Fz$P}H`(ezz=J*z~dqvIrMrb0|}s29J}n*Pt<|IfmycxU={>m<Rl%H<NX
zFCkdqj$A*l_=`wF9{`RLPET)^kc^n#tOu=PYH?z4sviH}U;4PYmqD^97_-j&t?jkV
ztsULgl@))W%r>zE0C~^-;J%ttz>p@Txvj4k0i&8&qWR{mDT1Pm>c94LKZ)Uns!*$O
z7{yz>3W^4fE(%zM(4||VAcYV;TKoE0xNfHibN#b(m;s9~Ob|6$)kju<+8MtL;0nz@
zd7=LeG)>U@QsFfHH+AVqHc>m~(jrN+9~8RDa3ohYR=5fv1<A?5>FD{?^xLyI)$v$<
zS<z(cjScY+vu23;m|zn5*RSg8WW3XD&yXRR_gW*&rz1XR!z9^El`Z+I{b;YpJDXV9
z+yJj&ZA}})xDO$TKldxWIO?d<^O$$@Q@*P?GYMg$KMIfGL`|6!E}Z~&p}t5*16%dU
zlI!nhr=`*Kr+NWow4Km6Q*Dg1^<~^(MK|o6ryMvO)i|1(A{*t&is(G)UCO{fuElm#
znW}jW)YY2fG5Oxsk1DD{tC*(hDne`sRHz?vzzi^=zsccIODamyqRR?WSU>+6mvm;7
zHHogBMt>tBduM)YclvGus(@PONrvnDA^GI+&)Ecw(+z=*jgIV0({5+28HXXy{{yXp
z@}lWa>*Z2!V*Q;5&>Gc0ZB;K35`_EgLzo-3+vz!R=ey(v^+>$Ve)jPQ%$UU~5B=iA
zC%X*rjYQ*aX4^1S_;-N4`4~CyGyrOj4Et*VjDpv+QgQ0a$dz;8i-1Lv&|_e?MGPWq
z6+G3_=BU$)u+`?$l8?uTI8zrGM5MNO7Pu09g}XIQh!mRT%p0I7ACQ1qVqC_%oV%Kn
zZ*n-;GpW1LzdfvEeVc?@G2LgEb|EX20snm0Tim|2Az9oy90oeQLtkGn8Lcd!oIlsC
zP4{8EUliRY*u4)`Zqi|#3wt_^CS646X9UJQ9%Un<)BXWgU!*8sc<G~l{@gH4V9dWd
zo=O%sp1MCnMTZFJ2vq7}609^Bea8!V_=9r-N&d$|Bz3$W#YbicOMXgw86L(hp7|S#
zIGBC_VIBN>y&jwOSJ&UL0Mju<n9nwE=ZGLn^3s~-i+FL7iiVke?nHb_NP>XMBS;-!
zjl&3Q9JT#DVIa}z0V8D(C0f=x4yM7MB|GbtAiAO(Er%T0uv5G~zl_rljsj*6`YXuI
zYka9FXRFn6SN~dMCufA{HLcf%ZSx@PNV7F;bkl@}1U?enRa~&D1(_}Fk7$nFFopQS
z1^voP5nVLbkIPGlGXk`|P834Ns1Bu6tyQS-hw#yhSRap+w%?=9?;%{50LKu|(XU}O
zhA`FoRbe)fOcs@xmyxI|OMpE$woM@wzU%SoJslT%m8*7_UbYc577Ee(L2c5fEIvBy
zw!&FCC-h)aq0;?11qW;L3nn@1B@Tm@FYbns(9)|PUVrQQJErkUwpH>Yxp>m^@TGnc
z)09Fxjd4utHo;%WZcp}(eU`-k^O<v^?(MW?|0_%pY8y(F!Y^vH)3Mo~zT3%JGRU&p
zHhuSihCW`VEr{ZIVfIw81B(g^EyD2axG>Uor+E)fFhg}uP53)3+s7&ABZcYoH|mR}
z&uo_{lOz*xfFN|AlymV)WXpcpPg3wvpHh@w=fgPPQHFbBqBHqy^HgL_>4e3#^5&oE
zg~#~x^DpTU)3<6@^qp3pUcaeo6}$fRBjyYD1o9xU2|oErJN>lMvCGwW#0<fhtXHmI
z_EtA=MH1<u$ElmQti#%YrFD0f%t@fD*x+!Q?aw0vl_97F5O;<9X%qnEBs(OTr~sFz
z;3ceMJ{OmmA{$W+A7h^m()#hF#n8TL6)^W_Chr>y3)-8iM|;^OwnU9<Z5)SPp=)vQ
z6@$G-A(b~xGpmy7<l@m_#+5UqHS^HHSIzKKtDpvCi7!NKtGRh=TZf7wYNzowhrXkQ
zN(C0sIkMJ@u=yZ7{h_*1QKK(YB{d4gBq&P;qUNGd(MtrtyPCFG@A!azaG6_t2DO+N
zWz>5RUNJ9JvYBHUmBOvS-4qLP9V1MK4GAU9qV4IsA!UfwA7^5L-q7%AHNY>=`ww2t
zju1B$VoTdJ$|-RZvh@yX2E-N@Kb7gag4HF^kCcwG$rrx#l-0;CuS7)!xR=hDS|XX5
zj8pL}5_4j(FFo}o(}FJ?hK_(a1(A5in*b$}!o3Mjs{||nQvA?*C?uai3>XI{4X1&Z
zrCd=y8~;AjH=Bn?t$lhjdZD+E_~4+QJR4>D!5}=L#EMfH!QZ@p`ew?noh@yk*G|W}
zAej<TNsTxs5x(i#>b&<t4maE<+^^9RlL0~;;J*+L_j2ZXM|%n8-lADS+ytIg>w05v
z#6`uce|%4D_V7N%0e-<6g!PX%aWyS}$p3Z05UGQUzQD~Y^^kA^5NChBiE!~EG(B1d
znB_b6oFcC}!wJF4)J({`Q8ox<LKEQz(PP{Sr%8c1UGVCcY*fPC?1ypD+Pi4wT%DK{
zsezVf0Z=AxA|R_CH+?Iab6dU0YkAuL53qa9d@rr=(gRs>&VGLVs@J+l46I%pEO{=(
z08`8RljNEipP0C#i<2OQ<Tf9b^C=?DbNm26{mgmlEef1?6X%MKcjyT7AGdut7}JC@
zLwO|kOO89W8Iip`YO}(;Bs)*0Krh(BNq}ZLqumxS^P4o1t-$*qyn2$^WUg?HKl~U;
z{u}(!U+D`w&>e5PHgv|1l~vNx(tXr};lmI7Y+}<O=ll-{K_Ry83~99R2mOUL;6EZR
z9m<>f@5Rh%ZxT%Mrw{^domu<A5Bxv~h|lDIDE4~wd*4V7;UwH{gfrm6tj1iOOGjY`
z?d<3BhH#6-00q|pflFHs-^GpU%iXNSUh=K$;HJXD(Q!}IRYW3%<Sr7_iz4RH5A7k@
z^@w_E(y4VEy*<B_CG$uS`UR)H2U<WJ<1}M-h$2cS!8*}4&U>K6P=+%XK+^V$QQV42
zKHv&Zl1L!6_fHe_Eq5a<5W365U8dvtSW{ZG4QO{1)d-EbMsrc5KPZ%v%TRqrMcvJq
ztg>~ajm2eVa?%G{o(ytW+_#|#fEz#0oT|JD8mma%4sN>MySlNk_^f_V@->Z>T}9$i
zex-w+S|G+}>B&)x7@zA!aDQ#{j^8B|ey<%^DPjdR`Ay_oqfJ5^1WFmVT)X?oji*e5
zI}w@l8s&pOTAH&Aj=_?QTsQthHI^6!l)MUbN}LRYFfFO&R)qPxaT~D5i)UsIyIHmI
z0a{q!T!Fd$um(;?2AJqv3Vv63uZRlC=)aXbJ#?!*j_;y)AN(1Y4MiY=5POKgVR3QS
z7MFI+mpjx&$7-pN3^$+2DTPt_KG4WvD)*R?jb*3rH33zqtFzWRLA;qFVff(}nvh|m
z1T#_yBpSOLp*o?!MX<5HM@N9jmVoO_la}H?m?`6OV17#0u+0!haR!}U3e=Jnwp-mA
zaVOeU`)k%_+$AcE+u6c*(|Xk|$FyCZ$pNBp!S*Q5S^zLl;xTWsq^x%-cS`kI<`AR}
z4P6Oe1k-YtlW7BEK+}#oBbJ*?B-2mf{wZ{oaZEp@97T%?t`K&;RFjH2(413>d2UI6
zn*ML~HU9`Ti5HfMR@46o*RiOL5@H~XNsvIRqeWM7R-7sIO<65<d^%WcqbSKvjtk5=
z;)aTCJph7#DTb5t1AeNvGM><Gm$+*Zgp~R*w=dFhq>s2hI!ZJ9*)VIhb}&8k%LSnU
z(|Orm+|eQ75N6FClp$;o*+p%9(9d356Xq0T0DYGkJ?1*zJ^hyqBsU2c6~At$wKvX`
zUF}vYS!VG?e3}|HqLFFi9JemBclCBFVKlUpA7$bL8?`nP5sFda;z~CNsels6b1!QP
zJ=x24d2K@45sjVe?bFYr^yW(lkM`fg821GosD@rUf8wc+nTETiP+wfAQ&I2p^mnNw
zD9}m@moRlgIAhC)HBCaJWFYqdQO9Yk=#i^9vvvh(l7aj?Jy$X}RD_a(*RShWTchDH
z<LIj7>F21NFhWX0H?B~KJwy+KkMmRn<4g2-fStjIYo{=5s>p4*FW%zx&AYy&bCaWN
zGfdM)VVBUaX0@^AUn<6>-P}SYD??_k#PUTX_Oq+mdQAoNnRO+4;t=K8OPve*6Bq2`
zH?%G^S-_Nf)FFEhIHRQZ^ePcXVp5Amn$b~PKTY4n7z+DdidGKIxnCd*%y91!)ay#?
zC>sjx_K?y(YnbaK-zKbgi3zKeA{Df7TZN&qF%SLttAE1b{*}aPZiubu%u6<w%X%A0
z7n@!$EVBprAQ^0R+$voWxTXNw6CxFzR+#GU17ReO7@$KHV_fV^P+Wi4)L}{);A3aS
z=$tlck$gam*YNq_MJRzM9(z*Mzj5OQA20?}G*Ae;Cfl~+!o~X(|4JS@1oc+y$hd8E
z?uk@mL(^wWK?qyx@YY)g*mN1-Yep+xFSbQzs8&uy)esa-qe+D`8MP~GSZh*{qup-{
zgq1`u$!so$w!>&6|312`M+V*dJE7*;tJ@Kj#r!$FUs7UT(Fs_OQnA5t(Izv^sy^mC
zcL&AcgXorliR0d$j4i!`2I*AoWHLeFX%bw%A#%v6B`W}rP`f|(a~4WI=p+6g>jG?c
z;_`8SNz4HC@+L6AQoe5<F^!Ddl^b3z2D3$c7-xx{is|gb)Kp+ePfCrLANpKtKkaa4
zU}WHNZW2n@b0KsWuiq%6YPXl?bPAh-0O|L5*J(@6OTnwFygqO&7=Q^t=aD0gu}ODu
zl}Hj@ck`@?a#WV$3>ik8PPXP#nZXt~f#5W}{__<7fUwHFCEHB2qBc72&bk<)J&~oE
zCoQ2v9w}xZY^lCQ(K?o{%E>!_Q7%ClBMKvQqDcZjd?aI^bCYg&--jhD#Lwd+iv%Xp
z;hqCTYa-)FW~QMzOmPkT01=k6x&~(<zTN^{lfJ?NwbjFc6{Z3%m(zC<DFS1ff9OVM
zP9}{C@AgwDlXxiMW-~nj^wM4rdKIZuLDA2yKAYlt(eZ3uruH;&2TXBH)gYLEjk=DA
zWf2PW{s^6w1OgX-6u1Fsu^dfyiL{qMLlkJmKp=nV<l!5&Y`w~t&F*3AGLwDl<Ba9b
z5eiFWpt{r4_W4_-vy!(-j%}2gJDCCqqDcHly8abK(64^$kJ3K1j)~eP9OC07w=5WG
z)orxTLyM%tHXKji$%wp|y52|DRQSLL0~70VaPu=-N#d@yfMt=KmQM~NR6{@l)U$od
zlIRZRcJ<f{=+&RFj1_AvjjZ8Q(z#xE0q^JesFNNqk9CJ(`n{{{cnu9=R<~DrUK{hU
zq);>yI?>k2b%1tvcf}7++_>j+Tf*uz5y`1n>%7sgC69|Vnoo5~MM&$=A8*pZKa5Pq
zh~RPSvD3E#@&vn39tauIq7Ua@0wQ9z8-BGKxQVXQw`#%AAW(o(&w5;c9D%dXzvSIf
zTdtp071#|fFVGfvV=qPB_$&jShy4*Ds?;r%li|53$f*Z2)XV49AknM%yjetnfdJfu
zFzg(J3xX9=7RorK%!^?Pv@-F%3sg=*lA8QcNiGOP5Gy^_2<K#Hjygk3IO>A{<Cqi}
zS?OrSica|&5g@I#bs0|K|M5lz8WLZ~^B^--+~w$3kd0{Ok)EY@At!M*{Z~EavyVNg
zH7dE1OpImT$~tcVy43oH@k^|$1``P-9`zo4+tg8lF#4<e-;}pvhjEupOwATi8bI&c
zSy|NkE&-ASO$b%>(j5(Jl@eHV$gGr>$-a>V6-<MoH$Sq}q}O4aiicOdeOO?J=%Tpq
zSZq=b^!^mV=l8+R(rV9FkQThHPT<5yF=>71P1AmSP|-;7^<OL3a{U*+iNra$Nhuly
zjPY@z5GG-9hBL>QuYBALPp5tjiJPLThK3RXMR`S0$&i5vENeVed3hy{f+*0So9Noe
z84_w7MK*Dsh`w0!h^VtEl!%fvqU4GN^up^|Tiy7Am%OAfzrIe#%QR<}$y_sJvg9Sy
zzs;Y$MT9Kw3ge>urY4b)>A%(Z8zJZ#sscWSc$P2|YY3GRq@+$!OmKJqbP#YtU@ZXZ
zOn`v`1F8FHDd-m~d=-I%xgnOZe>qQcm&6-H1MUKBaAQpVquX}E%`HuO25_^tztA7X
zmiyycsV@t~TcTqVq4089!|jNf8FYONPpV0CqGa_5E@KmhU?6Pb$Dfh()D|0;Tvsm^
zCK<KUZ!Mw400d?WQiT_OBV&a(BJ`8f)qRBEhF=(X9BkV-gM0U$P6GwB8yC(WqF=Lg
z$Z5VA$89jvM4Av@XU*VeBW`=dcKkj-X6UuzZ==E@$ewEXZ~pNQfAcc`Yq`7DKhf4;
z;lDuNBuL?q@RnBvj9Z1&KG=<DBwzx~BMbd0Ax#Lw`Cn>Y&HLiuC`unhfAE*SBKj)X
z=ue?~WZ~;!nzV{^_SUmjD*ug|SA?rfQ=8k|T=Uv*mNY!BWw07@d*;q4&P=jK=f@?x
z{!6_@sGerQnG%mn>?KkEVSrW0esfE`H*$yRNu5w_UXX3G(W1u${vxIJ$4?}@_k@xq
z!ueVoC|0ZylWc($S3fmC6hZx3@x`lmr7)kpt^a>?Cb88&%s70LsoVPf;LBP+UA%_b
z0D6u3(27g}{?57UBT4|qeIiNyEm_5WH`O3G;pSIFy383lLgBK==N+V9f~0VJDic71
zHqG==O{K#F3?43yoNdBT`qdEhahxERh$37|DuQ>+OI8j;pIKU3-1eiE_eqAO{jN-5
zL{PMQ=3r|laBU_0V#NH+)0&T?>OL_~l?@-)H+L&O?}?9nOuu;gr5IQPm7~7Pv|s=%
z@ptpyvP|TDXbZR|FErI@(lm0-lf_08uN)j=vE~)sxeAk;aXhK4=^QplUub~E4I<mD
zHy_3|wqs@Q$cHF^qO~6N0{6?dC35v0x+fndfU{^*#@(A_j#ClZ7km8Ki~2l*wlua1
z{)a+(Y_k^y2}Du6;rTM)!J9$iR@Y6LaylKD%nQbA>r5%;Q?aY*r<6}W(wl0dh_cqR
zy2GVtr|EBZ<E9oFABjExUMzV{UeNypk3ge=GP>?^HWy23^nDauMo*VA9Mvd8Zn?<+
zdEbxN#owMV1JY6OE|@0;(Ad8DSCD@(-}AFdXwEBHdS110`sC_-V~1p#ztkn}=ItV0
z2}$V*W1>j@r4nGZ)`@``cpwS9T~niiG#jbbh;S&%irKH`rG@JCHV<=_n#_L!SG}BQ
zp^<G3l0XVA`Q(^UB%|3YqzOSBX|p2)fVyj$d0C6DOE8kyLS0=mwGNy&fa7ACkA^;l
zlq&@Tqvo1*%rxF1kUx`{F0-JTD?p(MZ1UY?r^6%symB8<9GtF~clAc?J+;Dp4Y!8g
zm#54o%~pS{7;To>Qyk+gBH{Bm_otFicBNHIK<o{ri_skh><%I?@7Yb=veQUTZ!F$1
z8R$J9!+AsquL=7%!=vcj80xuR`fI8jy0m6M*OyG*S)W(>Sxm?XD2-H)NV*G}A)SOn
zne~1_lk;;ZuTEy`Oy9|X_|0Lg@y{QBUflGbzrFF4{)?&!5y6rQDvj*i^uAtm_X1l>
zKX6Ecj5{ir5HN6B2<RAB(=eac5@(qbX0>CtPx74BFQO7oB4OMM@P|s&%~^dKhmlXx
zA8)TVy&3G$4BFg+ozf(x*k>H-$-^amA+~r~HJgPYPi2E($I=hH*?Z>yjp<MHVnK()
zK1yR_A5@n=GCzqdsyxh0R?vGZ&7Q08h5wL6<v;~4hc$t;TvPpi)>`$Kw<w43?l2i6
zY`ycWcHJrtmO=@|yK5X#E2Zy4QskXJvsUUg&`R`A!$Js7plZId7qtOt%0|tVU%z;i
zkPOU0NMVA}(*|UZ4*)C-z^8DO=_8zi7DNKhx-TKIH(amjn<<r4t({7TH+K>kPfxHO
zb#?w0`-uuxs}1x%s+hCRq|WrGqw#D<=G_?uvG%o#>u+9vufn{C1JLh&URj}xj#TAD
z(<5~AEz=q7W~M@g+fXHk>9Vle6j>J<U-Xvb*<HWlg)6V50kUHM=$q)@Z-YYNGNwre
zAQ07<51<gKG_PFPoY&PNOcU5Kq-NF#uio;)&Il-&`unXm{psSSO@B6=)iu)yFKss@
z8`JZ=@@5&H%dI3CS1YIP$UZJw2f>s0L_Tz$z4ZvC?{_JY(qok7ypCD#d_n;wgk2dK
z&cBy#T0K+8dYHC!K&Z<los6^PI+t|%oBS}^I9ZKxm!uMtE)jJSL0Yu-7DcD`X=`Ib
zP+NYY6_tUJ)fEE9EYCqA=HScmU9vI$%F<CQLwjgxXdch}R|KX$aBE{;n$*OdYwaKJ
z4|lDe5$&W92**^+=koO3T2kAYv{UN_vNun8N1u;Nh$pMv;$%m5)jLOV)&904Ow(Q~
zLH~6sxL9qu9BoAC-2)tA!{b^0LkLT9Fm}uKMHO6J3dY@no4$^87{fghH>l*|S1csC
z0w{!uw*?S2n767A>d;D}B##6<cIPlJDI@{;S|JfjHS7CK>D@K`aU5D34VdX#I=%tq
zG>IwqP*YbQ4t$btS$uZ2xzqH`iHI7dMZV<7y2vv`{>F3!O;>yNEe;ZYep{7-hTDRR
znys_&D}eAiIIc^dg8FviY+&5HswNF{yj<|9a7udm(2BXA*>OOfV(N51X3u4V#a)WE
z&CQ4rQjvZvBV`i7s2aW!rGmJbNsOap(t_Cd6QORouo)I1^T|5g78p)g84Nrekx?Dx
zN``$KoYyYH)zQ{r+rcX8&F(c3lmIcJTQE_Zo@92_GGnqCba?sT4Sj-qdDRF7v!q2%
z?~gMI-&`^u6?dLZ`7QFmtIyXAWUtuzbD#U1&yTIO!i#4!>Jk_%(P~rmQm-4q#M)gX
z>#Ozt{@qfCw+GP9uR8$%RL8IldY3iDO6q19mxOprWP6YM58m*XkOp79e#L^)W;tq}
za3B7tI+(ogOR+9E5|Kx6DuGfwC_~jQ0nwRZ6;0D5B~t^tb{B>ypqBnY70p<6nAYO0
zP&^6_MlmWhc<jl?ysTm+i&_LeMJ)y1xZBxb1HL`Q4@r44eW%oL3gqm=#1tbnAZ|(9
zo>r;RJB}=3!pfB`brUBY^`kM80TE0wZJC3RzzwWJzZ1;RIQ!sW4sc1hp#{)#@5X=E
z_M%=8BcQ~mK&VI7j%_Z@d6x_{<!;3;OECMfPrdY*o{u%H7}7zUH2tylUk%?hrJ0Nu
zK)fV;andp^jx+Y5f<o7?YEN?PGBl;(->_uut&iY~#7Ld<aP`j$h<b;V?w26X&OWmZ
zdl713^~<e5o5@(uTTNIu!1zZGUe(U%hs-;1ti3)e@vw9I9W%z-kmM4fpY)Y;glb;%
z(k%pVqd&iRw@1Gx{8kBt7e6jq93V!#$`A^E(IyEQK?f{s)+TkBJaJ8=r^$|y>YPL8
zGCeI+gd%<K{WqS@CIO%cy%|A{hzbgM01N8ntLV!l5rCx>c_*Vu6?rkh*y6J^p@@42
zxvOV<nBL{&I7deDqUdbrhc}3H1X=I6j~=K(o#^x@{J~qHoMv2Ggs@-&^}5S}K+H6q
zwx%m>A^~V2R8ty8j!GWmgaJ8vvOLpt(kKIfRYIdT_$^?y+OPq%Y32`)ra$pYN>#s}
zB7}XM)#4ZQPpG*bt&HfedOxy_C8covdVW)5WK!8g!>Z|<RTC?aB9HAfQ%EuXz09lF
z4?BqGZ|$aWL~qy>hdml{mH42V_t(P@7Ju6$Zk=P_YLUb9)`M^ABEIm6&zgIdPy;&d
zD}wj>3q^kVrxH(j?|y{r!l1Ck$>D{=QKa+MHs|#D;kYECz_T1?NLaY2US?UaQ^VVm
zeFBQ^>gP7iLEB+0<JISh7%HH(Jn_-zb<CVS$zZPrV?%A*Qrc1HSfC}KNfH8rf>JRU
z^|KZK)#{eceM(~Z7OxDGHBT>cvJ1{Ga=$~W(v)itDWs15(TyPKA@t?T*WWcgB*a8|
zZm#vqdYzBC*p*eGXUUQho=MNbv`EIjQZvo>{4!d%R2)yFqw7UZ>;g>i?3NGiIUHfX
zaIJ(&7<o8LpOI_BILZ(;<|FwDw3Nu&Xzqv^wy*$7be3w7Fy$c4enwZoK_K92FT)1G
z8^5@Tp&Y04*k|;9?j6_Jv>_XN3V<z*KGorQ6IheyWi~sT==Qq>?J|86Oi_V#O^#V^
zu2&T-q%^z5#u<i6p{n*cibuZmr7ww$s*slOk+TK=VBlyu0li6ogLoTWpgv6LaqN|s
zH@~RQ^{HDej={t4PtFioc};=e)Es2!hI(lc&pHU7)+t?F(iWvlOQ>x<V;?II<9$E4
zi(mq10n3MRH!lMK_ZY?)ZAN5+n0y_a9!a2TY3(TpK1yX@N-o)G4QT-PZ3TK}`a#V@
zGtf*gY)?jenL=tr+>A7d8D@RwwkSt)-F0YAlLY^M<st+5jIPFV)iK|=@w|83rLRcD
zmlPzYU^1Bbv{kwgf?I?(dDuf$xa!*vzZW+(QUTqhF}-<*M>(U4?SDt&#d?JO(pjiP
zG48T^({AIYmq@}~e8^vLxkLVbB<gcz5)Tt+B^?G|V%GVoM0EFt3v+(Pt-8>wA5r}F
zRWr6?o&Lm9KO_KFa$r@xtsNqIHBA0YPff1@SBSmP)L|~}7FaT6u#tVa9FE+VO_~;?
z<G*NH(WkW^gX8+Zsr1igH^S{?fzoM*VQ*rZdx>{!=kr^}8K${*bCQlX##rsNe}L6h
zz#;fvE-MLH)}lLrI1EcxW~uze_kF}%#&IwMzh}ML+InMkBR-9D8>TD_?bU!uO`EHv
z;eccKK|6})o2!+@9`#S+LWBdFSq=m{%08tSHe{6|vcGFmlMrD=zbvwo5JV*|9riw6
zZ&s-+GSqyU)m<*Oa2J;b7)#DmR3<zkw%Su079aR%lEl0$BmAH+zKXrlIXYtWm8YbU
zbgD%jhm0&aMRH5T;~%lf=+&Ab^9Yyswh6|M!UkSlY?__OrXqN~XVW6Qtxos&AVXm^
zX$jv&D)TzJg2QTXesn`1ZV{J|s7Mx<(c)2cUX~z@qdnB%>3-bg?r!dOu9d-`u$sPJ
zOA!9_>F0g6GN!yVmY`O^>AsL2)@RluUJ{Y8C+ItV!#7RZ9An1sk6^k`K;k44=~>(u
zCRoN%O1wg!XfjgIFNHOkxpDovA6shX6F!3BU80}&o_xvc-2u^|vn1(D@>D<3K>!!Q
z=Su|MRHwZ<eIEU-b#AKPuuRmD-Zg$ba85A6Cy{|$ZEk9DXs}G)EYViOf~l2jVghlA
z&B8f%=Ci4Ssvkha_;t@gGrvyNuw?&Dr^t@r0%xU}(!&TBz4j2Kj4#zKCc?&Wi7k9o
ztmcb(7!X^?e?9{Cly3s-kkU4UVdQ<W;>iK&23(fnJC;ga2L?=r-1Sn-X0LId#!{rd
zMZ)P%R8(gYg3z6TN%24fDmAyD2cPvKzelq(FKSD+k)J2CY*zfno=f+KAo8vC8k(Da
z0H`WAf)$WboyO|SPM4w=U4(-*S<&%%jW~2V?N^|{JmpSW2!;&f(#F9T7x<jRb9q@y
zna0@8y2~;1IK1#NZGY2v0c(jf??22F_RGLBV9G&HF!ngA6`Ff%Y14nN!(p$#4ilo_
zMZ+}`fF+EF*B2K@<HFOM!ePM*F{Pjl9q>|sD!QC~4?+r6L5Z;-4^@<({wNnnfUEQf
z8(mbuG~he%^M1ObL?pemRQ#C?gsZn6@+<|9+%&D6{}Q^ewcfi+SRqpMV*jWh)Xolv
zs$n&EaCS2*X%YU+9F0C6#>#)bedDRr>jA|jyHzwjX{1&F?V5|EH}f_ucEc9*)!b1h
zS*Q0EdC`+@;frOM1Ql=|reN?<Vu~A+=|A^?LnrF&FeX_B1MIfwQ0T?uAP_gK?Fj6&
zXoyTYEL;P<-hc*f&{}J!_pErMzjuheTMM|Ut|V-07Hte_YEE5Yl3iik{v#z2kXP~T
z;z`Jvy?)j5i*ge`NyD^!j>{;py_&d`isyxvC_8XFyjhC|)R;Xb0yY*L;2rA@R=z`Y
z?^2=b=Mz@d&}8umkgU)+kM`46u0Yd@5>|#;$NRiYMxaD}R}=qwKWlY<7(r?^HK*mg
z{U?<=Gv8XtQr*d;Y~&zutLod7uVid$*Uv6`h3<i6dRy}%<N=)vO0ML`#}8i7lGW|j
zi&6QY*aS34K{k%N$&_EftH_Jfp$I5d<a4d8|8e|Iu(;PuGm_FAw2nslvuYEZjgDP<
z-5PcFLK^+jas!!pmT(rvo2g;LcdLPVYed?)mE{evFs1jYeoNg}HpLzQg#(c5ouR-&
zDWSZiXMAw;%h7YdmlmwLTQ6%R*q}n`1v(Dyee6@oFbHanyVhVbEWo_S>`m_*QV^W<
zx|Y@07oS2axpg8)X)5$F&f+*Tf9xe|J7BenNhf6j-ofh?IuJO3-q(n3F(%m#O#tuf
zEaBz!wbnqj2ny~c%|%pQ0`TDvGDg+!KK!1J``*15UwYo3V;nR`=J$c?QonDl93M!g
zwqA09ofKSh3<@ls_(3|d)ObafiNZ5^@mk((6fpK=)j_CEJslnAQ8#|^3gCcJu5;N-
zoW}T2DGhH@syuj!PT8=um#7#oksUvTvJvR3H9dGWi+I*c<M8Q^Yjg02|8O3mGZi7{
za|j%s;T`L%Yqi?7(Zv)a<*MD{ICX;ZG6pJa8nk|dqr%?{HvyK>L;8o|lOreM%czWC
zlXbPV4_mYaALq3L(&g>or1yy%S951T(A8u=AMWR6R6sA%N1b7*H>#cK30_3OP$@rj
zK7<=p8oRLv6s8u0F;nvDlo6p5k|U8jOv@2}VKK2w-wf^o{*Sijv)EFe&OgT*wB@^}
z5ekMrk3X3kAnZ$TuL6-=D2+MS`+*#P;-gR2hXJZGDJX$V%P&AT^_=&p7(;AoX<=g}
zph8a%fjG#1XV-sU=IYY;u=D@2cPD;wl!u-FpWM!^+{(Ef$Bt#k!JS(<oD|?gI0iak
z8zr%m0*ozej5-Kx*|C#nW+WI1ff^%;5jrU0(SZS}00~AK9ZxipEie%UiO^tYlc&3<
zdwS3idb8Oqo6Tl_pXd8j582<}ka@k{Ra5=+Q&rDdcl8R6LqnD25~l8&xZ;M-4EB<)
z<_y-2t4<GJ`<b$X<w+QF;^uHZpj@%{+`uynp|gFbp8qU(Awc}C4U%{TPXL$e%&$8S
z?!OIhaN-x4vT_EYe;Vw!T(3th!}`-JvVU8Cg+18#ZfoWlF8&WP;#$sD4o-uNYtKKN
z4CZtQMKri($r~ujYV)zFBe!0+a&S@Z8n%_38&K!YEJJy}hd1mFHaYy{<=~!{tF9ef
ze+%!XizQF@$kUf)gSwdA*6ZU_4WU2ex@j3uWMQjTCgM|{<g}UU`SqVzxR%7dbnK{w
zPp{?-)Zl6-R}Q%1GdN3G{vNKH3NMEXxkEnOx^l2P@S)e=WN^z0ee0?#Yav}6t?34k
zcO)SWTanj)4kT<p=pP1GW4LKbj}_30)bHHFPE>gOBktC5z9UrKx4iYYwEE$<H$#D6
zd`+0`mArCfVRx1nFs{@lo)?3zDvx(>;4P43Z0G^vZIt{J#lo9IUjH3a_R0o3eV2q;
z>Hv&%zWJ?hFYDePu|vc@9d8dAY-2eSbF22$v>keK;l`88Ih>9^ToHcALuIR1tb~%#
zo5pAlWvk_ul=#AD{mA9p-c;r;ujDOjP3W4ruQ%i;H+K(Kk&@*p-jE&aHO6PNdDK}=
zn{H&`^k#UkXgQx7!y}3d58LqMA4iB~H3<*0F8sJ2cUAF{hn1@a^?FBHpBGN=mDPzW
z3=4Z}Y`1Tk+NZq@4()j)yG+FgR_dYy`wk-u?}Xs}Crav)H;1(k4Qp_8^zoH94W8ir
z;3o%nD^miTq0`%wF5`%i_r8@4nwJz0y5{RwEo`|i=e1Q~;h~HD_+XETGn#`Pjur9Q
zxzO|JoLabC%As{x7~E^Zi>5+N3EBPrD=&W6VE5oQ)|vx~L4UwAfrGtX&Me;WIl8wp
zFS);&NBYa2brp|?vv<1iF!+afn5X<6TYwv;_Lj3RkC@dC!*8dnJLClV>X4Mx;ZAFP
z!aFSnwQ_3RVBe0HnJ?6t!F~h{k1nCis7pCnU4G-;w+t4~EurjUhE{dcZG&r^m+0ch
zRqwvy17($|^P?=ET)6dzE`g^v%hl&jE%GYNmYeu!$mPd3tmc3~XA}5|gLoC60=$M}
zw6HSax9`{w7#xs<LxT9i1#X5f7kOnkvWu_tAmY2O9b6auefGdZnQ)F>$4G41;%*V&
zdgmL<<b5)}nJT;ji3V0C|3jQFEN33hW^><GsCVz9l$G7QIp^bpL~%)t3o;>?TOSvm
z;NZ!w+jPol$s3ch{BH<9s!~3NK(`iNwEh45!o?MwO@BR8c;_Y$%av7vOX=YOp@j$T
z*}^TH<O~P(Hx3@?Cov0m((&S_!FfQPD%37QndqxOaMj>}h@FE6^}|Ns`3)hz*u)Gf
z%4Mqt)%O;D4rFkKkkh9j!Qow}gB|0WxFztGm1U<FHja8{Lac&ePrt0RQx9=vn%NHy
zeRNRHcNc!srN*)I!VivI#ZQ`qr>^6hE^fH?V}t#9u32(Q^z~bCd73I5`Gu8Y|LEGm
zTM-wIR0mC~yg)rj-Ng$(WN`hBgJy6gj~~AG{pIR%hZ|3V#<w%CzV=gtpI70oeQI@h
zONO>wZ~ekW<%-1{{<$MTImPiSw<HbPF88|KIJmfZ75ycr=|X0&y6LvTqPY^uJ?3F6
z_5Q2MO7kx6X&f9AaXiJ(@RmvWDDQk*DACaT|3lNgQ9E$DtApLUa@ERfkNj3eIcq(L
zU8Zvt*N<-InpC;2ToT~U!cbSj-HbQu3<b~Fzy8)HS`_DZS=rDT)K?DP*K_HL8>qqM
zJj2rrq2-qI3^yGuT(gaDQZ7Tnt3%@3-@;Q^g9^ocp6tViue|5RTUOl~4m!%p%Zq=?
zw#hCQ#Xi_o;$0epN<agDs~%kB(3W~pruYNlTwb}*RV(R!2OE((?|3LbpW?Fbb=n&U
z_ra(&u<KAJi8Buz@+|b?;W31;16@YtNR3Tk$Ps?=D^$f5T+`Kltqyke0wGpx)yiw@
zAvE06I@l*<`>v;cSACZKjlmm1bY7yIw@$Pxw@F0VN%3<6+ODxygPTuo<3~ip3cT};
z7lqOFD;9PFAE*2W7vk&O4siY8t|9LC;jWKC17R~nqltT%HO|C@3}8><qvdW9N1?pd
zn&Jv)`WNaEUnt`bZde(lYZX^mLt44|ah<Y+cY&_dC%1)5u4T5y=Rg(?LCCa)H#LTv
z7+(KvlxuHW%{7uRL-O$>A1Uj@rI%bCa)k5A;X&nco*YurkHj1L!*3)noZ<-E8fD})
z3%mI==Rxtai_A67avt>3;Z`TTBZviFM~xVq_6$2x&#!0iT{n4#hEDzt-W?I{*o^1-
zc)_-AMGDWgmUZ^MJX1;@hqEN%Ttj##<INw{Nw?7OKQTD;;AWt~HF+JM=>%MPYK0R+
z{2&&OgHs&ks_F+b`7NSylDBV$XFvwKJ<C?E)QgzHwql_>(4ikU4~1O3I^4=JD30(X
zY3Ra2*%wd7YhdBk6Jh$39F>L$m#zr+)|54Xci??Y?TU)SeRlK{q4KS$tD2K>*W4DS
zPwyJM{hrO)!Y|~k;zsf?LyiGiXjlX{2Mqe>cilF4c!So@dBQN8&wpX?MDlxY8*HPm
z)^XRMm2#?wyT-29`(e4BqCJZmKgXn(R=)c^Wv$`H$3a!$T5WiaBFz1^g<m+jBD^2>
z#<Jeu9FAVY&I?J^m4euta{6j;PXzb2ay2Y&?>@^D>R~C`#_R3*%Wk+yZ$RaaoWb$e
zm2^f6&*NQu6OS&e8a$lEebl;nBObxrc&+v>Xj%(5nTL0#$2a0}|JR}TdyQAG;1DC^
z-zskXQk!GPBtCK$ZUyG4sj{?cp${wn=k)38=fo)j_AkRMc+hN6v5H;l<Mdl06eV{U
z#~ayJt{fZ=g)LENf8mugH?7jWDAcCg2HS)hi&)sxX0K*&=7Bqad5;BGUUXRYzDvq_
zS))6;WmPy-FLx2b?J-<Qy`Ei!vfsPp_4j4baWA|rnL8Bg%Bd@^=GOws1oCEt8wPif
zT^ioA7?O802kB+M$_AUnmQ6FhwNUq@Uh|%EVOMiHyquRF+ANzO>ENiJ->TrX2%+BG
zc-`O<6)!oc56&!es)aHrtJ!D58{FtR%PPf|i68N%zYe>d<=XJG9-Id*E9j+HToG0_
zJnj;TlOKV-b<pL84qUtEoCId`T*lVjPlK&*tX>O02&aPuUPZ{MpTV+j<f4%ddw4Gj
zxBZlBNnacu6xdj1NqET@rw7YG4&N>hzr<Be>XJ7*LgnP(z$IiC=a}?cjAdFrHnl6v
zkREfAjV70I%a!1Ec%DcrmsMNA!y-4X4vAjDvwRCV9eVY67s6mQct676`sYO_9}W??
zp-5-pco6iK%LcotZ+pl0mPL1Ixci~pZoE<V&V));ZnWr5>9Rhx@H3Rx-8k5P3@;7h
z7VaB9Q7-sD-ox}cE|i5ivT7>^ZTf>Zae;kB=*KVBnHAcBo|r1vf~~?0gIn2dyiT3n
zdu~{$AFFr|4UcyX&a_>2UD*JrChX~QbQWI_8tzw*x8Q^e;uP25xS5LsAt%^(yF1iK
z-rQPd{;J!w|IEGbgR`8Rs4l}UUU*9;m)Hi!beCThLc^~P-#BRGAL4XmnTpWf2V1FP
zeBr?|9v&L(PKAr%W%J<{%)!}2evEkGVSvlI^SONM4L@G9a-pd8fG`(v7H;q1>`Odx
z#uhE?@`n`i$m3wOsO5v&A5J?j^fN0s7%7h!bw=p_|HN6bvp5(EPj4wIv*MzQ-ckN{
z-NHL=i*59N$IAN2S;}zAvrO5$DTVNTUYtWKA5`J1F1!5Vt8{x6o#p#(92`7-`t+`A
zKN-FqUVFFj68cNn5f~KrHJ6v!8(s&fjWrMY4Gu>y<`6S1Hrxin10rSPq5L^(sdhCK
zA1Z$S?s6uVvriFsUa$Y0cDKlYg{?;se#v-nzZ!4a(@#H!+3@)F4IH?#@%#@n<sB!3
zf+sj`P{NHx+|D?-vnR|T{uKKLwllg1HPpZlzxARrTUh<@a}puPd27hRLqe<wHz}3*
z`9^9@nJ~_bbK`1RDL6dkOPmYh;Iy0_z2;}x{V5~zN;7p{7vFjry{+C1A6ic6$hcq?
z?t>y`IcGK|?4yMB;P*6c*3L1ngy9Fs=$HTB9V1@*T`uiY+(iG;&#66g56caFb#S`@
zXYv-Fet6#>a7I}#DBwik;JQ)mT~`iDmCJ6|mrLNNg&QkE4;E@OI}LhYQMe+tu&WvF
zRvcWvU`0cVFYD&DHw|{_>9$sWOotn_t1c_^f(LlC)5ZQle1jvkaM_h*R4%ZEV)@*{
z7L#*K3vVyrq~FH}+riq3&+^El&UEmmsJO4nj98#vITG%^EA!@79^B)F4PgQCx%H5>
z+<mfeDZGBmr*EEmG^8_JKw9W<S--F;D%YBmf`k3Mo49>`;Slg*PJ}Hy^>Fb`oQ2Uk
zu_|x;^FhrDB^TcibR+wwVMXE@gzz6budlzXhiw<<;uiK<!}TO>xi5)FdAx^YP@yil
z_O`*zwU-bj<k?%^`p030dP~LQ<yzNXf2-N$H{Qz6BZUVL!?#$!n`4>2iwhA87ysG4
z{pjHCPTeEKoeP7FF|Qcn7$hu-`<TLrd3+P28^Yk$Ft|1hZVZE)!(e3?+!6+>!r-Pb
z_-Gic2!rdx;I=UMR2bC5;NxNNu`swO46X}<&xXO<!{9Sv@aZu4WElKX7`!zMmW07C
zhCww9ek%+v3xlh}X0A9M{{3(m{Bjsv8U}9+gNqfMhkrj127ev~UkHPDguz?FV70z8
z5&r#C{d*|<t6#w_j)i~U83ykTgUiF976zXP-+bPNf8P}bzZnK^3xoG7WH|hLNf=zE
zY31W@hv}aSgExi2pM}8{Vel(j?&IO#_vqir@b9mM!LNqF((vKbL*d^m!(f@dbR_(n
zhQY6g!F$8tH^Sf=1s@Fm{;vK#8ivo0gb)5W3_cPDdZk2hBK-ToF!-G?cwZR&z9K!L
zNE<cWqEDU*fxj09zo6;I!f<Mr{y!Xs&u<8Uf2faR^gHzNo%-a_5cmgS@F$uc5C8tW
zzOXj@`}r_f9tJ-f28q6;_XxM_z2TEL>$|t>bG;U$crtwaNBUr&hMB%Hs>*9Oh0yQS
zCwGV8Y5frN$;t44q18V5aQx<0eQAvXzNp~?8r~m+w<~x|AL;+15&oartKn{it_utM
zi}3HMtufXPO{~@Mt}ra(+TExB6*_f9|KFqGrZAlPl7?FRsgdyiscrh?%VAjT48zkg
z%kR}E59*Ux!kW+N<N74VQW2e2@=o6qS4*=f6#TqaYwDP0qy-mB<n!ypC(p+f(gLRT
zhfk&+*EeGV#`Vd%Fg&GVKKWGm|4dw%^74$9QYeF_4l3?u#X1~@rw@eT$y}eS_?}nX
zB4+D^zPTq1pD)Wn3p=S?EFM+x!x|pa@IXx8SRCnFC$++-l?%n~3fN@Bmy0;xm`SSL
zQ<M5!>ooO9oQJ-2I+k5*0x=tg!dIWwa7y*zq{{KMO5~J2KeZuz=d?aQ`&eAEs^4@>
zoz+L1RgX@`dhtblv_`MKY}xg36&}=RG=|--x;dr%KdT>mZ4GOtQ_AkCSj+DZOFpH#
zFcoWGT)vIL%CfVu%$46KRkhBlOrFz>URo0u{&*O@dP3iw8V$2OJE@N*PbhT?cus3@
zTEWi^DdMoQpwPTd#pS7pPAcuEW2hpWRIQti)l?;LQnP+(ZCnHOB2!AyOKSKpscJr_
zmT~f6{F1ttvuXk-72&j+&vPp2m$aN|mFw9farG2?O6B^s7`!i}XG-Na73aMxe)SP0
zR5kdl>i5%X-R%)Au6Rh9qI_y?QdO8zeSSgNe^Qn7ITgam!*Pbnlxb!8d1e1ut@88}
zA-FxH3U*5Ce^xE#8ygfbqHn#VhIU?!=sC59v$581Q8Lcz)t9sPYZ)iCZd1cr?Com&
zuO3yc{F-Xr*|L&rX45MDDLs_b8dk=gRH00%LpY_)z-j$|T9xqRG0k_A`iSSWjx8H&
z-n|MLR;s5|XlE4hX|=ZVig8{U{#<O5!}`)GRqJVG*-NS-(@Mc9-S5(}WkMR7`)1|u
z)0;H0ZakF2d1e18rS04<6`N9fUNQWkyTW(d!%FD1YEEZ-uRa@56@C7J_=R!J^8vjq
zyk%n_bx5C$DwAJ|EAqfTech_|zYv!(uC8}lb><nh&*n(X>sS$#y)zS<IIj|Ej>PV2
zSevZ&NG!(vn)Qg1UnuKe8qpdI$MRPfe*VkhqnQa6*4Zpf_zBg-_J~^2S+%XR%Biz4
z*EPp!eaY`uhWLpuX<<XEGtb0MF!nve>p~_zqb+cIL`!Q~CA2*pYu)HhtyPTs++E6Z
zebgR)Fif0P-FiB%=#khL#;$x+{huGtweXRcx-Dv+okx{rCzXrS%8J%~TDzCjft^>E
z*u8IeT<`nBs5zk)dS1EEc})FX+s1xIncg10GkmYRMHx7yyl;;vf#=og&&3@ImFC&l
z52!Ft#@eoC`E_OA$ymSde^fKv9$V3x&0)$<s6{`&BP9OWC*wyd?dM~*3~7I(yIIL@
zj;Q)}wy7IEs|we)D$cJerPHy~RMDN?9_H4v%Bl8{Dqm|z>1_|GyO>t@^rAAhJ*-}I
zN_DV3q~2ya7WD8_TJJ4dg;x)&gq~5Ccj|ZuX^$vjtsylLe?YaQb%(Oyv}W<@QML0k
zDoKA(DQ~S)arlFAi6dn#P*yb$mrGXf)E!gZnct!Lx9(I&Kf5VbgyAoSRhS!6JU^kr
zI;pPWrC4DOYt-GLN<F1*!TB9AXH<D+H^q)cC!adw%F|bmt_$DrM^y#pHpdDvq^X$+
z)!MUa^qohw!T*ZZrMq3V<D6Fd)uUzmRHgiSxu>x)=2tBGaTST5P^Ei1R<I#ehx4kU
zGZV2@+1OW(s_SVUj(x#iwU$`nv?_jF898&b%w)~HJ*<xE^b=a{kc!J6(gOcln=3ya
z*JY1_I=Q-z^ACit_jYNy?ICq&&nVJqHMFU+th9RVVXe^FZ6Qu`MD=TCLiyJoigOs&
zmguZ%cV|MK;q2YY%+?xJ{L|XfwTINXw}<wGS#>t3c>KYbM#btJJ{XdHUcL9Zm^7uj
zWi<ysu8`Bp=9X2FKO1Xm+!ctc6*IJLRZ*KGyTUwA#o98i-eY!?a-g*?3qeJ!4C5--
z*6p!z=?i`$mdS|nygS;^itmpl5IdNY>Lt!AciKZLwHMVt`Uw@l%!K;7&KKjBN1t`J
zt4f|#)%Fug^I7dgw1y6aMfi!w!iqf|_c%5Uh0of<F>%9cx+k@>aW3|Faf|Yzc8^bM
z*XOiaY<pPswlx%cve+*T#k#yDX6S?(?n%|hxrepoX-=w<omI&_Up7>YI@`;3pl#BW
z`l{9%b-wLk6>57(dEQy8Wi^i|0nHK3>giZJ53AjD?ojD>pHOdeRtxrJRy8yaKcN}>
zgDUc~ntkhbbp!3;n3?K<3RT>%Xh|~@O8wuckv*pcwKkPoE2W?{8q=$Mn;q90H4m4~
zU#s$x%D6qGD&}`9D+=||&7(1BM74NkLJ62TtOa+TQkT;nQInZDtjgbcRN2~&ZN~48
z9n~TAw(XI)Q@>6t`RY+6;-%OW)R#1m#$B!vmHm*mKQG4Y-=U3JcT7E6^O%}l_r7x5
zqFtt^W9KlUguObUM*mFNnibT!ORLt~t=8qo%bf(Jr9Gms&S7;boukT$<_R6`yr6pS
z$F&)n*5-RkwYalib?g~Uw1-uTTkFdC#ilo*o~Sjd6t#xS2CddH`-tj`KbVJVKdq#+
zw&;+f^MI1s9?=%<>9`}H$}oFeAv2TudUGW1YwHUymU*mVm>W@RJFOn#s|r6K>)W_;
zx;>;#{K;4uj;NlrhsssgPQ;XE*V(NMncc14v9(oOhLftKGm~*!cR2PWdsI%Haiz`g
zRdsHUsEE%gXFBWj+1wrK!Cq9~H+P4U)7hZh_Y<mfEvs3zhqP9m_3C{)cPlTRQK8Kq
z+aEinvUgBMc1N|fX->o~$3dlY=1A-Swx~FoBV~skH}^_>Yg9S)qPD}Gjf&OTxJC27
zTQz0o$c8Z0+M;rvZKx^q?odK!PAJc2Cbh19LUrW4imGYVSp2xAT0`ooo-K1<^|H5T
zD9pHdLf`QFRacsm+7<A-V-r_%e<tSb2CY?VyZQrvP%ALML&bXGPNljVJK2|1`JYu|
z`kGeWkE`mnN0dk1T&Zmyi7jD5omzWX>)RUAH#<+n3K_R2?O_#H_faLYwMCnd>DX6I
zsA_a>j|V#Hq`n$^$xX`pnPb{#>onBro5v<Xn%hHKbZd(`s^*Efu=QI**sI4h$L6H^
zwU=U3+osQEjwtiH>r|XGM^x8)yVSh>L~OgG`buj^`++ZLJAYDN@2pjY?W|RSPitcC
z4z2K6ZSwrMQrRBa7*@kis4cdJCbhyNO6}Z;Ds1;IRlwGL`ex@Ttz2vLq43R~RRPY~
z`@_`SNIXJZqw;BOeKdU7+N?rp59!0^gc?ToG1c(%>T{;F_4pgj&>z;fJEN)ve!uo<
zpV6vz4ydr^*K0;I$6|Tz(Wo5{rkY3NhD&SO8dYtXA6CiuJ<7P52`%zzHHveppv{q3
zpAS9|R<{{zjo+>8_6OCRyE|jC-laP1CseGx`?Mn69WirvD9_qMYVoI3z1t(Iy}c7E
z>iNAY8Glfz>FrWCF#CWCvKMo1_96AS=alMcwUln$L48KYu+M3>orcQAPsEn8R{KZK
zs2X0_qr|=(gSwAv=Chkr*?a4?S!o_?gmlh6rlM|bQ6rgMr{kQn+PIul^8A6=U`O=X
zw0bLlDAu>lnM!uOrrK6n)gD=+`Nj=iXMNlZ?NMT9x2w3HR)^NvsM<QMwV%(Fr0!1T
zMt7SQ-OaU9osBx0eJQTtc<f;gsxr=R*1|g5j%!6m)jf1aV~*{J*>g}!Ysa%{owe#d
zXSb?`c210hcz(Z1(2uKhd;2xB<}qyydQYix_;IaRYgC2&%4U7DyIs9ap>%iGsp+26
zj9)#fO4Qk@pym-Zwf3;Gx-+R1w1@TK%!EFizf*I(ut9aBc_QYoRYyCwO;gQSi)Od!
zfb^__+QS-kjwr=*o3+Z_F)h3un{9JaweICybDlY(z22{Dju&=mq3sc^$=n@kk~2q?
zZJmu}-aoD_b*u{UIHo<MUahlT2|lIP?e~{Gv%WZUOci5puX=@+Rd?3fs=VurYtQVI
z_Ml%-f6<&&`{;a0m8-LDn+kK2`fI;m1>M;g*I@4*A*^{!m8LbSTGiUD%HDlsS6J0k
zvF~|6z3c45DuCIz6EYVYSu0akbhfH$zOs2N%yQ;PY*UZweASEUVmim-*#OO>HKxRL
z4yg3HHkRJunAO8-Bkg!lK66YN=?`k}^cxE5*rQ=7osFt#%@O5@kF9iWtGWz-NNH<r
z(VF}HO5$wnHG5xHZ|e^#7v@K`l3&$Kx@%%~ZCAtZZdH~vC(BjP;l(-SWPA8AC3HfC
z-f5_Vo&S>7YvzRd-YMllb5cd$JgPOA+o8p@hc%O#V=Aidgfe3`wo^Z$?si&ru{*Lh
zTy5>|h;790iSvxRInBu(%7$F^uNyZd7d9%R&Z>H}Hfg=uBXKQb>9?(Va=%A)p&id=
zpVFH9@t9o`I>Y$4N}?ZEg_z&j2n*|OIuS-bHtV_jl#W*)QL%J)#I}A|{Xl20O1e9w
zCf|v@Pj|E0Pw#HEt<I?WfzHv`PU8l$JyLE!^v&75%D(QH($jrBHu{G89e+#>)lX=1
z{Iqhmvr+ln9#N4st@fEtX)D-yO7mzQQ?h0s&{St_Y$ll!GP_ChpWPaJz%gZPZ=dRP
z=eSn3J*+0t&6SBw8#CuVWscwTc*vaQVO65eS{>k@Qb9J4X{*tSyO`|}C9a#Pl3ut|
zT|o1g4ryOJq(U9v5=O19YWwYBEp2X#>WiPyhuutb?u^DVJFGJ3tk<&V;_hl^_p#85
zUx+PF%WSQQ2Sti>Vf`j8Ws_FF6^o<0L2K1pr^NJ5=qq!vb@dLY;9BdHYjY2((KnAB
z4s+?=tsI?SqsrF0T?uXv#k@G74c@bw<=ms{Heb*lLU&ut`F*M_FURvzb0bQcKc?gL
z(<(lHP~BT6?jy}^SIzF+qcvy`sU@_BwE^qpTFJRRF~M6^pV~ue=F^JS-L9UdHM%w?
zKNj$)W-_}+c|E&MxzHX~Ms@Ct9oj~92=i;y6MRKqY;9I8oZq5!G)L5`W{<~p9@jee
z9*ZTlJ}x!(Di=0tWnX?mh1olyW9D<Jc(eCw*|Y1EugxRcD9mkE%j?~#>f1b`%=QPh
zH7~Ti@O#vub=JlTu}8gQ_mJwv+*VD^?}+VuuRd%(pf=OZ6>m0{#QccDI(MlRH;*Xy
z=El{=PRChpS4R2$asNZb*tshn5ve72GcBcaT;(vIDWR<~_4Vy|vbOuE3Zi!+R@ben
zfSqwwt5?=*eOl|)lXtdjCFkx?Zg*nIbvMO~%CxStkE%bIePU9@IjW`i;*nVEcD1MZ
z+hcu+dzqbv%B;6Xm9Bdv9^S@u@820_(z#0|cOmB2{5GxMg(p<Tt@Wx5y}e3#+p50L
zZc<|X9?iKuQdUTHLl=fs!Td4Rp4m0pv`(pyeO8CcFQ`nr4`||9&9=4qJ|$+eDpGgH
zt}x~Is5F|#;;J53nr0@qgs|CT>M8wxwXxPZ72hjc)zrSB6`mVY4QkrBv{9{aYgBpu
z@}WnxfMd$@rj1Q!i@MKq3ToY_%<9~8EPSz(YbJBYmB3CTw(@ZuHlEbl_4cVQbk?eK
zYsH=Y=1A-@Mzm1BPo?PhtBBe|N@}~@oF7)L@%!V4@%&qRSi5WIltay<`bK+1m9n!o
zR{kS8q4BJO=I>P7nw`{CcT;Rm$CN>xyR>IEGoev0S9Y~F$6^~P+p^l&>?S33=7d^y
z=h0X>w`v`_Pbzh@kLa|<Gb)b@Pb%TfN##k)DvhlzajC~t68@k%$NBs6keBm2VnT-W
zmH9DcUu#oL)>Ha$F79GCt-ks4A@vBI{Yp@GyIOx|zbZ`gVC=Vs;u3Z$33JC)p8l!z
zVR_w+s_&<i*3Q~EuUPQ&nX>PkQZRE=*9D$a>G%nq?>VhCZVxNFX5(d;=3!NunRpD;
zS|4k%R;oFv;%jYD^_`8orZb1N<hgCS7<EqHXzo;})r%*g{V`SP`TMoB&QTSkA6NJ3
z$5n4T4ORE%iI_mUE~M<`Ct^(+)_(RgYP2(xv9V-Y|MP0rO&i<M(M=(iKe#rGn(>Q1
zwxjtqTIl?kiecukPEMRuQ=chMSdA-MYqMt3y-(k7Syk})9XgnKMyonEthzG$l$PCD
zue|6ypzg5qn7W&}wXvf(pvu)NcQ_BLzvv!R5<B}<yJoYvxF@un&baDk=ZHRRT9sRC
zb6n6yHQ3pQb*|;CzTbUJv-IOyqn967TW-aj7(cF>Ff*y<-rc9woZGHh`grz!Zku{o
zzei`&&nau#Lt5wVqp`lm!@lmQ3b6Z-R-v2eK>L)c#O&QFDnAih<6&ju%#j%TUX@jE
zor=dFTNBEqJ**ni8C4yg+oA)dvzq^fM`C_GsP1Hbm#RxQ({g%opTE03R@F&uKKyQ_
zzdamxagHgeorCHFF6>oXZfzI~OYI&|md;t_-ruPz`3aS0_i<&-?A==2%!E!AoYlT!
zXR}8BfM(U*t}>n*i#vi!cW>_(!|Z1t)^7SawVYQr9}J&$HmZ60cs0GZOFMhL{W_C>
zQnkG^sy=1*DQ%6<scUFXD0^m(#^xN?&5vsp+wpW|cS89&GpW*>-5Luq_Jcm2$L~zU
zY<ft|urr~V&EKc|nU5DlnpQc|x=&lU&ViWVjmoXg5oK>{Ypj5ys$RYQT8Zv9wZ>D*
z`Q|ZQ8GlCgVkS21wvCC~psLVaqgi$C)>?KSSE;uq)cLmKW%JJM`mpz;BKZAf>Xdt}
zO)=}gqzL{g<#ubm`o;@~V~?Y9YsSNdud3?I#Z5#j*0BrsZr4%Q0~^#TyKxV)dst_=
zUR0%QS~Zr|kj^SS8*5d(;58Hb_|AUSvCdI-NxcL5e$T3c&m33!n}?OZbL%wA&RTug
z8q@XCld2z2$3}8gt-N(mD?YnMWprV;c8<TgS5+s|H@e$(750?sb9Y3WfR<HLX^+HW
z+pkWvcT6pR{{C2`Yt&(MpHOX_dq`_FyIzZ#ji-3$x2t>W9*hMyp|s8K)r$K)>K40C
z#p=~iP3k_R1$8&6aJpmK3U=>Q8)(KZb}nYd-2JiIPsExu9y6mnlfPa$+bK_|?$Msz
z+@#KFzo_%uCsk9MlQA1M#-<;y7Iou(MYF6lcd2g9#qMhU@wm>rRV=e{+u1oBcaU}}
zJ9>Ln!)N1hY-gJ?;KEu}p>L>K^^WUU>e+Z+K<nL#CyzTv)Kko_QUBDnTFLIC%Anqm
z3gp58CFa$ms$-pTbpig6K5UJ~zGAYh(#p}+7G-*OgTh*Osa|!)Rc77C^x@2;O5Vql
zqH_=GkmqYkM)Rl|W9PBW>VCGX9{OWiYVRJ!eq~H+>-WWKc2EhP+oA=${FK68J-RJK
z_fM+2&FzZouvSx@{mQ*=ELC5g+gPt$pE=wJ@w#`bm}d7XBRcU&ycsV8wTHEy?IBf3
z|Aew-Hg+NYP~3^ytbV0Cs&#4Ip>KBXP*rXoQFqzN_03M~d1oe7U0*$+T<s03!}L#T
z>D?!kXn#o2dk2&g-Mf_9&I$D_vj?^E{@9ojeN4T1v%FGsD5hb5EZSTh%-nt2%FaxV
zhX|dg)a1HHHP7CiyY=B?TBXionss-b)_Qhp+)&)7POBSlspzbYxf{=t`Y-D;=Q%a-
ze^3kcd$iEbVKr1A&loh1DOPX2c6Xl9iSzSXax-2iZmk*CdK`@vVKO$;tvW~k>_bZE
zZgruvn>43a#$v_Zt}N*tQqsCx)Na~CvH9#(jq!W5BE2IIhq=t{Q8k~LROYqTDxrR#
zif+!TAe-@WZfC!~(%qms*@;W<99Mnq?NN(<`H+^-iG@5H>vv~|^02c%_V@=?ApV$U
z-Hx{e^x{de=1!$@_F;9Wy}MNMvxihC{ZrcBbjLP?weIfISNwkUofjTat?+yFVS89T
z?z1{ScTNqwbEk^G_oTkwwA;cw=C`YlIHjJic~mLx#k1*iTlK}63H30Y16t<XHl@*z
zs{wRwSIwW_tW?d#?7NVwwC3+u0kyWrmbP0}xBIy2*X(iaqMp)5>KXL|ofDZ#V$;?z
zx^Sl|=Y@xKn(36*zq?CSV{V&TM(;qZD2KJ?oyWDX3&V=moKzmnOsE_?qw0$~M^pp5
z+tmY}FLOSo?Xc3_J)jEvw5miqUf62IBhQ(5lU?Tym0R<uDnW0bYE&y;k7^BT{<BYN
zp)bcwXj<ik->n^^=T(T!<EnbCovL%4c!^+slh(W)Z{#?wT$(v@PsoeTJ?dFI<Ejb0
z1DfO9{VIT$^LPxTCfON{O?kaG%3o8+*OaTXN3|>;Go(4GJnJ1-D?O#N@_(ng(Tpdn
zn{nT<vr!e`!o6Dg)A1smDxiN%<?iFosNWy^)Ny?{yIq||@1gkNj@W{8^?$uxI+1Wr
zea(EXB+tat9NkQXJokvYKR>R~EAbR;XUq1G5#23X@447bU)UQjX6l%v8E;jbJ+2(+
zoroE?OD7bbQ}+4&F@uNI4rh0%_nJMZjBdpmGPgtX>TXxdXzhqKYeMzDIjOvu-4rX|
zlbYw;6R}1fQLUKUre3Fer@q<TsReb9sJ(O##Y{O6d*H`ZD!q+L*4&rW;9pr2JEncH
zy~KNq=B&<JeO((0|G2&}w{=w2Vn_@3ag*<#R84Nie!p|OlF&J#?r`?VgStMVT<<-s
zMt53gOU`Oe&B<6)k80)SGS%w2FR7@$rrEZ(Xy@T2C88B~CuibqK669L<mc6BI#0zd
z(?Mlq?{Q7d?TX#hL7jJaS~m?mtCseHy3N^m`$X@ATE~3srTlTNQ?tCx9nYS(cB^=2
z4l5<yoiTNfDZhIsV&UDZO4$37%HriGRE^q0nssNRminv;bY@aDpmR^`&&QN#e^5=c
zvsNqK*{|Bs8C7ZdC)G*#r&O-Zk)6s@<%mD1y4$o`h2|kO)XrmCg;z#wyee|9a@UWm
z)MhuT?R`x>)BJ<_W@o+nuueR|JsS^nn>+7SVISApzIs&K$WvOOcDz>Ki#Oz6h?i;R
z$KyA&&)12&ZQa;CU&uB8=7gGX_bFv?H!i!oMcwZFHZ{}kf%uhp0ri#pm0`UN%9~z1
z!uPR=o0B79_1f`N+02nx?KUbxXH5BU;mc}x7h(_K;}PJTRXdvBp>O(lPr%HCHmcpt
z%8uSXRnED1Y3=2>0i56Yh_WM|^XrbPtz7uBBFw~lVVdzATQ}Z-^GbP6@=+D#>}H(|
zcsjQ1-Ekk)s)hFAzTEtMstDaD)YLA-Gl$J%D)aV`a<v`r8R;EYweFoz1)7T)F~3#4
zb9Ym$M{zqbJGnKyo$S_?{A>w77Ie!+AOHBLZn}tH7cXD_5uV&LUkHQ${QJ)w_|F{p
z&m7?4*H7Mj));$&yl}=Ou*l%xVg4$9r}9@kAd3Vwq^QGU6b+0a!#?DgfWsLSn1)9W
zeY^&9(wJ|>8}L0?gdfIY{1hs96PDoZsNz?#6u*PyWL&>x{QrK`@Bu8xhmqo=Sb>{R
z$E{e6&%<IEYj8V8u^t<6HyZdNHen0Ka6h(VCo(*UaqPi9d>M~mKXN>dgLo1XID(@%
z4u_LCg|FZYp24#?ivlm;Z}B3g@pb$i{vIAJ%%F=NUdBBB5q<m<{u%#**YL0SH~c%y
zlri6gZ^pOaTk&o9c6<lkfbYb2;k)rY_+ES;z8{P51NcGw5Pldxf*-|?VKIIjKY^db
zPvNKWGgD=~;{Tt;8}V~^6W)xU$6K%jZ^hg23wS$z5f|YdsN$FK%lH-iDt-;Wj^Dsi
z{3d=2zm4C)JMp{tJtVjom*7&o3-88z@Lnv#Wq4n?zVGM%%W(y+L=As{t8g_wfDhtB
zxCYDdhxjA>F+Pls;7{<UNO3K$!}a(mZotRzajd|N_yj(QoA4>zjFqV4&u|M?;a1#+
zPvbLKjnCq9_;Y+7U%+4BFJY&Qg&ATDGt3xfgfYw-#xQFc!`#jo<_^X%ql{tJF@{;s
z80JpKcg8%k8<@U}X_&hi!)#;>(_jp94`Y}wGKRUAG0Y~$Fq;{}Y+(#@A7hv?#xPqM
z!`#mpW*cLe?Tle|FoxO580G=SFd1W*U5sHKWDN5q#xUcIVRkcy*~1uSFJqW}jA0&P
z4D)5iFb^|^d4w^{ql{tpGln_97$#>7^B7~8#~H&s!5HQsW0*sXVV-0R^Auy43C1vo
z8N(bY>**xZFh?1~9AgY~oH5J^#xRaC%wI8vImsAiiZRS7#xSQD!+eD?%vTx1oM8;}
zG-H@&7{h#xG0d}!Vg8yi%vr`T&oPE67{ffz80H1WFn_}s=5HCpoMR00B4e2IjA5o3
z!@R^8=Ie}MzQGvg?-;}USH>`Z&lu((7{hqRFiplVEygfy#xOICVLFUqx{P6F8N>7#
z!^|;;d6_ZH1;#M*jA3434D*kSVP0hn(`O9x-x$OE6JwbF&KTyO8N>V!#xVcF80LR6
zhIx%K%>QBx^RJ9y{x@Toe`5^ue;C93J7bvt%NP`}Nnqh%zKLmAIGAr{8Ws-bTbPE0
zgZWmbVc}rDjcHi8Z#$(vKKXX0;b6Xla9B8)H!uwg2lJgw!@|LQ7t^qCFyGBIEF8@D
zFbxX_^Sw;N!ohqW)39(b-_JBG9Lyr7Vc}qYfN5Abm>*;s77pfzn1+Re`C+DE;b4A*
zX;?UzA7vUA4(7+0hJ}M!%rq<<%#Sk-3kUNPOvA##{3O$`a4<i`G%Os<Pcsb*2lF#b
z!@|M*7p7t1U@A<*!omD3)39(bZ)6%44(8{WhJ}N96VtG8FmGlW77pg;nTCafc?;99
za4<`lhJ}N9E7P!WFmGcT77peYn1+Rec{|gva4^5fG%Os<MNGrO!MuZMSU8v})74l{
z9siqOVj30>=9ig<g@gGOreWb=ewArhIGA5!8Ws-b*O`WegZT}nVc}qwG7Sp{^P5b=
z!omC&)39(bzs)o(9L(=94GRbJPNreuV1AcrSU8y9V;U9?CSe*D4(4K}Vc}pdVHy?=
z=2E6%;b7jyG%Os<yP1ZCgLx0ruy8Q%Wf~R^W*O74a4?rK4GRbJKBi&eV1A!zSU8yX
zGYty|b2-zna4=Ue4GRZzCDX8QFg2!O;b8uNX;?UztC)s`gSnb%SU8vuFbxX_^FgLz
z;b1<*G%Os<HB7_8!7OJQ77peQnTCaf`6H%b;b8ulX;?Uz4>Ju52lEl8Vc}r@glSkf
zm_KD277iw58Ws-bTBc#)V6I~t77pfmreWb=KFTyK9Lx<&!@|LQjA>Xnn2$3J3kS1;
zX;?Uz8<~cMgZTv0uy8P+WEvI@<|d|L;b1<+G%Os<%}m3>!K`E&77nJ)G%Os<pD_&!
z2XhP4uy8P|n1+Rexs_>HIGEd*hJ}OqG}EwfFrQ%>77k`L)39(bpJf^r4(4-A!@|M*
zIn%IkFrQ}{77peMOvA##`~}mna4>(#G%Ot0xib|cNMVt|!NUv@jsz(zGB|jcVZxCh
zg+&Gj4>LkI5~Q%m;NW4_5RL>XEHXHFn6-o>K?;iu4j$%q!jT|_MFs~Ca|hu_kisH^
zgNGR<90^ibWN`2>>j+1J6c!mAJj{B+ksyUd1_uvwC*eqt!Xkr%huJ_l5~Q%m;NW5I
zA{+@)SY&YUFn1G<1Su>sICz+igd;%;iwq7Pra?Foq_D`~;9>3|90^ibWN`2>UnCp}
zQdneg@G$ohjsz(zGB|jcO@t#s3X2R59%eJ)NRYxJgM)|JLO2qnu*l%xVeTUw2~t>O
zaPTl=gd;%;iwq7PW-H-HkisH^gNM1Fa3n}!k-@>kY$F^AQdneg@G#p6M}ia<85}&!
z4#JTjg+&Gj53`eSBuHVA!NJ2kKsXYlu*l%xVKTyzAcaK+2M@E0a3n}!k-@>kJV-ba
zq_D`~;9<T*I1;3=$l%~%#tBD)6c!mAJj`ywksyUd1_uwbhj1iFVUfYX!|Wv-2~t>O
zaPTnu2uFex78x8o%tM4DK?;iu4j$&qgd;%;iwq7P=3&B-AcaK+2M_ZI;Yg6eB7=j6
zd6aM@NMVt|!Ncq)90^ibWN`2>2M9-k6c!mAJWNh_9`}D!{<p~B;9(vk9LZzxJC^@5
zICz-H2}gny78x8o%oBtoK?;iu4j$$p;Yg6eB7=j6IYc-Tq_D`~;9;I590^ibWN`2>
zPZ5p;DJ(KLc$f*oksyUd1_uvwm~bRWVUfYX!yF+T2~t>OaPTmbgd;%;iwq7P<|yGv
zkisH^gNHdrI1;3=$l%~%juVaqDJ(KLc$gD}BS8v_3=ST~5sm~YEHXHFn7<+%2~t>O
zaPTlE2}gny78x8o%oO2BkisH^gNHdqI1;3=$l%~%P7{s<DJ(KLc$lvcjsz(zGB|jc
zuM&;~DJ(KLc$hPUBS8v_3=SUVX~K~pg+&Gj5AzJ+NRYxJgM)|p8sSKg!Xkr%hk2H8
zBuHVA!NJ4)HQ`8*!Xkr%hdE0)5~Q%m;NW4NBOD1*SY&YUFa_aAkisH^gNJ#Za3n}!
zk-@>kyg)b-q_D`~;9>rTa3n}!k-@>k{4L=~kisH^gNHdsI1;3=$l%~%UL+g|Qdneg
z@G$2IM}ia<85}&!G~q~)!Xkr%hk1!`BuHVA!NJ3Pop2;bVUfYX!+e8qBuHVA!NJ4)
z9pOlj!Xkr%hxxCBBS8v_3=SUV?+HhO6c!mAJj_23jsz(zGB|k9$V>$ZQdneg@Gwon
zksyUd1_uw*A{+@)SY&YUFm1w-AcaK+2M;qtI1;3=$l%~%I)o!Z3X2R59;Qn;5~Q%m
z;NW3q2}gny78x8oOpkCRNMVt|!NbfEjsz(zGB|jcmkCFL6c!mAJj?~cksyUd1_uu_
zPdE~!u*l%xVO}8|2~t>OaPTnyNH`Ltu*l%xVO}L12~t>OaPTmF!jT|_MFs~C^WO+Z
zf)o}R96ZcF5sm~YEHXHFnEy^V5~Q%m;NW5YnQ$aXVUfYX!~74zksyUd1_uxGFN7mO
z3X2R59_D`%jsz(zGB|jc*9b>~6c!mAJk0+h90^ibWN`2>|4KL#q_D`~;9>qZ;Yg6e
zB7=j6`8UFmAcaK+2M_aq2uFex78x8o%)b+k1Su>sICz-<OE?mwu*l%x!3NeWLIqVM
zs3Aoi77b*`;ZVS%5A#jLM+H?Rs3Aoi77b*`;ZVS%5A)5$M+H?Rs3Aoi77b*`;ZVS%
z5A!X=M+H?Rs3Aoi77b*`;ZVS%5A&_WM+H?Rs3Aoi77b*`;ZVS%5A$urM+H?Rs3Aoi
z77b*`;ZVS%5A*HBM+H?Rs3Aoi77b*`;ZVS%5Az+wM+H?Rs3Aoi77b*`;ZVS%5Az1%
zqk<|D)R3YMiv}{}a46u>hxtz8qk<|D)R3YMiv}{}a46u>hxsnzqk<|D)R3YMiv}{}
za46u>hxu;eqk<|D)R3YMiv}{}a46u>hxs1jqk<|D)R3YMiv}{}a46u>hxuOOqk<|D
z)R3YMiv}{}a46u>hxtC@qk<|D)R3YMiv}{}a46u>hxvZuqk<|D)R3YMiv}{}a46u>
zhgn2?R8U2N8dB6@(LjbA4h1~=Fh4+iR8U2N8dB6@(LjbA4h1~=Fh59qR8U2N8dB6@
z(LjbA4h1~=Fh4|mR8U2N8dB6@(LjbA4h1~=Fh5LuR8U2N8dB6@(LjbA4h1~=Fh4?k
zR8U2N8dB6@(LjbA4h1~=Fh5FsR8U2N8dB6@(LjbA4h1~=Fh53oR8U2N8dB6@(LjbA
z4h1~=FpG(g3aUs@Ly9^q8px2tp@2sp=EsST3aUs@Ly9^q8px2tp@2sp<|l}c3aUs@
zLy9^q8px2tp@2sp<|m1d3aUs@Ly9^q8px2tp@2sp=BJ2{3aUs@Ly9^q8px2tp@2sp
z=BJ5|3aUs@Ly9^q8px2tp@2sp=4XhH3aUs@Ly9^q8px2tp@2sp=D!dh6;zR+h7@&J
zG>{>OLjjLIIJh>8P(c+5YDiIsMFSafI27>c!~87qQ9%_6YDiIsMFSafI27>c!@QCB
zsGy1jHKeG+qJa!K913{!VSbMIsGy1jHKeG+qJa!K913{!VctZ1R8U2N8dB6@(LjbA
z4h1~=FmEP4DySks4JqodXdpulhXNjbn4c#;DySks4JqodXdpulhXNjbn70rg6;zR+
zh7@&JG>{>OLjjLI%o5_Gf+`Z!kfIKY1~TMuDB#hDc`NZzK@|yVNKuDH0~vBS6!7T7
zyp8y%po#=Fq^QH9febku3V8Hkeu4O?po#=Fq^QH9febku3V8Hk-cEc}P(^|oQq*D5
zK!zL+1w8sNzes#kP(^|oQq*D5K!zL+1w8sN7ZD#7RFR;D6m?iMkRgXd0gpb+JBW`8
zsz^{niaIPB$dJRKfJYyuN_<pMMS>bq)M3#;h8zwBJo+%dM0`|GMS>bq)M3#;h8zwB
zJo+%dOng*OMS>bq)M3#;h8zwBJo+%dLVQ$EMS>bq)M3#;h8zwBJo+%dN_<pMMS>bq
z)M3#;h8zwBJo+%dMtoFIMS>bq)M3#;h8zwBJo+%dPJC2QMS>bq)M3#;h8zwBJo+%d
zL3~tDMS>bq)M3#;h8zwBJo+$8iH{1ZNKiwHIxHH<ki(&XM<3=liH{1ZNKiwHIxHH<
zki(&XM<3?5h>r@YNKiwHIxHH<ki(&XM<3?5iH{1ZNKiwHIxHH<ki(&XM<3>Qh>r@Y
zNKiwHIxHH<ki(&XM<3>$#76~HB&Z=p9Tp8_$l*}HqYv}D#76~HB&Z=p9Tp8_$l*}H
zqYv|Y#76~HB&Z=p9Tp8_$l*}HqYqB3n?<OgiUc*JsKcUx3^^PMc=TZ|CO#^tB0&u)
z>ab`aLk@=m9(|Zgh>r@YNKiwHIxHH<ki(&XM<3==;-i8p64a2Q4vPjd<Zvk9(T8~#
z@lin)32I1DheZP!ayS(5=)=65_^6<Y1U00n!=ix<IUEXj^kLpZd{j_Hf*MlPVbMT_
z91aCM`Y`V$J}RgpK@BPDuxKDd4u=9BeVAp$M+H?Rs3Aoi77b*`;ZVS%4|5suQ9%_6
zYDiIsMFSafI27>c!@Q69sGy1jHKeG+qJa!K913{!VSb<ZsGy1jHKeG+qJa!K913{!
zVct)CR8U2N8dB6@(LjbA4h1~=Fqab_6;zR+h7@&JG>{>OLjjLI%oW5(1yv-dAw?Y)
z4P?mSP{5-Pb0zUnK@|yVNKuDH0~vBS6!7T7)QFD?sz^{niaIPB$dJRKfJYzZ4~UNn
zsz^{niaIPB$dJRKfJYzZD&nJpDiYL?q7I7&GURY5;L(S<n)s-oiUc*JsKcUx3^^PM
zc=TaDKzvkCMS>bq)M3#;h8zwBJo+#nBt9yrB0&u)>ab`aLk@=m9(|Y(5g!#)k)VbY
zbyzfzA%{Z&k3P&b#76~HB&Z=p9Tp8_$l*}HqYty3_^6<Y1U00n!=ix<IUEXj^kM#x
z_^6<Y1U00n!=ix<IUEXj^kM#p_^6<Y1U00n!=ix<IUEXj^kM#(_^6<Y1U00n!=ix<
zIUEXj^kF_sd{j_Hf*MlPVbMT_91aCM`Y<0MJ}RgpK@BPDuxKDd4u=9BeV9KXJ}Rgp
zK@BPDuxKDd4u=9BeV9KbJ}RgpK@BPDuxKDd4u=9BeQ;6TEJ6iUB&Z=p9Tp8_$l*}H
zqYrZ}@lin)32I1DheZP!ayS(5=)+t`d{j_Hf*MlPVbMT_91aCM`Y_iM9~D%QpoSE6
zSTv9!heH96KFmjnj|!?tP(zA3EE>p=!=ZpjALa()qk<|D)R3YMiv}{}a46u>hxr)s
zQ9%_6YDiIsMFSafI27>c!+f0hsGy1jHKeG+qJa!K913{!VO9_y6;zR+h7@&JG>{>O
zLjk|ybbNmAM&hG_DiYL?q7I7&GURY5;L(Tq1o2Tp6$xrcQHMnX8FDxj@aV&QlK7~g
ziUc*JsKcUx3^^PMc=TazB0egpB0&u)>ab`aLk@=m9(|Zk5g!#)k)VbYbyzfzA%{Z&
zk3P)J#76~HB&Z=p9Tp8_$l*}HqYtx^_^6<Y1U00n!=ix<IUEXj^kM46M+H?Rs3Aoi
z77b*`;ZVS%5A$clM+H?Rs3Aoi77b*`;ZVS%4|5CgQ9%_6YDiIsMFSafI27>c!>l4c
zDySks4JqodXdpulhXNjbm|KaD3aUs@Ly9^q8px2tp@2sp<~HJ^f+`Z!kfIKY1~TMu
zDB#hD`84rSK@|yVNKuDH0~vBS6!7T7e1`a_po#=Fq^QH9febku3V8HkRudl;RFR;D
z6m?iMkRgXd0gpb+XNivrsz^{niaIPB$dJRKfJYzZbHqmlRV1h(MI9CmWXR!Az@rcI
z=fp<^RV1h(MI9CmWXR!Az@rcIdE%pjDiYL?q7I7&GURY5;L(Tq0`XBn6$xrcQHMnX
z8FDxj@aV(*1@Tcq6$xrcQHMnX8FDxj@aV(*CGk-~6$xrcQHMnX8FDxj@aTgZ>dYIk
z2#Zm{5>&Ai36`OT<w&stb*zTPC>j_;hJDB}0f#dvFb$6$`gjdyh~;Au7NdeCsA4G+
zEJF>;kzxhvSPhF&G%$t?`;cP-4rfqc8Xi6L@fyr9%f})tMg>by#Zn|#h8mV5#R}B1
z8Wy8yU<?`dA;$z9&Y-|FJbLKkHJA~Wk40FF3YMUXrAV+0H7rMp6{ur1EJo457&7cb
zjtMxNL4j#_^w7s^Fl$&o7GW_eSb{2+BEd4$upB8?ppMnB7)1kP$gmGNCg5-e1*YNA
zLm#iftY!IFgvF>}3949%1j|sva->*+I#$DC6b+0a!#?DgfWsLSn1)9WeY^&9JIlu+
zEJg)OP{mRtScV#wBgG2Tu^JYmXkZK(_94dv9L}J?G(39f<29H&SUwhEF)CPsDwZO_
zGSsjfDORA4)vy>v17par4>=~_a0UgY;n71MufdG6d@RCZRImh9EJcE4s9`x$tUw*B
zVKIsZ#*kqja!kPC3<^xcqlZ3TgIUM&u?UM%!4g!l6bY80hUG}H0(Gp0#V8sWLxz3G
zF#(4&C@>9=9{P9<W<ATtA}mG)OHjp9Bv^(TmLtUq)Ug^CqiA3Z8TKK^1RTzwz%)F1
z=&vvLpYLS(ScJu>U<s;NiUiA0!*ZlpfjU;hViXOGA;Ui8n1I6>6qtrb4}H7_vw`Jf
z5f-C@C8%O45-dXv%aLLQ>R1hnQ8X}y4EvB{0uE<TU>Y7h^zj<ZT`V7quox9AK^04p
zU>Ryyjub0U$7)!NqJc4F*oPbwa5#ek)9~n_kJn)CX8Blz#i(Ejs#uBy%TU8|q*#GE
zR>NWx4U8efKIE8y!x<EqhDQ&5yauz8<zo>Rqk<)<Vkr_VLk-K3Vg>404U17UFoq2K
zkYfT4XHZ}o9zFE&8cc)bV-Xgkf+eV8DH1G04a<>Y1?pH0i%~Q%h79|VV*(CmP+%G!
zJ@oM!%snh0i?A3KEI}1Zkzg5WSdJ7cP{(RmjG}=tWY~us6L2_#0@Lv5p^w*KzR2>i
z2#Zm{5>&Ai36`OT<w&stb*%pXIQoO|2;v9+=iBl_5IWUp2cbhl2T?<VMX)Mux6hNq
z9&*@24tvO94>{~1hdtylhaBdR!#t0uX@xF=K`;oWZLl=`ciN^k2$rS{O>1e&AXu8>
z^*FqqXK(N4vxj|lXLr6!N=8n>ijs<&H4QBrI=1xe82pX;1cXGyBn%lbX2O&ia~3Q~
z$;c^KQBqN}rlDm+$CjQQg9m*ALLy=khKv|9Vakj-3znp0<P@wZsi;}g(6XUpOV5tM
z-=t4KNJLD+kP%}hOqnrf!IG4WoPrf46*X%bS~hfS>De*(U+5DM5)qRyWW<;WQ)bLr
zup}iTr(i`%Ma`OqmJJ<SdUg!{W_<!eB4QGTj2JUv%8WS+mZW6l6s#zzs9Dp{vY}&3
z&yGQ%Pe4dSOu~>6V<t?QF=xS&l#HB$6(tolYZ_WMbZqI_G5BBV6A%&+lQ3k&m<dy6
z%vrD`B_pR`MM*`?nueAQ9b0;K4E|U81cXGyBn%lbX2O&ia~3Q~$;c^KQBqN}rlDm+
z$CjQQgTF<efRKoogdrowOqeob&VnT=894<jN-ApBG_-8!*wV9O@TgBfNJLD+kP%}h
zOqnrf!IG4WoPrf46*X%bS~hfS>De*(TlEPDiHJ!UGGfexDKq9QSdx;FQ?R0>qGnA)
z%Z83EJv#<}n?3;{5itovMvR#-WyYKZOHwj&3RaX<)U0V}+0e12XUE`wtxrHmL`=ev
z5o0DynK5U<l9Y^`f)yndHESALHgs(1+2Llo!5#r25itovMvR#-WyYKZOHwj&3RaX<
z)U0V}+0e12XUE`g*C!w(A|_$Th%pnU%$T!aNlHde!HSZKnl%kA8#=c1>=^tV`UHeT
z#3T$EF=oP)8FLmaNy*45SW!|@v!<bCL&uh$9fQA9pMa2vn1mrC#!Q$pW6pvlDH%Bh
zD@rPA)-<$i=-AS;WALO;KuAPP!jKVTCQO+zXTg$`jGTfMB^5Pm8d^4VZ0XrC_`CE8
z2#JVE7&2nagef!TELf6~kyEguq@rd`L(7JaEj>F1f44pXArUbNLq?35FlEM^1xr#g
zatc<IRMf0#XxY%QrDw<B@6jh9BqAnZ$cQl$rp%bLU`a|wPQi+jikdYIEgL$v^z0av
z`UHeT#3T$EF=oP)8FLmaNy*45SW!|@v!<bCL&uh$9fQADpMa2vn1mrC#!Q$pW6pvl
zDH%BhD@rPA)-<$i=-AS;WAOLs6A%&+lQ3k&m<dy6%vrD`B_pR`MM*`?nueAQ9b0;K
z4E}z70zx8U5{8TzGhxb%ISZDgWaJdAD5<Df)6lY^V@uDD!LvRAArUbNLq?35FlEM^
z1xr#gatc<IRMf0#XxY%QrDw<BAJ8WtBqAnZ$cQl$rp%bLU`a|wPQi+jikdYIEgL$v
z^z0b?gZc!7M8qTv88K$Ulo@jtEJ?}8DOgcbQM0C@Wkbi7o*jdKNS}a^h?s;SBgRaa
zGGoqyB`Fy>1uIG_YSuKgZ0Ok1vtv-{6A%&+lQ3k&m<dy6%vrD`B_pR`MM*`?nueAQ
z9b0;K4E|w#0zx8U5{8TzGhxb%ISZDgWaJdAD5<Df)6lY^V@uDD!9Su;KuAPP!jKVT
zCQO+zXTg$`jGTfMB^5Pm8d^4VZ0XrC_($~#2#JVE7&2nagef!TELf6~kyEguq@rd`
zL(7JaEj>F1FZu+8M8qTv88K$Ulo@jtEJ?}8DOgcbQM0C@Wkbi7o*jdKOrL;|h?s;S
zBgRaaGGoqyB`Fy>1uIG_YSuKgZ0Ok1vt#g&>k|+X5tA@v#Fz<FX3SZzBqbxKU`0tq
z&6<Xm4INv0b`1UreF8!vViJao7&BqYj5!OIq-5k2tSG6dS<}$6p<_$WjzO(YKuAPP
z!jKVTCQO+zXTg$`jGTfMB^5Pm8d^4VZ0XrC_}}Og5E2oSFl5A-2~%dwS+FD}Bd1_R
zNkz??hL#N-TY7d3{<r!Bgha$73>h(I!ju_v7A#50$SGJ+Qc<&}p=CqImYyAh|D8So
zArUbNLq?35FlEM^1xr#gatc<IRMf0#XxY%QrDw<BRiA*6h?s;SBgRaaGGoqyB`Fy>
z1uIG_YSuKgZ0Ok1vt#g2>JtzW5tA@v#Fz<FX3SZzBqbxKU`0tq&6<Xm4INv0b`1V0
zeF8!vViJao7&BqYj5!OIq-5k2tSG6dS<}$6p<_$Wj=?{zPe4dSOu~>6V<t?QF=xS&
zl#HB$6(tolYZ_WMbZqI_;UJyC9swZ{F$qIPjF~WH#+(I9QZjN1R+LoKtZ8W3(6Oaw
z$KapQCm<vuCSk~kF%zcDn6qF>N=8n>ijs<&H4QBrI=1xe82q#P1cXGyBn%lbX2O&i
za~3Q~$;c^KQBqN}rlDm+$CjQQga5rg0U;4F2}4GVnJ{I>oCQl#GI9!5lvLELX=vHd
zv889n;7y-^kcgOsAtT02m@;F|f+Z;#IRz_9Dr(j=v~1|u(z9dmf6yl&BqAnZ$cQl$
zrp%bLU`a|wPQi+jikdYIEgL$v^z0b?5q$zeB4QGTj2JUv%8WS+mZW6l6s#zzs9Dp{
zvY}&3&yK<WQJ;X2h?s;SBgRaaGGoqyB`Fy>1uIG_YSuKgZ0Ok1v%`}!*drh$A|_$T
zh%pnU%$T!aNlHde!HSZKnl%kA8#=c1>=^u?^a%)wh)EbSV$6goGv+K<l9G{Au%e`*
zW=%uOhK?;gI|lz}eF8!vViJao7&BqYj5!OIq-5k2tSG6dS<}$6p<_$Wj=}#$pMa2v
zn1mrC#!Q$pW6pvlDH%BhD@rPA)-<$i=-AS;WALs|KuAPP!jKVTCQO+zXTg$`jGTfM
zB^5Pm8d^4VZ0XrC_`m8C5E2oSFl5A-2~%dwS+FD}Bd1_RNkz??hL#N-TY7d3{%`sO
zgha$73>h(I!ju_v7A#50$SGJ+Qc<&}p=CqImYyAhe@>r(kcgOsAtT02m@;F|f+Z;#
zIRz_9Dr(j=v~1|u(zC-0<6w_~kcgOsAtT02m@;F|f+Z;#IRz_9Dr(j=v~1|u(z9dm
z&+8Kq5)qRyWW<;WQ)bLrup}iTr(i`%Ma`OqmJJ<SdUg!{1$_cSB4QGTj2JUv%8WS+
zmZW6l6s#zzs9Dp{vY}&3&yK;rs82vhL`=ev5o0DynK5U<l9Y^`f)yndHESALHgs(1
z*)jOgCm<vuCSk~kF%zcDn6qF>N=8n>ijs<&H4QBrI=1xe82n561cXGyBn%lbX2O&i
za~3Q~$;c^KQBqN}rlDm+$CjQQgMV3{fRKoogdrowOqeob&VnT=894<jN-ApBG_-8!
z*wV9O@UQ3-5E2oSFl5A-2~%dwS+FD}Bd1_RNkz??hL#N-TY7eQQ#9BkAS5CtVaSLv
z6Q<0VvtUU|Moz(sl8Tx&4J{iww)E^6{Hyu|gha$73>h(I!ju_v7A#50$SGJ+Qc<&}
zp=CqImYyAhe@&l&kcgOsAtT02m@;F|f+Z;#IRz_9Dr(j=v~1|u(z9dmf7d4<BqAnZ
z$cQl$rp%bLU`a|wPQi+jikdYIEgL$v^z0aX>JtzW5tA@v#Fz<FX3SZzBqbxKU`0tq
z&6<Xm4INv0b`1U>`UHeT#3T$EF=oP)8FLmaNy*45SW!|@v!<bCL&uh$9fN;epMa2v
zn1mrC#!Q$pW6pvlDH%BhD@rPA)-<$i=-AS;WAJb26A%&+lQ3k&m<dy6%vrD`B_pR`
zMM*`?nueAQ9b0;K_&_w+BOoLqCSk~kF%zcDn6qF>N=8n>ijs<&H4QBrI=1xe82p?1
z1cXGyBn%lbX2O&ia~3Q~$;c^KQBqN}rlDm+$CjQQgMUk(fRKoogdrowOqeob&VnT=
z894<jN-ApBG_-8!*wV9O@c+~&AS5CtVaSLv6Q<0VvtUU|Moz(sl8Tx&4J{iww)E^6
zeCZPq5)qRyWW<;WQ)bLrup}iTr(i`%Ma`OqmJJ<SdUg!{U-|@uM8qTv88K$Ulo@jt
zEJ?}8DOgcbQM0C@Wkbi7o*je#w>|+O5itovMvR#-WyYKZOHwj&3RaX<)U0V}+0e12
zXUE{*)+Zn&A|_$Th%pnU%$T!aNlHde!HSZKnl%kA8#=c1>~Jw;utz{hL`=ev5o0Dy
znK5U<l9Y^`f)yndHESALHgs(1*)jNc^a%)wh)EbSV$6goGv+K<l9G{Au%e`*W=%uO
zhK?;gI|l!*J^>*SF$qIPjF~WH#+(I9QZjN1R+LoKtZ8W3(6Oaw$Kd~?Pe4dSOu~>6
zV<t?QF=xS&l#HB$6(tolYZ_WMbZqI_G5FReAS5CtVaSLv6Q<0VvtUU|Moz(sl8Tx&
z4J{iww)E^6{D1Wc2#JVE7&2nagef!TELf6~kyEguq@rd`L(7JaEj>F1|DHYpArUbN
zLq?35FlEM^1xr#gatc<IRMf0#XxY%QrDw<B-`6J~BqAnZ$cQl$rp%bLU`a|wPQi+j
zikdYIEgL$v^z3jUbg)N2NJLD+kP%}hOqnrf!IG4WoPrf46*X%bS~hfS>De*(5A+EL
ziHJ!UGGfexDKq9QSdx;FQ?R0>qGnA)%Z83EJv#>fp*{g25itovMvR#-WyYKZOHwj&
z3RaX<)U0V}+0e12XUE__(kCD!A|_$Th%pnU%$T!aNlHde!HSZKnl%kA8#=c1>=^v$
z6A%&+lQ3k&m<dy6%vrD`B_pR`MM*`?nueAQ9b0;K4E|$%0zx8U5{8TzGhxb%ISZDg
zWaJdAD5<Df)6lY^V@uDD!GEGpKuAPP!jKVTCQO+zXTg$`jGTfMB^5Pm8d^4VZ0XrC
z_)ql-2#JVE7&2nagef!TELf6~kyEguq@rd`L(7JaEj>GY2{PCtAS5CtVaSLv6Q<0V
zvtUU|Moz(sl8Tx&4J{iww)E^6{Ac<Egha$73>h(I!ju_v7A#50$SGJ+Qc<&}p=CqI
zmYyAh|6HGdkcgOsAtT02m@;F|f+Z;#IRz_9Dr(j=v~1|u(z9dm|I;TRBqAnZ$cQl$
zrp%bLU`a|wPQi+jikdYIEgL$v^z0b?>JtzW5tA@v#Fz<FX3SZzBqbxKU`0tq&6<Xm
z4INv0b`1Xi`UHeT#3T$EF=oP)8FLmaNy*45SW!|@v!<bCL&uh$9fSWupMa2vn1mrC
z#!Q$pW6pvlDH%BhD@rPA)-<$i=-AS;WAI<<6A%&+lQ3k&m<dy6%vrD`B_pR`MM*`?
znueAQ9b0;K_yTnB2lm(};DC@rB94eTCgFr3r;Ipb%sCS-m~zRCE9P9Y;D#l)q}-8l
zPtF4ckF0p2<e7>WYF=6MM#DQTA8h!f<BKid^!%{nm%)F-e)iZW;DC@rB94eTCgFr3
zr;Ipb%sCS-m~zRCE9P9Y;D#l)q}-8lPtF4ckF0p2<e7>WYF=6MM#DQTA8h!f<BKid
z^!%{nm%$%pKYQ#Ga6rf*5l6%vlW@Y2Q%0OI=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5J
zM^-#h@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP%ixc;pFQ>oI3VPZh$CW-NjPE1DI?Ap
zbIybdrd%@PiaFOTxM9gHDR*Stlk-5qBP*ULd8Xopnpf7m(eO^o2OB==_+ra9JwNRD
zW$*|4*<+u8140goI3nhlgcF9GGUALe=S;X@$|W<dm~+j78<yOXa!1BJIS&*(vf_!7
zXDVK(d1cKT4ezvku;G)AFSdNs^TUo`27iqG?6FV40U?J(91(L&!U;o88F9v#b0%Cc
z<&qg!%(-U44NGoGxg+DAoCgXXS@A^4GZin?yt3wvhId*%*zifm7hAsR`C-Q|gFn`O
z_Sh%jfRIBXj)*xX;e;Wlj5uS=ITJ3Ja><M<=3KMjh9$S8+>vol&I1LHtazg2nTi)`
zURm=-!#gb>Z1|+(i!I;u{IKJf!5?Qod+ZZ%K*%8xN5mYHaKeyNMw~I`oCz09xn#x_
zbFNu%!;)K4?#Q?&=YfJpRy<MiOvMW|udI2a;hmNbHhj|Y#g=b+e%SGgA3qKLz#jVq
z91wCy#1S#aB%Cnhlo4l)IcLHJQ!bft#hhyv+_2=9lshu+$$6mQkrhvrJX7&P%`0o(
zXn3dPgAJc_e6i)5o*#DnGWg@|XODdX4hT6U;)s}I5>6O$%7`<@oHOBqDVNN+V$L-S
zZdh_l${iW^<UCOD$ciUQo~d}D=9M*XG`!RD!G=#dzS#0j&ks9(8T<+Mv&TLG2ZS6F
zaYW292`3CWWyBd{&Y5t*luKq@G3S~EH!Qg&<&KPdavmsnWW^ID&s4ln^U9hx8s2I7
zV8bUJUu^lN=Z77?4E{v>*<+u8140goI3nhlgcF9GGUALe=S;X@$|W<dm~+j78<yOX
za!1BJIS&*(vf_!7XDVK(d1cKT4ezvku;G)AFSdNs^TUo`2K)B2$36iEgd7raM9eV>
zCk#1d#2I7GnQ+0BOJ-a#=b8mKEV(7+j*NS99w>NZ#S<mZRJ>60%9=MC-f8(@!zUeI
zZ26|=haJBR{v`X^W1oNnLJo;IBIcNc6Na2J;*2rpOt@gmB{Qy=bIpPqmfVtZN5(xl
z4-`DI;)#-HDqg60Wz8E6@3efd;ggOpwtUm`!;W7Df3p4Tu}{DOA%{d95pzt!2}4d9
zamJW)CR{M(k{MUbxn{u)OKwTIBjcW&2MQip@kGfp6))7hvgVD3cUnH!@JYuPTfXV}
zVaG3n|EB%yu}{DOA%{d95pzt!2}4d9amJW)CR{M(k{MUbxn{u)OKwTIBjcW&2MQip
z@kGfp6))7hvgVD3cUnH!@JYuPTfXV}VaG3i_&oRnd+ZZ%K*%8xN5mYHaKeyNMw~I`
zoCz09xn#x_bFNu%!;)K4?#Q?&=YfJpRy<MiOvMW|udI2a;hmNbHhj|Y#g=b+e%SHL
z;J;-*d+ZZ%K*%8xN5mYHaKeyNMw~I`oCz09xn#x_bFNu%!;)K4?#Q?&=YfJpRy<Mi
zOvMW|udI2a;hmNbHhj|Y#g=b+e%SHL;7_riJ@yGWAmosUBVvw8IAO>sBhDCe&V&o5
zTr%T|IoB+>VaY8icVyg?^FYBPE1oEMrs9Q~*FWWt{MTm?zj@s7PRj=yKI!;k%Qrng
z?D%Ey-?pDU_6ay3<dBFXVvb2TVaO>X&KPsfgbSuzGUJLl*DSbU$t@{&WZaYUK*1v`
zo+x>y;)R-5*1Xa1PRj=yKI!;k%Qrng?D%DHU_X286L3JtArVK!9FuUukW)sSG3J~J
z7fiWi#uam}S#ZOWTT<@GxF_d<f=5<7QSwa13pKB-d86T-mJc?3((%QXZ+d>%@yp;(
zwVyrq2{<6+kccB<j!8IS$SEVv7<0~q3#ME$<BB=gEVyCGEh%?o+>`S_!6Pf4D0!yh
zg_>8^ywUJZ%Lf}i>G)#HH$6Y>_+{|lv7bHm2{<6+kccB<j!8IS$SEVv7<0~q3#ME$
z<BB=gEVyCGEh%?o+>`S_!6Pf4D0!yhg_>8^ywUJZ%Lf}i>G)#HH$6Y>_+{|lwVyrq
z2{<6+kccB<j!8IS$SEVv7<0~q3#ME$<BB=gEVyCGEh%?o+>`S_!6Pf4D0!yhg_>8^
zywUJZ%Lf}i>G)#HH$6Y>_{B|6gFmpxJ^=@W91?Lv%rOZk3^`@Q8Dq|waKV&IW?V7n
zngus3xh3U}jC*n(D0pPW6D7}7yioJXnl~EWY58EoCmmmG`KITG9ls3zH2c|OpMV2G
z4v9D-=9q*NhMY3uj4|g-xM0d9Gp?9(&4L@2+>&xf#yvR?6g;xxiIQh3UZ{Cx%^MBx
zw0yAPla4R8eADy8j$a0Uy8Z02Prv~oheR9^b4<bsLrxiS#+Y*^TrlO58CT4?X2A_h
zZb`W#<DQ%c3LaVUM9DK1FVwuU=8cATT0YqDNyiskzUld4$1j6F!+!SIC*XjPLn4ld
zIVRzRA*YNuW6U`dE|_x3j4S3`v*3m$x1`*WaZk<z1&^$FqU4#17iwNv^G3rvEgx+7
zq~nV%-}L;j<Cnpq{p_(%zyTqLL>v)wOu`96P8o5=m~$pvFy)dNSIoI)!3|4pNx38A
zo}32?9$E23$uku%)V#9hjfQtxKG^U{#}`|^>G@&DFM~hRe)iZW;DC@rB94eTCgFr3
zr;Ipb%sCS-m~zRCE9P9Y;D#l)q}-8lPtF4ckF0p2<e7>WYF=6MM#DQTA8h!f<BKid
z^!%{nm%*Q9KYQ#Ga6rf*5l6%vlW@Y2Q%0OI=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5J
zM^-#h@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP%izCfKYQ#Ga6rf*5l6%vlW@Y2Q%0OI
z=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5JM^-#h@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP
ziyNK?e_)S&0uBf{B;tseV-ijna>|G^#+)<Zf+?5GxMI#V3vO6)OUfM?_vAcK@W_fM
zN}j2Bq2`q}Z#2Bq^1+5rI=<NQP0tTIei{7P_Or)60SAN}5^+S#F$pIOIc3BdW6qgy
z!IVp8TruaG1vf0YCFPEcdvYEqcx1&BCC^m6Q1i-~HyYk)`C!8*9batursszpzYP8y
z``KfkfCEAfi8vzWn1mCCoHF8!G3QLUV9F&ku9$Ppf*Y3Hl5$7JJvk2)JhI}6l4mMj
zsCi}08x8NYe6Zn@jxV-+)APfQUk3ku``KfkfCEAfi8vzWn1mCCoHF8!G3QLUV9F&k
zu9$Ppf*Y3Hl5$7JJvk2)JhI}6l4mMjsCi}08x8NYe6Zn@jxV-+)APfQUj|3^v&TLG
z2ZS6FaYW292`3CWWyBd{&Y5t*luKq@G3S~EH!Qg&<&KPdavmsnWW^ID&s4ln^U9hx
z8s2I7V8bUJUu^lN=Z77?4E|jE*<+u8140goI3nhlgcF9GGUALe=S;X@$|W<dm~+j7
z8<yOXa!1BJIS&*(vf_!7XDVK(d1cKT4ezvku;G)AFSdNs^TUo`27jLY?6FV40U?J(
z91(L&!U;o88F9v#b0%Cc<&qg!%(-U44NGoGxg+DAoCgXXS@A^4GZin?yt3wvhId*%
z*zifm7hAsR`C-Q|ga3j3?6FV40U?J(91(L&!U;o88F9v#b0%Cc<&qg!%(-U44NGoG
zxg+DAoCgXXS@A^4GZin?yt3wvhId*%*zifm7hAsR`C-Q|4m%nAfj#yKI3VPZh$CW-
zNjPE1DI?ApbIybdrd%@PiaFOTxM9gHDR*Stlk-5qBP*ULd8Xopnpf7m(eO^o2OB==
z_+ra9JwNRDW$@?Q&mQ{(91wCy#1S#aB%Cnhlo4l)IcLHJQ!bft#hhyv+_2=9lshu+
z$$6mQkrhvrJX7&P%`0o(Xn3dPgAJc_e6i)5o*#DnGWZMZXODdX4hT6U;)s}I5>6O$
z%7`<@oHOBqDVNN+V$L-SZdh_l${iW^<UCOD$ciUQo~d}D=9M*XG`!RD!G=#dzS#0j
z&ks9(8T^Izv&TLG2ZS6FaYW292`3CWWyBd{&Y5t*luKq@G3S~EH!Qg&<&KPdavmsn
zWW^ID&s4ln^U9hx8s2I7V8bUJUu^lN=Z77?436z*k9`6T2stF;h?rv%P8f2^h%?5V
zGvR_Mm&~|g&NT~eSaM6s9U1rJJW%k+iYH2*sd%C0l{IfPywmc*hEF=a*z!%!4?BJt
z{6+S&$36iEgd7raM9eV>Ck#1d#2I7GnQ+0BOJ-a#=b8mKEV(7+j*NS99w>NZ#S<mZ
zRJ>60%9=MC-f8(@!zUeIZ26|=haJBR{)hIn$36iEgd7raM9eV>Ck#1d#2I7GnQ+0B
zOJ-a#=b8mKEV(7+j*NS99w>NZ#S<mZRJ>60%9=MC-f8(@!zUeIZ26|=haJBR{$l&t
zW1oNnLJo;IBIcNc6Na2J;*2rpOt@gmB{Qy=bIpPqmfVtZN5(xl4-`DI;)#-HDqg60
zWz8E6@3efd;ggOpwtUm`!;W7ZbT;?{d+ZZ%K*%8xN5mYHaKeyNMw~I`oCz09xn#x_
zbFNu%!;)K4?#Q?&=YfJpRy<MiOvMW|udI2a;hmNbHhj|Y#g=b+e%SHL;4iVCJ@yGW
zAmosUBVvw8IAO>sBhDCe&V&o5Tr%T|IoB+>VaY8icVyg?^FYBPE1oEMrs9Q~SJu4I
z@J`DI8$RjyV#_x@KkWEr@R!=p9{U6w5OPSw5i!RkoG|2+5oe4!XTk+jE}3z~oNE@`
zu;iAMJ2LLcd7$8t6;G5rQ}II0D{J0pc&Fuq4WD#;vE`edA9nmQ_#fHN9{U6w5OPSw
z5i!RkoG|2+5oe4!XTk+jE}3z~oNE@`u;iAMJ2LLcd7$8t6;G5rQ}II0D{J0pc&Fuq
z4WD#;vE`edA9nmQII*8S_6ay3<dBFXVvb2TVaO>X&KPsfgbSuzGUJLl*DSbU$t@{&
zWZaYUK*1v`o+x>y;)R-5*1Xa1PRj=yKI!;k%Qrng?D%Eym)XxA`ve>ia!AAxF~=mF
zFyxdGXN);#!Ua<<nQ_IOYZlzF<d&2>GVaNFpx}`ePn0}U@j}fjYu;#hr{#kUpLBe&
z<(r-#cKkB<%k5{6eF6>$IV9qUm}3%77;?&pGsc`V;esib%(!CCH4AQ7a!bk`8TaHo
zQ1Hl#CrX~Fc%kN%HE%S$)AGTFPddKX@=ebVJAN7b751~oJ^=@W91?Lv%rOZk3^`@Q
z8Dq|waKV&IW?V7nngus3xh3U}jC*n(D0pPW6D7}7yioJXnl~EWY58EoCmmmG`KITG
z9lto_bnplE*eBqCkV7Jlh&d+VgdwMlIAhE?6E2u?$&4%JT(jVYCAXy9k#SGX0|k$)
zc%tN)iWh2LS@TB2J1rk<_@v{DE#LI~u;Z7(Uui#k>=ST6$RQC&#2k}w!jMx&oH6E{
z2^UPcWX2V9u32!yl3P;l$harxfr3X?JW=vY#S1mBta+p1ot6(aeA4m7mT!7~*zwEY
zud<&#_6ay3<dBFXVvb2TVaO>X&KPsfgbSuzGUJLl*DSbU$t@{&WZaYUK*1v`o+x>y
z;)R-5*1Xa1PRj=yKI!;k%Qrng?D%EySKH4X`ve>ia!AAxF~=mFFyxdGXN);#!Ua<<
znQ_IOYZlzF<d&2>GVaNFpx}`ePn0}U@j}fjYu;#hr{#kUpLBe&<(r-#cKk9pwVyrq
z2{<6+kccB<j!8IS$SEVv7<0~q3#ME$<BB=gEVyCGEh%?o+>`S_!6Pf4D0!yhg_>8^
zywUJZ%Lf}i>G)#HH$6Y>_+{|d*v}sO1RM}@NW>8_$0VFE<dhL-j5%k*1ye4WamAc#
z7TmDpmXtd(?#X$e;E@$ilsr@MLd`2{-e`EI<%12MbbPVpo1Pzb{4)4!?PrgD0uBf{
zB;tseV-ijna>|G^#+)<Zf+?5GxMI#V3vO6)OUfM?_vAcK@W_fMN}j2Bq2`q}Z#2Bq
z^1+5rI=<NQP0tTIei{6A_Or)60SAN}5^+S#F$pIOIc3BdW6qgy!IVp8TruaG1vf0Y
zCFPEcdvYEqcx1&BCC^m6Q1i-~HyYk)`C!8*9batursszpzc}E0@CWwTC*XjPLn4ld
zIVRzRA*YNuW6U`dE|_x3j4S3`v*3m$x1`*WaZk<z1&^$FqU4#17iwNv^G3rvEgx+7
zq~nV%-}L;j<Cno-Z$EqN6L3JtArVK!9FuUukW)sSG3J~J7fiWi#uam}S#ZOWTT<@G
zxF_d<f=5<7QSwa13pKB-d86T-mJc?3((%QXZ+d>%@yp<UY(IPK6L3JtArVK!9FuUu
zkW)sSG3J~J7fiWi#uam}S#ZOWTT<@GxF_d<f=5<7QSwa13pKB-d86T-mJc?3((%QX
zZ+d>%@yp<Eu%A8l2{<6+kccB<j!8IS$SEVv7<0~q3#ME$<BB=gEVyCGEh%?o+>`S_
z!6Pf4D0!yhg_>8^ywUJZ%Lf}i>G)#HH$6Y>_+@ZrKYQ#Ga6rf*5l6%vlW@Y2Q%0OI
z=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5JM^-#h@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP
z%iw=vKYQ#Ga6rf*5l6%vlW@Y2Q%0OI=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5JM^-#h
z@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP%iw=%KYQ#Ga6rf*5l6%vlW@Y2Q%0OI=9~!^
zOu1yn6?3jxaKn;YQtrsOC+C5JM^-#h@=V1GHLt9Bqv4&F4>o+#@x_*JdVbjP%iw=z
zKYQ#Ga6rf*5l6%vlW@Y2Q%0OI=9~!^Ou1yn6?3jxaKn;YQtrsOC+C5JM^-#h@=V1G
zHLt9Bqv4&F4>o+#@x_*JdVbjPi$6CR{DD382{<6+kccB<j!8IS$SEVv7<0~q3#ME$
z<BB=gEVyCGEh%?o+>`S_!6Pf4D0!yhg_>8^ywUJZ%Lf}i>G)#HH$6Y>_+{`vx1T-s
z2{<6+kccB<j!8IS$bZGq9lkaJz;S*KL!?NNB0-7-DN-azks?Kk1PKx(xSY#e=H*=G
zGMDpmUgmOM=H*<@%UtF%m-BKibD7Ir=7JO{Qlv<cB1M7}DN>|Jks?Kk6e&`qNRcA<
zym;XE_QUrN_(g~kBTj-ODbi%fk|R%nA|=XHs8XX&gC;H7bm-D!ifLw;WsZ3k=(ETY
z%dD`<8tZJZ$rjt}u*)8S3;G#gkRgH$6Jmr>!i+J_1d~LF5+hE6Bq`El$dV&Zfg&Z!
zRH#y;PJ<>b+H~mBV~S~Jm}QQ67U;9c63eWx${Oozu*nwN?6Auof&W-P0}L`mkYPfM
zFiMy)#+hJ}2vK6hNsuH(nhaTT<S9_3M41XzYSd}aq(z$!U3yG0%?z{5G0y^h7FlAM
z6;@eeoeehGVw)Xy*&~q9&j5oA5oDMUBa9MejBzHIBtn!JaS|j+ktRcy9C->9DN&|E
zl^S&#G-=VMLzf;?Of$nQbIh|qpGB5fW`$MOSZ9Mxw%BHeUG@n41N{sz$Pht>2{FPb
zVa6C|f=MDoi4iA3k`!q&WXX}IK#>w<DpaXar$LhzZ8~)6F~u}9%reJ33-noJiDgz;
zWsP+<*kp@scGzW)z(xHGFvt)=h6ypkC}GAJXM#y0M2Qh6L6Q_{GGxh-r$CVsWhzvu
zQKvzZ7Hv9o=`qDLGt4r_JPY($WQk=~SY?fMHrQl~ZFbmYkHCMTp8*CLBFHcyMi?c`
z7~@PZNrWgd;v`6tB29)YIr0=JQld<SDmCgfXwsrhhb}#)m}Z7q=9p)JK8q}|%nGZl
zvCamYY_ZJ_yX@hySzsRn3^GKJVM2^BN|-UmnP8F#QDVeNkR(N#3|VsIDNv+DnF>{E
z)M?P9MVk&?dQ36R471EJ&jNiGSz?(LR#{`64K~?gn;mx9Bk-T<XMjP52r^8F5k?6!
z#yAs95+O>AI0=%ZNRuH;jywg5lqgf7N{u=VnzU%sp-Yb`rkP=uIp$fQ&mv1Kv%)HC
zth2!;TWqt#E_(zn>1TjJh6plDh!I8!GsZX*OcEhVj5rCBq)3w?OO8ARij*i*p-PQ9
z4VtuQ)1ga`DW;iWmO18GpwA*pEVIHYYpk=uCR=Q?!!COS{xkgyFvt)=h6ypkC}GAJ
zXM#y0M2Qh6L6Q_{GGxh-r$CVsWhzvuQKvzZ7Hv9o=`qDLGt4r_JPY($WQk=~SY?fM
zHrQl~ZFbmYk3dR40}L`mkYPfMFiMy)#+hJ}2vK6hNsuH(nhaTT<S9_3M41XzYSd}a
zq(z$!U3yG0%?z{5G0y^h7FlAM6;@eeoeehGVw)Xy*(30u>t}#Lh6plDh!I8!GsZX*
zOcEhVj5rCBq)3w?OO8ARij*i*p-PQ94VtuQ)1ga`DW;iWmO18GpwA*pEVIHYYpk=u
zCR=Q?!!COSF6(E2L52u2Oo$Oi2{Xnx6HF2zN{l!OlB7tJAxn-t1&Wj?Q=v+YIt`k%
zXw#ufk13{^VU{`OS)k7%ODwa(Dr>B>!6sX5v%@ZX1pc9Z1{h?BAj5<hVU#dqj5EO`
z5u(J1lORcoG#Rqw$Wx$5i82+c)Tq;-NsBfey7ZW0ni*!9W1a>2EV9HhE3C4{IvZ@V
z#Wp+avWKUpfqe`x$Pht>2{FPbVa6C|f=MDoi4iA3k`!q&WXX}IK#>w<DpaXar$Lhz
zZ8~)6F~u}9%reJ33-noJiDgz;WsP+<*kp@scGzW)z(3N@0D}w>WS9^mj1p#yaVD4~
zLX;SB5+q5HCPS7Sc?uLMQKmwb8g&{pY0;)bmmX70Gs7%%%(FnBMV44*g;myAXM;_)
z*k*@a_6S_j&j5oA5oDMUBa9MejBzHIBtn!JaS|j+ktRcy9C->9DN&|El^S&#G-=VM
zLzf;?Of$nQbIh|qpGB5fW`$MOSZ9Mxw%BHeUG@n47y21skRgH$6Jmr>!i+J_1d~LF
z5+hE6Bq`El$dV&Zfg&Z!RH#y;PJ<>b+H~mBV~S~Jm}QQ67U;9c63eWx${Oozu*nwN
z?6AuofsB3z7-Wbb!-N=NlrUqAGr=SgqQr=kAW4ce8M5TaQ=mwRG8L-SsMDZHi#8p)
z^q69r8D^Pdo(1|Wvcxhgtg^;B8*H-0HaqOHN8rEI&j5oA5oDMUBa9MejBzHIBtn!J
zaS|j+ktRcy9C->9DN&|El^S&#G-=VMLzf;?Of$nQbIh|qpGB5fW`$MOSZ9Mxw%BHe
zUG@lE)z1Kf3=w3Q5F?BdW{hzrm?T1!7;zFLNs%T)mK=Es6e&@rLX{eI8Z>FqrbCw=
zQ%p0%EOX4WK%YgHSZ0M))>vnQO}5x(hh6pv{8#!JV2~k#3=?96QNoNd&IFT0h!P`C
zf+Q)@WXO^uPk|yO%2cRQqfUb+E!uSG(qoEgW|(D;c^2ri$P&w}u*w?iY_Q1|+w8E*
z9v+$p_A$U9Lj)Nn#0aB=8DpFYCW#OwMw|pmQl!a{B}bkDMM{*ZP^Ctl22EPD>CmOe
z6w}Nw%N+A8&}We)mRVtyHP+c+lP$K{VV6Au|FwPw7-Wbb!-N=NlrUqAGr=SgqQr=k
zAW4ce8M5TaQ=mwRG8L-SsMDZHi#8p)^q69r8D^Pdo(1|Wvcxhgtg^;B8*H-0HaqOH
zN8p-%1{h?BAj5<hVU#dqj5EO`5u(J1lORcoG#Rqw$Wx$5i82+c)Tq;-NsBfey7ZW0
zni*!9W1a>2EV9HhE3C4{IvZ@V#Wp+avPa;*(a!*b3=w3Q5F?BdW{hzrm?T1!7;zFL
zNs%T)mK=Es6e&@rLX{eI8Z>FqrbCw=Q%p0%EOX4WK%YgHSZ0M))>vnQO}5x(hh6pv
z<n%MZAVUNhCd3G%gc)O;2_}gUB}SYCNm8WAkR?Z+0!2!csZgaxod!)>wCT{L#}w1d
zFv}eCEYN3>C6-xXl{MDcV3RGj*<qJG0{>V)0}L`mkYPfMFiMy)#+hJ}2vK6hNsuH(
znhaTT<S9_3M41XzYSd}aq(z$!U3yG0%?z{5G0y^h7FlAM6;@eeoeehGVw)Xy*&}dW
zKLZRhM37-Zj4(==F~*r-k_b^^#7U4OMVbs*a^xvcq(qqtRch2}(4<A14qbXoG0hCK
z%rVabeHK|_nH5%9W1S5)*<za=cG)BFPxLduAVUNhCd3G%gc)O;2_}gUB}SYCNm8WA
zkR?Z+0!2!csZgaxod!)>wCT{L#}w1dFv}eCEYN3>C6-xXl{MDcV3RGj*<qJG0(t!m
zFvt)=h6ypkC}GAJXM#y0M2Qh6L6Q_{GGxh-r$CVsWhzvuQKvzZ7Hv9o=`qDLGt4r_
zJPY($WQk=~SY?fMHrQl~ZFbmYkHA0G&j5oA5oDMUBa9MejBzHIBtn!JaS|j+ktRcy
z9C->9DN&|El^S&#G-=VMLzf;?Of$nQbIh|qpGB5fW`$MOSZ9Mxw%BHeUG@mv(9Zya
z3=w3Q5F?BdW{hzrm?T1!7;zFLNs%T)mK=Es6e&@rLX{eI8Z>FqrbCw=Q%p0%EOX4W
zK%YgHSZ0M))>vnQO}5x(hh6pv{4@OwFvt)=h6ypkC}GAJXM#y0M2Qh6L6Q_{GGxh-
zr$CVsWhzvuQKvzZ7Hv9o=`qDLGt4r_JPY($WQk=~SY?fMHrQl~ZFbmYk3c~`0}L`m
zkYPfMFiMy)#+hJ}2vK6hNsuH(nhaTT<S9_3M41XzYSd}aq(z$!U3yG0%?z{5G0y^h
z7FlAM6;@eeoeehGVw)Xy*(31Z>SusKh6plDh!I8!GsZX*OcEhVj5rCBq)3w?OO8AR
zij*i*p-PQ94VtuQ)1ga`DW;iWmO18GpwA*pEVIHYYpk=uCR=Q?!!COSZt7=%L52u2
zOo$Oi2{Xnx6HF2zN{l!OlB7tJAxn-t1&Wj?Q=v+YIt`k%Xw#ufk13{^VU{`OS)k7%
zODwa(Dr>B>!6sX5v%@ZX1pYhy3^2$LL52x2!YE<J7-xb>B1DN1Cqa@FX)<KVk*7eB
z5@jk>sZpmvlNN0{bm=k0G&9UH$2<%4S!9W2R#;_?bvD>!i*0t;Wsg8nKLZRhM37-Z
zj4(==F~*r-k_b^^#7U4OMVbs*a^xvcq(qqtRch2}(4<A14qbXoG0hCK%rVabeHK|_
znH5%9W1S5)*<za=cG)BF&-F9FAVUNhCd3G%gc)O;2_}gUB}SYCNm8WAkR?Z+0!2!c
zsZgaxod!)>wCT{L#}w1dFv}eCEYN3>C6-xXl{MDcV3RGj*<qJG0=M)tz#u~e879OC
zql6h_oCzk05G6*O1W8h)$&e*So&rTml&MgqMx6#tTD0lVrN<Q0%rMIw^DNM3ktLQ{
zVU;!3*<h0`w%K8qJp%tiKLZRhM37-Zj4(==F~*r-k_b^^#7U4OMVbs*a^xvcq(qqt
zRch2}(4<A14qbXoG0hCK%rVabeHK|_nH5%9W1S5)*<za=cG)9P($4^c3=w3Q5F?Bd
zW{hzrm?T1!7;zFLNs%T)mK=Es6e&@rLX{eI8Z>FqrbCw=Q%p0%EOX4WK%YgHSZ0M)
z)>vnQO}5x(hh6pv{P+49V2~k#3=?96QNoNd&IFT0h!P`Cf+Q)@WXO^uPk|yO%2cRQ
zqfUb+E!uSG(qoEgW|(D;c^2ri$P&w}u*w?iY_Q1|+w8E*9)a8X8DNkhf(#R4gi*qb
zG0p^&M2He2PJ$#U(qzbzBTs=MCCXH&Qln0TCN0`@=+a|~X=a#Zj(Ha7v&a(5tgy-&
z>uj*e7TfHw%N~J$sh<G`86wCqAx0P_%oyWLFiC_cG2$dhk|IrpEIINNC{m(Kg(@}b
zG-%SIO@}T$rkG}iS>~8$fj)~YvCImqtg+4pn{2Vo4!i6TDC=i{L52u2Oo$Oi2{Xnx
z6HF2zN{l!OlB7tJAxn-t1&Wj?Q=v+YIt`k%Xw#ufk13{^VU{`OS)k7%ODwa(Dr>B>
z!6sX5v%@ZX1pbwN1{h?BAj5<hVU#dqj5EO`5u(J1lORcoG#Rqw$Wx$5i82+c)Tq;-
zNsBfey7ZW0ni*!9W1a>2EV9HhE3C4{IvZ@V#Wp+avPa;Keg+t1h#<p+7-5t!V~jJw
zBoU&-h?5{miZmIr<j7N?NQp8Ps??~{ph=519lG?GVwxFdnPZ*>`Yf`<GApdI#yT5p
zvc)z#?6ODTf6&hWgA5U5m=Gh35@w8XCYU5blo)XmBuSAbLzWzQ3KS_(rb3k(bs98j
z(WXO}9#c#+!z^>mvp}CkmRM$mRn}N%gH5*BW`|w&2vqblz#u~e879OCql6h_oCzk0
z5G6*O1W8h)$&e*So&rTml&MgqMx6#tTD0lVrN<Q0%rMIw^DNM3ktLQ{VU;!3*<h0`
zw%K8qJp%t)KLZRhM37-Zj4(==F~*r-k_b^^#7U4OMVbs*a^xvcq(qqtRch2}(4<A1
z4qbXoG0hCK%rVabeHK|_nH5%9W1S5)*<za=cG)9vS3d&`GDMJJLX0p<m@&qgV3G(?
zV#G<1Bt@DGS#snlP^3he3RP;<Y0#uan+{!iOfk(2v&=Ei0(}-)Vwn|IS!10IHrZmE
z9d_9x@Ne`pz#u~e879OCql6h_oCzk05G6*O1W8h)$&e*So&rTml&MgqMx6#tTD0lV
zrN<Q0%rMIw^DNM3ktLQ{VU;!3*<h0`w%K8qJp$GDf&Yko>}P-j404bm4iV%q!yF;R
zQARk%D8~tNf-z1q&M781%_L`taF!_Nh;g1c7f5iCB$r5WnKW0(aFs0A$Z?%KHz;tE
zBDW}Un=*H(aF;6gsBxb<4`}d^CXZ<Gm^M%7@RTmk=<%E>UNFr|W_ZOcubJZw^Sot&
zcl3GBA|F`dBg=eZh0m<=g*Cpi&Nnvr&L%(D;wRhuVu#=C@`pYC68InO&p!4uzySt1
z$PkAJa+qO`5aK8!9AlK@ggL<&CmH7y6P#v}GekH`lyk&5Pn-)RxJZ&qq_|9)D`dDz
zmTTm=PM#YSxJi*)l(<cqJ5;z!m3!2<Pn`!ecu12+w0KOLCv<p9muK{N&J-`0<|Q+{
zVwTs;@rHTcvcNm~yl0UQEb)<LKC!}QR{6pjUs>lH8+>PzA8hfHZGN%CZ+7{^9)Ahk
zvp@UT&j1G)<RC*FBFJHeIYNk|jBt!mjuYktW1M81Q%rE0NzM@AEK$x8<2-RLkl-Ro
zE|KCgX|9mrDp{_P<2re6P~av-Zc*YkW$sYnE>-SP<34pB(BL6W9?{}4ZJyBKDP5k?
z<2h5jV49cA@QPVpGshd|dCLOt=<}XMKCr|`mifdApIPM#YkXy$Z*1_LO@6S&Pqz8R
z4!_yu4}1J2@ITp~ee7p|0}OJIAr2AbFvA=n#8E~##wf=LbAmBWGR`R`IL#zyh;WuD
z=ZJBhI2TB8ktCN$ahWt%$Z(Y`*T`|5JU1wClOne$aho!CsBo7m_o#87IuB^@kS334
z@t8JG=<t*-&*<@-DPAzmOJ;b*EU%g44fDKZfp_$I&mtdK;v>s^VujDF@`W|Nvd%X)
z_|7Ij*y1PK{9=dS?DB^_{t~F!pMC6SfCCJ2kRc8c<S@e=A;eKeIL0W)33GxmPBP9Z
zCOFL`XNYi?DCdZAo;VjsaFHaJNO74oSIBUcEZ4|!ojf-vaFZgpC~=!Icc^feD)*>y
zpE?g{@Q@~tXz`dfPw4QJF3;%koGD%~%}Zu@#VoIx;|=q?Wr26}dCwvrSmGngd}4*q
ztn!65zOv3YHu%maKiJ|Y+x%jO-|X^-J^m8-pY6{+_A|f%206$OhX`_*VU7^uC?gzW
zl;eas!5AkQ=M)p1W|A{RI7^gs#5hl!3naKml1rqxOqwfXxJs65<hV|r8x*)nkz16w
zO_@7XxJ#9L)VNQb2Q+v{lSj08Oq(ZkcuJRN^mxt`FPP>fGrVG!*Ua&TdET<XJNmq5
zkq<2Kk!3!y!e>_b!Wv&$=NlV*XOkan@sn+SvBPh6`NJN63Ea0o``FI_2N>iaLmVQ=
zVTL(Eh@*^fj8Tpg<^*G$WSmn>aGFWZ5aBFQ&Jp80aW0VHB1tZh;xcKjkl`v>u94$9
zd2UePCPi*h;x=XOP~k3B?os1Dbso^*Ax$39;xTQW(BUavp3&nuQ@miBm(1{rSza^8
z8|Hb-0`KVao<%;e#7CC-#0sBT<qK<kWu0$q@SRP5u*FZd`Na;u+2s#={3Y<e*q?pu
zXMh6?a*!bo5#%t#93jL}MmWYO#|d+SF-|hhDJD40Bxi_lmMG_lah^CANN|xPmq>A$
zG*`%Ql`PlDah*IjC~%V^w<vL&GIywOmn!$Dai2O5Xz-9Gk7)6jHc#mAlrGQc@ti4M
zFwIM5c*QKQnd1%fyk&uR^m)%BA6Vie%Y0&m&#dx=HNLXWH#YdrCO_EXC)@mDhu`e-
zhdurhsN0`?>}P-j404bm4iV%q!yF;RQARk%D8~tNf-z1q&M781%_L`taF!_Nh;g1c
z7f5iCB$r5WnKW0(aFs0A$Z?%KHz;tEBDW}Un=*H(aF;6gsBxb<4`}d^CXZ<Gm^M%7
z@RTmk=<%E>UNFr|W_ZOcubJZw^Sot&cl3GBA|F`dBg=eZh0m<=g*Cpi&Nnvr&L%(D
z;wRhuVu#=C@`pYC68K;3&p!4uzySt1$PkAJa+qO`5aK8!9AlK@ggL<&CmH7y6P#v}
zGekH`lyk&5Pn-)RxJZ&qq_|9)D`dDzmTTm=PM#YSxJi*)l(<cqJ5;z!m3!2<Pn`!e
zcu12+w0KOLCv<p9muK{N&J-`0<|Q+{VwTs;@rHTcvcNm~yl0UQEb)<LKC!}QR{6pj
zUs>lH8+>PzA8hfHZGN%CZ+7{^9)Afuus{3Q&j1G)<RC*FBFJHeIYNk|jBt!mjuYkt
zW1M81Q%rE0NzM@AEK$x8<2-RLkl-RoE|KCgX|9mrDp{_P<2re6P~av-Zc*YkW$sYn
zE>-SP<34pB(BL6W9?{}4ZJyBKDP5k?<2h5jV49cA@QPVpGshd|dCLOt=<}XMKCr|`
zmifdApIPM#YkXy$Z*1_LO@6S&Pqz8R4!_yu4}1J2@W0ugee7p|0}OJIAr2AbFvA=n
z#8E~##wf=LbAmBWGR`R`IL#zyh;WuD=ZJBhI2TB8ktCN$ahWt%$Z(Y`*T`|5JU1wC
zlOne$aho!CsBo7m_o#87IuB^@kS334@t8JG=<t*-&*<@-DPAzmOJ;b*EU%g44fDKZ
zfp_$I&mtdK;v>s^VujDF@`W|Nvd%X)_|7Ij*y1PK{9=dS?DB^_{t{@|pMC6SfCCJ2
zkRc8c<S@e=A;eKeIL0W)33GxmPBP9ZCOFL`XNYi?DCdZAo;VjsaFHaJNO74oSIBUc
zEZ4|!ojf-vaFZgpC~=!Icc^feD)*>ypE?g{@Q@~tXz`dfPw4QJF3;%koGD%~%}Zu@
z#VoIx;|=q?Wr26}dCwvrSmGngd}4*qtn!65zOv3YHu%maKiJ|Y+x%jO-|X^-J^m8-
zxAtcr`x)Q>gB)atLj*a@Fh>Y+lo5_G%5lP+V2qQDbBYO0GszhuoF&RRVw@+=1rl5&
z$t6-;Ce0NxTqVmja$G0R4GP?($Sq3Trpz5G+@;DrYTT#J0~$P}$s<}krp*&NJf+Js
zdOT-}7fkb#8D25VYvy>vJa1Xx9ev)j$Oo4A$TFW;;WMjzVU4e>^NkI@v&j#(_{lcE
z*x@(3{9%v31RmO-ee7p|0}OJIAr2AbFvA=n#8E~##wf=LbAmBWGR`R`IL#zyh;WuD
z=ZJBhI2TB8ktCN$ahWt%$Z(Y`*T`|5JU1wClOne$aho!CsBo7m_o#87IuB^@kS334
z@t8JG=<t*-&*<@-DPAzmOJ;b*EU%g44fDKZfp_$I&mtdK;v>s^VujDF@`W|Nvd%X)
z_|7Ij*y1PK{9=dS?DB^_{u20i_Gcga8Q=he9At<?1UbwwM+kA05sop+al)KnjFXIW
ziV03L$r&P?CCWKsoF~o&5?mz7B~n}_%@s0SCCfE(Tqn;B3f!c~ElS*`%pEG+rOG{O
z+^5b18a$-QBU(JB%@aC2rOPvVJZFj*O!JZ%UNOsS=6J(AZ&~0SecrRk2bTEAGM`xC
zGpl@IjjycpjSarD$q%;p$u_^(;WxYdVUNEAn)YWO`x)Q>gB)atLj*a@Fh>Y+lo5_G
z%5lP+V2qQDbBYO0GszhuoF&RRVw@+=1rl5&$t6-;Ce0NxTqVmja$G0R4GP?($Sq3T
zrpz5G+@;DrYTT#J0~$P}$s<}krp*&NJf+JsdOT-}7fkb#8D25VYvy>vJa1Xx9ev)j
z$Oo4A$TFW;;WMjzVU4e>^NkI@v&j#(_{lcE*x@(3{9%v31pdAK*~fkcIKUtW8R8H@
z4l~RVLL6m;V~lc~Feez}B;%Z7g40ZLh6rbga*i11iF1Jj7fEu76qiYJg$!58a*Z6<
z$#a7OHz{(961ORHhYEM8{`>dge|Ga@pys|$od+~{NRvmjcubooba+aaXY_c^6fc<O
zB{RHYme<VjhI!tyz&rZ9XORyq@sVXdvBGCo`NA4sS?3!Yd}os%Z1IzAezC)EcKO2|
ze+fLYKl|9v00$W4AVVA?$YF*#LWrY`aEwuo6XpbCoMfC+OmLb>&Jf`&QO*(LJaI0N
z;37#bk>WCGu8`p>S+0@eI(cqT;3h?GQQ|gb?oi<_Rqj#aK6M_@;2}*O(c&>}p3vbb
zU7pe7Ia9n~nwQM*idkMW#~bE(%L4D{^PWXMu*64}`NRsJS>+3Bd}W<)Z19~;ez3()
zw)w>lzuDyvd;BHvzuTXE>}P-j404bm4iV%q!yF;RQARk%D8~tNf-z1q&M781%_L`t
zaF!_Nh;g1c7f5iCB$r5WnKW0(aFs0A$Z?%KHz;tEBDW}Un=*H(aF;6gsBxb<4`}d^
zCXZ<Gm^M%7@RTmk=<%E>UNFr|W_ZOcubJZw^Sot&cl3GBA|F`dBg=eZh0m<=g*Cpi
z&Nnvr&L%(D;wRhuVu#=C@`pYC5@^|<ee7p|0}OJIAr2AbFvA=n#8E~##wf=LbAmBW
zGR`R`IL#zyh;WuD=ZJBhI2TB8ktCN$ahWt%$Z(Y`*T`|5JU1wClOne$aho!CsBo7m
z_o#87IuB^@kS334@t8JG=<t*-&*<@-DPAzmOJ;b*EU%g44fDKZfp_$I&mtdK;v>s^
zVujDF@`W|Nvd%X)_|7Ij*y1PK{9=dS?DB^_{u1~<?9V>-Gr$1`Imi%)2y&QVju7G~
zBOGIt<AgcE7$+I$6ce0gk~2g&OO$iOI8U4lB)CYDOQg6=nk!_uN|tNnxK5rM6u3!|
zTa>s>nLAXtOO<=nxKEu2G<ZmpN3?iMn<sR5N|$H!c+M0rnC2xjykeHu%<+bK-m<_u
z`n+e64=nMKWj?XOXIA;b8eduG8ykFQlOJsHlWl&n!*6!^!ybPLJhngk*v|k57~~*B
z93seJhB-orql|EjQH~Sl1Y?|JoKsA2nn}(O;Ve<k5#u~@E|B0NNiLD%GHI@m;VN0K
zk>fgfZcyMRMQ%~zHf8Qm;VxC~QR6;!9?;++O&-zWF>RjE;VE67(c?K&ykMG_%<zg?
zUNgrV=6TBk@96WMMLw{^N0#}-3ZGf!3u}C3oo{ULolSnQ#ZR{R#SXvO<qv!OCGdaR
zpMC6SfCCJ2kRc8c<S@e=A;eKeIL0W)33GxmPBP9ZCOFL`XNYi?DCdZAo;VjsaFHaJ
zNO74oSIBUcEZ4|!ojf-vaFZgpC~=!Icc^feD)*>ypE?g{@Q@~tXz`dfPw4QJF3;%k
zoGD%~%}Zu@#VoIx;|=q?Wr26}dCwvrSmGngd}4*qtn!65zOv3YHu%maKiJ|Y+x%jO
z-|X^-J^m7C+n;^xXMh6?a*!bo5#%t#93jL}MmWYO#|d+SF-|hhDJD40Bxi_lmMG_l
zah^CANN|xPmq>A$G*`%Ql`PlDah*IjC~%V^w<vL&GIywOmn!$Dai2O5Xz-9Gk7)6j
zHc#mAlrGQc@ti4MFwIM5c*QKQnd1%fyk&uR^m)%BA6Vie%Y0&m&#dx=HNLXWH#Ydr
zCO_EXC)@mDhu`e-hdurh_`mGWKK3)f0R}n95QhkIm|>0(;wU2=W0d2BIl&kw8Rrxe
zoMw_UL^w;7bHq4LoC_qlNRmsWxJ;TWWVlL}Yvj02o*NXnNs(KWxJ{WmRJco(d(^m3
zod+~{NRvmjcubooba+aaXY_c^6fc<OB{RHYme<VjhI!tyz&rZ9XORyq@sVXdvBGCo
z`NA4sS?3!Yd}os%Z1IzAezC)EcKO2|e+fLXKl|9v00$W4AVVA?$YF*#LWrY`aEwuo
z6XpbCoMfC+OmLb>&Jf`&QO*(LJaI0N;37#bk>WCGu8`p>S+0@eI(cqT;3h?GQQ|gb
z?oi<_Rqj#aK6M_@;2}*O(c&>}p3vbbU7pe7Ia9n~nwQM*idkMW#~bE(%L4D{^PWXM
zu*64}`NRsJS>+3Bd}W<)Z19~;ez3()w)w>lzuDyvd;BHvf7_pZ>}P-j404bm4iV%q
z!yF;RQARk%D8~tNf-z1q&M781%_L`taF!_Nh;g1c7f5iCB$r5WnKW0(aFs0A$Z?%K
zHz;tEBDW}Un=*H(aF;6gsBxb<4`}d^CXZ<Gm^M%7@RTmk=<%E>UNFr|W_ZOcubJZw
z^Sot&cl3GBA|F`dBg=eZh0m<=g*Cpi&Nnvr&L%(D;wRhuVu#=C@`pYC66n~Uee7p|
z0}OJIAr2AbFvA=n#8E~##wf=LbAmBWGR`R`IL#zyh;WuD=ZJBhI2TB8ktCN$ahWt%
z$Z(Y`*T`|5JU1wClOne$aho!CsBo7m_o#87IuB^@kS334@t8JG=<t*-&*<@-DPAzm
zOJ;b*EU%g44fDKZfp_$I&mtdK;v>s^VujDF@`W|Nvd%X)_|7Ij*y1PK{9=dS?DB^_
z{u20q?9V>-Gr$1`Imi%)2y&QVju7G~BOGIt<AgcE7$+I$6ce0gk~2g&OO$iOI8U4l
zB)CYDOQg6=nk!_uN|tNnxK5rM6u3!|Ta>s>nLAXtOO<=nxKEu2G<ZmpN3?iMn<sR5
zN|$H!c+M0rnC2xjykeHu%<+bK-m<_u`n+e64=nMKWj?XOXIA;b8eduG8ykFQlOJsH
zlWl&n!*6!^!ybPLJheaj*v|k57~~*B93seJhB-orql|EjQH~Sl1Y?|JoKsA2nn}(O
z;Ve<k5#u~@E|B0NNiLD%GHI@m;VN0Kk>fgfZcyMRMQ%~zHf8Qm;VxC~QR6;!9?;++
zO&-zWF>RjE;VE67(c?K&ykMG_%<zg?UNgrV=6TBk@96WMMLw{^N0#}-3ZGf!3u}C3
zoo{ULolSnQ#ZR{R#SXvO<qv!OCGh{+pMC6SfCCJ2kRc8c<S@e=A;eKeIL0W)33Gxm
zPBP9ZCOFL`XNYi?DCdZAo;VjsaFHaJNO74oSIBUcEZ4|!ojf-vaFZgpC~=!Icc^fe
zD)*>ypE?g{@Q@~tXz`dfPw4QJF3;%koGD%~%}Zu@#VoIx;|=q?Wr26}dCwvrSmGng
zd}4*qtn!65zOv3YHu%maKiJ|Y+x%jO-|X^-J^m8t+Mj*wXMh6?a*!bo5#%t#93jL}
zMmWYO#|d+SF-|hhDJD40Bxi_lmMG_lah^CANN|xPmq>A$G*`%Ql`PlDah*IjC~%V^
zw<vL&GIywOmn!$Dai2O5Xz-9Gk7)6jHc#mAlrGQc@ti4MFwIM5c*QKQnd1%fyk&uR
z^m)%BA6Vie%Y0&m&#dx=HNLXWH#YdrCO_EXC)@mDhu`e-hdurh`2XzBKK3)f0R}n9
z5QhkIm|>0(;wU2=W0d2BIl&kw8RrxeoMw_UL^w;7bHq4LoC_qlNRmsWxJ;TWWVlL}
zYvj02o*NXnNs(KWxJ{WmRJco(d(^m3od+~{NRvmjcubooba+aaXY_c^6fc<OB{RHY
zme<VjhI!tyz&rZ9XORyq@sVXdvBGCo`NA4sS?3!Yd}os%Z1IzAezC)EcKO2|e+fLZ
zKl|9v00$W4AVVA?$YF*#LWrY`aEwuo6XpbCoMfC+OmLb>&Jf`&QO*(LJaI0N;37#b
zk>WCGu8`p>S+0@eI(cqT;3h?GQQ|gb?oi<_Rqj#aK6M_@;2}*O(c&>}p3vbbU7pe7
zIa9n~nwQM*idkMW#~bE(%L4D{^PWXMu*64}`NRsJS>+3Bd}W<)Z19~;ez3()w)w>l
zzuDyvd;BHv|J$E^>}P-j404bm4iV%q!yF;RQARk%D8~tNf-z1q&M781%_L`taF!_N
zh;g1c7f5iCB$r5WnKW0(aFs0A$Z?%KHz;tEBDW}Un=*H(aF;6gsBxb<4`}d^CXZ<G
zm^M%7@RTmk=<%E>UNFr|W_ZOcubJZw^Sot&cl3GBA|F`dBg=eZh0m<=g*Cpi&Nnvr
z&L%(D;wRhuVu#=C@`pYC66o2Vee7p|0}OJIAr2AbFvA=n#8E~##wf=LbAmBWGR`R`
zIL#zyh;WuD=ZJBhI2TB8ktCN$ahWt%$Z(Y`*T`|5JU1wClOne$aho!CsBo7m_o#87
zIuB^@kS334@%TS5Y{#zv|8aod^-Yo_Ns=T<k|arzBuSDaNs=T<k|arzBuSDaNs=VV
z%sJ<rGcz+YGcz-D&Y3yqoRcI;k|gOKKRo_|=chiO&-?v)G0hCK%rVabi!8Cs3ahNK
z&IX%ovCR&<?6J=Qha7Rt38$QK&IOlTam@|4+;PtXk38|r3$MKK&Ig}-@y!pv{1Ncq
z=}#a*1QS9iVT2PwBvC{YLo9K`lRzR#B$GlaX{3`uCRt>YLoRvbQ$Qg_6jMSeWt3Av
zB~?^YLoIdG(?BClG}A&WZM4%tCtY;YLoa>wGr%B23^T$gV~jJwBvVW?!z^>mv%n%t
zEVIHYYpk=uCR=Q?!!CR5bHE`-9CN}cXPk4vC0AT?!!38*^S~odJoCaUZ@lxtCtrN?
z!!LgXEa^`mK?D;*C}D&XK_pQ`6GJR<#FIcGNhFg(DruyXK_*#blS3|f<WoQ)MHEv)
zDP@#XK_yjGQ$sCv)YCvCO*GR&D{ZvXK_^{w(?c(P^fSO9Lku&*C}WH>!6Z{mGs7%%
z%(K8EODwa(Dr>B>!6sX5v%@ZX>~p{&M;vp)DQBE>!6jE*bHgon-1ERAPdxL&D{s8>
z!6#pQ^TRKH1pN2<6G#xjgb+#?;Y1Kg6w$;GOC0eekVq2Aq>xG)>12>e7TM&GOCI?Y
zP)HHQlu$|;<y25f71h*GOC9wz&`1-_w9rZ$?R3yd7v1#GOCS9VFvt+Yj4;X=<4iEg
z6w}Nw%N+A8u*ee2tgy-&>uj*e7TfHw%O3k2aL5tIoN&q+=Ui~f71!Kw%N_ST@W>O-
zyzt5!?|ksd7vKEw%O3&D`V&YH!GsV>7~w<^Nfgn<5KA2KB#=lF$)u1<8tG(^Nfz1U
zkV_u<6i`SJ#gtG=8Rb+^Nfp)9P)i;4G|)&B&9u-;8|`$^Nf+Jp&`Tfv3^2$L!;CP>
z7~@PZ$rRJfFv}eCEU?HD%dD`<8tZJZ$rjt}u*)9%9B{}H$DDA=8RuMZ$rab!aLXO{
zJn+a9&%E%;8}EGZ$rs=J@XH?o|3ZHP2_l#fLJ1?B2qK9hniyh<Bc22jNg|mPQb{A7
z3^K_gn;de<BcB2aDWaGXN-3k93M#3hni^`Uqn-vDX`-1HT4|%54m#<gn;v@Uqn`l=
z8Df|bMj2zA2_~6hni*!9W1a;TSz?(LR#{`64K~?gn;mx9W1j;KIpUZTPC4V83og0h
znj3Dp<DLf|dE%KDUU}o44?g+gn;(ApBVa{;0tq6R5JCwfoCqR`BAOUti6fo_5=kPN
z6jDheoeVO`BAXm?$s?Zv3MrzP5=tqfoC+$bqM90NsiU3-8fl`L7FubeoenzbqMIIi
z>7$<k1{q?Q5k?tfoCzkGVwxFdnPZ*>7FlAM6;@eeoeehGVw)Xy*<+sr4mskO6HYnf
zoC_|w;+h+7x#OM(9(m%K7hZYeoew_w;+r3S`6J+8>Q5j+1QS9iVT2PwBvC{YLo9K`
zlRzR#B$GlaX{3`uCRt>YLoRvbQ$Qg_6jMSeWt3AvB~?^YLoIdG(?BClG}A&WZM4%t
zCtY;YLoa>wGr%B23^T$gV~jJwBvVW?!z^>mv%n%tEVIHYYpk=uCR=Q?!!CR5bHE`-
z9CN}cXPk4vC0AT?!!38*^S~odJoCaUZ@lxtCtrN?!!LgXtm;o7K?D;*C}D&XK_pQ`
z6GJR<#FIcGNhFg(DruyXK_*#blS3|f<WoQ)MHEv)DP@#XK_yjGQ$sCv)YCvCO*GR&
zD{ZvXK_^{w(?c(P^fSO9Lku&*C}WH>!6Z{mGs7%%%(K8EODwa(Dr>B>!6sX5v%@ZX
z>~p{&M;vp)DQBE>!6jE*bHgon-1ERAPdxL&D{s8>!6#pQ^TRKH1pF)g2_%SMLI@>{
za3Y8#ifCepC60I!NF<45Qb;9@bTY^!i)?bpC69axD5QvDN+_j_aw@2#ifU@8rH*<U
zXrzf|T4<$>b~@;!i*9=8rH_6F7-WcHMi^y`aVD5#ifLw;WsZ3kSY(N1R#;_?bvD>!
zi*0t;WsiLhIOK?9PB`U^b1t~#ifeAT<&JwEc;ty^UU=n=cRu*!i*J7T<&S_h{Rt$9
zU_uBbjBp}|B#LNah$W7A5=bP8WKu{ajdU`|B#Ufv$R&?_3MizAVoE5bjB+Zdq>5^4
zsHKj28fc`6W?E>ajdnWdq>FBP=%tT-1{h?BVMZ8bjBzHIWQu8Km}QQ67FcA7WmZ^a
zjdeEIWQ%Qf*kzA>4mjk9V@^2bjB_ry<ce!<xaE#}9(d%5XI^;ajdwoy<cn{9_~nm)
zf2}`(1QARKp@b1m1d&7$O$@Qb5l;e%B#}%Csicuk2AO1$O%A!_kxv1I6j4kGrIb-l
z1(j4$O%1iwQBMPnG|@~8t+dfj2c2}$O%J{F(a!*b3^B|Iql_`m1d~iL%?z{5G0y^v
zEV0ZAtE{ok2AgcL%?`WlvCjdA9C6GEr<`%l1(#fL%?-EQanA#fJn_s6ue|Zj2cLZL
z%@4o)5wNa5fdmmu2%&@#P6Uxe5lsxS#1T&di6oIs3aO-#P6nA|kxdS{<dIJSg%nXt
z38j=#P6d@zQB4iC)KO0ZjWp3r3$3)#P6wTI(M=D%^wG}%gA6gu2&0TK&IFT8G0hCK
z%rVabi!8Cs3ahNK&IX%ovCR&<?6J=Qha7Rt38$QK&IOlTam@|4+;PtXk38|r3$MKK
z&Ig}-@y!pv{1NbP^e2!Yf(ap%Fv5u-k|?5yA(lAeNg$CVl1U+zG}6f+lPt2yA(uSz
zDWH%diYcL#GRmo-k}9gHp_V%8X`qoNnrWexHrnZ+lP<dHp_e}T8DNkhh8ba$F~*r-
zk}0N{VU{`OSzwVRmRVtyHP+c+lP$K{VV6DjIpB~Zjyd6!GtRl-k}Iyc;g&n@dEk*J
zo_XPwH{SW+lP|vc;g>%GHuNWuAc6@YlrX}HAd)Dei6NFa;z=NpB$7!Xl{C`HAd@Vz
z$sw0K@+qK@B8n-YlrqYxppq)8siBrS>S>^nCYouXl{VVxpp!1T>7kcC`Wax5A%+=Y
zlrhGcV3H}OnPHYW=2>8oC6-xXl{MDcV3RGj*<qJG_Br5?BaS)Ylrzq`;F2q@x#5;O
z?s?#mC!TrXl{en`;FB-D`Qev80{*T31QJ9rA%qe}I1xk=MKm$Q5=T4<B$7ljDWsA{
zIvHe=MK(F)l1DxT6jDSnC6rP|ITch=MKv|lQb#=vG}1&fEws``I~{b=MK?Y4(nmi7
z3^K$pBaAY}I1@}V#WXX_GRHg%EV9HhE3C4{IvZ@V#Wp+avd2CL9CE}lC!BJ|ITu`V
z#WgqFa>qRnJo3adFTC=`J0E=V#Wz3v@<+g?{sa<4Fd>8zMmP~f5=AsI#1cn52_%w4
zGAX2zMmiZ}l0`N-<dR1|1r$<5F(s5zMmZH!Qbjd2)KW)14K&h3GcB~zMmrsJ(nU8t
z^wLK^0}L|6Fe8jI#yAs9GQ~7A%reJ33oNq4GApdI#yT5pvc)z#?6Su`2OM(5F(;gI
z#yJ;Ua>X?_+;Yb~4?Ob3GcUaI#ycN;^2Ikl{PIV@|DZpC1QARKp@b1m1d&7$O$@Qb
z5l;e%B#}%Csicuk2AO1$O%A!_kxv1I6j4kGrIb-l1(j4$O%1iwQBMPnG|@~8t+dfj
z2c2}$O%J{F(a!*b3^B|Iql_`m1d~iL%?z{5G0y^vEV0ZAtE{ok2AgcL%?`WlvCjdA
z9C6GEr<`%l1(#fL%?-EQanA#fJn_s6ue|Zj2cLZL%@4o)5wN8{fdmmu2%&@#P6Uxe
z5lsxS#1T&di6oIs3aO-#P6nA|kxdS{<dIJSg%nXt38j=#P6d@zQB4iC)KO0ZjWp3r
z3$3)#P6wTI(M=D%^wG}%gA6gu2&0TK&IFT8G0hCK%rVabi!8Cs3ahNK&IX%ovCR&<
z?6J=Qha7Rt38$QK&IOlTam@|4+;PtXk38|r3$MKK&Ig}-@y!pv{1Na!>Q5j+1QS9i
zVT2PwBvC{YLo9K`lRzR#B$GlaX{3`uCRt>YLoRvbQ$Qg_6jMSeWt3AvB~?^YLoIdG
z(?BClG}A&WZM4%tCtY;YLoa>wGr%B23^T$gV~jJwBvVW?!z^>mv%n%tEVIHYYpk=u
zCR=Q?!!CR5bHE`-9CN}cXPk4vC0AT?!!38*^S~odJoCaUZ@lxtCtrN?!!LgXZ0k=T
zK?D;*C}D&XK_pQ`6GJR<#FIcGNhFg(DruyXK_*#blS3|f<WoQ)MHEv)DP@#XK_yjG
zQ$sCv)YCvCO*GR&D{ZvXK_^{w(?c(P^fSO9Lku&*C}WH>!6Z{mGs7%%%(K8EODwa(
zDr>B>!6sX5v%@ZX>~p{&M;vp)DQBE>!6jE*bHgon-1ERAPdxL&D{s8>!6#pQ^TRKH
z1pH6>6G#xjgb+#?;Y1Kg6w$;GOC0eekVq2Aq>xG)>12>e7TM&GOCI?YP)HHQlu$|;
z<y25f71h*GOC9wz&`1-_w9rZ$?R3yd7v1#GOCS9VFvt+Yj4;X=<4iEg6w}Nw%N+A8
zu*ee2tgy-&>uj*e7TfHw%O3k2aL5tIoN&q+=Ui~f71!Kw%N_ST@W>O-yzt5!?|ksd
z7vKEw%O3$d`V&YH!GsV>7~w<^Nfgn<5KA2KB#=lF$)u1<8tG(^Nfz1UkV_u<6i`SJ
z#gtG=8Rb+^Nfp)9P)i;4G|)&B&9u-;8|`$^Nf+Jp&`Tfv3^2$L!;CP>7~@PZ$rRJf
zFv}eCEU?HD%dD`<8tZJZ$rjt}u*)9%9B{}H$DDA=8RuMZ$rab!aLXO{Jn+a9&%E%;
z8}EGZ$rs=J@XH?o|FixC5=1Z|gc3$L5kwM2G%>^yM?486l0-5oq>@HD8Dx@0HaX;y
zM?M7<QbaK&lu|}H6;x71H8s>yM?DQR(nK>Yw9-a99dyz~H$C*yM?V7$GQ=<=j55YJ
z6HGG2G&9UH$2<!xvcxhgtg^;B8*H-0HaqOH$36!fa>OwwoN~rF7hH11H8<RH$2|``
z^29SQyz<67AAIt~H$VLHN5HQB1QJ9rA%qe}I1xk=MKm$Q5=T4<B$7ljDWsA{IvHe=
zMK(F)l1DxT6jDSnC6rP|ITch=MKv|lQb#=vG}1&fEws``I~{b=MK?Y4(nmi73^K$p
zBaAY}I1@}V#WXX_GRHg%EV9HhE3C4{IvZ@V#Wp+avd2CL9CE}lC!BJ|ITu`V#WgqF
za>qRnJo3adFTC=`J0E=V#Wz3v@<+h`qCbHI5ljf7gb_{zkwg(q46(!!PXdV~kxUAy
zq>)YrnPibo4!Pu!PXUD#QA`P?lu=Fvl~hqp4Ykx!PXmoK(M$`iw9!rnopjMn554r!
z&j5oAG0X^~j4{pxlT0zq471EJ&jO1qvCImqtg+4pn{2Vo4!i8J&jE)Vam)#)oN>+t
zmt1kp4Y%BJ&jXJ<@yrXayz$NlpM3Gn55N2ou%|zP1QARKp@b1m1d&7$O$@Qb5l;e%
zB#}%Csicuk2AO1$O%A!_kxv1I6j4kGrIb-l1(j4$O%1iwQBMPnG|@~8t+dfj2c2}$
zO%J{F(a!*b3^B|Iql_`m1d~iL%?z{5G0y^vEV0ZAtE{ok2AgcL%?`WlvCjdA9C6GE
zr<`%l1(#fL%?-EQanA#fJn_s6ue|Zj2cLZL%@4o)5%9n2Par`A6GA9qgcCs|QA86%
zEOEq>Kq5&blR_$Kq?17=S!9z#E_virKp{mGQ$i_alv6<^Ra8?$Ep^n>KqE~w(?Tn4
zw9`Q+U3Ak!FMaegz#u~mGr}lij5EO`Q%p0%EOX4Wz#>a5v%)HCth2!;TWqt#E_>{A
zz#&H*bHXWSoO8h?S6p+$EqC1Wz#~sQ^TI1{yz{{)Uwre!FMkB=>rWs-1QS9iVT2Pw
zBvC{YLo9K`lRzR#B$GlaX{3`uCRt>YLoRvbQ$Qg_6jMSeWt3AvB~?^YLoIdG(?BCl
zG}A&WZM4%tCtY;YLoa>wGr%B23^T$gV~jJwBvVW?!z^>mv%n%tEVIHYYpk=uCR=Q?
z!!CR5bHE`-9CN}cXPk4vC0AT?!!38*^S~odJoCaUZ@lxtCtrN?!!LgX{BQabND#q<
z5K0)~L=Z_7(Zmo-9PuQOND|4UkV+cqWROV~+2oK*9{ChdND;-9P)Zr)R8UD3)znZ+
z9rZNONE6Mp&`KNabkIo`-Sp5)AN>q4$PmMfFv=L?Ofbn5)66i-9P=!&$P&w}u*w?i
zY_Q1|+w8E*9{U_{$Pve!aLO6yTyV)1*W7T+9rrx&$P>@J@X8zSeDKK^-~8~)9{~sY
z6G#xjgb+#?;Y1Kg6w$;GOC0eekVq2Aq>xG)>12>e7TM&GOCI?YP)HHQlu$|;<y25f
z71h*GOC9wz&`1-_w9rZ$?R3yd7v1#GOCS9VFvt+Yj4;X=<4iEg6w}Nw%N+A8u*ee2
ztgy-&>uj*e7TfHw%O3k2aL5tIoN&q+=Ui~f71!Kw%N_ST@W>O-yzt5!?|ksd7vKEw
z%O3&%yZ!_cL@*(Q5=J-?L=r_bF~kx_JP9O{L^3I)l14fiWRgWTIpmT@J_Qs~L@_0l
zQbsuyR8mDXHPli^Jq<L{L^Cb4(ndQSbkapPJ@nE?KLZRh#4sa_GR8O)OftnZGt4r_
zJPRzc#4;<avc@_aY_i2RJM6N@J_j6f#4#tFa>h9qTyn)VH{5c^Jr6wc#4|6v^2R$K
zeDcLNKm77Xz@h#G5=1Z|gc3$L5kwM2G%>^yM?486l0-5oq>@HD8Dx@0HaX;yM?M7<
zQbaK&lu|}H6;x71H8s>yM?DQR(nK>Yw9-a99dyz~H$C*yM?V7$GQ=<=j55YJ6HGG2
zG&9UH$2<!xvcxhgtg^;B8*H-0HaqOH$36!fa>OwwoN~rF7hH11H8<RH$2|``^29SQ
zyz<67AAIt~H$VLHN5KD~KY;`hObDTb5l#e=L=jC4vBVKi0*NG%ObV%_kxmAgWRXn{
zx#W>g0fiJ%ObMlwQBDPwR8dV0wbW5h1C2D%Obe~F(M|`QbkR)@z4Xz~0D}xM%m|~5
zG0p^&Ofk(2v&=Ei0*frM%nGZlvCamYY_ZJ_yX>*g0f!uM%n7HQan1#oTyf0}x7=~h
z1CKoM%nPr)@y-XIeDTc>zx)w!q(6ZK5ljf7gb_{zkwg(q46(!!PXdV~kxUAyq>)Yr
znPibo4!Pu!PXUD#QA`P?lu=Fvl~hqp4Ykx!PXmoK(M$`iw9!rnopjMn554r!&j5oA
zG0X^~j4{pxlT0zq471EJ&jO1qvCImqtg+4pn{2Vo4!i8J&jE)Vam)#)oN>+tmt1kp
z4Y%BJ&jXJ<@yrXayz$NlpM3Gn55N2o@PF!0AVCBZLMUN`6G0?VL=!_Sam15AB1t5Z
zLMmyblR+j~WRpWKdE`?-Aw?8ZLMdgGQ$ZzFR8vDOb=1>9BTY2ZLMv^w(?KU)bkjpG
zee^THAVUl@!YE^mGr=TNOf$nQbIh~AB1<f@!YXU5v%w}?Y_r2Id+c+-Ax9i@!YOB*
zbHOE7Tyw)Mcii*9BTqc@!YgmQ^T8)yeDlLEe|X>m{vnVcf(ap%Fv5u-k|?5yA(lAe
zNg$CVl1U+zG}6f+lPt2yA(uSzDWH%diYcL#GRmo-k}9gHp_V%8X`qoNnrWexHrnZ+
zlP<dHp_e}T8DNkhh8ba$F~*r-k}0N{VU{`OSzwVRmRVtyHP+c+lP$K{VV6DjIpB~Z
zjyd6!GtRl-k}Iyc;g&n@dEk*Jo_XPwH{SW+lP|vc;g>%G{xAIrB#2-_2qlbgB8Vi4
zXkv&Zj(8GCB#C5FNF|MQGRP#0Y;wpYk9-O!q=;flD5Z>YDyXE2YHFyZj(Qqsq={x)
zXr+yII_RW}ZhGjYkA4OiWQbu#7-fucCYWT3X=a#Zj(HYXWQk=~SY?fMHrQl~ZFbmY
zk9`g}<cMQVIOU9UF1X~1Yi_vZj(Z+><cViqc;$_EKKSH|Z+`gY4=<a5e+VRqU_uBb
zjBp}|B#LNah$W7A5=bP8WKu{ajdU`|B#Ufv$R&?_3MizAVoE5bjB+Zdq>5^4sHKj2
z8fc`6W?E>ajdnWdq>FBP=%tT-1{h?BVMZ8bjBzHIWQu8Km}QQ67FcA7WmZ^ajdeEI
zWQ%Qf*kzA>4mjk9V@^2bjB_ry<ce!<xaE#}9(d%5XI^;ajdwoy<cn{9_~nm)|66|o
z2_l#fLJ1?B2qK9hniyh<Bc22jNg|mPQb{A73^K_gn;de<BcB2aDWaGXN-3k93M#3h
zni^`Uqn-vDX`-1HT4|%54m#<gn;v@Uqn`l=8Df|bMj2zA2_~6hni*!9W1a;TSz?(L
zR#{`64K~?gn;mx9W1j;KIpUZTPC4V83og0hnj3Dp<DLf|dE%KDUU}o44?g+gn;(Ap
z!-r<TKLip)Fd>8zMmP~f5=AsI#1cn52_%w4GAX2zMmiZ}l0`N-<dR1|1r$<5F(s5z
zMmZH!Qbjd2)KW)14K&h3GcB~zMmrsJ(nU8t^wLK^0}L|6Fe8jI#yAs9GQ~7A%reJ3
z3oNq4GApdI#yT5pvc)z#?6Su`2OM(5F(;gI#yJ;Ua>X?_+;Yb~4?Ob3GcUaI#ycN;
z^2Ikl{PIV@|D!*F1QARKp@b1m1d&7$O$@Qb5l;e%B#}%Csicuk2AO1$O%A!_kxv1I
z6j4kGrIb-l1(j4$O%1iwQBMPnG|@~8t+dfj2c2}$O%J{F(a!*b3^B|Iql_`m1d~iL
z%?z{5G0y^vEV0ZAtE{ok2AgcL%?`WlvCjdA9C6GEr<`%l1(#fL%?-EQanA#fJn_s6
zue|Zj2cLZL%@4o);U_cT9|8#?m=Hn<Bb*2#i6WX9Vu>T31QJOinG{k<Bb^K~$s(H^
za>*l~0tzXjm=a1UqnrvVsiK-1YN?~11{!IinHE}Uqn!>q>7tt+dg-H|0R|ajm=Q)9
zW1I;lnPQq5W|?E21r}LinH5%9W1S5)*<za=cG+W}0}eUjm=jJp<D3gFx#F4|Zn@*0
z2OfFinHOGp<DCya`Qn=&e)%Kd|J9#Bf(RyrP{If&f=Hr>CWctzh$n$Wl1L_nRMJQ%
zgG{o>CWl<|$ftlpiYTUpQpzZ&f=a5WriNPTsHcHOnrNnlR@!K%gHF2WriWho=x2aI
zh8SjqQN|c&f=Q;BW`<ejm}h}SmRM$mRn}N%gH5*BW`|w&*yn&ljyUFoQ_eW&f=jNr
z=7wADxaWaKo_OYkSKfH%gHOKr=7(SYaKIGs4}k;`ObDTb5l#e=L=jC4vBVKi0*NG%
zObV%_kxmAgWRXn{x#W>g0fiJ%ObMlwQBDPwR8dV0wbW5h1C2D%Obe~F(M|`QbkR)@
zz4Xz~0D}xM%m|~5G0p^&Ofk(2v&=Ei0*frM%nGZlvCamYY_ZJ_yX>*g0f!uM%n7HQ
zan1#oTyf0}x7=~h1CKoM%nPr)@y-XIeDTc>zx)yK@AM~-Ac6@YlrX}HAd)Dei6NFa
z;z=NpB$7!Xl{C`HAd@Vz$sw0K@+qK@B8n-YlrqYxppq)8siBrS>S>^nCYouXl{VVx
zpp!1T>7kcC`Wax5A%+=YlrhGcV3H}OnPHYW=2>8oC6-xXl{MDcV3RGj*<qJG_Br5?
zBaS)Ylrzq`;F2q@x#5;O?s?#mC!TrXl{en`;FB-D`Qev894-g^Lm)u}6GA9qgcCs|
zQA86%EOEq>Kq5&blR_$Kq?17=S!9z#E_virKp{mGQ$i_alv6<^Ra8?$Ep^n>KqE~w
z(?Tn4w9`Q+U3Ak!FMaegz#u~mGr}lij5EO`Q%p0%EOX4Wz#>a5v%)HCth2!;TWqt#
zE_>{Az#&H*bHXWSoO8h?S6p+$EqC1Wz#~sQ^TI1{yz{{)Uwre!FMkC5d;JL{h+skp
zC5&()h$M<=Vu&S<coIk?iDXhpC5?14$Rvwwa>yl*d<rO}h+;}8rHpbasHBQ&YN(}-
zdKze?iDp`8rHytv=%kBodg!H(eg+t1h+#$;WsGqqm}H7+W|(D;c@|h?iDgz;WsP+<
z*kp@scGzW)eGWL}h+|GT<&1MKxa5j!Zn))+dmeb?iDzDT<&Aef_~eUke)#1N7fS*E
z5J(Wggb+#?;Y1Kg6w$;GOC0eekVq2Aq>xG)>12>e7TM&GOCI?YP)HHQlu$|;<y25f
z71h*GOC9wz&`1-_w9rZ$?R3yd7v1#GOCS9VFvt+Yj4;X=<4iEg6w}Nw%N+A8u*ee2
ztgy-&>uj*e7TfHw%O3k2aL5tIoN&q+=Ui~f71!Kw%N_ST@W>O-yzt5!?|ksd7vKEw
z%O3&%L4N`XBA5_D2_u{cB8eiJ7-ESdo&*v}BAFCYNh6&MGRY#F9CFDcp8^UgqL>m&
zDWjYUDygEH8fvMdo(39eqL~(2X``JEI_aXD9(w7cp8*CLVwe#|8DpFYCYfTI8D^Pd
zo&^?JVwn|IS!10IHrZmE9d_Acp92m#;+PXoIpdrQF1g~G8*aJdo(CRz;+Yp-dE=cA
zKKbICAAb46rE<VO1QJ9rA%qe}I1xk=MKm$Q5=T4<B$7ljDWsA{IvHe=MK(F)l1DxT
z6jDSnC6rP|ITch=MKv|lQb#=vG}1&fEws``I~{b=MK?Y4(nmi73^K$pBaAY}I1@}V
z#WXX_GRHg%EV9HhE3C4{IvZ@V#Wp+avd2CL9CE}lC!BJ|ITu`V#WgqFa>qRnJo3ad
zFTC=`J0E=V#Wz3v@<+h`r$2!N5ljf7gb_{zkwg(q46(!!PXdV~kxUAyq>)YrnPibo
z4!Pu!PXUD#QA`P?lu=Fvl~hqp4Ykx!PXmoK(M$`iw9!rnopjMn554r!&j5oAG0X^~
zj4{pxlT0zq471EJ&jO1qvCImqtg+4pn{2Vo4!i8J&jE)Vam)#)oN>+tmt1kp4Y%BJ
z&jXJ<@yrXayz$NlpM3Gn55N3jpd{cQ0tq6R5JCwfoCqR`BAOUti6fo_5=kPN6jDhe
zoeVO`BAXm?$s?Zv3MrzP5=tqfoC+$bqM90NsiU3-8fl`L7FubeoenzbqMIIi>7$<k
z1{q?Q5k?tfoCzkGVwxFdnPZ*>7FlAM6;@eeoeehGVw)Xy*<+sr4mskO6HYnfoC_|w
z;+h+7x#OM(9(m%K7hZYeoew_w;+r3S`6J-}*PlRw2quJ3!U!jVNTP@)hFIc=CxJwg
zNG63;(nu$ROtQ!(hg|Z=r+`9=D5iu`$|$FTN~)-)hFa>Vr-4SAXr_f$+GwYPPP*u(
zhhF;VXMjP57-ob~#u#UUNv4=)hFRvAXMshQSZ0M))>vnQO}5x(hh6sA=YT_wIOc>?
z&N%0SORl))hFk8q=YdC_c;<yy-gxJOPrmr(hhP3MOcwADfdmmu2%&@#P6Uxe5lsxS
z#1T&di6oIs3aO-#P6nA|kxdS{<dIJSg%nXt38j=#P6d@zQB4iC)KO0ZjWp3r3$3)#
zP6wTI(M=D%^wG}%gA6gu2&0TK&IFT8G0hCK%rVabi!8Cs3ahNK&IX%ovCR&<?6J=Q
zha7Rt38$QK&IOlTam@|4+;PtXk38|r3$MKK&Ig}-@y!pv{1NaU^(T-Zf(ap%Fv5u-
zk|?5yA(lAeNg$CVl1U+zG}6f+lPt2yA(uSzDWH%diYcL#GRmo-k}9gHp_V%8X`qoN
znrWexHrnZ+lP<dHp_e}T8DNkhh8ba$F~*r-k}0N{VU{`OSzwVRmRVtyHP+c+lP$K{
zVV6DjIpB~Zjyd6!GtRl-k}Iyc;g&n@dEk*Jo_XPwH{SW+lP|vc;g>%Qk_P-kAVCBZ
zLMUN`6G0?VL=!_Sam15AB1t5ZLMmyblR+j~WRpWKdE`?-Aw?8ZLMdgGQ$ZzFR8vDO
zb=1>9BTY2ZLMv^w(?KU)bkjpGee^THAVUl@!YE^mGr=TNOf$nQbIh~AB1<f@!YXU5
zv%w}?Y_r2Id+c+-Ax9i@!YOB*bHOE7Tyw)Mcii*9BTqc@!YgmQ^T8)yeDlLEe+2v|
z{Rt$9U_uBbjBp}|B#LNah$W7A5=bP8WKu{ajdU`|B#Ufv$R&?_3MizAVoE5bjB+Zd
zq>5^4sHKj28fc`6W?E>ajdnWdq>FBP=%tT-1{h?BVMZ8bjBzHIWQu8Km}QQ67FcA7
zWmZ^ajdeEIWQ%Qf*kzA>4mjk9V@^2bjB_ry<ce!<xaE#}9(d%5XI^;ajdwoy<cn{9
z_~j2n<N^N>ND#q<5K0)~L=Z_7(Zmo-9PuQOND|4UkV+cqWROV~+2oK*9{ChdND;-9
zP)Zr)R8UD3)znZ+9rZNONE6Mp&`KNabkIo`-Sp5)AN>q4$PmMfFv=L?Ofbn5)66i-
z9P=!&$P&w}u*w?iY_Q1|+w8E*9{U_{$Pve!aLO6yTyV)1*W7T+9rrx&$P>@J@X8zS
zeDKK^-~8~)9|8Yae*y_2m=Hn<Bb*2#i6WX9Vu>T31QJOinG{k<Bb^K~$s(H^a>*l~
z0tzXjm=a1UqnrvVsiK-1YN?~11{!IinHE}Uqn!>q>7tt+dg-H|0R|ajm=Q)9W1I;l
znPQq5W|?E21r}LinH5%9W1S5)*<za=cG+W}0}eUjm=jJp<D3gFx#F4|Zn@*02OfFi
znHOGp<DCya`Qn=&e)+=!Nx(k@5=1Z|gc3$L5kwM2G%>^yM?486l0-5oq>@HD8Dx@0
zHaX;yM?M7<QbaK&lu|}H6;x71H8s>yM?DQR(nK>Yw9-a99dyz~H$C*yM?V7$GQ=<=
zj55YJ6HGG2G&9UH$2<!xvcxhgtg^;B8*H-0HaqOH$36!fa>OwwoN~rF7hH11H8<RH
z$2|``^29SQyz<67AAIt~H$VLHN5Fs4pFn~LCWKJJ2q%I_qKGDjSmKB$fkcu>CWTbe
zNGF3#vdAWfT=K}LfI^BWri4<;D5ru-s;H)hTI#5$fkv8WriE78Xs3fty6C2dUi#=~
zfI)^BW`t437-xb>rkG}iS>~8$fkl>BW`$MOSZ9Mxw%BHeUG~`LfJ2Tr=7dwuIOl>(
zuDIrgTkg2$fk&Qr=7m?@c;|ypzWC;cU;eN>7Vr;&1QARKp@b1m1d&7$O$@Qb5l;e%
zB#}%Csicuk2AO1$O%A!_kxv1I6j4kGrIb-l1(j4$O%1iwQBMPnG|@~8t+dfj2c2}$
zO%J{F(a!*b3^B|Iql_`m1d~iL%?z{5G0y^vEV0ZAtE{ok2AgcL%?`WlvCjdA9C6GE
zr<`%l1(#fL%?-EQanA#fJn_s6ue|Zj2cLZL%@4o)5%6F2Cy*e52_cj)!igY~D58lW
zmN?=`Adw`JNg<Uq(#ar`EV9WVmpt+*ppYVpDWQ}y%Bi4|DypfWmOAQbpphn;X`z)i
z+UcN^F1qQVmp=L#V2~k(8DW$$#+hJ}DW;iWmO18GV38%3Sz(nm*4bc_Ew<TVmp%45
z;E*GZIpLHu&bi={E3UcWmOJiw;E^YudEu2e-ud8@FTVNVmp?3y2K+-HK?D;*C}D&X
zK_pQ`6GJR<#FIcGNhFg(DruyXK_*#blS3|f<WoQ)MHEv)DP@#XK_yjGQ$sCv)YCvC
zO*GR&D{ZvXK_^{w(?c(P^fSO9Lku&*C}WH>!6g5!fjbHX1_0AEsy|7RBuSDaNs=T<
zk|arzBuSDaNk)<+Ns=T<k|arzBuSDaNs=T<k|arzjIq{QYps!_Jr}3G`{(+5r<i7j
zS>~8$fkl>BW`$MOSZ9Mxw%BHeUG~`LfJ2Tr=7dwuIOl>(uDIrgTkg2$fk&Qr=7m?@
zc;|ypzWC;cU;YUAPyGobh+skpC5&()h$M<=Vu&S<coIk?iDXhpC5?14$Rvwwa>yl*
zd<rO}h+;}8rHpbasHBQ&YN(}-dKze?iDp`8rHytv=%kBodg!H(eg+t1h+#$;WsGqq
zm}H7+W|(D;c@|h?iDgz;WsP+<*kp@scGzW)eGWL}h+|GT<&1MKxa5j!Zn))+dmeb?
ziDzDT<&Aef_~eUke)#1NPvZgq2qcJLLI@>{a3Y8#ifCepC60I!NF<45Qb;9@bTY^!
zi)?bpC69axD5QvDN+_j_aw@2#ifU@8rH*<UXrzf|T4<$>b~@;!i*9=8rH_6F7-WcH
zMi^y`aVD5#ifLw;WsZ3kSY(N1R#;_?bvD>!i*0t;WsiLhIOK?9PB`U^b1t~#ifeAT
z<&JwEc;ty^UU=n=cRu*!i*J7T<&S{>(w{(r2quJ3!U!jVNTP@)hFIc=CxJwgNG63;
z(nu$ROtQ!(hg|Z=r+`9=D5iu`$|$FTN~-?rDEU9y&H`%OwbW5h1C2D%Obe~F(M|`Q
zbkR)@z4Xz~0D}xM%m|~5G0p^&Ofk(2v&=Ei0*frM%nGZlvCamYY_ZJ_yX>*g0f!uM
z%n7HQan1#oTyf0}x7=~h1CKoM%nPr)@y-XIeDTc>zx-h^BH$l^1QARKp@b1m1d&7$
zO$@Qb5l;e%B#}%Csicuk2AO1$O%A!_kxv1I6j4kGrIb-l1(j4$O%1iwQBMPnG|@~8
zt+dfj2c2}$O%J{F(a!*b3^B|Iql_`m1d~iL%?z{5G0y^vEV0ZAtE{ok2AgcL%?`Wl
zvCjdA9C6GEr<`%l1(#fL%?-EQanA#fJn_s6ue|Zj2cLZL%@4o)5%53sCy*e52_cj)
z!igY~D58lWmN?=`Adw`JNg<Uq(#ar`EV9WVmpt+*ppYVpDWQ}y%Bi4|DypfWmOAQb
zpphn;X`z)i+UcN^F1qQVmp=L#V2~k(8DW$$#+hJ}DW;iWmO18GV38%3Sz(nm*4bc_
zEw<TVmp%45;E*GZIpLHu&bi={E3UcWmOJiw;E^YudEu2e-ud8@FTVNVmp_cf1pFhA
zAc6@YlrX}HAd)Dei6NFa;z=NpB$7!Xl{C`HAd@Vz$sw0K@+qK@B8n-YlrqYxppq)8
zsiBrS>S>^nCYouXl{VVxpp!1T>7kcC`Wax5A%+=YlrhGcV3H}OnPHYW=2>8oC6-xX
zl{MDcV3RGj*<qJG_Br5?BaS)Ylrzq`;F2q@x#5;O?s?#mC!TrXl{en`;FB-D`Qev8
z0{*A|1QJ9rA%qe}I1xk=MKm$Q5=T4<B$7ljDWsA{IvHe=MK(F)l1DxT6jDSnC6rP|
zITch=MKv|lQb#=vG}1&fEws``I~{b=MK?Y4(nmi73^K$pBaAY}I1@}V#WXX_GRHg%
zEV9HhE3C4{IvZ@V#Wp+avd2CL9CE}lC!BJ|ITu`V#WgqFa>qRnJo3adFTC=`J0E=V
z#Wz3v@`s_QfPVxML@*(Q5=J-?L=r_bF~kx_JP9O{L^3I)l14fiWRgWTIpmT@J_Qs~
zL@_0lQbsuyR8mDXHPli^Jq<L{L^Cb4(ndQSbkapPJ@nE?KLZRh#4sa_GR8O)OftnZ
zGt4r_JPRzc#4;<avc@_aY_i2RJM6N@J_j6f#4#tFa>h9qTyn)VH{5c^Jr6wc#4|6v
z^2R$KeDcLNKm77X!2i;pK!OM+giyi=CxS?#h$e<u;)o}KM3P7*g;df=Cxc9~$R>we
z^2n!vLW(G+gi^{Vr-DkVsHTQm>ZqrIMw)1*g;v^Vr-M$q=%$BW`sinXL53J+gi*#A
zXM#zlm}Z7q=9p)JMV44*g;myAXM;_)*k*@a_SoluLykD+gj3Eq=YmVFxaNji?zrcH
zN1k}*g;(Bq=Yvna_~wUS{xA|3@Q*-(2quJ3!U!jVNTP@)hFIc=CxJwgNG63;(nu$R
zOtQ!(hg|Z=r+`9=D5iu`$|$FTN~)-)hFa>Vr-4SAXr_f$+GwYPPP*u(hhF;VXMjP5
z7-ob~#u#UUNv4=)hFRvAXMshQSZ0M))>vnQO}5x(hh6sA=YT_wIOc>?&N%0SORl))
zhFk8q=YdC_c;<yy-gxJOPrmr(hhP2(_}}^yND#q<5K0)~L=Z_7(Zmo-9PuQOND|4U
zkV+cqWROV~+2oK*9{ChdND;-9P)Zr)R8UD3)znZ+9rZNONE6Mp&`KNabkIo`-Sp5)
zAN>q4$PmMfFv=L?Ofbn5)66i-9P=!&$P&w}u*w?iY_Q1|+w8E*9{U_{$Pve!aLO6y
zTyV)1*W7T+9rrx&$P>@J@X8zSeDKK^-~8~)9|j@={t-wJ!GsV>7~w<^Nfgn<5KA2K
zB#=lF$)u1<8tG(^Nfz1UkV_u<6i`SJ#gtG=8Rb+^Nfp)9P)i;4G|)&B&9u-;8|`$^
zNf+Jp&`Tfv3^2$L!;CP>7~@PZ$rRJfFv}eCEU?HD%dD`<8tZJZ$rjt}u*)9%9B{}H
z$DDA=8RuMZ$rab!aLXO{Jn+a9&%E%;8}EGZ$rs=J@XH?o|3`lU2_l#fLJ1?B2qK9h
zniyh<Bc22jNg|mPQb{A73^K_gn;de<BcB2aDWaGXN-3k93M#3hni^`Uqn-vDX`-1H
zT4|%54m#<gn;v@Uqn`l=8Df|bMj2zA2_~6hni*!9W1a;TSz?(LR#{`64K~?gn;mx9
zW1j;KIpUZTPC4V83og0hnj3Dp<DLf|dE%KDUU}o44?g+gn;(Ap!#HffKLQCNm=Hn<
zBb*2#i6WX9Vu>T31QJOinG{k<Bb^K~$s(H^a>*l~0tzXjm=a1UqnrvVsiK-1YN?~1
z1{!IinHE}Uqn!>q>7tt+dg-H|0R|ajm=Q)9W1I;lnPQq5W|?E21r}LinH5%9W1S5)
z*<za=cG+W}0}eUjm=jJp<D3gFx#F4|Zn@*02OfFinHOGp<DCya`Qn=&e)%Kd|LRX5
zK?D;*C}D&XK_pQ`6GJR<#FIcGNhFg(DruyXK_*#blS3|f<WoQ)MHEv)DP@#XK_yjG
zQ$sCv)YCvCO*GR&D{ZvXK_^{w(?c(P^fSO9Lku&*C}WH>!6Z{mGs7%%%(K8EODwa(
zDr>B>!6sX5v%@ZX>~p{&M;vp)DQBE>!6jE*bHgon-1ERAPdxL&D{s8>!6#pQ^TRKH
z7={k`M<7826GA9qgcCs|QA86%EOEq>Kq5&blR_$Kq?17=S!9z#E_virKp{mGQ$i_a
zlv6<^Ra8?$Ep^n>KqE~w(?Tn4w9`Q+U3Ak!FMaegz#u~mGr}lij5EO`Q%p0%EOX4W
zz#>a5v%)HCth2!;TWqt#E_>{Az#&H*bHXWSoO8h?S6p+$EqC1Wz#~sQ^TI1{yz{{)
zUwre!FMkC5Km7?Lh+skpC5&()h$M<=Vu&S<coIk?iDXhpC5?14$Rvwwa>yl*d<rO}
zh+;}8rHpbasHBQ&YN(}-dKze?iDp`8rHytv=%kBodg!H(eg+t1h+#$;WsGqqm}H7+
zW|(D;c@|h?iDgz;WsP+<*kp@scGzW)eGWL}h+|GT<&1MKxa5j!Zn))+dmeb?iDzDT
z<&Aef_~eUke)#1NqwoR$2qcJLLI@>{a3Y8#ifCepC60I!NF<45Qb;9@bTY^!i)?bp
zC69axD5QvDN+_j_aw@2#ifU@8rH*<UXrzf|T4<$>b~@;!i*9=8rH_6F7-WcHMi^y`
zaVD5#ifLw;WsZ3kSY(N1R#;_?bvD>!i*0t;WsiLhIOK?9PB`U^b1t~#ifeAT<&JwE
zc;ty^UU=n=cRu*!i*J7T<&S{>)}KIv2quJ3!U!jVNTP@)hFIc=CxJwgNG63;(nu$R
zOtQ!(hg|Z=r+`9=D5iu`$|$FTN~)-)hFa>Vr-4SAXr_f$+GwYPPP*u(hhF;VXMjP5
z7-ob~#u#UUNv4=)hFRvAXMshQSZ0M))>vnQO}5x(hh6sA=YT_wIOc>?&N%0SORl))
bhFk8q=YdC_c;<yy-gxJOPrms6@1y?@wPD!l

literal 0
HcmV?d00001

diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index 9831b13d4b039..bc2100efb485c 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -26,10 +26,10 @@ llama_build_and_test_executable(test-quantize-fns.cpp)
 llama_build_and_test_executable(test-quantize-perf.cpp)
 llama_build_and_test_executable(test-sampling.cpp)
 llama_build_executable(test-tokenizer-0.cpp)
-llama_test_executable(test-tokenizer-0.llama test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.bin)
+llama_test_executable(test-tokenizer-0.llama test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
 llama_build_executable(test-tokenizer-1.cpp)
-llama_test_executable(test-tokenizer-1.llama test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.bin)
-llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.bin)
+llama_test_executable(test-tokenizer-1.llama  test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
+#llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf)
 llama_build_and_test_executable(test-grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp)
 llama_build_and_test_executable(test-grad0.cpp) # SLOW
 # llama_build_and_test_executable(test-opt.cpp) # SLOW
diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp
index f973271a34eb0..9f5382679e380 100644
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -89,6 +89,8 @@ int main(int argc, char **argv) {
         return 2;
     }
 
+    bool success = true;
+
     for (const auto & test_kv : k_tests()) {
         std::vector<llama_token> res = llama_tokenize(ctx, test_kv.first, true);
         fprintf(stderr, "%s : '%s' tokenized to '%s'\n",
@@ -103,7 +105,8 @@ int main(int argc, char **argv) {
         }
 
         if (!correct) {
-            fprintf(stderr, "%s : failed test: '%s'\n", __func__, test_kv.first.c_str());
+            fprintf(stderr, "%s : failed test:    '%s'\n", __func__, test_kv.first.c_str());
+            fprintf(stderr, "%s : detokenized to: '%s'\n", __func__, unescape_whitespace(ctx, test_kv.second).c_str());
             fprintf(stderr, "%s : expected tokens: ", __func__);
             for (const auto & t : test_kv.second) {
                 fprintf(stderr, "%6d, ", t);
@@ -115,9 +118,7 @@ int main(int argc, char **argv) {
             }
             fprintf(stderr, "\n");
 
-            llama_free_model(model);
-            llama_free(ctx);
-            return 3;
+            success = false;
         }
     }
 
@@ -126,5 +127,5 @@ int main(int argc, char **argv) {
 
     llama_backend_free();
 
-    return 0;
+    return success ? 0 : 3;
 }

From 2ddd9681d662df23b8bb3e2399c1be3097e066f6 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 17:22:43 +0300
Subject: [PATCH 169/242] convert.py : update to support GGUF output

---
 convert-new.py | 1013 ------------------------------------------------
 convert.py     |  868 +++++++++++++----------------------------
 2 files changed, 268 insertions(+), 1613 deletions(-)
 delete mode 100755 convert-new.py
 mode change 100644 => 100755 convert.py

diff --git a/convert-new.py b/convert-new.py
deleted file mode 100755
index b243356f0db28..0000000000000
--- a/convert-new.py
+++ /dev/null
@@ -1,1013 +0,0 @@
-#!/usr/bin/env python
-
-import gguf
-import argparse
-import concurrent.futures
-import copy
-import enum
-import faulthandler
-import functools
-import io
-import itertools
-import json
-import math
-import mmap
-import pickle
-import re
-import signal
-import struct
-import sys
-import zipfile
-import numpy as np
-
-from abc import ABCMeta, abstractmethod
-from dataclasses import dataclass
-from pathlib import Path
-from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List, Literal, Optional, Sequence, Tuple, TypeVar, Union)
-from sentencepiece import SentencePieceProcessor  # type: ignore
-
-if TYPE_CHECKING:
-    from typing_extensions import TypeAlias
-
-if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
-    faulthandler.register(signal.SIGUSR1)
-
-NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
-
-ARCH=gguf.MODEL_ARCH.LLAMA
-NAMES=gguf.MODEL_TENSOR_NAMES[ARCH]
-
-#
-# data types
-#
-
-@dataclass(frozen=True)
-class UnquantizedDataType:
-    name: str
-
-DT_F16  = UnquantizedDataType('F16')
-DT_F32  = UnquantizedDataType('F32')
-DT_I32  = UnquantizedDataType('I32')
-DT_BF16 = UnquantizedDataType('BF16')
-
-DataType = Union[UnquantizedDataType]
-
-DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
-    DT_BF16: np.dtype(np.uint16),
-    DT_F16:  np.dtype(np.float16),
-    DT_F32:  np.dtype(np.float32),
-    DT_I32:  np.dtype(np.int32),
-}
-
-NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
-    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
-
-SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
-    'BF16': DT_BF16,
-    'F16': DT_F16,
-    'F32': DT_F32,
-    'I32': DT_I32,
-}
-
-class GGMLFileType(enum.Enum):
-    AllF32    = 0
-    MostlyF16 = 1  # except 1d tensors
-
-    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
-        if len(tensor.shape) == 1:
-            # 1D tensors are always F32.
-            return DT_F32
-        elif self == GGMLFileType.AllF32:
-            return DT_F32
-        elif self == GGMLFileType.MostlyF16:
-            return DT_F16
-        else:
-            raise ValueError(self)
-
-
-#
-# hparams loading
-#
-
-@dataclass
-class Params:
-    n_vocab:    int
-    n_embd:     int
-    n_mult:     int
-    n_layer:    int
-    n_ctx:      int
-    n_ff:       int
-    n_head:     int
-    n_head_kv:  int
-    f_norm_eps: float
-
-    @staticmethod
-    def find_n_mult(n_ff: int, n_embd: int) -> int:
-        # hardcoded magic range
-        for n_mult in range(8192, 1, -1):
-            calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
-            if calc_ff == n_ff:
-                return n_mult
-        raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
-
-    @staticmethod
-    def guessed(model: 'LazyModel') -> 'Params':
-        # try transformer naming first
-        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
-
-        # try transformer naming first
-        if "model.layers.0.self_attn.q_proj.weight" in model:
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
-        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
-        else:
-            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
-
-        if n_layer < 1:
-            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
-                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
-
-        n_head = n_embd // 128 # guessed
-        n_mult = 256           # guessed
-
-        # TODO: verify this
-        n_ff = int(2 * (4 * n_embd) / 3)
-        n_ff = n_mult * ((n_ff + n_mult - 1) // n_mult)
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = -1,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head,
-            f_norm_eps = 1e-5,
-        )
-
-    @staticmethod
-    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab    = config["vocab_size"];
-        n_embd     = config["hidden_size"];
-        n_layer    = config["num_hidden_layers"];
-        n_ff       = config["intermediate_size"];
-        n_head     = config["num_attention_heads"];
-        n_head_kv  = config["num_key_value_heads"];
-        f_norm_eps = config["rms_norm_eps"];
-
-        n_mult = Params.find_n_mult(n_ff, n_embd);
-
-        if "max_sequence_length" in config:
-            n_ctx = config["max_sequence_length"]
-        elif "max_position_embeddings" in config:
-            n_ctx = config["max_position_embeddings"]
-        else:
-            raise Exception("failed to guess 'n_ctx'. This model is unknown or unsupported.\n"
-                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = n_ctx,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head_kv,
-            f_norm_eps = f_norm_eps,
-        )
-
-    # LLaMA v2 70B params.json
-    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
-    @staticmethod
-    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab    = config["vocab_size"];
-        n_embd     = config["dim"];
-        n_layer    = config["n_layers"];
-        n_mult     = config["multiple_of"];
-        n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
-        n_ff       = -1;
-        n_head     = config["n_heads"];
-        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head;
-        f_norm_eps = config["norm_eps"];
-
-        if n_vocab == -1:
-            n_vocab = model["tok_embeddings.weight"].shape[0]
-
-        if n_ff == -1:
-            n_ff = model["layers.0.feed_forward.w1.weight"].shape[0]
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = n_ctx,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head_kv,
-            f_norm_eps = f_norm_eps,
-        )
-
-    @staticmethod
-    def load(model_plus: 'ModelPlus') -> 'Params':
-        hf_config_path   = model_plus.paths[0].parent / "config.json"
-        orig_config_path = model_plus.paths[0].parent / "params.json"
-
-        if hf_config_path.exists():
-            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
-        elif orig_config_path.exists():
-            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
-        else:
-            params = Params.guessed(model_plus.model)
-
-        return params
-
-
-#
-# vocab
-#
-
-class BpeVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
-        self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
-        added_tokens: Dict[str, int]
-        if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
-        else:
-            added_tokens = {}
-        vocab_size: int = len(self.bpe_tokenizer)
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
-        self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
-
-    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        tokenizer = self.bpe_tokenizer
-        from transformers.models.gpt2 import tokenization_gpt2
-        byte_encoder = tokenization_gpt2.bytes_to_unicode()
-        byte_decoder = {v: k for k, v in byte_encoder.items()}
-        for i, item in enumerate(tokenizer):
-            text: bytes = item.encode("utf-8")
-            score: float = -i
-            yield text, score
-
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        yield from self.bpe_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
-class SentencePieceVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
-        self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
-        added_tokens: Dict[str, int]
-        if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
-        else:
-            added_tokens = {}
-        vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids   = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
-
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
-        self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
-
-    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        tokenizer = self.sentencepiece_tokenizer
-        for i in range(tokenizer.vocab_size()):
-            piece = tokenizer.id_to_piece(i)
-            text: bytes = piece.encode("utf-8")
-            score: float = tokenizer.get_score(i)
-            yield text, score
-
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        yield from self.sentencepiece_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-Vocab = Union[BpeVocab, SentencePieceVocab]
-
-
-#
-# data loading
-# TODO: reuse (probably move to gguf.py?)
-#
-
-def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
-    if n_head_kv is not None and n_head != n_head_kv:
-        n_head //= n_head_kv
-    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                .swapaxes(1, 2)
-                .reshape(weights.shape))
-
-
-class Tensor(metaclass=ABCMeta):
-    data_type: DataType
-
-    @abstractmethod
-    def astype(self, data_type: DataType) -> 'Tensor': ...
-    @abstractmethod
-    def permute(self, n_head: int, n_head_kv: int) -> 'Tensor': ...
-    @abstractmethod
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def part(self, n_part: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
-
-
-def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
-    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
-    fp32_arr = bf16_arr.astype(np.uint32) << 16
-    return fp32_arr.view(np.float32)
-
-
-class UnquantizedTensor(Tensor):
-    def __init__(self, ndarray: NDArray) -> None:
-        assert isinstance(ndarray, np.ndarray)
-        self.ndarray = ndarray
-        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
-
-    def astype(self, data_type: DataType) -> Tensor:
-        dtype = DATA_TYPE_TO_NUMPY[data_type]
-        if self.data_type == DT_BF16:
-            self.ndarray = bf16_to_fp32(self.ndarray)
-        return UnquantizedTensor(self.ndarray.astype(dtype))
-
-    def to_ggml(self) -> 'UnquantizedTensor':
-        return self
-
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
-
-    def part(self, n_part: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
-
-    def permute(self, n_head: int, n_head_kv: int) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head, n_head_kv))
-
-
-def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
-    tensor = lazy_tensor.load()
-    assert isinstance(tensor, UnquantizedTensor)
-
-    # double-check:
-    actual_shape = list(tensor.ndarray.shape)
-    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
-    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
-        if convert:
-            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
-        else:
-            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
-
-    return tensor.ndarray
-
-
-GGMLCompatibleTensor = Union[UnquantizedTensor]
-
-
-class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int, n_head_kv: int) -> None:
-        self.base = base
-        self.n_head = n_head
-        self.data_type = self.base.data_type
-
-    def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head, self.n_head_kv)
-
-    def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head, self.n_head_kv)
-
-    def permute(self, n_head: int, n_head_kv: int) -> Tensor:
-        raise Exception("shouldn't permute twice")
-
-
-@dataclass
-class LazyTensor:
-    _load: Callable[[], Tensor]
-    shape: List[int]
-    data_type: DataType
-    description: str
-
-    def load(self) -> Tensor:
-        ret = self._load()
-        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
-        return ret
-
-    def astype(self, data_type: DataType) -> 'LazyTensor':
-        self.validate_conversion_to(data_type)
-
-        def load() -> Tensor:
-            return self.load().astype(data_type)
-        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
-
-    def validate_conversion_to(self, data_type: DataType) -> None:
-        if data_type == self.data_type:
-            return
-
-
-LazyModel = Dict[str, LazyTensor]
-
-
-@dataclass
-class ModelPlus:
-    model: LazyModel
-    paths: List[Path]  # Where this was read from.
-    format: Literal['ggml', 'torch', 'safetensors']
-    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
-
-
-def merge_sharded(models: List[LazyModel]) -> LazyModel:
-    # Original LLaMA models have each file contain one part of each tensor.
-    # Use a dict instead of a set to preserve order.
-    names = {name: None for model in models for name in model}
-
-    def convert(name: str) -> LazyTensor:
-        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
-        if len(lazy_tensors) == 1:
-            # only one file; don't go through this procedure since there might
-            # be quantized tensors
-            return lazy_tensors[0]
-        if len(lazy_tensors[0].shape) == 1:
-            # the tensor is just duplicated in every file
-            return lazy_tensors[0]
-        if name.startswith('tok_embeddings.') or \
-           name.endswith('.attention.wo.weight') or \
-           name.endswith('.feed_forward.w2.weight'):
-            # split by columns
-            axis = 1
-        else:
-            # split by rows
-            axis = 0
-        concatenated_shape = list(lazy_tensors[0].shape)
-        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
-
-        def load() -> UnquantizedTensor:
-            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
-            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
-            return UnquantizedTensor(concatenated)
-        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
-        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
-    return {name: convert(name) for name in names}
-
-
-def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
-    formats = set(mp.format for mp in models_plus)
-    assert len(formats) == 1, "different formats?"
-    format = formats.pop()
-    paths = [path for mp in models_plus for path in mp.paths]
-    # Use the first non-None vocab, if any.
-    try:
-        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
-    except StopIteration:
-        vocab = None
-
-    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
-        # Transformers models put different tensors in different files, but
-        # don't split indivdual tensors between files.
-        model: LazyModel = {}
-        for mp in models_plus:
-            model.update(mp.model)
-    else:
-        model = merge_sharded([mp.model for mp in models_plus])
-
-    return ModelPlus(model, paths, format, vocab)
-
-
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_head_kv: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head, n_head_kv)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_head_kv}) ' + lazy_tensor.description)
-
-def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute_part(n_part, n_head)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
-
-def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().part(n_part)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
-
-
-# Functionality that simulates `torch.load` but where individual tensors are
-# only loaded into memory on demand, not all at once.
-# PyTorch can't do this natively as of time of writing:
-# - https://github.com/pytorch/pytorch/issues/64327
-# This allows us to de-shard without multiplying RAM usage, and also
-# conveniently drops the PyTorch dependency (though we still need numpy).
-
-
-@dataclass
-class LazyStorageKind:
-    data_type: DataType
-
-
-@dataclass
-class LazyStorage:
-    load: Callable[[int, int], NDArray]
-    kind: LazyStorageKind
-    description: str
-
-
-class LazyUnpickler(pickle.Unpickler):
-    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
-        super().__init__(fp)
-        self.data_base_path = data_base_path
-        self.zip_file = zip_file
-
-    def persistent_load(self, pid: Any) -> Any:
-        assert pid[0] == 'storage'
-        assert isinstance(pid[1], LazyStorageKind)
-        data_type = pid[1].data_type
-        filename_stem = pid[2]
-        filename = self.data_base_path + '/' + filename_stem
-        info = self.zip_file.getinfo(filename)
-
-        def load(offset: int, elm_count: int) -> NDArray:
-            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
-            if dtype is None:
-                raise Exception("tensor stored in unsupported format")
-            fp = self.zip_file.open(info)
-            fp.seek(offset * dtype.itemsize)
-            size = elm_count * dtype.itemsize
-            data = fp.read(size)
-            assert len(data) == size
-            return np.frombuffer(data, dtype)
-        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
-        return LazyStorage(load=load, kind=pid[1], description=description)
-
-    # @staticmethod
-    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
-                               # pyright: ignore[reportSelfClsParameterName]
-                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
-        assert isinstance(storage, LazyStorage)
-
-        def load() -> UnquantizedTensor:
-            elm_count = stride[0] * size[0]
-            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
-        description = f'pickled storage_offset={storage_offset} in {storage.description}'
-        return LazyTensor(load, list(size), storage.kind.data_type, description)
-
-    # @staticmethod
-    def rebuild_from_type_v2(func, new_type, args, state):
-        return func(*args)
-
-    CLASSES: Dict[Any, Any] = {
-        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
-        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
-        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
-        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
-        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
-        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
-        ('torch', 'Tensor'): LazyTensor,
-    }
-
-    def find_class(self, module: str, name: str) -> Any:
-        if not module.startswith('torch'):
-            return super().find_class(module, name)
-        return self.CLASSES[(module, name)]
-
-
-def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
-    zf = zipfile.ZipFile(outer_fp)
-    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
-    assert len(pickle_paths) == 1, pickle_paths
-    pickle_fp = zf.open(pickle_paths[0], 'r')
-    unpickler = LazyUnpickler(pickle_fp,
-                              data_base_path=pickle_paths[0][:-4],
-                              zip_file=zf)
-    model = unpickler.load()
-    as_dict = dict(model.items())
-    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
-
-
-def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
-    header_size, = struct.unpack('<Q', fp.read(8))
-    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
-    # Use mmap for the actual data to avoid race conditions with the file offset.
-    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
-    byte_buf = mapped[8 + header_size:]
-
-    def convert(info: Dict[str, Any]) -> LazyTensor:
-        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
-        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
-        shape: List[int] = info['shape']
-        begin, end = info['data_offsets']
-        assert 0 <= begin <= end <= len(byte_buf)
-        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
-        buf = byte_buf[begin:end]
-
-        def load() -> UnquantizedTensor:
-            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
-        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
-        return LazyTensor(load, shape, data_type, description)
-    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
-    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
-
-
-def must_read(fp: IO[bytes], length: int) -> bytes:
-    ret = fp.read(length)
-    if len(ret) < length:
-        raise Exception("unexpectedly reached end of file")
-    return ret
-
-
-@functools.lru_cache(maxsize=None)
-def lazy_load_file(path: Path) -> ModelPlus:
-    fp = open(path, 'rb')
-    first8 = fp.read(8)
-    fp.seek(0)
-    if first8[:2] == b'PK':
-        # A zip file, i.e. PyTorch format
-        return lazy_load_torch_file(fp, path)
-    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
-        # Probably safetensors
-        return lazy_load_safetensors_file(fp, path)
-    else:
-        raise ValueError(f"unknown format: {path}")
-
-
-In = TypeVar('In')
-Out = TypeVar('Out')
-
-def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
-    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
-    fast enough, this will stop calling `func` at some point rather than
-    letting results pile up in memory.  Specifically, there is a max of one
-    output value buffered per thread.'''
-    with concurrent.futures.ThreadPoolExecutor() as executor:
-        futures: List[concurrent.futures.Future[Out]] = []
-        items_rev = list(iterable)[::-1]
-        for i in range(min(concurrency, len(items_rev))):
-            futures.append(executor.submit(func, items_rev.pop()))
-        while futures:
-            result = futures.pop(0).result()
-            if items_rev:
-                futures.append(executor.submit(func, items_rev.pop()))
-            yield result
-
-
-def check_vocab_size(params: Params, vocab: Vocab) -> None:
-    if params.n_vocab != vocab.vocab_size:
-        assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
-        if params.n_vocab == vocab.vocab_size_base:
-            print("Ignoring added_tokens.json since model matches vocab size without it.")
-            vocab.added_tokens_list = []
-            vocab.vocab_size = vocab.vocab_size_base
-            return
-        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
-        if vocab.fname_added_tokens is not None:
-            msg += f" combined with {vocab.fname_added_tokens}"
-        msg += f" has {vocab.vocab_size})."
-        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
-            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
-        raise Exception(msg)
-
-
-class OutputFile:
-    def __init__(self, fname_out: Path) -> None:
-        self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
-
-    def add_meta_arch(self, params: Params) -> None:
-        self.gguf.add_context_length      (params.n_ctx)
-        self.gguf.add_embedding_length    (params.n_embd)
-        self.gguf.add_block_count         (params.n_layer)
-        self.gguf.add_feed_forward_length (params.n_ff)
-        self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
-        self.gguf.add_head_count          (params.n_head)
-        self.gguf.add_head_count_kv       (params.n_head_kv)
-        self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
-
-    def add_meta_vocab(self, vocab: Vocab) -> None:
-        tokens = []
-        scores = []
-        for text, score in vocab.all_tokens():
-            tokens.append(text)
-            scores.append(score)
-
-        self.gguf.add_tokenizer_model("llama")
-        self.gguf.add_token_list(tokens)
-        self.gguf.add_token_scores(scores)
-        #self.gguf.add_token_types(toktypes) # TODO: add this
-
-        # TODO: added / special tokens
-
-    def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
-        n_elements = 1
-        for dim in tensor.shape:
-            n_elements *= dim
-        data_type = DATA_TYPE_TO_NUMPY[tensor.data_type]
-        data_nbytes = n_elements * data_type.itemsize
-        self.gguf.add_tensor_info(name, tensor.shape, data_type, data_nbytes)
-
-    def write_meta(self) -> None:
-        self.gguf.write_header_to_file()
-        self.gguf.write_kv_data_to_file()
-
-    def write_tensor_info(self) -> None:
-        self.gguf.write_ti_data_to_file()
-
-    def close(self) -> None:
-        self.gguf.close()
-
-    @staticmethod
-    def write_vocab_only(fname_out: Path, params: Params, vocab: Vocab) -> None:
-        check_vocab_size(params, vocab)
-
-        of = OutputFile(fname_out)
-
-        # meta data
-        of.add_meta_arch(params)
-        of.add_meta_vocab(vocab)
-        of.write_meta()
-
-        of.close()
-
-    @staticmethod
-    def write_all(fname_out: Path, params: Params, model: LazyModel, vocab: Vocab) -> None:
-        check_vocab_size(params, vocab)
-
-        of = OutputFile(fname_out)
-
-        # meta data
-        of.add_meta_arch(params)
-        of.add_meta_vocab(vocab)
-
-        # tensor info
-        for name, lazy_tensor in model.items():
-            of.add_tensor_info(name, lazy_tensor)
-
-        of.write_meta()
-        of.write_tensor_info()
-
-        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
-            name, lazy_tensor = item
-            return lazy_tensor.load().to_ggml().ndarray
-
-        # tensor data
-        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8)
-        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
-            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
-            padi = len(str(len(model)))
-            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
-            of.gguf.write_tensor_data(ndarray)
-
-        of.close()
-
-def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
-    wq_type = model[NAMES[gguf.MODEL_TENSOR.ATTN_Q].format(bid=0)+".weight"].data_type
-
-    if output_type_str == "f32" or (output_type_str is None and wq_type == DT_F32):
-        return GGMLFileType.AllF32
-    if output_type_str == "f16" or (output_type_str is None and wq_type in (DT_F16, DT_BF16)):
-        return GGMLFileType.MostlyF16
-
-    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
-
-    raise Exception(f"Unexpected combination of types: {name_to_type}")
-
-def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
-    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
-            for (name, tensor) in model.items()}
-
-def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
-    tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
-
-    out: LazyModel = {}
-    for name, lazy_tensor in model.items():
-        name_new = name
-
-        if name in tmap:
-            name_new = tmap[name]
-        elif name.endswith(".weight") and name[:-7] in tmap:
-            name_new = tmap[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tmap:
-            name_new = tmap[name[:-5]] + ".bias"
-        else:
-            raise Exception(f"Unexpected tensor name: {name}")
-
-        if gguf.should_skip_tensor(ARCH, params.n_layer, name_new):
-            print(f"skipping tensor {name_new}")
-        else:
-            print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
-            out[name_new] = lazy_tensor
-
-    return out
-
-def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the nth path in the model.
-    '''
-    # Support the following patterns:
-    patterns: List[Tuple[str, str]] = [
-        # - x.00.pth, x.01.pth, etc.
-        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
-        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
-        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
-        # x.bin, x.bin.1, etc.
-        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
-    ]
-    for regex, replacement in patterns:
-        if re.search(regex, path.name):
-            new_path = path.with_name(re.sub(regex, replacement, path.name))
-            if new_path.exists():
-                return new_path
-    return None
-
-
-def find_multifile_paths(path: Path) -> List[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the whole list of paths in the model.
-    '''
-    ret: List[Path] = []
-    for i in itertools.count():
-        nth_path = nth_multifile_path(path, i)
-        if nth_path is None:
-            break
-        ret.append(nth_path)
-    if not ret:
-        # No matches.  This should only happen if the file was named, e.g.,
-        # foo.0, and there was no file named foo.  Oh well, try to process it
-        # as a single file.
-        return [path]
-    return ret
-
-
-def load_some_model(path: Path) -> ModelPlus:
-    '''Load a model of any supported format.'''
-    # Be extra-friendly and accept either a file or a directory:
-    if path.is_dir():
-        # Check if it's a set of safetensors files first
-        files = list(path.glob("model-00001-of-*.safetensors"))
-        if not files:
-            # Try the PyTorch patterns too, with lower priority
-            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
-            files = [file for glob in globs for file in path.glob(glob)]
-        if not files:
-            raise Exception(f"Can't find model in directory {path}")
-        if len(files) > 1:
-            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
-        path = files[0]
-
-    paths = find_multifile_paths(path)
-    models_plus: List[ModelPlus] = []
-    for path in paths:
-        print(f"Loading model file {path}")
-        models_plus.append(lazy_load_file(path))
-
-    model_plus = merge_multifile_models(models_plus)
-    return model_plus
-
-
-def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
-    # Be extra-friendly and accept either a file or a directory.  Also, if it's
-    # a directory, it might be the model directory, and tokenizer.model might
-    # be in the parent of that.
-    if path.is_dir():
-        vocab_file = "tokenizer.model"
-        if vocabtype == 'bpe':
-            vocab_file = "vocab.json"
-        path2 = path / vocab_file
-        # Use `.parent` instead of /.. to handle the symlink case better.
-        path3 = path.parent / vocab_file
-        if path2.exists():
-            path = path2
-        elif path3.exists():
-            path = path3
-        else:
-            raise FileNotFoundError(
-                f"Could not find tokenizer.model in {path} or its parent; "
-                "if it's in another directory, pass the directory as --vocab-dir")
-
-    print(f"Loading vocab file '{path}', type '{vocabtype}'")
-
-    added_tokens_path = path.parent / "added_tokens.json"
-    if vocabtype == "bpe":
-        return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
-    elif vocabtype == "spm":
-        return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None)
-    else:
-        raise ValueError(f"Unsupported vocabulary type {vocabtype}")
-
-
-def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
-    namestr = {
-        GGMLFileType.AllF32:    "f32",
-        GGMLFileType.MostlyF16: "f16",
-    }[file_type]
-    ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
-    if ret in model_paths:
-        sys.stderr.write(
-            f"Error: Default output path ({ret}) would overwrite the input. "
-            "Please explicitly specify a path using --outfile.\n")
-        sys.exit(1)
-    return ret
-
-
-def do_dump_model(model_plus: ModelPlus) -> None:
-    print(f"model_plus.paths = {model_plus.paths!r}")
-    print(f"model_plus.format = {model_plus.format!r}")
-    print(f"model_plus.vocab = {model_plus.vocab!r}")
-    for name, lazy_tensor in model_plus.model.items():
-        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
-
-
-def main(args_in: Optional[List[str]] = None) -> None:
-    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
-    parser.add_argument("--dump",        action="store_true",    help="don't convert, just show what's in the model")
-    parser.add_argument("--dump-single", action="store_true",    help="don't convert, just show what's in a single model file")
-    parser.add_argument("--vocab-only",  action="store_true",    help="extract only the vocab")
-    parser.add_argument("--outtype",     choices=["f32", "f16"], help="output format (default: based on input)")
-    parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
-    parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
-    parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)", default="spm")
-    parser.add_argument("--ctx",         type=int,               help="model training context (default: based on input)")
-    args = parser.parse_args(args_in)
-
-    if args.dump_single:
-        model_plus = lazy_load_file(args.model)
-        do_dump_model(model_plus)
-
-    model_plus = load_some_model(args.model)
-
-    params = Params.load(model_plus)
-    if params.n_ctx == -1:
-        if args.ctx is None:
-            raise Exception("The model doesn't have a context size, and you didn't specify one with --ctx\n"
-                            "Please specify one with --ctx:\n"
-                            " - LLaMA v1: --ctx 2048\n"
-                            " - LLaMA v2: --ctx 4096\n")
-        params.n_ctx = args.ctx
-
-    print(f"params = {params}")
-
-    vocab: Vocab
-    if args.vocab_only:
-        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
-        assert args.outfile, "need --outfile if using --vocab-only"
-        outfile = args.outfile
-        OutputFile.write_vocab_only(outfile, params, vocab)
-        print(f"Wrote {outfile}")
-    else:
-        if args.dump:
-            do_dump_model(model_plus)
-            return
-
-        if model_plus.vocab is not None and args.vocab_dir is None:
-            vocab = model_plus.vocab
-        else:
-            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
-            vocab = load_vocab(vocab_dir, args.vocabtype)
-
-        model       = model_plus.model
-        model       = convert_model_names(model, params)
-        output_type = pick_output_type(model, args.outtype)
-        model       = convert_to_output_type(model, output_type)
-        outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
-
-        OutputFile.write_all(outfile, params, model, vocab)
-        print(f"Wrote {outfile}")
-
-
-if __name__ == '__main__':
-    main()
diff --git a/convert.py b/convert.py
old mode 100644
new mode 100755
index d9d0c0b38215e..b243356f0db28
--- a/convert.py
+++ b/convert.py
@@ -1,4 +1,6 @@
 #!/usr/bin/env python
+
+import gguf
 import argparse
 import concurrent.futures
 import copy
@@ -16,13 +18,12 @@
 import struct
 import sys
 import zipfile
+import numpy as np
+
 from abc import ABCMeta, abstractmethod
 from dataclasses import dataclass
 from pathlib import Path
-from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List,
-                    Literal, Optional, Sequence, Tuple, TypeVar, Union)
-
-import numpy as np
+from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List, Literal, Optional, Sequence, Tuple, TypeVar, Union)
 from sentencepiece import SentencePieceProcessor  # type: ignore
 
 if TYPE_CHECKING:
@@ -33,57 +34,44 @@
 
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
+ARCH=gguf.MODEL_ARCH.LLAMA
+NAMES=gguf.MODEL_TENSOR_NAMES[ARCH]
+
+#
+# data types
+#
 
 @dataclass(frozen=True)
 class UnquantizedDataType:
     name: str
 
-
-DT_F16 = UnquantizedDataType('F16')
-DT_F32 = UnquantizedDataType('F32')
-DT_I32 = UnquantizedDataType('I32')
+DT_F16  = UnquantizedDataType('F16')
+DT_F32  = UnquantizedDataType('F32')
+DT_I32  = UnquantizedDataType('I32')
 DT_BF16 = UnquantizedDataType('BF16')
 
-
-@dataclass(frozen=True)
-class QuantizedDataType:
-    groupsize: int
-    have_addends: bool
-    have_g_idx: bool
-
-
-DT_Q4_0 = QuantizedDataType(groupsize=32, have_addends=False, have_g_idx=False)
-DT_Q4_1 = QuantizedDataType(groupsize=32, have_addends=True, have_g_idx=False)
-
-DataType = Union[UnquantizedDataType, QuantizedDataType]
-
-DATA_TYPE_TO_FTYPE: Dict[DataType, int] = {
-    DT_F32: 0,
-    DT_F16: 1,
-    DT_Q4_0: 2,
-    DT_Q4_1: 3,
-}
-
-FTYPE_TO_DATA_TYPE: Dict[int, DataType] = \
-    {ftype: dtype for (dtype, ftype) in DATA_TYPE_TO_FTYPE.items()}
+DataType = Union[UnquantizedDataType]
 
 DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
     DT_BF16: np.dtype(np.uint16),
-    DT_F16: np.dtype(np.float16),
-    DT_F32: np.dtype(np.float32),
-    DT_I32: np.dtype(np.int32),
+    DT_F16:  np.dtype(np.float16),
+    DT_F32:  np.dtype(np.float32),
+    DT_I32:  np.dtype(np.int32),
 }
 
 NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
     {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
 
+SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
+    'BF16': DT_BF16,
+    'F16': DT_F16,
+    'F32': DT_F32,
+    'I32': DT_I32,
+}
 
 class GGMLFileType(enum.Enum):
-    AllF32 = 0
+    AllF32    = 0
     MostlyF16 = 1  # except 1d tensors
-    MostlyQ4_0 = 2  # except 1d tensors
-    MostlyQ4_1 = 3  # except 1d tensors
-    PerLayerIsQ4_1 = 4  # but tok_embeddings.weight and output.weight are F16
 
     def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
         if len(tensor.shape) == 1:
@@ -93,60 +81,34 @@ def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
             return DT_F32
         elif self == GGMLFileType.MostlyF16:
             return DT_F16
-        elif self == GGMLFileType.MostlyQ4_0:
-            return DT_Q4_0
-        elif self == GGMLFileType.MostlyQ4_1:
-            return DT_Q4_1
-        elif self == GGMLFileType.PerLayerIsQ4_1:
-            if name in ('output.weight', 'tok_embeddings.weight'):
-                return DT_F16
-            else:
-                return DT_Q4_1
         else:
             raise ValueError(self)
 
 
-def make_tensors_list() -> List[str]:
-    ret = [
-        'tok_embeddings.weight',
-        'norm.weight',
-        'output.weight',
-    ]
-    for i in range(80):  # maximum number of layer
-        ret += [
-            f'layers.{i}.attention.wq.weight',
-            f'layers.{i}.attention.wk.weight',
-            f'layers.{i}.attention.wv.weight',
-            f'layers.{i}.attention.wo.weight',
-            f'layers.{i}.attention_norm.weight',
-            f'layers.{i}.feed_forward.w1.weight',
-            f'layers.{i}.feed_forward.w2.weight',
-            f'layers.{i}.feed_forward.w3.weight',
-            f'layers.{i}.ffn_norm.weight',
-        ]
-    return ret
-
-
-TENSORS_LIST = make_tensors_list()
-TENSORS_SET = set(TENSORS_LIST)
-
-
-def find_n_mult(n_ff: int, n_embd: int) -> int:
-    # hardcoded magic range
-    for n_mult in range(8192, 1, -1):
-        calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
-        if calc_ff == n_ff:
-            return n_mult
-    raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
+#
+# hparams loading
+#
 
 @dataclass
 class Params:
-    n_vocab:   int
-    n_embd:    int
-    n_mult:    int
-    n_head:    int
-    n_layer:   int
-    n_kv_head: Optional[int]  # This parameter is only used for Llama 2
+    n_vocab:    int
+    n_embd:     int
+    n_mult:     int
+    n_layer:    int
+    n_ctx:      int
+    n_ff:       int
+    n_head:     int
+    n_head_kv:  int
+    f_norm_eps: float
+
+    @staticmethod
+    def find_n_mult(n_ff: int, n_embd: int) -> int:
+        # hardcoded magic range
+        for n_mult in range(8192, 1, -1):
+            calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
+            if calc_ff == n_ff:
+                return n_mult
+        raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
 
     @staticmethod
     def guessed(model: 'LazyModel') -> 'Params':
@@ -165,37 +127,57 @@ def guessed(model: 'LazyModel') -> 'Params':
             raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
                             "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
 
-        n_head=n_embd // 128 # guessed
+        n_head = n_embd // 128 # guessed
+        n_mult = 256           # guessed
+
+        # TODO: verify this
+        n_ff = int(2 * (4 * n_embd) / 3)
+        n_ff = n_mult * ((n_ff + n_mult - 1) // n_mult)
 
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = 256,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = -1,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head,
+            f_norm_eps = 1e-5,
         )
 
     @staticmethod
     def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab = config["vocab_size"];
-        n_embd  = config["hidden_size"];
-        n_head  = config["num_attention_heads"];
-        n_layer = config["num_hidden_layers"];
-        n_ff    = config["intermediate_size"];
-        n_kv_head = config.get("num_key_value_heads")
+        n_vocab    = config["vocab_size"];
+        n_embd     = config["hidden_size"];
+        n_layer    = config["num_hidden_layers"];
+        n_ff       = config["intermediate_size"];
+        n_head     = config["num_attention_heads"];
+        n_head_kv  = config["num_key_value_heads"];
+        f_norm_eps = config["rms_norm_eps"];
 
-        n_mult = find_n_mult(n_ff, n_embd);
+        n_mult = Params.find_n_mult(n_ff, n_embd);
+
+        if "max_sequence_length" in config:
+            n_ctx = config["max_sequence_length"]
+        elif "max_position_embeddings" in config:
+            n_ctx = config["max_position_embeddings"]
+        else:
+            raise Exception("failed to guess 'n_ctx'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
 
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = n_kv_head,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
         )
 
     # LLaMA v2 70B params.json
@@ -204,22 +186,32 @@ def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
     def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab   = config["vocab_size"];
-        n_embd    = config["dim"];
-        n_head    = config["n_heads"];
-        n_layer   = config["n_layers"];
-        n_mult    = config["multiple_of"];
+        n_vocab    = config["vocab_size"];
+        n_embd     = config["dim"];
+        n_layer    = config["n_layers"];
+        n_mult     = config["multiple_of"];
+        n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
+        n_ff       = -1;
+        n_head     = config["n_heads"];
+        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head;
+        f_norm_eps = config["norm_eps"];
 
         if n_vocab == -1:
             n_vocab = model["tok_embeddings.weight"].shape[0]
 
+        if n_ff == -1:
+            n_ff = model["layers.0.feed_forward.w1.weight"].shape[0]
+
         return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
         )
 
     @staticmethod
@@ -234,10 +226,13 @@ def load(model_plus: 'ModelPlus') -> 'Params':
         else:
             params = Params.guessed(model_plus.model)
 
-        print(f'params: n_vocab:{params.n_vocab} n_embd:{params.n_embd} n_mult:{params.n_mult} n_head:{params.n_head} n_layer:{params.n_layer}')
         return params
 
 
+#
+# vocab
+#
+
 class BpeVocab:
     def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
         self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
@@ -291,9 +286,10 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) ->
             added_tokens = {}
         vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
         expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids = sorted(added_tokens.values())
+        actual_ids   = sorted(added_tokens.values())
         if expected_ids != actual_ids:
             raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+
         items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
         self.added_tokens_list = [text for (text, idx) in items]
         self.vocab_size_base: int = vocab_size
@@ -321,76 +317,29 @@ def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
     def __repr__(self) -> str:
         return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
 
+Vocab = Union[BpeVocab, SentencePieceVocab]
 
-class GGMLVocab:
-    def __init__(self, tokens: List[Tuple[bytes, float]]):
-        self.tokens = tokens
-        self.vocab_size = len(tokens)
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        return self.tokens
-
-    def __repr__(self) -> str:
-        return f"<GGMLVocab with {self.vocab_size} tokens>"
-
-
-Vocab = Union[BpeVocab, SentencePieceVocab, GGMLVocab]
 
+#
+# data loading
+# TODO: reuse (probably move to gguf.py?)
+#
 
-def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
-    if n_kv_head is not None and n_head != n_kv_head:
-        n_head //= n_kv_head
+def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
+    if n_head_kv is not None and n_head != n_head_kv:
+        n_head //= n_head_kv
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
                 .swapaxes(1, 2)
                 .reshape(weights.shape))
 
 
-def dequantize_q4(qvalues_pack32: NDArray, scales: NDArray, addends: Optional[NDArray], g_idx: Optional[NDArray]) -> NDArray:
-    # First reinterpret each row from a list of int32s containing 8 values each
-    # to a list of uint8s containing 2 values each.
-    qvalues_pack8 = qvalues_pack32.view(np.uint8)
-
-    # Then split out the two values per int8 (which requires an actual
-    # conversion because numpy doesn't natively support int4s).
-    qvalues = np.zeros([qvalues_pack8.shape[0], qvalues_pack8.shape[1] * 2], dtype=np.uint8)
-    qvalues[:, 0::2] = qvalues_pack8 & 0xf
-    qvalues[:, 1::2] = qvalues_pack8 >> 4
-
-    assert addends is None or addends.shape == scales.shape
-    assert qvalues.shape[0] == scales.shape[0]
-    assert qvalues.shape[1] % scales.shape[1] == 0
-    if g_idx is None:
-        repeat_count = qvalues.shape[1] // scales.shape[1]
-        scales = scales[:, :, np.newaxis]
-        if addends is not None:
-            addends = addends[:, :, np.newaxis]
-        # Reshape so that the below computation broadcasts over scales and addends:
-        qvalues.shape = (qvalues.shape[0], scales.shape[1], int(repeat_count))
-    else:
-        # In this case the scale and addend is selected for each column by g_idx:
-        assert addends is not None
-        scales = scales[:, g_idx]
-        addends = addends[:, g_idx]
-    if addends is None:
-        # Q4_0
-        qvalues = qvalues.view(np.int8)
-        qvalues -= 8
-    # And do the actual 'value = scale * qvalue + addend' computation.
-    values = scales * qvalues
-    if addends is not None:
-        values += addends
-    if g_idx is None:
-        values.shape = (values.shape[0], values.shape[1] * values.shape[2])
-    return values
-
-
 class Tensor(metaclass=ABCMeta):
     data_type: DataType
 
     @abstractmethod
     def astype(self, data_type: DataType) -> 'Tensor': ...
     @abstractmethod
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'Tensor': ...
+    def permute(self, n_head: int, n_head_kv: int) -> 'Tensor': ...
     @abstractmethod
     def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
     @abstractmethod
@@ -428,8 +377,8 @@ def part(self, n_part: int) -> 'UnquantizedTensor':
         r = self.ndarray.shape[0] // 3
         return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
 
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head, n_kv_head))
+    def permute(self, n_head: int, n_head_kv: int) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head, n_head_kv))
 
 
 def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
@@ -448,183 +397,25 @@ def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, conv
     return tensor.ndarray
 
 
-class GGMLQuantizedTensor(Tensor):
-    data_type: QuantizedDataType
-
-    def __init__(self, ndarray: NDArray, shape: List[int], data_type: DataType) -> None:
-        rows, columns = shape
-        assert data_type in (DT_Q4_1, DT_Q4_0)  # for now
-        assert isinstance(data_type, QuantizedDataType)  # redundant, but mypy complains without this
-        assert columns % data_type.groupsize == 0
-        words_in_block = 6 if data_type == DT_Q4_1 else 5
-        self.ndarray = ndarray.view(dtype=np.uint32).reshape((rows, columns // data_type.groupsize, words_in_block))
-        self.shape = shape[:]
-        self.data_type = data_type
-
-    def astype(self, data_type: DataType) -> Tensor:
-        if data_type == self.data_type:
-            return self
-        scales = self.ndarray[:, :, 0].view(np.float32)
-        if self.data_type.have_addends:
-            addends = self.ndarray[:, :, 1].view(np.float32)
-        else:
-            addends = None
-        qweights = self.ndarray[:, :, -4:].reshape([self.shape[0], self.shape[1] // 8])
-
-        dq = dequantize_q4(qweights, scales, addends, g_idx=None)
-        return UnquantizedTensor(dq).astype(data_type)
-
-    def to_ggml(self) -> 'GGMLQuantizedTensor':
-        return self
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'GGMLQuantizedTensor':
-        return GGMLQuantizedTensor(permute(self.ndarray, n_head, n_kv_head), self.shape, self.data_type)
-
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
-
-    def part(self, n_part: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
-
-GGMLCompatibleTensor = Union[UnquantizedTensor, GGMLQuantizedTensor]
+GGMLCompatibleTensor = Union[UnquantizedTensor]
 
 
 class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int, n_kv_head: Optional[int] = None) -> None:
+    def __init__(self, base: Tensor, n_head: int, n_head_kv: int) -> None:
         self.base = base
         self.n_head = n_head
-        self.n_kv_head = n_kv_head
         self.data_type = self.base.data_type
 
     def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head, self.n_kv_head)
+        return self.base.astype(data_type).permute(self.n_head, self.n_head_kv)
 
     def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head, self.n_kv_head)
+        return self.base.to_ggml().permute(self.n_head, self.n_head_kv)
 
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
+    def permute(self, n_head: int, n_head_kv: int) -> Tensor:
         raise Exception("shouldn't permute twice")
 
 
-class GPTQForLLaMaQuantizedTensor(Tensor):
-    def __init__(self, model: 'LazyModel', namebase: str) -> None:
-        qweight = load_unquantized(model[f"{namebase}.qweight"], np.int32)
-        scales = load_unquantized(model[f"{namebase}.scales"], np.float32, convert=True)
-
-        bias = model.get(f"{namebase}.bias")
-        if bias is not None:
-            # Q4_1 does not support bias; good thing the bias is always all zeros.
-            assert not np.any(load_unquantized(bias))
-
-        if f"{namebase}.zeros" in model:
-            zeros = load_unquantized(model[f"{namebase}.zeros"], np.float32)
-        else:
-            qzeros = load_unquantized(model[f"{namebase}.qzeros"], np.int32)
-            assert qzeros.dtype == np.int32
-            zeros = dequantize_q4(qzeros, scales, scales, g_idx=None)
-            assert zeros.dtype == np.float32
-
-        assert zeros.shape == scales.shape
-
-        # Output is transposed compared to the input, and addends have their sign flipped.
-        # Scales and zeros similarly must be transposed but only for newer
-        # versions of GPTQ-for-LLaMa; the older versions can be identified by
-        # having shape (n_embd, 1).
-        qweight = qweight.T
-        if scales.shape[1] != 1:
-            scales = scales.T
-            zeros = zeros.T
-
-        # Output also has signs flipped for the addends.
-        self.qweight = qweight
-        self.scales = scales
-        self.addends = -zeros
-
-        self.g_idx: Optional[NDArray]
-        if f"{namebase}.g_idx" in model:
-            self.g_idx = load_unquantized(model[f"{namebase}.g_idx"], np.int32)
-            assert self.g_idx.shape == (qweight.shape[1] * 8,)
-        else:
-            self.g_idx = None
-
-        self.shape = [self.qweight.shape[0], self.qweight.shape[1] * 8]
-        self.data_type = QuantizedDataType(groupsize=self.groupsize(), have_addends=True,
-                                           have_g_idx=(self.g_idx is not None))
-
-    def inspect(self, row: int, col: int) -> None:
-        '''For debugging.'''
-        qweight = (self.qweight[row, col // 8] >> (4 * (col & 7))) & 0xf
-        if self.g_idx is not None:
-            group = self.g_idx[col]
-        else:
-            group = int(col // self.groupsize())
-        scale = self.scales[row, group]
-        addend = self.addends[row, group]
-        with np.printoptions(precision=None, suppress=True):
-            print(f'scale:{scale} addend:{addend} qweight:{qweight}')
-            print('possible values:', np.arange(16) * scale + addend)
-            print('actual value:', qweight * scale + addend)
-
-    def astype(self, data_type: DataType) -> Tensor:
-        if isinstance(data_type, QuantizedDataType):
-            assert self.g_idx is None and data_type.have_addends is True and data_type.have_g_idx is False
-            return self.regroup(data_type.groupsize)
-
-        dequantized = dequantize_q4(np.ascontiguousarray(self.qweight), self.scales, self.addends, self.g_idx)
-        return UnquantizedTensor(dequantized).astype(data_type)
-
-    def groupsize(self) -> int:
-        assert self.addends.shape == self.scales.shape
-        assert self.shape[1] % self.scales.shape[1] == 0
-        return self.shape[1] // self.scales.shape[1]
-
-    def regroup(self, new_groupsize: int = 32) -> 'GPTQForLLaMaQuantizedTensor':
-        # Old versions of GPTQ-for-LLaMa shared scales and addends between all the
-        # columns in a row.  Newer versions share them between every set of N
-        # columns in a row, where N is the `groupsize` parameter, usually 128.  The
-        # output format shares them between every set of 32 columns.  To handle
-        # this, duplicate scales and addends for every smaller group.
-        # (In the above, 'row' and 'column' are in the sense of the output.)
-        assert self.g_idx is None
-        old_groupsize = self.groupsize()
-        assert old_groupsize >= new_groupsize and old_groupsize % new_groupsize == 0, old_groupsize
-        ret = copy.copy(self)
-        ret.addends = self.addends.repeat(old_groupsize // new_groupsize, axis=1)
-        ret.scales = self.scales.repeat(old_groupsize // new_groupsize, axis=1)
-        ret.data_type = QuantizedDataType(groupsize=new_groupsize, have_addends=True, have_g_idx=False)
-        return ret
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
-        return DeferredPermutedTensor(self, n_head, n_kv_head)
-
-    def to_ggml(self) -> GGMLQuantizedTensor:
-        # The output format looks like this:
-        # For each row:
-        #   For each group of 32 columns:
-        #     - addend (float32, 4 bytes)
-        #     - scale (float32, 4 bytes)
-        #     - weights (int4 * 32, 16 bytes)
-
-        if self.groupsize() != 32:
-            raise Exception("should have been regrouped before converting to ggml")
-
-        # Since the output format is mixed between integers and floats, we have
-        # to hackily view the floats as int32s just so numpy will let us
-        # concatenate them.
-        addends_view = self.addends.view(dtype=np.int32)[:, :, np.newaxis]
-        scales_view = self.scales.view(dtype=np.int32)[:, :, np.newaxis]
-
-        # Split into groups of 4 columns (i.e. 32 columns of quantized data):
-        grouped = self.qweight.reshape([self.qweight.shape[0], self.qweight.shape[1] // 4, 4])
-
-        # And concatenate:
-        grouped = np.concatenate([scales_view, addends_view, grouped], axis=2, casting='no')
-
-        return GGMLQuantizedTensor(grouped, self.shape, DT_Q4_1)
-
-
 @dataclass
 class LazyTensor:
     _load: Callable[[], Tensor]
@@ -647,17 +438,6 @@ def load() -> Tensor:
     def validate_conversion_to(self, data_type: DataType) -> None:
         if data_type == self.data_type:
             return
-        if isinstance(data_type, QuantizedDataType):
-            if not isinstance(self.data_type, QuantizedDataType):
-                raise Exception(f"Can't turn an unquantized tensor into a quantized type ({data_type})")
-            if self.data_type.have_g_idx:
-                sys.stderr.write(
-                    "Error: Input uses the newer GPTQ-for-LLaMa format (using g_idx), "
-                    "which is not yet natively supported by GGML. "
-                    "For now you can still convert this model by passing `--outtype f16` to dequantize, "
-                    "but that will result in a much larger output file for no quality benefit.\n")
-                sys.exit(1)
-            assert not data_type.have_g_idx and self.data_type.have_addends and data_type.have_addends
 
 
 LazyModel = Dict[str, LazyTensor]
@@ -728,10 +508,10 @@ def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
     return ModelPlus(model, paths, format, vocab)
 
 
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_kv_head: Optional[int] = None) -> LazyTensor:
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_head_kv: int) -> LazyTensor:
     def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head, n_kv_head)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_kv_head}) ' + lazy_tensor.description)
+        return lazy_tensor.load().permute(n_head, n_head_kv)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_head_kv}) ' + lazy_tensor.description)
 
 def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
     def load() -> Tensor:
@@ -747,66 +527,6 @@ def load() -> Tensor:
     s[0] = s[0] // 3
     return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
 
-def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
-    out: LazyModel = {}
-    out["tok_embeddings.weight"] = model["model.embed_tokens.weight"]
-    out["norm.weight"] = model["model.norm.weight"]
-    out["output.weight"] = model["lm_head.weight"]
-
-    for i in itertools.count():
-        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_kv_head)
-            out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
-        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head)
-            out[f"layers.{i}.attention.wv.weight"] = part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
-        else:
-            break
-
-        out[f"layers.{i}.attention.wo.weight"] = model[f"model.layers.{i}.self_attn.o_proj.weight"]
-
-        out[f"layers.{i}.feed_forward.w1.weight"] = model[f"model.layers.{i}.mlp.gate_proj.weight"]
-        out[f"layers.{i}.feed_forward.w2.weight"] = model[f"model.layers.{i}.mlp.down_proj.weight"]
-        out[f"layers.{i}.feed_forward.w3.weight"] = model[f"model.layers.{i}.mlp.up_proj.weight"]
-
-        out[f"layers.{i}.attention_norm.weight"] = model[f"model.layers.{i}.input_layernorm.weight"]
-        out[f"layers.{i}.ffn_norm.weight"] = model[f"model.layers.{i}.post_attention_layernorm.weight"]
-    return out
-
-
-def handle_quantization(model: LazyModel) -> LazyModel:
-    '''Convert a model with entries for 'foo.qweight', 'foo.scales', etc.
-    (which resolve to UnquantizedTensors with the raw data) to one with entries
-    for 'foo.weight' (which resolve to QuantizedTensors).
-    '''
-    def convert(name: str) -> Tuple[str, LazyTensor]:
-        if name.endswith(".qweight"):
-            namebase = name.rsplit('.', 1)[0]
-            orig_name = namebase + ".weight"
-
-            lazy_tensor = model[name]
-            assert len(lazy_tensor.shape) == 2
-            real_shape = [lazy_tensor.shape[1], lazy_tensor.shape[0] * 8]
-
-            # Calculate type.  This replicates the logic in
-            # GPTQForLLaMaQuantizedTensor (which is executed when the modelis
-            # actually loaded).
-            lazy_scales = model[f"{namebase}.scales"]
-            scales_width = 1 if lazy_scales.shape[1] == 1 else lazy_scales.shape[0]
-            assert real_shape[1] % scales_width == 0
-            groupsize = real_shape[1] // scales_width
-            have_g_idx = f"{namebase}.g_idx" in model
-            data_type = QuantizedDataType(groupsize=groupsize, have_addends=True, have_g_idx=have_g_idx)
-
-            def load() -> Tensor:
-                return GPTQForLLaMaQuantizedTensor(model, namebase)
-
-            return (orig_name, LazyTensor(load, real_shape, data_type, '[quantized]'))
-        else:
-            return (name, model[name])
-    return dict(convert(name) for name in model)
 
 # Functionality that simulates `torch.load` but where individual tensors are
 # only loaded into memory on demand, not all at once.
@@ -900,14 +620,6 @@ def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
     return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
 
 
-SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
-    'BF16': DT_BF16,
-    'F16': DT_F16,
-    'F32': DT_F32,
-    'I32': DT_I32,
-}
-
-
 def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
     header_size, = struct.unpack('<Q', fp.read(8))
     header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
@@ -939,84 +651,6 @@ def must_read(fp: IO[bytes], length: int) -> bytes:
     return ret
 
 
-def lazy_load_ggml_file(fp: io.BufferedReader, path: Path) -> ModelPlus:
-    magic = must_read(fp, 4)[::-1]
-    if magic in (b'ggmf', b'ggjt'):
-        version, = struct.unpack("i", must_read(fp, 4))
-        assert version == 1
-    else:
-        assert magic == b'ggml'
-        version = None
-    n_vocab, n_embd, n_mult, n_head, n_layer, rot, file_type = struct.unpack('<7i', must_read(fp, 28))
-
-    tokens: List[Tuple[bytes, float]] = []
-    for i in range(n_vocab):
-        if i == 32000:
-            # HACK: GPT4All messed with the format without changing the magic
-            # number.  Specifically, they changed the vocab section to contain
-            # `n_vocab - 1` tokens instead of `n_vocab` (i.e. omitting the
-            # extra pad token).  Try to detect if we're reading a file like
-            # this.
-            orig_pos = fp.tell()
-            fp.seek(20, io.SEEK_CUR)
-            is_gpt4all = fp.read(21) == b'tok_embeddings.weight'
-            fp.seek(orig_pos)
-            if is_gpt4all:
-                break
-
-        length, = struct.unpack("i", must_read(fp, 4))
-        text = must_read(fp, length)
-        if magic != b'ggml':
-            score, = struct.unpack("f", must_read(fp, 4))
-            tokens.append((text, score))
-    vocab = GGMLVocab(tokens) if magic != b'ggml' else None
-
-    model: LazyModel = {}
-    # Use mmap for the actual data to avoid race conditions with the file offset.
-    off = fp.raw.tell()
-    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
-    fp.raw.seek(off)  # needed on Windows
-
-    def read_tensor() -> None:  # this is a function so that variables captured in `load` don't change
-        shape_len, name_len, ftype = struct.unpack("iii", must_read(fp, 12))
-        assert 0 <= shape_len <= 3
-        shape: List[int] = list(struct.unpack(f"{shape_len}i", must_read(fp, 4 * shape_len)))
-        shape = shape[::-1]
-        name = must_read(fp, name_len).decode('utf-8')
-        data_type = FTYPE_TO_DATA_TYPE[ftype]
-
-        if magic == b'ggjt':
-            fp.seek((fp.tell() + 31) & -32)
-
-        if data_type == DT_Q4_1:
-            # See GPTQForLLaMaQuantizedTensor.ggml_ndarray()
-            size = 24 * (shape[1] // 32) * shape[0]
-        elif data_type == DT_Q4_0:
-            size = 20 * (shape[1] // 32) * shape[0]
-        else:
-            numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
-            elm_count = math.prod(shape)
-            size = elm_count * numpy_dtype.itemsize
-        offset = fp.tell()
-        buf = mapped[offset:offset+size]
-        fp.seek(size, io.SEEK_CUR)
-
-        def load() -> Tensor:
-            if isinstance(data_type, QuantizedDataType):
-                ndarray = np.frombuffer(buf, dtype=np.uint32)
-                return GGMLQuantizedTensor(ndarray, shape, data_type)
-            else:
-                return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
-        description = f'ggml offset={offset} type={data_type} path={path}'
-        model[name] = LazyTensor(load, shape, data_type, description)
-
-    while fp.read(1) != b'':
-        fp.seek(-1, io.SEEK_CUR)
-        read_tensor()
-
-    return ModelPlus(model=model, paths=[path], format='ggml', vocab=vocab)
-
-
 @functools.lru_cache(maxsize=None)
 def lazy_load_file(path: Path) -> ModelPlus:
     fp = open(path, 'rb')
@@ -1025,9 +659,6 @@ def lazy_load_file(path: Path) -> ModelPlus:
     if first8[:2] == b'PK':
         # A zip file, i.e. PyTorch format
         return lazy_load_torch_file(fp, path)
-    elif first8[2:4] == b'gg':
-        # GGML format
-        return lazy_load_ggml_file(fp, path)
     elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
         # Probably safetensors
         return lazy_load_safetensors_file(fp, path)
@@ -1038,7 +669,6 @@ def lazy_load_file(path: Path) -> ModelPlus:
 In = TypeVar('In')
 Out = TypeVar('Out')
 
-
 def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
     '''Parallel map, but with backpressure.  If the caller doesn't call `next`
     fast enough, this will stop calling `func` at some point rather than
@@ -1058,7 +688,6 @@ def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], conc
 
 def check_vocab_size(params: Params, vocab: Vocab) -> None:
     if params.n_vocab != vocab.vocab_size:
-        # GGMLVocab comes from the same file as the model so shouldn't mismatch:
         assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
         if params.n_vocab == vocab.vocab_size_base:
             print("Ignoring added_tokens.json since model matches vocab size without it.")
@@ -1076,98 +705,133 @@ def check_vocab_size(params: Params, vocab: Vocab) -> None:
 
 class OutputFile:
     def __init__(self, fname_out: Path) -> None:
-        self.fout = open(fname_out, "wb")
-
-    def write_file_header(self, params: Params, file_type: GGMLFileType) -> None:
-        self.fout.write(b"ggjt"[::-1])  # magic
-        values = [
-            1,  # file version
-            params.n_vocab,
-            params.n_embd,
-            params.n_mult,
-            params.n_head,
-            params.n_layer,
-            params.n_embd // params.n_head,  # rot (obsolete)
-            file_type.value,
-        ]
-        self.fout.write(struct.pack("i" * len(values), *values))
-
-    def write_tensor_header(self, name: str, shape: Sequence[int], data_type: DataType) -> None:
-        sname = name.encode('utf-8')
-        self.fout.write(struct.pack("iii", len(shape), len(sname), DATA_TYPE_TO_FTYPE[data_type]))
-        self.fout.write(struct.pack("i" * len(shape), *shape[::-1]))
-        self.fout.write(sname)
-        self.fout.seek((self.fout.tell() + 31) & -32)
-
-    def write_vocab(self, vocab: Vocab) -> None:
+        self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
+
+    def add_meta_arch(self, params: Params) -> None:
+        self.gguf.add_context_length      (params.n_ctx)
+        self.gguf.add_embedding_length    (params.n_embd)
+        self.gguf.add_block_count         (params.n_layer)
+        self.gguf.add_feed_forward_length (params.n_ff)
+        self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
+        self.gguf.add_head_count          (params.n_head)
+        self.gguf.add_head_count_kv       (params.n_head_kv)
+        self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
+
+    def add_meta_vocab(self, vocab: Vocab) -> None:
+        tokens = []
+        scores = []
         for text, score in vocab.all_tokens():
-            self.fout.write(struct.pack("i", len(text)))
-            self.fout.write(text)
-            self.fout.write(struct.pack("f", score))
+            tokens.append(text)
+            scores.append(score)
+
+        self.gguf.add_tokenizer_model("llama")
+        self.gguf.add_token_list(tokens)
+        self.gguf.add_token_scores(scores)
+        #self.gguf.add_token_types(toktypes) # TODO: add this
+
+        # TODO: added / special tokens
+
+    def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
+        n_elements = 1
+        for dim in tensor.shape:
+            n_elements *= dim
+        data_type = DATA_TYPE_TO_NUMPY[tensor.data_type]
+        data_nbytes = n_elements * data_type.itemsize
+        self.gguf.add_tensor_info(name, tensor.shape, data_type, data_nbytes)
+
+    def write_meta(self) -> None:
+        self.gguf.write_header_to_file()
+        self.gguf.write_kv_data_to_file()
+
+    def write_tensor_info(self) -> None:
+        self.gguf.write_ti_data_to_file()
+
+    def close(self) -> None:
+        self.gguf.close()
 
     @staticmethod
-    def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
-        of = OutputFile(fname_out)
-        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0, n_kv_head=None)
+    def write_vocab_only(fname_out: Path, params: Params, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+
         of = OutputFile(fname_out)
-        of.write_file_header(params, file_type=GGMLFileType.AllF32)
-        of.write_vocab(vocab)
-        of.fout.close()
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+        of.write_meta()
+
+        of.close()
 
     @staticmethod
-    def write_all(fname_out: Path, params: Params, file_type: GGMLFileType, model: LazyModel, vocab: Vocab) -> None:
+    def write_all(fname_out: Path, params: Params, model: LazyModel, vocab: Vocab) -> None:
         check_vocab_size(params, vocab)
+
         of = OutputFile(fname_out)
-        of.write_file_header(params, file_type)
-        print("Writing vocab...")
-        of.write_vocab(vocab)
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+
+        # tensor info
+        for name, lazy_tensor in model.items():
+            of.add_tensor_info(name, lazy_tensor)
+
+        of.write_meta()
+        of.write_tensor_info()
 
         def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
             name, lazy_tensor = item
             return lazy_tensor.load().to_ggml().ndarray
 
+        # tensor data
         ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=8)
         for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
             size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
             padi = len(str(len(model)))
             print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
-            of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
-            ndarray.tofile(of.fout)
-        of.fout.close()
+            of.gguf.write_tensor_data(ndarray)
 
+        of.close()
 
 def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
-    wq_type = model["layers.0.attention.wq.weight"].data_type
-    if output_type_str == "f32" or (output_type_str is None and wq_type in (DT_F32, DT_BF16)):
+    wq_type = model[NAMES[gguf.MODEL_TENSOR.ATTN_Q].format(bid=0)+".weight"].data_type
+
+    if output_type_str == "f32" or (output_type_str is None and wq_type == DT_F32):
         return GGMLFileType.AllF32
-    if output_type_str == "f16" or (output_type_str is None and wq_type == DT_F16):
+    if output_type_str == "f16" or (output_type_str is None and wq_type in (DT_F16, DT_BF16)):
         return GGMLFileType.MostlyF16
-    if output_type_str == "q4_1" or (output_type_str is None and isinstance(wq_type, QuantizedDataType) and
-                                     wq_type.have_addends):
-        if isinstance(model["output.weight"].data_type, QuantizedDataType):
-            return GGMLFileType.MostlyQ4_1
-        else:
-            return GGMLFileType.PerLayerIsQ4_1
-    if output_type_str == "q4_0" or (output_type_str is None and isinstance(wq_type, QuantizedDataType)):
-        return GGMLFileType.MostlyQ4_0
-    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
-    raise Exception(f"Unexpected combination of types: {name_to_type}")
-
 
-def do_necessary_conversions(model: LazyModel, params: Params) -> LazyModel:
-    model = handle_quantization(model)
-
-    if "lm_head.weight" in model:
-        model = convert_transformers_to_orig(model, params)
-    model = filter_and_sort_tensors(model)
-
-    return model
+    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
 
+    raise Exception(f"Unexpected combination of types: {name_to_type}")
 
 def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
     return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
             for (name, tensor) in model.items()}
 
+def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
+    tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
+
+    out: LazyModel = {}
+    for name, lazy_tensor in model.items():
+        name_new = name
+
+        if name in tmap:
+            name_new = tmap[name]
+        elif name.endswith(".weight") and name[:-7] in tmap:
+            name_new = tmap[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tmap:
+            name_new = tmap[name[:-5]] + ".bias"
+        else:
+            raise Exception(f"Unexpected tensor name: {name}")
+
+        if gguf.should_skip_tensor(ARCH, params.n_layer, name_new):
+            print(f"skipping tensor {name_new}")
+        else:
+            print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
+            out[name_new] = lazy_tensor
+
+    return out
 
 def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
     '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
@@ -1218,11 +882,6 @@ def load_some_model(path: Path) -> ModelPlus:
             # Try the PyTorch patterns too, with lower priority
             globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
             files = [file for glob in globs for file in path.glob(glob)]
-        if not files:
-            # Try GGML too, but with lower priority, since if both a non-GGML
-            # model and a GGML model exist in the same directory, we assume the
-            # latter was converted from the former.
-            files = list(path.glob("ggml-model*.bin*"))
         if not files:
             raise Exception(f"Can't find model in directory {path}")
         if len(files) > 1:
@@ -1239,19 +898,14 @@ def load_some_model(path: Path) -> ModelPlus:
     return model_plus
 
 
-def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
-    return {name: model[name] for name in TENSORS_LIST if name in model}
-
-
 def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
-    print(f"vocabtype: {vocabtype}")
     # Be extra-friendly and accept either a file or a directory.  Also, if it's
     # a directory, it might be the model directory, and tokenizer.model might
     # be in the parent of that.
     if path.is_dir():
         vocab_file = "tokenizer.model"
         if vocabtype == 'bpe':
-          vocab_file = "vocab.json"
+            vocab_file = "vocab.json"
         path2 = path / vocab_file
         # Use `.parent` instead of /.. to handle the symlink case better.
         path3 = path.parent / vocab_file
@@ -1263,8 +917,10 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, Sentence
             raise FileNotFoundError(
                 f"Could not find tokenizer.model in {path} or its parent; "
                 "if it's in another directory, pass the directory as --vocab-dir")
+
+    print(f"Loading vocab file '{path}', type '{vocabtype}'")
+
     added_tokens_path = path.parent / "added_tokens.json"
-    print(f"Loading vocab file {path}")
     if vocabtype == "bpe":
         return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
     elif vocabtype == "spm":
@@ -1275,11 +931,8 @@ def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, Sentence
 
 def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
     namestr = {
-        GGMLFileType.AllF32: "f32",
+        GGMLFileType.AllF32:    "f32",
         GGMLFileType.MostlyF16: "f16",
-        GGMLFileType.MostlyQ4_0: "q4_0",
-        GGMLFileType.MostlyQ4_1: "q4_1",
-        GGMLFileType.PerLayerIsQ4_1: "q4_1",
     }[file_type]
     ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
     if ret in model_paths:
@@ -1300,44 +953,59 @@ def do_dump_model(model_plus: ModelPlus) -> None:
 
 def main(args_in: Optional[List[str]] = None) -> None:
     parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
-    parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
-    parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
-    parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
-    parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)")
-    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
-    parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
-    parser.add_argument("model", type=Path,
-                        help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
+    parser.add_argument("--dump",        action="store_true",    help="don't convert, just show what's in the model")
+    parser.add_argument("--dump-single", action="store_true",    help="don't convert, just show what's in a single model file")
+    parser.add_argument("--vocab-only",  action="store_true",    help="extract only the vocab")
+    parser.add_argument("--outtype",     choices=["f32", "f16"], help="output format (default: based on input)")
+    parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
+    parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
+    parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
+    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)", default="spm")
+    parser.add_argument("--ctx",         type=int,               help="model training context (default: based on input)")
     args = parser.parse_args(args_in)
 
-    vocab: Vocab
     if args.dump_single:
         model_plus = lazy_load_file(args.model)
         do_dump_model(model_plus)
-    elif args.vocab_only:
+
+    model_plus = load_some_model(args.model)
+
+    params = Params.load(model_plus)
+    if params.n_ctx == -1:
+        if args.ctx is None:
+            raise Exception("The model doesn't have a context size, and you didn't specify one with --ctx\n"
+                            "Please specify one with --ctx:\n"
+                            " - LLaMA v1: --ctx 2048\n"
+                            " - LLaMA v2: --ctx 4096\n")
+        params.n_ctx = args.ctx
+
+    print(f"params = {params}")
+
+    vocab: Vocab
+    if args.vocab_only:
         vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
         assert args.outfile, "need --outfile if using --vocab-only"
         outfile = args.outfile
-        OutputFile.write_vocab_only(outfile, vocab)
+        OutputFile.write_vocab_only(outfile, params, vocab)
         print(f"Wrote {outfile}")
     else:
-        model_plus = load_some_model(args.model)
         if args.dump:
             do_dump_model(model_plus)
             return
+
         if model_plus.vocab is not None and args.vocab_dir is None:
             vocab = model_plus.vocab
         else:
             vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
             vocab = load_vocab(vocab_dir, args.vocabtype)
-        params = Params.load(model_plus)
-        model = model_plus.model
-        model = do_necessary_conversions(model, params)
+
+        model       = model_plus.model
+        model       = convert_model_names(model, params)
         output_type = pick_output_type(model, args.outtype)
-        model = convert_to_output_type(model, output_type)
-        outfile = args.outfile or default_outfile(model_plus.paths, output_type)
-        OutputFile.write_all(outfile, params, output_type, model, vocab)
+        model       = convert_to_output_type(model, output_type)
+        outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
+
+        OutputFile.write_all(outfile, params, model, vocab)
         print(f"Wrote {outfile}")
 
 

From dd016cc246d01e58b433f2c8a1b0487fb6de62ac Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 17:23:16 +0300
Subject: [PATCH 170/242] Revert "ci : disable CI temporary to not waste
 energy"

This reverts commit 7e82d25f40386540c2c15226300ad998ecd871ea.
---
 .github/ISSUE_TEMPLATE/custom.md   | 185 +++++++++
 .github/workflows/build.yml        | 632 +++++++++++++++++++++++++++++
 .github/workflows/docker.yml       |  65 +++
 .github/workflows/editorconfig.yml |  17 +
 .github/workflows/tidy-post.yml    |  20 +
 .github/workflows/tidy-review.yml  |  23 ++
 6 files changed, 942 insertions(+)
 create mode 100644 .github/ISSUE_TEMPLATE/custom.md
 create mode 100644 .github/workflows/build.yml
 create mode 100644 .github/workflows/docker.yml
 create mode 100644 .github/workflows/editorconfig.yml
 create mode 100644 .github/workflows/tidy-post.yml
 create mode 100644 .github/workflows/tidy-review.yml

diff --git a/.github/ISSUE_TEMPLATE/custom.md b/.github/ISSUE_TEMPLATE/custom.md
new file mode 100644
index 0000000000000..8fd9553567780
--- /dev/null
+++ b/.github/ISSUE_TEMPLATE/custom.md
@@ -0,0 +1,185 @@
+---
+name: Issue and enhancement template
+about: Used to report issues and request enhancements for llama.cpp
+title: "[User] Insert summary of your issue or enhancement.."
+labels: ''
+assignees: ''
+
+---
+
+# Prerequisites
+
+Please answer the following questions for yourself before submitting an issue.
+
+- [ ] I am running the latest code. Development is very rapid so there are no tagged versions as of now.
+- [ ] I carefully followed the [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md).
+- [ ] I [searched using keywords relevant to my issue](https://docs.github.com/en/issues/tracking-your-work-with-issues/filtering-and-searching-issues-and-pull-requests) to make sure that I am creating a new issue that is not already open (or closed).
+- [ ] I reviewed the [Discussions](https://github.com/ggerganov/llama.cpp/discussions), and have a new bug or useful enhancement to share.
+
+# Expected Behavior
+
+Please provide a detailed written description of what you were trying to do, and what you expected `llama.cpp` to do.
+
+# Current Behavior
+
+Please provide a detailed written description of what `llama.cpp` did, instead.
+
+# Environment and Context
+
+Please provide detailed information about your computer setup. This is important in case the issue is not reproducible except for under certain specific conditions.
+
+* Physical (or virtual) hardware you are using, e.g. for Linux:
+
+`$ lscpu`
+
+* Operating System, e.g. for Linux:
+
+`$ uname -a`
+
+* SDK version, e.g. for Linux:
+
+```
+$ python3 --version
+$ make --version
+$ g++ --version
+```
+
+# Failure Information (for bugs)
+
+Please help provide information about the failure if this is a bug. If it is not a bug, please remove the rest of this template.
+
+# Steps to Reproduce
+
+Please provide detailed steps for reproducing the issue. We are not sitting in front of your screen, so the more detail the better.
+
+1. step 1
+2. step 2
+3. step 3
+4. etc.
+
+# Failure Logs
+
+Please include any relevant log snippets or files. If it works under one configuration but not under another, please provide logs for both configurations and their corresponding outputs so it is easy to see where behavior changes.
+
+Also, please try to **avoid using screenshots** if at all possible. Instead, copy/paste the console output and use [Github's markdown](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax) to cleanly format your logs for easy readability.
+
+Example environment info:
+```
+llama.cpp$ git log | head -1
+commit 2af23d30434a677c6416812eea52ccc0af65119c
+
+llama.cpp$ lscpu | egrep "AMD|Flags"
+Vendor ID:                       AuthenticAMD
+Model name:                      AMD Ryzen Threadripper 1950X 16-Core Processor
+Flags:                           fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid amd_dcm aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb hw_pstate ssbd ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf xsaveerptr arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov succor smca sme sev
+Virtualization:                  AMD-V
+
+llama.cpp$ python3 --version
+Python 3.10.9
+
+llama.cpp$ pip list | egrep "torch|numpy|sentencepiece"
+numpy                         1.24.2
+numpydoc                      1.5.0
+sentencepiece                 0.1.97
+torch                         1.13.1
+torchvision                   0.14.1
+
+llama.cpp$ make --version | head -1
+GNU Make 4.3
+
+$ md5sum ./models/65B/ggml-model-q4_0.bin
+dbdd682cce80e2d6e93cefc7449df487  ./models/65B/ggml-model-q4_0.bin
+```
+
+Example run with the Linux command [perf](https://www.brendangregg.com/perf.html)
+```
+llama.cpp$ perf stat ./main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p "Please close your issue when it has been answered."
+main: seed = 1679149377
+llama_model_load: loading model from './models/65B/ggml-model-q4_0.bin' - please wait ...
+llama_model_load: n_vocab = 32000
+llama_model_load: n_ctx   = 512
+llama_model_load: n_embd  = 8192
+llama_model_load: n_mult  = 256
+llama_model_load: n_head  = 64
+llama_model_load: n_layer = 80
+llama_model_load: n_rot   = 128
+llama_model_load: f16     = 2
+llama_model_load: n_ff    = 22016
+llama_model_load: n_parts = 8
+llama_model_load: ggml ctx size = 41477.73 MB
+llama_model_load: memory_size =  2560.00 MB, n_mem = 40960
+llama_model_load: loading model part 1/8 from './models/65B/ggml-model-q4_0.bin'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 2/8 from './models/65B/ggml-model-q4_0.bin.1'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 3/8 from './models/65B/ggml-model-q4_0.bin.2'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 4/8 from './models/65B/ggml-model-q4_0.bin.3'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 5/8 from './models/65B/ggml-model-q4_0.bin.4'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 6/8 from './models/65B/ggml-model-q4_0.bin.5'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 7/8 from './models/65B/ggml-model-q4_0.bin.6'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 8/8 from './models/65B/ggml-model-q4_0.bin.7'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+
+system_info: n_threads = 16 / 32 | AVX = 1 | AVX2 = 1 | AVX512 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | VSX = 0 |
+
+main: prompt: 'Please close your issue when it has been answered.'
+main: number of tokens in prompt = 11
+     1 -> ''
+ 12148 -> 'Please'
+  3802 -> ' close'
+   596 -> ' your'
+  2228 -> ' issue'
+   746 -> ' when'
+   372 -> ' it'
+   756 -> ' has'
+  1063 -> ' been'
+  7699 -> ' answered'
+ 29889 -> '.'
+
+sampling parameters: temp = 0.800000, top_k = 40, top_p = 0.950000, repeat_last_n = 64, repeat_penalty = 1.300000
+
+
+Please close your issue when it has been answered.
+@duncan-donut: I'm trying to figure out what kind of "support" you need for this script and why, exactly? Is there a question about how the code works that hasn't already been addressed in one or more comments below this ticket, or are we talking something else entirely like some sorta bugfixing job because your server setup is different from mine??
+I can understand if your site needs to be running smoothly and you need help with a fix of sorts but there should really be nothing wrong here that the code itself could not handle. And given that I'm getting reports about how it works perfectly well on some other servers, what exactly are we talking? A detailed report will do wonders in helping us get this resolved for ya quickly so please take your time and describe the issue(s) you see as clearly & concisely as possible!!
+@duncan-donut: I'm not sure if you have access to cPanel but you could try these instructions. It is worth a shot! Let me know how it goes (or what error message, exactly!) when/if ya give that code a go? [end of text]
+
+
+main: mem per token = 71159620 bytes
+main:     load time = 19309.95 ms
+main:   sample time =   168.62 ms
+main:  predict time = 223895.61 ms / 888.47 ms per token
+main:    total time = 246406.42 ms
+
+ Performance counter stats for './main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p Please close your issue when it has been answered.':
+
+        3636882.89 msec task-clock                #   14.677 CPUs utilized
+             13509      context-switches          #    3.714 /sec
+              2436      cpu-migrations            #    0.670 /sec
+          10476679      page-faults               #    2.881 K/sec
+    13133115082869      cycles                    #    3.611 GHz                      (16.77%)
+       29314462753      stalled-cycles-frontend   #    0.22% frontend cycles idle     (16.76%)
+    10294402631459      stalled-cycles-backend    #   78.39% backend cycles idle      (16.74%)
+    23479217109614      instructions              #    1.79  insn per cycle
+                                                  #    0.44  stalled cycles per insn  (16.76%)
+     2353072268027      branches                  #  647.002 M/sec                    (16.77%)
+        1998682780      branch-misses             #    0.08% of all branches          (16.76%)
+
+     247.802177522 seconds time elapsed
+
+    3618.573072000 seconds user
+      18.491698000 seconds sys
+```
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
new file mode 100644
index 0000000000000..84faad37ab95a
--- /dev/null
+++ b/.github/workflows/build.yml
@@ -0,0 +1,632 @@
+name: CI
+
+on:
+  workflow_dispatch: # allows manual triggering
+    inputs:
+      create_release:
+        description: 'Create new release'
+        required: true
+        type: boolean
+  push:
+    branches:
+      - master
+    paths: ['.github/workflows/**', '**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu']
+  pull_request:
+    types: [opened, synchronize, reopened]
+    paths: ['**/CMakeLists.txt', '**/Makefile', '**/*.h', '**/*.hpp', '**/*.c', '**/*.cpp', '**/*.cu']
+
+env:
+  BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
+  GGML_NLOOP: 3
+  GGML_NITER: 1
+  GGML_N_THREADS: 1
+
+jobs:
+  ubuntu-focal-make:
+    runs-on: ubuntu-20.04
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential gcc-8
+
+      - name: Build
+        id: make_build
+        run: |
+          CC=gcc-8 make
+
+  ubuntu-latest-cmake:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential
+
+      - name: Build
+        id: cmake_build
+        run: |
+          mkdir build
+          cd build
+          cmake ..
+          cmake --build . --config Release
+
+      - name: Test
+        id: cmake_test
+        run: |
+          cd build
+          ctest --verbose --timeout 900
+
+  ubuntu-latest-cmake-sanitizer:
+    runs-on: ubuntu-latest
+
+    continue-on-error: true
+
+    strategy:
+      matrix:
+        sanitizer: [ADDRESS, THREAD, UNDEFINED]
+        build_type: [Debug, Release]
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential
+
+      - name: Build
+        id: cmake_build
+        run: |
+          mkdir build
+          cd build
+          cmake .. -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+          cmake --build . --config ${{ matrix.build_type }}
+
+      - name: Test
+        id: cmake_test
+        run: |
+          cd build
+          ctest --verbose --timeout 900
+
+  ubuntu-latest-cmake-mpi:
+    runs-on: ubuntu-latest
+
+    continue-on-error: true
+
+    strategy:
+      matrix:
+        mpi_library: [mpich, libopenmpi-dev]
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential ${{ matrix.mpi_library }}
+
+      - name: Build
+        id: cmake_build
+        run: |
+          mkdir build
+          cd build
+          cmake -DLLAMA_MPI=ON ..
+          cmake --build . --config Release
+
+      - name: Test
+        id: cmake_test
+        run: |
+          cd build
+          ctest --verbose
+
+  macOS-latest-make:
+    runs-on: macos-latest
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        continue-on-error: true
+        run: |
+          brew update
+
+      - name: Build
+        id: make_build
+        run: |
+          make
+
+  macOS-latest-cmake:
+    runs-on: macos-latest
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Dependencies
+        id: depends
+        continue-on-error: true
+        run: |
+          brew update
+
+      - name: Build
+        id: cmake_build
+        run: |
+          sysctl -a
+          mkdir build
+          cd build
+          cmake -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF ..
+          cmake --build . --config Release
+
+      - name: Test
+        id: cmake_test
+        run: |
+          cd build
+          ctest --verbose --timeout 900
+
+  windows-latest-cmake:
+    runs-on: windows-latest
+
+    env:
+      OPENBLAS_VERSION: 0.3.23
+      OPENCL_VERSION: 2023.04.17
+      CLBLAST_VERSION: 1.6.0
+
+    strategy:
+      matrix:
+        include:
+          - build: 'noavx'
+            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF'
+          - build: 'avx2'
+            defines: '-DLLAMA_BUILD_SERVER=ON'
+          - build: 'avx'
+            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX2=OFF'
+          - build: 'avx512'
+            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUILD_SHARED_LIBS=ON'
+          - build: 'clblast'
+            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_CLBLAST=ON -DCMAKE_PREFIX_PATH="$env:RUNNER_TEMP/clblast"'
+          - build: 'openblas'
+            defines: '-DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - name: Download OpenCL SDK
+        id: get_opencl
+        if: ${{ matrix.build == 'clblast' }}
+        run: |
+          curl.exe -o $env:RUNNER_TEMP/opencl.zip -L "https://github.com/KhronosGroup/OpenCL-SDK/releases/download/v${env:OPENCL_VERSION}/OpenCL-SDK-v${env:OPENCL_VERSION}-Win-x64.zip"
+          mkdir $env:RUNNER_TEMP/opencl
+          tar.exe -xvf $env:RUNNER_TEMP/opencl.zip --strip-components=1 -C $env:RUNNER_TEMP/opencl
+
+      - name: Download CLBlast
+        id: get_clblast
+        if: ${{ matrix.build == 'clblast' }}
+        run: |
+          curl.exe -o $env:RUNNER_TEMP/clblast.7z -L "https://github.com/CNugteren/CLBlast/releases/download/${env:CLBLAST_VERSION}/CLBlast-${env:CLBLAST_VERSION}-windows-x64.7z"
+          curl.exe -o $env:RUNNER_TEMP/CLBlast.LICENSE.txt -L "https://github.com/CNugteren/CLBlast/raw/${env:CLBLAST_VERSION}/LICENSE"
+          7z x "-o${env:RUNNER_TEMP}" $env:RUNNER_TEMP/clblast.7z
+          rename-item $env:RUNNER_TEMP/CLBlast-${env:CLBLAST_VERSION}-windows-x64 clblast
+          foreach ($f in (gci -Recurse -Path "$env:RUNNER_TEMP/clblast" -Filter '*.cmake')) {
+            $txt = Get-Content -Path $f -Raw
+            $txt.Replace('C:/vcpkg/packages/opencl_x64-windows/', "$($env:RUNNER_TEMP.Replace('\','/'))/opencl/") | Set-Content -Path $f -Encoding UTF8
+          }
+
+      - name: Download OpenBLAS
+        id: get_openblas
+        if: ${{ matrix.build == 'openblas' }}
+        run: |
+          curl.exe -o $env:RUNNER_TEMP/openblas.zip -L "https://github.com/xianyi/OpenBLAS/releases/download/v${env:OPENBLAS_VERSION}/OpenBLAS-${env:OPENBLAS_VERSION}-x64.zip"
+          curl.exe -o $env:RUNNER_TEMP/OpenBLAS.LICENSE.txt -L "https://github.com/xianyi/OpenBLAS/raw/v${env:OPENBLAS_VERSION}/LICENSE"
+          mkdir $env:RUNNER_TEMP/openblas
+          tar.exe -xvf $env:RUNNER_TEMP/openblas.zip -C $env:RUNNER_TEMP/openblas
+          $vcdir = $(vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath)
+          $msvc = $(join-path $vcdir $('VC\Tools\MSVC\'+$(gc -raw $(join-path $vcdir 'VC\Auxiliary\Build\Microsoft.VCToolsVersion.default.txt')).Trim()))
+          $lib =  $(join-path $msvc 'bin\Hostx64\x64\lib.exe')
+          & $lib /machine:x64 "/def:${env:RUNNER_TEMP}/openblas/lib/libopenblas.def" "/out:${env:RUNNER_TEMP}/openblas/lib/openblas.lib" /name:openblas.dll
+
+      - name: Build
+        id: cmake_build
+        run: |
+          mkdir build
+          cd build
+          cmake .. ${{ matrix.defines }}
+          cmake --build . --config Release
+
+      - name: Add clblast.dll
+        id: add_clblast_dll
+        if: ${{ matrix.build == 'clblast' }}
+        run: |
+          cp $env:RUNNER_TEMP/clblast/lib/clblast.dll ./build/bin/Release
+          cp $env:RUNNER_TEMP/CLBlast.LICENSE.txt ./build/bin/Release/CLBlast-${env:CLBLAST_VERSION}.txt
+
+      - name: Add libopenblas.dll
+        id: add_libopenblas_dll
+        if: ${{ matrix.build == 'openblas' }}
+        run: |
+          cp $env:RUNNER_TEMP/openblas/bin/libopenblas.dll ./build/bin/Release/openblas.dll
+          cp $env:RUNNER_TEMP/OpenBLAS.LICENSE.txt ./build/bin/Release/OpenBLAS-${env:OPENBLAS_VERSION}.txt
+
+      - name: Check AVX512F support
+        id: check_avx512f
+        if: ${{ matrix.build == 'avx512' }}
+        continue-on-error: true
+        run: |
+          cd build
+          $vcdir = $(vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath)
+          $msvc = $(join-path $vcdir $('VC\Tools\MSVC\'+$(gc -raw $(join-path $vcdir 'VC\Auxiliary\Build\Microsoft.VCToolsVersion.default.txt')).Trim()))
+          $cl =  $(join-path $msvc 'bin\Hostx64\x64\cl.exe')
+          echo 'int main(void){unsigned int a[4];__cpuid(a,7);return !(a[1]&65536);}' >> avx512f.c
+          & $cl /O2 /GS- /kernel avx512f.c /link /nodefaultlib /entry:main
+          .\avx512f.exe && echo "AVX512F: YES" && ( echo HAS_AVX512F=1 >> $env:GITHUB_ENV ) || echo "AVX512F: NO"
+
+      - name: Test
+        id: cmake_test
+        if: ${{ matrix.build != 'clblast' && (matrix.build != 'avx512' || env.HAS_AVX512F == '1') }} # Test AVX-512 only when possible
+        run: |
+          cd build
+          ctest -C Release --verbose --timeout 900
+
+      - name: Get commit hash
+        id: commit
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        uses: pr-mpt/actions-commit-hash@v2
+
+      - name: Pack artifacts
+        id: pack_artifacts
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        run: |
+          Copy-Item LICENSE .\build\bin\Release\llama.cpp.txt
+          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-x64.zip .\build\bin\Release\*
+
+      - name: Upload artifacts
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        uses: actions/upload-artifact@v3
+        with:
+          path: |
+            llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-x64.zip
+
+  windows-latest-cmake-cublas:
+    runs-on: windows-latest
+
+    strategy:
+      matrix:
+        cuda: ['12.1.0', '11.7.1']
+        build: ['cublas']
+
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v1
+
+      - uses: Jimver/cuda-toolkit@v0.2.10
+        id: cuda-toolkit
+        with:
+          cuda: ${{ matrix.cuda }}
+          # TODO(green-sky): _dev seems to fail, and non dev are not enought
+          #sub-packages: '["nvcc", "cudart", "cublas", "cudart_dev", "cublas_dev"]'
+
+      - name: Build
+        id: cmake_build
+        run: |
+          mkdir build
+          cd build
+          cmake .. -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUBLAS=ON
+          cmake --build . --config Release
+
+      - name: Get commit hash
+        id: commit
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        uses: pr-mpt/actions-commit-hash@v2
+
+      - name: Pack artifacts
+        id: pack_artifacts
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        run: |
+          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-cu${{ matrix.cuda }}-x64.zip .\build\bin\Release\*
+
+      - name: Upload artifacts
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        uses: actions/upload-artifact@v3
+        with:
+          path: |
+            llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-${{ matrix.build }}-cu${{ matrix.cuda }}-x64.zip
+
+      - name: Copy and pack Cuda runtime
+        if: ${{ matrix.cuda == '12.1.0' }}
+        # TODO(green-sky): paths are cuda 12 specific
+        run: |
+          echo "Cuda install location: ${{steps.cuda-toolkit.outputs.CUDA_PATH}}"
+          mkdir '.\build\bin\cudart\'
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cudart64_12.dll" '.\build\bin\cudart\'
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublas64_12.dll" '.\build\bin\cudart\'
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublasLt64_12.dll" '.\build\bin\cudart\'
+          7z a cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip .\build\bin\cudart\*
+
+      - name: Copy and pack Cuda runtime
+        if: ${{ matrix.cuda == '11.7.1' }}
+        # TODO(green-sky): paths are cuda 11 specific
+        run: |
+          echo "Cuda install location: ${{steps.cuda-toolkit.outputs.CUDA_PATH}}"
+          mkdir '.\build\bin\cudart\'
+          ls "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin"
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cudart64_110.dll" '.\build\bin\cudart\'
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublas64_11.dll" '.\build\bin\cudart\'
+          cp "${{steps.cuda-toolkit.outputs.CUDA_PATH}}\bin\cublasLt64_11.dll" '.\build\bin\cudart\'
+          7z a cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip .\build\bin\cudart\*
+
+      - name: Upload Cuda runtime
+        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+        uses: actions/upload-artifact@v3
+        with:
+          path: |
+            cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip
+
+  release:
+    if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
+
+    runs-on: ubuntu-latest
+
+    needs:
+      - ubuntu-focal-make
+      - ubuntu-latest-cmake
+      - macOS-latest-make
+      - macOS-latest-cmake
+      - windows-latest-cmake
+      - windows-latest-cmake-cublas
+
+    steps:
+      - name: Download artifacts
+        id: download-artifact
+        uses: actions/download-artifact@v3
+
+      - name: Get commit hash
+        id: commit
+        uses: pr-mpt/actions-commit-hash@v2
+
+      - name: Create release
+        id: create_release
+        uses: anzz1/action-create-release@v1
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          tag_name: ${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}
+
+      - name: Upload release
+        id: upload_release
+        uses: actions/github-script@v3
+        with:
+          github-token: ${{secrets.GITHUB_TOKEN}}
+          script: |
+            const path = require('path');
+            const fs = require('fs');
+            const release_id = '${{ steps.create_release.outputs.id }}';
+            for (let file of await fs.readdirSync('./artifact')) {
+              if (path.extname(file) === '.zip') {
+                console.log('uploadReleaseAsset', file);
+                await github.repos.uploadReleaseAsset({
+                  owner: context.repo.owner,
+                  repo: context.repo.repo,
+                  release_id: release_id,
+                  name: file,
+                  data: await fs.readFileSync(`./artifact/${file}`)
+                });
+              }
+            }
+
+#  ubuntu-latest-gcc:
+#    runs-on: ubuntu-latest
+#
+#    strategy:
+#      matrix:
+#        build: [Debug, Release]
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Dependencies
+#        run: |
+#          sudo apt-get update
+#          sudo apt-get install build-essential
+#          sudo apt-get install cmake
+#
+#      - name: Configure
+#        run: cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }}
+#
+#      - name: Build
+#        run: |
+#          make
+#
+#  ubuntu-latest-clang:
+#    runs-on: ubuntu-latest
+#
+#    strategy:
+#      matrix:
+#        build: [Debug, Release]
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Dependencies
+#        run: |
+#          sudo apt-get update
+#          sudo apt-get install build-essential
+#          sudo apt-get install cmake
+#
+#      - name: Configure
+#        run: cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }} -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_C_COMPILER=clang
+#
+#      - name: Build
+#        run: |
+#          make
+#
+#  ubuntu-latest-gcc-sanitized:
+#    runs-on: ubuntu-latest
+#
+#    strategy:
+#      matrix:
+#        sanitizer: [ADDRESS, THREAD, UNDEFINED]
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Dependencies
+#        run: |
+#          sudo apt-get update
+#          sudo apt-get install build-essential
+#          sudo apt-get install cmake
+#
+#      - name: Configure
+#        run: cmake . -DCMAKE_BUILD_TYPE=Debug -DLLAMA_SANITIZE_${{ matrix.sanitizer }}=ON
+#
+#      - name: Build
+#        run: |
+#          make
+#
+#  windows:
+#    runs-on: windows-latest
+#
+#    strategy:
+#      matrix:
+#        build: [Release]
+#        arch: [Win32, x64]
+#        include:
+#          - arch: Win32
+#            s2arc: x86
+#          - arch: x64
+#            s2arc: x64
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Add msbuild to PATH
+#        uses: microsoft/setup-msbuild@v1
+#
+#      - name: Configure
+#        run: >
+#          cmake -S . -B ./build -A ${{ matrix.arch }}
+#          -DCMAKE_BUILD_TYPE=${{ matrix.build }}
+#
+#      - name: Build
+#        run: |
+#          cd ./build
+#          msbuild ALL_BUILD.vcxproj -t:build -p:configuration=${{ matrix.build }} -p:platform=${{ matrix.arch }}
+#
+#      - name: Upload binaries
+#        uses: actions/upload-artifact@v1
+#        with:
+#          name: llama-bin-${{ matrix.arch }}
+#          path: build/bin/${{ matrix.build }}
+#
+#  windows-blas:
+#    runs-on: windows-latest
+#
+#    strategy:
+#      matrix:
+#        build: [Release]
+#        arch: [Win32, x64]
+#        blas: [ON]
+#        include:
+#          - arch: Win32
+#            obzip: https://github.com/xianyi/OpenBLAS/releases/download/v0.3.21/OpenBLAS-0.3.21-x86.zip
+#            s2arc: x86
+#          - arch: x64
+#            obzip: https://github.com/xianyi/OpenBLAS/releases/download/v0.3.21/OpenBLAS-0.3.21-x64.zip
+#            s2arc: x64
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Add msbuild to PATH
+#        uses: microsoft/setup-msbuild@v1
+#
+#      - name: Fetch OpenBLAS
+#        if: matrix.blas == 'ON'
+#        run: |
+#          C:/msys64/usr/bin/wget.exe -qO blas.zip ${{ matrix.obzip }}
+#          7z x blas.zip -oblas -y
+#          copy blas/include/cblas.h .
+#          copy blas/include/openblas_config.h .
+#          echo "blasdir=$env:GITHUB_WORKSPACE/blas" >> $env:GITHUB_ENV
+#
+#      - name: Configure
+#        run: >
+#          cmake -S . -B ./build -A ${{ matrix.arch }}
+#          -DCMAKE_BUILD_TYPE=${{ matrix.build }}
+#          -DLLAMA_SUPPORT_OPENBLAS=${{ matrix.blas }}
+#          -DCMAKE_LIBRARY_PATH="$env:blasdir/lib"
+#
+#      - name: Build
+#        run: |
+#          cd ./build
+#          msbuild ALL_BUILD.vcxproj -t:build -p:configuration=${{ matrix.build }} -p:platform=${{ matrix.arch }}
+#
+#      - name: Copy libopenblas.dll
+#        if: matrix.blas == 'ON'
+#        run: copy "$env:blasdir/bin/libopenblas.dll" build/bin/${{ matrix.build }}
+#
+#      - name: Upload binaries
+#        if: matrix.blas == 'ON'
+#        uses: actions/upload-artifact@v1
+#        with:
+#          name: llama-blas-bin-${{ matrix.arch }}
+#          path: build/bin/${{ matrix.build }}
+#
+#  emscripten:
+#    runs-on: ubuntu-latest
+#
+#    strategy:
+#      matrix:
+#        build: [Release]
+#
+#    steps:
+#      - name: Clone
+#        uses: actions/checkout@v1
+#
+#      - name: Dependencies
+#        run: |
+#          wget -q https://github.com/emscripten-core/emsdk/archive/master.tar.gz
+#          tar -xvf master.tar.gz
+#          emsdk-master/emsdk update
+#          emsdk-master/emsdk install latest
+#          emsdk-master/emsdk activate latest
+#
+#      - name: Configure
+#        run: echo "tmp"
+#
+#      - name: Build
+#        run: |
+#          pushd emsdk-master
+#          source ./emsdk_env.sh
+#          popd
+#          emcmake cmake . -DCMAKE_BUILD_TYPE=${{ matrix.build }}
+#          make
diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
new file mode 100644
index 0000000000000..379fbd7ad35f1
--- /dev/null
+++ b/.github/workflows/docker.yml
@@ -0,0 +1,65 @@
+# This workflow uses actions that are not certified by GitHub.
+# They are provided by a third-party and are governed by
+# separate terms of service, privacy policy, and support
+# documentation.
+
+# GitHub recommends pinning actions to a commit SHA.
+# To get a newer version, you will need to update the SHA.
+# You can also reference a tag or branch, but the action may change without warning.
+
+name: Publish Docker image
+
+on:
+  pull_request:
+  push:
+    branches:
+      - master
+
+jobs:
+  push_to_registry:
+    name: Push Docker image to Docker Hub
+    if: github.event.pull_request.draft == false
+
+    runs-on: ubuntu-latest
+    env:
+      COMMIT_SHA: ${{ github.sha }}
+    strategy:
+      matrix:
+        config:
+          - { tag: "light", dockerfile: ".devops/main.Dockerfile" }
+          - { tag: "full", dockerfile: ".devops/full.Dockerfile" }
+    steps:
+      - name: Check out the repo
+        uses: actions/checkout@v3
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v2
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+
+      - name: Log in to Docker Hub
+        uses: docker/login-action@v2
+        with:
+          registry: ghcr.io
+          username: ${{ github.repository_owner }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build and push Docker image (versioned)
+        if: github.event_name == 'push'
+        uses: docker/build-push-action@v4
+        with:
+          context: .
+          push: true
+          platforms: linux/amd64,linux/arm64
+          tags: "ghcr.io/ggerganov/llama.cpp:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }}"
+          file: ${{ matrix.config.dockerfile }}
+
+      - name: Build and push Docker image (tagged)
+        uses: docker/build-push-action@v4
+        with:
+          context: .
+          push: ${{ github.event_name == 'push' }}
+          platforms: linux/amd64,linux/arm64
+          tags: "ghcr.io/ggerganov/llama.cpp:${{ matrix.config.tag }}"
+          file: ${{ matrix.config.dockerfile }}
diff --git a/.github/workflows/editorconfig.yml b/.github/workflows/editorconfig.yml
new file mode 100644
index 0000000000000..b4e535acf1f64
--- /dev/null
+++ b/.github/workflows/editorconfig.yml
@@ -0,0 +1,17 @@
+name: EditorConfig Checker
+
+on:
+  push:
+    branches:
+      - master
+  pull_request:
+    branches:
+      - master
+
+jobs:
+  editorconfig:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: editorconfig-checker/action-editorconfig-checker@main
+      - run: editorconfig-checker
diff --git a/.github/workflows/tidy-post.yml b/.github/workflows/tidy-post.yml
new file mode 100644
index 0000000000000..03652760c80dc
--- /dev/null
+++ b/.github/workflows/tidy-post.yml
@@ -0,0 +1,20 @@
+name: clang-tidy review post comments
+
+on:
+  workflow_dispatch:
+    workflows: ["clang-tidy-review"]
+    types:
+      - completed
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: ZedThree/clang-tidy-review/post@v0.13.0
+        # lgtm_comment_body, max_comments, and annotations need to be set on the posting workflow in a split setup
+        with:
+          # adjust options as necessary
+          lgtm_comment_body: ''
+          annotations: false
+          max_comments: 25
diff --git a/.github/workflows/tidy-review.yml b/.github/workflows/tidy-review.yml
new file mode 100644
index 0000000000000..a4bc8d976560e
--- /dev/null
+++ b/.github/workflows/tidy-review.yml
@@ -0,0 +1,23 @@
+name: clang-tidy-review
+
+on:
+  pull_request:
+    branches:
+      - master
+
+jobs:
+  clang-tidy-review:
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - uses: ZedThree/clang-tidy-review@v0.13.0
+      id: review
+      with:
+        lgtm_comment_body: ''
+        build_dir: build
+        cmake_command: cmake . -B build -DCMAKE_EXPORT_COMPILE_COMMANDS=on
+        split_workflow: true
+
+    - uses: ZedThree/clang-tidy-review/upload@v0.13.0

From d646c4efcea098adbda055ae2df2c9399d19b971 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 17:20:36 +0200
Subject: [PATCH 171/242] convert.py : n_head_kv optional and .gguf file
 extension

---
 convert.py | 41 +++++++++++++++++++++++------------------
 1 file changed, 23 insertions(+), 18 deletions(-)

diff --git a/convert.py b/convert.py
index b243356f0db28..86f1a97476186 100755
--- a/convert.py
+++ b/convert.py
@@ -150,15 +150,20 @@ def guessed(model: 'LazyModel') -> 'Params':
     def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab    = config["vocab_size"];
-        n_embd     = config["hidden_size"];
-        n_layer    = config["num_hidden_layers"];
-        n_ff       = config["intermediate_size"];
-        n_head     = config["num_attention_heads"];
-        n_head_kv  = config["num_key_value_heads"];
-        f_norm_eps = config["rms_norm_eps"];
+        n_vocab    = config["vocab_size"]
+        n_embd     = config["hidden_size"]
+        n_layer    = config["num_hidden_layers"]
+        n_ff       = config["intermediate_size"]
+        n_head     = config["num_attention_heads"]
+
+        if "num_key_value_heads" in config:
+            n_head_kv  = config["num_key_value_heads"]
+        else:
+            n_head_kv = None
+
+        f_norm_eps = config["rms_norm_eps"]
 
-        n_mult = Params.find_n_mult(n_ff, n_embd);
+        n_mult = Params.find_n_mult(n_ff, n_embd)
 
         if "max_sequence_length" in config:
             n_ctx = config["max_sequence_length"]
@@ -186,15 +191,15 @@ def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
     def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         config = json.load(open(config_path))
 
-        n_vocab    = config["vocab_size"];
-        n_embd     = config["dim"];
-        n_layer    = config["n_layers"];
-        n_mult     = config["multiple_of"];
+        n_vocab    = config["vocab_size"]
+        n_embd     = config["dim"]
+        n_layer    = config["n_layers"]
+        n_mult     = config["multiple_of"]
         n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
-        n_ff       = -1;
-        n_head     = config["n_heads"];
-        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head;
-        f_norm_eps = config["norm_eps"];
+        n_ff       = -1
+        n_head     = config["n_heads"]
+        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head
+        f_norm_eps = config["norm_eps"]
 
         if n_vocab == -1:
             n_vocab = model["tok_embeddings.weight"].shape[0]
@@ -714,7 +719,7 @@ def add_meta_arch(self, params: Params) -> None:
         self.gguf.add_feed_forward_length (params.n_ff)
         self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
         self.gguf.add_head_count          (params.n_head)
-        self.gguf.add_head_count_kv       (params.n_head_kv)
+        if params.n_head_kv is not None: self.gguf.add_head_count_kv(params.n_head_kv)
         self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
 
     def add_meta_vocab(self, vocab: Vocab) -> None:
@@ -934,7 +939,7 @@ def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
         GGMLFileType.AllF32:    "f32",
         GGMLFileType.MostlyF16: "f16",
     }[file_type]
-    ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
+    ret = model_paths[0].parent / f"ggml-model-{namestr}.gguf"
     if ret in model_paths:
         sys.stderr.write(
             f"Error: Default output path ({ret}) would overwrite the input. "

From 8ace03ad3d8ba62dc3cb27a7e739f42eeb97c09c Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 18:47:06 +0300
Subject: [PATCH 172/242] convert.py : better always have n_head_kv and default
 it to n_head

---
 convert.py | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

diff --git a/convert.py b/convert.py
index 86f1a97476186..adc5fdd26d804 100755
--- a/convert.py
+++ b/convert.py
@@ -155,12 +155,7 @@ def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
         n_layer    = config["num_hidden_layers"]
         n_ff       = config["intermediate_size"]
         n_head     = config["num_attention_heads"]
-
-        if "num_key_value_heads" in config:
-            n_head_kv  = config["num_key_value_heads"]
-        else:
-            n_head_kv = None
-
+        n_head_kv  = config["num_key_value_heads"] if "num_key_value_heads" in config else n_head
         f_norm_eps = config["rms_norm_eps"]
 
         n_mult = Params.find_n_mult(n_ff, n_embd)
@@ -719,7 +714,7 @@ def add_meta_arch(self, params: Params) -> None:
         self.gguf.add_feed_forward_length (params.n_ff)
         self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
         self.gguf.add_head_count          (params.n_head)
-        if params.n_head_kv is not None: self.gguf.add_head_count_kv(params.n_head_kv)
+        self.gguf.add_head_count_kv       (params.n_head_kv)
         self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
 
     def add_meta_vocab(self, vocab: Vocab) -> None:

From 11bf4366c202e5931cb1a67efa444b84cd205bad Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:03:15 +0300
Subject: [PATCH 173/242] llama : sync with recent PRs on master

---
 ggml-alloc.c         |  42 +-
 ggml-alloc.h         |   4 +
 ggml-metal.h         |   9 +-
 ggml-metal.m         | 195 ++++-----
 ggml-metal.metal     | 969 +++++++++++++++++++++----------------------
 llama.cpp            |  68 ++-
 tests/CMakeLists.txt |   1 +
 7 files changed, 616 insertions(+), 672 deletions(-)

diff --git a/ggml-alloc.c b/ggml-alloc.c
index 4121f3dbab20f..3ee98d03dea4d 100644
--- a/ggml-alloc.c
+++ b/ggml-alloc.c
@@ -67,6 +67,8 @@ struct ggml_allocr {
     struct hash_node hash_table[GGML_GRAPH_HASHTABLE_SIZE];
     size_t max_size;
     bool measure;
+    int parse_seq[GGML_MAX_NODES];
+    bool has_parse_seq;
 
 #ifdef GGML_ALLOCATOR_DEBUG
     struct ggml_tensor * allocated_tensors[1024];
@@ -111,10 +113,10 @@ void ggml_allocr_alloc(struct ggml_allocr * alloc, struct ggml_tensor * tensor)
 
     size_t max_avail = 0;
 
-    // find the best fitting free block
+    // find the best fitting free block besides the last block
     int best_fit_block = -1;
     size_t best_fit_size = SIZE_MAX;
-    for (int i = 0; i < alloc->n_free_blocks; i++) {
+    for (int i = 0; i < alloc->n_free_blocks - 1; i++) {
         struct free_block * block = &alloc->free_blocks[i];
         max_avail = MAX(max_avail, block->size);
         if (block->size >= size && block->size <= best_fit_size) {
@@ -126,10 +128,17 @@ void ggml_allocr_alloc(struct ggml_allocr * alloc, struct ggml_tensor * tensor)
     AT_PRINTF("block %d\n", best_fit_block);
 
     if (best_fit_block == -1) {
-        fprintf(stderr, "%s: not enough space in the buffer (needed %zu, largest block available %zu)\n",
-                __func__, size, max_avail);
-        GGML_ASSERT(!"not enough space in the buffer");
+        // the last block is our last resort
+        struct free_block * block = &alloc->free_blocks[alloc->n_free_blocks - 1];
+        if (block->size >= size) {
+            best_fit_block = alloc->n_free_blocks - 1;
+            max_avail = MAX(max_avail, block->size);
+        } else {
+            fprintf(stderr, "%s: not enough space in the buffer (needed %zu, largest block available %zu)\n",
+                    __func__, size, max_avail);
+            GGML_ASSERT(!"not enough space in the buffer");
         return;
+        }
     }
     struct free_block * block = &alloc->free_blocks[best_fit_block];
     void * addr = block->addr;
@@ -229,6 +238,17 @@ static void ggml_allocator_free_tensor(struct ggml_allocr * alloc, struct ggml_t
     alloc->n_free_blocks++;
 }
 
+void ggml_allocr_set_parse_seq(struct ggml_allocr * alloc, int * list, int n) {
+    int pos = 0;
+    for (int i = 0; i < n; i++) {
+        if (list[i] != -1) {
+            alloc->parse_seq[pos] = list[i];
+            pos++;
+        }
+    }
+    alloc->has_parse_seq = true;
+}
+
 void ggml_allocr_reset(struct ggml_allocr * alloc) {
     alloc->n_free_blocks = 1;
     size_t align_offset = aligned_offset(alloc->data, 0, alloc->alignment);
@@ -248,6 +268,8 @@ struct ggml_allocr * ggml_allocr_new(void * data, size_t size, size_t alignment)
         /*.hash_table    = */ {{0}},
         /*.max_size      = */ 0,
         /*.measure       = */ false,
+        /*.parse_seq     = */ {0},
+        /*.has_parse_seq = */ false,
 #ifdef GGML_ALLOCATOR_DEBUG
         /*.allocated_tensors = */ = {0},
 #endif
@@ -275,6 +297,8 @@ struct ggml_allocr * ggml_allocr_new_measure(size_t alignment) {
         /*.hash_table    = */ {{0}},
         /*.max_size      = */ 0,
         /*.measure       = */ true,
+        /*.parse_seq     = */ {0},
+        /*.has_parse_seq = */ false,
 #ifdef GGML_ALLOCATOR_DEBUG
         /*.allocated_tensors = */ = {0},
 #endif
@@ -473,7 +497,13 @@ static size_t ggml_allocator_alloc_graph_tensors_n(
                 allocate_node(alloc, input);
             }
         }
-        for (int i = 0; i < gf->n_nodes; i++) {
+        for (int ind = 0; ind < gf->n_nodes; ind++) {
+            int i;
+            if (alloc->has_parse_seq) {
+                i = alloc->parse_seq[ind];
+            } else {
+                i = ind;
+            }
             struct ggml_tensor * node = gf->nodes[i];
 
             // allocate parents (leafs)
diff --git a/ggml-alloc.h b/ggml-alloc.h
index a5ec8f87a9453..14a4350ac2e96 100644
--- a/ggml-alloc.h
+++ b/ggml-alloc.h
@@ -10,6 +10,10 @@ extern "C" {
 GGML_API struct ggml_allocr * ggml_allocr_new(void * data, size_t size, size_t alignment);
 GGML_API struct ggml_allocr * ggml_allocr_new_measure(size_t alignment);
 
+// tell the allocator to parse nodes following the order described in the list
+// you should call this if your graph are optimized to execute out-of-order
+GGML_API void   ggml_allocr_set_parse_seq(struct ggml_allocr * alloc, int * list, int n);
+
 GGML_API void   ggml_allocr_free(struct ggml_allocr * alloc);
 GGML_API bool   ggml_allocr_is_measure(struct ggml_allocr * alloc);
 GGML_API void   ggml_allocr_reset(struct ggml_allocr * alloc);
diff --git a/ggml-metal.h b/ggml-metal.h
index 1a5d96c339913..00202b787c804 100644
--- a/ggml-metal.h
+++ b/ggml-metal.h
@@ -66,10 +66,13 @@ void ggml_metal_get_tensor(struct ggml_metal_context * ctx, struct ggml_tensor *
 
 // try to find operations that can be run concurrently in the graph
 // you should run it again if the topology of your graph changes
-void ggml_metal_graph_find_concurrency(struct ggml_metal_context * ctx, struct ggml_cgraph * gf);
+void ggml_metal_graph_find_concurrency(struct ggml_metal_context * ctx, struct ggml_cgraph * gf, bool check_mem);
 
-// if the graph has been optimized for concurrently dispatch
-bool ggml_metal_if_optimized(struct ggml_metal_context * ctx);
+// if the graph has been optimized for concurrently dispatch, return length of the concur_list if optimized
+int ggml_metal_if_optimized(struct ggml_metal_context * ctx);
+
+// output the concur_list for ggml_alloc
+int * ggml_metal_get_concur_list(struct ggml_metal_context * ctx);
 
 // same as ggml_graph_compute but uses Metal
 // creates gf->n_threads command buffers in parallel
diff --git a/ggml-metal.m b/ggml-metal.m
index 9cc4d9e030463..835c5f297cf95 100644
--- a/ggml-metal.m
+++ b/ggml-metal.m
@@ -5,7 +5,6 @@
 #import <Foundation/Foundation.h>
 
 #import <Metal/Metal.h>
-#import <MetalPerformanceShaders/MetalPerformanceShaders.h>
 
 #undef MIN
 #undef MAX
@@ -79,6 +78,14 @@
     GGML_METAL_DECL_KERNEL(mul_mat_q4_K_f32);
     GGML_METAL_DECL_KERNEL(mul_mat_q5_K_f32);
     GGML_METAL_DECL_KERNEL(mul_mat_q6_K_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_f16_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q4_0_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q4_1_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q2_K_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q3_K_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q4_K_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q5_K_f32);
+    GGML_METAL_DECL_KERNEL(mul_mm_q6_K_f32);
     GGML_METAL_DECL_KERNEL(rope);
     GGML_METAL_DECL_KERNEL(alibi_f32);
     GGML_METAL_DECL_KERNEL(cpy_f32_f16);
@@ -110,13 +117,6 @@ @implementation GGMLMetalClass
     ctx->n_buffers = 0;
     ctx->concur_list_len = 0;
 
-    // determine if we can use MPS
-    if (MPSSupportsMTLDevice(ctx->device)) {
-        fprintf(stderr, "%s: using MPS\n", __func__);
-    } else {
-        fprintf(stderr, "%s: not using MPS\n", __func__);
-        GGML_ASSERT(false && "MPS not supported");
-    }
 
 #if 0
     // compile from source string and show compile log
@@ -163,10 +163,15 @@ @implementation GGMLMetalClass
 
     // load kernels
     {
+        NSError * error = nil;
 #define GGML_METAL_ADD_KERNEL(name) \
         ctx->function_##name = [ctx->library newFunctionWithName:@"kernel_"#name]; \
-        ctx->pipeline_##name = [ctx->device newComputePipelineStateWithFunction:ctx->function_##name error:nil]; \
-        fprintf(stderr, "%s: loaded %-32s %16p\n", __func__, "kernel_"#name, (void *) ctx->pipeline_##name);
+        ctx->pipeline_##name = [ctx->device newComputePipelineStateWithFunction:ctx->function_##name error:&error]; \
+        fprintf(stderr, "%s: loaded %-32s %16p\n", __func__, "kernel_"#name, (void *) ctx->pipeline_##name); \
+        if (error) { \
+            fprintf(stderr, "%s: load pipeline error: %s\n", __func__, [[error description] UTF8String]); \
+            return NULL; \
+        }
 
         GGML_METAL_ADD_KERNEL(add);
         GGML_METAL_ADD_KERNEL(add_row);
@@ -196,6 +201,14 @@ @implementation GGMLMetalClass
         GGML_METAL_ADD_KERNEL(mul_mat_q4_K_f32);
         GGML_METAL_ADD_KERNEL(mul_mat_q5_K_f32);
         GGML_METAL_ADD_KERNEL(mul_mat_q6_K_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_f16_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q4_0_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q4_1_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q2_K_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q3_K_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q4_K_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q5_K_f32);
+        GGML_METAL_ADD_KERNEL(mul_mm_q6_K_f32);
         GGML_METAL_ADD_KERNEL(rope);
         GGML_METAL_ADD_KERNEL(alibi_f32);
         GGML_METAL_ADD_KERNEL(cpy_f32_f16);
@@ -243,11 +256,12 @@ void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb) {
     ctx->n_cb = n_cb;
 }
 
-bool ggml_metal_if_optimized(struct ggml_metal_context * ctx) {
-    if (ctx->concur_list_len) {
-        return true;
-    }
-    return false;
+int ggml_metal_if_optimized(struct ggml_metal_context * ctx) {
+    return ctx->concur_list_len;
+}
+
+int * ggml_metal_get_concur_list(struct ggml_metal_context * ctx) {
+    return ctx->concur_list;
 }
 
 // finds the Metal buffer that contains the tensor data on the GPU device
@@ -390,7 +404,7 @@ void ggml_metal_get_tensor(
 
 void ggml_metal_graph_find_concurrency(
         struct ggml_metal_context * ctx,
-        struct ggml_cgraph * gf) {
+        struct ggml_cgraph * gf, bool check_mem) {
     int search_depth = gf->n_nodes; //we only find concurrency in this range to avoid wasting too much time
     int nodes_unused[GGML_MAX_CONCUR];
 
@@ -437,7 +451,7 @@ void ggml_metal_graph_find_concurrency(
                         }
                     }
                 }
-                if (exe_flag) {
+                if (exe_flag && check_mem) {
                     // check if nodes[i]'s data will be overwritten by a node before nodes[i].
                     // if node[5] and node[3] write to the same memory region, then we can't issue node[5] before node[3]
                     int64_t data_start = (int64_t) gf->nodes[i]->data;
@@ -521,7 +535,7 @@ void ggml_metal_graph_compute(
 
             id<MTLCommandBuffer> command_buffer = command_buffers[cb_idx];
 
-            id<MTLComputeCommandEncoder> encoder = nil;
+            id<MTLComputeCommandEncoder> encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
 
             const int node_start =                                  (cb_idx + 0) * n_nodes_per_cb;
             const int node_end   = (cb_idx == n_cb - 1) ? n_nodes : (cb_idx + 1) * n_nodes_per_cb;
@@ -530,10 +544,6 @@ void ggml_metal_graph_compute(
                 const int i = has_concur ? ctx->concur_list[ind] : ind;
 
                 if (i == -1) {
-                    if (encoder == nil) {
-                        encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                        continue;
-                    }
                     [encoder memoryBarrierWithScope:MTLBarrierScopeBuffers];
                     continue;
                 }
@@ -607,10 +617,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_ADD:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             if (ggml_nelements(src1) == ne10) {
                                 // src1 is a row
                                 [encoder setComputePipelineState:ctx->pipeline_add_row];
@@ -628,10 +634,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_MUL:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             if (ggml_nelements(src1) == ne10) {
                                 // src1 is a row
                                 [encoder setComputePipelineState:ctx->pipeline_mul_row];
@@ -649,10 +651,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_SCALE:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const float scale = *(const float *) src1->data;
 
                             [encoder setComputePipelineState:ctx->pipeline_scale];
@@ -668,10 +666,6 @@ void ggml_metal_graph_compute(
                         switch (ggml_get_unary_op(gf->nodes[i])) {
                             case GGML_UNARY_OP_SILU:
                                 {
-                                    if (encoder == nil) {
-                                        encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                                    }
-
                                     [encoder setComputePipelineState:ctx->pipeline_silu];
                                     [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
                                     [encoder setBuffer:id_dst  offset:offs_dst  atIndex:1];
@@ -682,10 +676,6 @@ void ggml_metal_graph_compute(
                                 } break;
                             case GGML_UNARY_OP_RELU:
                                 {
-                                    if (encoder == nil) {
-                                        encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                                    }
-
                                     [encoder setComputePipelineState:ctx->pipeline_relu];
                                     [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
                                     [encoder setBuffer:id_dst  offset:offs_dst  atIndex:1];
@@ -696,10 +686,6 @@ void ggml_metal_graph_compute(
                                 } break;
                             case GGML_UNARY_OP_GELU:
                                 {
-                                    if (encoder == nil) {
-                                        encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                                    }
-
                                     [encoder setComputePipelineState:ctx->pipeline_gelu];
                                     [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
                                     [encoder setBuffer:id_dst  offset:offs_dst  atIndex:1];
@@ -716,10 +702,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_SOFT_MAX:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const int nth = 32;
 
                             [encoder setComputePipelineState:ctx->pipeline_soft_max];
@@ -734,10 +716,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_DIAG_MASK_INF:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const int n_past = ((int32_t *)(dst->op_params))[0];
 
                             [encoder setComputePipelineState:ctx->pipeline_diag_mask_inf];
@@ -755,53 +733,43 @@ void ggml_metal_graph_compute(
 
                             GGML_ASSERT(ne00 == ne10);
                             // GGML_ASSERT(ne02 == ne12); // Should be checked on individual data types until broadcast is implemented everywhere
+                            uint gqa = ne12/ne02;
                             GGML_ASSERT(ne03 == ne13);
 
+                            // for now the matrix-matrix multiplication kernel only works on A14+/M1+ SoCs
+                            // AMD GPU and older A-chips will reuse matrix-vector multiplication kernel
                             if (ggml_is_contiguous(src0) &&
                                 ggml_is_contiguous(src1) &&
-                                (src0t == GGML_TYPE_F32 || src0t == GGML_TYPE_F16) && ne11 > 1) {
-
-                                if (encoder != nil) {
-                                    [encoder endEncoding];
-                                    encoder = nil;
-                                }
-
-                                MPSDataType src0dt = src0t == GGML_TYPE_F32 ? MPSDataTypeFloat32 : MPSDataTypeFloat16;
-                                MPSDataType src1dt = src1t == GGML_TYPE_F32 ? MPSDataTypeFloat32 : MPSDataTypeFloat16;
-
-                                // for F32 x F32 we use MPS
-                                MPSMatrixDescriptor * desc0 = [MPSMatrixDescriptor
-                                    matrixDescriptorWithRows:ne01 columns:ne00 rowBytes:src0->nb[1] dataType:src0dt];
-
-                                MPSMatrixDescriptor * desc1 = [MPSMatrixDescriptor
-                                    matrixDescriptorWithRows:ne11 columns:ne10 rowBytes:src1->nb[1] dataType:src1dt];
-
-                                MPSMatrixDescriptor * desc  = [MPSMatrixDescriptor
-                                    matrixDescriptorWithRows:ne1 columns:ne0 rowBytes:dst->nb[1] dataType:MPSDataTypeFloat32];
-
-                                MPSMatrixMultiplication * mul = [[MPSMatrixMultiplication alloc]
-                                    initWithDevice:ctx->device transposeLeft:false transposeRight:true
-                                        resultRows:ne11 resultColumns:ne01 interiorColumns:ne00 alpha:1.0 beta:0.0];
-
-                                // we need to do ne12 multiplications
-                                // TODO: is there a way to do this in parallel - currently very slow ..
-                                // TODO: might be possible to offload part of the computation to ANE using Accelerate's CBLAS
-                                for (int64_t i02 = 0; i02 < ne12; ++i02) {
-                                    size_t offs_src0_cur = offs_src0 + i02/(ne12/ne02)*nb02; // gqa not used for now
-                                    size_t offs_src1_cur = offs_src1 + i02*nb12;
-                                    size_t offs_dst_cur  = offs_dst  + i02*nb2;
-
-                                    MPSMatrix * mat_src0 = [[MPSMatrix alloc] initWithBuffer:id_src0 offset:offs_src0_cur descriptor:desc0];
-                                    MPSMatrix * mat_src1 = [[MPSMatrix alloc] initWithBuffer:id_src1 offset:offs_src1_cur descriptor:desc1];
-                                    MPSMatrix * mat_dst  = [[MPSMatrix alloc] initWithBuffer:id_dst  offset:offs_dst_cur  descriptor:desc ];
-
-                                    [mul encodeToCommandBuffer:command_buffer leftMatrix:mat_src1 rightMatrix:mat_src0 resultMatrix:mat_dst];
-                                }
-                            } else {
-                                if (encoder == nil) {
-                                    encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
+                                src1t == GGML_TYPE_F32 &&
+                                [ctx->device supportsFamily:MTLGPUFamilyApple7] &&
+                                ne00%32 == 0 &&
+                                ne11 > 1) {
+                                    switch (src0->type) {
+                                        case GGML_TYPE_F16:  [encoder setComputePipelineState:ctx->pipeline_mul_mm_f16_f32]; break;
+                                        case GGML_TYPE_Q4_0: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q4_0_f32]; break;
+                                        case GGML_TYPE_Q4_1: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q4_1_f32]; break;
+                                        case GGML_TYPE_Q2_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q2_K_f32]; break;
+                                        case GGML_TYPE_Q3_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q3_K_f32]; break;
+                                        case GGML_TYPE_Q4_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q4_K_f32]; break;
+                                        case GGML_TYPE_Q5_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q5_K_f32]; break;
+                                        case GGML_TYPE_Q6_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q6_K_f32]; break;
+                                        default: GGML_ASSERT(false && "MUL MAT-MAT not implemented");
+                                    }
+                                    [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
+                                    [encoder setBuffer:id_src1 offset:offs_src1 atIndex:1];
+                                    [encoder setBuffer:id_dst  offset:offs_dst  atIndex:2];
+                                    [encoder setBytes:&ne00 length:sizeof(ne00) atIndex:3];
+                                    [encoder setBytes:&ne02 length:sizeof(ne02) atIndex:4];
+                                    [encoder setBytes:&nb01 length:sizeof(nb01) atIndex:5];
+                                    [encoder setBytes:&nb02 length:sizeof(nb02) atIndex:6];
+                                    [encoder setBytes:&ne12 length:sizeof(ne12) atIndex:7];
+                                    [encoder setBytes:&ne0 length:sizeof(ne0) atIndex:8];
+                                    [encoder setBytes:&ne1 length:sizeof(ne1) atIndex:9];
+                                    [encoder setBytes:&gqa length:sizeof(gqa) atIndex:10];
+                                    [encoder setThreadgroupMemoryLength:8192 atIndex:0];
+                                    [encoder dispatchThreadgroups:MTLSizeMake( (ne11+31)/32, (ne01+63) / 64, ne12) threadsPerThreadgroup:MTLSizeMake(128, 1, 1)];
                                 }
-
+                            else {
                                 int nth0 = 32;
                                 int nth1 = 1;
 
@@ -900,23 +868,24 @@ void ggml_metal_graph_compute(
                                 [encoder setBytes:&nb12 length:sizeof(nb12) atIndex:14];
                                 [encoder setBytes:&ne0  length:sizeof(ne0)  atIndex:15];
                                 [encoder setBytes:&ne1  length:sizeof(ne1)  atIndex:16];
+                                [encoder setBytes:&gqa length:sizeof(gqa) atIndex:17];
 
                                 if (src0t == GGML_TYPE_Q4_0 || src0t == GGML_TYPE_Q4_1 ||
                                     src0t == GGML_TYPE_Q2_K || src0t == GGML_TYPE_Q4_K) {
-                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7) / 8, ne11, 1) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
+                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7) / 8, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
                                 }
                                 else if (src0t == GGML_TYPE_Q3_K) {
 #ifdef GGML_QKK_64
-                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+1)/2, ne11, 1) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
+                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+1)/2, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
 #else
-                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+3)/4, ne11, 1) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
+                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+3)/4, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
 #endif
                                 }
                                 else if (src0t == GGML_TYPE_Q5_K) {
-                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 3) / 4, ne11, 1) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
+                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 3) / 4, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
                                 }
                                 else if (src0t == GGML_TYPE_Q6_K) {
-                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+1)/2, ne11, 1) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
+                                    [encoder dispatchThreadgroups:MTLSizeMake((ne01+1)/2, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
                                 } else {
                                     [encoder setThreadgroupMemoryLength:nth0*sizeof(float) atIndex:0];
                                     [encoder dispatchThreadgroups:MTLSizeMake(ne01, ne11, ne12) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)];
@@ -925,10 +894,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_GET_ROWS:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             switch (src0->type) {
                                 case GGML_TYPE_F16:  [encoder setComputePipelineState:ctx->pipeline_get_rows_f16]; break;
                                 case GGML_TYPE_Q4_0: [encoder setComputePipelineState:ctx->pipeline_get_rows_q4_0]; break;
@@ -954,10 +919,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_RMS_NORM:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             float eps;
                             memcpy(&eps, dst->op_params, sizeof(float));
 
@@ -977,10 +938,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_NORM:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const float eps = 1e-5f;
 
                             const int nth = 256;
@@ -999,10 +956,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_ALIBI:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             GGML_ASSERT((src0t == GGML_TYPE_F32));
 
                             const int n_past = ((int32_t *) dst->op_params)[0]; UNUSED(n_past);
@@ -1042,10 +995,6 @@ void ggml_metal_graph_compute(
                         } break;
                     case GGML_OP_ROPE:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const int n_past = ((int32_t *) dst->op_params)[0];
                             const int n_dims = ((int32_t *) dst->op_params)[1];
                             const int mode   = ((int32_t *) dst->op_params)[2];
@@ -1086,10 +1035,6 @@ void ggml_metal_graph_compute(
                     case GGML_OP_CPY:
                     case GGML_OP_CONT:
                         {
-                            if (encoder == nil) {
-                                encoder = [command_buffer computeCommandEncoderWithDescriptor: edesc];
-                            }
-
                             const int nth = 32;
 
                             switch (src0t) {
diff --git a/ggml-metal.metal b/ggml-metal.metal
index 8d26b5ec2dfa4..3f3125236f197 100644
--- a/ggml-metal.metal
+++ b/ggml-metal.metal
@@ -18,47 +18,6 @@ typedef struct {
     uint8_t qs[QK4_1 / 2];  // nibbles / quants
 } block_q4_1;
 
-static void dequantize_row_q4_0(device const block_q4_0 * x, device float * y, int k) {
-    const int qk = QK4_0;
-
-    assert(k % qk == 0);
-
-    const int nb = k / qk;
-
-    for (int i = 0; i < nb; i++) {
-        const half d = x[i].d;
-
-        for (int j = 0; j < qk/2; ++j) {
-            const int x0 = (x[i].qs[j] & 0x0F) - 8;
-            const int x1 = (x[i].qs[j] >>   4) - 8;
-
-            y[i*qk + j + 0   ] = x0*d;
-            y[i*qk + j + qk/2] = x1*d;
-        }
-    }
-}
-
-static void dequantize_row_q4_1(device const block_q4_1 * x, device float * y, int k) {
-    const int qk = QK4_1;
-
-    assert(k % qk == 0);
-
-    const int nb = k / qk;
-
-    for (int i = 0; i < nb; i++) {
-        const half d = x[i].d;
-        const half m = x[i].m;
-
-        for (int j = 0; j < qk/2; ++j) {
-            const int x0 = (x[i].qs[j] & 0x0F);
-            const int x1 = (x[i].qs[j] >>   4);
-
-            y[i*qk + j + 0   ] = x0*d + m;
-            y[i*qk + j + qk/2] = x1*d + m;
-        }
-    }
-}
-
 kernel void kernel_add(
         device const float * src0,
         device const float * src1,
@@ -219,54 +178,6 @@ kernel void kernel_diag_mask_inf(
     }
 }
 
-kernel void kernel_get_rows_f16(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    for (int j = 0; j < ne00; j++) {
-        dst[i*nb1 + j] = ((device half *) ((device char *) src0 + r*nb01))[j];
-    }
-}
-
-kernel void kernel_get_rows_q4_0(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q4_0(
-            (device const block_q4_0 *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
-kernel void kernel_get_rows_q4_1(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q4_1(
-            (device const block_q4_1 *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
 kernel void kernel_norm(
         device const  void * src0,
         device       float * dst,
@@ -432,14 +343,16 @@ inline float block_q_n_dot_y(device const block_q4_1 * qb_curr, float sumy, thre
 //      N_DST, so this is another explicit assumption of the implementation.
 template<typename block_q_type, int nr, int nsg, int nw>
 void mul_vec_q_n_f32(device const void * src0, device const float * src1, device float * dst,
-                    int64_t ne00, int64_t ne10, int64_t ne0, int64_t ne01,
-                    uint2 tgpig, uint tiisg, uint sgitg) {
+                    int64_t ne00, int64_t ne01, int64_t ne02, int64_t ne10, int64_t ne12, int64_t ne0, int64_t ne1, uint gqa,
+                    uint3 tgpig, uint tiisg, uint sgitg) {
     const int nb = ne00/QK4_0;
     const int r0 = tgpig.x;
     const int r1 = tgpig.y;
+    const int im = tgpig.z;
     const int first_row = (r0 * nsg + sgitg) * nr;
-    device const block_q_type * x = (device const block_q_type *) src0 + first_row * nb;
-    device const float      * y = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = first_row * nb + im/gqa*(nb*ne0);
+    device const block_q_type * x = (device const block_q_type *) src0 + offset0;
+    device const float      * y = (device const float      *) src1 + r1*ne10 + im*ne00*ne1;
     float yl[16];       // src1 vector cache
     float sumf[nr]={0.f};
 
@@ -470,7 +383,7 @@ void mul_vec_q_n_f32(device const void * src0, device const float * src1, device
     for (int row = 0; row < nr; ++row) {
         const float tot = simd_sum(sumf[row]);
         if (tiisg == 0 && first_row + row < ne01) {
-            dst[r1*ne0 + first_row + row] = tot;
+            dst[r1*ne0 + im*ne0*ne1 + first_row + row] = tot;
         }
     }
 }
@@ -480,13 +393,17 @@ kernel void kernel_mul_mat_q4_0_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
         constant   int64_t & ne01[[buffer(4)]],
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
-    mul_vec_q_n_f32<block_q4_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne10,ne0,ne01,tgpig,tiisg,sgitg);
+    mul_vec_q_n_f32<block_q4_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,gqa,tgpig,tiisg,sgitg);
 }
 
 kernel void kernel_mul_mat_q4_1_f32(
@@ -494,13 +411,17 @@ kernel void kernel_mul_mat_q4_1_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
         constant   int64_t & ne01[[buffer(4)]],
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
-     mul_vec_q_n_f32<block_q4_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne10,ne0,ne01,tgpig,tiisg,sgitg);
+     mul_vec_q_n_f32<block_q4_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,gqa,tgpig,tiisg,sgitg);
 }
 
 kernel void kernel_mul_mat_f16_f32(
@@ -869,354 +790,6 @@ static inline uchar4 get_scale_min_k4(int j, device const uint8_t * q) {
     return r;
 }
 
-//========================================== dequantization =============================
-
-static void dequantize_row_q2_K(device const block_q2_K * x, device float * y, int k) {
-    assert(k % QK_K == 0);
-    const int nb = k / QK_K;
-
-    for (int i = 0; i < nb; i++) {
-
-        const float d = x[i].d;
-        const float min = x[i].dmin;
-
-        device const uint8_t * q = x[i].qs;
-
-#if QK_K == 256
-        int is = 0;
-        float dl, ml;
-        for (int n = 0; n < QK_K; n += 128) {
-            int shift = 0;
-            for (int j = 0; j < 4; ++j) {
-
-                uint8_t sc = x[i].scales[is++];
-                dl = d * (sc & 0xF); ml = min * (sc >> 4);
-                for (int l = 0; l < 16; ++l) *y++ = dl * ((int8_t)((q[l] >> shift) & 3)) - ml;
-
-                sc = x[i].scales[is++];
-                dl = d * (sc & 0xF); ml = min * (sc >> 4);
-                for (int l = 0; l < 16; ++l) *y++ = dl * ((int8_t)((q[l+16] >> shift) & 3)) - ml;
-
-                shift += 2;
-            }
-            q += 32;
-        }
-#else
-        float dl1 = d * (x[i].scales[0] & 0xF), ml1 = min * (x[i].scales[0] >> 4);
-        float dl2 = d * (x[i].scales[1] & 0xF), ml2 = min * (x[i].scales[1] >> 4);
-        float dl3 = d * (x[i].scales[2] & 0xF), ml3 = min * (x[i].scales[2] >> 4);
-        float dl4 = d * (x[i].scales[3] & 0xF), ml4 = min * (x[i].scales[3] >> 4);
-        for (int l = 0; l < 16; ++l) {
-            y[l+ 0] = dl1 * ((q[l] >> 0) & 3) - ml1;
-            y[l+16] = dl2 * ((q[l] >> 2) & 3) - ml2;
-            y[l+32] = dl3 * ((q[l] >> 4) & 3) - ml3;
-            y[l+48] = dl4 * ((q[l] >> 6) & 3) - ml4;
-        }
-        y += QK_K;
-#endif
-
-    }
-}
-
-static void dequantize_row_q3_K(device const block_q3_K * x, device float * y, int k) {
-    assert(k % QK_K == 0);
-    const int nb = k / QK_K;
-
-#if QK_K == 256
-
-    const uint16_t kmask1 = 0x0303;
-    const uint16_t kmask2 = 0x0f0f;
-
-    uint16_t aux[8];
-    thread const int8_t * scales = (thread const int8_t*)aux;
-
-    for (int i = 0; i < nb; i++) {
-
-        const float d_all = (float)(x[i].d);
-
-        device const uint8_t * q = x[i].qs;
-        device const uint8_t * h = x[i].hmask;
-        uint8_t m = 1;
-
-        device const uint16_t * a = (device const uint16_t *)x[i].scales;
-        aux[0] = (a[0] & kmask2) | (((a[4] >> 0) & kmask1) << 4);
-        aux[1] = (a[1] & kmask2) | (((a[5] >> 0) & kmask1) << 4);
-        aux[2] = (a[2] & kmask2) | (((a[4] >> 2) & kmask1) << 4);
-        aux[3] = (a[3] & kmask2) | (((a[5] >> 2) & kmask1) << 4);
-        aux[4] = ((a[0] >> 4) & kmask2) | (((a[4] >> 4) & kmask1) << 4);
-        aux[5] = ((a[1] >> 4) & kmask2) | (((a[5] >> 4) & kmask1) << 4);
-        aux[6] = ((a[2] >> 4) & kmask2) | (((a[4] >> 6) & kmask1) << 4);
-        aux[7] = ((a[3] >> 4) & kmask2) | (((a[5] >> 6) & kmask1) << 4);
-
-        int is = 0;
-        float dl;
-        for (int n = 0; n < QK_K; n += 128) {
-            int shift = 0;
-            for (int j = 0; j < 4; ++j) {
-
-                dl = d_all * (scales[is++] - 32);
-                for (int l = 0; l < 16; ++l) {
-                    *y++ = dl * ((int8_t)((q[l+ 0] >> shift) & 3) - ((h[l+ 0] & m) ? 0 : 4));
-                }
-
-                dl = d_all * (scales[is++] - 32);
-                for (int l = 0; l < 16; ++l) {
-                    *y++ = dl * ((int8_t)((q[l+16] >> shift) & 3) - ((h[l+16] & m) ? 0 : 4));
-                }
-
-                shift += 2;
-                m <<= 1;
-            }
-            q += 32;
-        }
-    }
-#else
-    for (int i = 0; i < nb; i++) {
-
-        const float d_all = (float)(x[i].d);
-
-        device const uint8_t * q = x[i].qs;
-        device const uint8_t * hm = x[i].hmask;
-
-        const float d1 = d_all * ((x[i].scales[0] & 0xF) - 8);
-        const float d2 = d_all * ((x[i].scales[0] >>  4) - 8);
-        const float d3 = d_all * ((x[i].scales[1] & 0xF) - 8);
-        const float d4 = d_all * ((x[i].scales[1] >>  4) - 8);
-
-        for (int l = 0; l < 8; ++l) {
-            uint8_t h = hm[l];
-            y[l+ 0] = d1 * ((int8_t)((q[l+0] >> 0) & 3) - ((h & 0x01) ? 0 : 4));
-            y[l+ 8] = d1 * ((int8_t)((q[l+8] >> 0) & 3) - ((h & 0x02) ? 0 : 4));
-            y[l+16] = d2 * ((int8_t)((q[l+0] >> 2) & 3) - ((h & 0x04) ? 0 : 4));
-            y[l+24] = d2 * ((int8_t)((q[l+8] >> 2) & 3) - ((h & 0x08) ? 0 : 4));
-            y[l+32] = d3 * ((int8_t)((q[l+0] >> 4) & 3) - ((h & 0x10) ? 0 : 4));
-            y[l+40] = d3 * ((int8_t)((q[l+8] >> 4) & 3) - ((h & 0x20) ? 0 : 4));
-            y[l+48] = d4 * ((int8_t)((q[l+0] >> 6) & 3) - ((h & 0x40) ? 0 : 4));
-            y[l+56] = d4 * ((int8_t)((q[l+8] >> 6) & 3) - ((h & 0x80) ? 0 : 4));
-        }
-        y += QK_K;
-    }
-#endif
-
-}
-
-static void dequantize_row_q4_K(device const block_q4_K * x, device float * y, int k) {
-    assert(k % QK_K == 0);
-    const int nb = k / QK_K;
-
-    for (int i = 0; i < nb; i++) {
-
-        device const uint8_t * q = x[i].qs;
-
-#if QK_K == 256
-        const float d = x[i].d;
-        const float min = x[i].dmin;
-
-        device const uint8_t * scales = x[i].scales;
-
-        int is = 0;
-        for (int j = 0; j < QK_K; j += 64) {
-            const uchar4 sc = get_scale_min_k4(is, scales);
-            const float d1 = d * sc[0]; const float m1 = min * sc[1];
-            const float d2 = d * sc[2]; const float m2 = min * sc[3];
-            for (int l = 0; l < 32; ++l) *y++ = d1 * (q[l] & 0xF) - m1;
-            for (int l = 0; l < 32; ++l) *y++ = d2 * (q[l]  >> 4) - m2;
-            q += 32; is += 2;
-        }
-#else
-        device const uint8_t * s = x[i].scales;
-        device const half2 * dh = (device const half2 *)x[i].d;
-        const float2 d = (float2)dh[0];
-        const float d1 = d[0] * (s[0] & 0xF);
-        const float d2 = d[0] * (s[1] & 0xF);
-        const float m1 = d[1] * (s[0] >>  4);
-        const float m2 = d[1] * (s[1] >>  4);
-        for (int l = 0; l < 32; ++l) {
-            y[l+ 0] = d1 * (q[l] & 0xF) - m1;
-            y[l+32] = d2 * (q[l] >>  4) - m2;
-        }
-        y += QK_K;
-#endif
-
-    }
-}
-
-static void dequantize_row_q5_K(device const block_q5_K * x, device float * y, int k) {
-    assert(k % QK_K == 0);
-    const int nb = k / QK_K;
-
-#if QK_K == 256
-   for (int i = 0; i < nb; i++) {
-
-        const float d = (float)(x[i].d);
-        const float min = (float)(x[i].dmin);
-
-        device const uint8_t * ql = x[i].qs;
-        device const uint8_t * qh = x[i].qh;
-
-        int is = 0;
-        uint8_t u1 = 1, u2 = 2;
-        for (int j = 0; j < QK_K; j += 64) {
-            const uchar4 sc = get_scale_min_k4(is, x[i].scales);
-            const float d1 = d * sc[0]; const float m1 = min * sc[1];
-            const float d2 = d * sc[2]; const float m2 = min * sc[3];
-            for (int l = 0; l < 32; ++l) *y++ = d1 * ((ql[l] & 0xF) + (qh[l] & u1 ? 16 : 0)) - m1;
-            for (int l = 0; l < 32; ++l) *y++ = d2 * ((ql[l]  >> 4) + (qh[l] & u2 ? 16 : 0)) - m2;
-            ql += 32; is += 2;
-            u1 <<= 2; u2 <<= 2;
-        }
-    }
-#else
-    for (int i = 0; i < nb; i++) {
-
-        const float d = (float)x[i].d;
-
-        device const uint8_t * ql = x[i].qs;
-        device const uint8_t * qh = x[i].qh;
-        device const int8_t  * sc = x[i].scales;
-
-        for (int l = 0; l < 8; ++l) {
-            y[l+ 0] = d * sc[0] * ((ql[l+ 0] & 0xF) - (qh[l] & 0x01 ? 0 : 16));
-            y[l+ 8] = d * sc[0] * ((ql[l+ 8] & 0xF) - (qh[l] & 0x02 ? 0 : 16));
-            y[l+16] = d * sc[1] * ((ql[l+16] & 0xF) - (qh[l] & 0x04 ? 0 : 16));
-            y[l+24] = d * sc[1] * ((ql[l+24] & 0xF) - (qh[l] & 0x08 ? 0 : 16));
-            y[l+32] = d * sc[2] * ((ql[l+ 0] >>  4) - (qh[l] & 0x10 ? 0 : 16));
-            y[l+40] = d * sc[2] * ((ql[l+ 8] >>  4) - (qh[l] & 0x20 ? 0 : 16));
-            y[l+48] = d * sc[3] * ((ql[l+16] >>  4) - (qh[l] & 0x40 ? 0 : 16));
-            y[l+56] = d * sc[3] * ((ql[l+24] >>  4) - (qh[l] & 0x80 ? 0 : 16));
-        }
-        y += QK_K;
-    }
-#endif
-
-}
-
-static void dequantize_row_q6_K(device const block_q6_K * x, device float * y, int k) {
-    assert(k % QK_K == 0);
-    const int nb = k / QK_K;
-
-    for (int i = 0; i < nb; i++) {
-
-        device const uint8_t * ql = x[i].ql;
-        device const uint8_t * qh = x[i].qh;
-        device const int8_t  * sc = x[i].scales;
-
-        const float d = x[i].d;
-
-#if QK_K == 256
-        for (int n = 0; n < QK_K; n += 128) {
-            for (int l = 0; l < 32; ++l) {
-                int is = l/16;
-                const int8_t q1 = (int8_t)((ql[l +  0] & 0xF) | (((qh[l] >> 0) & 3) << 4)) - 32;
-                const int8_t q2 = (int8_t)((ql[l + 32] & 0xF) | (((qh[l] >> 2) & 3) << 4)) - 32;
-                const int8_t q3 = (int8_t)((ql[l +  0]  >> 4) | (((qh[l] >> 4) & 3) << 4)) - 32;
-                const int8_t q4 = (int8_t)((ql[l + 32]  >> 4) | (((qh[l] >> 6) & 3) << 4)) - 32;
-                y[l +  0] = d * sc[is + 0] * q1;
-                y[l + 32] = d * sc[is + 2] * q2;
-                y[l + 64] = d * sc[is + 4] * q3;
-                y[l + 96] = d * sc[is + 6] * q4;
-            }
-            y  += 128;
-            ql += 64;
-            qh += 32;
-            sc += 8;
-        }
-#else
-        for (int l = 0; l < 16; ++l) {
-            const int8_t q1 = (int8_t)((ql[l+ 0] & 0xF) | (((qh[l] >> 0) & 3) << 4)) - 32;
-            const int8_t q2 = (int8_t)((ql[l+16] & 0xF) | (((qh[l] >> 2) & 3) << 4)) - 32;
-            const int8_t q3 = (int8_t)((ql[l+ 0]  >> 4) | (((qh[l] >> 4) & 3) << 4)) - 32;
-            const int8_t q4 = (int8_t)((ql[l+16]  >> 4) | (((qh[l] >> 6) & 3) << 4)) - 32;
-            y[l+ 0] = d * sc[0] * q1;
-            y[l+16] = d * sc[1] * q2;
-            y[l+32] = d * sc[2] * q3;
-            y[l+48] = d * sc[3] * q4;
-        }
-        y  += 64;
-#endif
-    }
-}
-
-kernel void kernel_get_rows_q2_K(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q2_K(
-            (device const block_q2_K *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
-kernel void kernel_get_rows_q3_K(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q3_K(
-            (device const block_q3_K *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
-kernel void kernel_get_rows_q4_K(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q4_K(
-            (device const block_q4_K *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
-kernel void kernel_get_rows_q5_K(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q5_K(
-            (device const block_q5_K *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
-kernel void kernel_get_rows_q6_K(
-        device const  void * src0,
-        device const   int * src1,
-        device       float * dst,
-        constant   int64_t & ne00,
-        constant  uint64_t & nb01,
-        constant  uint64_t & nb1,
-        uint tpig[[thread_position_in_grid]]) {
-    const int i = tpig;
-    const int r = ((device int32_t *) src1)[i];
-
-    dequantize_row_q6_K(
-            (device const block_q6_K *) ((device char *) src0 + r*nb01),
-                       (device float *) ((device char *)  dst + i*nb1), ne00);
-}
-
 //====================================== dot products =========================
 
 kernel void kernel_mul_mat_q2_K_f32(
@@ -1224,21 +797,27 @@ kernel void kernel_mul_mat_q2_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
         constant   int64_t & ne01[[buffer(4)]],
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
     const int nb = ne00/QK_K;
     const int r0 = tgpig.x;
     const int r1 = tgpig.y;
+    const int r2 = tgpig.z;
 
     const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
     const int ib_row = first_row * nb;
-    device const block_q2_K * x = (device const block_q2_K *) src0 + ib_row;
-    device const float      * y = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q2_K * x = (device const block_q2_K *) src0 + ib_row + offset0;
+    device const float      * y = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
     float yl[32];
     float sumf[N_DST]={0.f}, all_sum;
 
@@ -1351,7 +930,7 @@ kernel void kernel_mul_mat_q2_K_f32(
     for (int row = 0; row < N_DST; ++row) {
         all_sum = simd_sum(sumf[row]);
         if (tiisg == 0) {
-            dst[r1*ne0 + first_row + row] = all_sum;
+            dst[r1*ne0 + r2*ne0*ne1 + first_row + row] = all_sum;
         }
     }
 }
@@ -1362,10 +941,14 @@ kernel void kernel_mul_mat_q3_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
-        constant   int64_t & ne1,
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne01[[buffer(4)]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1373,11 +956,12 @@ kernel void kernel_mul_mat_q3_K_f32(
 
     const int64_t r0 = tgpig.x;
     const int64_t r1 = tgpig.y;
+    const int64_t r2 = tgpig.z;
 
     const int first_row = (r0 * N_SIMDGROUP + sgitg) * 2;
-
-    device const block_q3_K * x = (device const block_q3_K *) src0 + first_row*nb;
-    device const float     * yy = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q3_K * x = (device const block_q3_K *) src0 + first_row*nb + offset0;
+    device const float     * yy = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
 
     float yl[16];
 
@@ -1465,7 +1049,7 @@ kernel void kernel_mul_mat_q3_K_f32(
         const float sumf = (sumf1[row] - 32.f*sumf2[row]) / (1 << shift);
         const float tot = simd_sum(sumf);
         if (tiisg == 0) {
-            dst[r1*ne0 + first_row + row] = tot;
+            dst[r1*ne0 + r2*ne0*ne1 + first_row + row] = tot;
         }
     }
 }
@@ -1475,10 +1059,14 @@ kernel void kernel_mul_mat_q3_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
-        constant   int64_t & ne1,
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne01[[buffer(4)]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1486,11 +1074,12 @@ kernel void kernel_mul_mat_q3_K_f32(
 
     const int64_t r0 = tgpig.x;
     const int64_t r1 = tgpig.y;
+    const int64_t r2 = tgpig.z;
 
     const int row = 2 * r0 + sgitg;
-
-    device const block_q3_K * x = (device const block_q3_K *) src0 + row*nb;
-    device const float     * yy = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q3_K * x = (device const block_q3_K *) src0 + row*nb + offset0;
+    device const float     * yy = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
     const int ix = tiisg/4;
     const int il = 4 * (tiisg%4);// 0, 4, 8, 12
     const int im = il/8;         // 0, 0, 1, 1
@@ -1529,7 +1118,7 @@ kernel void kernel_mul_mat_q3_K_f32(
 
     const float tot = simd_sum(sumf);
     if (tiisg == 0) {
-        dst[r1*ne0 + row] = tot;
+        dst[r1*ne0 + r2*ne0*ne1 + row] = tot;
     }
 
 }
@@ -1541,10 +1130,14 @@ kernel void kernel_mul_mat_q4_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
         constant   int64_t & ne01[[buffer(4)]],
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1560,10 +1153,12 @@ kernel void kernel_mul_mat_q4_K_f32(
     const int nb = ne00/QK_K;
     const int r0 = tgpig.x;
     const int r1 = tgpig.y;
+    const int r2 = tgpig.z;
     const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
     const int ib_row = first_row * nb;
-    device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row;
-    device const float      * y = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row + offset0;
+    device const float      * y = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
     float yl[16];
     float yh[16];
     float sumf[N_DST]={0.f}, all_sum;
@@ -1630,7 +1225,7 @@ kernel void kernel_mul_mat_q4_K_f32(
     for (int row = 0; row < N_DST; ++row) {
         all_sum = simd_sum(sumf[row]);
         if (tiisg == 0) {
-            dst[r1*ne0 + first_row + row] = all_sum;
+            dst[r1*ne0 + r2*ne0*ne1 + first_row + row] = all_sum;
         }
     }
 }
@@ -1640,10 +1235,14 @@ kernel void kernel_mul_mat_q4_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
         constant   int64_t & ne01[[buffer(4)]],
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1653,10 +1252,12 @@ kernel void kernel_mul_mat_q4_K_f32(
     const int nb = ne00/QK_K;
     const int r0 = tgpig.x;
     const int r1 = tgpig.y;
+    const int r2 = tgpig.z;
     const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
     const int ib_row = first_row * nb;
-    device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row;
-    device const float      * y = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row + offset0;
+    device const float      * y = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
     float yl[8];
     float yh[8];
     float sumf[N_DST]={0.f}, all_sum;
@@ -1712,7 +1313,7 @@ kernel void kernel_mul_mat_q4_K_f32(
     for (int row = 0; row < N_DST; ++row) {
         all_sum = simd_sum(sumf[row]);
         if (tiisg == 0) {
-            dst[r1*ne0 + first_row + row] = all_sum;
+            dst[r1*ne0+ r2*ne0*ne1 + first_row + row] = all_sum;
         }
     }
 }
@@ -1723,9 +1324,14 @@ kernel void kernel_mul_mat_q5_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne01[[buffer(4)]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1733,11 +1339,12 @@ kernel void kernel_mul_mat_q5_K_f32(
 
     const int64_t r0 = tgpig.x;
     const int64_t r1 = tgpig.y;
+    const int r2 = tgpig.z;
 
     const int first_row = (r0 * N_SIMDGROUP + sgitg) * 2;
-
-    device const block_q5_K * x = (device const block_q5_K *) src0 + first_row*nb;
-    device const float     * yy = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q5_K * x = (device const block_q5_K *) src0 + first_row*nb + offset0;
+    device const float     * yy = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
 
     float sumf[2]={0.f};
 
@@ -1871,7 +1478,7 @@ kernel void kernel_mul_mat_q5_K_f32(
     for (int row = 0; row < 2; ++row) {
         const float tot = simd_sum(sumf[row]);
         if (tiisg == 0) {
-            dst[r1*ne0 + first_row + row] = tot;
+            dst[r1*ne0 + r2*ne0*ne1 + first_row + row] = tot;
         }
     }
 
@@ -1882,9 +1489,14 @@ kernel void kernel_mul_mat_q6_K_f32(
         device const float * src1,
         device       float * dst,
         constant   int64_t & ne00,
-        constant   int64_t & ne10,
-        constant   int64_t & ne0,
-        uint2 tgpig[[threadgroup_position_in_grid]],
+        constant   int64_t & ne01[[buffer(4)]],
+        constant   int64_t & ne02[[buffer(5)]],
+        constant   int64_t & ne10[[buffer(9)]],
+        constant   int64_t & ne12[[buffer(11)]],
+        constant   int64_t & ne0[[buffer(15)]],
+        constant   int64_t & ne1[[buffer(16)]],
+        constant   uint    & gqa[[buffer(17)]],
+        uint3 tgpig[[threadgroup_position_in_grid]],
         uint tiisg[[thread_index_in_simdgroup]],
         uint sgitg[[simdgroup_index_in_threadgroup]]) {
 
@@ -1897,11 +1509,12 @@ kernel void kernel_mul_mat_q6_K_f32(
 
     const int64_t r0 = tgpig.x;
     const int64_t r1 = tgpig.y;
+    const int r2 = tgpig.z;
 
     const int row = 2 * r0 + sgitg;
-
-    device const block_q6_K * x = (device const block_q6_K *) src0 + row * nb; //r0*nb;
-    device const float     * yy = (device const float      *) src1 + r1*ne10;
+    const uint offset0 = r2/gqa*(nb*ne0);
+    device const block_q6_K * x = (device const block_q6_K *) src0 + row * nb + offset0;
+    device const float     * yy = (device const float      *) src1 + r1*ne10 + r2*ne00*ne1;
 
     float sumf = 0;
 
@@ -1967,6 +1580,366 @@ kernel void kernel_mul_mat_q6_K_f32(
 
     const float tot = simd_sum(sumf);
     if (tiisg == 0) {
-        dst[r1*ne0 + row] = tot;
+        dst[r1*ne0 + r2*ne0*ne1 + row] = tot;
+    }
+}
+
+//============================= templates and their specializations =============================
+
+template <typename type4x4>
+void dequantize_f16(device const half4x4 * src, short il, thread type4x4 & reg) {
+    half4x4 temp = *(((device half4x4 *)src));
+    for (int i = 0; i < 16; i++){
+        reg[i/4][i%4] = temp[i/4][i%4];
+    }
+}
+
+template <typename type4x4>
+void dequantize_q4_0(device const block_q4_0 *xb, short il, thread type4x4 & reg) {
+    device const uint16_t * qs = ((device const uint16_t *)xb + 1);
+    const half d = il ? (xb->d / 16.h) : xb->d;
+    const half m = il ? (-8.h * 16.h) : -8.h;
+    const ushort mask0 = il ? 0x00F0 : 0x000F;
+    const ushort mask1 = il ? 0xF000 : 0x0F00;
+
+    for (int i=0;i<8;i++) {
+        reg[i/2][2*(i%2)] = (((qs[i] & mask0)) + m) * d;
+        reg[i/2][2*(i%2)+1] = (((qs[i] & mask1) >> 8) + m) * d;
+    }
+}
+
+template <typename type4x4>
+void dequantize_q4_1(device const block_q4_1 *xb, short il, thread type4x4 & reg) {
+    device const uint16_t * qs = ((device const uint16_t *)xb + 2);
+    const half d = il ? (xb->d / 16.h) : xb->d;
+    const half m = xb->m;
+    const ushort mask0 = il ? 0x00F0 : 0x000F;
+    const ushort mask1 = il ? 0xF000 : 0x0F00;
+
+    for (int i=0;i<8;i++) {
+        reg[i/2][2*(i%2)] = (((qs[i] & mask0)) * d) + m;
+        reg[i/2][2*(i%2)+1] = (((qs[i] & mask1) >> 8) * d) + m;
+    }
+}
+
+template <typename type4x4>
+void dequantize_q2_K(device const block_q2_K *xb, short il, thread type4x4 & reg) {
+    const half d = xb->d;
+    const half min = xb->dmin;
+    device const uint8_t * q = (device const uint8_t *)xb->qs;
+    half dl, ml;
+    uint8_t sc = xb->scales[il];
+
+#if QK_K == 256
+    q = q + 32*(il/8) + 16*(il&1);
+    il = (il/2)%4;
+#endif
+    half  coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
+    uchar mask = il>1 ? (il>2 ? 192    : 48)     : (il>0 ? 12    : 3);
+    dl = d * (sc & 0xF) * coef, ml = min * (sc >> 4);
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = dl * (q[i] & mask) - ml;
+    }
+}
+
+template <typename type4x4>
+void dequantize_q3_K(device const block_q3_K *xb, short il, thread type4x4 & reg) {
+    const float d_all = (float)(xb->d);
+    device const uint8_t * q = (device const uint8_t *)xb->qs;
+    device const uint8_t * h = (device const uint8_t *)xb->hmask;
+    device const int8_t * scales = (device const int8_t *)xb->scales;
+
+#if QK_K == 256
+    q = q + 32 * (il/8) + 16 * (il&1);
+    h = h + 16 * (il&1);
+    uint8_t m = 1 << (il/2);
+    uint16_t kmask1 = (il/4)>1 ? ((il/4)>2 ? 192 : 48) : \
+                                 ((il/4)>0 ? 12  : 3);
+    uint16_t kmask2 = il/8 ? 0xF0 : 0x0F;
+    uint16_t scale_2 = scales[il%8], scale_1 = scales[8 + il%4];
+    int16_t  dl_int = (il/4)&1 ? (scale_2&kmask2) | ((scale_1&kmask1) << 2) : \
+                                 (scale_2&kmask2) | ((scale_1&kmask1) << 4);
+    float dl = il<8 ? d_all * (dl_int - 32.f) : d_all * (dl_int / 16.f - 32.f);
+
+    il = (il/2)%4;
+    float   coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
+    uint8_t mask = il>1 ? (il>2 ? 192    : 48)     : (il>0 ? 12    : 3);
+
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = coef * dl * ((q[i] & mask) - ((h[i] & m) ? 0 : 4.f/coef));
+    }
+#else
+    float    kcoef = il&1 ? 1.f/16.f : 1.f;
+    uint16_t kmask = il&1 ? 0xF0     : 0x0F;
+    float    dl = d_all * ((scales[il/2] & kmask) * kcoef - 8);
+    float    coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
+    uint8_t  mask = il>1 ? (il>2 ? 192    : 48)     : (il>0 ? 12    : 3);
+    uint8_t  m = 1<<(il*2);
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = coef * dl * ((q[i] & mask) - ((h[i%8] & (m * (1 + i/8))) ? 0 : 4.f/coef));
+    }
+#endif
+}
+
+template <typename type4x4>
+void dequantize_q4_K(device const block_q4_K *xb, short il, thread type4x4 & reg) {
+    device const uint8_t * q = xb->qs;
+
+#if QK_K == 256
+    const float d = (float)(xb->d);
+    const float min = (float)(xb->dmin);
+    short is = (il/4) * 2;
+    q = q + (il/4) * 32 + 16 * (il&1);
+    il = il%4;
+    const uchar4 sc = get_scale_min_k4(is, xb->scales);
+    const float dl = il<2 ? d * sc[0]   : d * sc[2]/16.h;
+    const float ml = il<2 ? min * sc[1] : min * sc[3];
+#else
+    q = q + 16 * (il&1);
+    device const uint8_t * s = xb->scales;
+    device const half2 * dh = (device const half2 *)xb->d;
+    const float2 d = (float2)dh[0];
+    const float dl = il<2 ? d[0] * (s[0]&0xF) : d[0] * (s[1]&0xF)/16.h;
+    const float ml = il<2 ? d[1] * (s[0]>>4)  : d[1 ]* (s[1]>>4);
+#endif
+    const ushort mask = il<2 ? 0x0F : 0xF0;
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = dl * (q[i] & mask) - ml;
+    }
+}
+
+template <typename type4x4>
+void dequantize_q5_K(device const block_q5_K *xb, short il, thread type4x4 & reg) {
+    device const uint8_t * q  = xb->qs;
+    device const uint8_t * qh = xb->qh;
+
+#if QK_K == 256
+    const float d = (float)(xb->d);
+    const float min = (float)(xb->dmin);
+    short is = (il/4) * 2;
+    q  = q + 32 * (il/4) + 16 * (il&1);
+    qh = qh + 16 * (il&1);
+    uint8_t ul = 1 << (il/2);
+    il = il%4;
+    const uchar4 sc = get_scale_min_k4(is, xb->scales);
+    const float dl = il<2 ? d * sc[0]   : d * sc[2]/16.h;
+    const float ml = il<2 ? min * sc[1] : min * sc[3];
+
+    const ushort mask   = il<2 ? 0x0F : 0xF0;
+    const float  qh_val = il<2 ? 16.f : 256.f;
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = dl * ((q[i] & mask) + (qh[i] & ul ? qh_val : 0)) - ml;
+    }
+#else
+    q = q + 16 * (il&1);
+    device const int8_t * s = xb->scales;
+    const float dl = xb->d * s[il];
+    uint8_t m = 1<<(il*2);
+    const float  coef = il<2 ? 1.f  : 1.f/16.f;
+    const ushort mask = il<2 ? 0x0F : 0xF0;
+    for (int i = 0; i < 16; ++i) {
+        reg[i/4][i%4] = coef * dl * ((q[i] & mask) - (qh[i%8] & (m*(1+i/8)) ? 0.f : 16.f/coef));
+    }
+#endif
+}
+
+template <typename type4x4>
+void dequantize_q6_K(device const block_q6_K *xb, short il, thread type4x4 & reg) {
+    const float d_all = (float)(xb->d);
+    device const uint8_t * ql = (device const uint8_t *)xb->ql;
+    device const uint8_t * qh = (device const uint8_t *)xb->qh;
+    device const int8_t * scales = (device const int8_t *)xb->scales;
+
+#if QK_K == 256
+    ql = ql + 64*(il/8) + 32*((il/2)&1) + 16*(il&1);
+    qh = qh + 32*(il/8) + 16*(il&1);
+    float sc = scales[(il%2) + 2 * ((il/2))];
+    il = (il/2)%4;
+#else
+    ql = ql + 16 * (il&1);
+    float sc = scales[il];
+#endif
+    for (int i = 0; i < 16; ++i) {
+        uint16_t  kmask1 = il>1 ? (il>2 ? 192 : 48) : (il>0 ? 12 : 3);
+        uint16_t  kmask2 = il>1 ? 0xF0              : 0x0F;
+        const float coef = il>1 ? 1.f/16.f          : 1.f;
+        float q = il&1 ? ((ql[i]&kmask2)|((qh[i]&kmask1)<<2)) - 32.f/coef : \
+                         ((ql[i]&kmask2)|((qh[i]&kmask1)<<4)) - 32.f/coef;
+        reg[i/4][i%4] = d_all * sc * q * coef;
+    }
+}
+
+template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread float4x4 &)>
+kernel void kernel_get_rows(
+        device const  void * src0,
+        device const   int * src1,
+        device       float * dst,
+        constant   int64_t & ne00,
+        constant  uint64_t & nb01,
+        constant  uint64_t & nb1,
+        uint                 tgpig[[threadgroup_position_in_grid]],
+        uint                 tiitg[[thread_index_in_threadgroup]],
+        uint                 tptg[[threads_per_threadgroup]]) {
+    const int i = tgpig;
+    const int r = ((device int32_t *) src1)[i];
+
+    for (int ind = tiitg; ind < ne00/16; ind += tptg) {
+        float4x4 temp;
+        dequantize_func(
+            ((device const block_q *) ((device char *) src0 + r*nb01)) + ind/nl, ind%nl, temp);
+        *(((device float4x4 *) ((device char *) dst + i*nb1)) + ind) = temp;
+    }
+}
+
+#define BLOCK_SIZE_M 64 // 8 simdgroup matrices from matrix A
+#define BLOCK_SIZE_N 32 // 4 simdgroup matrices from matrix A
+#define BLOCK_SIZE_K 32
+#define THREAD_MAT_M 4 // each thread take 4 simdgroup matrices from matrix A
+#define THREAD_MAT_N 2 // each thread take 2 simdgroup matrices from matrix B
+#define THREAD_PER_BLOCK 128
+#define THREAD_PER_ROW 2 // 2 thread for each row in matrix A to load numbers
+#define THREAD_PER_COL 4 // 4 thread for each row in matrix B to load numbers
+#define SG_MAT_SIZE 64 // simdgroup matrix is of shape 8x8
+#define SG_MAT_ROW 8
+
+// each block_q contains 16*nl weights
+template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
+kernel void kernel_mul_mm(device const  uchar * src0,
+                           device const  float * src1,
+                           device        float * dst,
+                           constant    int64_t & ne00,
+                           constant    int64_t & ne02,
+                           constant    int64_t & nb01,
+                           constant    int64_t & nb02,
+                           constant    int64_t & ne12,
+                           constant    int64_t & ne0,
+                           constant    int64_t & ne1,
+                           constant    uint & gqa,
+                           threadgroup   uchar * shared_memory [[threadgroup(0)]],
+                           uint3                 tgpig[[threadgroup_position_in_grid]],
+                           uint                  tiitg[[thread_index_in_threadgroup]],
+                           uint                  sgitg[[simdgroup_index_in_threadgroup]]) {
+
+    threadgroup half * sa = ((threadgroup half *)shared_memory);
+    threadgroup float * sb = (threadgroup float *)(shared_memory + 4096);
+
+    const uint r0 = tgpig.y;
+    const uint r1 = tgpig.x;
+    const uint im = tgpig.z;
+    // if this block is of 64x32 shape or smaller
+    short n_rows = (ne0 - r0 * BLOCK_SIZE_M < BLOCK_SIZE_M) ? (ne0 - r0 * BLOCK_SIZE_M) : BLOCK_SIZE_M;
+    short n_cols = (ne1 - r1 * BLOCK_SIZE_N < BLOCK_SIZE_N) ? (ne1 - r1 * BLOCK_SIZE_N) : BLOCK_SIZE_N;
+    // a thread shouldn't load data outside of the matrix
+    short thread_row = ((short)tiitg/THREAD_PER_ROW) < n_rows ? ((short)tiitg/THREAD_PER_ROW) : n_rows - 1;
+    short thread_col = ((short)tiitg/THREAD_PER_COL) < n_cols ? ((short)tiitg/THREAD_PER_COL) : n_cols - 1;
+
+    simdgroup_half8x8 ma[4];
+    simdgroup_float8x8 mb[2];
+    simdgroup_float8x8 c_res[8];
+    for (int i = 0; i < 8; i++){
+        c_res[i] = make_filled_simdgroup_matrix<float, 8>(0.f);
+    }
+
+    short il = (tiitg % THREAD_PER_ROW);
+    uint offset0 = im/gqa*nb02; ushort offset1 = il/nl;
+    device const block_q  * x = (device const block_q  *)(src0 + (r0 * BLOCK_SIZE_M + thread_row) * nb01 + offset0) + offset1;
+    device const float * y = src1 + (r1 * BLOCK_SIZE_N + thread_col) * ne00 \
+                             + BLOCK_SIZE_K / THREAD_PER_COL * (tiitg % THREAD_PER_COL) + im * ne00 * ne1;
+
+    for (int loop_k = 0; loop_k < ne00; loop_k += BLOCK_SIZE_K) {
+        //load data and store to threadgroup memory
+        half4x4 temp_a;
+        dequantize_func(x, il, temp_a);
+        #pragma unroll(16)
+        for (int i = 0; i < 16; i++) {
+            *(sa + SG_MAT_SIZE * ((tiitg / THREAD_PER_ROW / 8) \
+            + 16 * (tiitg % THREAD_PER_ROW) + 8 * (i / 8)) \
+            + (tiitg / THREAD_PER_ROW) % 8 + (i & 7) * 8) = temp_a[i/4][i%4];
+        }
+        *(threadgroup float2x4 *)(sb + (tiitg % THREAD_PER_COL) * 8 * 32 + 8 * (tiitg / THREAD_PER_COL)) \
+                = *((device float2x4 *)y);
+        il = (il + 2 < nl) ? il + 2 : il % 2;
+        x  = (il < 2) ? x + (2+nl-1)/nl : x;
+        y += BLOCK_SIZE_K;
+
+        threadgroup_barrier(mem_flags::mem_threadgroup);
+        //load matrices from threadgroup memory and conduct outer products
+        threadgroup half  * lsma = (sa + THREAD_MAT_M * SG_MAT_SIZE * (sgitg % 2));
+        threadgroup float * lsmb = (sb + THREAD_MAT_N * SG_MAT_SIZE * (sgitg / 2));
+        #pragma unroll(4)
+        for (int ik = 0; ik < BLOCK_SIZE_K / 8; ik++) {
+            #pragma unroll(4)
+            for (int i = 0; i < 4; i++) {
+                simdgroup_load(ma[i],lsma + SG_MAT_SIZE * i);
+            }
+            simdgroup_barrier(mem_flags::mem_none);
+            #pragma unroll(2)
+            for (int i = 0; i < 2; i++) {
+                simdgroup_load(mb[i],lsmb + SG_MAT_SIZE * i);
+            }
+
+            lsma += BLOCK_SIZE_M / SG_MAT_ROW * SG_MAT_SIZE;
+            lsmb += BLOCK_SIZE_N / SG_MAT_ROW * SG_MAT_SIZE;
+            #pragma unroll(8)
+            for (int i = 0; i < 8; i++){
+                simdgroup_multiply_accumulate(c_res[i], mb[i/4], ma[i%4], c_res[i]);
+            }
+        }
+    }
+
+    if ((r0 + 1) * BLOCK_SIZE_M <= ne0 && (r1 + 1) * BLOCK_SIZE_N <= ne1) {
+        device float *C = dst + BLOCK_SIZE_M * r0 + 32 * (sgitg&1) \
+                          + (BLOCK_SIZE_N * r1 + 16 * (sgitg>>1)) * ne0 + im*ne1*ne0;
+        for (int i = 0; i < 8; i++) {
+            simdgroup_store(c_res[i], C + 8 * (i%4) + 8 * ne0 * (i/4), ne0);
+        }
+    } else {
+        // block is smaller than 64x32, we should avoid writing data outside of the matrix
+        threadgroup float *temp_str = ((threadgroup float *)shared_memory) \
+                                      + 32 * (sgitg&1) + (16 * (sgitg>>1)) * BLOCK_SIZE_M;
+        for (int i = 0; i < 8; i++) {
+            simdgroup_store(c_res[i], temp_str + 8 * (i%4) + 8 * BLOCK_SIZE_M * (i/4), BLOCK_SIZE_M);
+        }
+
+        threadgroup_barrier(mem_flags::mem_threadgroup);
+        device float *C = dst + BLOCK_SIZE_M * r0 + (BLOCK_SIZE_N * r1) * ne0 + im*ne1*ne0;
+        if (sgitg==0) {
+            for (int i = 0; i < n_rows; i++) {
+                for (int j = tiitg; j< n_cols; j += BLOCK_SIZE_N) {
+                    *(C + i + j * ne0) = *(temp_str + i + j * BLOCK_SIZE_M);
+                }
+            }
+        }
     }
 }
+
+#if QK_K == 256
+#define QK_NL 16
+#else
+#define QK_NL 4
+#endif
+
+typedef void (get_rows_t)(device const void *, device const int *, device float *, constant int64_t &, \
+                          constant uint64_t &, constant uint64_t &, uint, uint, uint);
+
+template [[host_name("kernel_get_rows_f16")]] kernel get_rows_t kernel_get_rows<half4x4, 1, dequantize_f16>;
+template [[host_name("kernel_get_rows_q4_0")]] kernel get_rows_t kernel_get_rows<block_q4_0, 2, dequantize_q4_0>;
+template [[host_name("kernel_get_rows_q4_1")]] kernel get_rows_t kernel_get_rows<block_q4_1, 2, dequantize_q4_1>;
+template [[host_name("kernel_get_rows_q2_K")]] kernel get_rows_t kernel_get_rows<block_q2_K, QK_NL, dequantize_q2_K>;
+template [[host_name("kernel_get_rows_q3_K")]] kernel get_rows_t kernel_get_rows<block_q3_K, QK_NL, dequantize_q3_K>;
+template [[host_name("kernel_get_rows_q4_K")]] kernel get_rows_t kernel_get_rows<block_q4_K, QK_NL, dequantize_q4_K>;
+template [[host_name("kernel_get_rows_q5_K")]] kernel get_rows_t kernel_get_rows<block_q5_K, QK_NL, dequantize_q5_K>;
+template [[host_name("kernel_get_rows_q6_K")]] kernel get_rows_t kernel_get_rows<block_q6_K, QK_NL, dequantize_q6_K>;
+
+typedef void (mat_mm_t)(device const uchar *, device const float *, device float *, constant int64_t &,\
+                             constant int64_t &, constant int64_t &, constant int64_t &, constant int64_t &, \
+                             constant int64_t &, constant int64_t &, constant uint &, threadgroup uchar *, uint3, uint, uint);
+
+template [[host_name("kernel_mul_mm_f16_f32")]] kernel mat_mm_t kernel_mul_mm<half4x4, 1, dequantize_f16>;
+template [[host_name("kernel_mul_mm_q4_0_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_0, 2, dequantize_q4_0>;
+template [[host_name("kernel_mul_mm_q4_1_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_1, 2, dequantize_q4_1>;
+template [[host_name("kernel_mul_mm_q2_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q2_K, QK_NL, dequantize_q2_K>;
+template [[host_name("kernel_mul_mm_q3_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q3_K, QK_NL, dequantize_q3_K>;
+template [[host_name("kernel_mul_mm_q4_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_K, QK_NL, dequantize_q4_K>;
+template [[host_name("kernel_mul_mm_q5_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q5_K, QK_NL, dequantize_q5_K>;
+template [[host_name("kernel_mul_mm_q6_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q6_K, QK_NL, dequantize_q6_K>;
diff --git a/llama.cpp b/llama.cpp
index 0c3d412e941eb..1d1642bd8ae8c 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -11,7 +11,7 @@
 
 #include "ggml.h"
 
-#if !defined(GGML_USE_CUBLAS) && !defined(GGML_USE_METAL)
+#if !defined(GGML_USE_CUBLAS)
 #  include "ggml-alloc.h"
 #  define LLAMA_USE_ALLOCATOR
 #else
@@ -1895,11 +1895,11 @@ static struct ggml_cgraph * llama_build_graph(
             ggml_set_name(Q, "Q");
 
             struct ggml_tensor * K =
-                ggml_permute(ctx0,
-                        ggml_reshape_3d(ctx0,
-                            ggml_view_1d(ctx0, kv_self.k, (n_past + N)*n_embd_gqa, il*n_ctx*ggml_element_size(kv_self.k)*n_embd_gqa),
-                            n_embd_head, n_head_kv, n_past + N),
-                        0, 2, 1, 3);
+                ggml_view_3d(ctx0, kv_self.k,
+                        n_embd_head, n_past + N, n_head_kv,
+                        ggml_element_size(kv_self.k)*n_embd_gqa,
+                        ggml_element_size(kv_self.k)*n_embd_head,
+                        ggml_element_size(kv_self.k)*n_embd_gqa*n_ctx*il);
             offload_func_kq(K);
             ggml_set_name(K, "K");
 
@@ -1928,9 +1928,9 @@ static struct ggml_cgraph * llama_build_graph(
             struct ggml_tensor * V =
                 ggml_view_3d(ctx0, kv_self.v,
                         n_past + N, n_embd_head, n_head_kv,
-                        n_ctx*ggml_element_size(kv_self.v),
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_head,
-                        n_ctx*ggml_element_size(kv_self.v)*n_embd_gqa*il);
+                        ggml_element_size(kv_self.v)*n_ctx,
+                        ggml_element_size(kv_self.v)*n_ctx*n_embd_head,
+                        ggml_element_size(kv_self.v)*n_ctx*n_embd_gqa*il);
             offload_func_v(V);
             ggml_set_name(V, "V");
 
@@ -2131,11 +2131,7 @@ static bool llama_eval_internal(
 #endif
 
 #ifdef GGML_USE_METAL
-    if (lctx.ctx_metal && N == 1) {
-        // TODO: disabled until #2413 is resolved
-        //if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
-        //    ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
-        //}
+    if (lctx.ctx_metal) {
         ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
         ggml_metal_graph_compute(lctx.ctx_metal, gf);
         ggml_metal_get_tensor   (lctx.ctx_metal, res);
@@ -2143,22 +2139,6 @@ static bool llama_eval_internal(
             ggml_metal_get_tensor(lctx.ctx_metal, embeddings);
         }
     } else {
-        // IMPORTANT:
-        // Since we don't have efficient Matrix x Matrix Metal multiplication yet, we fallback to vanilla
-        // ggml_graph_compute(). It uses Apple's Accelerate CBLAS API which takes advantage of the ANE or the AMX
-        // coprocessor.
-        //
-        // When we implement Matrix x Matrix Metal multiplication, we can avoid this branch.
-        // But for now, we have focused only on Matrix x Vector Metal multiplication.
-        //
-        // TODO: avoid these syncs via shared memory (ref #1696)
-        //
-        if (lctx.ctx_metal) {
-            // We need to sync the GPU KV cache with the CPU KV cache
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.k);
-            ggml_metal_get_tensor(lctx.ctx_metal, kv_self.v);
-        }
-
         ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
     }
 #else
@@ -4141,7 +4121,18 @@ struct llama_context * llama_new_context_with_model(
             int n_past = hparams.n_ctx - n_tokens;
             llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
             ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
-
+#ifdef GGML_USE_METAL
+            if (params.n_gpu_layers > 0) {
+                ctx->ctx_metal = ggml_metal_init(1);
+                if (!ctx->ctx_metal) {
+                    LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
+                    llama_free(ctx);
+                    return NULL;
+                }
+                ggml_metal_graph_find_concurrency(ctx->ctx_metal, gf, false);
+                ggml_allocr_set_parse_seq(ctx->alloc, ggml_metal_get_concur_list(ctx->ctx_metal), ggml_metal_if_optimized(ctx->ctx_metal));
+            }
+#endif
             // measure memory requirements for the graph
             size_t alloc_size = ggml_allocr_alloc_graph(ctx->alloc, gf) + tensor_alignment;
 
@@ -4159,6 +4150,11 @@ struct llama_context * llama_new_context_with_model(
 
             ctx->buf_alloc.resize(alloc_size);
             ctx->alloc = ggml_allocr_new(ctx->buf_alloc.data, ctx->buf_alloc.size, tensor_alignment);
+#ifdef GGML_USE_METAL
+            if (ctx->ctx_metal) {
+                ggml_allocr_set_parse_seq(ctx->alloc, ggml_metal_get_concur_list(ctx->ctx_metal), ggml_metal_if_optimized(ctx->ctx_metal));
+            }
+#endif
         }
 #else
         ctx->buf_compute.resize(MEM_REQ_EVAL().at(ctx->model.type) + ggml_graph_overhead());
@@ -4173,13 +4169,6 @@ struct llama_context * llama_new_context_with_model(
 #ifdef GGML_USE_METAL
     if (params.n_gpu_layers > 0) {
         // this allocates all Metal resources and memory buffers
-        ctx->ctx_metal = ggml_metal_init(1);
-
-        if (!ctx->ctx_metal) {
-            LLAMA_LOG_ERROR("%s: ggml_metal_init() failed\n", __func__);
-            llama_free(ctx);
-            return NULL;
-        }
 
         void * data_ptr  = NULL;
         size_t data_size = 0;
@@ -4208,8 +4197,7 @@ struct llama_context * llama_new_context_with_model(
         LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "eval", ctx->buf_compute.data, ctx->buf_compute.size, 0));
         LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "kv",   ctx->kv_self.buf.data, ctx->kv_self.buf.size, 0));
 
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr0", ctx->buf_scratch[0].data, ctx->buf_scratch[0].size, 0));
-        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "scr1", ctx->buf_scratch[1].data, ctx->buf_scratch[1].size, 0));
+        LLAMA_METAL_CHECK_BUF(ggml_metal_add_buffer(ctx->ctx_metal, "alloc", ctx->buf_alloc.data, ctx->buf_alloc.size, 0));
 #undef LLAMA_METAL_CHECK_BUF
     }
 #endif
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index bc2100efb485c..d19ec5af0ef20 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -31,5 +31,6 @@ llama_build_executable(test-tokenizer-1.cpp)
 llama_test_executable(test-tokenizer-1.llama  test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
 #llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf)
 llama_build_and_test_executable(test-grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp)
+llama_build_and_test_executable(test-llama-grammar.cpp  ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../llama.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/common.cpp)
 llama_build_and_test_executable(test-grad0.cpp) # SLOW
 # llama_build_and_test_executable(test-opt.cpp) # SLOW

From c3b739374e90b4be579292bac88c4e1cbd86aa83 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:17:25 +0300
Subject: [PATCH 174/242] editorconfig : ignore models folder

ggml-ci
---
 models/.editorconfig | 1 +
 1 file changed, 1 insertion(+)
 create mode 100644 models/.editorconfig

diff --git a/models/.editorconfig b/models/.editorconfig
new file mode 100644
index 0000000000000..78b36ca0838fc
--- /dev/null
+++ b/models/.editorconfig
@@ -0,0 +1 @@
+root = true

From dd9e2fc98892e90d8ef97ec5e1f502f885c560ec Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:32:14 +0300
Subject: [PATCH 175/242] ci : update ".bin" to ".gguf" extension

ggml-ci
---
 README.md                                 | 22 ++++++------
 ci/run.sh                                 | 44 +++++++++++------------
 docs/token_generation_performance_tips.md |  6 ++--
 examples/common.h                         |  2 +-
 examples/metal/metal.cpp                  |  2 +-
 examples/server/README.md                 |  6 ++--
 llama.cpp                                 |  2 +-
 7 files changed, 42 insertions(+), 42 deletions(-)

diff --git a/README.md b/README.md
index 6900b1152e736..a24e91653f1e2 100644
--- a/README.md
+++ b/README.md
@@ -284,7 +284,7 @@ When built with Metal support, you can enable GPU inference with the `--gpu-laye
 Any value larger than 0 will offload the computation to the GPU. For example:
 
 ```bash
-./main -m ./models/7B/ggml-model-q4_0.bin -n 128 -ngl 1
+./main -m ./models/7B/ggml-model-q4_0.gguf -n 128 -ngl 1
 ```
 
 ### MPI Build
@@ -323,7 +323,7 @@ The above will distribute the computation across 2 processes on the first host a
 Finally, you're ready to run a computation using `mpirun`:
 
 ```bash
-mpirun -hostfile hostfile -n 3 ./main -m ./models/7B/ggml-model-q4_0.bin -n 128
+mpirun -hostfile hostfile -n 3 ./main -m ./models/7B/ggml-model-q4_0.gguf -n 128
 ```
 
 ### BLAS Build
@@ -506,10 +506,10 @@ python3 convert.py models/7B/
   python convert.py models/7B/ --vocabtype bpe
 
 # quantize the model to 4-bits (using q4_0 method)
-./quantize ./models/7B/ggml-model-f16.bin ./models/7B/ggml-model-q4_0.bin q4_0
+./quantize ./models/7B/ggml-model-f16.gguf ./models/7B/ggml-model-q4_0.gguf q4_0
 
 # run the inference
-./main -m ./models/7B/ggml-model-q4_0.bin -n 128
+./main -m ./models/7B/ggml-model-q4_0.gguf -n 128
 ```
 
 When running the larger models, make sure you have enough disk space to store all the intermediate files.
@@ -565,7 +565,7 @@ Here is an example of a few-shot interaction, invoked with the command
 ./examples/chat-13B.sh
 
 # custom arguments using a 13B model
-./main -m ./models/13B/ggml-model-q4_0.bin -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
+./main -m ./models/13B/ggml-model-q4_0.gguf -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
 ```
 
 Note the use of `--color` to distinguish between user input and generated text. Other parameters are explained in more detail in the [README](examples/main/README.md) for the `main` example program.
@@ -628,6 +628,8 @@ OpenLLaMA is an openly licensed reproduction of Meta's original LLaMA model. It
 
 ### Using [GPT4All](https://github.com/nomic-ai/gpt4all)
 
+*Note: these instructions are likely obsoleted by the GGUF update*
+
 - Obtain the `tokenizer.model` file from LLaMA model and put it to `models`
 - Obtain the `added_tokens.json` file from Alpaca model and put it to `models`
 - Obtain the `gpt4all-lora-quantized.bin` file from GPT4All model and put it to `models/gpt4all-7B`
@@ -703,7 +705,7 @@ If your issue is with model generation quality, then please at least scan the fo
 #### How to run
 
 1. Download/extract: https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip?ref=salesforce-research
-2. Run `./perplexity -m models/7B/ggml-model-q4_0.bin -f wiki.test.raw`
+2. Run `./perplexity -m models/7B/ggml-model-q4_0.gguf -f wiki.test.raw`
 3. Output:
 ```
 perplexity : calculating perplexity over 655 chunks
@@ -802,13 +804,13 @@ docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --all-in-
 On completion, you are ready to play!
 
 ```bash
-docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.bin -p "Building a website can be done in 10 simple steps:" -n 512
+docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512
 ```
 
 or with a light image:
 
 ```bash
-docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:light -m /models/7B/ggml-model-q4_0.bin -p "Building a website can be done in 10 simple steps:" -n 512
+docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:light -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512
 ```
 
 ### Docker With CUDA
@@ -839,8 +841,8 @@ The resulting images, are essentially the same as the non-CUDA images:
 After building locally, Usage is similar to the non-CUDA examples, but you'll need to add the `--gpus` flag. You will also want to use the `--n-gpu-layers` flag.
 
 ```bash
-docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.bin -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
-docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.bin -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
+docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
+docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1
 ```
 
 ### Contributing
diff --git a/ci/run.sh b/ci/run.sh
index 8dc3949648ff9..54ba6d7100e8f 100644
--- a/ci/run.sh
+++ b/ci/run.sh
@@ -159,17 +159,17 @@ function gg_run_open_llama_3b_v2 {
 
     python3 ../convert.py ${path_models}
 
-    model_f16="${path_models}/ggml-model-f16.bin"
-    model_q8_0="${path_models}/ggml-model-q8_0.bin"
-    model_q4_0="${path_models}/ggml-model-q4_0.bin"
-    model_q4_1="${path_models}/ggml-model-q4_1.bin"
-    model_q5_0="${path_models}/ggml-model-q5_0.bin"
-    model_q5_1="${path_models}/ggml-model-q5_1.bin"
-    model_q2_k="${path_models}/ggml-model-q2_k.bin"
-    model_q3_k="${path_models}/ggml-model-q3_k.bin"
-    model_q4_k="${path_models}/ggml-model-q4_k.bin"
-    model_q5_k="${path_models}/ggml-model-q5_k.bin"
-    model_q6_k="${path_models}/ggml-model-q6_k.bin"
+    model_f16="${path_models}/ggml-model-f16.gguf"
+    model_q8_0="${path_models}/ggml-model-q8_0.gguf"
+    model_q4_0="${path_models}/ggml-model-q4_0.gguf"
+    model_q4_1="${path_models}/ggml-model-q4_1.gguf"
+    model_q5_0="${path_models}/ggml-model-q5_0.gguf"
+    model_q5_1="${path_models}/ggml-model-q5_1.gguf"
+    model_q2_k="${path_models}/ggml-model-q2_k.gguf"
+    model_q3_k="${path_models}/ggml-model-q3_k.gguf"
+    model_q4_k="${path_models}/ggml-model-q4_k.gguf"
+    model_q5_k="${path_models}/ggml-model-q5_k.gguf"
+    model_q6_k="${path_models}/ggml-model-q6_k.gguf"
 
     wiki_test_60="${path_wiki}/wiki.test-60.raw"
 
@@ -285,17 +285,17 @@ function gg_run_open_llama_7b_v2 {
 
     python3 ../convert.py ${path_models}
 
-    model_f16="${path_models}/ggml-model-f16.bin"
-    model_q8_0="${path_models}/ggml-model-q8_0.bin"
-    model_q4_0="${path_models}/ggml-model-q4_0.bin"
-    model_q4_1="${path_models}/ggml-model-q4_1.bin"
-    model_q5_0="${path_models}/ggml-model-q5_0.bin"
-    model_q5_1="${path_models}/ggml-model-q5_1.bin"
-    model_q2_k="${path_models}/ggml-model-q2_k.bin"
-    model_q3_k="${path_models}/ggml-model-q3_k.bin"
-    model_q4_k="${path_models}/ggml-model-q4_k.bin"
-    model_q5_k="${path_models}/ggml-model-q5_k.bin"
-    model_q6_k="${path_models}/ggml-model-q6_k.bin"
+    model_f16="${path_models}/ggml-model-f16.gguf"
+    model_q8_0="${path_models}/ggml-model-q8_0.gguf"
+    model_q4_0="${path_models}/ggml-model-q4_0.gguf"
+    model_q4_1="${path_models}/ggml-model-q4_1.gguf"
+    model_q5_0="${path_models}/ggml-model-q5_0.gguf"
+    model_q5_1="${path_models}/ggml-model-q5_1.gguf"
+    model_q2_k="${path_models}/ggml-model-q2_k.gguf"
+    model_q3_k="${path_models}/ggml-model-q3_k.gguf"
+    model_q4_k="${path_models}/ggml-model-q4_k.gguf"
+    model_q5_k="${path_models}/ggml-model-q5_k.gguf"
+    model_q6_k="${path_models}/ggml-model-q6_k.gguf"
 
     wiki_test="${path_wiki}/wiki.test.raw"
 
diff --git a/docs/token_generation_performance_tips.md b/docs/token_generation_performance_tips.md
index 69ba6173c0c26..c9acff7d4f18c 100644
--- a/docs/token_generation_performance_tips.md
+++ b/docs/token_generation_performance_tips.md
@@ -3,7 +3,7 @@
 ## Verifying that the model is running on the GPU with cuBLAS
 Make sure you compiled llama with the correct env variables according to [this guide](../README.md#cublas), so that llama accepts the `-ngl N` (or `--n-gpu-layers N`) flag. When running llama, you may configure `N` to be very large, and llama will offload the maximum possible number of layers to the GPU, even if it's less than the number you configured. For example:
 ```shell
-./main -m "path/to/model.bin" -ngl 200000 -p "Please sir, may I have some "
+./main -m "path/to/model.gguf" -ngl 200000 -p "Please sir, may I have some "
 ```
 
 When running llama, before it starts the inference work, it will output diagnostic information that shows whether cuBLAS is offloading work to the GPU. Look for these lines:
@@ -25,9 +25,9 @@ GPU: A6000 (48GB VRAM)
 CPU: 7 physical cores
 RAM: 32GB
 
-Model: `TheBloke_Wizard-Vicuna-30B-Uncensored-GGML/Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_0.bin` (30B parameters, 4bit quantization, GGML)
+Model: `TheBloke_Wizard-Vicuna-30B-Uncensored-GGML/Wizard-Vicuna-30B-Uncensored.q4_0.gguf` (30B parameters, 4bit quantization, GGML)
 
-Run command: `./main -m "path/to/model.bin" -p "-p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]`
+Run command: `./main -m "path/to/model.gguf" -p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]`
 
 Result:
 
diff --git a/examples/common.h b/examples/common.h
index b8510dfe453e0..62772eff7e541 100644
--- a/examples/common.h
+++ b/examples/common.h
@@ -52,7 +52,7 @@ struct gpt_params {
     std::string cfg_negative_prompt;       // string to help guidance
     float       cfg_scale         = 1.f;   // How strong is guidance
 
-    std::string model             = "models/7B/ggml-model.bin"; // model path
+    std::string model             = "models/7B/ggml-model-f16.bin"; // model path
     std::string model_alias       = "unknown"; // model alias
     std::string prompt            = "";
     std::string path_prompt_cache = "";  // path to file for saving/loading prompt eval state
diff --git a/examples/metal/metal.cpp b/examples/metal/metal.cpp
index 7438defdefcdf..c05a4fa933d31 100644
--- a/examples/metal/metal.cpp
+++ b/examples/metal/metal.cpp
@@ -2,7 +2,7 @@
 //
 // - First, export a LLaMA graph:
 //
-//  $ ./bin/main -m ../models/7B/ggml-model-q4_0.bin --export
+//  $ ./bin/main -m ../models/7B/ggml-model-q4_0.gguf --export
 //
 // - Run this tool to evaluate the exported graph:
 //
diff --git a/examples/server/README.md b/examples/server/README.md
index 1559dd3f2639a..ce2d2c9c69257 100644
--- a/examples/server/README.md
+++ b/examples/server/README.md
@@ -5,7 +5,7 @@ This example demonstrates a simple HTTP API server and a simple web front end to
 Command line options:
 
 -   `--threads N`, `-t N`: Set the number of threads to use during computation.
--   `-m FNAME`, `--model FNAME`: Specify the path to the LLaMA model file (e.g., `models/7B/ggml-model.bin`).
+-   `-m FNAME`, `--model FNAME`: Specify the path to the LLaMA model file (e.g., `models/7B/ggml-model.gguf`).
 -   `-m ALIAS`, `--alias ALIAS`: Set an alias for the model. The alias will be returned in API responses.
 -   `-c N`, `--ctx-size N`: Set the size of the prompt context. The default is 512, but LLaMA models were built with a context of 2048, which will provide better results for longer input/inference. The size may differ in other models, for example, baichuan models were build with a context of 4096.
 -   `-ngl N`, `--n-gpu-layers N`: When compiled with appropriate support (currently CLBlast or cuBLAS), this option allows offloading some layers to the GPU for computation. Generally results in increased performance.
@@ -48,14 +48,12 @@ To get started right away, run the following command, making sure to use the cor
 ### Unix-based systems (Linux, macOS, etc.):
 
 ```bash
-./server -m models/7B/ggml-model.bin -c 2048
+./server -m models/7B/ggml-model.gguf -c 2048
 ```
 
 ### Windows:
 
 ```powershell
-server.exe -m models\7B\ggml-model.bin -c 2048
-```
 
 The above command will start a server that by default listens on `127.0.0.1:8080`.
 You can consume the endpoints with Postman or NodeJS with axios library. You can visit the web front end at the same url.
diff --git a/llama.cpp b/llama.cpp
index 1d1642bd8ae8c..018f52e71a137 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -3575,7 +3575,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             } else {
                 size_t counter = 0;
                 new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements] () {
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size]() { // NOLINT
                     std::vector<int64_t> local_hist;
                     size_t local_size = 0;
                     while (true) {

From 81a2c2a6f4ae2a6a89a0456b5b405c426f59db75 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:49:02 +0300
Subject: [PATCH 176/242] llama : fix llama_model_loader memory leak

---
 llama.cpp | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/llama.cpp b/llama.cpp
index 018f52e71a137..753d130d65de1 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1083,6 +1083,15 @@ struct llama_model_loader {
         this->use_mmap = use_mmap;
     }
 
+    ~llama_model_loader() {
+        if (ctx_gguf) {
+            gguf_free(ctx_gguf);
+        }
+        if (ctx_meta) {
+            ggml_free(ctx_meta);
+        }
+    }
+
     const char * get_tensor_name(int i) const {
         return gguf_get_tensor_name(ctx_gguf, i);
     }

From 93f285bdf19ce3413811d2e89e99a272332eee94 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:38:48 +0300
Subject: [PATCH 177/242] gptneox : move as a WIP example

---
 .gitignore                                                   | 1 -
 Makefile                                                     | 5 +----
 .../gptneox-wip/cmpnct_gpt2bpe.hpp                           | 0
 gptneox-main.cpp => examples/gptneox-wip/gptneox-main.cpp    | 0
 4 files changed, 1 insertion(+), 5 deletions(-)
 rename cmpnct_gpt2bpe.hpp => examples/gptneox-wip/cmpnct_gpt2bpe.hpp (100%)
 rename gptneox-main.cpp => examples/gptneox-wip/gptneox-main.cpp (100%)

diff --git a/.gitignore b/.gitignore
index 107caa228b89e..1cfec94d94c84 100644
--- a/.gitignore
+++ b/.gitignore
@@ -50,7 +50,6 @@ models-mnt
 /embd-input-test
 /gguf
 /gguf-llama-simple
-/gptneox-main
 /libllama.so
 build-info.h
 arm_neon.h
diff --git a/Makefile b/Makefile
index 0f768fd1990cf..7199912f16b4f 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple server embd-input-test gguf gptneox-main
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple server embd-input-test gguf
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-llama-grammar tests/test-grammar-parser tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0
@@ -388,9 +388,6 @@ embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-te
 gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
-gptneox-main: gptneox-main.cpp ggml.o $(OBJS)
-	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
-
 train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
diff --git a/cmpnct_gpt2bpe.hpp b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
similarity index 100%
rename from cmpnct_gpt2bpe.hpp
rename to examples/gptneox-wip/cmpnct_gpt2bpe.hpp
diff --git a/gptneox-main.cpp b/examples/gptneox-wip/gptneox-main.cpp
similarity index 100%
rename from gptneox-main.cpp
rename to examples/gptneox-wip/gptneox-main.cpp

From 899f9a5350937db8b80900a1604edec5997e4b01 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 19:49:21 +0300
Subject: [PATCH 178/242] llama : fix lambda capture

ggml-ci
---
 llama.cpp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 753d130d65de1..c8403e9c538d2 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -3576,7 +3576,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             new_data = work.data();
             std::vector<int64_t> hist_cur(1 << 4, 0);
 
-            const int chunk_size = 32 * 512;
+            static const int chunk_size = 32 * 512;
             const int nchunk = (nelements + chunk_size - 1)/chunk_size;
             const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
             if (nthread_use < 2) {
@@ -3584,7 +3584,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             } else {
                 size_t counter = 0;
                 new_size = 0;
-                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size]() { // NOLINT
+                auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements]() {
                     std::vector<int64_t> local_hist;
                     size_t local_size = 0;
                     while (true) {

From e72c8c21240b4d2c0d9dd5aed85d4e18c1bd808d Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 20:13:12 +0300
Subject: [PATCH 179/242] ggml : fix bug in gguf_set_kv

ggml-ci
---
 ggml.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml.c b/ggml.c
index f7e72ed84ddc6..228271fc221c2 100644
--- a/ggml.c
+++ b/ggml.c
@@ -19290,7 +19290,7 @@ void gguf_set_kv(struct gguf_context * ctx, struct gguf_context * src) {
                         }
                         gguf_set_arr_str(ctx, src->kv[i].key.data, data, src->kv[i].value.arr.n);
                         free(data);
-                    } if (src->kv[i].value.arr.type == GGUF_TYPE_ARRAY) {
+                    } else if (src->kv[i].value.arr.type == GGUF_TYPE_ARRAY) {
                         GGML_ASSERT(false && "nested arrays not supported");
                     } else {
                         gguf_set_arr_data(ctx, src->kv[i].key.data, src->kv[i].value.arr.type, src->kv[i].value.arr.data, src->kv[i].value.arr.n);

From fb11dd3f9263b9542513987d6aa5bfaed3960a98 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 19:16:35 +0200
Subject: [PATCH 180/242] common.h : .bin --> .gguf

---
 examples/common.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/examples/common.h b/examples/common.h
index 62772eff7e541..de04b4c5b2f03 100644
--- a/examples/common.h
+++ b/examples/common.h
@@ -52,7 +52,7 @@ struct gpt_params {
     std::string cfg_negative_prompt;       // string to help guidance
     float       cfg_scale         = 1.f;   // How strong is guidance
 
-    std::string model             = "models/7B/ggml-model-f16.bin"; // model path
+    std::string model             = "models/7B/ggml-model-f16.gguf"; // model path
     std::string model_alias       = "unknown"; // model alias
     std::string prompt            = "";
     std::string path_prompt_cache = "";  // path to file for saving/loading prompt eval state

From 78e1e57862cf8c0d046823b1bc45537e6a33047b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 19:18:24 +0200
Subject: [PATCH 181/242] quantize-stats.cpp : .bin --> .gguf

---
 examples/quantize-stats/quantize-stats.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/examples/quantize-stats/quantize-stats.cpp b/examples/quantize-stats/quantize-stats.cpp
index a330b20df2963..60ace876503ab 100644
--- a/examples/quantize-stats/quantize-stats.cpp
+++ b/examples/quantize-stats/quantize-stats.cpp
@@ -25,7 +25,7 @@
 #endif
 
 struct quantize_stats_params {
-    std::string model = "models/7B/ggml-model-f16.bin";
+    std::string model = "models/7B/ggml-model-f16.gguf";
     bool verbose = false;
     bool per_layer_stats = false;
     bool print_histogram = false;

From acaa98234ac80480cd15ebf1f8e99f81cd5ec442 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 21:06:45 +0300
Subject: [PATCH 182/242] convert.py : fix HF tensor permuting / unpacking

ggml-ci
---
 convert.py | 20 +++++++++++++++++++-
 gguf.py    |  6 +++++-
 2 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/convert.py b/convert.py
index adc5fdd26d804..f6237579d0954 100755
--- a/convert.py
+++ b/convert.py
@@ -812,6 +812,23 @@ def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyM
 def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
     tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
 
+    tmp = model
+
+    # HF models permut or pack some of the tensors, so we need to undo that
+    for i in itertools.count():
+        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
+            print(f"Permuting layer {i}")
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
+           #tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
+        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
+            print(f"Unpacking and permuting layer {i}")
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
+        else:
+            break
+
     out: LazyModel = {}
     for name, lazy_tensor in model.items():
         name_new = name
@@ -825,8 +842,9 @@ def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
         else:
             raise Exception(f"Unexpected tensor name: {name}")
 
-        if gguf.should_skip_tensor(ARCH, params.n_layer, name_new):
+        if gguf.should_skip_tensor_TMP(ARCH, params.n_layer, name_new):
             print(f"skipping tensor {name_new}")
+            continue
         else:
             print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
             out[name_new] = lazy_tensor
diff --git a/gguf.py b/gguf.py
index a4dd108722d6c..2a6806a9188a0 100644
--- a/gguf.py
+++ b/gguf.py
@@ -148,7 +148,11 @@ class MODEL_TENSOR(IntEnum):
         ],
     }
 
-def should_skip_tensor(arch : MODEL_ARCH, n_blocks : int, name : str) -> bool:
+# TODO: the following helper functions should be removed
+#       instead, get_tensor_name_map should return tuples of (name, MODEL_TENSOR)
+#       however, my Python is very bad, and I couldn't figure out how to do this, hence these functions
+# REMOVE
+def should_skip_tensor_TMP(arch : MODEL_ARCH, n_blocks : int, name : str) -> bool:
     for skip in MODEL_TENSOR_SKIP.get(arch, []):
         for i in range(n_blocks):
             if name == MODEL_TENSOR_NAMES[arch][skip].format(bid=i):

From b3cc182990254e17d6a33171666343503e1b2607 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 20:27:50 +0200
Subject: [PATCH 183/242] llama.cpp : typo

---
 llama.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llama.cpp b/llama.cpp
index c8403e9c538d2..44ff5e43a4b94 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1057,7 +1057,7 @@ struct llama_model_loader {
         // print meta data
         // TODO: make optional
         {
-            LLAMA_LOG_INFO("%s: loaded meta data with %d key-value paris and %d tensors from %s (version %s)\n",
+            LLAMA_LOG_INFO("%s: loaded meta data with %d key-value pairs and %d tensors from %s (version %s)\n",
                     __func__, n_kv, n_tensors, fname.c_str(), llama_file_version_name(file_version));
 
             for (int i = 0; i < n_tensors; i++) {

From 57eaadb853dd03e674104f6eda9e177cfeb0cf47 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 21:31:52 +0300
Subject: [PATCH 184/242] llama : throw error if gguf fails to init from file

ggml-ci
---
 llama.cpp | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/llama.cpp b/llama.cpp
index 44ff5e43a4b94..b0d23abbe33df 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1042,6 +1042,9 @@ struct llama_model_loader {
         };
 
         ctx_gguf = gguf_init_from_file(fname.c_str(), params);
+        if (!ctx_gguf) {
+            throw std::runtime_error(format("%s: failed to load model from %s\n", __func__, fname.c_str()));
+        }
 
         n_kv      = gguf_get_n_kv(ctx_gguf);
         n_tensors = gguf_get_n_tensors(ctx_gguf);

From 5484737d58ea111f48cb773b7aae7b299b3552f8 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Thu, 17 Aug 2023 21:40:51 +0300
Subject: [PATCH 185/242] llama : fix tensor name grepping during quantization

ggml-ci
---
 llama.cpp | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/llama.cpp b/llama.cpp
index b0d23abbe33df..35945d3574acb 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -3432,6 +3432,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
 
         const std::string name = ggml_get_name(meta);
 
+        // TODO: avoid hardcoded tensor names - use the TN_* constants
         if (name.find("attn_v.weight") != std::string::npos) {
             ++n_attention_wv;
         }
@@ -3510,6 +3511,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         } else {
             new_type = quantized_type;
 #ifdef GGML_USE_K_QUANTS
+            // TODO: avoid hardcoded tensor names - use the TN_* constants
             if (name == TN_OUTPUT) {
                 int nx = tensor->ne[0];
                 int ny = tensor->ne[1];
@@ -3524,7 +3526,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
                 else if (QK_K == 64 && (ftype == LLAMA_FTYPE_MOSTLY_Q4_K_S || ftype == LLAMA_FTYPE_MOSTLY_Q3_K_S) &&
                         (i_attention_wv < n_attention_wv/8 || i_attention_wv >= 7*n_attention_wv/8)) new_type = GGML_TYPE_Q6_K;
                 ++i_attention_wv;
-            } else if (name.find("feed_forward.w2.weight") != std::string::npos) {
+            } else if (name.find("ffn_down.weight") != std::string::npos) {
                 if      (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q2_K) new_type = GGML_TYPE_Q4_K;
                 else if (ftype == LLAMA_FTYPE_MOSTLY_Q3_K_L) new_type = GGML_TYPE_Q5_K;
                 else if ((ftype == LLAMA_FTYPE_MOSTLY_Q4_K_M || ftype == LLAMA_FTYPE_MOSTLY_Q5_K_M) &&

From fc3a52321144abc2f6775fee293bcac52507a78d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=2E=20Yusuf=20Sar=C4=B1g=C3=B6z?= <yusufsarigoz@gmail.com>
Date: Thu, 17 Aug 2023 21:57:39 +0300
Subject: [PATCH 186/242] gguf.py : write tensors in a single pass (#2644)

* gguf : single pass for writing tensors + refactoring writer

* gguf : single pass for writing tensors + refactoring writer

* gguf : single pass for writing tensors + refactoring writer

* gguf : style fixes in simple conversion script

* gguf : refactor gptneox conversion script

* gguf : rename h5 to hf (for HuggingFace)

* gguf : refactor pth to gguf conversion script

* gguf : rm file_type key and method

* gguf.py : fix vertical alignment

* gguf.py : indentation

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---
 ...o-gguf.py => convert-gptneox-hf-to-gguf.py | 117 ++-----
 convert-llama-7b-pth-to-gguf.py               | 120 +++----
 ...-to-gguf.py => convert-llama-hf-to-gguf.py | 140 +++-----
 gguf.py                                       | 329 ++++++++++--------
 4 files changed, 301 insertions(+), 405 deletions(-)
 rename convert-gptneox-h5-to-gguf.py => convert-gptneox-hf-to-gguf.py (68%)
 rename convert-llama-h5-to-gguf.py => convert-llama-hf-to-gguf.py (67%)

diff --git a/convert-gptneox-h5-to-gguf.py b/convert-gptneox-hf-to-gguf.py
similarity index 68%
rename from convert-gptneox-h5-to-gguf.py
rename to convert-gptneox-hf-to-gguf.py
index 79876eee3e188..11cf19282aedc 100644
--- a/convert-gptneox-h5-to-gguf.py
+++ b/convert-gptneox-hf-to-gguf.py
@@ -13,6 +13,8 @@
 from transformers import AutoTokenizer
 
 # ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
+
+
 def bytes_to_unicode():
     """
     Returns list of utf-8 byte and a corresponding list of unicode strings.
@@ -34,6 +36,7 @@ def bytes_to_unicode():
     cs = [chr(n) for n in cs]
     return dict(zip(bs, cs))
 
+
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
     for filename in os.listdir(dir_model):
@@ -44,6 +47,7 @@ def count_model_parts(dir_model: str) -> int:
         print("gguf: found " + str(num_parts) + " model parts")
     return num_parts
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
@@ -58,7 +62,7 @@ def count_model_parts(dir_model: str) -> int:
 # possible tensor data types
 #   ftype == 0 -> float32
 #   ftype == 1 -> float16
-#
+
 # map from ftype to string
 ftype_str = ["f32", "f16"]
 
@@ -67,6 +71,7 @@ def count_model_parts(dir_model: str) -> int:
     ftype = int(sys.argv[2])
     if ftype < 0 or ftype > 1:
         print("Invalid ftype: " + str(ftype))
+
         sys.exit(1)
 
 fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
@@ -77,30 +82,30 @@ def count_model_parts(dir_model: str) -> int:
     hparams = json.load(f)
 
 if hparams["architectures"][0] != "GPTNeoXForCausalLM":
-    print("Model architecture not supported: " + hparams["architectures"][0] )
+    print("Model architecture not supported: " + hparams["architectures"][0])
+
     sys.exit()
 
 # get number of model parts
 num_parts = count_model_parts(dir_model)
 
-gguf_writer = gguf.GGUFWriter.open(fname_out)
+llm_arch = "gptneox"
+gguf_writer = gguf.GGUFWriter(fname_out, arch=llm_arch)
 
 print("gguf: get model metadata")
 
-llm_arch    = "gptneox"
 block_count = hparams["num_hidden_layers"]
 
-gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
-gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
-gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
-gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_block_count(llm_arch, block_count)
-gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
-gguf_writer.add_rope_dimension_count(llm_arch, int( hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])) )
-gguf_writer.add_head_count(llm_arch, hparams["num_attention_heads"])
-gguf_writer.add_parallel_residual(llm_arch, hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
-gguf_writer.add_layer_norm_eps(llm_arch, hparams["layer_norm_eps"])
+gguf_writer.add_context_length(hparams["max_position_embeddings"])
+gguf_writer.add_embedding_length(hparams["hidden_size"])
+gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(int(hparams["rotary_pct"]*(hparams["hidden_size"]//hparams["num_attention_heads"])))
+gguf_writer.add_head_count(hparams["num_attention_heads"])
+gguf_writer.add_parallel_residual(hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
+gguf_writer.add_layer_norm_eps(hparams["layer_norm_eps"])
 
 # TOKENIZATION
 
@@ -124,14 +129,14 @@ def count_model_parts(dir_model: str) -> int:
 
     print("gguf: get gpt2 tokenizer vocab")
 
-    vocab_size = len( tokenizer_json["model"]["vocab"] )
+    vocab_size = len(tokenizer_json["model"]["vocab"])
 
     # ref: https://github.com/cmp-nct/ggllm.cpp/blob/master/falcon_convert.py
     tokenizer = AutoTokenizer.from_pretrained(dir_model)
 
     reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()}
     byte_encoder = bytes_to_unicode()
-    byte_decoder = {v:k for k, v in byte_encoder.items()}
+    byte_decoder = {v: k for k, v in byte_encoder.items()}
 
     for i in range(vocab_size):
         if i in reverse_vocab:
@@ -146,8 +151,9 @@ def count_model_parts(dir_model: str) -> int:
                         text.extend(c.encode('utf-8'))
         else:
             print(f"Key {i} not in tokenizer vocabulary. Padding with an arbitrary token.")
-            padding_token = f"[PAD{i}]".encode("utf8")
-            text = bytearray(padding_token)
+            pad_token = f"[PAD{i}]".encode("utf8")
+            text = bytearray(pad_token)
+
         tokens.append(text)
 
     gguf_writer.add_token_list(tokens)
@@ -201,7 +207,7 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
+    print("gguf: loading model part '" + part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
@@ -223,11 +229,12 @@ def count_model_parts(dir_model: str) -> int:
         elif name.endswith(".bias") and name[:-5] in tensor_map:
             name = tensor_map[name[:-5]] + ".bias"
         else:
-            print( "Can not map tensor '" + name + "'" )
+            print("Can not map tensor '" + name + "'")
             sys.exit()
 
         n_dims = len(data.shape)
         data_dtype = data.dtype
+        old_dtype = data_dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -241,77 +248,21 @@ def count_model_parts(dir_model: str) -> int:
         if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
             data_dtype = np.float16
 
-        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data_dtype))
+
+        data = data.astype(data_dtype)
 
-        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
+        gguf_writer.add_tensor(name, data)
 
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
 print("gguf: write metadata")
 gguf_writer.write_kv_data_to_file()
-print("gguf: write tensor metadata")
-gguf_writer.write_ti_data_to_file()
-
-# tensor data
-print("gguf: convert and write tensor data")
-
-if num_parts == 0:
-    part_names = ("pytorch_model.bin",)
-else:
-    part_names = (
-        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
-    )
-
-for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
-    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
-
-    for name in model_part.keys():
-        data = model_part[name]
-
-        old_dtype = data.dtype
-
-        # we don't need these
-        if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
-            continue
-
-        # convert any unsupported data types to float32
-        if data.dtype != torch.float16 and data.dtype != torch.float32:
-            data = data.to(torch.float32)
-
-        data = data.squeeze().numpy()
-
-        # map tensor names
-        if name.endswith(".weight") and name[:-7] in tensor_map:
-            name = tensor_map[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tensor_map:
-            name = tensor_map[name[:-5]] + ".bias"
-        else:
-            print( "Can not map tensor '" + name + "'" )
-            sys.exit()
-
-        n_dims = len(data.shape)
-        data_dtype = data.dtype
-
-        # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data = data.astype(np.float32)
-
-        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data = data.astype(np.float32)
-
-        # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-
-        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
-
-        gguf_writer.write_tensor_to_file(data)
+print("gguf: write tensors")
+gguf_writer.write_tensors_to_file()
 
 gguf_writer.close()
 
-
-print("gguf: model successfully exported to '" + fname_out + "'" )
+print("gguf: model successfully exported to '" + fname_out + "'")
 print("")
diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 9afea8a7e4bf1..72ad4d4ea7069 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -18,6 +18,7 @@
 # compatible with python < 3.9
 NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
 
+
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
     for filename in os.listdir(dir_model):
@@ -28,10 +29,12 @@ def count_model_parts(dir_model: str) -> int:
         print("gguf: found " + str(num_parts) + " model parts")
     return num_parts
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
     print("  ftype == 1 -> float16")
+
     sys.exit(1)
 
 
@@ -43,7 +46,7 @@ def count_model_parts(dir_model: str) -> int:
 # possible tensor data types
 #   ftype == 0 -> float32
 #   ftype == 1 -> float16
-#
+
 # map from ftype to string
 ftype_str = ["f32", "f16"]
 
@@ -52,6 +55,7 @@ def count_model_parts(dir_model: str) -> int:
     ftype = int(sys.argv[2])
     if ftype < 0 or ftype > 1:
         print("Invalid ftype: " + str(ftype))
+
         sys.exit(1)
 
 fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
@@ -70,14 +74,14 @@ def count_model_parts(dir_model: str) -> int:
 
 if num_parts > 1:
     print("gguf: Only models with a single datafile are supported.")
-    sys.exit()
 
-gguf_writer = gguf.GGUFWriter.open(fname_out)
+    sys.exit()
+llm_arch = "llama"
+gguf_writer = gguf.GGUFWriter(fname_out, arch=llm_arch)
 
 
 print("gguf: get model metadata")
 
-llm_arch = "llama"
 block_count = hparams["num_hidden_layers"]
 head_count = hparams["num_attention_heads"]
 
@@ -89,21 +93,20 @@ def count_model_parts(dir_model: str) -> int:
 if "_name_or_path" in hparams:
     hf_repo = hparams["_name_or_path"]
 else:
-    hf_repo=""
+    hf_repo = ""
 
-gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
-gguf_writer.add_file_type( "All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
-gguf_writer.add_tensor_data_layout(llm_arch, "Meta AI original pth")
-gguf_writer.add_context_length(llm_arch, hparams["max_position_embeddings"])
-gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_block_count(llm_arch, block_count)
-gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
-gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
-gguf_writer.add_head_count(llm_arch, head_count)
-gguf_writer.add_head_count_kv(llm_arch, head_count_kv)
-gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
+gguf_writer.add_tensor_data_layout("Meta AI original pth")
+gguf_writer.add_context_length(hparams["max_position_embeddings"])
+gguf_writer.add_embedding_length(hparams["hidden_size"])
+gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(hparams["hidden_size"] // hparams["num_attention_heads"])
+gguf_writer.add_head_count(head_count)
+gguf_writer.add_head_count_kv(head_count_kv)
+gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
 
 # TOKENIZATION
@@ -125,19 +128,23 @@ def count_model_parts(dir_model: str) -> int:
         score: float
 
         piece = tokenizer.id_to_piece(i)
-        text  = piece.encode("utf-8")
+        text = piece.encode("utf-8")
         score = tokenizer.get_score(i)
 
-        toktype = 1 # defualt to normal token type
-        if tokenizer.is_unknown(i): toktype = 2
-        if tokenizer.is_control(i): toktype = 3
+        toktype = 1  # defualt to normal token type
+        if tokenizer.is_unknown(i):
+            toktype = 2
+        if tokenizer.is_control(i):
+            toktype = 3
 
         # TODO: How to determinate if a token is user defined?
         # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
         # if tokenizer.is_user_defined(i): toktype = 4
 
-        if tokenizer.is_unused(i):  toktype = 5
-        if tokenizer.is_byte(i):    toktype = 6
+        if tokenizer.is_unused(i):
+            toktype = 5
+        if tokenizer.is_byte(i):
+            toktype = 6
 
         tokens.append(text)
         scores.append(score)
@@ -193,10 +200,10 @@ def count_model_parts(dir_model: str) -> int:
 # tensor info
 print("gguf: get tensor metadata")
 
-part_names = ( f"consolidated.{n:02}.pth" for n in range(0, num_parts) )
+part_names = (f"consolidated.{n:02}.pth" for n in range(0, num_parts))
 
 for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
+    print("gguf: loading model part '" + part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
@@ -218,11 +225,12 @@ def count_model_parts(dir_model: str) -> int:
         elif name.endswith(".bias") and name[:-5] in tensor_map:
             name = tensor_map[name[:-5]] + ".bias"
         else:
-            print( "Can not map tensor '" + name + "'" )
+            print("Can not map tensor '" + name + "'")
             sys.exit()
 
         n_dims = len(data.shape)
         data_dtype = data.dtype
+        old_dtype = data_dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -236,69 +244,19 @@ def count_model_parts(dir_model: str) -> int:
         if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
             data_dtype = np.float16
 
-        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data_dtype))
+
+        data = data.astype(data_dtype)
 
-        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
+        gguf_writer.add_tensor(name, data)
 
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
 print("gguf: write metadata")
 gguf_writer.write_kv_data_to_file()
-print("gguf: write tensor metadata")
-gguf_writer.write_ti_data_to_file()
-
-# tensor data
-print("gguf: convert and write tensor data")
-
-part_names = ( f"consolidated.{n:02}.pth" for n in range(0, num_parts) )
-
-for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
-    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
-
-    for name in model_part.keys():
-        data = model_part[name]
-
-        old_dtype = data.dtype
-
-        # we don't need these
-        if name == "rope.freqs":
-            continue
-
-        # convert any unsupported data types to float32
-        if data.dtype != torch.float16 and data.dtype != torch.float32:
-            data = data.to(torch.float32)
-
-        data = data.squeeze().numpy()
-
-        # map tensor names
-        if name.endswith(".weight") and name[:-7] in tensor_map:
-            name = tensor_map[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tensor_map:
-            name = tensor_map[name[:-5]] + ".bias"
-        else:
-            print( "Can not map tensor '" + name + "'" )
-            sys.exit()
-
-        n_dims = len(data.shape)
-        data_dtype = data.dtype
-
-        # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data = data.astype(np.float32)
-
-        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data = data.astype(np.float32)
-
-        # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-
-        print( name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
-
-        gguf_writer.write_tensor_data(data)
+print("gguf: write tensors")
+gguf_writer.write_tensors_to_file()
 
 gguf_writer.close()
 
diff --git a/convert-llama-h5-to-gguf.py b/convert-llama-hf-to-gguf.py
similarity index 67%
rename from convert-llama-h5-to-gguf.py
rename to convert-llama-hf-to-gguf.py
index a2b3f9a3061fb..885dd640abe0a 100644
--- a/convert-llama-h5-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -18,26 +18,35 @@
 
 # reverse HF permute back to original pth layout
 # https://github.com/huggingface/transformers/blob/main/src/transformers/models/llama/convert_llama_weights_to_hf.py
+
+
 def reverse_hf_permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
-    if n_kv_head is not None and n_head != n_kv_head: n_head //= n_kv_head
+    if n_kv_head is not None and n_head != n_kv_head:
+        n_head //= n_kv_head
+
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                .swapaxes(1, 2)
-                .reshape(weights.shape))
+            .swapaxes(1, 2)
+            .reshape(weights.shape))
+
 
 def count_model_parts(dir_model: str) -> int:
     num_parts = 0
+
     for filename in os.listdir(dir_model):
         if filename.startswith("pytorch_model-"):
             num_parts += 1
 
     if num_parts > 0:
         print("gguf: found " + str(num_parts) + " model parts")
+
     return num_parts
 
+
 if len(sys.argv) < 3:
     print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
     print("  ftype == 0 -> float32")
     print("  ftype == 1 -> float16")
+
     sys.exit(1)
 
 
@@ -49,7 +58,8 @@ def count_model_parts(dir_model: str) -> int:
 # possible tensor data types
 #   ftype == 0 -> float32
 #   ftype == 1 -> float16
-#
+
+
 # map from ftype to string
 ftype_str = ["f32", "f16"]
 
@@ -58,6 +68,7 @@ def count_model_parts(dir_model: str) -> int:
     ftype = int(sys.argv[2])
     if ftype < 0 or ftype > 1:
         print("Invalid ftype: " + str(ftype))
+
         sys.exit(1)
 
 fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
@@ -69,17 +80,17 @@ def count_model_parts(dir_model: str) -> int:
 
 if hparams["architectures"][0] != "LlamaForCausalLM":
     print("Model architecture not supported: " + hparams["architectures"][0])
+
     sys.exit()
 
 # get number of model parts
 num_parts = count_model_parts(dir_model)
 
-gguf_writer = gguf.GGUFWriter.open(fname_out)
+gguf_writer = gguf.GGUFWriter(fname_out, arch="llama")
 
 
 print("gguf: get model metadata")
 
-llm_arch = "llama"
 block_count = hparams["num_hidden_layers"]
 head_count = hparams["num_attention_heads"]
 
@@ -91,7 +102,7 @@ def count_model_parts(dir_model: str) -> int:
 if "_name_or_path" in hparams:
     hf_repo = hparams["_name_or_path"]
 else:
-    hf_repo=""
+    hf_repo = ""
 
 if "max_sequence_length" in hparams:
     ctx_length = hparams["max_sequence_length"]
@@ -99,22 +110,22 @@ def count_model_parts(dir_model: str) -> int:
     ctx_length = hparams["max_position_embeddings"]
 else:
     print("gguf: can not find ctx length parameter.")
+
     sys.exit()
 
 
-gguf_writer.add_architecture(llm_arch)
+gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
-gguf_writer.add_file_type("All tensors F32" if ftype == 0 else "Most tensors F16, some F32")
 gguf_writer.add_source_hf_repo(hf_repo)
-gguf_writer.add_tensor_data_layout(llm_arch, "Meta AI original pth")
-gguf_writer.add_context_length(llm_arch, ctx_length)
-gguf_writer.add_embedding_length(llm_arch, hparams["hidden_size"])
-gguf_writer.add_block_count(llm_arch, block_count)
-gguf_writer.add_feed_forward_length(llm_arch, hparams["intermediate_size"])
-gguf_writer.add_rope_dimension_count(llm_arch, hparams["hidden_size"] // hparams["num_attention_heads"])
-gguf_writer.add_head_count(llm_arch, head_count)
-gguf_writer.add_head_count_kv(llm_arch, head_count_kv)
-gguf_writer.add_layer_norm_rms_eps(llm_arch, hparams["rms_norm_eps"])
+gguf_writer.add_tensor_data_layout("Meta AI original pth")
+gguf_writer.add_context_length(ctx_length)
+gguf_writer.add_embedding_length(hparams["hidden_size"])
+gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(hparams["intermediate_size"])
+gguf_writer.add_rope_dimension_count(hparams["hidden_size"] // hparams["num_attention_heads"])
+gguf_writer.add_head_count(head_count)
+gguf_writer.add_head_count_kv(head_count_kv)
+gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
 
 # TOKENIZATION
@@ -136,19 +147,23 @@ def count_model_parts(dir_model: str) -> int:
         score: float
 
         piece = tokenizer.id_to_piece(i)
-        text  = piece.encode("utf-8")
+        text = piece.encode("utf-8")
         score = tokenizer.get_score(i)
 
-        toktype = 1 # defualt to normal token type
-        if tokenizer.is_unknown(i): toktype = 2
-        if tokenizer.is_control(i): toktype = 3
+        toktype = 1  # defualt to normal token type
+        if tokenizer.is_unknown(i):
+            toktype = 2
+        if tokenizer.is_control(i):
+            toktype = 3
 
         # TODO: How to determinate if a token is user defined?
         # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
         # if tokenizer.is_user_defined(i): toktype = 4
 
-        if tokenizer.is_unused(i):  toktype = 5
-        if tokenizer.is_byte(i):    toktype = 6
+        if tokenizer.is_unused(i):
+            toktype = 5
+        if tokenizer.is_byte(i):
+            toktype = 6
 
         tokens.append(text)
         scores.append(score)
@@ -212,7 +227,7 @@ def count_model_parts(dir_model: str) -> int:
     )
 
 for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
+    print("gguf: loading model part '" + part_name + "'")
     model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
 
     for name in model_part.keys():
@@ -238,11 +253,13 @@ def count_model_parts(dir_model: str) -> int:
         elif name.endswith(".bias") and name[:-5] in tensor_map:
             name = tensor_map[name[:-5]] + ".bias"
         else:
-            print( "Can not map tensor '" + name + "'" )
+            print("Can not map tensor '" + name + "'")
+
             sys.exit()
 
         n_dims = len(data.shape)
         data_dtype = data.dtype
+        old_dtype = data_dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data.dtype == np.float16:
@@ -256,78 +273,19 @@ def count_model_parts(dir_model: str) -> int:
         if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
             data_dtype = np.float16
 
-        data_nbytes = data.size * 2 if data_dtype == np.float16 else data.size * 4
+        data = data.astype(data_dtype)
+
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
-        gguf_writer.add_tensor_info(name, data.shape, data_dtype, data_nbytes)
+        gguf_writer.add_tensor(name, data)
 
 
 print("gguf: write header")
 gguf_writer.write_header_to_file()
 print("gguf: write metadata")
 gguf_writer.write_kv_data_to_file()
-print("gguf: write tensor metadata")
-gguf_writer.write_ti_data_to_file()
-
-# tensor data
-print("gguf: convert and write tensor data")
-
-if num_parts == 0:
-    part_names = ("pytorch_model.bin",)
-else:
-    part_names = (
-        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
-    )
-
-for part_name in part_names:
-    print("gguf: loading model part '"+ part_name + "'")
-    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
-
-    for name in model_part.keys():
-        data = model_part[name]
-
-        old_dtype = data.dtype
-
-        # we don't need these
-        if name.endswith(".rotary_emb.inv_freq"):
-            continue
-
-        # convert any unsupported data types to float32
-        if data.dtype != torch.float16 and data.dtype != torch.float32:
-            data = data.to(torch.float32)
-
-        data = data.squeeze().numpy()
-
-        # reverse permute these
-        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
-            data = reverse_hf_permute(data, head_count, head_count_kv)
-
-        # map tensor names
-        if name.endswith(".weight") and name[:-7] in tensor_map:
-            name = tensor_map[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tensor_map:
-            name = tensor_map[name[:-5]] + ".bias"
-        else:
-            print( "Can not map tensor '" + name + "'" )
-            sys.exit()
-
-        n_dims = len(data.shape)
-        data_dtype = data.dtype
-
-        # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data = data.astype(np.float32)
-
-        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data = data.astype(np.float32)
-
-        # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data = data.astype(np.float16)
-
-        print(name + ", shape " + str(len(data.shape)) + ", " + str(old_dtype) + " --> " + str(data.dtype))
-
-        gguf_writer.write_tensor_to_file(data)
+print("gguf: write tensors")
+gguf_writer.write_tensors_to_file()
 
 gguf_writer.close()
 
diff --git a/gguf.py b/gguf.py
index 2a6806a9188a0..70418ef5f94d5 100644
--- a/gguf.py
+++ b/gguf.py
@@ -1,11 +1,7 @@
-"""TODOs
-1. Implement writers for known architectures, LLaMA in particular.
-2. Add docstrings from the format specs.
-3. After development is done, Convert it to a proper pip-installable Python package, and possibly move it to its own repo under ggml-org.
-"""
-
+import shutil
 import sys
 import struct
+import tempfile
 import numpy as np
 
 from enum import IntEnum, auto
@@ -27,30 +23,29 @@
 KEY_GENERAL_AUTHOR               = "general.author"
 KEY_GENERAL_URL                  = "general.url"
 KEY_GENERAL_DESCRIPTION          = "general.description"
-KEY_GENERAL_FILE_TYPE            = "general.file_type"
 KEY_GENERAL_LICENSE              = "general.license"
 KEY_GENERAL_SOURCE_URL           = "general.source.url"
 KEY_GENERAL_SOURCE_HF_REPO       = "general.source.hugginface.repository"
 
 # LLM
-KEY_LLM_CONTEXT_LENGTH           = "{arch}.context_length"
-KEY_LLM_EMBEDDING_LENGTH         = "{arch}.embedding_length"
-KEY_LLM_BLOCK_COUNT              = "{arch}.block_count"
-KEY_LLM_FEED_FORWARD_LENGTH      = "{arch}.feed_forward_length"
-KEY_LLM_USE_PARALLEL_RESIDUAL    = "{arch}.use_parallel_residual"
-KEY_LLM_TENSOR_DATA_LAYOUT       = "{arch}.tensor_data_layout"
+KEY_LLM_CONTEXT_LENGTH        = "{arch}.context_length"
+KEY_LLM_EMBEDDING_LENGTH      = "{arch}.embedding_length"
+KEY_LLM_BLOCK_COUNT           = "{arch}.block_count"
+KEY_LLM_FEED_FORWARD_LENGTH   = "{arch}.feed_forward_length"
+KEY_LLM_USE_PARALLEL_RESIDUAL = "{arch}.use_parallel_residual"
+KEY_LLM_TENSOR_DATA_LAYOUT    = "{arch}.tensor_data_layout"
 
 # attention
-KEY_ATTENTION_HEAD_COUNT         = "{arch}.attention.head_count"
-KEY_ATTENTION_HEAD_COUNT_KV      = "{arch}.attention.head_count_kv"
-KEY_ATTENTION_MAX_ALIBI_BIAS     = "{arch}.attention.max_alibi_bias"
-KEY_ATTENTION_CLAMP_KQV          = "{arch}.attention.clamp_kqv"
-KEY_ATTENTION_LAYERNORM_EPS      = "{arch}.attention.layer_norm_epsilon"
-KEY_ATTENTION_LAYERNORM_RMS_EPS  = "{arch}.attention.layer_norm_rms_epsilon"
+KEY_ATTENTION_HEAD_COUNT        = "{arch}.attention.head_count"
+KEY_ATTENTION_HEAD_COUNT_KV     = "{arch}.attention.head_count_kv"
+KEY_ATTENTION_MAX_ALIBI_BIAS    = "{arch}.attention.max_alibi_bias"
+KEY_ATTENTION_CLAMP_KQV         = "{arch}.attention.clamp_kqv"
+KEY_ATTENTION_LAYERNORM_EPS     = "{arch}.attention.layer_norm_epsilon"
+KEY_ATTENTION_LAYERNORM_RMS_EPS = "{arch}.attention.layer_norm_rms_epsilon"
 
 # RoPE
-KEY_ROPE_DIMENSION_COUNT         = "{arch}.rope.dimension_count"
-KEY_ROPE_SCALE                   = "{arch}.rope.scale"
+KEY_ROPE_DIMENSION_COUNT = "{arch}.rope.dimension_count"
+KEY_ROPE_SCALE           = "{arch}.rope.scale"
 
 # tokenization
 KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
@@ -70,6 +65,7 @@
 # recommended mapping of model tensor names for storage in gguf
 #
 
+
 class MODEL_ARCH(IntEnum):
     LLAMA   = auto()
     FALCON  = auto()
@@ -78,81 +74,84 @@ class MODEL_ARCH(IntEnum):
     GPTNEOX = auto()
     MPT     = auto()
 
+
 class MODEL_TENSOR(IntEnum):
-    TOKEN_EMBD        = auto()
-    POS_EMBD          = auto()
-    OUTPUT            = auto()
-    OUTPUT_NORM       = auto()
-    ROPE_FREQS        = auto()
-    ATTN_Q            = auto()
-    ATTN_K            = auto()
-    ATTN_V            = auto()
-    ATTN_QKV          = auto()
-    ATTN_OUT          = auto()
-    ATTN_NORM         = auto()
-    ATTN_NORM_2       = auto()
-    ATTN_ROT_EMBD     = auto()
-    FFN_GATE          = auto()
-    FFN_DOWN          = auto()
-    FFN_UP            = auto()
-    FFN_NORM          = auto()
+    TOKEN_EMBD    = auto()
+    POS_EMBD      = auto()
+    OUTPUT        = auto()
+    OUTPUT_NORM   = auto()
+    ROPE_FREQS    = auto()
+    ATTN_Q        = auto()
+    ATTN_K        = auto()
+    ATTN_V        = auto()
+    ATTN_QKV      = auto()
+    ATTN_OUT      = auto()
+    ATTN_NORM     = auto()
+    ATTN_NORM_2   = auto()
+    ATTN_ROT_EMBD = auto()
+    FFN_GATE      = auto()
+    FFN_DOWN      = auto()
+    FFN_UP        = auto()
+    FFN_NORM      = auto()
+
 
 MODEL_ARCH_NAMES = {
-    MODEL_ARCH.LLAMA   : "llama",
-    MODEL_ARCH.FALCON  : "falcon",
-    MODEL_ARCH.GPT2    : "gpt2",
-    MODEL_ARCH.GPTJ    : "gptj",
-    MODEL_ARCH.GPTNEOX : "gptneox",
-    MODEL_ARCH.MPT     : "mpt",
-    }
+    MODEL_ARCH.LLAMA:   "llama",
+    MODEL_ARCH.FALCON:  "falcon",
+    MODEL_ARCH.GPT2:    "gpt2",
+    MODEL_ARCH.GPTJ:    "gptj",
+    MODEL_ARCH.GPTNEOX: "gptneox",
+    MODEL_ARCH.MPT:     "mpt",
+}
 
 MODEL_TENSOR_NAMES = {
-    MODEL_ARCH.LLAMA  : {
-        MODEL_TENSOR.TOKEN_EMBD        : "token_embd",
-        MODEL_TENSOR.OUTPUT_NORM       : "output_norm",
-        MODEL_TENSOR.OUTPUT            : "output",
-        MODEL_TENSOR.ROPE_FREQS        : "rope_freqs",
-        MODEL_TENSOR.ATTN_NORM         : "blk.{bid}.attn_norm",
-        MODEL_TENSOR.ATTN_Q            : "blk.{bid}.attn_q",
-        MODEL_TENSOR.ATTN_K            : "blk.{bid}.attn_k",
-        MODEL_TENSOR.ATTN_V            : "blk.{bid}.attn_v",
-        MODEL_TENSOR.ATTN_OUT          : "blk.{bid}.attn_output",
-        MODEL_TENSOR.ATTN_ROT_EMBD     : "blk.{bid}.attn_rot_embd",
-        MODEL_TENSOR.FFN_NORM          : "blk.{bid}.ffn_norm",
-        MODEL_TENSOR.FFN_GATE          : "blk.{bid}.ffn_gate",
-        MODEL_TENSOR.FFN_DOWN          : "blk.{bid}.ffn_down",
-        MODEL_TENSOR.FFN_UP            : "blk.{bid}.ffn_up",
-        },
-    MODEL_ARCH.FALCON : {
-        MODEL_TENSOR.TOKEN_EMBD  : "token_embd",
-        MODEL_TENSOR.OUTPUT_NORM : "output_norm",
-        MODEL_TENSOR.OUTPUT      : "output",
-        MODEL_TENSOR.ATTN_NORM   : "blk.{bid}.attn_norm",
-        MODEL_TENSOR.ATTN_NORM_2 : "blk.{bid}.attn_norm_2",
-        MODEL_TENSOR.ATTN_QKV    : "blk.{bid}.attn_qkv",
-        MODEL_TENSOR.ATTN_OUT    : "blk.{bid}.attn_output",
-        MODEL_TENSOR.FFN_DOWN    : "blk.{bid}.ffn_down",
-        MODEL_TENSOR.FFN_UP      : "blk.{bid}.ffn_up",
-        },
-    MODEL_ARCH.GPT2 : {
+    MODEL_ARCH.LLAMA: {
+        MODEL_TENSOR.TOKEN_EMBD:    "token_embd",
+        MODEL_TENSOR.OUTPUT_NORM:   "output_norm",
+        MODEL_TENSOR.OUTPUT:        "output",
+        MODEL_TENSOR.ROPE_FREQS:    "rope_freqs",
+        MODEL_TENSOR.ATTN_NORM:     "blk.{bid}.attn_norm",
+        MODEL_TENSOR.ATTN_Q:        "blk.{bid}.attn_q",
+        MODEL_TENSOR.ATTN_K:        "blk.{bid}.attn_k",
+        MODEL_TENSOR.ATTN_V:        "blk.{bid}.attn_v",
+        MODEL_TENSOR.ATTN_OUT:      "blk.{bid}.attn_output",
+        MODEL_TENSOR.ATTN_ROT_EMBD: "blk.{bid}.attn_rot_embd",
+        MODEL_TENSOR.FFN_NORM:      "blk.{bid}.ffn_norm",
+        MODEL_TENSOR.FFN_GATE:      "blk.{bid}.ffn_gate",
+        MODEL_TENSOR.FFN_DOWN:      "blk.{bid}.ffn_down",
+        MODEL_TENSOR.FFN_UP:        "blk.{bid}.ffn_up",
+    },
+    MODEL_ARCH.FALCON: {
+        MODEL_TENSOR.TOKEN_EMBD:  "token_embd",
+        MODEL_TENSOR.OUTPUT_NORM: "output_norm",
+        MODEL_TENSOR.OUTPUT:      "output",
+        MODEL_TENSOR.ATTN_NORM:   "blk.{bid}.attn_norm",
+        MODEL_TENSOR.ATTN_NORM_2: "blk.{bid}.attn_norm_2",
+        MODEL_TENSOR.ATTN_QKV:    "blk.{bid}.attn_qkv",
+        MODEL_TENSOR.ATTN_OUT:    "blk.{bid}.attn_output",
+        MODEL_TENSOR.FFN_DOWN:    "blk.{bid}.ffn_down",
+        MODEL_TENSOR.FFN_UP:      "blk.{bid}.ffn_up",
+    },
+    MODEL_ARCH.GPT2: {
         # TODO
-        },
+    },
     # TODO
-    }
+}
 
 # tensors that will not be serialized
 MODEL_TENSOR_SKIP = {
-    MODEL_ARCH.LLAMA : [
+    MODEL_ARCH.LLAMA: [
         MODEL_TENSOR.ROPE_FREQS,
         MODEL_TENSOR.ATTN_ROT_EMBD,
-        ],
-    }
+    ],
+}
+
 
 # TODO: the following helper functions should be removed
 #       instead, get_tensor_name_map should return tuples of (name, MODEL_TENSOR)
 #       however, my Python is very bad, and I couldn't figure out how to do this, hence these functions
 # REMOVE
-def should_skip_tensor_TMP(arch : MODEL_ARCH, n_blocks : int, name : str) -> bool:
+def should_skip_tensor_TMP(arch: MODEL_ARCH, n_blocks: int, name: str) -> bool:
     for skip in MODEL_TENSOR_SKIP.get(arch, []):
         for i in range(n_blocks):
             if name == MODEL_TENSOR_NAMES[arch][skip].format(bid=i):
@@ -160,151 +159,152 @@ def should_skip_tensor_TMP(arch : MODEL_ARCH, n_blocks : int, name : str) -> boo
 
     return False
 
-def get_tensor_name_map(arch : MODEL_ARCH, n_blocks : int) -> dict:
+
+def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict:
     tensor_map = {}
 
     # Token embeddings
     mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.TOKEN_EMBD, None)
 
-    tensor_map["gpt_neox.embed_in"]           = mapped_to # gptneox
-    tensor_map["transformer.wte"]             = mapped_to # gpt2 mpt
-    tensor_map["transformer.word_embeddings"] = mapped_to # falcon
-    tensor_map["model.embed_tokens"]          = mapped_to # llama-hf
-    tensor_map["tok_embeddings"]              = mapped_to # llama-pth
+    tensor_map["gpt_neox.embed_in"]           = mapped_to  # gptneox
+    tensor_map["transformer.wte"]             = mapped_to  # gpt2 mpt
+    tensor_map["transformer.word_embeddings"] = mapped_to  # falcon
+    tensor_map["model.embed_tokens"]          = mapped_to  # llama-hf
+    tensor_map["tok_embeddings"]              = mapped_to  # llama-pth
 
     # Position embeddings
     mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.POS_EMBD, None)
 
-    tensor_map["transformer.wpe"] = mapped_to # gpt2
+    tensor_map["transformer.wpe"] = mapped_to  # gpt2
 
     # Output
     mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT, None)
 
-    tensor_map["embed_out"] = mapped_to # gptneox
-    tensor_map["lm_head"]   = mapped_to # gpt2 mpt falcon llama-hf
-    tensor_map["output"]    = mapped_to # llama-pth
+    tensor_map["embed_out"] = mapped_to  # gptneox
+    tensor_map["lm_head"]   = mapped_to  # gpt2 mpt falcon llama-hf
+    tensor_map["output"]    = mapped_to  # llama-pth
 
     # Output norm
     mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT_NORM, None)
 
-    tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox
-    tensor_map["transformer.ln_f"]          = mapped_to # gpt2 falcon
-    tensor_map["transformer.norm_f"]        = mapped_to # mpt
-    tensor_map["model.norm"]                = mapped_to # llama-hf
-    tensor_map["norm"]                      = mapped_to # llama-pth
+    tensor_map["gpt_neox.final_layer_norm"] = mapped_to  # gptneox
+    tensor_map["transformer.ln_f"]          = mapped_to  # gpt2 falcon
+    tensor_map["transformer.norm_f"]        = mapped_to  # mpt
+    tensor_map["model.norm"]                = mapped_to  # llama-hf
+    tensor_map["norm"]                      = mapped_to  # llama-pth
 
     # Rope frequencies
     mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ROPE_FREQS, None)
 
-    tensor_map["rope.freqs"] = mapped_to # llama-pth
+    tensor_map["rope.freqs"] = mapped_to  # llama-pth
 
     # Attention and feed-forward blocks
-    for i in range(0,n_blocks):
+    for i in range(0, n_blocks):
         # Attention norm
         # TODO: is there are simpler way to write these 2 lines in Python?
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to # falcon40b
-        tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to # llama-pth
+        tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to  # falcon7b
+        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to  # falcon40b
+        tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to  # llama-pth
 
         # Attention norm 2
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM_2, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b
+        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to  # falcon40b
 
         # Attention query-key-value
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_QKV, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon
+        tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"]    = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_attn"]                    = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"]                 = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to  # falcon
 
         # Attention query
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_Q, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to # llama-pth
+        tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention.wq"]           = mapped_to  # llama-pth
 
         # Attention key
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_K, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to # llama-pth
+        tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention.wk"]           = mapped_to  # llama-pth
 
         # Attention value
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_V, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to # llama-pth
+        tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention.wv"]           = mapped_to  # llama-pth
 
         # Attention output
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_OUT, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to # llama-pth
+        tensor_map["gpt_neox.layers."+str(i)+".attention.dense"]    = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".attn.c_proj"]          = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".attn.out_proj"]   = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to  # falcon
+        tensor_map["model.layers."+str(i)+".self_attn.o_proj"]      = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention.wo"]                = mapped_to  # llama-pth
 
         # Rotary embeddings
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_ROT_EMBD, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["model.layers."+str(i)+".self_attn.rotary_emb.inv_freq"]  = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".attention.inner_attention.rope.freqs"] = mapped_to # llama-pth
+        tensor_map["model.layers."+str(i)+".self_attn.rotary_emb.inv_freq"]  = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".attention.inner_attention.rope.freqs"] = mapped_to  # llama-pth
 
         # Feed-forward norm
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_NORM, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to # mpt
-        tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to # llama-pth
+        tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".ln_2"]                       = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".norm_2"]                = mapped_to  # mpt
+        tensor_map["model.layers."+str(i)+".post_attention_layernorm"]    = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".ffn_norm"]                          = mapped_to  # llama-pth
 
         # Feed-forward up
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_UP, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to # llama-pth
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_fc"]            = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"]    = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"]   = mapped_to  # falcon
+        tensor_map["model.layers."+str(i)+".mlp.up_proj"]          = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w3"]            = mapped_to  # llama-pth
 
         # Feed-forward gate
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_GATE, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to # llama-pth
+        tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w1"]     = mapped_to  # llama-pth
 
         # Feed-forward down
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_DOWN, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox
-        tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to # gpt2
-        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to # mpt
-        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"]   = mapped_to # falcon
-        tensor_map["model.layers."+str(i)+".mlp.down_proj"]        = mapped_to # llama-hf
-        tensor_map["layers."+str(i)+".feed_forward.w2"]            = mapped_to # llama-pth
+        tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to  # gptneox
+        tensor_map["transformer.h."+str(i)+".mlp.c_proj"]          = mapped_to  # gpt2
+        tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"]  = mapped_to  # mpt
+        tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"]   = mapped_to  # falcon
+        tensor_map["model.layers."+str(i)+".mlp.down_proj"]        = mapped_to  # llama-hf
+        tensor_map["layers."+str(i)+".feed_forward.w2"]            = mapped_to  # llama-pth
 
     return tensor_map
 
@@ -312,6 +312,7 @@ def get_tensor_name_map(arch : MODEL_ARCH, n_blocks : int) -> dict:
 # implementation
 #
 
+
 class GGMLQuantizationType(IntEnum):
     F32 = 0
     F16 = 1
@@ -481,6 +482,19 @@ def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.
         self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
         self.ti_data_count += 1
 
+    def add_tensor(self, name: str, tensor: np.ndarray):
+        if not hasattr(self, "temp_file"):
+            self.temp_file = tempfile.SpooledTemporaryFile(mode="w+b", max_size=256*1024*1024)
+            self.temp_file.seek(0)
+
+        self.add_tensor_info(name, tensor.shape, tensor.dtype, tensor.nbytes)
+
+        tensor.tofile(self.temp_file)
+
+        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
+        if pad != 0:
+            self.temp_file.write(bytes([0] * pad))
+
     def write_tensor_data(self, tensor: np.ndarray):
         pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
         if pad != 0:
@@ -492,6 +506,19 @@ def write_tensor_data(self, tensor: np.ndarray):
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
+    def write_tensors_to_file(self):
+        self.write_ti_data_to_file()
+
+        pad = GGUFWriter.ggml_pad(self.fout.tell(), self.data_alignment) - self.fout.tell()
+        if pad != 0:
+            self.fout.write(bytes([0] * pad))
+
+        self.temp_file.seek(0)
+
+        shutil.copyfileobj(self.temp_file, self.fout)
+        self.flush()
+        self.temp_file.close()
+
     def flush(self):
         self.fout.flush()
 
@@ -513,9 +540,6 @@ def add_url(self, url: str):
     def add_description(self, description: str):
         self.add_string(KEY_GENERAL_DESCRIPTION, description)
 
-    def add_file_type(self, file_type: str):
-        self.add_string(KEY_GENERAL_FILE_TYPE, file_type)
-
     def add_source_url(self, url: str):
         self.add_string(KEY_GENERAL_SOURCE_URL, url)
 
@@ -618,23 +642,28 @@ def add_sep_token_id(self, id: int):
     def add_pad_token_id(self, id: int):
         self.add_uint32(KEY_TOKENIZER_PAD_ID, id)
 
+
 # Example usage:
 if __name__ == "__main__":
     # Example usage with a file
     gguf_writer = GGUFWriter("example.gguf", "llama")
 
+    gguf_writer.add_architecture()
+    gguf_writer.add_block_count(12)
     gguf_writer.add_uint32("answer", 42)  # Write a 32-bit integer
     gguf_writer.add_float32("answer_in_float", 42.0)  # Write a 32-bit float
     gguf_writer.add_custom_alignment(64)
+
     tensor1 = np.ones((32,), dtype=np.float32) * 100.0
-    tensor2 = np.ones((32,), dtype=np.float32) * 101.0
-    gguf_writer.add_tensor_info("tensor0", tensor1)
-    gguf_writer.add_tensor_info("tensor1", tensor2)
+    tensor2 = np.ones((64,), dtype=np.float32) * 101.0
+    tensor3 = np.ones((96,), dtype=np.float32) * 102.0
+
+    gguf_writer.add_tensor("tensor1", tensor1)
+    gguf_writer.add_tensor("tensor2", tensor2)
+    gguf_writer.add_tensor("tensor3", tensor3)
 
     gguf_writer.write_header_to_file()
     gguf_writer.write_kv_data_to_file()
-    gguf_writer.write_ti_data_to_file()
-    gguf_writer.write_tensor_data(tensor1)
-    gguf_writer.write_tensor_data(tensor2)
+    gguf_writer.write_tensors_to_file()
 
     gguf_writer.close()

From b668cd32960f32dc0c1481b44da6e1d15b046814 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 21:42:26 +0200
Subject: [PATCH 187/242] convert-gptneox-hf-to-gguf.py : fixes

---
 convert-gptneox-hf-to-gguf.py | 27 +++++++++++++--------------
 1 file changed, 13 insertions(+), 14 deletions(-)

diff --git a/convert-gptneox-hf-to-gguf.py b/convert-gptneox-hf-to-gguf.py
index 11cf19282aedc..1704b1f41075f 100644
--- a/convert-gptneox-hf-to-gguf.py
+++ b/convert-gptneox-hf-to-gguf.py
@@ -89,8 +89,8 @@ def count_model_parts(dir_model: str) -> int:
 # get number of model parts
 num_parts = count_model_parts(dir_model)
 
-llm_arch = "gptneox"
-gguf_writer = gguf.GGUFWriter(fname_out, arch=llm_arch)
+ARCH=gguf.MODEL_ARCH.GPTNEOX
+gguf_writer = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
 
 print("gguf: get model metadata")
 
@@ -194,7 +194,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = gguf.get_tensor_name_map(block_count)
+tensor_map = gguf.get_tensor_name_map(ARCH,block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
@@ -217,6 +217,8 @@ def count_model_parts(dir_model: str) -> int:
         if name.endswith(".attention.masked_bias") or name.endswith(".attention.bias") or name.endswith(".attention.rotary_emb.inv_freq"):
             continue
 
+        old_dtype = data.dtype
+
         # convert any unsupported data types to float32
         if data.dtype != torch.float16 and data.dtype != torch.float32:
             data = data.to(torch.float32)
@@ -233,24 +235,21 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype
-        old_dtype = data_dtype
+        data_dtype = data.dtype 
 
         # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data_dtype = np.float32
+        if ftype == 0 and data_dtype == np.float16:
+            data = data.astype(np.float32)
 
         # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
-        if ftype == 1 and data.dtype == np.float16 and n_dims == 1:
-            data_dtype = np.float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
 
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data_dtype = np.float16
-
-        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data_dtype))
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
 
-        data = data.astype(data_dtype)
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
         gguf_writer.add_tensor(name, data)
 

From 640ddc4259f11735c9f2d8cd44e2f81f939e7dc3 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 21:43:10 +0200
Subject: [PATCH 188/242] gguf.py : gptneox mapping

---
 gguf.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/gguf.py b/gguf.py
index 70418ef5f94d5..5fe4feb1bf642 100644
--- a/gguf.py
+++ b/gguf.py
@@ -121,6 +121,17 @@ class MODEL_TENSOR(IntEnum):
         MODEL_TENSOR.FFN_DOWN:      "blk.{bid}.ffn_down",
         MODEL_TENSOR.FFN_UP:        "blk.{bid}.ffn_up",
     },
+    MODEL_ARCH.GPTNEOX: {
+        MODEL_TENSOR.TOKEN_EMBD:    "token_embd",
+        MODEL_TENSOR.OUTPUT_NORM:   "output_norm",
+        MODEL_TENSOR.OUTPUT:        "output",
+        MODEL_TENSOR.ATTN_NORM:     "blk.{bid}.attn_norm",
+        MODEL_TENSOR.ATTN_QKV:      "blk.{bid}.attn_qkv",
+        MODEL_TENSOR.ATTN_OUT:      "blk.{bid}.attn_output",
+        MODEL_TENSOR.FFN_NORM:      "blk.{bid}.ffn_norm",
+        MODEL_TENSOR.FFN_DOWN:      "blk.{bid}.ffn_down",
+        MODEL_TENSOR.FFN_UP:        "blk.{bid}.ffn_up",
+    },
     MODEL_ARCH.FALCON: {
         MODEL_TENSOR.TOKEN_EMBD:  "token_embd",
         MODEL_TENSOR.OUTPUT_NORM: "output_norm",

From 9e2d4dd48e23870a056becde4886a2ba076914e8 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 21:43:48 +0200
Subject: [PATCH 189/242] convert-llama-hf-to-gguf.py : fixes

---
 convert-llama-hf-to-gguf.py | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index 885dd640abe0a..cc97bb3de61d8 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -86,8 +86,8 @@ def count_model_parts(dir_model: str) -> int:
 # get number of model parts
 num_parts = count_model_parts(dir_model)
 
-gguf_writer = gguf.GGUFWriter(fname_out, arch="llama")
-
+ARCH=gguf.MODEL_ARCH.LLAMA
+gguf_writer = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
 
 print("gguf: get model metadata")
 
@@ -214,7 +214,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = gguf.get_tensor_name_map(block_count)
+tensor_map = gguf.get_tensor_name_map(ARCH,block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
@@ -237,6 +237,8 @@ def count_model_parts(dir_model: str) -> int:
         if name.endswith(".rotary_emb.inv_freq"):
             continue
 
+        old_dtype = data.dtype
+
         # convert any unsupported data types to float32
         if data.dtype != torch.float16 and data.dtype != torch.float32:
             data = data.to(torch.float32)
@@ -254,26 +256,22 @@ def count_model_parts(dir_model: str) -> int:
             name = tensor_map[name[:-5]] + ".bias"
         else:
             print("Can not map tensor '" + name + "'")
-
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype
-        old_dtype = data_dtype
+        data_dtype = data.dtype 
 
         # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data_dtype = np.float32
+        if ftype == 0 and data_dtype == np.float16:
+            data = data.astype(np.float32)
 
         # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
         if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data_dtype = np.float32
+            data = data.astype(np.float32)
 
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data_dtype = np.float16
-
-        data = data.astype(data_dtype)
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
 
         print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 

From 3c1b7217a9ceddf06b2edc1b8825737561488d5b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 21:44:34 +0200
Subject: [PATCH 190/242] convert-llama-7b-pth-to-gguf.py : fixes

---
 convert-llama-7b-pth-to-gguf.py | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 72ad4d4ea7069..34b21d2f7537d 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -76,8 +76,9 @@ def count_model_parts(dir_model: str) -> int:
     print("gguf: Only models with a single datafile are supported.")
 
     sys.exit()
-llm_arch = "llama"
-gguf_writer = gguf.GGUFWriter(fname_out, arch=llm_arch)
+
+ARCH=gguf.MODEL_ARCH.LLAMA
+gguf_writer = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
 
 
 print("gguf: get model metadata")
@@ -195,7 +196,7 @@ def count_model_parts(dir_model: str) -> int:
 
 # TENSORS
 
-tensor_map = gguf.get_tensor_name_map(block_count)
+tensor_map = gguf.get_tensor_name_map(ARCH,block_count)
 
 # tensor info
 print("gguf: get tensor metadata")
@@ -213,6 +214,8 @@ def count_model_parts(dir_model: str) -> int:
         if name == "rope.freqs":
             continue
 
+        old_dtype = data.dtype
+
         # convert any unsupported data types to float32
         if data.dtype != torch.float16 and data.dtype != torch.float32:
             data = data.to(torch.float32)
@@ -229,24 +232,21 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype
-        old_dtype = data_dtype
+        data_dtype = data.dtype 
 
         # if f32 desired, convert any float16 to float32
-        if ftype == 0 and data.dtype == np.float16:
-            data_dtype = np.float32
+        if ftype == 0 and data_dtype == np.float16:
+            data = data.astype(np.float32)
 
         # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
         if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
-            data_dtype = np.float32
+            data = data.astype(np.float32)
 
         # if f16 desired, convert any float32 2-dim weight tensors to float16
-        if ftype == 1 and data.dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
-            data_dtype = np.float16
-
-        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data_dtype))
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
 
-        data = data.astype(data_dtype)
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data.dtype))
 
         gguf_writer.add_tensor(name, data)
 

From c20ae49b5917e5a961140a01b08a0e56e4c0fb6d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 23:23:17 +0200
Subject: [PATCH 191/242] ggml.h : reverse GGUF_MAGIC

---
 ggml.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml.h b/ggml.h
index ad12c133ee12d..27f4b0ed80c5e 100644
--- a/ggml.h
+++ b/ggml.h
@@ -214,7 +214,7 @@
 #define GGML_EXIT_SUCCESS 0
 #define GGML_EXIT_ABORTED 1
 
-#define GGUF_MAGIC   0x47475546 // "GGUF"
+#define GGUF_MAGIC   0x46554747 // "GGUF"
 #define GGUF_VERSION 1
 
 #define GGUF_DEFAULT_ALIGNMENT 32

From 147a99bd3a77bc587c9645d96aa516ea94fe4546 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 23:24:04 +0200
Subject: [PATCH 192/242] gguf.py : reverse GGUF_MAGIC

---
 gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gguf.py b/gguf.py
index 5fe4feb1bf642..c4290d64a2f4e 100644
--- a/gguf.py
+++ b/gguf.py
@@ -11,7 +11,7 @@
 # constants
 #
 
-GGUF_MAGIC             = 0x47475546
+GGUF_MAGIC             = 0x46554747
 GGUF_VERSION           = 1
 GGUF_DEFAULT_ALIGNMENT = 32
 

From d9e6890a51808a9e676f5b02d3073dcadcc88220 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Thu, 17 Aug 2023 23:34:21 +0200
Subject: [PATCH 193/242] test-tokenizer-0.cpp : fix warning

---
 tests/test-tokenizer-0.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp
index 9f5382679e380..e368ec3a67821 100644
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -41,7 +41,7 @@ static const std::map<std::string, std::vector<llama_token>> & k_tests() {
     };
 
     return _k_tests;
-};
+}
 
 int main(int argc, char **argv) {
     if (argc < 2) {

From 306070c89627950952d5376742d60fae30c0cb65 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 18 Aug 2023 01:06:27 +0200
Subject: [PATCH 194/242] llama.cpp : print kv general.name

---
 llama.cpp | 41 +++++++++++++++++++++++++----------------
 1 file changed, 25 insertions(+), 16 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 35945d3574acb..e97796f52b3be 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1329,6 +1329,8 @@ static void llama_model_load_internal(
 
     auto & hparams = model.hparams;
 
+    std::string general_name = "n/a";
+
     // read hparams
     {
         struct gguf_context * ctx = ml->ctx_gguf;
@@ -1347,6 +1349,10 @@ static void llama_model_load_internal(
             } \
         }
 
+        // get general kv
+        GGUF_GET(general_name, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.name");
+
+        // get hparams kv
         GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
         GGUF_GET(hparams.n_ctx_train,    gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
         GGUF_GET(hparams.n_embd,         gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.embedding_length");
@@ -1359,6 +1365,7 @@ static void llama_model_load_internal(
         // n_head_kv is optional, default to n_head
         hparams.n_head_kv = hparams.n_head;
         GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
+
 #undef GGUF_GET
 
         switch (hparams.n_layer) {
@@ -1422,22 +1429,24 @@ static void llama_model_load_internal(
     }
 
     {
-        LLAMA_LOG_INFO("%s: format      = %s\n",    __func__, llama_file_version_name(ml->file_version));
-        LLAMA_LOG_INFO("%s: n_vocab     = %u\n",    __func__, hparams.n_vocab);
-        LLAMA_LOG_INFO("%s: n_ctx_train = %u\n",    __func__, hparams.n_ctx_train);
-        LLAMA_LOG_INFO("%s: n_ctx       = %u\n",    __func__, hparams.n_ctx);
-        LLAMA_LOG_INFO("%s: n_embd      = %u\n",    __func__, hparams.n_embd);
-        LLAMA_LOG_INFO("%s: n_head      = %u\n",    __func__, hparams.n_head);
-        LLAMA_LOG_INFO("%s: n_head_kv   = %u\n",    __func__, hparams.n_head_kv);
-        LLAMA_LOG_INFO("%s: n_layer     = %u\n",    __func__, hparams.n_layer);
-        LLAMA_LOG_INFO("%s: n_rot       = %u\n",    __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        LLAMA_LOG_INFO("%s: n_gqa       = %u\n",    __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: f_norm_eps  = %.1e\n",  __func__, hparams.f_norm_rms_eps);
-        LLAMA_LOG_INFO("%s: n_ff        = %u\n",    __func__, hparams.n_ff);
-        LLAMA_LOG_INFO("%s: freq_base   = %.1f\n",  __func__, hparams.rope_freq_base);
-        LLAMA_LOG_INFO("%s: freq_scale  = %g\n",    __func__, hparams.rope_freq_scale);
-        LLAMA_LOG_INFO("%s: model type  = %s\n",    __func__, llama_model_type_name(model.type));
-        LLAMA_LOG_INFO("%s: model size  = %.2fB\n", __func__, ml->n_elements*1e-9);
+        LLAMA_LOG_INFO("%s: general.name = %s\n",    __func__, general_name.c_str());
+
+        LLAMA_LOG_INFO("%s: format       = %s\n",    __func__, llama_file_version_name(ml->file_version));
+        LLAMA_LOG_INFO("%s: n_vocab      = %u\n",    __func__, hparams.n_vocab);
+        LLAMA_LOG_INFO("%s: n_ctx_train  = %u\n",    __func__, hparams.n_ctx_train);
+        LLAMA_LOG_INFO("%s: n_ctx        = %u\n",    __func__, hparams.n_ctx);
+        LLAMA_LOG_INFO("%s: n_embd       = %u\n",    __func__, hparams.n_embd);
+        LLAMA_LOG_INFO("%s: n_head       = %u\n",    __func__, hparams.n_head);
+        LLAMA_LOG_INFO("%s: n_head_kv    = %u\n",    __func__, hparams.n_head_kv);
+        LLAMA_LOG_INFO("%s: n_layer      = %u\n",    __func__, hparams.n_layer);
+        LLAMA_LOG_INFO("%s: n_rot        = %u\n",    __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        LLAMA_LOG_INFO("%s: n_gqa        = %u\n",    __func__, hparams.n_gqa());
+        LLAMA_LOG_INFO("%s: f_norm_eps   = %.1e\n",  __func__, hparams.f_norm_rms_eps);
+        LLAMA_LOG_INFO("%s: n_ff         = %u\n",    __func__, hparams.n_ff);
+        LLAMA_LOG_INFO("%s: freq_base    = %.1f\n",  __func__, hparams.rope_freq_base);
+        LLAMA_LOG_INFO("%s: freq_scale   = %g\n",    __func__, hparams.rope_freq_scale);
+        LLAMA_LOG_INFO("%s: model type   = %s\n",    __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: model size   = %.2f B\n", __func__, ml->n_elements*1e-9);
 
         // TODO: print number of tensors for each quantization
     }

From b275de745d687a50003342d76066db42de5b7352 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 18 Aug 2023 03:34:30 +0200
Subject: [PATCH 195/242] llama.cpp : get special token kv and linefeed token
 id

---
 llama.cpp | 45 +++++++++++++++++++++++++++++++++++++--------
 1 file changed, 37 insertions(+), 8 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index e97796f52b3be..11519e8a54385 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -761,7 +761,6 @@ struct llama_kv_cache {
 struct llama_vocab {
     // TODO:
     // - add a vector of merges
-    // - add members for bos/eos/pad/sep tokens
     //   so that we can pass it to different types of tokenizers with a common interface
 
     using id    = int32_t;
@@ -774,6 +773,14 @@ struct llama_vocab {
 
     std::unordered_map<token, id> token_to_id;
     std::vector<token_score> id_to_token;
+
+    id special_bos_id = -1;
+    id special_eos_id = -1;
+    id special_unk_id = -1;
+    id special_sep_id = -1;
+    id special_pad_id = -1;
+
+    id linefeed_id = -1;
 };
 
 struct llama_model {
@@ -1349,9 +1356,6 @@ static void llama_model_load_internal(
             } \
         }
 
-        // get general kv
-        GGUF_GET(general_name, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.name");
-
         // get hparams kv
         GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
         GGUF_GET(hparams.n_ctx_train,    gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
@@ -1366,6 +1370,16 @@ static void llama_model_load_internal(
         hparams.n_head_kv = hparams.n_head;
         GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
 
+        // get general kv
+        GGUF_GET(general_name, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.name");
+
+        // special tokens
+        GGUF_GET(vocab.special_bos_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.bos_token_id");
+        GGUF_GET(vocab.special_eos_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.eos_token_id");
+        GGUF_GET(vocab.special_unk_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.unknown_token_id");
+        GGUF_GET(vocab.special_sep_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.separator_token_id");
+        GGUF_GET(vocab.special_pad_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.padding_token_id");
+
 #undef GGUF_GET
 
         switch (hparams.n_layer) {
@@ -1407,12 +1421,12 @@ static void llama_model_load_internal(
 
         const int token_idx = gguf_find_key(ctx, "tokenizer.ggml.tokens");
         if (token_idx == -1) {
-            throw std::runtime_error("cannot find token list in GGUF file\n");
+            throw std::runtime_error("cannot find tokenizer vocab in model file\n");
         }
 
         const int score_idx = gguf_find_key(ctx, "tokenizer.ggml.scores");
         if (score_idx == -1) {
-            throw std::runtime_error("cannot find token scores list in GGUF file\n");
+            throw std::runtime_error("cannot find tokenizer scores in model file\n");
         }
 
         const float * scores = (const float * ) gguf_get_arr_data(ctx, score_idx);
@@ -1425,12 +1439,16 @@ static void llama_model_load_internal(
             auto & tok_score = vocab.id_to_token[i];
             tok_score.tok = std::move(word);
             tok_score.score = scores[i];
+
+            if( tok_score.tok == "<0x0A>" ) {
+                vocab.linefeed_id = i;
+            }
         }
+
     }
 
     {
-        LLAMA_LOG_INFO("%s: general.name = %s\n",    __func__, general_name.c_str());
-
+        // hparams
         LLAMA_LOG_INFO("%s: format       = %s\n",    __func__, llama_file_version_name(ml->file_version));
         LLAMA_LOG_INFO("%s: n_vocab      = %u\n",    __func__, hparams.n_vocab);
         LLAMA_LOG_INFO("%s: n_ctx_train  = %u\n",    __func__, hparams.n_ctx_train);
@@ -1449,6 +1467,17 @@ static void llama_model_load_internal(
         LLAMA_LOG_INFO("%s: model size   = %.2f B\n", __func__, ml->n_elements*1e-9);
 
         // TODO: print number of tensors for each quantization
+
+        // general kv
+        LLAMA_LOG_INFO("%s: general.name = %s\n",    __func__, general_name.c_str());
+
+        // special tokens
+        if( vocab.special_bos_id != -1 ) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].tok.c_str() ); }
+        if( vocab.special_eos_id != -1 ) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].tok.c_str() ); }
+        if( vocab.special_unk_id != -1 ) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].tok.c_str() ); }
+        if( vocab.special_sep_id != -1 ) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].tok.c_str() ); }
+        if( vocab.special_pad_id != -1 ) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].tok.c_str() ); }
+        if( vocab.linefeed_id    != -1 ) { LLAMA_LOG_INFO( "%s: LF token  = %d '%s'\n", __func__, vocab.linefeed_id,    vocab.id_to_token[vocab.linefeed_id].tok.c_str() ); }
     }
 
     if (vocab_only) {

From aa3efe87c8da17e9f20aaf744629e9a009c8d209 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 10:36:45 +0300
Subject: [PATCH 196/242] llama : print number of tensors per type + print arch
 + style

---
 llama.cpp | 72 ++++++++++++++++++++++++++++++-------------------------
 1 file changed, 40 insertions(+), 32 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 11519e8a54385..d57791ed9a25f 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -126,13 +126,6 @@ static void llama_log_callback_default(llama_log_level level, const char * text,
 // helpers
 //
 
-template<typename T>
-static std::string to_string(const T & val) {
-    std::stringstream ss;
-    ss << val;
-    return ss.str();
-}
-
 static void zeros(std::ofstream & file, size_t n) {
     char zero = 0;
     for (size_t i = 0; i < n; ++i) {
@@ -1070,10 +1063,14 @@ struct llama_model_loader {
             LLAMA_LOG_INFO("%s: loaded meta data with %d key-value pairs and %d tensors from %s (version %s)\n",
                     __func__, n_kv, n_tensors, fname.c_str(), llama_file_version_name(file_version));
 
+            std::map<enum ggml_type, uint32_t> n_type;
+
             for (int i = 0; i < n_tensors; i++) {
                 const char * name = gguf_get_tensor_name(ctx_gguf, i);
                 struct ggml_tensor * meta = ggml_get_tensor(ctx_meta, name);
 
+                n_type[meta->type]++;
+
                 LLAMA_LOG_INFO("%s: - tensor %4d: %32s %-8s [ %s ]\n", __func__, i, name, ggml_type_name(meta->type), llama_format_tensor_shape(meta).c_str());
             }
 
@@ -1083,6 +1080,15 @@ struct llama_model_loader {
 
                 LLAMA_LOG_INFO("%s: - kv %3d: %42s %-8s\n", __func__, i, name, gguf_type_name(type));
             }
+
+            // print type counts
+            for (auto & kv : n_type) {
+                if (kv.second == 0) {
+                    continue;
+                }
+
+                LLAMA_LOG_INFO("%s: - type %4s: %4d tensors\n", __func__, ggml_type_name(kv.first), kv.second);
+            }
         }
 
         if (!llama_mmap::SUPPORTED) {
@@ -1337,6 +1343,7 @@ static void llama_model_load_internal(
     auto & hparams = model.hparams;
 
     std::string general_name = "n/a";
+    std::string general_arch = "n/a";
 
     // read hparams
     {
@@ -1372,6 +1379,7 @@ static void llama_model_load_internal(
 
         // get general kv
         GGUF_GET(general_name, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.name");
+        GGUF_GET(general_arch, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.architecture");
 
         // special tokens
         GGUF_GET(vocab.special_bos_id, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "tokenizer.ggml.bos_token_id");
@@ -1440,47 +1448,47 @@ static void llama_model_load_internal(
             tok_score.tok = std::move(word);
             tok_score.score = scores[i];
 
-            if( tok_score.tok == "<0x0A>" ) {
+            // determine the newline token: 0x0A == 10 == '\n'
+            if (tok_score.tok == "<0x0A>") {
                 vocab.linefeed_id = i;
             }
         }
-
     }
 
     {
         // hparams
-        LLAMA_LOG_INFO("%s: format       = %s\n",    __func__, llama_file_version_name(ml->file_version));
-        LLAMA_LOG_INFO("%s: n_vocab      = %u\n",    __func__, hparams.n_vocab);
-        LLAMA_LOG_INFO("%s: n_ctx_train  = %u\n",    __func__, hparams.n_ctx_train);
-        LLAMA_LOG_INFO("%s: n_ctx        = %u\n",    __func__, hparams.n_ctx);
-        LLAMA_LOG_INFO("%s: n_embd       = %u\n",    __func__, hparams.n_embd);
-        LLAMA_LOG_INFO("%s: n_head       = %u\n",    __func__, hparams.n_head);
-        LLAMA_LOG_INFO("%s: n_head_kv    = %u\n",    __func__, hparams.n_head_kv);
-        LLAMA_LOG_INFO("%s: n_layer      = %u\n",    __func__, hparams.n_layer);
-        LLAMA_LOG_INFO("%s: n_rot        = %u\n",    __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
-        LLAMA_LOG_INFO("%s: n_gqa        = %u\n",    __func__, hparams.n_gqa());
-        LLAMA_LOG_INFO("%s: f_norm_eps   = %.1e\n",  __func__, hparams.f_norm_rms_eps);
-        LLAMA_LOG_INFO("%s: n_ff         = %u\n",    __func__, hparams.n_ff);
-        LLAMA_LOG_INFO("%s: freq_base    = %.1f\n",  __func__, hparams.rope_freq_base);
-        LLAMA_LOG_INFO("%s: freq_scale   = %g\n",    __func__, hparams.rope_freq_scale);
-        LLAMA_LOG_INFO("%s: model type   = %s\n",    __func__, llama_model_type_name(model.type));
+        LLAMA_LOG_INFO("%s: format       = %s\n",     __func__, llama_file_version_name(ml->file_version));
+        LLAMA_LOG_INFO("%s: arch         = %s\n",     __func__, general_arch.c_str());
+        LLAMA_LOG_INFO("%s: n_vocab      = %u\n",     __func__, hparams.n_vocab);
+        LLAMA_LOG_INFO("%s: n_ctx_train  = %u\n",     __func__, hparams.n_ctx_train);
+        LLAMA_LOG_INFO("%s: n_ctx        = %u\n",     __func__, hparams.n_ctx);
+        LLAMA_LOG_INFO("%s: n_embd       = %u\n",     __func__, hparams.n_embd);
+        LLAMA_LOG_INFO("%s: n_head       = %u\n",     __func__, hparams.n_head);
+        LLAMA_LOG_INFO("%s: n_head_kv    = %u\n",     __func__, hparams.n_head_kv);
+        LLAMA_LOG_INFO("%s: n_layer      = %u\n",     __func__, hparams.n_layer);
+        LLAMA_LOG_INFO("%s: n_rot        = %u\n",     __func__, hparams.n_rot); // a.k.a. n_embd_head, n_head_dim
+        LLAMA_LOG_INFO("%s: n_gqa        = %u\n",     __func__, hparams.n_gqa());
+        LLAMA_LOG_INFO("%s: f_norm_eps   = %.1e\n",   __func__, hparams.f_norm_rms_eps);
+        LLAMA_LOG_INFO("%s: n_ff         = %u\n",     __func__, hparams.n_ff);
+        LLAMA_LOG_INFO("%s: freq_base    = %.1f\n",   __func__, hparams.rope_freq_base);
+        LLAMA_LOG_INFO("%s: freq_scale   = %g\n",     __func__, hparams.rope_freq_scale);
+        LLAMA_LOG_INFO("%s: model type   = %s\n",     __func__, llama_model_type_name(model.type));
         LLAMA_LOG_INFO("%s: model size   = %.2f B\n", __func__, ml->n_elements*1e-9);
 
-        // TODO: print number of tensors for each quantization
-
         // general kv
         LLAMA_LOG_INFO("%s: general.name = %s\n",    __func__, general_name.c_str());
 
         // special tokens
-        if( vocab.special_bos_id != -1 ) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].tok.c_str() ); }
-        if( vocab.special_eos_id != -1 ) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].tok.c_str() ); }
-        if( vocab.special_unk_id != -1 ) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].tok.c_str() ); }
-        if( vocab.special_sep_id != -1 ) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].tok.c_str() ); }
-        if( vocab.special_pad_id != -1 ) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].tok.c_str() ); }
-        if( vocab.linefeed_id    != -1 ) { LLAMA_LOG_INFO( "%s: LF token  = %d '%s'\n", __func__, vocab.linefeed_id,    vocab.id_to_token[vocab.linefeed_id].tok.c_str() ); }
+        if (vocab.special_bos_id != -1) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].tok.c_str() ); }
+        if (vocab.special_eos_id != -1) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].tok.c_str() ); }
+        if (vocab.special_unk_id != -1) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].tok.c_str() ); }
+        if (vocab.special_sep_id != -1) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].tok.c_str() ); }
+        if (vocab.special_pad_id != -1) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].tok.c_str() ); }
+        if (vocab.linefeed_id    != -1) { LLAMA_LOG_INFO( "%s: LF token  = %d '%s'\n", __func__, vocab.linefeed_id,    vocab.id_to_token[vocab.linefeed_id].tok.c_str() );    }
     }
 
     if (vocab_only) {
+        LLAMA_LOG_INFO("%s: vocab only - skipping tensors\n", __func__);
         return;
     }
 

From e35f8c744ee91974ec7ee60dc57f46088917acf8 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 12:39:09 +0300
Subject: [PATCH 197/242] tests : update vocab file with new magic

---
 models/ggml-vocab-llama.gguf | Bin 467382 -> 467382 bytes
 1 file changed, 0 insertions(+), 0 deletions(-)

diff --git a/models/ggml-vocab-llama.gguf b/models/ggml-vocab-llama.gguf
index 8548445f8eac41d07a65e74ef3fc5d4a098c2615..c50db67dc52023444e607bde7b684da5d631e564 100644
GIT binary patch
delta 58
zcmdmXSZ3Q{85Vc<P`8OJevASegE<*l)*MwgZ%$xrPhbRLCLm@8Viq7~1!A`C35@Jv
F3IG(F5Gw!x

delta 58
zcmdmXSZ3Q{85XxtclU`bevBL&gE<*l_L{euHYYH)CoqCA6A&{4F$)m00x{e61V;8S
F1pp3&5CZ@J


From dea5be61d7244792007b081f12558184a4eefea7 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 12:42:38 +0300
Subject: [PATCH 198/242] editorconfig : fix whitespaces

---
 convert-gptneox-hf-to-gguf.py   | 2 +-
 convert-llama-7b-pth-to-gguf.py | 2 +-
 convert-llama-hf-to-gguf.py     | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/convert-gptneox-hf-to-gguf.py b/convert-gptneox-hf-to-gguf.py
index 1704b1f41075f..e49a619cf289d 100644
--- a/convert-gptneox-hf-to-gguf.py
+++ b/convert-gptneox-hf-to-gguf.py
@@ -235,7 +235,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:
diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 34b21d2f7537d..89b58e132ef74 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -232,7 +232,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:
diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index cc97bb3de61d8..7d4571f4731dc 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -259,7 +259,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:

From 660ca9bbcae88d2602de9b755d7b3a82c97e35ce Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 14:56:36 +0300
Subject: [PATCH 199/242] llama : re-order functions

---
 llama.cpp                  | 284 +++++++++++++++++++------------------
 llama.h                    |  86 +++++------
 tests/test-tokenizer-1.cpp |   3 +-
 3 files changed, 187 insertions(+), 186 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 35e11fb7e5e14..790f21ce9683b 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -880,7 +880,6 @@ struct llama_context {
     std::vector<uint8_t> work_buffer;
 
     // memory buffers used to evaluate the model
-    // TODO: move in llama_state
     llama_buffer buf_compute;
 
 #ifdef LLAMA_USE_ALLOCATOR
@@ -2547,7 +2546,7 @@ struct llama_tokenizer {
     std::map<std::string, std::pair<int, int> > rev_merge;
 };
 
-static std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & raw_text, bool bos, bool escape) {
+static std::vector<llama_vocab::id> llama_tokenize_internal(const llama_vocab & vocab, const std::string & raw_text, bool bos, bool escape) {
     llama_tokenizer tokenizer(vocab);
     std::vector<llama_vocab::id> output;
 
@@ -4399,6 +4398,30 @@ void llama_free(struct llama_context * ctx) {
     delete ctx;
 }
 
+int llama_n_vocab(const struct llama_context * ctx) {
+    return ctx->model.vocab.id_to_token.size();
+}
+
+int llama_n_ctx(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_ctx;
+}
+
+int llama_n_embd(const struct llama_context * ctx) {
+    return ctx->model.hparams.n_embd;
+}
+
+int llama_n_vocab_from_model(const struct llama_model * model) {
+    return model->vocab.id_to_token.size();
+}
+
+int llama_n_ctx_from_model(const struct llama_model * model) {
+    return model->hparams.n_ctx;
+}
+
+int llama_n_embd_from_model(const struct llama_model * model) {
+    return model->hparams.n_embd;
+}
+
 int llama_model_quantize(
         const char * fname_inp,
         const char * fname_out,
@@ -4876,25 +4899,33 @@ int llama_eval_export(struct llama_context * ctx, const char * fname) {
     return 0;
 }
 
-int llama_tokenize_with_model(
-    const struct llama_model * model,
-                  const char * text,
-                 llama_token * tokens,
-                         int   n_max_tokens,
-                        bool   add_bos) {
-    auto escape = llama_vocab_type(model->vocab) == "spm";
-    auto res = llama_tokenize(model->vocab, text, add_bos, escape);
+float * llama_get_logits(struct llama_context * ctx) {
+    return ctx->logits.data();
+}
 
-    if (n_max_tokens < (int) res.size()) {
-        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
-        return -((int) res.size());
-    }
+float * llama_get_embeddings(struct llama_context * ctx) {
+    return ctx->embedding.data();
+}
 
-    for (size_t i = 0; i < res.size(); i++) {
-        tokens[i] = res[i];
-    }
+int llama_get_vocab(
+        const struct llama_context * ctx,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
+}
 
-    return res.size();
+int llama_get_vocab_from_model(
+        const struct llama_model * model,
+        const char * * strings,
+        float  * scores,
+        int capacity) {
+    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
+    for (int i = 0; i<n; ++i) {
+        strings[i] = model->vocab.id_to_token[i].tok.c_str();
+        scores[i]  = model->vocab.id_to_token[i].score;
+    }
+    return n;
 }
 
 int llama_tokenize(
@@ -4906,31 +4937,13 @@ int llama_tokenize(
     return llama_tokenize_with_model(&ctx->model, text, tokens, n_max_tokens, add_bos);
 }
 
-std::vector<llama_token> llama_tokenize(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos) {
-    int length = text.length() + add_bos;
-    std::vector<llama_token> result(length);
-    length = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -length);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(length);
-    }
-    return result;
-}
-
 int llama_tokenize_bpe(
         struct llama_context * ctx,
                   const char * text,
                  llama_token * tokens,
                          int   n_max_tokens,
                         bool   add_bos) {
-    auto res = llama_tokenize(ctx->model.vocab, text, add_bos, false);
+    auto res = llama_tokenize_internal(ctx->model.vocab, text, add_bos, false);
 
     if (n_max_tokens < (int) res.size()) {
         LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
@@ -4944,75 +4957,41 @@ int llama_tokenize_bpe(
     return res.size();
 }
 
-std::vector<llama_token> llama_tokenize_bpe(
-        struct llama_context * ctx,
-           const std::string & text,
+int llama_tokenize_with_model(
+    const struct llama_model * model,
+                  const char * text,
+                 llama_token * tokens,
+                         int   n_max_tokens,
                         bool   add_bos) {
-    int length = text.length() + add_bos;
-    std::vector<llama_token> result(length);
-    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -length);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(length);
-    }
-    return result;
-}
-
-int llama_n_vocab_from_model(const struct llama_model * model) {
-    return model->vocab.id_to_token.size();
-}
-
-int llama_n_ctx_from_model(const struct llama_model * model) {
-    return model->hparams.n_ctx;
-}
-
-int llama_n_embd_from_model(const struct llama_model * model) {
-    return model->hparams.n_embd;
-}
-
-int llama_n_vocab(const struct llama_context * ctx) {
-    return ctx->model.vocab.id_to_token.size();
-}
-
-int llama_n_ctx(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_ctx;
-}
+    auto escape = llama_vocab_type(model->vocab) == "spm";
+    auto res = llama_tokenize_internal(model->vocab, text, add_bos, escape);
 
-int llama_n_embd(const struct llama_context * ctx) {
-    return ctx->model.hparams.n_embd;
-}
+    if (n_max_tokens < (int) res.size()) {
+        LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
+        return -((int) res.size());
+    }
 
-int llama_get_vocab_from_model(
-        const struct llama_model * model,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
-    for (int i = 0; i<n; ++i) {
-        strings[i] = model->vocab.id_to_token[i].tok.c_str();
-        scores[i]  = model->vocab.id_to_token[i].score;
+    for (size_t i = 0; i < res.size(); i++) {
+        tokens[i] = res[i];
     }
-    return n;
-}
 
-int llama_get_vocab(
-        const struct llama_context * ctx,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
+    return res.size();
 }
 
-float * llama_get_logits(struct llama_context * ctx) {
-    return ctx->logits.data();
+int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * buf, int length) {
+    return llama_token_to_str_with_model(&ctx->model, token, buf, length);
 }
 
-float * llama_get_embeddings(struct llama_context * ctx) {
-    return ctx->embedding.data();
+int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * buf, int length) {
+    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
+        std::string result = ctx->model.vocab.id_to_token[token].tok;
+        if (length < (int) result.length()) {
+            return -result.length();
+        }
+        memcpy(buf, result.c_str(), result.length());
+        return result.length();
+    }
+    return 0;
 }
 
 // does not write null-terminator to str
@@ -5049,50 +5028,6 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
     return 0;
 }
 
-int llama_token_to_str(const struct llama_context * ctx, llama_token token, char * buf, int length) {
-    return llama_token_to_str_with_model(&ctx->model, token, buf, length);
-}
-
-std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_token_to_str(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
-int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * buf, int length) {
-    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
-        std::string result = ctx->model.vocab.id_to_token[token].tok;
-        if (length < (int) result.length()) {
-            return -result.length();
-        }
-        memcpy(buf, result.c_str(), result.length());
-        return result.length();
-    }
-    return 0;
-}
-
-std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
 llama_token llama_token_bos(void) {
     return 1;
 }
@@ -5165,6 +5100,73 @@ const char * llama_print_system_info(void) {
     return s.c_str();
 }
 
+
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    // upper limit for the number of tokens
+    int n_tokens = text.length() + add_bos;
+    std::vector<llama_token> result(n_tokens);
+    n_tokens = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -n_tokens);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(n_tokens);
+    }
+    return result;
+}
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int length = text.length() + add_bos;
+    std::vector<llama_token> result(length);
+    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        assert(check == -length);
+        GGML_UNUSED(check);
+    } else {
+        result.resize(length);
+    }
+    return result;
+}
+
+std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
+    if (length < 0) {
+        result.resize(-length);
+        int check = llama_token_to_str(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
+    } else {
+        result.resize(length);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
+std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+    if (length < 0) {
+        result.resize(-length);
+        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -length);
+    } else {
+        result.resize(length);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
+
 // For internal test use
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
     return ctx->model.tensors_by_name;
diff --git a/llama.h b/llama.h
index 2e407a1db3ca3..f6015a324ad15 100644
--- a/llama.h
+++ b/llama.h
@@ -199,20 +199,15 @@ extern "C" {
     LLAMA_API struct llama_context_params llama_context_default_params(void);
     LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
 
-    LLAMA_API int  llama_max_devices(void);
-    LLAMA_API bool llama_mmap_supported(void);
-    LLAMA_API bool llama_mlock_supported(void);
-
     // TODO: not great API - very likely to change
     // Initialize the llama + ggml backend
     // If numa is true, use NUMA optimizations
     // Call once at the start of the program
     LLAMA_API void llama_backend_init(bool numa);
+
     // Call once at the end of the program - currently only used for MPI
     LLAMA_API void llama_backend_free(void);
 
-    LLAMA_API int64_t llama_time_us(void);
-
     LLAMA_API struct llama_model * llama_load_model_from_file(
                              const char * path_model,
             struct llama_context_params   params);
@@ -223,10 +218,23 @@ extern "C" {
                      struct llama_model * model,
             struct llama_context_params   params);
 
-
     // Frees all allocated memory
     LLAMA_API void llama_free(struct llama_context * ctx);
 
+    LLAMA_API int64_t llama_time_us(void);
+
+    LLAMA_API int  llama_max_devices    (void);
+    LLAMA_API bool llama_mmap_supported (void);
+    LLAMA_API bool llama_mlock_supported(void);
+
+    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
+    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
+    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
+
+    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
+    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
+    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
+
     // Returns 0 on success
     LLAMA_API int llama_model_quantize(
             const char * fname_inp,
@@ -300,6 +308,31 @@ extern "C" {
     // IMPORTANT: do not use for anything else other than debugging and testing!
     LLAMA_API int llama_eval_export(struct llama_context * ctx, const char * fname);
 
+    // Token logits obtained from the last call to llama_eval()
+    // The logits for the last token are stored in the last row
+    // Can be mutated in order to change the probabilities of the next token
+    // Rows: n_tokens
+    // Cols: n_vocab
+    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
+
+    // Get the embeddings for the input
+    // shape: [n_embd] (1-dimensional)
+    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
+
+    // Get the vocabulary as output parameters.
+    // Returns number of results.
+    LLAMA_API int llama_get_vocab(
+            const struct llama_context * ctx,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
+    LLAMA_API int llama_get_vocab_from_model(
+              const struct llama_model * model,
+                          const char * * strings,
+                                 float * scores,
+                                   int   capacity);
+
     // Convert the provided text into tokens.
     // The tokens pointer must be large enough to hold the resulting tokens.
     // Returns the number of tokens on success, no more than n_max_tokens
@@ -326,39 +359,6 @@ extern "C" {
                              int   n_max_tokens,
                             bool   add_bos);
 
-    LLAMA_API int llama_n_vocab(const struct llama_context * ctx);
-    LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
-    LLAMA_API int llama_n_embd (const struct llama_context * ctx);
-
-    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
-    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
-    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
-
-    // Get the vocabulary as output parameters.
-    // Returns number of results.
-    LLAMA_API int llama_get_vocab(
-            const struct llama_context * ctx,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    LLAMA_API int llama_get_vocab_from_model(
-              const struct llama_model * model,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
-
-    // Token logits obtained from the last call to llama_eval()
-    // The logits for the last token are stored in the last row
-    // Can be mutated in order to change the probabilities of the next token
-    // Rows: n_tokens
-    // Cols: n_vocab
-    LLAMA_API float * llama_get_logits(struct llama_context * ctx);
-
-    // Get the embeddings for the input
-    // shape: [n_embd] (1-dimensional)
-    LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
-
     // Token Id -> String. Uses the vocabulary in the provided context
     // Does not write null terminator to the buffer
     LLAMA_API int llama_token_to_str(
@@ -379,9 +379,9 @@ extern "C" {
                                   char * buf,
                                   int    length);
     // Special tokens
-    LLAMA_API llama_token llama_token_bos(void);  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos(void);  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl(void);   // next-line
+    LLAMA_API llama_token llama_token_bos(/*struct llama_model * model*/ void);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(/*struct llama_model * model*/ void);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl (/*struct llama_model * model*/ void);  // next-line
 
     // Grammar
     //
diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
index a44f38cd714ad..620e3f2757e4b 100644
--- a/tests/test-tokenizer-1.cpp
+++ b/tests/test-tokenizer-1.cpp
@@ -114,8 +114,7 @@ int main(int argc, char **argv) {
         }
         std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str), false);
         if (tokens.size() == 1) {
-            fprintf(stderr, "%s : info: %s tokenized to %d \n",
-                __func__, str.c_str(), tokens[0]);
+            fprintf(stderr, "%s : info: %s tokenized to %d \n", __func__, str.c_str(), tokens[0]);
         }
     }
 

From 2d6c2c757ca414fd11456f35292c2df141018d4f Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 16:22:48 +0300
Subject: [PATCH 200/242] llama : remove C++ API + reorganize common source in
 /common dir

---
 CMakeLists.txt                             | 12 ++-
 Makefile                                   | 14 ++--
 common/CMakeLists.txt                      | 20 +++++
 {examples => common}/common.cpp            | 72 ++++++++++++++++
 {examples => common}/common.h              | 23 ++++-
 {examples => common}/console.cpp           |  0
 {examples => common}/console.h             |  0
 {examples => common}/grammar-parser.cpp    |  0
 {examples => common}/grammar-parser.h      |  0
 examples/CMakeLists.txt                    | 21 -----
 examples/quantize-stats/quantize-stats.cpp |  1 -
 llama.cpp                                  | 98 +++++-----------------
 llama.h                                    | 31 +------
 tests/CMakeLists.txt                       | 12 +--
 tests/test-grammar-parser.cpp              |  3 +-
 tests/test-llama-grammar.cpp               |  6 +-
 tests/test-tokenizer-0.cpp                 |  2 +-
 tests/test-tokenizer-1.cpp                 |  2 +-
 18 files changed, 169 insertions(+), 148 deletions(-)
 create mode 100644 common/CMakeLists.txt
 rename {examples => common}/common.cpp (93%)
 rename {examples => common}/common.h (90%)
 rename {examples => common}/console.cpp (100%)
 rename {examples => common}/console.h (100%)
 rename {examples => common}/grammar-parser.cpp (100%)
 rename {examples => common}/grammar-parser.h (100%)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 6c5a3e09e9c9a..bb63ef98e3013 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -497,9 +497,11 @@ else()
 endif()
 
 #
-# Build libraries
+# libraries
 #
 
+# ggml
+
 add_library(ggml OBJECT
             ggml.c
             ggml.h
@@ -524,6 +526,8 @@ if (BUILD_SHARED_LIBS)
     install(TARGETS ggml_shared LIBRARY)
 endif()
 
+# llama
+
 add_library(llama
             llama.cpp
             llama.h
@@ -545,6 +549,10 @@ if (BUILD_SHARED_LIBS)
     install(TARGETS llama LIBRARY)
 endif()
 
+#
+# install
+#
+
 include(GNUInstallDirs)
 install(
     FILES convert.py
@@ -583,6 +591,8 @@ endif()
 # programs, examples and tests
 #
 
+add_subdirectory(common)
+
 if (LLAMA_BUILD_TESTS AND NOT CMAKE_JS_VERSION)
     include(CTest)
     add_subdirectory(tests)
diff --git a/Makefile b/Makefile
index e212c63cae531..d31acc450b261 100644
--- a/Makefile
+++ b/Makefile
@@ -45,8 +45,8 @@ OPT = -Ofast
 else
 OPT = -O3
 endif
-CFLAGS   = -I.              $(OPT) -std=c11   -fPIC
-CXXFLAGS = -I. -I./examples $(OPT) -std=c++11 -fPIC
+CFLAGS   = -I.            $(OPT) -std=c11   -fPIC
+CXXFLAGS = -I. -I./common $(OPT) -std=c++11 -fPIC
 LDFLAGS  =
 
 ifdef LLAMA_DEBUG
@@ -332,13 +332,13 @@ OBJS += ggml-alloc.o
 llama.o: llama.cpp ggml.h ggml-alloc.h ggml-cuda.h ggml-metal.h llama.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
-common.o: examples/common.cpp examples/common.h
+common.o: common/common.cpp common/common.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
-console.o: examples/console.cpp examples/console.h
+console.o: common/console.cpp common/console.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
-grammar-parser.o: examples/grammar-parser.cpp examples/grammar-parser.h
+grammar-parser.o: common/grammar-parser.cpp common/grammar-parser.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
 libllama.so: llama.o ggml.o $(OBJS)
@@ -388,7 +388,7 @@ embd-input-test: $(LIB_PRE)embdinput$(DSO_EXT) examples/embd-input/embd-input-te
 gguf: examples/gguf/gguf.cpp                                  build-info.h ggml.o llama.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
-train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o $(OBJS)
+train-text-from-scratch: examples/train-text-from-scratch/train-text-from-scratch.cpp    build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
 convert-llama2c-to-ggml: examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp    build-info.h ggml.o llama.o $(OBJS)
@@ -421,7 +421,7 @@ vdot: pocs/vdot/vdot.cpp ggml.o $(OBJS)
 tests/test-llama-grammar: tests/test-llama-grammar.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.txt,$^) -o $@ $(LDFLAGS)
 
-tests/test-grammar-parser: tests/test-grammar-parser.cpp examples/grammar-parser.cpp build-info.h ggml.o llama.o common.o $(OBJS)
+tests/test-grammar-parser: tests/test-grammar-parser.cpp build-info.h ggml.o llama.o common.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.txt,$^) -o $@ $(LDFLAGS)
 
 tests/test-double-float: tests/test-double-float.cpp build-info.h ggml.o llama.o common.o $(OBJS)
diff --git a/common/CMakeLists.txt b/common/CMakeLists.txt
new file mode 100644
index 0000000000000..dead56118bac8
--- /dev/null
+++ b/common/CMakeLists.txt
@@ -0,0 +1,20 @@
+# common
+
+set(TARGET common)
+
+add_library(${TARGET} OBJECT
+    common.h
+    common.cpp
+    console.h
+    console.cpp
+    grammar-parser.h
+    grammar-parser.cpp
+    )
+
+if (BUILD_SHARED_LIBS)
+    set_target_properties(${TARGET} PROPERTIES POSITION_INDEPENDENT_CODE ON)
+endif()
+
+target_include_directories(${TARGET} PUBLIC .)
+target_compile_features(${TARGET} PUBLIC cxx_std_11)
+target_link_libraries(${TARGET} PRIVATE llama)
diff --git a/examples/common.cpp b/common/common.cpp
similarity index 93%
rename from examples/common.cpp
rename to common/common.cpp
index ea6c9d4991fe2..8ea7bdda0e51b 100644
--- a/examples/common.cpp
+++ b/common/common.cpp
@@ -636,6 +636,10 @@ std::string gpt_random_prompt(std::mt19937 & rng) {
     return "The";
 }
 
+//
+// Model utils
+//
+
 struct llama_context_params llama_context_params_from_gpt_params(const gpt_params & params) {
     auto lparams = llama_context_default_params();
 
@@ -689,3 +693,71 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
 
     return std::make_tuple(model, lctx);
 }
+
+//
+// Vocab utils
+//
+
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    // upper limit for the number of tokens
+    int n_tokens = text.length() + add_bos;
+    std::vector<llama_token> result(n_tokens);
+    n_tokens = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+    return result;
+}
+
+std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int n_tokens = llama_token_to_str(ctx, token, result.data(), result.size());
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_token_to_str(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos) {
+    int n_tokens = text.length() + add_bos;
+    std::vector<llama_token> result(n_tokens);
+    n_tokens = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+    return result;
+}
+
+std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int n_tokens = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
diff --git a/examples/common.h b/common/common.h
similarity index 90%
rename from examples/common.h
rename to common/common.h
index de04b4c5b2f03..50145c93290f6 100644
--- a/examples/common.h
+++ b/common/common.h
@@ -2,7 +2,6 @@
 
 #pragma once
 
-#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include <string>
@@ -105,3 +104,25 @@ std::string gpt_random_prompt(std::mt19937 & rng);
 
 std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(const gpt_params & params);
 struct llama_context_params llama_context_params_from_gpt_params(const gpt_params & params);
+
+//
+// Vocab utils
+//
+
+std::vector<llama_token> llama_tokenize(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::vector<llama_token> llama_tokenize_bpe(
+        struct llama_context * ctx,
+           const std::string & text,
+                        bool   add_bos);
+
+std::string llama_token_to_str(
+        const struct llama_context * ctx,
+                       llama_token   token);
+
+std::string llama_token_to_str_bpe(
+    const struct llama_context * ctx,
+                   llama_token   token);
diff --git a/examples/console.cpp b/common/console.cpp
similarity index 100%
rename from examples/console.cpp
rename to common/console.cpp
diff --git a/examples/console.h b/common/console.h
similarity index 100%
rename from examples/console.h
rename to common/console.h
diff --git a/examples/grammar-parser.cpp b/common/grammar-parser.cpp
similarity index 100%
rename from examples/grammar-parser.cpp
rename to common/grammar-parser.cpp
diff --git a/examples/grammar-parser.h b/common/grammar-parser.h
similarity index 100%
rename from examples/grammar-parser.h
rename to common/grammar-parser.h
diff --git a/examples/CMakeLists.txt b/examples/CMakeLists.txt
index d53652815e83e..d2176c910c299 100644
--- a/examples/CMakeLists.txt
+++ b/examples/CMakeLists.txt
@@ -6,27 +6,6 @@ find_package(Threads REQUIRED)
 
 # ...
 
-# common
-
-set(TARGET common)
-
-add_library(${TARGET} OBJECT
-    common.h
-    common.cpp
-    console.h
-    console.cpp
-    grammar-parser.h
-    grammar-parser.cpp
-    )
-
-if (BUILD_SHARED_LIBS)
-    set_target_properties(${TARGET} PROPERTIES POSITION_INDEPENDENT_CODE ON)
-endif()
-
-target_include_directories(${TARGET} PUBLIC .)
-target_compile_features(${TARGET} PUBLIC cxx_std_11)
-target_link_libraries(${TARGET} PRIVATE llama)
-
 # examples
 
 include_directories(${CMAKE_CURRENT_SOURCE_DIR})
diff --git a/examples/quantize-stats/quantize-stats.cpp b/examples/quantize-stats/quantize-stats.cpp
index 60ace876503ab..06ce18f09a346 100644
--- a/examples/quantize-stats/quantize-stats.cpp
+++ b/examples/quantize-stats/quantize-stats.cpp
@@ -1,7 +1,6 @@
 #include "ggml.h"
 #include "build-info.h"
 
-#define LLAMA_API_CPP // TODO: eliminate me
 #define LLAMA_API_INTERNAL
 #include "llama.h"
 
diff --git a/llama.cpp b/llama.cpp
index 2509c9dcf844b..87e2d5b11ac0d 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -6,7 +6,6 @@
 #include <cstdio>
 #endif
 
-#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
 
 #include "ggml.h"
@@ -277,7 +276,7 @@ struct llama_file {
         }
     }
 
-    uint32_t read_u32() {
+    uint32_t read_u32() const {
         uint32_t ret;
         read_raw(&ret, sizeof(ret));
         return ret;
@@ -559,10 +558,24 @@ struct llama_mlock {
 
 typedef void (*offload_func_t)(struct ggml_tensor * tensor);
 
-void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
+static void llama_nop(struct ggml_tensor * tensor) { // don't offload by default
     (void) tensor;
 }
 
+static std::string llama_token_to_text(const struct llama_context * ctx, llama_token token) {
+    std::vector<char> result(8, 0);
+    const int n_tokens = llama_token_to_str(ctx, token, result.data(), result.size());
+    if (n_tokens < 0) {
+        result.resize(-n_tokens);
+        int check = llama_token_to_str(ctx, token, result.data(), result.size());
+        GGML_ASSERT(check == -n_tokens);
+    } else {
+        result.resize(n_tokens);
+    }
+
+    return std::string(result.data(), result.size());
+}
+
 //
 // globals
 //
@@ -3286,16 +3299,16 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
     std::vector<llama_grammar_candidate>                              candidates_grammar;
 
     for (size_t i = 0; i < candidates->size; ++i) {
-        const llama_token id  = candidates->data[i].id;
-        std::string       str = llama_token_to_str(ctx, id);
+        const llama_token id   = candidates->data[i].id;
+        const std::string text = llama_token_to_text(ctx, id);
         if (id == eos) {
             if (!allow_eos) {
                 candidates->data[i].logit = -INFINITY;
             }
-        } else if (str.empty()) {
+        } else if (text.empty()) {
             candidates->data[i].logit = -INFINITY;
         } else {
-            candidates_decoded.push_back(decode_utf8(str.c_str(), grammar->partial_utf8));
+            candidates_decoded.push_back(decode_utf8(text.c_str(), grammar->partial_utf8));
             candidates_grammar.push_back({ i, candidates_decoded.back().first.data(), candidates_decoded.back().second });
         }
     }
@@ -3495,10 +3508,10 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
         GGML_ASSERT(false);
     }
 
-    const std::string str = llama_token_to_str(ctx, token);
+    const std::string text = llama_token_to_text(ctx, token);
 
     // Note terminating 0 in decoded string
-    const auto   decoded     = decode_utf8(str.c_str(), grammar->partial_utf8);
+    const auto   decoded     = decode_utf8(text.c_str(), grammar->partial_utf8);
     const auto & code_points = decoded.first;
     for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
         grammar->stacks = llama_grammar_accept(grammar->rules, grammar->stacks, *it);
@@ -5144,73 +5157,6 @@ const char * llama_print_system_info(void) {
     return s.c_str();
 }
 
-
-std::vector<llama_token> llama_tokenize(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos) {
-    // upper limit for the number of tokens
-    int n_tokens = text.length() + add_bos;
-    std::vector<llama_token> result(n_tokens);
-    n_tokens = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (n_tokens < 0) {
-        result.resize(-n_tokens);
-        int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -n_tokens);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(n_tokens);
-    }
-    return result;
-}
-
-std::vector<llama_token> llama_tokenize_bpe(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos) {
-    int length = text.length() + add_bos;
-    std::vector<llama_token> result(length);
-    length = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_tokenize_bpe(ctx, text.c_str(), result.data(), result.size(), add_bos);
-        assert(check == -length);
-        GGML_UNUSED(check);
-    } else {
-        result.resize(length);
-    }
-    return result;
-}
-
-std::string llama_token_to_str(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        int check = llama_token_to_str(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
-std::string llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token) {
-    std::vector<char> result(8, 0);
-    const int length = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-    if (length < 0) {
-        result.resize(-length);
-        const int check = llama_token_to_str_bpe(ctx, token, result.data(), result.size());
-        GGML_ASSERT(check == -length);
-    } else {
-        result.resize(length);
-    }
-
-    return std::string(result.data(), result.size());
-}
-
-
 // For internal test use
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx) {
     return ctx->model.tensors_by_name;
diff --git a/llama.h b/llama.h
index 8e5c339d3e3fd..5126b8193ab41 100644
--- a/llama.h
+++ b/llama.h
@@ -472,43 +472,16 @@ extern "C" {
 }
 #endif
 
-// C++ API, will be moving to common.h soon (TM)
-#ifdef LLAMA_API_CPP
+// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
+#ifdef LLAMA_API_INTERNAL
 
 #include <vector>
 #include <string>
 
-//
-// Vocab utils
-//
-
-std::vector<llama_token> llama_tokenize(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos);
-
-std::vector<llama_token> llama_tokenize_bpe(
-        struct llama_context * ctx,
-           const std::string & text,
-                        bool   add_bos);
-
-std::string llama_token_to_str(
-        const struct llama_context * ctx,
-                       llama_token   token);
-
-std::string llama_token_to_str_bpe(
-    const struct llama_context * ctx,
-                   llama_token   token);
-
-// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
-#ifdef LLAMA_API_INTERNAL
-
 struct ggml_tensor;
 
 const std::vector<std::pair<std::string, struct ggml_tensor *>>& llama_internal_get_tensor_map(struct llama_context * ctx);
 
-#endif // LLAMA_API_CPP
-
 #endif // LLAMA_API_INTERNAL
 
 #endif // LLAMA_H
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index d19ec5af0ef20..4ccefe9322322 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -2,7 +2,7 @@ function(llama_build_executable source)
     get_filename_component(TEST_TARGET ${source} NAME_WE)
     add_executable(${TEST_TARGET} ${source})
     install(TARGETS ${TEST_TARGET} RUNTIME)
-    target_link_libraries(${TEST_TARGET} PRIVATE llama)
+    target_link_libraries(${TEST_TARGET} PRIVATE llama common)
 endfunction()
 
 function(llama_test_executable name source)
@@ -17,7 +17,7 @@ function(llama_build_and_test_executable source)
     get_filename_component(TEST_TARGET ${source} NAME_WE)
     add_executable(${TEST_TARGET} ${source})
     install(TARGETS ${TEST_TARGET} RUNTIME)
-    target_link_libraries(${TEST_TARGET} PRIVATE llama)
+    target_link_libraries(${TEST_TARGET} PRIVATE llama common)
     add_test(NAME ${TEST_TARGET} COMMAND $<TARGET_FILE:${TEST_TARGET}> ${ARGN})
 endfunction()
 
@@ -26,11 +26,11 @@ llama_build_and_test_executable(test-quantize-fns.cpp)
 llama_build_and_test_executable(test-quantize-perf.cpp)
 llama_build_and_test_executable(test-sampling.cpp)
 llama_build_executable(test-tokenizer-0.cpp)
-llama_test_executable(test-tokenizer-0.llama test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
+llama_test_executable (test-tokenizer-0.llama test-tokenizer-0.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
 llama_build_executable(test-tokenizer-1.cpp)
-llama_test_executable(test-tokenizer-1.llama  test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
+llama_test_executable (test-tokenizer-1.llama test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
 #llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf)
-llama_build_and_test_executable(test-grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp)
-llama_build_and_test_executable(test-llama-grammar.cpp  ${CMAKE_CURRENT_SOURCE_DIR}/../examples/grammar-parser.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../llama.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../examples/common.cpp)
+llama_build_and_test_executable(test-grammar-parser.cpp)
+llama_build_and_test_executable(test-llama-grammar.cpp)
 llama_build_and_test_executable(test-grad0.cpp) # SLOW
 # llama_build_and_test_executable(test-opt.cpp) # SLOW
diff --git a/tests/test-grammar-parser.cpp b/tests/test-grammar-parser.cpp
index 7022988b40b10..a0b5b043df868 100644
--- a/tests/test-grammar-parser.cpp
+++ b/tests/test-grammar-parser.cpp
@@ -3,7 +3,8 @@
 #endif
 
 #include "llama.h"
-#include "examples/grammar-parser.cpp"
+#include "grammar-parser.h"
+
 #include <cassert>
 
 int main()
diff --git a/tests/test-llama-grammar.cpp b/tests/test-llama-grammar.cpp
index 81c31e9e2e5d2..73dd33dd286a5 100644
--- a/tests/test-llama-grammar.cpp
+++ b/tests/test-llama-grammar.cpp
@@ -2,9 +2,9 @@
 #undef NDEBUG
 #endif
 
-#include "llama.cpp"
-#include "examples/common.cpp"
-#include "examples/grammar-parser.cpp"
+#include "llama.cpp" // TODO: not great
+#include "grammar-parser.h"
+
 #include <cassert>
 
 int main()
diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp
index e368ec3a67821..81764565b5710 100644
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -1,5 +1,5 @@
-#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
+#include "common.h"
 
 #include <cstdio>
 #include <string>
diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
index 620e3f2757e4b..5841f73399b76 100644
--- a/tests/test-tokenizer-1.cpp
+++ b/tests/test-tokenizer-1.cpp
@@ -1,5 +1,5 @@
-#define LLAMA_API_CPP // TODO: eliminate me
 #include "llama.h"
+#include "common.h"
 
 #include <cassert>
 #include <cstdio>

From 035d51145736d44e68ac0ef26eb00a8d159df47d Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 17:06:34 +0300
Subject: [PATCH 201/242] llama : minor API updates

---
 llama.cpp | 64 +++++++++++++++++++++++++++++--------------------------
 llama.h   | 37 +++++++++++++++++++++-----------
 2 files changed, 58 insertions(+), 43 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 87e2d5b11ac0d..4f2af46647891 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -3002,7 +3002,7 @@ void llama_grammar_free(struct llama_grammar * grammar) {
 //
 
 void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(candidates->size > 0);
+    GGML_ASSERT(candidates->size > 0);
 
     const int64_t t_start_sample_us = ggml_time_us();
 
@@ -3282,7 +3282,7 @@ void llama_sample_frequency_and_presence_penalties(struct llama_context * ctx, l
 }
 
 void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * candidates, const struct llama_grammar * grammar) {
-    assert(ctx);
+    GGML_ASSERT(ctx);
     const int64_t t_start_sample_us = ggml_time_us();
 
     bool allow_eos = false;
@@ -3342,10 +3342,12 @@ void llama_sample_classifier_free_guidance(
                          float   scale) {
     int64_t t_start_sample_us = ggml_time_us();
 
-    assert(ctx);
+    GGML_ASSERT(ctx);
+
     auto n_vocab = llama_n_vocab(ctx);
-    assert(n_vocab == (int)candidates->size);
-    assert(!candidates->sorted);
+
+    GGML_ASSERT(n_vocab == (int)candidates->size);
+    GGML_ASSERT(!candidates->sorted);
 
     std::vector<float> logits_base;
     logits_base.reserve(candidates->size);
@@ -3369,7 +3371,8 @@ void llama_sample_classifier_free_guidance(
 }
 
 llama_token llama_sample_token_mirostat(struct llama_context * ctx, llama_token_data_array * candidates, float tau, float eta, int m, float * mu) {
-    assert(ctx);
+    GGML_ASSERT(ctx);
+
     auto N = float(llama_n_vocab(ctx));
     int64_t t_start_sample_us;
     t_start_sample_us = ggml_time_us();
@@ -3475,7 +3478,8 @@ llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_da
 }
 
 llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
-    assert(ctx);
+    GGML_ASSERT(ctx);
+
     const int64_t t_start_sample_us = ggml_time_us();
     llama_sample_softmax(nullptr, candidates);
 
@@ -4463,18 +4467,22 @@ int llama_n_embd(const struct llama_context * ctx) {
     return ctx->model.hparams.n_embd;
 }
 
-int llama_n_vocab_from_model(const struct llama_model * model) {
+int llama_model_n_vocab(const struct llama_model * model) {
     return model->vocab.id_to_token.size();
 }
 
-int llama_n_ctx_from_model(const struct llama_model * model) {
+int llama_model_n_ctx(const struct llama_model * model) {
     return model->hparams.n_ctx;
 }
 
-int llama_n_embd_from_model(const struct llama_model * model) {
+int llama_model_n_embd(const struct llama_model * model) {
     return model->hparams.n_embd;
 }
 
+int llama_model_type(const struct llama_model * model, char * buf, size_t buf_size) {
+    return snprintf(buf, buf_size, "LLaMA %s %s", llama_model_type_name(model->type), llama_model_ftype_name(model->ftype));
+}
+
 int llama_model_quantize(
         const char * fname_inp,
         const char * fname_out,
@@ -4965,14 +4973,10 @@ int llama_get_vocab(
         const char * * strings,
         float  * scores,
         int capacity) {
-    return llama_get_vocab_from_model(&ctx->model, strings, scores, capacity);
-}
-
-int llama_model_type(const struct llama_model * model, char * buf, size_t buf_size) {
-    return snprintf(buf, buf_size, "LLaMA %s %s", llama_model_type_name(model->type), llama_model_ftype_name(model->ftype));
+    return llama_model_get_vocab(&ctx->model, strings, scores, capacity);
 }
 
-int llama_get_vocab_from_model(
+int llama_model_get_vocab(
         const struct llama_model * model,
         const char * * strings,
         float  * scores,
@@ -4985,6 +4989,18 @@ int llama_get_vocab_from_model(
     return n;
 }
 
+llama_token llama_token_bos(void) {
+    return 1;
+}
+
+llama_token llama_token_eos(void) {
+    return 2;
+}
+
+llama_token llama_token_nl(void) {
+    return 13;
+}
+
 int llama_tokenize(
         struct llama_context * ctx,
                   const char * text,
@@ -5040,7 +5056,7 @@ int llama_token_to_str(const struct llama_context * ctx, llama_token token, char
 }
 
 int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * buf, int length) {
-    if (0 <= token && token < llama_n_vocab_from_model(&ctx->model)) {
+    if (0 <= token && token < llama_model_n_vocab(&ctx->model)) {
         std::string result = ctx->model.vocab.id_to_token[token].tok;
         if (length < (int) result.length()) {
             return -result.length();
@@ -5053,7 +5069,7 @@ int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token,
 
 // does not write null-terminator to str
 int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * buf, int length) {
-    if (0 <= token && token < llama_n_vocab_from_model(model)) {
+    if (0 <= token && token < llama_model_n_vocab(model)) {
         if (llama_is_normal_token(model->vocab, token)) {
             std::string result = model->vocab.id_to_token[token].tok;
             if (llama_vocab_type(model->vocab) == "spm") {
@@ -5085,18 +5101,6 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
     return 0;
 }
 
-llama_token llama_token_bos(void) {
-    return 1;
-}
-
-llama_token llama_token_eos(void) {
-    return 2;
-}
-
-llama_token llama_token_nl(void) {
-    return 13;
-}
-
 struct llama_timings llama_get_timings(struct llama_context * ctx) {
     struct llama_timings result = {
         /*.t_start_ms  =*/ 1e-3 * ctx->t_start_us,
diff --git a/llama.h b/llama.h
index 5126b8193ab41..e2b28afbb1c45 100644
--- a/llama.h
+++ b/llama.h
@@ -199,7 +199,6 @@ extern "C" {
     LLAMA_API struct llama_context_params llama_context_default_params(void);
     LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
 
-    // TODO: not great API - very likely to change
     // Initialize the llama + ggml backend
     // If numa is true, use NUMA optimizations
     // Call once at the start of the program
@@ -231,9 +230,9 @@ extern "C" {
     LLAMA_API int llama_n_ctx  (const struct llama_context * ctx);
     LLAMA_API int llama_n_embd (const struct llama_context * ctx);
 
-    LLAMA_API int llama_n_vocab_from_model(const struct llama_model * model);
-    LLAMA_API int llama_n_ctx_from_model  (const struct llama_model * model);
-    LLAMA_API int llama_n_embd_from_model (const struct llama_model * model);
+    LLAMA_API int llama_model_n_vocab(const struct llama_model * model);
+    LLAMA_API int llama_model_n_ctx  (const struct llama_model * model);
+    LLAMA_API int llama_model_n_embd (const struct llama_model * model);
 
     // Get a string describing the model type
     LLAMA_API int llama_model_type(const struct llama_model * model, char * buf, size_t buf_size);
@@ -259,9 +258,9 @@ extern "C" {
 
     LLAMA_API int llama_model_apply_lora_from_file(
             const struct llama_model * model,
-                      const char * path_lora,
-                      const char * path_base_model,
-                             int   n_threads);
+                          const char * path_lora,
+                          const char * path_base_model,
+                                 int   n_threads);
 
     // Returns the number of tokens in the KV cache
     LLAMA_API int llama_get_kv_cache_token_count(const struct llama_context * ctx);
@@ -322,6 +321,10 @@ extern "C" {
     // shape: [n_embd] (1-dimensional)
     LLAMA_API float * llama_get_embeddings(struct llama_context * ctx);
 
+    //
+    // Vocab
+    //
+
     // Get the vocabulary as output parameters.
     // Returns number of results.
     LLAMA_API int llama_get_vocab(
@@ -330,17 +333,25 @@ extern "C" {
                                  float * scores,
                                    int   capacity);
 
-    LLAMA_API int llama_get_vocab_from_model(
+    LLAMA_API int llama_model_get_vocab(
               const struct llama_model * model,
                           const char * * strings,
                                  float * scores,
                                    int   capacity);
 
+    // Special tokens
+    LLAMA_API llama_token llama_token_bos(/*struct llama_model * model*/ void);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(/*struct llama_model * model*/ void);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl (/*struct llama_model * model*/ void);  // next-line
+
+    //
+    // Tokenization
+    //
+
     // Convert the provided text into tokens.
     // The tokens pointer must be large enough to hold the resulting tokens.
     // Returns the number of tokens on success, no more than n_max_tokens
     // Returns a negative number on failure - the number of tokens that would have been returned
-    // TODO: not sure if correct
     LLAMA_API int llama_tokenize(
             struct llama_context * ctx,
                       const char * text,
@@ -381,13 +392,11 @@ extern "C" {
                            llama_token   token,
                                   char * buf,
                                   int    length);
-    // Special tokens
-    LLAMA_API llama_token llama_token_bos(/*struct llama_model * model*/ void);  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos(/*struct llama_model * model*/ void);  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl (/*struct llama_model * model*/ void);  // next-line
 
+    //
     // Grammar
     //
+
     LLAMA_API struct llama_grammar * llama_grammar_init(
             const llama_grammar_element ** rules,
                                  size_t    n_rules,
@@ -395,7 +404,9 @@ extern "C" {
 
     LLAMA_API void llama_grammar_free(struct llama_grammar * grammar);
 
+    //
     // Sampling functions
+    //
 
     /// @details Repetition penalty described in CTRL academic paper https://arxiv.org/abs/1909.05858, with negative logit fix.
     LLAMA_API void llama_sample_repetition_penalty(struct llama_context * ctx, llama_token_data_array * candidates, const llama_token * last_tokens, size_t last_tokens_size, float penalty);

From 5d2656d6707399860c7902c2f48fd1c23b3fcd40 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 17:29:20 +0300
Subject: [PATCH 202/242] llama : avoid hardcoded special tokens

---
 common/common.cpp                             |  8 +++-
 common/common.h                               |  6 ++-
 examples/embd-input/embd-input-lib.cpp        |  6 +--
 examples/llama-bench/llama-bench.cpp          |  4 +-
 examples/main/main.cpp                        | 19 ++++----
 examples/perplexity/perplexity.cpp            |  2 +-
 examples/server/server.cpp                    | 14 +++---
 examples/simple/simple.cpp                    |  2 +-
 .../train-text-from-scratch.cpp               | 16 +++----
 llama.cpp                                     | 43 ++++++++-----------
 llama.h                                       |  6 +--
 11 files changed, 61 insertions(+), 65 deletions(-)

diff --git a/common/common.cpp b/common/common.cpp
index 8ea7bdda0e51b..d7e1a5725b483 100644
--- a/common/common.cpp
+++ b/common/common.cpp
@@ -427,7 +427,7 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
             }
             params.hellaswag_tasks = std::stoi(argv[i]);
         } else if (arg == "--ignore-eos") {
-            params.logit_bias[llama_token_eos()] = -INFINITY;
+            params.ignore_eos = true;
         } else if (arg == "--no-penalize-nl") {
             params.penalize_nl = false;
         } else if (arg == "-l" || arg == "--logit-bias") {
@@ -662,7 +662,7 @@ struct llama_context_params llama_context_params_from_gpt_params(const gpt_param
     return lparams;
 }
 
-std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(const gpt_params & params) {
+std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(gpt_params & params) {
     auto lparams = llama_context_params_from_gpt_params(params);
 
     llama_model * model  = llama_load_model_from_file(params.model.c_str(), lparams);
@@ -691,6 +691,10 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
         }
     }
 
+    if (params.ignore_eos) {
+        params.logit_bias[llama_token_eos(lctx)] = -INFINITY;
+    }
+
     return std::make_tuple(model, lctx);
 }
 
diff --git a/common/common.h b/common/common.h
index 50145c93290f6..c50a6edfc4124 100644
--- a/common/common.h
+++ b/common/common.h
@@ -32,7 +32,6 @@ struct gpt_params {
     float   rope_freq_scale                 = 1.0f;     // RoPE frequency scaling factor
 
     // sampling parameters
-    std::unordered_map<llama_token, float> logit_bias; // logit bias for specific tokens
     int32_t top_k             = 40;    // <= 0 to use vocab size
     float   top_p             = 0.95f; // 1.0 = disabled
     float   tfs_z             = 1.00f; // 1.0 = disabled
@@ -46,6 +45,8 @@ struct gpt_params {
     float   mirostat_tau      = 5.00f; // target entropy
     float   mirostat_eta      = 0.10f; // learning rate
 
+    std::unordered_map<llama_token, float> logit_bias; // logit bias for specific tokens
+
     // Classifier-Free Guidance
     // https://arxiv.org/abs/2306.17806
     std::string cfg_negative_prompt;       // string to help guidance
@@ -81,6 +82,7 @@ struct gpt_params {
     bool simple_io         = false; // improves compatibility with subprocesses and limited consoles
 
     bool input_prefix_bos  = false; // prefix BOS to user inputs, preceding input_prefix
+    bool ignore_eos        = false; // ignore generated EOS tokens
     bool instruct          = false; // instruction mode (used for Alpaca models)
     bool penalize_nl       = true;  // consider newlines as a repeatable token
     bool perplexity        = false; // compute perplexity over the prompt
@@ -102,7 +104,7 @@ std::string gpt_random_prompt(std::mt19937 & rng);
 // Model utils
 //
 
-std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(const gpt_params & params);
+std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(gpt_params & params);
 struct llama_context_params llama_context_params_from_gpt_params(const gpt_params & params);
 
 //
diff --git a/examples/embd-input/embd-input-lib.cpp b/examples/embd-input/embd-input-lib.cpp
index 2185b9b0e2839..8a6ad882e8fa8 100644
--- a/examples/embd-input/embd-input-lib.cpp
+++ b/examples/embd-input/embd-input-lib.cpp
@@ -167,7 +167,7 @@ llama_token sampling_id(struct MyModel* mymodel) {
         llama_token_data_array candidates_p = { candidates.data(), candidates.size(), false };
 
         // TODO: Apply penalties
-        // float nl_logit = logits[llama_token_nl()];
+        // float nl_logit = logits[llama_token_nl(ctx)];
         // auto last_n_repeat = std::min(std::min((int)last_n_tokens.size(), repeat_last_n), n_ctx);
         // llama_sample_repetition_penalty(ctx, &candidates_p,
         //      last_n_tokens.data() + last_n_tokens.size() - last_n_repeat,
@@ -176,7 +176,7 @@ llama_token sampling_id(struct MyModel* mymodel) {
         // last_n_tokens.data() + last_n_tokens.size() - last_n_repeat,
         // last_n_repeat, alpha_frequency, alpha_presence);
         // if (!penalize_nl) {
-        //     logits[llama_token_nl()] = nl_logit;
+        //     logits[llama_token_nl(ctx)] = nl_logit;
         // }
 
         if (temp <= 0) {
@@ -211,7 +211,7 @@ const char * sampling(struct MyModel * mymodel) {
     llama_context * ctx = mymodel->ctx;
     int id = sampling_id(mymodel);
     static std::string ret;
-    if (id == llama_token_eos()) {
+    if (id == llama_token_eos(ctx)) {
         ret = "</s>";
     } else {
         ret = llama_token_to_str(ctx, id);
diff --git a/examples/llama-bench/llama-bench.cpp b/examples/llama-bench/llama-bench.cpp
index 1e2b892fae51b..d11fff2889f65 100755
--- a/examples/llama-bench/llama-bench.cpp
+++ b/examples/llama-bench/llama-bench.cpp
@@ -851,7 +851,7 @@ struct sql_printer : public printer {
 };
 
 static void test_prompt(llama_context * ctx, int n_prompt, int n_past, int n_batch, int n_threads) {
-    std::vector<llama_token> tokens(n_batch, llama_token_bos());
+    std::vector<llama_token> tokens(n_batch, llama_token_bos(ctx));
     int n_processed = 0;
     while (n_processed < n_prompt) {
         int n_tokens = std::min(n_prompt - n_processed, n_batch);
@@ -861,7 +861,7 @@ static void test_prompt(llama_context * ctx, int n_prompt, int n_past, int n_bat
 }
 
 static void test_gen(llama_context * ctx, int n_gen, int n_past, int n_threads) {
-    llama_token token = llama_token_bos();
+    llama_token token = llama_token_bos(ctx);
     for (int i = 0; i < n_gen; i++) {
         llama_eval(ctx, &token, 1, n_past + i, n_threads);
     }
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index 5c2f64883cc2c..388e1f7d7fe02 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -143,7 +143,7 @@ int main(int argc, char ** argv) {
         {
             fprintf(stderr, "%s: testing memory usage for n_batch = %d, n_ctx = %d\n", __func__, params.n_batch, params.n_ctx);
 
-            const std::vector<llama_token> tmp(params.n_batch, llama_token_bos());
+            const std::vector<llama_token> tmp(params.n_batch, llama_token_bos(ctx));
             llama_eval(ctx, tmp.data(), tmp.size(), params.n_ctx, params.n_threads);
         }
 
@@ -345,10 +345,9 @@ int main(int argc, char ** argv) {
         fprintf(stderr, "\n");
 
         {
-            auto it = params.logit_bias.find(llama_token_eos());
+            auto it = params.logit_bias.find(llama_token_eos(ctx));
             if (it != params.logit_bias.end() && it->second == -INFINITY) {
-                fprintf(stderr,
-                    "%s: warning: EOS token is disabled, which will cause most grammars to fail\n", __func__);
+                fprintf(stderr, "%s: warning: EOS token is disabled, which will cause most grammars to fail\n", __func__);
             }
         }
 
@@ -398,7 +397,7 @@ int main(int argc, char ** argv) {
 
     // do one empty run to warm up the model
     {
-        const std::vector<llama_token> tmp = { llama_token_bos(), };
+        const std::vector<llama_token> tmp = { llama_token_bos(ctx), };
         llama_eval(ctx, tmp.data(), tmp.size(), 0, params.n_threads);
         llama_reset_timings(ctx);
     }
@@ -582,7 +581,7 @@ int main(int argc, char ** argv) {
                 }
 
                 // Apply penalties
-                float nl_logit = logits[llama_token_nl()];
+                float nl_logit = logits[llama_token_nl(ctx)];
                 auto last_n_repeat = std::min(std::min((int)last_n_tokens.size(), repeat_last_n), n_ctx);
                 llama_sample_repetition_penalty(ctx, &candidates_p,
                     last_n_tokens.data() + last_n_tokens.size() - last_n_repeat,
@@ -591,7 +590,7 @@ int main(int argc, char ** argv) {
                     last_n_tokens.data() + last_n_tokens.size() - last_n_repeat,
                     last_n_repeat, alpha_frequency, alpha_presence);
                 if (!penalize_nl) {
-                    logits[llama_token_nl()] = nl_logit;
+                    logits[llama_token_nl(ctx)] = nl_logit;
                 }
 
                 if (grammar != NULL) {
@@ -697,7 +696,7 @@ int main(int argc, char ** argv) {
             }
 
             // deal with end of text token in interactive mode
-            if (last_n_tokens.back() == llama_token_eos()) {
+            if (last_n_tokens.back() == llama_token_eos(ctx)) {
                 if (params.interactive) {
                     if (params.antiprompt.size() != 0) {
                         // tokenize and inject first reverse prompt
@@ -721,7 +720,7 @@ int main(int argc, char ** argv) {
                 }
 
                 if (params.input_prefix_bos) {
-                    embd_inp.push_back(llama_token_bos());
+                    embd_inp.push_back(llama_token_bos(ctx));
                 }
 
                 std::string buffer;
@@ -786,7 +785,7 @@ int main(int argc, char ** argv) {
         }
 
         // end of text token
-        if (!embd.empty() && embd.back() == llama_token_eos() && !(params.instruct || params.interactive)) {
+        if (!embd.empty() && embd.back() == llama_token_eos(ctx) && !(params.instruct || params.interactive)) {
             fprintf(stderr, " [end of text]\n");
             break;
         }
diff --git a/examples/perplexity/perplexity.cpp b/examples/perplexity/perplexity.cpp
index b9b28a20b58ae..9eadbeaa9c3bb 100644
--- a/examples/perplexity/perplexity.cpp
+++ b/examples/perplexity/perplexity.cpp
@@ -63,7 +63,7 @@ void perplexity(llama_context * ctx, const gpt_params & params) {
 
             // add BOS token for the first batch of each chunk
             if (j == 0) {
-                tokens[batch_start] = llama_token_bos();
+                tokens[batch_start] = llama_token_bos(ctx);
             }
 
             if (llama_eval(ctx, tokens.data() + batch_start, batch_size, j * n_batch, params.n_threads)) {
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 9337e2104b9b7..a04f1910c0d8d 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -279,7 +279,7 @@ struct llama_server_context
             grammar_parser::print_grammar(stderr, parsed_grammar);
 
             {
-                auto it = params.logit_bias.find(llama_token_eos());
+                auto it = params.logit_bias.find(llama_token_eos(ctx));
                 if (it != params.logit_bias.end() && it->second == -INFINITY) {
                     LOG_WARNING("EOS token is disabled, which will cause most grammars to fail", {});
                 }
@@ -402,7 +402,7 @@ struct llama_server_context
         if (params.n_predict == 0)
         {
             has_next_token = false;
-            result.tok = llama_token_eos();
+            result.tok = llama_token_eos(ctx);
             return result;
         }
 
@@ -442,7 +442,7 @@ struct llama_server_context
             llama_token_data_array candidates_p = {candidates.data(), candidates.size(), false};
 
             // Apply penalties
-            float nl_logit = logits[llama_token_nl()];
+            float nl_logit = logits[llama_token_nl(ctx)];
             auto last_n_repeat = std::min(std::min((int)last_n_tokens.size(), repeat_last_n), params.n_ctx);
             llama_sample_repetition_penalty(ctx, &candidates_p,
                                             last_n_tokens.data() + last_n_tokens.size() - last_n_repeat,
@@ -452,7 +452,7 @@ struct llama_server_context
                                                           last_n_repeat, alpha_frequency, alpha_presence);
             if (!penalize_nl)
             {
-                logits[llama_token_nl()] = nl_logit;
+                logits[llama_token_nl(ctx)] = nl_logit;
             }
 
             if (grammar != nullptr) {
@@ -515,7 +515,7 @@ struct llama_server_context
         // decrement remaining sampling budget
         --n_remain;
 
-        if (!embd.empty() && embd.back() == llama_token_eos())
+        if (!embd.empty() && embd.back() == llama_token_eos(ctx))
         {
             // stopping_word = llama_token_to_str(ctx, embd.back());
             has_next_token = false;
@@ -949,7 +949,7 @@ static void server_params_parse(int argc, char **argv, server_params &sparams,
 
 static json format_generation_settings(llama_server_context &llama)
 {
-    const auto eos_bias = llama.params.logit_bias.find(llama_token_eos());
+    const auto eos_bias = llama.params.logit_bias.find(llama_token_eos(llama.ctx));
     const bool ignore_eos = eos_bias != llama.params.logit_bias.end() &&
                             eos_bias->second < 0.0f && std::isinf(eos_bias->second);
 
@@ -1084,7 +1084,7 @@ static void parse_options_completion(const json &body, llama_server_context &lla
     llama.params.logit_bias.clear();
     if (body.value("ignore_eos", false))
     {
-        llama.params.logit_bias[llama_token_eos()] = -INFINITY;
+        llama.params.logit_bias[llama_token_eos(llama.ctx)] = -INFINITY;
     }
 
     const auto &logit_bias = body.find("logit_bias");
diff --git a/examples/simple/simple.cpp b/examples/simple/simple.cpp
index 476b31b2e9754..132f7fbf912bb 100644
--- a/examples/simple/simple.cpp
+++ b/examples/simple/simple.cpp
@@ -106,7 +106,7 @@ int main(int argc, char ** argv) {
         new_token_id = llama_sample_token_greedy(ctx , &candidates_p);
 
         // is it an end of stream ?
-        if (new_token_id == llama_token_eos()) {
+        if (new_token_id == llama_token_eos(ctx)) {
             fprintf(stderr, " [end of text]\n");
             break;
         }
diff --git a/examples/train-text-from-scratch/train-text-from-scratch.cpp b/examples/train-text-from-scratch/train-text-from-scratch.cpp
index 79599951cb174..922518da4d380 100644
--- a/examples/train-text-from-scratch/train-text-from-scratch.cpp
+++ b/examples/train-text-from-scratch/train-text-from-scratch.cpp
@@ -1996,7 +1996,7 @@ void print_tokens_batch(struct llama_context* ctx, struct ggml_tensor * tokens)
     }
 }
 
-void get_example_targets(const int * train_samples, size_t n_train_samples, const llama_token * train_data, size_t n_train_data, int example_id, struct ggml_tensor * tokens_input, struct ggml_tensor * target_logits, struct ggml_tensor * target_probs) {
+void get_example_targets(struct llama_context * lctx, const int * train_samples, size_t n_train_samples, const llama_token * train_data, size_t n_train_data, int example_id, struct ggml_tensor * tokens_input, struct ggml_tensor * target_logits, struct ggml_tensor * target_probs) {
     int n_tokens = tokens_input->ne[0];
     int n_vocab  = target_logits->ne[0];
 
@@ -2005,7 +2005,7 @@ void get_example_targets(const int * train_samples, size_t n_train_samples, cons
 
     ggml_set_f32(target_logits, -1.0f/n_vocab);
     ggml_set_f32(target_probs, 0.0f);
-    ggml_set_i32_1d(tokens_input, 0, llama_token_bos());
+    ggml_set_i32_1d(tokens_input, 0, llama_token_bos(lctx));
     for (int i=1; i<n_tokens+1; ++i) {
         int token = clamp(train_data[sample+i-1], 0, n_vocab-1);
         set_f32_2d(target_logits, token, i-1, +1.0f);
@@ -2016,7 +2016,7 @@ void get_example_targets(const int * train_samples, size_t n_train_samples, cons
     }
 }
 
-void get_example_targets_batch(struct llama_context * /*lctx*/, const int * train_samples, size_t n_train_samples, const llama_token * train_data, size_t n_train_data, int example_id, struct ggml_tensor * tokens_input, struct ggml_tensor * target_logits, struct ggml_tensor * target_probs) {
+void get_example_targets_batch(struct llama_context * lctx, const int * train_samples, size_t n_train_samples, const llama_token * train_data, size_t n_train_data, int example_id, struct ggml_tensor * tokens_input, struct ggml_tensor * target_logits, struct ggml_tensor * target_probs) {
     GGML_ASSERT(tokens_input->n_dims  == 2);
     GGML_ASSERT(target_logits->n_dims == 3);
     GGML_ASSERT(target_probs->n_dims  == 3);
@@ -2036,7 +2036,7 @@ void get_example_targets_batch(struct llama_context * /*lctx*/, const int * trai
         size_t sample = train_samples[(example_id*n_batch + k) % n_train_samples];
         GGML_ASSERT(sample+n_tokens-1 < n_train_data);
 
-        set_i32_2d(tokens_input, 0, k, llama_token_bos());
+        set_i32_2d(tokens_input, 0, k, llama_token_bos(lctx));
         for (int i=1; i<n_tokens+1; ++i) {
             int token = clamp(train_data[sample+i-1], 0, n_vocab-1);
             // print_token(lctx, token);
@@ -2294,7 +2294,7 @@ llama_token sample(struct my_llama_sampler * sampler, float * logits, const llam
     const auto params = sampler->params;
 
     // Apply penalties
-    const float nl_logit = logits[llama_token_nl()];
+    const float nl_logit = logits[llama_token_nl(ctx)];
 
     const int n_last = std::min(std::min(n_last_tokens, params.repeat_last_n), sampler->n_ctx);
 
@@ -2313,7 +2313,7 @@ llama_token sample(struct my_llama_sampler * sampler, float * logits, const llam
         params.alpha_presence);
 
     if (!params.penalize_nl) {
-        logits[llama_token_nl()] = nl_logit;
+        logits[llama_token_nl(ctx)] = nl_logit;
     }
 
     llama_token token = 0;
@@ -3181,7 +3181,7 @@ int main(int argc, char ** argv) {
     std::vector<int> train_samples;
     train_samples.push_back(0);
     for (int i = 1; i < (int) train_tokens.size() - n_tokens; ++i) {
-        if (!params.samples_start_after_nl || (train_tokens[i-1] == llama_token_nl())) {
+        if (!params.samples_start_after_nl || (train_tokens[i-1] == llama_token_nl(lctx))) {
             train_samples.push_back(i);
         }
     }
@@ -3341,7 +3341,7 @@ int main(int argc, char ** argv) {
         struct ggml_tensor * target_logits = ggml_new_tensor_2d(model.ctx, GGML_TYPE_F32, n_vocab,  n_tokens);
         struct ggml_tensor * target_probs  = ggml_new_tensor_2d(model.ctx, GGML_TYPE_F32, n_vocab,  n_tokens);
 
-        get_example_targets(train_samples.data(), train_samples.size(), train_tokens.data(), train_tokens.size(), rand()%train_samples.size(), tokens_input, target_logits, target_probs);
+        get_example_targets(lctx, train_samples.data(), train_samples.size(), train_tokens.data(), train_tokens.size(), rand()%train_samples.size(), tokens_input, target_logits, target_probs);
         for (int i=sample_ctx; i<n_tokens; ++i) {
             ggml_set_i32_1d(tokens_input, i, n_vocab/2);
         }
diff --git a/llama.cpp b/llama.cpp
index 4f2af46647891..c028c90238354 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -780,13 +780,14 @@ struct llama_vocab {
     std::unordered_map<token, id> token_to_id;
     std::vector<token_score> id_to_token;
 
-    id special_bos_id = -1;
-    id special_eos_id = -1;
+    // default LLaMA special tokens
+    id special_bos_id = 1;
+    id special_eos_id = 2;
     id special_unk_id = -1;
     id special_sep_id = -1;
     id special_pad_id = -1;
 
-    id linefeed_id = -1;
+    id linefeed_id = 13;
 };
 
 struct llama_model {
@@ -2351,21 +2352,11 @@ static bool llama_is_control_token(const llama_vocab & vocab, llama_token token)
 }
 
 static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 1;
-    }
-
-    // TODO: improve?
-    return false;
+    return token == vocab.special_bos_id;
 }
 
 static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 2;
-    }
-
-    // TODO: improve?
-    return false;
+    return token == vocab.special_eos_id;
 }
 
 static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
@@ -2608,7 +2599,7 @@ static std::vector<llama_vocab::id> llama_tokenize_internal(const llama_vocab &
     }
 
     if (bos) {
-        output.push_back(llama_token_bos());
+        output.push_back(vocab.special_bos_id);
     }
 
     std::string text;
@@ -3293,7 +3284,7 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
         }
     }
 
-    const llama_token eos = llama_token_eos();
+    const llama_token eos = llama_token_eos(ctx);
 
     std::vector<std::pair<std::vector<uint32_t>, llama_partial_utf8>> candidates_decoded;
     std::vector<llama_grammar_candidate>                              candidates_grammar;
@@ -3503,7 +3494,7 @@ llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_arra
 void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
     const int64_t t_start_sample_us = ggml_time_us();
 
-    if (token == llama_token_eos()) {
+    if (token == llama_token_eos(ctx)) {
         for (const auto & stack : grammar->stacks) {
             if (stack.empty()) {
                 return;
@@ -4340,7 +4331,7 @@ struct llama_context * llama_new_context_with_model(
             // build worst-case graph
             int n_tokens = std::min((int)hparams.n_ctx, params.n_batch);
             int n_past = hparams.n_ctx - n_tokens;
-            llama_token token = llama_token_bos(); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
+            llama_token token = llama_token_bos(ctx); // not actually used by llama_build_graph, but required to choose between token and embedding inputs graph
             ggml_cgraph * gf = llama_build_graph(*ctx, &token, NULL, n_tokens, n_past);
 #ifdef GGML_USE_METAL
             if (params.n_gpu_layers > 0) {
@@ -4950,7 +4941,7 @@ int llama_eval_export(struct llama_context * ctx, const char * fname) {
     const int n_batch = 1;
     const int n_ctx   = 512 - n_batch;
 
-    const std::vector<llama_token> tmp(n_batch, llama_token_bos());
+    const std::vector<llama_token> tmp(n_batch, llama_token_bos(ctx));
 
     if (!llama_eval_internal(*ctx, tmp.data(), nullptr, tmp.size(), n_ctx, 1, fname)) {
         LLAMA_LOG_ERROR("%s: failed to eval\n", __func__);
@@ -4989,16 +4980,16 @@ int llama_model_get_vocab(
     return n;
 }
 
-llama_token llama_token_bos(void) {
-    return 1;
+llama_token llama_token_bos(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_bos_id;
 }
 
-llama_token llama_token_eos(void) {
-    return 2;
+llama_token llama_token_eos(const struct llama_context * ctx) {
+    return ctx->model.vocab.special_eos_id;
 }
 
-llama_token llama_token_nl(void) {
-    return 13;
+llama_token llama_token_nl(const struct llama_context * ctx) {
+    return ctx->model.vocab.linefeed_id;
 }
 
 int llama_tokenize(
diff --git a/llama.h b/llama.h
index e2b28afbb1c45..54081840b846c 100644
--- a/llama.h
+++ b/llama.h
@@ -340,9 +340,9 @@ extern "C" {
                                    int   capacity);
 
     // Special tokens
-    LLAMA_API llama_token llama_token_bos(/*struct llama_model * model*/ void);  // beginning-of-sentence
-    LLAMA_API llama_token llama_token_eos(/*struct llama_model * model*/ void);  // end-of-sentence
-    LLAMA_API llama_token llama_token_nl (/*struct llama_model * model*/ void);  // next-line
+    LLAMA_API llama_token llama_token_bos(const struct llama_context * ctx);  // beginning-of-sentence
+    LLAMA_API llama_token llama_token_eos(const struct llama_context * ctx);  // end-of-sentence
+    LLAMA_API llama_token llama_token_nl (const struct llama_context * ctx);  // next-line
 
     //
     // Tokenization

From a4ad2bf35c8504ca0347b40ad5b209063c82a0ee Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 17:34:27 +0300
Subject: [PATCH 203/242] llama : fix MPI build

ggml-ci
---
 llama.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llama.cpp b/llama.cpp
index c028c90238354..9c1b2a93e7ecf 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -4419,7 +4419,7 @@ struct llama_context * llama_new_context_with_model(
 
     if (ggml_mpi_rank(ctx->ctx_mpi) > 0) {
         // Enter a blocking eval loop with dummy input, letting rank=0 drive the process
-        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos());
+        const std::vector<llama_token> tmp(ctx->model.hparams.n_ctx, llama_token_bos(ctx));
         while (!llama_eval(ctx, tmp.data(), tmp.size(), 0, 0)) {};
         llama_backend_free();
         exit(1);

From 25b8a8922d2a042f013737381be5ddda1bb5756a Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 18 Aug 2023 18:46:38 +0300
Subject: [PATCH 204/242] llama : introduce enum llama_vocab_type + remove
 hardcoded string constants

---
 llama.cpp | 68 ++++++++++++++++++++++++++++++-------------------
 llama.h   | 75 +++++++++++++++++++++++++++++--------------------------
 2 files changed, 82 insertions(+), 61 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 9c1b2a93e7ecf..86c943fc6e3c5 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -777,8 +777,10 @@ struct llama_vocab {
         float score;
     };
 
+    llama_vocab_type type = LLAMA_VOCAB_TYPE_SPM;
+
     std::unordered_map<token, id> token_to_id;
-    std::vector<token_score> id_to_token;
+    std::vector<token_score>      id_to_token;
 
     // default LLaMA special tokens
     id special_bos_id = 1;
@@ -1406,6 +1408,19 @@ static void llama_model_load_internal(
             } \
         }
 
+        std::string tokenizer_name;
+        GGUF_GET(tokenizer_name, gguf_get_val_str, GGUF_TYPE_STRING, true, "tokenizer.ggml.model");
+
+        if (tokenizer_name == "llama") {
+            vocab.type = LLAMA_VOCAB_TYPE_SPM;
+        } else if (tokenizer_name == "gpt2") {
+            vocab.type = LLAMA_VOCAB_TYPE_BPE;
+        } else {
+            LLAMA_LOG_WARN("%s: unknown tokenizer: '%s'", __func__, tokenizer_name.c_str());
+            LLAMA_LOG_WARN("%s: using default tokenizer: 'llama'", __func__);
+            vocab.type = LLAMA_VOCAB_TYPE_SPM;
+        }
+
         // get hparams kv
         GGUF_GET(hparams.n_vocab,        gguf_get_arr_n,   GGUF_TYPE_ARRAY,   true, "tokenizer.ggml.tokens");
         GGUF_GET(hparams.n_ctx_train,    gguf_get_val_u32, GGUF_TYPE_UINT32,  true, "llama.context_length");
@@ -1504,6 +1519,7 @@ static void llama_model_load_internal(
         // hparams
         LLAMA_LOG_INFO("%s: format       = %s\n",     __func__, llama_file_version_name(ml->fver));
         LLAMA_LOG_INFO("%s: arch         = %s\n",     __func__, general_arch.c_str());
+        LLAMA_LOG_INFO("%s: vocab type   = %s\n",     __func__, vocab.type == LLAMA_VOCAB_TYPE_SPM ? "SPM" : "BPE"); // TODO: fix
         LLAMA_LOG_INFO("%s: n_vocab      = %u\n",     __func__, hparams.n_vocab);
         LLAMA_LOG_INFO("%s: n_ctx_train  = %u\n",     __func__, hparams.n_ctx_train);
         LLAMA_LOG_INFO("%s: n_ctx        = %u\n",     __func__, hparams.n_ctx);
@@ -2317,46 +2333,46 @@ static bool llama_eval_internal(
 // tokenizer
 //
 
-static std::string llama_vocab_type(const llama_vocab & vocab) {
-    return vocab.token_to_id.size() == 32000 ? "spm": "bpe";
+static enum llama_vocab_type llama_vocab_get_type(const llama_vocab & vocab) {
+    return vocab.type;
 }
 
 static bool llama_is_normal_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
         return token >= 259;
     }
 
-    if (llama_vocab_type(vocab) == "bpe") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
         return token >= 95;
     }
 
     return false;
 }
 
-static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 0;
-    }
+static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
+    return token == vocab.special_bos_id;
+}
 
-    // TODO: improve?
-    return false;
+static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
+    return token == vocab.special_eos_id;
 }
 
 static bool llama_is_control_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
-        return token == 1 || token == 2;
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
+        return token == llama_is_bos_token(vocab, token) || token == llama_is_eos_token(vocab, token);
     }
 
     // TODO: improve?
     return false;
 }
 
-static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
-    return token == vocab.special_bos_id;
-}
+static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
+        return token == 0;
+    }
 
-static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
-    return token == vocab.special_eos_id;
+    // TODO: improve?
+    return false;
 }
 
 static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
@@ -2374,11 +2390,11 @@ static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token)
 }
 
 static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_type(vocab) == "spm") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
         return 3 <= token && token < 259;
     }
 
-    if (llama_vocab_type(vocab) == "bpe") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
         return 1 <= token && token < 95;
     }
 
@@ -2386,11 +2402,11 @@ static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
 }
 
 static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
-    if (llama_vocab_type(vocab) == "spm") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
         return byte - 3;
     }
 
-    if (llama_vocab_type(vocab) == "bpe") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
         return byte + 32;
     }
 
@@ -2398,11 +2414,11 @@ static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
 }
 
 static uint8_t llama_char_to_byte(const llama_vocab & vocab, uint8_t ch) {
-    if (llama_vocab_type(vocab) == "spm") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
         return ch + 3;
     }
 
-    if (llama_vocab_type(vocab) == "bpe") {
+    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
         return ch - 32;
     }
 
@@ -5027,7 +5043,7 @@ int llama_tokenize_with_model(
                  llama_token * tokens,
                          int   n_max_tokens,
                         bool   add_bos) {
-    auto escape = llama_vocab_type(model->vocab) == "spm";
+    auto escape = llama_vocab_get_type(model->vocab) == LLAMA_VOCAB_TYPE_SPM;
     auto res = llama_tokenize_internal(model->vocab, text, add_bos, escape);
 
     if (n_max_tokens < (int) res.size()) {
@@ -5063,7 +5079,7 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
     if (0 <= token && token < llama_model_n_vocab(model)) {
         if (llama_is_normal_token(model->vocab, token)) {
             std::string result = model->vocab.id_to_token[token].tok;
-            if (llama_vocab_type(model->vocab) == "spm") {
+            if (llama_vocab_get_type(model->vocab) == LLAMA_VOCAB_TYPE_SPM) {
                 result = llama_unescape_whitespace(result);
             }
             if (length < (int) result.length()) {
diff --git a/llama.h b/llama.h
index 54081840b846c..0ea65c1b56710 100644
--- a/llama.h
+++ b/llama.h
@@ -61,6 +61,40 @@ extern "C" {
 
     typedef int llama_token;
 
+    enum llama_log_level {
+        LLAMA_LOG_LEVEL_ERROR = 2,
+        LLAMA_LOG_LEVEL_WARN  = 3,
+        LLAMA_LOG_LEVEL_INFO  = 4
+    };
+
+    enum llama_vocab_type {
+        LLAMA_VOCAB_TYPE_SPM = 0, // SentencePiece
+        LLAMA_VOCAB_TYPE_BPE = 1, // Byte Pair Encoding
+    };
+
+    // model file types
+    enum llama_ftype {
+        LLAMA_FTYPE_ALL_F32              = 0,
+        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
+        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
+        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
+        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
+        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
+    };
+
     typedef struct llama_token_data {
         llama_token id; // token id
         float logit;    // log-odds of the token
@@ -75,19 +109,6 @@ extern "C" {
 
     typedef void (*llama_progress_callback)(float progress, void *ctx);
 
-    enum llama_log_level {
-        LLAMA_LOG_LEVEL_ERROR = 2,
-        LLAMA_LOG_LEVEL_WARN  = 3,
-        LLAMA_LOG_LEVEL_INFO  = 4
-    };
-
-    // Signature for logging events
-    // Note that text includes the new line character at the end for most events.
-    // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
-    // if it exists.
-    // It might not exist for progress report where '.' is output repeatedly.
-    typedef void (*llama_log_callback)(enum llama_log_level level, const char * text, void * user_data);
-
     struct llama_context_params {
         uint32_t seed;         // RNG seed, -1 for random
         int32_t  n_ctx;        // text context
@@ -117,28 +138,12 @@ extern "C" {
         bool embedding;  // embedding mode only
     };
 
-    // model file types
-    enum llama_ftype {
-        LLAMA_FTYPE_ALL_F32              = 0,
-        LLAMA_FTYPE_MOSTLY_F16           = 1, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_0          = 2, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1          = 3, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4, // tok_embeddings.weight and output.weight are F16
-        // LLAMA_FTYPE_MOSTLY_Q4_2       = 5, // support has been removed
-        // LLAMA_FTYPE_MOSTLY_Q4_3       = 6, // support has been removed
-        LLAMA_FTYPE_MOSTLY_Q8_0          = 7, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_0          = 8, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_1          = 9, // except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q2_K          = 10,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17,// except 1d tensors
-        LLAMA_FTYPE_MOSTLY_Q6_K          = 18,// except 1d tensors
-    };
+    // Signature for logging events
+    // Note that text includes the new line character at the end for most events.
+    // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
+    // if it exists.
+    // It might not exist for progress report where '.' is output repeatedly.
+    typedef void (*llama_log_callback)(enum llama_log_level level, const char * text, void * user_data);
 
     // model quantization parameters
     typedef struct llama_model_quantize_params {

From fb7c883cd3dad5715074685b66863ec699af5f40 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Fri, 18 Aug 2023 20:14:01 +0200
Subject: [PATCH 205/242] convert-falcon-hf-to-gguf.py : falcon HF --> gguf
 conversion, not tested

---
 convert-falcon-hf-to-gguf.py | 285 +++++++++++++++++++++++++++++++++++
 1 file changed, 285 insertions(+)
 create mode 100644 convert-falcon-hf-to-gguf.py

diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py
new file mode 100644
index 0000000000000..7d50d8c5caaff
--- /dev/null
+++ b/convert-falcon-hf-to-gguf.py
@@ -0,0 +1,285 @@
+# HF falcon--> gguf conversion
+
+import gguf
+import os
+import sys
+import struct
+import json
+import numpy as np
+import torch
+
+from typing import Any, List
+from pathlib import Path
+from transformers import AutoTokenizer
+
+def bytes_to_unicode():
+    # ref: https://github.com/openai/gpt-2/blob/master/src/encoder.py
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a significant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1))
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8+n)
+            n += 1
+    cs = [chr(n) for n in cs]
+    return dict(zip(bs, cs))
+
+
+def count_model_parts(dir_model: str) -> int:
+    num_parts = 0
+    for filename in os.listdir(dir_model):
+        if filename.startswith("pytorch_model-"):
+            num_parts += 1
+
+    if num_parts > 0:
+        print("gguf: found " + str(num_parts) + " model parts")
+    return num_parts
+
+
+if len(sys.argv) < 3:
+    print("Usage: convert-h5-to-ggml.py dir-model ftype\n")
+    print("  ftype == 0 -> float32")
+    print("  ftype == 1 -> float16")
+    sys.exit(1)
+
+
+# output in the same directory as the model
+dir_model = sys.argv[1]
+last_dir = os.path.basename(os.path.normpath(dir_model))
+
+# possible tensor data types
+#   ftype == 0 -> float32
+#   ftype == 1 -> float16
+
+# map from ftype to string
+ftype_str = ["f32", "f16"]
+
+ftype = 1
+if len(sys.argv) > 2:
+    ftype = int(sys.argv[2])
+    if ftype < 0 or ftype > 1:
+        print("Invalid ftype: " + str(ftype))
+
+        sys.exit(1)
+
+fname_out = sys.argv[1] + "/ggml-model-" + ftype_str[ftype] + ".gguf"
+
+print("gguf: loading model "+last_dir)
+
+with open(dir_model + "/config.json", "r", encoding="utf-8") as f:
+    hparams = json.load(f)
+
+if hparams["architectures"][0] != "RWForCausalLM":
+    print("Model architecture not supported: " + hparams["architectures"][0])
+
+    sys.exit()
+
+# get number of model parts
+num_parts = count_model_parts(dir_model)
+
+ARCH=gguf.MODEL_ARCH.FALCON
+gguf_writer = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
+
+print("gguf: get model metadata")
+
+block_count = hparams["n_layer"]
+
+gguf_writer.add_architecture()
+gguf_writer.add_name(last_dir)
+gguf_writer.add_context_length(2048)
+gguf_writer.add_embedding_length(hparams["hidden_size"])
+gguf_writer.add_feed_forward_length(4 * hparams["hidden_size"])
+gguf_writer.add_block_count(block_count)
+gguf_writer.add_head_count(hparams["n_head"])
+if "n_head_kv" in hparams: gguf_writer.add_head_count_kv(hparams["n_head_kv"])
+gguf_writer.add_layer_norm_eps(hparams["layer_norm_epsilon"])
+
+# TOKENIZATION
+
+print("gguf: get tokenizer metadata")
+
+tokens: List[str] = []
+merges: List[str] = []
+
+
+if Path(dir_model + "/tokenizer.json").is_file():
+    # gpt2 tokenizer
+    gguf_writer.add_tokenizer_model("gpt2")
+
+    print("gguf: get gpt2 tokenizer merges")
+
+    with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
+        tokenizer_json = json.load(f)
+    merges = tokenizer_json["model"]["merges"]
+
+    gguf_writer.add_token_merges(merges)
+
+    print("gguf: get gpt2 tokenizer vocab")
+
+    vocab_size = len(tokenizer_json["model"]["vocab"])
+
+    # ref: https://github.com/cmp-nct/ggllm.cpp/blob/master/falcon_convert.py
+    tokenizer = AutoTokenizer.from_pretrained(dir_model)
+
+    reverse_vocab = {id: encoded_tok for encoded_tok, id in tokenizer.vocab.items()}
+    byte_encoder = bytes_to_unicode()
+    byte_decoder = {v: k for k, v in byte_encoder.items()}
+
+    for i in range(vocab_size):
+        if i in reverse_vocab:
+            try:
+                text = bytearray([byte_decoder[c] for c in reverse_vocab[i]])
+            except KeyError:
+                text = bytearray()
+                for c in reverse_vocab[i]:
+                    if ord(c) < 256:  # single byte character
+                        text.append(byte_decoder[ord(c)])
+                    else:  # multibyte special token character
+                        text.extend(c.encode('utf-8'))
+        else:
+            print(f"Key {i} not in tokenizer vocabulary. Padding with an arbitrary token.")
+            pad_token = f"[PAD{i}]".encode("utf8")
+            text = bytearray(pad_token)
+
+        tokens.append(text)
+
+    gguf_writer.add_token_list(tokens)
+
+    if "added_tokens" in tokenizer_json and Path(dir_model + "/tokenizer_config.json").is_file():
+        print("gguf: get special token ids")
+
+        with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
+            tokenizer_config = json.load(f)
+
+        # find special token ids
+
+        if "bos_token" in tokenizer_config:
+            for key in tokenizer_json["added_tokens"]:
+                if key["content"] == tokenizer_config["bos_token"]:
+                    gguf_writer.add_bos_token_id(key["id"])
+
+        if "eos_token" in tokenizer_config:
+            for key in tokenizer_json["added_tokens"]:
+                if key["content"] == tokenizer_config["eos_token"]:
+                    gguf_writer.add_eos_token_id(key["id"])
+
+        if "unk_token" in tokenizer_config:
+            for key in tokenizer_json["added_tokens"]:
+                if key["content"] == tokenizer_config["unk_token"]:
+                    gguf_writer.add_unk_token_id(key["id"])
+
+        if "sep_token" in tokenizer_config:
+            for key in tokenizer_json["added_tokens"]:
+                if key["content"] == tokenizer_config["sep_token"]:
+                    gguf_writer.add_sep_token_id(key["id"])
+
+        if "pad_token" in tokenizer_config:
+            for key in tokenizer_json["added_tokens"]:
+                if key["content"] == tokenizer_config["pad_token"]:
+                    gguf_writer.add_pad_token_id(key["id"])
+
+
+# TENSORS
+
+tensor_map = gguf.get_tensor_name_map(ARCH,block_count)
+
+# params for qkv transform
+if "n_head_kv" in hparams:
+    n_head_kv = hparams["n_head_kv"]
+else:
+    n_head_kv = 1
+n_head = hparams["n_head"]
+head_dim = hparams["hidden_size"] // n_head
+
+# tensor info
+print("gguf: get tensor metadata")
+
+if num_parts == 0:
+    part_names = ("pytorch_model.bin",)
+else:
+    part_names = (
+        f"pytorch_model-{n:05}-of-{num_parts:05}.bin" for n in range(1, num_parts + 1)
+    )
+
+for part_name in part_names:
+    print("gguf: loading model part '" + part_name + "'")
+    model_part = torch.load(f"{dir_model}/{part_name}", map_location="cpu")
+
+    for name in model_part.keys():
+        data = model_part[name]
+
+        old_dtype = data.dtype
+
+        # convert any unsupported data types to float32
+        if data.dtype != torch.float16 and data.dtype != torch.float32:
+            data = data.to(torch.float32)
+
+        # QKV tensor transform
+        # The original query_key_value tensor contains n_head_kv "kv groups",
+        # each consisting of n_head/n_head_kv query weights followed by one key
+        # and one value weight (shared by all query heads in the kv group).
+        # This layout makes it a big pain to work with in GGML.
+        # So we rearrange them here,, so that we have n_head query weights
+        # followed by n_head_kv key weights followed by n_head_kv value weights,
+        # in contiguous fashion.
+        # ref: https://github.com/jploski/ggml/blob/d5295b477fb36c69468c3fecb0393a8d7980b7c8/examples/falcon/convert-hf-to-ggml.py#L107-L123
+
+        if "query_key_value" in name:
+            qkv = data.view(n_head_kv, n_head // n_head_kv + 2, head_dim, head_dim * n_head)
+            q = qkv[:, :-2 ].reshape(n_head * head_dim, head_dim * n_head)
+            k = qkv[:, [-2]].reshape(n_head_kv * head_dim, head_dim * n_head)
+            v = qkv[:, [-1]].reshape(n_head_kv * head_dim, head_dim * n_head)
+            data = torch.cat((q,k,v)).reshape_as(data)
+
+        data = data.squeeze().numpy()
+
+        # map tensor names
+        if name.endswith(".weight") and name[:-7] in tensor_map:
+            name = tensor_map[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tensor_map:
+            name = tensor_map[name[:-5]] + ".bias"
+        else:
+            print("Can not map tensor '" + name + "'")
+            sys.exit()
+
+        n_dims = len(data.shape)
+        data_dtype = data.dtype 
+
+        # if f32 desired, convert any float16 to float32
+        if ftype == 0 and data_dtype == np.float16:
+            data = data.astype(np.float32)
+
+        # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32
+        if ftype == 1 and data_dtype == np.float16 and n_dims == 1:
+            data = data.astype(np.float32)
+
+        # if f16 desired, convert any float32 2-dim weight tensors to float16
+        if ftype == 1 and data_dtype == np.float32 and name.endswith(".weight") and n_dims == 2:
+            data = data.astype(np.float16)
+
+        print(name + ", n_dims = " + str(n_dims) + ", " + str(old_dtype) + " --> " + str(data.dtype))
+
+        gguf_writer.add_tensor(name, data)
+
+
+print("gguf: write header")
+gguf_writer.write_header_to_file()
+print("gguf: write metadata")
+gguf_writer.write_kv_data_to_file()
+print("gguf: write tensors")
+gguf_writer.write_tensors_to_file()
+
+gguf_writer.close()
+
+print("gguf: model successfully exported to '" + fname_out + "'")
+print("")

From d5e976c12ba5ec2e5510fdab9c432d9e8062159d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 00:02:18 +0200
Subject: [PATCH 206/242] falcon-main.cpp : falcon inference example

---
 examples/gptneox-wip/falcon-main.cpp | 1094 ++++++++++++++++++++++++++
 1 file changed, 1094 insertions(+)
 create mode 100644 examples/gptneox-wip/falcon-main.cpp

diff --git a/examples/gptneox-wip/falcon-main.cpp b/examples/gptneox-wip/falcon-main.cpp
new file mode 100644
index 0000000000000..c97aa602acbd9
--- /dev/null
+++ b/examples/gptneox-wip/falcon-main.cpp
@@ -0,0 +1,1094 @@
+#include "ggml.h"
+#include "cmpnct_gpt2bpe.hpp"
+
+#include <cassert>
+#include <cmath>
+#include <cstdio>
+#include <cstring>
+#include <cinttypes>
+#include <fstream>
+#include <map>
+#include <string>
+#include <vector>
+#include <thread>
+#include <random>
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244 4267) // possible loss of data
+#endif
+
+// default hparams
+struct falcon_hparams {
+    size_t n_merges = 0;
+    size_t n_vocab  = 0;
+    uint32_t n_ctx    = 0;
+    uint32_t n_embd   = 0;
+    uint32_t n_head   = 0;
+    uint32_t n_head_kv = 1; // Needs to be 1 for 7B model
+    uint32_t n_ff = 0;
+    uint32_t n_block  = 0;
+    float norm_eps = 1e-5;
+};
+struct falcon_block {
+    // normalization
+    struct ggml_tensor* input_layernorm;
+    struct ggml_tensor* input_layernorm_b;
+    struct ggml_tensor* attention_norm;    // Falcon-40B only
+    struct ggml_tensor* attention_norm_b;  // Falcon-40B only
+
+    // attention
+    struct ggml_tensor* query_key_value;
+    struct ggml_tensor* wo;
+
+    // ff
+    struct ggml_tensor* ffn_up;
+    struct ggml_tensor* ffn_down;
+};
+
+struct falcon_model {
+    falcon_hparams hparams;
+
+    struct ggml_tensor* tok_embeddings;
+    struct ggml_tensor* output_norm;
+    struct ggml_tensor* output_norm_b;
+    struct ggml_tensor* lm_head;
+
+    std::vector<falcon_block> blocks;
+
+    // key + value memory
+    struct ggml_tensor* memory_k;
+    struct ggml_tensor* memory_v;
+
+    struct gguf_context * ggufctx;
+    struct ggml_context * ctx;
+    struct ggml_context * kvctx;
+
+    std::map<std::string, struct ggml_tensor*> tensors;
+};
+
+struct gpt_params {
+    int32_t seed      = -1;  // RNG seed
+    int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
+    uint32_t n_predict = 200; // new tokens to predict
+    uint32_t n_batch   = 512;   // batch size for prompt processing
+
+    // sampling parameters
+    int32_t top_k          = 40;
+    float top_p            = 1.0f;
+    float temp             = 0.8f;
+    int32_t repeat_last_n  = 64;
+    float repeat_penalty   = 1.02f;
+
+    std::string model      = ""; // model path
+    std::string prompt     = "";
+
+    std::string token_test = "";
+    bool    interactive      = false;
+    int32_t interactive_port = -1;
+    int32_t n_gpu_layers     = 0;
+};
+
+void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
+    fprintf(stderr, "usage: %s [options]\n", argv[0]);
+    fprintf(stderr, "\n");
+    fprintf(stderr, "options:\n");
+    fprintf(stderr, "  -h, --help            show this help message and exit\n");
+    fprintf(stderr, "  -s SEED, --seed SEED  RNG seed (default: -1)\n");
+    fprintf(stderr, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
+    fprintf(stderr, "  -ngl N, --gpu-layers N  number of layers to offload to GPU on supported models (default: %d)\n", params.n_gpu_layers);
+    fprintf(stderr, "  -p PROMPT, --prompt PROMPT\n");
+    fprintf(stderr, "                        prompt to start generation with (default: random)\n");
+    fprintf(stderr, "  -f FNAME, --file FNAME\n");
+    fprintf(stderr, "                        load prompt from a file\n");
+    fprintf(stderr, "  -tt TOKEN_TEST, --token_test TOKEN_TEST\n");
+    fprintf(stderr, "                        test tokenization\n");
+    fprintf(stderr, "  -n N, --n_predict N   number of tokens to predict (default: %d)\n", params.n_predict);
+    fprintf(stderr, "  --top_k N             top-k sampling, 0 = n_vocab (default: %d)\n", params.top_k);
+    fprintf(stderr, "  --top_p N             top-p sampling (default: %.1f)\n", params.top_p);
+    fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
+    fprintf(stderr, "  --repeat-last-n N     last n tokens to consider for penalize (default: %d, 0 = disabled)\n", params.repeat_last_n);
+    fprintf(stderr, "  --repeat-penalty N    penalize repeat sequence of tokens (default: %.2f, 1.0 = disabled)\n", (double)params.repeat_penalty);
+    fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
+    fprintf(stderr, "  -m FNAME, --model FNAME\n");
+    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
+    fprintf(stderr, "\n");
+}
+
+// Function to check if the next argument exists
+std::string get_next_arg(int& i, int argc, char** argv, const std::string& flag, gpt_params& params) {
+    if (i + 1 < argc && argv[i + 1][0] != '-') {
+        return argv[++i];
+    } else {
+        fprintf(stderr, "error: %s requires one argument.\n", flag.c_str());
+        gpt_print_usage(argc, argv, params);
+        exit(0);
+    }
+}
+
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
+    for (int i = 1; i < argc; i++) {
+        std::string arg = argv[i];
+
+        if (arg == "-s" || arg == "--seed") {
+            params.seed = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-t" || arg == "--threads") {
+            params.n_threads = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-ngl" || arg == "--gpu-layers" || arg == "--n-gpu-layers") {
+            params.n_gpu_layers = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-p" || arg == "--prompt") {
+            params.prompt = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-n" || arg == "--n_predict") {
+            params.n_predict = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_k") {
+            params.top_k = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--top_p") {
+            params.top_p = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--temp") {
+            params.temp = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-last-n") {
+            params.repeat_last_n = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "--repeat-penalty") {
+            params.repeat_penalty = std::stof(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-b" || arg == "--batch_size") {
+            params.n_batch= std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-m" || arg == "--model") {
+            params.model = get_next_arg(i, argc, argv, arg, params);
+        } else if (arg == "-i" || arg == "--interactive") {
+            params.interactive = true;
+        } else if (arg == "-ip" || arg == "--interactive-port") {
+            params.interactive = true;
+            params.interactive_port = std::stoi(get_next_arg(i, argc, argv, arg, params));
+        } else if (arg == "-h" || arg == "--help") {
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        } else if (arg == "-f" || arg == "--file") {
+            get_next_arg(i, argc, argv, arg, params);
+            std::ifstream file(argv[i]);
+            if (!file) {
+                fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
+                break;
+            }
+            std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
+            if (params.prompt.back() == '\n') {
+                params.prompt.pop_back();
+            }
+        } else if (arg == "-tt" || arg == "--token_test") {
+            params.token_test = get_next_arg(i, argc, argv, arg, params);
+        }
+        else {
+            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
+            gpt_print_usage(argc, argv, params);
+            exit(0);
+        }
+    }
+
+    return true;
+}
+
+gpt2bpe_vocab::id sample_top_k_top_p_repeat(
+        const gpt2bpe_vocab & vocab,
+        const float * logits,
+        const int32_t * last_n_tokens_data,
+        size_t last_n_tokens_data_size,
+        int    top_k,
+        double top_p,
+        double temp,
+        int repeat_last_n,
+        float repeat_penalty,
+        std::mt19937 & rng) {
+
+    int n_logits = vocab.id_to_token.size();
+
+    const auto * plogits = logits;
+
+    const auto last_n_tokens = std::vector<int32_t>(last_n_tokens_data, last_n_tokens_data + last_n_tokens_data_size);
+
+    if (temp <= 0) {
+        // select the token with the highest logit directly
+        float max_logit = plogits[0];
+        gpt2bpe_vocab::id max_id = 0;
+
+        for (int i = 1; i < n_logits; ++i) {
+            if (plogits[i] > max_logit) {
+                max_logit = plogits[i];
+                max_id = i;
+            }
+        }
+        return max_id;
+    }
+
+
+    std::vector<std::pair<double, gpt2bpe_vocab::id>> logits_id;
+    logits_id.reserve(n_logits);
+
+    {
+        const float scale = 1.0f/temp;
+        for (int i = 0; i < n_logits; ++i) {
+            // repetition penalty from ctrl paper (https://arxiv.org/abs/1909.05858)
+            // credit https://github.com/facebookresearch/llama/compare/main...shawwn:llama:main
+            if (repeat_last_n > 0 && std::find(last_n_tokens.end()-repeat_last_n, last_n_tokens.end(), i) != last_n_tokens.end()) {
+                // if score < 0 then repetition penalty has to multiplied to reduce the previous token probability
+                if (plogits[i] < 0.0f) {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale*repeat_penalty, i));
+                } else {
+                    logits_id.push_back(std::make_pair(plogits[i]*scale/repeat_penalty, i));
+                }
+            } else {
+                logits_id.push_back(std::make_pair(plogits[i]*scale, i));
+            }
+        }
+    }
+
+    // find the top K tokens
+    std::partial_sort(
+            logits_id.begin(),
+            logits_id.begin() + top_k, logits_id.end(),
+            [](const std::pair<double, gpt2bpe_vocab::id> & a, const std::pair<double, gpt2bpe_vocab::id> & b) {
+        return a.first > b.first;
+    });
+
+    logits_id.resize(top_k);
+
+    double maxl = -INFINITY;
+    for (const auto & kv : logits_id) {
+        maxl = std::max(maxl, kv.first);
+    }
+
+    // compute probs for the top K tokens
+    std::vector<double> probs;
+    probs.reserve(logits_id.size());
+
+    double sum = 0.0;
+    for (const auto & kv : logits_id) {
+        double p = exp(kv.first - maxl);
+        probs.push_back(p);
+        sum += p;
+    }
+
+    // normalize the probs
+    for (auto & p : probs) {
+        p /= sum;
+    }
+
+    if (top_p < 1.0f) {
+        double cumsum = 0.0f;
+        for (int i = 0; i < top_k; i++) {
+            cumsum += probs[i];
+            if (cumsum >= top_p) {
+                top_k = i + 1;
+                probs.resize(top_k);
+                logits_id.resize(top_k);
+                break;
+            }
+        }
+
+        cumsum = 1.0/cumsum;
+        for (int i = 0; i < (int) probs.size(); i++) {
+            probs[i] *= cumsum;
+        }
+    }
+
+//    printf("\n");
+//    for (int i = 0; i < (int) probs.size(); i++) {
+//    for (int i = 0; i < 10; i++) {
+//        printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
+//    }
+
+    std::discrete_distribution<> dist(probs.begin(), probs.end());
+    int idx = dist(rng);
+
+    return logits_id[idx].second;
+
+}
+
+struct ggml_tensor * get_tensor_ex( struct ggml_context * ctx, std::string name){
+
+    struct ggml_tensor * cur = ggml_get_tensor(ctx, name.c_str());
+    if( cur == NULL ) {
+        fprintf(stdout, "%s: tensor '%s' not found!\n", __func__, name.c_str());
+    } else {
+//        fprintf(stdout, "%s: n_dims = %d, name = '%s'\n", __func__, cur->n_dims, cur->name);
+    }
+
+    return cur;
+}
+
+// load the model's weights from a file
+bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_vocab & vocab) {
+    printf("%s: loading model from '%s'..\n", __func__, fname.c_str());
+
+    model.ctx = NULL;
+
+    struct gguf_init_params ggufparams = {
+        /*.no_alloc = */ false,
+        /*.ctx      = */ &model.ctx,
+    };
+
+    auto & ggufctx = model.ggufctx;
+
+    ggufctx  = gguf_init_from_file(fname.c_str(), ggufparams);
+
+    if (!ggufctx) {
+        fprintf(stderr, "%s: gguf_init_from_file() failed\n", __func__);
+        return false;
+    }
+
+    fprintf(stdout, "%s: gguf version     = %d\n", __func__, gguf_get_version(ggufctx));
+    fprintf(stdout, "%s: gguf alignment   = %zu\n", __func__, gguf_get_alignment(ggufctx));
+    fprintf(stdout, "%s: gguf data offset = %zu\n", __func__, gguf_get_data_offset(ggufctx));
+
+    // print all kv
+    #if 0
+    {
+        const int n_kv = gguf_get_n_kv(ggufctx);
+
+        fprintf(stdout, "%s: n_kv: %d\n", __func__, n_kv);
+
+        for (int i = 0; i < n_kv; ++i) {
+            const char * key = gguf_get_key(ggufctx, i);
+
+            fprintf(stdout, "%s: kv[%d]: key = %s\n", __func__, i, key);
+        }
+    }
+    #endif
+
+    // print some standard metadata
+    {
+        int keyidx;
+
+        keyidx = gguf_find_key(ggufctx, "general.name");
+        if (keyidx != -1) { fprintf(stdout, "%s: model name           = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.description");
+        if (keyidx != -1) { fprintf(stdout, "%s: model description    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.author");
+        if (keyidx != -1) { fprintf(stdout, "%s: model author         = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.license");
+        if (keyidx != -1) { fprintf(stdout, "%s: model license        = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.architecture");
+        if (keyidx != -1) { fprintf(stdout, "%s: model architecture   = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.file_type");
+        if (keyidx != -1) { fprintf(stdout, "%s: model file type      = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "gptneox.tensor_data_layout");
+        if (keyidx != -1) { fprintf(stdout, "%s: model data layout    = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+        keyidx = gguf_find_key(ggufctx, "general.source.hugginface.repository");
+        if (keyidx != -1) { fprintf(stdout, "%s: model source HF repo = %s\n", __func__, gguf_get_val_str(ggufctx, keyidx)); }
+    }
+
+    // check required metadata
+    {
+        int keyidx;
+
+        keyidx = gguf_find_key(ggufctx, "general.architecture");
+        if (keyidx != -1) {
+            if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "falcon") != 0) {
+                fprintf(stdout, "%s: model architecture not supported!\n", __func__);
+                return false;
+            }
+        } else {
+            fprintf(stdout, "%s: gguf model architecture not found!\n", __func__);
+            return false;
+        }
+
+    }
+
+    // load hparams
+    {
+        auto & hparams = model.hparams;
+
+        bool ok = true;
+        int keyidx;
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.context_length");
+                  if (keyidx != -1) { hparams.n_ctx = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.embedding_length");
+                  if (keyidx != -1) { hparams.n_embd = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.attention.head_count");
+                  if (keyidx != -1) { hparams.n_head = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.feed_forward_length");
+                  if (keyidx != -1) { hparams.n_ff = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.block_count");
+                  if (keyidx != -1) { hparams.n_block = gguf_get_val_u32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (ok) { keyidx = gguf_find_key(ggufctx, "falcon.attention.layer_norm_epsilon");
+                  if (keyidx != -1) { hparams.norm_eps= gguf_get_val_f32(ggufctx, keyidx); } else { ok = false; }  }
+
+        if (!ok) {
+            fprintf(stderr, "%s: required hparam missing!\n", __func__);
+            return false;
+        }
+
+        keyidx = gguf_find_key(ggufctx, "falcon.attention.head_count_kv");
+        if (keyidx != -1) { hparams.n_head_kv = gguf_get_val_u32(ggufctx, keyidx); }
+
+
+        printf("%s: n_ctx      = %d\n", __func__, hparams.n_ctx);
+        printf("%s: n_embd     = %d\n", __func__, hparams.n_embd);
+        printf("%s: n_head     = %d\n", __func__, hparams.n_head);
+        printf("%s: n_head_kv  = %d\n", __func__, hparams.n_head_kv);
+        printf("%s: n_block    = %d\n", __func__, hparams.n_block);
+        printf("%s: norm_eps   = %g\n", __func__, hparams.norm_eps);
+
+    }
+
+    // load vocab
+    {
+        auto & hparams = model.hparams;
+
+        int keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.model");
+
+        if (keyidx != -1) {
+            if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "gpt2") != 0) {
+                fprintf(stdout, "%s: tokenizer model not supported!\n", __func__);
+                return false;
+            }
+        } else {
+            fprintf(stdout, "%s: tokenizer model not found!\n", __func__);
+            return false;
+        }
+
+
+        int tokens_keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.tokens");
+
+        if (tokens_keyidx == -1) {
+            fprintf(stdout, "%s: gpt2 tokenizer vocab not found!\n", __func__);
+            return false;
+        }
+
+        int merges_keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.merges");
+
+        if (merges_keyidx == -1) {
+            fprintf(stdout, "%s: gpt2 tokenizer merges not found!\n", __func__);
+            return false;
+        }
+
+        hparams.n_vocab = gguf_get_arr_n(ggufctx,tokens_keyidx);
+        hparams.n_merges = gguf_get_arr_n(ggufctx,merges_keyidx);
+
+        fprintf(stdout, "%s: gpt2 tokenizer vocab  = %zu\n", __func__, hparams.n_vocab);
+        fprintf(stdout, "%s: gpt2 tokenizer merges = %zu\n", __func__, hparams.n_merges);
+
+        for (size_t i = 0; i < hparams.n_vocab; i++) {
+            std::string word = gguf_get_arr_str(ggufctx, tokens_keyidx, i);
+
+//            printf("token %d = '%s'\n",i,word.c_str() );
+
+            vocab.token_to_id[word] = i;
+            vocab.id_to_token[i] = word;
+
+        }
+
+        std::vector<std::pair<std::string, std::string>> bpe_merges;
+
+        for (size_t i = 0; i < hparams.n_merges; i++) {
+
+            std::string word = gguf_get_arr_str(ggufctx, merges_keyidx, i);
+
+            // Split the merges
+            std::string first, second;
+            size_t pos = word.find(' ', 1); // Start the search from the second character
+            if (pos != std::string::npos) {
+                first = word.substr(0, pos);
+                second = word.substr(pos + 1);
+            }
+
+            bpe_merges.push_back(std::make_pair(first, second));
+        }
+
+        vocab.populate_bpe_ranks(bpe_merges);
+
+
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_bos=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_eos=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_unk=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_sep=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_pad=true; }
+
+        if( vocab.special_have_bos ) { fprintf(stdout, "%s: bos token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
+        if( vocab.special_have_eos ) { fprintf(stdout, "%s: eos token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
+        if( vocab.special_have_unk ) { fprintf(stdout, "%s: unk token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
+        if( vocab.special_have_sep ) { fprintf(stdout, "%s: sep token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
+        if( vocab.special_have_pad ) { fprintf(stdout, "%s: pad token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
+    }
+
+
+    auto & ctx = model.ctx;
+    size_t ctx_size = ggml_get_mem_size(ctx);
+
+    printf("%s: ggml ctx size = %6.2f MB\n", __func__, ctx_size/(1024.0*1024.0));
+
+    // print tensor info
+    #if 0
+    {
+        const int n_tensors = gguf_get_n_tensors(ggufctx);
+
+        fprintf(stdout, "%s: n_tensors: %d\n", __func__, n_tensors);
+
+        for (int i = 0; i < n_tensors; ++i) {
+            const char * name   = gguf_get_tensor_name  (ggufctx, i);
+            const size_t offset = gguf_get_tensor_offset(ggufctx, i);
+
+            fprintf(stdout, "%s: tensor[%d]: name = %s, offset = %zu\n", __func__, i, name, offset);
+        }
+    }
+    #endif
+
+    // prepare memory for the weights
+    {
+
+        auto & hparams = model.hparams;
+
+        const int n_block = hparams.n_block;
+
+        model.blocks.resize(n_block);
+
+        model.tok_embeddings = ggml_get_tensor(ctx, "token_embd.weight");
+
+        model.output_norm = ggml_get_tensor(ctx, "output_norm.weight");
+        model.output_norm_b = ggml_get_tensor(ctx, "output_norm.bias");
+        model.lm_head = ggml_get_tensor(ctx, "output.weight");
+
+        // map by name
+        model.tensors["token_embd.weight"] = model.tok_embeddings;
+        model.tensors["output_norm.weight"] = model.output_norm;
+        model.tensors["output_norm.bias"] = model.output_norm_b;
+        model.tensors["output.weight"] = model.lm_head;
+
+        for (int i = 0; i < n_block; ++i) {
+
+            auto& block = model.blocks[i];
+            std::string blocknamestart = "blk." + std::to_string(i) + ".";
+
+            block.input_layernorm   =  get_tensor_ex(ctx, blocknamestart + "attn_norm.weight" );
+            block.input_layernorm_b =  get_tensor_ex(ctx, blocknamestart + "attn_norm.bias" );
+
+            if ( hparams.n_head_kv == 8 ) { // Falcon-40B
+                block.attention_norm   =  get_tensor_ex(ctx, blocknamestart + "attn_norm_2.weight" );
+                block.attention_norm_b =  get_tensor_ex(ctx, blocknamestart + "attn_norm_2.bias" );
+            }
+
+            // query_key_value shape for config.multi_query == True:
+            block.query_key_value = get_tensor_ex(ctx, blocknamestart + "attn_qkv.weight" );
+            block.wo = get_tensor_ex(ctx, blocknamestart + "attn_output.weight" );
+
+            block.ffn_up = get_tensor_ex(ctx, blocknamestart + "ffn_up.weight" );
+            block.ffn_down = get_tensor_ex(ctx, blocknamestart + "ffn_down.weight" );
+
+            // map by name
+            if ( hparams.n_head_kv == 8 ) { // Falcon-40B
+                // Falcon-40B:
+                model.tensors[blocknamestart + "attn_norm.weight"] = block.input_layernorm;
+                model.tensors[blocknamestart + "attn_norm.bias"] = block.input_layernorm_b;
+                model.tensors[blocknamestart + "attn_norm_2.weight"] = block.attention_norm;
+                model.tensors[blocknamestart + "attn_norm_2.bias"] = block.attention_norm_b;
+            } else {
+                // Falcon-7B:
+                model.tensors[blocknamestart + "attn_norm.weight"] = block.input_layernorm;
+                model.tensors[blocknamestart + "attn_norm.bias"] = block.input_layernorm_b;
+            }
+
+            model.tensors[blocknamestart + "attn_qkv.weight"] = block.query_key_value;
+            model.tensors[blocknamestart + "attn_output.weight"] = block.wo;
+
+            model.tensors[blocknamestart + "ffn_up.weight"] = block.ffn_up;
+            model.tensors[blocknamestart + "ffn_down.weight"] = block.ffn_down;
+        }
+    }
+
+    // key + value memory
+    {
+        const auto & kvctx = model.kvctx;
+        const auto & hparams = model.hparams;
+
+        const int n_block = hparams.n_block;
+        const int n_ctx   = hparams.n_ctx;
+        const int n_head_kv = hparams.n_head_kv;
+        const int head_dim = hparams.n_embd / hparams.n_head;
+
+        const int64_t n_mem      = n_block*n_ctx;
+        const int64_t n_elements = head_dim*n_mem;
+
+        // create the ggml context
+        {
+            struct ggml_init_params params = {
+                /*.mem_size   =*/ size_t(n_elements*4+ggml_tensor_overhead()*2),
+                /*.mem_buffer =*/ NULL,
+                /*.no_alloc   =*/ false,
+            };
+
+            model.kvctx = ggml_init(params);
+            if (!model.kvctx) {
+                fprintf(stderr, "%s: kv ggml_init() failed\n", __func__);
+                return false;
+            }
+
+        }
+
+
+        model.memory_k = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_head_kv * n_elements);
+        model.memory_v = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_head_kv * n_elements);
+
+        const size_t memory_size = ggml_nbytes(model.memory_k) + ggml_nbytes(model.memory_v);
+
+        printf("%s: memory_size = %8.2f MB, n_mem = %" PRId64 "\n", __func__, memory_size/1024.0/1024.0, n_mem);
+    }
+
+    return true;
+}
+
+
+// evaluate the transformer
+//
+//   - model:     the model
+//   - n_threads: number of threads to use
+//   - n_past:    the context size so far
+//   - embd_inp:  the embeddings of the tokens in the context
+//   - embd_w:    the predicted logits for the next token
+//
+bool falcon_eval(
+        const falcon_model & model,
+        const int n_threads,
+        const int n_past,
+        const std::vector<gpt2bpe_vocab::id> & embd_inp,
+              std::vector<float>         & embd_w,
+              size_t                     & mem_per_token) {
+
+
+    const int N = embd_inp.size();
+
+    const auto & hparams = model.hparams;
+
+    const int n_embd  = hparams.n_embd;
+    const int n_block = hparams.n_block;
+    const int n_ctx   = hparams.n_ctx;
+    const int n_head  = hparams.n_head;
+    const int n_head_kv = hparams.n_head_kv;
+    const int n_vocab = hparams.n_vocab;
+    const size_t head_dim = n_embd / n_head;
+
+    static size_t buf_size = 256u*1024*1024;
+    static void * buf = malloc(buf_size);
+
+    // use 2 scratch buffers
+    // TODO: very hacky solution - reimplement in a more elegant way
+    static size_t scr0_size = 256u*1024*1024;
+    static void * scr0 = malloc(scr0_size);
+
+    static size_t scr1_size = 256u*1024*1024;
+    static void * scr1 = malloc(scr1_size);
+
+    if (mem_per_token > 0 && mem_per_token*N > buf_size) {
+        const size_t buf_size_new = 1.1*(mem_per_token*N); // add 10% to account for ggml object overhead
+        //printf("\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, buf_size, buf_size_new);
+
+        // reallocate
+        buf_size = buf_size_new;
+        buf = realloc(buf, buf_size);
+        if (buf == nullptr) {
+            fprintf(stderr, "%s: failed to allocate %zu bytes\n", __func__, buf_size);
+            return false;
+        }
+    }
+
+    struct ggml_init_params params = {
+        /*.mem_size   =*/ buf_size,
+        /*.mem_buffer =*/ buf,
+        /*.no_alloc   =*/ false,
+    };
+
+    struct ggml_context * ctx0 = ggml_init(params);
+    struct ggml_cgraph gf = {};
+//    gf.n_threads = n_threads;
+
+    struct ggml_tensor * embd = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, N);
+    memcpy(embd->data, embd_inp.data(), N*ggml_element_size(embd));
+
+    // wte
+    struct ggml_tensor * inpL = ggml_get_rows(ctx0, model.tok_embeddings, embd);
+//    struct ggml_tensor* repeat_dummy = ggml_new_tensor_3d(ctx0, inpL->type, head_dim, N + n_past, n_head);
+
+    ggml_type wtype = GGML_TYPE_F32;
+    const int sizeof_wtype = ggml_type_sizef(wtype);
+
+    for (int il = 0; il < n_block; ++il) {
+        struct ggml_tensor * cur;
+        struct ggml_tensor * layernorm_output;
+
+        ggml_set_scratch(ctx0, { 0, scr0_size, scr0, });
+
+        // self-attention
+        {
+            layernorm_output = ggml_norm(ctx0, inpL);
+
+            layernorm_output = ggml_add(ctx0,
+                    ggml_mul(ctx0,
+                        ggml_repeat(ctx0, model.blocks[il].input_layernorm, layernorm_output),
+                        layernorm_output),
+                    ggml_repeat(ctx0, model.blocks[il].input_layernorm_b, layernorm_output));
+
+            if ( hparams.n_head_kv == 8 ) { // Falcon-40B
+                cur = ggml_norm(ctx0, inpL);
+
+                cur = ggml_add(ctx0,
+                        ggml_mul(ctx0,
+                            ggml_repeat(ctx0, model.blocks[il].attention_norm, cur),
+                            cur),
+                        ggml_repeat(ctx0, model.blocks[il].attention_norm_b, cur));
+            }
+            else { // Falcon 7B
+                cur = layernorm_output;
+            }
+
+            // compute QKV
+
+            cur = ggml_mul_mat(ctx0, model.blocks[il].query_key_value, cur);
+
+            // Note that the strides for Kcur, Vcur are set up so that the
+            // resulting views are misaligned with the tensor's storage
+            // (by applying the K/V offset we shift the tensor's original
+            // view to stick out behind the viewed QKV tensor's allocated
+            // memory, so to say). This is ok because no actual accesses
+            // happen to that out-of-range memory, but it can require some
+            // trickery when trying to accurately dump these views for
+            // debugging.
+
+            struct ggml_tensor * Qcur = ggml_view_3d(
+                ctx0, cur, head_dim, n_head, N,
+                head_dim * sizeof_wtype,
+                head_dim * (n_head + 2 * n_head_kv) * sizeof_wtype,
+                0);
+
+            struct ggml_tensor * Kcur = ggml_view_3d(
+                ctx0, cur, head_dim, n_head_kv, N,
+                head_dim * sizeof_wtype,
+                head_dim * (n_head + 2 * n_head_kv) * sizeof_wtype,
+                head_dim * n_head * sizeof_wtype);
+
+            struct ggml_tensor * Vcur = ggml_view_3d(
+                ctx0, cur, head_dim, n_head_kv, N,
+                head_dim * sizeof_wtype,
+                head_dim * (n_head + 2 * n_head_kv) * sizeof_wtype,
+                head_dim * (n_head + n_head_kv) * sizeof_wtype);
+
+            // using mode = 2 for neox mode
+            Qcur = ggml_rope_inplace(ctx0, Qcur, n_past, head_dim, 2, 0);
+            Kcur = ggml_rope_inplace(ctx0, Kcur, n_past, head_dim, 2, 0);
+
+            // store key and value to memory
+            {
+                struct ggml_tensor* k = ggml_view_1d(
+                    ctx0, model.memory_k, N * n_head_kv * head_dim,
+                    (ggml_element_size(model.memory_k) * n_head_kv * head_dim) *
+                        (il * n_ctx + n_past));
+                struct ggml_tensor* v = ggml_view_1d(
+                    ctx0, model.memory_v, N * n_head_kv * head_dim,
+                    (ggml_element_size(model.memory_v) * n_head_kv * head_dim) *
+                        (il * n_ctx + n_past));
+
+                ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Kcur, k));
+                ggml_build_forward_expand(&gf, ggml_cpy(ctx0, Vcur, v));
+            }
+
+            struct ggml_tensor * K = ggml_permute(
+                ctx0,
+                ggml_reshape_3d(
+                    ctx0,
+                    ggml_view_1d(ctx0, model.memory_k, (n_past + N) * n_head_kv * head_dim,
+                                 il * n_ctx *
+                                     ggml_element_size(model.memory_k) *
+                                     n_head_kv *
+                                     head_dim),
+                    head_dim, n_head_kv, n_past + N),
+                0, 2, 1, 3);
+
+            // K * Q
+
+//            K = ggml_cont(ctx0, ggml_repeat2(ctx0, K, repeat_dummy));
+
+            struct ggml_tensor * Q = ggml_permute(ctx0, Qcur, 0, 2, 1, 3);
+            struct ggml_tensor * KQ = ggml_mul_mat(ctx0, K, Q);
+
+            // KQ_scaled = KQ / sqrt(n_embd/n_head)
+            struct ggml_tensor * KQ_scaled =
+                ggml_scale_inplace(ctx0,
+                        KQ,
+                        ggml_new_f32(ctx0, 1.0f/sqrt(float(head_dim)))
+                        );
+
+            // KQ_masked = mask_past(KQ_scaled)
+            struct ggml_tensor * KQ_masked = ggml_diag_mask_inf_inplace(ctx0, KQ_scaled, n_past);
+
+            // KQ = soft_max(KQ_masked)
+            struct ggml_tensor * KQ_soft_max = ggml_soft_max_inplace(ctx0, KQ_masked);
+
+            // V_trans = Vmem.view(n_embd/n_head, n_head, n_past + N).permute(1, 2, 0, 3).contiguous()
+            struct ggml_tensor* V = ggml_permute(
+                ctx0,
+                ggml_reshape_3d(
+                    ctx0,
+                    ggml_view_1d(ctx0, model.memory_v, (n_past + N) * n_head_kv * head_dim,
+                                 il * n_ctx *
+                                     ggml_element_size(model.memory_v) *
+                                     n_head_kv *
+                                     head_dim),
+                    head_dim, n_head_kv, n_past + N),
+                0, 2, 1, 3);
+
+//            V = ggml_cont(ctx0, ggml_transpose(ctx0, ggml_repeat2(ctx0, V, repeat_dummy)));
+            V = ggml_cont(ctx0, ggml_transpose(ctx0, V));
+
+            // KQV = transpose(V) * KQ_soft_max
+            struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ_soft_max);
+
+            // KQV_merged = KQV.permute(0, 2, 1, 3)
+            struct ggml_tensor * KQV_merged = ggml_permute(ctx0, KQV, 0, 2, 1, 3);
+
+            // cur = KQV_merged.contiguous().view(n_embd, N)
+            cur = ggml_cpy(ctx0,
+                    KQV_merged,
+                    ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+
+            // projection
+            {
+                cur = ggml_mul_mat(ctx0,
+                        model.blocks[il].wo,
+                        cur);
+            }
+        }
+
+        ggml_set_scratch(ctx0, { 0, scr1_size, scr1, });
+
+        struct ggml_tensor* inpFF = layernorm_output;
+        struct ggml_tensor* attn_out = ggml_cpy(
+            ctx0, cur, ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, n_embd, N));
+
+        {
+            cur = ggml_mul_mat(ctx0, model.blocks[il].ffn_up, inpFF);
+            cur = ggml_gelu(ctx0, cur);
+            cur = ggml_mul_mat(ctx0, model.blocks[il].ffn_down, cur);
+        }
+
+        cur = ggml_add(ctx0, cur, attn_out);
+        cur = ggml_add(ctx0, cur, inpL);
+        // input for next layer
+        inpL = cur;
+    }
+
+    ggml_set_scratch(ctx0, { 0, scr0_size, scr0, });
+
+    // norm
+    {
+        inpL = ggml_norm(ctx0, inpL);
+
+        // inpL = ln_f_g*inpL + ln_f_b
+        inpL = ggml_add(ctx0,
+                ggml_mul(ctx0,
+                    ggml_repeat(ctx0, model.output_norm, inpL),
+                    inpL),
+                ggml_repeat(ctx0, model.output_norm_b, inpL));
+    }
+
+    ggml_set_scratch(ctx0, { 0, 0, nullptr, });
+
+    // lm_head
+    {
+        inpL = ggml_mul_mat(ctx0, model.lm_head, inpL);
+
+        //inpL = ggml_add(ctx0,
+        //        ggml_repeat(ctx0, model.lmh_b, inpL),
+        //        inpL);
+    }
+
+    // logits -> probs
+    //inpL = ggml_soft_max_inplace(ctx0, inpL);
+
+    // run the computation
+    ggml_build_forward_expand(&gf, inpL);
+//    ggml_graph_compute       (ctx0, &gf);
+    ggml_graph_compute_with_ctx(ctx0, &gf, n_threads);
+
+    //if (n_past%100 == 0) {
+    //    ggml_graph_print   (&gf);
+    //    ggml_graph_dump_dot(&gf, NULL, "gpt-2.dot");
+    //}
+
+    // return result for just the last token
+    embd_w.resize(n_vocab);
+    memcpy(embd_w.data(), (float *)ggml_get_data(inpL) + (n_vocab * (N - 1)), sizeof(float) * n_vocab);
+
+    if (mem_per_token == 0) {
+        mem_per_token = ggml_used_mem(ctx0)/N;
+    }
+    //printf("used_mem = %zu\n", ggml_used_mem(ctx0));
+
+    ggml_free(ctx0);
+
+    return true;
+}
+
+int main(int argc, char ** argv) {
+    ggml_time_init();
+
+    const int64_t t_main_start_us = ggml_time_us();
+
+    gpt_params params;
+
+    if (gpt_params_parse(argc, argv, params) == false) {
+        return 1;
+    }
+
+    int64_t t_load_us = 0;
+
+    gpt2bpe_vocab vocab;
+    falcon_model model;
+
+    // load the model
+    {
+        const int64_t t_start_us = ggml_time_us();
+
+        if (!falcon_model_load(params.model, model, vocab)) {
+            fprintf(stderr, "%s: failed to load model from '%s'\n", __func__, params.model.c_str());
+            return 1;
+        }
+
+        t_load_us = ggml_time_us() - t_start_us;
+
+    }
+
+    if (params.seed < 0) {
+        params.seed = time(NULL);
+    }
+
+    if (params.top_k == 0) {
+        params.top_k = model.hparams.n_vocab;
+    }
+
+    printf("%s: seed           = %d\n",   __func__, params.seed);
+    printf("%s: temp           = %.3f\n", __func__, params.temp);
+    printf("%s: top_k          = %d\n",   __func__, params.top_k);
+    printf("%s: top_p          = %.3f\n", __func__, params.top_p);
+    printf("%s: repeat_last_n  = %d\n",   __func__, params.repeat_last_n);
+    printf("%s: repeat_penalty = %.3f\n", __func__, params.repeat_penalty);
+
+    std::mt19937 rng(params.seed);
+
+    if (params.prompt.empty()) {
+        params.prompt = "Once upon";
+    }
+
+    std::vector<int32_t> last_n_tokens(model.hparams.n_ctx);
+    std::fill(last_n_tokens.begin(), last_n_tokens.end(), 0);
+
+    int n_past = 0;
+
+    int64_t t_sample_us  = 0;
+    int64_t t_predict_us = 0;
+
+    std::vector<float> logits;
+
+    // tokenize the prompt
+    std::vector<gpt2bpe_vocab::id> embd_inp = gpt2bpe_tokenize(vocab, params.prompt,false, false);
+
+    params.n_predict = std::min(params.n_predict, model.hparams.n_ctx - (int) embd_inp.size());
+
+    printf("%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
+//    for (size_t i = 0; i < embd_inp.size(); i++) {
+//        printf("%s: token[%zu] = %6d, %s\n", __func__, i, embd_inp[i], vocab.id_to_token[embd_inp[i]].c_str());
+//    }
+
+    if( model.hparams.n_ctx < params.n_predict+embd_inp.size() ) {
+        params.n_predict = model.hparams.n_ctx-embd_inp.size();
+    }
+
+    printf("%s: n_predict = %d\n", __func__, params.n_predict);
+    printf("\n");
+
+    std::vector<gpt2bpe_vocab::id> embd;
+
+    // determine the required inference memory per token:
+    size_t mem_per_token = 0;
+    falcon_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
+
+    for (size_t i = embd.size(); i < embd_inp.size() + params.n_predict; i++) {
+        // predict
+        if (embd.size() > 0) {
+            const int64_t t_start_us = ggml_time_us();
+
+            if (!falcon_eval(model, params.n_threads, n_past, embd, logits, mem_per_token)) {
+                printf("Failed to predict\n");
+                return 1;
+            }
+
+            t_predict_us += ggml_time_us() - t_start_us;
+        }
+
+        n_past += embd.size();
+        embd.clear();
+
+        if (i >= embd_inp.size()) {
+            // sample next token
+            const int   top_k = params.top_k;
+            const float top_p = params.top_p;
+            const float temp  = params.temp;
+            const int repeat_last_n = params.repeat_last_n;
+            const float repeat_penalty = params.repeat_penalty;
+
+            const int n_vocab = model.hparams.n_vocab;
+
+            gpt2bpe_vocab::id id = 0;
+
+            {
+                const int64_t t_start_sample_us = ggml_time_us();
+
+                id = sample_top_k_top_p_repeat(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens.data(), last_n_tokens.size(), top_k, top_p, temp, repeat_last_n, repeat_penalty, rng);
+
+                last_n_tokens.erase(last_n_tokens.begin());
+                last_n_tokens.push_back(id);
+
+                t_sample_us += ggml_time_us() - t_start_sample_us;
+            }
+
+            // add it to the context
+            embd.push_back(id);
+        } else {
+            // if here, it means we are still processing the input prompt
+            for (size_t k = i; k < embd_inp.size(); k++) {
+                embd.push_back(embd_inp[k]);
+                if (embd.size() > params.n_batch) {
+                    break;
+                }
+            }
+            i += embd.size() - 1;
+        }
+
+        // display text
+        for (auto id : embd) {
+            printf("%s", vocab.id_to_token[id].c_str()  );
+        }
+        fflush(stdout);
+
+        // end of text token
+        if (vocab.special_have_eos && embd.back() == vocab.special_eos_id) {
+            break;
+        }
+    }
+
+    // report timing
+    {
+        const int64_t t_main_end_us = ggml_time_us();
+
+        printf("\n\n");
+        printf("%s: mem per token = %8zu bytes\n", __func__, mem_per_token);
+        printf("%s:     load time = %8.2f ms\n", __func__, t_load_us/1000.0f);
+        printf("%s:   sample time = %8.2f ms\n", __func__, t_sample_us/1000.0f);
+        printf("%s:  predict time = %8.2f ms / %.2f ms per token\n", __func__, t_predict_us/1000.0f, t_predict_us/1000.0f/n_past);
+        printf("%s:    total time = %8.2f ms\n", __func__, (t_main_end_us - t_main_start_us)/1000.0f);
+    }
+
+    ggml_free(model.ctx);
+
+    return 0;
+}

From 16ab9ba3b3438960fe69e15fbc60515cc0b8467f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 00:31:28 +0200
Subject: [PATCH 207/242] convert-falcon-hf-to-gguf.py : remove extra kv

---
 convert-falcon-hf-to-gguf.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py
index 7d50d8c5caaff..f06c6b395a4df 100644
--- a/convert-falcon-hf-to-gguf.py
+++ b/convert-falcon-hf-to-gguf.py
@@ -94,7 +94,6 @@ def count_model_parts(dir_model: str) -> int:
 
 block_count = hparams["n_layer"]
 
-gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
 gguf_writer.add_context_length(2048)
 gguf_writer.add_embedding_length(hparams["hidden_size"])

From c0e4ca630b09ed0bfd3f648247afa46347de5e78 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 00:31:56 +0200
Subject: [PATCH 208/242] convert-gptneox-hf-to-gguf.py : remove extra kv

---
 convert-gptneox-hf-to-gguf.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/convert-gptneox-hf-to-gguf.py b/convert-gptneox-hf-to-gguf.py
index e49a619cf289d..a7cefc6f35e04 100644
--- a/convert-gptneox-hf-to-gguf.py
+++ b/convert-gptneox-hf-to-gguf.py
@@ -96,7 +96,6 @@ def count_model_parts(dir_model: str) -> int:
 
 block_count = hparams["num_hidden_layers"]
 
-gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
 gguf_writer.add_context_length(hparams["max_position_embeddings"])
 gguf_writer.add_embedding_length(hparams["hidden_size"])

From 593b04fdcd5df4882b268b9cf22f43bf8321f23d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 00:32:27 +0200
Subject: [PATCH 209/242] convert-llama-7b-pth-to-gguf.py : remove extra kv

---
 convert-llama-7b-pth-to-gguf.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 89b58e132ef74..7c767df609301 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -96,7 +96,6 @@ def count_model_parts(dir_model: str) -> int:
 else:
     hf_repo = ""
 
-gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
 gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_tensor_data_layout("Meta AI original pth")

From 281d6d1105ca26d060b61e28c28f2087c8fad94e Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 00:32:56 +0200
Subject: [PATCH 210/242] convert-llama-hf-to-gguf.py : remove extra kv

---
 convert-llama-hf-to-gguf.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index 7d4571f4731dc..a998a6f496e91 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -114,7 +114,6 @@ def count_model_parts(dir_model: str) -> int:
     sys.exit()
 
 
-gguf_writer.add_architecture()
 gguf_writer.add_name(last_dir)
 gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_tensor_data_layout("Meta AI original pth")

From bd5a57901bc9699f929ff89e70b9dc885c80fcaa Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 01:01:52 +0200
Subject: [PATCH 211/242] gguf.py : fix for falcon 40b

---
 gguf.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gguf.py b/gguf.py
index c4290d64a2f4e..e5eb85ded3f59 100644
--- a/gguf.py
+++ b/gguf.py
@@ -220,7 +220,7 @@ def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict:
         tensor_map["transformer.h."+str(i)+".ln_1"]              = mapped_to  # gpt2
         tensor_map["transformer.blocks."+str(i)+".norm_1"]       = mapped_to  # mpt
         tensor_map["transformer.h."+str(i)+".input_layernorm"]   = mapped_to  # falcon7b
-        tensor_map["transformer.h."+str(i)+".ln_attn"]           = mapped_to  # falcon40b
+        tensor_map["transformer.h."+str(i)+".ln_mlp"]            = mapped_to  # falcon40b
         tensor_map["model.layers."+str(i)+".input_layernorm"]    = mapped_to  # llama-hf
         tensor_map["layers."+str(i)+".attention_norm"]           = mapped_to  # llama-pth
 
@@ -228,7 +228,7 @@ def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict:
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM_2, None)
         mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None
 
-        tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to  # falcon40b
+        tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to  # falcon40b
 
         # Attention query-key-value
         mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_QKV, None)

From 1d80eea57456f9b62d8256219baee85488708db7 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 01:03:37 +0200
Subject: [PATCH 212/242] falcon-main.cpp : fix for falcon 40b

---
 examples/gptneox-wip/falcon-main.cpp | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/examples/gptneox-wip/falcon-main.cpp b/examples/gptneox-wip/falcon-main.cpp
index c97aa602acbd9..5be192a48dd16 100644
--- a/examples/gptneox-wip/falcon-main.cpp
+++ b/examples/gptneox-wip/falcon-main.cpp
@@ -605,11 +605,10 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
 
         const int n_block = hparams.n_block;
         const int n_ctx   = hparams.n_ctx;
-        const int n_head_kv = hparams.n_head_kv;
-        const int head_dim = hparams.n_embd / hparams.n_head;
+        const int n_embd = hparams.n_embd;
 
         const int64_t n_mem      = n_block*n_ctx;
-        const int64_t n_elements = head_dim*n_mem;
+        const int64_t n_elements = n_embd*n_mem;
 
         // create the ggml context
         {
@@ -628,8 +627,8 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
         }
 
 
-        model.memory_k = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_head_kv * n_elements);
-        model.memory_v = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_head_kv * n_elements);
+        model.memory_k = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_elements);
+        model.memory_v = ggml_new_tensor_1d(kvctx, GGML_TYPE_F16, n_elements);
 
         const size_t memory_size = ggml_nbytes(model.memory_k) + ggml_nbytes(model.memory_v);
 

From 2c8055b65bcf327742267acf48ce3919a5af557b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 01:08:39 +0200
Subject: [PATCH 213/242] convert-falcon-hf-to-gguf.py : update ref

---
 convert-falcon-hf-to-gguf.py | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py
index f06c6b395a4df..1ec470695b2e1 100644
--- a/convert-falcon-hf-to-gguf.py
+++ b/convert-falcon-hf-to-gguf.py
@@ -193,11 +193,8 @@ def count_model_parts(dir_model: str) -> int:
 tensor_map = gguf.get_tensor_name_map(ARCH,block_count)
 
 # params for qkv transform
-if "n_head_kv" in hparams:
-    n_head_kv = hparams["n_head_kv"]
-else:
-    n_head_kv = 1
 n_head = hparams["n_head"]
+n_head_kv = hparams["n_head_kv"] if "n_head_kv" in hparams else 1
 head_dim = hparams["hidden_size"] // n_head
 
 # tensor info
@@ -231,7 +228,7 @@ def count_model_parts(dir_model: str) -> int:
         # So we rearrange them here,, so that we have n_head query weights
         # followed by n_head_kv key weights followed by n_head_kv value weights,
         # in contiguous fashion.
-        # ref: https://github.com/jploski/ggml/blob/d5295b477fb36c69468c3fecb0393a8d7980b7c8/examples/falcon/convert-hf-to-ggml.py#L107-L123
+        # ref: https://github.com/jploski/ggml/blob/falcon40b/examples/falcon/convert-hf-to-ggml.py
 
         if "query_key_value" in name:
             qkv = data.view(n_head_kv, n_head // n_head_kv + 2, head_dim, head_dim * n_head)

From b3a7a2b4867eec1d4e3371f598d54d3a77ae9ec4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 12:05:11 +0200
Subject: [PATCH 214/242] convert-falcon-hf-to-gguf.py : add tensor data layout

---
 convert-falcon-hf-to-gguf.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py
index 1ec470695b2e1..e0da3a04de811 100644
--- a/convert-falcon-hf-to-gguf.py
+++ b/convert-falcon-hf-to-gguf.py
@@ -95,7 +95,8 @@ def count_model_parts(dir_model: str) -> int:
 block_count = hparams["n_layer"]
 
 gguf_writer.add_name(last_dir)
-gguf_writer.add_context_length(2048)
+gguf_writer.add_context_length(2048) # not in config.json
+gguf_writer.add_tensor_data_layout("jploski") # qkv tensor transform
 gguf_writer.add_embedding_length(hparams["hidden_size"])
 gguf_writer.add_feed_forward_length(4 * hparams["hidden_size"])
 gguf_writer.add_block_count(block_count)

From dadf098b5a4325bc53d90ef00801614b547ee1d4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 12:06:22 +0200
Subject: [PATCH 215/242] cmpnct_gpt2bpe.hpp : fixes

---
 examples/gptneox-wip/cmpnct_gpt2bpe.hpp | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)

diff --git a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
index ac5f8c672a305..64648b646085f 100644
--- a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
+++ b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
@@ -445,17 +445,13 @@ struct gpt2bpe_vocab {
     std::vector<std::pair<std::string, std::string>> bpe_merges;
     std::map<std::string, int> special_tokens;
 
-    id special_bos_id = 0;
-    id special_eos_id = 0;
-    id special_unk_id = 0;
-    id special_sep_id = 0;
-    id special_pad_id = 0;
-
-    bool special_have_bos = false;
-    bool special_have_eos = false;
-    bool special_have_unk = false;
-    bool special_have_sep = false;
-    bool special_have_pad = false;
+    id special_bos_id = -1;
+    id special_eos_id = -1;
+    id special_unk_id = -1;
+    id special_sep_id = -1;
+    id special_pad_id = -1;
+
+    id linefeed_id = -1;
 
     std::unordered_map<token, id> token_to_id;
     std::unordered_map<id, token> id_to_token;
@@ -1002,7 +998,7 @@ static std::vector<gpt2bpe_vocab::id> gpt2bpe_tokenize(const gpt2bpe_vocab & voc
         return output;
     }
 
-    if (bos && vocab.special_have_bos) {
+    if (bos && vocab.special_bos_id != -1) {
         output.push_back(vocab.special_bos_id);
     }
 

From 781bf2481fc4af617fcf7adbe654cb11b570e4f4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 12:08:17 +0200
Subject: [PATCH 216/242] falcon-main.cpp : fixes

---
 examples/gptneox-wip/falcon-main.cpp | 40 ++++++++++++++++++++--------
 1 file changed, 29 insertions(+), 11 deletions(-)

diff --git a/examples/gptneox-wip/falcon-main.cpp b/examples/gptneox-wip/falcon-main.cpp
index 5be192a48dd16..43b6a29f312ed 100644
--- a/examples/gptneox-wip/falcon-main.cpp
+++ b/examples/gptneox-wip/falcon-main.cpp
@@ -378,6 +378,7 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
     {
         int keyidx;
 
+        // check model architecture kv
         keyidx = gguf_find_key(ggufctx, "general.architecture");
         if (keyidx != -1) {
             if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "falcon") != 0) {
@@ -389,6 +390,18 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
             return false;
         }
 
+        // check model tensor data layout kv
+        keyidx = gguf_find_key(ggufctx, "falcon.tensor_data_layout");
+        if (keyidx != -1) {
+            if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "jploski") != 0) {
+                fprintf(stdout, "%s: model tensor data layout not supported!\n", __func__);
+                return false;
+            }
+        } else {
+            fprintf(stdout, "%s: gguf model tensor data layout not found!\n", __func__);
+            return false;
+        }
+
     }
 
     // load hparams
@@ -479,6 +492,9 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
             vocab.token_to_id[word] = i;
             vocab.id_to_token[i] = word;
 
+            if( vocab.id_to_token[i] == "\n" ) {
+                vocab.linefeed_id = i;
+            }
         }
 
         std::vector<std::pair<std::string, std::string>> bpe_merges;
@@ -501,17 +517,19 @@ bool falcon_model_load(const std::string & fname, falcon_model & model, gpt2bpe_
         vocab.populate_bpe_ranks(bpe_merges);
 
 
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_bos=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_eos=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_unk=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_sep=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_pad=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+
+        if( vocab.special_bos_id != -1 ) { fprintf(stdout, "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
+        if( vocab.special_eos_id != -1 ) { fprintf(stdout, "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
+        if( vocab.special_unk_id != -1 ) { fprintf(stdout, "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
+        if( vocab.special_sep_id != -1 ) { fprintf(stdout, "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
+        if( vocab.special_pad_id != -1 ) { fprintf(stdout, "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
+        if( vocab.linefeed_id    != -1 ) { fprintf(stdout, "%s: LF token  = %d\n",      __func__, vocab.linefeed_id ); }
 
-        if( vocab.special_have_bos ) { fprintf(stdout, "%s: bos token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
-        if( vocab.special_have_eos ) { fprintf(stdout, "%s: eos token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
-        if( vocab.special_have_unk ) { fprintf(stdout, "%s: unk token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
-        if( vocab.special_have_sep ) { fprintf(stdout, "%s: sep token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
-        if( vocab.special_have_pad ) { fprintf(stdout, "%s: pad token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
     }
 
 
@@ -1070,7 +1088,7 @@ int main(int argc, char ** argv) {
         fflush(stdout);
 
         // end of text token
-        if (vocab.special_have_eos && embd.back() == vocab.special_eos_id) {
+        if (vocab.special_eos_id != -1 && embd.back() == vocab.special_eos_id) {
             break;
         }
     }

From 8945d47f52ef6ed852dcc4690b10f0fed8585a51 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 12:09:24 +0200
Subject: [PATCH 217/242] gptneox-main.cpp : fixes

---
 examples/gptneox-wip/gptneox-main.cpp | 28 ++++++++++++++++-----------
 1 file changed, 17 insertions(+), 11 deletions(-)

diff --git a/examples/gptneox-wip/gptneox-main.cpp b/examples/gptneox-wip/gptneox-main.cpp
index aca8a5b0a3bd5..04af50245ff77 100644
--- a/examples/gptneox-wip/gptneox-main.cpp
+++ b/examples/gptneox-wip/gptneox-main.cpp
@@ -391,6 +391,7 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
     {
         int keyidx;
 
+        // check model architecture kv
         keyidx = gguf_find_key(ggufctx, "general.architecture");
         if (keyidx != -1) {
             if ( strcmp(gguf_get_val_str(ggufctx, keyidx), "gptneox") != 0) {
@@ -492,6 +493,9 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
             vocab.token_to_id[word] = i;
             vocab.id_to_token[i] = word;
 
+            if( vocab.id_to_token[i] == "\n" ) {
+                vocab.linefeed_id = i;
+            }
         }
 
         std::vector<std::pair<std::string, std::string>> bpe_merges;
@@ -514,17 +518,18 @@ bool gpt_neox_model_load(const std::string & fname, gpt_neox_model & model, gpt2
         vocab.populate_bpe_ranks(bpe_merges);
 
 
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_bos=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_eos=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_unk=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_sep=true; }
-        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); vocab.special_have_pad=true; }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.bos_token_id"); if( keyidx != -1 ) {       vocab.special_bos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.eos_token_id"); if( keyidx != -1 ) {       vocab.special_eos_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.unknown_token_id"); if( keyidx != -1 ) {   vocab.special_unk_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.separator_token_id"); if( keyidx != -1 ) { vocab.special_sep_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
+        keyidx = gguf_find_key(ggufctx, "tokenizer.ggml.padding_token_id"); if( keyidx != -1 ) {   vocab.special_pad_id = (int32_t)gguf_get_val_u32(ggufctx, keyidx); }
 
-        if( vocab.special_have_bos ) { fprintf(stdout, "%s: bos token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
-        if( vocab.special_have_eos ) { fprintf(stdout, "%s: eos token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
-        if( vocab.special_have_unk ) { fprintf(stdout, "%s: unk token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
-        if( vocab.special_have_sep ) { fprintf(stdout, "%s: sep token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
-        if( vocab.special_have_pad ) { fprintf(stdout, "%s: pad token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
+        if( vocab.special_bos_id != -1 ) { fprintf(stdout, "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].c_str() ); }
+        if( vocab.special_eos_id != -1 ) { fprintf(stdout, "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].c_str() ); }
+        if( vocab.special_unk_id != -1 ) { fprintf(stdout, "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].c_str() ); }
+        if( vocab.special_sep_id != -1 ) { fprintf(stdout, "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].c_str() ); }
+        if( vocab.special_pad_id != -1 ) { fprintf(stdout, "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].c_str() ); }
+        if( vocab.linefeed_id    != -1 ) { fprintf(stdout, "%s: LF token  = %d\n",      __func__, vocab.linefeed_id ); }
     }
 
 
@@ -656,6 +661,7 @@ ggml_tensor * gpt_neox_ff(
         const gpt_neox_block &block,
         ggml_context * ctx0,
         ggml_tensor * inp) {
+
     ggml_tensor * cur = ggml_norm(ctx0, inp);
 
     cur = ggml_add(ctx0, ggml_mul(ctx0, ggml_repeat(ctx0, block.ln_2_g, cur), cur), ggml_repeat(ctx0, block.ln_2_b, cur));
@@ -1053,7 +1059,7 @@ int main(int argc, char ** argv) {
         fflush(stdout);
 
         // end of text token
-        if (vocab.special_have_eos && embd.back() == vocab.special_eos_id) {
+        if (vocab.special_eos_id != -1 && embd.back() == vocab.special_eos_id) {
             break;
         }
     }

From 6a2e520095356a8d23682f8fba421f955c4ce091 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 13:19:02 +0200
Subject: [PATCH 218/242] cmpnct_gpt2bpe.hpp : remove non-general stuff

---
 examples/gptneox-wip/cmpnct_gpt2bpe.hpp | 120 ++++--------------------
 1 file changed, 18 insertions(+), 102 deletions(-)

diff --git a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
index 64648b646085f..81ea443b2a20a 100644
--- a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
+++ b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
@@ -14,6 +14,7 @@
 //-----
 // Unicode GPT2 Byte Pair Encoding Tokenizer
 // Adapted from https://github.com/cmp-nct/ggllm.cpp
+// Removed loading of merges from HF json and parts made for a specific vocab
 //-----
 
 // Unicode library (from cmpnct_unicode.cpp)
@@ -439,11 +440,10 @@ struct Trie {
 struct gpt2bpe_vocab {
     using id = int32_t;
     using token = std::string;
-    std::map<std::string, uint32_t> max_token_length; // max length, for each 2byte prefix
 
+    std::map<std::string, uint32_t> max_token_length; // max length, for each 2byte prefix
     std::map<std::pair<std::string,std::string>, int> bpe_ranks;
     std::vector<std::pair<std::string, std::string>> bpe_merges;
-    std::map<std::string, int> special_tokens;
 
     id special_bos_id = -1;
     id special_eos_id = -1;
@@ -476,22 +476,6 @@ struct gpt2bpe_vocab {
             bpe_ranks.emplace(bpe_merges_[i], i);
         }
         bpe_merges = bpe_merges_;
-
-        // populate special tokens too (0-11 and if available 65024++)
-
-        #if 0
-        for (int i = 0; i < 12; i++) {
-            special_tokens[id_to_token[i].tok] = i;
-        }
-        for (int i = 65024; i < (int)id_to_token.size(); i++) {
-            special_tokens[id_to_token[i].tok] = i;
-        }
-        #endif
-
-        // token_to_id["</s>"] = 11; // bugfix for TII instruct training (blocks stopwords)
-        // special_tokens["</s>"] = 11; // bugfix for TII instruct training (blocks stopwords)
-
-
         return bpe_merges_.size();
     }
 
@@ -508,10 +492,6 @@ struct gpt2bpe_vocab {
         }).base(), str.end());
     }
 
-    // removed, merges loaded from gguf model file:
-    // requires the standard HF type tokenizer.json (pretty printed)
-    // std::vector<std::pair<std::string, std::string>> parse_json_to_bpe_merges(const std::string& filename) {
-
     // get max token length available for a prefix of 2 bytes (string at least 2 bytes long)
     int get_max_token_length(const std::string& string) const {
         if (string.size() < 2)
@@ -609,45 +589,27 @@ struct gpt2bpe_tokenizer {
         {
             work_queue_ = ggllm_bpe_bigram::queue();
             symbols_.clear();
-            bool is_special = false;
-            for (auto it = vocab_.special_tokens.begin(); it != vocab_.special_tokens.end(); ++it)
-            {
-                std::string special_token = it->first;
-                if (word.compare(special_token) == 0)
-                {
-                    ggllm_bpe_symbol sym;
-                    sym.text = word.c_str();
-                    sym.n = word.size();
-                    sym.prev = -1;
-                    sym.next = -1;
-                    symbols_.emplace_back(sym);
-                    is_special = true;
-                    break;
-                }
-            }
 
             int index = 0;
             size_t offset = 0;
-            if (!is_special)
-            {
 
-                while (offset < word.size())
-                {
-                    ggllm_bpe_symbol sym;
-                    size_t char_len = std::min(word.size() - offset, (size_t) CNCTUnicode::utf8_len(word[offset]));
-                    sym.text = word.c_str() + offset;
-                    sym.n = 1;
-                    sym.n = char_len;
-                    offset += sym.n;
-                    sym.prev = index - 1;
-                    sym.next = offset == word.size() ? -1 : index + 1;
-                    index++;
-                    symbols_.emplace_back(sym);
-                }
-                for (size_t i = 1; i < symbols_.size(); ++i) {
-                    add_new_bigram(i - 1, i);
-                }
+            while (offset < word.size())
+            {
+                ggllm_bpe_symbol sym;
+                size_t char_len = std::min(word.size() - offset, (size_t) CNCTUnicode::utf8_len(word[offset]));
+                sym.text = word.c_str() + offset;
+                sym.n = 1;
+                sym.n = char_len;
+                offset += sym.n;
+                sym.prev = index - 1;
+                sym.next = offset == word.size() ? -1 : index + 1;
+                index++;
+                symbols_.emplace_back(sym);
             }
+            for (size_t i = 1; i < symbols_.size(); ++i) {
+                add_new_bigram(i - 1, i);
+            }
+
             // build token(s)
             while (!work_queue_.empty())
             {
@@ -790,17 +752,6 @@ struct gpt2bpe_tokenizer {
         bpe_encoded_words.reserve(text.size());
 
         text_utf = CNCTUnicode::split_utf8_enhanced(text);
-        std::map<std::string, int> special_tokens = vocab_.special_tokens;
-        int smallest_len_special_tokens = 0;
-        if (special_tokens.size())
-        {
-            smallest_len_special_tokens = special_tokens.begin()->first.size();
-            for (auto it = special_tokens.begin(); it != special_tokens.end(); ++it)
-            {
-                if (it->first.size() < (size_t)smallest_len_special_tokens)
-                    smallest_len_special_tokens = it->first.size();
-            }
-        }
 
         for (int i = 0; i < (int)text_utf.size(); i++)
         {
@@ -813,41 +764,6 @@ struct gpt2bpe_tokenizer {
             const CNCTString &utf_char_next_next = (i+2 < (int)text_utf.size()) ? text_utf[i+2] : CNCTString();
             // const CNCTString &utf_char_prev = (i > 0) ? text_utf[i-1] : CNCTString();
 
-        // handling special tokens
-            bool special_token_found = false;
-            if (bytes_remain >= (int)smallest_len_special_tokens)
-            for (auto it = special_tokens.begin(); it != special_tokens.end(); ++it)
-            {
-                if ((bytes_remain) < (int)it->first.size())
-                    continue;
-
-                if (str_is_equal(text_pos, it->first.c_str(), it->first.size()))
-                {
-                    if (token.size())
-                    {
-                        bpe_words.emplace_back(token); // push previous content as token
-                        token.clear();
-                        collecting = false;
-                        collecting_letter = false;
-                        collecting_numeric = false;
-                        collecting_special = false;
-                        collecting_whitespace_lookahead = false;
-                    }
-
-                    bpe_words.emplace_back(it->first); // push special token as token
-
-                    // we now advance i until the token is fulfilled by the utf_chars
-                    int st_bytes = (int)it->first.size();
-                    for (;st_bytes;st_bytes -= text_utf[i++].str.size());
-                    i--;
-                    special_token_found = true;
-                    break;
-                }
-            }
-
-            if (special_token_found)   continue;
-
-
             // handling contractions
             if (!split_condition && bytes_remain >= 2)
             {

From c0a1269b7f1fdd013dc23b831b9c688c7fe8340f Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 15:27:37 +0200
Subject: [PATCH 219/242] Update examples/server/README.md

Co-authored-by: slaren <slarengh@gmail.com>
---
 examples/server/README.md | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/examples/server/README.md b/examples/server/README.md
index ce2d2c9c69257..4d97db2e480eb 100644
--- a/examples/server/README.md
+++ b/examples/server/README.md
@@ -54,7 +54,8 @@ To get started right away, run the following command, making sure to use the cor
 ### Windows:
 
 ```powershell
-
+server.exe -m models\7B\ggml-model.gguf -c 2048
+```
 The above command will start a server that by default listens on `127.0.0.1:8080`.
 You can consume the endpoints with Postman or NodeJS with axios library. You can visit the web front end at the same url.
 

From 28b8c265eb3934396856ea1d1996a0b076a3d60a Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sat, 19 Aug 2023 18:26:51 +0200
Subject: [PATCH 220/242] cmpnct_gpt2bpe.hpp : cleanup

---
 examples/gptneox-wip/cmpnct_gpt2bpe.hpp | 573 ++++++++++++++++--------
 1 file changed, 390 insertions(+), 183 deletions(-)

diff --git a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
index 81ea443b2a20a..9d433f4b1acf0 100644
--- a/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
+++ b/examples/gptneox-wip/cmpnct_gpt2bpe.hpp
@@ -11,13 +11,15 @@
 #include <queue>
 #include <cstring>
 
-//-----
+
 // Unicode GPT2 Byte Pair Encoding Tokenizer
-// Adapted from https://github.com/cmp-nct/ggllm.cpp
+// Adapted from https://github.com/cmp-nct/ggllm.cpp [MIT License]
 // Removed loading of merges from HF json and parts made for a specific vocab
-//-----
 
+
+//-----------------
 // Unicode library (from cmpnct_unicode.cpp)
+//-----------------
 
 // Minimal library for high performance handling and categorization of UTF8 strings and characters
 // Using std::string
@@ -70,25 +72,221 @@ struct CNCTUnicode {
 };
 
 static const std::vector<std::pair<int, int>> digit_ranges = {
-{0x30, 0x39}, {0xB2, 0xB3}, {0xB9, 0xB9}, {0x660, 0x669}, {0x6F0, 0x6F9}, {0x7C0, 0x7C9}, {0x966, 0x96F}, {0x9E6, 0x9EF}, {0xA66, 0xA6F}, {0xAE6, 0xAEF}, {0xB66, 0xB6F}, {0xBE6, 0xBEF}, {0xC66, 0xC6F}, {0xCE6, 0xCEF}, {0xD66, 0xD6F}, {0xDE6, 0xDEF}, {0xE50, 0xE59}, {0xED0, 0xED9}, {0xF20, 0xF29}, {0x1040, 0x1049}, {0x1090, 0x1099}, {0x1369, 0x1371}, {0x17E0, 0x17E9}, {0x1810, 0x1819}, {0x1946, 0x194F}, {0x19D0, 0x19DA}, {0x1A80, 0x1A89}, {0x1A90, 0x1A99}, {0x1B50, 0x1B59}, {0x1BB0, 0x1BB9}, {0x1C40, 0x1C49}, {0x1C50, 0x1C59}, {0x2070, 0x2070}, {0x2074, 0x2079}, {0x2080, 0x2089}, {0x2460, 0x2468}, {0x2474, 0x247C}, {0x2488, 0x2490}, {0x24EA, 0x24EA}, {0x24F5, 0x24FD}, {0x24FF, 0x24FF}, {0x2776, 0x277E}, {0x2780, 0x2788}, {0x278A, 0x2792}, {0xA620, 0xA629}, {0xA8D0, 0xA8D9}, {0xA900, 0xA909}, {0xA9D0, 0xA9D9}, {0xA9F0, 0xA9F9}, {0xAA50, 0xAA59}, {0xABF0, 0xABF9}, {0xFF10, 0xFF19}, {0x104A0, 0x104A9}, {0x10A40, 0x10A43}, {0x10D30, 0x10D39}, {0x10E60, 0x10E68}, {0x11052, 0x1105A}, {0x11066, 0x1106F}, {0x110F0, 0x110F9}, {0x11136, 0x1113F}, {0x111D0, 0x111D9}, {0x112F0, 0x112F9}, {0x11450, 0x11459}, {0x114D0, 0x114D9}, {0x11650, 0x11659}, {0x116C0, 0x116C9}, {0x11730, 0x11739}, {0x118E0, 0x118E9}, {0x11950, 0x11959}, {0x11C50, 0x11C59}, {0x11D50, 0x11D59}, {0x11DA0, 0x11DA9}, {0x16A60, 0x16A69}, {0x16B50, 0x16B59}, {0x1D7CE, 0x1D7FF}, {0x1E140, 0x1E149}, {0x1E2F0, 0x1E2F9}, {0x1E950, 0x1E959}, {0x1F100, 0x1F10A}, {0x1FBF0, 0x1FBF9}, };
+{0x30, 0x39}, {0xB2, 0xB3}, {0xB9, 0xB9}, {0x660, 0x669}, {0x6F0, 0x6F9}, {0x7C0, 0x7C9}, {0x966, 0x96F}, {0x9E6, 0x9EF}, {0xA66, 0xA6F}, {0xAE6, 0xAEF}, {0xB66, 0xB6F}, {0xBE6, 0xBEF}, {0xC66, 0xC6F},
+{0xCE6, 0xCEF}, {0xD66, 0xD6F}, {0xDE6, 0xDEF}, {0xE50, 0xE59}, {0xED0, 0xED9}, {0xF20, 0xF29}, {0x1040, 0x1049}, {0x1090, 0x1099}, {0x1369, 0x1371}, {0x17E0, 0x17E9}, {0x1810, 0x1819}, {0x1946, 0x194F},
+{0x19D0, 0x19DA}, {0x1A80, 0x1A89}, {0x1A90, 0x1A99}, {0x1B50, 0x1B59}, {0x1BB0, 0x1BB9}, {0x1C40, 0x1C49}, {0x1C50, 0x1C59}, {0x2070, 0x2070}, {0x2074, 0x2079}, {0x2080, 0x2089}, {0x2460, 0x2468},
+{0x2474, 0x247C}, {0x2488, 0x2490}, {0x24EA, 0x24EA}, {0x24F5, 0x24FD}, {0x24FF, 0x24FF}, {0x2776, 0x277E}, {0x2780, 0x2788}, {0x278A, 0x2792}, {0xA620, 0xA629}, {0xA8D0, 0xA8D9}, {0xA900, 0xA909},
+{0xA9D0, 0xA9D9}, {0xA9F0, 0xA9F9}, {0xAA50, 0xAA59}, {0xABF0, 0xABF9}, {0xFF10, 0xFF19}, {0x104A0, 0x104A9}, {0x10A40, 0x10A43}, {0x10D30, 0x10D39}, {0x10E60, 0x10E68}, {0x11052, 0x1105A},
+{0x11066, 0x1106F}, {0x110F0, 0x110F9}, {0x11136, 0x1113F}, {0x111D0, 0x111D9}, {0x112F0, 0x112F9}, {0x11450, 0x11459}, {0x114D0, 0x114D9}, {0x11650, 0x11659}, {0x116C0, 0x116C9}, {0x11730, 0x11739},
+{0x118E0, 0x118E9}, {0x11950, 0x11959}, {0x11C50, 0x11C59}, {0x11D50, 0x11D59}, {0x11DA0, 0x11DA9}, {0x16A60, 0x16A69}, {0x16B50, 0x16B59}, {0x1D7CE, 0x1D7FF}, {0x1E140, 0x1E149}, {0x1E2F0, 0x1E2F9},
+{0x1E950, 0x1E959}, {0x1F100, 0x1F10A}, {0x1FBF0, 0x1FBF9},
+};
 
 static const std::vector<std::pair<int, int>> letter_ranges = {
-{0x41, 0x5A}, {0x61, 0x7A}, {0xAA, 0xAA}, {0xB5, 0xB5}, {0xBA, 0xBA}, {0xC0, 0xD6}, {0xD8, 0xF6}, {0xF8, 0x2C1}, {0x2C6, 0x2D1}, {0x2E0, 0x2E4}, {0x2EC, 0x2EC}, {0x2EE, 0x2EE}, {0x370, 0x374}, {0x376, 0x377}, {0x37A, 0x37D}, {0x37F, 0x37F}, {0x386, 0x386}, {0x388, 0x38A}, {0x38C, 0x38C}, {0x38E, 0x3A1}, {0x3A3, 0x3F5}, {0x3F7, 0x481}, {0x48A, 0x52F}, {0x531, 0x556}, {0x559, 0x559}, {0x560, 0x588}, {0x5D0, 0x5EA}, {0x5EF, 0x5F2}, {0x620, 0x64A}, {0x66E, 0x66F}, {0x671, 0x6D3}, {0x6D5, 0x6D5}, {0x6E5, 0x6E6}, {0x6EE, 0x6EF}, {0x6FA, 0x6FC}, {0x6FF, 0x6FF}, {0x710, 0x710}, {0x712, 0x72F}, {0x74D, 0x7A5}, {0x7B1, 0x7B1}, {0x7CA, 0x7EA}, {0x7F4, 0x7F5}, {0x7FA, 0x7FA}, {0x800, 0x815}, {0x81A, 0x81A}, {0x824, 0x824}, {0x828, 0x828}, {0x840, 0x858}, {0x860, 0x86A}, {0x8A0, 0x8B4}, {0x8B6, 0x8C7}, {0x904, 0x939}, {0x93D, 0x93D}, {0x950, 0x950}, {0x958, 0x961}, {0x971, 0x980}, {0x985, 0x98C}, {0x98F, 0x990}, {0x993, 0x9A8}, {0x9AA, 0x9B0}, {0x9B2, 0x9B2}, {0x9B6, 0x9B9}, {0x9BD, 0x9BD}, {0x9CE, 0x9CE}, {0x9DC, 0x9DD}, {0x9DF, 0x9E1}, {0x9F0, 0x9F1}, {0x9FC, 0x9FC}, {0xA05, 0xA0A}, {0xA0F, 0xA10}, {0xA13, 0xA28}, {0xA2A, 0xA30}, {0xA32, 0xA33}, {0xA35, 0xA36}, {0xA38, 0xA39}, {0xA59, 0xA5C}, {0xA5E, 0xA5E}, {0xA72, 0xA74}, {0xA85, 0xA8D}, {0xA8F, 0xA91}, {0xA93, 0xAA8}, {0xAAA, 0xAB0}, {0xAB2, 0xAB3}, {0xAB5, 0xAB9}, {0xABD, 0xABD}, {0xAD0, 0xAD0}, {0xAE0, 0xAE1}, {0xAF9, 0xAF9}, {0xB05, 0xB0C}, {0xB0F, 0xB10}, {0xB13, 0xB28}, {0xB2A, 0xB30}, {0xB32, 0xB33}, {0xB35, 0xB39}, {0xB3D, 0xB3D}, {0xB5C, 0xB5D}, {0xB5F, 0xB61}, {0xB71, 0xB71}, {0xB83, 0xB83}, {0xB85, 0xB8A}, {0xB8E, 0xB90}, {0xB92, 0xB95}, {0xB99, 0xB9A}, {0xB9C, 0xB9C}, {0xB9E, 0xB9F}, {0xBA3, 0xBA4}, {0xBA8, 0xBAA}, {0xBAE, 0xBB9}, {0xBD0, 0xBD0}, {0xC05, 0xC0C}, {0xC0E, 0xC10}, {0xC12, 0xC28}, {0xC2A, 0xC39}, {0xC3D, 0xC3D}, {0xC58, 0xC5A}, {0xC60, 0xC61}, {0xC80, 0xC80}, {0xC85, 0xC8C}, {0xC8E, 0xC90}, {0xC92, 0xCA8}, {0xCAA, 0xCB3}, {0xCB5, 0xCB9}, {0xCBD, 0xCBD}, {0xCDE, 0xCDE}, {0xCE0, 0xCE1}, {0xCF1, 0xCF2}, {0xD04, 0xD0C}, {0xD0E, 0xD10}, {0xD12, 0xD3A}, {0xD3D, 0xD3D}, {0xD4E, 0xD4E}, {0xD54, 0xD56}, {0xD5F, 0xD61}, {0xD7A, 0xD7F}, {0xD85, 0xD96}, {0xD9A, 0xDB1}, {0xDB3, 0xDBB}, {0xDBD, 0xDBD}, {0xDC0, 0xDC6}, {0xE01, 0xE30}, {0xE32, 0xE33}, {0xE40, 0xE46}, {0xE81, 0xE82}, {0xE84, 0xE84}, {0xE86, 0xE8A}, {0xE8C, 0xEA3}, {0xEA5, 0xEA5}, {0xEA7, 0xEB0}, {0xEB2, 0xEB3}, {0xEBD, 0xEBD}, {0xEC0, 0xEC4}, {0xEC6, 0xEC6}, {0xEDC, 0xEDF}, {0xF00, 0xF00}, {0xF40, 0xF47}, {0xF49, 0xF6C}, {0xF88, 0xF8C}, {0x1000, 0x102A}, {0x103F, 0x103F}, {0x1050, 0x1055}, {0x105A, 0x105D}, {0x1061, 0x1061}, {0x1065, 0x1066}, {0x106E, 0x1070}, {0x1075, 0x1081}, {0x108E, 0x108E}, {0x10A0, 0x10C5}, {0x10C7, 0x10C7}, {0x10CD, 0x10CD}, {0x10D0, 0x10FA}, {0x10FC, 0x1248}, {0x124A, 0x124D}, {0x1250, 0x1256}, {0x1258, 0x1258}, {0x125A, 0x125D}, {0x1260, 0x1288}, {0x128A, 0x128D}, {0x1290, 0x12B0}, {0x12B2, 0x12B5}, {0x12B8, 0x12BE}, {0x12C0, 0x12C0}, {0x12C2, 0x12C5}, {0x12C8, 0x12D6}, {0x12D8, 0x1310}, {0x1312, 0x1315}, {0x1318, 0x135A}, {0x1380, 0x138F}, {0x13A0, 0x13F5}, {0x13F8, 0x13FD}, {0x1401, 0x166C}, {0x166F, 0x167F}, {0x1681, 0x169A}, {0x16A0, 0x16EA}, {0x16F1, 0x16F8}, {0x1700, 0x170C}, {0x170E, 0x1711}, {0x1720, 0x1731}, {0x1740, 0x1751}, {0x1760, 0x176C}, {0x176E, 0x1770}, {0x1780, 0x17B3}, {0x17D7, 0x17D7}, {0x17DC, 0x17DC}, {0x1820, 0x1878}, {0x1880, 0x1884}, {0x1887, 0x18A8}, {0x18AA, 0x18AA}, {0x18B0, 0x18F5}, {0x1900, 0x191E}, {0x1950, 0x196D}, {0x1970, 0x1974}, {0x1980, 0x19AB}, {0x19B0, 0x19C9}, {0x1A00, 0x1A16}, {0x1A20, 0x1A54}, {0x1AA7, 0x1AA7}, {0x1B05, 0x1B33}, {0x1B45, 0x1B4B}, {0x1B83, 0x1BA0}, {0x1BAE, 0x1BAF}, {0x1BBA, 0x1BE5}, {0x1C00, 0x1C23}, {0x1C4D, 0x1C4F}, {0x1C5A, 0x1C7D}, {0x1C80, 0x1C88}, {0x1C90, 0x1CBA}, {0x1CBD, 0x1CBF}, {0x1CE9, 0x1CEC}, {0x1CEE, 0x1CF3}, {0x1CF5, 0x1CF6}, {0x1CFA, 0x1CFA}, {0x1D00, 0x1DBF}, {0x1E00, 0x1F15}, {0x1F18, 0x1F1D}, {0x1F20, 0x1F45}, {0x1F48, 0x1F4D}, {0x1F50, 0x1F57}, {0x1F59, 0x1F59}, {0x1F5B, 0x1F5B}, {0x1F5D, 0x1F5D}, {0x1F5F, 0x1F7D}, {0x1F80, 0x1FB4}, {0x1FB6, 0x1FBC}, {0x1FBE, 0x1FBE}, {0x1FC2, 0x1FC4}, {0x1FC6, 0x1FCC}, {0x1FD0, 0x1FD3}, {0x1FD6, 0x1FDB}, {0x1FE0, 0x1FEC}, {0x1FF2, 0x1FF4}, {0x1FF6, 0x1FFC}, {0x2071, 0x2071}, {0x207F, 0x207F}, {0x2090, 0x209C}, {0x2102, 0x2102}, {0x2107, 0x2107}, {0x210A, 0x2113}, {0x2115, 0x2115}, {0x2119, 0x211D}, {0x2124, 0x2124}, {0x2126, 0x2126}, {0x2128, 0x2128}, {0x212A, 0x212D}, {0x212F, 0x2139}, {0x213C, 0x213F}, {0x2145, 0x2149}, {0x214E, 0x214E}, {0x2183, 0x2184}, {0x2C00, 0x2C2E}, {0x2C30, 0x2C5E}, {0x2C60, 0x2CE4}, {0x2CEB, 0x2CEE}, {0x2CF2, 0x2CF3}, {0x2D00, 0x2D25}, {0x2D27, 0x2D27}, {0x2D2D, 0x2D2D}, {0x2D30, 0x2D67}, {0x2D6F, 0x2D6F}, {0x2D80, 0x2D96}, {0x2DA0, 0x2DA6}, {0x2DA8, 0x2DAE}, {0x2DB0, 0x2DB6}, {0x2DB8, 0x2DBE}, {0x2DC0, 0x2DC6}, {0x2DC8, 0x2DCE}, {0x2DD0, 0x2DD6}, {0x2DD8, 0x2DDE}, {0x2E2F, 0x2E2F}, {0x3005, 0x3006}, {0x3031, 0x3035}, {0x303B, 0x303C}, {0x3041, 0x3096}, {0x309D, 0x309F}, {0x30A1, 0x30FA}, {0x30FC, 0x30FF}, {0x3105, 0x312F}, {0x3131, 0x318E}, {0x31A0, 0x31BF}, {0x31F0, 0x31FF}, {0x3400, 0x4DBF}, {0x4E00, 0x9FFC}, {0xA000, 0xA48C}, {0xA4D0, 0xA4FD}, {0xA500, 0xA60C}, {0xA610, 0xA61F}, {0xA62A, 0xA62B}, {0xA640, 0xA66E}, {0xA67F, 0xA69D}, {0xA6A0, 0xA6E5}, {0xA717, 0xA71F}, {0xA722, 0xA788}, {0xA78B, 0xA7BF}, {0xA7C2, 0xA7CA}, {0xA7F5, 0xA801}, {0xA803, 0xA805}, {0xA807, 0xA80A}, {0xA80C, 0xA822}, {0xA840, 0xA873}, {0xA882, 0xA8B3}, {0xA8F2, 0xA8F7}, {0xA8FB, 0xA8FB}, {0xA8FD, 0xA8FE}, {0xA90A, 0xA925}, {0xA930, 0xA946}, {0xA960, 0xA97C}, {0xA984, 0xA9B2}, {0xA9CF, 0xA9CF}, {0xA9E0, 0xA9E4}, {0xA9E6, 0xA9EF}, {0xA9FA, 0xA9FE}, {0xAA00, 0xAA28}, {0xAA40, 0xAA42}, {0xAA44, 0xAA4B}, {0xAA60, 0xAA76}, {0xAA7A, 0xAA7A}, {0xAA7E, 0xAAAF}, {0xAAB1, 0xAAB1}, {0xAAB5, 0xAAB6}, {0xAAB9, 0xAABD}, {0xAAC0, 0xAAC0}, {0xAAC2, 0xAAC2}, {0xAADB, 0xAADD}, {0xAAE0, 0xAAEA}, {0xAAF2, 0xAAF4}, {0xAB01, 0xAB06}, {0xAB09, 0xAB0E}, {0xAB11, 0xAB16}, {0xAB20, 0xAB26}, {0xAB28, 0xAB2E}, {0xAB30, 0xAB5A}, {0xAB5C, 0xAB69}, {0xAB70, 0xABE2}, {0xAC00, 0xD7A3}, {0xD7B0, 0xD7C6}, {0xD7CB, 0xD7FB}, {0xF900, 0xFA6D}, {0xFA70, 0xFAD9}, {0xFB00, 0xFB06}, {0xFB13, 0xFB17}, {0xFB1D, 0xFB1D}, {0xFB1F, 0xFB28}, {0xFB2A, 0xFB36}, {0xFB38, 0xFB3C}, {0xFB3E, 0xFB3E}, {0xFB40, 0xFB41}, {0xFB43, 0xFB44}, {0xFB46, 0xFBB1}, {0xFBD3, 0xFD3D}, {0xFD50, 0xFD8F}, {0xFD92, 0xFDC7}, {0xFDF0, 0xFDFB}, {0xFE70, 0xFE74}, {0xFE76, 0xFEFC}, {0xFF21, 0xFF3A}, {0xFF41, 0xFF5A}, {0xFF66, 0xFFBE}, {0xFFC2, 0xFFC7}, {0xFFCA, 0xFFCF}, {0xFFD2, 0xFFD7}, {0xFFDA, 0xFFDC}, {0x10000, 0x1000B}, {0x1000D, 0x10026}, {0x10028, 0x1003A}, {0x1003C, 0x1003D}, {0x1003F, 0x1004D}, {0x10050, 0x1005D}, {0x10080, 0x100FA}, {0x10280, 0x1029C}, {0x102A0, 0x102D0}, {0x10300, 0x1031F}, {0x1032D, 0x10340}, {0x10342, 0x10349}, {0x10350, 0x10375}, {0x10380, 0x1039D}, {0x103A0, 0x103C3}, {0x103C8, 0x103CF}, {0x10400, 0x1049D}, {0x104B0, 0x104D3}, {0x104D8, 0x104FB}, {0x10500, 0x10527}, {0x10530, 0x10563}, {0x10600, 0x10736}, {0x10740, 0x10755}, {0x10760, 0x10767}, {0x10800, 0x10805}, {0x10808, 0x10808}, {0x1080A, 0x10835}, {0x10837, 0x10838}, {0x1083C, 0x1083C}, {0x1083F, 0x10855}, {0x10860, 0x10876}, {0x10880, 0x1089E}, {0x108E0, 0x108F2}, {0x108F4, 0x108F5}, {0x10900, 0x10915}, {0x10920, 0x10939}, {0x10980, 0x109B7}, {0x109BE, 0x109BF}, {0x10A00, 0x10A00}, {0x10A10, 0x10A13}, {0x10A15, 0x10A17}, {0x10A19, 0x10A35}, {0x10A60, 0x10A7C}, {0x10A80, 0x10A9C}, {0x10AC0, 0x10AC7}, {0x10AC9, 0x10AE4}, {0x10B00, 0x10B35}, {0x10B40, 0x10B55}, {0x10B60, 0x10B72}, {0x10B80, 0x10B91}, {0x10C00, 0x10C48}, {0x10C80, 0x10CB2}, {0x10CC0, 0x10CF2}, {0x10D00, 0x10D23}, {0x10E80, 0x10EA9}, {0x10EB0, 0x10EB1}, {0x10F00, 0x10F1C}, {0x10F27, 0x10F27}, {0x10F30, 0x10F45}, {0x10FB0, 0x10FC4}, {0x10FE0, 0x10FF6}, {0x11003, 0x11037}, {0x11083, 0x110AF}, {0x110D0, 0x110E8}, {0x11103, 0x11126}, {0x11144, 0x11144}, {0x11147, 0x11147}, {0x11150, 0x11172}, {0x11176, 0x11176}, {0x11183, 0x111B2}, {0x111C1, 0x111C4}, {0x111DA, 0x111DA}, {0x111DC, 0x111DC}, {0x11200, 0x11211}, {0x11213, 0x1122B}, {0x11280, 0x11286}, {0x11288, 0x11288}, {0x1128A, 0x1128D}, {0x1128F, 0x1129D}, {0x1129F, 0x112A8}, {0x112B0, 0x112DE}, {0x11305, 0x1130C}, {0x1130F, 0x11310}, {0x11313, 0x11328}, {0x1132A, 0x11330}, {0x11332, 0x11333}, {0x11335, 0x11339}, {0x1133D, 0x1133D}, {0x11350, 0x11350}, {0x1135D, 0x11361}, {0x11400, 0x11434}, {0x11447, 0x1144A}, {0x1145F, 0x11461}, {0x11480, 0x114AF}, {0x114C4, 0x114C5}, {0x114C7, 0x114C7}, {0x11580, 0x115AE}, {0x115D8, 0x115DB}, {0x11600, 0x1162F}, {0x11644, 0x11644}, {0x11680, 0x116AA}, {0x116B8, 0x116B8}, {0x11700, 0x1171A}, {0x11800, 0x1182B}, {0x118A0, 0x118DF}, {0x118FF, 0x11906}, {0x11909, 0x11909}, {0x1190C, 0x11913}, {0x11915, 0x11916}, {0x11918, 0x1192F}, {0x1193F, 0x1193F}, {0x11941, 0x11941}, {0x119A0, 0x119A7}, {0x119AA, 0x119D0}, {0x119E1, 0x119E1}, {0x119E3, 0x119E3}, {0x11A00, 0x11A00}, {0x11A0B, 0x11A32}, {0x11A3A, 0x11A3A}, {0x11A50, 0x11A50}, {0x11A5C, 0x11A89}, {0x11A9D, 0x11A9D}, {0x11AC0, 0x11AF8}, {0x11C00, 0x11C08}, {0x11C0A, 0x11C2E}, {0x11C40, 0x11C40}, {0x11C72, 0x11C8F}, {0x11D00, 0x11D06}, {0x11D08, 0x11D09}, {0x11D0B, 0x11D30}, {0x11D46, 0x11D46}, {0x11D60, 0x11D65}, {0x11D67, 0x11D68}, {0x11D6A, 0x11D89}, {0x11D98, 0x11D98}, {0x11EE0, 0x11EF2}, {0x11FB0, 0x11FB0}, {0x12000, 0x12399}, {0x12480, 0x12543}, {0x13000, 0x1342E}, {0x14400, 0x14646}, {0x16800, 0x16A38}, {0x16A40, 0x16A5E}, {0x16AD0, 0x16AED}, {0x16B00, 0x16B2F}, {0x16B40, 0x16B43}, {0x16B63, 0x16B77}, {0x16B7D, 0x16B8F}, {0x16E40, 0x16E7F}, {0x16F00, 0x16F4A}, {0x16F50, 0x16F50}, {0x16F93, 0x16F9F}, {0x16FE0, 0x16FE1}, {0x16FE3, 0x16FE3}, {0x17000, 0x187F7}, {0x18800, 0x18CD5}, {0x18D00, 0x18D08}, {0x1B000, 0x1B11E}, {0x1B150, 0x1B152}, {0x1B164, 0x1B167}, {0x1B170, 0x1B2FB}, {0x1BC00, 0x1BC6A}, {0x1BC70, 0x1BC7C}, {0x1BC80, 0x1BC88}, {0x1BC90, 0x1BC99}, {0x1D400, 0x1D454}, {0x1D456, 0x1D49C}, {0x1D49E, 0x1D49F}, {0x1D4A2, 0x1D4A2}, {0x1D4A5, 0x1D4A6}, {0x1D4A9, 0x1D4AC}, {0x1D4AE, 0x1D4B9}, {0x1D4BB, 0x1D4BB}, {0x1D4BD, 0x1D4C3}, {0x1D4C5, 0x1D505}, {0x1D507, 0x1D50A}, {0x1D50D, 0x1D514}, {0x1D516, 0x1D51C}, {0x1D51E, 0x1D539}, {0x1D53B, 0x1D53E}, {0x1D540, 0x1D544}, {0x1D546, 0x1D546}, {0x1D54A, 0x1D550}, {0x1D552, 0x1D6A5}, {0x1D6A8, 0x1D6C0}, {0x1D6C2, 0x1D6DA}, {0x1D6DC, 0x1D6FA}, {0x1D6FC, 0x1D714}, {0x1D716, 0x1D734}, {0x1D736, 0x1D74E}, {0x1D750, 0x1D76E}, {0x1D770, 0x1D788}, {0x1D78A, 0x1D7A8}, {0x1D7AA, 0x1D7C2}, {0x1D7C4, 0x1D7CB}, {0x1E100, 0x1E12C}, {0x1E137, 0x1E13D}, {0x1E14E, 0x1E14E}, {0x1E2C0, 0x1E2EB}, {0x1E800, 0x1E8C4}, {0x1E900, 0x1E943}, {0x1E94B, 0x1E94B}, {0x1EE00, 0x1EE03}, {0x1EE05, 0x1EE1F}, {0x1EE21, 0x1EE22}, {0x1EE24, 0x1EE24}, {0x1EE27, 0x1EE27}, {0x1EE29, 0x1EE32}, {0x1EE34, 0x1EE37}, {0x1EE39, 0x1EE39}, {0x1EE3B, 0x1EE3B}, {0x1EE42, 0x1EE42}, {0x1EE47, 0x1EE47}, {0x1EE49, 0x1EE49}, {0x1EE4B, 0x1EE4B}, {0x1EE4D, 0x1EE4F}, {0x1EE51, 0x1EE52}, {0x1EE54, 0x1EE54}, {0x1EE57, 0x1EE57}, {0x1EE59, 0x1EE59}, {0x1EE5B, 0x1EE5B}, {0x1EE5D, 0x1EE5D}, {0x1EE5F, 0x1EE5F}, {0x1EE61, 0x1EE62}, {0x1EE64, 0x1EE64}, {0x1EE67, 0x1EE6A}, {0x1EE6C, 0x1EE72}, {0x1EE74, 0x1EE77}, {0x1EE79, 0x1EE7C}, {0x1EE7E, 0x1EE7E}, {0x1EE80, 0x1EE89}, {0x1EE8B, 0x1EE9B}, {0x1EEA1, 0x1EEA3}, {0x1EEA5, 0x1EEA9}, {0x1EEAB, 0x1EEBB}, {0x20000, 0x2A6DD}, {0x2A700, 0x2B734}, {0x2B740, 0x2B81D}, {0x2B820, 0x2CEA1}, {0x2CEB0, 0x2EBE0}, {0x2F800, 0x2FA1D}, {0x30000, 0x3134A}, };
+{0x41, 0x5A}, {0x61, 0x7A}, {0xAA, 0xAA}, {0xB5, 0xB5}, {0xBA, 0xBA}, {0xC0, 0xD6}, {0xD8, 0xF6}, {0xF8, 0x2C1}, {0x2C6, 0x2D1}, {0x2E0, 0x2E4}, {0x2EC, 0x2EC}, {0x2EE, 0x2EE}, {0x370, 0x374},
+{0x376, 0x377}, {0x37A, 0x37D}, {0x37F, 0x37F}, {0x386, 0x386}, {0x388, 0x38A}, {0x38C, 0x38C}, {0x38E, 0x3A1}, {0x3A3, 0x3F5}, {0x3F7, 0x481}, {0x48A, 0x52F}, {0x531, 0x556}, {0x559, 0x559},
+{0x560, 0x588}, {0x5D0, 0x5EA}, {0x5EF, 0x5F2}, {0x620, 0x64A}, {0x66E, 0x66F}, {0x671, 0x6D3}, {0x6D5, 0x6D5}, {0x6E5, 0x6E6}, {0x6EE, 0x6EF}, {0x6FA, 0x6FC}, {0x6FF, 0x6FF}, {0x710, 0x710},
+{0x712, 0x72F}, {0x74D, 0x7A5}, {0x7B1, 0x7B1}, {0x7CA, 0x7EA}, {0x7F4, 0x7F5}, {0x7FA, 0x7FA}, {0x800, 0x815}, {0x81A, 0x81A}, {0x824, 0x824}, {0x828, 0x828}, {0x840, 0x858}, {0x860, 0x86A},
+{0x8A0, 0x8B4}, {0x8B6, 0x8C7}, {0x904, 0x939}, {0x93D, 0x93D}, {0x950, 0x950}, {0x958, 0x961}, {0x971, 0x980}, {0x985, 0x98C}, {0x98F, 0x990}, {0x993, 0x9A8}, {0x9AA, 0x9B0}, {0x9B2, 0x9B2},
+{0x9B6, 0x9B9}, {0x9BD, 0x9BD}, {0x9CE, 0x9CE}, {0x9DC, 0x9DD}, {0x9DF, 0x9E1}, {0x9F0, 0x9F1}, {0x9FC, 0x9FC}, {0xA05, 0xA0A}, {0xA0F, 0xA10}, {0xA13, 0xA28}, {0xA2A, 0xA30}, {0xA32, 0xA33},
+{0xA35, 0xA36}, {0xA38, 0xA39}, {0xA59, 0xA5C}, {0xA5E, 0xA5E}, {0xA72, 0xA74}, {0xA85, 0xA8D}, {0xA8F, 0xA91}, {0xA93, 0xAA8}, {0xAAA, 0xAB0}, {0xAB2, 0xAB3}, {0xAB5, 0xAB9}, {0xABD, 0xABD},
+{0xAD0, 0xAD0}, {0xAE0, 0xAE1}, {0xAF9, 0xAF9}, {0xB05, 0xB0C}, {0xB0F, 0xB10}, {0xB13, 0xB28}, {0xB2A, 0xB30}, {0xB32, 0xB33}, {0xB35, 0xB39}, {0xB3D, 0xB3D}, {0xB5C, 0xB5D}, {0xB5F, 0xB61},
+{0xB71, 0xB71}, {0xB83, 0xB83}, {0xB85, 0xB8A}, {0xB8E, 0xB90}, {0xB92, 0xB95}, {0xB99, 0xB9A}, {0xB9C, 0xB9C}, {0xB9E, 0xB9F}, {0xBA3, 0xBA4}, {0xBA8, 0xBAA}, {0xBAE, 0xBB9}, {0xBD0, 0xBD0},
+{0xC05, 0xC0C}, {0xC0E, 0xC10}, {0xC12, 0xC28}, {0xC2A, 0xC39}, {0xC3D, 0xC3D}, {0xC58, 0xC5A}, {0xC60, 0xC61}, {0xC80, 0xC80}, {0xC85, 0xC8C}, {0xC8E, 0xC90}, {0xC92, 0xCA8}, {0xCAA, 0xCB3},
+{0xCB5, 0xCB9}, {0xCBD, 0xCBD}, {0xCDE, 0xCDE}, {0xCE0, 0xCE1}, {0xCF1, 0xCF2}, {0xD04, 0xD0C}, {0xD0E, 0xD10}, {0xD12, 0xD3A}, {0xD3D, 0xD3D}, {0xD4E, 0xD4E}, {0xD54, 0xD56}, {0xD5F, 0xD61},
+{0xD7A, 0xD7F}, {0xD85, 0xD96}, {0xD9A, 0xDB1}, {0xDB3, 0xDBB}, {0xDBD, 0xDBD}, {0xDC0, 0xDC6}, {0xE01, 0xE30}, {0xE32, 0xE33}, {0xE40, 0xE46}, {0xE81, 0xE82}, {0xE84, 0xE84}, {0xE86, 0xE8A},
+{0xE8C, 0xEA3}, {0xEA5, 0xEA5}, {0xEA7, 0xEB0}, {0xEB2, 0xEB3}, {0xEBD, 0xEBD}, {0xEC0, 0xEC4}, {0xEC6, 0xEC6}, {0xEDC, 0xEDF}, {0xF00, 0xF00}, {0xF40, 0xF47}, {0xF49, 0xF6C}, {0xF88, 0xF8C},
+{0x1000, 0x102A}, {0x103F, 0x103F}, {0x1050, 0x1055}, {0x105A, 0x105D}, {0x1061, 0x1061}, {0x1065, 0x1066}, {0x106E, 0x1070}, {0x1075, 0x1081}, {0x108E, 0x108E}, {0x10A0, 0x10C5}, {0x10C7, 0x10C7},
+{0x10CD, 0x10CD}, {0x10D0, 0x10FA}, {0x10FC, 0x1248}, {0x124A, 0x124D}, {0x1250, 0x1256}, {0x1258, 0x1258}, {0x125A, 0x125D}, {0x1260, 0x1288}, {0x128A, 0x128D}, {0x1290, 0x12B0}, {0x12B2, 0x12B5},
+{0x12B8, 0x12BE}, {0x12C0, 0x12C0}, {0x12C2, 0x12C5}, {0x12C8, 0x12D6}, {0x12D8, 0x1310}, {0x1312, 0x1315}, {0x1318, 0x135A}, {0x1380, 0x138F}, {0x13A0, 0x13F5}, {0x13F8, 0x13FD}, {0x1401, 0x166C},
+{0x166F, 0x167F}, {0x1681, 0x169A}, {0x16A0, 0x16EA}, {0x16F1, 0x16F8}, {0x1700, 0x170C}, {0x170E, 0x1711}, {0x1720, 0x1731}, {0x1740, 0x1751}, {0x1760, 0x176C}, {0x176E, 0x1770}, {0x1780, 0x17B3},
+{0x17D7, 0x17D7}, {0x17DC, 0x17DC}, {0x1820, 0x1878}, {0x1880, 0x1884}, {0x1887, 0x18A8}, {0x18AA, 0x18AA}, {0x18B0, 0x18F5}, {0x1900, 0x191E}, {0x1950, 0x196D}, {0x1970, 0x1974}, {0x1980, 0x19AB},
+{0x19B0, 0x19C9}, {0x1A00, 0x1A16}, {0x1A20, 0x1A54}, {0x1AA7, 0x1AA7}, {0x1B05, 0x1B33}, {0x1B45, 0x1B4B}, {0x1B83, 0x1BA0}, {0x1BAE, 0x1BAF}, {0x1BBA, 0x1BE5}, {0x1C00, 0x1C23}, {0x1C4D, 0x1C4F},
+{0x1C5A, 0x1C7D}, {0x1C80, 0x1C88}, {0x1C90, 0x1CBA}, {0x1CBD, 0x1CBF}, {0x1CE9, 0x1CEC}, {0x1CEE, 0x1CF3}, {0x1CF5, 0x1CF6}, {0x1CFA, 0x1CFA}, {0x1D00, 0x1DBF}, {0x1E00, 0x1F15}, {0x1F18, 0x1F1D},
+{0x1F20, 0x1F45}, {0x1F48, 0x1F4D}, {0x1F50, 0x1F57}, {0x1F59, 0x1F59}, {0x1F5B, 0x1F5B}, {0x1F5D, 0x1F5D}, {0x1F5F, 0x1F7D}, {0x1F80, 0x1FB4}, {0x1FB6, 0x1FBC}, {0x1FBE, 0x1FBE}, {0x1FC2, 0x1FC4},
+{0x1FC6, 0x1FCC}, {0x1FD0, 0x1FD3}, {0x1FD6, 0x1FDB}, {0x1FE0, 0x1FEC}, {0x1FF2, 0x1FF4}, {0x1FF6, 0x1FFC}, {0x2071, 0x2071}, {0x207F, 0x207F}, {0x2090, 0x209C}, {0x2102, 0x2102}, {0x2107, 0x2107},
+{0x210A, 0x2113}, {0x2115, 0x2115}, {0x2119, 0x211D}, {0x2124, 0x2124}, {0x2126, 0x2126}, {0x2128, 0x2128}, {0x212A, 0x212D}, {0x212F, 0x2139}, {0x213C, 0x213F}, {0x2145, 0x2149}, {0x214E, 0x214E},
+{0x2183, 0x2184}, {0x2C00, 0x2C2E}, {0x2C30, 0x2C5E}, {0x2C60, 0x2CE4}, {0x2CEB, 0x2CEE}, {0x2CF2, 0x2CF3}, {0x2D00, 0x2D25}, {0x2D27, 0x2D27}, {0x2D2D, 0x2D2D}, {0x2D30, 0x2D67}, {0x2D6F, 0x2D6F},
+{0x2D80, 0x2D96}, {0x2DA0, 0x2DA6}, {0x2DA8, 0x2DAE}, {0x2DB0, 0x2DB6}, {0x2DB8, 0x2DBE}, {0x2DC0, 0x2DC6}, {0x2DC8, 0x2DCE}, {0x2DD0, 0x2DD6}, {0x2DD8, 0x2DDE}, {0x2E2F, 0x2E2F}, {0x3005, 0x3006},
+{0x3031, 0x3035}, {0x303B, 0x303C}, {0x3041, 0x3096}, {0x309D, 0x309F}, {0x30A1, 0x30FA}, {0x30FC, 0x30FF}, {0x3105, 0x312F}, {0x3131, 0x318E}, {0x31A0, 0x31BF}, {0x31F0, 0x31FF}, {0x3400, 0x4DBF},
+{0x4E00, 0x9FFC}, {0xA000, 0xA48C}, {0xA4D0, 0xA4FD}, {0xA500, 0xA60C}, {0xA610, 0xA61F}, {0xA62A, 0xA62B}, {0xA640, 0xA66E}, {0xA67F, 0xA69D}, {0xA6A0, 0xA6E5}, {0xA717, 0xA71F}, {0xA722, 0xA788},
+{0xA78B, 0xA7BF}, {0xA7C2, 0xA7CA}, {0xA7F5, 0xA801}, {0xA803, 0xA805}, {0xA807, 0xA80A}, {0xA80C, 0xA822}, {0xA840, 0xA873}, {0xA882, 0xA8B3}, {0xA8F2, 0xA8F7}, {0xA8FB, 0xA8FB}, {0xA8FD, 0xA8FE},
+{0xA90A, 0xA925}, {0xA930, 0xA946}, {0xA960, 0xA97C}, {0xA984, 0xA9B2}, {0xA9CF, 0xA9CF}, {0xA9E0, 0xA9E4}, {0xA9E6, 0xA9EF}, {0xA9FA, 0xA9FE}, {0xAA00, 0xAA28}, {0xAA40, 0xAA42}, {0xAA44, 0xAA4B},
+{0xAA60, 0xAA76}, {0xAA7A, 0xAA7A}, {0xAA7E, 0xAAAF}, {0xAAB1, 0xAAB1}, {0xAAB5, 0xAAB6}, {0xAAB9, 0xAABD}, {0xAAC0, 0xAAC0}, {0xAAC2, 0xAAC2}, {0xAADB, 0xAADD}, {0xAAE0, 0xAAEA}, {0xAAF2, 0xAAF4},
+{0xAB01, 0xAB06}, {0xAB09, 0xAB0E}, {0xAB11, 0xAB16}, {0xAB20, 0xAB26}, {0xAB28, 0xAB2E}, {0xAB30, 0xAB5A}, {0xAB5C, 0xAB69}, {0xAB70, 0xABE2}, {0xAC00, 0xD7A3}, {0xD7B0, 0xD7C6}, {0xD7CB, 0xD7FB},
+{0xF900, 0xFA6D}, {0xFA70, 0xFAD9}, {0xFB00, 0xFB06}, {0xFB13, 0xFB17}, {0xFB1D, 0xFB1D}, {0xFB1F, 0xFB28}, {0xFB2A, 0xFB36}, {0xFB38, 0xFB3C}, {0xFB3E, 0xFB3E}, {0xFB40, 0xFB41}, {0xFB43, 0xFB44},
+{0xFB46, 0xFBB1}, {0xFBD3, 0xFD3D}, {0xFD50, 0xFD8F}, {0xFD92, 0xFDC7}, {0xFDF0, 0xFDFB}, {0xFE70, 0xFE74}, {0xFE76, 0xFEFC}, {0xFF21, 0xFF3A}, {0xFF41, 0xFF5A}, {0xFF66, 0xFFBE}, {0xFFC2, 0xFFC7},
+{0xFFCA, 0xFFCF}, {0xFFD2, 0xFFD7}, {0xFFDA, 0xFFDC}, {0x10000, 0x1000B}, {0x1000D, 0x10026}, {0x10028, 0x1003A}, {0x1003C, 0x1003D}, {0x1003F, 0x1004D}, {0x10050, 0x1005D}, {0x10080, 0x100FA},
+{0x10280, 0x1029C}, {0x102A0, 0x102D0}, {0x10300, 0x1031F}, {0x1032D, 0x10340}, {0x10342, 0x10349}, {0x10350, 0x10375}, {0x10380, 0x1039D}, {0x103A0, 0x103C3}, {0x103C8, 0x103CF}, {0x10400, 0x1049D},
+{0x104B0, 0x104D3}, {0x104D8, 0x104FB}, {0x10500, 0x10527}, {0x10530, 0x10563}, {0x10600, 0x10736}, {0x10740, 0x10755}, {0x10760, 0x10767}, {0x10800, 0x10805}, {0x10808, 0x10808}, {0x1080A, 0x10835},
+{0x10837, 0x10838}, {0x1083C, 0x1083C}, {0x1083F, 0x10855}, {0x10860, 0x10876}, {0x10880, 0x1089E}, {0x108E0, 0x108F2}, {0x108F4, 0x108F5}, {0x10900, 0x10915}, {0x10920, 0x10939}, {0x10980, 0x109B7},
+{0x109BE, 0x109BF}, {0x10A00, 0x10A00}, {0x10A10, 0x10A13}, {0x10A15, 0x10A17}, {0x10A19, 0x10A35}, {0x10A60, 0x10A7C}, {0x10A80, 0x10A9C}, {0x10AC0, 0x10AC7}, {0x10AC9, 0x10AE4}, {0x10B00, 0x10B35},
+{0x10B40, 0x10B55}, {0x10B60, 0x10B72}, {0x10B80, 0x10B91}, {0x10C00, 0x10C48}, {0x10C80, 0x10CB2}, {0x10CC0, 0x10CF2}, {0x10D00, 0x10D23}, {0x10E80, 0x10EA9}, {0x10EB0, 0x10EB1}, {0x10F00, 0x10F1C},
+{0x10F27, 0x10F27}, {0x10F30, 0x10F45}, {0x10FB0, 0x10FC4}, {0x10FE0, 0x10FF6}, {0x11003, 0x11037}, {0x11083, 0x110AF}, {0x110D0, 0x110E8}, {0x11103, 0x11126}, {0x11144, 0x11144}, {0x11147, 0x11147},
+{0x11150, 0x11172}, {0x11176, 0x11176}, {0x11183, 0x111B2}, {0x111C1, 0x111C4}, {0x111DA, 0x111DA}, {0x111DC, 0x111DC}, {0x11200, 0x11211}, {0x11213, 0x1122B}, {0x11280, 0x11286}, {0x11288, 0x11288},
+{0x1128A, 0x1128D}, {0x1128F, 0x1129D}, {0x1129F, 0x112A8}, {0x112B0, 0x112DE}, {0x11305, 0x1130C}, {0x1130F, 0x11310}, {0x11313, 0x11328}, {0x1132A, 0x11330}, {0x11332, 0x11333}, {0x11335, 0x11339},
+{0x1133D, 0x1133D}, {0x11350, 0x11350}, {0x1135D, 0x11361}, {0x11400, 0x11434}, {0x11447, 0x1144A}, {0x1145F, 0x11461}, {0x11480, 0x114AF}, {0x114C4, 0x114C5}, {0x114C7, 0x114C7}, {0x11580, 0x115AE},
+{0x115D8, 0x115DB}, {0x11600, 0x1162F}, {0x11644, 0x11644}, {0x11680, 0x116AA}, {0x116B8, 0x116B8}, {0x11700, 0x1171A}, {0x11800, 0x1182B}, {0x118A0, 0x118DF}, {0x118FF, 0x11906}, {0x11909, 0x11909},
+{0x1190C, 0x11913}, {0x11915, 0x11916}, {0x11918, 0x1192F}, {0x1193F, 0x1193F}, {0x11941, 0x11941}, {0x119A0, 0x119A7}, {0x119AA, 0x119D0}, {0x119E1, 0x119E1}, {0x119E3, 0x119E3}, {0x11A00, 0x11A00},
+{0x11A0B, 0x11A32}, {0x11A3A, 0x11A3A}, {0x11A50, 0x11A50}, {0x11A5C, 0x11A89}, {0x11A9D, 0x11A9D}, {0x11AC0, 0x11AF8}, {0x11C00, 0x11C08}, {0x11C0A, 0x11C2E}, {0x11C40, 0x11C40}, {0x11C72, 0x11C8F},
+{0x11D00, 0x11D06}, {0x11D08, 0x11D09}, {0x11D0B, 0x11D30}, {0x11D46, 0x11D46}, {0x11D60, 0x11D65}, {0x11D67, 0x11D68}, {0x11D6A, 0x11D89}, {0x11D98, 0x11D98}, {0x11EE0, 0x11EF2}, {0x11FB0, 0x11FB0},
+{0x12000, 0x12399}, {0x12480, 0x12543}, {0x13000, 0x1342E}, {0x14400, 0x14646}, {0x16800, 0x16A38}, {0x16A40, 0x16A5E}, {0x16AD0, 0x16AED}, {0x16B00, 0x16B2F}, {0x16B40, 0x16B43}, {0x16B63, 0x16B77},
+{0x16B7D, 0x16B8F}, {0x16E40, 0x16E7F}, {0x16F00, 0x16F4A}, {0x16F50, 0x16F50}, {0x16F93, 0x16F9F}, {0x16FE0, 0x16FE1}, {0x16FE3, 0x16FE3}, {0x17000, 0x187F7}, {0x18800, 0x18CD5}, {0x18D00, 0x18D08},
+{0x1B000, 0x1B11E}, {0x1B150, 0x1B152}, {0x1B164, 0x1B167}, {0x1B170, 0x1B2FB}, {0x1BC00, 0x1BC6A}, {0x1BC70, 0x1BC7C}, {0x1BC80, 0x1BC88}, {0x1BC90, 0x1BC99}, {0x1D400, 0x1D454}, {0x1D456, 0x1D49C},
+{0x1D49E, 0x1D49F}, {0x1D4A2, 0x1D4A2}, {0x1D4A5, 0x1D4A6}, {0x1D4A9, 0x1D4AC}, {0x1D4AE, 0x1D4B9}, {0x1D4BB, 0x1D4BB}, {0x1D4BD, 0x1D4C3}, {0x1D4C5, 0x1D505}, {0x1D507, 0x1D50A}, {0x1D50D, 0x1D514},
+{0x1D516, 0x1D51C}, {0x1D51E, 0x1D539}, {0x1D53B, 0x1D53E}, {0x1D540, 0x1D544}, {0x1D546, 0x1D546}, {0x1D54A, 0x1D550}, {0x1D552, 0x1D6A5}, {0x1D6A8, 0x1D6C0}, {0x1D6C2, 0x1D6DA}, {0x1D6DC, 0x1D6FA},
+{0x1D6FC, 0x1D714}, {0x1D716, 0x1D734}, {0x1D736, 0x1D74E}, {0x1D750, 0x1D76E}, {0x1D770, 0x1D788}, {0x1D78A, 0x1D7A8}, {0x1D7AA, 0x1D7C2}, {0x1D7C4, 0x1D7CB}, {0x1E100, 0x1E12C}, {0x1E137, 0x1E13D},
+{0x1E14E, 0x1E14E}, {0x1E2C0, 0x1E2EB}, {0x1E800, 0x1E8C4}, {0x1E900, 0x1E943}, {0x1E94B, 0x1E94B}, {0x1EE00, 0x1EE03}, {0x1EE05, 0x1EE1F}, {0x1EE21, 0x1EE22}, {0x1EE24, 0x1EE24}, {0x1EE27, 0x1EE27},
+{0x1EE29, 0x1EE32}, {0x1EE34, 0x1EE37}, {0x1EE39, 0x1EE39}, {0x1EE3B, 0x1EE3B}, {0x1EE42, 0x1EE42}, {0x1EE47, 0x1EE47}, {0x1EE49, 0x1EE49}, {0x1EE4B, 0x1EE4B}, {0x1EE4D, 0x1EE4F}, {0x1EE51, 0x1EE52},
+{0x1EE54, 0x1EE54}, {0x1EE57, 0x1EE57}, {0x1EE59, 0x1EE59}, {0x1EE5B, 0x1EE5B}, {0x1EE5D, 0x1EE5D}, {0x1EE5F, 0x1EE5F}, {0x1EE61, 0x1EE62}, {0x1EE64, 0x1EE64}, {0x1EE67, 0x1EE6A}, {0x1EE6C, 0x1EE72},
+{0x1EE74, 0x1EE77}, {0x1EE79, 0x1EE7C}, {0x1EE7E, 0x1EE7E}, {0x1EE80, 0x1EE89}, {0x1EE8B, 0x1EE9B}, {0x1EEA1, 0x1EEA3}, {0x1EEA5, 0x1EEA9}, {0x1EEAB, 0x1EEBB}, {0x20000, 0x2A6DD}, {0x2A700, 0x2B734},
+{0x2B740, 0x2B81D}, {0x2B820, 0x2CEA1}, {0x2CEB0, 0x2EBE0}, {0x2F800, 0x2FA1D}, {0x30000, 0x3134A},
+};
 
 static const std::vector<std::pair<int, int>> whitespace_ranges = {
-{0x9, 0xD}, {0x1C, 0x20}, {0x85, 0x85}, {0xA0, 0xA0}, {0x1680, 0x1680}, {0x2000, 0x200A}, {0x2028, 0x2029}, {0x202F, 0x202F}, {0x205F, 0x205F}, {0x3000, 0x3000}, };
+{0x9, 0xD}, {0x1C, 0x20}, {0x85, 0x85}, {0xA0, 0xA0}, {0x1680, 0x1680}, {0x2000, 0x200A}, {0x2028, 0x2029}, {0x202F, 0x202F}, {0x205F, 0x205F}, {0x3000, 0x3000},
+};
 
 static const std::vector<std::pair<int, int>> accent_mark_ranges = {
-{0x300, 0x36F}, {0x483, 0x489}, {0x591, 0x5BD}, {0x5BF, 0x5BF}, {0x5C1, 0x5C2}, {0x5C4, 0x5C5}, {0x5C7, 0x5C7}, {0x610, 0x61A}, {0x64B, 0x65F}, {0x670, 0x670}, {0x6D6, 0x6DC}, {0x6DF, 0x6E4}, {0x6E7, 0x6E8}, {0x6EA, 0x6ED}, {0x711, 0x711}, {0x730, 0x74A}, {0x7A6, 0x7B0}, {0x7EB, 0x7F3}, {0x7FD, 0x7FD}, {0x816, 0x819}, {0x81B, 0x823}, {0x825, 0x827}, {0x829, 0x82D}, {0x859, 0x85B}, {0x8D3, 0x8E1}, {0x8E3, 0x903}, {0x93A, 0x93C}, {0x93E, 0x94F}, {0x951, 0x957}, {0x962, 0x963}, {0x981, 0x983}, {0x9BC, 0x9BC}, {0x9BE, 0x9C4}, {0x9C7, 0x9C8}, {0x9CB, 0x9CD}, {0x9D7, 0x9D7}, {0x9E2, 0x9E3}, {0x9FE, 0x9FE}, {0xA01, 0xA03}, {0xA3C, 0xA3C}, {0xA3E, 0xA42}, {0xA47, 0xA48}, {0xA4B, 0xA4D}, {0xA51, 0xA51}, {0xA70, 0xA71}, {0xA75, 0xA75}, {0xA81, 0xA83}, {0xABC, 0xABC}, {0xABE, 0xAC5}, {0xAC7, 0xAC9}, {0xACB, 0xACD}, {0xAE2, 0xAE3}, {0xAFA, 0xAFF}, {0xB01, 0xB03}, {0xB3C, 0xB3C}, {0xB3E, 0xB44}, {0xB47, 0xB48}, {0xB4B, 0xB4D}, {0xB55, 0xB57}, {0xB62, 0xB63}, {0xB82, 0xB82}, {0xBBE, 0xBC2}, {0xBC6, 0xBC8}, {0xBCA, 0xBCD}, {0xBD7, 0xBD7}, {0xC00, 0xC04}, {0xC3E, 0xC44}, {0xC46, 0xC48}, {0xC4A, 0xC4D}, {0xC55, 0xC56}, {0xC62, 0xC63}, {0xC81, 0xC83}, {0xCBC, 0xCBC}, {0xCBE, 0xCC4}, {0xCC6, 0xCC8}, {0xCCA, 0xCCD}, {0xCD5, 0xCD6}, {0xCE2, 0xCE3}, {0xD00, 0xD03}, {0xD3B, 0xD3C}, {0xD3E, 0xD44}, {0xD46, 0xD48}, {0xD4A, 0xD4D}, {0xD57, 0xD57}, {0xD62, 0xD63}, {0xD81, 0xD83}, {0xDCA, 0xDCA}, {0xDCF, 0xDD4}, {0xDD6, 0xDD6}, {0xDD8, 0xDDF}, {0xDF2, 0xDF3}, {0xE31, 0xE31}, {0xE34, 0xE3A}, {0xE47, 0xE4E}, {0xEB1, 0xEB1}, {0xEB4, 0xEBC}, {0xEC8, 0xECD}, {0xF18, 0xF19}, {0xF35, 0xF35}, {0xF37, 0xF37}, {0xF39, 0xF39}, {0xF3E, 0xF3F}, {0xF71, 0xF84}, {0xF86, 0xF87}, {0xF8D, 0xF97}, {0xF99, 0xFBC}, {0xFC6, 0xFC6}, {0x102B, 0x103E}, {0x1056, 0x1059}, {0x105E, 0x1060}, {0x1062, 0x1064}, {0x1067, 0x106D}, {0x1071, 0x1074}, {0x1082, 0x108D}, {0x108F, 0x108F}, {0x109A, 0x109D}, {0x135D, 0x135F}, {0x1712, 0x1714}, {0x1732, 0x1734}, {0x1752, 0x1753}, {0x1772, 0x1773}, {0x17B4, 0x17D3}, {0x17DD, 0x17DD}, {0x180B, 0x180D}, {0x1885, 0x1886}, {0x18A9, 0x18A9}, {0x1920, 0x192B}, {0x1930, 0x193B}, {0x1A17, 0x1A1B}, {0x1A55, 0x1A5E}, {0x1A60, 0x1A7C}, {0x1A7F, 0x1A7F}, {0x1AB0, 0x1AC0}, {0x1B00, 0x1B04}, {0x1B34, 0x1B44}, {0x1B6B, 0x1B73}, {0x1B80, 0x1B82}, {0x1BA1, 0x1BAD}, {0x1BE6, 0x1BF3}, {0x1C24, 0x1C37}, {0x1CD0, 0x1CD2}, {0x1CD4, 0x1CE8}, {0x1CED, 0x1CED}, {0x1CF4, 0x1CF4}, {0x1CF7, 0x1CF9}, {0x1DC0, 0x1DF9}, {0x1DFB, 0x1DFF}, {0x20D0, 0x20F0}, {0x2CEF, 0x2CF1}, {0x2D7F, 0x2D7F}, {0x2DE0, 0x2DFF}, {0x302A, 0x302F}, {0x3099, 0x309A}, {0xA66F, 0xA672}, {0xA674, 0xA67D}, {0xA69E, 0xA69F}, {0xA6F0, 0xA6F1}, {0xA802, 0xA802}, {0xA806, 0xA806}, {0xA80B, 0xA80B}, {0xA823, 0xA827}, {0xA82C, 0xA82C}, {0xA880, 0xA881}, {0xA8B4, 0xA8C5}, {0xA8E0, 0xA8F1}, {0xA8FF, 0xA8FF}, {0xA926, 0xA92D}, {0xA947, 0xA953}, {0xA980, 0xA983}, {0xA9B3, 0xA9C0}, {0xA9E5, 0xA9E5}, {0xAA29, 0xAA36}, {0xAA43, 0xAA43}, {0xAA4C, 0xAA4D}, {0xAA7B, 0xAA7D}, {0xAAB0, 0xAAB0}, {0xAAB2, 0xAAB4}, {0xAAB7, 0xAAB8}, {0xAABE, 0xAABF}, {0xAAC1, 0xAAC1}, {0xAAEB, 0xAAEF}, {0xAAF5, 0xAAF6}, {0xABE3, 0xABEA}, {0xABEC, 0xABED}, {0xFB1E, 0xFB1E}, {0xFE00, 0xFE0F}, {0xFE20, 0xFE2F}, {0x101FD, 0x101FD}, {0x102E0, 0x102E0}, {0x10376, 0x1037A}, {0x10A01, 0x10A03}, {0x10A05, 0x10A06}, {0x10A0C, 0x10A0F}, {0x10A38, 0x10A3A}, {0x10A3F, 0x10A3F}, {0x10AE5, 0x10AE6}, {0x10D24, 0x10D27}, {0x10EAB, 0x10EAC}, {0x10F46, 0x10F50}, {0x11000, 0x11002}, {0x11038, 0x11046}, {0x1107F, 0x11082}, {0x110B0, 0x110BA}, {0x11100, 0x11102}, {0x11127, 0x11134}, {0x11145, 0x11146}, {0x11173, 0x11173}, {0x11180, 0x11182}, {0x111B3, 0x111C0}, {0x111C9, 0x111CC}, {0x111CE, 0x111CF}, {0x1122C, 0x11237}, {0x1123E, 0x1123E}, {0x112DF, 0x112EA}, {0x11300, 0x11303}, {0x1133B, 0x1133C}, {0x1133E, 0x11344}, {0x11347, 0x11348}, {0x1134B, 0x1134D}, {0x11357, 0x11357}, {0x11362, 0x11363}, {0x11366, 0x1136C}, {0x11370, 0x11374}, {0x11435, 0x11446}, {0x1145E, 0x1145E}, {0x114B0, 0x114C3}, {0x115AF, 0x115B5}, {0x115B8, 0x115C0}, {0x115DC, 0x115DD}, {0x11630, 0x11640}, {0x116AB, 0x116B7}, {0x1171D, 0x1172B}, {0x1182C, 0x1183A}, {0x11930, 0x11935}, {0x11937, 0x11938}, {0x1193B, 0x1193E}, {0x11940, 0x11940}, {0x11942, 0x11943}, {0x119D1, 0x119D7}, {0x119DA, 0x119E0}, {0x119E4, 0x119E4}, {0x11A01, 0x11A0A}, {0x11A33, 0x11A39}, {0x11A3B, 0x11A3E}, {0x11A47, 0x11A47}, {0x11A51, 0x11A5B}, {0x11A8A, 0x11A99}, {0x11C2F, 0x11C36}, {0x11C38, 0x11C3F}, {0x11C92, 0x11CA7}, {0x11CA9, 0x11CB6}, {0x11D31, 0x11D36}, {0x11D3A, 0x11D3A}, {0x11D3C, 0x11D3D}, {0x11D3F, 0x11D45}, {0x11D47, 0x11D47}, {0x11D8A, 0x11D8E}, {0x11D90, 0x11D91}, {0x11D93, 0x11D97}, {0x11EF3, 0x11EF6}, {0x16AF0, 0x16AF4}, {0x16B30, 0x16B36}, {0x16F4F, 0x16F4F}, {0x16F51, 0x16F87}, {0x16F8F, 0x16F92}, {0x16FE4, 0x16FE4}, {0x16FF0, 0x16FF1}, {0x1BC9D, 0x1BC9E}, {0x1D165, 0x1D169}, {0x1D16D, 0x1D172}, {0x1D17B, 0x1D182}, {0x1D185, 0x1D18B}, {0x1D1AA, 0x1D1AD}, {0x1D242, 0x1D244}, {0x1DA00, 0x1DA36}, {0x1DA3B, 0x1DA6C}, {0x1DA75, 0x1DA75}, {0x1DA84, 0x1DA84}, {0x1DA9B, 0x1DA9F}, {0x1DAA1, 0x1DAAF}, {0x1E000, 0x1E006}, {0x1E008, 0x1E018}, {0x1E01B, 0x1E021}, {0x1E023, 0x1E024}, {0x1E026, 0x1E02A}, {0x1E130, 0x1E136}, {0x1E2EC, 0x1E2EF}, {0x1E8D0, 0x1E8D6}, {0x1E944, 0x1E94A}, {0xE0100, 0xE01EF}, };
+{0x300, 0x36F}, {0x483, 0x489}, {0x591, 0x5BD}, {0x5BF, 0x5BF}, {0x5C1, 0x5C2}, {0x5C4, 0x5C5}, {0x5C7, 0x5C7}, {0x610, 0x61A}, {0x64B, 0x65F}, {0x670, 0x670}, {0x6D6, 0x6DC}, {0x6DF, 0x6E4},
+{0x6E7, 0x6E8}, {0x6EA, 0x6ED}, {0x711, 0x711}, {0x730, 0x74A}, {0x7A6, 0x7B0}, {0x7EB, 0x7F3}, {0x7FD, 0x7FD}, {0x816, 0x819}, {0x81B, 0x823}, {0x825, 0x827}, {0x829, 0x82D}, {0x859, 0x85B},
+{0x8D3, 0x8E1}, {0x8E3, 0x903}, {0x93A, 0x93C}, {0x93E, 0x94F}, {0x951, 0x957}, {0x962, 0x963}, {0x981, 0x983}, {0x9BC, 0x9BC}, {0x9BE, 0x9C4}, {0x9C7, 0x9C8}, {0x9CB, 0x9CD}, {0x9D7, 0x9D7},
+{0x9E2, 0x9E3}, {0x9FE, 0x9FE}, {0xA01, 0xA03}, {0xA3C, 0xA3C}, {0xA3E, 0xA42}, {0xA47, 0xA48}, {0xA4B, 0xA4D}, {0xA51, 0xA51}, {0xA70, 0xA71}, {0xA75, 0xA75}, {0xA81, 0xA83}, {0xABC, 0xABC},
+{0xABE, 0xAC5}, {0xAC7, 0xAC9}, {0xACB, 0xACD}, {0xAE2, 0xAE3}, {0xAFA, 0xAFF}, {0xB01, 0xB03}, {0xB3C, 0xB3C}, {0xB3E, 0xB44}, {0xB47, 0xB48}, {0xB4B, 0xB4D}, {0xB55, 0xB57}, {0xB62, 0xB63},
+{0xB82, 0xB82}, {0xBBE, 0xBC2}, {0xBC6, 0xBC8}, {0xBCA, 0xBCD}, {0xBD7, 0xBD7}, {0xC00, 0xC04}, {0xC3E, 0xC44}, {0xC46, 0xC48}, {0xC4A, 0xC4D}, {0xC55, 0xC56}, {0xC62, 0xC63}, {0xC81, 0xC83},
+{0xCBC, 0xCBC}, {0xCBE, 0xCC4}, {0xCC6, 0xCC8}, {0xCCA, 0xCCD}, {0xCD5, 0xCD6}, {0xCE2, 0xCE3}, {0xD00, 0xD03}, {0xD3B, 0xD3C}, {0xD3E, 0xD44}, {0xD46, 0xD48}, {0xD4A, 0xD4D}, {0xD57, 0xD57},
+{0xD62, 0xD63}, {0xD81, 0xD83}, {0xDCA, 0xDCA}, {0xDCF, 0xDD4}, {0xDD6, 0xDD6}, {0xDD8, 0xDDF}, {0xDF2, 0xDF3}, {0xE31, 0xE31}, {0xE34, 0xE3A}, {0xE47, 0xE4E}, {0xEB1, 0xEB1}, {0xEB4, 0xEBC},
+{0xEC8, 0xECD}, {0xF18, 0xF19}, {0xF35, 0xF35}, {0xF37, 0xF37}, {0xF39, 0xF39}, {0xF3E, 0xF3F}, {0xF71, 0xF84}, {0xF86, 0xF87}, {0xF8D, 0xF97}, {0xF99, 0xFBC}, {0xFC6, 0xFC6}, {0x102B, 0x103E},
+{0x1056, 0x1059}, {0x105E, 0x1060}, {0x1062, 0x1064}, {0x1067, 0x106D}, {0x1071, 0x1074}, {0x1082, 0x108D}, {0x108F, 0x108F}, {0x109A, 0x109D}, {0x135D, 0x135F}, {0x1712, 0x1714}, {0x1732, 0x1734},
+{0x1752, 0x1753}, {0x1772, 0x1773}, {0x17B4, 0x17D3}, {0x17DD, 0x17DD}, {0x180B, 0x180D}, {0x1885, 0x1886}, {0x18A9, 0x18A9}, {0x1920, 0x192B}, {0x1930, 0x193B}, {0x1A17, 0x1A1B}, {0x1A55, 0x1A5E},
+{0x1A60, 0x1A7C}, {0x1A7F, 0x1A7F}, {0x1AB0, 0x1AC0}, {0x1B00, 0x1B04}, {0x1B34, 0x1B44}, {0x1B6B, 0x1B73}, {0x1B80, 0x1B82}, {0x1BA1, 0x1BAD}, {0x1BE6, 0x1BF3}, {0x1C24, 0x1C37}, {0x1CD0, 0x1CD2},
+{0x1CD4, 0x1CE8}, {0x1CED, 0x1CED}, {0x1CF4, 0x1CF4}, {0x1CF7, 0x1CF9}, {0x1DC0, 0x1DF9}, {0x1DFB, 0x1DFF}, {0x20D0, 0x20F0}, {0x2CEF, 0x2CF1}, {0x2D7F, 0x2D7F}, {0x2DE0, 0x2DFF}, {0x302A, 0x302F},
+{0x3099, 0x309A}, {0xA66F, 0xA672}, {0xA674, 0xA67D}, {0xA69E, 0xA69F}, {0xA6F0, 0xA6F1}, {0xA802, 0xA802}, {0xA806, 0xA806}, {0xA80B, 0xA80B}, {0xA823, 0xA827}, {0xA82C, 0xA82C}, {0xA880, 0xA881},
+{0xA8B4, 0xA8C5}, {0xA8E0, 0xA8F1}, {0xA8FF, 0xA8FF}, {0xA926, 0xA92D}, {0xA947, 0xA953}, {0xA980, 0xA983}, {0xA9B3, 0xA9C0}, {0xA9E5, 0xA9E5}, {0xAA29, 0xAA36}, {0xAA43, 0xAA43}, {0xAA4C, 0xAA4D},
+{0xAA7B, 0xAA7D}, {0xAAB0, 0xAAB0}, {0xAAB2, 0xAAB4}, {0xAAB7, 0xAAB8}, {0xAABE, 0xAABF}, {0xAAC1, 0xAAC1}, {0xAAEB, 0xAAEF}, {0xAAF5, 0xAAF6}, {0xABE3, 0xABEA}, {0xABEC, 0xABED}, {0xFB1E, 0xFB1E},
+{0xFE00, 0xFE0F}, {0xFE20, 0xFE2F}, {0x101FD, 0x101FD}, {0x102E0, 0x102E0}, {0x10376, 0x1037A}, {0x10A01, 0x10A03}, {0x10A05, 0x10A06}, {0x10A0C, 0x10A0F}, {0x10A38, 0x10A3A}, {0x10A3F, 0x10A3F},
+{0x10AE5, 0x10AE6}, {0x10D24, 0x10D27}, {0x10EAB, 0x10EAC}, {0x10F46, 0x10F50}, {0x11000, 0x11002}, {0x11038, 0x11046}, {0x1107F, 0x11082}, {0x110B0, 0x110BA}, {0x11100, 0x11102}, {0x11127, 0x11134},
+{0x11145, 0x11146}, {0x11173, 0x11173}, {0x11180, 0x11182}, {0x111B3, 0x111C0}, {0x111C9, 0x111CC}, {0x111CE, 0x111CF}, {0x1122C, 0x11237}, {0x1123E, 0x1123E}, {0x112DF, 0x112EA}, {0x11300, 0x11303},
+{0x1133B, 0x1133C}, {0x1133E, 0x11344}, {0x11347, 0x11348}, {0x1134B, 0x1134D}, {0x11357, 0x11357}, {0x11362, 0x11363}, {0x11366, 0x1136C}, {0x11370, 0x11374}, {0x11435, 0x11446}, {0x1145E, 0x1145E},
+{0x114B0, 0x114C3}, {0x115AF, 0x115B5}, {0x115B8, 0x115C0}, {0x115DC, 0x115DD}, {0x11630, 0x11640}, {0x116AB, 0x116B7}, {0x1171D, 0x1172B}, {0x1182C, 0x1183A}, {0x11930, 0x11935}, {0x11937, 0x11938},
+{0x1193B, 0x1193E}, {0x11940, 0x11940}, {0x11942, 0x11943}, {0x119D1, 0x119D7}, {0x119DA, 0x119E0}, {0x119E4, 0x119E4}, {0x11A01, 0x11A0A}, {0x11A33, 0x11A39}, {0x11A3B, 0x11A3E}, {0x11A47, 0x11A47},
+{0x11A51, 0x11A5B}, {0x11A8A, 0x11A99}, {0x11C2F, 0x11C36}, {0x11C38, 0x11C3F}, {0x11C92, 0x11CA7}, {0x11CA9, 0x11CB6}, {0x11D31, 0x11D36}, {0x11D3A, 0x11D3A}, {0x11D3C, 0x11D3D}, {0x11D3F, 0x11D45},
+{0x11D47, 0x11D47}, {0x11D8A, 0x11D8E}, {0x11D90, 0x11D91}, {0x11D93, 0x11D97}, {0x11EF3, 0x11EF6}, {0x16AF0, 0x16AF4}, {0x16B30, 0x16B36}, {0x16F4F, 0x16F4F}, {0x16F51, 0x16F87}, {0x16F8F, 0x16F92},
+{0x16FE4, 0x16FE4}, {0x16FF0, 0x16FF1}, {0x1BC9D, 0x1BC9E}, {0x1D165, 0x1D169}, {0x1D16D, 0x1D172}, {0x1D17B, 0x1D182}, {0x1D185, 0x1D18B}, {0x1D1AA, 0x1D1AD}, {0x1D242, 0x1D244}, {0x1DA00, 0x1DA36},
+{0x1DA3B, 0x1DA6C}, {0x1DA75, 0x1DA75}, {0x1DA84, 0x1DA84}, {0x1DA9B, 0x1DA9F}, {0x1DAA1, 0x1DAAF}, {0x1E000, 0x1E006}, {0x1E008, 0x1E018}, {0x1E01B, 0x1E021}, {0x1E023, 0x1E024}, {0x1E026, 0x1E02A},
+{0x1E130, 0x1E136}, {0x1E2EC, 0x1E2EF}, {0x1E8D0, 0x1E8D6}, {0x1E944, 0x1E94A}, {0xE0100, 0xE01EF},
+};
 
 static const std::vector<std::pair<int, int>> punctuation_ranges = {
-{0x21, 0x23}, {0x25, 0x2A}, {0x2C, 0x2F}, {0x3A, 0x3B}, {0x3F, 0x40}, {0x5B, 0x5D}, {0x5F, 0x5F}, {0x7B, 0x7B}, {0x7D, 0x7D}, {0xA1, 0xA1}, {0xA7, 0xA7}, {0xAB, 0xAB}, {0xB6, 0xB7}, {0xBB, 0xBB}, {0xBF, 0xBF}, {0x37E, 0x37E}, {0x387, 0x387}, {0x55A, 0x55F}, {0x589, 0x58A}, {0x5BE, 0x5BE}, {0x5C0, 0x5C0}, {0x5C3, 0x5C3}, {0x5C6, 0x5C6}, {0x5F3, 0x5F4}, {0x609, 0x60A}, {0x60C, 0x60D}, {0x61B, 0x61B}, {0x61E, 0x61F}, {0x66A, 0x66D}, {0x6D4, 0x6D4}, {0x700, 0x70D}, {0x7F7, 0x7F9}, {0x830, 0x83E}, {0x85E, 0x85E}, {0x964, 0x965}, {0x970, 0x970}, {0x9FD, 0x9FD}, {0xA76, 0xA76}, {0xAF0, 0xAF0}, {0xC77, 0xC77}, {0xC84, 0xC84}, {0xDF4, 0xDF4}, {0xE4F, 0xE4F}, {0xE5A, 0xE5B}, {0xF04, 0xF12}, {0xF14, 0xF14}, {0xF3A, 0xF3D}, {0xF85, 0xF85}, {0xFD0, 0xFD4}, {0xFD9, 0xFDA}, {0x104A, 0x104F}, {0x10FB, 0x10FB}, {0x1360, 0x1368}, {0x1400, 0x1400}, {0x166E, 0x166E}, {0x169B, 0x169C}, {0x16EB, 0x16ED}, {0x1735, 0x1736}, {0x17D4, 0x17D6}, {0x17D8, 0x17DA}, {0x1800, 0x180A}, {0x1944, 0x1945}, {0x1A1E, 0x1A1F}, {0x1AA0, 0x1AA6}, {0x1AA8, 0x1AAD}, {0x1B5A, 0x1B60}, {0x1BFC, 0x1BFF}, {0x1C3B, 0x1C3F}, {0x1C7E, 0x1C7F}, {0x1CC0, 0x1CC7}, {0x1CD3, 0x1CD3}, {0x2010, 0x2027}, {0x2030, 0x2043}, {0x2045, 0x2051}, {0x2053, 0x205E}, {0x207D, 0x207E}, {0x208D, 0x208E}, {0x2308, 0x230B}, {0x2329, 0x232A}, {0x2768, 0x2775}, {0x27C5, 0x27C6}, {0x27E6, 0x27EF}, {0x2983, 0x2998}, {0x29D8, 0x29DB}, {0x29FC, 0x29FD}, {0x2CF9, 0x2CFC}, {0x2CFE, 0x2CFF}, {0x2D70, 0x2D70}, {0x2E00, 0x2E2E}, {0x2E30, 0x2E4F}, {0x2E52, 0x2E52}, {0x3001, 0x3003}, {0x3008, 0x3011}, {0x3014, 0x301F}, {0x3030, 0x3030}, {0x303D, 0x303D}, {0x30A0, 0x30A0}, {0x30FB, 0x30FB}, {0xA4FE, 0xA4FF}, {0xA60D, 0xA60F}, {0xA673, 0xA673}, {0xA67E, 0xA67E}, {0xA6F2, 0xA6F7}, {0xA874, 0xA877}, {0xA8CE, 0xA8CF}, {0xA8F8, 0xA8FA}, {0xA8FC, 0xA8FC}, {0xA92E, 0xA92F}, {0xA95F, 0xA95F}, {0xA9C1, 0xA9CD}, {0xA9DE, 0xA9DF}, {0xAA5C, 0xAA5F}, {0xAADE, 0xAADF}, {0xAAF0, 0xAAF1}, {0xABEB, 0xABEB}, {0xFD3E, 0xFD3F}, {0xFE10, 0xFE19}, {0xFE30, 0xFE52}, {0xFE54, 0xFE61}, {0xFE63, 0xFE63}, {0xFE68, 0xFE68}, {0xFE6A, 0xFE6B}, {0xFF01, 0xFF03}, {0xFF05, 0xFF0A}, {0xFF0C, 0xFF0F}, {0xFF1A, 0xFF1B}, {0xFF1F, 0xFF20}, {0xFF3B, 0xFF3D}, {0xFF3F, 0xFF3F}, {0xFF5B, 0xFF5B}, {0xFF5D, 0xFF5D}, {0xFF5F, 0xFF65}, {0x10100, 0x10102}, {0x1039F, 0x1039F}, {0x103D0, 0x103D0}, {0x1056F, 0x1056F}, {0x10857, 0x10857}, {0x1091F, 0x1091F}, {0x1093F, 0x1093F}, {0x10A50, 0x10A58}, {0x10A7F, 0x10A7F}, {0x10AF0, 0x10AF6}, {0x10B39, 0x10B3F}, {0x10B99, 0x10B9C}, {0x10EAD, 0x10EAD}, {0x10F55, 0x10F59}, {0x11047, 0x1104D}, {0x110BB, 0x110BC}, {0x110BE, 0x110C1}, {0x11140, 0x11143}, {0x11174, 0x11175}, {0x111C5, 0x111C8}, {0x111CD, 0x111CD}, {0x111DB, 0x111DB}, {0x111DD, 0x111DF}, {0x11238, 0x1123D}, {0x112A9, 0x112A9}, {0x1144B, 0x1144F}, {0x1145A, 0x1145B}, {0x1145D, 0x1145D}, {0x114C6, 0x114C6}, {0x115C1, 0x115D7}, {0x11641, 0x11643}, {0x11660, 0x1166C}, {0x1173C, 0x1173E}, {0x1183B, 0x1183B}, {0x11944, 0x11946}, {0x119E2, 0x119E2}, {0x11A3F, 0x11A46}, {0x11A9A, 0x11A9C}, {0x11A9E, 0x11AA2}, {0x11C41, 0x11C45}, {0x11C70, 0x11C71}, {0x11EF7, 0x11EF8}, {0x11FFF, 0x11FFF}, {0x12470, 0x12474}, {0x16A6E, 0x16A6F}, {0x16AF5, 0x16AF5}, {0x16B37, 0x16B3B}, {0x16B44, 0x16B44}, {0x16E97, 0x16E9A}, {0x16FE2, 0x16FE2}, {0x1BC9F, 0x1BC9F}, {0x1DA87, 0x1DA8B}, {0x1E95E, 0x1E95F}, };
+{0x21, 0x23}, {0x25, 0x2A}, {0x2C, 0x2F}, {0x3A, 0x3B}, {0x3F, 0x40}, {0x5B, 0x5D}, {0x5F, 0x5F}, {0x7B, 0x7B}, {0x7D, 0x7D}, {0xA1, 0xA1}, {0xA7, 0xA7}, {0xAB, 0xAB}, {0xB6, 0xB7}, {0xBB, 0xBB},
+{0xBF, 0xBF}, {0x37E, 0x37E}, {0x387, 0x387}, {0x55A, 0x55F}, {0x589, 0x58A}, {0x5BE, 0x5BE}, {0x5C0, 0x5C0}, {0x5C3, 0x5C3}, {0x5C6, 0x5C6}, {0x5F3, 0x5F4}, {0x609, 0x60A}, {0x60C, 0x60D},
+{0x61B, 0x61B}, {0x61E, 0x61F}, {0x66A, 0x66D}, {0x6D4, 0x6D4}, {0x700, 0x70D}, {0x7F7, 0x7F9}, {0x830, 0x83E}, {0x85E, 0x85E}, {0x964, 0x965}, {0x970, 0x970}, {0x9FD, 0x9FD}, {0xA76, 0xA76},
+{0xAF0, 0xAF0}, {0xC77, 0xC77}, {0xC84, 0xC84}, {0xDF4, 0xDF4}, {0xE4F, 0xE4F}, {0xE5A, 0xE5B}, {0xF04, 0xF12}, {0xF14, 0xF14}, {0xF3A, 0xF3D}, {0xF85, 0xF85}, {0xFD0, 0xFD4}, {0xFD9, 0xFDA},
+{0x104A, 0x104F}, {0x10FB, 0x10FB}, {0x1360, 0x1368}, {0x1400, 0x1400}, {0x166E, 0x166E}, {0x169B, 0x169C}, {0x16EB, 0x16ED}, {0x1735, 0x1736}, {0x17D4, 0x17D6}, {0x17D8, 0x17DA}, {0x1800, 0x180A},
+{0x1944, 0x1945}, {0x1A1E, 0x1A1F}, {0x1AA0, 0x1AA6}, {0x1AA8, 0x1AAD}, {0x1B5A, 0x1B60}, {0x1BFC, 0x1BFF}, {0x1C3B, 0x1C3F}, {0x1C7E, 0x1C7F}, {0x1CC0, 0x1CC7}, {0x1CD3, 0x1CD3}, {0x2010, 0x2027},
+{0x2030, 0x2043}, {0x2045, 0x2051}, {0x2053, 0x205E}, {0x207D, 0x207E}, {0x208D, 0x208E}, {0x2308, 0x230B}, {0x2329, 0x232A}, {0x2768, 0x2775}, {0x27C5, 0x27C6}, {0x27E6, 0x27EF}, {0x2983, 0x2998},
+{0x29D8, 0x29DB}, {0x29FC, 0x29FD}, {0x2CF9, 0x2CFC}, {0x2CFE, 0x2CFF}, {0x2D70, 0x2D70}, {0x2E00, 0x2E2E}, {0x2E30, 0x2E4F}, {0x2E52, 0x2E52}, {0x3001, 0x3003}, {0x3008, 0x3011}, {0x3014, 0x301F},
+{0x3030, 0x3030}, {0x303D, 0x303D}, {0x30A0, 0x30A0}, {0x30FB, 0x30FB}, {0xA4FE, 0xA4FF}, {0xA60D, 0xA60F}, {0xA673, 0xA673}, {0xA67E, 0xA67E}, {0xA6F2, 0xA6F7}, {0xA874, 0xA877}, {0xA8CE, 0xA8CF},
+{0xA8F8, 0xA8FA}, {0xA8FC, 0xA8FC}, {0xA92E, 0xA92F}, {0xA95F, 0xA95F}, {0xA9C1, 0xA9CD}, {0xA9DE, 0xA9DF}, {0xAA5C, 0xAA5F}, {0xAADE, 0xAADF}, {0xAAF0, 0xAAF1}, {0xABEB, 0xABEB}, {0xFD3E, 0xFD3F},
+{0xFE10, 0xFE19}, {0xFE30, 0xFE52}, {0xFE54, 0xFE61}, {0xFE63, 0xFE63}, {0xFE68, 0xFE68}, {0xFE6A, 0xFE6B}, {0xFF01, 0xFF03}, {0xFF05, 0xFF0A}, {0xFF0C, 0xFF0F}, {0xFF1A, 0xFF1B}, {0xFF1F, 0xFF20},
+{0xFF3B, 0xFF3D}, {0xFF3F, 0xFF3F}, {0xFF5B, 0xFF5B}, {0xFF5D, 0xFF5D}, {0xFF5F, 0xFF65}, {0x10100, 0x10102}, {0x1039F, 0x1039F}, {0x103D0, 0x103D0}, {0x1056F, 0x1056F}, {0x10857, 0x10857},
+{0x1091F, 0x1091F}, {0x1093F, 0x1093F}, {0x10A50, 0x10A58}, {0x10A7F, 0x10A7F}, {0x10AF0, 0x10AF6}, {0x10B39, 0x10B3F}, {0x10B99, 0x10B9C}, {0x10EAD, 0x10EAD}, {0x10F55, 0x10F59}, {0x11047, 0x1104D},
+{0x110BB, 0x110BC}, {0x110BE, 0x110C1}, {0x11140, 0x11143}, {0x11174, 0x11175}, {0x111C5, 0x111C8}, {0x111CD, 0x111CD}, {0x111DB, 0x111DB}, {0x111DD, 0x111DF}, {0x11238, 0x1123D}, {0x112A9, 0x112A9},
+{0x1144B, 0x1144F}, {0x1145A, 0x1145B}, {0x1145D, 0x1145D}, {0x114C6, 0x114C6}, {0x115C1, 0x115D7}, {0x11641, 0x11643}, {0x11660, 0x1166C}, {0x1173C, 0x1173E}, {0x1183B, 0x1183B}, {0x11944, 0x11946},
+{0x119E2, 0x119E2}, {0x11A3F, 0x11A46}, {0x11A9A, 0x11A9C}, {0x11A9E, 0x11AA2}, {0x11C41, 0x11C45}, {0x11C70, 0x11C71}, {0x11EF7, 0x11EF8}, {0x11FFF, 0x11FFF}, {0x12470, 0x12474}, {0x16A6E, 0x16A6F},
+{0x16AF5, 0x16AF5}, {0x16B37, 0x16B3B}, {0x16B44, 0x16B44}, {0x16E97, 0x16E9A}, {0x16FE2, 0x16FE2}, {0x1BC9F, 0x1BC9F}, {0x1DA87, 0x1DA8B}, {0x1E95E, 0x1E95F},
+};
 
 static const std::vector<std::pair<int, int>> symbol_ranges = {
-{0x24, 0x24}, {0x2B, 0x2B}, {0x3C, 0x3E}, {0x5E, 0x5E}, {0x60, 0x60}, {0x7C, 0x7C}, {0x7E, 0x7E}, {0xA2, 0xA6}, {0xA8, 0xA9}, {0xAC, 0xAC}, {0xAE, 0xB1}, {0xB4, 0xB4}, {0xB8, 0xB8}, {0xD7, 0xD7}, {0xF7, 0xF7}, {0x2C2, 0x2C5}, {0x2D2, 0x2DF}, {0x2E5, 0x2EB}, {0x2ED, 0x2ED}, {0x2EF, 0x2FF}, {0x375, 0x375}, {0x384, 0x385}, {0x3F6, 0x3F6}, {0x482, 0x482}, {0x58D, 0x58F}, {0x606, 0x608}, {0x60B, 0x60B}, {0x60E, 0x60F}, {0x6DE, 0x6DE}, {0x6E9, 0x6E9}, {0x6FD, 0x6FE}, {0x7F6, 0x7F6}, {0x7FE, 0x7FF}, {0x9F2, 0x9F3}, {0x9FA, 0x9FB}, {0xAF1, 0xAF1}, {0xB70, 0xB70}, {0xBF3, 0xBFA}, {0xC7F, 0xC7F}, {0xD4F, 0xD4F}, {0xD79, 0xD79}, {0xE3F, 0xE3F}, {0xF01, 0xF03}, {0xF13, 0xF13}, {0xF15, 0xF17}, {0xF1A, 0xF1F}, {0xF34, 0xF34}, {0xF36, 0xF36}, {0xF38, 0xF38}, {0xFBE, 0xFC5}, {0xFC7, 0xFCC}, {0xFCE, 0xFCF}, {0xFD5, 0xFD8}, {0x109E, 0x109F}, {0x1390, 0x1399}, {0x166D, 0x166D}, {0x17DB, 0x17DB}, {0x1940, 0x1940}, {0x19DE, 0x19FF}, {0x1B61, 0x1B6A}, {0x1B74, 0x1B7C}, {0x1FBD, 0x1FBD}, {0x1FBF, 0x1FC1}, {0x1FCD, 0x1FCF}, {0x1FDD, 0x1FDF}, {0x1FED, 0x1FEF}, {0x1FFD, 0x1FFE}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207A, 0x207C}, {0x208A, 0x208C}, {0x20A0, 0x20BF}, {0x2100, 0x2101}, {0x2103, 0x2106}, {0x2108, 0x2109}, {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211E, 0x2123}, {0x2125, 0x2125}, {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212E, 0x212E}, {0x213A, 0x213B}, {0x2140, 0x2144}, {0x214A, 0x214D}, {0x214F, 0x214F}, {0x218A, 0x218B}, {0x2190, 0x2307}, {0x230C, 0x2328}, {0x232B, 0x2426}, {0x2440, 0x244A}, {0x249C, 0x24E9}, {0x2500, 0x2767}, {0x2794, 0x27C4}, {0x27C7, 0x27E5}, {0x27F0, 0x2982}, {0x2999, 0x29D7}, {0x29DC, 0x29FB}, {0x29FE, 0x2B73}, {0x2B76, 0x2B95}, {0x2B97, 0x2BFF}, {0x2CE5, 0x2CEA}, {0x2E50, 0x2E51}, {0x2E80, 0x2E99}, {0x2E9B, 0x2EF3}, {0x2F00, 0x2FD5}, {0x2FF0, 0x2FFB}, {0x3004, 0x3004}, {0x3012, 0x3013}, {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303E, 0x303F}, {0x309B, 0x309C}, {0x3190, 0x3191}, {0x3196, 0x319F}, {0x31C0, 0x31E3}, {0x3200, 0x321E}, {0x322A, 0x3247}, {0x3250, 0x3250}, {0x3260, 0x327F}, {0x328A, 0x32B0}, {0x32C0, 0x33FF}, {0x4DC0, 0x4DFF}, {0xA490, 0xA4C6}, {0xA700, 0xA716}, {0xA720, 0xA721}, {0xA789, 0xA78A}, {0xA828, 0xA82B}, {0xA836, 0xA839}, {0xAA77, 0xAA79}, {0xAB5B, 0xAB5B}, {0xAB6A, 0xAB6B}, {0xFB29, 0xFB29}, {0xFBB2, 0xFBC1}, {0xFDFC, 0xFDFD}, {0xFE62, 0xFE62}, {0xFE64, 0xFE66}, {0xFE69, 0xFE69}, {0xFF04, 0xFF04}, {0xFF0B, 0xFF0B}, {0xFF1C, 0xFF1E}, {0xFF3E, 0xFF3E}, {0xFF40, 0xFF40}, {0xFF5C, 0xFF5C}, {0xFF5E, 0xFF5E}, {0xFFE0, 0xFFE6}, {0xFFE8, 0xFFEE}, {0xFFFC, 0xFFFD}, {0x10137, 0x1013F}, {0x10179, 0x10189}, {0x1018C, 0x1018E}, {0x10190, 0x1019C}, {0x101A0, 0x101A0}, {0x101D0, 0x101FC}, {0x10877, 0x10878}, {0x10AC8, 0x10AC8}, {0x1173F, 0x1173F}, {0x11FD5, 0x11FF1}, {0x16B3C, 0x16B3F}, {0x16B45, 0x16B45}, {0x1BC9C, 0x1BC9C}, {0x1D000, 0x1D0F5}, {0x1D100, 0x1D126}, {0x1D129, 0x1D164}, {0x1D16A, 0x1D16C}, {0x1D183, 0x1D184}, {0x1D18C, 0x1D1A9}, {0x1D1AE, 0x1D1E8}, {0x1D200, 0x1D241}, {0x1D245, 0x1D245}, {0x1D300, 0x1D356}, {0x1D6C1, 0x1D6C1}, {0x1D6DB, 0x1D6DB}, {0x1D6FB, 0x1D6FB}, {0x1D715, 0x1D715}, {0x1D735, 0x1D735}, {0x1D74F, 0x1D74F}, {0x1D76F, 0x1D76F}, {0x1D789, 0x1D789}, {0x1D7A9, 0x1D7A9}, {0x1D7C3, 0x1D7C3}, {0x1D800, 0x1D9FF}, {0x1DA37, 0x1DA3A}, {0x1DA6D, 0x1DA74}, {0x1DA76, 0x1DA83}, {0x1DA85, 0x1DA86}, {0x1E14F, 0x1E14F}, {0x1E2FF, 0x1E2FF}, {0x1ECAC, 0x1ECAC}, {0x1ECB0, 0x1ECB0}, {0x1ED2E, 0x1ED2E}, {0x1EEF0, 0x1EEF1}, {0x1F000, 0x1F02B}, {0x1F030, 0x1F093}, {0x1F0A0, 0x1F0AE}, {0x1F0B1, 0x1F0BF}, {0x1F0C1, 0x1F0CF}, {0x1F0D1, 0x1F0F5}, {0x1F10D, 0x1F1AD}, {0x1F1E6, 0x1F202}, {0x1F210, 0x1F23B}, {0x1F240, 0x1F248}, {0x1F250, 0x1F251}, {0x1F260, 0x1F265}, {0x1F300, 0x1F6D7}, {0x1F6E0, 0x1F6EC}, {0x1F6F0, 0x1F6FC}, {0x1F700, 0x1F773}, {0x1F780, 0x1F7D8}, {0x1F7E0, 0x1F7EB}, {0x1F800, 0x1F80B}, {0x1F810, 0x1F847}, {0x1F850, 0x1F859}, {0x1F860, 0x1F887}, {0x1F890, 0x1F8AD}, {0x1F8B0, 0x1F8B1}, {0x1F900, 0x1F978}, {0x1F97A, 0x1F9CB}, {0x1F9CD, 0x1FA53}, {0x1FA60, 0x1FA6D}, {0x1FA70, 0x1FA74}, {0x1FA78, 0x1FA7A}, {0x1FA80, 0x1FA86}, {0x1FA90, 0x1FAA8}, {0x1FAB0, 0x1FAB6}, {0x1FAC0, 0x1FAC2}, {0x1FAD0, 0x1FAD6}, {0x1FB00, 0x1FB92}, {0x1FB94, 0x1FBCA}, };
+{0x24, 0x24}, {0x2B, 0x2B}, {0x3C, 0x3E}, {0x5E, 0x5E}, {0x60, 0x60}, {0x7C, 0x7C}, {0x7E, 0x7E}, {0xA2, 0xA6}, {0xA8, 0xA9}, {0xAC, 0xAC}, {0xAE, 0xB1}, {0xB4, 0xB4}, {0xB8, 0xB8}, {0xD7, 0xD7},
+{0xF7, 0xF7}, {0x2C2, 0x2C5}, {0x2D2, 0x2DF}, {0x2E5, 0x2EB}, {0x2ED, 0x2ED}, {0x2EF, 0x2FF}, {0x375, 0x375}, {0x384, 0x385}, {0x3F6, 0x3F6}, {0x482, 0x482}, {0x58D, 0x58F}, {0x606, 0x608},
+{0x60B, 0x60B}, {0x60E, 0x60F}, {0x6DE, 0x6DE}, {0x6E9, 0x6E9}, {0x6FD, 0x6FE}, {0x7F6, 0x7F6}, {0x7FE, 0x7FF}, {0x9F2, 0x9F3}, {0x9FA, 0x9FB}, {0xAF1, 0xAF1}, {0xB70, 0xB70}, {0xBF3, 0xBFA},
+{0xC7F, 0xC7F}, {0xD4F, 0xD4F}, {0xD79, 0xD79}, {0xE3F, 0xE3F}, {0xF01, 0xF03}, {0xF13, 0xF13}, {0xF15, 0xF17}, {0xF1A, 0xF1F}, {0xF34, 0xF34}, {0xF36, 0xF36}, {0xF38, 0xF38}, {0xFBE, 0xFC5},
+{0xFC7, 0xFCC}, {0xFCE, 0xFCF}, {0xFD5, 0xFD8}, {0x109E, 0x109F}, {0x1390, 0x1399}, {0x166D, 0x166D}, {0x17DB, 0x17DB}, {0x1940, 0x1940}, {0x19DE, 0x19FF}, {0x1B61, 0x1B6A}, {0x1B74, 0x1B7C},
+{0x1FBD, 0x1FBD}, {0x1FBF, 0x1FC1}, {0x1FCD, 0x1FCF}, {0x1FDD, 0x1FDF}, {0x1FED, 0x1FEF}, {0x1FFD, 0x1FFE}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207A, 0x207C}, {0x208A, 0x208C}, {0x20A0, 0x20BF},
+{0x2100, 0x2101}, {0x2103, 0x2106}, {0x2108, 0x2109}, {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211E, 0x2123}, {0x2125, 0x2125}, {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212E, 0x212E}, {0x213A, 0x213B},
+{0x2140, 0x2144}, {0x214A, 0x214D}, {0x214F, 0x214F}, {0x218A, 0x218B}, {0x2190, 0x2307}, {0x230C, 0x2328}, {0x232B, 0x2426}, {0x2440, 0x244A}, {0x249C, 0x24E9}, {0x2500, 0x2767}, {0x2794, 0x27C4},
+{0x27C7, 0x27E5}, {0x27F0, 0x2982}, {0x2999, 0x29D7}, {0x29DC, 0x29FB}, {0x29FE, 0x2B73}, {0x2B76, 0x2B95}, {0x2B97, 0x2BFF}, {0x2CE5, 0x2CEA}, {0x2E50, 0x2E51}, {0x2E80, 0x2E99}, {0x2E9B, 0x2EF3},
+{0x2F00, 0x2FD5}, {0x2FF0, 0x2FFB}, {0x3004, 0x3004}, {0x3012, 0x3013}, {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303E, 0x303F}, {0x309B, 0x309C}, {0x3190, 0x3191}, {0x3196, 0x319F}, {0x31C0, 0x31E3},
+{0x3200, 0x321E}, {0x322A, 0x3247}, {0x3250, 0x3250}, {0x3260, 0x327F}, {0x328A, 0x32B0}, {0x32C0, 0x33FF}, {0x4DC0, 0x4DFF}, {0xA490, 0xA4C6}, {0xA700, 0xA716}, {0xA720, 0xA721}, {0xA789, 0xA78A},
+{0xA828, 0xA82B}, {0xA836, 0xA839}, {0xAA77, 0xAA79}, {0xAB5B, 0xAB5B}, {0xAB6A, 0xAB6B}, {0xFB29, 0xFB29}, {0xFBB2, 0xFBC1}, {0xFDFC, 0xFDFD}, {0xFE62, 0xFE62}, {0xFE64, 0xFE66}, {0xFE69, 0xFE69},
+{0xFF04, 0xFF04}, {0xFF0B, 0xFF0B}, {0xFF1C, 0xFF1E}, {0xFF3E, 0xFF3E}, {0xFF40, 0xFF40}, {0xFF5C, 0xFF5C}, {0xFF5E, 0xFF5E}, {0xFFE0, 0xFFE6}, {0xFFE8, 0xFFEE}, {0xFFFC, 0xFFFD}, {0x10137, 0x1013F},
+{0x10179, 0x10189}, {0x1018C, 0x1018E}, {0x10190, 0x1019C}, {0x101A0, 0x101A0}, {0x101D0, 0x101FC}, {0x10877, 0x10878}, {0x10AC8, 0x10AC8}, {0x1173F, 0x1173F}, {0x11FD5, 0x11FF1}, {0x16B3C, 0x16B3F},
+{0x16B45, 0x16B45}, {0x1BC9C, 0x1BC9C}, {0x1D000, 0x1D0F5}, {0x1D100, 0x1D126}, {0x1D129, 0x1D164}, {0x1D16A, 0x1D16C}, {0x1D183, 0x1D184}, {0x1D18C, 0x1D1A9}, {0x1D1AE, 0x1D1E8}, {0x1D200, 0x1D241},
+{0x1D245, 0x1D245}, {0x1D300, 0x1D356}, {0x1D6C1, 0x1D6C1}, {0x1D6DB, 0x1D6DB}, {0x1D6FB, 0x1D6FB}, {0x1D715, 0x1D715}, {0x1D735, 0x1D735}, {0x1D74F, 0x1D74F}, {0x1D76F, 0x1D76F}, {0x1D789, 0x1D789},
+{0x1D7A9, 0x1D7A9}, {0x1D7C3, 0x1D7C3}, {0x1D800, 0x1D9FF}, {0x1DA37, 0x1DA3A}, {0x1DA6D, 0x1DA74}, {0x1DA76, 0x1DA83}, {0x1DA85, 0x1DA86}, {0x1E14F, 0x1E14F}, {0x1E2FF, 0x1E2FF}, {0x1ECAC, 0x1ECAC},
+{0x1ECB0, 0x1ECB0}, {0x1ED2E, 0x1ED2E}, {0x1EEF0, 0x1EEF1}, {0x1F000, 0x1F02B}, {0x1F030, 0x1F093}, {0x1F0A0, 0x1F0AE}, {0x1F0B1, 0x1F0BF}, {0x1F0C1, 0x1F0CF}, {0x1F0D1, 0x1F0F5}, {0x1F10D, 0x1F1AD},
+{0x1F1E6, 0x1F202}, {0x1F210, 0x1F23B}, {0x1F240, 0x1F248}, {0x1F250, 0x1F251}, {0x1F260, 0x1F265}, {0x1F300, 0x1F6D7}, {0x1F6E0, 0x1F6EC}, {0x1F6F0, 0x1F6FC}, {0x1F700, 0x1F773}, {0x1F780, 0x1F7D8},
+{0x1F7E0, 0x1F7EB}, {0x1F800, 0x1F80B}, {0x1F810, 0x1F847}, {0x1F850, 0x1F859}, {0x1F860, 0x1F887}, {0x1F890, 0x1F8AD}, {0x1F8B0, 0x1F8B1}, {0x1F900, 0x1F978}, {0x1F97A, 0x1F9CB}, {0x1F9CD, 0x1FA53},
+{0x1FA60, 0x1FA6D}, {0x1FA70, 0x1FA74}, {0x1FA78, 0x1FA7A}, {0x1FA80, 0x1FA86}, {0x1FA90, 0x1FAA8}, {0x1FAB0, 0x1FAB6}, {0x1FAC0, 0x1FAC2}, {0x1FAD0, 0x1FAD6}, {0x1FB00, 0x1FB92}, {0x1FB94, 0x1FBCA},
+};
 
 static const std::vector<std::pair<int, int>> control_ranges = {
-{0x0, 0x8}, {0xE, 0x1B}, {0x7F, 0x84}, {0x86, 0x9F}, {0xAD, 0xAD}, {0x378, 0x379}, {0x380, 0x383}, {0x38B, 0x38B}, {0x38D, 0x38D}, {0x3A2, 0x3A2}, {0x530, 0x530}, {0x557, 0x558}, {0x58B, 0x58C}, {0x590, 0x590}, {0x5C8, 0x5CF}, {0x5EB, 0x5EE}, {0x5F5, 0x605}, {0x61C, 0x61D}, {0x6DD, 0x6DD}, {0x70E, 0x70F}, {0x74B, 0x74C}, {0x7B2, 0x7BF}, {0x7FB, 0x7FC}, {0x82E, 0x82F}, {0x83F, 0x83F}, {0x85C, 0x85D}, {0x85F, 0x85F}, {0x86B, 0x89F}, {0x8B5, 0x8B5}, {0x8C8, 0x8D2}, {0x8E2, 0x8E2}, {0x984, 0x984}, {0x98D, 0x98E}, {0x991, 0x992}, {0x9A9, 0x9A9}, {0x9B1, 0x9B1}, {0x9B3, 0x9B5}, {0x9BA, 0x9BB}, {0x9C5, 0x9C6}, {0x9C9, 0x9CA}, {0x9CF, 0x9D6}, {0x9D8, 0x9DB}, {0x9DE, 0x9DE}, {0x9E4, 0x9E5}, {0x9FF, 0xA00}, {0xA04, 0xA04}, {0xA0B, 0xA0E}, {0xA11, 0xA12}, {0xA29, 0xA29}, {0xA31, 0xA31}, {0xA34, 0xA34}, {0xA37, 0xA37}, {0xA3A, 0xA3B}, {0xA3D, 0xA3D}, {0xA43, 0xA46}, {0xA49, 0xA4A}, {0xA4E, 0xA50}, {0xA52, 0xA58}, {0xA5D, 0xA5D}, {0xA5F, 0xA65}, {0xA77, 0xA80}, {0xA84, 0xA84}, {0xA8E, 0xA8E}, {0xA92, 0xA92}, {0xAA9, 0xAA9}, {0xAB1, 0xAB1}, {0xAB4, 0xAB4}, {0xABA, 0xABB}, {0xAC6, 0xAC6}, {0xACA, 0xACA}, {0xACE, 0xACF}, {0xAD1, 0xADF}, {0xAE4, 0xAE5}, {0xAF2, 0xAF8}, {0xB00, 0xB00}, {0xB04, 0xB04}, {0xB0D, 0xB0E}, {0xB11, 0xB12}, {0xB29, 0xB29}, {0xB31, 0xB31}, {0xB34, 0xB34}, {0xB3A, 0xB3B}, {0xB45, 0xB46}, {0xB49, 0xB4A}, {0xB4E, 0xB54}, {0xB58, 0xB5B}, {0xB5E, 0xB5E}, {0xB64, 0xB65}, {0xB78, 0xB81}, {0xB84, 0xB84}, {0xB8B, 0xB8D}, {0xB91, 0xB91}, {0xB96, 0xB98}, {0xB9B, 0xB9B}, {0xB9D, 0xB9D}, {0xBA0, 0xBA2}, {0xBA5, 0xBA7}, {0xBAB, 0xBAD}, {0xBBA, 0xBBD}, {0xBC3, 0xBC5}, {0xBC9, 0xBC9}, {0xBCE, 0xBCF}, {0xBD1, 0xBD6}, {0xBD8, 0xBE5}, {0xBFB, 0xBFF}, {0xC0D, 0xC0D}, {0xC11, 0xC11}, {0xC29, 0xC29}, {0xC3A, 0xC3C}, {0xC45, 0xC45}, {0xC49, 0xC49}, {0xC4E, 0xC54}, {0xC57, 0xC57}, {0xC5B, 0xC5F}, {0xC64, 0xC65}, {0xC70, 0xC76}, {0xC8D, 0xC8D}, {0xC91, 0xC91}, {0xCA9, 0xCA9}, {0xCB4, 0xCB4}, {0xCBA, 0xCBB}, {0xCC5, 0xCC5}, {0xCC9, 0xCC9}, {0xCCE, 0xCD4}, {0xCD7, 0xCDD}, {0xCDF, 0xCDF}, {0xCE4, 0xCE5}, {0xCF0, 0xCF0}, {0xCF3, 0xCFF}, {0xD0D, 0xD0D}, {0xD11, 0xD11}, {0xD45, 0xD45}, {0xD49, 0xD49}, {0xD50, 0xD53}, {0xD64, 0xD65}, {0xD80, 0xD80}, {0xD84, 0xD84}, {0xD97, 0xD99}, {0xDB2, 0xDB2}, {0xDBC, 0xDBC}, {0xDBE, 0xDBF}, {0xDC7, 0xDC9}, {0xDCB, 0xDCE}, {0xDD5, 0xDD5}, {0xDD7, 0xDD7}, {0xDE0, 0xDE5}, {0xDF0, 0xDF1}, {0xDF5, 0xE00}, {0xE3B, 0xE3E}, {0xE5C, 0xE80}, {0xE83, 0xE83}, {0xE85, 0xE85}, {0xE8B, 0xE8B}, {0xEA4, 0xEA4}, {0xEA6, 0xEA6}, {0xEBE, 0xEBF}, {0xEC5, 0xEC5}, {0xEC7, 0xEC7}, {0xECE, 0xECF}, {0xEDA, 0xEDB}, {0xEE0, 0xEFF}, {0xF48, 0xF48}, {0xF6D, 0xF70}, {0xF98, 0xF98}, {0xFBD, 0xFBD}, {0xFCD, 0xFCD}, {0xFDB, 0xFFF}, {0x10C6, 0x10C6}, {0x10C8, 0x10CC}, {0x10CE, 0x10CF}, {0x1249, 0x1249}, {0x124E, 0x124F}, {0x1257, 0x1257}, {0x1259, 0x1259}, {0x125E, 0x125F}, {0x1289, 0x1289}, {0x128E, 0x128F}, {0x12B1, 0x12B1}, {0x12B6, 0x12B7}, {0x12BF, 0x12BF}, {0x12C1, 0x12C1}, {0x12C6, 0x12C7}, {0x12D7, 0x12D7}, {0x1311, 0x1311}, {0x1316, 0x1317}, {0x135B, 0x135C}, {0x137D, 0x137F}, {0x139A, 0x139F}, {0x13F6, 0x13F7}, {0x13FE, 0x13FF}, {0x169D, 0x169F}, {0x16F9, 0x16FF}, {0x170D, 0x170D}, {0x1715, 0x171F}, {0x1737, 0x173F}, {0x1754, 0x175F}, {0x176D, 0x176D}, {0x1771, 0x1771}, {0x1774, 0x177F}, {0x17DE, 0x17DF}, {0x17EA, 0x17EF}, {0x17FA, 0x17FF}, {0x180E, 0x180F}, {0x181A, 0x181F}, {0x1879, 0x187F}, {0x18AB, 0x18AF}, {0x18F6, 0x18FF}, {0x191F, 0x191F}, {0x192C, 0x192F}, {0x193C, 0x193F}, {0x1941, 0x1943}, {0x196E, 0x196F}, {0x1975, 0x197F}, {0x19AC, 0x19AF}, {0x19CA, 0x19CF}, {0x19DB, 0x19DD}, {0x1A1C, 0x1A1D}, {0x1A5F, 0x1A5F}, {0x1A7D, 0x1A7E}, {0x1A8A, 0x1A8F}, {0x1A9A, 0x1A9F}, {0x1AAE, 0x1AAF}, {0x1AC1, 0x1AFF}, {0x1B4C, 0x1B4F}, {0x1B7D, 0x1B7F}, {0x1BF4, 0x1BFB}, {0x1C38, 0x1C3A}, {0x1C4A, 0x1C4C}, {0x1C89, 0x1C8F}, {0x1CBB, 0x1CBC}, {0x1CC8, 0x1CCF}, {0x1CFB, 0x1CFF}, {0x1DFA, 0x1DFA}, {0x1F16, 0x1F17}, {0x1F1E, 0x1F1F}, {0x1F46, 0x1F47}, {0x1F4E, 0x1F4F}, {0x1F58, 0x1F58}, {0x1F5A, 0x1F5A}, {0x1F5C, 0x1F5C}, {0x1F5E, 0x1F5E}, {0x1F7E, 0x1F7F}, {0x1FB5, 0x1FB5}, {0x1FC5, 0x1FC5}, {0x1FD4, 0x1FD5}, {0x1FDC, 0x1FDC}, {0x1FF0, 0x1FF1}, {0x1FF5, 0x1FF5}, {0x1FFF, 0x1FFF}, {0x200B, 0x200F}, {0x202A, 0x202E}, {0x2060, 0x206F}, {0x2072, 0x2073}, {0x208F, 0x208F}, {0x209D, 0x209F}, {0x20C0, 0x20CF}, {0x20F1, 0x20FF}, {0x218C, 0x218F}, {0x2427, 0x243F}, {0x244B, 0x245F}, {0x2B74, 0x2B75}, {0x2B96, 0x2B96}, {0x2C2F, 0x2C2F}, {0x2C5F, 0x2C5F}, {0x2CF4, 0x2CF8}, {0x2D26, 0x2D26}, {0x2D28, 0x2D2C}, {0x2D2E, 0x2D2F}, {0x2D68, 0x2D6E}, {0x2D71, 0x2D7E}, {0x2D97, 0x2D9F}, {0x2DA7, 0x2DA7}, {0x2DAF, 0x2DAF}, {0x2DB7, 0x2DB7}, {0x2DBF, 0x2DBF}, {0x2DC7, 0x2DC7}, {0x2DCF, 0x2DCF}, {0x2DD7, 0x2DD7}, {0x2DDF, 0x2DDF}, {0x2E53, 0x2E7F}, {0x2E9A, 0x2E9A}, {0x2EF4, 0x2EFF}, {0x2FD6, 0x2FEF}, {0x2FFC, 0x2FFF}, {0x3040, 0x3040}, {0x3097, 0x3098}, {0x3100, 0x3104}, {0x3130, 0x3130}, {0x318F, 0x318F}, {0x31E4, 0x31EF}, {0x321F, 0x321F}, {0x9FFD, 0x9FFF}, {0xA48D, 0xA48F}, {0xA4C7, 0xA4CF}, {0xA62C, 0xA63F}, {0xA6F8, 0xA6FF}, {0xA7C0, 0xA7C1}, {0xA7CB, 0xA7F4}, {0xA82D, 0xA82F}, {0xA83A, 0xA83F}, {0xA878, 0xA87F}, {0xA8C6, 0xA8CD}, {0xA8DA, 0xA8DF}, {0xA954, 0xA95E}, {0xA97D, 0xA97F}, {0xA9CE, 0xA9CE}, {0xA9DA, 0xA9DD}, {0xA9FF, 0xA9FF}, {0xAA37, 0xAA3F}, {0xAA4E, 0xAA4F}, {0xAA5A, 0xAA5B}, {0xAAC3, 0xAADA}, {0xAAF7, 0xAB00}, {0xAB07, 0xAB08}, {0xAB0F, 0xAB10}, {0xAB17, 0xAB1F}, {0xAB27, 0xAB27}, {0xAB2F, 0xAB2F}, {0xAB6C, 0xAB6F}, {0xABEE, 0xABEF}, {0xABFA, 0xABFF}, {0xD7A4, 0xD7AF}, {0xD7C7, 0xD7CA}, {0xD7FC, 0xF8FF}, {0xFA6E, 0xFA6F}, {0xFADA, 0xFAFF}, {0xFB07, 0xFB12}, {0xFB18, 0xFB1C}, {0xFB37, 0xFB37}, {0xFB3D, 0xFB3D}, {0xFB3F, 0xFB3F}, {0xFB42, 0xFB42}, {0xFB45, 0xFB45}, {0xFBC2, 0xFBD2}, {0xFD40, 0xFD4F}, {0xFD90, 0xFD91}, {0xFDC8, 0xFDEF}, {0xFDFE, 0xFDFF}, {0xFE1A, 0xFE1F}, {0xFE53, 0xFE53}, {0xFE67, 0xFE67}, {0xFE6C, 0xFE6F}, {0xFE75, 0xFE75}, {0xFEFD, 0xFF00}, {0xFFBF, 0xFFC1}, {0xFFC8, 0xFFC9}, {0xFFD0, 0xFFD1}, {0xFFD8, 0xFFD9}, {0xFFDD, 0xFFDF}, {0xFFE7, 0xFFE7}, {0xFFEF, 0xFFFB}, {0xFFFE, 0xFFFF}, {0x1000C, 0x1000C}, {0x10027, 0x10027}, {0x1003B, 0x1003B}, {0x1003E, 0x1003E}, {0x1004E, 0x1004F}, {0x1005E, 0x1007F}, {0x100FB, 0x100FF}, {0x10103, 0x10106}, {0x10134, 0x10136}, {0x1018F, 0x1018F}, {0x1019D, 0x1019F}, {0x101A1, 0x101CF}, {0x101FE, 0x1027F}, {0x1029D, 0x1029F}, {0x102D1, 0x102DF}, {0x102FC, 0x102FF}, {0x10324, 0x1032C}, {0x1034B, 0x1034F}, {0x1037B, 0x1037F}, {0x1039E, 0x1039E}, {0x103C4, 0x103C7}, {0x103D6, 0x103FF}, {0x1049E, 0x1049F}, {0x104AA, 0x104AF}, {0x104D4, 0x104D7}, {0x104FC, 0x104FF}, {0x10528, 0x1052F}, {0x10564, 0x1056E}, {0x10570, 0x105FF}, {0x10737, 0x1073F}, {0x10756, 0x1075F}, {0x10768, 0x107FF}, {0x10806, 0x10807}, {0x10809, 0x10809}, {0x10836, 0x10836}, {0x10839, 0x1083B}, {0x1083D, 0x1083E}, {0x10856, 0x10856}, {0x1089F, 0x108A6}, {0x108B0, 0x108DF}, {0x108F3, 0x108F3}, {0x108F6, 0x108FA}, {0x1091C, 0x1091E}, {0x1093A, 0x1093E}, {0x10940, 0x1097F}, {0x109B8, 0x109BB}, {0x109D0, 0x109D1}, {0x10A04, 0x10A04}, {0x10A07, 0x10A0B}, {0x10A14, 0x10A14}, {0x10A18, 0x10A18}, {0x10A36, 0x10A37}, {0x10A3B, 0x10A3E}, {0x10A49, 0x10A4F}, {0x10A59, 0x10A5F}, {0x10AA0, 0x10ABF}, {0x10AE7, 0x10AEA}, {0x10AF7, 0x10AFF}, {0x10B36, 0x10B38}, {0x10B56, 0x10B57}, {0x10B73, 0x10B77}, {0x10B92, 0x10B98}, {0x10B9D, 0x10BA8}, {0x10BB0, 0x10BFF}, {0x10C49, 0x10C7F}, {0x10CB3, 0x10CBF}, {0x10CF3, 0x10CF9}, {0x10D28, 0x10D2F}, {0x10D3A, 0x10E5F}, {0x10E7F, 0x10E7F}, {0x10EAA, 0x10EAA}, {0x10EAE, 0x10EAF}, {0x10EB2, 0x10EFF}, {0x10F28, 0x10F2F}, {0x10F5A, 0x10FAF}, {0x10FCC, 0x10FDF}, {0x10FF7, 0x10FFF}, {0x1104E, 0x11051}, {0x11070, 0x1107E}, {0x110BD, 0x110BD}, {0x110C2, 0x110CF}, {0x110E9, 0x110EF}, {0x110FA, 0x110FF}, {0x11135, 0x11135}, {0x11148, 0x1114F}, {0x11177, 0x1117F}, {0x111E0, 0x111E0}, {0x111F5, 0x111FF}, {0x11212, 0x11212}, {0x1123F, 0x1127F}, {0x11287, 0x11287}, {0x11289, 0x11289}, {0x1128E, 0x1128E}, {0x1129E, 0x1129E}, {0x112AA, 0x112AF}, {0x112EB, 0x112EF}, {0x112FA, 0x112FF}, {0x11304, 0x11304}, {0x1130D, 0x1130E}, {0x11311, 0x11312}, {0x11329, 0x11329}, {0x11331, 0x11331}, {0x11334, 0x11334}, {0x1133A, 0x1133A}, {0x11345, 0x11346}, {0x11349, 0x1134A}, {0x1134E, 0x1134F}, {0x11351, 0x11356}, {0x11358, 0x1135C}, {0x11364, 0x11365}, {0x1136D, 0x1136F}, {0x11375, 0x113FF}, {0x1145C, 0x1145C}, {0x11462, 0x1147F}, {0x114C8, 0x114CF}, {0x114DA, 0x1157F}, {0x115B6, 0x115B7}, {0x115DE, 0x115FF}, {0x11645, 0x1164F}, {0x1165A, 0x1165F}, {0x1166D, 0x1167F}, {0x116B9, 0x116BF}, {0x116CA, 0x116FF}, {0x1171B, 0x1171C}, {0x1172C, 0x1172F}, {0x11740, 0x117FF}, {0x1183C, 0x1189F}, {0x118F3, 0x118FE}, {0x11907, 0x11908}, {0x1190A, 0x1190B}, {0x11914, 0x11914}, {0x11917, 0x11917}, {0x11936, 0x11936}, {0x11939, 0x1193A}, {0x11947, 0x1194F}, {0x1195A, 0x1199F}, {0x119A8, 0x119A9}, {0x119D8, 0x119D9}, {0x119E5, 0x119FF}, {0x11A48, 0x11A4F}, {0x11AA3, 0x11ABF}, {0x11AF9, 0x11BFF}, {0x11C09, 0x11C09}, {0x11C37, 0x11C37}, {0x11C46, 0x11C4F}, {0x11C6D, 0x11C6F}, {0x11C90, 0x11C91}, {0x11CA8, 0x11CA8}, {0x11CB7, 0x11CFF}, {0x11D07, 0x11D07}, {0x11D0A, 0x11D0A}, {0x11D37, 0x11D39}, {0x11D3B, 0x11D3B}, {0x11D3E, 0x11D3E}, {0x11D48, 0x11D4F}, {0x11D5A, 0x11D5F}, {0x11D66, 0x11D66}, {0x11D69, 0x11D69}, {0x11D8F, 0x11D8F}, {0x11D92, 0x11D92}, {0x11D99, 0x11D9F}, {0x11DAA, 0x11EDF}, {0x11EF9, 0x11FAF}, {0x11FB1, 0x11FBF}, {0x11FF2, 0x11FFE}, {0x1239A, 0x123FF}, {0x1246F, 0x1246F}, {0x12475, 0x1247F}, {0x12544, 0x12FFF}, {0x1342F, 0x143FF}, {0x14647, 0x167FF}, {0x16A39, 0x16A3F}, {0x16A5F, 0x16A5F}, {0x16A6A, 0x16A6D}, {0x16A70, 0x16ACF}, {0x16AEE, 0x16AEF}, {0x16AF6, 0x16AFF}, {0x16B46, 0x16B4F}, {0x16B5A, 0x16B5A}, {0x16B62, 0x16B62}, {0x16B78, 0x16B7C}, {0x16B90, 0x16E3F}, {0x16E9B, 0x16EFF}, {0x16F4B, 0x16F4E}, {0x16F88, 0x16F8E}, {0x16FA0, 0x16FDF}, {0x16FE5, 0x16FEF}, {0x16FF2, 0x16FFF}, {0x187F8, 0x187FF}, {0x18CD6, 0x18CFF}, {0x18D09, 0x1AFFF}, {0x1B11F, 0x1B14F}, {0x1B153, 0x1B163}, {0x1B168, 0x1B16F}, {0x1B2FC, 0x1BBFF}, {0x1BC6B, 0x1BC6F}, {0x1BC7D, 0x1BC7F}, {0x1BC89, 0x1BC8F}, {0x1BC9A, 0x1BC9B}, {0x1BCA0, 0x1CFFF}, {0x1D0F6, 0x1D0FF}, {0x1D127, 0x1D128}, {0x1D173, 0x1D17A}, {0x1D1E9, 0x1D1FF}, {0x1D246, 0x1D2DF}, {0x1D2F4, 0x1D2FF}, {0x1D357, 0x1D35F}, {0x1D379, 0x1D3FF}, {0x1D455, 0x1D455}, {0x1D49D, 0x1D49D}, {0x1D4A0, 0x1D4A1}, {0x1D4A3, 0x1D4A4}, {0x1D4A7, 0x1D4A8}, {0x1D4AD, 0x1D4AD}, {0x1D4BA, 0x1D4BA}, {0x1D4BC, 0x1D4BC}, {0x1D4C4, 0x1D4C4}, {0x1D506, 0x1D506}, {0x1D50B, 0x1D50C}, {0x1D515, 0x1D515}, {0x1D51D, 0x1D51D}, {0x1D53A, 0x1D53A}, {0x1D53F, 0x1D53F}, {0x1D545, 0x1D545}, {0x1D547, 0x1D549}, {0x1D551, 0x1D551}, {0x1D6A6, 0x1D6A7}, {0x1D7CC, 0x1D7CD}, {0x1DA8C, 0x1DA9A}, {0x1DAA0, 0x1DAA0}, {0x1DAB0, 0x1DFFF}, {0x1E007, 0x1E007}, {0x1E019, 0x1E01A}, {0x1E022, 0x1E022}, {0x1E025, 0x1E025}, {0x1E02B, 0x1E0FF}, {0x1E12D, 0x1E12F}, {0x1E13E, 0x1E13F}, {0x1E14A, 0x1E14D}, {0x1E150, 0x1E2BF}, {0x1E2FA, 0x1E2FE}, {0x1E300, 0x1E7FF}, {0x1E8C5, 0x1E8C6}, {0x1E8D7, 0x1E8FF}, {0x1E94C, 0x1E94F}, {0x1E95A, 0x1E95D}, {0x1E960, 0x1EC70}, {0x1ECB5, 0x1ED00}, {0x1ED3E, 0x1EDFF}, {0x1EE04, 0x1EE04}, {0x1EE20, 0x1EE20}, {0x1EE23, 0x1EE23}, {0x1EE25, 0x1EE26}, {0x1EE28, 0x1EE28}, {0x1EE33, 0x1EE33}, {0x1EE38, 0x1EE38}, {0x1EE3A, 0x1EE3A}, {0x1EE3C, 0x1EE41}, {0x1EE43, 0x1EE46}, {0x1EE48, 0x1EE48}, {0x1EE4A, 0x1EE4A}, {0x1EE4C, 0x1EE4C}, {0x1EE50, 0x1EE50}, {0x1EE53, 0x1EE53}, {0x1EE55, 0x1EE56}, {0x1EE58, 0x1EE58}, {0x1EE5A, 0x1EE5A}, {0x1EE5C, 0x1EE5C}, {0x1EE5E, 0x1EE5E}, {0x1EE60, 0x1EE60}, {0x1EE63, 0x1EE63}, {0x1EE65, 0x1EE66}, {0x1EE6B, 0x1EE6B}, {0x1EE73, 0x1EE73}, {0x1EE78, 0x1EE78}, {0x1EE7D, 0x1EE7D}, {0x1EE7F, 0x1EE7F}, {0x1EE8A, 0x1EE8A}, {0x1EE9C, 0x1EEA0}, {0x1EEA4, 0x1EEA4}, {0x1EEAA, 0x1EEAA}, {0x1EEBC, 0x1EEEF}, {0x1EEF2, 0x1EFFF}, {0x1F02C, 0x1F02F}, {0x1F094, 0x1F09F}, {0x1F0AF, 0x1F0B0}, {0x1F0C0, 0x1F0C0}, {0x1F0D0, 0x1F0D0}, {0x1F0F6, 0x1F0FF}, {0x1F1AE, 0x1F1E5}, {0x1F203, 0x1F20F}, {0x1F23C, 0x1F23F}, {0x1F249, 0x1F24F}, {0x1F252, 0x1F25F}, {0x1F266, 0x1F2FF}, {0x1F6D8, 0x1F6DF}, {0x1F6ED, 0x1F6EF}, {0x1F6FD, 0x1F6FF}, {0x1F774, 0x1F77F}, {0x1F7D9, 0x1F7DF}, {0x1F7EC, 0x1F7FF}, {0x1F80C, 0x1F80F}, {0x1F848, 0x1F84F}, {0x1F85A, 0x1F85F}, {0x1F888, 0x1F88F}, {0x1F8AE, 0x1F8AF}, {0x1F8B2, 0x1F8FF}, {0x1F979, 0x1F979}, {0x1F9CC, 0x1F9CC}, {0x1FA54, 0x1FA5F}, {0x1FA6E, 0x1FA6F}, {0x1FA75, 0x1FA77}, {0x1FA7B, 0x1FA7F}, {0x1FA87, 0x1FA8F}, {0x1FAA9, 0x1FAAF}, {0x1FAB7, 0x1FABF}, {0x1FAC3, 0x1FACF}, {0x1FAD7, 0x1FAFF}, {0x1FB93, 0x1FB93}, {0x1FBCB, 0x1FBEF}, {0x1FBFA, 0x1FFFF}, {0x2A6DE, 0x2A6FF}, {0x2B735, 0x2B73F}, {0x2B81E, 0x2B81F}, {0x2CEA2, 0x2CEAF}, {0x2EBE1, 0x2F7FF}, {0x2FA1E, 0x2FFFF}, {0x3134B, 0xE00FF}, {0xE01F0, 0x10FFFF}, };
+{0x0, 0x8}, {0xE, 0x1B}, {0x7F, 0x84}, {0x86, 0x9F}, {0xAD, 0xAD}, {0x378, 0x379}, {0x380, 0x383}, {0x38B, 0x38B}, {0x38D, 0x38D}, {0x3A2, 0x3A2}, {0x530, 0x530}, {0x557, 0x558}, {0x58B, 0x58C},
+{0x590, 0x590}, {0x5C8, 0x5CF}, {0x5EB, 0x5EE}, {0x5F5, 0x605}, {0x61C, 0x61D}, {0x6DD, 0x6DD}, {0x70E, 0x70F}, {0x74B, 0x74C}, {0x7B2, 0x7BF}, {0x7FB, 0x7FC}, {0x82E, 0x82F}, {0x83F, 0x83F},
+{0x85C, 0x85D}, {0x85F, 0x85F}, {0x86B, 0x89F}, {0x8B5, 0x8B5}, {0x8C8, 0x8D2}, {0x8E2, 0x8E2}, {0x984, 0x984}, {0x98D, 0x98E}, {0x991, 0x992}, {0x9A9, 0x9A9}, {0x9B1, 0x9B1}, {0x9B3, 0x9B5},
+{0x9BA, 0x9BB}, {0x9C5, 0x9C6}, {0x9C9, 0x9CA}, {0x9CF, 0x9D6}, {0x9D8, 0x9DB}, {0x9DE, 0x9DE}, {0x9E4, 0x9E5}, {0x9FF, 0xA00}, {0xA04, 0xA04}, {0xA0B, 0xA0E}, {0xA11, 0xA12}, {0xA29, 0xA29},
+{0xA31, 0xA31}, {0xA34, 0xA34}, {0xA37, 0xA37}, {0xA3A, 0xA3B}, {0xA3D, 0xA3D}, {0xA43, 0xA46}, {0xA49, 0xA4A}, {0xA4E, 0xA50}, {0xA52, 0xA58}, {0xA5D, 0xA5D}, {0xA5F, 0xA65}, {0xA77, 0xA80},
+{0xA84, 0xA84}, {0xA8E, 0xA8E}, {0xA92, 0xA92}, {0xAA9, 0xAA9}, {0xAB1, 0xAB1}, {0xAB4, 0xAB4}, {0xABA, 0xABB}, {0xAC6, 0xAC6}, {0xACA, 0xACA}, {0xACE, 0xACF}, {0xAD1, 0xADF}, {0xAE4, 0xAE5},
+{0xAF2, 0xAF8}, {0xB00, 0xB00}, {0xB04, 0xB04}, {0xB0D, 0xB0E}, {0xB11, 0xB12}, {0xB29, 0xB29}, {0xB31, 0xB31}, {0xB34, 0xB34}, {0xB3A, 0xB3B}, {0xB45, 0xB46}, {0xB49, 0xB4A}, {0xB4E, 0xB54},
+{0xB58, 0xB5B}, {0xB5E, 0xB5E}, {0xB64, 0xB65}, {0xB78, 0xB81}, {0xB84, 0xB84}, {0xB8B, 0xB8D}, {0xB91, 0xB91}, {0xB96, 0xB98}, {0xB9B, 0xB9B}, {0xB9D, 0xB9D}, {0xBA0, 0xBA2}, {0xBA5, 0xBA7},
+{0xBAB, 0xBAD}, {0xBBA, 0xBBD}, {0xBC3, 0xBC5}, {0xBC9, 0xBC9}, {0xBCE, 0xBCF}, {0xBD1, 0xBD6}, {0xBD8, 0xBE5}, {0xBFB, 0xBFF}, {0xC0D, 0xC0D}, {0xC11, 0xC11}, {0xC29, 0xC29}, {0xC3A, 0xC3C},
+{0xC45, 0xC45}, {0xC49, 0xC49}, {0xC4E, 0xC54}, {0xC57, 0xC57}, {0xC5B, 0xC5F}, {0xC64, 0xC65}, {0xC70, 0xC76}, {0xC8D, 0xC8D}, {0xC91, 0xC91}, {0xCA9, 0xCA9}, {0xCB4, 0xCB4}, {0xCBA, 0xCBB},
+{0xCC5, 0xCC5}, {0xCC9, 0xCC9}, {0xCCE, 0xCD4}, {0xCD7, 0xCDD}, {0xCDF, 0xCDF}, {0xCE4, 0xCE5}, {0xCF0, 0xCF0}, {0xCF3, 0xCFF}, {0xD0D, 0xD0D}, {0xD11, 0xD11}, {0xD45, 0xD45}, {0xD49, 0xD49},
+{0xD50, 0xD53}, {0xD64, 0xD65}, {0xD80, 0xD80}, {0xD84, 0xD84}, {0xD97, 0xD99}, {0xDB2, 0xDB2}, {0xDBC, 0xDBC}, {0xDBE, 0xDBF}, {0xDC7, 0xDC9}, {0xDCB, 0xDCE}, {0xDD5, 0xDD5}, {0xDD7, 0xDD7},
+{0xDE0, 0xDE5}, {0xDF0, 0xDF1}, {0xDF5, 0xE00}, {0xE3B, 0xE3E}, {0xE5C, 0xE80}, {0xE83, 0xE83}, {0xE85, 0xE85}, {0xE8B, 0xE8B}, {0xEA4, 0xEA4}, {0xEA6, 0xEA6}, {0xEBE, 0xEBF}, {0xEC5, 0xEC5},
+{0xEC7, 0xEC7}, {0xECE, 0xECF}, {0xEDA, 0xEDB}, {0xEE0, 0xEFF}, {0xF48, 0xF48}, {0xF6D, 0xF70}, {0xF98, 0xF98}, {0xFBD, 0xFBD}, {0xFCD, 0xFCD}, {0xFDB, 0xFFF}, {0x10C6, 0x10C6}, {0x10C8, 0x10CC},
+{0x10CE, 0x10CF}, {0x1249, 0x1249}, {0x124E, 0x124F}, {0x1257, 0x1257}, {0x1259, 0x1259}, {0x125E, 0x125F}, {0x1289, 0x1289}, {0x128E, 0x128F}, {0x12B1, 0x12B1}, {0x12B6, 0x12B7}, {0x12BF, 0x12BF},
+{0x12C1, 0x12C1}, {0x12C6, 0x12C7}, {0x12D7, 0x12D7}, {0x1311, 0x1311}, {0x1316, 0x1317}, {0x135B, 0x135C}, {0x137D, 0x137F}, {0x139A, 0x139F}, {0x13F6, 0x13F7}, {0x13FE, 0x13FF}, {0x169D, 0x169F},
+{0x16F9, 0x16FF}, {0x170D, 0x170D}, {0x1715, 0x171F}, {0x1737, 0x173F}, {0x1754, 0x175F}, {0x176D, 0x176D}, {0x1771, 0x1771}, {0x1774, 0x177F}, {0x17DE, 0x17DF}, {0x17EA, 0x17EF}, {0x17FA, 0x17FF},
+{0x180E, 0x180F}, {0x181A, 0x181F}, {0x1879, 0x187F}, {0x18AB, 0x18AF}, {0x18F6, 0x18FF}, {0x191F, 0x191F}, {0x192C, 0x192F}, {0x193C, 0x193F}, {0x1941, 0x1943}, {0x196E, 0x196F}, {0x1975, 0x197F},
+{0x19AC, 0x19AF}, {0x19CA, 0x19CF}, {0x19DB, 0x19DD}, {0x1A1C, 0x1A1D}, {0x1A5F, 0x1A5F}, {0x1A7D, 0x1A7E}, {0x1A8A, 0x1A8F}, {0x1A9A, 0x1A9F}, {0x1AAE, 0x1AAF}, {0x1AC1, 0x1AFF}, {0x1B4C, 0x1B4F},
+{0x1B7D, 0x1B7F}, {0x1BF4, 0x1BFB}, {0x1C38, 0x1C3A}, {0x1C4A, 0x1C4C}, {0x1C89, 0x1C8F}, {0x1CBB, 0x1CBC}, {0x1CC8, 0x1CCF}, {0x1CFB, 0x1CFF}, {0x1DFA, 0x1DFA}, {0x1F16, 0x1F17}, {0x1F1E, 0x1F1F},
+{0x1F46, 0x1F47}, {0x1F4E, 0x1F4F}, {0x1F58, 0x1F58}, {0x1F5A, 0x1F5A}, {0x1F5C, 0x1F5C}, {0x1F5E, 0x1F5E}, {0x1F7E, 0x1F7F}, {0x1FB5, 0x1FB5}, {0x1FC5, 0x1FC5}, {0x1FD4, 0x1FD5}, {0x1FDC, 0x1FDC},
+{0x1FF0, 0x1FF1}, {0x1FF5, 0x1FF5}, {0x1FFF, 0x1FFF}, {0x200B, 0x200F}, {0x202A, 0x202E}, {0x2060, 0x206F}, {0x2072, 0x2073}, {0x208F, 0x208F}, {0x209D, 0x209F}, {0x20C0, 0x20CF}, {0x20F1, 0x20FF},
+{0x218C, 0x218F}, {0x2427, 0x243F}, {0x244B, 0x245F}, {0x2B74, 0x2B75}, {0x2B96, 0x2B96}, {0x2C2F, 0x2C2F}, {0x2C5F, 0x2C5F}, {0x2CF4, 0x2CF8}, {0x2D26, 0x2D26}, {0x2D28, 0x2D2C}, {0x2D2E, 0x2D2F},
+{0x2D68, 0x2D6E}, {0x2D71, 0x2D7E}, {0x2D97, 0x2D9F}, {0x2DA7, 0x2DA7}, {0x2DAF, 0x2DAF}, {0x2DB7, 0x2DB7}, {0x2DBF, 0x2DBF}, {0x2DC7, 0x2DC7}, {0x2DCF, 0x2DCF}, {0x2DD7, 0x2DD7}, {0x2DDF, 0x2DDF},
+{0x2E53, 0x2E7F}, {0x2E9A, 0x2E9A}, {0x2EF4, 0x2EFF}, {0x2FD6, 0x2FEF}, {0x2FFC, 0x2FFF}, {0x3040, 0x3040}, {0x3097, 0x3098}, {0x3100, 0x3104}, {0x3130, 0x3130}, {0x318F, 0x318F}, {0x31E4, 0x31EF},
+{0x321F, 0x321F}, {0x9FFD, 0x9FFF}, {0xA48D, 0xA48F}, {0xA4C7, 0xA4CF}, {0xA62C, 0xA63F}, {0xA6F8, 0xA6FF}, {0xA7C0, 0xA7C1}, {0xA7CB, 0xA7F4}, {0xA82D, 0xA82F}, {0xA83A, 0xA83F}, {0xA878, 0xA87F},
+{0xA8C6, 0xA8CD}, {0xA8DA, 0xA8DF}, {0xA954, 0xA95E}, {0xA97D, 0xA97F}, {0xA9CE, 0xA9CE}, {0xA9DA, 0xA9DD}, {0xA9FF, 0xA9FF}, {0xAA37, 0xAA3F}, {0xAA4E, 0xAA4F}, {0xAA5A, 0xAA5B}, {0xAAC3, 0xAADA},
+{0xAAF7, 0xAB00}, {0xAB07, 0xAB08}, {0xAB0F, 0xAB10}, {0xAB17, 0xAB1F}, {0xAB27, 0xAB27}, {0xAB2F, 0xAB2F}, {0xAB6C, 0xAB6F}, {0xABEE, 0xABEF}, {0xABFA, 0xABFF}, {0xD7A4, 0xD7AF}, {0xD7C7, 0xD7CA},
+{0xD7FC, 0xF8FF}, {0xFA6E, 0xFA6F}, {0xFADA, 0xFAFF}, {0xFB07, 0xFB12}, {0xFB18, 0xFB1C}, {0xFB37, 0xFB37}, {0xFB3D, 0xFB3D}, {0xFB3F, 0xFB3F}, {0xFB42, 0xFB42}, {0xFB45, 0xFB45}, {0xFBC2, 0xFBD2},
+{0xFD40, 0xFD4F}, {0xFD90, 0xFD91}, {0xFDC8, 0xFDEF}, {0xFDFE, 0xFDFF}, {0xFE1A, 0xFE1F}, {0xFE53, 0xFE53}, {0xFE67, 0xFE67}, {0xFE6C, 0xFE6F}, {0xFE75, 0xFE75}, {0xFEFD, 0xFF00}, {0xFFBF, 0xFFC1},
+{0xFFC8, 0xFFC9}, {0xFFD0, 0xFFD1}, {0xFFD8, 0xFFD9}, {0xFFDD, 0xFFDF}, {0xFFE7, 0xFFE7}, {0xFFEF, 0xFFFB}, {0xFFFE, 0xFFFF}, {0x1000C, 0x1000C}, {0x10027, 0x10027}, {0x1003B, 0x1003B},
+{0x1003E, 0x1003E}, {0x1004E, 0x1004F}, {0x1005E, 0x1007F}, {0x100FB, 0x100FF}, {0x10103, 0x10106}, {0x10134, 0x10136}, {0x1018F, 0x1018F}, {0x1019D, 0x1019F}, {0x101A1, 0x101CF}, {0x101FE, 0x1027F},
+{0x1029D, 0x1029F}, {0x102D1, 0x102DF}, {0x102FC, 0x102FF}, {0x10324, 0x1032C}, {0x1034B, 0x1034F}, {0x1037B, 0x1037F}, {0x1039E, 0x1039E}, {0x103C4, 0x103C7}, {0x103D6, 0x103FF}, {0x1049E, 0x1049F},
+{0x104AA, 0x104AF}, {0x104D4, 0x104D7}, {0x104FC, 0x104FF}, {0x10528, 0x1052F}, {0x10564, 0x1056E}, {0x10570, 0x105FF}, {0x10737, 0x1073F}, {0x10756, 0x1075F}, {0x10768, 0x107FF}, {0x10806, 0x10807},
+{0x10809, 0x10809}, {0x10836, 0x10836}, {0x10839, 0x1083B}, {0x1083D, 0x1083E}, {0x10856, 0x10856}, {0x1089F, 0x108A6}, {0x108B0, 0x108DF}, {0x108F3, 0x108F3}, {0x108F6, 0x108FA}, {0x1091C, 0x1091E},
+{0x1093A, 0x1093E}, {0x10940, 0x1097F}, {0x109B8, 0x109BB}, {0x109D0, 0x109D1}, {0x10A04, 0x10A04}, {0x10A07, 0x10A0B}, {0x10A14, 0x10A14}, {0x10A18, 0x10A18}, {0x10A36, 0x10A37}, {0x10A3B, 0x10A3E},
+{0x10A49, 0x10A4F}, {0x10A59, 0x10A5F}, {0x10AA0, 0x10ABF}, {0x10AE7, 0x10AEA}, {0x10AF7, 0x10AFF}, {0x10B36, 0x10B38}, {0x10B56, 0x10B57}, {0x10B73, 0x10B77}, {0x10B92, 0x10B98}, {0x10B9D, 0x10BA8},
+{0x10BB0, 0x10BFF}, {0x10C49, 0x10C7F}, {0x10CB3, 0x10CBF}, {0x10CF3, 0x10CF9}, {0x10D28, 0x10D2F}, {0x10D3A, 0x10E5F}, {0x10E7F, 0x10E7F}, {0x10EAA, 0x10EAA}, {0x10EAE, 0x10EAF}, {0x10EB2, 0x10EFF},
+{0x10F28, 0x10F2F}, {0x10F5A, 0x10FAF}, {0x10FCC, 0x10FDF}, {0x10FF7, 0x10FFF}, {0x1104E, 0x11051}, {0x11070, 0x1107E}, {0x110BD, 0x110BD}, {0x110C2, 0x110CF}, {0x110E9, 0x110EF}, {0x110FA, 0x110FF},
+{0x11135, 0x11135}, {0x11148, 0x1114F}, {0x11177, 0x1117F}, {0x111E0, 0x111E0}, {0x111F5, 0x111FF}, {0x11212, 0x11212}, {0x1123F, 0x1127F}, {0x11287, 0x11287}, {0x11289, 0x11289}, {0x1128E, 0x1128E},
+{0x1129E, 0x1129E}, {0x112AA, 0x112AF}, {0x112EB, 0x112EF}, {0x112FA, 0x112FF}, {0x11304, 0x11304}, {0x1130D, 0x1130E}, {0x11311, 0x11312}, {0x11329, 0x11329}, {0x11331, 0x11331}, {0x11334, 0x11334},
+{0x1133A, 0x1133A}, {0x11345, 0x11346}, {0x11349, 0x1134A}, {0x1134E, 0x1134F}, {0x11351, 0x11356}, {0x11358, 0x1135C}, {0x11364, 0x11365}, {0x1136D, 0x1136F}, {0x11375, 0x113FF}, {0x1145C, 0x1145C},
+{0x11462, 0x1147F}, {0x114C8, 0x114CF}, {0x114DA, 0x1157F}, {0x115B6, 0x115B7}, {0x115DE, 0x115FF}, {0x11645, 0x1164F}, {0x1165A, 0x1165F}, {0x1166D, 0x1167F}, {0x116B9, 0x116BF}, {0x116CA, 0x116FF},
+{0x1171B, 0x1171C}, {0x1172C, 0x1172F}, {0x11740, 0x117FF}, {0x1183C, 0x1189F}, {0x118F3, 0x118FE}, {0x11907, 0x11908}, {0x1190A, 0x1190B}, {0x11914, 0x11914}, {0x11917, 0x11917}, {0x11936, 0x11936},
+{0x11939, 0x1193A}, {0x11947, 0x1194F}, {0x1195A, 0x1199F}, {0x119A8, 0x119A9}, {0x119D8, 0x119D9}, {0x119E5, 0x119FF}, {0x11A48, 0x11A4F}, {0x11AA3, 0x11ABF}, {0x11AF9, 0x11BFF}, {0x11C09, 0x11C09},
+{0x11C37, 0x11C37}, {0x11C46, 0x11C4F}, {0x11C6D, 0x11C6F}, {0x11C90, 0x11C91}, {0x11CA8, 0x11CA8}, {0x11CB7, 0x11CFF}, {0x11D07, 0x11D07}, {0x11D0A, 0x11D0A}, {0x11D37, 0x11D39}, {0x11D3B, 0x11D3B},
+{0x11D3E, 0x11D3E}, {0x11D48, 0x11D4F}, {0x11D5A, 0x11D5F}, {0x11D66, 0x11D66}, {0x11D69, 0x11D69}, {0x11D8F, 0x11D8F}, {0x11D92, 0x11D92}, {0x11D99, 0x11D9F}, {0x11DAA, 0x11EDF}, {0x11EF9, 0x11FAF},
+{0x11FB1, 0x11FBF}, {0x11FF2, 0x11FFE}, {0x1239A, 0x123FF}, {0x1246F, 0x1246F}, {0x12475, 0x1247F}, {0x12544, 0x12FFF}, {0x1342F, 0x143FF}, {0x14647, 0x167FF}, {0x16A39, 0x16A3F}, {0x16A5F, 0x16A5F},
+{0x16A6A, 0x16A6D}, {0x16A70, 0x16ACF}, {0x16AEE, 0x16AEF}, {0x16AF6, 0x16AFF}, {0x16B46, 0x16B4F}, {0x16B5A, 0x16B5A}, {0x16B62, 0x16B62}, {0x16B78, 0x16B7C}, {0x16B90, 0x16E3F}, {0x16E9B, 0x16EFF},
+{0x16F4B, 0x16F4E}, {0x16F88, 0x16F8E}, {0x16FA0, 0x16FDF}, {0x16FE5, 0x16FEF}, {0x16FF2, 0x16FFF}, {0x187F8, 0x187FF}, {0x18CD6, 0x18CFF}, {0x18D09, 0x1AFFF}, {0x1B11F, 0x1B14F}, {0x1B153, 0x1B163},
+{0x1B168, 0x1B16F}, {0x1B2FC, 0x1BBFF}, {0x1BC6B, 0x1BC6F}, {0x1BC7D, 0x1BC7F}, {0x1BC89, 0x1BC8F}, {0x1BC9A, 0x1BC9B}, {0x1BCA0, 0x1CFFF}, {0x1D0F6, 0x1D0FF}, {0x1D127, 0x1D128}, {0x1D173, 0x1D17A},
+{0x1D1E9, 0x1D1FF}, {0x1D246, 0x1D2DF}, {0x1D2F4, 0x1D2FF}, {0x1D357, 0x1D35F}, {0x1D379, 0x1D3FF}, {0x1D455, 0x1D455}, {0x1D49D, 0x1D49D}, {0x1D4A0, 0x1D4A1}, {0x1D4A3, 0x1D4A4}, {0x1D4A7, 0x1D4A8},
+{0x1D4AD, 0x1D4AD}, {0x1D4BA, 0x1D4BA}, {0x1D4BC, 0x1D4BC}, {0x1D4C4, 0x1D4C4}, {0x1D506, 0x1D506}, {0x1D50B, 0x1D50C}, {0x1D515, 0x1D515}, {0x1D51D, 0x1D51D}, {0x1D53A, 0x1D53A}, {0x1D53F, 0x1D53F},
+{0x1D545, 0x1D545}, {0x1D547, 0x1D549}, {0x1D551, 0x1D551}, {0x1D6A6, 0x1D6A7}, {0x1D7CC, 0x1D7CD}, {0x1DA8C, 0x1DA9A}, {0x1DAA0, 0x1DAA0}, {0x1DAB0, 0x1DFFF}, {0x1E007, 0x1E007}, {0x1E019, 0x1E01A},
+{0x1E022, 0x1E022}, {0x1E025, 0x1E025}, {0x1E02B, 0x1E0FF}, {0x1E12D, 0x1E12F}, {0x1E13E, 0x1E13F}, {0x1E14A, 0x1E14D}, {0x1E150, 0x1E2BF}, {0x1E2FA, 0x1E2FE}, {0x1E300, 0x1E7FF}, {0x1E8C5, 0x1E8C6},
+{0x1E8D7, 0x1E8FF}, {0x1E94C, 0x1E94F}, {0x1E95A, 0x1E95D}, {0x1E960, 0x1EC70}, {0x1ECB5, 0x1ED00}, {0x1ED3E, 0x1EDFF}, {0x1EE04, 0x1EE04}, {0x1EE20, 0x1EE20}, {0x1EE23, 0x1EE23}, {0x1EE25, 0x1EE26},
+{0x1EE28, 0x1EE28}, {0x1EE33, 0x1EE33}, {0x1EE38, 0x1EE38}, {0x1EE3A, 0x1EE3A}, {0x1EE3C, 0x1EE41}, {0x1EE43, 0x1EE46}, {0x1EE48, 0x1EE48}, {0x1EE4A, 0x1EE4A}, {0x1EE4C, 0x1EE4C}, {0x1EE50, 0x1EE50},
+{0x1EE53, 0x1EE53}, {0x1EE55, 0x1EE56}, {0x1EE58, 0x1EE58}, {0x1EE5A, 0x1EE5A}, {0x1EE5C, 0x1EE5C}, {0x1EE5E, 0x1EE5E}, {0x1EE60, 0x1EE60}, {0x1EE63, 0x1EE63}, {0x1EE65, 0x1EE66}, {0x1EE6B, 0x1EE6B},
+{0x1EE73, 0x1EE73}, {0x1EE78, 0x1EE78}, {0x1EE7D, 0x1EE7D}, {0x1EE7F, 0x1EE7F}, {0x1EE8A, 0x1EE8A}, {0x1EE9C, 0x1EEA0}, {0x1EEA4, 0x1EEA4}, {0x1EEAA, 0x1EEAA}, {0x1EEBC, 0x1EEEF}, {0x1EEF2, 0x1EFFF},
+{0x1F02C, 0x1F02F}, {0x1F094, 0x1F09F}, {0x1F0AF, 0x1F0B0}, {0x1F0C0, 0x1F0C0}, {0x1F0D0, 0x1F0D0}, {0x1F0F6, 0x1F0FF}, {0x1F1AE, 0x1F1E5}, {0x1F203, 0x1F20F}, {0x1F23C, 0x1F23F}, {0x1F249, 0x1F24F},
+{0x1F252, 0x1F25F}, {0x1F266, 0x1F2FF}, {0x1F6D8, 0x1F6DF}, {0x1F6ED, 0x1F6EF}, {0x1F6FD, 0x1F6FF}, {0x1F774, 0x1F77F}, {0x1F7D9, 0x1F7DF}, {0x1F7EC, 0x1F7FF}, {0x1F80C, 0x1F80F}, {0x1F848, 0x1F84F},
+{0x1F85A, 0x1F85F}, {0x1F888, 0x1F88F}, {0x1F8AE, 0x1F8AF}, {0x1F8B2, 0x1F8FF}, {0x1F979, 0x1F979}, {0x1F9CC, 0x1F9CC}, {0x1FA54, 0x1FA5F}, {0x1FA6E, 0x1FA6F}, {0x1FA75, 0x1FA77}, {0x1FA7B, 0x1FA7F},
+{0x1FA87, 0x1FA8F}, {0x1FAA9, 0x1FAAF}, {0x1FAB7, 0x1FABF}, {0x1FAC3, 0x1FACF}, {0x1FAD7, 0x1FAFF}, {0x1FB93, 0x1FB93}, {0x1FBCB, 0x1FBEF}, {0x1FBFA, 0x1FFFF}, {0x2A6DE, 0x2A6FF}, {0x2B735, 0x2B73F},
+{0x2B81E, 0x2B81F}, {0x2CEA2, 0x2CEAF}, {0x2EBE1, 0x2F7FF}, {0x2FA1E, 0x2FFFF}, {0x3134B, 0xE00FF}, {0xE01F0, 0x10FFFF},
+};
 
 //String
 bool CNCTString::operator==(const std::string& other) const {
@@ -136,8 +334,7 @@ struct CRCompare {
 };
 
 // binary search for code range
-bool CNCTUnicode::check_code_range(int c, const std::vector<std::pair<int, int>> &ranges)
-{
+bool CNCTUnicode::check_code_range(int c, const std::vector<std::pair<int, int>> &ranges) {
     auto it = std::upper_bound(ranges.begin(), ranges.end(), c, CRCompare());
     if (it != ranges.begin()) {
         --it;
@@ -146,22 +343,28 @@ bool CNCTUnicode::check_code_range(int c, const std::vector<std::pair<int, int>>
 }
 
 // these are binary searches, it takes only a few operations
-CNCTCharType CNCTUnicode::get_code_type(int c)
-{
-    if (check_code_range(c, letter_ranges))
+CNCTCharType CNCTUnicode::get_code_type(int c) {
+    if (check_code_range(c, letter_ranges)) {
         return LETTER;
-    if (check_code_range(c, digit_ranges))
+    }
+    if (check_code_range(c, digit_ranges)) {
         return DIGIT;
-    if (check_code_range(c, whitespace_ranges))
+    }
+    if (check_code_range(c, whitespace_ranges)) {
         return WHITESPACE;
-    if (check_code_range(c, punctuation_ranges))
+    }
+    if (check_code_range(c, punctuation_ranges)) {
         return PUNCTUATION;
-    if (check_code_range(c, symbol_ranges))
+    }
+    if (check_code_range(c, symbol_ranges)) {
         return SYMBOL;
-    if (check_code_range(c, accent_mark_ranges))
+    }
+    if (check_code_range(c, accent_mark_ranges)) {
         return ACCENT_MARK;
-    if (check_code_range(c, control_ranges))
+    }
+    if (check_code_range(c, control_ranges)) {
         return CONTROL;
+    }
     return UNIDENTIFIED;
 }
 
@@ -180,34 +383,34 @@ static int utf8_to_unicode(const std::string& utf8_char) {
     return c;
 }
 
-CNCTCharType CNCTUnicode::get_code_type(const std::string &utf8_char)
-{
+CNCTCharType CNCTUnicode::get_code_type(const std::string &utf8_char) {
     return get_code_type(utf8_to_unicode(utf8_char));
 }
 
 int CNCTUnicode::utf8_len(const char c)
 {
-    if ((c & 0x80) == 0)
+    if ((c & 0x80) == 0) {
         return 1; // ASCII character
-    if ((c & 0xE0) == 0xC0)
+    }
+    if ((c & 0xE0) == 0xC0) {
         return 2; // 2-byte character
-    if ((c & 0xF0) == 0xE0)
+    }
+    if ((c & 0xF0) == 0xE0) {
         return 3; // 3-byte character
-    if ((c & 0xF0) == 0xF0)
+    }
+    if ((c & 0xF0) == 0xF0) {
         return 4; // 4-byte character
+    }
     return 1;     // not valid utf8
     // static const uint8_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
     // return lookup[static_cast<uint8_t>(c) >> 4];
 }
 
-int CNCTUnicode::strlen_utf8(const std::string src)
-{
+int CNCTUnicode::strlen_utf8(const std::string src) {
     int len = 0;
-    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
-    {
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it) {
         int char_len = utf8_len(*it);
-        if (char_len > 1)
-        {
+        if (char_len > 1) {
             it += char_len - 1;
         }
         len += 1;
@@ -216,16 +419,13 @@ int CNCTUnicode::strlen_utf8(const std::string src)
 }
 
 // split a string into unicode strings
-std::vector<std::string> CNCTUnicode::split_utf8(const std::string &src)
-{
+std::vector<std::string> CNCTUnicode::split_utf8(const std::string &src) {
     std::vector<std::string> result;
-    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
-    {
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it) {
         int char_len = utf8_len(*it);
         std::string str(it, it + char_len);
         result.push_back(str);
-        if (char_len > 1)
-        {
+        if (char_len > 1) {
             it += char_len - 1;
         }
     }
@@ -233,13 +433,11 @@ std::vector<std::string> CNCTUnicode::split_utf8(const std::string &src)
 }
 
 // split a string into unicode strings (CNCTString) with sequence information
-std::vector<CNCTString> CNCTUnicode::split_utf8_enhanced(const std::string &src)
-{
+std::vector<CNCTString> CNCTUnicode::split_utf8_enhanced(const std::string &src) {
     std::vector<CNCTString> result;
     int seq_offset_bytes=0;
     int seq_offset_utf8_chars=0;
-    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it)
-    {
+    for (std::string::const_iterator it = src.begin(); it != src.end(); ++it) {
         int char_len = utf8_len(*it);
         std::string str(it, it + char_len);
         CNCTString cnct_str;
@@ -275,12 +473,10 @@ std::vector<CNCTString> CNCTUnicode::split_utf8_enhanced(const std::string &src)
         }
         #endif
 
-
         result.push_back(cnct_str);
         seq_offset_bytes += char_len;
         seq_offset_utf8_chars += 1;
-        if (char_len > 1)
-        {
+        if (char_len > 1) {
             it += char_len - 1;
         }
 
@@ -289,57 +485,41 @@ std::vector<CNCTString> CNCTUnicode::split_utf8_enhanced(const std::string &src)
 }
 
 // return the type of the string
-CNCTCharType CNCTUnicode::string_identify(const std::string &str)
-{
+CNCTCharType CNCTUnicode::string_identify(const std::string &str) {
     CNCTCharType result = UNIDENTIFIED;
     std::string::const_iterator it = str.begin();
-    while (it != str.end())
-    {
+    while (it != str.end()) {
         int len = utf8_len(*it);
         int c = 0;
-        for (int i = 0; i < len && it != str.end(); ++i, ++it)
-        {
+        for (int i = 0; i < len && it != str.end(); ++i, ++it) {
             c = (c << 8) | static_cast<unsigned char>(*it);
         }
-        switch (get_code_type(c))
-        {
+        switch (get_code_type(c)) {
         case DIGIT:
-            if (result == UNIDENTIFIED)
-            {
+            if (result == UNIDENTIFIED) {
                 result = DIGIT;
-            }
-            else if (result != DIGIT)
-            {
+            } else if (result != DIGIT) {
                 return MIXED;
             }
             break;
         case LETTER:
-            if (result == UNIDENTIFIED)
-            {
+            if (result == UNIDENTIFIED) {
                 result = LETTER;
-            }
-            else if (result != LETTER)
-            {
+            } else if (result != LETTER) {
                 return MIXED;
             }
             break;
         case WHITESPACE:
-            if (result == UNIDENTIFIED)
-            {
+            if (result == UNIDENTIFIED) {
                 result = WHITESPACE;
-            }
-            else if (result != WHITESPACE)
-            {
+            } else if (result != WHITESPACE) {
                 return MIXED;
             }
             break;
         case PUNCTUATION:
-            if (result == UNIDENTIFIED)
-            {
+            if (result == UNIDENTIFIED) {
                 result = PUNCTUATION;
-            }
-            else if (result != PUNCTUATION)
-            {
+            } else if (result != PUNCTUATION) {
                 return MIXED;
             }
             break;
@@ -355,23 +535,22 @@ CNCTCharType CNCTUnicode::string_identify(const std::string &str)
 bool CNCTUnicode::string_test(const std::string &str, CNCTCharType chartype)
 {
     std::string::const_iterator it = str.begin();
-    while (it != str.end())
-    {
+    while (it != str.end()) {
         int len = utf8_len(*it);
         int c = 0;
-        for (int i = 0; i < len && it != str.end(); ++i, ++it)
-        {
+        for (int i = 0; i < len && it != str.end(); ++i, ++it) {
             c = (c << 8) | static_cast<unsigned char>(*it);
         }
-        if (get_code_type(c) != chartype)
-        {
+        if (get_code_type(c) != chartype) {
             return false;
         }
     }
     return true;
 }
 
+//-----------------
 // llama.cpp GPT2 vocab (from libfalcon.cpp)
+//-----------------
 
 std::string replaceAll(std::string str, const std::string& from, const std::string& to) {
     size_t start_pos = 0;
@@ -391,11 +570,13 @@ struct Trie {
     TrieNode *root;
 
     Trie() : root(new TrieNode()) {}
+
     ~Trie() {
         if(root)
         deleteTrie(root);
     }
-        // Move constructor
+
+    // Move constructor
     Trie(Trie&& other) noexcept : root(other.root) {
         other.root = nullptr;
     }
@@ -414,15 +595,15 @@ struct Trie {
     void insert(const std::string &token, int32_t Id) {
         TrieNode* current = root;
         for(auto ch : token) {
-            if(current->map.find(ch) == current->map.end())
+            if(current->map.find(ch) == current->map.end()) {
                 current->map[ch] = new TrieNode();
+            }
             current = current->map[ch];
         }
         current->Id = Id;
     }
 
-    void reset()
-    {
+    void reset() {
         deleteTrie(root);
         root = new TrieNode();
     }
@@ -463,8 +644,7 @@ struct gpt2bpe_vocab {
         trie.reset();
         for (const auto& pair : token_to_id) {
             trie.insert(pair.first, pair.second);
-            if (pair.first.size() >= 2)
-            {
+            if (pair.first.size() >= 2) {
                 std::string prefix = pair.first.substr(0, 2);
                 max_token_length[prefix] = std::max(max_token_length[prefix], (uint32_t)pair.first.size());
             }
@@ -494,17 +674,18 @@ struct gpt2bpe_vocab {
 
     // get max token length available for a prefix of 2 bytes (string at least 2 bytes long)
     int get_max_token_length(const std::string& string) const {
-        if (string.size() < 2)
+        if (string.size() < 2) {
             return -1;
+        }
         std::string prefix = string.substr(0, 2);
-        if (max_token_length.find(prefix) == max_token_length.end())
+        if (max_token_length.find(prefix) == max_token_length.end()) {
             return 0;
+        }
         return max_token_length.at(prefix);
     }
 
     // function to find if two tokens match in bpe_rank, return rank or -1
-    int find_bpe_rank(const std::string& token1, const std::string& token2) const
-    {
+    int find_bpe_rank(const std::string& token1, const std::string& token2) const {
         std::string left_token = token1;
         std::string right_token = token2;
         left_token = replaceAll(left_token, " ", "Ġ");
@@ -513,8 +694,9 @@ struct gpt2bpe_vocab {
         right_token = replaceAll(right_token, "\n", "Ċ");
 
         auto it = bpe_ranks.find(std::make_pair(left_token, right_token));
-        if (it == bpe_ranks.end())
+        if (it == bpe_ranks.end()) {
             return -1;
+        }
         return it->second;
     }
 
@@ -555,10 +737,8 @@ struct ggllm_bpe_symbol {
 static_assert(std::is_trivially_copyable<ggllm_bpe_symbol>::value, "ggllm_bpe_symbol is not trivially copyable");
 
 struct ggllm_bpe_bigram {
-    struct comparator
-    {
-        bool operator()(ggllm_bpe_bigram & l, ggllm_bpe_bigram & r)
-        {
+    struct comparator {
+        bool operator()(ggllm_bpe_bigram & l, ggllm_bpe_bigram & r) {
             return l.rank > r.rank || (l.rank == r.rank && l.left > r.left);
         }
     };
@@ -582,19 +762,16 @@ struct gpt2bpe_tokenizer {
         // auto end = ggml_time_us();
         // fprintf(stderr, "%s: preprocessing took %0.3f ms\n", __func__, (end - start) / 1000.0);
 
-
         symbols_final.clear();
 
-        for (auto & word : word_collection)
-        {
+        for (auto & word : word_collection) {
             work_queue_ = ggllm_bpe_bigram::queue();
             symbols_.clear();
 
             int index = 0;
             size_t offset = 0;
 
-            while (offset < word.size())
-            {
+            while (offset < word.size()) {
                 ggllm_bpe_symbol sym;
                 size_t char_len = std::min(word.size() - offset, (size_t) CNCTUnicode::utf8_len(word[offset]));
                 sym.text = word.c_str() + offset;
@@ -611,8 +788,7 @@ struct gpt2bpe_tokenizer {
             }
 
             // build token(s)
-            while (!work_queue_.empty())
-            {
+            while (!work_queue_.empty()) {
                 auto bigram = work_queue_.top();
                 work_queue_.pop();
 
@@ -641,16 +817,13 @@ struct gpt2bpe_tokenizer {
                 add_new_bigram(left_symbol.prev, bigram.left);  // left side of current symbol
                 add_new_bigram(bigram.left, left_symbol.next);  // right side of current symbol
             }
-            // add the fnished tokens to the final list keeping correct order for next and prev
 
-            for (auto & sym : symbols_)
-            {
-                if (sym.n > 0)
-                {
+            // add the fnished tokens to the final list keeping correct order for next and prev
+            for (auto & sym : symbols_) {
+                if (sym.n > 0) {
                     sym.prev = final_prev_index;
                     sym.next = -1;
-                    if (final_prev_index != -1)
-                    {
+                    if (final_prev_index != -1) {
                         symbols_final[final_prev_index].next = symbols_final.size();
                     }
                     symbols_final.emplace_back(sym);
@@ -659,7 +832,6 @@ struct gpt2bpe_tokenizer {
             }
         }
 
-
         symbols_ = symbols_final;
         if (symbols_.size())
         for (int i = 0; i != -1; i = symbols_[i].next) {
@@ -687,8 +859,7 @@ struct gpt2bpe_tokenizer {
     }
 
 private:
-    void add_new_bigram(int left, int right)
-    {
+    void add_new_bigram(int left, int right) {
         if (left == -1 || right == -1)  return;
 
         std::string left_token = std::string(symbols_[left].text, symbols_[left].n);
@@ -711,12 +882,74 @@ struct gpt2bpe_tokenizer {
     }
 
     std::unordered_map<unsigned char, std::string> bytes_to_unicode() {
-        static std::unordered_map<unsigned char, std::string> hex_map = { { 0x21, "\x21" }, { 0x22, "\x22" }, { 0x23, "\x23" }, { 0x24, "\x24" }, { 0x25, "\x25" }, { 0x26, "\x26" }, { 0x27, "\x27" }, { 0x28, "\x28" }, { 0x29, "\x29" }, { 0x2A, "\x2A" }, { 0x2B, "\x2B" }, { 0x2C, "\x2C" }, { 0x2D, "\x2D" }, { 0x2E, "\x2E" }, { 0x2F, "\x2F" }, { 0x30, "\x30" }, { 0x31, "\x31" }, { 0x32, "\x32" }, { 0x33, "\x33" }, { 0x34, "\x34" }, { 0x35, "\x35" }, { 0x36, "\x36" }, { 0x37, "\x37" }, { 0x38, "\x38" }, { 0x39, "\x39" }, { 0x3A, "\x3A" }, { 0x3B, "\x3B" }, { 0x3C, "\x3C" }, { 0x3D, "\x3D" }, { 0x3E, "\x3E" }, { 0x3F, "\x3F" }, { 0x40, "\x40" }, { 0x41, "\x41" }, { 0x42, "\x42" }, { 0x43, "\x43" }, { 0x44, "\x44" }, { 0x45, "\x45" }, { 0x46, "\x46" }, { 0x47, "\x47" }, { 0x48, "\x48" }, { 0x49, "\x49" }, { 0x4A, "\x4A" }, { 0x4B, "\x4B" }, { 0x4C, "\x4C" }, { 0x4D, "\x4D" }, { 0x4E, "\x4E" }, { 0x4F, "\x4F" }, { 0x50, "\x50" }, { 0x51, "\x51" }, { 0x52, "\x52" }, { 0x53, "\x53" }, { 0x54, "\x54" }, { 0x55, "\x55" }, { 0x56, "\x56" }, { 0x57, "\x57" }, { 0x58, "\x58" }, { 0x59, "\x59" }, { 0x5A, "\x5A" }, { 0x5B, "\x5B" }, { 0x5C, "\x5C" }, { 0x5D, "\x5D" }, { 0x5E, "\x5E" }, { 0x5F, "\x5F" }, { 0x60, "\x60" }, { 0x61, "\x61" }, { 0x62, "\x62" }, { 0x63, "\x63" }, { 0x64, "\x64" }, { 0x65, "\x65" }, { 0x66, "\x66" }, { 0x67, "\x67" }, { 0x68, "\x68" }, { 0x69, "\x69" }, { 0x6A, "\x6A" }, { 0x6B, "\x6B" }, { 0x6C, "\x6C" }, { 0x6D, "\x6D" }, { 0x6E, "\x6E" }, { 0x6F, "\x6F" }, { 0x70, "\x70" }, { 0x71, "\x71" }, { 0x72, "\x72" }, { 0x73, "\x73" }, { 0x74, "\x74" }, { 0x75, "\x75" }, { 0x76, "\x76" }, { 0x77, "\x77" }, { 0x78, "\x78" }, { 0x79, "\x79" }, { 0x7A, "\x7A" }, { 0x7B, "\x7B" }, { 0x7C, "\x7C" }, { 0x7D, "\x7D" }, { 0x7E, "\x7E" }, { 0xA1, "\xC2\xA1" }, { 0xA2, "\xC2\xA2" }, { 0xA3, "\xC2\xA3" }, { 0xA4, "\xC2\xA4" }, { 0xA5, "\xC2\xA5" }, { 0xA6, "\xC2\xA6" }, { 0xA7, "\xC2\xA7" }, { 0xA8, "\xC2\xA8" }, { 0xA9, "\xC2\xA9" }, { 0xAA, "\xC2\xAA" }, { 0xAB, "\xC2\xAB" }, { 0xAC, "\xC2\xAC" }, { 0xAE, "\xC2\xAE" }, { 0xAF, "\xC2\xAF" }, { 0xB0, "\xC2\xB0" }, { 0xB1, "\xC2\xB1" }, { 0xB2, "\xC2\xB2" }, { 0xB3, "\xC2\xB3" }, { 0xB4, "\xC2\xB4" }, { 0xB5, "\xC2\xB5" }, { 0xB6, "\xC2\xB6" }, { 0xB7, "\xC2\xB7" }, { 0xB8, "\xC2\xB8" }, { 0xB9, "\xC2\xB9" }, { 0xBA, "\xC2\xBA" }, { 0xBB, "\xC2\xBB" }, { 0xBC, "\xC2\xBC" }, { 0xBD, "\xC2\xBD" }, { 0xBE, "\xC2\xBE" }, { 0xBF, "\xC2\xBF" }, { 0xC0, "\xC3\x80" }, { 0xC1, "\xC3\x81" }, { 0xC2, "\xC3\x82" }, { 0xC3, "\xC3\x83" }, { 0xC4, "\xC3\x84" }, { 0xC5, "\xC3\x85" }, { 0xC6, "\xC3\x86" }, { 0xC7, "\xC3\x87" }, { 0xC8, "\xC3\x88" }, { 0xC9, "\xC3\x89" }, { 0xCA, "\xC3\x8A" }, { 0xCB, "\xC3\x8B" }, { 0xCC, "\xC3\x8C" }, { 0xCD, "\xC3\x8D" }, { 0xCE, "\xC3\x8E" }, { 0xCF, "\xC3\x8F" }, { 0xD0, "\xC3\x90" }, { 0xD1, "\xC3\x91" }, { 0xD2, "\xC3\x92" }, { 0xD3, "\xC3\x93" }, { 0xD4, "\xC3\x94" }, { 0xD5, "\xC3\x95" }, { 0xD6, "\xC3\x96" }, { 0xD7, "\xC3\x97" }, { 0xD8, "\xC3\x98" }, { 0xD9, "\xC3\x99" }, { 0xDA, "\xC3\x9A" }, { 0xDB, "\xC3\x9B" }, { 0xDC, "\xC3\x9C" }, { 0xDD, "\xC3\x9D" }, { 0xDE, "\xC3\x9E" }, { 0xDF, "\xC3\x9F" }, { 0xE0, "\xC3\xA0" }, { 0xE1, "\xC3\xA1" }, { 0xE2, "\xC3\xA2" }, { 0xE3, "\xC3\xA3" }, { 0xE4, "\xC3\xA4" }, { 0xE5, "\xC3\xA5" }, { 0xE6, "\xC3\xA6" }, { 0xE7, "\xC3\xA7" }, { 0xE8, "\xC3\xA8" }, { 0xE9, "\xC3\xA9" }, { 0xEA, "\xC3\xAA" }, { 0xEB, "\xC3\xAB" }, { 0xEC, "\xC3\xAC" }, { 0xED, "\xC3\xAD" }, { 0xEE, "\xC3\xAE" }, { 0xEF, "\xC3\xAF" }, { 0xF0, "\xC3\xB0" }, { 0xF1, "\xC3\xB1" }, { 0xF2, "\xC3\xB2" }, { 0xF3, "\xC3\xB3" }, { 0xF4, "\xC3\xB4" }, { 0xF5, "\xC3\xB5" }, { 0xF6, "\xC3\xB6" }, { 0xF7, "\xC3\xB7" }, { 0xF8, "\xC3\xB8" }, { 0xF9, "\xC3\xB9" }, { 0xFA, "\xC3\xBA" }, { 0xFB, "\xC3\xBB" }, { 0xFC, "\xC3\xBC" }, { 0xFD, "\xC3\xBD" }, { 0xFE, "\xC3\xBE" }, { 0xFF, "\xC3\xBF" }, { 0x00, "\xC4\x80" }, { 0x01, "\xC4\x81" }, { 0x02, "\xC4\x82" }, { 0x03, "\xC4\x83" }, { 0x04, "\xC4\x84" }, { 0x05, "\xC4\x85" }, { 0x06, "\xC4\x86" }, { 0x07, "\xC4\x87" }, { 0x08, "\xC4\x88" }, { 0x09, "\xC4\x89" }, { 0x0A, "\xC4\x8A" }, { 0x0B, "\xC4\x8B" }, { 0x0C, "\xC4\x8C" }, { 0x0D, "\xC4\x8D" }, { 0x0E, "\xC4\x8E" }, { 0x0F, "\xC4\x8F" }, { 0x10, "\xC4\x90" }, { 0x11, "\xC4\x91" }, { 0x12, "\xC4\x92" }, { 0x13, "\xC4\x93" }, { 0x14, "\xC4\x94" }, { 0x15, "\xC4\x95" }, { 0x16, "\xC4\x96" }, { 0x17, "\xC4\x97" }, { 0x18, "\xC4\x98" }, { 0x19, "\xC4\x99" }, { 0x1A, "\xC4\x9A" }, { 0x1B, "\xC4\x9B" }, { 0x1C, "\xC4\x9C" }, { 0x1D, "\xC4\x9D" }, { 0x1E, "\xC4\x9E" }, { 0x1F, "\xC4\x9F" }, { 0x20, "\xC4\xA0" }, { 0x7F, "\xC4\xA1" }, { 0x80, "\xC4\xA2" }, { 0x81, "\xC4\xA3" }, { 0x82, "\xC4\xA4" }, { 0x83, "\xC4\xA5" }, { 0x84, "\xC4\xA6" }, { 0x85, "\xC4\xA7" }, { 0x86, "\xC4\xA8" }, { 0x87, "\xC4\xA9" }, { 0x88, "\xC4\xAA" }, { 0x89, "\xC4\xAB" }, { 0x8A, "\xC4\xAC" }, { 0x8B, "\xC4\xAD" }, { 0x8C, "\xC4\xAE" }, { 0x8D, "\xC4\xAF" }, { 0x8E, "\xC4\xB0" }, { 0x8F, "\xC4\xB1" }, { 0x90, "\xC4\xB2" }, { 0x91, "\xC4\xB3" }, { 0x92, "\xC4\xB4" }, { 0x93, "\xC4\xB5" }, { 0x94, "\xC4\xB6" }, { 0x95, "\xC4\xB7" }, { 0x96, "\xC4\xB8" }, { 0x97, "\xC4\xB9" }, { 0x98, "\xC4\xBA" }, { 0x99, "\xC4\xBB" }, { 0x9A, "\xC4\xBC" }, { 0x9B, "\xC4\xBD" }, { 0x9C, "\xC4\xBE" }, { 0x9D, "\xC4\xBF" }, { 0x9E, "\xC5\x80" }, { 0x9F, "\xC5\x81" }, { 0xA0, "\xC5\x82" }, { 0xAD, "\xC5\x83" }};
+        static std::unordered_map<unsigned char, std::string> hex_map = {
+            { 0x21, "\x21" }, { 0x22, "\x22" }, { 0x23, "\x23" }, { 0x24, "\x24" }, { 0x25, "\x25" }, { 0x26, "\x26" }, { 0x27, "\x27" }, { 0x28, "\x28" }, { 0x29, "\x29" }, { 0x2A, "\x2A" },
+            { 0x2B, "\x2B" }, { 0x2C, "\x2C" }, { 0x2D, "\x2D" }, { 0x2E, "\x2E" }, { 0x2F, "\x2F" }, { 0x30, "\x30" }, { 0x31, "\x31" }, { 0x32, "\x32" }, { 0x33, "\x33" }, { 0x34, "\x34" },
+            { 0x35, "\x35" }, { 0x36, "\x36" }, { 0x37, "\x37" }, { 0x38, "\x38" }, { 0x39, "\x39" }, { 0x3A, "\x3A" }, { 0x3B, "\x3B" }, { 0x3C, "\x3C" }, { 0x3D, "\x3D" }, { 0x3E, "\x3E" },
+            { 0x3F, "\x3F" }, { 0x40, "\x40" }, { 0x41, "\x41" }, { 0x42, "\x42" }, { 0x43, "\x43" }, { 0x44, "\x44" }, { 0x45, "\x45" }, { 0x46, "\x46" }, { 0x47, "\x47" }, { 0x48, "\x48" },
+            { 0x49, "\x49" }, { 0x4A, "\x4A" }, { 0x4B, "\x4B" }, { 0x4C, "\x4C" }, { 0x4D, "\x4D" }, { 0x4E, "\x4E" }, { 0x4F, "\x4F" }, { 0x50, "\x50" }, { 0x51, "\x51" }, { 0x52, "\x52" },
+            { 0x53, "\x53" }, { 0x54, "\x54" }, { 0x55, "\x55" }, { 0x56, "\x56" }, { 0x57, "\x57" }, { 0x58, "\x58" }, { 0x59, "\x59" }, { 0x5A, "\x5A" }, { 0x5B, "\x5B" }, { 0x5C, "\x5C" },
+            { 0x5D, "\x5D" }, { 0x5E, "\x5E" }, { 0x5F, "\x5F" }, { 0x60, "\x60" }, { 0x61, "\x61" }, { 0x62, "\x62" }, { 0x63, "\x63" }, { 0x64, "\x64" }, { 0x65, "\x65" }, { 0x66, "\x66" },
+            { 0x67, "\x67" }, { 0x68, "\x68" }, { 0x69, "\x69" }, { 0x6A, "\x6A" }, { 0x6B, "\x6B" }, { 0x6C, "\x6C" }, { 0x6D, "\x6D" }, { 0x6E, "\x6E" }, { 0x6F, "\x6F" }, { 0x70, "\x70" },
+            { 0x71, "\x71" }, { 0x72, "\x72" }, { 0x73, "\x73" }, { 0x74, "\x74" }, { 0x75, "\x75" }, { 0x76, "\x76" }, { 0x77, "\x77" }, { 0x78, "\x78" }, { 0x79, "\x79" }, { 0x7A, "\x7A" },
+            { 0x7B, "\x7B" }, { 0x7C, "\x7C" }, { 0x7D, "\x7D" }, { 0x7E, "\x7E" }, { 0xA1, "\xC2\xA1" }, { 0xA2, "\xC2\xA2" }, { 0xA3, "\xC2\xA3" }, { 0xA4, "\xC2\xA4" }, { 0xA5, "\xC2\xA5" },
+            { 0xA6, "\xC2\xA6" }, { 0xA7, "\xC2\xA7" }, { 0xA8, "\xC2\xA8" }, { 0xA9, "\xC2\xA9" }, { 0xAA, "\xC2\xAA" }, { 0xAB, "\xC2\xAB" }, { 0xAC, "\xC2\xAC" }, { 0xAE, "\xC2\xAE" },
+            { 0xAF, "\xC2\xAF" }, { 0xB0, "\xC2\xB0" }, { 0xB1, "\xC2\xB1" }, { 0xB2, "\xC2\xB2" }, { 0xB3, "\xC2\xB3" }, { 0xB4, "\xC2\xB4" }, { 0xB5, "\xC2\xB5" }, { 0xB6, "\xC2\xB6" },
+            { 0xB7, "\xC2\xB7" }, { 0xB8, "\xC2\xB8" }, { 0xB9, "\xC2\xB9" }, { 0xBA, "\xC2\xBA" }, { 0xBB, "\xC2\xBB" }, { 0xBC, "\xC2\xBC" }, { 0xBD, "\xC2\xBD" }, { 0xBE, "\xC2\xBE" },
+            { 0xBF, "\xC2\xBF" }, { 0xC0, "\xC3\x80" }, { 0xC1, "\xC3\x81" }, { 0xC2, "\xC3\x82" }, { 0xC3, "\xC3\x83" }, { 0xC4, "\xC3\x84" }, { 0xC5, "\xC3\x85" }, { 0xC6, "\xC3\x86" },
+            { 0xC7, "\xC3\x87" }, { 0xC8, "\xC3\x88" }, { 0xC9, "\xC3\x89" }, { 0xCA, "\xC3\x8A" }, { 0xCB, "\xC3\x8B" }, { 0xCC, "\xC3\x8C" }, { 0xCD, "\xC3\x8D" }, { 0xCE, "\xC3\x8E" },
+            { 0xCF, "\xC3\x8F" }, { 0xD0, "\xC3\x90" }, { 0xD1, "\xC3\x91" }, { 0xD2, "\xC3\x92" }, { 0xD3, "\xC3\x93" }, { 0xD4, "\xC3\x94" }, { 0xD5, "\xC3\x95" }, { 0xD6, "\xC3\x96" },
+            { 0xD7, "\xC3\x97" }, { 0xD8, "\xC3\x98" }, { 0xD9, "\xC3\x99" }, { 0xDA, "\xC3\x9A" }, { 0xDB, "\xC3\x9B" }, { 0xDC, "\xC3\x9C" }, { 0xDD, "\xC3\x9D" }, { 0xDE, "\xC3\x9E" },
+            { 0xDF, "\xC3\x9F" }, { 0xE0, "\xC3\xA0" }, { 0xE1, "\xC3\xA1" }, { 0xE2, "\xC3\xA2" }, { 0xE3, "\xC3\xA3" }, { 0xE4, "\xC3\xA4" }, { 0xE5, "\xC3\xA5" }, { 0xE6, "\xC3\xA6" },
+            { 0xE7, "\xC3\xA7" }, { 0xE8, "\xC3\xA8" }, { 0xE9, "\xC3\xA9" }, { 0xEA, "\xC3\xAA" }, { 0xEB, "\xC3\xAB" }, { 0xEC, "\xC3\xAC" }, { 0xED, "\xC3\xAD" }, { 0xEE, "\xC3\xAE" },
+            { 0xEF, "\xC3\xAF" }, { 0xF0, "\xC3\xB0" }, { 0xF1, "\xC3\xB1" }, { 0xF2, "\xC3\xB2" }, { 0xF3, "\xC3\xB3" }, { 0xF4, "\xC3\xB4" }, { 0xF5, "\xC3\xB5" }, { 0xF6, "\xC3\xB6" },
+            { 0xF7, "\xC3\xB7" }, { 0xF8, "\xC3\xB8" }, { 0xF9, "\xC3\xB9" }, { 0xFA, "\xC3\xBA" }, { 0xFB, "\xC3\xBB" }, { 0xFC, "\xC3\xBC" }, { 0xFD, "\xC3\xBD" }, { 0xFE, "\xC3\xBE" },
+            { 0xFF, "\xC3\xBF" }, { 0x00, "\xC4\x80" }, { 0x01, "\xC4\x81" }, { 0x02, "\xC4\x82" }, { 0x03, "\xC4\x83" }, { 0x04, "\xC4\x84" }, { 0x05, "\xC4\x85" }, { 0x06, "\xC4\x86" },
+            { 0x07, "\xC4\x87" }, { 0x08, "\xC4\x88" }, { 0x09, "\xC4\x89" }, { 0x0A, "\xC4\x8A" }, { 0x0B, "\xC4\x8B" }, { 0x0C, "\xC4\x8C" }, { 0x0D, "\xC4\x8D" }, { 0x0E, "\xC4\x8E" },
+            { 0x0F, "\xC4\x8F" }, { 0x10, "\xC4\x90" }, { 0x11, "\xC4\x91" }, { 0x12, "\xC4\x92" }, { 0x13, "\xC4\x93" }, { 0x14, "\xC4\x94" }, { 0x15, "\xC4\x95" }, { 0x16, "\xC4\x96" },
+            { 0x17, "\xC4\x97" }, { 0x18, "\xC4\x98" }, { 0x19, "\xC4\x99" }, { 0x1A, "\xC4\x9A" }, { 0x1B, "\xC4\x9B" }, { 0x1C, "\xC4\x9C" }, { 0x1D, "\xC4\x9D" }, { 0x1E, "\xC4\x9E" },
+            { 0x1F, "\xC4\x9F" }, { 0x20, "\xC4\xA0" }, { 0x7F, "\xC4\xA1" }, { 0x80, "\xC4\xA2" }, { 0x81, "\xC4\xA3" }, { 0x82, "\xC4\xA4" }, { 0x83, "\xC4\xA5" }, { 0x84, "\xC4\xA6" },
+            { 0x85, "\xC4\xA7" }, { 0x86, "\xC4\xA8" }, { 0x87, "\xC4\xA9" }, { 0x88, "\xC4\xAA" }, { 0x89, "\xC4\xAB" }, { 0x8A, "\xC4\xAC" }, { 0x8B, "\xC4\xAD" }, { 0x8C, "\xC4\xAE" },
+            { 0x8D, "\xC4\xAF" }, { 0x8E, "\xC4\xB0" }, { 0x8F, "\xC4\xB1" }, { 0x90, "\xC4\xB2" }, { 0x91, "\xC4\xB3" }, { 0x92, "\xC4\xB4" }, { 0x93, "\xC4\xB5" }, { 0x94, "\xC4\xB6" },
+            { 0x95, "\xC4\xB7" }, { 0x96, "\xC4\xB8" }, { 0x97, "\xC4\xB9" }, { 0x98, "\xC4\xBA" }, { 0x99, "\xC4\xBB" }, { 0x9A, "\xC4\xBC" }, { 0x9B, "\xC4\xBD" }, { 0x9C, "\xC4\xBE" },
+            { 0x9D, "\xC4\xBF" }, { 0x9E, "\xC5\x80" }, { 0x9F, "\xC5\x81" }, { 0xA0, "\xC5\x82" }, { 0xAD, "\xC5\x83" }
+        };
         return hex_map;
     }
 
     std::unordered_map<std::string, unsigned char> unicode_to_bytes() {
-        static std::unordered_map<std::string, unsigned char> hex_map = { { "\x21", 0x21 }, { "\x22", 0x22 }, { "\x23", 0x23 }, { "\x24", 0x24 }, { "\x25", 0x25 }, { "\x26", 0x26 }, { "\x27", 0x27 }, { "\x28", 0x28 }, { "\x29", 0x29 }, { "\x2A", 0x2A }, { "\x2B", 0x2B }, { "\x2C", 0x2C }, { "\x2D", 0x2D }, { "\x2E", 0x2E }, { "\x2F", 0x2F }, { "\x30", 0x30 }, { "\x31", 0x31 }, { "\x32", 0x32 }, { "\x33", 0x33 }, { "\x34", 0x34 }, { "\x35", 0x35 }, { "\x36", 0x36 }, { "\x37", 0x37 }, { "\x38", 0x38 }, { "\x39", 0x39 }, { "\x3A", 0x3A }, { "\x3B", 0x3B }, { "\x3C", 0x3C }, { "\x3D", 0x3D }, { "\x3E", 0x3E }, { "\x3F", 0x3F }, { "\x40", 0x40 }, { "\x41", 0x41 }, { "\x42", 0x42 }, { "\x43", 0x43 }, { "\x44", 0x44 }, { "\x45", 0x45 }, { "\x46", 0x46 }, { "\x47", 0x47 }, { "\x48", 0x48 }, { "\x49", 0x49 }, { "\x4A", 0x4A }, { "\x4B", 0x4B }, { "\x4C", 0x4C }, { "\x4D", 0x4D }, { "\x4E", 0x4E }, { "\x4F", 0x4F }, { "\x50", 0x50 }, { "\x51", 0x51 }, { "\x52", 0x52 }, { "\x53", 0x53 }, { "\x54", 0x54 }, { "\x55", 0x55 }, { "\x56", 0x56 }, { "\x57", 0x57 }, { "\x58", 0x58 }, { "\x59", 0x59 }, { "\x5A", 0x5A }, { "\x5B", 0x5B }, { "\x5C", 0x5C }, { "\x5D", 0x5D }, { "\x5E", 0x5E }, { "\x5F", 0x5F }, { "\x60", 0x60 }, { "\x61", 0x61 }, { "\x62", 0x62 }, { "\x63", 0x63 }, { "\x64", 0x64 }, { "\x65", 0x65 }, { "\x66", 0x66 }, { "\x67", 0x67 }, { "\x68", 0x68 }, { "\x69", 0x69 }, { "\x6A", 0x6A }, { "\x6B", 0x6B }, { "\x6C", 0x6C }, { "\x6D", 0x6D }, { "\x6E", 0x6E }, { "\x6F", 0x6F }, { "\x70", 0x70 }, { "\x71", 0x71 }, { "\x72", 0x72 }, { "\x73", 0x73 }, { "\x74", 0x74 }, { "\x75", 0x75 }, { "\x76", 0x76 }, { "\x77", 0x77 }, { "\x78", 0x78 }, { "\x79", 0x79 }, { "\x7A", 0x7A }, { "\x7B", 0x7B }, { "\x7C", 0x7C }, { "\x7D", 0x7D }, { "\x7E", 0x7E }, { "\xC2\xA1", 0xA1 }, { "\xC2\xA2", 0xA2 }, { "\xC2\xA3", 0xA3 }, { "\xC2\xA4", 0xA4 }, { "\xC2\xA5", 0xA5 }, { "\xC2\xA6", 0xA6 }, { "\xC2\xA7", 0xA7 }, { "\xC2\xA8", 0xA8 }, { "\xC2\xA9", 0xA9 }, { "\xC2\xAA", 0xAA }, { "\xC2\xAB", 0xAB }, { "\xC2\xAC", 0xAC }, { "\xC2\xAE", 0xAE }, { "\xC2\xAF", 0xAF }, { "\xC2\xB0", 0xB0 }, { "\xC2\xB1", 0xB1 }, { "\xC2\xB2", 0xB2 }, { "\xC2\xB3", 0xB3 }, { "\xC2\xB4", 0xB4 }, { "\xC2\xB5", 0xB5 }, { "\xC2\xB6", 0xB6 }, { "\xC2\xB7", 0xB7 }, { "\xC2\xB8", 0xB8 }, { "\xC2\xB9", 0xB9 }, { "\xC2\xBA", 0xBA }, { "\xC2\xBB", 0xBB }, { "\xC2\xBC", 0xBC }, { "\xC2\xBD", 0xBD }, { "\xC2\xBE", 0xBE }, { "\xC2\xBF", 0xBF }, { "\xC3\x80", 0xC0 }, { "\xC3\x81", 0xC1 }, { "\xC3\x82", 0xC2 }, { "\xC3\x83", 0xC3 }, { "\xC3\x84", 0xC4 }, { "\xC3\x85", 0xC5 }, { "\xC3\x86", 0xC6 }, { "\xC3\x87", 0xC7 }, { "\xC3\x88", 0xC8 }, { "\xC3\x89", 0xC9 }, { "\xC3\x8A", 0xCA }, { "\xC3\x8B", 0xCB }, { "\xC3\x8C", 0xCC }, { "\xC3\x8D", 0xCD }, { "\xC3\x8E", 0xCE }, { "\xC3\x8F", 0xCF }, { "\xC3\x90", 0xD0 }, { "\xC3\x91", 0xD1 }, { "\xC3\x92", 0xD2 }, { "\xC3\x93", 0xD3 }, { "\xC3\x94", 0xD4 }, { "\xC3\x95", 0xD5 }, { "\xC3\x96", 0xD6 }, { "\xC3\x97", 0xD7 }, { "\xC3\x98", 0xD8 }, { "\xC3\x99", 0xD9 }, { "\xC3\x9A", 0xDA }, { "\xC3\x9B", 0xDB }, { "\xC3\x9C", 0xDC }, { "\xC3\x9D", 0xDD }, { "\xC3\x9E", 0xDE }, { "\xC3\x9F", 0xDF }, { "\xC3\xA0", 0xE0 }, { "\xC3\xA1", 0xE1 }, { "\xC3\xA2", 0xE2 }, { "\xC3\xA3", 0xE3 }, { "\xC3\xA4", 0xE4 }, { "\xC3\xA5", 0xE5 }, { "\xC3\xA6", 0xE6 }, { "\xC3\xA7", 0xE7 }, { "\xC3\xA8", 0xE8 }, { "\xC3\xA9", 0xE9 }, { "\xC3\xAA", 0xEA }, { "\xC3\xAB", 0xEB }, { "\xC3\xAC", 0xEC }, { "\xC3\xAD", 0xED }, { "\xC3\xAE", 0xEE }, { "\xC3\xAF", 0xEF }, { "\xC3\xB0", 0xF0 }, { "\xC3\xB1", 0xF1 }, { "\xC3\xB2", 0xF2 }, { "\xC3\xB3", 0xF3 }, { "\xC3\xB4", 0xF4 }, { "\xC3\xB5", 0xF5 }, { "\xC3\xB6", 0xF6 }, { "\xC3\xB7", 0xF7 }, { "\xC3\xB8", 0xF8 }, { "\xC3\xB9", 0xF9 }, { "\xC3\xBA", 0xFA }, { "\xC3\xBB", 0xFB }, { "\xC3\xBC", 0xFC }, { "\xC3\xBD", 0xFD }, { "\xC3\xBE", 0xFE }, { "\xC3\xBF", 0xFF }, { "\xC4\x80", 0x00 }, { "\xC4\x81", 0x01 }, { "\xC4\x82", 0x02 }, { "\xC4\x83", 0x03 }, { "\xC4\x84", 0x04 }, { "\xC4\x85", 0x05 }, { "\xC4\x86", 0x06 }, { "\xC4\x87", 0x07 }, { "\xC4\x88", 0x08 }, { "\xC4\x89", 0x09 }, { "\xC4\x8A", 0x0A }, { "\xC4\x8B", 0x0B }, { "\xC4\x8C", 0x0C }, { "\xC4\x8D", 0x0D }, { "\xC4\x8E", 0x0E }, { "\xC4\x8F", 0x0F }, { "\xC4\x90", 0x10 }, { "\xC4\x91", 0x11 }, { "\xC4\x92", 0x12 }, { "\xC4\x93", 0x13 }, { "\xC4\x94", 0x14 }, { "\xC4\x95", 0x15 }, { "\xC4\x96", 0x16 }, { "\xC4\x97", 0x17 }, { "\xC4\x98", 0x18 }, { "\xC4\x99", 0x19 }, { "\xC4\x9A", 0x1A }, { "\xC4\x9B", 0x1B }, { "\xC4\x9C", 0x1C }, { "\xC4\x9D", 0x1D }, { "\xC4\x9E", 0x1E }, { "\xC4\x9F", 0x1F }, { "\xC4\xA0", 0x20 }, { "\xC4\xA1", 0x7F }, { "\xC4\xA2", 0x80 }, { "\xC4\xA3", 0x81 }, { "\xC4\xA4", 0x82 }, { "\xC4\xA5", 0x83 }, { "\xC4\xA6", 0x84 }, { "\xC4\xA7", 0x85 }, { "\xC4\xA8", 0x86 }, { "\xC4\xA9", 0x87 }, { "\xC4\xAA", 0x88 }, { "\xC4\xAB", 0x89 }, { "\xC4\xAC", 0x8A }, { "\xC4\xAD", 0x8B }, { "\xC4\xAE", 0x8C }, { "\xC4\xAF", 0x8D }, { "\xC4\xB0", 0x8E }, { "\xC4\xB1", 0x8F }, { "\xC4\xB2", 0x90 }, { "\xC4\xB3", 0x91 }, { "\xC4\xB4", 0x92 }, { "\xC4\xB5", 0x93 }, { "\xC4\xB6", 0x94 }, { "\xC4\xB7", 0x95 }, { "\xC4\xB8", 0x96 }, { "\xC4\xB9", 0x97 }, { "\xC4\xBA", 0x98 }, { "\xC4\xBB", 0x99 }, { "\xC4\xBC", 0x9A }, { "\xC4\xBD", 0x9B }, { "\xC4\xBE", 0x9C }, { "\xC4\xBF", 0x9D }, { "\xC5\x80", 0x9E }, { "\xC5\x81", 0x9F }, { "\xC5\x82", 0xA0 }, { "\xC5\x83", 0xAD }};
+        static std::unordered_map<std::string, unsigned char> hex_map = {
+            { "\x21", 0x21 }, { "\x22", 0x22 }, { "\x23", 0x23 }, { "\x24", 0x24 }, { "\x25", 0x25 }, { "\x26", 0x26 }, { "\x27", 0x27 }, { "\x28", 0x28 }, { "\x29", 0x29 }, { "\x2A", 0x2A },
+            { "\x2B", 0x2B }, { "\x2C", 0x2C }, { "\x2D", 0x2D }, { "\x2E", 0x2E }, { "\x2F", 0x2F }, { "\x30", 0x30 }, { "\x31", 0x31 }, { "\x32", 0x32 }, { "\x33", 0x33 }, { "\x34", 0x34 },
+            { "\x35", 0x35 }, { "\x36", 0x36 }, { "\x37", 0x37 }, { "\x38", 0x38 }, { "\x39", 0x39 }, { "\x3A", 0x3A }, { "\x3B", 0x3B }, { "\x3C", 0x3C }, { "\x3D", 0x3D }, { "\x3E", 0x3E },
+            { "\x3F", 0x3F }, { "\x40", 0x40 }, { "\x41", 0x41 }, { "\x42", 0x42 }, { "\x43", 0x43 }, { "\x44", 0x44 }, { "\x45", 0x45 }, { "\x46", 0x46 }, { "\x47", 0x47 }, { "\x48", 0x48 },
+            { "\x49", 0x49 }, { "\x4A", 0x4A }, { "\x4B", 0x4B }, { "\x4C", 0x4C }, { "\x4D", 0x4D }, { "\x4E", 0x4E }, { "\x4F", 0x4F }, { "\x50", 0x50 }, { "\x51", 0x51 }, { "\x52", 0x52 },
+            { "\x53", 0x53 }, { "\x54", 0x54 }, { "\x55", 0x55 }, { "\x56", 0x56 }, { "\x57", 0x57 }, { "\x58", 0x58 }, { "\x59", 0x59 }, { "\x5A", 0x5A }, { "\x5B", 0x5B }, { "\x5C", 0x5C },
+            { "\x5D", 0x5D }, { "\x5E", 0x5E }, { "\x5F", 0x5F }, { "\x60", 0x60 }, { "\x61", 0x61 }, { "\x62", 0x62 }, { "\x63", 0x63 }, { "\x64", 0x64 }, { "\x65", 0x65 }, { "\x66", 0x66 },
+            { "\x67", 0x67 }, { "\x68", 0x68 }, { "\x69", 0x69 }, { "\x6A", 0x6A }, { "\x6B", 0x6B }, { "\x6C", 0x6C }, { "\x6D", 0x6D }, { "\x6E", 0x6E }, { "\x6F", 0x6F }, { "\x70", 0x70 },
+            { "\x71", 0x71 }, { "\x72", 0x72 }, { "\x73", 0x73 }, { "\x74", 0x74 }, { "\x75", 0x75 }, { "\x76", 0x76 }, { "\x77", 0x77 }, { "\x78", 0x78 }, { "\x79", 0x79 }, { "\x7A", 0x7A },
+            { "\x7B", 0x7B }, { "\x7C", 0x7C }, { "\x7D", 0x7D }, { "\x7E", 0x7E }, { "\xC2\xA1", 0xA1 }, { "\xC2\xA2", 0xA2 }, { "\xC2\xA3", 0xA3 }, { "\xC2\xA4", 0xA4 }, { "\xC2\xA5", 0xA5 },
+            { "\xC2\xA6", 0xA6 }, { "\xC2\xA7", 0xA7 }, { "\xC2\xA8", 0xA8 }, { "\xC2\xA9", 0xA9 }, { "\xC2\xAA", 0xAA }, { "\xC2\xAB", 0xAB }, { "\xC2\xAC", 0xAC }, { "\xC2\xAE", 0xAE },
+            { "\xC2\xAF", 0xAF }, { "\xC2\xB0", 0xB0 }, { "\xC2\xB1", 0xB1 }, { "\xC2\xB2", 0xB2 }, { "\xC2\xB3", 0xB3 }, { "\xC2\xB4", 0xB4 }, { "\xC2\xB5", 0xB5 }, { "\xC2\xB6", 0xB6 },
+            { "\xC2\xB7", 0xB7 }, { "\xC2\xB8", 0xB8 }, { "\xC2\xB9", 0xB9 }, { "\xC2\xBA", 0xBA }, { "\xC2\xBB", 0xBB }, { "\xC2\xBC", 0xBC }, { "\xC2\xBD", 0xBD }, { "\xC2\xBE", 0xBE },
+            { "\xC2\xBF", 0xBF }, { "\xC3\x80", 0xC0 }, { "\xC3\x81", 0xC1 }, { "\xC3\x82", 0xC2 }, { "\xC3\x83", 0xC3 }, { "\xC3\x84", 0xC4 }, { "\xC3\x85", 0xC5 }, { "\xC3\x86", 0xC6 },
+            { "\xC3\x87", 0xC7 }, { "\xC3\x88", 0xC8 }, { "\xC3\x89", 0xC9 }, { "\xC3\x8A", 0xCA }, { "\xC3\x8B", 0xCB }, { "\xC3\x8C", 0xCC }, { "\xC3\x8D", 0xCD }, { "\xC3\x8E", 0xCE },
+            { "\xC3\x8F", 0xCF }, { "\xC3\x90", 0xD0 }, { "\xC3\x91", 0xD1 }, { "\xC3\x92", 0xD2 }, { "\xC3\x93", 0xD3 }, { "\xC3\x94", 0xD4 }, { "\xC3\x95", 0xD5 }, { "\xC3\x96", 0xD6 },
+            { "\xC3\x97", 0xD7 }, { "\xC3\x98", 0xD8 }, { "\xC3\x99", 0xD9 }, { "\xC3\x9A", 0xDA }, { "\xC3\x9B", 0xDB }, { "\xC3\x9C", 0xDC }, { "\xC3\x9D", 0xDD }, { "\xC3\x9E", 0xDE },
+            { "\xC3\x9F", 0xDF }, { "\xC3\xA0", 0xE0 }, { "\xC3\xA1", 0xE1 }, { "\xC3\xA2", 0xE2 }, { "\xC3\xA3", 0xE3 }, { "\xC3\xA4", 0xE4 }, { "\xC3\xA5", 0xE5 }, { "\xC3\xA6", 0xE6 },
+            { "\xC3\xA7", 0xE7 }, { "\xC3\xA8", 0xE8 }, { "\xC3\xA9", 0xE9 }, { "\xC3\xAA", 0xEA }, { "\xC3\xAB", 0xEB }, { "\xC3\xAC", 0xEC }, { "\xC3\xAD", 0xED }, { "\xC3\xAE", 0xEE },
+            { "\xC3\xAF", 0xEF }, { "\xC3\xB0", 0xF0 }, { "\xC3\xB1", 0xF1 }, { "\xC3\xB2", 0xF2 }, { "\xC3\xB3", 0xF3 }, { "\xC3\xB4", 0xF4 }, { "\xC3\xB5", 0xF5 }, { "\xC3\xB6", 0xF6 },
+            { "\xC3\xB7", 0xF7 }, { "\xC3\xB8", 0xF8 }, { "\xC3\xB9", 0xF9 }, { "\xC3\xBA", 0xFA }, { "\xC3\xBB", 0xFB }, { "\xC3\xBC", 0xFC }, { "\xC3\xBD", 0xFD }, { "\xC3\xBE", 0xFE },
+            { "\xC3\xBF", 0xFF }, { "\xC4\x80", 0x00 }, { "\xC4\x81", 0x01 }, { "\xC4\x82", 0x02 }, { "\xC4\x83", 0x03 }, { "\xC4\x84", 0x04 }, { "\xC4\x85", 0x05 }, { "\xC4\x86", 0x06 },
+            { "\xC4\x87", 0x07 }, { "\xC4\x88", 0x08 }, { "\xC4\x89", 0x09 }, { "\xC4\x8A", 0x0A }, { "\xC4\x8B", 0x0B }, { "\xC4\x8C", 0x0C }, { "\xC4\x8D", 0x0D }, { "\xC4\x8E", 0x0E },
+            { "\xC4\x8F", 0x0F }, { "\xC4\x90", 0x10 }, { "\xC4\x91", 0x11 }, { "\xC4\x92", 0x12 }, { "\xC4\x93", 0x13 }, { "\xC4\x94", 0x14 }, { "\xC4\x95", 0x15 }, { "\xC4\x96", 0x16 },
+            { "\xC4\x97", 0x17 }, { "\xC4\x98", 0x18 }, { "\xC4\x99", 0x19 }, { "\xC4\x9A", 0x1A }, { "\xC4\x9B", 0x1B }, { "\xC4\x9C", 0x1C }, { "\xC4\x9D", 0x1D }, { "\xC4\x9E", 0x1E },
+            { "\xC4\x9F", 0x1F }, { "\xC4\xA0", 0x20 }, { "\xC4\xA1", 0x7F }, { "\xC4\xA2", 0x80 }, { "\xC4\xA3", 0x81 }, { "\xC4\xA4", 0x82 }, { "\xC4\xA5", 0x83 }, { "\xC4\xA6", 0x84 },
+            { "\xC4\xA7", 0x85 }, { "\xC4\xA8", 0x86 }, { "\xC4\xA9", 0x87 }, { "\xC4\xAA", 0x88 }, { "\xC4\xAB", 0x89 }, { "\xC4\xAC", 0x8A }, { "\xC4\xAD", 0x8B }, { "\xC4\xAE", 0x8C },
+            { "\xC4\xAF", 0x8D }, { "\xC4\xB0", 0x8E }, { "\xC4\xB1", 0x8F }, { "\xC4\xB2", 0x90 }, { "\xC4\xB3", 0x91 }, { "\xC4\xB4", 0x92 }, { "\xC4\xB5", 0x93 }, { "\xC4\xB6", 0x94 },
+            { "\xC4\xB7", 0x95 }, { "\xC4\xB8", 0x96 }, { "\xC4\xB9", 0x97 }, { "\xC4\xBA", 0x98 }, { "\xC4\xBB", 0x99 }, { "\xC4\xBC", 0x9A }, { "\xC4\xBD", 0x9B }, { "\xC4\xBE", 0x9C },
+            { "\xC4\xBF", 0x9D }, { "\xC5\x80", 0x9E }, { "\xC5\x81", 0x9F }, { "\xC5\x82", 0xA0 }, { "\xC5\x83", 0xAD }
+        };
         return hex_map;
     }
 
@@ -730,13 +963,11 @@ struct gpt2bpe_tokenizer {
         return true;
     }
 
-    std::vector<std::string> bpe_gpt2_preprocess(const std::string& text)
-    {
+    std::vector<std::string> bpe_gpt2_preprocess(const std::string& text) {
         static std::unordered_map< unsigned char, std::string> byte_encoder = bytes_to_unicode();
         std::vector<std::string> bpe_words;
         std::vector<std::string> bpe_encoded_words;
 
-
         std::string token="";
         const char *raw_text_p = text.c_str();
         // GPT2 system regex:  's|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+
@@ -753,8 +984,7 @@ struct gpt2bpe_tokenizer {
 
         text_utf = CNCTUnicode::split_utf8_enhanced(text);
 
-        for (int i = 0; i < (int)text_utf.size(); i++)
-        {
+        for (int i = 0; i < (int)text_utf.size(); i++) {
             const CNCTString &utf_char = text_utf[i];
             bool split_condition = false;
             const char *text_pos = raw_text_p + utf_char.seq_offset_bytes;
@@ -765,15 +995,15 @@ struct gpt2bpe_tokenizer {
             // const CNCTString &utf_char_prev = (i > 0) ? text_utf[i-1] : CNCTString();
 
             // handling contractions
-            if (!split_condition && bytes_remain >= 2)
-            {
+            if (!split_condition && bytes_remain >= 2) {
                 // 's|'t|'m|'d
-                if (utf_char == '\'' && (utf_char_next == 's' || utf_char_next == 't' || utf_char_next == 'm' || utf_char_next == 'd'))
+                if (utf_char == '\'' && (utf_char_next == 's' || utf_char_next == 't' || utf_char_next == 'm' || utf_char_next == 'd')) {
                     split_condition = true;
-                if (split_condition)
-                {
-                    if (token.size())
+                }
+                if (split_condition) {
+                    if (token.size()) {
                         bpe_words.emplace_back(token); // push previous content as token
+                    }
                     token = utf_char.str + utf_char_next.str;
                     bpe_words.emplace_back(token);
                     token="";
@@ -781,20 +1011,20 @@ struct gpt2bpe_tokenizer {
                     continue;
                 }
             }
-            if (!split_condition && bytes_remain >= 3)
-            {
+            if (!split_condition && bytes_remain >= 3) {
                 // 're|'ve|'ll
                 if (utf_char == '\'' && (
-                    (utf_char_next == 'r' || utf_char_next_next == 'e') ||
-                    (utf_char_next == 'v' || utf_char_next_next == 'e') ||
-                    (utf_char_next == 'l' || utf_char_next_next == 'l'))
-                )
+                                          (utf_char_next == 'r' || utf_char_next_next == 'e') ||
+                                          (utf_char_next == 'v' || utf_char_next_next == 'e') ||
+                                          (utf_char_next == 'l' || utf_char_next_next == 'l'))
+                                        ) {
                     split_condition = true;
-                if (split_condition)
-                {
+                }
+                if (split_condition) {
                     // current token + next token can be defined
-                    if (token.size())
+                    if (token.size()) {
                         bpe_words.emplace_back(token); // push previous content as token
+                    }
                     token = utf_char.str + utf_char_next.str + utf_char_next_next.str;
                     bpe_words.emplace_back(token); // the contraction
                     token="";
@@ -803,78 +1033,60 @@ struct gpt2bpe_tokenizer {
                 }
             }
 
-            if (!split_condition && !collecting)
-            {
-                if (utf_char.char_type == CNCTCharType::LETTER || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::LETTER))
-                {
+            if (!split_condition && !collecting) {
+                if (utf_char.char_type == CNCTCharType::LETTER || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::LETTER)) {
                     collecting_letter = true;
                     collecting = true;
-                }
-                else if (utf_char.char_type == CNCTCharType::DIGIT || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::DIGIT))
-                {
+                } else if (utf_char.char_type == CNCTCharType::DIGIT || (!token.size() && utf_char==" " && utf_char_next.char_type == CNCTCharType::DIGIT)) {
                     collecting_numeric = true;
                     collecting = true;
-                }
-                else if (
-                    ((utf_char.char_type != CNCTCharType::LETTER && utf_char.char_type != CNCTCharType::DIGIT) && (utf_char.char_type != CNCTCharType::WHITESPACE)) ||
-                    (!token.size() && utf_char==" " && utf_char_next.char_type != CNCTCharType::LETTER && utf_char_next.char_type != CNCTCharType::DIGIT && utf_char_next.char_type != CNCTCharType::WHITESPACE)
-                    )
-                {
+                } else if (
+                           ((utf_char.char_type != CNCTCharType::LETTER && utf_char.char_type != CNCTCharType::DIGIT) && (utf_char.char_type != CNCTCharType::WHITESPACE)) ||
+                           (!token.size() && utf_char==" " && utf_char_next.char_type != CNCTCharType::LETTER && utf_char_next.char_type != CNCTCharType::DIGIT && utf_char_next.char_type != CNCTCharType::WHITESPACE)
+                          ) {
                     collecting_special = true;
                     collecting = true;
-                }
-                else if (utf_char.char_type == CNCTCharType::WHITESPACE && utf_char_next.char_type == CNCTCharType::WHITESPACE)
-                {
+                } else if (utf_char.char_type == CNCTCharType::WHITESPACE && utf_char_next.char_type == CNCTCharType::WHITESPACE) {
                     collecting_whitespace_lookahead = true;
                     collecting = true;
-                } else if (utf_char.char_type == CNCTCharType::WHITESPACE)
-                {
+                } else if (utf_char.char_type == CNCTCharType::WHITESPACE) {
                     split_condition = true;
                 }
-            } else
-            if (!split_condition && collecting)
-            {
-                if (collecting_letter && utf_char.char_type != CNCTCharType::LETTER)
-                {
+            } else if (!split_condition && collecting) {
+                if (collecting_letter && utf_char.char_type != CNCTCharType::LETTER) {
                     split_condition = true;
-                }
-                else if (collecting_numeric && utf_char.char_type != CNCTCharType::DIGIT)
-                {
+                } else if (collecting_numeric && utf_char.char_type != CNCTCharType::DIGIT) {
                     split_condition = true;
-                }
-                else if (collecting_special && (utf_char.char_type == CNCTCharType::LETTER || utf_char.char_type == CNCTCharType::DIGIT || utf_char.char_type == CNCTCharType::WHITESPACE))
-                {
+                } else if (collecting_special && (utf_char.char_type == CNCTCharType::LETTER || utf_char.char_type == CNCTCharType::DIGIT || utf_char.char_type == CNCTCharType::WHITESPACE)) {
                     split_condition = true;
-                }
-                else if (collecting_whitespace_lookahead && utf_char_next.char_type != CNCTCharType::WHITESPACE)
-                {
+                } else if (collecting_whitespace_lookahead && utf_char_next.char_type != CNCTCharType::WHITESPACE) {
                     split_condition = true;
                 }
             }
-            if(utf_char_next.str.size() == 0)
-            {
+
+            if(utf_char_next.str.size() == 0) {
                 split_condition = true; // final
                 token += utf_char.str;
             }
 
-            if (split_condition)
-            {
-                if (token.size())
+            if (split_condition) {
+                if (token.size()) {
                     bpe_words.emplace_back(token);
+                }
                 token = utf_char.str;
                 collecting = false;
                 collecting_letter = false;
                 collecting_numeric = false;
                 collecting_special = false;
                 collecting_whitespace_lookahead = false;
-            } else token += utf_char.str;
+            } else {
+                token += utf_char.str;
+            }
         }
 
-        for (std::string& word : bpe_words)
-        {
+        for (std::string& word : bpe_words) {
             std::string encoded_token="";
-            for (char& c : word)
-            {
+            for (char& c : word) {
                 encoded_token += byte_encoder[c];
             }
             bpe_encoded_words.emplace_back(encoded_token);
@@ -884,21 +1096,17 @@ struct gpt2bpe_tokenizer {
     }
 
     // decoder (for one token)
-    std::string decode_token(const std::string& token)
-    {
+    std::string decode_token(const std::string& token) {
         static std::unordered_map< std::string, unsigned char> byte_decoder = unicode_to_bytes();
         std::string decoded_token="";
         auto unicode_seqeunces = CNCTUnicode::split_utf8(token);
-        for (auto& unicode_sequence : unicode_seqeunces)
-        {
+        for (auto& unicode_sequence : unicode_seqeunces) {
             decoded_token += byte_decoder[unicode_sequence];
-
         }
 
         return decoded_token;
     }
 
-
     const gpt2bpe_vocab & vocab_;
     std::vector<ggllm_bpe_symbol> symbols_;
     std::vector<ggllm_bpe_symbol> symbols_final;
@@ -922,5 +1130,4 @@ static std::vector<gpt2bpe_vocab::id> gpt2bpe_tokenize(const gpt2bpe_vocab & voc
     return output;
 }
 
-
 #endif // CMPNCT_GPT2BPE

From 76b46627e2c9b0f1d1ad777ab21b360d33dc0a4b Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 20 Aug 2023 16:54:42 +0200
Subject: [PATCH 221/242] convert-llama-hf-to-gguf.py : special tokens

---
 convert-llama-hf-to-gguf.py | 27 +++++++++++++++++++++++----
 1 file changed, 23 insertions(+), 4 deletions(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index a998a6f496e91..9289efd28cf54 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -173,18 +173,20 @@ def count_model_parts(dir_model: str) -> int:
     gguf_writer.add_token_scores(scores)
     gguf_writer.add_token_types(toktypes)
 
+
+print("gguf: get special token ids")
+
 if Path(dir_model + "/tokenizer.json").is_file():
+    # Look for special tokens in tokenizer.json if it exists
+
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
         tokenizer = json.load(f)
 
     if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
-        print("gguf: get special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
             tokenizer_config = json.load(f)
 
-        # find special token ids
-
         if "bos_token" in tokenizer_config and tokenizer_config["bos_token"] != None:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["bos_token"]["content"]:
@@ -209,6 +211,23 @@ def count_model_parts(dir_model: str) -> int:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["pad_token"]["content"]:
                     gguf_writer.add_pad_token_id(key["id"])
+else:
+    # If no tokenizer.json: Look for special tokens in config.json
+
+    if "bos_token_id" in hparams and hparams["bos_token_id"] != None:
+        gguf_writer.add_bos_token_id(hparams["bos_token_id"])
+
+    if "eos_token_id" in hparams and hparams["eos_token_id"] != None:
+        gguf_writer.add_eos_token_id(hparams["eos_token_id"])
+
+    if "unk_token_id" in hparams and hparams["unk_token_id"] != None:
+        gguf_writer.add_unk_token_id(hparams["unk_token_id"])
+
+    if "sep_token_id" in hparams and hparams["sep_token_id"] != None:
+        gguf_writer.add_sep_token_id(hparams["sep_token_id"])
+
+    if "pad_token_id" in hparams and hparams["pad_token_id"] != None:
+        gguf_writer.add_pad_token_id(hparams["pad_token_id"])
 
 
 # TENSORS
@@ -258,7 +277,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype
+        data_dtype = data.dtype 
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:

From f838faa874d0ab07fee00a6d02b8ca1b332f36ae Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Sun, 20 Aug 2023 16:56:48 +0200
Subject: [PATCH 222/242] convert-llama-7b-pth-to-gguf.py : special tokens

---
 convert-llama-7b-pth-to-gguf.py | 39 ++++++++++++++++++++++++++++-----
 1 file changed, 34 insertions(+), 5 deletions(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 7c767df609301..77edd026c6243 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -1,4 +1,4 @@
-# 7b pth llama --> gguf conversion, GQA/70b not supported
+# 7b pth llama --> gguf conversion
 # Only models with a single datafile are supported, like 7B
 # HF files required in the model dir: config.json tokenizer_config.json tokenizer.json tokenizer.model
 
@@ -96,10 +96,20 @@ def count_model_parts(dir_model: str) -> int:
 else:
     hf_repo = ""
 
+if "max_sequence_length" in hparams:
+    ctx_length = hparams["max_sequence_length"]
+elif "max_position_embeddings" in hparams:
+    ctx_length = hparams["max_position_embeddings"]
+else:
+    print("gguf: can not find ctx length parameter.")
+
+    sys.exit()
+
+
 gguf_writer.add_name(last_dir)
 gguf_writer.add_source_hf_repo(hf_repo)
 gguf_writer.add_tensor_data_layout("Meta AI original pth")
-gguf_writer.add_context_length(hparams["max_position_embeddings"])
+gguf_writer.add_context_length(ctx_length)
 gguf_writer.add_embedding_length(hparams["hidden_size"])
 gguf_writer.add_block_count(block_count)
 gguf_writer.add_feed_forward_length(hparams["intermediate_size"])
@@ -155,18 +165,20 @@ def count_model_parts(dir_model: str) -> int:
     gguf_writer.add_token_scores(scores)
     gguf_writer.add_token_types(toktypes)
 
+
+print("gguf: get special token ids")
+
 if Path(dir_model + "/tokenizer.json").is_file():
+    # Look for special tokens in tokenizer.json if it exists
+
     with open(dir_model + "/tokenizer.json", "r", encoding="utf-8") as f:
         tokenizer = json.load(f)
 
     if "added_tokens" in tokenizer and Path(dir_model + "/tokenizer_config.json").is_file():
-        print("gguf: get special token ids")
 
         with open(dir_model + "/tokenizer_config.json", "r", encoding="utf-8") as f:
             tokenizer_config = json.load(f)
 
-        # find special token ids
-
         if "bos_token" in tokenizer_config and tokenizer_config["bos_token"] != None:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["bos_token"]["content"]:
@@ -191,6 +203,23 @@ def count_model_parts(dir_model: str) -> int:
             for key in tokenizer["added_tokens"]:
                 if key["content"] == tokenizer_config["pad_token"]["content"]:
                     gguf_writer.add_pad_token_id(key["id"])
+else:
+    # If no tokenizer.json: Look for special tokens in config.json
+
+    if "bos_token_id" in hparams and hparams["bos_token_id"] != None:
+        gguf_writer.add_bos_token_id(hparams["bos_token_id"])
+
+    if "eos_token_id" in hparams and hparams["eos_token_id"] != None:
+        gguf_writer.add_eos_token_id(hparams["eos_token_id"])
+
+    if "unk_token_id" in hparams and hparams["unk_token_id"] != None:
+        gguf_writer.add_unk_token_id(hparams["unk_token_id"])
+
+    if "sep_token_id" in hparams and hparams["sep_token_id"] != None:
+        gguf_writer.add_sep_token_id(hparams["sep_token_id"])
+
+    if "pad_token_id" in hparams and hparams["pad_token_id"] != None:
+        gguf_writer.add_pad_token_id(hparams["pad_token_id"])
 
 
 # TENSORS

From 5a02b9625af73a449091880b64cb5d0b79ad02fe Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 03:24:29 +0200
Subject: [PATCH 223/242] convert-permute-debug.py : permute debug print

---
 convert-permute-debug.py | 1032 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 1032 insertions(+)
 create mode 100644 convert-permute-debug.py

diff --git a/convert-permute-debug.py b/convert-permute-debug.py
new file mode 100644
index 0000000000000..14927e8e4976d
--- /dev/null
+++ b/convert-permute-debug.py
@@ -0,0 +1,1032 @@
+#!/usr/bin/env python
+
+import gguf
+import argparse
+import concurrent.futures
+import copy
+import enum
+import faulthandler
+import functools
+import io
+import itertools
+import json
+import math
+import mmap
+import pickle
+import re
+import signal
+import struct
+import sys
+import zipfile
+import numpy as np
+
+from abc import ABCMeta, abstractmethod
+from dataclasses import dataclass
+from pathlib import Path
+from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List, Literal, Optional, Sequence, Tuple, TypeVar, Union)
+from sentencepiece import SentencePieceProcessor  # type: ignore
+
+if TYPE_CHECKING:
+    from typing_extensions import TypeAlias
+
+if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
+    faulthandler.register(signal.SIGUSR1)
+
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+
+ARCH=gguf.MODEL_ARCH.LLAMA
+NAMES=gguf.MODEL_TENSOR_NAMES[ARCH]
+
+#
+# data types
+#
+
+@dataclass(frozen=True)
+class UnquantizedDataType:
+    name: str
+
+DT_F16  = UnquantizedDataType('F16')
+DT_F32  = UnquantizedDataType('F32')
+DT_I32  = UnquantizedDataType('I32')
+DT_BF16 = UnquantizedDataType('BF16')
+
+DataType = Union[UnquantizedDataType]
+
+DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
+    DT_BF16: np.dtype(np.uint16),
+    DT_F16:  np.dtype(np.float16),
+    DT_F32:  np.dtype(np.float32),
+    DT_I32:  np.dtype(np.int32),
+}
+
+NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
+    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
+
+SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
+    'BF16': DT_BF16,
+    'F16': DT_F16,
+    'F32': DT_F32,
+    'I32': DT_I32,
+}
+
+class GGMLFileType(enum.Enum):
+    AllF32    = 0
+    MostlyF16 = 1  # except 1d tensors
+
+    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
+        if len(tensor.shape) == 1:
+            # 1D tensors are always F32.
+            return DT_F32
+        elif self == GGMLFileType.AllF32:
+            return DT_F32
+        elif self == GGMLFileType.MostlyF16:
+            return DT_F16
+        else:
+            raise ValueError(self)
+
+
+#
+# hparams loading
+#
+
+@dataclass
+class Params:
+    n_vocab:    int
+    n_embd:     int
+    n_mult:     int
+    n_layer:    int
+    n_ctx:      int
+    n_ff:       int
+    n_head:     int
+    n_head_kv:  int
+    f_norm_eps: float
+
+    @staticmethod
+    def find_n_mult(n_ff: int, n_embd: int) -> int:
+        # hardcoded magic range
+        for n_mult in range(8192, 1, -1):
+            calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
+            if calc_ff == n_ff:
+                return n_mult
+        raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
+
+    @staticmethod
+    def guessed(model: 'LazyModel') -> 'Params':
+        # try transformer naming first
+        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
+
+        # try transformer naming first
+        if "model.layers.0.self_attn.q_proj.weight" in model:
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
+        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
+        else:
+            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
+
+        if n_layer < 1:
+            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
+
+        n_head = n_embd // 128 # guessed
+        n_mult = 256           # guessed
+
+        # TODO: verify this
+        n_ff = int(2 * (4 * n_embd) / 3)
+        n_ff = n_mult * ((n_ff + n_mult - 1) // n_mult)
+
+        return Params(
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = -1,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head,
+            f_norm_eps = 1e-5,
+        )
+
+    @staticmethod
+    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab    = config["vocab_size"]
+        n_embd     = config["hidden_size"]
+        n_layer    = config["num_hidden_layers"]
+        n_ff       = config["intermediate_size"]
+        n_head     = config["num_attention_heads"]
+        n_head_kv  = config["num_key_value_heads"] if "num_key_value_heads" in config else n_head
+        f_norm_eps = config["rms_norm_eps"]
+
+        n_mult = Params.find_n_mult(n_ff, n_embd)
+
+        if "max_sequence_length" in config:
+            n_ctx = config["max_sequence_length"]
+        elif "max_position_embeddings" in config:
+            n_ctx = config["max_position_embeddings"]
+        else:
+            raise Exception("failed to guess 'n_ctx'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
+
+        return Params(
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
+        )
+
+    # LLaMA v2 70B params.json
+    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
+    @staticmethod
+    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab    = config["vocab_size"]
+        n_embd     = config["dim"]
+        n_layer    = config["n_layers"]
+        n_mult     = config["multiple_of"]
+        n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
+        n_ff       = -1
+        n_head     = config["n_heads"]
+        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head
+        f_norm_eps = config["norm_eps"]
+
+        if n_vocab == -1:
+            n_vocab = model["tok_embeddings.weight"].shape[0]
+
+        if n_ff == -1:
+            n_ff = model["layers.0.feed_forward.w1.weight"].shape[0]
+
+        return Params(
+            n_vocab    = n_vocab,
+            n_embd     = n_embd,
+            n_mult     = n_mult,
+            n_layer    = n_layer,
+            n_ctx      = n_ctx,
+            n_ff       = n_ff,
+            n_head     = n_head,
+            n_head_kv  = n_head_kv,
+            f_norm_eps = f_norm_eps,
+        )
+
+    @staticmethod
+    def load(model_plus: 'ModelPlus') -> 'Params':
+        hf_config_path   = model_plus.paths[0].parent / "config.json"
+        orig_config_path = model_plus.paths[0].parent / "params.json"
+
+        if hf_config_path.exists():
+            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
+        elif orig_config_path.exists():
+            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
+        else:
+            params = Params.guessed(model_plus.model)
+
+        return params
+
+
+#
+# vocab
+#
+
+class BpeVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
+        else:
+            added_tokens = {}
+        vocab_size: int = len(self.bpe_tokenizer)
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
+
+    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.bpe_tokenizer
+        from transformers.models.gpt2 import tokenization_gpt2
+        byte_encoder = tokenization_gpt2.bytes_to_unicode()
+        byte_decoder = {v: k for k, v in byte_encoder.items()}
+        for i, item in enumerate(tokenizer):
+            text: bytes = item.encode("utf-8")
+            score: float = -i
+            yield text, score
+
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.bpe_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+
+class SentencePieceVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
+        self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
+        else:
+            added_tokens = {}
+        vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids   = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
+
+    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.sentencepiece_tokenizer
+        for i in range(tokenizer.vocab_size()):
+            piece = tokenizer.id_to_piece(i)
+            text: bytes = piece.encode("utf-8")
+            score: float = tokenizer.get_score(i)
+            yield text, score
+
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.sentencepiece_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+Vocab = Union[BpeVocab, SentencePieceVocab]
+
+
+#
+# data loading
+# TODO: reuse (probably move to gguf.py?)
+#
+
+def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
+    print( "permute " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_head_kv) )
+    if n_head_kv is not None and n_head != n_head_kv:
+        n_head //= n_head_kv
+    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
+                .swapaxes(1, 2)
+                .reshape(weights.shape))
+
+
+class Tensor(metaclass=ABCMeta):
+    data_type: DataType
+
+    @abstractmethod
+    def astype(self, data_type: DataType) -> 'Tensor': ...
+    @abstractmethod
+    def permute(self, n_head: int, n_head_kv: int) -> 'Tensor': ...
+    @abstractmethod
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def part(self, n_part: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
+
+
+def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
+    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
+    fp32_arr = bf16_arr.astype(np.uint32) << 16
+    return fp32_arr.view(np.float32)
+
+
+class UnquantizedTensor(Tensor):
+    def __init__(self, ndarray: NDArray) -> None:
+        assert isinstance(ndarray, np.ndarray)
+        self.ndarray = ndarray
+        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
+
+    def astype(self, data_type: DataType) -> Tensor:
+        dtype = DATA_TYPE_TO_NUMPY[data_type]
+        if self.data_type == DT_BF16:
+            self.ndarray = bf16_to_fp32(self.ndarray)
+        return UnquantizedTensor(self.ndarray.astype(dtype))
+
+    def to_ggml(self) -> 'UnquantizedTensor':
+        return self
+
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
+
+    def part(self, n_part: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
+
+    def permute(self, n_head: int, n_head_kv: int) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head, n_head_kv))
+
+
+def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
+    tensor = lazy_tensor.load()
+    assert isinstance(tensor, UnquantizedTensor)
+
+    # double-check:
+    actual_shape = list(tensor.ndarray.shape)
+    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
+    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
+        if convert:
+            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
+        else:
+            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
+
+    return tensor.ndarray
+
+
+GGMLCompatibleTensor = Union[UnquantizedTensor]
+
+
+class DeferredPermutedTensor(Tensor):
+    def __init__(self, base: Tensor, n_head: int, n_head_kv: int) -> None:
+        self.base = base
+        self.n_head = n_head
+        self.data_type = self.base.data_type
+
+    def astype(self, data_type: DataType) -> Tensor:
+        return self.base.astype(data_type).permute(self.n_head, self.n_head_kv)
+
+    def to_ggml(self) -> GGMLCompatibleTensor:
+        return self.base.to_ggml().permute(self.n_head, self.n_head_kv)
+
+    def permute(self, n_head: int, n_head_kv: int) -> Tensor:
+        raise Exception("shouldn't permute twice")
+
+
+@dataclass
+class LazyTensor:
+    _load: Callable[[], Tensor]
+    shape: List[int]
+    data_type: DataType
+    description: str
+
+    def load(self) -> Tensor:
+        ret = self._load()
+        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
+        return ret
+
+    def astype(self, data_type: DataType) -> 'LazyTensor':
+        self.validate_conversion_to(data_type)
+
+        def load() -> Tensor:
+            return self.load().astype(data_type)
+        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
+
+    def validate_conversion_to(self, data_type: DataType) -> None:
+        if data_type == self.data_type:
+            return
+
+
+LazyModel = Dict[str, LazyTensor]
+
+
+@dataclass
+class ModelPlus:
+    model: LazyModel
+    paths: List[Path]  # Where this was read from.
+    format: Literal['ggml', 'torch', 'safetensors']
+    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
+
+
+def merge_sharded(models: List[LazyModel]) -> LazyModel:
+    # Original LLaMA models have each file contain one part of each tensor.
+    # Use a dict instead of a set to preserve order.
+    names = {name: None for model in models for name in model}
+
+    def convert(name: str) -> LazyTensor:
+        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
+        if len(lazy_tensors) == 1:
+            # only one file; don't go through this procedure since there might
+            # be quantized tensors
+            return lazy_tensors[0]
+        if len(lazy_tensors[0].shape) == 1:
+            # the tensor is just duplicated in every file
+            return lazy_tensors[0]
+        if name.startswith('tok_embeddings.') or \
+           name.endswith('.attention.wo.weight') or \
+           name.endswith('.feed_forward.w2.weight'):
+            # split by columns
+            axis = 1
+        else:
+            # split by rows
+            axis = 0
+        concatenated_shape = list(lazy_tensors[0].shape)
+        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
+
+        def load() -> UnquantizedTensor:
+            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
+            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
+            return UnquantizedTensor(concatenated)
+        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
+        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
+    return {name: convert(name) for name in names}
+
+
+def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
+    formats = set(mp.format for mp in models_plus)
+    assert len(formats) == 1, "different formats?"
+    format = formats.pop()
+    paths = [path for mp in models_plus for path in mp.paths]
+    # Use the first non-None vocab, if any.
+    try:
+        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
+    except StopIteration:
+        vocab = None
+
+    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
+        # Transformers models put different tensors in different files, but
+        # don't split indivdual tensors between files.
+        model: LazyModel = {}
+        for mp in models_plus:
+            model.update(mp.model)
+    else:
+        model = merge_sharded([mp.model for mp in models_plus])
+
+    return ModelPlus(model, paths, format, vocab)
+
+
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_head_kv: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute(n_head, n_head_kv)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_head_kv}) ' + lazy_tensor.description)
+
+def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute_part(n_part, n_head)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
+
+def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().part(n_part)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
+
+
+# Functionality that simulates `torch.load` but where individual tensors are
+# only loaded into memory on demand, not all at once.
+# PyTorch can't do this natively as of time of writing:
+# - https://github.com/pytorch/pytorch/issues/64327
+# This allows us to de-shard without multiplying RAM usage, and also
+# conveniently drops the PyTorch dependency (though we still need numpy).
+
+
+@dataclass
+class LazyStorageKind:
+    data_type: DataType
+
+
+@dataclass
+class LazyStorage:
+    load: Callable[[int, int], NDArray]
+    kind: LazyStorageKind
+    description: str
+
+
+class LazyUnpickler(pickle.Unpickler):
+    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
+        super().__init__(fp)
+        self.data_base_path = data_base_path
+        self.zip_file = zip_file
+
+    def persistent_load(self, pid: Any) -> Any:
+        assert pid[0] == 'storage'
+        assert isinstance(pid[1], LazyStorageKind)
+        data_type = pid[1].data_type
+        filename_stem = pid[2]
+        filename = self.data_base_path + '/' + filename_stem
+        info = self.zip_file.getinfo(filename)
+
+        def load(offset: int, elm_count: int) -> NDArray:
+            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
+            if dtype is None:
+                raise Exception("tensor stored in unsupported format")
+            fp = self.zip_file.open(info)
+            fp.seek(offset * dtype.itemsize)
+            size = elm_count * dtype.itemsize
+            data = fp.read(size)
+            assert len(data) == size
+            return np.frombuffer(data, dtype)
+        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
+        return LazyStorage(load=load, kind=pid[1], description=description)
+
+    # @staticmethod
+    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
+                               # pyright: ignore[reportSelfClsParameterName]
+                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
+        assert isinstance(storage, LazyStorage)
+
+        def load() -> UnquantizedTensor:
+            elm_count = stride[0] * size[0]
+            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
+        description = f'pickled storage_offset={storage_offset} in {storage.description}'
+        return LazyTensor(load, list(size), storage.kind.data_type, description)
+
+    # @staticmethod
+    def rebuild_from_type_v2(func, new_type, args, state):
+        return func(*args)
+
+    CLASSES: Dict[Any, Any] = {
+        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
+        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
+        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
+        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
+        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
+        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
+        ('torch', 'Tensor'): LazyTensor,
+    }
+
+    def find_class(self, module: str, name: str) -> Any:
+        if not module.startswith('torch'):
+            return super().find_class(module, name)
+        return self.CLASSES[(module, name)]
+
+
+def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
+    zf = zipfile.ZipFile(outer_fp)
+    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
+    assert len(pickle_paths) == 1, pickle_paths
+    pickle_fp = zf.open(pickle_paths[0], 'r')
+    unpickler = LazyUnpickler(pickle_fp,
+                              data_base_path=pickle_paths[0][:-4],
+                              zip_file=zf)
+    model = unpickler.load()
+    as_dict = dict(model.items())
+    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
+
+
+def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
+    header_size, = struct.unpack('<Q', fp.read(8))
+    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
+    # Use mmap for the actual data to avoid race conditions with the file offset.
+    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
+    byte_buf = mapped[8 + header_size:]
+
+    def convert(info: Dict[str, Any]) -> LazyTensor:
+        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
+        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
+        shape: List[int] = info['shape']
+        begin, end = info['data_offsets']
+        assert 0 <= begin <= end <= len(byte_buf)
+        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
+        buf = byte_buf[begin:end]
+
+        def load() -> UnquantizedTensor:
+            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
+        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
+        return LazyTensor(load, shape, data_type, description)
+    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
+    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
+
+
+def must_read(fp: IO[bytes], length: int) -> bytes:
+    ret = fp.read(length)
+    if len(ret) < length:
+        raise Exception("unexpectedly reached end of file")
+    return ret
+
+
+@functools.lru_cache(maxsize=None)
+def lazy_load_file(path: Path) -> ModelPlus:
+    fp = open(path, 'rb')
+    first8 = fp.read(8)
+    fp.seek(0)
+    if first8[:2] == b'PK':
+        # A zip file, i.e. PyTorch format
+        return lazy_load_torch_file(fp, path)
+    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
+        # Probably safetensors
+        return lazy_load_safetensors_file(fp, path)
+    else:
+        raise ValueError(f"unknown format: {path}")
+
+
+In = TypeVar('In')
+Out = TypeVar('Out')
+
+def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
+    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
+    fast enough, this will stop calling `func` at some point rather than
+    letting results pile up in memory.  Specifically, there is a max of one
+    output value buffered per thread.'''
+    with concurrent.futures.ThreadPoolExecutor() as executor:
+        futures: List[concurrent.futures.Future[Out]] = []
+        items_rev = list(iterable)[::-1]
+        for i in range(min(concurrency, len(items_rev))):
+            futures.append(executor.submit(func, items_rev.pop()))
+        while futures:
+            result = futures.pop(0).result()
+            if items_rev:
+                futures.append(executor.submit(func, items_rev.pop()))
+            yield result
+
+
+def check_vocab_size(params: Params, vocab: Vocab) -> None:
+    if params.n_vocab != vocab.vocab_size:
+        assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
+        if params.n_vocab == vocab.vocab_size_base:
+            print("Ignoring added_tokens.json since model matches vocab size without it.")
+            vocab.added_tokens_list = []
+            vocab.vocab_size = vocab.vocab_size_base
+            return
+        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
+        if vocab.fname_added_tokens is not None:
+            msg += f" combined with {vocab.fname_added_tokens}"
+        msg += f" has {vocab.vocab_size})."
+        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
+            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
+        raise Exception(msg)
+
+
+class OutputFile:
+    def __init__(self, fname_out: Path) -> None:
+        self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
+
+    def add_meta_arch(self, params: Params) -> None:
+        self.gguf.add_context_length      (params.n_ctx)
+        self.gguf.add_embedding_length    (params.n_embd)
+        self.gguf.add_block_count         (params.n_layer)
+        self.gguf.add_feed_forward_length (params.n_ff)
+        self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
+        self.gguf.add_head_count          (params.n_head)
+        self.gguf.add_head_count_kv       (params.n_head_kv)
+        self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
+
+    def add_meta_vocab(self, vocab: Vocab) -> None:
+        tokens = []
+        scores = []
+        for text, score in vocab.all_tokens():
+            tokens.append(text)
+            scores.append(score)
+
+        self.gguf.add_tokenizer_model("llama")
+        self.gguf.add_token_list(tokens)
+        self.gguf.add_token_scores(scores)
+        #self.gguf.add_token_types(toktypes) # TODO: add this
+
+        # TODO: added / special tokens
+
+    def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
+        n_elements = 1
+        for dim in tensor.shape:
+            n_elements *= dim
+        data_type = DATA_TYPE_TO_NUMPY[tensor.data_type]
+        data_nbytes = n_elements * data_type.itemsize
+        self.gguf.add_tensor_info(name, tensor.shape, data_type, data_nbytes)
+
+    def write_meta(self) -> None:
+        self.gguf.write_header_to_file()
+        self.gguf.write_kv_data_to_file()
+
+    def write_tensor_info(self) -> None:
+        self.gguf.write_ti_data_to_file()
+
+    def close(self) -> None:
+        self.gguf.close()
+
+    @staticmethod
+    def write_vocab_only(fname_out: Path, params: Params, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+
+        of = OutputFile(fname_out)
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+        of.write_meta()
+
+        of.close()
+
+    @staticmethod
+    def write_all(fname_out: Path, params: Params, model: LazyModel, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+
+        of = OutputFile(fname_out)
+
+        # meta data
+        of.add_meta_arch(params)
+        of.add_meta_vocab(vocab)
+
+        # tensor info
+        for name, lazy_tensor in model.items():
+            of.add_tensor_info(name, lazy_tensor)
+
+        of.write_meta()
+        of.write_tensor_info()
+
+        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
+            name, lazy_tensor = item
+            return lazy_tensor.load().to_ggml().ndarray
+
+        # tensor data
+        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=1)
+        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
+            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
+            padi = len(str(len(model)))
+            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
+            of.gguf.write_tensor_data(ndarray)
+
+        of.close()
+
+def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
+    wq_type = model[NAMES[gguf.MODEL_TENSOR.ATTN_Q].format(bid=0)+".weight"].data_type
+
+    if output_type_str == "f32" or (output_type_str is None and wq_type == DT_F32):
+        return GGMLFileType.AllF32
+    if output_type_str == "f16" or (output_type_str is None and wq_type in (DT_F16, DT_BF16)):
+        return GGMLFileType.MostlyF16
+
+    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
+
+    raise Exception(f"Unexpected combination of types: {name_to_type}")
+
+def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
+    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
+            for (name, tensor) in model.items()}
+
+def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
+    tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
+
+    tmp = model
+
+    # HF models permut or pack some of the tensors, so we need to undo that
+    for i in itertools.count():
+        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
+            print(f"Permuting layer {i}")
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
+           #tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
+        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
+            print(f"Unpacking and permuting layer {i}")
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
+        else:
+            break
+
+    out: LazyModel = {}
+    for name, lazy_tensor in model.items():
+        name_new = name
+
+        if name in tmap:
+            name_new = tmap[name]
+        elif name.endswith(".weight") and name[:-7] in tmap:
+            name_new = tmap[name[:-7]] + ".weight"
+        elif name.endswith(".bias") and name[:-5] in tmap:
+            name_new = tmap[name[:-5]] + ".bias"
+        else:
+            raise Exception(f"Unexpected tensor name: {name}")
+
+        if gguf.should_skip_tensor_TMP(ARCH, params.n_layer, name_new):
+            print(f"skipping tensor {name_new}")
+            continue
+        else:
+            print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
+            out[name_new] = lazy_tensor
+
+    return out
+
+def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the nth path in the model.
+    '''
+    # Support the following patterns:
+    patterns: List[Tuple[str, str]] = [
+        # - x.00.pth, x.01.pth, etc.
+        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
+        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
+        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
+        # x.bin, x.bin.1, etc.
+        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
+    ]
+    for regex, replacement in patterns:
+        if re.search(regex, path.name):
+            new_path = path.with_name(re.sub(regex, replacement, path.name))
+            if new_path.exists():
+                return new_path
+    return None
+
+
+def find_multifile_paths(path: Path) -> List[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the whole list of paths in the model.
+    '''
+    ret: List[Path] = []
+    for i in itertools.count():
+        nth_path = nth_multifile_path(path, i)
+        if nth_path is None:
+            break
+        ret.append(nth_path)
+    if not ret:
+        # No matches.  This should only happen if the file was named, e.g.,
+        # foo.0, and there was no file named foo.  Oh well, try to process it
+        # as a single file.
+        return [path]
+    return ret
+
+
+def load_some_model(path: Path) -> ModelPlus:
+    '''Load a model of any supported format.'''
+    # Be extra-friendly and accept either a file or a directory:
+    if path.is_dir():
+        # Check if it's a set of safetensors files first
+        files = list(path.glob("model-00001-of-*.safetensors"))
+        if not files:
+            # Try the PyTorch patterns too, with lower priority
+            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
+            files = [file for glob in globs for file in path.glob(glob)]
+        if not files:
+            raise Exception(f"Can't find model in directory {path}")
+        if len(files) > 1:
+            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
+        path = files[0]
+
+    paths = find_multifile_paths(path)
+    models_plus: List[ModelPlus] = []
+    for path in paths:
+        print(f"Loading model file {path}")
+        models_plus.append(lazy_load_file(path))
+
+    model_plus = merge_multifile_models(models_plus)
+    return model_plus
+
+
+def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
+    # Be extra-friendly and accept either a file or a directory.  Also, if it's
+    # a directory, it might be the model directory, and tokenizer.model might
+    # be in the parent of that.
+    if path.is_dir():
+        vocab_file = "tokenizer.model"
+        if vocabtype == 'bpe':
+            vocab_file = "vocab.json"
+        path2 = path / vocab_file
+        # Use `.parent` instead of /.. to handle the symlink case better.
+        path3 = path.parent / vocab_file
+        if path2.exists():
+            path = path2
+        elif path3.exists():
+            path = path3
+        else:
+            raise FileNotFoundError(
+                f"Could not find tokenizer.model in {path} or its parent; "
+                "if it's in another directory, pass the directory as --vocab-dir")
+
+    print(f"Loading vocab file '{path}', type '{vocabtype}'")
+
+    added_tokens_path = path.parent / "added_tokens.json"
+    if vocabtype == "bpe":
+        return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    elif vocabtype == "spm":
+        return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None)
+    else:
+        raise ValueError(f"Unsupported vocabulary type {vocabtype}")
+
+
+def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
+    namestr = {
+        GGMLFileType.AllF32:    "f32",
+        GGMLFileType.MostlyF16: "f16",
+    }[file_type]
+    ret = model_paths[0].parent / f"ggml-model-{namestr}.gguf"
+    if ret in model_paths:
+        sys.stderr.write(
+            f"Error: Default output path ({ret}) would overwrite the input. "
+            "Please explicitly specify a path using --outfile.\n")
+        sys.exit(1)
+    return ret
+
+
+def do_dump_model(model_plus: ModelPlus) -> None:
+    print(f"model_plus.paths = {model_plus.paths!r}")
+    print(f"model_plus.format = {model_plus.format!r}")
+    print(f"model_plus.vocab = {model_plus.vocab!r}")
+    for name, lazy_tensor in model_plus.model.items():
+        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
+
+
+def main(args_in: Optional[List[str]] = None) -> None:
+    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
+    parser.add_argument("--dump",        action="store_true",    help="don't convert, just show what's in the model")
+    parser.add_argument("--dump-single", action="store_true",    help="don't convert, just show what's in a single model file")
+    parser.add_argument("--vocab-only",  action="store_true",    help="extract only the vocab")
+    parser.add_argument("--outtype",     choices=["f32", "f16"], help="output format (default: based on input)")
+    parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
+    parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
+    parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
+    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)", default="spm")
+    parser.add_argument("--ctx",         type=int,               help="model training context (default: based on input)")
+    args = parser.parse_args(args_in)
+
+    if args.dump_single:
+        model_plus = lazy_load_file(args.model)
+        do_dump_model(model_plus)
+
+    model_plus = load_some_model(args.model)
+
+    params = Params.load(model_plus)
+    if params.n_ctx == -1:
+        if args.ctx is None:
+            raise Exception("The model doesn't have a context size, and you didn't specify one with --ctx\n"
+                            "Please specify one with --ctx:\n"
+                            " - LLaMA v1: --ctx 2048\n"
+                            " - LLaMA v2: --ctx 4096\n")
+        params.n_ctx = args.ctx
+
+    print(f"params = {params}")
+
+    vocab: Vocab
+    if args.vocab_only:
+        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
+        assert args.outfile, "need --outfile if using --vocab-only"
+        outfile = args.outfile
+        OutputFile.write_vocab_only(outfile, params, vocab)
+        print(f"Wrote {outfile}")
+    else:
+        if args.dump:
+            do_dump_model(model_plus)
+            return
+
+        if model_plus.vocab is not None and args.vocab_dir is None:
+            vocab = model_plus.vocab
+        else:
+            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
+            vocab = load_vocab(vocab_dir, args.vocabtype)
+
+        model       = model_plus.model
+        model       = convert_model_names(model, params)
+        output_type = pick_output_type(model, args.outtype)
+        model       = convert_to_output_type(model, output_type)
+        outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
+
+        OutputFile.write_all(outfile, params, model, vocab)
+        print(f"Wrote {outfile}")
+
+
+if __name__ == '__main__':
+    main()

From 4f92488dd655d1ec768a8030ad1cd2e4b812a315 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 03:44:16 +0200
Subject: [PATCH 224/242] convert-permute-debug-master.py : permute debug for
 master

---
 convert-permute-debug-master.py | 1327 +++++++++++++++++++++++++++++++
 1 file changed, 1327 insertions(+)
 create mode 100644 convert-permute-debug-master.py

diff --git a/convert-permute-debug-master.py b/convert-permute-debug-master.py
new file mode 100644
index 0000000000000..7d64b22529c17
--- /dev/null
+++ b/convert-permute-debug-master.py
@@ -0,0 +1,1327 @@
+#!/usr/bin/env python
+import argparse
+import concurrent.futures
+import copy
+import enum
+import faulthandler
+import functools
+import io
+import itertools
+import json
+import math
+import mmap
+import pickle
+import re
+import signal
+import struct
+import sys
+import zipfile
+from abc import ABCMeta, abstractmethod
+from dataclasses import dataclass
+from pathlib import Path
+from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List,
+                    Literal, Optional, Sequence, Tuple, TypeVar, Union)
+
+import numpy as np
+from sentencepiece import SentencePieceProcessor  # type: ignore
+
+if TYPE_CHECKING:
+    from typing_extensions import TypeAlias
+
+if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
+    faulthandler.register(signal.SIGUSR1)
+
+NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
+
+
+@dataclass(frozen=True)
+class UnquantizedDataType:
+    name: str
+
+
+DT_F16 = UnquantizedDataType('F16')
+DT_F32 = UnquantizedDataType('F32')
+DT_I32 = UnquantizedDataType('I32')
+DT_BF16 = UnquantizedDataType('BF16')
+
+
+@dataclass(frozen=True)
+class QuantizedDataType:
+    groupsize: int
+    have_addends: bool
+    have_g_idx: bool
+
+
+DT_Q4_0 = QuantizedDataType(groupsize=32, have_addends=False, have_g_idx=False)
+DT_Q4_1 = QuantizedDataType(groupsize=32, have_addends=True, have_g_idx=False)
+
+DataType = Union[UnquantizedDataType, QuantizedDataType]
+
+DATA_TYPE_TO_FTYPE: Dict[DataType, int] = {
+    DT_F32: 0,
+    DT_F16: 1,
+    DT_Q4_0: 2,
+    DT_Q4_1: 3,
+}
+
+FTYPE_TO_DATA_TYPE: Dict[int, DataType] = \
+    {ftype: dtype for (dtype, ftype) in DATA_TYPE_TO_FTYPE.items()}
+
+DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
+    DT_BF16: np.dtype(np.uint16),
+    DT_F16: np.dtype(np.float16),
+    DT_F32: np.dtype(np.float32),
+    DT_I32: np.dtype(np.int32),
+}
+
+NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
+    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
+
+
+class GGMLFileType(enum.Enum):
+    AllF32 = 0
+    MostlyF16 = 1  # except 1d tensors
+    MostlyQ4_0 = 2  # except 1d tensors
+    MostlyQ4_1 = 3  # except 1d tensors
+    PerLayerIsQ4_1 = 4  # but tok_embeddings.weight and output.weight are F16
+
+    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
+        if len(tensor.shape) == 1:
+            # 1D tensors are always F32.
+            return DT_F32
+        elif self == GGMLFileType.AllF32:
+            return DT_F32
+        elif self == GGMLFileType.MostlyF16:
+            return DT_F16
+        elif self == GGMLFileType.MostlyQ4_0:
+            return DT_Q4_0
+        elif self == GGMLFileType.MostlyQ4_1:
+            return DT_Q4_1
+        elif self == GGMLFileType.PerLayerIsQ4_1:
+            if name in ('output.weight', 'tok_embeddings.weight'):
+                return DT_F16
+            else:
+                return DT_Q4_1
+        else:
+            raise ValueError(self)
+
+
+def make_tensors_list() -> List[str]:
+    ret = [
+        'tok_embeddings.weight',
+        'norm.weight',
+        'output.weight',
+    ]
+    for i in range(80):  # maximum number of layer
+        ret += [
+            f'layers.{i}.attention.wq.weight',
+            f'layers.{i}.attention.wk.weight',
+            f'layers.{i}.attention.wv.weight',
+            f'layers.{i}.attention.wo.weight',
+            f'layers.{i}.attention_norm.weight',
+            f'layers.{i}.feed_forward.w1.weight',
+            f'layers.{i}.feed_forward.w2.weight',
+            f'layers.{i}.feed_forward.w3.weight',
+            f'layers.{i}.ffn_norm.weight',
+        ]
+    return ret
+
+
+TENSORS_LIST = make_tensors_list()
+TENSORS_SET = set(TENSORS_LIST)
+
+
+def find_n_mult(n_ff: int, n_embd: int) -> int:
+    # hardcoded magic range
+    for n_mult in range(8192, 1, -1):
+        calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
+        if calc_ff == n_ff:
+            return n_mult
+    raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
+
+@dataclass
+class Params:
+    n_vocab:   int
+    n_embd:    int
+    n_mult:    int
+    n_head:    int
+    n_layer:   int
+    n_kv_head: Optional[int]  # This parameter is only used for Llama 2
+
+    @staticmethod
+    def guessed(model: 'LazyModel') -> 'Params':
+        # try transformer naming first
+        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
+
+        # try transformer naming first
+        if "model.layers.0.self_attn.q_proj.weight" in model:
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
+        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
+            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
+        else:
+            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
+
+        if n_layer < 1:
+            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
+                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
+
+        n_head=n_embd // 128 # guessed
+
+        return Params(
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = 256,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = None,
+        )
+
+    @staticmethod
+    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab = config["vocab_size"];
+        n_embd  = config["hidden_size"];
+        n_head  = config["num_attention_heads"];
+        n_layer = config["num_hidden_layers"];
+        n_ff    = config["intermediate_size"];
+        n_kv_head = config.get("num_key_value_heads")
+
+        n_mult = find_n_mult(n_ff, n_embd);
+
+        return Params(
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = n_mult,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = n_kv_head,
+        )
+
+    # LLaMA v2 70B params.json
+    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
+    @staticmethod
+    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
+        config = json.load(open(config_path))
+
+        n_vocab   = config["vocab_size"];
+        n_embd    = config["dim"];
+        n_head    = config["n_heads"];
+        n_layer   = config["n_layers"];
+        n_mult    = config["multiple_of"];
+
+        if n_vocab == -1:
+            n_vocab = model["tok_embeddings.weight"].shape[0]
+
+        return Params(
+            n_vocab   = n_vocab,
+            n_embd    = n_embd,
+            n_mult    = n_mult,
+            n_head    = n_head,
+            n_layer   = n_layer,
+            n_kv_head = None,
+        )
+
+    @staticmethod
+    def load(model_plus: 'ModelPlus') -> 'Params':
+        hf_config_path   = model_plus.paths[0].parent / "config.json"
+        orig_config_path = model_plus.paths[0].parent / "params.json"
+
+        if hf_config_path.exists():
+            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
+        elif orig_config_path.exists():
+            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
+        else:
+            params = Params.guessed(model_plus.model)
+
+        print(f'params: n_vocab:{params.n_vocab} n_embd:{params.n_embd} n_mult:{params.n_mult} n_head:{params.n_head} n_layer:{params.n_layer}')
+        return params
+
+
+class SentencePieceVocab:
+    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vocabtype: Optional[str]) -> None:
+        self.vocabtype = vocabtype
+        if self.vocabtype == "bpe":
+          self.sentencepiece_tokenizer = json.loads(open(str(fname_tokenizer)).read())
+        else:
+          self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
+        added_tokens: Dict[str, int]
+        if fname_added_tokens is not None:
+            added_tokens = json.load(open(fname_added_tokens))
+        else:
+            added_tokens = {}
+        if self.vocabtype == "bpe":
+          vocab_size: int = len(self.sentencepiece_tokenizer)
+        else:
+          vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
+        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids = sorted(added_tokens.values())
+        if expected_ids != actual_ids:
+            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
+        self.added_tokens_list = [text for (text, idx) in items]
+        self.vocab_size_base: int = vocab_size
+        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer = fname_tokenizer
+        self.fname_added_tokens = fname_added_tokens
+
+    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        tokenizer = self.sentencepiece_tokenizer
+        if self.vocabtype == "bpe":
+          from transformers.models.gpt2 import tokenization_gpt2
+          byte_encoder = tokenization_gpt2.bytes_to_unicode()
+          byte_decoder = {v: k for k, v in byte_encoder.items()}
+          for i, item in enumerate(tokenizer):
+            text: bytes
+            text = b''.join([x.to_bytes(1, byteorder='big') for x in [byte_decoder[y] for y in item]])
+            score: float = -i
+            yield text, score
+        else:
+          for i in range(tokenizer.vocab_size()):
+              text: bytes
+              if tokenizer.is_unknown(i):
+                  text = " \u2047 ".encode("utf-8")
+              elif tokenizer.is_control(i):
+                  text = b""
+              elif tokenizer.is_byte(i):
+                  piece = tokenizer.id_to_piece(i)
+                  if len(piece) != 6:
+                      raise Exception(f"Invalid token: {piece}")
+                  byte_value = int(piece[3:-1], 16)
+                  text = struct.pack("B", byte_value)
+              else:
+                  text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
+              score: float = tokenizer.get_score(i)
+              yield text, score
+
+    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        for text in self.added_tokens_list:
+            score = -1000.0
+            yield text.encode("utf-8"), score
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        yield from self.sentencepiece_tokens()
+        yield from self.added_tokens()
+
+    def __repr__(self) -> str:
+        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
+
+
+class GGMLVocab:
+    def __init__(self, tokens: List[Tuple[bytes, float]]):
+        self.tokens = tokens
+        self.vocab_size = len(tokens)
+
+    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+        return self.tokens
+
+    def __repr__(self) -> str:
+        return f"<GGMLVocab with {self.vocab_size} tokens>"
+
+
+Vocab = Union[SentencePieceVocab, GGMLVocab]
+
+
+def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
+    print( "permute " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_kv_head) )
+    if n_kv_head is not None and n_head != n_kv_head:
+        n_head //= n_kv_head
+    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
+                .swapaxes(1, 2)
+                .reshape(weights.shape))
+
+
+def dequantize_q4(qvalues_pack32: NDArray, scales: NDArray, addends: Optional[NDArray], g_idx: Optional[NDArray]) -> NDArray:
+    # First reinterpret each row from a list of int32s containing 8 values each
+    # to a list of uint8s containing 2 values each.
+    qvalues_pack8 = qvalues_pack32.view(np.uint8)
+
+    # Then split out the two values per int8 (which requires an actual
+    # conversion because numpy doesn't natively support int4s).
+    qvalues = np.zeros([qvalues_pack8.shape[0], qvalues_pack8.shape[1] * 2], dtype=np.uint8)
+    qvalues[:, 0::2] = qvalues_pack8 & 0xf
+    qvalues[:, 1::2] = qvalues_pack8 >> 4
+
+    assert addends is None or addends.shape == scales.shape
+    assert qvalues.shape[0] == scales.shape[0]
+    assert qvalues.shape[1] % scales.shape[1] == 0
+    if g_idx is None:
+        repeat_count = qvalues.shape[1] // scales.shape[1]
+        scales = scales[:, :, np.newaxis]
+        if addends is not None:
+            addends = addends[:, :, np.newaxis]
+        # Reshape so that the below computation broadcasts over scales and addends:
+        qvalues.shape = (qvalues.shape[0], scales.shape[1], int(repeat_count))
+    else:
+        # In this case the scale and addend is selected for each column by g_idx:
+        assert addends is not None
+        scales = scales[:, g_idx]
+        addends = addends[:, g_idx]
+    if addends is None:
+        # Q4_0
+        qvalues = qvalues.view(np.int8)
+        qvalues -= 8
+    # And do the actual 'value = scale * qvalue + addend' computation.
+    values = scales * qvalues
+    if addends is not None:
+        values += addends
+    if g_idx is None:
+        values.shape = (values.shape[0], values.shape[1] * values.shape[2])
+    return values
+
+
+class Tensor(metaclass=ABCMeta):
+    data_type: DataType
+
+    @abstractmethod
+    def astype(self, data_type: DataType) -> 'Tensor': ...
+    @abstractmethod
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'Tensor': ...
+    @abstractmethod
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def part(self, n_part: int) -> 'UnquantizedTensor': ...
+    @abstractmethod
+    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
+
+
+def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
+    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
+    fp32_arr = bf16_arr.astype(np.uint32) << 16
+    return fp32_arr.view(np.float32)
+
+
+class UnquantizedTensor(Tensor):
+    def __init__(self, ndarray: NDArray) -> None:
+        assert isinstance(ndarray, np.ndarray)
+        self.ndarray = ndarray
+        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
+
+    def astype(self, data_type: DataType) -> Tensor:
+        dtype = DATA_TYPE_TO_NUMPY[data_type]
+        if self.data_type == DT_BF16:
+            self.ndarray = bf16_to_fp32(self.ndarray)
+        return UnquantizedTensor(self.ndarray.astype(dtype))
+
+    def to_ggml(self) -> 'UnquantizedTensor':
+        return self
+
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
+
+    def part(self, n_part: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
+
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'UnquantizedTensor':
+        return UnquantizedTensor(permute(self.ndarray, n_head, n_kv_head))
+
+
+def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
+    tensor = lazy_tensor.load()
+    assert isinstance(tensor, UnquantizedTensor)
+
+    # double-check:
+    actual_shape = list(tensor.ndarray.shape)
+    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
+    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
+        if convert:
+            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
+        else:
+            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
+
+    return tensor.ndarray
+
+
+class GGMLQuantizedTensor(Tensor):
+    data_type: QuantizedDataType
+
+    def __init__(self, ndarray: NDArray, shape: List[int], data_type: DataType) -> None:
+        rows, columns = shape
+        assert data_type in (DT_Q4_1, DT_Q4_0)  # for now
+        assert isinstance(data_type, QuantizedDataType)  # redundant, but mypy complains without this
+        assert columns % data_type.groupsize == 0
+        words_in_block = 6 if data_type == DT_Q4_1 else 5
+        self.ndarray = ndarray.view(dtype=np.uint32).reshape((rows, columns // data_type.groupsize, words_in_block))
+        self.shape = shape[:]
+        self.data_type = data_type
+
+    def astype(self, data_type: DataType) -> Tensor:
+        if data_type == self.data_type:
+            return self
+        scales = self.ndarray[:, :, 0].view(np.float32)
+        if self.data_type.have_addends:
+            addends = self.ndarray[:, :, 1].view(np.float32)
+        else:
+            addends = None
+        qweights = self.ndarray[:, :, -4:].reshape([self.shape[0], self.shape[1] // 8])
+
+        dq = dequantize_q4(qweights, scales, addends, g_idx=None)
+        return UnquantizedTensor(dq).astype(data_type)
+
+    def to_ggml(self) -> 'GGMLQuantizedTensor':
+        return self
+
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'GGMLQuantizedTensor':
+        return GGMLQuantizedTensor(permute(self.ndarray, n_head, n_kv_head), self.shape, self.data_type)
+
+    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
+
+    def part(self, n_part: int) -> 'UnquantizedTensor':
+        r = self.ndarray.shape[0] // 3
+        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
+
+GGMLCompatibleTensor = Union[UnquantizedTensor, GGMLQuantizedTensor]
+
+
+class DeferredPermutedTensor(Tensor):
+    def __init__(self, base: Tensor, n_head: int, n_kv_head: Optional[int] = None) -> None:
+        self.base = base
+        self.n_head = n_head
+        self.n_kv_head = n_kv_head
+        self.data_type = self.base.data_type
+
+    def astype(self, data_type: DataType) -> Tensor:
+        return self.base.astype(data_type).permute(self.n_head, self.n_kv_head)
+
+    def to_ggml(self) -> GGMLCompatibleTensor:
+        return self.base.to_ggml().permute(self.n_head, self.n_kv_head)
+
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
+        raise Exception("shouldn't permute twice")
+
+
+class GPTQForLLaMaQuantizedTensor(Tensor):
+    def __init__(self, model: 'LazyModel', namebase: str) -> None:
+        qweight = load_unquantized(model[f"{namebase}.qweight"], np.int32)
+        scales = load_unquantized(model[f"{namebase}.scales"], np.float32, convert=True)
+
+        bias = model.get(f"{namebase}.bias")
+        if bias is not None:
+            # Q4_1 does not support bias; good thing the bias is always all zeros.
+            assert not np.any(load_unquantized(bias))
+
+        if f"{namebase}.zeros" in model:
+            zeros = load_unquantized(model[f"{namebase}.zeros"], np.float32)
+        else:
+            qzeros = load_unquantized(model[f"{namebase}.qzeros"], np.int32)
+            assert qzeros.dtype == np.int32
+            zeros = dequantize_q4(qzeros, scales, scales, g_idx=None)
+            assert zeros.dtype == np.float32
+
+        assert zeros.shape == scales.shape
+
+        # Output is transposed compared to the input, and addends have their sign flipped.
+        # Scales and zeros similarly must be transposed but only for newer
+        # versions of GPTQ-for-LLaMa; the older versions can be identified by
+        # having shape (n_embd, 1).
+        qweight = qweight.T
+        if scales.shape[1] != 1:
+            scales = scales.T
+            zeros = zeros.T
+
+        # Output also has signs flipped for the addends.
+        self.qweight = qweight
+        self.scales = scales
+        self.addends = -zeros
+
+        self.g_idx: Optional[NDArray]
+        if f"{namebase}.g_idx" in model:
+            self.g_idx = load_unquantized(model[f"{namebase}.g_idx"], np.int32)
+            assert self.g_idx.shape == (qweight.shape[1] * 8,)
+        else:
+            self.g_idx = None
+
+        self.shape = [self.qweight.shape[0], self.qweight.shape[1] * 8]
+        self.data_type = QuantizedDataType(groupsize=self.groupsize(), have_addends=True,
+                                           have_g_idx=(self.g_idx is not None))
+
+    def inspect(self, row: int, col: int) -> None:
+        '''For debugging.'''
+        qweight = (self.qweight[row, col // 8] >> (4 * (col & 7))) & 0xf
+        if self.g_idx is not None:
+            group = self.g_idx[col]
+        else:
+            group = int(col // self.groupsize())
+        scale = self.scales[row, group]
+        addend = self.addends[row, group]
+        with np.printoptions(precision=None, suppress=True):
+            print(f'scale:{scale} addend:{addend} qweight:{qweight}')
+            print('possible values:', np.arange(16) * scale + addend)
+            print('actual value:', qweight * scale + addend)
+
+    def astype(self, data_type: DataType) -> Tensor:
+        if isinstance(data_type, QuantizedDataType):
+            assert self.g_idx is None and data_type.have_addends is True and data_type.have_g_idx is False
+            return self.regroup(data_type.groupsize)
+
+        dequantized = dequantize_q4(np.ascontiguousarray(self.qweight), self.scales, self.addends, self.g_idx)
+        return UnquantizedTensor(dequantized).astype(data_type)
+
+    def groupsize(self) -> int:
+        assert self.addends.shape == self.scales.shape
+        assert self.shape[1] % self.scales.shape[1] == 0
+        return self.shape[1] // self.scales.shape[1]
+
+    def regroup(self, new_groupsize: int = 32) -> 'GPTQForLLaMaQuantizedTensor':
+        # Old versions of GPTQ-for-LLaMa shared scales and addends between all the
+        # columns in a row.  Newer versions share them between every set of N
+        # columns in a row, where N is the `groupsize` parameter, usually 128.  The
+        # output format shares them between every set of 32 columns.  To handle
+        # this, duplicate scales and addends for every smaller group.
+        # (In the above, 'row' and 'column' are in the sense of the output.)
+        assert self.g_idx is None
+        old_groupsize = self.groupsize()
+        assert old_groupsize >= new_groupsize and old_groupsize % new_groupsize == 0, old_groupsize
+        ret = copy.copy(self)
+        ret.addends = self.addends.repeat(old_groupsize // new_groupsize, axis=1)
+        ret.scales = self.scales.repeat(old_groupsize // new_groupsize, axis=1)
+        ret.data_type = QuantizedDataType(groupsize=new_groupsize, have_addends=True, have_g_idx=False)
+        return ret
+
+    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
+        return DeferredPermutedTensor(self, n_head, n_kv_head)
+
+    def to_ggml(self) -> GGMLQuantizedTensor:
+        # The output format looks like this:
+        # For each row:
+        #   For each group of 32 columns:
+        #     - addend (float32, 4 bytes)
+        #     - scale (float32, 4 bytes)
+        #     - weights (int4 * 32, 16 bytes)
+
+        if self.groupsize() != 32:
+            raise Exception("should have been regrouped before converting to ggml")
+
+        # Since the output format is mixed between integers and floats, we have
+        # to hackily view the floats as int32s just so numpy will let us
+        # concatenate them.
+        addends_view = self.addends.view(dtype=np.int32)[:, :, np.newaxis]
+        scales_view = self.scales.view(dtype=np.int32)[:, :, np.newaxis]
+
+        # Split into groups of 4 columns (i.e. 32 columns of quantized data):
+        grouped = self.qweight.reshape([self.qweight.shape[0], self.qweight.shape[1] // 4, 4])
+
+        # And concatenate:
+        grouped = np.concatenate([scales_view, addends_view, grouped], axis=2, casting='no')
+
+        return GGMLQuantizedTensor(grouped, self.shape, DT_Q4_1)
+
+
+@dataclass
+class LazyTensor:
+    _load: Callable[[], Tensor]
+    shape: List[int]
+    data_type: DataType
+    description: str
+
+    def load(self) -> Tensor:
+        ret = self._load()
+        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
+        return ret
+
+    def astype(self, data_type: DataType) -> 'LazyTensor':
+        self.validate_conversion_to(data_type)
+
+        def load() -> Tensor:
+            return self.load().astype(data_type)
+        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
+
+    def validate_conversion_to(self, data_type: DataType) -> None:
+        if data_type == self.data_type:
+            return
+        if isinstance(data_type, QuantizedDataType):
+            if not isinstance(self.data_type, QuantizedDataType):
+                raise Exception(f"Can't turn an unquantized tensor into a quantized type ({data_type})")
+            if self.data_type.have_g_idx:
+                sys.stderr.write(
+                    "Error: Input uses the newer GPTQ-for-LLaMa format (using g_idx), "
+                    "which is not yet natively supported by GGML. "
+                    "For now you can still convert this model by passing `--outtype f16` to dequantize, "
+                    "but that will result in a much larger output file for no quality benefit.\n")
+                sys.exit(1)
+            assert not data_type.have_g_idx and self.data_type.have_addends and data_type.have_addends
+
+
+LazyModel = Dict[str, LazyTensor]
+
+
+@dataclass
+class ModelPlus:
+    model: LazyModel
+    paths: List[Path]  # Where this was read from.
+    format: Literal['ggml', 'torch', 'safetensors']
+    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
+
+
+def merge_sharded(models: List[LazyModel]) -> LazyModel:
+    # Original LLaMA models have each file contain one part of each tensor.
+    # Use a dict instead of a set to preserve order.
+    names = {name: None for model in models for name in model}
+
+    def convert(name: str) -> LazyTensor:
+        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
+        if len(lazy_tensors) == 1:
+            # only one file; don't go through this procedure since there might
+            # be quantized tensors
+            return lazy_tensors[0]
+        if len(lazy_tensors[0].shape) == 1:
+            # the tensor is just duplicated in every file
+            return lazy_tensors[0]
+        if name.startswith('tok_embeddings.') or \
+           name.endswith('.attention.wo.weight') or \
+           name.endswith('.feed_forward.w2.weight'):
+            # split by columns
+            axis = 1
+        else:
+            # split by rows
+            axis = 0
+        concatenated_shape = list(lazy_tensors[0].shape)
+        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
+
+        def load() -> UnquantizedTensor:
+            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
+            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
+            return UnquantizedTensor(concatenated)
+        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
+        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
+    return {name: convert(name) for name in names}
+
+
+def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
+    formats = set(mp.format for mp in models_plus)
+    assert len(formats) == 1, "different formats?"
+    format = formats.pop()
+    paths = [path for mp in models_plus for path in mp.paths]
+    # Use the first non-None vocab, if any.
+    try:
+        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
+    except StopIteration:
+        vocab = None
+
+    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
+        # Transformers models put different tensors in different files, but
+        # don't split indivdual tensors between files.
+        model: LazyModel = {}
+        for mp in models_plus:
+            model.update(mp.model)
+    else:
+        model = merge_sharded([mp.model for mp in models_plus])
+
+    return ModelPlus(model, paths, format, vocab)
+
+
+def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_kv_head: Optional[int] = None) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute(n_head, n_kv_head)
+    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_kv_head}) ' + lazy_tensor.description)
+
+def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().permute_part(n_part, n_head)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
+
+def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
+    def load() -> Tensor:
+        return lazy_tensor.load().part(n_part)
+    s = lazy_tensor.shape.copy()
+    s[0] = s[0] // 3
+    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
+
+def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
+    out: LazyModel = {}
+    out["tok_embeddings.weight"] = model["model.embed_tokens.weight"]
+    out["norm.weight"] = model["model.norm.weight"]
+    out["output.weight"] = model["lm_head.weight"]
+
+    for i in itertools.count():
+        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
+            out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
+            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_kv_head)
+            out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
+        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
+            out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)
+            out[f"layers.{i}.attention.wk.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head)
+            out[f"layers.{i}.attention.wv.weight"] = part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
+        else:
+            break
+
+        out[f"layers.{i}.attention.wo.weight"] = model[f"model.layers.{i}.self_attn.o_proj.weight"]
+
+        out[f"layers.{i}.feed_forward.w1.weight"] = model[f"model.layers.{i}.mlp.gate_proj.weight"]
+        out[f"layers.{i}.feed_forward.w2.weight"] = model[f"model.layers.{i}.mlp.down_proj.weight"]
+        out[f"layers.{i}.feed_forward.w3.weight"] = model[f"model.layers.{i}.mlp.up_proj.weight"]
+
+        out[f"layers.{i}.attention_norm.weight"] = model[f"model.layers.{i}.input_layernorm.weight"]
+        out[f"layers.{i}.ffn_norm.weight"] = model[f"model.layers.{i}.post_attention_layernorm.weight"]
+    return out
+
+
+def handle_quantization(model: LazyModel) -> LazyModel:
+    '''Convert a model with entries for 'foo.qweight', 'foo.scales', etc.
+    (which resolve to UnquantizedTensors with the raw data) to one with entries
+    for 'foo.weight' (which resolve to QuantizedTensors).
+    '''
+    def convert(name: str) -> Tuple[str, LazyTensor]:
+        if name.endswith(".qweight"):
+            namebase = name.rsplit('.', 1)[0]
+            orig_name = namebase + ".weight"
+
+            lazy_tensor = model[name]
+            assert len(lazy_tensor.shape) == 2
+            real_shape = [lazy_tensor.shape[1], lazy_tensor.shape[0] * 8]
+
+            # Calculate type.  This replicates the logic in
+            # GPTQForLLaMaQuantizedTensor (which is executed when the modelis
+            # actually loaded).
+            lazy_scales = model[f"{namebase}.scales"]
+            scales_width = 1 if lazy_scales.shape[1] == 1 else lazy_scales.shape[0]
+            assert real_shape[1] % scales_width == 0
+            groupsize = real_shape[1] // scales_width
+            have_g_idx = f"{namebase}.g_idx" in model
+            data_type = QuantizedDataType(groupsize=groupsize, have_addends=True, have_g_idx=have_g_idx)
+
+            def load() -> Tensor:
+                return GPTQForLLaMaQuantizedTensor(model, namebase)
+
+            return (orig_name, LazyTensor(load, real_shape, data_type, '[quantized]'))
+        else:
+            return (name, model[name])
+    return dict(convert(name) for name in model)
+
+# Functionality that simulates `torch.load` but where individual tensors are
+# only loaded into memory on demand, not all at once.
+# PyTorch can't do this natively as of time of writing:
+# - https://github.com/pytorch/pytorch/issues/64327
+# This allows us to de-shard without multiplying RAM usage, and also
+# conveniently drops the PyTorch dependency (though we still need numpy).
+
+
+@dataclass
+class LazyStorageKind:
+    data_type: DataType
+
+
+@dataclass
+class LazyStorage:
+    load: Callable[[int, int], NDArray]
+    kind: LazyStorageKind
+    description: str
+
+
+class LazyUnpickler(pickle.Unpickler):
+    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
+        super().__init__(fp)
+        self.data_base_path = data_base_path
+        self.zip_file = zip_file
+
+    def persistent_load(self, pid: Any) -> Any:
+        assert pid[0] == 'storage'
+        assert isinstance(pid[1], LazyStorageKind)
+        data_type = pid[1].data_type
+        filename_stem = pid[2]
+        filename = self.data_base_path + '/' + filename_stem
+        info = self.zip_file.getinfo(filename)
+
+        def load(offset: int, elm_count: int) -> NDArray:
+            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
+            if dtype is None:
+                raise Exception("tensor stored in unsupported format")
+            fp = self.zip_file.open(info)
+            fp.seek(offset * dtype.itemsize)
+            size = elm_count * dtype.itemsize
+            data = fp.read(size)
+            assert len(data) == size
+            return np.frombuffer(data, dtype)
+        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
+        return LazyStorage(load=load, kind=pid[1], description=description)
+
+    # @staticmethod
+    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
+                               # pyright: ignore[reportSelfClsParameterName]
+                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
+        assert isinstance(storage, LazyStorage)
+
+        def load() -> UnquantizedTensor:
+            elm_count = stride[0] * size[0]
+            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
+        description = f'pickled storage_offset={storage_offset} in {storage.description}'
+        return LazyTensor(load, list(size), storage.kind.data_type, description)
+
+    # @staticmethod
+    def rebuild_from_type_v2(func, new_type, args, state):
+        return func(*args)
+
+    CLASSES: Dict[Any, Any] = {
+        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
+        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
+        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
+        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
+        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
+        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
+        ('torch', 'Tensor'): LazyTensor,
+    }
+
+    def find_class(self, module: str, name: str) -> Any:
+        if not module.startswith('torch'):
+            return super().find_class(module, name)
+        return self.CLASSES[(module, name)]
+
+
+def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
+    zf = zipfile.ZipFile(outer_fp)
+    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
+    assert len(pickle_paths) == 1, pickle_paths
+    pickle_fp = zf.open(pickle_paths[0], 'r')
+    unpickler = LazyUnpickler(pickle_fp,
+                              data_base_path=pickle_paths[0][:-4],
+                              zip_file=zf)
+    model = unpickler.load()
+    as_dict = dict(model.items())
+    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
+
+
+SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
+    'BF16': DT_BF16,
+    'F16': DT_F16,
+    'F32': DT_F32,
+    'I32': DT_I32,
+}
+
+
+def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
+    header_size, = struct.unpack('<Q', fp.read(8))
+    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
+    # Use mmap for the actual data to avoid race conditions with the file offset.
+    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
+    byte_buf = mapped[8 + header_size:]
+
+    def convert(info: Dict[str, Any]) -> LazyTensor:
+        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
+        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
+        shape: List[int] = info['shape']
+        begin, end = info['data_offsets']
+        assert 0 <= begin <= end <= len(byte_buf)
+        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
+        buf = byte_buf[begin:end]
+
+        def load() -> UnquantizedTensor:
+            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
+        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
+        return LazyTensor(load, shape, data_type, description)
+    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
+    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
+
+
+def must_read(fp: IO[bytes], length: int) -> bytes:
+    ret = fp.read(length)
+    if len(ret) < length:
+        raise Exception("unexpectedly reached end of file")
+    return ret
+
+
+def lazy_load_ggml_file(fp: io.BufferedReader, path: Path) -> ModelPlus:
+    magic = must_read(fp, 4)[::-1]
+    if magic in (b'ggmf', b'ggjt'):
+        version, = struct.unpack("i", must_read(fp, 4))
+        assert version == 1
+    else:
+        assert magic == b'ggml'
+        version = None
+    n_vocab, n_embd, n_mult, n_head, n_layer, rot, file_type = struct.unpack('<7i', must_read(fp, 28))
+
+    tokens: List[Tuple[bytes, float]] = []
+    for i in range(n_vocab):
+        if i == 32000:
+            # HACK: GPT4All messed with the format without changing the magic
+            # number.  Specifically, they changed the vocab section to contain
+            # `n_vocab - 1` tokens instead of `n_vocab` (i.e. omitting the
+            # extra pad token).  Try to detect if we're reading a file like
+            # this.
+            orig_pos = fp.tell()
+            fp.seek(20, io.SEEK_CUR)
+            is_gpt4all = fp.read(21) == b'tok_embeddings.weight'
+            fp.seek(orig_pos)
+            if is_gpt4all:
+                break
+
+        length, = struct.unpack("i", must_read(fp, 4))
+        text = must_read(fp, length)
+        if magic != b'ggml':
+            score, = struct.unpack("f", must_read(fp, 4))
+            tokens.append((text, score))
+    vocab = GGMLVocab(tokens) if magic != b'ggml' else None
+
+    model: LazyModel = {}
+    # Use mmap for the actual data to avoid race conditions with the file offset.
+    off = fp.raw.tell()
+    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
+    fp.raw.seek(off)  # needed on Windows
+
+    def read_tensor() -> None:  # this is a function so that variables captured in `load` don't change
+        shape_len, name_len, ftype = struct.unpack("iii", must_read(fp, 12))
+        assert 0 <= shape_len <= 3
+        shape: List[int] = list(struct.unpack(f"{shape_len}i", must_read(fp, 4 * shape_len)))
+        shape = shape[::-1]
+        name = must_read(fp, name_len).decode('utf-8')
+        data_type = FTYPE_TO_DATA_TYPE[ftype]
+
+        if magic == b'ggjt':
+            fp.seek((fp.tell() + 31) & -32)
+
+        if data_type == DT_Q4_1:
+            # See GPTQForLLaMaQuantizedTensor.ggml_ndarray()
+            size = 24 * (shape[1] // 32) * shape[0]
+        elif data_type == DT_Q4_0:
+            size = 20 * (shape[1] // 32) * shape[0]
+        else:
+            numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
+            elm_count = math.prod(shape)
+            size = elm_count * numpy_dtype.itemsize
+        offset = fp.tell()
+        buf = mapped[offset:offset+size]
+        fp.seek(size, io.SEEK_CUR)
+
+        def load() -> Tensor:
+            if isinstance(data_type, QuantizedDataType):
+                ndarray = np.frombuffer(buf, dtype=np.uint32)
+                return GGMLQuantizedTensor(ndarray, shape, data_type)
+            else:
+                return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
+        description = f'ggml offset={offset} type={data_type} path={path}'
+        model[name] = LazyTensor(load, shape, data_type, description)
+
+    while fp.read(1) != b'':
+        fp.seek(-1, io.SEEK_CUR)
+        read_tensor()
+
+    return ModelPlus(model=model, paths=[path], format='ggml', vocab=vocab)
+
+
+@functools.lru_cache(maxsize=None)
+def lazy_load_file(path: Path) -> ModelPlus:
+    fp = open(path, 'rb')
+    first8 = fp.read(8)
+    fp.seek(0)
+    if first8[:2] == b'PK':
+        # A zip file, i.e. PyTorch format
+        return lazy_load_torch_file(fp, path)
+    elif first8[2:4] == b'gg':
+        # GGML format
+        return lazy_load_ggml_file(fp, path)
+    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
+        # Probably safetensors
+        return lazy_load_safetensors_file(fp, path)
+    else:
+        raise ValueError(f"unknown format: {path}")
+
+
+In = TypeVar('In')
+Out = TypeVar('Out')
+
+
+def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
+    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
+    fast enough, this will stop calling `func` at some point rather than
+    letting results pile up in memory.  Specifically, there is a max of one
+    output value buffered per thread.'''
+    with concurrent.futures.ThreadPoolExecutor() as executor:
+        futures: List[concurrent.futures.Future[Out]] = []
+        items_rev = list(iterable)[::-1]
+        for i in range(min(concurrency, len(items_rev))):
+            futures.append(executor.submit(func, items_rev.pop()))
+        while futures:
+            result = futures.pop(0).result()
+            if items_rev:
+                futures.append(executor.submit(func, items_rev.pop()))
+            yield result
+
+
+def check_vocab_size(params: Params, vocab: Vocab) -> None:
+    if params.n_vocab != vocab.vocab_size:
+        # GGMLVocab comes from the same file as the model so shouldn't mismatch:
+        assert isinstance(vocab, SentencePieceVocab)
+        if params.n_vocab == vocab.vocab_size_base:
+            print("Ignoring added_tokens.json since model matches vocab size without it.")
+            vocab.added_tokens_list = []
+            vocab.vocab_size = vocab.vocab_size_base
+            return
+        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
+        if vocab.fname_added_tokens is not None:
+            msg += f" combined with {vocab.fname_added_tokens}"
+        msg += f" has {vocab.vocab_size})."
+        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
+            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
+        raise Exception(msg)
+
+
+class OutputFile:
+    def __init__(self, fname_out: Path) -> None:
+        self.fout = open(fname_out, "wb")
+
+    def write_file_header(self, params: Params, file_type: GGMLFileType) -> None:
+        self.fout.write(b"ggjt"[::-1])  # magic
+        values = [
+            1,  # file version
+            params.n_vocab,
+            params.n_embd,
+            params.n_mult,
+            params.n_head,
+            params.n_layer,
+            params.n_embd // params.n_head,  # rot (obsolete)
+            file_type.value,
+        ]
+        self.fout.write(struct.pack("i" * len(values), *values))
+
+    def write_tensor_header(self, name: str, shape: Sequence[int], data_type: DataType) -> None:
+        sname = name.encode('utf-8')
+        self.fout.write(struct.pack("iii", len(shape), len(sname), DATA_TYPE_TO_FTYPE[data_type]))
+        self.fout.write(struct.pack("i" * len(shape), *shape[::-1]))
+        self.fout.write(sname)
+        self.fout.seek((self.fout.tell() + 31) & -32)
+
+    def write_vocab(self, vocab: Vocab) -> None:
+        for text, score in vocab.all_tokens():
+            self.fout.write(struct.pack("i", len(text)))
+            self.fout.write(text)
+            self.fout.write(struct.pack("f", score))
+
+    @staticmethod
+    def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
+        of = OutputFile(fname_out)
+        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0)
+        of = OutputFile(fname_out)
+        of.write_file_header(params, file_type=GGMLFileType.AllF32)
+        of.write_vocab(vocab)
+        of.fout.close()
+
+    @staticmethod
+    def write_all(fname_out: Path, params: Params, file_type: GGMLFileType, model: LazyModel, vocab: Vocab) -> None:
+        check_vocab_size(params, vocab)
+        of = OutputFile(fname_out)
+        of.write_file_header(params, file_type)
+        print("Writing vocab...")
+        of.write_vocab(vocab)
+
+        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
+            name, lazy_tensor = item
+            return lazy_tensor.load().to_ggml().ndarray
+
+        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=1)
+        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
+            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
+            padi = len(str(len(model)))
+            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
+            of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
+            ndarray.tofile(of.fout)
+        of.fout.close()
+
+
+def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
+    wq_type = model["layers.0.attention.wq.weight"].data_type
+    if output_type_str == "f32" or (output_type_str is None and wq_type in (DT_F32, DT_BF16)):
+        return GGMLFileType.AllF32
+    if output_type_str == "f16" or (output_type_str is None and wq_type == DT_F16):
+        return GGMLFileType.MostlyF16
+    if output_type_str == "q4_1" or (output_type_str is None and isinstance(wq_type, QuantizedDataType) and
+                                     wq_type.have_addends):
+        if isinstance(model["output.weight"].data_type, QuantizedDataType):
+            return GGMLFileType.MostlyQ4_1
+        else:
+            return GGMLFileType.PerLayerIsQ4_1
+    if output_type_str == "q4_0" or (output_type_str is None and isinstance(wq_type, QuantizedDataType)):
+        return GGMLFileType.MostlyQ4_0
+    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
+    raise Exception(f"Unexpected combination of types: {name_to_type}")
+
+
+def do_necessary_conversions(model: LazyModel, params: Params) -> LazyModel:
+    model = handle_quantization(model)
+
+    if "lm_head.weight" in model:
+        model = convert_transformers_to_orig(model, params)
+    model = filter_and_sort_tensors(model)
+
+    return model
+
+
+def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
+    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
+            for (name, tensor) in model.items()}
+
+
+def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the nth path in the model.
+    '''
+    # Support the following patterns:
+    patterns: List[Tuple[str, str]] = [
+        # - x.00.pth, x.01.pth, etc.
+        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
+        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
+        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
+        # x.bin, x.bin.1, etc.
+        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
+    ]
+    for regex, replacement in patterns:
+        if re.search(regex, path.name):
+            new_path = path.with_name(re.sub(regex, replacement, path.name))
+            if new_path.exists():
+                return new_path
+    return None
+
+
+def find_multifile_paths(path: Path) -> List[Path]:
+    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
+    the whole list of paths in the model.
+    '''
+    ret: List[Path] = []
+    for i in itertools.count():
+        nth_path = nth_multifile_path(path, i)
+        if nth_path is None:
+            break
+        ret.append(nth_path)
+    if not ret:
+        # No matches.  This should only happen if the file was named, e.g.,
+        # foo.0, and there was no file named foo.  Oh well, try to process it
+        # as a single file.
+        return [path]
+    return ret
+
+
+def load_some_model(path: Path) -> ModelPlus:
+    '''Load a model of any supported format.'''
+    # Be extra-friendly and accept either a file or a directory:
+    if path.is_dir():
+        # Check if it's a set of safetensors files first
+        files = list(path.glob("model-00001-of-*.safetensors"))
+        if not files:
+            # Try the PyTorch patterns too, with lower priority
+            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
+            files = [file for glob in globs for file in path.glob(glob)]
+        if not files:
+            # Try GGML too, but with lower priority, since if both a non-GGML
+            # model and a GGML model exist in the same directory, we assume the
+            # latter was converted from the former.
+            files = list(path.glob("ggml-model*.bin*"))
+        if not files:
+            raise Exception(f"Can't find model in directory {path}")
+        if len(files) > 1:
+            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
+        path = files[0]
+
+    paths = find_multifile_paths(path)
+    models_plus: List[ModelPlus] = []
+    for path in paths:
+        print(f"Loading model file {path}")
+        models_plus.append(lazy_load_file(path))
+
+    model_plus = merge_multifile_models(models_plus)
+    return model_plus
+
+
+def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
+    return {name: model[name] for name in TENSORS_LIST if name in model}
+
+
+def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
+    print(f"vocabtype: {vocabtype}")
+    # Be extra-friendly and accept either a file or a directory.  Also, if it's
+    # a directory, it might be the model directory, and tokenizer.model might
+    # be in the parent of that.
+    if path.is_dir():
+        vocab_file = "tokenizer.model"
+        if vocabtype == 'bpe':
+          vocab_file = "vocab.json"
+        path2 = path / vocab_file
+        # Use `.parent` instead of /.. to handle the symlink case better.
+        path3 = path.parent / vocab_file
+        if path2.exists():
+            path = path2
+        elif path3.exists():
+            path = path3
+        else:
+            raise FileNotFoundError(
+                f"Could not find tokenizer.model in {path} or its parent; "
+                "if it's in another directory, pass the directory as --vocab-dir")
+    added_tokens_path = path.parent / "added_tokens.json"
+    print(f"Loading vocab file {path}")
+    return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None,
+                              vocabtype)
+
+
+def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
+    namestr = {
+        GGMLFileType.AllF32: "f32",
+        GGMLFileType.MostlyF16: "f16",
+        GGMLFileType.MostlyQ4_0: "q4_0",
+        GGMLFileType.MostlyQ4_1: "q4_1",
+        GGMLFileType.PerLayerIsQ4_1: "q4_1",
+    }[file_type]
+    ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
+    if ret in model_paths:
+        sys.stderr.write(
+            f"Error: Default output path ({ret}) would overwrite the input. "
+            "Please explicitly specify a path using --outfile.\n")
+        sys.exit(1)
+    return ret
+
+
+def do_dump_model(model_plus: ModelPlus) -> None:
+    print(f"model_plus.paths = {model_plus.paths!r}")
+    print(f"model_plus.format = {model_plus.format!r}")
+    print(f"model_plus.vocab = {model_plus.vocab!r}")
+    for name, lazy_tensor in model_plus.model.items():
+        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
+
+
+def main(args_in: Optional[List[str]] = None) -> None:
+    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
+    parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
+    parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
+    parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
+    parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)")
+    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
+    parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
+    parser.add_argument("model", type=Path,
+                        help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
+    parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
+    args = parser.parse_args(args_in)
+
+    vocab: Vocab
+    if args.dump_single:
+        model_plus = lazy_load_file(args.model)
+        do_dump_model(model_plus)
+    elif args.vocab_only:
+        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
+        assert args.outfile, "need --outfile if using --vocab-only"
+        outfile = args.outfile
+        OutputFile.write_vocab_only(outfile, vocab)
+        print(f"Wrote {outfile}")
+    else:
+        model_plus = load_some_model(args.model)
+        if args.dump:
+            do_dump_model(model_plus)
+            return
+        if model_plus.vocab is not None and args.vocab_dir is None:
+            vocab = model_plus.vocab
+        else:
+            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
+            vocab = load_vocab(vocab_dir, args.vocabtype)
+        params = Params.load(model_plus)
+        model = model_plus.model
+        model = do_necessary_conversions(model, params)
+        output_type = pick_output_type(model, args.outtype)
+        model = convert_to_output_type(model, output_type)
+        outfile = args.outfile or default_outfile(model_plus.paths, output_type)
+        OutputFile.write_all(outfile, params, output_type, model, vocab)
+        print(f"Wrote {outfile}")
+
+
+if __name__ == '__main__':
+    main()

From 7de7cb4bd86fb2d48391246ab9646a05fc05e556 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 04:06:59 +0200
Subject: [PATCH 225/242] convert-permute-debug.py : change permute type of
 attn_q

---
 convert-permute-debug.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/convert-permute-debug.py b/convert-permute-debug.py
index 14927e8e4976d..fcbb5f2da1173 100644
--- a/convert-permute-debug.py
+++ b/convert-permute-debug.py
@@ -819,12 +819,12 @@ def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
     for i in itertools.count():
         if f"model.layers.{i}.self_attn.q_proj.weight" in model:
             print(f"Permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
            #tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
         elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
             print(f"Unpacking and permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
             tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
         else:

From d5c8fcfd8ab2db5405c615cae2c74334e03afb65 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 04:33:33 +0200
Subject: [PATCH 226/242] convert.py : 70b model working (change attn_q
 permute)

---
 convert.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/convert.py b/convert.py
index f6237579d0954..df589928baf5f 100755
--- a/convert.py
+++ b/convert.py
@@ -326,6 +326,7 @@ def __repr__(self) -> str:
 #
 
 def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
+    #print( "permute debug " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_kv_head) )
     if n_head_kv is not None and n_head != n_head_kv:
         n_head //= n_head_kv
     return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
@@ -818,12 +819,12 @@ def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
     for i in itertools.count():
         if f"model.layers.{i}.self_attn.q_proj.weight" in model:
             print(f"Permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
            #tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
         elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
             print(f"Unpacking and permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head_kv)
+            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head)
             tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
             tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
         else:

From 287db5101539f1eb87b3ae8218ef1a5cceade991 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 04:34:39 +0200
Subject: [PATCH 227/242] Delete convert-permute-debug-master.py

---
 convert-permute-debug-master.py | 1327 -------------------------------
 1 file changed, 1327 deletions(-)
 delete mode 100644 convert-permute-debug-master.py

diff --git a/convert-permute-debug-master.py b/convert-permute-debug-master.py
deleted file mode 100644
index 7d64b22529c17..0000000000000
--- a/convert-permute-debug-master.py
+++ /dev/null
@@ -1,1327 +0,0 @@
-#!/usr/bin/env python
-import argparse
-import concurrent.futures
-import copy
-import enum
-import faulthandler
-import functools
-import io
-import itertools
-import json
-import math
-import mmap
-import pickle
-import re
-import signal
-import struct
-import sys
-import zipfile
-from abc import ABCMeta, abstractmethod
-from dataclasses import dataclass
-from pathlib import Path
-from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List,
-                    Literal, Optional, Sequence, Tuple, TypeVar, Union)
-
-import numpy as np
-from sentencepiece import SentencePieceProcessor  # type: ignore
-
-if TYPE_CHECKING:
-    from typing_extensions import TypeAlias
-
-if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
-    faulthandler.register(signal.SIGUSR1)
-
-NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
-
-
-@dataclass(frozen=True)
-class UnquantizedDataType:
-    name: str
-
-
-DT_F16 = UnquantizedDataType('F16')
-DT_F32 = UnquantizedDataType('F32')
-DT_I32 = UnquantizedDataType('I32')
-DT_BF16 = UnquantizedDataType('BF16')
-
-
-@dataclass(frozen=True)
-class QuantizedDataType:
-    groupsize: int
-    have_addends: bool
-    have_g_idx: bool
-
-
-DT_Q4_0 = QuantizedDataType(groupsize=32, have_addends=False, have_g_idx=False)
-DT_Q4_1 = QuantizedDataType(groupsize=32, have_addends=True, have_g_idx=False)
-
-DataType = Union[UnquantizedDataType, QuantizedDataType]
-
-DATA_TYPE_TO_FTYPE: Dict[DataType, int] = {
-    DT_F32: 0,
-    DT_F16: 1,
-    DT_Q4_0: 2,
-    DT_Q4_1: 3,
-}
-
-FTYPE_TO_DATA_TYPE: Dict[int, DataType] = \
-    {ftype: dtype for (dtype, ftype) in DATA_TYPE_TO_FTYPE.items()}
-
-DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
-    DT_BF16: np.dtype(np.uint16),
-    DT_F16: np.dtype(np.float16),
-    DT_F32: np.dtype(np.float32),
-    DT_I32: np.dtype(np.int32),
-}
-
-NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
-    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
-
-
-class GGMLFileType(enum.Enum):
-    AllF32 = 0
-    MostlyF16 = 1  # except 1d tensors
-    MostlyQ4_0 = 2  # except 1d tensors
-    MostlyQ4_1 = 3  # except 1d tensors
-    PerLayerIsQ4_1 = 4  # but tok_embeddings.weight and output.weight are F16
-
-    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
-        if len(tensor.shape) == 1:
-            # 1D tensors are always F32.
-            return DT_F32
-        elif self == GGMLFileType.AllF32:
-            return DT_F32
-        elif self == GGMLFileType.MostlyF16:
-            return DT_F16
-        elif self == GGMLFileType.MostlyQ4_0:
-            return DT_Q4_0
-        elif self == GGMLFileType.MostlyQ4_1:
-            return DT_Q4_1
-        elif self == GGMLFileType.PerLayerIsQ4_1:
-            if name in ('output.weight', 'tok_embeddings.weight'):
-                return DT_F16
-            else:
-                return DT_Q4_1
-        else:
-            raise ValueError(self)
-
-
-def make_tensors_list() -> List[str]:
-    ret = [
-        'tok_embeddings.weight',
-        'norm.weight',
-        'output.weight',
-    ]
-    for i in range(80):  # maximum number of layer
-        ret += [
-            f'layers.{i}.attention.wq.weight',
-            f'layers.{i}.attention.wk.weight',
-            f'layers.{i}.attention.wv.weight',
-            f'layers.{i}.attention.wo.weight',
-            f'layers.{i}.attention_norm.weight',
-            f'layers.{i}.feed_forward.w1.weight',
-            f'layers.{i}.feed_forward.w2.weight',
-            f'layers.{i}.feed_forward.w3.weight',
-            f'layers.{i}.ffn_norm.weight',
-        ]
-    return ret
-
-
-TENSORS_LIST = make_tensors_list()
-TENSORS_SET = set(TENSORS_LIST)
-
-
-def find_n_mult(n_ff: int, n_embd: int) -> int:
-    # hardcoded magic range
-    for n_mult in range(8192, 1, -1):
-        calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
-        if calc_ff == n_ff:
-            return n_mult
-    raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
-
-@dataclass
-class Params:
-    n_vocab:   int
-    n_embd:    int
-    n_mult:    int
-    n_head:    int
-    n_layer:   int
-    n_kv_head: Optional[int]  # This parameter is only used for Llama 2
-
-    @staticmethod
-    def guessed(model: 'LazyModel') -> 'Params':
-        # try transformer naming first
-        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
-
-        # try transformer naming first
-        if "model.layers.0.self_attn.q_proj.weight" in model:
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
-        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
-        else:
-            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
-
-        if n_layer < 1:
-            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
-                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
-
-        n_head=n_embd // 128 # guessed
-
-        return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = 256,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
-        )
-
-    @staticmethod
-    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab = config["vocab_size"];
-        n_embd  = config["hidden_size"];
-        n_head  = config["num_attention_heads"];
-        n_layer = config["num_hidden_layers"];
-        n_ff    = config["intermediate_size"];
-        n_kv_head = config.get("num_key_value_heads")
-
-        n_mult = find_n_mult(n_ff, n_embd);
-
-        return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = n_kv_head,
-        )
-
-    # LLaMA v2 70B params.json
-    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
-    @staticmethod
-    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab   = config["vocab_size"];
-        n_embd    = config["dim"];
-        n_head    = config["n_heads"];
-        n_layer   = config["n_layers"];
-        n_mult    = config["multiple_of"];
-
-        if n_vocab == -1:
-            n_vocab = model["tok_embeddings.weight"].shape[0]
-
-        return Params(
-            n_vocab   = n_vocab,
-            n_embd    = n_embd,
-            n_mult    = n_mult,
-            n_head    = n_head,
-            n_layer   = n_layer,
-            n_kv_head = None,
-        )
-
-    @staticmethod
-    def load(model_plus: 'ModelPlus') -> 'Params':
-        hf_config_path   = model_plus.paths[0].parent / "config.json"
-        orig_config_path = model_plus.paths[0].parent / "params.json"
-
-        if hf_config_path.exists():
-            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
-        elif orig_config_path.exists():
-            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
-        else:
-            params = Params.guessed(model_plus.model)
-
-        print(f'params: n_vocab:{params.n_vocab} n_embd:{params.n_embd} n_mult:{params.n_mult} n_head:{params.n_head} n_layer:{params.n_layer}')
-        return params
-
-
-class SentencePieceVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path], vocabtype: Optional[str]) -> None:
-        self.vocabtype = vocabtype
-        if self.vocabtype == "bpe":
-          self.sentencepiece_tokenizer = json.loads(open(str(fname_tokenizer)).read())
-        else:
-          self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
-        added_tokens: Dict[str, int]
-        if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens))
-        else:
-            added_tokens = {}
-        if self.vocabtype == "bpe":
-          vocab_size: int = len(self.sentencepiece_tokenizer)
-        else:
-          vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
-        self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
-
-    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        tokenizer = self.sentencepiece_tokenizer
-        if self.vocabtype == "bpe":
-          from transformers.models.gpt2 import tokenization_gpt2
-          byte_encoder = tokenization_gpt2.bytes_to_unicode()
-          byte_decoder = {v: k for k, v in byte_encoder.items()}
-          for i, item in enumerate(tokenizer):
-            text: bytes
-            text = b''.join([x.to_bytes(1, byteorder='big') for x in [byte_decoder[y] for y in item]])
-            score: float = -i
-            yield text, score
-        else:
-          for i in range(tokenizer.vocab_size()):
-              text: bytes
-              if tokenizer.is_unknown(i):
-                  text = " \u2047 ".encode("utf-8")
-              elif tokenizer.is_control(i):
-                  text = b""
-              elif tokenizer.is_byte(i):
-                  piece = tokenizer.id_to_piece(i)
-                  if len(piece) != 6:
-                      raise Exception(f"Invalid token: {piece}")
-                  byte_value = int(piece[3:-1], 16)
-                  text = struct.pack("B", byte_value)
-              else:
-                  text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
-              score: float = tokenizer.get_score(i)
-              yield text, score
-
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        yield from self.sentencepiece_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
-class GGMLVocab:
-    def __init__(self, tokens: List[Tuple[bytes, float]]):
-        self.tokens = tokens
-        self.vocab_size = len(tokens)
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        return self.tokens
-
-    def __repr__(self) -> str:
-        return f"<GGMLVocab with {self.vocab_size} tokens>"
-
-
-Vocab = Union[SentencePieceVocab, GGMLVocab]
-
-
-def permute(weights: NDArray, n_head: int, n_kv_head: Optional[int] = None) -> NDArray:
-    print( "permute " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_kv_head) )
-    if n_kv_head is not None and n_head != n_kv_head:
-        n_head //= n_kv_head
-    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                .swapaxes(1, 2)
-                .reshape(weights.shape))
-
-
-def dequantize_q4(qvalues_pack32: NDArray, scales: NDArray, addends: Optional[NDArray], g_idx: Optional[NDArray]) -> NDArray:
-    # First reinterpret each row from a list of int32s containing 8 values each
-    # to a list of uint8s containing 2 values each.
-    qvalues_pack8 = qvalues_pack32.view(np.uint8)
-
-    # Then split out the two values per int8 (which requires an actual
-    # conversion because numpy doesn't natively support int4s).
-    qvalues = np.zeros([qvalues_pack8.shape[0], qvalues_pack8.shape[1] * 2], dtype=np.uint8)
-    qvalues[:, 0::2] = qvalues_pack8 & 0xf
-    qvalues[:, 1::2] = qvalues_pack8 >> 4
-
-    assert addends is None or addends.shape == scales.shape
-    assert qvalues.shape[0] == scales.shape[0]
-    assert qvalues.shape[1] % scales.shape[1] == 0
-    if g_idx is None:
-        repeat_count = qvalues.shape[1] // scales.shape[1]
-        scales = scales[:, :, np.newaxis]
-        if addends is not None:
-            addends = addends[:, :, np.newaxis]
-        # Reshape so that the below computation broadcasts over scales and addends:
-        qvalues.shape = (qvalues.shape[0], scales.shape[1], int(repeat_count))
-    else:
-        # In this case the scale and addend is selected for each column by g_idx:
-        assert addends is not None
-        scales = scales[:, g_idx]
-        addends = addends[:, g_idx]
-    if addends is None:
-        # Q4_0
-        qvalues = qvalues.view(np.int8)
-        qvalues -= 8
-    # And do the actual 'value = scale * qvalue + addend' computation.
-    values = scales * qvalues
-    if addends is not None:
-        values += addends
-    if g_idx is None:
-        values.shape = (values.shape[0], values.shape[1] * values.shape[2])
-    return values
-
-
-class Tensor(metaclass=ABCMeta):
-    data_type: DataType
-
-    @abstractmethod
-    def astype(self, data_type: DataType) -> 'Tensor': ...
-    @abstractmethod
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'Tensor': ...
-    @abstractmethod
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def part(self, n_part: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
-
-
-def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
-    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
-    fp32_arr = bf16_arr.astype(np.uint32) << 16
-    return fp32_arr.view(np.float32)
-
-
-class UnquantizedTensor(Tensor):
-    def __init__(self, ndarray: NDArray) -> None:
-        assert isinstance(ndarray, np.ndarray)
-        self.ndarray = ndarray
-        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
-
-    def astype(self, data_type: DataType) -> Tensor:
-        dtype = DATA_TYPE_TO_NUMPY[data_type]
-        if self.data_type == DT_BF16:
-            self.ndarray = bf16_to_fp32(self.ndarray)
-        return UnquantizedTensor(self.ndarray.astype(dtype))
-
-    def to_ggml(self) -> 'UnquantizedTensor':
-        return self
-
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
-
-    def part(self, n_part: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head, n_kv_head))
-
-
-def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
-    tensor = lazy_tensor.load()
-    assert isinstance(tensor, UnquantizedTensor)
-
-    # double-check:
-    actual_shape = list(tensor.ndarray.shape)
-    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
-    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
-        if convert:
-            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
-        else:
-            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
-
-    return tensor.ndarray
-
-
-class GGMLQuantizedTensor(Tensor):
-    data_type: QuantizedDataType
-
-    def __init__(self, ndarray: NDArray, shape: List[int], data_type: DataType) -> None:
-        rows, columns = shape
-        assert data_type in (DT_Q4_1, DT_Q4_0)  # for now
-        assert isinstance(data_type, QuantizedDataType)  # redundant, but mypy complains without this
-        assert columns % data_type.groupsize == 0
-        words_in_block = 6 if data_type == DT_Q4_1 else 5
-        self.ndarray = ndarray.view(dtype=np.uint32).reshape((rows, columns // data_type.groupsize, words_in_block))
-        self.shape = shape[:]
-        self.data_type = data_type
-
-    def astype(self, data_type: DataType) -> Tensor:
-        if data_type == self.data_type:
-            return self
-        scales = self.ndarray[:, :, 0].view(np.float32)
-        if self.data_type.have_addends:
-            addends = self.ndarray[:, :, 1].view(np.float32)
-        else:
-            addends = None
-        qweights = self.ndarray[:, :, -4:].reshape([self.shape[0], self.shape[1] // 8])
-
-        dq = dequantize_q4(qweights, scales, addends, g_idx=None)
-        return UnquantizedTensor(dq).astype(data_type)
-
-    def to_ggml(self) -> 'GGMLQuantizedTensor':
-        return self
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'GGMLQuantizedTensor':
-        return GGMLQuantizedTensor(permute(self.ndarray, n_head, n_kv_head), self.shape, self.data_type)
-
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
-
-    def part(self, n_part: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
-
-GGMLCompatibleTensor = Union[UnquantizedTensor, GGMLQuantizedTensor]
-
-
-class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int, n_kv_head: Optional[int] = None) -> None:
-        self.base = base
-        self.n_head = n_head
-        self.n_kv_head = n_kv_head
-        self.data_type = self.base.data_type
-
-    def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head, self.n_kv_head)
-
-    def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head, self.n_kv_head)
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
-        raise Exception("shouldn't permute twice")
-
-
-class GPTQForLLaMaQuantizedTensor(Tensor):
-    def __init__(self, model: 'LazyModel', namebase: str) -> None:
-        qweight = load_unquantized(model[f"{namebase}.qweight"], np.int32)
-        scales = load_unquantized(model[f"{namebase}.scales"], np.float32, convert=True)
-
-        bias = model.get(f"{namebase}.bias")
-        if bias is not None:
-            # Q4_1 does not support bias; good thing the bias is always all zeros.
-            assert not np.any(load_unquantized(bias))
-
-        if f"{namebase}.zeros" in model:
-            zeros = load_unquantized(model[f"{namebase}.zeros"], np.float32)
-        else:
-            qzeros = load_unquantized(model[f"{namebase}.qzeros"], np.int32)
-            assert qzeros.dtype == np.int32
-            zeros = dequantize_q4(qzeros, scales, scales, g_idx=None)
-            assert zeros.dtype == np.float32
-
-        assert zeros.shape == scales.shape
-
-        # Output is transposed compared to the input, and addends have their sign flipped.
-        # Scales and zeros similarly must be transposed but only for newer
-        # versions of GPTQ-for-LLaMa; the older versions can be identified by
-        # having shape (n_embd, 1).
-        qweight = qweight.T
-        if scales.shape[1] != 1:
-            scales = scales.T
-            zeros = zeros.T
-
-        # Output also has signs flipped for the addends.
-        self.qweight = qweight
-        self.scales = scales
-        self.addends = -zeros
-
-        self.g_idx: Optional[NDArray]
-        if f"{namebase}.g_idx" in model:
-            self.g_idx = load_unquantized(model[f"{namebase}.g_idx"], np.int32)
-            assert self.g_idx.shape == (qweight.shape[1] * 8,)
-        else:
-            self.g_idx = None
-
-        self.shape = [self.qweight.shape[0], self.qweight.shape[1] * 8]
-        self.data_type = QuantizedDataType(groupsize=self.groupsize(), have_addends=True,
-                                           have_g_idx=(self.g_idx is not None))
-
-    def inspect(self, row: int, col: int) -> None:
-        '''For debugging.'''
-        qweight = (self.qweight[row, col // 8] >> (4 * (col & 7))) & 0xf
-        if self.g_idx is not None:
-            group = self.g_idx[col]
-        else:
-            group = int(col // self.groupsize())
-        scale = self.scales[row, group]
-        addend = self.addends[row, group]
-        with np.printoptions(precision=None, suppress=True):
-            print(f'scale:{scale} addend:{addend} qweight:{qweight}')
-            print('possible values:', np.arange(16) * scale + addend)
-            print('actual value:', qweight * scale + addend)
-
-    def astype(self, data_type: DataType) -> Tensor:
-        if isinstance(data_type, QuantizedDataType):
-            assert self.g_idx is None and data_type.have_addends is True and data_type.have_g_idx is False
-            return self.regroup(data_type.groupsize)
-
-        dequantized = dequantize_q4(np.ascontiguousarray(self.qweight), self.scales, self.addends, self.g_idx)
-        return UnquantizedTensor(dequantized).astype(data_type)
-
-    def groupsize(self) -> int:
-        assert self.addends.shape == self.scales.shape
-        assert self.shape[1] % self.scales.shape[1] == 0
-        return self.shape[1] // self.scales.shape[1]
-
-    def regroup(self, new_groupsize: int = 32) -> 'GPTQForLLaMaQuantizedTensor':
-        # Old versions of GPTQ-for-LLaMa shared scales and addends between all the
-        # columns in a row.  Newer versions share them between every set of N
-        # columns in a row, where N is the `groupsize` parameter, usually 128.  The
-        # output format shares them between every set of 32 columns.  To handle
-        # this, duplicate scales and addends for every smaller group.
-        # (In the above, 'row' and 'column' are in the sense of the output.)
-        assert self.g_idx is None
-        old_groupsize = self.groupsize()
-        assert old_groupsize >= new_groupsize and old_groupsize % new_groupsize == 0, old_groupsize
-        ret = copy.copy(self)
-        ret.addends = self.addends.repeat(old_groupsize // new_groupsize, axis=1)
-        ret.scales = self.scales.repeat(old_groupsize // new_groupsize, axis=1)
-        ret.data_type = QuantizedDataType(groupsize=new_groupsize, have_addends=True, have_g_idx=False)
-        return ret
-
-    def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> Tensor:
-        return DeferredPermutedTensor(self, n_head, n_kv_head)
-
-    def to_ggml(self) -> GGMLQuantizedTensor:
-        # The output format looks like this:
-        # For each row:
-        #   For each group of 32 columns:
-        #     - addend (float32, 4 bytes)
-        #     - scale (float32, 4 bytes)
-        #     - weights (int4 * 32, 16 bytes)
-
-        if self.groupsize() != 32:
-            raise Exception("should have been regrouped before converting to ggml")
-
-        # Since the output format is mixed between integers and floats, we have
-        # to hackily view the floats as int32s just so numpy will let us
-        # concatenate them.
-        addends_view = self.addends.view(dtype=np.int32)[:, :, np.newaxis]
-        scales_view = self.scales.view(dtype=np.int32)[:, :, np.newaxis]
-
-        # Split into groups of 4 columns (i.e. 32 columns of quantized data):
-        grouped = self.qweight.reshape([self.qweight.shape[0], self.qweight.shape[1] // 4, 4])
-
-        # And concatenate:
-        grouped = np.concatenate([scales_view, addends_view, grouped], axis=2, casting='no')
-
-        return GGMLQuantizedTensor(grouped, self.shape, DT_Q4_1)
-
-
-@dataclass
-class LazyTensor:
-    _load: Callable[[], Tensor]
-    shape: List[int]
-    data_type: DataType
-    description: str
-
-    def load(self) -> Tensor:
-        ret = self._load()
-        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
-        return ret
-
-    def astype(self, data_type: DataType) -> 'LazyTensor':
-        self.validate_conversion_to(data_type)
-
-        def load() -> Tensor:
-            return self.load().astype(data_type)
-        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
-
-    def validate_conversion_to(self, data_type: DataType) -> None:
-        if data_type == self.data_type:
-            return
-        if isinstance(data_type, QuantizedDataType):
-            if not isinstance(self.data_type, QuantizedDataType):
-                raise Exception(f"Can't turn an unquantized tensor into a quantized type ({data_type})")
-            if self.data_type.have_g_idx:
-                sys.stderr.write(
-                    "Error: Input uses the newer GPTQ-for-LLaMa format (using g_idx), "
-                    "which is not yet natively supported by GGML. "
-                    "For now you can still convert this model by passing `--outtype f16` to dequantize, "
-                    "but that will result in a much larger output file for no quality benefit.\n")
-                sys.exit(1)
-            assert not data_type.have_g_idx and self.data_type.have_addends and data_type.have_addends
-
-
-LazyModel = Dict[str, LazyTensor]
-
-
-@dataclass
-class ModelPlus:
-    model: LazyModel
-    paths: List[Path]  # Where this was read from.
-    format: Literal['ggml', 'torch', 'safetensors']
-    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
-
-
-def merge_sharded(models: List[LazyModel]) -> LazyModel:
-    # Original LLaMA models have each file contain one part of each tensor.
-    # Use a dict instead of a set to preserve order.
-    names = {name: None for model in models for name in model}
-
-    def convert(name: str) -> LazyTensor:
-        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
-        if len(lazy_tensors) == 1:
-            # only one file; don't go through this procedure since there might
-            # be quantized tensors
-            return lazy_tensors[0]
-        if len(lazy_tensors[0].shape) == 1:
-            # the tensor is just duplicated in every file
-            return lazy_tensors[0]
-        if name.startswith('tok_embeddings.') or \
-           name.endswith('.attention.wo.weight') or \
-           name.endswith('.feed_forward.w2.weight'):
-            # split by columns
-            axis = 1
-        else:
-            # split by rows
-            axis = 0
-        concatenated_shape = list(lazy_tensors[0].shape)
-        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
-
-        def load() -> UnquantizedTensor:
-            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
-            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
-            return UnquantizedTensor(concatenated)
-        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
-        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
-    return {name: convert(name) for name in names}
-
-
-def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
-    formats = set(mp.format for mp in models_plus)
-    assert len(formats) == 1, "different formats?"
-    format = formats.pop()
-    paths = [path for mp in models_plus for path in mp.paths]
-    # Use the first non-None vocab, if any.
-    try:
-        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
-    except StopIteration:
-        vocab = None
-
-    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
-        # Transformers models put different tensors in different files, but
-        # don't split indivdual tensors between files.
-        model: LazyModel = {}
-        for mp in models_plus:
-            model.update(mp.model)
-    else:
-        model = merge_sharded([mp.model for mp in models_plus])
-
-    return ModelPlus(model, paths, format, vocab)
-
-
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_kv_head: Optional[int] = None) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head, n_kv_head)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_kv_head}) ' + lazy_tensor.description)
-
-def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute_part(n_part, n_head)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
-
-def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().part(n_part)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
-
-def convert_transformers_to_orig(model: LazyModel, params: Params) -> LazyModel:
-    out: LazyModel = {}
-    out["tok_embeddings.weight"] = model["model.embed_tokens.weight"]
-    out["norm.weight"] = model["model.norm.weight"]
-    out["output.weight"] = model["lm_head.weight"]
-
-    for i in itertools.count():
-        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_kv_head)
-            out[f"layers.{i}.attention.wv.weight"] = model[f"model.layers.{i}.self_attn.v_proj.weight"]
-        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
-            out[f"layers.{i}.attention.wq.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head)
-            out[f"layers.{i}.attention.wk.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head)
-            out[f"layers.{i}.attention.wv.weight"] = part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
-        else:
-            break
-
-        out[f"layers.{i}.attention.wo.weight"] = model[f"model.layers.{i}.self_attn.o_proj.weight"]
-
-        out[f"layers.{i}.feed_forward.w1.weight"] = model[f"model.layers.{i}.mlp.gate_proj.weight"]
-        out[f"layers.{i}.feed_forward.w2.weight"] = model[f"model.layers.{i}.mlp.down_proj.weight"]
-        out[f"layers.{i}.feed_forward.w3.weight"] = model[f"model.layers.{i}.mlp.up_proj.weight"]
-
-        out[f"layers.{i}.attention_norm.weight"] = model[f"model.layers.{i}.input_layernorm.weight"]
-        out[f"layers.{i}.ffn_norm.weight"] = model[f"model.layers.{i}.post_attention_layernorm.weight"]
-    return out
-
-
-def handle_quantization(model: LazyModel) -> LazyModel:
-    '''Convert a model with entries for 'foo.qweight', 'foo.scales', etc.
-    (which resolve to UnquantizedTensors with the raw data) to one with entries
-    for 'foo.weight' (which resolve to QuantizedTensors).
-    '''
-    def convert(name: str) -> Tuple[str, LazyTensor]:
-        if name.endswith(".qweight"):
-            namebase = name.rsplit('.', 1)[0]
-            orig_name = namebase + ".weight"
-
-            lazy_tensor = model[name]
-            assert len(lazy_tensor.shape) == 2
-            real_shape = [lazy_tensor.shape[1], lazy_tensor.shape[0] * 8]
-
-            # Calculate type.  This replicates the logic in
-            # GPTQForLLaMaQuantizedTensor (which is executed when the modelis
-            # actually loaded).
-            lazy_scales = model[f"{namebase}.scales"]
-            scales_width = 1 if lazy_scales.shape[1] == 1 else lazy_scales.shape[0]
-            assert real_shape[1] % scales_width == 0
-            groupsize = real_shape[1] // scales_width
-            have_g_idx = f"{namebase}.g_idx" in model
-            data_type = QuantizedDataType(groupsize=groupsize, have_addends=True, have_g_idx=have_g_idx)
-
-            def load() -> Tensor:
-                return GPTQForLLaMaQuantizedTensor(model, namebase)
-
-            return (orig_name, LazyTensor(load, real_shape, data_type, '[quantized]'))
-        else:
-            return (name, model[name])
-    return dict(convert(name) for name in model)
-
-# Functionality that simulates `torch.load` but where individual tensors are
-# only loaded into memory on demand, not all at once.
-# PyTorch can't do this natively as of time of writing:
-# - https://github.com/pytorch/pytorch/issues/64327
-# This allows us to de-shard without multiplying RAM usage, and also
-# conveniently drops the PyTorch dependency (though we still need numpy).
-
-
-@dataclass
-class LazyStorageKind:
-    data_type: DataType
-
-
-@dataclass
-class LazyStorage:
-    load: Callable[[int, int], NDArray]
-    kind: LazyStorageKind
-    description: str
-
-
-class LazyUnpickler(pickle.Unpickler):
-    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
-        super().__init__(fp)
-        self.data_base_path = data_base_path
-        self.zip_file = zip_file
-
-    def persistent_load(self, pid: Any) -> Any:
-        assert pid[0] == 'storage'
-        assert isinstance(pid[1], LazyStorageKind)
-        data_type = pid[1].data_type
-        filename_stem = pid[2]
-        filename = self.data_base_path + '/' + filename_stem
-        info = self.zip_file.getinfo(filename)
-
-        def load(offset: int, elm_count: int) -> NDArray:
-            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
-            if dtype is None:
-                raise Exception("tensor stored in unsupported format")
-            fp = self.zip_file.open(info)
-            fp.seek(offset * dtype.itemsize)
-            size = elm_count * dtype.itemsize
-            data = fp.read(size)
-            assert len(data) == size
-            return np.frombuffer(data, dtype)
-        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
-        return LazyStorage(load=load, kind=pid[1], description=description)
-
-    # @staticmethod
-    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
-                               # pyright: ignore[reportSelfClsParameterName]
-                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
-        assert isinstance(storage, LazyStorage)
-
-        def load() -> UnquantizedTensor:
-            elm_count = stride[0] * size[0]
-            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
-        description = f'pickled storage_offset={storage_offset} in {storage.description}'
-        return LazyTensor(load, list(size), storage.kind.data_type, description)
-
-    # @staticmethod
-    def rebuild_from_type_v2(func, new_type, args, state):
-        return func(*args)
-
-    CLASSES: Dict[Any, Any] = {
-        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
-        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
-        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
-        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
-        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
-        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
-        ('torch', 'Tensor'): LazyTensor,
-    }
-
-    def find_class(self, module: str, name: str) -> Any:
-        if not module.startswith('torch'):
-            return super().find_class(module, name)
-        return self.CLASSES[(module, name)]
-
-
-def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
-    zf = zipfile.ZipFile(outer_fp)
-    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
-    assert len(pickle_paths) == 1, pickle_paths
-    pickle_fp = zf.open(pickle_paths[0], 'r')
-    unpickler = LazyUnpickler(pickle_fp,
-                              data_base_path=pickle_paths[0][:-4],
-                              zip_file=zf)
-    model = unpickler.load()
-    as_dict = dict(model.items())
-    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
-
-
-SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
-    'BF16': DT_BF16,
-    'F16': DT_F16,
-    'F32': DT_F32,
-    'I32': DT_I32,
-}
-
-
-def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
-    header_size, = struct.unpack('<Q', fp.read(8))
-    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
-    # Use mmap for the actual data to avoid race conditions with the file offset.
-    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
-    byte_buf = mapped[8 + header_size:]
-
-    def convert(info: Dict[str, Any]) -> LazyTensor:
-        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
-        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
-        shape: List[int] = info['shape']
-        begin, end = info['data_offsets']
-        assert 0 <= begin <= end <= len(byte_buf)
-        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
-        buf = byte_buf[begin:end]
-
-        def load() -> UnquantizedTensor:
-            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
-        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
-        return LazyTensor(load, shape, data_type, description)
-    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
-    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
-
-
-def must_read(fp: IO[bytes], length: int) -> bytes:
-    ret = fp.read(length)
-    if len(ret) < length:
-        raise Exception("unexpectedly reached end of file")
-    return ret
-
-
-def lazy_load_ggml_file(fp: io.BufferedReader, path: Path) -> ModelPlus:
-    magic = must_read(fp, 4)[::-1]
-    if magic in (b'ggmf', b'ggjt'):
-        version, = struct.unpack("i", must_read(fp, 4))
-        assert version == 1
-    else:
-        assert magic == b'ggml'
-        version = None
-    n_vocab, n_embd, n_mult, n_head, n_layer, rot, file_type = struct.unpack('<7i', must_read(fp, 28))
-
-    tokens: List[Tuple[bytes, float]] = []
-    for i in range(n_vocab):
-        if i == 32000:
-            # HACK: GPT4All messed with the format without changing the magic
-            # number.  Specifically, they changed the vocab section to contain
-            # `n_vocab - 1` tokens instead of `n_vocab` (i.e. omitting the
-            # extra pad token).  Try to detect if we're reading a file like
-            # this.
-            orig_pos = fp.tell()
-            fp.seek(20, io.SEEK_CUR)
-            is_gpt4all = fp.read(21) == b'tok_embeddings.weight'
-            fp.seek(orig_pos)
-            if is_gpt4all:
-                break
-
-        length, = struct.unpack("i", must_read(fp, 4))
-        text = must_read(fp, length)
-        if magic != b'ggml':
-            score, = struct.unpack("f", must_read(fp, 4))
-            tokens.append((text, score))
-    vocab = GGMLVocab(tokens) if magic != b'ggml' else None
-
-    model: LazyModel = {}
-    # Use mmap for the actual data to avoid race conditions with the file offset.
-    off = fp.raw.tell()
-    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
-    fp.raw.seek(off)  # needed on Windows
-
-    def read_tensor() -> None:  # this is a function so that variables captured in `load` don't change
-        shape_len, name_len, ftype = struct.unpack("iii", must_read(fp, 12))
-        assert 0 <= shape_len <= 3
-        shape: List[int] = list(struct.unpack(f"{shape_len}i", must_read(fp, 4 * shape_len)))
-        shape = shape[::-1]
-        name = must_read(fp, name_len).decode('utf-8')
-        data_type = FTYPE_TO_DATA_TYPE[ftype]
-
-        if magic == b'ggjt':
-            fp.seek((fp.tell() + 31) & -32)
-
-        if data_type == DT_Q4_1:
-            # See GPTQForLLaMaQuantizedTensor.ggml_ndarray()
-            size = 24 * (shape[1] // 32) * shape[0]
-        elif data_type == DT_Q4_0:
-            size = 20 * (shape[1] // 32) * shape[0]
-        else:
-            numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
-            elm_count = math.prod(shape)
-            size = elm_count * numpy_dtype.itemsize
-        offset = fp.tell()
-        buf = mapped[offset:offset+size]
-        fp.seek(size, io.SEEK_CUR)
-
-        def load() -> Tensor:
-            if isinstance(data_type, QuantizedDataType):
-                ndarray = np.frombuffer(buf, dtype=np.uint32)
-                return GGMLQuantizedTensor(ndarray, shape, data_type)
-            else:
-                return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
-        description = f'ggml offset={offset} type={data_type} path={path}'
-        model[name] = LazyTensor(load, shape, data_type, description)
-
-    while fp.read(1) != b'':
-        fp.seek(-1, io.SEEK_CUR)
-        read_tensor()
-
-    return ModelPlus(model=model, paths=[path], format='ggml', vocab=vocab)
-
-
-@functools.lru_cache(maxsize=None)
-def lazy_load_file(path: Path) -> ModelPlus:
-    fp = open(path, 'rb')
-    first8 = fp.read(8)
-    fp.seek(0)
-    if first8[:2] == b'PK':
-        # A zip file, i.e. PyTorch format
-        return lazy_load_torch_file(fp, path)
-    elif first8[2:4] == b'gg':
-        # GGML format
-        return lazy_load_ggml_file(fp, path)
-    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
-        # Probably safetensors
-        return lazy_load_safetensors_file(fp, path)
-    else:
-        raise ValueError(f"unknown format: {path}")
-
-
-In = TypeVar('In')
-Out = TypeVar('Out')
-
-
-def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
-    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
-    fast enough, this will stop calling `func` at some point rather than
-    letting results pile up in memory.  Specifically, there is a max of one
-    output value buffered per thread.'''
-    with concurrent.futures.ThreadPoolExecutor() as executor:
-        futures: List[concurrent.futures.Future[Out]] = []
-        items_rev = list(iterable)[::-1]
-        for i in range(min(concurrency, len(items_rev))):
-            futures.append(executor.submit(func, items_rev.pop()))
-        while futures:
-            result = futures.pop(0).result()
-            if items_rev:
-                futures.append(executor.submit(func, items_rev.pop()))
-            yield result
-
-
-def check_vocab_size(params: Params, vocab: Vocab) -> None:
-    if params.n_vocab != vocab.vocab_size:
-        # GGMLVocab comes from the same file as the model so shouldn't mismatch:
-        assert isinstance(vocab, SentencePieceVocab)
-        if params.n_vocab == vocab.vocab_size_base:
-            print("Ignoring added_tokens.json since model matches vocab size without it.")
-            vocab.added_tokens_list = []
-            vocab.vocab_size = vocab.vocab_size_base
-            return
-        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
-        if vocab.fname_added_tokens is not None:
-            msg += f" combined with {vocab.fname_added_tokens}"
-        msg += f" has {vocab.vocab_size})."
-        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
-            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
-        raise Exception(msg)
-
-
-class OutputFile:
-    def __init__(self, fname_out: Path) -> None:
-        self.fout = open(fname_out, "wb")
-
-    def write_file_header(self, params: Params, file_type: GGMLFileType) -> None:
-        self.fout.write(b"ggjt"[::-1])  # magic
-        values = [
-            1,  # file version
-            params.n_vocab,
-            params.n_embd,
-            params.n_mult,
-            params.n_head,
-            params.n_layer,
-            params.n_embd // params.n_head,  # rot (obsolete)
-            file_type.value,
-        ]
-        self.fout.write(struct.pack("i" * len(values), *values))
-
-    def write_tensor_header(self, name: str, shape: Sequence[int], data_type: DataType) -> None:
-        sname = name.encode('utf-8')
-        self.fout.write(struct.pack("iii", len(shape), len(sname), DATA_TYPE_TO_FTYPE[data_type]))
-        self.fout.write(struct.pack("i" * len(shape), *shape[::-1]))
-        self.fout.write(sname)
-        self.fout.seek((self.fout.tell() + 31) & -32)
-
-    def write_vocab(self, vocab: Vocab) -> None:
-        for text, score in vocab.all_tokens():
-            self.fout.write(struct.pack("i", len(text)))
-            self.fout.write(text)
-            self.fout.write(struct.pack("f", score))
-
-    @staticmethod
-    def write_vocab_only(fname_out: Path, vocab: Vocab) -> None:
-        of = OutputFile(fname_out)
-        params = Params(n_vocab=vocab.vocab_size, n_embd=0, n_mult=0, n_head=1, n_layer=0)
-        of = OutputFile(fname_out)
-        of.write_file_header(params, file_type=GGMLFileType.AllF32)
-        of.write_vocab(vocab)
-        of.fout.close()
-
-    @staticmethod
-    def write_all(fname_out: Path, params: Params, file_type: GGMLFileType, model: LazyModel, vocab: Vocab) -> None:
-        check_vocab_size(params, vocab)
-        of = OutputFile(fname_out)
-        of.write_file_header(params, file_type)
-        print("Writing vocab...")
-        of.write_vocab(vocab)
-
-        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
-            name, lazy_tensor = item
-            return lazy_tensor.load().to_ggml().ndarray
-
-        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=1)
-        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
-            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
-            padi = len(str(len(model)))
-            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
-            of.write_tensor_header(name, lazy_tensor.shape, lazy_tensor.data_type)
-            ndarray.tofile(of.fout)
-        of.fout.close()
-
-
-def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
-    wq_type = model["layers.0.attention.wq.weight"].data_type
-    if output_type_str == "f32" or (output_type_str is None and wq_type in (DT_F32, DT_BF16)):
-        return GGMLFileType.AllF32
-    if output_type_str == "f16" or (output_type_str is None and wq_type == DT_F16):
-        return GGMLFileType.MostlyF16
-    if output_type_str == "q4_1" or (output_type_str is None and isinstance(wq_type, QuantizedDataType) and
-                                     wq_type.have_addends):
-        if isinstance(model["output.weight"].data_type, QuantizedDataType):
-            return GGMLFileType.MostlyQ4_1
-        else:
-            return GGMLFileType.PerLayerIsQ4_1
-    if output_type_str == "q4_0" or (output_type_str is None and isinstance(wq_type, QuantizedDataType)):
-        return GGMLFileType.MostlyQ4_0
-    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
-    raise Exception(f"Unexpected combination of types: {name_to_type}")
-
-
-def do_necessary_conversions(model: LazyModel, params: Params) -> LazyModel:
-    model = handle_quantization(model)
-
-    if "lm_head.weight" in model:
-        model = convert_transformers_to_orig(model, params)
-    model = filter_and_sort_tensors(model)
-
-    return model
-
-
-def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
-    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
-            for (name, tensor) in model.items()}
-
-
-def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the nth path in the model.
-    '''
-    # Support the following patterns:
-    patterns: List[Tuple[str, str]] = [
-        # - x.00.pth, x.01.pth, etc.
-        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
-        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
-        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
-        # x.bin, x.bin.1, etc.
-        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
-    ]
-    for regex, replacement in patterns:
-        if re.search(regex, path.name):
-            new_path = path.with_name(re.sub(regex, replacement, path.name))
-            if new_path.exists():
-                return new_path
-    return None
-
-
-def find_multifile_paths(path: Path) -> List[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the whole list of paths in the model.
-    '''
-    ret: List[Path] = []
-    for i in itertools.count():
-        nth_path = nth_multifile_path(path, i)
-        if nth_path is None:
-            break
-        ret.append(nth_path)
-    if not ret:
-        # No matches.  This should only happen if the file was named, e.g.,
-        # foo.0, and there was no file named foo.  Oh well, try to process it
-        # as a single file.
-        return [path]
-    return ret
-
-
-def load_some_model(path: Path) -> ModelPlus:
-    '''Load a model of any supported format.'''
-    # Be extra-friendly and accept either a file or a directory:
-    if path.is_dir():
-        # Check if it's a set of safetensors files first
-        files = list(path.glob("model-00001-of-*.safetensors"))
-        if not files:
-            # Try the PyTorch patterns too, with lower priority
-            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
-            files = [file for glob in globs for file in path.glob(glob)]
-        if not files:
-            # Try GGML too, but with lower priority, since if both a non-GGML
-            # model and a GGML model exist in the same directory, we assume the
-            # latter was converted from the former.
-            files = list(path.glob("ggml-model*.bin*"))
-        if not files:
-            raise Exception(f"Can't find model in directory {path}")
-        if len(files) > 1:
-            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
-        path = files[0]
-
-    paths = find_multifile_paths(path)
-    models_plus: List[ModelPlus] = []
-    for path in paths:
-        print(f"Loading model file {path}")
-        models_plus.append(lazy_load_file(path))
-
-    model_plus = merge_multifile_models(models_plus)
-    return model_plus
-
-
-def filter_and_sort_tensors(model: LazyModel) -> LazyModel:
-    return {name: model[name] for name in TENSORS_LIST if name in model}
-
-
-def load_vocab(path: Path, vocabtype: Optional[str]) -> SentencePieceVocab:
-    print(f"vocabtype: {vocabtype}")
-    # Be extra-friendly and accept either a file or a directory.  Also, if it's
-    # a directory, it might be the model directory, and tokenizer.model might
-    # be in the parent of that.
-    if path.is_dir():
-        vocab_file = "tokenizer.model"
-        if vocabtype == 'bpe':
-          vocab_file = "vocab.json"
-        path2 = path / vocab_file
-        # Use `.parent` instead of /.. to handle the symlink case better.
-        path3 = path.parent / vocab_file
-        if path2.exists():
-            path = path2
-        elif path3.exists():
-            path = path3
-        else:
-            raise FileNotFoundError(
-                f"Could not find tokenizer.model in {path} or its parent; "
-                "if it's in another directory, pass the directory as --vocab-dir")
-    added_tokens_path = path.parent / "added_tokens.json"
-    print(f"Loading vocab file {path}")
-    return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None,
-                              vocabtype)
-
-
-def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
-    namestr = {
-        GGMLFileType.AllF32: "f32",
-        GGMLFileType.MostlyF16: "f16",
-        GGMLFileType.MostlyQ4_0: "q4_0",
-        GGMLFileType.MostlyQ4_1: "q4_1",
-        GGMLFileType.PerLayerIsQ4_1: "q4_1",
-    }[file_type]
-    ret = model_paths[0].parent / f"ggml-model-{namestr}.bin"
-    if ret in model_paths:
-        sys.stderr.write(
-            f"Error: Default output path ({ret}) would overwrite the input. "
-            "Please explicitly specify a path using --outfile.\n")
-        sys.exit(1)
-    return ret
-
-
-def do_dump_model(model_plus: ModelPlus) -> None:
-    print(f"model_plus.paths = {model_plus.paths!r}")
-    print(f"model_plus.format = {model_plus.format!r}")
-    print(f"model_plus.vocab = {model_plus.vocab!r}")
-    for name, lazy_tensor in model_plus.model.items():
-        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
-
-
-def main(args_in: Optional[List[str]] = None) -> None:
-    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
-    parser.add_argument("--dump", action="store_true", help="don't convert, just show what's in the model")
-    parser.add_argument("--dump-single", action="store_true", help="don't convert, just show what's in a single model file")
-    parser.add_argument("--vocab-only", action="store_true", help="extract only the vocab")
-    parser.add_argument("--outtype", choices=["f32", "f16", "q4_1", "q4_0"], help="output format (default: based on input)")
-    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file")
-    parser.add_argument("--outfile", type=Path, help="path to write to; default: based on input")
-    parser.add_argument("model", type=Path,
-                        help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype", default='spm', choices=["spm", "bpe"], help="vocab format (default: spm)")
-    args = parser.parse_args(args_in)
-
-    vocab: Vocab
-    if args.dump_single:
-        model_plus = lazy_load_file(args.model)
-        do_dump_model(model_plus)
-    elif args.vocab_only:
-        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
-        assert args.outfile, "need --outfile if using --vocab-only"
-        outfile = args.outfile
-        OutputFile.write_vocab_only(outfile, vocab)
-        print(f"Wrote {outfile}")
-    else:
-        model_plus = load_some_model(args.model)
-        if args.dump:
-            do_dump_model(model_plus)
-            return
-        if model_plus.vocab is not None and args.vocab_dir is None:
-            vocab = model_plus.vocab
-        else:
-            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
-            vocab = load_vocab(vocab_dir, args.vocabtype)
-        params = Params.load(model_plus)
-        model = model_plus.model
-        model = do_necessary_conversions(model, params)
-        output_type = pick_output_type(model, args.outtype)
-        model = convert_to_output_type(model, output_type)
-        outfile = args.outfile or default_outfile(model_plus.paths, output_type)
-        OutputFile.write_all(outfile, params, output_type, model, vocab)
-        print(f"Wrote {outfile}")
-
-
-if __name__ == '__main__':
-    main()

From 58bde5c5c1c4d7f1201819a16734ea1b933279f4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 04:35:06 +0200
Subject: [PATCH 228/242] Delete convert-permute-debug.py

---
 convert-permute-debug.py | 1032 --------------------------------------
 1 file changed, 1032 deletions(-)
 delete mode 100644 convert-permute-debug.py

diff --git a/convert-permute-debug.py b/convert-permute-debug.py
deleted file mode 100644
index fcbb5f2da1173..0000000000000
--- a/convert-permute-debug.py
+++ /dev/null
@@ -1,1032 +0,0 @@
-#!/usr/bin/env python
-
-import gguf
-import argparse
-import concurrent.futures
-import copy
-import enum
-import faulthandler
-import functools
-import io
-import itertools
-import json
-import math
-import mmap
-import pickle
-import re
-import signal
-import struct
-import sys
-import zipfile
-import numpy as np
-
-from abc import ABCMeta, abstractmethod
-from dataclasses import dataclass
-from pathlib import Path
-from typing import (IO, TYPE_CHECKING, Any, Callable, Dict, Iterable, List, Literal, Optional, Sequence, Tuple, TypeVar, Union)
-from sentencepiece import SentencePieceProcessor  # type: ignore
-
-if TYPE_CHECKING:
-    from typing_extensions import TypeAlias
-
-if hasattr(faulthandler, 'register') and hasattr(signal, 'SIGUSR1'):
-    faulthandler.register(signal.SIGUSR1)
-
-NDArray: 'TypeAlias' = 'np.ndarray[Any, Any]'
-
-ARCH=gguf.MODEL_ARCH.LLAMA
-NAMES=gguf.MODEL_TENSOR_NAMES[ARCH]
-
-#
-# data types
-#
-
-@dataclass(frozen=True)
-class UnquantizedDataType:
-    name: str
-
-DT_F16  = UnquantizedDataType('F16')
-DT_F32  = UnquantizedDataType('F32')
-DT_I32  = UnquantizedDataType('I32')
-DT_BF16 = UnquantizedDataType('BF16')
-
-DataType = Union[UnquantizedDataType]
-
-DATA_TYPE_TO_NUMPY: Dict[DataType, 'np.dtype[Any]'] = {
-    DT_BF16: np.dtype(np.uint16),
-    DT_F16:  np.dtype(np.float16),
-    DT_F32:  np.dtype(np.float32),
-    DT_I32:  np.dtype(np.int32),
-}
-
-NUMPY_TYPE_TO_DATA_TYPE: Dict['np.dtype[Any]', DataType] = \
-    {dtype: data_type for (data_type, dtype) in DATA_TYPE_TO_NUMPY.items()}
-
-SAFETENSORS_DATA_TYPES: Dict[str, DataType] = {
-    'BF16': DT_BF16,
-    'F16': DT_F16,
-    'F32': DT_F32,
-    'I32': DT_I32,
-}
-
-class GGMLFileType(enum.Enum):
-    AllF32    = 0
-    MostlyF16 = 1  # except 1d tensors
-
-    def type_for_tensor(self, name: str, tensor: 'LazyTensor') -> DataType:
-        if len(tensor.shape) == 1:
-            # 1D tensors are always F32.
-            return DT_F32
-        elif self == GGMLFileType.AllF32:
-            return DT_F32
-        elif self == GGMLFileType.MostlyF16:
-            return DT_F16
-        else:
-            raise ValueError(self)
-
-
-#
-# hparams loading
-#
-
-@dataclass
-class Params:
-    n_vocab:    int
-    n_embd:     int
-    n_mult:     int
-    n_layer:    int
-    n_ctx:      int
-    n_ff:       int
-    n_head:     int
-    n_head_kv:  int
-    f_norm_eps: float
-
-    @staticmethod
-    def find_n_mult(n_ff: int, n_embd: int) -> int:
-        # hardcoded magic range
-        for n_mult in range(8192, 1, -1):
-            calc_ff = (((8*n_embd) // 3 + n_mult - 1) // n_mult)*n_mult
-            if calc_ff == n_ff:
-                return n_mult
-        raise Exception(f"failed to find n_mult for (n_ff={n_ff}, n_embd={n_embd}).")
-
-    @staticmethod
-    def guessed(model: 'LazyModel') -> 'Params':
-        # try transformer naming first
-        n_vocab, n_embd = model["model.embed_tokens.weight"].shape if "model.embed_tokens.weight" in model else model["tok_embeddings.weight"].shape
-
-        # try transformer naming first
-        if "model.layers.0.self_attn.q_proj.weight" in model:
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.q_proj.weight" not in model)
-        elif "model.layers.0.self_attn.W_pack.weight" in model:   # next: try baichuan naming
-            n_layer=next(i for i in itertools.count() if f"model.layers.{i}.self_attn.W_pack.weight" not in model)
-        else:
-            n_layer=next(i for i in itertools.count() if f"layers.{i}.attention.wq.weight" not in model)
-
-        if n_layer < 1:
-            raise Exception("failed to guess 'n_layer'. This model is unknown or unsupported.\n"
-                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
-
-        n_head = n_embd // 128 # guessed
-        n_mult = 256           # guessed
-
-        # TODO: verify this
-        n_ff = int(2 * (4 * n_embd) / 3)
-        n_ff = n_mult * ((n_ff + n_mult - 1) // n_mult)
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = -1,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head,
-            f_norm_eps = 1e-5,
-        )
-
-    @staticmethod
-    def loadHFTransformerJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab    = config["vocab_size"]
-        n_embd     = config["hidden_size"]
-        n_layer    = config["num_hidden_layers"]
-        n_ff       = config["intermediate_size"]
-        n_head     = config["num_attention_heads"]
-        n_head_kv  = config["num_key_value_heads"] if "num_key_value_heads" in config else n_head
-        f_norm_eps = config["rms_norm_eps"]
-
-        n_mult = Params.find_n_mult(n_ff, n_embd)
-
-        if "max_sequence_length" in config:
-            n_ctx = config["max_sequence_length"]
-        elif "max_position_embeddings" in config:
-            n_ctx = config["max_position_embeddings"]
-        else:
-            raise Exception("failed to guess 'n_ctx'. This model is unknown or unsupported.\n"
-                            "Suggestion: provide 'config.json' of the model in the same directory containing model files.")
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = n_ctx,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head_kv,
-            f_norm_eps = f_norm_eps,
-        )
-
-    # LLaMA v2 70B params.json
-    # {"dim": 8192, "multiple_of": 4096, "ffn_dim_multiplier": 1.3, "n_heads": 64, "n_kv_heads": 8, "n_layers": 80, "norm_eps": 1e-05, "vocab_size": -1
-    @staticmethod
-    def loadOriginalParamsJson(model: 'LazyModel', config_path: 'Path') -> 'Params':
-        config = json.load(open(config_path))
-
-        n_vocab    = config["vocab_size"]
-        n_embd     = config["dim"]
-        n_layer    = config["n_layers"]
-        n_mult     = config["multiple_of"]
-        n_ctx      = 2048 if config["norm_eps"] == 1e-06 else 4096 # hack to determine LLaMA v1 vs v2
-        n_ff       = -1
-        n_head     = config["n_heads"]
-        n_head_kv  = config["n_kv_heads"] if "n_kv_heads" in config else n_head
-        f_norm_eps = config["norm_eps"]
-
-        if n_vocab == -1:
-            n_vocab = model["tok_embeddings.weight"].shape[0]
-
-        if n_ff == -1:
-            n_ff = model["layers.0.feed_forward.w1.weight"].shape[0]
-
-        return Params(
-            n_vocab    = n_vocab,
-            n_embd     = n_embd,
-            n_mult     = n_mult,
-            n_layer    = n_layer,
-            n_ctx      = n_ctx,
-            n_ff       = n_ff,
-            n_head     = n_head,
-            n_head_kv  = n_head_kv,
-            f_norm_eps = f_norm_eps,
-        )
-
-    @staticmethod
-    def load(model_plus: 'ModelPlus') -> 'Params':
-        hf_config_path   = model_plus.paths[0].parent / "config.json"
-        orig_config_path = model_plus.paths[0].parent / "params.json"
-
-        if hf_config_path.exists():
-            params = Params.loadHFTransformerJson(model_plus.model, hf_config_path)
-        elif orig_config_path.exists():
-            params = Params.loadOriginalParamsJson(model_plus.model, orig_config_path)
-        else:
-            params = Params.guessed(model_plus.model)
-
-        return params
-
-
-#
-# vocab
-#
-
-class BpeVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
-        self.bpe_tokenizer = json.loads(open(str(fname_tokenizer), encoding="utf-8").read())
-        added_tokens: Dict[str, int]
-        if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
-        else:
-            added_tokens = {}
-        vocab_size: int = len(self.bpe_tokenizer)
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
-        self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
-
-    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        tokenizer = self.bpe_tokenizer
-        from transformers.models.gpt2 import tokenization_gpt2
-        byte_encoder = tokenization_gpt2.bytes_to_unicode()
-        byte_decoder = {v: k for k, v in byte_encoder.items()}
-        for i, item in enumerate(tokenizer):
-            text: bytes = item.encode("utf-8")
-            score: float = -i
-            yield text, score
-
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        yield from self.bpe_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"BpeVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-
-class SentencePieceVocab:
-    def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) -> None:
-        self.sentencepiece_tokenizer = SentencePieceProcessor(str(fname_tokenizer))
-        added_tokens: Dict[str, int]
-        if fname_added_tokens is not None:
-            added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
-        else:
-            added_tokens = {}
-        vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids   = sorted(added_tokens.values())
-        if expected_ids != actual_ids:
-            raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
-
-        items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
-        self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
-
-    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        tokenizer = self.sentencepiece_tokenizer
-        for i in range(tokenizer.vocab_size()):
-            piece = tokenizer.id_to_piece(i)
-            text: bytes = piece.encode("utf-8")
-            score: float = tokenizer.get_score(i)
-            yield text, score
-
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        for text in self.added_tokens_list:
-            score = -1000.0
-            yield text.encode("utf-8"), score
-
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
-        yield from self.sentencepiece_tokens()
-        yield from self.added_tokens()
-
-    def __repr__(self) -> str:
-        return f"<SentencePieceVocab with {self.vocab_size_base} base tokens and {len(self.added_tokens_list)} added tokens>"
-
-Vocab = Union[BpeVocab, SentencePieceVocab]
-
-
-#
-# data loading
-# TODO: reuse (probably move to gguf.py?)
-#
-
-def permute(weights: NDArray, n_head: int, n_head_kv: int) -> NDArray:
-    print( "permute " + str(weights.shape[0]) + " x " + str(weights.shape[1]) + " nhead " + str(n_head) + " nheadkv " + str(n_head_kv) )
-    if n_head_kv is not None and n_head != n_head_kv:
-        n_head //= n_head_kv
-    return (weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
-                .swapaxes(1, 2)
-                .reshape(weights.shape))
-
-
-class Tensor(metaclass=ABCMeta):
-    data_type: DataType
-
-    @abstractmethod
-    def astype(self, data_type: DataType) -> 'Tensor': ...
-    @abstractmethod
-    def permute(self, n_head: int, n_head_kv: int) -> 'Tensor': ...
-    @abstractmethod
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def part(self, n_part: int) -> 'UnquantizedTensor': ...
-    @abstractmethod
-    def to_ggml(self) -> 'GGMLCompatibleTensor': ...
-
-
-def bf16_to_fp32(bf16_arr: np.ndarray) -> np.ndarray:
-    assert bf16_arr.dtype == np.uint16, f"Input array should be of dtype uint16, but got {bf16_arr.dtype}"
-    fp32_arr = bf16_arr.astype(np.uint32) << 16
-    return fp32_arr.view(np.float32)
-
-
-class UnquantizedTensor(Tensor):
-    def __init__(self, ndarray: NDArray) -> None:
-        assert isinstance(ndarray, np.ndarray)
-        self.ndarray = ndarray
-        self.data_type = NUMPY_TYPE_TO_DATA_TYPE[ndarray.dtype]
-
-    def astype(self, data_type: DataType) -> Tensor:
-        dtype = DATA_TYPE_TO_NUMPY[data_type]
-        if self.data_type == DT_BF16:
-            self.ndarray = bf16_to_fp32(self.ndarray)
-        return UnquantizedTensor(self.ndarray.astype(dtype))
-
-    def to_ggml(self) -> 'UnquantizedTensor':
-        return self
-
-    def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head))
-
-    def part(self, n_part: int) -> 'UnquantizedTensor':
-        r = self.ndarray.shape[0] // 3
-        return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...])
-
-    def permute(self, n_head: int, n_head_kv: int) -> 'UnquantizedTensor':
-        return UnquantizedTensor(permute(self.ndarray, n_head, n_head_kv))
-
-
-def load_unquantized(lazy_tensor: 'LazyTensor', expected_dtype: Any = None, convert: bool = False) -> NDArray:
-    tensor = lazy_tensor.load()
-    assert isinstance(tensor, UnquantizedTensor)
-
-    # double-check:
-    actual_shape = list(tensor.ndarray.shape)
-    assert actual_shape == lazy_tensor.shape, (actual_shape, lazy_tensor.shape)
-    if expected_dtype is not None and expected_dtype != tensor.ndarray.dtype:
-        if convert:
-            tensor.ndarray = tensor.ndarray.astype(expected_dtype)
-        else:
-            raise ValueError(f'expected this tensor to have dtype {expected_dtype}, got {tensor.ndarray.dtype}')
-
-    return tensor.ndarray
-
-
-GGMLCompatibleTensor = Union[UnquantizedTensor]
-
-
-class DeferredPermutedTensor(Tensor):
-    def __init__(self, base: Tensor, n_head: int, n_head_kv: int) -> None:
-        self.base = base
-        self.n_head = n_head
-        self.data_type = self.base.data_type
-
-    def astype(self, data_type: DataType) -> Tensor:
-        return self.base.astype(data_type).permute(self.n_head, self.n_head_kv)
-
-    def to_ggml(self) -> GGMLCompatibleTensor:
-        return self.base.to_ggml().permute(self.n_head, self.n_head_kv)
-
-    def permute(self, n_head: int, n_head_kv: int) -> Tensor:
-        raise Exception("shouldn't permute twice")
-
-
-@dataclass
-class LazyTensor:
-    _load: Callable[[], Tensor]
-    shape: List[int]
-    data_type: DataType
-    description: str
-
-    def load(self) -> Tensor:
-        ret = self._load()
-        assert ret.data_type == self.data_type, (self.data_type, ret.data_type, self.description)
-        return ret
-
-    def astype(self, data_type: DataType) -> 'LazyTensor':
-        self.validate_conversion_to(data_type)
-
-        def load() -> Tensor:
-            return self.load().astype(data_type)
-        return LazyTensor(load, self.shape, data_type, f'convert({data_type}) {self.description}')
-
-    def validate_conversion_to(self, data_type: DataType) -> None:
-        if data_type == self.data_type:
-            return
-
-
-LazyModel = Dict[str, LazyTensor]
-
-
-@dataclass
-class ModelPlus:
-    model: LazyModel
-    paths: List[Path]  # Where this was read from.
-    format: Literal['ggml', 'torch', 'safetensors']
-    vocab: Optional[Vocab]  # For GGML models (which have vocab built in), the vocab.
-
-
-def merge_sharded(models: List[LazyModel]) -> LazyModel:
-    # Original LLaMA models have each file contain one part of each tensor.
-    # Use a dict instead of a set to preserve order.
-    names = {name: None for model in models for name in model}
-
-    def convert(name: str) -> LazyTensor:
-        lazy_tensors: List[LazyTensor] = [model[name] for model in models]
-        if len(lazy_tensors) == 1:
-            # only one file; don't go through this procedure since there might
-            # be quantized tensors
-            return lazy_tensors[0]
-        if len(lazy_tensors[0].shape) == 1:
-            # the tensor is just duplicated in every file
-            return lazy_tensors[0]
-        if name.startswith('tok_embeddings.') or \
-           name.endswith('.attention.wo.weight') or \
-           name.endswith('.feed_forward.w2.weight'):
-            # split by columns
-            axis = 1
-        else:
-            # split by rows
-            axis = 0
-        concatenated_shape = list(lazy_tensors[0].shape)
-        concatenated_shape[axis] = sum(tensor.shape[axis] for tensor in lazy_tensors)
-
-        def load() -> UnquantizedTensor:
-            ndarrays = [load_unquantized(tensor) for tensor in lazy_tensors]
-            concatenated: NDArray = np.concatenate(ndarrays, axis=axis)
-            return UnquantizedTensor(concatenated)
-        description = 'concatenated[[' + '] | ['.join(lt.description for lt in lazy_tensors) + ']]'
-        return LazyTensor(load, concatenated_shape, lazy_tensors[0].data_type, description)
-    return {name: convert(name) for name in names}
-
-
-def merge_multifile_models(models_plus: List[ModelPlus]) -> ModelPlus:
-    formats = set(mp.format for mp in models_plus)
-    assert len(formats) == 1, "different formats?"
-    format = formats.pop()
-    paths = [path for mp in models_plus for path in mp.paths]
-    # Use the first non-None vocab, if any.
-    try:
-        vocab = next(mp.vocab for mp in models_plus if mp.vocab is not None)
-    except StopIteration:
-        vocab = None
-
-    if any("model.embed_tokens.weight" in mp.model for mp in models_plus):
-        # Transformers models put different tensors in different files, but
-        # don't split indivdual tensors between files.
-        model: LazyModel = {}
-        for mp in models_plus:
-            model.update(mp.model)
-    else:
-        model = merge_sharded([mp.model for mp in models_plus])
-
-    return ModelPlus(model, paths, format, vocab)
-
-
-def permute_lazy(lazy_tensor: LazyTensor, n_head: int, n_head_kv: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute(n_head, n_head_kv)
-    return LazyTensor(load, lazy_tensor.shape, lazy_tensor.data_type, f'permute({n_head}, {n_head_kv}) ' + lazy_tensor.description)
-
-def permute_part_lazy(lazy_tensor: LazyTensor, n_part: int, n_head: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().permute_part(n_part, n_head)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, f'permute({n_head}) ' + lazy_tensor.description)
-
-def part_lazy(lazy_tensor: LazyTensor, n_part: int) -> LazyTensor:
-    def load() -> Tensor:
-        return lazy_tensor.load().part(n_part)
-    s = lazy_tensor.shape.copy()
-    s[0] = s[0] // 3
-    return LazyTensor(load, s, lazy_tensor.data_type, 'part ' + lazy_tensor.description)
-
-
-# Functionality that simulates `torch.load` but where individual tensors are
-# only loaded into memory on demand, not all at once.
-# PyTorch can't do this natively as of time of writing:
-# - https://github.com/pytorch/pytorch/issues/64327
-# This allows us to de-shard without multiplying RAM usage, and also
-# conveniently drops the PyTorch dependency (though we still need numpy).
-
-
-@dataclass
-class LazyStorageKind:
-    data_type: DataType
-
-
-@dataclass
-class LazyStorage:
-    load: Callable[[int, int], NDArray]
-    kind: LazyStorageKind
-    description: str
-
-
-class LazyUnpickler(pickle.Unpickler):
-    def __init__(self, fp: IO[bytes], data_base_path: str, zip_file: zipfile.ZipFile):
-        super().__init__(fp)
-        self.data_base_path = data_base_path
-        self.zip_file = zip_file
-
-    def persistent_load(self, pid: Any) -> Any:
-        assert pid[0] == 'storage'
-        assert isinstance(pid[1], LazyStorageKind)
-        data_type = pid[1].data_type
-        filename_stem = pid[2]
-        filename = self.data_base_path + '/' + filename_stem
-        info = self.zip_file.getinfo(filename)
-
-        def load(offset: int, elm_count: int) -> NDArray:
-            dtype = DATA_TYPE_TO_NUMPY.get(data_type)
-            if dtype is None:
-                raise Exception("tensor stored in unsupported format")
-            fp = self.zip_file.open(info)
-            fp.seek(offset * dtype.itemsize)
-            size = elm_count * dtype.itemsize
-            data = fp.read(size)
-            assert len(data) == size
-            return np.frombuffer(data, dtype)
-        description = f'storage data_type={data_type} path-in-zip={filename} path={self.zip_file.filename}'
-        return LazyStorage(load=load, kind=pid[1], description=description)
-
-    # @staticmethod
-    def lazy_rebuild_tensor_v2(storage: Any, storage_offset: Any, size: Any, stride: Any,
-                               # pyright: ignore[reportSelfClsParameterName]
-                               requires_grad: Any, backward_hooks: Any, metadata: Any = None) -> LazyTensor:
-        assert isinstance(storage, LazyStorage)
-
-        def load() -> UnquantizedTensor:
-            elm_count = stride[0] * size[0]
-            return UnquantizedTensor(storage.load(storage_offset, elm_count).reshape(size))
-        description = f'pickled storage_offset={storage_offset} in {storage.description}'
-        return LazyTensor(load, list(size), storage.kind.data_type, description)
-
-    # @staticmethod
-    def rebuild_from_type_v2(func, new_type, args, state):
-        return func(*args)
-
-    CLASSES: Dict[Any, Any] = {
-        ('torch._tensor', '_rebuild_from_type_v2'): rebuild_from_type_v2,
-        ('torch._utils', '_rebuild_tensor_v2'): lazy_rebuild_tensor_v2,
-        ('torch', 'BFloat16Storage'): LazyStorageKind(DT_BF16),
-        ('torch', 'HalfStorage'): LazyStorageKind(DT_F16),
-        ('torch', 'FloatStorage'): LazyStorageKind(DT_F32),
-        ('torch', 'IntStorage'): LazyStorageKind(DT_I32),
-        ('torch', 'Tensor'): LazyTensor,
-    }
-
-    def find_class(self, module: str, name: str) -> Any:
-        if not module.startswith('torch'):
-            return super().find_class(module, name)
-        return self.CLASSES[(module, name)]
-
-
-def lazy_load_torch_file(outer_fp: IO[bytes], path: Path) -> ModelPlus:
-    zf = zipfile.ZipFile(outer_fp)
-    pickle_paths = [name for name in zf.namelist() if name.endswith('.pkl')]
-    assert len(pickle_paths) == 1, pickle_paths
-    pickle_fp = zf.open(pickle_paths[0], 'r')
-    unpickler = LazyUnpickler(pickle_fp,
-                              data_base_path=pickle_paths[0][:-4],
-                              zip_file=zf)
-    model = unpickler.load()
-    as_dict = dict(model.items())
-    return ModelPlus(model=as_dict, paths=[path], format='torch', vocab=None)
-
-
-def lazy_load_safetensors_file(fp: IO[bytes], path: Path) -> ModelPlus:
-    header_size, = struct.unpack('<Q', fp.read(8))
-    header: Dict[str, Dict[str, Any]] = json.loads(fp.read(header_size))
-    # Use mmap for the actual data to avoid race conditions with the file offset.
-    mapped = memoryview(mmap.mmap(fp.fileno(), 0, access=mmap.ACCESS_READ))
-    byte_buf = mapped[8 + header_size:]
-
-    def convert(info: Dict[str, Any]) -> LazyTensor:
-        data_type = SAFETENSORS_DATA_TYPES[info['dtype']]
-        numpy_dtype = DATA_TYPE_TO_NUMPY[data_type]
-        shape: List[int] = info['shape']
-        begin, end = info['data_offsets']
-        assert 0 <= begin <= end <= len(byte_buf)
-        assert end - begin == math.prod(shape) * numpy_dtype.itemsize
-        buf = byte_buf[begin:end]
-
-        def load() -> UnquantizedTensor:
-            return UnquantizedTensor(np.frombuffer(buf, dtype=numpy_dtype).reshape(shape))
-        description = f'safetensors begin={begin} end={end} type={data_type} path={path}'
-        return LazyTensor(load, shape, data_type, description)
-    model = {name: convert(info) for (name, info) in header.items() if name != '__metadata__'}
-    return ModelPlus(model=model, paths=[path], format='safetensors', vocab=None)
-
-
-def must_read(fp: IO[bytes], length: int) -> bytes:
-    ret = fp.read(length)
-    if len(ret) < length:
-        raise Exception("unexpectedly reached end of file")
-    return ret
-
-
-@functools.lru_cache(maxsize=None)
-def lazy_load_file(path: Path) -> ModelPlus:
-    fp = open(path, 'rb')
-    first8 = fp.read(8)
-    fp.seek(0)
-    if first8[:2] == b'PK':
-        # A zip file, i.e. PyTorch format
-        return lazy_load_torch_file(fp, path)
-    elif struct.unpack('<Q', first8)[0] < 16 * 1024 * 1024:
-        # Probably safetensors
-        return lazy_load_safetensors_file(fp, path)
-    else:
-        raise ValueError(f"unknown format: {path}")
-
-
-In = TypeVar('In')
-Out = TypeVar('Out')
-
-def bounded_parallel_map(func: Callable[[In], Out], iterable: Iterable[In], concurrency: int) -> Iterable[Out]:
-    '''Parallel map, but with backpressure.  If the caller doesn't call `next`
-    fast enough, this will stop calling `func` at some point rather than
-    letting results pile up in memory.  Specifically, there is a max of one
-    output value buffered per thread.'''
-    with concurrent.futures.ThreadPoolExecutor() as executor:
-        futures: List[concurrent.futures.Future[Out]] = []
-        items_rev = list(iterable)[::-1]
-        for i in range(min(concurrency, len(items_rev))):
-            futures.append(executor.submit(func, items_rev.pop()))
-        while futures:
-            result = futures.pop(0).result()
-            if items_rev:
-                futures.append(executor.submit(func, items_rev.pop()))
-            yield result
-
-
-def check_vocab_size(params: Params, vocab: Vocab) -> None:
-    if params.n_vocab != vocab.vocab_size:
-        assert isinstance(vocab, BpeVocab) or isinstance(vocab, SentencePieceVocab)
-        if params.n_vocab == vocab.vocab_size_base:
-            print("Ignoring added_tokens.json since model matches vocab size without it.")
-            vocab.added_tokens_list = []
-            vocab.vocab_size = vocab.vocab_size_base
-            return
-        msg = f"Vocab size mismatch (model has {params.n_vocab}, but {vocab.fname_tokenizer}"
-        if vocab.fname_added_tokens is not None:
-            msg += f" combined with {vocab.fname_added_tokens}"
-        msg += f" has {vocab.vocab_size})."
-        if vocab.vocab_size < params.n_vocab < vocab.vocab_size + 20 and vocab.fname_added_tokens is None:
-            msg += f"  Most likely you are missing added_tokens.json (should be in {vocab.fname_tokenizer.parent})."
-        raise Exception(msg)
-
-
-class OutputFile:
-    def __init__(self, fname_out: Path) -> None:
-        self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
-
-    def add_meta_arch(self, params: Params) -> None:
-        self.gguf.add_context_length      (params.n_ctx)
-        self.gguf.add_embedding_length    (params.n_embd)
-        self.gguf.add_block_count         (params.n_layer)
-        self.gguf.add_feed_forward_length (params.n_ff)
-        self.gguf.add_rope_dimension_count(params.n_embd // params.n_head)
-        self.gguf.add_head_count          (params.n_head)
-        self.gguf.add_head_count_kv       (params.n_head_kv)
-        self.gguf.add_layer_norm_rms_eps  (params.f_norm_eps)
-
-    def add_meta_vocab(self, vocab: Vocab) -> None:
-        tokens = []
-        scores = []
-        for text, score in vocab.all_tokens():
-            tokens.append(text)
-            scores.append(score)
-
-        self.gguf.add_tokenizer_model("llama")
-        self.gguf.add_token_list(tokens)
-        self.gguf.add_token_scores(scores)
-        #self.gguf.add_token_types(toktypes) # TODO: add this
-
-        # TODO: added / special tokens
-
-    def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
-        n_elements = 1
-        for dim in tensor.shape:
-            n_elements *= dim
-        data_type = DATA_TYPE_TO_NUMPY[tensor.data_type]
-        data_nbytes = n_elements * data_type.itemsize
-        self.gguf.add_tensor_info(name, tensor.shape, data_type, data_nbytes)
-
-    def write_meta(self) -> None:
-        self.gguf.write_header_to_file()
-        self.gguf.write_kv_data_to_file()
-
-    def write_tensor_info(self) -> None:
-        self.gguf.write_ti_data_to_file()
-
-    def close(self) -> None:
-        self.gguf.close()
-
-    @staticmethod
-    def write_vocab_only(fname_out: Path, params: Params, vocab: Vocab) -> None:
-        check_vocab_size(params, vocab)
-
-        of = OutputFile(fname_out)
-
-        # meta data
-        of.add_meta_arch(params)
-        of.add_meta_vocab(vocab)
-        of.write_meta()
-
-        of.close()
-
-    @staticmethod
-    def write_all(fname_out: Path, params: Params, model: LazyModel, vocab: Vocab) -> None:
-        check_vocab_size(params, vocab)
-
-        of = OutputFile(fname_out)
-
-        # meta data
-        of.add_meta_arch(params)
-        of.add_meta_vocab(vocab)
-
-        # tensor info
-        for name, lazy_tensor in model.items():
-            of.add_tensor_info(name, lazy_tensor)
-
-        of.write_meta()
-        of.write_tensor_info()
-
-        def do_item(item: Tuple[str, LazyTensor]) -> NDArray:
-            name, lazy_tensor = item
-            return lazy_tensor.load().to_ggml().ndarray
-
-        # tensor data
-        ndarrays = bounded_parallel_map(do_item, model.items(), concurrency=1)
-        for i, ((name, lazy_tensor), ndarray) in enumerate(zip(model.items(), ndarrays)):
-            size = ' x '.join(f"{dim:6d}" for dim in lazy_tensor.shape)
-            padi = len(str(len(model)))
-            print(f"[{i+1:{padi}d}/{len(model)}] Writing tensor {name:38s} | size {size:16} | type {lazy_tensor.data_type}")
-            of.gguf.write_tensor_data(ndarray)
-
-        of.close()
-
-def pick_output_type(model: LazyModel, output_type_str: Optional[str]) -> GGMLFileType:
-    wq_type = model[NAMES[gguf.MODEL_TENSOR.ATTN_Q].format(bid=0)+".weight"].data_type
-
-    if output_type_str == "f32" or (output_type_str is None and wq_type == DT_F32):
-        return GGMLFileType.AllF32
-    if output_type_str == "f16" or (output_type_str is None and wq_type in (DT_F16, DT_BF16)):
-        return GGMLFileType.MostlyF16
-
-    name_to_type = {name: lazy_tensor.data_type for (name, lazy_tensor) in model.items()}
-
-    raise Exception(f"Unexpected combination of types: {name_to_type}")
-
-def convert_to_output_type(model: LazyModel, output_type: GGMLFileType) -> LazyModel:
-    return {name: tensor.astype(output_type.type_for_tensor(name, tensor))
-            for (name, tensor) in model.items()}
-
-def convert_model_names(model: LazyModel, params: Params) -> LazyModel:
-    tmap = gguf.get_tensor_name_map(ARCH, params.n_layer)
-
-    tmp = model
-
-    # HF models permut or pack some of the tensors, so we need to undo that
-    for i in itertools.count():
-        if f"model.layers.{i}.self_attn.q_proj.weight" in model:
-            print(f"Permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.q_proj.weight"], params.n_head, params.n_head)
-            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_lazy(model[f"model.layers.{i}.self_attn.k_proj.weight"], params.n_head, params.n_head_kv)
-           #tmp[f"model.layers.{i}.self_attn.v_proj.weight"] =              model[f"model.layers.{i}.self_attn.v_proj.weight"]
-        elif f"model.layers.{i}.self_attn.W_pack.weight" in model:
-            print(f"Unpacking and permuting layer {i}")
-            tmp[f"model.layers.{i}.self_attn.q_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 0, params.n_head, params.n_head)
-            tmp[f"model.layers.{i}.self_attn.k_proj.weight"] = permute_part_lazy(model[f"model.layers.{i}.self_attn.W_pack.weight"], 1, params.n_head, params.n_head_kv)
-            tmp[f"model.layers.{i}.self_attn.v_proj.weight"] = part_lazy        (model[f"model.layers.{i}.self_attn.W_pack.weight"], 2)
-        else:
-            break
-
-    out: LazyModel = {}
-    for name, lazy_tensor in model.items():
-        name_new = name
-
-        if name in tmap:
-            name_new = tmap[name]
-        elif name.endswith(".weight") and name[:-7] in tmap:
-            name_new = tmap[name[:-7]] + ".weight"
-        elif name.endswith(".bias") and name[:-5] in tmap:
-            name_new = tmap[name[:-5]] + ".bias"
-        else:
-            raise Exception(f"Unexpected tensor name: {name}")
-
-        if gguf.should_skip_tensor_TMP(ARCH, params.n_layer, name_new):
-            print(f"skipping tensor {name_new}")
-            continue
-        else:
-            print(f"{name:48s} -> {name_new:40s} | {lazy_tensor.data_type} | {lazy_tensor.shape}")
-            out[name_new] = lazy_tensor
-
-    return out
-
-def nth_multifile_path(path: Path, n: int) -> Optional[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the nth path in the model.
-    '''
-    # Support the following patterns:
-    patterns: List[Tuple[str, str]] = [
-        # - x.00.pth, x.01.pth, etc.
-        (r'\.[0-9]{2}\.pth$', f'.{n:02}.pth'),
-        # - x-00001-of-00002.bin, x-00002-of-00002.bin, etc.
-        (r'-[0-9]{5}-of-(.*)$', fr'-{n:05}-of-\1'),
-        # x.bin, x.bin.1, etc.
-        (r'(\.[0-9]+)?$', r'\1' if n == 0 else fr'\1.{n}')
-    ]
-    for regex, replacement in patterns:
-        if re.search(regex, path.name):
-            new_path = path.with_name(re.sub(regex, replacement, path.name))
-            if new_path.exists():
-                return new_path
-    return None
-
-
-def find_multifile_paths(path: Path) -> List[Path]:
-    '''Given any path belonging to a multi-file model (e.g. foo.bin.1), return
-    the whole list of paths in the model.
-    '''
-    ret: List[Path] = []
-    for i in itertools.count():
-        nth_path = nth_multifile_path(path, i)
-        if nth_path is None:
-            break
-        ret.append(nth_path)
-    if not ret:
-        # No matches.  This should only happen if the file was named, e.g.,
-        # foo.0, and there was no file named foo.  Oh well, try to process it
-        # as a single file.
-        return [path]
-    return ret
-
-
-def load_some_model(path: Path) -> ModelPlus:
-    '''Load a model of any supported format.'''
-    # Be extra-friendly and accept either a file or a directory:
-    if path.is_dir():
-        # Check if it's a set of safetensors files first
-        files = list(path.glob("model-00001-of-*.safetensors"))
-        if not files:
-            # Try the PyTorch patterns too, with lower priority
-            globs = ["consolidated.00.pth", "pytorch_model-00001-of-*.bin", "*.pt", "pytorch_model.bin"]
-            files = [file for glob in globs for file in path.glob(glob)]
-        if not files:
-            raise Exception(f"Can't find model in directory {path}")
-        if len(files) > 1:
-            raise Exception(f"Found multiple models in {path}, not sure which to pick: {files}")
-        path = files[0]
-
-    paths = find_multifile_paths(path)
-    models_plus: List[ModelPlus] = []
-    for path in paths:
-        print(f"Loading model file {path}")
-        models_plus.append(lazy_load_file(path))
-
-    model_plus = merge_multifile_models(models_plus)
-    return model_plus
-
-
-def load_vocab(path: Path, vocabtype: Optional[str]) -> Union[BpeVocab, SentencePieceVocab]:
-    # Be extra-friendly and accept either a file or a directory.  Also, if it's
-    # a directory, it might be the model directory, and tokenizer.model might
-    # be in the parent of that.
-    if path.is_dir():
-        vocab_file = "tokenizer.model"
-        if vocabtype == 'bpe':
-            vocab_file = "vocab.json"
-        path2 = path / vocab_file
-        # Use `.parent` instead of /.. to handle the symlink case better.
-        path3 = path.parent / vocab_file
-        if path2.exists():
-            path = path2
-        elif path3.exists():
-            path = path3
-        else:
-            raise FileNotFoundError(
-                f"Could not find tokenizer.model in {path} or its parent; "
-                "if it's in another directory, pass the directory as --vocab-dir")
-
-    print(f"Loading vocab file '{path}', type '{vocabtype}'")
-
-    added_tokens_path = path.parent / "added_tokens.json"
-    if vocabtype == "bpe":
-        return BpeVocab(path, added_tokens_path if added_tokens_path.exists() else None)
-    elif vocabtype == "spm":
-        return SentencePieceVocab(path, added_tokens_path if added_tokens_path.exists() else None)
-    else:
-        raise ValueError(f"Unsupported vocabulary type {vocabtype}")
-
-
-def default_outfile(model_paths: List[Path], file_type: GGMLFileType) -> Path:
-    namestr = {
-        GGMLFileType.AllF32:    "f32",
-        GGMLFileType.MostlyF16: "f16",
-    }[file_type]
-    ret = model_paths[0].parent / f"ggml-model-{namestr}.gguf"
-    if ret in model_paths:
-        sys.stderr.write(
-            f"Error: Default output path ({ret}) would overwrite the input. "
-            "Please explicitly specify a path using --outfile.\n")
-        sys.exit(1)
-    return ret
-
-
-def do_dump_model(model_plus: ModelPlus) -> None:
-    print(f"model_plus.paths = {model_plus.paths!r}")
-    print(f"model_plus.format = {model_plus.format!r}")
-    print(f"model_plus.vocab = {model_plus.vocab!r}")
-    for name, lazy_tensor in model_plus.model.items():
-        print(f"{name}: shape={lazy_tensor.shape} type={lazy_tensor.data_type}; {lazy_tensor.description}")
-
-
-def main(args_in: Optional[List[str]] = None) -> None:
-    parser = argparse.ArgumentParser(description="Convert a LLaMa model to a GGML compatible file")
-    parser.add_argument("--dump",        action="store_true",    help="don't convert, just show what's in the model")
-    parser.add_argument("--dump-single", action="store_true",    help="don't convert, just show what's in a single model file")
-    parser.add_argument("--vocab-only",  action="store_true",    help="extract only the vocab")
-    parser.add_argument("--outtype",     choices=["f32", "f16"], help="output format (default: based on input)")
-    parser.add_argument("--vocab-dir",   type=Path,              help="directory containing tokenizer.model, if separate from model file")
-    parser.add_argument("--outfile",     type=Path,              help="path to write to; default: based on input")
-    parser.add_argument("model",         type=Path,              help="directory containing model file, or model file itself (*.pth, *.pt, *.bin)")
-    parser.add_argument("--vocabtype",   choices=["spm", "bpe"], help="vocab format (default: spm)", default="spm")
-    parser.add_argument("--ctx",         type=int,               help="model training context (default: based on input)")
-    args = parser.parse_args(args_in)
-
-    if args.dump_single:
-        model_plus = lazy_load_file(args.model)
-        do_dump_model(model_plus)
-
-    model_plus = load_some_model(args.model)
-
-    params = Params.load(model_plus)
-    if params.n_ctx == -1:
-        if args.ctx is None:
-            raise Exception("The model doesn't have a context size, and you didn't specify one with --ctx\n"
-                            "Please specify one with --ctx:\n"
-                            " - LLaMA v1: --ctx 2048\n"
-                            " - LLaMA v2: --ctx 4096\n")
-        params.n_ctx = args.ctx
-
-    print(f"params = {params}")
-
-    vocab: Vocab
-    if args.vocab_only:
-        vocab = load_vocab(args.vocab_dir or args.model, args.vocabtype)
-        assert args.outfile, "need --outfile if using --vocab-only"
-        outfile = args.outfile
-        OutputFile.write_vocab_only(outfile, params, vocab)
-        print(f"Wrote {outfile}")
-    else:
-        if args.dump:
-            do_dump_model(model_plus)
-            return
-
-        if model_plus.vocab is not None and args.vocab_dir is None:
-            vocab = model_plus.vocab
-        else:
-            vocab_dir = args.vocab_dir if args.vocab_dir else model_plus.paths[0].parent
-            vocab = load_vocab(vocab_dir, args.vocabtype)
-
-        model       = model_plus.model
-        model       = convert_model_names(model, params)
-        output_type = pick_output_type(model, args.outtype)
-        model       = convert_to_output_type(model, output_type)
-        outfile     = args.outfile or default_outfile(model_plus.paths, output_type)
-
-        OutputFile.write_all(outfile, params, model, vocab)
-        print(f"Wrote {outfile}")
-
-
-if __name__ == '__main__':
-    main()

From c818c405e025a766c0ab12afb18c4baec7810a6d Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 04:42:09 +0200
Subject: [PATCH 229/242] convert-llama-hf-to-gguf.py : fix attn_q permute

---
 convert-llama-hf-to-gguf.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index 9289efd28cf54..6ee5fdf76d434 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -264,7 +264,9 @@ def count_model_parts(dir_model: str) -> int:
         data = data.squeeze().numpy()
 
         # reverse permute these
-        if name.endswith(".q_proj.weight") or name.endswith(".k_proj.weight"):
+        if name.endswith(".q_proj.weight"):
+            data = reverse_hf_permute(data, head_count)
+        if name.endswith(".k_proj.weight"):
             data = reverse_hf_permute(data, head_count, head_count_kv)
 
         # map tensor names

From 6a69a693cbd9695faec0c449f237686f3bdfa281 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 13:23:10 +0200
Subject: [PATCH 230/242] gguf.py : fix rope scale kv

---
 gguf.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/gguf.py b/gguf.py
index e5eb85ded3f59..d461b8d40babd 100644
--- a/gguf.py
+++ b/gguf.py
@@ -45,7 +45,7 @@
 
 # RoPE
 KEY_ROPE_DIMENSION_COUNT = "{arch}.rope.dimension_count"
-KEY_ROPE_SCALE           = "{arch}.rope.scale"
+KEY_ROPE_SCALE_LINEAR    = "{arch}.rope.scale_linear"
 
 # tokenization
 KEY_TOKENIZER_MODEL      = "tokenizer.ggml.model"
@@ -620,8 +620,8 @@ def add_rope_dimension_count(self, count: int):
         self.add_uint32(
             KEY_ROPE_DIMENSION_COUNT.format(arch=self.arch), count)
 
-    def add_rope_scale(self, value:  float):
-        self.add_float32(KEY_ROPE_SCALE.format(arch=self.arch), value)
+    def add_rope_scale_linear(self, value:  float):
+        self.add_float32(KEY_ROPE_SCALE_LINEAR.format(arch=self.arch), value)
 
     def add_tokenizer_model(self, model: str):
         self.add_string(KEY_TOKENIZER_MODEL, model)

From 5f6ff387ca90c5d6f690030d19751e25e0bbc024 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 13:25:14 +0200
Subject: [PATCH 231/242] convert-llama-hf-to-gguf.py : rope scale and added
 tokens

---
 convert-llama-hf-to-gguf.py | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index 6ee5fdf76d434..3a7d4c6c861a3 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -126,6 +126,11 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_head_count_kv(head_count_kv)
 gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
+if "rope_scaling" in hparams and "factor" in hparams["rope_scaling"]:
+    if "type" in hparams["rope_scaling"]:
+        if hparams["rope_scaling"]["type"] == "linear":
+            gguf_writer.add_rope_scale_linear(hparams["rope_scaling"]["factor"])
+
 
 # TOKENIZATION
 
@@ -155,9 +160,7 @@ def count_model_parts(dir_model: str) -> int:
         if tokenizer.is_control(i):
             toktype = 3
 
-        # TODO: How to determinate if a token is user defined?
-        # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
-        # if tokenizer.is_user_defined(i): toktype = 4
+        # toktype = 4 is user-defined = tokens from added_tokens.json
 
         if tokenizer.is_unused(i):
             toktype = 5
@@ -168,6 +171,18 @@ def count_model_parts(dir_model: str) -> int:
         scores.append(score)
         toktypes.append(toktype)
 
+    if Path(dir_model + "/added_tokens.json").is_file():
+        with open(dir_model + "/added_tokens.json", "r", encoding="utf-8") as f:
+            addtokens_json = json.load(f)
+
+            print("gguf: get added tokens")
+
+            for key in addtokens_json:
+                tokens.append( key.encode("utf-8") ) 
+                scores.append(-1000.0)
+                toktypes.append(4) # user-defined token type
+
+
     gguf_writer.add_tokenizer_model("llama")
     gguf_writer.add_token_list(tokens)
     gguf_writer.add_token_scores(scores)

From dc1f0510134ba743d5a87bf8c62c23023e1f44f7 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 13:27:53 +0200
Subject: [PATCH 232/242] convert-llama-7b-pth-to-gguf.py : rope scale and
 added tokens

---
 convert-llama-7b-pth-to-gguf.py | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 77edd026c6243..9e2f2099e66ac 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -118,6 +118,11 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_head_count_kv(head_count_kv)
 gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
+if "rope_scaling" in hparams and "factor" in hparams["rope_scaling"]:
+    if "type" in hparams["rope_scaling"]:
+        if hparams["rope_scaling"]["type"] == "linear":
+            gguf_writer.add_rope_scale_linear(hparams["rope_scaling"]["factor"])
+
 
 # TOKENIZATION
 
@@ -147,9 +152,7 @@ def count_model_parts(dir_model: str) -> int:
         if tokenizer.is_control(i):
             toktype = 3
 
-        # TODO: How to determinate if a token is user defined?
-        # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
-        # if tokenizer.is_user_defined(i): toktype = 4
+        # toktype = 4 is user-defined = tokens from added_tokens.json
 
         if tokenizer.is_unused(i):
             toktype = 5
@@ -160,6 +163,17 @@ def count_model_parts(dir_model: str) -> int:
         scores.append(score)
         toktypes.append(toktype)
 
+    if Path(dir_model + "/added_tokens.json").is_file():
+        with open(dir_model + "/added_tokens.json", "r", encoding="utf-8") as f:
+            addtokens_json = json.load(f)
+
+            print("gguf: get added tokens")
+
+            for key in addtokens_json:
+                tokens.append( key.encode("utf-8") ) 
+                scores.append(-1000.0)
+                toktypes.append(4) # user-defined token type
+
     gguf_writer.add_tokenizer_model("llama")
     gguf_writer.add_token_list(tokens)
     gguf_writer.add_token_scores(scores)

From c082b9fa0b6b55ed4be0c1f90461f44e266542dc Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 13:30:03 +0200
Subject: [PATCH 233/242] llama.cpp : use rope scale kv

---
 llama.cpp | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/llama.cpp b/llama.cpp
index 86c943fc6e3c5..ec954b84fe180 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1435,6 +1435,14 @@ static void llama_model_load_internal(
         hparams.n_head_kv = hparams.n_head;
         GGUF_GET(hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, "llama.attention.head_count_kv");
 
+        // TODO: manually setting rope scale should override this
+        // rope_freq_scale (inverse of the kv) is optional
+        float ropescale = 1.0f;
+        GGUF_GET(ropescale, gguf_get_val_f32, GGUF_TYPE_FLOAT32, false, "llama.rope.scale_linear");
+        if (ropescale != 1.0f) {
+            rope_freq_scale = 1.0f/ropescale;
+        }
+
         // get general kv
         GGUF_GET(general_name, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.name");
         GGUF_GET(general_arch, gguf_get_val_str, GGUF_TYPE_STRING, false, "general.architecture");

From 9070e330abc410e21d3296d7ca1a4d5f8facc8f4 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 14:11:22 +0200
Subject: [PATCH 234/242] convert-llama-7b-pth-to-gguf.py : rope scale fix

---
 convert-llama-7b-pth-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 9e2f2099e66ac..3a606b55df916 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -118,7 +118,7 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_head_count_kv(head_count_kv)
 gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
-if "rope_scaling" in hparams and "factor" in hparams["rope_scaling"]:
+if "rope_scaling" in hparams and hparams["rope_scaling"] != None and "factor" in hparams["rope_scaling"]:
     if "type" in hparams["rope_scaling"]:
         if hparams["rope_scaling"]["type"] == "linear":
             gguf_writer.add_rope_scale_linear(hparams["rope_scaling"]["factor"])

From 7a7d1ba68ab511269f5e686b7d4d2b7c6b9a3e45 Mon Sep 17 00:00:00 2001
From: klosax <131523366+klosax@users.noreply.github.com>
Date: Mon, 21 Aug 2023 14:12:02 +0200
Subject: [PATCH 235/242] convert-llama-hf-to-gguf.py : rope scale fix

---
 convert-llama-hf-to-gguf.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index 3a7d4c6c861a3..d5b3897c7e8ea 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -126,7 +126,7 @@ def count_model_parts(dir_model: str) -> int:
 gguf_writer.add_head_count_kv(head_count_kv)
 gguf_writer.add_layer_norm_rms_eps(hparams["rms_norm_eps"])
 
-if "rope_scaling" in hparams and "factor" in hparams["rope_scaling"]:
+if "rope_scaling" in hparams and hparams["rope_scaling"] != None and "factor" in hparams["rope_scaling"]:
     if "type" in hparams["rope_scaling"]:
         if hparams["rope_scaling"]["type"] == "linear":
             gguf_writer.add_rope_scale_linear(hparams["rope_scaling"]["factor"])

From 6490ff7198ced6fed47865d90eed866817cad7da Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 21 Aug 2023 16:42:27 +0300
Subject: [PATCH 236/242] py : fix whitespace

---
 convert-falcon-hf-to-gguf.py    | 2 +-
 convert-llama-7b-pth-to-gguf.py | 2 +-
 convert-llama-hf-to-gguf.py     | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py
index e0da3a04de811..b3e190a0fd83c 100644
--- a/convert-falcon-hf-to-gguf.py
+++ b/convert-falcon-hf-to-gguf.py
@@ -250,7 +250,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:
diff --git a/convert-llama-7b-pth-to-gguf.py b/convert-llama-7b-pth-to-gguf.py
index 3a606b55df916..ab5c80b69ede2 100644
--- a/convert-llama-7b-pth-to-gguf.py
+++ b/convert-llama-7b-pth-to-gguf.py
@@ -170,7 +170,7 @@ def count_model_parts(dir_model: str) -> int:
             print("gguf: get added tokens")
 
             for key in addtokens_json:
-                tokens.append( key.encode("utf-8") ) 
+                tokens.append( key.encode("utf-8") )
                 scores.append(-1000.0)
                 toktypes.append(4) # user-defined token type
 
diff --git a/convert-llama-hf-to-gguf.py b/convert-llama-hf-to-gguf.py
index d5b3897c7e8ea..f8cfdaa800c85 100644
--- a/convert-llama-hf-to-gguf.py
+++ b/convert-llama-hf-to-gguf.py
@@ -178,7 +178,7 @@ def count_model_parts(dir_model: str) -> int:
             print("gguf: get added tokens")
 
             for key in addtokens_json:
-                tokens.append( key.encode("utf-8") ) 
+                tokens.append( key.encode("utf-8") )
                 scores.append(-1000.0)
                 toktypes.append(4) # user-defined token type
 
@@ -294,7 +294,7 @@ def count_model_parts(dir_model: str) -> int:
             sys.exit()
 
         n_dims = len(data.shape)
-        data_dtype = data.dtype 
+        data_dtype = data.dtype
 
         # if f32 desired, convert any float16 to float32
         if ftype == 0 and data_dtype == np.float16:

From e06cbcee73dfd32ea2c159a79da62b3a5afc7640 Mon Sep 17 00:00:00 2001
From: Kerfuffle <44031344+KerfuffleV2@users.noreply.github.com>
Date: Mon, 21 Aug 2023 08:45:52 -0600
Subject: [PATCH 237/242] gguf : add Python script to convert GGMLv3 LLaMA
 models to GGUF (#2682)

* First pass at converting GGMLv3 LLaMA models to GGUF

* Cleanups, better output during conversion

* Fix vocab space conversion logic

* More vocab conversion fixes

* Add description to converted GGUF files

* Improve help text, expand warning

* Allow specifying name and description for output GGUF

* Allow overriding vocab and hyperparams from original model metadata

* Use correct params override var name

* Fix wrong type size for Q8_K

Better handling of original style metadata

* Set default value for gguf add_tensor raw_shape KW arg
---
 convert-llama-ggmlv3-to-gguf.py | 334 ++++++++++++++++++++++++++++++++
 gguf.py                         |  52 +++--
 2 files changed, 374 insertions(+), 12 deletions(-)
 create mode 100644 convert-llama-ggmlv3-to-gguf.py

diff --git a/convert-llama-ggmlv3-to-gguf.py b/convert-llama-ggmlv3-to-gguf.py
new file mode 100644
index 0000000000000..30038072f4bd5
--- /dev/null
+++ b/convert-llama-ggmlv3-to-gguf.py
@@ -0,0 +1,334 @@
+import sys, struct, math, argparse
+from pathlib import Path
+
+import numpy as np
+
+import gguf
+
+# Note: Does not support GGML_QKK_64
+QK_K = 256
+# Items here are (block size, type size)
+GGML_QUANT_SIZES = {
+    gguf.GGMLQuantizationType.F32  : (1, 4),
+    gguf.GGMLQuantizationType.F16  : (1, 2),
+    gguf.GGMLQuantizationType.Q4_0 : (32, 2 + 16),
+    gguf.GGMLQuantizationType.Q4_1 : (32, 2 + 2 + 16),
+    gguf.GGMLQuantizationType.Q5_0 : (32, 2 + 4 + 16),
+    gguf.GGMLQuantizationType.Q5_1 : (32, 2 + 2 + 4 + 16),
+    gguf.GGMLQuantizationType.Q8_0 : (32, 2 + 32),
+    gguf.GGMLQuantizationType.Q8_1 : (32, 4 + 4 + 32),
+    gguf.GGMLQuantizationType.Q2_K : (256, 2 + 2 + QK_K // 16 + QK_K // 4),
+    gguf.GGMLQuantizationType.Q3_K : (256, 2 + QK_K // 4 + QK_K // 8 + 12),
+    gguf.GGMLQuantizationType.Q4_K : (256, 2 + 2 + QK_K // 2 + 12),
+    gguf.GGMLQuantizationType.Q5_K : (256, 2 + 2 + QK_K // 2 + QK_K // 8 + 12),
+    gguf.GGMLQuantizationType.Q6_K : (256, 2 + QK_K // 2 + QK_K // 4 + QK_K // 16),
+    gguf.GGMLQuantizationType.Q8_K : (256, 4 + QK_K + QK_K // 8),
+}
+
+class Hyperparameters:
+    def __init__(self):
+        self.n_vocab = self.n_embd = self.n_mult = self.n_head = self.n_layer = self.n_rot = self.ftype = 0
+        self.n_ff = 0
+
+    def set_n_ff(self, model):
+        ff_tensor_idx = model.tensor_map.get(b'layers.0.feed_forward.w1.weight')
+        assert ff_tensor_idx is not None, 'Missing layer 0 FF tensor'
+        ff_tensor = model.tensors[ff_tensor_idx]
+        self.n_ff = ff_tensor.dims[1]
+
+    def load(self, data, offset):
+        (
+            self.n_vocab,
+            self.n_embd,
+            self.n_mult,
+            self.n_head,
+            self.n_layer,
+            self.n_rot,
+            self.ftype,
+        ) = struct.unpack('<7I', data[offset:offset + (4 * 7)])
+        return 4 * 7
+
+    def __str__(self):
+        return f'<Hyperparameters: n_vocab={self.n_vocab}, n_embd={self.n_embd}, n_mult={self.n_mult}, n_head={self.n_head}, n_layer={self.n_layer}, n_rot={self.n_rot}, n_ff={self.n_ff}, ftype={self.ftype}>'
+
+class Vocab:
+    def __init__(self):
+        self.items = []
+
+    def load(self, data, offset, n_vocab):
+        orig_offset = offset
+        for _ in range(n_vocab):
+            itemlen = struct.unpack('<I', data[offset:offset + 4])[0]
+            assert itemlen < 4096, 'Absurd vocab item length'
+            offset += 4
+            vocab = bytes(data[offset:offset + itemlen])
+            offset += itemlen
+            score = struct.unpack('<f', data[offset:offset + 4])[0]
+            offset += 4
+            self.items.append((vocab, score))
+        return offset - orig_offset
+
+class Tensor:
+    def __init__(self):
+        self.name = None
+        self.dims = ()
+        self.dtype = None
+        self.start_offset = 0
+        self.len_bytes = 0
+
+    def load(self, data, offset):
+        orig_offset = offset
+        (n_dims, name_len, dtype) = struct.unpack('<3I', data[offset:offset + 12])
+        assert n_dims >= 0 and n_dims <= 4, f'Invalid tensor dimensions {n_dims}'
+        assert name_len < 4096, 'Absurd tensor name length'
+        quant = GGML_QUANT_SIZES.get(dtype)
+        assert quant is not None, 'Unknown tensor type'
+        (blksize, tysize) = quant
+        offset += 12
+        self.dtype= dtype
+        self.dims = struct.unpack(f'<{n_dims}I', data[offset:offset + (4 * n_dims)])
+        offset += 4 * n_dims
+        self.name = bytes(data[offset:offset + name_len])
+        offset += name_len
+        pad = ((offset + 31) & ~31) - offset
+        offset += pad
+        n_elems = np.prod(self.dims)
+        n_bytes = (n_elems * tysize) // blksize
+        self.start_offset = offset
+        self.len_bytes = n_bytes
+        offset += n_bytes
+        # print(n_dims, name_len, dtype, self.dims, self.name, pad)
+        return offset - orig_offset
+
+class GGMLV3Model:
+    def __init__(self):
+        self.hyperparameters = None
+        self.vocab = None
+        self.tensor_map = {}
+        self.tensors = []
+
+    def validate_header(self, data, offset):
+        if bytes(data[offset:offset + 4]) != b'tjgg' or struct.unpack('<I', data[offset + 4:offset + 8])[0] != 3:
+            raise ValueError('Only GGJTv3 supported')
+        return 8
+
+    def load(self, data, offset):
+        offset += self.validate_header(data, offset)
+        hp = Hyperparameters()
+        offset += hp.load(data, offset)
+        vocab = Vocab()
+        offset += vocab.load(data, offset, hp.n_vocab)
+        tensors = []
+        tensor_map = {}
+        while offset < len(data):
+            tensor = Tensor()
+            offset += tensor.load(data, offset)
+            tensor_map[tensor.name] = len(tensors)
+            tensors.append(tensor)
+        self.hyperparameters = hp
+        self.vocab = vocab
+        self.tensors = tensors
+        self.tensor_map = tensor_map
+        hp.set_n_ff(self)
+        return offset
+
+class GGMLToGGUF:
+    def __init__(self, ggml_model, data, cfg, params_override = None, vocab_override = None):
+        hp = ggml_model.hyperparameters
+        self.model = ggml_model
+        self.data = data
+        self.cfg = cfg
+        self.params_override = params_override
+        self.vocab_override = vocab_override
+        if params_override is not None:
+            n_kv_head = params_override.n_head_kv
+        else:
+            if cfg.gqa == 1:
+                n_kv_head = hp.n_head
+            else:
+                gqa = float(cfg.gqa)
+                n_kv_head = None
+                for x in range(1, 256):
+                    if float(hp.n_head) / float(x) == gqa:
+                        n_kv_head = x
+                assert n_kv_head is not None, "Couldn't determine n_kv_head from GQA param"
+                print(f'- Guessed n_kv_head = {n_kv_head} based on GQA {cfg.gqa}')
+        self.n_kv_head = n_kv_head
+        self.name_map = gguf.get_tensor_name_map(gguf.MODEL_ARCH.LLAMA, ggml_model.hyperparameters.n_layer)
+
+    def save(self):
+        print('* Preparing to save GGUF file')
+        gguf_writer = gguf.GGUFWriter(self.cfg.output, gguf.MODEL_ARCH_NAMES[gguf.MODEL_ARCH.LLAMA], use_temp_file = False)
+        self.add_params(gguf_writer)
+        self.add_vocab(gguf_writer)
+        self.add_tensors(gguf_writer)
+        print("    gguf: write header")
+        gguf_writer.write_header_to_file()
+        print("    gguf: write metadata")
+        gguf_writer.write_kv_data_to_file()
+        print("    gguf: write tensors")
+        gguf_writer.write_tensors_to_file()
+        gguf_writer.close()
+
+    def add_params(self, gguf_writer):
+        hp = self.model.hyperparameters
+        cfg = self.cfg
+        desc = cfg.desc if cfg.desc is not None else 'converted from legacy GGJTv3 format'
+        try:
+            # Filenames aren't necessarily valid UTF8.
+            name = cfg.name if cfg.name is not None else cfg.input.name
+        except UnicodeDecodeError:
+            name = None
+        print('* Adding model parameters and KV items')
+        if name is not None:
+            gguf_writer.add_name(name)
+        gguf_writer.add_description(desc)
+        if self.params_override is not None:
+            po = self.params_override
+            assert po.n_embd == hp.n_embd, 'Model hyperparams mismatch'
+            assert po.n_layer == hp.n_layer, 'Model hyperparams mismatch'
+            assert po.n_head == hp.n_head, 'Model hyperparams mismatch'
+            gguf_writer.add_context_length      (po.n_ctx)
+            gguf_writer.add_embedding_length    (po.n_embd)
+            gguf_writer.add_block_count         (po.n_layer)
+            gguf_writer.add_feed_forward_length (po.n_ff)
+            gguf_writer.add_rope_dimension_count(po.n_embd // po.n_head)
+            gguf_writer.add_head_count          (po.n_head)
+            gguf_writer.add_head_count_kv       (po.n_head_kv)
+            gguf_writer.add_layer_norm_rms_eps  (po.f_norm_eps)
+            return
+        gguf_writer.add_context_length(cfg.context_length)
+        gguf_writer.add_embedding_length(hp.n_embd)
+        gguf_writer.add_block_count(hp.n_layer)
+        gguf_writer.add_feed_forward_length(hp.n_ff)
+        gguf_writer.add_rope_dimension_count(hp.n_embd // hp.n_head)
+        gguf_writer.add_head_count(hp.n_head)
+        gguf_writer.add_head_count_kv(self.n_kv_head)
+        gguf_writer.add_layer_norm_rms_eps(float(cfg.eps))
+
+    def add_vocab(self, gguf_writer):
+        hp = self.model.hyperparameters
+        gguf_writer.add_tokenizer_model('llama')
+        tokens = []
+        scores = []
+        toktypes = []
+        if self.vocab_override is not None:
+            vo = self.vocab_override
+            print('* Adding vocab item(s)')
+            for (idx, vitem) in enumerate(vo.all_tokens()):
+                if len(vitem) == 3:
+                    tokens.append(vitem[0])
+                    scores.append(vitem[1])
+                    toktypes.append(vitem[2])
+                else:
+                    # Maybe try to guess the token type here?
+                    tokens.append(vitem[0])
+                    scores.append(vitem[1])
+            assert len(tokens) == hp.n_vocab, f'Override vocab has a different number of items than hyperparameters - override = {len(tokens)} but n_vocab={hp.n_vocab}'
+            gguf_writer.add_token_list(tokens)
+            gguf_writer.add_token_scores(scores)
+            if len(toktypes) > 0:
+                gguf_writer.add_token_types(toktypes)
+            return
+        print(f'* Adding {hp.n_vocab} vocab item(s)')
+        for (tokid, (vbytes, vscore)) in enumerate(self.model.vocab.items):
+            tt = 1 # Normal
+            if len(vbytes) == 0:
+                tt = 3 # Control
+            elif tokid >= 3 and tokid <= 258 and len(vbytes) == 1:
+                hv = hex(vbytes[0])[2:].upper()
+                vbytes = bytes(f'<0x{hv}>', encoding = 'UTF-8')
+                tt = 6 # Byte
+            else:
+                vbytes = vbytes.replace(b' ', b'\xe2\x96\x81')
+            toktypes.append(tt)
+            tokens.append(vbytes)
+            scores.append(vscore)
+        gguf_writer.add_token_list(tokens)
+        gguf_writer.add_token_scores(scores)
+        gguf_writer.add_token_types(toktypes)
+
+    def add_tensors(self, gguf_writer):
+        nm = self.name_map
+        data = self.data
+        print(f'* Adding {len(self.model.tensors)} tensor(s)')
+        for tensor in self.model.tensors:
+            name = str(tensor.name, 'UTF-8')
+            if name.endswith('.weight'):
+                name = name[:-7]
+                suffix = '.weight'
+            elif name.endswith('.bias'):
+                name = name[:-5]
+                suffix = '.bias'
+            mapped_name = nm.get(name)
+            assert mapped_name is not None, f'Bad name {name}'
+            mapped_name += suffix
+            tempdims = list(tensor.dims[:])
+            if len(tempdims) > 1:
+                temp = tempdims[1]
+                tempdims[1] = tempdims[0]
+                tempdims[0] = temp
+            # print(f'+ {tensor.name} | {mapped_name} {tensor.dims} :: {tempdims}')
+            gguf_writer.add_tensor(mapped_name, data[tensor.start_offset:tensor.start_offset + tensor.len_bytes], raw_shape = tempdims, raw_dtype = tensor.dtype)
+
+def handle_metadata(cfg, hp):
+    import convert
+    assert cfg.model_metadata_dir.is_dir(), 'Metadata dir is not a directory'
+    hf_config_path   = cfg.model_metadata_dir / "config.json"
+    orig_config_path = cfg.model_metadata_dir / "params.json"
+    # We pass a fake model here. "original" mode will check the shapes of some
+    # tensors if information is missing in the .json file: other than that, the
+    # model data isn't used so this should be safe (at least for now).
+    fakemodel = {
+        'tok_embeddings.weight': convert.LazyTensor.__new__(convert.LazyTensor),
+        'layers.0.feed_forward.w1.weight': convert.LazyTensor.__new__(convert.LazyTensor),
+    }
+    fakemodel['tok_embeddings.weight'].shape = [hp.n_vocab]
+    fakemodel['layers.0.feed_forward.w1.weight'].shape = [hp.n_ff]
+    if hf_config_path.exists():
+        params = convert.Params.loadHFTransformerJson(fakemodel, hf_config_path)
+    elif orig_config_path.exists():
+        params = convert.Params.loadOriginalParamsJson(fakemodel, orig_config_path)
+    else:
+        raise ValueError('Unable to load metadata')
+    vocab = convert.load_vocab(cfg.vocab_dir if cfg.vocab_dir is not None else cfg.model_metadata_dir, cfg.vocabtype)
+    convert.check_vocab_size(params, vocab)
+    return (params, vocab)
+
+def handle_args():
+    parser = argparse.ArgumentParser(description = 'Convert GGMLv3 models to GGUF')
+    parser.add_argument('--input', '-i', type = Path, help = 'Input GGMLv3 filename')
+    parser.add_argument('--output', '-o', type = Path, help ='Output GGUF filename')
+    parser.add_argument('--name', help = 'Set model name')
+    parser.add_argument('--desc', help = 'Set model description')
+    parser.add_argument('--gqa', type = int, default = 1, help = 'grouped-query attention factor (use 8 for LLaMA2 70B)')
+    parser.add_argument('--eps', default = '5.0e-06', help = 'RMS norm eps: Use 1e-6 for LLaMA1 and OpenLLaMA, use 1e-5 for LLaMA2')
+    parser.add_argument('--context-length', '-c', type=int, default = 2048, help = 'Default max context length: LLaMA1 is typically 2048, LLaMA2 is typically 4096')
+    parser.add_argument('--model-metadata-dir', '-m', type = Path, help ='Load HuggingFace/.pth vocab and metadata from the specified directory')
+    parser.add_argument("--vocab-dir", type=Path, help="directory containing tokenizer.model, if separate from model file - only meaningful with --model-metadata-dir")
+    parser.add_argument("--vocabtype", choices=["spm", "bpe"], help="vocab format - only meaningful with --model-metadata-dir and/or --vocab-dir (default: spm)", default="spm")
+    return parser.parse_args()
+
+def main():
+    cfg = handle_args()
+    print(f'* Using config: {cfg}')
+    print('\n=== WARNING === Be aware that this conversion script is best-effort. Use a native GGUF model if possible. === WARNING ===\n')
+    data = np.memmap(cfg.input, mode = 'r')
+    model = GGMLV3Model()
+    print('* Scanning GGML input file')
+    offset = model.load(data, 0)
+    print(f'* GGML model hyperparameters: {model.hyperparameters}')
+    vocab_override = None
+    params_override = None
+    if cfg.model_metadata_dir is not None:
+        (params_override, vocab_override) = handle_metadata(cfg, model.hyperparameters)
+        print('!! Note: When overriding params the --gqa, --eps and --context-length options are ignored.')
+        print(f'* Overriding params: {params_override}')
+        print(f'* Overriding vocab: {vocab_override}')
+    else:
+        print('\n=== WARNING === Special tokens may not be converted correctly. Use --model-metadata-dir if possible === WARNING ===\n')
+    converter = GGMLToGGUF(model, data, cfg, params_override = params_override, vocab_override = vocab_override)
+    converter.save()
+    print(f'* Successful completion. Output saved to: {cfg.output}')
+
+main()
diff --git a/gguf.py b/gguf.py
index d461b8d40babd..60ee52f095b6c 100644
--- a/gguf.py
+++ b/gguf.py
@@ -5,7 +5,7 @@
 import numpy as np
 
 from enum import IntEnum, auto
-from typing import Any, IO, List
+from typing import Any, IO, List, Optional
 
 #
 # constants
@@ -325,8 +325,20 @@ def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict:
 
 
 class GGMLQuantizationType(IntEnum):
-    F32 = 0
-    F16 = 1
+    F32  = 0
+    F16  = 1
+    Q4_0 = 2
+    Q4_1 = 3
+    Q5_0 = 6
+    Q5_1 = 7
+    Q8_0 = 8
+    Q8_1 = 9
+    Q2_K = 10
+    Q3_K = 11
+    Q4_K = 12
+    Q5_K = 13
+    Q6_K = 14
+    Q8_K = 15
 
 
 class GGUFValueType(IntEnum):
@@ -359,7 +371,7 @@ def get_type(val):
 
 
 class GGUFWriter:
-    def __init__(self, path: str, arch: str):
+    def __init__(self, path: str, arch: str, use_temp_file = True):
         self.fout = open(path, "wb")
         self.arch = arch
         self.offset_tensor = 0
@@ -369,6 +381,8 @@ def __init__(self, path: str, arch: str):
         self.ti_data = b""
         self.ti_data_count = 0
         self.add_architecture()
+        self.use_temp_file = use_temp_file
+        self.tensors = []
 
     def write_header_to_file(self):
         self.fout.write(struct.pack("<I", GGUF_MAGIC))
@@ -476,8 +490,8 @@ def add_val(self: str, val: Any, vtype: GGUFValueType = None, add_vtype: bool =
     def ggml_pad(x: int, n: int) -> int:
         return ((x + n - 1) // n) * n
 
-    def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int):
-        assert tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
+    def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int, raw_dtype: Optional[GGMLQuantizationType] = None):
+        assert raw_dtype is not None or tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now"
 
         encoded_name = name.encode("utf8")
         self.ti_data += struct.pack("<I", len(encoded_name))
@@ -486,23 +500,30 @@ def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.
         self.ti_data += struct.pack("<I", n_dims)
         for i in range(n_dims):
             self.ti_data += struct.pack("<I", tensor_shape[n_dims - 1 - i])
-
-        dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
+        if raw_dtype is None:
+            dtype = GGMLQuantizationType.F32 if tensor_dtype == np.float32 else GGMLQuantizationType.F16
+        else:
+            dtype = raw_dtype
         self.ti_data += struct.pack("<I", dtype)
         self.ti_data += struct.pack("<Q", self.offset_tensor)
         self.offset_tensor += GGUFWriter.ggml_pad(tensor_nbytes, self.data_alignment)
         self.ti_data_count += 1
 
-    def add_tensor(self, name: str, tensor: np.ndarray):
-        if not hasattr(self, "temp_file"):
+    def add_tensor(self, name: str, tensor: np.ndarray, raw_shape: Optional[np.ndarray] = None, raw_dtype: Optional[GGMLQuantizationType] = None):
+        if self.use_temp_file and not hasattr(self, "temp_file"):
             self.temp_file = tempfile.SpooledTemporaryFile(mode="w+b", max_size=256*1024*1024)
             self.temp_file.seek(0)
 
-        self.add_tensor_info(name, tensor.shape, tensor.dtype, tensor.nbytes)
+        self.add_tensor_info(name, raw_shape if raw_shape is not None else tensor.shape, tensor.dtype, tensor.nbytes, raw_dtype = raw_dtype)
+
+        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
+
+        if not self.use_temp_file:
+            self.tensors.append((tensor, pad))
+            return
 
         tensor.tofile(self.temp_file)
 
-        pad = GGUFWriter.ggml_pad(tensor.nbytes, self.data_alignment) - tensor.nbytes
         if pad != 0:
             self.temp_file.write(bytes([0] * pad))
 
@@ -524,6 +545,13 @@ def write_tensors_to_file(self):
         if pad != 0:
             self.fout.write(bytes([0] * pad))
 
+        if not self.use_temp_file:
+            for (currtensor, currpad) in self.tensors:
+                currtensor.tofile(self.fout)
+                if currpad != 0:
+                    self.fout.write(bytes([0] * currpad))
+            return
+
         self.temp_file.seek(0)
 
         shutil.copyfileobj(self.temp_file, self.fout)

From 8d177eddeb8c0b284a47a6840ca820ed846708c1 Mon Sep 17 00:00:00 2001
From: goerch <jhr.walter@t-online.de>
Date: Mon, 21 Aug 2023 17:56:02 +0200
Subject: [PATCH 238/242] llama : improve token type support (#2668)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)

* Improved tokenizer test

But does it work on MacOS?

* Improve token type support

- Added @klosax code to convert.py
- Improved token type support in vocabulary

* Exclude platform dependent tests

* More sentencepiece compatibility by eliminating magic numbers

* Restored accidentally removed comment
---
 convert.py                   |  30 ++++++--
 llama.cpp                    | 128 ++++++++++++++---------------------
 models/ggml-vocab-llama.gguf | Bin 467382 -> 595423 bytes
 tests/test-tokenizer-1.cpp   |  34 ++++++----
 4 files changed, 94 insertions(+), 98 deletions(-)

diff --git a/convert.py b/convert.py
index df589928baf5f..f680f85963b0e 100755
--- a/convert.py
+++ b/convert.py
@@ -261,12 +261,12 @@ def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for i, item in enumerate(tokenizer):
             text: bytes = item.encode("utf-8")
             score: float = -i
-            yield text, score
+            yield text, score, 4
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
             score = -1000.0
-            yield text.encode("utf-8"), score
+            yield text.encode("utf-8"), score, 4
 
     def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
         yield from self.bpe_tokens()
@@ -303,12 +303,28 @@ def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
             piece = tokenizer.id_to_piece(i)
             text: bytes = piece.encode("utf-8")
             score: float = tokenizer.get_score(i)
-            yield text, score
+
+            toktype = 1  # defualt to normal token type
+            if tokenizer.is_unknown(i):
+                toktype = 2
+            if tokenizer.is_control(i):
+                toktype = 3
+
+            # NOTE: I think added_tokens are user defined.
+            # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
+            # if tokenizer.is_user_defined(i): toktype = 4
+
+            if tokenizer.is_unused(i):
+                toktype = 5
+            if tokenizer.is_byte(i):
+                toktype = 6
+
+            yield text, score, toktype
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
             score = -1000.0
-            yield text.encode("utf-8"), score
+            yield text.encode("utf-8"), score, 4
 
     def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
         yield from self.sentencepiece_tokens()
@@ -721,14 +737,16 @@ def add_meta_arch(self, params: Params) -> None:
     def add_meta_vocab(self, vocab: Vocab) -> None:
         tokens = []
         scores = []
-        for text, score in vocab.all_tokens():
+        toktypes = []
+        for text, score, toktype in vocab.all_tokens():
             tokens.append(text)
             scores.append(score)
+            toktypes.append(toktype)
 
         self.gguf.add_tokenizer_model("llama")
         self.gguf.add_token_list(tokens)
         self.gguf.add_token_scores(scores)
-        #self.gguf.add_token_types(toktypes) # TODO: add this
+        self.gguf.add_token_types(toktypes)
 
         # TODO: added / special tokens
 
diff --git a/llama.cpp b/llama.cpp
index ec954b84fe180..1785025f0078b 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -772,15 +772,16 @@ struct llama_vocab {
     using id    = int32_t;
     using token = std::string;
 
-    struct token_score {
+    struct token_data {
         token tok;
         float score;
+        int toktype;
     };
 
     llama_vocab_type type = LLAMA_VOCAB_TYPE_SPM;
 
     std::unordered_map<token, id> token_to_id;
-    std::vector<token_score>      id_to_token;
+    std::vector<token_data>       id_to_token;
 
     // default LLaMA special tokens
     id special_bos_id = 1;
@@ -1507,17 +1508,25 @@ static void llama_model_load_internal(
 
         const float * scores = (const float * ) gguf_get_arr_data(ctx, score_idx);
 
+        const int toktype_idx = gguf_find_key(ctx, "tokenizer.ggml.token_type");
+        if (toktype_idx == -1) {
+            throw std::runtime_error("cannot find token type list in GGUF file\n");
+        }
+
+        const int * toktypes = (const int * ) gguf_get_arr_data(ctx, toktype_idx);
+
         for (uint32_t i = 0; i < hparams.n_vocab; i++) {
             std::string word = gguf_get_arr_str(ctx, token_idx, i);
 
             vocab.token_to_id[word] = i;
 
-            auto & tok_score = vocab.id_to_token[i];
-            tok_score.tok = std::move(word);
-            tok_score.score = scores[i];
+            auto & token_data = vocab.id_to_token[i];
+            token_data.tok = std::move(word);
+            token_data.score = scores[i];
+            token_data.toktype = toktypes[i];
 
             // determine the newline token: 0x0A == 10 == '\n'
-            if (tok_score.tok == "<0x0A>") {
+            if (token_data.tok == "<0x0A>") {
                 vocab.linefeed_id = i;
             }
         }
@@ -2345,92 +2354,57 @@ static enum llama_vocab_type llama_vocab_get_type(const llama_vocab & vocab) {
     return vocab.type;
 }
 
-static bool llama_is_normal_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return token >= 259;
-    }
-
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
-        return token >= 95;
-    }
-
-    return false;
+static bool llama_is_normal_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 1;
 }
 
-static bool llama_is_bos_token(const llama_vocab & vocab, llama_token token) {
-    return token == vocab.special_bos_id;
+static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 2;
 }
 
-static bool llama_is_eos_token(const llama_vocab & vocab, llama_token token) {
-    return token == vocab.special_eos_id;
+static bool llama_is_control_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 3;
 }
 
-static bool llama_is_control_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return token == llama_is_bos_token(vocab, token) || token == llama_is_eos_token(vocab, token);
-    }
-
-    // TODO: improve?
-    return false;
+static bool llama_is_bos_token(const llama_vocab & vocab, llama_token id) {
+    GGML_ASSERT(llama_is_control_token(vocab, id));
+    return id == vocab.special_bos_id;
 }
 
-static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return token == 0;
-    }
-
-    // TODO: improve?
-    return false;
+static bool llama_is_eos_token(const llama_vocab & vocab, llama_token id ) {
+    GGML_ASSERT(llama_is_control_token(vocab, id));
+    return id == vocab.special_eos_id;
 }
 
-static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token token) {
-    GGML_UNUSED(vocab);
-    GGML_UNUSED(token);
-    // TODO: improve?
-    return false;
+static bool llama_is_pad_token(const llama_vocab & vocab, llama_token id ) {
+    GGML_ASSERT(id < 0 || llama_is_control_token(vocab, id));
+    return id == vocab.special_pad_id;
 }
 
-static bool llama_is_unused_token(const llama_vocab & vocab, llama_token token) {
-    GGML_UNUSED(vocab);
-    GGML_UNUSED(token);
-    // TODO: improve?
-    return false;
+static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 4;
 }
 
-static bool llama_is_byte_token(const llama_vocab & vocab, llama_token token) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return 3 <= token && token < 259;
-    }
-
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
-        return 1 <= token && token < 95;
-    }
-
-    return false;
+static bool llama_is_unused_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 5;
 }
 
-static uint8_t llama_byte_to_char(const llama_vocab & vocab, uint8_t byte) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return byte - 3;
-    }
-
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
-        return byte + 32;
-    }
-
-    return false;
+static bool llama_is_byte_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].toktype == 6;
 }
 
-static uint8_t llama_char_to_byte(const llama_vocab & vocab, uint8_t ch) {
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_SPM) {
-        return ch + 3;
-    }
-
-    if (llama_vocab_get_type(vocab) == LLAMA_VOCAB_TYPE_BPE) {
-        return ch - 32;
-    }
+static uint8_t llama_token_to_byte(const llama_vocab & vocab, llama_token id) {
+    GGML_ASSERT(llama_is_byte_token(vocab, id));
+    const auto& token_data = vocab.id_to_token.at(id);
+    auto buf = token_data.tok.substr(3, 2);
+    return strtol(buf.c_str(), NULL, 16);
+}
 
-    return false;
+static llama_token llama_byte_to_token(const llama_vocab & vocab, uint8_t ch) {
+    char buf[7];
+    int result = snprintf(buf, sizeof(buf), "<0x%02X>", ch);
+    GGML_ASSERT(0 <= result && result < 7);
+    return vocab.token_to_id.at(buf);
 }
 
 static std::string llama_escape_whitespace(const std::string& text) {
@@ -2569,7 +2543,7 @@ struct llama_tokenizer {
         if (p == rev_merge.end()) {
             // output any symbols that did not form tokens as bytes.
             for (int j = 0; j < (int)symbol.n; ++j) {
-                llama_vocab::id token_id = llama_char_to_byte(vocab_, symbol.text[j]);
+                llama_vocab::id token_id = llama_byte_to_token(vocab_, symbol.text[j]);
                 output.push_back(token_id);
             }
             return;
@@ -2595,12 +2569,12 @@ struct llama_tokenizer {
             return;
         }
 
-        const auto &tok_score = vocab_.id_to_token[(*token).second];
+        const auto &tok_data = vocab_.id_to_token[(*token).second];
 
         llama_sp_bigram bigram;
         bigram.left = left;
         bigram.right = right;
-        bigram.score = tok_score.score;
+        bigram.score = tok_data.score;
         bigram.size = text.size();
         work_queue_.push(bigram);
 
@@ -5109,7 +5083,7 @@ int llama_token_to_str_with_model(const struct llama_model * model, llama_token
             if (length < 1) {
                 return -1;
             }
-            buf[0] = llama_byte_to_char(model->vocab, token);
+            buf[0] = llama_token_to_byte(model->vocab, token);
             return 1;
         }
     }
diff --git a/models/ggml-vocab-llama.gguf b/models/ggml-vocab-llama.gguf
index c50db67dc52023444e607bde7b684da5d631e564..63bfaf672f382c0f5bbcffe54736e2698ef3ac55 100644
GIT binary patch
delta 129117
zcmeI(y=qf&0Egk6endOu0tyZe&fUZ-&|ZM(YL-yMPo#^2oI%jp>u_?>$;k`xR{Td@
zU6R1NJi?)fG^c&jlRW2F{Py$KuP=92mzR$oOpCvNR-ae*r<-3No=qn|?tQ%bVLg3X
zPut^Y9LM808OP~3mfzRYjmNj=^YMAtubypQK7F@&b8&U`;`zm&zdpHs`+9RW{_%AC
z!0#7}qw(SR-?1DY4|B|idY~O2XPo~)-^p%|Gan}7ZYLk-p40DsInLjH0iOSTegL<x
zI{O`?|9%g!=jU&q|MCl%%af0D&*^tRZLj~dfagEW@A+@@yZ*iYrv<$Jr}<s~ZT<!S
z@n2bh|Jxf7*MIrbznI+rwKpLA*8;@rzt{h=fa~A&Ul#EG@BP0l!2kST7T~{9KjZhW
z<%`cax98vUUl!p1QopW${%_xaT>q~BvViN~^<Ng?|MKqF@BiHYm3Kex|M{=fkNZFV
zEA`|5fAAmwl?C{Z|H=aV$A4u3{^P&00RNTxvH$U3sUQ0v|CRc&|M6d`ANwEw@n2bh
z|M;&gz<>N#7T`brD+};nsUQ0v|CRc&|M6d`ANwEwmHM&&@gM({1^AEu$^!hye`Nvw
z<G->1|CRc&|M6d`ANwEwmHM&&@n5MQ`yc=DUs-_v_^&L$fBaV#;6MH=3-Dj5ANwEw
zmHM&&@n5MQ`yc<6`mz7<AODpF_>cd}0{q8+WdZ)<zp?=TmHM&&@n5MQ`yc<6`mz7<
zU#TDaAOG=RS%CleuPnfS{8tv>KmIEV@L#DP`yc<6`mz7<U#TDaAODs5vH$TO|CI&!
zkN?U7{KtP~0siB^vH<^;`mz7<U#TDaAODs5vH$U3sUQ0v|M6d0fdBZfEWm&KR~Fzu
z{woXcU#TDaAODs5vH$U3sUQ0v|CRc&|M4IHl?C{Z|H=aV$A4u3{^P&00RNTxvH$U3
zsUQ0v|CRc&|M6d`ANwEw@n2bh|M;&gz<>N#7T`brD+};nsUQ0v|CRc&|M6d`ANwEw
zmHM&&@gM({1^AEu$^!hye`Nvw<G->1|CRc&|M6d`ANwEwmHM&&@n5MQ`yc=DUs-_v
z_^&L$fBaV#;6MH=3-Dj5ANwEwmHM&&@n5MQ`yc<6`mz7<AODpF_>cd}0{q8+WdZ)<
zzp?=TmHM&&@n5MQ`yc<6`mz7<U#TDaAOG=RS%CleuPnfS{8tv>KmIEV@L#DP`yc<6
z`mz7<U#TDaAODs5vH$TO|CI&!kN?U7{KtP~0siB^vH<^;`mz7<U#TDaAODs5vH$U3
zsUQ0v|M6d0fdBZfEWm&KR~Fzu{woXcU#TDaAODs5vH$U3sUQ0v|CRc&|M4IHl?C{Z
z|H=aV$A4u3{^P&00RNTxvH$U3sUQ0v|CRc&|M6d`ANwEw@n2bh|M;&gz<>N#7T`br
zD+};nsUQ0v|CRc&|M6d`ANwEwmHM&&@gM({1^AEu$^!hye`Nvw<G->1|CRc&|M6d`
zANwEwmHM&&@n5MQ`yc=DUs-_v_^&L$fBaV#;6MH=3-Dj5ANwEwmHM&&@n5MQ`yc<6
z`mz7<AODpF_>cd}0{q8+WdZ)<zp?=TmHM&&@n5MQ`yc<6`mz7<U#TDaAOG=RS%Cle
zuPnfS{8tv>KmIEV@L#DP`yc<6`mz7<U#TDaAODs5vH$TO|CI&!kN?U7{MY*L{J$?g
FzX4J`kwX9g

delta 67
zcmcb=T4mc|8D4kyP&Y;f2;iB>>&hswF_@E)WzA7_^X3G`_5?-{W&&bnAZ7t#Rv>2E
Kp1{Z+rT_p&>=0T2

diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
index 5841f73399b76..a8a7e88988260 100644
--- a/tests/test-tokenizer-1.cpp
+++ b/tests/test-tokenizer-1.cpp
@@ -10,10 +10,6 @@
 #include <vector>
 #include <locale>
 
-static std::string vocab_type(llama_context * ctx) {
-    return llama_n_vocab(ctx) == 32000 ? "spm": "bpe";
-}
-
 static std::string escape_whitespace(const std::string& text) {
     std::string result;
     bool escaping = false;
@@ -91,8 +87,8 @@ int main(int argc, char **argv) {
                 return 2;
             }
         } else {
-            if ((vocab_type(ctx) == "spm" && i <= 258) ||
-                (vocab_type(ctx) == "bpe" && (i == 0 || i >= 100000))) {
+            // TODO: needs access to token types
+            if (0 <= i && i < 259) {
                 fprintf(stderr, "%s : info: token %d is string %s and bpe returns tokens %s\n",
                     __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
             } else {
@@ -103,20 +99,28 @@ int main(int argc, char **argv) {
         }
     }
 
-    std::wstring_convert<typename std::codecvt_utf8<wchar_t>, wchar_t> converter;
-    for (wchar_t ch = 0x0000; ch < 0xffff; ++ch) {
-        std::wstring wstr(1, ch);
-        std::string str;
-        try {
-            str = converter.to_bytes(wstr);
-        } catch (std::exception & e) {
-            continue;
+#ifdef _WIN32
+    std::wstring_convert<typename std::codecvt_utf8<char16_t>, char16_t> u16converter;
+    for (char16_t ch = 0x0000; ch < 0xffff; ++ch) {
+        std::u16string u16str(1, ch);
+        std::string str = u16converter.to_bytes(u16str);
+        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
+        if (tokens.size() == 1) {
+            fprintf(stderr, "%s : info: %s tokenized to %d \n",
+                __func__, str.c_str(), tokens[0]);
         }
-        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str), false);
+    }
+
+    std::wstring_convert<typename std::codecvt_utf8<char32_t>, char32_t> u32converter;
+    for (char32_t ch = 0x0000; ch < 0x0010ffff; ++ch) {
+        std::u32string u32str(1, ch);
+        std::string str = u32converter.to_bytes(u32str);
+        std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
         if (tokens.size() == 1) {
             fprintf(stderr, "%s : info: %s tokenized to %d \n", __func__, str.c_str(), tokens[0]);
         }
     }
+#endif
 
     llama_free_model(model);
     llama_free(ctx);

From 0b53b8b08d951da4b24f80e06a7982c3394d66f4 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 21 Aug 2023 19:35:31 +0300
Subject: [PATCH 239/242] llama : add API for token type

ggml-ci
---
 convert.py                                    | 33 +++----
 .../convert-llama2c-to-ggml.cpp               | 44 +++++-----
 .../train-text-from-scratch.cpp               | 33 +++----
 gguf.py                                       | 10 +++
 llama.cpp                                     | 85 +++++++++----------
 llama.h                                       | 26 +++---
 6 files changed, 115 insertions(+), 116 deletions(-)

diff --git a/convert.py b/convert.py
index f680f85963b0e..4ba36f2801891 100755
--- a/convert.py
+++ b/convert.py
@@ -241,17 +241,19 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) ->
             added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
         else:
             added_tokens = {}
+
         vocab_size: int = len(self.bpe_tokenizer)
-        expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
-        actual_ids = sorted(added_tokens.values())
+        expected_ids    = list(range(vocab_size, vocab_size + len(added_tokens)))
+        actual_ids      = sorted(added_tokens.values())
         if expected_ids != actual_ids:
             raise Exception(f"Expected added token IDs to be sequential and start at {len(added_tokens)}; got {actual_ids}")
+
         items = sorted(added_tokens.items(), key=lambda text_idx: text_idx[1])
-        self.added_tokens_list = [text for (text, idx) in items]
+        self.added_tokens_list    = [text for (text, idx) in items]
         self.vocab_size_base: int = vocab_size
-        self.vocab_size: int = self.vocab_size_base + len(self.added_tokens_list)
-        self.fname_tokenizer = fname_tokenizer
-        self.fname_added_tokens = fname_added_tokens
+        self.vocab_size: int      = self.vocab_size_base + len(self.added_tokens_list)
+        self.fname_tokenizer      = fname_tokenizer
+        self.fname_added_tokens   = fname_added_tokens
 
     def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
         tokenizer = self.bpe_tokenizer
@@ -261,12 +263,12 @@ def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for i, item in enumerate(tokenizer):
             text: bytes = item.encode("utf-8")
             score: float = -i
-            yield text, score, 4
+            yield text, score, gguf.TokenType.USER_DEFINED
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
             score = -1000.0
-            yield text.encode("utf-8"), score, 4
+            yield text.encode("utf-8"), score, gguf.TokenType.USER_DEFINED
 
     def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
         yield from self.bpe_tokens()
@@ -304,27 +306,27 @@ def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
             text: bytes = piece.encode("utf-8")
             score: float = tokenizer.get_score(i)
 
-            toktype = 1  # defualt to normal token type
+            toktype = gguf.TokenType.NORMAL
             if tokenizer.is_unknown(i):
-                toktype = 2
+                toktype = gguf.TokenType.UNKNOWN
             if tokenizer.is_control(i):
-                toktype = 3
+                toktype = gguf.TokenType.CONTROL
 
             # NOTE: I think added_tokens are user defined.
             # ref: https://github.com/google/sentencepiece/blob/master/src/sentencepiece_model.proto
-            # if tokenizer.is_user_defined(i): toktype = 4
+            # if tokenizer.is_user_defined(i): toktype = gguf.TokenType.USER_DEFINED
 
             if tokenizer.is_unused(i):
-                toktype = 5
+                toktype = gguf.TokenType.UNUSED
             if tokenizer.is_byte(i):
-                toktype = 6
+                toktype = gguf.TokenType.BYTE
 
             yield text, score, toktype
 
     def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
         for text in self.added_tokens_list:
             score = -1000.0
-            yield text.encode("utf-8"), score, 4
+            yield text.encode("utf-8"), score, gguf.TokenType.USER_DEFINED
 
     def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
         yield from self.sentencepiece_tokens()
@@ -725,6 +727,7 @@ def __init__(self, fname_out: Path) -> None:
         self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
 
     def add_meta_arch(self, params: Params) -> None:
+        self.gguf.add_name                ("llama")
         self.gguf.add_context_length      (params.n_ctx)
         self.gguf.add_embedding_length    (params.n_embd)
         self.gguf.add_block_count         (params.n_layer)
diff --git a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
index af493e15bfc88..469d6e3de3006 100644
--- a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
+++ b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
@@ -139,14 +139,16 @@ void print_sample_weights(TransformerWeights *w){
 struct llama_vocab {
     using id    = int32_t;
     using token = std::string;
+    using ttype = llama_token_type;
 
-    struct token_score {
-        token tok;
+    struct token_data {
+        token text;
         float score;
+        ttype type;
     };
 
     std::unordered_map<token, id> token_to_id;
-    std::vector<token_score> id_to_token;
+    std::vector<token_data> id_to_token;
 };
 
 struct my_llama_hparams {
@@ -516,36 +518,30 @@ void load_vocab(const char *filename, Config *config, struct llama_vocab *vocab)
         struct llama_model * lmodel = llama_load_model_from_file(filename, llama_params);
         struct llama_context * lctx = llama_new_context_with_model(lmodel, llama_params);
 
-        std::vector<const char *> strings;
-        std::vector<float> scores;
-        int n_vocab = llama_n_vocab(lctx);
-        strings.resize(n_vocab, NULL);
-        scores.resize(n_vocab, 0);
-        n_vocab = llama_get_vocab(lctx, strings.data(), scores.data(), n_vocab);
-        GGML_ASSERT(n_vocab == llama_n_vocab(lctx));
+        const int n_vocab = llama_n_vocab(lctx);
         vocab->id_to_token.resize(n_vocab);
         for (int i=0; i<n_vocab; ++i) {
-            std::string tok   = std::string(strings[i]);
-            float       score = scores[i];
-            vocab->id_to_token[i].tok   = tok;
-            vocab->id_to_token[i].score = score;
-            vocab->token_to_id.emplace(tok, i);
+            vocab->id_to_token[i].text  = llama_token_get_text(lctx, i);
+            vocab->id_to_token[i].score = llama_token_get_score(lctx, i);
+            vocab->id_to_token[i].type  = llama_token_get_type(lctx, i);
+            vocab->token_to_id.emplace(vocab->id_to_token[i].text, i);
         }
         llama_free(lctx);
         llama_free_model(lmodel);
     } else { // assume llama2.c vocabulary
         printf("Assuming llama2.c vocabulary since %s is not a ggml file\n", filename);
         llama_file file(filename, "rb");
-        uint32_t n_vocab = config->vocab_size;
+        const int  n_vocab = config->vocab_size;
         /* uint32_t max_token_length =  */ file.read_u32(); // unused
         vocab->id_to_token.resize(n_vocab);
-        for (uint32_t i=0; i<n_vocab; ++i) {
+        for (int i=0; i<n_vocab; ++i) {
             float_t score = file.read_f32();
             uint32_t len = file.read_u32();
-            std::string tok = file.read_string(len);
-            vocab->id_to_token[i].tok = tok;
+            std::string text = file.read_string(len);
+            vocab->id_to_token[i].text = text;
             vocab->id_to_token[i].score = score;
-            vocab->token_to_id.emplace(tok, i);
+            vocab->id_to_token[i].type = LLAMA_TOKEN_TYPE_UNDEFINED;
+            vocab->token_to_id.emplace(text, i);
         }
     }
 }
@@ -611,10 +607,10 @@ void save_as_llama_model(struct llama_vocab * vocab, struct my_llama_model * mod
 //    // write_vocab - for now we are just writing the existing BPE voc. assuming karpathy's vocabulary is the same. idk.
 //    uint32_t n_vocab = model->hparams.n_vocab;
 //    for (uint32_t i = 0; i < n_vocab; i++) {
-//        const auto & token_score = vocab->id_to_token.at(i);
-//        file.write_u32((uint32_t) token_score.tok.size());
-//        file.write_raw(token_score.tok.data(), token_score.tok.size());
-//        file.write_raw(&token_score.score, sizeof(token_score.score));
+//        const auto & token_data = vocab->id_to_token.at(i);
+//        file.write_u32((uint32_t) token_data.tok.size());
+//        file.write_raw(token_data.tok.data(), token_data.tok.size());
+//        file.write_raw(&token_data.score, sizeof(token_data.score));
 //    }
 //
 //    // stuff AK weights into GG weights one by one.
diff --git a/examples/train-text-from-scratch/train-text-from-scratch.cpp b/examples/train-text-from-scratch/train-text-from-scratch.cpp
index 922518da4d380..31d6620a23501 100644
--- a/examples/train-text-from-scratch/train-text-from-scratch.cpp
+++ b/examples/train-text-from-scratch/train-text-from-scratch.cpp
@@ -170,14 +170,16 @@ struct ggml_tensor * randomize_tensor_uniform(struct ggml_tensor * tensor, struc
 struct llama_vocab {
     using id    = int32_t;
     using token = std::string;
+    using ttype = llama_token_type;
 
-    struct token_score {
-        token tok;
+    struct token_data {
+        token text;
         float score;
+        ttype type;
     };
 
     std::unordered_map<token, id> token_to_id;
-    std::vector<token_score> id_to_token;
+    std::vector<token_data> id_to_token;
 };
 
 struct my_llama_hparams {
@@ -2629,10 +2631,10 @@ void save_as_llama_model(struct llama_vocab * vocab, struct my_llama_model * mod
 //    // write_vocab
 //    uint32_t n_vocab = model->hparams.n_vocab;
 //    for (uint32_t i = 0; i < n_vocab; i++) {
-//        const auto & token_score = vocab->id_to_token.at(i);
-//        file.write_u32((uint32_t) token_score.tok.size());
-//        file.write_raw(token_score.tok.data(), token_score.tok.size());
-//        file.write_raw(&token_score.score, sizeof(token_score.score));
+//        const auto & token_data = vocab->id_to_token.at(i);
+//        file.write_u32((uint32_t) token_data.tok.size());
+//        file.write_raw(token_data.tok.data(), token_data.tok.size());
+//        file.write_raw(&token_data.score, sizeof(token_data.score));
 //    }
 //    // write tensors
 //    write_tensor(&file, model->tok_embeddings);
@@ -3055,20 +3057,13 @@ int main(int argc, char ** argv) {
 
     struct llama_vocab vocab;
     {
-        std::vector<const char *> strings;
-        std::vector<float> scores;
-        int n_vocab = llama_n_vocab(lctx);
-        strings.resize(n_vocab, NULL);
-        scores.resize(n_vocab, 0);
-        n_vocab = llama_get_vocab(lctx, strings.data(), scores.data(), n_vocab);
-        GGML_ASSERT(n_vocab == llama_n_vocab(lctx));
+        const int n_vocab = llama_n_vocab(lctx);
         vocab.id_to_token.resize(n_vocab);
         for (int i=0; i<n_vocab; ++i) {
-            std::string tok   = std::string(strings[i]);
-            float       score = scores[i];
-            vocab.id_to_token[i].tok   = tok;
-            vocab.id_to_token[i].score = score;
-            vocab.token_to_id.emplace(tok, i);
+            vocab.id_to_token[i].text  = llama_token_get_text(lctx, i);
+            vocab.id_to_token[i].score = llama_token_get_score(lctx, i);
+            vocab.id_to_token[i].type  = llama_token_get_type(lctx, i);
+            vocab.token_to_id.emplace(vocab.id_to_token[i].text, i);
         }
     }
 
diff --git a/gguf.py b/gguf.py
index 60ee52f095b6c..9776649c76119 100644
--- a/gguf.py
+++ b/gguf.py
@@ -61,6 +61,7 @@
 KEY_TOKENIZER_HF_JSON    = "tokenizer.huggingface.json"
 KEY_TOKENIZER_RWKV       = "tokenizer.rwkv.world"
 
+
 #
 # recommended mapping of model tensor names for storage in gguf
 #
@@ -319,6 +320,15 @@ def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict:
 
     return tensor_map
 
+
+class TokenType(IntEnum):
+    NORMAL       = 1
+    UNKNOWN      = 2
+    CONTROL      = 3
+    USER_DEFINED = 4
+    UNUSED       = 5
+    BYTE         = 6
+
 #
 # implementation
 #
diff --git a/llama.cpp b/llama.cpp
index 1785025f0078b..c97aaee6967e1 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -771,11 +771,12 @@ struct llama_vocab {
 
     using id    = int32_t;
     using token = std::string;
+    using ttype = llama_token_type;
 
     struct token_data {
-        token tok;
+        token text;
         float score;
-        int toktype;
+        ttype type;
     };
 
     llama_vocab_type type = LLAMA_VOCAB_TYPE_SPM;
@@ -1521,12 +1522,12 @@ static void llama_model_load_internal(
             vocab.token_to_id[word] = i;
 
             auto & token_data = vocab.id_to_token[i];
-            token_data.tok = std::move(word);
+            token_data.text  = std::move(word);
             token_data.score = scores[i];
-            token_data.toktype = toktypes[i];
+            token_data.type  = (llama_token_type) toktypes[i];
 
             // determine the newline token: 0x0A == 10 == '\n'
-            if (token_data.tok == "<0x0A>") {
+            if (token_data.text == "<0x0A>") {
                 vocab.linefeed_id = i;
             }
         }
@@ -1558,12 +1559,12 @@ static void llama_model_load_internal(
         LLAMA_LOG_INFO("%s: general.name = %s\n",    __func__, general_name.c_str());
 
         // special tokens
-        if (vocab.special_bos_id != -1) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].tok.c_str() ); }
-        if (vocab.special_eos_id != -1) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].tok.c_str() ); }
-        if (vocab.special_unk_id != -1) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].tok.c_str() ); }
-        if (vocab.special_sep_id != -1) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].tok.c_str() ); }
-        if (vocab.special_pad_id != -1) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].tok.c_str() ); }
-        if (vocab.linefeed_id    != -1) { LLAMA_LOG_INFO( "%s: LF token  = %d '%s'\n", __func__, vocab.linefeed_id,    vocab.id_to_token[vocab.linefeed_id].tok.c_str() );    }
+        if (vocab.special_bos_id != -1) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].text.c_str() ); }
+        if (vocab.special_eos_id != -1) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].text.c_str() ); }
+        if (vocab.special_unk_id != -1) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].text.c_str() ); }
+        if (vocab.special_sep_id != -1) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].text.c_str() ); }
+        if (vocab.special_pad_id != -1) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].text.c_str() ); }
+        if (vocab.linefeed_id    != -1) { LLAMA_LOG_INFO( "%s: LF token  = %d '%s'\n", __func__, vocab.linefeed_id,    vocab.id_to_token[vocab.linefeed_id].text.c_str() );    }
     }
 
     if (vocab_only) {
@@ -2355,15 +2356,27 @@ static enum llama_vocab_type llama_vocab_get_type(const llama_vocab & vocab) {
 }
 
 static bool llama_is_normal_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 1;
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_NORMAL;
 }
 
 static bool llama_is_unknown_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 2;
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_UNKNOWN;
 }
 
 static bool llama_is_control_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 3;
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_CONTROL;
+}
+
+static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_USER_DEFINED;
+}
+
+static bool llama_is_unused_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_UNUSED;
+}
+
+static bool llama_is_byte_token(const llama_vocab & vocab, llama_token id) {
+    return vocab.id_to_token[id].type == LLAMA_TOKEN_TYPE_BYTE;
 }
 
 static bool llama_is_bos_token(const llama_vocab & vocab, llama_token id) {
@@ -2381,22 +2394,10 @@ static bool llama_is_pad_token(const llama_vocab & vocab, llama_token id ) {
     return id == vocab.special_pad_id;
 }
 
-static bool llama_is_user_defined_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 4;
-}
-
-static bool llama_is_unused_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 5;
-}
-
-static bool llama_is_byte_token(const llama_vocab & vocab, llama_token id) {
-    return vocab.id_to_token[id].toktype == 6;
-}
-
 static uint8_t llama_token_to_byte(const llama_vocab & vocab, llama_token id) {
     GGML_ASSERT(llama_is_byte_token(vocab, id));
     const auto& token_data = vocab.id_to_token.at(id);
-    auto buf = token_data.tok.substr(3, 2);
+    auto buf = token_data.text.substr(3, 2);
     return strtol(buf.c_str(), NULL, 16);
 }
 
@@ -2709,6 +2710,7 @@ static std::pair<bool, const llama_grammar_element *> llama_grammar_match_char(
 
     bool found            = false;
     bool is_positive_char = pos->type == LLAMA_GRETYPE_CHAR;
+
     GGML_ASSERT(is_positive_char || pos->type == LLAMA_GRETYPE_CHAR_NOT); // NOLINT
 
     do {
@@ -4957,25 +4959,16 @@ float * llama_get_embeddings(struct llama_context * ctx) {
     return ctx->embedding.data();
 }
 
-int llama_get_vocab(
-        const struct llama_context * ctx,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    return llama_model_get_vocab(&ctx->model, strings, scores, capacity);
+const char * llama_token_get_text(const struct llama_context * ctx, llama_token token) {
+    return ctx->model.vocab.id_to_token[token].text.c_str();
+}
+
+float llama_token_get_score(const struct llama_context * ctx, llama_token token) {
+    return ctx->model.vocab.id_to_token[token].score;
 }
 
-int llama_model_get_vocab(
-        const struct llama_model * model,
-        const char * * strings,
-        float  * scores,
-        int capacity) {
-    int n = std::min(capacity, (int) model->vocab.id_to_token.size());
-    for (int i = 0; i<n; ++i) {
-        strings[i] = model->vocab.id_to_token[i].tok.c_str();
-        scores[i]  = model->vocab.id_to_token[i].score;
-    }
-    return n;
+llama_token_type llama_token_get_type(const struct llama_context * ctx, llama_token token) {
+    return ctx->model.vocab.id_to_token[token].type;
 }
 
 llama_token llama_token_bos(const struct llama_context * ctx) {
@@ -5046,7 +5039,7 @@ int llama_token_to_str(const struct llama_context * ctx, llama_token token, char
 
 int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token, char * buf, int length) {
     if (0 <= token && token < llama_model_n_vocab(&ctx->model)) {
-        std::string result = ctx->model.vocab.id_to_token[token].tok;
+        std::string result = ctx->model.vocab.id_to_token[token].text;
         if (length < (int) result.length()) {
             return -result.length();
         }
@@ -5060,7 +5053,7 @@ int llama_token_to_str_bpe(const struct llama_context * ctx, llama_token token,
 int llama_token_to_str_with_model(const struct llama_model * model, llama_token token, char * buf, int length) {
     if (0 <= token && token < llama_model_n_vocab(model)) {
         if (llama_is_normal_token(model->vocab, token)) {
-            std::string result = model->vocab.id_to_token[token].tok;
+            std::string result = model->vocab.id_to_token[token].text;
             if (llama_vocab_get_type(model->vocab) == LLAMA_VOCAB_TYPE_SPM) {
                 result = llama_unescape_whitespace(result);
             }
diff --git a/llama.h b/llama.h
index 0ea65c1b56710..aa5b7d69ca81a 100644
--- a/llama.h
+++ b/llama.h
@@ -72,6 +72,16 @@ extern "C" {
         LLAMA_VOCAB_TYPE_BPE = 1, // Byte Pair Encoding
     };
 
+    enum llama_token_type {
+        LLAMA_TOKEN_TYPE_UNDEFINED    = 0,
+        LLAMA_TOKEN_TYPE_NORMAL       = 1,
+        LLAMA_TOKEN_TYPE_UNKNOWN      = 2,
+        LLAMA_TOKEN_TYPE_CONTROL      = 3,
+        LLAMA_TOKEN_TYPE_USER_DEFINED = 4,
+        LLAMA_TOKEN_TYPE_UNUSED       = 5,
+        LLAMA_TOKEN_TYPE_BYTE         = 6,
+    };
+
     // model file types
     enum llama_ftype {
         LLAMA_FTYPE_ALL_F32              = 0,
@@ -330,19 +340,11 @@ extern "C" {
     // Vocab
     //
 
-    // Get the vocabulary as output parameters.
-    // Returns number of results.
-    LLAMA_API int llama_get_vocab(
-            const struct llama_context * ctx,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
+    LLAMA_API const char * llama_token_get_text(const struct llama_context * ctx, llama_token token);
 
-    LLAMA_API int llama_model_get_vocab(
-              const struct llama_model * model,
-                          const char * * strings,
-                                 float * scores,
-                                   int   capacity);
+    LLAMA_API float llama_token_get_score(const struct llama_context * ctx, llama_token token);
+
+    LLAMA_API llama_token_type llama_token_get_type(const struct llama_context * ctx, llama_token token);
 
     // Special tokens
     LLAMA_API llama_token llama_token_bos(const struct llama_context * ctx);  // beginning-of-sentence

From 49c25cce19ce24c4ec90352a81fc01c84462301a Mon Sep 17 00:00:00 2001
From: goerch <jhr.walter@t-online.de>
Date: Mon, 21 Aug 2023 19:11:14 +0200
Subject: [PATCH 240/242] tests : use new tokenizer type API (#2692)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)

* Improved tokenizer test

But does it work on MacOS?

* Improve token type support

- Added @klosax code to convert.py
- Improved token type support in vocabulary

* Exclude platform dependent tests

* More sentencepiece compatibility by eliminating magic numbers

* Restored accidentally removed comment

* Improve commentary

* Use token type API in test-tokenizer-1.cpp
---
 convert.py                 | 4 ++--
 tests/test-tokenizer-1.cpp | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)
 mode change 100755 => 100644 convert.py

diff --git a/convert.py b/convert.py
old mode 100755
new mode 100644
index 4ba36f2801891..0428c229f6ddb
--- a/convert.py
+++ b/convert.py
@@ -741,6 +741,8 @@ def add_meta_vocab(self, vocab: Vocab) -> None:
         tokens = []
         scores = []
         toktypes = []
+        # NOTE: `all_tokens` returns the the base vocabulary and added tokens
+        # TODO: add special tokens?
         for text, score, toktype in vocab.all_tokens():
             tokens.append(text)
             scores.append(score)
@@ -751,8 +753,6 @@ def add_meta_vocab(self, vocab: Vocab) -> None:
         self.gguf.add_token_scores(scores)
         self.gguf.add_token_types(toktypes)
 
-        # TODO: added / special tokens
-
     def add_tensor_info(self, name: str, tensor: LazyTensor) -> None:
         n_elements = 1
         for dim in tensor.shape:
diff --git a/tests/test-tokenizer-1.cpp b/tests/test-tokenizer-1.cpp
index a8a7e88988260..d8db7cd96eaa4 100644
--- a/tests/test-tokenizer-1.cpp
+++ b/tests/test-tokenizer-1.cpp
@@ -87,8 +87,8 @@ int main(int argc, char **argv) {
                 return 2;
             }
         } else {
-            // TODO: needs access to token types
-            if (0 <= i && i < 259) {
+            llama_token_type type = llama_token_get_type(ctx, i);
+            if (type == LLAMA_TOKEN_TYPE_UNKNOWN || type == LLAMA_TOKEN_TYPE_CONTROL || type == LLAMA_TOKEN_TYPE_BYTE) {
                 fprintf(stderr, "%s : info: token %d is string %s and bpe returns tokens %s\n",
                     __func__, i, llama_token_to_str(ctx, i).c_str(), unescape_whitespace(ctx, tokens).c_str());
             } else {

From 811f653f95b570eff690ea0b3335f0ff2a7d304f Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 21 Aug 2023 20:40:08 +0300
Subject: [PATCH 241/242] py : cosmetics

---
 convert.py | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/convert.py b/convert.py
index 0428c229f6ddb..c29c032cd23ab 100644
--- a/convert.py
+++ b/convert.py
@@ -255,7 +255,7 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) ->
         self.fname_tokenizer      = fname_tokenizer
         self.fname_added_tokens   = fname_added_tokens
 
-    def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def bpe_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         tokenizer = self.bpe_tokenizer
         from transformers.models.gpt2 import tokenization_gpt2
         byte_encoder = tokenization_gpt2.bytes_to_unicode()
@@ -265,12 +265,12 @@ def bpe_tokens(self) -> Iterable[Tuple[bytes, float]]:
             score: float = -i
             yield text, score, gguf.TokenType.USER_DEFINED
 
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def added_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         for text in self.added_tokens_list:
             score = -1000.0
             yield text.encode("utf-8"), score, gguf.TokenType.USER_DEFINED
 
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def all_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         yield from self.bpe_tokens()
         yield from self.added_tokens()
 
@@ -286,6 +286,7 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) ->
             added_tokens = json.load(open(fname_added_tokens, encoding="utf-8"))
         else:
             added_tokens = {}
+
         vocab_size: int = self.sentencepiece_tokenizer.vocab_size()
         expected_ids = list(range(vocab_size, vocab_size + len(added_tokens)))
         actual_ids   = sorted(added_tokens.values())
@@ -299,7 +300,7 @@ def __init__(self, fname_tokenizer: Path, fname_added_tokens: Optional[Path]) ->
         self.fname_tokenizer = fname_tokenizer
         self.fname_added_tokens = fname_added_tokens
 
-    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         tokenizer = self.sentencepiece_tokenizer
         for i in range(tokenizer.vocab_size()):
             piece = tokenizer.id_to_piece(i)
@@ -323,12 +324,12 @@ def sentencepiece_tokens(self) -> Iterable[Tuple[bytes, float]]:
 
             yield text, score, toktype
 
-    def added_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def added_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         for text in self.added_tokens_list:
             score = -1000.0
             yield text.encode("utf-8"), score, gguf.TokenType.USER_DEFINED
 
-    def all_tokens(self) -> Iterable[Tuple[bytes, float]]:
+    def all_tokens(self) -> Iterable[Tuple[bytes, float, gguf.TokenType]]:
         yield from self.sentencepiece_tokens()
         yield from self.added_tokens()
 
@@ -727,7 +728,7 @@ def __init__(self, fname_out: Path) -> None:
         self.gguf = gguf.GGUFWriter(fname_out, gguf.MODEL_ARCH_NAMES[ARCH])
 
     def add_meta_arch(self, params: Params) -> None:
-        self.gguf.add_name                ("llama")
+        self.gguf.add_name                ("LLaMA")
         self.gguf.add_context_length      (params.n_ctx)
         self.gguf.add_embedding_length    (params.n_embd)
         self.gguf.add_block_count         (params.n_layer)

From 66a66a05a8c509e57e170659182f96165d2e8112 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Mon, 21 Aug 2023 22:11:00 +0300
Subject: [PATCH 242/242] readme : add notice about new file format

ggml-ci
---
 README.md | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 7b72290e4b7fa..82e070ac3ef0d 100644
--- a/README.md
+++ b/README.md
@@ -9,11 +9,17 @@
 
 Inference of [LLaMA](https://arxiv.org/abs/2302.13971) model in pure C/C++
 
-### 🚧 Incoming breaking change + refactoring:
+### Hot topics
 
-See PR https://github.com/ggerganov/llama.cpp/pull/2398 for more info.
+A new file format has been introduced: [GGUF](https://github.com/ggerganov/llama.cpp/pull/2398)
 
-To devs: avoid making big changes to `llama.h` / `llama.cpp` until merged
+Last revision compatible with the old format: [dadbed9](https://github.com/ggerganov/llama.cpp/commit/dadbed99e65252d79f81101a392d0d6497b86caa)
+
+### Current `master` should be considered in Beta - expect some issues for a few days!
+
+### Be prepared to re-convert and / or re-quantize your GGUF models while this notice is up!
+
+### Issues with non-GGUF models will be considered with low priority!
 
 ----